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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
f50bf9a7-0170-4ea9-a4f6-0ee74ae7ea5a | docformer-end-to-end-transformer-for-document | 2106.11539 | null | https://arxiv.org/abs/2106.11539v2 | https://arxiv.org/pdf/2106.11539v2.pdf | DocFormer: End-to-End Transformer for Document Understanding | We present DocFormer -- a multi-modal transformer based architecture for the task of Visual Document Understanding (VDU). VDU is a challenging problem which aims to understand documents in their varied formats (forms, receipts etc.) and layouts. In addition, DocFormer is pre-trained in an unsupervised fashion using carefully designed tasks which encourage multi-modal interaction. DocFormer uses text, vision and spatial features and combines them using a novel multi-modal self-attention layer. DocFormer also shares learned spatial embeddings across modalities which makes it easy for the model to correlate text to visual tokens and vice versa. DocFormer is evaluated on 4 different datasets each with strong baselines. DocFormer achieves state-of-the-art results on all of them, sometimes beating models 4x its size (in no. of parameters). | ['R. Manmatha', 'Yusheng Xie', 'Bhargava Urala Kota', 'Bhavan Jasani', 'Srikar Appalaraju'] | 2021-06-22 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Appalaraju_DocFormer_End-to-End_Transformer_for_Document_Understanding_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Appalaraju_DocFormer_End-to-End_Transformer_for_Document_Understanding_ICCV_2021_paper.pdf | iccv-2021-1 | ['document-image-classification'] | ['computer-vision'] | [-1.26122132e-01 -9.86198038e-02 8.28227252e-02 -2.48685941e-01
-6.84568942e-01 -1.14412677e+00 1.28299510e+00 2.91809112e-01
-1.69558063e-01 8.54025260e-02 7.93523192e-01 -4.15978849e-01
1.24367788e-01 -4.51419950e-01 -1.04182053e+00 -3.92606258e-01
2.53195167e-01 7.91896999e-01 1.26994234e-02 -8.39500576e-02
4.37221289e-01 2.33931348e-01 -1.34512103e+00 1.21011841e+00
3.79387975e-01 8.11136961e-01 6.42072558e-01 1.17071378e+00
-6.47055328e-01 8.92043650e-01 -3.87391448e-01 -2.69664735e-01
-1.77492380e-01 7.51236677e-02 -1.08239591e+00 -3.71931493e-02
1.08648276e+00 -5.02552927e-01 -5.53933442e-01 4.15110230e-01
5.07531345e-01 2.58934528e-01 1.13597929e+00 -9.31081951e-01
-1.73136771e+00 7.54968643e-01 -7.03922510e-01 2.06041589e-01
6.52273417e-01 1.86268508e-01 1.27958846e+00 -1.28537178e+00
9.19826448e-01 1.64140046e+00 2.61311978e-01 4.18794036e-01
-1.28412926e+00 -8.46439004e-02 4.52428132e-01 2.23956451e-01
-9.97587860e-01 -3.18159550e-01 4.66396481e-01 -5.26764393e-01
1.48942149e+00 3.27577800e-01 3.39480698e-01 1.57587194e+00
8.38192925e-02 1.49152601e+00 8.19866180e-01 -6.66578293e-01
-7.46435374e-02 2.99376220e-01 3.41839463e-01 4.89081234e-01
-7.45913312e-02 -3.82471263e-01 -8.45597804e-01 4.05809730e-01
8.34034145e-01 1.13017462e-01 -3.11423182e-01 -4.91122991e-01
-1.49280703e+00 4.36865836e-01 6.33343041e-01 6.75284922e-01
-6.61039874e-02 4.25504655e-01 4.86015201e-01 1.36800393e-01
1.33085683e-01 4.28550988e-01 -3.68745834e-01 -3.57564062e-01
-6.49408042e-01 6.29716367e-02 1.83414683e-01 1.25028324e+00
3.90690804e-01 -1.09776994e-02 -6.85836017e-01 9.09965396e-01
5.46991706e-01 6.67706549e-01 4.03212696e-01 -3.60569686e-01
9.59063828e-01 4.87110198e-01 2.35698789e-01 -7.38189220e-01
-3.47452432e-01 -3.04467052e-01 -6.24117255e-01 1.50801525e-01
3.97544473e-01 2.23437816e-01 -1.38083994e+00 1.27071834e+00
-2.81629235e-01 -3.41473520e-01 -1.86802000e-01 8.02534163e-01
1.12729442e+00 9.04556870e-01 2.70336032e-01 5.45341074e-01
1.42012084e+00 -1.26298916e+00 -8.29493701e-01 -4.48044270e-01
5.59492409e-01 -8.95222723e-01 1.77510202e+00 3.49558979e-01
-1.11780953e+00 -7.31801987e-01 -8.00589144e-01 -8.60854566e-01
-9.70505178e-01 3.60290825e-01 4.88373518e-01 3.39484990e-01
-1.27966082e+00 1.73598245e-01 -6.34281993e-01 -5.75170457e-01
6.25216901e-01 3.08418982e-02 -5.54785609e-01 -2.68595934e-01
-5.48622608e-01 9.33465421e-01 2.31807068e-01 -5.51669486e-02
-7.00557292e-01 -9.03948843e-01 -9.61978078e-01 1.95759967e-01
-8.12492445e-02 -7.52789795e-01 1.27613044e+00 -6.12383246e-01
-1.15582788e+00 9.46025491e-01 -4.99327481e-01 6.57439902e-02
5.22487104e-01 -6.83695078e-01 -3.64418775e-01 8.21267739e-02
-8.61563087e-02 8.39698374e-01 8.27124476e-01 -1.51581728e+00
-2.80886203e-01 -4.74689007e-01 6.73355535e-02 3.09777200e-01
-5.74798644e-01 -2.12112710e-01 -9.45625246e-01 -6.12280428e-01
-2.06565902e-01 -5.21073997e-01 4.30867881e-01 1.93994835e-01
-7.66258121e-01 -3.87392342e-01 1.28348815e+00 -5.21123886e-01
9.66901124e-01 -2.13600302e+00 4.08159643e-01 -5.88190705e-02
2.59082288e-01 8.53440464e-02 -6.00985527e-01 8.34525406e-01
-1.20895825e-01 2.09270462e-01 3.68102908e-01 -9.41640615e-01
4.84041244e-01 -1.37920618e-01 -6.18601203e-01 6.43352643e-02
2.84150634e-02 1.47484386e+00 -5.76273680e-01 -3.59598935e-01
7.85692275e-01 7.31314421e-01 -4.48030591e-01 3.19079280e-01
-5.46421885e-01 5.26816165e-03 -2.14179441e-01 6.78142309e-01
8.17961037e-01 -6.81343794e-01 1.42338619e-01 -6.34926260e-01
-1.07356675e-01 2.35890597e-02 -9.05212402e-01 2.11961579e+00
-9.39041734e-01 1.30914891e+00 -2.91368037e-01 -5.67514658e-01
3.90823275e-01 1.67356044e-01 -4.23276536e-02 -9.68754947e-01
2.14308366e-01 -2.98094213e-01 -5.45526862e-01 -5.94194293e-01
9.28782046e-01 4.46756542e-01 1.69481471e-01 7.40689337e-01
3.05942774e-01 1.54769495e-01 5.93615398e-02 7.76142716e-01
8.26897323e-01 3.40112597e-01 -1.49341265e-03 -1.64980382e-01
-1.15017465e-03 -2.66900420e-01 -7.09438682e-01 9.42401528e-01
2.50016153e-01 9.00680125e-01 3.61565143e-01 -3.92629623e-01
-9.99542594e-01 -1.45517099e+00 -1.28240526e-01 1.50146842e+00
2.32350037e-01 -6.05111837e-01 -1.56463221e-01 -7.90610075e-01
2.12395266e-01 1.05081964e+00 -1.10564530e+00 2.97312886e-01
-1.02219887e-01 -2.56361187e-01 1.44332618e-01 1.14721680e+00
1.44034013e-01 -1.13006628e+00 -5.42655528e-01 -6.55694455e-02
8.89657531e-03 -1.16678882e+00 -6.54159069e-01 3.28571945e-01
-5.08716643e-01 -9.95004654e-01 -8.55975389e-01 -7.65410602e-01
7.94630826e-01 5.49093544e-01 1.39640760e+00 -1.82795182e-01
-1.87403724e-01 9.11936164e-01 -4.27983373e-01 -3.88431132e-01
1.52018994e-01 -3.92681025e-02 -5.29600739e-01 -2.13586003e-01
2.49310970e-01 -2.38865167e-01 -5.78513563e-01 -7.82597139e-02
-1.01990545e+00 2.53682405e-01 5.18397450e-01 8.42144191e-01
3.09035808e-01 -6.47594452e-01 -1.32456526e-01 -9.50019240e-01
6.88317299e-01 -2.32537508e-01 -2.16958106e-01 6.87300801e-01
-5.40736131e-02 1.77090153e-01 6.27122402e-01 -4.15098161e-01
-1.13677013e+00 -3.25095594e-01 3.20181578e-01 -5.08709490e-01
-2.59498060e-01 2.54270226e-01 -2.69301325e-01 2.56485969e-01
6.19548440e-01 3.12888116e-01 -6.03579640e-01 -6.26003206e-01
1.12283778e+00 6.80568993e-01 6.09101117e-01 -3.91882956e-01
7.03047395e-01 5.21282732e-01 -5.03463507e-01 -9.15305018e-01
-7.02909231e-01 -4.35631812e-01 -7.89007306e-01 -1.93071976e-01
9.89191651e-01 -6.92629158e-01 -9.61341441e-01 3.51288885e-01
-1.49818742e+00 -5.70598900e-01 3.09688319e-02 3.82209197e-02
-3.00755531e-01 7.14057088e-02 -4.89626259e-01 -8.36771727e-01
-1.72938019e-01 -8.57627034e-01 1.62441504e+00 1.41854674e-01
-2.89355367e-01 -1.42758715e+00 -8.71752203e-02 3.63181770e-01
4.84846473e-01 3.12212519e-02 9.98094320e-01 -3.23737442e-01
-7.66494691e-01 2.60185629e-01 -8.79804671e-01 -1.30840287e-01
1.01514727e-01 5.41737080e-02 -1.40638828e+00 -3.49735975e-01
-8.82382631e-01 -6.45203829e-01 1.28404081e+00 5.43657005e-01
1.47941875e+00 -6.74480200e-02 -6.26431704e-01 5.95628798e-01
1.48951042e+00 1.85775265e-01 6.13804042e-01 5.46822727e-01
1.18499541e+00 2.65688330e-01 1.47738263e-01 3.84756893e-01
6.54754281e-01 7.82801092e-01 6.12270594e-01 -3.14667314e-01
-2.89554387e-01 -3.22970718e-01 2.53787518e-01 2.71011502e-01
2.44392261e-01 -9.34716284e-01 -1.07802284e+00 8.23319972e-01
-1.97494495e+00 -1.12054706e+00 -2.72969007e-01 1.62925589e+00
3.69411945e-01 -7.14926869e-02 -2.26340055e-01 -1.46627411e-01
1.55337289e-01 4.85357016e-01 -3.75767350e-01 -8.59609425e-01
-4.43119645e-01 9.06898752e-02 1.06276944e-01 3.69555503e-01
-1.21118319e+00 1.05021548e+00 6.63707447e+00 4.82132792e-01
-1.18560171e+00 -5.41676916e-02 5.53231657e-01 -2.40371436e-01
-7.22564936e-01 -3.10283870e-01 -5.49211264e-01 2.91531324e-01
5.71838200e-01 2.05798775e-01 3.90237391e-01 5.36978126e-01
-9.49382223e-03 -2.14409888e-01 -1.36403549e+00 1.22023535e+00
4.64102685e-01 -1.85173929e+00 4.51325864e-01 -3.12844098e-01
7.11300731e-01 3.44391689e-02 6.28436983e-01 1.34819970e-01
4.23263073e-01 -1.44901097e+00 1.06915331e+00 5.93575299e-01
1.01057887e+00 -4.37540978e-01 3.47811431e-01 -1.77276149e-01
-1.19474268e+00 -9.00461618e-03 -8.08359608e-02 2.70251989e-01
2.45911345e-01 1.76046446e-01 -6.64148211e-01 5.48863828e-01
9.99796808e-01 1.03574812e+00 -8.16656232e-01 7.50916660e-01
8.79987180e-02 1.04728304e-01 1.21181300e-02 -1.77790254e-01
4.18320924e-01 2.39573836e-01 2.69687682e-01 1.71100378e+00
2.12056994e-01 -2.96512097e-01 -1.04530662e-01 8.82866681e-01
-1.79779917e-01 9.35753807e-03 -7.36622810e-01 -3.60571355e-01
4.31721687e-01 1.06532598e+00 -3.83305579e-01 -2.57878751e-01
-4.59793657e-01 1.50621414e+00 4.47204232e-01 7.04859734e-01
-6.39032722e-01 -4.26796079e-01 4.21434134e-01 -8.47825501e-03
6.97799444e-01 -3.18621188e-01 -4.66624528e-01 -1.13654959e+00
9.68753621e-02 -5.84167838e-01 3.85309309e-01 -1.53070462e+00
-1.11633694e+00 8.70340526e-01 -2.08735958e-01 -1.08437181e+00
-1.57329544e-01 -1.06068206e+00 -6.32917345e-01 9.15861189e-01
-1.53821540e+00 -1.99280107e+00 -4.69018131e-01 6.35855258e-01
8.66290331e-01 -1.98109359e-01 9.73668635e-01 -3.20970453e-02
-4.85508204e-01 7.32815087e-01 4.16879296e-01 3.93247902e-02
9.82891798e-01 -1.80178607e+00 5.24740160e-01 6.45421922e-01
5.42354941e-01 7.78446794e-01 5.98273218e-01 -2.74153829e-01
-1.72124326e+00 -9.99872983e-01 8.32927883e-01 -1.12695348e+00
6.49215341e-01 -7.89009511e-01 -7.30814397e-01 1.07575333e+00
9.82047200e-01 -2.25240290e-02 6.46873593e-01 4.70446587e-01
-9.33194578e-01 2.62367994e-01 -6.26350880e-01 7.28324950e-01
9.75003958e-01 -9.28909719e-01 -6.15801156e-01 5.08738220e-01
4.10101652e-01 -6.52667344e-01 -4.90774333e-01 -4.63943601e-01
8.79562140e-01 -9.23329413e-01 1.06184602e+00 -9.19866502e-01
8.15799356e-01 -1.89280465e-01 -2.45172560e-01 -1.40724313e+00
-4.93216157e-01 -3.35095316e-01 -3.57717186e-01 1.25818884e+00
4.40477222e-01 -6.33977875e-02 4.26276952e-01 2.81322420e-01
-8.18737000e-02 -3.93171638e-01 -4.97758090e-01 -6.44474626e-01
-5.87070845e-02 -6.29173338e-01 3.82629752e-01 9.47620749e-01
2.00844347e-01 5.01858473e-01 -3.59224975e-01 8.54905881e-03
3.16373110e-01 2.27382302e-01 7.75456309e-01 -8.54089737e-01
-2.65394747e-01 -6.04845166e-01 -1.03523873e-01 -1.44365144e+00
8.56870413e-02 -9.43578660e-01 -2.42514923e-01 -2.20976448e+00
3.72089595e-01 -9.83612612e-02 -2.60602832e-01 6.75569415e-01
4.29718057e-03 1.46628037e-01 5.77864528e-01 5.48865870e-02
-8.78922999e-01 6.43670380e-01 1.43600094e+00 -7.99081147e-01
-1.25181228e-01 -6.50130630e-01 -7.82802522e-01 3.14816654e-01
3.96162122e-01 3.20373088e-01 -5.37484646e-01 -1.27904105e+00
9.54446718e-02 -3.00165921e-01 5.43529570e-01 -7.04083741e-01
3.27660367e-02 -2.09173001e-02 9.87185717e-01 -1.09689486e+00
4.89236742e-01 -8.11418474e-01 -4.40120459e-01 -1.70869008e-01
-5.48048675e-01 3.48682851e-01 6.95039272e-01 5.74129581e-01
-4.03165966e-02 2.11686015e-01 3.53138775e-01 1.42999724e-01
-1.03647697e+00 -2.73667313e-02 -4.02272969e-01 -1.18430175e-01
6.81658685e-01 -3.48554581e-01 -9.31253791e-01 -5.75964928e-01
-6.25232577e-01 3.32884908e-01 3.68082702e-01 1.04511940e+00
7.30290234e-01 -1.33003175e+00 -4.28098917e-01 6.23928681e-02
5.87543845e-01 -1.90201864e-01 5.43660104e-01 4.82148349e-01
-3.46463770e-01 9.53241587e-01 -2.34589234e-01 -6.91803038e-01
-1.26502597e+00 4.93565112e-01 4.73182648e-02 -1.98194489e-01
-7.06938446e-01 1.08916950e+00 4.79067117e-01 -2.83223480e-01
4.08406168e-01 -4.92905974e-01 -3.85622144e-01 1.72763079e-01
8.68840754e-01 -4.45045568e-02 6.12423196e-02 -4.50941324e-01
-4.71034199e-01 7.29390562e-01 -3.94623071e-01 -2.82613069e-01
1.31531382e+00 -1.88636288e-01 1.52217403e-01 5.95455527e-01
1.44844484e+00 -7.80095905e-02 -1.20335102e+00 -1.22812115e-01
-7.60964006e-02 -4.78090495e-01 1.90752298e-01 -1.32275331e+00
-9.53586102e-01 1.26286483e+00 5.70244312e-01 -4.52453718e-02
8.16664755e-01 1.33087233e-01 3.04113656e-01 5.48961282e-01
-1.80875331e-01 -1.04212391e+00 6.12881601e-01 7.69559801e-01
1.44428039e+00 -1.36150861e+00 -3.40857923e-01 9.16988123e-03
-7.93475568e-01 1.22412753e+00 7.75069594e-01 2.71504849e-01
4.14576173e-01 1.75219402e-01 2.73208648e-01 -2.49949247e-01
-9.89633977e-01 -8.17024056e-03 6.97687864e-01 8.91091406e-01
8.76197398e-01 -1.13420278e-01 3.72670621e-01 4.47662979e-01
7.87757046e-04 -4.00999039e-01 4.20683593e-01 8.88515234e-01
-1.40823126e-01 -1.06062102e+00 -3.83705050e-01 5.30099332e-01
1.84084147e-01 -2.56724775e-01 -5.33703685e-01 8.68661344e-01
-2.35805601e-01 8.02372634e-01 4.23549056e-01 -1.43998533e-01
3.80894840e-01 3.62021923e-02 6.70967758e-01 -4.67582494e-01
-5.55206537e-01 7.74966404e-02 3.72152869e-03 -7.17943668e-01
-7.14475811e-02 -6.44053638e-01 -1.12695682e+00 -1.21441558e-01
-6.68872520e-02 -3.74469131e-01 5.79084098e-01 8.90755117e-01
4.52056617e-01 1.00164163e+00 1.15663603e-01 -8.53345394e-01
7.65757412e-02 -1.04483390e+00 -4.11729723e-01 6.54791355e-01
4.37035054e-01 -5.79651296e-01 1.87832370e-01 2.37221867e-01] | [11.339566230773926, 2.110960006713867] |
210a14bf-bdc8-48b7-833d-d16ae19cfbb0 | incorporating-domain-knowledge-in-deep-neural | 2306.00016 | null | https://arxiv.org/abs/2306.00016v1 | https://arxiv.org/pdf/2306.00016v1.pdf | Incorporating Domain Knowledge in Deep Neural Networks for Discrete Choice Models | Discrete choice models (DCM) are widely employed in travel demand analysis as a powerful theoretical econometric framework for understanding and predicting choice behaviors. DCMs are formed as random utility models (RUM), with their key advantage of interpretability. However, a core requirement for the estimation of these models is a priori specification of the associated utility functions, making them sensitive to modelers' subjective beliefs. Recently, machine learning (ML) approaches have emerged as a promising avenue for learning unobserved non-linear relationships in DCMs. However, ML models are considered "black box" and may not correspond with expected relationships. This paper proposes a framework that expands the potential of data-driven approaches for DCM by supporting the development of interpretable models that incorporate domain knowledge and prior beliefs through constraints. The proposed framework includes pseudo data samples that represent required relationships and a loss function that measures their fulfillment, along with observed data, for model training. The developed framework aims to improve model interpretability by combining ML's specification flexibility with econometrics and interpretable behavioral analysis. A case study demonstrates the potential of this framework for discrete choice analysis. | ['Tomer Toledo', 'Omar Mansour', 'Shadi Haj-Yahia'] | 2023-05-30 | null | null | null | null | ['econometrics'] | ['miscellaneous'] | [-3.76820974e-02 3.61268491e-01 -9.87811863e-01 -8.94350231e-01
-3.95321459e-01 -3.58359367e-01 5.25956333e-01 6.22817054e-02
-1.33576632e-01 8.82206202e-01 2.45441571e-01 -6.35513783e-01
-6.46681845e-01 -8.26176465e-01 -4.05461013e-01 -4.49767321e-01
-1.83821499e-01 6.33183241e-01 -4.79205936e-01 -1.81775376e-01
5.68024702e-02 2.71456391e-01 -1.39985490e+00 -4.96179052e-03
1.34334147e+00 9.84816611e-01 2.30233699e-01 1.34196058e-01
2.09712964e-02 8.93549085e-01 1.51953399e-01 -6.24121249e-01
2.06436262e-01 -1.47924766e-01 -6.50544286e-01 4.36315060e-01
-4.63998348e-01 -4.48084205e-01 -2.03070045e-01 6.71657801e-01
-9.25404113e-03 3.55655074e-01 1.09156334e+00 -1.74617434e+00
-8.02092135e-01 7.62881517e-01 8.79106596e-02 -2.36977905e-01
4.87337500e-01 2.06738263e-01 1.28954518e+00 -5.72771668e-01
3.93669516e-01 1.46000004e+00 3.70344639e-01 1.73166946e-01
-1.69640839e+00 -3.45536321e-01 3.19648713e-01 4.55165595e-01
-1.20077491e+00 -3.97747964e-01 8.50330651e-01 -5.40342867e-01
7.07900524e-01 5.53773701e-01 5.16436815e-01 1.13572466e+00
3.88231575e-02 1.11696422e+00 1.37093711e+00 -5.21624506e-01
6.07215464e-01 6.75696075e-01 3.16557556e-01 1.59891665e-01
2.63853163e-01 3.48573774e-01 -9.65373218e-02 -2.20466390e-01
7.33085275e-01 1.33432895e-01 1.66311741e-01 -6.60120964e-01
-6.02309585e-01 1.21639001e+00 1.92742169e-01 -2.82254130e-01
-8.58881831e-01 -1.88797906e-01 9.36206132e-02 5.04813313e-01
4.99440074e-01 3.40608060e-01 -4.28822011e-01 2.64097424e-03
-6.36644483e-01 4.48824048e-01 8.81777048e-01 1.08645749e+00
7.64617383e-01 7.47745782e-02 -2.58810282e-01 7.13792682e-01
9.34645832e-01 2.41420999e-01 3.75582546e-01 -8.29521716e-01
3.71434778e-01 8.31472397e-01 5.31574309e-01 -9.72607493e-01
-2.84714520e-01 -4.79723126e-01 -5.62368989e-01 9.73262936e-02
4.24909234e-01 -1.70343801e-01 -6.82110727e-01 1.72495818e+00
2.66438782e-01 -1.33927211e-01 9.94948000e-02 7.72010088e-01
1.81393608e-01 6.37344301e-01 2.39925534e-01 -1.53003782e-01
1.07504439e+00 -6.40313387e-01 -6.95070744e-01 -1.39461502e-01
6.28180861e-01 -1.83555335e-01 1.16892958e+00 3.22775841e-01
-1.04663539e+00 -4.15649474e-01 -6.35939419e-01 1.77690729e-01
-1.87469825e-01 -2.31571496e-01 1.01174033e+00 1.00824976e+00
-7.14376152e-01 2.19840884e-01 -7.28037059e-01 -1.14612259e-01
1.90939575e-01 6.13763571e-01 2.32553676e-01 -1.83098897e-01
-1.36851287e+00 1.06522179e+00 1.77405581e-01 1.85739368e-01
-8.38294148e-01 -3.13459069e-01 -1.11012459e+00 3.33492905e-01
4.76835489e-01 -6.29274607e-01 1.24250388e+00 -1.18917096e+00
-1.47545028e+00 4.32126433e-01 -1.57942504e-01 -6.42017066e-01
7.02869236e-01 3.03478807e-01 -6.26137614e-01 -3.67098510e-01
1.69123009e-01 3.65610749e-01 6.36395156e-01 -1.37855589e+00
-5.41213334e-01 -1.70613080e-01 2.58943796e-01 9.09188017e-02
8.50262567e-02 -6.42314628e-02 7.32504576e-02 -5.10068357e-01
8.96511301e-02 -1.01018262e+00 -5.30716777e-01 -2.55570382e-01
-2.13895336e-01 -3.68605882e-01 1.19110130e-01 -5.64598322e-01
1.44315422e+00 -2.09250689e+00 -2.65688747e-01 7.26425350e-01
-9.50052366e-02 -2.82520264e-01 4.37538363e-02 6.08652055e-01
1.84898987e-01 4.28839214e-02 -1.93321601e-01 -3.11187953e-01
7.51519084e-01 5.13607204e-01 7.57751102e-03 4.24065858e-01
2.84239769e-01 9.09169912e-01 -6.88728392e-01 -2.44256496e-01
5.28009415e-01 4.29063253e-02 -6.44980013e-01 2.76226342e-01
-3.96890908e-01 4.50745463e-01 -5.67755759e-01 5.01723349e-01
8.65096271e-01 -9.96688157e-02 6.94860578e-01 3.00888121e-01
-2.35746324e-01 4.68866915e-01 -1.58193552e+00 9.79327917e-01
-5.74543715e-01 2.72773355e-01 -9.38312709e-02 -1.43123424e+00
1.04897392e+00 4.04603481e-01 3.48126352e-01 -6.77214563e-01
-4.94793393e-02 1.89579874e-01 1.46496594e-01 -8.49212646e-01
3.13708633e-01 -2.17540756e-01 -3.04311156e-01 3.07383299e-01
-2.29862258e-01 7.51353130e-02 1.08532332e-01 -4.88940358e-01
3.87807548e-01 3.15754801e-01 6.16651177e-01 -4.34578896e-01
5.13275802e-01 1.07997105e-01 7.90192425e-01 6.18611574e-01
2.04273611e-01 2.47847605e-02 5.47111571e-01 -4.12775666e-01
-9.03302491e-01 -9.97290850e-01 -3.54447871e-01 7.98346877e-01
1.98791511e-02 1.87879473e-01 -3.98796409e-01 -2.99765825e-01
2.31864423e-01 1.35117865e+00 -5.21563292e-01 -4.75274548e-02
-1.50967851e-01 -7.32101500e-01 1.67206764e-01 4.80357587e-01
2.42879704e-01 -7.63306260e-01 -4.70597804e-01 4.10858124e-01
-2.26066828e-01 -7.55555034e-01 -7.20729455e-02 1.74926236e-01
-9.04207587e-01 -9.43643808e-01 -3.30079943e-01 -5.96794307e-01
7.74841607e-01 1.66628152e-01 1.01265907e+00 -1.77716970e-01
2.94850707e-01 3.60516310e-01 -3.88775170e-01 -3.51536989e-01
-5.66950142e-01 -1.54824927e-01 1.09501019e-01 4.02753353e-01
5.20490468e-01 -4.74986494e-01 -5.13723433e-01 6.28808320e-01
-1.08836639e+00 5.88301457e-02 6.50208056e-01 8.82738471e-01
1.40614063e-01 2.85271138e-01 9.69121337e-01 -9.66581106e-01
9.84693587e-01 -1.09967458e+00 -7.48562157e-01 3.84053320e-01
-1.30022538e+00 2.89472699e-01 5.62569797e-01 -3.66217345e-01
-1.52536118e+00 -1.91717193e-01 5.67111559e-03 1.62234679e-01
-4.94807929e-01 1.06281269e+00 -3.89143735e-01 2.88218647e-01
3.15770596e-01 7.29801804e-02 1.60995483e-01 -5.12299478e-01
2.01083004e-01 7.38000751e-01 -4.11884487e-02 -8.19080889e-01
3.44586253e-01 3.93482074e-02 6.29578233e-02 -4.90137100e-01
-3.40161681e-01 -4.31973547e-01 -6.74297273e-01 -4.24210280e-02
4.22965944e-01 -8.39628041e-01 -8.03836107e-01 -3.19200270e-02
-4.95392442e-01 -3.22994173e-01 -1.69342339e-01 9.29851055e-01
-7.80891359e-01 6.08318374e-02 -2.88726807e-01 -1.51723742e+00
3.38037252e-01 -1.37423778e+00 3.16931218e-01 1.26351386e-01
-7.10196316e-01 -1.49729598e+00 -2.03947753e-01 5.25270820e-01
2.60932356e-01 1.74481273e-01 1.41278982e+00 -6.63229764e-01
-5.83539486e-01 -2.82488525e-01 5.39062656e-02 8.97008404e-02
1.00652806e-01 -1.48138255e-01 -6.73709512e-01 -5.42332307e-02
2.57670094e-04 -2.00155884e-01 3.02666903e-01 8.54657233e-01
9.79654729e-01 -8.33966851e-01 -1.79233596e-01 2.22817749e-01
1.50987923e+00 6.13572180e-01 5.95101595e-01 3.93581748e-01
-7.57989287e-03 1.02976429e+00 7.73483157e-01 7.58660436e-01
1.06711090e+00 7.79983997e-01 4.75391984e-01 -9.75998268e-02
6.96108282e-01 -5.04313290e-01 2.62535542e-01 3.40237826e-01
-1.36263549e-01 -2.69221306e-01 -6.55600309e-01 5.19598484e-01
-2.15692925e+00 -1.04241073e+00 -3.67399812e-01 2.33115768e+00
4.94292676e-01 2.44975179e-01 5.74207425e-01 1.70336545e-01
4.81294721e-01 -3.28038812e-01 -6.03773654e-01 -8.73138964e-01
9.70187560e-02 -5.50491251e-02 4.31891620e-01 7.35752523e-01
-6.09211147e-01 4.56968307e-01 6.58213854e+00 4.68093544e-01
-5.03332615e-01 -2.65962154e-01 8.97036135e-01 5.12444913e-01
-1.08278120e+00 3.35345834e-01 -5.15869617e-01 5.85139990e-01
9.55209017e-01 -2.15556026e-01 5.08500755e-01 8.74991238e-01
9.79220271e-01 -2.71408886e-01 -1.24409401e+00 4.80626971e-01
-5.50802588e-01 -9.16909337e-01 -1.89070329e-02 4.53568757e-01
7.65856802e-01 -6.44944191e-01 3.09308946e-01 5.21940947e-01
6.75552726e-01 -1.23384738e+00 8.03510487e-01 6.61884069e-01
3.21744323e-01 -7.89409876e-01 9.04409230e-01 7.07858086e-01
-7.08481729e-01 -3.59102249e-01 -3.97225559e-01 -7.35838473e-01
2.33006984e-01 2.89713383e-01 -1.13557100e+00 5.46164572e-01
3.42023522e-02 5.90648413e-01 -1.86642215e-01 9.25111651e-01
7.32698366e-02 7.93662131e-01 -1.20855138e-01 -2.08650440e-01
4.83326405e-01 -6.97289228e-01 1.58695504e-01 9.85585272e-01
3.18223685e-01 -1.05664355e-03 1.97280377e-01 1.28989565e+00
3.52603585e-01 3.04677505e-02 -5.45012295e-01 7.79622570e-02
6.39120638e-01 7.33839571e-01 -3.55776966e-01 -9.51312035e-02
-7.52661765e-01 5.06199598e-01 -1.00681931e-01 5.39282680e-01
-8.12045753e-01 3.39760214e-01 7.00393796e-01 2.39254519e-01
-6.05142713e-02 -2.40984291e-01 -5.01973450e-01 -1.03804290e+00
-1.81013212e-01 -9.54759359e-01 3.82339776e-01 -3.83617580e-01
-1.36255610e+00 3.08544282e-02 4.93190914e-01 -1.27557206e+00
-6.43928289e-01 -4.38934863e-01 -4.47301298e-01 9.46141362e-01
-1.62211490e+00 -1.19218433e+00 3.90352309e-02 5.59722483e-01
7.34926641e-01 9.80478749e-02 8.86828482e-01 1.18085995e-01
-5.49117744e-01 5.27410805e-01 4.71563786e-01 -3.50823224e-01
4.37836237e-02 -1.29459143e+00 1.72454014e-01 4.07670438e-01
-2.57322401e-01 6.14259601e-01 8.91496658e-01 -5.56009829e-01
-1.31263840e+00 -5.86764991e-01 9.79832828e-01 -1.73875347e-01
4.76446748e-01 -3.18726808e-01 -6.12414062e-01 9.21494007e-01
-1.43120289e-01 -6.24001861e-01 1.03422630e+00 3.61811161e-01
2.56003529e-01 1.40278805e-02 -1.38020051e+00 7.26544499e-01
6.14297986e-01 -2.26700932e-01 -5.00289142e-01 -7.08637014e-03
3.09543580e-01 1.20094195e-01 -9.25406158e-01 1.03417121e-01
8.30762386e-01 -9.70710397e-01 9.49426591e-01 -8.29160154e-01
8.32100436e-02 2.90504973e-02 -2.58634031e-01 -1.41276002e+00
-6.49075925e-01 -6.05723739e-01 1.34822980e-01 1.20491087e+00
7.46600091e-01 -7.72275686e-01 6.63318396e-01 1.64285600e+00
5.66823483e-02 -5.63340783e-01 -7.77744114e-01 -7.35740781e-01
-1.00553736e-01 -7.92577326e-01 9.56454396e-01 1.11089861e+00
2.55494237e-01 5.61925508e-02 -6.07094347e-01 6.61417516e-03
6.49297953e-01 1.64317377e-02 6.90997064e-01 -1.44886553e+00
-3.44516486e-01 -2.98988044e-01 -1.83362007e-01 -1.20861089e+00
3.51498693e-01 -7.29394317e-01 -3.11795592e-01 -1.41379499e+00
-7.95785934e-02 -8.35966825e-01 -2.06704706e-01 4.46845107e-02
2.10741475e-01 -5.09987473e-01 -3.66225094e-02 1.16414778e-01
-1.07424155e-01 5.47614515e-01 1.10457242e+00 2.25230996e-02
-5.12039959e-01 8.99698615e-01 -7.26363003e-01 6.39370799e-01
9.15533602e-01 -4.48741585e-01 -9.02477741e-01 -1.42361656e-01
2.85938799e-01 6.06226683e-01 5.14905870e-01 -2.13127583e-01
-6.58771247e-02 -1.01930773e+00 1.59968317e-01 -3.34941715e-01
2.76727736e-01 -1.23405743e+00 7.34191775e-01 3.46028805e-01
-7.68157959e-01 2.26301864e-01 -2.17499763e-01 5.48173189e-01
-1.44409999e-01 -3.60540837e-01 2.98706889e-01 -2.55864650e-01
-8.07171166e-01 1.62088588e-01 -8.55444193e-01 -3.55643243e-01
9.94004726e-01 -5.35455883e-01 3.23620915e-01 -8.76164675e-01
-1.02338195e+00 4.23894852e-01 3.08002532e-01 2.84273893e-01
5.82097590e-01 -1.44811118e+00 -5.68901181e-01 3.65860254e-01
2.99439877e-02 -2.08253890e-01 1.30403921e-01 8.48893344e-01
-2.30550930e-01 9.27032530e-01 -3.23709786e-01 -1.05270758e-01
-5.81152201e-01 6.27833366e-01 8.74155611e-02 -2.57121563e-01
-2.79417187e-01 2.31893629e-01 1.17291592e-01 -6.34127021e-01
3.30474898e-02 -6.61858976e-01 -2.17715427e-01 -7.24716857e-02
-1.24541216e-01 7.43781090e-01 -4.46496844e-01 -6.24218225e-01
-2.17211783e-01 -1.59251660e-01 3.60508084e-01 -2.88449287e-01
1.36208010e+00 -8.65745902e-01 2.01445848e-01 6.06769383e-01
8.17326128e-01 -2.63352096e-01 -1.42396462e+00 -3.46114308e-01
5.32378674e-01 -6.24615133e-01 -6.83326423e-02 -7.44707286e-01
-3.90580714e-01 4.93549377e-01 2.61397153e-01 5.55099249e-01
9.78870213e-01 -2.85150081e-01 3.00542206e-01 2.13321820e-01
3.20694000e-01 -1.27757919e+00 -3.94480318e-01 -3.61868702e-02
5.88061929e-01 -1.59950411e+00 -2.83574045e-01 -3.27387333e-01
-9.51690316e-01 9.36202109e-01 2.52915204e-01 3.23671103e-02
8.02073479e-01 -2.62837052e-01 -1.23500347e-01 1.39202848e-01
-8.19285333e-01 -2.19451323e-01 6.00989580e-01 8.50622952e-01
3.77390414e-01 6.38938367e-01 -5.80821931e-01 9.54508543e-01
-8.22168738e-02 2.11081401e-01 4.43337977e-01 6.17623448e-01
-2.28342831e-01 -1.23194265e+00 -3.59828562e-01 5.59251547e-01
-2.26568401e-01 1.28453821e-02 3.26552019e-02 9.51967955e-01
-6.50518611e-02 1.42109048e+00 -9.44976285e-02 -2.43979752e-01
2.63230652e-01 9.22252759e-02 1.86107692e-03 -4.26839054e-01
-1.88034460e-01 1.52563259e-01 4.58715141e-01 -1.18688688e-01
-5.13126433e-01 -9.86422598e-01 -7.80266583e-01 -5.91630042e-01
-3.66335988e-01 2.91474581e-01 4.36891466e-01 1.24578691e+00
1.64479002e-01 1.34190798e-01 1.09209871e+00 -2.85928875e-01
-9.89956081e-01 -7.62595773e-01 -8.68374407e-01 5.04732013e-01
3.06512058e-01 -8.43564332e-01 -2.27659672e-01 -1.17582001e-01] | [9.059266090393066, 5.496286869049072] |
956b5243-88d4-4da8-a414-711b8cf465cb | a-new-search-paradigm-for-natural-language | null | null | https://openreview.net/forum?id=YeaNQgfFwlQ | https://openreview.net/pdf?id=YeaNQgfFwlQ | A New Search Paradigm for Natural Language Code Search | Code search can accelerate the efficiency of software development by finding code snippets for the given query. The dominant code search paradigm is to learn the semantic matching between code snippets and queries by neural networks. However, this search paradigm causes the gap transferring and expansion between code snippets and queries because researchers utilize pairs of code snippets and code descriptions (e.g., comments and documentation) to train their models and evaluate the trained models on the query which is different from the code description in writing style and application scenario. To remedy the issue, we propose a new simple but effective search paradigm, Query2Desc, which entirely depends on natural language and conducts code search by performing the semantic matching between code descriptions and queries. Experimental results on dataset CoSQA show that the state-of-the-art model CodeBERT gets improvement of 17.48\% in terms of the average MRR when applying it on Query2Desc. Moreover, baseline models on Query2Desc can return the right results in top-$10$ search results for at least 95\% of queries in the test set of CoSQA. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['code-search', 'code-search'] | ['computer-code', 'computer-vision'] | [-2.58955956e-01 -5.19148171e-01 -3.48369867e-01 -3.47383291e-01
-8.46486449e-01 -6.36126816e-01 1.41155720e-01 1.36918485e-01
-2.94633985e-01 -1.84378907e-01 -6.28052503e-02 -4.41022724e-01
-2.26138942e-02 -5.56170821e-01 -6.70492411e-01 2.63647228e-01
2.44923994e-01 2.83656448e-01 5.40892720e-01 -4.85928625e-01
7.90069342e-01 -3.39489698e-01 -1.49185610e+00 7.62827456e-01
1.53325915e+00 8.25787485e-01 1.09064829e+00 4.68021423e-01
-9.04550552e-01 6.76696956e-01 -7.82566369e-01 -8.06368172e-01
5.54500893e-02 -3.05550516e-01 -8.57518733e-01 -6.41069651e-01
3.82425994e-01 -3.69255506e-02 -9.28799883e-02 1.48060083e+00
3.63838106e-01 -4.33815956e-01 2.46332154e-01 -1.05349469e+00
-1.30053532e+00 7.09746659e-01 -6.62315845e-01 2.34387457e-01
7.47970462e-01 -1.67035043e-01 1.13957775e+00 -1.27572048e+00
6.08559370e-01 7.98304141e-01 6.11525655e-01 6.84949636e-01
-6.66940689e-01 -9.17749763e-01 -2.45802790e-01 3.20840478e-01
-1.63737106e+00 -2.59574682e-01 4.82481331e-01 -7.71042705e-01
1.41961110e+00 3.31483305e-01 1.56389832e-01 6.78536892e-01
-7.70400614e-02 8.65085781e-01 1.97400138e-01 -6.73180103e-01
-9.30786878e-03 3.88643384e-01 2.29325384e-01 9.86092865e-01
-5.95705584e-02 -2.10096925e-01 -3.70492131e-01 -2.93188870e-01
2.42325097e-01 9.54065099e-02 -3.20759118e-01 8.49288050e-03
-8.84110212e-01 8.45770001e-01 3.83060873e-01 4.69209999e-01
1.02819189e-01 1.44539818e-01 7.56988585e-01 4.87365603e-01
-5.93109578e-02 9.65498090e-01 -7.24016428e-01 -4.59698290e-01
-1.08578157e+00 3.02405417e-01 8.17924678e-01 1.82592285e+00
8.97029102e-01 -2.32181549e-01 -2.70921767e-01 1.25879121e+00
4.05521840e-01 9.27060485e-01 8.61673474e-01 -6.80074155e-01
1.16803265e+00 1.26162839e+00 -1.13912508e-01 -1.14484560e+00
2.46711791e-01 -6.09230280e-01 -2.50554055e-01 -3.34265918e-01
-2.05042735e-01 2.37120822e-01 -5.84388793e-01 1.31430531e+00
-5.23839772e-01 -2.57370949e-01 1.16161607e-01 6.19493723e-01
1.08564651e+00 5.26534498e-01 -1.44654155e-01 2.76139617e-01
1.22898555e+00 -1.44536614e+00 -5.16084790e-01 -5.19590497e-01
1.03597176e+00 -1.15485454e+00 1.54868114e+00 5.64645082e-02
-8.80370915e-01 -7.77163565e-01 -1.00807834e+00 -6.07832894e-02
-4.62253124e-01 9.81593549e-01 3.82490486e-01 4.35665190e-01
-1.16741204e+00 2.58423477e-01 -6.10434532e-01 -4.45825994e-01
1.68276969e-02 6.33559376e-02 2.24454738e-02 -2.56543696e-01
-1.03518271e+00 4.98857319e-01 3.22278917e-01 -3.55685323e-01
-8.78054976e-01 -8.65463197e-01 -8.58006418e-01 5.22821546e-01
3.26120019e-01 -7.76196420e-02 1.33690882e+00 -1.02235532e+00
-7.33899295e-01 8.94116700e-01 -1.61629930e-01 1.04970355e-02
-2.28064694e-02 -2.81585097e-01 -7.90379226e-01 -6.49815500e-02
6.58842444e-01 3.66048992e-01 4.39670801e-01 -9.49106693e-01
-7.94081867e-01 5.88634871e-02 2.81802386e-01 -2.32323736e-01
-6.13798082e-01 5.40643454e-01 -1.29289043e+00 -5.49324393e-01
-1.70625776e-01 -9.21004415e-01 2.32688099e-01 -1.53492838e-01
-1.73361585e-01 -5.17428756e-01 6.09724820e-01 -8.88929546e-01
2.22892714e+00 -2.43341565e+00 -2.17932358e-01 6.66960794e-03
1.09442912e-01 3.81294489e-01 -4.15016592e-01 5.86896658e-01
-6.94882795e-02 3.23042452e-01 -1.57428861e-01 2.45003104e-02
7.01194257e-02 -2.55699128e-01 -4.52971876e-01 -2.52565771e-01
1.29195705e-01 9.93735790e-01 -9.58374679e-01 -6.45718813e-01
-5.42569518e-01 -1.35934323e-01 -9.51925159e-01 3.31452698e-01
-3.78967166e-01 -3.65819842e-01 -8.43830049e-01 8.64728391e-01
4.83246267e-01 -6.96190417e-01 -2.18517005e-01 2.43103087e-01
-6.55757785e-02 2.86844462e-01 -5.74059367e-01 2.36359358e+00
-8.87462258e-01 6.89411283e-01 -3.01410973e-01 -6.20721579e-01
1.09542680e+00 3.22924554e-01 3.96243334e-02 -1.11011386e+00
-5.52638412e-01 7.75159061e-01 -1.47064894e-01 -1.21090460e+00
3.08159828e-01 5.85322738e-01 -3.64751726e-01 4.76094604e-01
-1.01030320e-01 -8.89405310e-02 3.50952029e-01 2.33348042e-01
1.55583012e+00 9.03484970e-02 -5.35467751e-02 -4.45817024e-01
8.36566567e-01 3.82503718e-01 3.25466245e-01 8.62005472e-01
1.32294223e-01 4.76742417e-01 2.60012686e-01 -2.96625227e-01
-8.43055367e-01 -6.15865588e-01 1.05240554e-01 1.39164925e+00
3.71131629e-01 -7.79263198e-01 -9.29852724e-01 -1.21615434e+00
-6.35329336e-02 6.67455912e-01 -4.86726701e-01 -5.07133424e-01
-7.02364683e-01 -1.04339741e-01 8.03781152e-01 6.42301857e-01
6.33139431e-01 -1.04192054e+00 -6.36938632e-01 6.71762154e-02
-2.84047544e-01 -7.68642485e-01 -1.14656961e+00 -1.83689594e-01
-4.71171379e-01 -1.26100540e+00 -7.67657995e-01 -1.20600116e+00
7.98258007e-01 3.52409959e-01 1.71666455e+00 9.10765290e-01
-1.59998223e-01 1.05726011e-01 -9.09258664e-01 -2.08500803e-01
-6.41553342e-01 3.57828170e-01 -7.61505544e-01 -6.67312622e-01
8.43257248e-01 -3.06345988e-02 -7.33361542e-01 4.27646905e-01
-9.63008642e-01 -2.27042977e-02 7.51273096e-01 6.01494789e-01
-4.99561429e-02 -1.85220748e-01 5.49444437e-01 -6.31433308e-01
8.73130202e-01 -7.52515495e-01 -7.54042745e-01 1.03051221e+00
-1.09508717e+00 3.72394294e-01 8.28960657e-01 -1.32022575e-01
-9.15093601e-01 -1.08685978e-01 -1.07725248e-01 -3.41423780e-01
2.92968959e-01 1.16755974e+00 3.58001053e-01 -2.92865504e-02
1.02963984e+00 6.55333340e-01 -3.31286848e-01 -8.20954204e-01
-1.83757976e-01 1.05714571e+00 4.51504350e-01 -7.34584928e-01
6.07628822e-01 -7.72956684e-02 -8.64769936e-01 -9.53833833e-02
-2.53862590e-01 -9.42773581e-01 -4.43370566e-02 -9.01718810e-03
9.40620184e-01 -9.09298062e-01 -3.02821875e-01 4.58999053e-02
-1.56214297e+00 1.67246073e-01 4.69048679e-01 2.69446522e-01
-1.33582592e-01 3.19964051e-01 -3.17684382e-01 -2.66750783e-01
-4.29837316e-01 -1.69101930e+00 1.35638928e+00 1.45301461e-01
-7.53461346e-02 -6.36128128e-01 2.10634083e-01 2.65543163e-01
7.43346751e-01 -4.61098462e-01 1.53899801e+00 -8.73572409e-01
-6.65268660e-01 -5.98238826e-01 -3.61473680e-01 2.86678523e-01
1.78480837e-02 1.55549824e-01 -5.88024020e-01 -4.23604280e-01
-1.43013269e-01 -3.47482771e-01 4.17945564e-01 -4.18465734e-01
1.51036704e+00 -1.09012440e-01 -4.99085367e-01 5.58791637e-01
1.93792236e+00 8.78427804e-01 5.33124030e-01 2.09556282e-01
4.06510115e-01 4.19746876e-01 6.50415361e-01 3.06711465e-01
4.82888639e-01 9.48276758e-01 4.61336136e-01 2.84182101e-01
-1.04103401e-01 -4.03239578e-01 3.26144993e-01 1.40073192e+00
4.02001768e-01 1.10976992e-03 -1.48212421e+00 8.44770432e-01
-1.59184110e+00 -4.18894440e-01 1.85392201e-02 2.01471233e+00
1.06298888e+00 -5.93151301e-02 -4.34308231e-01 -4.80311304e-01
6.53325438e-01 -3.99564117e-01 -4.35256928e-01 -2.12206140e-01
1.81755006e-01 2.79840499e-01 2.43160248e-01 1.50770918e-01
-6.03977442e-01 8.01740110e-01 5.68236876e+00 1.25740659e+00
-1.01412010e+00 1.54506564e-01 6.99310377e-02 2.96124965e-01
-5.35517573e-01 1.78996116e-01 -7.82984316e-01 9.79649186e-01
7.39925086e-01 -4.52064425e-01 5.52001953e-01 1.46156931e+00
-1.65475860e-01 1.03415951e-01 -1.34827423e+00 1.20297098e+00
3.21179032e-01 -1.50058413e+00 2.50107143e-02 -4.45023566e-01
8.50935340e-01 1.27800226e-01 -2.04889908e-01 1.01255834e+00
2.19687656e-01 -9.51289356e-01 6.66227639e-01 4.12016302e-01
1.04342353e+00 -2.51660168e-01 8.77416372e-01 4.07853276e-01
-1.43408251e+00 -2.76191980e-01 -4.67197329e-01 4.40843523e-01
-6.31616414e-01 -1.90269612e-02 -7.21288502e-01 4.22084004e-01
1.03144634e+00 7.70182312e-01 -1.17732918e+00 1.30450523e+00
5.44220284e-02 3.25850874e-01 2.42974699e-01 -4.95984584e-01
2.63605624e-01 1.06737405e-01 2.58372687e-02 1.43997765e+00
8.83770108e-01 -4.51003820e-01 5.76923117e-02 1.56743264e+00
-1.94305941e-01 2.54877716e-01 -4.85142052e-01 -2.40031704e-01
8.08990598e-01 7.01944888e-01 -3.98591012e-01 -3.61303538e-01
-1.05193400e+00 9.64275062e-01 3.41141820e-01 5.52386463e-01
-8.46641839e-01 -1.18621171e+00 2.94228584e-01 -5.67048863e-02
3.07793558e-01 2.49557465e-01 4.36185822e-02 -1.11038685e+00
7.85470486e-01 -1.28626597e+00 2.77658552e-01 -9.37057257e-01
-8.77238393e-01 7.58154213e-01 -9.65594277e-02 -1.54453599e+00
-2.13895082e-01 -4.10955876e-01 -7.20305085e-01 9.96257484e-01
-1.31093872e+00 -6.69936538e-01 -5.42964518e-01 3.45358521e-01
1.09605157e+00 -7.51869738e-01 8.13683808e-01 8.26786995e-01
-2.07465276e-01 9.36259329e-01 2.65088946e-01 4.92070645e-01
5.31164348e-01 -1.26494598e+00 7.12435722e-01 8.44581842e-01
1.01785503e-01 1.24605536e+00 2.87590832e-01 -7.04415262e-01
-1.47961831e+00 -8.16368818e-01 1.13031268e+00 -5.37476480e-01
6.48830771e-01 -2.96659619e-01 -9.84742403e-01 1.94113538e-01
2.25700155e-01 -9.59503427e-02 5.31323731e-01 -2.38348931e-01
-5.73814809e-01 -7.24193081e-02 -6.77649319e-01 3.92609954e-01
9.43933845e-01 -1.13094842e+00 -5.14206290e-01 3.63038689e-01
1.16688848e+00 -3.74506414e-01 -7.05261290e-01 3.31825137e-01
3.62102866e-01 -8.23543549e-01 7.87930965e-01 -4.56042022e-01
7.80248821e-01 -5.10773122e-01 -3.46586019e-01 -9.09304261e-01
1.52105078e-01 -3.04094434e-01 1.32063270e-01 1.11726558e+00
9.17831659e-01 -4.58821515e-03 6.38230205e-01 5.44728875e-01
-4.90129620e-01 -8.52306008e-01 -5.26439667e-01 -9.34383512e-01
-1.39981997e-03 -3.27561975e-01 9.43316996e-01 1.02304196e+00
3.34463030e-01 7.29060993e-02 2.99889177e-01 -1.71274453e-01
-9.70703065e-02 4.64052588e-01 4.32383120e-01 -8.08802009e-01
-5.97511470e-01 -6.66450202e-01 -9.34756994e-02 -1.49753702e+00
3.08846980e-01 -1.36842942e+00 1.70586094e-01 -1.20389831e+00
4.87452775e-01 -5.98685741e-01 -1.59752205e-01 3.89212757e-01
-2.89718181e-01 -4.63405877e-01 8.46063644e-02 6.28989398e-01
-8.29541147e-01 2.92109817e-01 9.21087682e-01 -5.79954565e-01
4.46675904e-02 1.13807194e-01 -6.96363330e-01 3.01831841e-01
2.78082073e-01 -6.61786139e-01 -7.29109228e-01 -1.22082686e+00
8.52334201e-01 3.68312359e-01 2.03442364e-03 -9.83613789e-01
6.67405784e-01 6.22314513e-02 -3.84995371e-01 -3.33923876e-01
-5.44850171e-01 -9.42435086e-01 -1.06702223e-01 6.68652177e-01
-8.13973725e-01 7.96529055e-01 1.77537471e-01 3.90413344e-01
-6.36011720e-01 -1.18370748e+00 2.69366592e-01 -3.69893610e-01
-9.49740648e-01 1.59062725e-02 1.07185856e-01 5.74263692e-01
5.67855120e-01 -1.69055700e-01 -6.44919753e-01 -5.57148941e-02
-9.50461477e-02 3.88321280e-01 5.02323866e-01 9.06753063e-01
7.91389704e-01 -1.33965719e+00 -6.60215795e-01 2.02459440e-01
9.62797284e-01 -2.94611961e-01 -1.89877748e-01 4.34683472e-01
-8.87221992e-01 8.59188080e-01 1.13713011e-01 -5.20019233e-01
-1.15871346e+00 4.58758414e-01 2.95358062e-01 -1.03671454e-01
-3.89533229e-02 1.19023502e+00 2.18773425e-01 -5.85325062e-01
3.20528239e-01 -4.91089255e-01 -2.27246881e-01 -5.82151473e-01
5.38033366e-01 1.33716866e-01 2.98617899e-01 -1.68432534e-01
-5.53505540e-01 7.37921596e-01 -2.75184274e-01 4.49769765e-01
1.02920902e+00 1.49201930e-01 -4.10091817e-01 -1.25185221e-01
1.74662268e+00 1.10570274e-01 -7.57188857e-01 -2.86990374e-01
6.54054523e-01 -7.62211084e-01 -2.99824595e-01 -1.19515157e+00
-1.28279793e+00 8.81203771e-01 7.17033267e-01 2.38435313e-01
1.08454382e+00 2.79046655e-01 8.13736916e-01 7.74298370e-01
6.19508266e-01 -1.17297792e+00 4.33016986e-01 5.43986917e-01
1.13435698e+00 -1.32985914e+00 -4.64123547e-01 -2.53523141e-01
-4.25489306e-01 1.17646563e+00 1.12984943e+00 1.20863706e-01
5.32478392e-01 1.39307499e-01 1.65157080e-01 -6.83618724e-01
-7.97237456e-01 1.13076814e-01 4.64819640e-01 3.75761628e-01
7.12675691e-01 -3.81854832e-01 -2.42616251e-01 7.25550592e-01
-1.48403635e-02 -1.76378965e-01 2.77281940e-01 1.00833619e+00
-4.12268996e-01 -1.11115074e+00 2.50111341e-01 5.67758083e-01
-5.13449907e-01 -7.80234516e-01 -3.63177627e-01 6.23239636e-01
1.77816898e-01 9.13426042e-01 -1.98564932e-01 -7.43686378e-01
5.21341741e-01 6.83612153e-02 -2.60579586e-01 -1.05925202e+00
-8.59259725e-01 -3.25455964e-01 -2.95800060e-01 -6.86061621e-01
1.82141438e-02 -3.21835488e-01 -1.34426510e+00 2.03768000e-01
-8.12624633e-01 6.49299264e-01 8.17830741e-01 6.34371161e-01
7.14119256e-01 4.19142425e-01 6.48948967e-01 3.39911461e-01
-7.38698125e-01 -8.96345258e-01 -2.69959401e-03 5.82656920e-01
2.26916417e-01 -4.84411508e-01 -2.56724626e-01 2.41726190e-01] | [7.503477573394775, 8.080648422241211] |
3a3ca133-e5a3-4e5d-918d-6e381f08ebb4 | transprompt-towards-an-automatic-transferable | null | null | https://aclanthology.org/2021.emnlp-main.221 | https://aclanthology.org/2021.emnlp-main.221.pdf | TransPrompt: Towards an Automatic Transferable Prompting Framework for Few-shot Text Classification | Recent studies have shown that prompts improve the performance of large pre-trained language models for few-shot text classification. Yet, it is unclear how the prompting knowledge can be transferred across similar NLP tasks for the purpose of mutual reinforcement. Based on continuous prompt embeddings, we propose TransPrompt, a transferable prompting framework for few-shot learning across similar tasks. In TransPrompt, we employ a multi-task meta-knowledge acquisition procedure to train a meta-learner that captures cross-task transferable knowledge. Two de-biasing techniques are further designed to make it more task-agnostic and unbiased towards any tasks. After that, the meta-learner can be adapted to target tasks with high accuracy. Extensive experiments show that TransPrompt outperforms single-task and cross-task strong baselines over multiple NLP tasks and datasets. We further show that the meta-learner can effectively improve the performance on previously unseen tasks; and TransPrompt also outperforms strong fine-tuning baselines when learning with full training sets. | ['Ming Gao', 'Jun Huang', 'Minghui Qiu', 'Jianing Wang', 'Chengyu Wang'] | null | null | null | null | emnlp-2021-11 | ['few-shot-text-classification'] | ['natural-language-processing'] | [ 2.70455480e-01 6.53265938e-02 -4.66138899e-01 -4.91842508e-01
-1.18026507e+00 -4.33017582e-01 9.23601329e-01 3.51186186e-01
-9.23785150e-01 6.82669342e-01 5.58667481e-01 -7.49009550e-02
-6.01070793e-03 -5.60304523e-01 -6.75566316e-01 -3.73770833e-01
2.87008673e-01 5.26314199e-01 5.00972390e-01 -4.83704150e-01
4.07753624e-02 -2.66642839e-01 -1.09544599e+00 5.91336131e-01
7.83695698e-01 6.61579490e-01 4.06195283e-01 8.71987402e-01
-3.41984987e-01 1.00679004e+00 -5.16239822e-01 -5.63605785e-01
-8.59026238e-02 -3.14922720e-01 -1.06147158e+00 -4.83301103e-01
3.20656151e-01 -3.48716974e-01 -3.56919706e-01 6.38493657e-01
7.33349442e-01 8.60208690e-01 9.46144164e-01 -1.17989087e+00
-1.13637972e+00 8.21699619e-01 -3.64984781e-01 4.81317580e-01
1.45700827e-01 3.38595748e-01 1.26507771e+00 -1.23091543e+00
4.94196624e-01 1.23047447e+00 5.59602380e-01 1.06134307e+00
-1.26701868e+00 -7.71322072e-01 3.18203837e-01 1.65397257e-01
-8.90157580e-01 -4.13303137e-01 5.34432173e-01 -2.28718042e-01
1.28905129e+00 -1.58392966e-01 -5.34270564e-03 1.87547851e+00
8.96125436e-02 1.13430965e+00 1.04572725e+00 -7.55831957e-01
2.78774351e-01 1.47081271e-01 6.72329962e-01 5.10355890e-01
4.68316488e-02 6.40827268e-02 -9.20799375e-01 -2.70741463e-01
1.83119819e-01 7.58209378e-02 -2.32893839e-01 -2.21653730e-01
-1.28352761e+00 1.01308584e+00 2.58732259e-01 4.87325460e-01
-1.50507465e-01 1.55445039e-01 6.79203212e-01 3.59631717e-01
8.42967272e-01 8.99760664e-01 -8.70609462e-01 -3.58850032e-01
-6.24257684e-01 1.21602401e-01 8.81118298e-01 9.93721962e-01
7.65722275e-01 -8.98345113e-02 -1.07069492e+00 1.13806629e+00
-9.96178612e-02 3.76759768e-01 9.15078163e-01 -6.34641707e-01
6.84089899e-01 2.37941116e-01 1.53825074e-01 -3.08429271e-01
-3.44596386e-01 -9.61901173e-02 -4.27320540e-01 -2.17966825e-01
6.16700947e-02 -5.46895564e-01 -9.69457865e-01 1.82585883e+00
1.64237335e-01 3.75669897e-01 2.70846754e-01 4.99849141e-01
9.86087263e-01 8.95284474e-01 8.34400177e-01 2.51011718e-02
1.41949379e+00 -1.36536503e+00 -7.17441440e-01 -6.80094659e-01
9.74957407e-01 -3.92440677e-01 1.57883167e+00 -4.79428433e-02
-7.61647165e-01 -6.16844356e-01 -8.59396219e-01 -4.11742210e-01
-5.61435521e-01 -2.19597891e-01 3.07003349e-01 1.78068057e-01
-7.07414865e-01 4.90084797e-01 -5.26829600e-01 -4.62703407e-01
5.43355644e-01 -1.84284851e-01 -1.66258737e-01 -1.92417234e-01
-1.71152008e+00 1.13410997e+00 7.49483109e-01 -6.94507599e-01
-1.27030683e+00 -1.32349920e+00 -9.83561456e-01 3.34493488e-01
4.26418692e-01 -7.68448532e-01 1.89612365e+00 -5.93717277e-01
-1.51445758e+00 8.48677933e-01 -1.54006451e-01 -5.18700778e-01
3.46209824e-01 -5.50708055e-01 -2.32432544e-01 2.70117121e-03
2.94675052e-01 9.44612324e-01 1.03498507e+00 -9.65870619e-01
-8.02525640e-01 7.41770640e-02 -9.67479870e-03 4.48760420e-01
-8.63834381e-01 1.81225482e-02 -1.72924370e-01 -6.37066066e-01
-1.03111923e+00 -6.05768204e-01 -1.37350142e-01 -1.78608999e-01
-3.03306896e-02 -8.62353563e-01 7.83914328e-01 -2.71054536e-01
9.27570879e-01 -2.12395048e+00 -4.96173054e-02 -6.76276028e-01
9.22903121e-02 6.80184245e-01 -7.74480700e-01 5.64690828e-01
1.33881912e-01 3.92292961e-02 -1.45464271e-01 -4.33969051e-01
1.85281590e-01 3.24154407e-01 -6.17258966e-01 -1.08141936e-01
3.53599489e-01 1.57036924e+00 -1.43775129e+00 -3.84443969e-01
1.30180031e-01 3.08229327e-01 -2.08105743e-01 6.29363716e-01
-5.59895575e-01 3.39832753e-02 -5.44334173e-01 8.60325247e-02
1.46274224e-01 -4.54243362e-01 -2.42554277e-01 2.19078258e-01
3.99258673e-01 3.47783536e-01 -4.14279848e-01 2.03068423e+00
-8.07621360e-01 4.66366976e-01 -3.46961468e-01 -1.04589236e+00
9.89121497e-01 5.92063785e-01 8.79424363e-02 -8.12794268e-01
8.08822811e-02 -1.19810365e-01 -2.55084783e-01 -4.85073835e-01
6.78421259e-01 -3.69766653e-01 -2.11608067e-01 9.17613626e-01
6.97587192e-01 -1.70065999e-01 5.75386919e-02 5.59494317e-01
1.24070644e+00 4.47762571e-02 4.92292374e-01 -1.80975184e-01
2.93635223e-02 1.36483843e-02 2.19837695e-01 1.08818316e+00
-6.48085177e-01 3.50817531e-01 8.97276103e-02 -1.75734609e-01
-8.46355617e-01 -9.89172995e-01 1.32923916e-01 2.25510454e+00
-1.55208677e-01 -2.55243331e-01 -3.18600565e-01 -1.20245588e+00
3.29899877e-01 1.10532868e+00 -1.10002017e+00 -6.05612695e-01
-2.69441828e-02 -5.92472792e-01 4.48539227e-01 9.11838651e-01
1.93814039e-01 -1.36724114e+00 -5.45175850e-01 3.95819366e-01
-9.44165140e-02 -1.33807802e+00 -8.05138767e-01 7.19943821e-01
-7.19000220e-01 -8.78694713e-01 -1.07832360e+00 -1.01979208e+00
3.32706362e-01 6.78987324e-01 1.37978983e+00 -2.12580308e-01
-1.80935636e-01 8.47313225e-01 -7.30273128e-01 -6.67686999e-01
-3.13775718e-01 3.60912979e-01 3.88293117e-02 -2.85981745e-01
9.70978618e-01 -3.46541703e-01 -3.34809810e-01 3.03986501e-02
-6.86733425e-01 -1.61013752e-01 2.63047338e-01 1.24873948e+00
3.96843031e-02 -4.34989631e-01 9.30323362e-01 -1.11515999e+00
1.18211102e+00 -7.10253835e-01 -9.11696255e-02 4.21587944e-01
-5.38545847e-01 2.34420672e-01 5.88084877e-01 -8.15473974e-01
-1.54709506e+00 -3.21793318e-01 2.79346377e-01 -4.31072116e-01
-7.11473152e-02 3.68718207e-01 2.48045385e-01 2.47532770e-01
1.19310200e+00 2.64524728e-01 -4.34055090e-01 -3.93371016e-01
9.22137618e-01 8.12061131e-01 3.63550901e-01 -8.35022569e-01
6.46190107e-01 2.78908789e-01 -6.83173358e-01 -7.07613468e-01
-1.70998096e+00 -8.79652202e-01 -5.43765843e-01 1.64304674e-01
8.47133636e-01 -1.12740755e+00 -2.54556477e-01 2.02099040e-01
-1.30985284e+00 -9.83716667e-01 -5.44905543e-01 3.40392470e-01
-5.16127706e-01 2.19011195e-02 -6.49807215e-01 -5.83939254e-01
-7.10691869e-01 -5.58426738e-01 1.11639166e+00 1.82712883e-01
-5.30517459e-01 -1.53579807e+00 6.98526323e-01 1.16568856e-01
5.62999547e-01 -4.90363538e-01 9.05075014e-01 -1.37837946e+00
1.42955959e-01 1.59426838e-01 -2.76393205e-01 3.68289798e-01
1.77081138e-01 -5.39678633e-01 -1.29779398e+00 -3.18099976e-01
-1.04252473e-01 -1.32023168e+00 1.11949897e+00 1.79282174e-01
9.53004718e-01 -2.20914707e-01 -4.77838069e-01 2.66546577e-01
1.07378590e+00 -1.97127476e-01 1.52708128e-01 4.86116767e-01
5.21361113e-01 6.01360202e-01 7.50661433e-01 3.30840200e-01
3.94450694e-01 3.99644166e-01 -1.61805958e-01 1.37407839e-01
-2.85453230e-01 -5.84189057e-01 5.19075036e-01 7.49461234e-01
4.57863599e-01 -1.52049825e-01 -1.02043867e+00 8.53577614e-01
-2.02196932e+00 -1.00082898e+00 4.93506879e-01 1.75476646e+00
1.33938098e+00 7.43747950e-02 -1.07246079e-01 -3.93152595e-01
6.31996036e-01 5.71781933e-01 -7.23087370e-01 -5.74379027e-01
1.66426197e-01 5.35678744e-01 1.34912357e-01 5.12315214e-01
-1.10949481e+00 1.26508999e+00 6.23792171e+00 8.13662469e-01
-8.55399132e-01 7.01529264e-01 3.20361286e-01 -1.60787344e-01
-2.93378413e-01 -1.01516709e-01 -1.13443840e+00 2.69862235e-01
1.05901229e+00 -7.95112729e-01 3.73777524e-02 1.05019283e+00
-2.83321619e-01 3.11418116e-01 -1.34414089e+00 6.27624035e-01
2.89769858e-01 -1.04481983e+00 1.12112619e-01 -4.31798011e-01
1.26238883e+00 4.63967830e-01 -2.73403320e-02 1.34434843e+00
1.02886462e+00 -8.67736697e-01 1.84413403e-01 2.21218571e-01
6.09443128e-01 -4.87755358e-01 4.17669952e-01 6.64127409e-01
-8.64394069e-01 -1.97178945e-01 -7.07164347e-01 -5.48461862e-02
2.33614028e-01 3.14633995e-01 -1.08540642e+00 2.88018018e-01
4.24668014e-01 7.85012603e-01 -5.25079370e-01 6.46297157e-01
-6.67110264e-01 6.77836299e-01 2.60633349e-01 -5.24442792e-01
4.39418375e-01 6.17360890e-01 2.32497975e-01 1.52422953e+00
-1.09667748e-01 2.06458762e-01 4.45136368e-01 7.09810078e-01
-5.78038514e-01 1.96256831e-01 -7.07817554e-01 -2.36689970e-01
6.43600941e-01 1.31737542e+00 1.46948006e-02 -7.51427114e-01
-6.37790143e-01 1.07300782e+00 1.02222443e+00 6.07032478e-01
-4.70069855e-01 -4.75838214e-01 4.42949176e-01 -2.88297921e-01
4.32514131e-01 1.53571321e-02 3.27572301e-02 -1.38403487e+00
-3.35406840e-01 -6.84705436e-01 6.50766313e-01 -8.71449351e-01
-1.94174719e+00 3.98107469e-01 2.62119249e-02 -7.84605205e-01
-2.90078431e-01 -7.38616526e-01 -9.65543389e-01 8.65590155e-01
-1.95093429e+00 -1.16925812e+00 -1.93664163e-01 6.83039725e-01
1.07817411e+00 -2.97137290e-01 1.15572929e+00 -3.92102510e-01
-3.00683260e-01 6.03057981e-01 1.45867139e-01 3.10145199e-01
1.38198376e+00 -1.37500918e+00 6.21928871e-01 6.77035451e-01
3.37355137e-01 5.62636554e-01 5.14133275e-01 -6.91494644e-01
-1.10478234e+00 -1.34021831e+00 9.16140497e-01 -1.02789390e+00
1.02287161e+00 -5.02927065e-01 -1.31330335e+00 1.06328785e+00
6.68960452e-01 3.38592559e-01 8.72622073e-01 6.97956622e-01
-8.60366106e-01 8.24430436e-02 -5.96586227e-01 4.30429012e-01
8.18903685e-01 -8.54076684e-01 -1.58210135e+00 6.19154692e-01
1.13208568e+00 -1.12848908e-01 -6.33305728e-01 1.49391025e-01
1.57447487e-01 -2.75577694e-01 8.17306876e-01 -1.32627416e+00
6.58745110e-01 4.73961979e-01 6.63050860e-02 -1.81535745e+00
-5.65663815e-01 -6.04802668e-01 -5.42612612e-01 1.20129418e+00
5.33978164e-01 -4.50780451e-01 4.72969234e-01 6.03726983e-01
1.52564403e-02 -4.55714256e-01 -7.54972577e-01 -1.14514756e+00
5.70849717e-01 -2.47511029e-01 1.12215281e-01 1.19594848e+00
6.07399523e-01 1.15089095e+00 -4.19136882e-01 -3.35119337e-01
5.85916102e-01 -8.93025920e-02 8.26087773e-01 -1.29023850e+00
-2.42108002e-01 -1.27461001e-01 2.69307107e-01 -9.92265642e-01
7.33671069e-01 -1.23464262e+00 4.00921077e-01 -1.45326173e+00
6.99635267e-01 -1.88327804e-01 -9.08164501e-01 9.35314059e-01
-9.34141517e-01 -2.55912483e-01 2.39240959e-01 7.04784542e-02
-1.13529849e+00 1.02548707e+00 1.24470544e+00 -2.55534291e-01
-2.37671897e-01 -3.63708347e-01 -9.45229471e-01 4.53849405e-01
7.90573418e-01 -8.19125533e-01 -6.23413861e-01 -6.03213251e-01
-2.49049496e-02 -3.23213518e-01 1.41734645e-01 -6.32285535e-01
4.68827724e-01 -8.81328657e-02 2.59251148e-01 -1.06175259e-01
3.82178366e-01 -3.84137303e-01 -1.08881462e+00 3.93372655e-01
-9.64575112e-01 -2.83157140e-01 3.30081135e-01 8.52598071e-01
6.90520108e-02 -5.12113690e-01 7.60811627e-01 -2.33118698e-01
-1.02630746e+00 4.89493430e-01 -1.14674501e-01 9.06050801e-01
1.00647056e+00 3.73619050e-01 -7.02449441e-01 -3.32275718e-01
-4.95599419e-01 5.54816902e-01 8.85718018e-02 9.74257350e-01
4.99224842e-01 -1.28764367e+00 -8.37039113e-01 -2.55864590e-01
8.23078334e-01 -2.10902795e-01 2.68151641e-01 4.53641474e-01
4.95153069e-01 4.89359617e-01 -1.31361157e-01 -4.72706527e-01
-9.60497081e-01 8.93593371e-01 7.20642805e-02 -5.73999226e-01
-6.28278434e-01 1.03150189e+00 3.48754674e-01 -7.01291919e-01
2.89605826e-01 -6.05669850e-03 -1.69107392e-01 2.26529837e-01
1.05762494e+00 5.93889169e-02 -4.64125201e-02 3.17307375e-02
-6.94641396e-02 2.68685669e-01 -7.66835451e-01 -3.71156275e-01
1.34772015e+00 -1.61331128e-02 5.57255447e-01 7.83876061e-01
1.28901613e+00 -3.85171950e-01 -1.56241667e+00 -8.49375308e-01
1.33203343e-01 -1.45245910e-01 4.16094139e-02 -1.13303769e+00
-2.11530298e-01 1.23861134e+00 1.00194559e-01 -1.61688238e-01
5.45656145e-01 7.42505044e-02 8.42011988e-01 7.64037788e-01
1.87415361e-01 -1.29819095e+00 6.97826743e-01 9.98915970e-01
6.94970489e-01 -1.61357462e+00 -3.56968552e-01 1.85575992e-01
-1.11974597e+00 9.10181224e-01 9.37994719e-01 -2.29409616e-02
5.24375796e-01 8.81624594e-03 1.14228837e-01 5.95846632e-03
-1.43113148e+00 -2.71348864e-01 3.26915890e-01 8.24558616e-01
4.95893031e-01 -1.11182943e-01 -1.16821034e-02 9.27526414e-01
2.75778621e-01 2.83183604e-01 2.00030446e-01 1.18532777e+00
-8.48463655e-01 -9.15718734e-01 8.88388380e-02 3.13369453e-01
-2.36902043e-01 -5.22100985e-01 -3.34054053e-01 6.39985263e-01
-4.31071311e-01 8.31839740e-01 -1.11864120e-01 -6.48663640e-02
3.33960295e-01 7.09485114e-01 3.68794471e-01 -1.34048772e+00
-6.76452756e-01 -5.99425256e-01 5.16346507e-02 -3.42370391e-01
-1.58291340e-01 -1.12855777e-01 -9.44874823e-01 1.26807600e-01
-1.92879707e-01 4.80318278e-01 1.61074460e-01 1.16233635e+00
4.17095214e-01 6.75768375e-01 4.66703326e-01 -4.97106522e-01
-1.11519969e+00 -1.33641231e+00 -2.94805557e-01 5.82808137e-01
3.01033974e-01 -6.27286077e-01 -4.23220932e-01 1.21416692e-02] | [10.831937789916992, 7.932953834533691] |
c3bfa77d-692a-490d-bb36-eb071825ee08 | multi-view-silhouette-and-depth-decomposition | 1802.09987 | null | http://arxiv.org/abs/1802.09987v3 | http://arxiv.org/pdf/1802.09987v3.pdf | Multi-View Silhouette and Depth Decomposition for High Resolution 3D Object Representation | We consider the problem of scaling deep generative shape models to
high-resolution. Drawing motivation from the canonical view representation of
objects, we introduce a novel method for the fast up-sampling of 3D objects in
voxel space through networks that perform super-resolution on the six
orthographic depth projections. This allows us to generate high-resolution
objects with more efficient scaling than methods which work directly in 3D. We
decompose the problem of 2D depth super-resolution into silhouette and depth
prediction to capture both structure and fine detail. This allows our method to
generate sharp edges more easily than an individual network. We evaluate our
work on multiple experiments concerning high-resolution 3D objects, and show
our system is capable of accurately predicting novel objects at resolutions as
large as 512$\mathbf{\times}$512$\mathbf{\times}$512 -- the highest resolution
reported for this task. We achieve state-of-the-art performance on 3D object
reconstruction from RGB images on the ShapeNet dataset, and further demonstrate
the first effective 3D super-resolution method. | ['Scott Fujimoto', 'Edward Smith', 'David Meger'] | 2018-02-27 | multi-view-silhouette-and-depth-decomposition-1 | http://papers.nips.cc/paper/7883-multi-view-silhouette-and-depth-decomposition-for-high-resolution-3d-object-representation | http://papers.nips.cc/paper/7883-multi-view-silhouette-and-depth-decomposition-for-high-resolution-3d-object-representation.pdf | neurips-2018-12 | ['3d-object-reconstruction', '3d-object-super-resolution'] | ['computer-vision', 'computer-vision'] | [ 3.59764934e-01 4.46727902e-01 5.48828244e-01 -3.99646878e-01
-1.03129792e+00 -2.40467638e-01 6.10038877e-01 -3.52296233e-01
-1.90332323e-01 6.40336275e-01 1.73598483e-01 1.78749293e-01
5.19876406e-02 -1.35271120e+00 -9.16480482e-01 -3.60836565e-01
-4.00851555e-02 1.17148256e+00 5.23255706e-01 -8.02782848e-02
4.94832881e-02 1.23895812e+00 -1.82153487e+00 5.78511238e-01
2.10151359e-01 8.90193760e-01 2.53854901e-01 7.08858073e-01
-1.52131602e-01 1.91901609e-01 -3.12940508e-01 -3.80028218e-01
5.76692879e-01 -2.27833226e-01 -6.96985245e-01 2.55963266e-01
9.30836916e-01 -7.74725258e-01 -3.35582018e-01 8.22360933e-01
5.91600657e-01 -6.47669062e-02 7.83322871e-01 -5.22415459e-01
-7.38571644e-01 4.85729098e-01 -8.12795997e-01 2.38758996e-01
2.73726553e-01 4.89352159e-02 6.84570849e-01 -1.21589553e+00
1.03788507e+00 1.54921269e+00 6.65382683e-01 7.46523023e-01
-1.70401406e+00 -5.56918561e-01 4.00929153e-03 -2.87395775e-01
-1.33732128e+00 -2.97195196e-01 7.31370091e-01 -3.28524441e-01
1.32318056e+00 8.33763182e-02 8.45387161e-01 9.68486428e-01
9.10525993e-02 2.55453020e-01 1.22823095e+00 -1.40714228e-01
1.62642300e-01 -2.26396605e-01 -3.48138720e-01 6.70034766e-01
3.68624508e-01 2.33987078e-01 -5.22246718e-01 6.37147715e-03
1.99797201e+00 -7.63003454e-02 -9.20314118e-02 -3.59088570e-01
-1.20858014e+00 6.64887547e-01 6.32834971e-01 1.94726139e-02
-5.14668286e-01 4.46010292e-01 -1.78691909e-01 -2.79984653e-01
7.51673579e-01 5.30474603e-01 -2.64006317e-01 3.97366792e-04
-9.49798405e-01 4.35473382e-01 3.76507372e-01 1.19698465e+00
7.06229985e-01 2.67373949e-01 2.38427743e-01 6.28663898e-01
1.93674751e-02 3.72169048e-01 -4.00084397e-03 -1.39802909e+00
1.16911128e-01 4.10799503e-01 4.26336601e-02 -6.10405743e-01
-6.06916070e-01 -4.41061944e-01 -9.79704320e-01 7.98486710e-01
2.58065104e-01 3.27027947e-01 -1.25780237e+00 1.42863321e+00
5.81018925e-01 7.87745938e-02 -2.14107960e-01 9.26403761e-01
9.99884486e-01 5.84263802e-01 -3.51902366e-01 -8.62833709e-02
1.38948548e+00 -4.66745406e-01 -1.25981212e-01 9.44873225e-03
-1.90121263e-01 -6.18816853e-01 1.01551843e+00 4.42317933e-01
-1.82376444e+00 -7.11104572e-01 -9.50200319e-01 -4.57732737e-01
1.01538621e-01 -5.16622245e-01 6.32508874e-01 2.78860152e-01
-1.17656553e+00 9.16469097e-01 -1.06483436e+00 -2.42542215e-02
8.46536815e-01 3.05552453e-01 -3.92861575e-01 -8.79201666e-02
-5.50174296e-01 7.89769590e-01 1.59264445e-01 -3.44525814e-01
-1.05968285e+00 -1.17668414e+00 -6.24488592e-01 -8.59781355e-03
2.65863501e-02 -1.05078602e+00 1.18665004e+00 -2.07643121e-01
-1.26552570e+00 1.24078488e+00 -7.13226274e-02 -3.69132280e-01
5.59812903e-01 9.19129252e-02 7.69732567e-03 4.05052781e-01
1.68097559e-02 9.33778286e-01 8.04616749e-01 -1.67442381e+00
-5.29250920e-01 -8.61659467e-01 1.58385243e-02 2.39336267e-01
1.26628503e-01 -3.74115676e-01 -3.62165719e-01 -5.84338665e-01
7.04246879e-01 -5.85894823e-01 -3.90105605e-01 2.62444735e-01
-1.35922968e-01 1.39646843e-01 5.89833677e-01 -5.11938393e-01
4.22859907e-01 -1.73517597e+00 3.11062515e-01 1.45750612e-01
6.41610205e-01 -2.15935811e-01 7.56789893e-02 -1.68228030e-01
-4.51546274e-02 7.90980011e-02 -2.31433302e-01 -6.49704874e-01
-2.05448881e-01 1.72223896e-01 -1.77020893e-01 2.39852354e-01
1.84886739e-01 9.98684883e-01 -5.45833051e-01 -2.46787474e-01
2.04872102e-01 1.15799141e+00 -6.03472233e-01 -4.10055462e-03
-3.64894807e-01 4.78289664e-01 -3.38312596e-01 5.12697101e-01
9.18274462e-01 -5.01391768e-01 -1.14817001e-01 -5.75559020e-01
-5.00469543e-02 2.65702069e-01 -1.26612830e+00 1.93634343e+00
-3.93739641e-01 2.30552092e-01 2.74759173e-01 -2.84543037e-01
1.06599259e+00 2.18306184e-02 5.67494392e-01 -6.23172641e-01
-6.31931052e-02 5.48050217e-02 -3.96141827e-01 1.42204821e-01
5.74710190e-01 -7.01851308e-01 1.42047107e-01 6.55466795e-01
1.10848941e-01 -9.37312663e-01 -1.64489403e-01 5.57207651e-02
9.42130923e-01 5.29736102e-01 -1.10789262e-01 -1.58682927e-01
-1.18871748e-01 -5.80477566e-02 1.35874981e-02 4.90984917e-01
3.82658213e-01 1.07535541e+00 1.46530852e-01 -7.83419371e-01
-1.81141627e+00 -1.56681132e+00 -4.40643966e-01 6.33129716e-01
2.24968344e-02 -1.49304658e-01 -6.88790858e-01 -2.54645586e-01
2.03744382e-01 6.55909181e-01 -7.13457942e-01 2.01613709e-01
-6.73184812e-01 -7.74066806e-01 3.73844475e-01 8.21048141e-01
4.43627059e-01 -1.03847611e+00 -8.90912950e-01 2.15989977e-01
1.33613825e-01 -1.17052305e+00 2.88638007e-02 3.27835828e-01
-1.30278802e+00 -6.00024879e-01 -8.09186995e-01 -6.33929074e-01
7.86277533e-01 1.00901797e-01 1.45442092e+00 -1.83443382e-01
-7.39501476e-01 3.10057282e-01 1.54775172e-01 -1.65369138e-01
-4.35063869e-01 -1.12868167e-01 2.42570359e-02 -5.09041429e-01
8.79309624e-02 -1.07067299e+00 -6.55853271e-01 7.20576718e-02
-8.55479419e-01 5.77975273e-01 4.59356695e-01 4.30381000e-01
1.31973815e+00 1.22488841e-01 1.16703764e-01 -7.58057177e-01
2.45536998e-01 1.36615992e-01 -8.37601721e-01 -3.04596156e-01
-3.82743686e-01 2.45339632e-01 3.79244298e-01 -2.41448343e-01
-1.24780297e+00 1.41328961e-01 -4.62795645e-01 -6.85927689e-01
-4.20297623e-01 -3.20948452e-01 9.39195976e-02 -1.23321645e-01
9.24507320e-01 2.01096848e-01 -9.94493291e-02 -7.44075537e-01
5.57694018e-01 -1.27017628e-02 6.20235562e-01 -6.22572303e-01
9.07502472e-01 9.84603047e-01 5.73732615e-01 -7.85148084e-01
-7.60006905e-01 2.35405996e-01 -1.02718878e+00 -2.69959737e-02
1.05905390e+00 -9.89813626e-01 -6.86927259e-01 1.90850675e-01
-1.09585452e+00 -4.32644755e-01 -7.77536035e-01 1.90059766e-01
-9.12421286e-01 -2.72074938e-02 -9.09280300e-01 -5.93489647e-01
-2.61636585e-01 -1.10162532e+00 1.42467117e+00 -4.55257706e-02
-1.13776200e-01 -5.83115637e-01 -1.13367990e-01 2.92022943e-01
4.50502127e-01 6.65290594e-01 1.06943679e+00 1.20808557e-01
-1.21493816e+00 1.83865264e-01 -5.36434650e-01 1.07733294e-01
-9.91512686e-02 -1.73808351e-01 -1.08829319e+00 -1.63092270e-01
9.19584632e-02 -3.13631147e-01 8.95176232e-01 5.25207222e-01
1.29943752e+00 -1.84774976e-02 -2.44051337e-01 9.66163099e-01
1.60217476e+00 -1.24116860e-01 7.24700809e-01 1.08891763e-01
8.65455687e-01 3.58407319e-01 1.02008872e-01 5.97679675e-01
2.67832339e-01 7.51029372e-01 7.20322728e-01 -1.40738323e-01
-6.98461354e-01 -3.64258409e-01 -3.93397331e-01 2.50959009e-01
-6.40247047e-01 2.82684803e-01 -8.15526962e-01 3.82970840e-01
-1.17509210e+00 -9.62620676e-01 -2.03188628e-01 2.14228702e+00
8.87926459e-01 3.71835023e-01 2.25333750e-01 -1.88143283e-01
5.44531465e-01 -5.81732905e-03 -7.11144149e-01 -2.63777673e-01
-1.13251448e-01 7.84161389e-01 3.11303377e-01 7.66479075e-01
-6.61170185e-01 9.38743830e-01 6.46481752e+00 7.04254985e-01
-7.74820089e-01 1.58950359e-01 8.05385232e-01 -5.89105606e-01
-7.51793563e-01 -4.73380864e-01 -1.10207629e+00 -2.66868938e-02
6.47434652e-01 2.40150671e-02 5.54220915e-01 8.82286608e-01
-2.31703624e-01 -4.12528822e-03 -1.24398732e+00 1.13130307e+00
4.91030477e-02 -1.70646799e+00 3.12197357e-01 3.17227900e-01
1.02002287e+00 7.66553953e-02 1.40000418e-01 -1.61137193e-01
4.02752727e-01 -1.29344833e+00 7.79368162e-01 4.86173362e-01
1.28288031e+00 -9.96982157e-01 -4.81611229e-02 3.08922142e-01
-1.19754744e+00 2.42113680e-01 -7.97079206e-01 1.49015382e-01
4.63502973e-01 7.85973489e-01 -9.60358202e-01 3.18473488e-01
9.56490993e-01 3.57049704e-01 -2.94681668e-01 7.00642705e-01
-1.21971086e-01 -2.62477160e-01 -6.48497045e-01 3.05045366e-01
-1.73415169e-01 9.48849842e-02 4.63824362e-01 9.98008966e-01
5.61970711e-01 5.77404976e-01 -2.97810882e-01 1.40980911e+00
-3.06106716e-01 -3.22861552e-01 -5.68130195e-01 3.45934272e-01
3.18410248e-01 1.14959013e+00 -1.11011672e+00 -4.17781115e-01
5.31548113e-02 1.08038664e+00 4.85262066e-01 1.56222889e-02
-6.96368277e-01 8.96402523e-02 6.53187156e-01 7.29641080e-01
6.67850316e-01 -5.47438085e-01 -7.30400503e-01 -9.47135329e-01
-2.39645075e-02 -3.66919130e-01 -7.56137597e-04 -1.24580276e+00
-1.18058002e+00 9.94239867e-01 3.00049067e-01 -9.27610099e-01
-2.68211305e-01 -6.35110021e-01 1.91332415e-01 9.59864438e-01
-1.18777847e+00 -1.25721598e+00 -5.38692951e-01 5.23508132e-01
5.14877021e-01 1.96035415e-01 9.72205758e-01 9.02512521e-02
4.02373075e-01 1.89880297e-01 -2.87041903e-01 -3.29003304e-01
2.11736873e-01 -1.11471319e+00 1.00049400e+00 5.73824167e-01
2.57324517e-01 2.01448485e-01 4.86851305e-01 -8.39728773e-01
-1.25716853e+00 -9.18500364e-01 3.67193878e-01 -8.94577384e-01
1.00687794e-01 -5.48962593e-01 -8.23004484e-01 7.46997535e-01
-3.01761091e-01 1.70033753e-01 1.22360066e-01 -2.90893495e-01
-5.51115990e-01 7.89422691e-02 -1.62732565e+00 4.39579666e-01
1.64248371e+00 -3.59293610e-01 -4.86470371e-01 1.96535021e-01
8.41817677e-01 -9.80197310e-01 -1.21404290e+00 5.27031362e-01
6.88219011e-01 -1.39018047e+00 1.51631296e+00 -4.12169278e-01
8.38148534e-01 -1.90031871e-01 -3.08845341e-01 -9.84590769e-01
-7.07550883e-01 -8.51687789e-02 -3.18710297e-01 6.19932771e-01
2.39698812e-01 -2.33763307e-01 1.13121367e+00 6.18840158e-01
-2.43058011e-01 -9.03793514e-01 -1.03725791e+00 -4.90457296e-01
1.49106875e-01 -4.83237475e-01 8.85887086e-01 5.98131537e-01
-7.39999294e-01 1.93841025e-01 5.34135941e-03 1.95607632e-01
1.14376032e+00 3.90101522e-01 5.81535578e-01 -1.37902737e+00
-4.67251092e-01 -5.99188268e-01 -4.10008013e-01 -1.15737510e+00
-2.21103624e-01 -7.29749322e-01 -1.96543097e-01 -1.92016602e+00
2.58781254e-01 -4.23872948e-01 2.93338031e-01 2.26020336e-01
3.42567891e-01 9.25478518e-01 2.91886199e-02 5.29852994e-02
-1.62559077e-01 3.55326086e-01 1.77478421e+00 3.05940390e-01
-1.56623736e-01 -2.15080827e-01 -7.23836422e-01 8.66754234e-01
3.46592546e-01 -2.46782318e-01 -3.48970145e-01 -7.78732657e-01
2.16690823e-01 1.30310267e-01 5.56041896e-01 -9.27192807e-01
-1.46299720e-01 -7.83616975e-02 1.20765257e+00 -1.08859372e+00
1.08203220e+00 -7.64274299e-01 5.91473341e-01 1.47075802e-01
-2.52163261e-01 3.65748741e-02 3.74350041e-01 2.98731297e-01
4.97219235e-01 1.29046842e-01 1.20316362e+00 -7.62854338e-01
-4.76327062e-01 6.07895672e-01 6.33399263e-02 -3.71285416e-02
8.72373462e-01 -4.32875574e-01 -2.43358567e-01 -2.36947030e-01
-1.20842588e+00 -3.91545087e-01 9.38148320e-01 1.35002702e-01
1.12990713e+00 -1.59554815e+00 -7.96919465e-01 5.74824572e-01
-2.91802347e-01 7.43881345e-01 4.59845155e-01 1.80680513e-01
-7.50083864e-01 1.45393029e-01 -5.25156200e-01 -6.83041394e-01
-1.07728314e+00 4.44461048e-01 4.88365382e-01 -1.07706644e-01
-1.15065062e+00 1.31391251e+00 4.93761837e-01 -1.51702926e-01
4.27065715e-02 -3.50317955e-01 3.19530994e-01 -3.04775894e-01
5.67638397e-01 4.06750739e-01 2.00958446e-01 -6.16145790e-01
-1.82878196e-01 9.35186744e-01 2.01999545e-02 -4.55286860e-01
1.86706388e+00 -6.33223653e-02 -1.38505846e-01 2.71647185e-01
1.09857357e+00 -2.31731296e-01 -1.71352887e+00 -2.00858071e-01
-7.70630836e-01 -7.72160530e-01 1.16839983e-01 -7.63953626e-01
-1.03732371e+00 1.01701617e+00 5.82381606e-01 2.08478719e-01
8.80954504e-01 5.01865268e-01 7.06914186e-01 9.23324004e-02
7.88034439e-01 -7.32762516e-01 3.55563313e-01 4.21712756e-01
1.18360090e+00 -8.36124718e-01 2.68632352e-01 -4.33541477e-01
-1.48376435e-01 1.11589217e+00 7.69072592e-01 -5.29426754e-01
6.01584971e-01 6.32156014e-01 -4.92155194e-01 -5.57100117e-01
-5.27185619e-01 -1.53061328e-02 3.34127456e-01 7.93102324e-01
2.45568439e-01 2.76926309e-02 4.35797125e-01 3.53822321e-01
-5.26178062e-01 -2.45656893e-01 5.55761158e-01 6.66200638e-01
-4.79107648e-01 -6.91528022e-01 -5.41371107e-01 4.47110653e-01
-2.66768664e-01 3.49375904e-02 -1.62568569e-01 7.44805157e-01
1.11830704e-01 5.05275913e-02 5.83729684e-01 -1.56319454e-01
5.78968525e-01 -4.47748927e-03 1.39986944e+00 -8.37046266e-01
-3.08762155e-02 3.21242511e-01 -7.81378448e-02 -8.35709035e-01
-2.72864580e-01 -4.74421173e-01 -1.43285143e+00 -4.07093406e-01
2.45891765e-01 -4.67471480e-01 7.64826000e-01 4.12355900e-01
4.03158545e-01 6.35153592e-01 3.32140505e-01 -1.88985479e+00
-2.20498174e-01 -6.23237789e-01 -6.92907631e-01 3.83103937e-01
1.92443684e-01 -7.00258672e-01 -1.92949742e-01 1.05817050e-01] | [8.901082992553711, -3.354342460632324] |
f2a4fc1f-001a-466d-a352-3c17ff17a699 | systemic-formalisation-of-cyber-physical | 2104.05710 | null | https://arxiv.org/abs/2104.05710v1 | https://arxiv.org/pdf/2104.05710v1.pdf | Systemic formalisation of Cyber-Physical-Social System (CPSS): A systematic literature review | The notion of Cyber-Physical-Social System (CPSS) is an emerging concept developed as a result of the need to understand the impact of Cyber-Physical Systems (CPS) on humans and vice versa. This paradigm shift from CPS to CPSS was mainly attributed to the increasing use of sensor-enabled smart devices and the tight link with the users. The concept of CPSS has been around for over a decade and it has gained increasing attention over the past few years. The evolution to incorporate human aspects in the CPS research has unlocked a number of research challenges. Particularly human dynamics brings additional complexity that is yet to be explored. The exploration to conceptualise the notion of CPSS has been partially addressed in few scientific literatures. Although its conceptualisation has always been use-case dependent. Thus, there is a lack of generic view as most works focus on specific domains. Furthermore, the systemic core and design principles linking it with the theory of systems are loose. This work aims at addressing these issues by first exploring and analysing scientific literature to understand the complete spectrum of CPSS through a Systematic Literature Review (SLR). Thereby identifying the state-of-the-art perspectives on CPSS regarding definitions, underlining principles and application areas. Subsequently, based on the findings of the SLR, we propose a domain-independent definition and a meta-model for CPSS, grounded in the Theory of Systems. Finally, a discussion on feasible future research directions is presented based on the systemic notion and the proposed meta-models. | ['Yannick Naudet', 'Hervé Panetto', 'Bereket Abera Yilma'] | 2021-04-11 | null | null | null | null | ['human-dynamics'] | ['computer-vision'] | [ 1.65553689e-01 3.51171702e-01 -2.65663087e-01 3.12862396e-01
1.05770715e-01 -7.17992604e-01 1.04657519e+00 4.49812412e-01
-4.76568304e-02 4.52707052e-01 3.06702197e-01 -4.78066266e-01
-6.40922487e-01 -8.08153629e-01 -3.28176498e-01 -4.78058398e-01
8.91197324e-02 -7.91981351e-03 3.19444895e-01 -5.61121106e-01
2.24073097e-01 5.22017002e-01 -1.83927333e+00 -3.18099469e-01
6.56910837e-01 7.48005986e-01 4.09119099e-01 1.63453832e-01
2.51058992e-02 4.27758962e-01 -6.06541693e-01 -1.40108079e-01
5.72550148e-02 -3.64628792e-01 -6.09016895e-01 -1.11587547e-01
-5.01232147e-01 2.42986143e-01 -1.95837885e-01 8.88834655e-01
2.52781779e-01 -7.62940422e-02 6.58774003e-02 -1.92077398e+00
-5.62260509e-01 1.14298649e-01 -8.96437168e-02 8.11147969e-03
7.71547854e-01 4.28152055e-01 7.32260585e-01 -2.27786735e-01
6.73929751e-01 1.15696597e+00 4.96089518e-01 6.80308491e-02
-1.10132098e+00 -4.63577032e-01 3.87463599e-01 1.84679210e-01
-1.05414832e+00 1.63362265e-01 6.22236252e-01 -6.31445229e-01
1.10764158e+00 4.43509698e-01 1.33622658e+00 1.24710500e+00
3.37081909e-01 3.06134641e-01 1.42066288e+00 -5.58866560e-01
4.35883850e-01 4.19177592e-01 6.12593293e-01 -3.65025818e-01
1.05726743e+00 4.34429884e-01 -2.83009280e-02 2.48216372e-02
3.18878561e-01 -3.34703401e-02 -1.47051485e-02 -4.17593449e-01
-1.06938720e+00 4.37855631e-01 1.41719636e-03 9.87733126e-01
-5.09223223e-01 -1.53879181e-01 6.47355318e-01 2.56146658e-02
7.56430179e-02 4.07072186e-01 -3.92540097e-01 -7.14026988e-01
-3.39540213e-01 4.38554347e-01 9.50392962e-01 7.37106800e-01
2.76792318e-01 -5.26372902e-02 3.01676631e-01 7.31117129e-02
6.93179846e-01 4.64411557e-01 1.89177722e-01 -6.03702068e-01
1.01298332e-01 9.85417902e-01 4.04449403e-01 -1.17054331e+00
-6.54144585e-01 -2.54922718e-01 -4.11565751e-01 1.48832127e-01
1.38255641e-01 -2.37320364e-01 -3.33870947e-01 1.72720361e+00
4.70670044e-01 1.39302835e-01 1.41307801e-01 6.35044754e-01
6.66014433e-01 5.13309419e-01 4.86831516e-01 -7.87159801e-02
1.61462665e+00 -5.16098201e-01 -9.44393039e-01 -1.14479326e-01
3.31541866e-01 -3.55736047e-01 8.96959126e-01 4.61113304e-01
-7.70026267e-01 -5.22756219e-01 -1.18741119e+00 4.21203107e-01
-1.03997064e+00 -3.57489616e-01 4.21419203e-01 1.34864223e+00
-1.01389825e+00 1.70764223e-01 -1.01873922e+00 -1.10824585e+00
-1.03854157e-01 -2.82633267e-02 -3.87685336e-02 1.72259167e-01
-1.47111106e+00 1.60269880e+00 6.45251453e-01 -1.73870265e-01
-3.17533493e-01 -7.54020751e-01 -7.04026639e-01 -1.86529219e-01
2.74889797e-01 -8.67931664e-01 8.84125471e-01 -5.58185458e-01
-1.65129077e+00 4.03521091e-01 2.70166576e-01 -4.91302282e-01
4.20118511e-01 -6.35568947e-02 -1.19096529e+00 -4.76905443e-02
5.28603420e-02 -1.54934481e-01 1.87567174e-01 -1.34878707e+00
-9.91358101e-01 -3.52568358e-01 4.92780775e-01 -1.81613475e-01
-5.88388890e-02 3.57318521e-01 2.14865103e-01 -2.86913455e-01
-6.13428392e-02 -1.14003456e+00 -2.13832349e-01 -7.64254272e-01
-9.46397036e-02 -2.72558123e-01 1.11002314e+00 -3.12787980e-01
1.68070352e+00 -2.04618406e+00 -3.15943398e-02 1.28900439e-01
1.87905371e-01 5.34068108e-01 1.70066163e-01 1.51441133e+00
-4.31282014e-01 6.82620183e-02 8.84690788e-03 1.43861860e-01
4.79546607e-01 2.54647523e-01 -5.94181299e-01 5.90962231e-01
1.52671020e-02 7.24349737e-01 -1.27910304e+00 1.60024852e-01
9.03060973e-01 6.85675144e-01 2.10512411e-02 -9.68019366e-02
2.14109898e-01 6.07552171e-01 -3.53477448e-01 4.85855162e-01
7.39712656e-01 9.54241678e-02 3.23209673e-01 5.23620956e-02
-9.39398527e-01 4.03780311e-01 -1.29942036e+00 1.25316417e+00
-3.29351664e-01 4.42184210e-01 -5.87191656e-02 -1.00486326e+00
6.16173625e-01 8.28440547e-01 7.90553391e-01 -7.89967835e-01
2.65451223e-01 3.59514266e-01 3.13980728e-01 -6.86912656e-01
3.40817720e-01 -1.26233041e-01 -3.87427032e-01 2.31113106e-01
-2.18070611e-01 -3.18975687e-01 2.25079536e-01 2.11682473e-03
1.20476925e+00 5.44078827e-01 8.56042564e-01 -3.92852038e-01
7.89428353e-01 1.19594201e-01 5.34856498e-01 2.84742594e-01
-6.19932353e-01 -1.88170061e-01 5.05105615e-01 -1.15529738e-01
-7.84540594e-01 -9.90795612e-01 -4.17959362e-01 4.83511031e-01
4.31271642e-01 -7.45729387e-01 -7.90100396e-01 -2.58346707e-01
-3.82421538e-02 1.11307812e+00 -4.62921500e-01 -8.24713930e-02
-3.83392394e-01 -4.89233613e-01 4.46189612e-01 1.54509023e-01
3.79823744e-01 -9.67343807e-01 -1.49255228e+00 3.19240957e-01
8.93997028e-02 -1.42102075e+00 8.19742620e-01 -3.03372145e-01
-5.60676575e-01 -1.24871469e+00 -2.43040711e-01 -2.49519721e-01
3.50477427e-01 3.31443071e-01 7.70249724e-01 -2.24180207e-01
-3.20722461e-02 9.06469226e-01 -7.57304788e-01 -9.91672456e-01
-5.24677753e-01 -2.26370111e-01 3.56012642e-01 -4.12381947e-01
4.69619066e-01 -6.91986263e-01 -4.32744056e-01 2.15658963e-01
-1.01723123e+00 -5.22521995e-02 3.48658144e-01 1.06864586e-01
-1.87684953e-01 4.86551821e-01 8.85259330e-01 -2.48821467e-01
9.46302414e-01 -8.16604614e-01 -5.15930891e-01 2.17170231e-02
-6.24513090e-01 -5.76790988e-01 2.67179489e-01 -2.11918801e-01
-1.02945054e+00 -6.38594627e-01 1.28891887e-02 2.22687036e-01
-7.94345140e-01 6.91082299e-01 -3.51811379e-01 7.50157535e-02
3.53414923e-01 -1.20895103e-01 1.51695954e-02 -8.84942040e-02
3.78207564e-01 5.45677483e-01 -2.28105672e-02 -7.15989590e-01
8.44419777e-01 4.41531479e-01 5.95451333e-02 -1.15995240e+00
-2.65778303e-01 -4.07386750e-01 -4.81947124e-01 -6.45409465e-01
8.72627556e-01 -5.16802669e-01 -9.97220218e-01 2.80069351e-01
-1.05303800e+00 -6.40367046e-02 -6.61985159e-01 6.29664123e-01
-4.02779639e-01 3.75090510e-01 -2.39969984e-01 -1.31746769e+00
6.81582093e-02 -1.28620052e+00 6.93828881e-01 1.61199838e-01
-8.14469576e-01 -9.78438675e-01 4.52349126e-01 1.68636620e-01
4.92396593e-01 8.96114230e-01 6.77323818e-01 -4.95098919e-01
-2.73527175e-01 -4.39260632e-01 -2.53704246e-02 1.84128404e-01
4.92312342e-01 1.05193719e-01 -7.01314867e-01 -2.76190013e-01
4.03911352e-01 4.00374204e-01 -2.93709934e-01 1.15684532e-01
1.83472618e-01 -3.79505791e-02 -5.78022182e-01 -3.71304274e-01
1.70919657e+00 7.23287225e-01 8.10721397e-01 8.48695636e-01
1.05267078e-01 1.16547155e+00 1.07198751e+00 4.73226786e-01
5.59842229e-01 6.08666182e-01 5.49029946e-01 3.42485458e-01
-1.00975465e-02 -1.04007423e-01 3.05066317e-01 9.38096046e-01
-3.80757928e-01 -2.66757309e-02 -1.23177469e+00 6.92096353e-01
-1.98156261e+00 -8.62912059e-01 -4.77846503e-01 2.09575081e+00
3.97274317e-03 2.97341943e-01 4.76340473e-01 4.28581744e-01
7.34966457e-01 2.37200838e-02 -8.61790478e-02 -3.33757162e-01
2.68488467e-01 8.31049830e-02 2.85995811e-01 1.51888549e-01
-7.87192941e-01 6.50460720e-01 6.32531548e+00 2.46948093e-01
-9.71842825e-01 2.62763768e-01 -3.30943286e-01 4.28746253e-01
-1.85423922e-02 4.81840461e-01 -4.74505812e-01 6.87761188e-01
1.30667222e+00 -5.92197657e-01 -2.45493557e-02 6.19064271e-01
7.64587343e-01 -3.16228718e-01 -7.00728595e-01 5.49152076e-01
-4.32362229e-01 -7.07508385e-01 -3.74908715e-01 4.90337670e-01
6.42567337e-01 -8.24886933e-02 1.21909641e-02 2.89757967e-01
1.81476116e-01 -4.77786869e-01 1.12458992e+00 3.77045482e-01
2.12923557e-01 -5.42144597e-01 7.39070773e-01 4.19091940e-01
-1.48650384e+00 -4.42048639e-01 2.53910750e-01 -5.81063986e-01
8.04515362e-01 4.77128446e-01 -3.30046088e-01 1.14085853e+00
6.46142542e-01 4.76518840e-01 -5.46727963e-02 1.04242361e+00
-3.69510621e-01 5.10345876e-01 -2.26786882e-01 -1.89346284e-01
2.85129696e-01 -5.74629188e-01 8.97856772e-01 1.31421280e+00
3.95148128e-01 2.11489737e-01 -1.83298122e-02 9.83873904e-01
1.22205222e+00 -1.42098799e-01 -8.45987737e-01 -4.05225277e-01
5.82941592e-01 1.21836078e+00 -9.53181505e-01 -1.51765525e-01
-7.66454041e-01 1.90586790e-01 -5.99685669e-01 2.95474261e-01
-1.01403952e+00 -2.83378124e-01 1.06758213e+00 4.87521708e-01
-7.52321333e-02 -5.04949629e-01 -4.17037129e-01 -7.00236678e-01
-1.60975337e-01 -8.46768796e-01 3.78716877e-03 -5.38414538e-01
-1.10456157e+00 2.77699590e-01 6.35760844e-01 -1.45158291e+00
-1.60933182e-01 -4.99847472e-01 -4.09477472e-01 6.35053217e-01
-1.11649334e+00 -1.35539329e+00 -3.16391200e-01 8.36503580e-02
-3.14069577e-02 2.95651972e-01 1.04393172e+00 1.99553683e-01
-5.85026860e-01 -3.16182435e-01 -2.27179497e-01 -5.30029714e-01
1.64831072e-01 -9.11312938e-01 5.73228955e-01 1.00454545e+00
-8.79493773e-01 1.07017100e+00 1.24540675e+00 -1.11041117e+00
-1.52675700e+00 -4.45403904e-01 8.70823741e-01 -6.60394430e-01
1.22506678e+00 -2.42697835e-01 -4.89946723e-01 7.71979094e-01
7.07345486e-01 -5.33384323e-01 8.84232283e-01 -1.16537608e-01
9.67467495e-04 1.16609991e-01 -1.49962437e+00 8.17669570e-01
1.08272195e+00 -3.57432812e-01 -9.73565280e-01 -1.50070116e-01
9.11413968e-01 -4.50586230e-02 -1.11772728e+00 4.05817628e-01
4.76170897e-01 -9.69474494e-01 9.52429712e-01 -3.13165307e-01
-2.25066096e-01 -6.07048273e-01 -1.12660035e-01 -1.12956047e+00
-3.52837712e-01 -6.41516924e-01 -1.49848640e-01 1.26676977e+00
-1.60405993e-01 -1.53902435e+00 2.72806436e-01 9.86583531e-01
-3.78115028e-01 -4.62052733e-01 -9.51333642e-01 -1.03759325e+00
-1.51953503e-01 -8.62714767e-01 9.68419135e-01 1.22433197e+00
9.49988306e-01 -1.31079122e-01 3.66940379e-01 1.31204396e-01
4.41741854e-01 -3.31547350e-01 8.43239188e-01 -1.64045715e+00
5.31213060e-02 -5.90373695e-01 -9.34042931e-01 -3.03279102e-01
-2.25292042e-01 -4.54517931e-01 -4.58161682e-01 -1.96939766e+00
-3.13565165e-01 -1.25141397e-01 -1.82060257e-01 3.99957672e-02
4.02037591e-01 -3.91866326e-01 6.87382579e-01 -4.44256477e-02
-1.79142311e-01 2.89692849e-01 8.11077058e-01 4.32707846e-01
-3.57773334e-01 -1.17118366e-01 -8.35030258e-01 8.84068787e-01
9.03358340e-01 -2.39107206e-01 -5.80184877e-01 4.54992056e-01
5.98939002e-01 -2.92330924e-02 4.53228563e-01 -1.36430562e+00
1.28198102e-01 -4.13897842e-01 -4.37548995e-01 -6.77610636e-01
2.58586168e-01 -1.53429830e+00 8.57599437e-01 8.46888781e-01
4.11524951e-01 3.59259456e-01 3.60928595e-01 4.45205152e-01
-9.99230072e-02 -1.78765744e-01 4.06456530e-01 6.87804222e-02
-5.98953724e-01 -2.54991442e-01 -8.51116300e-01 -3.55833590e-01
1.63209450e+00 -6.45235479e-01 -3.22714537e-01 6.12337282e-03
-6.58727765e-01 -8.02975968e-02 4.99450624e-01 8.33470643e-01
8.84143263e-02 -1.08587432e+00 -1.04913432e-02 -1.55732810e-01
1.36309355e-01 -5.03197432e-01 3.89456183e-01 8.01942706e-01
-3.11040342e-01 1.19379938e+00 -4.85452950e-01 -3.24943304e-01
-7.68795013e-01 8.18879128e-01 -1.16303489e-01 -1.43360943e-01
-8.21570277e-01 -2.08332390e-01 5.59637584e-02 -4.11882013e-01
-8.81171133e-03 -7.25085318e-01 -4.23563242e-01 1.54524952e-01
4.37073201e-01 1.05125999e+00 -2.08558604e-01 -1.00497413e+00
-5.72893620e-01 4.84992325e-01 7.88667619e-01 -4.76179600e-01
1.41037011e+00 -3.56475979e-01 -3.55367631e-01 6.85864449e-01
6.78693414e-01 -7.94781595e-02 -7.02238917e-01 2.69074142e-01
3.55364323e-01 -1.96083680e-01 -4.28007454e-01 -8.19663346e-01
-4.39176083e-01 3.20293695e-01 6.62002623e-01 1.04336691e+00
9.10289109e-01 -1.05307668e-01 5.45090139e-01 -2.75626004e-01
9.51650262e-01 -1.36615539e+00 -2.76344508e-01 4.38801378e-01
9.72777545e-01 -4.54907477e-01 -1.07634582e-01 -8.44033659e-01
-3.58730137e-01 8.78597736e-01 3.35512340e-01 -2.26111457e-01
1.18737590e+00 5.00537395e-01 -5.55281222e-01 -3.75950634e-01
-1.59938127e-01 -4.45044249e-01 -2.26365194e-01 9.97550309e-01
3.82815242e-01 4.49227154e-01 -1.02353692e+00 1.02321720e+00
-3.84180099e-01 3.47485065e-01 4.89518195e-01 1.27631545e+00
-2.65639812e-01 -1.42472351e+00 -7.09360063e-01 -1.89964101e-01
-2.16253921e-01 5.78094184e-01 -1.78929418e-01 1.24801826e+00
4.23360795e-01 1.43073189e+00 -2.42004246e-01 -5.13574123e-01
9.87607121e-01 -8.38778540e-02 3.34679514e-01 -3.52464795e-01
-8.26235592e-01 -4.03539389e-01 1.88716531e-01 -5.04203498e-01
-7.90366292e-01 -1.14588368e+00 -1.22942638e+00 -3.91692758e-01
-2.42052689e-01 3.04905735e-02 1.01145017e+00 1.13842595e+00
5.34163594e-01 9.54237938e-01 6.86315000e-02 -7.87441194e-01
-1.25226825e-01 -7.45773077e-01 -5.78946412e-01 7.65033364e-02
-7.96900541e-02 -9.04980898e-01 -2.33584031e-01 -4.46256965e-01] | [8.859177589416504, 6.649916172027588] |
478248df-35cf-4432-9531-8456a575f3b8 | controllable-lexical-simplification-for | 2302.02900 | null | https://arxiv.org/abs/2302.02900v1 | https://arxiv.org/pdf/2302.02900v1.pdf | Controllable Lexical Simplification for English | Fine-tuning Transformer-based approaches have recently shown exciting results on sentence simplification task. However, so far, no research has applied similar approaches to the Lexical Simplification (LS) task. In this paper, we present ConLS, a Controllable Lexical Simplification system fine-tuned with T5 (a Transformer-based model pre-trained with a BERT-style approach and several other tasks). The evaluation results on three datasets (LexMTurk, BenchLS, and NNSeval) have shown that our model performs comparable to LSBert (the current state-of-the-art) and even outperforms it in some cases. We also conducted a detailed comparison on the effectiveness of control tokens to give a clear view of how each token contributes to the model. | ['Horacio Saggion', 'Daniel Ferrés', 'Kim Cheng SHEANG'] | 2023-02-06 | null | null | null | null | ['lexical-simplification'] | ['natural-language-processing'] | [-1.94141790e-01 3.39309484e-01 1.32675260e-01 -4.25522476e-01
-9.39858317e-01 -4.03478354e-01 6.77674890e-01 9.83977690e-02
-4.87425685e-01 9.33753312e-01 5.59450209e-01 -3.44510198e-01
1.19208321e-01 -6.01636887e-01 -5.54230869e-01 -1.74114078e-01
5.40879846e-01 8.36525202e-01 1.80777147e-01 -1.04238284e+00
2.02231228e-01 3.15587401e-01 -1.12531269e+00 4.33595926e-01
1.23848212e+00 2.99760282e-01 -4.77069104e-03 4.80909437e-01
-2.84853995e-01 8.29546034e-01 -1.16097319e+00 -1.13099873e+00
1.33293033e-01 -2.47530341e-01 -8.98571551e-01 -5.02945960e-01
7.91078508e-01 -2.52089854e-02 -3.28819662e-01 8.44868124e-01
7.33063996e-01 2.81399250e-01 4.99111742e-01 -6.99221373e-01
-7.91052699e-01 1.20079577e+00 -2.28095930e-02 3.33223939e-01
4.97643739e-01 5.75636402e-02 9.56565022e-01 -7.77123511e-01
1.00043547e+00 1.63955057e+00 9.84555364e-01 5.89670002e-01
-1.18293679e+00 -5.79073727e-01 1.88626796e-01 2.15078369e-01
-1.33669865e+00 -5.44750392e-01 3.79578859e-01 2.50852685e-02
1.63947749e+00 4.36125934e-01 5.65000355e-01 1.00892997e+00
5.09252608e-01 7.49746501e-01 9.23893571e-01 -6.23510122e-01
-3.16213459e-01 2.30747938e-01 1.80590495e-01 5.62943399e-01
9.95914638e-02 -1.32687479e-01 -4.98257756e-01 1.65011749e-01
1.11031711e-01 -3.95848393e-01 -7.30557814e-02 -2.31886432e-02
-9.66615319e-01 7.83816934e-01 2.55832881e-01 4.28647637e-01
-5.18177152e-02 -2.80154254e-02 9.99891937e-01 8.41845214e-01
8.57890248e-01 8.46970081e-01 -7.02521384e-01 -2.53327757e-01
-1.08661985e+00 5.45323014e-01 1.01223099e+00 1.31722987e+00
4.80386943e-01 4.75929886e-01 -9.47238386e-01 9.68693674e-01
-4.44949389e-01 4.58966553e-01 4.03149217e-01 -7.94030786e-01
7.89636314e-01 4.71020401e-01 -6.08191453e-02 -5.25054574e-01
-3.98105949e-01 -4.03268725e-01 -8.11149240e-01 -4.26328555e-02
5.59643172e-02 7.75604858e-04 -9.26025867e-01 1.72454989e+00
-1.60297930e-01 -4.39376771e-01 -1.42459407e-01 4.79547948e-01
1.06352711e+00 4.55270052e-01 3.48878324e-01 -1.51201427e-01
1.11152458e+00 -1.18947935e+00 -1.13451123e+00 7.52633512e-02
9.48715091e-01 -1.14246142e+00 1.56028914e+00 5.59368908e-01
-1.35826635e+00 -4.32084560e-01 -9.05914903e-01 -7.43589938e-01
-6.95534348e-01 1.48259684e-01 6.84190392e-01 6.20224476e-01
-1.11578095e+00 9.97964978e-01 -5.35712540e-01 -6.84358239e-01
2.16936097e-01 2.87249446e-01 -4.03402239e-01 -1.21094048e-01
-1.49227870e+00 1.73385203e+00 2.67241508e-01 -2.48129740e-01
-8.61572862e-01 -8.81440401e-01 -8.80266726e-01 1.63797885e-01
2.80693203e-01 -1.16369939e+00 1.56234992e+00 -4.67068732e-01
-1.78429639e+00 8.58907402e-01 -4.58291739e-01 -8.11357141e-01
4.94799942e-01 -6.23094738e-01 -3.66305560e-01 -4.24787551e-01
1.32627651e-01 4.79877204e-01 7.41575003e-01 -7.59067893e-01
-3.62631977e-01 1.31753668e-01 4.20770913e-01 3.76986653e-01
-2.28055716e-01 3.81317347e-01 -2.95368403e-01 -9.70042765e-01
-7.42753983e-01 -6.81870937e-01 -1.13468394e-01 -6.57614470e-01
-8.10996890e-01 -5.12692451e-01 6.02647841e-01 -7.40077972e-01
1.85924661e+00 -1.73005664e+00 5.63673437e-01 -2.94040859e-01
1.88114941e-02 1.06365132e+00 -2.62704283e-01 8.36254418e-01
-2.76920259e-01 6.36328638e-01 -2.07124040e-01 -1.08236146e+00
1.22810706e-01 2.83601999e-01 -5.25675893e-01 8.30411538e-03
2.23433431e-02 1.06208968e+00 -8.53155017e-01 -4.87777710e-01
6.25641644e-01 3.20178568e-01 -6.12586141e-01 7.73014873e-03
-1.92148983e-01 9.21686217e-02 -1.78714573e-01 4.88280624e-01
5.57584763e-01 6.73695445e-01 -6.58930093e-02 -3.00378174e-01
-3.03046823e-01 8.03054988e-01 -5.62073171e-01 1.83647144e+00
-8.30938101e-01 7.52798736e-01 -3.72557670e-01 -4.99269158e-01
5.75928152e-01 3.40221763e-01 -2.40064308e-01 -9.11651015e-01
1.97373375e-01 5.88933565e-02 -1.93264693e-01 -3.65459591e-01
1.00829601e+00 -2.85960227e-01 -3.92203838e-01 2.01500043e-01
3.46761286e-01 -8.56521487e-01 7.56357074e-01 5.47886312e-01
1.28197956e+00 1.00968905e-01 5.30981123e-01 -3.52112353e-01
5.60682714e-01 6.95051774e-02 5.19877493e-01 9.19778228e-01
-5.68453632e-02 6.87236786e-01 4.48217183e-01 -3.09696913e-01
-1.26561975e+00 -9.14634168e-01 1.86059822e-03 1.28699696e+00
-3.57536823e-01 -1.22128689e+00 -1.28627551e+00 -6.42663658e-01
-8.27523917e-02 1.53593361e+00 -5.90341508e-01 -3.18548888e-01
-8.94512951e-01 -5.37523508e-01 9.80325639e-01 1.03345245e-01
2.08369642e-01 -1.40068340e+00 -7.37871006e-02 1.99040368e-01
-3.90312076e-01 -1.02535367e+00 -6.95248485e-01 1.11546963e-01
-8.14400733e-01 -7.15690196e-01 -1.55615360e-01 -5.64847469e-01
1.10108778e-01 9.29853246e-02 1.68498814e+00 -3.45229842e-02
-1.68777272e-01 -2.46688649e-01 -4.31120157e-01 -7.58856535e-01
-5.22797406e-01 7.52192020e-01 2.78945286e-02 -8.70989382e-01
3.32836211e-01 -5.15298724e-01 6.65001720e-02 -4.95021008e-02
-9.18642521e-01 -1.00905620e-01 2.62726039e-01 8.69545341e-01
4.94143367e-01 -1.13956057e-01 3.24539661e-01 -1.54078472e+00
1.19493735e+00 8.89766812e-02 -2.12924585e-01 3.99123907e-01
-5.36427557e-01 2.20229909e-01 1.06157398e+00 -1.46284580e-01
-1.27055192e+00 -6.18438721e-01 -4.37738419e-01 -3.61893564e-01
1.42134443e-01 5.17489731e-01 -2.99244195e-01 3.96623686e-02
7.16825306e-01 -1.39188588e-01 -4.94923145e-01 -8.26508343e-01
7.53465593e-01 4.39713120e-01 4.10568297e-01 -4.80097950e-01
8.76353204e-01 1.55428782e-01 -2.31097460e-01 -7.52856016e-01
-1.46841872e+00 -1.21172063e-01 -7.05221176e-01 4.00700688e-01
4.45322841e-01 -8.55033815e-01 -2.91449904e-01 4.31453347e-01
-1.33146322e+00 -4.38387483e-01 -7.72386670e-01 9.32072848e-03
-4.07835305e-01 2.11965591e-01 -8.36307049e-01 -2.58037359e-01
-8.67689192e-01 -9.76312160e-01 1.02812827e+00 -6.63641468e-02
-6.72627687e-01 -1.13666558e+00 2.48733297e-01 1.99214742e-01
8.54645371e-01 1.07847124e-01 1.04167283e+00 -6.01255834e-01
-1.80412516e-01 -1.14839651e-01 -2.18636859e-02 5.72331667e-01
-7.72188976e-02 3.66015285e-01 -8.49344671e-01 -1.47329271e-01
-7.01802373e-02 -3.72404724e-01 1.04900575e+00 4.05925333e-01
1.12334788e+00 -2.46467844e-01 -1.54016197e-01 8.59292805e-01
1.03246415e+00 -3.58438492e-01 9.89187717e-01 5.47521293e-01
9.70085859e-01 1.89791426e-01 8.81137550e-01 1.90230295e-01
5.84414601e-01 8.82241547e-01 5.38252993e-03 -2.87044466e-01
-8.67478490e-01 -3.51053268e-01 4.16670382e-01 1.36423826e+00
3.41078192e-02 -5.75781405e-01 -4.38683242e-01 4.22353446e-01
-1.64139807e+00 -1.01254153e+00 -3.21723223e-01 1.95142078e+00
1.15772521e+00 4.64854866e-01 -2.49761149e-01 1.53675929e-01
4.77008611e-01 1.68697745e-01 -2.25192755e-01 -1.33592808e+00
-5.88822603e-01 7.63392329e-01 4.52955961e-01 8.58185291e-01
-1.09569716e+00 1.85952461e+00 7.47776031e+00 1.44000268e+00
-8.45427096e-01 3.19389045e-01 1.73115805e-01 -3.22571188e-01
-4.85279888e-01 6.12859912e-02 -9.13721204e-01 2.01664120e-02
9.13116157e-01 -5.20839453e-01 5.95274627e-01 5.47962606e-01
4.38116997e-01 5.07360473e-02 -1.17213070e+00 5.25017560e-01
2.21318498e-01 -1.02568316e+00 7.02774227e-01 -5.28568864e-01
7.96143591e-01 1.75435990e-01 -1.27313910e-02 1.00748825e+00
5.20895779e-01 -1.10294187e+00 9.14902985e-01 6.55144453e-01
1.01885056e+00 -9.94055569e-01 7.79128313e-01 2.87760377e-01
-9.92963433e-01 2.45505199e-01 -6.93320572e-01 -3.07266980e-01
4.78858173e-01 8.99089396e-01 -6.99186563e-01 7.85871029e-01
7.36005247e-01 9.00213242e-01 -7.25282371e-01 7.80565023e-01
-7.68699825e-01 7.30299115e-01 -1.47642106e-01 2.41811439e-01
7.32116997e-02 -1.00760289e-01 9.73750353e-01 1.75093913e+00
-5.94773032e-02 -5.56159094e-02 -2.42734402e-01 6.28373206e-01
-3.19680452e-01 1.90815955e-01 -6.08783245e-01 1.62481815e-01
6.17832482e-01 1.24996603e+00 -3.93174216e-03 -5.47150850e-01
-1.07225224e-01 9.88776445e-01 8.08423758e-01 2.18134031e-01
-8.50441098e-01 -7.93550313e-01 6.72809720e-01 -1.03318132e-01
3.94058406e-01 1.06178477e-01 -6.92510545e-01 -1.38948441e+00
-1.15877494e-01 -1.15253186e+00 2.15857521e-01 -8.12148631e-01
-1.21441460e+00 8.65067244e-01 1.18588157e-01 -8.90255928e-01
-1.35894656e-01 -1.86862499e-01 -7.67750025e-01 1.15221620e+00
-1.65488005e+00 -1.20519269e+00 2.45519932e-02 4.48144823e-01
7.08266139e-01 -1.44515246e-01 1.17186701e+00 3.94269526e-01
-8.21930349e-01 1.13330412e+00 2.10789889e-01 -2.91154087e-01
1.20137370e+00 -1.35079420e+00 1.03270698e+00 8.77861977e-01
-2.49620572e-01 8.51439536e-01 1.14232528e+00 -5.99451959e-01
-8.29821706e-01 -1.31255293e+00 1.64324343e+00 -9.17927802e-01
4.79318142e-01 -4.92192239e-01 -8.92424822e-01 9.07862723e-01
5.75233877e-01 -6.70083761e-01 3.30212414e-01 5.00422120e-01
-3.42515230e-01 -3.12167853e-01 -1.22082102e+00 7.67511070e-01
1.28925622e+00 -4.46274102e-01 -1.08944070e+00 3.09059501e-01
1.12169933e+00 -7.90010095e-01 -8.64296496e-01 3.03728104e-01
1.27793431e-01 -1.14128506e+00 8.78837943e-01 -6.43696845e-01
4.83667672e-01 -1.94109103e-03 1.87812984e-01 -2.01321864e+00
-4.69341338e-01 -8.02972019e-01 1.00009039e-01 1.53023601e+00
3.42875749e-01 -5.06122291e-01 2.46502638e-01 4.17003751e-01
-8.06197405e-01 -6.21146142e-01 -8.85723889e-01 -8.01774323e-01
5.01035094e-01 -2.99409002e-01 8.23240280e-01 8.20610166e-01
2.98475325e-02 7.38454580e-01 -3.44231963e-01 -6.06687546e-01
2.75016755e-01 -1.42440170e-01 9.03915524e-01 -9.50311482e-01
1.81767374e-01 -5.57311952e-01 5.87690771e-02 -6.15413070e-01
4.95328635e-01 -1.04384542e+00 -3.26548904e-01 -1.65392244e+00
2.07833618e-01 -3.26310337e-01 -9.36280563e-03 7.07783759e-01
-3.93420994e-01 5.38430475e-02 4.41515535e-01 1.00926273e-02
-7.27061331e-01 8.49744022e-01 1.43751180e+00 -1.90223649e-01
-8.10870752e-02 -1.71390101e-01 -9.34389472e-01 6.52689993e-01
8.08721125e-01 -7.35570967e-01 -1.89716518e-01 -9.89995360e-01
1.07547574e-01 -3.26112449e-01 -2.59789050e-01 -7.38862157e-01
8.85527954e-03 2.23373860e-01 -2.27185875e-01 -6.74198449e-01
3.61396283e-01 -1.66938573e-01 -5.86617365e-03 3.48104626e-01
-3.93028021e-01 2.26810947e-01 4.56337601e-01 -1.54305935e-01
-3.56955409e-01 -2.10769162e-01 7.41982102e-01 -1.51853383e-01
-2.93990612e-01 6.06035180e-02 -4.93071020e-01 5.36875069e-01
5.64118266e-01 9.11737531e-02 -5.20794451e-01 -3.53092849e-01
-5.25245547e-01 2.15819776e-01 4.74169612e-01 4.30812508e-01
3.37548912e-01 -1.06312251e+00 -9.76055324e-01 -1.80158228e-01
6.91434518e-02 -1.34104624e-01 -4.16710898e-02 5.45558691e-01
-5.33291698e-01 8.63777876e-01 -6.65407330e-02 -9.07118618e-03
-1.23885381e+00 7.02286780e-01 4.52247262e-01 -9.16085541e-01
-4.93480265e-01 8.86393309e-01 1.51680345e-02 -9.81533527e-01
1.28200233e-01 -5.65096736e-01 -2.70393819e-01 -3.08508296e-02
3.49788994e-01 6.61581397e-01 6.71346366e-01 -4.80381817e-01
-2.42460489e-01 4.06286240e-01 -4.77170378e-01 1.75498158e-01
1.24248731e+00 -1.99796885e-01 -3.94003242e-01 3.50489944e-01
9.70209360e-01 1.23972692e-01 -4.21650410e-01 -3.81074250e-01
-2.29472741e-01 -2.04602808e-01 2.60587670e-02 -1.07400703e+00
-8.50813270e-01 9.10973907e-01 -1.54348657e-01 5.19615971e-02
1.14056015e+00 -4.62984055e-01 9.67392445e-01 6.17702365e-01
3.99027348e-01 -1.13765621e+00 -3.60494196e-01 1.36098969e+00
9.99801934e-01 -8.69789004e-01 2.08704010e-01 -6.68083251e-01
-9.26067710e-01 8.02948713e-01 5.01790643e-01 -2.88282216e-01
3.59614521e-01 1.94218859e-01 1.19360968e-01 -7.65600353e-02
-1.04234171e+00 -1.74372181e-01 1.20745428e-01 5.94667077e-01
7.16053903e-01 2.87628531e-01 -6.42278731e-01 4.91680503e-01
-9.31451201e-01 4.64981571e-02 6.31126761e-01 6.23618364e-01
-2.54440904e-01 -1.57391334e+00 -1.13905430e-01 5.72562158e-01
-6.23655438e-01 -7.64950335e-01 -6.28689945e-01 1.02063012e+00
5.62621579e-02 9.69132066e-01 -4.51789796e-01 -2.87664175e-01
1.15336359e+00 -1.22429065e-01 6.42194927e-01 -1.00478959e+00
-1.46278358e+00 -2.65662611e-01 6.16869092e-01 -6.63580060e-01
4.56814654e-02 -6.43930912e-01 -8.59508097e-01 -9.97604489e-01
-2.95678705e-01 2.39201605e-01 3.69717151e-01 1.03820920e+00
2.76382536e-01 1.01587129e+00 2.92500615e-01 -9.90327716e-01
-6.09370351e-01 -1.29732585e+00 -4.50145841e-01 6.34822130e-01
2.13199817e-02 -4.92456257e-01 -2.73740590e-01 -1.78013220e-01] | [11.057673454284668, 10.353551864624023] |
5d82ca71-86bd-402f-af79-0209ea15d592 | scaling-up-visual-and-vision-language | 2102.05918 | null | https://arxiv.org/abs/2102.05918v2 | https://arxiv.org/pdf/2102.05918v2.pdf | Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision | Pre-trained representations are becoming crucial for many NLP and perception tasks. While representation learning in NLP has transitioned to training on raw text without human annotations, visual and vision-language representations still rely heavily on curated training datasets that are expensive or require expert knowledge. For vision applications, representations are mostly learned using datasets with explicit class labels such as ImageNet or OpenImages. For vision-language, popular datasets like Conceptual Captions, MSCOCO, or CLIP all involve a non-trivial data collection (and cleaning) process. This costly curation process limits the size of datasets and hence hinders the scaling of trained models. In this paper, we leverage a noisy dataset of over one billion image alt-text pairs, obtained without expensive filtering or post-processing steps in the Conceptual Captions dataset. A simple dual-encoder architecture learns to align visual and language representations of the image and text pairs using a contrastive loss. We show that the scale of our corpus can make up for its noise and leads to state-of-the-art representations even with such a simple learning scheme. Our visual representation achieves strong performance when transferred to classification tasks such as ImageNet and VTAB. The aligned visual and language representations enables zero-shot image classification and also set new state-of-the-art results on Flickr30K and MSCOCO image-text retrieval benchmarks, even when compared with more sophisticated cross-attention models. The representations also enable cross-modality search with complex text and text + image queries. | ['Tom Duerig', 'Zhen Li', 'YunHsuan Sung', 'Quoc V. Le', 'Hieu Pham', 'Zarana Parekh', 'Yi-Ting Chen', 'Ye Xia', 'Yinfei Yang', 'Chao Jia'] | 2021-02-11 | null | null | null | null | ['zero-shot-transfer-image-classification', 'zero-shot-cross-modal-retrieval'] | ['computer-vision', 'miscellaneous'] | [ 5.27208924e-01 -4.55171913e-02 -3.43103617e-01 -2.94243008e-01
-1.15843904e+00 -7.14820206e-01 1.01298141e+00 1.89600959e-01
-6.81941628e-01 4.95444745e-01 1.99105188e-01 -2.97574878e-01
1.77546784e-01 -5.34213960e-01 -9.69450116e-01 -5.12933671e-01
5.04640222e-01 5.65639079e-01 -1.03984125e-01 -2.01782718e-01
-1.22674718e-01 1.24361113e-01 -2.00030780e+00 5.13623536e-01
3.55113178e-01 1.30946207e+00 3.98296297e-01 7.81700730e-01
-4.39183086e-01 8.02014887e-01 -3.22131664e-01 -4.43459988e-01
4.07315999e-01 -1.52427718e-01 -8.03577542e-01 2.19210703e-02
1.16177094e+00 -2.59780228e-01 -5.65083563e-01 9.37664092e-01
5.78862667e-01 1.42861962e-01 7.26526737e-01 -1.30066395e+00
-1.42982459e+00 2.59741366e-01 -6.14353418e-01 -3.34857255e-02
1.50738388e-01 3.79610419e-01 1.27851534e+00 -9.87501264e-01
9.32099998e-01 1.17757213e+00 4.83746231e-01 6.31454647e-01
-1.50471437e+00 -5.10528386e-01 1.22159883e-01 2.76035309e-01
-1.33701348e+00 -5.82407832e-01 3.46880525e-01 -6.04782641e-01
1.13612103e+00 1.02504887e-01 4.37163174e-01 1.60380769e+00
-3.46725762e-01 8.51410627e-01 8.97113383e-01 -4.78770018e-01
-6.48713857e-02 2.41473824e-01 1.76857784e-01 6.63678944e-01
2.67654866e-01 -7.72416294e-02 -3.64976078e-01 1.58210203e-01
6.18724465e-01 3.12293023e-01 -4.13167626e-01 -6.45373762e-01
-1.25442839e+00 9.28810239e-01 7.33953297e-01 2.79125035e-01
-1.81487843e-01 6.28427804e-01 6.78849280e-01 3.53810936e-01
2.98097134e-01 4.65414554e-01 -4.26771939e-01 -4.38383501e-03
-1.06353259e+00 -7.44420961e-02 5.08611679e-01 1.13248873e+00
9.18554246e-01 -9.18316767e-02 -2.51522273e-01 9.59419131e-01
-5.11698611e-02 9.43922758e-01 5.13443828e-01 -7.43464410e-01
5.06739914e-01 4.95884985e-01 -1.42515883e-01 -8.55488777e-01
-1.37843162e-01 -2.23892778e-01 -8.77339125e-01 1.49540842e-01
3.80028874e-01 3.32932919e-01 -1.64789271e+00 1.58511674e+00
-1.11422189e-01 -5.47764823e-02 2.10741013e-01 9.89466190e-01
1.11083293e+00 6.92306459e-01 1.92074299e-01 1.10790573e-01
1.52809060e+00 -1.11276150e+00 -4.95939612e-01 -5.65526247e-01
5.06725013e-01 -7.15816617e-01 1.50453091e+00 2.44195595e-01
-8.04463685e-01 -4.82930720e-01 -8.86560619e-01 -8.56622875e-01
-8.04943204e-01 7.68732578e-02 5.90295196e-01 1.59500122e-01
-1.01871216e+00 2.23227903e-01 -4.98213142e-01 -6.91992640e-01
7.46706009e-01 2.11790297e-02 -6.70638502e-01 -5.47621191e-01
-8.49851906e-01 9.46394265e-01 2.34706014e-01 -1.88680530e-01
-1.17515802e+00 -8.73642385e-01 -1.13627827e+00 4.00453582e-02
5.13655841e-01 -7.85220563e-01 1.09608912e+00 -1.27180409e+00
-9.51409101e-01 1.26589870e+00 -1.09413322e-02 -7.14660048e-01
4.55246717e-01 -4.54343826e-01 -1.36778414e-01 4.20416892e-01
2.66325325e-01 1.12434423e+00 1.24524879e+00 -1.32365918e+00
-2.57261008e-01 -2.03468651e-01 1.93309948e-01 4.32857648e-02
-2.62555301e-01 -4.04278710e-02 -9.10260081e-01 -5.09727597e-01
-2.39501849e-01 -8.55970025e-01 -8.67313445e-02 4.10828799e-01
-2.67288923e-01 -1.99889671e-02 8.88138592e-01 -6.79881573e-01
7.04136908e-01 -2.16582751e+00 2.23857433e-01 -1.54415920e-01
1.08700171e-01 3.11493218e-01 -6.99611545e-01 4.37817365e-01
-1.07251637e-01 1.20080970e-01 -1.98829532e-01 -5.47574520e-01
8.93978849e-02 3.86066377e-01 -6.85148776e-01 4.34566855e-01
3.19369435e-01 1.31753147e+00 -8.77022803e-01 -5.73895216e-01
5.73392093e-01 7.92980433e-01 -4.94528085e-01 1.15343191e-01
-5.55765748e-01 -2.94447728e-02 -1.22508332e-01 8.25352132e-01
2.92863041e-01 -7.77368009e-01 -1.10166401e-01 -4.95687127e-01
2.40465343e-01 7.51260482e-03 -6.44758224e-01 2.24770093e+00
-7.72129178e-01 9.93513405e-01 2.46937182e-02 -1.18698621e+00
5.63089848e-01 1.01697445e-01 2.96834439e-01 -1.06015289e+00
2.21841037e-01 -3.85622401e-03 -5.30487955e-01 -5.53535819e-01
7.11738348e-01 1.16704375e-01 -2.14216158e-01 3.14584434e-01
5.17416418e-01 -2.36084104e-01 2.36370832e-01 4.52787966e-01
9.29614067e-01 1.95782572e-01 1.70300931e-01 1.94605321e-01
1.64260983e-01 2.29070112e-01 4.67839837e-02 9.21716869e-01
1.26554500e-02 1.04967034e+00 3.20212781e-01 -4.76447165e-01
-1.30753386e+00 -9.23534274e-01 -2.07139298e-01 1.31620526e+00
5.01739122e-02 -5.28699279e-01 -3.97854149e-01 -7.38399386e-01
1.91855371e-01 6.12840414e-01 -9.20221508e-01 -5.12008332e-02
-1.40357792e-01 -3.67127120e-01 6.09521806e-01 4.94056195e-01
2.46158600e-01 -1.02698994e+00 -5.92625797e-01 -1.47136390e-01
-2.02478677e-01 -1.47977829e+00 -3.11423421e-01 2.78870523e-01
-3.35408688e-01 -1.21219957e+00 -8.67046773e-01 -7.19220757e-01
6.74697459e-01 5.25334120e-01 1.29615998e+00 6.46645501e-02
-8.42649460e-01 8.92383218e-01 -4.74594921e-01 -3.97842050e-01
-1.28263891e-01 -6.73911124e-02 -2.54115403e-01 -6.73236325e-03
3.52842003e-01 -3.76175284e-01 -7.38842845e-01 -9.32565033e-02
-1.07101083e+00 1.47104755e-01 8.08410406e-01 1.14308500e+00
8.17462265e-01 -7.32294440e-01 1.62647069e-01 -6.04810715e-01
2.26465344e-01 -4.05000508e-01 -6.07518613e-01 4.62820381e-01
-4.93771940e-01 1.68766007e-01 4.39245164e-01 -6.55627787e-01
-5.98878860e-01 1.76514730e-01 3.27039838e-01 -1.14561963e+00
-1.09691262e-01 3.34506631e-01 1.66889668e-01 1.18424892e-02
7.99228907e-01 3.00059170e-01 4.39363904e-02 -4.71416235e-01
1.00534046e+00 6.53321385e-01 8.26566100e-01 -2.90316403e-01
8.29582632e-01 6.37447357e-01 -1.50393590e-01 -1.10027111e+00
-1.17045426e+00 -5.69497049e-01 -4.81650352e-01 1.56014726e-01
1.20147216e+00 -1.33157182e+00 -3.05166423e-01 -3.50940116e-02
-1.11780107e+00 -2.31948972e-01 -5.78692973e-01 2.05204844e-01
-5.86503983e-01 3.19900602e-01 -3.15612614e-01 -5.89563370e-01
-5.68582416e-01 -1.05196404e+00 1.53205848e+00 7.33436085e-03
9.04098898e-02 -6.64049327e-01 -1.51429117e-01 6.30008399e-01
3.71651083e-01 1.90263450e-01 7.74679840e-01 -5.07339358e-01
-8.26038778e-01 -1.89263314e-01 -8.58063757e-01 4.76199210e-01
-2.74974018e-01 -2.15257317e-01 -1.31972122e+00 -3.83569598e-01
-6.67312622e-01 -1.08730209e+00 1.38277876e+00 1.37498155e-01
1.37137449e+00 -1.47287235e-01 -2.86004156e-01 8.92022491e-01
1.66250396e+00 -3.44032645e-01 6.50312722e-01 2.68306732e-01
1.05048501e+00 4.84030575e-01 4.00092304e-01 6.03356920e-02
3.12547058e-01 6.07232332e-01 6.46176100e-01 -2.62973398e-01
-4.09804463e-01 -4.41908300e-01 1.50684312e-01 4.90454882e-01
1.11668229e-01 -3.26302916e-01 -1.05720818e+00 9.00248945e-01
-1.85682940e+00 -9.45117891e-01 2.22493306e-01 2.01670098e+00
7.67177522e-01 -2.94535339e-01 -4.12994355e-01 -2.25101843e-01
3.91989022e-01 3.99185240e-01 -6.08306170e-01 -1.48146227e-01
-4.17353392e-01 3.70832652e-01 7.81350970e-01 2.31929183e-01
-1.19449544e+00 1.12404430e+00 5.85426950e+00 7.74474382e-01
-1.28333998e+00 2.21447185e-01 4.80405331e-01 -4.52563792e-01
-2.81316996e-01 -1.11551337e-01 -6.07978702e-01 1.48572251e-01
8.69966686e-01 9.91780683e-02 5.72918296e-01 9.47974563e-01
-2.98592538e-01 1.52799219e-03 -1.34708667e+00 1.74875987e+00
4.67154950e-01 -1.61884093e+00 4.41036314e-01 1.37132511e-01
7.00203717e-01 6.81984961e-01 1.67134970e-01 5.48230469e-01
3.57574254e-01 -1.44082367e+00 6.33749306e-01 5.49454153e-01
1.36817205e+00 -2.46887654e-01 4.78349268e-01 -8.67830589e-03
-9.69804823e-01 -1.67653680e-01 -6.00150943e-01 2.97779918e-01
1.67415310e-02 3.32448751e-01 -6.82040155e-01 4.37925845e-01
8.79750848e-01 9.32545066e-01 -8.91139448e-01 6.55269384e-01
-2.11644843e-02 3.02520931e-01 -2.66881734e-01 9.00256932e-02
4.02874529e-01 1.98195614e-02 2.74942756e-01 1.19489849e+00
1.67303190e-01 -1.88602567e-01 1.92434758e-01 8.51012170e-01
-4.65712100e-01 1.06301472e-01 -9.66193259e-01 -5.74690819e-01
2.36912787e-01 1.26394010e+00 -3.68669093e-01 -5.18329620e-01
-6.50740862e-01 1.12435555e+00 5.58372676e-01 6.84976280e-01
-7.26946175e-01 -2.46136412e-01 8.54312479e-01 1.71664115e-02
6.33473277e-01 -1.24706246e-01 -8.80207196e-02 -1.41836071e+00
-2.44036876e-02 -8.57259572e-01 3.13686520e-01 -1.10646629e+00
-1.41572201e+00 5.30819058e-01 -1.19297646e-01 -1.04736412e+00
-2.49113902e-01 -7.74609804e-01 -1.28144458e-01 5.74196458e-01
-1.76153982e+00 -1.61852264e+00 -6.52058661e-01 8.00468326e-01
6.58787489e-01 -1.58528462e-01 8.88112068e-01 3.87200475e-01
-3.00055057e-01 5.72542787e-01 1.74845561e-01 3.73943090e-01
8.44466388e-01 -1.04006350e+00 2.43785143e-01 6.09489620e-01
6.50970995e-01 4.08289075e-01 5.05694211e-01 -2.05370992e-01
-1.73278654e+00 -1.17835796e+00 5.64295888e-01 -6.90276206e-01
8.37425947e-01 -6.53120935e-01 -8.99104953e-01 7.97590792e-01
4.21938717e-01 6.33178234e-01 4.29294199e-01 6.10521846e-02
-9.55803096e-01 -9.74519625e-02 -6.46187305e-01 4.73228067e-01
1.16073728e+00 -1.01453769e+00 -6.88585579e-01 5.22517979e-01
1.03529608e+00 -1.97965711e-01 -5.25548041e-01 2.68429697e-01
6.29470170e-01 -4.88663614e-01 1.28949082e+00 -9.06262755e-01
6.24514282e-01 -2.25929886e-01 -4.96780694e-01 -9.99722958e-01
-3.32536213e-02 -3.10877949e-01 6.14744462e-02 1.14415205e+00
4.25713837e-01 -2.12768704e-01 5.20660877e-01 4.27913606e-01
8.43520612e-02 -4.50170040e-01 -8.00292671e-01 -6.33179486e-01
-9.39806104e-02 -6.54956400e-01 1.61436111e-01 1.07044053e+00
-5.42818666e-01 6.32565856e-01 -4.81879383e-01 -9.52653289e-02
6.95141912e-01 2.25247130e-01 9.98517096e-01 -1.15041137e+00
-2.05897853e-01 -2.73214698e-01 -5.83403766e-01 -8.69884551e-01
3.13624144e-01 -1.18308520e+00 -3.72980610e-02 -1.76742458e+00
2.33788475e-01 -2.89413512e-01 -3.44628632e-01 7.41943479e-01
1.24280110e-01 6.12180948e-01 4.96073842e-01 1.47820160e-01
-9.71818626e-01 6.31256878e-01 9.71405625e-01 -6.26749992e-01
5.45256771e-02 -8.94699514e-01 -7.12267280e-01 5.38836122e-01
4.69057500e-01 -2.90440351e-01 -4.86579508e-01 -8.35698187e-01
2.83878535e-01 -3.10226291e-01 9.12221313e-01 -7.91363955e-01
1.67963400e-01 -9.93989781e-03 5.40882230e-01 -5.51316500e-01
7.13624001e-01 -9.39942122e-01 -6.66893348e-02 1.47719279e-01
-4.50903118e-01 -1.57087654e-01 2.12078050e-01 9.01344419e-01
-3.49458903e-01 5.39303571e-02 6.63373649e-01 -2.17428759e-01
-9.31645751e-01 4.20020670e-01 1.88952208e-01 3.78976613e-01
9.45308924e-01 -9.64904502e-02 -7.28880286e-01 -4.52561110e-01
-6.25229836e-01 3.76696438e-01 6.12679183e-01 8.08980942e-01
6.96946263e-01 -1.16569185e+00 -5.52118063e-01 1.57173976e-01
7.31078982e-01 1.07911296e-01 2.25732982e-01 7.18308985e-01
-4.41615909e-01 5.05029559e-01 -2.01711610e-01 -8.71420145e-01
-1.30900645e+00 9.02284086e-01 1.53172851e-01 -1.14600517e-01
-8.19425642e-01 7.66890764e-01 4.91885245e-01 -2.05379069e-01
4.56180960e-01 -4.42006558e-01 3.14185768e-02 3.97717059e-01
6.77201509e-01 -1.31050602e-01 -9.29329693e-02 -7.87855029e-01
-2.55136579e-01 7.37600029e-01 -1.75953150e-01 1.04575772e-02
1.38514924e+00 -1.44786268e-01 8.34768191e-02 3.91321480e-01
1.54357421e+00 -5.01706719e-01 -1.13959789e+00 -4.45367008e-01
-2.11090937e-01 -4.78485197e-01 2.63763964e-01 -7.37300456e-01
-9.96082246e-01 1.28001225e+00 6.90357268e-01 -1.27939001e-01
9.64142323e-01 3.21969181e-01 7.34304845e-01 7.99777687e-01
2.20475093e-01 -1.08820117e+00 3.02092224e-01 5.25772512e-01
1.13589740e+00 -1.77425039e+00 -4.16583642e-02 -4.47153784e-02
-7.44787216e-01 8.44261050e-01 4.46389645e-01 -9.07234848e-02
3.61205101e-01 -1.45586744e-01 1.23864248e-01 -2.38411725e-01
-9.06594634e-01 -7.11130917e-01 5.19470334e-01 7.69629121e-01
1.40510932e-01 -1.24130100e-01 2.33132973e-01 2.97556430e-01
-2.64626816e-02 -8.40122476e-02 3.12120855e-01 7.74627268e-01
-3.38703364e-01 -7.74409831e-01 -1.78365618e-01 4.96426165e-01
-3.77148718e-01 -4.92016375e-01 -3.94017667e-01 9.19805944e-01
-2.71525327e-02 6.27076387e-01 2.94195354e-01 -7.53889307e-02
2.06011310e-01 1.85758665e-01 4.87618297e-01 -6.01921797e-01
-3.34640950e-01 -1.58588514e-01 1.25637904e-01 -8.83755744e-01
-4.56867129e-01 -2.94932276e-01 -9.82805431e-01 4.71411757e-02
-1.44780830e-01 -2.33504310e-01 8.97747278e-01 7.06902921e-01
5.30647159e-01 4.30156559e-01 -4.63810302e-02 -1.03103316e+00
-4.18705821e-01 -8.40111494e-01 -2.31282488e-01 8.44158590e-01
6.64952695e-01 -6.94834888e-01 -2.93102264e-01 3.26469183e-01] | [10.554529190063477, 1.6556657552719116] |
76730c98-1ea1-4847-819a-2eca5557ab2b | jointist-simultaneous-improvement-of-multi | 2302.00286 | null | https://arxiv.org/abs/2302.00286v2 | https://arxiv.org/pdf/2302.00286v2.pdf | Jointist: Simultaneous Improvement of Multi-instrument Transcription and Music Source Separation via Joint Training | In this paper, we introduce Jointist, an instrument-aware multi-instrument framework that is capable of transcribing, recognizing, and separating multiple musical instruments from an audio clip. Jointist consists of an instrument recognition module that conditions the other two modules: a transcription module that outputs instrument-specific piano rolls, and a source separation module that utilizes instrument information and transcription results. The joint training of the transcription and source separation modules serves to improve the performance of both tasks. The instrument module is optional and can be directly controlled by human users. This makes Jointist a flexible user-controllable framework. Our challenging problem formulation makes the model highly useful in the real world given that modern popular music typically consists of multiple instruments. Its novelty, however, necessitates a new perspective on how to evaluate such a model. In our experiments, we assess the proposed model from various aspects, providing a new evaluation perspective for multi-instrument transcription. Our subjective listening study shows that Jointist achieves state-of-the-art performance on popular music, outperforming existing multi-instrument transcription models such as MT3. We conducted experiments on several downstream tasks and found that the proposed method improved transcription by more than 1 percentage points (ppt.), source separation by 5 SDR, downbeat detection by 1.8 ppt., chord recognition by 1.4 ppt., and key estimation by 1.4 ppt., when utilizing transcription results obtained from Jointist. Demo available at \url{https://jointist.github.io/Demo}. | ['Dorien Herremans', 'Yun-Ning Hung', 'Ju-Chiang Wang', 'Minz Won', 'Bochen Li', 'Qiuqiang Kong', 'Keunwoo Choi', 'Kin Wai Cheuk'] | 2023-02-01 | null | null | null | null | ['chord-recognition', 'instrument-recognition', 'music-source-separation'] | ['audio', 'audio', 'music'] | [ 1.42263412e-01 -3.82722408e-01 -6.56878874e-02 2.63243765e-01
-1.47195327e+00 -1.08900654e+00 2.25117475e-01 -5.47677279e-02
-1.53145120e-01 3.91904384e-01 1.61136791e-01 4.24866118e-02
-3.03688228e-01 -2.33814478e-01 -5.06374717e-01 -6.81499839e-01
1.35960905e-02 2.60995835e-01 2.71531083e-02 -2.92136222e-01
3.25854093e-01 3.37444335e-01 -1.53054690e+00 3.41721684e-01
6.13921940e-01 1.15149581e+00 2.84963876e-01 1.13287926e+00
2.47749761e-01 4.36026096e-01 -8.98807168e-01 -1.38893843e-01
3.15618545e-01 -5.90428889e-01 -5.24190247e-01 -2.59285361e-01
1.48102626e-01 -1.62881330e-01 2.73499787e-01 7.32046187e-01
1.00730860e+00 8.78193006e-02 4.99402881e-01 -9.54565108e-01
-7.10743889e-02 1.02061415e+00 -5.17319322e-01 1.53956965e-01
6.22871101e-01 8.59455168e-02 1.38634157e+00 -7.78040946e-01
1.88567251e-01 7.87512720e-01 7.44478703e-01 1.56480838e-02
-1.08574283e+00 -9.11212385e-01 -3.93461227e-01 -3.44263501e-02
-1.60620725e+00 -6.76744521e-01 8.98015440e-01 -2.94129044e-01
5.65427184e-01 7.19590068e-01 6.07021272e-01 9.42600071e-01
-7.24650174e-02 7.75954247e-01 9.10798669e-01 -4.99708176e-01
3.89521755e-02 -1.79845706e-01 -9.50637013e-02 1.22654587e-01
-2.00004935e-01 -1.32984549e-01 -9.05991375e-01 -3.01755399e-01
9.10195887e-01 -4.12950397e-01 -5.45672655e-01 2.37725690e-01
-1.27379656e+00 4.15753812e-01 -1.30453229e-01 4.45570558e-01
-3.95366788e-01 1.95749238e-01 6.29898190e-01 4.64297891e-01
1.65740535e-01 6.37059331e-01 -4.05924231e-01 -6.18295312e-01
-1.39007366e+00 3.54985565e-01 7.71760404e-01 7.04862416e-01
2.76656806e-01 2.73368299e-01 -2.47081876e-01 1.09626973e+00
-1.56329144e-02 6.64227068e-01 7.27687597e-01 -1.02472675e+00
4.42024767e-01 2.15776116e-01 1.36556104e-01 -8.73923421e-01
-3.86355430e-01 -8.76682222e-01 -7.89466500e-01 -3.88491787e-02
4.61112916e-01 -2.50245690e-01 -3.55002671e-01 1.66302562e+00
3.25429207e-03 2.28762984e-01 -1.26001000e-01 1.03015506e+00
6.30015612e-01 6.22570097e-01 -4.98205423e-01 -5.62770426e-01
1.54674673e+00 -8.70894611e-01 -7.90197432e-01 1.10939696e-01
2.63621688e-01 -1.44072509e+00 1.28490496e+00 1.01977348e+00
-1.38337898e+00 -9.39818084e-01 -1.08180857e+00 1.54544845e-01
5.06901205e-01 6.68267131e-01 3.34696025e-01 5.95352590e-01
-6.76944017e-01 6.67268813e-01 -6.61568999e-01 -7.84374177e-02
-2.25408733e-01 3.14456701e-01 -4.83427346e-02 6.34858310e-01
-1.01698136e+00 2.23798886e-01 3.21597844e-01 -5.72434179e-02
-7.65342951e-01 -4.97789532e-01 -4.27455395e-01 1.77053615e-01
4.68252122e-01 -5.33703625e-01 1.60565400e+00 -9.66252744e-01
-2.00375414e+00 5.66810489e-01 -1.99654371e-01 -3.10442030e-01
5.23221552e-01 -6.74847782e-01 -6.37992024e-01 3.06609809e-01
-2.51289420e-02 -5.17884158e-02 1.10397363e+00 -9.89181221e-01
-6.61730886e-01 -1.23172365e-01 -1.48012072e-01 3.15401614e-01
-2.14604557e-01 1.76410362e-01 -8.01736295e-01 -1.10910761e+00
1.76203519e-01 -1.19384325e+00 2.42864475e-01 -5.74384451e-01
-6.92797899e-01 1.87305138e-01 3.82938206e-01 -7.89989769e-01
1.61786461e+00 -2.44878197e+00 3.27265084e-01 2.97686219e-01
-2.26489425e-01 1.16356112e-01 -7.32173473e-02 6.02366984e-01
-6.61121458e-02 -3.93609218e-02 -3.35562788e-02 -5.29938936e-01
-2.19720434e-02 -2.29399189e-01 -5.67904115e-01 3.43321711e-01
-3.24224263e-01 4.78934765e-01 -4.01113182e-01 -3.39198768e-01
-1.12744318e-02 3.06602299e-01 -5.46377420e-01 2.98882544e-01
1.46093294e-01 6.80814743e-01 -1.12895198e-01 8.53448331e-01
3.78989697e-01 1.07032403e-01 2.69232541e-01 -2.51368880e-01
-3.73425364e-01 3.65808368e-01 -1.87919068e+00 2.05162239e+00
-4.74605203e-01 4.12453949e-01 2.74095416e-01 -5.37050188e-01
1.13674808e+00 8.23951006e-01 3.94284546e-01 -2.36956850e-01
1.68836042e-01 5.30635357e-01 2.49846637e-01 -4.56646323e-01
6.88282728e-01 -2.04025954e-03 -3.42721313e-01 5.33157945e-01
1.50941521e-01 -3.83437157e-01 3.38772804e-01 -3.82265747e-02
8.73030543e-01 1.53397340e-02 4.82883960e-01 -7.45064095e-02
4.66840774e-01 -4.43322033e-01 6.25619113e-01 6.68047369e-01
-1.30177774e-02 9.27006662e-01 3.69087815e-01 1.43616423e-01
-6.37629211e-01 -1.06878090e+00 2.45216921e-01 1.36010587e+00
-1.15288630e-01 -8.81716669e-01 -5.87166905e-01 4.97365519e-02
-2.08864316e-01 4.03264254e-01 -1.89915210e-01 1.02759786e-01
-6.53091252e-01 -3.78255308e-01 1.09972882e+00 4.60307568e-01
2.65123248e-01 -9.51685131e-01 -5.29463887e-01 2.14783579e-01
-7.94198275e-01 -8.07304740e-01 -7.78685391e-01 3.62100989e-01
-8.14664185e-01 -8.21308851e-01 -7.15825677e-01 -4.87441927e-01
-2.71660775e-01 1.63826406e-01 8.39977324e-01 -2.16309056e-01
1.89923141e-02 1.55586824e-01 -5.11712790e-01 -5.15317440e-01
-2.42520511e-01 3.70726466e-01 3.87063742e-01 1.56354740e-01
-2.54970163e-01 -1.17320585e+00 -5.93407691e-01 4.75352168e-01
-7.58346200e-01 -1.44760594e-01 6.50629401e-01 5.01022637e-01
5.98855197e-01 -5.89614846e-02 7.66617179e-01 -3.62703502e-01
8.49428296e-01 -2.01908514e-01 -3.87206793e-01 -1.17833562e-01
-2.56407917e-01 -2.55068213e-01 8.66580606e-01 -6.58726215e-01
-7.33945847e-01 8.48349705e-02 -3.20152849e-01 -3.92233491e-01
3.99190784e-02 4.58521575e-01 -2.52088398e-01 3.42692256e-01
6.17810369e-01 2.70267129e-01 -2.48722345e-01 -8.52830112e-01
2.90800571e-01 9.90985990e-01 9.65110540e-01 -6.52039170e-01
7.82328010e-01 2.16737330e-01 -2.95956075e-01 -8.67972136e-01
-5.98607421e-01 -7.48462856e-01 -5.39856613e-01 -2.09458604e-01
4.42536563e-01 -1.04840362e+00 -9.56471503e-01 4.22022104e-01
-9.77423131e-01 -1.12778656e-01 -2.08533376e-01 7.71652818e-01
-8.07130158e-01 2.21884429e-01 -7.18901575e-01 -8.93919826e-01
-7.81380415e-01 -1.05069864e+00 1.05904639e+00 8.37398991e-02
-8.78032267e-01 -4.63407159e-01 2.37985834e-01 6.13927782e-01
1.42254815e-01 5.87953292e-02 4.57231343e-01 -5.71623385e-01
-2.22218141e-01 -1.23956643e-01 4.08616215e-01 4.58350509e-01
2.17632487e-01 1.62429780e-01 -1.36478519e+00 -3.33068669e-01
1.75738893e-02 -3.29226673e-01 6.51649237e-01 2.91139096e-01
9.52000260e-01 -1.24610998e-01 9.90955010e-02 6.56204820e-01
1.15179956e+00 1.57121405e-01 5.32642186e-01 1.86662495e-01
3.99149984e-01 1.08256169e-01 6.81783736e-01 8.55936468e-01
-8.81270915e-02 1.02394271e+00 2.22178340e-01 1.12248406e-01
-1.96478531e-01 -2.11545527e-01 5.43764412e-01 1.31774426e+00
-5.55689514e-01 -2.43424997e-01 -6.48811638e-01 5.39760530e-01
-1.70493615e+00 -1.03516197e+00 -2.50394940e-01 2.31949568e+00
1.13118422e+00 1.64930031e-01 5.20035267e-01 8.32090557e-01
5.60651183e-01 8.15753490e-02 -3.59670520e-01 -5.35121620e-01
-4.58067395e-02 5.91359913e-01 1.54715106e-01 2.90329278e-01
-9.68359590e-01 6.04984522e-01 5.67066813e+00 1.05076480e+00
-1.40594542e+00 4.70842682e-02 -4.33113649e-02 -5.10515988e-01
2.89918613e-02 -1.45335004e-01 -5.37804246e-01 3.91451031e-01
8.78371537e-01 -1.91952318e-01 4.68619347e-01 6.41909063e-01
5.37488997e-01 -1.01347929e-02 -1.01056778e+00 1.30862308e+00
-4.88730259e-02 -9.68088388e-01 -9.26369429e-02 -9.29958895e-02
3.84984672e-01 -2.30226770e-01 9.35513228e-02 2.55549163e-01
-3.41337293e-01 -8.07347119e-01 1.21469367e+00 4.78608340e-01
8.91087770e-01 -9.28370357e-01 4.85514045e-01 4.87339854e-01
-1.52036846e+00 -1.76211953e-01 1.96487308e-01 -2.78789073e-01
2.99153060e-01 6.67775273e-01 -7.72944152e-01 9.66917038e-01
6.10667288e-01 3.47233742e-01 -2.58881599e-01 1.11044097e+00
-4.19721693e-01 1.18784702e+00 -4.50812489e-01 5.26105702e-01
-2.55863786e-01 -1.64903933e-04 8.29859138e-01 1.35327458e+00
6.49351120e-01 -1.93486940e-02 2.52874881e-01 4.75048989e-01
-1.41960932e-02 5.56339443e-01 -6.67594299e-02 1.22747578e-01
6.29204094e-01 1.42174399e+00 -6.14660203e-01 -1.34253010e-01
7.30202645e-02 1.13120532e+00 -8.53721574e-02 1.81525424e-01
-1.02077711e+00 -6.99858427e-01 5.63873351e-01 -1.14494346e-01
3.55240911e-01 -2.81804413e-01 -3.62831801e-01 -1.01458395e+00
1.21820189e-01 -1.30152297e+00 3.01672280e-01 -6.69790566e-01
-6.96635902e-01 5.87643802e-01 -3.03033590e-01 -1.71060467e+00
-6.01628721e-01 -1.22287691e-01 -6.77098215e-01 9.44035590e-01
-8.61861348e-01 -9.71096992e-01 -5.86499199e-02 6.03900969e-01
5.95042408e-01 -5.24453707e-02 1.07953298e+00 6.37457967e-01
-5.63293338e-01 7.93817937e-01 9.52100307e-02 7.65918493e-02
1.14974976e+00 -1.16711366e+00 1.73572883e-01 7.04624414e-01
6.15289271e-01 6.09310985e-01 8.66627812e-01 -3.08288485e-01
-1.35111296e+00 -6.46985173e-01 7.32590854e-01 -1.50441960e-01
4.95980978e-01 -3.24894518e-01 -7.04993606e-01 5.17002583e-01
6.82142079e-02 -3.40165347e-01 9.84821796e-01 2.89307803e-01
-2.90881544e-01 -2.73582965e-01 -6.41730845e-01 5.32902241e-01
7.98491955e-01 -6.69193864e-01 -5.77811658e-01 -1.56815037e-01
3.56525958e-01 -4.45904225e-01 -9.60771084e-01 2.88921982e-01
1.03840256e+00 -1.11969221e+00 8.76283467e-01 4.34831642e-02
2.03664139e-01 -5.36222935e-01 -3.03722858e-01 -1.39840281e+00
-2.40995780e-01 -1.10539770e+00 -2.23105326e-01 1.55242169e+00
3.77523154e-01 -3.14527273e-01 2.17521474e-01 -1.61555454e-01
-1.85558140e-01 -3.54623795e-01 -7.27195203e-01 -9.90378141e-01
-2.83242822e-01 -8.00872207e-01 3.32359701e-01 7.42002666e-01
1.77948281e-01 5.45285404e-01 -7.88034379e-01 2.32969493e-01
3.81405681e-01 5.07157624e-01 9.12881553e-01 -1.17483354e+00
-1.03264570e+00 -3.41907948e-01 -1.51435688e-01 -1.13167644e+00
-2.30683431e-01 -7.00419307e-01 -1.74642995e-01 -1.05711734e+00
7.23176524e-02 -9.57716480e-02 -4.44401532e-01 4.76871520e-01
-5.11736497e-02 6.16925240e-01 6.37498796e-01 5.61116517e-01
-3.79709125e-01 3.21322024e-01 1.03276086e+00 -6.92239404e-02
-6.78354919e-01 5.28778672e-01 -7.06797719e-01 8.99708092e-01
1.14876401e+00 -4.25404668e-01 -4.19161528e-01 -2.44059682e-01
1.37984246e-01 3.54110479e-01 2.22880132e-02 -1.34936154e+00
1.60905495e-01 2.11808428e-01 1.05459586e-01 -5.21777213e-01
7.10932672e-01 -4.12231833e-01 3.99573475e-01 2.99672455e-01
-2.42949367e-01 -8.42725039e-02 4.20138747e-01 1.25237867e-01
-4.80470777e-01 -9.99568254e-02 5.96784592e-01 6.74512014e-02
-2.41356015e-01 -2.76055962e-01 -3.94207180e-01 5.66076720e-03
5.29025316e-01 -1.92496091e-01 1.61580905e-01 -6.06344581e-01
-8.77893388e-01 -2.08366051e-01 6.38933852e-02 3.40402752e-01
3.02400231e-01 -1.22306466e+00 -8.17588568e-01 7.59838298e-02
4.34795581e-02 -3.40728045e-01 3.22733790e-01 9.93836582e-01
-1.71058863e-01 2.11091578e-01 -1.09427467e-01 -5.41239023e-01
-1.69749284e+00 5.52643947e-02 8.63437504e-02 -2.61509150e-01
-3.42435330e-01 7.98365295e-01 -2.12380458e-02 -2.17263609e-01
3.13389122e-01 -5.81081152e-01 -1.72555760e-01 3.79153281e-01
5.86122632e-01 5.76968491e-01 1.60807386e-01 -5.10207057e-01
-2.91150063e-01 8.01225364e-01 2.01189265e-01 -6.32237375e-01
1.05802906e+00 -8.30924213e-02 -1.52515089e-02 1.04646707e+00
8.56003881e-01 8.51457894e-01 -7.42024481e-01 2.39174627e-02
-1.85604379e-01 -2.54538387e-01 -9.20397341e-02 -1.00165558e+00
-7.17725456e-01 6.56494975e-01 3.86607885e-01 2.27972165e-01
1.53714693e+00 -2.87370414e-01 8.77078176e-01 2.41743982e-01
3.58023345e-01 -1.01206076e+00 1.19065098e-01 5.77468276e-01
1.21852577e+00 -6.49870694e-01 -1.22207247e-01 -3.42297673e-01
-5.70445836e-01 1.10606563e+00 1.36789903e-01 1.45339713e-01
3.69864702e-01 4.08710122e-01 3.33430290e-01 1.44795060e-01
-5.61203122e-01 -1.61650047e-01 4.44321841e-01 2.95184813e-02
7.65047252e-01 2.10701779e-01 -3.52839857e-01 1.17839253e+00
-7.48732150e-01 4.72257137e-02 3.69331926e-01 7.15820789e-01
-3.71432960e-01 -1.26097918e+00 -8.65996540e-01 -2.54325327e-02
-8.92636836e-01 -1.91786930e-01 -4.35743690e-01 5.03148079e-01
1.90990284e-01 1.26866949e+00 -2.27619141e-01 -5.80076993e-01
5.46759725e-01 1.94145083e-01 3.69793445e-01 -4.19157118e-01
-1.06537914e+00 7.15430617e-01 1.08799212e-01 -5.17867565e-01
-2.67836511e-01 -6.18265986e-01 -1.22212410e+00 -1.45971015e-01
-3.67475003e-01 3.56438726e-01 6.13849819e-01 5.93246639e-01
3.24517846e-01 7.68904865e-01 7.27142334e-01 -9.99209821e-01
-6.63280368e-01 -1.31518745e+00 -9.47090030e-01 1.82014704e-01
3.33882898e-01 -2.39962712e-01 -3.35320145e-01 3.13529074e-01] | [15.787206649780273, 5.426917552947998] |
a9ae7feb-0863-4097-9022-02a89c126ea2 | up-to-down-network-fusing-multi-scale-context | null | null | https://ieeexplore.ieee.org/abstract/document/9635888 | https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9635888 | Up-to-Down Network: Fusing Multi-Scale Context for 3D Semantic Scene Completion | An efficient 3D scene perception algorithm is a vital component for autonomous driving and robotics systems. In this paper, we focus on semantic scene completion, which is a task of jointly estimating the volumetric occupancy and semantic labels of objects. Since the real-world data is sparse and occluded, this is an extremely challenging task. We propose a novel framework, named Up-to-Down network (UDNet), to achieve the large-scale semantic scene completion with an encoder-decoder architecture for voxel grids. The novel up-to-down block can effectively aggregate multi-scale context information to improve labeling coherence, and the atrous spatial pyramid pooling module is leveraged to expand the receptive field while preserving detailed geometric information. Besides, the proposed multi-scale fusion mechanism efficiently aggregates global background information and improves the semantic completion accuracy. Moreover, to further satisfy the needs of different tasks, our UDNet can accomplish the multi-resolution semantic completion, achieving faster but coarser completion. Detailed experiments in the semantic scene completion benchmark of SemanticKITTI illustrate that our proposed framework surpasses the state-of-the-art methods with remarkable margins and a real-time inference speed by using only voxel grids as input. | ['Hongbo Zhang', 'Feng Wen', 'Wanlong Li', 'Yong liu', 'Chujuan Zhang', 'Tianxin Huang', 'Xuemeng Yang', 'Hao Zou'] | 2021-09-27 | null | null | null | ieee-international-workshop-on-intelligent | ['3d-semantic-scene-completion'] | ['computer-vision'] | [ 1.08190998e-01 -1.53713584e-01 8.83359686e-02 -4.58992004e-01
-6.63059592e-01 -9.92977917e-02 5.72825372e-01 1.41032666e-01
-4.18264896e-01 4.40314293e-01 1.68908983e-01 -3.40919830e-02
5.15814163e-02 -1.02944136e+00 -8.71223867e-01 -6.45111978e-01
2.49216929e-01 2.78669745e-01 7.48473465e-01 -2.04422325e-01
2.29897201e-01 4.39226866e-01 -1.90684843e+00 2.63397872e-01
1.01626682e+00 1.21042573e+00 9.43022549e-01 2.55950451e-01
-3.96065682e-01 8.08379650e-01 -4.91989292e-02 2.36449078e-01
1.31073892e-01 2.02360079e-01 -3.85591924e-01 1.93225026e-01
5.85283875e-01 -4.14323330e-01 -5.39334655e-01 1.13385582e+00
2.54532129e-01 3.63590896e-01 4.08874691e-01 -1.00419354e+00
-4.33730394e-01 6.56222031e-02 -7.71529973e-01 8.56783018e-02
2.48997100e-02 5.48379496e-02 7.29118645e-01 -1.12800992e+00
3.58210772e-01 1.54682827e+00 2.87251800e-01 1.07657425e-01
-9.33361173e-01 -8.18397582e-01 5.31037986e-01 3.46467942e-01
-1.59521985e+00 -3.17020535e-01 7.86700368e-01 -3.65790039e-01
9.35041487e-01 -7.03082308e-02 5.13471365e-01 5.26421249e-01
3.39104950e-01 6.64134979e-01 1.18272829e+00 1.89290673e-01
4.03468221e-01 -3.00668418e-01 -2.46257149e-02 8.80530357e-01
2.78919935e-01 -1.75468355e-01 -5.93547702e-01 2.27546915e-01
1.04330444e+00 4.74132806e-01 -1.93896051e-02 -4.69695330e-01
-1.31418264e+00 7.33006060e-01 9.45357442e-01 9.31014419e-02
-4.29765046e-01 4.88821775e-01 3.54953855e-01 -3.45240772e-01
4.69498962e-01 4.42263037e-02 -1.91879064e-01 3.75185281e-01
-8.28709602e-01 3.44270676e-01 3.16217721e-01 1.15370810e+00
1.21397209e+00 1.01341657e-01 -3.41476947e-01 7.54911900e-01
4.25735861e-01 6.48170769e-01 1.69089422e-01 -1.12689841e+00
6.33246303e-01 7.10866690e-01 1.20049760e-01 -1.00367665e+00
-4.98230129e-01 -6.34021878e-01 -1.05023503e+00 3.10400128e-01
-7.03310296e-02 4.65609372e-01 -1.34224916e+00 1.57951677e+00
6.46390080e-01 6.63179219e-01 2.96950247e-02 1.10665178e+00
1.06657958e+00 7.68804610e-01 4.02164519e-01 8.81008133e-02
1.65149426e+00 -1.23185325e+00 -7.11154819e-01 -5.73796928e-01
3.76113445e-01 -5.15031874e-01 8.34957361e-01 1.79541126e-01
-7.82728434e-01 -9.26956475e-01 -1.29603124e+00 -7.14844108e-01
-2.85959631e-01 2.22149678e-02 9.03407633e-01 9.92200077e-02
-8.40830386e-01 1.28599003e-01 -9.32682991e-01 -1.51164517e-01
8.46240759e-01 1.35865211e-01 -5.32953322e-01 -7.38623857e-01
-9.20973837e-01 7.60613561e-01 6.64685547e-01 6.85438812e-02
-1.17927718e+00 -7.03061223e-01 -1.25104618e+00 1.68292612e-01
5.59725881e-01 -7.09298491e-01 9.59909797e-01 -1.89891865e-03
-1.00695157e+00 5.78318596e-01 -4.52483773e-01 -2.73684412e-01
1.93581015e-01 -1.17340565e-01 -6.74684122e-02 8.10757056e-02
5.05170107e-01 9.58026469e-01 6.17049873e-01 -1.44019079e+00
-8.78242016e-01 -6.66329026e-01 7.16340840e-02 6.91395819e-01
1.45052215e-02 -6.89592302e-01 -8.42450440e-01 -3.74574631e-01
6.37668252e-01 -4.63924140e-01 -5.75943291e-01 2.43484005e-02
-1.52842581e-01 -3.80880058e-01 9.03394341e-01 -7.09404349e-01
7.88355052e-01 -2.33102751e+00 6.07221089e-02 -1.34277180e-01
5.91203749e-01 -1.58400625e-01 -8.29531178e-02 -2.36444604e-02
4.01002556e-01 -4.42066073e-01 -4.96851087e-01 -7.15696633e-01
1.28243119e-01 4.46883202e-01 -2.66473770e-01 5.22933006e-01
1.60945430e-01 1.00429738e+00 -1.04589713e+00 -5.59166193e-01
8.20277333e-01 6.59534216e-01 -7.61749387e-01 1.95490748e-01
-3.82569015e-01 7.30894923e-01 -7.84482598e-01 5.67451537e-01
1.07096851e+00 -2.41554976e-01 -3.11186194e-01 -3.28576267e-01
-3.07667315e-01 1.51969627e-01 -1.13788223e+00 2.49886012e+00
-6.96709931e-01 3.16130221e-01 2.67039388e-01 -1.05823088e+00
1.09642696e+00 -1.74025774e-01 3.05261731e-01 -9.62663472e-01
1.50653079e-01 2.39742175e-01 -5.55858493e-01 -2.94729710e-01
7.09700525e-01 -1.11580305e-01 -3.00206363e-01 -1.73976526e-01
-1.44779623e-01 -5.32063365e-01 -4.95664552e-02 1.29411757e-01
9.36215222e-01 2.16662019e-01 1.82607546e-01 -3.73388648e-01
5.83061099e-01 7.66129047e-02 8.30194414e-01 4.08397347e-01
-1.23090737e-01 4.54578072e-01 8.41341633e-03 -2.54972190e-01
-1.11327100e+00 -1.18904245e+00 -8.45925137e-02 8.73741865e-01
8.48119617e-01 -2.06191912e-01 -5.50982296e-01 -2.37669006e-01
1.97690621e-01 6.22941613e-01 -4.94405240e-01 -4.65557836e-02
-4.57014680e-01 -4.26338404e-01 1.24935888e-01 6.73448861e-01
1.12068248e+00 -8.46735179e-01 -6.65593147e-01 3.41959625e-01
-4.36111629e-01 -1.66272676e+00 -3.18693787e-01 1.19515799e-01
-7.35952616e-01 -7.63399959e-01 -3.82594675e-01 -9.98485148e-01
5.99423885e-01 7.31405854e-01 8.31349134e-01 -1.69725493e-01
-3.37830812e-01 -1.52311638e-01 -2.91246057e-01 -2.16939315e-01
1.25863984e-01 -1.00136660e-02 -1.18909419e-01 -9.43569690e-02
3.01770687e-01 -7.49298990e-01 -8.88702154e-01 1.57180563e-01
-1.01999307e+00 6.97626233e-01 6.86007261e-01 5.88207722e-01
9.76697028e-01 3.37186933e-01 4.68411058e-01 -6.09899998e-01
-4.28715646e-02 -4.57765549e-01 -6.92239106e-01 -1.28709689e-01
-1.67100504e-01 9.49652866e-02 5.74648023e-01 2.64874458e-01
-1.22715819e+00 3.23568344e-01 -3.96212667e-01 -7.15269387e-01
-2.89535433e-01 1.91960931e-01 -4.99848455e-01 -4.35564592e-02
2.30328009e-01 3.82039398e-01 -3.25446934e-01 -5.64751506e-01
6.83863163e-01 4.70202297e-01 8.48673403e-01 -4.70335960e-01
6.30492151e-01 9.15654659e-01 2.73318052e-01 -6.19563997e-01
-9.86468613e-01 -7.50823259e-01 -6.46942675e-01 -4.42583598e-02
1.21224034e+00 -1.54478955e+00 -6.42650604e-01 4.82975960e-01
-1.25979519e+00 -3.05346340e-01 -1.60864785e-01 4.82512325e-01
-6.18249893e-01 3.48862886e-01 -3.94723713e-01 -5.33398330e-01
-1.84004799e-01 -1.33034277e+00 1.53586376e+00 3.33115309e-01
5.74476063e-01 -5.61175048e-01 -3.13448668e-01 5.21336913e-01
2.78056949e-01 3.55180055e-01 6.93211555e-01 2.34434940e-03
-1.15705454e+00 1.77925467e-01 -9.23270762e-01 1.21755883e-01
5.99620864e-02 -7.93703198e-01 -1.15035951e+00 -2.64412194e-01
4.17020917e-02 -1.88373297e-01 1.21731246e+00 2.75921285e-01
1.50654578e+00 2.35507965e-01 -3.09361756e-01 8.41311216e-01
1.57120717e+00 -1.47360235e-01 5.54676294e-01 9.35763866e-03
1.23598123e+00 5.96836150e-01 8.99133801e-01 5.17999232e-01
9.71358836e-01 5.36116242e-01 8.48462462e-01 -2.50018716e-01
-5.04287183e-01 -5.86180091e-01 -9.83328521e-02 7.61114955e-01
3.40571463e-01 6.87420070e-02 -7.15105116e-01 5.99343896e-01
-2.05930567e+00 -5.68909764e-01 -8.96234661e-02 1.92167675e+00
3.47764730e-01 1.20399557e-01 -5.47961831e-01 -1.13362014e-01
5.88112414e-01 5.03362596e-01 -8.22078228e-01 9.62275490e-02
6.76333606e-02 2.57176787e-01 7.65393019e-01 6.23165667e-01
-1.11051857e+00 1.31212282e+00 5.07906103e+00 1.23898244e+00
-8.14803720e-01 4.86572266e-01 6.26461446e-01 1.88018009e-01
-4.09694791e-01 -1.62242085e-01 -8.68674219e-01 2.39586845e-01
2.55050898e-01 1.70982584e-01 3.62408310e-01 9.47394729e-01
2.38199949e-01 -3.82026911e-01 -8.19278002e-01 1.37961996e+00
1.94483727e-01 -1.41654491e+00 3.74472961e-02 5.75224198e-02
8.99018526e-01 3.81202340e-01 -2.27311343e-01 3.71105343e-01
4.67840880e-01 -1.02155268e+00 8.10192347e-01 4.59166437e-01
9.88465726e-01 -7.78887033e-01 5.29049516e-01 4.34118509e-01
-1.96171284e+00 -8.83479938e-02 -7.02343583e-01 -1.36124417e-01
3.33804548e-01 9.54834998e-01 -4.63411510e-01 6.74880564e-01
7.36374319e-01 9.43464339e-01 -3.80046785e-01 1.01797867e+00
-2.21968189e-01 -6.99412003e-02 -3.28197598e-01 1.45508781e-01
4.23915207e-01 -3.06941383e-02 2.61637300e-01 1.05504036e+00
3.29173028e-01 2.78231829e-01 7.42029071e-01 9.68722045e-01
-2.91766543e-02 -5.66217639e-02 -4.53112245e-01 4.31992322e-01
5.55755734e-01 1.45270252e+00 -8.68759990e-01 -5.69608331e-01
-3.44466180e-01 1.06210697e+00 6.01696193e-01 3.89219373e-01
-8.48981917e-01 -2.26119086e-01 7.91686654e-01 8.38605985e-02
4.48251545e-01 -7.50385880e-01 -6.90586269e-01 -1.09432459e+00
-9.25935060e-03 -1.92187950e-01 2.04170141e-02 -9.14135456e-01
-8.36542964e-01 3.89828801e-01 -8.63159522e-02 -9.54708338e-01
4.47156042e-01 -5.27392864e-01 -2.44550303e-01 9.52076733e-01
-2.13305616e+00 -1.26757717e+00 -1.05376387e+00 7.55526066e-01
8.54184389e-01 2.85100549e-01 3.88437957e-01 4.32610035e-01
-2.02016443e-01 -6.57136813e-02 -1.37847289e-01 -1.62191331e-01
3.35052162e-01 -1.02444780e+00 4.17359978e-01 8.29210579e-01
-2.64342248e-01 1.21298186e-01 4.14168835e-01 -7.40936399e-01
-1.36315703e+00 -1.66371000e+00 4.56404060e-01 -1.55857787e-01
2.56270021e-01 -6.52950168e-01 -7.75957048e-01 3.89907897e-01
-2.35286355e-01 2.33503774e-01 1.16796205e-02 -3.03132147e-01
-4.22579169e-01 -3.01765352e-01 -1.03857934e+00 4.49975908e-01
1.42410290e+00 -5.85068047e-01 -5.45078039e-01 3.72399181e-01
1.25040281e+00 -7.10445285e-01 -6.77451193e-01 6.55041873e-01
2.02761516e-01 -8.86378169e-01 1.13441980e+00 1.60833672e-01
4.66174811e-01 -7.70991862e-01 -5.97129405e-01 -9.70172644e-01
-5.23186624e-01 4.95399758e-02 3.41265015e-02 8.53063226e-01
-1.81867778e-01 -6.43532097e-01 7.90383041e-01 2.66373515e-01
-7.34450579e-01 -7.66346633e-01 -1.12017071e+00 -4.71285254e-01
-2.60869414e-01 -6.75657749e-01 7.74459660e-01 6.68969810e-01
-4.49671924e-01 3.40791553e-01 -1.60484120e-01 4.85453218e-01
9.38652873e-01 2.38626331e-01 8.14419389e-01 -1.04494095e+00
7.46549219e-02 -1.94765478e-01 -6.85710132e-01 -1.64827061e+00
7.94019550e-02 -9.42068338e-01 4.77317184e-01 -2.03336358e+00
2.31101155e-01 -7.05950439e-01 -4.16098833e-01 4.16095644e-01
-2.38856032e-01 5.34600258e-01 7.55609870e-02 6.54416829e-02
-9.21693563e-01 1.05106044e+00 1.66819322e+00 -1.88178360e-01
-2.87319869e-02 -6.64300323e-01 -5.82201540e-01 6.21870756e-01
5.61861277e-01 -1.65110320e-01 -5.56416273e-01 -5.73208094e-01
-1.23619750e-01 -4.73091006e-02 7.14101255e-01 -1.34534442e+00
3.64000589e-01 -1.64412007e-01 3.95264000e-01 -1.09995735e+00
7.87497938e-01 -7.26147950e-01 -6.75806254e-02 4.30511475e-01
1.72194839e-01 -2.81455278e-01 2.67943770e-01 8.34972143e-01
-3.13243717e-01 3.60816926e-01 7.60248482e-01 -2.61993498e-01
-1.42481732e+00 9.31377649e-01 4.83299680e-02 -1.08993210e-01
1.11335087e+00 -8.45305175e-02 -2.95701027e-01 6.54632524e-02
-5.04462302e-01 6.21079922e-01 5.16903698e-01 6.04270279e-01
8.82795036e-01 -1.39299381e+00 -7.05969512e-01 3.46821308e-01
3.25657278e-01 9.47974682e-01 7.85537958e-01 7.32050180e-01
-6.81912303e-01 5.18282771e-01 -2.11576030e-01 -9.70475793e-01
-8.59503865e-01 4.53907162e-01 1.16794743e-01 -1.99785739e-01
-9.58876431e-01 9.54649389e-01 1.13430679e+00 -4.74227488e-01
9.66078341e-02 -4.74084914e-01 -1.67859644e-01 -3.72893155e-01
5.06264448e-01 1.45786911e-01 5.91990389e-02 -7.24010587e-01
-3.82816076e-01 6.91886604e-01 4.46825512e-02 1.31247565e-01
1.16579056e+00 -4.53505695e-01 -1.78988099e-01 3.14178139e-01
1.11911905e+00 -4.08262551e-01 -1.66915715e+00 -4.88453656e-01
-6.01859987e-01 -7.06902385e-01 4.10376102e-01 -2.69980669e-01
-9.73143876e-01 1.12989795e+00 4.93329972e-01 -3.78636986e-01
1.03711724e+00 8.15209746e-02 9.98770356e-01 2.56498873e-01
7.53157020e-01 -9.92899656e-01 1.44847289e-01 5.50843954e-01
7.95723319e-01 -1.35637677e+00 1.49934933e-01 -8.80944550e-01
-5.01779377e-01 5.87776482e-01 8.33942950e-01 -3.95038873e-01
6.01551831e-01 7.02988356e-02 -3.97913545e-01 -3.71949852e-01
-4.71092761e-01 -4.45650697e-01 2.17210501e-01 5.19910336e-01
-1.36486351e-01 3.58144552e-01 8.17763060e-02 4.72255200e-01
-1.47550493e-01 -7.83192068e-02 2.83366982e-02 6.30284488e-01
-9.47465837e-01 -5.47041953e-01 -3.27877730e-01 5.24109185e-01
1.49344921e-01 -1.35480404e-01 4.79943633e-01 4.17004585e-01
6.04930282e-01 9.64766741e-01 2.74108827e-01 -3.74060303e-01
2.48525098e-01 -3.77108485e-01 4.19337004e-01 -7.95333564e-01
1.97191104e-01 2.66175997e-02 -1.98430404e-01 -9.46811795e-01
-3.48431259e-01 -4.47574675e-01 -1.72263002e+00 -2.14142606e-01
-9.72319543e-02 -1.91079170e-01 8.45978320e-01 1.17891777e+00
3.58262181e-01 1.00124681e+00 5.57032466e-01 -1.10981953e+00
3.41070592e-02 -8.42712045e-01 -6.94337547e-01 2.68999457e-01
3.44264030e-01 -1.06687796e+00 1.46529926e-02 -1.24858305e-01] | [8.397635459899902, -2.732337474822998] |
8c297c22-7bad-4f1b-97d6-020865a2a12a | stochastic-conservative-contextual-linear | 2203.15629 | null | https://arxiv.org/abs/2203.15629v1 | https://arxiv.org/pdf/2203.15629v1.pdf | Stochastic Conservative Contextual Linear Bandits | Many physical systems have underlying safety considerations that require that the strategy deployed ensures the satisfaction of a set of constraints. Further, often we have only partial information on the state of the system. We study the problem of safe real-time decision making under uncertainty. In this paper, we formulate a conservative stochastic contextual bandit formulation for real-time decision making when an adversary chooses a distribution on the set of possible contexts and the learner is subject to certain safety/performance constraints. The learner observes only the context distribution and the exact context is unknown, and the goal is to develop an algorithm that selects a sequence of optimal actions to maximize the cumulative reward without violating the safety constraints at any time step. By leveraging the UCB algorithm for this setting, we propose a conservative linear UCB algorithm for stochastic bandits with context distribution. We prove an upper bound on the regret of the algorithm and show that it can be decomposed into three terms: (i) an upper bound for the regret of the standard linear UCB algorithm, (ii) a constant term (independent of time horizon) that accounts for the loss of being conservative in order to satisfy the safety constraint, and (ii) a constant term (independent of time horizon) that accounts for the loss for the contexts being unknown and only the distribution being known. To validate the performance of our approach we perform extensive simulations on synthetic data and on real-world maize data collected through the Genomes to Fields (G2F) initiative. | ['Baskar Ganapathysubramanian', 'Soumik Sarkar', 'Shana Moothedath', 'Talukder Jubery', 'Xian Yeow Lee', 'Jiabin Lin'] | 2022-03-29 | null | null | null | null | ['decision-making-under-uncertainty', 'decision-making-under-uncertainty'] | ['medical', 'reasoning'] | [ 4.82471466e-01 3.48125488e-01 -3.12505037e-01 -1.55290633e-01
-8.22775841e-01 -9.30130661e-01 1.85630828e-01 4.05072927e-01
-2.67072886e-01 1.01166999e+00 -4.16606158e-01 -8.34871650e-01
-7.07990348e-01 -9.14395452e-01 -1.10578561e+00 -1.00646806e+00
-1.86274499e-01 4.30877626e-01 2.81990580e-02 3.49910744e-02
-2.15085223e-02 3.50519240e-01 -1.03848326e+00 -1.32941946e-01
6.15631700e-01 1.46080458e+00 5.27355708e-02 6.84777617e-01
3.79491448e-01 6.68523788e-01 -5.25618851e-01 -8.87236074e-02
5.79780519e-01 -3.99235189e-01 -7.89397478e-01 2.11536959e-01
-2.15620935e-01 -5.59586942e-01 -4.88911755e-02 1.31458247e+00
3.58789265e-01 2.21139073e-01 2.27923214e-01 -1.39295352e+00
-4.61039832e-03 8.35461617e-01 -6.78585887e-01 -1.18704047e-02
-5.33390231e-02 8.63299519e-02 1.00377369e+00 4.55431461e-01
3.22533041e-01 1.01092720e+00 1.01564519e-01 5.36466360e-01
-1.48821068e+00 -4.69666451e-01 6.20967031e-01 -1.06500156e-01
-1.16453528e+00 -1.94950730e-01 3.45844716e-01 -2.91149110e-01
4.29156601e-01 6.45430624e-01 4.83049870e-01 8.92327666e-01
6.70652390e-02 7.82761097e-01 1.07340753e+00 -5.68223715e-01
9.72465813e-01 5.64605705e-02 3.14053297e-01 1.75467417e-01
4.22092974e-01 7.19093919e-01 -2.33218834e-01 -5.41349769e-01
3.18668634e-01 -3.75677347e-02 -3.63024890e-01 -6.00455165e-01
-6.56064928e-01 7.65526175e-01 4.15930487e-02 -2.06511885e-01
-5.21566153e-01 5.11496544e-01 2.13477030e-01 3.63847315e-01
3.42973053e-01 1.60834655e-01 -4.85074669e-01 9.66057703e-02
-6.19411469e-01 4.18428630e-01 7.79096186e-01 8.93779814e-01
2.96482503e-01 -1.29524842e-01 -5.59757590e-01 2.08966717e-01
1.63141545e-03 7.76780963e-01 -2.10231528e-01 -1.03297067e+00
5.62353730e-01 -1.01376042e-01 1.00454021e+00 -5.24940610e-01
-5.84449880e-02 -4.28562135e-01 -3.31766129e-01 4.09264088e-01
5.77737272e-01 -5.09304106e-01 -7.93483555e-01 2.24612880e+00
3.66504222e-01 1.68447927e-01 2.18757987e-02 7.81590819e-01
-4.11427617e-01 9.22340214e-01 -1.79225087e-01 -8.00699711e-01
9.13104713e-01 -2.15345472e-01 -6.35959506e-01 -2.42773101e-01
3.38965893e-01 -1.72441721e-01 5.69645524e-01 5.68759561e-01
-1.01943958e+00 3.12225908e-01 -1.25827134e+00 6.78812444e-01
-2.15403852e-03 -2.56223291e-01 4.96793419e-01 9.70360518e-01
-5.54418802e-01 4.37943876e-01 -7.73526132e-01 -1.35655075e-01
1.91511869e-01 3.48162740e-01 1.23804919e-02 6.73617125e-02
-9.31696534e-01 5.67480683e-01 6.20948076e-01 2.29744032e-01
-1.34832442e+00 -6.73125923e-01 -4.71176416e-01 5.42671442e-01
1.15801394e+00 -3.81016850e-01 1.48188317e+00 -7.45592475e-01
-1.52804387e+00 3.03438842e-01 2.50474960e-01 -8.80568266e-01
8.46123219e-01 -7.58948401e-02 1.51999682e-01 -1.03839330e-01
-1.43311903e-01 -1.66190520e-01 6.01668835e-01 -1.06836331e+00
-1.06193960e+00 -5.38072288e-01 6.52037561e-01 -7.16353506e-02
7.70771950e-02 -3.15395147e-01 1.46240450e-03 -2.87324697e-01
-1.40515625e-01 -1.35827792e+00 -4.64825839e-01 -1.85282037e-01
-5.46241760e-01 2.67023832e-01 4.30235088e-01 -4.08310324e-01
1.14356017e+00 -2.13666153e+00 -6.13979809e-02 3.93348366e-01
-5.15026927e-01 4.84347865e-02 7.30964076e-03 3.54733407e-01
5.77436797e-02 1.04749046e-01 -3.79684895e-01 -2.71368064e-02
3.33666243e-03 2.01866776e-01 -7.89962232e-01 6.42541051e-01
-2.57054120e-01 3.03172410e-01 -7.95057893e-01 2.39185035e-01
1.16490163e-01 -2.67557651e-01 -3.42188001e-01 1.81866005e-01
-7.36098051e-01 1.16521806e-01 -7.38817632e-01 2.17275515e-01
7.62333453e-01 1.89425483e-01 4.54130054e-01 3.30251396e-01
-1.36044785e-01 -7.35796764e-02 -1.38593340e+00 9.40852344e-01
-3.79208684e-01 3.67079340e-02 4.56802726e-01 -1.18880653e+00
9.99909416e-02 3.60873878e-01 3.89257312e-01 -3.28609109e-01
3.12623173e-01 2.73639467e-02 -1.71132967e-01 -8.10461789e-02
-7.99606517e-02 -2.64314681e-01 -3.97267073e-01 5.28519452e-01
-2.69197315e-01 -6.06256053e-02 -9.22063738e-02 9.42824483e-02
1.12700582e+00 -8.54194164e-02 4.35654908e-01 -3.96996796e-01
1.91509485e-01 -1.17112711e-01 8.51566970e-01 1.09221458e+00
-2.87988603e-01 1.71178654e-02 1.00923431e+00 -1.62634984e-01
-7.38167048e-01 -7.55716562e-01 1.42078280e-01 7.77003944e-01
2.69803792e-01 1.85611039e-01 -4.94719326e-01 -6.40298545e-01
3.28102440e-01 1.30589318e+00 -9.06754017e-01 -2.81052411e-01
5.18793315e-02 -6.57467008e-01 -5.03825489e-04 1.49521276e-01
2.52001464e-01 -4.19230074e-01 -1.05361068e+00 1.94434285e-01
3.70243639e-02 -1.06475794e+00 -4.69589144e-01 5.19283235e-01
-4.35393572e-01 -8.48134458e-01 -1.95043534e-01 1.84676021e-01
5.71848869e-01 1.94679752e-01 4.38568205e-01 -4.93993551e-01
8.84914622e-02 2.86465049e-01 -3.33734415e-02 -8.84242058e-01
-2.86971688e-01 -2.95343697e-01 -1.53498113e-01 1.40099689e-01
-3.50986958e-01 -2.01904893e-01 -3.60242665e-01 1.77764401e-01
-9.02430534e-01 -3.20728198e-02 1.61431104e-01 7.80802786e-01
7.56525576e-01 6.46117628e-01 4.07931983e-01 -1.01224875e+00
3.94782931e-01 -3.75088990e-01 -1.57940996e+00 6.78910375e-01
-6.46858335e-01 2.29434311e-01 4.18181092e-01 -6.01723552e-01
-1.21463203e+00 2.77807921e-01 5.78425348e-01 -2.14054763e-01
1.47643313e-01 5.30352473e-01 -5.46233356e-01 6.29795194e-02
2.28197441e-01 6.76224530e-02 -2.85554200e-01 -1.43077627e-01
3.21591228e-01 4.97579306e-01 3.76178354e-01 -1.11694145e+00
5.83673298e-01 5.00500739e-01 4.55158859e-01 -4.24906939e-01
-1.07248402e+00 -9.91748720e-02 4.26373519e-02 -1.62694365e-01
3.92766118e-01 -5.86584151e-01 -1.23119307e+00 2.50106603e-01
-6.25102997e-01 -5.19764364e-01 -4.16208059e-01 3.44384819e-01
-1.00152504e+00 6.77853301e-02 -1.98460758e-01 -1.75740814e+00
-6.07918836e-02 -8.85659695e-01 5.86851776e-01 2.75919616e-01
2.85717010e-01 -4.47685122e-01 -1.53498843e-01 2.34730676e-01
1.20217785e-01 5.77586532e-01 9.78986561e-01 -3.55166763e-01
-8.93431365e-01 -2.75643677e-01 1.74843132e-01 2.41816133e-01
-7.61502907e-02 -1.21915884e-01 -7.55748928e-01 -4.82891023e-01
2.26253808e-01 -1.84095562e-01 5.37284136e-01 6.41013801e-01
1.26816702e+00 -8.28083813e-01 -4.74832803e-01 3.73015106e-01
1.56514597e+00 8.15052927e-01 2.04011410e-01 2.72978753e-01
-3.66856195e-02 5.28196573e-01 1.00829363e+00 7.83122659e-01
-8.86464044e-02 6.52185321e-01 8.13516259e-01 4.30835217e-01
8.83723438e-01 -1.94094032e-01 3.67950946e-01 -4.96863931e-01
3.29956710e-01 -6.68242991e-01 -5.29967368e-01 8.43649387e-01
-2.28664970e+00 -9.15808856e-01 4.66231138e-01 3.06420255e+00
8.78585160e-01 3.58729303e-01 2.18623385e-01 1.51571035e-01
7.07661152e-01 -2.91642457e-01 -1.01451230e+00 -8.03102553e-01
8.28779563e-02 -1.53952003e-01 1.18439353e+00 6.72712266e-01
-1.02386379e+00 5.14370322e-01 5.60799980e+00 8.71254921e-01
-1.09055555e+00 -3.97300385e-02 8.85868728e-01 -5.92893243e-01
-6.67095929e-02 3.09436440e-01 -5.72579086e-01 5.85445821e-01
1.12677407e+00 -7.22244322e-01 8.21110547e-01 7.22212136e-01
3.05000991e-01 -4.59204316e-01 -1.27934158e+00 3.36698413e-01
-7.05330908e-01 -1.03012145e+00 -4.77284908e-01 2.72640169e-01
7.47939408e-01 -2.48865947e-01 2.03590453e-01 9.82829705e-02
9.08002853e-01 -6.44123912e-01 1.29778552e+00 2.30341062e-01
5.20035625e-01 -1.25462258e+00 6.16566181e-01 8.53469491e-01
-6.57501459e-01 -6.37996614e-01 -8.74652416e-02 -1.24229498e-01
1.40667230e-01 7.36484885e-01 -6.55153692e-01 6.39429986e-01
4.31466699e-01 -2.39662737e-01 3.92675757e-01 1.03450465e+00
-3.10641944e-01 7.74936259e-01 -7.33064175e-01 1.31010090e-03
3.91679466e-01 -2.55397797e-01 6.42451286e-01 6.72901750e-01
3.16535145e-01 4.51586068e-01 3.17719132e-01 8.05099964e-01
2.88470864e-01 -2.93261379e-01 -3.18449974e-01 -1.76154271e-01
6.80815041e-01 7.54645050e-01 -6.22016490e-01 -3.62751223e-02
-6.96717352e-02 7.12791026e-01 6.44630641e-02 5.78928232e-01
-9.00456965e-01 -1.25345513e-01 7.74536133e-01 -2.13864848e-01
4.91940290e-01 1.14859894e-01 -4.00427103e-01 -5.75018406e-01
1.01805873e-01 -4.64954644e-01 7.22710729e-01 -4.58530039e-01
-9.87648845e-01 -7.78730735e-02 2.44398087e-01 -6.69295192e-01
-2.44528472e-01 -2.68610746e-01 -4.25993681e-01 9.29668725e-01
-1.21971536e+00 -7.93625414e-01 4.10637587e-01 3.08096379e-01
5.16163036e-02 3.30438614e-01 6.16835058e-01 -1.48336545e-01
-7.84208298e-01 3.57112318e-01 5.75064838e-01 -3.93680394e-01
1.47022232e-01 -1.12795067e+00 -2.14310452e-01 1.05811727e+00
-3.89416277e-01 3.68573844e-01 1.13051689e+00 -5.05024552e-01
-1.29367244e+00 -1.11434126e+00 3.19849372e-01 4.23596613e-02
6.94094777e-01 -3.64324421e-01 -3.27868521e-01 8.59399080e-01
-3.20205331e-01 6.12767451e-02 3.73495251e-01 -1.80742592e-02
-1.25025660e-01 -3.94786149e-01 -1.51488614e+00 4.38351572e-01
6.11617744e-01 -2.60843188e-01 -2.78281718e-02 3.19149792e-01
9.33680594e-01 -4.91424471e-01 -3.54785711e-01 4.12612140e-01
5.51519573e-01 -5.20412087e-01 5.80649972e-01 -8.59394073e-01
-1.19003959e-01 -3.67454618e-01 -4.55045044e-01 -1.31604350e+00
-2.00226933e-01 -1.12829411e+00 -3.25922400e-01 1.10992014e+00
4.26724613e-01 -7.02847838e-01 6.93819463e-01 1.15402555e+00
4.56501842e-01 -7.41919696e-01 -1.39441383e+00 -1.12079549e+00
3.63514312e-02 -3.39598000e-01 7.48630166e-01 5.33348083e-01
2.76155416e-02 -1.10174276e-01 -7.07899272e-01 7.08745360e-01
7.22688019e-01 6.23073399e-01 2.48583555e-01 -7.43155062e-01
-6.68782234e-01 -7.20707253e-02 5.26223369e-02 -6.53676569e-01
1.06364056e-01 -1.65229812e-01 4.35096651e-01 -9.89512265e-01
3.68421704e-01 -5.64822614e-01 -6.26282036e-01 5.00747144e-01
2.66828928e-02 -8.14325988e-01 3.26590449e-01 -4.27150100e-01
-4.72770452e-01 2.82632530e-01 7.41223395e-01 -2.33572125e-01
-2.61811733e-01 7.74395406e-01 -8.75546575e-01 4.09700960e-01
5.95477462e-01 -5.42357147e-01 -8.34828258e-01 -1.76196292e-01
3.04965794e-01 8.19643021e-01 1.71937257e-01 -4.95307177e-01
-3.27186510e-02 -9.83030200e-01 -2.59808123e-01 -6.56476915e-01
2.43227392e-01 -1.31430006e+00 4.25585836e-01 8.52902651e-01
-6.55649185e-01 -6.41945422e-01 1.10600859e-01 1.05733979e+00
4.35637414e-01 -3.36703181e-01 8.03095818e-01 1.72573403e-01
-1.91715553e-01 2.91608930e-01 -4.23749298e-01 -2.52467632e-01
1.45922387e+00 3.98333460e-01 -2.91450948e-01 -7.56074369e-01
-6.96639717e-01 7.19787776e-01 2.24483475e-01 7.43199959e-02
1.47019178e-02 -8.87354016e-01 -4.18870807e-01 -1.37663692e-01
-1.22953601e-01 -1.57613471e-01 2.02527195e-01 4.44969445e-01
-8.34572315e-03 7.10806847e-01 1.07744418e-01 -2.88121164e-01
-1.21985984e+00 1.04896402e+00 4.37741488e-01 -2.91184813e-01
-5.16892448e-02 6.88902557e-01 2.97890872e-01 1.83798552e-01
3.99514496e-01 -4.16272581e-01 3.61491084e-01 -1.46925375e-01
4.84625667e-01 3.87629598e-01 1.70435570e-02 -7.97677040e-02
-2.65354574e-01 -1.20899707e-01 1.04895063e-01 -4.71711129e-01
1.24674976e+00 -4.19945359e-01 1.43863440e-01 3.16436470e-01
6.02314293e-01 1.39031298e-02 -1.41488290e+00 -2.82010823e-01
5.45361638e-02 -5.91175616e-01 2.78736055e-01 -1.15921855e+00
-9.34178352e-01 4.06921476e-01 6.70896471e-01 4.73993599e-01
1.33333015e+00 -4.10801113e-01 4.73954052e-01 3.95411402e-01
9.03158009e-01 -1.02034545e+00 -7.18919158e-01 2.50161886e-01
5.32956004e-01 -6.82512045e-01 -6.74271882e-02 -3.64650011e-01
-5.35942852e-01 5.98252714e-01 2.19343379e-01 1.26519755e-01
6.24756336e-01 5.18518925e-01 -3.97100180e-01 4.34875995e-01
-1.09621656e+00 -1.93627521e-01 -1.49753541e-01 3.80034447e-01
-2.77618408e-01 7.12954342e-01 -4.14307922e-01 9.73934293e-01
7.63172135e-02 2.37394765e-01 6.46083891e-01 1.08052456e+00
-4.99467313e-01 -8.98395658e-01 -5.58214366e-01 2.42138043e-01
-7.66297221e-01 3.29070777e-01 -2.18235418e-01 3.09002817e-01
1.80687495e-02 1.19530046e+00 -3.08917463e-01 1.32835642e-01
3.11466843e-01 -1.27499968e-01 4.85337734e-01 -4.74030674e-01
-4.34012450e-02 1.81660533e-01 2.71967411e-01 -7.02710271e-01
9.22045633e-02 -6.28536046e-01 -9.59599137e-01 -2.77269542e-01
-6.64702117e-01 3.54321063e-01 7.44735062e-01 9.69971955e-01
1.65650740e-01 4.77015108e-01 8.37263763e-01 -1.97105661e-01
-1.35783195e+00 -4.88822460e-01 -8.81611586e-01 -5.71882240e-02
5.18762827e-01 -4.23994899e-01 -3.69524956e-01 -2.74029881e-01] | [4.568202972412109, 3.1694788932800293] |
50db2290-bf38-43fb-932e-e7bfea3ca650 | infant-brain-mri-segmentation-with-dilated | 1912.12570 | null | https://arxiv.org/abs/1912.12570v2 | https://arxiv.org/pdf/1912.12570v2.pdf | Infant brain MRI segmentation with dilated convolution pyramid downsampling and self-attention | In this paper, we propose a dual aggregation network to adaptively aggregate different information in infant brain MRI segmentation. More precisely, we added two modules based on 3D-UNet to better model information at different levels and locations. The dilated convolution pyramid downsampling module is mainly to solve the problem of loss of spatial information on the downsampling process, and it can effectively save details while reducing the resolution. The self-attention module can integrate the remote dependence on the feature maps in two dimensions of spatial and channel, effectively improving the representation ability and discriminating ability of the model. Our results are compared to the winners of iseg2017's first evaluation, the DICE ratio of WM and GM increased by 0.7%, and CSF is comparable.In the latest evaluation of the iseg-2019 cross-dataset challenge,we achieve the first place in the DICE of WM and GM, and the DICE of CSF is second. | ['Ying WEI', 'Zhihao Lei', 'Yunlong Zhou', 'Lin Qi'] | 2019-12-29 | null | null | null | null | ['infant-brain-mri-segmentation'] | ['medical'] | [-1.30568653e-01 6.19490333e-02 2.22793102e-01 -4.29413348e-01
-2.83673644e-01 -1.20640494e-01 2.07115576e-01 8.15023109e-02
-7.46074319e-01 5.01758337e-01 4.78686750e-01 1.44533724e-01
-7.27624223e-02 -4.90179658e-01 -4.93515581e-01 -7.36487806e-01
-2.93023497e-01 9.68632475e-02 5.22408307e-01 9.69425440e-02
1.54478282e-01 5.22229791e-01 -1.28914726e+00 6.39081776e-01
1.16305363e+00 1.19794452e+00 4.59006339e-01 1.76055849e-01
-7.39067718e-02 5.34338832e-01 -3.62726569e-01 -3.11385274e-01
1.81920737e-01 -2.35326126e-01 -8.87624800e-01 -5.64649463e-01
3.78728449e-01 -6.08093321e-01 -3.73148233e-01 1.26522279e+00
9.86692846e-01 -1.12425901e-01 8.32128525e-01 -8.39965880e-01
-7.52682149e-01 8.20640862e-01 -9.76586521e-01 8.19769323e-01
-1.34497315e-01 6.13553785e-02 2.97333628e-01 -7.54558265e-01
4.28189397e-01 1.18365729e+00 7.88757980e-01 5.06427407e-01
-9.93654370e-01 -9.94883060e-01 2.04678893e-01 3.50970447e-01
-1.26664627e+00 -3.03347588e-01 4.80560571e-01 -5.89591205e-01
8.31636250e-01 1.71561047e-01 7.38764048e-01 7.56293714e-01
3.34902525e-01 8.35373163e-01 1.23157835e+00 4.00143154e-02
4.45872312e-03 -3.18545729e-01 2.05438763e-01 5.16550899e-01
2.41947860e-01 -5.29715791e-03 -2.61505276e-01 2.09623292e-01
1.05504060e+00 4.98264059e-02 -3.39875877e-01 -1.79834828e-01
-1.35566008e+00 5.92944682e-01 8.42732072e-01 6.75735056e-01
-4.70191926e-01 -3.06329340e-01 4.25509661e-01 3.77708748e-02
9.07794833e-01 3.91849130e-01 -5.17380536e-01 2.41273746e-01
-1.16267622e+00 -7.32739568e-02 8.96893218e-02 6.71689153e-01
3.06912839e-01 -2.83463597e-01 -5.15161097e-01 9.68414128e-01
2.14412913e-01 1.31682456e-01 9.07100677e-01 -9.24290299e-01
7.15643525e-01 4.29472506e-01 -4.54997718e-01 -8.85979712e-01
-1.14510584e+00 -6.01093888e-01 -1.11648238e+00 1.46143019e-01
3.99353266e-01 -2.70414084e-01 -1.05660784e+00 1.78746045e+00
2.61919796e-01 2.03277767e-01 -5.54074407e-01 1.04126704e+00
1.27414775e+00 1.17892422e-01 2.26636559e-01 -2.19131425e-01
1.26895714e+00 -1.14255476e+00 -7.29970455e-01 -1.27427340e-01
6.12623692e-01 -5.16018510e-01 6.91347063e-01 4.79896460e-03
-1.41126907e+00 -6.21427000e-01 -1.06197584e+00 -5.21325730e-02
-4.00469214e-01 -2.51432303e-02 4.92514640e-01 4.51031268e-01
-1.47007978e+00 8.40250850e-01 -8.73416185e-01 1.50118247e-01
9.76677835e-01 3.96391690e-01 -5.52969038e-01 -1.26484230e-01
-1.20168567e+00 1.16668546e+00 3.68609935e-01 -2.97512189e-02
-4.66306895e-01 -1.30509043e+00 -5.05266964e-01 1.42169118e-01
-2.42586106e-01 -5.90132058e-01 7.86335528e-01 -4.40398902e-01
-1.09085226e+00 9.82519805e-01 1.41970143e-01 -3.75049740e-01
7.46694744e-01 -1.66689068e-01 -3.93402815e-01 3.91122252e-01
4.58176047e-01 1.07519686e+00 5.29246569e-01 -7.46255338e-01
-3.10927719e-01 -9.13415134e-01 -1.54671609e-01 2.21657231e-01
-4.02494781e-02 1.78035110e-01 -2.96431363e-01 -8.12997043e-01
5.64547718e-01 -4.84416217e-01 -2.25376487e-01 -7.69267976e-02
7.47051835e-03 3.65338996e-02 4.90289718e-01 -1.20627916e+00
1.04884684e+00 -2.25015759e+00 -1.34493992e-01 -8.58278796e-02
6.99520111e-01 1.86738878e-01 -2.76260614e-01 -5.10301471e-01
-6.46396995e-01 2.69355327e-01 -4.42035288e-01 -2.57467598e-01
-4.26654637e-01 -3.19700122e-01 2.23209038e-01 6.57346427e-01
2.89072156e-01 9.14005756e-01 -8.49234939e-01 -5.46199501e-01
2.82417163e-02 5.99995613e-01 -7.57609129e-01 -8.83517861e-02
5.90746641e-01 8.05887163e-01 -1.69855461e-01 1.78570464e-01
1.29204130e+00 -1.53289270e-02 -2.11897120e-01 -4.54168737e-01
-1.84482858e-01 1.11783773e-01 -8.32878411e-01 1.99935377e+00
-2.34542415e-01 4.01113391e-01 5.81424497e-02 -9.66335952e-01
5.86484909e-01 3.35422009e-01 7.73982227e-01 -1.17613673e+00
3.16990972e-01 1.92389727e-01 4.62518722e-01 -3.86347443e-01
-2.50241160e-01 1.34253860e-01 4.65552777e-01 5.29783249e-01
2.22182572e-01 -5.32458909e-02 5.62543534e-02 -5.74908368e-02
8.59215319e-01 -3.35795097e-02 4.67370600e-02 -8.73822927e-01
3.14901531e-01 -7.32932866e-01 4.54744935e-01 5.25501311e-01
-4.22973305e-01 1.10873592e+00 5.80461740e-01 -4.51294750e-01
-8.37472022e-01 -1.08229470e+00 -5.10597110e-01 8.37234318e-01
-7.36849615e-04 9.73678846e-03 -1.26564181e+00 -7.38424480e-01
-2.21048728e-01 4.71467078e-01 -1.02325070e+00 -2.11221322e-01
-6.98332787e-01 -1.22885263e+00 4.02006716e-01 7.52385616e-01
9.22396481e-01 -9.39776301e-01 -8.03403139e-01 7.76406378e-02
-3.21351320e-01 -1.05317891e+00 -6.89833105e-01 1.57353282e-01
-1.07068634e+00 -8.91295731e-01 -1.24641967e+00 -6.60519242e-01
6.66186631e-01 2.40499862e-02 8.69491637e-01 1.62200913e-01
-3.42824101e-01 -2.98047662e-01 -3.23528916e-01 -4.01928842e-01
1.07883729e-01 1.83678284e-01 -1.34248748e-01 -3.38110328e-01
-6.75801770e-04 -7.98157036e-01 -1.10415494e+00 1.98145971e-01
-7.42070496e-01 2.76454896e-01 4.23290670e-01 7.58447349e-01
3.40828389e-01 -2.94163615e-01 6.82589889e-01 -5.31366527e-01
5.36800086e-01 -5.74837744e-01 -2.28979081e-01 1.52938589e-01
-5.23627579e-01 -4.82560843e-02 3.16860437e-01 -4.20099139e-01
-8.93000007e-01 -3.04228723e-01 -2.17107967e-01 -9.84933078e-02
-8.22863877e-02 1.05124578e-01 -1.05865568e-01 -1.19727202e-01
2.35193089e-01 -1.01854503e-01 5.68143018e-02 -8.43427181e-01
1.74051479e-01 3.02822560e-01 3.68529588e-01 -2.34937817e-01
3.17704976e-02 3.80593568e-01 -1.00883000e-01 -2.96356678e-01
-7.16535032e-01 -4.41295542e-02 -1.14323938e+00 2.07945630e-02
1.29658067e+00 -8.49875629e-01 -3.78604084e-01 7.13146150e-01
-1.41733027e+00 -2.68811226e-01 -1.51858315e-01 7.19925463e-01
-3.38905960e-01 3.38909358e-01 -7.97364891e-01 -9.27755758e-02
-5.62762082e-01 -1.56549895e+00 7.13823438e-01 3.66783589e-01
-2.34669051e-03 -7.56581604e-01 -1.08776011e-01 2.21666083e-01
7.32572258e-01 1.82078913e-01 1.09723687e+00 -6.35252655e-01
5.31611927e-02 1.18913069e-01 -8.47765684e-01 3.69389743e-01
2.09855405e-03 -5.24287462e-01 -1.16713798e+00 -2.64324397e-01
2.93449134e-01 1.92909360e-01 1.16283786e+00 8.92261982e-01
1.74938107e+00 1.25790741e-02 -1.64103106e-01 9.02580142e-01
9.16970134e-01 4.47490156e-01 7.11942732e-01 6.85026795e-02
6.64677441e-01 7.83379197e-01 5.17077670e-02 1.54913783e-01
5.06419718e-01 5.39779603e-01 2.95973927e-01 -4.91292298e-01
-7.84713566e-01 3.76297310e-02 -1.64125010e-01 1.09682584e+00
-1.76934600e-01 3.77046406e-01 -9.46561813e-01 4.75938439e-01
-1.58066320e+00 -7.85514414e-01 -5.68871610e-02 1.87607586e+00
7.87503958e-01 -2.73284823e-01 -3.81303206e-03 -1.24720708e-01
8.42189014e-01 1.13671869e-01 -5.87680578e-01 -3.10757548e-01
-2.01945961e-01 4.70811367e-01 5.19041717e-01 3.48634273e-01
-1.15917158e+00 5.99976957e-01 7.14051008e+00 8.69686544e-01
-1.11913371e+00 6.13664210e-01 1.10549831e+00 -2.97626823e-01
-5.55586927e-02 -6.58846021e-01 -3.75000328e-01 8.42458308e-01
6.76993906e-01 5.86041547e-02 4.07439858e-01 3.63680899e-01
-1.18641313e-02 -4.13120091e-02 -8.97391915e-01 9.82787311e-01
1.12099750e-02 -1.13384080e+00 -1.90129802e-01 -5.57786413e-02
7.71522164e-01 4.84171391e-01 2.18305752e-01 1.54236153e-01
-2.46099755e-01 -1.33641219e+00 6.49066091e-01 6.66800380e-01
9.70807195e-01 -6.76504791e-01 8.72651041e-01 2.67487019e-01
-9.51483846e-01 2.76343394e-02 -3.65876496e-01 7.07477704e-02
-7.10497722e-02 6.89698875e-01 -4.25497591e-01 3.95318389e-01
1.19850302e+00 4.91267532e-01 -8.62199366e-01 1.48309171e+00
1.08959759e-02 1.58461824e-01 -2.81607956e-01 3.70366037e-01
1.37736678e-01 -2.72671610e-01 1.96761042e-01 1.34778714e+00
3.54326785e-01 2.96499014e-01 -3.32608074e-01 8.98222506e-01
-5.30427918e-02 2.24540144e-01 -2.18071818e-01 6.71580970e-01
1.58688605e-01 1.13318896e+00 -9.22367394e-01 -4.94921893e-01
-3.71892899e-01 7.65518665e-01 3.50326538e-01 3.60800564e-01
-9.04094696e-01 -3.45441103e-01 4.82550651e-01 1.49528101e-01
2.55471706e-01 1.48425817e-01 -8.02589297e-01 -1.13654256e+00
-1.08633012e-01 -5.85762441e-01 1.85409129e-01 -7.32572556e-01
-1.08011067e+00 8.77985001e-01 8.59553218e-02 -6.63352907e-01
2.16537297e-01 -3.89280379e-01 -5.76968491e-01 1.22587883e+00
-1.32618678e+00 -6.58235550e-01 -1.23963825e-01 4.56274629e-01
1.46499649e-01 -8.41747746e-02 6.98137641e-01 8.73045921e-01
-6.43606186e-01 7.72542238e-01 -6.90659210e-02 3.01612645e-01
5.61740041e-01 -9.46363866e-01 4.74798173e-01 8.66849899e-01
-2.83108711e-01 6.25123143e-01 2.52884150e-01 -5.97449541e-01
-3.00085962e-01 -8.09632540e-01 5.22518754e-01 -2.09814340e-01
3.98368388e-01 -1.29567847e-01 -1.07521129e+00 3.68558466e-01
1.88144907e-01 2.43497714e-01 4.58283037e-01 -2.03849241e-01
-3.63263547e-01 -2.46035866e-02 -1.59734142e+00 3.89740467e-01
1.18622780e+00 -2.53361523e-01 -6.43872261e-01 1.32955268e-01
1.07771802e+00 -4.51554537e-01 -1.13138497e+00 7.97876060e-01
5.67086041e-01 -1.15862656e+00 1.18703270e+00 -5.81316352e-01
6.10750318e-01 -9.88896713e-02 5.08924499e-02 -1.52963150e+00
-7.27037787e-01 6.65123165e-02 1.06945649e-01 9.65988696e-01
3.70195061e-01 -7.20479846e-01 4.92878735e-01 5.98625124e-01
-3.53647172e-01 -1.07723224e+00 -1.35109067e+00 -4.28655595e-01
5.66797256e-01 -2.85046726e-01 1.20612824e+00 9.57948506e-01
-1.95651159e-01 5.15910387e-02 9.82181281e-02 -8.99417475e-02
6.69093013e-01 -2.57483810e-01 -2.67687708e-01 -1.10616696e+00
1.77833498e-01 -8.88484180e-01 -3.46826971e-01 -8.76546204e-01
-8.68232250e-02 -1.14173877e+00 -3.49247843e-01 -1.62689555e+00
6.07254386e-01 -2.42684394e-01 -7.44848669e-01 4.18760151e-01
-2.53452748e-01 3.55455548e-01 3.90555352e-01 -5.08563668e-02
-3.35419029e-01 3.68168086e-01 1.87384224e+00 -8.32588300e-02
7.95790832e-03 -1.61525980e-01 -8.61751556e-01 9.34792221e-01
7.91514039e-01 -3.98412198e-01 -2.07145065e-01 -1.00919759e+00
-3.12757224e-01 -7.66953006e-02 1.92179829e-01 -1.03254974e+00
4.74772193e-02 3.33052903e-01 1.00109875e+00 -7.17375934e-01
3.72549035e-02 -6.30320728e-01 -2.04075232e-01 6.17246866e-01
-1.94762036e-01 1.47286683e-01 3.39719355e-01 -1.98069677e-01
-1.94965191e-02 -9.77883488e-02 1.22956049e+00 -1.14347935e-01
-2.52642155e-01 7.12772429e-01 -8.53215456e-02 3.05358499e-01
7.72978544e-01 -8.70526358e-02 -4.37776476e-01 1.15449265e-01
-9.02664483e-01 1.92425266e-01 2.13168174e-01 3.81657183e-01
6.40033066e-01 -1.30892563e+00 -7.33346283e-01 4.62825567e-01
-3.72300446e-01 -5.45938574e-02 7.38847852e-01 1.39831042e+00
-6.03832960e-01 2.88161814e-01 -7.63998866e-01 -4.52015311e-01
-8.11051905e-01 3.45208675e-01 4.71380830e-01 -3.75950813e-01
-6.64138079e-01 1.10500288e+00 6.88118994e-01 -3.58990490e-01
1.04941554e-01 -5.57206333e-01 -6.56750560e-01 3.76841009e-01
1.04105377e+00 6.45628512e-01 2.99962282e-01 -7.16997504e-01
-6.47780478e-01 7.75175273e-01 -2.06396818e-01 2.62089521e-02
1.68924284e+00 -2.50415117e-01 -4.74044085e-01 -5.70412241e-02
1.54796350e+00 -2.95609027e-01 -1.24796391e+00 -1.03324585e-01
-1.47321492e-01 -3.45896006e-01 2.82138646e-01 -1.12126923e+00
-1.56926048e+00 1.30077755e+00 1.22233713e+00 4.06779796e-02
1.19451070e+00 -9.45370924e-03 8.50545764e-01 -5.45953989e-01
2.57022142e-01 -8.98996234e-01 -2.37495184e-01 5.74661434e-01
1.09913492e+00 -1.22162962e+00 -3.77691276e-02 -1.36389762e-01
-6.63275778e-01 1.00736320e+00 6.98831260e-01 -1.53539017e-01
9.05676723e-01 3.94041061e-01 -2.58496702e-02 -2.74288446e-01
-2.41659001e-01 1.35919437e-01 7.17987061e-01 8.58932734e-01
5.66137075e-01 1.64735571e-01 -5.76598942e-01 1.10441434e+00
-2.70201087e-01 -1.70783073e-01 9.69215482e-02 1.52141675e-01
-3.49063218e-01 -4.41924185e-01 -1.81723699e-01 7.75697827e-01
-7.58618116e-01 -4.47942078e-01 1.62644342e-01 4.92812008e-01
6.73599184e-01 6.26658201e-01 2.40687653e-01 -2.26237088e-01
3.21049631e-01 -9.04459208e-02 5.72203159e-01 -4.71798211e-01
-6.47129893e-01 1.21257463e-02 -3.61053556e-01 -8.94374788e-01
-3.09821635e-01 -7.26614654e-01 -1.32273221e+00 -1.61877930e-01
-4.66824360e-02 -1.32617787e-01 5.74511766e-01 8.80988955e-01
4.00621593e-01 8.17913473e-01 4.49199021e-01 -8.17773044e-01
-2.19408736e-01 -1.14365876e+00 -7.02232420e-01 3.01875919e-01
3.71608824e-01 -8.41951191e-01 -1.90321848e-01 -4.64643598e-01] | [14.305098533630371, -2.3886730670928955] |
24b70ddf-53b8-47dd-84f8-0f3bb245415d | hierarchical-explanations-for-video-action | 2301.00436 | null | https://arxiv.org/abs/2301.00436v3 | https://arxiv.org/pdf/2301.00436v3.pdf | Hierarchical Explanations for Video Action Recognition | To interpret deep neural networks, one main approach is to dissect the visual input and find the prototypical parts responsible for the classification. However, existing methods often ignore the hierarchical relationship between these prototypes, and thus can not explain semantic concepts at both higher level (e.g., water sports) and lower level (e.g., swimming). In this paper inspired by human cognition system, we leverage hierarchal information to deal with uncertainty: When we observe water and human activity, but no definitive action it can be recognized as the water sports parent class. Only after observing a person swimming can we definitively refine it to the swimming action. To this end, we propose HIerarchical Prototype Explainer (HIPE) to build hierarchical relations between prototypes and classes. HIPE enables a reasoning process for video action classification by dissecting the input video frames on multiple levels of the class hierarchy, our method is also applicable to other video tasks. The faithfulness of our method is verified by reducing accuracy-explainability trade off on ActivityNet and UCF-101 while providing multi-level explanations. | ['Nanne van Noord', 'Teng Long', 'Sadaf Gulshad'] | 2023-01-01 | null | null | null | null | ['action-classification'] | ['computer-vision'] | [ 9.56393927e-02 4.18979973e-01 -1.96532056e-01 -4.36834127e-01
1.96233138e-01 -6.44669354e-01 5.60882092e-01 2.72039890e-01
-9.58168283e-02 3.68018121e-01 3.68496835e-01 -2.54115671e-01
-2.44855955e-01 -8.83747458e-01 -7.55361676e-01 -4.52832580e-01
1.78207159e-01 2.70257473e-01 4.87413079e-01 5.83848581e-02
2.52708137e-01 2.03583658e-01 -2.00546932e+00 9.83591378e-01
7.32714057e-01 8.89701128e-01 9.20739770e-02 5.36257565e-01
-9.10173655e-02 1.39741898e+00 -4.96399224e-01 -3.35482806e-01
-5.98758385e-02 -7.45208681e-01 -1.24135900e+00 3.12130570e-01
4.79868740e-01 -4.56180036e-01 -1.93010822e-01 1.17064583e+00
-2.42049545e-01 1.88187346e-01 6.98457301e-01 -1.87703085e+00
-8.32098901e-01 1.02018440e+00 -1.73803896e-01 9.08555016e-02
3.07319611e-01 1.70360021e-02 1.15045106e+00 -6.63889647e-01
5.39302409e-01 1.47976780e+00 6.13141775e-01 7.49827981e-01
-9.30279493e-01 -6.23641074e-01 6.71182275e-01 9.70365584e-01
-1.18099225e+00 -2.41157692e-02 4.85744625e-01 -6.50874674e-01
7.72543550e-01 4.55206752e-01 1.13479364e+00 1.07323134e+00
6.24423735e-02 8.44140410e-01 1.01969492e+00 -5.73191978e-02
6.78204358e-01 -1.63938209e-01 5.49854696e-01 8.40173721e-01
5.07073224e-01 -8.98837820e-02 -7.21601963e-01 2.48947129e-01
8.38176906e-01 3.59955013e-01 -3.49703997e-01 -3.70212674e-01
-1.12639904e+00 6.34761453e-01 7.14619935e-01 2.96198905e-01
-3.05635124e-01 3.19904417e-01 2.40772232e-01 -4.11632657e-02
-1.93702832e-01 5.21744430e-01 -3.77533495e-01 1.96085740e-02
-7.76221573e-01 1.29056633e-01 6.41876221e-01 8.77345800e-01
8.30292046e-01 -6.04238175e-02 -2.27182671e-01 2.92472333e-01
3.71750832e-01 -5.71632385e-02 6.15933418e-01 -1.48527765e+00
3.21074314e-02 1.01203144e+00 -7.32793435e-02 -9.47793305e-01
-4.62616652e-01 -1.69183955e-01 -9.39887702e-01 3.52506161e-01
2.17458203e-01 3.12246472e-01 -1.03564143e+00 1.66176713e+00
1.57690793e-01 4.40417141e-01 3.24227512e-01 1.19730747e+00
1.06904423e+00 4.24495816e-01 3.60237360e-01 -5.86487390e-02
1.64911962e+00 -1.10871172e+00 -5.86962104e-01 -2.97627240e-01
3.55724066e-01 -1.32975532e-02 1.04624403e+00 5.31177461e-01
-6.89058423e-01 -8.76221299e-01 -1.11617851e+00 -2.28380278e-01
-2.74894476e-01 2.00461242e-02 7.18047917e-01 2.33703300e-01
-9.49921489e-01 7.33010352e-01 -1.00299764e+00 -5.27278006e-01
3.43849957e-01 1.15740128e-01 -5.05350649e-01 4.26308550e-02
-1.20536005e+00 7.77742624e-01 7.66999602e-01 1.22523010e-01
-1.10158992e+00 -5.28442502e-01 -8.68577242e-01 4.22390491e-01
3.66731524e-01 -8.81138802e-01 1.13501501e+00 -1.11476612e+00
-1.00243926e+00 4.98123318e-01 -6.38505965e-02 -7.23356783e-01
3.29549789e-01 -2.69272447e-01 -1.09855555e-01 4.71758515e-01
1.90728664e-01 1.08546925e+00 6.87252045e-01 -1.55694520e+00
-1.01008928e+00 -3.08808953e-01 6.69471085e-01 3.00850779e-01
-3.64640117e-01 -4.01112407e-01 -3.57079268e-01 -3.63765627e-01
6.79426908e-01 -8.68952870e-01 -1.86809599e-02 2.45817676e-01
-3.77107412e-01 -3.03664178e-01 8.35867405e-01 -5.10590017e-01
1.17987049e+00 -1.98540819e+00 2.18739733e-01 -4.12892662e-02
5.76518714e-01 -7.75520578e-02 9.41713601e-02 4.36428189e-02
-1.58190683e-01 4.05158043e-01 -2.25937307e-01 -2.39262488e-02
-1.76794324e-02 7.18139529e-01 -3.15416068e-01 1.06037416e-01
1.87693238e-01 7.03975439e-01 -1.08670139e+00 -6.75293624e-01
2.42622718e-01 3.60436350e-01 -8.14984381e-01 1.37159973e-01
-1.46938011e-01 2.13897780e-01 -3.15279275e-01 5.02113581e-01
2.91970044e-01 -3.84051830e-01 3.38475466e-01 -6.54060245e-01
-4.98463539e-03 -4.10555005e-02 -1.30465341e+00 1.42243004e+00
-1.70507610e-01 7.02233136e-01 -4.27668929e-01 -1.04245400e+00
7.36956775e-01 1.89535990e-01 1.65637538e-01 -2.60045469e-01
-6.11582436e-02 -1.15440644e-01 2.75983989e-01 -7.88654506e-01
4.31987941e-01 -8.90088081e-02 1.69610813e-01 1.98796347e-01
3.88111733e-02 1.49780428e-02 2.61158586e-01 3.39463115e-01
1.05957460e+00 4.76705849e-01 5.77077985e-01 -2.83829063e-01
4.37185735e-01 2.55219698e-01 8.61803532e-01 8.54744971e-01
-3.75061780e-01 7.27794349e-01 7.54036546e-01 -7.63673663e-01
-7.66362607e-01 -1.02808046e+00 1.92236468e-01 1.03936601e+00
7.57770360e-01 -5.63143194e-01 -9.04139161e-01 -6.90109074e-01
-1.60294488e-01 7.64591217e-01 -1.01452279e+00 -4.15880799e-01
-4.22576219e-01 -2.56906629e-01 4.92133796e-01 9.19515967e-01
7.73254514e-01 -1.12372434e+00 -1.16726017e+00 -4.20921780e-02
-5.93475342e-01 -1.14615345e+00 -4.69669029e-02 2.25077808e-01
-8.09479177e-01 -1.43976212e+00 -9.45380256e-02 -5.78258932e-01
7.13718176e-01 3.07197988e-01 1.19090927e+00 6.33485079e-01
-1.09357126e-01 3.73822808e-01 -5.91869533e-01 -2.13397279e-01
-3.51112336e-01 -3.28498453e-01 5.81723116e-02 7.50701427e-02
4.12296861e-01 -5.00224411e-01 -8.27890575e-01 4.29983824e-01
-9.25531209e-01 4.40581590e-01 5.36381841e-01 6.09387219e-01
5.95086455e-01 3.26958477e-01 3.56962234e-02 -6.09550059e-01
3.18093598e-01 -3.98163140e-01 -5.78907989e-02 5.27659118e-01
-4.63650703e-01 2.37264186e-01 5.39963961e-01 -5.92662871e-01
-1.00453436e+00 1.19092382e-01 6.88145757e-02 -4.82738674e-01
-5.28410017e-01 3.13918859e-01 -2.23741204e-01 4.69605178e-01
7.06384063e-01 2.05855399e-01 -4.44741428e-01 -8.70187432e-02
4.61457133e-01 4.70058978e-01 7.74856031e-01 -5.41103065e-01
4.26310182e-01 6.47278845e-01 1.35360926e-01 -5.83141804e-01
-1.12755728e+00 -2.57057637e-01 -8.06883216e-01 -4.67106909e-01
1.32757390e+00 -8.86599004e-01 -1.06323552e+00 2.55020726e-02
-1.31361973e+00 -1.21110156e-01 -2.07896248e-01 4.67241406e-01
-5.97766697e-01 4.42404628e-01 -6.60395741e-01 -6.86231673e-01
-3.92855667e-02 -1.13954401e+00 1.02989089e+00 2.46174246e-01
-6.83986366e-01 -6.54862225e-01 -4.64235097e-01 4.38710451e-01
-8.89606550e-02 3.06441158e-01 1.14386737e+00 -6.80863082e-01
-4.76161242e-01 2.61440128e-01 -3.33428562e-01 1.79139823e-01
-1.45122269e-02 1.62366778e-01 -8.10806811e-01 1.44729257e-01
-7.01716095e-02 -2.59927630e-01 8.91427159e-01 1.58441857e-01
1.26902735e+00 -4.72437501e-01 -3.31450582e-01 3.10856372e-01
1.11830544e+00 3.32520485e-01 6.72488391e-01 4.66465443e-01
8.09769928e-01 8.45284700e-01 6.37878954e-01 3.44685167e-01
6.37027740e-01 4.83036667e-01 8.71868134e-01 6.54523969e-02
-2.81957060e-01 -3.58825147e-01 2.57358044e-01 4.21110421e-01
-2.38784984e-01 -1.68864831e-01 -8.87818456e-01 4.46013868e-01
-2.25361848e+00 -1.32559705e+00 -4.20240939e-01 1.61830354e+00
5.84388375e-01 9.52003896e-02 -7.80951679e-02 4.65495974e-01
7.70169020e-01 -1.69699267e-01 -5.88019550e-01 -1.22133128e-01
4.47642356e-02 -3.01538557e-01 -1.41797230e-01 5.08310616e-01
-8.43484461e-01 9.24107552e-01 5.80821609e+00 4.52168077e-01
-7.60833561e-01 3.14139053e-02 4.93432909e-01 1.93765685e-01
-3.46097797e-01 3.01797897e-01 -6.48843944e-01 1.96389616e-01
2.56551594e-01 2.89856214e-02 3.21122795e-01 1.04225075e+00
8.51694494e-02 -9.34445113e-02 -1.63542068e+00 9.04757738e-01
1.85123414e-01 -1.30132997e+00 6.22208297e-01 -2.89401025e-01
4.49831784e-01 -6.66972756e-01 -3.56494039e-01 4.60458308e-01
2.34553501e-01 -1.06192684e+00 1.24231589e+00 4.93929178e-01
1.54416531e-01 -4.54712421e-01 5.89832008e-01 5.24177611e-01
-1.39682460e+00 -3.68782043e-01 -4.16722804e-01 -3.73770922e-01
-8.75830874e-02 6.60787746e-02 -4.61280763e-01 4.29728955e-01
1.05381978e+00 8.49755883e-01 -6.38328910e-01 8.04444730e-01
-7.74844229e-01 4.10543948e-01 5.71402088e-02 5.32974824e-02
1.25546753e-01 6.41764002e-03 2.06889525e-01 9.66814935e-01
3.30282032e-01 5.29709160e-01 1.98149130e-01 9.50910091e-01
2.03225344e-01 -4.70403492e-01 -2.24627957e-01 2.84708947e-01
4.99428928e-01 1.08763063e+00 -1.03195369e+00 -8.00313771e-01
-1.14533767e-01 1.07727432e+00 3.15955371e-01 3.21667761e-01
-1.11041880e+00 -8.12546909e-02 7.57046700e-01 1.24730705e-03
1.55617267e-01 -4.15717997e-02 -3.74522537e-01 -1.20910203e+00
-5.03549017e-02 -8.55599701e-01 5.90667248e-01 -1.29354215e+00
-9.15033698e-01 7.84345090e-01 2.45842144e-01 -1.36375976e+00
-1.58226162e-01 -7.89854050e-01 -5.07010818e-01 2.02430487e-01
-1.08595407e+00 -1.12978899e+00 -8.42119992e-01 6.35454416e-01
8.42498004e-01 2.79094279e-01 6.48561358e-01 -7.64613450e-02
-3.09186369e-01 1.31755292e-01 -8.05758357e-01 1.77478775e-01
5.07625192e-02 -1.30241549e+00 4.20810878e-02 9.76174593e-01
5.06792963e-01 8.02331507e-01 9.70291615e-01 -5.93343973e-01
-8.75541031e-01 -1.08592069e+00 6.61064863e-01 -4.50853348e-01
6.36655092e-01 -1.49285510e-01 -1.05752075e+00 6.47068858e-01
2.33822748e-01 -1.99672967e-01 6.38549924e-01 2.13772035e-03
-6.60671473e-01 4.63794656e-02 -8.98600876e-01 8.34938765e-01
1.48483849e+00 -3.74011606e-01 -1.12773061e+00 1.25805646e-01
8.21702361e-01 -2.08247900e-01 -7.28648603e-01 5.48475266e-01
7.89087355e-01 -1.40832686e+00 8.78648281e-01 -9.78868544e-01
9.56394672e-01 -8.16186726e-01 -2.48949781e-01 -1.12053120e+00
-4.98213321e-01 1.70840919e-02 -3.50744516e-01 1.00735569e+00
1.75521046e-01 -1.03315309e-01 8.99312019e-01 6.79310739e-01
-3.22271377e-01 -5.75724304e-01 -7.69429028e-01 -8.01697612e-01
-2.98175156e-01 -5.25543272e-01 7.01091826e-01 9.35976267e-01
3.00591052e-01 2.90614456e-01 -4.06632096e-01 5.35980403e-01
6.56744838e-01 3.90114903e-01 4.86619413e-01 -1.28298748e+00
-3.71950269e-01 -5.09162247e-01 -7.07634330e-01 -1.01931393e+00
1.61556020e-01 -6.41784251e-01 3.92688781e-01 -1.91778648e+00
5.05770028e-01 4.46177572e-02 -3.09312910e-01 9.74840164e-01
-1.93504900e-01 2.38447472e-01 3.98075759e-01 4.50029641e-01
-8.11006069e-01 3.64501297e-01 1.12878525e+00 -2.55076051e-01
8.82108323e-03 -3.83560330e-01 -8.30424070e-01 1.16905153e+00
7.81312585e-01 -4.87624884e-01 -5.50306201e-01 -6.31694019e-01
2.30325684e-01 -6.08134596e-03 7.84808397e-01 -1.32813346e+00
3.13499898e-01 -4.08404559e-01 3.74623656e-01 -3.09235573e-01
1.93760648e-01 -1.04637849e+00 3.53942841e-01 8.58982384e-01
-5.32206953e-01 1.65327359e-02 -8.14897716e-02 6.63265049e-01
-3.99024099e-01 -2.35667065e-01 5.75438619e-01 -3.01843494e-01
-1.26625383e+00 1.18296847e-01 -6.15679622e-01 -2.10436136e-01
1.00311899e+00 -4.97066796e-01 -5.91030300e-01 -2.86926001e-01
-1.13032448e+00 2.79737920e-01 3.09720337e-01 6.73966646e-01
8.01232815e-01 -1.41647255e+00 -2.74728209e-01 -6.97127357e-02
3.54015887e-01 -1.30918130e-01 3.41076910e-01 6.47772610e-01
-6.29826248e-01 9.27293152e-02 -5.17710507e-01 -8.06384742e-01
-1.37699783e+00 6.37410402e-01 4.92141992e-01 2.27961794e-01
-5.82118988e-01 7.65515924e-01 6.24715209e-01 -5.16782068e-02
4.20682967e-01 -8.51148546e-01 -7.16575384e-01 -3.77091579e-02
5.95587373e-01 3.23596895e-01 -4.13955361e-01 -6.88113987e-01
-5.20621836e-01 5.81732929e-01 3.24315578e-01 1.15169778e-01
1.05831063e+00 -1.10385880e-01 -2.36654788e-01 4.79802698e-01
7.81225860e-01 -6.08674347e-01 -1.38129759e+00 2.83684283e-01
5.28254732e-02 -2.92484701e-01 -3.18009615e-01 -5.99097669e-01
-9.34769034e-01 1.04776764e+00 5.38038492e-01 3.45144302e-01
1.08320510e+00 1.73217297e-01 2.77487844e-01 5.87784588e-01
2.60482848e-01 -8.81875753e-01 3.64604563e-01 3.96289438e-01
9.47278082e-01 -1.13622451e+00 -1.11829918e-02 -6.74039304e-01
-9.11795795e-01 1.35587299e+00 1.12579274e+00 5.61826974e-02
2.48630390e-01 6.20301552e-02 -2.08308473e-02 -3.75095397e-01
-8.31739545e-01 -2.96879083e-01 2.84075052e-01 5.01080751e-01
2.92840034e-01 8.36979300e-02 -2.21671104e-01 9.64873672e-01
-1.85193688e-01 -1.04773335e-01 7.28607178e-01 6.95535302e-01
-7.68284798e-01 -5.43126404e-01 -2.43243739e-01 2.01455891e-01
-7.06315935e-02 -7.13881478e-03 -5.69733322e-01 8.32058668e-01
7.22098053e-01 1.08970606e+00 3.54520530e-01 -6.90482974e-01
3.88116956e-01 -5.89468926e-02 3.51325899e-01 -6.35318518e-01
-4.12686288e-01 -4.15900558e-01 -1.10819280e-01 -6.95654631e-01
-8.30694795e-01 -5.09365797e-01 -1.81629062e+00 -7.66326785e-02
9.22887295e-04 3.18737566e-01 1.92327291e-01 1.25841188e+00
1.44660905e-01 5.42904854e-01 1.50781311e-02 -6.09880686e-01
-1.15178451e-01 -5.86599708e-01 -3.78188670e-01 8.82729292e-01
1.47504598e-01 -7.84502387e-01 -4.03741270e-01 6.65961206e-01] | [8.945298194885254, 0.9106769561767578] |
6d8f5f2f-1b0f-427f-9082-67b022b2229b | target-specified-sequence-labeling-with-multi | null | null | https://aclanthology.org/2021.naacl-main.145 | https://aclanthology.org/2021.naacl-main.145.pdf | Target-specified Sequence Labeling with Multi-head Self-attention for Target-oriented Opinion Words Extraction | Opinion target extraction and opinion term extraction are two fundamental tasks in Aspect Based Sentiment Analysis (ABSA). Many recent works on ABSA focus on Target-oriented Opinion Words (or Terms) Extraction (TOWE), which aims at extracting the corresponding opinion words for a given opinion target. TOWE can be further applied to Aspect-Opinion Pair Extraction (AOPE) which aims at extracting aspects (i.e., opinion targets) and opinion terms in pairs. In this paper, we propose Target-Specified sequence labeling with Multi-head Self-Attention (TSMSA) for TOWE, in which any pre-trained language model with multi-head self-attention can be integrated conveniently. As a case study, we also develop a Multi-Task structure named MT-TSMSA for AOPE by combining our TSMSA with an aspect and opinion term extraction module. Experimental results indicate that TSMSA outperforms the benchmark methods on TOWE significantly; meanwhile, the performance of MT-TSMSA is similar or even better than state-of-the-art AOPE baseline models. | ['He Liu', 'Ninghua Wang', 'Yuyao Tang', 'Yanghui Rao', 'Yuhao Feng'] | 2021-06-01 | null | null | null | naacl-2021-4 | ['target-oriented-opinion-words-extraction'] | ['natural-language-processing'] | [ 3.88592571e-01 -3.68604176e-02 -5.39411865e-02 -5.08175194e-01
-1.27846229e+00 -6.08930767e-01 6.79575086e-01 8.11053962e-02
-2.61234850e-01 2.83332437e-01 3.90613616e-01 -5.23977935e-01
4.25985157e-01 -6.38278067e-01 -3.70710641e-01 -7.81500936e-01
3.37554216e-01 3.66997898e-01 -5.90764694e-02 -5.42663455e-01
3.39277118e-01 -9.46210846e-02 -1.28194070e+00 6.33242607e-01
9.88123357e-01 1.13406539e+00 -1.98043808e-01 5.65325618e-01
-5.03348053e-01 6.56777680e-01 -7.70144582e-01 -1.03234410e+00
-1.18681908e-01 -3.70862275e-01 -7.95427978e-01 2.93511808e-01
6.24726750e-02 6.84665740e-02 4.87735778e-01 9.78706777e-01
5.39658368e-01 -2.17247624e-02 8.56957853e-01 -1.24513531e+00
-8.14009249e-01 7.16003120e-01 -8.96329999e-01 -9.62349772e-02
2.51102924e-01 -2.61736009e-02 1.58028793e+00 -1.36858070e+00
2.54865617e-01 1.40561831e+00 4.14908946e-01 2.85163283e-01
-6.13747418e-01 -4.84583527e-01 8.31436276e-01 1.61734223e-01
-7.71325529e-01 -3.61171037e-01 7.55949557e-01 -9.71066356e-02
1.39624298e+00 3.36727142e-01 7.86329687e-01 9.03227925e-01
5.53077757e-01 1.54213464e+00 1.18396688e+00 -5.63392520e-01
1.39872402e-01 2.59398758e-01 7.65079379e-01 4.85687464e-01
1.03808455e-01 -5.26168764e-01 -8.29122365e-01 -3.66557427e-02
-2.58499943e-02 -2.08719864e-01 -1.23465285e-01 1.05364695e-01
-1.19567752e+00 7.58193851e-01 2.57841218e-02 3.91689092e-01
-5.60471237e-01 -3.08415145e-01 6.68494523e-01 5.76941669e-01
1.22890913e+00 4.85894531e-01 -1.11441076e+00 8.91466662e-02
-4.75347728e-01 8.54347721e-02 9.99809444e-01 1.19678247e+00
6.59531593e-01 8.67300555e-02 -6.71831012e-01 7.16602802e-01
4.79311675e-01 9.27736402e-01 6.02311671e-01 -2.51548469e-01
4.73260880e-01 8.29079509e-01 9.20197293e-02 -7.47136235e-01
-2.02880070e-01 -5.27262568e-01 -5.67047894e-01 -7.64893666e-02
-2.69835681e-01 -4.40440029e-01 -1.25478840e+00 1.24634838e+00
4.21765000e-01 -4.39398706e-01 3.94945532e-01 6.24677241e-01
1.02980459e+00 9.15179670e-01 -8.54694098e-02 -5.51717281e-01
1.87247264e+00 -1.74430108e+00 -9.69795763e-01 -7.99845278e-01
7.45789409e-01 -1.07166350e+00 1.08660913e+00 3.86480600e-01
-8.92347991e-01 -2.38155648e-01 -8.07139933e-01 -1.27543220e-02
-7.05611587e-01 1.93284363e-01 6.94594443e-01 5.45972586e-01
-1.03610098e+00 -1.74570724e-01 -5.18387973e-01 -1.06233977e-01
1.90293163e-01 2.96759754e-01 -1.25418335e-01 1.10887839e-02
-1.29193914e+00 7.44535625e-01 -3.06288563e-02 3.44112724e-01
-6.15771413e-01 -5.77113748e-01 -1.28013313e+00 -1.38319179e-01
5.17889738e-01 -8.40846479e-01 1.49335408e+00 -1.28568518e+00
-1.56252277e+00 8.02035749e-01 -1.05963171e+00 3.89878079e-02
-4.03108865e-01 -6.05788231e-01 -6.72047496e-01 -3.04549903e-01
5.70278585e-01 3.95001560e-01 1.06211030e+00 -1.30864334e+00
-7.51867592e-01 -4.77268249e-01 3.21159899e-01 4.79531139e-01
-5.65529525e-01 5.52137852e-01 -6.05714738e-01 -7.26449907e-01
-2.07817510e-01 -7.55788684e-01 -4.25522268e-01 -5.16773582e-01
-5.47436416e-01 -6.93221152e-01 7.19913065e-01 -5.28576255e-01
1.40978527e+00 -1.82370532e+00 1.98070243e-01 1.35299459e-01
-1.87788028e-02 4.29946810e-01 -6.75185502e-01 4.47991520e-01
-1.34908035e-01 1.26529068e-01 -2.92187154e-01 -6.13685608e-01
8.19496736e-02 -1.84950829e-01 -4.67268527e-01 -1.95549369e-01
5.49673498e-01 1.28330863e+00 -1.07164180e+00 -5.07515311e-01
-2.56305963e-01 3.43383759e-01 -1.86884373e-01 3.58355612e-01
-3.54960918e-01 -3.64676379e-02 -6.55238330e-01 1.09200478e+00
5.24035454e-01 -3.25877160e-01 1.76425744e-02 -3.62683296e-01
-1.59054592e-01 7.54824340e-01 -5.74551165e-01 1.23767436e+00
-1.02594924e+00 4.19883221e-01 -5.47276065e-02 -6.12911284e-01
8.76585066e-01 4.47822034e-01 1.12347029e-01 -6.43593788e-01
3.84249538e-01 2.34446049e-01 -1.08286859e-02 -3.12529743e-01
7.09051907e-01 -3.83922875e-01 -1.84986189e-01 5.90589464e-01
6.95046633e-02 -3.18743914e-01 5.18030405e-01 2.15108171e-01
8.58880579e-01 -7.21899644e-02 5.12432754e-01 -2.78252453e-01
9.97759461e-01 1.38317812e-02 5.68674505e-01 2.98572451e-01
-9.01047066e-02 4.60655540e-01 5.51563263e-01 -2.11144596e-01
-5.42749822e-01 -5.08109987e-01 3.11057836e-01 1.28888452e+00
6.32408867e-03 -7.27889180e-01 -4.47402894e-01 -1.21675658e+00
-5.15824676e-01 8.26110482e-01 -6.73900902e-01 1.99539766e-01
-4.34593290e-01 -8.85736823e-01 -2.70884186e-01 5.03813684e-01
3.41809690e-01 -1.43680108e+00 9.70219746e-02 2.30345488e-01
-3.70445728e-01 -1.14904892e+00 -9.81784880e-01 1.30156979e-01
-5.01804173e-01 -7.15912402e-01 -9.49914634e-01 -1.07685590e+00
6.44292831e-01 5.83261490e-01 1.52147889e+00 -1.98049709e-01
5.30093372e-01 2.93312877e-01 -9.15872037e-01 -9.80679214e-01
8.52685496e-02 3.44885290e-01 -2.70915896e-01 4.32662070e-01
8.39355171e-01 -3.97428781e-01 -5.49923897e-01 1.58639010e-02
-9.26736295e-01 2.53793187e-02 1.03564131e+00 7.01114655e-01
6.80059731e-01 -3.05837333e-01 6.52184010e-01 -1.23422754e+00
8.88709188e-01 -2.73093611e-01 -3.08917791e-01 4.66274828e-01
-7.93172002e-01 -7.05191642e-02 5.10740578e-01 -3.13240767e-01
-1.15690589e+00 -4.24857706e-01 -5.05109489e-01 7.96302184e-02
1.86959758e-01 1.05169213e+00 -4.85463411e-01 3.76001030e-01
-1.18482120e-01 4.42340463e-01 -2.72003800e-01 -1.42557248e-01
4.56491202e-01 1.00734544e+00 -1.35770306e-01 -5.66976294e-02
6.13698065e-01 3.35393965e-01 -4.09376889e-01 -6.91251457e-01
-1.69601417e+00 -8.43763649e-01 -2.56498784e-01 -1.34110570e-01
8.14210951e-01 -1.07933688e+00 -3.78952205e-01 9.53546524e-01
-1.46767926e+00 9.86063108e-02 -1.78629067e-02 1.60179988e-01
-4.44452763e-02 3.16143543e-01 -6.69695795e-01 -1.09870338e+00
-1.33345795e+00 -1.22346222e+00 1.64095283e+00 2.72737175e-01
-3.37404668e-01 -1.20434213e+00 1.31380647e-01 6.09820366e-01
4.08459187e-01 -2.79304117e-01 7.60348439e-01 -9.13944364e-01
-2.16320410e-01 -2.12483913e-01 2.35814359e-02 6.43357694e-01
3.69628131e-01 -1.29848957e-01 -8.77303958e-01 -1.37066275e-01
2.81619728e-01 -3.20548832e-01 9.72546577e-01 2.69597858e-01
5.21692693e-01 -4.99663204e-01 -2.10983083e-01 1.52529821e-01
1.00302303e+00 3.03377390e-01 4.57110882e-01 5.35739243e-01
9.04786646e-01 7.34388351e-01 1.17010939e+00 5.68089075e-02
7.93023109e-01 3.71525079e-01 2.00533256e-01 -2.70651191e-01
1.26111597e-01 1.61247954e-01 1.14976060e+00 1.80221856e+00
3.33669662e-01 -5.99082053e-01 -5.07287025e-01 1.12051988e+00
-1.72267318e+00 -4.23664570e-01 -4.14444476e-01 1.54689872e+00
1.00792551e+00 2.98759192e-01 -1.38969332e-01 -1.46314930e-02
4.21763182e-01 7.65977740e-01 -6.23085439e-01 -8.28847051e-01
-1.60657912e-01 3.71105462e-01 -1.23800449e-01 6.18397653e-01
-1.36454570e+00 1.02410436e+00 4.87742853e+00 1.26382315e+00
-8.76324356e-01 2.37173900e-01 7.08303034e-01 3.10162216e-01
-8.60680223e-01 1.22268815e-02 -8.75236332e-01 9.12698880e-02
9.00220811e-01 -3.39362949e-01 -1.81580991e-01 1.06425154e+00
9.46960300e-02 1.23893723e-01 -7.49619067e-01 4.99945879e-01
5.14842331e-01 -7.67755508e-01 4.76743460e-01 -1.65739968e-01
9.63031352e-01 -1.10669672e-01 6.70017824e-02 5.64129353e-01
1.81480795e-01 -6.07123196e-01 5.34106672e-01 1.61878899e-01
5.26731551e-01 -9.32727218e-01 1.40476501e+00 5.36767431e-02
-1.57084286e+00 3.11222017e-01 -8.91415030e-02 -5.09291096e-03
6.02142215e-01 1.08002079e+00 -1.99792862e-01 9.31267083e-01
5.95924258e-01 1.16561592e+00 -3.58047277e-01 6.15206599e-01
-8.48636448e-01 8.05159688e-01 1.07600160e-01 -5.43805122e-01
4.84683990e-01 -4.25155431e-01 6.58018231e-01 1.22775292e+00
2.55389661e-01 -7.42191821e-02 -5.99852018e-02 2.69507825e-01
-1.14680335e-01 5.83407819e-01 -4.46393520e-01 -3.15682471e-01
2.08316147e-02 1.76853645e+00 -5.58113754e-01 -7.34174192e-01
-7.38818705e-01 1.13501799e+00 1.68585733e-01 3.77358258e-01
-4.89837438e-01 -7.18677819e-01 8.93059731e-01 -4.58835125e-01
7.91809142e-01 2.02241287e-01 -1.63387269e-01 -1.51418984e+00
2.77914196e-01 -1.38030446e+00 2.09057599e-01 -8.39028239e-01
-1.66422105e+00 1.23594224e+00 -4.42340076e-01 -1.43295729e+00
-1.60819381e-01 -7.63701022e-01 -1.00823247e+00 9.82398152e-01
-1.83063996e+00 -1.71989095e+00 3.47698927e-01 2.37090930e-01
1.17037416e+00 1.07002079e-01 7.33664811e-01 -5.12996055e-02
-6.91594303e-01 6.07236445e-01 -3.88162881e-01 1.52323330e-02
6.52539551e-01 -1.31545925e+00 8.98312688e-01 1.05497658e+00
1.12610318e-01 7.44759262e-01 5.13708353e-01 -6.15976989e-01
-1.36210155e+00 -1.31329846e+00 1.69369996e+00 -6.32262230e-01
9.95096087e-01 -4.17051136e-01 -6.93274081e-01 7.30280697e-01
8.39362264e-01 -5.72885334e-01 8.97971451e-01 4.73705858e-01
-4.10583436e-01 -2.01271370e-01 -4.98959750e-01 8.12109053e-01
6.23910367e-01 -6.58114016e-01 -7.02342927e-01 4.60141033e-01
1.35766304e+00 -1.99823171e-01 -7.24209368e-01 7.79880941e-01
5.13535798e-01 -7.81069338e-01 5.28140008e-01 -5.40878057e-01
6.34810567e-01 -4.37396199e-01 1.28917143e-01 -1.61522973e+00
-1.03475945e-02 -5.82180798e-01 -2.04155415e-01 1.65195405e+00
1.03982520e+00 -5.81316650e-01 1.77609757e-01 3.16233963e-01
-2.28238255e-01 -1.34555817e+00 -4.60098743e-01 -4.34119225e-01
6.58916589e-03 -5.77574372e-01 5.42953670e-01 5.10975659e-01
1.10942520e-01 1.36418998e+00 -3.43602926e-01 2.32109651e-01
2.89423078e-01 6.57463193e-01 3.82160932e-01 -4.80029613e-01
-2.45776027e-01 -4.42558765e-01 1.57461733e-01 -1.23260176e+00
2.88453728e-01 -5.12434244e-01 3.78333300e-01 -1.88859665e+00
3.96555275e-01 3.44653904e-01 -4.00074989e-01 3.81609887e-01
-9.95806277e-01 2.25412816e-01 -4.69153887e-03 -1.57692537e-01
-1.02161574e+00 1.09470046e+00 1.35907531e+00 -4.70576525e-01
-8.67254287e-02 4.04269964e-01 -1.11298931e+00 8.26022148e-01
6.86326027e-01 -4.19462889e-01 -3.70742798e-01 -4.20467228e-01
6.01936638e-01 -3.96399766e-01 -5.70908427e-01 -1.74002334e-01
6.74335882e-02 1.04186222e-01 -2.66224772e-01 -1.04429460e+00
1.19314656e-01 -4.41964358e-01 -8.31249893e-01 1.71180382e-01
-1.45554066e-01 4.60226059e-01 1.10842653e-01 3.68340701e-01
-6.86897874e-01 -2.60765135e-01 7.94530436e-02 -1.39163047e-01
-6.26771390e-01 3.40333730e-01 -8.17452431e-01 9.53319073e-02
6.12002790e-01 3.80183995e-01 -3.97913724e-01 -5.51463783e-01
-2.88226128e-01 5.26309609e-01 5.53874187e-02 5.86647868e-01
7.39752233e-01 -1.19764102e+00 -7.01413691e-01 -5.35700507e-02
5.73929667e-01 7.34845921e-02 9.17425975e-02 9.06618953e-01
1.78212389e-01 5.23772120e-01 6.28094077e-01 -1.13848343e-01
-1.55657852e+00 6.48336411e-01 -4.67219949e-02 -9.78515208e-01
1.91528484e-01 1.10543191e+00 5.03182232e-01 -7.66225278e-01
-1.18524753e-01 -3.67731899e-01 -7.33960271e-01 3.07946354e-01
7.04065382e-01 -1.62193492e-01 2.41312578e-01 -7.34316766e-01
-5.67549050e-01 8.03639829e-01 -5.61444163e-01 -2.21912846e-01
1.17461812e+00 -3.17747653e-01 -8.21985781e-01 5.92957973e-01
1.13289428e+00 2.99867958e-01 -3.94653529e-01 -3.52397501e-01
7.01792166e-02 1.11509636e-01 3.98457870e-02 -7.89042771e-01
-1.10514915e+00 7.86406457e-01 -4.72146682e-02 1.89292908e-01
1.42148077e+00 1.21357106e-01 1.20181167e+00 4.54919070e-01
1.59676388e-01 -7.85594642e-01 1.26910269e-01 9.97873425e-01
1.07511640e+00 -1.47705269e+00 -1.43842056e-01 -6.19582295e-01
-9.95032728e-01 6.53663039e-01 6.78568304e-01 2.00160235e-01
9.27624404e-01 1.06290862e-01 7.37724185e-01 -4.08168346e-01
-1.18514037e+00 -7.77178764e-01 7.31510699e-01 3.74807298e-01
8.04752052e-01 -1.18930280e-01 -7.64250100e-01 9.42613184e-01
-1.56360015e-01 -4.61709708e-01 1.14587359e-01 1.12758219e+00
-2.57792175e-01 -1.50960422e+00 5.04072495e-02 5.73803544e-01
-7.11550832e-01 -9.49403465e-01 -6.91067576e-01 2.85980076e-01
-1.24420986e-01 1.48366618e+00 -3.43870372e-01 -5.70486367e-01
4.85303253e-01 1.48477867e-01 -1.21325560e-01 -8.54504108e-01
-1.01509619e+00 3.17415863e-01 5.80574036e-01 -4.87075478e-01
-8.70383739e-01 -4.95845109e-01 -8.51190865e-01 1.78746060e-01
-7.60253370e-01 4.73670095e-01 5.28198123e-01 1.44808745e+00
4.33170646e-01 6.64966166e-01 9.34644222e-01 -6.24114931e-01
-8.53402987e-02 -1.32871306e+00 -4.15686816e-01 1.79071695e-01
4.21851605e-01 -8.76357313e-03 -3.39660376e-01 -2.63249706e-02] | [11.47574520111084, 6.632970809936523] |
4b0b917c-b97f-4526-8d1a-079b337f8f4f | localize-to-binauralize-audio-spatialization | null | null | http://openaccess.thecvf.com//content/ICCV2021/html/Rachavarapu_Localize_to_Binauralize_Audio_Spatialization_From_Visual_Sound_Source_Localization_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Rachavarapu_Localize_to_Binauralize_Audio_Spatialization_From_Visual_Sound_Source_Localization_ICCV_2021_paper.pdf | Localize to Binauralize: Audio Spatialization From Visual Sound Source Localization | Videos with binaural audios provide an immersive viewing experience by enabling 3D sound sensation. Recent works attempt to generate binaural audio in a multimodal learning framework using large quantities of videos with accompanying binaural audio. In contrast, we attempt a more challenging problem -- synthesizing binaural audios for a video with monaural audio in a weakly supervised setting and weakly semi-supervised setting. Our key idea is that any down-stream task that can be solved only using binaural audios can be used to provide proxy supervision for binaural audio generation, thereby reducing the reliance on explicit supervision. In this work, as a proxy-task for weak supervision, we use Sound Source Localization with only audio. We design a two-stage architecture called Localize-to-Binauralize Network (L2BNet). The first stage of L2BNet is a Stereo Generation (SG) network employed to generate two-stream audio from monaural audio using visual frame information as guidance. In the second stage, an Audio Localization (AL) network is designed to use the synthesized two-stream audio to localize sound sources in visual frames. The entire network is trained end-to-end so that the AL network provides necessary supervision for the SG network. We experimentally show that our weakly-supervised framework generates two-stream audio containing binaural cues. Through user study, we further validate that our proposed approach generates binaural-quality audio using as little as 10% of explicit binaural supervision data for the SG network. | ['A. N. Rajagopalan', 'Vignesh Sundaresha', 'Aakanksha', 'Kranthi Kumar Rachavarapu'] | 2021-01-01 | null | null | null | iccv-2021-1 | ['audio-generation'] | ['audio'] | [ 3.61857951e-01 6.35288060e-02 2.48784155e-01 -8.96402225e-02
-1.30220568e+00 -6.11708999e-01 3.20595920e-01 -2.84888476e-01
-5.34132458e-02 5.48976362e-01 4.11416411e-01 -1.16108455e-01
3.73921126e-01 -5.36715567e-01 -1.23954439e+00 -6.39304698e-01
2.12766640e-02 -1.79835707e-01 2.67224580e-01 -2.08847091e-01
-1.02584109e-01 4.30009961e-02 -2.25107193e+00 7.05023050e-01
5.44235826e-01 1.25221431e+00 4.07555670e-01 1.34938991e+00
3.90636832e-01 7.02391863e-01 -8.54827225e-01 -7.56134093e-02
5.46891332e-01 -8.03267062e-01 -3.92153025e-01 9.36847106e-02
1.05028212e+00 -4.52089518e-01 -1.87250316e-01 9.09812450e-01
9.80457664e-01 2.41789162e-01 3.63969058e-01 -1.36146879e+00
-1.45134822e-01 6.51063740e-01 -1.35174155e-01 -8.69763196e-02
8.84691775e-01 5.27770042e-01 1.10963833e+00 -9.50307250e-01
3.11369270e-01 1.26489544e+00 3.76490295e-01 6.11334801e-01
-1.06046629e+00 -8.19459558e-01 -5.25507778e-02 9.81551111e-02
-1.34213972e+00 -7.07145095e-01 9.51910496e-01 -2.96054035e-01
5.95030785e-01 4.08391774e-01 8.32645416e-01 1.11519825e+00
-1.10574022e-01 7.70983398e-01 9.31983113e-01 -4.28876311e-01
3.90161395e-01 8.16786289e-02 -6.31774783e-01 3.71558458e-01
-4.36812490e-01 3.31036866e-01 -1.05823517e+00 1.01712540e-01
8.92793953e-01 -2.73321837e-01 -6.82404041e-01 -2.28767365e-01
-9.68930244e-01 5.65396369e-01 5.17577708e-01 5.07261157e-02
-1.79807544e-01 3.57223213e-01 2.42289200e-01 4.28110152e-01
3.25738400e-01 2.68101573e-01 -1.39414653e-01 -3.31264526e-01
-1.22898221e+00 1.12887628e-01 6.65525317e-01 7.81904280e-01
6.03460789e-01 5.31772792e-01 -7.56909847e-02 1.01470482e+00
3.61539602e-01 5.13944268e-01 4.36352879e-01 -1.24179733e+00
4.96640772e-01 -2.67512918e-01 -6.02318556e-04 -5.18805981e-01
9.64563563e-02 -5.43857276e-01 -4.53727424e-01 5.31694472e-01
2.75324613e-01 -4.04483050e-01 -9.34869826e-01 1.86678874e+00
4.45537984e-01 6.15461111e-01 -4.31648865e-02 1.45691800e+00
8.07852805e-01 7.92657852e-01 -2.86814660e-01 -2.49983117e-01
1.02987599e+00 -1.11898983e+00 -5.64104557e-01 5.33565432e-02
1.93233505e-01 -1.05083692e+00 1.41695118e+00 4.77501422e-01
-1.47589183e+00 -9.98087227e-01 -9.06523228e-01 -1.53573332e-04
3.83812375e-02 -8.25442523e-02 1.38711587e-01 4.10894215e-01
-1.30753875e+00 2.68389046e-01 -5.34640491e-01 -3.28776948e-02
-1.61910581e-03 -1.96888261e-02 -2.05449596e-01 7.51086473e-02
-1.18003201e+00 2.37001374e-01 2.22975444e-02 -1.20557590e-04
-1.72047138e+00 -9.83585000e-01 -1.24001336e+00 8.05754066e-02
3.58517051e-01 -7.05904245e-01 1.57796586e+00 -1.39361417e+00
-1.75387561e+00 4.72737849e-01 -2.17289984e-01 -5.01227140e-01
3.73147219e-01 -3.16782653e-01 -3.16447914e-01 6.87702656e-01
1.65174618e-01 9.89787400e-01 1.27498806e+00 -1.60183096e+00
-8.16076279e-01 2.09967852e-01 3.76672626e-01 5.12630522e-01
-6.65238127e-02 -6.90219030e-02 -3.12713027e-01 -8.69039714e-01
-1.16450034e-01 -7.35949457e-01 1.29239485e-01 3.02850194e-02
-3.45762312e-01 1.15249105e-01 7.78929830e-01 -5.21892130e-01
1.02407134e+00 -2.27373195e+00 6.83819503e-02 1.09819928e-02
6.43102378e-02 4.76698615e-02 -3.39531243e-01 1.13874055e-01
-3.96956533e-01 -3.10954571e-01 -8.78943428e-02 -6.93834245e-01
1.50466366e-02 -1.99808240e-01 -6.73978627e-01 1.39067154e-02
1.00549407e-01 5.28096497e-01 -1.16049910e+00 -4.14408296e-01
2.33749583e-01 6.87014759e-01 -9.74046648e-01 7.11630106e-01
-1.56558231e-01 3.78295839e-01 2.40803495e-01 9.29217994e-01
4.34495896e-01 1.04199089e-01 -2.52209455e-01 -3.31006944e-01
-1.40501052e-01 2.94566035e-01 -1.36952615e+00 1.86472809e+00
-7.51150668e-01 6.50943756e-01 5.24263322e-01 -5.78440487e-01
6.79464400e-01 6.52891338e-01 1.69299036e-01 -5.62503815e-01
-4.37635817e-02 3.04922134e-01 -1.05869338e-01 -4.60983127e-01
3.57670307e-01 -3.91128629e-01 -5.71523979e-02 2.96424180e-01
4.49459553e-01 -4.51736629e-01 6.22820891e-02 2.38712758e-01
8.92460644e-01 2.93654978e-01 -6.20678738e-02 4.04918998e-01
4.82213467e-01 -5.98632276e-01 3.30784291e-01 7.38996863e-01
-1.04868479e-01 1.16279507e+00 1.05053619e-01 2.56520569e-01
-8.43711376e-01 -1.43941653e+00 6.04941398e-02 1.40941143e+00
-6.25773370e-02 -7.18883634e-01 -1.00304389e+00 -5.81041217e-01
-4.10591722e-01 4.06588465e-01 -1.61748260e-01 -6.19181581e-02
-2.63851821e-01 2.26086881e-02 5.67466259e-01 4.69459832e-01
3.28277320e-01 -9.53338444e-01 -4.88287836e-01 7.74003714e-02
-3.83736163e-01 -9.30940509e-01 -6.85652852e-01 1.79571167e-01
-3.81908983e-01 -8.39285076e-01 -1.15775812e+00 -7.45497644e-01
3.05242509e-01 3.04280370e-01 8.70518386e-01 -4.60370690e-01
-1.70087144e-01 8.14998627e-01 -3.53772312e-01 -2.87897140e-01
-3.91294330e-01 -3.27645391e-01 3.46444339e-01 3.82429153e-01
-2.22504333e-01 -9.75917697e-01 -8.23394120e-01 3.88199270e-01
-9.55033541e-01 9.82494429e-02 4.13433015e-01 6.58262908e-01
6.56049669e-01 -7.40870088e-02 7.09789574e-01 -6.39374629e-02
3.79410088e-01 -4.29288000e-01 -4.23462391e-01 -3.63211691e-01
1.53798997e-01 -5.39182246e-01 1.02328289e+00 -6.58275187e-01
-9.76694047e-01 2.47081175e-01 -4.05099005e-01 -9.09776270e-01
-4.77215767e-01 2.90792912e-01 -2.88677007e-01 -2.05822475e-02
8.01336110e-01 2.28797317e-01 -2.83997774e-01 -4.13141698e-01
3.72085273e-01 1.02410090e+00 7.89083898e-01 -4.04366285e-01
7.09884822e-01 4.18745518e-01 -1.23030469e-01 -8.73529196e-01
-1.03597414e+00 -3.44379038e-01 -8.12343955e-02 -7.44261861e-01
7.70587325e-01 -1.36414242e+00 -6.68981552e-01 2.83652484e-01
-8.58170927e-01 -5.59028625e-01 -3.89761627e-01 6.29976511e-01
-8.85942996e-01 1.37787491e-01 -5.40341198e-01 -1.06627178e+00
9.13609564e-02 -1.22120142e+00 1.35042500e+00 2.67609328e-01
-1.22417770e-01 -5.82857311e-01 1.36653259e-01 6.71586931e-01
1.74788266e-01 9.99284759e-02 2.76717037e-01 -3.18258405e-01
-6.74189925e-01 4.62747142e-02 2.17428327e-01 6.92441523e-01
1.29895359e-01 -3.63828927e-01 -1.57453656e+00 -2.62439847e-01
-6.27532080e-02 -7.74676502e-01 6.72145724e-01 4.64439183e-01
1.21001256e+00 -3.00032228e-01 3.37140977e-01 7.21964359e-01
9.74837840e-01 1.89041600e-01 4.04628545e-01 -1.48112446e-01
8.84456158e-01 5.06700218e-01 8.66643071e-01 2.85798877e-01
2.28932425e-01 7.78507054e-01 6.87250555e-01 -3.26855004e-01
-6.33898973e-01 -6.91271007e-01 8.46149206e-01 8.53186131e-01
-1.45467473e-02 -8.75630453e-02 -4.65865850e-01 6.60339057e-01
-1.60832953e+00 -9.11625564e-01 2.20268294e-01 2.36052370e+00
1.32077420e+00 1.11579850e-01 3.44556600e-01 3.85949731e-01
7.54002213e-01 8.39414671e-02 -2.28356525e-01 -2.98148692e-01
-1.01503856e-01 4.44551080e-01 -2.13817626e-01 7.64363170e-01
-9.76442516e-01 8.42360377e-01 5.48398352e+00 9.67068374e-01
-1.65857220e+00 3.71276326e-02 3.26422662e-01 -7.64050901e-01
-2.46271163e-01 -1.04931988e-01 -6.39390826e-01 6.32139683e-01
9.82683837e-01 1.63809821e-01 6.01505756e-01 8.60468447e-01
6.47360802e-01 -1.29396901e-01 -1.36288536e+00 1.17967200e+00
3.41654271e-01 -1.12083626e+00 3.00394565e-01 -1.55005649e-01
6.04762852e-01 -2.56601542e-01 3.08431655e-01 2.16708958e-01
1.41085293e-02 -8.92122746e-01 1.10636616e+00 3.84432942e-01
1.15039814e+00 -8.09332490e-01 3.67890447e-01 1.76001549e-01
-1.40542150e+00 -4.05522361e-02 -2.88748257e-02 -9.63746682e-02
5.43941081e-01 5.31633139e-01 -8.93286347e-01 4.20780420e-01
9.86320496e-01 4.84664381e-01 -2.02096894e-01 1.38839996e+00
-4.02411371e-01 1.05267394e+00 -4.56028759e-01 4.70950723e-01
1.63836658e-01 2.46461689e-01 9.72224116e-01 1.24238288e+00
4.82271880e-01 -1.38095170e-01 2.16753826e-01 5.71149170e-01
-8.36939439e-02 1.15649141e-02 -7.04213321e-01 2.57917970e-01
3.58352065e-01 1.17420959e+00 -2.11858407e-01 -2.94941038e-01
-1.76175579e-01 9.44216430e-01 -2.39251122e-01 5.16504824e-01
-1.02798975e+00 -5.64846992e-01 6.05129600e-01 2.94324666e-01
3.94878507e-01 1.05696313e-01 1.77014992e-01 -9.76936758e-01
1.28717244e-01 -1.10610914e+00 3.40068609e-01 -1.55894983e+00
-1.02579927e+00 6.94035232e-01 -2.20987737e-01 -1.69673026e+00
-6.15268648e-01 -3.17761838e-01 -6.58829451e-01 7.81075120e-01
-1.69747174e+00 -1.16625667e+00 -3.69533151e-01 1.01437306e+00
7.77357340e-01 -4.84857783e-02 5.95502138e-01 4.50539619e-01
-1.31254733e-01 7.82988429e-01 -3.37504894e-01 -6.37503266e-02
1.19693601e+00 -1.28767884e+00 -3.41221333e-01 8.37779343e-01
2.75818229e-01 3.91290694e-01 8.88935149e-01 -3.71442765e-01
-1.00433087e+00 -1.31560135e+00 6.29598916e-01 -1.55669376e-01
5.85629106e-01 -7.37282872e-01 -6.23849928e-01 4.28594023e-01
3.00987542e-01 1.53504804e-01 9.15247440e-01 -3.26898426e-01
-3.20586622e-01 -4.45542008e-01 -7.77646363e-01 6.78856611e-01
8.72377574e-01 -8.70804548e-01 -5.95435977e-01 1.49567008e-01
1.09451902e+00 -5.59181690e-01 -6.49870276e-01 2.82357007e-01
3.30887645e-01 -1.01067317e+00 9.42141056e-01 -1.33855045e-01
8.04765463e-01 -8.21979940e-01 -2.45984450e-01 -1.58601463e+00
2.61399031e-01 -1.16587496e+00 -2.98458517e-01 1.29079115e+00
2.58843362e-01 -1.47469833e-01 4.14730728e-01 -1.80256173e-01
-5.57177901e-01 -3.91722649e-01 -8.69392633e-01 -6.36684179e-01
-4.13895845e-01 -8.53019118e-01 2.74865270e-01 7.87476003e-01
1.15556173e-01 3.84885669e-01 -5.26795089e-01 2.88676351e-01
7.48343945e-01 1.57964513e-01 9.47558582e-01 -7.30598867e-01
-7.66138136e-01 1.84900314e-02 -2.91578919e-01 -1.41882885e+00
1.02282137e-01 -7.34978616e-01 4.83832866e-01 -1.07831824e+00
-3.43899995e-01 -2.83524748e-02 -2.88103789e-01 3.38646680e-01
-2.83614220e-03 8.23187113e-01 3.35206956e-01 -1.32264823e-01
-6.30488217e-01 7.32121825e-01 1.33795011e+00 1.35792062e-01
-3.45408291e-01 1.43330887e-01 -5.69361985e-01 8.84017110e-01
4.10273045e-01 -2.70439625e-01 -5.86129665e-01 -1.99802414e-01
1.19005546e-01 4.05750275e-01 8.45176399e-01 -1.18519866e+00
3.68470341e-01 1.67897344e-01 2.85341352e-01 -5.40783346e-01
7.82133162e-01 -6.19573593e-01 -1.83724865e-01 -1.73846945e-01
-4.70000356e-01 -4.93710220e-01 1.83348671e-01 5.19261897e-01
-6.56698465e-01 -5.99211454e-02 6.47863805e-01 -1.27656888e-02
-3.34908783e-01 1.00009166e-01 -5.06878614e-01 5.19731827e-02
6.98641717e-01 -3.44361871e-01 -1.03040915e-02 -9.99871790e-01
-1.17233956e+00 4.34587039e-02 1.92870036e-01 3.47857326e-01
7.60793567e-01 -1.49778795e+00 -4.85384673e-01 2.51263291e-01
-6.61526471e-02 1.73694074e-01 4.48286057e-01 6.41210139e-01
-1.46506205e-01 1.83805719e-01 -1.26022175e-01 -1.09051824e+00
-1.33707595e+00 3.07255775e-01 3.73211145e-01 4.04013813e-01
-3.27866703e-01 1.04823291e+00 6.43630862e-01 -1.14690706e-01
6.30359590e-01 -1.81780130e-01 -6.03681691e-02 3.36250626e-02
5.54426312e-01 1.67044953e-01 -1.06472410e-01 -4.76305008e-01
4.21594828e-02 4.27735627e-01 3.65653694e-01 -6.82586730e-01
1.25191832e+00 -2.29998484e-01 2.55373240e-01 7.41992831e-01
1.20507646e+00 3.58788937e-01 -1.76296389e+00 1.72075685e-02
-7.81246543e-01 -5.45434177e-01 2.68678278e-01 -8.77290845e-01
-1.02334404e+00 1.12276256e+00 6.23325944e-01 2.59251624e-01
1.52905345e+00 1.28926756e-02 8.75757396e-01 4.00623530e-02
1.29042044e-01 -1.01179469e+00 5.64622521e-01 3.29051852e-01
9.72422838e-01 -9.29620326e-01 -8.51904154e-01 -3.83261323e-01
-6.34455740e-01 9.54035103e-01 8.26269090e-01 1.34836780e-02
5.15733480e-01 4.06689018e-01 2.55451024e-01 1.05717771e-01
-7.48464048e-01 -5.17185986e-01 4.49850827e-01 7.05761254e-01
3.56324017e-01 -3.03880423e-01 4.52439249e-01 7.40821004e-01
-7.47687519e-01 -1.34963050e-01 4.90304410e-01 7.98472822e-01
-4.24842566e-01 -6.69202924e-01 -6.17170930e-01 -1.79800436e-01
-4.17642325e-01 -4.12451178e-01 -3.82340640e-01 2.28687927e-01
3.71889740e-01 1.10809112e+00 1.75674483e-01 -3.65689903e-01
3.93129021e-01 1.51217595e-01 4.83879596e-01 -6.87351465e-01
-6.75402284e-01 8.55696738e-01 6.53650844e-03 -7.69008338e-01
-3.06041390e-01 -4.91795063e-01 -1.05670798e+00 2.68959314e-01
-1.49469137e-01 2.65473932e-01 5.37543356e-01 7.05842376e-01
2.39090681e-01 8.49920869e-01 9.47014093e-01 -1.29828036e+00
-1.64936170e-01 -8.34849417e-01 -5.03823876e-01 2.86601186e-01
8.12473238e-01 -4.29289997e-01 -8.46101642e-01 5.10115504e-01] | [14.980798721313477, 5.0881667137146] |
f9be64ff-be04-41b3-98c2-1d38113c34c6 | quatde-dynamic-quaternion-embedding-for | 2105.09002 | null | https://arxiv.org/abs/2105.09002v2 | https://arxiv.org/pdf/2105.09002v2.pdf | QuatDE: Dynamic Quaternion Embedding for Knowledge Graph Completion | Knowledge graph embedding has been an active research topic for knowledge base completion (KGC), with progressive improvement from the initial TransE, TransH, RotatE et al to the current state-of-the-art QuatE. However, QuatE ignores the multi-faceted nature of the entity and the complexity of the relation, only using rigorous operation on quaternion space to capture the interaction between entitiy pair and relation, leaving opportunities for better knowledge representation which will finally help KGC. In this paper, we propose a novel model, QuatDE, with a dynamic mapping strategy to explicitly capture the variety of relational patterns and separate different semantic information of the entity, using transition vectors to adjust the point position of the entity embedding vectors in the quaternion space via Hamilton product, enhancing the feature interaction capability between elements of the triplet. Experiment results show QuatDE achieves state-of-the-art performance on three well-established knowledge graph completion benchmarks. In particular, the MR evaluation has relatively increased by 26% on WN18 and 15% on WN18RR, which proves the generalization of QuatDE. | ['Ke Qin', 'Jim Wilson Owusu', 'Rufai Yusuf Zakari', 'Yuxue Yang', 'Kun Yang', 'Haipeng Gao'] | 2021-05-19 | null | null | null | null | ['knowledge-base-completion', 'knowledge-base-completion'] | ['graphs', 'knowledge-base'] | [-3.42417806e-01 1.32745862e-01 -3.62106115e-01 1.28158748e-01
-7.76252747e-02 -5.26372135e-01 4.44611400e-01 1.77068144e-01
-4.31905836e-01 5.55146277e-01 3.38814139e-01 6.12661541e-02
-5.27472079e-01 -1.03845310e+00 -6.01379156e-01 -5.09841740e-01
-2.84732580e-01 5.88445008e-01 1.61427513e-01 -7.36368835e-01
-1.23224713e-01 2.82188892e-01 -1.03636706e+00 -1.33544266e-01
8.86648357e-01 7.84366548e-01 -1.82158113e-01 1.78166702e-02
-3.23102586e-02 6.90145373e-01 -3.73392940e-01 -1.00609708e+00
1.63296580e-01 -1.37160078e-01 -9.32284296e-01 -8.98419097e-02
2.10099801e-01 1.07156351e-01 -7.86318123e-01 1.03544962e+00
3.84631366e-01 -1.54497223e-02 5.97872972e-01 -1.36982858e+00
-1.03739488e+00 6.77824318e-01 -4.89208549e-01 -1.62752569e-02
4.93031025e-01 -2.72395998e-01 1.45953166e+00 -1.21451795e+00
9.77910101e-01 1.18756938e+00 5.78701138e-01 6.13827966e-02
-1.01902199e+00 -6.23691082e-01 6.05196133e-02 9.92781997e-01
-1.91658056e+00 7.01389387e-02 7.30413318e-01 -2.94539660e-01
8.83924544e-01 1.43373460e-01 8.95809650e-01 3.89537185e-01
-5.36716953e-02 6.29644990e-01 8.02003741e-01 -2.99133301e-01
2.57639810e-02 1.73336137e-02 3.94563913e-01 1.05170739e+00
6.64610028e-01 -2.41416410e-01 -7.00761974e-01 1.20766461e-01
7.19857931e-01 -1.84209514e-02 -5.43299198e-01 -8.04957926e-01
-1.52180719e+00 7.59921610e-01 7.92518198e-01 1.56027898e-01
-3.73970568e-01 5.42645305e-02 2.51378596e-01 3.36531252e-01
-5.83489612e-02 4.96574998e-01 -2.75041401e-01 -1.37039721e-01
-3.55106622e-01 8.67791623e-02 9.40861404e-01 1.24531317e+00
9.24536586e-01 -3.57777439e-02 -2.39284411e-02 4.48094487e-01
1.26098335e-01 4.77061510e-01 2.22813115e-01 -5.16543686e-01
7.01199174e-01 1.11635029e+00 -8.84382650e-02 -1.49927342e+00
-4.27033186e-01 -6.57676697e-01 -9.99367833e-01 -5.24811387e-01
-1.64992083e-02 7.56480768e-02 -5.92926562e-01 1.52620792e+00
3.59168410e-01 2.42647365e-01 2.70107985e-01 7.87018538e-01
9.30045724e-01 5.50292850e-01 -2.93567032e-01 -7.68327191e-02
1.58082235e+00 -8.60751867e-01 -9.12271857e-01 1.90968126e-01
8.07098925e-01 -5.07929325e-01 7.48087227e-01 5.18783852e-02
-7.59192884e-01 -4.07960922e-01 -1.37657142e+00 -1.18619293e-01
-7.17729330e-01 1.15503639e-01 1.07700825e+00 6.13904297e-01
-7.34458506e-01 3.83914202e-01 -7.17218339e-01 -1.07341655e-01
1.16203986e-01 3.76356363e-01 -7.50635445e-01 -4.58699346e-01
-1.79307151e+00 1.03664088e+00 7.82446861e-01 3.70838076e-01
-2.27548584e-01 -8.79685104e-01 -1.00736201e+00 1.49181008e-01
8.21735382e-01 -7.39255786e-01 4.20612872e-01 -2.73368396e-02
-1.29855478e+00 2.94034451e-01 2.70466417e-01 -2.27092981e-01
2.29266077e-01 -3.03358585e-01 -8.16983163e-01 2.48853385e-01
1.46321394e-02 4.59681213e-01 4.83259767e-01 -1.04615188e+00
-3.28315914e-01 -3.35408926e-01 3.88696223e-01 4.98102188e-01
-5.36220312e-01 -4.46090043e-01 -9.42801178e-01 -5.65362751e-01
3.79393995e-01 -1.06167150e+00 2.65728474e-01 -1.66727036e-01
-3.11680228e-01 -3.97180319e-01 7.39513993e-01 -6.34393930e-01
1.52211666e+00 -2.22855020e+00 6.20034277e-01 3.91122460e-01
2.74678141e-01 2.44064003e-01 -8.55689123e-03 9.63469923e-01
-1.97005525e-01 1.21243941e-02 -5.92988543e-02 -6.99428516e-03
2.19473198e-01 4.30842251e-01 -2.04792202e-01 4.67277318e-01
1.76244169e-01 1.16832733e+00 -9.91588533e-01 -4.70142394e-01
-8.35664943e-03 5.23739219e-01 -4.95582759e-01 -1.62921667e-01
1.75331980e-01 -2.14463294e-01 -4.48150992e-01 5.07311642e-01
6.83008075e-01 -4.39775705e-01 4.64316338e-01 -8.58394682e-01
1.58145890e-01 4.34595570e-02 -1.70834267e+00 1.68361545e+00
-2.40252733e-01 1.09624729e-01 -3.85164827e-01 -7.41335034e-01
7.70078421e-01 1.32418722e-01 4.63361561e-01 -6.28379881e-01
-1.38501331e-01 1.82387866e-02 5.44206873e-02 -8.74358863e-02
8.99837911e-01 1.10789575e-01 -9.98945534e-02 1.97125990e-02
1.07286334e-01 -7.62201101e-02 4.08763051e-01 6.80212796e-01
8.50333154e-01 1.36795864e-01 2.14893058e-01 -2.99445540e-01
6.12212718e-01 -1.15392506e-01 6.33463562e-01 1.57830521e-01
1.51847839e-01 2.67061114e-01 6.49801552e-01 -5.71818836e-02
-5.36272526e-01 -1.05303776e+00 -7.98785612e-02 4.81743991e-01
6.50597572e-01 -1.00623834e+00 -3.66523862e-01 -6.66842401e-01
1.41074643e-01 5.29298842e-01 -6.28778577e-01 -4.77985144e-01
-5.15753925e-01 -8.38418245e-01 5.53378522e-01 4.03095275e-01
8.49648774e-01 -5.73889256e-01 6.28843009e-02 1.19069941e-01
-3.54820728e-01 -1.33273029e+00 -4.88443404e-01 -1.60321191e-01
-5.51440120e-01 -1.29713762e+00 -4.05014634e-01 -7.54280210e-01
6.34097099e-01 2.69075781e-01 8.93037081e-01 -4.66844663e-02
-1.21056691e-01 5.74715555e-01 -5.48032045e-01 -2.94926018e-02
2.33315244e-01 1.92999929e-01 1.58225015e-01 1.38600349e-01
5.52134365e-02 -4.81492013e-01 -6.37711585e-01 4.38011825e-01
-1.04791176e+00 1.38216063e-01 6.25542343e-01 1.01328957e+00
6.84939682e-01 4.59378839e-01 5.76373458e-01 -7.36049294e-01
4.23465759e-01 -3.97931516e-01 -3.32244217e-01 6.51044250e-01
-8.93024385e-01 2.63817459e-01 4.68010694e-01 -1.74509406e-01
-7.95310974e-01 -3.12743515e-01 2.57586479e-01 -2.88065463e-01
8.21413696e-01 1.01836801e+00 -3.85073692e-01 -2.14437857e-01
2.87829310e-01 3.66793931e-01 -1.21331468e-01 -2.90865779e-01
8.21625710e-01 1.67695701e-01 4.24908280e-01 -6.46992743e-01
1.09576094e+00 5.69343984e-01 4.30382580e-01 -7.48289406e-01
-6.68304205e-01 -4.59840238e-01 -6.12031102e-01 7.37128779e-02
6.11582696e-01 -1.11651456e+00 -9.44435537e-01 3.22295487e-01
-9.33963239e-01 4.08182114e-01 -2.83194959e-01 6.69655621e-01
-1.53698951e-01 7.06068099e-01 -4.66514200e-01 -4.30491678e-02
-4.30429488e-01 -9.65782583e-01 6.88751996e-01 2.23000720e-02
1.76199466e-01 -9.60257590e-01 6.03855914e-03 4.57564712e-01
1.14393242e-01 1.50139824e-01 1.22515309e+00 -4.35209662e-01
-1.01386416e+00 -3.94216597e-01 -4.23657596e-01 2.42631927e-01
1.15861773e-01 -1.35098919e-01 -2.27531388e-01 -4.00832027e-01
-3.72759491e-01 -9.63159502e-02 6.63437426e-01 -4.94945526e-01
5.82804918e-01 -2.32913509e-01 -2.43252590e-01 6.27309561e-01
1.67437828e+00 -1.69792369e-01 8.13585460e-01 2.77700812e-01
1.17977107e+00 2.85471529e-01 6.76118255e-01 1.44412801e-01
1.10353434e+00 9.29889083e-01 3.44358921e-01 2.27083102e-01
-3.00435930e-01 -3.77783686e-01 2.65285105e-01 1.50661576e+00
-4.83793586e-01 7.23895505e-02 -7.64345288e-01 5.27845085e-01
-1.94673693e+00 -6.54349089e-01 -1.24819361e-01 2.00663495e+00
9.08184946e-01 1.50804706e-02 -2.47973397e-01 3.34580913e-02
5.26557744e-01 2.20051944e-01 -2.11418584e-01 2.42295831e-01
-3.19929481e-01 2.92808622e-01 7.47675419e-01 3.82693172e-01
-6.91147268e-01 1.11207533e+00 5.47704124e+00 1.01445460e+00
-8.19212437e-01 -3.25200669e-02 -3.04074317e-01 3.97647083e-01
-4.67271507e-01 2.28828758e-01 -8.34901631e-01 8.49913731e-02
2.62847245e-01 -3.72913837e-01 6.20418549e-01 4.35165942e-01
-5.12082458e-01 5.90346679e-02 -9.81226444e-01 1.17171371e+00
2.42315069e-01 -1.30604649e+00 4.56892073e-01 9.16656703e-02
7.83520699e-01 -2.49884516e-01 -6.75382838e-02 6.53825283e-01
2.50661075e-02 -7.61445045e-01 4.13434267e-01 7.16640711e-01
7.56655574e-01 -8.74563813e-01 7.94607341e-01 -6.98229373e-02
-1.51457298e+00 3.23983818e-01 -4.39279854e-01 1.69200301e-01
-4.98709343e-02 3.40425402e-01 -8.29733729e-01 1.56860805e+00
4.17743653e-01 9.25632358e-01 -8.76600981e-01 6.75361633e-01
-4.32660460e-01 3.85993332e-01 -2.32470155e-01 -8.30492377e-02
1.28999993e-01 -4.69286144e-01 4.58278000e-01 9.54756081e-01
3.27149197e-03 9.59060267e-02 1.25329554e-01 4.27743554e-01
-2.64893204e-01 3.28775972e-01 -2.56361306e-01 -2.52396256e-01
6.18399620e-01 1.23106861e+00 -4.45610076e-01 -2.49916941e-01
-4.83405411e-01 9.34650362e-01 5.03436148e-01 5.35670996e-01
-9.15335536e-01 -6.76095665e-01 5.27841926e-01 -1.23011842e-01
6.34793341e-01 -5.87381840e-01 -1.63490847e-02 -1.39147198e+00
4.09080744e-01 -7.24597275e-01 4.53229010e-01 -4.94789034e-01
-9.66066241e-01 3.25378180e-01 1.69238091e-01 -1.05135679e+00
7.37164319e-02 -5.21517575e-01 -7.05758259e-02 7.58326888e-01
-1.60620356e+00 -1.38765919e+00 -2.52672195e-01 8.36334586e-01
-2.28443280e-01 -1.29660785e-01 8.88299763e-01 5.55897713e-01
-6.70580864e-01 7.91568160e-01 2.44770050e-01 2.71728098e-01
6.12850249e-01 -1.24664605e+00 2.05410913e-01 7.22500503e-01
3.32108289e-01 9.63796675e-01 3.26502740e-01 -6.48359179e-01
-2.12706447e+00 -9.77470219e-01 8.40290129e-01 -4.02525574e-01
9.72443283e-01 -3.35136682e-01 -8.98156404e-01 6.92479193e-01
-7.60997087e-03 -1.47233512e-02 5.41574001e-01 4.11407292e-01
-6.47159934e-01 -4.03403431e-01 -6.56150699e-01 7.88416266e-01
1.17707622e+00 -6.34695768e-01 -6.94131017e-01 2.52192736e-01
9.12472248e-01 -3.74878943e-01 -1.45171106e+00 6.31855071e-01
4.82639641e-01 -5.43688476e-01 1.15912962e+00 -5.38244307e-01
7.77765587e-02 -7.34390259e-01 -2.87901968e-01 -1.22802389e+00
-4.99290228e-01 -4.48599309e-01 -4.92941201e-01 1.11819935e+00
3.29393834e-01 -6.94321752e-01 7.32691646e-01 3.42878163e-01
3.84945273e-02 -9.41341400e-01 -9.30210352e-01 -9.63778973e-01
-2.37940535e-01 -1.06432088e-01 7.98909783e-01 1.28542638e+00
2.01277569e-01 7.62807190e-01 -4.77529973e-01 2.64292717e-01
5.25509477e-01 2.66904742e-01 7.55707145e-01 -1.01092839e+00
-2.89275527e-01 -1.35328472e-01 -8.87614489e-01 -1.03600669e+00
-3.71655934e-02 -1.17707479e+00 -7.15297878e-01 -1.81089258e+00
7.12672621e-02 -3.41555744e-01 -3.03641826e-01 5.13230026e-01
-3.35893065e-01 1.15654938e-01 4.37099189e-01 3.30999404e-01
-7.72854805e-01 1.11784863e+00 1.58413875e+00 -2.53453821e-01
1.35061331e-02 -5.34184098e-01 -7.61853695e-01 3.41412246e-01
3.82667840e-01 -1.74734682e-01 -6.33390367e-01 -3.30877274e-01
7.10665643e-01 -8.19235519e-02 2.16187358e-01 -8.72047544e-01
4.35562074e-01 1.17103215e-02 -1.99883003e-02 -5.03780961e-01
6.44765615e-01 -9.33620393e-01 3.74163687e-01 3.71947974e-01
3.30851763e-01 2.11225048e-01 -3.03770658e-02 9.21831012e-01
-5.31129122e-01 -2.51596030e-02 1.95745766e-01 2.88866073e-01
-1.07402849e+00 4.74212050e-01 4.20672089e-01 1.40534267e-01
9.02796924e-01 -1.99234188e-01 -4.61992532e-01 -1.63195625e-01
-5.73868275e-01 4.53856349e-01 2.86642104e-01 4.33407068e-01
6.35849476e-01 -1.66590798e+00 -6.87928855e-01 -4.26740712e-03
4.51899350e-01 1.60411615e-02 3.57827395e-01 8.79128695e-01
-5.60360253e-01 5.41782558e-01 -9.35156196e-02 -1.88902304e-01
-8.79473090e-01 6.48494303e-01 -8.04674905e-03 -4.71602648e-01
-6.34024084e-01 6.00962341e-01 2.77644061e-02 -4.76027578e-01
7.60807768e-02 -2.41763070e-01 -3.70475709e-01 2.86775082e-01
-1.19672045e-02 6.94471776e-01 2.01082200e-01 -7.46195495e-01
-4.83479559e-01 5.76591730e-01 -3.18272084e-01 1.75143465e-01
1.08909523e+00 1.06437542e-02 -4.00866926e-01 2.14956179e-01
1.17775333e+00 1.71310425e-01 -6.04651988e-01 -5.23925126e-01
-5.66927157e-02 -2.12400064e-01 -1.49205506e-01 -7.74331093e-01
-1.04016161e+00 4.14256662e-01 2.64701784e-01 -7.14579597e-02
7.92147815e-01 -1.45246327e-01 7.37734854e-01 8.75304580e-01
5.90269685e-01 -1.11920500e+00 9.22495052e-02 6.62779570e-01
1.11319447e+00 -8.87108088e-01 5.37770927e-01 -9.57318306e-01
-8.28861117e-01 8.50476444e-01 4.72933590e-01 -9.73862261e-02
9.51066375e-01 -5.12881875e-01 -3.07469755e-01 -4.75057155e-01
-3.03682834e-01 -2.67987460e-01 5.78816056e-01 3.87319386e-01
8.08880255e-02 3.57830912e-01 -6.77340984e-01 5.13209760e-01
-3.61056477e-01 -2.63073742e-01 3.65490079e-01 7.63186276e-01
-4.11245264e-02 -1.29189849e+00 -5.96835315e-02 1.51280031e-01
-7.52842575e-02 -3.27054769e-01 -3.52411777e-01 1.27716494e+00
1.00603692e-01 6.22941196e-01 -3.59924287e-01 -5.07445037e-01
6.98569059e-01 -1.27361730e-01 6.01039350e-01 -4.89372373e-01
-2.03542337e-01 -3.10178190e-01 2.33472556e-01 -3.39449972e-01
-3.70029479e-01 -4.22810853e-01 -1.43688464e+00 -3.24236900e-01
-4.52175349e-01 3.45920563e-01 5.55907845e-01 7.04921544e-01
6.29019082e-01 5.28151453e-01 5.65040827e-01 -2.63662133e-02
-6.19709790e-01 -7.96878934e-01 -8.88243675e-01 4.72890139e-01
-2.00109452e-01 -1.06116986e+00 -3.95188071e-02 -2.69691885e-01] | [8.659262657165527, 7.791315078735352] |
e6e736a4-d1b1-4160-a35d-1e5f98833fab | detection-recovery-in-online-multi-object | 2205.00968 | null | https://arxiv.org/abs/2205.00968v2 | https://arxiv.org/pdf/2205.00968v2.pdf | Detection Recovery in Online Multi-Object Tracking with Sparse Graph Tracker | In existing joint detection and tracking methods, pairwise relational features are used to match previous tracklets to current detections. However, the features may not be discriminative enough for a tracker to identify a target from a large number of detections. Selecting only high-scored detections for tracking may lead to missed detections whose confidence score is low. Consequently, in the online setting, this results in disconnections of tracklets which cannot be recovered. In this regard, we present Sparse Graph Tracker (SGT), a novel online graph tracker using higher-order relational features which are more discriminative by aggregating the features of neighboring detections and their relations. SGT converts video data into a graph where detections, their connections, and the relational features of two connected nodes are represented by nodes, edges, and edge features, respectively. The strong edge features allow SGT to track targets with tracking candidates selected by top-K scored detections with large K. As a result, even low-scored detections can be tracked, and the missed detections are also recovered. The robustness of K value is shown through the extensive experiments. In the MOT16/17/20 and HiEve Challenge, SGT outperforms the state-of-the-art trackers with real-time inference speed. Especially, a large improvement in MOTA is shown in the MOT20 and HiEve Challenge. Code is available at https://github.com/HYUNJS/SGT. | ['Dit-yan Yeung', 'Dongyoon Wee', 'Myunggu Kang', 'Jeongseok Hyun'] | 2022-05-02 | null | null | null | null | ['online-multi-object-tracking'] | ['computer-vision'] | [-4.41241354e-01 -2.71698684e-01 -3.55876267e-01 1.58245864e-04
-5.43654799e-01 -7.38148510e-01 4.48919505e-01 2.11157724e-01
-1.51853904e-01 6.61855400e-01 -3.96451615e-02 2.98633844e-01
-2.90820956e-01 -7.34741509e-01 -7.67103553e-01 -7.26520360e-01
-7.48453736e-01 3.32409918e-01 9.65818644e-01 1.42352402e-01
-2.30137020e-01 4.32979286e-01 -1.29331601e+00 -2.69709855e-01
5.09288728e-01 1.00003850e+00 1.72800213e-01 5.61507165e-01
2.10115120e-01 3.97554994e-01 -5.52605152e-01 -4.04244393e-01
5.22165596e-01 -2.07616299e-01 1.04839519e-01 -1.49220690e-01
8.64884138e-01 -2.91070998e-01 -1.00190806e+00 1.22329366e+00
2.96642929e-01 -9.98270977e-03 2.89556295e-01 -1.65085340e+00
-3.11091989e-01 6.96009994e-01 -9.34697986e-01 4.92683887e-01
2.68182456e-01 8.21632743e-02 9.26449299e-01 -7.35580862e-01
5.34096360e-01 1.22663569e+00 9.67662573e-01 1.72327444e-01
-1.06294549e+00 -1.16186869e+00 4.82448190e-01 2.44233131e-01
-1.80996311e+00 -2.78261483e-01 3.87234837e-01 -3.46732467e-01
3.84455621e-01 3.47126305e-01 9.84162390e-01 6.99454367e-01
8.26424211e-02 6.95263565e-01 5.33434272e-01 2.46715993e-01
-1.99278608e-01 -1.98242366e-01 2.18680143e-01 8.80530000e-01
1.05392277e+00 5.17229915e-01 -8.92981827e-01 -3.55098426e-01
6.43054962e-01 2.63651282e-01 -2.17435375e-01 -4.64138120e-01
-1.48471689e+00 6.68125927e-01 9.57563937e-01 -7.44068064e-03
-3.84892821e-01 3.48202407e-01 1.36584774e-01 2.84657836e-01
2.56950021e-01 -1.35686725e-01 3.56063470e-02 6.90654889e-02
-8.91113281e-01 2.23244447e-03 5.17836392e-01 1.15015650e+00
7.88332462e-01 6.09842874e-02 -5.80960333e-01 3.02701265e-01
4.96124059e-01 1.04108286e+00 -2.61223525e-01 -6.56620920e-01
2.69016504e-01 6.72945797e-01 3.25285763e-01 -1.45749569e+00
-4.03698981e-01 -9.05388474e-01 -7.04361141e-01 -8.03021044e-02
2.16379479e-01 -1.62424952e-01 -7.79143989e-01 1.77639484e+00
7.57371783e-01 8.90753210e-01 -9.68901068e-02 9.57598507e-01
9.00854588e-01 5.02797842e-01 -2.26166919e-01 -3.58866096e-01
1.24025059e+00 -8.09726954e-01 -6.76353276e-01 -2.18256250e-01
4.11918402e-01 -7.70365477e-01 -1.34016156e-01 -5.52184097e-02
-6.05346262e-01 -6.58098936e-01 -8.82307410e-01 7.57657409e-01
-9.31109861e-02 5.50415039e-01 5.49846828e-01 2.28597164e-01
-1.05827773e+00 2.50187635e-01 -1.05228126e+00 -5.47299266e-01
4.90561932e-01 3.33600789e-01 -2.00090244e-01 -1.48267001e-01
-9.51140225e-01 5.41106582e-01 4.29615229e-01 2.08784461e-01
-1.03909421e+00 -4.03834522e-01 -7.02414155e-01 -1.25424758e-01
7.65165091e-01 -3.61286461e-01 8.39790106e-01 -1.89270765e-01
-7.08830059e-01 3.09638947e-01 -3.44603956e-01 -7.59501159e-01
5.24751723e-01 -2.17195153e-01 -7.43257940e-01 -5.21935448e-02
4.23873931e-01 6.07291281e-01 6.82454765e-01 -1.03618443e+00
-1.18150294e+00 -1.84070706e-01 -7.82171190e-02 3.73577438e-02
-2.46462166e-01 -2.37408623e-01 -8.72193336e-01 -6.50452375e-01
3.83880883e-01 -1.16457880e+00 5.67522924e-03 3.61246288e-01
-4.25791979e-01 -4.98941839e-01 1.06663394e+00 -2.08815709e-01
1.21895158e+00 -2.29519892e+00 -9.08052772e-02 2.84923434e-01
5.55633962e-01 3.11032563e-01 -1.25461683e-01 4.00050104e-01
5.35845935e-01 -2.74476379e-01 4.27990109e-01 -2.35437557e-01
-6.11075200e-02 2.09607020e-01 -4.49927263e-02 7.73887098e-01
-1.57830626e-01 7.84277081e-01 -1.21813321e+00 -5.97431421e-01
2.60573715e-01 3.87102872e-01 -1.70751289e-01 5.14986180e-02
1.36821598e-01 3.57896656e-01 -8.01710486e-01 8.83696556e-01
7.98394740e-01 -3.82410258e-01 7.73819461e-02 -4.43520635e-01
-2.70990938e-01 2.03465391e-02 -1.40886343e+00 1.23702288e+00
3.65845442e-01 7.11932957e-01 1.23660706e-01 -5.87013841e-01
8.64902794e-01 4.38015424e-02 7.26082802e-01 -4.74264383e-01
-1.24484211e-01 -6.06584409e-03 7.33411079e-03 5.21025434e-02
4.21467245e-01 6.95287466e-01 -1.85188502e-02 1.27736434e-01
-1.08739801e-01 7.29653060e-01 5.41524887e-01 7.29007125e-01
1.41534805e+00 1.50059536e-02 -1.73299517e-02 1.05631709e-01
1.92843720e-01 2.36773297e-01 1.11443675e+00 1.12698364e+00
-3.39211404e-01 1.15435801e-01 -7.13751763e-02 -2.69826353e-01
-3.35488290e-01 -1.33947599e+00 3.01032104e-02 8.21164489e-01
7.62559354e-01 -7.89722621e-01 -5.24112917e-02 -5.93399644e-01
5.24299383e-01 3.40158604e-02 -5.01520216e-01 -3.75316113e-01
-3.69312733e-01 -5.18808424e-01 5.37418604e-01 4.01629359e-01
4.37179357e-01 -6.73058569e-01 -4.44984704e-01 3.54867488e-01
-1.43851101e-01 -1.32967782e+00 -7.89179981e-01 -1.83171332e-01
-7.35386848e-01 -1.32806396e+00 -4.37676579e-01 -5.02589405e-01
7.16973126e-01 1.09654951e+00 8.30013633e-01 3.52564901e-01
-1.95072860e-01 4.70679104e-01 -5.23972929e-01 -2.29392231e-01
6.83199167e-02 -1.75542802e-01 4.91353303e-01 1.29926592e-01
2.61348963e-01 -3.09313446e-01 -6.60710990e-01 7.69559145e-01
-1.98768690e-01 -1.63769469e-01 6.79250360e-01 5.65223575e-01
8.32427859e-01 9.75250304e-02 3.97598445e-01 -2.72590727e-01
-1.17889076e-01 -6.04964077e-01 -1.03840661e+00 2.97000378e-01
-2.66220063e-01 -2.74084002e-01 2.22263262e-01 -7.46527195e-01
-5.09077191e-01 2.27912381e-01 5.31997800e-01 -9.01448369e-01
3.79681855e-01 3.82562846e-01 1.48050398e-01 -5.34839213e-01
4.83779699e-01 1.89847544e-01 -2.70037293e-01 -3.91220599e-01
6.61754310e-02 5.35659157e-02 6.08319640e-01 -1.81947067e-01
1.50336671e+00 5.88863850e-01 2.00254083e-01 -5.48768044e-01
-9.19085264e-01 -8.03539991e-01 -3.36936653e-01 -6.59294426e-01
2.62132585e-01 -1.42603600e+00 -7.38522172e-01 2.34507665e-01
-8.28236103e-01 4.46458757e-02 -7.83624575e-02 9.33033645e-01
2.18474045e-01 3.75601918e-01 -3.55748802e-01 -1.01377749e+00
-7.63142928e-02 -7.94091344e-01 1.01899600e+00 7.59663522e-01
1.72990948e-01 -5.98022997e-01 7.85740614e-02 -1.63938925e-01
2.49233767e-01 3.89757842e-01 -3.93139184e-01 -5.49968302e-01
-1.15636492e+00 -6.41514361e-01 -3.34907532e-01 -3.24745476e-01
2.34202176e-01 1.45600796e-01 -4.46077108e-01 -8.64895284e-01
-7.82545865e-01 -8.31305087e-02 1.18804228e+00 5.73352039e-01
4.63327110e-01 -1.51929557e-01 -1.24641466e+00 4.97260362e-01
1.15200043e+00 -3.17559540e-02 6.09448887e-02 -2.07860991e-02
9.05109227e-01 -5.52198067e-02 1.21228874e+00 4.70838994e-01
5.32853186e-01 9.23398256e-01 6.14427924e-01 6.58477247e-02
-4.41912740e-01 -5.10023594e-01 7.75628209e-01 5.85993409e-01
-4.55401540e-02 -2.54134029e-01 -5.37129819e-01 6.75196290e-01
-2.28413010e+00 -1.15185356e+00 -6.38736308e-01 2.58971477e+00
3.43849480e-01 4.02006954e-01 3.78291368e-01 -4.33719069e-01
1.21074235e+00 6.46816716e-02 -6.82422698e-01 8.54309559e-01
-8.04289579e-02 -4.17984366e-01 9.56272542e-01 3.15724552e-01
-1.20289779e+00 8.43020618e-01 5.21078682e+00 9.69029844e-01
-6.33053303e-01 1.29621476e-01 -2.78188914e-01 -2.57977843e-01
2.52103686e-01 2.79269278e-01 -1.47208476e+00 7.47327209e-01
5.65586925e-01 -2.30817392e-01 2.51751661e-01 7.10915208e-01
9.34143290e-02 -1.99919254e-01 -7.48777866e-01 9.10804927e-01
-1.51365578e-01 -1.36011958e+00 -2.54494250e-01 2.32099831e-01
6.00138724e-01 6.11740530e-01 -2.08137959e-01 4.17569578e-01
7.37819135e-01 -4.28358316e-01 6.61442876e-01 5.14648378e-01
3.34570795e-01 -5.03058314e-01 6.36143148e-01 2.92025983e-01
-2.24394369e+00 -5.94772659e-02 -6.93354726e-01 2.24812880e-01
2.48759568e-01 7.28736162e-01 -6.06412709e-01 9.35979605e-01
1.07415771e+00 1.23793554e+00 -6.85041010e-01 1.86800873e+00
-7.16810301e-02 6.80264652e-01 -9.32492197e-01 -4.01528329e-02
1.31986782e-01 -1.77499298e-02 1.15122676e+00 1.08414984e+00
5.49506962e-01 2.28713844e-02 1.00216579e+00 5.45455396e-01
-8.35972745e-03 -3.59612018e-01 -5.91356575e-01 1.32608116e-01
1.10737884e+00 1.57966185e+00 -7.99332917e-01 -3.72080743e-01
-4.32428211e-01 2.96015471e-01 3.00144374e-01 2.30427936e-01
-1.22879720e+00 4.45297994e-02 7.45604932e-01 5.53037412e-02
6.68409109e-01 -2.15139344e-01 4.80391473e-01 -1.20944226e+00
1.31677851e-01 -3.42197120e-01 7.58464515e-01 -3.68669927e-01
-1.31547701e+00 4.87544566e-01 -6.63490780e-03 -1.81485009e+00
2.00997338e-01 -3.94576266e-02 -7.04399288e-01 3.84044617e-01
-1.21272826e+00 -1.19891560e+00 -6.78875625e-01 6.32960618e-01
4.65976186e-02 -1.15932390e-01 1.94484010e-01 5.34344435e-01
-6.56066000e-01 6.47143900e-01 4.42273803e-02 4.82080311e-01
7.18910396e-01 -8.36848140e-01 3.27490002e-01 1.11353993e+00
4.58508492e-01 2.98237085e-01 7.81282663e-01 -1.21687520e+00
-1.66983843e+00 -1.53419709e+00 4.74034011e-01 -2.70895630e-01
8.73114347e-01 -1.75471440e-01 -7.06338406e-01 7.75209725e-01
-1.30301684e-01 6.82367027e-01 2.41223961e-01 7.53686056e-02
-3.13278556e-01 -3.44262093e-01 -7.95260251e-01 3.76676679e-01
1.51939797e+00 -4.13780510e-02 -2.30385974e-01 4.66174573e-01
4.82065469e-01 -4.81911540e-01 -9.11412060e-01 5.28044045e-01
5.67383766e-01 -6.43887460e-01 1.11714911e+00 5.94450673e-03
-8.71915638e-01 -1.17229021e+00 2.32603848e-02 -1.10774517e+00
-5.76539218e-01 -5.73296726e-01 -6.30237699e-01 1.38872850e+00
2.21946299e-01 -8.08950126e-01 6.65951073e-01 -2.26001292e-01
-3.42966244e-02 -2.88094491e-01 -1.13093054e+00 -1.31402886e+00
-8.40344310e-01 4.23194095e-02 2.63850898e-01 7.07830131e-01
-3.63069832e-01 1.29256040e-01 -7.35189915e-01 7.59089231e-01
1.33922899e+00 4.79525298e-01 1.07215416e+00 -1.72146571e+00
-2.79187769e-01 -1.14119470e-01 -7.84832597e-01 -1.30229390e+00
-2.16714889e-01 -9.26910996e-01 1.30670577e-01 -1.36322641e+00
4.13925827e-01 -7.33935297e-01 -4.43857938e-01 7.14852333e-01
-2.07661346e-01 4.45369601e-01 3.36693585e-01 5.89596987e-01
-1.24416566e+00 4.52303141e-01 1.01488554e+00 -3.19515765e-01
-1.03951797e-01 3.28204006e-01 -2.69643813e-01 4.14403707e-01
7.06382394e-01 -1.05066085e+00 6.82605505e-02 -1.44712493e-01
-1.14661735e-03 1.53405830e-01 7.83899307e-01 -1.39610600e+00
7.71152437e-01 8.85452107e-02 6.76317334e-01 -1.20397997e+00
4.83311087e-01 -8.31997216e-01 6.51731312e-01 7.24349201e-01
1.45067021e-01 -6.84449077e-02 2.60285109e-01 1.21666718e+00
-1.09336279e-01 1.79085106e-01 3.70309472e-01 2.56741703e-01
-7.85568058e-01 8.37779999e-01 -4.43303138e-02 -9.17615592e-02
1.27714515e+00 -3.44423294e-01 -5.57872534e-01 -3.63877803e-01
-6.33755505e-01 9.79424298e-01 4.23172832e-01 5.86999297e-01
3.95086676e-01 -1.70226061e+00 -9.09024835e-01 -2.28128597e-01
3.41004819e-01 -1.04208834e-01 1.93377256e-01 1.30953777e+00
1.70585483e-01 1.33925721e-01 -3.48482244e-02 -1.13406563e+00
-1.64096570e+00 4.05255467e-01 2.10485041e-01 -1.73891395e-01
-8.66522968e-01 6.74471259e-01 2.04370230e-01 9.18154642e-02
5.54646313e-01 -5.62704839e-02 -7.54712150e-02 2.42344663e-02
5.85273087e-01 2.86433727e-01 -4.07883167e-01 -7.99597859e-01
-7.56904423e-01 7.67502367e-01 -2.12579258e-02 3.96578938e-01
1.02954018e+00 -2.09537968e-01 2.70111442e-01 3.62303108e-02
5.07791877e-01 1.92758039e-01 -1.63137853e+00 -5.93977094e-01
-6.95565641e-02 -6.97222471e-01 4.71820682e-02 -3.68681550e-01
-1.55892217e+00 1.65714547e-01 7.95660555e-01 3.89951527e-01
7.83602476e-01 2.89057195e-01 6.32151246e-01 3.93378913e-01
8.15746784e-01 -5.75356007e-01 6.32650703e-02 3.13431114e-01
3.77387702e-01 -1.12208593e+00 1.77943319e-01 -5.91099977e-01
-2.26207554e-01 8.79076779e-01 8.60868931e-01 -2.63581932e-01
5.71827412e-01 3.18532318e-01 -2.79076099e-01 -3.65288198e-01
-6.63566947e-01 -7.34914660e-01 4.11254436e-01 6.57175660e-01
-1.35557055e-01 3.07348549e-01 7.39639625e-03 1.68478712e-01
1.41822845e-01 -3.28243375e-01 1.90193623e-01 7.29003668e-01
-7.35974610e-01 -7.82272637e-01 -7.17351496e-01 5.74465036e-01
2.77737211e-02 2.03599289e-01 -4.72730875e-01 8.17804575e-01
1.17520757e-01 1.14674759e+00 2.91278828e-02 -6.47029102e-01
3.18439424e-01 -6.96871042e-01 2.78242469e-01 -2.98136681e-01
-3.52488667e-01 2.77069986e-01 9.54017118e-02 -6.94310606e-01
-6.17708802e-01 -7.74494112e-01 -1.24060738e+00 -4.90702212e-01
-7.44439244e-01 3.77175808e-01 1.05844438e-01 5.96745074e-01
5.94296217e-01 5.74225962e-01 5.56043684e-01 -7.18022466e-01
-2.23204374e-01 -8.75262499e-01 -5.16918898e-01 8.45855474e-03
3.47684324e-01 -1.37570560e+00 -2.91063547e-01 -3.78586918e-01] | [6.323513507843018, -2.122504949569702] |
89325676-761c-4755-a9ee-12b5e0c49bae | a-marketplace-for-web-scale-analytics-and | null | null | https://aclanthology.org/C14-2022 | https://aclanthology.org/C14-2022.pdf | A Marketplace for Web Scale Analytics and Text Annotation Services | null | ['Holger D{\\"u}wiger', 'er', 'Alex L{\\"o}ser', 'Peter Adolphs', 'Torsten Kilias', 'Holmer Hemsen', 'Johannes Kirschnick', 'Heiko Ehrig'] | 2014-08-01 | a-marketplace-for-web-scale-analytics-and-1 | https://aclanthology.org/C14-2022 | https://aclanthology.org/C14-2022.pdf | coling-2014-8 | ['text-annotation'] | ['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] | [-7.292387962341309, 3.742838144302368] |
e07a0751-3415-4198-b74d-9d0af204a54f | pre-training-strategies-and-datasets-for | 2103.16554 | null | https://arxiv.org/abs/2103.16554v2 | https://arxiv.org/pdf/2103.16554v2.pdf | Pre-training strategies and datasets for facial representation learning | What is the best way to learn a universal face representation? Recent work on Deep Learning in the area of face analysis has focused on supervised learning for specific tasks of interest (e.g. face recognition, facial landmark localization etc.) but has overlooked the overarching question of how to find a facial representation that can be readily adapted to several facial analysis tasks and datasets. To this end, we make the following 4 contributions: (a) we introduce, for the first time, a comprehensive evaluation benchmark for facial representation learning consisting of 5 important face analysis tasks. (b) We systematically investigate two ways of large-scale representation learning applied to faces: supervised and unsupervised pre-training. Importantly, we focus our evaluations on the case of few-shot facial learning. (c) We investigate important properties of the training datasets including their size and quality (labelled, unlabelled or even uncurated). (d) To draw our conclusions, we conducted a very large number of experiments. Our main two findings are: (1) Unsupervised pre-training on completely in-the-wild, uncurated data provides consistent and, in some cases, significant accuracy improvements for all facial tasks considered. (2) Many existing facial video datasets seem to have a large amount of redundancy. We will release code, and pre-trained models to facilitate future research. | ['Georgios Tzimiropoulos', 'Enrique Sanchez', 'Andrew Garbett', 'Jing Yang', 'Shiyang Cheng', 'Adrian Bulat'] | 2021-03-30 | null | null | null | null | ['face-alignment', '3d-facial-landmark-localization', 'facial-action-unit-detection', 'unsupervised-pre-training'] | ['computer-vision', 'computer-vision', 'computer-vision', 'methodology'] | [ 2.12049231e-01 -5.48326038e-02 -1.66437671e-01 -6.98247492e-01
-7.20457911e-01 -2.96065062e-01 7.57570803e-01 -1.66212097e-01
-1.27369776e-01 4.61167604e-01 1.92192852e-01 1.22310624e-01
-2.50710219e-01 -4.18809801e-01 -4.77870822e-01 -6.24617338e-01
-4.33987349e-01 3.59607309e-01 -2.16349155e-01 -2.17034414e-01
6.65730983e-02 1.00193226e+00 -2.20997334e+00 3.03501219e-01
-2.32664286e-03 1.25408697e+00 -3.97820085e-01 3.71381402e-01
1.31195292e-01 9.70801890e-01 -4.15760636e-01 -6.38277531e-01
3.43007267e-01 -4.30053174e-01 -1.15820932e+00 5.14263630e-01
8.34513366e-01 -3.96067739e-01 -2.80238986e-01 9.88525391e-01
7.04179525e-01 1.41689792e-01 8.06077182e-01 -1.45140493e+00
-5.18108368e-01 1.78966284e-01 -6.42129660e-01 1.92264065e-01
1.91262439e-01 1.04875170e-01 7.26926148e-01 -9.98394907e-01
7.84426928e-01 1.44810641e+00 6.77713037e-01 9.20011997e-01
-1.07895601e+00 -7.66755462e-01 -9.66912508e-02 1.40914708e-01
-1.61302507e+00 -1.36762404e+00 7.69953609e-01 -5.08337975e-01
7.14599311e-01 1.51485398e-01 1.80185184e-01 1.20541024e+00
-1.47206336e-01 6.06169522e-01 1.20645857e+00 -5.78717947e-01
1.27357274e-01 -8.70520547e-02 -3.95475700e-02 9.25543547e-01
8.41309279e-02 4.58520763e-02 -4.45272148e-01 -1.43926620e-01
6.42262459e-01 -1.15971349e-01 -5.66946007e-02 -4.19157654e-01
-7.23428071e-01 9.08571184e-01 1.12237729e-01 5.48675478e-01
-1.82747781e-01 2.21294180e-01 6.97223008e-01 6.18609250e-01
6.17815077e-01 1.60416260e-01 -4.22102749e-01 -1.51513472e-01
-1.00443947e+00 -5.69515713e-02 6.55038059e-01 8.71640742e-01
9.98303652e-01 3.24117780e-01 9.32825282e-02 1.12234378e+00
8.81172493e-02 2.56383210e-01 5.85834503e-01 -8.72030556e-01
5.40563203e-02 1.01628393e-01 -2.67608762e-01 -1.08745754e+00
-5.57280064e-01 2.12496743e-02 -8.10154080e-01 4.26844686e-01
4.65566903e-01 -2.79906482e-01 -9.55608547e-01 1.86403191e+00
1.62800238e-01 2.82086194e-01 -2.05255505e-02 5.13020396e-01
1.10111237e+00 2.41406590e-01 1.44349277e-01 -2.56327242e-01
1.40077233e+00 -7.72942424e-01 -5.97874284e-01 8.15278385e-03
7.41135359e-01 -7.86982894e-01 5.22221446e-01 1.19922765e-01
-8.93806398e-01 -7.28004694e-01 -6.75386906e-01 2.26195939e-02
-5.16262293e-01 2.39888698e-01 8.32574368e-01 7.57881224e-01
-1.35747540e+00 7.28577852e-01 -5.79907477e-01 -8.84819329e-01
7.76051700e-01 5.10144293e-01 -1.02954984e+00 -2.12608784e-01
-7.69675076e-01 8.66718531e-01 4.94290069e-02 -7.23628849e-02
-1.25282276e+00 -4.00305688e-01 -1.03152251e+00 -3.53260934e-02
4.94156927e-01 -3.67359519e-01 1.17089093e+00 -1.56917512e+00
-1.38822889e+00 1.43349028e+00 -3.04686934e-01 -1.80669829e-01
1.40304089e-01 4.41794619e-02 -5.21173298e-01 3.15623075e-01
-1.68816838e-02 8.48928869e-01 1.16240263e+00 -1.34224010e+00
-4.12952036e-01 -7.66380727e-01 -5.57861291e-02 -2.91375428e-01
-4.26999837e-01 5.19269466e-01 -3.95298392e-01 -7.32058942e-01
-1.85226023e-01 -8.09814811e-01 1.38816029e-01 1.40784010e-01
3.80909480e-02 -4.92325336e-01 7.94385016e-01 -4.23042417e-01
8.81114304e-01 -2.25181627e+00 7.09678903e-02 2.90881544e-02
1.64808437e-01 4.50503945e-01 -5.31303465e-01 4.34072703e-01
-5.97650588e-01 1.70155361e-01 -2.17077598e-01 -7.72104144e-01
-4.06938829e-02 2.37505183e-01 -1.73212290e-01 7.45943964e-01
4.25086945e-01 9.78781521e-01 -7.45693147e-01 -5.32536507e-01
-5.49304597e-02 5.83261549e-01 -2.25151971e-01 4.67866249e-02
2.22593993e-01 2.88233250e-01 -1.70444056e-01 1.18244147e+00
6.40206993e-01 1.64306592e-02 6.85252547e-02 -1.72935456e-01
1.49999365e-01 -1.98089942e-01 -9.93360043e-01 1.55789435e+00
-3.30917150e-01 7.68245816e-01 2.09330365e-01 -1.15338087e+00
8.61132264e-01 5.13418972e-01 7.03059077e-01 -6.57013357e-01
2.61758119e-01 1.57469764e-01 -4.03401181e-02 -5.70684731e-01
1.65379182e-01 -3.57256800e-01 3.63768607e-01 6.90166473e-01
8.25808048e-01 3.16818804e-01 1.54368252e-01 -1.60001159e-01
8.88203025e-01 -1.97568722e-02 4.78846371e-01 -3.68827194e-01
5.27367532e-01 -4.87973630e-01 2.53221810e-01 3.34015757e-01
-4.75080729e-01 7.67939091e-01 7.79857039e-01 -7.33300865e-01
-8.50055754e-01 -4.07527536e-01 -3.50674212e-01 1.59702587e+00
-5.34160912e-01 -5.34907877e-01 -8.23217928e-01 -8.89059842e-01
5.89540638e-02 6.95433021e-02 -1.15797400e+00 4.59848084e-02
-3.69822621e-01 -6.94275022e-01 7.34361589e-01 6.28735185e-01
1.60738915e-01 -1.26620471e+00 -3.52508515e-01 -2.91502416e-01
2.69246340e-01 -1.02478099e+00 -1.79831326e-01 1.52595282e-01
-8.71125400e-01 -1.23169541e+00 -7.90599942e-01 -9.39569652e-01
7.87444949e-01 3.96762252e-01 1.19884312e+00 4.22104061e-01
-3.28356653e-01 9.51616228e-01 -4.34570432e-01 -4.44427043e-01
-1.35567799e-01 -1.49754599e-01 2.17275783e-01 4.27747965e-01
6.52062118e-01 -5.10264337e-01 -2.16590583e-01 3.40106100e-01
-9.19123411e-01 -6.51160419e-01 5.00834286e-01 7.90415347e-01
4.44597602e-01 -1.92812644e-02 6.92730546e-01 -8.26417327e-01
5.02004623e-01 -5.33647358e-01 -3.25639486e-01 2.06467032e-01
-2.52005130e-01 -1.03929095e-01 3.31716746e-01 -1.61756650e-01
-7.17305660e-01 2.48946175e-01 -4.08833593e-01 -6.51407778e-01
-4.93074447e-01 2.77966917e-01 -1.53518066e-01 -4.19922382e-01
6.57149017e-01 1.32929474e-01 5.14494061e-01 -5.70915282e-01
3.16076368e-01 5.46361327e-01 2.84583420e-01 -7.92277813e-01
6.73724353e-01 8.01367521e-01 2.45290756e-01 -1.28042626e+00
-5.12816489e-01 -4.16244119e-01 -1.16909063e+00 -1.88415423e-01
6.24153614e-01 -8.34816337e-01 -5.31352103e-01 5.29441714e-01
-9.06298757e-01 -3.54186356e-01 -3.30344558e-01 1.24022506e-01
-7.13567495e-01 4.89641488e-01 -3.87401849e-01 -9.06864226e-01
-1.77988246e-01 -1.01146746e+00 1.26009738e+00 5.08107282e-02
-2.64417887e-01 -1.02758718e+00 7.41228834e-02 9.45928097e-02
3.53401780e-01 4.85651970e-01 5.95702887e-01 -7.04318047e-01
-8.52139294e-02 -2.18345091e-01 -1.80090994e-01 5.09004414e-01
3.54966104e-01 2.32008532e-01 -1.51907241e+00 -5.31189978e-01
-1.21664636e-01 -8.47932935e-01 1.04995334e+00 2.62212455e-01
1.34735727e+00 -1.14933081e-01 -1.43466741e-01 6.27741575e-01
1.20211291e+00 -1.22894254e-02 8.90780807e-01 1.48270294e-01
4.92182076e-01 1.07958293e+00 3.47899765e-01 3.85753453e-01
-5.63991368e-02 8.15798044e-01 3.65911365e-01 1.24172056e-02
-3.81489873e-01 -5.86271659e-02 3.22895050e-01 4.14468795e-01
-5.73750019e-01 1.61203831e-01 -6.79231524e-01 6.09746516e-01
-1.60577929e+00 -1.16533291e+00 4.45148826e-01 2.19763398e+00
5.31541586e-01 -3.46963823e-01 5.22632480e-01 2.16181830e-01
5.88098228e-01 2.83721387e-01 -8.99299309e-02 -3.36018056e-01
-5.61190806e-02 6.52198911e-01 1.17600843e-01 1.59944072e-01
-1.36072063e+00 1.00428617e+00 6.69362879e+00 8.07920516e-01
-1.31773305e+00 2.05625027e-01 8.46606195e-01 1.56862009e-02
1.69922724e-01 -3.26052666e-01 -7.51996398e-01 2.73603797e-01
9.21306670e-01 5.63353300e-03 2.46109650e-01 9.64279771e-01
-2.15905562e-01 2.41304323e-01 -1.50202417e+00 1.38020837e+00
4.16660935e-01 -1.15782320e+00 2.23580256e-01 8.28877240e-02
7.36515880e-01 -1.12726428e-01 2.71453917e-01 4.04381007e-01
-1.66819453e-01 -1.67943180e+00 4.92595375e-01 4.70800728e-01
1.11100101e+00 -9.55837429e-01 7.90135860e-01 -1.56536847e-01
-1.29012728e+00 -1.63428038e-01 -5.66834152e-01 3.24405134e-02
-2.26071671e-01 1.07062586e-01 -3.35567832e-01 5.83247542e-01
6.40475869e-01 9.47462916e-01 -8.21885765e-01 8.54170859e-01
5.00649400e-02 4.40438032e-01 -6.32055700e-02 3.93921852e-01
3.93912286e-01 5.07682860e-02 1.08497798e-01 1.22589684e+00
2.58188158e-01 9.43040475e-02 -1.57030255e-01 3.73811871e-01
-4.73610550e-01 3.26019228e-01 -9.90990341e-01 -1.40293807e-01
1.20921366e-01 1.43293130e+00 -7.19425440e-01 -1.43588707e-01
-7.25443244e-01 6.17798388e-01 5.16491115e-01 2.68725902e-01
-3.34950238e-01 -2.32028663e-01 8.71315598e-01 2.89748818e-01
4.20070112e-01 5.19148596e-02 1.39309391e-01 -9.38348055e-01
-1.52528226e-01 -1.01211596e+00 5.49344480e-01 -5.01783907e-01
-1.47713494e+00 7.95589626e-01 2.77217524e-03 -9.69698250e-01
-3.29654008e-01 -9.38396454e-01 -6.74094379e-01 5.21519125e-01
-1.52085805e+00 -1.22541797e+00 -2.24122182e-01 7.97341824e-01
5.23848653e-01 -7.76503265e-01 1.07773221e+00 4.82596427e-01
-5.93128204e-01 9.04121995e-01 5.17862290e-02 5.45042813e-01
8.68641555e-01 -7.69898951e-01 1.62821710e-01 5.10001421e-01
4.77195263e-01 6.36913955e-01 2.77942181e-01 -1.91142142e-01
-1.50047922e+00 -1.19998741e+00 7.37161219e-01 -6.02916360e-01
5.01173317e-01 -2.03850821e-01 -9.51284051e-01 9.09318209e-01
3.96195464e-02 5.51646531e-01 8.97837222e-01 3.45160156e-01
-5.60256243e-01 -2.32878923e-01 -1.33188105e+00 2.73928463e-01
1.10313034e+00 -6.88236833e-01 -3.43423367e-01 2.97798544e-01
4.35399488e-02 7.78948218e-02 -9.10702944e-01 4.63059336e-01
7.65071452e-01 -1.03750479e+00 9.26725686e-01 -1.04001546e+00
5.38539827e-01 9.70764533e-02 -2.53336400e-01 -1.03427517e+00
-3.05051714e-01 -6.67655468e-01 -2.10138097e-01 1.33191895e+00
-1.38540920e-02 -5.48385024e-01 7.16162860e-01 3.80450934e-01
1.04594044e-01 -1.03345501e+00 -1.06082523e+00 -7.76370049e-01
1.08480074e-01 -2.87387550e-01 6.69331491e-01 1.08834779e+00
-2.02148274e-01 3.35441172e-01 -6.13045335e-01 -1.84548095e-01
5.12551844e-01 -1.63357537e-02 9.60961223e-01 -1.48519266e+00
2.89910734e-01 -5.68965793e-01 -9.05133843e-01 -7.31755316e-01
5.89649796e-01 -9.67999041e-01 -2.91918308e-01 -1.06358016e+00
3.39391947e-01 -2.38106683e-01 -3.64876688e-01 7.74788857e-01
1.85090333e-01 6.04179442e-01 1.21875808e-01 1.49464369e-01
-6.39192224e-01 4.49542969e-01 1.04028761e+00 1.82596147e-02
3.16062808e-01 -5.81787117e-02 -9.33695376e-01 7.27059662e-01
6.65016949e-01 -3.15111846e-01 -3.20980459e-01 -3.80252540e-01
-2.95241833e-01 -3.11439157e-01 2.98155129e-01 -8.62805903e-01
-2.19936892e-02 -6.23149537e-02 5.38241744e-01 -9.88557711e-02
3.16590965e-01 -7.18403220e-01 -2.52712339e-01 1.38248056e-01
-7.06487447e-02 -4.01418805e-02 4.20835882e-01 3.61180305e-01
-3.53686303e-01 -4.23705012e-01 1.19792056e+00 -2.20550194e-01
-1.12935078e+00 6.13494158e-01 -2.35621229e-01 1.72102600e-02
1.11985731e+00 -1.96355596e-01 -5.34702018e-02 -5.04025877e-01
-6.84088826e-01 -2.06745297e-01 4.15283769e-01 5.55613339e-01
5.48777938e-01 -1.51650906e+00 -8.61314297e-01 2.65408635e-01
2.83729613e-01 -3.49686444e-01 1.55675828e-01 6.88089192e-01
-3.44242573e-01 5.51788926e-01 -6.19666517e-01 -5.20537913e-01
-1.58195138e+00 7.02411711e-01 3.73026103e-01 2.38204762e-01
-4.33089525e-01 8.36984098e-01 5.16190045e-02 -2.42466152e-01
3.98254275e-01 1.77653924e-01 -4.36500847e-01 5.53043365e-01
9.33043122e-01 3.79296780e-01 2.72111595e-01 -1.39674723e+00
-4.40277696e-01 8.03022265e-01 -3.55157740e-02 4.06420469e-01
1.60903740e+00 1.34331882e-01 -3.33814323e-01 1.89087212e-01
1.58393502e+00 -2.09089294e-01 -1.03984928e+00 -1.45677939e-01
7.11540729e-02 -5.94022989e-01 -1.77783400e-01 -2.16497242e-01
-1.55911160e+00 1.03019643e+00 7.50071883e-01 -1.24635454e-02
1.20849109e+00 1.89867690e-01 2.89846182e-01 4.43432570e-01
4.35534865e-01 -1.01965892e+00 2.53110468e-01 5.37099719e-01
1.08135164e+00 -1.44921851e+00 2.00010121e-01 -1.31944269e-01
-6.01644397e-01 1.41941619e+00 4.62929487e-01 1.27312154e-01
9.05995011e-01 -2.71855090e-02 -5.10470420e-02 -5.14202237e-01
-6.34209514e-01 -6.19079471e-01 3.38493258e-01 8.54362249e-01
7.02403486e-01 -2.91471183e-01 1.04846343e-01 3.98546487e-01
-5.64022213e-02 5.86776026e-02 1.89672545e-01 9.99509633e-01
-3.37204009e-01 -1.07325733e+00 -3.91178817e-01 3.78386348e-01
-5.70286334e-01 2.79947340e-01 -6.27145469e-01 9.63453948e-01
1.71346501e-01 8.79382968e-01 4.93271872e-02 -2.74041235e-01
1.67625412e-01 4.14780259e-01 8.31768513e-01 -7.87789285e-01
-1.99596390e-01 -1.17023550e-01 -6.76640496e-02 -5.55321574e-01
-6.91425562e-01 -6.87726021e-01 -6.81773365e-01 -4.30477142e-01
-3.64644118e-02 -7.90621787e-02 3.66901904e-01 8.77222776e-01
3.14470202e-01 2.15834573e-01 6.33574784e-01 -1.12969530e+00
-4.62245613e-01 -1.12382710e+00 -8.91722500e-01 7.08124518e-01
3.35287631e-01 -1.07844353e+00 -3.32714945e-01 1.89745188e-01] | [13.39468765258789, 1.1921461820602417] |
37fa7fe2-d7ab-4355-a4c5-b2f5945a440f | multi-task-learning-for-radar-signal | 2306.13105 | null | https://arxiv.org/abs/2306.13105v1 | https://arxiv.org/pdf/2306.13105v1.pdf | Multi-task Learning for Radar Signal Characterisation | Radio signal recognition is a crucial task in both civilian and military applications, as accurate and timely identification of unknown signals is an essential part of spectrum management and electronic warfare. The majority of research in this field has focused on applying deep learning for modulation classification, leaving the task of signal characterisation as an understudied area. This paper addresses this gap by presenting an approach for tackling radar signal classification and characterisation as a multi-task learning (MTL) problem. We propose the IQ Signal Transformer (IQST) among several reference architectures that allow for simultaneous optimisation of multiple regression and classification tasks. We demonstrate the performance of our proposed MTL model on a synthetic radar dataset, while also providing a first-of-its-kind benchmark for radar signal characterisation. | ['Terrence Martin', 'Clinton Fookes', 'Simon Denman', 'Akila Pemasiri', 'Zi Huang'] | 2023-06-19 | null | null | null | null | ['classification-1', 'multi-task-learning', 'management'] | ['methodology', 'methodology', 'miscellaneous'] | [ 1.03943336e+00 -5.36746025e-01 -1.05891176e-01 -3.14029098e-01
-1.20083594e+00 -4.63778138e-01 8.38240027e-01 -1.51138306e-01
-4.11081761e-01 7.01132655e-01 -1.29095107e-01 -7.16893494e-01
-6.67549014e-01 -3.84705901e-01 -1.47719279e-01 -8.84521127e-01
-3.63270432e-01 4.74176943e-01 -2.66508553e-02 -3.97947371e-01
2.54387200e-01 8.48730326e-01 -1.25200117e+00 2.96101421e-01
4.13832843e-01 1.51376283e+00 -1.56964526e-01 8.08873296e-01
3.72188747e-01 8.19981813e-01 -1.00657475e+00 -2.59412736e-01
3.89331639e-01 -4.94230896e-01 -3.92568588e-01 -4.17186052e-01
3.10437530e-01 1.37597755e-01 -2.96276838e-01 8.31577599e-01
9.02090967e-01 -1.84267342e-01 8.06791008e-01 -1.20788658e+00
8.81196484e-02 3.67633402e-01 -4.01759535e-01 5.65587461e-01
-2.82965600e-02 -1.17856413e-01 1.14654028e+00 -3.52696478e-01
-1.06286602e-02 7.18262553e-01 5.34372211e-01 1.34748757e-01
-1.40798748e+00 -1.13125682e+00 -4.34512347e-01 6.17146015e-01
-1.38087487e+00 -7.20427155e-01 8.72424543e-01 -4.13189143e-01
6.97936475e-01 4.34858114e-01 3.08144480e-01 1.20319510e+00
5.52084684e-01 8.11636984e-01 1.44029260e+00 -2.60586113e-01
1.14486799e-01 -2.37927809e-01 5.65654822e-02 1.97127268e-01
-1.19946308e-01 7.08055437e-01 -4.03650522e-01 -2.05273062e-01
4.92084861e-01 -4.48345751e-01 -4.37585473e-01 -1.55370802e-01
-1.24541378e+00 8.61402810e-01 7.32268617e-02 5.19366384e-01
-4.82593089e-01 3.56247336e-01 3.01679730e-01 1.00685358e+00
3.95231903e-01 5.96958756e-01 -3.30575764e-01 -2.47101739e-01
-1.38917124e+00 2.88201064e-01 7.99461007e-01 3.20715666e-01
3.22430223e-01 6.73484862e-01 -1.59180239e-01 8.68853688e-01
-1.82388928e-02 7.91647136e-01 2.07822114e-01 -2.21949503e-01
1.00806378e-01 -7.45252669e-02 -1.54687211e-01 -9.09995675e-01
-8.47416937e-01 -1.36754346e+00 -1.22665334e+00 3.59500706e-01
1.52629748e-01 -3.73517871e-01 -7.91464865e-01 1.39566064e+00
-3.07380915e-01 5.27991951e-01 9.17638019e-02 8.95615458e-01
6.80613220e-01 6.18382037e-01 -3.59572053e-01 -1.93448648e-01
1.35830629e+00 -2.66987264e-01 -4.68485326e-01 -5.23023903e-01
1.80854157e-01 -8.33855808e-01 -2.38224994e-02 9.20090377e-01
-4.96366858e-01 -5.09850621e-01 -1.44103432e+00 6.35849774e-01
4.75274213e-02 3.52016240e-02 5.25100052e-01 1.01900673e+00
-6.38795197e-01 2.84244835e-01 -3.18024516e-01 2.35911965e-01
4.24767941e-01 4.56466109e-01 -5.25584146e-02 1.31465480e-01
-1.52012753e+00 1.13469338e+00 1.91409603e-01 2.02623427e-01
-1.02827108e+00 -7.76615739e-01 -5.91634274e-01 -1.23334743e-01
3.20379436e-01 -3.27201039e-01 1.18239641e+00 -8.67380381e-01
-1.48649967e+00 6.26104951e-01 3.05684060e-01 -1.05083764e+00
3.15766782e-01 -3.93169336e-02 -1.14279449e+00 -2.55363882e-01
-3.91644359e-01 -7.01375585e-03 1.66582465e+00 -9.78262484e-01
-7.72921979e-01 -8.21048319e-02 -2.64790922e-01 -1.71342626e-01
2.49487206e-01 1.94084510e-01 5.17068923e-01 -9.68772054e-01
-1.26962364e-01 -6.15438104e-01 -6.30369261e-02 -6.27453566e-01
-2.94658959e-01 3.14862579e-01 7.75264204e-01 -6.48665130e-01
1.13266814e+00 -1.99523509e+00 3.27791333e-01 5.34832120e-01
-3.65044102e-02 4.82314199e-01 -1.80635422e-01 4.13508922e-01
-3.22972417e-01 -5.17088413e-01 -3.93396586e-01 1.44220144e-01
4.04238068e-02 -3.71840060e-01 -7.79578149e-01 7.35739887e-01
2.76922554e-01 7.28905320e-01 -4.32365000e-01 5.21740541e-02
1.34181872e-01 3.51714104e-01 4.99864630e-02 -3.77609544e-02
-6.38896301e-02 7.17751265e-01 -2.09079340e-01 7.97466755e-01
4.06644076e-01 2.24450216e-01 5.01546673e-02 -5.07780433e-01
-8.35337490e-02 1.48890167e-01 -1.14702678e+00 1.10197425e+00
-7.84077287e-01 1.15189219e+00 3.01096112e-01 -1.61430907e+00
1.05972254e+00 3.75761062e-01 7.67912209e-01 -1.15329731e+00
3.89168888e-01 3.54671538e-01 6.99910283e-01 -9.03746560e-02
2.54288584e-01 -7.63908148e-01 -4.33526188e-01 7.17176259e-01
2.72108167e-02 -2.69475400e-01 -2.41032079e-01 -3.51167530e-01
1.37435663e+00 -3.42967302e-01 4.67612207e-01 -1.45840570e-01
7.58681655e-01 -1.71246156e-01 2.37088576e-01 8.00915480e-01
-1.06471188e-01 2.59517450e-02 1.65820956e-01 -3.56162935e-01
-7.29978085e-01 -9.62517798e-01 -7.05143213e-01 9.74428594e-01
-2.32091933e-01 2.10446462e-01 -1.29864171e-01 -2.29354456e-01
7.68839493e-02 6.92506969e-01 -1.38083205e-01 -2.13620409e-01
-6.35555148e-01 -1.31357658e+00 1.03750610e+00 5.40634757e-03
2.88070917e-01 -1.02260935e+00 -8.87606859e-01 4.66258109e-01
5.38931275e-03 -1.30064368e+00 1.38373896e-01 6.42929852e-01
-3.41616243e-01 -1.01824820e+00 -6.05606735e-01 -4.97942746e-01
-2.58175641e-01 1.06817245e-01 1.05157220e+00 -3.60738218e-01
-7.37465799e-01 1.55258521e-01 -3.70743036e-01 -5.59427440e-01
-5.65270662e-01 5.59097156e-02 1.42739609e-01 5.77313721e-01
7.61115551e-02 -7.56145954e-01 -2.76762813e-01 4.37140226e-01
-8.51657152e-01 -2.75804758e-01 1.18406749e+00 8.37500334e-01
6.87346235e-03 2.22441003e-01 1.05968821e+00 -6.07822359e-01
7.53461063e-01 -2.42574632e-01 -9.31556642e-01 1.11521155e-01
-1.47185281e-01 1.36369139e-01 6.75478399e-01 -1.63912863e-01
-4.80449498e-01 -9.85159278e-02 -5.17279267e-01 4.73036729e-02
-1.30703747e-01 6.30975604e-01 -1.22695170e-01 -7.84966111e-01
6.64472163e-01 6.01382971e-01 -1.16900340e-01 -2.14279652e-01
9.94700789e-02 9.01829183e-01 5.80265522e-01 -2.24857420e-01
1.23877621e+00 2.34320581e-01 5.97653866e-01 -1.47635901e+00
-8.44963193e-01 -5.30160367e-01 -3.83153796e-01 -3.24107826e-01
4.32628334e-01 -7.46014535e-01 -7.79634774e-01 5.20646393e-01
-8.74531090e-01 -2.11759120e-01 1.73307434e-01 4.86304194e-01
-6.04772449e-01 3.34409595e-01 5.49633466e-02 -1.08781958e+00
-3.81291837e-01 -1.11834204e+00 1.02293968e+00 -5.99086761e-01
-1.99141547e-01 -9.59637165e-01 2.87855387e-01 4.93532926e-01
8.09620678e-01 3.36576372e-01 9.00477588e-01 -8.22651684e-01
-7.18263686e-01 -3.91854048e-01 -1.47645399e-01 4.42419678e-01
-3.65796983e-02 -7.15989292e-01 -1.25440705e+00 -5.78500688e-01
3.33228022e-01 -3.71360779e-01 1.26425290e+00 5.15866101e-01
8.74165595e-01 1.96648240e-01 -1.11506216e-01 8.63270819e-01
1.26571381e+00 4.04916048e-01 7.02269256e-01 3.32570970e-01
1.10329777e-01 3.65890026e-01 5.37263334e-01 3.95899028e-01
-2.46415868e-01 1.00764358e+00 4.47687894e-01 -8.82390663e-02
3.67497131e-02 6.80067003e-01 4.64558691e-01 3.65887672e-01
1.29162520e-01 -2.54980385e-01 -9.48405743e-01 1.89510986e-01
-1.44623423e+00 -1.25595999e+00 -8.01266357e-02 2.14807034e+00
4.72144693e-01 2.91102052e-01 1.70391854e-02 7.42238760e-01
4.65262353e-01 4.58644807e-01 -3.04581642e-01 -1.33002609e-01
-2.65055776e-01 8.27768564e-01 5.97362578e-01 5.65539956e-01
-1.66308129e+00 5.00106812e-01 6.03421974e+00 1.18923581e+00
-1.51982188e+00 -1.56811714e-01 2.74877816e-01 1.75832175e-02
2.47848943e-01 -2.93425500e-01 -6.37039661e-01 2.51315713e-01
1.17971134e+00 1.67503774e-01 4.88747746e-01 8.86520073e-02
1.48620848e-02 7.02790357e-03 -9.85012770e-01 1.30502152e+00
2.96534687e-01 -1.16368651e+00 -1.35697842e-01 1.44636616e-01
2.21363157e-01 9.14709419e-02 3.94936919e-01 3.43883753e-01
-2.08132282e-01 -1.31450307e+00 6.71622872e-01 5.88074267e-01
7.33210504e-01 -1.00084281e+00 8.71877968e-01 4.05838788e-01
-1.01306605e+00 -3.74846607e-01 1.45993963e-01 4.15670127e-02
3.85385960e-01 8.84391427e-01 -8.90343726e-01 1.00253367e+00
9.34540704e-02 4.60065246e-01 -4.27688420e-01 1.24366331e+00
-3.93076129e-02 7.35899627e-01 -3.20942700e-03 9.62901935e-02
2.50293374e-01 -1.53526336e-01 1.00941443e+00 1.36802065e+00
4.95091081e-01 -2.08895892e-01 2.25848079e-01 3.47236335e-01
1.93331897e-01 -2.45415092e-01 -2.82791346e-01 -2.16648027e-01
2.64128149e-01 1.33628356e+00 -4.99422669e-01 2.35508904e-01
-1.35811806e-01 4.70188588e-01 -4.16928411e-01 8.41041505e-02
-9.64474678e-01 -7.10916460e-01 6.62208915e-01 -1.68734133e-01
2.89354384e-01 -3.46888840e-01 -1.87385276e-01 -7.15596199e-01
-2.83971429e-01 -1.21063411e+00 2.38137394e-01 -2.82053761e-02
-1.14899290e+00 8.44212472e-01 -1.03385687e-01 -1.52810454e+00
-4.85680729e-01 -7.97254264e-01 -3.16331118e-01 8.64716709e-01
-1.90480745e+00 -1.15545845e+00 -1.27021093e-02 4.42517370e-01
3.22555572e-01 -7.90759027e-01 7.95852304e-01 6.05411768e-01
-2.81327724e-01 5.17278135e-01 1.27578601e-01 1.23986751e-01
6.87935174e-01 -1.06409550e+00 5.31406999e-01 7.02722967e-01
3.69459510e-01 2.45308764e-02 9.79850829e-01 -2.34436706e-01
-1.51683652e+00 -1.05421650e+00 5.61222553e-01 -2.63263762e-01
9.81107295e-01 -5.45792758e-01 -4.92353678e-01 2.47229114e-01
6.26827553e-02 -4.46059406e-01 1.00048125e+00 -8.32774565e-02
-9.70904902e-02 -4.34178531e-01 -7.52565980e-01 2.19149292e-01
5.81924438e-01 -6.16272986e-01 -3.97426367e-01 2.45052934e-01
-1.90964136e-02 -1.27690315e-01 -6.69868350e-01 6.22599185e-01
7.48481810e-01 -9.66749072e-01 1.22816122e+00 -1.72613397e-01
-1.91143155e-01 -3.81109327e-01 -3.54802966e-01 -1.58475828e+00
-3.32465529e-01 -8.12915564e-01 -1.68864951e-01 7.54451156e-01
4.39540356e-01 -5.59278548e-01 6.60071433e-01 -7.99796700e-01
-2.57434130e-01 -5.81082344e-01 -1.24761403e+00 -1.03663933e+00
-3.05860519e-01 -1.01723278e+00 2.75813073e-01 7.57805765e-01
-4.64523315e-01 7.57023573e-01 -9.20579255e-01 2.32108518e-01
9.91403639e-01 3.76646101e-01 7.15334117e-01 -1.66900003e+00
-6.98679030e-01 -6.17857933e-01 -9.24270272e-01 -8.45117331e-01
3.15397054e-01 -1.19636202e+00 5.14001064e-02 -1.13185191e+00
-4.05871928e-01 -4.00023311e-01 -3.40408325e-01 1.26470566e-01
4.81021851e-01 6.50231779e-01 5.23827560e-02 -1.83814317e-02
-3.01387817e-01 4.81702983e-01 6.17407739e-01 -5.54901242e-01
1.45253688e-01 7.51869678e-01 -5.01531661e-01 3.60234082e-01
7.76256680e-01 -3.54583770e-01 -1.03651322e-01 -1.33696496e-01
3.20284903e-01 3.06006551e-01 7.37779319e-01 -1.55185771e+00
2.83072352e-01 3.41692492e-02 3.80004704e-01 -6.43604398e-01
7.22944140e-01 -9.04652894e-01 2.10583135e-01 6.73301041e-01
-1.44556388e-01 -1.21273085e-01 2.91384369e-01 4.58192945e-01
-3.60209495e-01 -2.08404362e-02 1.00965893e+00 4.76357222e-01
-8.39965224e-01 1.20034285e-01 -7.34458089e-01 -5.89984376e-03
9.47324336e-01 -7.57352561e-02 -2.75415927e-01 -5.37890434e-01
-5.38442850e-01 9.27714333e-02 -4.62482750e-01 3.76663893e-01
8.79812956e-01 -1.09634125e+00 -1.23976886e+00 4.63654906e-01
1.40708461e-01 -7.60703087e-01 1.21035501e-01 9.74741220e-01
-1.86657026e-01 8.95403028e-01 -3.65333378e-01 -6.63244545e-01
-1.29896665e+00 1.23589605e-01 6.86693132e-01 -3.24202955e-01
-3.88571739e-01 7.24594593e-01 -1.90452978e-01 -2.25749940e-01
3.48222941e-01 4.05143797e-02 -3.64356905e-01 3.30818087e-01
5.50755024e-01 3.62918437e-01 6.11883223e-01 -6.52208924e-01
-5.59973061e-01 5.46498060e-01 8.42857957e-02 -3.54826421e-01
1.35848832e+00 4.62884426e-01 -1.08883262e-01 3.09742689e-01
1.07801735e+00 -2.02518746e-01 -5.08955300e-01 -2.32654542e-01
4.90639120e-01 -2.06438795e-01 6.17772818e-01 -1.11758745e+00
-6.99517310e-01 1.07971168e+00 7.82530010e-01 4.34228092e-01
1.34181535e+00 -3.74798715e-01 8.46971631e-01 7.24245369e-01
5.39736986e-01 -9.08152342e-01 -5.60848787e-02 7.81378388e-01
8.82717669e-01 -9.28400397e-01 1.62413448e-01 1.58073083e-01
-4.32889640e-01 1.12422264e+00 -2.59221733e-01 1.69606134e-02
7.30935812e-01 4.22218651e-01 9.15702954e-02 -2.47158766e-01
-6.87505007e-01 -4.03756857e-01 6.58809423e-01 6.97541177e-01
4.34985369e-01 -1.01957455e-01 1.25191510e-01 4.17879313e-01
-3.99518877e-01 -3.70730996e-01 2.00140119e-01 9.51749980e-01
-7.77301550e-01 -1.36173153e+00 -6.46054029e-01 8.22116673e-01
-6.19539738e-01 -1.84102833e-01 -5.38698971e-01 4.76550937e-01
-9.86235961e-02 9.84249115e-01 -1.03715524e-01 -6.00537002e-01
4.06839520e-01 -8.02550931e-03 6.74880683e-01 -3.27987134e-01
-6.24600887e-01 -1.02304453e-02 2.78879225e-01 -1.39267698e-01
-3.58468592e-01 -5.61329782e-01 -4.16521668e-01 1.82566881e-01
-2.17450947e-01 2.63251215e-01 7.74708569e-01 1.16940272e+00
-4.29267883e-02 1.11080050e+00 9.71804142e-01 -8.15607607e-01
-1.03436792e+00 -7.61043429e-01 -7.76013553e-01 -3.32540244e-01
9.96715724e-01 -6.34384274e-01 -2.56689966e-01 -4.49674964e-01] | [6.524192810058594, 1.5140364170074463] |
6522ef93-b0a7-4230-abb1-63708a9e9a6d | model-driven-deep-learning-for-physical-layer | 1809.06059 | null | http://arxiv.org/abs/1809.06059v2 | http://arxiv.org/pdf/1809.06059v2.pdf | Model-Driven Deep Learning for Physical Layer Communications | Intelligent communication is gradually considered as the mainstream direction
in future wireless communications. As a major branch of machine learning, deep
learning (DL) has been applied in physical layer communications and has
demonstrated an impressive performance improvement in recent years. However,
most of the existing works related to DL focus on data-driven approaches, which
consider the communication system as a black box and train it by using a huge
volume of data. Training a network requires sufficient computing resources and
extensive time, both of which are rarely found in communication devices. By
contrast, model-driven DL approaches combine communication domain knowledge
with DL to reduce the demand for computing resources and training time. This
article reviews the recent advancements in the application of model-driven DL
approaches in physical layer communications, including transmission scheme,
receiver design, and channel information recovery. Several open issues for
further research are also highlighted after presenting the comprehensive
survey. | ['Chao-Kai Wen', 'Hengtao He', 'Zongben Xu', 'Feifei Gao', 'Geoffrey Ye Li', 'Shi Jin'] | 2018-09-17 | null | null | null | null | ['intelligent-communication'] | ['time-series'] | [ 3.37625831e-01 1.64536253e-01 -8.15417767e-01 -3.73738587e-01
-5.61731398e-01 8.29213783e-02 2.15444431e-01 -9.49509367e-02
-2.59223819e-01 1.10506761e+00 -1.74650893e-01 -8.47739398e-01
-2.17001796e-01 -8.33243132e-01 -3.29043627e-01 -9.17382240e-01
-3.94713581e-01 -1.39845116e-02 -2.63353944e-01 -5.30879721e-02
-1.29863366e-01 4.83002573e-01 -5.91293395e-01 1.34537607e-01
6.12824023e-01 1.60104060e+00 2.56585270e-01 7.66404033e-01
-1.33919910e-01 8.91600251e-01 -8.61325920e-01 -2.42851660e-01
4.25994955e-02 -6.90569997e-01 -5.22803783e-01 5.17596379e-02
-4.54984486e-01 -2.36210003e-01 -9.81742084e-01 7.76821971e-01
1.06910419e+00 -4.83751535e-01 3.94548237e-01 -1.11867857e+00
-2.02706218e-01 8.83746207e-01 -4.02575731e-01 1.36630490e-01
4.05997373e-02 -4.44161803e-01 6.50257230e-01 -5.53544223e-01
-6.45066099e-03 1.09234571e+00 7.13490248e-01 4.87198085e-01
-8.70706379e-01 -1.12626803e+00 5.80703206e-02 4.05644953e-01
-1.35507357e+00 -6.01727009e-01 9.07907665e-01 -8.56377333e-02
7.40295947e-01 1.93652567e-02 3.92895997e-01 1.05634761e+00
2.56646127e-01 1.13143516e+00 6.91278040e-01 -7.55973220e-01
4.62233990e-01 1.94332525e-01 -9.21657011e-02 5.43778837e-01
9.61065590e-02 4.70858127e-01 -3.46382022e-01 -5.00829294e-02
7.10564137e-01 -2.05318928e-01 -1.36103749e-01 -2.01558679e-01
-8.35945964e-01 6.80280864e-01 4.79047835e-01 2.31429577e-01
-2.79753745e-01 4.32314366e-01 3.36765289e-01 6.90661252e-01
4.03728843e-01 2.24431127e-01 -5.27304351e-01 -1.49394453e-01
-9.24305975e-01 -1.15124144e-01 1.03177083e+00 1.37401712e+00
4.38698471e-01 3.34921300e-01 -5.57240993e-02 6.94075286e-01
2.93160886e-01 7.03697979e-01 -2.73581706e-02 -6.65446520e-01
7.99677312e-01 1.40439510e-01 -3.13735723e-01 -7.85790920e-01
-8.55249286e-01 -1.10503066e+00 -1.60955036e+00 -1.81439176e-01
-1.82182137e-02 -1.03550792e+00 -4.04250681e-01 1.39562285e+00
-2.39761636e-01 3.14390510e-01 5.25039136e-01 4.99849826e-01
7.44813919e-01 7.70500124e-01 -1.97631553e-01 -4.25710022e-01
6.33118510e-01 -1.00055408e+00 -8.72979820e-01 -1.75416529e-01
9.32242393e-01 -6.59395456e-01 1.58704653e-01 5.66816270e-01
-9.86691058e-01 -5.39026856e-01 -1.50313854e+00 2.96697617e-01
-3.01866755e-02 2.38585562e-01 9.06681895e-01 1.11434460e+00
-5.83660662e-01 2.98362225e-01 -6.04064703e-01 -1.30457908e-01
6.83206975e-01 7.90188968e-01 2.65898079e-01 -1.96063131e-01
-1.34454298e+00 5.12793124e-01 2.67080575e-01 3.16323757e-01
-9.51867163e-01 -5.41027129e-01 -6.07430041e-01 1.46978736e-01
3.90262425e-01 -4.20408607e-01 1.42802489e+00 -6.60238743e-01
-1.86201322e+00 1.23186387e-01 -1.15271859e-01 -6.95556879e-01
3.03817272e-01 -7.45309070e-02 -9.69131827e-01 -2.17592791e-01
-6.25022888e-01 -2.44013853e-02 7.22555101e-01 -1.24457622e+00
-1.03740644e+00 8.42582732e-02 7.99860805e-02 -1.76693782e-01
-4.66996342e-01 -2.99265563e-01 -5.58414102e-01 -6.14327490e-01
1.46836648e-03 -6.70918524e-01 -4.69920993e-01 -6.61814399e-03
-5.64227521e-01 -9.44009516e-03 1.01172543e+00 -3.52939337e-01
1.70701861e+00 -2.07580924e+00 -5.25249518e-04 6.26925707e-01
2.85282135e-01 5.27100980e-01 1.39299646e-01 6.19848847e-01
2.04832062e-01 -3.86381187e-02 6.25789091e-02 -3.07463944e-01
-1.24643542e-01 2.41334170e-01 -1.11405894e-01 4.86541241e-01
-4.13926899e-01 6.70201778e-01 -7.44398057e-01 -1.24845363e-01
3.14120889e-01 3.13186675e-01 -4.99366581e-01 3.33626837e-01
-1.99719891e-01 8.38922024e-01 -6.83888376e-01 3.31742376e-01
8.61949086e-01 -4.08436805e-01 4.16627079e-01 -2.15824723e-01
1.32129192e-02 2.92136878e-01 -8.82788062e-01 1.57914412e+00
-1.00683045e+00 9.04064536e-01 3.06072116e-01 -1.52283299e+00
8.38199377e-01 7.51737416e-01 6.26623452e-01 -1.12048721e+00
4.00891602e-01 3.72551739e-01 2.56472826e-01 -6.30273402e-01
-1.10651828e-01 -4.61090393e-02 -2.59065598e-01 4.53333467e-01
-1.98111668e-01 1.36347577e-01 -2.04531759e-01 4.69835699e-02
1.00445199e+00 -5.23168802e-01 2.80545831e-01 3.19749489e-02
9.08873796e-01 -3.51854265e-01 6.67900443e-01 9.65510070e-01
1.86329335e-01 -2.13809889e-02 2.57240921e-01 -2.54021794e-01
-9.20498431e-01 -7.61410117e-01 -2.20375106e-01 7.67364860e-01
5.23958862e-01 -2.83415943e-01 -5.60037851e-01 -4.60542560e-01
-2.21671581e-01 3.31367463e-01 -6.34295642e-02 -2.89401293e-01
-3.92610282e-01 -1.03012276e+00 6.93933249e-01 4.53383058e-01
1.01120067e+00 -6.73129916e-01 6.88284561e-02 5.30280113e-01
2.18669802e-01 -1.39525378e+00 1.19173534e-01 5.09252846e-01
-5.99470735e-01 -6.05311573e-01 -9.08382535e-01 -1.00301301e+00
5.52719831e-01 1.36648923e-01 7.35354304e-01 2.90374547e-01
-2.15011120e-01 7.26792961e-02 -4.82981592e-01 -6.55690253e-01
-4.20282185e-01 4.32212919e-01 1.37003571e-01 -3.62391844e-02
7.20428452e-02 -5.78660071e-01 -4.34529811e-01 1.76493362e-01
-3.08165163e-01 1.03655890e-01 1.08265996e+00 7.92504489e-01
5.93543313e-02 8.29931200e-01 6.87785566e-01 -9.27826941e-01
7.48183012e-01 -6.60780072e-01 -5.35788059e-01 2.37779170e-01
-3.74061078e-01 1.04370318e-01 8.93438280e-01 -1.55083150e-01
-1.09324408e+00 -3.67304355e-01 -5.84240079e-01 4.00633216e-01
1.78843364e-01 6.36668444e-01 -4.34828669e-01 -6.36128128e-01
3.93552810e-01 9.59516615e-02 -1.75859839e-01 -5.34280181e-01
-1.18101895e-01 1.27945256e+00 1.42773613e-02 -5.06127477e-01
9.01814938e-01 3.05859596e-01 2.74120480e-01 -1.25054967e+00
-1.27905738e+00 -2.97846854e-01 -5.68293691e-01 -1.97516717e-02
3.52192551e-01 -9.15795743e-01 -9.16857600e-01 5.08494318e-01
-1.08509099e+00 -6.85382307e-01 2.49698818e-01 9.12656963e-01
-2.08564237e-01 2.50654012e-01 -4.64293361e-01 -8.67032170e-01
-3.06107908e-01 -1.04514492e+00 5.58134854e-01 1.12101622e-01
2.24795476e-01 -1.43642807e+00 -3.91707957e-01 2.56278496e-02
7.48357892e-01 -6.05669357e-02 1.22674406e+00 -2.10772604e-01
-5.47054529e-01 -3.12879801e-01 -4.92298096e-01 4.65186208e-01
2.08177760e-01 -7.30884671e-01 -1.02778280e+00 -5.55997074e-01
-5.99969476e-02 -1.50757924e-01 5.75881600e-01 5.24694741e-01
1.72206950e+00 -1.52487755e-01 -8.65271151e-01 8.90472233e-01
1.34996593e+00 5.83320975e-01 3.93825412e-01 -1.79895505e-01
3.68501991e-01 -9.33972299e-02 2.46142060e-01 7.78030574e-01
3.18709463e-01 6.89830542e-01 3.51183265e-01 -7.81943679e-01
-9.89999771e-02 1.83193968e-03 1.05294257e-01 7.50692964e-01
1.53527901e-01 -7.61006415e-01 -5.88161826e-01 -1.27421558e-01
-1.66989553e+00 -7.84253240e-01 1.16124943e-01 1.90335035e+00
5.52084327e-01 4.77456421e-01 -1.47206590e-01 6.86188757e-01
4.16759789e-01 2.77659446e-02 -6.65532053e-01 -1.32738099e-01
4.21645232e-02 1.73718795e-01 8.34546983e-01 6.40719235e-01
-1.08219957e+00 6.55839086e-01 7.07677746e+00 1.03094709e+00
-1.43515849e+00 -1.34479059e-02 5.31713068e-01 2.91065574e-01
6.33597076e-02 -2.59268641e-01 -8.39993179e-01 4.68978524e-01
9.88580585e-01 -5.43170981e-02 2.94156373e-01 7.18644202e-01
5.43650329e-01 -5.95144592e-02 -1.25100827e+00 1.35516500e+00
-2.16611385e-01 -1.50450373e+00 -2.08642140e-01 2.92409092e-01
6.22458398e-01 -6.40806556e-02 1.70504659e-01 4.33546782e-01
-4.07917751e-03 -1.12872744e+00 3.05265069e-01 3.55836868e-01
7.42004156e-01 -8.80664289e-01 1.12617683e+00 6.06477439e-01
-1.20477533e+00 -4.29816484e-01 -3.15446049e-01 -5.84217072e-01
3.63690674e-01 9.13112700e-01 -6.17048204e-01 7.50360668e-01
3.39328468e-01 8.99064660e-01 -9.26864073e-02 1.32054651e+00
-9.45804492e-02 9.18834746e-01 -1.81219801e-01 -2.14734152e-01
2.52961665e-01 2.62957186e-01 1.03048131e-01 1.24270904e+00
3.61400038e-01 4.47808914e-02 6.00804508e-01 1.67542398e-01
-2.77465641e-01 -3.57884437e-01 -2.89340019e-01 1.86261460e-01
7.54141808e-01 7.95428872e-01 2.78760977e-02 -1.96071148e-01
-8.56549323e-01 5.71510196e-01 -2.59285450e-01 5.20434856e-01
-7.06199884e-01 -5.46750963e-01 5.28405547e-01 -1.92531586e-01
-5.37503185e-03 -6.56309128e-01 -4.24160659e-01 -6.10361457e-01
-2.13125288e-01 -5.91807306e-01 6.98682144e-02 -5.72781190e-02
-1.06250906e+00 3.74398738e-01 -2.42480338e-01 -1.28785825e+00
6.48022890e-02 -7.13077188e-01 -3.28531057e-01 6.32409573e-01
-2.06260824e+00 -8.53229582e-01 -4.21058774e-01 5.10235667e-01
4.65594530e-01 -6.17089093e-01 1.02325237e+00 8.11734676e-01
-7.86121428e-01 1.03574824e+00 5.77317834e-01 4.45954174e-01
2.17900902e-01 -7.00253963e-01 1.22652575e-01 3.81480873e-01
-1.94066554e-01 3.12328011e-01 4.68010992e-01 -1.50585203e-02
-1.51428056e+00 -9.76317942e-01 6.95923626e-01 4.52724040e-01
4.44090158e-01 -6.07446909e-01 -3.32189023e-01 3.18910867e-01
2.21710980e-01 9.11890343e-02 9.49436486e-01 1.43529028e-01
2.43477210e-01 -5.36668003e-01 -8.58127296e-01 4.54741776e-01
9.93703604e-01 -1.98063836e-01 5.07281125e-01 4.41282451e-01
2.53813505e-01 -4.62364614e-01 -9.13445234e-01 4.19403255e-01
4.75077003e-01 -6.77576900e-01 8.53568614e-01 -5.69063686e-02
-2.47752607e-01 5.89359142e-02 -1.63686097e-01 -1.14869177e+00
-9.81943160e-02 -1.03975248e+00 -4.36843365e-01 8.46507013e-01
7.00441301e-01 -3.21448177e-01 1.10203946e+00 1.02208359e-02
-1.91371188e-01 -1.04573143e+00 -6.83673382e-01 -8.90849054e-01
1.48408599e-02 -8.05983543e-01 5.49763560e-01 2.60072798e-01
2.12935582e-01 5.41341245e-01 -7.73190975e-01 3.00752372e-01
6.46043062e-01 -1.76621914e-01 8.38645518e-01 -1.27386534e+00
-3.74088705e-01 -2.10475400e-01 -2.74123430e-01 -1.99317062e+00
2.33858541e-01 -8.29966605e-01 -1.16174802e-01 -1.52590907e+00
-1.48319066e-01 -1.18206906e+00 -2.06464410e-01 7.27858394e-02
4.97090548e-01 -2.75476053e-02 -1.51120543e-01 -3.08105703e-02
-7.07113266e-01 6.20360374e-01 1.26277721e+00 -3.08597594e-01
-1.39765173e-01 8.92461181e-01 -6.83361590e-01 6.72634661e-01
1.11983716e+00 -8.32352862e-02 -5.07680058e-01 -6.97915673e-01
2.48478904e-01 3.12124282e-01 -7.09140673e-02 -1.45684958e+00
4.86307293e-01 -7.76149631e-02 4.59723920e-01 -3.24163467e-01
5.05189180e-01 -1.36382985e+00 -3.25653046e-01 7.61426926e-01
-3.65013838e-01 -5.87653935e-01 -4.98427823e-02 6.42736912e-01
-2.07468018e-01 7.32078850e-02 9.44455087e-01 1.69477105e-01
-1.01992023e+00 6.00781322e-01 -6.86841488e-01 -1.90121219e-01
1.09460211e+00 -4.51406687e-02 1.14912622e-01 -7.67526031e-01
-7.50972807e-01 3.63115937e-01 -4.60330933e-01 2.48552904e-01
5.44402719e-01 -1.13681078e+00 -4.01193351e-01 2.98342973e-01
-9.21427608e-02 -1.68752581e-01 7.08444342e-02 7.90698647e-01
-3.27954531e-01 9.27466750e-01 3.19902480e-01 -5.18816590e-01
-1.07715368e+00 3.23979974e-01 6.79014802e-01 -2.79575616e-01
-6.13858342e-01 9.24747944e-01 -8.01781863e-02 -6.97397888e-02
1.04926372e+00 -1.13634162e-01 -2.05798745e-02 -6.18626118e-01
5.04236996e-01 3.22200239e-01 5.91640249e-02 -1.09502479e-01
-2.79071778e-01 5.33310592e-01 -5.18660285e-02 1.73319772e-01
1.07142997e+00 -5.31960487e-01 2.38449037e-01 1.50386721e-01
1.38647532e+00 -2.19962552e-01 -1.12995434e+00 -8.87659490e-01
2.16057658e-01 -1.38973491e-02 5.74242949e-01 -7.23077834e-01
-1.29501140e+00 1.15507925e+00 7.02652395e-01 1.12368681e-01
1.27633345e+00 -1.11613736e-01 1.14299297e+00 7.97518969e-01
9.18745220e-01 -1.26360834e+00 1.26055926e-01 7.25865245e-01
2.91205853e-01 -1.27835095e+00 1.83781698e-01 -5.74055731e-01
5.95220774e-02 1.23954737e+00 2.38571033e-01 2.68936425e-01
1.47760749e+00 8.29240263e-01 1.93987973e-02 9.40001532e-02
-5.10655105e-01 -1.59435943e-01 -9.01753083e-02 7.71759391e-01
6.62148952e-01 2.18638964e-02 -1.78554684e-01 6.17448986e-01
-2.06505686e-01 1.36058116e-02 1.76464871e-01 1.01847637e+00
-5.77916980e-01 -1.65975714e+00 -3.21003683e-02 5.59436619e-01
-5.13655245e-01 -1.55653656e-01 -1.86890289e-02 5.62938392e-01
2.23620445e-01 1.42172432e+00 -2.56896466e-01 -5.33426464e-01
1.44277170e-01 -8.47602069e-01 4.02942657e-01 -6.33790195e-01
7.78730661e-02 -2.24788412e-01 2.43579090e-01 -1.75828591e-01
-4.44182187e-01 -5.79834878e-02 -1.44921708e+00 -5.61063766e-01
-3.31310540e-01 3.57203573e-01 7.60053992e-01 1.38320172e+00
2.59228617e-01 9.67176020e-01 1.03404820e+00 -8.52838755e-01
-1.96345165e-01 -7.01837778e-01 -7.68156648e-01 -4.75530654e-01
9.67995465e-01 -5.88939309e-01 -1.02308296e-01 -2.11818710e-01] | [6.30243444442749, 1.4449907541275024] |
b0f9252b-e4df-46c2-9b72-bd4dcc66ce1a | time-reversed-diffusion-tensor-transformer-a | 2210.16897 | null | https://arxiv.org/abs/2210.16897v1 | https://arxiv.org/pdf/2210.16897v1.pdf | Time-rEversed diffusioN tEnsor Transformer: A new TENET of Few-Shot Object Detection | In this paper, we tackle the challenging problem of Few-shot Object Detection. Existing FSOD pipelines (i) use average-pooled representations that result in information loss; and/or (ii) discard position information that can help detect object instances. Consequently, such pipelines are sensitive to large intra-class appearance and geometric variations between support and query images. To address these drawbacks, we propose a Time-rEversed diffusioN tEnsor Transformer (TENET), which i) forms high-order tensor representations that capture multi-way feature occurrences that are highly discriminative, and ii) uses a transformer that dynamically extracts correlations between the query image and the entire support set, instead of a single average-pooled support embedding. We also propose a Transformer Relation Head (TRH), equipped with higher-order representations, which encodes correlations between query regions and the entire support set, while being sensitive to the positional variability of object instances. Our model achieves state-of-the-art results on PASCAL VOC, FSOD, and COCO. | ['Piotr Koniusz', 'Lei Wang', 'Naila Murray', 'Shan Zhang'] | 2022-10-30 | null | null | null | null | ['few-shot-object-detection'] | ['computer-vision'] | [ 6.44014999e-02 -4.79491532e-01 7.79702421e-03 -2.95901984e-01
-6.23595417e-01 -6.33144975e-01 6.22905135e-01 1.62429839e-01
-3.20262790e-01 -9.53058824e-02 -9.52639803e-02 4.66593713e-01
-2.49447629e-01 -6.92650080e-01 -5.51353514e-01 -6.72413170e-01
-2.85432577e-01 4.07186240e-01 1.08200157e+00 -9.67239738e-02
2.40797594e-01 7.07162559e-01 -1.88703763e+00 5.79594791e-01
6.51983619e-01 1.45983768e+00 2.92932093e-01 4.56301868e-01
-1.71587437e-01 8.66909623e-01 -5.70383728e-01 -4.82043386e-01
2.49805361e-01 -2.16993764e-02 -7.16283619e-01 2.33600676e-01
1.09658146e+00 -3.53140116e-01 -6.24563038e-01 9.68065560e-01
3.41363311e-01 3.61660831e-02 6.38315260e-01 -1.28125274e+00
-5.87504804e-01 9.04436186e-02 -7.04916120e-01 6.50465012e-01
5.79601452e-02 4.01477039e-01 1.19740248e+00 -1.20761120e+00
8.33070636e-01 1.20529163e+00 5.63365459e-01 1.89294487e-01
-1.33678067e+00 -2.76317269e-01 3.76400769e-01 5.69339693e-01
-1.56743801e+00 -4.58763808e-01 9.19642806e-01 -7.47857213e-01
1.08751726e+00 2.88645595e-01 7.47255325e-01 9.79564130e-01
2.40992919e-01 1.11934960e+00 6.94880188e-01 8.08653608e-02
2.28074387e-01 -7.59178102e-02 4.10678148e-01 6.92805469e-01
2.08327234e-01 -1.46557041e-03 -7.87855685e-01 -1.46258742e-01
3.16905379e-01 3.08319122e-01 -1.99001759e-01 -7.84557760e-01
-1.19906557e+00 6.06262982e-01 6.45195961e-01 3.21995825e-01
-4.82089728e-01 9.44001526e-02 4.58676279e-01 -1.05545972e-03
3.19449157e-01 2.40498751e-01 -2.05344349e-01 -1.04135565e-01
-9.56849337e-01 3.58162284e-01 4.84987617e-01 9.88864839e-01
9.70479608e-01 -2.41216078e-01 -7.32260048e-01 8.83293867e-01
5.14489710e-02 2.83545762e-01 2.75070906e-01 -6.87272310e-01
3.73261541e-01 1.01682055e+00 -9.35251266e-02 -1.31338739e+00
-2.56280601e-01 -5.78230023e-01 -3.55619341e-01 -2.08737895e-01
2.40889952e-01 5.58509290e-01 -9.38937962e-01 1.38904524e+00
5.13357520e-01 5.73476069e-02 -2.57046461e-01 1.12731743e+00
8.55409265e-01 3.63380760e-01 -2.25350484e-01 5.64587303e-02
1.56612241e+00 -1.01504540e+00 -5.84575176e-01 -4.11592960e-01
5.53856552e-01 -7.08318174e-01 1.07214296e+00 8.20716321e-02
-9.90867198e-01 -4.51582253e-01 -1.04440308e+00 -3.41076016e-01
-5.13655782e-01 2.34368607e-01 5.49323440e-01 2.06428558e-01
-7.61581838e-01 5.89921772e-01 -8.68513107e-01 -2.52388805e-01
4.95670289e-01 1.25506803e-01 -2.69258887e-01 -3.60300928e-01
-7.17954099e-01 6.21963620e-01 1.91717833e-01 1.50345579e-01
-1.00900686e+00 -7.44938195e-01 -8.72786880e-01 2.22931057e-01
7.65186191e-01 -2.76856184e-01 8.58864784e-01 -5.00501454e-01
-1.09073627e+00 7.31484771e-01 -2.74554342e-01 -1.26984164e-01
3.70639205e-01 -3.85749996e-01 -1.76040977e-01 4.22118664e-01
3.19741517e-01 4.57092732e-01 8.86462331e-01 -9.65081275e-01
-4.90098268e-01 -8.23121548e-01 6.39009625e-02 4.63931449e-02
-4.93640274e-01 7.75789469e-02 -8.92279506e-01 -6.59723461e-01
3.85854959e-01 -9.50959444e-01 -1.79451592e-02 3.45717371e-01
-4.71039265e-01 -4.14931029e-01 1.24296379e+00 -3.55431914e-01
1.28819859e+00 -2.46005869e+00 3.17381680e-01 -3.30683105e-02
3.47707331e-01 2.98031479e-01 -3.52059901e-01 4.18988049e-01
3.58302921e-01 -1.92404062e-01 -1.03943825e-01 -3.76256317e-01
-5.24381287e-02 3.15378666e-01 -2.73295879e-01 5.07552981e-01
5.70650578e-01 8.81594479e-01 -9.55502748e-01 -6.69227481e-01
1.32452667e-01 4.57050294e-01 -4.62814718e-01 1.85735181e-01
-2.46788293e-01 -4.48585004e-02 -4.90686029e-01 9.27861214e-01
7.46867537e-01 -4.14916217e-01 -5.00746220e-02 -7.96912551e-01
-2.89157242e-01 2.58227706e-01 -1.18130803e+00 1.71782696e+00
1.14476159e-01 5.57793736e-01 -1.53481841e-01 -7.59824038e-01
8.41646612e-01 -1.88071117e-01 5.83541095e-01 -6.44974053e-01
9.35967937e-02 7.97220692e-02 -1.81586474e-01 -5.08059680e-01
6.99333310e-01 4.87728953e-01 2.04042032e-01 2.50201494e-01
2.87600249e-01 8.33356306e-02 5.12980580e-01 4.09493357e-01
1.32673907e+00 5.74279241e-02 -4.08678390e-02 -1.84078798e-01
3.39670449e-01 -1.59826472e-01 7.32565463e-01 7.30511606e-01
-3.18919450e-01 8.24711978e-01 7.78496861e-01 -5.35036504e-01
-6.54213190e-01 -1.14808953e+00 -1.84055761e-01 1.10798180e+00
4.88468289e-01 -6.57793045e-01 -3.28374773e-01 -9.50692892e-01
2.63572723e-01 4.19191718e-01 -7.78326690e-01 -2.96203941e-01
-5.43070614e-01 -5.62317789e-01 3.13070893e-01 6.33855760e-01
3.07314336e-01 -5.51174879e-01 -8.77132297e-01 2.37730905e-01
-4.54304740e-02 -1.33807111e+00 -7.75452912e-01 1.64044619e-01
-7.36066997e-01 -1.29858947e+00 -5.83018005e-01 -4.27838087e-01
5.61292410e-01 7.51719356e-01 1.06009376e+00 -7.38539025e-02
-7.58817136e-01 5.08790851e-01 -4.88833696e-01 -7.65487701e-02
3.36033851e-01 -1.53354034e-01 -1.61642492e-01 4.10813689e-01
5.37918508e-01 -2.71155119e-01 -7.03407228e-01 6.67749584e-01
-8.94603014e-01 -2.91829914e-01 4.92169529e-01 8.86960208e-01
8.34427774e-01 -2.86037177e-01 -2.28824746e-02 -5.25456131e-01
8.33287016e-02 -2.14195549e-01 -6.67107999e-01 4.67104077e-01
-3.94778371e-01 7.48254806e-02 2.59247512e-01 -6.03847027e-01
-7.87848651e-01 1.08078994e-01 4.57761556e-01 -9.18690503e-01
3.46358150e-01 1.86223507e-01 4.83523272e-02 -1.42717659e-01
6.41482413e-01 3.96168292e-01 -1.66152924e-01 -7.05299377e-01
2.53778428e-01 3.38662505e-01 3.53180736e-01 -5.52834511e-01
6.43853188e-01 8.39965582e-01 5.09446021e-03 -9.70785975e-01
-9.95225191e-01 -8.46086025e-01 -9.88647461e-01 -1.99161336e-01
7.27491200e-01 -7.85115242e-01 -3.81867051e-01 5.89100420e-01
-1.19821954e+00 3.53723541e-02 -4.32723612e-01 4.25303549e-01
-2.67683148e-01 4.04132843e-01 -7.19308913e-01 -6.62644148e-01
-1.02802858e-01 -1.23406231e+00 1.52395248e+00 1.14053287e-01
1.91322848e-01 -3.69623572e-01 -4.52255737e-03 2.48281509e-01
2.32439771e-01 1.97680101e-01 6.07339084e-01 -4.33599889e-01
-9.80521560e-01 -3.81619185e-01 -4.37679172e-01 2.49004543e-01
-1.48186490e-01 2.93775171e-01 -9.58882928e-01 -4.75847721e-01
-2.78189957e-01 -3.33217144e-01 1.14535642e+00 2.01981068e-01
1.11200583e+00 -1.71596229e-01 -4.57830131e-01 6.55478418e-01
1.11173034e+00 -2.45636970e-01 6.48477435e-01 6.43541664e-02
7.82652617e-01 7.63114333e-01 8.20384264e-01 4.67496186e-01
4.63322818e-01 1.13608575e+00 4.81413037e-01 1.87253505e-01
-2.55256802e-01 -1.35747343e-01 3.65564257e-01 6.46917284e-01
1.56346895e-02 1.40680419e-02 -6.60770118e-01 7.61920869e-01
-1.80090082e+00 -8.97875965e-01 -1.85534850e-01 2.03976536e+00
4.11259234e-01 1.05402529e-01 1.75449386e-01 -7.81717226e-02
4.59584415e-01 5.35749614e-01 -6.82883203e-01 1.67328060e-01
-2.47772157e-01 -9.11729187e-02 3.81095052e-01 -8.57806057e-02
-1.08395898e+00 9.45415676e-01 5.89983559e+00 7.26605237e-01
-1.30691147e+00 2.02786922e-01 5.15869744e-02 -4.14774001e-01
-8.00027102e-02 -4.16484959e-02 -9.09356117e-01 4.25519586e-01
3.08120608e-01 3.60256769e-02 2.29368702e-01 1.05631530e+00
-2.89651811e-01 -1.35303602e-01 -1.16699231e+00 9.05534387e-01
2.34860227e-01 -1.14609683e+00 8.32689460e-03 2.94022281e-02
4.80092883e-01 2.07947344e-01 1.28175974e-01 2.48634651e-01
-2.21423358e-01 -4.93075520e-01 8.98168206e-01 4.26229596e-01
5.68890810e-01 -3.94394249e-01 5.99873841e-01 1.05791822e-01
-1.50216591e+00 -2.32065871e-01 -7.21707284e-01 3.67744654e-01
-9.54237804e-02 7.95019388e-01 -6.82627678e-01 4.83595699e-01
1.13684368e+00 9.71117198e-01 -1.01450992e+00 1.13238072e+00
5.32841124e-03 2.57412970e-01 -3.24100375e-01 5.53071685e-02
2.00195566e-01 4.18576179e-03 8.92409325e-01 1.25679350e+00
1.33298114e-01 -1.86340928e-01 2.53337741e-01 9.69282866e-01
2.17379257e-01 9.67011414e-03 -4.64050442e-01 -1.72500864e-01
3.74286503e-01 1.30591595e+00 -7.59985864e-01 -2.61641026e-01
-4.44638103e-01 1.10344350e+00 5.39665103e-01 3.87797624e-01
-6.01915479e-01 -2.65672237e-01 8.76281321e-01 2.21106827e-01
7.71084070e-01 -3.31807643e-01 6.36615306e-02 -1.38239193e+00
5.60705423e-01 -5.36053896e-01 5.74187934e-01 -5.44211447e-01
-1.34860706e+00 5.69823503e-01 -2.51200814e-02 -1.32086980e+00
1.96482152e-01 -6.44278526e-01 -5.07605076e-01 3.95911306e-01
-1.44576538e+00 -1.31299436e+00 -4.13887411e-01 5.11948943e-01
5.38304210e-01 -1.14812125e-02 5.52520871e-01 4.98184711e-01
-8.84496391e-01 6.04089200e-01 -1.27765223e-01 5.76548614e-02
5.26475668e-01 -1.03252351e+00 8.90187323e-02 8.55681717e-01
1.89924002e-01 7.49229789e-01 3.99021119e-01 -5.25408804e-01
-1.69111371e+00 -1.28786922e+00 6.50426090e-01 -5.16793013e-01
5.60464680e-01 -3.99128497e-01 -1.11589682e+00 5.48895776e-01
-4.46700603e-01 7.58856535e-01 4.78307188e-01 2.34877765e-01
-9.38476622e-01 -4.88025367e-01 -1.01294780e+00 4.02829051e-01
1.18423140e+00 -6.95970356e-01 -4.84841406e-01 3.14986289e-01
6.14904583e-01 -2.82239884e-01 -7.52421558e-01 5.56534410e-01
6.66310191e-01 -1.23772776e+00 1.04166114e+00 -5.76323390e-01
2.35838532e-01 -6.43943250e-01 -3.65309387e-01 -8.59478056e-01
-5.04272103e-01 -2.87808269e-01 -6.08123899e-01 1.10151005e+00
7.19124824e-02 -2.46391565e-01 5.77969968e-01 3.13151300e-01
-1.58674672e-01 -1.01888227e+00 -1.11701763e+00 -9.75041866e-01
-5.28778315e-01 -2.99990058e-01 5.31701803e-01 8.20085526e-01
-3.02462339e-01 2.05944300e-01 -2.50036776e-01 2.52386004e-01
7.28026330e-01 2.62511998e-01 6.84101820e-01 -1.16327441e+00
-2.43531436e-01 -3.68388325e-01 -8.94006729e-01 -1.13053453e+00
-2.17525125e-01 -6.79785430e-01 9.38573256e-02 -1.21890378e+00
4.02034789e-01 -3.59266877e-01 -4.97388780e-01 4.68687952e-01
-1.56140670e-01 3.62902105e-01 2.12123781e-01 4.62497205e-01
-1.00365734e+00 8.27576578e-01 1.13802922e+00 -3.75073373e-01
-8.33055843e-03 -3.38359147e-01 -2.47125432e-01 5.46387434e-01
1.36396408e-01 -5.77163935e-01 -2.52756447e-01 -5.80405772e-01
2.12480687e-02 -2.73961484e-01 7.20836818e-01 -9.77837503e-01
2.39848539e-01 -1.77871525e-01 4.50983405e-01 -1.04773533e+00
5.72111785e-01 -6.58009529e-01 -1.89445570e-01 3.90381217e-01
-1.20730557e-01 -6.47719502e-02 -5.71279787e-02 7.01645494e-01
-2.86048681e-01 -1.05558909e-01 9.16089594e-01 2.74668466e-02
-8.46688688e-01 5.16303301e-01 7.00396299e-02 8.94617066e-02
1.12879705e+00 -2.17984855e-01 -4.68000799e-01 1.96113363e-01
-2.77022034e-01 3.79713118e-01 5.43569803e-01 6.92504644e-01
7.12667108e-01 -1.37992930e+00 -4.72743958e-01 2.02305242e-01
7.27179706e-01 1.19994901e-01 4.93683964e-01 9.52212989e-01
-2.31775150e-01 1.43548474e-01 -2.24785469e-02 -1.11632097e+00
-1.24159050e+00 6.09347582e-01 3.22228998e-01 -1.47241950e-01
-8.25940073e-01 1.16149485e+00 3.66579413e-01 -1.76901683e-01
1.78096175e-01 -1.80840552e-01 -2.30921134e-01 4.78959501e-01
8.03491354e-01 4.86890197e-01 8.34510326e-02 -8.26168358e-01
-6.11840606e-01 8.23921561e-01 -4.18628156e-01 3.29871833e-01
1.33673584e+00 2.39583272e-02 -1.97016835e-01 4.86614436e-01
1.33028138e+00 -1.84516639e-01 -1.43786705e+00 -6.85596049e-01
1.80219993e-01 -9.93049681e-01 1.97687209e-01 -4.26204145e-01
-1.22528780e+00 9.69886184e-01 7.12959409e-01 7.39161894e-02
9.18494165e-01 2.90027261e-01 7.19675481e-01 3.20782393e-01
2.94832468e-01 -1.17177200e+00 6.03345752e-01 4.83610690e-01
9.99643207e-01 -1.03127611e+00 8.99179578e-02 -5.49498260e-01
-7.07492232e-01 1.19687009e+00 7.52399504e-01 -2.83807572e-02
4.70692068e-01 4.82567325e-02 -2.85285145e-01 -6.80116355e-01
-8.31370831e-01 -4.34191018e-01 6.92414761e-01 5.46979904e-01
2.46442109e-01 -7.86723644e-02 -1.13642886e-01 3.69199038e-01
2.14422673e-01 -1.12540975e-01 1.61608621e-01 1.01208949e+00
-4.35457855e-01 -7.04359531e-01 -2.17084110e-01 6.03512228e-01
-1.93148136e-01 1.52532712e-01 -4.72049862e-01 3.94121587e-01
2.16572031e-01 7.54603803e-01 2.30956450e-01 -5.82652092e-01
5.92035770e-01 -1.12133570e-01 4.64632392e-01 -6.06511056e-01
-3.47956330e-01 3.07665281e-02 -1.44847080e-01 -9.98573184e-01
-1.93963259e-01 -7.81224012e-01 -8.14791441e-01 7.32316077e-02
-4.48626876e-01 -1.65251628e-01 4.92382824e-01 6.60727143e-01
6.55909657e-01 5.29166937e-01 5.66515207e-01 -9.17080343e-01
-7.55149126e-01 -9.24217641e-01 -6.90535903e-01 4.91461486e-01
3.78014266e-01 -1.07128572e+00 -3.36741090e-01 -2.40976080e-01] | [9.40718936920166, 0.9685880541801453] |
5afbcb0a-0658-4eb7-9222-bab25063a867 | is-a-pet-all-you-need-a-multi-modal-study-for | 2207.02094 | null | https://arxiv.org/abs/2207.02094v1 | https://arxiv.org/pdf/2207.02094v1.pdf | Is a PET all you need? A multi-modal study for Alzheimer's disease using 3D CNNs | Alzheimer's Disease (AD) is the most common form of dementia and often difficult to diagnose due to the multifactorial etiology of dementia. Recent works on neuroimaging-based computer-aided diagnosis with deep neural networks (DNNs) showed that fusing structural magnetic resonance images (sMRI) and fluorodeoxyglucose positron emission tomography (FDG-PET) leads to improved accuracy in a study population of healthy controls and subjects with AD. However, this result conflicts with the established clinical knowledge that FDG-PET better captures AD-specific pathologies than sMRI. Therefore, we propose a framework for the systematic evaluation of multi-modal DNNs and critically re-evaluate single- and multi-modal DNNs based on FDG-PET and sMRI for binary healthy vs. AD, and three-way healthy/mild cognitive impairment/AD classification. Our experiments demonstrate that a single-modality network using FDG-PET performs better than MRI (accuracy 0.91 vs 0.87) and does not show improvement when combined. This conforms with the established clinical knowledge on AD biomarkers, but raises questions about the true benefit of multi-modal DNNs. We argue that future work on multi-modal fusion should systematically assess the contribution of individual modalities following our proposed evaluation framework. Finally, we encourage the community to go beyond healthy vs. AD classification and focus on differential diagnosis of dementia, where fusing multi-modal image information conforms with a clinical need. | ['Christian Wachinger', 'Igor Yakushev', 'Aldana Lizarraga', 'Sebastian Pölsterl', 'Ignacio Sarasua', 'Marla Narazani'] | 2022-07-05 | null | null | null | null | ['clinical-knowledge'] | ['miscellaneous'] | [ 8.5127898e-02 -2.4631521e-01 -2.9800743e-01 -3.9773268e-01
-6.2191308e-01 -3.4697267e-01 5.1654935e-01 -1.1261822e-01
-8.7291658e-01 1.1246879e+00 4.6353388e-01 -3.4983775e-01
-4.0188140e-01 -8.7090886e-01 -7.0212975e-02 -5.7902050e-01
-3.3386078e-01 1.0248049e+00 3.1341752e-01 2.7523133e-01
-4.7762915e-01 5.9351808e-01 -1.3385745e+00 3.9882112e-01
7.5921685e-01 9.5392489e-01 3.0814144e-01 4.3392944e-01
1.1463249e-01 4.8739842e-01 -2.0142031e-01 -1.6443391e-01
3.8119573e-02 -4.8297912e-01 -7.7288580e-01 -1.7131132e-01
4.7525546e-01 -9.5527500e-01 -2.7759457e-01 1.0636874e+00
8.9964676e-01 -1.9319576e-01 8.5059094e-01 -9.9726272e-01
-7.6077867e-01 2.5375342e-01 -1.4856839e-01 1.0564405e+00
4.5327730e-02 4.1790771e-01 5.7375085e-01 -5.6774169e-01
6.0187936e-01 1.2564495e+00 8.7113541e-01 6.3760853e-01
-9.7185779e-01 -3.0855870e-01 3.5331666e-01 8.7324393e-01
-1.0379924e+00 -2.0838678e-01 1.8865889e-01 -6.2406284e-01
9.7539639e-01 2.1776947e-01 1.3492799e+00 1.3526182e+00
4.1300958e-01 2.8952345e-01 1.4051437e+00 -9.9438779e-02
4.3560842e-01 -3.3132336e-01 4.8918533e-01 4.7827399e-01
7.6880091e-01 2.9354697e-01 3.4357116e-01 -6.0181582e-01
8.4811795e-01 2.5285104e-01 -4.0205353e-01 -1.6664292e-01
-1.3281474e+00 8.5406464e-01 4.2284402e-01 8.4563279e-01
-7.1002841e-01 4.8183300e-02 6.4406180e-01 -5.9781689e-03
2.2644053e-01 -4.5458570e-02 -3.4718373e-01 1.1823327e-01
-1.2980208e+00 3.4905735e-01 3.1220856e-01 4.5297351e-02
1.2805593e-01 -1.0394384e-01 3.1172683e-02 1.0513506e+00
8.0499905e-01 6.2308550e-01 1.0092561e+00 -1.2444413e+00
8.1103370e-02 4.6030256e-01 -1.6294904e-01 -3.6538035e-01
-8.9014423e-01 -6.6240066e-01 -1.1051658e+00 3.9016774e-01
4.9229285e-01 -2.1487497e-01 -8.4619367e-01 1.6000391e+00
-7.9349987e-02 -1.1775713e-01 -2.0951506e-01 1.3976592e+00
6.2402368e-01 -3.4074432e-01 5.5922729e-01 -1.8726380e-01
1.9521147e+00 -5.7484317e-01 -6.3688135e-01 -3.5262343e-01
7.3805732e-01 1.1126803e-01 6.9569594e-01 1.7156966e-01
-9.1581774e-01 -1.7379294e-01 -1.0421278e+00 2.7768850e-02
-3.5740137e-01 1.0848549e-01 4.9852931e-01 8.9254957e-01
-1.3943774e+00 2.7784896e-01 -1.1160488e+00 -6.5372139e-01
7.7430034e-01 5.2065051e-01 -5.6815404e-01 -3.4361824e-01
-1.3821456e+00 1.5265732e+00 2.3670542e-01 -1.0158860e-01
-1.0734150e+00 -9.5621955e-01 -2.7384156e-01 -4.2222312e-01
-1.5919605e-01 -1.5731730e+00 1.1545390e+00 -8.6889482e-01
-7.9654598e-01 1.0118561e+00 -9.5368460e-02 -7.7227908e-01
8.3454216e-01 -1.3838321e-01 -6.4701194e-01 7.7047533e-01
3.2523191e-01 1.0519488e+00 5.0027978e-02 -8.3037215e-01
-5.9192437e-01 -1.2312541e+00 -8.5352911e-03 8.3323218e-02
-1.7788868e-02 2.9273130e-02 7.0372027e-01 -3.4039620e-01
5.1540498e-02 -9.3715650e-01 -3.4002936e-01 4.3045813e-01
-6.0328931e-02 -1.5239061e-01 9.0998000e-01 -9.0753758e-01
7.4735183e-01 -1.6340634e+00 -2.7617362e-01 -2.0327052e-01
9.0049499e-01 1.8864343e-01 7.6659247e-02 -5.2525526e-01
-4.4294035e-01 2.4610947e-01 -3.1242037e-01 2.5454989e-02
-4.3946397e-03 2.3631011e-01 5.1372999e-01 7.6863647e-01
-3.9695341e-02 1.1957844e+00 -7.6375365e-01 -3.1841907e-01
2.7506143e-01 7.5945735e-01 -2.8013179e-01 -4.2777529e-01
1.1556418e-01 4.6980420e-01 -2.8086659e-01 7.1058303e-01
5.8198488e-01 -5.3041059e-01 4.4241834e-01 -4.5109406e-01
-3.7213314e-02 1.7092207e-02 -5.2308637e-01 1.2438506e+00
7.8800537e-02 4.2109203e-01 9.5843012e-03 -1.1215954e+00
4.5827591e-01 6.1089152e-01 5.7464975e-01 -9.9039030e-01
3.9250654e-01 5.4019064e-01 7.2174519e-01 -4.7045293e-01
-5.3478390e-01 -6.1391437e-01 5.7194149e-01 4.4711965e-01
6.8980112e-04 7.4345541e-01 2.0784682e-01 -2.5429875e-01
1.4257734e+00 -2.6422504e-01 4.3881112e-01 -3.2882792e-01
4.6692854e-01 -1.1299050e-01 2.0831817e-01 6.7300153e-01
-8.8785249e-01 5.8887434e-01 3.8166928e-01 -4.3519190e-01
-1.2029390e+00 -1.2546443e+00 -4.4299352e-01 5.8238775e-01
-4.7821581e-01 2.5200945e-01 -6.6937757e-01 -7.9935765e-01
-6.1196923e-02 5.0313854e-01 -7.3756790e-01 -1.4166468e-01
-4.8937207e-01 -1.6205221e+00 8.3961946e-01 9.2282635e-01
8.6685348e-01 -4.9408928e-01 -8.0017406e-01 2.6983133e-01
-2.1255590e-01 -9.5119333e-01 3.4298629e-01 2.1040401e-01
-1.3196744e+00 -1.4716251e+00 -1.7670703e+00 -7.2298586e-01
3.9613113e-01 1.1818195e-01 9.5454317e-01 -1.3797750e-01
-1.3534439e-01 7.5324374e-01 -9.1028854e-02 2.0740306e-01
-2.5107712e-01 -1.5682831e-01 2.1120819e-01 -4.5090023e-01
6.6635287e-01 -1.1073890e+00 -1.0682263e+00 2.3959789e-01
-6.7659843e-01 -3.4173140e-01 9.1634548e-01 3.1901687e-01
6.8598282e-01 -2.8069559e-01 9.1895336e-01 -1.8221553e-01
7.1602136e-01 -8.1119072e-01 2.5292423e-02 2.3183517e-01
-5.9511060e-01 -3.5186097e-01 8.9569196e-02 -3.6741519e-01
-8.0011100e-01 -3.1818947e-01 -2.6922834e-01 -1.4812392e-01
-6.4683330e-01 3.7306052e-01 -2.7547976e-01 7.0098594e-02
6.7928118e-01 5.6475136e-02 2.1299221e-01 -4.8719621e-01
1.3193713e-01 4.2873779e-01 3.9064449e-01 -2.1028227e-01
-1.7720781e-01 7.2327697e-01 9.2365272e-02 -6.0012704e-01
-4.2117825e-01 -7.5523093e-02 -7.2036886e-01 -1.8317300e-01
1.2530559e+00 -9.9883181e-01 -3.6553138e-01 3.5694072e-01
-1.0701398e+00 -3.0368459e-01 -1.3589285e-01 8.6825448e-01
-4.7584227e-01 5.3433847e-01 -5.5763555e-01 -3.4594876e-01
-6.1475337e-01 -1.3621918e+00 6.5256089e-01 -2.3185199e-01
-5.0476068e-01 -1.2871531e+00 3.0157429e-01 3.2254678e-01
5.5756015e-01 1.0552221e-01 1.2686200e+00 -8.6381233e-01
-7.0091248e-02 -1.2194093e-04 -8.2790744e-01 4.3552932e-01
2.8782627e-01 -7.1396697e-01 -6.7052257e-01 1.5078849e-01
2.6292774e-01 1.1782087e-01 8.5114574e-01 8.8075364e-01
5.2651298e-01 6.2684938e-02 -4.3958417e-01 -9.2011891e-02
1.2630017e+00 6.0072470e-01 8.0095726e-01 8.7094593e-01
4.5657536e-01 3.2811168e-01 -1.7786205e-01 1.7591231e-01
7.3387319e-01 6.0114902e-01 6.0723639e-01 2.6073265e-01
-6.5144831e-01 6.9184464e-01 3.6751258e-01 1.7493130e-01
-3.6327854e-01 -1.3243932e-01 -1.0659136e+00 4.9683264e-01
-1.4981896e+00 -8.5872740e-01 -5.9889317e-01 1.7105402e+00
4.2848727e-01 -9.5462307e-02 5.7335019e-01 1.8511939e-01
9.0576869e-01 -1.7843652e-01 -7.6796633e-01 2.1622831e-01
-5.0809634e-01 1.1811262e-01 3.6752826e-01 5.2097008e-02
-9.9691343e-01 -4.4247814e-02 7.0078082e+00 1.5523702e-01
-1.1490523e+00 1.0405960e+00 6.3767666e-01 -1.5721349e-01
-9.3201123e-02 -4.9775648e-01 -5.8513987e-01 6.8413275e-01
1.5799181e+00 8.7252580e-02 3.1611953e-02 6.4662594e-01
4.4795540e-01 -4.0685034e-01 -1.0986791e+00 7.3138291e-01
-9.8004099e-03 -1.0815957e+00 1.3229419e-01 4.1141462e-01
2.5369638e-01 5.5785048e-01 -6.6301599e-02 9.4567956e-03
1.1329938e-02 -9.0425175e-01 6.1017001e-01 8.4783560e-01
8.0430377e-01 -1.7765670e-01 1.1415402e+00 -1.8175286e-01
-7.8462660e-01 6.4504258e-02 1.2114953e-02 -4.6206608e-02
4.9656400e-01 7.1363169e-01 -7.1984762e-01 1.6270341e-01
6.8549365e-01 6.7604363e-01 -7.7121234e-01 1.3559022e+00
-1.4074770e-01 4.2082739e-01 -2.3152453e-01 3.3989045e-01
2.5610018e-01 8.7914601e-02 4.7019330e-01 1.0347552e+00
5.9593797e-01 3.2680467e-02 -8.7115392e-02 1.0476135e+00
4.1817725e-01 -2.3253398e-01 -3.1543556e-01 -5.4357599e-02
5.5292811e-02 1.0011513e+00 -9.8947656e-01 -7.2924745e-01
-6.7179066e-01 7.5425172e-01 -2.1482606e-01 1.3305312e-01
-7.0033383e-01 3.7625208e-01 3.6685285e-01 6.0931784e-01
1.0388846e-01 -7.0313461e-02 -6.0005051e-01 -9.1518074e-01
-1.5732972e-01 -6.3218737e-01 4.4350952e-01 -9.4290590e-01
-1.5722653e+00 6.5639055e-01 1.2531580e-01 -9.2381394e-01
-1.4005047e-01 -8.6777538e-01 -2.3011383e-01 7.7899683e-01
-1.5271832e+00 -1.0071567e+00 -3.9996767e-01 5.6097680e-01
-5.2767299e-02 -1.2499905e-01 1.0224720e+00 7.9944032e-01
-4.3550470e-01 -4.7958223e-03 -1.7171670e-01 -8.2223434e-03
5.2601439e-01 -1.0971187e+00 -1.6312122e-01 4.0382907e-01
-9.6638930e-01 4.2698765e-01 2.0464525e-01 -1.0592545e+00
-5.4563421e-01 -1.2852994e+00 6.0716075e-01 -2.0844060e-01
7.3088533e-01 5.6063497e-01 -8.3592117e-01 6.4860851e-01
1.7683908e-01 1.0300555e-01 1.0046272e+00 -4.1837355e-01
-3.4315720e-02 2.7729884e-01 -1.6224687e+00 2.9725277e-01
1.1656803e+00 -3.6529839e-01 -9.5895815e-01 3.7322462e-01
2.1764471e-01 4.2207882e-01 -1.1862701e+00 5.5608118e-01
7.0026022e-01 -1.0918294e+00 1.2086374e+00 -5.4240757e-01
3.0028111e-01 -9.8206595e-02 -5.2296788e-01 -1.0693249e+00
-5.2747548e-01 7.4228758e-01 -2.0330918e-01 5.4929912e-01
-1.3227756e-01 -9.1521883e-01 6.4466453e-01 5.7001400e-01
-3.0684295e-01 -4.8759133e-01 -1.2557852e+00 -8.5236442e-01
3.7495485e-01 -4.5530227e-01 3.2426322e-01 7.6131266e-01
-5.0947452e-01 8.7201893e-02 2.3823655e-01 1.0242685e-01
5.8276266e-01 -6.4871353e-01 -4.2980716e-01 -1.8946937e+00
2.6228151e-01 -6.4423978e-01 -8.4746742e-01 -5.1286981e-02
2.2515973e-01 -1.1011183e+00 -7.7278423e-01 -2.1400578e+00
5.6486243e-01 -1.5450469e-01 -3.7662789e-01 4.0105540e-01
2.7935728e-01 7.5935334e-01 -8.4723495e-02 2.1314730e-01
-5.0048912e-01 3.3486599e-01 1.2336640e+00 -1.6604047e-01
7.3916562e-02 -3.5086945e-01 -8.1612915e-01 9.3805140e-01
8.3697063e-01 -2.5223401e-01 -3.1629103e-01 -2.1090540e-01
-1.6440041e-01 -2.0710455e-01 1.2784832e+00 -1.3989491e+00
-2.1532854e-01 3.0137008e-01 9.9107438e-01 -4.8583558e-01
4.7282666e-01 -8.2775861e-01 3.2812220e-01 8.6427981e-01
1.8752448e-01 1.0233327e-01 3.4072265e-02 2.2314119e-01
8.2198642e-02 -2.3490205e-01 1.0089109e+00 -7.8236526e-01
-6.8441254e-01 1.8520899e-01 -1.0671160e+00 -9.5311431e-03
8.8495976e-01 -4.4088000e-01 -6.6512293e-01 5.8140349e-02
-1.5853409e+00 -2.6047030e-01 1.4174674e-01 1.3294363e-01
4.9769607e-01 -1.5298202e+00 -4.4991076e-01 -2.8256330e-01
-3.7309220e-01 -6.2278527e-01 6.1471307e-01 1.8050529e+00
-3.5943422e-01 9.3422079e-01 -7.0087767e-01 -6.6986895e-01
-1.0068557e+00 2.0479399e-01 9.6358532e-01 -3.2469407e-01
-7.9239690e-01 3.7661248e-01 2.1811686e-01 -5.6599718e-02
-3.5526693e-02 -4.5245168e-01 -3.1689504e-01 6.5405422e-01
9.7162002e-01 8.8752973e-01 4.9023983e-01 -6.4222276e-01
-7.8356296e-01 1.1627054e-01 -2.4745698e-01 -3.6533874e-01
1.4602307e+00 -3.3955455e-01 -4.0717798e-01 2.7925780e-01
1.0888915e+00 -8.5672843e-01 -7.0566761e-01 1.5162835e-01
-2.2151990e-02 2.6625010e-01 5.2844256e-01 -1.4436114e+00
-1.1951568e+00 7.8538042e-01 1.6625366e+00 -3.0090306e-02
8.9366424e-01 1.2930502e-01 7.5547284e-01 1.8403988e-01
3.7180623e-01 -5.4842937e-01 -1.0693755e-01 1.6194148e-01
8.8591439e-01 -8.6766201e-01 1.6971665e-02 4.2306680e-02
-3.5020602e-01 1.2199718e+00 5.4261816e-01 -2.2528431e-01
7.3713112e-01 4.6142627e-02 -6.3732073e-02 -2.9365280e-01
-2.7538434e-01 -3.0915815e-01 1.4967506e-01 1.0025914e+00
3.3609515e-01 3.6376190e-01 -7.0867109e-01 1.2550268e+00
-1.5545795e-03 6.2379187e-01 2.0524701e-01 9.6723735e-01
-6.1801249e-01 -1.1234183e+00 -6.4164883e-01 1.0218213e+00
-5.1912600e-01 -4.7295149e-02 -5.1938838e-01 8.6055803e-01
6.9717962e-01 5.4737753e-01 2.6846859e-01 7.9151362e-02
2.1419030e-01 5.2696711e-01 7.1126974e-01 -3.8029841e-01
-2.6635844e-01 1.0430453e-01 2.8778887e-01 -3.3023956e-01
-1.0090631e+00 -1.0043764e+00 -1.2514712e+00 -2.4801934e-02
3.3893099e-01 -2.6813352e-01 3.4845456e-01 1.3081908e+00
4.6561456e-01 8.7573183e-01 -4.5466745e-01 -8.8383865e-01
-5.2045945e-02 -1.1025447e+00 -6.7484927e-01 -4.6772397e-01
3.7104627e-01 -1.2291713e+00 -4.2528546e-01 -1.8822329e-01] | [14.139301300048828, -1.769507646560669] |
66d9c66d-40ac-46b8-a395-118a7a05f701 | improving-conversational-passage-re-ranking | 2304.13290 | null | https://arxiv.org/abs/2304.13290v1 | https://arxiv.org/pdf/2304.13290v1.pdf | Improving Conversational Passage Re-ranking with View Ensemble | This paper presents ConvRerank, a conversational passage re-ranker that employs a newly developed pseudo-labeling approach. Our proposed view-ensemble method enhances the quality of pseudo-labeled data, thus improving the fine-tuning of ConvRerank. Our experimental evaluation on benchmark datasets shows that combining ConvRerank with a conversational dense retriever in a cascaded manner achieves a good balance between effectiveness and efficiency. Compared to baseline methods, our cascaded pipeline demonstrates lower latency and higher top-ranking effectiveness. Furthermore, the in-depth analysis confirms the potential of our approach to improving the effectiveness of conversational search. | ['Chuan-Ju Wang', 'Ming-Feng Tsai', 'Sheng-Chieh Lin', 'Jia-Huei Ju'] | 2023-04-26 | null | null | null | null | ['passage-re-ranking', 'conversational-search'] | ['natural-language-processing', 'natural-language-processing'] | [-2.95767546e-01 -1.24243021e-01 -1.79085344e-01 -5.69528759e-01
-1.37327325e+00 -9.24147010e-01 9.76293504e-01 5.36230654e-02
-6.04589403e-01 7.69712627e-01 1.04724944e+00 -2.69551307e-01
3.97357997e-03 -5.52827001e-01 -3.59998137e-01 -2.30219677e-01
2.43719459e-01 9.04803753e-01 2.89155394e-01 -5.39926946e-01
5.13722062e-01 -4.85175662e-02 -1.54896164e+00 1.02785003e+00
9.18036163e-01 3.67400467e-01 3.39964516e-02 9.05791044e-01
-1.60469130e-01 9.49028432e-01 -7.67256975e-01 -5.75020134e-01
-3.23852189e-02 -3.48492473e-01 -1.41480601e+00 -5.69223583e-01
6.06883824e-01 -5.68371713e-01 -3.41907650e-01 3.79731387e-01
6.92249477e-01 5.70271075e-01 3.64794284e-01 -7.59234190e-01
-5.62141001e-01 1.02855432e+00 -1.11869276e-01 5.83326578e-01
7.54689276e-01 -1.57169402e-01 1.55548048e+00 -9.25512969e-01
7.40820587e-01 1.18512237e+00 4.51487273e-01 4.81123298e-01
-1.11173975e+00 -5.41745305e-01 6.72502592e-02 3.61276120e-01
-1.23107243e+00 -5.15380979e-01 3.73172909e-01 -1.28619492e-01
1.44948494e+00 4.90697622e-01 5.16156316e-01 1.24424326e+00
-2.50463814e-01 9.08491075e-01 1.18781292e+00 -4.55950558e-01
-5.32819740e-02 1.37845635e-01 5.44329703e-01 5.42951345e-01
-1.97694764e-01 -2.36988500e-01 -1.05937552e+00 -4.35412258e-01
9.17194784e-02 -5.49066216e-02 -5.84001184e-01 -1.24094635e-01
-9.79908884e-01 4.34874535e-01 3.38612944e-01 3.74113590e-01
-1.12975016e-02 -1.41927330e-02 5.50565481e-01 5.41865110e-01
8.21482480e-01 8.82998705e-01 -4.78957415e-01 -7.18943119e-01
-1.01090848e+00 2.37761453e-01 1.12958407e+00 8.54567349e-01
5.14436662e-01 -7.46886253e-01 -6.61764264e-01 1.17551029e+00
-8.56963694e-02 4.19119865e-01 3.94868612e-01 -1.05576789e+00
4.97353345e-01 8.07445168e-01 2.55314529e-01 -7.08257020e-01
-4.06378567e-01 -8.19666445e-01 -2.06698343e-01 -4.44418192e-01
-6.39252514e-02 3.22339296e-01 -3.66052717e-01 1.54414809e+00
8.55341926e-02 1.34626821e-01 3.36621106e-01 8.68393064e-01
1.20656097e+00 6.30963564e-01 -5.68064675e-02 -1.72456995e-01
1.28983140e+00 -1.69299579e+00 -5.51201344e-01 7.28769898e-02
7.25994229e-01 -8.13283682e-01 1.47652090e+00 1.34147912e-01
-8.90393019e-01 -3.82940471e-01 -9.00556564e-01 -4.68863755e-01
-1.75314501e-01 1.31675992e-02 5.53070307e-01 4.35064226e-01
-1.49671412e+00 5.28849602e-01 -6.28649712e-01 -5.02006888e-01
-2.78139282e-02 9.84184742e-02 -2.53854156e-01 -1.11802444e-01
-1.18422520e+00 7.71635950e-01 1.09350614e-01 -1.35864049e-01
-8.58979940e-01 -7.60709703e-01 -2.97797203e-01 3.11329752e-01
3.31414402e-01 -8.93742502e-01 1.70282662e+00 -2.54508376e-01
-1.55747390e+00 7.74204135e-01 -4.29508418e-01 -4.72545445e-01
5.39075911e-01 -5.15929699e-01 -3.17836016e-01 2.76646644e-01
8.86700749e-02 5.55498481e-01 1.34918511e-01 -1.15316582e+00
-8.42519581e-01 1.06733449e-01 5.93011081e-01 7.89463937e-01
-7.05990434e-01 -3.27858746e-01 -9.32784438e-01 -1.45491466e-01
-2.56437331e-01 -1.00919700e+00 8.92703235e-02 -9.24477816e-01
-2.02977598e-01 -7.21954226e-01 6.01859748e-01 -5.48490286e-01
1.60615027e+00 -1.87560582e+00 1.70255050e-01 -3.77261825e-03
4.36913550e-01 3.32210928e-01 -2.89757222e-01 1.10197675e+00
4.53352898e-01 2.13236645e-01 1.15568556e-01 -7.02558041e-01
-1.93002686e-01 -3.83475609e-02 -2.62709349e-01 -1.79620106e-02
-4.73876536e-01 1.16487801e+00 -1.44420993e+00 -4.82840061e-01
1.83586419e-01 3.91422957e-01 -5.76781631e-01 4.30798084e-01
-3.07909042e-01 3.28822076e-01 -3.60895067e-01 2.14431763e-01
2.35363811e-01 -5.88356614e-01 3.79066676e-01 -9.03526992e-02
-1.11668095e-01 1.10208857e+00 -3.16981226e-01 2.05741978e+00
-8.54997396e-01 8.21394324e-01 -3.37202102e-01 -2.51710474e-01
5.44766545e-01 2.45795682e-01 2.15999484e-01 -9.36853170e-01
-2.02244639e-01 7.25406855e-02 -5.37226140e-01 -3.88119906e-01
1.14198649e+00 5.18728316e-01 -6.94274381e-02 9.77811575e-01
2.26486046e-02 2.08640084e-01 3.44438374e-01 1.05240464e+00
1.54027188e+00 1.02027640e-01 3.01250909e-02 -3.92378509e-01
5.50115824e-01 3.07789475e-01 1.21084347e-01 1.16162241e+00
1.91794693e-01 4.01409060e-01 2.68963665e-01 -2.70961076e-01
-8.49016130e-01 -7.62228370e-01 9.32876989e-02 1.48706973e+00
2.43784785e-01 -9.35134172e-01 -7.23197699e-01 -9.56251085e-01
-2.17906192e-01 7.64848173e-01 -5.47876418e-01 -9.18145627e-02
-5.97640216e-01 -5.64567685e-01 6.68525040e-01 3.97995383e-01
6.86261833e-01 -9.63499129e-01 -2.24656567e-01 1.69781893e-01
-9.86391902e-01 -1.24206936e+00 -4.85381216e-01 -3.08049411e-01
-5.80388427e-01 -1.05512190e+00 -4.34815347e-01 -7.34947026e-01
3.54675651e-01 1.00248611e+00 1.75720787e+00 3.91115457e-01
2.77075052e-01 7.04214752e-01 -9.49085414e-01 1.65949523e-01
-4.13117051e-01 7.67332315e-01 -1.97002158e-01 -4.19701934e-01
4.03181672e-01 -6.79853082e-01 -1.21689999e+00 3.62435699e-01
-4.78435606e-01 1.62834138e-01 2.66336262e-01 9.52655315e-01
1.50406808e-01 -3.96298856e-01 7.12924123e-01 -1.10944152e+00
1.27672267e+00 -2.29830608e-01 -2.00254127e-01 5.78626633e-01
-1.12511075e+00 4.99012023e-02 3.95549446e-01 -8.06869194e-03
-1.22541153e+00 -4.62538302e-01 -1.08694561e-01 6.13921061e-02
2.95652568e-01 4.79061276e-01 5.01559079e-01 1.54991239e-01
7.25465059e-01 8.76436755e-02 -2.57120341e-01 -7.63121605e-01
7.76122928e-01 9.27718103e-01 5.28107285e-01 -4.00933176e-01
2.50494808e-01 4.57174957e-01 -6.74836397e-01 -4.68670785e-01
-1.14228916e+00 -1.01377189e+00 -2.73180723e-01 -5.15817106e-01
4.39556092e-01 -1.23472321e+00 -1.07846093e+00 1.24995120e-01
-9.71659541e-01 -4.17346418e-01 3.05706114e-02 3.10846448e-01
-1.64518971e-02 3.24927360e-01 -9.04595673e-01 -6.01368010e-01
-9.88548338e-01 -9.01579738e-01 1.18871784e+00 2.18089551e-01
-3.77753437e-01 -8.76155615e-01 5.59385777e-01 1.13400996e+00
5.00754476e-01 -4.99225914e-01 6.65512502e-01 -1.03204441e+00
-5.27602196e-01 -1.39722854e-01 -2.52714932e-01 6.96096420e-02
-1.82372227e-01 -3.19902003e-01 -1.12959039e+00 -4.85990793e-01
-5.97451687e-01 -5.22279263e-01 1.06778312e+00 -1.99590042e-01
7.49399781e-01 -1.67072073e-01 -3.88898909e-01 7.34986290e-02
1.14326668e+00 -6.50348961e-02 2.39904091e-01 4.47731793e-01
5.20824671e-01 4.52742398e-01 6.73861742e-01 1.89802170e-01
9.12265301e-01 8.17564011e-01 1.67329654e-01 -4.19375785e-02
-3.48328829e-01 -3.89709502e-01 3.25614125e-01 1.48257709e+00
6.49784356e-02 -7.50486553e-01 -8.32956851e-01 5.86033702e-01
-2.09821796e+00 -1.00529075e+00 -5.40050818e-03 1.96619189e+00
9.00414884e-01 -1.38020024e-01 2.01270636e-02 -3.46279234e-01
4.82946694e-01 2.90773481e-01 -3.19289207e-01 -4.00970995e-01
-8.09029639e-02 1.76003560e-01 4.47194017e-02 9.31522846e-01
-7.19855309e-01 1.14992213e+00 7.40058994e+00 7.99752533e-01
-6.10274017e-01 2.90859938e-01 3.55986178e-01 -4.59108740e-01
-5.90985417e-01 5.37254624e-02 -8.31695318e-01 1.48590982e-01
9.14964795e-01 -3.33242089e-01 7.04867363e-01 7.83587456e-01
3.91237587e-02 3.01499553e-02 -1.17002392e+00 8.49358439e-01
3.75540316e-01 -1.66740644e+00 1.28463417e-01 -1.50820345e-01
8.40319872e-01 5.21778166e-01 -3.51587683e-01 7.43019760e-01
8.34825754e-01 -5.07564068e-01 2.98579842e-01 3.59207302e-01
4.20494497e-01 -7.57046521e-01 1.03922689e+00 3.28886002e-01
-9.86227810e-01 1.18214451e-01 -5.77226328e-03 -4.27137539e-02
2.43559107e-01 4.99285787e-01 -1.44235635e+00 6.84832692e-01
8.85045946e-01 6.82274997e-01 -7.00069547e-01 8.06965053e-01
-2.84046799e-01 7.00466633e-01 -1.51962101e-01 -3.88386160e-01
1.39073104e-01 2.30145967e-03 6.27241254e-01 1.77320802e+00
-1.62088141e-01 3.32318634e-01 2.93140709e-01 3.19138199e-01
-5.04359245e-01 3.10629964e-01 -3.04243296e-01 3.06024086e-02
8.09599459e-01 1.45581186e+00 -4.65759248e-01 -5.93781173e-01
-1.89655557e-01 1.02408707e+00 8.34320784e-01 3.68247122e-01
-4.68184561e-01 1.08676087e-02 4.16923821e-01 -3.01951081e-01
4.14443947e-03 5.79719571e-03 2.74475999e-02 -1.22700083e+00
7.48772770e-02 -1.04254639e+00 6.40091121e-01 -8.13926935e-01
-1.20671952e+00 1.21327460e+00 1.90147292e-02 -1.04991424e+00
-4.22211677e-01 5.42955995e-02 -2.57958829e-01 6.46893501e-01
-1.33616841e+00 -1.09717488e+00 -5.44888794e-01 3.41181904e-01
9.29574668e-01 3.70237753e-02 1.02740777e+00 4.15940821e-01
-4.32861924e-01 6.76972747e-01 2.14544222e-01 6.49803830e-03
9.04500186e-01 -1.33738911e+00 4.52338576e-01 6.27151072e-01
4.03384656e-01 1.02391660e+00 6.45775437e-01 -4.23413604e-01
-1.21394455e+00 -7.92058647e-01 1.08843029e+00 -7.82491922e-01
4.44291890e-01 -4.86029446e-01 -6.90778732e-01 3.77826393e-01
8.25297892e-01 -7.54147530e-01 9.41352785e-01 9.01065707e-01
-5.79582036e-01 -2.39239052e-01 -6.52247965e-01 7.23562300e-01
1.21678221e+00 -1.01577318e+00 -8.26383710e-01 4.91564661e-01
1.12532222e+00 -4.54990417e-01 -6.51174188e-01 5.00719309e-01
7.23391533e-01 -1.07449675e+00 9.05656278e-01 -6.31592631e-01
3.59808385e-01 -1.02141887e-01 1.11851230e-01 -1.42892766e+00
-3.23203683e-01 -6.14380598e-01 -1.32938713e-01 1.44329345e+00
6.41817987e-01 -4.40064847e-01 6.96474552e-01 3.27060729e-01
-1.59487039e-01 -7.64264822e-01 -7.78650343e-01 -4.55748528e-01
-3.44430417e-01 -1.83395654e-01 5.13918757e-01 8.72693241e-01
4.00640965e-01 8.76166046e-01 -4.41120446e-01 -1.09106496e-01
2.03159228e-01 4.58087653e-01 9.06364918e-01 -1.03189862e+00
-2.09031656e-01 -3.20624471e-01 1.01959065e-01 -1.43282235e+00
2.44269818e-01 -1.08160341e+00 -2.11081449e-02 -1.94186497e+00
7.54975677e-01 -4.70741868e-01 -6.41176820e-01 3.26259673e-01
-4.21117693e-01 3.89899135e-01 -1.25304520e-01 4.49188530e-01
-1.53143859e+00 4.90711778e-01 9.62326884e-01 -4.00909223e-02
-3.96216065e-01 -2.99418718e-01 -8.04007530e-01 2.66595006e-01
5.56559265e-01 -5.93147576e-01 -6.36405766e-01 -6.42207205e-01
5.05850554e-01 1.44649418e-02 -7.05692545e-02 -6.94503784e-01
4.34862405e-01 3.60411644e-01 -2.93358505e-01 -8.11421514e-01
4.37657565e-01 -3.00979257e-01 -2.08584696e-01 8.57854858e-02
-8.33105743e-01 3.67818058e-01 -3.16260452e-03 7.74464905e-01
-2.63484955e-01 1.72711551e-01 1.06722541e-01 -6.43889979e-02
-6.58682644e-01 -1.10923305e-01 -3.64143014e-01 2.05384895e-01
2.68838018e-01 3.81908387e-01 -8.83376360e-01 -8.08354974e-01
-3.69547278e-01 6.59803867e-01 3.63495350e-01 8.44935596e-01
2.32391194e-01 -1.05468261e+00 -8.12868297e-01 -3.92614394e-01
4.14640963e-01 -3.61234784e-01 3.21259677e-01 5.43140233e-01
-4.45527256e-01 6.44997299e-01 2.69885272e-01 -6.34551466e-01
-1.64084339e+00 1.70857310e-01 1.18542992e-01 -8.25304747e-01
-6.20094478e-01 9.38809276e-01 -2.35238671e-01 -6.65070415e-01
3.07303101e-01 1.11894354e-01 -5.25945127e-01 1.26572102e-01
7.05664396e-01 6.70732498e-01 4.07678217e-01 -2.00092882e-01
-4.21612173e-01 4.39518457e-03 -5.63482463e-01 -5.22417426e-01
1.23413563e+00 -5.70484579e-01 -4.20220375e-01 4.79204208e-01
1.18850160e+00 3.48282486e-01 -6.44823670e-01 -4.79514778e-01
5.69523461e-02 -2.41832584e-01 1.84339628e-01 -1.29683030e+00
-6.76356316e-01 3.78287256e-01 3.13077360e-01 8.92407894e-02
1.00346911e+00 5.28359450e-02 8.61976087e-01 9.31836188e-01
5.54219365e-01 -9.89328802e-01 3.60466719e-01 6.38213158e-01
7.22328544e-01 -1.11428428e+00 1.01886600e-01 -3.40481907e-01
-6.39924169e-01 8.43605697e-01 7.42702007e-01 1.20409250e-01
2.91307479e-01 8.24289545e-02 4.14897323e-01 -3.42508435e-01
-1.39471674e+00 -1.40142128e-01 4.19285506e-01 -9.37289745e-02
5.96488059e-01 -5.18435938e-03 -7.35085428e-01 2.92956144e-01
-3.91970783e-01 -1.68067873e-01 2.82815635e-01 8.33669841e-01
-2.77101219e-01 -1.42502701e+00 2.62638897e-01 2.69061238e-01
-4.31997806e-01 -6.24617934e-01 -7.43522227e-01 4.84029859e-01
-4.80679423e-01 1.51344502e+00 7.57203698e-02 -7.33406842e-01
3.97529483e-01 2.18407229e-01 1.00162894e-01 -6.00766778e-01
-1.27166724e+00 -5.91997057e-02 7.58452892e-01 -8.01516056e-01
-6.13025069e-01 -4.29794788e-01 -9.27238762e-01 -4.66208518e-01
-6.36734307e-01 9.65966463e-01 4.84808326e-01 9.63823915e-01
9.72702265e-01 4.60521311e-01 7.87789881e-01 -2.96438992e-01
-2.91558951e-01 -1.22654772e+00 1.61842734e-01 4.71854329e-01
2.58771330e-01 -3.23868483e-01 -3.30166370e-01 -3.11141908e-01] | [12.039189338684082, 7.813551902770996] |
74fe67bf-b9ed-4d9c-8932-a2daeca41f9e | viewpoint-invariant-change-captioning | 1901.02527 | null | http://arxiv.org/abs/1901.02527v2 | http://arxiv.org/pdf/1901.02527v2.pdf | Robust Change Captioning | Describing what has changed in a scene can be useful to a user, but only if
generated text focuses on what is semantically relevant. It is thus important
to distinguish distractors (e.g. a viewpoint change) from relevant changes
(e.g. an object has moved). We present a novel Dual Dynamic Attention Model
(DUDA) to perform robust Change Captioning. Our model learns to distinguish
distractors from semantic changes, localize the changes via Dual Attention over
"before" and "after" images, and accurately describe them in natural language
via Dynamic Speaker, by adaptively focusing on the necessary visual inputs
(e.g. "before" or "after" image). To study the problem in depth, we collect a
CLEVR-Change dataset, built off the CLEVR engine, with 5 types of scene
changes. We benchmark a number of baselines on our dataset, and systematically
study different change types and robustness to distractors. We show the
superiority of our DUDA model in terms of both change captioning and
localization. We also show that our approach is general, obtaining
state-of-the-art results on the recent realistic Spot-the-Diff dataset which
has no distractors. | ['Anna Rohrbach', 'Trevor Darrell', 'Dong Huk Park'] | 2019-01-08 | robust-change-captioning | http://openaccess.thecvf.com/content_ICCV_2019/html/Park_Robust_Change_Captioning_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Park_Robust_Change_Captioning_ICCV_2019_paper.pdf | iccv-2019-10 | ['natural-language-visual-grounding'] | ['reasoning'] | [ 1.97978705e-01 -3.82296085e-01 1.94445401e-01 -3.34391117e-01
-8.34347546e-01 -9.78817105e-01 9.12657201e-01 1.10033348e-01
-4.20912951e-01 4.85291213e-01 4.50300574e-01 -1.18079416e-01
4.11280960e-01 -4.85528052e-01 -1.14066815e+00 -6.63391471e-01
4.71056737e-02 4.04405862e-01 6.11862838e-01 -4.41354007e-01
2.95029044e-01 6.21064007e-01 -1.79489601e+00 3.93450916e-01
2.19897687e-01 6.07342541e-01 5.72751939e-01 6.96928322e-01
5.83268888e-02 7.89310694e-01 -7.73986876e-01 1.52449226e-02
1.81122348e-01 -4.50866491e-01 -1.00720072e+00 2.99055248e-01
9.47999299e-01 -3.11162859e-01 -4.18852419e-01 1.02187824e+00
6.00007951e-01 4.21524972e-01 5.42532384e-01 -9.94122863e-01
-1.14479923e+00 2.69843191e-01 -5.69627345e-01 8.41313660e-01
6.38584137e-01 5.48084736e-01 9.30591762e-01 -1.06082880e+00
8.88222039e-01 1.48442149e+00 1.16595186e-01 6.81078017e-01
-1.39300215e+00 -3.26353252e-01 7.91239262e-01 6.43657804e-01
-1.21364033e+00 -8.00892055e-01 8.00812125e-01 -6.72945559e-01
1.17364609e+00 5.03425360e-01 3.02587539e-01 1.45530200e+00
1.69769719e-01 6.98587775e-01 7.35481024e-01 -3.52132887e-01
3.18136424e-01 -4.33875024e-02 1.83360334e-02 2.58579463e-01
2.99039315e-02 -2.51474772e-02 -2.93566227e-01 3.68340492e-01
4.91544574e-01 -1.83607832e-01 -7.42973089e-01 -5.05333066e-01
-1.62167537e+00 3.64575833e-01 5.88991940e-01 3.79307926e-01
-3.56508315e-01 5.02073228e-01 2.90787727e-01 3.74221623e-01
4.33104575e-01 5.23315847e-01 -6.38961136e-01 -2.07777262e-01
-3.77082288e-01 1.17052339e-01 3.10593218e-01 1.18025935e+00
7.05245197e-01 -9.37570855e-02 -6.12413645e-01 7.70395219e-01
-2.23157238e-02 7.17539787e-01 4.77047384e-01 -8.31734240e-01
4.41983372e-01 7.41471499e-02 5.89527726e-01 -9.83759224e-01
-1.74438655e-01 -1.67314261e-01 -5.52006900e-01 2.05615804e-01
1.49313509e-01 2.44473949e-01 -1.20301163e+00 2.13489318e+00
2.52234280e-01 2.61637956e-01 -1.09111674e-01 1.09293365e+00
9.30454791e-01 7.27338433e-01 4.31494527e-02 -2.28913948e-01
1.41773844e+00 -1.01549661e+00 -7.42692292e-01 -5.79401791e-01
4.51383352e-01 -5.82474947e-01 1.65575480e+00 2.53290702e-02
-9.95329022e-01 -7.40683556e-01 -6.34747446e-01 -3.24417472e-01
-5.08619010e-01 -9.69950855e-02 3.33213024e-02 3.73593234e-02
-1.28446209e+00 1.71661034e-01 -8.54559898e-01 -6.86289549e-01
1.38456061e-01 -3.50356176e-02 -3.81145716e-01 -2.34265536e-01
-1.12758493e+00 1.14432144e+00 9.87287536e-02 -1.21503942e-01
-1.25872195e+00 -6.10914409e-01 -1.14127791e+00 1.76523682e-02
5.45279682e-01 -8.45504105e-01 1.47468722e+00 -1.33774710e+00
-1.16987693e+00 1.12283480e+00 -5.61689377e-01 -2.62396067e-01
4.97158557e-01 -1.72984004e-01 -3.99504453e-01 1.70193508e-01
2.58321106e-01 9.25209820e-01 9.49707568e-01 -1.63634467e+00
-5.03080726e-01 -1.48430645e-01 4.36097234e-01 4.12247777e-01
3.21233302e-01 2.62407064e-01 -7.44602978e-01 -6.49171412e-01
-6.35415912e-02 -9.91366386e-01 9.12084207e-02 1.57305136e-01
-2.45476246e-01 -2.35604960e-02 9.64228332e-01 -6.28610313e-01
1.12424016e+00 -2.29009891e+00 2.69333333e-01 -4.37314183e-01
-3.69930360e-03 1.78486127e-02 -3.37366074e-01 3.28508556e-01
-4.45124000e-01 1.34084657e-01 -3.73554260e-01 -4.27261174e-01
-1.41113013e-01 3.02787453e-01 -4.31304932e-01 5.19422650e-01
2.02611640e-01 8.82161915e-01 -1.15359104e+00 -9.00103003e-02
4.02892292e-01 3.80524337e-01 -6.19500339e-01 1.46295503e-01
-4.76552188e-01 5.67408979e-01 -7.69502148e-02 5.60804367e-01
5.47378421e-01 -2.25513414e-01 -4.26751763e-01 -3.34596395e-01
-1.75066590e-01 3.00070018e-01 -9.41603720e-01 1.67548001e+00
-5.64187944e-01 1.06101906e+00 -2.25360960e-01 -5.05629361e-01
4.04399663e-01 4.58064344e-04 9.94276106e-02 -7.76302934e-01
-8.30624253e-02 -3.07876468e-01 -7.43576512e-02 -7.57726610e-01
5.03843427e-01 9.88819227e-02 -1.85683951e-01 2.09373966e-01
-2.44152769e-01 -1.54684559e-01 1.88104481e-01 2.44018897e-01
1.20899749e+00 1.30722851e-01 5.71696281e-01 -2.06085056e-01
5.23126960e-01 -1.60760749e-02 3.27265412e-01 9.44435775e-01
-4.02988315e-01 9.37539160e-01 2.92938471e-01 -3.93548012e-01
-6.72706962e-01 -1.17347336e+00 1.27066746e-01 1.18744183e+00
5.82950294e-01 -1.78357080e-01 -3.53066981e-01 -7.52300918e-01
-1.04444385e-01 1.29830718e+00 -9.98463869e-01 -2.76091367e-01
-5.50125659e-01 -2.73892611e-01 7.15451613e-02 5.03443241e-01
4.61220562e-01 -1.36567962e+00 -9.26188111e-01 1.49990730e-02
-4.65460002e-01 -1.14114642e+00 -9.48441029e-01 -3.57064158e-02
-1.64621979e-01 -9.59708035e-01 -3.92418236e-01 -6.81523502e-01
7.42326856e-01 6.74880326e-01 1.48658681e+00 -3.63940895e-01
-1.56484872e-01 8.58651638e-01 -4.76504356e-01 -2.74592549e-01
-3.94315273e-01 -4.78249133e-01 -6.85127974e-02 6.72588870e-02
1.33710995e-01 -3.20840180e-01 -5.43913841e-01 3.89189243e-01
-1.06128716e+00 1.72643065e-01 3.53957832e-01 5.11822879e-01
7.12975085e-01 -1.04914971e-01 2.37488702e-01 -7.17196524e-01
3.00463647e-01 -5.85145235e-01 -4.09794122e-01 2.59177953e-01
-1.52015626e-01 1.80581808e-01 5.02206266e-01 -5.34771919e-01
-1.06032741e+00 -2.57108826e-02 -1.41384108e-02 -6.96683764e-01
-4.90632743e-01 2.30083570e-01 -5.15756190e-01 3.01164478e-01
7.80633926e-01 3.55272055e-01 -5.73221147e-01 -4.27207351e-01
7.37366319e-01 4.58545715e-01 8.24318171e-01 -3.74601632e-01
6.51573777e-01 8.19172382e-01 -3.69614631e-01 -7.92806506e-01
-8.08848500e-01 -7.00142682e-01 -6.52080894e-01 -1.36363015e-01
8.61937582e-01 -9.82128441e-01 -3.11403275e-01 5.06744087e-01
-1.41736281e+00 -4.99654621e-01 -4.53419238e-01 8.07971284e-02
-7.05907762e-01 3.75629842e-01 -3.24229747e-01 -2.05192462e-01
1.18228607e-01 -1.23228145e+00 1.30580640e+00 -8.70831981e-02
-1.40109599e-01 -9.12093699e-01 2.48311505e-01 -5.70770986e-02
4.43028122e-01 3.53510141e-01 7.57082164e-01 -3.37301463e-01
-6.72953367e-01 2.89030403e-01 -2.54520178e-01 1.88865110e-01
6.14857614e-01 -1.17186941e-01 -9.81594443e-01 -4.45225984e-01
-6.27438575e-02 5.51440194e-02 1.00249290e+00 2.72259802e-01
1.22006071e+00 -4.37657088e-01 -3.63726020e-01 5.57394922e-01
1.44657588e+00 3.03806484e-01 8.17120433e-01 4.04610395e-01
8.25014830e-01 2.70463884e-01 6.27413690e-01 3.34292322e-01
6.20487452e-01 1.02116239e+00 7.18097210e-01 -6.02618381e-02
-6.55791640e-01 -1.02216743e-01 5.43385684e-01 3.51063132e-01
3.15794706e-01 -7.50676334e-01 -8.89854848e-01 8.98898363e-01
-1.68320417e+00 -1.12249112e+00 -3.81777734e-02 2.16444802e+00
8.66714537e-01 -7.79185295e-02 -2.13569239e-01 -3.45827371e-01
9.45680618e-01 5.32615840e-01 -7.58586884e-01 -2.91548699e-01
-2.66020358e-01 -9.39133242e-02 2.85342127e-01 6.50388479e-01
-1.20970976e+00 1.16246438e+00 6.00462580e+00 3.41865748e-01
-1.29256463e+00 2.17881337e-01 2.69295067e-01 -2.72800326e-01
-4.69187260e-01 -7.37427846e-02 -5.61851442e-01 5.84431469e-01
6.46014512e-01 -7.26846829e-02 5.47891498e-01 6.09881461e-01
5.78403056e-01 -3.80817950e-01 -1.45502794e+00 1.24743867e+00
6.28990531e-01 -9.80324566e-01 3.15668464e-01 -4.10522133e-01
7.52777338e-01 2.69065589e-01 -2.55520339e-03 3.33402753e-01
3.63019466e-01 -6.80656910e-01 9.87965584e-01 5.91297448e-01
9.42275465e-01 -2.50205517e-01 4.16802406e-01 1.65181369e-01
-1.05595386e+00 2.40625486e-01 -1.13331176e-01 -1.06984572e-02
2.22581789e-01 2.02001035e-01 -6.49328947e-01 3.43326032e-01
8.68008971e-01 9.98777032e-01 -8.87676060e-01 8.95832241e-01
-3.66859227e-01 4.08447266e-01 -1.86023906e-01 2.50843525e-01
1.42540753e-01 1.36998042e-01 9.34655428e-01 1.35475266e+00
2.83381194e-01 5.83913848e-02 -2.43994687e-02 9.22249019e-01
-1.39948234e-01 -1.62891388e-01 -6.94980025e-01 2.88039625e-01
3.98225874e-01 7.83322096e-01 -4.55391765e-01 -3.95559758e-01
-4.47977960e-01 1.62788022e+00 2.23194271e-01 5.94182909e-01
-1.03547919e+00 -1.90522164e-01 9.74306822e-01 2.50234455e-01
7.05568194e-01 -1.33037895e-01 2.98257381e-01 -1.42676699e+00
1.64666519e-01 -7.90065408e-01 3.78822356e-01 -1.43458760e+00
-1.04046190e+00 5.88011682e-01 2.45397598e-01 -1.16924644e+00
-1.32369936e-01 -2.77697027e-01 -5.20497799e-01 7.65418231e-01
-1.68587887e+00 -1.02929449e+00 -6.92843020e-01 6.15764380e-01
1.06226873e+00 4.22857612e-01 5.41170120e-01 2.23647341e-01
-4.40489113e-01 3.32018346e-01 -3.15437987e-02 -9.16995034e-02
1.03917408e+00 -1.36181021e+00 6.81555808e-01 1.25451863e+00
2.60888249e-01 3.51573378e-01 1.01448584e+00 -4.64089274e-01
-1.22371364e+00 -1.44991970e+00 7.66072035e-01 -9.29768205e-01
5.68931520e-01 -3.76383185e-01 -1.05737221e+00 1.16660798e+00
2.95732856e-01 1.37449160e-01 3.31234746e-03 -1.31480768e-01
-6.26529157e-01 3.49660590e-02 -1.05381358e+00 8.73234689e-01
1.32971561e+00 -5.05199194e-01 -8.61706793e-01 3.21024835e-01
1.10161388e+00 -6.21307373e-01 -4.11555730e-02 2.02802092e-01
1.24583011e-02 -9.65398610e-01 9.94118750e-01 -7.26956189e-01
2.82209814e-01 -7.76586413e-01 -3.29068869e-01 -1.71624815e+00
-7.46155977e-01 -4.88805294e-01 -4.26783785e-02 1.18911827e+00
1.25414833e-01 -5.37476718e-01 -1.55282365e-02 3.49032104e-01
-4.50646996e-01 -1.24024920e-01 -9.34879839e-01 -8.60247254e-01
-1.64055154e-01 -4.79167670e-01 4.59346443e-01 1.03851843e+00
-4.64237303e-01 4.61364299e-01 -3.36059213e-01 5.16419351e-01
-2.67924536e-02 1.01409979e-01 7.90306687e-01 -6.37074113e-01
-3.33912909e-01 -3.85867327e-01 -5.59359908e-01 -1.31184673e+00
2.52687037e-01 -7.18767524e-01 3.17291588e-01 -1.77870083e+00
2.58400917e-01 6.03325665e-02 -2.99022615e-01 6.53875053e-01
-3.86981308e-01 -5.18517010e-03 1.71296492e-01 2.56979823e-01
-8.98013115e-01 5.65931082e-01 1.23198378e+00 -3.16318870e-01
-2.64470905e-01 -4.13691878e-01 -7.15311289e-01 4.74792540e-01
6.14964902e-01 -3.36496830e-01 -3.94881308e-01 -7.90621161e-01
1.50551409e-01 -1.73165753e-01 6.27039135e-01 -9.01453912e-01
7.27379881e-03 -4.18911964e-01 2.71948613e-02 -5.61250091e-01
3.40819716e-01 -7.87550807e-01 1.50549546e-01 2.22528413e-01
-4.76924568e-01 3.19714934e-01 5.06756604e-01 6.99132442e-01
-8.38739872e-02 1.50252476e-01 9.46770191e-01 -1.57754630e-01
-1.30110979e+00 3.15269142e-01 -3.86165917e-01 2.72260487e-01
1.19809508e+00 3.02910223e-03 -5.61357617e-01 -6.27460539e-01
-8.40191066e-01 2.70812213e-01 8.07061195e-01 7.84089148e-01
5.00592589e-01 -1.32277310e+00 -8.25676799e-01 6.90664053e-02
7.71346807e-01 -1.95937138e-02 4.12140369e-01 5.55421054e-01
-4.62651640e-01 3.90266091e-01 -4.72273417e-02 -7.38239348e-01
-1.25070190e+00 1.09449327e+00 5.05474985e-01 3.25888485e-01
-7.12657332e-01 8.48680735e-01 8.54472578e-01 3.45488749e-02
1.11884266e-01 -7.54015088e-01 -6.22470006e-02 -2.08731610e-02
6.65534735e-01 -6.22754432e-02 -1.42794717e-02 -8.74391496e-01
-5.31845629e-01 5.88937998e-01 -7.09991306e-02 -9.64437202e-02
9.43359733e-01 -6.91231787e-01 8.39276537e-02 8.02661180e-01
1.12295544e+00 2.13823229e-01 -1.44197345e+00 -3.50248009e-01
-3.82543832e-01 -8.22747350e-01 -2.10199077e-02 -1.10804224e+00
-8.32092285e-01 7.95596004e-01 6.61815763e-01 1.26740709e-02
1.27762449e+00 4.72437531e-01 3.42101306e-01 3.83613199e-01
3.41573983e-01 -7.37933218e-01 2.17978939e-01 5.34779310e-01
1.61807692e+00 -1.33351493e+00 -3.87341946e-01 -2.90578038e-01
-8.34101796e-01 6.34870231e-01 6.75694942e-01 3.79726700e-02
3.13946187e-01 -1.39575144e-02 1.75223842e-01 -1.95363179e-01
-1.02896726e+00 -6.02488875e-01 2.36302733e-01 6.95008636e-01
6.20282441e-02 1.90522466e-02 1.73222587e-01 1.34414092e-01
1.61762293e-02 -4.73730683e-01 8.14079881e-01 8.67786348e-01
-4.14944679e-01 -5.84682345e-01 -3.87859732e-01 2.55179014e-02
-4.23169211e-02 -2.08897427e-01 -6.00329578e-01 8.55597138e-01
1.97517708e-01 9.18136418e-01 3.44617754e-01 2.25579878e-03
6.73957944e-01 -1.82666242e-01 6.16328478e-01 -9.02934194e-01
-3.04568380e-01 -7.88549706e-02 2.59369519e-02 -7.91216493e-01
-4.34995115e-01 -9.68533933e-01 -1.04091084e+00 9.26506743e-02
2.43382566e-02 -2.88025945e-01 3.82304162e-01 7.17249095e-01
5.88560641e-01 7.61630416e-01 5.90305984e-01 -7.61282742e-01
-2.21974611e-01 -9.62479413e-01 -1.95481956e-01 9.58765447e-01
7.73118138e-01 -7.92516708e-01 -6.46541595e-01 4.29631829e-01] | [10.351526260375977, 1.2268550395965576] |
c3aee6b1-abf6-474e-aeb2-cb390a7b3c2a | feature-engineering-based-detection-of-buffer | 2306.07981 | null | https://arxiv.org/abs/2306.07981v1 | https://arxiv.org/pdf/2306.07981v1.pdf | Feature Engineering-Based Detection of Buffer Overflow Vulnerability in Source Code Using Neural Networks | One of the most significant challenges in the field of software code auditing is the presence of vulnerabilities in software source code. Every year, more and more software flaws are discovered, either internally in proprietary code or publicly disclosed. These flaws are highly likely to be exploited and can lead to system compromise, data leakage, or denial of service. To create a large-scale machine learning system for function level vulnerability identification, we utilized a sizable dataset of C and C++ open-source code containing millions of functions with potential buffer overflow exploits. We have developed an efficient and scalable vulnerability detection method based on neural network models that learn features extracted from the source codes. The source code is first converted into an intermediate representation to remove unnecessary components and shorten dependencies. We maintain the semantic and syntactic information using state of the art word embedding algorithms such as GloVe and fastText. The embedded vectors are subsequently fed into neural networks such as LSTM, BiLSTM, LSTM Autoencoder, word2vec, BERT, and GPT2 to classify the possible vulnerabilities. We maintain the semantic and syntactic information using state of the art word embedding algorithms such as GloVe and fastText. The embedded vectors are subsequently fed into neural networks such as LSTM, BiLSTM, LSTM Autoencoder, word2vec, BERT, and GPT2 to classify the possible vulnerabilities. Furthermore, we have proposed a neural network model that can overcome issues associated with traditional neural networks. We have used evaluation metrics such as F1 score, precision, recall, accuracy, and total execution time to measure the performance. We have conducted a comparative analysis between results derived from features containing a minimal text representation and semantic and syntactic information. | ['Alfredo Cuzzocrea', 'Sheikh Iqbal Ahamed', 'Juan Rodriguez Cardenas', 'Hossain Shahriar', 'Mst Shapna Akter'] | 2023-06-01 | null | null | null | null | ['feature-engineering', 'vulnerability-detection'] | ['methodology', 'miscellaneous'] | [-2.00672299e-01 1.14890918e-01 6.53113611e-03 -8.13457370e-02
-3.75472039e-01 -6.79455459e-01 9.68748182e-02 5.59811115e-01
-2.64226168e-01 1.98757827e-01 1.27536625e-01 -1.01529408e+00
1.45403251e-01 -1.00943744e+00 -6.44421756e-01 -4.14910577e-02
-3.83465618e-01 -4.55171078e-01 3.07778716e-01 -3.40284526e-01
5.96785724e-01 2.03474998e-01 -1.33018684e+00 3.36717159e-01
5.07620752e-01 7.78404713e-01 -9.70377475e-02 7.99054265e-01
-5.16871452e-01 7.75439739e-01 -8.04091275e-01 -4.75104004e-01
1.63552031e-01 7.82893077e-02 -8.84744823e-01 -7.05964744e-01
-1.41879052e-01 -1.52927995e-01 -4.22505826e-01 1.49695992e+00
3.32274318e-01 -2.87477165e-01 2.03015804e-01 -1.27009952e+00
-8.94130170e-01 6.85133457e-01 -3.72676700e-01 3.68421286e-01
3.75250131e-01 1.63507357e-01 7.07431853e-01 -6.56702459e-01
2.94893593e-01 1.18841481e+00 9.32777345e-01 4.72122997e-01
-7.25005865e-01 -6.20398641e-01 -1.29098177e-01 3.33762646e-01
-9.47921515e-01 7.55630359e-02 8.01702440e-01 -7.40171194e-01
1.79524624e+00 -1.59978587e-03 2.75840402e-01 9.40470099e-01
8.32755387e-01 1.82315975e-01 5.63731194e-01 -7.11421907e-01
2.00265393e-01 5.85820794e-01 8.10927212e-01 9.89118874e-01
4.65889513e-01 2.07562774e-01 7.30733350e-02 -7.56315351e-01
1.38156652e-01 3.32925618e-01 -9.67869535e-02 1.70053571e-01
-7.24043548e-01 1.13964581e+00 3.28609347e-01 7.62668312e-01
-7.37640411e-02 2.66624629e-01 1.03998232e+00 6.47806346e-01
1.26105607e-01 6.52597964e-01 -8.71123731e-01 -2.35033020e-01
-5.80890834e-01 -3.64300430e-01 8.28730047e-01 6.36268795e-01
9.42107916e-01 3.18580061e-01 1.52913108e-01 5.07711351e-01
6.46903753e-01 1.70797721e-01 1.10028601e+00 -2.20480502e-01
5.52478909e-01 1.20756817e+00 -4.47717756e-01 -1.53063440e+00
-7.06121549e-02 -6.55873790e-02 -3.55085045e-01 5.88629603e-01
-3.88111353e-01 -1.91183239e-01 -8.44454885e-01 1.39132106e+00
3.83325778e-02 -2.73439917e-03 3.98518920e-01 9.52395648e-02
6.90566301e-01 6.80159271e-01 2.73059867e-02 4.02848512e-01
1.41466761e+00 -6.77456856e-01 -6.27252400e-01 -5.41619658e-02
1.06562245e+00 -7.49245882e-01 9.24846649e-01 1.54231757e-01
-4.90338981e-01 -4.94592220e-01 -1.34346330e+00 1.50025800e-01
-1.25107396e+00 -1.68713167e-01 5.04511833e-01 1.42186499e+00
-1.09118283e+00 7.18772292e-01 -7.66021729e-01 -3.11226636e-01
2.34488741e-01 4.38386381e-01 -5.92088759e-01 3.66481811e-01
-1.15898228e+00 8.13241422e-01 9.09955680e-01 -4.57797609e-02
-9.64558482e-01 -5.45789480e-01 -1.38335478e+00 3.99064958e-01
2.24070568e-02 4.48467731e-02 8.72561038e-01 -8.53193760e-01
-9.87111866e-01 3.36993456e-01 2.10286781e-01 -4.36618954e-01
-4.44943011e-01 1.03978682e-02 -9.38739419e-01 -3.05070400e-01
-1.94454476e-01 -2.14146487e-02 6.78172708e-01 -8.65134597e-01
-5.31761944e-01 -3.27216029e-01 1.43626764e-01 -6.64541483e-01
-1.17810678e+00 5.17187357e-01 2.65264541e-01 -5.46647727e-01
-4.58982080e-01 -5.34713328e-01 1.83303691e-02 -3.72817457e-01
-3.85638893e-01 -4.97775152e-02 1.35272062e+00 -1.16201484e+00
1.63628113e+00 -2.35919881e+00 -2.54334182e-01 3.95906776e-01
2.75355875e-01 7.30036139e-01 -3.07960898e-01 5.58471262e-01
-5.66385150e-01 6.39125764e-01 -3.31121773e-01 2.10997969e-01
-2.88482532e-02 1.21908836e-01 -5.38795233e-01 3.58146578e-01
1.96920648e-01 7.81310678e-01 -5.74903131e-01 -1.39103547e-01
1.70105726e-01 5.34781158e-01 -4.59821135e-01 2.97772676e-01
-1.58537090e-01 -5.75158179e-01 -4.86962646e-01 7.41074741e-01
3.61816347e-01 -4.83619720e-02 -1.04187191e-01 1.10209789e-02
-1.17370628e-01 2.76309103e-01 -8.73691618e-01 1.26700830e+00
-7.91985810e-01 9.24569011e-01 -3.83881927e-01 -9.07790422e-01
1.14401841e+00 6.86922014e-01 -2.51303315e-01 -4.61186320e-01
3.34026903e-01 1.87390015e-01 -1.12834729e-01 -8.48715723e-01
3.89541090e-01 2.39179209e-01 -2.38501713e-01 4.95684683e-01
2.45629773e-01 4.88279492e-01 -2.23381311e-01 3.34059969e-02
1.60359788e+00 -1.98344126e-01 2.45443135e-01 2.15867464e-03
8.57662141e-01 -1.06892940e-02 3.30256760e-01 2.87079334e-01
-3.00287101e-02 3.33506241e-02 7.78494000e-01 -6.62660480e-01
-1.13954759e+00 -5.48240423e-01 2.20091611e-01 7.32672453e-01
-3.85959774e-01 -5.26403368e-01 -9.40320730e-01 -1.00238550e+00
-3.78455222e-03 7.45805442e-01 -8.15480649e-01 -7.53821313e-01
-4.51864332e-01 -5.63185513e-01 1.07604063e+00 6.81287348e-01
2.92833000e-01 -1.31481457e+00 -7.43096471e-01 4.09355283e-01
4.26722139e-01 -6.25135124e-01 -2.68228441e-01 3.26014847e-01
-8.86436939e-01 -1.33062756e+00 2.90173963e-02 -7.90263474e-01
5.78042507e-01 -1.46823272e-01 7.63227344e-01 3.78413558e-01
-6.85287476e-01 1.33033097e-01 -5.66231132e-01 -2.83552617e-01
-6.48032367e-01 -2.26682857e-01 -1.16380811e-01 -2.57417589e-01
6.19921386e-01 -4.84425098e-01 -1.36517987e-01 -1.12013690e-01
-1.15248120e+00 -8.97940159e-01 5.59492052e-01 8.49733233e-01
-7.73186097e-03 4.13256884e-01 3.98061633e-01 -8.68766487e-01
9.87939179e-01 -7.37315357e-01 -7.51651108e-01 3.48002791e-01
-7.83692122e-01 3.69004756e-01 7.94851482e-01 -4.83201385e-01
-8.52344513e-01 -4.07548755e-01 -5.06638765e-01 -3.74113321e-01
-3.84158618e-03 9.31664944e-01 -1.81936786e-01 -5.66973805e-01
8.00657332e-01 2.44031787e-01 -7.71546923e-03 -6.09425187e-01
1.29785419e-01 9.79442954e-01 1.69389486e-01 -3.08519065e-01
1.04919899e+00 -8.90223905e-02 -6.11224234e-01 -5.97336888e-01
-1.16988853e-01 -1.34379938e-01 -3.19526672e-01 1.78344250e-01
9.51975584e-01 -3.80391032e-01 -6.50912702e-01 3.85665447e-01
-1.43586814e+00 1.43228456e-01 -3.53185087e-02 1.56005636e-01
1.37216687e-01 8.48783731e-01 -7.34686255e-01 -7.59825408e-01
-8.71223569e-01 -1.38339233e+00 3.69771630e-01 2.30928913e-01
-3.03705513e-01 -1.19141364e+00 3.71530533e-01 -1.06653579e-01
6.66130900e-01 4.50058103e-01 1.68424332e+00 -1.14806104e+00
-1.60891086e-01 -7.23722219e-01 -1.09791242e-01 8.71811867e-01
3.07224035e-01 2.93326199e-01 -9.18173671e-01 -3.73714983e-01
2.44337320e-01 -6.00766875e-02 4.70631242e-01 -2.22012356e-01
1.23453450e+00 -6.22396588e-01 -3.98169190e-01 7.28200018e-01
1.77145302e+00 7.35615075e-01 8.08885813e-01 7.39296615e-01
6.77324176e-01 4.42124486e-01 8.47896859e-02 9.61593017e-02
1.20023891e-01 8.88672005e-03 6.99491262e-01 2.37499431e-01
2.91771054e-01 -2.33337417e-01 6.87275529e-01 8.58075440e-01
4.15285230e-01 1.58806960e-03 -1.21132314e+00 7.56014168e-01
-1.39682162e+00 -7.88144171e-01 1.92039795e-02 2.21146321e+00
6.13549531e-01 3.44417423e-01 -1.70645282e-01 6.38690114e-01
6.74559653e-01 -1.31430486e-02 -7.19960630e-02 -1.21354222e+00
3.66133571e-01 5.02482414e-01 6.30279005e-01 2.57930368e-01
-1.01729631e+00 7.81518638e-01 5.67369080e+00 6.31849766e-01
-1.31773984e+00 3.83950382e-01 2.91901529e-01 5.28264940e-01
-5.30530930e-01 3.25602889e-01 -5.88024974e-01 5.49994826e-01
1.55962110e+00 -1.74606055e-01 9.34551135e-02 1.27011538e+00
-3.38829666e-01 2.59835482e-01 -8.58652472e-01 4.98353183e-01
8.62310678e-02 -1.12636399e+00 -8.34888369e-02 2.89060995e-02
2.16366902e-01 -8.47857445e-02 5.51709644e-02 5.68980396e-01
2.91410923e-01 -1.17170966e+00 3.42110604e-01 9.11384895e-02
5.91309726e-01 -8.68685842e-01 1.11747205e+00 1.89756677e-02
-1.12580025e+00 -5.80106020e-01 -5.66252530e-01 -3.90414447e-02
-2.51334786e-01 2.60573357e-01 -7.44776368e-01 4.74658757e-01
8.33313704e-01 4.85923469e-01 -7.68752694e-01 8.39057684e-01
-2.02405676e-01 5.42347312e-01 8.83862283e-03 -3.79777849e-01
4.04377311e-01 3.05652589e-01 2.84183174e-01 1.31704664e+00
4.52195346e-01 -3.81910294e-01 -2.86104321e-01 1.00987363e+00
1.24048464e-01 5.31620197e-02 -9.08651352e-01 -5.78991950e-01
3.56759012e-01 1.15232027e+00 -4.82305050e-01 -2.78184444e-01
-7.48093247e-01 5.43928146e-01 1.49279341e-01 2.42643148e-01
-7.22841024e-01 -1.26739073e+00 7.51795590e-01 -1.31638095e-01
3.55639488e-01 -1.01964451e-01 -1.69755042e-01 -9.98625755e-01
3.26562226e-01 -9.33482885e-01 5.27713835e-01 -3.56120676e-01
-1.04284215e+00 1.17393649e+00 -1.82400987e-01 -9.95046735e-01
-6.89896941e-02 -9.53081429e-01 -1.21716869e+00 1.07618892e+00
-1.44645119e+00 -8.57414842e-01 -3.12148165e-02 5.28744638e-01
2.54554629e-01 -9.37385976e-01 1.09881198e+00 4.33317602e-01
-7.84231305e-01 8.50682318e-01 7.12479725e-02 5.83164215e-01
1.91397801e-01 -1.11488104e+00 8.17100406e-01 1.04905713e+00
-1.71496734e-01 9.96098876e-01 3.79650831e-01 -8.40979457e-01
-1.44404507e+00 -1.27315271e+00 7.86625683e-01 -3.64586592e-01
9.27072704e-01 -4.80429709e-01 -1.26231372e+00 8.61907601e-01
1.68615803e-01 3.90737802e-02 8.50038826e-01 -2.05559313e-01
-8.76631558e-01 1.59164473e-01 -1.39802456e+00 2.10040763e-01
1.15475051e-01 -9.48572040e-01 -9.74762857e-01 1.22242875e-01
1.16497934e+00 1.06042065e-01 -9.73103225e-01 1.42963320e-01
3.81913722e-01 -8.28928947e-01 8.94634426e-01 -7.36044765e-01
4.61362362e-01 -1.55430108e-01 -2.35488534e-01 -1.00336611e+00
-2.33468339e-01 -3.00627559e-01 -1.76557347e-01 1.29984319e+00
7.30301499e-01 -1.05832803e+00 5.18149137e-01 5.90301275e-01
-1.06306218e-01 -5.29409051e-01 -8.01347136e-01 -8.40309978e-01
1.98463038e-01 -4.31448877e-01 6.75321341e-01 1.14112282e+00
3.27196121e-01 -2.18950570e-01 -3.18771265e-02 3.06554288e-01
4.35968816e-01 -2.82657027e-01 2.72363126e-01 -9.63736653e-01
-3.53005171e-01 -3.55604678e-01 -9.28172231e-01 -2.85167992e-02
3.60515654e-01 -8.69656265e-01 -3.65346879e-01 -1.25271356e+00
6.69838637e-02 -8.41618329e-02 -6.77278161e-01 1.04712641e+00
3.58132496e-02 -1.90823659e-01 -2.18977556e-01 -1.56600460e-01
1.09115742e-01 3.01222533e-01 2.09536031e-01 -4.45589900e-01
-1.03665799e-01 -2.19370842e-01 -6.20740116e-01 6.94588363e-01
8.85909259e-01 -9.34719443e-01 -1.10813603e-01 -4.79458183e-01
3.46925795e-01 1.21293262e-01 2.74485588e-01 -1.03248203e+00
1.77762955e-01 1.72930852e-01 2.11475387e-01 -2.64607549e-01
-1.25085339e-01 -9.30719614e-01 -1.15510069e-01 8.84311855e-01
-7.63413822e-03 4.95945841e-01 7.59372413e-01 4.10611451e-01
-3.86829793e-01 -8.73389125e-01 3.40696871e-01 -1.16739847e-01
-8.05495441e-01 1.89478174e-01 -6.00896120e-01 -4.29064274e-01
1.20016873e+00 -4.09246892e-01 -4.60239172e-01 1.99889913e-01
-3.60397071e-01 -1.85999870e-02 4.01555002e-01 9.28244591e-01
9.94060516e-01 -1.15034103e+00 -2.48891562e-01 5.56556642e-01
2.50025928e-01 -6.90240860e-01 1.27028465e-01 9.05458108e-02
-6.96886241e-01 4.37286258e-01 -5.30020297e-01 1.68983594e-01
-1.51112533e+00 1.10033894e+00 3.01158428e-01 -7.01235533e-02
-4.79610413e-01 6.50694728e-01 -2.11548552e-01 -6.65679812e-01
1.33084893e-01 -4.92198139e-01 -4.83820975e-01 -3.90841842e-01
8.19177985e-01 2.70101368e-01 9.32947695e-02 -3.77547413e-01
-6.31727576e-01 5.23998022e-01 -1.54571697e-01 2.18285397e-01
1.43258345e+00 6.64289832e-01 -6.76158071e-01 5.23399934e-02
1.67427588e+00 -1.20343499e-01 -3.44106585e-01 3.32865342e-02
4.07204777e-01 -3.45730036e-01 2.27906272e-01 -7.33463645e-01
-1.14561760e+00 1.19801331e+00 8.80436361e-01 4.09584522e-01
9.28036392e-01 -3.69959652e-01 9.41900790e-01 5.26872575e-01
2.86646426e-01 -7.10629404e-01 9.62947775e-03 5.90530455e-01
3.41966808e-01 -1.05815518e+00 -4.71280783e-01 9.15647522e-02
-5.39212599e-02 1.51563978e+00 6.62982941e-01 -2.51679450e-01
9.13776755e-01 6.95984602e-01 2.08070427e-02 -2.68844515e-01
-6.04918063e-01 4.46205527e-01 1.47044078e-01 6.50648773e-01
5.77837050e-01 -2.11300805e-01 -9.02613699e-02 8.07552099e-01
-3.44712399e-02 -2.49206081e-01 8.75397801e-01 1.25244772e+00
-5.75780809e-01 -1.36846399e+00 -4.89051610e-01 4.23016489e-01
-7.58742750e-01 -3.39629591e-01 -2.91768700e-01 5.93472660e-01
1.10601179e-01 9.20696974e-01 -3.03009510e-01 -9.30591464e-01
3.38675797e-01 2.38015145e-01 -6.60653189e-02 -8.64771664e-01
-1.15236914e+00 -5.88003457e-01 -1.53462619e-01 -4.56160694e-01
3.95635396e-01 -6.09431081e-02 -1.18118954e+00 -1.93570614e-01
-4.18507308e-01 2.79852450e-01 1.02012467e+00 7.76325047e-01
4.88431841e-01 8.19778204e-01 4.88227725e-01 -3.38713199e-01
-7.04964340e-01 -1.01732075e+00 -4.07914966e-02 2.09910437e-01
4.05133545e-01 -3.95938069e-01 -4.33073103e-01 -5.11538237e-03] | [7.057379245758057, 7.776902675628662] |
83e709f5-9328-436f-a445-24839f027d2b | dcase-2022-challenge-task-6b-language-based | 2206.06108 | null | https://arxiv.org/abs/2206.06108v3 | https://arxiv.org/pdf/2206.06108v3.pdf | Language-based Audio Retrieval Task in DCASE 2022 Challenge | Language-based audio retrieval is a task, where natural language textual captions are used as queries to retrieve audio signals from a dataset. It has been first introduced into DCASE 2022 Challenge as Subtask 6B of task 6, which aims at developing computational systems to model relationships between audio signals and free-form textual descriptions. Compared with audio captioning (Subtask 6A), which is about generating audio captions for audio signals, language-based audio retrieval (Subtask 6B) focuses on ranking audio signals according to their relevance to natural language textual captions. In DCASE 2022 Challenge, the provided baseline system for Subtask 6B was significantly outperformed, with top performance being 0.276 in mAP@10. This paper presents the outcome of Subtask 6B in terms of submitted systems' performance and analysis. | ['Tuomas Virtanen', 'Samuel Lipping', 'Huang Xie'] | 2022-06-13 | null | null | null | null | ['audio-captioning'] | ['audio'] | [ 5.26318312e-01 3.50779714e-03 3.26706320e-01 -1.52614221e-01
-2.22903609e+00 -7.69491971e-01 6.94716454e-01 3.31521034e-01
-1.24502957e-01 6.25048339e-01 7.93061554e-01 1.52375758e-01
-1.24251775e-01 -8.87225196e-02 -8.34304810e-01 5.88714750e-03
-5.26182234e-01 4.86790329e-01 2.84027904e-01 -3.71710271e-01
1.96967900e-01 -9.81240198e-02 -1.68885517e+00 1.02907479e+00
-1.23651095e-01 1.40490067e+00 8.01232308e-02 1.12962174e+00
-7.62639195e-03 8.51808488e-01 -9.08112288e-01 -3.57993357e-02
-2.80507833e-01 -5.28202057e-01 -1.20131516e+00 -4.98531967e-01
6.49591446e-01 2.05231551e-03 -4.55616981e-01 6.99238002e-01
8.73332977e-01 3.13365832e-02 7.33211160e-01 -1.59778821e+00
-4.17724162e-01 9.48990941e-01 2.36279126e-02 3.19822669e-01
1.41089511e+00 -1.81280330e-01 1.55241287e+00 -1.17101038e+00
4.10119921e-01 1.20745897e+00 5.08554876e-01 6.64393842e-01
-9.76290762e-01 -8.18277657e-01 -7.02079237e-02 4.17240053e-01
-1.97679102e+00 -7.83732593e-01 6.59484327e-01 -6.46821916e-01
8.74546826e-01 8.49762321e-01 1.99054807e-01 1.09470189e+00
-4.37400013e-01 1.06779528e+00 3.22510213e-01 -4.86049920e-01
7.32701868e-02 3.53226811e-02 -1.97758321e-02 -3.42067987e-01
-4.83964264e-01 -3.23841929e-01 -1.26400685e+00 -4.85896260e-01
1.09071225e-01 -9.93291974e-01 -4.01604593e-01 4.74537432e-01
-1.48356318e+00 4.75140542e-01 9.41673890e-02 2.05745026e-01
-4.85512227e-01 5.22103071e-01 8.74556780e-01 5.03017962e-01
2.49183983e-01 9.38253164e-01 -1.26738504e-01 -4.66654927e-01
-1.03859591e+00 6.93608940e-01 6.01137996e-01 1.16101158e+00
5.09195924e-02 2.32970379e-02 -8.98730099e-01 1.18845522e+00
3.45844448e-01 5.66711366e-01 5.99102855e-01 -8.55997562e-01
7.37483561e-01 -9.72164869e-02 4.00237679e-01 -8.72927010e-01
-6.86256513e-02 -3.16299736e-01 -3.93103659e-01 -7.91057169e-01
-8.34623799e-02 -1.25468671e-01 -4.97646332e-01 1.62642157e+00
-3.30589056e-01 2.06260458e-01 2.49767393e-01 9.68757391e-01
1.43311703e+00 1.34265780e+00 3.89900394e-02 -2.15694189e-01
1.42379510e+00 -7.23743021e-01 -7.14814246e-01 -2.17146978e-01
2.40866423e-01 -1.17793143e+00 1.06118953e+00 5.72357118e-01
-1.27813983e+00 -6.78751588e-01 -8.92031848e-01 1.18897922e-01
-1.39580071e-01 -4.97119687e-02 1.38169779e-02 2.50336975e-01
-1.41553307e+00 -2.32354268e-01 -1.12549834e-01 -1.95615128e-01
-1.84081450e-01 4.90896925e-02 -1.93584427e-01 2.50923246e-01
-1.89459693e+00 2.22015366e-01 4.32935923e-01 -7.89416954e-02
-1.52808523e+00 -1.06205523e+00 -7.17485368e-01 1.46207407e-01
3.00004393e-01 -2.35206783e-01 1.92092216e+00 -4.40759033e-01
-1.20787275e+00 7.44758368e-01 -8.71755779e-02 -8.87010276e-01
2.05495253e-01 -3.86289299e-01 -7.75727212e-01 5.60856819e-01
1.82601705e-01 1.23086596e+00 7.20813513e-01 -1.12503207e+00
-6.71051264e-01 3.66785169e-01 6.24952205e-02 3.40420991e-01
-3.05856496e-01 5.40940821e-01 -6.29485846e-01 -8.17319095e-01
-1.33899584e-01 -8.01057220e-01 2.08573148e-01 -3.00637364e-01
-4.64749098e-01 -6.19021595e-01 7.20667362e-01 -6.82341278e-01
1.64064419e+00 -2.35974979e+00 -1.84078082e-01 8.92703906e-02
-1.94279030e-01 1.35100260e-02 -4.73107010e-01 8.98684859e-01
-3.01679134e-01 2.78649569e-01 4.72479761e-02 -1.19255953e-01
5.16555548e-01 -5.82727194e-01 -1.07462990e+00 -3.23523998e-01
3.78594071e-01 7.22477615e-01 -1.08631325e+00 -5.86136162e-01
-2.39474669e-01 4.88947898e-01 -3.28749597e-01 4.29872602e-01
-4.56742495e-01 4.12915573e-02 -3.28906119e-01 5.10872126e-01
6.22659959e-02 1.35414705e-01 -4.16344434e-01 -1.62670743e-02
1.45844474e-01 7.80256331e-01 -9.54203606e-01 1.76770306e+00
-4.59375679e-01 1.19196057e+00 5.08797206e-02 -5.47142982e-01
1.10846937e+00 1.28972328e+00 7.34567404e-01 -6.01901591e-01
-5.06209970e-01 1.69287667e-01 -5.27349949e-01 -6.38189375e-01
6.50346994e-01 -2.17903219e-03 -8.77409458e-01 4.01904076e-01
4.70831990e-02 -8.12648833e-01 2.09609538e-01 4.20994610e-01
1.21294487e+00 -1.21135533e-01 -1.44425586e-01 7.80652538e-02
8.14787865e-01 1.88040594e-03 -2.06111595e-01 8.54659140e-01
-1.70658335e-01 1.27941072e+00 1.87688738e-01 6.33854568e-02
-6.76090956e-01 -1.08989501e+00 1.75084293e-01 1.42272902e+00
-2.03599200e-01 -8.19859684e-01 -4.94455040e-01 8.94480012e-03
-3.02374095e-01 6.11262500e-01 -2.99079269e-01 -1.41797662e-01
-3.23760420e-01 -6.81364164e-02 1.33291781e+00 2.70418823e-01
1.69589505e-01 -1.32802534e+00 -3.30254734e-01 4.12287146e-01
-9.24281180e-01 -1.27605176e+00 -9.59820688e-01 -3.86707508e-03
-3.39857548e-01 -7.70298600e-01 -1.12023020e+00 -1.06443202e+00
-1.36539340e-01 9.50850323e-02 1.34654582e+00 -3.49861979e-01
-2.72548616e-01 9.91032600e-01 -7.66609013e-01 -8.32154036e-01
-3.59231919e-01 3.69013190e-01 -4.84641045e-02 8.62188712e-02
2.96519428e-01 -4.76308733e-01 -3.85678977e-01 2.69871742e-01
-1.12750554e+00 -2.33555242e-01 3.33316833e-01 4.75384325e-01
5.13521791e-01 -3.18103522e-01 1.09358215e+00 -3.87942344e-02
1.34168839e+00 -4.40546423e-01 -1.46378800e-01 1.27567485e-01
1.49145373e-03 -3.30416828e-01 4.79224712e-01 -5.33015966e-01
-4.75230455e-01 2.89234877e-01 -1.54015511e-01 -3.67543101e-01
-1.91517934e-01 7.38398314e-01 1.68955103e-02 4.67221469e-01
8.34211648e-01 5.04647791e-01 -3.81755531e-01 -5.63702226e-01
6.62454888e-02 1.16697228e+00 9.71125245e-01 -7.78926253e-01
7.35840619e-01 -2.15505555e-01 -4.25871611e-01 -1.05146241e+00
-7.41657913e-01 -9.56766903e-01 -7.79611394e-02 -7.51365662e-01
7.74804413e-01 -1.29664505e+00 -4.84354734e-01 -1.74291376e-02
-1.40586734e+00 1.55371353e-01 -2.58226305e-01 5.21855474e-01
-8.35604310e-01 -1.05891027e-01 -3.94214809e-01 -1.09456933e+00
-5.67919374e-01 -8.43423605e-01 1.66555142e+00 -3.09981018e-01
-7.51204193e-01 -3.86260360e-01 2.68464208e-01 5.14315188e-01
2.97652900e-01 1.34145647e-01 6.60433888e-01 -1.05007160e+00
-1.41231254e-01 -6.04836285e-01 4.70238402e-02 3.41291696e-01
-1.44191921e-01 -3.44281882e-01 -1.45815086e+00 -1.97953552e-01
-5.84898710e-01 -9.64847624e-01 5.35859406e-01 1.51685784e-02
1.29634023e+00 -2.77206123e-01 6.71620890e-02 -2.83243328e-01
8.66174042e-01 5.08756936e-01 6.11278296e-01 1.40056729e-01
1.50911175e-02 8.07113707e-01 9.88935769e-01 4.19578850e-01
3.28847580e-02 1.17534244e+00 3.16491783e-01 3.11548263e-01
-3.17983925e-01 -6.31673157e-01 7.13930070e-01 1.32882738e+00
4.78364199e-01 -2.69255400e-01 -1.15697789e+00 9.28298771e-01
-1.62847972e+00 -1.00228393e+00 -8.77463892e-02 1.99874878e+00
1.10506940e+00 1.78079963e-01 4.40719157e-01 4.98480916e-01
8.19556952e-01 8.32589492e-02 -2.01307923e-01 -3.05870444e-01
1.11920089e-01 -8.14193918e-04 -3.46119672e-01 4.82165664e-01
-1.05389345e+00 4.21031475e-01 7.06842184e+00 1.11278725e+00
-9.58575785e-01 -7.74790533e-03 4.11186874e-01 -2.25286424e-01
-1.56313702e-01 -4.52800035e-01 -4.79926616e-01 3.61838222e-01
1.45261717e+00 -7.32250929e-01 2.37015575e-01 3.92085224e-01
4.06110317e-01 3.81603628e-01 -1.58021665e+00 1.22621667e+00
3.12597096e-01 -1.04281044e+00 6.15751207e-01 -5.57189941e-01
2.76912600e-01 -1.59880325e-01 3.18721861e-01 7.22262800e-01
-3.81628960e-01 -1.12112296e+00 1.18950856e+00 4.91287529e-01
9.67386842e-01 -6.59785986e-01 6.96046710e-01 3.23281437e-02
-1.54213035e+00 8.60719159e-02 -6.69211745e-02 -1.25396103e-02
3.79079193e-01 1.88901082e-01 -1.48525763e+00 5.16621172e-01
8.75553310e-01 2.73555905e-01 -4.76266831e-01 1.53510332e+00
1.19957551e-02 9.89173353e-01 -2.78014928e-01 -2.12160751e-01
3.82548273e-01 6.71666086e-01 8.20577204e-01 1.64245796e+00
4.79527622e-01 -1.01603210e-01 1.08826019e-01 4.18118507e-01
-2.83418924e-01 3.68282944e-01 -5.44908047e-01 -4.42479610e-01
8.11390281e-01 6.59263849e-01 -1.63483828e-01 -2.92329520e-01
6.73359483e-02 4.83757019e-01 -5.99387586e-01 3.47047806e-01
-7.70656645e-01 -8.53903294e-01 2.47333035e-01 3.55184257e-01
-6.01225272e-02 2.21939936e-01 4.43190455e-01 -5.47814071e-01
1.75570562e-01 -1.00978053e+00 5.31997859e-01 -1.57772541e+00
-1.23142326e+00 1.00704074e+00 4.38111871e-01 -1.81652391e+00
-7.64057159e-01 -1.39258146e-01 -1.77564979e-01 6.44778907e-01
-1.12957847e+00 -8.06909084e-01 -9.34102014e-02 5.52273452e-01
1.03979349e+00 -2.63960689e-01 1.12823486e+00 8.32831621e-01
2.93197930e-01 6.75887167e-01 -2.41930187e-01 1.11979105e-01
1.05046010e+00 -9.50337052e-01 3.09614033e-01 2.53691703e-01
5.61943293e-01 1.41101599e-01 1.11775994e+00 -1.70622259e-01
-1.16190088e+00 -1.27068758e+00 1.41002083e+00 -4.18913960e-01
1.06222892e+00 -5.85713267e-01 -6.80760562e-01 3.00389081e-01
3.08043778e-01 -3.46572220e-01 8.53231370e-01 -2.28465170e-01
-4.21725035e-01 -3.44034940e-01 -5.62289774e-01 2.64335692e-01
8.01710248e-01 -1.27908397e+00 -8.69742990e-01 6.45077705e-01
1.30443037e+00 -2.84476668e-01 -9.72753406e-01 3.64361018e-01
5.53535223e-01 -1.62036193e-03 1.18998981e+00 -5.77806890e-01
3.75044882e-01 -4.48766083e-01 -5.14582515e-01 -9.43260193e-01
1.43108383e-01 -1.19873893e+00 1.64635658e-01 1.53005981e+00
8.33230376e-01 1.93107113e-01 4.15824682e-01 2.73153167e-02
-2.59737998e-01 -5.17175049e-02 -1.29401731e+00 -8.90894055e-01
-2.07658142e-01 -9.06733692e-01 6.42567635e-01 5.88520408e-01
3.14605623e-01 6.90746307e-01 -4.87736821e-01 2.67663561e-02
1.29686529e-02 -1.42245963e-01 6.19734943e-01 -1.26201260e+00
-1.08038045e-01 -3.77393991e-01 -6.27443790e-01 -1.05758619e+00
2.94036176e-02 -9.20787752e-01 7.35467136e-01 -1.38936102e+00
2.30883304e-02 7.59181455e-02 -5.32546997e-01 4.89343822e-01
2.93939590e-01 9.43277478e-01 4.21877295e-01 1.90040529e-01
-1.10229349e+00 3.82049471e-01 7.24597454e-01 -7.88958073e-01
-3.21319848e-01 3.27327281e-01 -7.13618815e-01 1.09792687e-01
5.91363847e-01 -4.94057417e-01 -5.46710610e-01 -3.42649668e-01
4.40709263e-01 5.73338151e-01 2.59649396e-01 -1.24600923e+00
4.34495449e-01 1.77315325e-01 -3.30435187e-01 -8.83880973e-01
7.85911024e-01 -6.58941805e-01 1.95339695e-01 9.34475958e-02
-1.23061156e+00 1.57152951e-01 4.30779576e-01 3.76194447e-01
-1.07805860e+00 -3.50279838e-01 3.05087622e-02 2.41909131e-01
-4.16344136e-01 5.21486811e-03 -9.60709691e-01 4.14811075e-01
6.01545095e-01 1.77436069e-01 5.48370034e-02 -1.27600443e+00
-1.04874635e+00 3.46264184e-01 -6.14997804e-01 7.49037266e-01
9.72123981e-01 -1.79520571e+00 -1.29550898e+00 -4.57759023e-01
6.96657836e-01 -1.75222754e-01 1.32465698e-02 3.00435185e-01
-2.00720802e-01 1.19127977e+00 1.54825538e-01 -7.46428549e-01
-1.38422477e+00 7.71473870e-02 7.84217492e-02 -5.83010167e-02
-5.92155345e-02 9.89798069e-01 2.30038026e-03 5.86312376e-02
9.98001814e-01 -2.48749867e-01 -4.51244980e-01 3.09786797e-01
8.45514059e-01 7.67092332e-02 2.37678573e-01 -7.74904668e-01
-4.37534690e-01 2.60475188e-01 2.25573212e-01 -8.83439183e-01
1.09971607e+00 -5.17804995e-02 8.17322433e-02 9.68412519e-01
1.26331449e+00 -2.13173360e-01 -4.74518538e-01 -1.51311591e-01
4.62794155e-01 7.64918700e-02 1.55896749e-02 -1.03972578e+00
-3.37777734e-01 1.02228010e+00 5.55307269e-01 5.43318689e-01
1.26555884e+00 2.08267972e-01 8.89370203e-01 6.90546811e-01
3.94344866e-01 -1.00255728e+00 4.19801533e-01 6.82675004e-01
1.62758195e+00 -8.14351797e-01 -3.42530102e-01 -8.23354200e-02
-5.76914430e-01 9.64805603e-01 5.24723589e-01 2.77006626e-01
3.67073536e-01 -1.20913815e-02 1.70678005e-01 -5.64359240e-02
-1.12900722e+00 6.97290748e-02 8.94234598e-01 6.26696467e-01
5.84214568e-01 -4.44118157e-02 9.68095567e-03 9.08088446e-01
-5.48310876e-01 -1.06585458e-01 2.56684422e-01 7.54230678e-01
-4.38759029e-01 -8.59508157e-01 -7.04690337e-01 7.55482242e-02
-5.53149402e-01 -2.34516636e-01 -1.01249397e+00 4.09720093e-01
-4.38153774e-01 1.32983983e+00 -4.33579162e-02 -5.64798117e-01
5.52159429e-01 4.33329254e-01 -2.45698467e-02 -8.26422930e-01
-8.53032291e-01 2.06669658e-01 4.02120113e-01 -3.89054239e-01
-3.54580462e-01 -4.10532385e-01 -9.14302886e-01 6.70947790e-01
-5.34006208e-02 1.22155726e+00 5.80676258e-01 4.75608379e-01
2.52975583e-01 5.70906460e-01 7.07513034e-01 -7.63763666e-01
-4.46468711e-01 -1.20356357e+00 -2.97295988e-01 1.87507853e-01
5.26764333e-01 -1.08810633e-01 -4.01864350e-01 4.01100606e-01] | [15.262482643127441, 4.8621649742126465] |
2a1f3469-e17e-45d9-8157-815139ec9f95 | gensf-simultaneous-adaptation-of-generative | 2106.07055 | null | https://arxiv.org/abs/2106.07055v1 | https://arxiv.org/pdf/2106.07055v1.pdf | GenSF: Simultaneous Adaptation of Generative Pre-trained Models and Slot Filling | In transfer learning, it is imperative to achieve strong alignment between a pre-trained model and a downstream task. Prior work has done this by proposing task-specific pre-training objectives, which sacrifices the inherent scalability of the transfer learning paradigm. We instead achieve strong alignment by simultaneously modifying both the pre-trained model and the formulation of the downstream task, which is more efficient and preserves the scalability of transfer learning. We present GenSF (Generative Slot Filling), which leverages a generative pre-trained open-domain dialog model for slot filling. GenSF (1) adapts the pre-trained model by incorporating inductive biases about the task and (2) adapts the downstream task by reformulating slot filling to better leverage the pre-trained model's capabilities. GenSF achieves state-of-the-art results on two slot filling datasets with strong gains in few-shot and zero-shot settings. We achieve a 9 F1 score improvement in zero-shot slot filling. This highlights the value of strong alignment between the pre-trained model and the downstream task. | ['Maxine Eskenazi', 'Shikib Mehri'] | 2021-06-13 | null | https://aclanthology.org/2021.sigdial-1.51 | https://aclanthology.org/2021.sigdial-1.51.pdf | sigdial-acl-2021-7 | ['zero-shot-slot-filling', 'open-domain-dialog'] | ['natural-language-processing', 'natural-language-processing'] | [ 3.79038692e-01 9.06871498e-01 -3.02087575e-01 -5.74045300e-01
-1.04674494e+00 -3.75566244e-01 7.64531076e-01 -3.53983082e-02
-5.52271962e-01 6.91386461e-01 6.43929482e-01 -5.81236959e-01
2.49154299e-01 -7.91597545e-01 -5.93588352e-01 -3.00755471e-01
2.56158829e-01 8.90337408e-01 3.68400544e-01 -3.79918694e-01
1.62436105e-02 -4.02450413e-01 -1.16443849e+00 4.17693555e-01
9.37494159e-01 7.82101631e-01 5.23642480e-01 5.01137972e-01
-3.72302443e-01 6.16689324e-01 -4.03696477e-01 -3.29120249e-01
2.00408906e-01 -6.44545734e-01 -1.14711761e+00 2.96547502e-01
1.62887052e-01 -4.05202985e-01 -1.82767034e-01 5.29333472e-01
3.61220717e-01 3.25845242e-01 5.56838751e-01 -1.08697450e+00
-8.44973147e-01 7.39963710e-01 -1.76260188e-01 7.16533959e-02
-1.17275901e-01 4.35161330e-02 1.21024573e+00 -8.37141395e-01
5.95071018e-01 1.17820191e+00 6.97397530e-01 8.25090289e-01
-1.45852447e+00 -6.05142653e-01 2.58543164e-01 -2.42520541e-01
-8.58720362e-01 -6.15742862e-01 4.27849710e-01 -4.64472860e-01
1.17863786e+00 -2.56110877e-01 3.07017982e-01 1.06331229e+00
-1.28953084e-01 9.19277668e-01 9.56199408e-01 -6.92957163e-01
4.11048502e-01 2.21683338e-01 1.66664675e-01 3.72249842e-01
-1.82670295e-01 1.25308976e-01 -6.06042862e-01 -1.40276819e-01
7.13035226e-01 -1.10021405e-01 1.61647454e-01 -5.88570535e-01
-9.09487665e-01 1.19954598e+00 2.98351973e-01 2.97316760e-01
-1.32858753e-01 -5.63094951e-02 5.35295069e-01 2.69375175e-01
8.30131054e-01 5.91562331e-01 -6.74127400e-01 -3.23115021e-01
-1.08514822e+00 1.98311687e-01 8.72080743e-01 1.04382479e+00
8.70543718e-01 -1.16430573e-01 -6.77912056e-01 9.68213141e-01
3.99850726e-01 9.45113599e-02 6.30126536e-01 -9.46185052e-01
5.53891718e-01 3.95918310e-01 -3.68623110e-03 1.37503102e-01
-1.66062981e-01 -3.71646106e-01 -2.98570469e-02 -1.02969550e-01
2.55925804e-01 -2.97663361e-01 -1.38251078e+00 2.17829490e+00
3.78669292e-01 1.99892186e-02 4.78117675e-01 5.91191828e-01
8.17551017e-01 6.13444090e-01 8.18391263e-01 1.99128836e-01
1.37844610e+00 -1.53025949e+00 -6.88805938e-01 -9.67018306e-01
8.93445134e-01 -5.61852098e-01 1.42762399e+00 -3.82715017e-01
-9.88667786e-01 -5.36063671e-01 -9.40884233e-01 -5.22204101e-01
-2.07412645e-01 -1.41999125e-02 6.12170160e-01 5.02093077e-01
-1.16916800e+00 4.43453282e-01 -8.16350818e-01 -4.53242332e-01
4.41448599e-01 2.14974210e-01 -2.54841387e-01 -1.00069240e-01
-1.30702984e+00 9.73005176e-01 4.25373048e-01 -5.80092132e-01
-8.64412010e-01 -1.02768350e+00 -1.24255419e+00 3.34350169e-01
4.93459195e-01 -8.59389007e-01 1.99098516e+00 -6.54564857e-01
-1.64092910e+00 1.03473389e+00 -2.05710441e-01 -4.23167109e-01
4.06094760e-01 -3.31133217e-01 9.58750993e-02 -2.95980364e-01
4.17734236e-01 1.18092144e+00 6.04853451e-01 -1.04571068e+00
-5.30638754e-01 -2.86844492e-01 3.93183418e-02 5.61281621e-01
-3.92781436e-01 -4.31269258e-01 -4.83944446e-01 -4.48101789e-01
-4.36169803e-02 -9.00556266e-01 -2.45493323e-01 -2.76205003e-01
-2.46813729e-01 -3.02764684e-01 8.82770181e-01 -7.09684074e-01
1.07763588e+00 -2.26824737e+00 -9.35563073e-02 -4.22368914e-01
1.49758328e-02 4.62404042e-01 -4.76554930e-01 6.79077029e-01
9.64643136e-02 -1.08983152e-01 -4.27758455e-01 -1.12818873e+00
2.72701681e-01 5.20121336e-01 -5.64364076e-01 -4.88016047e-02
2.54192412e-01 1.21921659e+00 -8.44070196e-01 -3.96890551e-01
9.51293036e-02 3.11096966e-01 -9.19110835e-01 6.83595002e-01
-6.02587819e-01 5.41747391e-01 -2.82437384e-01 2.59921193e-01
3.96248817e-01 -5.12090743e-01 4.17958766e-01 1.07505068e-01
1.74296156e-01 8.41829598e-01 -6.32965446e-01 2.16841841e+00
-7.22182989e-01 3.16901296e-01 4.86283228e-02 -1.00395501e+00
9.56735432e-01 4.46753025e-01 2.99308419e-01 -1.00265634e+00
-2.85403361e-03 1.29683644e-01 -1.64176226e-01 -2.10034996e-01
8.07442844e-01 -4.97624755e-01 -3.96656066e-01 7.81134903e-01
6.84373677e-01 -1.21436387e-01 -1.32827193e-01 3.24817687e-01
9.49205518e-01 5.19934654e-01 3.61431390e-01 -2.35031232e-01
-1.16641782e-01 7.05658719e-02 5.92515111e-01 7.37852454e-01
-2.22678110e-01 5.44827104e-01 3.00849974e-01 4.99641486e-02
-1.17946219e+00 -1.27089739e+00 -5.52806035e-02 1.95443344e+00
-3.81786004e-02 -3.75493079e-01 -7.99193561e-01 -8.52101505e-01
2.18679577e-01 8.40983927e-01 -8.11451733e-01 -3.09947312e-01
-3.24498206e-01 -3.40689391e-01 4.66269225e-01 8.09065580e-01
6.07005894e-01 -1.04610598e+00 -5.07807612e-01 3.21943998e-01
-2.51029104e-01 -1.25202405e+00 -4.86908168e-01 5.66357255e-01
-1.02769947e+00 -5.37507832e-01 -6.20937943e-01 -8.09778929e-01
5.86436927e-01 1.58144772e-01 1.47583306e+00 -1.51326014e-02
6.06040694e-02 1.43750459e-01 -4.23549712e-01 -3.10088962e-01
-5.23024261e-01 5.16767025e-01 -3.56059790e-01 -4.79889780e-01
6.84204817e-01 -5.28421223e-01 -3.58187169e-01 2.98287451e-01
-8.13192368e-01 4.21907127e-01 6.67823136e-01 1.10814559e+00
2.15544477e-01 -6.65895998e-01 8.90364826e-01 -1.34073460e+00
6.48683250e-01 -6.27056360e-01 -3.17487806e-01 1.99745834e-01
-9.36802804e-01 2.24353477e-01 1.19175576e-01 -2.49585867e-01
-1.50005126e+00 2.29675565e-02 -1.51245192e-01 -1.44988984e-01
-8.45199823e-02 4.01203692e-01 -1.35360911e-01 4.55167800e-01
7.54253089e-01 -7.02822208e-02 2.62639541e-02 -7.04843879e-01
6.77133918e-01 8.45765769e-01 7.00431168e-01 -8.29817653e-01
5.66156447e-01 2.11780384e-01 -6.27879381e-01 -3.74166667e-01
-1.30429959e+00 -6.12454414e-01 -6.25986874e-01 3.28542292e-01
9.66838062e-01 -1.07712495e+00 -1.73050120e-01 1.97804093e-01
-9.28873956e-01 -1.22005451e+00 -6.24401093e-01 6.16371743e-02
-6.71755552e-01 -6.15540221e-02 -8.67404759e-01 -7.19610751e-01
-4.26345140e-01 -8.23965132e-01 1.36660576e+00 2.88783848e-01
-4.06254321e-01 -1.39900243e+00 4.01277453e-01 6.43927455e-01
8.77885520e-01 -3.86017799e-01 1.01847577e+00 -1.16491950e+00
-2.51736969e-01 3.93526286e-01 -3.40749174e-01 1.02226399e-01
1.05793424e-01 -7.93617725e-01 -1.35088503e+00 -2.11631775e-01
-3.43114771e-02 -8.46725702e-01 1.06858432e+00 2.86117941e-01
6.11544788e-01 -6.39849380e-02 -3.46689999e-01 3.28863621e-01
9.83372331e-01 3.88219990e-02 5.92857897e-01 5.17227888e-01
2.88642764e-01 7.19331920e-01 7.66740203e-01 2.23368287e-01
8.21049869e-01 7.33902752e-01 -2.73981616e-02 -4.53516841e-01
-1.98757693e-01 -9.38965619e-01 1.64827466e-01 4.44180548e-01
5.98595202e-01 -3.05443592e-02 -8.90403450e-01 7.64955401e-01
-2.15125155e+00 -5.68990648e-01 6.21711016e-01 1.91552293e+00
1.28508234e+00 1.85133457e-01 2.41549741e-02 -3.15722734e-01
4.54198390e-01 3.74394238e-01 -7.84848690e-01 -4.85263973e-01
3.62690747e-01 5.61515510e-01 2.94163883e-01 7.50115931e-01
-1.18016458e+00 1.66278791e+00 6.87635660e+00 6.67526722e-01
-9.67441916e-01 5.25479555e-01 6.58357263e-01 8.73850659e-02
-5.40786564e-01 3.46009314e-01 -1.06797242e+00 1.95762545e-01
1.14424956e+00 -2.18054086e-01 2.08087787e-01 9.43393290e-01
-2.13961080e-01 -2.09810823e-01 -1.32828736e+00 3.23194355e-01
-5.70104867e-02 -1.16769004e+00 -1.17881194e-01 2.48055816e-01
7.22722590e-01 1.44703180e-01 1.73226550e-01 1.28150225e+00
8.25883031e-01 -1.07355905e+00 5.60652018e-01 4.31745835e-02
6.69812739e-01 -2.52522051e-01 4.85601246e-01 4.26712364e-01
-8.80255997e-01 7.33693466e-02 -3.06996405e-01 -2.71141171e-01
5.09228349e-01 4.12328929e-01 -1.24233842e+00 1.59358442e-01
2.86272436e-01 3.90307009e-01 -3.91559452e-01 5.89066744e-01
-3.62780422e-01 7.16731012e-01 -7.83298388e-02 5.18664658e-01
4.06417578e-01 -7.65559301e-02 3.38737220e-02 1.18931937e+00
5.40326163e-02 2.62579769e-02 5.60005248e-01 1.08714151e+00
-1.60667345e-01 -2.96268128e-02 -4.64656442e-01 -3.11304271e-01
6.98842227e-01 1.13660979e+00 -4.72435445e-01 -4.95633483e-01
-5.41154563e-01 9.58615124e-01 5.48565269e-01 3.21769595e-01
-5.72180927e-01 -8.28744844e-02 7.80891061e-01 -2.63615400e-02
5.36027133e-01 -1.80032387e-01 -6.02507651e-01 -1.06117535e+00
-3.60789299e-01 -5.63335419e-01 5.59205055e-01 -5.84688902e-01
-1.03621233e+00 5.15986741e-01 4.37662005e-02 -6.99326336e-01
-7.58171201e-01 -3.14224422e-01 -7.04498827e-01 1.14897621e+00
-1.76236284e+00 -1.50015903e+00 -1.15075618e-01 3.97579134e-01
8.72669697e-01 -3.43065336e-02 1.20721602e+00 1.48968637e-01
-1.78072035e-01 6.21544123e-01 -2.39784226e-01 1.13197163e-01
9.54061091e-01 -1.30814672e+00 9.96578455e-01 6.56969607e-01
3.92248407e-02 4.42076117e-01 6.77885354e-01 -8.24518323e-01
-7.27399707e-01 -1.11966681e+00 1.20209420e+00 -4.52970594e-01
5.09618521e-01 -7.01713681e-01 -1.03856134e+00 1.15777779e+00
2.45786890e-01 -3.42360735e-02 1.02403152e+00 6.26858950e-01
-5.59909642e-01 4.90188956e-01 -1.03500533e+00 4.18152422e-01
1.07027876e+00 -9.45044816e-01 -1.08268917e+00 2.02304944e-01
1.11225069e+00 -7.38395631e-01 -7.10890412e-01 3.11587006e-01
2.46687576e-01 -8.04043829e-01 8.65476787e-01 -7.87880838e-01
5.95988393e-01 3.79827142e-01 -1.90237373e-01 -1.46696198e+00
-3.15346986e-01 -3.74892354e-01 -1.73023418e-01 1.26073241e+00
6.68891191e-01 -4.88122195e-01 1.05842936e+00 1.07177460e+00
-4.26087677e-01 -6.65314615e-01 -7.13931501e-01 -8.11841547e-01
4.26774174e-01 -1.07659347e-01 4.91042912e-01 9.31806505e-01
3.58496547e-01 7.71618307e-01 -3.09764832e-01 -2.82995105e-01
1.77193418e-01 1.44242316e-01 9.32347953e-01 -1.16466236e+00
-7.09362507e-01 -7.87767991e-02 3.32182169e-01 -1.39474106e+00
2.57489920e-01 -9.55540955e-01 3.29091460e-01 -1.60464752e+00
3.15526605e-01 -8.30911458e-01 -2.41263956e-01 9.50755477e-01
-4.16489810e-01 9.19704288e-02 3.34960669e-01 3.33309919e-01
-7.09937811e-01 7.75608420e-01 1.13954902e+00 2.49492660e-01
-5.38264394e-01 -1.58183962e-01 -1.12700617e+00 3.06704670e-01
6.20493293e-01 -3.96780133e-01 -7.14093149e-01 -6.55852854e-01
-3.28333914e-01 -1.75851230e-02 -4.23176549e-02 -6.90563023e-01
1.53870925e-01 1.39703630e-02 7.47316256e-02 4.64989208e-02
6.26798093e-01 -1.88776314e-01 -2.43871987e-01 2.72206455e-01
-7.63096631e-01 -3.34434628e-01 5.07652462e-01 4.86987352e-01
-1.64856479e-01 -3.32370996e-01 7.06057608e-01 -1.77879840e-01
-7.97016144e-01 1.06572993e-01 -5.49126901e-02 4.83380169e-01
7.57809460e-01 -3.53053808e-02 -5.90287030e-01 -4.23506320e-01
-8.92492294e-01 3.69619489e-01 5.35569966e-01 5.80623090e-01
1.19285407e-02 -1.24706757e+00 -3.90665829e-01 3.49194378e-01
2.93296158e-01 9.72217321e-02 1.55574962e-01 6.62345290e-01
2.55020887e-01 5.61177075e-01 -2.76576847e-01 -5.36421120e-01
-9.54175830e-01 2.24061996e-01 2.35547021e-01 -6.52003944e-01
-6.42892480e-01 8.84400249e-01 4.61912334e-01 -9.76831317e-01
1.67240262e-01 -9.81536210e-02 -1.86784212e-02 -5.82470782e-02
3.15466881e-01 -1.59111828e-01 7.20867962e-02 -2.92822003e-01
-8.14407226e-03 -1.24323003e-01 -3.99370372e-01 -5.51724553e-01
1.39670646e+00 -2.08609357e-01 4.47734296e-01 4.05673712e-01
1.07581031e+00 -3.38928223e-01 -1.73131180e+00 -8.13274741e-01
1.69001222e-01 -3.27716857e-01 1.59597591e-01 -1.02079761e+00
-6.28219485e-01 1.02065825e+00 2.63344258e-01 -1.94211408e-01
6.73071742e-01 2.46893376e-01 1.19000959e+00 3.07317734e-01
3.71360809e-01 -1.34154260e+00 3.87428671e-01 8.61926079e-01
4.68154967e-01 -1.36695874e+00 -4.33261216e-01 -3.47295046e-01
-9.10347044e-01 6.56900644e-01 8.64734232e-01 8.24016985e-03
4.80317801e-01 2.82818437e-01 2.91603923e-01 -2.16060981e-01
-8.32518697e-01 -3.69606376e-01 1.12600334e-01 5.51291466e-01
6.96668506e-01 7.61009008e-02 -2.07623377e-01 8.87085378e-01
-2.89911389e-01 1.71976745e-01 5.09983040e-02 1.35127449e+00
-6.55779541e-01 -1.48568940e+00 3.28887440e-03 2.89228022e-01
-5.25971465e-02 -3.31172019e-01 -2.99221426e-01 6.14404678e-01
-2.23562971e-01 8.18665266e-01 2.43577585e-01 -2.45497286e-01
1.49565399e-01 6.31788373e-01 2.26424441e-01 -1.26759124e+00
-4.97384340e-01 8.90223682e-02 2.99588919e-01 -6.61042809e-01
-7.69433305e-02 -3.81437600e-01 -1.30363941e+00 3.06928921e-02
-1.05603293e-01 3.87629598e-01 6.49018705e-01 1.29704046e+00
7.72571445e-01 5.17633498e-01 1.25026032e-01 -9.13732231e-01
-8.52030933e-01 -1.13318288e+00 -4.77229208e-01 6.09192193e-01
4.13669571e-02 -9.08097684e-01 -7.12677985e-02 -2.27310884e-04] | [12.582409858703613, 7.4868011474609375] |
0ab1ab3d-098f-4aa6-b78d-d4ed4012caa8 | molecule-morphology-contrastive-pretraining | 2305.09790 | null | https://arxiv.org/abs/2305.09790v2 | https://arxiv.org/pdf/2305.09790v2.pdf | Molecule-Morphology Contrastive Pretraining for Transferable Molecular Representation | Image-based profiling techniques have become increasingly popular over the past decade for their applications in target identification, mechanism-of-action inference, and assay development. These techniques have generated large datasets of cellular morphologies, which are typically used to investigate the effects of small molecule perturbagens. In this work, we extend the impact of such dataset to improving quantitative structure-activity relationship (QSAR) models by introducing Molecule-Morphology Contrastive Pretraining (MoCoP), a framework for learning multi-modal representation of molecular graphs and cellular morphologies. We scale MoCoP to approximately 100K molecules and 600K morphological profiles using data from the JUMP-CP Consortium and show that MoCoP consistently improves performances of graph neural networks (GNNs) on molecular property prediction tasks in ChEMBL20 across all dataset sizes. The pretrained GNNs are also evaluated on internal GSK pharmacokinetic data and show an average improvement of 2.6% and 6.3% in AUPRC for full and low data regimes, respectively. Our findings suggest that integrating cellular morphologies with molecular graphs using MoCoP can significantly improve the performance of QSAR models, ultimately expanding the deep learning toolbox available for QSAR applications. | ['Kim M. Branson', 'Dante Pertusi', 'Cuong Q. Nguyen'] | 2023-04-27 | null | null | null | null | ['molecular-property-prediction'] | ['miscellaneous'] | [ 3.37702900e-01 -4.76257563e-01 -5.26704729e-01 3.36818509e-02
-7.52972543e-01 -1.10150707e+00 6.58937454e-01 8.73987317e-01
-2.97675967e-01 8.59468043e-01 -5.74044436e-02 -5.51983595e-01
-3.58281843e-02 -6.84200108e-01 -1.06502068e+00 -8.05280685e-01
-1.42685771e-01 3.88246000e-01 7.25021735e-02 -4.24117409e-02
2.21221283e-01 8.09290230e-01 -7.84534752e-01 4.92533654e-01
9.48659301e-01 9.07695055e-01 -3.06428403e-01 7.00998187e-01
8.91807526e-02 5.21805942e-01 -5.02434373e-01 -3.66622657e-01
-2.66961493e-02 -1.61028489e-01 -5.91031015e-01 -5.08523583e-01
8.95108402e-01 2.00120300e-01 -2.50498682e-01 7.76082814e-01
9.53393221e-01 1.25619024e-01 9.22550738e-01 -6.53303027e-01
-7.52420127e-01 2.00815365e-01 -5.23497820e-01 4.64558631e-01
3.76565576e-01 6.19524717e-01 9.13614273e-01 -5.88903904e-01
7.40950704e-01 1.23943377e+00 7.34336793e-01 4.27525759e-01
-1.72869587e+00 -8.21432471e-01 -1.71951622e-01 1.45095348e-01
-1.27630055e+00 -4.31853682e-01 3.16690773e-01 -7.81053305e-01
1.43148649e+00 1.22985654e-01 6.13053501e-01 1.28492701e+00
6.36596024e-01 4.34048593e-01 1.00782931e+00 1.75092459e-01
3.54049176e-01 -5.32344878e-01 -1.46621754e-02 7.53313243e-01
4.47803169e-01 1.31927609e-01 -5.58746457e-01 -3.76688510e-01
7.06867337e-01 -8.13785102e-03 -7.67290518e-02 -2.79880166e-01
-9.45932806e-01 8.84055614e-01 6.29420936e-01 -8.33988637e-02
2.14280211e-03 3.62498850e-01 4.68319893e-01 -4.61258367e-02
4.25577492e-01 1.21340787e+00 -6.41366422e-01 1.22322932e-01
-5.34820318e-01 3.66485208e-01 5.07851541e-01 5.13420999e-01
4.32787746e-01 5.54178469e-02 -8.64560902e-02 7.32292652e-01
1.97774172e-01 4.70432252e-01 3.16507697e-01 -5.41287959e-01
3.04105848e-01 8.14969599e-01 -9.42689627e-02 -9.29888487e-01
-8.47950518e-01 -5.52171588e-01 -5.87466955e-01 -1.58742756e-01
5.88324308e-01 -5.19202910e-02 -1.18185902e+00 1.65637338e+00
3.98559511e-01 2.33635768e-01 3.60394157e-02 4.74580199e-01
1.12788868e+00 5.89057446e-01 7.95736015e-01 2.62668505e-02
1.16022348e+00 -6.33678913e-01 -2.38525033e-01 7.78235123e-02
1.20026553e+00 -4.28210229e-01 9.90691245e-01 5.89152396e-01
-6.90630496e-01 -2.47064918e-01 -1.13578534e+00 -1.20880514e-01
-9.37358499e-01 -6.25399724e-02 1.02103698e+00 7.22560525e-01
-7.17031777e-01 9.68106747e-01 -7.72876143e-01 -1.03697002e-01
1.18711126e+00 9.20669436e-01 -5.21602035e-01 -2.71038972e-02
-1.05923092e+00 7.94924259e-01 3.48660529e-01 -2.16316357e-01
-1.23321629e+00 -1.38489437e+00 -7.83843338e-01 6.75855204e-02
1.08596735e-01 -7.75492072e-01 7.50836313e-01 -2.94544637e-01
-1.52232385e+00 6.01659358e-01 1.74665213e-01 -6.39093339e-01
-8.26690346e-02 1.75684348e-01 -4.08833861e-01 1.71715379e-01
-1.46126091e-01 7.92670488e-01 3.01840991e-01 -7.15655923e-01
-1.75093964e-01 -7.00532615e-01 -2.66148802e-02 1.43320888e-01
-2.47878551e-01 -2.68821716e-01 -1.47530824e-01 -3.75896394e-01
-5.68167150e-01 -8.29890728e-01 -2.37776026e-01 -2.59008229e-01
-4.96238381e-01 8.92360359e-02 5.69065034e-01 -3.81135404e-01
9.29939508e-01 -1.64357054e+00 4.33622330e-01 1.69690594e-01
6.26495481e-01 4.77883160e-01 -5.63014448e-01 4.34053123e-01
-3.55244666e-01 4.75591332e-01 -1.36207297e-01 6.95314854e-02
-5.00901639e-01 -3.32916707e-01 7.26246610e-02 5.66808522e-01
3.51850957e-01 1.44085228e+00 -8.72805834e-01 -1.52034670e-01
1.25265256e-01 6.61213517e-01 -6.09574914e-01 -2.58107096e-01
-8.32007408e-01 7.28634298e-01 -5.22428215e-01 8.55800569e-01
6.22051179e-01 -5.73243201e-01 3.52067411e-01 -3.27769876e-01
3.57733890e-02 1.65062755e-01 -3.67135644e-01 1.59635699e+00
-2.47459799e-01 3.37817669e-01 -5.69310308e-01 -7.17936933e-01
5.14220476e-01 -2.37794966e-02 7.05091357e-01 -9.42194402e-01
1.58661529e-01 -1.79731213e-02 5.16253948e-01 8.09069630e-03
-1.81507245e-01 -3.08989584e-01 1.30135760e-01 -2.09406018e-01
1.39268026e-01 -3.76725593e-03 3.32119495e-01 8.03886950e-02
1.28090763e+00 -2.03468818e-02 9.50094089e-02 -3.06667447e-01
6.28361166e-01 2.96326965e-01 2.64411002e-01 2.85405666e-01
-5.04919738e-02 2.09403545e-01 9.35021281e-01 -4.02166247e-01
-9.59836185e-01 -8.40898395e-01 -3.27654958e-01 1.09635687e+00
-1.29125983e-01 -6.21314287e-01 -6.31274223e-01 -6.52799129e-01
3.09814841e-01 1.49497256e-01 -9.46441650e-01 -3.33579421e-01
-1.94389522e-01 -1.31034565e+00 1.03061318e+00 4.12401140e-01
9.46207643e-02 -6.20676696e-01 2.77003944e-01 2.31373385e-01
6.29356682e-01 -1.17567718e+00 -3.05392474e-01 4.29884344e-01
-8.25807333e-01 -1.46568108e+00 -5.38385987e-01 -4.10330176e-01
4.64380145e-01 -2.06520185e-02 7.69760609e-01 -2.29046866e-01
-4.31473941e-01 -7.69209564e-02 -6.97838590e-02 -6.18474364e-01
-4.03980851e-01 3.43066067e-01 1.25353724e-01 -3.25869977e-01
3.89206484e-02 -6.45080626e-01 -8.77165675e-01 1.79437146e-01
-9.23285961e-01 -9.89827439e-02 6.03013635e-01 7.43952990e-01
1.06900597e+00 -2.43491068e-01 8.31353605e-01 -1.24153745e+00
6.84068859e-01 -3.76155704e-01 -9.08260584e-01 6.72857761e-02
-6.78265274e-01 2.55383730e-01 1.01873147e+00 -4.97722149e-01
-5.52526116e-01 6.15464486e-02 -2.16169655e-01 -3.79566550e-01
-3.96124050e-02 8.91363561e-01 -3.09881091e-01 -7.65205562e-01
8.82757246e-01 -8.37161951e-03 1.14744440e-01 -2.70315349e-01
3.78948361e-01 1.78174034e-01 1.48162678e-01 -6.24993503e-01
4.55317140e-01 2.59836853e-01 8.66485715e-01 -8.02637577e-01
-7.20153153e-01 -5.17747402e-01 -3.67319912e-01 3.31321597e-01
8.27004135e-01 -1.02332592e+00 -1.14459884e+00 4.03464526e-01
-7.33544886e-01 -7.70986974e-01 1.73128143e-01 1.94866866e-01
-3.38974237e-01 5.28795600e-01 -6.23746991e-01 -8.75652358e-02
-6.15578175e-01 -1.46490824e+00 1.12657475e+00 1.69271946e-01
-3.81781496e-02 -1.27078462e+00 5.18166602e-01 5.37443936e-01
7.74646327e-02 7.90050328e-01 1.37941706e+00 -9.08016801e-01
-4.54666853e-01 -2.66124398e-01 -2.55171001e-01 -7.15644881e-02
3.04668337e-01 5.49568310e-02 -1.04048860e+00 -2.40448549e-01
-9.11486566e-01 -4.95188057e-01 1.08576691e+00 6.61539674e-01
1.36634970e+00 -1.27342075e-01 -7.23763585e-01 9.80645955e-01
1.40952456e+00 3.55855912e-01 7.30936706e-01 6.63315952e-02
1.23329818e+00 8.39521289e-02 2.34155133e-01 1.53683513e-01
5.42048104e-02 6.49821460e-01 6.18022621e-01 -3.11611652e-01
-1.68508232e-01 -3.76793683e-01 3.74652177e-01 1.26643494e-01
3.40280533e-02 -5.50278783e-01 -9.92051303e-01 3.38803530e-02
-1.55180776e+00 -6.98854744e-01 -2.15225905e-01 2.11096215e+00
1.03587925e+00 -8.77825320e-02 4.55107361e-01 -4.20970440e-01
2.92439252e-01 8.57880563e-02 -1.08949137e+00 -3.92667145e-01
-4.31333005e-01 7.12753594e-01 7.52321899e-01 4.22232687e-01
-1.14327800e+00 1.14960420e+00 6.57753420e+00 1.23781025e+00
-1.41898549e+00 -3.09273154e-01 1.13303161e+00 3.83549891e-02
-9.08062905e-02 -2.27486685e-01 -8.88049066e-01 3.44440371e-01
1.48594117e+00 1.25895336e-01 3.33448648e-01 4.82666850e-01
4.29689616e-01 -1.24429718e-01 -1.40039754e+00 8.54313552e-01
-2.94068158e-01 -2.22383714e+00 4.87461001e-01 6.23567879e-01
9.02971148e-01 2.72367716e-01 4.78903621e-01 1.72390506e-01
2.36921445e-01 -1.80919075e+00 -1.16570465e-01 2.69330055e-01
1.09571624e+00 -8.46642375e-01 5.26110947e-01 -3.91351897e-03
-1.07470632e+00 1.48753062e-01 -5.07704020e-01 3.26908797e-01
-4.87339318e-01 2.93824553e-01 -1.19330275e+00 6.59870327e-01
5.46608455e-02 1.06015480e+00 -9.68926072e-01 1.07578778e+00
2.04434022e-01 6.29117489e-01 -2.73300409e-01 -1.07043713e-01
2.12583244e-01 -1.98305011e-01 3.65152843e-02 1.22858846e+00
-2.12170929e-01 -1.08219393e-01 1.11620888e-01 8.20727766e-01
-4.79162425e-01 3.11841637e-01 -5.25671005e-01 -8.84593666e-01
2.75359958e-01 1.26763213e+00 -7.04101384e-01 -1.22985981e-01
-2.04335704e-01 5.62310219e-01 3.21643084e-01 3.29110593e-01
-9.18175101e-01 -1.17271356e-01 8.61247063e-01 3.72893006e-01
1.80940643e-01 -1.88293800e-01 -3.22712302e-01 -7.64568686e-01
-7.45551407e-01 -1.14721298e+00 5.32367349e-01 -4.72071618e-01
-1.28755689e+00 -5.40224556e-03 -2.23478332e-01 -8.24731767e-01
5.23992956e-01 -1.30841029e+00 -4.96210247e-01 9.23066080e-01
-1.39871371e+00 -1.27904904e+00 -3.62523906e-02 2.34195575e-01
1.86102763e-01 -4.30767909e-02 1.03490257e+00 2.90413558e-01
-1.02886319e+00 5.40265679e-01 3.96796525e-01 -1.05736472e-01
7.92366564e-01 -1.26295507e+00 4.24802631e-01 2.47053936e-01
1.33904383e-01 8.06201041e-01 4.63017046e-01 -9.02220905e-01
-1.81114876e+00 -1.36607635e+00 -1.07769288e-01 -9.73269641e-01
7.83392370e-01 -4.83946919e-01 -6.86233759e-01 2.81726152e-01
-2.39051849e-01 1.14499755e-01 1.35740519e+00 1.15897506e-01
-6.77212954e-01 1.68051012e-02 -1.01620209e+00 6.41615689e-01
9.24348831e-01 -4.95690554e-01 3.34019065e-01 6.69815719e-01
7.06972003e-01 -5.75156450e-01 -1.60113156e+00 6.54147565e-01
5.07213771e-01 -3.57527167e-01 1.21320212e+00 -1.22453403e+00
3.86984050e-01 -4.62609053e-01 -2.26494372e-02 -1.44800878e+00
-4.78932858e-01 -6.08697355e-01 -3.56821001e-01 5.48003256e-01
9.01979804e-01 -4.62568462e-01 1.01981819e+00 6.19066767e-02
-1.53935701e-01 -1.26611149e+00 -6.83221996e-01 -5.76222181e-01
6.59833252e-01 -3.61409523e-02 4.02866006e-01 8.01437020e-01
5.78816794e-02 9.39198613e-01 1.71081528e-01 3.12311901e-03
1.78827479e-01 -1.42176658e-01 8.60457659e-01 -1.26407969e+00
-4.90505666e-01 -7.13605762e-01 -5.31114578e-01 -7.83259988e-01
1.31075621e-01 -1.25725603e+00 -6.65612936e-01 -1.53250790e+00
4.25047785e-01 -7.95376822e-02 -4.17877376e-01 4.70681220e-01
-2.28839740e-01 1.94825068e-01 -2.41665721e-01 -3.21491435e-02
-7.62866259e-01 4.27254587e-01 1.60925066e+00 -5.47272086e-01
-2.54171640e-01 -3.79813194e-01 -8.90163720e-01 2.87941158e-01
7.49837279e-01 -1.53376192e-01 -4.88448650e-01 5.63958623e-02
4.30120140e-01 -2.25827470e-01 2.10362911e-01 -8.85982037e-01
-1.99258238e-01 -2.81745464e-01 1.09279251e+00 -2.52939910e-01
4.25427526e-01 -2.21854493e-01 2.68008262e-01 5.13062239e-01
-2.95802027e-01 -2.91255653e-01 9.21223998e-01 9.73151326e-01
2.07663029e-01 4.19203162e-01 9.16337073e-01 -2.33732704e-02
-2.48203129e-01 9.14571583e-01 -3.30433577e-01 -1.20653100e-02
7.86123633e-01 -3.54683042e-01 -7.50057936e-01 -3.80586870e-02
-7.25515902e-01 1.90982342e-01 5.44240892e-01 -1.17886551e-01
3.91722918e-01 -9.62208569e-01 -2.71705806e-01 -1.59472540e-01
5.01689017e-01 -2.60817539e-02 1.89358011e-01 9.73710120e-01
-8.00541341e-01 8.38442266e-01 -1.28182471e-01 -5.81427455e-01
-1.29796660e+00 6.21687770e-01 6.85180783e-01 -6.08780801e-01
1.63763568e-01 8.10760498e-01 6.78050041e-01 -2.56211579e-01
-1.68598980e-01 -5.06257653e-01 -3.32517952e-01 1.23333363e-02
2.36913517e-01 2.04031870e-01 2.53502220e-01 -4.33921576e-01
-4.82449710e-01 6.79016113e-01 -4.21383202e-01 5.70222735e-01
1.30218565e+00 6.41181469e-01 1.27698034e-01 5.93035715e-03
1.19695735e+00 8.37124065e-02 -1.22719252e+00 2.56099164e-01
-1.04186364e-01 1.07972361e-01 -9.94300935e-03 -1.20493114e+00
-6.49968147e-01 8.87248456e-01 6.91789746e-01 -3.83961529e-01
6.67124927e-01 -5.82305454e-02 6.50703609e-01 6.88931465e-01
7.13321716e-02 -6.00025535e-01 3.53015512e-01 4.20675308e-01
5.73780537e-01 -1.01772511e+00 4.59273040e-01 -4.72199649e-01
-3.55861396e-01 9.19300258e-01 5.89395344e-01 3.90040129e-02
2.20386207e-01 -7.36753941e-02 -4.27410364e-01 -7.17572570e-01
-7.53824294e-01 -3.66879813e-02 5.62111735e-01 9.44817662e-01
6.76056027e-01 7.32269585e-02 -2.39327122e-02 4.87513751e-01
8.83332416e-02 -2.36140281e-01 3.44315678e-01 5.49390912e-01
-3.87155682e-01 -1.37355256e+00 7.16071948e-02 5.35908997e-01
-7.23583341e-01 -4.72593546e-01 -1.13079250e+00 7.76836455e-01
7.46675059e-02 6.05807185e-01 -3.31829756e-01 -4.48211223e-01
3.03806603e-01 -4.43125963e-02 9.30921435e-01 -8.23577106e-01
-5.42465091e-01 1.79820016e-01 2.16657177e-01 -3.60719383e-01
-1.52709663e-01 -2.88270056e-01 -1.47111452e+00 -3.85450274e-01
-4.80914354e-01 5.15108034e-02 4.58346784e-01 5.81232250e-01
8.66954386e-01 7.16682553e-01 3.06284666e-01 -5.09592116e-01
1.28365634e-02 -7.88697779e-01 -2.65075982e-01 4.78833057e-02
1.28793538e-01 -7.72171319e-01 1.77297711e-01 1.19052462e-01] | [5.147570610046387, 5.860888481140137] |
44fd0ca1-9435-4f6f-85ac-11296a886c07 | identity-expression-ambiguity-in-3d-morphable | 2109.14203 | null | https://arxiv.org/abs/2109.14203v1 | https://arxiv.org/pdf/2109.14203v1.pdf | Identity-Expression Ambiguity in 3D Morphable Face Models | 3D Morphable Models are a class of generative models commonly used to model faces. They are typically applied to ill-posed problems such as 3D reconstruction from 2D data. Several ambiguities in this problem's image formation process have been studied explicitly. We demonstrate that non-orthogonality of the variation in identity and expression can cause identity-expression ambiguity in 3D Morphable Models, and that in practice expression and identity are far from orthogonal and can explain each other surprisingly well. Whilst previously reported ambiguities only arise in an inverse rendering setting, identity-expression ambiguity emerges in the 3D shape generation process itself. We demonstrate this effect with 3D shapes directly as well as through an inverse rendering task, and use two popular models built from high quality 3D scans as well as a model built from a large collection of 2D images and videos. We explore this issue's implications for inverse rendering and observe that it cannot be resolved by a purely statistical prior on identity and expression deformations. | ['Joshua Tenenbaum', 'Safa C. Medin', 'Skylar Sutherland', 'Bernhard Egger'] | 2021-09-29 | null | null | null | null | ['3d-shape-generation'] | ['computer-vision'] | [ 6.37706697e-01 5.71110487e-01 4.48916823e-01 -3.93665969e-01
-6.29373193e-01 -7.71818697e-01 9.41180527e-01 -5.03789186e-01
-2.23123953e-02 4.67713624e-01 2.46168688e-01 -6.25519361e-03
-7.35941753e-02 -6.80515528e-01 -5.83898962e-01 -6.71583176e-01
4.69277576e-02 8.13638866e-01 -1.29457623e-01 -4.52988327e-01
1.42849013e-01 7.04447329e-01 -1.64302409e+00 6.11305200e-02
5.56146920e-01 7.15149403e-01 -2.32697308e-01 6.60943687e-01
-1.69029802e-01 1.21453613e-01 -3.73642951e-01 -4.66305584e-01
6.58923924e-01 -6.75046265e-01 -7.00916350e-01 6.20641589e-01
7.37781942e-01 -2.05908254e-01 9.74902138e-02 8.80178928e-01
4.85946387e-01 -6.36766106e-02 1.05119765e+00 -1.16960001e+00
-8.84148777e-01 -3.37468654e-01 -6.43721700e-01 -1.76148579e-01
6.36758983e-01 -1.16582908e-01 9.07255054e-01 -9.77143526e-01
1.04838049e+00 1.57065213e+00 5.79491317e-01 8.10834527e-01
-1.80508006e+00 -4.35943455e-01 -2.23866031e-01 -4.66355473e-01
-1.23760879e+00 -6.98998690e-01 1.03193140e+00 -7.55674064e-01
4.32617635e-01 4.75260705e-01 7.75801122e-01 1.12637854e+00
1.16703749e-01 2.73802459e-01 1.55796409e+00 -5.16402423e-01
2.19206009e-02 -9.08331387e-03 -4.27815288e-01 7.62982190e-01
5.92041435e-03 2.78716445e-01 -5.29721081e-01 -4.03622538e-01
1.39357209e+00 -5.85908711e-01 -1.64282575e-01 -5.81912398e-01
-9.94015694e-01 7.87825644e-01 7.70911127e-02 -3.05002034e-02
-3.04455757e-01 1.13459490e-01 -2.75884360e-01 3.15455019e-01
7.07422078e-01 5.98986864e-01 -1.62501514e-01 -1.19675517e-01
-5.57607234e-01 6.24234259e-01 7.60965168e-01 9.20693517e-01
1.04525089e+00 2.43458763e-01 3.26653004e-01 8.52362275e-01
4.63842183e-01 4.55402255e-01 5.95276095e-02 -1.36400259e+00
-1.91982575e-02 4.35412079e-01 5.79585545e-02 -1.07579505e+00
-8.72423798e-02 -1.00140013e-01 -5.17834246e-01 7.57353961e-01
5.77659965e-01 2.63898484e-02 -1.16773021e+00 2.06152749e+00
4.82035041e-01 6.72170147e-02 -1.98778167e-01 9.61973071e-01
8.43879521e-01 2.52209842e-01 -1.30884022e-01 -2.02827930e-01
1.36464345e+00 -1.02645062e-01 -5.70684552e-01 -2.26790324e-01
1.81181431e-01 -1.03467572e+00 7.92240858e-01 1.90496176e-01
-1.55145323e+00 -1.83714241e-01 -7.35787570e-01 -2.56500691e-01
1.45852700e-01 -5.47038257e-01 5.36754549e-01 6.76828802e-01
-1.20508718e+00 6.21362686e-01 -4.99891341e-01 -4.00710970e-01
4.12372500e-01 4.53827858e-01 -7.09018350e-01 1.55696958e-01
-7.95106471e-01 9.85961318e-01 -3.87346566e-01 1.12474762e-01
-4.90481079e-01 -6.00903451e-01 -9.20823812e-01 -4.94254768e-01
9.46622416e-02 -1.03981757e+00 1.16029370e+00 -1.18339300e+00
-1.52273667e+00 1.75282288e+00 -3.46819669e-01 2.89453179e-01
5.91637552e-01 6.98141828e-02 -1.15071341e-01 -3.70313674e-02
5.33591695e-02 7.30742037e-01 1.00170231e+00 -1.87341750e+00
1.79350704e-01 -7.39915133e-01 1.52657896e-01 3.34716648e-01
4.14360791e-01 -9.92121249e-02 -1.91512510e-01 -8.69692743e-01
7.47900963e-01 -1.17758346e+00 -2.46370301e-01 3.66730273e-01
-4.31984782e-01 3.60114813e-01 8.40850830e-01 -6.34689927e-01
5.44998527e-01 -1.77394462e+00 4.23733503e-01 4.37894315e-01
2.05832750e-01 -1.58903360e-01 -2.89876074e-01 3.29021573e-01
-2.48899862e-01 5.13964772e-01 -5.26856244e-01 -5.52425206e-01
2.26542637e-01 6.36472881e-01 -1.63190335e-01 5.65337896e-01
6.48890436e-01 1.05566788e+00 -5.42739213e-01 -3.17671895e-01
1.14365034e-02 7.73627341e-01 -8.55485380e-01 2.22465590e-01
-2.67868012e-01 1.00785577e+00 -3.49358976e-01 6.66544557e-01
7.33127236e-01 -2.47845110e-02 9.16476846e-02 -1.89941213e-01
-1.48544852e-02 2.41054654e-01 -1.14079797e+00 1.60881722e+00
-3.10310096e-01 4.21961188e-01 6.03340447e-01 -7.62283027e-01
9.29321826e-01 4.50923115e-01 4.68353391e-01 -6.41719997e-01
2.00853068e-02 2.99702018e-01 1.62604868e-01 -5.01555324e-01
4.26611960e-01 -1.11309171e+00 9.08119828e-02 7.02195287e-01
-4.11920436e-02 -9.85513210e-01 -2.73160338e-01 -8.76342058e-02
7.25768447e-01 4.14080024e-01 6.13400675e-02 -3.94614607e-01
1.94606170e-01 -1.59747362e-01 3.41937840e-01 2.50788957e-01
2.75263280e-01 1.12555087e+00 5.29202580e-01 -4.34855551e-01
-1.21369076e+00 -1.25851285e+00 -3.97329777e-01 4.27677423e-01
2.48266235e-01 -1.81185395e-01 -6.62871242e-01 -1.86510041e-01
2.47948449e-02 5.51874459e-01 -9.45056140e-01 -3.04904557e-03
-6.87418699e-01 -9.99905586e-01 4.09976006e-01 2.15059798e-02
8.59181210e-02 -8.42970908e-01 -4.39619690e-01 7.60051981e-02
2.68292893e-02 -1.08923495e+00 -3.45502228e-01 -1.55342296e-01
-8.83912504e-01 -1.11914456e+00 -5.53226888e-01 -5.07036865e-01
9.63803709e-01 -6.51925802e-02 1.39652777e+00 2.44902179e-01
-4.62310225e-01 7.90089250e-01 5.30752987e-02 -2.42497355e-01
-9.70105767e-01 -5.60304165e-01 4.25876006e-02 1.29650384e-01
-2.05158606e-01 -1.18911731e+00 -3.97243321e-01 5.11734486e-01
-1.11446261e+00 2.54263937e-01 2.66983926e-01 6.41091526e-01
5.71107030e-01 -3.69881511e-01 4.42894548e-01 -9.13278162e-01
7.02351928e-01 -1.87710136e-01 -3.77430439e-01 -2.25557059e-01
-4.51238334e-01 2.59633809e-01 1.90267842e-02 -4.20136243e-01
-1.21347141e+00 1.59176742e-03 -2.83868343e-01 -3.70405525e-01
-4.04552877e-01 6.64415583e-02 -3.38502109e-01 -3.60813826e-01
5.92060030e-01 -2.00258978e-02 5.19637585e-01 -6.08112097e-01
3.62481982e-01 2.05255345e-01 4.52069193e-01 -8.76604617e-01
1.21688664e+00 8.45188081e-01 5.13240457e-01 -1.08391297e+00
-3.86962980e-01 1.82912320e-01 -8.54916990e-01 -7.29721263e-02
9.29242074e-01 -6.98869705e-01 -4.16563570e-01 3.18271369e-01
-1.38273060e+00 -2.71535188e-01 -4.93555427e-01 3.20911333e-02
-9.25276935e-01 2.94765919e-01 -4.33082819e-01 -8.07039380e-01
1.42953157e-01 -1.20188093e+00 1.51717257e+00 -9.68196243e-02
-6.69728577e-01 -1.17915392e+00 1.58533141e-01 4.30374801e-01
3.10720623e-01 8.60448122e-01 1.31038678e+00 5.83661422e-02
-6.88610196e-01 6.86586276e-02 8.76230933e-03 6.60051554e-02
4.28929955e-01 1.24273628e-01 -1.10367656e+00 3.82480584e-03
1.70506284e-01 -9.57815349e-02 5.05372405e-01 2.55707830e-01
5.16421378e-01 -2.94610053e-01 8.14493932e-03 6.27511501e-01
1.26079476e+00 1.83490947e-01 7.33982146e-01 -3.14398706e-02
8.65309358e-01 1.00588894e+00 7.20031112e-02 2.04502523e-01
2.97326505e-01 1.08394670e+00 5.44454038e-01 -2.12980330e-01
-1.84638977e-01 -3.39030772e-01 9.00922641e-02 7.24747479e-01
-5.54796100e-01 6.99807927e-02 -7.40709662e-01 2.80013084e-01
-1.39378762e+00 -8.80250692e-01 -3.47631693e-01 2.29225254e+00
7.76533008e-01 -2.13818610e-01 1.73585452e-02 6.14908198e-03
4.75886285e-01 1.29122913e-01 -3.23976010e-01 -5.69778502e-01
-4.73813891e-01 4.65935916e-01 5.97478123e-03 8.45085621e-01
-6.68548107e-01 6.71568453e-01 7.28568935e+00 4.77652133e-01
-1.09098125e+00 7.09626265e-03 6.80209756e-01 8.74397084e-02
-1.13038898e+00 2.41432115e-01 -2.92735457e-01 -4.34044413e-02
3.34828138e-01 -1.28528878e-01 3.68889391e-01 2.82500744e-01
2.49356747e-01 -1.10956438e-01 -1.35349309e+00 1.10900712e+00
1.11740068e-01 -1.25088835e+00 2.16861725e-01 4.71684545e-01
7.62007177e-01 -5.04633367e-01 2.11031765e-01 -2.18287632e-01
8.90720263e-02 -1.51717782e+00 9.25877571e-01 4.72550273e-01
1.02128875e+00 -4.10906762e-01 -5.13318069e-02 2.80751646e-01
-8.82661223e-01 5.34236252e-01 8.24736953e-02 -2.43706241e-01
6.16469681e-01 3.36079091e-01 -6.39557421e-01 3.11859518e-01
2.50626504e-01 4.04272050e-01 -3.70463461e-01 4.55629855e-01
-1.64302304e-01 2.56640911e-01 -3.91726583e-01 5.80674291e-01
-6.84874579e-02 -7.30797291e-01 9.63231921e-01 6.92141473e-01
4.87500846e-01 5.14096141e-01 -3.13463986e-01 1.36084986e+00
1.74553975e-01 -9.88877192e-02 -8.90138984e-01 -1.54615156e-02
-2.47412715e-02 9.95920777e-01 -6.59337997e-01 1.17751360e-01
-1.51263282e-01 9.99781370e-01 -2.05719806e-02 4.52691823e-01
-5.60376406e-01 4.00336266e-01 1.03961468e+00 6.37290299e-01
7.81275332e-02 -6.35149956e-01 -2.21175209e-01 -1.13549149e+00
9.80849490e-02 -7.37469077e-01 -2.56711990e-01 -1.04113591e+00
-1.36375380e+00 5.26018500e-01 2.44021311e-01 -1.00383997e+00
-4.75698262e-01 -5.96775591e-01 -5.95932662e-01 1.06716001e+00
-1.22187960e+00 -1.25326240e+00 -1.20317742e-01 3.14516038e-01
2.04769209e-01 3.35002422e-01 1.10458112e+00 1.46561921e-01
-1.12227164e-01 1.70761228e-01 -4.21966016e-01 -2.94900775e-01
4.91169631e-01 -1.19660759e+00 6.29650056e-01 4.22549099e-01
3.85567188e-01 6.58511162e-01 9.09645975e-01 -5.26549459e-01
-1.67881882e+00 -6.02408051e-01 5.97969115e-01 -8.63686562e-01
1.47938430e-01 -6.13362074e-01 -8.95084202e-01 6.73357189e-01
-5.95351495e-03 2.33026832e-01 6.88045561e-01 -2.73923337e-01
-3.25021625e-01 5.82297146e-01 -1.32927656e+00 6.37362063e-01
1.65716434e+00 -6.83879495e-01 -4.95304227e-01 4.32268716e-02
3.50928366e-01 -6.46899819e-01 -9.53654170e-01 4.57464695e-01
7.64452815e-01 -1.32666349e+00 1.12351489e+00 -6.07862651e-01
4.32890117e-01 -1.82907447e-01 -3.03042561e-01 -1.27466118e+00
-1.28650710e-01 -1.02161169e+00 2.04049855e-01 1.16172540e+00
2.33832210e-01 -6.60611928e-01 8.61946046e-01 9.34046686e-01
1.05794822e-03 -7.20684409e-01 -1.12344205e+00 -6.90180004e-01
3.77411216e-01 -4.06371921e-01 5.26961327e-01 1.10526347e+00
-5.40953279e-01 3.91721338e-01 -2.60942131e-01 -1.60588492e-02
6.56743705e-01 2.19764411e-01 9.72790718e-01 -1.52678633e+00
-2.54782647e-01 -3.38679135e-01 -5.12148023e-01 -1.05630755e+00
3.42232049e-01 -8.68273497e-01 -1.28383711e-01 -1.18477845e+00
-1.34267658e-01 -7.56382644e-01 7.33451784e-01 9.83180180e-02
2.49014378e-01 5.62272489e-01 8.04185122e-02 2.75481492e-01
1.79012164e-01 7.37848103e-01 1.77027035e+00 2.07802609e-01
-3.24985906e-02 -1.91623330e-01 -7.88885295e-01 9.35077190e-01
4.09503937e-01 -3.39166999e-01 -4.78664368e-01 -4.85842347e-01
3.79718602e-01 2.15802565e-01 7.78184354e-01 -2.70017952e-01
-3.35660189e-01 -2.70359367e-01 3.71870965e-01 -3.21284756e-02
8.94342422e-01 -7.08100140e-01 8.67230117e-01 -1.31450221e-01
-8.18341970e-02 1.87514976e-01 2.00941369e-01 4.19030041e-01
6.65035658e-03 -7.97832757e-02 7.82094777e-01 -4.49258119e-01
-4.16838437e-01 4.33096528e-01 -2.98037261e-01 2.98571318e-01
7.41210997e-01 -7.01487422e-01 9.05495808e-02 -7.42138505e-01
-1.01591408e+00 -5.02607882e-01 1.16456568e+00 3.72726142e-01
6.29229426e-01 -1.76181221e+00 -6.76397741e-01 6.65929735e-01
-1.33371264e-01 2.74460882e-01 7.14783147e-02 8.11146021e-01
-4.01431352e-01 -9.47695002e-02 -2.62589753e-01 -6.43694043e-01
-1.29040933e+00 -4.40799445e-02 5.59082270e-01 1.14378430e-01
-4.25707519e-01 7.37057805e-01 6.56240165e-01 -7.46489346e-01
-5.07499456e-01 -1.96342468e-01 2.01770261e-01 -3.59793454e-02
3.89448851e-02 5.17200939e-02 -1.15945980e-01 -1.49378169e+00
-2.33804300e-01 1.26107693e+00 5.11571884e-01 -6.19387031e-01
1.29180157e+00 -1.92397282e-01 -3.76513034e-01 3.83805007e-01
1.20273006e+00 2.52311289e-01 -1.26322794e+00 3.93585190e-02
-2.95701772e-01 -7.59674728e-01 -3.59309047e-01 -3.22829932e-01
-9.76549804e-01 8.47514689e-01 2.43995085e-01 3.54350328e-01
7.79336751e-01 2.32939914e-01 5.84439874e-01 -2.23205715e-01
3.45906556e-01 -6.04662895e-01 1.92534536e-01 3.21444809e-01
1.40031600e+00 -1.13690829e+00 1.77374040e-03 -7.41872489e-01
-2.84965187e-01 9.14708674e-01 3.83428395e-01 -2.23050132e-01
7.05406010e-01 3.19565415e-01 1.86837107e-01 -4.25502956e-01
-5.17399788e-01 1.03008404e-01 5.54157794e-01 8.54199052e-01
6.09671056e-01 -1.28739387e-01 -3.53995636e-02 2.38296509e-01
-5.33288240e-01 -4.00538713e-01 5.48600554e-01 8.85196805e-01
1.03978045e-01 -1.56357813e+00 -6.37738168e-01 1.07500248e-01
-3.24010938e-01 1.80534214e-01 -7.13421822e-01 8.33402455e-01
2.76522368e-01 5.10434151e-01 3.79278213e-01 -2.18627855e-01
3.70760918e-01 1.14585571e-01 1.05869925e+00 -7.32468605e-01
-1.63193464e-01 2.34676555e-01 2.54378140e-01 -4.42375571e-01
-7.27164447e-01 -7.97726989e-01 -1.08223259e+00 -2.11447105e-01
-1.87702894e-01 -1.82863593e-01 7.15627372e-01 7.91115046e-01
4.39526290e-01 8.50797594e-02 3.30241591e-01 -1.35786664e+00
-2.31336102e-01 -5.05226076e-01 -7.29685545e-01 8.21586668e-01
3.85985941e-01 -9.45539474e-01 -6.93581462e-01 3.93253893e-01] | [12.70760726928711, -0.31205445528030396] |
8c918f1b-b56a-48d6-b0f6-b4d459471980 | convolutional-sparse-coding-fast | 2106.15296 | null | https://arxiv.org/abs/2106.15296v1 | https://arxiv.org/pdf/2106.15296v1.pdf | Convolutional Sparse Coding Fast Approximation with Application to Seismic Reflectivity Estimation | In sparse coding, we attempt to extract features of input vectors, assuming that the data is inherently structured as a sparse superposition of basic building blocks. Similarly, neural networks perform a given task by learning features of the training data set. Recently both data-driven and model-driven feature extracting methods have become extremely popular and have achieved remarkable results. Nevertheless, practical implementations are often too slow to be employed in real-life scenarios, especially for real-time applications. We propose a speed-up upgraded version of the classic iterative thresholding algorithm, that produces a good approximation of the convolutional sparse code within 2-5 iterations. The speed advantage is gained mostly from the observation that most solvers are slowed down by inefficient global thresholding. The main idea is to normalize each data point by the local receptive field energy, before applying a threshold. This way, the natural inclination towards strong feature expressions is suppressed, so that one can rely on a global threshold that can be easily approximated, or learned during training. The proposed algorithm can be employed with a known predetermined dictionary, or with a trained dictionary. The trained version is implemented as a neural net designed as the unfolding of the proposed solver. The performance of the proposed solution is demonstrated via the seismic inversion problem in both synthetic and real data scenarios. We also provide theoretical guarantees for a stable support recovery. Namely, we prove that under certain conditions the true support is perfectly recovered within the first iteration. | ['Anthony A. Vassiliou', 'Israel Cohen', 'Deborah Pereg'] | 2021-06-29 | null | null | null | null | ['seismic-inversion'] | ['miscellaneous'] | [ 3.72021765e-01 1.65306285e-01 -1.51382508e-02 -2.59476930e-01
-3.43782902e-01 -1.95813909e-01 4.77118611e-01 3.57491642e-01
-5.27903080e-01 7.31057525e-01 -1.19941369e-01 1.82906359e-01
-3.43081415e-01 -9.52494442e-01 -7.05806196e-01 -9.46043432e-01
-2.01646730e-01 4.89608169e-01 -8.93700197e-02 -3.41743082e-01
2.54974425e-01 6.37762725e-01 -1.90458632e+00 3.21009494e-02
9.72729504e-01 1.20878875e+00 2.45124653e-01 2.58360118e-01
-6.29683137e-02 7.35394418e-01 -2.28833750e-01 1.96437374e-01
4.26280737e-01 -3.27558696e-01 -4.28049445e-01 3.34618658e-01
1.14302948e-01 -3.06831915e-02 -7.47497529e-02 1.17396975e+00
2.72278994e-01 2.85241812e-01 5.06252348e-01 -7.12064743e-01
1.18791878e-01 3.46759617e-01 -5.33306360e-01 6.26464486e-02
1.42761096e-01 -1.42455578e-01 8.75928938e-01 -1.18264258e+00
6.71905518e-01 5.32532275e-01 6.27509594e-01 6.21915315e-05
-1.32305062e+00 -1.87025964e-01 -1.26195878e-01 2.25016683e-01
-1.53176391e+00 -6.38272166e-01 9.87637520e-01 -3.97972733e-01
6.98932827e-01 2.42780566e-01 8.10162485e-01 6.44588053e-01
-6.66875467e-02 5.16937971e-01 1.00697875e+00 -5.20186663e-01
5.21383882e-01 1.91778839e-01 9.27181393e-02 7.80820549e-01
4.59402323e-01 1.96119472e-02 -5.63346982e-01 -1.83718339e-01
7.50902891e-01 4.79276590e-02 -6.17874444e-01 -5.74991345e-01
-1.03071713e+00 9.52653646e-01 5.65456867e-01 7.76500821e-01
-7.53405154e-01 -1.22788057e-01 2.88332671e-01 2.25200802e-01
3.98561478e-01 2.01096639e-01 -6.05391003e-02 7.81365708e-02
-1.52825058e+00 1.56406581e-01 8.61719728e-01 4.46127117e-01
1.00744164e+00 4.34315532e-01 4.28167105e-01 7.04462349e-01
7.97177013e-03 4.89647299e-01 5.88832736e-01 -7.25997150e-01
1.58263862e-01 4.50232744e-01 1.15598194e-01 -1.48438048e+00
-5.25727034e-01 -9.51854110e-01 -1.34557962e+00 4.24429864e-01
4.64406073e-01 -2.44838428e-02 -6.39275789e-01 1.55969858e+00
3.81069005e-01 2.80252874e-01 1.23797588e-01 9.55012858e-01
3.62664700e-01 7.37275124e-01 -5.26691079e-01 -5.18016934e-01
1.01296902e+00 -4.51572120e-01 -6.03642225e-01 -2.74627149e-01
4.46986228e-01 -6.45692289e-01 6.05053604e-01 8.61669898e-01
-1.06455994e+00 -6.18737817e-01 -1.09771073e+00 3.41227949e-01
-1.20858558e-01 3.49796295e-01 4.53389019e-01 3.51825088e-01
-7.82601893e-01 7.77513087e-01 -8.47107887e-01 -1.30111217e-01
1.50490448e-01 4.07578647e-01 -5.29572427e-01 2.56341714e-02
-8.68322194e-01 7.54031539e-01 4.92259175e-01 5.18362343e-01
-5.75528979e-01 -4.06822294e-01 -8.21789742e-01 3.47867399e-01
2.67376721e-01 -5.40741205e-01 7.22368956e-01 -1.34179819e+00
-1.28367984e+00 7.43830562e-01 -2.63364553e-01 -6.47674441e-01
3.54678154e-01 -1.00071594e-01 -2.68513616e-02 4.35989290e-01
-7.75760338e-02 1.37007758e-01 1.45422316e+00 -1.22419810e+00
-3.10383171e-01 -1.78694770e-01 -1.83974952e-01 -4.98618605e-03
-5.78506649e-01 -4.10890907e-01 -2.92641539e-02 -7.28581011e-01
5.11886001e-01 -7.83703446e-01 -5.16476512e-01 -7.16479346e-02
-1.11247271e-01 3.75023812e-01 5.89087844e-01 -5.81058860e-01
9.98420715e-01 -2.24512839e+00 4.46390986e-01 8.09174955e-01
1.43500999e-01 1.24172971e-01 1.57746732e-01 3.19265217e-01
-3.18783283e-01 -5.21792948e-01 -7.24509299e-01 -3.12886506e-01
-4.68109548e-01 1.67430148e-01 -5.12684703e-01 7.81382859e-01
1.23757459e-01 2.22051457e-01 -7.09527612e-01 -3.34488362e-01
2.56335497e-01 5.87852836e-01 -7.63249397e-01 1.53367192e-01
6.75652027e-02 4.94282573e-01 -4.04590249e-01 2.42804438e-01
6.96006656e-01 -2.22107738e-01 2.38136470e-01 -3.16088468e-01
-4.48756903e-01 -8.71925894e-03 -1.75583422e+00 1.74458659e+00
-7.19326854e-01 5.37809551e-01 4.94486362e-01 -1.74965894e+00
1.22615063e+00 3.21540356e-01 6.19498253e-01 -4.39057201e-01
2.95469970e-01 5.50573647e-01 -8.32737833e-02 -3.34245265e-01
3.05450678e-01 -3.00608367e-01 2.12383419e-01 2.16993019e-01
7.85201415e-02 -9.93574187e-02 2.77139276e-01 -7.13484660e-02
9.94020045e-01 -3.80676016e-02 6.13884509e-01 -4.81515080e-01
7.13765025e-01 1.17287479e-01 4.95197266e-01 5.51264346e-01
4.71005470e-01 6.78230762e-01 3.11621159e-01 -5.11790931e-01
-9.79694843e-01 -5.38894892e-01 -2.65433222e-01 5.17251968e-01
-7.29563236e-02 -9.56534669e-02 -7.32447028e-01 -2.14393750e-01
-1.17477380e-01 3.46651495e-01 -6.19223237e-01 -1.50415421e-01
-7.50961781e-01 -7.99133539e-01 9.32984576e-02 1.06986165e-01
5.20256996e-01 -1.01880646e+00 -1.07742012e+00 4.49605495e-01
-6.90431595e-02 -8.42221439e-01 3.22122902e-01 6.55159771e-01
-1.23357093e+00 -7.87558496e-01 -7.09793270e-01 -8.14314902e-01
7.71767020e-01 2.66614497e-01 8.36244166e-01 2.96237260e-01
-1.78928807e-01 5.69166243e-02 -3.77877057e-01 1.93171665e-01
-1.35847300e-01 5.28995618e-02 3.01540494e-01 6.05337977e-01
-1.15557656e-01 -1.04959822e+00 -4.28103149e-01 -4.84248810e-02
-9.57638621e-01 1.96812730e-02 6.80220842e-01 1.05740416e+00
5.64906836e-01 1.32022098e-01 6.24429405e-01 -6.02480590e-01
2.52586752e-01 -4.77753848e-01 -7.43282080e-01 -1.36650294e-01
-2.64327586e-01 3.14543873e-01 1.13480055e+00 -1.82799920e-01
-9.69306946e-01 4.63057041e-01 -3.53605747e-01 -4.56905246e-01
-1.36932731e-01 9.25520718e-01 2.54050791e-02 -4.41824257e-01
7.66915023e-01 5.14668405e-01 5.58385290e-02 -5.21225214e-01
1.43265992e-01 4.50100780e-01 6.02842391e-01 -6.64255738e-01
1.00163376e+00 7.68300653e-01 2.38554657e-01 -1.17190552e+00
-6.17724597e-01 -5.02457142e-01 -5.72747409e-01 -2.49105319e-01
2.14463487e-01 -7.47384012e-01 -4.94417995e-01 2.81167924e-01
-1.00506639e+00 -9.48949307e-02 -5.83146036e-01 5.16719639e-01
-5.51670492e-01 5.10079861e-01 -2.17973188e-01 -7.11543918e-01
-2.81300336e-01 -8.59062254e-01 6.99600577e-01 6.08427450e-02
-2.76660528e-02 -8.95344377e-01 1.22093119e-01 -1.38751283e-01
5.29783487e-01 3.64058495e-01 6.61704540e-01 -5.92307270e-01
-3.64927739e-01 -3.27592790e-01 1.17547521e-02 3.73068422e-01
3.85782938e-03 -1.36309624e-01 -9.70963776e-01 -4.59757954e-01
6.96790278e-01 -2.79648125e-01 9.35077727e-01 4.30633068e-01
1.04261887e+00 -2.69788861e-01 -1.30002245e-01 7.98346579e-01
1.60616803e+00 -1.56634614e-01 3.34026575e-01 2.20565438e-01
3.10601354e-01 6.05129421e-01 5.35074890e-01 8.66573334e-01
-2.58221239e-01 5.69628537e-01 6.30632997e-01 -1.76904470e-01
2.32990086e-01 1.04073711e-01 3.88510078e-02 1.02261543e+00
-3.93854529e-01 1.26494348e-01 -9.55635607e-01 6.94042981e-01
-1.82275629e+00 -9.51540291e-01 -1.72157899e-01 2.46899009e+00
6.04987919e-01 3.74556810e-01 -1.26636252e-01 7.03005791e-01
4.87724066e-01 2.00239360e-01 -2.34357834e-01 -1.33053169e-01
-1.72258630e-01 5.92094243e-01 2.39810422e-01 5.18511772e-01
-9.60885823e-01 4.10867214e-01 5.14578819e+00 7.11691856e-01
-1.40596175e+00 7.35823363e-02 4.37227190e-01 -7.22777843e-02
-9.39418599e-02 2.59639509e-02 -4.25530285e-01 2.26866931e-01
6.50317550e-01 -1.56666890e-01 4.15352732e-01 1.06427836e+00
2.78119177e-01 -2.49152556e-01 -8.35851550e-01 1.09148538e+00
3.58090401e-02 -1.44610918e+00 -1.48923606e-01 8.43235292e-03
6.34964466e-01 -1.87247187e-01 -5.89802302e-03 -3.86090018e-02
-5.43909192e-01 -7.36320972e-01 7.53717005e-01 5.81202149e-01
6.52729928e-01 -8.13625038e-01 7.23305702e-01 6.18842363e-01
-1.20363605e+00 -2.41319373e-01 -5.45116782e-01 -3.29884529e-01
1.74249604e-01 1.24562585e+00 -4.08839583e-01 5.99149168e-01
4.13967341e-01 8.31716835e-01 -8.02942067e-02 1.17886603e+00
7.07523674e-02 6.58532560e-01 -7.33251989e-01 1.79167286e-01
4.29692060e-01 -4.78613615e-01 6.87551737e-01 1.09506845e+00
5.92969418e-01 6.83848411e-02 -2.25094296e-02 6.60002410e-01
1.67668998e-01 3.18772495e-01 -8.05738688e-01 2.50278443e-01
1.04209624e-01 1.42397678e+00 -9.80307043e-01 -2.57896692e-01
-4.22671705e-01 8.23259175e-01 2.50597268e-01 3.57577890e-01
-5.57714403e-01 -5.17820358e-01 3.16515207e-01 1.92329720e-01
7.46627927e-01 -3.08464825e-01 -4.13590550e-01 -1.24185026e+00
2.29438618e-01 -9.42687452e-01 1.04349211e-01 -4.06886071e-01
-8.23824763e-01 7.94397891e-01 -1.28982306e-01 -1.46503472e+00
-3.30477029e-01 -5.40960908e-01 -5.75454056e-01 6.40338004e-01
-1.44967008e+00 -5.54550231e-01 -5.66949785e-01 7.17002869e-01
2.62589782e-01 -2.17753604e-01 8.17327321e-01 5.00688910e-01
-3.66094261e-01 1.30538359e-01 4.01232362e-01 -1.09353855e-01
2.48120353e-01 -8.82717490e-01 -1.42463237e-01 9.20518398e-01
2.68056810e-01 6.48921132e-01 9.12495792e-01 -4.26433414e-01
-1.24804246e+00 -6.59936488e-01 7.31350899e-01 4.79807347e-01
6.44366741e-01 -2.44996637e-01 -1.21379805e+00 3.22118461e-01
1.39101865e-02 3.16892028e-01 3.56493562e-01 -1.10204749e-01
3.18270959e-02 -3.39888334e-01 -1.05771327e+00 1.76296625e-02
6.93597138e-01 -3.54329556e-01 -6.08740807e-01 3.16904455e-01
6.81188926e-02 -2.69662321e-01 -7.59613335e-01 3.89668018e-01
3.60160142e-01 -1.24249065e+00 8.56996357e-01 -2.62009412e-01
3.55219215e-01 -4.13259625e-01 -1.76618367e-01 -1.20680606e+00
-1.72227472e-01 -5.71542144e-01 -1.97310582e-01 7.67482400e-01
1.98382549e-02 -5.81976056e-01 9.00706232e-01 5.35776354e-02
-1.60112038e-01 -8.90788496e-01 -1.27629745e+00 -6.44899368e-01
-4.37541962e-01 -2.47274205e-01 1.65193692e-01 1.05017602e+00
-4.09535738e-03 3.42925750e-02 -4.10624236e-01 2.79452920e-01
7.84653187e-01 4.80062872e-01 4.29138720e-01 -1.31011641e+00
-6.05184019e-01 -3.06226045e-01 -5.72325647e-01 -1.04106474e+00
1.68679599e-02 -7.67049134e-01 1.61584660e-01 -1.13116038e+00
-8.02611709e-02 -6.14690661e-01 -2.13850141e-01 3.44963938e-01
2.21004769e-01 3.64652365e-01 5.71120456e-02 2.73695171e-01
2.95968540e-02 5.98984480e-01 7.48524308e-01 2.56071854e-02
-1.63311541e-01 1.22286476e-01 -1.69892669e-01 9.12948132e-01
8.39087844e-01 -5.77253819e-01 -2.63300836e-01 -1.92992687e-01
5.60815394e-01 1.17109105e-01 3.88418794e-01 -1.41196167e+00
3.67544085e-01 5.14275096e-02 1.38073251e-01 -3.33706647e-01
4.10871655e-01 -1.07871771e+00 2.39472449e-01 6.43966079e-01
4.07508239e-02 -3.05166215e-01 -2.07325928e-02 3.28908026e-01
-5.43378770e-01 -6.89043820e-01 8.59922230e-01 -8.15164819e-02
-4.60384250e-01 -3.37669104e-02 -4.43159103e-01 -2.49253169e-01
7.34382629e-01 -2.95798182e-01 3.13103080e-01 -3.89558375e-01
-9.66274023e-01 -3.28371972e-01 3.82893294e-01 -3.59706610e-01
6.76153719e-01 -1.20262241e+00 -5.44289529e-01 5.04293919e-01
-1.56144843e-01 6.47464469e-02 2.12866902e-01 1.09646976e+00
-5.94388664e-01 1.85254037e-01 -3.41316134e-01 -7.28825986e-01
-8.23419392e-01 5.32135546e-01 3.24927568e-01 -2.58040905e-01
-7.10569441e-01 6.04934752e-01 -3.16565894e-02 -3.61505561e-02
1.16558351e-01 -4.25124496e-01 -2.44641349e-01 2.28760988e-01
4.91819054e-01 2.94362903e-01 4.24845397e-01 -6.11031771e-01
-2.29157507e-01 8.00239265e-01 3.08521330e-01 5.86734414e-02
1.72127950e+00 2.72079825e-01 -2.89031774e-01 3.34305823e-01
1.13560808e+00 1.86335161e-01 -9.67345536e-01 -3.26726019e-01
-9.25293863e-02 -3.47952604e-01 3.39216799e-01 1.33987507e-02
-1.21549213e+00 8.33324492e-01 5.37503004e-01 4.39822257e-01
1.42932773e+00 -3.33282799e-01 6.59015954e-01 7.41303146e-01
5.09546757e-01 -9.46370721e-01 -1.07815623e-01 3.31253380e-01
9.39601362e-01 -9.83463943e-01 1.70263097e-01 -3.11119407e-01
-1.81608334e-01 1.37299979e+00 1.35741070e-01 -5.69121599e-01
6.56208634e-01 4.02182430e-01 -2.74503112e-01 -1.73654646e-01
-3.03315729e-01 -2.45715737e-01 1.22207776e-01 1.61715925e-01
1.64729416e-01 -3.01614195e-01 -3.63389313e-01 3.02091181e-01
-8.84946361e-02 4.80029881e-02 4.62280869e-01 8.73999119e-01
-7.80111730e-01 -8.99376750e-01 -5.89964807e-01 4.15992916e-01
-2.19104916e-01 -7.15557188e-02 1.13642022e-01 7.27008820e-01
1.19538628e-01 4.76784259e-01 -2.12910920e-02 -6.56543598e-02
2.48890519e-01 2.26809680e-02 3.38969171e-01 -4.84845728e-01
-4.74236041e-01 1.03809878e-01 -1.69296503e-01 -6.26992762e-01
-6.19946539e-01 -7.15576291e-01 -1.35618627e+00 -5.07638119e-02
-3.20995748e-01 5.01694858e-01 7.43341565e-01 9.19392347e-01
8.68051127e-02 2.72452086e-01 7.44598627e-01 -1.26342773e+00
-4.74139810e-01 -7.40426600e-01 -5.69396317e-01 2.23029003e-01
4.86838669e-01 -8.45693767e-01 -5.73068202e-01 1.14409365e-01] | [11.599015235900879, -2.207329750061035] |
02955497-946b-486a-a526-353ecdc9b0f8 | hilmeme-a-human-in-the-loop-machine | 2211.05201 | null | https://arxiv.org/abs/2211.05201v1 | https://arxiv.org/pdf/2211.05201v1.pdf | HilMeMe: A Human-in-the-Loop Machine Translation Evaluation Metric Looking into Multi-Word Expressions | With the fast development of Machine Translation (MT) systems, especially the new boost from Neural MT (NMT) models, the MT output quality has reached a new level of accuracy. However, many researchers criticised that the current popular evaluation metrics such as BLEU can not correctly distinguish the state-of-the-art NMT systems regarding quality differences. In this short paper, we describe the design and implementation of a linguistically motivated human-in-the-loop evaluation metric looking into idiomatic and terminological Multi-word Expressions (MWEs). MWEs have played a bottleneck in many Natural Language Processing (NLP) tasks including MT. MWEs can be used as one of the main factors to distinguish different MT systems by looking into their capabilities in recognising and translating MWEs in an accurate and meaning equivalent manner. | ['Lifeng Han'] | 2022-11-09 | null | null | null | null | ['nmt'] | ['computer-code'] | [ 2.85662264e-01 -7.85509299e-04 -4.60312963e-01 -5.08800387e-01
-9.81117725e-01 -7.56894290e-01 9.71733928e-01 2.87632793e-01
-5.38622081e-01 8.11992943e-01 3.02948356e-01 -7.32024670e-01
5.22972569e-02 -4.87997085e-01 -4.56554353e-01 -1.63093746e-01
5.47424674e-01 8.64683867e-01 -2.84863740e-01 -5.89646578e-01
4.39043909e-01 3.63953948e-01 -1.10990775e+00 6.61282778e-01
1.22166717e+00 5.15686333e-01 1.77899465e-01 5.07583380e-01
-6.43040121e-01 6.30641282e-01 -7.25739360e-01 -9.49041307e-01
2.20901668e-01 -7.03251719e-01 -1.21165574e+00 -6.21007502e-01
5.24428070e-01 1.60215244e-01 2.24671304e-01 1.10298181e+00
5.82271338e-01 -2.33577386e-01 6.28983259e-01 -9.43550110e-01
-9.75316465e-01 9.30763245e-01 -2.31207594e-01 2.81932712e-01
4.71693367e-01 2.29024485e-01 1.01962447e+00 -1.08826160e+00
9.22178566e-01 1.48456132e+00 4.13436890e-01 6.72283471e-01
-1.32890892e+00 -4.35146153e-01 -3.78451079e-01 4.32790369e-01
-9.48144436e-01 -6.71237648e-01 3.46222967e-01 -3.61830264e-01
1.34479010e+00 2.77180970e-01 3.56526881e-01 1.18180561e+00
5.76099992e-01 6.37434959e-01 1.65898740e+00 -9.90525484e-01
-1.86777875e-01 3.21904808e-01 -1.56556144e-01 4.48497355e-01
2.27202624e-01 4.22491059e-02 -6.89922512e-01 7.56904185e-02
4.70560491e-01 -7.80927956e-01 -4.90633473e-02 3.65666717e-01
-1.67147112e+00 7.76267469e-01 -3.17755491e-02 8.27709436e-01
-3.39548320e-01 -1.49549082e-01 5.83268225e-01 1.02345419e+00
3.82197559e-01 8.67925823e-01 -6.66760325e-01 -5.74090064e-01
-8.75071645e-01 -3.82293314e-02 6.69305503e-01 6.43228829e-01
4.25529748e-01 -2.09848117e-02 -4.13331240e-01 1.03967285e+00
3.08402367e-02 7.23188102e-01 1.00709963e+00 -8.11493456e-01
7.75723994e-01 7.41596878e-01 -1.05940647e-01 -8.06660295e-01
-1.07022621e-01 -4.23977315e-01 -5.54969907e-01 1.19189702e-01
6.13013864e-01 -4.54243645e-02 -7.62165010e-01 1.84583342e+00
8.23930278e-02 -9.95143771e-01 2.09136710e-01 8.20753038e-01
4.80427146e-01 8.12715769e-01 8.53888690e-02 -2.63061941e-01
1.44698453e+00 -9.80249763e-01 -8.19763839e-01 -4.39509183e-01
9.03806448e-01 -1.33801174e+00 1.50735724e+00 2.01099753e-01
-1.42556858e+00 -7.03634381e-01 -1.03169394e+00 -2.79089570e-01
-6.12787068e-01 8.21452290e-02 3.34796369e-01 6.71016037e-01
-1.16019988e+00 7.28972733e-01 -5.47951102e-01 -6.04801059e-01
-7.13239089e-02 2.01459050e-01 -4.81295049e-01 -1.85029823e-02
-1.34915268e+00 1.93710756e+00 4.10805434e-01 1.03427060e-01
-4.14841771e-01 -2.57307500e-01 -5.85153461e-01 -2.84134239e-01
-5.08002341e-02 -8.21276069e-01 1.55588055e+00 -1.55332243e+00
-1.60322189e+00 1.35390031e+00 -3.74479681e-01 -2.62121081e-01
4.84255075e-01 -9.60613042e-02 -6.71021879e-01 -2.33056948e-01
2.82561570e-01 7.41255522e-01 5.40141582e-01 -7.68819392e-01
-7.38125086e-01 -3.36495131e-01 -1.97107106e-01 2.64682710e-01
-5.19461110e-02 6.11139178e-01 3.71596336e-01 -5.61313689e-01
-3.07541508e-02 -7.84567416e-01 2.06006959e-01 -3.01039129e-01
3.80244851e-02 -4.28201348e-01 3.70484263e-01 -9.30567384e-01
1.19994724e+00 -1.68494892e+00 3.86932015e-01 -2.34436527e-01
-1.64679781e-01 4.50706035e-01 -3.18543345e-01 7.20626295e-01
-2.96796635e-02 3.59832764e-01 -1.05208412e-01 -1.59032851e-01
2.58189023e-01 2.98772335e-01 -7.87034854e-02 8.31738189e-02
4.46692437e-01 1.41039371e+00 -9.63307738e-01 -5.05490780e-01
-1.13948481e-02 6.43095598e-02 2.24407673e-01 1.20475374e-01
-1.90195158e-01 4.60225552e-01 -1.16810158e-01 5.24538517e-01
2.25849599e-01 1.98763728e-01 1.92771196e-01 2.30365396e-01
-4.22022372e-01 8.46971393e-01 -4.17213440e-01 1.72831893e+00
-6.49776757e-01 9.71064508e-01 -3.39660406e-01 -7.84684479e-01
8.49619091e-01 5.49879253e-01 -1.69008985e-01 -1.33492422e+00
9.43752527e-02 8.83367836e-01 6.25052154e-01 -2.89281875e-01
6.54505134e-01 -3.19140792e-01 -7.93297961e-02 5.97801864e-01
5.25504500e-02 -7.78671801e-02 3.86535197e-01 -2.56578147e-01
6.91834748e-01 1.94799021e-01 4.20313030e-01 -5.88813007e-01
6.64350688e-01 5.14710367e-01 3.52118403e-01 3.66706103e-01
-9.76297036e-02 2.96808749e-01 1.75356448e-01 -5.30125797e-01
-1.28734064e+00 -9.89566386e-01 -1.11345379e-02 1.15658593e+00
-4.09386545e-01 -6.77661225e-02 -1.01411498e+00 -5.10018766e-01
-5.47367394e-01 1.08440387e+00 -2.87429810e-01 -1.90824389e-01
-9.71957386e-01 -6.11395955e-01 1.01403058e+00 2.54464895e-01
4.13333893e-01 -1.19102252e+00 -7.35590816e-01 4.72325146e-01
-7.18533993e-01 -9.92203653e-01 -2.90458113e-01 7.89199024e-02
-1.05270350e+00 -4.94353175e-01 -8.47836554e-01 -9.39405918e-01
3.09040248e-01 4.01156433e-02 1.47348881e+00 -1.54024288e-01
2.70262331e-01 -3.04735035e-01 -5.60825646e-01 -6.60232425e-01
-1.25086546e+00 3.41722727e-01 1.15376540e-01 -3.49515498e-01
4.96951371e-01 -3.84416610e-01 -2.55995005e-01 3.13425750e-01
-8.13211441e-01 3.39970738e-01 9.44272280e-01 6.35236561e-01
2.80073464e-01 -6.03447556e-01 6.54856622e-01 -5.80931425e-01
1.11784947e+00 -1.09369993e-01 -1.55383036e-01 6.06202543e-01
-9.49436128e-01 3.98077041e-01 6.28501654e-01 -5.28052509e-01
-9.47282851e-01 -4.98622954e-01 -2.27081269e-01 1.68100148e-01
-1.04800418e-01 6.63150311e-01 -1.63267836e-01 6.28537908e-02
8.91272962e-01 3.46811980e-01 -5.32131307e-02 -4.37336206e-01
5.54362774e-01 8.89073193e-01 5.43325067e-01 -8.04560423e-01
6.56574845e-01 -1.92895964e-01 -1.72057569e-01 -4.18017745e-01
-6.18441105e-01 -1.21871546e-01 -6.96642339e-01 -1.39731482e-01
7.07396746e-01 -5.69911778e-01 -1.71106294e-01 4.40253139e-01
-1.59328091e+00 -2.63253927e-01 -3.07624102e-01 3.64698887e-01
-4.89755601e-01 3.32598537e-01 -7.64031112e-01 -4.63166624e-01
-7.62319744e-01 -1.17676878e+00 8.50850284e-01 4.04412253e-03
-7.68636823e-01 -9.12378490e-01 3.49976093e-01 4.13642257e-01
7.59929180e-01 6.20974600e-02 1.51338410e+00 -7.10262477e-01
1.20860143e-02 9.08394530e-02 -2.09761232e-01 3.14617693e-01
8.95279944e-02 -1.04079008e-01 -8.40003073e-01 4.19170111e-02
2.60928422e-01 -1.23208374e-01 2.43280381e-01 7.00110942e-02
1.09121814e-01 -5.82178831e-01 9.52511206e-02 1.34827778e-01
1.28375697e+00 2.36438408e-01 6.41684115e-01 6.59892619e-01
2.84466773e-01 9.10880387e-01 5.95087230e-01 -5.44311404e-01
4.06789273e-01 8.32145691e-01 -9.97016113e-03 -7.79004917e-02
-3.30631495e-01 -2.88082451e-01 9.41035390e-01 1.39993632e+00
-3.03329136e-02 -2.26826862e-01 -1.18081188e+00 5.34583330e-01
-1.80719101e+00 -6.73173964e-01 -4.13877875e-01 2.11502218e+00
1.08857095e+00 2.01577261e-01 1.06230413e-03 8.74157771e-02
7.44109392e-01 -4.29531038e-02 1.94137767e-02 -1.45924866e+00
-4.28278029e-01 4.31917310e-01 2.34497115e-01 3.78354371e-01
-4.49680090e-01 1.25487459e+00 6.30643797e+00 9.57675338e-01
-1.30733418e+00 3.98903936e-01 5.39845645e-01 4.27986145e-01
-2.30556190e-01 1.46274902e-02 -4.32063609e-01 3.61359149e-01
1.56799817e+00 -2.65644759e-01 4.73247111e-01 5.11686802e-01
2.62028277e-01 1.43974470e-02 -1.35002220e+00 8.48971188e-01
5.55742607e-02 -1.05183518e+00 4.34936583e-01 6.11397251e-02
6.13525212e-01 2.96441883e-01 -3.54723334e-02 3.60517949e-01
1.64817087e-02 -1.11592376e+00 1.00243759e+00 2.11598068e-01
9.80546117e-01 -5.55984795e-01 9.24559891e-01 4.68102217e-01
-7.23353505e-01 1.86940640e-01 -4.13449466e-01 -4.42682862e-01
1.45284817e-01 4.19584841e-01 -8.43311727e-01 7.25572765e-01
2.00574622e-01 2.42898017e-02 -5.22482693e-01 5.74470103e-01
-4.27181184e-01 5.89682460e-01 -8.79557710e-03 -3.28733593e-01
5.14837444e-01 -3.37320715e-01 6.12270474e-01 1.50133562e+00
4.08122957e-01 -3.16377133e-01 -2.84909785e-01 8.47483933e-01
-5.63573614e-02 7.29840636e-01 -5.76342046e-01 -5.11224389e-01
2.34571159e-01 1.16346717e+00 -6.72023833e-01 -4.32644606e-01
-2.87613958e-01 1.20902121e+00 3.66383702e-01 -8.43244717e-02
-5.60079396e-01 -2.36892223e-01 6.48795784e-01 -5.97255379e-02
-2.94207692e-01 -3.86272520e-01 -5.09115875e-01 -9.60748911e-01
4.09532964e-01 -1.47678649e+00 7.86340460e-02 -7.25706875e-01
-1.30119824e+00 1.00959814e+00 -1.75190270e-01 -1.02655673e+00
-4.83333826e-01 -8.63029957e-01 -4.07667130e-01 1.13557243e+00
-1.42013466e+00 -1.22074282e+00 2.61187196e-01 1.65613309e-01
8.18363786e-01 -1.50741756e-01 1.06550598e+00 2.50331640e-01
-2.70052403e-01 7.09813535e-01 9.24938619e-02 1.96170285e-01
8.54328811e-01 -1.16288233e+00 8.78067553e-01 8.66221368e-01
5.93854964e-01 5.59564173e-01 7.28676379e-01 -4.51849014e-01
-1.23465991e+00 -7.22861111e-01 1.89496911e+00 -8.19121957e-01
6.27061486e-01 -2.47222856e-01 -7.04481602e-01 3.06141973e-01
6.30442917e-01 -8.46472442e-01 3.74602675e-01 2.08644196e-02
-5.20001769e-01 8.63681436e-02 -1.07351065e+00 8.15486789e-01
9.99353766e-01 -7.18530715e-01 -1.21807683e+00 2.23810896e-01
7.23422766e-01 -1.81845546e-01 -7.50152111e-01 5.33563495e-01
8.02408576e-01 -7.10673809e-01 5.37645519e-01 -7.85509109e-01
6.38954878e-01 -8.32197070e-02 -6.98360801e-02 -1.60704613e+00
-3.16847295e-01 -6.67291462e-01 3.12293082e-01 1.19209182e+00
9.00519311e-01 -7.51168549e-01 1.36643291e-01 3.55276972e-01
-1.53732851e-01 -6.44674242e-01 -1.08667374e+00 -9.48119998e-01
5.16430676e-01 -3.48508567e-01 5.52992940e-01 1.10508645e+00
2.83632755e-01 9.41292882e-01 -1.55865163e-01 -5.59229910e-01
8.59355368e-03 -2.51629911e-02 4.99560595e-01 -1.17335463e+00
-1.90106466e-01 -1.01395428e+00 -3.76816958e-01 -5.65754950e-01
2.62740552e-02 -1.18689740e+00 -1.23022191e-01 -1.56422734e+00
2.70454079e-01 1.43317074e-01 -1.33105487e-01 3.34762901e-01
-7.78864995e-02 2.28604391e-01 3.78497452e-01 3.81770492e-01
-1.57193214e-01 1.32369742e-01 1.21538460e+00 -5.12867980e-02
9.82521921e-02 -3.69129866e-01 -5.47939539e-01 5.36820292e-01
1.06914043e+00 -7.25742698e-01 -7.72583038e-02 -1.01151359e+00
5.98203540e-01 -1.06554985e-01 -1.86451033e-01 -8.22745442e-01
3.70220579e-02 -2.65466690e-01 2.06029058e-01 -1.41211599e-01
-4.65478608e-03 -5.28696477e-01 2.06444755e-01 5.97670436e-01
-4.44216192e-01 9.51316535e-01 2.44280353e-01 -3.04947883e-01
-2.67984062e-01 -3.76630038e-01 6.23900473e-01 -3.39377135e-01
-4.35297966e-01 -2.21795201e-01 -6.22513711e-01 1.42995566e-01
4.77787793e-01 -5.46345949e-01 -4.74377364e-01 -2.04963371e-01
-2.17330486e-01 2.38943491e-02 5.51759541e-01 8.27913702e-01
2.33463332e-01 -1.11266959e+00 -1.16574061e+00 -1.38799816e-01
2.23938793e-01 -6.17792904e-01 -3.71215671e-01 1.02294445e+00
-5.93973398e-01 7.73154140e-01 -6.31519079e-01 -2.89299488e-01
-1.17164707e+00 3.45096171e-01 3.32723320e-01 -6.42430723e-01
-7.26467445e-02 4.62835163e-01 -2.99073398e-01 -7.02662826e-01
-1.69951633e-01 -2.67452687e-01 8.11558440e-02 -6.34468999e-03
3.81026894e-01 4.61442024e-01 4.12155956e-01 -8.39147747e-01
-1.95467353e-01 3.99951160e-01 -1.90328788e-02 -6.83761656e-01
8.75280917e-01 -1.64986715e-01 -4.23961490e-01 8.30240786e-01
8.71614695e-01 -4.54370631e-03 -2.74114441e-02 -1.41157776e-01
7.12372184e-01 -2.73460925e-01 -3.11704516e-01 -1.36673892e+00
-2.58881837e-01 1.27948058e+00 6.11753643e-01 -1.23866387e-02
8.66156876e-01 -1.85699895e-01 1.10052288e+00 4.95711207e-01
4.77249473e-01 -1.47878098e+00 -3.59197915e-01 8.50809276e-01
8.21599722e-01 -1.21713316e+00 -5.79634011e-01 -6.81803748e-02
-5.82865715e-01 1.18601716e+00 3.22856426e-01 4.45399731e-01
-1.73343539e-01 6.40614778e-02 5.84701777e-01 7.53551126e-02
-7.66736388e-01 -1.19976416e-01 4.43956167e-01 5.25171638e-01
1.01078010e+00 3.38807881e-01 -1.13188040e+00 2.42041916e-01
-5.26464462e-01 -1.96094483e-01 2.32672662e-01 6.34057164e-01
-4.48856264e-01 -1.73369861e+00 -3.45553845e-01 1.76826909e-01
-8.63140702e-01 -3.71103138e-01 -9.46457207e-01 8.01001966e-01
3.99101302e-02 1.07236779e+00 -2.89330751e-01 -4.87356603e-01
3.85246545e-01 5.91785371e-01 6.56120658e-01 -5.93609989e-01
-1.04258859e+00 -3.14427197e-01 4.88362163e-01 -4.74419057e-01
-4.12245274e-01 -5.59998453e-01 -9.23623145e-01 -3.88650239e-01
-2.50197113e-01 3.80193740e-01 1.00139666e+00 1.24419403e+00
3.12719285e-01 -2.63650194e-02 3.85372937e-01 -4.51089025e-01
-7.19169617e-01 -1.41466892e+00 2.12746292e-01 4.42069948e-01
-1.24103986e-01 -1.36330381e-01 -1.00689247e-01 -1.24323718e-01] | [11.471527099609375, 10.215839385986328] |
2fb35184-6e4e-4817-b500-435ba203d9af | learning-to-answer-by-learning-to-ask-getting | 1911.02365 | null | https://arxiv.org/abs/1911.02365v1 | https://arxiv.org/pdf/1911.02365v1.pdf | Learning to Answer by Learning to Ask: Getting the Best of GPT-2 and BERT Worlds | Automatic question generation aims at the generation of questions from a context, with the corresponding answers being sub-spans of the given passage. Whereas, most of the methods mostly rely on heuristic rules to generate questions, more recently also neural network approaches have been proposed. In this work, we propose a variant of the self-attention Transformer network architectures model to generate meaningful and diverse questions. To this end, we propose an easy to use model consisting of the conjunction of the Transformer decoder GPT-2 model with Transformer encoder BERT for the downstream task for question answering. The model is trained in an end-to-end fashion, where the language model is trained to produce a question-answer-aware input representation that facilitates to generate an answer focused question. Our result of neural question generation from text on the SQuAD 1.1 dataset suggests that our method can produce semantically correct and diverse questions. Additionally, we assessed the performance of our proposed method for the downstream task of question answering. The analysis shows that our proposed generation & answering collaboration framework relatively improves both tasks and is particularly powerful in the semi-supervised setup. The results further suggest a robust and comparably lean pipeline facilitating question generation in the small-data regime. | ['Moin Nabi', 'Tassilo Klein'] | 2019-11-06 | null | null | null | null | ['small-data'] | ['computer-vision'] | [ 3.22931051e-01 8.09565306e-01 6.27971411e-01 -3.68820578e-01
-1.39434493e+00 -6.43997312e-01 1.04976547e+00 1.09370574e-01
-2.53749251e-01 8.20501149e-01 6.21763170e-01 -3.93222898e-01
-1.10140547e-01 -1.06351960e+00 -7.14833379e-01 -1.72367185e-01
6.48320735e-01 7.93399751e-01 1.79828733e-01 -7.25809038e-01
3.27224523e-01 -1.96931899e-01 -1.60892403e+00 7.37321794e-01
1.01047802e+00 9.63816226e-01 3.81918699e-01 1.04472470e+00
-5.71215451e-01 1.39896822e+00 -8.78921628e-01 -7.54134357e-01
1.15886435e-01 -1.04648423e+00 -1.46125817e+00 -2.41196871e-01
5.82709014e-01 -2.73724884e-01 1.57209747e-02 3.39260340e-01
6.45572901e-01 3.34732324e-01 5.07890463e-01 -9.19475257e-01
-1.07415497e+00 9.10616040e-01 3.89400452e-01 1.57086119e-01
8.28050911e-01 2.42064834e-01 1.30086088e+00 -9.42298949e-01
6.53120577e-01 1.15587127e+00 4.80682164e-01 7.13509321e-01
-1.08790588e+00 -1.08920544e-01 -1.60401285e-01 3.04495156e-01
-7.89365411e-01 -4.99471605e-01 6.46871746e-01 -1.01595439e-01
1.19219422e+00 2.50928372e-01 3.45680475e-01 1.16593289e+00
3.30734886e-02 7.61718631e-01 1.04933763e+00 -6.17741466e-01
2.29749829e-01 2.00425819e-01 9.19745937e-02 3.75909716e-01
-2.46227041e-01 -1.49985000e-01 -4.94372457e-01 -3.63667831e-02
3.05978656e-01 -5.26677907e-01 -1.93077937e-01 2.55094022e-01
-1.23111773e+00 1.07658041e+00 4.18666661e-01 4.35289115e-01
-5.70942819e-01 1.87630430e-01 3.34480166e-01 4.92716432e-01
5.01387000e-01 9.19142067e-01 -4.12845403e-01 -1.03055798e-01
-1.12694883e+00 8.15302193e-01 1.21190929e+00 9.84760761e-01
4.50861901e-01 -1.55238524e-01 -1.12952006e+00 6.44452453e-01
2.42149994e-01 2.04875469e-02 6.49872959e-01 -8.90058398e-01
7.69319475e-01 7.47803330e-01 1.16989449e-01 -7.18529582e-01
-2.04228535e-01 -4.01956499e-01 -4.56617147e-01 -1.74472570e-01
6.86470926e-01 -2.93145657e-01 -6.84017122e-01 1.78295040e+00
3.87797743e-01 -2.14835048e-01 3.86675566e-01 6.67193472e-01
1.26831746e+00 8.32214117e-01 1.10469326e-01 3.50710541e-01
1.65589547e+00 -1.29965413e+00 -7.49601781e-01 -1.14279635e-01
4.67020720e-01 -8.61082137e-01 1.29834163e+00 -1.24538783e-02
-1.46836519e+00 -8.65493536e-01 -6.17676258e-01 -7.78697550e-01
-3.87678981e-01 2.37644389e-01 4.59611602e-03 3.31003845e-01
-1.44559157e+00 3.98967385e-01 -8.47304165e-02 -4.68613893e-01
2.06055507e-01 -3.99693213e-02 -3.29120494e-02 1.77513883e-02
-1.35674953e+00 1.13070059e+00 5.81296980e-01 4.64916714e-02
-9.09082353e-01 -6.40185237e-01 -7.26195276e-01 2.11218655e-01
1.56067491e-01 -1.37380886e+00 1.61203873e+00 -8.57908189e-01
-1.71836412e+00 7.61276484e-01 -1.90648124e-01 -8.49719942e-01
3.78011525e-01 -2.28389248e-01 1.52426530e-02 4.99527365e-01
3.15672070e-01 9.91764605e-01 7.54183054e-01 -9.66713071e-01
-5.36901593e-01 3.08783092e-02 6.02983296e-01 2.05631211e-01
1.37396883e-02 -2.41905022e-02 2.61562467e-01 -5.90683877e-01
-4.16472256e-01 -5.27958333e-01 -1.70733139e-01 -4.18168247e-01
-4.79407281e-01 -6.78105354e-01 5.13336480e-01 -1.12135100e+00
1.05134213e+00 -1.43583882e+00 5.23790047e-02 -2.54448116e-01
-8.76628682e-02 1.67225108e-01 -3.58742952e-01 9.74704802e-01
-1.04991913e-01 7.69220442e-02 -5.16136229e-01 -3.95866901e-01
2.18724534e-01 -7.36354291e-02 -7.45393693e-01 -4.42288280e-01
7.13978827e-01 1.34025669e+00 -8.76614392e-01 -4.27427769e-01
-1.47739977e-01 2.74699718e-01 -7.23997355e-01 8.55453789e-01
-8.90845358e-01 4.91052896e-01 -3.23470414e-01 8.14425871e-02
1.73242077e-01 -2.70963460e-01 -2.34745860e-01 2.62781717e-02
-3.23688835e-02 9.46972132e-01 -6.73837125e-01 1.98834908e+00
-7.61422813e-01 4.95306045e-01 -2.34468088e-01 -7.86417842e-01
1.15295649e+00 6.99006855e-01 -3.32961977e-01 -8.09082091e-01
-2.22022715e-03 2.71034360e-01 -8.06111395e-02 -8.08486223e-01
9.50522780e-01 -4.30346638e-01 -9.02937725e-02 7.09877253e-01
4.31644559e-01 -5.08673012e-01 4.58645374e-01 3.00039202e-01
1.17438769e+00 6.59251451e-01 3.52962524e-01 -2.46343628e-01
8.08811069e-01 2.20992282e-01 -4.20014441e-01 8.34601402e-01
2.38836631e-01 9.34294879e-01 4.22275007e-01 -7.45576844e-02
-1.02254963e+00 -8.95829856e-01 3.08820128e-01 1.13793707e+00
-4.08637851e-01 -2.93040782e-01 -1.13085067e+00 -9.02671814e-01
-4.11816329e-01 1.33467340e+00 -6.42566204e-01 -8.53869990e-02
-7.21726954e-01 -1.22813530e-01 7.40126967e-01 3.22908789e-01
5.46556652e-01 -1.54358006e+00 -6.18872166e-01 5.26445866e-01
-6.84181392e-01 -1.04205811e+00 -3.27433676e-01 -1.31240442e-01
-7.84406245e-01 -9.15157139e-01 -9.13828373e-01 -9.70118165e-01
3.22678089e-01 -1.28471032e-01 1.55867648e+00 7.63570443e-02
2.44262382e-01 5.08338690e-01 -6.70674086e-01 -2.10808307e-01
-6.79587841e-01 4.63499904e-01 -7.97431171e-01 1.07281245e-01
1.04242072e-01 -4.81083184e-01 -7.06781447e-01 -2.61059310e-02
-1.13535273e+00 2.41616830e-01 4.47400510e-01 8.04862022e-01
4.16295916e-01 -5.92579186e-01 1.37692225e+00 -7.98669875e-01
1.33626521e+00 -8.79923522e-01 -1.03690960e-01 4.09062356e-01
-2.92066514e-01 5.25640547e-01 1.03869712e+00 -2.52802614e-02
-1.44313884e+00 -3.27717721e-01 -7.70300567e-01 2.45039210e-01
-4.06503767e-01 3.86985272e-01 -1.33094132e-01 5.22712111e-01
8.86130273e-01 3.70529175e-01 -1.83635831e-01 -4.99960750e-01
8.47932935e-01 6.49998903e-01 4.74459618e-01 -6.67143226e-01
7.55527794e-01 1.36680678e-01 -2.74517655e-01 -4.57562298e-01
-1.18377590e+00 -1.26847073e-01 -2.39419207e-01 -2.35774443e-01
1.14746356e+00 -7.36099780e-01 -4.00602758e-01 5.92794344e-02
-1.50968862e+00 -4.26186353e-01 -8.07675421e-01 -1.27510160e-01
-8.46035242e-01 2.10235566e-01 -4.48412836e-01 -7.41886199e-01
-9.53675449e-01 -7.55399227e-01 1.17577255e+00 3.83945525e-01
-4.57265317e-01 -1.11957574e+00 3.37443650e-01 8.90625834e-01
7.61955738e-01 3.00495267e-01 1.07369721e+00 -1.16323352e+00
-6.14329875e-01 9.09097567e-02 -1.38945088e-01 4.14880157e-01
-4.42446768e-02 -4.89066958e-01 -1.13624263e+00 2.08216995e-01
2.49721378e-01 -7.15292871e-01 7.41605520e-01 -1.84819862e-01
9.44027364e-01 -5.70468664e-01 3.70587468e-01 -3.70581821e-02
1.13005722e+00 -1.85979202e-01 8.54882121e-01 2.21229598e-01
4.10944074e-01 1.18330586e+00 4.07047570e-01 6.41896650e-02
9.67465878e-01 5.03442824e-01 3.66473556e-01 1.34114712e-01
-5.65173268e-01 -7.46231437e-01 3.87722224e-01 7.49386370e-01
3.45527589e-01 -4.09134448e-01 -6.55523181e-01 9.09664690e-01
-1.62573099e+00 -1.16411376e+00 -4.89820272e-01 1.83386731e+00
1.18469012e+00 -1.85234770e-01 1.22841366e-01 6.06618263e-02
4.86885190e-01 1.01493426e-01 -1.15329355e-01 -6.61560893e-01
4.01215674e-03 8.73087585e-01 -4.74667490e-01 5.75059414e-01
-5.27527988e-01 8.54951680e-01 5.85587978e+00 7.21285701e-01
-5.60854316e-01 2.88191259e-01 5.14198124e-01 -1.36839161e-02
-6.91165805e-01 2.33575270e-01 -7.55079985e-01 4.10656899e-01
1.38443291e+00 -2.45669439e-01 7.45339915e-02 5.71905553e-01
2.92685091e-01 -7.66851753e-02 -1.23677766e+00 4.62506443e-01
3.85329336e-01 -1.36110246e+00 5.36189318e-01 -4.78008837e-01
6.25042260e-01 -4.25216317e-01 -1.45206496e-01 6.12838566e-01
1.19809963e-01 -1.07674384e+00 9.55535293e-01 8.98898602e-01
4.06046569e-01 -6.33991659e-01 7.32980847e-01 6.27493739e-01
-9.31927919e-01 -4.25559469e-02 -1.96931496e-01 -2.55811065e-01
5.65358400e-01 4.69686270e-01 -1.22017467e+00 7.64238715e-01
2.87473083e-01 1.39161319e-01 -9.66093361e-01 8.81910503e-01
-6.28741443e-01 7.47387350e-01 -7.08891153e-02 -1.75047666e-01
3.61218154e-01 -3.20722796e-02 3.25595498e-01 1.10421586e+00
5.11669457e-01 1.09091327e-02 -2.82041520e-01 1.25406492e+00
-2.54204810e-01 3.81789744e-01 -4.81110066e-01 -2.86457930e-02
1.32739127e-01 1.41137123e+00 -3.19289774e-01 -5.07553160e-01
-1.03919171e-01 9.12916124e-01 5.47653675e-01 2.02366590e-01
-6.95407689e-01 -5.42713225e-01 -2.65491121e-02 1.56304881e-01
3.65143001e-01 1.04549944e-01 -2.33172774e-01 -1.03468442e+00
2.06405193e-01 -1.17028975e+00 4.09582496e-01 -1.14151430e+00
-1.33039582e+00 9.73684967e-01 -4.41344529e-02 -8.16174746e-01
-8.73072207e-01 -3.39236408e-01 -1.01260424e+00 1.13831794e+00
-1.79937041e+00 -1.36782372e+00 -2.34063610e-01 4.62492824e-01
8.44553411e-01 -5.12984954e-03 8.56648684e-01 1.07377745e-01
-1.62610441e-01 4.44099605e-01 -4.17643875e-01 -7.79611319e-02
5.70879817e-01 -1.51321459e+00 7.46772528e-01 9.40283895e-01
1.87274203e-01 5.87404191e-01 6.72783256e-01 -4.71855193e-01
-9.64668572e-01 -1.20534706e+00 1.70020485e+00 -7.02761889e-01
4.66054738e-01 -3.27792376e-01 -9.24675584e-01 4.58504260e-01
1.04346693e+00 -5.95737517e-01 6.43130660e-01 -3.37657750e-01
-1.72745258e-01 1.78893343e-01 -1.19785285e+00 5.38726687e-01
8.01941454e-01 -5.75418949e-01 -1.31757188e+00 3.59794259e-01
1.02241862e+00 -2.29477674e-01 -7.87527502e-01 -1.20902546e-02
1.36042103e-01 -1.14930058e+00 6.73376739e-01 -6.60810590e-01
1.11826694e+00 -3.24846625e-01 -2.74041034e-02 -1.24031889e+00
-2.06420571e-02 -7.26073742e-01 -1.95176318e-01 1.67045891e+00
7.54952610e-01 -4.00616139e-01 4.54838097e-01 4.77824658e-01
-3.41394901e-01 -7.07940817e-01 -9.49437797e-01 -3.59435529e-01
3.18121344e-01 -2.40942091e-01 8.16791594e-01 2.74546206e-01
-2.05797598e-01 9.73641694e-01 -1.36048898e-01 -4.15966421e-01
2.49655113e-01 2.82111615e-01 7.67094970e-01 -8.61931384e-01
-5.55371046e-01 -3.29438329e-01 3.90041053e-01 -1.22923982e+00
1.56153128e-01 -1.19987476e+00 2.25933850e-01 -2.11623001e+00
-1.53560519e-01 6.59683421e-02 3.46031755e-01 1.70026898e-01
-5.72328031e-01 6.25695661e-02 2.63081342e-01 -1.97104007e-01
-5.37784219e-01 8.15123141e-01 1.34510648e+00 2.00783625e-01
1.02855869e-01 1.95382927e-02 -1.12031758e+00 2.30194464e-01
9.40972507e-01 -4.00987625e-01 -4.84452546e-01 -5.76481044e-01
6.36873782e-01 1.74720719e-01 5.88229179e-01 -8.79496276e-01
9.12631825e-02 3.63993108e-01 1.26093134e-01 -5.62462389e-01
1.30667314e-01 -3.61883193e-01 -2.71308154e-01 1.69541433e-01
-8.87902081e-01 3.02317858e-01 2.98757385e-02 2.79422641e-01
-4.17097509e-01 -6.28757894e-01 4.54073220e-01 -3.88990283e-01
-1.15090355e-01 -1.72299147e-01 -4.08498526e-01 6.85192347e-01
5.70624828e-01 -4.01104875e-02 -3.31875473e-01 -8.38198900e-01
-4.00579542e-01 3.57607394e-01 -7.15722740e-02 5.01474440e-01
5.78261495e-01 -1.28351700e+00 -1.24387598e+00 -2.68798351e-01
1.49872482e-01 8.85710195e-02 2.55670786e-01 4.81887579e-01
-4.07743812e-01 7.26051927e-01 -2.85398518e-03 -1.28137767e-01
-7.05786407e-01 3.28370541e-01 4.65088755e-01 -7.68756688e-01
-3.05188864e-01 9.21925724e-01 6.59996048e-02 -7.66597569e-01
-1.95836797e-01 -5.49775422e-01 -6.74147964e-01 3.07304829e-01
6.14861548e-01 4.58834141e-01 8.62573162e-02 -3.87709886e-01
2.09433258e-01 8.33334029e-02 8.60293731e-02 -4.80078846e-01
1.11094368e+00 -2.20744938e-01 -2.02058882e-01 1.65231675e-01
9.67689872e-01 -1.56830341e-01 -9.14600194e-01 -5.42264767e-02
2.79118150e-01 1.69107154e-01 -4.80115443e-01 -1.07937050e+00
-4.26847279e-01 9.81481969e-01 4.68040444e-02 6.34628356e-01
1.09995985e+00 1.50627643e-01 1.13850427e+00 4.13649827e-01
3.37497741e-02 -8.56066704e-01 3.00696343e-01 8.74447227e-01
1.45410562e+00 -9.66439784e-01 -6.93374157e-01 6.75736666e-02
-5.37171602e-01 1.14396262e+00 7.35415995e-01 -2.48863131e-01
-1.03140049e-01 -2.65759736e-01 -2.55983733e-02 -2.43329749e-01
-1.09706724e+00 -4.18944120e-01 3.53354961e-01 4.93746340e-01
7.90554166e-01 -3.48268241e-01 -4.49034631e-01 7.77205527e-01
-8.54020655e-01 9.93292481e-02 5.28402388e-01 5.97167671e-01
-4.07055587e-01 -1.30156040e+00 -3.54146928e-01 3.93694073e-01
-5.36926687e-01 -5.01857817e-01 -6.54137790e-01 5.04665911e-01
5.06876074e-02 1.49209023e+00 -2.13700473e-01 5.02472045e-03
5.84616065e-01 6.73004091e-01 4.22538996e-01 -9.91350353e-01
-1.52387393e+00 -6.75104856e-01 5.98376870e-01 -3.24616164e-01
-2.99946398e-01 -4.09991473e-01 -1.15015852e+00 1.84649527e-01
-1.92529961e-01 5.05487561e-01 4.90437865e-01 1.27133060e+00
6.76220417e-01 6.75927520e-01 3.95776570e-01 -4.49463189e-01
-8.07354927e-01 -1.33886075e+00 1.75858781e-01 4.19404149e-01
2.16076463e-01 6.74666688e-02 -2.21505657e-01 1.91723332e-01] | [11.481911659240723, 8.238142013549805] |
d07527ba-19fa-40fc-9c9a-80e999195477 | gcnnmatch-graph-convolutional-neural-networks | 2010.00067 | null | https://arxiv.org/abs/2010.00067v4 | https://arxiv.org/pdf/2010.00067v4.pdf | GCNNMatch: Graph Convolutional Neural Networks for Multi-Object Tracking via Sinkhorn Normalization | This paper proposes a novel method for online Multi-Object Tracking (MOT) using Graph Convolutional Neural Network (GCNN) based feature extraction and end-to-end feature matching for object association. The Graph based approach incorporates both appearance and geometry of objects at past frames as well as the current frame into the task of feature learning. This new paradigm enables the network to leverage the "context" information of the geometry of objects and allows us to model the interactions among the features of multiple objects. Another central innovation of our proposed framework is the use of the Sinkhorn algorithm for end-to-end learning of the associations among objects during model training. The network is trained to predict object associations by taking into account constraints specific to the MOT task. Experimental results demonstrate the efficacy of the proposed approach in achieving top performance on the MOT '15, '16, '17 and '20 Challenges among state-of-the-art online approaches. The code is available at https://github.com/IPapakis/GCNNMatch. | ['Abhijit Sarkar', 'Ioannis Papakis', 'Anuj Karpatne'] | 2020-09-30 | null | null | null | null | ['online-multi-object-tracking'] | ['computer-vision'] | [-1.47348568e-01 -2.42214039e-01 -2.29898214e-01 -3.15696180e-01
-2.80565113e-01 -3.07545692e-01 3.60481024e-01 2.27072313e-01
-3.45352948e-01 2.96578497e-01 -2.32568964e-01 1.41364589e-01
-1.21380962e-01 -7.89223850e-01 -9.37508881e-01 -3.62996429e-01
-2.98438996e-01 4.78771001e-01 6.43428743e-01 -1.22886583e-01
4.38343510e-02 7.54247963e-01 -1.83281696e+00 1.18676297e-01
3.52359653e-01 1.40283608e+00 3.69866490e-01 1.00618684e+00
-2.22864468e-02 7.04261541e-01 -3.00498251e-02 -2.20283002e-01
6.20134711e-01 1.56548545e-01 -5.93595505e-01 -2.49338057e-02
1.15639889e+00 -3.07112157e-01 -7.17786491e-01 8.35476458e-01
4.49294806e-01 2.79341727e-01 2.27280378e-01 -1.42215788e+00
-2.45475873e-01 1.75401717e-01 -3.94411534e-01 5.04394054e-01
5.94024025e-02 2.79184729e-01 1.15107000e+00 -9.11179185e-01
7.24988461e-01 1.13685405e+00 8.31810772e-01 3.77117246e-01
-7.75516629e-01 -7.68599272e-01 3.90761942e-01 5.64155996e-01
-1.39516830e+00 -4.14271533e-01 9.08022940e-01 -5.34734011e-01
7.92451978e-01 1.42657369e-01 1.13366652e+00 4.95773852e-01
2.51075000e-01 7.87175000e-01 4.37076777e-01 -2.80237615e-01
-1.44499823e-01 -2.32344180e-01 2.95925975e-01 1.17128372e+00
2.33366206e-01 3.48060578e-01 -6.66353822e-01 -2.00183094e-01
6.47229433e-01 2.71688819e-01 2.97528505e-01 -7.95340598e-01
-1.02596569e+00 6.00864708e-01 9.45452332e-01 1.38272718e-01
-4.03369755e-01 6.84854090e-01 2.16823220e-01 4.08388376e-02
3.65622520e-01 -1.49872974e-01 -4.63669956e-01 1.52997777e-01
-6.24454498e-01 4.27175343e-01 5.48403442e-01 1.13994336e+00
1.02189231e+00 -1.44221798e-01 -2.32124940e-01 4.30238932e-01
6.81951404e-01 5.05097449e-01 -6.65786043e-02 -8.97806346e-01
3.73719096e-01 8.03327680e-01 1.96553990e-01 -1.12133992e+00
-5.48857987e-01 -4.91875798e-01 -2.62714356e-01 2.19102010e-01
4.63490874e-01 -3.01519074e-02 -8.52601528e-01 1.64627802e+00
1.04770839e+00 5.83926022e-01 -3.29032063e-01 7.76461601e-01
1.02630818e+00 1.36058569e-01 1.68344453e-01 2.62081295e-01
1.34271789e+00 -1.11201882e+00 -5.13824582e-01 -1.84898883e-01
5.86012602e-01 -6.79379821e-01 2.98848033e-01 -1.37292534e-01
-1.00513649e+00 -9.37847614e-01 -9.07507658e-01 5.36784865e-02
-6.99205995e-01 1.32650703e-01 9.49203491e-01 2.86912560e-01
-1.04519701e+00 7.77217090e-01 -1.06061375e+00 -5.50423622e-01
7.93783963e-01 1.00836086e+00 -1.00738175e-01 -3.19185182e-02
-6.87662661e-01 7.87676930e-01 5.10571539e-01 2.82617658e-01
-8.59897077e-01 -7.42063224e-01 -7.48669147e-01 -1.13301486e-01
4.41329390e-01 -9.99387980e-01 1.16026950e+00 -8.87029529e-01
-9.61949646e-01 5.83607137e-01 -1.56553343e-01 -3.93703938e-01
5.24038851e-01 -5.21768630e-01 -2.33770654e-01 1.50580198e-01
3.33697125e-02 9.19825673e-01 6.99775040e-01 -1.03825927e+00
-1.02843332e+00 -3.51264894e-01 2.17617810e-01 1.70399785e-01
-2.44219750e-01 4.70802709e-02 -6.67088568e-01 -2.71008372e-01
1.03219241e-01 -1.09582102e+00 -1.74147844e-01 6.41226172e-01
-1.24634676e-01 -4.56788838e-01 1.43432426e+00 -4.04960454e-01
7.92996645e-01 -1.89302027e+00 -8.91679898e-02 9.03397202e-02
3.55384707e-01 2.65977561e-01 -2.33722419e-01 3.32199752e-01
2.14109272e-01 -4.43125188e-01 4.43926036e-01 -4.49478298e-01
-1.12851620e-01 1.61378130e-01 2.40113840e-01 6.80253506e-01
2.68651366e-01 1.25368392e+00 -1.04583323e+00 -7.65558600e-01
5.73554397e-01 6.19694650e-01 -5.48846662e-01 3.58092934e-01
-5.55847526e-01 5.82029521e-01 -6.26404405e-01 7.98988879e-01
6.99402153e-01 -4.51961040e-01 1.63972646e-01 -5.38195312e-01
-2.22174242e-01 -2.12905388e-02 -1.27544713e+00 1.74001062e+00
-8.84336308e-02 5.12182891e-01 -1.67513490e-01 -6.93841636e-01
6.88897491e-01 1.38698667e-01 9.57896411e-01 -5.09925604e-01
3.05787861e-01 -7.50310570e-02 2.52891555e-02 -4.21564519e-01
5.27812839e-01 3.80848348e-01 3.23576510e-01 1.02686375e-01
2.52889395e-01 4.32608873e-01 2.44773477e-01 2.17277557e-01
9.66537237e-01 5.48391342e-01 1.97541758e-01 -1.46918163e-01
4.85957414e-01 4.80648987e-02 6.46208882e-01 7.59634256e-01
-2.36902580e-01 2.07480699e-01 -3.48390996e-01 -8.68479550e-01
-7.95707107e-01 -8.42961133e-01 2.96243310e-01 1.37457299e+00
4.51816291e-01 -3.26702565e-01 -2.15158775e-01 -6.99559450e-01
3.24068964e-01 7.45263919e-02 -8.49918067e-01 5.50633594e-02
-8.87917578e-01 -2.89236665e-01 6.83182254e-02 8.61494124e-01
4.23961550e-01 -9.09543514e-01 -8.82590055e-01 3.40546906e-01
-1.20237179e-01 -1.35410118e+00 -4.80795532e-01 -1.94954127e-02
-9.34009969e-01 -1.36309457e+00 -1.15517110e-01 -6.58295274e-01
6.76411748e-01 4.49937642e-01 9.24834251e-01 6.18913949e-01
-6.00031018e-01 7.59525657e-01 -2.95094669e-01 -5.57787418e-01
-1.05850413e-01 4.68272269e-02 -5.78559516e-03 1.28540471e-01
2.13601232e-01 -6.01623476e-01 -8.25332701e-01 3.23335499e-01
-4.82442915e-01 1.26039058e-01 5.00533581e-01 4.29293215e-01
7.03883350e-01 -3.90137583e-01 2.52275258e-01 -4.53728288e-01
-3.86820614e-01 -4.19982672e-01 -9.41686928e-01 2.73420751e-01
-4.06775326e-01 -1.31494001e-01 4.49067175e-01 -5.90963781e-01
-6.93918765e-01 6.45178199e-01 2.31884733e-01 -8.36322188e-01
-9.59044993e-02 1.10318162e-01 -1.83553118e-02 -6.92440271e-01
2.30346173e-01 -5.17444350e-02 -1.50066823e-01 -5.16489208e-01
3.57560486e-01 1.89592481e-01 5.55093110e-01 -4.87713158e-01
1.00665450e+00 7.52368748e-01 3.84456664e-01 -4.26958472e-01
-1.02204263e+00 -9.13970530e-01 -1.11911702e+00 -7.92840779e-01
9.09565270e-01 -1.03864574e+00 -1.02269506e+00 4.33519125e-01
-1.20515668e+00 -2.11762980e-01 -1.71129718e-01 4.71090913e-01
-5.38076580e-01 2.14752659e-01 -2.14573979e-01 -8.37722123e-01
-5.37778139e-01 -7.37851799e-01 9.84557092e-01 5.91817856e-01
2.11511314e-01 -1.10725439e+00 -1.10049680e-01 3.71238887e-01
3.12674880e-01 4.36442405e-01 3.80424619e-01 -7.24172115e-01
-1.24021935e+00 -2.93727607e-01 -2.88146466e-01 -1.52904168e-01
1.41833037e-01 7.25091323e-02 -9.04900253e-01 -5.73043346e-01
-5.66053331e-01 -1.41827419e-01 6.59292758e-01 4.84073400e-01
8.89244974e-01 -1.64958671e-01 -7.90651143e-01 7.42532670e-01
1.73166168e+00 -7.83763975e-02 2.77805805e-01 2.88839281e-01
1.17431211e+00 3.46353680e-01 9.76758480e-01 5.90414822e-01
6.83141232e-01 9.89559352e-01 9.58120883e-01 1.81587234e-01
-4.52604979e-01 -1.43159464e-01 4.53977883e-02 6.89653277e-01
-1.75913349e-01 -7.46649206e-02 -9.19138074e-01 8.41372609e-01
-2.24020243e+00 -8.99449646e-01 -3.18472862e-01 1.98111296e+00
1.44798309e-01 6.80615306e-02 3.65889847e-01 -5.57334185e-01
8.64403546e-01 9.14112702e-02 -6.69290304e-01 7.72576854e-02
3.04114908e-01 8.17787647e-02 7.20860898e-01 2.75154948e-01
-1.46549940e+00 9.13910925e-01 5.01348686e+00 3.56010556e-01
-8.86274099e-01 1.03373498e-01 1.86016411e-02 -1.12880707e-01
4.46719259e-01 1.74886182e-01 -1.24754798e+00 1.49103761e-01
7.64435649e-01 8.58576447e-02 4.04027730e-01 7.34308362e-01
8.18927586e-02 -3.27796526e-02 -1.18481851e+00 7.01992869e-01
-7.49514773e-02 -1.59374809e+00 -2.04063535e-01 -1.27551276e-02
6.19916737e-01 5.15199542e-01 -9.14213359e-02 7.17881694e-02
3.26387167e-01 -5.24004400e-01 8.44598055e-01 7.49307692e-01
4.16595072e-01 -5.06913602e-01 4.56255049e-01 1.23845741e-01
-2.11640549e+00 -2.16001168e-01 -2.80733049e-01 -8.27579647e-02
-1.48775622e-01 -8.00272450e-03 -1.07393336e+00 1.05025923e+00
8.20962727e-01 9.41947103e-01 -8.79448712e-01 1.37265420e+00
3.06544214e-01 2.54033804e-01 -5.78921974e-01 5.70938457e-03
1.42253801e-01 2.60043144e-01 6.04744077e-01 1.14291179e+00
1.54977039e-01 -1.09434389e-01 6.42082274e-01 6.33552372e-01
-3.10639460e-02 -5.76266423e-02 -5.65048993e-01 1.18631765e-01
4.63787735e-01 1.52658713e+00 -8.02066684e-01 -3.23366404e-01
-6.19429231e-01 3.63472521e-01 5.59985995e-01 2.37757768e-02
-9.15055811e-01 3.32802124e-02 5.78925908e-01 1.45802677e-01
1.09247768e+00 -2.25308090e-01 9.38141998e-03 -7.17350423e-01
1.23325847e-01 -4.15372610e-01 6.82070613e-01 -6.20898902e-01
-1.17508554e+00 2.43719503e-01 3.58434319e-02 -1.50619292e+00
-3.81604657e-02 -5.59809029e-01 -7.05982029e-01 5.56048453e-01
-1.66100764e+00 -1.85785115e+00 -6.64678276e-01 6.46955729e-01
4.48409677e-01 1.69640817e-02 4.94728625e-01 3.70365709e-01
-2.64477819e-01 4.51417923e-01 1.67973563e-02 3.62922430e-01
2.74287224e-01 -9.41563845e-01 3.83361876e-01 7.30110824e-01
1.26474500e-01 2.65184104e-01 5.14695346e-01 -9.27939713e-01
-1.74485457e+00 -1.52617550e+00 6.47514403e-01 -5.69587409e-01
6.41497135e-01 -4.24028218e-01 -5.91355681e-01 7.95490801e-01
-2.44401023e-02 7.52614319e-01 3.98085982e-01 -3.56246419e-02
-2.62911052e-01 -2.76459694e-01 -9.40929592e-01 3.90995890e-01
1.39269078e+00 -1.80999786e-01 -1.45650163e-01 5.32933652e-01
6.88538313e-01 -7.59505153e-01 -1.04141128e+00 5.45719683e-01
9.49153602e-01 -6.46948278e-01 1.20413470e+00 -6.42807484e-01
-2.54074950e-02 -5.19035280e-01 -1.89891934e-01 -6.26209557e-01
-4.74340796e-01 -5.70995331e-01 -6.95719779e-01 9.87020016e-01
4.85498309e-02 -3.80983591e-01 9.68620598e-01 3.84761333e-01
-1.88126177e-01 -8.52757871e-01 -9.62818980e-01 -7.79260337e-01
-4.78188336e-01 -3.28466713e-01 4.77686852e-01 6.96676791e-01
-6.00680351e-01 -1.58153451e-03 -3.56904149e-01 7.04242408e-01
9.40262198e-01 4.20362890e-01 1.05414510e+00 -1.50283408e+00
-3.24549705e-01 -8.82864967e-02 -8.71004283e-01 -9.37508047e-01
4.27906141e-02 -9.89653707e-01 -4.11920398e-02 -1.55551374e+00
2.21351475e-01 -5.56516051e-01 -5.53560853e-01 6.35480702e-01
-1.95527524e-01 3.33601266e-01 6.50480032e-01 1.30528599e-01
-1.22808599e+00 4.13559854e-01 1.22921944e+00 -1.51481032e-01
-3.37399617e-02 1.19642891e-01 -2.45211065e-01 5.55105209e-01
7.83272445e-01 -7.62078345e-01 -6.42731646e-03 -5.23259938e-01
-1.67403892e-01 -9.52172652e-02 8.99497151e-01 -1.39298820e+00
6.40497029e-01 -1.98442921e-01 5.50684690e-01 -1.08413255e+00
6.60433054e-01 -1.19106936e+00 4.73440588e-01 6.60167158e-01
1.42271956e-02 3.13371986e-01 4.65064138e-01 8.22210491e-01
2.84184456e-01 7.37910271e-02 6.79563522e-01 -1.56959713e-01
-1.14463067e+00 7.60695279e-01 2.58173525e-01 -1.34377643e-01
1.06757736e+00 -4.60363150e-01 -2.30588168e-01 -7.72931650e-02
-6.66905224e-01 5.33144534e-01 3.13325107e-01 6.20430410e-01
6.54401183e-01 -1.44676399e+00 -5.23944199e-01 4.46026847e-02
1.93098366e-01 4.06372622e-02 2.94838160e-01 1.03812718e+00
-1.92147508e-01 2.94942498e-01 -3.41301471e-01 -9.79540467e-01
-1.74225712e+00 4.92939472e-01 3.81446093e-01 -2.91761756e-01
-6.84390664e-01 7.95835257e-01 9.30958614e-02 -2.39882737e-01
3.31409365e-01 -1.29042134e-01 -1.09651603e-01 -2.57602990e-01
1.93844542e-01 3.25240016e-01 -8.60540420e-02 -8.56465042e-01
-5.93396664e-01 7.48325884e-01 -3.53479743e-01 3.59131575e-01
1.54395330e+00 -1.26508415e-01 2.14331970e-01 2.90750653e-01
1.14645123e+00 -3.79060209e-01 -1.50315893e+00 -5.76136112e-01
-2.33935677e-02 -6.57294750e-01 -1.44178933e-02 -6.12480462e-01
-1.28915167e+00 6.01282060e-01 1.06859756e+00 -1.83619067e-01
7.95653880e-01 8.72698650e-02 7.86819398e-01 5.24080396e-01
3.70963007e-01 -1.04001606e+00 2.66573459e-01 4.95026916e-01
4.84047949e-01 -1.23323190e+00 2.17601076e-01 -4.57324505e-01
-6.05326816e-02 1.12650156e+00 9.67000604e-01 -1.94677398e-01
7.28642166e-01 1.03887856e-01 3.34672481e-02 -4.99620467e-01
-7.21735716e-01 -5.90356410e-01 5.89180589e-01 6.68965816e-01
1.12423569e-01 -1.80890724e-01 1.97821334e-01 -9.13213417e-02
1.89247563e-01 1.09744623e-01 -9.51287150e-02 1.29741573e+00
-3.69512379e-01 -1.09480655e+00 -2.05257058e-01 5.51315069e-01
-2.96028167e-01 1.52154803e-01 -1.96079507e-01 8.83632183e-01
4.69899684e-01 6.99827671e-01 -4.67839697e-03 -4.57380921e-01
3.10986668e-01 3.00421827e-02 7.15972900e-01 -4.84219432e-01
-7.99606740e-01 -8.47975090e-02 4.93146852e-02 -8.35521519e-01
-8.61733615e-01 -8.03267360e-01 -1.30056214e+00 -1.15626179e-01
-6.75597429e-01 -3.06809843e-01 6.91161513e-01 9.74592924e-01
6.11151338e-01 6.31380260e-01 4.40561861e-01 -1.23407638e+00
-1.30464867e-01 -7.42881238e-01 -6.05144911e-02 4.86065865e-01
5.06328523e-01 -1.09387708e+00 2.81791717e-01 5.47027178e-02] | [6.3703179359436035, -2.1100995540618896] |
9abbf8a0-5ed2-4beb-988a-a319d2d92872 | a-classical-approach-to-handcrafted-feature | 2201.10102 | null | https://arxiv.org/abs/2201.10102v1 | https://arxiv.org/pdf/2201.10102v1.pdf | A Classical Approach to Handcrafted Feature Extraction Techniques for Bangla Handwritten Digit Recognition | Bangla Handwritten Digit recognition is a significant step forward in the development of Bangla OCR. However, intricate shape, structural likeness and distinctive composition style of Bangla digits makes it relatively challenging to distinguish. Thus, in this paper, we benchmarked four rigorous classifiers to recognize Bangla Handwritten Digit: K-Nearest Neighbor (KNN), Support Vector Machine (SVM), Random Forest (RF), and Gradient-Boosted Decision Trees (GBDT) based on three handcrafted feature extraction techniques: Histogram of Oriented Gradients (HOG), Local Binary Pattern (LBP), and Gabor filter on four publicly available Bangla handwriting digits datasets: NumtaDB, CMARTdb, Ekush and BDRW. Here, handcrafted feature extraction methods are used to extract features from the dataset image, which are then utilized to train machine learning classifiers to identify Bangla handwritten digits. We further fine-tuned the hyperparameters of the classification algorithms in order to acquire the finest Bangla handwritten digits recognition performance from these algorithms, and among all the models we employed, the HOG features combined with SVM model (HOG+SVM) attained the best performance metrics across all datasets. The recognition accuracy of the HOG+SVM method on the NumtaDB, CMARTdb, Ekush and BDRW datasets reached 93.32%, 98.08%, 95.68% and 89.68%, respectively as well as we compared the model performance with recent state-of-art methods. | ['Md. Shohanur Islam Sobuj', 'Md. Fahim Shahriar', 'Md. Ferdous Wahid'] | 2022-01-25 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [-2.65577585e-01 -8.55539918e-01 -2.26020768e-01 -4.48058575e-01
-2.19448999e-01 -7.24909425e-01 7.52661109e-01 -1.43045202e-01
-3.03099006e-01 8.28759789e-01 -6.90979809e-02 -4.18643087e-01
-3.18753481e-01 -8.29602361e-01 -8.35026577e-02 -1.12467611e+00
1.30309388e-01 1.54275253e-01 2.12396264e-01 -8.59259069e-02
8.87782753e-01 1.35270941e+00 -1.36186874e+00 4.70263362e-01
5.82488239e-01 1.27549994e+00 -9.20773447e-02 8.80202830e-01
1.11479260e-01 1.03175557e+00 -5.98672986e-01 -5.00651002e-01
2.80990094e-01 -3.39881212e-01 -6.70099199e-01 3.28071147e-01
1.94311619e-01 -5.08873641e-01 -7.29081333e-01 6.85609162e-01
6.59414947e-01 3.52955237e-02 1.11838078e+00 -8.57363403e-01
-1.24021232e+00 1.58033907e-01 -7.75989473e-01 3.42137724e-01
-1.59057245e-01 1.24066360e-01 4.27788019e-01 -1.19125366e+00
5.66731811e-01 1.09084082e+00 7.36674428e-01 4.45062220e-01
-7.34463930e-01 -5.71340084e-01 -4.40273762e-01 4.64979708e-01
-1.29768014e+00 -2.77153105e-01 7.56449282e-01 -6.49132490e-01
1.05560136e+00 2.62959599e-01 4.33836520e-01 9.42572296e-01
5.47281861e-01 6.08389854e-01 1.93699872e+00 -7.68018901e-01
1.86690897e-01 2.48545736e-01 6.55550063e-01 7.46390164e-01
2.83756167e-01 4.34267670e-02 -3.84819388e-01 5.34984283e-02
8.38302135e-01 3.75951454e-02 1.78581238e-01 1.80665091e-01
-1.00314283e+00 8.36244345e-01 2.28673309e-01 3.09793472e-01
-3.41671884e-01 -5.26421607e-01 3.30649614e-01 3.03646058e-01
-8.15977380e-02 1.80084646e-01 -4.79347438e-01 -4.67861407e-02
-6.11356735e-01 1.58259958e-01 6.62787676e-01 8.84829044e-01
6.31931305e-01 1.68876722e-01 -2.83367038e-01 1.38174856e+00
4.20712292e-01 7.14512408e-01 7.80229807e-01 -2.83118039e-01
3.02322596e-01 5.54032624e-01 -2.94716153e-02 -1.41055715e+00
-1.07144117e-01 5.28097935e-02 -9.60932374e-01 2.71160692e-01
6.43056273e-01 9.47015535e-04 -1.38778210e+00 4.92371053e-01
7.99862668e-02 -6.93294525e-01 3.81972671e-01 1.19936323e+00
7.04864502e-01 8.79296064e-01 -1.94652826e-02 1.02754578e-01
1.36670613e+00 -1.01422155e+00 -5.37269950e-01 2.32107695e-02
3.12106520e-01 -1.14859414e+00 8.61804664e-01 7.57035494e-01
-4.15114850e-01 -7.70065904e-01 -1.25024259e+00 9.42981318e-02
-6.52926624e-01 7.48993874e-01 7.73525000e-01 9.05246735e-01
-4.07038420e-01 4.48160678e-01 -5.33510327e-01 -4.88933831e-01
4.94532228e-01 3.49618971e-01 -6.30634248e-01 -3.94628167e-01
-6.55587494e-01 8.88787866e-01 3.17914903e-01 2.87384599e-01
-6.42396390e-01 9.32597890e-02 -4.96453851e-01 -5.43495536e-01
-4.08325315e-01 4.59265441e-01 6.51233554e-01 -6.95592165e-01
-1.63596010e+00 1.05257881e+00 1.77474886e-01 5.68546541e-03
4.09386039e-01 9.92876217e-02 -7.39839792e-01 1.17854789e-01
-4.02779609e-01 2.50617892e-01 8.17458808e-01 -8.95235717e-01
-8.07397425e-01 -7.11563289e-01 -6.20531023e-01 -3.42665792e-01
-3.11927497e-01 5.17006099e-01 4.17841561e-02 -9.41558063e-01
3.72902513e-01 -5.95473230e-01 3.51670623e-01 -1.73552677e-01
-1.79938316e-01 -3.79862338e-01 1.35780931e+00 -1.20796335e+00
8.93058658e-01 -2.35639834e+00 -1.81572676e-01 3.87822956e-01
-5.03769994e-01 6.12823665e-01 -4.73667197e-02 4.15851712e-01
3.33631247e-01 -2.67647296e-01 4.53821979e-02 6.40351713e-01
-6.77231923e-02 3.04313809e-01 -3.36333781e-01 6.71339512e-01
4.65975642e-01 5.56987405e-01 -4.93607938e-01 -6.08317912e-01
2.94284612e-01 4.19010639e-01 9.46002826e-02 1.28760457e-01
4.26772535e-01 -1.21808253e-01 -3.03679049e-01 1.36725926e+00
9.26834881e-01 2.77899355e-01 1.28934577e-01 -1.86072320e-01
-2.63964236e-01 -1.87666774e-01 -1.28458297e+00 5.45743763e-01
-1.37731835e-01 6.32688046e-01 -4.14641112e-01 -1.05408514e+00
1.95082963e+00 -6.70346990e-03 -2.37443149e-01 -8.85811567e-01
1.62909538e-01 6.92730665e-01 -2.46557910e-02 -7.13610530e-01
2.59678006e-01 1.24976493e-01 -1.08484495e-02 1.19333036e-01
2.61554271e-01 1.88153416e-01 3.89224328e-02 -4.86538708e-01
6.23971820e-01 -6.40433505e-02 3.64244461e-01 -2.95293093e-01
8.02716136e-01 3.46714616e-01 3.21712911e-01 6.92882776e-01
-4.40519482e-01 6.04548454e-01 3.19168538e-01 -8.25154483e-01
-1.20526326e+00 -6.58921540e-01 -5.61629832e-01 1.01015985e+00
-1.20901771e-01 6.13964736e-01 -4.15435106e-01 -8.58440161e-01
2.85279989e-01 1.86948299e-01 -3.47914457e-01 7.73027390e-02
-7.33529687e-01 -9.61173952e-01 7.81799436e-01 7.43486345e-01
1.15684247e+00 -1.24269581e+00 -3.51021320e-01 1.52580410e-01
5.35302281e-01 -9.67409074e-01 -1.83419302e-01 4.25724953e-01
-7.84153104e-01 -8.52699339e-01 -1.14858377e+00 -1.37840891e+00
6.21115983e-01 -5.64182773e-02 1.56262532e-01 -3.21483493e-01
-7.54106700e-01 -1.13787986e-01 -6.02093458e-01 -3.06427360e-01
-2.60262698e-01 -3.23924333e-01 2.49096355e-03 2.50511020e-01
6.06519461e-01 3.01641747e-02 -2.96541721e-01 6.34758890e-01
-5.63510835e-01 -4.22660381e-01 1.05192971e+00 1.14249396e+00
4.55693066e-01 2.60464579e-01 4.43830758e-01 -4.30612981e-01
3.75613451e-01 -1.23764716e-01 -7.18104482e-01 3.28972131e-01
-3.70566785e-01 -1.52529851e-01 9.35390770e-01 -6.15333140e-01
-1.06196570e+00 1.48402408e-01 2.16026828e-02 6.20982088e-02
-5.71488023e-01 3.58433604e-01 -1.48319393e-01 -5.52402198e-01
6.84994161e-01 7.32720673e-01 -1.79674163e-01 -6.02799296e-01
-2.47330099e-01 1.77676940e+00 7.34699488e-01 -4.48876023e-01
3.49645197e-01 2.14450806e-01 -3.80077995e-02 -1.06700504e+00
-1.79057732e-01 -4.81697917e-01 -9.48266685e-01 -2.54383117e-01
6.77454710e-01 -5.61839461e-01 -4.21250522e-01 1.60551572e+00
-8.36510062e-01 -6.22227378e-02 2.36971602e-01 6.06427312e-01
-2.10465983e-01 3.33952576e-01 -8.95417035e-01 -1.07271016e+00
-2.90577739e-01 -9.43429291e-01 6.38834119e-01 4.01787728e-01
5.24388142e-02 -5.99649370e-01 -3.08880508e-01 4.55349594e-01
3.75061780e-01 4.92401809e-01 1.21166253e+00 -7.42833078e-01
-1.35466427e-01 -5.23637354e-01 -6.71207130e-01 6.62767112e-01
5.00970662e-01 5.53777754e-01 -8.90914440e-01 -2.13857934e-01
-3.42259377e-01 -6.08651042e-01 7.30603456e-01 -6.13912158e-02
9.28677678e-01 -5.63073218e-01 1.09501228e-01 4.62871820e-01
1.46344209e+00 8.72429430e-01 8.99917006e-01 7.38075376e-01
3.88712883e-01 4.06720459e-01 8.49324286e-01 6.94417477e-01
1.03795528e-02 5.35923004e-01 -2.23415494e-01 3.13437283e-01
-2.15003774e-01 -1.62301704e-01 3.00174445e-01 8.23785901e-01
-5.89236557e-01 1.88128129e-01 -1.18562138e+00 4.13464844e-01
-1.30120218e+00 -5.96378267e-01 -1.26598537e-01 1.87689483e+00
9.79621053e-01 -8.08095112e-02 3.34261358e-01 9.17494237e-01
7.39253163e-01 7.42379762e-03 -1.71149328e-01 -1.01176381e+00
-4.75214899e-01 4.34763759e-01 7.55791426e-01 -4.50805984e-02
-1.59786391e+00 7.58795738e-01 5.30791903e+00 7.50142932e-01
-1.73853922e+00 -4.68815714e-01 7.02472389e-01 8.81309152e-01
5.59796333e-01 -4.18341547e-01 -1.08117068e+00 4.64697570e-01
6.48730874e-01 4.09402311e-01 3.89489174e-01 1.05737281e+00
-2.30407864e-01 -3.13432738e-02 -6.19427800e-01 1.03644097e+00
-1.34240778e-03 -1.10999286e+00 2.09302425e-01 3.93839851e-02
8.02538395e-01 -4.77384984e-01 6.75688684e-02 1.91232085e-01
2.14219868e-01 -1.15733314e+00 5.76296151e-01 4.35415000e-01
7.27552235e-01 -8.78344953e-01 1.10582864e+00 2.44406030e-01
-5.52104414e-01 -3.41978699e-01 -7.86987841e-01 -5.96254766e-02
-7.86602974e-01 2.96962291e-01 -6.69461727e-01 2.79832602e-01
9.38734114e-01 5.56880951e-01 -6.35730684e-01 9.64984238e-01
-9.83988419e-02 6.34216309e-01 1.26346443e-02 -7.09970295e-01
5.71373522e-01 1.36771843e-01 -8.65337253e-02 1.67205417e+00
2.62759000e-01 3.33408475e-01 -2.36720517e-01 3.31900001e-01
2.33161435e-01 2.46667400e-01 -6.66896552e-02 -4.41344917e-01
3.27561259e-01 1.21246123e+00 -8.48465145e-01 -9.41208750e-02
-3.44182342e-01 8.56546879e-01 4.21713144e-02 1.68029159e-01
-6.42712951e-01 -1.28437972e+00 1.14525601e-01 -2.67325222e-01
7.96462655e-01 -4.02433336e-01 -3.48008573e-01 -9.33404148e-01
1.46982763e-02 -1.32919288e+00 5.21581829e-01 -4.25714701e-01
-1.53856599e+00 3.52428257e-01 -3.94227803e-01 -9.19389665e-01
1.65317357e-01 -1.42455029e+00 -3.86383653e-01 1.00796425e+00
-1.41246104e+00 -1.23772955e+00 -5.72348118e-01 6.65768683e-01
4.94022787e-01 -7.50742733e-01 7.39004314e-01 1.75382733e-01
-6.93750203e-01 9.08091664e-01 5.69492877e-01 7.75495052e-01
6.05563819e-01 -9.33017671e-01 -4.56559621e-02 7.03677952e-01
-1.92394868e-01 3.41163069e-01 1.72732428e-01 -4.13375556e-01
-1.83832335e+00 -1.05782998e+00 9.99358833e-01 1.59266442e-01
4.41134006e-01 -1.89074755e-01 -7.69406319e-01 3.64822149e-01
-2.76183218e-01 4.28364128e-01 4.98309314e-01 -5.89019775e-01
-5.14574111e-01 -4.16132927e-01 -1.57018983e+00 7.39644989e-02
2.57149816e-01 -3.34453017e-01 -6.09094262e-01 1.09879106e-01
-3.64705682e-01 -1.86000943e-01 -1.17863715e+00 2.40930468e-01
1.06885588e+00 -8.26552570e-01 6.60255015e-01 -7.84101427e-01
6.21709228e-01 -2.98781574e-01 -6.67355359e-01 -7.53582478e-01
-5.46762705e-01 -1.75569862e-01 -5.53505197e-02 1.48009682e+00
1.67614773e-01 -8.41854274e-01 5.51455021e-01 1.93513587e-01
1.66992143e-01 -8.97800148e-01 -5.87374687e-01 -1.10935855e+00
1.61619440e-01 3.03215772e-01 4.88678575e-01 1.12406099e+00
-1.09667525e-01 -3.19139332e-01 -4.32248473e-01 2.54680693e-01
6.67986989e-01 2.09799930e-01 6.26551986e-01 -7.53875315e-01
-1.63447797e-01 -3.36527586e-01 -1.16633630e+00 -5.95045149e-01
-3.72066259e-01 -5.81183434e-01 -2.25956082e-01 -1.23346090e+00
1.82167798e-01 -6.11849785e-01 -3.50797534e-01 6.94666386e-01
2.08334073e-01 5.72554827e-01 1.23903573e-01 3.91571581e-01
1.27235517e-01 -1.42334020e-02 1.22685015e+00 -3.64757895e-01
-6.02464005e-02 7.54869804e-02 -2.91944057e-01 4.07774538e-01
8.22922826e-01 -1.92581773e-01 3.73946965e-01 -2.86292672e-01
-7.72467673e-01 -1.30914524e-01 3.79604220e-01 -9.50690866e-01
1.90372765e-01 -5.15689492e-01 1.24973798e+00 -5.67229807e-01
-8.24469551e-02 -4.33571458e-01 -2.89849371e-01 7.16341734e-01
-1.95091695e-01 -2.87477016e-01 1.02782160e-01 -2.55701728e-02
-4.13435608e-01 -2.59610772e-01 1.31581712e+00 2.25458011e-01
-1.32659078e+00 -1.44865483e-01 -4.81822163e-01 -4.80963022e-01
1.23984194e+00 -7.50448704e-01 -5.02347350e-01 2.34367892e-01
-3.85162890e-01 -4.49597329e-01 -2.34013144e-02 3.24743003e-01
9.22088325e-01 -1.34439147e+00 -7.88658023e-01 5.54468751e-01
1.97365269e-01 -5.97905219e-01 3.44528770e-03 5.93929589e-01
-1.22786820e+00 7.35481024e-01 -1.11803198e+00 -3.58794659e-01
-1.42876995e+00 3.56468379e-01 1.14153378e-01 1.35459185e-01
-2.61752665e-01 7.14330375e-01 -7.69091427e-01 -4.04670238e-01
3.83213401e-01 -1.05055191e-01 -3.87752563e-01 -2.77677663e-02
4.96217698e-01 9.37075615e-01 3.61321360e-01 -7.65940845e-01
-6.53553486e-01 9.56452906e-01 -3.64949197e-01 2.74157107e-01
1.40274346e+00 4.28555101e-01 -2.05271810e-01 7.87917376e-02
1.18127394e+00 5.74584771e-03 -8.55545938e-01 4.17669713e-02
1.22359365e-01 -9.00266171e-01 -9.18702334e-02 -1.17863238e+00
-7.92901635e-01 9.35369551e-01 7.72556543e-01 3.10227890e-02
1.27237093e+00 -2.62128562e-01 5.54177165e-01 7.41007090e-01
5.44607282e-01 -1.30377674e+00 -1.49071202e-01 5.68974078e-01
9.93272662e-01 -1.05617416e+00 1.19741738e-01 -7.72121176e-02
-8.13492417e-01 1.90709186e+00 6.44527733e-01 -4.45223600e-01
5.06064534e-01 2.90574491e-01 3.97119105e-01 3.93913627e-01
-2.25609571e-01 2.73919880e-01 3.59979063e-01 8.36670339e-01
3.43790829e-01 3.73495892e-02 -4.13583964e-01 7.87810028e-01
-1.09767720e-01 1.52890950e-01 3.71286213e-01 1.36467147e+00
-6.79120183e-01 -7.69477129e-01 -6.73208416e-01 4.40614551e-01
-5.12371778e-01 3.77615243e-01 -6.98741615e-01 9.57869709e-01
2.50306707e-02 6.72853708e-01 -1.51676744e-01 -5.42231083e-01
4.01317477e-01 4.38112766e-02 6.35410249e-01 1.10956699e-01
-4.49295670e-01 -1.99579656e-01 -7.00119361e-02 7.06954002e-02
-1.72134444e-01 -6.06458426e-01 -9.63276446e-01 -4.75715578e-01
-3.92633677e-01 -1.28159806e-01 9.01944160e-01 7.50622332e-01
-3.17059904e-02 -1.31391168e-01 1.09831607e+00 -5.44395506e-01
-9.88072455e-01 -1.03636742e+00 -1.05516946e+00 2.37059727e-01
1.58928379e-01 -4.62570876e-01 -2.99741626e-01 2.62939602e-01] | [11.850691795349121, 2.6709163188934326] |
042dd3ab-a98e-4956-b6ab-fd6f12fa6e67 | re-2-region-aware-relation-extraction-from | 2305.14590 | null | https://arxiv.org/abs/2305.14590v1 | https://arxiv.org/pdf/2305.14590v1.pdf | RE$^2$: Region-Aware Relation Extraction from Visually Rich Documents | Current research in form understanding predominantly relies on large pre-trained language models, necessitating extensive data for pre-training. However, the importance of layout structure (i.e., the spatial relationship between the entity blocks in the visually rich document) to relation extraction has been overlooked. In this paper, we propose REgion-Aware Relation Extraction (RE$^2$) that leverages region-level spatial structure among the entity blocks to improve their relation prediction. We design an edge-aware graph attention network to learn the interaction between entities while considering their spatial relationship defined by their region-level representations. We also introduce a constraint objective to regularize the model towards consistency with the inherent constraints of the relation extraction task. Extensive experiments across various datasets, languages and domains demonstrate the superiority of our proposed approach. | ['Lifu Huang', 'Joy Rimchala', 'Lalla Mouatadid', 'Sijia Wang', 'Pritika Ramu'] | 2023-05-24 | null | null | null | null | ['graph-attention', 'relation-extraction'] | ['graphs', 'natural-language-processing'] | [ 1.33538306e-01 2.73843646e-01 -3.56266618e-01 -4.11629975e-01
-2.45836768e-02 -6.47663593e-01 6.22764468e-01 4.20821637e-01
-1.21422730e-01 3.06463927e-01 3.88053179e-01 -6.18514895e-01
-2.31524974e-01 -1.26365745e+00 -8.63209426e-01 -1.17085571e-03
-3.09902459e-01 3.37320119e-02 -9.90340561e-02 -7.86972940e-02
1.72563657e-01 6.42299354e-01 -1.05860710e+00 3.49285990e-01
1.03086174e+00 8.67794454e-01 1.31202802e-01 3.79848450e-01
-4.05081302e-01 9.79340374e-01 -4.86848325e-01 -4.90505993e-01
1.47231221e-01 -1.57593474e-01 -9.26953435e-01 2.69489437e-01
3.59704405e-01 -2.33533829e-01 -5.98266065e-01 7.73953140e-01
7.33096227e-02 2.11262614e-01 6.43089056e-01 -8.72324765e-01
-1.31026924e+00 6.96296573e-01 -1.00186467e+00 2.23518416e-01
4.15144742e-01 -1.23628981e-01 1.60488260e+00 -1.07509851e+00
8.80566955e-01 1.12383318e+00 3.27690125e-01 -4.43763845e-02
-1.15411437e+00 -3.57584625e-01 8.66680324e-01 4.38553900e-01
-1.85900974e+00 -2.20313281e-01 1.20497024e+00 -4.86567974e-01
1.51421762e+00 1.42474890e-01 6.32284164e-01 6.77585125e-01
-2.31146291e-01 7.94948459e-01 5.85709810e-01 -6.53018057e-01
-1.07691757e-01 1.71716630e-01 3.71344298e-01 8.48707139e-01
4.43253726e-01 -2.47007638e-01 -5.70026398e-01 3.58607978e-01
9.01628375e-01 -2.16764316e-01 -1.84249654e-01 -4.99847412e-01
-7.50910103e-01 4.91125852e-01 9.69380558e-01 3.05871218e-01
-2.54561722e-01 -2.01627910e-02 1.58709943e-01 -3.01371396e-01
5.30778468e-01 4.89730090e-01 -2.84530699e-01 2.82672346e-01
-6.67677999e-01 -5.43376170e-02 4.87800926e-01 1.43161857e+00
8.74240160e-01 -3.38682234e-02 -2.76146322e-01 8.06912959e-01
6.30205333e-01 8.18998292e-02 -2.53197253e-01 -2.72262663e-01
9.77233529e-01 1.42772496e+00 -1.22125313e-01 -1.56374145e+00
-3.50772798e-01 -7.14718044e-01 -7.24937022e-01 -1.59481466e-01
1.17913239e-01 4.53749299e-02 -8.29517484e-01 1.48838568e+00
3.44914466e-01 3.19656767e-02 -1.33788317e-01 7.84763336e-01
9.68567193e-01 4.65478629e-01 3.97581041e-01 1.69629976e-01
1.44620824e+00 -1.01243937e+00 -8.13726902e-01 -5.76921999e-01
8.16838205e-01 -3.57696861e-01 1.15689278e+00 -3.65623906e-02
-8.08220029e-01 -5.21238804e-01 -1.24740303e+00 -5.70493996e-01
-6.70261323e-01 4.39800948e-01 9.80822444e-01 4.39217418e-01
-6.11090004e-01 1.61627561e-01 -5.65861881e-01 -2.43404850e-01
7.11569726e-01 2.77955234e-01 -3.41653556e-01 -1.77754298e-01
-1.05495477e+00 8.93671811e-01 4.19163764e-01 4.19338107e-01
-1.59300208e-01 -5.14753461e-01 -1.20437920e+00 1.94727063e-01
6.79031730e-01 -6.79065108e-01 5.39330661e-01 -6.17066622e-01
-8.56307864e-01 7.80987322e-01 -4.14235294e-01 -1.77255869e-01
-4.71490948e-03 -4.84275550e-01 -4.12016183e-01 6.72363862e-02
-4.37660776e-02 3.69739950e-01 2.74264187e-01 -1.64514875e+00
-5.07666886e-01 -4.79653209e-01 5.44849336e-01 3.97367895e-01
-4.74935323e-01 -4.67975996e-02 -1.02950084e+00 -8.09932768e-01
2.24091753e-01 -4.77852136e-01 -9.92324427e-02 -2.92827636e-02
-9.46883321e-01 -3.89137745e-01 8.62737179e-01 -8.54365826e-01
1.81455255e+00 -2.12154341e+00 1.56631097e-01 5.41910350e-01
5.06469905e-01 1.07827540e-02 -2.07847834e-01 4.29722279e-01
-8.01768601e-02 3.95886928e-01 -1.17986128e-01 -1.92740977e-01
2.59604305e-02 1.43502221e-01 -2.27506116e-01 1.42177656e-01
6.41864181e-01 1.28750324e+00 -8.50123465e-01 -7.68926084e-01
2.21525416e-01 6.09980822e-01 -4.15907681e-01 1.99599937e-01
-3.01840693e-01 7.03782663e-02 -6.63701773e-01 7.94657052e-01
6.90879285e-01 -5.93577027e-01 4.57495093e-01 -5.46139479e-01
1.22392423e-01 4.59285349e-01 -1.03046072e+00 1.55520165e+00
-6.24282300e-01 6.88161492e-01 -2.13658035e-01 -8.31154406e-01
1.02177048e+00 -8.41635391e-02 3.03653449e-01 -7.58574426e-01
-5.54531887e-02 -4.37881649e-01 1.82577327e-01 -5.18347740e-01
6.89324319e-01 3.01929027e-01 7.62715861e-02 3.08399916e-01
-2.14347184e-01 2.37877846e-01 3.53282988e-01 6.55688703e-01
1.03030503e+00 4.72520500e-01 4.57902700e-01 -1.31545039e-02
3.22337180e-01 -7.26489797e-02 5.08249760e-01 5.73694944e-01
1.56150356e-01 4.00691390e-01 5.82101762e-01 -4.36946720e-01
-6.76817954e-01 -8.98593485e-01 8.02045390e-02 9.02161598e-01
4.33759153e-01 -8.15172732e-01 -4.70549315e-01 -9.67542291e-01
-7.81657696e-02 7.41448402e-01 -7.96999872e-01 -9.81200188e-02
-8.33788812e-01 -5.12992263e-01 3.45377415e-01 9.87112701e-01
4.37205166e-01 -8.83902729e-01 -3.13909739e-01 2.84431018e-02
-5.35028726e-02 -1.45136678e+00 -3.24626654e-01 9.71897021e-02
-5.33691049e-01 -1.09952343e+00 -1.37070939e-01 -8.17129135e-01
1.11825454e+00 2.73613244e-01 1.37388897e+00 3.57839584e-01
-1.75355509e-01 1.29715592e-01 -4.98818159e-01 -1.35938823e-01
5.48124135e-01 2.64218837e-01 -5.72761059e-01 -6.94887713e-02
6.14924073e-01 -5.65253913e-01 -5.41821241e-01 1.13985471e-01
-6.76541686e-01 3.41269195e-01 7.98339069e-01 3.91323149e-01
7.46457815e-01 2.88632363e-01 1.20271862e-01 -1.09449434e+00
6.08443379e-01 -4.62087721e-01 -3.34819913e-01 7.06826031e-01
-5.15610635e-01 1.50569841e-01 4.06652153e-01 -2.46729493e-01
-1.10986459e+00 6.81536868e-02 4.11449134e-01 -1.30632222e-01
-1.30917415e-01 9.75522935e-01 -8.77439737e-01 2.40628570e-01
2.72329181e-01 2.15510130e-02 -8.48132670e-01 -4.43712711e-01
7.54474461e-01 3.38668972e-01 3.40598881e-01 -7.39016593e-01
8.91384661e-01 3.69150281e-01 3.98359401e-03 -6.75676048e-01
-8.68609846e-01 -2.00155079e-01 -9.84532714e-01 -5.75246885e-02
8.61500025e-01 -8.79766941e-01 -4.31349397e-01 -1.84853449e-01
-1.40117669e+00 -1.69308856e-01 -1.32849127e-01 1.73603311e-01
9.06646177e-02 3.25477004e-01 -4.63977784e-01 -9.14070547e-01
-1.38558030e-01 -7.21098840e-01 9.26580429e-01 2.54112095e-01
-3.68400753e-01 -1.11581361e+00 -1.52527109e-01 3.01273406e-01
-7.33566061e-02 3.29804122e-01 1.38257039e+00 -3.01382959e-01
-1.14940298e+00 -1.77520275e-01 -9.02774274e-01 -3.11291337e-01
5.09626925e-01 2.25193381e-01 -7.12396443e-01 3.80993009e-01
-6.90378368e-01 1.64832443e-01 8.61947060e-01 2.39757985e-01
1.41939020e+00 -4.10192937e-01 -7.19425797e-01 7.14447796e-01
1.31725276e+00 1.80051655e-01 5.43596864e-01 3.91751707e-01
1.44654286e+00 8.65791440e-01 5.21841764e-01 4.02850628e-01
7.75714874e-01 7.12763608e-01 3.38912487e-01 -3.91185135e-01
-3.28548968e-01 -6.68152571e-01 -1.96284905e-01 4.79943037e-01
-2.21372306e-01 -5.92490852e-01 -1.01190996e+00 7.24032462e-01
-1.77381051e+00 -6.29719794e-01 -3.16667199e-01 1.72517025e+00
7.39645839e-01 2.04360381e-01 -2.02345699e-01 -3.28968838e-02
5.11850357e-01 3.37739050e-01 -4.05324459e-01 -3.30778420e-01
-2.77897090e-01 1.78943276e-01 2.24896297e-01 5.85113883e-01
-1.17199838e+00 1.34236419e+00 5.55075645e+00 7.68882036e-01
-9.01325822e-01 -3.47107440e-01 6.17106974e-01 -3.35122980e-02
-7.68295646e-01 1.81280822e-01 -7.72845089e-01 -4.98726554e-02
2.14750096e-01 -2.37649083e-02 3.99107397e-01 6.28953636e-01
2.99877733e-01 -5.32829911e-02 -1.19180560e+00 6.73792720e-01
5.80302030e-02 -1.18079031e+00 2.99214065e-01 2.97344834e-01
5.39148271e-01 -5.74161410e-01 -2.64981040e-03 1.49847537e-01
4.73420501e-01 -1.39698672e+00 6.33541167e-01 3.84876311e-01
8.47348273e-01 -7.68117666e-01 3.94501716e-01 -1.77319832e-02
-1.81978416e+00 1.81850463e-01 -9.29526314e-02 -1.26191705e-01
2.50905544e-01 6.11559153e-01 -8.77344966e-01 9.11831439e-01
4.92339492e-01 1.04112089e+00 -8.78780603e-01 5.60361028e-01
-7.40416706e-01 4.39300448e-01 -2.58502334e-01 8.45954716e-02
5.02744392e-02 -3.40456158e-01 2.04714224e-01 1.28767288e+00
2.60618851e-02 2.76126146e-01 5.37409186e-02 1.20241785e+00
-2.27274865e-01 3.30920964e-01 -8.53666782e-01 -2.89847940e-01
5.95459878e-01 1.25351071e+00 -9.07537282e-01 -1.71277210e-01
-8.72507215e-01 7.96870112e-01 9.68578339e-01 7.04313815e-01
-7.22455263e-01 -4.73595232e-01 4.65296775e-01 3.22909564e-01
6.03403687e-01 -4.99287426e-01 -8.98995221e-01 -9.85070467e-01
4.06641304e-01 -4.83848304e-01 3.93114716e-01 -8.66406798e-01
-1.09401906e+00 7.67239332e-01 -3.57104875e-02 -1.03522217e+00
1.52841687e-01 -5.33484697e-01 -6.48181915e-01 9.69386637e-01
-1.49304223e+00 -1.73014331e+00 -2.51532763e-01 3.36797059e-01
3.00930738e-01 1.67990774e-01 6.86193883e-01 2.09958345e-01
-1.00615871e+00 7.04247177e-01 -5.66774547e-01 6.25774920e-01
1.42161638e-01 -1.37252569e+00 6.12269998e-01 1.11299479e+00
6.77654028e-01 1.24206519e+00 3.57048839e-01 -9.35881257e-01
-1.26514709e+00 -1.13207710e+00 1.11507416e+00 -4.32765752e-01
6.29485846e-01 -6.92775309e-01 -9.12651002e-01 8.46297383e-01
3.11358064e-01 2.31791109e-01 8.86528969e-01 6.42765582e-01
-8.61514986e-01 9.86429900e-02 -5.88528395e-01 8.21763098e-01
1.52390647e+00 -7.14639246e-01 -3.80464733e-01 -3.13281193e-02
6.89991951e-01 -1.67239070e-01 -7.84256279e-01 4.34804708e-01
4.73662943e-01 -5.76440454e-01 1.18915975e+00 -6.90653741e-01
7.54926741e-01 -4.11160439e-01 -8.38810951e-02 -1.03128576e+00
-5.34693778e-01 -3.05219106e-02 -7.65763521e-01 1.34414077e+00
8.44590425e-01 1.47395516e-02 8.77052248e-01 7.90700436e-01
9.05591473e-02 -9.91959274e-01 -4.76907223e-01 -4.84240294e-01
-1.43269375e-01 -4.01745111e-01 6.58924103e-01 1.09788060e+00
2.01017305e-01 8.18773210e-01 -2.26540685e-01 6.17288291e-01
1.62602499e-01 4.27374244e-01 6.58904672e-01 -1.04346550e+00
-4.36836481e-01 -5.00740290e-01 -1.45082548e-01 -1.49065542e+00
1.72276571e-01 -7.73377895e-01 -2.55343467e-01 -2.12763476e+00
1.84181243e-01 -4.94628429e-01 -2.68622309e-01 6.32492840e-01
-5.22751093e-01 -1.15766235e-01 -2.92219669e-02 -4.57995012e-02
-7.47253537e-01 5.90901554e-01 1.30665731e+00 -4.77370977e-01
-3.31869483e-01 -4.02679950e-01 -9.97875571e-01 6.43423140e-01
5.51499903e-01 -1.48668557e-01 -6.31446958e-01 -9.74470079e-01
5.34807086e-01 -2.98309833e-01 1.94940850e-01 -4.80001628e-01
1.67365879e-01 -3.02843004e-01 5.01900911e-01 -7.50868380e-01
2.31976077e-01 -1.07636571e+00 -2.25628480e-01 -2.59562433e-01
-3.90156806e-01 2.26795506e-02 2.88512379e-01 6.95726097e-01
-2.04014063e-01 5.83162010e-02 5.78105785e-02 1.27097843e-02
-8.48164201e-01 2.77808338e-01 5.81706166e-02 7.40847290e-02
8.64608109e-01 -3.53244513e-01 -3.55871946e-01 -2.75943965e-01
-6.32907987e-01 2.00465620e-01 3.49035054e-01 5.26237130e-01
7.94418693e-01 -1.33926105e+00 -3.98185313e-01 1.92271993e-01
5.26405215e-01 4.26676989e-01 1.23083420e-01 3.68963361e-01
-4.00198996e-01 2.98659235e-01 2.02571511e-01 -1.63466558e-01
-1.41048992e+00 7.08144188e-01 1.16696782e-01 -5.37219763e-01
-6.64786875e-01 9.59650159e-01 4.14600581e-01 -1.14840060e-01
2.51481801e-01 -4.48443651e-01 -4.94604945e-01 -1.53822899e-01
2.99101204e-01 -4.27931373e-04 -9.41913426e-02 -7.41423607e-01
-5.13503134e-01 6.17309391e-01 -3.29989374e-01 2.58491617e-02
1.25166965e+00 -2.13628694e-01 -5.01554422e-02 1.46973908e-01
9.56660450e-01 3.05437624e-01 -1.15877020e+00 -4.68589842e-01
5.73520660e-01 -5.69793999e-01 1.68449908e-01 -7.95475304e-01
-1.10468423e+00 9.00519609e-01 -5.11315539e-02 1.83765948e-01
9.21286583e-01 1.43642962e-01 5.12975037e-01 2.99767613e-01
1.62262827e-01 -9.31254566e-01 6.77765682e-02 4.81754452e-01
9.29676652e-01 -1.17672002e+00 4.56619203e-01 -1.24751735e+00
-4.83375162e-01 7.76310265e-01 1.01431584e+00 6.70091212e-02
7.22444534e-01 3.06168377e-01 -1.19519778e-01 -4.54507649e-01
-5.66854119e-01 -5.35077512e-01 9.08684313e-01 8.29948485e-01
8.85446548e-01 1.39574558e-01 -1.84751838e-01 8.28537643e-01
-3.68781723e-02 -3.74307960e-01 -1.19481020e-01 1.01712549e+00
-1.44396991e-01 -1.17134273e+00 4.14458290e-02 2.89737493e-01
-1.23094358e-01 -4.13895696e-01 -1.00409818e+00 7.94735014e-01
3.04658204e-01 9.40507472e-01 2.05709770e-01 -4.01852369e-01
4.52882499e-01 -3.05448979e-01 4.27787542e-01 -7.87629128e-01
-3.74330044e-01 4.68468107e-02 3.38331431e-01 -3.80067319e-01
-2.30659723e-01 -2.86017001e-01 -1.41963303e+00 -5.90164922e-02
-3.39211285e-01 -1.83964267e-01 2.24307060e-01 9.60608065e-01
5.78773558e-01 7.46066153e-01 3.14990193e-01 -2.12777257e-01
3.58494103e-01 -9.30769265e-01 -5.43973744e-01 4.61289108e-01
-3.52268219e-02 -6.86849713e-01 1.17097855e-01 1.58808023e-01] | [9.216463088989258, 8.207855224609375] |
9107a8da-1896-4f77-8b7f-891c25ef50c7 | ynu-hpcc-at-semeval-2021-task-10-using-a | null | null | https://aclanthology.org/2021.semeval-1.184 | https://aclanthology.org/2021.semeval-1.184.pdf | YNU-HPCC at SemEval-2021 Task 10: Using a Transformer-based Source-Free Domain Adaptation Model for Semantic Processing | Data sharing restrictions are common in NLP datasets. The purpose of this task is to develop a model trained in a source domain to make predictions for a target domain with related domain data. To address the issue, the organizers provided the models that fine-tuned a large number of source domain data on pre-trained models and the dev data for participants. But the source domain data was not distributed. This paper describes the provided model to the NER (Name entity recognition) task and the ways to develop the model. As a little data provided, pre-trained models are suitable to solve the cross-domain tasks. The models fine-tuned by large number of another domain could be effective in new domain because the task had no change. | ['Xuejie Zhang', 'Jin Wang', 'Zhewen Yu'] | 2021-08-01 | null | null | null | semeval-2021 | ['source-free-domain-adaptation'] | ['computer-vision'] | [-4.28934395e-01 5.41486979e-01 -9.28753689e-02 -9.26612914e-01
-6.58421516e-01 -9.26108122e-01 5.80352664e-01 1.22317925e-01
-8.48822057e-01 1.36040962e+00 4.33387756e-01 9.10116509e-02
-2.33368110e-03 -8.16676378e-01 -5.97198904e-01 9.69771743e-02
4.21452016e-01 1.12169015e+00 5.20857215e-01 -5.93661845e-01
2.58586943e-01 3.86598766e-01 -1.13935673e+00 7.63251722e-01
9.70118761e-01 2.50029892e-01 4.05849636e-01 4.54783082e-01
-7.25531578e-01 2.14363351e-01 -1.01524949e+00 -4.91315663e-01
3.90628487e-01 -1.68858603e-01 -1.23731554e+00 -7.11943269e-01
2.48539403e-01 7.69594982e-02 -7.42331892e-02 8.15965593e-01
7.63968945e-01 4.32683200e-01 7.55933702e-01 -1.42870569e+00
-1.25111115e+00 8.69027793e-01 8.89607593e-02 2.58049637e-01
3.88446480e-01 -6.53391600e-01 4.80928093e-01 -7.52995849e-01
1.15383422e+00 1.09738708e+00 7.27387011e-01 1.02996397e+00
-9.63237166e-01 -1.04789996e+00 -5.90872429e-02 1.42337054e-01
-1.26834738e+00 -2.95338690e-01 4.86425608e-01 -3.68895203e-01
1.30122852e+00 -1.95137151e-02 -2.18630388e-01 1.32968140e+00
-2.82932043e-01 3.66758592e-02 1.08184040e+00 -6.97372317e-01
8.06800053e-02 9.97329712e-01 3.02008599e-01 -1.34777993e-01
3.13639045e-01 1.35916650e-01 -3.42475742e-01 -2.55461305e-01
4.84930664e-01 -2.57782072e-01 1.53536499e-01 -2.46737257e-01
-9.77527380e-01 7.93502510e-01 2.26875350e-01 9.16838288e-01
-1.43343329e-01 -8.57563376e-01 4.81742084e-01 6.79729819e-01
5.36867857e-01 1.10325158e+00 -1.45508587e+00 5.96702769e-02
-8.70525181e-01 4.89762574e-01 1.29239368e+00 1.24507129e+00
6.82085812e-01 -5.33923447e-01 -5.73629998e-02 1.16574955e+00
8.83775055e-02 2.81491071e-01 9.18102920e-01 -5.55448472e-01
8.83541226e-01 7.28991985e-01 4.79561210e-01 -5.71630657e-01
-6.40768528e-01 -2.82307379e-02 -5.00475764e-01 -3.28261629e-02
7.57502615e-01 -5.99181116e-01 -9.52556610e-01 1.78176939e+00
3.44907016e-01 -1.40121698e-01 5.24723053e-01 4.73963797e-01
1.18640924e+00 7.81154811e-01 7.10064054e-01 1.86975434e-01
1.26554453e+00 -8.52716565e-01 -8.40066075e-01 -1.71865642e-01
8.88942540e-01 -8.09335828e-01 8.90654922e-01 1.34783775e-01
-7.05561161e-01 -9.69907224e-01 -7.13732958e-01 -2.09017932e-01
-1.26210856e+00 3.17646638e-02 2.78168321e-01 6.06165946e-01
-9.01921034e-01 6.34821415e-01 -1.17568307e-01 -1.16280007e+00
-4.91224714e-02 3.08156073e-01 -7.59887099e-01 -1.27705082e-01
-1.88792145e+00 1.42881739e+00 1.08057618e+00 -7.27916002e-01
-6.00529671e-01 -8.60292137e-01 -5.99563301e-01 -9.85997915e-02
-1.86404616e-01 -4.81535643e-01 1.15772533e+00 -8.94037008e-01
-1.17439890e+00 1.08622682e+00 4.83171865e-02 -4.65038121e-01
3.96111310e-01 -1.04463339e-01 -1.29709208e+00 -5.26957750e-01
6.36143446e-01 7.13447511e-01 2.97517419e-01 -8.38318944e-01
-9.67096925e-01 -3.95309627e-01 -1.95053101e-01 2.45292768e-01
-5.09003580e-01 4.11561996e-01 6.91504925e-02 -5.73825836e-01
-5.19701540e-01 -5.87697089e-01 -1.91938177e-01 -8.27456355e-01
-1.12670541e-01 -5.74519336e-01 5.25615692e-01 -8.32552493e-01
1.39926350e+00 -1.82399237e+00 -3.51890117e-01 -2.12839134e-02
-2.14325637e-01 4.99530345e-01 -3.78100306e-01 6.09469414e-01
-4.85162824e-01 7.18362272e-01 2.28605717e-01 1.96621835e-01
-6.45099953e-02 7.98233598e-02 -2.28028879e-01 -1.74710602e-01
8.86610597e-02 3.76733601e-01 -5.49344122e-01 -4.10465360e-01
-1.58996224e-01 1.05569698e-01 -4.82991457e-01 5.09996533e-01
-1.06852308e-01 3.48880142e-01 -6.70303762e-01 2.34154150e-01
8.83053184e-01 2.03440860e-01 1.99409291e-01 -1.77163020e-01
-2.56184608e-01 4.77883875e-01 -1.35877454e+00 1.93181312e+00
-3.45374972e-01 4.76377666e-01 -1.98105499e-01 -1.07961845e+00
1.37149191e+00 6.30006850e-01 2.94539332e-01 -7.46898890e-01
-1.08626299e-01 3.29218626e-01 -7.00154603e-02 -8.34856093e-01
5.30332208e-01 -2.16155902e-01 -5.77277184e-01 2.65816271e-01
5.33266664e-01 1.97167248e-01 1.30730391e-01 -9.28777009e-02
1.06096792e+00 2.01320872e-01 6.61031902e-01 -3.27068001e-01
5.04936755e-01 7.96477854e-01 7.66400635e-01 6.97722852e-01
-3.55093539e-01 3.46921176e-01 -3.40839550e-02 -6.03987396e-01
-1.46072543e+00 -6.94190204e-01 -4.66255099e-01 1.72160149e+00
-2.11317644e-01 -2.41355076e-01 -7.30396450e-01 -1.08647513e+00
-1.70993298e-01 9.79887068e-01 -6.50964200e-01 1.62810504e-01
-4.99633998e-01 -3.98277044e-01 7.88564682e-01 3.54124486e-01
6.43737197e-01 -1.28549457e+00 9.98763591e-02 4.28039402e-01
-3.69581580e-01 -9.76492882e-01 -2.52481848e-01 3.11152518e-01
-6.85725331e-01 -9.01403666e-01 -8.64546478e-01 -1.21733093e+00
4.89596486e-01 -2.08433315e-01 1.25024235e+00 -5.27964234e-01
1.08986013e-01 -2.78449543e-02 -4.69884574e-01 -8.39677751e-01
-4.38811451e-01 5.35814285e-01 1.08542420e-01 -7.25562036e-01
1.22287405e+00 -3.67173821e-01 3.99243720e-02 5.83762765e-01
-7.19814360e-01 -4.26941872e-01 3.00075769e-01 6.07212424e-01
2.93107480e-01 2.66274691e-01 1.02845776e+00 -1.46926820e+00
1.02255738e+00 -9.04430389e-01 -3.12571228e-01 7.09748387e-01
-5.63977540e-01 2.67656028e-01 6.16959751e-01 -6.03169143e-01
-1.56992555e+00 1.00314088e-01 -9.33213383e-02 1.82657167e-01
-8.49395216e-01 4.18129265e-01 -6.31766498e-01 2.19555542e-01
1.38423514e+00 -4.92348224e-01 -7.27617621e-01 -1.10593688e+00
4.59737182e-01 9.26191270e-01 2.63327271e-01 -7.91782022e-01
6.80563509e-01 -3.17758024e-01 -7.57263422e-01 -2.90240228e-01
-8.87816370e-01 -6.45902991e-01 -1.10741222e+00 4.68606234e-01
1.07017565e+00 -1.01372516e+00 5.44945113e-02 -3.59850749e-02
-1.41641486e+00 -2.15219125e-01 -3.53166252e-01 5.47694445e-01
-1.68124270e-02 -1.39523163e-01 -7.01965094e-02 -3.23038548e-01
-1.28460705e-01 -3.34484488e-01 3.06690395e-01 5.72829962e-01
-4.82892543e-01 -1.28940725e+00 7.27970898e-01 3.61092687e-01
5.86735845e-01 1.72033563e-01 8.86296332e-01 -2.01499343e+00
3.05091619e-01 -1.60250396e-01 -1.55155569e-01 2.84956247e-01
7.80860037e-02 -2.88810730e-01 -9.16247904e-01 8.83509368e-02
8.16530362e-03 -3.20077598e-01 3.01337183e-01 -3.60484682e-02
9.62094724e-01 -2.14481741e-01 -7.46244311e-01 2.02984765e-01
1.38997948e+00 4.36854690e-01 4.75782037e-01 8.47322166e-01
3.30746382e-01 8.46765161e-01 8.19043636e-01 2.04368815e-01
3.99268687e-01 5.87426305e-01 -4.03343588e-01 2.12935116e-02
-8.68346617e-02 -6.53730333e-01 1.00285120e-01 2.27666721e-01
1.30494401e-01 -3.62793326e-01 -1.15202975e+00 9.06941533e-01
-1.74560285e+00 -1.00729990e+00 -1.31447807e-01 1.93695724e+00
1.06778085e+00 -3.71374860e-02 1.40885815e-01 -7.23134518e-01
1.13821435e+00 -4.83190864e-01 -5.52341342e-01 -8.16141605e-01
-1.89057320e-01 4.86964911e-01 6.08926952e-01 2.75898218e-01
-1.04242933e+00 1.17139030e+00 7.17268276e+00 6.34999096e-01
-6.04802966e-01 3.91701132e-01 1.49805874e-01 2.62851685e-01
-3.29192758e-01 1.96470067e-01 -1.56807244e+00 5.27144670e-01
1.47228026e+00 -5.69274366e-01 8.40918869e-02 1.21197152e+00
-1.19791731e-01 4.88817751e-01 -1.26448047e+00 7.10197270e-01
-1.20107315e-01 -1.12444305e+00 9.42142308e-02 7.35750645e-02
8.02105367e-01 2.54761726e-01 -5.01812279e-01 1.08016562e+00
9.95578110e-01 -1.04115474e+00 3.15630101e-02 4.27078754e-01
6.60223663e-01 -5.11109531e-01 9.36606824e-01 7.08250403e-01
-7.10779071e-01 -2.84525845e-02 -1.03228283e+00 -1.14252493e-02
-1.67001206e-02 4.25101429e-01 -1.38921404e+00 4.96640414e-01
9.86453891e-01 4.28942084e-01 -6.70029461e-01 9.31818604e-01
-1.71622843e-01 3.19010913e-01 -2.94749945e-01 1.78470328e-01
-9.88364890e-02 2.52323180e-01 3.76264811e-01 1.35995877e+00
6.40972137e-01 1.15605094e-01 7.40336850e-02 7.37558067e-01
-3.70397598e-01 4.07628000e-01 -9.70487654e-01 -4.23818864e-02
1.02374244e+00 1.24822330e+00 -5.76781891e-02 -2.79726744e-01
-5.76836824e-01 8.44864607e-01 6.49961233e-01 4.30801481e-01
-6.23904288e-01 -6.67315722e-01 5.49900413e-01 4.01774317e-01
1.84436124e-02 2.67345399e-01 -2.83429086e-01 -1.31172705e+00
-3.54847044e-01 -7.70557404e-01 9.73840177e-01 -7.82520473e-01
-1.90722227e+00 8.37253451e-01 2.42307067e-01 -1.07309127e+00
-3.10685575e-01 -6.44040585e-01 -5.36375225e-01 1.30207300e+00
-1.54780614e+00 -1.13334870e+00 -4.66582812e-02 1.02461362e+00
3.69702607e-01 -7.22876132e-01 1.34668875e+00 6.79413259e-01
-2.93221027e-01 6.54913366e-01 2.77682364e-01 6.81331277e-01
1.79553056e+00 -1.16219616e+00 3.99024695e-01 6.59510553e-01
-1.02868415e-01 9.00238931e-01 7.71764338e-01 -1.02226448e+00
-6.88371435e-02 -1.12159598e+00 1.89742792e+00 -9.65698421e-01
3.63893360e-01 -2.36397922e-01 -1.07968831e+00 8.43387365e-01
6.72552764e-01 -1.26899078e-01 1.35048258e+00 5.05369484e-01
-3.31733167e-01 2.18291990e-02 -1.75935066e+00 1.08499655e-04
9.03436661e-01 -2.98844308e-01 -1.60205185e+00 4.01442468e-01
7.36864507e-01 -3.03192943e-01 -1.14398837e+00 1.06301829e-01
6.72198385e-02 -4.18318450e-01 8.99435997e-01 -1.42496049e+00
1.59375980e-01 -1.89296812e-01 -1.74002737e-01 -1.61816406e+00
-5.97813129e-01 -1.87672868e-01 4.91915703e-01 2.04100251e+00
1.02333105e+00 -6.36282086e-01 4.88331586e-01 1.53181529e+00
5.35621084e-02 4.81512964e-01 -7.06239402e-01 -8.90948892e-01
5.71765602e-01 -2.05334686e-02 9.31488454e-01 1.61569858e+00
1.44531518e-01 6.18770003e-01 -1.41359255e-01 1.40398785e-01
1.55010045e-01 -2.04115584e-01 7.49803603e-01 -1.81069291e+00
2.31364772e-01 1.71736836e-01 1.01089828e-01 -4.80627269e-01
3.94307822e-01 -1.06288135e+00 -3.37592930e-01 -1.67410362e+00
-1.79854743e-02 -8.30190659e-01 -5.50657868e-01 8.15555453e-01
-2.20272597e-02 -4.41445529e-01 9.88834351e-02 2.67899066e-01
-4.69085604e-01 2.64810938e-02 7.94516563e-01 2.03807086e-01
-5.81560254e-01 6.41010851e-02 -1.22027266e+00 6.75680041e-01
1.03028834e+00 -1.21421087e+00 -3.60854387e-01 -6.41131103e-01
2.32105792e-01 -2.48521432e-01 -9.23436210e-02 -1.03818369e+00
4.91763890e-01 -3.38031173e-01 8.96361709e-01 -4.33356375e-01
-1.22985825e-01 -1.08071208e+00 1.91611066e-01 -1.00316018e-01
-8.01928341e-01 5.33926859e-02 4.59523648e-01 2.52379507e-01
-4.06684786e-01 -6.08994424e-01 6.93524778e-01 -4.41893607e-01
-1.30097544e+00 1.00344859e-01 4.21573818e-02 5.44911861e-01
1.18829906e+00 -2.37681478e-01 -4.02600348e-01 1.29525915e-01
-1.21576476e+00 1.98102355e-01 3.16101164e-01 8.15385282e-01
1.08744398e-01 -1.51228464e+00 -8.78479898e-01 -3.02950805e-03
3.05084527e-01 -2.25645185e-01 2.39201799e-01 -1.84869185e-01
-1.49296492e-01 7.46101379e-01 -9.51714814e-01 2.01932833e-01
-1.09223771e+00 6.95867300e-01 2.95612842e-01 -7.48012900e-01
-1.82582498e-01 8.29984725e-01 1.24222979e-01 -1.38499761e+00
-6.87065162e-03 -4.24487479e-02 -8.95889580e-01 1.58501536e-01
7.29282022e-01 2.18658403e-01 6.12177700e-02 -2.99327165e-01
-4.64175373e-01 1.78193167e-01 -3.02648813e-01 -1.78293824e-01
1.63544619e+00 -4.06262666e-01 1.86122045e-01 1.68935552e-01
1.18084574e+00 -1.50286686e-02 -7.34214008e-01 -5.90291142e-01
3.42797697e-01 -2.01870143e-01 -3.65863204e-01 -1.58752620e+00
-5.22243500e-01 6.26715422e-01 8.14036012e-01 1.01738274e-01
9.00054097e-01 8.28012973e-02 5.62685788e-01 5.41478813e-01
3.42683136e-01 -1.70671344e+00 -4.46111530e-01 9.03077722e-01
7.83148468e-01 -1.48108721e+00 -2.57879972e-01 -1.34582743e-01
-1.02274978e+00 1.19352567e+00 1.15690899e+00 -9.55532268e-02
1.05137539e+00 1.59968197e-01 7.55869299e-02 -2.77011786e-02
-6.06372833e-01 -1.18008396e-03 3.30200732e-01 1.36947441e+00
7.27503240e-01 4.49889600e-02 -5.28113961e-01 1.45622802e+00
-3.28121722e-01 4.12130773e-01 3.15079182e-01 5.77744126e-01
-4.56621200e-01 -1.80953813e+00 -2.90060043e-01 1.66348964e-01
-6.34111285e-01 -2.47320414e-01 -8.02776337e-01 1.11199677e+00
6.04247510e-01 1.01249254e+00 -9.16821510e-02 -1.57225639e-01
8.54523242e-01 5.07819593e-01 -1.35407269e-01 -1.08920968e+00
-1.03073680e+00 -5.23340166e-01 4.56045270e-01 -2.30960459e-01
-3.80327016e-01 -3.48509103e-01 -1.12856829e+00 -3.05997342e-01
5.45647964e-02 6.42479897e-01 6.56465173e-01 8.45689595e-01
7.22364366e-01 1.57129124e-01 2.71799952e-01 -7.76700526e-02
-3.34604412e-01 -1.29517615e+00 -6.46825194e-01 8.23456049e-01
-2.55532086e-01 -4.59066659e-01 -3.72027233e-03 2.22262725e-01] | [9.81234073638916, 9.488790512084961] |
aee28961-ea2b-45bf-9f2c-3bb4c6470818 | experimental-evidence-supporting-a-new | 1201.0912 | null | https://arxiv.org/abs/1201.0912v2 | https://arxiv.org/pdf/1201.0912v2.pdf | Experimental Evidence Supporting a New "Osmosis Law & Theory" Derived New Formula that Improves van't Hoff Osmotic Pressure Equation | Experimental data were used to support a new concept of osmotic force and a new osmotic law that can explain the osmotic process without the difficulties encountered with van't Hoff osmotic pressure theory. Derived new osmotic formula with curvilinear equation (via new osmotic law) overcomes the limitations and incompleteness of van't Hoff (linear) osmotic pressure equation, $\pi=(n/v)RT$, (for ideal dilute solution only). The application of this classical theory often resulted in contradiction regardless of miscellaneous explaining efforts. This is due to the lack of a scientific concept like "osmotic force" that we believe can elaborate the osmotic process. Via this new concept, the proposed new osmotic law and derived new osmotic pressure equation will greatly complete and improve the theoretical consistency within the scientific framework of osmosis. | ['Hung-Chung Huang', 'Gaochao Lin', 'Rongqing Xie'] | 2012-01-02 | null | null | null | null | ['miscellaneous'] | ['miscellaneous'] | [-6.52363300e-02 -1.15379527e-01 -1.22199476e-01 -1.07796378e-01
6.14338398e-01 -4.85015422e-01 4.15731594e-03 7.41526842e-01
-4.51462626e-01 1.25992703e+00 -2.95552909e-01 -6.73143327e-01
-2.14829624e-01 -6.30724072e-01 -4.92333204e-01 -6.36685312e-01
-2.49478921e-01 2.49326870e-01 9.85392649e-03 -4.11231846e-01
9.77628887e-01 5.03433049e-01 -1.59223461e+00 -3.18302572e-01
1.44496846e+00 3.85166228e-01 4.41462129e-01 7.37658083e-01
-6.66495383e-01 2.72418827e-01 -3.02796513e-01 -1.08563386e-01
6.83143213e-02 -4.59827751e-01 -5.70800662e-01 -5.00135303e-01
-3.69600616e-02 -2.15274900e-01 2.09537223e-01 7.68258035e-01
3.07196826e-01 -7.54555315e-02 9.71838176e-01 -1.10122383e+00
-9.83943701e-01 3.91325146e-01 -7.79023707e-01 -4.46920320e-02
4.12415117e-01 5.75805530e-02 3.42233598e-01 -6.68351710e-01
8.35007608e-01 1.03655255e+00 4.80086505e-01 6.37455821e-01
-9.41478789e-01 -5.51109791e-01 -5.11331446e-02 -1.41304359e-01
-1.04947531e+00 2.62768596e-01 5.05039513e-01 -1.90497234e-01
9.77659762e-01 4.00195241e-01 9.87730801e-01 2.63405014e-02
9.35194612e-01 3.13150376e-01 1.20881069e+00 -4.67716068e-01
3.74407113e-01 4.76029277e-01 -3.86967473e-02 2.87580699e-01
1.28535402e+00 -3.64863425e-01 -1.89093888e-01 -5.99271692e-02
7.03351557e-01 -2.00706199e-01 -2.38757655e-01 -6.22137845e-01
-3.26686919e-01 5.91601193e-01 1.24812059e-01 5.59461236e-01
-6.62357286e-02 1.53220326e-01 5.31355143e-01 1.57123163e-01
-9.21574757e-02 6.63167953e-01 -8.24811280e-01 -1.07566141e-01
-5.66678107e-01 6.35980964e-01 1.22974515e+00 6.72117651e-01
3.57802838e-01 3.76602352e-01 8.69252563e-01 5.63403904e-01
8.80463004e-01 6.68490291e-01 5.77586412e-01 -7.56221473e-01
1.06777176e-01 6.95637405e-01 4.31973338e-01 -6.02359712e-01
-2.12221235e-01 -3.04624557e-01 -3.65209699e-01 5.37105739e-01
6.41011775e-01 -8.17068815e-02 -9.48698223e-01 1.37604904e+00
2.71703899e-01 -4.68819439e-01 4.34896350e-01 6.11041427e-01
8.15002322e-01 5.18047810e-01 7.84452677e-01 -6.64541781e-01
1.09172857e+00 -2.29476258e-01 -1.06241417e+00 6.85082376e-02
7.99933016e-01 -7.64413118e-01 6.67916119e-01 4.87977415e-01
-1.29838312e+00 -1.47935254e-02 -1.25043273e+00 1.04225621e-01
-8.02435994e-01 -7.53480732e-01 1.06631899e+00 8.36267948e-01
-7.38340437e-01 8.67215991e-01 -4.28881168e-01 -5.92873335e-01
-2.01767847e-01 5.50998569e-01 -4.12764579e-01 2.35037282e-01
-1.12011778e+00 1.30527902e+00 5.11992037e-01 2.42531121e-01
5.61206304e-02 -7.54751742e-01 -8.62424076e-01 1.36607038e-02
-2.53123522e-01 -7.32040703e-01 5.55757344e-01 -1.76459789e-01
-1.27846754e+00 7.03976989e-01 -4.38472331e-01 -4.31502014e-01
3.89105648e-01 -5.37248217e-02 -2.55012989e-01 3.06316733e-01
-1.00510940e-01 4.51367587e-01 -2.08900452e-01 -1.74672234e+00
-1.54946998e-01 -5.28425753e-01 -5.09655833e-01 1.40900180e-01
4.18687850e-01 -2.49177843e-01 6.43944800e-01 9.91101339e-02
9.55251217e-01 -3.43856007e-01 -4.25860405e-01 -1.43166661e-01
1.97073698e-01 -2.24145338e-01 8.88485193e-01 -3.95390987e-01
8.82170916e-01 -1.76171291e+00 -3.92162681e-01 2.17595220e-01
2.13389710e-01 3.60856146e-01 3.45420271e-01 9.87965524e-01
-4.38122302e-01 6.91163242e-01 -2.69920200e-01 5.20475209e-01
8.72781277e-02 2.12256029e-01 -5.68170212e-02 4.17659611e-01
-1.79389082e-02 4.69492853e-01 -1.03139102e+00 -4.47661936e-01
5.03130913e-01 5.50204992e-01 -2.34992892e-01 -3.03434014e-01
1.25594109e-01 3.14039886e-02 -1.18018895e-01 7.79886901e-01
1.63718641e+00 2.84997582e-01 6.42130852e-01 -1.44418245e-02
-4.71776813e-01 2.26849178e-03 -1.27265227e+00 1.09987116e+00
3.17476839e-01 8.12936276e-02 -5.93501423e-03 -8.30019295e-01
1.44926643e+00 2.69362837e-01 5.85226238e-01 -6.53030932e-01
2.95471847e-01 8.62462819e-01 3.73866677e-01 -8.65310788e-01
4.74558592e-01 -9.12967741e-01 5.61733305e-01 -2.02582374e-01
-2.74064485e-02 -5.25536776e-01 3.05546433e-01 1.47535115e-01
3.36773843e-01 6.77579939e-01 4.83718514e-01 -7.98203230e-01
6.64278030e-01 -3.75007570e-01 5.68615735e-01 2.63962179e-01
-8.04174781e-01 -7.26980492e-02 6.08660936e-01 -2.32982591e-01
-1.31030941e+00 -1.01604247e+00 -8.06918323e-01 3.40084672e-01
5.97057283e-01 -1.95530355e-01 -2.20758513e-01 3.09743434e-01
7.29531169e-01 5.46424329e-01 -6.20405734e-01 2.61123151e-01
-1.56589285e-01 -9.30105269e-01 2.90331930e-01 2.19527125e-01
1.42555028e-01 -7.03230262e-01 -3.48491818e-01 4.07530904e-01
3.90628248e-01 -5.58371365e-01 7.26461411e-01 5.37181139e-01
-1.34805322e+00 -1.15156293e+00 -5.58671415e-01 -4.17886585e-01
6.24360442e-01 1.86472669e-01 4.24280375e-01 4.77193594e-01
-3.97075891e-01 -6.71457946e-02 -2.60836631e-01 -9.26008463e-01
-3.73318642e-01 -5.05179703e-01 9.76179540e-02 -1.08561409e+00
9.36408281e-01 -8.93837988e-01 -9.90105033e-01 -8.96768197e-02
-8.47364128e-01 -5.30052543e-01 1.35509208e-01 2.63529062e-01
1.71370625e-01 -2.45264858e-01 1.23955202e+00 -7.61492789e-01
4.67241764e-01 -3.74150693e-01 -2.74344862e-01 -2.48790368e-01
-1.08635557e+00 9.95461345e-02 3.95936728e-01 -1.12901047e-01
-7.66612887e-01 -6.25755668e-01 -1.94962382e-01 3.30881745e-01
-7.09240437e-02 4.43878353e-01 4.89808619e-02 -2.94091970e-01
4.00660962e-01 1.55635402e-01 1.38881728e-01 -3.37708205e-01
-1.17990606e-01 3.94384235e-01 2.50819385e-01 -5.80051124e-01
5.52937269e-01 2.50095814e-01 7.43811727e-01 -6.08379483e-01
-3.97827476e-02 -3.75099331e-01 -6.72931373e-01 1.80444848e-02
7.37053096e-01 -4.50499445e-01 -1.50403249e+00 1.63039684e-01
-7.43873596e-01 2.60272771e-01 -1.54105574e-01 8.71974885e-01
-5.08378804e-01 1.00630331e+00 -7.83541024e-01 -1.75464988e+00
-4.98457760e-01 -6.31177723e-01 5.97113296e-02 7.62915730e-01
-1.34267852e-01 -1.20985317e+00 1.06470801e-01 1.88832581e-01
5.65772712e-01 6.99786663e-01 9.59031820e-01 -3.26878667e-01
-3.31926882e-01 -3.50619167e-01 -2.77458549e-01 1.20000228e-01
-6.14095479e-02 6.49353921e-01 -7.87354231e-01 8.13603178e-02
1.06638908e-01 7.43943900e-02 6.91872776e-01 5.58459640e-01
5.82521260e-01 2.16459900e-01 -3.10743123e-01 3.18069547e-01
2.35717010e+00 9.93657231e-01 1.08344972e+00 6.51883841e-01
1.20302111e-01 3.80056828e-01 6.48306549e-01 4.82824624e-01
3.43427539e-01 -3.11143726e-01 5.55470109e-01 -5.57654304e-03
4.45976913e-01 -2.56387889e-01 -1.94066241e-01 8.37168872e-01
-7.04576135e-01 -9.80557576e-02 -5.73905885e-01 4.32805359e-01
-1.54313862e+00 -9.89699304e-01 -9.25121069e-01 2.43048525e+00
8.19797814e-01 2.04188332e-01 -1.42817721e-01 4.24991697e-01
4.15973216e-01 -4.27858114e-01 -4.02156979e-01 -1.34531951e+00
-2.37941027e-01 4.37401980e-01 7.60665536e-01 8.57361376e-01
-4.17726904e-01 6.39443159e-01 7.86875439e+00 4.04613055e-02
-1.12160873e+00 -4.69682723e-01 -1.57252848e-01 4.49427515e-01
-5.79540789e-01 4.95239913e-01 -9.32510018e-01 6.00792825e-01
8.76031697e-01 -7.17286408e-01 -2.70402193e-01 5.68145216e-01
4.67437834e-01 -9.93821025e-01 -8.17826629e-01 5.62695622e-01
-4.91995588e-02 -8.84852052e-01 9.37574878e-02 3.74492317e-01
4.09134150e-01 -4.64437187e-01 -2.20944092e-01 1.75133824e-01
-1.20935142e-01 -7.99116552e-01 -7.00326487e-02 3.85052234e-01
3.38164955e-01 -6.96551561e-01 1.07168031e+00 4.89203893e-02
-1.07376826e+00 1.80115715e-01 -8.41758430e-01 -5.56151867e-01
1.98017627e-01 3.69682759e-01 -9.82578695e-01 6.15918338e-01
2.14012280e-01 3.12006474e-02 -1.01052493e-01 1.47180784e+00
1.66594818e-01 -6.14525266e-02 -5.69736719e-01 -1.46962866e-01
1.96760610e-01 -7.30656803e-01 2.13301599e-01 9.38066602e-01
3.57512087e-01 1.33941069e-01 -5.86947262e-01 8.75418246e-01
3.29116881e-01 9.01412785e-01 -8.09958041e-01 -8.78635421e-02
3.67265165e-01 9.15703475e-01 -7.69397974e-01 -4.16653544e-01
-1.43527523e-01 4.07822102e-01 -3.71291161e-01 3.35054606e-01
-5.21777868e-01 -5.89220524e-01 3.26437980e-01 3.48673344e-01
4.71683182e-02 -2.39712015e-01 -5.90534985e-01 -1.03454852e+00
-2.83036530e-01 7.96937346e-02 3.11354958e-02 -8.55305970e-01
-7.43578434e-01 -2.50251561e-01 -8.92550871e-03 -7.57335663e-01
1.50725007e-01 -8.36169422e-01 -6.64351285e-01 1.19259953e+00
-1.60900950e+00 -7.82125831e-01 -3.91049683e-02 -1.06071487e-01
8.09294656e-02 3.75351757e-01 8.44600976e-01 8.04226175e-02
-2.75611766e-02 3.00716043e-01 3.90013546e-01 -5.87007463e-01
1.01409876e+00 -1.55352485e+00 -4.40225571e-01 2.50116289e-01
-1.51877308e+00 9.31616008e-01 1.42109990e+00 -9.87151563e-01
-1.49211979e+00 4.80628898e-03 1.15833735e+00 -1.74933344e-01
5.46870708e-01 5.86925112e-02 -1.20177245e+00 2.76908547e-01
2.51664460e-01 -6.32861197e-01 1.23142338e+00 2.27132067e-02
-1.25194177e-01 4.12136409e-03 -1.77712882e+00 5.01709104e-01
3.21208090e-01 1.33458510e-01 -6.19576633e-01 2.33556837e-01
6.03531599e-01 -1.99870884e-01 -1.32037890e+00 5.83222926e-01
1.07352912e+00 -8.08470786e-01 7.62755036e-01 -5.81162453e-01
4.14136231e-01 -3.68317902e-01 -2.47824676e-02 -5.87243617e-01
-4.49774824e-02 -3.56276006e-01 3.31314087e-01 9.35701489e-01
6.22521043e-01 -1.13232076e+00 5.53482771e-01 1.31267381e+00
-8.42462033e-02 -9.86595273e-01 -7.34984517e-01 -5.25983572e-01
5.92091858e-01 7.44625255e-02 4.61846650e-01 1.32065082e+00
9.26017642e-01 -2.63397306e-01 -9.78438929e-02 -2.53860891e-01
7.29417205e-01 -3.15875858e-01 6.39151633e-01 -1.51216185e+00
7.05108196e-02 -1.71257660e-01 -6.38824582e-01 -4.22153026e-01
-4.84867752e-01 -4.14760411e-01 -5.42670250e-01 -1.84615242e+00
1.89804494e-01 -3.55362624e-01 -4.79076236e-01 -5.54089360e-02
-1.40197843e-01 -2.10150793e-01 3.96026611e-01 2.93876976e-02
2.75073975e-01 2.26444334e-01 1.65935457e+00 3.96554977e-01
-6.10867560e-01 -8.00561845e-01 -9.65649545e-01 6.83730841e-01
8.56197655e-01 -2.89457619e-01 -4.88786906e-01 2.29596660e-01
6.38889134e-01 1.42512932e-01 -2.78409570e-01 -6.95399165e-01
-8.31669867e-02 -6.62163317e-01 6.75485194e-01 -7.80375481e-01
1.48033738e-01 -7.32071280e-01 1.51583821e-01 9.20025826e-01
3.62177938e-01 -7.93268457e-02 3.66828173e-01 1.45192787e-01
1.37797803e-01 -8.63166690e-01 7.66652286e-01 -3.83440256e-01
-3.77872199e-01 -2.20846087e-02 -3.41684908e-01 -7.37822115e-01
1.30836296e+00 -1.02090561e+00 -7.29147315e-01 -1.61479264e-01
-1.14828610e+00 3.47742260e-01 8.37019384e-01 -2.46936619e-01
5.18669784e-01 -9.49636281e-01 -2.27227286e-01 1.72634330e-02
-3.04500520e-01 -2.32770696e-01 5.85672557e-01 1.00553870e+00
-1.43944287e+00 5.13352096e-01 -5.98670781e-01 -2.66948760e-01
-1.10719168e+00 8.41735840e-01 5.13185501e-01 4.61391248e-02
-4.18684572e-01 3.82043600e-01 2.33965561e-01 -1.44873291e-01
-4.24310446e-01 -3.99933815e-01 -3.27249378e-01 -3.10462207e-01
2.73629010e-01 4.61776644e-01 -5.77994227e-01 -1.38297811e-01
-4.75344926e-01 7.72756577e-01 -9.87423807e-02 8.30025375e-02
1.26745188e+00 -3.72130007e-01 -4.28616434e-01 6.82027638e-01
5.70583224e-01 2.77469069e-01 -4.02083933e-01 6.51714444e-01
-4.36004996e-01 -8.15729320e-01 -5.60781538e-01 -9.41015422e-01
-1.53613821e-01 8.44726682e-01 5.58814883e-01 3.42943072e-01
6.44232571e-01 -5.84744573e-01 5.38372040e-01 1.46146178e-01
6.59042969e-02 -1.30222690e+00 -6.08697772e-01 2.12370396e-01
6.12909794e-01 -1.03010321e+00 3.90910000e-01 -1.05397141e+00
-3.46407801e-01 1.33135772e+00 9.32432950e-01 -4.02086198e-01
6.72157288e-01 5.30338168e-01 1.72703430e-01 -6.51062354e-02
-9.14165378e-01 7.65121281e-02 -4.89098161e-01 8.21057975e-01
1.27873766e+00 5.51064946e-02 -1.94129813e+00 1.94157317e-01
-5.87463826e-02 5.03100097e-01 1.11306024e+00 1.25872421e+00
-1.28389180e+00 -1.25428271e+00 -3.65472049e-01 1.74780130e-01
-3.82987797e-01 2.26996481e-01 -1.76792026e-01 1.25850725e+00
3.40825677e-01 6.45489991e-01 -1.41768053e-01 1.54833868e-01
1.20636995e-03 5.80224574e-01 7.62655795e-01 1.27861455e-01
-2.09531441e-01 4.89244610e-01 -1.65444955e-01 1.32904679e-01
-5.65511286e-01 -2.54641026e-01 -1.92778385e+00 -8.53927791e-01
-4.70544159e-01 7.91970491e-01 1.50495768e+00 8.87527466e-01
-2.22058827e-03 3.44977304e-02 2.55287141e-01 -2.90318638e-01
-1.82738632e-01 -1.06607282e+00 -1.39223886e+00 3.15290153e-01
-1.60020545e-01 -6.58774972e-01 -5.95073938e-01 -1.12157561e-01] | [4.816150665283203, 5.012392997741699] |
d182fe48-32b5-4995-a661-3afaba73b3e0 | optimizing-credit-limit-adjustments-under | 2306.15585 | null | https://arxiv.org/abs/2306.15585v1 | https://arxiv.org/pdf/2306.15585v1.pdf | Optimizing Credit Limit Adjustments Under Adversarial Goals Using Reinforcement Learning | Reinforcement learning has been explored for many problems, from video games with deterministic environments to portfolio and operations management in which scenarios are stochastic; however, there have been few attempts to test these methods in banking problems. In this study, we sought to find and automatize an optimal credit card limit adjustment policy by employing reinforcement learning techniques. In particular, because of the historical data available, we considered two possible actions per customer, namely increasing or maintaining an individual's current credit limit. To find this policy, we first formulated this decision-making question as an optimization problem in which the expected profit was maximized; therefore, we balanced two adversarial goals: maximizing the portfolio's revenue and minimizing the portfolio's provisions. Second, given the particularities of our problem, we used an offline learning strategy to simulate the impact of the action based on historical data from a super-app (i.e., a mobile application that offers various services from goods deliveries to financial products) in Latin America to train our reinforcement learning agent. Our results show that a Double Q-learning agent with optimized hyperparameters can outperform other strategies and generate a non-trivial optimal policy reflecting the complex nature of this decision. Our research not only establishes a conceptual structure for applying reinforcement learning framework to credit limit adjustment, presenting an objective technique to make these decisions primarily based on data-driven methods rather than relying only on expert-driven systems but also provides insights into the effect of alternative data usage for determining these modifications. | ['Cristián Bravo', 'Kristina P. Sendova', 'Alejandro Correa-Bahnsen', 'Jesús Solano', 'Sherly Alfonso-Sánchez'] | 2023-06-27 | null | null | null | null | ['q-learning', 'decision-making'] | ['methodology', 'reasoning'] | [-5.24800308e-02 6.17381670e-02 -3.89124215e-01 1.01201572e-02
-3.70169014e-01 -6.54594302e-01 2.36592516e-01 3.85333300e-02
-7.30572104e-01 8.82479370e-01 -1.35982290e-01 -9.19810176e-01
-4.07680243e-01 -1.09849584e+00 -6.09180450e-01 -5.22940576e-01
-1.37053683e-01 5.52830279e-01 5.30958168e-05 -4.89080101e-01
5.82139790e-01 4.93288368e-01 -1.28423691e+00 -1.08077481e-01
7.32966483e-01 9.36272800e-01 1.12139016e-01 6.49444759e-01
-2.78668031e-02 7.20689714e-01 -7.91567266e-01 -6.46573961e-01
6.10053241e-01 -3.13923150e-01 -4.49440569e-01 1.89494178e-01
-4.82409865e-01 -8.62858295e-01 -8.83737132e-02 7.74531007e-01
5.17102420e-01 2.86354482e-01 4.86004025e-01 -1.32223332e+00
-3.28527570e-01 5.71695566e-01 -3.70743334e-01 3.51728618e-01
5.35961032e-01 5.20676374e-01 1.01556933e+00 1.05090281e-02
2.14552209e-01 8.81473362e-01 3.32069695e-01 5.57673335e-01
-9.38644946e-01 -4.14596885e-01 1.95138454e-01 2.03301311e-01
-7.90154755e-01 3.79037447e-02 4.51249957e-01 -3.39878112e-01
1.17409647e+00 3.43058594e-02 9.44546103e-01 7.21885145e-01
3.36830050e-01 6.92136347e-01 1.00463021e+00 -7.52714157e-01
6.12199962e-01 3.34834635e-01 -5.44320464e-01 9.76431817e-02
1.48006722e-01 4.83206272e-01 2.94434261e-02 -3.35088432e-01
8.88358474e-01 -1.05471194e-01 1.34307928e-02 -2.92843163e-01
-4.91999298e-01 1.09396338e+00 -1.95126057e-01 -5.60251288e-02
-7.30615795e-01 3.28944027e-02 3.45231980e-01 5.27648687e-01
-1.81733638e-01 6.19564235e-01 -5.64787388e-01 -5.04554331e-01
-6.10886037e-01 5.16541183e-01 1.11378026e+00 5.28868794e-01
4.11909968e-01 3.03255975e-01 -9.08174366e-03 5.47216177e-01
2.37403512e-01 2.78451622e-01 5.02367258e-01 -1.17114592e+00
7.12450981e-01 1.58217847e-01 5.80222368e-01 -7.07268357e-01
-2.31373906e-01 -8.70730802e-02 -1.42641753e-01 5.51833928e-01
6.69557691e-01 -9.21811223e-01 -4.66920376e-01 1.49949467e+00
1.54571056e-01 1.12956930e-02 1.67175680e-01 9.16265249e-01
-1.90308064e-01 6.16170049e-01 1.02450080e-01 -5.59861422e-01
1.01056266e+00 -4.50562835e-01 -5.55435538e-01 -2.31931685e-03
4.81246948e-01 -3.86079848e-01 9.29790735e-01 6.90237284e-01
-1.27354717e+00 -1.20416336e-01 -1.02497280e+00 1.09966385e+00
-9.86089930e-02 -3.95886391e-01 6.42974555e-01 1.14991724e+00
-7.27868676e-01 8.08461428e-01 -6.59187853e-01 -1.84201121e-01
1.22635208e-01 4.81622934e-01 3.25836331e-01 3.01894665e-01
-1.57517278e+00 9.68811870e-01 2.72118509e-01 -1.34821847e-01
-7.56184816e-01 -3.66759360e-01 -4.31072235e-01 3.38480771e-01
8.52158487e-01 -5.24635911e-01 1.58074093e+00 -1.40919185e+00
-1.82553816e+00 2.24462256e-01 6.77102327e-01 -7.55041540e-01
7.54003644e-01 1.38975903e-01 -3.33864540e-01 6.17345124e-02
-1.04993843e-01 1.65006667e-01 6.57155871e-01 -8.32392514e-01
-8.93728554e-01 -1.93835065e-01 8.16615522e-01 6.05150223e-01
-2.95814604e-01 3.24160568e-02 -7.38683417e-02 -6.44144177e-01
-5.25619447e-01 -9.78625059e-01 -5.76124191e-01 -6.17360532e-01
6.76543713e-02 1.22553304e-01 1.51177526e-01 -5.05285025e-01
1.36445034e+00 -1.77630675e+00 -2.33022466e-01 4.62952673e-01
-4.79876816e-01 2.53153294e-01 4.17714193e-02 8.28627348e-01
1.41904071e-01 2.97629505e-01 2.78375689e-02 3.01738083e-01
6.91513345e-02 2.55273193e-01 -2.90634930e-01 7.57319778e-02
2.27903891e-02 4.78847712e-01 -7.27567554e-01 -1.37624860e-01
1.64291322e-01 -2.32647017e-01 -8.62248540e-01 3.49984944e-01
-3.36202055e-01 7.11011365e-02 -5.81006110e-01 7.70107925e-01
3.85032326e-01 2.24913463e-01 5.09428859e-01 3.82462978e-01
6.27537593e-02 -8.38382319e-02 -1.55192053e+00 8.11001301e-01
-6.24226809e-01 7.00326860e-02 1.17384195e-02 -1.35709202e+00
7.24210441e-01 2.62681097e-01 8.87050152e-01 -8.71669114e-01
1.80711463e-01 2.00876556e-02 1.13420181e-01 -7.43945301e-01
4.65800673e-01 -5.75279891e-01 -1.71384290e-01 7.98945248e-01
-2.98570156e-01 -1.03497408e-01 2.56221503e-01 -7.45683834e-02
1.19881129e+00 2.89712518e-01 3.79009932e-01 1.29102319e-01
3.03140640e-01 -6.64277375e-03 7.15016544e-01 1.05420446e+00
-4.56848800e-01 1.79851428e-01 1.00635147e+00 -2.80063212e-01
-7.55629361e-01 -9.12473679e-01 2.05456227e-01 9.73921895e-01
4.19527888e-02 2.12054193e-01 -8.30557287e-01 -6.88170195e-01
3.95199955e-01 1.05289423e+00 -3.14459503e-01 -1.02355056e-01
-4.92028803e-01 -8.86201501e-01 2.13867486e-01 3.22625816e-01
4.11418766e-01 -1.40600991e+00 -1.05839086e+00 6.88889384e-01
2.46219680e-01 -8.41000736e-01 -5.64049482e-01 2.42306545e-01
-7.92688787e-01 -1.24914324e+00 -4.96690422e-01 -9.91815925e-02
2.68999279e-01 -1.62789494e-01 9.88451958e-01 -1.08705200e-01
-1.74157340e-02 7.34162807e-01 -4.00173366e-01 -6.28861308e-01
-5.93501687e-01 1.93753075e-02 1.16945013e-01 -5.01600057e-02
3.13251972e-01 -4.04630065e-01 -6.94487393e-01 4.99655366e-01
-1.18788457e+00 -5.97249150e-01 5.22337973e-01 7.83973277e-01
2.67503470e-01 2.88955092e-01 1.33227515e+00 -1.16909099e+00
1.22792995e+00 -6.64893687e-01 -7.98201323e-01 4.12933439e-01
-1.06338263e+00 -3.64834704e-02 7.65328348e-01 -6.49444342e-01
-1.09857368e+00 -7.11372793e-02 -1.96762830e-02 -2.01253429e-01
5.02804760e-03 6.01979733e-01 -2.72557467e-01 1.83251902e-01
3.63168269e-01 9.98190343e-02 3.46434742e-01 1.32669270e-01
4.84555736e-02 7.81652629e-01 3.83482799e-02 -7.99743056e-01
3.13492090e-01 -3.05466168e-02 -2.19695479e-01 -2.85692602e-01
-3.09777528e-01 -1.33330509e-01 -2.30898723e-01 -4.33949590e-01
3.90507191e-01 -5.01602530e-01 -1.15481257e+00 3.75193417e-01
-5.21701872e-01 -4.78643537e-01 -4.94771928e-01 4.98381644e-01
-1.08431089e+00 3.04196417e-01 -5.54815650e-01 -1.02485073e+00
-1.01368621e-01 -9.70799148e-01 5.90070970e-02 5.06603539e-01
-4.34631407e-02 -1.07888126e+00 2.46732280e-01 3.98223817e-01
3.99812788e-01 2.11785123e-01 8.92701805e-01 -6.65001333e-01
-3.89688402e-01 -1.90204233e-01 3.10379148e-01 3.98633778e-01
1.69238567e-01 8.38343576e-02 -4.54346269e-01 -5.22334874e-01
9.68364254e-03 -2.67264754e-01 -5.66296503e-02 4.50856417e-01
8.94844770e-01 -5.79580665e-01 1.79808572e-01 2.93373704e-01
1.60225773e+00 9.42545593e-01 6.45301342e-01 1.11253083e+00
-4.71487716e-02 5.82888842e-01 1.20616496e+00 1.05870259e+00
3.25350940e-01 7.23336160e-01 7.82763898e-01 2.90455312e-01
7.62742877e-01 -1.19641826e-01 4.57247376e-01 -1.48687899e-01
-2.18125075e-01 -3.28701526e-01 -6.57584608e-01 2.51835465e-01
-1.89057863e+00 -1.18592381e+00 7.45006800e-01 2.47475934e+00
6.24668121e-01 5.67779064e-01 6.72838569e-01 -3.96163799e-02
6.66513145e-01 -2.78090358e-01 -9.31513488e-01 -1.12719667e+00
1.92937091e-01 5.69086336e-02 8.66057515e-01 1.88580692e-01
-7.15861976e-01 5.98133028e-01 6.41609955e+00 6.33494079e-01
-1.01441133e+00 -4.03173536e-01 8.07072282e-01 -3.41537744e-01
-3.90116513e-01 1.24428406e-01 -5.17008841e-01 6.85749233e-01
1.11181247e+00 -3.50070924e-01 7.67481267e-01 6.97874665e-01
7.78137505e-01 -3.75681311e-01 -7.88691401e-01 4.98366296e-01
-4.79825139e-01 -1.10174870e+00 -3.94607745e-02 3.08043867e-01
5.24584711e-01 -5.03142059e-01 -6.33686036e-02 6.90598071e-01
5.78235149e-01 -8.08935046e-01 7.09785283e-01 5.09010673e-01
3.60727400e-01 -1.23889303e+00 9.13692653e-01 5.94745755e-01
-5.70524693e-01 -7.23377585e-01 -1.36039793e-01 -3.88980567e-01
2.59654701e-01 6.13160133e-02 -9.15127695e-01 2.89050996e-01
3.82400960e-01 2.53103729e-02 5.20145409e-02 1.00029767e+00
4.13610712e-02 7.28545308e-01 -1.38947830e-01 -2.68739522e-01
3.68640125e-01 -4.76594597e-01 2.59290606e-01 9.53904212e-01
3.49429309e-01 4.58195150e-01 1.04026765e-01 5.09965301e-01
2.52583712e-01 2.15419158e-01 -4.81985122e-01 8.76027718e-02
5.77216268e-01 9.41148043e-01 -7.36251116e-01 1.60397768e-01
-6.36174917e-01 6.06057823e-01 1.57007016e-02 4.91222113e-01
-9.98765945e-01 -3.99746507e-01 6.53628469e-01 4.10502613e-01
4.42139745e-01 4.70820591e-02 -1.68147326e-01 -7.05148518e-01
-3.83033194e-02 -1.14021623e+00 6.07450724e-01 -3.86818618e-01
-9.96484578e-01 9.55301300e-02 1.53297901e-01 -1.19982517e+00
-8.90976787e-01 -4.66359645e-01 -8.05595338e-01 7.71061003e-01
-1.38702095e+00 -3.89802933e-01 3.53229433e-01 4.63374138e-01
3.81272554e-01 -5.36170423e-01 4.80062962e-01 5.90859614e-02
-6.46536350e-01 6.87134802e-01 4.29848313e-01 -8.99061486e-02
3.58813375e-01 -1.20793879e+00 -5.28388517e-03 6.58815503e-01
-3.15297216e-01 1.25615343e-01 6.87931240e-01 -5.91220200e-01
-1.24236107e+00 -4.47331190e-01 2.13020682e-01 -2.98646707e-02
8.31544876e-01 1.61097080e-01 -7.66317427e-01 3.25250387e-01
2.24693045e-02 -4.88859415e-01 5.39562166e-01 -2.67796487e-01
4.35417444e-01 -2.41951928e-01 -1.46501338e+00 6.32029951e-01
4.74682510e-01 -2.35481630e-03 -2.30556965e-01 1.70073602e-02
3.12680334e-01 -3.91897440e-01 -8.39908242e-01 2.03857556e-01
4.43692207e-01 -1.03437936e+00 7.50346959e-01 -9.26236987e-01
2.51873791e-01 2.90029556e-01 -1.00317761e-01 -1.40134346e+00
-1.09000333e-01 -7.95615911e-01 -6.27720058e-02 1.09999549e+00
4.07279611e-01 -7.72917509e-01 1.15772629e+00 1.09470129e+00
2.60700554e-01 -1.24253094e+00 -9.76600766e-01 -6.67293906e-01
3.10582727e-01 -3.10029715e-01 8.61467600e-01 6.54965401e-01
6.68331683e-02 -3.53171945e-01 -3.25987339e-01 1.97274774e-01
3.58773589e-01 4.72416244e-02 5.82858384e-01 -4.19400036e-01
-7.74166644e-01 -4.72006589e-01 -1.56513110e-01 -7.12068498e-01
1.51378706e-01 -1.78478956e-01 -1.31140769e-01 -1.10516942e+00
-1.22659937e-01 -7.40342140e-01 -4.08945352e-01 2.47491196e-01
-3.95699367e-02 -5.69195032e-01 5.17270148e-01 1.04369737e-01
-2.01831847e-01 2.64512569e-01 1.09445190e+00 -3.80389020e-02
-5.89743555e-01 9.74322081e-01 -8.15680146e-01 4.63205814e-01
9.43729460e-01 -2.89577991e-01 -5.68540871e-01 4.23440672e-02
3.98225248e-01 8.12647998e-01 -2.48169918e-02 -5.29889345e-01
1.78316265e-01 -9.33629572e-01 -9.19090211e-02 3.79845351e-02
-1.32142633e-01 -8.73104036e-01 6.07574023e-02 5.98832488e-01
-1.80483162e-01 1.72147140e-01 2.06189305e-01 6.19660258e-01
-1.09930791e-01 -1.00045252e+00 4.76747543e-01 -2.98728883e-01
-6.83258355e-01 9.51072425e-02 -9.13477182e-01 7.65763298e-02
1.37521148e+00 -6.01955950e-01 2.57089913e-01 -9.00771916e-01
-7.97257781e-01 4.93676603e-01 3.16063434e-01 2.80455947e-01
4.64742571e-01 -8.66893113e-01 -6.03308439e-01 -9.13198441e-02
-4.26197410e-01 -4.13230360e-01 1.24097444e-01 1.20989822e-01
-6.57509863e-01 2.36827493e-01 -4.47721660e-01 1.14167454e-02
-7.81845689e-01 5.36755383e-01 9.02650356e-01 -5.89324236e-01
-8.81972536e-02 1.74494088e-01 -3.91573906e-01 -2.26514533e-01
2.99189299e-01 2.06935070e-02 -4.92041945e-01 3.17356378e-01
1.74417198e-01 4.75731820e-01 1.26429006e-01 2.36642957e-02
-3.45847718e-02 1.66869029e-01 -7.59615675e-02 -3.35377663e-01
1.30045605e+00 -2.53280908e-01 6.14350498e-01 7.96588659e-02
4.29843962e-01 -2.02661946e-01 -1.62961829e+00 5.63390478e-02
2.15513840e-01 -5.34422934e-01 -1.29072726e-01 -7.74525583e-01
-1.21576226e+00 3.11770111e-01 5.57415426e-01 5.74231446e-01
1.32702065e+00 -6.86673820e-01 4.90898222e-01 3.20659041e-01
5.61073780e-01 -1.62541425e+00 8.39481205e-02 2.07805216e-01
4.28473473e-01 -1.34344542e+00 -6.31555915e-02 6.26047105e-02
-1.05408871e+00 1.25790787e+00 7.47854710e-01 -2.33889706e-02
5.98142207e-01 2.63298392e-01 2.03372642e-01 2.86462814e-01
-7.26212740e-01 4.72598821e-02 -4.51719373e-01 6.63387120e-01
-1.53506948e-02 2.49488577e-01 -3.77470404e-01 6.87129080e-01
1.62098855e-02 3.54028076e-01 1.11127877e+00 1.11555421e+00
-3.04802358e-01 -1.28246319e+00 -4.70565587e-01 7.06067622e-01
-7.20327139e-01 2.41085768e-01 1.07748516e-01 1.05422390e+00
2.06999257e-02 9.68615532e-01 1.42634302e-01 -4.87151146e-02
6.92144871e-01 -2.53829867e-01 3.28245789e-01 -4.92083907e-01
-7.96290934e-01 -1.62553862e-01 6.55215457e-02 -3.58635873e-01
-2.21845955e-01 -1.09903359e+00 -1.27535224e+00 -3.72310221e-01
-2.54727364e-01 2.30585828e-01 4.80429888e-01 1.03764403e+00
-1.01743685e-02 3.93482476e-01 1.10760593e+00 -6.02907419e-01
-1.46571195e+00 -4.68249112e-01 -9.34824824e-01 2.04816297e-01
-9.45554078e-02 -5.32595992e-01 -4.38552052e-01 -3.85137528e-01] | [4.322490692138672, 2.6020255088806152] |
4a38594b-be6e-4924-98c7-28a2cdf52008 | are-pre-trained-transformers-robust-in-intent | null | null | https://aclanthology.org/2022.nlp4convai-1.2 | https://aclanthology.org/2022.nlp4convai-1.2.pdf | Are Pre-trained Transformers Robust in Intent Classification? A Missing Ingredient in Evaluation of Out-of-Scope Intent Detection | Pre-trained Transformer-based models were reported to be robust in intent classification. In this work, we first point out the importance of in-domain out-of-scope detection in few-shot intent recognition tasks and then illustrate the vulnerability of pre-trained Transformer-based models against samples that are in-domain but out-of-scope (ID-OOS). We construct two new datasets, and empirically show that pre-trained models do not perform well on both ID-OOS examples and general out-of-scope examples, especially on fine-grained few-shot intent detection tasks. | ['Philip Yu', 'Caiming Xiong', 'Ye Liu', 'Zhiwei Liu', 'Yao Wan', 'Kazuma Hashimoto', 'JianGuo Zhang'] | null | null | null | null | nlp4convai-acl-2022-5 | ['intent-recognition'] | ['natural-language-processing'] | [ 2.38854110e-01 -1.75901219e-01 -3.20670038e-01 -4.20977086e-01
-9.37339365e-01 -2.70674020e-01 6.96161985e-01 1.08372048e-01
-3.64422500e-01 3.95694703e-01 3.84446591e-01 -1.24337360e-01
-5.46282642e-02 -6.05483174e-01 -9.22553465e-02 -1.85679540e-01
-1.17631257e-01 4.50305432e-01 5.48503697e-01 -6.42846704e-01
5.28980851e-01 8.97684470e-02 -1.42126334e+00 5.38708508e-01
7.53614724e-01 8.80811036e-01 -2.79488146e-01 6.97575212e-01
-1.55782923e-02 1.28962469e+00 -1.19910204e+00 -3.46174002e-01
3.03974506e-02 -3.53063941e-01 -5.85409343e-01 9.65302438e-02
6.11835599e-01 -6.49951160e-01 -5.25324225e-01 8.19378972e-01
2.47590780e-01 4.13350940e-01 1.21087909e+00 -1.32379782e+00
-8.29248428e-01 1.37462348e-01 -5.65818325e-02 9.31134164e-01
8.33238184e-01 4.08548027e-01 1.28063869e+00 -1.03331780e+00
7.31225014e-01 1.00108540e+00 1.09655488e+00 4.35791105e-01
-1.03642642e+00 -4.92947876e-01 -2.33046040e-02 3.84832144e-01
-1.23742127e+00 -5.19192636e-01 8.42479765e-01 -4.77486670e-01
1.79759300e+00 2.02892184e-01 2.33679444e-01 1.69147074e+00
5.18381238e-01 9.37120855e-01 9.06084239e-01 -4.38669622e-01
6.62717283e-01 -1.02804519e-01 1.05325019e+00 5.71131945e-01
2.59801984e-01 1.37601987e-01 -6.86617672e-01 -4.73783255e-01
1.36899114e-01 5.05464196e-01 -3.07750702e-03 -1.09147690e-01
-1.03703928e+00 1.28284526e+00 2.58722097e-01 7.05821514e-01
-2.11225569e-01 -1.63523495e-01 7.89753377e-01 6.03142977e-01
6.99505389e-01 5.28914571e-01 -2.63433188e-01 -3.86274189e-01
-1.28743517e+00 8.32245946e-02 7.98537016e-01 1.02007997e+00
6.94905102e-01 1.86538175e-01 -7.50447154e-01 9.24467623e-01
-2.19513550e-01 6.34537339e-02 1.01372051e+00 -3.40258062e-01
4.14963454e-01 8.20348024e-01 -1.31444097e-01 -5.38282573e-01
-2.15899870e-01 -2.94658422e-01 -5.52105784e-01 -5.77698797e-02
5.77198900e-02 2.05941305e-01 -1.20710158e+00 1.36719799e+00
-2.79197156e-01 -5.97684970e-03 2.56041586e-01 4.66714501e-01
8.76173615e-01 4.39922988e-01 3.45221668e-01 -1.47340104e-01
1.73409200e+00 -1.04499495e+00 -8.46591532e-01 -8.36038411e-01
9.77714717e-01 -3.87948126e-01 1.42652464e+00 -1.52717695e-01
-5.41202068e-01 -5.03825486e-01 -1.15906918e+00 -4.50353697e-02
-8.03115726e-01 2.93480009e-02 4.51813430e-01 7.25158989e-01
-7.48733401e-01 3.15215528e-01 -3.94941360e-01 -8.07631433e-01
7.56444931e-01 -5.90562932e-02 -2.11760953e-01 -1.92890644e-01
-1.48218405e+00 1.04569650e+00 4.71180826e-01 -8.71607125e-01
-1.28581536e+00 -9.54575479e-01 -1.08841777e+00 2.17248157e-01
5.32401025e-01 -5.55725217e-01 1.43365991e+00 -3.49815011e-01
-7.27028131e-01 9.96591568e-01 -2.46583998e-01 -6.52610779e-01
1.43619597e-01 -8.97864178e-02 -8.14167380e-01 9.37403142e-02
5.27036011e-01 3.47063601e-01 1.30554008e+00 -7.07762301e-01
-4.97917861e-01 -2.76488274e-01 2.97872603e-01 2.05396917e-02
-6.72464967e-01 2.89652705e-01 3.20037842e-01 -9.88360822e-01
-4.11926448e-01 -5.06147146e-01 1.27256170e-01 -3.79454553e-01
-4.73334104e-01 -4.93565649e-01 1.46889842e+00 -3.35834593e-01
1.45270479e+00 -2.06117225e+00 -7.26728320e-01 -6.08643055e-01
1.92675889e-02 3.41383427e-01 -4.29411381e-02 6.85212016e-01
-5.25126867e-02 4.91401888e-02 -1.77082807e-01 -4.22701210e-01
2.77229369e-01 3.02474439e-01 -7.60945559e-01 1.93667158e-01
2.30792448e-01 9.83547091e-01 -9.02932227e-01 -5.76171815e-01
4.68850553e-01 -4.00530696e-02 -3.20319653e-01 4.65091705e-01
-3.09059352e-01 -2.23125324e-01 -4.31707978e-01 1.09458029e+00
1.40744016e-01 -3.95816594e-01 -4.53147411e-01 -2.24251091e-01
2.27433443e-01 8.36047351e-01 -2.88647830e-01 1.54724872e+00
-5.46894848e-01 7.98976302e-01 -6.10447228e-01 -8.67214024e-01
7.12222576e-01 3.90691400e-01 -1.04908042e-01 -7.47815549e-01
7.40130320e-02 3.94460186e-02 -1.02639452e-01 -3.30530465e-01
7.89257705e-01 -8.52899611e-01 -6.91928804e-01 7.44833112e-01
5.79030752e-01 -1.61466107e-01 4.27623510e-01 3.67145896e-01
1.57275856e+00 -5.67165792e-01 1.04910862e+00 -2.04714999e-01
3.45899850e-01 3.13410908e-01 1.33046255e-01 1.07106650e+00
-6.92456663e-01 7.75210202e-01 3.36686343e-01 -4.74298269e-01
-8.70953023e-01 -9.23204422e-01 -4.55131158e-02 1.64936841e+00
-1.70837611e-01 -6.43422127e-01 -3.00944328e-01 -1.47210646e+00
-2.30597824e-01 1.51377118e+00 -7.21064031e-01 -4.78893131e-01
-2.00072333e-01 -6.33933127e-01 9.10630643e-01 5.17561138e-01
3.36281508e-01 -1.12638187e+00 -1.07836401e+00 2.09902689e-01
-2.89970189e-01 -1.26072097e+00 -7.40598023e-01 7.12859035e-01
-8.22992802e-01 -9.35168087e-01 -6.82696640e-01 -5.29731452e-01
3.16214919e-01 4.26259428e-01 1.26443100e+00 -2.31136009e-02
-6.37993157e-01 4.75409925e-01 -6.41715705e-01 -3.59252274e-01
-4.15937006e-01 -1.22232204e-02 8.95315595e-03 -3.71021271e-01
9.30636823e-01 -5.06671786e-01 -1.08708329e-01 4.79820997e-01
-8.30215633e-01 -5.23817778e-01 4.39513624e-01 1.13403058e+00
-1.39100090e-01 1.77900076e-01 5.46535313e-01 -6.65552258e-01
9.24897909e-01 -4.98350948e-01 1.03255175e-01 2.97529191e-01
-4.78510320e-01 -4.48531583e-02 8.19533587e-01 -6.28710806e-01
-1.07271647e+00 -5.28229475e-01 -1.90869853e-01 -8.93692255e-01
-5.88998199e-01 2.80607671e-01 2.60200024e-01 1.54018611e-01
9.96096313e-01 6.57392085e-01 -5.10625601e-01 -2.57097274e-01
-1.47535384e-01 7.77806818e-01 1.33474961e-01 -7.84313679e-02
7.05010772e-01 5.07141292e-01 -3.37119132e-01 -1.22300506e+00
-1.30392468e+00 -8.30984950e-01 -6.24329031e-01 2.08036937e-02
6.15037858e-01 -8.94289970e-01 1.43424287e-01 5.92168570e-01
-9.61407602e-01 -3.30395550e-01 -7.93359578e-01 -3.96663547e-02
-6.10697508e-01 3.28627110e-01 -7.08230853e-01 -9.69988823e-01
-5.63309014e-01 -9.08820033e-01 1.31808543e+00 7.00336769e-02
-8.48971665e-01 -1.32963479e+00 2.85876155e-01 3.44839633e-01
4.27498311e-01 -3.10145855e-01 9.34331298e-01 -1.54269516e+00
-8.44104365e-02 -5.89483559e-01 -3.50997932e-02 1.80938765e-01
3.41813892e-01 -5.35961926e-01 -1.18242085e+00 -1.59830660e-01
6.21181369e-01 -8.53237092e-01 1.01397622e+00 1.16090633e-01
4.67704833e-01 -2.99002200e-01 -4.45205092e-01 2.67685235e-01
1.22365820e+00 2.59254277e-01 8.17152321e-01 3.11402708e-01
3.59392345e-01 3.18755150e-01 1.10683715e+00 4.20936197e-01
1.01088166e-01 8.86871338e-01 1.48346676e-02 3.90379697e-01
-3.24331999e-01 -3.04925650e-01 6.42501891e-01 2.69112438e-01
5.05257726e-01 -2.90453374e-01 -9.01038706e-01 8.85989547e-01
-1.55530989e+00 -1.35840833e+00 1.50003225e-01 1.70179939e+00
5.85818946e-01 4.32722360e-01 2.93066114e-01 1.23150870e-01
7.24139273e-01 7.25363910e-01 -4.10463512e-01 -5.10006905e-01
3.48722339e-02 2.51068681e-01 6.14978112e-02 2.47458424e-02
-1.41078091e+00 9.49586034e-01 7.77617979e+00 1.27397442e+00
-6.24730825e-01 5.38852513e-01 5.29854298e-01 3.15890834e-02
-9.50456113e-02 -3.86562794e-02 -8.99444342e-01 5.05195498e-01
1.05511963e+00 -8.71389583e-02 -1.57260984e-01 1.35308611e+00
-3.41782212e-01 -1.31341994e-01 -1.26955795e+00 6.92850828e-01
5.38003325e-01 -1.09920728e+00 -9.21269786e-03 -1.79068632e-02
6.61348760e-01 -8.89082924e-02 -8.31529200e-02 1.13132548e+00
2.10911483e-01 -7.92127669e-01 4.99897093e-01 -5.62620088e-02
9.35477734e-01 -3.60941947e-01 8.13844979e-01 6.27908587e-01
-1.20812345e+00 -1.75003290e-01 -5.45468330e-01 -2.96507269e-01
1.40536830e-01 5.97244143e-01 -9.93689954e-01 2.78579026e-01
5.18482983e-01 9.10786271e-01 -7.97240615e-01 6.20203316e-01
-1.74558893e-01 6.20485902e-01 -1.63781047e-01 -1.89654469e-01
3.80006969e-01 3.00312221e-01 7.76531994e-01 1.50907958e+00
2.98352867e-01 7.09366277e-02 1.08522795e-01 1.06203246e+00
1.74822509e-01 -4.88764405e-01 -9.51593697e-01 -3.59835625e-01
1.22256838e-01 1.07095611e+00 -7.44475782e-01 -8.52761805e-01
-4.57926691e-01 9.33796465e-01 2.87173420e-01 1.90377347e-02
-9.15758610e-01 -4.35943902e-01 9.05277789e-01 9.92589295e-02
6.62572443e-01 2.50608802e-01 -2.54063487e-01 -1.62857056e+00
-3.44267190e-01 -8.17707896e-01 9.33029175e-01 -7.75360882e-01
-1.70684159e+00 3.88880134e-01 2.45959476e-01 -1.49258745e+00
-6.34422958e-01 -4.26673084e-01 -1.30177665e+00 4.54092771e-01
-1.32475948e+00 -1.07206941e+00 -2.00437188e-01 6.14405930e-01
1.28820455e+00 -3.30522388e-01 9.93911266e-01 5.18211257e-03
-3.72171193e-01 7.00937867e-01 -3.43813002e-01 2.08822027e-01
6.71163082e-01 -9.42485988e-01 8.39644372e-01 9.32552755e-01
5.10288700e-02 7.74702251e-01 8.88635755e-01 -8.13307106e-01
-1.09806275e+00 -1.15507638e+00 9.50125813e-01 -6.27036154e-01
9.39738393e-01 -4.30531770e-01 -8.77038598e-01 1.07570386e+00
3.63292880e-02 7.95377865e-02 7.69799411e-01 2.31822848e-01
-8.16040933e-01 5.21268308e-01 -1.46043849e+00 5.08522213e-01
1.24152803e+00 -9.74825442e-01 -1.77417362e+00 4.44448739e-01
5.77395737e-01 -9.12673771e-02 -4.35899109e-01 5.01111090e-01
3.14767927e-01 -1.44147420e+00 1.28638816e+00 -6.60286009e-01
4.25691158e-01 1.93914771e-01 -2.88618892e-01 -1.14489901e+00
-2.81777889e-01 -1.28687024e-01 -5.44679880e-01 9.07387376e-01
-1.28219590e-01 -6.02842510e-01 8.04037511e-01 2.13491693e-01
-4.16066796e-01 -3.93018246e-01 -1.06425679e+00 -1.32124186e+00
-3.01726341e-01 -6.82492197e-01 8.41386318e-02 8.54406059e-01
5.44503689e-01 7.67931521e-01 -3.99834543e-01 -3.43375623e-01
7.05390751e-01 2.04386115e-01 6.27638400e-01 -1.16659784e+00
-1.71701014e-01 -2.39695668e-01 -6.37164533e-01 -9.25344467e-01
2.51274616e-01 -5.08732498e-01 9.03554112e-02 -1.48255742e+00
4.07274663e-01 -8.64052027e-03 -2.46701375e-01 8.36931765e-01
-3.71892393e-01 4.41661060e-01 6.10978864e-02 5.44014335e-01
-1.03672922e+00 5.49746513e-01 5.76987863e-01 -5.02882898e-01
1.43577307e-01 3.79811488e-02 -4.54371154e-01 7.22299337e-01
5.38585544e-01 -8.42568457e-01 -5.15362561e-01 6.23850822e-01
-4.81502861e-01 -9.33324620e-02 3.73396218e-01 -1.35259688e+00
6.24316232e-03 -1.06034629e-01 9.78887156e-02 -9.86293435e-01
7.17251599e-01 -6.91298842e-01 -2.62647659e-01 8.82673621e-01
-3.33987176e-01 -4.15660203e-01 1.11340545e-01 6.33386135e-01
-2.13893622e-01 -8.45368087e-01 7.08860159e-01 -3.40633243e-01
-1.28674018e+00 -1.25354901e-01 -7.70100653e-01 6.26541317e-01
1.16301024e+00 -5.32844126e-01 -7.11565673e-01 -5.14915586e-01
-4.23950464e-01 -3.53765637e-01 6.01920545e-01 4.21556324e-01
7.65829444e-01 -1.07034683e+00 -2.21716821e-01 3.71138990e-01
9.61895406e-01 -8.05343747e-01 5.91423750e-01 7.80442476e-01
-4.17708186e-03 9.68264282e-01 -9.03620124e-02 -6.14024520e-01
-1.32409394e+00 1.03742242e+00 1.50682479e-01 -8.60983491e-01
-6.97647929e-01 7.25777686e-01 5.61190903e-01 -5.03897071e-01
5.59221357e-02 -3.45760226e-01 1.52608869e-03 1.68167740e-01
7.58231401e-01 3.11866194e-01 1.43049583e-01 -6.23579979e-01
-7.41927862e-01 2.27876514e-01 -4.17253584e-01 2.04616666e-01
9.27327156e-01 1.31573468e-01 4.67511773e-01 7.49220431e-01
1.49656546e+00 -3.49101305e-01 -7.83994496e-01 1.57662705e-02
-5.89595817e-04 -5.36985874e-01 -7.16598630e-02 -8.04194927e-01
-2.69516349e-01 8.49903941e-01 2.43344694e-01 4.94532168e-01
9.38275635e-01 1.04135282e-01 9.14942026e-01 6.39474869e-01
5.74901104e-01 -1.01151967e+00 8.09545755e-01 1.05240762e+00
5.15793741e-01 -1.36628211e+00 4.74315733e-02 -2.28855371e-01
-9.59879756e-01 9.37662721e-01 4.55588490e-01 -3.53244632e-01
3.99305820e-01 2.59095222e-01 -2.72261590e-01 -5.97128272e-01
-9.89265084e-01 -4.21875656e-01 3.62393677e-01 7.97917247e-01
1.60553053e-01 -2.38155037e-01 1.70559511e-01 4.71701324e-01
-4.84691234e-03 1.66225985e-01 5.79770267e-01 1.43144822e+00
-6.64133072e-01 -4.88690853e-01 -5.30905783e-01 8.84439588e-01
-2.61270612e-01 -4.31720644e-01 -4.98545706e-01 8.97446275e-01
-3.84114981e-01 1.12267506e+00 4.32440862e-02 -6.36098266e-01
2.50783712e-01 7.59936750e-01 1.21152259e-01 -1.00225306e+00
-7.46485174e-01 -3.89344066e-01 3.52228701e-01 -3.81819576e-01
-1.33202046e-01 -4.43958551e-01 -7.43209243e-01 -3.83244492e-02
-3.94481838e-01 -1.22251615e-01 5.26651461e-03 1.24669456e+00
3.22221369e-01 6.55797124e-01 2.68905163e-01 -7.76547074e-01
-8.72684598e-01 -1.26888800e+00 -6.38834000e-01 6.31768048e-01
5.63489199e-01 -6.94976330e-01 -1.12203789e+00 -2.60156304e-01] | [12.01559066772461, 7.519038200378418] |
4fcc68db-d44a-48dd-bb0c-b8267c1e4a6d | semantic-parsing-of-colonoscopy-videos-with | 2306.06960 | null | https://arxiv.org/abs/2306.06960v1 | https://arxiv.org/pdf/2306.06960v1.pdf | Semantic Parsing of Colonoscopy Videos with Multi-Label Temporal Networks | Following the successful debut of polyp detection and characterization, more advanced automation tools are being developed for colonoscopy. The new automation tasks, such as quality metrics or report generation, require understanding of the procedure flow that includes activities, events, anatomical landmarks, etc. In this work we present a method for automatic semantic parsing of colonoscopy videos. The method uses a novel DL multi-label temporal segmentation model trained in supervised and unsupervised regimes. We evaluate the accuracy of the method on a test set of over 300 annotated colonoscopy videos, and use ablation to explore the relative importance of various method's components. | ['Roman Goldenberg', 'Ehud Rivlin', 'Or Weinstein', 'Ori Kelner'] | 2023-06-12 | null | null | null | null | ['semantic-parsing'] | ['natural-language-processing'] | [ 5.12891293e-01 1.10301889e-01 -2.69213527e-01 -5.13329148e-01
-7.18693912e-01 -9.72683251e-01 4.94439423e-01 9.20030355e-01
-3.80314857e-01 4.94766176e-01 4.26920056e-01 -4.29286391e-01
-2.42121950e-01 -5.69908798e-01 -5.67376137e-01 -3.37174088e-01
-3.45649302e-01 2.63194710e-01 4.46957558e-01 3.17388147e-01
2.82851428e-01 1.32973850e-01 -1.08628356e+00 6.75800025e-01
6.29638433e-01 6.97500587e-01 1.55248344e-01 1.19354916e+00
-6.40652776e-02 8.52853239e-01 -6.42352879e-01 -1.88417077e-01
-6.35445341e-02 -6.26926899e-01 -1.06427133e+00 1.76715374e-01
3.56140547e-02 -1.73944965e-01 1.52999848e-01 1.06418347e+00
2.48663038e-01 3.13046239e-02 6.44083977e-01 -5.23034155e-01
1.16069792e-02 8.34102929e-01 -7.34672770e-02 4.61191684e-01
7.85060167e-01 -1.55737191e-01 5.83960712e-01 -1.07694775e-01
9.87601519e-01 7.42724538e-01 6.94434166e-01 2.47040167e-01
-8.15163732e-01 -7.93261826e-02 7.84770921e-02 -2.80426025e-01
-8.06968510e-01 -1.52534368e-02 3.56821537e-01 -8.31780195e-01
7.22393692e-01 2.73812473e-01 7.94673085e-01 1.07087970e+00
3.53260458e-01 7.04929292e-01 1.08440912e+00 -7.63253689e-01
1.62989855e-01 3.12945604e-01 4.20873284e-01 1.07970631e+00
5.68674326e-01 5.21470867e-02 -1.85895309e-01 -1.54433608e-01
7.52099156e-01 -1.02098227e-01 -2.25336075e-01 -3.24527025e-01
-1.37888432e+00 7.56854594e-01 2.15405375e-02 6.72067881e-01
-8.76179934e-02 1.18663371e-01 6.80236936e-01 -1.57941177e-01
9.51678529e-02 6.01863146e-01 -3.03210527e-01 -2.00239092e-01
-1.05761516e+00 -2.07774788e-01 1.22511017e+00 1.14281344e+00
7.22410828e-02 -6.61265075e-01 -2.92449921e-01 3.13108265e-01
2.90867776e-01 -3.12253945e-02 8.52162004e-01 -8.56130302e-01
1.73558712e-01 7.72017539e-01 1.89349964e-01 -8.11455846e-01
-8.63517702e-01 -3.20689172e-01 -4.04214829e-01 -3.72784495e-01
3.86129767e-01 -1.58100933e-01 -1.05312884e+00 1.17019844e+00
1.44415170e-01 -7.80903175e-02 5.84663823e-03 5.17586827e-01
9.63743150e-01 3.02693367e-01 4.64306980e-01 -2.99253643e-01
1.50437987e+00 -1.02772355e+00 -8.37146461e-01 -1.05542712e-01
9.63515103e-01 -9.31641638e-01 5.00728846e-01 5.64819515e-01
-9.98797357e-01 -6.11504614e-01 -9.79176164e-01 1.55126616e-01
-4.85815763e-01 5.36002994e-01 7.79947460e-01 8.23780179e-01
-7.42844224e-01 5.70950091e-01 -1.21306491e+00 -5.67826390e-01
2.49972254e-01 3.59745622e-01 -2.27529794e-01 2.31388863e-02
-7.84241438e-01 8.18923175e-01 6.83659554e-01 -3.51847038e-02
-9.88207221e-01 -1.94139823e-01 -1.11719108e+00 -2.77915180e-01
4.17224020e-01 -6.47033751e-01 1.66028190e+00 -8.48390102e-01
-1.43780386e+00 1.18854749e+00 2.28241414e-01 -4.85280424e-01
4.71998274e-01 -3.01102847e-01 -4.68810171e-01 5.09726107e-01
-3.49875423e-03 3.63360226e-01 4.55466241e-01 -9.18437779e-01
-8.41341794e-01 -6.98506385e-02 2.11143121e-01 -1.07693952e-02
6.90311193e-02 1.80651732e-02 -4.20471400e-01 -4.98517424e-01
1.73883364e-02 -1.13586557e+00 -6.14259720e-01 -1.77786872e-01
-4.12516683e-01 2.77584940e-01 1.44372240e-01 -7.52807140e-01
1.43208623e+00 -2.20041060e+00 -8.72013792e-02 2.52680689e-01
-2.48435307e-02 -5.50584756e-02 4.16831970e-01 2.36122131e-01
3.01678479e-02 1.58974364e-01 -2.75710523e-01 -5.86980022e-02
-3.06926310e-01 2.01385692e-01 2.13875815e-01 4.73991573e-01
-1.45721123e-01 6.96925700e-01 -1.25979662e+00 -8.94400418e-01
6.03313327e-01 -2.68646665e-02 -6.34614706e-01 2.11603776e-01
-3.22431892e-01 5.69513977e-01 -7.98288763e-01 7.00005889e-01
-4.38779630e-02 -5.02991438e-01 5.66863775e-01 -3.57226551e-01
-2.82994181e-01 2.26512790e-01 -9.21527565e-01 2.42318153e+00
-4.30985212e-01 3.78011078e-01 -3.47170144e-01 -7.49211192e-01
4.12932962e-01 5.97902477e-01 7.00198531e-01 -1.43695295e-01
3.73030722e-01 3.05730850e-01 3.31950635e-02 -1.09838414e+00
4.44123238e-01 1.52364552e-01 -3.71331781e-01 9.59808901e-02
2.78407902e-01 -1.87166944e-01 5.21748126e-01 5.50005697e-02
1.33752441e+00 5.62340856e-01 1.04625297e+00 -4.15040135e-01
6.07995391e-01 6.04424179e-01 1.75259933e-01 8.17242980e-01
-4.83317912e-01 5.55062473e-01 4.53494251e-01 -5.92782974e-01
-5.32094002e-01 -6.96522355e-01 -1.30494103e-01 5.36633849e-01
2.52563655e-01 -4.54194069e-01 -7.86583245e-01 -1.12489462e+00
-4.99948680e-01 5.38648844e-01 -6.11145318e-01 -4.38330090e-03
-7.56383479e-01 -1.07444501e+00 3.11021090e-01 3.14757466e-01
1.58851326e-01 -1.04534376e+00 -1.27835989e+00 6.59943283e-01
-2.41458237e-01 -1.42047548e+00 -2.76102692e-01 2.58804381e-01
-1.07885349e+00 -1.48344731e+00 -5.05116701e-01 -9.61891115e-01
8.20196927e-01 -1.56359792e-01 1.23934758e+00 -1.27426848e-01
-7.99975455e-01 6.97989762e-01 -7.35754490e-01 -3.37489575e-01
-6.82989061e-01 3.96939740e-02 -7.26040661e-01 -3.89765084e-01
7.64716193e-02 2.50106659e-02 -7.22234011e-01 1.07931860e-01
-9.59629476e-01 1.52168915e-01 6.39734387e-01 5.37515581e-01
5.76101661e-01 -2.68604130e-01 4.53215279e-02 -1.44738424e+00
4.83559877e-01 -4.35222745e-01 -5.36106586e-01 5.42286456e-01
-3.17487389e-01 2.46568665e-01 2.35636681e-01 -3.79565895e-01
-1.09456170e+00 4.49054033e-01 -6.81239069e-02 2.00741678e-01
-4.33360338e-01 6.98525965e-01 5.03014743e-01 -1.08923607e-01
7.52486706e-01 5.02479402e-03 -4.29526329e-01 -2.78149962e-01
2.69584179e-01 2.94897914e-01 4.25485164e-01 -3.37163955e-01
2.46167611e-02 5.45864165e-01 -1.16760924e-01 -4.79523659e-01
-1.03349268e+00 -7.81339288e-01 -5.02308428e-01 -3.19823354e-01
1.08728373e+00 -6.48690343e-01 -3.58610302e-01 6.00741990e-02
-1.18380332e+00 -1.85818911e-01 -4.47183102e-01 9.71936166e-01
-5.63463092e-01 4.09885556e-01 -8.80709589e-01 -4.47986454e-01
-1.86561331e-01 -1.39413142e+00 1.03013086e+00 3.48479033e-01
-3.47479135e-01 -1.23095083e+00 4.45898682e-01 1.69954732e-01
2.05055736e-02 6.21975005e-01 7.36405194e-01 -7.96934128e-01
-6.24347270e-01 -3.59456092e-01 -3.30549851e-02 1.68254867e-01
2.91987985e-01 8.41172412e-02 -7.01019406e-01 -2.34123692e-02
2.76535124e-01 1.26811981e-01 5.04356742e-01 7.46413410e-01
1.22294962e+00 2.19782330e-02 -7.72801936e-01 5.64388812e-01
1.54126930e+00 5.71160853e-01 3.15861166e-01 3.47698420e-01
1.54035464e-01 5.79053760e-01 7.96883941e-01 2.18380451e-01
2.28518933e-01 2.94266164e-01 3.55163008e-01 3.88475694e-02
-1.86460659e-01 -4.05404018e-03 7.05546737e-02 1.04276168e+00
-2.44735166e-01 -3.44420254e-01 -1.03241909e+00 5.89855969e-01
-1.75612354e+00 -4.64116305e-01 5.11526838e-02 1.97294581e+00
6.00704670e-01 3.65832657e-01 -7.47745037e-02 -1.78599760e-01
4.38985646e-01 -1.14563808e-01 1.46003682e-02 -3.64701957e-01
2.87469804e-01 2.36313969e-01 7.19169974e-01 3.67729187e-01
-1.49261272e+00 4.99162346e-01 7.36337328e+00 2.66099781e-01
-8.12354326e-01 2.84677416e-01 5.04699707e-01 2.88559854e-01
1.26635462e-01 -1.07414030e-01 -6.52721107e-01 3.12455654e-01
1.03374612e+00 4.99171242e-02 -4.84531969e-02 6.63076878e-01
1.63497671e-01 -5.37822247e-01 -1.17901075e+00 7.12737918e-01
2.44675443e-01 -1.36746514e+00 -1.78263217e-01 -2.60045975e-01
7.01269209e-01 -1.44367129e-01 -3.69845152e-01 1.00764669e-01
3.09637010e-01 -6.91188335e-01 5.27323008e-01 5.36870778e-01
6.46448314e-01 -3.71129289e-02 9.66941893e-01 4.58730347e-02
-1.24642467e+00 7.87835941e-03 2.96781182e-01 3.61976653e-01
2.82815278e-01 3.30359459e-01 -1.11824155e+00 9.33059454e-01
3.35034996e-01 8.33312511e-01 -5.98494589e-01 1.55031896e+00
-3.09766531e-01 5.28682292e-01 -1.00891240e-01 1.72347273e-03
3.80629778e-01 -9.63176563e-02 4.14266437e-01 1.77943754e+00
4.85643089e-01 1.12159066e-01 4.14712250e-01 1.96371034e-01
2.92564720e-01 3.93302292e-01 -4.99273181e-01 -4.00701225e-01
-8.86215791e-02 1.32788360e+00 -1.37026167e+00 -4.16382730e-01
-5.56714654e-01 8.22102427e-01 -3.00041139e-01 5.23550585e-02
-9.25876081e-01 -5.06003425e-02 -3.32749963e-01 -5.50345033e-02
-3.72774564e-02 -2.72611141e-01 -8.68556052e-02 -1.04351830e+00
-5.47885820e-02 -6.92264795e-01 8.26776505e-01 -6.08454466e-01
-6.38390779e-01 6.99096441e-01 1.51534751e-01 -1.53802156e+00
-3.75081539e-01 -7.54934251e-01 -2.67099619e-01 1.98265806e-01
-1.45859385e+00 -1.06705225e+00 -5.09794295e-01 2.63642013e-01
7.89353073e-01 1.60843089e-01 1.17234445e+00 4.60523158e-01
-2.36145899e-01 -6.94734827e-02 -1.99968025e-01 2.11128980e-01
5.73805988e-01 -1.34225106e+00 -8.93380791e-02 8.36644232e-01
1.61977157e-01 2.15773016e-01 6.80646896e-01 -6.52241409e-01
-9.16652560e-01 -8.93805385e-01 7.80217171e-01 -4.47079718e-01
6.08841419e-01 7.42033124e-02 -1.93587288e-01 8.74829173e-01
8.12119544e-02 -4.21889611e-02 8.15891087e-01 -2.83173382e-01
1.19189255e-01 2.13962913e-01 -1.04561090e+00 1.88714191e-01
9.24785137e-01 -1.91766188e-01 -8.30286503e-01 7.02960908e-01
6.87726200e-01 -8.52111101e-01 -9.70584691e-01 5.43302953e-01
4.01969820e-01 -7.84619093e-01 8.23422670e-01 -4.02816892e-01
3.88798267e-01 -1.52835667e-01 1.16980277e-01 -8.84082079e-01
-3.97220533e-03 -5.39232373e-01 6.07299693e-02 3.98493826e-01
5.13827622e-01 -3.05458814e-01 7.03183115e-01 1.46908507e-01
-3.90848309e-01 -4.09264624e-01 -6.25300229e-01 -3.30362082e-01
-6.82102561e-01 -1.95737600e-01 -1.11072078e-01 8.59397709e-01
1.79964930e-01 -1.36873901e-01 -3.40288831e-03 2.35536635e-01
4.11013097e-01 4.10211772e-01 5.88218041e-04 -9.81865942e-01
-4.46105719e-01 -2.49817207e-01 -5.04445434e-01 -6.61247373e-01
-5.61395586e-01 -7.18820393e-01 3.32737155e-02 -1.80956197e+00
1.82563156e-01 -2.37062722e-01 -3.51262420e-01 2.12292880e-01
-1.48304123e-02 -8.45523849e-02 -1.74889758e-01 -4.04365797e-04
-9.79089797e-01 -3.16834748e-01 1.21531701e+00 -6.65922686e-02
-3.80937785e-01 3.17098588e-01 -2.31708944e-01 9.86762166e-01
6.02266908e-01 -7.00715899e-01 -3.16587925e-01 -2.87181258e-01
3.53867918e-01 4.08969283e-01 2.11938843e-01 -1.17562795e+00
2.89462626e-01 5.92817739e-02 5.47229871e-02 -3.93554479e-01
-2.14514858e-03 -9.39831555e-01 4.17488277e-01 1.07998562e+00
-4.69996363e-01 1.13051608e-01 2.98137575e-01 6.47032201e-01
-5.47958255e-01 -7.31641889e-01 5.02388597e-01 -7.27609813e-01
-7.29238272e-01 -8.59911367e-02 -3.97069097e-01 -7.61950016e-03
1.30114472e+00 -5.99484183e-02 -1.23160183e-02 1.98446572e-01
-1.27751291e+00 5.20760417e-02 1.48527533e-01 3.66999865e-01
3.45776200e-01 -6.73129380e-01 -4.44692820e-01 5.95265925e-02
6.79666772e-02 1.46086991e-01 2.67550796e-01 1.08127558e+00
-1.16367567e+00 6.25745296e-01 4.84165959e-02 -7.25196898e-01
-1.26672089e+00 5.83130240e-01 5.07887483e-01 -8.87329459e-01
-6.82394087e-01 5.58672190e-01 1.57734290e-01 -7.20660016e-02
2.27022544e-01 -9.74097013e-01 -5.83914816e-01 -7.09546581e-02
2.67713040e-01 5.12905605e-02 3.74705680e-02 4.23373021e-02
-3.43673319e-01 4.54280704e-01 -2.69572753e-02 -4.97289076e-02
1.11498177e+00 -4.56959456e-02 1.28804609e-01 5.82570553e-01
6.69234335e-01 -8.02962258e-02 -7.67538011e-01 2.22523943e-01
5.86364150e-01 -5.31888865e-02 -3.33707482e-02 -1.14965725e+00
-5.44798672e-01 5.34232616e-01 7.26258993e-01 3.67014319e-01
1.16291440e+00 -2.93365903e-02 4.36283201e-01 1.25725120e-02
6.47221863e-01 -9.29411709e-01 -1.49121493e-01 2.80251391e-02
3.10726941e-01 -1.23343194e+00 1.53006867e-01 -7.86119044e-01
-4.44507539e-01 1.39942133e+00 5.17687351e-02 2.33786646e-03
7.11463690e-01 3.03460479e-01 5.16707711e-02 -4.63646859e-01
-2.85402268e-01 3.33413035e-02 4.04986352e-01 1.53169781e-01
6.69310927e-01 3.10749918e-01 -6.92797303e-01 5.71939647e-01
3.22792798e-01 5.77746212e-01 5.88951170e-01 1.42934000e+00
-3.95226210e-01 -1.06105077e+00 -4.31534722e-02 4.41438377e-01
-8.88182878e-01 1.42975859e-02 9.90391672e-02 9.97734547e-01
3.99924546e-01 7.89793193e-01 -4.30518776e-01 1.45406798e-01
4.30114895e-01 -7.22546875e-02 8.17672253e-01 -9.48824108e-01
-9.56016243e-01 3.82610947e-01 4.98715788e-01 -6.23216212e-01
-1.08758581e+00 -8.00971329e-01 -1.20231998e+00 8.19144845e-01
-1.45622209e-01 2.85441488e-01 9.22373354e-01 8.97908986e-01
-8.40519965e-02 9.66190994e-01 1.57872796e-01 -3.23650897e-01
-3.17929417e-01 -7.27289975e-01 -1.65636227e-01 5.39369941e-01
1.19407117e-01 -3.11099857e-01 -3.05811912e-01 6.70378149e-01] | [14.100101470947266, -3.2736968994140625] |
ecab1b7b-7546-4d39-b16a-ab508c94efe0 | mlc-at-hecktor-2022-the-effect-and-importance | 2211.16834 | null | https://arxiv.org/abs/2211.16834v1 | https://arxiv.org/pdf/2211.16834v1.pdf | MLC at HECKTOR 2022: The Effect and Importance of Training Data when Analyzing Cases of Head and Neck Tumors using Machine Learning | Head and neck cancers are the fifth most common cancer worldwide, and recently, analysis of Positron Emission Tomography (PET) and Computed Tomography (CT) images has been proposed to identify patients with a prognosis. Even though the results look promising, more research is needed to further validate and improve the results. This paper presents the work done by team MLC for the 2022 version of the HECKTOR grand challenge held at MICCAI 2022. For Task 1, the automatic segmentation task, our approach was, in contrast to earlier solutions using 3D segmentation, to keep it as simple as possible using a 2D model, analyzing every slice as a standalone image. In addition, we were interested in understanding how different modalities influence the results. We proposed two approaches; one using only the CT scans to make predictions and another using a combination of the CT and PET scans. For Task 2, the prediction of recurrence-free survival, we first proposed two approaches, one where we only use patient data and one where we combined the patient data with segmentations from the image model. For the prediction of the first two approaches, we used Random Forest. In our third approach, we combined patient data and image data using XGBoost. Low kidney function might worsen cancer prognosis. In this approach, we therefore estimated the kidney function of the patients and included it as a feature. Overall, we conclude that our simple methods were not able to compete with the highest-ranking submissions, but we still obtained reasonably good scores. We also got interesting insights into how the combination of different modalities can influence the segmentation and predictions. | ['Michael A. Riegler', 'Pål Halvorsen', 'Steven A. Hicks', 'Andrea M. Storås', 'Vajira Thambawita'] | 2022-11-30 | null | null | null | null | ['kidney-function'] | ['medical'] | [-4.62403409e-02 2.86223888e-01 -4.24689651e-01 -5.89859068e-01
-9.52506185e-01 -3.95141274e-01 5.40124714e-01 4.15554255e-01
-7.52637386e-01 7.86917150e-01 3.15511018e-01 -6.47863507e-01
-1.96103722e-01 -6.98699653e-01 -4.71763253e-01 -8.32775712e-01
1.19579703e-01 9.05152023e-01 5.00163078e-01 2.25725234e-01
5.69560640e-02 6.19352698e-01 -1.08351922e+00 4.65764403e-01
6.77527606e-01 9.44174528e-01 4.15238023e-01 6.42237186e-01
-1.03079051e-01 8.68202269e-01 -1.39999449e-01 -9.97337252e-02
1.36268571e-01 -5.73370337e-01 -9.99832869e-01 -1.04124941e-01
1.18352249e-01 -9.71301738e-03 2.25808814e-01 5.35876155e-01
4.72289711e-01 -3.58212113e-01 5.68932712e-01 -1.10906589e+00
3.52906495e-01 5.02009213e-01 -4.58281517e-01 1.98316202e-01
1.41458333e-01 1.03576154e-01 6.68992400e-01 -5.75445831e-01
8.75330508e-01 6.40742064e-01 8.30920994e-01 5.94522595e-01
-1.13593447e+00 -4.26856309e-01 8.15583840e-02 2.73874164e-01
-1.19156301e+00 -2.73656070e-01 9.97117758e-02 -4.56381917e-01
9.07149017e-01 6.88370526e-01 9.57378328e-01 8.22907329e-01
3.72208714e-01 7.14024007e-01 1.44645822e+00 -5.82631767e-01
2.77798742e-01 3.63819420e-01 9.06552151e-02 7.20106006e-01
-2.76045930e-02 1.16605215e-01 9.38792750e-02 -3.81655306e-01
2.33996987e-01 6.66360855e-02 -3.65574837e-01 -4.60542381e-01
-1.28822052e+00 8.49624574e-01 7.15803504e-01 5.87933719e-01
-4.48594034e-01 1.41864702e-01 3.74493599e-01 -4.58175875e-02
4.38092411e-01 3.26155722e-01 -5.76230049e-01 9.38446373e-02
-1.34689140e+00 1.64915204e-01 7.67034352e-01 2.00737610e-01
2.98497498e-01 -8.85094821e-01 -2.90682107e-01 6.91636562e-01
3.59518409e-01 -8.00294280e-02 9.65553701e-01 -5.68429232e-01
2.44030841e-02 4.29317474e-01 -1.19346879e-01 -3.25351179e-01
-1.12166429e+00 -4.01488423e-01 -6.48436368e-01 1.87643781e-01
8.17377567e-01 -1.25050560e-01 -1.50702357e+00 1.42667842e+00
1.29785299e-01 -2.25441325e-02 -2.42577121e-01 1.06810772e+00
7.50064909e-01 3.86003666e-02 5.08389235e-01 -2.91029334e-01
1.54907620e+00 -1.18041563e+00 -3.50118726e-01 -2.52717528e-02
1.24179959e+00 -7.46466041e-01 6.07393444e-01 3.66239190e-01
-1.02669418e+00 -1.15300119e-01 -8.13058436e-01 2.57060021e-01
-3.32156152e-01 6.64758310e-02 7.68989921e-01 8.03085208e-01
-1.16868150e+00 8.60138774e-01 -1.18632805e+00 -7.60895789e-01
4.11257416e-01 5.21753192e-01 -4.37174439e-01 -2.18698546e-01
-9.76263881e-01 1.42445540e+00 2.12655410e-01 -2.76973516e-01
-5.86528063e-01 -7.29339659e-01 -3.10117602e-01 -1.67538881e-01
2.98507273e-01 -9.51692462e-01 1.28574598e+00 -8.42081010e-01
-1.18893766e+00 9.91216123e-01 -2.31666118e-01 -4.50861305e-01
8.03535461e-01 3.11835319e-01 -2.69115418e-02 -2.75244135e-02
1.38525158e-01 7.93989837e-01 4.67785336e-02 -1.01042759e+00
-7.69240320e-01 -6.30160987e-01 -4.16619718e-01 3.29483032e-01
4.32242751e-02 -1.21084698e-01 -7.01781213e-01 -2.54328012e-01
4.30248499e-01 -1.24006104e+00 -6.98550105e-01 -4.99671362e-02
-3.60004574e-01 -9.87532958e-02 3.83796066e-01 -7.45801449e-01
9.38423872e-01 -1.75524127e+00 4.69140746e-02 4.80920970e-01
6.42564893e-02 -1.49303138e-01 2.59406149e-01 3.23776305e-02
-3.04252416e-01 5.50107539e-01 -1.69409692e-01 -4.66625601e-01
-4.01468605e-01 1.22281037e-01 3.78322601e-01 4.25489515e-01
-1.04428321e-01 9.24826682e-01 -7.10464120e-01 -7.84066916e-01
1.12018995e-01 2.51748890e-01 -2.92332023e-01 -2.26798192e-01
-1.34186715e-01 7.19001889e-01 -3.98203373e-01 6.19382858e-01
5.78778028e-01 -3.11097622e-01 3.22217911e-01 -2.47347560e-02
-6.44668937e-02 2.27713272e-01 -6.37587786e-01 1.59625828e+00
-3.52364421e-01 2.38081947e-01 -1.12151831e-01 -8.97300720e-01
4.40583885e-01 4.81465816e-01 9.47987795e-01 -9.20214832e-01
4.72646281e-02 4.10790533e-01 3.10957044e-01 -3.62332433e-01
3.15966434e-03 -6.39050126e-01 1.19138524e-01 3.21770728e-01
-2.66468048e-01 -4.42837417e-01 2.63397396e-02 8.07891488e-02
1.40881836e+00 9.50989779e-03 4.42343086e-01 -2.37808511e-01
3.80336881e-01 3.37369740e-01 4.36255544e-01 7.26903439e-01
-2.73624092e-01 8.99498224e-01 6.55373454e-01 -4.05531019e-01
-7.78797805e-01 -8.59495938e-01 -5.03329039e-01 4.84763265e-01
-1.74541295e-01 -3.00719678e-01 -6.23381078e-01 -1.16709650e+00
-7.65581205e-02 6.91722989e-01 -8.60506117e-01 1.47697747e-01
-4.04674441e-01 -1.35844243e+00 4.61677998e-01 4.44049180e-01
1.35527134e-01 -8.29939365e-01 -5.92956901e-01 1.83408245e-01
-2.22409785e-01 -8.66543233e-01 -2.78125461e-02 6.98012292e-01
-1.31117070e+00 -1.18465245e+00 -1.01621473e+00 -3.90354753e-01
5.60380936e-01 -1.07207246e-01 1.11234355e+00 4.81421381e-01
-3.16449672e-01 8.42870250e-02 -3.38724166e-01 -6.11779511e-01
-3.71532887e-01 3.16852868e-01 -3.59825313e-01 -4.37946528e-01
1.89203799e-01 -1.37459710e-01 -6.57362342e-01 4.98385757e-01
-7.85715044e-01 3.15021306e-01 8.44381750e-01 7.28995085e-01
8.46063495e-01 -3.62489447e-02 1.92053378e-01 -1.26276231e+00
2.80692011e-01 -4.53923911e-01 -2.00144470e-01 2.76122361e-01
-8.17476690e-01 -7.48912841e-02 4.25737202e-01 1.10709719e-01
-7.17643023e-01 5.44190228e-01 -4.15044427e-01 6.20164946e-02
-2.72708744e-01 6.75459206e-01 7.48765469e-02 -7.19791725e-02
5.83414495e-01 -2.55743206e-01 1.73249319e-01 -4.96617049e-01
-2.03312144e-01 4.43742633e-01 -7.53822923e-02 -1.99372798e-01
2.21270770e-01 5.56968987e-01 3.53881896e-01 -4.80566442e-01
-7.42362142e-01 -6.51112556e-01 -5.94958365e-01 -1.52050883e-01
1.05013692e+00 -6.33936882e-01 -3.06184798e-01 4.43337917e-01
-7.80131698e-01 -6.01826072e-01 -1.90510631e-01 7.49570847e-01
-5.27292848e-01 2.24367961e-01 -4.68551785e-01 -4.04451102e-01
-2.61639446e-01 -1.39860308e+00 8.60492766e-01 2.04939574e-01
-3.43903393e-01 -1.06611288e+00 2.77537525e-01 4.91273046e-01
5.59259295e-01 2.51590252e-01 9.72313702e-01 -8.98126602e-01
-3.93763155e-01 -3.47176522e-01 -2.24380866e-01 -8.74629691e-02
-3.28483991e-02 -6.14554025e-02 -9.53729153e-01 -1.62928402e-01
-2.99481787e-02 -8.85318667e-02 1.14594781e+00 6.61983073e-01
1.37722349e+00 3.42902631e-01 -9.40631866e-01 3.66672367e-01
1.42124450e+00 2.46450946e-01 8.97906303e-01 5.83008349e-01
4.75785583e-01 5.53480864e-01 6.53952241e-01 1.55761708e-02
4.24259633e-01 1.00535929e+00 5.70168674e-01 -3.45779598e-01
-1.31063849e-01 2.12162882e-02 -3.79250348e-02 2.89404452e-01
-4.35103476e-01 -2.92761475e-01 -1.27201927e+00 3.59756261e-01
-1.71771073e+00 -6.25463128e-01 -5.21565914e-01 2.21568775e+00
3.83910030e-01 3.22771609e-01 2.09190086e-01 9.94702131e-02
3.68863702e-01 -3.31376880e-01 -2.75071144e-01 -2.44732857e-01
1.45565003e-01 3.51897180e-01 8.64494264e-01 3.42747241e-01
-9.72490430e-01 4.48507845e-01 6.52242041e+00 6.67183340e-01
-1.30872560e+00 2.71197796e-01 1.02776945e+00 -2.27112472e-01
-1.17829636e-01 3.08594465e-01 -5.91179848e-01 4.82624412e-01
1.22404552e+00 3.35073113e-01 -1.38011143e-01 6.59133315e-01
3.66076529e-01 -8.56742799e-01 -1.10014558e+00 2.80563205e-01
-1.36621773e-01 -1.17717505e+00 -3.94831061e-01 4.50567871e-01
3.19719881e-01 4.24363136e-01 -3.48072886e-01 3.03125560e-01
1.65206462e-01 -1.21681869e+00 4.86298352e-01 9.29570138e-01
2.35935301e-01 -2.24433258e-01 1.23177338e+00 6.19494915e-01
-6.05583429e-01 9.10815895e-02 4.79294285e-02 1.77634165e-01
9.78147984e-02 6.71525717e-01 -1.27485478e+00 8.03060293e-01
7.34524071e-01 4.24274117e-01 -9.41388011e-01 1.59316182e+00
8.92182998e-03 6.80994570e-01 -4.37642783e-01 1.96980294e-02
9.06456485e-02 3.17210317e-01 1.62363335e-01 1.01395226e+00
3.80130231e-01 1.18731022e-01 2.07926184e-01 2.62478858e-01
2.86471039e-01 3.06241542e-01 -8.57031941e-02 2.75802553e-01
-8.89162868e-02 1.38238013e+00 -1.31676698e+00 -3.77978832e-01
-4.05614614e-01 7.56959140e-01 1.08036995e-01 -1.97640494e-01
-8.35752130e-01 3.38834792e-01 -8.84797573e-02 5.28281748e-01
5.10461479e-02 3.31678331e-01 -6.03908062e-01 -9.07571197e-01
-2.79645324e-01 -5.33106804e-01 7.77685404e-01 -7.42095768e-01
-1.13955510e+00 5.29288352e-01 2.55541522e-02 -9.38540161e-01
-1.28774434e-01 -5.45066714e-01 -6.31559849e-01 1.03504169e+00
-1.54575276e+00 -9.84316528e-01 -3.93547982e-01 1.95461467e-01
1.89926013e-01 4.52448815e-01 1.06152725e+00 2.29107618e-01
-5.18134654e-01 4.58654314e-01 -3.30701992e-02 -1.13743968e-01
7.94147372e-01 -1.39681280e+00 -7.66917691e-02 2.30666921e-01
-7.56343976e-02 1.21580273e-01 4.54256415e-01 -7.21127748e-01
-7.96171665e-01 -8.00220013e-01 9.63153005e-01 -5.54612398e-01
3.72023433e-01 1.39591143e-01 -7.43850410e-01 4.38313782e-01
-2.10867096e-02 1.10332891e-01 9.22101319e-01 1.74651314e-02
4.22068149e-01 2.50441939e-01 -1.28520167e+00 1.31334141e-01
4.49438959e-01 1.88239723e-01 -1.88681811e-01 3.56593072e-01
1.21274181e-01 -5.47759473e-01 -9.28627074e-01 7.38920152e-01
6.06091976e-01 -1.00903940e+00 8.05224061e-01 -4.62291062e-01
3.35751712e-01 -1.51506066e-01 -6.29450977e-02 -1.38670921e+00
-2.77494639e-01 5.49748957e-01 4.92236853e-01 7.06543088e-01
8.92596900e-01 -5.44801176e-01 1.41104376e+00 9.33570266e-01
-1.63762331e-01 -1.17984819e+00 -9.44990635e-01 -6.39297009e-01
3.95629048e-01 -3.72352242e-01 3.39434296e-01 7.50459433e-01
-8.36122483e-02 8.43290891e-03 7.29277404e-03 -1.16809070e-01
2.61527568e-01 -7.35748187e-02 4.19142276e-01 -1.32311285e+00
-4.39667642e-01 -5.39475441e-01 -5.25187373e-01 -3.46348256e-01
-2.16355667e-01 -1.17726636e+00 -7.91227743e-02 -1.89719427e+00
6.90102279e-01 -7.27725804e-01 -5.38681746e-01 6.75933778e-01
-1.42872110e-01 4.76092815e-01 2.91263461e-01 3.90909195e-01
-2.51415819e-01 -8.25447589e-02 1.25779974e+00 4.83266339e-02
-1.08102523e-01 4.82377559e-01 -6.15096688e-01 8.35276842e-01
7.78912902e-01 -7.08196163e-01 -6.89414367e-02 -4.09546085e-02
-8.02722871e-02 3.95206124e-01 5.87502062e-01 -1.02530646e+00
2.70407140e-01 -1.34702519e-01 8.44901204e-01 -7.26534128e-01
3.80036891e-01 -8.97574365e-01 4.18281764e-01 9.08828378e-01
-2.99751963e-02 -1.64249614e-01 2.03400180e-01 2.55619824e-01
-1.28131762e-01 -5.12031436e-01 7.90533066e-01 -5.68993270e-01
-4.35843229e-01 1.33350879e-01 -5.60436368e-01 -4.14060801e-01
1.27689040e+00 -3.32393378e-01 -6.34758398e-02 -2.09824845e-01
-1.34879172e+00 3.71367067e-01 6.65253758e-01 -1.73364300e-02
2.54791677e-01 -9.39733326e-01 -6.45498395e-01 -2.66715288e-01
-1.82758365e-02 -3.82289179e-02 3.07533652e-01 1.56268442e+00
-5.76114058e-01 5.49998403e-01 -1.15941420e-01 -6.47709668e-01
-1.49422836e+00 3.85938823e-01 6.63051188e-01 -1.03950500e+00
-3.82869750e-01 7.76860178e-01 -7.28032067e-02 -5.48715353e-01
6.83151856e-02 -2.41641611e-01 -3.63905549e-01 2.38167822e-01
2.38893270e-01 6.44435585e-02 6.82401955e-01 -4.05092716e-01
-7.27117896e-01 3.57275665e-01 -3.49719793e-01 -2.49187663e-01
1.39911187e+00 1.09180354e-01 -1.53599709e-01 2.86565989e-01
1.13773274e+00 -1.68893322e-01 -5.96350670e-01 2.11443827e-01
4.27606851e-01 -2.12189198e-01 3.84643495e-01 -1.38606036e+00
-1.10277843e+00 6.47038579e-01 9.06066239e-01 1.33285612e-01
1.25556636e+00 1.21976927e-01 4.74513412e-01 -8.87427777e-02
3.04743141e-01 -7.38706946e-01 -3.98742676e-01 2.11804807e-01
4.75086957e-01 -1.40818799e+00 3.46938431e-01 -5.45736849e-01
-7.34051526e-01 1.15598273e+00 4.99316931e-01 2.21731007e-01
6.70326412e-01 1.89903870e-01 2.48214856e-01 -2.30357990e-01
-8.70536149e-01 -2.45504782e-01 1.46698400e-01 2.58250535e-01
7.54020929e-01 2.72551328e-01 -6.34518862e-01 6.03593767e-01
-2.66215503e-01 4.54199314e-01 4.98375803e-01 9.67941523e-01
-3.62503201e-01 -1.63690841e+00 -3.93103600e-01 9.16708231e-01
-6.47341311e-01 -2.44076885e-02 -6.06017947e-01 1.13759470e+00
2.26423755e-01 5.42038560e-01 -9.79806781e-02 -2.83834130e-01
4.74255502e-01 2.36931384e-01 4.49116677e-01 -7.22072423e-01
-7.60831177e-01 -8.71144049e-03 1.64316267e-01 -5.36308706e-01
-4.47007120e-01 -9.36548054e-01 -1.23432183e+00 2.30148584e-02
-4.77210790e-01 3.90717506e-01 1.22078335e+00 9.43361819e-01
-7.16297030e-02 6.30041957e-01 4.09674168e-01 -7.35370040e-01
-2.38537699e-01 -8.79556417e-01 -4.65707362e-01 -3.83673869e-02
-1.01082092e-02 -5.48900723e-01 -3.20906609e-01 -2.74539530e-01] | [14.79146957397461, -2.565415143966675] |
d8a78f33-d94d-4f22-8e58-93f4bc951df1 | property-inference-attack-graph-neural | 2209.01100 | null | https://arxiv.org/abs/2209.01100v1 | https://arxiv.org/pdf/2209.01100v1.pdf | Property inference attack; Graph neural networks; Privacy attacks and defense; Trustworthy machine learning | With the fast adoption of machine learning (ML) techniques, sharing of ML models is becoming popular. However, ML models are vulnerable to privacy attacks that leak information about the training data. In this work, we focus on a particular type of privacy attacks named property inference attack (PIA) which infers the sensitive properties of the training data through the access to the target ML model. In particular, we consider Graph Neural Networks (GNNs) as the target model, and distribution of particular groups of nodes and links in the training graph as the target property. While the existing work has investigated PIAs that target at graph-level properties, no prior works have studied the inference of node and link properties at group level yet. In this work, we perform the first systematic study of group property inference attacks (GPIA) against GNNs. First, we consider a taxonomy of threat models under both black-box and white-box settings with various types of adversary knowledge, and design six different attacks for these settings. We evaluate the effectiveness of these attacks through extensive experiments on three representative GNN models and three real-world graphs. Our results demonstrate the effectiveness of these attacks whose accuracy outperforms the baseline approaches. Second, we analyze the underlying factors that contribute to GPIA's success, and show that the target model trained on the graphs with or without the target property represents some dissimilarity in model parameters and/or model outputs, which enables the adversary to infer the existence of the property. Further, we design a set of defense mechanisms against the GPIA attacks, and demonstrate that these mechanisms can reduce attack accuracy effectively with small loss on GNN model accuracy. | ['Wendy Hui Wang', 'Xiuling Wang'] | 2022-09-02 | null | null | null | null | ['inference-attack'] | ['adversarial'] | [ 3.39171320e-01 2.80346781e-01 -3.53103340e-01 -8.36660415e-02
-2.74887592e-01 -1.10260475e+00 5.37928760e-01 2.10743576e-01
6.05701879e-02 5.68769753e-01 -4.63367820e-01 -8.82499993e-01
-3.14653099e-01 -1.21786797e+00 -1.05648994e+00 -6.26689970e-01
-5.24277627e-01 -2.41074041e-02 3.85124922e-01 1.02913432e-01
2.47219913e-02 7.25327373e-01 -8.35862517e-01 -8.94853771e-02
5.27780116e-01 7.98842907e-01 -5.53071618e-01 5.39788187e-01
4.49694186e-01 7.48199880e-01 -7.94936180e-01 -9.40935373e-01
7.20373929e-01 -2.89365798e-01 -6.41828299e-01 -4.28543866e-01
5.18260479e-01 -2.82241136e-01 -6.31362438e-01 1.60398579e+00
2.93171018e-01 -3.10522348e-01 3.60672772e-01 -1.95243251e+00
-2.96391279e-01 9.27084804e-01 -5.65910995e-01 2.85197096e-03
2.11160615e-01 7.25294575e-02 9.99293268e-01 -3.83200403e-03
3.64996225e-01 1.09610724e+00 8.24459851e-01 5.92823446e-01
-1.37656641e+00 -1.26903307e+00 1.48559317e-01 -1.94083005e-01
-1.62638378e+00 -2.43826747e-01 8.01343203e-01 -1.44877180e-01
5.43573916e-01 6.33890510e-01 2.68589407e-01 1.26474118e+00
3.18928927e-01 3.86930257e-01 9.82207239e-01 -1.16683744e-01
2.74395198e-01 3.00347984e-01 3.09289366e-01 6.94112122e-01
1.11854351e+00 3.36701095e-01 -2.75483131e-01 -1.03447497e+00
7.71000564e-01 -2.38479659e-01 -4.03779119e-01 -7.04772711e-01
-5.04726529e-01 8.08739305e-01 3.82304430e-01 -7.53893033e-02
6.10679723e-02 3.68550897e-01 3.29777628e-01 4.47481811e-01
1.72024742e-01 4.66321558e-01 -5.02692938e-01 5.46433985e-01
-4.08976018e-01 7.43338466e-02 1.30626500e+00 1.02511239e+00
7.08833635e-01 -1.22957557e-01 1.86760008e-01 -1.10919094e-02
4.65528160e-01 4.44838345e-01 -1.40042156e-01 -3.06809098e-01
5.11212349e-01 5.49952507e-01 -2.50834793e-01 -1.68003798e+00
-1.64060876e-01 -5.23120463e-01 -7.79940426e-01 7.12251291e-03
5.50398290e-01 -3.96759123e-01 -6.14924729e-01 2.25536704e+00
4.04140383e-01 5.93772113e-01 1.49392843e-01 3.62971455e-01
3.66797328e-01 3.15325409e-01 1.63479730e-01 -5.44179976e-03
1.11844265e+00 -5.49543858e-01 -4.27159756e-01 -8.67801607e-02
8.78289282e-01 -3.74838002e-02 6.41663730e-01 5.85277192e-02
-6.73616052e-01 5.70814386e-02 -1.30855191e+00 4.05423880e-01
-5.83839834e-01 -4.67642099e-01 8.13785791e-01 1.44770062e+00
-9.41251576e-01 7.14005768e-01 -8.86035085e-01 -3.31398219e-01
6.46714687e-01 6.58234835e-01 -5.56915164e-01 3.81478257e-02
-1.55599511e+00 4.22117114e-01 5.01017988e-01 -1.72424596e-02
-1.13684058e+00 -9.03053284e-01 -8.69451284e-01 1.40536264e-01
7.23878324e-01 -6.26756907e-01 6.51790857e-01 -8.32230031e-01
-8.86096478e-01 6.57220662e-01 3.79235446e-01 -8.04118454e-01
4.58171934e-01 1.78834081e-01 -6.80488765e-01 8.45607668e-02
-3.70928258e-01 -1.48625106e-01 8.99787784e-01 -1.50411201e+00
-1.74674988e-01 -4.25440341e-01 5.50319493e-01 -3.09233278e-01
-5.09503722e-01 8.16254169e-02 -1.09651998e-01 -3.82605672e-01
-1.58361956e-01 -8.09269965e-01 -1.55436769e-01 -1.65461414e-02
-1.05206287e+00 1.98243946e-01 1.12213552e+00 -3.25122684e-01
1.51590490e+00 -2.29346061e+00 -4.75807995e-01 9.14529681e-01
5.57503343e-01 5.47796488e-01 1.28312968e-02 6.71874523e-01
-6.30525872e-03 8.18581343e-01 -2.14699909e-01 -2.14041427e-01
1.03359707e-01 2.67459810e-01 -6.66858137e-01 6.80983424e-01
-1.60322770e-01 8.12670648e-01 -6.49722338e-01 -2.15535462e-01
-2.76736408e-01 2.82933086e-01 -4.90431130e-01 1.17159687e-01
-2.27321699e-01 1.78771988e-01 -7.81500578e-01 6.12274468e-01
9.99185622e-01 -4.80702639e-01 7.12614954e-01 -1.24807812e-01
6.07897401e-01 2.08006099e-01 -9.75212693e-01 7.71270573e-01
7.35432357e-02 1.54477775e-01 2.27496400e-01 -5.78915894e-01
8.61502588e-01 1.83972791e-01 -1.86630990e-02 -4.94765677e-02
2.40228549e-01 5.92028536e-02 2.61131167e-01 -1.14864476e-01
-9.31953937e-02 1.91925079e-01 -1.95522830e-01 6.26912355e-01
-2.48711526e-01 4.34053540e-01 -4.52536464e-01 3.45438987e-01
1.50241506e+00 -3.49463850e-01 5.29782414e-01 -4.38404024e-01
5.07608891e-01 -4.09308642e-01 6.36055231e-01 1.30196333e+00
-4.54342037e-01 1.53464347e-01 1.02565861e+00 -2.63237655e-01
-7.27725267e-01 -1.01972950e+00 6.07544370e-02 6.46733582e-01
4.25156236e-01 -7.14761317e-01 -8.51949155e-01 -1.25038290e+00
3.60189348e-01 6.48013949e-01 -6.98289275e-01 -5.55827856e-01
-2.66637206e-01 -7.22447276e-01 1.26156926e+00 2.12870881e-01
7.73929238e-01 -6.95910692e-01 -8.03242922e-02 -4.26496476e-01
2.68689603e-01 -1.23641992e+00 -2.89744705e-01 -1.87529549e-01
-5.10340214e-01 -1.42921090e+00 3.28058451e-01 -4.83739525e-01
9.16618824e-01 9.07828733e-02 7.47252226e-01 5.57303369e-01
4.58751358e-02 4.19021279e-01 -1.61439627e-02 -4.97653306e-01
-5.95577896e-01 2.71378070e-01 8.85162279e-02 2.90255427e-01
3.23738813e-01 -8.80773306e-01 -2.13094309e-01 3.43859553e-01
-1.01915622e+00 -2.35113889e-01 4.45309103e-01 1.58744454e-01
2.09178224e-01 5.29228985e-01 3.66929829e-01 -1.58896065e+00
7.86652625e-01 -6.58175111e-01 -8.74311209e-01 5.45895040e-01
-6.74025893e-01 -8.22535232e-02 8.11123848e-01 -6.02606714e-01
-4.47902560e-01 -2.71428794e-01 3.05279493e-01 -5.50582290e-01
3.85295078e-02 6.06008887e-01 -8.64029646e-01 -6.12608194e-01
5.60179949e-01 1.98392617e-03 7.88340494e-02 -2.38987878e-01
-3.69849498e-04 2.15948701e-01 2.73121268e-01 -7.74865329e-01
1.36556196e+00 4.80239451e-01 6.41496718e-01 -5.27206302e-01
-6.86804712e-01 1.27664313e-01 -1.93506584e-01 1.13039374e-01
4.39227313e-01 -4.25721407e-01 -1.02841198e+00 8.21847796e-01
-8.48867536e-01 -3.13060254e-01 2.51911640e-01 1.46848187e-01
-5.71096167e-02 7.07452476e-01 -6.50611758e-01 -8.49842906e-01
-3.67738724e-01 -9.47567523e-01 3.55876476e-01 1.15918040e-01
-3.01230350e-03 -1.33128119e+00 -2.66064852e-01 1.28879696e-01
3.46717894e-01 6.81222200e-01 1.21891320e+00 -1.41584063e+00
-8.10627162e-01 -5.88150203e-01 -2.25222871e-01 2.34792784e-01
1.95300266e-01 -1.19605124e-01 -9.67319250e-01 -7.27708757e-01
2.12516829e-01 -1.15892783e-01 3.54356319e-01 -2.66659316e-02
1.33798480e+00 -8.10735464e-01 -5.02161086e-01 9.19735551e-01
1.59652317e+00 8.14570785e-02 6.75306797e-01 1.74944088e-01
9.19641972e-01 3.94872636e-01 1.54301360e-01 1.62668958e-01
8.48562084e-03 3.42887193e-01 7.87323296e-01 -6.34382218e-02
4.74541664e-01 -8.54859293e-01 2.68200427e-01 1.51505321e-01
2.65644580e-01 -5.83375394e-01 -7.71634221e-01 7.71241114e-02
-1.73308504e+00 -7.36543059e-01 2.99527515e-02 2.52647352e+00
5.90787530e-01 2.47827753e-01 -3.58369797e-02 -6.56593218e-02
8.55948389e-01 4.01941478e-01 -6.08893216e-01 -2.97936827e-01
6.86019193e-03 7.52518922e-02 1.20869541e+00 4.60266888e-01
-1.22920513e+00 7.81167448e-01 6.23212957e+00 8.21951210e-01
-8.24413419e-01 -2.64202565e-01 5.53386152e-01 3.40744376e-01
-4.00268376e-01 4.58700866e-01 -8.78070354e-01 3.09870690e-01
1.00659323e+00 -5.49381375e-01 5.62744379e-01 8.27720940e-01
-3.32946956e-01 3.78229439e-01 -1.22981644e+00 4.46116090e-01
2.69958924e-04 -1.00230014e+00 2.93573409e-01 6.41038239e-01
4.56167310e-01 -3.64591628e-01 2.07488239e-01 2.37346575e-01
7.93943405e-01 -1.13349485e+00 2.36148015e-01 2.85029829e-01
6.98907256e-01 -1.07390988e+00 6.47334278e-01 4.49778616e-01
-9.82437611e-01 -5.74467182e-02 -3.62906784e-01 9.73788202e-02
-3.24879229e-01 1.66393518e-01 -6.87094212e-01 9.93333995e-01
3.96222681e-01 3.35023820e-01 -7.77678072e-01 6.45762682e-01
-4.74806398e-01 1.03617537e+00 -5.21467626e-01 7.62996748e-02
8.67273808e-02 -1.66088209e-01 6.88594162e-01 7.30097532e-01
-1.28083870e-01 -6.43984005e-02 2.67667651e-01 1.06712425e+00
-4.45456982e-01 -2.16393232e-01 -8.93191397e-01 -3.07807773e-01
8.85849595e-01 1.13044095e+00 -5.11050045e-01 7.95383602e-02
-2.86916852e-01 5.63341558e-01 4.11311239e-01 4.58689451e-01
-9.11097467e-01 -4.35685247e-01 9.52375352e-01 1.71243668e-01
1.16121471e-01 -8.35128576e-02 -9.88494381e-02 -1.01955032e+00
-6.32067472e-02 -1.16222131e+00 7.51084447e-01 -1.56970292e-01
-1.53805315e+00 3.49407852e-01 1.13300085e-01 -8.05384159e-01
1.13595933e-01 -4.26592290e-01 -7.82594800e-01 8.52386594e-01
-1.20353127e+00 -1.37580597e+00 9.19631720e-02 7.10205793e-01
-7.40833461e-01 -7.96875134e-02 9.01776671e-01 9.33629200e-02
-8.10946107e-01 1.31124961e+00 -1.55653432e-01 7.33373582e-01
3.79798412e-01 -8.70676100e-01 7.46900678e-01 1.24746966e+00
2.06025630e-01 1.17753208e+00 6.37104988e-01 -9.44245100e-01
-1.52516401e+00 -1.18023098e+00 6.17509127e-01 -4.39718813e-01
9.07931864e-01 -8.74098122e-01 -1.06065226e+00 1.20472920e+00
-3.20945859e-01 2.61442661e-01 7.88307905e-01 5.46501093e-02
-8.79320681e-01 -4.24870513e-02 -1.70784545e+00 6.55163765e-01
1.10465658e+00 -7.61154056e-01 1.42907470e-01 1.08719856e-01
1.05645454e+00 -2.07824886e-01 -8.99978220e-01 5.81049502e-01
4.88956660e-01 -8.76516879e-01 9.61633623e-01 -1.03779030e+00
-2.47662678e-01 -2.82837510e-01 -2.34729707e-01 -7.42068291e-01
-1.01961769e-01 -9.52992678e-01 -3.26374501e-01 1.41983974e+00
6.33505046e-01 -1.26988065e+00 9.11051810e-01 8.82530332e-01
6.25816107e-01 -6.63161337e-01 -7.20773518e-01 -9.47997510e-01
8.36779401e-02 -2.66980916e-01 1.14233994e+00 1.31159401e+00
-2.98231184e-01 -3.00775301e-02 -5.24472177e-01 9.68549371e-01
8.93454611e-01 -8.03369135e-02 1.08784115e+00 -1.15159047e+00
-5.06469846e-01 -1.94809467e-01 -7.77980387e-01 -4.18073505e-01
3.01135510e-01 -1.00784516e+00 -7.03133345e-01 -9.24746931e-01
1.94456309e-01 -4.43042368e-01 -4.55435961e-01 6.96480513e-01
-1.23351542e-02 -4.22966033e-02 1.21833228e-01 1.24712422e-01
-2.34724849e-01 9.57361311e-02 7.22391129e-01 7.78725818e-02
-1.81537885e-02 4.32738006e-01 -9.08549905e-01 5.66604495e-01
9.43093121e-01 -6.07240796e-01 -8.52752388e-01 1.43285662e-01
3.28195006e-01 -1.21165298e-01 7.09790170e-01 -8.67298543e-01
2.93812186e-01 -1.30454034e-01 -1.98038239e-02 -1.34831384e-01
-2.44986899e-02 -1.17151082e+00 5.63565075e-01 5.37803054e-01
-3.20439517e-01 -2.03953311e-01 2.18951479e-01 9.78117883e-01
2.66718715e-01 -1.29441127e-01 6.71116173e-01 1.81118816e-01
-2.85135925e-01 7.33788431e-01 1.55130267e-01 1.60290942e-01
1.24908066e+00 -1.70031175e-01 -7.40061700e-01 -4.59608853e-01
-4.74002868e-01 2.23146871e-01 7.57717073e-01 2.20195696e-01
3.76056969e-01 -1.03842187e+00 -4.00893569e-01 5.30277431e-01
1.23303711e-01 -2.63305426e-01 -7.82707259e-02 5.96767962e-01
-3.98090214e-01 4.32952568e-02 -2.39987411e-02 -7.88970664e-02
-1.50658846e+00 9.38039422e-01 5.90653360e-01 -4.66531515e-01
-4.90755200e-01 6.40144408e-01 3.94323409e-01 -5.14506638e-01
3.55791658e-01 1.06798373e-01 9.72496942e-02 -6.26687109e-01
3.82890761e-01 2.53229231e-01 -3.24246466e-01 -5.25948644e-01
-3.37001711e-01 3.70524712e-02 -2.91870028e-01 3.28393668e-01
7.48436987e-01 1.00859493e-01 -2.43634567e-01 1.71044469e-02
1.22415376e+00 2.65879899e-01 -9.76900280e-01 -3.34674329e-01
7.76758417e-02 -6.84591413e-01 -2.76124239e-01 -6.55009151e-01
-1.15293467e+00 5.01571357e-01 2.74721459e-02 6.15534902e-01
1.02330315e+00 -1.65471628e-01 6.43773735e-01 2.16151178e-01
7.37782180e-01 -3.58391732e-01 -4.74838406e-01 1.36123985e-01
2.89546847e-01 -7.30129182e-01 1.00484192e-01 -9.93264496e-01
-2.67826259e-01 6.29286468e-01 7.11351454e-01 -1.29313052e-01
9.38212097e-01 3.18497151e-01 -1.73442349e-01 -2.92964786e-01
-6.41015708e-01 4.58604872e-01 2.66747717e-02 7.75233746e-01
-3.87182146e-01 1.78500608e-01 4.02694643e-02 8.69582355e-01
-2.43847907e-01 -3.93326908e-01 4.75959241e-01 9.09530163e-01
-2.36621555e-02 -1.20270586e+00 -1.68990344e-02 4.23871070e-01
-8.83610070e-01 -2.26272661e-02 -9.63742197e-01 1.09407628e+00
-2.91183323e-01 8.43800783e-01 -3.80590349e-01 -8.03627372e-01
1.38551757e-01 -1.32854775e-01 1.64461195e-01 -4.17483360e-01
-8.08035254e-01 -4.85554367e-01 3.04621071e-01 -5.62472701e-01
8.94134119e-02 -3.46156090e-01 -1.00929022e+00 -8.70244861e-01
-5.69847286e-01 2.09612161e-01 3.35203618e-01 5.90748906e-01
4.58289564e-01 3.91390845e-02 8.70534539e-01 9.40929130e-02
-9.65375304e-01 -5.22624075e-01 -1.04198074e+00 4.70321596e-01
1.76154271e-01 -4.79808539e-01 -9.77370739e-01 -5.74407816e-01] | [5.98232889175415, 7.2423624992370605] |
495e2190-11de-448f-85f2-7623794b91ae | lingglewrite-a-coaching-system-for-essay | null | null | https://aclanthology.org/2020.acl-demos.17 | https://aclanthology.org/2020.acl-demos.17.pdf | LinggleWrite: a Coaching System for Essay Writing | This paper presents LinggleWrite, a writing coach that provides writing suggestions, assesses writing proficiency levels, detects grammatical errors, and offers corrective feedback in response to user{'}s essay. The method involves extracting grammar patterns, training models for automated essay scoring (AES) and grammatical error detection (GED), and finally retrieving plausible corrections from a n-gram search engine. Experiments on public test sets indicate that both AES and GED models achieve state-of-the-art performance. These results show that LinggleWrite is potentially useful in helping learners improve their writing skills. | ['Ching-Yu Yang', 'Jhih-Jie Chen', 'Chung-Ting Tsai', 'Jason S. Chang'] | 2020-07-01 | null | null | null | acl-2020-6 | ['automated-essay-scoring', 'grammatical-error-detection'] | ['natural-language-processing', 'natural-language-processing'] | [-9.78869870e-02 -4.43506353e-02 -1.76353812e-01 -3.99952143e-01
-8.92696559e-01 -7.33829558e-01 2.12168574e-01 7.69775152e-01
-4.28972781e-01 8.50132465e-01 2.93561459e-01 -7.86971629e-01
-1.85551509e-01 -5.29073000e-01 -5.26661053e-02 3.89547944e-01
9.53832090e-01 4.49957758e-01 2.26621434e-01 -6.10338509e-01
1.41283631e+00 3.31556618e-01 -1.36603129e+00 2.65109748e-01
1.72002900e+00 2.01612085e-01 3.35566580e-01 1.39336717e+00
-4.12923306e-01 1.19227517e+00 -1.02174973e+00 -1.01339865e+00
-4.77185994e-01 -7.62078583e-01 -7.59888411e-01 -5.45058012e-01
1.00783181e+00 -3.31126630e-01 -1.81603566e-01 1.12192166e+00
4.28892821e-01 3.10574740e-01 7.55204856e-01 -4.43606824e-01
-1.23840773e+00 5.72868526e-01 1.94195792e-01 6.85950696e-01
9.16741967e-01 1.03271380e-01 8.63451660e-01 -1.25816941e+00
2.99431503e-01 6.59533799e-01 9.20614004e-01 1.12321341e+00
-8.11853290e-01 -4.79490697e-01 -3.09482574e-01 9.21900943e-02
-7.38049030e-01 -4.68293458e-01 3.17481309e-01 -6.10983372e-01
1.57024753e+00 5.83670020e-01 4.24890816e-01 8.08017612e-01
6.22568965e-01 8.11449409e-01 1.01770735e+00 -1.11968505e+00
-2.51559019e-01 7.11394176e-02 9.88771617e-01 1.24047947e+00
1.57105863e-01 -1.21756375e-01 -1.17532790e+00 -2.62499768e-02
-5.43134063e-02 -5.05662858e-01 -2.76515692e-01 9.43691969e-01
-3.65286112e-01 5.62841177e-01 -5.12070775e-01 1.03623532e-02
-4.86922003e-02 -9.56685096e-02 2.26067871e-01 8.11415851e-01
3.12573731e-01 1.04988658e+00 -1.55624330e-01 -1.01777911e+00
-9.70722973e-01 2.70476013e-01 1.08628023e+00 1.00083792e+00
-1.93503965e-02 8.41635019e-02 -6.45246625e-01 1.09096253e+00
5.33587098e-01 4.29666489e-01 1.11903954e+00 -4.54103619e-01
9.07361746e-01 8.44969273e-01 -1.86638609e-01 -7.13496923e-01
-1.58626109e-01 -2.29979098e-01 -6.36145920e-02 -8.00636485e-02
4.33448464e-01 -4.74949367e-02 -3.54166895e-01 1.24397779e+00
-4.80875373e-01 -3.95237774e-01 -1.47306696e-01 2.90577024e-01
1.21634638e+00 3.04770052e-01 3.00585985e-01 -2.71582156e-01
8.53737772e-01 -9.59302962e-01 -9.53507483e-01 -5.08429885e-01
1.16819775e+00 -1.06116986e+00 1.51465619e+00 9.54525292e-01
-1.74011993e+00 -7.53968179e-01 -9.22703564e-01 -6.24523401e-01
3.53489444e-02 6.26621723e-01 1.15031324e-01 1.07506824e+00
-9.95020568e-01 8.02188098e-01 -6.59434855e-01 -2.03003228e-01
1.34358108e-01 2.06857864e-02 3.97961810e-02 8.43371674e-02
-7.50259161e-01 1.07738912e+00 -8.59056637e-02 -3.60016376e-01
-8.80668908e-02 -7.26397932e-01 -5.93676865e-01 2.25265652e-01
-3.17831427e-01 -3.15873504e-01 1.88910198e+00 -2.66233534e-01
-1.69579792e+00 8.21221709e-01 -3.57018173e-01 1.13530755e-01
3.97100210e-01 -4.41392183e-01 -4.42175448e-01 -3.80589396e-01
-5.98678924e-02 -1.77504003e-01 3.71914417e-01 -7.78298005e-02
-7.73139834e-01 -5.63074768e-01 -5.76740801e-01 4.05040354e-01
-8.40465248e-01 5.28666317e-01 -1.76411033e-01 -8.25058639e-01
1.17523171e-01 -3.84034365e-01 3.99592757e-01 -5.91481805e-01
-3.38814825e-01 -1.20996583e+00 1.34875759e-01 -1.40995932e+00
2.57883859e+00 -1.57875264e+00 -9.60524306e-02 2.93359995e-01
9.94040072e-02 6.11530960e-01 -1.57422945e-01 5.73555648e-01
3.72423261e-01 3.79789650e-01 2.22499877e-01 -3.53344738e-01
5.77248707e-02 -3.15178633e-01 -2.28297189e-01 -3.12391967e-01
-1.59000725e-01 1.03480625e+00 -1.01393855e+00 -3.28245491e-01
-2.88181603e-01 -3.34327817e-01 -5.43074787e-01 5.26998341e-01
-1.90030739e-01 -2.52556622e-01 -2.90930033e-01 6.89632893e-01
-2.09219709e-01 1.15883715e-01 1.43466562e-01 8.96401405e-01
-2.33172208e-01 1.29110253e+00 -5.56856632e-01 1.14458466e+00
-4.58125830e-01 8.67202759e-01 -7.02065349e-01 -6.66352287e-02
1.56717515e+00 7.71044986e-03 -6.82916701e-01 -9.68187094e-01
-1.35711148e-01 6.54751718e-01 2.12076697e-02 -9.05993342e-01
1.08130801e+00 2.26706073e-01 3.51620317e-02 1.15070820e+00
-6.07044511e-02 -4.62240010e-01 4.82043535e-01 3.53140891e-01
1.41614473e+00 -8.67166594e-02 5.43223262e-01 -2.59739578e-01
6.63172901e-01 -2.13290542e-01 -1.23088226e-01 1.12677515e+00
-1.80925895e-02 1.91993326e-01 2.19194159e-01 -6.75810575e-02
-7.38320768e-01 -8.42998803e-01 4.15990353e-01 1.41425169e+00
-5.44978142e-01 -8.59710455e-01 -8.08788300e-01 -9.89255726e-01
1.33812845e-01 1.46777892e+00 -1.16819330e-01 -5.65116286e-01
-6.86890304e-01 -1.70028105e-01 8.10193837e-01 9.12526667e-01
1.29478902e-01 -1.36595070e+00 -4.17692780e-01 4.88585055e-01
-1.72598988e-01 -2.03552276e-01 -9.87685263e-01 2.67155886e-01
-1.05515540e+00 -1.03531086e+00 6.35425299e-02 -9.52733696e-01
7.41139889e-01 4.51670922e-02 1.47074866e+00 9.98508573e-01
-9.53433365e-02 3.83717805e-01 -3.76553863e-01 -6.36591792e-01
-9.01721776e-01 3.14236879e-01 7.87122101e-02 -1.23407722e+00
1.14995635e+00 -1.23515099e-01 6.86592683e-02 -2.59023085e-02
-9.94236320e-02 -2.16044888e-01 1.92525595e-01 8.86842608e-01
2.07304880e-01 -5.14942646e-01 8.33226740e-01 -1.11977088e+00
1.97028756e+00 -9.99276191e-02 -3.30991060e-01 8.17015052e-01
-1.49282503e+00 3.03781498e-02 8.66551042e-01 -3.89182195e-02
-1.15756786e+00 -4.81158674e-01 -4.75253403e-01 4.17726547e-01
5.82939424e-02 7.08373189e-01 4.64733392e-01 -3.04179490e-01
1.27138126e+00 5.23788750e-01 -1.09290011e-01 -5.98642051e-01
-3.11295033e-01 1.19460714e+00 8.28200817e-01 -6.58360064e-01
2.31165200e-01 -9.57018852e-01 -4.15180862e-01 -4.89281744e-01
-1.25341153e+00 -5.10815084e-01 -6.54738009e-01 -4.08426970e-01
7.08531365e-02 -3.42352837e-01 -7.83083498e-01 4.22878772e-01
-1.01019228e+00 -5.23703754e-01 -1.49034232e-01 3.81144971e-01
-3.63127768e-01 4.83086407e-01 -1.02058136e+00 -8.50701571e-01
-6.61450267e-01 -7.06495762e-01 2.87469715e-01 7.81434298e-01
-1.01636636e+00 -1.06803155e+00 4.39869434e-01 7.82862604e-01
9.88195688e-02 -9.68071580e-01 1.19826174e+00 -9.57491577e-01
2.99743265e-02 -1.93348885e-01 9.52960104e-02 5.02140880e-01
-1.56947434e-01 1.26111403e-01 -5.21872997e-01 8.69732723e-02
-1.52247861e-01 -7.90075004e-01 5.61460376e-01 -8.66747424e-02
1.78964436e+00 -5.76662123e-01 3.52387577e-01 5.30596197e-01
9.25093412e-01 2.03942284e-01 5.16251266e-01 2.87984371e-01
5.34437895e-01 4.57985520e-01 6.70015097e-01 3.46395582e-01
5.20301223e-01 3.92252117e-01 -2.04194531e-01 1.04474914e+00
-3.25233430e-01 -7.62028575e-01 6.77344084e-01 1.49483514e+00
-6.52549462e-03 -7.09332585e-01 -1.14821839e+00 4.46208090e-01
-1.43788409e+00 -9.63289618e-01 -7.31803656e-01 1.95971525e+00
1.26405621e+00 7.70247057e-02 -5.20283170e-02 1.96726874e-01
1.77223846e-01 -3.03607672e-01 -3.11731875e-01 -1.50857496e+00
1.46716624e-01 1.01245975e+00 2.77705759e-01 8.97782385e-01
-2.79998392e-01 1.00631106e+00 6.89251089e+00 5.88331938e-01
-6.85811162e-01 -3.25441845e-02 3.83027881e-01 7.16166571e-02
-5.11705995e-01 -5.50159514e-01 -1.40548706e+00 5.62204421e-01
1.21547520e+00 -5.44297576e-01 4.20210809e-01 6.78034902e-01
1.55654177e-01 -7.76905119e-02 -9.80883718e-01 7.40354419e-01
6.58597529e-01 -1.19684231e+00 7.89701715e-02 -4.42239106e-01
7.84547091e-01 -2.76654541e-01 -3.03594880e-02 6.87782407e-01
5.75093329e-01 -1.29933596e+00 7.40477681e-01 1.12021601e+00
8.52771282e-01 -8.38415563e-01 5.04019797e-01 7.85419405e-01
-3.64236414e-01 -1.35772064e-01 -5.41277230e-01 -5.70273459e-01
-4.65890855e-01 -8.09840742e-04 -1.18627059e+00 -9.87111852e-02
2.51294374e-01 5.09539843e-01 -1.24629188e+00 1.00384927e+00
-1.07517803e+00 1.27270532e+00 1.29791632e-01 -1.00569320e+00
-2.14935347e-01 2.73202509e-02 3.28812510e-01 1.51070750e+00
6.87673151e-01 6.75368607e-01 -1.70180589e-01 7.51599967e-01
-4.02579218e-01 4.15011764e-01 -1.24254890e-01 -3.19728911e-01
1.01465273e+00 7.52971053e-01 -1.18186794e-01 -1.48930043e-01
-1.33100644e-01 1.08306217e+00 6.97015166e-01 -1.65249974e-01
2.52390709e-02 -8.55150342e-01 2.62988687e-01 -3.85985374e-02
-2.93546706e-01 -1.35935202e-01 -1.16133738e+00 -8.83249223e-01
2.02720344e-01 -1.17632163e+00 5.04475832e-01 -8.69648635e-01
-1.34879732e+00 1.01960927e-01 -7.72647738e-01 -6.93510711e-01
-4.74559069e-01 -8.42102051e-01 -1.13772488e+00 1.48695779e+00
-1.10014856e+00 -2.79545993e-01 -5.88240504e-01 1.47576869e-01
1.12090659e+00 -7.39728451e-01 9.13925052e-01 -9.01210457e-02
-6.31890535e-01 1.47623122e+00 1.59087420e-01 -9.32020545e-02
8.15658987e-01 -1.65696859e+00 6.42532587e-01 1.07921159e+00
2.02801853e-01 8.34051967e-01 4.56019431e-01 -1.10053647e+00
-1.11268592e+00 -8.21636677e-01 2.10017967e+00 -1.06918073e+00
5.34231484e-01 2.40236744e-01 -1.11795712e+00 5.14772534e-01
-8.65999162e-02 -7.25358009e-01 1.09228051e+00 2.46296510e-01
-1.59053981e-01 4.63484854e-01 -9.01622295e-01 5.85753977e-01
1.02747869e+00 -8.84641707e-01 -1.17030668e+00 3.37684482e-01
-4.97775488e-02 -6.98220491e-01 -5.89053214e-01 -1.58358797e-01
6.23171270e-01 -8.58036280e-01 3.22789013e-01 -1.07679129e+00
1.04072821e+00 3.23186219e-01 3.88824582e-01 -1.73824096e+00
-5.49141347e-01 -6.80407107e-01 -6.08221769e-01 1.07101262e+00
5.39236963e-01 -1.09274112e-01 8.19068789e-01 9.42366123e-01
-7.49642372e-01 -9.22710478e-01 -4.66120958e-01 -7.06461906e-01
2.56731570e-01 -4.75685924e-01 4.53088462e-01 9.21975851e-01
8.34868371e-01 2.78159589e-01 2.53092706e-01 -3.28324080e-01
6.33152798e-02 -4.79690492e-01 5.22773981e-01 -1.48820138e+00
-1.86842158e-01 -9.39205468e-01 7.82070234e-02 -1.25770593e+00
3.60519618e-01 -1.21450973e+00 1.48726285e-01 -1.41112316e+00
9.96711180e-02 -3.31251830e-01 4.73321564e-02 6.60029113e-01
-7.30261862e-01 1.11585714e-01 1.41368201e-02 5.51114343e-02
-3.83346111e-01 2.90696651e-01 9.08087134e-01 1.29161417e-01
-3.59786719e-01 3.15427244e-01 -1.08058846e+00 6.54205501e-01
9.02475536e-01 -4.63736981e-01 -3.45655918e-01 -4.56481248e-01
7.23978460e-01 2.04855680e-01 -1.37517303e-01 -7.96947658e-01
7.45168865e-01 -4.03797746e-01 3.83882552e-01 -4.62714881e-01
-6.85095906e-01 1.61691099e-01 -7.25518703e-01 4.50709522e-01
-9.17968571e-01 8.56513023e-01 3.53988856e-01 -4.08305265e-02
2.30504069e-02 -1.33346617e+00 4.90376502e-01 1.37133067e-02
-4.38017309e-01 -2.78222740e-01 -6.76770329e-01 2.18975857e-01
3.01489592e-01 -3.99250805e-01 -7.79217243e-01 -3.69108498e-01
-3.24531347e-01 2.70344764e-01 1.93489179e-01 4.88957107e-01
1.03781462e+00 -1.05612946e+00 -9.59437490e-01 4.39780682e-01
3.42367619e-01 -5.16613424e-01 -2.47779310e-01 5.70142210e-01
-9.18118596e-01 6.52050853e-01 -2.12105632e-01 1.16706289e-01
-2.00513339e+00 -5.43829143e-01 2.60413647e-01 -4.46438879e-01
-2.86282331e-01 1.65809727e+00 -8.14011157e-01 -6.53536141e-01
3.53510290e-01 -6.65871352e-02 -5.87620616e-01 -1.52137339e-01
1.04572713e+00 8.50044429e-01 7.19723225e-01 5.88446781e-02
2.33212590e-01 -9.08126906e-02 -3.88338447e-01 1.68194119e-02
9.72072661e-01 -1.17440097e-01 -1.11859821e-01 5.08104026e-01
2.23329365e-01 6.85527623e-01 -4.57942754e-01 1.00105122e-01
5.01276314e-01 -6.58595741e-01 3.90960276e-02 -1.65617990e+00
-2.12955162e-01 9.29144740e-01 7.20664337e-02 -8.60564262e-02
7.85006702e-01 -6.00868523e-01 6.98517859e-01 1.15734541e+00
1.27483746e-02 -1.67211103e+00 3.25566590e-01 1.37123191e+00
8.16499352e-01 -9.13746357e-01 -1.07634932e-01 -5.05355299e-02
-3.72240543e-01 1.86855626e+00 1.32183123e+00 1.66240692e-01
5.78897372e-02 -1.46790138e-02 3.05529777e-02 -9.59171504e-02
-1.04119468e+00 5.39525032e-01 7.94233680e-01 2.27853522e-01
1.27255213e+00 2.75559835e-02 -9.75998878e-01 1.01032805e+00
-8.51304412e-01 4.73135784e-02 1.06671882e+00 7.31287122e-01
-9.44657385e-01 -1.23377681e+00 -1.86298698e-01 1.16918993e+00
-3.95394981e-01 -7.14819849e-01 -1.06949925e+00 3.03607732e-01
-3.45181018e-01 8.85990322e-01 -1.63834900e-01 -3.77030283e-01
4.98986274e-01 8.64985168e-01 8.11392188e-01 -1.30840433e+00
-1.56337559e+00 -8.10522556e-01 3.62791687e-01 -3.14570606e-01
5.60389757e-01 -8.22516561e-01 -1.18438864e+00 -5.04302144e-01
-5.33580244e-01 4.11841124e-01 4.81470466e-01 9.98761594e-01
8.57390165e-02 5.28027892e-01 2.05973357e-01 3.61529589e-01
-1.34619236e+00 -1.43285584e+00 -2.97146142e-01 4.31313552e-02
-1.20512009e-01 -4.38479111e-02 -2.30901688e-01 -1.39361128e-01] | [11.326905250549316, 9.303570747375488] |
e3860700-af9b-476c-a5d0-b125b515dd8a | fast-improving-controllability-for-text | 2210.03167 | null | https://arxiv.org/abs/2210.03167v1 | https://arxiv.org/pdf/2210.03167v1.pdf | FAST: Improving Controllability for Text Generation with Feedback Aware Self-Training | Controllable text generation systems often leverage control codes to direct various properties of the output like style and length. Inspired by recent work on causal inference for NLP, this paper reveals a previously overlooked flaw in these control code-based conditional text generation algorithms. Spurious correlations in the training data can lead models to incorrectly rely on parts of the input other than the control code for attribute selection, significantly undermining downstream generation quality and controllability. We demonstrate the severity of this issue with a series of case studies and then propose two simple techniques to reduce these correlations in training sets. The first technique is based on resampling the data according to an example's propensity towards each linguistic attribute (IPS). The second produces multiple counterfactual versions of each example and then uses an additional feedback mechanism to remove noisy examples (feedback aware self-training, FAST). We evaluate on 3 tasks -- news headline, meta review, and search ads generation -- and demonstrate that FAST can significantly improve the controllability and language quality of generated outputs when compared to state-of-the-art controllable text generation approaches. | ['Yi Liu', 'Chenguang Zhu', 'Konstantin Golobokov', 'Victor Ye Dong', 'Reid Pryzant', 'Junyi Chai'] | 2022-10-06 | null | null | null | null | ['conditional-text-generation'] | ['natural-language-processing'] | [ 6.68147266e-01 7.82255292e-01 -4.39292997e-01 -3.63722175e-01
-1.02579212e+00 -8.92391026e-01 1.17654598e+00 2.32496083e-01
-1.08481839e-01 1.25024009e+00 8.26012015e-01 -5.64296842e-01
-1.56386092e-01 -8.99725556e-01 -1.01882768e+00 -4.23250616e-01
2.99794286e-01 6.00551426e-01 -3.25214535e-01 -3.11268121e-01
4.23173249e-01 -7.21073225e-02 -1.63483024e+00 5.92480123e-01
1.21381509e+00 4.84739125e-01 6.67267367e-02 7.47477055e-01
-1.62133977e-01 9.10772502e-01 -8.38557661e-01 -6.52574599e-01
1.49436712e-01 -7.61120915e-01 -4.92785662e-01 -2.29302824e-01
4.51410472e-01 -1.34542599e-01 1.13193892e-01 9.63934124e-01
5.36824822e-01 -8.39951634e-03 1.06804693e+00 -1.27090085e+00
-9.51110840e-01 1.57113993e+00 -3.03511992e-02 -1.62526876e-01
4.32753593e-01 2.99226314e-01 1.44305527e+00 -5.29321492e-01
8.78193557e-01 1.53486097e+00 4.84615684e-01 7.63580263e-01
-1.64707470e+00 -8.55432749e-01 2.09735036e-01 -4.76271749e-01
-8.35855961e-01 -6.36002958e-01 5.73688090e-01 -5.47010660e-01
8.33723485e-01 3.49871635e-01 5.62783718e-01 1.77270234e+00
2.43220448e-01 7.36665308e-01 1.25268400e+00 -6.58601642e-01
3.73008013e-01 4.82867539e-01 -2.66678065e-01 5.24817228e-01
4.52650458e-01 5.98701894e-01 -7.59256721e-01 -3.68678182e-01
3.48044455e-01 -6.87810779e-01 -2.25040391e-01 2.54434105e-02
-1.34413970e+00 1.15287936e+00 1.35087922e-01 -9.11019370e-03
-2.24046513e-01 3.69460225e-01 2.18550369e-01 3.22831988e-01
5.91801405e-01 1.23429751e+00 -7.35893786e-01 -2.63828576e-01
-8.58946681e-01 9.18748796e-01 1.02801633e+00 1.24435520e+00
2.95293212e-01 2.02500261e-02 -9.27350998e-01 6.55122817e-01
1.40893191e-01 6.57947302e-01 6.45265341e-01 -6.94810033e-01
7.35782385e-01 4.91476804e-01 1.61966816e-01 -6.37516856e-01
-3.13882753e-02 -4.59700316e-01 -3.84906799e-01 -1.50482655e-01
4.25480038e-01 -5.38169920e-01 -1.05669832e+00 1.98446465e+00
5.28167747e-02 -5.07915020e-01 -2.63414979e-02 5.86017489e-01
3.45120996e-01 4.66311753e-01 3.36578161e-01 -3.65248919e-01
8.70609820e-01 -6.44901216e-01 -7.32283771e-01 -1.81657389e-01
7.70253301e-01 -7.13058889e-01 1.37087083e+00 3.51842016e-01
-9.69229043e-01 -2.50738502e-01 -1.08418667e+00 1.44764647e-01
-3.96379828e-01 7.35705271e-02 8.60100985e-01 8.32287014e-01
-8.11956704e-01 1.01386070e+00 -1.73845306e-01 4.05724254e-03
3.85631382e-01 1.29693955e-01 1.45311847e-01 2.64679104e-01
-1.53064704e+00 6.47515655e-01 2.42425844e-01 -4.10383791e-01
-7.81113505e-01 -1.13717496e+00 -5.63365161e-01 9.27241594e-02
4.46033448e-01 -1.02496171e+00 1.58395445e+00 -1.00996709e+00
-1.62694025e+00 2.72607535e-01 -5.61731122e-02 -3.82915288e-01
8.88252676e-01 -3.79178524e-01 -1.61267713e-01 -4.80850756e-01
4.87672776e-01 7.24050999e-01 1.05673671e+00 -1.31699014e+00
-6.41706526e-01 -5.24329357e-02 -2.31061161e-01 2.01407045e-01
-2.05980361e-01 -3.52870911e-01 6.28252849e-02 -9.10986662e-01
-4.79677886e-01 -1.00516880e+00 -3.25241417e-01 -5.82039595e-01
-1.11140347e+00 -2.74539202e-01 2.26468638e-01 -1.95037171e-01
1.42129016e+00 -1.61770105e+00 1.78630859e-01 2.33673885e-01
-1.63803399e-02 -1.60592318e-01 -1.07629448e-01 3.46839249e-01
-1.39970705e-01 8.76154780e-01 -3.42720263e-02 -2.79173404e-01
1.13208137e-01 -1.75931975e-01 -8.19366395e-01 -2.05779642e-01
4.26509738e-01 9.03844178e-01 -1.17392790e+00 -3.54715884e-01
-1.51083618e-01 -1.82112288e-02 -9.40686285e-01 1.86974466e-01
-9.76155400e-01 2.10766926e-01 -4.13330138e-01 3.56373101e-01
-1.91168825e-03 -1.05862981e-02 5.37740439e-02 2.01699942e-01
-1.85406879e-01 8.19831312e-01 -8.53261292e-01 1.33938980e+00
-6.81807816e-01 5.02432466e-01 -5.84183514e-01 -3.20792973e-01
7.11172581e-01 2.44669393e-01 -1.30478367e-02 -3.79104406e-01
2.58020870e-02 2.41848439e-01 7.71995485e-02 -4.13951963e-01
7.12108374e-01 -2.49987200e-01 -2.73568422e-01 7.29248345e-01
-2.59074960e-02 -5.80954075e-01 3.35811526e-01 5.21578550e-01
1.05548823e+00 3.15744162e-01 1.32309169e-01 -4.53586966e-01
-7.97914565e-02 1.66600391e-01 4.03026313e-01 1.30353856e+00
3.70467305e-01 5.93114078e-01 1.03435826e+00 1.37781799e-01
-1.33933079e+00 -1.03707731e+00 3.03207505e-02 1.04625189e+00
-5.05736470e-01 -6.10492468e-01 -7.91445374e-01 -9.62722242e-01
3.93874109e-01 1.73658788e+00 -1.08976305e+00 -4.02969331e-01
-2.05535695e-01 -7.22904742e-01 6.16732359e-01 4.02177960e-01
6.48180246e-02 -1.10900235e+00 -3.09553266e-01 2.66527057e-01
-2.76955992e-01 -6.33103609e-01 -6.38236761e-01 2.27766052e-01
-7.77137399e-01 -6.42322361e-01 -2.38960698e-01 -1.04634076e-01
5.97677886e-01 -5.06944537e-01 1.35749245e+00 -7.45525286e-02
-6.36087060e-02 -8.33350345e-02 -3.13454449e-01 -9.16088820e-01
-1.03674304e+00 3.84228349e-01 -6.36759354e-03 -2.61536092e-01
-1.96731556e-02 -3.80069196e-01 -3.53283525e-01 -1.27521947e-01
-6.23072445e-01 2.65659869e-01 7.41636693e-01 1.08181858e+00
2.92206645e-01 -2.29471609e-01 9.03162539e-01 -1.50836480e+00
1.36023772e+00 -6.42687440e-01 -4.43948388e-01 2.18126088e-01
-1.28582454e+00 6.15254104e-01 7.46606946e-01 -6.78811133e-01
-1.25030267e+00 8.90491251e-03 4.30222303e-01 -1.02193676e-01
3.28559689e-02 5.34315944e-01 -8.14803615e-02 8.79578471e-01
1.15263307e+00 4.70263064e-02 -7.34262988e-02 -1.42782152e-01
6.99961543e-01 5.58282077e-01 2.11669654e-01 -7.52515972e-01
6.98408723e-01 -3.74497734e-02 -1.10178091e-01 -3.79081249e-01
-1.11584127e+00 2.76483744e-01 -1.83867320e-01 -1.24516204e-01
6.82594955e-01 -7.15007782e-01 -4.50841933e-01 3.11001819e-02
-1.10996938e+00 -6.03879452e-01 -5.23989975e-01 1.06030911e-01
-5.79407632e-01 -4.93180037e-01 -2.96483099e-01 -1.04965293e+00
-4.09441561e-01 -1.00195122e+00 1.17350519e+00 4.08137105e-02
-7.52796233e-01 -8.70526016e-01 2.51718201e-02 2.14268360e-02
5.09344995e-01 2.04769969e-01 1.34900916e+00 -8.32677305e-01
-4.69758421e-01 -1.95421576e-01 1.15725890e-01 8.26996937e-02
7.33263195e-02 3.16964090e-01 -1.02886724e+00 2.05399498e-01
-3.03659290e-01 -4.02287722e-01 8.93024981e-01 4.19075727e-01
9.20366049e-01 -7.97388613e-01 -5.65160334e-01 1.22819379e-01
1.08700681e+00 2.92279646e-02 4.00473863e-01 5.93435615e-02
5.63323140e-01 7.66726077e-01 6.12295270e-01 4.82403338e-01
1.16488561e-01 5.65636516e-01 5.83932474e-02 1.84342086e-01
-1.43581882e-01 -1.13004959e+00 4.86112416e-01 2.22781986e-01
1.11771822e-01 -5.03934383e-01 -4.22524363e-01 6.15581036e-01
-1.73235238e+00 -1.06153011e+00 -2.01756671e-01 2.26539731e+00
1.47327590e+00 4.04176116e-01 3.65313217e-02 3.35396677e-02
5.69355190e-01 -7.18157440e-02 -4.87998962e-01 -5.52770793e-01
-1.14090189e-01 1.60882622e-01 6.31057560e-01 6.72147274e-01
-6.94894969e-01 1.01695180e+00 6.48899555e+00 7.69492328e-01
-7.30976343e-01 -1.72675744e-01 8.80487680e-01 -3.00271481e-01
-1.11645222e+00 1.50581315e-01 -9.91118550e-01 6.52852535e-01
1.06898248e+00 -6.22833252e-01 5.36540687e-01 8.04188550e-01
4.55221623e-01 -7.50865191e-02 -1.64646113e+00 2.19392508e-01
-1.62291571e-01 -1.51034093e+00 5.27241290e-01 7.58737624e-02
1.02350283e+00 -5.47694683e-01 1.96696624e-01 4.32960838e-01
8.82393181e-01 -1.24675989e+00 1.21770120e+00 5.35859823e-01
1.02615118e+00 -7.33746469e-01 4.01087195e-01 2.70226389e-01
-3.61907214e-01 -3.34188551e-01 -7.80221671e-02 -4.59572375e-02
9.78500992e-02 9.12373483e-01 -1.27664316e+00 1.65234178e-01
2.86620289e-01 3.19129169e-01 -6.24035180e-01 3.70665044e-01
-6.96239591e-01 9.54638124e-01 -4.59522195e-02 -5.89207828e-01
-1.20768115e-01 1.76288232e-01 7.45872736e-01 1.26975811e+00
2.82583892e-01 -5.47001064e-02 -3.40791255e-01 1.61154926e+00
-3.07191730e-01 -9.07155350e-02 -9.17396784e-01 -4.05772269e-01
7.35342622e-01 9.38904464e-01 -2.31296167e-01 -4.99026597e-01
1.76892295e-01 7.17752099e-01 2.04578251e-01 2.81681210e-01
-6.31956935e-01 -4.49077338e-01 4.82114226e-01 2.38434985e-01
1.17351703e-01 2.59720802e-01 -8.81926656e-01 -9.05633509e-01
-2.78644592e-01 -1.08314109e+00 1.56650960e-01 -8.36015821e-01
-1.25125957e+00 2.80444324e-01 7.74047375e-02 -8.18442464e-01
-7.75950909e-01 -2.19598785e-01 -4.92363155e-01 1.07804513e+00
-1.00996900e+00 -8.69381845e-01 1.38905033e-01 1.41958341e-01
6.99907660e-01 -2.10026741e-01 8.44922423e-01 -2.55364925e-01
-2.63480961e-01 8.67753625e-01 -2.30565488e-01 -2.08973557e-01
9.30695415e-01 -1.67686582e+00 6.05730176e-01 6.48508370e-01
1.31426388e-02 9.93672192e-01 1.21328270e+00 -1.10328102e+00
-1.19145346e+00 -1.09471846e+00 1.40853262e+00 -9.32561219e-01
7.04010963e-01 -6.47854388e-01 -3.06727946e-01 4.83197957e-01
2.28137553e-01 -7.97751963e-01 4.60456789e-01 4.05033708e-01
-4.49779600e-01 1.44578263e-01 -1.04677963e+00 1.15425253e+00
1.19158196e+00 -2.19130978e-01 -4.43008333e-01 4.68613625e-01
1.28663290e+00 -2.93548137e-01 -5.38825572e-01 1.55370981e-01
6.10810578e-01 -7.77847588e-01 4.47769672e-01 -8.89006734e-01
1.30763793e+00 -1.35474512e-02 -1.40407244e-02 -2.03183317e+00
-2.25093752e-01 -1.07020462e+00 1.50605619e-01 1.46495485e+00
1.24507809e+00 -4.49986130e-01 5.24847269e-01 8.35992813e-01
-1.40696570e-01 -6.35359883e-01 -4.02234197e-01 -5.11837304e-01
4.08411413e-01 -4.37651813e-01 7.33908415e-01 7.43866742e-01
2.57226914e-01 8.49959850e-01 -2.26237953e-01 -2.39682525e-01
5.92315793e-01 2.66035087e-02 7.32049942e-01 -1.05769849e+00
-4.30097848e-01 -5.12449026e-01 3.37943494e-01 -7.03143597e-01
8.66333693e-02 -9.35611486e-01 2.08356470e-01 -1.20679688e+00
1.31224573e-01 -6.20888650e-01 3.54854435e-01 4.59247559e-01
-5.03130257e-01 -3.52096349e-01 2.33849049e-01 -5.43548241e-02
2.39265785e-02 4.99100268e-01 1.32098925e+00 -5.00256307e-02
-4.66362745e-01 1.23080932e-01 -1.32069242e+00 2.61909574e-01
7.31581092e-01 -5.86583436e-01 -6.73298717e-01 -1.62233204e-01
6.80107236e-01 1.04300968e-01 2.50758559e-01 -4.67498243e-01
-9.19001251e-02 -3.64980102e-01 3.96233976e-01 -1.05359510e-01
-2.01488256e-01 -3.29311311e-01 -3.24483179e-02 3.64144087e-01
-1.44318366e+00 1.62256747e-01 1.06840558e-01 7.68062413e-01
2.76328921e-01 -1.67513311e-01 3.23795408e-01 -1.98735654e-01
2.22982466e-01 -1.60043105e-01 -5.19475222e-01 3.43354225e-01
5.73828220e-01 4.26437885e-01 -5.80365300e-01 -5.28211534e-01
-3.62372190e-01 8.43793750e-02 2.34175280e-01 7.06314683e-01
3.80137146e-01 -1.18277478e+00 -9.21509743e-01 1.79342434e-01
1.07338123e-01 -2.87147254e-01 -3.18938881e-01 4.30215746e-01
1.84221536e-01 5.04805803e-01 3.44996452e-01 -2.48119146e-01
-7.82188594e-01 4.46552366e-01 1.64474934e-01 -3.34139287e-01
-1.27853960e-01 9.22107756e-01 -9.58537832e-02 -2.55823374e-01
3.64230797e-02 -4.28906500e-01 9.97290835e-02 1.92357868e-01
1.84348136e-01 9.26524475e-02 -4.46157716e-03 1.86201289e-01
1.00996234e-01 -2.39943191e-01 -3.01846415e-01 -8.32708955e-01
1.02943230e+00 2.32377633e-01 1.45075396e-01 6.86149180e-01
7.98130453e-01 3.01052868e-01 -1.15544939e+00 2.05795288e-01
6.06298484e-02 -4.91326571e-01 6.54203491e-03 -1.47794402e+00
-5.25298059e-01 5.73590934e-01 4.32571135e-02 4.31709558e-01
5.74683905e-01 -4.68067564e-02 3.18664730e-01 1.34181425e-01
1.49852023e-01 -1.30521607e+00 9.61125642e-02 2.53337711e-01
1.24906623e+00 -9.89994884e-01 6.18458018e-02 -3.40750098e-01
-1.03315806e+00 7.80693352e-01 5.61541855e-01 1.66923523e-01
3.14244419e-01 3.79372060e-01 -1.66890264e-01 4.29581217e-02
-1.48397672e+00 1.80883020e-01 1.99152142e-01 4.55335826e-01
8.65923762e-01 2.73451328e-01 -4.98302758e-01 7.36029387e-01
-8.70748043e-01 -1.41964450e-01 7.14225709e-01 5.19350827e-01
-9.93551910e-02 -1.10538256e+00 -1.12710200e-01 1.05878401e+00
-4.82054323e-01 -6.09256327e-01 -8.80986094e-01 6.68295562e-01
1.62623581e-02 1.18330336e+00 -7.10532665e-02 -6.08136654e-01
3.65779907e-01 3.93413872e-01 3.74826133e-01 -7.98180759e-01
-8.21402252e-01 -5.93875833e-02 6.08703971e-01 -3.78469557e-01
7.54567459e-02 -1.08773482e+00 -9.36718166e-01 -2.83723325e-01
-5.39860010e-01 1.89036310e-01 7.50160813e-01 7.94169068e-01
4.86561090e-01 6.57634556e-01 6.10383153e-01 -5.53182483e-01
-1.13360894e+00 -1.26992261e+00 -2.49700189e-01 5.81802011e-01
2.71769077e-01 -6.25021756e-01 -8.19682717e-01 2.51670301e-01] | [11.586310386657715, 8.893105506896973] |
95e95cc9-4fa6-4137-8382-095ff68dee49 | towards-efficient-modularity-in-industrial | 2210.01971 | null | https://arxiv.org/abs/2210.01971v3 | https://arxiv.org/pdf/2210.01971v3.pdf | Towards Efficient Modularity in Industrial Drying: A Combinatorial Optimization Viewpoint | The industrial drying process consumes approximately 12% of the total energy used in manufacturing, with the potential for a 40% reduction in energy usage through improved process controls and the development of new drying technologies. To achieve cost-efficient and high-performing drying, multiple drying technologies can be combined in a modular fashion with optimal sequencing and control parameters for each. This paper presents a mathematical formulation of this optimization problem and proposes a framework based on the Maximum Entropy Principle (MEP) to simultaneously solve for both optimal values of control parameters and optimal sequence. The proposed algorithm addresses the combinatorial optimization problem with a non-convex cost function riddled with multiple poor local minima. Simulation results on drying distillers dried grain (DDG) products show up to 12% improvement in energy consumption compared to the most efficient single-stage drying process. The proposed algorithm converges to local minima and is designed heuristically to reach the global minimum. | ['Srinivasa Salapaka', 'Hao Feng', 'Amir Malvandi', 'Amber Srivastava', 'Alisina Bayati'] | 2022-10-05 | null | null | null | null | ['total-energy'] | ['miscellaneous'] | [ 3.75156790e-01 8.81975666e-02 -1.85726330e-01 -2.46554658e-01
-2.77177900e-01 -5.73869944e-01 1.69701148e-02 9.14838374e-01
-1.21002614e-01 7.27129757e-01 -3.65245074e-01 -2.51479328e-01
-6.12033248e-01 -9.73304272e-01 -2.95547307e-01 -7.84413218e-01
-7.81169608e-02 4.65159088e-01 -5.00523865e-01 -1.58728063e-01
3.95341009e-01 5.46634912e-01 -1.31573749e+00 -4.43560630e-01
1.37875760e+00 8.47214103e-01 9.80548799e-01 6.49992049e-01
-5.64468019e-02 -1.27969384e-01 -3.47836465e-01 7.99417198e-02
4.28105652e-01 -5.48485935e-01 -4.66446429e-01 2.68250138e-01
-1.57002971e-01 -1.78558435e-02 7.37509131e-01 1.07893980e+00
4.13719088e-01 5.26147068e-01 4.86079514e-01 -7.32717097e-01
-4.58756268e-01 1.95044532e-01 -7.58104742e-01 -1.78327277e-01
-2.33226232e-02 -8.81768614e-02 8.66495073e-01 -3.94604474e-01
2.61594981e-01 8.75490427e-01 6.20935783e-02 2.49538168e-01
-1.27227068e+00 -1.80366710e-01 -5.74287400e-02 -1.59952223e-01
-1.06146610e+00 -2.86400057e-02 3.95362526e-01 -1.88613974e-03
1.09032428e+00 4.16181564e-01 1.08832073e+00 -1.36634469e-01
7.01013625e-01 2.03894779e-01 1.20806754e+00 -4.82459188e-01
6.26448810e-01 -2.72298098e-01 6.84973225e-03 6.42892599e-01
8.79969239e-01 1.12399235e-01 1.73613489e-01 2.39466816e-01
5.14402390e-01 -5.90432324e-02 -1.72341496e-01 -5.89020312e-01
-6.05961621e-01 9.69989598e-01 4.02481318e-01 1.46232218e-01
-7.62359202e-01 -2.08511516e-01 7.89634734e-02 1.78241640e-01
4.63830203e-01 7.91185141e-01 -7.71233737e-01 1.16242647e-01
-8.82674515e-01 1.97026044e-01 1.03112638e+00 7.75160611e-01
3.89810234e-01 6.48855567e-02 2.54102319e-01 7.27937758e-01
4.70531136e-01 6.49800360e-01 -5.35210334e-02 -9.42768514e-01
4.23290968e-01 6.21361375e-01 5.42130113e-01 -7.75970459e-01
-4.56801713e-01 -2.04070210e-01 -7.93431818e-01 3.02609831e-01
6.03803694e-02 -3.91591281e-01 -9.25131023e-01 1.27070284e+00
5.41947246e-01 -9.45013940e-01 -2.96207052e-02 9.90896463e-01
-4.43408489e-02 1.15369022e+00 2.16498435e-01 -1.07113230e+00
1.17856765e+00 -9.05640960e-01 -1.28879428e+00 -1.68183997e-01
9.86192748e-02 -8.93809140e-01 5.53287983e-01 5.53093851e-01
-1.50116932e+00 -2.41566777e-01 -1.46916831e+00 1.33524567e-01
-5.45232594e-01 1.39007613e-01 5.35790980e-01 6.86536014e-01
-6.71071112e-01 8.10907125e-01 -8.43112707e-01 -4.96238947e-01
-1.90760121e-01 7.05723166e-01 2.28283674e-01 -1.94209799e-01
-4.43647593e-01 1.32549834e+00 5.47092557e-01 6.54524446e-01
-2.83425331e-01 -6.83289289e-01 -8.20648849e-01 1.19850606e-01
3.50475490e-01 -6.67070627e-01 1.05745530e+00 -3.67239416e-01
-1.95366657e+00 4.86333013e-01 -2.91850060e-01 -1.18913651e-01
3.11617941e-01 -4.57479686e-01 -3.05682510e-01 8.67981836e-02
-2.88325071e-01 2.52203494e-01 3.97239774e-01 -1.17471516e+00
-7.52491415e-01 -3.27992439e-01 -1.78927332e-01 1.86670795e-01
1.70486405e-01 -1.31039500e-01 3.75095397e-01 -1.63200215e-01
1.84057876e-01 -7.60827422e-01 -7.10997701e-01 -4.44136173e-01
-1.47544131e-01 -7.88972452e-02 7.07769573e-01 -9.13811862e-01
1.16249776e+00 -1.53024364e+00 6.14661098e-01 4.20014352e-01
-3.55015308e-01 1.48284554e-01 1.21134013e-01 5.78291655e-01
9.53364596e-02 1.07548758e-01 -3.22709590e-01 9.41757336e-02
-3.16445641e-02 5.07413626e-01 4.42905724e-01 6.56453073e-01
3.09613496e-01 3.46302599e-01 -1.16392779e+00 -9.06117931e-02
8.29513550e-01 3.80961627e-01 -3.45812500e-01 2.19090566e-01
-1.49955690e-01 5.67355417e-02 -4.47924316e-01 8.57521057e-01
8.75992477e-01 3.66391391e-01 6.55392468e-01 -4.34897929e-01
-5.98814189e-01 -2.26070255e-01 -1.32377148e+00 1.39279330e+00
-8.00733626e-01 -6.84784427e-02 8.88615429e-01 -1.08681107e+00
1.28182805e+00 1.68636799e-01 6.42587662e-01 -6.60286784e-01
3.58401626e-01 3.84879440e-01 -1.64928079e-01 -3.66048574e-01
7.31819808e-01 -5.49775720e-01 2.10919291e-01 -5.07160909e-02
-2.78408639e-03 -4.97123539e-01 7.00117886e-01 -4.73274797e-01
5.11348069e-01 2.65330821e-01 6.40447021e-01 -9.96259451e-01
5.43056011e-01 1.12340100e-01 7.17609644e-01 -4.02258597e-02
6.82114661e-02 5.80376238e-02 -2.07008645e-01 -1.99989006e-01
-1.25851512e+00 -7.61280537e-01 -1.23706743e-01 3.54437202e-01
5.12769222e-01 -3.31412777e-02 -5.04274964e-01 1.60891652e-01
-1.56871472e-02 1.16840661e+00 -7.03695491e-02 -8.33416954e-02
-6.81937218e-01 -1.09207392e+00 -7.92583644e-01 3.32704261e-02
4.78701711e-01 -5.27640522e-01 -1.19113755e+00 5.90679228e-01
-4.64897789e-02 -6.84249341e-01 -1.29574925e-01 7.95132041e-01
-1.31818140e+00 -1.01416850e+00 -4.15824056e-01 -6.63535893e-01
9.88324285e-01 2.30732396e-01 1.00777614e+00 -8.26866180e-02
-6.98063076e-01 -1.17028028e-01 -3.26623142e-01 -5.51788330e-01
-4.48564112e-01 -2.28391830e-02 -1.04308851e-01 -6.27596378e-01
1.95193335e-01 -1.45955965e-01 -8.26302707e-01 1.00924753e-01
-5.53410590e-01 -5.10352738e-02 4.43357140e-01 5.37992060e-01
7.53586709e-01 8.27458024e-01 4.82991338e-01 -3.84197414e-01
7.43925095e-01 -3.70034844e-01 -1.03282320e+00 4.64372873e-01
-1.10709596e+00 2.81617463e-01 8.69208276e-01 -1.28457695e-01
-1.15577471e+00 1.22162797e-01 4.33538318e-01 1.10361569e-01
1.33998007e-01 2.35198244e-01 -2.01381162e-01 3.03143524e-02
-1.69247851e-01 -2.15382829e-01 1.28935441e-01 -6.66624784e-01
2.69765079e-01 1.01120554e-01 2.90798753e-01 -6.30670130e-01
7.56412327e-01 -1.09017529e-01 5.07629097e-01 -6.09483540e-01
-5.02277374e-01 -2.97480196e-01 -3.87933493e-01 -3.68573964e-01
1.06573975e+00 -4.29498702e-01 -9.40339446e-01 1.46303969e-02
-5.56087673e-01 -2.39745095e-01 -1.91135153e-01 5.60578942e-01
-5.26343048e-01 1.34959161e-01 -4.64043319e-01 -1.21434236e+00
-8.72304499e-01 -1.01192987e+00 6.16327405e-01 7.15247214e-01
-2.71140099e-01 -9.69695389e-01 -3.00604161e-02 1.65019631e-01
4.29728955e-01 7.42700517e-01 1.23639715e+00 2.15292349e-01
-4.25643504e-01 -7.49089643e-02 2.44253039e-01 2.93478787e-01
6.70758605e-01 4.86501455e-01 -1.56216606e-01 -6.97984457e-01
2.05449224e-01 2.20563188e-01 3.57324421e-01 8.77130747e-01
4.63185400e-01 -2.25284725e-01 -2.20774442e-01 4.83307272e-01
2.44699430e+00 9.23756778e-01 1.45770654e-01 4.34524208e-01
6.44485950e-02 6.35744333e-01 1.25127375e+00 6.45301342e-01
-1.11639760e-01 2.44192004e-01 7.85942256e-01 -1.50090978e-01
3.83641660e-01 -7.57282088e-03 1.54101953e-01 9.07484114e-01
-4.96509254e-01 -2.99550653e-01 -3.79006594e-01 6.51769817e-01
-1.50719035e+00 -7.19957411e-01 -1.93959072e-01 2.44574571e+00
6.28793299e-01 -2.54489720e-01 -1.78301603e-01 2.66917765e-01
8.28889191e-01 -2.01723471e-01 -3.26949000e-01 -1.43142021e+00
2.35362798e-01 6.08790278e-01 1.17345822e+00 7.20847070e-01
-8.75970662e-01 3.91603649e-01 6.68542957e+00 2.66576797e-01
-8.68522286e-01 -4.47993249e-01 3.73249680e-01 -4.49360728e-01
-1.52585767e-02 2.73747463e-02 -6.97413385e-01 3.82107288e-01
1.00075817e+00 -5.49052894e-01 8.97031605e-01 4.79574412e-01
8.32725763e-01 -6.08733535e-01 -8.00799191e-01 5.34476757e-01
-5.28457940e-01 -1.01754797e+00 -3.24770629e-01 3.09920937e-01
9.68860209e-01 -4.64037120e-01 -4.71038252e-01 -2.04831079e-01
6.09517217e-01 -6.40737593e-01 7.33950257e-01 1.64031237e-01
1.35512158e-01 -1.40252912e+00 7.54488230e-01 1.24220602e-01
-1.41314900e+00 -4.50983703e-01 -5.43118596e-01 -8.73968080e-02
6.35936081e-01 8.49528193e-01 -3.92761141e-01 9.95954335e-01
5.99056244e-01 2.01698884e-01 2.27584794e-01 9.54697132e-01
-7.76901245e-02 1.43658882e-02 -6.39574111e-01 -1.99970946e-01
3.07936221e-01 -1.39875638e+00 2.61411399e-01 8.78853977e-01
6.06697321e-01 3.03309619e-01 3.03700119e-01 8.34319711e-01
3.99416387e-01 3.61143410e-01 -1.65698215e-01 -1.66379288e-01
4.38652545e-01 1.36755776e+00 -1.02845418e+00 1.21962287e-01
-5.96260466e-02 6.40050828e-01 -2.35768706e-01 9.80561301e-02
-8.32856536e-01 -4.01691616e-01 6.49348974e-01 -2.78528363e-01
3.57771546e-01 -5.42074144e-01 -3.46726447e-01 -3.87108505e-01
-7.14592934e-02 -4.11129415e-01 1.74144909e-01 -4.04028028e-01
-9.09313977e-01 -1.67320371e-01 1.26615241e-01 -5.87817073e-01
-5.47573604e-02 -5.85021019e-01 -5.48499882e-01 1.03015339e+00
-1.70998693e+00 -4.62908030e-01 -1.94219217e-01 3.68061773e-02
8.02224457e-01 3.69519472e-01 8.26373994e-01 2.13158235e-01
-8.37646723e-01 -1.80405125e-01 5.61568856e-01 -9.16335225e-01
1.52220458e-01 -1.55204260e+00 -3.26306731e-01 8.46996725e-01
-9.91261601e-01 4.53800321e-01 1.21781147e+00 -5.64987600e-01
-2.12255788e+00 -8.47739339e-01 8.91353130e-01 6.80604875e-01
3.75418127e-01 1.15873814e-01 -3.88576418e-01 7.31509551e-02
4.36336964e-01 -8.08384597e-01 3.82247090e-01 -2.10440546e-01
7.66424239e-01 -2.76595026e-01 -1.73243046e+00 4.76236045e-01
5.54242432e-01 3.28758627e-01 -2.73028582e-01 3.24397534e-01
3.41266543e-01 -2.79703885e-01 -1.36940026e+00 5.53027511e-01
3.68231028e-01 -3.92226338e-01 8.72851908e-01 1.90369855e-03
4.20067579e-01 -1.86086401e-01 -1.54405594e-01 -1.56602895e+00
-3.93310368e-01 -9.49370861e-01 -2.48036429e-01 1.33791876e+00
1.79224223e-01 -3.30790281e-01 3.07895094e-01 8.42797101e-01
-8.39633942e-02 -9.99662459e-01 -5.60394883e-01 -7.64835000e-01
-9.14450735e-02 6.17570460e-01 3.75717372e-01 7.04769790e-01
1.13329276e-01 1.90790534e-01 -1.13920152e-01 2.84295678e-01
8.41534972e-01 4.28334683e-01 5.30957393e-02 -1.14090121e+00
1.03678748e-01 -3.50278318e-01 2.86708057e-01 -5.88034093e-01
-3.07561457e-01 -5.88020921e-01 3.36946398e-01 -2.13748169e+00
-6.93552792e-02 -2.10390240e-01 -1.79646477e-01 1.46477073e-01
-2.22443640e-01 -5.34123480e-01 4.82072741e-01 -4.50894594e-01
2.42624775e-01 4.74233210e-01 1.59193289e+00 -1.31599978e-01
-5.68120003e-01 -1.36156768e-01 -4.58871186e-01 2.15684637e-01
1.07912517e+00 -2.90961504e-01 -5.65575957e-01 -2.21103713e-01
2.39419103e-01 3.34731847e-01 -3.63793820e-01 -7.02602208e-01
-3.94744635e-01 -7.10528910e-01 2.81228393e-01 -6.98475540e-01
1.27784058e-01 -1.45649683e+00 5.79780519e-01 8.52007985e-01
7.67236948e-02 2.53311604e-01 8.79679546e-02 4.14170235e-01
-4.43471558e-02 -5.03474414e-01 9.97746348e-01 -1.91235393e-01
-4.27195042e-01 -1.25357181e-01 -4.33785886e-01 -8.18129063e-01
1.40818477e+00 -4.62821007e-01 -2.54924387e-01 2.28826687e-01
-7.11597085e-01 6.69122577e-01 4.90376472e-01 3.72606933e-01
1.34163573e-01 -9.75874007e-01 -4.59961325e-01 -1.22190259e-01
-7.41190195e-01 1.51098613e-02 1.37083516e-01 5.47801912e-01
-1.03214622e+00 5.04303277e-01 -3.37257713e-01 -3.10197800e-01
-1.24209118e+00 7.29262948e-01 4.35017377e-01 -6.32128894e-01
-5.27779639e-01 3.34990144e-01 -6.01073444e-01 -3.37317921e-02
-4.14460957e-01 -7.49626517e-01 1.19670399e-01 2.06463769e-01
-5.68233803e-03 8.58357072e-01 2.29092240e-01 3.17455605e-02
-2.01086730e-01 7.69205689e-01 3.49339455e-01 3.31428111e-01
1.50928342e+00 -3.57190669e-01 -3.34913909e-01 4.36851084e-02
1.05181706e+00 -1.86176866e-01 -1.12030876e+00 5.57632923e-01
7.50956088e-02 -3.78008723e-01 4.68401700e-01 -9.10990655e-01
-1.30219328e+00 2.98144877e-01 7.73918569e-01 2.75151968e-01
1.51523328e+00 -7.64569402e-01 7.11753666e-01 2.69301981e-01
4.61643040e-01 -1.49295735e+00 -5.03278375e-01 7.14477971e-02
5.21446824e-01 -8.58271420e-01 4.53656286e-01 -5.75990498e-01
-1.64811805e-01 1.45966077e+00 4.14218575e-01 -2.77203441e-01
4.58243072e-01 3.70321453e-01 -3.03084582e-01 -1.10614486e-01
-4.74940002e-01 3.20198294e-03 -3.54131460e-01 2.13929608e-01
5.68809450e-01 5.25814295e-01 -1.27645874e+00 -5.28191887e-02
5.08120982e-03 7.35495538e-02 2.13584572e-01 1.42817259e+00
-7.49730587e-01 -1.23495686e+00 -4.33916658e-01 3.53739738e-01
-4.89440084e-01 2.30717331e-01 1.59868270e-01 8.92378688e-01
6.06448650e-02 1.40383494e+00 8.88188109e-02 1.95736736e-01
4.81310636e-01 1.23597018e-01 7.48569131e-01 -2.38348588e-01
-6.64484560e-01 2.95578986e-01 2.86989897e-01 -4.80232865e-01
-7.40513682e-01 -6.00936294e-01 -1.42302871e+00 -4.82466489e-01
-6.63904071e-01 3.64386350e-01 1.26995540e+00 7.82949924e-01
4.35781330e-02 7.85409093e-01 9.68637705e-01 -7.82736242e-01
-5.77444434e-01 -5.57184160e-01 -7.65690506e-01 -1.10152187e-02
1.05028294e-01 -5.96547663e-01 -2.45356187e-01 -1.01504223e-02] | [5.689939498901367, 3.031780242919922] |
cccb0759-51dc-462d-b1a6-b800cd18c5fd | binaural-lcmv-beamforming-with-partial-noise | 1905.04050 | null | https://arxiv.org/abs/1905.04050v2 | https://arxiv.org/pdf/1905.04050v2.pdf | Binaural LCMV Beamforming with Partial Noise Estimation | Besides reducing undesired sources (interfering sources and background noise), another important objective of a binaural beamforming algorithm is to preserve the spatial impression of the acoustic scene, which can be achieved by preserving the binaural cues of all sound sources. While the binaural minimum variance distortionless response (BMVDR) beamformer provides a good noise reduction performance and preserves the binaural cues of the desired source, it does not allow to control the reduction of the interfering sources and distorts the binaural cues of the interfering sources and the background noise. Hence, several extensions have been proposed. First, the binaural linearly constrained minimum variance (BLCMV) beamformer uses additional constraints, enabling to control the reduction of the interfering sources while preserving their binaural cues. Second, the BMVDR with partial noise estimation (BMVDR-N) mixes the output signals of the BMVDR with the noisy reference microphone signals, enabling to control the binaural cues of the background noise. Merging the advantages of both extensions, in this paper we propose the BLCMV with partial noise estimation (BLCMV-N). We show that the output signals of the BLCMV-N can be interpreted as a mixture of the noisy reference microphone signals and the output signals of a BLCMV using an adjusted interference scaling parameter. We provide a theoretical comparison between the BMVDR, the BLCMV, the BMVDR-N and the proposed BLCMV-N in terms of noise and interference reduction performance and binaural cue preservation. Experimental results using recorded signals as well as the results of a perceptual listening test show that the BLCMV-N is able to preserve the binaural cues of an interfering source (like the BLCMV), while enabling to trade off between noise reduction performance and binaural cue preservation of the background noise (like the BMVDR-N). | ['Simon Doclo', 'Sharon Gannot', 'Elior Hadad', 'Nico Gößling'] | 2019-05-10 | null | null | null | null | ['noise-estimation'] | ['medical'] | [ 1.50924191e-01 -2.49055594e-01 6.81711614e-01 3.07411700e-01
-6.71395957e-01 -5.13353109e-01 4.18249965e-01 7.58265406e-02
-2.66750395e-01 4.90545630e-01 5.55428147e-01 -2.14213267e-01
-4.87531453e-01 -6.64845407e-01 -5.15073299e-01 -1.09520471e+00
5.83672039e-02 -3.96862507e-01 5.74452579e-01 -2.08735526e-01
-7.23227486e-02 4.68301386e-01 -2.14171028e+00 -2.57000439e-02
7.22111642e-01 1.30721104e+00 5.63845634e-01 7.77115703e-01
1.98573902e-01 2.31266826e-01 -8.02254438e-01 1.37555301e-01
4.35369760e-01 -6.71487808e-01 2.56982654e-01 -1.58290714e-01
4.99662608e-01 1.49719164e-01 -2.19324455e-01 1.07187831e+00
9.78756130e-01 4.33012098e-01 5.48597932e-01 -1.07484949e+00
-2.15083852e-01 2.05356851e-01 -4.03495461e-01 -2.08866112e-02
3.69186908e-01 7.04418719e-02 5.61739266e-01 -7.06634581e-01
1.06787048e-01 1.08414900e+00 3.19873482e-01 3.48305792e-01
-1.32179403e+00 -7.30244458e-01 -1.58181101e-01 2.07584158e-01
-1.56210148e+00 -6.45653844e-01 8.21339369e-01 -3.57480228e-01
3.86332840e-01 9.05693889e-01 5.72406948e-01 4.26957488e-01
9.83592868e-02 1.30809650e-01 1.20378375e+00 -7.53982902e-01
3.63588631e-01 1.43184811e-01 -9.12970230e-02 -5.56880143e-03
1.59568638e-01 3.98049921e-01 -4.16593492e-01 -3.03565152e-02
4.42701072e-01 -3.92727554e-01 -1.08253443e+00 -4.64539200e-01
-6.91509843e-01 2.07759291e-01 2.99749851e-01 8.43553007e-01
-3.01766068e-01 -7.09381104e-02 -2.51814604e-01 2.28426769e-01
7.27296993e-02 2.64516056e-01 -1.08837802e-02 1.88070208e-01
-8.78328323e-01 1.90480143e-01 6.42904162e-01 6.30575359e-01
6.45318329e-01 4.59497333e-01 -4.15081710e-01 1.18183088e+00
4.50112671e-01 1.00423908e+00 3.90171260e-01 -8.01831722e-01
2.13116586e-01 -5.22325486e-02 2.20106944e-01 -1.08928764e+00
-2.42235541e-01 -8.31305623e-01 -8.58028769e-01 7.63458014e-01
3.58311534e-01 -1.32364988e-01 -7.91370153e-01 2.01395631e+00
2.71001995e-01 8.65512416e-02 9.30323824e-02 1.07015836e+00
5.13308287e-01 8.43925238e-01 -3.97702157e-01 -8.54908228e-01
1.01056695e+00 -4.40582424e-01 -1.13836730e+00 -1.57712638e-01
-9.94605646e-02 -1.33464241e+00 7.06420898e-01 6.02086008e-01
-1.19806361e+00 -9.50693071e-01 -8.75198007e-01 4.81893927e-01
-6.12898394e-02 -2.55481631e-01 -2.64935404e-01 8.54069650e-01
-1.26126063e+00 -1.36821046e-01 -4.59484786e-01 2.04938293e-01
-4.89950806e-01 4.63353097e-02 -2.24023730e-01 -7.26623908e-02
-7.92662323e-01 8.00268590e-01 -1.90952554e-01 2.70636797e-01
-7.30921090e-01 -7.71390796e-01 -7.69206941e-01 3.41089129e-01
2.09937721e-01 -2.40511060e-01 9.78443861e-01 -9.04017568e-01
-1.54096305e+00 9.78742316e-02 -4.38563973e-01 -1.92967534e-01
2.60623574e-01 -2.82983370e-02 -7.59090602e-01 -1.81435458e-02
-5.76168932e-02 4.01003771e-02 7.88097322e-01 -1.63936961e+00
-5.14643133e-01 -1.30370781e-01 -4.93308455e-01 2.76553005e-01
-6.38822913e-02 -2.13430911e-01 -2.35782549e-01 -8.25810254e-01
4.97535080e-01 -6.03225529e-01 -1.06555909e-01 -2.11048886e-01
-2.47582212e-01 1.48735791e-01 7.61994421e-01 -4.53703970e-01
1.30439627e+00 -2.67632031e+00 -2.02367425e-01 3.03936779e-01
-2.14675248e-01 4.72473770e-01 -2.38910466e-01 2.48172179e-01
-3.46016258e-01 -4.67181176e-01 -8.50235298e-02 -2.67528981e-01
-1.86922312e-01 -2.95051793e-03 -1.83514401e-01 4.89045084e-01
-2.98734605e-01 2.92217564e-02 -6.37310505e-01 -1.11760259e-01
4.08796608e-01 8.18727255e-01 -6.28399789e-01 4.68467623e-01
2.61029035e-01 3.27932209e-01 2.37055257e-01 2.88661361e-01
9.82313156e-01 8.90254557e-01 -2.17985839e-01 -2.59839624e-01
-4.07990128e-01 -3.02797090e-02 -1.75498044e+00 8.08777332e-01
-5.43634713e-01 5.64643800e-01 8.49315882e-01 -5.06566703e-01
1.26740086e+00 5.10506332e-01 1.64895386e-01 -9.76495445e-01
-8.69161338e-02 4.81180221e-01 3.46129298e-01 -1.75781921e-01
1.49682477e-01 -5.45778513e-01 1.85862079e-01 -3.01464666e-02
1.19225435e-01 -2.44470358e-01 -1.24920497e-03 -7.75111020e-02
4.89805520e-01 -4.07946140e-01 3.46243143e-01 -2.46840566e-01
9.15447235e-01 -9.63951588e-01 5.92078626e-01 6.33969307e-01
-1.12440616e-01 7.76434064e-01 4.57727350e-02 5.17939389e-01
-5.98464429e-01 -1.24098635e+00 -2.26700753e-01 7.71100581e-01
1.62853166e-01 -1.20512303e-02 -7.50550270e-01 1.58782765e-01
-2.43960395e-01 1.17812979e+00 4.08603810e-02 -6.75390959e-02
-4.85319376e-01 -2.89599240e-01 3.10368866e-01 -2.29782127e-02
1.17473826e-01 -3.71620744e-01 -3.84651840e-01 2.22700641e-01
-3.86700064e-01 -6.87034369e-01 -4.91328359e-01 3.69106799e-01
-2.81164795e-01 -8.64613295e-01 -6.75223410e-01 -5.70195794e-01
3.40801001e-01 3.64994258e-01 4.75066632e-01 -3.82605344e-01
-2.47547384e-02 5.52864552e-01 -3.04640532e-01 -5.66825569e-01
-4.36040789e-01 -9.98180628e-01 2.21532658e-01 5.11931121e-01
-2.18674064e-01 -1.04447258e+00 -5.27995527e-01 6.30702972e-01
-9.41305161e-01 -2.90708840e-01 3.92056018e-01 4.73478407e-01
7.22968817e-01 4.84607428e-01 3.73625219e-01 1.55443564e-01
5.48665345e-01 -1.71686992e-01 -7.36885548e-01 -3.33811611e-01
-3.57486218e-01 -4.14009362e-01 7.64551342e-01 -6.57398582e-01
-1.20539558e+00 -9.67860669e-02 -5.67195117e-01 -4.63130504e-01
-4.81861204e-01 -3.99875157e-02 -8.00767243e-01 -1.02918893e-01
6.77682877e-01 4.28129315e-01 -2.57063001e-01 -8.19675505e-01
2.92496890e-01 9.42410648e-01 9.08898354e-01 -4.38137054e-02
6.80473089e-01 2.78951287e-01 1.76152200e-01 -1.09605122e+00
-2.84472823e-01 -6.56413853e-01 -5.48932031e-02 -4.55746055e-01
5.88502109e-01 -6.29207075e-01 -3.68074179e-01 5.27175188e-01
-8.91961932e-01 -2.90320683e-02 -4.84470248e-01 9.62757885e-01
-5.18772125e-01 3.83074552e-01 -4.24456894e-02 -1.49848819e+00
6.99908361e-02 -1.21714938e+00 5.43147504e-01 3.03348482e-01
3.91331222e-03 -5.54042757e-01 1.28806263e-01 3.14100608e-02
6.16957843e-01 2.04299569e-01 8.88069093e-01 -4.02856976e-01
-3.34079236e-01 -2.68019050e-01 4.25176352e-01 7.08131194e-01
2.56389737e-01 -4.06227082e-01 -1.05942404e+00 -2.97198892e-01
5.69126427e-01 4.09153849e-01 5.06214321e-01 8.85371745e-01
5.22454798e-01 -4.52987641e-01 -1.86538100e-02 4.91754472e-01
1.72510493e+00 7.72550702e-01 9.45975542e-01 1.04366504e-02
1.34788007e-01 5.96211016e-01 5.33591151e-01 1.38661578e-01
-2.07457826e-01 8.52271795e-01 6.59258127e-01 -2.20852122e-01
-6.85448170e-01 -6.69365376e-02 1.74045816e-01 9.40257609e-01
1.32945821e-01 -5.33196688e-01 -4.52145517e-01 7.78808296e-01
-1.47358549e+00 -7.77118862e-01 -5.83053648e-01 2.69621801e+00
6.24373853e-01 -8.28143023e-03 -1.18429683e-01 7.62626886e-01
8.96178961e-01 2.01713860e-01 -9.59832221e-03 -7.99267530e-01
-2.49880791e-01 4.83854860e-01 2.52922684e-01 1.11420119e+00
-6.32292271e-01 1.25369653e-01 5.54092598e+00 1.02065098e+00
-1.13546467e+00 1.94603190e-01 -2.97735985e-02 -3.82058382e-01
-2.56768942e-01 -1.15116283e-01 -6.35972679e-01 5.89224279e-01
7.38146901e-01 -1.30708858e-01 4.65489745e-01 5.77405572e-01
5.78490674e-01 -5.43763518e-01 -8.21824491e-01 1.01410854e+00
1.64978281e-01 -4.93360728e-01 -1.05473042e-01 8.38083550e-02
4.44022566e-01 -4.91728544e-01 1.78201959e-01 -4.25281189e-02
-1.69421867e-01 -7.30980277e-01 9.63039637e-01 7.60657430e-01
5.18476963e-01 -9.42507982e-01 8.08918178e-01 5.47434092e-01
-1.06569731e+00 -1.76812083e-01 -2.35444099e-01 -2.04042912e-01
1.86776206e-01 1.05556285e+00 -4.56181914e-01 5.76623857e-01
6.98763728e-01 -3.37582380e-01 -1.13680027e-01 1.59223688e+00
-3.14980894e-01 6.02577448e-01 -3.02492291e-01 2.37252280e-01
1.22443689e-02 -1.22646540e-01 1.23694229e+00 1.28774786e+00
6.26151443e-01 1.93904743e-01 -2.09236830e-01 8.03650379e-01
4.13135052e-01 3.55896652e-01 -4.00268316e-01 5.10026336e-01
4.50944126e-01 9.82535958e-01 -1.29117653e-01 2.10629720e-02
-7.10712001e-02 4.20550823e-01 -7.07433105e-01 6.46256864e-01
-5.16202271e-01 -7.57701397e-01 9.83356476e-01 4.01057661e-01
4.17094082e-01 2.39774603e-02 -9.50618610e-02 -4.04682487e-01
4.07474451e-02 -9.37184811e-01 1.46092065e-02 -1.17009163e+00
-8.68302464e-01 6.43839061e-01 -1.47752792e-01 -1.48871565e+00
-1.77028373e-01 -3.27157468e-01 -5.42111278e-01 1.45738852e+00
-1.44694197e+00 -6.40582442e-01 -4.89007123e-02 7.91747570e-01
3.10455352e-01 1.00953571e-01 6.48655355e-01 3.84857267e-01
6.03151135e-02 4.81246829e-01 2.81075776e-01 -2.57919461e-01
7.81098425e-01 -1.05068004e+00 -6.62068665e-01 1.20320547e+00
-3.37316751e-01 5.07949352e-01 1.23634815e+00 -3.06371063e-01
-9.54777360e-01 -9.32625055e-01 1.09164274e+00 2.76762992e-01
2.45884657e-01 -4.17250603e-01 -9.21349883e-01 1.26459338e-02
2.12148577e-01 -2.51621664e-01 7.74943352e-01 -4.49482799e-01
-1.16841622e-01 -7.25714326e-01 -8.80992830e-01 5.80683768e-01
5.03581345e-01 -4.46454018e-01 -7.61655807e-01 -7.85656497e-02
6.39312088e-01 -1.61405772e-01 -3.73385608e-01 4.97912437e-01
4.90470856e-01 -1.38814604e+00 8.58498573e-01 5.36287308e-01
-4.92504127e-02 -8.58635008e-01 -5.23766696e-01 -1.61817479e+00
-4.82706070e-01 -5.91523886e-01 1.06673107e-01 1.58739471e+00
8.41525346e-02 -7.59677231e-01 -7.48851448e-02 1.24317020e-01
-4.01758820e-01 -1.15289867e-01 -1.06910360e+00 -8.77836645e-01
-2.06550464e-01 -7.03172147e-01 2.02589646e-01 5.63411832e-01
-1.38857856e-01 6.29116148e-02 -3.37491542e-01 6.87568545e-01
6.71887159e-01 -5.52855320e-02 6.23704553e-01 -9.08021986e-01
-7.19610393e-01 -5.81586421e-01 -3.32906157e-01 -1.02646136e+00
-5.56070328e-01 -3.65404338e-01 5.35420954e-01 -1.49182844e+00
-5.58207929e-01 -1.29833981e-01 -6.01126492e-01 9.47085172e-02
8.28718580e-03 2.83551633e-01 4.87702310e-01 -1.71477243e-01
1.80451974e-01 4.01836157e-01 1.23293424e+00 5.99937141e-02
-4.83524173e-01 3.98147136e-01 -6.11297429e-01 8.58351886e-01
1.79997578e-01 -4.65804935e-01 -3.27539206e-01 -1.32455409e-01
-1.47942320e-01 3.33229750e-01 3.12883288e-01 -1.25975871e+00
4.52476710e-01 1.04219228e-01 2.27333933e-01 -7.53960609e-01
6.22788727e-01 -1.12833655e+00 5.76885700e-01 4.13102299e-01
-3.13568413e-02 -6.47669673e-01 2.49695450e-01 6.40122771e-01
-4.82860327e-01 -1.92004964e-01 1.29904652e+00 4.55414318e-02
-1.51539385e-01 -4.78925407e-01 -9.48891938e-01 -5.68075240e-01
7.68580556e-01 -3.36999893e-01 -1.27980992e-01 -8.73770058e-01
-7.52997279e-01 -1.84871420e-01 -2.74032876e-02 1.50321126e-01
5.95745385e-01 -1.09719396e+00 -5.58567941e-01 4.48812097e-01
-3.38252962e-01 -1.38320088e-01 3.60868782e-01 9.59222376e-01
-1.88216902e-02 4.65293854e-01 -1.86122417e-01 -6.02451980e-01
-1.46841156e+00 6.82250798e-01 6.66390836e-01 2.21692264e-01
-4.10509631e-02 8.00876737e-01 7.07997262e-01 -2.53273232e-04
4.90683079e-01 -3.07663113e-01 -5.36830842e-01 -6.23255298e-02
7.52570748e-01 6.78850353e-01 2.72411972e-01 -7.29611099e-01
-2.38209218e-01 6.13332689e-01 7.11415052e-01 -4.64214951e-01
1.13000619e+00 -6.17380321e-01 -2.73603320e-01 4.14468914e-01
1.13415909e+00 1.07347226e+00 -7.18503654e-01 -8.06196481e-02
-4.66810405e-01 -6.53151333e-01 4.79342818e-01 -1.08478463e+00
-7.63147175e-01 9.61006880e-01 1.17574692e+00 4.46547329e-01
2.03123569e+00 -3.60192120e-01 4.27896082e-01 -2.78698921e-01
2.97868311e-01 -6.98171854e-01 2.18639672e-02 1.89382270e-01
8.95488858e-01 -1.91795856e-01 -5.04727244e-01 -4.68751699e-01
-2.04950884e-01 8.64247262e-01 2.93202251e-01 -5.42231314e-02
6.52603328e-01 4.05033201e-01 2.45741650e-01 3.74293536e-01
-3.81088525e-01 -5.74928701e-01 5.03443599e-01 7.57663310e-01
2.14734495e-01 -6.90997094e-02 -6.38239503e-01 8.88928890e-01
-3.51744592e-01 -5.65838575e-01 5.09500027e-01 4.47259009e-01
-8.53123069e-01 -8.85243237e-01 -1.14147592e+00 -8.21801350e-02
-4.96336788e-01 -1.90818593e-01 -2.79200643e-01 3.12040746e-01
7.75375009e-01 1.52978742e+00 7.43952617e-02 -3.56252939e-01
1.02638912e+00 3.05059552e-01 1.15875602e-01 -3.04501384e-01
-4.12952751e-01 1.29039085e+00 -1.11115821e-01 -3.49055558e-01
-3.03864956e-01 -3.62577230e-01 -6.59148276e-01 8.32452327e-02
-6.73204064e-01 4.17175204e-01 7.31741965e-01 5.87808549e-01
-2.02059094e-02 7.65380859e-01 8.25269759e-01 -6.32870138e-01
-8.60071480e-02 -9.20654774e-01 -1.01996303e+00 1.73952252e-01
7.57889807e-01 -3.72754008e-01 -9.05881822e-01 -5.35713620e-02] | [15.126298904418945, 5.7947211265563965] |
5ed70a80-e2c6-45b9-b1d4-703c2b9105b1 | trust-but-verify-cross-modality-fusion-for-hd | 2212.07312 | null | https://arxiv.org/abs/2212.07312v1 | https://arxiv.org/pdf/2212.07312v1.pdf | Trust, but Verify: Cross-Modality Fusion for HD Map Change Detection | High-definition (HD) map change detection is the task of determining when sensor data and map data are no longer in agreement with one another due to real-world changes. We collect the first dataset for the task, which we entitle the Trust, but Verify (TbV) dataset, by mining thousands of hours of data from over 9 months of autonomous vehicle fleet operations. We present learning-based formulations for solving the problem in the bird's eye view and ego-view. Because real map changes are infrequent and vector maps are easy to synthetically manipulate, we lean on simulated data to train our model. Perhaps surprisingly, we show that such models can generalize to real world distributions. The dataset, consisting of maps and logs collected in six North American cities, is one of the largest AV datasets to date with more than 7.8 million images. We make the data available to the public at https://www.argoverse.org/av2.html#mapchange-link, along with code and models at https://github.com/johnwlambert/tbv under the the CC BY-NC-SA 4.0 license. | ['James Hays', 'John Lambert'] | 2022-12-14 | null | null | null | null | ['change-detection'] | ['computer-vision'] | [-3.40699971e-01 -2.47360760e-04 -1.06768340e-01 -8.64910126e-01
-6.30955279e-01 -8.53584588e-01 8.58605683e-01 1.47816360e-01
-2.89212853e-01 8.00815344e-01 3.09353825e-02 -4.39769208e-01
-1.08678630e-02 -9.98575211e-01 -1.19595623e+00 -3.63494158e-01
-5.55957437e-01 7.34858692e-01 4.72796410e-01 -3.55020583e-01
2.23552473e-02 3.99085313e-01 -1.63035381e+00 -5.26516745e-03
5.96164644e-01 7.98386931e-01 3.04446071e-01 8.26478899e-01
4.30686116e-01 6.60658658e-01 -1.56744719e-01 -1.72697589e-01
6.57382727e-01 4.03747320e-01 -4.68603820e-01 -4.51908633e-02
4.68748510e-01 -6.31961405e-01 -6.09684348e-01 1.06702030e+00
2.14645803e-01 -1.56845316e-01 3.89604539e-01 -2.00951409e+00
-5.51402330e-01 3.16175729e-01 -5.34215450e-01 3.87758851e-01
2.51223654e-01 5.03739893e-01 8.47575307e-01 -7.62286544e-01
7.54679680e-01 9.05397594e-01 7.66367018e-01 -4.38206419e-02
-1.15874100e+00 -6.49953902e-01 1.01356953e-01 2.98184246e-01
-1.63632524e+00 -7.13580608e-01 2.98129112e-01 -7.86110103e-01
9.24038410e-01 3.35571676e-01 4.85712558e-01 1.07039416e+00
3.31655651e-01 6.50417268e-01 9.29613113e-01 1.75362512e-01
2.43597999e-01 2.50343651e-01 -8.34498629e-02 5.61593592e-01
3.61547112e-01 2.54429162e-01 -3.70548189e-01 -1.97301060e-01
3.23994935e-01 1.25459194e-01 3.96988634e-03 -6.16018236e-01
-1.25108230e+00 7.03035057e-01 5.12157917e-01 -3.64631444e-01
-5.25945127e-01 3.89093220e-01 1.72353521e-01 5.15983522e-01
5.69829226e-01 7.67986327e-02 -5.93949258e-01 -5.00374019e-01
-5.50235987e-01 6.33950293e-01 7.28214800e-01 1.22011590e+00
1.14866734e+00 -2.33808145e-01 6.53617203e-01 3.56931806e-01
2.62684077e-01 1.05041838e+00 -1.10496581e-01 -1.08839798e+00
5.53934038e-01 3.30198765e-01 5.96221089e-01 -1.18108249e+00
-4.75637645e-01 9.97566283e-02 -6.08551025e-01 2.77114481e-01
2.71219492e-01 -3.17399800e-01 -8.33720922e-01 1.58762145e+00
2.73570299e-01 2.34316573e-01 3.53694633e-02 5.64558804e-01
5.55508614e-01 6.80137873e-01 -4.05099303e-01 2.01275393e-01
8.79843533e-01 -3.38181973e-01 -5.71050465e-01 -5.24093449e-01
7.14312255e-01 -3.19146484e-01 9.60857391e-01 1.06828116e-01
-5.63844502e-01 -1.75571084e-01 -1.13595271e+00 4.40175951e-01
-8.33814383e-01 -3.20563555e-01 5.18696904e-01 2.14624368e-02
-1.35505378e+00 1.27882818e-02 -1.13240337e+00 -7.88253307e-01
5.71913362e-01 5.08140922e-02 -5.65538466e-01 -1.81690961e-01
-1.21779799e+00 1.07431304e+00 1.89444035e-01 -1.12716677e-02
-1.05387998e+00 -5.98962426e-01 -9.35610533e-01 -5.88286340e-01
3.68945926e-01 -5.08239530e-02 1.28158808e+00 -4.32781070e-01
-5.34786046e-01 8.26198459e-01 -1.30630344e-01 -4.78432089e-01
7.45349050e-01 9.84636042e-03 -6.59912407e-01 -3.72044265e-01
5.47165453e-01 7.28119910e-01 4.84020948e-01 -1.43825114e+00
-1.08009958e+00 -4.11110938e-01 1.14450701e-01 -1.22150861e-01
3.07968795e-01 -1.56300738e-01 -2.45040134e-01 -5.04592694e-02
2.52222009e-02 -1.16679502e+00 -8.15210119e-02 1.43932542e-02
-3.75041634e-01 -1.48079656e-02 8.35552692e-01 -7.44917154e-01
9.30487752e-01 -2.04529381e+00 -4.04238999e-01 3.60616475e-01
3.64454806e-01 -2.54884988e-01 -1.35228485e-01 6.90364480e-01
1.56604305e-01 1.58878461e-01 -2.82280803e-01 -1.10550866e-01
2.06475243e-01 4.20137972e-01 -3.98260832e-01 8.77571464e-01
-1.06291719e-01 8.72528970e-01 -9.32747066e-01 1.16924345e-02
2.25722417e-01 -6.44304752e-02 -7.50223026e-02 -7.22001269e-02
-2.51334697e-01 3.10716718e-01 -2.57859915e-01 7.09820569e-01
9.70685780e-01 -1.02747962e-01 -4.54660431e-02 5.86325377e-02
-4.97597396e-01 1.08988941e-01 -1.16823721e+00 1.30273163e+00
-8.53974670e-02 1.04830945e+00 1.11690722e-01 -5.73080242e-01
9.04438257e-01 -1.79512203e-01 5.00333548e-01 -9.06110227e-01
-2.78991580e-01 5.69256544e-02 -1.57561257e-01 -5.70700824e-01
6.12429976e-01 4.22517508e-01 -4.56618488e-01 5.71284771e-01
-5.29273033e-01 -4.10738289e-01 3.07605088e-01 3.44562918e-01
1.45057166e+00 -2.56305784e-01 1.64469242e-01 -1.47578239e-01
-2.10244328e-01 5.27222693e-01 6.47350848e-01 7.55259156e-01
-4.26735520e-01 2.29185000e-01 5.12382030e-01 -8.20893347e-01
-1.37263501e+00 -1.32881820e+00 -3.84301901e-01 7.52342045e-01
3.14314604e-01 -3.00439268e-01 -3.43168527e-01 -3.23602021e-01
6.78999305e-01 1.07958937e+00 -5.85371375e-01 2.89458901e-01
-2.13769168e-01 -6.53757751e-01 6.70207500e-01 3.97960126e-01
4.85585541e-01 -5.50239325e-01 -8.02889049e-01 -5.97727634e-02
-2.04150990e-01 -1.05166149e+00 -4.44316566e-02 -1.75639018e-02
-2.41410717e-01 -1.14606714e+00 -2.13638619e-02 -2.43354082e-01
6.10761642e-01 3.55294943e-01 1.04770231e+00 -2.13084847e-01
-1.94875747e-01 5.72286725e-01 -3.19095075e-01 -6.96548522e-01
-3.35683823e-01 -6.35054111e-02 5.10442436e-01 -2.49224871e-01
7.47048438e-01 -6.34656012e-01 -1.99353486e-01 5.08728087e-01
-6.09408081e-01 9.18305889e-02 3.37565929e-01 2.70757943e-01
6.68206751e-01 2.45850921e-01 4.42608207e-01 -5.91739595e-01
4.26086932e-01 -1.12126446e+00 -1.18544614e+00 3.17557417e-02
-8.71779442e-01 -3.53858680e-01 -5.37059791e-02 1.86236188e-01
-7.44202018e-01 1.95515856e-01 1.39965817e-01 -3.11918288e-01
-4.70761329e-01 5.77342749e-01 1.02803949e-02 1.56319335e-01
7.84560680e-01 1.95132032e-01 1.92445144e-01 -1.32837683e-01
4.50371265e-01 9.58033860e-01 7.64273942e-01 -1.80437952e-01
1.20242381e+00 7.71855295e-01 -3.76246691e-01 -7.10539520e-01
-3.61888409e-01 -4.56758976e-01 -7.93346524e-01 -5.12877464e-01
6.06933951e-01 -1.26737893e+00 -5.82221270e-01 6.58080876e-01
-9.09253538e-01 -9.45546627e-01 -8.36625844e-02 2.89919257e-01
-7.06321001e-01 -9.95643958e-02 -1.42301843e-01 -5.76992214e-01
2.07340881e-01 -9.04416621e-01 7.90367842e-01 -7.26449043e-02
-1.65787026e-01 -7.28596985e-01 4.09543514e-01 -3.08015454e-03
6.98073626e-01 5.69377959e-01 3.34704429e-01 -4.97808337e-01
-8.59188616e-01 -4.87039268e-01 -2.34368056e-01 7.11196288e-02
3.36852461e-01 1.72760487e-01 -9.16185975e-01 -4.82922912e-01
-3.14166576e-01 -2.06254616e-01 6.70481443e-01 2.83660650e-01
7.69735157e-01 -5.21255612e-01 -6.14298582e-01 6.24092937e-01
1.40595007e+00 9.22542959e-02 5.23207009e-01 7.70049155e-01
5.96127868e-01 5.46218336e-01 9.89750981e-01 7.95523703e-01
1.07806730e+00 7.17053711e-01 9.09430146e-01 1.21783707e-02
3.90428215e-01 -3.56615424e-01 2.72544175e-01 2.50531316e-01
2.37585440e-01 -3.05937141e-01 -1.71826518e+00 8.54329407e-01
-2.20555329e+00 -1.08151984e+00 -3.82366419e-01 2.27446938e+00
4.05930579e-01 2.58834511e-01 1.01583801e-01 -5.05768359e-01
5.30937552e-01 4.64956403e-01 -1.01830816e+00 1.06248528e-01
-1.65745988e-01 -6.18612289e-01 1.30000269e+00 6.24460518e-01
-1.35893357e+00 8.75791430e-01 5.81821108e+00 1.82959080e-01
-1.00405777e+00 2.47713849e-01 3.84405017e-01 -2.61795133e-01
-5.07829189e-01 3.82310450e-02 -6.60574555e-01 6.17331922e-01
1.30122268e+00 -4.64085549e-01 6.05061889e-01 8.80114973e-01
5.53211987e-01 -4.32737380e-01 -1.01883793e+00 9.08445835e-01
-9.55019072e-02 -1.47083747e+00 -6.10043228e-01 3.18653882e-01
7.01359630e-01 1.22439444e+00 1.06694244e-01 1.37611493e-01
8.40217471e-01 -9.15976703e-01 9.60701942e-01 6.77942872e-01
1.03488934e+00 -6.54945791e-01 8.12628329e-01 5.51072598e-01
-1.13096058e+00 -4.90336232e-02 -3.95026505e-01 -1.31375194e-01
3.66816908e-01 5.70220172e-01 -1.01753175e+00 2.01968580e-01
1.21838653e+00 1.11742330e+00 -8.23917091e-01 8.78283858e-01
-1.02268986e-01 5.07935345e-01 -6.54922664e-01 2.60062307e-01
1.72019482e-01 -9.93543118e-02 7.74428248e-01 9.36112344e-01
2.91979373e-01 -2.25465208e-01 1.21491261e-01 9.06501889e-01
1.18963793e-01 -6.46441936e-01 -1.32608843e+00 2.51032919e-01
9.51005220e-01 1.05845439e+00 -2.27905750e-01 -1.47198573e-01
-4.54372197e-01 5.51239669e-01 1.67202130e-01 4.04270470e-01
-9.49180543e-01 -3.24809223e-01 1.09403789e+00 4.25487071e-01
2.06147775e-01 -4.37427610e-01 -2.16612536e-02 -8.27654839e-01
3.09587151e-01 -5.52570760e-01 3.66717540e-02 -9.52938020e-01
-1.18643427e+00 3.71869862e-01 4.34301674e-01 -1.37942028e+00
-6.33149683e-01 -3.17915112e-01 -5.73648214e-01 6.56980693e-01
-1.49993575e+00 -1.18065214e+00 -6.89579844e-01 4.62675303e-01
3.49205106e-01 -2.21313849e-01 5.53106427e-01 2.29914755e-01
-3.30570787e-01 2.46786326e-01 5.17524242e-01 6.04011640e-02
7.44921863e-01 -1.09476602e+00 1.17155421e+00 1.00765646e+00
-1.34245858e-01 1.87589467e-01 9.05022025e-01 -8.59561145e-01
-1.50282300e+00 -1.47872210e+00 7.33643830e-01 -9.39475417e-01
1.05998027e+00 -6.25694215e-01 -8.17067981e-01 1.49134231e+00
-1.49718327e-02 1.31683514e-01 2.01787323e-01 -8.43252093e-02
-2.43458197e-01 -3.80570292e-01 -1.20860028e+00 3.19961876e-01
1.06585467e+00 -5.15140057e-01 -3.18397105e-01 5.84972739e-01
5.86646199e-01 -4.62977409e-01 -8.20670426e-01 4.31363106e-01
4.80078369e-01 -9.69363928e-01 5.83284974e-01 -3.14512193e-01
-4.83625755e-02 -7.76419282e-01 -7.74929702e-01 -1.52491486e+00
-3.95244062e-01 -3.16541195e-01 1.57146633e-01 1.18353868e+00
7.38070130e-01 -1.14579058e+00 5.90251327e-01 7.90546656e-01
-1.80645630e-01 -2.79708266e-01 -1.04514825e+00 -9.35621619e-01
-1.72665745e-01 -7.20887721e-01 1.06750059e+00 1.08952749e+00
-1.55807167e-01 -1.81736305e-01 -2.33136013e-01 7.74898052e-01
7.08954215e-01 -2.72379428e-01 1.41350889e+00 -1.20584667e+00
1.40338883e-01 -5.97144738e-02 -7.84249127e-01 -7.90755033e-01
9.90323424e-02 -7.21036673e-01 2.88068861e-01 -1.62777996e+00
4.99455109e-02 -9.74438548e-01 9.42636505e-02 8.36458981e-01
3.32694858e-01 -1.57098979e-01 -1.95951730e-01 5.81238151e-01
-5.85076988e-01 3.80672514e-01 3.83721054e-01 -3.49233121e-01
-3.17490101e-02 2.30407089e-01 -4.44430560e-01 5.59388220e-01
1.44292426e+00 -5.22851110e-01 -4.91381168e-01 -7.13752329e-01
4.90108550e-01 -3.26641798e-01 5.56965768e-01 -1.01547182e+00
4.69629049e-01 -4.56059277e-01 9.38004851e-02 -9.51679468e-01
2.56817102e-01 -8.20434332e-01 6.48351192e-01 3.37874800e-01
-5.99851087e-02 7.16357589e-01 4.23359931e-01 6.42037392e-01
-4.77691628e-02 2.28032902e-01 4.00585800e-01 -7.98540562e-03
-1.45243394e+00 5.41601479e-01 -4.55037683e-01 1.09222420e-02
1.40162444e+00 -8.19169506e-02 -9.71014380e-01 -7.30340004e-01
-2.58060277e-01 7.78151453e-01 9.78498936e-01 6.86535358e-01
5.57566762e-01 -1.23947632e+00 -1.04824042e+00 4.68478441e-01
7.84724236e-01 1.41096160e-01 1.71259820e-01 8.00265908e-01
-6.53323829e-01 2.68081844e-01 -2.45934442e-01 -8.37207913e-01
-8.46642077e-01 3.54197621e-01 3.93903881e-01 4.80063677e-01
-7.08634138e-01 5.42500913e-01 -3.41026373e-02 -1.00877786e+00
-3.77749987e-02 -2.27309793e-01 3.30619723e-01 -4.11899984e-02
6.17497087e-01 4.67381746e-01 -1.25670554e-02 -9.11386728e-01
-5.32479882e-01 6.38750941e-02 6.69150203e-02 -2.63531864e-01
1.58006239e+00 -5.41552663e-01 -8.10766444e-02 7.21189916e-01
9.70769703e-01 -6.25519678e-02 -1.57407713e+00 -9.29397047e-02
1.04522845e-02 -7.50913501e-01 -3.87043394e-02 -7.49366939e-01
-8.17583978e-01 2.52620906e-01 9.37916458e-01 2.58649200e-01
5.59316576e-01 3.62544924e-01 5.02838969e-01 6.02613032e-01
7.53657579e-01 -1.11420858e+00 -4.66908634e-01 4.93567705e-01
9.41921294e-01 -1.62836468e+00 1.04634399e-02 5.18209189e-02
-7.20074654e-01 4.09936696e-01 6.81387126e-01 7.34840184e-02
9.77593064e-01 5.55205166e-01 1.54484436e-01 -4.78045762e-01
-1.10168958e+00 -1.68168306e-01 -4.77618545e-01 1.06103325e+00
-4.49679911e-01 6.69292092e-01 6.25713587e-01 -4.39511351e-02
-5.39188921e-01 -5.17049097e-02 8.61580610e-01 9.97818232e-01
-6.40851974e-01 -5.50663471e-01 -1.44106314e-01 7.65168190e-01
2.88577855e-01 7.92170316e-02 -1.51007205e-01 1.04767835e+00
-4.42350507e-02 1.04096711e+00 5.65377891e-01 -7.66893029e-01
5.37490547e-01 -3.47174674e-01 -4.05632444e-02 -2.78788179e-01
3.34228516e-01 -6.94557488e-01 3.63393456e-01 -8.19491863e-01
-1.01751931e-01 -1.07759166e+00 -1.13941228e+00 -1.00056374e+00
3.13584566e-01 -2.55676508e-01 9.40495253e-01 4.77094322e-01
6.09741569e-01 1.04860654e-02 7.66658366e-01 -8.29219759e-01
-4.30346906e-01 -9.45329309e-01 -7.87661970e-01 1.72021270e-01
6.92358494e-01 -6.59472585e-01 -4.98003185e-01 -6.72594458e-02] | [7.7961626052856445, -1.9003843069076538] |
ee9793f1-b33c-48a1-bf69-7b518c7547b6 | when-is-nontrivial-estimation-possible-for | 1604.01871 | null | http://arxiv.org/abs/1604.01871v1 | http://arxiv.org/pdf/1604.01871v1.pdf | When is Nontrivial Estimation Possible for Graphons and Stochastic Block Models? | Block graphons (also called stochastic block models) are an important and
widely-studied class of models for random networks. We provide a lower bound on
the accuracy of estimators for block graphons with a large number of blocks. We
show that, given only the number $k$ of blocks and an upper bound $\rho$ on the
values (connection probabilities) of the graphon, every estimator incurs error
at least on the order of $\min(\rho, \sqrt{\rho k^2/n^2})$ in the $\delta_2$
metric with constant probability, in the worst case over graphons. In
particular, our bound rules out any nontrivial estimation (that is, with
$\delta_2$ error substantially less than $\rho$) when $k\geq n\sqrt{\rho}$.
Combined with previous upper and lower bounds, our results characterize, up to
logarithmic terms, the minimax accuracy of graphon estimation in the $\delta_2$
metric. A similar lower bound to ours was obtained independently by Klopp,
Tsybakov and Verzelen (2016). | ['Adam Smith', 'Audra McMillan'] | 2016-04-07 | null | null | null | null | ['graphon-estimation'] | ['graphs'] | [ 6.20236956e-02 5.91741085e-01 -3.67323875e-01 8.64756182e-02
-5.57296813e-01 -7.35233366e-01 2.37901732e-02 3.17003101e-01
-2.84861982e-01 8.57778609e-01 -5.33268094e-01 -6.64056659e-01
-6.67113900e-01 -1.19757736e+00 -9.18238461e-01 -7.92851985e-01
-7.28945136e-01 4.16424066e-01 3.96140218e-01 -1.21243507e-01
1.39751539e-01 3.92555565e-01 -1.07902670e+00 -6.48738384e-01
7.72018492e-01 1.18163013e+00 -2.06145570e-01 1.00834930e+00
2.59964079e-01 5.97243845e-01 -3.20096403e-01 -9.84505832e-01
5.28697908e-01 -6.74794316e-01 -6.20529771e-01 -2.48946548e-01
2.82015026e-01 -4.80641499e-02 -7.98063576e-01 1.56181812e+00
1.71676889e-01 -8.37918669e-02 7.13302016e-01 -1.14579093e+00
-2.03532368e-01 1.13289917e+00 -8.37712348e-01 3.44910175e-01
5.24089858e-02 -5.77538982e-02 1.27881694e+00 -1.68324202e-01
4.82838750e-01 7.67891407e-01 8.69187355e-01 1.90077767e-01
-1.59911740e+00 -1.20137417e+00 1.55446663e-01 3.48171219e-02
-1.74403751e+00 -3.69223468e-02 4.48552102e-01 -3.95899117e-01
6.51153922e-01 3.47373784e-01 6.97089612e-01 3.60327125e-01
1.32355779e-01 2.31914118e-01 8.79149616e-01 -5.12672305e-01
1.20965391e-01 -6.05326109e-02 3.88669878e-01 1.00093853e+00
1.02489603e+00 -7.17851624e-04 -2.72869319e-01 -1.80992290e-01
7.22024381e-01 -2.41629854e-01 -2.07662269e-01 -3.33383888e-01
-9.11005199e-01 7.81792879e-01 3.79078865e-01 1.97889730e-01
-2.73233876e-02 1.01370382e+00 2.02439532e-01 3.00282210e-01
2.72637039e-01 3.50041091e-02 -4.04864728e-01 -2.96023220e-01
-8.33113134e-01 -5.88854169e-03 1.03514636e+00 1.22169888e+00
8.06396842e-01 -6.52289540e-02 3.08191448e-01 3.74889702e-01
1.60141200e-01 9.21037018e-01 -4.85465139e-01 -9.58177745e-01
6.56939268e-01 3.90011936e-01 1.26973823e-01 -1.08558095e+00
-3.05506796e-01 -6.19568348e-01 -1.13929832e+00 -1.08868562e-01
8.49373460e-01 -4.29533571e-02 -4.76892769e-01 2.04907274e+00
-8.77577290e-02 -1.93091005e-01 -5.34508169e-01 3.04430038e-01
1.48946658e-01 5.43298721e-01 -1.87181249e-01 -5.07425845e-01
1.12541854e+00 -5.17108381e-01 -2.47376785e-01 -1.82825267e-01
6.01398408e-01 -4.05682981e-01 6.36602223e-01 5.69725513e-01
-1.15079141e+00 -3.44218500e-02 -1.12582898e+00 4.42106813e-01
-6.69539720e-02 -2.83951372e-01 7.29450762e-01 1.21531141e+00
-1.14993238e+00 7.22263038e-01 -7.42369711e-01 -2.07530931e-01
3.42353016e-01 5.96654534e-01 -1.95203796e-01 -1.79505974e-01
-8.87487531e-01 4.44442004e-01 -7.74144083e-02 1.82545289e-01
-7.47517109e-01 -5.30089974e-01 -6.16575241e-01 1.96482807e-01
5.43432474e-01 -1.00634187e-01 6.69128358e-01 -4.90769774e-01
-9.61337686e-01 5.87468088e-01 -6.57333657e-02 -7.71838784e-01
4.42136586e-01 3.56665194e-01 -1.75068006e-01 2.80861080e-01
3.93483564e-02 5.67631610e-02 4.81050849e-01 -9.09982920e-01
-2.48330519e-01 -6.22650921e-01 4.28710997e-01 -2.93137312e-01
-1.12877697e-01 -1.27528325e-01 -1.92644432e-01 -2.23822057e-01
2.74147183e-01 -1.05837274e+00 -3.16680878e-01 -6.98527992e-02
-4.30874020e-01 -6.73456565e-02 9.29261073e-02 -4.62520540e-01
1.53791678e+00 -1.84515560e+00 2.94732209e-02 8.95174265e-01
7.01035261e-01 -6.15656376e-02 1.54942334e-01 6.25334799e-01
-5.17208269e-03 5.28588951e-01 -2.57807195e-01 1.40324682e-01
3.58284153e-02 -8.29363912e-02 1.47324353e-01 9.04616833e-01
-5.40697575e-01 5.40706277e-01 -8.87287676e-01 -1.63193613e-01
-1.05043352e-02 1.88784495e-01 -3.88562202e-01 -4.09918249e-01
7.36413524e-02 -5.91218844e-02 -3.65928262e-01 1.99702591e-01
8.36152613e-01 -8.13628495e-01 6.04243994e-01 1.57407194e-01
2.67249852e-01 2.88430512e-01 -1.53288805e+00 1.00462496e+00
-4.10840422e-01 7.66318023e-01 4.09068137e-01 -1.21621823e+00
4.73760366e-01 8.52260515e-02 5.05034089e-01 -2.82391816e-01
2.25314289e-01 2.61440843e-01 2.76890486e-01 1.65051237e-01
-2.96027120e-02 -3.13067019e-01 -1.96003541e-01 7.35930324e-01
-2.11674362e-01 3.47260498e-02 5.29644787e-01 5.07881105e-01
1.79951847e+00 -5.81783712e-01 2.93435812e-01 -6.17016017e-01
3.22845966e-01 -3.35501015e-01 3.61790180e-01 1.10609925e+00
-3.43814313e-01 8.48303437e-02 1.49535167e+00 6.61498830e-02
-1.10605085e+00 -1.08673775e+00 -1.64174438e-01 9.02827203e-01
4.38405573e-01 -7.23601997e-01 -9.19507682e-01 -5.43900371e-01
1.73270449e-01 2.35985979e-01 -9.17083800e-01 -1.49073273e-01
-3.95877004e-01 -9.62511480e-01 6.61551297e-01 5.38162529e-01
5.39894819e-01 -6.48271739e-01 -1.02095373e-01 -4.16429266e-02
2.03266546e-01 -1.03182304e+00 -4.88202184e-01 3.74592960e-01
-9.14518595e-01 -1.27746964e+00 -4.08235967e-01 -4.76051092e-01
8.30056667e-01 2.17968896e-01 1.02528811e+00 3.18039149e-01
-1.50107637e-01 1.50863454e-01 -1.74712315e-01 -2.53749583e-02
-2.57241398e-01 2.98338979e-01 -6.75550774e-02 -4.65731621e-01
1.86536655e-01 -7.92755604e-01 -9.05094266e-01 2.61718810e-01
-5.93802631e-01 -1.53103814e-01 4.45743114e-01 5.43115318e-01
4.78774279e-01 3.34742218e-01 3.13701421e-01 -9.42807376e-01
1.55659914e-01 -4.91987199e-01 -1.32038367e+00 3.02670505e-02
-8.77548277e-01 2.19354853e-01 8.85309458e-01 -1.21248014e-01
4.36295792e-02 -2.39554644e-01 -1.77475195e-02 -1.33646457e-02
5.45863271e-01 3.59314620e-01 2.99703013e-02 -4.46199536e-01
5.24755657e-01 1.19469918e-01 -1.60709992e-01 -1.91186294e-01
3.03459525e-01 3.22632194e-01 4.82965307e-03 -2.85530120e-01
8.30829322e-01 6.79635346e-01 7.50545263e-01 -7.21751153e-01
-6.20329380e-01 -8.61834586e-02 -3.41994733e-01 -9.78552625e-02
5.65609753e-01 -6.50978386e-01 -1.35750389e+00 1.73493475e-01
-6.03696823e-01 -6.49391711e-01 -5.96308447e-02 4.18612719e-01
-5.68514884e-01 5.04597664e-01 -6.75427437e-01 -1.22590196e+00
-5.86123578e-02 -8.60170901e-01 4.55554843e-01 -6.55595735e-02
2.63908389e-03 -9.07666326e-01 -2.88147479e-01 2.63499588e-01
2.71222502e-01 3.31229359e-01 1.01888585e+00 -3.15102011e-01
-8.81402433e-01 -5.85577726e-01 -6.90219283e-01 3.47035557e-01
-2.27305964e-01 -7.75349960e-02 -3.17429513e-01 -3.28116447e-01
-3.66334379e-01 2.54678488e-01 8.39824378e-01 5.07126749e-01
1.33430982e+00 -5.34296811e-01 -5.76847970e-01 4.63733792e-01
1.63580239e+00 -2.15447880e-02 6.96634114e-01 -1.28932670e-01
5.22325814e-01 1.45408571e-01 1.03106737e-01 3.49569261e-01
1.94147587e-01 5.33587277e-01 4.95149076e-01 5.03344595e-01
2.22541355e-02 -3.47671956e-01 1.70833662e-01 7.87859499e-01
-1.22948401e-01 -5.11593163e-01 -7.94844151e-01 6.57490373e-01
-1.58854854e+00 -8.14514279e-01 -4.21231866e-01 2.58106422e+00
7.09817529e-01 5.93350112e-01 3.30451757e-01 2.92677432e-01
9.76589680e-01 2.85770506e-01 -4.56210136e-01 -3.73198807e-01
-1.54891387e-02 7.51752138e-01 1.49680769e+00 7.69837022e-01
-5.12170136e-01 6.18497610e-01 6.05002880e+00 1.22491741e+00
-5.02916098e-01 1.12085521e-01 5.75326383e-01 -1.88379347e-01
-3.55746180e-01 4.05989230e-01 -6.93030238e-01 9.71144319e-01
1.23834991e+00 -2.38614216e-01 7.82436907e-01 8.47333133e-01
-8.84920359e-02 -4.63127494e-01 -8.05676103e-01 8.69642556e-01
-9.38629061e-02 -1.17531943e+00 -6.05857909e-01 7.24670827e-01
8.83987248e-01 -7.80228451e-02 -7.01586828e-02 -4.50312011e-02
8.87157500e-01 -1.12756300e+00 4.28556204e-01 -5.61053343e-02
1.04337275e+00 -9.30338264e-01 5.84646285e-01 3.26806933e-01
-1.48460317e+00 -1.26972035e-01 -3.51081550e-01 -5.40822111e-02
4.40009423e-02 9.08746660e-01 -3.76223683e-01 3.09230238e-01
6.35730565e-01 4.26980495e-01 -1.94100991e-01 8.53817105e-01
-3.10182244e-01 8.16202223e-01 -8.81019294e-01 -3.92189175e-01
5.91955781e-02 -3.86104107e-01 5.20538092e-01 9.07312572e-01
2.81761676e-01 3.85487825e-01 -2.01846361e-01 4.90534037e-01
-5.01236737e-01 -8.90978053e-02 -4.25406873e-01 -9.62135494e-02
7.54708648e-01 9.13650036e-01 -1.11270547e+00 -4.75714654e-02
-1.08774528e-01 5.63227177e-01 3.16143125e-01 1.42847165e-01
-8.66043985e-01 -8.98087382e-01 5.53079844e-01 6.55196309e-01
9.07998681e-01 -5.27059197e-01 -5.86950541e-01 -8.64416957e-01
2.05953375e-01 -4.35211271e-01 2.24730477e-01 -2.23420829e-01
-1.02928758e+00 2.75687516e-01 6.12304583e-02 -8.05023491e-01
-6.55927584e-02 -8.01797807e-01 -2.59505033e-01 6.15492284e-01
-9.70213532e-01 -5.98612666e-01 -4.65660878e-02 1.81114957e-01
-4.20093834e-01 2.98381597e-01 4.20840889e-01 2.96706200e-01
-4.69415188e-01 6.99342310e-01 6.51038289e-01 3.03012162e-01
4.63012755e-02 -1.23348844e+00 1.92873120e-01 9.30945039e-01
3.12589407e-02 3.19712520e-01 9.51817751e-01 -5.64106226e-01
-1.43965626e+00 -7.01287925e-01 6.31150305e-01 -2.95874566e-01
9.15217876e-01 -7.38375127e-01 -4.05733228e-01 8.05249631e-01
-1.75294444e-01 1.63673654e-01 5.86214244e-01 2.03269914e-01
-4.45472360e-01 -4.66504127e-01 -1.26405358e+00 5.72634101e-01
1.48360205e+00 -4.12065744e-01 4.19069231e-01 4.13623363e-01
4.07959819e-01 -1.05409868e-01 -1.17508888e+00 4.08706993e-01
6.77761376e-01 -1.13131499e+00 6.56903386e-01 -3.81748259e-01
-5.15967002e-03 -1.41638890e-01 -3.20448995e-01 -7.30692744e-01
-1.12346381e-01 -8.96545649e-01 -3.64083439e-01 6.47163630e-01
6.51557744e-01 -6.90522015e-01 1.07985115e+00 3.08899581e-01
4.23950493e-01 -9.15611327e-01 -1.24483180e+00 -1.08463979e+00
4.04120564e-01 -6.44844234e-01 3.04323167e-01 3.18203241e-01
2.20663711e-01 1.32933790e-02 -1.90718502e-01 5.25977984e-02
9.20553625e-01 -1.34389967e-01 6.67568862e-01 -1.21931159e+00
-5.49704909e-01 -7.11209059e-01 -7.86755145e-01 -1.25094795e+00
8.18135440e-02 -8.02685618e-01 -1.43182397e-01 -1.40883577e+00
4.91699040e-01 -8.09819758e-01 -2.54833043e-01 9.03466940e-02
2.42895588e-01 4.36919481e-01 6.95113540e-02 -7.38657042e-02
-7.99738109e-01 1.03670530e-01 9.56034422e-01 1.95248932e-01
2.93426812e-01 6.33739755e-02 -8.30726206e-01 6.00621700e-01
5.58680415e-01 -6.51902854e-01 -1.27341092e-01 -1.52839884e-01
1.02611101e+00 3.31288546e-01 2.76255250e-01 -8.84981394e-01
1.61148012e-02 1.74940955e-02 -2.61306181e-03 -3.88274908e-01
2.90084332e-01 -6.04632139e-01 2.09532559e-01 6.48306489e-01
-3.19654703e-01 1.46401882e-01 -1.83064505e-01 1.04570043e+00
3.61548066e-01 -2.89872140e-01 9.47532713e-01 2.63066944e-02
1.75946131e-01 3.65482241e-01 -2.64830977e-01 1.21472336e-01
1.16796696e+00 -1.82516009e-01 -5.07998824e-01 -9.57846463e-01
-6.05038345e-01 3.05742621e-01 4.69762325e-01 -3.66361052e-01
-1.03267983e-01 -1.06242406e+00 -4.15226400e-01 -5.90408370e-02
-1.47887859e-02 -1.04816571e-01 3.47377360e-01 1.23268175e+00
-7.72246838e-01 4.41455424e-01 3.29483539e-01 -4.31799799e-01
-8.67143750e-01 4.07277912e-01 4.23639566e-01 -6.73496842e-01
-1.00903034e-01 1.01204646e+00 -8.95396248e-02 1.65738776e-01
5.89505024e-02 -1.51166126e-01 7.02512264e-01 -3.20999265e-01
2.35990584e-01 8.64088178e-01 -2.26327963e-02 -3.98889750e-01
-4.12173569e-01 5.13395250e-01 -5.32804877e-02 -3.75468254e-01
1.13349664e+00 -2.99259603e-01 -4.40890938e-01 2.53169715e-01
1.25671923e+00 3.61524761e-01 -1.12622571e+00 -1.22597264e-02
-2.18298018e-01 -4.48187649e-01 -3.69641006e-01 -4.21310663e-01
-1.13573062e+00 7.45466530e-01 3.42460096e-01 1.05615449e+00
7.39547670e-01 3.95998567e-01 5.25007188e-01 1.46140605e-01
8.32914054e-01 -1.01677012e+00 -3.31582576e-01 3.60582769e-01
2.23107174e-01 -9.03597236e-01 1.79335028e-01 -6.98298931e-01
3.35839055e-02 8.07252169e-01 3.74070942e-01 -5.37997186e-01
1.18975878e+00 1.29241675e-01 -7.06095994e-01 -1.06928296e-01
-6.25787079e-01 -1.09910764e-01 -1.37305990e-01 4.14047837e-01
3.99605483e-01 2.37868249e-01 -6.45779192e-01 4.96081263e-01
-4.92095381e-01 -3.34371001e-01 6.26145840e-01 5.29504478e-01
-6.29368007e-01 -1.15089524e+00 1.03909828e-01 8.85869086e-01
-5.78195155e-01 -2.07619786e-01 -2.28306144e-01 1.05365539e+00
-4.58423570e-02 9.06839430e-01 1.44091129e-01 -3.35648477e-01
-3.77711244e-02 -3.24520320e-01 8.63161087e-01 -4.02707309e-01
-2.84540892e-01 -2.40078285e-01 5.49433045e-02 -3.13561082e-01
-9.28693637e-02 -4.44296032e-01 -1.00940943e+00 -1.16464853e+00
-5.35018563e-01 3.88616055e-01 5.15743732e-01 8.43164444e-01
2.04583347e-01 4.93334904e-02 6.00338101e-01 -9.90432203e-02
-4.57544178e-01 -1.02937698e+00 -1.10995317e+00 -8.45568627e-02
5.10902889e-02 -4.97132093e-01 -9.34595406e-01 -3.53889018e-01] | [6.736847400665283, 5.022206783294678] |
aee928db-b4fb-4bda-9fcc-b6ce10af6850 | single-image-object-counting-and-localizing | 2111.08383 | null | https://arxiv.org/abs/2111.08383v1 | https://arxiv.org/pdf/2111.08383v1.pdf | Single Image Object Counting and Localizing using Active-Learning | The need to count and localize repeating objects in an image arises in different scenarios, such as biological microscopy studies, production lines inspection, and surveillance recordings analysis. The use of supervised Convoutional Neural Networks (CNNs) achieves accurate object detection when trained over large class-specific datasets. The labeling effort in this approach does not pay-off when the counting is required over few images of a unique object class. We present a new method for counting and localizing repeating objects in single-image scenarios, assuming no pre-trained classifier is available. Our method trains a CNN over a small set of labels carefully collected from the input image in few active-learning iterations. At each iteration, the latent space of the network is analyzed to extract a minimal number of user-queries that strives to both sample the in-class manifold as thoroughly as possible as well as avoid redundant labels. Compared with existing user-assisted counting methods, our active-learning iterations achieve state-of-the-art performance in terms of counting and localizing accuracy, number of user mouse clicks, and running-time. This evaluation was performed through a large user study over a wide range of image classes with diverse conditions of illumination and occlusions. | ['Raanan Fattal', 'Inbar Huberman-Spiegelglas'] | 2021-11-16 | null | null | null | null | ['object-counting'] | ['computer-vision'] | [ 4.75919873e-01 -2.25468472e-01 9.74861719e-03 -3.72791678e-01
-7.13147759e-01 -7.72206068e-01 3.91039819e-01 4.87998694e-01
-1.23184013e+00 5.01811266e-01 -6.27582371e-01 -7.19982460e-02
2.23054990e-01 -5.93077064e-01 -7.28758395e-01 -7.40550280e-01
6.57953024e-02 5.93196630e-01 4.22672719e-01 6.03072584e-01
4.49790299e-01 8.28565657e-01 -1.47951233e+00 9.59087454e-04
6.12201273e-01 7.72485852e-01 5.18842518e-01 8.69206786e-01
-2.96510965e-01 9.28309739e-01 -8.20096970e-01 -1.97618604e-01
-4.98445630e-02 -1.09087221e-01 -6.67663693e-01 3.83222580e-01
7.38863051e-01 -4.12812442e-01 3.72310102e-01 1.09275520e+00
3.74311656e-01 7.61687979e-02 6.19462609e-01 -8.68157923e-01
-4.43588108e-01 -6.25303239e-02 -6.36922061e-01 6.69376075e-01
-1.56663328e-01 3.31453294e-01 6.30705416e-01 -1.15715396e+00
5.79173028e-01 9.69233096e-01 5.89757919e-01 5.31295776e-01
-1.58077765e+00 -8.20494711e-01 -4.38789017e-02 -1.95182860e-01
-1.50486636e+00 -4.08770442e-01 4.33974504e-01 -5.12140751e-01
8.21680784e-01 7.94935748e-02 6.47843301e-01 7.37504005e-01
-1.39612332e-01 7.39450932e-01 7.49395192e-01 -6.54223442e-01
5.32091379e-01 2.20565423e-01 3.01023215e-01 1.13138914e+00
4.49158430e-01 -3.90723020e-01 -5.51518261e-01 -3.54596287e-01
9.94141519e-01 2.72920161e-01 2.21547306e-01 -4.34795737e-01
-1.24078596e+00 7.56000996e-01 1.42013192e-01 2.62610912e-01
-1.57645449e-01 2.43704244e-01 3.80356699e-01 -1.12182051e-01
7.82045186e-01 5.65802932e-01 -3.78978997e-01 1.31090835e-01
-1.14693117e+00 -1.46855533e-01 7.22259998e-01 9.54544008e-01
8.55659962e-01 -3.00855756e-01 -3.97604942e-01 7.63184726e-01
-6.73122779e-02 4.32432353e-01 -9.26550329e-02 -1.07075822e+00
2.12592766e-01 9.04652834e-01 3.88141185e-01 -9.09379721e-01
-3.44803780e-01 -4.25252318e-01 -8.48669827e-01 2.88819045e-01
7.81847894e-01 5.99544756e-02 -9.31566119e-01 1.49988103e+00
4.34607625e-01 9.06581134e-02 -6.64512932e-01 5.12963593e-01
4.53135818e-01 4.11431670e-01 2.43431166e-01 -2.84756571e-01
1.39973283e+00 -8.44391763e-01 -5.26505053e-01 -3.58216882e-01
7.55523801e-01 -7.03048289e-01 1.24080527e+00 3.24448943e-01
-8.75337362e-01 -5.63282490e-01 -9.44571257e-01 -2.06319794e-01
-6.18841887e-01 7.97079742e-01 6.17985547e-01 5.93540013e-01
-8.10269892e-01 4.85303909e-01 -1.04049790e+00 -4.72119272e-01
9.77414906e-01 5.65633774e-01 -2.32032925e-01 1.12035032e-02
-2.78162003e-01 6.85830414e-01 2.54045278e-01 1.47348672e-01
-1.13112426e+00 -4.93023783e-01 -6.11456633e-01 2.08939776e-01
5.19048870e-01 -2.87010908e-01 1.14041674e+00 -6.95469439e-01
-1.04639661e+00 1.20629501e+00 -4.08054680e-01 -5.84972873e-02
5.09824336e-01 -2.14647308e-01 2.35229388e-01 2.09388956e-01
3.43273699e-01 8.93792748e-01 5.47811925e-01 -9.58031535e-01
-5.80974936e-01 -4.49912578e-01 8.30681697e-02 -6.72236457e-02
-5.86233258e-01 2.31479466e-01 -6.75260603e-01 -1.85905591e-01
-1.21353328e-01 -7.44072080e-01 -2.36433849e-01 5.79076469e-01
-3.41050923e-01 -5.13671398e-01 1.02062881e+00 -2.87906099e-02
7.07734406e-01 -2.01992679e+00 -3.08524221e-01 -1.53714597e-01
6.15844429e-01 5.08128643e-01 -1.30882010e-01 9.99965593e-02
1.89407960e-01 2.01264191e-02 -1.18816316e-01 -9.96234357e-01
-4.24677789e-01 8.31144676e-02 3.39544341e-02 8.76176476e-01
4.95773464e-01 7.73697853e-01 -1.06975389e+00 -1.09089887e+00
4.45426852e-01 4.28465486e-01 -3.11141193e-01 4.11772102e-01
-2.40188718e-01 7.47562230e-01 -8.68549570e-02 7.96059251e-01
4.90350515e-01 -9.93572414e-01 -2.12349053e-02 -2.19478384e-02
-2.01097682e-01 -2.63666481e-01 -1.06624484e+00 1.57105649e+00
-4.66829121e-01 7.95284986e-01 -2.87924767e-01 -9.89542127e-01
7.11595654e-01 3.80665436e-02 3.17206264e-01 -4.87415940e-01
3.73060286e-01 1.13754198e-01 -3.44191134e-01 -6.50641620e-01
1.47408232e-01 4.78656471e-01 1.84289977e-01 6.04645550e-01
3.67613107e-01 2.67195255e-01 8.77823234e-01 1.66565493e-01
1.17996609e+00 -1.15957804e-01 4.64938551e-01 -3.60072196e-01
4.03099626e-01 -4.48534340e-02 4.27464038e-01 1.10862207e+00
-3.61177504e-01 4.25891399e-01 2.76843369e-01 -7.71538615e-01
-1.19999766e+00 -7.71717191e-01 -2.60943592e-01 1.44037676e+00
4.23870161e-02 4.43873852e-02 -8.53284538e-01 -6.11467898e-01
-4.28488791e-01 3.20534468e-01 -7.76888311e-01 3.25630009e-01
-6.59413338e-01 -6.44637883e-01 4.94374692e-01 5.21680951e-01
5.48571229e-01 -1.25196815e+00 -1.09569192e+00 6.46135211e-02
-1.32438347e-01 -1.27602720e+00 -4.59580332e-01 3.98467511e-01
-7.93908894e-01 -1.37905872e+00 -7.83318639e-01 -9.28648055e-01
1.22651160e+00 2.55669594e-01 1.22091281e+00 3.25701445e-01
-8.58110785e-01 2.76482314e-01 -2.34294571e-02 -6.91522539e-01
1.86028853e-02 1.89184278e-01 -2.81291157e-01 -3.47589180e-02
8.13237131e-01 -1.79692045e-01 -7.11154401e-01 2.90977806e-01
-7.48870194e-01 -2.37001553e-01 4.57541943e-01 7.72657156e-01
6.39330447e-01 -2.72361726e-01 1.69697627e-01 -9.30855215e-01
3.20622176e-01 -8.78255591e-02 -1.16606069e+00 4.70577419e-01
-1.52853221e-01 -2.20336869e-01 6.92840695e-01 -1.01985765e+00
-5.91190577e-01 4.43458468e-01 3.94825429e-01 -4.23642546e-01
-2.71253526e-01 -1.26698539e-01 2.51521975e-01 -2.17699394e-01
9.24412668e-01 2.80101955e-01 -1.32835522e-01 -1.86248615e-01
2.70427261e-02 4.86468822e-01 4.13937658e-01 -2.94364423e-01
4.81370181e-01 8.82645965e-01 7.40515664e-02 -9.29675937e-01
-1.21350813e+00 -1.04582179e+00 -8.83535147e-01 -4.91555959e-01
9.67713296e-01 -6.06797218e-01 -1.29517293e+00 4.89857167e-01
-1.54553854e+00 -3.61366570e-01 -1.66419312e-01 4.45260555e-01
-3.75143766e-01 3.93096179e-01 -5.50849259e-01 -1.27633595e+00
-4.86127049e-01 -1.17649519e+00 1.41908991e+00 2.92401850e-01
-3.57676268e-01 -8.47390234e-01 6.39823750e-02 1.14931520e-02
3.87107879e-01 1.47860557e-01 7.32685149e-01 -7.31137216e-01
-9.30941164e-01 -6.25296116e-01 -4.18117851e-01 2.52683014e-01
1.17403656e-01 -3.08414409e-03 -1.15191793e+00 -4.23136085e-01
-2.00462103e-01 -6.66794002e-01 6.36437654e-01 3.30015779e-01
1.66316712e+00 -1.35734677e-01 -3.35918069e-01 3.73904765e-01
1.45803761e+00 1.17988393e-01 4.25225168e-01 -7.08257407e-02
5.20559669e-01 5.03593624e-01 4.29906815e-01 1.80629686e-01
-2.47205183e-01 2.67832279e-01 4.41102505e-01 -5.80530345e-01
2.63596624e-01 1.20990500e-01 -3.59575391e-01 3.62523764e-01
-1.28032848e-01 -3.83766115e-01 -9.44367707e-01 6.46659315e-01
-1.62124741e+00 -8.52096736e-01 1.85598001e-01 2.41749644e+00
6.75668418e-01 1.95195928e-01 1.32589743e-01 1.72915906e-01
1.02812815e+00 -2.00140730e-01 -6.93079531e-01 2.01894313e-01
1.82375640e-01 1.50260091e-01 6.88050926e-01 3.23096752e-01
-1.24430621e+00 7.48334348e-01 6.48490381e+00 8.95605564e-01
-1.08728993e+00 2.21038878e-01 9.21492934e-01 -2.55605280e-01
6.98752880e-01 -1.44278660e-01 -1.02420950e+00 3.89180720e-01
3.95354748e-01 4.49677050e-01 2.81982183e-01 1.08353162e+00
4.55325395e-02 -4.82871324e-01 -1.31353509e+00 1.27891886e+00
1.26279369e-01 -1.42991066e+00 -2.14111090e-01 7.73032829e-02
3.45052361e-01 2.29765959e-02 -1.89296216e-01 2.20587421e-02
2.62476563e-01 -9.56791520e-01 6.59788489e-01 4.96427357e-01
9.87244666e-01 -5.03954470e-01 7.28448570e-01 8.21118057e-01
-9.81993020e-01 -4.92909886e-02 -5.25214136e-01 1.08570531e-01
9.14470479e-03 7.99255252e-01 -8.89441729e-01 -4.15837288e-01
7.70682633e-01 2.02718347e-01 -6.16742194e-01 1.12732649e+00
9.70561653e-02 5.46301842e-01 -5.65527320e-01 -5.65318048e-01
-1.26054566e-02 -6.38626441e-02 2.06136912e-01 1.55569136e+00
-1.58859696e-02 9.00678933e-02 2.31097981e-01 1.14112103e+00
-2.08769456e-01 9.82458144e-02 -5.61282098e-01 -3.05487931e-01
5.69824815e-01 1.52935863e+00 -1.54360855e+00 -5.47761619e-01
-1.55824304e-01 9.28525388e-01 8.63863230e-01 2.76086777e-01
-6.80549860e-01 -5.42782366e-01 -9.52140540e-02 1.67613760e-01
2.83571690e-01 -4.99729544e-01 -1.91789374e-01 -9.96562481e-01
-5.30797541e-02 -2.50492275e-01 2.38699973e-01 -6.36466563e-01
-1.07727230e+00 2.96241909e-01 6.57578334e-02 -1.07454646e+00
1.51789993e-01 -7.94523954e-01 -6.72683656e-01 6.14339948e-01
-1.16980422e+00 -9.37879205e-01 -6.95490539e-01 3.16329062e-01
6.92898929e-01 -7.60337636e-02 9.64202106e-01 4.35505807e-01
-5.66036701e-01 5.21532118e-01 -8.25222805e-02 5.82834065e-01
3.61243606e-01 -1.06236064e+00 1.87701806e-01 9.04548883e-01
1.78651839e-01 9.70010877e-01 4.37685698e-01 -3.81359011e-01
-9.00496006e-01 -1.00539887e+00 9.89563882e-01 -6.12567961e-01
2.49845743e-01 -9.84164178e-01 -6.33196354e-01 4.75055456e-01
-1.08529821e-01 5.88700950e-01 6.99716270e-01 1.38938576e-01
-2.99517006e-01 7.15689808e-02 -1.27134991e+00 3.73331547e-01
9.53134954e-01 -5.57383597e-01 6.51672930e-02 7.36218214e-01
3.10514271e-01 -1.36446476e-01 -2.99932212e-01 1.12146474e-01
4.35638070e-01 -7.72327006e-01 8.60534370e-01 -4.33897316e-01
2.87377477e-01 -3.13758165e-01 1.33625597e-01 -5.17214179e-01
-3.35711926e-01 -2.34454691e-01 -7.72283077e-02 8.57390881e-01
3.32043111e-01 -3.83567899e-01 1.07029617e+00 2.89151579e-01
2.58262098e-01 -7.58145273e-01 -7.68364787e-01 -7.13731349e-01
-3.62553269e-01 -3.14601585e-02 1.32855773e-01 8.92328560e-01
-3.89689505e-01 5.19896090e-01 -5.74976504e-02 5.83086200e-02
1.13628089e+00 -1.78460166e-01 9.89309430e-01 -1.26622641e+00
-6.92750961e-02 -2.93341037e-02 -5.20905256e-01 -1.07724929e+00
-1.58284843e-01 -5.50134301e-01 8.00514817e-02 -1.29445159e+00
7.13010728e-01 -2.39829645e-01 2.14847624e-02 5.26760161e-01
-9.99388322e-02 6.30108953e-01 -2.91192867e-02 4.43555087e-01
-1.15053689e+00 2.45357547e-02 1.11860371e+00 -1.87926844e-01
9.10253748e-02 -9.82854292e-02 -8.92965589e-03 8.72639835e-01
6.58291817e-01 -8.30044866e-01 -2.63901681e-01 -5.38178504e-01
3.04785311e-01 -1.66823134e-01 6.60047889e-01 -1.04772651e+00
6.28476560e-01 -1.31852403e-01 7.65507102e-01 -9.27818358e-01
4.57689613e-01 -6.85838342e-01 -2.96284050e-01 5.33330142e-01
-4.92280900e-01 -1.97110191e-01 -7.09338188e-02 6.42835915e-01
1.88068151e-01 -4.59837765e-01 1.21868110e+00 -5.54778457e-01
-3.55662793e-01 2.47422457e-01 -3.66284460e-01 -4.21387069e-02
9.22053397e-01 -6.02537334e-01 -2.88383216e-01 1.13240648e-02
-7.31332421e-01 2.43959054e-02 3.10916781e-01 -7.14760274e-02
2.75514543e-01 -8.23787451e-01 -3.99043888e-01 5.66314571e-02
4.33006257e-01 4.10616845e-01 -4.91889147e-03 7.07574427e-01
-7.58843958e-01 4.08408612e-01 1.52114742e-02 -1.15923142e+00
-1.36900592e+00 5.58582485e-01 3.59776825e-01 -4.30282593e-01
-4.07315910e-01 9.34712827e-01 2.83593476e-01 -5.54548442e-01
5.37817597e-01 -2.57072538e-01 -2.57129043e-01 1.38260964e-02
7.06975758e-01 5.31382680e-01 1.14533707e-01 -1.18705593e-01
-3.29469115e-01 4.50803310e-01 -2.52904564e-01 2.02282757e-01
1.15018618e+00 -1.79979485e-03 1.47876777e-02 7.38999188e-01
1.46179819e+00 -1.97418720e-01 -1.23768377e+00 -4.12444085e-01
4.52338206e-03 -5.38730204e-01 -1.79369882e-01 -4.13482517e-01
-9.15312350e-01 9.66209531e-01 9.05189574e-01 2.57606030e-01
7.33989656e-01 1.33184224e-01 1.81154191e-01 8.32807124e-01
5.29202938e-01 -1.21368802e+00 6.36970043e-01 4.48200732e-01
3.34534526e-01 -1.63148487e+00 1.66605815e-01 -4.35920328e-01
-1.91764981e-01 1.09109688e+00 7.17880309e-01 -2.81746030e-01
4.08229262e-01 4.29736733e-01 6.44579604e-02 -6.01950645e-01
-4.37452704e-01 -1.71080559e-01 -1.49443030e-01 4.33837146e-01
4.38747376e-01 -2.38705352e-01 2.96671204e-02 -3.00788552e-01
3.16465199e-01 2.19125256e-01 3.23402941e-01 1.18836808e+00
-6.84778333e-01 -3.73125672e-01 -2.86766857e-01 6.11868024e-01
-5.29031098e-01 2.68372800e-02 -2.93400019e-01 7.30410695e-01
2.66095549e-01 8.00699115e-01 6.18873119e-01 3.48084509e-01
-4.15071808e-02 -1.16867974e-01 5.38218439e-01 -9.91107285e-01
-2.76786208e-01 -1.74364820e-02 -2.53499299e-01 -3.84317100e-01
-7.57258117e-01 -3.10258746e-01 -9.08963621e-01 -1.69575483e-01
-9.50148463e-01 -1.99117512e-01 6.90996766e-01 1.05581605e+00
-1.44271757e-02 2.43491173e-01 4.13058043e-01 -1.04121459e+00
-3.71628642e-01 -1.15049171e+00 -5.09232223e-01 3.99309665e-01
2.70517975e-01 -6.46647573e-01 -5.54710329e-01 3.54184121e-01] | [9.052270889282227, 0.528303325176239] |
370856c9-80ff-4f2e-99c6-83f33c5cd669 | sense-disambiguation-of-compound-constituents | 2204.00429 | null | https://arxiv.org/abs/2204.00429v1 | https://arxiv.org/pdf/2204.00429v1.pdf | Sense disambiguation of compound constituents | In distributional semantic accounts of the meaning of noun-noun compounds (e.g. starfish, bank account, houseboat) the important role of constituent polysemy remains largely unaddressed(cf. the meaning of star in starfish vs. star cluster vs. star athlete). Instead of semantic vectors that average over the different meanings of a constituent, disambiguated vectors of the constituents would be needed in order to see what these more specific constituent meanings contribute to the meaning of the compound as a whole. This paper presents a novel approach to this specific problem of word sense disambiguation: set expansion. We build on the approach developed by Mahabal et al. (2018) which was originally designed to solve the analogy problem. We modified their method in such a way that it can address the problem of sense disambiguation of compound constituents. The results of experiments with a data set of almost 9000 compounds (LADEC, Gagn\'e et al. 2019) suggest that this approach is successful, yet the success is sensitive to the frequency with which the compounds are attested. | ['Ingo Plag', 'Stefan Conrad', 'Carlo Schackow'] | 2022-04-01 | null | null | null | null | ['word-sense-disambiguation'] | ['natural-language-processing'] | [ 3.12969126e-02 -1.06453128e-01 -4.38371114e-02 -1.44447029e-01
-1.95744082e-01 -1.19718480e+00 8.49724174e-01 6.17066622e-01
-7.56378651e-01 5.55207849e-01 6.24489486e-01 -3.78342837e-01
-2.90702730e-01 -8.85223091e-01 -3.00768554e-01 -7.12919176e-01
2.75424808e-01 5.73064804e-01 4.95517582e-01 -7.39114642e-01
4.82091188e-01 3.27694505e-01 -1.59803629e+00 9.12567824e-02
5.10421276e-01 4.87212002e-01 5.17774165e-01 8.86361003e-02
-5.26894569e-01 9.70401317e-02 -6.68406010e-01 -5.12656808e-01
2.63320744e-01 -3.41425031e-01 -9.34354603e-01 3.85386981e-02
3.84990811e-01 4.86365348e-01 1.05431944e-01 1.32041180e+00
4.17089015e-01 2.41879910e-01 7.30897427e-01 -8.62828553e-01
-5.46525359e-01 1.14490342e+00 -2.76271343e-01 5.39409518e-01
3.35190564e-01 -6.34830222e-02 1.72934163e+00 -8.95059288e-01
6.76968455e-01 1.35632241e+00 3.78685176e-01 5.30869901e-01
-1.28770339e+00 -6.42274320e-01 3.13678116e-01 1.87520176e-01
-1.43548644e+00 -1.52656630e-01 4.32730854e-01 -4.69185323e-01
1.06852043e+00 2.82550156e-01 8.69201362e-01 8.75120580e-01
-6.13345541e-02 4.09896314e-01 1.09658265e+00 -5.37408352e-01
3.35158497e-01 -2.11473089e-02 4.80368197e-01 -1.52978510e-01
8.07932317e-01 -7.40582943e-02 -3.82013530e-01 -2.08545730e-01
3.44367325e-01 -2.27069363e-01 -1.71299204e-01 -3.04975033e-01
-1.13425601e+00 1.07819474e+00 3.43896091e-01 9.31084514e-01
-1.66733935e-01 -4.03504912e-03 4.59098786e-01 6.86801672e-02
3.59240711e-01 8.15812230e-01 -6.70172870e-01 5.05332928e-03
-5.92941105e-01 8.83840322e-01 8.34708810e-01 6.95502520e-01
7.12642908e-01 -1.35941193e-01 1.07338049e-01 7.38124907e-01
4.63614374e-01 2.20795572e-01 8.18598449e-01 -6.25522316e-01
1.59428671e-01 5.47794402e-01 7.40620866e-02 -8.14419329e-01
-3.75300229e-01 -1.69756398e-01 -6.34739362e-03 -7.14612529e-02
4.92242336e-01 3.10779065e-02 -7.84440279e-01 1.90915227e+00
4.21064913e-01 -2.65219454e-02 3.34127873e-01 9.47153568e-01
7.84282386e-01 4.05845374e-01 6.27912462e-01 -9.64024663e-02
1.81188858e+00 -2.37291649e-01 -6.80669904e-01 -5.30500054e-01
6.07364357e-01 -8.31587732e-01 1.35256433e+00 1.73358127e-01
-7.76765049e-01 -1.59556463e-01 -1.05408967e+00 9.09357890e-02
-7.71544218e-01 -5.68091273e-01 8.94260108e-01 7.61091709e-01
-8.05930853e-01 5.47311008e-01 -4.06789124e-01 -8.09942305e-01
-3.05167474e-02 3.40173066e-01 -2.58038968e-01 1.01225734e-01
-1.41277969e+00 1.21072984e+00 9.98133361e-01 -1.28426060e-01
-3.41474801e-01 -5.45449972e-01 -9.57304597e-01 -4.48603444e-02
5.29342830e-01 -4.01487619e-01 1.12204242e+00 -9.94701147e-01
-8.92565370e-01 1.10833740e+00 -7.86581263e-02 -2.98660547e-01
-2.90618092e-01 -2.05405280e-01 -4.17834193e-01 -2.13886261e-01
5.04291534e-01 4.87317830e-01 2.98188746e-01 -1.12491584e+00
-7.20232666e-01 -5.88669002e-01 1.86727718e-01 3.16339821e-01
-2.00927287e-01 5.80350459e-01 1.86826158e-02 -8.64783525e-01
2.75283396e-01 -1.03569138e+00 -3.40554416e-02 -7.71183968e-01
-3.23603638e-02 -6.64907813e-01 3.94546241e-01 -1.95067853e-01
1.25056231e+00 -2.22867393e+00 2.87945211e-01 1.68964580e-01
2.03588903e-01 1.67613894e-01 -1.92877665e-01 7.03384757e-01
-5.61100066e-01 2.68175960e-01 -5.08541405e-01 2.59536505e-01
1.54823557e-01 6.43199861e-01 -2.31654555e-01 4.33672607e-01
2.15280116e-01 7.99005806e-01 -1.11532235e+00 -1.88348144e-01
-5.05702058e-03 2.97012597e-01 -4.28988516e-01 -3.33270490e-01
-1.95832923e-01 -9.75725651e-02 -2.74476975e-01 4.37282443e-01
3.90553802e-01 8.89763832e-02 4.70020950e-01 -2.99828500e-01
-3.86137187e-01 8.37352216e-01 -1.40570736e+00 1.35526335e+00
-1.30461261e-01 4.68673170e-01 -1.40219986e-01 -7.20776558e-01
6.10149443e-01 2.58997709e-01 2.20882311e-01 -4.47273642e-01
1.61635503e-01 5.17660677e-01 6.85248673e-01 -3.20015460e-01
8.49955261e-01 -8.88437629e-01 -3.49908650e-01 2.51500249e-01
2.80568041e-02 -4.19427186e-01 4.46296781e-01 3.04150313e-01
7.77031839e-01 3.64283174e-02 7.57248878e-01 -9.47204292e-01
4.45416659e-01 3.26063573e-01 7.27352440e-01 3.02676797e-01
-3.19303870e-02 5.07055998e-01 2.89689273e-01 -1.46361858e-01
-8.80145371e-01 -1.25849724e+00 -4.97711778e-01 1.16197753e+00
3.20643246e-01 -8.91062021e-01 -5.48954248e-01 -4.17794049e-01
3.20728309e-02 1.11866772e+00 -6.36449099e-01 -5.02129942e-02
-6.39876425e-01 -8.13972592e-01 2.84914136e-01 4.87824023e-01
-2.85267472e-01 -1.17429829e+00 -7.88739026e-01 3.91282171e-01
1.84324123e-02 -9.02828217e-01 -2.86100268e-01 3.50861192e-01
-6.73015416e-01 -1.01128423e+00 -2.34202817e-01 -5.68460941e-01
2.25130305e-01 1.83889329e-01 1.27041423e+00 -4.86021526e-02
-2.49456167e-01 3.24599296e-01 -7.07356632e-01 -7.42508113e-01
-2.56904989e-01 -2.89141476e-01 3.15050095e-01 -2.49226496e-01
9.60172296e-01 -6.10105038e-01 -2.23454282e-01 -5.62587492e-02
-1.29071081e+00 -7.82893181e-01 2.87351221e-01 6.07719719e-01
5.00458300e-01 3.98025215e-02 2.82674581e-01 -1.06714237e+00
8.16289961e-01 -6.92463219e-01 -4.39198881e-01 -3.05172522e-02
-6.11638904e-01 2.40903169e-01 1.95406362e-01 -7.01195419e-01
-6.79206967e-01 -1.09601133e-01 -2.83433974e-01 -2.70351786e-02
-1.03471331e-01 6.09493792e-01 -3.74856502e-01 3.02913576e-01
7.58980036e-01 4.90458570e-02 -3.45057786e-01 -6.35660827e-01
4.79591668e-01 3.81631911e-01 3.87860000e-01 -7.29241371e-01
6.45612061e-01 2.05756724e-01 -1.00833498e-01 -7.40076363e-01
-7.94241250e-01 -7.03266382e-01 -6.43667817e-01 1.96557701e-01
1.17904794e+00 -7.95307875e-01 -3.76942396e-01 1.29439691e-02
-1.29002655e+00 2.57403672e-01 -4.79418814e-01 4.80252504e-01
-2.57200629e-01 3.97735447e-01 -3.16234797e-01 -5.52947700e-01
2.92595252e-02 -1.01279902e+00 9.24686015e-01 1.78369728e-03
-8.28262448e-01 -1.05927587e+00 1.75131083e-01 -1.43788621e-01
2.50067413e-01 2.26030484e-01 1.23318100e+00 -1.27066731e+00
-4.20989394e-02 2.41667300e-01 1.44250318e-01 1.45039096e-01
3.81497622e-01 -2.86284953e-01 -7.41376638e-01 -1.09514825e-01
3.79653871e-01 8.77916366e-02 7.84890413e-01 1.37060180e-01
2.48710275e-01 -3.84751499e-01 -2.38845348e-01 2.26330943e-02
1.63736296e+00 1.55764699e-01 4.90798354e-01 4.88382101e-01
6.72506154e-01 1.07966936e+00 7.43896365e-01 8.21309239e-02
3.20717305e-01 7.38922238e-01 4.13731217e-01 5.28898418e-01
-7.82462656e-02 -1.05523817e-01 4.23808187e-01 5.97572446e-01
1.69513330e-01 -3.42750043e-01 -1.01936948e+00 9.92663682e-01
-1.57705081e+00 -8.71170521e-01 -5.30637860e-01 2.19369483e+00
8.11486363e-01 1.37000531e-01 3.11651349e-01 1.95137277e-01
6.57437086e-01 2.27002084e-01 -3.20634851e-03 -6.14414215e-01
-3.25619906e-01 4.91006672e-01 3.83000791e-01 6.52325094e-01
-7.89492011e-01 1.39344406e+00 6.40419292e+00 9.27706003e-01
-8.54120970e-01 1.19723819e-01 -1.70527473e-01 1.20389454e-01
-8.09469104e-01 4.19819295e-01 -8.25681806e-01 4.76253301e-01
7.87701190e-01 -2.43436247e-01 2.78923362e-01 4.32260066e-01
-1.44719347e-01 -2.35677898e-01 -1.15284503e+00 8.33180606e-01
2.04667047e-01 -8.68333340e-01 1.09451786e-01 3.01336963e-02
2.53046066e-01 1.12753548e-01 -2.43435100e-01 -9.06323735e-03
3.91225517e-01 -1.00683165e+00 1.06541502e+00 -7.39755109e-02
3.41718376e-01 -5.95765531e-01 5.94459653e-01 9.98521373e-02
-1.09761596e+00 -4.11280096e-02 -5.52362680e-01 -3.96004170e-01
2.32033923e-01 5.18152535e-01 -6.02370620e-01 4.12407160e-01
6.14518642e-01 5.60299516e-01 -5.49414694e-01 5.82185507e-01
-5.15486121e-01 6.82555318e-01 -4.01253790e-01 -2.62629002e-01
3.95256430e-01 -3.67919862e-01 9.20315444e-01 1.16567183e+00
1.39981031e-01 3.41977507e-01 1.38962325e-02 9.71138537e-01
4.51610714e-01 5.02822459e-01 -6.08951986e-01 -3.24779451e-01
5.90216875e-01 1.18514383e+00 -9.73094761e-01 -2.34316781e-01
-4.59395647e-01 7.73939669e-01 2.05622226e-01 7.73798451e-02
-3.92805308e-01 -2.31021091e-01 1.02724159e+00 2.49524027e-01
4.27594543e-01 -2.63204604e-01 -2.89621204e-01 -8.56139719e-01
-2.39420403e-02 -7.61061549e-01 5.58420956e-01 -4.92125869e-01
-1.61932075e+00 5.01873314e-01 2.59303182e-01 -1.02379572e+00
-2.71550506e-01 -8.80068779e-01 -4.81594384e-01 1.01989162e+00
-1.14998722e+00 -9.70830619e-01 3.98015827e-01 3.86053592e-01
4.27683532e-01 7.71507025e-02 8.65180790e-01 6.41544238e-02
8.43950454e-03 9.69053134e-02 -2.64560610e-01 -1.17617868e-01
6.56962335e-01 -1.45566189e+00 3.46914530e-01 8.70835662e-01
4.25646842e-01 9.38089967e-01 1.23028541e+00 -8.56051564e-01
-1.13414252e+00 -4.80803162e-01 1.47506845e+00 -9.82482374e-01
1.18840778e+00 -3.61903727e-01 -8.94430935e-01 6.10760987e-01
4.26519156e-01 -3.11123192e-01 9.28567231e-01 3.14534068e-01
-6.93783045e-01 2.45208398e-01 -9.75944281e-01 7.53513455e-01
1.11830342e+00 -3.46349061e-01 -1.24521565e+00 9.21083912e-02
1.06492341e+00 6.41691405e-03 -7.26397693e-01 3.02652121e-01
2.68742919e-01 -8.57546091e-01 9.43999946e-01 -8.27754200e-01
2.35477030e-01 -4.85245198e-01 -5.41675687e-01 -1.43791807e+00
-3.33895773e-01 -1.91689193e-01 4.80356842e-01 1.33715093e+00
5.00970244e-01 -6.78895652e-01 1.37318820e-01 5.32562435e-01
-2.87067115e-01 -3.12745035e-01 -9.76304770e-01 -8.78432930e-01
5.15497029e-01 -5.66493988e-01 7.96991110e-01 1.10141969e+00
2.79945403e-01 7.78286934e-01 2.28849053e-01 -1.39600337e-01
2.97460467e-01 1.61115468e-01 5.03041148e-02 -1.42395806e+00
-3.69626462e-01 -6.09043360e-01 -6.91183567e-01 -5.71326971e-01
4.62259859e-01 -1.25637424e+00 -2.24238038e-02 -1.28301513e+00
1.59019887e-01 -4.79109138e-01 -3.01948398e-01 4.88846779e-01
-4.08281088e-01 2.05314472e-01 5.51836669e-01 1.30693659e-01
-2.48795539e-01 1.64161325e-01 9.07051206e-01 -5.04735112e-02
-1.56786263e-01 -3.55690688e-01 -1.26556158e+00 7.09974349e-01
7.99621582e-01 -7.04767525e-01 -3.50020945e-01 -4.82527733e-01
6.00751817e-01 -5.02070248e-01 3.43983471e-01 -4.88640100e-01
-3.51352990e-02 -3.96186918e-01 -7.44640604e-02 -1.60197377e-01
2.36816496e-01 -9.83732283e-01 2.43826538e-01 4.67260480e-01
-2.35059753e-01 5.93472838e-01 2.83423036e-01 4.31823999e-01
-3.24448526e-01 -5.85534394e-01 5.39478421e-01 -4.00078237e-01
-1.00088990e+00 -1.88345835e-01 -5.19937336e-01 4.19628352e-01
8.85841608e-01 -2.38530099e-01 -1.67000666e-01 2.82300506e-02
-6.43813968e-01 -1.36341289e-01 6.17665529e-01 7.28067219e-01
3.90080273e-01 -1.32744658e+00 -7.52176881e-01 -1.43386647e-01
4.48424459e-01 -2.90708333e-01 -2.69768149e-01 5.21704733e-01
-1.84914112e-01 2.32216969e-01 -2.18559802e-02 -2.30366781e-01
-1.12210441e+00 8.92595649e-01 1.09981686e-01 9.95870158e-02
-5.63619614e-01 8.35262597e-01 5.00617266e-01 -2.73131847e-01
-3.60919386e-01 -3.71598542e-01 -4.75632250e-01 4.76156145e-01
4.23274547e-01 1.11397721e-01 1.22752935e-02 -1.22603178e+00
-6.43640041e-01 5.75058997e-01 1.57965913e-01 -3.41638863e-01
1.29326391e+00 -6.25909641e-02 -5.75077176e-01 8.27017546e-01
1.02346563e+00 5.27209938e-01 -3.52748185e-01 -8.28726441e-02
5.38361430e-01 -4.28629905e-01 -3.29838932e-01 -8.25142086e-01
-6.48181558e-01 4.82420981e-01 3.99639726e-01 1.68592915e-01
8.30565095e-01 5.72243929e-01 5.61461568e-01 7.52579570e-02
2.75204301e-01 -8.54716837e-01 -3.73391658e-01 7.73794830e-01
6.90205634e-01 -7.87833214e-01 -1.07879929e-01 -6.56907737e-01
-6.86428130e-01 9.29802299e-01 3.20918322e-01 -2.10158736e-01
4.21013147e-01 1.39750853e-01 8.17637984e-03 -5.99213481e-01
-4.91061836e-01 -7.88552403e-01 2.74988204e-01 5.13501227e-01
7.53769934e-01 4.38968062e-01 -1.33297467e+00 7.51943886e-01
-5.92231095e-01 -7.41608560e-01 6.27303123e-01 8.40819955e-01
-5.63819468e-01 -1.47261965e+00 -5.73778868e-01 3.43325168e-01
-6.33398592e-01 -5.92802405e-01 -8.15431476e-01 9.46208477e-01
3.10805708e-01 9.33716118e-01 1.86425641e-01 -3.43731523e-01
2.98723608e-01 3.29311579e-01 5.17563820e-01 -1.20904136e+00
-8.26924264e-01 1.83596954e-01 2.58494645e-01 -2.48964772e-01
-6.67724192e-01 -9.16379690e-01 -1.62433946e+00 1.24795683e-01
-3.14543843e-01 4.85840321e-01 4.70714450e-01 1.10061538e+00
-1.08767122e-01 1.21791944e-01 -2.53969915e-02 -4.95036989e-01
-4.79012042e-01 -8.91959906e-01 -9.90960836e-01 8.93812239e-01
-5.50852343e-02 -8.65934134e-01 -4.62508827e-01 -1.39845148e-01] | [10.323955535888672, 9.081868171691895] |
001f33bf-c4cc-4921-91d1-7b37d1add753 | a-step-by-step-gradient-penalty-with | null | null | https://link.springer.com/article/10.1007/s11063-022-11031-0 | https://link.springer.com/article/10.1007/s11063-022-11031-0 | A Step-by-Step Gradient Penalty with Similarity Calculation for Text Summary Generation | The summary generation model equipped with gradient penalty avoids overfitting and makes the model more stable. However, the traditional gradient penalty faces two issues: (i) calculating the gradient twice increases training time, and (ii) the disturbance factor requires repeated trials to find the best value. To this end, we propose a step-by-step gradient penalty model with similarity calculation (S2SGP). Firstly, the step-by-step gradient penalty is applied to the summary generation model, effectively reducing the training time without sacrificing accuracy. Secondly, the similarity score between reference and candidate summary is calculated as disturbance factor. To show the performance of our proposed solution, we conduct experiments on four summary generation datasets, among which the EDUSum dataset is newly produced by us. Experimental results show that S2SGP effectively reduces training time, and the disturbance factors do not rely on repeated trials. Especially, our model outperforms the baseline by more than 2.4 ROUGE-L points when tested on the CSL dataset. | ['Shuai Zhao'] | 2022-09-20 | null | null | null | neural-processing-letters-2022-9 | ['abstractive-text-summarization'] | ['natural-language-processing'] | [ 5.02372161e-02 -3.57419074e-01 -3.51134896e-01 -2.93477297e-01
-9.40949380e-01 -4.68717963e-01 3.91263783e-01 3.46958935e-01
-6.03673220e-01 7.81318128e-01 1.59189746e-01 -1.97693914e-01
-1.29906207e-01 -6.89753413e-01 -4.48451370e-01 -4.82937038e-01
2.01707199e-01 1.02914058e-01 5.92232943e-01 -8.69375616e-02
6.27952039e-01 2.01308697e-01 -1.28544188e+00 9.48255584e-02
1.43025291e+00 1.10942984e+00 4.66573089e-01 3.11938345e-01
-1.20805852e-01 4.54327255e-01 -8.06366324e-01 -5.04709840e-01
1.71243161e-01 -5.33271253e-01 -4.74744707e-01 2.93749701e-02
2.92231321e-01 -4.66702670e-01 -1.20637357e-01 9.72045779e-01
5.80024362e-01 4.34787422e-01 4.12356168e-01 -1.22621441e+00
-5.88589966e-01 5.71850896e-01 -5.96157193e-01 -1.02799684e-02
3.62292916e-01 -6.42273426e-02 1.08530498e+00 -1.10480082e+00
4.45641160e-01 1.04827821e+00 5.80743611e-01 2.48312086e-01
-1.01449728e+00 -5.78830838e-01 2.24376753e-01 1.79482549e-01
-1.43830407e+00 -1.62265927e-01 7.14165986e-01 -9.51447934e-02
9.00985599e-01 4.77406114e-01 4.33385313e-01 4.53547806e-01
-1.34406969e-01 1.06759465e+00 6.72373772e-01 -2.77792364e-01
1.97297752e-01 -5.90976700e-02 3.28990072e-01 7.53089368e-01
4.31658655e-01 -2.29143932e-01 -2.66821593e-01 -3.38216692e-01
5.23211420e-01 -1.99632049e-02 -3.83734703e-01 -9.10720676e-02
-1.17929709e+00 8.54602098e-01 3.15187544e-01 2.08360359e-01
-2.02410609e-01 1.36802614e-01 5.09377658e-01 2.19686151e-01
4.53254759e-01 4.69203621e-01 -4.35970783e-01 -3.68327528e-01
-1.14008403e+00 2.87975878e-01 4.88803536e-01 8.46012175e-01
7.45555818e-01 -3.57237607e-02 -4.68932092e-01 1.25830615e+00
9.65231434e-02 3.75292689e-01 7.67517030e-01 -8.85945618e-01
6.24590099e-01 7.91513741e-01 1.63078636e-01 -1.23339641e+00
-2.75753677e-01 -4.99931067e-01 -7.97800243e-01 -3.17549616e-01
2.39316776e-01 -2.21831158e-01 -5.09287059e-01 1.69229388e+00
1.33582413e-01 3.42564493e-01 -2.40799695e-01 7.59602666e-01
9.34559047e-01 8.24518085e-01 -3.28352511e-01 -6.16911113e-01
1.06473005e+00 -1.37897587e+00 -6.57390118e-01 -2.00990915e-01
8.80281031e-01 -8.85698736e-01 1.45601857e+00 2.88143039e-01
-1.13462257e+00 -5.06320059e-01 -1.09983516e+00 -7.13308388e-03
5.85483722e-02 7.70470202e-01 6.18268490e-01 4.59973007e-01
-8.71453285e-01 8.25508118e-01 -6.56930208e-01 -1.91843793e-01
4.57275845e-02 1.86905533e-01 -1.04499862e-01 5.50985336e-02
-1.09591222e+00 4.87667829e-01 7.60491371e-01 -7.79431611e-02
-2.12461412e-01 -6.13876045e-01 -7.03120470e-01 4.20855552e-01
5.09988248e-01 -5.46500564e-01 1.17679763e+00 -5.01696825e-01
-1.54502296e+00 2.90606588e-01 -2.84622997e-01 -1.95981145e-01
6.26594603e-01 -4.13536072e-01 -2.30498627e-01 -5.55258729e-02
1.75145268e-01 4.11999375e-01 5.21308780e-01 -1.05426860e+00
-7.13299572e-01 1.03753516e-02 -4.61172909e-02 1.83171749e-01
-5.10229111e-01 -2.16282934e-01 -8.59497964e-01 -8.92324805e-01
1.68021306e-01 -8.76070976e-01 -1.01104774e-01 -3.18088561e-01
-3.47062021e-01 -4.16761339e-01 5.18972158e-01 -5.24168789e-01
2.12059665e+00 -2.29909134e+00 -2.38527805e-01 7.10581616e-02
-6.82719692e-04 5.49760818e-01 -3.59298408e-01 4.32844222e-01
1.23218164e-01 1.88118264e-01 -2.34890655e-01 -2.40862474e-01
-1.24675659e-02 8.95834644e-04 -1.82087645e-01 -1.10454507e-01
5.77192679e-02 7.02470958e-01 -1.05791831e+00 -5.49706638e-01
-1.71886817e-01 -7.02749193e-02 -8.26006711e-01 1.62672341e-01
-7.59035200e-02 -2.21308231e-01 -4.25378174e-01 2.47922391e-01
6.96215570e-01 -4.88428503e-01 1.98793739e-01 -3.28296930e-01
-1.46811739e-01 5.33983886e-01 -1.22090614e+00 1.63776076e+00
-3.81684154e-01 3.15427214e-01 -4.63268548e-01 -6.92954719e-01
1.10951054e+00 -1.33057535e-02 1.92313179e-01 -5.08971274e-01
4.12293971e-02 4.99572903e-01 -1.12389073e-01 -2.60570168e-01
8.95796418e-01 1.64990470e-01 -1.18787766e-01 5.30189276e-01
-3.04284453e-01 -2.95977239e-02 7.47343540e-01 4.40633237e-01
1.11515939e+00 5.50714992e-02 3.04107636e-01 3.02323345e-02
6.36440456e-01 -1.65272430e-01 9.29070115e-01 6.78144276e-01
-1.57443464e-01 5.80053926e-01 6.66523039e-01 -2.00650856e-01
-6.82152748e-01 -7.62194455e-01 1.50599852e-01 8.16175818e-01
3.50765675e-01 -7.89517760e-01 -5.46597838e-01 -9.44906652e-01
-3.14987972e-02 9.51289356e-01 -1.91464767e-01 -3.71812493e-01
-5.64435303e-01 -7.75508404e-01 4.10030693e-01 5.37452161e-01
7.97767639e-01 -7.73665965e-01 -2.93634146e-01 3.48227054e-01
-5.55086195e-01 -9.82891798e-01 -9.95059013e-01 -3.84485424e-01
-1.01575947e+00 -8.57851923e-01 -6.30799055e-01 -7.22355008e-01
7.29594171e-01 6.33299828e-01 9.47740138e-01 4.45330501e-01
2.61185139e-01 -5.42749800e-02 -4.14476156e-01 -3.42776850e-02
-1.75024197e-01 1.92140758e-01 -8.78848135e-02 -4.00435299e-01
1.85988918e-01 -4.53623354e-01 -6.14810228e-01 3.87344301e-01
-7.29601443e-01 1.58602260e-02 6.89135969e-01 1.01109266e+00
5.96252859e-01 8.36460590e-02 9.37087178e-01 -7.96378076e-01
1.08199787e+00 -2.40659326e-01 -5.87455750e-01 5.19060135e-01
-9.63019431e-01 1.91405769e-02 8.35622966e-01 -4.00390178e-01
-9.99336004e-01 -1.88826516e-01 -4.12677415e-02 -1.35643736e-01
4.70273495e-01 7.75255084e-01 1.31625226e-02 2.57569551e-01
4.10295516e-01 4.02734727e-01 -1.74153075e-01 -4.57953900e-01
2.01418906e-01 6.08372092e-01 4.08199191e-01 -3.82839054e-01
6.01523221e-01 -3.74276750e-02 -1.22173093e-01 -4.26137954e-01
-8.46646607e-01 -3.37233931e-01 -2.40740955e-01 6.17626347e-02
3.38893622e-01 -7.06726491e-01 -5.96985638e-01 3.97238046e-01
-1.13716662e+00 -1.07241489e-01 -3.23056988e-02 6.76749051e-01
-3.41217488e-01 8.47182572e-01 -6.66059256e-01 -6.55906737e-01
-6.17985725e-01 -9.25619483e-01 7.14669406e-01 3.16043884e-01
-1.41747534e-01 -6.75453424e-01 -4.90266047e-02 1.88377678e-01
3.21787566e-01 2.07766745e-04 9.58334982e-01 -6.05889857e-01
-3.83431524e-01 -4.17604208e-01 -4.19986993e-01 4.21812296e-01
2.58897245e-01 1.11862719e-01 -3.90662104e-01 -3.03961784e-01
-1.54544920e-01 -2.25939661e-01 8.87288213e-01 1.90361664e-01
1.30042779e+00 -4.11533177e-01 -1.58517301e-01 4.68123436e-01
1.24216568e+00 2.78734624e-01 5.23083985e-01 3.55033368e-01
5.97763658e-01 1.95431840e-02 1.07778084e+00 5.30726850e-01
4.91897792e-01 6.69159889e-01 1.13888510e-01 3.90734263e-02
-1.54458746e-01 -4.72225964e-01 3.07947218e-01 1.34297872e+00
4.13101800e-02 -3.41356605e-01 -5.38292289e-01 4.96415406e-01
-2.02657270e+00 -7.48719454e-01 -1.93471700e-01 2.53197646e+00
9.20742869e-01 4.13758874e-01 1.10959977e-01 1.80007711e-01
6.79686666e-01 1.84558317e-01 -4.74256039e-01 -3.88524979e-01
1.44108996e-01 -2.37237498e-01 1.91277832e-01 3.77382249e-01
-8.25771213e-01 9.11603749e-01 6.02384043e+00 1.25042808e+00
-1.18048680e+00 -2.01682344e-01 4.41731006e-01 -1.79596424e-01
-3.13954324e-01 7.05537945e-02 -7.92657971e-01 8.48543823e-01
5.44061065e-01 -6.48858547e-01 3.66205603e-01 1.02687645e+00
2.31549636e-01 -1.25513107e-01 -8.47502112e-01 1.04593909e+00
1.12109490e-01 -1.09941852e+00 2.91995645e-01 -3.49392802e-01
5.93102753e-01 -2.29779914e-01 -2.08993196e-01 5.05504549e-01
1.74387068e-01 -3.54397386e-01 4.94214475e-01 2.11449996e-01
6.52443469e-01 -9.91207063e-01 7.96817243e-01 4.40905809e-01
-1.20976973e+00 4.15151566e-02 -6.26166999e-01 1.35637254e-01
2.63550431e-01 9.39549863e-01 -7.80961394e-01 8.17884624e-01
2.49917775e-01 5.81672430e-01 -6.23745084e-01 1.26171589e+00
-4.23207551e-01 6.55421674e-01 -2.64703125e-01 -3.69679779e-01
2.61858851e-01 -5.26845813e-01 5.46371877e-01 1.21991873e+00
7.08283901e-01 1.09931771e-02 3.37246716e-01 6.13627493e-01
-2.49554828e-01 2.60764003e-01 -6.76707774e-02 -7.03794584e-02
8.64452243e-01 1.30309248e+00 -6.90251172e-01 -5.30906439e-01
-2.43512571e-01 9.88875091e-01 3.62218887e-01 2.37953812e-01
-9.51828599e-01 -9.29495633e-01 1.27568975e-01 -7.49940723e-02
4.16884631e-01 -2.37903774e-01 -2.94728905e-01 -1.17573178e+00
1.85401171e-01 -6.80554748e-01 2.76051819e-01 -6.85469985e-01
-1.18365109e+00 6.89735293e-01 -4.99515757e-02 -1.38276386e+00
-1.52645618e-01 -1.20809026e-01 -7.48592496e-01 7.09236801e-01
-1.44431138e+00 -7.34127760e-01 -5.83657086e-01 1.58952922e-01
6.06014132e-01 -4.10728902e-02 5.83784819e-01 5.75453699e-01
-8.27648282e-01 9.91754949e-01 1.39403671e-01 -1.18397519e-01
7.95124114e-01 -1.05170202e+00 4.47710246e-01 8.38851392e-01
-1.78670570e-01 8.01146746e-01 5.10224700e-01 -6.85156524e-01
-9.26434278e-01 -1.06064749e+00 1.15815115e+00 1.95733737e-02
5.86233437e-01 1.31831067e-02 -1.04449034e+00 2.92900592e-01
-1.80037633e-01 -2.29385227e-01 7.90251076e-01 8.88999999e-02
-1.88161269e-01 -2.51009643e-01 -9.80439544e-01 7.60368586e-01
1.01380777e+00 -8.38163644e-02 -4.62654561e-01 3.85228783e-01
7.80240893e-01 -2.93317318e-01 -7.92454898e-01 4.76240814e-01
6.08973324e-01 -8.64026845e-01 5.95125198e-01 -2.23891884e-01
5.87607503e-01 -5.24894059e-01 -1.16704643e-01 -1.50911856e+00
-4.98203069e-01 -5.98016798e-01 -2.00672820e-01 1.53552377e+00
6.41556680e-01 -7.45976865e-01 5.27427673e-01 5.21286011e-01
-2.66445726e-01 -1.14255726e+00 -6.12565577e-01 -1.21446526e+00
-1.96181133e-01 -1.26811877e-01 9.22916532e-01 8.65659714e-01
4.15146984e-02 5.98216891e-01 -3.92365605e-01 -3.20514739e-01
2.61865437e-01 2.89662600e-01 8.96718025e-01 -9.61131573e-01
-4.34295088e-01 -4.62110937e-01 6.74485136e-03 -1.57328475e+00
-2.22126633e-01 -8.29237103e-01 -8.27953499e-03 -1.73406446e+00
2.63278067e-01 -5.97738862e-01 -2.88442284e-01 4.55161035e-01
-9.64003325e-01 -1.10764794e-01 3.69878441e-01 3.43471199e-01
-7.67149568e-01 7.42637277e-01 1.21203101e+00 2.41944820e-01
-3.42054725e-01 4.82652225e-02 -7.86068499e-01 7.01216936e-01
9.07741427e-01 -5.20046771e-01 -5.97899675e-01 -4.72792357e-01
3.63081783e-01 -5.12439385e-02 -1.58906162e-01 -8.01177561e-01
2.30994210e-01 -1.62503332e-01 9.51682329e-02 -9.11568582e-01
1.24554835e-01 -4.69478577e-01 -1.19527981e-01 4.85015810e-01
-2.36815289e-01 2.51460969e-01 1.06370397e-01 4.48824942e-01
-2.30098888e-01 -6.02431536e-01 5.21854579e-01 1.46514088e-01
-4.15931821e-01 1.39658958e-01 3.57122570e-02 4.01022732e-02
7.55022943e-01 -1.26265839e-01 -4.44055945e-01 -3.25475514e-01
-1.56084299e-01 5.45871198e-01 5.21416843e-01 2.72921503e-01
5.07146835e-01 -1.67069638e+00 -6.84824407e-01 -4.82762419e-03
9.71088260e-02 -1.24863654e-01 8.55404586e-02 7.84634948e-01
-4.95241672e-01 2.94688910e-01 2.70921618e-01 -3.73244584e-01
-1.40890181e+00 5.25345325e-01 -1.23427995e-01 -6.57246292e-01
-3.65893036e-01 8.80543828e-01 -2.80544832e-02 -4.17058825e-01
2.02602684e-01 -4.15372163e-01 -2.07279980e-01 8.78840312e-02
5.81633925e-01 7.76491344e-01 3.81064378e-02 -1.25696003e-01
-3.49139661e-01 4.19591874e-01 -4.18210417e-01 -1.45434421e-02
1.30117106e+00 -5.98648861e-02 -8.09490085e-02 3.78372252e-01
1.11275315e+00 2.37433389e-01 -1.02162850e+00 -2.07769707e-01
4.02629152e-02 -7.28267729e-01 -2.23739501e-02 -8.20633948e-01
-9.82608616e-01 5.43502927e-01 2.14454412e-01 3.70073199e-01
1.32545960e+00 -6.31129861e-01 1.28140497e+00 4.14295971e-01
2.21721902e-01 -1.23981953e+00 2.20014215e-01 5.98736942e-01
8.54118228e-01 -1.09413350e+00 2.79212147e-01 -5.69089413e-01
-8.23304236e-01 9.37051117e-01 5.93924761e-01 -1.34048015e-01
2.52874166e-01 -1.63191378e-01 -2.03734353e-01 1.63277507e-01
-8.26115668e-01 9.36574116e-02 3.56853068e-01 6.49753287e-02
4.88329411e-01 -1.84784196e-02 -1.29272580e+00 7.92971492e-01
-3.44662040e-01 1.88294083e-01 4.49905753e-01 8.99825454e-01
-5.74522495e-01 -1.32384050e+00 1.27343023e-02 6.86953187e-01
-3.17025661e-01 -2.41242900e-01 -3.61829698e-01 5.52085161e-01
-7.21881092e-02 1.16148210e+00 -7.68470690e-02 -4.97153431e-01
6.25408292e-01 -1.83644891e-01 2.92539537e-01 -4.96830046e-01
-5.52093148e-01 1.30198613e-01 2.20105037e-01 -3.02445143e-01
-7.69320205e-02 -4.00420725e-01 -1.44002342e+00 -4.05953258e-01
-8.73957098e-01 5.33438981e-01 5.29694021e-01 6.43967330e-01
6.84557140e-01 5.44526458e-01 9.81302381e-01 -4.40322131e-01
-8.91659319e-01 -1.01459050e+00 -4.96906400e-01 3.73962045e-01
-1.35137260e-01 -5.89644074e-01 -5.32116294e-01 -2.98988849e-01] | [12.071192741394043, 8.96774673461914] |
d20f79b5-d969-4426-995b-521fbca2f8d8 | learning-spatiotemporal-features-via-video | 2001.05691 | null | https://arxiv.org/abs/2001.05691v3 | https://arxiv.org/pdf/2001.05691v3.pdf | Learning Spatiotemporal Features via Video and Text Pair Discrimination | Current video representations heavily rely on learning from manually annotated video datasets which are time-consuming and expensive to acquire. We observe videos are naturally accompanied by abundant text information such as YouTube titles and Instagram captions. In this paper, we leverage this visual-textual connection to learn spatiotemporal features in an efficient weakly-supervised manner. We present a general cross-modal pair discrimination (CPD) framework to capture this correlation between a video and its associated text. Specifically, we adopt noise-contrastive estimation to tackle the computational issue imposed by the huge amount of pair instance classes and design a practical curriculum learning strategy. We train our CPD models on both standard video dataset (Kinetics-210k) and uncurated web video dataset (Instagram-300k) to demonstrate its effectiveness. Without further fine-tuning, the learnt models obtain competitive results for action classification on Kinetics under the linear classification protocol. Moreover, our visual model provides an effective initialization to fine-tune on downstream tasks, which yields a remarkable performance gain for action recognition on UCF101 and HMDB51, compared with the existing state-of-the-art self-supervised training methods. In addition, our CPD model yields a new state of the art for zero-shot action recognition on UCF101 by directly utilizing the learnt visual-textual embeddings. The code will be made available at https://github.com/MCG-NJU/CPD-Video. | ['Li-Min Wang', 'Tianhao Li'] | 2020-01-16 | null | https://openreview.net/forum?id=Bw7VC-DJUM | https://openreview.net/pdf?id=Bw7VC-DJUM | null | ['zero-shot-action-recognition'] | ['computer-vision'] | [ 9.57936123e-02 -5.24860144e-01 -6.51740909e-01 -2.18821675e-01
-1.08597875e+00 -6.03739858e-01 7.09732711e-01 -1.52425051e-01
-6.25978529e-01 6.02625847e-01 2.96686411e-01 -4.38679866e-02
9.41675529e-02 -3.88807178e-01 -1.00901520e+00 -7.04566300e-01
-1.25849172e-01 2.16830254e-01 1.82308644e-01 -7.34205963e-03
-3.59155983e-02 1.32103581e-02 -1.65540802e+00 5.00262499e-01
5.79192936e-01 1.27043378e+00 1.85297966e-01 7.29775608e-01
1.25267923e-01 1.03264868e+00 -2.61901140e-01 -3.19060951e-01
3.01438034e-01 -3.81531537e-01 -6.71126366e-01 3.22124988e-01
7.03843534e-01 -6.20416701e-01 -9.61701453e-01 8.16282511e-01
3.65808576e-01 2.21686602e-01 7.08353043e-01 -1.50415385e+00
-7.30742574e-01 2.65668809e-01 -6.92051351e-01 6.11735284e-01
3.52129072e-01 4.28659022e-01 1.06601119e+00 -1.05940521e+00
6.92538917e-01 8.35697055e-01 4.92608219e-01 5.38060367e-01
-1.06616008e+00 -7.60912538e-01 2.72268891e-01 6.50645018e-01
-1.35914183e+00 -4.95496333e-01 7.44081736e-01 -5.79402566e-01
7.57357657e-01 1.38523564e-01 7.08663523e-01 1.92129958e+00
-3.82586807e-01 1.22974873e+00 8.28342557e-01 -2.46214524e-01
5.38127571e-02 -7.57719129e-02 -1.59035046e-02 8.50463927e-01
-1.16352059e-01 -2.01243088e-02 -8.40712428e-01 1.05019860e-01
9.22286928e-01 3.16216379e-01 -5.26263773e-01 -4.98592436e-01
-1.31852913e+00 7.23401308e-01 2.34452531e-01 2.50458002e-01
-6.78842291e-02 3.63815069e-01 7.82827795e-01 3.47721547e-01
4.71818537e-01 1.57043025e-01 -5.54573238e-01 -6.77559912e-01
-7.59115100e-01 -2.20638029e-02 4.69606787e-01 9.23964262e-01
4.46131915e-01 -2.12232601e-02 -2.88810104e-01 8.59073460e-01
1.68925703e-01 6.25949621e-01 7.70043790e-01 -9.53860044e-01
7.74404824e-01 3.95664603e-01 -1.84929650e-02 -9.57257330e-01
1.36070000e-02 -1.65073857e-01 -6.90835416e-01 -1.36695564e-01
6.22399449e-01 5.22479527e-02 -8.58130038e-01 1.72524893e+00
2.21468955e-01 8.04490387e-01 4.68058214e-02 9.33702230e-01
8.63687396e-01 6.63617611e-01 -3.83964798e-04 -9.70056430e-02
1.28253484e+00 -1.34680128e+00 -5.58940589e-01 -2.77064666e-02
8.92611384e-01 -3.70923162e-01 1.33614814e+00 2.05214649e-01
-6.55874491e-01 -5.83875954e-01 -8.95737946e-01 -1.56561479e-01
-2.44540870e-01 4.57744867e-01 4.58514839e-01 1.76607519e-01
-5.46461463e-01 5.79991877e-01 -1.03217673e+00 -3.80564302e-01
8.68999302e-01 6.29317462e-02 -7.50030696e-01 -3.25176597e-01
-1.16538596e+00 3.64263088e-01 2.44821444e-01 -7.45331272e-02
-1.31455278e+00 -7.09825993e-01 -1.05101562e+00 -1.36724249e-01
6.75097287e-01 -2.14182094e-01 1.20419478e+00 -1.09355152e+00
-1.37806010e+00 9.53326941e-01 4.11641635e-02 -4.11492676e-01
7.78998315e-01 -3.34832162e-01 -3.65317464e-01 6.45068228e-01
1.27106115e-01 6.19801641e-01 1.03271079e+00 -8.62495601e-01
-5.09645283e-01 -2.04938084e-01 2.28062123e-01 2.37058118e-01
-7.85666883e-01 -1.76938921e-01 -7.57544279e-01 -8.52417827e-01
-3.92654091e-01 -8.78207684e-01 2.46501788e-01 3.41367841e-01
-1.64667815e-01 -3.08535159e-01 8.58855903e-01 -4.71890360e-01
1.06034303e+00 -2.42596579e+00 2.38012120e-01 -3.00185561e-01
8.41814950e-02 4.89314705e-01 -4.09560561e-01 3.92948657e-01
-5.65340482e-02 -8.37721080e-02 -1.94560047e-02 -3.80260229e-01
3.81530672e-02 3.00898910e-01 -3.04836184e-01 5.92502832e-01
2.91728079e-01 1.06300223e+00 -1.06090045e+00 -6.21700168e-01
1.61401927e-01 4.11504805e-01 -5.74563682e-01 4.53916848e-01
-1.97953805e-01 4.44984615e-01 -5.11948109e-01 7.64338672e-01
1.63415834e-01 -5.77141166e-01 2.50122339e-01 -3.25021535e-01
1.24352314e-01 2.69739307e-03 -7.99417794e-01 2.11792731e+00
-2.25629479e-01 7.30167627e-01 -3.06929260e-01 -1.46846950e+00
5.32133162e-01 2.98608214e-01 5.68914950e-01 -6.73137605e-01
1.43782258e-01 4.73113135e-02 -3.73882204e-01 -9.43632722e-01
1.32924512e-01 2.18622655e-01 -1.31819591e-01 2.97464550e-01
5.23179650e-01 4.44894373e-01 3.16997200e-01 3.60078514e-01
1.37615764e+00 5.46066105e-01 1.51110634e-01 7.68020824e-02
3.90323967e-01 -7.56159276e-02 5.78154385e-01 6.17817700e-01
-5.22974491e-01 6.94828451e-01 6.22256041e-01 -3.14781249e-01
-8.96031797e-01 -9.42494333e-01 -8.03116187e-02 1.20843554e+00
1.16561495e-01 -6.08029485e-01 -5.23392379e-01 -1.02714503e+00
3.81399840e-02 1.48359343e-01 -8.94151449e-01 -8.83512720e-02
-4.18046176e-01 -4.05622631e-01 5.55263758e-01 8.92808855e-01
6.21859312e-01 -9.00059104e-01 -3.29804420e-01 -5.62579520e-02
-3.63560408e-01 -1.43797684e+00 -7.00895786e-01 2.81627849e-02
-6.03670239e-01 -1.28773725e+00 -8.32021415e-01 -7.66966939e-01
4.56726879e-01 4.56664115e-01 8.05299461e-01 -4.94433641e-02
-4.47879970e-01 7.38033891e-01 -6.77478373e-01 9.09492299e-02
1.24921322e-01 -5.62255122e-02 1.75848350e-01 2.96629488e-01
5.34487486e-01 -6.76650643e-01 -6.58512950e-01 3.97151619e-01
-8.60494137e-01 9.46487933e-02 4.77832168e-01 1.09933770e+00
6.64739668e-01 -2.71759152e-01 3.20151597e-01 -5.77489018e-01
1.83920357e-02 -7.35081553e-01 -3.79550606e-01 2.82196641e-01
-3.28604281e-02 -1.06301382e-01 6.37039125e-01 -8.08505714e-01
-7.62434244e-01 1.54970542e-01 2.20015928e-01 -1.18172073e+00
-2.97898203e-01 5.10133564e-01 -8.84771496e-02 7.34162256e-02
4.89281744e-01 4.87946212e-01 3.43736969e-02 -4.47551936e-01
3.05272400e-01 7.17803359e-01 7.28180051e-01 -6.63001776e-01
8.08410823e-01 5.21062016e-01 -3.82242888e-01 -6.98719084e-01
-1.02800536e+00 -6.78649724e-01 -7.25628495e-01 -3.60625833e-01
1.01171565e+00 -1.32952714e+00 -6.15095377e-01 5.41809976e-01
-7.86039770e-01 -7.17804968e-01 -1.51395857e-01 6.87999427e-01
-7.64369547e-01 4.98513103e-01 -7.59199560e-01 -4.72263366e-01
7.06142113e-02 -1.03052402e+00 1.06484413e+00 -2.86092982e-02
5.44227771e-02 -8.31413090e-01 1.26353845e-01 8.08841527e-01
-1.69029962e-02 2.41804972e-01 3.62379521e-01 -7.98302710e-01
-5.62249660e-01 -2.25379586e-01 -3.59498292e-01 4.84056354e-01
5.80473356e-02 -1.68721098e-02 -8.62357795e-01 -3.56419355e-01
-4.42314893e-01 -1.06463170e+00 1.03640378e+00 -1.86436232e-02
1.44309449e+00 -2.12272689e-01 -2.06004888e-01 7.31390536e-01
1.27726781e+00 -1.01215310e-01 5.47129154e-01 3.91676247e-01
9.15550470e-01 2.79003322e-01 8.31656992e-01 6.49672508e-01
4.34128970e-01 8.86887670e-01 3.74661118e-01 1.40071556e-01
-2.64947824e-02 -4.96966600e-01 6.00952208e-01 8.19799185e-01
-3.08956325e-01 -2.76211888e-01 -8.63944411e-01 5.48011541e-01
-2.19733548e+00 -1.15340126e+00 8.82141441e-02 1.94290400e+00
8.78883362e-01 -1.63838118e-02 2.41559297e-01 6.77542016e-03
5.04905939e-01 4.53944832e-01 -5.10999620e-01 3.36014241e-01
5.03854863e-02 1.40361125e-02 3.45548570e-01 7.23373368e-02
-1.59037626e+00 8.39342296e-01 4.95623827e+00 1.07774889e+00
-1.03834891e+00 3.49730909e-01 5.68085015e-01 -5.29791057e-01
1.45448565e-01 -3.79003525e-01 -6.49272442e-01 8.08051765e-01
9.73124385e-01 -1.96565110e-02 2.41073355e-01 9.35833812e-01
1.57743767e-01 1.85190111e-01 -1.24662101e+00 1.36549878e+00
3.33384573e-01 -1.41934645e+00 -4.20896895e-03 -3.39273401e-02
6.46190166e-01 1.80275381e-01 5.59537336e-02 6.68207288e-01
5.42432405e-02 -9.22549605e-01 5.94264328e-01 3.06911975e-01
1.02383447e+00 -4.21836376e-01 4.65660781e-01 1.07672177e-01
-1.24777246e+00 -2.59820819e-01 -3.65920126e-01 -6.82244301e-02
1.11510657e-01 8.76161456e-02 -1.61656231e-01 4.87233907e-01
9.34396923e-01 1.41198015e+00 -4.55512524e-01 8.23941648e-01
-3.00755382e-01 6.74471915e-01 -4.51299101e-02 1.15314804e-01
4.93108809e-01 -3.77821401e-02 2.45094374e-01 1.13181758e+00
2.09635943e-01 2.51084805e-01 4.39403802e-01 3.74552280e-01
-4.30631101e-01 9.19695944e-02 -5.62762678e-01 -4.87072468e-01
2.70463794e-01 1.18688190e+00 -4.53277886e-01 -5.19834936e-01
-7.78503895e-01 1.22230434e+00 7.91879535e-01 4.22990918e-01
-1.24309158e+00 -7.42570236e-02 8.98685694e-01 -4.58794087e-02
7.93176770e-01 -2.07574025e-01 4.92557824e-01 -1.76699007e+00
1.85878843e-01 -1.07336366e+00 4.65245396e-01 -7.00996757e-01
-1.38380718e+00 4.20439810e-01 6.31534383e-02 -1.68250835e+00
1.15141477e-02 -8.13192487e-01 -4.35119420e-01 8.64204317e-02
-1.47576761e+00 -1.14215493e+00 -5.50449789e-01 9.11716998e-01
7.24312901e-01 -3.08089286e-01 7.16917276e-01 6.66616321e-01
-9.30508554e-01 8.55337560e-01 2.81739056e-01 6.35370791e-01
9.29698467e-01 -1.04082084e+00 -5.79273812e-02 6.51704490e-01
4.30486143e-01 2.11213306e-01 3.13688397e-01 -3.20067257e-01
-1.48481178e+00 -1.19999003e+00 3.02763224e-01 -5.72067499e-01
1.16794884e+00 -5.44290006e-01 -8.83584082e-01 8.27078402e-01
6.14534877e-03 6.59872770e-01 9.19028938e-01 -8.32645744e-02
-7.64896333e-01 -1.26971930e-01 -5.91468155e-01 4.68791783e-01
1.47264421e+00 -7.57557213e-01 -6.03523374e-01 6.14306152e-01
6.71265781e-01 -3.68624389e-01 -9.18196440e-01 3.55913430e-01
6.94902360e-01 -6.64809942e-01 9.06682611e-01 -1.11297047e+00
9.34859455e-01 -1.06639251e-01 -2.97833294e-01 -1.04545569e+00
-1.22140549e-01 -4.11190271e-01 -4.54837829e-01 1.21472716e+00
2.80632257e-01 -3.12033623e-01 8.17066014e-01 3.05461645e-01
-7.53929541e-02 -8.97432566e-01 -9.22280729e-01 -1.01587212e+00
-1.72813743e-01 -4.17069763e-01 -3.71871982e-04 1.22898734e+00
1.25347257e-01 2.26510242e-01 -7.51903117e-01 -3.54152508e-02
6.39193237e-01 1.33663073e-01 8.87341678e-01 -7.69213617e-01
-6.28480017e-01 -1.86693087e-01 -8.21483552e-01 -1.33835471e+00
4.23116803e-01 -8.28953803e-01 -5.54010011e-02 -9.82661247e-01
4.85181272e-01 -1.58084914e-01 -6.46327317e-01 5.68380237e-01
-1.83081061e-01 5.83181500e-01 1.97600111e-01 3.25392872e-01
-1.15401220e+00 8.51877928e-01 1.08369756e+00 -2.90622503e-01
1.72142535e-01 -3.78901482e-01 -2.61639982e-01 5.43715477e-01
7.11108029e-01 -4.49786931e-01 -5.71139038e-01 -5.08971512e-01
-2.13685930e-01 1.26795501e-01 5.53258777e-01 -9.28852260e-01
8.94643590e-02 -1.20078087e-01 3.97398770e-01 -2.71015167e-01
5.37445068e-01 -6.60406947e-01 -2.51854390e-01 2.46756956e-01
-5.57139874e-01 -2.30026647e-01 5.00732586e-02 1.12141418e+00
-4.35578555e-01 -7.63641968e-02 4.83601570e-01 -5.60661145e-02
-1.01273024e+00 7.13970482e-01 -2.41612986e-01 3.67710948e-01
1.27291691e+00 1.25473682e-02 -4.97779727e-01 -4.42795306e-01
-7.10415900e-01 3.94396365e-01 4.56772506e-01 5.40341496e-01
5.73978662e-01 -1.58200312e+00 -5.32505929e-01 1.10060312e-01
5.06465912e-01 -3.05274189e-01 4.85380262e-01 1.16074812e+00
-3.25703323e-01 2.85496771e-01 -3.83899212e-01 -6.37285888e-01
-1.39924896e+00 6.55169606e-01 -5.52106043e-03 -2.77940005e-01
-7.05984652e-01 9.51446235e-01 2.81980395e-01 -1.59205452e-01
4.86608148e-01 -9.62288678e-02 -1.25099510e-01 2.57927060e-01
7.45234609e-01 1.59184292e-01 -3.67655188e-01 -6.74167037e-01
-4.29227859e-01 4.71508652e-01 -1.47562578e-01 9.82623547e-02
1.40133154e+00 1.14249602e-01 4.72960323e-01 5.47356009e-01
1.69306576e+00 -4.39408809e-01 -1.75743961e+00 -4.21576977e-01
-1.93339601e-01 -7.38953114e-01 -8.08511600e-02 -3.72411221e-01
-1.21251392e+00 9.41929221e-01 4.82470959e-01 -1.78489283e-01
8.79048288e-01 1.23528577e-01 8.02636206e-01 5.53585827e-01
2.67850846e-01 -1.10564375e+00 7.05650687e-01 2.72839010e-01
7.68596649e-01 -1.69357395e+00 -1.62942022e-01 -2.72360861e-01
-8.87304664e-01 1.04847109e+00 7.27014542e-01 -1.55719116e-01
5.55108547e-01 -3.39928530e-02 5.95880672e-03 -1.35209098e-01
-9.18382585e-01 -2.57799745e-01 2.68931746e-01 4.69866782e-01
2.31300279e-01 -1.01894997e-01 -7.75017217e-02 5.78497648e-01
3.66774887e-01 2.62704551e-01 3.74248266e-01 1.02874637e+00
-1.59627438e-01 -9.31771219e-01 1.68149307e-01 4.41358447e-01
-4.74619627e-01 -3.66263986e-02 -1.30177706e-01 8.32405686e-01
-2.17293248e-01 6.41558051e-01 1.51054710e-01 -5.45364797e-01
1.22192234e-01 1.20895408e-01 5.16753793e-01 -4.55460370e-01
-2.95065381e-02 -3.29261459e-02 2.11955160e-01 -9.57796335e-01
-7.22071528e-01 -7.07862020e-01 -9.25975025e-01 -1.41157269e-01
-4.66303080e-02 -4.09455337e-02 1.80635363e-01 9.29698884e-01
4.22555447e-01 3.48102957e-01 6.35639489e-01 -1.00079739e+00
-6.40744805e-01 -8.80733669e-01 -5.31329155e-01 7.81606555e-01
2.24215329e-01 -1.04281926e+00 -4.93144810e-01 2.83035100e-01] | [8.817800521850586, 0.852346658706665] |
74cec07c-0c73-4c0c-9b94-08ceac9da82e | personalized-face-modeling-for-improved-face | 2007.06759 | null | https://arxiv.org/abs/2007.06759v2 | https://arxiv.org/pdf/2007.06759v2.pdf | Personalized Face Modeling for Improved Face Reconstruction and Motion Retargeting | Traditional methods for image-based 3D face reconstruction and facial motion retargeting fit a 3D morphable model (3DMM) to the face, which has limited modeling capacity and fail to generalize well to in-the-wild data. Use of deformation transfer or multilinear tensor as a personalized 3DMM for blendshape interpolation does not address the fact that facial expressions result in different local and global skin deformations in different persons. Moreover, existing methods learn a single albedo per user which is not enough to capture the expression-specific skin reflectance variations. We propose an end-to-end framework that jointly learns a personalized face model per user and per-frame facial motion parameters from a large corpus of in-the-wild videos of user expressions. Specifically, we learn user-specific expression blendshapes and dynamic (expression-specific) albedo maps by predicting personalized corrections on top of a 3DMM prior. We introduce novel constraints to ensure that the corrected blendshapes retain their semantic meanings and the reconstructed geometry is disentangled from the albedo. Experimental results show that our personalization accurately captures fine-grained facial dynamics in a wide range of conditions and efficiently decouples the learned face model from facial motion, resulting in more accurate face reconstruction and facial motion retargeting compared to state-of-the-art methods. | ['Linda Shapiro', 'Bindita Chaudhuri', 'Noranart Vesdapunt', 'Baoyuan Wang'] | 2020-07-14 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2986_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123500137.pdf | eccv-2020-8 | ['motion-retargeting'] | ['computer-vision'] | [-2.10066125e-01 -5.71529940e-02 -7.98386112e-02 -7.44501233e-01
-3.68183017e-01 -5.14583826e-01 3.81626397e-01 -9.49834824e-01
1.50018156e-01 1.70409635e-01 3.36421251e-01 5.06951809e-01
2.02726811e-01 -3.97102982e-01 -6.26360953e-01 -7.39565551e-01
8.84135142e-02 4.29122806e-01 -3.69397908e-01 -3.51102769e-01
-2.00392976e-01 9.18736994e-01 -1.52232480e+00 1.78397477e-01
4.76755232e-01 9.38520432e-01 -4.60315645e-01 6.81653380e-01
3.50849479e-02 4.30779517e-01 -1.54045209e-01 -6.01562142e-01
6.69950247e-01 -4.80058134e-01 -4.66182441e-01 5.63173532e-01
1.25338805e+00 -6.36225760e-01 -5.10916829e-01 8.03222537e-01
5.40764630e-01 2.02086717e-01 6.15336299e-01 -1.32865047e+00
-7.81189919e-01 -3.67996752e-01 -8.66926253e-01 -4.54087198e-01
4.58627045e-01 2.06484035e-01 5.76227963e-01 -1.06316698e+00
8.15555811e-01 1.69776332e+00 9.42163765e-01 1.15494430e+00
-1.50679755e+00 -6.92595899e-01 2.37434134e-01 -2.69228607e-01
-1.61737740e+00 -7.77556658e-01 1.01720273e+00 -6.17144823e-01
4.96559680e-01 4.09314275e-01 9.56827521e-01 1.14887834e+00
2.32745223e-02 3.14211547e-01 9.71926928e-01 -1.21749409e-01
-2.23243818e-01 -1.05826296e-01 -5.97370327e-01 1.18894792e+00
-2.45503113e-01 -5.64948618e-02 -9.22272027e-01 -2.28762865e-01
1.29865503e+00 1.96107570e-02 -2.64836222e-01 -5.15799761e-01
-9.54148173e-01 5.07561088e-01 2.92800307e-01 -1.38845071e-01
-2.66711205e-01 4.30940062e-01 -5.96275069e-02 1.13078304e-01
8.96711767e-01 -6.71201050e-02 -6.77813232e-01 1.28268078e-01
-9.92352664e-01 5.91018438e-01 4.05086488e-01 7.74800122e-01
1.29592133e+00 2.72715986e-01 -1.10913314e-01 7.37133205e-01
5.63193083e-01 9.68271434e-01 1.32683739e-01 -1.58786571e+00
5.72967753e-02 3.86755228e-01 2.29057282e-01 -1.19686615e+00
-3.72155190e-01 6.89426437e-02 -5.72289526e-01 2.94977218e-01
3.86635721e-01 -1.44450963e-01 -1.00138819e+00 2.21186900e+00
8.70033681e-01 4.48965818e-01 -4.44854975e-01 1.11918139e+00
6.78883493e-01 4.19469088e-01 -6.01453185e-02 -2.55825341e-01
1.25438523e+00 -7.93424308e-01 -7.31841862e-01 -9.59302410e-02
2.47050703e-01 -8.23112667e-01 1.02404022e+00 2.70254817e-02
-1.17353964e+00 -4.55668867e-01 -3.77866387e-01 -2.48962298e-01
2.51090944e-01 -5.10415882e-02 6.69045210e-01 7.50223279e-01
-1.11014903e+00 5.81420839e-01 -8.65271807e-01 -3.99551302e-01
3.69064152e-01 4.48057532e-01 -6.72347903e-01 1.33282140e-01
-7.36199021e-01 7.31761873e-01 -5.45036793e-01 3.10785621e-01
-8.16729367e-01 -1.20940149e+00 -9.07074749e-01 -4.12712395e-01
2.06372961e-02 -1.12961864e+00 1.12796080e+00 -1.48623621e+00
-2.15364718e+00 1.19443893e+00 -4.66045052e-01 5.44234633e-01
6.63297713e-01 -2.06520319e-01 -2.34692514e-01 2.00669244e-01
-2.04389766e-01 7.22527802e-01 1.31651449e+00 -1.34369302e+00
1.06894404e-01 -7.50593960e-01 -1.81484535e-01 4.15091038e-01
-4.15481448e-01 4.96335365e-02 -7.63794065e-01 -8.09726298e-01
1.15602158e-01 -1.12565327e+00 1.26363650e-01 7.58471370e-01
4.19555008e-02 3.37043971e-01 9.33384418e-01 -8.90713990e-01
8.78567636e-01 -1.91404271e+00 6.13347173e-01 1.66857123e-01
1.91415608e-01 -4.26114574e-02 -4.67658252e-01 -6.23958744e-02
-1.37487248e-01 6.93350583e-02 -2.14141700e-02 -8.88717175e-01
5.44751696e-02 3.21019948e-01 -5.10980785e-02 7.94819653e-01
1.42358527e-01 9.30479169e-01 -8.18205416e-01 -3.41762334e-01
4.01389524e-02 1.09044349e+00 -9.65808630e-01 3.74699384e-01
-3.44101012e-01 1.04451728e+00 -4.30651665e-01 7.61033058e-01
9.76702988e-01 1.50401443e-01 1.56184480e-01 -5.29048741e-01
2.22628132e-01 -2.96237588e-01 -9.54627514e-01 2.05698991e+00
-4.65161651e-01 3.58504742e-01 4.37862784e-01 -2.68626124e-01
7.75349557e-01 3.32427144e-01 1.03469503e+00 -4.60805804e-01
1.12058416e-01 6.82513267e-02 -6.42972887e-01 -7.15707242e-01
2.45097771e-01 -5.10984004e-01 2.81389862e-01 2.65514374e-01
2.02035040e-01 -2.56372213e-01 -7.07462490e-01 -2.86293328e-01
4.42570776e-01 8.00999582e-01 -3.20919454e-01 -2.30409399e-01
2.95266718e-01 -7.21347094e-01 8.38489711e-01 -2.98087060e-01
-2.93921325e-02 9.03039396e-01 2.81911284e-01 -6.16723120e-01
-9.89827096e-01 -1.09155345e+00 9.81661840e-04 1.15421164e+00
-1.40020296e-01 -3.49352926e-01 -1.10287368e+00 -5.69114983e-01
2.01809049e-01 1.25310138e-01 -8.95082772e-01 2.24547777e-02
-6.29062414e-01 -7.00677633e-01 5.22920370e-01 1.60393104e-01
4.13350016e-01 -3.47002476e-01 -2.41567060e-01 -2.47059181e-01
-3.40182841e-01 -1.25325716e+00 -1.18519807e+00 -8.49564314e-01
-7.24553108e-01 -9.36842144e-01 -6.89898193e-01 -3.84510756e-01
9.87528443e-01 2.84681708e-01 8.10841024e-01 8.23192671e-02
-3.41547132e-01 7.28700519e-01 1.78005751e-02 2.82351598e-02
-3.51249874e-01 -5.26705503e-01 4.65090811e-01 8.64400506e-01
-9.93907675e-02 -7.34836221e-01 -7.90349841e-01 5.25989413e-01
-7.87317634e-01 1.97473526e-01 -2.26126626e-01 4.62276310e-01
5.15876234e-01 -5.77101588e-01 9.55239113e-04 -7.32804418e-01
-4.77881581e-02 -8.83356482e-02 -5.11070549e-01 2.05746517e-01
-3.77812654e-01 -3.67332771e-02 1.61419541e-01 -7.37117589e-01
-1.34312248e+00 2.71788061e-01 -1.13840900e-01 -9.86678720e-01
1.29761711e-01 -1.34909093e-01 -4.91254568e-01 -6.55663431e-01
6.51393473e-01 -2.69668698e-01 4.05163467e-01 -5.25578320e-01
6.14496648e-01 2.15788394e-01 4.91539896e-01 -7.73545921e-01
1.07248294e+00 8.51984322e-01 4.24957931e-01 -9.76648450e-01
-7.83795476e-01 -1.18929353e-02 -9.83453214e-01 -5.60130954e-01
9.72213149e-01 -1.09383166e+00 -9.76863146e-01 8.61890674e-01
-1.17487645e+00 -7.31691539e-01 -2.46200293e-01 1.29451349e-01
-6.76580429e-01 4.32377428e-01 -5.02383471e-01 -7.52790213e-01
-1.94383428e-01 -9.88367617e-01 1.50222874e+00 1.73471808e-01
-1.49749428e-01 -1.04836941e+00 8.09444413e-02 4.48825389e-01
3.19624573e-01 9.13179994e-01 6.17580175e-01 6.25058472e-01
-5.34269035e-01 -1.57724731e-02 1.85558170e-01 2.42110848e-01
3.43448967e-01 3.58181059e-01 -1.20711648e+00 -2.91678160e-01
-1.62718952e-01 -4.08694968e-02 5.61853170e-01 3.22584510e-01
1.04286575e+00 -7.69387603e-01 3.70138437e-02 1.32412529e+00
1.11924219e+00 -5.94088316e-01 6.08554363e-01 -3.60339344e-01
1.32838583e+00 9.44339573e-01 2.26104513e-01 6.47337854e-01
6.91426098e-01 1.18855655e+00 5.04243374e-01 -3.03636827e-02
-4.16335285e-01 -4.93750334e-01 7.85833061e-01 5.36867023e-01
-6.37462318e-01 3.37511569e-01 -4.36737776e-01 1.22010529e-01
-1.72747958e+00 -9.20099556e-01 9.96188596e-02 2.43466759e+00
8.91623497e-01 -7.89441705e-01 3.37301254e-01 -6.11262679e-01
3.26782376e-01 1.91936761e-01 -6.24387324e-01 -1.12672299e-01
-2.08484963e-01 2.44564340e-01 3.34107339e-01 9.70858335e-01
-6.72830403e-01 1.21858168e+00 5.66711807e+00 3.76080662e-01
-1.47666204e+00 4.52587903e-01 4.87482518e-01 -4.01452214e-01
-8.95882785e-01 2.05664779e-03 -6.89989567e-01 1.16429158e-01
6.77650094e-01 2.71590263e-01 8.53965878e-01 4.78218615e-01
4.38808024e-01 3.58442217e-01 -9.51324582e-01 1.26282907e+00
2.96705872e-01 -1.11188614e+00 1.39181912e-01 2.31090993e-01
1.00346804e+00 -3.22951764e-01 2.46685758e-01 -2.16619819e-01
-1.48536488e-01 -1.01721263e+00 1.10986257e+00 1.12893581e+00
1.21578896e+00 -4.76988107e-01 -1.60004497e-01 1.96922421e-02
-1.05513561e+00 2.55176812e-01 -1.20820165e-01 2.18025699e-01
8.56806338e-02 2.12991953e-01 -5.03340781e-01 2.98747748e-01
7.57675648e-01 6.97930276e-01 -2.93952793e-01 2.51802415e-01
-7.01238513e-02 1.74236804e-01 -3.77399325e-01 5.92579305e-01
-3.68771315e-01 -4.29203451e-01 6.87782347e-01 9.61290836e-01
4.90300745e-01 1.79012269e-01 -1.25843644e-01 9.17211056e-01
-2.97851026e-01 7.50321224e-02 -3.02365184e-01 1.31967098e-01
1.42908365e-01 1.34001720e+00 -2.35733073e-02 2.09037140e-01
-2.35859290e-01 1.43626559e+00 2.30448648e-01 7.42743015e-01
-7.91278183e-01 4.86903101e-01 1.59060156e+00 6.95881784e-01
5.88619374e-02 -5.23940086e-01 9.79343206e-02 -1.56875110e+00
-2.01148652e-02 -6.83360875e-01 -7.32986107e-02 -8.05548370e-01
-1.21932554e+00 4.69151676e-01 1.39510119e-02 -8.17262173e-01
-2.48135060e-01 -6.79913402e-01 -3.44854176e-01 9.66217399e-01
-1.45533240e+00 -1.85454488e+00 -7.43637443e-01 8.89262557e-01
1.44666851e-01 3.27396556e-03 9.77961481e-01 2.88493782e-01
-6.18284345e-01 9.32911992e-01 -3.69932532e-01 -5.15264750e-04
1.08549690e+00 -8.25433910e-01 2.98563421e-01 5.60371041e-01
-9.20690037e-03 5.87012589e-01 6.37725234e-01 -5.04762411e-01
-1.94851756e+00 -1.29317760e+00 6.46589994e-01 -9.72758949e-01
2.01727808e-01 -5.18986583e-01 -7.49084473e-01 8.75327170e-01
-1.96113631e-01 6.81992531e-01 5.86912394e-01 -8.11257884e-02
-8.83790255e-01 -5.09045184e-01 -1.30068910e+00 7.51234353e-01
1.37875092e+00 -7.02678382e-01 2.45654747e-01 2.65847385e-01
4.75704759e-01 -7.92999387e-01 -1.07373846e+00 3.34512621e-01
1.03463197e+00 -9.39008117e-01 1.03617311e+00 -7.69174039e-01
-8.52126107e-02 -4.38775986e-01 -3.80451411e-01 -1.12600243e+00
-1.60082355e-01 -1.37789261e+00 -3.12245280e-01 1.16085505e+00
-9.57812890e-02 -3.70413750e-01 8.68856370e-01 1.11134982e+00
1.46975473e-01 -6.51530266e-01 -9.29029107e-01 -4.33768481e-01
6.20637871e-02 -3.33768487e-01 9.25664485e-01 1.17665708e+00
-5.60077429e-01 4.87585925e-02 -8.42025697e-01 3.48796695e-01
8.09091151e-01 -2.56021190e-02 1.21031821e+00 -1.24626696e+00
-2.22403005e-01 -3.43082815e-01 -3.67774248e-01 -9.75946426e-01
6.69067800e-01 -8.46454501e-01 -2.69554585e-01 -8.76309156e-01
1.55240104e-01 -3.35545301e-01 4.65933025e-01 7.11909235e-01
7.64200324e-03 6.10343397e-01 2.71834791e-01 1.95343018e-01
3.46925035e-02 7.86700308e-01 1.57857382e+00 2.08154127e-01
-2.43673548e-01 -2.33197600e-01 -5.65579951e-01 8.51467729e-01
2.69232064e-01 -1.30101219e-01 -3.53044122e-01 -7.23710239e-01
2.83916712e-01 1.46377698e-01 7.04435468e-01 -4.80646938e-01
-2.22222462e-01 -6.65166259e-01 6.07312500e-01 1.65826008e-01
7.94249415e-01 -7.30864346e-01 6.15279555e-01 -1.57312453e-01
6.31131753e-02 -1.08249165e-01 2.44898051e-01 4.43556577e-01
3.59713793e-01 4.68590677e-01 1.05595326e+00 5.36986394e-03
-3.55511725e-01 1.13463831e+00 2.70149589e-01 -8.03465471e-02
7.22163260e-01 -2.83759892e-01 1.79877296e-01 -5.30036688e-01
-8.28976274e-01 -1.55921772e-01 9.90993023e-01 6.15294814e-01
4.72045362e-01 -1.68100536e+00 -8.62597287e-01 4.61449653e-01
-1.03136878e-02 5.03302403e-02 6.13985777e-01 7.71858633e-01
-6.35556638e-01 -2.88048208e-01 -2.66423434e-01 -6.00828886e-01
-1.42806518e+00 1.07769735e-01 1.02058947e+00 4.01075363e-01
-5.44685245e-01 1.00022852e+00 4.61228698e-01 -7.19366848e-01
-1.15282290e-01 5.78552391e-03 4.13773358e-01 -4.73914221e-02
3.66377175e-01 1.91175282e-01 -9.25074071e-02 -1.46527410e+00
-3.08495820e-01 1.48498797e+00 4.78674710e-01 -2.45658740e-01
1.26930559e+00 -3.94727737e-01 -3.30775112e-01 2.43976116e-01
1.42384350e+00 3.52077723e-01 -1.94708669e+00 -1.89648017e-01
-7.31903613e-01 -9.43835855e-01 1.46680102e-02 -4.83229101e-01
-1.49902439e+00 8.22870910e-01 5.49208403e-01 -6.43958807e-01
1.28311038e+00 -1.85294241e-01 7.37833619e-01 -1.26036897e-01
3.26806277e-01 -8.98578525e-01 3.58588219e-01 3.77455086e-01
1.21456146e+00 -9.68864918e-01 -3.54831144e-02 -5.12252629e-01
-6.07770324e-01 1.24013925e+00 6.35915339e-01 6.28230423e-02
9.04155314e-01 6.91071600e-02 2.59227812e-01 -4.15192805e-02
-3.28294456e-01 2.15855598e-01 6.45757854e-01 6.95970118e-01
3.76740932e-01 1.22431539e-01 3.94245028e-01 2.95024306e-01
-2.47871816e-01 -1.14441797e-01 1.04714364e-01 3.17300528e-01
9.97549221e-02 -1.19540191e+00 -4.76674646e-01 -2.51188964e-01
-1.55229568e-01 3.94759685e-01 -2.76246607e-01 6.16929710e-01
3.58762056e-01 4.26009476e-01 1.12115465e-01 -4.55721051e-01
4.04428840e-01 1.69048384e-01 1.24922431e+00 -4.74667430e-01
-2.28175551e-01 3.37234825e-01 -2.81649441e-01 -9.88752782e-01
-5.58771551e-01 -7.95797169e-01 -1.26583648e+00 -7.63526380e-01
1.24785870e-01 -4.46832806e-01 6.14605725e-01 6.26850545e-01
7.27370799e-01 -4.74567153e-02 9.12871540e-01 -1.45506990e+00
-2.63393134e-01 -6.88418508e-01 -7.09829032e-01 6.96993649e-01
6.91663146e-01 -8.21748674e-01 -1.37579724e-01 4.07789320e-01] | [12.809592247009277, -0.31808707118034363] |
363e0b47-6ad1-46d7-98ec-8f3e08cdc385 | covariate-balancing-using-the-integral | 2305.13715 | null | https://arxiv.org/abs/2305.13715v1 | https://arxiv.org/pdf/2305.13715v1.pdf | Covariate balancing using the integral probability metric for causal inference | Weighting methods in causal inference have been widely used to achieve a desirable level of covariate balancing. However, the existing weighting methods have desirable theoretical properties only when a certain model, either the propensity score or outcome regression model, is correctly specified. In addition, the corresponding estimators do not behave well for finite samples due to large variance even when the model is correctly specified. In this paper, we consider to use the integral probability metric (IPM), which is a metric between two probability measures, for covariate balancing. Optimal weights are determined so that weighted empirical distributions for the treated and control groups have the smallest IPM value for a given set of discriminators. We prove that the corresponding estimator can be consistent without correctly specifying any model (neither the propensity score nor the outcome regression model). In addition, we empirically show that our proposed method outperforms existing weighting methods with large margins for finite samples. | ['Yongdai Kim', 'Kwonsang Lee', 'Joonhyuk Jung', 'Yuha Park', 'Insung Kong'] | 2023-05-23 | null | null | null | null | ['causal-inference', 'causal-inference'] | ['knowledge-base', 'miscellaneous'] | [ 3.52253675e-01 1.93686616e-02 -9.82319117e-01 -5.51721096e-01
-7.74796605e-01 -2.74506032e-01 4.78267938e-01 3.85025650e-01
-3.59466732e-01 9.66395378e-01 4.37412381e-01 -3.96512687e-01
-6.25335097e-01 -1.01266098e+00 -5.33109426e-01 -7.38341630e-01
-7.91722015e-02 4.10456836e-01 6.58703670e-02 3.52005512e-01
4.13398236e-01 2.55843729e-01 -1.26339543e+00 -4.44178402e-01
1.34093535e+00 4.77665633e-01 -4.14668508e-02 1.80590674e-01
2.66021341e-01 4.55627263e-01 -2.25179374e-01 -4.18915868e-01
1.41781718e-01 -5.58225572e-01 -4.59650546e-01 -4.33040708e-01
4.99045819e-01 -2.06518516e-01 5.29939979e-02 1.46604753e+00
3.04834485e-01 -6.95070773e-02 1.26229787e+00 -1.56395888e+00
-5.39435804e-01 9.54822004e-01 -9.23692167e-01 -2.17342317e-01
1.88185632e-01 -2.76093930e-01 1.16420269e+00 -5.26508749e-01
2.85464913e-01 1.46255767e+00 5.83236516e-01 2.40175501e-01
-1.55574512e+00 -1.04968750e+00 1.73478186e-01 -1.65449098e-01
-1.26248050e+00 -2.08264396e-01 5.20762861e-01 -7.42153227e-01
8.72629061e-02 3.27026218e-01 3.16890776e-01 7.41939783e-01
5.77882469e-01 2.64256954e-01 1.21517563e+00 -4.69178617e-01
4.22159195e-01 1.22382216e-01 5.29217303e-01 3.78178328e-01
8.41981173e-01 5.58425665e-01 -2.12191373e-01 -8.67620528e-01
6.88699186e-01 1.01931177e-01 -3.46414745e-01 -6.87125504e-01
-1.29475439e+00 1.19771242e+00 9.39637050e-02 1.26654878e-01
-4.06578839e-01 2.66160786e-01 1.51444942e-01 5.60606420e-02
2.52086967e-01 6.24015667e-02 -2.90706366e-01 3.63852620e-01
-9.26290214e-01 6.14396274e-01 5.84274471e-01 9.09006476e-01
4.43491280e-01 -2.16906846e-01 -6.52234197e-01 6.76022410e-01
4.59040314e-01 8.24259877e-01 1.41302466e-01 -1.12045085e+00
6.09365940e-01 4.84018952e-01 5.66594005e-01 -9.36658740e-01
-4.60653991e-01 -2.43257567e-01 -1.01792467e+00 2.11063281e-01
9.44727898e-01 -2.40482375e-01 -6.07397258e-01 2.41497779e+00
3.85414541e-01 1.12545967e-01 -1.45860136e-01 7.25958943e-01
2.18429208e-01 3.74889106e-01 6.17732644e-01 -5.50171077e-01
1.45300496e+00 -2.74164915e-01 -8.12594354e-01 -1.70368865e-01
5.91902375e-01 -6.18740857e-01 9.50927913e-01 1.65422365e-01
-1.07929981e+00 -1.81582183e-01 -9.05712903e-01 2.75218785e-01
1.80260867e-01 -1.24384671e-01 4.91865695e-01 9.50327158e-01
-4.93631154e-01 5.79606056e-01 -4.50792283e-01 -3.08161438e-01
2.19836801e-01 3.30279469e-01 -1.10668428e-01 -2.89346911e-02
-1.34930587e+00 7.27316141e-01 1.20594185e-02 -2.96736449e-01
-7.33723342e-01 -9.89556789e-01 -6.76097631e-01 4.23349172e-01
8.42937455e-02 -7.97351062e-01 1.04865873e+00 -7.66070724e-01
-8.59918058e-01 6.67673349e-01 -2.48298585e-01 -2.33675256e-01
8.19127917e-01 1.27799496e-01 -1.93396807e-01 -2.80650616e-01
2.67270416e-01 2.35504791e-01 6.15501225e-01 -1.04379725e+00
-8.20854425e-01 -5.47738969e-01 9.10376832e-02 1.55902468e-02
-1.80162221e-01 2.81528503e-01 4.34085011e-01 -7.59335101e-01
1.11926384e-01 -6.75688744e-01 -4.16212976e-01 -1.05192885e-01
-4.87923950e-01 -2.31788993e-01 1.17180929e-01 -3.76179069e-01
1.48305619e+00 -2.05398536e+00 -1.12590492e-01 4.21309501e-01
3.10013033e-02 -3.00567657e-01 9.09062400e-02 2.68304288e-01
-3.32728863e-01 1.26731023e-01 -5.10158241e-01 1.88032791e-01
1.92781597e-01 -2.30792150e-01 -1.98484555e-01 9.11662281e-01
-2.36078128e-01 3.13729554e-01 -8.37759912e-01 -6.81166053e-01
8.44251812e-02 -2.76269279e-02 -6.32594943e-01 1.20300740e-01
3.16266626e-01 7.18772039e-02 -4.78385717e-01 2.32892141e-01
9.41100538e-01 -6.01869971e-02 4.67095524e-01 -7.01288953e-02
-2.14113846e-01 9.90193561e-02 -1.51327729e+00 8.05858612e-01
-2.30523825e-01 2.58625075e-02 -9.93317068e-02 -1.12067461e+00
8.68631124e-01 4.45898771e-01 4.90930140e-01 -1.88540518e-01
2.31385842e-01 2.96137124e-01 2.08215117e-01 -2.32318163e-01
2.02435762e-01 -6.24833822e-01 -4.31682587e-01 4.17229295e-01
-3.58077824e-01 3.17645639e-01 1.18558541e-01 2.55750846e-02
8.15064669e-01 -3.96590084e-01 8.99834454e-01 -8.00300658e-01
3.85967642e-01 -1.42030090e-01 1.07326806e+00 1.04917955e+00
-3.75300169e-01 5.88145375e-01 9.44883883e-01 1.48743391e-01
-9.34989989e-01 -1.28794289e+00 -7.29594111e-01 7.12532222e-01
2.48076200e-01 2.08423406e-01 -5.96121073e-01 -6.85055375e-01
3.97150010e-01 8.57603431e-01 -7.41386592e-01 -3.24104577e-01
-2.39550054e-01 -1.07291782e+00 2.72944421e-01 4.36795503e-01
2.27709219e-01 -3.88790458e-01 -4.75603819e-01 2.52001941e-01
-1.03260696e-01 -3.09174359e-01 -7.45317578e-01 -4.12971340e-02
-1.07683480e+00 -1.32794988e+00 -9.19387043e-01 -5.94919443e-01
6.91029131e-01 2.74141371e-01 7.97290564e-01 -1.32863835e-01
3.59437019e-01 -1.35448307e-01 -5.43329567e-02 -2.49353364e-01
-3.49173367e-01 -2.33231530e-01 1.32921383e-01 -1.35083511e-01
4.93581057e-01 -4.86044049e-01 -7.49133825e-01 5.50610542e-01
-7.23617792e-01 -8.48211795e-02 2.30964750e-01 9.82225895e-01
9.13073197e-02 6.23289198e-02 1.01322222e+00 -1.06712914e+00
4.98338997e-01 -6.69238627e-01 -8.52692425e-01 5.17373979e-01
-7.88479805e-01 1.17912047e-01 3.31119806e-01 -5.59558451e-01
-1.15095890e+00 -3.26510638e-01 2.51385063e-01 2.81161487e-01
-3.36634577e-03 4.84564900e-01 -4.87181246e-01 3.02304894e-01
4.41720247e-01 -4.94603723e-01 -2.21071199e-01 -4.65035707e-01
-1.99806783e-02 7.06103444e-01 1.92043811e-01 -7.48924613e-01
6.78215683e-01 4.06348199e-01 3.32838535e-01 -6.12634830e-02
-7.72750318e-01 -1.88414469e-01 -3.82206440e-01 5.34336418e-02
7.39448607e-01 -6.74625516e-01 -9.34973300e-01 2.36284941e-01
-8.21934342e-01 -4.36051562e-02 -3.96915004e-02 1.15621817e+00
-5.36141098e-01 2.88224131e-01 -2.57494032e-01 -1.27267945e+00
-3.81416045e-02 -1.09144747e+00 6.82178855e-01 2.33926356e-01
-5.23757696e-01 -1.02991569e+00 3.28958817e-02 -3.33731025e-02
1.55776739e-01 4.32401448e-01 1.47203422e+00 -4.95481074e-01
-1.33919358e-01 -3.37626368e-01 -4.53478575e-01 -1.26559898e-01
2.32548699e-01 8.85746628e-02 -4.90751058e-01 -3.00012141e-01
-1.01766065e-01 7.13384524e-02 6.90070331e-01 1.01570332e+00
1.12573481e+00 -3.10616016e-01 -5.74954569e-01 2.11553827e-01
1.41501403e+00 4.58581656e-01 3.93799514e-01 1.26871407e-01
2.15405911e-01 8.36465299e-01 8.59480798e-01 6.07510626e-01
3.78708810e-01 5.55597186e-01 1.85150042e-01 3.17051299e-02
3.03963393e-01 -3.95146251e-01 7.26575702e-02 2.30913505e-01
1.35217294e-01 -2.02823594e-01 -7.48202860e-01 6.90179527e-01
-1.98358834e+00 -9.68447387e-01 -5.07835090e-01 2.97145391e+00
9.63305891e-01 1.31536379e-01 1.54871121e-01 1.65124238e-01
1.11203599e+00 -2.45137159e-02 -3.98864239e-01 -2.95032889e-01
1.30602494e-01 5.65578900e-02 9.97384489e-01 6.56216383e-01
-9.31310713e-01 1.97187930e-01 7.09930706e+00 7.79493392e-01
-4.40725207e-01 9.59186032e-02 5.53355336e-01 1.65124491e-01
-5.96562505e-01 3.84672344e-01 -8.00971329e-01 7.61538386e-01
5.63844800e-01 -7.40182996e-01 -1.55460939e-01 6.21295393e-01
6.91575110e-01 -4.81875598e-01 -1.21011984e+00 3.62979919e-01
-5.36720216e-01 -8.38412046e-01 -2.03847498e-01 2.77689964e-01
8.38052034e-01 -8.22552264e-01 -3.01568361e-04 4.15947214e-02
7.68572152e-01 -1.00326920e+00 7.66830623e-01 5.86715460e-01
7.70873845e-01 -8.49114895e-01 8.68224025e-01 3.76858294e-01
-8.51387203e-01 -5.29906079e-02 -4.81385112e-01 -8.54122713e-02
2.97236323e-01 1.05938137e+00 -1.32495612e-01 3.94018799e-01
3.32818300e-01 3.39541852e-01 -3.43942386e-03 1.28661537e+00
-1.80918291e-01 8.34351242e-01 -1.97095662e-01 1.09913357e-01
4.15277258e-02 -2.86978692e-01 4.95385826e-01 7.17278719e-01
5.90046167e-01 6.07666224e-02 4.75627966e-02 7.73090780e-01
2.74996702e-02 2.39563033e-01 -4.82163876e-01 3.56246561e-01
8.79009843e-01 7.74222493e-01 -5.16619086e-01 -2.81488389e-01
-5.00032783e-01 1.80155903e-01 1.75889045e-01 3.84791344e-01
-8.78700435e-01 -2.97138631e-01 7.90155172e-01 1.36082500e-01
-3.28618735e-01 2.06856743e-01 -6.82739437e-01 -9.48007882e-01
-2.83408195e-01 -5.34740388e-01 6.85341656e-01 -3.95857066e-01
-1.51202881e+00 -2.83439875e-01 5.91781437e-01 -1.09070456e+00
2.64992714e-02 -2.98545003e-01 -7.27322578e-01 1.16192782e+00
-1.21014667e+00 -6.70476079e-01 1.47752156e-02 2.90725380e-01
-1.32538795e-01 9.93641019e-02 6.45553052e-01 3.54068398e-01
-6.19465351e-01 7.11205781e-01 3.68054628e-01 -2.01187551e-01
1.04695749e+00 -1.40494609e+00 -2.56093442e-01 5.81636965e-01
-6.57347322e-01 8.03074062e-01 1.00580668e+00 -9.47024524e-01
-8.98845494e-01 -8.06888402e-01 1.26710439e+00 8.19154270e-03
6.32477105e-01 -1.74139708e-01 -8.34051847e-01 5.47277391e-01
-1.25771716e-01 -3.68289530e-01 6.89602911e-01 4.82255310e-01
-2.83286929e-01 -2.02549040e-01 -1.56391752e+00 7.48702645e-01
9.91392612e-01 6.07836200e-03 -6.87152803e-01 2.58690864e-01
5.43469727e-01 5.32689616e-02 -1.02383876e+00 7.06707954e-01
9.02520299e-01 -9.32991445e-01 8.95617843e-01 -7.50119209e-01
4.04201835e-01 -1.94868729e-01 -2.79235214e-01 -1.21436584e+00
-5.57776332e-01 -1.01148307e-01 3.78545552e-01 1.40687466e+00
4.29241270e-01 -8.67565989e-01 5.86277664e-01 9.31804717e-01
5.39224863e-01 -3.92348707e-01 -1.09304118e+00 -8.78764868e-01
4.31133896e-01 -3.46214622e-01 9.02184367e-01 1.01071405e+00
2.12084278e-01 1.23857498e-01 -5.88974059e-01 2.16363370e-01
1.18323553e+00 2.88299531e-01 4.77423698e-01 -1.54008996e+00
-4.50858921e-02 -5.55509865e-01 -1.60423875e-01 -5.38790822e-01
2.47519597e-01 -5.73691070e-01 -1.59472063e-01 -1.36146617e+00
8.70328546e-01 -7.99479783e-01 -2.98671454e-01 2.32685447e-01
-6.15341187e-01 -1.73171669e-01 -1.26054898e-01 6.30834401e-02
3.23031992e-01 4.34451848e-01 1.06549633e+00 -8.42140522e-03
-7.63304308e-02 3.59235376e-01 -9.38601792e-01 7.80236363e-01
8.03549170e-01 -8.78341258e-01 -4.17649239e-01 6.32904768e-02
1.37424529e-01 5.27601302e-01 4.24781680e-01 -5.92244565e-01
-3.27741988e-02 -7.57947445e-01 2.02298179e-01 -4.48940545e-01
-1.94851011e-01 -1.02219474e+00 5.27280450e-01 5.88393211e-01
-7.39042222e-01 -2.05350310e-01 -3.27960044e-01 6.26228511e-01
4.60489206e-02 -5.79026520e-01 1.08540213e+00 4.15846854e-01
1.62481785e-01 1.23085432e-01 -3.79336238e-01 -2.03852989e-02
1.08715189e+00 -2.06924863e-02 -3.04603726e-01 -4.71769452e-01
-3.76965076e-01 4.86983567e-01 4.82666343e-01 1.53706044e-01
1.11685835e-01 -1.75781798e+00 -8.00919771e-01 -1.23968847e-01
1.91658273e-01 -5.65483749e-01 1.79025486e-01 1.00667095e+00
5.03851958e-02 3.29449922e-01 5.68630989e-04 -1.79364651e-01
-1.11770022e+00 5.62961936e-01 2.15311542e-01 -3.95749271e-01
-2.70895481e-01 1.47026658e-01 7.66957581e-01 -3.29631329e-01
1.89536795e-01 -3.32383126e-01 -1.58537075e-01 1.41813263e-01
5.93793631e-01 7.83127666e-01 -4.25012678e-01 -4.82048452e-01
-3.70899677e-01 5.65983653e-01 2.93964714e-01 -2.89777189e-01
1.03370976e+00 -1.87389374e-01 -1.18679807e-01 3.81880134e-01
1.08858323e+00 2.14469954e-01 -1.05790949e+00 -1.73907235e-01
2.37920120e-01 -9.13746119e-01 -1.24989320e-02 -5.90100110e-01
-8.24760020e-01 5.22708356e-01 5.69005907e-01 2.85395145e-01
9.66251969e-01 -2.88902611e-01 3.10361564e-01 -3.91681582e-01
5.39386332e-01 -8.40655863e-01 -7.10674345e-01 4.28555794e-02
6.06883049e-01 -1.14086843e+00 2.34518647e-01 -4.62594509e-01
-3.75112414e-01 7.09645271e-01 4.18505102e-01 -2.96513855e-01
8.53386581e-01 2.32462928e-01 -3.13794196e-01 1.36625439e-01
-4.23426747e-01 2.83742752e-02 3.83101404e-01 6.29172087e-01
7.41652250e-01 5.24888039e-01 -1.18290222e+00 7.83694446e-01
-2.01143056e-01 -4.39263955e-02 4.56017137e-01 5.11940658e-01
-5.41939974e-01 -1.20334172e+00 -7.82048941e-01 8.05004060e-01
-5.40971279e-01 6.42253608e-02 9.21657234e-02 8.98497343e-01
-1.11961216e-01 1.09571648e+00 6.49429485e-02 1.08741127e-01
4.73297387e-01 -1.17930416e-02 3.91479909e-01 -3.70026439e-01
-2.77195513e-01 9.22301188e-02 6.45073503e-02 -2.89548665e-01
-6.28359675e-01 -1.01091015e+00 -8.75832796e-01 -7.29389429e-01
-5.37849188e-01 3.24834943e-01 4.32499386e-02 8.40864420e-01
-2.67206877e-01 1.06517136e-01 7.95788825e-01 -2.46104583e-01
-8.93649220e-01 -9.19704020e-01 -1.11859572e+00 1.32856309e-01
1.38980523e-01 -1.07048368e+00 -4.76099730e-01 -2.65598625e-01] | [7.930051803588867, 5.207375526428223] |
94c08e85-7a49-4095-95ee-064ad6407e5a | gta-global-temporal-attention-for-video | 2012.08510 | null | https://arxiv.org/abs/2012.08510v3 | https://arxiv.org/pdf/2012.08510v3.pdf | GTA: Global Temporal Attention for Video Action Understanding | Self-attention learns pairwise interactions to model long-range dependencies, yielding great improvements for video action recognition. In this paper, we seek a deeper understanding of self-attention for temporal modeling in videos. We first demonstrate that the entangled modeling of spatio-temporal information by flattening all pixels is sub-optimal, failing to capture temporal relationships among frames explicitly. To this end, we introduce Global Temporal Attention (GTA), which performs global temporal attention on top of spatial attention in a decoupled manner. We apply GTA on both pixels and semantically similar regions to capture temporal relationships at different levels of spatial granularity. Unlike conventional self-attention that computes an instance-specific attention matrix, GTA directly learns a global attention matrix that is intended to encode temporal structures that generalize across different samples. We further augment GTA with a cross-channel multi-head fashion to exploit channel interactions for better temporal modeling. Extensive experiments on 2D and 3D networks demonstrate that our approach consistently enhances temporal modeling and provides state-of-the-art performance on three video action recognition datasets. | ['Abhinav Shrivastava', 'Ser-Nam Lim', 'Hao Chen', 'Zuxuan Wu', 'Xitong Yang', 'Bo He'] | 2020-12-15 | null | null | null | null | ['action-understanding'] | ['computer-vision'] | [ 9.62276757e-03 -2.27512777e-01 -5.82629561e-01 -3.27360719e-01
-6.20187521e-01 -4.31418359e-01 8.08104634e-01 -1.90489128e-01
-1.16452619e-01 2.89955676e-01 7.78819025e-01 -1.68610781e-01
-4.93428111e-02 -4.59535182e-01 -9.67640340e-01 -6.44590795e-01
-5.60244858e-01 -2.07469743e-02 2.89424807e-01 -2.34177746e-02
6.98102117e-02 3.90463740e-01 -1.14612758e+00 7.68654406e-01
3.20520014e-01 9.37703729e-01 -3.33869308e-02 8.60607803e-01
1.14605181e-01 1.38464856e+00 -1.64926708e-01 -1.02119796e-01
1.58575609e-01 -5.34760177e-01 -1.11491263e+00 2.71358371e-01
6.02528453e-01 -6.05820060e-01 -1.01634049e+00 3.96427333e-01
-6.80667581e-03 2.89992571e-01 4.79612440e-01 -1.05351639e+00
-9.42984581e-01 4.07824576e-01 -8.09336424e-01 7.48283744e-01
1.99353367e-01 4.35544312e-01 1.26935279e+00 -5.52654743e-01
4.93045777e-01 1.34450722e+00 6.29767895e-01 5.39835870e-01
-1.33462119e+00 -4.27985549e-01 7.62337029e-01 5.90259790e-01
-1.06018245e+00 -2.53409237e-01 6.94875419e-01 -5.52545488e-01
1.34069455e+00 -2.44726669e-02 8.85984898e-01 1.29870784e+00
4.18286711e-01 1.15748000e+00 6.59838021e-01 -1.03817508e-01
-3.04787625e-02 -8.02406132e-01 2.81403899e-01 5.47259450e-01
-4.74724144e-01 1.18350312e-01 -9.80814517e-01 2.31957108e-01
1.23260605e+00 2.77246565e-01 -3.74422744e-02 -3.22008550e-01
-1.45595455e+00 5.49982429e-01 5.72312534e-01 3.60713124e-01
-5.31312823e-01 9.02311385e-01 5.28522730e-01 1.86849087e-01
5.63674629e-01 1.63497850e-01 -5.00524521e-01 -4.81518745e-01
-7.96397805e-01 6.08269423e-02 1.16838783e-01 8.52006495e-01
5.39795101e-01 2.05381270e-02 -6.15635276e-01 4.35213655e-01
6.20897822e-02 5.46791777e-02 3.03880036e-01 -1.29541266e+00
3.17666680e-01 2.40622029e-01 5.08908965e-02 -7.17753649e-01
-3.04303914e-01 -4.17399049e-01 -7.76178956e-01 -1.13383137e-01
4.14354146e-01 1.85680747e-01 -1.08583999e+00 2.06822705e+00
8.27622861e-02 9.00698900e-01 -1.88036904e-01 8.33276212e-01
3.12186748e-01 7.57325649e-01 4.01639044e-01 -7.43641332e-02
1.13186097e+00 -1.42973351e+00 -7.39958942e-01 -2.05171436e-01
7.55172968e-01 -1.92359015e-01 8.41707408e-01 1.11753426e-01
-1.22371948e+00 -7.62775481e-01 -7.11037338e-01 -3.82417262e-01
-6.84574768e-02 -9.67792049e-02 8.26627791e-01 1.51603890e-03
-1.09852850e+00 7.58609772e-01 -1.44967234e+00 -5.16840339e-01
5.99148810e-01 2.71436363e-01 -4.31935251e-01 -2.35156249e-02
-1.00637352e+00 5.90391040e-01 -6.64249957e-02 -4.32014912e-02
-1.28052318e+00 -9.26652193e-01 -9.33666646e-01 -5.92241250e-02
3.13535213e-01 -7.40471482e-01 1.30872202e+00 -1.00392604e+00
-1.32270324e+00 6.64676726e-01 -6.39716864e-01 -8.50668609e-01
2.43649289e-01 -5.19373417e-01 -2.43595392e-01 3.01527023e-01
1.73587188e-01 8.88476372e-01 7.09992349e-01 -8.39032948e-01
-3.52734923e-01 -3.69536847e-01 3.35615307e-01 5.45585752e-02
-1.54363558e-01 4.79464978e-02 -9.35756743e-01 -1.04325449e+00
1.40184751e-02 -9.33673322e-01 -2.32816428e-01 2.04501837e-01
-1.43890202e-01 -4.08490673e-02 9.24128532e-01 -5.72442055e-01
1.42218101e+00 -2.04027247e+00 5.51832795e-01 -2.38532022e-01
2.29787424e-01 -5.43992259e-02 -3.58521640e-01 4.52584535e-01
-3.85313392e-01 3.59775871e-02 -1.34234697e-01 -6.08800232e-01
-1.97817072e-01 5.92508614e-01 -3.51855546e-01 3.22704971e-01
5.32485843e-01 1.32546651e+00 -1.00473523e+00 -3.39219064e-01
2.95963228e-01 6.03944600e-01 -9.71214414e-01 3.99978191e-04
-3.35435659e-01 8.21025670e-01 -5.56926131e-01 6.31317139e-01
2.12527648e-01 -6.95289016e-01 1.77837804e-01 -3.65864336e-01
-1.11732379e-01 4.48656559e-01 -5.82028687e-01 2.03629804e+00
-4.35519516e-01 9.01261270e-01 -3.02681237e-01 -1.12646723e+00
2.96294838e-01 3.62215996e-01 1.17525291e+00 -8.62570405e-01
-1.68251731e-02 -3.67527306e-01 -1.14162348e-01 -5.60414910e-01
3.36761117e-01 1.65002942e-02 3.21866386e-02 6.70273066e-01
1.83144286e-01 6.50232255e-01 6.14360459e-02 3.43324512e-01
1.27605951e+00 5.63086629e-01 9.66979116e-02 -1.80178344e-01
2.36502647e-01 -2.47947037e-01 6.53801143e-01 7.75425196e-01
-4.09613043e-01 6.55828357e-01 5.80922842e-01 -8.38985741e-01
-9.85926032e-01 -9.11729455e-01 2.34791175e-01 1.28627086e+00
8.12809095e-02 -6.72844410e-01 -4.61674660e-01 -7.81160951e-01
-6.53283894e-02 3.99262667e-01 -1.19479942e+00 -4.76936474e-02
-8.05025160e-01 -4.49029893e-01 2.99268246e-01 1.14106739e+00
5.17461479e-01 -8.94439816e-01 -7.84581780e-01 3.00900131e-01
-3.05486113e-01 -1.21047270e+00 -9.02478635e-01 1.94694266e-01
-9.05774653e-01 -9.71404374e-01 -7.73516893e-01 -3.55755329e-01
2.70573407e-01 5.10805547e-01 1.15036607e+00 -1.04448564e-01
-2.03289583e-01 6.91343784e-01 -4.68923271e-01 1.63447365e-01
2.22253427e-01 -1.32732168e-01 -7.61783794e-02 3.12823236e-01
4.15445060e-01 -8.68594170e-01 -6.32748306e-01 3.48655462e-01
-7.89754093e-01 1.52142659e-01 3.16839963e-01 8.95927310e-01
5.39050519e-01 -2.90749192e-01 1.40793055e-01 -4.39510167e-01
-4.62034121e-02 -5.92784047e-01 -1.70914680e-01 1.24766566e-01
1.42717093e-01 2.99479812e-01 2.88825244e-01 -6.28297508e-01
-9.63322222e-01 -5.41719086e-02 1.47586882e-01 -1.00545192e+00
5.05546890e-02 4.34251368e-01 1.59738272e-01 6.18184507e-02
2.10758880e-01 3.62261385e-01 -8.55554081e-03 -3.85641575e-01
3.88338745e-01 -8.84758011e-02 4.50233817e-01 -6.97107911e-01
3.77183199e-01 8.93756330e-01 4.64420617e-02 -7.06616640e-01
-1.08094120e+00 -4.08629388e-01 -1.00774360e+00 -3.05157840e-01
1.34000778e+00 -1.02557671e+00 -7.64160216e-01 6.13586485e-01
-1.11718357e+00 -1.01000941e+00 -3.24408829e-01 5.90676844e-01
-8.40229809e-01 2.75696933e-01 -8.36506665e-01 -5.29538631e-01
2.87107229e-01 -1.10010326e+00 1.36238289e+00 -2.62367249e-01
-2.15938270e-01 -1.16322923e+00 1.17392071e-01 2.88669825e-01
1.45892397e-01 1.01391718e-01 5.67045033e-01 -2.96229180e-02
-1.04544628e+00 2.89766163e-01 -2.33750015e-01 8.60464722e-02
1.07714735e-01 3.49676944e-02 -7.58729994e-01 -1.12939291e-01
-2.14716509e-01 -2.52275020e-01 1.25765693e+00 8.58873010e-01
1.41595674e+00 -2.33774930e-01 -3.81170154e-01 8.12424719e-01
1.03886890e+00 2.62531757e-01 8.27136219e-01 2.98371464e-01
9.80955184e-01 4.27167386e-01 5.17204821e-01 6.16596758e-01
5.10814250e-01 9.89081800e-01 3.59555632e-01 -2.74614841e-02
-1.88029617e-01 -1.92017570e-01 4.79033887e-01 3.39307308e-01
-3.33731592e-01 -2.89612919e-01 -9.32045817e-01 7.27583528e-01
-2.15643501e+00 -1.46336627e+00 -6.27425089e-02 1.91128790e+00
5.01276553e-01 -1.44531233e-02 3.84627342e-01 -2.17224643e-01
3.29076529e-01 7.43686795e-01 -6.17469311e-01 -8.51830840e-02
-5.60465194e-02 -2.50352882e-02 5.50970674e-01 6.03791475e-01
-1.32142293e+00 1.08455157e+00 7.07514620e+00 4.27750081e-01
-1.05146658e+00 1.75877005e-01 7.62733161e-01 -4.89083767e-01
-2.79973716e-01 7.54964277e-02 -4.62625921e-01 3.99493873e-01
7.88718998e-01 8.50258321e-02 3.29657137e-01 2.88220406e-01
4.64220047e-01 1.40012264e-01 -1.30657625e+00 9.06027496e-01
-1.50652021e-01 -1.59116518e+00 2.46437132e-01 1.92470878e-01
9.19135869e-01 1.30088821e-01 1.72219321e-01 1.71119943e-01
3.33614260e-01 -1.00167346e+00 7.51034200e-01 7.85136402e-01
4.24356431e-01 -5.03262281e-01 1.99256986e-01 8.28334019e-02
-1.62307358e+00 -2.76033789e-01 1.32253900e-01 -3.22611749e-01
4.88988400e-01 1.97817013e-02 -5.44092916e-02 4.38530087e-01
8.99645269e-01 1.54836011e+00 -3.67126793e-01 6.54037535e-01
2.79370546e-02 5.35914660e-01 -8.06924775e-02 4.92970854e-01
8.38002682e-01 -3.46409790e-02 3.77853930e-01 1.16064894e+00
1.64986625e-01 5.03322482e-01 1.27232343e-01 6.48829579e-01
1.82362914e-01 -3.75494361e-01 -5.22383213e-01 -1.28405020e-01
2.89181098e-02 5.66341519e-01 -3.55877101e-01 -5.43887556e-01
-8.45835745e-01 1.29380322e+00 4.55970109e-01 7.05127776e-01
-1.29831219e+00 4.99083698e-01 1.18809676e+00 2.03236729e-01
7.12898731e-01 -7.14029491e-01 -9.76272002e-02 -1.34703159e+00
-2.81733014e-02 -6.10636532e-01 5.84355950e-01 -9.28837657e-01
-1.09428561e+00 2.81049639e-01 4.83996086e-02 -1.36126256e+00
-1.62880078e-01 -5.15333235e-01 -6.52466834e-01 6.23463273e-01
-1.25117266e+00 -1.42314577e+00 -2.49077305e-01 8.04924190e-01
8.91269803e-01 2.23315775e-01 3.92981529e-01 3.07884157e-01
-6.20145679e-01 5.71860731e-01 -2.17555031e-01 2.00143933e-01
5.72390854e-01 -1.08291841e+00 6.46640956e-01 9.64464724e-01
3.62106889e-01 6.33374453e-01 4.82084066e-01 -5.89569330e-01
-1.34718800e+00 -1.26718795e+00 7.75480270e-01 -6.14724100e-01
9.51318920e-01 -2.82723345e-02 -9.64632809e-01 1.29905593e+00
3.32015783e-01 4.41193759e-01 5.50038517e-01 2.84675121e-01
-7.16201901e-01 2.60299128e-02 -3.59871507e-01 6.15695894e-01
1.57139933e+00 -9.16653991e-01 -3.08126211e-01 3.10613751e-01
7.70691991e-01 -1.71781212e-01 -1.01398408e+00 2.85566092e-01
7.71466494e-01 -1.10383177e+00 1.12805390e+00 -1.06338441e+00
7.12067485e-01 -2.72406280e-01 -2.95659333e-01 -8.31226110e-01
-7.83716440e-01 -8.00044596e-01 -5.21325290e-01 8.01933467e-01
1.58604950e-01 -2.69638002e-01 8.97448480e-01 5.88061452e-01
-3.57042789e-01 -7.99422979e-01 -8.36218357e-01 -8.79766226e-01
8.84243846e-02 -7.05601454e-01 2.83577710e-01 1.00193524e+00
1.46038786e-01 1.19190626e-01 -8.06376100e-01 2.71757782e-01
5.13177693e-01 6.21746331e-02 6.50698841e-01 -5.03332198e-01
-5.69740057e-01 -5.62703609e-01 -6.16581202e-01 -1.93957126e+00
2.46352971e-01 -4.55418736e-01 -1.54940948e-01 -1.27601004e+00
4.19903725e-01 -2.65535805e-02 -5.48670590e-01 6.04683638e-01
-1.26084849e-01 3.90019417e-01 1.37779057e-01 3.04821342e-01
-8.74996126e-01 8.07806015e-01 1.39892292e+00 -2.69431233e-01
-1.09160155e-01 -3.39079767e-01 -4.24672902e-01 4.50946778e-01
4.37968016e-01 -1.54296562e-01 -6.06648266e-01 -9.14704382e-01
-3.02299887e-01 3.52039635e-01 7.70577729e-01 -1.06176984e+00
3.03742796e-01 -3.17375422e-01 3.21795970e-01 -7.58626044e-01
6.69297397e-01 -5.88047802e-01 1.24352224e-01 2.17679352e-01
-6.02539361e-01 8.07902887e-02 2.64971405e-01 8.68246615e-01
-4.47400451e-01 6.52634740e-01 6.67668700e-01 -1.83808744e-01
-9.59294498e-01 8.85064006e-01 -3.82523566e-01 -4.69269603e-02
1.10344470e+00 -2.94308454e-01 -9.64357406e-02 -6.05648994e-01
-1.23434973e+00 3.44538242e-01 4.31827188e-01 5.23758352e-01
3.48819762e-01 -1.56788611e+00 -4.14598137e-01 2.04111129e-01
1.49143890e-01 -3.82329583e-01 8.75339508e-01 1.12376106e+00
-2.20188603e-01 7.23519206e-01 -2.75069863e-01 -9.44197953e-01
-1.21220767e+00 7.70384908e-01 5.11637509e-01 -3.47923517e-01
-7.59699643e-01 1.03951538e+00 7.74619520e-01 2.27126613e-01
2.75992334e-01 -6.09686792e-01 1.06096528e-02 -1.19389810e-01
5.72826743e-01 7.23566338e-02 -4.67898607e-01 -8.75983715e-01
-3.86370391e-01 8.19397271e-01 -9.28421021e-02 -2.36802056e-01
1.33948505e+00 -2.98559934e-01 1.71489999e-01 5.49619973e-01
1.38506687e+00 -4.70917910e-01 -2.17751193e+00 -4.32236254e-01
-2.62805581e-01 -6.58470333e-01 9.26663727e-02 -4.28424597e-01
-1.24936008e+00 1.08395398e+00 1.56936288e-01 2.14984789e-01
1.20254779e+00 3.19188982e-01 7.07513988e-01 9.45559368e-02
2.12378219e-01 -7.57231057e-01 6.56168163e-01 7.28120446e-01
9.26720917e-01 -1.15900850e+00 -1.85980618e-01 -1.40999407e-01
-7.80133128e-01 9.97563243e-01 6.76287770e-01 -1.12453125e-01
7.49089420e-01 5.34508377e-02 -3.22834194e-01 -2.93329686e-01
-1.25229406e+00 -3.71701926e-01 4.16941494e-01 3.91527593e-01
5.87687612e-01 -1.75464854e-01 1.94685251e-01 2.61368841e-01
5.02352953e-01 6.30672053e-02 1.19738966e-01 9.48538363e-01
-5.60151190e-02 -1.05113816e+00 1.75607693e-03 1.43440962e-01
-3.60414594e-01 -1.32400006e-01 -8.25884417e-02 8.23530793e-01
-5.58979362e-02 5.97516298e-01 5.00638485e-01 -5.01775444e-01
-7.68091821e-04 -2.36551017e-02 6.88192129e-01 -4.06284809e-01
-2.55835235e-01 2.30637744e-01 1.81236099e-02 -1.25582325e+00
-7.31632411e-01 -9.98567104e-01 -1.09601009e+00 -2.24715590e-01
3.47383916e-01 -2.07858086e-01 5.15320599e-02 9.85997617e-01
6.78568423e-01 8.72474194e-01 3.89599025e-01 -1.03323913e+00
-4.76217121e-02 -7.89514899e-01 -3.05100977e-01 5.33867598e-01
7.70714939e-01 -7.89864838e-01 2.10018479e-03 4.34372306e-01] | [8.796674728393555, 0.5911589860916138] |
7009f50b-d8b0-40e6-847a-7e797b9bf77c | focus-or-not-a-baseline-for-anomaly-event | 2303.11668 | null | https://arxiv.org/abs/2303.11668v2 | https://arxiv.org/pdf/2303.11668v2.pdf | Focus or Not: A Baseline for Anomaly Event Detection On the Open Public Places with Satellite Images | In recent years, monitoring the world wide area with satellite images has been emerged as an important issue. Site monitoring task can be divided into two independent tasks; 1) Change Detection and 2) Anomaly Event Detection. Unlike to change detection research is actively conducted based on the numerous datasets(\eg LEVIR-CD, WHU-CD, S2Looking, xView2 and etc...) to meet up the expectations of industries or governments, research on AI models for detecting anomaly events is passively and rarely conducted. In this paper, we introduce a novel satellite imagery dataset(AED-RS) for detecting anomaly events on the open public places. AED-RS Dataset contains satellite images of normal and abnormal situations of 8 open public places from all over the world. Each places are labeled with different criteria based on the difference of characteristics of each places. With this dataset, we introduce a baseline model for our dataset TB-FLOW, which can be trained in weakly-supervised manner and shows reasonable performance on the AED-RS Dataset compared with the other NF(Normalizing-Flow) based anomaly detection models. Our dataset and code will be publicly open in \url{https://github.com/SIAnalytics/RS_AnomalyDetection.git}. | ['Junsik Kim', 'Hyunguk Choi', 'Doyoung Jeong', 'Youngtack Oh', 'Yongjin Jeon'] | 2023-03-21 | null | null | null | null | ['change-detection'] | ['computer-vision'] | [-1.35509238e-01 -3.08103025e-01 4.51813191e-02 -2.57296950e-01
-4.29022312e-01 -3.50117952e-01 9.05677617e-01 4.16231573e-01
-3.24252516e-01 5.30478597e-01 8.73747319e-02 -4.52857137e-01
2.12712660e-02 -1.06847143e+00 -3.03060502e-01 -7.19660103e-01
-5.56845367e-01 2.42070794e-01 4.22824860e-01 -3.75863105e-01
3.49875480e-01 7.29202449e-01 -1.67167425e+00 6.84361607e-02
9.75570619e-01 1.19050276e+00 -2.73764044e-01 3.24569821e-01
-3.52026582e-01 3.56720835e-01 -4.64500636e-01 1.61267877e-01
6.39060736e-01 -2.78667271e-01 -8.41579258e-01 -4.68200445e-02
3.79204333e-01 -4.28958982e-01 -4.36066657e-01 1.20104790e+00
3.96002591e-01 1.04539596e-01 7.26954937e-01 -1.55723822e+00
-3.92958432e-01 2.02247739e-01 -8.89407337e-01 1.22220504e+00
2.37428695e-01 1.78135738e-01 8.45908284e-01 -7.33652413e-01
3.92971307e-01 9.73166108e-01 6.74711049e-01 9.15308446e-02
-9.78565633e-01 -6.77686095e-01 6.09442353e-01 4.59561259e-01
-1.62676811e+00 -1.65413305e-01 5.11972487e-01 -5.66563487e-01
1.10435772e+00 5.39164007e-01 6.29799306e-01 1.06723905e+00
1.57945141e-01 7.89185405e-01 7.73029387e-01 -1.79980248e-01
3.68375778e-01 -2.12141842e-01 1.93867728e-01 3.20712000e-01
2.61993289e-01 -1.81819528e-01 -2.31441841e-01 -4.73059595e-01
5.03085852e-01 3.62619728e-01 -3.60105306e-01 2.18151808e-01
-1.12952411e+00 6.79162800e-01 3.73333961e-01 4.91738141e-01
-6.07923329e-01 -3.28096390e-01 5.42556167e-01 7.60642409e-01
1.08549368e+00 2.27305442e-02 -4.66113001e-01 -7.62189105e-02
-7.22853601e-01 3.90587449e-01 5.04834294e-01 5.88751554e-01
6.20027363e-01 2.21083924e-01 -1.47938719e-02 8.23563933e-01
5.54642260e-01 6.30202174e-01 6.69861794e-01 -3.28309625e-01
3.92403841e-01 9.44801688e-01 2.19786137e-01 -1.17058682e+00
-4.77303624e-01 -1.52608648e-01 -1.07182574e+00 1.62646279e-01
1.80917799e-01 -8.09077322e-02 -1.16933382e+00 1.29225445e+00
4.40024555e-01 3.90017271e-01 -2.38424510e-01 5.24856746e-01
8.21208656e-01 9.63931918e-01 1.65903643e-01 -7.86752105e-02
1.05839157e+00 -5.45016229e-01 -8.17977190e-01 -3.30419570e-01
7.73696005e-01 -4.11140203e-01 8.31255138e-01 2.49269024e-01
-3.71421099e-01 -6.61421791e-02 -7.68187582e-01 5.58962584e-01
-9.03960705e-01 -3.98351759e-01 3.83944213e-01 3.29402626e-01
-1.10102534e+00 1.66700155e-01 -1.16347325e+00 -1.05698121e+00
6.97252750e-01 1.47822555e-02 -4.59007621e-01 2.67597556e-01
-1.07374895e+00 7.59714663e-01 4.79016602e-01 1.06984273e-01
-8.42792392e-01 -6.19516134e-01 -8.87984157e-01 -2.54687250e-01
1.76606119e-01 2.51305327e-02 9.78361309e-01 -9.14691865e-01
-8.08957875e-01 1.09412789e+00 2.98134498e-02 -4.52450842e-01
4.31103379e-01 -2.73562163e-01 -1.07024431e+00 -1.66648492e-01
5.76122820e-01 1.63522035e-01 4.38631296e-01 -9.81582701e-01
-9.91825223e-01 -7.40867615e-01 -3.94464970e-01 4.32516821e-02
-2.31564388e-01 2.65341014e-01 -4.82666161e-04 -8.85582983e-01
3.50305825e-01 -8.16643596e-01 -2.74873793e-01 -5.84588982e-02
-4.72246230e-01 -3.63437891e-01 1.27477789e+00 -8.32538426e-01
1.64291990e+00 -2.25597811e+00 -5.87261796e-01 5.91236651e-01
-2.34033734e-01 4.05436784e-01 -2.06510887e-01 4.91492391e-01
-4.47216481e-01 3.72067481e-01 -7.05332637e-01 1.79611370e-02
-2.12133139e-01 3.10018390e-01 -4.70728725e-01 7.32667327e-01
1.84245273e-01 3.69639784e-01 -8.37996781e-01 -1.77497745e-01
2.36287668e-01 -3.29371542e-01 -3.45355481e-01 1.29258558e-01
-7.38251880e-02 5.93944490e-01 -8.12720358e-01 1.30082858e+00
8.22889686e-01 -1.24701532e-02 -4.39814806e-01 2.63117343e-01
-5.00510573e-01 5.42900852e-05 -1.41225028e+00 1.44877064e+00
6.63941652e-02 5.07276297e-01 -2.10403323e-01 -1.15285265e+00
9.92736816e-01 4.07241106e-01 7.09037602e-01 -7.52223372e-01
-1.66814223e-01 3.25949788e-01 -1.94549322e-01 -6.40634298e-01
3.06985915e-01 2.55740672e-01 -6.02787472e-02 4.88550156e-01
-2.24879459e-01 2.08641604e-01 1.97119430e-01 1.31402686e-01
1.35423231e+00 2.63309199e-02 6.07965052e-01 -4.28653032e-01
7.73353994e-01 3.14546168e-01 5.57099998e-01 8.70512187e-01
-7.06930637e-01 4.02873456e-01 2.86947668e-01 -9.43148673e-01
-8.61780941e-01 -1.15778184e+00 -6.38244927e-01 9.26896572e-01
3.69237214e-02 -2.61715919e-01 -1.45588517e-01 -7.95315623e-01
1.95269719e-01 8.76563489e-01 -6.32346392e-01 1.66243598e-01
-2.43678391e-01 -1.41851878e+00 7.64118314e-01 2.27116570e-01
1.01830852e+00 -1.22065532e+00 -1.33520335e-01 6.92367479e-02
-2.83533186e-01 -6.62802875e-01 5.21569587e-02 5.40126860e-02
-8.38774025e-01 -1.01917899e+00 -2.56061018e-01 -4.46396917e-01
5.93828440e-01 2.51092166e-01 1.06960976e+00 -2.52285730e-02
-3.10793698e-01 4.33506608e-01 -3.88409913e-01 -7.88462996e-01
-1.06014006e-01 5.23499027e-03 2.95514077e-01 1.67287320e-01
1.05265033e+00 -8.30896974e-01 -5.38457155e-01 5.72113276e-01
-1.25638044e+00 -6.06573999e-01 2.60499388e-01 2.68713355e-01
7.20804632e-01 7.88496360e-02 5.88288784e-01 -6.37498975e-01
2.85568237e-01 -1.34926772e+00 -5.60218751e-01 -1.04464307e-01
-4.67995822e-01 -3.35874021e-01 9.57952738e-02 1.17246576e-01
-1.01406133e+00 -2.12576509e-01 -2.89175510e-01 -1.46837905e-01
-9.48192537e-01 6.29648209e-01 -4.59155664e-02 1.99547842e-01
8.42476130e-01 3.49664479e-01 -1.09283663e-01 -7.68721998e-01
-1.63200200e-01 1.10645294e+00 5.60437202e-01 -2.85402656e-01
1.15833902e+00 7.70880282e-01 -3.64735186e-01 -1.18071353e+00
-8.39224696e-01 -9.95689452e-01 -6.09907746e-01 -2.07868114e-01
7.73123622e-01 -1.20816970e+00 7.81914294e-02 9.42622006e-01
-7.35000372e-01 -2.38654241e-01 -2.16590002e-01 3.21195781e-01
3.65111046e-02 4.01935786e-01 -2.77971447e-01 -8.16169202e-01
-3.37729782e-01 -5.17681539e-01 9.69945431e-01 1.23396248e-01
-1.68235779e-01 -1.01170993e+00 6.77359819e-01 -3.83292772e-02
7.37673700e-01 6.95009530e-01 3.93794298e-01 -1.17386305e+00
-3.89271259e-01 -4.06864107e-01 -8.80918577e-02 3.76296461e-01
5.47381401e-01 1.41815558e-01 -8.90600145e-01 -1.86515629e-01
-7.38722906e-02 1.27447903e-01 8.22682738e-01 3.58143568e-01
1.48912179e+00 -4.05161977e-01 -4.45744097e-01 5.39879501e-01
1.35617483e+00 2.90324867e-01 1.02882671e+00 1.30023634e+00
3.89250904e-01 1.72500148e-01 6.49026453e-01 8.03816438e-01
2.39361495e-01 3.60654235e-01 8.06216598e-01 2.42979340e-02
5.36608338e-01 2.71525502e-01 5.77208102e-01 2.70251602e-01
-4.89977419e-01 -4.42483872e-01 -1.47088242e+00 9.19809461e-01
-1.98425019e+00 -1.44761884e+00 -5.60038745e-01 2.13223648e+00
3.25350821e-01 -6.28303289e-02 7.46351779e-02 2.09097520e-01
7.46252060e-01 5.80090165e-01 -5.47155440e-01 -2.00036451e-01
-2.71837592e-01 -1.68433324e-01 4.06114280e-01 1.46360416e-02
-1.72240663e+00 6.15439892e-01 5.80029440e+00 5.50740421e-01
-1.08265567e+00 -4.16346677e-02 6.51346922e-01 2.72819791e-02
-1.12609170e-01 -3.51495385e-01 -6.99830115e-01 7.00634420e-01
8.71594191e-01 -1.07291810e-01 -1.42597616e-01 8.80593300e-01
4.92437094e-01 -2.54218042e-01 -6.05603874e-01 7.75159001e-01
-5.50380396e-03 -8.31940174e-01 2.65308559e-01 1.41585752e-01
8.17506015e-01 8.57174397e-01 -2.04475224e-01 3.20907950e-01
2.08450139e-01 -7.44615555e-01 2.81201571e-01 4.37626660e-01
4.50712949e-01 -4.51556295e-01 1.11679387e+00 2.55892962e-01
-1.16384137e+00 -9.51283276e-02 -7.78822377e-02 -1.14679985e-01
1.15935421e-02 7.82167256e-01 -3.70514423e-01 7.82161832e-01
1.55989587e+00 1.16131151e+00 -8.09308648e-01 1.33959711e+00
-1.47908172e-02 9.83583152e-01 -7.65885472e-01 5.06801128e-01
5.02042592e-01 -4.82043952e-01 1.09495282e+00 1.10349381e+00
7.41702080e-01 1.44337118e-01 2.14068919e-01 4.93206114e-01
3.32138807e-01 1.45152330e-01 -1.04069006e+00 2.19294041e-01
2.29862660e-01 1.13327789e+00 -4.68357563e-01 -5.35688400e-02
-7.37003565e-01 9.23319936e-01 -2.63897210e-01 3.38943899e-01
-7.36751556e-01 -2.57855594e-01 1.12299001e+00 3.06866407e-01
1.23798095e-01 -2.26683840e-02 1.98652204e-02 -1.37696052e+00
3.84556665e-03 -7.87150681e-01 9.97510791e-01 -4.88345236e-01
-1.65739155e+00 4.85536367e-01 1.67802796e-01 -1.67062974e+00
1.38477847e-01 -4.08417195e-01 -1.28958905e+00 6.17760122e-01
-1.78174388e+00 -8.93633902e-01 -7.42764056e-01 9.51144397e-01
5.16970754e-01 -3.89649540e-01 9.10977662e-01 5.89185834e-01
-1.03455329e+00 8.53790492e-02 3.21072519e-01 3.69908273e-01
6.21484399e-01 -1.30733776e+00 6.81521535e-01 1.27705276e+00
-9.87171009e-02 8.10792744e-02 4.88895327e-01 -8.07413101e-01
-4.28607076e-01 -1.54299843e+00 7.45643795e-01 -5.01842678e-01
1.03445721e+00 1.07127883e-01 -1.32553875e+00 1.19067407e+00
1.31731570e-01 3.78751457e-01 9.12823260e-01 -2.55092289e-02
-7.77667912e-04 -1.34774372e-01 -1.38714480e+00 4.29547727e-01
1.10711229e+00 -1.36128724e-01 -7.38740325e-01 4.55282986e-01
6.50132298e-02 -1.13149665e-01 -7.08658636e-01 4.13150012e-01
-1.82934582e-01 -8.38526964e-01 7.30142236e-01 -7.42547452e-01
5.37580624e-02 -7.26358712e-01 -2.97278345e-01 -1.28812349e+00
-3.72655720e-01 -2.01766506e-01 -1.24745362e-01 1.32238936e+00
2.70002007e-01 -1.07145202e+00 4.38617349e-01 3.11522037e-01
-3.51780206e-01 -2.20913813e-01 -1.15428245e+00 -8.49768519e-01
-3.00961405e-01 -5.58822095e-01 7.17196882e-01 1.24656904e+00
-2.75582463e-01 -2.49329597e-01 -4.13230546e-02 9.08710420e-01
5.08711696e-01 -9.95364934e-02 7.71478176e-01 -1.63796818e+00
3.78860116e-01 -3.75089645e-01 -9.06740010e-01 -5.78369617e-01
-1.77301213e-01 -7.79934824e-01 -2.13202879e-01 -1.43708038e+00
-9.13290009e-02 -5.78312635e-01 -5.34413993e-01 9.98079360e-01
-6.30731508e-02 2.90662646e-01 -5.87128937e-01 5.78724086e-01
-5.57380974e-01 6.56527340e-01 5.58016598e-01 -1.86519846e-01
-3.17590505e-01 2.33245179e-01 -2.75417209e-01 9.48240817e-01
1.14003432e+00 -3.99286330e-01 1.48435295e-01 -2.36824930e-01
2.51986533e-01 -5.62133074e-01 4.48875993e-01 -1.21498322e+00
-1.10104911e-01 -2.91556180e-01 2.56576598e-01 -9.13316011e-01
-3.98377806e-01 -8.27452660e-01 1.26449451e-01 4.96024370e-01
2.05370098e-01 3.34793836e-01 4.16507870e-01 7.70964980e-01
-5.62999368e-01 -3.84011567e-02 6.33743644e-01 -2.40965083e-01
-1.25484335e+00 8.26542020e-01 -7.87695706e-01 -6.33130968e-02
1.21060884e+00 -8.60210061e-02 -3.94429028e-01 -2.54816473e-01
-6.15588546e-01 5.92619777e-01 3.18524897e-01 5.36993861e-01
2.32357964e-01 -1.40319383e+00 -1.04036117e+00 3.26202750e-01
6.50642395e-01 3.16803515e-01 2.66425312e-01 8.65356922e-01
-8.44871998e-01 2.09173828e-01 -4.32490289e-01 -7.47783363e-01
-9.62910950e-01 1.55910268e-01 5.12691617e-01 -2.35068262e-01
-7.42882133e-01 3.84850979e-01 -1.05866015e-01 -6.11596346e-01
1.20384388e-01 -1.97809473e-01 -5.07806242e-01 1.91335812e-01
7.57453859e-01 5.00013053e-01 3.21972996e-01 -5.63958049e-01
-5.53197563e-01 8.95035118e-02 -1.73330143e-01 4.34407651e-01
1.84613943e+00 -2.66784191e-01 -2.33832613e-01 5.29979646e-01
7.14826107e-01 -2.05154970e-01 -6.85185909e-01 -4.06141132e-01
3.57020319e-01 -6.60638630e-01 2.16300249e-01 -6.00796998e-01
-1.09169006e+00 2.63574123e-01 1.15995479e+00 7.04399347e-01
1.14515483e+00 4.23143022e-02 5.99058807e-01 5.17401874e-01
1.52507886e-01 -1.07544553e+00 -1.46524698e-01 5.80731332e-01
1.02307940e+00 -1.69551754e+00 -7.96057805e-02 1.01321097e-02
-4.75332022e-01 8.04645717e-01 8.08769405e-01 -2.61566211e-02
9.55671012e-01 -8.00461993e-02 1.71159595e-01 -5.21811485e-01
-2.67670363e-01 -3.80838037e-01 4.88809682e-02 5.73727906e-01
1.09128423e-01 -1.28321633e-01 -1.01233013e-01 1.81480840e-01
2.44607851e-01 -2.71704674e-01 4.25052226e-01 1.14850855e+00
-6.28198266e-01 -8.24759722e-01 -6.19882107e-01 9.90709424e-01
-6.97544038e-01 -7.83188567e-02 -1.67066276e-01 8.49958241e-01
7.33661279e-02 7.69063652e-01 5.50523400e-01 9.16464329e-02
4.98054892e-01 2.35634327e-01 -4.86300647e-01 -4.69155282e-01
-3.59233558e-01 -5.37206769e-01 3.14810984e-02 -6.88554287e-01
-5.74250698e-01 -9.22093511e-01 -9.82904792e-01 -3.92150193e-01
-1.01164639e-01 4.37143743e-02 3.21574956e-01 7.84025431e-01
2.92008221e-01 8.36798549e-02 7.87696838e-01 -6.93666518e-01
-2.07572117e-01 -1.09514165e+00 -9.89292622e-01 6.02089643e-01
3.92468691e-01 -5.12955368e-01 -9.00344491e-01 -2.51372993e-01] | [7.61887788772583, 2.2360100746154785] |
1f6c5d9e-004b-4d3b-b3e3-267577c83a11 | closed-book-question-generation-via | 2210.06781 | null | https://arxiv.org/abs/2210.06781v2 | https://arxiv.org/pdf/2210.06781v2.pdf | Closed-book Question Generation via Contrastive Learning | Question Generation (QG) is a fundamental NLP task for many downstream applications. Recent studies on open-book QG, where supportive answer-context pairs are provided to models, have achieved promising progress. However, generating natural questions under a more practical closed-book setting that lacks these supporting documents still remains a challenge. In this work, we propose a new QG model for this closed-book setting that is designed to better understand the semantics of long-form abstractive answers and store more information in its parameters through contrastive learning and an answer reconstruction module. Through experiments, we validate the proposed QG model on both public datasets and a new WikiCQA dataset. Empirical results show that the proposed QG model outperforms baselines in both automatic evaluation and human evaluation. In addition, we show how to leverage the proposed model to improve existing question-answering systems. These results further indicate the effectiveness of our QG model for enhancing closed-book question-answering tasks. | ['James Caverlee', 'Jianling Wang', 'Jiaying Lu', 'Xiangjue Dong'] | 2022-10-13 | null | null | null | null | ['natural-questions', 'question-generation', 'open-domain-question-answering'] | ['miscellaneous', 'natural-language-processing', 'natural-language-processing'] | [ 3.40139261e-03 3.77995044e-01 -8.97921342e-03 -4.25367415e-01
-1.58703804e+00 -7.60229409e-01 6.58410251e-01 1.37898028e-01
-2.60773599e-01 8.25529516e-01 6.29233181e-01 -4.90720123e-01
-1.03059664e-01 -9.26196992e-01 -7.83573806e-01 7.55977854e-02
4.78435487e-01 7.06209898e-01 6.49576843e-01 -8.61435235e-01
4.18575644e-01 -3.87468398e-01 -1.41515315e+00 6.82596207e-01
1.45510244e+00 7.74636924e-01 3.35077256e-01 9.01184261e-01
-5.91253340e-01 1.17176199e+00 -6.77639127e-01 -9.71906006e-01
-9.91512462e-03 -4.88529086e-01 -1.33846772e+00 -3.14209819e-01
7.30242670e-01 -4.83705342e-01 -1.22099191e-01 6.14468873e-01
4.49091047e-01 4.21724111e-01 4.93414521e-01 -1.01012170e+00
-1.13547015e+00 7.51002133e-01 1.01872437e-01 1.85397968e-01
8.12801421e-01 1.37576744e-01 1.70609605e+00 -9.51642990e-01
6.41974568e-01 1.29148126e+00 3.80803913e-01 8.47341120e-01
-9.08839345e-01 -3.02122355e-01 4.12365124e-02 5.13156176e-01
-7.72351563e-01 -4.04567599e-01 6.03735566e-01 5.75100854e-02
1.13673651e+00 5.49923837e-01 2.69260049e-01 8.82877946e-01
-1.71156347e-01 1.05357170e+00 7.52107084e-01 -6.21287942e-01
-4.07772996e-02 4.23530340e-02 4.62246686e-01 6.09927237e-01
8.88248906e-03 -2.59795249e-01 -6.01573288e-01 -2.23997712e-01
2.45387420e-01 -3.60713571e-01 -4.29835081e-01 3.61316912e-02
-8.91227007e-01 1.00059831e+00 3.60171378e-01 -7.10372180e-02
-9.45257545e-02 1.57465339e-01 2.72910714e-01 4.64522630e-01
4.72162098e-01 1.13315618e+00 -5.25564492e-01 -1.65590152e-01
-6.63183570e-01 9.50215995e-01 1.31308627e+00 1.21977353e+00
4.52767044e-01 -6.26218557e-01 -9.85142171e-01 1.10557008e+00
2.98192799e-01 4.74039406e-01 3.43475431e-01 -1.34358728e+00
9.47236478e-01 8.40795338e-01 3.75284821e-01 -8.28598976e-01
-9.79954973e-02 -4.05250907e-01 -3.31294268e-01 -8.61768782e-01
5.33311546e-01 -1.56636029e-01 -3.25481921e-01 1.66663015e+00
4.01493877e-01 -6.05306961e-02 2.61026591e-01 7.74203062e-01
1.48024249e+00 1.01259315e+00 1.95755772e-02 1.84135169e-01
1.63381529e+00 -1.45671570e+00 -9.83948290e-01 -2.80295223e-01
8.28072906e-01 -7.12020278e-01 1.68927860e+00 5.32665476e-02
-1.15981531e+00 -5.55179894e-01 -6.69951916e-01 -7.74167061e-01
-1.11848995e-01 2.26072595e-01 3.40545177e-01 4.38334376e-01
-1.09790659e+00 9.95139629e-02 -1.83395386e-01 -4.43505406e-01
9.20582265e-02 -1.50426328e-01 6.17380925e-02 -4.73282158e-01
-1.62573981e+00 7.25782275e-01 1.59150019e-01 -6.02561310e-02
-5.45511425e-01 -7.84532964e-01 -8.89472783e-01 2.20752984e-01
7.13345230e-01 -1.22553647e+00 1.88600981e+00 -1.68668419e-01
-1.45200634e+00 6.65537536e-01 -4.47445095e-01 -5.33703446e-01
1.41017929e-01 -5.26418567e-01 -2.30181947e-01 4.38497365e-01
3.70525658e-01 8.08561325e-01 5.55069387e-01 -1.10934639e+00
-5.21523595e-01 -2.72481114e-01 7.68428922e-01 4.84279454e-01
-3.74898881e-01 1.23427073e-02 -4.89439815e-01 -5.22144377e-01
-5.81338303e-03 -5.13966739e-01 -9.47151482e-02 -2.82288224e-01
-4.18729365e-01 -9.76396501e-01 2.73518145e-01 -8.00508738e-01
1.48907804e+00 -1.58920717e+00 -2.31609151e-01 -2.42952168e-01
1.01425923e-01 1.73465729e-01 -4.71151173e-01 9.27960098e-01
5.84727526e-01 5.80429882e-02 -2.73998231e-01 -2.95578808e-01
3.25408697e-01 1.83170080e-01 -8.38574469e-01 -5.42484641e-01
4.32304472e-01 1.53127527e+00 -1.12711978e+00 -5.86263478e-01
-4.51946527e-01 -1.35330990e-01 -8.77246261e-01 7.43013501e-01
-9.02176142e-01 1.06364846e-01 -5.40351212e-01 4.62366611e-01
3.75777036e-01 -4.80628192e-01 -1.14562824e-01 2.31893018e-01
4.31571722e-01 9.38304961e-01 -7.62711287e-01 1.76811051e+00
-6.05848551e-01 1.81695223e-01 -1.26212701e-01 -6.02562666e-01
9.42618012e-01 2.49191359e-01 -3.30177158e-01 -9.75309134e-01
-2.66289294e-01 3.49898189e-01 -2.29416862e-01 -9.27726448e-01
1.15275335e+00 -8.79156142e-02 -1.38684258e-01 7.25034773e-01
1.58695728e-01 -6.81740403e-01 6.96602821e-01 8.31243932e-01
1.18841004e+00 3.65081251e-01 -3.53735611e-02 -1.30097404e-01
8.13409030e-01 1.15393803e-01 1.23558789e-01 1.16764510e+00
-1.22723483e-01 5.98376095e-01 5.01208544e-01 1.43226326e-01
-4.67849821e-01 -9.03144121e-01 1.67088583e-01 1.44302559e+00
1.00821927e-01 -6.76969886e-01 -8.79217029e-01 -1.00732911e+00
-1.33283243e-01 1.01792002e+00 -2.97443867e-01 -1.44148573e-01
-8.13523114e-01 -2.89579242e-01 6.19266212e-01 5.57279170e-01
4.84716743e-01 -1.18352234e+00 -2.90033847e-01 3.38405102e-01
-9.58958805e-01 -1.24753749e+00 -7.23362088e-01 -2.93583691e-01
-1.01703572e+00 -1.10298002e+00 -6.14894509e-01 -9.42764759e-01
3.30940098e-01 4.88296807e-01 1.87777007e+00 4.00775343e-01
3.55284870e-01 5.95630884e-01 -8.85973394e-01 -3.54965031e-01
-3.97040218e-01 5.55213332e-01 -7.55614281e-01 -3.11839283e-01
4.81463701e-01 -2.32038915e-01 -8.39036703e-01 2.81529635e-01
-1.05439675e+00 -2.13459115e-02 2.87057847e-01 9.12974834e-01
4.42179680e-01 -6.56984270e-01 1.48635018e+00 -1.33469212e+00
1.25043106e+00 -7.40808189e-01 -4.27839249e-01 6.06920123e-01
-4.41910714e-01 3.12686652e-01 7.44328260e-01 1.82681710e-01
-1.51657486e+00 -5.68283498e-01 -6.81452215e-01 4.36821282e-01
2.48265833e-01 5.05279422e-01 -3.19990665e-01 4.91170943e-01
7.89558113e-01 3.03964138e-01 -1.62904009e-01 -5.55782914e-01
1.05050755e+00 6.70587838e-01 6.08370364e-01 -9.60664093e-01
7.33790994e-01 1.60729647e-01 -5.26412904e-01 -5.10052741e-01
-1.55004203e+00 -8.59389663e-01 -3.72348465e-02 -4.77219839e-03
5.43231428e-01 -8.26592624e-01 -4.15654540e-01 -8.52013566e-03
-1.41257834e+00 -2.07095131e-01 -4.04875606e-01 -9.43447649e-02
-5.54145336e-01 4.90048289e-01 -6.98833406e-01 -7.84185946e-01
-8.07413578e-01 -7.36862361e-01 1.32394516e+00 3.23443562e-01
-4.07591879e-01 -1.09365153e+00 3.32306683e-01 1.32085621e+00
3.52041394e-01 -5.59371531e-01 1.17802918e+00 -8.51011336e-01
-9.32588816e-01 -3.30782533e-02 -2.42447242e-01 3.49704266e-01
-1.86072886e-01 -4.15548146e-01 -8.40801537e-01 1.64945483e-01
1.02571949e-01 -9.27371264e-01 1.13004136e+00 -5.84016331e-02
1.21754444e+00 -5.89542449e-01 1.08103216e-01 2.30243802e-01
1.07713330e+00 -2.01403901e-01 7.01190054e-01 2.53440976e-01
4.21532452e-01 1.12745094e+00 9.21678841e-01 1.23333924e-01
1.20412958e+00 3.95445764e-01 3.40478688e-01 3.02142233e-01
-3.37275356e-01 -7.70545006e-01 2.58976996e-01 1.12299132e+00
3.70192647e-01 -4.41561043e-01 -6.81688547e-01 8.72908294e-01
-1.82178378e+00 -9.30171609e-01 -4.24851894e-01 1.71484339e+00
1.30873775e+00 -2.19117522e-01 2.54805759e-03 -2.13443100e-01
3.04161787e-01 1.39404818e-01 -3.74626845e-01 -3.15293342e-01
-1.90315828e-01 6.53210163e-01 -3.58338803e-01 6.24179304e-01
-6.97760820e-01 1.03082442e+00 5.75098133e+00 7.75323093e-01
-1.78136572e-01 1.06424823e-01 4.40493733e-01 2.13367671e-01
-9.27741170e-01 1.47156969e-01 -1.08505285e+00 1.59067705e-01
8.53224456e-01 -2.64096588e-01 1.32534936e-01 7.04298973e-01
4.19718307e-03 -1.20423526e-01 -1.11004341e+00 6.89914763e-01
3.66193593e-01 -1.35043705e+00 4.06746745e-01 -4.36316967e-01
7.49097109e-01 -4.18037653e-01 -1.54697532e-02 9.86923218e-01
3.33698779e-01 -9.32419002e-01 4.61985499e-01 4.19313937e-01
3.97213817e-01 -5.13690889e-01 8.38457882e-01 6.16374314e-01
-9.13014948e-01 -1.71964526e-01 -4.63926136e-01 -1.93452045e-01
3.49079162e-01 2.97024548e-01 -9.96411502e-01 7.06621826e-01
5.04329622e-01 4.46346402e-01 -9.05129492e-01 8.98025751e-01
-8.07970464e-01 9.87910569e-01 8.71481076e-02 -5.16229033e-01
3.24976414e-01 -1.40289620e-01 3.03571373e-01 9.98100579e-01
2.07469597e-01 3.66054893e-01 -1.82745621e-01 1.11859453e+00
-5.92985630e-01 4.93888706e-01 -4.26655769e-01 3.10058258e-02
6.23369813e-01 1.28958774e+00 -9.46177542e-02 -3.40548575e-01
-5.39189816e-01 8.82559896e-01 6.77981734e-01 3.49429935e-01
-4.97335732e-01 -6.73573494e-01 2.36531034e-01 1.88481808e-01
7.17687383e-02 9.93682146e-02 -2.23028123e-01 -1.43434095e+00
4.89816457e-01 -1.17579186e+00 8.06797147e-01 -8.97251308e-01
-1.53562307e+00 3.93781900e-01 -2.63448432e-02 -8.05282593e-01
-7.33647823e-01 -3.67843002e-01 -5.43114007e-01 8.91595840e-01
-2.05212450e+00 -1.11397004e+00 -1.28723428e-01 4.21099961e-01
8.10274601e-01 9.71808881e-02 7.71948993e-01 1.79373212e-02
-2.01214209e-01 6.40242755e-01 2.69397479e-02 3.86505723e-02
8.28598022e-01 -1.50134957e+00 6.62034392e-01 9.07040536e-01
3.51733506e-01 9.49853480e-01 4.56717372e-01 -5.32329261e-01
-1.50305974e+00 -1.26557338e+00 1.51931334e+00 -1.01524925e+00
5.88510752e-01 -4.01847750e-01 -1.22430885e+00 4.80235636e-01
5.14774442e-01 -3.35252404e-01 8.65953147e-01 2.77761787e-01
-4.82944608e-01 -1.46026686e-01 -8.72620404e-01 5.90397298e-01
1.02896798e+00 -6.49608731e-01 -1.31995571e+00 4.77397352e-01
1.38428795e+00 -2.86576629e-01 -6.12280965e-01 2.19819903e-01
2.05241114e-01 -8.18416715e-01 8.29133451e-01 -7.22231150e-01
8.94080758e-01 -2.05129996e-01 -1.62407801e-01 -1.39945948e+00
1.23486586e-01 -6.74838424e-01 -5.59495032e-01 1.36455822e+00
6.14357710e-01 -4.49170440e-01 6.79571986e-01 6.29166186e-01
-3.86970639e-01 -9.75357652e-01 -7.82536268e-01 -7.13493407e-01
3.92791092e-01 -3.98045272e-01 8.54013860e-01 3.95522505e-01
1.56232893e-01 9.86375451e-01 7.72289112e-02 -2.30772078e-01
3.49460632e-01 3.68985832e-01 8.50705683e-01 -9.41820860e-01
-4.12726045e-01 -7.46285096e-02 4.55189437e-01 -1.80919683e+00
3.61935079e-01 -1.15961635e+00 2.54318595e-01 -2.09010196e+00
3.03377748e-01 -3.79814804e-01 1.82576805e-01 -3.04730516e-02
-1.01939738e+00 5.48531972e-02 1.60000026e-01 -7.39900619e-02
-1.18711030e+00 1.03644383e+00 1.58249545e+00 2.12120656e-02
1.08644582e-01 -4.54557836e-02 -1.26689351e+00 3.17981154e-01
6.46847963e-01 -2.33546495e-01 -8.86351526e-01 -8.44104350e-01
7.18038917e-01 3.79771769e-01 2.82865077e-01 -5.04953861e-01
2.80649334e-01 -2.13591997e-02 -2.83477485e-01 -6.74226701e-01
2.11823761e-01 -3.13684970e-01 -7.85753667e-01 -2.15198640e-02
-7.78038561e-01 5.01019880e-02 -7.64655620e-02 6.12033725e-01
-4.86330658e-01 -6.23511136e-01 2.37375855e-01 -2.37207472e-01
-5.26017427e-01 1.63906679e-01 -7.71051086e-03 1.09546471e+00
4.06419694e-01 1.22885861e-01 -7.40522742e-01 -9.25812066e-01
-8.76398757e-02 8.74160945e-01 -3.94367315e-02 5.39063811e-01
9.24833417e-01 -1.26153934e+00 -9.71379280e-01 -1.89545259e-01
8.15322876e-01 1.73397481e-01 2.73795724e-01 3.27354550e-01
-3.69440645e-01 8.27351511e-01 4.10215884e-01 -1.98351815e-01
-9.21252608e-01 1.86596498e-01 2.08007544e-01 -7.45937824e-01
-2.55314618e-01 1.14209878e+00 9.61025283e-02 -1.05352378e+00
7.41672665e-02 -4.82435524e-01 -6.38372719e-01 3.71884890e-02
7.58934855e-01 4.24310714e-01 2.88396001e-01 -1.62421644e-01
2.18954831e-01 1.03334159e-01 -2.35891759e-01 -3.17913055e-01
1.05595279e+00 -4.36960697e-01 -3.19097459e-01 1.35353953e-01
8.72328877e-01 -5.43966182e-02 -8.37198436e-01 -5.54276824e-01
4.14937347e-01 -3.26550871e-01 -4.89137739e-01 -1.00320423e+00
-3.62626314e-01 9.83807206e-01 -1.53870896e-01 1.85235530e-01
1.00788748e+00 3.56425852e-01 1.47797823e+00 9.61960077e-01
2.21799061e-01 -1.08895886e+00 5.73087871e-01 1.11433244e+00
1.16686964e+00 -1.25083554e+00 -5.89323461e-01 -4.71698791e-01
-5.60105205e-01 8.29065323e-01 1.11453056e+00 2.20286071e-01
5.14659248e-02 -4.69720542e-01 2.34693319e-01 -3.71887356e-01
-1.23172903e+00 -3.74414623e-01 3.84350896e-01 5.20796597e-01
7.54400671e-01 -1.87015846e-01 -4.91287619e-01 1.14248848e+00
-7.31556416e-01 -5.39198890e-02 5.91958225e-01 8.02128315e-01
-5.51712632e-01 -1.22848284e+00 -2.01908842e-01 6.42360866e-01
-4.44029301e-01 -5.69725633e-01 -5.24668992e-01 4.03435111e-01
-4.16529030e-01 1.55958140e+00 -2.73577958e-01 1.10597767e-01
6.25649393e-01 4.27999556e-01 4.96928364e-01 -1.16744530e+00
-9.14225698e-01 -6.02669775e-01 4.80828941e-01 -4.80647296e-01
-1.50803283e-01 -2.94036686e-01 -1.15576375e+00 1.12695694e-01
-4.95091558e-01 6.20060623e-01 2.69868881e-01 8.96418393e-01
5.60474753e-01 3.57381076e-01 4.04761463e-01 2.22385928e-01
-1.20862269e+00 -1.24809599e+00 -2.95266747e-01 5.63655794e-01
2.07173318e-01 -1.50992364e-01 -1.52698129e-01 5.39811030e-02] | [11.418457984924316, 8.063614845275879] |
f4e93679-3e47-43d6-a33c-5dc8e0b6572d | where-in-the-world-is-this-image-transformer | 2204.13861 | null | https://arxiv.org/abs/2204.13861v2 | https://arxiv.org/pdf/2204.13861v2.pdf | Where in the World is this Image? Transformer-based Geo-localization in the Wild | Predicting the geographic location (geo-localization) from a single ground-level RGB image taken anywhere in the world is a very challenging problem. The challenges include huge diversity of images due to different environmental scenarios, drastic variation in the appearance of the same location depending on the time of the day, weather, season, and more importantly, the prediction is made from a single image possibly having only a few geo-locating cues. For these reasons, most existing works are restricted to specific cities, imagery, or worldwide landmarks. In this work, we focus on developing an efficient solution to planet-scale single-image geo-localization. To this end, we propose TransLocator, a unified dual-branch transformer network that attends to tiny details over the entire image and produces robust feature representation under extreme appearance variations. TransLocator takes an RGB image and its semantic segmentation map as inputs, interacts between its two parallel branches after each transformer layer, and simultaneously performs geo-localization and scene recognition in a multi-task fashion. We evaluate TransLocator on four benchmark datasets - Im2GPS, Im2GPS3k, YFCC4k, YFCC26k and obtain 5.5%, 14.1%, 4.9%, 9.9% continent-level accuracy improvement over the state-of-the-art. TransLocator is also validated on real-world test images and found to be more effective than previous methods. | ['Rama Chellappa', 'Carlos D. Castillo', 'Joshua Gleason', 'Ewa M. Nowara', 'Shraman Pramanick'] | 2022-04-29 | null | null | null | null | ['scene-recognition'] | ['computer-vision'] | [ 1.38882408e-02 -4.13747221e-01 1.39644518e-01 -2.68978834e-01
-7.85302937e-01 -5.72913527e-01 4.78425980e-01 -3.46174613e-02
-3.62016857e-01 5.60405910e-01 -1.12258650e-01 -1.28287613e-01
-1.96299981e-02 -9.06585813e-01 -8.54004323e-01 -7.60558188e-01
3.39662582e-02 1.95022509e-01 4.09220934e-01 -1.58993497e-01
2.42414519e-01 3.19257915e-01 -1.54164565e+00 6.26341254e-02
9.23645139e-01 1.40980649e+00 3.72663647e-01 4.57107246e-01
-9.09749195e-02 4.08394754e-01 -3.14705431e-01 -8.85662958e-02
3.10678869e-01 2.22571455e-02 -5.90251267e-01 6.44957423e-02
5.30931890e-01 1.85730845e-01 -1.18702814e-01 1.05688393e+00
3.54063839e-01 -1.52795948e-02 5.57547331e-01 -1.23291135e+00
-6.82875216e-01 6.41816780e-02 -9.76053655e-01 4.93836813e-02
1.25065506e-01 -7.87103027e-02 5.80058932e-01 -7.00291514e-01
2.62096375e-01 9.31208789e-01 8.68948221e-01 -1.13162450e-01
-7.48183727e-01 -6.77504659e-01 3.02880317e-01 2.21412510e-01
-1.87349677e+00 -1.54716447e-01 6.72586977e-01 -2.48664811e-01
8.26217473e-01 2.35081419e-01 3.23216140e-01 6.72410190e-01
1.16324209e-01 4.50544149e-01 1.34726465e+00 -1.50468662e-01
1.60387725e-01 -2.21937206e-02 -1.98578864e-01 7.74781585e-01
8.43019560e-02 -4.18096542e-01 -4.28056717e-01 2.16778249e-01
4.81583357e-01 1.97320282e-01 -3.53394598e-01 -1.08502723e-01
-1.09410393e+00 3.33643764e-01 1.01942050e+00 2.40784004e-01
-4.46138650e-01 9.32231471e-02 -3.77955241e-03 -7.45043391e-03
5.53693712e-01 -8.37263092e-03 -5.15570104e-01 -1.28199374e-02
-9.65385199e-01 1.33948233e-02 3.32160592e-01 8.45112145e-01
9.81346309e-01 3.18129128e-03 2.82301456e-01 7.52090633e-01
3.20859253e-01 1.03718781e+00 4.98385310e-01 -3.50032508e-01
6.67315125e-01 6.61236286e-01 2.53726244e-01 -1.52459800e+00
-6.50899768e-01 -6.97130799e-01 -7.71035433e-01 -1.07404530e-01
2.93714464e-01 -4.52573560e-02 -1.12743378e+00 1.50025260e+00
4.51914042e-01 5.73010683e-01 3.35044153e-02 1.00014341e+00
8.37205410e-01 8.15669775e-01 1.10230438e-01 4.83043790e-01
1.32488096e+00 -1.10381043e+00 -2.38614175e-02 -4.84788030e-01
4.46408629e-01 -7.18711317e-01 9.19261992e-01 2.69005120e-01
-3.55361164e-01 -5.18036664e-01 -8.34593058e-01 1.03399932e-01
-9.60981727e-01 6.16591394e-01 6.57563090e-01 4.70628142e-01
-1.17193210e+00 2.49962240e-01 -5.71634114e-01 -7.18772769e-01
2.24892229e-01 3.07758927e-01 -6.13479316e-01 -3.98954719e-01
-9.24123347e-01 6.36839032e-01 4.70642507e-01 3.71816248e-01
-4.68806207e-01 -6.72755361e-01 -7.64284253e-01 -7.30280429e-02
1.52351320e-01 -4.29514676e-01 6.03462577e-01 -7.86838472e-01
-1.17185509e+00 8.79664421e-01 -1.00642599e-01 -2.30848357e-01
4.00969118e-01 -2.67302655e-02 -7.29175568e-01 -7.27102160e-02
5.44740081e-01 6.28229797e-01 5.13938606e-01 -1.10465848e+00
-1.03849053e+00 -6.52984023e-01 -4.55744416e-02 3.22573960e-01
-1.08888134e-01 -1.74031630e-01 -9.11147356e-01 -5.04829466e-01
4.68273252e-01 -1.11597693e+00 -1.37391299e-01 -2.30609819e-01
-5.89267612e-01 -1.33905085e-02 8.50943327e-01 -7.51022160e-01
7.81707168e-01 -2.27022457e+00 -7.83743337e-02 4.87320572e-02
-2.08439738e-01 1.63468257e-01 -6.66263252e-02 2.69014746e-01
1.56865552e-01 8.09110552e-02 -1.05948716e-01 -3.14018548e-01
-1.37967423e-01 -4.32083495e-02 -2.41473556e-01 7.21273124e-01
1.33219296e-02 8.55098307e-01 -7.07898796e-01 -3.68002951e-01
5.47163785e-01 5.81228435e-01 -1.10884465e-01 -1.11099854e-01
7.84963742e-02 5.74783146e-01 -6.08160913e-01 1.17534077e+00
9.65211630e-01 -3.10311824e-01 -2.52272010e-01 -3.75675708e-01
-4.44960862e-01 -2.83562809e-01 -1.22791731e+00 1.78125584e+00
-7.26069570e-01 5.15240073e-01 -2.07235947e-01 -9.26163971e-01
9.77214396e-01 -3.84344272e-02 3.36336315e-01 -1.26674366e+00
9.89549421e-03 3.76540273e-01 -5.93659580e-01 -4.27971959e-01
6.16427243e-01 1.77154541e-01 -4.52890009e-01 3.41231115e-02
-8.05164576e-02 1.05747297e-01 -3.05545870e-02 -1.24963515e-01
8.15833330e-01 8.29757527e-02 1.42544523e-01 -2.00510085e-01
5.51361918e-01 3.33188802e-01 5.65801322e-01 7.69415021e-01
-2.83369213e-01 1.02594721e+00 7.87151381e-02 -5.32648504e-01
-7.87448227e-01 -9.17218387e-01 -7.56906271e-02 9.59333897e-01
6.17946684e-01 -6.05856702e-02 -5.49971163e-01 -5.02805233e-01
3.02611440e-02 3.93800855e-01 -7.14516580e-01 1.21344090e-01
-3.25212389e-01 -1.15658045e+00 5.65167665e-01 4.24086034e-01
1.10931981e+00 -5.91041327e-01 -5.11630118e-01 1.23543538e-01
-3.88494730e-01 -1.47614133e+00 -1.40306950e-01 -1.15140587e-01
-4.72114503e-01 -1.05777419e+00 -7.57147193e-01 -6.43465519e-01
6.88601136e-01 6.65375292e-01 9.60579157e-01 -1.59038395e-01
-3.38497758e-01 1.54551402e-01 -3.65442783e-01 -2.63534755e-01
6.00889266e-01 2.43480057e-01 -1.69751465e-01 4.31477964e-01
1.38550863e-01 -5.54779470e-01 -8.30464363e-01 4.79575634e-01
-5.24146259e-01 2.43773192e-01 6.75004423e-01 4.59561765e-01
7.72367179e-01 4.36468631e-01 2.23920807e-01 -5.23445308e-01
-1.58358172e-01 -9.03851151e-01 -7.76483059e-01 4.30834293e-01
-3.21856141e-01 -3.24283272e-01 8.04230273e-01 -3.51417251e-02
-1.01474249e+00 5.04536569e-01 1.24736108e-01 -2.09618747e-01
-3.60854000e-01 5.44054270e-01 -2.66985357e-01 -5.17934620e-01
4.11040157e-01 5.37807822e-01 -5.94429851e-01 -5.04648209e-01
3.70736450e-01 7.82070160e-01 8.20540249e-01 -4.32798266e-01
9.39370930e-01 8.13544810e-01 -8.70187134e-02 -8.34226429e-01
-8.93299639e-01 -7.90427685e-01 -6.53694332e-01 -2.51227617e-01
7.26931095e-01 -1.38127589e+00 -4.99339670e-01 8.25037897e-01
-7.33260751e-01 -2.81334639e-01 4.99155790e-01 3.09214890e-01
-4.38144654e-02 -8.32119212e-02 -2.73722559e-02 -5.66412389e-01
-3.32064390e-01 -1.04897404e+00 1.44816160e+00 7.40961790e-01
4.76282239e-01 -8.72857571e-01 -2.76375115e-01 4.35023487e-01
6.51263416e-01 7.09053814e-01 3.95129830e-01 -1.08197689e-01
-8.43895555e-01 -2.79907227e-01 -7.40799785e-01 -1.05924442e-01
2.35028476e-01 -1.99009433e-01 -1.05306065e+00 -7.19757378e-02
-3.62530261e-01 -2.45438874e-01 9.26070869e-01 2.62415230e-01
1.11185908e+00 -1.55520989e-02 -4.26837474e-01 1.11371398e+00
2.01887798e+00 5.15932143e-02 4.42390382e-01 6.59147382e-01
1.08189631e+00 4.41555887e-01 8.15603733e-01 3.51185173e-01
9.01824653e-01 7.32607722e-01 7.94418454e-01 -1.94753721e-01
-1.72715381e-01 -1.18912056e-01 9.05577615e-02 4.52027291e-01
-2.67912894e-02 -4.27551836e-01 -1.13975322e+00 8.46814334e-01
-1.81473541e+00 -6.04385316e-01 -3.28137875e-01 1.97660196e+00
2.15995863e-01 -1.40879750e-01 -1.62349567e-01 -8.06038901e-02
6.63079381e-01 3.34423751e-01 -6.33780539e-01 2.18737617e-01
-5.46790004e-01 -9.14293006e-02 1.25940788e+00 1.63150758e-01
-1.62775326e+00 1.14478767e+00 4.82962275e+00 9.41293716e-01
-1.81034839e+00 2.25729287e-01 8.70991528e-01 -4.69400622e-02
9.83800590e-02 -2.04218358e-01 -7.62366235e-01 8.13082933e-01
6.63040996e-01 9.06841084e-02 4.76274967e-01 8.28517497e-01
1.24254614e-01 -4.79647070e-01 -2.58207887e-01 1.22975171e+00
1.72546610e-01 -1.24860632e+00 -3.59932095e-01 -6.32348843e-03
9.28474486e-01 6.79308236e-01 3.61891657e-01 1.88568071e-01
1.51068389e-01 -1.13894379e+00 8.89155924e-01 3.95954669e-01
8.53133440e-01 -7.65746891e-01 7.58984089e-01 2.70836979e-01
-1.64901614e+00 -1.01010986e-01 -3.47857177e-01 1.37525126e-01
4.37500440e-02 5.40973186e-01 -6.47268593e-01 9.97478485e-01
1.12702298e+00 8.63917470e-01 -1.03478444e+00 1.20762920e+00
-2.55420208e-01 4.88651723e-01 -7.02348650e-01 1.95316881e-01
7.39649773e-01 -6.45163804e-02 -4.94409092e-02 1.09972715e+00
7.62471199e-01 9.73640978e-02 3.72274250e-01 5.14226735e-01
1.03676155e-01 1.58858165e-01 -5.79402983e-01 4.94179875e-01
4.09026086e-01 1.44172597e+00 -1.02807581e+00 -5.10137044e-02
-3.99622172e-01 1.11749327e+00 3.00719827e-01 4.94028926e-01
-1.09833598e+00 -4.33355540e-01 5.10621488e-01 -4.25445288e-02
4.88036126e-01 -1.91637129e-01 -1.69367343e-01 -1.15734291e+00
1.96753159e-01 -3.40979546e-01 2.96674371e-01 -1.00163257e+00
-8.35400343e-01 6.74013376e-01 -2.40189582e-01 -1.16065550e+00
2.06264570e-01 -6.52083397e-01 -6.62989140e-01 9.97002661e-01
-1.97242951e+00 -1.71282840e+00 -8.42434883e-01 7.02635288e-01
2.27508962e-01 2.70591658e-02 6.35587096e-01 5.43417394e-01
-8.18483710e-01 4.60392654e-01 4.64186579e-01 6.56487048e-02
5.16189933e-01 -1.22766232e+00 4.92241621e-01 1.00390291e+00
3.51822339e-02 3.13061953e-01 4.12802041e-01 -4.53156710e-01
-1.52959251e+00 -1.67937303e+00 7.34862804e-01 -1.52524084e-01
6.28924906e-01 -2.75264263e-01 -5.36711872e-01 5.41703939e-01
-1.93518341e-01 4.90344703e-01 2.02591643e-01 -2.81626023e-02
-3.74497592e-01 -5.08488417e-01 -1.31568730e+00 3.96320820e-01
8.72042716e-01 -4.72876936e-01 4.56067659e-02 3.83074284e-01
4.95674700e-01 -6.18198812e-01 -7.37024963e-01 3.11191052e-01
3.68952543e-01 -1.00036180e+00 1.12740672e+00 1.42456159e-01
1.22402370e-01 -7.08641231e-01 -5.01242518e-01 -1.22722435e+00
-2.23172456e-01 -2.24644653e-02 5.18642426e-01 1.26099980e+00
2.68176258e-01 -7.86862016e-01 8.64739656e-01 3.85133147e-01
-1.24855638e-01 -5.76712966e-01 -9.63174641e-01 -6.91980064e-01
-2.35304430e-01 -4.55601811e-01 9.34604228e-01 9.13230836e-01
-5.92689335e-01 -1.79563820e-01 -4.59156692e-01 8.20393562e-01
7.21298039e-01 4.17928308e-01 7.58322239e-01 -1.02622020e+00
2.46432379e-01 -2.48113513e-01 -6.69767857e-01 -6.91872954e-01
-4.50953655e-02 -6.65875614e-01 1.38159012e-02 -1.57295668e+00
-6.79499954e-02 -8.68737400e-01 -5.69736540e-01 7.28548646e-01
-3.28830071e-02 8.09839070e-01 7.63779357e-02 1.86386660e-01
-8.88399899e-01 4.86157060e-01 7.22805500e-01 -2.75419027e-01
2.99218167e-02 -1.43635809e-01 -6.45490766e-01 5.82412124e-01
1.03072107e+00 -4.53913152e-01 -4.48323190e-01 -8.19979012e-01
1.43729389e-01 -2.51209468e-01 6.66529000e-01 -1.46549511e+00
5.15957654e-01 -2.37323061e-01 4.50226754e-01 -8.48434389e-01
3.70077431e-01 -7.35122442e-01 3.46415371e-01 2.12679163e-01
4.01836306e-01 8.49443302e-02 2.96276182e-01 6.20553553e-01
-4.42273498e-01 1.40228465e-01 6.24210536e-01 -2.79074788e-01
-1.53467274e+00 4.98192251e-01 1.92240968e-01 4.31838930e-02
1.05298400e+00 -3.32006454e-01 -3.86095434e-01 5.74474968e-02
-2.93219835e-01 3.43686700e-01 6.72415853e-01 5.21907270e-01
3.87624949e-01 -1.22214425e+00 -4.68199074e-01 1.83337480e-01
5.40262103e-01 2.32670397e-01 6.34884298e-01 9.01840746e-01
-7.49910474e-01 6.45553172e-01 -1.90792739e-01 -7.76755691e-01
-1.08190072e+00 1.14181675e-01 3.55272859e-01 -4.90089022e-02
-4.86997545e-01 9.03473616e-01 2.46428892e-01 -6.23538494e-01
3.33408564e-02 -3.13298345e-01 -1.84484422e-01 -3.27563174e-02
3.54261905e-01 4.71374243e-02 2.28947237e-01 -1.22745252e+00
-6.42246485e-01 1.14305329e+00 2.63127953e-01 1.75306246e-01
1.37315059e+00 -3.78625154e-01 -7.32931048e-02 2.74943233e-01
1.24204803e+00 -1.66837037e-01 -1.19095302e+00 -4.65740234e-01
-9.93792638e-02 -6.28022611e-01 8.32063034e-02 -1.03861475e+00
-1.46634388e+00 6.53679073e-01 7.83482492e-01 -2.75116283e-02
1.32922757e+00 -4.30158302e-02 8.02848876e-01 1.35950208e-01
7.96341598e-01 -9.81902361e-01 -2.55525947e-01 4.71272677e-01
5.86549759e-01 -1.50615728e+00 -5.80564886e-02 -3.17535371e-01
-5.04462183e-01 8.45157266e-01 5.69321454e-01 -2.42542513e-02
7.18426764e-01 -2.09277645e-01 1.63647547e-01 -1.90643683e-01
-1.39541104e-01 -3.16199690e-01 2.94591725e-01 5.74699521e-01
4.34941910e-02 4.23918009e-01 3.34144533e-01 2.92644262e-01
-2.89696008e-01 -7.76578039e-02 1.64710373e-01 6.86075449e-01
-3.15483868e-01 -6.08122528e-01 -6.26896322e-01 1.65489540e-01
-4.14333224e-01 -1.05282985e-01 -3.70727852e-02 8.52664888e-01
5.87222278e-01 1.03714406e+00 1.02785692e-01 -4.71112609e-01
2.19658777e-01 -4.26754713e-01 6.45542890e-02 -2.45378941e-01
-3.74614567e-01 -1.68067381e-01 -1.18085481e-01 -7.75281608e-01
-4.10952628e-01 -7.40739882e-01 -1.23122478e+00 -3.69329810e-01
-1.53064635e-02 -5.66359945e-02 1.09311843e+00 8.41138482e-01
6.91730261e-01 3.26245189e-01 6.88838601e-01 -9.95802462e-01
1.38183638e-01 -7.06453860e-01 -6.84147298e-01 2.18929037e-01
2.75112540e-01 -7.31159151e-01 -1.57567337e-01 -1.72726199e-01] | [7.682118892669678, -1.8271548748016357] |
836c6faf-3a59-4747-bd66-8816301c9d18 | text-is-all-you-need-personalizing-asr-models | 2303.14885 | null | https://arxiv.org/abs/2303.14885v1 | https://arxiv.org/pdf/2303.14885v1.pdf | Text is All You Need: Personalizing ASR Models using Controllable Speech Synthesis | Adapting generic speech recognition models to specific individuals is a challenging problem due to the scarcity of personalized data. Recent works have proposed boosting the amount of training data using personalized text-to-speech synthesis. Here, we ask two fundamental questions about this strategy: when is synthetic data effective for personalization, and why is it effective in those cases? To address the first question, we adapt a state-of-the-art automatic speech recognition (ASR) model to target speakers from four benchmark datasets representative of different speaker types. We show that ASR personalization with synthetic data is effective in all cases, but particularly when (i) the target speaker is underrepresented in the global data, and (ii) the capacity of the global model is limited. To address the second question of why personalized synthetic data is effective, we use controllable speech synthesis to generate speech with varied styles and content. Surprisingly, we find that the text content of the synthetic data, rather than style, is important for speaker adaptation. These results lead us to propose a data selection strategy for ASR personalization based on speech content. | ['Oncel Tuzel', 'Hema Swetha Koppula', 'Jen-Hao Rick Chang', 'Ting-yao Hu', 'Karren Yang'] | 2023-03-27 | null | null | null | null | ['text-to-speech-synthesis', 'speech-synthesis'] | ['speech', 'speech'] | [ 4.51177329e-01 6.30042180e-02 -1.15851812e-01 -3.66092980e-01
-7.51506805e-01 -5.73190928e-01 4.66566086e-01 -2.83944964e-01
-2.24921823e-01 6.60475791e-01 6.68596864e-01 -1.39424384e-01
8.56469572e-02 -3.37383866e-01 -5.34512460e-01 -6.64664626e-01
4.20493007e-01 6.20960176e-01 1.94844350e-01 -6.55486584e-01
9.88219529e-02 6.33786619e-01 -1.82620800e+00 1.87918991e-01
1.10925138e+00 5.14522731e-01 3.14398319e-01 7.74035037e-01
-2.19043136e-01 4.95497771e-02 -1.12178123e+00 -2.54417211e-01
1.70774981e-01 -7.26400614e-01 -6.27994597e-01 3.23956579e-01
5.23883224e-01 1.79016346e-03 -2.79508889e-01 9.32156801e-01
9.11500394e-01 3.26999962e-01 4.72143501e-01 -7.22961783e-01
-7.50355840e-01 9.41075325e-01 1.29001230e-01 3.92852336e-01
2.15420708e-01 1.68506369e-01 7.57276416e-01 -6.35711670e-01
3.71525586e-01 1.43500042e+00 2.63383776e-01 1.24425960e+00
-1.37674785e+00 -3.86227727e-01 4.75981027e-01 -8.13821107e-02
-1.36374795e+00 -1.18341339e+00 7.88618565e-01 -2.41238639e-01
6.27724886e-01 7.55838990e-01 4.16439265e-01 1.43813086e+00
-4.08761382e-01 5.80801845e-01 8.79656374e-01 -5.16643405e-01
4.38857704e-01 5.86796105e-01 4.89014648e-02 3.48065458e-02
-7.49788061e-02 3.35646123e-02 -9.40090120e-01 -1.22675933e-01
4.29343522e-01 -5.34320772e-01 -4.07575399e-01 -5.29939756e-02
-1.06130266e+00 6.53839886e-01 -2.91130692e-01 5.43470144e-01
-1.83043003e-01 -2.41137788e-01 2.62761921e-01 6.49594903e-01
6.24937177e-01 7.96930015e-01 -8.01065087e-01 -3.54031891e-01
-8.01379621e-01 3.06020021e-01 8.66770387e-01 1.05398715e+00
4.72174406e-01 5.26235461e-01 -4.39486712e-01 1.52026916e+00
-8.34348798e-02 7.06238747e-01 8.81119609e-01 -8.33854914e-01
4.58359033e-01 2.17526019e-01 2.27568168e-02 -3.75184894e-01
-5.71436733e-02 -6.22060478e-01 -4.65902597e-01 -2.83559471e-01
4.17964786e-01 -3.53114367e-01 -9.15102839e-01 2.11474872e+00
2.06917465e-01 -2.18455732e-01 1.33125260e-01 6.98282957e-01
6.29387081e-01 4.99393910e-01 -3.96937914e-02 -3.88733834e-01
1.04671812e+00 -7.25895643e-01 -6.84080303e-01 -4.71336871e-01
4.04983103e-01 -8.34131122e-01 1.40622509e+00 5.03974035e-02
-9.51424956e-01 -5.20840645e-01 -9.24867392e-01 4.10767198e-01
-3.19349408e-01 1.17502205e-01 1.77935645e-01 1.29955888e+00
-1.19028902e+00 4.79661047e-01 -4.21090722e-01 -7.84578919e-01
-5.41021787e-02 3.68689716e-01 -6.06045388e-02 1.75644740e-01
-1.34256959e+00 7.45791316e-01 -9.45343543e-03 -4.38702404e-01
-6.16364956e-01 -6.59613371e-01 -4.63000268e-01 1.72796190e-01
2.22125441e-01 -7.17213511e-01 1.23547757e+00 -1.38497174e+00
-2.24304152e+00 5.50763011e-01 -3.54489446e-01 -1.17421038e-01
5.23099124e-01 9.20502841e-02 -5.80625474e-01 -2.54964262e-01
-2.74361521e-01 3.79243851e-01 1.16829002e+00 -1.32313430e+00
-4.88176376e-01 -5.73405385e-01 -2.31228545e-01 4.25036073e-01
-7.81670630e-01 1.69998989e-01 -4.52503234e-01 -9.80672359e-01
-6.44456744e-02 -1.19362402e+00 -7.57484511e-02 -6.17660046e-01
-4.14894998e-01 -2.62309283e-01 6.03097916e-01 -7.03773797e-01
1.26824152e+00 -2.37423086e+00 3.23589116e-01 2.05639005e-01
-3.36372286e-01 4.01867062e-01 -3.03390652e-01 2.90891498e-01
-9.65474248e-02 3.10035944e-01 7.15574175e-02 -5.43391228e-01
3.51473242e-02 1.46232873e-01 -3.98416698e-01 1.92742255e-02
-1.63787365e-01 3.68282527e-01 -5.12217164e-01 -2.60036856e-01
-2.52596568e-02 4.19176668e-01 -6.82955027e-01 3.02355230e-01
-2.33408645e-01 7.54012048e-01 -4.45896119e-01 4.58086401e-01
3.14896882e-01 1.87448636e-01 -3.62447556e-03 6.30030185e-02
-2.85206661e-02 4.68478888e-01 -1.11584806e+00 1.37693381e+00
-4.77493703e-01 5.26192725e-01 2.25003347e-01 -7.22997069e-01
1.00688803e+00 4.35811609e-01 3.16589147e-01 -5.52261889e-01
-6.49872944e-02 2.80121922e-01 6.91917241e-02 -3.82973015e-01
5.37563622e-01 -2.33686164e-01 -6.02478832e-02 2.81589925e-01
3.57475393e-02 -2.70840019e-01 7.28484020e-02 -7.69515857e-02
9.23286855e-01 -3.50525528e-01 1.00410566e-01 -3.72816473e-01
3.66510242e-01 -4.12658513e-01 6.88056409e-01 1.03431690e+00
-2.82633066e-01 7.16226757e-01 2.05559701e-01 3.91667187e-02
-1.08088744e+00 -1.05101562e+00 2.73985276e-03 1.60863447e+00
-1.64324120e-01 -2.76880026e-01 -1.05548823e+00 -5.04838765e-01
-2.15425849e-01 1.00727344e+00 -4.07164007e-01 -2.75347680e-01
-6.58261538e-01 -5.46860576e-01 4.93706971e-01 3.47746104e-01
1.86393052e-01 -8.49978328e-01 -6.85709938e-02 1.05718963e-01
-2.26202920e-01 -1.04626036e+00 -1.08624017e+00 2.69147940e-02
-7.51942456e-01 -3.07429761e-01 -9.94267225e-01 -5.43476582e-01
5.28856158e-01 1.91161662e-01 9.65824544e-01 -1.51708648e-01
2.27759287e-01 6.36688352e-01 -5.53031325e-01 -3.48287940e-01
-1.06807125e+00 6.10941947e-01 3.68762463e-01 1.94090322e-01
5.37519567e-02 -5.96271455e-01 -3.07435453e-01 6.41077340e-01
-7.38398612e-01 -1.64505318e-01 4.04299736e-01 6.06339753e-01
1.15382016e-01 -5.16984016e-02 1.04338062e+00 -8.41251433e-01
8.43801022e-01 -5.85104153e-02 -2.96979219e-01 4.36016887e-01
-4.28766280e-01 1.58779800e-01 7.31530666e-01 -7.79818118e-01
-1.24529910e+00 1.03281274e-01 -3.68617296e-01 -2.83132285e-01
-2.18039677e-01 1.40499383e-01 -6.19672418e-01 1.97290927e-01
8.80177319e-01 3.21144849e-01 1.59500763e-02 -6.58099115e-01
3.72112423e-01 1.08984399e+00 1.57823399e-01 -7.21706808e-01
5.34830570e-01 9.56384558e-03 -4.65393782e-01 -1.39051604e+00
-4.17985797e-01 -2.05781639e-01 -3.79679322e-01 -3.46295536e-02
6.56323969e-01 -6.36222959e-01 -2.37155482e-01 4.91876245e-01
-9.93572235e-01 -6.08756959e-01 -3.95852447e-01 2.57060945e-01
-7.05062389e-01 2.07706898e-01 -2.58863240e-01 -9.67471004e-01
-2.96366781e-01 -1.45383632e+00 9.65843320e-01 1.36949971e-01
-4.89170104e-01 -7.52511322e-01 -5.33407032e-02 5.64231277e-01
7.75496066e-01 -5.63961089e-01 9.11308587e-01 -1.06380773e+00
-3.43475074e-01 -1.06152743e-01 4.01815206e-01 5.50354838e-01
5.10972500e-01 -1.30533978e-01 -1.10139263e+00 -3.52771640e-01
-9.58007127e-02 1.02069356e-01 5.27190566e-01 2.69725531e-01
1.17614734e+00 -5.18268824e-01 -1.75448060e-01 3.90399337e-01
5.12690961e-01 3.41453075e-01 4.03038263e-01 2.57544452e-03
5.88536501e-01 1.01838994e+00 2.07052022e-01 4.14732873e-01
2.49159068e-01 1.25993693e+00 -2.86262602e-01 3.83645773e-01
-4.84472841e-01 -2.49898896e-01 7.17383087e-01 9.05305207e-01
2.20547542e-01 -4.05468553e-01 -6.41425967e-01 5.91998756e-01
-1.55447757e+00 -7.88426280e-01 5.80029845e-01 2.62216830e+00
1.07171071e+00 -8.60939696e-02 4.35978204e-01 -8.47098380e-02
9.81179953e-01 1.17245473e-01 -6.70762539e-01 -3.89721662e-01
-3.63824278e-01 -9.67769697e-02 4.65955734e-01 6.06786907e-01
-7.09364772e-01 1.03332603e+00 6.58104467e+00 1.02238750e+00
-1.34074390e+00 2.02214420e-01 6.65353537e-01 -3.20195466e-01
-5.79092145e-01 -2.32918546e-01 -1.11018860e+00 7.35147119e-01
1.18020463e+00 -3.74845594e-01 7.83968747e-01 7.33002782e-01
2.59304315e-01 4.18184221e-01 -1.20643497e+00 1.00450826e+00
3.93359125e-01 -9.44075465e-01 3.08167934e-01 -5.21755032e-02
6.15949392e-01 -1.22878842e-01 3.03168863e-01 3.07150602e-01
1.61324501e-01 -8.54323447e-01 7.58184731e-01 3.92911702e-01
7.37570584e-01 -6.06875598e-01 2.64705598e-01 3.23645294e-01
-8.17895591e-01 -1.52918264e-01 -6.89617405e-03 5.58419526e-01
3.45664914e-03 4.30303127e-01 -9.34209526e-01 1.82610843e-02
6.54795825e-01 -4.61256392e-02 -7.08680987e-01 6.85248435e-01
4.38511856e-02 7.09617913e-01 -4.58461791e-01 -1.00405820e-01
-3.91530186e-01 1.75614327e-01 1.04219854e+00 1.10641181e+00
6.18671238e-01 -1.36717334e-01 -1.24333881e-01 6.48775160e-01
-1.17559582e-02 1.57779649e-01 -4.86042738e-01 -2.92313159e-01
8.32286000e-01 7.14597225e-01 -2.50802964e-01 -2.49365181e-01
-5.86607903e-02 1.00999928e+00 1.68046176e-01 5.42947054e-01
-1.85579583e-01 -1.93303645e-01 8.67097616e-01 2.53350884e-01
1.68009952e-01 -1.26048625e-01 -4.59186733e-01 -1.28053916e+00
-9.68100578e-02 -1.39513254e+00 3.18891108e-01 -4.35648501e-01
-1.15963733e+00 7.06615090e-01 -1.51428124e-02 -7.60065615e-01
-4.38017726e-01 -3.03875774e-01 -4.50372249e-01 9.40504253e-01
-8.33790064e-01 -9.78086174e-01 -5.81320673e-02 4.97655898e-01
1.00587559e+00 -7.13671923e-01 7.22244203e-01 4.17354286e-01
-6.52347863e-01 1.25235295e+00 9.80067179e-02 -3.34650367e-01
9.21671093e-01 -1.09313023e+00 6.39502168e-01 7.97657371e-01
-4.44441661e-02 7.68041372e-01 1.24803078e+00 -5.85906863e-01
-1.31001985e+00 -9.37521636e-01 9.21415031e-01 -7.27558970e-01
2.33200073e-01 -5.82633913e-01 -9.55989480e-01 4.74105626e-01
4.00304310e-02 -4.54902649e-01 7.80697942e-01 3.46591443e-01
-4.73018698e-02 -3.56433064e-01 -1.11876667e+00 1.05527055e+00
1.09264886e+00 -4.53128636e-01 -2.82005399e-01 1.84099913e-01
1.06056750e+00 -3.45475554e-01 -7.25649297e-01 3.40982080e-01
3.50123465e-01 -6.68485045e-01 8.08216870e-01 -5.06051660e-01
-1.61626965e-01 3.81124858e-03 -6.34592772e-01 -1.98035848e+00
-2.17885613e-01 -9.03958917e-01 1.58842221e-01 1.61250472e+00
7.26205468e-01 -8.26378524e-01 7.79576600e-01 8.13493252e-01
-2.94895798e-01 -3.55702430e-01 -1.15537298e+00 -9.64265108e-01
1.87139437e-01 -2.13090569e-01 1.03973925e+00 8.06543469e-01
-5.24656922e-02 4.71276104e-01 -5.16322851e-01 1.43256247e-01
3.47128659e-01 -2.69534171e-01 9.33691442e-01 -9.32619512e-01
-5.05582213e-01 -5.71433663e-01 1.31066493e-03 -1.18709791e+00
3.31282973e-01 -5.20551622e-01 1.37798116e-01 -1.04420435e+00
-1.64363086e-01 -5.05257189e-01 5.07643744e-02 1.56727850e-01
-3.37308794e-01 -3.34427088e-01 3.65739286e-01 1.26753166e-01
-1.73604891e-01 7.77709067e-01 1.08459949e+00 -8.73214453e-02
-7.17914820e-01 5.53260267e-01 -9.54677820e-01 3.48721743e-01
8.41861069e-01 -1.01383246e-01 -5.00202477e-01 -3.07154506e-01
1.09922083e-03 1.76470324e-01 -1.92890748e-01 -7.79522836e-01
1.52659774e-01 -3.84712189e-01 -9.30247083e-03 -1.37319237e-01
5.79530060e-01 -5.03208697e-01 5.80178248e-03 1.56031311e-01
-6.23413503e-01 -1.55855492e-01 1.63817987e-01 3.53415519e-01
1.41634524e-01 -2.50909805e-01 1.12053466e+00 8.04252997e-02
-4.16031033e-01 2.46278092e-01 -5.58296442e-01 1.22250922e-01
4.25342113e-01 -3.79640572e-02 -2.82967597e-01 -7.73113370e-01
-7.65435874e-01 -1.37661859e-01 4.95587140e-01 8.31297159e-01
2.47161552e-01 -1.20368302e+00 -8.56929779e-01 4.32096392e-01
3.63678128e-01 -4.04824227e-01 3.90033752e-01 4.39913362e-01
1.11837223e-01 3.62897664e-01 3.03189456e-01 -4.32362974e-01
-1.48398280e+00 3.85490179e-01 5.63492835e-01 2.84233481e-01
-1.45911410e-01 8.06833446e-01 2.84081846e-01 -5.39769650e-01
2.49582335e-01 -7.63218198e-03 -1.25174388e-01 -1.29653362e-03
4.41100925e-01 4.14774388e-01 2.89082766e-01 -7.45245397e-01
-2.94723183e-01 2.75830120e-01 -2.12772593e-01 -3.90721917e-01
1.07132733e+00 -5.20257175e-01 2.45484531e-01 5.19654393e-01
8.28021348e-01 3.03774148e-01 -1.09139752e+00 -3.72551918e-01
-2.74224460e-01 -5.27749777e-01 -1.11290082e-01 -7.43648589e-01
-9.14681315e-01 5.13936102e-01 6.44753754e-01 4.14588302e-01
9.16813672e-01 9.61607397e-02 5.82658172e-01 2.90545672e-01
2.76696473e-01 -1.46452820e+00 1.09175555e-01 6.80338681e-01
1.15532875e+00 -1.05963337e+00 -3.71588200e-01 -2.73092955e-01
-1.00737596e+00 7.05736518e-01 6.53138995e-01 4.51099485e-01
4.57820177e-01 -1.38440177e-01 1.87299192e-01 3.65309417e-01
-8.11001480e-01 -8.52921009e-02 3.02587748e-01 7.90654600e-01
4.42965597e-01 2.46498898e-01 -2.85953600e-02 6.11316741e-01
-7.85671890e-01 -4.58930820e-01 3.57755274e-01 5.28332949e-01
-5.58430970e-01 -1.32762158e+00 -5.38374066e-01 3.88111979e-01
-3.31929028e-01 -1.70748550e-02 -7.81352162e-01 2.24615276e-01
-1.78519696e-01 1.28743851e+00 -1.35084122e-01 -3.05191576e-01
5.66558480e-01 2.68766224e-01 4.26813155e-01 -9.12664592e-01
-4.95593488e-01 1.41971335e-01 3.76731485e-01 -5.82625158e-02
1.95494428e-01 -8.76772046e-01 -5.48715472e-01 -2.36494392e-01
-1.76686585e-01 1.78507879e-01 8.96714509e-01 9.95263696e-01
5.64527094e-01 5.96643686e-01 7.73196995e-01 -8.28029096e-01
-9.32826102e-01 -1.06074560e+00 -6.59704626e-01 3.34391564e-01
4.83687490e-01 -5.05488813e-01 -6.42004371e-01 -1.41337765e-02] | [14.588102340698242, 6.6042561531066895] |
827436e5-0e33-4a68-992b-3f0739df8660 | segmentation-and-tracking-of-vegetable-plants | 2306.13518 | null | https://arxiv.org/abs/2306.13518v2 | https://arxiv.org/pdf/2306.13518v2.pdf | Segmentation and Tracking of Vegetable Plants by Exploiting Vegetable Shape Feature for Precision Spray of Agricultural Robots | With the increasing deployment of agricultural robots, the traditional manual spray of liquid fertilizer and pesticide is gradually being replaced by agricultural robots. For robotic precision spray application in vegetable farms, accurate plant phenotyping through instance segmentation and robust plant tracking are of great importance and a prerequisite for the following spray action. Regarding the robust tracking of vegetable plants, to solve the challenging problem of associating vegetables with similar color and texture in consecutive images, in this paper, a novel method of Multiple Object Tracking and Segmentation (MOTS) is proposed for instance segmentation and tracking of multiple vegetable plants. In our approach, contour and blob features are extracted to describe unique feature of each individual vegetable, and associate the same vegetables in different images. By assigning a unique ID for each vegetable, it ensures the robot to spray each vegetable exactly once, while traversing along the farm rows. Comprehensive experiments including ablation studies are conducted, which prove its superior performance over two State-Of-The-Art (SOTA) MOTS methods. Compared to the conventional MOTS methods, the proposed method is able to re-identify objects which have gone out of the camera field of view and re-appear again using the proposed data association strategy, which is important to ensure each vegetable be sprayed only once when the robot travels back and forth. Although the method is tested on lettuce farm, it can be applied to other similar vegetables such as broccoli and canola. Both code and the dataset of this paper is publicly released for the benefit of the community: https://github.com/NanH5837/LettuceMOTS. | ['Yu Tan', 'Yongliang Qiao', 'Zimeng Wang', 'Huiyu Zhong', 'Xuechang Wang', 'Shuo Wang', 'Daobilige Su', 'Nan Hu'] | 2023-06-23 | null | null | null | null | ['object-tracking', 'multiple-object-tracking', 'plant-phenotyping', 'instance-segmentation'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 1.72016889e-01 -1.21283740e-01 -2.62866825e-01 8.21604878e-02
3.29945922e-01 -1.09887302e+00 -1.23214699e-01 8.92123699e-01
-1.58521667e-01 3.63027096e-01 -1.09878302e+00 -5.24316847e-01
-3.34361762e-01 -9.71420109e-01 -6.96712554e-01 -9.31831360e-01
-1.11644179e-01 4.54201102e-01 6.14607394e-01 -1.56653240e-01
-3.17132771e-02 7.31385469e-01 -1.56413233e+00 -1.45143479e-01
8.92895758e-01 5.11571169e-01 1.13334858e+00 5.85610092e-01
-1.65620516e-03 1.13586336e-01 -3.84447992e-01 1.50183141e-01
2.38449797e-01 -2.10072070e-01 -5.62300920e-01 2.45770082e-01
1.03370892e-02 -5.01265764e-01 4.78390247e-01 1.29337227e+00
9.46708322e-02 -2.10998669e-01 5.74726045e-01 -1.40307856e+00
-4.96577948e-01 6.64081693e-01 -1.16704392e+00 -2.48213798e-01
6.46530092e-02 -6.88784868e-02 4.96952713e-01 -2.83129901e-01
5.57565153e-01 9.91354764e-01 6.12503052e-01 2.47306451e-01
-1.20205474e+00 -7.62675047e-01 3.88846457e-01 -9.56555363e-03
-1.30310392e+00 2.99395751e-02 2.25785136e-01 -3.56870204e-01
1.13841288e-01 2.34703571e-01 7.65287340e-01 3.02926958e-01
3.28580856e-01 7.05339313e-01 4.37105000e-01 -3.67398590e-01
2.23725766e-01 1.37188837e-01 1.49975782e-02 6.73375607e-01
9.33784842e-01 -5.61532266e-02 2.17282668e-01 6.28737807e-02
8.11954856e-01 2.12334350e-01 -2.94372410e-01 -9.60045815e-01
-1.28487539e+00 6.69921875e-01 5.80832601e-01 5.01091301e-01
-7.07901180e-01 1.47886455e-01 2.88907915e-01 -1.73828304e-01
6.45904914e-02 1.83601260e-01 -5.73828101e-01 4.72302049e-01
-6.17391348e-01 3.08729172e-01 8.36891055e-01 1.38287103e+00
8.74488473e-01 -3.45893353e-01 -5.77195622e-02 2.96192795e-01
3.98700356e-01 8.98441195e-01 -9.97108817e-02 -7.89962113e-01
-6.80764392e-02 7.90586710e-01 5.75315893e-01 -1.36265719e+00
-4.31547970e-01 1.47523209e-01 -8.28548789e-01 4.58835512e-01
4.63157088e-01 -2.43222401e-01 -8.82299960e-01 1.31381691e+00
7.69863009e-01 5.60771264e-02 1.79667190e-01 8.16268086e-01
6.94548428e-01 6.83059990e-01 3.12198490e-01 -2.82739490e-01
1.86127329e+00 -4.80570674e-01 -9.23076272e-01 -1.49951652e-01
7.15648770e-01 -8.42048466e-01 2.82451242e-01 2.73627430e-01
-6.04129195e-01 -6.48928583e-01 -1.15429473e+00 5.24306536e-01
-6.92084312e-01 8.96348536e-01 7.01215327e-01 4.98738676e-01
-8.13793361e-01 5.61186552e-01 -1.18558061e+00 -1.01645517e+00
2.24999219e-01 5.78654110e-01 -2.77753472e-01 -1.44071817e-01
-5.41048110e-01 7.06152260e-01 6.39238596e-01 5.16002357e-01
-7.63984501e-01 -4.80586112e-01 -6.65420592e-01 1.87709238e-02
5.64137459e-01 -6.54006824e-02 1.12750149e+00 -6.07305944e-01
-1.44618285e+00 6.89106941e-01 -1.32433876e-01 -1.33495510e-01
1.06628112e-01 -3.54093075e-01 -1.25567615e-01 1.85659736e-01
3.28876227e-01 7.95823812e-01 6.90763772e-01 -1.52221012e+00
-1.19799721e+00 -3.66781473e-01 1.23699024e-01 3.79660130e-02
-1.98021382e-01 9.24386643e-03 -2.97592252e-01 -1.80470318e-01
4.20960039e-01 -1.02147603e+00 -3.34028512e-01 3.40593398e-01
-2.33521208e-01 -9.40935761e-02 1.35501623e+00 -3.86755675e-01
6.89265549e-01 -2.04971695e+00 1.85390592e-01 -1.30643383e-01
-7.35674351e-02 5.80965102e-01 -1.48918986e-01 2.42752582e-01
2.95894653e-01 -1.36461556e-01 -2.41154790e-01 2.69658417e-01
-4.31005150e-01 3.05815756e-01 8.67617428e-02 7.21321464e-01
3.66621614e-01 5.42381406e-01 -1.27775383e+00 -5.13817966e-01
5.26780844e-01 2.74151504e-01 6.42536022e-03 4.87412587e-02
-2.57310808e-01 5.61802864e-01 -8.25330675e-01 9.62018847e-01
1.43818843e+00 2.91895363e-02 5.54184675e-01 -4.04207826e-01
-6.01018786e-01 -7.31932104e-01 -1.37919593e+00 1.33120656e+00
-4.42245454e-02 7.12805018e-02 7.41699219e-01 -1.02109766e+00
1.00259721e+00 5.68514407e-01 6.80030525e-01 -1.15936093e-01
1.45494804e-01 3.57390255e-01 -1.62926629e-01 -5.24018347e-01
5.85495949e-01 6.10628664e-01 1.23684593e-01 -3.24657769e-03
-2.11872563e-01 -2.99136013e-01 6.95040941e-01 -1.05066657e-01
8.26958418e-01 6.42857790e-01 3.40335280e-01 -6.49617612e-01
5.12758613e-01 7.07213998e-01 6.50112092e-01 5.79315364e-01
-3.10868084e-01 -2.30354741e-02 9.21545178e-02 -9.85446200e-02
-7.74769604e-01 -7.21593142e-01 -1.70656726e-01 9.08143938e-01
1.04742491e+00 1.58387929e-01 -6.51985824e-01 -5.68071902e-01
3.53459626e-01 4.45004612e-01 -3.76915514e-01 1.68053016e-01
-5.44987261e-01 -7.17482567e-01 2.29732424e-01 3.47961336e-01
5.80967665e-01 -1.21764052e+00 -1.56992877e+00 5.98377705e-01
-1.53627574e-01 -1.09433413e+00 2.21868917e-01 6.68709755e-01
-8.87465894e-01 -1.22245371e+00 -9.05346394e-01 -9.61881101e-01
7.70034254e-01 9.92236972e-01 6.06271267e-01 3.41154426e-01
-4.50393111e-01 5.28902896e-02 -7.98872411e-01 -6.45955861e-01
-6.29959583e-01 2.71659851e-01 -1.20424516e-01 -3.66906494e-01
3.97799969e-01 -1.02200098e-01 -5.95181108e-01 5.42330086e-01
-9.47320700e-01 -5.62880561e-02 5.59623063e-01 4.95978206e-01
5.45483828e-01 2.81141430e-01 2.98912734e-01 -6.69204712e-01
-1.28146172e-01 -5.90434909e-01 -1.24523830e+00 6.48029685e-01
-4.63015363e-02 -3.77281040e-01 5.15524983e-01 -5.24564862e-01
-7.51740515e-01 5.89567244e-01 6.07908785e-01 -1.59672290e-01
-6.50106549e-01 4.02717113e-01 -7.14752078e-02 -1.10941455e-01
3.38994235e-01 -2.24165201e-01 5.02861403e-02 -3.61549020e-01
2.35187709e-01 5.01200557e-01 5.78869939e-01 -7.43963867e-02
8.00668836e-01 4.30143774e-01 1.99205682e-01 -9.01124895e-01
-3.38127732e-01 -7.16824532e-01 -9.44899201e-01 -2.87143022e-01
1.12191606e+00 -6.97483182e-01 -1.26810658e+00 5.55595756e-01
-1.39021611e+00 -3.50833207e-01 2.27078721e-01 6.95566773e-01
-8.64392221e-02 2.77309567e-01 -3.72950435e-01 -9.41252172e-01
-2.44594499e-01 -1.19275856e+00 1.04141808e+00 7.18346000e-01
1.91306137e-02 -6.38077617e-01 -2.19949961e-01 -3.38144034e-01
5.03094196e-02 2.91394621e-01 6.70340180e-01 -2.16132611e-01
-4.82930154e-01 -2.79062420e-01 -3.66914421e-01 -2.44553909e-01
8.72157931e-01 6.22489631e-01 -5.38140655e-01 -5.51136971e-01
-2.83369154e-01 1.45509765e-01 4.18951720e-01 7.38694608e-01
6.02511525e-01 2.61420608e-01 -1.08555746e+00 4.02245790e-01
1.55952513e+00 7.89590418e-01 1.45613506e-01 4.04175639e-01
5.84457576e-01 8.61369908e-01 1.50217438e+00 3.78775656e-01
2.07761899e-01 5.57904124e-01 1.02510297e+00 -3.07202876e-01
3.52270544e-01 3.23958784e-01 2.44831905e-01 2.55606949e-01
-5.40289432e-02 -6.73719883e-01 -8.33069205e-01 7.71048784e-01
-2.06333184e+00 -7.39829719e-01 -5.21757782e-01 1.93684685e+00
3.32014561e-01 -4.63057637e-01 -2.62349814e-01 1.34337321e-01
1.08286643e+00 -3.18454474e-01 -5.76549828e-01 -2.60234326e-01
1.47063822e-01 -2.85689607e-02 1.13513207e+00 1.48702323e-01
-1.41437340e+00 1.10841513e+00 4.91831589e+00 5.67286253e-01
-1.10262346e+00 -2.47299895e-01 1.29684210e-01 7.65613794e-01
2.65456259e-01 1.03011042e-01 -9.64722037e-01 -6.76839501e-02
2.40176275e-01 1.94506928e-01 3.70081007e-01 6.24386728e-01
3.00569087e-01 -8.36882591e-01 -8.61471593e-01 3.07695121e-01
-5.49037099e-01 -7.07126200e-01 -3.60492021e-01 -1.68801278e-01
4.64875668e-01 -4.06869084e-01 -2.99133122e-01 -1.39120758e-01
6.31216347e-01 -4.95129466e-01 6.91819429e-01 4.27464060e-02
5.41821361e-01 -3.89331460e-01 7.92450905e-01 4.36761856e-01
-1.72589672e+00 -2.06127211e-01 -7.16751397e-01 3.32058281e-01
4.88465540e-02 3.65984559e-01 -7.90783942e-01 1.16828251e+00
1.10302413e+00 7.51571894e-01 -4.99875933e-01 1.05245829e+00
-9.69911814e-02 2.51920283e-01 -5.10812044e-01 -1.14698865e-01
2.63832986e-01 -4.97107238e-01 3.76857400e-01 9.00153577e-01
7.83878446e-01 1.87824696e-01 4.22094017e-01 8.84014785e-01
5.74748039e-01 1.93387657e-01 -6.30303442e-01 -1.71208888e-01
6.32951438e-01 1.74467874e+00 -1.47991383e+00 -1.45428315e-01
-1.33078739e-01 1.02024543e+00 -1.95012048e-01 2.47326180e-01
-8.77969325e-01 -5.38148761e-01 2.66255438e-01 -2.58236736e-01
8.71390462e-01 -2.10002229e-01 3.88963558e-02 -3.78168046e-01
-2.53359020e-01 -4.88044918e-01 8.22010934e-02 -7.28548110e-01
-6.11872613e-01 3.32170576e-01 2.53356785e-01 -1.25434196e+00
1.13354087e-01 -7.04049885e-01 -3.83762419e-01 7.14247525e-01
-1.38402879e+00 -1.27252626e+00 -7.66017914e-01 4.81310077e-02
3.18744272e-01 2.13088483e-01 1.03179097e+00 1.94749370e-01
-5.58774054e-01 9.31921378e-02 4.03045654e-01 -2.84006447e-01
5.38314581e-01 -8.25771034e-01 2.66862381e-02 8.93472612e-01
-2.44318292e-01 2.80758023e-01 8.07999372e-01 -1.00933456e+00
-1.58843255e+00 -1.24060166e+00 3.43794227e-01 1.18467771e-01
4.10515398e-01 -1.72520086e-01 -8.61684561e-01 5.94528913e-01
1.62183836e-01 -5.42749129e-02 -4.91892062e-02 -6.72400236e-01
5.02208412e-01 -1.35389298e-01 -1.39446068e+00 3.85531723e-01
5.26228666e-01 4.39743370e-01 2.54923195e-01 2.52120882e-01
6.23154640e-01 -5.26032507e-01 -6.80937529e-01 8.08170319e-01
5.87349772e-01 -5.88317156e-01 7.94960022e-01 7.81605393e-02
7.27641070e-03 -8.81826460e-01 -3.47500220e-02 -1.08248329e+00
-3.88237923e-01 -2.83249736e-01 2.98929930e-01 1.39733303e+00
-3.25045809e-02 -5.00680983e-01 6.39877141e-01 -1.58555523e-01
7.85641223e-02 -2.57811815e-01 -4.18739647e-01 -7.99832761e-01
-2.39035949e-01 2.89276689e-01 9.34467018e-01 7.75140047e-01
-1.96891785e-01 -2.42731005e-01 -7.51028955e-02 1.00615561e+00
6.81235492e-01 3.77360612e-01 6.72597408e-01 -1.40137374e+00
1.30074710e-01 -1.00280456e-01 -3.44542980e-01 -9.19983566e-01
-2.17992783e-01 -6.83089197e-01 7.01206326e-01 -1.63687336e+00
5.02947606e-02 -7.80582130e-01 9.27385911e-02 6.96473479e-01
-2.11517245e-01 1.60211194e-02 1.55417144e-01 -7.25257769e-02
-2.43539065e-01 -2.54580518e-03 1.31098735e+00 -2.77244627e-01
-6.39275789e-01 4.28527743e-01 -2.03993306e-01 5.64548016e-01
1.05122948e+00 -7.34592676e-01 -2.29310334e-01 -6.38117194e-01
-1.06151953e-01 2.25210890e-01 6.02882504e-01 -1.05199170e+00
3.01504135e-01 -3.15635741e-01 1.66573361e-01 -1.13485932e+00
4.64556664e-02 -1.39311004e+00 4.35938209e-01 1.05488348e+00
2.34662056e-01 1.59256637e-01 6.18963778e-01 5.38978755e-01
2.80954480e-01 -7.60031581e-01 7.39522636e-01 -1.01540893e-01
-8.21163833e-01 1.03620462e-01 -7.56503344e-01 -8.44650805e-01
1.67706668e+00 -4.20458198e-01 -1.01981118e-01 2.88417101e-01
-4.88868952e-01 7.07281113e-01 4.55466568e-01 4.59312618e-01
2.84082204e-01 -9.42630112e-01 -7.28552580e-01 1.17169127e-01
1.57346219e-01 1.99108452e-01 1.45060703e-01 9.31215048e-01
-1.02319014e+00 5.03122449e-01 -4.14631635e-01 -1.09133112e+00
-1.50093675e+00 7.76124001e-01 4.97392043e-02 -3.43118142e-03
-5.67171752e-01 4.62805957e-01 3.64746839e-01 -4.27818775e-01
1.14259720e-01 -8.42355430e-01 -6.37618184e-01 8.85455310e-02
1.56149313e-01 4.47076619e-01 -1.35849565e-01 -5.42370141e-01
-5.31883776e-01 7.17474222e-01 1.63030669e-01 3.72193724e-01
1.21973431e+00 -2.72990197e-01 -2.56166309e-01 3.48636061e-01
3.14469963e-01 -2.99036384e-01 -1.09417486e+00 1.85041726e-01
2.52691269e-01 -2.21948668e-01 -2.29353532e-01 -5.07595181e-01
-1.17006242e+00 5.36670864e-01 1.10598350e+00 5.46972871e-01
1.11353338e+00 -1.64240018e-01 2.50041664e-01 4.01599467e-01
6.11647248e-01 -5.30946493e-01 -4.42630500e-01 2.67374128e-01
3.78816247e-01 -1.11866403e+00 8.08322430e-02 -9.24168587e-01
-3.77964050e-01 1.14921534e+00 6.80527031e-01 4.95342612e-02
4.21682954e-01 5.19599855e-01 1.35885164e-01 -2.97885060e-01
-1.35819778e-01 -1.95398584e-01 -4.57074851e-01 9.06985521e-01
3.38343799e-01 2.43404999e-01 -3.36192578e-01 1.13985971e-01
4.15327221e-01 1.87248915e-01 5.17523050e-01 1.37963712e+00
-6.64911687e-01 -9.82846498e-01 -7.23014176e-01 2.20202044e-01
-2.52437830e-01 3.74203742e-01 4.22709212e-02 8.89684618e-01
4.15295869e-01 1.17955756e+00 7.35034794e-02 1.17583580e-01
3.80540371e-01 -4.81995463e-01 3.59107971e-01 -6.80544555e-01
-7.04282522e-01 1.43494114e-01 -4.57611680e-01 -2.18428329e-01
-8.81782830e-01 -7.25647867e-01 -1.51112139e+00 -2.12845713e-01
-9.52055752e-01 1.42305300e-01 8.01095545e-01 4.65807110e-01
1.07218765e-01 6.16690934e-01 4.90788609e-01 -9.90192533e-01
-1.85158178e-01 -7.37707913e-01 -9.08443987e-01 -3.31748798e-02
2.67303944e-01 -9.15232122e-01 4.58768681e-02 2.82976151e-01] | [9.035737991333008, -1.5670366287231445] |
679a766a-75cd-4822-bf1f-8728a15d2baa | graph-differentiable-architecture-search-with | null | null | http://proceedings.neurips.cc/paper/2021/hash/8c9f32e03aeb2e3000825c8c875c4edd-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/8c9f32e03aeb2e3000825c8c875c4edd-Paper.pdf | Graph Differentiable Architecture Search with Structure Learning | Discovering ideal Graph Neural Networks (GNNs) architectures for different tasks is labor intensive and time consuming. To save human efforts, Neural Architecture Search (NAS) recently has been used to automatically discover adequate GNN architectures for certain tasks in order to achieve competitive or even better performance compared with manually designed architectures. However, existing works utilizing NAS to search GNN structures fail to answer the question: how NAS is able to select the desired GNN architectures? In this paper, we investigate this question to solve the problem, for the first time. We conduct a measurement study with experiments to discover that gradient based NAS methods tend to select proper architectures based on the usefulness of different types of information with respect to the target task. Our explorations further show that gradient based NAS also suffers from noises hidden in the graph, resulting in searching suboptimal GNN architectures. Based on our findings, we propose a Graph differentiable Architecture Search model with Structure Optimization (GASSO), which allows differentiable search of the architecture with gradient descent and is able to discover graph neural architectures with better performance through employing graph structure learning as a denoising process in the search procedure. The proposed GASSO model is capable of simultaneously searching the optimal architecture and adaptively adjusting graph structure by jointly optimizing graph architecture search and graph structure denoising. Extensive experiments on real-world graph datasets demonstrate that our proposed GASSO model is able to achieve state-of-the-art performance compared with existing baselines. | ['Wenwu Zhu', 'Zeyang Zhang', 'Xin Wang', 'Yijian Qin'] | 2021-12-01 | null | https://openreview.net/forum?id=kSv_AMdehh3 | https://openreview.net/pdf?id=kSv_AMdehh3 | neurips-2021-12 | ['graph-structure-learning'] | ['graphs'] | [ 1.25454426e-01 1.87993065e-01 1.14108855e-02 -1.75072432e-01
-7.61511996e-02 -4.01031315e-01 2.53391862e-01 -7.51517564e-02
-1.66680232e-01 1.49471194e-01 -1.72261283e-01 -4.91228461e-01
-4.58785534e-01 -7.95009732e-01 -6.41014516e-01 -5.50137997e-01
-2.67200116e-02 5.00563860e-01 2.41619721e-02 -4.51323718e-01
4.50914860e-01 4.20376837e-01 -1.34816110e+00 -2.06871063e-01
1.07661033e+00 9.67348516e-01 4.41542983e-01 5.13795435e-01
-3.47705036e-01 5.84816039e-01 -4.46884036e-01 -5.74008107e-01
7.01045752e-01 -5.87000310e-01 -6.09949350e-01 7.08233193e-02
6.84351981e-01 1.51338473e-01 -3.78596842e-01 1.57446921e+00
3.52650791e-01 1.63402185e-01 5.06713152e-01 -1.09674942e+00
-8.73065174e-01 8.95331085e-01 -4.80348170e-01 5.26272655e-01
-1.99416071e-01 2.25741729e-01 1.44787550e+00 -7.04955757e-01
2.95214593e-01 1.24067485e+00 6.08316660e-01 5.45631289e-01
-1.26492035e+00 -6.23772502e-01 4.92295623e-01 3.04106504e-01
-1.48630440e+00 -3.44664752e-01 1.35020721e+00 -4.79243658e-02
1.04671860e+00 1.72091663e-01 7.83465147e-01 1.08146429e+00
1.73601523e-01 6.78227305e-01 6.86517596e-01 -4.21189189e-01
2.29979351e-01 -1.86419785e-01 3.78338069e-01 1.18903041e+00
6.86902285e-01 1.80984996e-02 -4.81732041e-01 1.12292394e-01
8.71771574e-01 -3.75528663e-01 -3.54900479e-01 -3.64773750e-01
-8.00996184e-01 7.51917601e-01 9.90630686e-01 4.59370077e-01
-6.80022895e-01 3.24388385e-01 1.77900866e-01 6.75687909e-01
1.93113357e-01 9.05975342e-01 -3.07535768e-01 2.24650219e-01
-5.63459396e-01 -1.48164213e-01 7.41748691e-01 8.97480130e-01
7.20025659e-01 7.17535734e-01 1.23394042e-01 7.84108281e-01
4.48373795e-01 4.77151424e-02 5.86739063e-01 -5.86012363e-01
5.62544584e-01 1.24963629e+00 -6.60097778e-01 -1.38403642e+00
-5.61380446e-01 -1.13528991e+00 -1.12964630e+00 -2.59867567e-03
2.61718273e-01 -5.32696582e-02 -1.11134064e+00 1.74697959e+00
1.61241993e-01 3.82910728e-01 -3.45773697e-02 1.03140593e+00
9.49558675e-01 3.71862501e-01 -5.12667857e-02 1.78495068e-02
1.17674112e+00 -1.04121971e+00 -3.77910137e-01 -5.21046102e-01
5.94549596e-01 -2.43436366e-01 1.24762869e+00 5.51135719e-01
-8.39882612e-01 -6.33901298e-01 -1.21309173e+00 3.76983970e-01
-1.09119445e-01 1.08855799e-01 7.30969071e-01 9.13049936e-01
-1.50461721e+00 6.32076383e-01 -8.49075615e-01 -4.14335668e-01
3.20013314e-01 5.81864595e-01 -1.44757450e-01 1.96282610e-01
-1.00614715e+00 5.09467423e-01 6.35907948e-01 4.52316076e-01
-8.19924772e-01 -4.32041228e-01 -6.93641961e-01 3.26717466e-01
6.74366653e-01 -9.88117218e-01 7.38677979e-01 -1.27060127e+00
-1.44041502e+00 3.85245323e-01 2.42403060e-01 -7.58432150e-01
1.15868539e-01 2.16923058e-01 -4.12000537e-01 1.11084983e-01
-4.16644305e-01 4.51762408e-01 1.05028582e+00 -1.20867693e+00
-3.00254971e-01 -5.73248327e-01 1.16931573e-01 3.30446988e-01
-7.58896530e-01 -5.80253363e-01 -6.26304865e-01 -6.54326618e-01
6.20051324e-01 -1.02735031e+00 -3.80634993e-01 -6.36255264e-01
-6.13348663e-01 -4.20609802e-01 4.20385778e-01 -4.23963517e-01
1.51954949e+00 -1.88946617e+00 3.72937053e-01 6.87542677e-01
7.08895802e-01 3.70730132e-01 -4.87452298e-01 7.57054240e-02
6.72523165e-03 3.34071457e-01 8.48388150e-02 -2.46716440e-01
-1.05800003e-01 1.97283909e-01 1.06449760e-01 2.03726113e-01
3.03891208e-03 1.11580253e+00 -6.37752533e-01 -3.71571332e-01
-2.42265034e-02 3.53236943e-01 -4.45987076e-01 5.46496995e-02
-1.75118282e-01 2.33775511e-01 -6.68071747e-01 7.86144495e-01
2.94508994e-01 -6.21231794e-01 4.65936869e-01 -4.28426206e-01
4.06006753e-01 -1.07634395e-01 -1.09435213e+00 1.35453749e+00
-2.03570917e-01 4.46543097e-01 8.32850635e-02 -1.38304949e+00
1.37824202e+00 -1.29013151e-01 1.19803801e-01 -8.86024535e-01
4.02237326e-01 3.30333829e-01 4.87317085e-01 -3.23839456e-01
2.77538240e-01 1.80489078e-01 2.09759220e-01 1.18250623e-01
1.32439911e-01 3.92946571e-01 9.95963961e-02 3.98016982e-02
1.42726898e+00 -3.72614801e-01 2.13904008e-01 -4.27158713e-01
6.10162377e-01 1.54323084e-02 3.37261677e-01 8.76866579e-01
-2.59676516e-01 2.36911878e-01 4.34798360e-01 -7.06882238e-01
-8.27830195e-01 -7.67923176e-01 5.59686840e-01 1.01758444e+00
1.75872803e-01 -1.99976206e-01 -9.03123438e-01 -8.74214709e-01
-3.95086408e-01 6.59267128e-01 -6.45205915e-01 -3.86462033e-01
-7.90239632e-01 -6.80736780e-01 4.73984778e-01 3.80483866e-01
5.69673777e-01 -1.37707102e+00 -4.71447200e-01 1.62244961e-01
8.36794004e-02 -8.84947479e-01 -4.99027371e-01 2.57227063e-01
-1.19037473e+00 -1.08023405e+00 -4.66105342e-01 -1.04113066e+00
1.05925429e+00 3.36297810e-01 1.25917256e+00 7.29271770e-01
7.52701461e-02 2.81784058e-01 -3.77897590e-01 -1.81815922e-01
-4.61901128e-01 6.52928948e-01 -1.09206721e-01 1.71355397e-01
3.43356818e-01 -8.68965745e-01 -7.35646963e-01 2.12361798e-01
-9.44564402e-01 -5.49993590e-02 9.94138956e-01 7.36019969e-01
6.59367800e-01 3.92543554e-01 7.22261906e-01 -1.11683547e+00
1.26716590e+00 -3.68254811e-01 -8.26055944e-01 4.48343217e-01
-1.29409361e+00 5.56089461e-01 9.33900714e-01 -2.60439873e-01
-5.08491635e-01 -2.26324797e-02 6.61736429e-02 -8.47651541e-01
1.06616542e-01 9.43477213e-01 2.03085821e-02 -3.58000338e-01
8.78393888e-01 4.20404583e-01 2.28409767e-01 -5.03658175e-01
1.15050912e-01 -3.27769965e-02 4.54697251e-01 -3.42639089e-01
8.29483271e-01 1.61541789e-03 3.22359830e-01 -7.38008261e-01
-6.04182720e-01 -3.76116097e-01 -4.98757720e-01 -2.42380410e-01
5.84622741e-01 -5.64652324e-01 -6.20474339e-01 2.92640328e-01
-7.16338396e-01 -2.71176696e-01 2.46388465e-01 1.75076559e-01
-1.74421623e-01 4.84846324e-01 -3.59809458e-01 -6.19110346e-01
-9.36194718e-01 -1.23845911e+00 6.30587459e-01 3.24734867e-01
1.68003850e-02 -1.16253090e+00 -1.25212461e-01 2.69193143e-01
5.73289633e-01 1.36287078e-01 1.10733080e+00 -8.58423412e-01
-7.78514862e-01 -1.29167974e-01 -1.93223163e-01 2.45835513e-01
-1.50249712e-02 -3.65391582e-01 -3.88463646e-01 -4.81210828e-01
6.41300380e-02 -2.47853752e-02 9.59323645e-01 4.86342967e-01
1.11095023e+00 -5.34893751e-01 -1.36956960e-01 8.70065153e-01
1.74715555e+00 2.32270539e-01 6.12668335e-01 6.14010632e-01
1.12807655e+00 4.55605179e-01 1.60265505e-01 8.43863338e-02
2.87472993e-01 3.72945696e-01 9.42761302e-01 1.03902243e-01
-2.38480404e-01 -2.37620622e-01 1.14152908e-01 9.86242950e-01
-2.03261748e-01 -5.26891351e-01 -1.08155167e+00 4.83195186e-01
-1.77615738e+00 -4.21090066e-01 -5.29248789e-02 1.85831308e+00
2.78570443e-01 3.42721343e-01 1.13748990e-01 -4.22936715e-02
6.27187252e-01 1.84425965e-01 -8.25808108e-01 -3.27340394e-01
6.82870895e-02 1.64831221e-01 5.34275532e-01 2.11079732e-01
-7.53772080e-01 9.69638824e-01 5.50901270e+00 6.91864312e-01
-1.16092980e+00 -3.69436532e-01 5.53600669e-01 1.74249381e-01
-3.94278347e-01 -6.72561601e-02 -6.08178377e-01 1.84194446e-01
8.69869471e-01 -1.09219685e-01 9.37037766e-01 9.17764962e-01
-6.77990615e-02 6.73016846e-01 -7.82673299e-01 1.21868491e+00
1.14728861e-01 -1.26089203e+00 5.08792341e-01 1.30320385e-01
5.73443174e-01 4.74540256e-02 1.44160941e-01 4.01580691e-01
3.03240269e-01 -1.17883718e+00 6.59922481e-01 4.10587877e-01
1.67061031e-01 -7.60000944e-01 7.61477649e-01 3.10341239e-01
-1.49498498e+00 -4.24427509e-01 -4.16510701e-01 2.01823816e-01
-3.18218470e-01 2.35429212e-01 -1.14429069e+00 7.24068403e-01
5.47317266e-01 4.08376008e-01 -1.05716181e+00 1.10702765e+00
-1.82035521e-01 7.85952568e-01 -1.92876756e-01 -4.77078110e-01
5.71950436e-01 -4.20151740e-01 7.31913447e-01 6.89155817e-01
5.22911608e-01 -5.70145398e-02 2.37134725e-01 1.08304918e+00
-2.03647166e-01 3.04928690e-01 -6.37240529e-01 -2.94272095e-01
4.63758260e-01 1.21235895e+00 -1.17145872e+00 4.60785553e-02
-1.36843756e-01 6.71522021e-01 6.83535337e-01 4.73782003e-01
-5.33331513e-01 -3.62642109e-01 1.67266205e-01 1.70237094e-01
3.76826823e-01 -3.32531601e-01 -4.95262682e-01 -8.02833855e-01
3.79417874e-02 -1.13439643e+00 6.81508482e-01 -4.36537415e-01
-1.24787474e+00 1.13957441e+00 -2.54153073e-01 -1.00215876e+00
-9.69994590e-02 -5.13418734e-01 -6.60191596e-01 4.66418058e-01
-1.27305222e+00 -1.03763533e+00 -5.34923494e-01 5.37791908e-01
6.16820574e-01 -6.56968832e-01 3.11500907e-01 8.21865648e-02
-8.48761141e-01 8.93033981e-01 -3.42424780e-01 1.46615446e-01
8.20391253e-02 -1.32244706e+00 6.65162563e-01 1.06399679e+00
6.69932663e-01 7.86887050e-01 7.48130798e-01 -7.21189320e-01
-1.81665945e+00 -1.06617999e+00 2.89596587e-01 -8.68068635e-02
5.86785316e-01 -9.81306136e-02 -1.10804737e+00 4.50225949e-01
5.88970035e-02 -2.27850541e-01 2.08513796e-01 2.42005050e-01
-2.65098810e-01 -2.23582298e-01 -7.71861017e-01 7.01625109e-01
1.45046318e+00 -6.22057468e-02 -3.52868676e-01 6.71085378e-04
8.73581231e-01 -2.82618463e-01 -4.99484658e-01 5.03362417e-01
2.31642887e-01 -7.89155960e-01 9.41394389e-01 -4.96882498e-01
-1.39355259e-02 -2.81223416e-01 1.30410921e-02 -1.44219697e+00
-4.89290029e-01 -5.58475316e-01 -2.48604849e-01 9.08785105e-01
7.00156689e-01 -8.42926979e-01 1.19764304e+00 3.40962201e-01
-3.66850466e-01 -1.00112617e+00 -7.78395414e-01 -8.31468463e-01
-3.19993198e-01 -3.10983717e-01 6.97140455e-01 8.05414975e-01
-6.54509723e-01 7.27509916e-01 -7.31723979e-02 3.69452536e-01
7.96210706e-01 7.66685680e-02 7.02659786e-01 -1.43235767e+00
-3.66383493e-01 -9.51216400e-01 -5.76498032e-01 -8.95972967e-01
3.64091605e-01 -1.15153742e+00 -3.56010608e-02 -1.63217008e+00
-1.52606070e-01 -4.09791470e-01 -5.96855164e-01 4.86701638e-01
-2.73682743e-01 -6.89691980e-04 3.74188982e-02 2.89180487e-01
-5.90771556e-01 4.99250680e-01 1.16734827e+00 -3.20832521e-01
-4.81214404e-01 8.65629017e-02 -1.15301132e+00 5.70069551e-01
8.75201821e-01 -3.35987359e-01 -1.03230119e+00 -6.62305415e-01
5.30190229e-01 -1.84993535e-01 3.43275219e-01 -1.07725596e+00
3.23692352e-01 1.61162779e-01 8.94842669e-02 -2.86070287e-01
-1.08365282e-01 -9.15430307e-01 2.71345973e-01 5.55549920e-01
-2.78673053e-01 4.65741575e-01 6.35243505e-02 9.54237700e-01
-1.25945494e-01 -4.18648511e-01 5.76064229e-01 -2.34900892e-01
-1.04811776e+00 4.38880056e-01 6.78522512e-02 -1.44906521e-01
6.39247179e-01 -6.02915466e-01 -2.20404044e-01 -3.85403335e-01
-6.34942710e-01 3.68194103e-01 1.95755869e-01 4.58145678e-01
8.61509085e-01 -1.12326407e+00 -5.84321916e-01 1.80898488e-01
-7.42074251e-02 -8.01427588e-02 1.45552337e-01 6.91527605e-01
-6.90532625e-01 3.05385947e-01 -4.59333807e-02 -6.33221626e-01
-1.23365676e+00 6.16381109e-01 4.57259029e-01 -5.42146921e-01
-7.18351722e-01 1.10214257e+00 1.55927956e-01 -4.08266127e-01
4.24157619e-01 -1.97216585e-01 -5.28645456e-01 -3.14253449e-01
1.39086515e-01 2.42157772e-01 2.65633732e-01 -4.42249984e-01
-1.82035848e-01 4.48971331e-01 -1.24011569e-01 4.54632550e-01
1.31429851e+00 -1.96892962e-01 -9.74684060e-02 -9.65083912e-02
9.89696920e-01 -4.44888383e-01 -9.96752203e-01 -3.34999710e-01
3.52214485e-01 -1.50397718e-01 2.71351069e-01 -4.84375834e-01
-1.65898550e+00 4.23017323e-01 6.74405336e-01 6.33193314e-01
1.49463475e+00 -4.05729562e-02 7.55364418e-01 7.02703774e-01
4.90666598e-01 -8.29319656e-01 3.85438740e-01 4.35806334e-01
9.52740371e-01 -1.09339941e+00 -1.57301128e-01 -3.35748255e-01
-5.42769313e-01 1.22124732e+00 1.01837027e+00 -3.54869008e-01
5.86501539e-01 -2.92992741e-01 -4.28483114e-02 -8.63867223e-01
-7.51814723e-01 -2.19620541e-01 6.96880341e-01 4.34795082e-01
4.49849963e-02 -7.68981874e-03 -6.19851379e-03 4.78208959e-01
-4.71614301e-01 -6.02331698e-01 2.46593535e-01 4.76888120e-01
-5.40374041e-01 -9.68264878e-01 -1.46964550e-01 7.42346883e-01
-1.95492804e-01 -1.86811984e-01 -7.27573514e-01 7.28613794e-01
-4.08922344e-01 8.73374283e-01 -2.78322309e-01 -8.05600882e-01
5.29970884e-01 -4.00343053e-02 3.33741754e-01 -4.55082268e-01
-7.52873719e-01 3.70054170e-02 5.44699505e-02 -5.23843229e-01
-1.42545387e-01 -1.43585324e-01 -1.15240932e+00 -1.76667154e-01
-5.41262209e-01 7.37201124e-02 5.89908361e-01 8.22454214e-01
6.04713500e-01 8.40243399e-01 4.66707438e-01 -5.33083439e-01
-7.18334794e-01 -8.38010550e-01 -4.15159374e-01 2.84571528e-01
9.17869285e-02 -5.66285253e-01 -3.93547803e-01 -3.53144705e-01] | [8.623514175415039, 3.439345359802246] |
9a3a43de-a5ee-40f5-ad04-077d7adf9b6b | counterfactual-inference-for-text | null | null | https://aclanthology.org/2021.acl-long.422 | https://aclanthology.org/2021.acl-long.422.pdf | Counterfactual Inference for Text Classification Debiasing | Today{'}s text classifiers inevitably suffer from unintended dataset biases, especially the document-level label bias and word-level keyword bias, which may hurt models{'} generalization. Many previous studies employed data-level manipulations or model-level balancing mechanisms to recover unbiased distributions and thus prevent models from capturing the two types of biases. Unfortunately, they either suffer from the extra cost of data collection/selection/annotation or need an elaborate design of balancing strategies. Different from traditional factual inference in which debiasing occurs before or during training, counterfactual inference mitigates the influence brought by unintended confounders after training, which can make unbiased decisions with biased observations. Inspired by this, we propose a model-agnostic text classification debiasing framework {--} Corsair, which can effectively avoid employing data manipulations or designing balancing mechanisms. Concretely, Corsair first trains a base model on a training set directly, allowing the dataset biases {`}poison{'} the trained model. In inference, given a factual input document, Corsair imagines its two counterfactual counterparts to distill and mitigate the two biases captured by the poisonous model. Extensive experiments demonstrate Corsair{'}s effectiveness, generalizability and fairness. | ['Pengjun Xie', 'Chunping Ma', 'Lijie Wen', 'Fuli Feng', 'Chen Qian'] | 2021-08-01 | null | null | null | acl-2021-5 | ['counterfactual-inference'] | ['miscellaneous'] | [ 2.58760363e-01 2.25017026e-01 -8.73860836e-01 -5.82291126e-01
-6.08906746e-01 -5.30637562e-01 8.06692302e-01 2.08494291e-02
-6.62809193e-01 1.09683394e+00 3.45510006e-01 -7.10035622e-01
-1.32524073e-01 -8.08801889e-01 -8.34268332e-01 -7.60702252e-01
4.58752394e-01 1.54726833e-01 -5.40779054e-01 1.06656313e-01
6.56749010e-01 1.40815318e-01 -1.34451807e+00 1.28646761e-01
1.10095048e+00 8.51849854e-01 -2.21269384e-01 1.89646721e-01
-2.12518826e-01 1.20640230e+00 -7.94237673e-01 -1.06568921e+00
3.08854699e-01 -2.19871342e-01 -6.65038645e-01 -2.57321119e-01
4.46998805e-01 -6.24424517e-01 -3.96177620e-01 1.27886140e+00
5.97336829e-01 -4.94806767e-02 9.10546243e-01 -1.75940657e+00
-1.23141313e+00 1.21466744e+00 -8.43290925e-01 3.59834842e-02
-1.19532168e-01 4.49040115e-01 9.26848650e-01 -5.11224270e-01
3.24180096e-01 1.57208395e+00 7.68725157e-01 8.52309942e-01
-1.31607354e+00 -1.40324664e+00 4.53491271e-01 -1.67203054e-01
-1.12525666e+00 -5.27875662e-01 8.53958488e-01 -5.04465818e-01
1.54140681e-01 5.91856539e-01 1.48355588e-01 1.79845154e+00
3.33049625e-01 1.00856221e+00 1.48949337e+00 -2.27826208e-01
4.38651264e-01 4.48647082e-01 5.65465927e-01 2.17893317e-01
9.22212362e-01 5.91680288e-01 -4.90723252e-01 -5.51513493e-01
1.51856840e-01 1.69117942e-01 -4.12676871e-01 -2.71109611e-01
-1.13352466e+00 1.03569508e+00 4.72253770e-01 -8.46590549e-02
-2.39165559e-01 1.68160513e-01 6.26184106e-01 3.51344734e-01
8.03977668e-01 5.70676744e-01 -6.93887711e-01 4.51700062e-01
-1.03087509e+00 5.74762106e-01 5.17673731e-01 1.08693278e+00
4.90249395e-01 -8.42315853e-02 -7.92899191e-01 6.64325655e-01
2.68132925e-01 8.27125371e-01 6.82571054e-01 -7.37180591e-01
5.20036399e-01 3.87357533e-01 2.65562296e-01 -9.54550564e-01
-2.12897807e-01 -4.82581913e-01 -1.26264095e+00 1.05489969e-01
6.75733089e-01 -1.90070510e-01 -7.51959741e-01 2.07025170e+00
3.74211907e-01 -2.88207024e-01 -1.83969811e-01 9.21131730e-01
4.30069387e-01 1.21285073e-01 5.51456630e-01 -3.80803406e-01
1.41493249e+00 -4.60203648e-01 -9.91419673e-01 -2.55083382e-01
8.17168951e-01 -4.30998325e-01 1.49591017e+00 3.50185245e-01
-8.45986843e-01 -8.06151107e-02 -9.98953223e-01 -2.07990900e-01
-4.30696487e-01 -3.47847641e-01 6.87885404e-01 1.03596258e+00
-4.56012934e-01 6.16943002e-01 -9.24402922e-02 3.21114779e-01
9.34545398e-01 -2.80426554e-02 8.24107789e-03 -1.04553938e-01
-1.57537496e+00 8.93187106e-01 3.27416152e-01 7.79757500e-02
-9.32713985e-01 -1.36743367e+00 -5.66990495e-01 5.57542443e-02
5.75376809e-01 -9.84224141e-01 1.27851295e+00 -1.08342302e+00
-1.05531394e+00 1.02786088e+00 -5.22345901e-02 -4.35690045e-01
1.17451036e+00 -2.24093899e-01 -1.57549411e-01 -7.31883764e-01
2.47147888e-01 3.15142304e-01 1.02822876e+00 -1.44145298e+00
-3.96155149e-01 -6.90355301e-01 -1.80993080e-02 -1.07966848e-01
-3.30579787e-01 -3.57063375e-02 3.41289788e-01 -1.20720923e+00
-3.51868749e-01 -5.76948822e-01 -9.12305191e-02 2.08237335e-01
-7.96660662e-01 -1.43029302e-01 6.93833947e-01 -4.95055467e-01
1.36986387e+00 -1.90754032e+00 -5.28954089e-01 -6.42108694e-02
3.12849075e-01 1.84366137e-01 -9.74548701e-03 -7.45298639e-02
-2.31194749e-01 6.83085382e-01 -2.95777082e-01 -3.09892714e-01
3.07310194e-01 2.24473830e-02 -8.36189687e-01 6.08932972e-01
-1.76051632e-01 8.12129974e-01 -8.61873507e-01 -3.33781004e-01
-1.46642942e-02 2.68334020e-02 -6.38865590e-01 1.21963024e-01
-3.32778007e-01 8.73703137e-02 -3.16860259e-01 3.44250977e-01
1.01607907e+00 1.65301152e-02 1.20111987e-01 2.49453783e-02
1.25100479e-01 6.49504006e-01 -8.86085391e-01 1.13757825e+00
-5.84377587e-01 2.95561969e-01 1.10526800e-01 -9.44023192e-01
7.47431993e-01 1.03163876e-01 -1.58042818e-01 -6.20135784e-01
5.06072938e-01 1.18193500e-01 -1.80644348e-01 -4.87399071e-01
4.18108165e-01 -7.41843283e-01 -2.77003556e-01 8.79247487e-01
-4.08266872e-01 -5.48696220e-02 -5.65608740e-01 3.48516405e-01
5.82473755e-01 -2.28398383e-01 4.52553421e-01 -8.39740515e-01
2.37619847e-01 -4.60444801e-02 6.74689710e-01 1.22748566e+00
-3.44079226e-01 2.28577152e-01 6.56558037e-01 -4.55417246e-01
-8.75853002e-01 -8.03663433e-01 -4.26062733e-01 1.30517137e+00
8.92126001e-03 2.34994173e-01 -6.47101045e-01 -1.21790314e+00
5.05539656e-01 1.58222747e+00 -1.18092370e+00 -6.52061105e-01
-2.12815441e-02 -1.14458668e+00 7.88493872e-01 3.08055311e-01
5.04447162e-01 -6.89930439e-01 -2.68564463e-01 -1.91956326e-01
-2.63865292e-01 -2.08764553e-01 -7.65563488e-01 -5.96021898e-02
-7.48839140e-01 -1.15088856e+00 -5.21364629e-01 9.92927924e-02
5.29242456e-01 4.41023648e-01 1.21211231e+00 8.06811973e-02
1.02368355e-01 -2.40694448e-01 -9.25365463e-03 -1.13077247e+00
-6.00967288e-01 -1.49654239e-01 4.15263325e-02 -1.55425072e-01
7.27301657e-01 -3.83493960e-01 -7.91026115e-01 2.53679931e-01
-1.01112449e+00 -6.96147010e-02 3.77422899e-01 1.05241215e+00
-3.90411960e-03 -1.11604378e-01 9.93366241e-01 -1.47041500e+00
8.95932138e-01 -7.63495684e-01 -4.81655508e-01 3.22363347e-01
-1.17012203e+00 9.30955857e-02 6.56942308e-01 -6.83978140e-01
-1.41222239e+00 -7.14694679e-01 3.51476401e-01 -3.02758366e-01
-6.27801865e-02 2.18251824e-01 -6.75073266e-01 6.72879040e-01
1.00284278e+00 -1.92063585e-01 -7.21559953e-03 -3.87355208e-01
5.89507341e-01 1.25589395e+00 4.34310317e-01 -7.84186602e-01
6.74691260e-01 5.92804790e-01 -3.74726564e-01 -3.90258443e-04
-1.56148088e+00 -1.78730804e-02 -3.15683424e-01 1.35524049e-01
4.52891231e-01 -8.83796632e-01 -8.84819210e-01 5.08521557e-01
-1.07741797e+00 -2.29547188e-01 -3.88971657e-01 2.72181392e-01
-2.12056398e-01 5.55539913e-02 -2.56523758e-01 -1.19577253e+00
-4.35710877e-01 -6.85840130e-01 8.21518242e-01 -1.42312527e-01
-4.82880741e-01 -9.97629166e-01 -1.21711135e-01 5.59994519e-01
4.70533371e-01 4.68410552e-02 1.18614447e+00 -8.43738556e-01
-2.71207154e-01 -4.73952264e-01 -4.20754880e-01 3.24974239e-01
1.11112431e-01 -2.19381750e-01 -1.46262550e+00 -2.85098046e-01
3.87745678e-01 -3.39831084e-01 8.81167948e-01 3.93844366e-01
1.67503595e+00 -1.08284593e+00 -3.49723876e-01 3.43589187e-01
1.08119214e+00 4.55316305e-02 4.42277610e-01 1.94596067e-01
6.02426887e-01 8.17332625e-01 5.12903154e-01 5.59617817e-01
3.78384084e-01 2.94557422e-01 3.84950995e-01 9.63931065e-03
1.90097436e-01 -7.98009515e-01 1.98767513e-01 4.58460376e-02
3.87501329e-01 -2.28263304e-01 -5.43373406e-01 5.96142232e-01
-1.74125564e+00 -1.05099070e+00 -2.69554824e-01 2.46735144e+00
1.21640396e+00 3.20144184e-03 -2.10950840e-02 2.40153387e-01
7.96504140e-01 2.83830792e-01 -8.80044580e-01 -3.65397274e-01
-1.24705136e-01 -4.20656234e-01 7.03075826e-01 4.28809375e-01
-8.84994209e-01 5.87000072e-01 6.05083990e+00 9.61453736e-01
-9.51168478e-01 1.47903755e-01 9.95792985e-01 -5.58293581e-01
-9.97784078e-01 -1.85772739e-02 -6.65523112e-01 7.71621764e-01
7.48051047e-01 -7.01874435e-01 3.08835983e-01 9.69099462e-01
4.17746097e-01 -6.90639764e-02 -1.41610956e+00 7.14568317e-01
-3.09049189e-02 -1.23008573e+00 4.58727986e-01 1.47189811e-01
7.93707907e-01 -3.87121201e-01 1.70084104e-01 5.37591457e-01
1.02063334e+00 -1.18880606e+00 1.11894965e+00 4.73991781e-01
8.44826818e-01 -6.78815961e-01 7.26011813e-01 8.38062108e-01
-3.84517722e-02 -2.75047034e-01 -4.79379445e-01 -4.19687122e-01
-2.24140212e-01 1.16250610e+00 -3.19779962e-01 4.04025465e-01
5.38793802e-01 2.78570175e-01 -3.42151612e-01 4.17556375e-01
-3.01974744e-01 7.89641380e-01 2.40979865e-01 -1.92387268e-01
-1.57292143e-01 -1.10068191e-02 3.49928349e-01 1.23975754e+00
-8.09319317e-02 6.47863299e-02 -3.27141047e-01 1.17888367e+00
-6.62781954e-01 -3.33426259e-02 -9.02796388e-01 4.52791333e-01
8.78028393e-01 8.28923583e-01 -5.52654862e-02 -6.42395973e-01
-1.79974943e-01 5.55147946e-01 3.84108275e-01 4.35528874e-01
-7.44319320e-01 -1.42337814e-01 7.72843778e-01 1.34420052e-01
-4.73376095e-01 6.18911982e-01 -1.12559152e+00 -1.36523485e+00
-1.11123428e-01 -1.12156153e+00 6.38734162e-01 -6.46945536e-01
-1.86182129e+00 -1.83452144e-01 1.05810106e-01 -6.86757386e-01
2.57097155e-01 -4.44491565e-01 -5.21073878e-01 9.94532108e-01
-1.42427564e+00 -9.54823971e-01 2.82489099e-02 4.82498407e-01
2.86305398e-01 2.03851521e-01 5.07127225e-01 1.24387732e-02
-8.87234628e-01 9.57496405e-01 1.28361642e-01 9.59748551e-02
1.09422636e+00 -1.24966192e+00 2.13704824e-01 6.01792336e-01
-3.15243900e-01 1.10551476e+00 8.52863550e-01 -7.23153055e-01
-9.59662378e-01 -1.24765956e+00 1.08726251e+00 -8.79686415e-01
7.09654808e-01 -5.27420402e-01 -1.11571288e+00 8.12596142e-01
-2.30738224e-05 -3.66328537e-01 8.07550371e-01 3.73256952e-01
-9.77184057e-01 -2.39011690e-01 -1.48058224e+00 8.84333909e-01
1.26281977e+00 -5.54645002e-01 -9.18168724e-01 2.98286915e-01
7.52566278e-01 1.21114478e-01 -3.17650020e-01 2.97100216e-01
6.41975582e-01 -8.84003878e-01 6.86291277e-01 -1.32856047e+00
7.81962156e-01 1.29091129e-01 -1.19334079e-01 -1.40978098e+00
-2.49438509e-01 -4.99229968e-01 1.52591467e-01 1.58332098e+00
4.38560098e-01 -7.75253773e-01 4.64276642e-01 1.26970661e+00
3.25480849e-01 -2.17435032e-01 -8.57961297e-01 -7.18717575e-01
8.92758667e-01 -6.66701257e-01 1.15129209e+00 1.60943747e+00
1.33048698e-01 2.73493558e-01 -6.85612321e-01 -3.45766284e-02
9.13040280e-01 2.01375678e-01 8.58468652e-01 -1.06626797e+00
2.97797751e-02 -5.68384886e-01 4.59377468e-01 -8.32071066e-01
4.80115861e-01 -9.19440031e-01 4.87398989e-02 -8.52214277e-01
7.14521646e-01 -5.76617897e-01 -3.44274223e-01 4.18974757e-01
-7.86839724e-01 -1.41188681e-01 4.67623472e-02 1.54025182e-01
-7.46554285e-02 6.59451127e-01 1.25765741e+00 -2.43590012e-01
1.10474162e-01 6.82257712e-02 -1.65086234e+00 7.91971087e-01
7.01917589e-01 -7.64905393e-01 -5.46361804e-01 -3.38221133e-01
3.41678351e-01 -1.08263999e-01 8.10039163e-01 1.33378059e-01
-3.27723436e-02 -6.20838165e-01 2.36925006e-01 -1.88572094e-01
-4.04432863e-01 -8.09513807e-01 8.84681009e-03 4.88375485e-01
-9.23321903e-01 -3.61883819e-01 5.35404775e-03 7.57992089e-01
2.48371094e-01 -2.43343040e-01 8.51013958e-01 -1.34634480e-01
1.99913725e-01 1.16923362e-01 -3.09499770e-01 3.26699287e-01
7.70207345e-01 2.99569041e-01 -9.43848729e-01 -1.86823636e-01
-9.89015698e-02 3.09345245e-01 4.66084898e-01 4.75830317e-01
-8.65592510e-02 -1.15082276e+00 -7.80295551e-01 5.11008166e-02
2.02650145e-01 5.02451696e-02 4.00154144e-01 7.44040549e-01
2.88584501e-01 3.06099772e-01 1.19116805e-01 -5.23114055e-02
-8.66964579e-01 1.16187119e+00 3.09962392e-01 -3.02362293e-01
-9.45466310e-02 7.68404603e-01 6.62287831e-01 -6.56174660e-01
2.57768810e-01 -1.57349169e-01 3.31620783e-01 2.34914675e-01
7.70018041e-01 6.64321661e-01 -5.11937067e-02 5.92189096e-02
-1.91555694e-01 -2.00410426e-01 -3.21723700e-01 2.12372597e-02
7.43226171e-01 -3.17997754e-01 -5.61433956e-02 4.22498852e-01
9.29294705e-01 9.71223786e-02 -1.18669212e+00 -3.40656996e-01
8.34148973e-02 -8.27663720e-01 2.97267288e-01 -1.24970508e+00
-7.47919977e-01 1.00327420e+00 4.03628983e-02 1.29972845e-01
8.67575169e-01 -3.58165145e-01 3.54814023e-01 2.96038445e-02
3.44238311e-01 -8.82808924e-01 -4.34576958e-01 -5.91819063e-02
1.11589694e+00 -1.37855422e+00 1.14759296e-01 -1.97960034e-01
-7.26118863e-01 2.51283079e-01 4.69558477e-01 1.46820039e-01
5.97742319e-01 2.20790014e-01 2.56729305e-01 3.13518979e-02
-9.58329380e-01 3.65498483e-01 1.01848006e-01 5.68200946e-01
3.93466830e-01 2.01642707e-01 -6.01370335e-01 1.11567116e+00
-4.43894595e-01 1.75784722e-01 4.04089838e-01 5.05307913e-01
-3.60189751e-02 -7.32925713e-01 -6.41714215e-01 7.55569458e-01
-6.67166054e-01 -2.79321671e-01 -6.16382718e-01 1.01834440e+00
8.68384540e-02 9.86452520e-01 -1.04148626e-01 -1.70433626e-01
4.72024113e-01 2.95055479e-01 -6.08708262e-02 -2.80248523e-01
-7.06175387e-01 -4.24150646e-01 7.92780742e-02 -3.52156788e-01
-3.04142267e-01 -6.79675400e-01 -5.92191160e-01 -9.61942494e-01
-5.32942474e-01 1.29885897e-01 3.06675613e-01 1.03590977e+00
2.78727949e-01 2.33108699e-01 7.85159051e-01 -2.51284659e-01
-1.48597586e+00 -1.20067883e+00 -7.07580090e-01 8.18222225e-01
4.77989048e-01 -5.48898935e-01 -8.22432220e-01 -8.66526216e-02] | [10.271683692932129, 7.7074666023254395] |
b8ed3ee1-fe3d-423e-bd01-7015c6988625 | an-attentive-listening-system-with-android | null | null | https://aclanthology.org/2020.sigdial-1.15 | https://aclanthology.org/2020.sigdial-1.15.pdf | An Attentive Listening System with Android ERICA: Comparison of Autonomous and WOZ Interactions | We describe an attentive listening system for the autonomous android robot ERICA. The proposed system generates several types of listener responses: backchannels, repeats, elaborating questions, assessments, generic sentimental responses, and generic responses. In this paper, we report a subjective experiment with 20 elderly people. First, we evaluated each system utterance excluding backchannels and generic responses, in an offline manner. It was found that most of the system utterances were linguistically appropriate, and they elicited positive reactions from the subjects. Furthermore, 58.2% of the responses were acknowledged as being appropriate listener responses. We also compared the proposed system with a WOZ system where a human operator was operating the robot. From the subjective evaluation, the proposed system achieved comparable scores in basic skills of attentive listening such as encouragement to talk, focused on the talk, and actively listening. It was also found that there is still a gap between the system and the WOZ for more sophisticated skills such as dialogue understanding, showing interest, and empathy towards the user. | ['Tatsuya Kawahara', 'Katsuya Takanashi', 'Shizuka Nakamura', 'Kenta Yamamoto', 'Divesh Lala', 'Koji Inoue'] | null | null | null | null | sigdial-acl-2020-7 | ['dialogue-understanding'] | ['natural-language-processing'] | [-3.40823978e-01 9.56440687e-01 2.06163406e-01 -7.52309918e-01
-4.31822628e-01 -2.90478021e-01 1.16685137e-01 2.26797596e-01
-3.91126722e-01 9.07459021e-01 5.65809548e-01 1.88650578e-01
3.45294267e-01 -3.28645468e-01 -9.75622144e-03 -4.16772276e-01
2.21386209e-01 1.80266336e-01 -7.16003627e-02 -9.01798487e-01
3.71872842e-01 -4.76465225e-02 -1.39548576e+00 1.31945759e-01
7.57557034e-01 1.04536331e+00 2.38865972e-01 8.43684196e-01
-5.62149398e-02 1.57967126e+00 -1.09117973e+00 -3.45828414e-01
-4.60344672e-01 -2.31398776e-01 -1.19876504e+00 -2.30696537e-02
-3.74596328e-01 -5.43716192e-01 4.22248632e-01 1.02259707e+00
7.72577584e-01 2.82311976e-01 2.29246691e-01 -1.44434190e+00
-5.14148176e-01 5.95152915e-01 -2.28034072e-02 -1.73927009e-01
1.25145876e+00 1.28379688e-01 6.23775303e-01 -6.20103538e-01
2.49359533e-01 1.52862108e+00 7.35633492e-01 8.69799018e-01
-2.64341056e-01 -6.12518907e-01 1.19511243e-02 2.72456259e-02
-8.13394070e-01 -6.34130061e-01 6.71535671e-01 -2.92052418e-01
9.63727415e-01 3.58490497e-01 5.42897880e-01 1.07422698e+00
5.40406048e-01 5.35715282e-01 1.23046899e+00 -4.45814013e-01
6.90496445e-01 9.94898021e-01 8.16046417e-01 1.99245185e-01
-2.81534374e-01 -5.77581584e-01 -3.92617881e-01 -3.86436909e-01
1.23121291e-01 -1.73890829e-01 -3.33068192e-01 4.33657616e-01
-1.05867457e+00 1.01287889e+00 2.38078386e-01 1.57243878e-01
-1.02669704e+00 -1.97956488e-01 5.98614752e-01 3.15212518e-01
4.13585782e-01 3.55563581e-01 -4.78054851e-01 -6.68713510e-01
1.51869595e-01 -1.59830689e-01 1.33885479e+00 1.05427635e+00
4.99538571e-01 -2.16678917e-01 9.58955660e-02 1.04900420e+00
6.23900652e-01 7.34675467e-01 8.35266769e-01 -1.55699432e+00
-1.20044909e-01 5.43685079e-01 6.23687685e-01 -1.08211029e+00
-7.20407248e-01 1.21975414e-01 -3.08999747e-01 2.64488935e-01
-7.57450834e-02 -9.66182888e-01 5.66439442e-02 1.78500021e+00
4.25012827e-01 -1.14128768e+00 9.53948915e-01 9.11951542e-01
1.48420966e+00 8.18769276e-01 2.99187928e-01 -5.80520689e-01
1.87507439e+00 -9.96168077e-01 -1.65206802e+00 -5.69010615e-01
5.06882906e-01 -9.30298448e-01 1.67236245e+00 5.71504354e-01
-1.05143881e+00 -7.03508794e-01 -1.06815529e+00 1.71036422e-01
1.51003022e-02 2.86443174e-01 4.77890521e-01 8.13581347e-01
-1.21450484e+00 -1.76918939e-01 -2.26734668e-01 -1.19743848e+00
-3.84560138e-01 5.34932017e-01 -3.66277784e-01 3.71523023e-01
-1.37612772e+00 9.51928079e-01 3.47457230e-02 4.24747467e-02
-4.98144507e-01 2.08758801e-01 -7.66223133e-01 -2.69904375e-01
-3.93537339e-03 -3.82591039e-01 2.15280843e+00 -1.30491495e+00
-2.35859108e+00 7.93535948e-01 -8.61258656e-02 -2.77010173e-01
1.40347138e-01 -7.87995696e-01 -5.29288411e-01 5.47081470e-01
4.69962209e-01 8.16415966e-01 2.37859637e-01 -1.15123272e+00
-6.39290452e-01 -3.23070675e-01 6.75711811e-01 6.18258715e-01
-2.31448874e-01 4.76142019e-01 2.59708017e-01 -1.81747079e-01
-1.19013943e-01 -8.67226362e-01 -9.05649439e-02 -4.50470924e-01
-2.25893408e-01 -6.04492605e-01 6.39537632e-01 -7.28442073e-01
7.27922976e-01 -2.30419874e+00 -6.58654630e-01 -3.71940397e-02
8.24846923e-02 -3.16244572e-01 5.41706026e-01 5.19586265e-01
4.22173925e-02 -1.04157515e-01 2.13193983e-01 -7.01390803e-02
3.69915999e-02 1.72383785e-01 5.86995594e-02 2.07402334e-01
-2.94701725e-01 2.50316739e-01 -9.87987816e-01 -5.36671638e-01
-2.64830410e-01 2.33140379e-01 -3.92397702e-01 6.50557935e-01
2.24275827e-01 2.95279294e-01 -6.11501515e-01 5.99885225e-01
2.52034605e-01 6.93292022e-02 -8.88597220e-02 -9.55009554e-03
-5.55237010e-02 1.91565856e-01 -5.30118942e-01 9.64061499e-01
-6.93677366e-01 4.70740259e-01 3.65046918e-01 -3.75805795e-01
1.39407170e+00 8.01791608e-01 5.77872917e-02 -5.78848302e-01
8.08597088e-01 2.02247187e-01 -1.83345318e-01 -1.25106990e+00
6.95177376e-01 -1.47641316e-01 -5.24592340e-01 7.56662071e-01
-3.45706910e-01 -2.47085974e-01 -4.84759241e-01 3.22513252e-01
7.72432327e-01 -1.99842483e-01 8.12783182e-01 -4.20397222e-01
6.59321129e-01 1.20073721e-01 1.01748869e-01 4.45759416e-01
-8.42083216e-01 5.70081212e-02 3.61966461e-01 -8.70778225e-03
-1.81146786e-01 -2.77898490e-01 5.89268982e-01 1.53337300e+00
2.81066000e-01 -1.34585574e-01 -1.50174904e+00 -3.89127523e-01
-1.01860523e+00 9.99534607e-01 -3.67091477e-01 -2.88680881e-01
6.39151037e-03 -2.48648539e-01 3.74959141e-01 2.52350569e-01
9.49837387e-01 -1.69853961e+00 -1.13117075e+00 1.52939647e-01
-9.29935277e-01 -1.24650764e+00 -3.39991540e-01 2.93956667e-01
-7.03086913e-01 -5.54431140e-01 -7.13194609e-01 -1.27536130e+00
5.62885165e-01 3.05034161e-01 7.42703736e-01 -1.00793004e-01
5.46916187e-01 9.15420234e-01 -9.21090484e-01 -7.79615700e-01
-5.48645914e-01 -1.87599540e-01 3.89726758e-01 -2.92668343e-01
5.43436170e-01 -2.08107680e-01 -6.22732520e-01 5.64632595e-01
-2.53544390e-01 -3.83082509e-01 4.73263621e-01 3.19709599e-01
-8.11380446e-02 -1.95555121e-01 1.05237699e+00 -8.21388602e-01
1.56244099e+00 -5.81360579e-01 5.76521695e-01 -1.42063722e-01
-1.73524886e-01 -4.12199885e-01 2.73694903e-01 -5.82942128e-01
-1.55076909e+00 -1.41612872e-01 -5.95184565e-01 6.96325123e-01
-5.91480196e-01 2.72903323e-01 -4.59261507e-01 -3.61371115e-02
1.05423069e+00 -4.23673898e-01 4.65577453e-01 1.19644545e-01
7.80502483e-02 1.68639362e+00 6.85297072e-01 -4.79504794e-01
-2.45773673e-01 1.14470609e-01 -1.12012541e+00 -1.01735616e+00
-6.78366303e-01 -4.49014127e-01 1.44390345e-01 -8.55325162e-01
1.04812372e+00 -1.00847447e+00 -1.40484869e+00 6.42620862e-01
-1.33996344e+00 -5.88895321e-01 -1.04597397e-01 7.83594131e-01
-7.46001661e-01 3.70721132e-01 -7.27609634e-01 -1.33908355e+00
-8.55845571e-01 -1.09466600e+00 5.68320334e-01 7.41911590e-01
-1.23998678e+00 -6.37784660e-01 -6.69424906e-02 6.25133336e-01
5.61473727e-01 -1.11715809e-01 6.10620260e-01 -7.57979691e-01
7.95104682e-01 -4.78399068e-01 2.23552853e-01 2.70506620e-01
5.21866322e-01 -3.79529148e-01 -1.18158102e+00 2.69726902e-01
9.87671852e-01 -9.57531571e-01 -4.06850398e-01 2.44990498e-01
2.25832149e-01 -8.62580359e-01 -8.59900285e-03 -4.07751232e-01
4.88126129e-01 9.41460550e-01 8.24573338e-01 4.53158468e-01
-1.28256291e-01 1.29107594e+00 1.30696261e+00 5.06610096e-01
9.15229738e-01 1.71790928e-01 4.84295070e-01 -9.52618569e-02
5.46252787e-01 6.90781549e-02 1.14400661e+00 7.33727157e-01
-1.02713466e-01 -9.09888893e-02 -5.12064993e-01 2.09226176e-01
-1.80723214e+00 -5.80025613e-01 -4.02199298e-01 1.69596314e+00
7.96286464e-01 1.16808571e-01 3.15622091e-01 1.03945814e-01
8.69983196e-01 -3.77710044e-01 -4.78115112e-01 -1.06528366e+00
1.71480432e-01 -5.82126975e-02 1.15288906e-01 8.14856112e-01
-7.56580651e-01 9.75088656e-01 6.51613283e+00 -7.51852393e-02
-1.07103050e+00 2.21362770e-01 8.03341985e-01 2.60693043e-01
1.32840261e-01 -3.75359446e-01 -4.38669443e-01 7.02727363e-02
9.88207400e-01 -1.40263317e-02 2.34329695e-04 1.35308170e+00
5.12862563e-01 -5.77110291e-01 -7.60794818e-01 9.22928572e-01
1.65953100e-01 -1.42348126e-01 -5.59133649e-01 -5.26598752e-01
-4.02362365e-03 -4.25665140e-01 -2.13115379e-01 6.05648100e-01
3.08315545e-01 -5.43528974e-01 9.70455945e-01 5.11594355e-01
8.00496265e-02 -6.50978863e-01 1.56915474e+00 4.26488817e-01
-4.36005950e-01 4.58186790e-02 -2.09303468e-01 -7.92639673e-01
1.03485204e-01 4.46512550e-02 -8.72419357e-01 1.74181238e-02
1.26969135e+00 1.52614817e-01 -4.63244438e-01 3.81629616e-01
-4.23373133e-01 3.43404680e-01 -2.49089897e-01 -9.36693251e-01
-2.83472016e-02 1.03156768e-01 1.83828592e-01 9.64711368e-01
2.59432137e-01 6.36849821e-01 -5.96482083e-02 1.69893250e-01
3.96068394e-01 6.47031009e-01 -4.37172771e-01 3.42526466e-01
5.11324644e-01 1.44975591e+00 -6.99735463e-01 -4.22210574e-01
-9.44696590e-02 9.68802333e-01 -3.30722541e-01 3.18966717e-01
-7.22485483e-01 -6.91185534e-01 4.60104160e-02 -2.76994228e-01
-6.25281990e-01 3.20458174e-01 -2.90206641e-01 -3.48751366e-01
4.68284357e-03 -1.16796637e+00 1.21312149e-01 -1.48067498e+00
-1.04554808e+00 1.02922201e+00 -1.33201197e-01 -1.05304694e+00
-3.15916508e-01 -2.67905951e-01 -7.58565903e-01 5.17188311e-01
-8.52510989e-01 -4.60433125e-01 -8.43158960e-01 5.81393003e-01
7.06908286e-01 -6.89032525e-02 1.10287941e+00 5.36029460e-03
-3.19235474e-01 4.32357430e-01 -8.74211311e-01 -4.34301108e-01
1.11630929e+00 -8.07083249e-01 -4.73953694e-01 9.81851518e-02
-1.10274446e+00 9.03893590e-01 1.15863085e+00 -4.53829288e-01
-9.48312938e-01 -5.23356438e-01 8.94237697e-01 1.98551677e-02
2.86921442e-01 -8.99142325e-02 -6.39306486e-01 6.85276449e-01
8.98257136e-01 -7.52517104e-01 1.05217361e+00 -6.42564297e-02
4.92351562e-01 5.26348464e-02 -1.62081838e+00 4.77188557e-01
2.53156841e-01 -1.28713742e-01 -8.93580317e-01 3.68693948e-01
1.05176103e+00 -5.72486296e-02 -6.96795344e-01 -7.97640607e-02
6.30125940e-01 -1.12567484e+00 4.32334185e-01 1.34859666e-01
4.05110836e-01 -1.31769598e-01 -3.04665327e-01 -1.07807481e+00
1.15197264e-01 -9.49759364e-01 4.79436904e-01 1.41629171e+00
4.81471896e-01 -1.07892001e+00 2.82745570e-01 1.05188596e+00
-3.59949082e-01 -4.17382240e-01 -1.56091496e-01 -6.38620509e-03
-4.10525382e-01 -3.04840386e-01 2.05945551e-01 6.30187273e-01
1.22542977e+00 1.00666106e+00 -4.33894843e-01 1.32779777e-01
-1.46965459e-01 -7.14911401e-01 1.06708014e+00 -1.07594192e+00
-7.94498026e-02 -6.45228624e-02 -4.98390831e-02 -1.04947543e+00
-6.07315488e-02 9.32110939e-03 7.83034027e-01 -1.72896063e+00
6.01430750e-03 8.96079913e-02 4.65122372e-01 5.18181860e-01
-2.78161407e-01 -1.01568975e-01 -1.79752171e-01 3.34734380e-01
-6.09805465e-01 4.91436034e-01 7.67011285e-01 8.51113647e-02
-5.25325835e-01 3.06188524e-01 -1.40857279e+00 1.28313923e+00
1.30395770e+00 -1.22018114e-01 -4.76001590e-01 1.44971788e-01
3.79048675e-01 2.02589259e-01 3.63402814e-02 -9.56302464e-01
2.54479259e-01 4.95278351e-02 -1.83877096e-01 -3.14358145e-01
4.80380416e-01 -7.30914354e-01 -1.18203789e-01 5.22488534e-01
-3.69376183e-01 9.50889215e-02 5.08850627e-03 1.99702997e-02
-1.78116202e-01 -4.46568429e-01 6.86912894e-01 -2.05317363e-01
-6.42010808e-01 -7.23632693e-01 -1.07974482e+00 -3.39725882e-01
1.09993458e+00 -5.54173350e-01 -3.48625630e-01 -1.42992365e+00
-8.26572657e-01 3.70504975e-01 2.45105088e-01 2.51119554e-01
6.30453587e-01 -9.66785848e-01 -2.05895707e-01 -3.18986863e-01
1.68815717e-01 -1.10879660e-01 2.45519876e-01 1.02092862e+00
-6.46334708e-01 6.12804703e-02 -4.48768258e-01 -2.96955049e-01
-1.44740736e+00 1.04284339e-01 1.22458629e-01 3.53497446e-01
-4.08124804e-01 7.11382210e-01 3.92230093e-01 -5.83825588e-01
7.69643724e-01 -4.03823555e-01 -9.96749461e-01 1.28981560e-01
6.70961499e-01 4.36867803e-01 -2.53951222e-01 -8.46077919e-01
-4.72193778e-01 4.88622189e-01 1.63231343e-01 -3.73722762e-01
9.53330040e-01 -6.65613234e-01 -1.82617188e-01 9.68696833e-01
6.82145357e-01 3.26185018e-01 -6.24168932e-01 2.31727287e-01
-8.53076503e-02 3.01513880e-01 -3.71785223e-01 -8.79169285e-01
-5.05649328e-01 6.21214509e-01 7.27895737e-01 7.60044038e-01
1.15677536e+00 1.98078454e-01 8.64658833e-01 8.34664345e-01
6.12916231e-01 -1.41869903e+00 5.43135524e-01 5.43023765e-01
1.15140176e+00 -1.31337380e+00 -3.97709161e-01 -4.71020341e-01
-1.36016524e+00 1.08788610e+00 8.91525686e-01 2.79413909e-02
4.25956249e-01 8.83693323e-02 8.85057926e-01 -5.00330091e-01
-6.04540408e-01 2.17615038e-01 -6.19629204e-01 8.59740615e-01
6.49037361e-01 2.20973134e-01 -4.96794313e-01 1.35724723e+00
-9.57121491e-01 4.96568754e-02 1.05174768e+00 9.42041159e-01
-9.68622983e-01 -3.73614132e-01 -6.46430373e-01 -2.46083274e-01
-7.43097186e-01 3.50574225e-01 -8.57147753e-01 7.46418834e-01
-3.33507568e-01 1.91710973e+00 -4.42046911e-01 -4.99861240e-01
7.23951817e-01 -6.19719690e-03 -2.97509372e-01 -6.35655880e-01
-7.55550921e-01 -2.46704463e-02 5.76198041e-01 -3.59966487e-01
-8.06534469e-01 -3.62068176e-01 -1.64145112e+00 6.39426801e-03
-2.24883586e-01 7.64499843e-01 6.98181152e-01 9.28055584e-01
4.28811722e-02 3.63756090e-01 9.29601848e-01 -6.93421125e-01
-2.89835274e-01 -1.69814312e+00 -1.98527038e-01 1.27825454e-01
3.06518674e-01 -2.75664896e-01 -6.66836619e-01 5.04154712e-02] | [13.129096031188965, 7.665170192718506] |
19306d76-2728-4927-aaf6-ffa7f248198c | a-dual-source-approach-for-3d-human-pose | 1705.02883 | null | http://arxiv.org/abs/1705.02883v2 | http://arxiv.org/pdf/1705.02883v2.pdf | A Dual-Source Approach for 3D Human Pose Estimation from a Single Image | In this work we address the challenging problem of 3D human pose estimation
from single images. Recent approaches learn deep neural networks to regress 3D
pose directly from images. One major challenge for such methods, however, is
the collection of training data. Specifically, collecting large amounts of
training data containing unconstrained images annotated with accurate 3D poses
is infeasible. We therefore propose to use two independent training sources.
The first source consists of accurate 3D motion capture data, and the second
source consists of unconstrained images with annotated 2D poses. To integrate
both sources, we propose a dual-source approach that combines 2D pose
estimation with efficient 3D pose retrieval. To this end, we first convert the
motion capture data into a normalized 2D pose space, and separately learn a 2D
pose estimation model from the image data. During inference, we estimate the 2D
pose and efficiently retrieve the nearest 3D poses. We then jointly estimate a
mapping from the 3D pose space to the image and reconstruct the 3D pose. We
provide a comprehensive evaluation of the proposed method and experimentally
demonstrate the effectiveness of our approach, even when the skeleton
structures of the two sources differ substantially. | ['Björn Krüger', 'Hashim Yasin', 'Andreas Weber', 'Andreas Doering', 'Umar Iqbal', 'Juergen Gall'] | 2017-05-08 | null | null | null | null | ['monocular-3d-human-pose-estimation', 'pose-retrieval'] | ['computer-vision', 'computer-vision'] | [ 1.34571642e-01 -4.66895513e-02 -4.17738140e-01 -3.09120148e-01
-1.16081917e+00 -6.27566576e-01 2.74898052e-01 -3.59482080e-01
-6.52447224e-01 4.94664103e-01 2.16389984e-01 1.15094252e-01
1.25185177e-01 -5.25108635e-01 -9.88302827e-01 -3.34144592e-01
2.14970231e-01 9.31009471e-01 1.64233923e-01 1.18186593e-01
4.64322232e-02 6.92067921e-01 -1.34116340e+00 -2.83510983e-01
5.20296574e-01 1.05118978e+00 1.22722387e-01 8.56969595e-01
2.13721275e-01 1.89223588e-01 -4.91823494e-01 -2.92947218e-02
5.12778640e-01 -2.97696382e-01 -6.34896874e-01 4.46203083e-01
7.48133302e-01 -9.40505147e-01 -3.55340898e-01 9.09713268e-01
6.27172470e-01 1.36155263e-01 7.50856161e-01 -1.16799378e+00
-2.38851868e-02 -2.64430374e-01 -7.23280907e-01 -1.76310048e-01
4.99628395e-01 1.30802235e-02 7.48109758e-01 -9.83204544e-01
8.65944684e-01 1.25714350e+00 5.97304165e-01 6.96699321e-01
-1.09415126e+00 -3.55250180e-01 1.40839338e-01 -2.13322178e-01
-1.49672914e+00 -4.29212719e-01 9.92429554e-01 -6.57344341e-01
7.03364313e-01 -1.07941702e-01 8.61578763e-01 1.21884501e+00
-1.90839335e-01 1.03109467e+00 6.93511486e-01 -3.94813955e-01
1.14074193e-01 -3.48056734e-01 -3.14986855e-01 8.96233141e-01
2.96119779e-01 -7.31395185e-02 -5.27202189e-01 2.45520920e-02
1.28402531e+00 1.41841948e-01 -2.45468691e-01 -8.96688521e-01
-1.27997839e+00 5.17775714e-01 3.28819364e-01 -1.81310385e-01
-4.13132042e-01 4.16194379e-01 9.31933150e-02 -2.68840313e-01
4.70563531e-01 9.24584642e-02 -4.78694409e-01 -1.15826793e-01
-1.02470076e+00 5.95903397e-01 5.61715662e-01 9.90130007e-01
7.44150043e-01 -3.29589725e-01 9.22744721e-02 7.31248558e-01
5.53054988e-01 7.78441966e-01 2.93208301e-01 -1.27547348e+00
6.94037795e-01 5.32568574e-01 3.75777036e-01 -1.19536138e+00
-4.32067066e-01 -3.13183337e-01 -6.00925267e-01 -2.61601135e-02
6.70180678e-01 9.86994524e-03 -9.81226742e-01 1.95736468e+00
6.80884957e-01 -1.95494872e-02 -2.06284553e-01 1.29343581e+00
5.79575717e-01 3.61281246e-01 -2.22519174e-01 1.86069131e-01
1.05231071e+00 -9.62865353e-01 -4.67121989e-01 -5.79916894e-01
2.89916992e-01 -4.86130029e-01 7.18901277e-01 1.33137181e-01
-1.18876612e+00 -5.91592610e-01 -1.21664298e+00 -4.26045388e-01
3.90915833e-02 5.63152015e-01 2.16234118e-01 3.02963585e-01
-7.60664225e-01 4.46916133e-01 -1.18463600e+00 -2.88226575e-01
2.64185995e-01 4.46612269e-01 -5.51928401e-01 -7.32870400e-02
-9.89561379e-01 8.36616576e-01 2.52287775e-01 2.88544834e-01
-1.03094423e+00 -2.17626721e-01 -1.09735966e+00 -3.12703669e-01
6.76741064e-01 -1.02452004e+00 1.37320936e+00 -5.41132092e-01
-1.26971567e+00 1.17875218e+00 -2.31824547e-01 -1.65398955e-01
7.20871687e-01 -8.20444286e-01 4.28348482e-01 3.72023672e-01
2.95938551e-01 8.47974420e-01 9.55683947e-01 -1.41796279e+00
-3.57694149e-01 -7.55332649e-01 8.76815766e-02 4.34660763e-01
-1.97091654e-01 -4.65224832e-01 -1.18268490e+00 -5.47220707e-01
4.81021613e-01 -1.09811950e+00 -2.11537197e-01 4.78224516e-01
-4.69279379e-01 3.45617346e-02 4.95772868e-01 -7.59498894e-01
9.01384175e-01 -1.77976096e+00 7.07854986e-01 1.53673589e-01
3.62501979e-01 1.21664681e-01 -1.29722238e-01 -1.36425197e-02
1.82767883e-01 -2.17110291e-01 -1.60712153e-01 -7.81047523e-01
6.77416325e-02 4.12824214e-01 -3.08134686e-02 6.05101526e-01
3.30813736e-01 1.01742935e+00 -8.53470027e-01 -6.16659224e-01
2.02932939e-01 6.24219120e-01 -7.73827016e-01 5.22084713e-01
-1.79920763e-01 6.86907351e-01 -6.88359559e-01 6.81013584e-01
5.22114515e-01 -4.21861768e-01 1.26530394e-01 -5.51473618e-01
2.18811780e-01 1.05053551e-01 -1.25722420e+00 2.41094589e+00
-2.12872207e-01 3.03266317e-01 8.71960223e-02 -1.01465285e+00
7.20662534e-01 2.71306783e-01 8.88699472e-01 -1.48630038e-01
3.35309088e-01 2.56385505e-01 -6.23001397e-01 -5.30961394e-01
3.83818984e-01 -1.63161024e-01 -2.49204725e-01 5.25729954e-01
2.14006439e-01 -3.01602751e-01 -1.13107927e-01 -1.44826367e-01
8.70312870e-01 7.89201736e-01 4.72193241e-01 3.26508909e-01
4.36005473e-01 -3.38948034e-02 5.72763264e-01 3.90203059e-01
-3.18116546e-01 9.28446412e-01 4.03157711e-01 -4.89447922e-01
-1.21334708e+00 -1.37518620e+00 2.52430499e-01 5.38850427e-01
2.73432255e-01 -4.32254404e-01 -8.71652424e-01 -7.99193561e-01
-2.19614301e-02 -1.10765316e-01 -4.64292020e-01 -9.05031338e-02
-9.47846353e-01 -3.35211009e-01 5.08848965e-01 7.31917202e-01
5.50812006e-01 -3.99752021e-01 -8.69099379e-01 -1.55165195e-01
-6.30360007e-01 -1.37147856e+00 -8.13156188e-01 -8.57533887e-02
-1.01956463e+00 -1.17110527e+00 -9.83667433e-01 -7.31313229e-01
9.31018829e-01 1.79905772e-01 1.07915366e+00 1.28694758e-01
-3.04435134e-01 4.97695625e-01 -7.87767544e-02 -1.37745470e-01
-7.53715932e-02 5.57636991e-02 3.37576598e-01 -1.16443239e-01
1.98731765e-01 -5.61178029e-01 -6.70157313e-01 3.34187865e-01
-6.70536876e-01 8.23239610e-02 7.67253101e-01 7.65291274e-01
1.07089055e+00 -1.43065304e-01 1.43283784e-01 -5.03025532e-01
7.19833449e-02 -3.14260349e-02 -5.47556818e-01 4.63463031e-02
-4.19664085e-02 1.38656780e-01 1.06128663e-01 -3.91458422e-01
-7.91805029e-01 8.06626856e-01 -2.80057967e-01 -9.24331486e-01
-3.37518960e-01 4.19073254e-01 -4.38862592e-01 2.35628068e-01
4.69905794e-01 5.49842827e-02 1.64254025e-01 -7.20455110e-01
3.39644462e-01 5.38037181e-01 9.52498972e-01 -7.92585790e-01
9.62055147e-01 6.38435245e-01 1.41087890e-01 -6.53132379e-01
-1.15007353e+00 -4.65863287e-01 -1.32266378e+00 -3.28119576e-01
1.08114052e+00 -1.15014398e+00 -5.06363869e-01 5.28372169e-01
-1.31633770e+00 -2.16465369e-01 -5.52722998e-02 7.14588284e-01
-9.22477663e-01 5.17645359e-01 -5.22581398e-01 -6.97568357e-01
-1.64564714e-01 -1.28242886e+00 1.82028365e+00 -1.23373248e-01
-4.47310388e-01 -7.39788771e-01 2.35025004e-01 5.79204798e-01
-3.47231269e-01 6.03310049e-01 5.95284283e-01 -2.40020052e-01
-6.70361698e-01 -6.02314532e-01 -9.59011540e-02 2.22469568e-01
9.07353535e-02 -4.58257407e-01 -8.06315184e-01 -3.00836772e-01
9.99422092e-03 -5.87579370e-01 5.71001530e-01 5.38024485e-01
1.01590872e+00 -4.59023900e-02 -3.11387867e-01 7.39898562e-01
1.08700013e+00 -2.90170282e-01 2.61067033e-01 2.39894688e-01
9.04408634e-01 6.22945428e-01 6.97837651e-01 3.48889142e-01
4.98737782e-01 9.15459752e-01 4.17367220e-01 1.29672110e-01
-1.53638452e-01 -7.69422412e-01 2.13088214e-01 9.24360454e-01
-2.56259799e-01 -5.65376841e-02 -8.29251349e-01 3.40783030e-01
-1.89653540e+00 -5.07669270e-01 4.41598833e-01 2.31646466e+00
8.04560423e-01 2.03865826e-01 2.37124696e-01 7.81174451e-02
6.05113983e-01 7.08157644e-02 -8.90615582e-01 5.44209123e-01
2.62536138e-01 -3.98773737e-02 2.99630374e-01 4.99856889e-01
-1.22850120e+00 7.60797739e-01 6.35932016e+00 4.05162662e-01
-9.02418435e-01 -2.10124999e-01 2.01159239e-01 -5.08241773e-01
-7.94176199e-03 -1.60821691e-01 -7.95469105e-01 1.95229292e-01
4.25848901e-01 2.81754732e-01 4.71625142e-02 8.62298071e-01
9.92733762e-02 -3.74079943e-02 -1.42292559e+00 1.26616061e+00
2.45502517e-01 -1.01753342e+00 2.50536203e-01 2.51019090e-01
5.89070797e-01 -2.86823928e-01 -1.64286971e-01 5.79399057e-03
-1.42308429e-01 -9.10830617e-01 1.02102649e+00 6.69704735e-01
1.04588521e+00 -6.47652864e-01 4.06129777e-01 7.45841920e-01
-1.21933651e+00 2.66643435e-01 -1.63052171e-01 -1.07977509e-01
4.01849091e-01 3.48066449e-01 -4.28799510e-01 4.90705967e-01
6.85106993e-01 9.28420007e-01 -3.66571575e-01 8.47477615e-01
-4.13157761e-01 -9.67996046e-02 -4.61740732e-01 2.16805995e-01
-8.08650181e-02 -1.19555593e-01 6.42228484e-01 5.70805132e-01
4.32983249e-01 2.69013271e-02 4.59906787e-01 8.20520937e-01
-1.35213107e-01 -3.43449950e-01 -6.20846152e-01 3.30192521e-02
4.89107758e-01 1.04052651e+00 -7.12035239e-01 -1.94094881e-01
-3.01009983e-01 1.22816575e+00 4.33981478e-01 3.60326171e-01
-8.02667677e-01 -1.31260142e-01 7.03420818e-01 6.15778454e-02
2.71179527e-01 -8.02147031e-01 -6.03713505e-02 -1.39453673e+00
3.36809307e-01 -6.30832493e-01 2.59790212e-01 -9.11554515e-01
-1.06798398e+00 3.43619257e-01 3.49330574e-01 -1.39769292e+00
-6.70903265e-01 -8.18645954e-01 -5.98452091e-02 7.24268734e-01
-1.09784532e+00 -1.07143295e+00 -4.15093273e-01 4.75648522e-01
4.95262861e-01 2.12467626e-01 5.96032441e-01 3.72992098e-01
-4.48763520e-01 5.92976272e-01 -3.52498114e-01 4.57781971e-01
7.59026706e-01 -1.07346654e+00 4.58610058e-01 6.54766500e-01
2.59146482e-01 6.87463999e-01 3.62132549e-01 -7.15037227e-01
-1.66079664e+00 -9.31914091e-01 6.72330916e-01 -8.97914827e-01
9.95280519e-02 -5.00976682e-01 -4.27685112e-01 7.74685621e-01
-5.91508567e-01 9.95114446e-02 4.36712027e-01 -2.18966916e-01
-2.56654173e-01 7.50289187e-02 -8.53432775e-01 6.48362696e-01
1.31734884e+00 -5.40301859e-01 -7.29070783e-01 8.49173367e-02
6.06799066e-01 -8.61086190e-01 -8.10361683e-01 4.68215734e-01
7.53615022e-01 -6.82128012e-01 1.38164973e+00 -5.70144594e-01
4.20013934e-01 -4.22542244e-01 -3.87697518e-01 -1.04908991e+00
1.76460788e-01 -2.52186209e-01 -3.67066979e-01 5.88533342e-01
3.36062238e-02 4.56422567e-02 1.33606267e+00 5.85913420e-01
8.75029191e-02 -8.13148737e-01 -9.70689893e-01 -6.42039776e-01
3.16849165e-02 -5.09364426e-01 3.21036845e-01 3.70610088e-01
-4.38352793e-01 5.39215505e-01 -6.12306774e-01 1.46088764e-01
8.64965200e-01 3.75207663e-01 1.25577569e+00 -1.08799231e+00
-3.23396504e-01 -4.47210930e-02 -5.27071238e-01 -1.89335585e+00
2.77601004e-01 -6.54389083e-01 4.49876338e-01 -1.43624270e+00
3.96282941e-01 5.24046682e-02 2.23319650e-01 3.16698700e-01
-1.25595331e-01 4.00862604e-01 9.89680216e-02 4.32494730e-01
-4.91867393e-01 6.20011985e-01 1.43941665e+00 5.98909752e-03
-2.91438922e-02 -6.34856224e-02 -3.00697207e-01 1.03611743e+00
4.36584383e-01 -3.04342836e-01 -5.77884734e-01 -7.87668586e-01
8.33144560e-02 2.49523997e-01 7.17032731e-01 -9.33265209e-01
1.17793694e-01 -1.63603082e-01 8.68953049e-01 -1.08158946e+00
7.91765511e-01 -7.96876550e-01 -6.69659078e-02 3.92693132e-01
-2.81003773e-01 -2.62281075e-02 -1.34063125e-01 8.28752518e-01
-4.00198847e-02 -6.64704293e-02 5.76357365e-01 -4.52460527e-01
-5.98227262e-01 6.44920886e-01 -8.33021700e-02 2.27129415e-01
7.39597917e-01 -3.68679434e-01 3.21336299e-01 -5.50881267e-01
-8.51928473e-01 2.11609989e-01 7.15164244e-01 4.47186023e-01
8.78712356e-01 -1.69775128e+00 -4.22317564e-01 3.56826782e-01
4.06351946e-02 7.38437653e-01 4.33461219e-02 6.91709936e-01
-5.69609165e-01 4.46965843e-01 -2.60414690e-01 -9.62911427e-01
-1.05689549e+00 4.57829118e-01 3.24021876e-01 -1.76159963e-01
-5.50740361e-01 6.19867563e-01 2.63243437e-01 -7.50288606e-01
4.63451356e-01 -1.87681526e-01 4.58328985e-02 -2.71073580e-02
4.05362248e-01 2.86132544e-01 -9.92071927e-02 -1.01781690e+00
-4.00936633e-01 1.08679521e+00 1.65139839e-01 -3.52511495e-01
1.35239983e+00 -3.43274236e-01 6.35240451e-02 4.46341455e-01
1.64254642e+00 -2.60208875e-01 -1.66455805e+00 -3.20492625e-01
-1.87891334e-01 -7.73302674e-01 -2.01103836e-01 -2.87000716e-01
-1.10338187e+00 9.93123829e-01 3.51938725e-01 -6.76457226e-01
1.02291405e+00 1.70000404e-01 1.07539701e+00 5.64307809e-01
3.73876631e-01 -1.22329104e+00 6.07111335e-01 4.67191190e-01
8.03086281e-01 -1.26965618e+00 2.53423631e-01 -3.07405978e-01
-2.12908998e-01 1.01713431e+00 7.80648470e-01 -2.19420537e-01
5.05985200e-01 -3.62583548e-02 1.02077104e-01 -2.60456592e-01
-3.30759823e-01 -1.66811034e-01 6.83028698e-01 6.05169296e-01
2.42370799e-01 -1.97530687e-01 -9.45140272e-02 6.15517676e-01
-1.47544920e-01 5.44961728e-02 1.73844323e-01 1.12362087e+00
-2.89104700e-01 -1.12799847e+00 -6.06984913e-01 -7.26245623e-03
-2.59311318e-01 4.27445054e-01 -6.19136751e-01 8.16558838e-01
1.60901493e-03 4.96129513e-01 7.25843536e-04 -4.40770119e-01
4.92275894e-01 7.96063915e-02 8.51525426e-01 -5.95461309e-01
2.88202822e-01 5.24162412e-01 -1.15393393e-01 -8.34123194e-01
-5.82626939e-01 -5.84848344e-01 -1.21840262e+00 5.97887747e-02
-8.46533924e-02 -1.10927194e-01 7.03363955e-01 8.47882926e-01
2.03267992e-01 1.79264575e-01 2.88243741e-01 -1.50132430e+00
-7.48058915e-01 -6.08489513e-01 -4.12454396e-01 5.30421138e-01
5.99487424e-01 -1.12566769e+00 -1.24813043e-01 2.63324648e-01] | [6.952550411224365, -0.9700822234153748] |
55f8929c-18d5-40d1-ba36-20eec26289b2 | etrica-event-triggered-context-aware-story | 2210.12463 | null | https://arxiv.org/abs/2210.12463v1 | https://arxiv.org/pdf/2210.12463v1.pdf | EtriCA: Event-Triggered Context-Aware Story Generation Augmented by Cross Attention | One of the key challenges of automatic story generation is how to generate a long narrative that can maintain fluency, relevance, and coherence. Despite recent progress, current story generation systems still face the challenge of how to effectively capture contextual and event features, which has a profound impact on a model's generation performance. To address these challenges, we present EtriCA, a novel neural generation model, which improves the relevance and coherence of the generated stories through residually mapping context features to event sequences with a cross-attention mechanism. Such a feature capturing mechanism allows our model to better exploit the logical relatedness between events when generating stories. Extensive experiments based on both automatic and human evaluations show that our model significantly outperforms state-of-the-art baselines, demonstrating the effectiveness of our model in leveraging context and event features. | ['Zhihao Zhang', 'Frank Guerin', 'Henglin Huang', 'Chenghua Lin', 'Chen Tang'] | 2022-10-22 | null | null | null | null | ['story-generation'] | ['natural-language-processing'] | [ 2.27195442e-01 1.20398574e-01 -3.47832918e-01 -1.85932815e-01
-6.80365086e-01 -3.96668673e-01 1.20052135e+00 5.12114912e-02
7.75541551e-03 9.09749985e-01 1.26055694e+00 2.83222586e-01
1.89327210e-01 -1.00516498e+00 -5.08179903e-01 -6.34870157e-02
4.41687480e-02 1.99012548e-01 4.32407204e-03 -5.66093087e-01
4.43006486e-01 -1.50240839e-01 -1.63172269e+00 7.17862427e-01
9.88916755e-01 6.03911698e-01 2.50767022e-01 6.93807781e-01
-3.47161219e-02 1.49423194e+00 -7.53168046e-01 -4.64598417e-01
-2.87787914e-01 -1.04092324e+00 -9.19753015e-01 -1.94168329e-01
3.75509150e-02 -6.74134135e-01 -6.22182429e-01 3.57807308e-01
6.28300309e-01 3.83744299e-01 8.28765512e-01 -1.11584020e+00
-9.87729788e-01 1.27334034e+00 -1.81568325e-01 2.51436263e-01
8.49652171e-01 3.00520688e-01 1.20698690e+00 -7.82673597e-01
9.10380423e-01 1.14052749e+00 5.43288171e-01 5.61826646e-01
-1.14159381e+00 -6.53414488e-01 2.34677672e-01 5.92925668e-01
-1.10852194e+00 -6.15106225e-01 8.23374867e-01 -4.14143324e-01
1.15161157e+00 1.11717187e-01 1.04306602e+00 1.72976816e+00
3.51599395e-01 1.07161438e+00 6.59102976e-01 -3.59665900e-01
1.52551338e-01 -3.66626292e-01 -1.37734339e-01 3.66554618e-01
-1.70941025e-01 2.13196129e-01 -1.20512378e+00 1.39374867e-01
8.91117990e-01 -4.30500805e-01 -4.50985909e-01 2.89744347e-01
-1.52601099e+00 9.17597711e-01 3.34268957e-01 4.05664027e-01
-6.43830895e-01 4.03235197e-01 3.80550116e-01 -1.25889555e-01
2.72419691e-01 9.83910143e-01 2.42518008e-01 -7.31614292e-01
-1.03559327e+00 7.83305764e-01 5.25788307e-01 9.54462528e-01
6.75985888e-02 2.11236179e-01 -1.09837461e+00 7.49708891e-01
-5.82610331e-02 1.91166162e-01 7.28473067e-01 -8.09050322e-01
4.11840647e-01 5.56130171e-01 1.07525684e-01 -9.74301219e-01
-2.71189660e-01 -4.21151072e-01 -5.14233470e-01 -2.26945519e-01
-3.72829735e-02 -1.30476028e-01 -6.49495602e-01 2.20100570e+00
1.18905064e-02 4.70686585e-01 1.66063402e-02 7.15925992e-01
1.06050658e+00 7.85916507e-01 2.11345211e-01 -1.68877035e-01
1.23125529e+00 -1.11757219e+00 -9.80121017e-01 -5.07958174e-01
3.10041189e-01 -6.88933372e-01 1.18584645e+00 2.90536731e-02
-1.59308600e+00 -6.12090766e-01 -1.17746568e+00 -1.40886173e-01
6.23188466e-02 -1.28011890e-02 9.31528747e-01 9.76346806e-03
-8.59311402e-01 6.38813794e-01 -5.94433665e-01 -3.35821629e-01
4.01758105e-01 -1.77714497e-01 -2.04049408e-01 -2.31509190e-02
-1.50637722e+00 9.15214837e-01 7.09856272e-01 -1.85189947e-01
-8.90682220e-01 -8.75622451e-01 -9.43736136e-01 3.01821381e-01
2.19536155e-01 -1.11973357e+00 1.55913734e+00 -5.87845922e-01
-1.48698938e+00 3.29835534e-01 -3.43640566e-01 -3.72846931e-01
4.18404698e-01 -4.02487278e-01 -3.87297153e-01 1.32304907e-01
3.55526090e-01 7.84072578e-01 4.45489407e-01 -8.12375367e-01
-6.29604757e-01 2.33067468e-01 9.23989266e-02 5.65780461e-01
-3.84721458e-01 9.67498589e-03 -2.30047554e-01 -1.13364685e+00
-2.84844428e-01 -5.90480626e-01 -8.53136107e-02 -5.60201764e-01
-4.14206266e-01 -3.63120049e-01 3.78156066e-01 -6.68875396e-01
1.54890811e+00 -1.85615528e+00 3.94548774e-01 -4.95788753e-01
-2.35328618e-02 -1.28448084e-01 -3.86353195e-01 8.27505291e-01
1.01136811e-01 1.48651630e-01 -3.00895497e-02 -3.27790737e-01
8.51579234e-02 -3.22214633e-01 -6.94237709e-01 -3.38276118e-01
5.40520251e-01 1.34969425e+00 -1.27240682e+00 -4.77685839e-01
-3.91929783e-02 5.36969841e-01 -7.10926473e-01 4.66548026e-01
-5.59256613e-01 5.38159847e-01 -3.53784829e-01 2.74267167e-01
-1.09378420e-01 -3.40375483e-01 6.40311465e-02 5.99546768e-02
1.53720766e-01 9.31603789e-01 -8.04544330e-01 1.94093502e+00
-3.71940762e-01 6.64898694e-01 -8.51851404e-01 -3.24740201e-01
7.27680206e-01 5.99455953e-01 1.79336414e-01 -8.73472989e-01
1.06977344e-01 -1.03922263e-01 -1.94150105e-01 -3.55334669e-01
1.07677019e+00 -1.71641439e-01 -3.39086503e-01 9.89676178e-01
4.05511521e-02 -2.33941391e-01 4.59950387e-01 5.31891525e-01
1.18717611e+00 4.60154414e-01 4.06732172e-01 1.78041562e-01
1.12598473e-02 3.74836326e-02 5.04853606e-01 7.94622183e-01
2.18646135e-02 8.65797222e-01 4.88175273e-01 -1.42006740e-01
-8.66870224e-01 -1.05280054e+00 3.55061620e-01 1.10054374e+00
-1.04467325e-01 -6.85508609e-01 -6.89362526e-01 -4.04574782e-01
-4.24633503e-01 1.31339681e+00 -7.67967939e-01 -5.54391086e-01
-8.61138642e-01 -5.70281267e-01 5.34243107e-01 9.11593795e-01
4.03142661e-01 -1.61735451e+00 -6.72109008e-01 7.00309634e-01
-9.28159237e-01 -1.10973179e+00 -6.28852546e-01 -3.72885883e-01
-4.87427592e-01 -7.67575979e-01 -3.58766407e-01 -5.41345179e-01
4.07807492e-02 2.00000852e-01 1.55937219e+00 -7.65779465e-02
2.06654966e-02 1.95206590e-02 -6.09644830e-01 -3.22512090e-01
-5.09320736e-01 3.80155265e-01 -2.24617854e-01 -3.82325292e-01
6.36828095e-02 -9.32901084e-01 -5.59311211e-01 -7.83710033e-02
-9.29371238e-01 7.59237111e-01 3.51098955e-01 9.37413096e-01
3.66836429e-01 -8.76751766e-02 1.28741622e+00 -6.42430007e-01
1.07479179e+00 -6.72960579e-01 4.42303084e-02 1.87441200e-01
-3.48112434e-01 -5.17347790e-02 5.39855838e-01 -4.76531893e-01
-1.25524700e+00 -4.56528276e-01 6.00922033e-02 6.25024410e-03
1.43764257e-01 7.48957574e-01 -2.26312965e-01 7.15866804e-01
6.27771080e-01 3.90859842e-01 -4.75329489e-01 9.77597088e-02
7.64823318e-01 2.91525543e-01 9.56928074e-01 -6.98815465e-01
4.35431451e-01 1.27028823e-01 -3.25062454e-01 -2.92460918e-01
-9.75698590e-01 6.93793893e-02 -2.49108374e-01 -4.42681760e-01
7.92589784e-01 -1.12178183e+00 -2.26743981e-01 4.18311477e-01
-1.29778123e+00 -5.40853858e-01 -5.18588901e-01 3.68308544e-01
-9.15669680e-01 -5.85384071e-02 -8.30455422e-01 -5.24499118e-01
-3.88858676e-01 -7.58781552e-01 9.10519660e-01 3.79737020e-01
-9.77049470e-01 -8.41783345e-01 3.11737835e-01 4.36973900e-01
5.52293658e-01 6.33979619e-01 1.07047689e+00 -3.58030498e-01
-7.24405527e-01 -6.89059915e-03 -6.59620836e-02 -3.41978490e-01
1.13472261e-01 -1.72027022e-01 -7.78217375e-01 1.88842028e-01
-2.14633688e-01 -6.19507313e-01 9.66436207e-01 2.92639554e-01
7.59151757e-01 -3.70331496e-01 -2.39022940e-01 4.37315404e-01
9.34586763e-01 1.99728072e-01 8.45700920e-01 2.27357417e-01
6.79241657e-01 3.66940916e-01 6.07858837e-01 7.31617153e-01
8.56842756e-01 6.79147661e-01 1.74232781e-01 2.67002136e-01
-5.69552720e-01 -9.14198697e-01 4.66111064e-01 7.66842186e-01
-3.57236974e-02 -5.80322146e-01 -6.96050644e-01 9.34721231e-01
-2.16525221e+00 -1.63471675e+00 1.62794098e-01 1.69233954e+00
1.22982943e+00 1.72039255e-01 5.09492569e-02 1.23486377e-01
5.02005100e-01 6.11003458e-01 -3.55219305e-01 -4.01349664e-01
-2.28861675e-01 1.82126850e-01 -4.43628013e-01 1.39723986e-01
-9.16170537e-01 1.16323698e+00 7.21782827e+00 7.65401185e-01
-9.40705001e-01 -1.24710917e-01 5.31435430e-01 -5.80251694e-01
-5.94658434e-01 -1.84753731e-01 -5.38581133e-01 5.20339429e-01
8.81852388e-01 -6.68476343e-01 5.64421058e-01 5.97140253e-01
4.17588472e-01 4.70932834e-02 -1.43697846e+00 6.13274813e-01
3.40303659e-01 -1.79398966e+00 3.49051774e-01 -1.95589200e-01
1.15059006e+00 -4.89375919e-01 -4.41051263e-04 7.03789890e-01
6.00750148e-01 -1.34155881e+00 1.28073013e+00 5.98347962e-01
7.27056265e-01 -9.25613582e-01 4.11083877e-01 2.57560015e-01
-1.29689038e+00 8.25305134e-02 1.15366526e-01 -5.09251654e-01
7.52199709e-01 3.88016671e-01 -1.00356793e+00 4.23086822e-01
1.18966907e-01 8.29586327e-01 -4.59619939e-01 7.42103875e-01
-7.20136881e-01 6.28202796e-01 2.67503232e-01 3.15161049e-03
1.56945542e-01 2.71231204e-01 5.89848578e-01 1.26607680e+00
4.87608343e-01 3.14098835e-01 1.08193830e-01 1.25493824e+00
-2.99625278e-01 3.47072594e-02 -6.18811965e-01 -5.08438766e-01
8.50551367e-01 9.26321507e-01 -5.25179923e-01 -3.59615177e-01
-1.27189144e-01 1.03603399e+00 5.98966300e-01 1.91250116e-01
-1.07077420e+00 -2.12714389e-01 4.74147171e-01 9.64767188e-02
3.79544288e-01 -1.33734494e-01 -5.50849438e-01 -1.12113869e+00
-3.08494344e-02 -9.36319172e-01 2.63230205e-01 -9.87338722e-01
-1.17821467e+00 6.91579580e-01 -3.05691175e-02 -9.22951639e-01
-9.39363062e-01 2.60215789e-01 -1.08657491e+00 6.98549867e-01
-1.18873060e+00 -1.58723843e+00 -2.88200438e-01 2.79175758e-01
8.64120126e-01 -1.77849427e-01 7.84126997e-01 -1.18929662e-01
-3.83330166e-01 6.27181947e-01 -4.98973399e-01 -5.60914241e-02
7.44433403e-01 -1.15014589e+00 7.59203911e-01 1.07430828e+00
8.07767436e-02 4.30288523e-01 5.70806265e-01 -9.07635152e-01
-9.79490340e-01 -1.30508494e+00 1.32288659e+00 -4.59257275e-01
4.68161017e-01 -2.72868335e-01 -7.10067868e-01 8.21911454e-01
5.14595687e-01 -6.17464423e-01 8.74066353e-01 2.65133858e-01
-4.78471458e-01 4.44580138e-01 -7.22066045e-01 1.05254555e+00
1.43808186e+00 -5.64427793e-01 -8.93269658e-01 5.99211045e-02
9.78492498e-01 -3.86178464e-01 -6.78159475e-01 2.77040780e-01
6.58327401e-01 -9.53729808e-01 8.68075371e-01 -6.36652529e-01
1.41340244e+00 -9.79656577e-02 8.46087337e-02 -1.67111671e+00
-6.70154274e-01 -8.59541118e-01 -4.97714967e-01 1.54630518e+00
3.21279228e-01 -6.17010966e-02 2.87000030e-01 4.15656775e-01
-3.25403720e-01 -7.30951846e-01 -5.69103837e-01 -4.88041580e-01
6.70329705e-02 -3.89469773e-01 1.13627231e+00 9.09256339e-01
4.55986619e-01 7.97158003e-01 -5.72429121e-01 -4.51670706e-01
6.73382357e-02 3.18384916e-01 6.52768970e-01 -9.17832971e-01
-5.10128558e-01 -8.23322117e-01 -3.23024392e-02 -8.02302718e-01
3.37036699e-01 -9.26113069e-01 2.16378421e-01 -1.97185600e+00
6.00230813e-01 -1.24405026e-01 -1.07077442e-01 3.61932725e-01
-7.17284858e-01 1.85314104e-01 3.25534195e-01 -2.40905359e-02
-6.61835849e-01 9.70633388e-01 1.36005747e+00 3.35723981e-02
-3.10534567e-01 -4.87271011e-01 -1.23024213e+00 5.08745968e-01
6.68305218e-01 -2.01738209e-01 -6.46737874e-01 -7.03305721e-01
3.48267406e-01 3.50601554e-01 2.19319701e-01 -1.07087409e+00
1.76349029e-01 -4.80892390e-01 4.38141882e-01 -6.33380055e-01
4.52716619e-01 1.51690349e-01 3.27561587e-01 6.78013414e-02
-7.14602470e-01 2.11858258e-01 1.59882188e-01 5.07716715e-01
-2.66905040e-01 5.33093661e-02 4.31745023e-01 -1.41992137e-01
-6.62472367e-01 9.61531699e-02 -4.21006531e-01 3.19020361e-01
9.89243388e-01 -1.39372656e-02 -5.63143075e-01 -8.40092123e-01
-2.68072128e-01 2.25737423e-01 5.01790226e-01 8.22479784e-01
6.89609408e-01 -1.90218353e+00 -1.18480468e+00 -4.73530553e-02
2.94494987e-01 -1.20554142e-01 3.01685870e-01 2.98522204e-01
-1.12489171e-01 3.63427818e-01 -2.65939802e-01 -1.16595507e-01
-7.99414456e-01 2.09513873e-01 -2.09716074e-02 -6.76330268e-01
-6.64156199e-01 7.75559425e-01 2.13623285e-01 2.23979607e-01
-3.56790796e-02 -1.15740202e-01 -4.09730107e-01 1.07077695e-02
1.01061416e+00 2.51083851e-01 -2.16350138e-01 -5.47147334e-01
-1.22614775e-03 -2.38593412e-03 -1.78692043e-01 -5.47970235e-01
1.37307799e+00 3.59329991e-02 1.86407119e-01 4.13220435e-01
4.32855785e-01 4.81002685e-03 -1.32053518e+00 -4.85204570e-02
-1.16448641e-01 -3.64384115e-01 1.44973947e-02 -1.16946888e+00
-6.29283071e-01 6.22789025e-01 -1.87638670e-01 -8.05451125e-02
1.06864107e+00 -8.31471980e-02 1.27589107e+00 2.74582598e-02
3.30737799e-01 -9.20251667e-01 6.21446431e-01 8.24656844e-01
1.32014978e+00 -9.41053331e-01 -2.62588322e-01 -1.93931341e-01
-9.14901853e-01 8.89904439e-01 7.68465757e-01 -1.28051192e-01
2.04075556e-02 1.04332760e-01 -2.66781062e-01 5.53554930e-02
-1.23212862e+00 -9.52720940e-02 3.58553678e-01 5.50759792e-01
8.47918630e-01 2.45674133e-01 -4.39326435e-01 1.14205766e+00
-7.56962836e-01 2.68403888e-01 6.43295586e-01 8.89501393e-01
-3.07757854e-01 -1.06327927e+00 -5.61366640e-02 2.27317944e-01
-3.00888568e-01 -2.34942347e-01 -4.44396794e-01 5.85120618e-01
5.54626919e-02 1.06819069e+00 3.22332233e-02 -3.83002579e-01
2.51890481e-01 1.76222771e-01 6.69074297e-01 -7.86089718e-01
-7.86401808e-01 -7.55345225e-02 4.44459438e-01 -6.10000968e-01
-4.17649187e-02 -7.00613856e-01 -1.28541052e+00 -4.63970542e-01
-1.31818593e-01 -1.08668417e-01 2.49372259e-01 8.53137016e-01
5.75992227e-01 1.06306100e+00 4.89976168e-01 -7.67003298e-01
-3.01574349e-01 -1.14269555e+00 -2.05072537e-01 5.79425871e-01
-9.35253352e-02 -7.07176983e-01 1.36001512e-01 2.20937327e-01] | [11.644362449645996, 8.851446151733398] |
c453c668-4502-4297-accd-12a1306a8068 | controlling-hallucinations-at-word-level-in | 2102.02810 | null | https://arxiv.org/abs/2102.02810v2 | https://arxiv.org/pdf/2102.02810v2.pdf | Controlling Hallucinations at Word Level in Data-to-Text Generation | Data-to-Text Generation (DTG) is a subfield of Natural Language Generation aiming at transcribing structured data in natural language descriptions. The field has been recently boosted by the use of neural-based generators which exhibit on one side great syntactic skills without the need of hand-crafted pipelines; on the other side, the quality of the generated text reflects the quality of the training data, which in realistic settings only offer imperfectly aligned structure-text pairs. Consequently, state-of-art neural models include misleading statements - usually called hallucinations - in their outputs. The control of this phenomenon is today a major challenge for DTG, and is the problem addressed in the paper. Previous work deal with this issue at the instance level: using an alignment score for each table-reference pair. In contrast, we propose a finer-grained approach, arguing that hallucinations should rather be treated at the word level. Specifically, we propose a Multi-Branch Decoder which is able to leverage word-level labels to learn the relevant parts of each training instance. These labels are obtained following a simple and efficient scoring procedure based on co-occurrence analysis and dependency parsing. Extensive evaluations, via automated metrics and human judgment on the standard WikiBio benchmark, show the accuracy of our alignment labels and the effectiveness of the proposed Multi-Branch Decoder. Our model is able to reduce and control hallucinations, while keeping fluency and coherence in generated texts. Further experiments on a degraded version of ToTTo show that our model could be successfully used on very noisy settings. | ['Patrick Gallinari', 'Rossella Cancelliere', 'Geoffrey Scoutheeten', 'Laure Soulier', 'Marco Roberti', 'Clément Rebuffel'] | 2021-02-04 | null | null | null | null | ['table-to-text-generation', 'data-to-text-generation'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.87999541e-01 7.66486108e-01 2.10682005e-01 -2.43152410e-01
-9.21533525e-01 -4.45566833e-01 8.75862300e-01 3.26989204e-01
-3.08049530e-01 8.82518888e-01 5.26807249e-01 -1.10792421e-01
2.67053358e-02 -8.45603287e-01 -5.87308407e-01 -5.10095179e-01
2.80851781e-01 8.43718052e-01 -9.30644870e-02 -6.70063436e-01
2.09012926e-01 1.09915920e-01 -1.45616949e+00 5.61266422e-01
1.26948285e+00 6.77417338e-01 3.38863015e-01 3.38898510e-01
-3.15079361e-01 9.24028337e-01 -9.00373518e-01 -7.28340626e-01
-5.00065759e-02 -7.46719122e-01 -8.48837316e-01 2.80488342e-01
2.20888644e-01 4.27495874e-02 2.44587347e-01 1.02532899e+00
5.66548049e-01 -1.97957337e-01 6.52813613e-01 -8.66160393e-01
-4.62513447e-01 1.11766350e+00 -2.33729780e-02 -1.27317101e-01
5.75150967e-01 2.74397880e-01 1.18520904e+00 -9.65387702e-01
8.73152196e-01 1.16138256e+00 3.93408924e-01 6.33500397e-01
-1.38565934e+00 -3.03310007e-01 8.05745181e-03 2.43581966e-01
-1.11479986e+00 -5.43220699e-01 5.75185657e-01 -5.20573556e-01
1.14256418e+00 1.19366907e-01 4.16731447e-01 1.55022001e+00
6.14257008e-02 5.91233313e-01 1.28476930e+00 -6.67156041e-01
3.39162111e-01 3.31750959e-01 -1.85382828e-01 4.99204904e-01
2.93153405e-01 -1.12339199e-01 -7.14947820e-01 2.53965318e-01
2.51974702e-01 -7.38300800e-01 -4.03693289e-01 -1.02469459e-01
-1.24281216e+00 9.28064704e-01 1.99143857e-01 5.69246709e-01
-4.67279196e-01 -1.74297601e-01 4.91294026e-01 2.75242329e-01
4.72613990e-01 8.77090037e-01 -1.76388249e-01 -2.57105172e-01
-1.18416619e+00 4.70833033e-01 1.13668120e+00 1.08057988e+00
4.35019940e-01 6.47514984e-02 -6.61152184e-01 8.06750536e-01
-1.66065022e-02 2.98472822e-01 7.64831901e-01 -4.87442344e-01
9.11817491e-01 6.18703008e-01 7.93419629e-02 -9.42638695e-01
-3.84347290e-01 -6.65393353e-01 -8.90890718e-01 1.47570029e-01
5.32178938e-01 -1.65960684e-01 -7.42934585e-01 1.95922923e+00
5.50217368e-02 -3.33195478e-01 2.17358410e-01 6.58984005e-01
4.10219699e-01 5.33716679e-01 -3.48805599e-02 -2.98571795e-01
1.26073945e+00 -9.12477136e-01 -9.07204449e-01 -3.39634538e-01
8.38987052e-01 -6.35905743e-01 1.27831554e+00 6.42218888e-01
-1.24406171e+00 -4.42779630e-01 -1.11848831e+00 -3.37320901e-02
-3.93723637e-01 2.26737380e-01 6.64155483e-02 6.01206064e-01
-1.05115306e+00 7.13648915e-01 -4.82812345e-01 -3.67443591e-01
2.02209666e-01 1.49228215e-01 -3.60126674e-01 3.44331935e-02
-1.46099913e+00 1.26688230e+00 5.74382603e-01 2.05403715e-01
-6.65216208e-01 -3.29977006e-01 -8.64069521e-01 1.89686552e-01
5.16039848e-01 -7.68604279e-01 1.26743829e+00 -8.94351244e-01
-1.58509147e+00 7.05144942e-01 -5.94144277e-02 -5.99126995e-01
7.80804634e-01 -2.69119799e-01 -2.55953878e-01 1.16497828e-02
2.18944371e-01 5.45895815e-01 7.51968622e-01 -1.15034127e+00
-3.93524975e-01 -1.20063372e-01 2.64904257e-02 1.94697797e-01
-3.72397572e-01 -2.18394279e-01 3.33549990e-03 -7.38490105e-01
-2.56684870e-01 -8.31747830e-01 -2.62195915e-01 -6.25086248e-01
-7.65825927e-01 -2.23188907e-01 1.47564813e-01 -7.74390280e-01
1.36416411e+00 -1.79986167e+00 4.35832173e-01 3.66811976e-02
-3.81038487e-02 3.58488351e-01 -1.39384761e-01 7.90361941e-01
-8.54567662e-02 2.99664587e-01 -3.40232760e-01 -5.87573886e-01
2.19984785e-01 7.64374509e-02 -4.10957038e-01 1.56759692e-03
6.47217333e-01 8.52397621e-01 -9.57950413e-01 -4.18547332e-01
-1.43785281e-02 2.89732456e-01 -7.45902896e-01 4.61925566e-01
-4.58941251e-01 5.75849771e-01 -1.99198171e-01 9.32908431e-02
2.30355680e-01 -3.27418670e-02 3.39951068e-01 -2.44220849e-02
-1.11496985e-01 6.95496619e-01 -9.33798432e-01 1.79274082e+00
-9.11361158e-01 2.11966559e-01 -3.43927562e-01 -8.61624241e-01
9.58832741e-01 5.73641539e-01 -1.01894893e-01 -9.16454732e-01
7.24282339e-02 5.14521718e-01 2.27057979e-01 -5.92559695e-01
5.81661224e-01 -3.58328462e-01 -3.83116513e-01 3.05442005e-01
2.30705097e-01 -3.59774083e-01 4.69541907e-01 2.60485739e-01
1.22843015e+00 2.79727042e-01 5.75083137e-01 -2.04342857e-01
7.04929054e-01 1.39221787e-01 2.37677544e-01 6.97669506e-01
2.77574837e-01 8.66272569e-01 8.16293538e-01 1.06349215e-02
-1.10843360e+00 -6.42825127e-01 1.40825182e-01 7.00726032e-01
-3.92373323e-01 -6.12895846e-01 -1.26696479e+00 -6.87817991e-01
-4.28407341e-01 1.34250844e+00 -5.87412000e-01 -1.78253844e-01
-6.56699777e-01 -5.99684656e-01 7.26736665e-01 2.69017935e-01
9.13246498e-02 -1.48323250e+00 -7.00629413e-01 6.06175303e-01
-5.36112487e-01 -1.27443349e+00 -4.97993492e-02 3.65844041e-01
-6.69640064e-01 -7.09991455e-01 -5.37453055e-01 -4.10772085e-01
4.97647882e-01 -4.64888990e-01 1.38654149e+00 -4.82489765e-02
8.29199627e-02 -2.51179814e-01 -7.20788538e-01 -2.68079489e-01
-1.01114964e+00 4.77292418e-01 -1.68154985e-01 1.26990065e-01
6.43019453e-02 -7.74320960e-01 -1.96994334e-01 -1.40464932e-01
-1.08643103e+00 3.25856477e-01 8.97665322e-01 1.06452990e+00
2.03673303e-01 -2.60646880e-01 7.82531798e-01 -1.24030924e+00
1.04832387e+00 -4.60067034e-01 -3.99561197e-01 2.24900141e-01
-6.92358077e-01 4.80866343e-01 1.18935168e+00 -1.44560382e-01
-1.08749115e+00 -5.71958125e-02 -3.72233003e-01 3.52194160e-02
-3.55396897e-01 6.47966743e-01 -5.89450002e-01 6.48733854e-01
8.01817775e-01 2.96500325e-01 -2.65254527e-01 -4.36831295e-01
6.27170503e-01 7.05940604e-01 5.12253940e-01 -6.59903228e-01
7.74352193e-01 2.23244000e-02 -1.35729730e-01 -5.11350214e-01
-9.22706366e-01 -3.47102247e-02 -6.55012906e-01 -2.50805989e-02
7.86534131e-01 -7.52035975e-01 -1.76262215e-01 2.32955471e-01
-1.57656324e+00 -3.12655628e-01 -4.30822372e-01 3.36890891e-02
-6.95394218e-01 2.02925295e-01 -4.27367389e-01 -7.59385347e-01
-4.06582415e-01 -1.15490353e+00 1.12998986e+00 -1.33399487e-01
-5.95758259e-01 -9.61612165e-01 1.50760055e-01 3.88929963e-01
4.83709723e-01 2.45686889e-01 1.34456706e+00 -1.00346589e+00
-3.03504050e-01 -7.41861686e-02 1.58217456e-02 4.54446077e-01
-8.10294524e-02 -3.27992707e-01 -1.10891867e+00 2.04052553e-02
1.18848436e-01 -5.53063452e-01 6.21070147e-01 -1.56156525e-01
7.93165326e-01 -5.53670764e-01 1.42459095e-01 2.11575001e-01
1.30048299e+00 -1.55224964e-01 7.12576747e-01 2.11125076e-01
6.37381196e-01 1.10392058e+00 5.72339296e-01 5.15114665e-01
4.09641653e-01 1.01826811e+00 4.60832745e-01 7.53184333e-02
-4.00214553e-01 -4.91291463e-01 5.19274712e-01 1.01122069e+00
4.92802374e-02 -5.21985471e-01 -9.22016442e-01 6.17398262e-01
-1.75812519e+00 -8.66663516e-01 -3.71819794e-01 2.08091760e+00
1.10939777e+00 3.77055258e-01 -8.35924372e-02 3.87419581e-01
4.42180604e-01 1.24596674e-02 -2.60093249e-02 -6.04410827e-01
-1.62172019e-01 3.95342290e-01 1.66789498e-02 5.89097559e-01
-5.91498196e-01 9.80544448e-01 4.98459387e+00 8.65174294e-01
-9.80992854e-01 1.66156277e-01 4.47103113e-01 4.71402192e-03
-3.92614275e-01 -5.75205088e-02 -8.20149601e-01 5.34321904e-01
1.18712556e+00 -2.35493854e-01 3.61461788e-01 7.34149933e-01
5.28263390e-01 -1.20193753e-02 -1.43291616e+00 7.22725093e-01
2.43770853e-01 -1.10698366e+00 2.71531016e-01 1.61341634e-02
5.74573338e-01 -2.89949626e-01 -1.08018205e-01 4.33169782e-01
1.69285238e-01 -1.16293597e+00 1.17546928e+00 4.93614405e-01
6.74472213e-01 -6.57786250e-01 8.98615420e-01 6.48323715e-01
-6.71575129e-01 6.20387569e-02 -1.47060305e-01 -9.50785652e-02
3.70219797e-01 8.50522757e-01 -1.21318591e+00 7.47550607e-01
1.67511493e-01 2.84333050e-01 -7.60252178e-01 6.68484151e-01
-6.74584031e-01 5.72072327e-01 4.66566794e-02 -1.04152724e-01
3.14017415e-01 -1.00301728e-01 5.66089690e-01 1.55647290e+00
4.69651073e-01 -3.52844596e-01 1.71198566e-02 1.21965718e+00
-1.41303716e-02 3.22322816e-01 -6.17874384e-01 -4.91530932e-02
2.72006363e-01 1.30428624e+00 -4.42593873e-01 -3.46059710e-01
-5.16286120e-02 1.06317151e+00 6.49909735e-01 -4.20425199e-02
-6.21879816e-01 -2.88296133e-01 9.91845429e-02 2.91951478e-01
3.12426388e-01 2.76335701e-03 -4.75672573e-01 -1.24294400e+00
2.24857032e-01 -1.26724720e+00 -5.27423853e-03 -6.14323735e-01
-1.24263167e+00 1.11319840e+00 -1.29591823e-02 -1.13946211e+00
-9.07783091e-01 -4.55894411e-01 -3.62289459e-01 1.00225103e+00
-1.49735308e+00 -1.00888324e+00 -1.87945172e-01 3.50219697e-01
7.14502037e-01 -1.00246571e-01 1.09709346e+00 1.52575985e-01
-4.94881243e-01 5.29416084e-01 -3.35894674e-01 2.30546631e-02
6.01775050e-01 -1.56714869e+00 5.68661273e-01 1.03564405e+00
3.17328960e-01 5.55921197e-01 1.03270674e+00 -6.05769575e-01
-8.99475694e-01 -1.09726238e+00 1.51411915e+00 -4.10402417e-01
7.13618040e-01 -6.64425850e-01 -9.55146790e-01 2.43398458e-01
4.88633126e-01 -3.92656475e-01 4.83351082e-01 1.18049281e-02
-3.16160232e-01 1.43382385e-01 -8.90391052e-01 5.62463105e-01
9.92744923e-01 -4.60331708e-01 -8.14051092e-01 4.15253878e-01
5.41514277e-01 -3.63724709e-01 -5.40155113e-01 1.03287622e-01
1.32696584e-01 -1.19110382e+00 3.25349063e-01 -5.36089480e-01
9.50512826e-01 -1.19326234e-01 1.08302049e-01 -1.85994959e+00
-2.81379987e-02 -6.60313487e-01 1.13156036e-01 1.43349862e+00
7.34253109e-01 -2.35328212e-01 3.44424665e-01 4.11667317e-01
-3.59908998e-01 -6.42338097e-01 -9.26520407e-01 -7.93044508e-01
1.52591184e-01 -4.17245716e-01 4.07204121e-01 6.62648201e-01
3.72537524e-01 8.99716258e-01 -4.46246594e-01 -1.41438052e-01
3.18970799e-01 -1.75195068e-01 6.27672851e-01 -1.07572484e+00
-5.71061373e-01 -4.05715257e-01 -4.78291623e-02 -5.80004752e-01
2.53243238e-01 -1.03194523e+00 4.66130316e-01 -1.52421319e+00
-1.30404830e-01 -2.99718827e-01 2.00448483e-01 4.92406219e-01
-2.00878471e-01 9.18631107e-02 3.89773965e-01 3.76176983e-02
-4.19653773e-01 5.90516448e-01 1.07426035e+00 1.66130021e-01
-9.01217759e-02 -1.85434341e-01 -7.22886920e-01 5.32613933e-01
6.73528552e-01 -6.23114765e-01 -4.05048072e-01 -4.50880557e-01
5.34292400e-01 9.66722444e-02 2.32955024e-01 -1.22291386e+00
1.62602469e-01 1.81830421e-01 -6.43516779e-02 -1.26761645e-01
2.45623857e-01 -5.83934784e-01 9.48552638e-02 4.09458190e-01
-6.70597613e-01 1.53860480e-01 -1.31478012e-01 3.12872201e-01
-3.26607615e-01 -4.78612870e-01 6.21239007e-01 -2.59243608e-01
-2.37695888e-01 -2.33220860e-01 -4.09446537e-01 2.35133976e-01
7.14396179e-01 1.22624204e-01 -2.62482256e-01 -5.33039391e-01
-5.90553522e-01 -1.55908197e-01 3.38335246e-01 4.17760462e-01
3.52838337e-01 -1.07949305e+00 -1.00919378e+00 4.64013182e-02
3.80674690e-01 5.43648656e-03 -5.96873052e-02 8.79361987e-01
-2.76271015e-01 4.76943254e-01 -1.57880530e-01 -2.65540868e-01
-7.54419088e-01 3.74254197e-01 2.87194431e-01 -9.08804893e-01
-4.97056991e-01 4.99381363e-01 -1.24967825e-02 -3.17314714e-01
2.86061112e-02 -4.69599545e-01 -3.45980912e-01 3.70813787e-01
5.30236423e-01 8.06543753e-02 5.96354306e-01 -5.93002260e-01
-8.70936215e-02 -2.53931582e-02 -1.25920460e-01 -4.38849092e-01
1.29915762e+00 1.37937050e-02 9.57000256e-03 4.25013542e-01
7.40357876e-01 1.96530968e-01 -8.49609196e-01 5.71787991e-02
4.96504396e-01 -5.13810255e-02 -2.01202989e-01 -1.20262134e+00
-6.39009953e-01 1.05676389e+00 6.97282031e-02 3.87168705e-01
8.49234581e-01 -1.84307575e-01 7.30432570e-01 2.76182264e-01
3.22171390e-01 -1.06704330e+00 1.91874996e-01 6.53246105e-01
1.20127833e+00 -9.15601969e-01 -4.62151080e-01 -4.91780818e-01
-9.44907010e-01 1.11286879e+00 4.44195569e-01 1.08044716e-02
-8.18089470e-02 3.38758171e-01 1.41630948e-01 -1.18602905e-02
-1.03430355e+00 -3.59063119e-01 8.61449987e-02 5.43228209e-01
6.69537365e-01 1.07391523e-02 -5.99449337e-01 8.03063691e-01
-6.68003201e-01 -7.53045455e-02 8.04673016e-01 5.05591869e-01
-2.41001979e-01 -1.46937823e+00 -2.86721349e-01 1.52913153e-01
-3.56786400e-01 -4.43188637e-01 -5.25069177e-01 6.38062954e-01
1.83123454e-01 1.10809231e+00 -3.41022611e-01 -3.48155648e-01
6.46264017e-01 4.39205170e-01 3.11531514e-01 -1.06745553e+00
-1.02756536e+00 -2.60358807e-02 4.39244837e-01 -5.17919779e-01
-2.30007902e-01 -6.15803719e-01 -1.01299644e+00 1.07449338e-01
-2.10224912e-01 1.77036867e-01 7.22850800e-01 1.21204400e+00
2.98667878e-01 6.36035204e-01 4.82157379e-01 -8.10222328e-01
-1.04290891e+00 -1.28081036e+00 -3.55887622e-01 7.78649986e-01
-1.64912397e-03 -3.75387043e-01 -3.01976621e-01 1.05655782e-01] | [11.565491676330566, 8.983193397521973] |
680c4192-3555-4b9f-a11e-2b52c7a6b493 | tw-star-at-semeval-2017-task-4-sentiment | null | null | https://aclanthology.org/S17-2110 | https://aclanthology.org/S17-2110.pdf | Tw-StAR at SemEval-2017 Task 4: Sentiment Classification of Arabic Tweets | In this paper, we present our contribution in SemEval 2017 international workshop. We have tackled task 4 entitled {``}Sentiment analysis in Twitter{''}, specifically subtask 4A-Arabic. We propose two Arabic sentiment classification models implemented using supervised and unsupervised learning strategies. In both models, Arabic tweets were preprocessed first then various schemes of bag-of-N-grams were extracted to be used as features. The final submission was selected upon the best performance achieved by the supervised learning-based model. However, the results obtained by the unsupervised learning-based model are considered promising and evolvable if more rich lexica are adopted in further work. | ['Ismail Babaoglu', 'Mourad Gridach', 'Hatem Haddad', 'Hala Mulki'] | 2017-08-01 | null | null | null | semeval-2017-8 | ['twitter-sentiment-analysis'] | ['natural-language-processing'] | [ 1.31981105e-01 7.59209916e-02 7.93135446e-03 -8.05159986e-01
-6.73406422e-01 -7.62465954e-01 8.66035759e-01 9.63989258e-01
-9.23384428e-01 6.33213639e-01 2.64903247e-01 -2.24635780e-01
-1.89947709e-02 -7.53094256e-01 -2.34577090e-01 -4.85900491e-01
-2.85920411e-01 4.51841503e-01 1.15553603e-01 -1.10359812e+00
7.92811513e-01 2.30329499e-01 -1.76842463e+00 9.51164842e-01
7.98911810e-01 8.69690061e-01 -2.76049823e-01 6.42452717e-01
-2.90070295e-01 1.08865833e+00 -6.80626631e-01 -7.79952645e-01
3.15020978e-03 -4.90835071e-01 -1.09950602e+00 2.60267258e-01
-1.31436184e-01 2.85520881e-01 6.42513990e-01 8.99410188e-01
2.87543565e-01 2.09956601e-01 9.44410801e-01 -8.70581746e-01
-4.01110686e-02 1.07112014e+00 -3.95180881e-01 -4.98440005e-02
6.02154791e-01 -4.44027364e-01 1.11144328e+00 -1.18869436e+00
6.41610146e-01 9.93743360e-01 6.04206204e-01 2.73886085e-01
-6.09545469e-01 -3.34933788e-01 2.25936726e-01 2.69228548e-01
-8.30815375e-01 -4.03815240e-01 5.92425108e-01 -4.40433472e-01
1.05317211e+00 4.47088718e-01 6.52463436e-01 7.40314484e-01
1.93922132e-01 9.84815419e-01 1.68674457e+00 -9.90923643e-01
2.99176037e-01 6.21462822e-01 4.66451406e-01 5.36840200e-01
1.98878899e-01 -6.38742268e-01 -7.05027401e-01 -4.83612977e-02
-4.32139426e-01 -4.24356043e-01 3.58536839e-01 2.54001439e-01
-9.42321539e-01 1.12095439e+00 1.45708159e-01 5.88661194e-01
-4.90039647e-01 -5.46470940e-01 9.24607337e-01 5.81735253e-01
8.83716047e-01 5.47891498e-01 -6.66413784e-01 -1.71055615e-01
-9.23122048e-01 3.19074392e-01 9.69733834e-01 6.69430971e-01
6.92447007e-01 -1.49448171e-01 9.75666493e-02 9.48995233e-01
6.39560640e-01 3.76051396e-01 7.76688278e-01 -3.23037684e-01
3.57557088e-01 9.20907557e-01 2.15262994e-01 -1.14643061e+00
-6.84184670e-01 -6.79401085e-02 -3.89664680e-01 -8.72375257e-03
4.46480662e-01 -5.89950264e-01 -8.37984264e-01 7.51920521e-01
1.73541442e-01 -7.50624180e-01 3.91515464e-01 6.31101310e-01
1.02268910e+00 7.84656048e-01 1.14797287e-01 -4.21904385e-01
1.57740700e+00 -1.06910670e+00 -9.98139739e-01 -1.13517202e-01
1.09289706e+00 -1.40045214e+00 8.18212986e-01 8.90870094e-01
-9.57964122e-01 -2.85508484e-01 -1.22786760e+00 4.60174054e-01
-1.11296892e+00 3.42701912e-01 4.64367151e-01 1.29700625e+00
-9.56291497e-01 4.30777967e-01 -6.35839701e-01 -4.67844725e-01
4.55519706e-02 6.44161522e-01 -1.57882944e-01 1.77907273e-01
-1.30331838e+00 9.73021388e-01 4.21231925e-01 1.48293942e-01
-9.59979594e-02 2.05887824e-01 -8.54249775e-01 -5.39486229e-01
8.62117782e-02 3.64122689e-01 1.21673369e+00 -1.37911797e+00
-1.91087723e+00 1.14539015e+00 -2.20510706e-01 -4.40876424e-01
1.96555153e-01 -1.62167758e-01 -5.96273184e-01 5.61368465e-02
4.63172495e-02 1.49515390e-01 7.66105294e-01 -1.28782678e+00
-8.58126342e-01 -3.90478164e-01 1.21256821e-01 2.19477504e-01
-6.91408455e-01 7.86121547e-01 -1.49588838e-01 -7.88738668e-01
2.38620695e-02 -1.09961998e+00 -2.54853815e-01 -1.12291503e+00
-1.91556722e-01 -3.09355050e-01 4.28970218e-01 -6.23878121e-01
1.40319026e+00 -1.85127187e+00 5.58985136e-02 6.04562163e-01
-3.51980954e-01 3.52404445e-01 2.74373777e-02 8.15082490e-01
1.13964520e-01 7.23620728e-02 -5.47204055e-02 -4.05628681e-01
1.00303471e-01 5.95560223e-02 -1.27876490e-01 3.43472034e-01
2.66271412e-01 3.15964818e-01 -9.12774146e-01 -4.89684522e-01
-1.25859737e-01 1.81207493e-01 -2.70702004e-01 -3.05584073e-02
-1.66561469e-01 1.38020352e-01 -4.71276581e-01 8.28047872e-01
5.14860094e-01 1.34225652e-01 3.70243371e-01 -1.97558671e-01
-3.41327906e-01 4.00250286e-01 -9.99116719e-01 1.25081956e+00
-4.41813320e-01 3.68579835e-01 -1.62234217e-01 -8.67284656e-01
1.12964582e+00 2.08226308e-01 2.95007825e-01 -6.12487674e-01
6.21544063e-01 5.42599559e-01 3.29989903e-02 -4.50717568e-01
9.74272907e-01 -3.28395665e-02 -5.14890909e-01 5.87998569e-01
1.68525372e-02 -4.64923441e-01 8.73850405e-01 1.08519197e-01
4.33134973e-01 3.11211437e-01 3.36528122e-01 -6.15695357e-01
1.10622501e+00 5.25925398e-01 -1.46669269e-01 4.80060011e-01
-1.82088420e-01 3.65972996e-01 6.09224379e-01 -4.46868300e-01
-6.95176899e-01 -1.02415964e-01 -1.47380129e-01 1.82349980e+00
-3.79058123e-01 -9.89437103e-01 -9.81155515e-01 -1.22151732e+00
-4.43780005e-01 6.13596082e-01 -8.81164551e-01 2.72487164e-01
-5.20749748e-01 -1.32250369e+00 4.74191129e-01 3.74753028e-02
9.11337435e-02 -1.14612651e+00 -6.08824253e-01 2.52698421e-01
-9.24298987e-02 -7.79065490e-01 1.17285050e-01 6.26926243e-01
-4.98632520e-01 -7.02156901e-01 -4.96613532e-01 -7.97935307e-01
5.37476659e-01 -9.07453075e-02 9.57836151e-01 1.64594740e-01
2.55105048e-01 1.66970089e-01 -1.42232239e+00 -1.20971787e+00
-4.78378564e-01 4.82964605e-01 4.58210474e-03 2.92427182e-01
7.70756543e-01 2.95448035e-01 -2.00896427e-01 1.00216188e-01
-9.91833448e-01 -4.87173736e-01 2.03653857e-01 6.68104291e-01
2.38309354e-01 2.15807408e-02 7.00757384e-01 -1.42385101e+00
1.00026655e+00 -4.50503379e-01 -9.57717374e-02 5.93749061e-02
-6.14355147e-01 -2.50401527e-01 7.77320564e-01 1.38005480e-01
-9.96904135e-01 4.08728048e-02 -6.44461691e-01 7.82740593e-01
-8.61080289e-02 1.03977633e+00 1.22350238e-01 -2.17988677e-02
8.68979573e-01 -9.29688215e-02 7.75357767e-05 -9.85947400e-02
1.26434132e-01 1.21927404e+00 -2.84046084e-01 -3.95476311e-01
2.33938694e-01 2.90375173e-01 -2.14132860e-01 -8.83874357e-01
-1.14144409e+00 -6.23277187e-01 -7.70656943e-01 -6.00074351e-01
7.90818274e-01 -8.82937551e-01 -3.68844002e-01 8.17563891e-01
-8.36251795e-01 -9.13049802e-02 -1.58518985e-01 3.91485572e-01
-2.37410724e-01 2.80432433e-01 -5.86918056e-01 -1.08494425e+00
-4.15621728e-01 -1.14390206e+00 9.51097310e-01 1.29967153e-01
-7.24794567e-01 -1.19305027e+00 1.93827569e-01 4.98482078e-01
2.74354607e-01 1.51344910e-01 6.32275164e-01 -1.40831923e+00
6.09255612e-01 -6.94498062e-01 2.28464663e-01 4.25852180e-01
1.23140924e-01 3.63641739e-01 -1.14892995e+00 -2.41870463e-01
-1.03242971e-01 -6.89121664e-01 6.73069656e-01 -5.38428240e-02
7.90682375e-01 -3.38333905e-01 2.12688640e-01 -2.48465613e-01
1.25239396e+00 1.38576359e-01 4.81412828e-01 1.07200241e+00
1.93667844e-01 1.10034060e+00 1.36363316e+00 7.37996042e-01
5.77362716e-01 1.97595567e-01 2.89046019e-01 3.18310171e-01
5.76109648e-01 3.17314088e-01 7.82923818e-01 1.54267085e+00
-3.29272747e-01 -3.67700338e-01 -1.15391552e+00 4.65451688e-01
-1.90074456e+00 -5.52777529e-01 -6.46746933e-01 1.64315426e+00
8.36507857e-01 4.97128814e-01 5.39451301e-01 6.81954682e-01
1.52699664e-01 2.23563626e-01 4.36681122e-01 -1.13162982e+00
-4.88639116e-01 6.68918550e-01 2.56363064e-01 5.04007995e-01
-1.53413796e+00 1.04288626e+00 5.85494995e+00 7.73735523e-01
-1.04519415e+00 6.32969961e-02 5.74468136e-01 2.27465391e-01
-2.82984316e-01 -9.76797491e-02 -8.29460204e-01 3.75001401e-01
1.43113947e+00 3.86860639e-01 -2.19073340e-01 5.49408436e-01
1.27751693e-01 -4.60958481e-01 -3.97343904e-01 3.25881273e-01
5.38409829e-01 -1.08761764e+00 1.74325928e-01 -3.26087266e-01
7.92486608e-01 -7.75938630e-02 1.50526285e-01 4.29157078e-01
2.19768301e-01 -9.59706903e-01 7.80813873e-01 2.63429701e-01
1.48988292e-01 -1.07831049e+00 1.35536146e+00 2.71383286e-01
-7.28660882e-01 5.60468398e-02 -1.98526353e-01 -3.24585915e-01
-1.40430346e-01 4.50486600e-01 -5.44312716e-01 8.17653060e-01
8.67271245e-01 9.63877082e-01 -9.44550335e-01 6.25814319e-01
-1.89135954e-01 8.81799161e-01 -6.07686006e-02 -8.07512462e-01
8.20516706e-01 -4.33817416e-01 2.77771384e-01 1.94936717e+00
-1.24646295e-02 -1.02842756e-01 5.39546423e-02 -5.24157107e-01
2.57706523e-01 1.13406634e+00 -2.42819533e-01 -1.18549541e-01
-6.62220791e-02 1.50494516e+00 -1.39771342e+00 -4.18910742e-01
-3.21492195e-01 7.14590192e-01 2.97802445e-02 -6.94194511e-02
-3.90383452e-01 -7.32592821e-01 -2.25408673e-01 -9.94858518e-02
3.32655370e-01 -2.71689184e-02 -4.77493346e-01 -1.10176969e+00
-3.17294657e-01 -1.27606535e+00 6.31245375e-01 -5.29754281e-01
-1.30859864e+00 1.23576927e+00 -1.30222231e-01 -1.18818533e+00
-3.58732045e-01 -1.15948904e+00 -2.75670797e-01 6.28314316e-01
-1.44512939e+00 -1.34207046e+00 5.27568199e-02 3.54466647e-01
5.74492335e-01 -6.71849430e-01 1.15131140e+00 1.29664198e-01
-3.69449466e-01 4.73686665e-01 3.14988136e-01 -5.55246398e-02
1.17615139e+00 -1.39898527e+00 -2.30290309e-01 5.69805026e-01
4.83361594e-02 4.76182252e-01 8.82244408e-01 -4.91955876e-01
-1.21915364e+00 -8.76613617e-01 1.71566093e+00 -5.48460901e-01
9.63392496e-01 -2.82594860e-01 -5.12833714e-01 4.93379384e-01
7.80846536e-01 -6.14938140e-01 1.22662866e+00 1.18245378e-01
-3.74921486e-02 9.51638147e-02 -9.81746733e-01 3.04607958e-01
-1.21189885e-01 -2.69124538e-01 -6.23457551e-01 5.33054829e-01
1.26979157e-01 -2.57986605e-01 -9.61083949e-01 2.58992463e-01
4.85924214e-01 -9.13670778e-01 3.43104273e-01 -7.65090942e-01
6.90044701e-01 -9.51992795e-02 -3.74852002e-01 -1.43391716e+00
3.41567516e-01 -4.92507994e-01 4.08074349e-01 1.15462995e+00
1.00859678e+00 -3.74009103e-01 6.40499175e-01 -5.23762219e-03
4.36109304e-02 -7.60142565e-01 -3.93957734e-01 -1.87595993e-01
3.88461769e-01 -6.01386845e-01 3.17211509e-01 1.19176781e+00
6.09310567e-01 4.80658978e-01 -2.56359011e-01 -4.82386082e-01
6.19347468e-02 -1.01878814e-01 6.41644835e-01 -9.71361995e-01
2.10241497e-01 -4.97432679e-01 -2.43808314e-01 -4.11292851e-01
1.46013513e-01 -7.69633114e-01 2.13980563e-02 -1.29634964e+00
-2.06911296e-01 -3.38399112e-01 -2.28241131e-01 4.17265266e-01
-1.33341923e-01 7.19282746e-01 3.14510047e-01 -5.12845330e-02
-8.00586820e-01 3.04861590e-02 7.13870466e-01 -1.97313093e-02
-2.51692623e-01 2.98227310e-01 -6.93872631e-01 9.84756708e-01
9.56314862e-01 -5.00882387e-01 -1.60267558e-02 -3.27812582e-02
9.25761044e-01 -6.14646494e-01 -3.81218493e-01 -4.42539364e-01
1.13506511e-01 -1.34007767e-01 2.63748705e-01 -6.83057606e-01
2.59590387e-01 -7.68789768e-01 -5.17236412e-01 3.09498310e-01
-3.64323169e-01 3.95329326e-01 3.47165704e-01 4.23775567e-03
-5.15390933e-01 -8.15671861e-01 6.44722998e-01 -1.89293146e-01
-5.50210834e-01 -2.29584143e-01 -1.30266106e+00 -1.84461921e-01
9.40944552e-01 -6.50080070e-02 1.09503098e-01 -3.26167703e-01
-9.82060730e-01 3.41722108e-02 1.95400864e-01 3.28810930e-01
4.81124759e-01 -8.92090976e-01 -1.00056255e+00 -1.60409555e-01
4.55824971e-01 -5.74826539e-01 -9.73442271e-02 9.39103603e-01
-9.76770878e-01 3.44966024e-01 -3.15587610e-01 -1.67208999e-01
-1.35919392e+00 9.74953473e-02 -1.55918732e-01 -5.61725318e-01
3.74835730e-01 7.69107521e-01 -9.37777519e-01 -8.44953239e-01
1.05034038e-01 1.05858326e-01 -1.29964447e+00 1.11749315e+00
5.22014320e-01 4.10444289e-01 6.97745442e-01 -1.26295197e+00
-4.65449721e-01 3.44119400e-01 -3.62925917e-01 -4.50137794e-01
1.61749756e+00 -1.09116957e-01 -7.00988650e-01 8.10955346e-01
9.85228956e-01 5.02870619e-01 -2.75085509e-01 -6.67070448e-02
5.73314190e-01 -1.62449777e-01 -5.45025691e-02 -9.41556215e-01
-5.02541840e-01 6.57661319e-01 4.38505560e-01 7.41438925e-01
1.17260635e+00 -4.78809386e-01 2.82042116e-01 5.52878618e-01
1.95506856e-01 -1.56895506e+00 -5.24668396e-02 1.16067743e+00
7.36642301e-01 -1.44145989e+00 3.41410369e-01 -3.67202580e-01
-1.10103309e+00 1.53276598e+00 3.63294959e-01 -1.98622793e-01
1.14258409e+00 1.42015383e-01 4.11022425e-01 -2.81604439e-01
-4.66773868e-01 -3.55147123e-01 5.15620947e-01 3.72565836e-01
1.07164633e+00 6.04325011e-02 -1.02078342e+00 9.51279342e-01
-4.90657628e-01 -3.04368049e-01 9.06660795e-01 1.55198371e+00
-4.99430388e-01 -1.50041139e+00 -3.94348204e-01 5.10070920e-01
-9.38726425e-01 -1.53103665e-01 -9.39271867e-01 7.33920515e-01
1.08112305e-01 1.33059394e+00 -8.59987438e-02 -5.74427724e-01
4.62453932e-01 1.41288012e-01 3.65212858e-01 -7.63443649e-01
-1.56397212e+00 3.26356173e-01 6.83385730e-01 -7.39899129e-02
-1.23161387e+00 -1.03848469e+00 -1.13079560e+00 -2.38283932e-01
-3.86930078e-01 5.35602629e-01 9.48867798e-01 1.13881755e+00
-1.43038541e-01 6.90339580e-02 8.53853643e-01 -7.65837431e-01
-3.96974832e-01 -1.16494608e+00 -6.45357072e-01 3.12998146e-01
1.16762698e-01 -1.27774715e-01 -1.82772115e-01 4.13515717e-01] | [11.134836196899414, 6.914978504180908] |
d79f698a-91aa-4d01-bcd2-d53317e91064 | recognition-of-handwritten-chinese-text-by | 2207.14801 | null | https://arxiv.org/abs/2207.14801v1 | https://arxiv.org/pdf/2207.14801v1.pdf | Recognition of Handwritten Chinese Text by Segmentation: A Segment-annotation-free Approach | Online and offline handwritten Chinese text recognition (HTCR) has been studied for decades. Early methods adopted oversegmentation-based strategies but suffered from low speed, insufficient accuracy, and high cost of character segmentation annotations. Recently, segmentation-free methods based on connectionist temporal classification (CTC) and attention mechanism, have dominated the field of HCTR. However, people actually read text character by character, especially for ideograms such as Chinese. This raises the question: are segmentation-free strategies really the best solution to HCTR? To explore this issue, we propose a new segmentation-based method for recognizing handwritten Chinese text that is implemented using a simple yet efficient fully convolutional network. A novel weakly supervised learning method is proposed to enable the network to be trained using only transcript annotations; thus, the expensive character segmentation annotations required by previous segmentation-based methods can be avoided. Owing to the lack of context modeling in fully convolutional networks, we propose a contextual regularization method to integrate contextual information into the network during the training stage, which can further improve the recognition performance. Extensive experiments conducted on four widely used benchmarks, namely CASIA-HWDB, CASIA-OLHWDB, ICDAR2013, and SCUT-HCCDoc, show that our method significantly surpasses existing methods on both online and offline HCTR, and exhibits a considerably higher inference speed than CTC/attention-based approaches. | ['Jing Li', 'Shenggao Zhu', 'Hesuo Zhang', 'Canyu Xie', 'Weihong Ma', 'Lianwen Jin', 'Dezhi Peng'] | 2022-07-29 | null | null | null | null | ['handwritten-chinese-text-recognition', 'handwritten-chinese-text-recognition'] | ['computer-vision', 'natural-language-processing'] | [ 2.55153060e-01 -6.12458467e-01 -2.50464439e-01 -4.56317753e-01
-6.62776470e-01 -5.36768496e-01 4.15023655e-01 -4.12246324e-02
-6.62346542e-01 5.19300401e-01 1.96786858e-02 -5.84579170e-01
3.49280596e-01 -4.72994089e-01 -4.47641134e-01 -7.72268593e-01
6.31862700e-01 3.31979662e-01 4.36329931e-01 1.67153969e-01
5.42357564e-01 4.20945734e-01 -8.66121709e-01 4.74155903e-01
1.18183494e+00 1.17636967e+00 3.60729516e-01 5.59728265e-01
-5.79859078e-01 1.23609948e+00 -7.13268042e-01 -5.55695415e-01
-1.25310883e-01 -5.83749712e-01 -9.10027623e-01 1.78911686e-01
-3.89696681e-03 -3.61665159e-01 -5.57126641e-01 9.04000938e-01
3.61630976e-01 9.81025696e-02 4.34079677e-01 -4.81332272e-01
-8.83240581e-01 9.17436600e-01 -5.44375777e-01 6.02504313e-02
-4.57199942e-03 -2.73921248e-02 9.09962714e-01 -9.66818571e-01
4.50188369e-01 8.76609564e-01 6.72792435e-01 6.52762949e-01
-7.48221517e-01 -3.64152700e-01 5.82881093e-01 4.12045926e-01
-1.36058962e+00 -8.49428922e-02 6.73842967e-01 -9.46742371e-02
9.89819825e-01 3.97043079e-01 6.16355896e-01 1.14907324e+00
-1.26680180e-01 1.55995667e+00 1.05901325e+00 -5.91283202e-01
9.23356861e-02 -1.70601457e-01 4.05693203e-01 5.94447494e-01
-2.04298899e-01 -4.17293757e-01 -2.99432337e-01 3.32119673e-01
7.02700198e-01 5.71830384e-02 -3.42861444e-01 3.57774645e-01
-1.20626211e+00 6.07542038e-01 1.81789294e-01 5.90098381e-01
-2.43525561e-02 -6.59494549e-02 7.06714272e-01 -2.75067743e-02
4.86500710e-01 1.58754423e-01 -5.06241143e-01 -4.56442386e-01
-1.12170768e+00 -3.26167732e-01 7.53281355e-01 1.08929503e+00
4.09785956e-01 2.65972137e-01 -4.79017228e-01 1.07593393e+00
3.42769884e-02 3.24993372e-01 7.57583320e-01 -2.44590327e-01
7.72669077e-01 6.91357613e-01 -1.60523996e-01 -7.39654183e-01
-5.11574075e-02 -2.92269915e-01 -1.13304675e+00 -4.99468625e-01
4.36400682e-01 -1.40020534e-01 -1.33701074e+00 1.00471342e+00
-1.99947789e-01 -5.28740697e-02 -2.25482862e-02 9.97981429e-01
4.83688354e-01 8.98188710e-01 -2.47510374e-02 -1.49595588e-01
1.12206125e+00 -1.43800330e+00 -9.77011859e-01 -1.79658860e-01
7.39175618e-01 -8.44597936e-01 1.19243956e+00 6.33261979e-01
-7.14815199e-01 -5.29370070e-01 -1.06237304e+00 -3.15390736e-01
-4.55146313e-01 8.68398845e-01 4.73324418e-01 6.14873648e-01
-4.62269753e-01 6.21443212e-01 -1.08708644e+00 -2.13587910e-01
5.43603599e-01 5.05604185e-02 9.30564553e-02 -2.48155251e-01
-1.01513112e+00 6.99177384e-01 6.09326601e-01 7.94441581e-01
-7.69270182e-01 -8.73419940e-02 -5.05138695e-01 5.11393547e-02
6.03787720e-01 2.30435386e-01 1.09060800e+00 -1.24696755e+00
-1.97156835e+00 3.65528166e-01 -2.46865064e-01 -4.54897225e-01
9.45970774e-01 -3.86946410e-01 -4.17027980e-01 2.34579697e-01
-4.01044548e-01 4.33055133e-01 6.54615819e-01 -8.57281566e-01
-6.01098061e-01 -2.09515736e-01 -3.65988672e-01 1.38309464e-01
-5.23177564e-01 6.18430264e-02 -1.18258870e+00 -9.77307618e-01
1.73470080e-01 -7.41532326e-01 -1.92267269e-01 -2.02428192e-01
-7.17063606e-01 -3.06758225e-01 1.17420828e+00 -1.09002638e+00
1.41808021e+00 -2.18701458e+00 -3.62235047e-02 -5.12633007e-03
-2.99448729e-01 6.63460016e-01 -1.20166140e-02 3.13915282e-01
4.00086522e-01 3.06920320e-01 -4.54594344e-01 -3.62581432e-01
-1.35285974e-01 3.39045525e-01 -5.25915980e-01 3.73440564e-01
2.43742689e-01 1.01467299e+00 -6.04897738e-01 -6.82851553e-01
1.19803123e-01 2.86300987e-01 -1.70044526e-01 1.37101725e-01
-5.24739683e-01 3.09577525e-01 -6.21119857e-01 9.05394554e-01
6.32283986e-01 -2.17894182e-01 4.47054595e-01 -9.45828855e-02
-4.73716944e-01 1.80282190e-01 -7.36557305e-01 1.50388932e+00
-2.94119477e-01 7.77614295e-01 -3.85774165e-01 -1.21042645e+00
1.13263583e+00 2.78329164e-01 5.44106178e-02 -8.40761006e-01
2.40895495e-01 3.58499110e-01 -2.15480953e-01 -4.71533597e-01
7.42359579e-01 3.36583316e-01 -6.11072872e-03 1.87243894e-01
-3.56835157e-01 1.35123625e-01 1.26010418e-01 -9.29890797e-02
8.19671452e-01 5.30345559e-01 -1.77240774e-01 -1.71669144e-02
7.51820982e-01 1.62032872e-01 8.03167343e-01 7.57164955e-01
-1.80428892e-01 9.27895784e-01 4.87437874e-01 -4.62775767e-01
-9.92496789e-01 -3.62684458e-01 -1.40636235e-01 8.91001523e-01
2.40965188e-01 -3.00478876e-01 -8.76929700e-01 -8.98782909e-01
-3.14140737e-01 4.16701764e-01 -5.43270051e-01 1.55168191e-01
-9.02096510e-01 -9.51901317e-01 1.01919186e+00 9.77920532e-01
1.12396526e+00 -1.05151522e+00 -3.18215877e-01 4.70327348e-01
-3.75219434e-01 -1.47489738e+00 -7.46442974e-01 3.76837105e-01
-1.04502285e+00 -7.04119802e-01 -1.28762078e+00 -9.85841990e-01
7.91937709e-01 -4.82735671e-02 4.25889730e-01 2.17691287e-01
-2.42489070e-01 -2.11561739e-01 -8.30213547e-01 -9.98901203e-03
-1.27852887e-01 4.34995770e-01 -5.41990340e-01 2.98657745e-01
3.00968766e-01 1.95319533e-01 -4.09410626e-01 4.48395431e-01
-7.81074166e-01 1.79521099e-01 8.76342773e-01 1.15567899e+00
6.38685524e-01 -1.19875874e-02 4.79879797e-01 -1.07575238e+00
3.57821941e-01 1.10715017e-01 -7.29305387e-01 7.69590914e-01
-6.37152851e-01 -7.96913728e-02 1.09876132e+00 -6.64967179e-01
-1.47762311e+00 1.63766742e-01 -3.52407664e-01 -3.37933779e-01
-7.25916401e-02 8.76969039e-01 -2.58540899e-01 2.16134917e-02
1.88549384e-01 8.49504590e-01 -4.19159889e-01 -8.25746953e-01
9.35572609e-02 9.88099217e-01 6.54591799e-01 -6.59985065e-01
2.96947390e-01 4.17213231e-01 -6.92314148e-01 -8.73369813e-01
-7.97762394e-01 -3.56040269e-01 -7.40951836e-01 3.00136041e-02
1.07408679e+00 -7.90619731e-01 -5.89125395e-01 1.18331826e+00
-1.19661379e+00 -7.25385904e-01 1.87095314e-01 4.13632721e-01
-1.41470566e-01 9.24859583e-01 -1.12335086e+00 -8.32354903e-01
-4.27599400e-01 -1.03706813e+00 7.51777589e-01 2.97710657e-01
2.57304877e-01 -9.19586599e-01 -4.28516567e-01 4.19756830e-01
4.48349357e-01 -1.47346795e-01 9.37894046e-01 -7.52077758e-01
-7.46902704e-01 -3.33868384e-01 -5.45866549e-01 6.34431779e-01
2.74583548e-02 3.00038129e-01 -8.29830110e-01 -1.27065331e-01
-2.09581614e-01 -4.09217656e-01 1.08748603e+00 7.23459870e-02
1.81703377e+00 -3.17259371e-01 -2.57194936e-01 6.19471848e-01
1.49269176e+00 6.13435984e-01 9.82156992e-01 1.33426666e-01
9.81822789e-01 2.20591858e-01 6.61004961e-01 4.71627355e-01
3.05832326e-01 4.23207790e-01 -1.00852735e-01 -6.44842610e-02
-1.21026309e-02 -2.23768577e-01 3.80093545e-01 1.43751395e+00
-1.52610525e-01 -6.14486694e-01 -1.01785982e+00 5.02346754e-01
-1.96045029e+00 -5.42746127e-01 -1.98356271e-01 1.85362566e+00
1.04900169e+00 3.46990347e-01 -2.92095035e-01 1.89123362e-01
7.53821254e-01 1.55182824e-01 -6.61981285e-01 -3.47795397e-01
-5.19823730e-01 7.70159066e-02 5.72520614e-01 1.95570827e-01
-1.13998568e+00 1.38878942e+00 5.48471785e+00 1.23185110e+00
-1.49554658e+00 -9.14234146e-02 9.96039808e-01 6.08419895e-01
4.99708904e-03 1.51241660e-01 -1.02142656e+00 5.70528269e-01
7.18497872e-01 3.26761097e-01 2.51558036e-01 7.07919776e-01
1.52086511e-01 -1.18198358e-01 -9.03656960e-01 7.90245891e-01
2.07460240e-01 -1.40726471e+00 8.31141844e-02 -2.98020720e-01
7.95194447e-01 -9.97657254e-02 -7.91381076e-02 4.66814280e-01
6.76468238e-02 -9.01452303e-01 8.02967489e-01 4.70711976e-01
8.31274986e-01 -6.61503017e-01 9.28936601e-01 3.90362978e-01
-1.24072433e+00 -2.07778830e-02 -4.83911753e-01 2.40936354e-01
-1.18903875e-01 4.02159899e-01 -3.68139118e-01 7.77432621e-01
5.90818465e-01 1.01483834e+00 -6.50393188e-01 1.02314603e+00
-3.80609214e-01 1.08479941e+00 -2.21550558e-02 -4.02315438e-01
6.18854225e-01 -1.52289703e-01 -9.78549197e-02 1.54559863e+00
1.99652299e-01 1.23630494e-01 2.45480463e-01 7.38786042e-01
-1.26256168e-01 2.74855435e-01 1.02145761e-01 -4.21232492e-01
2.62638032e-01 1.07237828e+00 -1.18297529e+00 -5.12530684e-01
-4.11938250e-01 1.48317730e+00 2.89509237e-01 4.45238858e-01
-1.03129947e+00 -7.88214564e-01 -3.46444845e-01 -4.88275677e-01
7.30856478e-01 -4.31039363e-01 -5.11980295e-01 -1.56058002e+00
1.89874798e-01 -1.07992935e+00 2.85474122e-01 -4.92120296e-01
-1.09388733e+00 6.93616569e-01 -5.11002362e-01 -1.08569634e+00
2.73407340e-01 -6.30208790e-01 -6.52610064e-01 6.67514682e-01
-1.61936212e+00 -1.16695440e+00 -2.66656518e-01 5.09358525e-01
9.97496247e-01 -3.82317156e-02 4.71346617e-01 4.71456498e-01
-1.17822969e+00 9.01308775e-01 4.11355585e-01 8.12393188e-01
6.21628225e-01 -1.06771743e+00 4.32791859e-01 1.11082673e+00
1.35359764e-01 3.97471786e-01 2.89632706e-03 -7.63121545e-01
-1.49073541e+00 -1.24010015e+00 8.77985477e-01 -1.10226322e-03
3.94541204e-01 -5.07346570e-01 -1.23501134e+00 5.78584969e-01
1.57384545e-01 1.69444003e-03 3.14098924e-01 -2.26418346e-01
-3.35477382e-01 -1.26960844e-01 -6.35888517e-01 7.07739949e-01
7.06826210e-01 -6.30433202e-01 -4.16785508e-01 2.00207368e-01
5.35535455e-01 -5.08334696e-01 -6.36713803e-01 3.13803345e-01
5.44325650e-01 -5.62705696e-01 3.79163772e-01 -1.80383310e-01
4.43079352e-01 -1.96533680e-01 -7.48232603e-02 -7.01194584e-01
-4.38424340e-03 -5.29735804e-01 2.25730855e-02 1.42548490e+00
5.47004700e-01 -4.70783293e-01 8.62247527e-01 4.14897710e-01
-3.22913915e-01 -8.45684648e-01 -8.19346547e-01 -9.74306345e-01
2.72267878e-01 -3.87616128e-01 3.52302969e-01 9.78616118e-01
-1.00439444e-01 -2.78591383e-02 -5.28801262e-01 -5.83599396e-02
2.42715597e-01 2.70242453e-01 2.95100689e-01 -7.66995907e-01
-1.81390837e-01 -5.25941491e-01 4.59034415e-03 -1.74733102e+00
6.61012083e-02 -6.66025996e-01 4.31817412e-01 -1.35746610e+00
2.01425076e-01 -5.42023301e-01 -4.31401022e-02 7.51972854e-01
-1.63135678e-01 1.77342892e-01 1.75518975e-01 4.59860027e-01
-6.86158121e-01 6.68887496e-01 1.32361960e+00 -2.39138916e-01
-4.50005382e-02 -1.95949286e-01 -1.67942390e-01 4.92747933e-01
6.60776377e-01 -1.36028990e-01 -4.18016128e-02 -8.14839005e-01
-6.60017729e-02 5.26757054e-02 -3.00049856e-02 -8.97721291e-01
6.59104586e-01 -7.36943930e-02 4.99626160e-01 -9.07649398e-01
1.89357139e-02 -5.16296268e-01 -3.64097267e-01 4.38903421e-01
-4.61594760e-01 -1.37677073e-01 9.34006572e-02 5.88175118e-01
-5.47155559e-01 -3.02449614e-01 7.93043733e-01 -1.10417187e-01
-8.39161277e-01 2.95361102e-01 -7.35094428e-01 3.83153707e-02
6.62073493e-01 -2.70064563e-01 -3.60827088e-01 -5.37083596e-02
-4.34341013e-01 1.85990840e-01 1.45514309e-01 2.90966302e-01
7.26240337e-01 -1.02764845e+00 -5.81672430e-01 -3.59743237e-02
-3.03135514e-02 1.03950232e-01 2.97009945e-01 9.58347142e-01
-9.66018438e-01 6.69030726e-01 1.94937587e-01 -5.53306162e-01
-9.39083695e-01 3.79515946e-01 2.69985616e-01 -4.72501993e-01
-8.67440820e-01 6.15446031e-01 -1.63101301e-01 -2.13067472e-01
5.76357961e-01 -6.60766006e-01 -1.26968861e-01 8.46281415e-04
3.21530938e-01 2.14645565e-01 1.84210867e-01 -3.84551257e-01
-2.85226524e-01 6.34373009e-01 -6.03010774e-01 7.09295869e-02
1.08074415e+00 -1.15525715e-01 1.99470650e-02 4.50320214e-01
1.03882658e+00 -2.13216215e-01 -1.41264510e+00 -2.29434863e-01
4.55519557e-01 -3.61148179e-01 -7.85695314e-02 -1.07433152e+00
-1.23434317e+00 1.09646749e+00 2.72705138e-01 4.81338911e-02
1.13688493e+00 -5.05505741e-01 1.10510421e+00 7.47711003e-01
1.37896448e-01 -1.62362456e+00 2.48093724e-01 9.50300395e-01
5.41917503e-01 -1.24516296e+00 -1.40475333e-01 -3.01989228e-01
-8.60215306e-01 1.48528612e+00 7.97554433e-01 5.59249744e-02
3.91674727e-01 3.00945938e-01 1.12092242e-01 3.69486958e-01
-6.59366429e-01 -2.15177778e-02 1.14520043e-01 -2.24514343e-02
4.56844270e-01 -1.74784482e-01 -3.81943971e-01 7.77887762e-01
4.16209251e-01 6.76331595e-02 5.65071762e-01 1.17563355e+00
-1.85154423e-01 -1.16623676e+00 -1.54050970e-02 2.98652023e-01
-8.57728362e-01 -2.62392461e-01 -5.34595966e-01 6.48543954e-01
-9.55553874e-02 7.57503271e-01 2.68479809e-02 -2.18877748e-01
-4.35362533e-02 1.16923563e-01 3.37533683e-01 -3.36432010e-01
-6.67796373e-01 4.59724426e-01 -6.52381107e-02 -9.00276601e-02
-3.24192524e-01 -4.97365594e-01 -1.29317093e+00 1.81106851e-02
-8.84331942e-01 1.26036733e-01 6.08403146e-01 1.09793007e+00
1.54787764e-01 4.41268444e-01 6.93152547e-01 -3.93844932e-01
-5.64657271e-01 -9.53131020e-01 -3.43572497e-01 -1.63298089e-03
7.00235739e-02 7.54303783e-02 -1.63117737e-01 2.56114781e-01] | [11.90496826171875, 2.3856348991394043] |
b2664328-b6ea-48ac-82a0-2938ac4d32be | hakg-hierarchy-aware-knowledge-gated-network | 2204.04959 | null | https://arxiv.org/abs/2204.04959v1 | https://arxiv.org/pdf/2204.04959v1.pdf | HAKG: Hierarchy-Aware Knowledge Gated Network for Recommendation | Knowledge graph (KG) plays an increasingly important role to improve the recommendation performance and interpretability. A recent technical trend is to design end-to-end models based on information propagation schemes. However, existing propagation-based methods fail to (1) model the underlying hierarchical structures and relations, and (2) capture the high-order collaborative signals of items for learning high-quality user and item representations. In this paper, we propose a new model, called Hierarchy-Aware Knowledge Gated Network (HAKG), to tackle the aforementioned problems. Technically, we model users and items (that are captured by a user-item graph), as well as entities and relations (that are captured in a KG) in hyperbolic space, and design a hyperbolic aggregation scheme to gather relational contexts over KG. Meanwhile, we introduce a novel angle constraint to preserve characteristics of items in the embedding space. Furthermore, we propose a dual item embeddings design to represent and propagate collaborative signals and knowledge associations separately, and leverage the gated aggregation to distill discriminative information for better capturing user behavior patterns. Experimental results on three benchmark datasets show that, HAKG achieves significant improvement over the state-of-the-art methods like CKAN, Hyper-Know, and KGIN. Further analyses on the learned hyperbolic embeddings confirm that HAKG offers meaningful insights into the hierarchies of data. | ['Yunjun Gao', 'Baihua Zheng', 'Lu Chen', 'Xinjun Zhu', 'Yuntao Du'] | 2022-04-11 | null | null | null | null | ['knowledge-aware-recommendation'] | ['miscellaneous'] | [-5.72631121e-01 -3.59272398e-02 -5.77728748e-01 -2.18336895e-01
1.24303594e-01 -4.22227591e-01 2.40685463e-01 4.58558321e-01
-5.42027168e-02 3.08083981e-01 8.23236108e-01 -3.50010172e-02
-8.65991950e-01 -9.97485220e-01 -4.36317444e-01 -6.14555717e-01
-4.82951671e-01 2.11746663e-01 2.72028327e-01 -2.36125156e-01
-8.64421502e-02 5.07133231e-02 -9.62652087e-01 8.09898227e-02
9.18307722e-01 1.01824009e+00 -1.29808486e-01 2.14207903e-01
-1.27791509e-01 7.72586346e-01 2.53770091e-02 -7.05531359e-01
-1.14603490e-02 -2.91499227e-01 -6.33966088e-01 -7.64448419e-02
1.11069242e-02 -2.81827569e-01 -7.34712124e-01 8.99355590e-01
3.31934214e-01 2.43664458e-01 5.75634539e-01 -1.09012449e+00
-1.53280032e+00 1.14601099e+00 -5.24869680e-01 1.69175372e-01
1.92349523e-01 -3.22994918e-01 1.72243297e+00 -1.19935131e+00
4.63918805e-01 1.01859379e+00 7.71955431e-01 -8.38265345e-02
-1.13063312e+00 -6.59125805e-01 5.99829614e-01 5.31828642e-01
-1.77922344e+00 1.31510854e-01 9.68097210e-01 -4.35373247e-01
5.31732678e-01 1.87714666e-01 1.10877883e+00 9.61105824e-01
-3.03354144e-01 9.30274069e-01 5.01576602e-01 -4.92381565e-02
-8.48504435e-03 2.76906013e-01 6.56835198e-01 9.36295450e-01
7.26699948e-01 -1.21294692e-01 -7.60801971e-01 -2.53401846e-01
9.40518260e-01 5.54111362e-01 -5.91063142e-01 -6.14253819e-01
-1.11979365e+00 1.04979539e+00 1.02373362e+00 3.48424703e-01
-5.43242931e-01 -7.44129047e-02 1.23097613e-01 1.82063524e-02
3.97743344e-01 4.98184651e-01 -3.84701490e-01 1.86375737e-01
-4.64445889e-01 5.06750830e-02 7.30638623e-01 1.07920969e+00
8.05848300e-01 -1.75120696e-01 -3.43985498e-01 7.61395514e-01
8.28725398e-01 1.34994075e-01 4.72822517e-01 -2.05227554e-01
5.16067147e-01 1.02623498e+00 -7.10718483e-02 -1.71372688e+00
-4.96881992e-01 -1.13678968e+00 -9.14822102e-01 -1.01297414e+00
-4.77639139e-02 4.29442227e-02 -5.27493000e-01 1.43170452e+00
4.03438747e-01 6.40174031e-01 -8.54706913e-02 9.71580267e-01
9.91460025e-01 7.00600088e-01 -7.39530995e-02 -1.25804260e-01
1.28914332e+00 -1.04004014e+00 -8.06388974e-01 2.33473822e-01
7.77067184e-01 -2.58115768e-01 9.74145412e-01 2.45678991e-01
-7.05349147e-01 -5.22775769e-01 -1.04078722e+00 -4.11122143e-02
-4.66094017e-01 5.13064526e-02 9.00877297e-01 5.41601181e-01
-5.39958179e-01 3.59912038e-01 -7.16699302e-01 -1.32109731e-01
4.37370539e-01 2.82048047e-01 -2.49585882e-02 -3.32575142e-01
-1.70788467e+00 3.06897730e-01 5.94691753e-01 3.39101434e-01
-3.70386511e-01 -9.48767483e-01 -8.23998988e-01 2.49180540e-01
3.45523000e-01 -7.61961877e-01 6.02089882e-01 -2.69577831e-01
-1.16662335e+00 4.67399061e-02 1.11455873e-01 -1.74922302e-01
2.01506959e-03 -3.66864651e-01 -8.36527407e-01 -7.31232911e-02
-3.74742955e-01 -6.86263442e-02 4.52348739e-01 -1.49535871e+00
-6.16977453e-01 -4.39071417e-01 3.69663477e-01 2.91133791e-01
-1.32453108e+00 -4.28841799e-01 -8.55938137e-01 -8.55886459e-01
1.94480345e-01 -7.16011047e-01 -5.74509986e-02 -2.37531379e-01
-5.62491834e-01 -3.95279199e-01 6.56499386e-01 -6.63155258e-01
2.19745016e+00 -2.06467271e+00 4.31927532e-01 6.94719434e-01
6.90569520e-01 2.81435728e-01 7.85945356e-02 7.24317312e-01
3.70783657e-01 7.13951141e-02 1.34461284e-01 -4.12272900e-01
2.35887617e-01 5.32274723e-01 -2.51754075e-01 2.81961381e-01
-2.34471083e-01 1.24827862e+00 -1.26344967e+00 -3.13734412e-01
-1.75591949e-02 7.36506879e-01 -8.71341228e-01 1.19548924e-01
9.91348326e-02 2.05421895e-02 -7.63330281e-01 5.72365344e-01
5.78602314e-01 -6.93637192e-01 4.18163925e-01 -7.26506412e-01
2.68092036e-01 2.06743628e-01 -1.35211849e+00 1.52861917e+00
-4.62417275e-01 1.64665747e-02 -3.49191666e-01 -7.85423398e-01
8.81083667e-01 1.63236223e-02 4.10332680e-01 -4.59877133e-01
-6.34881184e-02 -2.33041830e-02 -8.04723129e-02 -4.69030291e-01
6.51907682e-01 1.99352607e-01 -3.82629298e-02 2.47087136e-01
1.37307361e-01 8.19269061e-01 -8.32491517e-02 6.76639616e-01
9.03820813e-01 -3.09714049e-01 1.43109784e-01 -1.56202510e-01
3.68702114e-01 -4.36981946e-01 7.56145120e-01 4.27007794e-01
2.16821328e-01 3.69769543e-01 3.12912285e-01 -2.79455781e-01
-4.77398992e-01 -1.11135972e+00 -2.47525144e-02 1.22852504e+00
4.87174869e-01 -1.07145870e+00 -1.14200383e-01 -9.43485260e-01
4.36797857e-01 4.69124556e-01 -9.18566227e-01 -5.44759095e-01
-3.20220053e-01 -8.89719605e-01 2.58716196e-01 9.06073749e-01
4.33060169e-01 -4.63876009e-01 3.29030991e-01 4.00089264e-01
-1.22641087e-01 -9.55991268e-01 -6.98421061e-01 -6.95798174e-02
-6.41206920e-01 -1.02579188e+00 -5.02128124e-01 -7.34898627e-01
6.61413193e-01 5.63780367e-01 9.61966574e-01 2.15781718e-01
1.87926814e-01 4.93775815e-01 -9.92565632e-01 -2.73883995e-03
5.58686554e-01 2.14571685e-01 2.27921769e-01 5.59056401e-01
4.43939000e-01 -8.56495678e-01 -1.01407003e+00 4.53406721e-01
-9.12959814e-01 -2.18686964e-02 7.70015180e-01 8.49187732e-01
6.07216299e-01 4.50585485e-01 7.43439615e-01 -1.16200781e+00
8.01429868e-01 -1.06810784e+00 -6.75938725e-02 3.81056070e-01
-1.00335491e+00 -5.78876175e-02 8.93759489e-01 -4.06080574e-01
-7.73826241e-01 -3.73111844e-01 1.16975307e-01 -6.06085896e-01
5.06854177e-01 1.22541595e+00 -2.46710688e-01 -1.73923690e-02
3.50930601e-01 3.20716381e-01 -6.54043734e-01 -7.62532055e-01
8.97513568e-01 5.83446205e-01 2.24538609e-01 -5.45318425e-01
9.35511947e-01 4.49304789e-01 -1.84371531e-01 -6.41576231e-01
-1.12806189e+00 -8.39211941e-01 -4.92712528e-01 1.41211674e-01
5.03944159e-01 -9.19483662e-01 -4.87353057e-01 5.27558327e-02
-7.03167975e-01 2.43326575e-01 -2.69411922e-01 8.07200909e-01
1.79688647e-01 3.90621066e-01 -6.90838099e-01 -6.14336789e-01
-2.93722928e-01 -5.20387113e-01 6.39424920e-01 3.41595739e-01
3.96610983e-02 -1.50359094e+00 1.66501805e-01 2.72334427e-01
3.89839232e-01 -8.04220214e-02 1.08259380e+00 -7.61364758e-01
-6.54912412e-01 -1.99947417e-01 -4.99254346e-01 2.54313082e-01
3.11402082e-01 -2.63688356e-01 -4.76723939e-01 -2.86215246e-01
-6.74107432e-01 -1.15488261e-01 9.98555601e-01 -6.28664494e-02
1.14186430e+00 -6.35968924e-01 -4.74352121e-01 7.72439599e-01
1.29241455e+00 -3.05421591e-01 2.89261580e-01 -9.45193917e-02
1.41061199e+00 4.27931428e-01 2.32384786e-01 6.72518015e-01
1.05748010e+00 8.21179152e-01 3.13224465e-01 1.35103390e-01
-5.13116047e-02 -1.02557480e+00 1.67016476e-01 1.68919837e+00
-3.23572725e-01 -2.28344396e-01 -5.83167970e-01 6.50211573e-01
-2.17433190e+00 -7.49705195e-01 -4.43635613e-01 1.96268463e+00
8.37678552e-01 -2.51024924e-02 3.36912870e-01 5.04840910e-02
3.84310395e-01 2.12940738e-01 -3.57783884e-01 2.41759911e-01
-1.95270255e-02 -3.20332460e-02 5.10169923e-01 3.77277762e-01
-9.19527709e-01 7.63797939e-01 5.07979155e+00 8.67274582e-01
-7.03680456e-01 1.21116288e-01 -8.93832594e-02 8.72644931e-02
-7.97492206e-01 -1.21770971e-01 -1.03630781e+00 6.31660104e-01
5.29981077e-01 -3.73170167e-01 2.65499175e-01 7.06228137e-01
-7.22279446e-03 6.79547966e-01 -9.18217719e-01 9.31801498e-01
2.13238120e-01 -1.26026464e+00 3.64073128e-01 1.62831217e-01
9.47483957e-01 -3.70369881e-01 1.98330045e-01 6.46171868e-01
5.83741903e-01 -7.42115021e-01 3.11703086e-01 7.44399786e-01
2.64765620e-01 -8.58359933e-01 7.44993865e-01 2.43726388e-01
-1.80455709e+00 -2.62594938e-01 -6.12803400e-01 1.05348215e-01
1.56515986e-01 7.64970422e-01 -6.01957083e-01 1.30852258e+00
6.15376055e-01 1.37983346e+00 -6.31502807e-01 1.22246802e+00
-4.79658693e-01 1.05419290e+00 -1.07466601e-01 -2.41614506e-01
3.30861330e-01 -2.37113282e-01 2.74983734e-01 1.36026931e+00
3.20017606e-01 2.46105358e-01 4.01180834e-01 7.36505568e-01
-3.74060243e-01 4.18690681e-01 -2.38754779e-01 -2.23152086e-01
7.29263484e-01 1.35435891e+00 -3.71089965e-01 -1.44782707e-01
-7.60024488e-01 8.92576754e-01 5.99441528e-01 5.93709528e-01
-7.62887657e-01 -4.13843006e-01 7.12500691e-01 2.00656608e-01
6.75105274e-01 -3.41887057e-01 1.20597202e-02 -1.32583904e+00
6.26964793e-02 -3.27634543e-01 7.55336285e-01 -3.57632875e-01
-1.69987774e+00 2.31813729e-01 -7.99127072e-02 -1.16908956e+00
3.05860937e-01 -4.07949686e-01 -4.48425114e-01 6.99865758e-01
-1.63375187e+00 -1.47575867e+00 -4.96690750e-01 8.07884514e-01
-1.07553480e-02 5.38078249e-02 6.84566975e-01 6.95992351e-01
-7.40474105e-01 9.56110239e-01 4.26497906e-01 2.73053944e-01
2.81177282e-01 -1.26686764e+00 -1.46195158e-01 4.65488613e-01
6.65025592e-01 1.25535667e+00 1.50911927e-01 -6.17541134e-01
-1.68885660e+00 -1.48977947e+00 6.82354093e-01 -4.32209015e-01
8.38505685e-01 -3.52486789e-01 -1.15698028e+00 8.30108762e-01
-1.11475073e-01 3.27678144e-01 1.30462360e+00 8.77022326e-01
-7.48020649e-01 -3.09009671e-01 -5.86225510e-01 3.56164187e-01
1.49804699e+00 -5.58550775e-01 -6.77102506e-01 2.42081136e-01
9.47288394e-01 -8.57444666e-03 -1.22596037e+00 3.27221096e-01
7.23991036e-01 -6.48604214e-01 1.10300922e+00 -9.15676892e-01
2.84617871e-01 -5.24688244e-01 -1.57121420e-01 -1.69272280e+00
-1.06679332e+00 -4.43390578e-01 -1.04534519e+00 1.36528897e+00
4.30570036e-01 -6.33622885e-01 7.69697368e-01 2.15842947e-01
-1.34621739e-01 -1.29378831e+00 -3.49055171e-01 -8.37644339e-01
-3.30715060e-01 -1.59873649e-01 7.51646578e-01 1.37026441e+00
3.28046679e-01 4.97821033e-01 -6.86913311e-01 4.78180826e-01
5.87007225e-01 3.41422319e-01 5.48510492e-01 -1.39169884e+00
-5.91225684e-01 -2.81285256e-01 -6.32737458e-01 -1.59449339e+00
-2.38223836e-01 -1.21262467e+00 -6.22811496e-01 -1.87441623e+00
2.60510117e-01 -7.71187723e-01 -1.02684784e+00 3.10558945e-01
-4.71790880e-01 1.05835915e-01 -7.26242810e-02 4.38765079e-01
-1.01152205e+00 8.99950564e-01 1.20850372e+00 -5.53906336e-03
-2.87722021e-01 -1.67710260e-01 -1.20207155e+00 6.12167120e-01
3.88710380e-01 -2.44239926e-01 -9.21786904e-01 -6.25768542e-01
7.82817841e-01 -4.85901147e-01 1.85962722e-01 -5.86494565e-01
6.10230267e-01 -1.24922609e-02 4.34295356e-01 -5.14167666e-01
2.67931223e-01 -9.91772056e-01 1.56051025e-01 -4.06959988e-02
-2.03924865e-01 -2.24125311e-01 -3.74953866e-01 1.17205501e+00
-2.50889689e-01 1.67442799e-01 1.29292682e-01 3.25441688e-01
-7.50762463e-01 8.65864456e-01 4.08521414e-01 8.11410248e-02
8.33508551e-01 3.57682630e-02 -3.17976087e-01 -3.32392722e-01
-7.83781826e-01 6.29485309e-01 8.51799473e-02 6.44747436e-01
8.28897834e-01 -1.97708404e+00 -4.40579593e-01 1.35094777e-01
6.08719707e-01 -2.82717887e-02 3.98869693e-01 9.11077440e-01
-5.33001870e-02 4.35602248e-01 3.12258095e-01 -1.51683196e-01
-8.87750208e-01 5.39287090e-01 -9.40330923e-02 -6.05301142e-01
-6.61973953e-01 1.18527019e+00 2.33242571e-01 -3.20269585e-01
2.20058128e-01 -2.96426326e-01 -6.12918198e-01 3.81255388e-01
5.54162681e-01 4.18800801e-01 -9.85420048e-02 -5.24712920e-01
-2.56965816e-01 5.50561786e-01 -4.26853091e-01 4.23627228e-01
1.38534069e+00 -2.31689140e-01 1.53284416e-01 4.57652032e-01
1.30953312e+00 3.80297422e-01 -8.29966009e-01 -9.43485737e-01
-3.48187201e-02 -7.02731609e-01 2.15352282e-01 -5.57910085e-01
-1.25337625e+00 6.98083043e-01 2.90044229e-02 6.07563734e-01
8.99918616e-01 7.99416825e-02 1.08948278e+00 3.38923395e-01
4.17882919e-01 -8.80164862e-01 1.95515886e-01 3.62903148e-01
6.46672487e-01 -6.50553524e-01 1.48757905e-01 -6.56384170e-01
-6.34448290e-01 8.31262112e-01 5.79432249e-01 -7.56746605e-02
1.15044177e+00 -3.78770441e-01 -4.25132632e-01 -4.52244729e-01
-5.34085870e-01 -3.67876887e-01 7.27972269e-01 4.98408943e-01
3.42021227e-01 2.96426713e-01 -2.91665018e-01 1.45502400e+00
-5.66916391e-02 -3.93638015e-02 -2.60139592e-02 6.75635397e-01
-4.42380607e-01 -1.00407386e+00 2.14926898e-01 6.66633546e-01
-6.33636937e-02 -2.40628883e-01 -4.14499164e-01 5.37137210e-01
3.65672708e-01 6.95452869e-01 -3.52518916e-01 -1.02474904e+00
5.34135699e-01 -2.06831306e-01 1.67777017e-01 -7.48838544e-01
-3.22265804e-01 -1.62642509e-01 9.63846128e-03 -3.37998778e-01
-1.08769767e-01 -2.65008986e-01 -1.15905333e+00 -2.68912166e-01
-5.50249338e-01 5.05897522e-01 1.39601439e-01 6.29020035e-01
6.68787241e-01 7.19902933e-01 8.48114610e-01 -4.28405911e-01
-4.67364490e-01 -8.88980746e-01 -1.15471315e+00 6.05540991e-01
2.56134421e-02 -1.01348495e+00 -3.16837400e-01 -2.83002675e-01] | [10.263258934020996, 5.671534061431885] |
1e34d3a8-33e6-4cfa-9aa2-54d6dbf14bf6 | anomaly-detection-in-video-sequence-with | 1908.06351 | null | https://arxiv.org/abs/1908.06351v1 | https://arxiv.org/pdf/1908.06351v1.pdf | Anomaly Detection in Video Sequence with Appearance-Motion Correspondence | Anomaly detection in surveillance videos is currently a challenge because of the diversity of possible events. We propose a deep convolutional neural network (CNN) that addresses this problem by learning a correspondence between common object appearances (e.g. pedestrian, background, tree, etc.) and their associated motions. Our model is designed as a combination of a reconstruction network and an image translation model that share the same encoder. The former sub-network determines the most significant structures that appear in video frames and the latter one attempts to associate motion templates to such structures. The training stage is performed using only videos of normal events and the model is then capable to estimate frame-level scores for an unknown input. The experiments on 6 benchmark datasets demonstrate the competitive performance of the proposed approach with respect to state-of-the-art methods. | ['Trong Nguyen Nguyen', 'Jean Meunier'] | 2019-08-17 | null | null | null | null | ['anomaly-detection-in-surveillance-videos', 'anomaly-detection-in-surveillance-videos'] | ['computer-vision', 'methodology'] | [ 3.07207763e-01 -2.29866073e-01 2.36715630e-01 -3.11144143e-01
-2.22451523e-01 -2.94673890e-01 7.83568025e-01 -1.09829932e-01
-4.40433443e-01 3.86737436e-01 -1.01664849e-01 -1.31907240e-01
2.62635678e-01 -5.38411021e-01 -1.19124126e+00 -7.61104882e-01
-1.81067988e-01 4.74671014e-02 9.10985589e-01 -5.58873080e-03
1.18185468e-01 5.98576784e-01 -1.59968448e+00 4.73492622e-01
3.69522363e-01 1.15415204e+00 1.05455834e-02 8.34235191e-01
6.14944845e-02 1.23386681e+00 -5.56325674e-01 -6.12844348e-01
4.00844187e-01 -4.16706443e-01 -5.00149250e-01 5.73009431e-01
9.16248500e-01 -6.71472967e-01 -6.23441458e-01 1.35008383e+00
-2.41334196e-02 8.29071030e-02 6.40143692e-01 -1.21600425e+00
-1.01869881e-01 -9.93342996e-02 -5.14788151e-01 7.82274306e-01
3.57855171e-01 1.24211572e-01 5.99570453e-01 -9.14258718e-01
7.21309543e-01 1.30748963e+00 4.40315902e-01 4.80376363e-01
-8.83607626e-01 -2.19985247e-01 3.27227324e-01 6.59446180e-01
-1.10829079e+00 -3.85663182e-01 7.09616065e-01 -6.27153158e-01
7.27181137e-01 5.84883019e-02 6.70707047e-01 1.45646751e+00
4.31313246e-01 6.89163089e-01 2.40584195e-01 -2.90636480e-01
2.66614288e-01 -2.18229160e-01 7.40892589e-02 8.41927171e-01
4.40886885e-01 5.14440238e-02 -1.42557532e-01 -9.33834165e-02
7.83325195e-01 3.56881142e-01 -3.26774836e-01 -5.26180863e-01
-8.76256764e-01 4.93319154e-01 2.41356134e-01 2.67456204e-01
-6.08182073e-01 2.27347076e-01 4.83392656e-01 1.76618546e-01
2.88389958e-02 -1.38531685e-01 -3.90245020e-01 1.56002894e-01
-9.14205194e-01 2.65330642e-01 6.20855927e-01 7.80868828e-01
5.14406145e-01 9.37123075e-02 -2.55961090e-01 3.97698432e-01
3.58113021e-01 2.38502458e-01 2.99740702e-01 -8.09477627e-01
4.01268870e-01 6.76590443e-01 2.80022770e-01 -1.14425015e+00
-1.71961844e-01 -4.30004895e-01 -8.36109221e-01 4.28056091e-01
6.42742395e-01 -8.23224634e-02 -9.07530725e-01 1.68608975e+00
4.58691090e-01 7.52376556e-01 5.08051887e-02 7.83745944e-01
6.84541345e-01 9.09952402e-01 1.02342457e-01 -2.10342154e-01
1.34697723e+00 -1.01107311e+00 -6.33184552e-01 -3.46641123e-01
9.09570232e-02 -7.75311887e-01 2.64884114e-01 3.48607183e-01
-1.13243818e+00 -9.69822168e-01 -8.35945427e-01 2.73460805e-01
-2.05694586e-01 3.42955291e-01 1.02357171e-01 3.45454156e-01
-9.14726853e-01 6.63068354e-01 -1.02768028e+00 -6.30067587e-01
2.55053043e-01 2.70004541e-01 -4.75708008e-01 -1.15033351e-01
-7.03682780e-01 7.48957634e-01 5.16207635e-01 4.29587632e-01
-1.26395810e+00 -3.69668715e-02 -9.88980830e-01 4.56442013e-02
3.73379052e-01 -6.61908031e-01 1.08995497e+00 -1.60185421e+00
-1.17378998e+00 7.53953576e-01 -2.60804564e-01 -5.32664776e-01
8.09465349e-01 -4.17719811e-01 -5.95817447e-01 3.40624779e-01
-8.72483477e-02 4.57063109e-01 1.24951565e+00 -1.13100386e+00
-1.24403191e+00 -1.38528064e-01 -6.11908883e-02 6.41465485e-02
-1.67938530e-01 3.95095736e-01 -7.24148393e-01 -6.48917139e-01
2.15667803e-02 -7.13641107e-01 -1.75771281e-01 2.85826623e-01
-2.79619813e-01 -5.83241880e-02 1.15312505e+00 -8.35105479e-01
1.03089952e+00 -2.07935596e+00 2.70495504e-01 1.19411677e-01
-1.50891012e-02 5.31111598e-01 -1.64383486e-01 1.32922575e-01
-3.90947968e-01 -4.20139790e-01 -1.17369331e-01 -2.71767437e-01
-4.47848231e-01 1.72782525e-01 -6.48050383e-02 6.59559548e-01
5.70597112e-01 5.40712953e-01 -8.62166047e-01 -4.50054228e-01
4.40967500e-01 2.44766489e-01 -3.30694437e-01 7.06367671e-01
-2.77198851e-01 4.55705822e-01 -2.42092162e-01 7.45805025e-01
7.36383200e-01 -1.05704531e-01 6.02814183e-02 -3.35918963e-01
-1.87703192e-01 -1.45321131e-01 -1.31209958e+00 1.34286797e+00
2.46854246e-01 6.51195705e-01 -4.39410023e-02 -1.06480765e+00
7.24915564e-01 3.80947620e-01 4.18599367e-01 -1.86498478e-01
2.99191713e-01 7.78819397e-02 3.35541636e-01 -9.27419484e-01
2.45078489e-01 4.47109878e-01 3.86171311e-01 -1.11011624e-01
3.06526423e-01 6.55140281e-01 4.02019799e-01 -1.55280679e-01
1.34952950e+00 1.81098059e-01 4.77340698e-01 -3.34700681e-02
1.07822418e+00 -2.69534588e-01 8.11053455e-01 6.39301002e-01
-4.08199251e-01 5.01949906e-01 6.08463824e-01 -1.16485989e+00
-1.11560762e+00 -1.10758483e+00 1.27519324e-01 6.76261127e-01
2.52153903e-01 -8.82931799e-02 -7.00121403e-01 -8.57258201e-01
-3.06601942e-01 3.64326954e-01 -7.82977283e-01 -1.46031231e-01
-9.26066458e-01 -5.73287368e-01 1.22806758e-01 5.38971722e-01
5.60022116e-01 -1.15959120e+00 -1.04762244e+00 2.57831037e-01
-2.64787167e-01 -1.47395563e+00 -3.88825387e-01 -8.08879584e-02
-9.37708080e-01 -1.34451151e+00 -5.83894908e-01 -1.02946949e+00
8.97490621e-01 2.52212405e-01 1.13554740e+00 1.66971534e-01
-3.26502830e-01 2.70857155e-01 -4.17944252e-01 -2.20939547e-01
-5.49838364e-01 -4.37295824e-01 -9.97828245e-02 7.94867337e-01
3.51126909e-01 -3.59668702e-01 -8.27563941e-01 3.61300766e-01
-1.17409837e+00 -2.04033524e-01 7.46088445e-01 8.17516327e-01
3.37900341e-01 5.26776053e-02 -7.57566765e-02 -4.54406649e-01
-8.33604559e-02 -5.51520705e-01 -8.96214604e-01 4.09315854e-01
3.38201851e-01 -1.17202088e-01 6.78874552e-01 -4.87386286e-01
-9.29625154e-01 5.55973947e-01 -1.10780876e-02 -7.83202589e-01
-6.60193682e-01 -4.80933562e-02 -3.19533050e-01 -5.20826411e-03
3.12478304e-01 3.91644388e-01 -2.64510334e-01 -2.57848442e-01
-6.96223825e-02 2.37072945e-01 9.49114501e-01 -2.35281304e-01
8.59312654e-01 5.81102848e-01 1.10329203e-01 -1.00886846e+00
-6.65843368e-01 -4.42240387e-01 -8.47895980e-01 -5.63320935e-01
1.18762100e+00 -7.19756067e-01 -3.34890276e-01 6.08797371e-01
-1.69486976e+00 1.33066714e-01 -3.78085747e-02 5.20358860e-01
-3.28807116e-01 6.60580099e-01 -4.92476016e-01 -8.65381956e-01
-1.84236661e-01 -1.14393318e+00 1.01397443e+00 2.97096848e-01
1.70434535e-01 -8.04124117e-01 6.37377501e-02 -1.70179009e-02
7.64116123e-02 6.34391129e-01 6.08579576e-01 -7.10229993e-01
-8.85847986e-01 -3.12198102e-01 -1.30172893e-01 4.54564631e-01
1.57197177e-01 4.74018902e-01 -8.56560707e-01 -3.73276919e-01
1.05407573e-02 1.73275784e-01 8.41279209e-01 5.97521126e-01
1.11359692e+00 -3.30743879e-01 -3.77236605e-01 6.60216749e-01
1.36099124e+00 5.84646940e-01 8.85364950e-01 4.37150329e-01
6.03605270e-01 4.24101442e-01 4.72681075e-01 4.22231048e-01
-2.01622993e-02 8.36898685e-01 9.62479591e-01 -1.84488982e-01
1.38933629e-01 1.17930323e-01 7.15476274e-01 2.96782881e-01
-2.09603190e-01 -3.52845669e-01 -6.33153319e-01 5.55708349e-01
-2.12596512e+00 -1.33798420e+00 -3.33724797e-01 2.23035645e+00
-9.66653507e-03 3.49977106e-01 2.47424796e-01 2.72670481e-02
9.60641205e-01 7.94928372e-02 -4.30476785e-01 -1.96633220e-01
-1.39376462e-01 -2.98804371e-03 2.83227175e-01 1.42475903e-01
-1.75199735e+00 7.89547622e-01 5.90354776e+00 2.86568165e-01
-9.45201457e-01 -1.92435041e-01 7.08564162e-01 2.12125421e-01
6.03399932e-01 -2.44436502e-01 -7.25809872e-01 6.29194796e-01
6.51208103e-01 2.61066288e-01 6.16203994e-02 8.50146115e-01
1.10045612e-01 2.02234387e-02 -1.28965664e+00 9.04366434e-01
2.98547357e-01 -1.08253372e+00 3.02033752e-01 -2.81347990e-01
6.52531862e-01 -1.32667974e-01 -2.11820632e-01 -1.45605624e-01
-1.10974640e-01 -7.37819552e-01 1.03739071e+00 7.83942580e-01
2.69111246e-01 -6.23678803e-01 1.07675946e+00 1.70494258e-01
-1.41060233e+00 -2.31238216e-01 -5.67726016e-01 6.86198696e-02
1.40437126e-01 1.70826122e-01 -5.02706289e-01 5.27973711e-01
8.41871977e-01 8.94444346e-01 -6.86017752e-01 1.48897839e+00
-2.10139230e-01 5.11786819e-01 -8.44812095e-02 2.33158365e-01
4.18337077e-01 -2.58497030e-01 7.69999146e-01 1.41556919e+00
4.79581058e-01 -1.73288509e-01 2.46049061e-01 6.70590878e-01
2.10402384e-01 -8.28666613e-03 -7.89320767e-01 4.83950704e-01
6.14254735e-02 1.30276728e+00 -7.19005287e-01 -5.12045085e-01
-8.80377769e-01 1.38643360e+00 2.09794819e-01 3.27093422e-01
-9.31848228e-01 2.86154598e-02 8.61107409e-01 -6.99361321e-03
7.47446299e-01 -1.59269616e-01 5.59153259e-01 -1.35347545e+00
3.21399063e-01 -8.22235763e-01 5.88142514e-01 -5.11032403e-01
-1.28537047e+00 8.25175047e-01 -7.09288940e-02 -1.69044304e+00
-3.97648931e-01 -9.18896198e-01 -1.02842402e+00 5.67865789e-01
-1.37057590e+00 -9.68677223e-01 -7.13413656e-01 8.12503159e-01
8.77714396e-01 -4.13000166e-01 5.23724616e-01 3.72392356e-01
-8.51592839e-01 2.83369452e-01 9.42886099e-02 5.03591359e-01
3.26215357e-01 -1.04198503e+00 6.58030629e-01 1.62958491e+00
9.91015360e-02 1.87436298e-01 7.00627923e-01 -6.13332927e-01
-1.15100503e+00 -1.42906308e+00 6.86452389e-01 -4.47462142e-01
4.59171146e-01 -1.35480434e-01 -1.01602352e+00 6.72895789e-01
2.69034445e-01 4.32646185e-01 2.22034216e-01 -6.81782186e-01
-1.41669065e-01 -2.26606235e-01 -8.68094683e-01 4.60519701e-01
9.23686266e-01 2.12686229e-02 -6.79872632e-01 1.50059894e-01
2.73662090e-01 -3.98910701e-01 -2.53909230e-01 2.97152251e-01
5.42338073e-01 -1.30035365e+00 1.03408217e+00 -9.22706842e-01
5.25669277e-01 -6.58849239e-01 -1.46458104e-01 -9.20891345e-01
-4.38317448e-01 -4.83636647e-01 -3.76956910e-01 9.58810329e-01
-3.49759199e-02 -1.77038729e-01 6.64127350e-01 3.45669121e-01
-3.48167494e-02 -3.49778950e-01 -9.71668065e-01 -8.26968491e-01
-5.87484241e-01 -7.67904818e-02 7.21537992e-02 6.03540003e-01
-6.64833546e-01 8.36903080e-02 -7.34336913e-01 6.35224402e-01
7.38948882e-01 -1.45709604e-01 8.45340252e-01 -1.25375903e+00
-3.74095947e-01 -2.81495959e-01 -1.06195748e+00 -8.95634711e-01
1.03804454e-01 -3.80395144e-01 -1.13773055e-03 -1.24781024e+00
1.83618516e-01 3.32108796e-01 -3.48966777e-01 1.29811242e-01
-3.35408121e-01 -2.68412735e-02 9.47572216e-02 -1.87717855e-03
-8.42872739e-01 3.45715106e-01 7.62863338e-01 -6.50372729e-02
-1.18794935e-02 2.75224328e-01 8.82427096e-02 1.04286885e+00
6.87740088e-01 -3.81926805e-01 -1.14913963e-01 -5.08202910e-01
-3.28285784e-01 1.22258306e-01 8.22020352e-01 -1.52850938e+00
2.38212049e-01 -4.70188446e-02 9.29404914e-01 -6.60376549e-01
1.94204763e-01 -1.25248539e+00 2.44464710e-01 8.44963849e-01
-1.31711990e-01 6.26292467e-01 1.91273749e-01 8.62121224e-01
-2.99304903e-01 -3.75646383e-01 9.64394510e-01 -1.29413903e-01
-9.34525371e-01 3.82510275e-01 -5.30409992e-01 -2.35848099e-01
1.25149059e+00 -1.79332107e-01 -9.10691097e-02 -3.45339864e-01
-6.08804524e-01 -1.91172093e-01 2.41803005e-01 5.51756501e-01
7.32133508e-01 -1.26207626e+00 -8.18445027e-01 3.40912342e-01
8.66735503e-02 -1.37863100e-01 3.58988009e-02 5.76990664e-01
-7.84879565e-01 1.55015379e-01 -6.47641003e-01 -6.98103726e-01
-1.46133244e+00 6.60874248e-01 7.06509233e-01 -1.92059368e-01
-7.05885231e-01 5.99666417e-01 4.14204359e-01 4.18980002e-01
5.02658427e-01 -5.27294815e-01 -4.42522407e-01 -3.62932712e-01
8.31910491e-01 3.92929494e-01 -1.05715524e-02 -9.24725235e-01
-4.75803733e-01 5.90802491e-01 -8.52715373e-02 1.94097489e-01
1.19134593e+00 1.87708527e-01 -1.00899599e-01 9.96169001e-02
1.10362601e+00 -5.02210975e-01 -1.67892396e+00 -2.12267727e-01
1.25025436e-01 -6.17467880e-01 -3.46925110e-01 -2.35337272e-01
-1.19463837e+00 7.40465641e-01 9.60981905e-01 9.10667852e-02
1.27356303e+00 -2.02016845e-01 5.77661335e-01 5.29974103e-01
5.57279773e-02 -7.31466413e-01 3.46236587e-01 4.85603511e-01
7.98699498e-01 -1.36675751e+00 -2.99768567e-01 -2.39173144e-01
-2.95986682e-01 1.61288655e+00 8.54055882e-01 -4.79295105e-01
4.88611490e-01 1.40927145e-02 -2.87993364e-02 -1.25457674e-01
-8.18704784e-01 -3.24635625e-01 6.17558777e-01 5.49123883e-01
2.66924381e-01 -3.33126724e-01 -1.83006614e-01 1.30533740e-01
4.59497809e-01 -1.33066133e-01 4.58041847e-01 8.93517673e-01
-4.66463625e-01 -9.41272140e-01 -6.83496773e-01 3.12950194e-01
-5.98030031e-01 3.01026344e-01 -3.05581927e-01 7.42915511e-01
5.04307628e-01 9.37893331e-01 2.58539915e-01 -2.99017012e-01
4.69514757e-01 4.49137017e-02 3.45779002e-01 -1.98087081e-01
-4.26604986e-01 2.57299423e-01 -1.47481427e-01 -8.35117221e-01
-5.56872725e-01 -9.81519759e-01 -6.13013089e-01 2.20327854e-01
-8.63318611e-03 -2.39809215e-01 3.18244368e-01 7.64123499e-01
6.73422683e-03 6.62332237e-01 6.37275398e-01 -8.52755249e-01
-2.57912606e-01 -7.66037703e-01 -2.39084393e-01 8.51223826e-01
6.45545483e-01 -6.69125021e-01 -1.18528679e-01 5.94256818e-01] | [7.913956642150879, 1.3588083982467651] |
51d1a6d9-08ac-496c-bc7c-41c3de301879 | a-new-sparse-auto-encoder-based-framework | 2201.12493 | null | https://arxiv.org/abs/2201.12493v1 | https://arxiv.org/pdf/2201.12493v1.pdf | A new Sparse Auto-encoder based Framework using Grey Wolf Optimizer for Data Classification Problem | One of the most important properties of deep auto-encoders (DAEs) is their capability to extract high level features from row data. Hence, especially recently, the autoencoders are preferred to be used in various classification problems such as image and voice recognition, computer security, medical data analysis, etc. Despite, its popularity and high performance, the training phase of autoencoders is still a challenging task, involving to select best parameters that let the model to approach optimal results. Different training approaches are applied to train sparse autoencoders. Previous studies and preliminary experiments reveal that those approaches may present remarkable results in same problems but also disappointing results can be obtained in other complex problems. Metaheuristic algorithms have emerged over the last two decades and are becoming an essential part of contemporary optimization techniques. Gray wolf optimization (GWO) is one of the current of those algorithms and is applied to train sparse auto-encoders for this study. This model is validated by employing several popular Gene expression databases. Results are compared with previous state-of-the art methods studied with the same data sets and also are compared with other popular metaheuristic algorithms, namely, Genetic Algorithms (GA), Particle Swarm Optimization (PSO) and Artificial Bee Colony (ABC). Results reveal that the performance of the trained model using GWO outperforms on both conventional models and models trained with most popular metaheuristic algorithms. | ['Ahmad Mozaffer Karim'] | 2022-01-29 | null | null | null | null | ['computer-security'] | ['miscellaneous'] | [-1.80413052e-02 -3.34298283e-01 1.41757075e-03 -4.84980382e-02
-9.54749212e-02 9.92763862e-02 4.47088510e-01 2.78182983e-01
-5.46022058e-01 1.14763987e+00 5.60806617e-02 3.59401494e-01
-7.06564367e-01 -9.06414986e-01 -5.23925006e-01 -1.12187278e+00
-1.98212504e-01 6.62318468e-01 -1.43079862e-01 -5.06980181e-01
4.98535275e-01 5.71065545e-01 -2.30460715e+00 8.57687294e-02
1.07299984e+00 9.36624646e-01 4.05145615e-01 3.84070128e-01
-3.54227901e-01 5.49477160e-01 -6.68017745e-01 -2.74025619e-01
-3.77575271e-02 -5.67134082e-01 -4.94222671e-01 2.75570583e-02
-4.50927764e-01 6.13888800e-01 4.60013114e-02 1.10089004e+00
6.24145746e-01 5.74682176e-01 5.95269024e-01 -1.04663265e+00
-5.16527593e-01 2.50973910e-01 -3.56890380e-01 5.34521103e-01
1.65344104e-01 -2.06137925e-01 6.58390820e-01 -8.00342977e-01
4.91472006e-01 9.78155732e-01 6.31942570e-01 3.53474289e-01
-7.99228549e-01 -3.80566955e-01 -5.69886029e-01 6.65489912e-01
-1.51811957e+00 -2.19810784e-01 7.84914970e-01 -3.84805202e-02
1.16872776e+00 2.89255649e-01 7.79127479e-01 6.90417469e-01
5.18883705e-01 6.28791273e-01 1.14336753e+00 -5.24129272e-01
5.82138956e-01 3.45697522e-01 -1.61421329e-01 9.20565486e-01
2.69670010e-01 1.83938518e-01 -5.62538207e-01 -1.69678345e-01
1.85066581e-01 -4.38582748e-02 -2.02137992e-01 -4.67225201e-02
-9.24255610e-01 1.26526916e+00 3.48261654e-01 1.04575574e+00
-1.00876641e+00 -2.58798063e-01 3.31932306e-01 1.15980022e-01
2.22847089e-01 7.79109895e-01 -2.75211483e-01 -4.32911754e-01
-9.36452091e-01 2.42294639e-01 7.25170612e-01 2.90382892e-01
5.63503146e-01 6.51267827e-01 1.28115430e-01 1.04529703e+00
-3.93042415e-02 2.07141504e-01 1.30461121e+00 -4.79903370e-01
-1.23657987e-01 8.85649741e-01 -2.91193992e-01 -1.28379726e+00
-4.32100862e-01 -8.75488281e-01 -1.26034749e+00 9.65296403e-02
-1.05097570e-01 -2.55846232e-01 -7.47478306e-01 1.21346247e+00
2.97948241e-01 3.90607953e-01 4.42847252e-01 6.72487974e-01
7.76181817e-01 1.11249256e+00 -3.48070748e-02 -3.91410232e-01
1.26299846e+00 -7.89595068e-01 -8.90234172e-01 -2.55818311e-02
3.37099016e-01 -8.17423761e-01 5.01954973e-01 7.57941842e-01
-1.03554261e+00 -7.25596249e-01 -1.03838408e+00 5.95824897e-01
-9.23051476e-01 3.08185220e-01 1.01303053e+00 8.22539985e-01
-9.95929778e-01 7.56170869e-01 -7.52959549e-01 -5.98848224e-01
4.85831708e-01 7.23789096e-01 -2.38072410e-01 1.01140440e-01
-1.04026651e+00 9.18816507e-01 8.34238231e-01 1.57913446e-01
-4.43778545e-01 -3.35016936e-01 -5.91499865e-01 2.99951315e-01
1.69819266e-01 -5.35744727e-01 4.00707245e-01 -1.11454785e+00
-1.60928285e+00 5.05160928e-01 -7.98561126e-02 -7.98778534e-01
-2.49840543e-01 1.23342440e-01 -6.71048403e-01 3.23157967e-03
-2.83882082e-01 4.86364245e-01 7.90201247e-01 -1.03819394e+00
-5.90099812e-01 -3.03383201e-01 -4.75263715e-01 1.65512308e-01
-6.72331929e-01 -1.06931582e-01 1.14701921e-02 -5.75562596e-01
-7.83271790e-02 -8.57740402e-01 -2.06883118e-01 -8.11594546e-01
-8.89562294e-02 -2.82496601e-01 8.37034583e-01 -4.46598262e-01
1.38005149e+00 -1.92799437e+00 4.83700037e-01 3.99717867e-01
-3.51691693e-01 5.89571774e-01 8.71886909e-02 5.35744846e-01
-2.27855429e-01 -1.03702538e-01 -3.90154928e-01 -4.66400906e-02
-3.95019859e-01 5.74445248e-01 2.93806225e-01 3.96648705e-01
1.69784024e-01 5.38194299e-01 -5.85006595e-01 -6.20359659e-01
3.08770776e-01 6.70889974e-01 -6.53068364e-01 1.79289967e-01
-7.73285180e-02 1.56901479e-01 -4.82996762e-01 8.37234557e-01
3.93285066e-01 -2.61608958e-01 8.99046734e-02 -3.13156635e-01
-2.53565609e-01 -5.46531081e-01 -1.30436468e+00 1.41660190e+00
-5.26619315e-01 5.04646719e-01 -2.05849543e-01 -1.74799669e+00
9.75284874e-01 4.90195930e-01 8.50440085e-01 -6.80345535e-01
6.44034922e-01 3.47171634e-01 3.66997510e-01 -7.11439550e-01
4.43876535e-01 -4.42398004e-02 5.04368901e-01 2.34528840e-01
4.08570379e-01 -8.75540376e-02 6.54313505e-01 -3.73759359e-01
5.97067177e-01 -2.63804585e-01 6.11121833e-01 -2.81416267e-01
1.08650172e+00 2.20352054e-01 4.85438108e-01 2.78474957e-01
1.56346366e-01 3.94415140e-01 1.27931520e-01 -6.80971146e-01
-1.00509155e+00 -2.76238948e-01 -1.91480175e-01 6.53369486e-01
-3.39285791e-01 -1.31175756e-01 -8.45421135e-01 -2.13029400e-01
-2.47252584e-01 6.83638334e-01 -7.16076374e-01 -1.81630686e-01
-5.02002716e-01 -1.54722714e+00 3.02131832e-01 1.40960857e-01
6.21780694e-01 -1.37059510e+00 -8.66395473e-01 4.33361977e-01
8.95795599e-02 -9.17015672e-01 4.57957745e-01 3.45673859e-01
-1.09104919e+00 -8.46262097e-01 -7.31397510e-01 -7.41367638e-01
6.24027252e-01 -9.75542963e-02 9.70731795e-01 4.35340047e-01
-5.03302991e-01 6.13134392e-02 -7.47982502e-01 -6.10064864e-01
-5.41094482e-01 2.67253280e-01 2.36416578e-01 2.76326299e-01
5.33540726e-01 -7.67909825e-01 -2.71519929e-01 9.48611870e-02
-1.11436701e+00 -4.38988239e-01 9.13367033e-01 1.29748440e+00
7.69391358e-01 8.49496543e-01 6.30968750e-01 -7.45213389e-01
9.19237673e-01 -6.67469263e-01 -7.41356850e-01 1.85955599e-01
-7.76045859e-01 3.63175899e-01 8.80585134e-01 -1.71358749e-01
-8.93729687e-01 -2.40223464e-02 -4.06394333e-01 -4.57723826e-01
-1.04908876e-01 7.13137150e-01 1.28112495e-01 -3.68648767e-01
5.05462587e-01 7.88081825e-01 2.83540189e-01 -3.66249532e-01
-3.97818387e-01 6.56350672e-01 1.36424869e-01 -2.96021044e-01
4.81177509e-01 3.20232987e-01 2.68608183e-01 -1.13544703e+00
-4.34751928e-01 -3.28317761e-01 -1.22722514e-01 -1.19028032e-01
9.44925129e-01 -2.92008519e-01 -8.31144452e-01 2.71562219e-01
-8.57007265e-01 4.84810054e-01 -1.67810455e-01 5.25885940e-01
-3.39435577e-01 7.56367818e-02 5.70316100e-03 -9.97707069e-01
-5.26962638e-01 -1.40600991e+00 6.02105260e-01 6.89316273e-01
6.15248922e-04 -1.07726729e+00 8.56981575e-02 2.38739803e-01
7.11097836e-01 2.62485772e-01 1.20543206e+00 -8.27789247e-01
-6.96612671e-02 -3.01360786e-01 4.49044138e-01 5.35207212e-01
1.31333858e-01 2.92466227e-02 -6.88645661e-01 -1.07425980e-01
2.36347675e-01 -2.60082901e-01 4.65240747e-01 6.40011072e-01
1.34482408e+00 -2.63566375e-01 -3.49529564e-01 6.29271150e-01
1.73117638e+00 8.09038103e-01 7.59855926e-01 6.07293427e-01
-6.19246205e-03 4.05580997e-01 5.81846952e-01 6.73212647e-01
-2.13931531e-01 4.64428246e-01 5.86159289e-01 1.89690337e-01
2.25148842e-01 6.08141907e-02 1.10523272e-02 1.22148013e+00
-6.04333401e-01 -4.57789838e-01 -6.75306559e-01 3.73446494e-01
-1.59410512e+00 -1.07874274e+00 2.31447831e-01 1.98931384e+00
3.48624110e-01 -1.26747265e-01 3.93472090e-02 7.36150861e-01
5.49617350e-01 1.08173087e-01 -2.80867726e-01 -8.22893918e-01
-4.41717237e-01 8.34554136e-01 2.31650785e-01 -1.37786502e-02
-9.73838210e-01 7.13296711e-01 5.42337990e+00 1.02003157e+00
-1.27160299e+00 2.46589016e-02 6.54927671e-01 -5.09851389e-02
4.63237539e-02 -3.83833200e-01 -5.78651309e-01 6.72380924e-01
1.12853014e+00 -1.07341573e-01 7.64810264e-01 9.42811370e-01
7.17307255e-02 -3.71500790e-01 -5.12534559e-01 1.26227164e+00
2.43067458e-01 -1.52014959e+00 9.77896899e-02 1.36727234e-02
1.15244222e+00 -2.26694927e-01 1.64086178e-01 1.78647533e-01
-5.20185009e-02 -1.44115698e+00 -7.27659240e-02 6.16672039e-01
4.30145524e-02 -1.30725861e+00 1.25947773e+00 3.41702729e-01
-8.31714571e-01 -4.42599684e-01 -6.74477220e-01 2.57616907e-01
-3.22656669e-02 6.43581688e-01 -6.94132984e-01 7.86415696e-01
8.00086379e-01 4.38978642e-01 -4.10363078e-01 1.27694368e+00
3.40531409e-01 5.05153298e-01 -3.14585298e-01 -8.78079116e-01
4.79042143e-01 -4.68620151e-01 6.24600708e-01 7.66978204e-01
7.26906359e-01 9.36898366e-02 -5.21860898e-01 6.49668217e-01
1.58030704e-01 5.86590290e-01 -4.61321086e-01 -5.64847350e-01
2.84450442e-01 1.11922503e+00 -7.33919024e-01 -2.84784228e-01
-3.14652413e-01 8.49730313e-01 2.26477347e-02 7.24602267e-02
-6.68195367e-01 -4.16794956e-01 5.56743264e-01 -2.09345236e-01
4.02429670e-01 2.50264741e-02 3.21724899e-02 -8.47429097e-01
-4.36415553e-01 -1.29015613e+00 5.68561256e-01 -5.74969292e-01
-1.00130522e+00 1.06617618e+00 -2.16117576e-01 -1.28355789e+00
-4.51502919e-01 -7.03002632e-01 -3.98977757e-01 5.68872631e-01
-1.30603755e+00 -5.59901059e-01 -2.56803364e-01 6.25086129e-01
7.92274117e-01 -1.06307185e+00 1.05491400e+00 3.59263688e-01
-7.20947921e-01 2.21879587e-01 5.47291934e-01 -3.18521887e-01
-6.40476421e-02 -9.25005615e-01 -5.82943857e-01 6.58169270e-01
4.43988681e-01 5.06166339e-01 9.58917916e-01 -2.90756613e-01
-1.80003428e+00 -4.38244760e-01 7.35209227e-01 3.60572249e-01
2.88527489e-01 2.08101124e-01 -7.91992784e-01 -1.75234266e-02
5.20106435e-01 -3.78864035e-02 8.61766756e-01 -1.44746333e-01
7.59011507e-01 -3.24454159e-01 -1.14344239e+00 1.25574261e-01
4.90565300e-01 1.55138776e-01 -5.19877374e-01 2.98225254e-01
-1.46257281e-01 -3.03256392e-01 -8.67721677e-01 3.87518704e-01
5.00576913e-01 -1.27528191e+00 9.43606138e-01 -7.24064708e-01
5.16143918e-01 -1.25332862e-01 -7.23903850e-02 -1.73171568e+00
-1.73647299e-01 -3.11560065e-01 -1.91061005e-01 9.72059131e-01
2.57102519e-01 -5.79481602e-01 1.07420123e+00 2.58652847e-02
-6.98098391e-02 -1.16489232e+00 -9.44305718e-01 -5.38364291e-01
-3.39849800e-01 6.79794252e-02 7.85167873e-01 9.15558934e-01
-3.12484562e-01 4.21740562e-02 -4.21952367e-01 -4.06118035e-02
4.06585813e-01 2.10920632e-01 4.06513751e-01 -1.34620965e+00
-3.18212599e-01 -5.62721252e-01 -7.60749936e-01 -1.71807349e-01
2.10583046e-01 -6.13944709e-01 -4.34090853e-01 -1.32218564e+00
-1.26610249e-01 -3.56451273e-01 -3.85516316e-01 1.08079776e-01
3.40271220e-02 8.13923255e-02 -1.92065090e-01 -1.37634277e-01
-3.00374795e-02 7.93357134e-01 1.03621399e+00 1.76305640e-02
-2.50594497e-01 1.55589521e-01 -5.41927338e-01 5.91083765e-01
9.05522764e-01 -5.73556960e-01 -5.22962689e-01 -1.39040112e-01
3.52703333e-01 4.53550741e-02 -1.97947755e-01 -1.52952552e+00
3.56331229e-01 -6.54094890e-02 4.74294275e-01 -3.98442447e-01
5.48822045e-01 -9.47699606e-01 2.94131398e-01 7.72570848e-01
1.00274824e-01 5.75555742e-01 2.25050434e-01 4.06181186e-01
-8.56650233e-01 -6.93963110e-01 8.25490594e-01 -3.76877904e-01
-9.13654566e-01 1.29038632e-01 -4.93794292e-01 -1.68109223e-01
1.17004395e+00 -5.56645811e-01 2.29280353e-01 -2.84747899e-01
-6.09418869e-01 -2.06688136e-01 -6.34717196e-02 2.99613118e-01
7.18453884e-01 -1.06501639e+00 -6.39131427e-01 1.71501845e-01
-2.32861072e-01 -3.71823967e-01 3.34060729e-01 9.16969180e-01
-7.79906929e-01 7.53606796e-01 -7.20092237e-01 -4.26399738e-01
-1.25770330e+00 5.99114060e-01 3.46259773e-01 -2.90957183e-01
-1.19781084e-01 7.47868896e-01 -5.66511929e-01 1.76573656e-02
-2.08843052e-02 1.16791643e-01 -9.21720147e-01 1.75008312e-01
3.10877115e-01 5.99214554e-01 4.17408347e-01 -7.12126136e-01
-3.60046834e-01 5.86563587e-01 2.22292423e-01 3.36709648e-01
1.98481464e+00 4.04305428e-01 -2.99514025e-01 3.63752991e-02
1.25605559e+00 -1.55495539e-01 -2.89269418e-01 2.01414928e-01
-6.67682216e-02 -3.15325528e-01 4.08409417e-01 -5.99976838e-01
-1.42179334e+00 8.29482257e-01 1.00602663e+00 3.49025398e-01
1.44376278e+00 -5.00412941e-01 7.19462752e-01 5.43898284e-01
2.22265095e-01 -1.36111283e+00 8.84077102e-02 3.20019096e-01
7.24194407e-01 -1.17625546e+00 1.02466404e-01 -2.22434215e-02
-6.73960507e-01 1.28997064e+00 5.00467241e-01 -1.60177782e-01
7.08320916e-01 3.81151855e-01 -3.40416998e-01 -2.61516899e-01
-6.44955635e-01 -2.83217937e-01 2.70877391e-01 4.24626470e-01
4.31697756e-01 -3.04451615e-01 -7.98339963e-01 4.85643387e-01
-2.79633015e-01 1.74225390e-01 2.39472851e-01 1.02110994e+00
-4.55884278e-01 -1.28787780e+00 -5.73955715e-01 6.79648280e-01
-7.11507142e-01 -1.65158696e-02 1.55789042e-02 8.30320835e-01
2.71592945e-01 7.00075030e-01 -1.19387016e-01 -4.40895200e-01
3.43860760e-02 8.25156644e-02 2.64825404e-01 -1.00635380e-01
-8.67497861e-01 -2.66999781e-01 -6.13852106e-02 -3.36180240e-01
-8.44852626e-01 -6.94334924e-01 -1.05859995e+00 -2.39727467e-01
-3.55380386e-01 8.86796772e-01 9.95446444e-01 1.00789714e+00
3.75704795e-01 6.99175000e-01 6.65740728e-01 -6.32974446e-01
-2.52093613e-01 -8.94788206e-01 -5.00251830e-01 1.46821439e-01
-1.74480692e-01 -1.00665319e+00 -1.96585715e-01 -7.86256865e-02] | [8.090047836303711, 3.280696153640747] |
2e2d66b4-bdcc-433d-ace9-e3459caa455a | amos-an-automated-model-order-selection | 1609.06457 | null | http://arxiv.org/abs/1609.06457v1 | http://arxiv.org/pdf/1609.06457v1.pdf | AMOS: An Automated Model Order Selection Algorithm for Spectral Graph Clustering | One of the longstanding problems in spectral graph clustering (SGC) is the
so-called model order selection problem: automated selection of the correct
number of clusters. This is equivalent to the problem of finding the number of
connected components or communities in an undirected graph. In this paper, we
propose AMOS, an automated model order selection algorithm for SGC. Based on a
recent analysis of clustering reliability for SGC under the random
interconnection model, AMOS works by incrementally increasing the number of
clusters, estimating the quality of identified clusters, and providing a series
of clustering reliability tests. Consequently, AMOS outputs clusters of minimal
model order with statistical clustering reliability guarantees. Comparing to
three other automated graph clustering methods on real-world datasets, AMOS
shows superior performance in terms of multiple external and internal
clustering metrics. | ['Pin-Yu Chen', 'Thibaut Gensollen', 'Alfred O. Hero III'] | 2016-09-21 | null | null | null | null | ['spectral-graph-clustering'] | ['graphs'] | [-7.44303167e-02 1.23387776e-01 6.11887313e-02 1.91740766e-01
-2.65571743e-01 -8.35578799e-01 2.39011392e-01 6.13612771e-01
6.63576052e-02 1.76268697e-01 -3.05947751e-01 -4.53566581e-01
-6.88873291e-01 -8.45568419e-01 -1.73084304e-01 -6.37463808e-01
-5.88895738e-01 1.02457690e+00 3.63048613e-01 2.23622650e-01
3.40007573e-01 4.87408429e-01 -1.33139133e+00 -1.88898921e-01
1.07028770e+00 5.44405520e-01 9.58339423e-02 6.64329171e-01
1.21859178e-01 3.82774442e-01 -4.87718821e-01 -1.42369896e-01
9.32163894e-02 -8.04718137e-01 -9.88499165e-01 6.23345375e-01
-3.99669737e-01 6.56240582e-01 4.54025455e-02 1.04372728e+00
2.56152421e-01 -1.97168216e-01 7.47051775e-01 -1.88691580e+00
-1.03138722e-01 9.50003386e-01 -5.95974863e-01 -1.15180530e-01
3.90536249e-01 -2.50328891e-02 1.14914978e+00 -4.87117112e-01
6.87740386e-01 1.16109550e+00 5.77390611e-01 1.37665540e-01
-1.69807506e+00 -3.52128357e-01 -3.60255055e-02 1.01118319e-01
-1.90464926e+00 -1.95648938e-01 6.26698732e-01 -7.04928160e-01
6.96740985e-01 3.17399830e-01 6.24916255e-01 1.56020463e-01
-2.78764427e-01 8.78091007e-02 1.06763268e+00 -6.37482166e-01
8.97040963e-01 -9.19723436e-02 3.01814705e-01 5.31667709e-01
6.30671620e-01 -2.80917764e-01 -1.17128581e-01 -5.22616804e-01
2.77538091e-01 -5.04625618e-01 -5.30697703e-02 -5.81833422e-01
-9.88491774e-01 7.41813660e-01 2.25064114e-01 4.30442661e-01
-9.83382314e-02 2.21891031e-01 2.08358854e-01 3.50718021e-01
3.18261981e-01 5.22720516e-01 -1.53757721e-01 2.22048149e-01
-8.74052107e-01 -3.36464196e-01 9.74011481e-01 1.17881322e+00
8.91660750e-01 -1.93965420e-01 5.22297442e-01 4.59836036e-01
3.61703694e-01 3.99706304e-01 -1.68868035e-01 -1.13884592e+00
2.82312721e-01 1.31034720e+00 -4.03687626e-01 -1.47581041e+00
-6.17640078e-01 -4.92237359e-01 -1.06780136e+00 -8.22183490e-02
1.72873497e-01 -8.00466444e-03 -4.81098920e-01 1.42603278e+00
4.10570562e-01 4.89374362e-02 -1.56156316e-01 4.92152572e-01
1.04982793e-01 3.68386120e-01 -2.41222292e-01 -5.94344616e-01
8.89528811e-01 -4.86888468e-01 -5.17883420e-01 1.50651991e-01
6.67305648e-01 -6.29040182e-01 5.02781928e-01 4.10917342e-01
-7.96346724e-01 -2.46381223e-01 -8.40969741e-01 7.25258648e-01
-2.37558991e-01 -5.95432706e-02 3.61480176e-01 8.84714663e-01
-1.63605666e+00 5.93767166e-01 -8.98052752e-01 -6.22709811e-01
-2.35847816e-01 4.37680304e-01 -2.81837195e-01 -8.66472349e-02
-5.46207428e-01 4.35195446e-01 6.78144813e-01 -1.69338599e-01
-5.25263309e-01 -1.60118088e-01 -6.01431668e-01 1.74094498e-01
5.26713192e-01 -4.23214644e-01 3.27666074e-01 -9.22809720e-01
-8.14087391e-01 6.03544652e-01 -9.30618420e-02 -4.45487142e-01
2.39092425e-01 6.07922316e-01 -5.58320105e-01 5.10663927e-01
2.39776611e-01 2.92564332e-01 5.73656023e-01 -1.60155129e+00
-4.36278552e-01 -4.28188264e-01 -4.54695553e-01 -2.13542394e-02
-1.86775103e-01 -4.78797480e-02 -7.41907895e-01 -3.42404902e-01
6.37745023e-01 -1.24839628e+00 -6.10056639e-01 -7.30886996e-01
-8.14562917e-01 -4.60348308e-01 7.01704800e-01 -3.89009625e-01
1.80973923e+00 -2.18820357e+00 4.30918723e-01 1.37845385e+00
5.79135895e-01 -3.18746924e-01 -7.81391561e-02 8.39966536e-01
-1.99576825e-01 6.33699417e-01 -3.45789313e-01 -1.97288975e-01
-3.94312553e-02 7.57250637e-02 3.29448640e-01 7.30117798e-01
-2.20572054e-02 3.20983618e-01 -8.91087532e-01 -7.70806730e-01
2.50023365e-01 -1.22498512e-01 -4.23031032e-01 -7.30190650e-02
1.75034761e-01 2.30633944e-01 -1.98862851e-02 3.46906006e-01
7.10468292e-01 -7.90306091e-01 1.08020079e+00 2.16812924e-01
8.99770111e-02 -2.22397760e-01 -1.69678009e+00 9.22649920e-01
1.92673758e-01 3.18695724e-01 2.52114773e-01 -1.07944858e+00
1.07019913e+00 3.27084452e-01 1.00772071e+00 -5.49652576e-02
8.34156759e-03 2.14876965e-01 1.79044545e-01 2.27413941e-02
6.62723184e-02 3.70348722e-01 -2.44281426e-01 8.00137758e-01
-2.11269483e-02 -3.18759819e-03 7.16397285e-01 9.37965930e-01
1.62198961e+00 -7.41558731e-01 3.61397475e-01 -7.59031832e-01
5.14355361e-01 2.24573538e-01 4.78473812e-01 5.24391055e-01
-1.21383645e-01 6.47582233e-01 7.88872898e-01 1.25645429e-01
-9.91526186e-01 -9.03013825e-01 2.88221717e-01 3.04434568e-01
2.01668113e-01 -1.08521438e+00 -1.05793107e+00 -4.69768614e-01
2.32822131e-02 2.26391658e-01 -4.78118896e-01 -8.07654187e-02
-2.76088595e-01 -4.74831194e-01 1.56620741e-01 1.34868771e-02
-1.21694401e-01 -6.88784420e-01 -9.60724354e-02 2.48781085e-01
-2.49275908e-01 -1.22670555e+00 -5.01154661e-01 1.67219982e-01
-8.31467032e-01 -1.79527915e+00 6.57762587e-02 -8.74841392e-01
1.16318107e+00 4.72687513e-01 1.27629459e+00 8.27739835e-01
-1.56510517e-01 4.96401340e-01 -7.47387230e-01 2.86655933e-01
-6.76992536e-01 3.06361556e-01 1.66445881e-01 8.89655873e-02
1.04400776e-01 -6.31461382e-01 -2.40152776e-01 5.46063662e-01
-7.73996890e-01 -8.57605040e-02 9.25271437e-02 2.84083307e-01
6.27698362e-01 1.02089894e+00 5.46988070e-01 -9.17437315e-01
7.76563942e-01 -5.45820534e-01 -7.00801134e-01 5.34056723e-01
-1.18982184e+00 -1.58310160e-02 7.63223767e-01 5.01632467e-02
-1.75228149e-01 3.31961751e-01 4.17916685e-01 -3.18528235e-01
5.55260926e-02 6.01973951e-01 -1.29686445e-01 1.09133653e-01
4.51493651e-01 4.23209220e-02 2.00050026e-01 -2.60512292e-01
3.75542849e-01 3.91291469e-01 3.30975205e-01 -2.99572825e-01
9.89570796e-01 3.69111866e-01 2.70454019e-01 -8.13339055e-01
-5.59961386e-02 -1.00233197e+00 -9.88199174e-01 -5.90907097e-01
6.23768806e-01 -6.29384875e-01 -8.78741860e-01 2.49569565e-01
-6.60429299e-01 -3.08359802e-01 1.84258208e-01 1.34315491e-01
-3.78753185e-01 6.80730581e-01 -3.80143851e-01 -9.44148183e-01
-1.65817723e-01 -9.62665737e-01 7.61862338e-01 -2.40175068e-01
-3.69707882e-01 -1.13346970e+00 -5.43138683e-02 9.08368528e-02
-3.78118753e-02 6.06560946e-01 1.07529938e+00 -4.89897281e-01
-4.12873328e-01 -2.07750723e-01 4.29318435e-02 -3.06406021e-02
1.76164329e-01 3.93093348e-01 -1.11193806e-01 -7.42820859e-01
-4.36120242e-01 3.51634562e-01 5.38568556e-01 4.21843648e-01
8.82062674e-01 -3.26673567e-01 -6.79110169e-01 2.32332140e-01
1.76759446e+00 3.34648043e-01 4.39894885e-01 -5.21752611e-02
5.33965886e-01 7.26369739e-01 3.78857315e-01 5.00298500e-01
3.16077441e-01 4.57798004e-01 3.24591815e-01 -9.22475383e-02
1.42768547e-01 -1.35408267e-01 5.05536720e-02 1.16149437e+00
7.18766153e-02 -3.64498556e-01 -1.41521811e+00 5.15989006e-01
-1.98039746e+00 -7.62635827e-01 -7.84026921e-01 2.22029471e+00
3.94892067e-01 1.30027890e-01 6.65712714e-01 6.03656948e-01
1.41232276e+00 -5.18596947e-01 -2.40196094e-01 -2.23943871e-03
-1.49348974e-01 -1.06536306e-01 5.98718226e-01 5.22755682e-01
-7.83477247e-01 6.45680964e-01 7.10631514e+00 7.31664181e-01
-4.27286834e-01 -1.57523021e-01 8.23596299e-01 2.84875929e-01
-2.41750583e-01 6.28921330e-01 -3.38713914e-01 6.71877801e-01
9.71806407e-01 -3.24041933e-01 7.51711667e-01 6.93148315e-01
4.24253523e-01 -3.09457421e-01 -8.71589124e-01 8.31987321e-01
-1.68666631e-01 -1.14109898e+00 -3.06244493e-01 5.44876814e-01
1.01255608e+00 -2.16511056e-01 -4.81512308e-01 -3.09354007e-01
8.89291883e-01 -9.58325386e-01 5.11205792e-01 2.58955091e-01
7.19667315e-01 -1.00414610e+00 5.79418361e-01 2.77312905e-01
-1.59971845e+00 -2.16307685e-01 -4.23805378e-02 1.33614331e-01
1.01037435e-02 9.83098924e-01 -7.71355689e-01 8.59778702e-01
8.84387672e-01 5.92576087e-01 -9.97949541e-01 1.17804337e+00
9.86851379e-02 9.24545646e-01 -5.20801008e-01 1.21863991e-01
-4.51160818e-02 -6.76434696e-01 3.39141130e-01 9.01712000e-01
2.03971326e-01 9.86053888e-03 5.24495840e-01 6.10282719e-01
2.79574543e-01 8.26781318e-02 -5.58073819e-01 -5.47099933e-02
1.25555217e+00 1.23309493e+00 -1.69915867e+00 -2.76111484e-01
1.72245070e-01 5.95476508e-01 2.02415049e-01 2.71011978e-01
-4.36351299e-01 -3.36669236e-01 -3.64687629e-02 3.80202085e-01
1.91238374e-01 -5.21162450e-01 -3.43778163e-01 -6.21949613e-01
-1.94341376e-01 -9.20996308e-01 6.87524736e-01 -3.73672962e-01
-1.29273367e+00 4.82684076e-01 -8.16165656e-02 -1.26829123e+00
-1.38461245e-02 -2.41076186e-01 -9.54862654e-01 3.44514132e-01
-4.74034190e-01 -7.98334420e-01 -2.20657393e-01 6.69809103e-01
-8.51342082e-02 -3.46934162e-02 5.14395714e-01 1.25080198e-02
-6.59803092e-01 1.90492332e-01 3.87193948e-01 -1.65839940e-02
3.83804679e-01 -1.57898974e+00 2.08417222e-01 1.20593262e+00
-1.77907035e-01 5.77546537e-01 7.62662888e-01 -8.10943127e-01
-1.25423825e+00 -1.02592492e+00 8.13955903e-01 -6.77016675e-02
7.57312715e-01 -4.83044595e-01 -5.21154284e-01 3.28008682e-01
1.56895489e-01 -3.94106090e-01 6.35723412e-01 2.23736510e-01
-1.21769883e-01 -1.71346530e-01 -1.05372429e+00 3.44669610e-01
1.15864289e+00 -1.90458611e-01 3.13802630e-01 3.18265736e-01
7.23356545e-01 5.72138488e-01 -1.24196649e+00 1.82663530e-01
-1.20762967e-01 -8.94429445e-01 6.63193583e-01 -8.73890072e-02
-1.36190012e-01 -7.84421682e-01 3.14027369e-02 -1.31943476e+00
-6.68990552e-01 -8.44189525e-01 8.64381492e-02 1.42872202e+00
4.61034447e-01 -5.47768772e-01 7.96912432e-01 1.86476618e-01
1.64910600e-01 -3.07444602e-01 -8.50310743e-01 -1.02589858e+00
-1.94512188e-01 -1.58768788e-01 6.50819898e-01 1.17608666e+00
3.92081857e-01 6.38476849e-01 2.97895796e-03 2.53656387e-01
1.19856989e+00 1.06354989e-01 7.46879399e-01 -1.77083611e+00
-1.72991276e-01 -5.75706482e-01 -4.94589001e-01 -7.00740069e-02
3.23638432e-02 -1.02605724e+00 -1.52057618e-01 -1.87480271e+00
2.64090389e-01 -7.20724344e-01 1.42701104e-01 2.72582024e-01
-3.51357050e-02 2.87385546e-02 1.19344234e-01 3.95633131e-01
-1.04677451e+00 2.90690754e-02 6.76875353e-01 -3.64783108e-02
-3.54738384e-01 1.16795979e-01 -5.48850656e-01 2.52795875e-01
1.03993559e+00 -4.81830418e-01 -5.00764728e-01 2.62881219e-01
4.94976550e-01 1.43224105e-01 1.89518794e-01 -1.12246180e+00
5.66150546e-01 -1.18545137e-01 -7.98403025e-02 -8.65980744e-01
-3.57975006e-01 -1.05612969e+00 9.29197609e-01 7.59692609e-01
-1.10243820e-01 5.02995491e-01 -3.22336435e-01 7.43164659e-01
-1.30971938e-01 2.26649828e-02 7.24326789e-01 -7.85875171e-02
-6.43076658e-01 8.34914595e-02 -6.91201866e-01 -2.63272896e-02
1.24294508e+00 -3.67030889e-01 -3.54160905e-01 -3.68292063e-01
-9.60582435e-01 7.15144992e-01 8.55941057e-01 1.45707339e-01
5.97438931e-01 -1.23961556e+00 -6.00008309e-01 -1.36813194e-01
2.44932413e-01 -8.77559036e-02 -1.07510507e-01 1.05757010e+00
-5.49341798e-01 1.85884297e-01 3.47129852e-01 -8.06811273e-01
-1.63749170e+00 8.46610069e-01 1.89226151e-01 -3.35706621e-01
-2.69431710e-01 1.01977475e-01 -3.56020987e-01 -4.02162403e-01
-5.58944754e-02 3.86817932e-01 1.57442968e-02 -1.16262972e-01
-6.06092550e-02 7.80431092e-01 4.68153358e-02 -7.63043463e-01
-6.32472515e-01 3.56876999e-01 4.66199189e-01 -1.12662733e-01
1.22529578e+00 -5.05478144e-01 -8.08421254e-01 1.19638562e-01
1.14090919e+00 -3.03886175e-01 -6.19486570e-01 5.68252243e-02
7.26504028e-01 -1.64792940e-01 -3.57471824e-01 -4.89249706e-01
-1.09742379e+00 2.80697614e-01 3.34738255e-01 9.83522713e-01
1.35757780e+00 3.51585031e-01 -8.41751620e-02 9.86645371e-02
5.46465158e-01 -1.48334062e+00 1.79669306e-01 2.22878486e-01
4.19409692e-01 -7.99375474e-01 6.25110120e-02 -7.64143944e-01
-4.64643329e-01 8.22123289e-01 5.18161058e-01 -7.91248083e-02
1.03924096e+00 2.87940383e-01 -3.65311235e-01 -5.97342074e-01
-8.07812393e-01 -3.66350025e-01 1.03454679e-01 8.75789821e-01
1.05169460e-01 4.78108108e-01 -4.99382079e-01 1.40348718e-01
-2.54632592e-01 -7.08402097e-01 6.93075478e-01 5.64813197e-01
-5.90873122e-01 -1.03001297e+00 -5.21082222e-01 4.79045779e-01
6.64449111e-03 1.77256674e-01 -1.10332847e+00 6.41478837e-01
-1.16859913e-01 1.75047028e+00 3.82560082e-02 -6.84088290e-01
1.36454543e-02 -3.67953092e-01 -3.07235252e-02 -5.74090004e-01
-4.24133569e-01 2.95009971e-01 1.22523636e-01 -3.87355804e-01
-6.08678699e-01 -5.84135473e-01 -1.29392338e+00 -7.08317935e-01
-7.73257673e-01 7.59318590e-01 4.70720887e-01 6.30041540e-01
6.39495134e-01 2.61083007e-01 1.25036228e+00 -2.62623757e-01
-1.22240493e-02 -8.14442337e-01 -1.02949059e+00 5.98006070e-01
-2.53021747e-01 -4.17340070e-01 -6.48400724e-01 -4.52344166e-03] | [7.03914737701416, 5.189591884613037] |
a579e9f7-774f-455e-ae1d-0f5d79aaaa6f | adversarial-semantic-collisions | 2011.04743 | null | https://arxiv.org/abs/2011.04743v1 | https://arxiv.org/pdf/2011.04743v1.pdf | Adversarial Semantic Collisions | We study semantic collisions: texts that are semantically unrelated but judged as similar by NLP models. We develop gradient-based approaches for generating semantic collisions and demonstrate that state-of-the-art models for many tasks which rely on analyzing the meaning and similarity of texts-- including paraphrase identification, document retrieval, response suggestion, and extractive summarization-- are vulnerable to semantic collisions. For example, given a target query, inserting a crafted collision into an irrelevant document can shift its retrieval rank from 1000 to top 3. We show how to generate semantic collisions that evade perplexity-based filtering and discuss other potential mitigations. Our code is available at https://github.com/csong27/collision-bert. | ['Vitaly Shmatikov', 'Alexander M. Rush', 'Congzheng Song'] | 2020-11-09 | null | https://aclanthology.org/2020.emnlp-main.344 | https://aclanthology.org/2020.emnlp-main.344.pdf | emnlp-2020-11 | ['paraphrase-identification'] | ['natural-language-processing'] | [ 4.99236763e-01 -6.20460622e-02 -1.19239055e-01 -3.83929878e-01
-1.55041742e+00 -8.51343334e-01 7.84397125e-01 8.17899764e-01
-4.97536182e-01 5.47795653e-01 1.00553012e+00 -2.92256504e-01
-3.25751007e-01 -5.94028711e-01 -5.45382679e-01 -2.14635819e-01
3.84450912e-01 7.05965936e-01 3.47667187e-01 -6.21318161e-01
1.16348565e+00 7.71702304e-02 -1.34588635e+00 1.00949574e+00
9.60857213e-01 3.07413310e-01 2.13442832e-01 1.03995883e+00
-4.75168318e-01 3.81706446e-01 -1.02819741e+00 -6.33023679e-01
-1.26985908e-01 -5.66321135e-01 -1.11048579e+00 -3.79568279e-01
5.84737957e-01 -1.05381086e-01 -5.22213578e-01 1.08296359e+00
7.92707801e-01 3.60834181e-01 1.01989508e+00 -9.63858843e-01
-6.79332793e-01 8.28149855e-01 -2.76892215e-01 6.95713699e-01
1.02582240e+00 -1.60414621e-01 1.27164638e+00 -1.02697396e+00
7.60707736e-01 1.62120032e+00 4.48895693e-01 6.96720421e-01
-1.19022524e+00 -4.58372176e-01 -1.12567078e-02 4.23310488e-01
-1.25556624e+00 -5.42393684e-01 5.33453763e-01 -6.69641048e-02
1.19655502e+00 7.42579281e-01 3.37433457e-01 1.37650752e+00
2.94565767e-01 1.03001499e+00 3.72098416e-01 -4.23630834e-01
2.10589215e-01 -3.09923649e-01 5.39916873e-01 2.15474904e-01
3.50996584e-01 -5.02902985e-01 -9.15531993e-01 -6.87628508e-01
-7.22625256e-02 -4.38022673e-01 -3.07831615e-01 1.68357998e-01
-1.03826535e+00 1.05382049e+00 3.27152945e-02 1.70910373e-01
-2.16056243e-01 3.76818748e-03 6.06627464e-01 3.76642883e-01
4.87876654e-01 1.01100230e+00 -3.01329434e-01 -2.76934564e-01
-1.06606579e+00 7.47321188e-01 9.82564747e-01 1.05802739e+00
2.83953905e-01 -4.91024733e-01 -7.04662561e-01 9.71957445e-01
-4.55457494e-02 6.85677648e-01 8.49993885e-01 -8.37626100e-01
7.54534900e-01 3.19915980e-01 2.14081436e-01 -1.10962152e+00
-3.73959094e-01 -1.59361318e-01 -3.28347117e-01 -6.82604253e-01
1.70311108e-01 -1.14543103e-01 -7.10682631e-01 1.50586069e+00
6.90943152e-02 -2.62070835e-01 2.50846028e-01 8.70002687e-01
1.03749692e+00 6.33507013e-01 2.58603632e-01 -1.74153060e-01
1.47669435e+00 -8.70173275e-01 -5.62510610e-01 -4.77048010e-01
1.07528508e+00 -1.37251520e+00 1.30927813e+00 1.92250520e-01
-1.26787162e+00 -1.09309614e-01 -8.36204410e-01 -4.10684317e-01
-1.12234965e-01 -1.40191808e-01 1.98455885e-01 2.90562749e-01
-9.96863663e-01 7.41654932e-01 -2.84587950e-01 -7.80127823e-01
1.14759617e-01 -1.53498456e-01 -7.55731016e-02 -1.40813857e-01
-1.60232818e+00 9.35631156e-01 3.06259692e-01 -5.32068372e-01
-4.56841916e-01 -5.59988320e-01 -5.53498507e-01 3.62173527e-01
4.27431047e-01 -1.19160378e+00 1.67070520e+00 -6.34543478e-01
-9.46047723e-01 6.80533171e-01 -5.96182644e-01 -4.02522773e-01
2.68082529e-01 -6.50588334e-01 -1.63262993e-01 5.39698422e-01
3.37795377e-01 4.60770816e-01 8.77340734e-01 -7.82215595e-01
-4.11318839e-01 -2.34229967e-01 -3.30487162e-01 8.43438387e-01
-4.20199841e-01 4.53865677e-01 -4.36566055e-01 -9.70240712e-01
2.24624664e-01 -7.78185427e-01 -1.18128017e-01 -6.45272613e-01
-7.51474202e-01 -5.89471579e-01 2.48727545e-01 -6.64885223e-01
1.62881649e+00 -1.90971076e+00 2.07399398e-01 2.11677790e-01
1.71926972e-02 1.16337791e-01 -6.14702344e-01 1.06078053e+00
1.27547517e-01 5.01158416e-01 -2.35478058e-02 -1.93069771e-01
3.92120965e-02 -3.60290676e-01 -6.25961721e-01 5.06757237e-02
-1.22739598e-01 9.75275874e-01 -1.17887270e+00 -5.23081005e-01
-1.20730914e-01 -5.84584363e-02 -5.45270860e-01 3.59206609e-02
-3.00013393e-01 -2.44492188e-01 -7.12861300e-01 2.88654000e-01
4.31488812e-01 -1.35593191e-01 3.82654890e-02 -1.12379692e-01
4.43065971e-01 1.03384137e+00 -6.04978502e-01 1.66834629e+00
-1.39095604e-01 5.82914829e-01 -3.18259358e-01 -8.03752363e-01
4.24691260e-01 1.42403375e-02 1.38425052e-01 -6.80898607e-01
4.38349731e-02 1.79165512e-01 -2.52086759e-01 -5.86429834e-01
1.09432983e+00 2.11034156e-02 -5.13426244e-01 6.43364608e-01
-2.20632836e-01 -6.70280874e-01 4.07587051e-01 1.04555559e+00
1.43398154e+00 -4.98490244e-01 2.17639148e-01 -2.71752238e-01
3.23991418e-01 4.62773353e-01 9.29876789e-02 1.47629273e+00
2.42171735e-02 7.47626901e-01 4.49108332e-01 9.76784006e-02
-1.21423912e+00 -1.20248330e+00 2.24030271e-01 1.28730071e+00
3.50292206e-01 -9.70783472e-01 -8.65719914e-01 -5.89817524e-01
2.86051810e-01 1.38572514e+00 -1.85520068e-01 -6.59027219e-01
-4.80209678e-01 -5.01265407e-01 9.41554964e-01 2.59518445e-01
-1.61368370e-01 -8.85343730e-01 -1.63487598e-01 1.99094981e-01
-8.29355538e-01 -8.72931421e-01 -8.66871655e-01 -1.54167399e-01
-9.32541549e-01 -1.04366136e+00 -6.14271045e-01 -4.91992623e-01
6.05529666e-01 8.38264823e-01 1.27283633e+00 2.61457831e-01
-2.78145045e-01 6.01432860e-01 -6.96812928e-01 -3.53778332e-01
-7.03486562e-01 2.40007624e-01 -1.18034780e-02 -7.58201301e-01
6.37152612e-01 -1.21203840e-01 -7.49663115e-01 -2.59703714e-02
-1.07203686e+00 -8.92365873e-02 2.41089761e-01 6.30353093e-01
2.22901583e-01 -3.12903464e-01 5.10623455e-01 -8.63624036e-01
1.67758083e+00 -6.12540543e-01 2.08059415e-01 5.66024005e-01
-5.53062975e-01 2.74216622e-01 4.96363640e-01 -4.25916135e-01
-8.73736918e-01 -3.79834563e-01 -1.47004142e-01 -3.18787992e-02
-1.24429584e-01 4.26291496e-01 4.92149115e-01 4.65403080e-01
1.16148937e+00 4.91688967e-01 -2.11052433e-01 -3.39033306e-01
6.85323119e-01 8.33172619e-01 4.62704599e-01 -6.00861728e-01
6.62964761e-01 2.75519133e-01 -3.94362479e-01 -7.96864688e-01
-1.10223675e+00 -1.00564158e+00 6.86621293e-02 -3.91008481e-02
5.85882604e-01 -7.97243357e-01 -2.85564542e-01 1.49968058e-01
-1.46877491e+00 4.74883392e-02 -2.16405243e-01 3.42880458e-01
-4.96018827e-01 1.01365709e+00 -7.40328848e-01 -4.18225437e-01
-8.87837768e-01 -6.77821875e-01 1.14827096e+00 1.21754959e-01
-1.16864502e+00 -6.21846080e-01 -5.29090874e-02 5.53484857e-01
3.43855172e-01 -5.93456089e-01 1.24975193e+00 -1.31326973e+00
4.41923887e-02 -2.28927419e-01 -4.25032303e-02 -9.12772939e-02
-9.46445316e-02 -4.37224656e-02 -4.07549590e-01 -2.05112815e-01
1.79466784e-01 -4.17996198e-01 1.04528999e+00 3.72537076e-01
1.00418651e+00 -7.11885273e-01 -4.42129970e-01 2.67876368e-02
8.53209257e-01 -1.26474246e-01 5.38327694e-01 3.49144042e-01
3.67783964e-01 7.87435710e-01 8.18540156e-01 6.87005579e-01
9.07860547e-02 6.08539045e-01 -4.52711843e-02 3.25260937e-01
-7.13111758e-02 -4.36328024e-01 2.68624723e-01 7.92970419e-01
7.11694062e-01 -8.82526457e-01 -7.39941716e-01 5.31970680e-01
-1.95101845e+00 -1.30544138e+00 -3.22615057e-01 2.19643331e+00
9.98268843e-01 2.39448130e-01 -1.28222346e-01 -1.71797633e-01
8.98683786e-01 1.20094016e-01 -3.41963083e-01 -7.03452170e-01
-2.36898899e-01 1.61247984e-01 3.18881512e-01 7.89655685e-01
-7.29557037e-01 1.40777004e+00 6.43127632e+00 1.31160331e+00
-5.75174630e-01 -1.14844874e-01 3.59834552e-01 -5.43150961e-01
-8.24635327e-01 1.80160534e-02 -8.32332373e-01 4.62731421e-01
8.78127158e-01 -7.89402902e-01 2.90805012e-01 7.30104506e-01
4.62125450e-01 -4.16507602e-01 -9.70208406e-01 7.09797561e-01
4.84147280e-01 -1.22954249e+00 8.48522604e-01 -5.69940567e-01
4.56999421e-01 -1.69257913e-02 -1.14313744e-01 2.16907084e-01
4.38576579e-01 -7.03052640e-01 7.48046279e-01 2.83005983e-01
2.78371423e-01 -5.47994852e-01 5.21278501e-01 4.26015854e-01
-7.08465457e-01 3.88435870e-02 -5.33002794e-01 2.69501328e-01
3.24131787e-01 5.87473571e-01 -9.22838151e-01 3.46490264e-01
4.56133008e-01 3.23327512e-01 -7.27468550e-01 9.79274869e-01
-3.43790382e-01 5.63085258e-01 -3.52221817e-01 -5.40839136e-01
2.18383834e-01 1.09651260e-01 1.00822902e+00 1.56800699e+00
5.79868078e-01 3.49934667e-01 -1.64196059e-01 5.87953269e-01
-8.99098888e-02 4.67378646e-01 -5.74371278e-01 -1.33407071e-01
9.69485164e-01 8.19858789e-01 -6.28934145e-01 -7.60956049e-01
-4.07927558e-02 1.44311368e+00 2.44772822e-01 3.72174054e-01
-7.16959655e-01 -6.57732904e-01 5.75135171e-01 4.03340869e-02
1.49062350e-01 6.39571920e-02 -3.28920722e-01 -1.29273772e+00
-2.86417007e-02 -1.12112606e+00 7.21362174e-01 -1.00548792e+00
-1.48089731e+00 2.40728483e-01 1.04165822e-01 -9.42230046e-01
-3.83428037e-01 -9.16188024e-03 -6.87827706e-01 7.25071013e-01
-9.48198259e-01 -2.46805817e-01 -4.17679772e-02 4.39185858e-01
1.18920493e+00 -1.56404637e-02 6.45358443e-01 -1.21484600e-01
-1.07489534e-01 6.45854771e-01 2.72005618e-01 -1.80276588e-01
1.05320728e+00 -9.94783640e-01 8.22254181e-01 8.34059238e-01
3.59042019e-01 9.97512639e-01 1.16138744e+00 -9.42633152e-01
-1.24937356e+00 -7.20551670e-01 1.61627162e+00 -5.70262313e-01
7.12870836e-01 -1.32694736e-01 -1.01974487e+00 1.00745775e-01
1.03049695e-01 -9.16524410e-01 7.54726350e-01 7.12536275e-02
-4.62363362e-01 4.51071084e-01 -8.25437307e-01 1.17237008e+00
1.27042019e+00 -7.61561930e-01 -1.22327793e+00 9.66151714e-01
7.97808826e-01 -3.83857608e-01 -2.36312985e-01 1.99629456e-01
3.93672585e-01 -8.31009328e-01 1.12390792e+00 -1.00413036e+00
5.34170866e-01 1.07292384e-01 7.23342225e-02 -1.45160568e+00
-5.55890977e-01 -7.19454229e-01 2.03586463e-02 7.82078624e-01
4.84959155e-01 -3.15833241e-01 7.18464315e-01 7.17872739e-01
-1.96716994e-01 -4.03724045e-01 -7.71322727e-01 -8.47518444e-01
2.03221053e-01 -3.54409039e-01 3.03907782e-01 6.82375431e-01
5.97471595e-01 8.02792847e-01 6.02030978e-02 -1.43099770e-01
3.88681889e-01 -6.40362203e-02 5.53100586e-01 -8.67718160e-01
-9.91464108e-02 -7.77649820e-01 -1.49542969e-02 -1.32093418e+00
2.82500386e-01 -9.55537736e-01 1.18994161e-01 -1.73954475e+00
5.59299707e-01 1.06801443e-01 -5.63177429e-02 1.02129661e-01
-6.31371737e-01 -1.13952793e-01 2.32332319e-01 4.82786298e-01
-8.78573835e-01 5.40547311e-01 8.65844905e-01 -9.16361511e-02
-8.60612467e-02 -3.48211415e-02 -1.05429685e+00 4.69570279e-01
1.06226420e+00 -1.03142464e+00 -5.64678013e-01 -4.20948982e-01
5.07960737e-01 1.66112229e-01 1.09974310e-01 -6.15530729e-01
4.39000130e-01 -1.45332962e-01 9.38586891e-02 -5.84162414e-01
1.09751172e-01 -5.46807833e-02 -2.69510895e-01 5.93684971e-01
-1.22157371e+00 4.41774487e-01 1.92212239e-01 7.00397015e-01
-6.05843551e-02 -9.02349293e-01 2.15076521e-01 -2.69665182e-01
-2.86034614e-01 -2.51783490e-01 -8.99994671e-01 7.80736268e-01
4.19683576e-01 5.79502210e-02 -7.92192459e-01 -9.41678405e-01
-3.47385317e-01 2.43923083e-01 4.11584020e-01 7.05104351e-01
8.25180531e-01 -9.19756055e-01 -1.04645288e+00 -3.64481360e-01
3.06258738e-01 -5.23340642e-01 2.08184257e-01 3.37150484e-01
-2.86545932e-01 6.78951800e-01 3.34620714e-01 -1.24353565e-01
-1.72346389e+00 3.94960821e-01 -9.17117521e-02 -2.45321557e-01
-4.20531064e-01 1.07681477e+00 1.30569652e-01 -6.34066164e-02
1.27335027e-01 -1.34059921e-01 -9.02310908e-02 4.18583602e-02
5.06987035e-01 3.96800578e-01 2.02829808e-01 -3.72523159e-01
-3.83786231e-01 2.13359952e-01 -5.48107564e-01 -5.06548226e-01
6.63902402e-01 -2.06055075e-01 -5.59027120e-02 -1.48682907e-01
1.23788428e+00 4.88224216e-02 -3.15352261e-01 -2.65806794e-01
2.78009444e-01 -5.07604599e-01 -1.29128456e-01 -6.48256898e-01
-2.09219992e-01 5.64435601e-01 -1.72687680e-01 4.04861093e-01
7.47255385e-01 2.42707714e-01 1.05931556e+00 9.66629505e-01
1.28462268e-02 -1.38321090e+00 4.30714518e-01 8.43649805e-01
1.08756304e+00 -8.52871239e-01 2.27566183e-01 -5.26978910e-01
-8.00084949e-01 1.02823222e+00 4.44465190e-01 6.27382472e-02
1.37686461e-01 -8.76000226e-02 5.70021756e-03 -2.28990987e-01
-1.09811091e+00 -1.13649722e-02 3.72005105e-01 2.58751541e-01
5.59721112e-01 -1.79457545e-01 -9.91574883e-01 5.04636884e-01
-4.61480588e-01 -4.38347161e-01 6.73448324e-01 9.59777534e-01
-7.04972744e-01 -9.83321965e-01 -3.84214073e-01 7.83921957e-01
-5.90426624e-01 -8.34552526e-01 -1.04054809e+00 3.08737606e-02
-8.79048049e-01 1.32874429e+00 1.16456009e-03 -3.48672092e-01
3.38685334e-01 3.33415270e-01 3.95416558e-01 -8.93513083e-01
-8.48436296e-01 -5.39214127e-02 6.47721827e-01 -5.85391104e-01
1.07583754e-01 -7.14392304e-01 -1.35177481e+00 -5.01368165e-01
-2.96141475e-01 5.05523860e-01 4.86781269e-01 8.63524616e-01
8.03110123e-01 -9.11199395e-03 3.90870929e-01 -4.91588801e-01
-1.18996012e+00 -1.05234194e+00 -9.54820067e-02 9.63269472e-01
-6.66850880e-02 -1.35091335e-01 -8.12943041e-01 -1.18956789e-01] | [12.155407905578613, 9.28640365600586] |
ec99b15f-6076-4f05-a646-c2d2f4ff31aa | generative-adversarial-networks-based-skin | 2305.18164 | null | https://arxiv.org/abs/2305.18164v1 | https://arxiv.org/pdf/2305.18164v1.pdf | Generative Adversarial Networks based Skin Lesion Segmentation | Skin cancer is a serious condition that requires accurate identification and treatment. One way to assist clinicians in this task is by using computer-aided diagnosis (CAD) tools that can automatically segment skin lesions from dermoscopic images. To this end, a new adversarial learning-based framework called EGAN has been developed. This framework uses an unsupervised generative network to generate accurate lesion masks. It consists of a generator module with a top-down squeeze excitation-based compound scaled path and an asymmetric lateral connection-based bottom-up path, and a discriminator module that distinguishes between original and synthetic masks. Additionally, a morphology-based smoothing loss is implemented to encourage the network to create smooth semantic boundaries of lesions. The framework is evaluated on the International Skin Imaging Collaboration (ISIC) Lesion Dataset 2018 and outperforms the current state-of-the-art skin lesion segmentation approaches with a Dice coefficient, Jaccard similarity, and Accuracy of 90.1%, 83.6%, and 94.5%, respectively. This represents a 2% increase in Dice Coefficient, 1% increase in Jaccard Index, and 1% increase in Accuracy. | ['Sharath Chandra Guntuku', 'Sanjay Talbar', 'Ujjwal Baid', 'Venu Pokuri', 'Bhakti Baheti', 'Prasad Dutande', 'Shubham Innani'] | 2023-05-29 | null | null | null | null | ['skin-lesion-segmentation', 'lesion-segmentation'] | ['medical', 'medical'] | [ 7.31884003e-01 5.07051706e-01 -7.33944261e-03 -1.40680268e-01
-6.67689860e-01 -5.43121576e-01 4.96144265e-01 1.61859453e-01
-2.95291632e-01 5.60480297e-01 -2.23149955e-01 -2.01908365e-01
9.61372256e-02 -8.42122257e-01 -3.06210160e-01 -7.98374295e-01
1.62370801e-01 1.03291739e-02 4.62846696e-01 1.38239592e-01
1.10129535e-01 5.68953693e-01 -8.66302490e-01 3.94163668e-01
1.64131320e+00 8.96320522e-01 -2.46488705e-01 6.13977551e-01
-1.34204909e-01 3.60064536e-01 -4.12871629e-01 -8.71684253e-01
2.40827829e-01 -6.25546038e-01 -6.90747380e-01 8.11633617e-02
4.38037932e-01 -2.27949709e-01 3.14797647e-02 1.12518358e+00
5.79657972e-01 -3.35506767e-01 1.10704005e+00 -9.29500580e-01
-5.45706451e-01 1.69269428e-01 -7.92312860e-01 -2.08124459e-01
1.44603342e-01 4.04991239e-01 3.53648931e-01 -3.14667523e-01
8.10767531e-01 8.29442978e-01 9.30841684e-01 8.66167307e-01
-1.14849830e+00 -4.89250362e-01 -4.09228116e-01 -6.90306649e-02
-1.28519082e+00 1.93713680e-01 5.56255877e-01 -5.62756777e-01
4.57270980e-01 5.84113359e-01 7.41708755e-01 9.26475763e-01
3.58148694e-01 5.84392190e-01 1.51217175e+00 -4.91233885e-01
2.46790826e-01 3.29594284e-01 -5.05024314e-01 6.84829473e-01
3.63132507e-01 -3.35169095e-03 2.37643421e-01 1.53613416e-02
9.65059459e-01 -4.00249474e-02 -1.04945689e-01 -8.62835199e-02
-3.45376700e-01 8.37105036e-01 7.54216492e-01 2.23121777e-01
-5.13691664e-01 -1.58123404e-01 2.37784371e-01 -3.04509550e-01
4.70255017e-01 4.44957942e-01 4.16179568e-01 3.38683426e-01
-1.12538695e+00 -5.25370054e-02 7.19845355e-01 2.50333875e-01
1.16425999e-01 -1.34289399e-01 -4.71953571e-01 9.09900725e-01
1.80314869e-01 3.00323308e-01 3.83257627e-01 -8.16413105e-01
5.98320477e-02 9.77796912e-01 -2.50528246e-01 -9.77907717e-01
-2.41605192e-01 -4.23957169e-01 -1.00333834e+00 4.58729297e-01
4.01887029e-01 -8.83805007e-02 -1.35836816e+00 1.34864330e+00
6.11626804e-01 3.14515620e-01 -2.09117860e-01 9.24484253e-01
5.69192052e-01 1.36352211e-01 4.33441818e-01 3.28141153e-02
1.14143395e+00 -9.70274985e-01 -6.29680514e-01 2.69344840e-02
3.75232905e-01 -7.65896499e-01 7.53054380e-01 2.99808294e-01
-1.18184614e+00 -2.01389149e-01 -1.04117703e+00 2.44533479e-01
-4.44299638e-01 3.88802946e-01 4.38165009e-01 8.89238179e-01
-8.69096041e-01 6.77311182e-01 -9.11542237e-01 -5.68266630e-01
8.47258329e-01 1.34395957e-01 -3.90847296e-01 -1.59684762e-01
-9.09108937e-01 8.90286624e-01 1.47058755e-01 1.87919049e-05
-4.23894346e-01 -1.05294514e+00 -7.52619028e-01 -4.13300306e-01
8.94409046e-02 -6.78544939e-01 7.15826869e-01 -1.13495898e+00
-1.61201131e+00 1.21918726e+00 1.69381872e-01 -5.79048753e-01
9.76353168e-01 5.77548109e-02 -3.93492371e-01 6.61196887e-01
-3.66455619e-03 8.57265472e-01 7.39859343e-01 -1.25339723e+00
-4.37814265e-01 -4.30885911e-01 -2.71154851e-01 2.45582446e-01
-2.71797389e-01 -1.12023883e-01 -4.05777276e-01 -8.48549128e-01
-2.31119186e-01 -9.51347828e-01 -3.99946868e-01 6.01507962e-01
-8.52114141e-01 2.79546797e-01 6.05125248e-01 -1.38271320e+00
1.27680540e+00 -1.88002157e+00 -2.27642074e-01 6.79321945e-01
2.40868539e-01 8.16124380e-01 -1.48431748e-01 2.03722581e-01
-6.15363568e-02 4.59140778e-01 -7.56251574e-01 -1.31788403e-01
-3.76134932e-01 -3.52114260e-01 2.28464440e-01 4.34674621e-01
5.96880138e-01 8.45956981e-01 -8.98231983e-01 -4.14810508e-01
4.80633825e-01 7.82723069e-01 -3.20268393e-01 -1.13136834e-02
-4.71792780e-02 3.48041892e-01 -1.59098789e-01 7.47674108e-01
9.93004620e-01 -6.28756210e-02 3.13595891e-01 -2.15555251e-01
1.43377364e-01 -2.89722562e-01 -9.60656285e-01 1.51042891e+00
-3.71623456e-01 3.25759023e-01 2.30694100e-01 -4.79949921e-01
7.95404255e-01 2.69188255e-01 4.31048691e-01 -3.89280945e-01
1.98838711e-01 2.56479144e-01 1.03297219e-01 -5.13785481e-01
-8.90049115e-02 -7.86255747e-02 4.92941231e-01 1.45093933e-01
-2.82289177e-01 -3.75880927e-01 1.79064929e-01 2.00695559e-01
1.09958780e+00 -1.18517004e-01 2.10700911e-02 -1.84294619e-02
5.92090726e-01 9.87890409e-04 2.96084464e-01 7.56205991e-02
-3.60075355e-01 8.34695756e-01 4.77304399e-01 5.33483969e-03
-1.02692068e+00 -1.11511302e+00 -2.79034048e-01 7.14946464e-02
1.91828206e-01 2.59240475e-02 -1.45880914e+00 -8.75235140e-01
8.08636472e-02 6.88943863e-01 -8.48968923e-01 -2.61200219e-01
-1.41653195e-01 -7.47224569e-01 7.08964825e-01 4.71629232e-01
7.85818458e-01 -8.73695254e-01 -2.17894524e-01 -3.83626446e-02
1.24724917e-01 -8.60342860e-01 -4.67459440e-01 -5.52949131e-01
-5.59836268e-01 -1.42496657e+00 -1.09531617e+00 -7.78096735e-01
1.20191908e+00 -2.12843895e-01 4.90443289e-01 2.09603794e-02
-1.07867801e+00 -6.42518327e-02 -2.59026647e-01 -3.29237670e-01
-6.92483425e-01 -7.14452863e-02 -4.24090564e-01 2.83420295e-01
1.25622019e-01 -3.51242989e-01 -1.13549769e+00 1.85936823e-01
-1.19410861e+00 8.75588804e-02 7.85344064e-01 8.68488967e-01
7.89727032e-01 -1.16933475e-03 4.87560898e-01 -1.09771335e+00
7.02747524e-01 -4.29924905e-01 -3.32031846e-01 3.52321416e-01
-6.47418618e-01 -4.60879534e-01 6.60077512e-01 -4.34207708e-01
-1.21078336e+00 2.65131652e-01 -1.99894458e-01 -3.22490633e-01
-2.23959118e-01 1.05891570e-01 1.34287581e-01 -3.69872421e-01
9.45441365e-01 6.99425489e-02 5.20506084e-01 -1.42098054e-01
3.58115017e-01 8.27810764e-01 5.86532593e-01 1.22067668e-01
7.85522699e-01 5.93141079e-01 5.89406043e-02 -5.59658706e-01
-5.13800681e-01 -3.23567361e-01 -6.37151778e-01 -3.67837638e-01
1.04955256e+00 -5.07550120e-01 -1.82016075e-01 1.01324940e+00
-7.84420133e-01 -4.69814509e-01 -1.32656753e-01 1.32456899e-01
-1.86883643e-01 5.45877993e-01 -8.05100024e-01 -7.04611719e-01
-8.91759396e-01 -9.44748521e-01 8.08731616e-01 9.01226163e-01
-4.24822897e-01 -1.17621017e+00 6.61767572e-02 5.73626995e-01
5.51195681e-01 1.04545879e+00 7.50742972e-01 -4.55162644e-01
-5.12193777e-02 -5.75859070e-01 -4.00113612e-01 8.45083773e-01
4.58600163e-01 4.71180290e-01 -7.68007874e-01 -3.38333666e-01
-4.45137560e-01 -5.96780516e-03 8.01268816e-01 5.63620329e-01
1.18502200e+00 -2.92867482e-01 -5.60196579e-01 7.08719969e-01
1.63102174e+00 4.13029641e-01 9.18299913e-01 -2.00051442e-01
6.44231379e-01 7.77395666e-01 3.96254361e-01 -3.94640528e-02
6.31273538e-02 3.22409093e-01 4.47369576e-01 -6.02710307e-01
-6.75817311e-01 -4.75687295e-01 -1.86541453e-01 2.44388938e-01
-8.92838743e-03 -7.62168765e-02 -7.69525468e-01 6.32226467e-01
-1.26399755e+00 -7.06663668e-01 -2.52549469e-01 2.21977019e+00
1.04236329e+00 1.11700013e-01 1.62166491e-01 3.19679119e-02
1.16476357e+00 -2.86950231e-01 -7.83122241e-01 -6.16505325e-01
9.47635770e-02 5.71445048e-01 5.25540709e-01 4.70162302e-01
-1.18844998e+00 1.00533879e+00 5.80301857e+00 1.16020477e+00
-1.47744596e+00 -2.57402420e-01 9.58896756e-01 2.46924266e-01
-2.61770636e-01 -3.13260943e-01 -2.77700096e-01 7.98261046e-01
5.60474455e-01 1.39076218e-01 2.46482834e-01 6.29755616e-01
1.76693127e-01 -3.66432101e-01 -4.19722825e-01 6.29045546e-01
1.85808763e-01 -1.38542652e+00 7.50728473e-02 1.14649802e-01
9.72808838e-01 -6.18841529e-01 3.25073451e-01 -4.12059039e-01
6.70609623e-02 -1.41341817e+00 6.65045679e-02 8.25007796e-01
1.46908474e+00 -7.67043948e-01 8.65559042e-01 -6.21866807e-02
-8.93530607e-01 3.03179175e-01 1.08387917e-01 6.28335357e-01
6.59699040e-03 7.96796620e-01 -1.10354877e+00 5.32645285e-01
3.40915591e-01 3.83749932e-01 -6.47701263e-01 1.45734549e+00
-6.18372321e-01 6.25728846e-01 -2.10297287e-01 -3.39877559e-03
9.85920876e-02 -4.61826771e-01 5.24212539e-01 1.18271637e+00
7.31804371e-02 -1.12969004e-01 -3.18422765e-01 1.04077673e+00
-2.56723468e-03 1.64396971e-01 -3.53514373e-01 9.64148268e-02
6.30410731e-01 1.67952418e+00 -7.46425986e-01 -2.38513704e-02
1.36346176e-01 1.43669116e+00 -1.88552856e-01 1.72391444e-01
-8.13710570e-01 -8.24852884e-01 3.73039454e-01 4.23258275e-01
-5.18413484e-02 3.72871190e-01 -4.80464667e-01 -4.77669358e-01
1.41852535e-02 -7.64683306e-01 3.02366942e-01 -6.01633251e-01
-1.46162200e+00 4.84579086e-01 -4.23473328e-01 -1.04603314e+00
-2.08431348e-01 -6.36637449e-01 -1.21377933e+00 1.19142163e+00
-1.36476552e+00 -1.50825083e+00 -6.49907947e-01 2.56850898e-01
1.81481317e-01 -1.00597674e-02 8.50365996e-01 5.25531843e-02
-8.56238544e-01 1.06892049e+00 2.06584439e-01 2.83834130e-01
8.02897573e-01 -1.44721532e+00 1.47151694e-01 8.45879972e-01
-5.35886407e-01 3.65604937e-01 3.07338715e-01 -7.79903829e-01
-7.77495921e-01 -1.50835419e+00 5.54792762e-01 -1.45272747e-01
4.66628075e-01 -4.63002408e-03 -7.86005080e-01 1.15583509e-01
1.47260457e-01 2.28109546e-02 9.48603511e-01 -7.67589927e-01
-2.56847113e-01 -8.84899572e-02 -2.06415963e+00 8.20056379e-01
5.78558326e-01 -2.24803925e-01 4.17153761e-02 3.33431840e-01
2.71644086e-01 -4.78767127e-01 -1.22356176e+00 5.66164196e-01
6.64361954e-01 -8.44518721e-01 8.92529964e-01 -2.55930126e-01
7.84153521e-01 -1.13666408e-01 5.08204281e-01 -1.45400405e+00
-1.10590458e-01 -6.47986531e-01 1.71469584e-01 1.30934024e+00
4.97675151e-01 -6.21997178e-01 9.64273512e-01 6.51259720e-01
3.64074367e-03 -1.36731315e+00 -6.63146257e-01 -6.02981567e-01
4.26318824e-01 4.27884758e-02 2.14968383e-01 7.93233871e-01
-1.39317155e-01 -2.72680640e-01 1.44951805e-01 -1.18369073e-01
7.35666811e-01 -4.80223596e-01 2.76970685e-01 -8.29825163e-01
1.23250991e-01 -5.98199368e-01 -3.67582023e-01 -1.89470321e-01
-2.91027993e-01 -9.56476808e-01 -3.54144931e-01 -1.68025029e+00
2.19816834e-01 -3.83823663e-01 -7.61458203e-02 5.11191368e-01
-3.53096545e-01 4.91239667e-01 -2.14926945e-03 -2.39959788e-02
8.29254985e-02 -6.91596093e-03 1.54615366e+00 -2.49060258e-01
-2.76379466e-01 4.98201475e-02 -7.67403960e-01 7.24002123e-01
1.17137778e+00 -1.56006426e-01 -2.00649485e-01 1.49291635e-01
-3.92775744e-01 -3.50137264e-01 5.32910109e-01 -1.02514887e+00
1.80689484e-01 -1.35955781e-01 4.99969304e-01 -1.40447468e-01
1.27762988e-01 -3.82541269e-01 1.17929399e-01 1.00047708e+00
-3.51056159e-01 -6.58898234e-01 2.14307591e-01 2.94654489e-01
-7.37258121e-02 -1.30604804e-01 1.29002428e+00 9.99185070e-02
-2.45671287e-01 3.34317178e-01 -1.57064110e-01 -1.93274826e-01
1.66718245e+00 -4.57139254e-01 -4.45802301e-01 -1.65567175e-01
-5.94890714e-01 5.69953509e-02 5.57635069e-01 7.92504176e-02
6.12541854e-01 -1.31572771e+00 -8.98948789e-01 1.26786798e-01
-7.91848227e-02 -2.81659011e-02 6.48372054e-01 1.03888273e+00
-1.06816840e+00 8.59220549e-02 -3.46843511e-01 -2.85455436e-01
-1.43897331e+00 -6.80188462e-02 6.28781855e-01 -5.35170019e-01
-2.99991637e-01 1.08160281e+00 -1.66768894e-01 -4.04909551e-01
1.25335352e-02 -6.99965358e-02 -1.57970414e-01 -2.58329153e-01
4.84110057e-01 6.19724214e-01 -1.34409498e-02 -3.90218794e-01
-2.01034680e-01 6.27885759e-01 -1.99626133e-01 5.82839958e-02
8.37430775e-01 3.18443090e-01 -1.30856723e-01 -3.38011563e-01
9.30834532e-01 3.77694406e-02 -1.31521583e+00 1.95209667e-01
-4.05341119e-01 -4.15850013e-01 -5.54019660e-02 -1.65672195e+00
-1.20581639e+00 8.76731277e-01 1.00817072e+00 3.84621352e-01
1.29927516e+00 -3.15090179e-01 1.01905394e+00 -6.48430407e-01
-1.50585547e-01 -1.13585067e+00 6.59895092e-02 -1.60451710e-01
9.69248474e-01 -9.17107940e-01 -1.97647825e-01 -1.04438734e+00
-7.77620792e-01 9.26274359e-01 6.98418319e-01 -3.55633348e-01
3.37843925e-01 2.50135005e-01 1.58732265e-01 1.13514349e-01
-5.78548461e-02 -4.70223837e-02 5.88610113e-01 9.27307248e-01
2.55227476e-01 2.22766459e-01 -8.12744796e-01 6.31576598e-01
7.29274331e-03 -1.58890039e-01 1.30720720e-01 5.58506906e-01
-1.54398546e-01 -1.17022574e+00 -1.92896709e-01 8.19788694e-01
-7.19424188e-01 -4.94298078e-02 -8.17083180e-01 8.87080431e-01
3.42534333e-01 9.11692798e-01 6.67218864e-02 -4.69873756e-01
7.70834386e-02 -1.92572236e-01 4.34049636e-01 -4.59679782e-01
-7.49105573e-01 -5.23323156e-02 -9.08552334e-02 -5.28561473e-01
-1.36133656e-01 -3.16432834e-01 -1.24904287e+00 -1.86681613e-01
-2.57439882e-01 -2.82202810e-01 1.02025366e+00 5.73427022e-01
4.63380426e-01 4.86605555e-01 6.91508353e-01 -3.03777903e-01
-4.47097719e-01 -8.60077322e-01 -6.14231646e-01 6.84940934e-01
-2.96518594e-01 -3.23804468e-01 -3.70025605e-01 1.23297319e-01] | [15.575652122497559, -2.9113235473632812] |
803e9aa3-c210-4a85-bb0d-813be0dfbfea | probabilistic-domain-adaptation-for | 2303.11790 | null | https://arxiv.org/abs/2303.11790v1 | https://arxiv.org/pdf/2303.11790v1.pdf | Probabilistic Domain Adaptation for Biomedical Image Segmentation | Segmentation is a key analysis tasks in biomedical imaging. Given the many different experimental settings in this field, the lack of generalization limits the use of deep learning in practice. Domain adaptation is a promising remedy: it trains a model for a given task on a source dataset with labels and adapts it to a target dataset without additional labels. We introduce a probabilistic domain adaptation method, building on self-training approaches and the Probabilistic UNet. We use the latter to sample multiple segmentation hypothesis to implement better pseudo-label filtering. We further study joint and separate source-target training strategies and evaluate our method on three challenging domain adaptation tasks for biomedical segmentation. | ['Constantin Pape', 'Anwai Archit'] | 2023-03-21 | null | null | null | null | ['pseudo-label'] | ['miscellaneous'] | [ 8.85687232e-01 4.10114050e-01 -4.93754178e-01 -8.15083206e-01
-1.03262150e+00 -5.89349806e-01 5.18250823e-01 1.65080085e-01
-8.61115336e-01 9.11736071e-01 2.97290217e-02 -1.03276856e-01
6.75718440e-03 -3.04965794e-01 -7.25553930e-01 -8.99471819e-01
3.45111191e-01 9.95618403e-01 7.11822689e-01 3.60213548e-01
6.91814423e-02 3.09869945e-01 -7.45438039e-01 4.64423001e-01
9.22178209e-01 7.63709903e-01 5.03607988e-01 5.04624724e-01
-7.82188401e-02 2.85447568e-01 -6.35079920e-01 -2.09100291e-01
8.55897367e-02 -5.13786674e-01 -1.19799614e+00 2.12029845e-01
8.46394897e-02 -2.86452919e-02 5.09047695e-02 1.20782888e+00
7.52499044e-01 2.05571860e-01 1.12045705e+00 -7.98146367e-01
-3.81387860e-01 5.66438735e-01 -5.36223590e-01 4.40603405e-01
-4.70572673e-02 -4.15486246e-02 5.36168754e-01 -6.24523818e-01
7.50424802e-01 9.66040611e-01 8.27799857e-01 9.60010767e-01
-1.65296531e+00 -5.80697119e-01 1.23924986e-01 -1.02500161e-02
-1.12265503e+00 -2.40350068e-01 7.57095695e-01 -6.14560962e-01
4.83493537e-01 -2.02316076e-01 1.27602577e-01 1.53188002e+00
1.68225333e-01 1.01916599e+00 1.52201426e+00 -5.04225433e-01
5.91686487e-01 3.94589007e-01 2.94315755e-01 2.83783615e-01
6.24889210e-02 3.58497119e-03 -2.20848560e-01 -4.69108224e-01
6.21381104e-01 -2.96384901e-01 -7.77889118e-02 -5.72680295e-01
-1.09455967e+00 8.55678856e-01 3.56182873e-01 3.55622381e-01
-3.21634680e-01 -2.47315869e-01 5.17516971e-01 8.16334970e-03
5.51607430e-01 5.56458712e-01 -8.23890567e-01 5.01785994e-01
-1.27232969e+00 1.35390684e-01 7.16660380e-01 8.89762580e-01
6.74761117e-01 -3.06356281e-01 -5.73379695e-01 1.13588393e+00
2.43429005e-01 2.65185356e-01 7.91254044e-01 -8.29516053e-01
-2.57275701e-02 1.91056386e-01 -1.14288092e-01 -2.17492253e-01
-8.34869742e-01 -6.31394207e-01 -7.16347098e-01 1.83802232e-01
7.12723136e-01 -2.73266613e-01 -1.35711730e+00 1.72607768e+00
4.92638290e-01 4.16354865e-01 -6.74144253e-02 6.98396087e-01
7.03992009e-01 2.48628885e-01 6.67832851e-01 -2.40121707e-01
1.28371727e+00 -9.16455448e-01 -7.36626148e-01 -6.92219853e-01
5.83850801e-01 -5.80348551e-01 9.95207667e-01 4.31291848e-01
-6.40091240e-01 -5.14028370e-01 -8.91291261e-01 1.73964202e-01
-3.27627301e-01 5.70289642e-02 3.52774739e-01 7.75916457e-01
-7.16191828e-01 7.11186349e-01 -1.03289163e+00 -5.23894548e-01
1.02899337e+00 5.79942703e-01 -1.96638167e-01 -1.96649522e-01
-1.07159841e+00 9.80269909e-01 7.37188935e-01 -3.59653085e-01
-7.76641667e-01 -7.29899466e-01 -6.27080798e-01 -4.37611282e-01
3.76951307e-01 -7.93580532e-01 1.46034896e+00 -1.03150856e+00
-1.61669731e+00 1.34629774e+00 -3.66065390e-02 -6.27649128e-01
4.57201093e-01 -6.44176453e-02 -2.53491908e-01 2.59503692e-01
3.32372993e-01 9.25159872e-01 1.07558477e+00 -1.27404034e+00
-5.85054159e-01 -3.87414843e-01 -3.94186974e-01 1.03812799e-01
-4.29243520e-02 -1.01916164e-01 -3.24585468e-01 -9.01198864e-01
1.14333086e-01 -1.14475942e+00 -5.65996587e-01 -2.49641389e-01
-4.45390731e-01 -2.45184332e-01 6.22027993e-01 -5.11020958e-01
7.89371789e-01 -2.05917025e+00 2.05831394e-01 1.38761967e-01
1.32372543e-01 2.83112556e-01 -1.49984941e-01 -1.95033029e-01
-1.91538647e-01 -9.28846896e-02 -8.89307797e-01 -4.59530860e-01
-1.45183355e-01 4.45881873e-01 -8.19256436e-03 5.72916150e-01
3.53339165e-01 8.03871036e-01 -1.05753434e+00 -8.09409142e-01
2.07799263e-02 2.20063940e-01 -5.25477827e-01 3.74033630e-01
-4.18634564e-01 1.06681442e+00 -6.41317189e-01 4.46947813e-01
8.29518616e-01 -3.43643606e-01 2.42029399e-01 -2.18014374e-01
5.31459928e-01 7.60936067e-02 -9.95657921e-01 2.04052401e+00
-1.89280108e-01 2.18246639e-01 -8.81986022e-02 -1.49867153e+00
6.57390535e-01 3.02417219e-01 7.05102682e-01 -3.94192725e-01
1.77132607e-01 3.04486006e-01 -9.73371938e-02 -5.96202195e-01
-2.23197609e-01 -6.82471275e-01 -1.71260431e-01 3.22809279e-01
6.08676493e-01 -2.22693831e-01 -4.00471576e-02 2.69637275e-02
1.16247463e+00 5.65987110e-01 5.97702026e-01 -5.03787875e-01
4.32327718e-01 3.36187072e-02 7.39862263e-01 9.74834800e-01
-6.17097378e-01 8.97540271e-01 2.24853575e-01 -2.34936640e-01
-7.78150380e-01 -1.19322371e+00 -6.27077281e-01 1.45178604e+00
-1.83479160e-01 1.82971746e-01 -9.55571234e-01 -1.34312367e+00
-1.13899313e-01 7.36697555e-01 -7.73449481e-01 -3.18680793e-01
-5.28588355e-01 -1.21963358e+00 4.94192511e-01 5.72010338e-01
3.77376735e-01 -1.18528581e+00 -4.61595863e-01 4.23173577e-01
-2.24644884e-01 -1.20436561e+00 -4.97869402e-01 7.41051733e-01
-1.10828817e+00 -1.03253984e+00 -1.13557065e+00 -1.02335000e+00
6.99474275e-01 -4.47362542e-01 1.24464202e+00 -3.79218757e-01
-1.10940479e-01 4.05939788e-01 -3.15379649e-01 -6.56070590e-01
-8.62034976e-01 4.09582764e-01 -1.32832646e-01 -5.70513494e-02
6.13134980e-01 -3.42304945e-01 -6.17762506e-01 2.21840411e-01
-9.42236900e-01 -2.13348210e-01 7.75020599e-01 1.21285295e+00
9.10483956e-01 -2.37118527e-01 7.94015050e-01 -1.68956232e+00
4.58406210e-01 -6.21460497e-01 -2.86672086e-01 3.00550997e-01
-5.25295079e-01 2.06824556e-01 2.72330940e-01 -7.34959066e-01
-1.13097858e+00 6.37184262e-01 -3.16060364e-01 -1.56125158e-01
-5.51189363e-01 3.52562934e-01 -1.27415091e-01 -6.88047633e-02
1.04877317e+00 -5.09596197e-03 -4.72053029e-02 -5.43809235e-01
2.45899618e-01 5.26285946e-01 6.12637222e-01 -6.00221336e-01
4.14788246e-01 4.46617484e-01 -1.39979526e-01 -5.89186013e-01
-1.22714067e+00 -6.22945070e-01 -1.14066744e+00 8.18836465e-02
1.10105681e+00 -7.37658918e-01 -5.87620400e-02 3.40623647e-01
-1.09044528e+00 -7.16881037e-01 -3.84063184e-01 5.05266190e-01
-7.71110952e-01 3.87051970e-01 -4.09231514e-01 -2.63670772e-01
-1.90567479e-01 -1.33961654e+00 1.15624130e+00 2.00302362e-01
-3.56752425e-01 -1.30973959e+00 2.92657048e-01 1.50527224e-01
2.38287106e-01 2.14633554e-01 7.77369559e-01 -1.33573091e+00
-4.22864929e-02 -1.73158869e-02 -2.27243707e-01 5.05780876e-01
2.14381784e-01 -5.92359185e-01 -1.19306672e+00 -3.00869703e-01
2.41407037e-01 -3.13462406e-01 1.07485962e+00 9.05964553e-01
1.31525970e+00 1.32676959e-01 -8.97928894e-01 4.60405797e-01
1.21332026e+00 8.15004036e-02 3.14479828e-01 4.40216720e-01
4.62761164e-01 6.95912421e-01 5.48121870e-01 8.54443163e-02
2.24465340e-01 6.60259247e-01 -4.81428429e-02 -3.52703512e-01
-4.39255923e-01 1.16429843e-01 -9.68382433e-02 3.53957832e-01
3.10593367e-01 4.20860052e-02 -1.19087315e+00 6.06504917e-01
-1.74220765e+00 -4.74605501e-01 1.57642692e-01 2.10722947e+00
1.34944129e+00 2.39291966e-01 2.96032488e-01 -2.50470102e-01
7.99740076e-01 -3.34459037e-01 -8.27466190e-01 2.64450349e-02
-5.33842295e-03 5.70295274e-01 7.19031811e-01 3.52669120e-01
-1.55566978e+00 1.06546688e+00 7.51703644e+00 8.98593843e-01
-8.67715955e-01 5.26100159e-01 8.82905722e-01 4.03689474e-01
2.05532119e-01 -3.01353216e-01 -7.48825192e-01 4.48640466e-01
1.03260136e+00 7.48809651e-02 4.02295180e-02 8.29170465e-01
-1.77466437e-01 -1.65961340e-01 -1.30918479e+00 7.44007647e-01
4.58991937e-02 -1.13107252e+00 -1.60145387e-01 -2.61089176e-01
8.88337493e-01 3.04440111e-01 -2.50179246e-02 3.80101472e-01
4.44089234e-01 -8.97209525e-01 3.47650558e-01 4.00791883e-01
6.20653570e-01 -2.74687886e-01 7.36310422e-01 2.91807622e-01
-5.42593718e-01 4.38932963e-02 -3.14026892e-01 5.90938270e-01
1.78710118e-01 7.02818334e-01 -1.15453732e+00 3.65101278e-01
6.65966094e-01 6.46621943e-01 -5.55196583e-01 1.32166862e+00
-2.94186234e-01 8.97956133e-01 -3.45406801e-01 4.59520221e-01
3.07217753e-03 -4.30874527e-02 5.66906750e-01 1.63529718e+00
-1.87356338e-01 -2.56083794e-02 4.74234164e-01 8.55744243e-01
5.92168944e-04 5.60882948e-02 -3.52130592e-01 2.67555773e-01
3.38090897e-01 1.19723487e+00 -1.16705441e+00 -5.47458410e-01
-1.93960115e-01 1.09962368e+00 2.63866127e-01 4.95404392e-01
-7.92691648e-01 -7.75807444e-03 3.03350002e-01 -6.55460656e-02
2.48684600e-01 1.42881777e-02 -6.37712538e-01 -8.49729002e-01
-3.97294581e-01 -7.95167208e-01 8.53809774e-01 -3.60072106e-01
-1.70901489e+00 6.16491556e-01 3.48713696e-01 -1.03648973e+00
-1.23538502e-01 -6.72442853e-01 -3.42437983e-01 6.94574654e-01
-1.70145988e+00 -1.11998987e+00 9.61401463e-02 5.20236671e-01
6.51403904e-01 -2.19786361e-01 9.66961741e-01 3.62694710e-01
-3.96983743e-01 6.04758739e-01 2.32450306e-01 1.21860653e-01
1.19736731e+00 -1.53487098e+00 1.97588682e-01 5.25769651e-01
-3.62770967e-02 3.50305408e-01 8.22807014e-01 -7.00915277e-01
-4.09631222e-01 -1.19204462e+00 4.47650522e-01 -6.64959311e-01
5.43622673e-01 -4.15859938e-01 -1.23651361e+00 7.54821301e-01
1.90118313e-01 1.64650425e-01 9.73968089e-01 7.21978024e-02
-1.31304190e-01 1.25451475e-01 -1.48356128e+00 2.43741781e-01
8.48678112e-01 -3.31945330e-01 -8.11963975e-01 7.73995161e-01
6.71012044e-01 -6.30845487e-01 -9.42461669e-01 4.77371454e-01
1.91985950e-01 -5.51921666e-01 1.03651571e+00 -6.67161822e-01
-3.22383270e-02 -1.08854145e-01 1.18491188e-01 -1.44280660e+00
-4.29794580e-01 -3.30048501e-01 1.24960810e-01 1.17762268e+00
4.83672172e-01 -6.23689175e-01 9.97268319e-01 7.21823454e-01
-1.79694638e-01 -4.59932268e-01 -1.07485843e+00 -8.17500055e-01
6.17766619e-01 -3.66078317e-01 2.45873734e-01 9.83857334e-01
-2.23596662e-01 3.37740213e-01 -2.13385560e-02 2.25401655e-01
8.87433171e-01 -1.59028262e-01 3.40413570e-01 -1.37961328e+00
-3.34110588e-01 -3.33525509e-01 -8.18307251e-02 -8.82441700e-01
5.29128671e-01 -1.19924545e+00 3.76892924e-01 -1.37181032e+00
2.25548476e-01 -6.20642662e-01 -6.92077637e-01 6.26554430e-01
-4.03450459e-01 4.37722087e-01 -3.06137860e-01 3.28815490e-01
-6.40788853e-01 2.17439547e-01 1.08794951e+00 -2.58805096e-01
-3.11648428e-01 3.48830372e-01 -6.31540179e-01 7.57031977e-01
8.14772546e-01 -1.06520391e+00 -3.45025063e-01 -2.61415035e-01
-3.50728035e-01 -2.93365717e-01 1.47083178e-01 -1.02801955e+00
7.72982240e-02 -4.04989868e-02 5.20012617e-01 -2.14133844e-01
1.02185318e-02 -7.91386127e-01 -2.42754132e-01 4.34822381e-01
-5.46206236e-01 -6.27358973e-01 3.50408882e-01 7.87076056e-01
3.41261551e-02 -5.20066261e-01 1.39048111e+00 -1.92121267e-01
-8.26903760e-01 2.06991717e-01 -3.39674473e-01 2.81722009e-01
1.00743437e+00 -2.70698488e-01 1.60653993e-01 2.89434761e-01
-1.46622121e+00 1.60454899e-01 2.90272534e-01 1.91780582e-01
2.43561745e-01 -1.15721762e+00 -7.30419576e-01 8.36746693e-02
2.94158876e-01 1.91533059e-01 1.62234813e-01 8.77037227e-01
-5.47532067e-02 1.92400157e-01 -2.77536660e-01 -9.38386917e-01
-9.35955644e-01 6.39953375e-01 5.84088743e-01 -5.15499651e-01
-6.58955753e-01 1.01300895e+00 5.50230503e-01 -7.58880794e-01
9.17518213e-02 -1.36376694e-01 -3.98204744e-01 -1.95782006e-01
3.64482790e-01 4.32495698e-02 1.80714116e-01 -3.50991249e-01
-5.39374948e-01 4.06784326e-01 -3.23131293e-01 -1.09696470e-01
1.29570365e+00 -1.37976035e-01 1.90266147e-01 4.49869782e-01
1.00243139e+00 -4.59727526e-01 -1.49409628e+00 -6.46270156e-01
6.10326111e-01 -1.22013830e-01 7.30364248e-02 -1.23454142e+00
-8.44536185e-01 6.62828326e-01 8.96200538e-01 -1.92522720e-01
1.14003193e+00 3.22377175e-01 5.17089784e-01 1.97372988e-01
1.82374805e-01 -1.36342931e+00 9.76317897e-02 3.09410155e-01
5.81923842e-01 -1.54611135e+00 8.13123435e-02 -4.36296791e-01
-8.59995961e-01 9.02925372e-01 3.75134766e-01 -2.40549326e-01
8.40012074e-01 2.59783119e-01 2.68282324e-01 -1.54423833e-01
-2.08370700e-01 -2.48415381e-01 5.13971388e-01 1.14817250e+00
4.47555751e-01 -2.07154229e-02 -3.18504155e-01 7.70376921e-01
3.37251127e-01 2.96210110e-01 3.39900434e-01 8.20567906e-01
-2.04609185e-01 -1.51209569e+00 -3.15100968e-01 5.87731540e-01
-6.58690214e-01 8.61290917e-02 -1.53622761e-01 5.79983354e-01
2.92162865e-01 6.47977531e-01 -2.66044289e-01 4.41269055e-02
2.43425488e-01 4.30595875e-01 5.99508882e-01 -1.09611404e+00
-4.26050097e-01 3.70202869e-01 -2.70656198e-01 -2.75045991e-01
-8.06895852e-01 -9.60268617e-01 -1.36794984e+00 5.98143041e-01
-2.62137204e-01 -5.67901768e-02 4.97018188e-01 1.22690046e+00
2.73454607e-01 6.19816899e-01 1.95154771e-01 -7.34862268e-01
-5.96705377e-01 -1.08411276e+00 -6.00168884e-01 6.75319135e-01
2.31013313e-01 -8.79248857e-01 -3.58072296e-02 4.78719592e-01] | [14.690811157226562, -1.9817419052124023] |
b059fffe-7f65-4275-bb44-99f962f657d0 | distilling-universal-and-joint-knowledge-for | 2307.03347 | null | https://arxiv.org/abs/2307.03347v1 | https://arxiv.org/pdf/2307.03347v1.pdf | Distilling Universal and Joint Knowledge for Cross-Domain Model Compression on Time Series Data | For many real-world time series tasks, the computational complexity of prevalent deep leaning models often hinders the deployment on resource-limited environments (e.g., smartphones). Moreover, due to the inevitable domain shift between model training (source) and deploying (target) stages, compressing those deep models under cross-domain scenarios becomes more challenging. Although some of existing works have already explored cross-domain knowledge distillation for model compression, they are either biased to source data or heavily tangled between source and target data. To this end, we design a novel end-to-end framework called Universal and joint knowledge distillation (UNI-KD) for cross-domain model compression. In particular, we propose to transfer both the universal feature-level knowledge across source and target domains and the joint logit-level knowledge shared by both domains from the teacher to the student model via an adversarial learning scheme. More specifically, a feature-domain discriminator is employed to align teacher's and student's representations for universal knowledge transfer. A data-domain discriminator is utilized to prioritize the domain-shared samples for joint knowledge transfer. Extensive experimental results on four time series datasets demonstrate the superiority of our proposed method over state-of-the-art (SOTA) benchmarks. | ['Zhenghua Chen', 'Kezhi Mao', 'XiaoLi Li', 'Min Wu', 'Qing Xu'] | 2023-07-07 | null | null | null | null | ['model-compression', 'transfer-learning'] | ['methodology', 'miscellaneous'] | [ 3.65754634e-01 -2.48015150e-01 -4.27227885e-01 -3.39140505e-01
-7.30713785e-01 -4.84570265e-01 4.37410116e-01 7.86895081e-02
-3.14814389e-01 7.13016927e-01 -7.20580518e-02 -3.17528009e-01
-2.90894747e-01 -8.42660308e-01 -7.96195149e-01 -6.22682869e-01
1.31255612e-01 5.28432190e-01 9.83043611e-02 -1.64339274e-01
-2.60286331e-01 1.77563921e-01 -1.34277415e+00 8.82193223e-02
1.28866374e+00 1.14095926e+00 2.51440823e-01 3.69613528e-01
-3.84593233e-02 7.40317881e-01 -6.17974997e-01 -6.92711234e-01
2.74662524e-01 -1.94418028e-01 -5.68334699e-01 -1.99257031e-01
1.99909374e-01 -4.21223730e-01 -7.34679878e-01 1.07017303e+00
6.16513073e-01 2.48147517e-01 7.37170219e-01 -1.36536217e+00
-8.43283772e-01 5.10766387e-01 -6.54336751e-01 3.49840105e-01
-1.27478838e-01 2.08339125e-01 4.20287788e-01 -6.49895787e-01
2.75130302e-01 1.07316697e+00 6.66221440e-01 5.85501790e-01
-1.20684648e+00 -1.30557597e+00 2.45363966e-01 3.65078300e-01
-1.30194306e+00 -3.05838555e-01 1.16876006e+00 -3.69503438e-01
6.76142097e-01 -6.25576302e-02 4.86754507e-01 1.30633640e+00
-5.22240810e-02 8.72463703e-01 1.05500138e+00 -3.16114910e-02
1.12968162e-01 3.46349835e-01 -1.30106732e-01 2.00019255e-01
1.38282344e-01 1.05950817e-01 -5.36074162e-01 -1.77701309e-01
7.53183901e-01 1.61708683e-01 -2.08183095e-01 -5.36730230e-01
-9.14270341e-01 5.95942259e-01 2.07910731e-01 1.18221767e-01
-3.97826403e-01 -5.85893393e-02 7.71271646e-01 6.31912649e-01
6.44537687e-01 1.37868272e-02 -7.79460013e-01 -4.33091193e-01
-1.07469928e+00 3.65172625e-01 6.70994699e-01 1.13713205e+00
4.64316368e-01 2.05857188e-01 -4.73800534e-03 1.02945340e+00
-1.17032103e-01 4.82159793e-01 9.72269475e-01 -3.64317536e-01
1.00860751e+00 5.26512444e-01 -1.52590200e-01 -9.54291999e-01
1.80546314e-01 -8.36970806e-01 -1.25418186e+00 -3.72102141e-01
1.13383867e-01 -3.04925323e-01 -7.60652959e-01 1.94943166e+00
5.85390747e-01 9.25877869e-01 3.49597871e-01 6.55521154e-01
5.30011892e-01 5.69859684e-01 2.92048275e-01 -1.91821367e-01
1.15937185e+00 -8.41319621e-01 -5.40065765e-01 -2.37837061e-01
4.68528807e-01 -6.71741784e-01 9.96421516e-01 4.36455458e-01
-1.16143727e+00 -8.34891081e-01 -9.60099339e-01 -3.21041830e-02
-4.17195499e-01 -1.72803015e-01 2.50530660e-01 2.42070049e-01
-2.64859617e-01 6.15760624e-01 -7.78818905e-01 2.91964598e-02
7.65554368e-01 3.02164257e-01 -3.06697130e-01 -2.53096402e-01
-1.58617222e+00 7.85984039e-01 6.47881806e-01 -5.06137908e-01
-1.13696504e+00 -1.32129896e+00 -6.10841095e-01 1.88460171e-01
1.42527193e-01 -7.87366569e-01 1.38658607e+00 -9.52418149e-01
-1.59471691e+00 6.06504261e-01 2.85990328e-01 -8.56589317e-01
6.68736637e-01 -4.60587651e-01 -6.94514811e-01 7.95627609e-02
-2.21906468e-01 4.00250465e-01 1.14668906e+00 -9.69590366e-01
-7.38532960e-01 -2.85186321e-01 -1.95159331e-01 3.14696342e-01
-9.46295500e-01 -3.44237775e-01 -3.58797103e-01 -1.19325614e+00
-1.74152568e-01 -6.43029034e-01 3.38881575e-02 -1.78473994e-01
-1.71182409e-01 -3.18570942e-01 1.25151622e+00 -9.47004676e-01
1.60760963e+00 -2.18133020e+00 2.21205667e-01 8.01626444e-02
1.80671558e-01 7.39199579e-01 -1.72010496e-01 3.16281855e-01
-3.93839031e-01 -3.03382844e-01 -3.40522587e-01 -5.29242218e-01
7.83689842e-02 4.63639587e-01 -8.23514581e-01 3.81407052e-01
1.42654449e-01 6.58197343e-01 -1.00731778e+00 -5.06170690e-01
2.08856001e-01 5.40864289e-01 -6.08100593e-01 4.42003220e-01
-1.38838723e-01 5.12399912e-01 -4.43566501e-01 4.53676432e-01
1.00582898e+00 -1.50479570e-01 5.63425245e-03 -1.19427659e-01
4.47285652e-01 4.42899615e-01 -1.05974638e+00 1.81560183e+00
-6.66777194e-01 2.24656761e-01 -1.62049606e-01 -1.36542678e+00
1.00824344e+00 4.06193733e-01 6.04039252e-01 -7.58236289e-01
-3.91673557e-02 1.77202076e-01 -6.55146986e-02 -3.03721070e-01
3.58675957e-01 -3.81074190e-01 -1.91234350e-01 4.00637984e-01
3.72922897e-01 -2.71367759e-01 -2.52516687e-01 -8.81029386e-03
9.11077976e-01 6.51177764e-02 2.22542226e-01 1.01096429e-01
4.58309889e-01 -1.61328137e-01 7.89366782e-01 2.22373262e-01
-3.78970712e-01 3.03008467e-01 1.48186907e-01 -3.59024823e-01
-1.00702143e+00 -1.13629031e+00 -3.46539058e-02 1.23713231e+00
-3.66275534e-02 -1.87718913e-01 -7.30983436e-01 -8.06633711e-01
4.19150978e-01 8.39444041e-01 -5.71080506e-01 -9.13545549e-01
-5.75200021e-01 -2.04199418e-01 9.65893269e-01 6.54907763e-01
7.23542869e-01 -6.38164163e-01 -5.49516737e-01 5.01119375e-01
-1.73359573e-01 -1.17521906e+00 -6.73417330e-01 2.55242765e-01
-1.15315592e+00 -8.98206055e-01 -7.73189247e-01 -7.38906503e-01
3.91148150e-01 2.63626784e-01 1.09083152e+00 -6.53286517e-01
1.00392334e-01 3.24088335e-01 -3.03318918e-01 -6.32583737e-01
-3.24425727e-01 1.68004662e-01 3.88754636e-01 2.51947921e-02
7.71374941e-01 -9.62757170e-01 -6.58587992e-01 1.88833565e-01
-1.12741065e+00 9.27465931e-02 5.61000645e-01 8.68807197e-01
6.85781956e-01 4.30427253e-01 8.36426258e-01 -6.97202086e-01
8.16178262e-01 -9.15742278e-01 -4.52032536e-01 2.36325085e-01
-5.47670305e-01 -9.16345268e-02 1.08998668e+00 -1.09459901e+00
-1.02355731e+00 -2.87219107e-01 1.85873508e-01 -1.24106801e+00
-1.42235667e-01 6.42312050e-01 -1.91786602e-01 3.08713555e-01
4.98595148e-01 7.51065314e-01 4.32262570e-02 -6.87729239e-01
2.39957437e-01 8.60577524e-01 9.04141724e-01 -8.19671690e-01
1.03253484e+00 1.50445208e-01 -2.85035640e-01 -3.65978807e-01
-6.92477286e-01 -2.01085120e-01 -5.00132501e-01 2.43855521e-01
4.49550807e-01 -1.34250295e+00 -2.11062461e-01 6.51642025e-01
-9.66747761e-01 -3.59423548e-01 -5.16010582e-01 5.23272336e-01
-6.33729160e-01 1.85336962e-01 -1.96806028e-01 -5.04101217e-01
-5.16762853e-01 -9.39690709e-01 7.07788825e-01 2.62707353e-01
-2.73785833e-02 -1.22894132e+00 2.25266650e-01 4.38142538e-01
4.61759329e-01 2.69689739e-01 1.05343235e+00 -9.94681120e-01
-1.50135653e-02 -2.20210224e-01 -1.19232744e-01 8.04109871e-01
2.31525302e-01 -4.62194532e-01 -9.35529172e-01 -4.86093521e-01
1.46046504e-01 -5.54117322e-01 6.19917929e-01 -8.35100114e-02
1.64346087e+00 -4.64874148e-01 -3.29986185e-01 7.10035145e-01
1.21233392e+00 1.15207694e-01 3.28162968e-01 1.16627946e-01
7.45717227e-01 4.06124830e-01 7.65107810e-01 7.64478385e-01
7.01849699e-01 5.87096155e-01 2.44708091e-01 9.66756567e-02
-1.21128224e-01 -6.84588969e-01 4.12084132e-01 1.25223494e+00
3.55002284e-01 5.59950173e-02 -9.95797515e-01 9.15931106e-01
-1.57283115e+00 -8.03570271e-01 5.40380776e-01 2.29273605e+00
1.52054060e+00 6.81636631e-02 6.58779144e-02 2.37758726e-01
6.45801663e-01 1.69416256e-02 -1.05564332e+00 -2.76044637e-01
1.93145841e-01 3.84932101e-01 4.50276941e-01 -4.38721664e-02
-1.18682206e+00 6.87322736e-01 4.60245848e+00 1.35018051e+00
-1.38817155e+00 3.64171356e-01 4.33049321e-01 -2.43378028e-01
-1.46118432e-01 -3.64520699e-01 -6.92743540e-01 8.31370652e-01
1.13938236e+00 -6.52343988e-01 4.23228949e-01 1.12921739e+00
-1.05747238e-01 4.51689750e-01 -1.16482103e+00 1.16279507e+00
-3.06994408e-01 -9.05042589e-01 1.18505113e-01 -5.89268208e-02
9.79800344e-01 -1.25967264e-01 3.93966764e-01 9.23066974e-01
4.81479496e-01 -8.48046958e-01 5.53870738e-01 3.97743732e-01
1.09453952e+00 -1.07425404e+00 4.43262935e-01 6.35350168e-01
-1.25012183e+00 -1.38017878e-01 -4.29739773e-01 1.45868942e-01
-2.73253620e-01 7.29383945e-01 -9.30462062e-01 7.96320856e-01
5.08159816e-01 8.10219288e-01 -2.87067622e-01 6.62656605e-01
1.38022780e-01 8.20003152e-01 -3.32549751e-01 5.52328348e-01
8.99432003e-02 1.06184050e-01 4.00460780e-01 1.21168458e+00
3.56321812e-01 -2.65421998e-02 1.87272415e-01 6.03700161e-01
-3.82212520e-01 -2.08826154e-01 -7.06416547e-01 -2.08899766e-01
9.17266488e-01 6.99699044e-01 1.38218492e-01 -5.25831521e-01
-4.38255548e-01 9.60243285e-01 4.95084316e-01 3.28000933e-01
-1.13342619e+00 -5.62835813e-01 1.16003776e+00 -5.42420382e-03
5.30627370e-01 -8.45615938e-02 -2.87648827e-01 -1.27308691e+00
4.23896611e-02 -1.05604839e+00 5.86526632e-01 -2.76036531e-01
-1.61379385e+00 1.63354203e-01 2.67289251e-01 -1.61774075e+00
-1.68596134e-01 -9.87587944e-02 -6.73654556e-01 1.08230066e+00
-1.92802119e+00 -1.17670274e+00 -1.91052586e-01 1.22337782e+00
4.88266438e-01 -3.77365559e-01 7.28883326e-01 6.97437286e-01
-4.34475154e-01 1.27409518e+00 4.36811328e-01 2.63206780e-01
8.60048652e-01 -1.00382864e+00 3.25995296e-01 6.52108908e-01
-2.87775874e-01 6.52011573e-01 5.65299153e-01 -5.10075986e-01
-1.53808749e+00 -1.63683689e+00 5.92735052e-01 -9.19634625e-02
6.89874053e-01 -2.52382100e-01 -1.33633840e+00 6.64452970e-01
3.29131559e-02 1.81826219e-01 1.03099275e+00 -1.74851656e-01
-6.77092910e-01 -5.21141469e-01 -1.34716523e+00 3.07253808e-01
8.16773236e-01 -7.33071029e-01 -7.94719338e-01 1.81751847e-01
9.54443336e-01 -6.03118896e-01 -1.32875991e+00 5.61639309e-01
5.28202057e-01 -5.71097672e-01 1.08405232e+00 -8.32140207e-01
6.88745141e-01 -6.56388626e-02 -2.27272600e-01 -1.58555055e+00
-2.24562902e-02 -4.64554906e-01 -8.30207169e-01 1.49800265e+00
-2.60839701e-01 -6.63850367e-01 6.27693951e-01 4.16491747e-01
-1.45506814e-01 -9.38703597e-01 -1.12428892e+00 -1.02642405e+00
5.74442744e-01 -5.00669658e-01 1.09541976e+00 1.35073161e+00
-2.03504547e-01 4.22331654e-02 -4.49018896e-01 1.81560636e-01
5.86261272e-01 1.49889931e-01 9.89247501e-01 -1.13469410e+00
-4.68610793e-01 -3.59904021e-01 -2.48659402e-01 -1.17674291e+00
1.39556065e-01 -8.49243462e-01 -3.53256315e-01 -9.67588246e-01
-1.30858690e-01 -7.26583183e-01 -7.81888783e-01 3.73769194e-01
-1.83567673e-01 -1.96046472e-01 1.94234177e-01 2.14081690e-01
-4.91464019e-01 9.83755350e-01 1.51555967e+00 -9.39567685e-02
-2.43329838e-01 -1.32355675e-01 -7.85459161e-01 5.10014832e-01
7.86736488e-01 -5.91215909e-01 -7.91618288e-01 -6.02661908e-01
-3.10380936e-01 1.51315361e-01 3.36454332e-01 -1.18910730e+00
4.72724915e-01 -2.85245299e-01 3.08218598e-01 -5.59927166e-01
3.16583961e-01 -1.18279481e+00 2.57726833e-02 4.65011239e-01
-2.27468967e-01 -1.59135118e-01 4.01590109e-01 7.37973630e-01
-6.23939931e-01 1.86403215e-01 1.04379046e+00 1.41450956e-01
-5.50284147e-01 6.36726916e-01 4.05367017e-01 4.00179118e-01
1.21337759e+00 -1.77350566e-01 -2.80112088e-01 -1.40837416e-01
-4.45066750e-01 3.74233216e-01 1.24221630e-01 5.90448201e-01
6.61290765e-01 -1.66629100e+00 -6.25697792e-01 4.67864454e-01
2.00369239e-01 5.44204950e-01 5.71664929e-01 7.79678941e-01
5.07084047e-03 3.67963016e-01 -3.09321076e-01 -4.37386602e-01
-1.05967581e+00 7.31428325e-01 1.67396396e-01 -6.66023195e-01
-3.43692333e-01 9.37876105e-01 2.79504627e-01 -6.50038958e-01
4.75526690e-01 -4.58329976e-01 1.49872541e-01 7.73466006e-02
5.34920275e-01 4.72086161e-01 -2.34665833e-02 -3.17174256e-01
-9.97100994e-02 3.67813736e-01 -4.59085673e-01 3.49624932e-01
1.36560202e+00 -2.74844631e-03 3.16280246e-01 3.01737010e-01
1.38374758e+00 -4.09677446e-01 -1.44965839e+00 -8.59737039e-01
-4.44807738e-01 -5.11635244e-01 -4.40024631e-03 -7.33771205e-01
-1.05245209e+00 1.14203477e+00 6.78903043e-01 -1.19456565e-02
1.58994758e+00 -3.82041812e-01 1.30848634e+00 1.25473842e-01
2.86207378e-01 -1.19520509e+00 1.92504764e-01 5.44579506e-01
6.86667144e-01 -1.11106014e+00 -2.07359537e-01 -2.73104645e-02
-7.38362968e-01 8.35158944e-01 9.01390731e-01 -1.64445668e-01
8.52810502e-01 9.72998515e-02 -2.65859425e-01 2.42739111e-01
-1.05219579e+00 2.46723235e-01 2.33819008e-01 8.34691465e-01
4.74880561e-02 2.50382155e-01 2.21923813e-01 1.17543221e+00
-1.92436323e-01 4.44548756e-01 -4.71035354e-02 8.61232996e-01
-1.89658687e-01 -1.08716834e+00 -3.32127571e-01 4.89889741e-01
-4.50763434e-01 9.42585915e-02 1.06046937e-01 6.37391269e-01
4.63221580e-01 6.81375682e-01 5.80451898e-02 -6.81801617e-01
4.67485875e-01 2.29888976e-01 3.16149592e-01 -4.38972503e-01
-6.69696629e-01 -2.95408279e-01 -4.41781878e-01 -1.78588063e-01
-1.46973610e-01 -6.69141471e-01 -9.78079796e-01 -5.42583048e-01
3.46897617e-02 9.85718668e-02 4.63741273e-01 7.55979538e-01
6.15829468e-01 6.99998558e-01 6.24135613e-01 -5.49867928e-01
-1.11712754e+00 -1.09269226e+00 -5.08675933e-01 6.23502553e-01
4.76751387e-01 -8.55458558e-01 -4.41703089e-02 2.66352564e-01] | [10.169921875, 3.120453357696533] |
7e5a9bf4-8820-43db-b7f2-ccb279120f9b | interpolated-convolutional-networks-for-3d | 1908.04512 | null | https://arxiv.org/abs/1908.04512v1 | https://arxiv.org/pdf/1908.04512v1.pdf | Interpolated Convolutional Networks for 3D Point Cloud Understanding | Point cloud is an important type of 3D representation. However, directly applying convolutions on point clouds is challenging due to the sparse, irregular and unordered data structure. In this paper, we propose a novel Interpolated Convolution operation, InterpConv, to tackle the point cloud feature learning and understanding problem. The key idea is to utilize a set of discrete kernel weights and interpolate point features to neighboring kernel-weight coordinates by an interpolation function for convolution. A normalization term is introduced to handle neighborhoods of different sparsity levels. Our InterpConv is shown to be permutation and sparsity invariant, and can directly handle irregular inputs. We further design Interpolated Convolutional Neural Networks (InterpCNNs) based on InterpConv layers to handle point cloud recognition tasks including shape classification, object part segmentation and indoor scene semantic parsing. Experiments show that the networks can capture both fine-grained local structures and global shape context information effectively. The proposed approach achieves state-of-the-art performance on public benchmarks including ModelNet40, ShapeNet Parts and S3DIS. | ['Hongsheng Li', 'Jiageng Mao', 'Xiaogang Wang'] | 2019-08-13 | interpolated-convolutional-networks-for-3d-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Mao_Interpolated_Convolutional_Networks_for_3D_Point_Cloud_Understanding_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Mao_Interpolated_Convolutional_Networks_for_3D_Point_Cloud_Understanding_ICCV_2019_paper.pdf | iccv-2019-10 | ['3d-part-segmentation'] | ['computer-vision'] | [ 2.13021822e-02 -3.83984685e-01 -1.44689053e-01 -8.06983232e-01
-3.22904766e-01 -5.86123109e-01 2.73387045e-01 2.50419259e-01
-1.15085848e-01 1.29959553e-01 -1.33506447e-01 -4.00847226e-01
-1.70492023e-01 -1.16689646e+00 -1.41236472e+00 -2.12116674e-01
-1.02833614e-01 4.14835721e-01 5.10121822e-01 6.28880560e-02
3.21819246e-01 1.30914736e+00 -1.62544477e+00 6.72103584e-01
9.97611582e-01 1.28089833e+00 4.99677509e-01 3.17756683e-01
-9.17442739e-01 1.58913925e-01 -2.04643875e-01 -4.70658876e-02
4.46089834e-01 6.05753601e-01 -7.18533814e-01 7.57642612e-02
7.19859838e-01 -1.01721115e-01 -1.17433578e-01 1.13866365e+00
2.14550629e-01 1.56479836e-01 3.66168261e-01 -1.18791211e+00
-9.00385320e-01 2.92139202e-01 -5.39243758e-01 -1.46440431e-01
-6.29983991e-02 2.46366281e-02 7.20091164e-01 -1.16834390e+00
5.03939807e-01 1.44211030e+00 1.07749867e+00 2.05496088e-01
-1.00407708e+00 -8.49621832e-01 4.32204902e-01 7.72102922e-03
-1.39752007e+00 1.22409808e-02 8.73386443e-01 -3.33103478e-01
1.25469184e+00 3.98552001e-01 6.26485288e-01 4.70435530e-01
-1.48441717e-01 6.55700386e-01 7.32396662e-01 -3.97172198e-02
2.88231313e-01 -2.57285923e-01 2.80079514e-01 7.45004714e-01
2.94984579e-01 -1.83910161e-01 -2.13557363e-01 -3.26045424e-01
1.39225137e+00 5.98599195e-01 -5.85879125e-02 -5.94465137e-01
-1.17863560e+00 6.90204203e-01 1.11141706e+00 1.14047304e-01
-3.46449226e-01 3.76970530e-01 3.19065660e-01 1.25572979e-01
5.06158888e-01 9.77911055e-02 -8.54102790e-01 1.26492530e-01
-6.58280492e-01 2.46858492e-01 5.72039127e-01 1.47301674e+00
1.01770759e+00 -1.09559886e-01 -7.14237317e-02 9.91110027e-01
2.71106213e-01 6.00862563e-01 -6.03452660e-02 -8.09261560e-01
6.30413294e-01 1.07974720e+00 -1.58874527e-01 -9.78457332e-01
-4.32657480e-01 -1.85600385e-01 -9.67208922e-01 3.89375210e-01
-6.63846508e-02 3.11069608e-01 -1.33369613e+00 1.22210121e+00
5.51525533e-01 7.08845198e-01 -2.41769776e-01 8.18703830e-01
1.34126282e+00 7.57970393e-01 2.59969495e-02 6.18738294e-01
1.28973329e+00 -8.34898651e-01 -8.59690681e-02 -7.12468997e-02
4.01422679e-01 -7.23536789e-01 1.04462028e+00 -1.76519215e-01
-9.49886620e-01 -7.30886400e-01 -9.65168655e-01 -4.18011546e-01
-5.16547382e-01 1.38641641e-01 1.02419126e+00 1.39941201e-01
-7.67362177e-01 7.87591457e-01 -1.12404895e+00 -1.92104265e-01
9.71169770e-01 7.27679610e-01 -3.94329786e-01 -3.29475403e-01
-4.62207496e-01 2.25042805e-01 3.83429974e-01 1.52776271e-01
-9.34532657e-02 -1.13265979e+00 -1.09956014e+00 2.37464294e-01
-5.81216626e-02 -8.05256963e-01 1.03985369e+00 -6.33821785e-01
-1.24089515e+00 7.42596149e-01 -3.44144434e-01 -1.93480372e-01
-6.52308911e-02 -3.44898105e-02 -1.70008540e-01 -4.26146574e-02
1.54895321e-01 8.18500102e-01 6.88079894e-01 -1.24687278e+00
-6.33911550e-01 -5.55857241e-01 -2.65361220e-02 6.73739836e-02
4.45455573e-02 -1.15500346e-01 -7.92744517e-01 -5.38349271e-01
7.58985996e-01 -6.60379648e-01 -3.65417838e-01 3.81997854e-01
-4.52165246e-01 -5.48388124e-01 1.15950716e+00 -2.18374223e-01
5.50479889e-01 -2.26370835e+00 -3.24675620e-01 5.55015802e-01
1.28727883e-01 2.90500641e-01 -2.12010637e-01 5.19264154e-02
-1.03722990e-01 1.31708503e-01 -3.46295744e-01 -3.81049961e-01
1.48038849e-01 5.87105393e-01 -3.00534904e-01 2.31820777e-01
6.19604588e-01 1.13060534e+00 -5.87710619e-01 -1.35538474e-01
5.69801569e-01 7.21945345e-01 -7.82297790e-01 -4.26267227e-03
-4.13689733e-01 2.38194302e-01 -7.28261769e-01 1.04201233e+00
1.49668717e+00 -3.47841650e-01 -4.17679936e-01 -5.36734521e-01
-4.53021169e-01 2.54885018e-01 -1.22052431e+00 2.06175327e+00
-5.59031367e-01 3.04710448e-01 2.16436669e-01 -8.37254047e-01
1.11211109e+00 -2.10731715e-01 4.95374918e-01 -3.98354232e-01
-1.92242395e-02 2.71346420e-01 -3.58557254e-01 -1.54096887e-01
4.09258068e-01 4.03451681e-01 5.79227693e-02 -6.86890408e-02
7.30601326e-02 -2.77025133e-01 -3.57511282e-01 -2.45098948e-01
9.54604149e-01 1.20421447e-01 -1.03270285e-01 -3.74405950e-01
4.27646488e-01 1.60563022e-01 5.91438949e-01 5.87541699e-01
2.91127592e-01 7.76836336e-01 2.27535665e-01 -9.45391357e-01
-9.45967615e-01 -1.16044366e+00 -4.34050798e-01 8.53219211e-01
3.40756029e-01 -2.80320197e-01 -4.30648178e-01 -4.90174353e-01
6.68274045e-01 4.41424459e-01 -3.40691954e-01 2.16903448e-01
-8.74784946e-01 -2.88660675e-01 2.48118713e-01 1.03873670e+00
6.78339958e-01 -1.14544356e+00 -3.38894784e-01 1.63603619e-01
3.86649430e-01 -1.31405008e+00 -5.60275257e-01 4.00993496e-01
-1.20135224e+00 -9.64606583e-01 -2.95472473e-01 -1.13742590e+00
8.87371421e-01 5.39406121e-01 1.29038215e+00 1.28448457e-01
-1.85779393e-01 2.23759249e-01 -2.85004407e-01 -5.21244705e-01
2.34332114e-01 1.32847071e-01 -2.08587274e-01 -1.32122412e-01
6.37261450e-01 -9.59112465e-01 -5.34167826e-01 2.70570904e-01
-1.02041900e+00 8.99952427e-02 6.25566065e-01 7.17491984e-01
1.32699442e+00 -2.09082112e-01 7.65575394e-02 -8.89411211e-01
2.57158220e-01 -3.15005749e-01 -8.74842525e-01 3.85277234e-02
5.25511652e-02 1.18596330e-01 5.42271614e-01 -3.00933629e-01
-6.99300706e-01 5.50273120e-01 -4.98381853e-01 -1.02213275e+00
-6.32942319e-01 2.82201767e-01 -3.25132728e-01 -6.16873562e-01
2.20728382e-01 9.94735584e-02 -1.75559491e-01 -8.10057580e-01
5.36825299e-01 5.17992258e-01 5.40228546e-01 -9.13646221e-01
8.77770782e-01 7.01279283e-01 1.15745820e-01 -9.55393970e-01
-5.95500708e-01 -6.04295135e-01 -9.60810542e-01 3.81244451e-01
9.34179544e-01 -9.20312226e-01 -9.78796959e-01 3.54637146e-01
-1.62871206e+00 -1.53446272e-01 -4.57397789e-01 2.11246490e-01
-3.91956747e-01 5.58831654e-02 -3.93052489e-01 -3.98732603e-01
-4.80386257e-01 -1.25228238e+00 1.43259025e+00 3.24322790e-01
4.41199541e-01 -6.99301660e-01 -3.22979808e-01 -1.35610983e-01
3.87359798e-01 4.13977146e-01 1.05927360e+00 -5.57646394e-01
-1.18360615e+00 -9.76170152e-02 -7.69232750e-01 3.24969858e-01
1.55631170e-01 -9.23719034e-02 -7.86476016e-01 -6.02502376e-02
-1.51469931e-01 -1.05179502e-02 8.48686814e-01 4.90699112e-01
1.95487356e+00 -6.81367069e-02 -4.75470126e-01 1.38984406e+00
1.52585101e+00 -2.32247841e-02 5.13053656e-01 1.33749589e-01
1.22616255e+00 1.34630099e-01 1.52744919e-01 2.82885879e-01
4.88958895e-01 3.91207725e-01 7.82187819e-01 -2.48400927e-01
-6.08454160e-02 -3.15435171e-01 -2.48931080e-01 8.86232436e-01
-1.49411708e-01 3.26443702e-01 -9.46201265e-01 4.12206829e-01
-1.80728614e+00 -5.98797143e-01 -4.42299128e-01 1.91275930e+00
3.33183348e-01 1.18029550e-01 -3.91877592e-01 -2.76621938e-01
6.97873950e-01 -8.95238575e-03 -6.24481499e-01 -4.42922562e-01
-4.32081148e-02 8.85638237e-01 8.64706576e-01 1.95658371e-01
-1.35190713e+00 9.81165051e-01 5.36768961e+00 9.44485366e-01
-1.17694664e+00 -6.32214323e-02 3.37272078e-01 2.65311778e-01
-2.47529984e-01 -1.73483059e-01 -8.49652290e-01 3.08611065e-01
2.04349697e-01 3.87393236e-01 2.81930238e-01 1.18991137e+00
-1.05465852e-01 4.25410151e-01 -1.07263267e+00 1.28132057e+00
-3.27427208e-01 -1.86930299e+00 2.02764541e-01 -1.62011489e-01
7.74311543e-01 5.67883134e-01 -2.41611227e-02 1.46976471e-01
1.77241445e-01 -1.17222118e+00 6.20474696e-01 3.71248424e-01
9.61470723e-01 -1.01118970e+00 4.99923050e-01 2.49854058e-01
-1.73632669e+00 9.80528742e-02 -9.62359905e-01 8.32383782e-02
-1.61034957e-01 5.62032223e-01 -6.52012527e-01 5.74699402e-01
9.86381114e-01 9.07694340e-01 -3.33205104e-01 1.34183025e+00
1.23092517e-01 3.49650830e-01 -8.10225070e-01 -3.99976373e-02
4.00085658e-01 -2.31472850e-01 3.01324636e-01 1.18537700e+00
3.96376759e-01 2.26647973e-01 4.41940159e-01 1.41059089e+00
-1.96189851e-01 -3.78259979e-02 -6.69633329e-01 2.47835085e-01
6.81725860e-01 1.24676776e+00 -9.02625740e-01 -1.14775300e-01
-6.13647878e-01 8.75576258e-01 4.42276210e-01 1.58890814e-01
-5.29408097e-01 -5.75310111e-01 1.21865070e+00 2.79589854e-02
9.04826462e-01 -5.48559606e-01 -6.96332753e-01 -9.52573538e-01
2.35892415e-01 -3.04449618e-01 -2.60982327e-02 -5.49227417e-01
-1.50430274e+00 5.83652616e-01 -9.86580085e-03 -1.33443868e+00
4.65666294e-01 -8.96553040e-01 -9.06551063e-01 1.05894208e+00
-1.87573934e+00 -1.53625643e+00 -5.79345644e-01 7.29131758e-01
5.72208524e-01 5.75722568e-02 7.37354577e-01 4.61231083e-01
-1.25679657e-01 4.15896684e-01 9.03480798e-02 2.82414496e-01
1.72053725e-02 -1.18594205e+00 1.00108242e+00 3.26902926e-01
6.79622069e-02 7.09118664e-01 -7.38019794e-02 -6.75732315e-01
-1.73243546e+00 -1.71488261e+00 4.20289308e-01 -2.86230862e-01
2.54784584e-01 -6.23719275e-01 -1.03757799e+00 6.02847397e-01
-3.23403120e-01 7.64536142e-01 3.79619986e-01 -4.04797606e-02
-5.38197875e-01 -5.04392758e-02 -1.36077249e+00 2.13308305e-01
1.58816028e+00 -3.66393328e-01 -4.44639921e-01 4.06898081e-01
1.30254793e+00 -9.24030006e-01 -9.01911736e-01 7.33884156e-01
2.39635199e-01 -7.45101929e-01 1.48860574e+00 -6.46675825e-01
2.35927761e-01 -4.55163270e-01 -4.76871908e-01 -1.03378308e+00
-6.52185678e-01 -1.14057206e-01 -1.88638214e-02 1.03359628e+00
1.61446542e-01 -6.29423082e-01 1.14736402e+00 4.38648343e-01
-7.11658061e-01 -9.73670006e-01 -1.00048232e+00 -8.21872652e-01
1.12381317e-01 -7.42743194e-01 1.29892123e+00 9.10654306e-01
-7.65874386e-01 -1.47717550e-01 3.82551521e-01 6.52578652e-01
6.78746641e-01 6.44720256e-01 7.34777749e-01 -1.50519264e+00
1.02016501e-01 -4.37012702e-01 -7.36256778e-01 -1.47094607e+00
2.40837872e-01 -1.24218261e+00 -1.16590880e-01 -1.62381399e+00
-2.81201273e-01 -1.00193393e+00 -2.35108733e-01 6.50286555e-01
1.04777180e-01 2.26340026e-01 1.58260718e-01 7.81216919e-02
-4.16277140e-01 5.37047148e-01 1.41626179e+00 -2.79116839e-01
-4.24098462e-01 1.74914852e-01 -3.45970213e-01 8.81678760e-01
6.88905954e-01 -2.87666261e-01 -2.36965090e-01 -9.50290740e-01
-1.40866533e-01 -4.02772576e-01 6.35919094e-01 -1.15427995e+00
3.75624806e-01 -2.34725341e-01 7.17978120e-01 -1.37329841e+00
5.48877478e-01 -1.15605879e+00 1.59780607e-01 -3.94778093e-03
1.60766423e-01 1.60590053e-01 6.21585965e-01 4.86855805e-01
-1.01200551e-01 -1.44590409e-02 4.52900410e-01 -2.31231198e-01
-8.27689409e-01 8.34433794e-01 4.36828047e-01 -2.84473181e-01
7.68978894e-01 -3.77637208e-01 -3.66789699e-02 3.64092410e-01
-5.50168812e-01 2.95953929e-01 4.79045004e-01 3.78451943e-01
1.12667477e+00 -1.82463336e+00 -3.60033214e-01 6.50802910e-01
1.48109943e-01 9.71939504e-01 2.80021638e-01 3.62793773e-01
-9.26502347e-01 3.16031903e-01 -2.52879173e-01 -1.11308479e+00
-1.08500373e+00 4.06674802e-01 1.72656775e-01 2.15144962e-01
-1.03432739e+00 1.11058664e+00 2.21189782e-01 -1.07251120e+00
2.85123378e-01 -1.17787325e+00 6.78123906e-02 -5.98093629e-01
2.51201093e-01 7.49734044e-02 4.04074699e-01 -5.64022839e-01
-4.28307235e-01 9.74341035e-01 9.43458900e-02 6.81485713e-01
1.70889950e+00 4.72135007e-01 -4.86078590e-01 7.44768530e-02
1.44056094e+00 -2.00328559e-01 -1.29075849e+00 -4.55353618e-01
-6.49377629e-02 -7.66641021e-01 -3.74612026e-02 -3.19581866e-01
-1.13931060e+00 7.93954074e-01 5.57826877e-01 8.04001614e-02
8.85563254e-01 2.50201881e-01 1.02736461e+00 7.10049450e-01
4.25541729e-01 -6.44713938e-01 -4.14621681e-01 8.75747919e-01
1.00600946e+00 -1.10201669e+00 -1.62857309e-01 -1.00109708e+00
8.78293253e-03 1.36645138e+00 7.46910095e-01 -5.66270828e-01
1.09244204e+00 4.50632334e-01 -3.48208487e-01 -5.13471723e-01
-3.20283562e-01 -2.06601575e-01 5.06158173e-01 7.68915415e-01
9.87404138e-02 2.64948010e-01 3.06371003e-01 6.90143645e-01
-1.98315859e-01 -1.22012556e-01 -2.37294003e-01 1.06512904e+00
-4.34933513e-01 -1.13305652e+00 -5.00936925e-01 7.62422979e-01
1.94702577e-02 -1.21135876e-01 -1.00351967e-01 6.66799366e-01
5.04290104e-01 2.20608220e-01 6.54214382e-01 -3.68262440e-01
7.10110426e-01 -7.42759556e-02 3.78727496e-01 -7.40423799e-01
-6.25638902e-01 -1.91006333e-01 -4.70516413e-01 -7.81874776e-01
-3.23552758e-01 -4.80535328e-01 -1.63176572e+00 -3.43188018e-01
-1.26824528e-01 -2.52467185e-01 9.83959138e-01 6.12707555e-01
8.47566366e-01 6.83805645e-01 5.43348372e-01 -1.30411780e+00
-3.38723481e-01 -6.58686936e-01 -3.41447592e-01 4.57868278e-01
2.66495973e-01 -6.21105075e-01 8.83812159e-02 -2.06384644e-01] | [7.946135520935059, -3.595989942550659] |
9ca819e1-cf1f-4e88-a16e-65f1652ce3bf | two-person-graph-convolutional-network-for | 2208.06174 | null | https://arxiv.org/abs/2208.06174v2 | https://arxiv.org/pdf/2208.06174v2.pdf | Two-person Graph Convolutional Network for Skeleton-based Human Interaction Recognition | Graph convolutional networks (GCNs) have been the predominant methods in skeleton-based human action recognition, including human-human interaction recognition. However, when dealing with interaction sequences, current GCN-based methods simply split the two-person skeleton into two discrete graphs and perform graph convolution separately as done for single-person action classification. Such operations ignore rich interactive information and hinder effective spatial inter-body relationship modeling. To overcome the above shortcoming, we introduce a novel unified two-person graph to represent inter-body and intra-body correlations between joints. Experiments show accuracy improvements in recognizing both interactions and individual actions when utilizing the proposed two-person graph topology. In addition, We design several graph labeling strategies to supervise the model to learn discriminant spatial-temporal interactive features. Finally, we propose a two-person graph convolutional network (2P-GCN). Our model achieves state-of-the-art results on four benchmarks of three interaction datasets: SBU, interaction subsets of NTU-RGB+D and NTU-RGB+D 120. | ['Jingyong Su', 'Tong Zhang', 'Linlin Tang', 'Yueran Li', 'Zhengcen Li'] | 2022-08-12 | null | null | null | null | ['action-classification', 'human-interaction-recognition'] | ['computer-vision', 'computer-vision'] | [ 2.24635869e-01 1.52365401e-01 -3.81261706e-01 -3.43698502e-01
-5.28786518e-02 -3.98301333e-02 4.96673733e-01 -2.90139318e-01
-4.29924816e-01 4.04983878e-01 4.23568875e-01 9.81297344e-02
8.13323408e-02 -8.08009565e-01 -5.20370603e-01 -3.67578685e-01
-1.83714181e-01 6.71912432e-01 5.51969528e-01 -3.88093233e-01
-1.54984608e-01 5.07803023e-01 -1.29274046e+00 3.50235760e-01
5.87572575e-01 7.80994117e-01 -4.75254834e-01 8.62153947e-01
8.07291120e-02 1.03109896e+00 -4.68806714e-01 -4.70956951e-01
2.91757613e-01 -7.90450394e-01 -8.53505075e-01 4.34426904e-01
5.86182892e-01 -3.48194122e-01 -1.04507220e+00 7.76258230e-01
6.21488810e-01 5.32585680e-01 6.79745793e-01 -1.59211409e+00
-4.41931754e-01 4.90282893e-01 -9.24601436e-01 6.17153905e-02
9.89770949e-01 4.38993037e-01 8.52328122e-01 -3.20522934e-01
6.52617514e-01 1.58529949e+00 7.42027819e-01 6.03119433e-01
-9.10150945e-01 -4.47814018e-01 4.40293849e-01 4.57520843e-01
-1.11529613e+00 2.71286219e-02 8.85803938e-01 -3.75816166e-01
1.14864635e+00 7.49063492e-02 1.34208477e+00 1.30232179e+00
3.59626591e-01 1.21923888e+00 4.95940447e-01 -3.20890963e-01
-2.19889820e-01 -1.15874565e+00 3.55992079e-01 1.28315365e+00
7.59317800e-02 -1.48612320e-01 -7.92825699e-01 1.16329707e-01
1.10587668e+00 1.42028585e-01 -2.10205689e-01 -7.32989728e-01
-1.31685185e+00 5.10641754e-01 5.25689602e-01 5.43737076e-02
-2.39752755e-01 6.10800087e-01 7.66102374e-01 4.01679128e-02
2.85621285e-01 -2.74859935e-01 -7.60997310e-02 -3.90123338e-01
-4.33317542e-01 3.27492744e-01 7.43440628e-01 9.31542695e-01
3.54850918e-01 -9.74830166e-02 -4.02643561e-01 7.66469479e-01
1.77335918e-01 2.29638740e-01 3.16017956e-01 -7.70849705e-01
5.69595158e-01 1.04237700e+00 -2.25185841e-01 -1.18149757e+00
-7.64741659e-01 -1.22124571e-02 -1.01852524e+00 2.83409636e-02
7.49456167e-01 1.28481105e-01 -1.24774623e+00 1.37515414e+00
4.16339606e-01 2.97110118e-02 -3.79617065e-01 1.12730694e+00
1.05269933e+00 2.70744823e-02 1.60149068e-01 3.36902738e-01
1.29609609e+00 -1.61774838e+00 -7.10314155e-01 -3.01599145e-01
7.37865448e-01 -1.26391321e-01 1.01208389e+00 2.69070894e-01
-8.86949956e-01 -9.24879730e-01 -8.97052884e-01 -3.09893727e-01
-2.44130149e-01 3.40209901e-01 1.00010157e+00 5.26003301e-01
-6.70839667e-01 7.07306743e-01 -1.10849202e+00 -6.46494150e-01
4.62882280e-01 4.57977027e-01 -7.80134857e-01 -4.70256470e-02
-1.16447532e+00 6.39831543e-01 3.30616742e-01 3.32937032e-01
-6.47221446e-01 1.04959689e-01 -1.17481089e+00 -2.12722361e-01
5.56101978e-01 -9.99358356e-01 9.30759549e-01 -6.83438540e-01
-1.55599010e+00 9.74594533e-01 1.20305024e-01 -3.79127622e-01
8.99264574e-01 -4.77341533e-01 -4.13818091e-01 1.65996358e-01
-3.44048403e-02 5.84043026e-01 6.44790173e-01 -7.20700979e-01
-2.74270505e-01 -6.58186495e-01 2.51425773e-01 4.44387227e-01
-6.13567308e-02 -1.16695292e-01 -1.01805019e+00 -7.22482026e-01
2.84015715e-01 -1.12782121e+00 -2.27648735e-01 2.95252055e-01
-7.02104628e-01 -4.49504286e-01 8.49007785e-01 -8.30486298e-01
1.16362870e+00 -1.80566657e+00 6.48804963e-01 9.78914350e-02
4.31116104e-01 2.12489620e-01 -2.20087886e-01 3.15745741e-01
-1.49954125e-01 -4.37081665e-01 -5.59703670e-02 -4.87055898e-01
7.42683783e-02 5.20957291e-01 5.39139271e-01 7.15220690e-01
-2.27598831e-01 1.31088567e+00 -9.66435730e-01 -6.65778816e-01
4.87435460e-01 4.76167947e-01 -3.28234315e-01 1.40032589e-01
-1.14030786e-01 7.54445672e-01 -4.87722397e-01 7.48984277e-01
2.79651999e-01 -2.57617891e-01 3.89651358e-01 -3.28316838e-01
5.24929941e-01 -1.01342350e-01 -1.25039542e+00 2.18304920e+00
7.37168193e-02 3.04828435e-01 -1.31233230e-01 -1.03417563e+00
6.60417080e-01 -2.98543144e-02 8.43460500e-01 -5.01067877e-01
3.78951758e-01 -3.30673993e-01 1.68506339e-01 -5.15100777e-01
2.17083573e-01 2.56502658e-01 -1.06637269e-01 4.04200941e-01
2.33648464e-01 1.71887264e-01 4.25046742e-01 2.59918898e-01
1.35990214e+00 6.62423670e-01 4.23503160e-01 7.25511909e-02
6.18849456e-01 -3.97788227e-01 5.12767315e-01 4.83334005e-01
-6.68272674e-01 7.06629813e-01 6.42152369e-01 -8.40668857e-01
-4.04259562e-01 -8.96460414e-01 6.87667727e-01 1.09263432e+00
2.40933642e-01 -7.19265282e-01 -8.21354032e-01 -1.20461583e+00
1.83880907e-02 1.70952156e-01 -8.01178336e-01 -2.93074876e-01
-9.05981541e-01 -3.29408795e-01 7.88748205e-01 1.03221047e+00
9.85494196e-01 -1.21786106e+00 -5.69934726e-01 7.23155439e-02
-4.90040570e-01 -1.30345547e+00 -8.52262735e-01 -2.84480095e-01
-6.23493910e-01 -1.59115434e+00 -7.94056296e-01 -5.94131708e-01
5.27248263e-01 7.21813887e-02 1.18649948e+00 1.09092705e-01
-5.15518367e-01 8.03969204e-01 -4.50775295e-01 1.11449473e-01
1.03220992e-01 -5.72107881e-02 6.63020909e-02 1.15550891e-01
3.86441380e-01 -7.15050161e-01 -7.29393005e-01 5.48180163e-01
-4.20665890e-01 1.96323618e-01 3.84539366e-01 7.93990076e-01
4.77333695e-01 8.30077007e-03 -1.43738583e-01 -6.15774155e-01
4.65576440e-01 1.47004992e-01 5.96628413e-02 4.44484502e-01
3.85924280e-02 -1.70829117e-01 3.69830817e-01 -5.09743750e-01
-9.24602926e-01 4.52031076e-01 3.53810303e-02 -5.66556573e-01
-2.75993705e-01 2.62853205e-01 -4.71314847e-01 -2.93624878e-01
6.20793760e-01 1.35717750e-01 6.69017956e-02 -2.79600024e-01
5.69589019e-01 1.34562477e-01 7.04595804e-01 -6.66341305e-01
3.94089818e-01 5.08045018e-01 2.54011273e-01 -8.23663771e-01
-5.83619237e-01 -6.68194830e-01 -1.38620520e+00 -7.13287652e-01
1.41296160e+00 -7.47074246e-01 -9.98534262e-01 1.09405088e+00
-1.16219866e+00 -6.98443651e-01 -1.73455536e-01 4.75899011e-01
-8.15796793e-01 1.01580131e+00 -7.83029914e-01 -7.39110589e-01
-1.89112410e-01 -1.03973138e+00 1.42940474e+00 -1.95805505e-02
-4.75217074e-01 -9.02513027e-01 3.18817198e-02 7.84681678e-01
-2.26517186e-01 7.61864543e-01 4.75495517e-01 -4.17642474e-01
-2.96212614e-01 -3.81716907e-01 -1.18737869e-01 2.59333462e-01
2.28641063e-01 -2.42546141e-01 -2.69202948e-01 -1.49198517e-01
-6.69126451e-01 -5.14987290e-01 1.00829113e+00 3.64855826e-01
1.24089837e+00 1.43484801e-01 -5.45856893e-01 5.81801176e-01
5.84986866e-01 -3.40890437e-02 8.76674950e-01 5.71418591e-02
1.45504212e+00 5.20118177e-01 5.17058790e-01 5.08687854e-01
4.22388852e-01 8.65556061e-01 2.43576258e-01 -1.20914549e-01
-5.57185233e-01 -3.63241136e-01 3.45148802e-01 4.47824448e-01
-9.02671814e-01 -3.67988348e-01 -9.37375486e-01 1.05537139e-01
-2.28107429e+00 -8.54370952e-01 -5.11936069e-01 1.83841705e+00
3.54792356e-01 3.49117905e-01 6.89419925e-01 1.06431685e-01
6.96792305e-01 4.80327755e-01 -5.35971344e-01 1.71226025e-01
-7.15176612e-02 1.80418581e-01 4.33218926e-01 1.43488377e-01
-1.40505779e+00 1.02373648e+00 6.02385044e+00 6.24291480e-01
-5.18636346e-01 -1.54115453e-01 2.14533314e-01 6.11636899e-02
4.43796694e-01 -2.49525920e-01 -4.56937432e-01 1.47421449e-01
4.00170863e-01 2.96005756e-01 2.52978593e-01 7.44142473e-01
1.19539611e-01 -2.03365535e-01 -1.27536500e+00 1.27245820e+00
2.31695876e-01 -7.40060508e-01 1.37764946e-01 -7.41890818e-02
3.66282195e-01 -4.82109219e-01 -4.64745343e-01 3.99646133e-01
3.97869736e-01 -9.75945890e-01 5.70321977e-01 6.84381783e-01
7.21909463e-01 -6.41657531e-01 4.58370954e-01 2.01838136e-01
-1.76726437e+00 1.62284017e-01 3.32305469e-02 -3.45384598e-01
4.88153517e-01 2.19517916e-01 -1.64196864e-01 9.75163400e-01
7.23452091e-01 1.16923261e+00 -6.99554682e-01 6.77062154e-01
-4.92709339e-01 2.38636285e-01 -1.66312814e-01 1.28086925e-01
9.37959179e-02 -3.45138997e-01 5.06036520e-01 1.02002704e+00
-1.18475929e-01 3.36280257e-01 7.01956451e-01 4.59383070e-01
1.50874704e-01 -9.92124677e-02 -5.26120722e-01 -2.72679836e-01
-4.44905967e-01 9.16047692e-01 -1.07778072e+00 -3.56469184e-01
-4.88906801e-01 1.68807280e+00 4.42668915e-01 3.73919725e-01
-1.07092524e+00 -2.17676982e-01 6.17936015e-01 1.72902465e-01
4.29632999e-02 -6.96706533e-01 9.68204364e-02 -1.42587137e+00
1.30162045e-01 -9.98009384e-01 7.35986650e-01 -6.59399509e-01
-1.13278604e+00 1.97374240e-01 2.10933179e-01 -1.08546507e+00
-7.32032955e-02 -8.94252777e-01 -5.63141108e-01 4.11557347e-01
-4.77930486e-01 -1.70733666e+00 -6.97256923e-01 1.04870582e+00
5.41246891e-01 -2.56763604e-02 6.79466784e-01 3.15657079e-01
-7.12089121e-01 6.40386939e-01 -8.20166290e-01 8.21492672e-01
5.83349943e-01 -1.24600303e+00 8.08518946e-01 7.51412451e-01
1.68456644e-01 5.45449495e-01 3.07428688e-01 -1.05010450e+00
-1.34469688e+00 -9.17828500e-01 5.45890570e-01 -4.23545122e-01
5.41243851e-01 -4.02064949e-01 -4.95533973e-01 1.06910956e+00
-4.90114465e-02 3.66199911e-01 5.55694878e-01 1.96115240e-01
-4.00329411e-01 2.35036910e-01 -8.44303966e-01 9.32677090e-01
2.18027401e+00 -4.13523495e-01 -3.77259582e-01 6.13870382e-01
3.63523483e-01 -7.09208012e-01 -8.39588404e-01 4.52403277e-01
8.56774032e-01 -1.12689352e+00 1.10137463e+00 -9.44176257e-01
2.99401850e-01 -3.36348653e-01 -5.13835019e-03 -1.04813969e+00
-3.76913577e-01 -5.89863420e-01 -4.46907818e-01 5.33115864e-01
-2.75993854e-01 -4.03896093e-01 1.26707864e+00 4.62322801e-01
-1.92010179e-01 -7.49798834e-01 -8.12981069e-01 -8.80759358e-01
-4.76753026e-01 -4.52552021e-01 2.84923285e-01 6.93880081e-01
1.82437584e-01 3.71378481e-01 -7.24585235e-01 -3.01707476e-01
8.16674531e-01 -1.60542399e-01 1.37886703e+00 -1.13073254e+00
-6.54298842e-01 -3.85761321e-01 -1.18439567e+00 -1.49485636e+00
4.68963206e-01 -7.47601688e-01 -9.68815982e-02 -1.88054609e+00
2.88745552e-01 2.81387687e-01 -1.14090249e-01 7.52306640e-01
-1.30335391e-01 2.46344656e-01 1.82546258e-01 -7.89610520e-02
-8.94760668e-01 7.15548396e-01 1.72178447e+00 -3.80709618e-01
-3.53493020e-02 5.03875688e-02 8.43555406e-02 8.01133871e-01
4.18914407e-01 8.42952728e-02 -5.93255520e-01 -8.68901238e-02
-3.01824778e-01 1.74023941e-01 7.72562265e-01 -1.45179689e+00
2.72591263e-01 -1.55577585e-01 5.55501640e-01 -7.62332976e-01
4.64271158e-01 -5.84812820e-01 2.91544497e-01 6.02282405e-01
-2.28984475e-01 -1.51587203e-01 -1.64277211e-01 9.14281547e-01
-2.26797804e-01 5.35491705e-01 6.21819615e-01 -4.19451445e-01
-9.90790486e-01 6.85647488e-01 -2.07372442e-01 -9.58434939e-02
1.02964818e+00 -5.55617154e-01 -4.48593572e-02 -5.97585201e-01
-1.28196824e+00 3.05219680e-01 2.21363321e-01 6.16771460e-01
6.93695128e-01 -1.55618536e+00 -3.61658990e-01 7.18506873e-02
2.68814921e-01 -1.13766946e-01 5.18032908e-01 1.07973635e+00
-7.17813015e-01 3.20649773e-01 -5.86731434e-01 -5.49437761e-01
-1.58376336e+00 3.42994809e-01 5.69213092e-01 -5.14559984e-01
-1.02500474e+00 9.66159761e-01 1.90207705e-01 -5.90949655e-01
4.47359025e-01 -3.51874799e-01 -8.37241784e-02 -2.16779649e-01
1.26895696e-01 7.43850708e-01 -2.56744117e-01 -9.44900215e-01
-4.50188488e-01 6.03330135e-01 2.03633219e-01 4.71283495e-02
9.27510321e-01 1.85537800e-01 -1.35509208e-01 3.72843534e-01
1.04986274e+00 -5.47906339e-01 -1.30920494e+00 1.70339905e-02
-2.65638381e-01 -5.28663516e-01 -4.61987883e-01 -4.84766692e-01
-1.32416058e+00 8.54440272e-01 4.32774365e-01 -1.49010122e-01
9.60358083e-01 -1.52945533e-01 1.11362612e+00 3.93974066e-01
6.10594451e-01 -1.11593258e+00 6.87076151e-01 5.32606423e-01
1.16164446e+00 -1.04273975e+00 2.89889544e-01 -7.45970249e-01
-6.52219474e-01 1.22525275e+00 1.05355513e+00 -1.02678314e-01
7.13986993e-01 -2.02980921e-01 -1.89652577e-01 -5.27606547e-01
1.01739421e-01 -4.80349749e-01 6.86349511e-01 7.48285174e-01
5.20911753e-01 2.71784306e-01 -5.47802448e-01 4.95870531e-01
4.07794937e-02 9.48143378e-02 2.38988381e-02 1.19744563e+00
-7.65565708e-02 -1.22783208e+00 -9.70569924e-02 3.98834646e-01
-9.05735195e-02 4.56681788e-01 -8.80345643e-01 1.09160316e+00
1.23549886e-01 7.46385098e-01 -2.75437117e-01 -7.44709909e-01
7.99424708e-01 1.45081535e-01 9.37149763e-01 -5.36329687e-01
-4.50972050e-01 -1.99596956e-02 3.65494400e-01 -1.28754902e+00
-6.93282783e-01 -6.62097216e-01 -1.44611907e+00 -3.30008060e-01
-6.77032769e-02 -4.51235384e-01 -7.63647482e-02 9.42059755e-01
1.63057789e-01 7.41705596e-01 -3.19710344e-01 -1.30087185e+00
-1.83598056e-01 -1.08400905e+00 -9.05273914e-01 7.41230130e-01
-1.78819373e-01 -1.17899954e+00 -4.47614715e-02 3.63243818e-02] | [7.933310031890869, 0.39629408717155457] |
ff969025-e3ef-468a-8be0-b693aa537237 | sequence-to-sequence-convolutional-neural | null | null | https://aclanthology.org/2019.rocling-1.12 | https://aclanthology.org/2019.rocling-1.12.pdf | Sequence to Sequence Convolutional Neural Network for Automatic Spelling Correction | null | ['Yuan-Fu Liao', 'Ján Staš', 'Matúš Pleva', 'Daniel Hládek'] | null | null | null | null | rocling-2019-10 | ['spelling-correction'] | ['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] | [-7.294840335845947, 3.683072090148926] |
9dcc72ee-9494-415d-9caa-00faac2b4555 | universal-morphology-control-via-contextual | 2302.11070 | null | https://arxiv.org/abs/2302.11070v1 | https://arxiv.org/pdf/2302.11070v1.pdf | Universal Morphology Control via Contextual Modulation | Learning a universal policy across different robot morphologies can significantly improve learning efficiency and generalization in continuous control. However, it poses a challenging multi-task reinforcement learning problem, as the optimal policy may be quite different across robots and critically depend on the morphology. Existing methods utilize graph neural networks or transformers to handle heterogeneous state and action spaces across different morphologies, but pay little attention to the dependency of a robot's control policy on its morphology context. In this paper, we propose a hierarchical architecture to better model this dependency via contextual modulation, which includes two key submodules: (1) Instead of enforcing hard parameter sharing across robots, we use hypernetworks to generate morphology-dependent control parameters; (2) We propose a morphology-dependent attention mechanism to modulate the interactions between different limbs in a robot. Experimental results show that our method not only improves learning performance on a diverse set of training robots, but also generalizes better to unseen morphologies in a zero-shot fashion. | ['Shimon Whiteson', 'Jacob Beck', 'Zheng Xiong'] | 2023-02-22 | null | null | null | null | ['continuous-control'] | ['playing-games'] | [ 5.96243590e-02 -1.44555017e-01 -1.29970804e-01 -1.31810037e-02
-2.46462896e-01 -7.13345826e-01 4.19463813e-01 5.12511022e-02
-4.99422103e-01 7.12542593e-01 3.38388532e-02 1.52513646e-02
-3.78063619e-01 -7.36355424e-01 -9.62917924e-01 -8.90160739e-01
-4.25629094e-02 4.47297722e-01 4.65265840e-01 -7.01983690e-01
1.23642176e-01 4.40442085e-01 -1.30894279e+00 -3.24556977e-01
1.02681780e+00 7.24027097e-01 6.31962836e-01 4.77401316e-01
2.72857785e-01 6.52363300e-01 -6.32842720e-01 -3.23716551e-02
2.74402052e-01 -3.61661971e-01 -6.47453606e-01 1.53170228e-01
1.84335887e-01 -6.05511889e-02 -4.88372415e-01 1.16414201e+00
4.45040464e-01 2.18187928e-01 8.56372356e-01 -1.25154471e+00
-1.03888679e+00 7.36630976e-01 -4.83377635e-01 -9.63386744e-02
-2.52434820e-01 7.05105722e-01 9.15538609e-01 -9.85464454e-02
4.91042584e-01 1.46180427e+00 4.33849424e-01 7.70569026e-01
-1.24033844e+00 -7.23123431e-01 7.12868869e-01 3.09897304e-01
-1.28336632e+00 -1.05520271e-01 7.73541451e-01 -2.18687728e-01
6.44664049e-01 -4.32593256e-01 8.43423784e-01 1.29100776e+00
5.00515580e-01 5.49627304e-01 7.54075766e-01 -3.83616947e-02
4.14622039e-01 -6.36585891e-01 -5.15613616e-01 9.68763649e-01
4.14655715e-01 5.40478304e-02 -1.57056138e-01 1.79450676e-01
1.19464016e+00 -1.30448490e-01 -3.62919360e-01 -7.17729330e-01
-1.33686161e+00 8.02313268e-01 7.45788336e-01 1.75505936e-01
-2.05401763e-01 6.49728179e-01 2.84761637e-01 4.28339034e-01
-1.92438573e-01 1.05785060e+00 -6.44824147e-01 1.68158755e-01
2.73302108e-01 2.58587360e-01 6.71387970e-01 1.12690747e+00
9.88678932e-01 3.38071764e-01 -1.10667743e-01 1.11695063e+00
2.06224620e-01 5.73073268e-01 5.66816390e-01 -1.07548559e+00
5.09530246e-01 5.93950272e-01 -3.12398523e-02 -9.63122368e-01
-5.96378624e-01 -3.03412318e-01 -7.16927588e-01 3.56698394e-01
2.92094678e-01 -5.55468500e-01 -1.18506944e+00 2.29942703e+00
2.60931373e-01 -1.78069994e-01 5.37846498e-02 8.42222750e-01
4.96383280e-01 3.89566451e-01 2.26063088e-01 1.20930038e-01
9.77677166e-01 -1.18953526e+00 -5.60265303e-01 -7.01308131e-01
5.40670633e-01 -1.67692125e-01 1.10671329e+00 8.40389431e-02
-8.22648942e-01 -3.67441982e-01 -1.28168583e+00 1.69766799e-01
-4.13294971e-01 2.93937828e-02 7.01665878e-01 3.99932899e-02
-1.12007070e+00 6.17078602e-01 -8.53894234e-01 -4.70139056e-01
4.34610575e-01 7.05415845e-01 -3.36396664e-01 1.45517141e-02
-1.00427043e+00 1.13119662e+00 5.64435363e-01 -8.87639895e-02
-1.27951646e+00 -2.06729695e-01 -1.08237362e+00 1.00196056e-01
7.97944665e-01 -9.59394217e-01 1.24450815e+00 -8.72768223e-01
-2.09842157e+00 2.44774476e-01 4.06268418e-01 -4.41305600e-02
7.90786445e-02 8.19327459e-02 4.23337519e-02 9.54316556e-03
-4.30226550e-02 1.06701159e+00 9.71156597e-01 -1.33796155e+00
-6.35928392e-01 -4.33984578e-01 4.14668530e-01 5.57371914e-01
-3.43731940e-01 -6.91780448e-01 -7.44133711e-01 -7.73369670e-01
-5.74837252e-02 -1.21742451e+00 -6.06595278e-01 1.40544742e-01
-1.86949283e-01 -3.67712468e-01 5.71769118e-01 8.89974087e-02
6.31983697e-01 -2.05861688e+00 9.34379876e-01 -2.04780236e-01
4.67697307e-02 6.42555729e-02 -6.83481216e-01 3.74103457e-01
5.50819397e-01 7.31216371e-02 -2.57046282e-01 8.31876621e-02
-1.45047121e-02 7.29841471e-01 -8.05080775e-03 4.58381087e-01
4.62725103e-01 1.04748309e+00 -9.59513664e-01 -4.60835248e-01
1.62755158e-02 2.35414714e-01 -6.87999249e-01 1.61377922e-01
-5.71747959e-01 6.45476580e-01 -8.15667808e-01 5.00074804e-01
2.05050632e-01 -2.64052719e-01 4.18456495e-01 -7.38409087e-02
4.15801555e-02 -1.38200238e-01 -8.67753923e-01 1.96066487e+00
-5.80133080e-01 9.80247706e-02 4.18646753e-01 -1.01874304e+00
8.53915751e-01 -5.22631146e-02 2.76883304e-01 -6.47241890e-01
4.70431775e-01 3.47291194e-02 4.10297811e-01 -4.37314570e-01
3.07526976e-01 9.45110917e-02 -3.47566247e-01 8.98927078e-02
3.10646415e-01 -5.32858789e-01 2.98132241e-01 -2.65971065e-01
1.23610878e+00 3.84559482e-01 3.55931759e-01 -3.37957203e-01
2.45063469e-01 9.94070396e-02 9.90963280e-01 7.52732933e-01
-2.82810241e-01 1.98903799e-01 4.50663626e-01 -1.90455228e-01
-9.11631882e-01 -9.47477400e-01 2.00179964e-01 1.45275092e+00
8.18693399e-01 -1.01662865e-02 -4.92813587e-01 -5.35065830e-01
1.62222162e-01 3.17576408e-01 -8.05082679e-01 -6.49369240e-01
-8.77258599e-01 -6.37007117e-01 3.88699204e-01 5.75484574e-01
4.81506526e-01 -1.29326093e+00 -7.27291942e-01 2.15822339e-01
2.34250575e-01 -1.13778782e+00 -6.87920868e-01 6.82820976e-01
-7.73875356e-01 -1.14472985e+00 -5.20640135e-01 -1.19552946e+00
7.53274620e-01 3.04408729e-01 7.61392176e-01 1.39540270e-01
-8.02433193e-02 4.04320568e-01 -4.61060971e-01 -2.42707551e-01
-3.20331037e-01 5.33189714e-01 3.88346128e-02 -3.54929984e-01
-3.84578943e-01 -8.62605989e-01 -3.97275835e-01 5.66115737e-01
-8.81087720e-01 -4.60078418e-02 1.09958744e+00 9.58434463e-01
5.66642821e-01 -4.68434244e-02 8.11200857e-01 -5.89575112e-01
7.47431278e-01 -4.42222774e-01 -7.37927854e-01 3.18612456e-01
-5.06491661e-01 6.21451557e-01 1.05160725e+00 -9.12829459e-01
-8.36846411e-01 2.40551382e-01 2.07230330e-01 -5.64373851e-01
-1.25589281e-01 2.57366598e-01 -4.34235543e-01 -3.20757091e-01
4.30249929e-01 1.50327429e-01 2.43506998e-01 -2.50291795e-01
5.15869379e-01 1.69754148e-01 5.33283949e-01 -9.32733715e-01
7.86768377e-01 2.79645026e-01 2.60939538e-01 -5.19120038e-01
-5.24207830e-01 2.79764663e-02 -6.79675341e-01 -2.98473909e-02
9.28120077e-01 -7.82974601e-01 -9.01534438e-01 6.23710632e-01
-8.10421407e-01 -1.08388138e+00 -3.78041938e-02 3.49333286e-01
-8.51519525e-01 5.45377769e-02 -7.16707528e-01 -2.39770249e-01
-1.08931206e-01 -1.43321371e+00 9.42796469e-01 4.53810483e-01
3.37755024e-01 -9.54162240e-01 2.20099799e-02 -2.00522408e-01
4.01223779e-01 1.13096409e-01 1.22877491e+00 -3.05434525e-01
-5.28006554e-01 2.50373125e-01 -4.36780341e-02 -4.71096076e-02
3.77341568e-01 -1.49417862e-01 -2.90704787e-01 -4.51781690e-01
-4.25217211e-01 -7.98282504e-01 7.77798533e-01 1.76195621e-01
1.30143464e+00 -2.33577654e-01 -4.17007834e-01 7.52284586e-01
1.17672098e+00 2.90110499e-01 3.55395079e-01 4.74843383e-01
8.84893775e-01 5.26755214e-01 4.35220748e-01 2.67436236e-01
7.37006128e-01 6.58175290e-01 1.04599833e+00 2.92127758e-01
-2.03043759e-01 -3.82321149e-01 4.76487905e-01 8.52486372e-01
1.18467890e-01 -2.57779509e-01 -6.10069752e-01 4.20555264e-01
-2.12819147e+00 -6.03978157e-01 4.57231104e-01 1.86582947e+00
8.78686488e-01 -4.28840481e-02 4.48064022e-02 -5.40315747e-01
8.17433298e-01 1.56354666e-01 -1.17524660e+00 -2.96020210e-02
-9.95328501e-02 -1.33164063e-01 5.90047657e-01 2.31518671e-01
-9.98861432e-01 1.36734831e+00 6.14416790e+00 8.08811247e-01
-1.36291397e+00 -1.56724378e-01 1.54337823e-01 -2.50467146e-03
4.01085056e-03 -1.97912917e-01 -6.33355439e-01 3.51383179e-01
2.83741117e-01 6.95305765e-02 8.35961521e-01 8.77249956e-01
-1.61991775e-01 1.42520428e-01 -9.64243233e-01 7.45728791e-01
-6.31103888e-02 -9.88815486e-01 1.16752893e-01 1.14727736e-01
6.95191264e-01 1.93996817e-01 1.15186125e-01 6.71126544e-01
8.70331168e-01 -9.53474879e-01 9.16757524e-01 3.61811072e-02
5.53036749e-01 -6.91207349e-01 2.34675393e-01 4.58125532e-01
-1.28629792e+00 -4.49156314e-01 -6.47950351e-01 -3.13297622e-02
-1.31560937e-01 -3.29646319e-01 -6.13727748e-01 3.03716213e-01
5.26903152e-01 7.54325926e-01 -6.49367809e-01 1.01005137e+00
-6.44892752e-01 5.66117242e-02 -1.46319404e-01 -5.45480847e-01
2.53328621e-01 -1.26675054e-01 4.08515930e-01 7.09480345e-01
1.81125402e-01 -3.00846081e-02 5.12214303e-01 6.63664162e-01
-2.03115314e-01 -1.44926682e-01 -6.89686179e-01 -2.18490860e-03
7.86540091e-01 1.33184814e+00 -7.61030793e-01 1.14459276e-01
-2.57789880e-01 8.42166841e-01 1.00382960e+00 4.94032115e-01
-8.76496315e-01 -3.93031120e-01 7.43906021e-01 -4.15407091e-01
8.17874968e-01 -5.63138306e-01 -9.31322500e-02 -1.03899813e+00
-2.50087082e-01 -7.70777047e-01 2.26117879e-01 -5.12415946e-01
-1.34707165e+00 3.94019067e-01 -9.73778516e-02 -1.16233778e+00
-6.21286407e-02 -7.90229857e-01 -5.24869502e-01 2.53891557e-01
-1.43634462e+00 -1.24204695e+00 -1.70722350e-01 6.27815425e-01
5.75940371e-01 -2.33302757e-01 6.21298492e-01 -3.10931467e-02
-8.41225386e-01 4.95333284e-01 -1.18546091e-01 -1.82282906e-02
6.95490837e-01 -1.31529617e+00 9.69588608e-02 4.30169880e-01
-1.34127513e-01 5.40611863e-01 5.55473506e-01 -5.05605698e-01
-1.86602664e+00 -1.42782056e+00 -1.71236828e-01 -1.55245081e-01
7.98775733e-01 -2.29865581e-01 -8.46226096e-01 6.45892620e-01
2.70600796e-01 5.47269126e-03 3.19671392e-01 2.19577178e-01
-4.80698377e-01 -1.22453064e-01 -8.14358056e-01 1.01511610e+00
1.44384921e+00 -1.14246577e-01 -5.64108193e-01 7.67921954e-02
8.58626187e-01 -4.07860279e-01 -8.38590980e-01 6.44322157e-01
3.99560779e-01 -4.19506490e-01 6.63957357e-01 -7.33463228e-01
4.10098135e-01 -5.13443649e-01 -1.89269390e-02 -1.85450637e+00
-7.77247906e-01 -4.54429239e-01 3.78487259e-02 8.26551557e-01
5.10816872e-01 -7.22779572e-01 5.11482596e-01 1.11480221e-01
-5.54328144e-01 -1.00776398e+00 -8.00653994e-01 -9.18641388e-01
4.78351206e-01 1.46315157e-01 5.86389780e-01 8.32784891e-01
-1.16531197e-02 8.17864597e-01 -3.42086166e-01 3.69041979e-01
2.20926464e-01 2.32373908e-01 9.34944391e-01 -1.08837318e+00
-4.62278873e-01 -8.04105222e-01 -1.69745013e-01 -1.32505631e+00
4.55934703e-01 -8.15508187e-01 7.59669662e-01 -1.42492878e+00
7.56367594e-02 -6.11507356e-01 -2.05003127e-01 7.47026145e-01
-2.96045095e-01 -1.27181754e-01 2.77235836e-01 2.66685367e-01
-1.03117406e+00 1.06014717e+00 1.90863872e+00 -2.14746758e-01
-2.04647392e-01 -2.74638355e-01 -8.55941236e-01 6.20893717e-01
9.96151984e-01 -3.42354834e-01 -6.89606011e-01 -9.86024439e-01
7.99070299e-02 1.51044037e-02 1.30608857e-01 -1.13756788e+00
3.94212663e-01 -6.23679638e-01 2.26196602e-01 1.25505432e-01
3.53654444e-01 -6.28947318e-01 -2.60178804e-01 6.61362827e-01
-3.14046025e-01 3.54460210e-01 1.63152963e-01 1.12340927e+00
1.88975483e-01 -1.85434729e-01 9.31305647e-01 -2.64912754e-01
-8.18059206e-01 6.24689877e-01 -5.74967980e-01 2.10216671e-01
1.09314454e+00 1.28599539e-01 -6.04594886e-01 -1.98942438e-01
-3.34976166e-01 8.11011970e-01 8.55300128e-01 6.81493104e-01
2.88832068e-01 -1.24230552e+00 -2.63646513e-01 -5.54941744e-02
2.53737330e-01 4.04289395e-01 3.66036110e-02 4.14804012e-01
-3.15924108e-01 1.34547189e-01 -6.12103343e-01 -5.11787653e-01
-7.02021658e-01 6.13757670e-01 3.99407506e-01 -6.13369718e-02
-4.04476255e-01 9.21335995e-01 4.98957813e-01 -6.26876652e-01
2.50975668e-01 -3.09968114e-01 -4.34728593e-01 -2.92365670e-01
2.57309638e-02 1.77539680e-02 -4.30786669e-01 -5.01468837e-01
-1.01092152e-01 8.50700080e-01 -1.73326179e-01 8.65842104e-02
1.31613541e+00 -4.70747799e-02 1.30165964e-01 3.28985423e-01
7.70875096e-01 -3.21774691e-01 -1.89032197e+00 -7.72314593e-02
-1.43847704e-01 -8.04572627e-02 -1.95382789e-01 -6.34369612e-01
-1.21028209e+00 5.21654308e-01 1.75288841e-01 1.39147207e-01
9.61113214e-01 1.89336300e-01 5.38061559e-01 9.09825325e-01
7.21510172e-01 -1.28231537e+00 6.70366049e-01 9.39289451e-01
1.04330707e+00 -1.24945974e+00 -1.61598384e-01 -2.50866354e-01
-7.31300116e-01 1.07432914e+00 1.21519184e+00 -3.35051179e-01
3.20846379e-01 1.04647473e-01 -1.36242611e-02 -1.81028411e-01
-1.02777731e+00 -5.25818884e-01 -6.66186810e-02 9.10358071e-01
1.06903957e-02 -1.05193786e-01 -1.19383916e-01 3.86553496e-01
-6.10678270e-02 -3.90993059e-01 3.78877670e-01 1.00724304e+00
-7.59499550e-01 -1.22161174e+00 4.44446802e-02 3.06788445e-01
-6.92069232e-02 3.60639930e-01 -4.47421908e-01 1.00952256e+00
1.66960403e-01 7.78019011e-01 -1.46576539e-01 -6.54052913e-01
3.29072326e-01 -3.81168872e-01 8.24057221e-01 -9.62561727e-01
-4.11487103e-01 -9.75122675e-02 -2.32742652e-01 -4.13166344e-01
-1.53439000e-01 -5.45345008e-01 -1.46365976e+00 8.06610286e-02
-2.35542804e-01 -1.14160404e-01 4.74284738e-01 9.42231178e-01
6.71121478e-01 8.34441662e-01 4.58066106e-01 -1.03164423e+00
-8.74375939e-01 -9.63006079e-01 -4.13584650e-01 3.48051816e-01
3.28029543e-01 -1.27312160e+00 -1.22804314e-01 -1.72613859e-01] | [4.3846893310546875, 1.0477122068405151] |
dfea1615-5e53-438b-9243-edb47947457e | understanding-autoencoders-with-information | 1804.00057 | null | https://arxiv.org/abs/1804.00057v3 | https://arxiv.org/pdf/1804.00057v3.pdf | Understanding Autoencoders with Information Theoretic Concepts | Despite their great success in practical applications, there is still a lack of theoretical and systematic methods to analyze deep neural networks. In this paper, we illustrate an advanced information theoretic methodology to understand the dynamics of learning and the design of autoencoders, a special type of deep learning architectures that resembles a communication channel. By generalizing the information plane to any cost function, and inspecting the roles and dynamics of different layers using layer-wise information quantities, we emphasize the role that mutual information plays in quantifying learning from data. We further suggest and also experimentally validate, for mean square error training, three fundamental properties regarding the layer-wise flow of information and intrinsic dimensionality of the bottleneck layer, using respectively the data processing inequality and the identification of a bifurcation point in the information plane that is controlled by the given data. Our observations have a direct impact on the optimal design of autoencoders, the design of alternative feedforward training methods, and even in the problem of generalization. | ['Jose C. Principe', 'Shujian Yu'] | 2018-03-30 | null | null | null | null | ['information-plane'] | ['methodology'] | [-4.11469080e-02 3.11098605e-01 1.94064915e-01 -6.33077472e-02
3.10000181e-01 -4.62391615e-01 6.45489097e-01 3.07900429e-01
-6.53074205e-01 6.28058851e-01 2.96724271e-02 -4.99751151e-01
-8.70640993e-01 -5.77964842e-01 -6.52523398e-01 -1.06642914e+00
-4.22481686e-01 1.57079548e-01 9.57625732e-02 -2.58250892e-01
2.81858683e-01 6.11988783e-01 -1.41719568e+00 -1.92619741e-01
4.35523152e-01 1.16324091e+00 1.40723050e-01 6.64013982e-01
-5.90449125e-02 5.93493283e-01 -5.51138520e-01 -4.26825255e-01
2.75670320e-01 -6.61926687e-01 -9.04797137e-01 -3.88345234e-02
-2.14609936e-01 -1.81397900e-01 -4.06327933e-01 1.02508283e+00
4.41394597e-01 -7.32313320e-02 8.54818821e-01 -9.13575768e-01
-4.09414440e-01 8.26148570e-01 9.74851921e-02 6.03217900e-01
-4.22573030e-01 -2.35759214e-01 9.50071156e-01 -5.49242079e-01
4.66164589e-01 7.78706729e-01 5.87547421e-01 5.62645316e-01
-1.16923106e+00 -4.78294611e-01 -1.19917452e-01 1.66837305e-01
-1.23662519e+00 -3.58659476e-01 7.87911415e-01 -7.58588135e-01
3.54834527e-01 -2.49159902e-01 7.29892671e-01 6.64444745e-01
3.62505049e-01 5.85747242e-01 9.37856376e-01 -7.56417692e-01
5.14232278e-01 4.67224658e-01 2.77233392e-01 6.85803354e-01
4.17579174e-01 3.01959962e-01 -5.08821249e-01 1.98501781e-01
9.47904170e-01 -3.08483213e-01 -3.68681014e-01 -6.30287230e-01
-8.39753211e-01 8.35995615e-01 5.00489891e-01 6.32089496e-01
-2.28937432e-01 8.67188051e-02 2.61323750e-01 7.48740971e-01
1.49273664e-01 5.53595066e-01 -4.33358103e-01 1.53062075e-01
-4.50228572e-01 -1.53666109e-01 8.97448242e-01 5.92229605e-01
8.80988777e-01 4.16077040e-02 5.01553059e-01 4.38129514e-01
2.10415363e-01 1.26268849e-01 5.65222740e-01 -7.73958564e-01
2.66743779e-01 5.90756714e-01 -2.00222149e-01 -7.99651027e-01
-6.35556698e-01 -9.70683277e-01 -1.02778637e+00 1.02980413e-01
7.87906826e-01 -6.55223727e-01 -1.85927957e-01 1.69196451e+00
3.85162272e-02 -3.24670404e-01 3.91023517e-01 8.69926989e-01
1.15025014e-01 3.72810870e-01 -3.70693982e-01 -2.87684232e-01
8.31984460e-01 -1.27456993e-01 -5.33983767e-01 4.77509126e-02
7.72697091e-01 -1.52530044e-01 6.67924464e-01 2.91265428e-01
-1.09363902e+00 -2.64171332e-01 -1.17887771e+00 2.77885288e-01
-5.27091086e-01 8.35976601e-02 4.39735770e-01 5.96013248e-01
-1.08482945e+00 1.04311979e+00 -7.47154534e-01 -2.14344889e-01
1.44315049e-01 6.00210011e-01 -2.90938854e-01 5.50196826e-01
-1.21190906e+00 8.90042722e-01 5.80018222e-01 2.95015097e-01
-6.33826077e-01 -6.81113839e-01 -3.49767566e-01 2.58916765e-01
-1.08500712e-01 -5.83856523e-01 1.01254177e+00 -1.20525968e+00
-1.59119713e+00 4.11588788e-01 2.83616334e-01 -6.66670084e-01
3.99441600e-01 5.27176354e-03 4.42345925e-02 2.20329270e-01
-5.96556783e-01 3.65424126e-01 6.39770985e-01 -1.04988050e+00
-2.61769235e-01 -5.63551903e-01 -1.21784985e-01 -1.75814182e-02
-8.47213030e-01 -4.65547413e-01 2.07888097e-01 -3.92057359e-01
1.17547549e-01 -5.99663079e-01 -2.08585650e-01 1.37736294e-02
-2.32082501e-01 -4.60648350e-02 4.84033138e-01 -2.86360115e-01
1.07289898e+00 -2.25669861e+00 4.53822285e-01 5.37427247e-01
3.42502624e-01 1.88036725e-01 1.85809750e-02 7.77403057e-01
-8.90250504e-02 1.09026730e-01 -2.55787700e-01 -7.76624656e-04
-2.15347290e-01 1.76980376e-01 -2.43214726e-01 6.01543784e-01
2.39635572e-01 5.03696561e-01 -6.15077138e-01 -2.64368858e-02
9.01817977e-02 7.35927165e-01 -5.71566403e-01 1.39086694e-01
1.34258717e-01 5.38198829e-01 -5.08851290e-01 -3.80803078e-01
1.60172686e-01 -3.74958187e-01 2.66561270e-01 3.33451629e-02
-4.86765534e-01 2.57990897e-01 -1.11742771e+00 1.19931078e+00
-5.88915884e-01 1.26420975e+00 1.72846377e-01 -1.37627292e+00
8.47229540e-01 2.07857281e-01 4.87021118e-01 -6.11346602e-01
5.28228819e-01 1.89003617e-01 3.91630203e-01 -3.44723523e-01
9.14956331e-02 -1.24873780e-01 3.96443844e-01 5.84893346e-01
5.32031655e-01 4.80358809e-01 1.05266474e-01 9.68171656e-02
9.02488291e-01 -7.63069570e-01 1.45237595e-01 -9.07721400e-01
5.02572238e-01 -4.42192674e-01 -7.62563869e-02 6.32557869e-01
7.30194673e-02 3.37116390e-01 9.08361793e-01 -2.39277184e-01
-1.12985361e+00 -6.83842123e-01 -5.19292355e-01 6.77484214e-01
6.43975213e-02 -1.45907044e-01 -9.38438058e-01 -4.61296365e-02
3.04029342e-02 2.88514704e-01 -8.04668009e-01 -4.83565122e-01
-2.28356078e-01 -7.31193662e-01 5.50223947e-01 1.77529126e-01
5.54284990e-01 -6.00025296e-01 -8.17251921e-01 1.76907897e-01
2.98488349e-01 -8.76865983e-01 9.43662897e-02 5.89784324e-01
-1.27405274e+00 -1.17600596e+00 -7.91731477e-01 -4.65638310e-01
6.63317502e-01 -1.73185199e-01 7.95708358e-01 1.05486307e-02
-1.64125770e-01 5.73973894e-01 -1.19035184e-01 -4.03680563e-01
-6.84158087e-01 2.61848956e-01 9.72644836e-02 8.76139328e-02
2.19547711e-02 -8.65473986e-01 -7.56643057e-01 3.96667629e-01
-1.00557637e+00 -1.03809051e-01 6.75874710e-01 6.70207202e-01
7.76772946e-02 3.03904533e-01 6.02560282e-01 -4.36225533e-01
8.78415644e-01 -4.15988773e-01 -9.48919356e-01 1.04206651e-01
-8.20114195e-01 8.51208031e-01 8.32194030e-01 -8.47792998e-02
-7.77722776e-01 -8.52554888e-02 -1.78536713e-01 2.57973410e-02
5.29394746e-02 5.22439420e-01 2.04794973e-01 -3.04122418e-01
6.64390862e-01 3.36354882e-01 3.98263425e-01 -5.16155720e-01
1.99417382e-01 5.07967412e-01 2.15018272e-01 -2.25843892e-01
4.88949120e-01 3.12655807e-01 2.22093880e-01 -1.22941709e+00
-4.78614688e-01 -2.95360506e-01 -7.72540808e-01 -4.71298367e-01
6.42584085e-01 -4.81924117e-01 -1.16437876e+00 4.43952858e-01
-1.09694624e+00 -3.78077984e-01 -4.00395960e-01 5.90247810e-01
-5.52640915e-01 -1.04371294e-01 -6.24623954e-01 -1.06892157e+00
-1.26971185e-01 -8.51032495e-01 3.04445326e-01 1.60191059e-01
2.50197828e-01 -1.47851872e+00 8.36196020e-02 -2.14052528e-01
5.18009067e-01 -2.64512837e-01 1.11181259e+00 -6.28199875e-01
-5.59195817e-01 -2.86146197e-02 -1.56492263e-01 5.05161047e-01
-3.20842266e-01 -5.78248538e-02 -9.36790109e-01 -8.99437815e-02
1.14513740e-01 7.33075887e-02 1.01398778e+00 5.69233298e-01
8.53105128e-01 -4.29201514e-01 -7.43310675e-02 3.97330731e-01
1.68947613e+00 2.22079501e-01 2.92008847e-01 1.82854831e-01
2.01006413e-01 9.47715878e-01 -6.26003802e-01 5.55439234e-01
-1.37537001e-02 2.38250360e-01 3.90134841e-01 2.40084797e-01
-1.68055948e-02 -1.81600973e-01 1.95633128e-01 1.20298409e+00
-1.30392626e-01 -9.66195315e-02 -7.84814715e-01 4.30722743e-01
-1.51182723e+00 -6.23501241e-01 5.12106009e-02 2.46708989e+00
7.25601733e-01 2.10632652e-01 1.81460440e-01 4.30109024e-01
5.68397701e-01 -2.30781481e-01 -4.44832236e-01 -3.11456412e-01
-1.11391939e-01 -2.09150631e-02 8.25230598e-01 6.38121486e-01
-6.76166952e-01 2.75309235e-01 7.12869024e+00 3.11334997e-01
-1.21645451e+00 -1.55015871e-01 4.23286289e-01 3.10530871e-01
-1.76555499e-01 -2.00519696e-01 -7.73902714e-01 4.08395946e-01
1.01886117e+00 -1.52638093e-01 4.55064207e-01 5.83100379e-01
7.17558041e-02 -1.56246111e-01 -1.07266796e+00 6.92066967e-01
-4.18908209e-01 -1.49692607e+00 -2.26033211e-01 2.15790227e-01
5.69851220e-01 9.53044370e-03 1.05287232e-01 -3.19121599e-01
6.63289279e-02 -6.18851781e-01 5.80562949e-01 5.33141851e-01
1.90629855e-01 -7.23821342e-01 6.05454683e-01 4.70940918e-01
-7.56350696e-01 -4.91503745e-01 -3.47639740e-01 -4.01600808e-01
-2.60679752e-01 7.96850801e-01 -5.35706580e-01 1.67984843e-01
2.87372261e-01 6.31362498e-01 -4.73896503e-01 1.10453832e+00
1.28890187e-01 6.48528874e-01 -3.94868970e-01 -4.62971449e-01
1.80010229e-01 -3.05042118e-01 4.40066010e-01 1.00856853e+00
9.88431498e-02 -2.25488413e-02 -7.73109972e-01 9.11833286e-01
-1.14412680e-01 -7.44530782e-02 -7.54146338e-01 -1.90711588e-01
3.25478643e-01 8.25748682e-01 -7.49857247e-01 2.03082219e-01
-3.12490910e-01 5.91232121e-01 2.59738892e-01 5.05309641e-01
-1.16824441e-01 -5.29957294e-01 7.52018392e-01 1.38789922e-01
3.10873210e-01 -5.22334337e-01 -3.31550449e-01 -9.37929213e-01
1.32705107e-01 -2.85735697e-01 7.49037834e-04 -1.43851221e-01
-9.21477318e-01 4.46565360e-01 1.95267722e-02 -8.15436244e-01
-4.20066059e-01 -9.20837879e-01 -5.34311950e-01 5.71380556e-01
-1.26233447e+00 -4.48248535e-02 2.04701900e-01 5.76437712e-01
-6.87456205e-02 -3.08417737e-01 6.05279505e-01 3.12514335e-01
-7.38611758e-01 5.75654566e-01 8.70951176e-01 3.69602114e-01
-1.40982717e-01 -1.10226572e+00 2.33165100e-02 6.23220026e-01
1.14263721e-01 6.47276163e-01 9.56257820e-01 1.40756872e-02
-1.51129746e+00 -5.14815211e-01 5.48929036e-01 -1.49120307e-02
1.02731490e+00 -5.04768789e-01 -7.66862690e-01 3.87794614e-01
1.77000463e-01 -4.74960297e-01 5.11218548e-01 2.82692283e-01
-4.01599333e-02 -3.60596985e-01 -6.42526925e-01 4.55453724e-01
8.37933481e-01 -5.72576642e-01 -2.81823248e-01 7.67773390e-02
3.75792056e-01 2.71596849e-01 -8.39714706e-01 7.26547539e-02
7.05653787e-01 -1.25110722e+00 6.14944339e-01 -6.00044549e-01
5.20096540e-01 3.50347936e-01 1.99387521e-01 -1.45636237e+00
-2.93896254e-02 -6.51174247e-01 3.33160609e-02 8.00881386e-01
5.56866765e-01 -8.89047623e-01 9.16091681e-01 5.24938941e-01
5.11210412e-02 -8.80518138e-01 -8.63818109e-01 -7.58273482e-01
4.45694923e-01 -2.02722996e-01 -1.57610953e-01 5.46664655e-01
3.03841889e-01 2.94366121e-01 3.23912688e-02 4.31061210e-03
6.63709462e-01 -3.04989666e-01 1.93892449e-01 -1.45562208e+00
-3.53268743e-01 -7.82254636e-01 -7.31343925e-01 -1.18662024e+00
1.75910871e-02 -6.59060895e-01 -1.24309950e-01 -9.31286931e-01
-2.93195337e-01 -3.71156514e-01 -2.89422214e-01 -2.24695504e-01
5.46935022e-01 -2.78207451e-01 2.55637784e-02 3.21433067e-01
-1.29091054e-01 5.29972494e-01 1.00638556e+00 2.88109392e-01
-2.56593525e-01 1.32762864e-01 -4.71472651e-01 6.46384001e-01
8.96990955e-01 -3.32967430e-01 -6.05715573e-01 -5.88829458e-01
6.98229134e-01 1.17194541e-01 4.21045810e-01 -1.24704599e+00
5.54047465e-01 2.99148709e-01 2.42835641e-01 -5.77487610e-02
1.65200710e-01 -9.49341238e-01 -2.11989552e-01 7.57337570e-01
-9.94695127e-01 4.58189175e-02 2.01296434e-02 6.18319929e-01
-2.39681885e-01 -6.53041422e-01 7.93245137e-01 2.26204991e-02
-4.29150164e-01 5.69118336e-02 -8.44414949e-01 2.34078094e-02
6.87151790e-01 1.77565813e-02 -1.12519659e-01 -4.39173698e-01
-7.92096257e-01 1.66872427e-01 -9.39717703e-03 5.07248566e-02
3.75233352e-01 -9.39955652e-01 -3.11962783e-01 5.68918288e-01
-2.20946997e-01 -3.39995503e-01 2.08707806e-02 8.04951131e-01
-3.78542632e-01 6.48919344e-01 -4.15521383e-01 -4.10256892e-01
-6.86686695e-01 2.76661992e-01 8.49408567e-01 5.90478182e-02
-3.76129657e-01 6.95843637e-01 1.75431177e-01 9.72743332e-02
5.67426026e-01 -2.52027959e-01 -2.45665640e-01 1.30463347e-01
5.31519711e-01 3.56004536e-01 2.31536359e-01 -1.45089880e-01
-9.78090987e-03 4.79020953e-01 1.93682760e-01 -3.18527222e-01
1.07847774e+00 -2.58150369e-01 -4.38295417e-02 7.67475307e-01
1.45601559e+00 -4.82528150e-01 -1.35374379e+00 -2.79650062e-01
2.03764170e-01 1.83520153e-01 1.89108223e-01 -3.27933460e-01
-1.27802157e+00 1.28288078e+00 6.40494645e-01 8.34931493e-01
1.03633726e+00 1.38508335e-01 1.86840355e-01 7.55849540e-01
6.01730309e-02 -1.03927755e+00 1.35218024e-01 5.98490655e-01
6.81642413e-01 -8.39847028e-01 -4.30487067e-01 -7.30366111e-02
-1.37196392e-01 1.51431012e+00 1.09211519e-01 -3.52192879e-01
1.31052685e+00 4.99663025e-01 -1.19115211e-01 -1.07205041e-01
-8.69469523e-01 -1.07144788e-01 1.00254871e-01 4.62966174e-01
5.33672988e-01 -2.25124002e-01 -4.47490126e-01 3.39674562e-01
-2.35278666e-01 -9.01352912e-02 5.27697504e-01 5.94586909e-01
-6.94165170e-01 -9.53730345e-01 1.62362829e-01 2.71782935e-01
-4.14480656e-01 4.16655988e-02 -4.60534394e-01 8.77607763e-01
-2.27073371e-01 5.78905344e-01 2.72062987e-01 -4.64690208e-01
9.81924869e-03 1.40113220e-01 5.11351705e-01 -2.24781297e-02
-3.71696353e-01 -5.58983624e-01 -3.05927962e-01 -7.44182542e-02
-4.26792592e-01 -5.62602818e-01 -1.05179787e+00 -3.80843341e-01
-4.23771858e-01 4.46853906e-01 1.05917788e+00 1.26574373e+00
4.47229713e-01 4.58040684e-01 6.99322522e-01 -5.63984454e-01
-7.28399634e-01 -6.92166030e-01 -7.27291763e-01 -6.20554313e-02
6.30328119e-01 -5.28075755e-01 -6.28238082e-01 -1.53630406e-01] | [7.950762748718262, 3.564326047897339] |
af07f3cf-96c3-4c51-915c-8802486e6dd9 | sample-efficient-on-policy-imitation-learning | 2306.09805 | null | https://arxiv.org/abs/2306.09805v1 | https://arxiv.org/pdf/2306.09805v1.pdf | Sample-Efficient On-Policy Imitation Learning from Observations | Imitation learning from demonstrations (ILD) aims to alleviate numerous shortcomings of reinforcement learning through the use of demonstrations. However, in most real-world applications, expert action guidance is absent, making the use of ILD impossible. Instead, we consider imitation learning from observations (ILO), where no expert actions are provided, making it a significantly more challenging problem to address. Existing methods often employ on-policy learning, which is known to be sample-costly. This paper presents SEILO, a novel sample-efficient on-policy algorithm for ILO, that combines standard adversarial imitation learning with inverse dynamics modeling. This approach enables the agent to receive feedback from both the adversarial procedure and a behavior cloning loss. We empirically demonstrate that our proposed algorithm requires fewer interactions with the environment to achieve expert performance compared to other state-of-the-art on-policy ILO and ILD methods. | ['Alexandros Kalousis', 'Naoya Takeishi', 'Lionel Blondé', 'João A. Cândido Ramos'] | 2023-06-16 | null | null | null | null | ['imitation-learning'] | ['methodology'] | [ 4.33268733e-02 2.80654758e-01 -3.67812276e-01 1.58871785e-01
-6.06642902e-01 -6.09617054e-01 6.52238250e-01 -2.26539001e-01
-6.17637575e-01 1.19088292e+00 -3.49842608e-01 -5.81763983e-01
-2.80788392e-02 -4.75407362e-01 -1.04567170e+00 -6.65760934e-01
-8.51669535e-02 4.00505573e-01 1.16645031e-01 -1.87956586e-01
1.78671047e-01 4.87420648e-01 -1.30766094e+00 -4.24495459e-01
1.18175769e+00 6.32783055e-01 2.62031347e-01 7.78239012e-01
2.77300119e-01 1.05004311e+00 -6.70776129e-01 2.04994261e-01
4.28629696e-01 -6.81913257e-01 -3.39730829e-01 1.46724746e-01
1.22056551e-01 -1.03735876e+00 -4.60237026e-01 1.06135261e+00
6.04441285e-01 3.70411456e-01 6.29068732e-01 -1.44828129e+00
-2.26240605e-01 3.20758909e-01 -2.96443969e-01 -1.92519888e-01
4.48428154e-01 7.29248881e-01 5.14698505e-01 -2.74046153e-01
5.37899792e-01 1.43183148e+00 3.06699216e-01 7.87635863e-01
-1.36155617e+00 -7.07834363e-01 2.80490935e-01 1.29179796e-02
-6.62907600e-01 -2.31008798e-01 6.63917720e-01 -3.46804112e-01
9.03315127e-01 -1.43038288e-01 7.69638956e-01 1.48322344e+00
1.94327086e-01 1.16778994e+00 1.53973997e+00 -3.30334902e-01
5.46150863e-01 1.28364608e-01 -5.79447448e-01 7.85463274e-01
1.17817283e-01 8.00779343e-01 -1.41054869e-01 -1.91575527e-01
1.08888865e+00 1.11887760e-01 -3.21476489e-01 -8.92763913e-01
-9.85444367e-01 6.18074358e-01 3.61235559e-01 -3.83120537e-01
-6.07632458e-01 5.89060247e-01 3.27144027e-01 8.51223946e-01
-1.26849404e-02 4.66706902e-01 -3.99655223e-01 -5.26704013e-01
-4.58800167e-01 6.27001941e-01 8.61796796e-01 8.82686138e-01
5.76351523e-01 4.76715624e-01 6.15541525e-02 4.07696992e-01
2.21841753e-01 7.11893082e-01 3.95537406e-01 -1.47939694e+00
4.46345240e-01 3.50033611e-01 7.95780540e-01 -2.65767604e-01
-5.77671751e-02 -2.03881070e-01 -2.32535228e-01 1.27015483e+00
4.48957920e-01 -5.16136587e-01 -7.92803586e-01 1.86931121e+00
5.43158829e-01 3.79639179e-01 3.19188178e-01 8.24360430e-01
-4.36713845e-02 3.24034542e-01 -1.56351313e-01 -2.81673670e-01
4.86906439e-01 -1.16735685e+00 -6.24916553e-01 -1.37002468e-01
4.83796120e-01 -4.46453393e-01 1.33963561e+00 4.67059374e-01
-9.34962153e-01 -3.83784860e-01 -1.08504426e+00 4.34459060e-01
-4.22819890e-02 -6.98129013e-02 4.27658916e-01 3.91494513e-01
-5.79528868e-01 1.12018740e+00 -1.20778871e+00 -1.75258741e-01
2.49912158e-01 5.44597864e-01 -6.39953762e-02 2.47354344e-01
-8.68627191e-01 1.16149664e+00 2.55245924e-01 -1.98879793e-01
-1.70158005e+00 -5.68933904e-01 -8.33164454e-01 -1.19975932e-01
1.15143597e+00 -4.58512872e-01 1.82380295e+00 -1.08746791e+00
-2.45535374e+00 -4.93550033e-04 1.83484584e-01 -5.65148771e-01
1.03842878e+00 -6.04561865e-01 1.42094001e-01 1.23078473e-01
-5.24691446e-03 4.49629813e-01 1.22247136e+00 -1.26094639e+00
-5.36119401e-01 3.92138660e-02 5.87017059e-01 3.41687560e-01
-8.85467082e-02 -4.67933834e-01 1.20160235e-02 -6.07150257e-01
-6.94680154e-01 -1.23665154e+00 -2.81661659e-01 3.49635541e-01
-1.99334234e-01 -2.00052038e-01 1.23035502e+00 -4.95532304e-01
7.09729314e-01 -2.03695250e+00 3.64225060e-01 9.23231337e-03
-2.19371114e-02 5.12887120e-01 -2.68587589e-01 6.90873623e-01
2.66714692e-01 -3.52145702e-01 -2.24132434e-01 -2.26896092e-01
3.32044035e-01 4.58727986e-01 -5.78808725e-01 4.90412712e-01
-6.26879605e-03 8.44264507e-01 -1.32990825e+00 -2.24985883e-01
4.33496833e-01 8.08037966e-02 -6.59957647e-01 6.82446003e-01
-6.83017075e-01 1.02065098e+00 -7.08868861e-01 5.69096506e-01
1.49374992e-01 1.54643744e-01 1.63210496e-01 5.28247476e-01
-1.95917338e-01 1.58749804e-01 -9.75837648e-01 1.45091736e+00
-7.89509356e-01 3.15318376e-01 3.22480351e-01 -9.82668757e-01
4.87613440e-01 5.54054976e-01 3.54124218e-01 -4.80748266e-01
1.03725918e-01 3.93869519e-01 2.37480953e-01 -5.82836628e-01
-1.48077467e-02 9.83367953e-03 -2.44040173e-02 5.29975593e-01
-1.05627351e-01 -4.09223437e-01 8.02217126e-02 -8.28191787e-02
1.28834212e+00 9.48717475e-01 4.99905437e-01 2.36478537e-01
3.64739597e-01 2.11849306e-02 6.29149735e-01 9.59289372e-01
-4.08684939e-01 -1.39113724e-01 4.76518154e-01 1.91424079e-02
-1.03300655e+00 -1.20492148e+00 4.77417320e-01 7.14347482e-01
2.63923526e-01 7.21571147e-02 -9.47521687e-01 -1.10781610e+00
3.17672133e-01 7.79813588e-01 -4.59556669e-01 -2.07857400e-01
-7.27947176e-01 1.57197013e-01 3.48259211e-01 5.54304361e-01
4.66615826e-01 -1.35225737e+00 -1.07600021e+00 4.15683508e-01
3.09400797e-01 -8.42381239e-01 -5.76337278e-01 4.07641791e-02
-9.60969090e-01 -1.16889620e+00 -7.46108234e-01 -4.27723229e-01
7.48861313e-01 1.99319750e-01 5.14246702e-01 -5.40090874e-02
-1.44174740e-01 8.10825586e-01 -2.42375568e-01 -4.39206958e-01
-7.53340065e-01 -2.50233293e-01 4.93413866e-01 -3.48570824e-01
-1.23984955e-01 -8.88434291e-01 -4.73665476e-01 2.70794541e-01
-5.82661450e-01 -1.56209931e-01 9.43134844e-01 1.30128777e+00
4.47164267e-01 -6.99092299e-02 6.77630842e-01 -5.89551806e-01
8.86752605e-01 -1.40286252e-01 -1.23075366e+00 4.24961373e-02
-6.55589342e-01 3.13232780e-01 1.22486353e+00 -1.03488398e+00
-9.27668154e-01 6.21610470e-02 1.95438132e-01 -9.31654096e-01
-1.97015613e-01 4.21305187e-02 -5.75546846e-02 -3.58432889e-01
4.74325061e-01 4.24798816e-01 4.59528863e-01 -3.30373168e-01
3.94270629e-01 6.13581836e-01 4.20307398e-01 -7.79958725e-01
9.87509072e-01 2.77930528e-01 1.73767600e-02 -6.95552647e-01
-4.89738137e-01 -8.86827409e-02 -3.35700035e-01 -2.53694713e-01
5.48241854e-01 -6.57455087e-01 -1.20691013e+00 2.12393120e-01
-7.42103815e-01 -1.07084036e+00 -4.72247571e-01 7.69977629e-01
-1.31202948e+00 3.67156476e-01 -3.38920623e-01 -1.14887428e+00
-7.56656602e-02 -1.38019001e+00 7.10530043e-01 2.50976533e-01
-1.40225425e-01 -8.50436211e-01 1.31332800e-01 1.23863434e-02
2.95206577e-01 4.14816350e-01 5.70634902e-01 -3.08623672e-01
-7.87798047e-01 -1.64037962e-02 3.50567102e-01 4.93362397e-01
1.60641104e-01 -4.68943000e-01 -6.09973848e-01 -8.56975496e-01
9.24783573e-02 -9.49807465e-01 3.26398760e-01 9.28872898e-02
1.03977251e+00 -5.46377122e-01 -2.43860662e-01 3.99805129e-01
1.13108158e+00 5.25291800e-01 2.56723285e-01 5.71212709e-01
4.52212065e-01 4.26167846e-01 1.09990394e+00 5.50808191e-01
-4.71745469e-02 6.93712234e-01 6.99179888e-01 9.04502049e-02
8.26416612e-02 -6.78486347e-01 9.09194291e-01 3.20824385e-01
-1.22970678e-01 9.66141224e-02 -3.60318065e-01 4.42226499e-01
-2.13231540e+00 -8.77967000e-01 5.26143312e-01 2.24488282e+00
7.67550468e-01 9.36952531e-02 2.91731417e-01 -1.52765825e-01
3.71212035e-01 -1.33868828e-01 -1.30617929e+00 -2.85646915e-01
6.43109858e-01 1.04788877e-01 5.88781595e-01 6.15205050e-01
-9.51591551e-01 8.73288810e-01 6.04563618e+00 7.44099319e-01
-1.03335702e+00 -2.36101095e-02 -1.68699875e-01 -2.61197805e-01
1.30177811e-01 1.18363358e-01 -3.79662722e-01 5.37651777e-01
6.83626175e-01 -1.57715484e-01 1.03957188e+00 1.23296595e+00
6.59199774e-01 -2.43782416e-01 -1.36893570e+00 6.21288955e-01
-3.86766285e-01 -5.97735226e-01 -4.34106261e-01 1.06822006e-01
6.99935853e-01 -1.60299361e-01 7.32490346e-02 7.11547852e-01
7.18333840e-01 -7.70949125e-01 7.10565269e-01 2.47186467e-01
7.01509774e-01 -7.79310107e-01 4.07431960e-01 8.31066072e-01
-8.39808285e-01 -3.93095076e-01 -1.55468509e-01 -2.43848458e-01
3.03535461e-01 -3.20641041e-01 -8.21422040e-01 3.45228225e-01
2.81224310e-01 6.02800071e-01 2.35622033e-01 7.95841098e-01
-8.25078249e-01 6.05332434e-01 -3.21661741e-01 -3.91661465e-01
5.94634771e-01 -2.36999139e-01 9.10257220e-01 5.68202972e-01
2.71477044e-01 -2.15041339e-01 7.80623019e-01 8.11026096e-01
3.83652635e-02 -3.18625271e-01 -9.35193181e-01 -2.02764779e-01
4.17312026e-01 7.59808898e-01 -1.02182515e-01 -4.00532722e-01
-3.09544355e-01 1.16369474e+00 5.04838109e-01 4.07873899e-01
-9.56325471e-01 -4.47156131e-01 8.66944790e-01 -1.52119160e-01
5.33861160e-01 -3.79300803e-01 4.35153008e-01 -1.06614161e+00
1.17619947e-01 -1.27229321e+00 -2.16860846e-01 -3.47778767e-01
-1.01926780e+00 -1.01390660e-01 3.90330665e-02 -1.71354294e+00
-8.29165101e-01 -4.62881356e-01 -5.92809856e-01 6.17419243e-01
-1.70461166e+00 -7.38991380e-01 5.19600697e-02 4.55311447e-01
7.96008766e-01 -2.34251171e-01 7.29199231e-01 8.49224813e-03
-4.67156589e-01 5.61872542e-01 4.90543574e-01 -2.51027286e-01
6.20621204e-01 -1.46625328e+00 1.60624728e-01 5.70350230e-01
-3.72483313e-01 5.19019485e-01 9.71815646e-01 -5.86146593e-01
-1.59923446e+00 -9.31688249e-01 -1.89485952e-01 -5.12990057e-02
9.10839379e-01 -1.91017002e-01 -6.32948995e-01 8.04333687e-01
1.26787245e-01 -9.04659256e-02 2.49275472e-02 -4.81136769e-01
-6.35506287e-02 4.17629145e-02 -1.14940655e+00 1.05515289e+00
9.59931910e-01 -4.60022300e-01 -7.26931334e-01 2.24638745e-01
6.10669136e-01 -5.06830573e-01 -6.28989160e-01 1.96300134e-01
6.28833652e-01 -6.10127687e-01 8.47064197e-01 -7.01122522e-01
2.09889755e-01 -4.69392329e-01 2.89898455e-01 -1.80116343e+00
1.94275767e-01 -1.17210937e+00 -5.54823458e-01 6.59759223e-01
6.18996993e-02 -9.26946759e-01 7.10517883e-01 2.09531888e-01
-6.69381395e-02 -7.45429575e-01 -8.98867428e-01 -1.50569510e+00
1.96696445e-03 4.19373401e-02 2.17739329e-01 6.55380845e-01
-3.17562893e-02 9.73637775e-02 -8.09836090e-01 2.44330138e-01
8.39900017e-01 2.55144864e-01 1.32394791e+00 -6.87286675e-01
-9.70663965e-01 -2.96696723e-01 -4.07890696e-03 -1.50129485e+00
4.92482960e-01 -5.23564279e-01 4.50441390e-01 -1.24592292e+00
-2.85849601e-01 -3.15937310e-01 -5.48825413e-02 4.28357840e-01
-1.40288427e-01 -4.20850217e-01 3.68674129e-01 8.60965326e-02
-4.65189993e-01 1.10898948e+00 1.66189933e+00 -8.75787735e-02
-3.44753146e-01 2.50162810e-01 -8.78754724e-03 9.41640675e-01
9.42623675e-01 -6.73014104e-01 -7.98914909e-01 -1.71712205e-01
-5.13013065e-01 4.29582208e-01 2.79490530e-01 -9.75557327e-01
-1.94454268e-02 -5.63520133e-01 -5.19571826e-02 -1.36882409e-01
5.05398989e-01 -8.73038650e-01 -2.47888908e-01 1.15019667e+00
-3.22452515e-01 -5.65005653e-02 4.30750959e-02 9.43210363e-01
4.15277220e-02 -3.38182777e-01 7.61247158e-01 -2.53247857e-01
-5.55452943e-01 1.79151013e-01 -5.96078753e-01 -8.49059299e-02
1.29743969e+00 9.04534534e-02 -1.52410895e-01 -5.98016500e-01
-5.77816486e-01 5.73835254e-01 6.81931913e-01 1.74751639e-01
5.51266372e-01 -1.13270366e+00 -2.14404970e-01 -1.43338740e-01
-2.56893337e-01 -1.03085525e-01 -1.37925118e-01 8.11549246e-01
-1.75788790e-01 2.10879773e-01 -3.23111385e-01 -2.36812890e-01
-1.21263528e+00 8.39694440e-01 3.84577751e-01 -4.23195601e-01
-8.93539429e-01 2.38766864e-01 1.83168069e-01 -7.01194167e-01
4.30964679e-01 -1.27105445e-01 6.68294579e-02 -6.10950172e-01
2.55853057e-01 5.74977517e-01 -6.08978391e-01 1.91637039e-01
1.63235515e-01 3.41145158e-01 -4.51275036e-02 -5.02515674e-01
1.19975126e+00 1.21274069e-01 3.52965295e-01 4.91270542e-01
6.67664766e-01 -2.16536224e-01 -2.10925937e+00 -4.25719619e-02
-1.44196853e-01 -5.49605548e-01 -2.08538175e-01 -7.49612391e-01
-4.37973410e-01 8.50430667e-01 5.60084820e-01 -7.91505873e-02
8.19734335e-01 -5.90085506e-01 8.48061979e-01 9.55355525e-01
7.28151619e-01 -1.36157072e+00 5.62014878e-01 4.50467616e-01
1.01186121e+00 -1.32442558e+00 -4.99365889e-02 -8.21522623e-02
-4.66284454e-01 8.45217347e-01 9.65219498e-01 -5.28743804e-01
1.77291617e-01 1.95427150e-01 6.46572411e-02 4.02683228e-01
-7.22958565e-01 -1.66960314e-01 -3.50129813e-01 7.32141793e-01
-3.00971419e-01 2.77896915e-02 -2.68523991e-01 -5.76086938e-02
2.70335346e-01 1.86559021e-01 6.10359430e-01 1.50421476e+00
-4.79392380e-01 -1.50498950e+00 -3.94174755e-01 1.91998199e-01
-3.37776005e-01 3.96436870e-01 -1.67390823e-01 1.06362593e+00
-2.77291566e-01 6.85242534e-01 -4.54501569e-01 1.36446429e-03
3.70905370e-01 -1.50110513e-01 7.60966897e-01 -5.21958172e-01
-2.85811871e-01 7.05887973e-02 1.62637886e-02 -9.43188667e-01
-2.58183092e-01 -5.37917018e-01 -1.26416838e+00 -1.23998493e-01
-2.39850685e-01 -4.14131247e-02 5.62097669e-01 9.27820563e-01
2.28046969e-01 7.85329401e-01 8.62848818e-01 -1.20876634e+00
-1.55982280e+00 -9.31017697e-01 -3.59011799e-01 2.90501446e-01
7.55043209e-01 -1.22475207e+00 -4.32829082e-01 -3.12490344e-01] | [4.311412334442139, 1.5652445554733276] |
dbf225e1-8b15-4707-9668-00467aa3d96c | counterfactual-data-augmentation-using | 2007.02863 | null | https://arxiv.org/abs/2007.02863v2 | https://arxiv.org/pdf/2007.02863v2.pdf | Counterfactual Data Augmentation using Locally Factored Dynamics | Many dynamic processes, including common scenarios in robotic control and reinforcement learning (RL), involve a set of interacting subprocesses. Though the subprocesses are not independent, their interactions are often sparse, and the dynamics at any given time step can often be decomposed into locally independent causal mechanisms. Such local causal structures can be leveraged to improve the sample efficiency of sequence prediction and off-policy reinforcement learning. We formalize this by introducing local causal models (LCMs), which are induced from a global causal model by conditioning on a subset of the state space. We propose an approach to inferring these structures given an object-oriented state representation, as well as a novel algorithm for Counterfactual Data Augmentation (CoDA). CoDA uses local structures and an experience replay to generate counterfactual experiences that are causally valid in the global model. We find that CoDA significantly improves the performance of RL agents in locally factored tasks, including the batch-constrained and goal-conditioned settings. | ['Animesh Garg', 'Silviu Pitis', 'Elliot Creager'] | 2020-07-06 | null | http://proceedings.neurips.cc/paper/2020/hash/294e09f267683c7ddc6cc5134a7e68a8-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/294e09f267683c7ddc6cc5134a7e68a8-Paper.pdf | neurips-2020-12 | ['multi-goal-reinforcement-learning'] | ['methodology'] | [ 2.76729792e-01 3.79264563e-01 -4.79664117e-01 -9.71657410e-02
-5.43942571e-01 -6.39711201e-01 1.32765090e+00 4.94750477e-02
-3.30196351e-01 1.19204366e+00 1.05871105e+00 -4.19166416e-01
-5.49524665e-01 -6.49202824e-01 -1.11156321e+00 -8.12184453e-01
-6.64219141e-01 6.64158821e-01 -1.76652715e-01 5.02838120e-02
8.46634880e-02 1.99174508e-01 -1.47947037e+00 1.90420970e-01
6.96652591e-01 2.61239648e-01 4.33553666e-01 8.41821134e-01
1.73318878e-01 1.31626880e+00 -3.10388446e-01 1.19424120e-01
2.26542816e-01 -8.05378914e-01 -8.49757612e-01 7.00360909e-02
-1.07484184e-01 -4.64531094e-01 -4.00557011e-01 8.20696533e-01
2.52362788e-01 4.59268540e-01 7.52175868e-01 -1.50160968e+00
-4.34543133e-01 1.10293710e+00 -4.27101225e-01 1.66152641e-02
3.93097907e-01 6.32787287e-01 1.03903592e+00 -2.08343416e-01
6.82075679e-01 2.04992771e+00 2.37975553e-01 7.47225463e-01
-1.74696827e+00 -6.78873241e-01 7.92876124e-01 5.96754961e-02
-3.93142313e-01 -3.30794692e-01 4.97079432e-01 -4.54355836e-01
8.03728282e-01 -2.10769653e-01 8.91163945e-01 1.71242154e+00
4.59594101e-01 1.03966618e+00 1.48476374e+00 -2.62427300e-01
8.38464618e-01 -6.82528198e-01 -2.62144506e-01 7.19484687e-01
1.20290399e-01 9.80916917e-01 -8.05073380e-01 -5.43252826e-01
1.11894941e+00 -8.14576447e-03 -1.33684039e-01 -6.85908318e-01
-1.54034722e+00 8.98105383e-01 1.96828827e-01 -2.69772440e-01
-8.05274308e-01 8.56031060e-01 1.90394908e-01 4.85500872e-01
1.59235835e-01 7.35188127e-01 -6.90431118e-01 -1.70725241e-01
-1.43637732e-01 8.85715246e-01 1.06632102e+00 8.13682020e-01
3.41134459e-01 -1.10877767e-01 -4.43247765e-01 3.43922347e-01
3.96822602e-01 5.23633242e-01 4.39653009e-01 -1.36782002e+00
3.14366877e-01 -7.01232394e-03 6.15600765e-01 -2.18325406e-01
-3.20008397e-01 1.25873834e-01 -4.78593200e-01 3.03787559e-01
5.75699985e-01 -5.69999993e-01 -1.07966959e+00 2.36957645e+00
5.12894809e-01 6.87628627e-01 2.15494826e-01 8.00349712e-01
-1.11854106e-01 5.94025433e-01 6.58648074e-01 -5.06706417e-01
8.59143138e-01 -8.20277095e-01 -7.62053370e-01 -4.71013576e-01
3.89411390e-01 -2.83019304e-01 1.08317626e+00 2.35654965e-01
-1.01569927e+00 -2.68240601e-01 -8.49268436e-01 3.49061251e-01
2.12973550e-01 -4.02239829e-01 1.13213134e+00 9.28998664e-02
-8.76627624e-01 8.67188096e-01 -1.13994384e+00 -3.21758002e-01
3.85541469e-01 1.77845791e-01 -1.33267269e-01 -1.41714975e-01
-1.13934505e+00 9.02483225e-01 5.22142708e-01 -2.57736951e-01
-2.03861547e+00 -7.90989459e-01 -6.71879172e-01 -5.75306825e-02
9.47256267e-01 -1.01976645e+00 1.88554621e+00 -7.01375604e-01
-1.80165648e+00 7.39919841e-02 -7.35219344e-02 -6.54996812e-01
4.17943329e-01 -4.70376283e-01 4.80921865e-02 -8.59728977e-02
1.49362400e-01 6.23783946e-01 1.18077564e+00 -1.31070161e+00
-8.34572375e-01 -2.24260285e-01 2.52513409e-01 6.95916057e-01
4.67785209e-01 -7.00265467e-02 2.70613264e-02 -5.46481431e-01
-2.08870158e-01 -1.20315492e+00 -8.37387979e-01 -2.45542869e-01
-3.77172887e-01 -2.82841086e-01 4.67567325e-01 -2.29074255e-01
4.97330695e-01 -1.75412083e+00 6.18138313e-01 -3.24784666e-02
1.46491915e-01 -4.02862400e-01 -6.16495848e-01 4.82158810e-01
-1.98240235e-01 7.13617504e-02 -1.18438318e-01 -1.97505236e-01
1.44567192e-01 6.38416588e-01 -7.86125660e-01 4.29085165e-01
2.45741889e-01 9.65921581e-01 -1.50379491e+00 -1.28647149e-01
2.60006785e-01 -2.00496376e-01 -7.97416210e-01 6.55486166e-01
-9.69950974e-01 8.73054206e-01 -6.32882953e-01 2.81445310e-02
5.68331517e-02 -2.06998754e-02 6.92906857e-01 5.48457026e-01
-7.32067740e-04 8.16020012e-01 -1.30256760e+00 1.81522882e+00
-4.55577999e-01 -4.97977883e-02 -4.53342907e-02 -7.72210836e-01
2.99018830e-01 6.37093663e-01 5.80425620e-01 -2.65826941e-01
-4.86997887e-02 -2.53762007e-01 3.28055739e-01 -5.19689798e-01
2.51987129e-01 -5.74589014e-01 -2.07397893e-01 9.33931231e-01
3.41688603e-01 -1.86004356e-01 1.02133393e-01 3.00538421e-01
1.38391054e+00 7.94176161e-01 5.72322607e-01 -2.44249329e-01
-2.54688859e-01 2.02955663e-01 7.66687870e-01 1.29860759e+00
-2.98844814e-01 3.32059921e-03 9.33953404e-01 -1.89259812e-01
-1.19874179e+00 -1.21453679e+00 4.99713928e-01 1.32582450e+00
1.00285158e-01 -3.20294023e-01 -4.67601448e-01 -7.68145859e-01
2.13193819e-01 1.11808884e+00 -7.64473736e-01 -3.59168768e-01
-4.33669299e-01 -8.38212132e-01 3.12648505e-01 6.69005752e-01
1.80570945e-01 -1.48404908e+00 -7.30629921e-01 5.17878234e-01
-1.41332662e-02 -7.25754380e-01 -3.25508565e-01 4.66594607e-01
-9.79590237e-01 -1.16013277e+00 -2.63931841e-01 -1.68875709e-01
2.87028074e-01 9.67801064e-02 1.08498168e+00 -4.61826086e-01
1.51938021e-01 6.19568110e-01 -2.54112966e-02 -5.55735469e-01
-6.70279682e-01 -4.94703591e-01 5.74278951e-01 -2.13442430e-01
-2.37165764e-02 -7.65701890e-01 -3.68462294e-01 9.24524218e-02
-5.66104591e-01 3.21988732e-01 7.17477977e-01 9.76181030e-01
4.79507416e-01 -9.20874849e-02 6.10888004e-01 -8.02079797e-01
9.41069424e-01 -7.11378634e-01 -6.14233553e-01 -1.35370707e-02
-4.28964019e-01 6.55951977e-01 4.76660192e-01 -8.68399620e-01
-1.72274935e+00 1.44027203e-01 5.89902163e-01 -4.66660827e-01
-4.74714011e-01 6.23192489e-01 -1.64611444e-01 8.03215086e-01
7.65841603e-01 -6.92859367e-02 1.88627332e-01 -3.02947283e-01
8.79567027e-01 -9.16098505e-02 3.99038494e-01 -1.13946676e+00
5.88419974e-01 4.35220242e-01 1.64129168e-01 -3.72489691e-01
-9.70480621e-01 -1.64747193e-01 -5.07558227e-01 -2.72201657e-01
6.97169423e-01 -1.06582713e+00 -1.05682600e+00 2.45948330e-01
-1.07728815e+00 -1.17754054e+00 -6.00148380e-01 7.85649300e-01
-1.22539461e+00 -1.97007120e-01 -6.25668049e-01 -1.15801156e+00
3.68751556e-01 -9.28839147e-01 9.32608306e-01 5.47709838e-02
-4.42822695e-01 -9.62251186e-01 3.88130724e-01 -1.33713543e-01
-1.39867797e-01 6.36401474e-02 1.00776911e+00 -5.38391948e-01
-6.58518195e-01 3.79282922e-01 2.82012910e-01 -9.68110263e-02
3.23091075e-02 -2.28013843e-01 -7.64550924e-01 -1.85898021e-01
4.72508818e-02 -6.94311380e-01 8.70259166e-01 6.91431403e-01
7.50797033e-01 -6.27844334e-01 -5.77798665e-01 -5.81936166e-02
9.41858172e-01 3.43662709e-01 3.15318435e-01 6.37206854e-03
5.47524333e-01 8.46681833e-01 8.58893216e-01 4.25407797e-01
2.86752075e-01 2.64148474e-01 5.39662659e-01 3.51116866e-01
7.89424405e-02 -9.18063760e-01 8.40434790e-01 1.71276376e-01
-2.78010994e-01 -4.51021157e-02 -7.14020967e-01 5.70406497e-01
-2.46604919e+00 -1.49895430e+00 -1.30009418e-02 2.04476023e+00
1.04120326e+00 -7.11956546e-02 1.86626434e-01 -4.09769326e-01
5.74892461e-01 2.02135429e-01 -9.58431065e-01 -2.82662474e-02
1.45049170e-01 6.95554391e-02 5.60975015e-01 6.78150833e-01
-1.05802345e+00 9.78444219e-01 7.13673067e+00 5.09090483e-01
-5.39113820e-01 2.15569556e-01 3.89140338e-01 -3.74751002e-01
-3.37574840e-01 3.46856773e-01 -5.39690912e-01 2.12714955e-01
1.03468132e+00 -2.51060069e-01 8.58452022e-01 8.22160304e-01
6.78870082e-01 -2.83573508e-01 -1.67099357e+00 3.98480654e-01
-6.00304127e-01 -1.33740747e+00 -8.91845524e-02 3.21547717e-01
1.12273705e+00 6.45383820e-02 -7.84961879e-02 6.03148699e-01
1.88652515e+00 -1.00770569e+00 9.22638237e-01 5.32951534e-01
3.67894560e-01 -7.14305401e-01 1.20435230e-01 7.41457939e-01
-6.28390968e-01 -5.93267918e-01 -2.46031821e-01 -5.13093412e-01
1.84661254e-01 1.82687417e-01 -8.70957077e-01 2.85521120e-01
4.67640013e-01 6.75797939e-01 2.44176328e-01 5.47664046e-01
-8.87671351e-01 8.95772696e-01 -2.28018180e-01 -1.92346349e-01
3.30746979e-01 -2.37517193e-01 7.18246639e-01 6.23318076e-01
-1.76347136e-01 3.01616460e-01 4.69762206e-01 1.05888939e+00
1.57011896e-01 -4.92168754e-01 -8.88806164e-01 -1.06124431e-01
4.85740185e-01 9.37403083e-01 -3.59703183e-01 -4.19984549e-01
-1.38169169e-01 7.59219229e-01 5.47695398e-01 6.16353750e-01
-8.32355857e-01 5.96873462e-01 1.07581651e+00 -3.54015797e-01
7.36614242e-02 -3.34668040e-01 -7.63416365e-02 -1.32425344e+00
-5.13193786e-01 -1.13778961e+00 4.78728861e-01 -7.58454740e-01
-1.45471585e+00 -1.95818186e-01 2.90738225e-01 -8.80414605e-01
-7.42094815e-01 -3.23209137e-01 -5.59605718e-01 7.90051460e-01
-1.15353036e+00 -9.23549175e-01 3.41483355e-01 6.42385125e-01
6.11540377e-01 9.82795879e-02 8.37653518e-01 -5.63334405e-01
-4.32075590e-01 -3.03447694e-01 -1.72386523e-02 -3.84187967e-01
6.07111335e-01 -1.54475498e+00 2.66637921e-01 8.70465994e-01
9.60498229e-02 8.14052522e-01 9.84885395e-01 -1.16730869e+00
-1.47890401e+00 -1.24158323e+00 2.58363485e-01 -6.79582477e-01
1.05307424e+00 -3.29921722e-01 -5.94414234e-01 1.12564051e+00
2.29638770e-01 -3.42813730e-01 3.57538402e-01 6.65171444e-01
-3.13540071e-01 5.14807463e-01 -8.95763457e-01 1.22862709e+00
1.45642245e+00 -4.37141538e-01 -1.08655190e+00 3.61980021e-01
8.81841779e-01 -2.72952557e-01 -5.46897233e-01 -4.07415591e-02
3.91323775e-01 -5.22452354e-01 8.67063046e-01 -1.42382181e+00
8.57524216e-01 -2.31294855e-01 6.51687384e-02 -1.90090692e+00
-4.78666723e-01 -9.96487439e-01 -5.60508132e-01 7.90342271e-01
2.47928858e-01 -4.60496455e-01 5.00136137e-01 6.28133953e-01
4.71069589e-02 -2.04299852e-01 -6.28751874e-01 -7.84089208e-01
1.56179816e-01 -5.42248547e-01 5.77023387e-01 9.45000470e-01
2.15891048e-01 7.72624016e-01 -5.47480464e-01 2.50122249e-01
9.08050358e-01 2.55488127e-01 8.72798085e-01 -1.06908751e+00
-7.40649521e-01 -1.94846451e-01 4.18987095e-01 -1.07018340e+00
4.26301926e-01 -6.72854841e-01 5.89368939e-01 -1.31988573e+00
4.15743798e-01 -5.49139500e-01 -1.23688661e-01 6.49390101e-01
-4.65368748e-01 -6.92263424e-01 3.05466533e-01 3.43917966e-01
-4.81018096e-01 9.84120965e-01 1.59996963e+00 8.52520317e-02
-4.25460637e-01 -6.73432425e-02 -6.42895579e-01 9.72099960e-01
7.26219654e-01 -5.86868346e-01 -7.79502511e-01 -6.10057153e-02
1.09571792e-01 6.72952831e-01 6.74611390e-01 -5.48004210e-01
1.27963662e-01 -9.10116971e-01 2.45256156e-01 -1.60495117e-01
1.44784853e-01 -4.89996433e-01 1.06728658e-01 7.21141160e-01
-1.06876159e+00 -1.86309099e-01 1.72849093e-02 1.42570221e+00
2.46648297e-01 -4.19103391e-02 3.97454947e-01 -4.03682709e-01
-7.07871377e-01 2.79224485e-01 -8.17269325e-01 -5.06545939e-02
9.86772299e-01 4.14104700e-01 -2.12558895e-01 -6.26445472e-01
-8.66374433e-01 4.65534955e-01 1.38755053e-01 5.63665032e-01
2.90360481e-01 -1.16710317e+00 -5.49831450e-01 -2.25729063e-01
-4.40164283e-02 9.98936072e-02 5.18598743e-02 6.35436118e-01
3.33626270e-01 2.98787385e-01 -3.52465749e-01 -2.72821248e-01
-7.14262426e-01 9.31272268e-01 1.64610654e-01 -5.57310581e-01
-5.87865651e-01 6.13541305e-01 6.16137028e-01 -5.87688982e-01
4.64624614e-02 -4.80145186e-01 -1.88267052e-01 -1.11931823e-01
4.57018733e-01 2.18672171e-01 -7.03804612e-01 1.68960705e-01
2.01411411e-01 -4.30409610e-01 4.08890732e-02 -8.59644949e-01
1.36832297e+00 -5.76794110e-02 2.92773396e-02 7.99499452e-01
3.51931632e-01 -2.63845354e-01 -2.16732931e+00 -1.86547071e-01
2.11022466e-01 -2.94974744e-01 -9.88598317e-02 -1.04280281e+00
-3.95585924e-01 4.09508586e-01 4.13835347e-02 9.80175957e-02
6.03347540e-01 1.62669763e-01 4.34018895e-02 4.54353869e-01
7.74275541e-01 -1.02910221e+00 2.92602271e-01 6.79509699e-01
1.08411002e+00 -8.98953438e-01 -1.46846950e-01 -8.95852298e-02
-8.54503870e-01 5.30606449e-01 5.62056065e-01 -3.77534091e-01
3.51024181e-01 4.09462988e-01 -4.66780096e-01 -1.75750434e-01
-1.63343740e+00 -3.97638887e-01 -2.04442233e-01 7.29729891e-01
4.33432013e-02 3.95464182e-01 -9.75762531e-02 5.71295440e-01
6.48530573e-02 3.91479172e-02 5.52437067e-01 1.04490376e+00
-1.71009779e-01 -1.00108683e+00 -4.62578654e-01 4.38165545e-01
-1.27684981e-01 -4.66475300e-02 -2.25141793e-01 7.23869324e-01
-1.79629028e-01 1.04022813e+00 2.31299162e-01 -9.73504409e-02
7.38835037e-02 2.40470469e-01 8.25396717e-01 -8.16570342e-01
-2.34501824e-01 1.96225330e-01 3.33607912e-01 -1.15880311e+00
-5.58128178e-01 -1.23904550e+00 -1.31420898e+00 -3.11805308e-01
7.36901835e-02 -1.58867531e-03 3.43413323e-01 1.16819179e+00
2.85806835e-01 7.87790358e-01 4.58779126e-01 -1.04495370e+00
-9.59851742e-01 -1.13930905e+00 -5.10901928e-01 4.13011432e-01
3.84119332e-01 -1.02690005e+00 -1.80372193e-01 4.26097572e-01] | [4.185120105743408, 1.5516901016235352] |
c4ea119d-16d5-4f33-9609-656215fc4d1b | mutual-contrastive-learning-for-visual | 2104.12565 | null | https://arxiv.org/abs/2104.12565v2 | https://arxiv.org/pdf/2104.12565v2.pdf | Mutual Contrastive Learning for Visual Representation Learning | We present a collaborative learning method called Mutual Contrastive Learning (MCL) for general visual representation learning. The core idea of MCL is to perform mutual interaction and transfer of contrastive distributions among a cohort of networks. A crucial component of MCL is Interactive Contrastive Learning (ICL). Compared with vanilla contrastive learning, ICL can aggregate cross-network embedding information and maximize the lower bound to the mutual information between two networks. This enables each network to learn extra contrastive knowledge from others, leading to better feature representations for visual recognition tasks. We emphasize that the resulting MCL is conceptually simple yet empirically powerful. It is a generic framework that can be applied to both supervised and self-supervised representation learning. Experimental results on image classification and transfer learning to object detection show that MCL can lead to consistent performance gains, demonstrating that MCL can guide the network to generate better feature representations. Code is available at https://github.com/winycg/MCL. | ['Yongjun Xu', 'Linhang Cai', 'Zhulin An', 'Chuanguang Yang'] | 2021-04-26 | null | null | null | null | ['self-supervised-image-classification'] | ['computer-vision'] | [ 5.58236651e-02 1.66185368e-02 -5.25925815e-01 -3.02797496e-01
-4.34098452e-01 -6.09875441e-01 7.78587878e-01 -1.99002791e-02
-1.55761927e-01 4.01986480e-01 1.42327458e-01 -2.26067707e-01
-4.01394695e-01 -7.36942410e-01 -5.70555389e-01 -5.87333322e-01
-5.44915915e-01 -8.89450498e-03 7.34937713e-02 2.51422767e-02
2.84238514e-02 6.91096127e-01 -1.40551591e+00 5.13842225e-01
6.34031355e-01 6.95912004e-01 2.57854491e-01 6.17864013e-01
-1.58003569e-01 1.24004483e+00 -3.16706717e-01 -4.23080549e-02
1.43737093e-01 -4.94932324e-01 -9.65475857e-01 5.55626824e-02
2.92148173e-01 5.74292541e-02 -3.81536186e-01 8.75376701e-01
3.17470253e-01 2.89905608e-01 1.26236284e+00 -1.66437173e+00
-9.65157092e-01 6.48915172e-01 -7.20836580e-01 3.18140864e-01
2.64521390e-01 9.69021469e-02 1.37988794e+00 -1.20355284e+00
4.24556464e-01 1.37266600e+00 4.79757220e-01 6.79150999e-01
-1.56381857e+00 -1.04552221e+00 3.00195843e-01 3.09690326e-01
-1.26024222e+00 -2.81482875e-01 9.34337437e-01 -6.09341919e-01
5.47834098e-01 2.77044088e-01 7.68915832e-01 8.46326172e-01
-5.95261483e-03 1.28806555e+00 1.20537758e+00 -5.62562168e-01
-3.61533239e-02 3.46818417e-01 9.23093036e-02 8.92100573e-01
-3.24729979e-02 3.26130867e-01 -6.72265947e-01 -6.52572364e-02
8.83268118e-01 4.37612981e-01 -1.85028076e-01 -8.78880382e-01
-9.29285586e-01 9.98015285e-01 1.14520431e+00 2.77745485e-01
-9.35899764e-02 2.74327964e-01 1.42415389e-01 5.69617987e-01
5.99497139e-01 5.14620543e-01 -5.25644459e-02 4.11839694e-01
-6.04687631e-01 -1.55856714e-01 6.55968308e-01 6.73696756e-01
1.07800174e+00 4.23961468e-02 -1.73366353e-01 9.41893578e-01
6.85663998e-01 3.50464702e-01 1.67245850e-01 -9.21569824e-01
4.53866944e-02 6.86577201e-01 -3.90421778e-01 -1.01065528e+00
-1.25630200e-01 -5.17925739e-01 -8.84311795e-01 6.48772717e-01
2.42731079e-01 -3.36491436e-01 -5.33250093e-01 1.84226143e+00
-9.05470327e-02 3.79139215e-01 -4.24557216e-02 5.54758728e-01
1.12222457e+00 6.09963596e-01 1.82387576e-01 9.46862698e-02
8.73098969e-01 -9.17988181e-01 -3.75591546e-01 -3.23155105e-01
5.17230928e-01 -6.01305962e-01 8.32772315e-01 8.82504787e-03
-9.11983192e-01 -5.04755616e-01 -1.09339607e+00 2.39674971e-01
-3.12476635e-01 -1.82726234e-01 9.32080328e-01 1.65690050e-01
-1.13076818e+00 5.03284812e-01 -6.31796539e-01 -1.94892630e-01
1.03038549e+00 2.95804709e-01 -4.82413739e-01 -2.54646569e-01
-9.95128155e-01 6.85663104e-01 3.82091135e-01 -1.65598512e-01
-9.53329682e-01 -8.73662651e-01 -9.43416893e-01 9.91641171e-03
1.57089114e-01 -6.04583263e-01 1.05587471e+00 -1.45355010e+00
-1.20203435e+00 8.37998807e-01 6.12805486e-02 -3.93858016e-01
4.28488374e-01 -7.30212703e-02 5.70030175e-02 1.39166668e-01
-7.90978745e-02 1.18089831e+00 9.01499271e-01 -1.62379098e+00
-3.32668960e-01 -5.67587204e-02 3.30956243e-02 4.91364151e-01
-5.34818411e-01 -1.49538919e-01 -3.73182118e-01 -6.02678835e-01
-7.94097781e-02 -7.99689531e-01 -2.00381577e-01 6.22199178e-01
-1.65870160e-01 -5.50525010e-01 9.58053291e-01 -1.99330732e-01
6.82874143e-01 -2.37370801e+00 3.12667608e-01 4.07254308e-01
6.49740040e-01 2.00174615e-01 -4.31805223e-01 6.09012008e-01
-3.76885623e-01 -5.02101751e-03 -8.31162632e-02 -2.19045237e-01
-2.12881073e-01 9.54607204e-02 -1.50732100e-01 4.89010632e-01
4.87638503e-01 1.31197548e+00 -1.18162608e+00 -3.73780221e-01
2.87650108e-01 6.28681660e-01 -5.79663694e-01 4.22702670e-01
6.48206919e-02 5.44150770e-01 -1.91201434e-01 3.31561238e-01
4.31953967e-01 -5.82098126e-01 3.21493119e-01 -3.37863773e-01
1.99300021e-01 -1.23487994e-01 -8.97116125e-01 1.33187139e+00
-3.52800965e-01 1.22880805e+00 -2.46957690e-01 -1.08163941e+00
1.04539847e+00 -5.40567748e-02 2.79552311e-01 -4.84340012e-01
7.58147612e-02 -3.11709404e-01 6.19605109e-02 -1.71110705e-01
-5.43734524e-03 3.17077823e-02 1.91145152e-01 6.80074513e-01
4.91007835e-01 -1.36662692e-01 -1.79869905e-01 6.06494427e-01
7.55643904e-01 -1.53219610e-01 5.81431210e-01 -2.99074024e-01
2.45613798e-01 -5.09069383e-01 3.63518119e-01 7.95076132e-01
-2.75642544e-01 2.44309783e-01 4.32000041e-01 -5.88307949e-03
-6.50796115e-01 -1.48217249e+00 9.24221203e-02 1.28665459e+00
1.42451927e-01 -3.63396764e-01 -1.76625833e-01 -8.68916988e-01
2.00260475e-01 3.98604393e-01 -8.95573914e-01 -4.96368825e-01
5.94843528e-04 -2.64107138e-01 1.92947134e-01 7.05686152e-01
4.96823162e-01 -1.23745728e+00 -3.87581438e-02 -3.11462820e-01
2.39433900e-01 -5.52423179e-01 -4.96763438e-01 2.24377528e-01
-7.84764826e-01 -1.19808006e+00 -6.28284931e-01 -1.10561824e+00
8.68669212e-01 7.04818249e-01 1.01527107e+00 2.05336332e-01
-5.06548941e-01 9.28505123e-01 -2.83848345e-01 -4.34130698e-01
-4.40919966e-01 -1.22685708e-01 -1.61371350e-01 3.59556898e-02
2.16388956e-01 -6.87351108e-01 -6.54828608e-01 2.25139707e-01
-8.13408911e-01 1.39217183e-01 5.59508502e-01 9.45142031e-01
3.25741470e-01 -2.39169195e-01 6.00565255e-01 -9.20718431e-01
7.87485600e-01 -6.51378334e-01 -3.46072346e-01 1.50943488e-01
-4.91905779e-01 -4.36563268e-02 2.85556108e-01 -5.55133462e-01
-8.17134976e-01 -3.36876027e-02 3.88639808e-01 -6.40933096e-01
4.03794013e-02 4.79932606e-01 1.69544235e-01 -2.25047216e-01
7.64265120e-01 1.70517027e-01 5.42723835e-01 -1.38223916e-01
8.97293687e-01 3.59353870e-01 3.02020192e-01 -2.41766572e-01
8.66961002e-01 2.72982895e-01 -1.37997344e-01 -7.82457054e-01
-8.06943476e-01 -4.43400711e-01 -7.48222947e-01 -6.08966112e-01
5.86627007e-01 -9.66654897e-01 -6.91213608e-01 2.93630421e-01
-6.83998346e-01 -6.46661699e-01 -4.59742069e-01 4.58398223e-01
-4.78912890e-01 2.40383193e-01 -4.98847276e-01 -6.87695801e-01
-1.53716385e-01 -7.68123925e-01 2.76481330e-01 4.50566292e-01
-2.62192935e-01 -1.52448893e+00 6.26639724e-02 5.53049259e-02
3.01594675e-01 2.17382815e-02 8.00427675e-01 -7.19559491e-01
-4.45234090e-01 -2.75528431e-03 -4.75884795e-01 6.24492288e-01
4.98424947e-01 1.63301274e-01 -9.97939527e-01 -6.16475403e-01
-5.56845367e-01 -6.95823729e-01 1.27419984e+00 3.60915244e-01
1.11948335e+00 -2.49356359e-01 -4.86821145e-01 4.08522785e-01
1.36153316e+00 -3.79398316e-02 5.53725481e-01 2.19205022e-02
8.49935234e-01 6.83015168e-01 4.07769799e-01 2.34223813e-01
3.66082013e-01 3.45712602e-01 3.36503565e-01 -5.15416861e-01
-4.21693951e-01 -2.49954343e-01 3.98490250e-01 8.21353197e-01
7.33391196e-02 5.77143170e-02 -1.01947176e+00 4.50923502e-01
-1.77751100e+00 -1.11312664e+00 2.75211573e-01 2.01167798e+00
8.39181483e-01 -1.06585070e-01 2.16148108e-01 -1.42039493e-01
7.95668185e-01 1.64152473e-01 -5.57126105e-01 -1.25494018e-01
-2.86620203e-03 1.51402384e-01 8.73489082e-02 5.70891321e-01
-1.22085261e+00 9.64449406e-01 6.64826488e+00 8.28610063e-01
-1.10618532e+00 8.13428909e-02 5.91621935e-01 -8.78601298e-02
-3.13690007e-01 -2.20402077e-01 -3.36814702e-01 2.20543444e-01
4.70478177e-01 -3.82668197e-01 2.85207003e-01 6.99147224e-01
-2.56466866e-01 1.93643853e-01 -1.25161266e+00 1.06589079e+00
2.73665003e-02 -1.66967154e+00 2.01050311e-01 -1.20741017e-01
8.96077931e-01 2.94170290e-01 4.34768021e-01 5.03130376e-01
7.64807820e-01 -1.06683433e+00 1.31041288e-01 7.94801891e-01
6.71740353e-01 -1.03113830e+00 4.14605886e-01 2.55097479e-01
-1.34855485e+00 -1.72539100e-01 -3.46471757e-01 1.12302907e-01
-4.11520422e-01 3.63261491e-01 -8.59758019e-01 3.03886503e-01
6.37309313e-01 1.29042935e+00 -6.91572249e-01 1.22245097e+00
-4.50403303e-01 5.87775588e-01 1.61982417e-01 -2.19118670e-01
9.29371417e-02 -1.03547849e-01 5.09805441e-01 1.41290271e+00
-2.23815396e-01 -2.93437898e-01 3.81610066e-01 9.89516616e-01
-3.56496096e-01 2.38154784e-01 -1.09093571e+00 -5.56722842e-02
7.37703264e-01 1.31620526e+00 -7.03595698e-01 -2.38566190e-01
-4.74260807e-01 9.89380598e-01 9.05887187e-01 4.85849947e-01
-5.51781416e-01 -3.66525501e-01 6.42271399e-01 -2.56259769e-01
4.29187953e-01 -1.89894825e-01 6.80069700e-02 -9.65839446e-01
-4.30210501e-01 -7.56472528e-01 4.31451976e-01 -5.12483418e-01
-1.85891974e+00 3.88816595e-01 1.08092532e-01 -1.49732816e+00
-1.56469300e-01 -5.70424736e-01 -7.80705452e-01 6.72943950e-01
-1.43893862e+00 -1.22809112e+00 -3.78707796e-01 7.99540102e-01
4.12082255e-01 -4.33919281e-01 7.92339742e-01 -2.86693964e-02
-4.58656728e-01 7.91219532e-01 1.49771318e-01 3.88369411e-01
7.97581553e-01 -1.30526638e+00 2.25216988e-02 4.96176869e-01
5.12504280e-01 6.18558705e-01 2.31391430e-01 -3.87621909e-01
-1.08079565e+00 -1.26268744e+00 2.42262140e-01 -2.67895103e-01
8.74583364e-01 -4.17441040e-01 -9.51306581e-01 8.54021192e-01
5.41526973e-01 2.38847449e-01 1.16901517e+00 2.64766842e-01
-8.63481939e-01 -1.01495184e-01 -8.77756596e-01 6.89885318e-01
1.01522827e+00 -9.10322845e-01 -3.89285386e-01 3.26819986e-01
4.31836218e-01 1.85589343e-01 -7.97284842e-01 2.08632573e-01
5.44723690e-01 -9.49623346e-01 1.08325195e+00 -6.65528059e-01
3.93525720e-01 -6.82737306e-02 -1.94960311e-02 -1.60427320e+00
-6.95008039e-01 -4.29082900e-01 -2.15709344e-01 1.25249910e+00
5.11305153e-01 -8.90415668e-01 4.99248862e-01 1.66952714e-01
3.11269790e-01 -7.69095540e-01 -4.34947163e-01 -9.17364657e-01
4.26684558e-01 -3.80897760e-01 6.39098808e-02 1.34463310e+00
9.91674513e-02 5.25745451e-01 -2.91902274e-01 -8.00027251e-02
8.95424664e-01 9.38945562e-02 7.10070670e-01 -1.49005806e+00
-4.32834655e-01 -6.55890703e-01 -6.03727758e-01 -1.07492268e+00
4.65581000e-01 -1.64491594e+00 -1.27633870e-01 -1.43970346e+00
7.03304708e-01 -4.72904563e-01 -6.37931228e-01 6.89662099e-01
-2.60232627e-01 5.57072341e-01 5.04454195e-01 4.41368192e-01
-7.55260170e-01 6.64020896e-01 1.24672127e+00 -3.36651027e-01
-3.02006632e-01 -4.11371142e-03 -8.91049087e-01 5.79554141e-01
1.01050305e+00 -3.87314260e-01 -7.47992694e-01 -3.15599889e-02
-5.34891617e-03 -4.36570913e-01 5.06266296e-01 -7.15723932e-01
3.38527560e-01 4.64746989e-02 7.11454391e-01 -3.12807649e-01
1.43457204e-01 -5.41941643e-01 -1.88601941e-01 5.63438535e-01
-6.88288450e-01 -3.49649459e-01 2.48682022e-01 7.13800848e-01
-1.74278930e-01 -4.11263779e-02 9.32217002e-01 -2.42786314e-02
-8.98622930e-01 4.36072111e-01 -4.83863980e-01 5.28954975e-02
1.10268354e+00 -1.49627268e-01 -2.63349354e-01 -6.50278330e-01
-1.01161993e+00 2.82700032e-01 9.81069505e-02 5.36050677e-01
9.50454533e-01 -1.61970055e+00 -8.51642311e-01 2.51823068e-01
3.40195894e-01 -3.29879284e-01 1.39606729e-01 6.89877272e-01
-1.27485931e-01 2.29510348e-02 -4.73481059e-01 -9.03615773e-01
-1.55400312e+00 3.00466359e-01 4.36971366e-01 -5.97308576e-02
-5.12740612e-01 1.16298425e+00 5.69235623e-01 -5.34173965e-01
2.40323678e-01 2.59454012e-01 -5.29237688e-01 2.72388279e-01
7.33687282e-01 2.98998892e-01 -4.95553464e-01 -4.04706061e-01
-2.69945085e-01 3.30805242e-01 -4.69473720e-01 4.30657230e-02
1.38319695e+00 -6.76220935e-03 -1.91452473e-01 7.07573771e-01
1.62267756e+00 -1.86560541e-01 -1.53284693e+00 -6.56865835e-01
-1.27145648e-01 -3.41882497e-01 2.20754296e-01 -4.48402256e-01
-1.35732281e+00 7.59960771e-01 8.64264846e-01 1.32081971e-01
9.75296974e-01 4.30518597e-01 -1.76557526e-01 4.83866364e-01
1.92787781e-01 -5.57936311e-01 7.54924834e-01 2.79244989e-01
1.17176378e+00 -1.35065782e+00 1.83497518e-01 -3.95178467e-01
-7.14176536e-01 1.04657447e+00 6.81224465e-01 -5.67867994e-01
1.11975956e+00 1.56755224e-01 2.86121890e-02 -3.41003031e-01
-8.62081230e-01 -2.86922693e-01 7.10605323e-01 9.00607169e-01
5.80769122e-01 1.52859807e-01 2.89687693e-01 4.06585559e-02
2.14091063e-01 -3.10553432e-01 1.77930176e-01 8.83200765e-01
-3.24055910e-01 -1.10858107e+00 9.09791216e-02 5.38408160e-01
6.57203840e-03 -1.02872163e-01 -6.48858130e-01 8.97812366e-01
-1.97035506e-01 8.75244617e-01 3.51011038e-01 -4.27283049e-01
-1.29492898e-02 -1.39587209e-01 8.19315732e-01 -8.68502676e-01
-4.44406003e-01 -1.38008922e-01 -2.68988103e-01 -4.32092309e-01
-5.85288823e-01 -5.98325610e-01 -1.05468047e+00 -1.98395774e-01
-3.97077590e-01 -2.13633999e-02 1.42372340e-01 6.26676798e-01
3.07897449e-01 5.13104320e-01 1.03831816e+00 -1.00104749e+00
-1.24171689e-01 -8.90929639e-01 -6.60833657e-01 3.80172640e-01
4.06007856e-01 -8.00868332e-01 -5.90606391e-01 -7.30245113e-02] | [9.465226173400879, 2.930659770965576] |
8b3d0ce1-95b0-4d89-83fd-1c6a7ce045c5 | agile-gesture-recognition-for-capacitive | 2305.07624 | null | https://arxiv.org/abs/2305.07624v1 | https://arxiv.org/pdf/2305.07624v1.pdf | Agile gesture recognition for capacitive sensing devices: adapting on-the-job | Automated hand gesture recognition has been a focus of the AI community for decades. Traditionally, work in this domain revolved largely around scenarios assuming the availability of the flow of images of the user hands. This has partly been due to the prevalence of camera-based devices and the wide availability of image data. However, there is growing demand for gesture recognition technology that can be implemented on low-power devices using limited sensor data instead of high-dimensional inputs like hand images. In this work, we demonstrate a hand gesture recognition system and method that uses signals from capacitive sensors embedded into the etee hand controller. The controller generates real-time signals from each of the wearer five fingers. We use a machine learning technique to analyse the time series signals and identify three features that can represent 5 fingers within 500 ms. The analysis is composed of a two stage training strategy, including dimension reduction through principal component analysis and classification with K nearest neighbour. Remarkably, we found that this combination showed a level of performance which was comparable to more advanced methods such as supervised variational autoencoder. The base system can also be equipped with the capability to learn from occasional errors by providing it with an additional adaptive error correction mechanism. The results showed that the error corrector improve the classification performance in the base system without compromising its performance. The system requires no more than 1 ms of computing time per input sample, and is smaller than deep neural networks, demonstrating the feasibility of agile gesture recognition systems based on this technology. | ['Ivan Y. Tyukin', 'Evgeny Mirkes', 'Alexander Gorban', 'Yuxiang Huang', 'Valeri A. Makarov', 'Liucheng Guo', 'Ying Liu'] | 2023-05-12 | null | null | null | null | ['hand-gesture-recognition', 'hand-gesture-recognition-1', 'gesture-recognition', 'dimensionality-reduction'] | ['computer-vision', 'computer-vision', 'computer-vision', 'methodology'] | [ 3.66958976e-01 -3.68590564e-01 2.10235700e-01 -7.09324107e-02
-4.11676764e-01 -4.90221679e-01 3.30967128e-01 -4.70308810e-01
-8.54088306e-01 2.44615465e-01 -1.99722290e-01 -1.23608798e-01
-2.11114749e-01 -4.71985430e-01 -4.10805941e-01 -8.40091169e-01
1.29674405e-01 4.12240326e-01 2.55252808e-01 -9.24490020e-02
2.49470606e-01 8.11761379e-01 -1.88006914e+00 2.01790899e-01
1.79327235e-01 1.00361168e+00 1.28049493e-01 1.11271048e+00
7.40808100e-02 4.02970999e-01 -6.73221946e-01 -6.72097206e-02
3.77968907e-01 -2.73520380e-01 -2.28868723e-01 -5.57431206e-02
1.76107258e-01 -9.53860700e-01 -6.84989274e-01 5.46077132e-01
8.98368657e-01 3.06057781e-01 4.73723710e-01 -1.10154700e+00
-1.98123321e-01 1.76851794e-01 -6.62292764e-02 1.61891609e-01
3.23688596e-01 4.55809355e-01 7.01881111e-01 -7.89750516e-01
6.75437391e-01 8.98634613e-01 7.23195314e-01 8.49287093e-01
-1.14897048e+00 -5.15028358e-01 -3.30418974e-01 2.95794815e-01
-1.32147956e+00 -4.02445704e-01 7.91003466e-01 -4.07298326e-01
1.51318955e+00 3.12405914e-01 9.46067631e-01 1.41319227e+00
7.40789250e-02 6.78229332e-01 8.11599553e-01 -6.75267637e-01
5.79249978e-01 5.17353714e-02 1.52231589e-01 3.86970103e-01
5.92425317e-02 1.87401101e-01 -5.01762629e-01 -3.01797502e-02
1.20537126e+00 4.41635847e-01 -2.60680735e-01 -1.70189831e-02
-1.00689137e+00 6.06124461e-01 1.73533276e-01 6.15899026e-01
-7.61081576e-01 2.59069771e-01 2.74536312e-01 2.77744532e-01
-9.65401009e-02 3.35041791e-01 -2.82072246e-01 -6.87823772e-01
-1.01304352e+00 2.19849127e-04 1.04167867e+00 6.44698083e-01
-9.81266797e-02 1.86492205e-01 2.14655146e-01 4.61653113e-01
3.72700036e-01 6.15017056e-01 7.84642398e-01 -1.02653039e+00
2.48776078e-01 6.22761905e-01 4.18180451e-02 -7.87886620e-01
-4.20358330e-01 2.18546212e-01 -6.99011803e-01 6.92264915e-01
8.79239917e-01 -2.48314947e-01 -9.66584504e-01 1.19592178e+00
1.19911835e-01 -1.38634592e-01 -9.83912349e-02 1.21327567e+00
1.75747663e-01 3.64593148e-01 -2.65303463e-01 -1.08795859e-01
1.06804514e+00 -2.52246827e-01 -7.37946451e-01 2.35978991e-01
-1.40230404e-02 -5.71007133e-01 1.15144527e+00 8.63864541e-01
-7.17212677e-01 -4.15655375e-01 -1.17576587e+00 3.61239985e-02
-3.41002792e-01 1.54929340e-01 6.07088447e-01 1.05021143e+00
-8.65589321e-01 9.54016030e-01 -1.49972141e+00 -5.09742796e-01
3.61722052e-01 8.37974727e-01 -2.12961972e-01 2.50401974e-01
-6.46579385e-01 7.66420007e-01 1.70851156e-01 3.88917804e-01
-2.09488153e-01 -3.81166816e-01 -3.93287867e-01 -4.91425358e-02
1.88697036e-02 -9.38804001e-02 1.11326420e+00 -1.11880386e+00
-2.08589172e+00 3.30690384e-01 -1.24425031e-01 -2.45438263e-01
6.20976090e-01 -2.55045652e-01 -3.01311612e-01 3.97579640e-01
-6.15607023e-01 3.40517044e-01 1.26821721e+00 -7.00018585e-01
-3.83212775e-01 -6.39835715e-01 -3.26170146e-01 -1.98192000e-01
-6.71841204e-01 -5.19155040e-02 -2.51548797e-01 -3.91772032e-01
1.67320415e-01 -1.07132244e+00 2.02673361e-01 8.50314051e-02
-1.52085070e-02 -1.29232109e-01 9.61627305e-01 -8.56190979e-01
1.06408978e+00 -2.03153706e+00 6.13499619e-03 5.78575552e-01
-5.06616421e-02 4.81696814e-01 1.10819183e-01 4.21397150e-01
1.55092508e-01 -2.34306082e-01 -2.03968249e-02 -4.32016235e-03
-1.12484694e-01 2.36864790e-01 -3.28397304e-01 3.99002075e-01
1.73808083e-01 7.90961623e-01 -6.46726906e-01 -1.12111084e-01
6.05090022e-01 1.01635230e+00 -4.44463849e-01 1.68130711e-01
6.47646040e-02 2.93482095e-01 -1.69459373e-01 7.66007721e-01
4.12476808e-02 -1.33012503e-01 4.29676175e-01 -3.10611606e-01
-1.02447346e-01 -1.54621350e-02 -1.48235250e+00 1.59243011e+00
-2.53803879e-01 1.07622051e+00 1.10662982e-01 -7.46777356e-01
7.34594464e-01 5.42546213e-01 7.21616507e-01 -6.34622455e-01
4.55922335e-01 3.51781905e-01 1.48765281e-01 -9.36897457e-01
2.09768429e-01 4.88725528e-02 4.37922239e-01 6.52450919e-01
-3.09932288e-02 -2.13538632e-02 -2.26348132e-01 -3.86022359e-01
1.15202546e+00 3.37172121e-01 -2.03406125e-01 2.73636311e-01
4.73257676e-02 1.78145051e-01 -4.03191373e-02 7.35610783e-01
-1.95838332e-01 4.88364309e-01 -1.32044807e-01 -4.44464356e-01
-1.25177658e+00 -9.11354005e-01 -6.11419044e-02 8.23358834e-01
-1.33683741e-01 -1.82634383e-01 -7.57582068e-01 -8.91842041e-03
1.03314310e-01 2.86124408e-01 -3.04046005e-01 1.95983812e-01
-7.80517995e-01 -5.11591852e-01 7.85474837e-01 9.95422244e-01
3.92903417e-01 -1.15105629e+00 -1.56095552e+00 5.20903587e-01
2.84749240e-01 -6.78266048e-01 -2.90688053e-02 2.40817621e-01
-1.21590042e+00 -1.08433402e+00 -7.93652177e-01 -4.30479527e-01
3.79459888e-01 -1.65064022e-01 3.59948009e-01 -1.06429905e-01
-8.25077593e-01 8.56412649e-01 -3.33260566e-01 -5.03552675e-01
2.49907281e-02 -8.04414973e-02 4.40450281e-01 -9.15187597e-02
8.45534742e-01 -7.50198364e-01 -6.36922538e-01 -8.96911323e-02
-7.98138916e-01 -4.79426861e-01 7.06937551e-01 1.00063050e+00
3.80028784e-01 -2.51763617e-04 1.42419055e-01 -7.39661232e-02
7.38840580e-01 -3.28135826e-02 -4.16733027e-01 3.79207246e-02
-7.77913213e-01 2.29077920e-01 4.08291966e-01 -1.04334617e+00
-9.20122206e-01 4.89744395e-01 -1.15597643e-01 -6.45112097e-01
-1.79764479e-01 2.38243788e-01 1.50891036e-01 -6.15618266e-02
6.69986725e-01 2.12213829e-01 5.05931854e-01 -4.92771775e-01
1.53361991e-01 1.29758477e+00 5.69973946e-01 -2.40062639e-01
2.55930513e-01 5.08313835e-01 -1.60657614e-01 -1.49350262e+00
3.37384224e-01 -3.76054406e-01 -7.81780005e-01 -5.10095716e-01
6.71265185e-01 -4.72909957e-01 -1.34142148e+00 8.58387709e-01
-1.09362078e+00 -5.70125222e-01 -3.15011680e-01 8.46874952e-01
-3.48049462e-01 1.45093530e-01 -7.47825444e-01 -1.30732667e+00
-5.05533397e-01 -9.17712629e-01 8.57044876e-01 3.09399068e-01
-5.94714224e-01 -6.15475059e-01 -2.75463201e-02 1.71501577e-01
5.32929003e-01 8.68772268e-02 4.42766726e-01 -6.81045115e-01
-5.10655344e-01 -6.07715011e-01 2.07435206e-01 3.86840045e-01
2.21067294e-01 1.05583526e-01 -1.28070962e+00 -4.36753452e-01
1.65462717e-01 -1.47421613e-01 5.21607399e-01 4.36706841e-01
8.83554578e-01 -1.28627390e-01 -2.50841022e-01 2.39984035e-01
1.42033708e+00 5.80193996e-01 6.68751955e-01 2.12363452e-01
6.65700018e-01 5.49234748e-01 -2.39350293e-02 3.95259947e-01
-7.50472248e-02 6.73303187e-01 6.22449666e-02 3.81335616e-01
1.02778770e-01 4.34946753e-02 4.75114137e-01 6.86029136e-01
-7.71316946e-01 -2.39048954e-02 -8.93649101e-01 3.85773063e-01
-1.74707460e+00 -1.20396566e+00 -1.48397256e-02 2.22971845e+00
4.84297156e-01 -8.19838643e-02 3.21027398e-01 7.44392455e-01
4.50876623e-01 -3.36722672e-01 -9.10321355e-01 -3.78386438e-01
2.52589077e-01 7.55162179e-01 4.17820990e-01 2.96015173e-01
-8.18616092e-01 4.48825210e-01 6.63035870e+00 2.41761744e-01
-1.54200852e+00 -2.12068617e-01 -1.58615395e-01 -6.46267593e-01
3.78936887e-01 -5.87111592e-01 -4.34985518e-01 6.59733951e-01
1.12004042e+00 3.43702465e-01 8.69467080e-01 8.83493543e-01
1.77414611e-01 -1.28483951e-01 -1.24523425e+00 1.29753900e+00
1.03647716e-01 -9.48238492e-01 -3.02358478e-01 2.93010890e-01
1.30104616e-01 -1.42331362e-01 -1.14485130e-01 -7.35116005e-03
-3.45178336e-01 -1.15219772e+00 5.05769193e-01 7.38575161e-01
7.51147807e-01 -4.13088083e-01 5.90198159e-01 3.98892611e-01
-1.00909019e+00 -3.77225697e-01 6.56922860e-03 -5.61207831e-01
-3.04837581e-02 1.44175440e-01 -8.33040416e-01 -1.67768210e-01
8.86457801e-01 2.35992029e-01 -1.24693111e-01 6.34779334e-01
9.21964943e-02 6.99094534e-01 -8.96794856e-01 -4.84557897e-01
-1.37534991e-01 -1.59067810e-01 5.27934551e-01 9.92103100e-01
3.92914683e-01 5.18929362e-01 -4.52039540e-01 6.60336971e-01
3.72057170e-01 -3.31768841e-01 -5.31919420e-01 -3.75638127e-01
6.20168269e-01 8.68127406e-01 -6.47875667e-01 -2.58264363e-01
-3.57332736e-01 1.38081539e+00 -2.53028005e-01 2.12115467e-01
-1.92564666e-01 -7.61768281e-01 5.53714216e-01 -2.82620527e-02
4.31721538e-01 -5.12431443e-01 -4.49221402e-01 -1.00166488e+00
5.89192331e-01 -8.06240082e-01 6.08032010e-02 -4.92519498e-01
-1.06098711e+00 3.37752491e-01 -3.91551346e-01 -9.87293243e-01
-7.66197145e-01 -1.17892909e+00 -3.58269542e-01 8.76858890e-01
-7.08790600e-01 -6.50770545e-01 -4.44328249e-01 7.78235018e-01
4.21943575e-01 -2.94249922e-01 1.12937307e+00 2.05028638e-01
-4.29371417e-01 5.17133713e-01 3.31151515e-01 3.94210607e-01
4.34648544e-01 -1.07000077e+00 9.82908085e-02 6.86699808e-01
1.72931850e-01 7.97571659e-01 5.21600008e-01 -5.69667220e-01
-2.03988433e+00 -2.81438470e-01 5.75773537e-01 -7.07738161e-01
4.53997195e-01 -2.25217164e-01 -8.75324428e-01 5.44260323e-01
-1.43004879e-01 -2.43927076e-01 6.19384289e-01 -1.58166543e-01
-1.43018588e-01 -4.35511172e-02 -1.09852552e+00 3.73991549e-01
8.90256524e-01 -8.76810968e-01 -8.29224229e-01 -1.10472992e-01
-1.20503493e-01 -4.43368286e-01 -9.10321951e-01 2.90673655e-02
1.48264301e+00 -7.68095195e-01 9.10260320e-01 -4.80398297e-01
1.54434845e-01 -1.77445691e-02 -2.46827662e-01 -9.06525254e-01
1.99261922e-02 -4.21039581e-01 -4.59420741e-01 9.15121794e-01
-5.64874429e-03 -5.80564022e-01 1.03161931e+00 1.36856258e+00
5.53279221e-01 -4.58762050e-01 -1.07019734e+00 -8.35213840e-01
-2.32012138e-01 -7.09224403e-01 2.90494204e-01 4.05803949e-01
3.51775557e-01 7.98906535e-02 -1.03377804e-01 4.02023233e-02
6.74805939e-01 -1.03555880e-01 4.44964200e-01 -1.38416791e+00
-4.38814044e-01 -5.42975008e-01 -7.42333889e-01 -7.40425944e-01
-2.07007915e-01 -4.72681195e-01 1.22146808e-01 -1.14882886e+00
-2.04902619e-01 -1.40566930e-01 -1.13911778e-01 4.74422365e-01
2.48761833e-01 2.49129474e-01 1.47904649e-01 4.82692629e-01
2.21432045e-01 9.27838162e-02 6.25630319e-01 -1.24292359e-01
-7.98266292e-01 -8.16913396e-02 4.74432558e-02 5.19286215e-01
7.13464081e-01 -1.47798628e-01 -2.34807417e-01 -4.13081497e-01
-2.93941852e-02 -3.22015062e-02 6.68678343e-01 -1.24779129e+00
6.02991819e-01 1.48707896e-01 8.42287719e-01 -1.91184103e-01
5.42458892e-01 -1.29058325e+00 3.80753756e-01 6.76968455e-01
-1.57622546e-01 -1.01175666e-01 2.08107248e-01 4.69414264e-01
5.75530082e-02 1.79031324e-02 3.66640627e-01 -2.26327199e-02
-8.41757834e-01 -2.07836658e-01 -6.53096259e-01 -6.56144917e-01
8.04951608e-01 -6.80042267e-01 5.03010079e-02 -1.76720232e-01
-8.46367538e-01 -2.88332283e-01 2.91532218e-01 3.84797692e-01
8.09085250e-01 -1.16343081e+00 -2.21453592e-01 6.35681868e-01
-3.22112024e-01 -2.48046979e-01 7.43573979e-02 7.12629378e-01
-5.44879377e-01 5.50077319e-01 -5.42982399e-01 -7.81156421e-01
-1.34243369e+00 1.98043436e-01 3.72109026e-01 3.39703262e-01
-9.76028144e-01 5.85704207e-01 -1.13415122e+00 -1.99511275e-02
4.89829212e-01 -3.21377635e-01 8.36480483e-02 1.62533462e-01
9.06244874e-01 7.16789901e-01 2.68726379e-01 -3.21704239e-01
-6.06958389e-01 5.58291137e-01 2.33141527e-01 -7.01446533e-01
1.51776350e+00 3.00286144e-01 2.67496347e-01 6.80929661e-01
9.52697217e-01 -4.79609430e-01 -1.50584280e+00 1.73888728e-01
-3.76350880e-02 -4.54243034e-01 2.58617848e-01 -8.71415079e-01
-7.86464632e-01 1.02516854e+00 1.30398571e+00 7.96492547e-02
1.05421400e+00 -3.88971627e-01 7.83676028e-01 9.08719778e-01
4.97440308e-01 -1.51293945e+00 1.00673102e-02 3.17074835e-01
7.00450420e-01 -1.15904272e+00 -2.11465955e-01 2.16174796e-01
-4.05281216e-01 1.54094887e+00 1.86195791e-01 -3.49218816e-01
6.88505113e-01 4.31376517e-01 1.85359165e-01 -2.34792307e-01
-3.17843229e-01 -6.72375262e-02 2.30231673e-01 7.44106948e-01
3.06623012e-01 2.19935775e-01 -5.27997650e-02 7.33482540e-01
-1.01679921e-01 6.06660664e-01 2.54159033e-01 1.34113717e+00
-1.98431954e-01 -9.14988995e-01 -5.33219278e-01 6.79652691e-01
-2.02966630e-01 1.94581389e-01 -3.25502038e-01 7.95081913e-01
-1.11044377e-01 1.04887283e+00 2.55718768e-01 -8.02923977e-01
4.15338278e-01 5.79091251e-01 1.01670456e+00 -1.02377005e-01
-6.03194416e-01 -5.30084632e-02 -4.28476840e-01 -8.56882572e-01
-4.01742905e-01 -7.83149660e-01 -1.26486778e+00 -3.15617889e-01
-3.31826061e-01 -4.22707558e-01 1.17589402e+00 1.05459309e+00
3.23999941e-01 3.40932459e-01 1.50071010e-01 -1.43934667e+00
-7.57995784e-01 -9.11532223e-01 -7.59505510e-01 2.07650870e-01
3.66547048e-01 -5.88801026e-01 -2.42377728e-01 3.02883416e-01] | [6.708060264587402, -0.011256803758442402] |
37c00db7-1bdb-4642-bb4f-c8d1de9f0f39 | intelligent-reflecting-surface-aided-mobile | 2205.02194 | null | https://arxiv.org/abs/2205.02194v1 | https://arxiv.org/pdf/2205.02194v1.pdf | Intelligent Reflecting Surface Aided Mobile Edge Computing With Binary Offloading: Energy Minimization for IoT Devices | Mobile edge computing (MEC) is envisioned as a promising technique to support computation-intensive and timecritical applications in future Internet of Things (IoT) era. However, the uplink transmission performance will be highly impacted by the hostile wireless channel, the low bandwidth, and the low transmission power of IoT devices. Recently, intelligent reflecting surface (IRS) has drawn much attention because of its capability to control the wireless environments so as to enhance the spectrum and energy efficiencies of wireless communications. In this paper, we consider an IRS-aided multidevice MEC system where each IoT device follows the binary offloading policy, i.e., a task has to be computed as a whole either locally or remotely at the edge server. We aim to minimize the total energy consumption of devices by jointly optimizing the binary offloading modes, the CPU frequencies, the offloading powers, the offloading times and the IRS phase shifts for all devices. Two algorithms, which are greedy-based and penalty-based, are proposed to solve the challenging nonconvex and discontinuous problem. It is found that the penalty-based method has only linear complexity with respect to the number of devices, but it performs close to the greedy-based method with cubic complexity with respect to number of devices. Furthermore, binary offloading via IRS indeed saves more energy compared to the case without IRS. | ['Yik-Chung Wu', 'Yi Gong', 'Yizhen Yang'] | 2022-05-04 | null | null | null | null | ['total-energy'] | ['miscellaneous'] | [ 2.21772969e-01 -5.76066934e-02 -3.06680262e-01 2.84502804e-01
-1.00753628e-01 -4.65702444e-01 -4.72614095e-02 -2.11575687e-01
-1.92641541e-01 6.98022366e-01 -1.97737023e-01 -5.17867208e-01
-3.74144673e-01 -8.42579722e-01 -3.83052498e-01 -1.12043881e+00
2.09074263e-02 3.25133950e-01 -2.91460659e-02 1.38242140e-01
-1.77863300e-01 3.57287794e-01 -1.24533045e+00 -4.84149873e-01
9.61671591e-01 1.54460049e+00 3.16239417e-01 1.70051530e-01
2.72968918e-01 4.26606871e-02 -1.37226060e-01 -1.74887270e-01
2.06780732e-01 -3.01840037e-01 3.96857923e-03 6.72609359e-02
-7.24722266e-01 -4.34875898e-02 -3.76231372e-01 1.07155025e+00
8.43271017e-01 -3.03908571e-04 5.30097604e-01 -1.49389780e+00
-1.34082690e-01 2.20265701e-01 -8.10070813e-01 1.56026036e-01
-6.09478503e-02 -2.32750058e-01 4.39910263e-01 -4.66158390e-01
3.67431611e-01 3.21967393e-01 4.08608586e-01 3.58012259e-01
-7.54191577e-01 -6.09776974e-01 -1.96049418e-02 4.32515621e-01
-1.32821572e+00 -5.50095320e-01 7.73177266e-01 -1.57178313e-01
7.53800809e-01 4.08205897e-01 9.70809340e-01 3.02871466e-01
2.16737345e-01 2.60583460e-01 6.12958193e-01 -3.48180979e-01
6.68348491e-01 -8.00927281e-02 -3.16446662e-01 4.58205163e-01
7.13163376e-01 -9.20878127e-02 -1.88712209e-01 -2.01277986e-01
4.40778255e-01 1.72195777e-01 -3.50850940e-01 -1.73212096e-01
-1.17800224e+00 2.81177372e-01 2.38882661e-01 3.86578947e-01
-7.32562959e-01 2.87242472e-01 2.18443573e-01 5.38791269e-02
5.67745149e-01 -1.38005376e-01 -6.33623958e-01 -2.20778212e-01
-6.88694835e-01 -3.86918157e-01 8.20788980e-01 1.13694859e+00
2.52018571e-01 -5.00660688e-02 -6.00525923e-02 3.60442668e-01
3.73265117e-01 9.14489210e-01 1.13766484e-01 -6.20308340e-01
5.43564558e-01 2.98943669e-01 3.50744963e-01 -7.46133327e-01
-7.23628223e-01 -7.22100973e-01 -1.20714855e+00 -4.00380403e-01
3.11282575e-02 -7.73963809e-01 -3.82929385e-01 1.59713483e+00
7.03524351e-01 4.18753028e-01 -2.17764713e-02 8.91240954e-01
3.06814969e-01 7.36241758e-01 -2.77567580e-02 -8.89203370e-01
1.43607557e+00 -7.71425664e-01 -9.44287896e-01 -9.13412645e-02
5.15745759e-01 -7.50292003e-01 4.66165125e-01 1.46419853e-01
-1.19722033e+00 1.24758907e-01 -1.13086855e+00 5.44582963e-01
-3.70727032e-02 4.58610624e-01 4.65405554e-01 1.14106333e+00
-8.82096291e-01 5.36330789e-03 -8.12234461e-01 -1.44576237e-01
4.86321032e-01 6.36687040e-01 4.19806510e-01 -2.08257623e-02
-7.21706152e-01 1.76561430e-01 8.40913877e-02 -4.55033453e-03
-1.62651822e-01 -5.59378505e-01 -3.11453760e-01 4.51981723e-01
5.31509638e-01 -9.63784456e-01 8.33345234e-01 -5.46548009e-01
-1.76441860e+00 2.77802408e-01 3.49856988e-02 -6.72597513e-02
4.28065777e-01 4.77248967e-01 -5.04003704e-01 1.21239811e-01
-2.91305780e-02 -6.37234673e-02 7.58094847e-01 -6.87999666e-01
-5.96794546e-01 -6.63238883e-01 -1.59435600e-01 3.19143176e-01
-7.20442116e-01 -2.14946553e-01 -4.16823328e-01 -5.27337134e-01
1.91378221e-01 -1.45252717e+00 -2.14047253e-01 -5.57591915e-02
-3.46668422e-01 -5.85502647e-02 1.16301489e+00 -3.21191192e-01
1.21753287e+00 -2.29835916e+00 9.98983830e-02 2.00177133e-01
-6.36197478e-02 6.11951516e-04 3.56895208e-01 1.76269919e-01
4.91658539e-01 -5.94205596e-02 -4.28652428e-02 -7.71773532e-02
-3.95791493e-02 -4.52505052e-02 1.60511494e-01 6.92208767e-01
-7.43307948e-01 6.14251196e-01 -6.57033503e-01 -2.07451820e-01
1.02484196e-01 4.82184678e-01 -6.44932032e-01 -1.93410702e-02
-1.44822923e-02 5.70705175e-01 -9.14806902e-01 7.18413413e-01
8.31506371e-01 -5.52821457e-01 4.41566825e-01 -2.90969074e-01
-2.20949635e-01 -2.98864752e-01 -1.13536835e+00 1.40074241e+00
-1.03359151e+00 3.04956943e-01 5.89137375e-01 -9.76854265e-01
3.42162699e-01 7.40009546e-01 1.14037061e+00 -7.35319495e-01
3.74182314e-01 5.08906722e-01 -2.66458243e-01 -5.46853364e-01
-4.60373573e-02 -2.13833064e-01 -1.32779077e-01 5.98952174e-01
-6.90341890e-01 3.53616744e-01 -7.26685673e-02 -7.55606145e-02
1.13619184e+00 -3.49668026e-01 2.82571584e-01 -5.05522132e-01
4.16305661e-01 -3.05996358e-01 7.20754147e-01 1.24540310e-02
-6.33483976e-02 -3.05400491e-01 3.44258323e-02 -1.21355556e-01
-7.45572627e-01 -8.45482767e-01 -9.64158475e-02 8.10857594e-01
9.16336894e-01 1.65817320e-01 -6.61782205e-01 -3.75462264e-01
-8.47175941e-02 5.56166232e-01 5.57140522e-02 -1.93699092e-01
-3.09541315e-01 -1.15377736e+00 -1.81765586e-01 1.85551554e-01
6.97827995e-01 -3.50171059e-01 -1.17825282e+00 3.20738077e-01
-2.62615651e-01 -1.54941142e+00 -8.74376595e-01 6.00058027e-02
-6.03836238e-01 -7.48988688e-01 -6.50415778e-01 -6.72041714e-01
9.00085688e-01 5.98473608e-01 6.42325819e-01 -1.33327935e-02
-1.67223170e-01 4.92162913e-01 -2.48555169e-01 -4.91495878e-01
3.82508695e-01 7.31442273e-02 2.52092987e-01 3.60366315e-01
-1.38177365e-01 -6.86111867e-01 -1.05666864e+00 6.64881527e-01
-6.86226845e-01 1.42008126e-01 3.98693264e-01 4.44149941e-01
7.65962780e-01 6.87448978e-01 9.14157093e-01 -3.66404146e-01
3.56023163e-01 -8.97848308e-01 -6.92505121e-01 1.32608190e-01
-5.24138451e-01 -2.40103468e-01 8.60944271e-01 -4.70137924e-01
-8.89444292e-01 2.58036911e-01 1.16262063e-01 -5.89868240e-02
5.20201743e-01 7.67202675e-02 -6.08012080e-01 -4.53446567e-01
-1.32077903e-01 1.36584872e-02 -6.67314708e-01 -8.05276409e-02
-2.74316240e-02 9.02151465e-01 1.67877581e-02 -3.86008531e-01
5.74220240e-01 8.01737249e-01 5.13758481e-01 -9.32800055e-01
-3.74009043e-01 -3.84562165e-01 2.83518642e-01 -6.15723848e-01
8.30699980e-01 -8.82322669e-01 -1.18101585e+00 1.11339301e-01
-9.46433723e-01 -2.31750205e-01 -9.31769833e-02 7.67026544e-01
-4.44748133e-01 6.21655770e-02 -8.71268883e-02 -9.88322735e-01
-6.80324018e-01 -1.09333706e+00 9.98994827e-01 3.98794204e-01
3.23966682e-01 -7.87968576e-01 -3.82101208e-01 2.50118256e-01
6.88256681e-01 3.79330099e-01 8.46058726e-01 2.30701715e-02
-6.63219988e-01 -1.24566242e-01 -3.53342146e-01 -3.87885720e-01
1.23989969e-01 -6.03025079e-01 -4.83682483e-01 -5.25521636e-01
3.04749936e-01 4.16039079e-01 1.82607740e-01 6.68932796e-01
1.31669366e+00 -5.34160316e-01 -8.27342927e-01 7.74750650e-01
1.64449298e+00 4.87851381e-01 3.25771779e-01 -1.00993380e-01
4.00356293e-01 1.59871697e-01 5.12668550e-01 1.03159583e+00
3.74150217e-01 8.57836127e-01 7.69280732e-01 3.90870459e-02
3.06042105e-01 2.81110793e-01 1.61986828e-01 9.10315633e-01
-1.62804216e-01 -9.12857533e-01 -4.67562944e-01 3.04445356e-01
-1.83125007e+00 -4.85510528e-01 -2.25958556e-01 2.34335780e+00
7.30820671e-02 -1.05087832e-01 1.02231428e-01 2.65053391e-01
9.29716945e-01 -8.80862996e-02 -9.67759550e-01 -1.77072823e-01
2.68232077e-01 -1.32798925e-01 1.03787982e+00 -1.63353890e-01
-5.55923939e-01 1.53072432e-01 4.79321194e+00 7.26933420e-01
-1.16414237e+00 7.05799341e-01 6.46581054e-01 -3.45090657e-01
-4.18395728e-01 -3.09756100e-01 -3.09246957e-01 1.11115217e+00
8.90576363e-01 -5.68831444e-01 7.77264476e-01 8.06953967e-01
6.42603159e-01 -1.80550173e-01 -6.76003456e-01 1.29569113e+00
-3.10823470e-01 -1.14725125e+00 -7.44456649e-01 4.43300545e-01
9.17582333e-01 -6.51673041e-03 -1.28605245e-02 -3.12220812e-01
-1.89381868e-01 -1.69395059e-01 6.03767693e-01 1.93048358e-01
1.18110490e+00 -9.07413900e-01 6.63116038e-01 5.52015841e-01
-1.68399918e+00 -3.80400091e-01 1.24929762e-02 -2.92094517e-02
7.49432683e-01 1.24697113e+00 -1.13360345e-01 4.54298377e-01
6.64615273e-01 3.14154893e-01 4.85775441e-01 9.93800819e-01
5.40140644e-02 2.74635822e-01 -7.20984399e-01 -4.31463331e-01
-1.78986162e-01 -5.84075570e-01 6.11921728e-01 5.93332589e-01
1.02173543e+00 5.61015189e-01 2.74065286e-01 2.32734904e-01
-4.37538803e-01 1.73252985e-01 -3.69455516e-01 1.11732833e-01
6.30105615e-01 1.33908296e+00 -1.13162816e+00 -1.80886775e-01
-4.62315053e-01 9.40840840e-01 -4.30009007e-01 3.71843845e-01
-1.06298256e+00 -3.44565630e-01 7.18378425e-01 3.48500192e-01
2.34373152e-01 -4.33845311e-01 -6.54674172e-01 -9.73821938e-01
2.64447331e-01 5.24455346e-02 1.62240639e-01 -3.78230065e-01
-8.25670421e-01 3.32389891e-01 -4.53707248e-01 -1.38348436e+00
3.20952654e-01 -9.70933363e-02 -6.24083579e-01 2.89939851e-01
-1.63783038e+00 -4.90205526e-01 -5.18335462e-01 5.86590946e-01
3.91667634e-01 1.73224956e-01 3.66491407e-01 1.09646416e+00
-7.89999247e-01 2.91567355e-01 5.53211570e-01 -7.42648840e-01
6.45223111e-02 -5.98684967e-01 -2.96358228e-01 6.25686705e-01
-4.64767337e-01 4.00913060e-02 7.27866411e-01 -5.07656753e-01
-2.06868505e+00 -1.14613891e+00 5.87557852e-01 5.87307096e-01
5.20220399e-01 -2.35832632e-01 -2.87033934e-02 1.90469816e-01
1.74699239e-02 3.80461425e-01 7.58146644e-01 -7.01968372e-01
5.75954914e-01 -3.08382899e-01 -1.46217477e+00 6.10041440e-01
1.32987475e+00 -1.37890298e-02 7.19351113e-01 7.80046165e-01
5.67989647e-01 -2.96303600e-01 -7.41644621e-01 5.24942994e-01
5.21124423e-01 -3.93076420e-01 7.29626358e-01 1.86091438e-01
-3.88212770e-01 -3.51902276e-01 -2.32735798e-01 -1.21687961e+00
-1.86327875e-01 -1.00132954e+00 -4.85479236e-01 9.65527534e-01
2.71051019e-01 -9.69882369e-01 9.08142745e-01 4.29652929e-01
-1.35678962e-01 -9.95834827e-01 -1.45083606e+00 -9.31677401e-01
-7.82897711e-01 -3.42605352e-01 6.39789045e-01 5.75470328e-01
1.44407973e-01 4.78407413e-01 -3.58699709e-01 4.05831069e-01
7.61460662e-01 1.33974299e-01 3.23198378e-01 -1.28070295e+00
-3.57150733e-01 -1.06419951e-01 -2.24655941e-01 -1.17022192e+00
-1.68531984e-01 -7.61521399e-01 -2.02394217e-01 -1.67223179e+00
-1.26420138e-02 -1.00813508e+00 -1.52143151e-01 1.02322534e-01
2.44979978e-01 3.95889372e-01 8.52393135e-02 1.27063707e-01
-7.57327318e-01 6.04033113e-01 1.15638196e+00 -9.06845182e-02
-3.56405020e-01 5.60724020e-01 -4.73183542e-01 6.48203552e-01
8.20651889e-01 -6.08874321e-01 -6.64349794e-01 -4.79887336e-01
6.32241607e-01 4.75472808e-01 -1.11389019e-01 -1.03737509e+00
4.20898676e-01 -1.99244514e-01 -1.98409930e-01 -3.39098990e-01
3.57220262e-01 -1.65978444e+00 6.97231948e-01 8.21590841e-01
5.12295306e-01 -5.21587916e-02 -2.20675543e-01 9.18974936e-01
3.94240439e-01 1.79878578e-01 6.93110287e-01 3.05432975e-01
-9.59781632e-02 6.87553048e-01 -4.47123647e-01 -1.60491362e-01
1.74682856e+00 -1.52788356e-01 -1.22051507e-01 -2.94751793e-01
-5.90638041e-01 4.08949226e-01 2.73342252e-01 4.70372066e-02
1.25235572e-01 -1.30038357e+00 -1.39774323e-01 -3.15005369e-02
-3.12321156e-01 -2.16135070e-01 6.59248173e-01 1.34065461e+00
-3.36645991e-01 5.45929730e-01 1.53161004e-01 -4.34170127e-01
-1.12647963e+00 5.74025154e-01 8.96391869e-02 -2.62171805e-01
-3.85907441e-01 4.33672965e-01 -6.25148788e-02 2.58024246e-01
6.12778291e-02 -4.11530770e-02 1.67098776e-01 -1.05718024e-01
2.04674855e-01 1.10502255e+00 2.72027135e-01 -3.46772373e-01
-5.54174542e-01 8.94414008e-01 7.51010180e-01 3.02357554e-01
1.29493523e+00 -6.07358932e-01 -7.42484257e-02 -2.44951323e-01
1.08050358e+00 1.90169781e-01 -1.02108264e+00 6.66826740e-02
-3.16328496e-01 -3.00095290e-01 5.18659770e-01 -4.00940329e-01
-1.55790710e+00 2.11969435e-01 7.33696103e-01 6.01963580e-01
1.73715532e+00 -5.33168204e-03 1.35489297e+00 2.06875708e-03
1.04856062e+00 -1.30551481e+00 -1.19702071e-01 -8.32564533e-02
1.71251297e-01 -7.41663039e-01 7.55547062e-02 -8.68958056e-01
-9.46830288e-02 9.08959687e-01 2.46779218e-01 3.14391464e-01
1.13295829e+00 5.96150577e-01 -6.02332056e-01 -1.10816974e-02
-5.30405283e-01 -8.77245516e-02 -2.48690963e-01 2.71862686e-01
-4.05146219e-02 5.32582521e-01 -7.90219665e-01 7.40856349e-01
2.24121660e-01 1.89060360e-01 1.81327745e-01 8.36407363e-01
-3.61597121e-01 -8.69612575e-01 -3.27040792e-01 7.82590628e-01
-5.86777925e-01 2.88778275e-01 3.44096810e-01 9.23138410e-02
5.45397103e-01 1.23983777e+00 2.40346789e-02 -2.55747467e-01
3.20894480e-01 -5.13333142e-01 2.88817108e-01 -3.29895914e-01
-2.51844674e-02 9.27452073e-02 2.48720199e-02 -4.11088914e-01
-4.31159765e-01 -4.64423537e-01 -1.42448533e+00 -3.20148110e-01
-6.40174985e-01 2.13057324e-01 1.16635883e+00 1.10465729e+00
8.47102344e-01 9.20795083e-01 1.16156447e+00 -8.82583737e-01
-5.09070233e-02 -2.68107235e-01 -7.13258922e-01 -3.64561945e-01
3.57835852e-02 -6.84630156e-01 -5.29618979e-01 -3.24371785e-01] | [5.938512325286865, 1.5986454486846924] |
17f3946b-74ad-4fda-8383-896b25333816 | bae-bert-based-adversarial-examples-for-text | 2004.01970 | null | https://arxiv.org/abs/2004.01970v3 | https://arxiv.org/pdf/2004.01970v3.pdf | BAE: BERT-based Adversarial Examples for Text Classification | Modern text classification models are susceptible to adversarial examples, perturbed versions of the original text indiscernible by humans which get misclassified by the model. Recent works in NLP use rule-based synonym replacement strategies to generate adversarial examples. These strategies can lead to out-of-context and unnaturally complex token replacements, which are easily identifiable by humans. We present BAE, a black box attack for generating adversarial examples using contextual perturbations from a BERT masked language model. BAE replaces and inserts tokens in the original text by masking a portion of the text and leveraging the BERT-MLM to generate alternatives for the masked tokens. Through automatic and human evaluations, we show that BAE performs a stronger attack, in addition to generating adversarial examples with improved grammaticality and semantic coherence as compared to prior work. | ['Goutham Ramakrishnan', 'Siddhant Garg'] | 2020-04-04 | null | https://aclanthology.org/2020.emnlp-main.498 | https://aclanthology.org/2020.emnlp-main.498.pdf | emnlp-2020-11 | ['adversarial-text'] | ['adversarial'] | [ 6.13924682e-01 6.81544185e-01 2.27708727e-01 -3.90603781e-01
-8.02119195e-01 -1.23570490e+00 1.07407069e+00 2.46833295e-01
-3.00737202e-01 1.05616558e+00 3.08738112e-01 -4.11209196e-01
3.94038796e-01 -7.49680281e-01 -7.49680400e-01 -3.37147176e-01
2.06936017e-01 7.22093880e-01 -1.03942007e-01 -5.47120392e-01
5.40746516e-03 4.14197981e-01 -1.20789993e+00 8.16998899e-01
9.86560106e-01 1.18532896e-01 -4.89272803e-01 7.91873634e-01
-1.75935864e-01 9.32181060e-01 -1.42056155e+00 -1.17560744e+00
5.70235908e-01 -2.11223766e-01 -8.86620283e-01 -3.11209947e-01
7.80369639e-01 -2.70115912e-01 -2.86915153e-01 1.21607876e+00
4.83375639e-01 1.00207850e-01 8.09267044e-01 -1.49163806e+00
-9.63739872e-01 1.40616131e+00 8.00703019e-02 -1.55652642e-01
7.35060811e-01 4.16814476e-01 7.41856694e-01 -7.47847438e-01
6.87084436e-01 1.86192799e+00 9.55545127e-01 1.05987942e+00
-1.58263302e+00 -9.13761616e-01 8.70324001e-02 -1.93258122e-01
-1.40336609e+00 -3.44657779e-01 5.42751789e-01 -1.66700482e-01
1.21814263e+00 8.38966012e-01 1.96649522e-01 1.93428910e+00
3.85935187e-01 5.39612353e-01 9.27983165e-01 -7.26179600e-01
1.90170020e-01 4.48269308e-01 -1.46806583e-01 3.00891012e-01
4.34797555e-01 4.95775044e-01 -2.63109535e-01 -8.89369965e-01
-1.66072007e-02 -3.89788032e-01 -5.40419891e-02 1.02313362e-01
-1.02366424e+00 8.50677133e-01 3.08129549e-01 1.16907686e-01
-1.24502458e-01 4.31529462e-01 5.97302198e-01 4.24414963e-01
3.50503325e-01 1.21767008e+00 -5.57297409e-01 2.13037997e-01
-6.29795969e-01 8.51321459e-01 8.99509907e-01 1.33418524e+00
2.87137210e-01 2.91319788e-01 -4.86901790e-01 5.23633480e-01
5.34085743e-03 7.17924774e-01 7.47082889e-01 -6.50590301e-01
5.87188184e-01 3.32433909e-01 2.90147454e-01 -7.36385763e-01
-1.25851884e-01 -2.95756847e-01 -5.16313851e-01 3.19188178e-01
4.85427767e-01 -2.33229801e-01 -8.92276943e-01 1.79509830e+00
3.00896823e-01 -2.51008213e-01 3.89989763e-01 2.58675575e-01
4.20944899e-01 2.87764192e-01 5.83016694e-01 2.46836349e-01
1.18854427e+00 -6.26911879e-01 -6.07067525e-01 -6.05312884e-01
1.05154061e+00 -8.13852608e-01 1.37666714e+00 3.11783433e-01
-8.49348545e-01 -2.60035664e-01 -1.14683187e+00 5.20901009e-02
-8.90022278e-01 -5.49951196e-01 3.18848670e-01 1.28658855e+00
-6.68460190e-01 7.73529589e-01 -2.53954381e-01 -1.99584663e-02
3.68084222e-01 2.12950394e-01 -3.95804048e-01 -9.16992500e-02
-1.92528510e+00 1.25440991e+00 8.46933842e-01 -2.70388097e-01
-6.53647006e-01 -1.00422919e+00 -1.09200943e+00 -2.32387900e-01
1.78089797e-01 -8.39349270e-01 1.44348669e+00 -1.30691659e+00
-1.07683265e+00 8.68763685e-01 -6.65985048e-03 -7.45271146e-01
1.17499864e+00 -2.51646370e-01 -7.59175777e-01 -2.46751830e-01
-2.69931811e-03 6.21582508e-01 1.28217924e+00 -1.37372208e+00
-1.98746935e-01 -1.71229526e-01 6.88391253e-02 1.97880790e-01
-1.16202310e-01 1.46951497e-01 6.13472581e-01 -1.56379390e+00
-6.30491614e-01 -7.89520442e-01 -2.85077244e-01 -2.43610382e-01
-9.57828999e-01 1.19941577e-01 8.39958906e-01 -6.97085083e-01
9.44857538e-01 -1.90034807e+00 -3.13017040e-01 2.77528703e-01
2.30202943e-01 6.25947773e-01 -2.64511436e-01 5.19002080e-01
-4.70560580e-01 8.70870292e-01 -4.75050598e-01 -3.12734276e-01
3.65653634e-01 1.94949761e-01 -9.88634586e-01 1.82117358e-01
2.58156359e-01 1.11398327e+00 -1.21429801e+00 -2.42823243e-01
1.21612296e-01 1.70014381e-01 -5.42631149e-01 1.59872055e-01
-5.77269793e-01 3.36411968e-02 -3.00515518e-02 3.87256384e-01
5.89750171e-01 5.53002656e-01 1.35052009e-02 1.58346102e-01
8.48851621e-01 4.89081055e-01 -9.25721407e-01 1.01983440e+00
-4.81704950e-01 3.20708424e-01 -3.49271178e-01 -4.07285690e-01
7.46764183e-01 3.58219773e-01 -5.09666383e-01 -1.05760112e-01
4.08750735e-02 2.98477441e-01 -7.23180547e-02 -1.66085139e-01
8.74779403e-01 -6.24799907e-01 -6.12158298e-01 6.04567051e-01
-2.52906203e-01 -7.00379789e-01 -1.25056416e-01 5.55942059e-01
1.26682568e+00 -3.06764990e-02 5.44838011e-01 -9.47964340e-02
5.12594879e-01 2.29165509e-01 3.07163090e-01 1.40693557e+00
-5.03675044e-02 3.41968119e-01 3.00073862e-01 -4.24865812e-01
-1.34885895e+00 -1.27672720e+00 2.06236281e-02 7.33610451e-01
-3.49341542e-01 -5.86182117e-01 -8.77542973e-01 -1.32152426e+00
4.92527455e-01 1.78189993e+00 -6.60706460e-01 -7.90719271e-01
-6.59508705e-01 -3.85020852e-01 1.62203360e+00 5.41318893e-01
1.20919161e-01 -1.20030010e+00 -1.16794653e-01 3.31058681e-01
-6.48180321e-02 -1.09504402e+00 -5.44938803e-01 1.15312375e-01
-4.62923855e-01 -9.12982404e-01 -8.60916078e-02 -5.25188982e-01
8.17992985e-01 -4.78337377e-01 1.32261622e+00 2.32152984e-01
-3.85112822e-01 -6.53680936e-02 -3.96331877e-01 -7.34182894e-01
-1.71129429e+00 -6.12538494e-03 1.85301825e-01 -4.53821689e-01
4.47841644e-01 -2.67882675e-01 1.27486676e-01 1.16881788e-01
-1.39263284e+00 -1.98852494e-01 8.05501118e-02 1.01954293e+00
-3.49259446e-03 -8.82925391e-02 7.24988878e-01 -1.80906165e+00
9.31940436e-01 -3.52059186e-01 -1.99607059e-01 3.27244997e-01
-5.20726383e-01 1.45441711e-01 1.42225957e+00 -8.63612771e-01
-9.33647692e-01 -1.72611326e-01 -1.14420570e-01 -3.40055436e-01
-3.74642700e-01 7.46994093e-02 -5.48253953e-01 5.29047661e-02
1.41386104e+00 -2.71807816e-02 -3.05905312e-01 -1.67003229e-01
1.01604331e+00 8.53166878e-01 6.66383743e-01 -8.96074176e-01
1.36516929e+00 2.06908092e-01 -4.55645621e-01 -4.26884919e-01
-6.28940403e-01 4.59468752e-01 -5.18523335e-01 2.53521681e-01
2.67337263e-01 -5.80660224e-01 -1.44706130e-01 6.52111411e-01
-1.32667708e+00 -3.99918109e-01 -6.46235347e-01 -1.24787658e-01
-3.28150421e-01 5.85471928e-01 -5.15006542e-01 -6.01501167e-01
-3.05039704e-01 -8.36230695e-01 8.78987253e-01 -4.12714899e-01
-1.03693748e+00 -1.00827396e+00 -2.15538934e-01 2.30692983e-01
4.21520948e-01 4.07589674e-01 1.51800787e+00 -1.51613879e+00
8.35430920e-02 -8.05734932e-01 4.11948979e-01 4.77736622e-01
2.80398637e-01 9.61069614e-02 -1.16145265e+00 -1.67431623e-01
-6.00523055e-02 -4.01554704e-01 4.13075805e-01 -5.88492215e-01
9.14831281e-01 -1.08298349e+00 -3.71536881e-01 4.26464587e-01
9.28067267e-01 1.05052494e-01 7.56124020e-01 1.91384703e-01
8.61053646e-01 5.49611747e-01 3.61969769e-01 2.90118486e-01
-1.48456097e-01 4.69465852e-01 2.03096122e-01 6.46570474e-02
-8.93040076e-02 -6.68354511e-01 4.26157802e-01 4.15187627e-02
8.33353162e-01 -6.13022089e-01 -9.24209297e-01 3.72317284e-01
-1.30344510e+00 -1.11389291e+00 8.53085220e-02 2.07876921e+00
1.37362981e+00 4.57455546e-01 -1.87062770e-01 3.05297345e-01
9.77925956e-01 1.92531291e-02 -5.55715382e-01 -9.00746465e-01
-3.11916441e-01 4.61975962e-01 5.93760192e-01 9.02185917e-01
-1.16199732e+00 1.48725581e+00 7.27602005e+00 9.40139949e-01
-6.72674894e-01 -1.31964711e-02 4.51636761e-01 -2.59701312e-01
-8.94169927e-01 8.45546126e-02 -6.68721497e-01 5.01013517e-01
8.60235214e-01 -6.57483697e-01 5.21060526e-01 7.90599525e-01
-1.25413150e-01 4.83955652e-01 -1.46665573e+00 4.15865570e-01
2.94682741e-01 -1.14729059e+00 8.33684802e-01 -3.70279342e-01
6.70594990e-01 -5.18829942e-01 1.56803299e-02 5.31375170e-01
1.15667975e+00 -1.49126148e+00 1.11625743e+00 1.83391497e-01
7.42428422e-01 -8.57042968e-01 8.04551959e-01 3.59915733e-01
-4.60545778e-01 -1.13404639e-01 -2.56870210e-01 -9.12205055e-02
-5.72330616e-02 4.49452162e-01 -1.11210608e+00 3.78470987e-01
1.48865432e-01 2.47869045e-01 -9.22671258e-01 2.04263255e-01
-6.20523095e-01 3.86495322e-01 -1.44795597e-01 -1.44354075e-01
1.41697666e-02 4.36309218e-01 9.85149026e-01 1.33764434e+00
-1.40868962e-01 -8.58097076e-02 5.53061850e-02 1.18641067e+00
-4.20371115e-01 -8.35851878e-02 -1.10580766e+00 -2.27835655e-01
9.93916750e-01 8.47444057e-01 -1.28095984e-01 -8.01644266e-01
2.22261935e-01 1.20617700e+00 2.79629350e-01 4.19129044e-01
-7.84625649e-01 -8.24431539e-01 7.53965020e-01 -9.14425030e-02
-9.39395577e-02 2.37335294e-01 -5.06073356e-01 -1.02470875e+00
1.98563486e-01 -1.59228098e+00 4.05755579e-01 -8.95700336e-01
-1.77103603e+00 7.39170313e-01 1.41375706e-01 -9.48960602e-01
-5.80624521e-01 -4.89644140e-01 -6.78216457e-01 1.17682135e+00
-6.05259001e-01 -1.11523020e+00 1.91652298e-01 4.50567842e-01
4.03488219e-01 -3.70842516e-01 1.20988810e+00 -2.09470019e-01
-2.60635048e-01 1.42698944e+00 -9.53178853e-02 2.78374344e-01
8.15630436e-01 -1.43700588e+00 1.32672143e+00 1.08401954e+00
1.92653220e-02 9.12759304e-01 1.10439432e+00 -1.14614415e+00
-7.07179666e-01 -1.55403686e+00 1.18938231e+00 -1.04443359e+00
8.74123514e-01 -8.22832584e-01 -1.13212144e+00 1.05300164e+00
4.17090468e-02 -2.05712184e-01 6.32254660e-01 -4.02250648e-01
-9.84967828e-01 3.89069974e-01 -1.88144171e+00 1.27636337e+00
1.10479760e+00 -8.28873932e-01 -1.21327341e+00 6.23825967e-01
1.08210826e+00 -5.19182324e-01 -4.75603610e-01 1.28198549e-01
3.45230013e-01 -3.55573446e-01 9.16943491e-01 -1.29042649e+00
3.65779549e-01 -2.87195832e-01 -2.19701335e-01 -1.71942139e+00
4.22584340e-02 -9.63602364e-01 1.82412774e-03 1.29660153e+00
6.63094401e-01 -1.06188488e+00 4.22129482e-01 1.03901577e+00
-7.79932039e-03 -2.41215751e-01 -9.00679469e-01 -1.10276663e+00
7.38620996e-01 -5.99419653e-01 1.06047440e+00 1.28342116e+00
1.38372883e-01 1.03458561e-01 -1.36583984e-01 1.45786077e-01
5.49868643e-01 -6.16032004e-01 6.55684769e-01 -7.30591416e-01
-1.73739180e-01 -3.69758874e-01 -4.40055668e-01 -2.96094179e-01
6.28908634e-01 -1.27977133e+00 1.77395016e-01 -9.42007065e-01
-3.83088201e-01 -4.45770353e-01 3.26475769e-01 7.54046321e-01
-6.05837882e-01 3.03761154e-01 3.08259785e-01 3.15598994e-02
2.24660322e-01 4.84035343e-01 6.38660252e-01 -5.61760604e-01
1.29151791e-01 -2.88761586e-01 -8.99706721e-01 8.32312703e-01
8.40243638e-01 -1.08731532e+00 -2.97652364e-01 -3.11755985e-01
3.53054047e-01 -5.28181195e-01 5.80899298e-01 -6.64624691e-01
-3.79549652e-01 -4.51918840e-02 5.29020965e-01 2.13274229e-02
9.53170881e-02 -6.51386678e-01 1.13282308e-01 6.57842934e-01
-9.68573093e-01 3.46416503e-01 4.33615625e-01 4.98037606e-01
2.02693582e-01 -5.64900637e-01 7.10277557e-01 -3.70989889e-01
-1.18314564e-01 -2.32666269e-01 -8.22176814e-01 4.68839467e-01
1.01125491e+00 -9.64285657e-02 -7.82990515e-01 -2.51254767e-01
-7.90971100e-01 -2.33475268e-02 7.90169239e-01 6.32497668e-01
3.60843420e-01 -1.19037771e+00 -9.74977195e-01 2.84995168e-01
2.07637563e-01 -3.35181177e-01 -7.10691363e-02 -1.87191084e-01
-6.63857758e-01 -1.89862065e-02 7.01351762e-02 1.18339196e-01
-1.28247654e+00 9.12384570e-01 6.43445432e-01 -1.59769133e-01
-5.87067008e-01 7.46244133e-01 -7.75761753e-02 -9.51817811e-01
1.49118736e-01 -1.64826259e-01 3.29162925e-01 -3.66617411e-01
5.15287101e-01 1.46424845e-01 2.45458648e-01 -3.38159978e-01
-5.41141808e-01 -2.25267708e-01 -4.23436612e-01 -3.12496245e-01
5.18021464e-01 3.24699461e-01 -4.39099036e-03 1.54668286e-01
8.60542119e-01 4.64568079e-01 -5.91708660e-01 -2.56521821e-01
1.35946974e-01 -5.61494946e-01 -5.75398028e-01 -1.32690012e+00
-2.93602705e-01 5.40706336e-01 8.31768066e-02 2.52995551e-01
5.49288452e-01 -3.50692570e-01 7.47929335e-01 7.44731247e-01
1.55898049e-01 -7.88509727e-01 -1.52536601e-01 5.51990151e-01
1.20615518e+00 -6.59203768e-01 -3.87405930e-03 -5.47739148e-01
-8.66865277e-01 7.15781748e-01 6.85172558e-01 -7.06396848e-02
2.33779550e-01 4.70578343e-01 4.76314723e-01 2.73575187e-01
-7.77300179e-01 5.47713339e-01 -1.03635946e-02 1.00410521e+00
1.57584548e-01 2.39664301e-01 -1.92328691e-02 6.85143054e-01
-1.00666738e+00 -6.47253513e-01 7.74158180e-01 9.92771983e-01
1.45931780e-01 -1.42865598e+00 -7.11647868e-01 5.55689692e-01
-4.92288262e-01 -7.35136509e-01 -1.18687916e+00 9.63601768e-01
8.58958289e-02 1.04829681e+00 -2.40885466e-01 -4.28620398e-01
2.94218153e-01 7.03954577e-01 4.88487571e-01 -8.56505156e-01
-1.45243168e+00 -8.25128078e-01 5.69102049e-01 -2.01200455e-01
4.03459549e-01 -4.30305213e-01 -1.16879332e+00 -4.90281045e-01
-1.94506198e-01 1.96002141e-01 3.21085691e-01 1.13370991e+00
3.85291010e-01 2.36392334e-01 5.07069230e-01 -5.08446217e-01
-1.24336910e+00 -1.10184371e+00 -2.87724227e-01 9.97318447e-01
1.87393442e-01 -2.00380147e-01 -7.25617528e-01 2.84032375e-02] | [6.018619060516357, 8.14523983001709] |
309ad20c-4fd1-4ba1-9d68-5adc5dd74cb3 | parametric-depth-based-feature-representation | 2307.04106 | null | https://arxiv.org/abs/2307.04106v1 | https://arxiv.org/pdf/2307.04106v1.pdf | Parametric Depth Based Feature Representation Learning for Object Detection and Segmentation in Bird's Eye View | Recent vision-only perception models for autonomous driving achieved promising results by encoding multi-view image features into Bird's-Eye-View (BEV) space. A critical step and the main bottleneck of these methods is transforming image features into the BEV coordinate frame. This paper focuses on leveraging geometry information, such as depth, to model such feature transformation. Existing works rely on non-parametric depth distribution modeling leading to significant memory consumption, or ignore the geometry information to address this problem. In contrast, we propose to use parametric depth distribution modeling for feature transformation. We first lift the 2D image features to the 3D space defined for the ego vehicle via a predicted parametric depth distribution for each pixel in each view. Then, we aggregate the 3D feature volume based on the 3D space occupancy derived from depth to the BEV frame. Finally, we use the transformed features for downstream tasks such as object detection and semantic segmentation. Existing semantic segmentation methods do also suffer from an hallucination problem as they do not take visibility information into account. This hallucination can be particularly problematic for subsequent modules such as control and planning. To mitigate the issue, our method provides depth uncertainty and reliable visibility-aware estimations. We further leverage our parametric depth modeling to present a novel visibility-aware evaluation metric that, when taken into account, can mitigate the hallucination problem. Extensive experiments on object detection and semantic segmentation on the nuScenes datasets demonstrate that our method outperforms existing methods on both tasks. | ['Miaomiao Liu', 'Jose M. Alvarez', 'Enze Xie', 'Jiayu Yang'] | 2023-07-09 | null | null | null | null | ['semantic-segmentation', 'object-detection', 'autonomous-driving', 'representation-learning'] | ['computer-vision', 'computer-vision', 'computer-vision', 'methodology'] | [ 6.22529984e-02 1.52110696e-01 -2.25144569e-02 -5.18430352e-01
-4.60924387e-01 -6.46488428e-01 6.75542474e-01 -7.96589553e-02
-3.79268050e-01 3.49357098e-01 2.39136014e-02 -1.04768582e-01
1.30178958e-01 -1.00391209e+00 -7.09358752e-01 -6.73609495e-01
4.97926623e-01 3.09752822e-01 6.66653633e-01 -1.15640007e-01
4.97016132e-01 5.36229670e-01 -1.93992031e+00 -1.09013215e-01
9.22581613e-01 1.09507143e+00 6.33157730e-01 5.36265194e-01
-1.02148831e-01 4.16845918e-01 -3.33905905e-01 8.11194722e-03
5.93063951e-01 -1.03773184e-01 -4.53989625e-01 4.14756954e-01
4.18878049e-01 -5.94194353e-01 -1.25896782e-01 1.17584229e+00
2.53219813e-01 3.93485248e-01 5.59150279e-01 -1.44611454e+00
-1.80747524e-01 -1.73561350e-01 -7.64350712e-01 1.54709384e-01
2.12745845e-01 1.55986130e-01 6.97266161e-01 -1.08955657e+00
7.20415056e-01 1.30442786e+00 3.19515526e-01 3.68393183e-01
-9.89838481e-01 -4.99044687e-01 4.44772959e-01 3.39120775e-01
-1.34847009e+00 -4.45374608e-01 8.16879213e-01 -5.88832915e-01
9.97048974e-01 8.96199197e-02 7.49711633e-01 5.62663257e-01
5.03900886e-01 5.79199731e-01 1.20480037e+00 -1.37286052e-01
3.68868053e-01 2.15654269e-01 -4.56722034e-03 7.76740074e-01
1.04595743e-01 2.94666886e-01 -5.30970275e-01 1.97265342e-01
7.84265339e-01 -2.49671526e-02 -2.39336848e-01 -7.77849734e-01
-1.21339452e+00 9.51395571e-01 5.32630563e-01 -3.22359920e-01
-2.42534027e-01 1.22231632e-01 5.31417392e-02 8.70340317e-02
6.38951898e-01 3.47423285e-01 -3.04278016e-01 2.40904149e-02
-6.83914721e-01 2.99607635e-01 4.04352099e-01 1.05974388e+00
1.25625265e+00 -5.87517172e-02 -7.56626278e-02 5.05584121e-01
5.38818300e-01 5.13191223e-01 1.24417827e-01 -1.51538265e+00
3.35466266e-01 5.29224932e-01 2.27310002e-01 -1.03451991e+00
-5.84401667e-01 -3.64620984e-01 -3.02889019e-01 7.36684144e-01
2.96007842e-01 1.11887999e-01 -1.23090816e+00 1.71678627e+00
6.33356452e-01 2.90674139e-02 2.12205812e-01 1.20279551e+00
8.06553602e-01 4.92063493e-01 -2.89349645e-01 2.23972201e-02
1.41623724e+00 -1.18988419e+00 -5.06489873e-01 -6.17993295e-01
6.15499258e-01 -4.92818743e-01 8.59238684e-01 3.34539324e-01
-1.04049850e+00 -5.00446856e-01 -1.15088034e+00 -3.47497821e-01
-3.28336537e-01 -1.02001503e-01 5.84066391e-01 5.85680842e-01
-1.06899023e+00 1.45456314e-01 -8.80057871e-01 -4.60099846e-01
2.62882084e-01 1.49510935e-01 -3.94862235e-01 -9.27015170e-02
-8.53938103e-01 1.03912592e+00 2.52214819e-01 4.36165631e-02
-9.18666363e-01 -5.55647910e-01 -1.25836289e+00 -1.11065507e-01
5.31127989e-01 -1.15398300e+00 1.05284262e+00 -5.26561856e-01
-1.39754188e+00 6.46781385e-01 -5.15776575e-01 -3.40876222e-01
5.52771032e-01 5.10905404e-03 1.10633530e-01 3.07199597e-01
3.41799349e-01 1.11069357e+00 8.06542575e-01 -1.67272115e+00
-8.00987780e-01 -6.96656406e-01 4.13203239e-01 6.97480083e-01
5.83966868e-03 -4.55633849e-01 -5.85984468e-01 -7.60724992e-02
7.30006099e-01 -8.59469891e-01 -4.65657920e-01 2.16430411e-01
-3.25389266e-01 8.98555890e-02 8.12412083e-01 -4.31196988e-01
5.94982922e-01 -2.04466748e+00 2.52044380e-01 3.69221088e-03
5.05607724e-01 -2.31111988e-01 3.02448235e-02 5.71745215e-03
4.13529217e-01 -1.66985452e-01 -9.50227827e-02 -7.23822176e-01
-1.40297323e-01 3.96987855e-01 -2.83646643e-01 5.95951617e-01
7.42101893e-02 8.24725628e-01 -8.65032732e-01 -4.28975374e-01
6.39877915e-01 4.16831344e-01 -8.63989949e-01 1.55969903e-01
-2.60157019e-01 6.59176409e-01 -5.46456933e-01 4.23320621e-01
1.04159486e+00 2.82047596e-02 -4.03144568e-01 -2.44693115e-01
-3.84238243e-01 1.17699817e-01 -8.93652976e-01 1.73077774e+00
-6.32741034e-01 4.97352660e-01 1.43818557e-01 -7.22307205e-01
8.12485039e-01 -2.00253263e-01 3.90960842e-01 -7.04803050e-01
1.82336077e-01 -3.32413241e-02 -1.26737371e-01 -3.42027903e-01
7.58148491e-01 -8.12413543e-02 1.43869147e-02 1.24751687e-01
-2.54989773e-01 -6.46024644e-01 -1.73208356e-01 1.19882688e-01
8.49007189e-01 2.67008841e-01 2.96748757e-01 -1.85854565e-02
3.05943072e-01 1.68071285e-01 7.07260132e-01 5.17281234e-01
-2.71911323e-01 9.43844080e-01 4.80901688e-01 -7.84102231e-02
-8.90131712e-01 -1.17695236e+00 -9.41435695e-02 6.48600757e-01
6.74990058e-01 -2.57362723e-01 -7.70902097e-01 -5.68812788e-01
-5.80314770e-02 9.69300747e-01 -6.58847272e-01 -2.23491102e-01
-1.46659777e-01 -3.99301231e-01 6.01106957e-02 5.97034335e-01
6.32289350e-01 -5.78024983e-01 -1.13813686e+00 1.32940963e-01
-2.44195655e-01 -1.45483136e+00 -3.60106707e-01 1.55658558e-01
-7.07022309e-01 -8.09638500e-01 -3.51948500e-01 -2.93010354e-01
6.97483957e-01 8.42119932e-01 5.83121419e-01 -2.81835645e-01
-1.69938415e-01 5.71216404e-01 -2.74680287e-01 -4.67024326e-01
-2.31552199e-01 -3.18645418e-01 1.67457704e-02 -2.71777194e-02
2.95634657e-01 -5.08471668e-01 -7.89793074e-01 4.42814320e-01
-7.65637457e-01 5.00633121e-01 3.06733191e-01 5.69844544e-01
7.23827839e-01 -1.87638447e-01 2.76788414e-01 -5.65158486e-01
1.01790257e-01 -3.91293585e-01 -9.01090205e-01 -2.25130185e-01
-6.04602218e-01 -4.71393578e-02 2.03403428e-01 -7.61945695e-02
-1.04283869e+00 3.35335076e-01 3.45260650e-02 -7.20460832e-01
-1.16998531e-01 2.95678258e-01 -4.68910635e-01 -7.19004720e-02
3.52891058e-01 2.76524633e-01 1.52344152e-01 -1.14342064e-01
4.57698613e-01 5.62171996e-01 2.59936154e-01 -1.70272440e-01
7.30574012e-01 1.02096212e+00 1.47390664e-01 -8.42931926e-01
-8.66117120e-01 -6.72739148e-01 -7.57862806e-01 -2.24586278e-01
1.29905951e+00 -1.14507079e+00 -5.29439569e-01 3.75095576e-01
-1.32239079e+00 -1.34509221e-01 -1.30303979e-01 6.70012295e-01
-9.85556722e-01 5.79032362e-01 -9.72865149e-03 -8.19875777e-01
9.35872495e-02 -1.51210332e+00 1.41650498e+00 1.36660516e-01
1.33427531e-01 -9.18276370e-01 -2.53704280e-01 5.60389519e-01
2.78017223e-01 2.26834521e-01 7.13141501e-01 -6.94483668e-02
-9.55539167e-01 2.09960029e-01 -4.69276994e-01 1.86836287e-01
-1.23952053e-01 -3.35284233e-01 -1.28858805e+00 -4.10103910e-02
3.28946888e-01 3.62165272e-02 1.14800060e+00 4.72876728e-01
1.00513995e+00 2.74655163e-01 -4.47820336e-01 9.88975227e-01
1.23969543e+00 9.82457474e-02 5.03710449e-01 3.98867577e-01
9.11134720e-01 8.81613970e-01 1.00140035e+00 4.39815372e-01
1.03516698e+00 9.62519705e-01 1.01994085e+00 -7.89060965e-02
-1.90392479e-01 -2.85199493e-01 3.86041522e-01 4.48944092e-01
3.86817083e-02 -2.89654106e-01 -8.68306696e-01 5.55140555e-01
-1.85407996e+00 -6.46121204e-01 -2.03920141e-01 2.22547412e+00
1.97961122e-01 1.24949433e-01 -2.14532822e-01 -1.01340614e-01
3.27677399e-01 1.23927258e-01 -7.41618931e-01 -2.91570008e-01
7.60715827e-02 -3.49955767e-01 7.59878218e-01 8.46718013e-01
-8.48019958e-01 1.13516605e+00 5.30773211e+00 3.66339654e-01
-9.65906858e-01 3.52600098e-01 3.25631469e-01 -8.67064372e-02
-6.42810047e-01 3.03189248e-01 -9.29763615e-01 1.66819900e-01
5.43725193e-01 -5.10064769e-04 3.60180348e-01 8.38257134e-01
4.24297482e-01 -6.73126757e-01 -1.08522236e+00 1.15331376e+00
2.59036392e-01 -9.21877861e-01 3.13773341e-02 3.80492032e-01
4.97995555e-01 2.71839917e-01 1.72498345e-01 3.70562486e-02
5.54734096e-02 -9.88162041e-01 9.87805188e-01 4.31685627e-01
5.91285586e-01 -5.89019537e-01 4.39466029e-01 5.43343067e-01
-1.17951000e+00 -6.05327412e-02 -6.18684053e-01 -1.54898018e-01
4.83091235e-01 6.05234265e-01 -9.52522159e-01 4.60843205e-01
5.31959891e-01 7.10022092e-01 -6.60940170e-01 9.64678347e-01
-6.15642183e-02 -6.76048845e-02 -3.53498429e-01 3.11925650e-01
2.76527703e-01 -2.35996142e-01 7.71709263e-01 5.26505828e-01
5.63170016e-01 -8.43287352e-03 2.62379467e-01 1.24495685e+00
4.52936798e-01 -2.51743406e-01 -8.39664042e-01 5.03436863e-01
4.10795301e-01 1.17467284e+00 -9.04595137e-01 -1.89672872e-01
-6.34366274e-01 1.09487331e+00 1.76841319e-01 5.72775066e-01
-9.28236783e-01 -1.20886765e-01 9.09815669e-01 3.66223931e-01
4.10899043e-01 -4.64124233e-01 -4.37696725e-01 -1.21950459e+00
1.13355152e-01 -1.64602280e-01 -1.91586554e-01 -9.99749839e-01
-7.51548171e-01 5.31755924e-01 3.50912273e-01 -1.31771684e+00
-3.98240536e-01 -5.47837913e-01 -2.79572457e-01 8.95963013e-01
-1.92914963e+00 -1.15183222e+00 -6.48398161e-01 4.53024745e-01
7.50652850e-01 2.14304224e-01 5.05026877e-01 -6.44756481e-02
-1.83138981e-01 1.91149279e-01 -3.10589433e-01 -4.39924419e-01
5.49173474e-01 -1.15160954e+00 5.13931155e-01 7.49664485e-01
-1.89718053e-01 2.92894483e-01 6.82169318e-01 -4.80901480e-01
-1.43618929e+00 -1.24828529e+00 4.70977485e-01 -8.53952765e-01
1.58210337e-01 -4.81895715e-01 -7.14038074e-01 6.65304363e-01
-1.39484510e-01 2.81749219e-01 2.34304711e-01 -3.19839746e-01
-2.59548724e-01 1.31937757e-01 -1.19190180e+00 5.29561937e-01
1.30011785e+00 -3.87156129e-01 -4.24236387e-01 7.82252625e-02
8.20821524e-01 -6.87435567e-01 -5.15917361e-01 4.70095783e-01
4.50252175e-01 -1.32234752e+00 1.00205243e+00 -1.98838431e-02
3.04958940e-01 -6.14685893e-01 -3.88938636e-01 -1.41482425e+00
-2.59309411e-02 -1.67009056e-01 2.70255730e-02 8.27204585e-01
2.71317601e-01 -8.40338588e-01 7.16235578e-01 5.78008950e-01
-5.18266022e-01 -5.70643008e-01 -1.12452638e+00 -6.26091301e-01
-6.72208965e-02 -6.95406616e-01 4.21093524e-01 6.42045856e-01
-3.29872906e-01 3.16575378e-01 2.24785618e-02 5.33515394e-01
6.89831555e-01 2.61475205e-01 1.03574133e+00 -1.19127762e+00
-1.91544801e-01 -2.56689668e-01 -7.17707157e-01 -1.37107670e+00
1.96389064e-01 -7.35937417e-01 3.69531035e-01 -1.81662965e+00
1.63524330e-01 -2.39933267e-01 1.97971329e-01 1.35785699e-01
-4.21663225e-02 2.52399951e-01 2.89076447e-01 8.23982060e-03
-5.01711011e-01 8.08712602e-01 1.53950906e+00 1.41174972e-01
-2.52148271e-01 -1.24966547e-01 -5.60864806e-01 9.99057233e-01
7.53058672e-01 -1.83833838e-01 -7.72565842e-01 -6.53920054e-01
1.07480086e-01 2.14374840e-01 6.62473738e-01 -8.98321509e-01
2.79593468e-01 -3.97088915e-01 2.58269817e-01 -8.12046468e-01
9.67080891e-01 -8.25009942e-01 -1.57832086e-01 1.51778206e-01
1.88356176e-01 -1.86026230e-01 1.26433492e-01 7.77066767e-01
-1.63108587e-01 -9.10749137e-02 7.66475141e-01 -1.88422903e-01
-1.01156282e+00 2.54787594e-01 -4.84517455e-01 -1.46031603e-01
1.35335004e+00 -6.97115362e-01 -2.76825815e-01 -5.69543421e-01
-7.05090463e-01 3.93174648e-01 9.55962718e-01 5.83109677e-01
1.02720821e+00 -1.09216142e+00 -4.96839076e-01 5.39112985e-01
3.69995087e-01 4.96369332e-01 2.59578437e-01 1.03354776e+00
-3.89596075e-01 4.00460094e-01 -1.12641476e-01 -9.85582471e-01
-1.09369731e+00 5.07218957e-01 4.25968617e-01 2.46670067e-01
-7.09991276e-01 6.34103417e-01 1.00173187e+00 -5.41921735e-01
-7.96217695e-02 -6.70134485e-01 -1.76604077e-01 5.25186919e-02
3.21272463e-01 2.25013003e-01 2.29214039e-03 -1.00508642e+00
-3.54601920e-01 9.71888244e-01 -4.50822972e-02 -3.86414647e-01
1.05345917e+00 -8.13507676e-01 8.49014595e-02 3.89551133e-01
1.06608188e+00 2.00323146e-02 -1.71726322e+00 -9.74312350e-02
-4.41849887e-01 -6.75242007e-01 4.80748802e-01 -3.51390243e-01
-1.04028141e+00 1.11773503e+00 6.79393530e-01 -7.99486265e-02
8.59149754e-01 1.84579417e-01 6.89184904e-01 1.95447922e-01
6.86757088e-01 -8.76412332e-01 2.56387275e-02 6.97882473e-01
6.55997038e-01 -1.47784603e+00 -9.73141640e-02 -8.65903020e-01
-6.17134333e-01 8.82876098e-01 1.01150727e+00 9.03601199e-02
6.16580546e-01 6.65529147e-02 1.67843774e-01 -2.51614481e-01
-6.38049960e-01 -3.99901420e-01 1.21924616e-01 7.93444097e-01
-1.10050529e-01 -7.54362717e-02 2.12794133e-02 3.89288634e-01
-3.61231476e-01 -3.21347237e-01 8.47255826e-01 7.82026052e-01
-7.58668661e-01 -6.94221675e-01 -3.82707506e-01 2.97543436e-01
8.02271292e-02 -1.57448016e-02 -3.68969329e-02 5.45337319e-01
2.72429794e-01 1.03922093e+00 3.50957483e-01 -4.95439202e-01
3.29610318e-01 -2.38741353e-01 6.07996523e-01 -9.98697937e-01
2.72575226e-02 2.64818240e-02 -1.86096221e-01 -8.09351087e-01
-3.80163044e-01 -7.19275832e-01 -1.60927713e+00 -3.08617912e-02
-3.13397855e-01 -2.81263202e-01 9.35770273e-01 8.54434013e-01
3.93993884e-01 4.17991638e-01 5.88746309e-01 -1.23962820e+00
-3.48354340e-01 -6.84177935e-01 -4.95305181e-01 1.87539738e-02
5.84259689e-01 -1.12800419e+00 -5.74927866e-01 -2.64142543e-01] | [8.233762741088867, -2.423077344894409] |
26e0148d-9968-4cd9-ab31-d2511d4b7005 | training-verifiers-to-solve-math-word | 2110.14168 | null | https://arxiv.org/abs/2110.14168v2 | https://arxiv.org/pdf/2110.14168v2.pdf | Training Verifiers to Solve Math Word Problems | State-of-the-art language models can match human performance on many tasks, but they still struggle to robustly perform multi-step mathematical reasoning. To diagnose the failures of current models and support research, we introduce GSM8K, a dataset of 8.5K high quality linguistically diverse grade school math word problems. We find that even the largest transformer models fail to achieve high test performance, despite the conceptual simplicity of this problem distribution. To increase performance, we propose training verifiers to judge the correctness of model completions. At test time, we generate many candidate solutions and select the one ranked highest by the verifier. We demonstrate that verification significantly improves performance on GSM8K, and we provide strong empirical evidence that verification scales more effectively with increased data than a finetuning baseline. | ['John Schulman', 'Christopher Hesse', 'Reiichiro Nakano', 'Jerry Tworek', 'Matthias Plappert', 'Lukasz Kaiser', 'Heewoo Jun', 'Mark Chen', 'Jacob Hilton', 'Mohammad Bavarian', 'Vineet Kosaraju', 'Karl Cobbe'] | 2021-10-27 | null | null | null | null | ['gsm8k', 'mathematical-reasoning'] | ['natural-language-processing', 'natural-language-processing'] | [-2.23056540e-01 -1.08149208e-01 -3.98259670e-01 -4.09113109e-01
-1.07788646e+00 -9.77873623e-01 4.32700038e-01 3.89428049e-01
-4.09703016e-01 8.40859711e-01 2.10604772e-01 -6.80586576e-01
-5.35767555e-01 -8.90980840e-01 -8.52657914e-01 8.13144818e-02
3.39130282e-01 9.31443274e-01 -6.14939779e-02 -2.23524615e-01
5.01708686e-01 1.73645794e-01 -1.37704158e+00 7.12081075e-01
1.25529861e+00 3.33952576e-01 -1.14035182e-01 8.23438227e-01
-6.03499338e-02 8.81108224e-01 -5.03580809e-01 -7.66332626e-01
1.58066005e-01 -5.20360656e-02 -1.19786417e+00 -3.87082785e-01
9.38666761e-01 -2.69509882e-01 -4.15888280e-01 1.02998161e+00
2.45055825e-01 3.26909751e-01 6.60537899e-01 -1.26418841e+00
-1.12321496e+00 1.19986212e+00 -1.37804165e-01 3.11047912e-01
6.46684885e-01 4.49912429e-01 1.42519426e+00 -1.02227080e+00
5.80051959e-01 1.33994830e+00 7.05841362e-01 6.34727955e-01
-1.44884241e+00 -1.00763905e+00 1.25275373e-01 1.49402082e-01
-1.43372691e+00 -4.48281914e-01 4.11850244e-01 -4.54020321e-01
1.00555110e+00 2.85021156e-01 4.64573622e-01 9.99350786e-01
-9.83711928e-02 8.15725923e-01 1.33766282e+00 -3.37611586e-01
1.09870806e-02 -2.86935329e-01 6.75416172e-01 1.12216949e+00
5.90293765e-01 -2.05904543e-01 -8.41766000e-01 -8.98277536e-02
4.59077120e-01 -6.26438856e-01 -1.05460875e-01 -3.13073695e-02
-1.14466214e+00 6.44378901e-01 5.11516333e-02 2.52335995e-01
2.40391195e-01 3.23594838e-01 -2.39220890e-03 6.60155714e-01
8.89862850e-02 1.26650584e+00 -6.71502888e-01 -4.29870039e-01
-1.09499478e+00 9.31593537e-01 7.47866452e-01 8.97824168e-01
1.63678452e-01 -1.14540018e-01 -4.17364597e-01 6.44065678e-01
3.12262803e-01 5.85040152e-01 5.71798265e-01 -1.28986979e+00
9.58970308e-01 8.47301424e-01 -3.48723144e-03 -8.95579934e-01
-3.51562023e-01 -5.37803948e-01 -4.93095726e-01 -1.71982348e-01
9.42166388e-01 2.36187726e-01 -7.46357024e-01 1.84530437e+00
-1.28588662e-01 3.17048609e-01 1.35478467e-01 6.22183144e-01
8.52304339e-01 5.36411703e-01 2.60447294e-01 2.80739695e-01
1.01718783e+00 -9.53411281e-01 -2.36703187e-01 -6.45198405e-01
1.07887888e+00 -6.25755012e-01 1.44369566e+00 7.03493118e-01
-1.53210783e+00 -5.16411901e-01 -9.63242590e-01 -4.20620680e-01
-2.22272471e-01 2.25185864e-02 8.36009085e-01 6.83748662e-01
-1.04747403e+00 6.52982950e-01 -6.17958188e-01 2.04043597e-01
2.36277908e-01 2.48926729e-01 -4.81990457e-01 -5.46521008e-01
-1.22202098e+00 1.07045269e+00 4.37593371e-01 -4.22267288e-01
-6.80133045e-01 -1.25127518e+00 -1.02776694e+00 2.84765065e-01
2.78252810e-01 -5.96811771e-01 1.52659667e+00 -1.24180034e-01
-1.12761188e+00 1.12705731e+00 -1.18314698e-01 -5.11882365e-01
6.43949270e-01 -1.76802218e-01 -1.44434378e-01 -1.63946196e-01
1.40108019e-01 3.74714255e-01 1.17526248e-01 -8.58651638e-01
-5.48011065e-01 -4.80089150e-02 1.56489998e-01 -1.05494924e-01
-3.38593215e-01 -1.51386550e-02 -2.99377710e-01 -6.64549649e-01
3.64923298e-01 -7.14890242e-01 -4.36386466e-02 -2.44084671e-01
-1.63500309e-01 -7.45367587e-01 -1.39498323e-01 -5.93564570e-01
1.52559221e+00 -1.77059639e+00 2.02723742e-01 3.55828315e-01
5.03894389e-01 4.21445549e-01 -5.14350951e-01 5.74274361e-03
-1.49571318e-02 5.86149037e-01 1.99753731e-01 -4.71142769e-01
5.93719125e-01 1.47457719e-02 -4.66797233e-01 2.79170871e-01
1.78491190e-01 1.37307143e+00 -1.02819967e+00 -4.50577945e-01
1.35944365e-02 -2.58487999e-01 -9.77575958e-01 -8.04046839e-02
-3.82751226e-01 -1.47606865e-01 -3.16724390e-01 4.91177827e-01
2.93383688e-01 -5.30813515e-01 1.74535140e-01 2.75692940e-01
3.28013420e-01 8.74615490e-01 -1.52772605e+00 1.68958056e+00
-3.48546296e-01 5.22540212e-01 -2.21746847e-01 -8.77393901e-01
6.07517600e-01 -6.59362897e-02 -1.49159819e-01 -8.30243111e-01
-2.66397476e-01 2.93343782e-01 4.09773499e-01 -2.88501352e-01
6.95346177e-01 1.90093488e-01 -2.13820443e-01 6.25967205e-01
-4.49680500e-02 -3.78401637e-01 5.42112589e-01 3.93913180e-01
1.22046161e+00 -1.09681666e-01 -1.47996321e-02 -4.47482109e-01
5.00589252e-01 9.08826441e-02 6.20840549e-01 1.19686508e+00
3.65583249e-03 3.11004460e-01 5.29954553e-01 -3.54538828e-01
-1.01304913e+00 -1.09723449e+00 1.63457274e-01 1.43974257e+00
-7.12342784e-02 -1.01584947e+00 -4.85007972e-01 -5.78661442e-01
4.04868811e-01 9.70055521e-01 -3.02281469e-01 -3.55776012e-01
-7.44355559e-01 -4.56765562e-01 1.08324850e+00 7.06704140e-01
2.30550200e-01 -6.60545647e-01 -3.18763629e-02 1.84684634e-01
-2.33927459e-01 -1.18835557e+00 -1.00281477e-01 -1.47989005e-01
-6.52503490e-01 -1.14884961e+00 -2.00960010e-01 -7.71762908e-01
5.85994840e-01 1.10757034e-02 1.70835733e+00 6.86441064e-01
-4.92691211e-02 2.41443843e-01 8.56303200e-02 -1.28830895e-01
-8.25500488e-01 2.27217793e-01 4.78782713e-01 -9.70466733e-01
4.29064304e-01 -3.83731484e-01 1.63425617e-02 3.60692739e-02
-4.03576791e-01 2.49339476e-01 1.86504155e-01 7.38904476e-01
2.46838048e-01 3.87173891e-01 4.05278325e-01 -6.78394854e-01
8.72961879e-01 -1.49319485e-01 -8.12523127e-01 6.29184425e-01
-8.58987868e-01 3.95264238e-01 6.69835806e-01 -5.03014266e-01
-3.79589707e-01 -4.37395215e-01 2.40665406e-01 -1.30289599e-01
3.65236066e-02 6.93142653e-01 1.22124180e-01 -1.05551682e-01
7.19195485e-01 1.42389342e-01 -4.55311090e-01 -4.47144955e-01
3.67060691e-01 1.65029794e-01 7.24492013e-01 -1.61870039e+00
1.09255326e+00 -4.06228721e-01 -3.24284285e-02 -2.54483908e-01
-1.23465729e+00 -6.05983958e-02 -2.10409939e-01 1.04198292e-01
4.00842458e-01 -1.08336437e+00 -1.28032279e+00 4.33360606e-01
-1.14014685e+00 -8.71432424e-01 7.89336488e-02 4.64564294e-01
-2.41261691e-01 -3.20917517e-02 -1.11639106e+00 -5.48089921e-01
-1.16375268e-01 -1.19804096e+00 7.61110902e-01 2.21756265e-01
-5.54556549e-01 -9.76552427e-01 -2.10358873e-01 1.05372560e+00
2.66206175e-01 -3.86814892e-01 1.52079010e+00 -8.34625602e-01
-7.03067660e-01 -7.86385383e-04 -1.73906341e-01 1.24942489e-01
-4.78488564e-01 1.40475631e-01 -6.43216133e-01 -1.67581603e-01
-5.56280196e-01 -9.52903807e-01 7.13628948e-01 -3.62800760e-03
1.48093152e+00 -1.45571798e-01 -7.80765265e-02 4.77535635e-01
1.05637574e+00 -5.23333013e-01 3.78789335e-01 3.66311342e-01
5.72436690e-01 1.06214330e-01 3.19048554e-01 -1.76913999e-02
8.21101904e-01 3.44911933e-01 -2.62020469e-01 4.26173776e-01
-1.85427263e-01 -6.78255796e-01 3.08357894e-01 9.89921451e-01
9.28182974e-02 -3.09701085e-01 -1.62516558e+00 6.22450650e-01
-1.65815961e+00 -8.27484787e-01 -1.02502987e-01 1.95251000e+00
1.32208753e+00 8.04507256e-01 -2.80676544e-01 2.81334788e-01
1.58228323e-01 -1.20454282e-01 -2.00865611e-01 -3.06051731e-01
-1.65797904e-01 8.41348827e-01 4.02266383e-01 9.64983225e-01
-8.38572741e-01 1.40635693e+00 7.11876535e+00 1.15041971e+00
-7.50319123e-01 -1.25983015e-01 6.26955450e-01 -2.39036009e-01
-6.43937588e-01 -2.77479947e-01 -1.04979587e+00 2.31163159e-01
9.98255670e-01 -4.53767985e-01 7.40451217e-01 6.75790429e-01
-8.15100297e-02 8.99186656e-02 -1.48814332e+00 9.21885550e-01
2.87208017e-02 -1.61807156e+00 1.21015191e-01 -1.06068306e-01
1.14923692e+00 -2.37148240e-01 1.27961800e-01 9.50010180e-01
9.14029777e-01 -1.79126084e+00 7.84920633e-01 5.02910614e-01
4.32369709e-01 -6.27076864e-01 3.77005488e-01 6.93173468e-01
-9.16997254e-01 -1.05876945e-01 -1.85670152e-01 -6.71285927e-01
-3.56948704e-01 1.76154882e-01 -7.10401118e-01 1.97245136e-01
2.67343283e-01 3.57247680e-01 -1.31213558e+00 9.55203533e-01
-6.75893009e-01 9.99966979e-01 -2.66640633e-01 -2.26024598e-01
3.26242238e-01 1.16767250e-01 1.34001076e-01 1.03809440e+00
1.03185020e-01 3.75318825e-01 4.73217785e-01 1.26215827e+00
-3.40853572e-01 -9.99678150e-02 -3.74617845e-01 -4.16017562e-01
8.09335411e-01 8.82978618e-01 -4.67467487e-01 -5.40580869e-01
-1.76192909e-01 5.81888199e-01 8.75821948e-01 2.13555172e-01
-7.65713334e-01 7.09766336e-03 7.28486955e-01 9.82922167e-02
-2.06497222e-01 -6.68558776e-01 -8.24367702e-01 -1.39627385e+00
1.71727911e-01 -1.37227833e+00 5.07515848e-01 -1.01298225e+00
-1.39431202e+00 -3.05282742e-01 -8.17591324e-02 -2.52495587e-01
-2.98286319e-01 -1.01263165e+00 -4.12322730e-01 9.60424900e-01
-1.16369069e+00 -9.84741390e-01 -8.75395834e-02 2.60143876e-01
4.56219345e-01 -2.42945865e-01 8.59418809e-01 3.26237202e-01
-4.25410837e-01 1.09698141e+00 -2.33299494e-01 5.23925483e-01
6.58887386e-01 -1.43164897e+00 7.97595203e-01 9.73837197e-01
4.29905474e-01 1.13055789e+00 7.11430430e-01 -7.59860456e-01
-1.58599591e+00 -7.88933218e-01 1.42077971e+00 -1.05753231e+00
9.77317572e-01 -3.25762212e-01 -8.79970133e-01 8.32367659e-01
-2.86296636e-01 -1.00831144e-01 4.82522130e-01 4.91369247e-01
-7.40247488e-01 2.78004885e-01 -9.09638643e-01 8.52788568e-01
1.25409126e+00 -7.81457424e-01 -1.22239828e+00 5.01517355e-01
7.33098924e-01 -8.85469675e-01 -9.57861066e-01 3.89105886e-01
4.26941156e-01 -3.27051759e-01 1.21007621e+00 -1.50258553e+00
1.02569687e+00 -1.08599164e-01 -2.50189662e-01 -1.09800744e+00
-6.11980915e-01 -3.67059886e-01 -2.88280427e-01 1.23769128e+00
6.56737745e-01 -2.73924381e-01 9.91588354e-01 1.29166913e+00
1.89163517e-02 -7.61817217e-01 -6.89149976e-01 -7.65833378e-01
7.38051236e-01 -8.04937005e-01 7.26935804e-01 1.27063966e+00
1.64377004e-01 2.28983790e-01 1.48800850e-01 2.47671694e-01
5.20023584e-01 3.51477325e-01 7.62782454e-01 -1.28814805e+00
-4.32511628e-01 -8.98777544e-01 -3.43614578e-01 -8.55377078e-01
8.14759254e-01 -1.34964931e+00 -2.91174948e-01 -1.59360623e+00
2.79498994e-01 -7.30755210e-01 -2.99306929e-01 8.26519787e-01
-5.43020129e-01 3.56748462e-01 1.75843313e-01 -4.29693639e-01
-9.43209171e-01 1.52591988e-01 1.22252202e+00 -3.55492562e-01
9.63656306e-02 -3.70720387e-01 -9.76533294e-01 7.89644361e-01
6.68474138e-01 -2.20771208e-01 -4.37321931e-01 -8.07548225e-01
6.15267277e-01 -3.36202160e-02 5.16962588e-01 -1.18022573e+00
5.22843897e-01 -7.00428009e-01 6.13422513e-01 -2.48022139e-01
5.55087924e-02 -1.54101983e-01 -3.39814723e-01 4.66679811e-01
-9.73556399e-01 4.67201054e-01 3.31288040e-01 -8.45220871e-03
1.86527178e-01 -3.56392205e-01 5.73393345e-01 -9.14487466e-02
-6.84865654e-01 1.26784280e-01 -1.38391972e-01 8.14457536e-01
5.66084623e-01 2.48022184e-01 -7.17641652e-01 -1.10725172e-01
-3.44041198e-01 6.17156029e-01 3.30439001e-01 5.30302644e-01
3.87388468e-01 -1.39719534e+00 -9.53268349e-01 1.44100159e-01
3.99332792e-02 2.03720257e-02 -1.62985995e-01 2.72927582e-01
-5.49010694e-01 6.67300284e-01 1.83391958e-01 -1.27616256e-01
-1.29278743e+00 3.23084086e-01 3.65834415e-01 -5.63262939e-01
-4.58375782e-01 1.20536554e+00 -4.65078413e-01 -6.97056115e-01
3.36856127e-01 -7.75678813e-01 1.20654330e-01 -4.10506099e-01
6.58423066e-01 3.67934108e-01 2.08579317e-01 -9.39087644e-02
-2.67659515e-01 3.09541881e-01 -1.13262916e-02 -1.25344157e-01
1.26680434e+00 5.36776900e-01 -1.09748289e-01 6.64866567e-02
6.26345754e-01 2.99713224e-01 -5.47749043e-01 -3.81332695e-01
2.80273587e-01 -4.11527038e-01 -1.13033131e-01 -1.12812793e+00
-7.07131982e-01 5.25595784e-01 -1.75209999e-01 -1.53743982e-01
4.07880932e-01 -2.90495992e-01 5.75096905e-01 9.27668333e-01
4.59057420e-01 -1.04875696e+00 7.14769587e-02 9.98977005e-01
5.98183632e-01 -1.17435789e+00 1.73515186e-01 -3.33042920e-01
-1.38627678e-01 9.09996450e-01 1.19591892e+00 -8.82010981e-02
1.49526030e-01 2.86469102e-01 -3.57027769e-01 -1.23718031e-01
-1.20224059e+00 2.90637374e-01 7.18572736e-01 2.24269331e-01
6.34814799e-01 2.91158736e-01 -2.03908026e-01 1.11545181e+00
-1.07683861e+00 3.81966569e-02 4.37143862e-01 6.12459421e-01
-3.98216426e-01 -9.95689094e-01 -4.26801085e-01 4.49727565e-01
-3.81780356e-01 -4.56410736e-01 -3.73280048e-01 7.42655277e-01
-3.40059511e-02 1.02995074e+00 -2.19494700e-01 -3.08596164e-01
1.72329336e-01 5.46342015e-01 9.16496694e-01 -6.39612317e-01
-5.73379219e-01 -8.11850071e-01 2.36356676e-01 -6.58892572e-01
3.62393647e-01 -3.69764358e-01 -1.30581295e+00 -1.04932773e+00
4.09151353e-02 2.48249426e-01 3.09363335e-01 1.32981932e+00
7.32164755e-02 4.56452817e-01 -2.25877732e-01 -5.76344244e-02
-1.09902751e+00 -7.04478204e-01 -1.99437603e-01 4.99179989e-01
7.72438049e-02 -4.89738315e-01 -1.80433467e-01 -1.80754811e-01] | [9.712186813354492, 7.365890979766846] |
222241d0-0677-4bf7-8f69-362bf949c8a3 | financial-risk-management-on-a-neutral-atom | 2212.03223 | null | https://arxiv.org/abs/2212.03223v1 | https://arxiv.org/pdf/2212.03223v1.pdf | Financial Risk Management on a Neutral Atom Quantum Processor | Machine Learning models capable of handling the large datasets collected in the financial world can often become black boxes expensive to run. The quantum computing paradigm suggests new optimization techniques, that combined with classical algorithms, may deliver competitive, faster and more interpretable models. In this work we propose a quantum-enhanced machine learning solution for the prediction of credit rating downgrades, also known as fallen-angels forecasting in the financial risk management field. We implement this solution on a neutral atom Quantum Processing Unit with up to 60 qubits on a real-life dataset. We report competitive performances against the state-of-the-art Random Forest benchmark whilst our model achieves better interpretability and comparable training times. We examine how to improve performance in the near-term validating our ideas with Tensor Networks-based numerical simulations. | ["Didier M'tamon", 'Hacene Isselnane', 'Oumaima Hammammi', 'Achraf Seddik', 'Roman Orus', 'Michel Kurek', 'Irene Caceres', 'Samuel Mugel', 'Andoni Duarte', 'Faysal Ishtiaq', 'Luc Andrea', 'Maitree Shah', 'Usman Ayub Sheikh', 'Gianni Del Bimbo', 'Loïc Henriet', 'Adrien Signoles', 'Vincent E. Elfving', 'Julia R. K. Cline', 'Boris Albrecht', 'Sebastian Grijalva', 'Luis Ortiz-Guitierrez', 'Lucas Leclerc'] | 2022-12-06 | null | null | null | null | ['tensor-networks'] | ['methodology'] | [-1.20518602e-01 2.85419285e-01 -6.42099511e-03 -5.95880628e-01
-8.77149582e-01 -4.82072622e-01 6.17311239e-01 4.29851115e-01
-3.91084611e-01 7.27714658e-01 2.18643323e-01 -7.86964417e-01
-3.82827342e-01 -1.21894455e+00 -4.46458787e-01 -5.14154494e-01
-1.67763814e-01 1.03575039e+00 -1.65323302e-01 -6.28891945e-01
8.47919703e-01 2.36486360e-01 -1.10446143e+00 6.67036295e-01
5.09756148e-01 1.22349966e+00 -7.08723307e-01 7.87144482e-01
4.04041678e-01 1.06227469e+00 -3.61005887e-02 -1.12179184e+00
6.82501674e-01 2.50053734e-01 -1.20298052e+00 -6.45184994e-01
2.42933676e-01 -3.66300851e-01 -3.88658255e-01 1.23776674e+00
3.73010516e-01 -9.42082424e-03 6.85803354e-01 -7.33991921e-01
-7.99830317e-01 7.23851621e-01 -7.02165887e-02 4.77907002e-01
1.01479828e-01 2.84561634e-01 1.74751973e+00 -4.81067806e-01
5.99320889e-01 1.10180604e+00 6.37372613e-01 2.12750331e-01
-1.34726965e+00 -3.63715291e-01 -8.40768814e-01 8.11548948e-01
-7.50369072e-01 -2.36289337e-01 2.90746808e-01 -3.22670966e-01
1.52585125e+00 1.94940776e-01 5.56750238e-01 7.57467031e-01
6.10516787e-01 1.39785096e-01 1.52401841e+00 -4.67150450e-01
5.23722172e-01 7.87748843e-02 5.08622706e-01 7.64094651e-01
4.73646522e-01 8.20732296e-01 -1.06668842e+00 -4.57494378e-01
3.35061640e-01 -2.00693697e-01 3.46265137e-01 1.10431813e-01
-1.23451662e+00 1.17219210e+00 7.67091453e-01 -1.61165521e-01
-4.50303763e-01 5.68943262e-01 3.73598695e-01 5.32086670e-01
6.70091689e-01 8.22008193e-01 -6.54026031e-01 -3.48412514e-01
-7.78236687e-01 5.18025041e-01 9.48556244e-01 3.12368900e-01
6.23662174e-01 -1.88992530e-01 2.29721308e-01 -2.01216519e-01
4.82680887e-01 3.56601328e-01 2.88758785e-01 -1.24841797e+00
5.98715067e-01 2.73553342e-01 2.65084565e-01 -5.96668482e-01
-6.70555115e-01 -6.38242245e-01 -8.16345572e-01 6.12389803e-01
6.17815256e-01 -1.26348779e-01 -5.07085323e-01 1.04690778e+00
1.84104592e-01 1.66710272e-01 3.98134947e-01 8.38621616e-01
9.27921236e-02 5.92953384e-01 -1.28550097e-01 -3.42317708e-02
1.61784613e+00 -6.79188669e-01 -4.31982070e-01 1.10755721e-02
1.13008356e+00 -6.73917651e-01 4.98577356e-01 9.71888959e-01
-8.57797682e-01 8.85157939e-03 -1.36247706e+00 -2.34337226e-01
-3.59945983e-01 -3.58694404e-01 1.45094669e+00 1.13114822e+00
-8.21813047e-01 1.58750165e+00 -1.05573344e+00 1.53374910e-01
3.42161864e-01 5.67345679e-01 -1.45593971e-01 3.52515668e-01
-1.51069582e+00 1.35247207e+00 2.07439825e-01 1.44707263e-01
-5.49908459e-01 -4.73233253e-01 -1.06634773e-01 1.94584265e-01
-5.93828373e-02 -1.03660202e+00 1.48269558e+00 -4.90034878e-01
-1.66618061e+00 7.45721221e-01 1.98536575e-01 -1.17444372e+00
3.30175579e-01 -6.01654388e-02 -3.01595718e-01 1.50724933e-01
-1.78393900e-01 9.74885598e-02 4.22531039e-01 -1.00891842e-02
-3.03529233e-01 -6.96295857e-01 2.87488997e-01 -2.14967370e-01
-2.36328945e-01 2.36823708e-01 8.87392759e-01 -3.45144212e-01
3.90420705e-01 -1.18927848e+00 -5.60970902e-01 -6.10935628e-01
-2.91864097e-01 -8.27578828e-02 -2.05347985e-01 -5.73741496e-01
8.77516568e-01 -1.32974327e+00 1.39967218e-01 3.70354086e-01
1.63907692e-01 -1.00985132e-02 4.09135610e-01 4.84635532e-01
-3.02785814e-01 2.35038996e-01 -2.19104998e-02 -2.02677026e-01
3.21496576e-01 -4.05205525e-02 -3.89472127e-01 6.14606619e-01
2.08090112e-01 9.79678035e-01 -6.78346694e-01 -1.26755267e-01
-2.29595050e-01 -3.67447287e-02 -9.10756826e-01 -1.97205886e-01
-2.49991510e-02 1.89103961e-01 -5.02915561e-01 6.62725210e-01
5.67340791e-01 -4.18377161e-01 1.44657731e-01 -7.46540353e-02
-8.65980610e-03 8.66619527e-01 -1.13683081e+00 1.37964559e+00
-2.71581858e-01 4.13716644e-01 -3.37889791e-01 -1.06359088e+00
5.55817783e-01 2.95220137e-01 3.48497182e-02 -8.22450042e-01
1.23127334e-01 4.77349550e-01 3.74398887e-01 -3.86320233e-01
6.17190838e-01 -8.21903110e-01 -5.20389639e-02 9.74857092e-01
1.42774507e-01 -2.52931237e-01 1.67137295e-01 2.26798698e-01
1.18693531e+00 6.03613630e-02 1.81840047e-01 -4.89327371e-01
1.44413069e-01 3.87512781e-02 2.38787800e-01 8.74160886e-01
-4.83796775e-01 3.39377731e-01 7.86324322e-01 -1.08326638e+00
-1.46703851e+00 -9.77245569e-01 -6.47209942e-01 8.59014630e-01
-2.18978271e-01 -7.50864148e-01 -7.32355654e-01 -2.88018346e-01
1.03808239e-01 8.25438440e-01 -5.80790222e-01 -2.46998489e-01
-2.55222499e-01 -1.94763029e+00 5.87132871e-01 3.21419716e-01
1.45077199e-01 -7.22828031e-01 -4.33087379e-01 4.15688783e-01
3.36664140e-01 -1.06532574e+00 7.04041302e-01 6.41004145e-01
-1.19395173e+00 -8.79458606e-01 -2.90760887e-03 3.70119363e-02
1.75740749e-01 -5.18935800e-01 1.12662351e+00 -1.25120074e-01
-9.89890248e-02 -5.55867910e-01 -2.34653220e-01 -4.66048807e-01
-6.44048333e-01 1.09942988e-01 3.75360429e-01 -1.44764930e-01
4.41037059e-01 -5.73227108e-01 -6.30192339e-01 -2.78647542e-01
-6.76295578e-01 8.76298770e-02 6.56426311e-01 9.66120243e-01
1.67590722e-01 -8.04670826e-02 3.34668905e-01 -1.07379544e+00
5.27374625e-01 -4.02529627e-01 -6.52623773e-01 1.29420087e-01
-1.25291216e+00 8.36367369e-01 7.49455273e-01 2.73103744e-01
-7.12761462e-01 -2.35889524e-01 -1.47977211e-02 5.23961604e-01
1.52672499e-01 8.27819169e-01 4.62214619e-01 -4.60534930e-01
9.61262465e-01 -4.71209019e-01 -2.80081898e-01 -5.95394015e-01
6.12253726e-01 8.20009589e-01 2.47891560e-01 -6.68128848e-01
9.39898372e-01 6.58490062e-01 8.41771901e-01 -2.32840523e-01
-1.24138963e+00 -8.64382237e-02 -7.82485843e-01 1.18423365e-01
9.63926911e-01 -7.33842731e-01 -9.32226837e-01 4.71992671e-01
-1.10463238e+00 2.05637261e-01 -5.84358238e-02 8.37719977e-01
-6.30596519e-01 3.09149057e-01 -1.16439462e+00 -9.51866627e-01
-6.60024643e-01 -1.02812171e+00 8.92306864e-01 3.25943619e-01
7.32893348e-02 -9.54243958e-01 3.50644827e-01 1.04097760e+00
4.55346555e-01 1.20706394e-01 9.72617626e-01 -7.60681450e-01
-9.30145562e-01 -4.54874218e-01 -3.48016113e-01 3.96400362e-01
-8.44357729e-01 -9.57960784e-02 -1.26612520e+00 -4.19282727e-02
1.38552785e-01 -4.91215467e-01 1.02542949e+00 -8.92169178e-02
7.62962878e-01 -1.97024092e-01 3.69512856e-01 6.58787131e-01
1.37179208e+00 -4.23974782e-01 6.46292508e-01 8.78107786e-01
4.48353142e-01 3.86669636e-01 5.68756223e-01 4.91231203e-01
4.67800051e-01 5.70212543e-01 6.66771591e-01 5.60639203e-01
5.53030789e-01 1.60849214e-01 5.54197252e-01 1.07811260e+00
-8.60185206e-01 3.18123609e-01 -1.18372750e+00 -1.35950297e-01
-1.70690084e+00 -1.29200637e+00 -4.89259988e-01 2.13078117e+00
7.11945832e-01 6.70822322e-01 -1.75045773e-01 2.04598039e-01
2.75137246e-01 -1.37744293e-01 -1.88422516e-01 -1.10455406e+00
-1.02877639e-01 1.00055540e+00 1.12319589e+00 5.15830636e-01
-1.06976032e+00 6.95421696e-01 6.45885706e+00 5.97939372e-01
-8.10069263e-01 5.51702201e-01 8.77862036e-01 -3.13637704e-01
-2.01577201e-01 8.52832377e-01 -6.48008764e-01 1.41934767e-01
1.90081751e+00 -2.06520781e-01 7.69067526e-01 6.40125394e-01
2.22119763e-02 -5.08443266e-02 -1.16665483e+00 6.40527844e-01
-4.23980087e-01 -1.61735344e+00 -2.32640535e-01 3.68955344e-01
6.61430717e-01 6.74813390e-01 1.19155563e-01 2.58790582e-01
3.15745801e-01 -1.24832392e+00 8.51351500e-01 7.75856555e-01
2.27137551e-01 -8.33171904e-01 1.05052149e+00 3.67243439e-01
-4.07076329e-01 -4.43969727e-01 -7.99215794e-01 -8.34196508e-01
2.04273596e-01 6.81913972e-01 -6.69562221e-01 9.26107705e-01
6.50392711e-01 6.19681954e-01 -9.32757914e-01 8.88335407e-01
-2.61429548e-01 8.69481802e-01 -4.43002671e-01 -3.99295419e-01
4.35625911e-01 -6.56698227e-01 2.49682337e-01 7.08049715e-01
2.21502841e-01 2.41725266e-01 -5.55756271e-01 8.72424960e-01
6.45893142e-02 -7.35674575e-02 -1.86013669e-01 -2.29639366e-01
-3.14981222e-01 1.28055191e+00 -5.56860089e-01 -5.19634545e-01
-1.48011670e-01 6.30936205e-01 4.28458184e-01 -1.19265348e-01
-7.26682007e-01 -4.08562213e-01 4.48200285e-01 -1.83880955e-01
4.70543168e-02 -3.54540199e-01 -9.44370389e-01 -1.67693353e+00
-1.38921171e-01 -6.64216995e-01 2.35772878e-01 -7.82696187e-01
-1.49159849e+00 3.49135429e-01 -4.07303691e-01 -1.01307809e+00
-4.34716225e-01 -1.31297684e+00 -4.67184037e-01 8.53286088e-01
-1.48682964e+00 -7.33103037e-01 4.25093889e-01 -1.35188356e-01
-4.62417603e-01 -2.58401245e-01 1.22710347e+00 2.68445194e-01
-6.18998468e-01 1.43711284e-01 8.11883807e-01 -3.26816402e-02
4.82297838e-01 -1.68727863e+00 7.80357242e-01 7.52008975e-01
5.56757212e-01 6.75270855e-01 1.00593114e+00 -2.94741631e-01
-1.42813230e+00 -5.28328717e-01 1.02676332e+00 -9.20435011e-01
1.36284161e+00 -4.22227353e-01 -5.26527584e-01 5.42477667e-01
1.57836359e-02 1.16098765e-02 8.35960269e-01 4.68131185e-01
-6.47530556e-01 -1.39254078e-01 -1.05588186e+00 1.00014113e-01
5.87866485e-01 -1.06976771e+00 -8.09758306e-01 1.04701078e+00
3.19491714e-01 -4.76483889e-02 -1.20874977e+00 1.74954966e-01
6.61311686e-01 -1.35733032e+00 9.44603324e-01 -1.30483389e+00
6.51315570e-01 8.44096765e-02 -1.54243350e-01 -1.18871546e+00
-2.35235006e-01 -9.00540173e-01 -2.44863294e-02 4.88124907e-01
6.58984601e-01 -7.49799550e-01 1.05626214e+00 8.67085099e-01
2.50223547e-01 -7.42426455e-01 -1.63443029e+00 -5.91353416e-01
5.77427745e-01 -7.17278183e-01 5.67594826e-01 7.13640869e-01
6.06240451e-01 4.72239941e-01 -2.25514516e-01 1.24392755e-01
1.12819922e+00 2.52473861e-01 8.80664885e-02 -1.34362853e+00
-7.63260424e-01 -3.54700804e-01 -9.13832784e-01 -2.17374682e-01
1.00599922e-01 -1.23143697e+00 -7.21286237e-01 -8.30725014e-01
3.12248528e-01 -3.56715977e-01 -5.34868896e-01 1.56704545e-01
5.81947975e-02 3.65140408e-01 3.01682293e-01 2.48863429e-01
-6.37568116e-01 4.39176768e-01 7.64583051e-01 -1.48000032e-01
6.21958017e-01 1.30966067e-01 -4.91357297e-01 7.41029263e-01
8.03477883e-01 -8.96176398e-01 3.67016464e-01 -3.72237444e-01
1.26584721e+00 1.59413680e-01 5.44968963e-01 -1.02081907e+00
1.20515212e-01 -1.21223191e-02 2.60068268e-01 -1.56251401e-01
1.67896166e-01 -1.42335191e-01 -1.64370701e-01 8.34658206e-01
-3.74002725e-01 2.78870851e-01 -4.64640737e-01 3.85524511e-01
-5.19167595e-02 -8.34630728e-01 6.70791745e-01 -3.62721950e-01
-9.23188850e-02 7.89059997e-02 -2.51174092e-01 -9.05962363e-02
7.16049612e-01 4.56926852e-01 -7.81774938e-01 -1.43033668e-01
-7.68667042e-01 -1.76476449e-01 3.44807625e-01 -2.74968389e-02
6.91090822e-02 -1.02527249e+00 -8.87775362e-01 -2.05512673e-01
-1.59085855e-01 -5.37517726e-01 9.35460702e-02 8.14935207e-01
-1.29631925e+00 9.95918155e-01 -2.72236198e-01 1.79225076e-02
-6.26683295e-01 4.11298156e-01 6.24716043e-01 -8.66227627e-01
-5.11459410e-02 8.32465112e-01 -4.24638242e-01 -4.23076868e-01
-5.46841562e-01 -5.46653330e-01 -7.94488192e-02 3.55201513e-02
4.95377153e-01 6.40958309e-01 6.05102420e-01 -5.68305135e-01
-2.62690485e-01 3.02854568e-01 4.80660982e-03 -4.42053676e-01
1.70305610e+00 2.74953246e-01 -4.28579837e-01 4.49129552e-01
1.02440536e+00 -3.64792407e-01 -6.81292713e-01 -4.32192534e-02
5.76282024e-01 -1.51103050e-01 4.96097803e-01 -7.80786753e-01
-5.87338030e-01 1.38164485e+00 7.84908175e-01 5.28872669e-01
4.88277942e-01 -3.09123963e-01 9.64060307e-01 1.11926198e+00
6.74882948e-01 -1.37868500e+00 -2.42286205e-01 4.88648444e-01
2.44788706e-01 -1.40893555e+00 4.86480623e-01 -6.54874020e-04
-4.42772865e-01 1.66042244e+00 -1.73550397e-01 -4.11096781e-01
6.66159570e-01 -1.52872711e-01 -1.08022526e-01 -5.11296034e-01
-1.22072089e+00 1.54459938e-01 2.52854321e-02 6.35522883e-03
4.12587404e-01 4.97893780e-01 -2.06267536e-01 6.81950867e-01
-7.83322573e-01 -5.44959158e-02 1.20913851e+00 7.42421091e-01
-3.28806877e-01 -1.40711617e+00 -4.23044741e-01 5.68678856e-01
-9.92513418e-01 -6.02354527e-01 -1.56138897e-01 1.18529238e-01
-1.31541118e-02 8.34959924e-01 -2.46265933e-01 -3.97378474e-01
-2.12764218e-02 3.53772163e-01 5.15748143e-01 -5.15603244e-01
-9.98101294e-01 -6.00069702e-01 3.76162857e-01 -6.77782893e-01
-4.91896123e-01 -9.83381689e-01 -1.28085160e+00 -7.18298197e-01
-5.23572683e-01 2.26275086e-01 1.06570148e+00 1.13438523e+00
2.77463615e-01 1.63704261e-01 5.46210289e-01 -7.95281768e-01
-1.76106882e+00 -9.34411049e-01 -9.00742710e-01 2.57956177e-01
1.08088084e-01 -4.88695264e-01 -5.57314813e-01 -2.60531098e-01] | [5.600744247436523, 5.033929347991943] |
b35a7c31-ba6e-4be6-a7e8-e72ca4df4a12 | a-mixer-layer-is-worth-one-graph-convolution | 2304.03532 | null | https://arxiv.org/abs/2304.03532v1 | https://arxiv.org/pdf/2304.03532v1.pdf | A Mixer Layer is Worth One Graph Convolution: Unifying MLP-Mixers and GCNs for Human Motion Prediction | The past few years has witnessed the dominance of Graph Convolutional Networks (GCNs) over human motion prediction, while their performance is still far from satisfactory. Recently, MLP-Mixers show competitive results on top of being more efficient and simple. To extract features, GCNs typically follow an aggregate-and-update paradigm, while Mixers rely on token mixing and channel mixing operations. The two research paths have been independently established in the community. In this paper, we develop a novel perspective by unifying Mixers and GCNs. We show that a mixer layer can be seen as a graph convolutional layer applied to a fully-connected graph with parameterized adjacency. Extending this theoretical finding to the practical side, we propose Meta-Mixing Network (M$^2$-Net). Assisted with a novel zero aggregation operation, our network is capable of capturing both the structure-agnostic and the structure-sensitive dependencies in a collaborative manner. Not only is it computationally efficient, but most importantly, it also achieves state-of-the-art performance. An extensive evaluation on the Human3.6M, AMASS, and 3DPW datasets shows that M$^2$-Net consistently outperforms all other approaches. We hope our work brings the community one step further towards truly predictable human motion. Our code will be publicly available. | ['Mengyuan Liu', 'Chen Chen', 'Shen Zhao', 'Xinshun Wang'] | 2023-04-07 | null | null | null | null | ['motion-prediction', 'human-motion-prediction'] | ['computer-vision', 'time-series'] | [-4.78270091e-02 1.36222020e-01 -4.32500780e-01 -7.26915747e-02
-3.97527218e-01 -3.28042388e-01 5.78772008e-01 7.33043477e-02
-4.27409530e-01 3.64605606e-01 4.20196921e-01 -3.55296373e-01
2.51274067e-03 -1.02682817e+00 -7.45606780e-01 -5.42672813e-01
-5.39841473e-01 3.32867384e-01 5.97341299e-01 -4.10432369e-01
-1.57951608e-01 3.72329831e-01 -1.17535007e+00 3.60031128e-01
4.40052778e-01 8.26676548e-01 -3.41055878e-02 9.56843078e-01
-3.02173253e-02 1.22581673e+00 -1.13544486e-01 -6.36018813e-01
1.38338894e-01 -3.66008341e-01 -9.95401919e-01 -2.75464714e-01
4.20703173e-01 -2.27387890e-01 -8.67652178e-01 6.31678402e-01
5.93584299e-01 1.69134632e-01 2.63183236e-01 -1.34915459e+00
-3.92513931e-01 9.24121439e-01 -4.14911270e-01 1.75675452e-01
2.50290185e-01 3.64079922e-01 1.30475402e+00 -4.34782922e-01
7.18146741e-01 1.27671707e+00 9.50771511e-01 5.03693163e-01
-9.69814062e-01 -5.66555202e-01 4.38504577e-01 3.18444192e-01
-1.20919192e+00 -2.47644752e-01 8.03376019e-01 -3.32752824e-01
1.31232870e+00 -8.47396106e-02 9.04304028e-01 1.21240914e+00
2.31489733e-01 1.04373026e+00 4.53347504e-01 -3.30370851e-02
1.24723531e-01 -5.89272082e-01 1.60725147e-01 1.25316429e+00
3.61939102e-01 6.78280890e-02 -6.87342405e-01 2.68661138e-02
6.19220316e-01 -6.66375309e-02 -1.24818787e-01 -3.56860876e-01
-1.37712240e+00 6.99856341e-01 7.26285398e-01 2.24633172e-01
-2.05097333e-01 9.31573689e-01 4.35826927e-01 1.93625852e-01
3.04894179e-01 -9.98217911e-02 -1.57437176e-01 -3.61992151e-01
-9.55277979e-01 3.61017913e-01 9.07653928e-01 8.81203473e-01
6.49078488e-01 3.15958308e-03 -8.58015716e-02 3.06696475e-01
4.31565911e-01 3.33659321e-01 2.90085971e-01 -8.34274054e-01
6.32645369e-01 6.56898677e-01 -2.47517228e-01 -1.38525760e+00
-8.00625145e-01 -5.54548204e-01 -1.13673532e+00 2.40166504e-02
4.59666729e-01 -1.67521983e-01 -8.67714167e-01 1.87033260e+00
2.59295225e-01 6.37237906e-01 -2.99442001e-02 7.41159081e-01
1.04546320e+00 5.42974055e-01 1.20345362e-01 3.33530158e-01
1.17663169e+00 -1.41383302e+00 -3.66278052e-01 -3.79904866e-01
8.69004846e-01 -3.92999262e-01 6.72523499e-01 1.97168589e-01
-1.11159122e+00 -6.90699160e-01 -1.14514947e+00 -2.38087267e-01
-2.52662778e-01 -6.11538775e-02 1.00802159e+00 5.92322767e-01
-1.41126359e+00 8.89429510e-01 -1.26680684e+00 -4.79164630e-01
5.81691384e-01 4.55259681e-01 -5.23856401e-01 1.27970036e-02
-1.06246865e+00 5.78483045e-01 2.03361526e-01 4.15919214e-01
-7.67368078e-01 -3.02868754e-01 -9.58360672e-01 1.66709796e-02
4.12094444e-01 -1.23191929e+00 9.93533909e-01 -4.96842563e-01
-1.59896195e+00 4.84752625e-01 -6.66443035e-02 -9.70076978e-01
7.00283825e-01 -2.92151332e-01 -3.50739151e-01 2.72878945e-01
-1.31236076e-01 7.30669558e-01 6.15748286e-01 -8.69720459e-01
-8.20971489e-01 -1.28507465e-01 2.49007329e-01 9.68935192e-02
-1.60830513e-01 -2.86630958e-01 -6.94805682e-01 -6.07587576e-01
3.84307727e-02 -1.16873145e+00 -5.56185126e-01 -1.35058880e-01
-4.79968369e-01 -1.95487700e-02 6.33968711e-01 -3.50966662e-01
1.59061325e+00 -1.93328679e+00 2.60415107e-01 1.85671851e-01
6.44396424e-01 3.74093562e-01 -2.47064665e-01 7.86538959e-01
2.10035384e-01 1.09219991e-01 -3.07099015e-01 -6.57710195e-01
4.52680662e-02 2.80706227e-01 -1.82231236e-02 5.24441779e-01
1.89700857e-01 1.39099979e+00 -1.09052134e+00 -4.77163136e-01
1.60368621e-01 5.95219254e-01 -8.93862128e-01 -2.06414446e-01
-1.99792087e-01 3.46227646e-01 -3.99233967e-01 6.36475265e-01
3.99552852e-01 -5.87450981e-01 3.90570849e-01 -1.36770129e-01
7.92469457e-02 2.07663134e-01 -1.13480818e+00 1.80720305e+00
-2.22670779e-01 7.71991849e-01 -2.45546669e-01 -1.00819945e+00
7.13690162e-01 9.47402716e-02 6.68643653e-01 -5.26181400e-01
9.53626260e-02 1.77062467e-01 7.89840445e-02 -4.14814472e-01
7.80610442e-01 2.52584070e-01 -8.30401182e-02 1.48211494e-01
6.68180212e-02 3.03502470e-01 2.11789817e-01 4.17367399e-01
1.47240233e+00 2.93475896e-01 2.15965807e-01 -9.60172713e-02
3.84663999e-01 -7.10743666e-02 4.50569451e-01 9.07692730e-01
-3.64859819e-01 5.48389554e-01 5.81676781e-01 -5.99900901e-01
-6.28644228e-01 -1.01477933e+00 5.38601220e-01 8.05912614e-01
2.23603860e-01 -8.13809931e-01 -5.51660597e-01 -7.43737578e-01
1.73466466e-02 6.52001053e-02 -5.76668084e-01 -5.41061983e-02
-1.09962881e+00 -6.72273278e-01 9.02908504e-01 8.84069264e-01
8.15569043e-01 -9.36405003e-01 -6.46719694e-01 5.59735596e-01
-3.62048507e-01 -1.41633284e+00 -4.38038051e-01 -1.18814312e-01
-1.01543105e+00 -1.11732161e+00 -5.97037733e-01 -7.84353673e-01
3.13463837e-01 3.70789438e-01 1.22600818e+00 3.70185673e-01
-1.14338011e-01 4.47707653e-01 -5.04957259e-01 -5.27789369e-02
-1.87109455e-01 5.70427358e-01 -2.41409674e-01 1.54134855e-01
1.08001865e-01 -9.40537512e-01 -9.72141743e-01 9.57249701e-02
-6.78924739e-01 1.46276623e-01 5.91386378e-01 5.89037061e-01
3.14890534e-01 -1.08292505e-01 2.35881075e-01 -8.31159353e-01
3.52885902e-01 -5.41781843e-01 -2.25892678e-01 -1.28695458e-01
-5.24978876e-01 1.66177362e-01 5.79916596e-01 -2.25879759e-01
-7.17572272e-01 3.38153243e-01 -3.74609917e-01 -2.23217696e-01
-6.03496330e-03 7.58283377e-01 5.70468754e-02 -1.44957021e-01
5.25527537e-01 5.93743660e-02 -1.57848373e-01 -2.38202721e-01
6.53615892e-01 1.27491415e-01 7.22366810e-01 -4.45135355e-01
8.41274679e-01 9.85880077e-01 3.67451727e-01 -8.89207602e-01
-4.08266902e-01 -5.93789697e-01 -6.35870457e-01 -3.45235914e-01
9.84942198e-01 -9.56941605e-01 -9.86295342e-01 7.14653790e-01
-1.05294132e+00 -8.05820584e-01 -1.60863698e-01 5.05594671e-01
-5.47293305e-01 7.07378626e-01 -8.34281623e-01 -6.31750822e-01
-2.88805902e-01 -1.07107437e+00 1.05946398e+00 1.86296538e-01
-1.35054946e-01 -1.21265447e+00 1.64575249e-01 2.90259063e-01
4.15672809e-01 4.85713094e-01 5.82308769e-01 -5.38055062e-01
-9.61644590e-01 -1.64357215e-01 -6.65406808e-02 -1.12983985e-02
-1.00912318e-01 8.42694789e-02 -7.73950458e-01 -2.45865569e-01
-6.84290230e-01 -1.58693641e-02 1.30984652e+00 5.71296632e-01
6.96843326e-01 -1.28287137e-01 -5.29528379e-01 8.35342526e-01
1.18487048e+00 -1.54348612e-01 8.51179123e-01 3.88172448e-01
1.11017132e+00 2.78382510e-01 9.62325335e-02 2.86377788e-01
1.00528860e+00 6.11477435e-01 5.64025342e-01 -1.20961986e-01
-5.00828624e-01 -4.39755052e-01 4.59454268e-01 9.70394135e-01
-5.37798762e-01 -5.96257567e-01 -9.79459882e-01 5.75758934e-01
-2.19766974e+00 -1.19686592e+00 -3.71255785e-01 1.77631950e+00
2.65985221e-01 4.23397660e-01 4.28537846e-01 1.16986744e-01
4.52515960e-01 7.04687119e-01 -1.89540744e-01 -1.34692520e-01
-1.04692690e-01 2.54642040e-01 5.57142556e-01 6.38369083e-01
-1.16664720e+00 1.19434214e+00 5.83161831e+00 5.44952095e-01
-1.02592206e+00 -1.59772605e-01 4.28535193e-01 -3.23760845e-02
-1.00760341e-01 1.81055591e-01 -8.33358228e-01 3.02319258e-01
7.72792459e-01 -2.97107380e-02 3.24455291e-01 6.76023901e-01
7.98585489e-02 2.39593647e-02 -1.05578840e+00 1.07402241e+00
-1.02029718e-01 -1.61659682e+00 1.12615250e-01 9.29712206e-02
4.33460921e-01 4.54242855e-01 -3.05880588e-02 2.53024817e-01
5.57128787e-01 -1.00697005e+00 8.25246394e-01 4.89022374e-01
2.80294240e-01 -6.03405714e-01 6.51560605e-01 2.49075010e-01
-2.03977633e+00 -2.68822815e-02 -8.18726420e-02 -3.08501303e-01
4.30165112e-01 4.36971992e-01 -4.76971775e-01 9.43214118e-01
7.09023118e-01 9.54168916e-01 -5.49674869e-01 1.04293644e+00
-1.92018211e-01 6.43659234e-01 -2.21493974e-01 -8.46079066e-02
5.08288145e-01 -2.65179132e-03 4.84687477e-01 1.62656903e+00
1.81206226e-01 5.00808731e-02 3.35052997e-01 3.87421161e-01
-1.87415287e-01 -9.60484445e-02 -5.38783848e-01 -1.79525778e-01
1.72466546e-01 1.13749611e+00 -1.01357651e+00 -4.04050946e-01
-4.46314901e-01 1.10857177e+00 4.56687480e-01 3.53838652e-01
-9.68465865e-01 -2.30221674e-01 7.11357415e-01 -3.72808911e-02
5.75507641e-01 -6.99901104e-01 -6.08615857e-03 -1.22558045e+00
2.53931973e-02 -7.03708231e-01 6.05887771e-01 -2.65216857e-01
-1.10414958e+00 6.97180033e-01 -3.41474228e-02 -1.16034317e+00
-1.94461852e-01 -6.61540151e-01 -6.00352287e-01 3.68569672e-01
-1.58589995e+00 -1.31894743e+00 -4.82825488e-01 7.03412592e-01
2.98633158e-01 1.48997664e-01 5.33366561e-01 4.81223673e-01
-5.63251376e-01 6.87085271e-01 -3.28390360e-01 5.04935026e-01
3.45682949e-01 -1.05165875e+00 1.12634659e+00 9.67109263e-01
4.00260657e-01 4.62273628e-01 4.04597819e-01 -7.72198915e-01
-1.51886547e+00 -1.22289062e+00 9.84222889e-01 -4.28977400e-01
8.35189104e-01 -2.65443474e-01 -6.66136265e-01 8.28157902e-01
1.50302544e-01 3.74708772e-01 3.37413013e-01 -5.07356878e-03
-4.31457639e-01 5.73170371e-03 -5.06295741e-01 8.35725188e-01
1.77207077e+00 -3.70522708e-01 -2.04336271e-01 1.14947364e-01
7.99096107e-01 -6.33094311e-01 -7.14614928e-01 4.35657948e-01
7.63936698e-01 -1.17840052e+00 1.04741144e+00 -7.06868052e-01
3.47475886e-01 -2.89079696e-01 -1.41330376e-01 -9.74781752e-01
-4.82747167e-01 -9.81795490e-01 -4.88814920e-01 7.83578992e-01
4.24703121e-01 -7.05172241e-01 1.15650475e+00 1.02603823e-01
-2.73302823e-01 -8.92293930e-01 -8.95338833e-01 -9.69498336e-01
2.49017756e-02 -7.87877202e-01 4.61834103e-01 6.47649229e-01
3.81043851e-02 2.47650594e-01 -5.93784153e-01 2.90870667e-01
5.56905925e-01 -4.96508516e-02 1.13439286e+00 -1.05074799e+00
-6.54587805e-01 -7.17997491e-01 -7.45341241e-01 -1.55159843e+00
4.83396091e-02 -9.54050481e-01 -7.02629983e-02 -1.84121430e+00
-4.98653278e-02 -1.94851875e-01 -1.23602934e-01 4.98976648e-01
-1.18971646e-01 2.90715337e-01 4.40981090e-01 1.36678204e-01
-7.68319011e-01 3.66285056e-01 1.15733790e+00 -2.18315288e-01
-3.21628302e-01 9.23511013e-02 -4.69256997e-01 8.30691755e-01
8.08694005e-01 -2.96618223e-01 -4.45095479e-01 -4.57791418e-01
4.70403135e-01 -9.51943733e-03 6.46898150e-01 -1.40350926e+00
5.69041491e-01 1.42003506e-01 2.06504390e-03 -5.63993812e-01
2.43542656e-01 -5.22353590e-01 2.14385733e-01 7.00191975e-01
-1.16362028e-01 2.44214773e-01 1.35385126e-01 9.29160714e-01
-1.46142960e-01 3.49309146e-01 4.75798190e-01 -2.08512053e-01
-1.00254929e+00 7.02987134e-01 -2.09469661e-01 6.51736930e-02
6.94962680e-01 -2.51828969e-01 -4.99827981e-01 -6.20269358e-01
-6.51972890e-01 1.85183942e-01 2.07463667e-01 3.96516085e-01
4.74973679e-01 -1.24886000e+00 -6.36985779e-01 -7.16386363e-02
-6.40175259e-03 4.22610529e-03 3.84445280e-01 1.04632115e+00
-7.35707223e-01 2.70305127e-01 7.19355345e-02 -6.79484367e-01
-1.00697470e+00 3.78335357e-01 3.77007902e-01 -6.03542805e-01
-1.01014423e+00 8.88090253e-01 -2.64205225e-02 -8.08586255e-02
3.75916034e-01 -7.23720551e-01 8.81387480e-03 -2.43715048e-02
2.58003712e-01 6.08560920e-01 -1.20177813e-01 -7.30290055e-01
-5.25834382e-01 5.75982749e-01 1.71546593e-01 -7.48341605e-02
1.22224116e+00 -4.72845994e-02 1.54025361e-01 2.56364524e-01
1.17520010e+00 9.43660215e-02 -1.30846560e+00 -1.25479490e-01
-2.23362818e-02 -5.26681617e-02 -3.76216620e-01 -2.91472495e-01
-1.37644148e+00 8.88426304e-01 2.47088686e-01 3.01177561e-01
9.79178846e-01 5.28596975e-02 1.12120497e+00 4.78025436e-01
4.28093553e-01 -9.56936896e-01 1.94207892e-01 6.63144469e-01
3.95819038e-01 -1.06633925e+00 -6.06278330e-02 -4.88157153e-01
-4.08650815e-01 1.15381110e+00 4.27446276e-01 -3.25421810e-01
7.68756270e-01 3.00150514e-01 -1.12210527e-01 -3.78391176e-01
-7.00033009e-01 -5.63303411e-01 3.49884719e-01 5.56901634e-01
4.58352566e-01 5.34573533e-02 -1.69333741e-01 4.97381479e-01
-2.70960659e-01 1.76877245e-01 4.03298497e-01 9.91226733e-01
-3.45936924e-01 -1.15482378e+00 1.63871482e-01 1.16768382e-01
-2.24248603e-01 -2.06210971e-01 -2.99130470e-01 1.05872691e+00
4.58300449e-02 9.65278327e-01 -1.25985235e-01 -8.98673534e-01
4.52099711e-01 -2.04179808e-01 3.40889513e-01 -2.53480136e-01
-6.69249177e-01 -1.40422046e-01 3.91894132e-01 -1.08691239e+00
-6.98464453e-01 -3.75186503e-01 -1.48241782e+00 -6.95617020e-01
5.73155796e-03 -1.47953510e-01 2.97373891e-01 7.86423743e-01
6.06098890e-01 6.51912808e-01 1.73373654e-01 -9.30627406e-01
-9.85434279e-02 -5.49198985e-01 -2.92058706e-01 1.92788050e-01
3.58494997e-01 -4.98425305e-01 -2.18631819e-01 1.63696371e-02] | [7.302338123321533, -0.06006191670894623] |
4ceeab14-c58d-4e4c-b44d-7fc29801a475 | analysis-open-published-17-june-2019 | null | null | https://www.nature.com/articles/s41597-019-0103-9 | https://www.nature.com/articles/s41597-019-0103-9.pdf | Analysis | OPEN | Published: 17 June 2019 Multitask learning and benchmarking with clinical time series data | Health care is one of the most exciting frontiers in data mining and machine learning. Successful adoption of electronic health records (EHRs) created an explosion in digital clinical data available for analysis, but progress in machine learning for healthcare research has been difficult to measure because of the absence of publicly available benchmark data sets. To address this problem, we propose four clinical prediction benchmarks using data derived from the publicly available Medical Information Mart for Intensive Care (MIMIC-III) database. These tasks cover a range of clinical problems including modeling risk of mortality, forecasting length of stay, detecting physiologic decline, and phenotype classification. We propose strong linear and neural baselines for all four tasks and evaluate the effect of deep supervision, multitask training and data-specific architectural modifications on the performance of neural models. | ['Hrant Khachatrian', 'Aram Galstyan', 'Hrayr Harutyunyan', 'Greg Ver Steeg', 'David C. Kale'] | 2019-06-17 | null | null | null | nature-scientific-data-2019-6 | ['computational-phenotyping', 'length-of-stay-prediction', 'phenotype-classification'] | ['medical', 'medical', 'medical'] | [ 3.13054800e-01 -5.78374118e-02 -3.89048904e-01 -6.68176532e-01
-1.02479899e+00 -3.02353084e-01 2.27289483e-01 9.57562804e-01
-5.22778749e-01 6.20725811e-01 4.46601182e-01 -7.14599609e-01
-2.35350326e-01 -4.18598890e-01 -4.94892567e-01 -3.44416469e-01
-3.59107137e-01 6.25527263e-01 -2.94809610e-01 3.08194548e-01
-2.29910582e-01 4.05876696e-01 -7.61415839e-01 7.73154438e-01
5.24145067e-01 1.12720847e+00 -2.22744748e-01 5.98873854e-01
3.11114401e-01 8.93073320e-01 -2.60798216e-01 -4.51079383e-02
8.32425728e-02 -4.30714399e-01 -6.15524769e-01 -3.07932287e-01
7.54459873e-02 -3.27766418e-01 -2.72190213e-01 4.45750237e-01
8.59008014e-01 -3.46386939e-01 4.32091683e-01 -8.30687106e-01
-4.55163598e-01 5.25159240e-01 -1.33931518e-01 4.51742619e-01
-3.56645994e-02 2.05329120e-01 7.21158385e-01 -3.16157967e-01
5.56038558e-01 7.86584377e-01 9.62286115e-01 5.96800327e-01
-1.31158984e+00 -2.98393309e-01 -2.64953911e-01 1.75978970e-02
-1.00677490e+00 -4.24728781e-01 1.41204387e-01 -7.00626373e-01
9.39212501e-01 1.25390872e-01 6.06951654e-01 1.33350050e+00
5.87305367e-01 4.43526059e-01 8.82851362e-01 -1.14216097e-01
5.85189387e-02 -7.26443306e-02 5.02523124e-01 6.51632726e-01
2.36439064e-01 9.68589187e-02 -2.75288701e-01 -6.45226955e-01
4.99423772e-01 5.58565259e-01 -3.67840856e-01 -1.25197023e-01
-1.40470910e+00 5.72955430e-01 1.56556770e-01 7.03955814e-02
-5.30759633e-01 -4.89724949e-02 8.37019801e-01 4.91965860e-01
4.08102721e-01 5.38136303e-01 -1.06193888e+00 -2.57176846e-01
-7.12049425e-01 2.07274616e-01 7.72078574e-01 3.12761307e-01
-3.61398011e-02 -1.92138180e-01 -3.27818066e-01 8.29804063e-01
1.42633736e-01 2.52002656e-01 7.61129022e-01 -7.78801203e-01
4.29161370e-01 8.14820588e-01 -2.53569148e-02 -5.39787173e-01
-1.01053679e+00 -5.49443662e-01 -9.66929257e-01 -5.17354235e-02
5.66207290e-01 -4.66660321e-01 -7.55369246e-01 1.38863111e+00
5.23088723e-02 6.89147264e-02 1.43937126e-01 6.34348989e-01
6.96939945e-01 2.93931037e-01 3.69867504e-01 -4.92216349e-02
1.31945682e+00 -5.53335130e-01 -4.59540069e-01 -3.33627425e-02
1.26486373e+00 -2.30959430e-01 8.32683384e-01 4.48029637e-01
-9.70871627e-01 -1.72812387e-01 -7.51070440e-01 -9.16727111e-02
-3.98762107e-01 1.01337232e-01 5.12173235e-01 3.20288181e-01
-7.18836427e-01 6.90077662e-01 -1.21227074e+00 -3.02343667e-01
8.25494051e-01 3.55335891e-01 -4.48559940e-01 -2.21371517e-01
-1.05745745e+00 9.46310341e-01 4.75678220e-02 -1.25353992e-01
-8.05890858e-01 -1.22243226e+00 -6.21135235e-01 2.39926189e-01
-4.62389998e-02 -1.10768950e+00 1.20485127e+00 -6.37558460e-01
-7.85652041e-01 9.84511375e-01 1.86124325e-01 -8.33189726e-01
4.81410801e-01 -3.75915468e-01 -4.10319149e-01 -1.62529886e-01
-2.43955284e-01 3.93021792e-01 2.18778238e-01 -5.36878109e-01
-4.88883823e-01 -7.62818754e-01 -6.85670614e-01 -1.53085485e-01
-3.33337098e-01 2.00893000e-01 1.31768286e-01 -3.79618466e-01
-3.92693132e-01 -7.78950572e-01 -6.06634855e-01 -1.83665425e-01
-2.28169441e-01 7.01486245e-02 4.64202404e-01 -9.16186690e-01
1.19575024e+00 -2.16190147e+00 -1.89402059e-01 -8.36935192e-02
5.97485006e-01 4.05526936e-01 5.16179316e-02 3.39087546e-01
-1.66084364e-01 1.39518037e-01 -1.01058178e-01 -2.29258493e-01
-5.36230206e-01 7.61802197e-02 -2.30385233e-02 3.91829193e-01
2.95391291e-01 1.08553529e+00 -6.74895287e-01 -2.58287936e-01
7.94422403e-02 6.51975572e-01 -6.37539089e-01 3.76079828e-01
-1.42217204e-01 6.63874090e-01 -5.67619681e-01 5.22699058e-01
-1.33731395e-01 -8.98212612e-01 3.68140876e-01 -4.94766645e-02
1.26096785e-01 6.38919592e-01 -2.46971279e-01 1.52440262e+00
-2.31846482e-01 2.96516269e-01 -1.01713680e-01 -1.07548976e+00
5.56871653e-01 6.39871538e-01 1.12795806e+00 -6.01762891e-01
2.64108121e-01 -9.00294483e-02 3.98074597e-01 -8.02140474e-01
-3.53683382e-01 -3.37755769e-01 -9.17253178e-03 3.29717308e-01
-2.51859009e-01 5.65850496e-01 -2.21398562e-01 -2.06186503e-01
1.81360328e+00 -2.67726302e-01 3.24342459e-01 -2.21180275e-01
1.04741175e-02 3.32716793e-01 8.75384510e-01 6.17557108e-01
-3.66426647e-01 5.75272799e-01 4.02197659e-01 -9.86066878e-01
-9.00066197e-01 -9.79963541e-01 -5.12732148e-01 9.36611474e-01
-5.29224396e-01 -3.71999085e-01 -4.87202048e-01 -4.96611476e-01
4.26445544e-01 3.25310409e-01 -7.72738218e-01 -3.93561065e-01
-4.96629447e-01 -1.18747079e+00 7.12306321e-01 6.34650052e-01
-5.11340350e-02 -1.10363448e+00 -1.12242985e+00 4.10351068e-01
9.46263820e-02 -1.06077075e+00 -2.43547574e-01 4.90886718e-01
-1.22428417e+00 -1.47464943e+00 -4.94668365e-01 -6.34068310e-01
2.09806696e-01 -2.70059675e-01 1.35035753e+00 7.11028129e-02
-8.21483076e-01 3.18648219e-01 -2.86711622e-02 -7.01948106e-01
-6.08373284e-01 3.32713038e-01 -1.32570788e-01 -1.86615989e-01
6.09905899e-01 -2.58615315e-01 -8.11012387e-01 3.54775190e-02
-7.14785933e-01 1.22772470e-01 6.72949493e-01 9.23482120e-01
6.52090967e-01 -6.11195683e-01 1.11299574e+00 -1.26111305e+00
7.74498224e-01 -8.60510170e-01 -3.04672539e-01 1.97423369e-01
-1.05703318e+00 -3.82988565e-02 6.55280530e-01 -1.00307561e-01
-6.98910356e-01 -5.34785762e-02 -1.65052861e-01 -1.68134242e-01
-4.14775968e-01 9.90955234e-01 3.45823556e-01 4.04430926e-01
6.68265760e-01 -1.73746094e-01 3.38814795e-01 -6.85355365e-01
-2.78378248e-01 7.14305699e-01 3.30922663e-01 -2.62411505e-01
-6.59771413e-02 2.14444354e-01 1.59348100e-01 -5.77144861e-01
-1.04169357e+00 -4.42738533e-01 -4.01024252e-01 2.01120540e-01
1.20892215e+00 -1.00267315e+00 -6.91088796e-01 2.45537117e-01
-8.19540203e-01 -6.14147663e-01 -3.16749036e-01 6.12893879e-01
-3.46581221e-01 5.94447600e-03 -8.84771466e-01 -3.11754942e-01
-6.64219141e-01 -1.13300800e+00 7.99869239e-01 -3.19827467e-01
-4.88034189e-01 -1.23515177e+00 3.69808853e-01 3.20969075e-01
6.39593840e-01 7.47861326e-01 1.51549244e+00 -1.01536763e+00
-1.58447579e-01 -4.86014873e-01 -1.63383991e-01 4.21686113e-01
4.04504895e-01 -2.04425201e-01 -5.83151221e-01 -2.65848100e-01
-2.13711970e-02 -5.34981668e-01 1.01593661e+00 5.83822966e-01
1.43456471e+00 -1.22687981e-01 -6.15870655e-01 9.17191148e-01
1.11560011e+00 3.26311797e-01 4.18746173e-01 2.15296552e-01
6.19192243e-01 3.33346277e-01 2.10241023e-02 5.16842663e-01
6.49714172e-01 3.52836996e-01 2.57282585e-01 -4.55364555e-01
1.88613862e-01 2.02604085e-01 -1.58466715e-02 6.79892302e-01
5.13844937e-02 3.47767845e-02 -1.48912311e+00 4.27945286e-01
-1.75711870e+00 -5.79347730e-01 -1.52672440e-01 2.21229267e+00
9.98446822e-01 -6.31785542e-02 1.91653624e-01 -9.32515711e-02
2.97584087e-01 -3.29012096e-01 -7.88265228e-01 -2.42495805e-01
2.77035218e-02 5.73392846e-02 4.64390516e-01 1.95657145e-02
-1.08596194e+00 2.61412770e-01 6.99679422e+00 -2.36017123e-01
-1.26429439e+00 1.34830862e-01 1.42249393e+00 -4.15733486e-01
1.97925061e-01 -5.08241475e-01 -4.85474885e-01 4.79555845e-01
1.43038142e+00 1.88794006e-02 1.62955746e-01 7.25943327e-01
5.46558499e-01 2.47833878e-01 -1.59589541e+00 8.95123720e-01
-2.73453951e-01 -1.44078732e+00 -1.52067348e-01 4.39424179e-02
5.59321523e-01 7.05497682e-01 9.07390788e-02 1.26609147e-01
3.25415581e-01 -1.26001227e+00 2.86233220e-02 6.49298429e-01
1.02808642e+00 -3.09196860e-01 8.36478531e-01 2.04117551e-01
-3.40372860e-01 -3.31687093e-01 4.39668223e-02 -5.72944432e-02
8.17490220e-02 5.32656312e-01 -9.60486650e-01 1.14127221e-02
6.87282026e-01 7.84959912e-01 -4.75213110e-01 1.08403826e+00
6.25799954e-01 1.02102339e+00 -7.74184391e-02 3.99292678e-01
1.27003053e-02 1.65813342e-02 8.89194012e-02 1.26369214e+00
1.44960701e-01 2.65823036e-01 2.66877711e-01 6.29333019e-01
-3.73147964e-01 9.00768042e-02 -4.95465845e-01 -1.84508234e-01
9.72859785e-02 1.15224171e+00 -2.93296009e-01 -2.56221563e-01
-6.62564278e-01 3.91543627e-01 3.20178509e-01 1.73213944e-01
-5.94793677e-01 1.30692810e-01 8.83738339e-01 6.22206092e-01
-3.70889276e-01 -4.63702567e-02 -5.67229569e-01 -1.11803424e+00
-3.68280351e-01 -1.09335721e+00 9.13197517e-01 -1.38281494e-01
-1.42555070e+00 3.43674541e-01 -2.63893455e-01 -7.00996816e-01
-3.99059445e-01 -6.52987242e-01 -4.10846770e-01 8.23016405e-01
-1.51185048e+00 -5.41524231e-01 -4.59541470e-01 4.08835143e-01
2.06778020e-01 -3.18880409e-01 1.18003607e+00 5.93668938e-01
-7.53549933e-01 5.33904374e-01 3.52576613e-01 3.17892015e-01
1.02051568e+00 -1.07292092e+00 5.20836532e-01 1.04081370e-01
-3.84022862e-01 4.85336304e-01 2.29780748e-01 -5.09050727e-01
-1.46092999e+00 -1.41277766e+00 7.79158354e-01 -7.17749834e-01
5.68083465e-01 -8.43450725e-02 -1.03419232e+00 8.64339352e-01
-1.88169703e-01 1.96837962e-01 1.18094397e+00 1.58155501e-01
-2.57021904e-01 -2.57614791e-01 -1.04510069e+00 1.63339674e-01
6.65135443e-01 -2.32594341e-01 -2.99678445e-01 3.62077951e-01
4.19891953e-01 -2.15749636e-01 -1.51223648e+00 7.22015917e-01
6.45327449e-01 -7.89143384e-01 9.31092501e-01 -1.27962708e+00
7.65902281e-01 3.12521964e-01 4.91756946e-02 -1.29125106e+00
-2.95912385e-01 -4.49774474e-01 -9.27879661e-02 5.99131763e-01
6.13169670e-01 -7.52358913e-01 7.83295274e-01 8.86952877e-01
-2.72166967e-01 -1.34240806e+00 -6.17934585e-01 -2.14996457e-01
3.15198004e-01 -3.52346227e-02 4.04492229e-01 1.22952366e+00
2.34744802e-01 5.19727886e-01 -2.54805446e-01 -2.43530393e-01
3.38412106e-01 4.24783677e-02 4.21154141e-01 -1.53860080e+00
-5.95559478e-01 -3.95428270e-01 -2.27823585e-01 -4.11789715e-01
-2.78149903e-01 -1.05840456e+00 -3.22000593e-01 -1.84168661e+00
3.95662040e-01 -6.77945197e-01 -7.86173165e-01 7.95999646e-01
-3.86927485e-01 -6.80175871e-02 -2.07644254e-01 2.48496324e-01
-5.55137932e-01 1.40195087e-01 7.90234685e-01 -1.33768832e-02
-3.22178006e-01 4.55087004e-03 -6.73525810e-01 5.81930637e-01
9.35663879e-01 -5.30072272e-01 -2.76524853e-02 -6.87205613e-01
1.20086327e-01 5.10075986e-01 2.78919786e-01 -1.02384841e+00
-1.20399691e-01 -3.49038616e-02 7.27687359e-01 -1.32133394e-01
5.26439995e-02 -7.67988682e-01 7.83019736e-02 8.73661816e-01
-8.14591885e-01 4.83815730e-01 2.01097578e-01 5.56102753e-01
-1.44481540e-01 4.69620228e-01 8.52405548e-01 -1.27601121e-02
-3.74186561e-02 4.39664304e-01 -4.72615957e-01 3.27844948e-01
8.65398228e-01 7.88098127e-02 -5.07465661e-01 -3.46944258e-02
-8.15000415e-01 2.91710615e-01 1.33885249e-01 4.82375026e-01
4.03094083e-01 -1.01578200e+00 -1.04045153e+00 1.95826069e-01
2.28678808e-01 -4.49021459e-02 1.36804536e-01 1.06969464e+00
-5.35469532e-01 6.05436444e-01 -2.53284276e-01 -5.18168390e-01
-1.18853390e+00 6.84303403e-01 5.01530230e-01 -5.23219109e-01
-8.56057227e-01 3.59675229e-01 1.53329983e-01 -2.67608106e-01
4.17110384e-01 -5.26031196e-01 -2.66488199e-03 -2.68242270e-01
4.72075820e-01 3.75660449e-01 2.86720067e-01 -8.70587304e-02
-3.59014779e-01 -1.70974046e-01 -2.72586763e-01 5.81161976e-01
1.67200780e+00 3.65602493e-01 1.54876877e-02 5.60379207e-01
1.26287675e+00 -7.34185100e-01 -9.24492300e-01 -1.26871616e-01
4.13027346e-01 1.26784891e-01 -3.31522222e-03 -1.14810884e+00
-8.89014184e-01 9.85162020e-01 8.44715476e-01 1.46543279e-01
1.04341805e+00 -1.54192165e-01 8.85242999e-01 5.90716004e-01
-1.04480892e-01 -6.53408349e-01 -1.85545132e-01 4.05622989e-01
4.04018521e-01 -1.41808808e+00 -2.36279041e-01 6.51430860e-02
-5.78356564e-01 8.47098947e-01 4.44388270e-01 -1.00030387e-02
9.99508321e-01 4.47831362e-01 3.94145012e-01 -2.93482780e-01
-1.27964485e+00 4.09350067e-01 9.44163054e-02 2.79412150e-01
9.60844994e-01 8.91032219e-02 -1.44805580e-01 9.34866369e-01
2.21263260e-01 4.02851582e-01 3.07621330e-01 8.67202640e-01
-2.51248062e-01 -1.05839646e+00 5.90323545e-02 1.25556803e+00
-1.19715261e+00 -3.14945817e-01 -3.20386052e-01 4.43919569e-01
-1.33552149e-01 6.46021366e-01 -6.05957061e-02 -1.00847065e-01
3.12058628e-01 5.07121563e-01 3.00426602e-01 -7.31044710e-01
-1.02062333e+00 -1.87607616e-01 1.96323544e-01 -4.76470083e-01
-7.74135860e-03 -6.14555001e-01 -1.23064971e+00 7.69078285e-02
2.12202355e-01 -3.22246626e-02 7.25954115e-01 8.92927647e-01
9.86983180e-01 9.01932657e-01 1.60409659e-01 -2.78705299e-01
-8.96783352e-01 -8.69031012e-01 -2.44718462e-01 5.68054676e-01
4.69638497e-01 -1.75776854e-01 1.19381048e-01 3.65967363e-01] | [7.963992118835449, 6.270141124725342] |
51b5d30c-1c75-48e3-9960-429a6ea964aa | temporal-convolutional-attention-neural | null | null | https://ieeexplore.ieee.org/abstract/document/9534351 | https://www.researchgate.net/profile/Yang-Lin-27/publication/354797495_Temporal_Convolutional_Attention_Neural_Networks_for_Time_Series_Forecasting/links/61558599ab3c1324134c8883/Temporal-Convolutional-Attention-Neural-Networks-for-Time-Series-Forecasting.pdf | Temporal Convolutional Attention Neural Networks for Time Series Forecasting | Temporal Convolutional Neural Networks (TCNNs) have been applied for various sequence modelling tasks including time series forecasting. However, TCNNs may require many convolutional layers if the input sequence is long and are not able to provide interpretable results. In this paper, we present TCAN, a novel deep learning approach that employs attention mechanism with temporal convolutions for probabilistic forecasting, and demonstrate its performance in a case study for solar power forecasting. TCAN uses the hierarchical convolutional structure of TCNN to extract temporal dependencies and then uses sparse attention to focus on the important timesteps. The sparse attention layer of TCAN enables an extended receptive field without requiring a deeper architecture and allows for interpretability of the forecasting results. An evaluation using three large solar power data sets demonstrates that TCAN outperforms several state-of-the-art deep learning forecasting models including TCNN in terms of accuracy. TCAN requires less number of convolutional layers than TCNN for an extended receptive field, is faster to train and is able to visualize the most important timesteps for the prediction. | ['Mashud Rana', 'Irena Koprinska', 'Yang Lin'] | 2021-09-23 | null | null | null | international-joint-conference-on-neural-1 | ['probabilistic-time-series-forecasting'] | ['time-series'] | [-1.55156046e-01 -2.74851352e-01 1.25804588e-01 -4.06286627e-01
9.99643579e-02 -5.92705786e-01 1.06901324e+00 -1.61965564e-01
4.00657840e-02 6.83536589e-01 4.20608193e-01 -7.21268773e-01
-2.83635497e-01 -9.41159368e-01 -6.58697605e-01 -9.09526587e-01
-2.84805864e-01 2.30632752e-01 2.18116686e-01 -1.36981800e-01
3.19612741e-01 6.63577080e-01 -1.63641691e+00 7.04696357e-01
6.52850747e-01 1.30050492e+00 2.49640882e-01 9.25024688e-01
-5.14782369e-01 1.37411869e+00 -7.64421701e-01 2.77456075e-01
-4.11798945e-03 -3.29234963e-03 -6.08241260e-01 -5.41155338e-01
1.93018571e-01 -1.25508383e-01 -3.31087261e-01 2.93490142e-01
3.50608826e-01 3.43430609e-01 8.01314592e-01 -1.26666260e+00
-8.66803646e-01 5.22343814e-01 -1.83263630e-01 7.73316383e-01
-1.70606300e-01 1.95957333e-01 5.27424932e-01 -4.46732432e-01
9.09393132e-02 1.15770006e+00 1.19715714e+00 7.73141310e-02
-9.38952148e-01 -6.39481306e-01 8.41699690e-02 2.99759716e-01
-8.19852173e-01 8.45558867e-02 4.92988229e-01 -7.29992688e-01
1.71837819e+00 1.04607329e-01 6.34009719e-01 1.09242642e+00
7.42919803e-01 3.72580796e-01 1.14415324e+00 -8.56079906e-02
4.06162977e-01 -5.16399741e-01 2.87128031e-01 2.77415574e-01
-4.13136274e-01 6.44008577e-01 -4.48061019e-01 9.12715048e-02
5.47860146e-01 3.44051719e-01 3.71989608e-02 6.05397642e-01
-1.18657076e+00 7.92474627e-01 7.57380784e-01 6.63010001e-01
-8.83869946e-01 7.07226217e-01 7.00093389e-01 2.19022557e-01
7.82234848e-01 1.70117661e-01 -7.29195654e-01 -2.61549473e-01
-1.18869305e+00 -6.15775809e-02 6.81355298e-01 4.49142963e-01
3.54192615e-01 6.75505102e-01 -5.83747745e-01 3.53565931e-01
-9.45488885e-02 5.07881284e-01 7.35092044e-01 -7.59269238e-01
-3.36463526e-02 5.86847067e-01 1.40074179e-01 -7.01655030e-01
-6.56807661e-01 -3.24222863e-01 -1.23584902e+00 5.55396914e-01
1.66754395e-01 -3.80464524e-01 -1.52498579e+00 1.20786333e+00
-8.84905756e-02 5.55336356e-01 3.35376769e-01 6.80277824e-01
9.92316604e-01 1.56907308e+00 3.61086428e-01 7.70203769e-02
1.27791548e+00 -1.09889662e+00 -6.39663041e-01 -1.29516393e-01
5.08673370e-01 -5.66186666e-01 7.01716721e-01 2.28591025e-01
-4.93567765e-01 -7.73001552e-01 -9.36592162e-01 -2.39071503e-01
-9.45397556e-01 3.82830024e-01 8.84442627e-01 2.37135381e-01
-1.33873749e+00 9.38567400e-01 -1.05297863e+00 -6.23143494e-01
2.78843403e-01 4.15772349e-01 2.53970665e-03 3.32730711e-01
-1.18291473e+00 1.10512912e+00 7.62459159e-01 2.87200093e-01
-9.71807599e-01 -8.93796742e-01 -6.86371148e-01 5.80613017e-01
-3.74336630e-01 -5.09283483e-01 1.44202816e+00 -1.09213758e+00
-1.64226580e+00 -1.65872321e-01 -3.05205584e-01 -7.82634258e-01
9.36094597e-02 -9.53899994e-02 -5.54019630e-01 -2.55377620e-01
-2.35104725e-01 7.74378359e-01 8.66934478e-01 -3.96148145e-01
-7.70086348e-01 1.32130489e-01 -1.44574851e-01 -2.52599090e-01
-2.10523143e-01 -2.50333287e-02 2.24549264e-01 -7.82468200e-01
-3.68300587e-01 -8.45881045e-01 -3.18528503e-01 -2.84814358e-01
2.69965734e-02 -8.54170501e-01 1.49449801e+00 -8.65542769e-01
1.10171103e+00 -1.94207764e+00 -1.12565830e-01 -1.47081956e-01
-9.41104144e-02 3.51884514e-01 -2.66349502e-02 6.28292143e-01
-3.58572483e-01 2.23146155e-01 -3.02830577e-01 -1.91479057e-01
-5.62354140e-02 6.23241067e-01 -7.46561170e-01 1.20589830e-01
3.40206206e-01 1.15135324e+00 -6.85986698e-01 1.26299694e-01
5.36536872e-01 8.24856460e-01 2.07746178e-01 1.86439693e-01
-7.82988787e-01 5.10940313e-01 -2.90495634e-01 2.75833040e-01
4.49915409e-01 -4.70937878e-01 -3.74099314e-02 -8.57238770e-02
-7.50388622e-01 3.45940202e-01 -2.76351452e-01 1.25094771e+00
-7.13447452e-01 1.34240282e+00 -8.55415046e-01 -9.69818830e-01
1.11831892e+00 5.35852194e-01 3.32439184e-01 -8.43774498e-01
7.61393085e-02 -6.57721609e-02 -2.23634824e-01 -4.65787590e-01
3.86623532e-01 -9.64427069e-02 5.11866137e-02 6.48589432e-01
5.72794341e-02 2.38208339e-01 1.07882731e-01 -1.95297316e-01
1.11894977e+00 4.43841249e-01 -9.63435993e-02 -4.34122324e-01
3.76342595e-01 7.45494068e-02 5.49126804e-01 4.64279979e-01
3.17849308e-01 3.61561388e-01 5.94686270e-01 -1.55942285e+00
-1.43157697e+00 -4.31479186e-01 1.89530566e-01 1.25720620e+00
-5.75514436e-01 -2.84724116e-01 -2.16666803e-01 -5.75841963e-01
-5.53996526e-02 1.18176317e+00 -1.13502967e+00 3.19431543e-01
-4.13792938e-01 -7.69101620e-01 4.63651270e-01 1.01179910e+00
4.71228927e-01 -1.67795432e+00 -1.00212395e+00 3.10482025e-01
1.12720020e-01 -8.07907343e-01 3.12763490e-02 3.85632455e-01
-1.02540398e+00 -6.05500281e-01 -5.93290210e-01 -4.21487272e-01
3.26160967e-01 -1.55090407e-01 1.36269820e+00 -1.92675069e-01
-4.24012505e-02 -6.87376410e-02 -4.29160267e-01 -7.85038948e-01
-2.33357877e-01 3.03635269e-01 -3.11395615e-01 -3.18265855e-01
5.54750383e-01 -9.27471519e-01 -5.59485614e-01 4.46977578e-02
-9.21115518e-01 2.27263615e-01 3.78411502e-01 8.23200107e-01
1.50852084e-01 1.31675139e-01 6.09424949e-01 -5.47763228e-01
4.68808949e-01 -7.80732632e-01 -9.42146301e-01 2.97216028e-01
-6.62224233e-01 4.27683502e-01 9.13758993e-01 -4.99631673e-01
-1.11657035e+00 -8.68584812e-02 -4.06432711e-03 -5.85917413e-01
-1.81689449e-02 7.06051350e-01 9.69563186e-01 8.90954733e-02
5.05989492e-01 4.60765541e-01 -1.73301086e-01 -5.51862299e-01
1.90232918e-01 3.83854300e-01 3.16091716e-01 -1.67094320e-01
3.25738609e-01 3.57735306e-01 1.90583676e-01 -6.51147544e-01
-7.89477468e-01 -1.42980322e-01 -8.27271998e-01 -3.24344099e-01
1.05010700e+00 -7.18225420e-01 -9.39143062e-01 4.67039973e-01
-1.41414428e+00 -6.75304174e-01 -2.03338295e-01 3.64047348e-01
-2.89328665e-01 -9.14572086e-03 -5.44345796e-01 -9.49637294e-01
-8.22708964e-01 -7.56942987e-01 9.65062320e-01 5.10061324e-01
-1.28099814e-01 -1.26685631e+00 8.31500143e-02 -3.10964942e-01
1.11388230e+00 4.60234165e-01 1.07362962e+00 -5.12523532e-01
-4.47864830e-01 -1.86721869e-02 -3.88344675e-01 1.98762655e-01
-1.04352474e-01 4.66138095e-01 -1.13858330e+00 -9.58010927e-02
-2.98968256e-01 -1.54337034e-01 1.04964602e+00 8.08655918e-01
1.50017214e+00 -4.23792928e-01 -3.30245227e-01 6.55160129e-01
1.35909629e+00 5.07331848e-01 8.74417305e-01 3.84385675e-01
5.84056854e-01 4.98353213e-01 2.25361705e-01 4.93595481e-01
2.99925745e-01 1.47631377e-01 5.89043140e-01 -7.81508014e-02
7.25855902e-02 5.30512780e-02 5.35794318e-01 6.68733299e-01
-3.53800356e-01 -5.61341524e-01 -1.10924530e+00 8.09797883e-01
-2.11789799e+00 -1.07528222e+00 -1.79066986e-01 1.81286979e+00
7.71439672e-02 -2.20018789e-01 -1.34805709e-01 -9.04817209e-02
4.31427211e-01 4.16142315e-01 -6.00368500e-01 -9.24549758e-01
-4.77726758e-02 4.39254969e-01 5.27756155e-01 2.54728973e-01
-9.79629457e-01 8.29841316e-01 6.60029364e+00 5.97926497e-01
-1.62001443e+00 2.29556099e-01 8.09166372e-01 -2.88370457e-02
-2.40178555e-01 1.52961416e-02 -5.95641315e-01 6.54907107e-01
1.72544265e+00 3.36548388e-02 1.91795155e-01 7.19016314e-01
3.62628222e-01 -5.64805530e-02 -9.26251411e-01 7.18721330e-01
-3.80872190e-01 -1.88592458e+00 -1.06500223e-01 -2.15166166e-01
8.34789932e-01 6.47161245e-01 -1.62288751e-02 6.45999551e-01
5.04976571e-01 -1.71858573e+00 4.10208136e-01 9.53398049e-01
5.36870718e-01 -7.97058821e-01 1.16954839e+00 3.37596089e-01
-1.30531096e+00 -3.83507013e-01 -6.49170756e-01 -5.31237125e-01
7.39122182e-02 4.71051186e-01 -9.84533787e-01 4.94599044e-01
1.41815138e+00 9.96433020e-01 -4.27070946e-01 7.24151850e-01
1.99646261e-02 9.89102066e-01 -6.78859055e-01 -3.55942458e-01
8.39580476e-01 -1.20366581e-01 1.16582640e-01 1.39928484e+00
9.30627286e-01 1.59614250e-01 3.55310552e-02 6.66174889e-01
2.46855319e-01 -2.90054709e-01 -6.20214701e-01 -3.91506076e-01
1.39282778e-01 1.17702949e+00 -6.16219521e-01 -5.75699985e-01
-3.21880698e-01 8.93430114e-01 2.47819722e-01 4.21613634e-01
-8.60792994e-01 -1.58798724e-01 4.96817529e-01 -2.84400862e-03
9.89442170e-01 -3.10811609e-01 -3.32346350e-01 -6.49069726e-01
-4.74717200e-01 -2.02837974e-01 5.31944156e-01 -1.55117941e+00
-1.25409770e+00 1.20996773e+00 -1.50768399e-01 -1.10553288e+00
-4.46615100e-01 -7.90422678e-01 -1.18595588e+00 1.36805940e+00
-1.63557506e+00 -1.52986085e+00 -5.49599111e-01 3.18967074e-01
8.20381641e-01 -2.21668318e-01 8.95610690e-01 -9.71871167e-02
-3.27544272e-01 -9.42891836e-02 4.35194701e-01 -1.59889370e-01
1.62565872e-01 -1.27683187e+00 8.13816726e-01 8.38147581e-01
-2.84148276e-01 1.92139313e-01 6.07745051e-01 -5.70818901e-01
-1.01524317e+00 -1.47593904e+00 1.14347708e+00 -4.04957026e-01
5.99560201e-01 5.66196851e-02 -1.03506351e+00 8.95692766e-01
9.24258888e-01 1.53364718e-01 4.73117292e-01 5.98298423e-02
-3.43095809e-01 -2.81451970e-01 -8.83173764e-01 7.43667260e-02
3.01913679e-01 -2.48541921e-01 -4.14874822e-01 5.34510195e-01
8.62901807e-01 -1.21427700e-01 -1.13675523e+00 4.97097164e-01
5.77995002e-01 -1.01113129e+00 6.43424392e-01 -6.08512223e-01
6.25599861e-01 -5.76476395e-01 1.68425053e-01 -1.17490172e+00
-8.68077695e-01 -5.48367381e-01 -4.52612966e-01 1.05334604e+00
4.85985696e-01 -5.85362732e-01 5.76006472e-01 2.39661887e-01
-3.31534624e-01 -6.86894298e-01 -8.85631144e-01 -5.73865950e-01
1.48793161e-01 -4.50033367e-01 8.38746428e-01 8.79648030e-01
-5.20171821e-01 2.03760430e-01 -6.22233927e-01 4.54455525e-01
5.91591932e-02 3.83493304e-01 2.02856094e-01 -1.33527935e+00
1.56196147e-01 -3.56382996e-01 -2.74676859e-01 -5.96618056e-01
8.23549405e-02 -4.88014549e-01 -1.18031278e-01 -1.70727980e+00
-2.11721450e-01 -3.35417598e-01 -5.84894955e-01 1.01604462e+00
3.53209227e-01 1.69306740e-01 6.40438795e-02 1.98672220e-01
-1.87243566e-01 6.68649077e-01 6.53386652e-01 -2.82892793e-01
-2.90573053e-02 -5.22203930e-03 1.28243249e-02 3.87165636e-01
1.11786520e+00 -3.52678359e-01 -4.47200835e-01 -5.92507899e-01
2.41082177e-01 -7.31277540e-02 3.97281945e-01 -1.14093876e+00
3.66273910e-01 -5.32192290e-02 9.67432082e-01 -1.24027157e+00
9.05212089e-02 -8.29661667e-01 7.59765565e-01 4.93759006e-01
-6.15831576e-02 5.38902283e-01 7.64300525e-01 4.02529269e-01
-1.42974541e-01 1.51044335e-02 4.61848199e-01 -6.09924085e-02
-1.13594413e+00 4.13708955e-01 -8.15661013e-01 -9.53875244e-01
9.79249656e-01 -8.97518545e-02 -3.68586063e-01 -3.57780993e-01
-6.27956390e-01 3.91871899e-01 4.49997969e-02 6.34133339e-01
2.56756544e-01 -1.24831176e+00 -6.88695371e-01 9.06825885e-02
-2.62323081e-01 -4.41313237e-02 4.81485397e-01 6.50826514e-01
-8.70305061e-01 7.66856074e-01 -4.36895609e-01 -8.31339478e-01
-9.76775885e-01 7.53616333e-01 5.07081151e-01 -4.19171304e-01
-7.96715856e-01 6.47103786e-01 2.35010400e-01 -4.20988411e-01
3.69226672e-02 -1.00174940e+00 -6.46088243e-01 3.75918411e-02
5.50192416e-01 3.11302572e-01 1.06771596e-01 -2.75280118e-01
-2.53667295e-01 5.76255798e-01 2.23564640e-01 2.52775133e-01
1.78442526e+00 1.68847337e-01 -2.63627708e-01 5.65970004e-01
7.79416800e-01 -8.98545623e-01 -1.66301823e+00 1.11108515e-02
2.06835121e-01 5.18115498e-02 4.40304458e-01 -1.26339638e+00
-1.19480121e+00 1.28380024e+00 8.14409018e-01 5.11737943e-01
1.38776326e+00 -5.21635950e-01 8.13562691e-01 2.92441666e-01
-1.74272358e-01 -6.47311449e-01 -3.49334657e-01 1.06428814e+00
1.04990363e+00 -1.09956276e+00 -2.35575989e-01 2.60662109e-01
-5.30116618e-01 1.83837724e+00 5.43807328e-01 -9.78469849e-02
8.17866981e-01 4.20979768e-01 -1.65578090e-02 -2.66934603e-01
-1.22397876e+00 -1.32739935e-02 4.01602685e-01 4.40625817e-01
5.59406757e-01 -6.48008194e-04 9.69239250e-02 3.08016032e-01
-8.66329446e-02 3.25415820e-01 1.69385165e-01 7.09920764e-01
-2.92660147e-01 -6.09736919e-01 -2.29986697e-01 5.11465967e-01
-5.16079068e-01 -3.69030118e-01 2.28907205e-02 3.97525758e-01
1.19978398e-01 6.60375118e-01 5.81096411e-01 -3.04630429e-01
-9.81540084e-02 3.41818750e-01 -3.83359343e-02 -1.16119616e-01
-9.40831304e-01 -2.20253676e-01 5.94566725e-02 -4.06439722e-01
-5.18453538e-01 -3.02152008e-01 -9.66117740e-01 -7.45396018e-01
1.62209451e-01 2.96563774e-01 8.08489799e-01 1.09946918e+00
7.90170372e-01 1.02317548e+00 3.68263066e-01 -1.25819063e+00
4.46459390e-02 -1.49614942e+00 -9.24155563e-02 -1.14003286e-01
5.81876636e-01 -4.54902768e-01 -9.12641883e-02 2.54633099e-01] | [6.6540141105651855, 2.848098039627075] |
65369582-e76a-4946-8af9-4250da27bd74 | fsd-fully-specialized-detector-via-neural | 2305.16649 | null | https://arxiv.org/abs/2305.16649v3 | https://arxiv.org/pdf/2305.16649v3.pdf | FSD: Fully-Specialized Detector via Neural Architecture Search | Most generic object detectors are mainly built for standard object detection tasks such as COCO and PASCAL VOC. They might not work well and/or efficiently on tasks of other domains consisting of images that are visually different from standard datasets. To this end, many advances have been focused on adapting a general-purposed object detector with limited domain-specific designs. However, designing a successful task-specific detector requires extraneous manual experiments and parameter tuning through trial and error. In this paper, we first propose and examine a fully-automatic pipeline to design a fully-specialized detector (FSD) which mainly incorporates a neural-architectural-searched model by exploring ideal network structures over the backbone and task-specific head. On the DeepLesion dataset, extensive results show that FSD can achieve 3.1 mAP gain while using approximately 40% fewer parameters on binary lesion detection task and improved the mAP by around 10% on multi-type lesion detection task via our region-aware graph modeling compared with existing general-purposed medical lesion detection networks. | ['Yudian Li', 'Zhe Huang'] | 2023-05-26 | null | null | null | null | ['architecture-search'] | ['methodology'] | [ 1.89216867e-01 2.51292080e-01 -2.95933843e-01 -2.14954272e-01
-5.78266442e-01 -1.32825851e-01 4.02860314e-01 5.48491292e-02
-5.36247790e-01 1.49530455e-01 -1.47510678e-01 -4.50033456e-01
2.91444719e-01 -4.92009163e-01 -3.81655604e-01 -4.83211637e-01
-1.49755254e-01 4.43424523e-01 1.21431005e+00 -1.28668159e-01
-1.21978968e-01 6.95828855e-01 -1.25671136e+00 2.08127350e-01
6.68233454e-01 8.43841612e-01 6.24265432e-01 6.36274457e-01
8.30026269e-02 5.46335042e-01 -5.36084116e-01 -3.44854563e-01
3.77960503e-01 -1.95950255e-01 -6.73059344e-01 2.36923829e-01
6.62771225e-01 -3.82474482e-01 -4.62094754e-01 1.21001446e+00
8.64044785e-01 -3.26934963e-01 7.70838499e-01 -1.03347862e+00
-5.18923402e-01 4.41241652e-01 -9.29234862e-01 5.26645362e-01
-1.63755462e-01 5.37660360e-01 7.03365028e-01 -7.38698959e-01
7.20488369e-01 1.23199117e+00 8.60742390e-01 7.47748017e-01
-1.07756639e+00 -5.00909686e-01 1.68774664e-01 1.65270165e-01
-1.40986061e+00 -1.46924734e-01 6.04511142e-01 -3.87779236e-01
1.09976554e+00 2.59530783e-01 6.64770186e-01 1.05948532e+00
1.98648468e-01 1.15436506e+00 7.71806300e-01 -3.67779046e-01
7.48238619e-03 2.23429784e-01 2.02866182e-01 1.05276489e+00
7.80711412e-01 -1.03983738e-01 -2.76772063e-02 1.34497657e-02
9.28753018e-01 -9.86452177e-02 -2.77288079e-01 -8.61780345e-01
-1.02824450e+00 8.43863308e-01 9.56246197e-01 3.51317197e-01
-2.03787848e-01 3.32151651e-02 6.17515326e-01 -2.13975683e-01
3.78208965e-01 4.17234629e-01 -1.38507500e-01 5.87487102e-01
-9.37691808e-01 -1.99620947e-02 6.06012762e-01 9.28607166e-01
3.36909473e-01 9.17550735e-03 -6.28801167e-01 9.54331815e-01
3.23893070e-01 2.66223162e-01 5.57875633e-01 -2.77625173e-01
2.24173769e-01 1.12353837e+00 -3.33564430e-01 -8.47189546e-01
-1.09545994e+00 -9.94316936e-01 -8.94368529e-01 3.97294551e-01
5.07282078e-01 2.09024325e-02 -1.59854460e+00 1.29576206e+00
4.92353022e-01 5.00149792e-03 -4.13417280e-01 1.15943253e+00
1.17604518e+00 -6.57226704e-03 3.14585179e-01 1.97017267e-01
1.68347573e+00 -1.23010325e+00 -4.28254992e-01 -6.79075003e-01
1.05799651e+00 -5.75959623e-01 1.13904023e+00 4.33978997e-02
-8.89346659e-01 -4.26308513e-01 -1.23277307e+00 -8.44707116e-02
-3.38014662e-01 4.61802334e-01 7.54182041e-01 9.97001946e-01
-1.15215933e+00 -5.06135598e-02 -9.50181723e-01 -7.46944845e-01
8.67071629e-01 1.94244459e-01 -3.12808096e-01 -2.73820519e-01
-6.40840232e-01 1.14208579e+00 6.55195773e-01 -7.40104392e-02
-1.13139558e+00 -8.33658636e-01 -7.03601539e-01 -1.65049940e-01
5.30970216e-01 -1.00712693e+00 1.21939433e+00 -4.90992397e-01
-9.77987587e-01 1.31641090e+00 2.88357258e-01 -5.77105343e-01
7.21638560e-01 2.26467744e-01 -3.17154378e-01 2.72960663e-01
5.52338995e-02 9.51540530e-01 7.48621047e-01 -9.52819824e-01
-6.34137094e-01 -3.33717734e-01 6.27097189e-02 2.49275357e-01
-4.03418094e-01 1.00694008e-01 -1.06697130e+00 -5.92881501e-01
-3.52765657e-02 -9.42431927e-01 -4.76048648e-01 6.52198851e-01
-6.95760429e-01 -3.25739503e-01 1.01675904e+00 -5.84595442e-01
9.89505410e-01 -2.09305835e+00 -1.64424747e-01 -4.11568135e-02
5.85752249e-01 6.99717104e-01 -2.54242927e-01 -2.00367853e-01
4.01587002e-02 -6.57472238e-02 -2.97924250e-01 -3.08308005e-01
-2.99604595e-01 -9.89109129e-02 4.94534880e-01 6.08967841e-01
2.77654856e-01 1.13763690e+00 -8.59924197e-01 -9.51570630e-01
3.39985818e-01 2.90463150e-01 -6.23669744e-01 -1.20060118e-02
-2.49149185e-02 -1.81119293e-02 -5.02106905e-01 9.75700200e-01
7.61951566e-01 -5.09836853e-01 1.28998578e-01 -4.72506553e-01
2.59354442e-01 -7.29601905e-02 -1.10148919e+00 1.90973330e+00
-1.98280513e-01 7.22051799e-01 3.37945074e-01 -9.23769712e-01
6.24083042e-01 -1.84975658e-02 2.54447728e-01 -6.49873137e-01
2.50021726e-01 2.41424918e-01 3.61286968e-01 -5.45685172e-01
8.09537899e-03 1.46912798e-01 2.58659273e-01 3.86220999e-02
1.88151643e-01 -1.43675163e-01 2.81061590e-01 3.49238694e-01
1.55649197e+00 -3.31752926e-01 4.60693568e-01 -3.30422133e-01
4.13292438e-01 2.88561970e-01 3.63010556e-01 9.11653876e-01
-7.96918094e-01 8.29444051e-01 4.38939303e-01 -3.09345543e-01
-6.73618913e-01 -7.93861270e-01 -4.98954713e-01 1.02658606e+00
2.79584646e-01 -1.69834390e-01 -6.54430807e-01 -1.22625589e+00
8.49337429e-02 6.80503249e-01 -6.44531667e-01 -1.67216524e-01
-5.21894932e-01 -1.23930645e+00 7.44189203e-01 6.06988251e-01
6.30931199e-01 -9.71032739e-01 -7.20208943e-01 5.33409677e-02
3.16384375e-01 -1.28558981e+00 -4.84012336e-01 2.23802418e-01
-7.53476679e-01 -1.23886788e+00 -1.14895415e+00 -1.10562348e+00
7.49622822e-01 6.92859471e-01 1.07398176e+00 1.11016840e-01
-1.11797965e+00 1.16696887e-01 -1.04765147e-01 -5.80922484e-01
-3.69927853e-01 2.71087408e-01 -3.16720188e-01 -2.12955102e-01
3.16408783e-01 -6.05428107e-02 -9.05612051e-01 4.31202739e-01
-8.34493637e-01 1.67240083e-01 1.14265561e+00 6.90257013e-01
3.85464340e-01 -3.25754821e-01 4.50656414e-01 -9.25073862e-01
6.98770523e-01 -2.90538639e-01 -4.09179628e-01 3.69312525e-01
-4.57301468e-01 -1.34663448e-01 8.88552666e-02 -4.73859191e-01
-9.81723070e-01 4.08484519e-01 -6.51754290e-02 -3.11719328e-01
-2.29336530e-01 1.85078792e-02 7.72622451e-02 -3.76201034e-01
1.20357645e+00 3.79194096e-02 6.63361046e-03 -4.57096636e-01
3.30590785e-01 6.31906807e-01 4.65268582e-01 -2.03982461e-02
7.11279690e-01 5.34142315e-01 2.70317681e-02 -7.58164108e-01
-7.22827554e-01 -8.54178488e-01 -6.70685530e-01 -2.63669938e-01
9.43861961e-01 -9.75152135e-01 -1.95514753e-01 5.73485970e-01
-9.89406943e-01 -4.34812129e-01 -1.43918335e-01 4.60379422e-01
-1.58312112e-01 3.49778354e-01 -5.39319575e-01 -3.61687094e-01
-4.61845070e-01 -1.27391946e+00 1.33496189e+00 2.47377619e-01
8.70006457e-02 -9.27512169e-01 -1.07693180e-01 2.03361541e-01
4.67507988e-01 1.00052245e-01 7.62053847e-01 -5.41441143e-01
-5.61220586e-01 -5.47762334e-01 -1.00857508e+00 3.04860324e-01
1.16305217e-01 -2.09403262e-01 -8.77777576e-01 -5.23595452e-01
-4.14667964e-01 -3.65549892e-01 1.27108204e+00 5.80929458e-01
1.19823122e+00 3.39710325e-01 -1.07489955e+00 7.19325781e-01
1.33798814e+00 -5.22682779e-02 2.81805336e-01 2.03062549e-01
7.20061183e-01 3.14575255e-01 2.37366229e-01 -1.24558799e-01
2.05001041e-01 7.43515968e-01 6.69071019e-01 -5.87374151e-01
-1.08117867e+00 -1.03913493e-01 2.79043755e-03 1.97912842e-01
3.45087610e-02 -2.83011436e-01 -1.17290223e+00 8.75162423e-01
-1.69196355e+00 -3.92820895e-01 -3.52165043e-01 1.67672741e+00
5.94686866e-01 4.64490265e-01 2.48095453e-01 -2.86788076e-01
8.39264691e-01 4.51693386e-02 -8.08164775e-01 5.53883202e-02
-3.95069197e-02 -1.30205303e-01 7.63271213e-01 4.54540510e-04
-1.43917572e+00 1.06343246e+00 6.42888689e+00 1.06018448e+00
-1.11462355e+00 3.39065999e-01 3.81945491e-01 -4.46698889e-02
2.20044017e-01 -4.11410570e-01 -9.66142833e-01 -3.78371887e-02
3.71690810e-01 2.37203270e-01 -3.13037992e-01 1.16655767e+00
-7.74674714e-02 -6.45226762e-02 -9.76885617e-01 1.09677172e+00
1.73297033e-01 -1.37094569e+00 -4.01965939e-02 3.44429053e-02
5.82361460e-01 3.47071111e-01 -6.98752254e-02 3.35728794e-01
2.60740727e-01 -8.30205142e-01 3.61513048e-01 2.67398078e-02
8.43901575e-01 -1.33597150e-01 6.79157615e-01 4.16996419e-01
-1.01601434e+00 1.13495693e-01 -6.58429623e-01 3.81302327e-01
1.56318489e-02 5.41545331e-01 -1.45706129e+00 2.65593439e-01
8.46186578e-01 5.90132296e-01 -1.20407450e+00 1.80891907e+00
-2.54369617e-01 6.75872862e-01 -4.81561303e-01 -1.82264313e-01
4.40704972e-01 5.15747845e-01 7.05299914e-01 1.71246362e+00
1.04594138e-02 -2.59007663e-01 2.60550290e-01 7.38876104e-01
-4.89552394e-02 4.83378768e-02 -5.31709075e-01 4.52003837e-01
1.51460662e-01 1.64505219e+00 -1.08481896e+00 -1.93597928e-01
-4.68822896e-01 8.53688240e-01 4.58218753e-01 1.93554401e-01
-9.97689009e-01 -2.74798930e-01 3.75703901e-01 3.40379715e-01
3.78622204e-01 -9.87445116e-02 -2.87882537e-01 -8.73603404e-01
-2.21743472e-02 -6.13038421e-01 5.47567427e-01 -4.89665538e-01
-1.36549962e+00 5.97435713e-01 -9.42621976e-02 -1.09069061e+00
1.59682184e-01 -9.55389798e-01 -6.87662899e-01 6.12308443e-01
-1.66940105e+00 -1.33333385e+00 -5.58683872e-01 5.76839447e-01
5.60012698e-01 -2.73968607e-01 6.60526216e-01 4.13103461e-01
-1.08227122e+00 8.39700222e-01 -3.75739813e-01 3.52984399e-01
7.61093378e-01 -1.15891576e+00 4.81899977e-01 9.83587921e-01
2.68798061e-02 6.64515868e-02 3.67070287e-01 -6.46159410e-01
-1.10759974e+00 -1.43978930e+00 2.73334622e-01 -4.32942390e-01
6.24446929e-01 -4.52488482e-01 -8.84499073e-01 5.05267680e-01
1.05031252e-01 5.33653498e-01 1.19136184e-01 -1.53229788e-01
-3.79673749e-01 1.82592403e-02 -1.33332312e+00 6.54449403e-01
1.57875824e+00 -1.60039634e-01 -3.44321668e-01 8.32152009e-01
6.58633173e-01 -4.98692721e-01 -5.08531272e-01 5.78589439e-01
8.34749490e-02 -7.41157591e-01 1.20182014e+00 -5.79842806e-01
-1.96851268e-01 -4.32645380e-01 1.77674249e-01 -1.12867284e+00
-6.59588993e-01 -2.10770950e-01 -1.31624192e-01 7.23862946e-01
4.52307910e-01 -5.21441758e-01 9.86294687e-01 2.56134182e-01
-5.60572088e-01 -7.80862808e-01 -9.03088748e-01 -9.27462637e-01
-1.18073650e-01 -5.39869606e-01 6.37443289e-02 6.27040148e-01
-4.23758328e-01 2.66295582e-01 2.35580318e-02 3.78044486e-01
7.69962966e-01 -2.93435603e-01 6.58488035e-01 -1.04454458e+00
-1.67678073e-01 -7.96553612e-01 -8.51993501e-01 -9.45998847e-01
-3.26290607e-01 -1.25224328e+00 3.46246213e-02 -1.79726744e+00
4.39756721e-01 -4.05467719e-01 -3.76328498e-01 7.22373068e-01
-3.16242963e-01 5.23550510e-01 6.89198449e-02 1.33274481e-01
-6.36159122e-01 1.23013534e-01 1.58704066e+00 -4.98094529e-01
-1.55193999e-01 -5.34452237e-02 -7.29446888e-01 9.00661349e-01
6.25476420e-01 -5.32956183e-01 -4.14186060e-01 -5.09916782e-01
-2.19958648e-01 -3.26751232e-01 7.15379119e-01 -1.31522214e+00
3.94902945e-01 1.18015237e-01 5.03072798e-01 -6.32529438e-01
2.44411886e-01 -5.84769666e-01 -3.21641982e-01 8.53035271e-01
-4.07634303e-02 -5.58279872e-01 4.02042687e-01 7.69475698e-01
3.43913622e-02 -2.55723864e-01 1.08945704e+00 -2.02622071e-01
-1.15105724e+00 3.20122838e-01 -1.01804860e-01 3.30361091e-02
1.49670184e+00 -3.28332335e-01 -5.11952341e-01 1.93042502e-01
-6.26236439e-01 2.43883789e-01 2.05395013e-01 6.41702175e-01
6.19680405e-01 -9.49253023e-01 -9.54721451e-01 1.13786131e-01
4.84540671e-01 2.46572495e-02 3.64699960e-01 9.17972088e-01
-7.40025818e-01 7.31602728e-01 -2.06316546e-01 -1.00770307e+00
-1.29426122e+00 5.77203035e-01 7.77568817e-01 -5.33506155e-01
-9.88515913e-01 1.19810605e+00 5.27792096e-01 -4.29636091e-01
3.61774027e-01 -5.23163259e-01 -1.69806436e-01 -1.63111798e-02
4.75788325e-01 2.24335298e-01 3.80855322e-01 -3.37118089e-01
-7.64771402e-01 4.45017695e-01 -3.92007977e-01 5.41960537e-01
1.11800241e+00 2.52397716e-01 4.08853024e-01 -2.02438921e-01
1.14127493e+00 -5.14114320e-01 -1.24502659e+00 -2.32994735e-01
-1.19144604e-01 -1.47744745e-01 4.89182591e-01 -9.40610170e-01
-1.39759469e+00 8.03710699e-01 1.11136556e+00 1.93442941e-01
1.04634917e+00 3.13354313e-01 5.95012009e-01 3.31118315e-01
4.21701342e-01 -8.75103712e-01 2.17933118e-01 1.07871890e-01
9.79803145e-01 -1.39286423e+00 8.18338990e-02 -7.19274461e-01
-5.28750360e-01 8.73232305e-01 9.51761425e-01 -1.69215634e-01
7.31844842e-01 1.68210119e-01 -5.83737120e-02 -5.37980258e-01
-2.99547255e-01 -6.41770422e-01 8.74437511e-01 9.26306784e-01
2.36461878e-01 2.86120206e-01 -2.53198028e-01 2.69687414e-01
2.61244476e-01 -1.69480473e-01 2.24224433e-01 8.39325249e-01
-6.20033860e-01 -7.40694463e-01 -2.45780617e-01 8.54593873e-01
-9.17422399e-02 1.18871152e-01 -4.06234384e-01 1.23802125e+00
1.40612707e-01 6.71132684e-01 -1.14022009e-01 -1.25010401e-01
5.37200630e-01 -1.13944612e-01 4.22116041e-01 -9.05142307e-01
-4.66652066e-01 -5.30357547e-02 1.60901710e-01 -5.41368783e-01
-1.56029835e-01 -4.79060829e-01 -1.12195051e+00 2.04528153e-01
-5.13372958e-01 -6.54411435e-01 7.62648761e-01 5.21800518e-01
3.93056750e-01 9.18748021e-01 -6.37028068e-02 -7.74045229e-01
-4.51650620e-01 -9.38823760e-01 -4.11651403e-01 2.74685770e-01
-1.45504046e-02 -8.04785073e-01 -3.84763479e-02 -4.04503345e-01] | [15.095234870910645, -2.379183769226074] |
083f8efd-9f97-479c-912f-030a0da67881 | sketch-a-shape-zero-shot-sketch-to-3d-shape | 2307.03869 | null | https://arxiv.org/abs/2307.03869v1 | https://arxiv.org/pdf/2307.03869v1.pdf | Sketch-A-Shape: Zero-Shot Sketch-to-3D Shape Generation | Significant progress has recently been made in creative applications of large pre-trained models for downstream tasks in 3D vision, such as text-to-shape generation. This motivates our investigation of how these pre-trained models can be used effectively to generate 3D shapes from sketches, which has largely remained an open challenge due to the limited sketch-shape paired datasets and the varying level of abstraction in the sketches. We discover that conditioning a 3D generative model on the features (obtained from a frozen large pre-trained vision model) of synthetic renderings during training enables us to effectively generate 3D shapes from sketches at inference time. This suggests that the large pre-trained vision model features carry semantic signals that are resilient to domain shifts, i.e., allowing us to use only RGB renderings, but generalizing to sketches at inference time. We conduct a comprehensive set of experiments investigating different design factors and demonstrate the effectiveness of our straightforward approach for generation of multiple 3D shapes per each input sketch regardless of their level of abstraction without requiring any paired datasets during training. | ['Saeid Asgari Taghanaki', 'Evan Atherton', 'Hooman Shayani', 'Joseph Lambourne', 'Arianna Rampini', 'Pradeep Kumar Jayaraman', 'Aditya Sanghi'] | 2023-07-08 | null | null | null | null | ['3d-shape-generation', 'text-to-shape-generation'] | ['computer-vision', 'computer-vision'] | [ 4.31356072e-01 2.01714769e-01 4.50386852e-01 -2.51657039e-01
-6.14332080e-01 -9.45711136e-01 1.12672317e+00 -3.43106866e-01
4.91536781e-02 2.83148825e-01 1.16692506e-01 -3.86682749e-01
1.73309445e-01 -9.43482280e-01 -1.09300125e+00 -3.80397677e-01
2.49977201e-01 6.50489986e-01 1.15672108e-02 -2.08807141e-01
2.57530272e-01 9.46399808e-01 -1.83732545e+00 4.80423987e-01
5.12592256e-01 7.96402752e-01 1.63980335e-01 9.68732595e-01
-3.80838513e-01 2.13118836e-01 -6.65863037e-01 -3.87875021e-01
6.46931648e-01 -4.23907667e-01 -4.30237561e-01 3.61975759e-01
7.13483870e-01 -6.02303088e-01 -2.00728193e-01 6.80350184e-01
5.18945575e-01 3.58332209e-02 1.02233672e+00 -1.12400186e+00
-8.93868983e-01 1.30358815e-01 -2.21412525e-01 -4.01624143e-01
2.45071724e-01 5.03908098e-01 9.40876842e-01 -1.00182021e+00
9.62083697e-01 1.43496025e+00 6.03197515e-01 7.54479885e-01
-1.65690863e+00 -5.20104587e-01 1.35646909e-01 -5.89420795e-01
-1.06864262e+00 -6.06790245e-01 9.24419284e-01 -6.16686404e-01
1.00225365e+00 1.34728238e-01 8.09951246e-01 1.23860002e+00
-2.15533331e-01 7.15920448e-01 1.14161146e+00 -4.27225381e-01
3.69787246e-01 1.81034207e-01 -3.15550834e-01 8.27453494e-01
8.35807323e-02 1.30800605e-01 -6.69951022e-01 -2.47606114e-01
1.50615239e+00 -1.88467488e-01 -8.24183822e-02 -5.00385344e-01
-1.08646774e+00 6.81334734e-01 3.75121504e-01 7.50773400e-02
-3.22939992e-01 3.64507467e-01 7.91565422e-03 3.52800339e-01
4.47782546e-01 8.17918718e-01 -3.89626145e-01 7.69358501e-03
-8.80756915e-01 7.88374424e-01 7.09078193e-01 1.34159780e+00
7.36878455e-01 2.12907851e-01 -2.20912606e-01 6.34076834e-01
1.73806250e-01 5.46591580e-01 5.04120775e-02 -1.06759894e+00
3.22616845e-01 5.91544151e-01 9.33391750e-02 -6.90937757e-01
5.75196519e-02 -1.43647656e-01 -5.03430068e-01 6.28820360e-01
6.48644328e-01 -2.81838961e-02 -1.17444456e+00 1.72231436e+00
1.66460305e-01 9.11017880e-02 1.15841087e-02 8.02490711e-01
7.56193817e-01 4.96461272e-01 -5.96845038e-02 4.38930809e-01
1.09257054e+00 -3.80549580e-01 9.89299268e-02 -9.96986181e-02
2.22035751e-01 -7.91160762e-01 1.52170038e+00 2.75988311e-01
-1.38872933e+00 -7.36427546e-01 -9.79193807e-01 -3.29483032e-01
-3.65237772e-01 8.35276991e-02 7.22008765e-01 5.99006772e-01
-1.05799747e+00 9.21608686e-01 -6.40848160e-01 -4.94463533e-01
8.27785254e-01 2.04660714e-01 -2.78755814e-01 -1.21105969e-01
-6.72540724e-01 7.11427689e-01 -1.69201151e-01 -4.10491787e-02
-9.81934309e-01 -9.11149919e-01 -5.70121646e-01 1.95761826e-02
3.79663967e-02 -1.29758298e+00 1.17137706e+00 -7.74232447e-01
-1.63532650e+00 1.04441786e+00 -1.79632470e-01 2.38067731e-02
5.91522753e-01 5.25442287e-02 2.16431499e-01 8.55577365e-02
-7.18276799e-02 9.94749606e-01 1.21768439e+00 -1.58382154e+00
-1.49686038e-01 -5.13674200e-01 3.25678349e-01 1.66249603e-01
-2.39230581e-02 -3.74902815e-01 -4.04910743e-01 -6.62425160e-01
2.57706195e-01 -1.10285568e+00 -1.78742692e-01 4.51465130e-01
-3.27082157e-01 -6.26255125e-02 3.31356347e-01 -2.26569504e-01
4.24858660e-01 -1.90774906e+00 3.81553441e-01 3.80984694e-01
2.31334083e-02 2.76410114e-02 -4.06671166e-01 4.75753158e-01
1.77462682e-01 4.14398611e-01 -1.85304031e-01 -5.58616936e-01
2.47814983e-01 1.52723655e-01 -8.58485699e-01 -4.09025103e-02
7.49599695e-01 1.13528383e+00 -7.91461647e-01 -1.59030631e-01
3.00364494e-01 5.33052862e-01 -7.22490907e-01 5.75183094e-01
-7.18478084e-01 6.75510764e-01 -4.97599632e-01 5.86815357e-01
6.21160150e-01 -1.98047385e-01 -4.26064059e-02 -1.58570021e-01
1.07143953e-01 2.48073101e-01 -8.59013319e-01 2.10046864e+00
-6.97188377e-01 4.90443438e-01 -9.40367579e-02 -6.68122292e-01
1.00270224e+00 2.07484871e-01 5.21115810e-02 -5.52318394e-01
-7.95832500e-02 2.20457345e-01 -2.44378641e-01 -2.19574213e-01
3.62418830e-01 -5.89535534e-01 -4.75995155e-04 7.94167221e-01
1.58822224e-01 -1.00055707e+00 -1.47737011e-01 1.51844800e-01
1.01515305e+00 6.82959378e-01 -4.73258585e-01 -2.86821909e-02
-3.20131779e-02 3.77385207e-02 -7.69277290e-02 9.30194855e-01
3.31160337e-01 1.05060983e+00 3.81672502e-01 -3.68213147e-01
-1.51417053e+00 -1.47447658e+00 1.71194575e-03 9.33455229e-01
-1.45707712e-01 -2.42567658e-01 -6.08813405e-01 -4.66514200e-01
2.72947550e-01 7.57072270e-01 -5.59068501e-01 -2.21282870e-01
-5.05717635e-01 -3.86070907e-01 6.64448440e-01 5.57603955e-01
3.77633162e-02 -1.06730485e+00 -6.23282313e-01 6.05960935e-02
4.93796855e-01 -1.01208735e+00 -6.39161319e-02 -7.35951737e-02
-1.17605102e+00 -7.08437562e-01 -9.06011403e-01 -5.83830655e-01
9.60176468e-01 2.51427948e-01 1.37659633e+00 1.52509451e-01
-3.45038950e-01 7.39464164e-01 -4.77318801e-02 -5.40250123e-01
-6.50676489e-01 -4.34538424e-02 -2.36084282e-01 -9.82521698e-02
-1.31748125e-01 -1.07430696e+00 -5.31039596e-01 1.26584560e-01
-8.94141436e-01 4.49502140e-01 5.73966682e-01 6.70510769e-01
4.69554096e-01 -2.98261434e-01 4.54283655e-01 -8.22427630e-01
8.63898277e-01 -1.50794744e-01 -5.75032175e-01 1.27228037e-01
-2.62342542e-01 4.90143508e-01 7.54329324e-01 -4.50301319e-01
-1.08300161e+00 1.30672053e-01 -1.07525848e-01 -7.88566291e-01
-3.74141157e-01 -7.99484402e-02 -1.47360697e-01 6.36958927e-02
6.69555962e-01 2.63976693e-01 6.33526221e-02 -5.49039185e-01
9.24344838e-01 2.72976100e-01 4.58287239e-01 -1.13487649e+00
1.12080634e+00 5.58190644e-01 3.34110439e-01 -9.27067280e-01
-4.37283486e-01 2.04460025e-01 -6.31397486e-01 -1.02980673e-01
7.55935371e-01 -7.98886001e-01 -5.74524879e-01 5.06992161e-01
-1.35696006e+00 -5.54910839e-01 -4.74428862e-01 -3.96270305e-02
-8.50641429e-01 3.67055763e-03 -5.32882214e-01 -7.52812862e-01
-2.29800567e-01 -1.06182277e+00 1.48626816e+00 -2.25542933e-02
-4.72990453e-01 -8.06141853e-01 -1.76827967e-01 1.21787131e-01
4.28814083e-01 3.56875598e-01 1.39885414e+00 -8.56856108e-02
-1.03341293e+00 -1.12174071e-01 -3.02633941e-01 1.99428961e-01
1.73979163e-01 1.06488086e-01 -1.39991641e+00 -5.24178073e-02
-3.09336066e-01 -5.75411797e-01 7.86687791e-01 1.18926421e-01
1.11925054e+00 -5.46003040e-03 -1.74038813e-01 6.56596482e-01
1.12371480e+00 -1.40925676e-01 4.59400117e-01 -2.47366861e-01
7.97866642e-01 5.69722354e-01 4.52931449e-02 3.64179045e-01
1.61305487e-01 4.16665465e-01 9.37217772e-02 -3.73941623e-02
-3.81855696e-01 -6.99785411e-01 9.08209756e-02 3.96231323e-01
-2.74502814e-01 -8.21750686e-02 -6.73428774e-01 3.89185786e-01
-1.47235048e+00 -7.98693597e-01 1.47734284e-01 2.16449785e+00
8.28154147e-01 3.25935662e-01 -4.41286340e-02 -8.93777907e-02
4.63136911e-01 5.55106476e-02 -7.53355265e-01 -4.75615352e-01
-3.33312452e-02 7.32101321e-01 3.52137126e-02 3.53783876e-01
-6.24204636e-01 9.63554442e-01 6.50026941e+00 4.91928577e-01
-1.23412228e+00 -3.33910733e-01 5.39608002e-01 -2.05442160e-01
-9.95488107e-01 1.18274979e-01 -4.81799960e-01 2.37587586e-01
5.49396515e-01 1.00549556e-01 7.94133306e-01 6.70449078e-01
1.97999924e-02 8.05604309e-02 -1.59519398e+00 8.83987367e-01
-1.64709121e-01 -1.59683454e+00 5.91965318e-01 9.51108858e-02
8.11081827e-01 -2.27442846e-01 2.53811598e-01 1.56524003e-01
4.58624274e-01 -1.20253789e+00 9.39520180e-01 6.91353500e-01
1.04531085e+00 -4.01988894e-01 -1.62760079e-01 4.23900247e-01
-9.02084291e-01 2.97528923e-01 -3.96757275e-01 -1.55358672e-01
4.61353734e-02 5.02977669e-01 -1.10685110e+00 2.66926348e-01
2.50329524e-01 4.46790189e-01 -3.45802665e-01 4.41263527e-01
-2.48631358e-01 3.71445179e-01 -4.60530221e-01 2.52159685e-02
9.89624187e-02 -1.16733789e-01 5.21582246e-01 8.62926126e-01
5.65137267e-01 4.17579301e-02 -1.17036402e-01 1.69631684e+00
-1.66867182e-01 -3.24519038e-01 -9.87099349e-01 -3.73618633e-01
5.36821246e-01 1.04551625e+00 -7.62181461e-01 -3.36523443e-01
-2.81607360e-01 1.16098142e+00 3.93721074e-01 5.09769201e-01
-5.40444493e-01 -1.71118632e-01 6.94296598e-01 3.39580119e-01
4.94026095e-01 -5.15710592e-01 -7.45003402e-01 -1.04678321e+00
7.92728439e-02 -5.05556762e-01 -2.84137666e-01 -1.18462265e+00
-1.40988064e+00 4.39927220e-01 -1.83518022e-01 -9.51725245e-01
-2.96845794e-01 -6.80794001e-01 -7.21871018e-01 1.16342676e+00
-1.21909010e+00 -1.45684564e+00 -3.00598174e-01 4.43406731e-01
4.86428708e-01 -1.27716020e-01 9.35816169e-01 -2.09316865e-01
-1.08596087e-01 4.50826943e-01 -2.88616091e-01 1.29807368e-03
5.22655904e-01 -1.21502316e+00 1.10536075e+00 6.30359292e-01
4.96739388e-01 9.15765703e-01 4.88102853e-01 -6.47093952e-01
-1.92144048e+00 -8.93055975e-01 4.39203173e-01 -9.91205156e-01
2.66036421e-01 -8.45144808e-01 -7.22342253e-01 5.67322671e-01
-1.37333378e-01 -6.87602162e-02 3.77168834e-01 3.20395753e-02
-6.86247289e-01 1.82940587e-01 -1.13963652e+00 8.99870813e-01
1.64717698e+00 -8.35675478e-01 -5.28806031e-01 -5.34998849e-02
6.64977193e-01 -4.19725507e-01 -7.90344417e-01 1.80920213e-01
8.77533078e-01 -1.00097477e+00 1.18169677e+00 -8.42229187e-01
6.90557599e-01 -2.71989733e-01 -1.85507491e-01 -1.20389116e+00
-2.67104506e-01 -5.80204368e-01 -2.89000403e-02 1.22789240e+00
3.29761147e-01 -2.26920187e-01 7.29710758e-01 1.02433360e+00
-1.61158055e-01 -6.61011338e-01 -5.55125117e-01 -6.07508898e-01
3.38941395e-01 -6.16243243e-01 8.32238019e-01 6.20626330e-01
-5.45807481e-01 3.92039031e-01 -1.24279493e-02 -8.73758569e-02
3.83996338e-01 6.69238448e-01 1.22162056e+00 -1.36499214e+00
-4.73968446e-01 -5.75487137e-01 -8.41903165e-02 -1.25333166e+00
2.43222505e-01 -1.13687730e+00 1.41390404e-02 -1.48409331e+00
-1.18049636e-01 -8.68848503e-01 1.45456254e-01 3.95633548e-01
1.66225377e-02 2.87882090e-01 4.58684862e-01 1.30945936e-01
4.83829621e-03 5.68152308e-01 1.61241603e+00 -7.54295141e-02
-2.59359717e-01 -1.29775656e-03 -9.43232059e-01 6.06349766e-01
4.23752934e-01 -2.79710084e-01 -6.11137986e-01 -7.05390871e-01
1.14173770e-01 -6.72090054e-02 7.79747188e-01 -7.90846884e-01
-2.30988234e-01 -1.70098454e-01 8.31551731e-01 -2.93325454e-01
5.73756993e-01 -6.50982738e-01 1.36666611e-01 5.08192368e-02
-4.11678463e-01 -1.40896767e-01 3.65361273e-01 4.49714690e-01
3.89166027e-01 -3.89319193e-03 5.95751226e-01 -6.21911943e-01
-4.44420218e-01 2.88535655e-01 -8.38383436e-02 4.20653559e-02
5.00420809e-01 -4.96933818e-01 -5.88788167e-02 -2.40525067e-01
-7.91570723e-01 -4.42369342e-01 9.18450058e-01 4.25168276e-01
6.10823631e-01 -1.45324683e+00 -5.77782810e-01 6.35071933e-01
1.85280479e-02 4.32681948e-01 -4.81450595e-02 1.30550331e-02
-2.55687863e-01 1.73033476e-01 -3.55781704e-01 -7.11876512e-01
-7.88560271e-01 4.17055964e-01 1.65970713e-01 1.35573670e-01
-7.51222372e-01 9.31151092e-01 3.86490077e-01 -4.69081372e-01
6.69568107e-02 -6.71127141e-01 6.69008076e-01 -1.97138548e-01
1.62489116e-01 8.00025389e-02 6.00157715e-02 -2.33305499e-01
-8.52193683e-02 8.60298693e-01 2.97420800e-01 -3.93310636e-01
1.44570887e+00 3.40925813e-01 2.78582871e-01 3.76546383e-01
1.09980297e+00 -1.23642255e-02 -1.63157463e+00 8.91063586e-02
-3.77733678e-01 -5.88692784e-01 -1.65393919e-01 -7.51922429e-01
-8.77038062e-01 1.22311425e+00 1.40485376e-01 -3.30942497e-02
7.58869827e-01 1.31701916e-01 6.70926213e-01 4.77680802e-01
6.64047182e-01 -7.61444986e-01 2.98902422e-01 3.29011381e-01
1.21889591e+00 -9.69961107e-01 -1.97975919e-01 -1.99039012e-01
-4.67442483e-01 1.20469415e+00 3.90184760e-01 -5.04030049e-01
4.13612396e-01 3.03529769e-01 -1.49060920e-01 -1.96676299e-01
-6.37319565e-01 -1.66869551e-01 3.35038722e-01 7.81772256e-01
3.73460442e-01 -4.77315076e-02 4.59316760e-01 4.37722474e-01
-3.63710582e-01 8.00566897e-02 3.17853808e-01 8.51336062e-01
-1.24091983e-01 -1.20859575e+00 -2.80476123e-01 4.55688208e-01
1.98219091e-01 -1.15606479e-01 -8.21572959e-01 5.58224738e-01
2.15109840e-01 4.91404772e-01 2.14546502e-01 -2.58264124e-01
4.85417902e-01 3.75720590e-01 1.20549297e+00 -7.64203668e-01
-3.52370650e-01 -1.02724463e-01 -6.56537572e-03 -3.75777274e-01
-1.58374861e-01 -6.61709607e-01 -1.23116338e+00 -2.07962543e-01
-3.00762132e-02 -3.31538081e-01 7.03204751e-01 6.80634856e-01
7.03586340e-01 2.86558986e-01 5.15149653e-01 -1.46087825e+00
-4.98238713e-01 -6.09271407e-01 -4.09612745e-01 7.33200967e-01
2.38516688e-01 -7.48538613e-01 -2.45423630e-01 2.40920082e-01] | [9.104023933410645, -3.498377799987793] |
ce4dffa9-621c-49e0-8a65-be29586d942f | energy-optimization-for-hvac-systems-in-multi | 2306.13333 | null | https://arxiv.org/abs/2306.13333v1 | https://arxiv.org/pdf/2306.13333v1.pdf | Energy Optimization for HVAC Systems in Multi-VAV Open Offices: A Deep Reinforcement Learning Approach | With more than 32% of the global energy used by commercial and residential buildings, there is an urgent need to revisit traditional approaches to Building Energy Management (BEM). With HVAC systems accounting for about 40% of the total energy cost in the commercial sector, we propose a low-complexity DRL-based model with multi-input multi-output architecture for the HVAC energy optimization of open-plan offices, which uses only a handful of controllable and accessible factors. The efficacy of our solution is evaluated through extensive analysis of the overall energy consumption and thermal comfort levels compared to a baseline system based on the existing HVAC schedule in a real building. This comparison shows that our method achieves 37% savings in energy consumption with minimum violation (<1%) of the desired temperature range during work hours. It takes only a total of 40 minutes for 5 epochs (about 7.75 minutes per epoch) to train a network with superior performance and covering diverse conditions for its low-complexity architecture; therefore, it easily adapts to changes in the building setups, weather conditions, occupancy rate, etc. Moreover, by enforcing smoothness on the control strategy, we suppress the frequent and unpleasant on/off transitions on HVAC units to avoid occupant discomfort and potential damage to the system. The generalizability of our model is verified by applying it to different building models and under various weather conditions. | ['Abolfazl Razi', 'Edward. Duffy', 'Natan Vital', 'Xiwen Chen', 'Hao Wang'] | 2023-06-23 | null | null | null | null | ['total-energy', 'energy-management'] | ['miscellaneous', 'time-series'] | [-5.27063385e-02 6.43192679e-02 1.19582705e-01 -1.75613418e-01
-4.20320958e-01 -6.34949625e-01 7.56021440e-02 2.09824115e-01
-9.95899513e-02 6.75917923e-01 -1.30898431e-01 -4.66772795e-01
-3.63058716e-01 -1.02546358e+00 -4.33627039e-01 -1.04305172e+00
9.55998898e-02 -6.63725520e-03 -2.83709884e-01 -2.35242456e-01
-5.28840363e-01 7.16575801e-01 -1.61669815e+00 -2.34931841e-01
7.58458376e-01 1.18173075e+00 3.97292167e-01 5.95121205e-01
7.21816123e-01 3.45361024e-01 -6.26798689e-01 4.91546690e-01
3.55520517e-01 -3.55745226e-01 -6.94139540e-01 6.91172853e-02
-6.81156144e-02 -3.87383044e-01 -1.08165048e-01 4.79199439e-01
8.80781114e-01 7.61629641e-01 3.28034103e-01 -1.29607236e+00
-2.39840001e-01 9.04751271e-02 6.09071180e-02 3.56244519e-02
1.15938284e-01 2.88703948e-01 8.36001635e-01 -1.65267780e-01
-2.65365064e-01 4.92852151e-01 3.59695494e-01 4.66308981e-01
-1.52811503e+00 -3.76593113e-01 2.85720974e-01 2.00863376e-01
-1.64448512e+00 -2.43981242e-01 7.99889803e-01 9.92883556e-03
1.38972926e+00 1.10489964e+00 1.16355240e+00 7.00719595e-01
2.32835248e-01 3.50664794e-01 1.11893785e+00 -2.62392491e-01
8.15450907e-01 1.56662837e-01 -9.51809287e-02 3.94436330e-01
-4.90350015e-02 7.46978745e-02 3.15250725e-01 -1.09869190e-01
3.91461909e-01 3.41869779e-02 -5.15924513e-01 -2.84455538e-01
-7.30369508e-01 4.95213866e-01 6.97598755e-01 4.32940304e-01
-1.03920192e-01 2.01689184e-01 1.00444168e-01 -8.51662904e-02
3.92227946e-03 7.00468838e-01 -6.21937752e-01 -1.81885567e-02
-9.02214825e-01 9.26237851e-02 6.57709837e-01 9.81156826e-01
5.24003088e-01 1.11930527e-01 -4.15045440e-01 8.81473899e-01
1.56570077e-01 5.42033613e-01 1.92684203e-01 -1.00118804e+00
1.07920714e-01 5.98972261e-01 4.92995739e-01 -6.34838045e-01
-8.23291957e-01 -4.56634372e-01 -1.23134339e+00 5.33775575e-02
-8.20401981e-02 -2.63570428e-01 -1.02821934e+00 1.71140099e+00
3.83991331e-01 -3.39977384e-01 -5.15099168e-01 7.34032154e-01
1.47385255e-01 1.08139193e+00 6.18908666e-02 -6.98298216e-01
1.32525527e+00 -8.48439336e-01 -8.89959693e-01 -2.12282792e-01
1.61347777e-01 -4.27477688e-01 1.07384539e+00 1.06202655e-01
-1.08959615e+00 -4.06292498e-01 -1.27834046e+00 -3.59368906e-03
-6.84200168e-01 -8.97051953e-03 1.83018863e-01 1.01661384e+00
-1.32499433e+00 4.95748520e-01 -9.30785477e-01 -2.37081885e-01
-4.57909033e-02 5.66249251e-01 4.78437126e-01 2.67229795e-01
-1.18409455e+00 1.00927806e+00 4.32045937e-01 6.84868932e-01
-8.06474864e-01 -7.83801138e-01 -6.71202838e-01 5.68365932e-01
6.06642902e-01 -7.52763569e-01 1.38580179e+00 -1.81995198e-01
-1.72870541e+00 -1.45366415e-01 1.06461190e-01 -4.67677712e-02
2.71736264e-01 -1.98768988e-01 -5.22166789e-01 -2.44084254e-01
-5.63229024e-01 2.57305413e-01 3.59040320e-01 -1.32293200e+00
-5.05215883e-01 3.04166190e-02 5.06979600e-02 9.67548937e-02
-4.63461399e-01 -4.58166659e-01 -7.90235400e-02 -2.59749353e-01
-2.24018678e-01 -1.26235867e+00 -4.66392875e-01 -5.12810767e-01
-4.19579029e-01 -1.84575543e-01 8.28675449e-01 -6.70010388e-01
1.63466609e+00 -1.88233066e+00 2.05000654e-01 6.98481560e-01
-4.61143881e-01 1.03906989e-02 3.67018998e-01 5.82324564e-01
-2.51333714e-01 1.66397944e-01 -6.10338509e-01 -2.60167867e-01
3.13465536e-01 4.70355392e-01 -4.34289500e-03 2.01042786e-01
-2.72110045e-01 5.46348035e-01 -7.75202274e-01 -3.14729325e-02
8.41772676e-01 5.81646204e-01 -2.32206747e-01 4.09849316e-01
-9.19595435e-02 2.62800992e-01 -3.71323824e-01 4.50388372e-01
4.30253267e-01 -8.52969736e-02 5.15925348e-01 -2.25787595e-01
-2.61857271e-01 1.76220402e-01 -1.37203586e+00 1.45402336e+00
-1.29344141e+00 3.03309172e-01 3.08634490e-01 -6.65954471e-01
5.07605433e-01 5.29985845e-01 7.91195691e-01 -8.33630502e-01
2.30996549e-01 -1.05595596e-01 -4.22027171e-01 -1.38487920e-01
3.86624694e-01 -3.18379812e-02 -1.42325893e-01 2.80999929e-01
-3.58293355e-01 -4.90296274e-01 -1.86122451e-02 -2.95230269e-01
1.00483036e+00 -2.15836063e-01 1.79968715e-01 -6.85263336e-01
5.83244860e-01 -5.69585562e-01 5.18165231e-01 1.26684636e-01
-1.90451756e-01 5.03327064e-02 -1.72148719e-01 -4.21348304e-01
-9.18451488e-01 -9.85523105e-01 -3.22361618e-01 8.67559612e-01
3.14690024e-02 -3.22953969e-01 -1.03201616e+00 -1.88909248e-01
-3.57931614e-01 1.49597347e+00 -1.94911391e-01 -3.82990092e-01
-7.31092513e-01 -1.14022565e+00 -2.20648184e-01 6.72185302e-01
6.71128988e-01 -7.12721944e-01 -1.21312892e+00 1.21371351e-01
-4.44320947e-01 -9.24674392e-01 -4.55348939e-01 7.16115236e-01
-5.15004456e-01 -8.05306554e-01 -2.25302637e-01 -4.27323014e-01
8.04872870e-01 3.28427856e-03 1.14056492e+00 1.12650476e-01
-5.13587415e-01 6.25432789e-01 2.18934357e-01 -2.21806094e-01
-1.67836547e-01 1.85076684e-01 1.66981779e-02 -3.27972412e-01
1.14456583e-02 -4.94077832e-01 -1.10555673e+00 6.21219158e-01
-7.64551878e-01 2.01153129e-01 8.10021907e-02 5.37924767e-01
5.01835227e-01 1.01578653e+00 2.87285477e-01 -1.20230451e-01
4.92126405e-01 -1.89020097e-01 -9.06710744e-01 4.10962820e-01
-1.12278640e+00 -6.22063205e-02 7.40219891e-01 -1.53760359e-01
-1.15748954e+00 3.85748416e-01 -1.28354190e-03 1.44435048e-01
-1.71166942e-01 -3.41454744e-01 -5.38550973e-01 8.59754235e-02
4.66249324e-03 8.71293712e-03 -7.51059771e-01 -2.47208849e-01
2.66262561e-01 3.43497902e-01 3.56937170e-01 -6.09773278e-01
1.18898129e+00 -4.05767336e-02 1.57194644e-01 -7.29477704e-01
-6.79030418e-01 -3.07318121e-01 -4.01925653e-01 -4.57773298e-01
1.11364794e+00 -8.48546803e-01 -1.30420732e+00 1.94934517e-01
-6.42289340e-01 -7.84502387e-01 -3.42158228e-01 6.67473674e-02
-5.64861834e-01 1.44115714e-02 -3.13755751e-01 -1.35334814e+00
-6.62110150e-01 -1.10018456e+00 7.40344465e-01 4.02030915e-01
-3.12595099e-01 -7.62102962e-01 8.80976394e-03 5.07806480e-01
6.26130581e-01 6.02586925e-01 1.09396446e+00 2.71899045e-01
-7.20818162e-01 2.28112340e-02 4.47727442e-01 6.67207122e-01
5.94131112e-01 -1.45246610e-01 -1.30811441e+00 -9.25013721e-01
2.95765609e-01 -8.84771496e-02 4.31327432e-01 4.39697057e-01
1.88138664e+00 -5.35809100e-01 -3.97038192e-01 2.73350328e-01
1.79465079e+00 7.32298315e-01 3.67219687e-01 3.01853865e-01
4.28088695e-01 3.94909441e-01 3.95837843e-01 6.77240968e-01
2.76432306e-01 6.32614374e-01 8.51104081e-01 -5.45045555e-01
7.15908945e-01 3.09699345e-02 1.27688572e-01 7.24352479e-01
-3.92711371e-01 -5.47494352e-01 -6.23800278e-01 5.97742677e-01
-1.59480917e+00 -6.94268763e-01 2.25242987e-01 2.55611467e+00
8.43614340e-01 -1.02793254e-01 -2.94194911e-02 3.78002554e-01
3.65840673e-01 3.38275194e-01 -4.90332723e-01 -6.26522779e-01
5.13080835e-01 2.48313293e-01 7.07827508e-01 3.64258319e-01
-1.03655195e+00 -1.62585482e-01 6.74222898e+00 4.82354999e-01
-6.68309033e-01 -4.03954163e-02 1.12345588e+00 -8.01269233e-01
-9.60276574e-02 -3.24848741e-01 -3.59521955e-01 6.91781163e-01
1.55932879e+00 -6.74727559e-02 1.10363126e+00 1.03590882e+00
5.47847390e-01 -2.50838369e-01 -1.62735438e+00 7.50550091e-01
-4.29314226e-01 -8.58499527e-01 -6.88427627e-01 1.47985116e-01
8.48733723e-01 -2.97061890e-01 -1.73023626e-01 3.50271225e-01
3.48709434e-01 -8.58056664e-01 3.59973907e-01 2.42071986e-01
3.74554724e-01 -1.32421982e+00 8.10924530e-01 4.05460745e-01
-1.52272248e+00 -5.09958088e-01 -5.87148443e-02 1.26553044e-01
3.58765841e-01 5.11270106e-01 -5.26286960e-01 7.74056673e-01
1.14612269e+00 -3.96476001e-01 -3.93664360e-01 5.28312445e-01
-1.90874815e-01 3.25953066e-01 -8.18187833e-01 -8.96412060e-02
1.54595703e-01 -2.97585428e-01 -1.25635669e-01 8.52964103e-01
4.84703988e-01 4.19775277e-01 2.86753178e-01 6.97180867e-01
1.00012563e-01 -2.09035665e-01 -2.99192816e-01 4.03842151e-01
5.05982757e-01 1.51852083e+00 -4.87030715e-01 -1.10388480e-01
-1.25829816e-01 8.28734219e-01 -3.30319703e-01 5.50593674e-01
-1.20943475e+00 -2.15913847e-01 7.62484908e-01 6.24248423e-02
1.96405560e-01 -1.78031281e-01 -1.82832599e-01 -3.52035195e-01
1.77299991e-01 -3.88933450e-01 2.25213706e-01 -8.07115495e-01
-7.42026687e-01 3.70968580e-01 3.18924636e-01 -7.31154799e-01
-1.93804398e-01 -3.76750827e-01 -7.32843697e-01 7.79002488e-01
-1.44853461e+00 -6.13229930e-01 -7.20555067e-01 5.29609025e-01
6.63941741e-01 4.78754848e-01 1.02181602e+00 5.61863601e-01
-1.10066926e+00 2.52796531e-01 4.05192584e-01 -5.46127856e-01
1.94694325e-01 -1.45654726e+00 2.66649812e-01 7.02019930e-01
-6.73063219e-01 3.32286954e-01 9.50019479e-01 -8.62787962e-02
-1.39352429e+00 -1.22318423e+00 5.68921864e-01 -3.14903200e-01
4.02143121e-01 -7.70788133e-01 -6.45610273e-01 5.18560350e-01
7.54099131e-01 -3.36278021e-01 7.29267955e-01 -5.62049598e-02
5.21114230e-01 -4.37039137e-01 -1.29705405e+00 6.92733943e-01
7.79830754e-01 -2.16549993e-01 3.61504848e-03 6.63217425e-01
7.40361929e-01 -4.43961799e-01 -1.26944590e+00 3.99740756e-01
1.52474985e-01 -5.67049801e-01 9.55282032e-01 3.28756683e-02
-1.44563153e-01 -5.54262161e-01 -2.34190017e-01 -1.44260705e+00
-7.55656779e-01 -7.87754953e-01 -3.70074362e-01 1.31379008e+00
3.05278003e-01 -5.49396276e-01 3.74794573e-01 1.39451814e+00
-3.22565794e-01 -1.09366000e+00 -1.26548886e+00 -6.97713673e-01
-3.14718217e-01 -2.16582060e-01 8.90979230e-01 7.58228183e-01
-2.27851793e-01 1.93923652e-01 -2.28973716e-01 6.35253489e-01
3.78579646e-01 3.83051187e-02 3.08498114e-01 -8.93698931e-01
-2.75020152e-01 -1.23722292e-01 3.42214525e-01 -5.42055368e-01
-2.52188981e-01 -2.93318242e-01 4.63019013e-01 -1.82663918e+00
5.68997972e-02 -2.44613588e-01 -8.46568227e-01 8.15839827e-01
5.02745397e-02 -2.86829293e-01 9.77047905e-02 -4.52342033e-01
-4.57684174e-02 9.99842405e-01 8.18914711e-01 -4.64800268e-01
-3.80524904e-01 2.25646123e-01 -3.10769051e-01 5.04922926e-01
9.45095003e-01 -1.91109508e-01 -8.49597812e-01 -2.47151315e-01
1.11489907e-01 -1.89415738e-01 4.20311719e-01 -1.15435183e+00
-2.33950213e-01 -6.59951508e-01 4.71388966e-01 -7.31741846e-01
3.76882017e-01 -1.80665600e+00 8.46092224e-01 6.38712645e-01
-4.62930091e-02 1.98602080e-01 4.23024058e-01 4.42610264e-01
4.05614644e-01 8.45831335e-02 9.07736063e-01 -4.82411571e-02
-3.31415504e-01 -6.54428676e-02 -5.74100673e-01 -6.52503788e-01
1.47720075e+00 -7.23028481e-02 -2.88791656e-01 -1.40691161e-01
-5.94782054e-01 8.26819539e-01 4.53595191e-01 3.74931157e-01
4.44507003e-02 -1.43543339e+00 3.48288129e-04 -4.38089296e-02
-2.55032778e-01 3.19119215e-01 7.22128212e-01 4.11700994e-01
-2.39391446e-01 6.59733415e-01 1.06738307e-01 -4.46054935e-01
-1.30300426e+00 8.66948307e-01 8.08410227e-01 -3.83553177e-01
-5.40505171e-01 2.65831023e-01 1.26251861e-01 -1.98405802e-01
1.53012216e-01 -9.07413602e-01 3.10644418e-01 -1.68276355e-01
1.05128750e-01 8.87097955e-01 3.63426268e-01 3.91919911e-03
-5.33209682e-01 2.88799286e-01 4.96224076e-01 3.60203147e-01
1.33060741e+00 -1.65486708e-01 2.78776959e-02 5.03128827e-01
9.76096928e-01 -3.97699893e-01 -1.33820200e+00 4.72284138e-01
-2.95894980e-01 -6.68873265e-02 5.14495850e-01 -1.17433548e+00
-1.16459084e+00 2.77805954e-01 1.18540192e+00 6.29981220e-01
2.04798603e+00 -2.89895266e-01 8.58688593e-01 5.46510160e-01
3.33719879e-01 -1.95556962e+00 -3.53703141e-01 -1.69476569e-01
8.44656348e-01 -8.23913813e-01 3.37400854e-01 -1.71126410e-01
-1.92920372e-01 8.41002405e-01 6.83120668e-01 4.58698481e-01
6.26866043e-01 7.32258260e-02 -2.88141340e-01 1.45932436e-01
-6.80655181e-01 3.76287371e-01 1.93499345e-02 2.00263739e-01
2.25655168e-01 4.46500391e-01 -5.57421520e-02 5.40483706e-02
-1.46688744e-01 -3.53397310e-01 2.12340176e-01 1.25324988e+00
-4.57318544e-01 -7.38198698e-01 -3.75192553e-01 1.11918129e-01
-5.36811911e-02 2.04929575e-01 1.71987265e-01 8.96065056e-01
2.18556166e-01 1.30911744e+00 1.22913934e-01 -1.07138157e-01
9.03910339e-01 3.02431226e-01 1.74402911e-02 -2.45584697e-01
-7.61059225e-01 1.91444471e-01 1.87102705e-01 -7.50530481e-01
-3.76444578e-01 -3.85965914e-01 -1.20875096e+00 -4.26364124e-01
-4.06338274e-01 4.75033000e-02 8.60519588e-01 8.01817119e-01
3.08802605e-01 1.56710851e+00 1.16186202e+00 -7.41705358e-01
-4.01472807e-01 -8.26154411e-01 -6.15373909e-01 -1.02896497e-01
4.49201792e-01 -6.52542412e-01 -4.78912145e-01 -2.44554564e-01] | [5.628588676452637, 2.447730302810669] |
6813cfa6-83df-47b1-a291-8b614b321ab7 | towards-pose-invariant-face-recognition-in | null | null | http://openaccess.thecvf.com/content_cvpr_2018/html/Zhao_Towards_Pose_Invariant_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Zhao_Towards_Pose_Invariant_CVPR_2018_paper.pdf | Towards Pose Invariant Face Recognition in the Wild | Pose variation is one key challenge in face recognition. As opposed to current techniques for pose invariant face recognition, which either directly extract pose invariant features for recognition, or first normalize profile face images to frontal pose before feature extraction, we argue that it is more desirable to perform both tasks jointly to allow them to benefit from each other. To this end, we propose a Pose Invariant Model (PIM) for face recognition in the wild, with three distinct novelties. First, PIM is a novel and unified deep architecture, containing a Face Frontalization sub-Net (FFN) and a Discriminative Learning sub-Net (DLN), which are jointly learned from end to end. Second, FFN is a well-designed dual-path Generative Adversarial Network (GAN) which simultaneously perceives global structures and local details, incorporated with an unsupervised cross-domain adversarial training and a "learning to learn" strategy for high-fidelity and identity-preserving frontal view synthesis. Third, DLN is a generic Convolutional Neural Network (CNN) for face recognition with our enforced cross-entropy optimization strategy for learning discriminative yet generalized feature representation. Qualitative and quantitative experiments on both controlled and in-the-wild benchmarks demonstrate the superiority of the proposed model over the state-of-the-arts. | ['ShengMei Shen', 'Sugiri Pranata', 'Yan Xu', 'Lin Xiong', 'Junliang Xing', 'Jian Zhao', 'Yu Cheng', 'Jianshu Li', 'Fang Zhao', 'Karlekar Jayashree', 'Jiashi Feng', 'Shuicheng Yan'] | 2018-06-01 | null | null | null | cvpr-2018-6 | ['robust-face-recognition'] | ['computer-vision'] | [ 4.02591079e-01 1.46438539e-01 2.56855208e-02 -7.52207339e-01
-7.72588015e-01 -5.76487303e-01 7.35133708e-01 -9.62663889e-01
6.98617473e-02 4.87719864e-01 3.33313406e-01 1.07853949e-01
-4.14967462e-02 -7.23270237e-01 -8.63919795e-01 -9.24881339e-01
8.71725455e-02 3.42621744e-01 -5.94712913e-01 -1.50639504e-01
-1.44735530e-01 1.30417991e+00 -1.43237209e+00 1.13313183e-01
3.70880544e-01 1.36917460e+00 -4.32001531e-01 3.04697543e-01
3.91977131e-01 4.81738269e-01 -3.67296815e-01 -7.54710317e-01
6.63729727e-01 -3.42755109e-01 -6.65342093e-01 2.57520854e-01
9.52387989e-01 -4.58770752e-01 -8.01888466e-01 9.87293363e-01
8.23880553e-01 1.32418051e-03 7.22675443e-01 -1.33674681e+00
-7.63813615e-01 -3.18442993e-02 -4.65523213e-01 -1.51919588e-01
4.70028609e-01 4.13655668e-01 6.38183177e-01 -1.11096191e+00
6.78598762e-01 1.53301525e+00 7.16123223e-01 9.88730192e-01
-1.31040919e+00 -8.32509160e-01 1.25029355e-01 -2.47048676e-01
-1.45592749e+00 -9.21445072e-01 1.14683104e+00 -4.46916163e-01
6.13731444e-01 3.42393637e-01 4.40207869e-01 1.60273659e+00
1.82176962e-01 4.52098519e-01 1.07040048e+00 -1.04341954e-01
-4.72352654e-02 -2.64824659e-01 -5.89699268e-01 8.60685647e-01
4.89853881e-02 5.75975955e-01 -6.96820796e-01 -6.02343343e-02
9.45433736e-01 4.44073200e-01 -4.68948126e-01 -5.75489104e-01
-6.67769134e-01 7.23869324e-01 4.65553433e-01 -1.07947901e-01
-4.41930115e-01 1.52547985e-01 3.91189098e-01 2.88040340e-01
5.46738863e-01 8.72170851e-02 -3.93338591e-01 2.57540345e-01
-1.05073786e+00 4.11262840e-01 7.40719378e-01 8.10792744e-01
7.39902139e-01 4.06963974e-01 -5.42255700e-01 6.81186020e-01
5.94572246e-01 6.23547971e-01 2.44485438e-01 -7.56147265e-01
2.24469900e-01 5.95558286e-01 -2.93425977e-01 -1.13816214e+00
-1.29322886e-01 -5.76661110e-01 -9.65785086e-01 3.19129646e-01
1.53527990e-01 -8.97149295e-02 -1.14198947e+00 2.20735884e+00
3.13783705e-01 2.23197088e-01 1.79518033e-02 7.67086983e-01
8.34187984e-01 2.17283264e-01 3.32351448e-03 -1.38288379e-01
1.33860958e+00 -8.50088418e-01 -5.47214329e-01 -1.94714457e-01
-2.12939993e-01 -6.94048464e-01 7.95727372e-01 6.06577657e-02
-1.04498279e+00 -5.67631900e-01 -1.03937280e+00 -5.62982298e-02
-4.33931023e-01 -9.08232480e-02 5.10214806e-01 8.31206083e-01
-1.16710329e+00 5.84983528e-01 -7.13338852e-01 -9.01294351e-02
9.23688173e-01 6.44344687e-01 -7.58656263e-01 -2.15700150e-01
-8.64452660e-01 5.97646832e-01 -4.00897712e-02 4.58498508e-01
-1.40368021e+00 -8.97679925e-01 -1.12802815e+00 1.56125695e-01
2.04782322e-01 -8.70662391e-01 8.72667968e-01 -1.17579257e+00
-1.87510228e+00 1.00717926e+00 -1.83823586e-01 -1.29802711e-02
5.33345699e-01 -2.05212981e-01 -4.13649857e-01 1.85702965e-01
-1.07234761e-01 6.21784151e-01 1.37627852e+00 -1.47893012e+00
5.35985008e-02 -8.16461980e-01 5.26295276e-03 -2.51048040e-02
-4.82629657e-01 1.53582528e-01 -5.75994790e-01 -9.31203365e-01
-1.29235968e-01 -8.01599085e-01 1.04250111e-01 2.23613054e-01
-3.78825366e-01 -1.01007866e-02 1.17922103e+00 -9.62457716e-01
7.62797773e-01 -2.07831216e+00 5.17959893e-01 2.68085957e-01
2.68886298e-01 3.62194210e-01 -4.40391749e-01 1.37340486e-01
-3.98170382e-01 -9.25663561e-02 -1.92734942e-01 -7.59071887e-01
2.28668541e-01 2.05823392e-01 -3.26183915e-01 7.91052878e-01
5.93562543e-01 1.21498704e+00 -5.74278951e-01 -1.71862736e-01
1.39995173e-01 1.07500780e+00 -7.88666964e-01 4.41341102e-01
-1.01936340e-01 7.29891300e-01 -2.97509730e-01 1.08240473e+00
1.08188462e+00 6.25568703e-02 5.24142794e-02 -5.57689428e-01
3.45936596e-01 -1.24617279e-01 -9.55119371e-01 1.64730251e+00
-4.34229255e-01 2.64440119e-01 3.70030284e-01 -9.47982848e-01
8.84000242e-01 5.19174039e-01 4.96327758e-01 -5.84108293e-01
2.36473814e-01 1.55390024e-01 -4.70919549e-01 -2.06557989e-01
-5.09834215e-02 -2.41200700e-01 1.12101726e-01 2.85111535e-02
5.91056645e-01 6.38152957e-02 -3.43982697e-01 -3.02872300e-01
9.56803679e-01 3.58020633e-01 1.01699434e-01 -3.03058296e-01
8.18358004e-01 -9.56149280e-01 8.10148120e-01 9.30236280e-02
-2.40453690e-01 7.52514541e-01 4.11313802e-01 -5.51022470e-01
-7.71992564e-01 -1.22630465e+00 -1.70203999e-01 9.76640165e-01
-1.27525270e-01 -1.37041613e-01 -9.07124043e-01 -1.03678477e+00
9.88553017e-02 1.36765568e-02 -9.35139358e-01 -2.95490265e-01
-6.33021057e-01 -3.02543670e-01 6.61673546e-01 6.56806588e-01
6.71950996e-01 -1.01740015e+00 -1.80261031e-01 -1.47478938e-01
3.17447126e-01 -1.11630201e+00 -9.10624981e-01 4.28269431e-02
-4.40012187e-01 -9.77470815e-01 -6.29941106e-01 -7.07290649e-01
8.51252735e-01 -1.54448763e-01 1.10823655e+00 -1.07526578e-01
-3.37426484e-01 5.57039320e-01 -1.41998917e-01 -2.60877341e-01
-8.75262097e-02 -6.72124848e-02 2.29823336e-01 5.66378772e-01
1.49822295e-01 -9.51981008e-01 -9.15666640e-01 3.13594967e-01
-9.63406026e-01 -2.70684898e-01 6.19409502e-01 1.23610961e+00
6.24435842e-01 -4.34387565e-01 4.92714554e-01 -6.73121274e-01
2.18566984e-01 -1.98549569e-01 -4.72381324e-01 2.51998037e-01
-3.63865256e-01 -3.67069542e-02 7.65540957e-01 -2.13393271e-01
-1.00908756e+00 4.17861640e-01 -4.27391380e-01 -1.00834465e+00
-1.77536279e-01 4.99683991e-02 -1.06739187e+00 -6.33766174e-01
4.58428800e-01 4.76271480e-01 3.52364093e-01 -4.05156076e-01
3.00449610e-01 3.39343011e-01 6.96714044e-01 -6.56342328e-01
1.30314445e+00 5.59609711e-01 3.09180051e-01 -6.83565378e-01
-7.69356668e-01 6.83830678e-02 -7.94698477e-01 -1.71149924e-01
8.71055841e-01 -9.84758914e-01 -1.00591457e+00 5.59068263e-01
-1.11078560e+00 -6.48824200e-02 -2.86186039e-01 -1.59863830e-02
-6.42303228e-01 8.93875435e-02 -4.33431000e-01 -5.84796846e-01
-5.57468355e-01 -1.21342397e+00 1.43729830e+00 4.81426239e-01
2.03148454e-01 -1.04220831e+00 -6.48173466e-02 3.96257669e-01
5.12525976e-01 8.11391234e-01 4.02638048e-01 -5.68458021e-01
-5.15272081e-01 -1.99678630e-01 -2.18770534e-01 7.36212432e-01
3.05445552e-01 -1.85858831e-01 -1.43341541e+00 -8.82889271e-01
1.04493640e-01 -4.82713670e-01 7.94835150e-01 -1.37566850e-02
1.39566827e+00 -6.90728009e-01 -2.90374041e-01 1.39486074e+00
1.35808229e+00 4.82066013e-02 7.73842573e-01 -2.67471462e-01
9.77477193e-01 4.60404038e-01 3.83013599e-02 3.47521752e-01
1.56994626e-01 9.14117992e-01 6.30678356e-01 -2.63411582e-01
-2.51748562e-01 -4.70915139e-01 6.06188476e-01 3.32952440e-01
-1.35324717e-01 2.60467269e-02 -5.25099218e-01 1.28808245e-01
-1.40651643e+00 -1.01124835e+00 9.56813335e-01 1.98674333e+00
6.88253939e-01 -3.21431041e-01 2.60547902e-02 4.25084680e-03
3.56834114e-01 4.15866107e-01 -7.22811759e-01 -2.49729812e-01
-4.70878817e-02 6.70100152e-01 2.51508147e-01 3.95501494e-01
-1.24007368e+00 8.11303318e-01 5.37393665e+00 8.94686282e-01
-1.55561626e+00 8.83812904e-02 8.05382609e-01 6.85653612e-02
-2.19683230e-01 -3.93983841e-01 -8.10453176e-01 3.09308171e-01
6.35169566e-01 7.84453973e-02 6.32109523e-01 9.58151400e-01
-2.03384116e-01 8.07147145e-01 -1.41230845e+00 1.05911994e+00
5.25384724e-01 -1.26316965e+00 4.58706260e-01 2.76206404e-01
6.76861167e-01 -4.13108379e-01 4.87460822e-01 2.36542732e-01
-8.32184553e-02 -1.65927815e+00 6.16135061e-01 7.67508209e-01
1.18298829e+00 -1.01370108e+00 4.52143788e-01 -1.52489260e-01
-1.44254220e+00 -3.84255080e-03 -1.65610313e-01 3.66960943e-01
-1.86611891e-01 2.63089150e-01 -2.61852622e-01 9.88616109e-01
4.84909117e-01 7.51343369e-01 -4.92894471e-01 4.40067410e-01
-7.58975968e-02 2.39671692e-01 -8.16424489e-02 6.33329511e-01
-6.16996409e-03 -4.45367508e-02 5.55034280e-01 1.12900102e+00
1.36902019e-01 -1.39975071e-01 1.90645561e-01 1.06078911e+00
-5.60949862e-01 -1.56579375e-01 -8.16069841e-01 -1.24444410e-01
3.86381000e-01 1.46354246e+00 -2.87902445e-01 8.64766613e-02
-4.11199689e-01 1.24769175e+00 3.29007417e-01 3.50151807e-01
-7.19411850e-01 -2.63555467e-01 1.17690802e+00 1.56649172e-01
4.57310826e-01 -1.59201231e-02 -4.61835563e-02 -1.15290511e+00
1.82180420e-01 -1.16760218e+00 1.97469145e-01 -1.13849774e-01
-1.52640438e+00 9.02495086e-01 -2.39765808e-01 -1.09678495e+00
-2.82040268e-01 -9.74622667e-01 -6.41669035e-01 1.15989566e+00
-1.68300557e+00 -1.89850414e+00 -3.84162754e-01 1.01520717e+00
4.16974992e-01 -5.90251923e-01 1.00417328e+00 5.17811298e-01
-6.94454491e-01 1.37549233e+00 -2.30707288e-01 4.99199957e-01
6.47822320e-01 -8.63556504e-01 3.11413080e-01 9.21130717e-01
3.13521214e-02 8.42976928e-01 2.74958372e-01 -4.05002713e-01
-1.96398592e+00 -1.57567561e+00 4.72047538e-01 -4.94119376e-01
-3.96174705e-03 -5.47691643e-01 -6.78517878e-01 7.69104779e-01
5.55243604e-02 7.37480462e-01 6.05714321e-01 -2.97405064e-01
-7.39644051e-01 -4.38709736e-01 -1.47072685e+00 5.37884951e-01
1.36817932e+00 -9.19009387e-01 -2.41869688e-01 2.89472133e-01
5.13873816e-01 -5.27590930e-01 -1.05757570e+00 8.15589845e-01
8.04212034e-01 -9.90374267e-01 1.32398570e+00 -5.26828468e-01
4.58718210e-01 -1.45146593e-01 -5.17195940e-01 -1.23524177e+00
-2.21086711e-01 -9.78227437e-01 -2.74821818e-01 1.46607971e+00
4.26478945e-02 -7.08290339e-01 9.33404982e-01 2.56087571e-01
-1.84002563e-01 -9.46951687e-01 -1.19819057e+00 -9.22275126e-01
1.75527245e-01 2.27489844e-02 8.86413574e-01 8.06730211e-01
-6.20442688e-01 2.70533532e-01 -4.43631798e-01 3.69460225e-01
7.01752424e-01 -4.30903211e-02 7.08726943e-01 -1.19478905e+00
-2.63983130e-01 -4.52666819e-01 -6.20475709e-01 -9.03623581e-01
6.66777432e-01 -9.41332400e-01 -9.27528590e-02 -7.94461727e-01
3.29585344e-01 2.46138144e-02 -2.31213406e-01 5.14805675e-01
5.29265217e-03 4.64045405e-01 1.65818736e-01 -9.68189463e-02
-1.53524339e-01 9.62920964e-01 1.54539073e+00 -1.51190087e-01
1.92737371e-01 -1.59033820e-01 -8.06414306e-01 5.68602920e-01
2.99471289e-01 -8.49008113e-02 -4.00128931e-01 -1.45014554e-01
-3.04710001e-01 5.53269275e-02 7.16981173e-01 -9.46501017e-01
6.19662814e-02 -1.33978888e-01 9.31970298e-01 -1.35099486e-01
5.78490376e-01 -9.15311158e-01 1.39327019e-01 3.70272547e-01
-8.04160163e-02 2.06600278e-04 3.11309516e-01 4.34756547e-01
-3.09839606e-01 3.92154068e-01 1.09426844e+00 1.30258799e-01
-3.16108227e-01 1.05948031e+00 4.40819949e-01 7.03664720e-02
9.31818485e-01 -2.35068843e-01 -5.80651127e-02 -2.01526612e-01
-6.06238961e-01 -2.00572997e-01 4.81377870e-01 5.47244251e-01
7.47914314e-01 -1.70007336e+00 -7.78292000e-01 8.32187235e-01
4.63004597e-02 1.48025341e-02 3.31376582e-01 5.47528088e-01
-2.66933262e-01 3.06243479e-01 -5.21184504e-01 -3.95460725e-01
-1.09558117e+00 5.77058911e-01 7.38085091e-01 -3.45734596e-01
-3.14283878e-01 1.03183818e+00 5.24380982e-01 -7.06674755e-01
2.95619637e-01 2.00334176e-01 -4.39223126e-02 -5.01464307e-02
5.14231205e-01 1.67558454e-02 1.16263479e-01 -9.36490119e-01
-4.93926167e-01 6.96197510e-01 -1.99077744e-02 6.12443611e-02
1.49361491e+00 2.10561156e-01 -2.44191945e-01 -2.06088394e-01
1.55204308e+00 -7.39862211e-03 -1.68265188e+00 -2.44537324e-01
-4.30406690e-01 -6.46059513e-01 -8.62327218e-02 -6.95432782e-01
-1.56698298e+00 7.37019658e-01 7.81238377e-01 -4.34404075e-01
1.38342750e+00 -2.13053271e-01 5.62635481e-01 6.22256957e-02
2.00650260e-01 -7.14230776e-01 4.04950827e-01 4.59937513e-01
1.55351198e+00 -1.09890211e+00 -1.87984303e-01 -4.86202985e-01
-1.85555696e-01 1.05198872e+00 6.95204377e-01 -1.70915708e-01
9.85821187e-01 3.38939339e-01 -1.12877309e-01 -2.78785080e-01
-4.94834751e-01 1.12137370e-01 7.33354986e-01 7.52444446e-01
3.69300753e-01 -2.15908121e-02 2.47733310e-01 8.11967313e-01
-2.65181661e-01 -9.16579962e-02 -3.70183736e-01 8.04114521e-01
2.92571783e-01 -1.23820674e+00 -2.79781848e-01 2.49647811e-01
-4.01664734e-01 8.58371854e-02 -4.35655594e-01 7.41774619e-01
3.75021607e-01 4.80286658e-01 -8.90422612e-02 -6.66454256e-01
4.63255018e-01 1.87387824e-01 8.46670508e-01 -5.81090391e-01
-6.46659732e-01 -1.51361048e-01 -3.97892952e-01 -8.81514907e-01
-9.09316316e-02 -6.41027391e-01 -6.14132702e-01 -3.95346314e-01
1.49086267e-01 -3.69350106e-01 4.72596556e-01 8.87805223e-01
4.96907353e-01 6.58875823e-01 1.10317957e+00 -1.37463641e+00
-8.21505964e-01 -8.97023320e-01 -5.10383725e-01 5.81677139e-01
4.66867357e-01 -9.14742172e-01 -3.36604565e-01 6.40940741e-02] | [13.056832313537598, 0.32404589653015137] |
849e6f98-6e87-4008-a5bb-0c10df5a8552 | a-multi-dimensional-cross-domain-and | null | null | https://aclanthology.org/2022.coling-1.45 | https://aclanthology.org/2022.coling-1.45.pdf | A Multi-Dimensional, Cross-Domain and Hierarchy-Aware Neural Architecture for ISO-Standard Dialogue Act Tagging | Dialogue Act tagging with the ISO 24617-2 standard is a difficult task that involves multi-label text classification across a diverse set of labels covering semantic, syntactic and pragmatic aspects of dialogue. The lack of an adequately sized training set annotated with this standard is a major problem when using the standard in practice. In this work we propose a neural architecture to increase classification accuracy, especially on low-frequency fine-grained tags. Our model takes advantage of the hierarchical structure of the ISO taxonomy and utilises syntactic information in the form of Part-Of-Speech and dependency tags, in addition to contextual information from previous turns. We train our architecture on an aggregated corpus of conversations from different domains, which provides a variety of dialogue interactions and linguistic registers. Our approach achieves state-of-the-art tagging results on the DialogBank benchmark data set, providing empirical evidence that this architecture can successfully generalise to different domains. | ['Alan Blair', 'Wayne Wobcke', 'Stefano Mezza'] | null | null | null | null | coling-2022-10 | ['multi-label-text-classification', 'multi-label-text-classification'] | ['methodology', 'natural-language-processing'] | [ 7.62723237e-02 4.49791729e-01 -1.05733119e-01 -6.16136372e-01
-7.30315447e-01 -9.40459728e-01 9.25087094e-01 2.49293655e-01
-7.37208128e-01 1.06281376e+00 8.79530966e-01 -2.38901585e-01
4.35098857e-02 -3.36150795e-01 1.86391518e-01 -1.96251288e-01
1.95486601e-02 9.36816931e-01 3.37185532e-01 -8.20220470e-01
-2.95522232e-02 6.37596622e-02 -1.03724849e+00 6.62603617e-01
4.44278955e-01 8.31972003e-01 -5.67073822e-02 6.39465451e-01
-5.72971404e-01 1.11662710e+00 -1.08938205e+00 -5.79705656e-01
-6.69416860e-02 -4.89832669e-01 -1.59499550e+00 7.84735680e-02
2.10970595e-01 -9.93611515e-02 1.83385804e-01 6.58854365e-01
4.49932337e-01 2.70378739e-01 5.07984042e-01 -8.25993121e-01
-6.86492026e-02 8.00457239e-01 7.85110742e-02 9.66137350e-02
6.68345869e-01 -1.68400526e-01 1.37888265e+00 -2.31450096e-01
7.41999090e-01 1.38316250e+00 9.22109723e-01 9.04544294e-01
-1.18608201e+00 -2.54944295e-01 -2.07460560e-02 -4.21706170e-01
-7.07169652e-01 -5.21959901e-01 4.02215362e-01 -4.90637451e-01
1.43207645e+00 1.74164400e-01 3.03092122e-01 1.23179138e+00
4.61330265e-02 5.62252939e-01 1.32832861e+00 -7.90270805e-01
5.06946445e-02 2.65237331e-01 4.06537265e-01 4.69876647e-01
-4.14344907e-01 -2.34694526e-01 -4.48563725e-01 -3.72165918e-01
3.22837919e-01 -5.10692060e-01 -8.95188823e-02 -3.97607945e-02
-9.52789783e-01 1.04376042e+00 1.24180429e-02 9.34376538e-01
-9.43463072e-02 -2.49500498e-01 1.08620000e+00 4.32788074e-01
7.45556116e-01 9.62772489e-01 -1.09139287e+00 -6.57895327e-01
-3.93797725e-01 1.73185304e-01 1.58411086e+00 6.25899255e-01
5.65970898e-01 -2.76966572e-01 -1.44803211e-01 1.47754896e+00
3.24522942e-01 -7.08069503e-02 7.02735186e-01 -1.28662574e+00
6.27395213e-01 7.30452061e-01 2.07777098e-01 -3.52203608e-01
-8.95979583e-01 8.19452405e-02 -5.31575561e-01 5.35595194e-02
8.64166915e-01 -5.97299695e-01 -3.80428374e-01 1.88626742e+00
3.04674208e-01 -4.70894933e-01 4.91984218e-01 4.08253610e-01
1.11721849e+00 4.52895284e-01 5.70840538e-01 -2.05110669e-01
1.66864872e+00 -8.48818898e-01 -9.04025674e-01 -4.42647398e-01
1.13357687e+00 -7.60078371e-01 8.80341291e-01 1.03914522e-01
-9.12341118e-01 -4.33999687e-01 -7.92473614e-01 -1.19155824e-01
-5.81259012e-01 -5.78800105e-02 7.93014765e-01 8.38009715e-01
-1.21488225e+00 3.95053238e-01 -4.49306190e-01 -5.38348317e-01
-1.29878759e-01 4.42108601e-01 -6.48144066e-01 3.32485825e-01
-1.59417522e+00 1.23032045e+00 6.80819511e-01 -1.75229579e-01
-7.39151686e-02 -2.92805314e-01 -1.16960740e+00 -1.19495638e-01
2.71832407e-01 -3.90158631e-02 1.90258622e+00 -8.70139122e-01
-1.89870441e+00 1.15522301e+00 2.59478748e-01 -4.52328086e-01
1.78384125e-01 -1.86568528e-01 -3.47553939e-01 -1.92301482e-01
6.92686066e-02 6.18825614e-01 8.01669136e-02 -8.32236052e-01
-9.89273012e-01 -1.12260960e-01 6.00941658e-01 4.08132523e-01
-3.21419597e-01 5.26879370e-01 1.06113344e-01 -1.43183753e-01
-6.16313398e-01 -9.63697433e-01 -2.30644494e-01 -8.48922789e-01
-1.16148688e-01 -7.84096897e-01 4.47279811e-01 -7.08521485e-01
1.28706777e+00 -1.86005330e+00 7.49203935e-02 -3.24269861e-01
1.16840795e-01 4.53197569e-01 1.40453249e-01 8.09626758e-01
1.28934786e-01 1.90826505e-02 1.50261000e-02 -4.89424378e-01
3.24645489e-01 5.16973734e-01 1.03083923e-01 1.98506653e-01
6.43065050e-02 6.14510715e-01 -9.29957688e-01 -5.15129864e-01
3.29274297e-01 2.27481857e-01 -3.32721174e-01 4.71219122e-01
-3.84174943e-01 5.61424553e-01 -4.14182603e-01 7.81202987e-02
-3.04488055e-02 -1.84203446e-01 7.00503945e-01 2.04105880e-02
-2.19264463e-01 1.16937244e+00 -6.67474270e-01 1.99682713e+00
-8.72611642e-01 5.16922712e-01 3.13347220e-01 -8.14885676e-01
8.78854811e-01 9.20713365e-01 4.98154432e-01 -3.43523681e-01
3.72075766e-01 2.38785282e-01 2.51678348e-01 -3.31661016e-01
5.89506626e-01 -1.72766715e-01 -7.90188849e-01 7.06097186e-01
6.31467223e-01 3.32797058e-02 5.10792077e-01 -1.65894683e-02
1.25747991e+00 6.03472255e-02 7.99060881e-01 -4.03472602e-01
7.19721377e-01 1.49827674e-01 4.27743375e-01 4.18316513e-01
-3.45479339e-01 -4.99942303e-02 5.60650885e-01 -5.70589840e-01
-8.64241004e-01 -3.07802469e-01 -3.68491352e-01 1.84250426e+00
-5.43447554e-01 -5.59606791e-01 -7.66397178e-01 -1.31839108e+00
-3.38570744e-01 5.47056437e-01 -6.77352190e-01 2.97456235e-01
-8.30697238e-01 -6.37637258e-01 8.80606353e-01 2.38741755e-01
5.26186466e-01 -1.52890694e+00 -5.19396067e-01 6.29863262e-01
-2.96086103e-01 -1.30852747e+00 -2.06753746e-01 7.25857317e-01
-3.36643249e-01 -1.14032638e+00 -2.46405110e-01 -8.43970299e-01
4.48279874e-03 -3.89450967e-01 1.68444395e+00 -1.04413731e-02
2.61526972e-01 2.53174573e-01 -7.64058352e-01 -2.30235696e-01
-1.21386278e+00 6.18138194e-01 -3.20265830e-01 -3.27987432e-01
7.66548097e-01 -1.15708530e-01 1.72448263e-01 4.24179822e-01
-5.50131619e-01 -2.04332873e-01 1.55467778e-01 1.17486608e+00
-4.21668708e-01 -2.31969222e-01 7.52873838e-01 -1.25634789e+00
9.50355589e-01 -2.58926600e-01 -3.24563205e-01 1.25700891e-01
-1.04510643e-01 1.00027382e-01 5.47574580e-01 -7.85022601e-02
-1.32253933e+00 3.73826623e-02 -7.05945551e-01 6.82478309e-01
-7.77336657e-01 4.98395681e-01 -3.61726671e-01 3.99495810e-02
6.78833485e-01 -4.38327521e-01 -1.69730503e-02 -7.26124883e-01
4.08772081e-01 1.23095536e+00 2.76514530e-01 -7.70141363e-01
-3.05436924e-02 -1.38798982e-01 -1.62489563e-01 -8.23003888e-01
-1.31370807e+00 -8.73780608e-01 -1.08223891e+00 -1.33506998e-01
1.10158479e+00 -7.39231765e-01 -8.63952637e-01 3.96493077e-01
-1.20203662e+00 -7.14116096e-01 -1.60968527e-01 1.90367907e-01
-5.17914355e-01 3.21482986e-01 -1.25265646e+00 -8.28818321e-01
-3.09364110e-01 -9.04779136e-01 9.83400404e-01 1.30537003e-01
-9.37301576e-01 -1.59883130e+00 3.68265063e-01 5.93049347e-01
3.15406799e-01 7.81733319e-02 7.25826859e-01 -1.50567412e+00
3.90075654e-01 -8.22625160e-02 1.65771395e-01 3.96788806e-01
3.05443913e-01 -5.13682842e-01 -1.08430064e+00 -2.39231378e-01
5.79559840e-02 -1.04540813e+00 4.89102453e-01 -3.64931114e-02
2.34933510e-01 -3.70972663e-01 -1.16998486e-01 -1.73718184e-01
1.01549578e+00 4.03943866e-01 2.94858664e-01 6.52781308e-01
4.40743476e-01 1.12093019e+00 6.22765601e-01 3.65056336e-01
6.00323677e-01 9.62593198e-01 8.62426981e-02 3.33218537e-02
9.93572325e-02 2.52044082e-01 1.97471768e-01 9.08834517e-01
1.56855315e-01 -2.93835670e-01 -1.14999652e+00 5.48069954e-01
-1.78195071e+00 -7.93050349e-01 -2.70140134e-02 1.77249920e+00
1.47716999e+00 1.89701796e-01 4.27050859e-01 3.25246863e-02
6.30418420e-01 3.33726585e-01 2.36565813e-01 -8.60248804e-01
1.26395047e-01 2.91623950e-01 2.76208997e-01 7.17650473e-01
-1.58544564e+00 1.12905633e+00 6.38950443e+00 6.28938019e-01
-6.41454816e-01 4.01576698e-01 4.13812131e-01 4.43677634e-01
3.01494569e-01 -2.19612777e-01 -1.13573194e+00 2.82802075e-01
1.56198895e+00 3.66124243e-01 2.26573631e-01 6.30688310e-01
-2.81861633e-01 -7.02970475e-02 -1.05732632e+00 3.47364694e-01
-3.20227407e-02 -1.22815776e+00 -6.76139772e-01 1.09088689e-01
4.85899836e-01 1.98895961e-01 -5.78695118e-01 6.53108358e-01
1.06207693e+00 -9.50503647e-01 3.79984289e-01 -8.39973018e-02
6.04316890e-01 -4.33098614e-01 1.26940835e+00 3.11234951e-01
-8.98325801e-01 8.45705420e-02 5.88637888e-02 -1.99124143e-01
2.83765525e-01 6.46992493e-03 -1.20502710e+00 4.91465211e-01
4.24077243e-01 6.43937588e-01 -2.87563354e-01 5.45774877e-01
-2.66927302e-01 5.87359071e-01 -2.57004559e-01 -2.08056718e-01
5.17189741e-01 5.29390126e-02 1.84711039e-01 1.68721259e+00
-3.05821478e-01 -6.82210326e-02 7.12083161e-01 -5.24452627e-02
-7.57695436e-02 2.46266395e-01 -4.43177521e-01 -3.57858390e-02
4.80737507e-01 1.40164304e+00 -4.68132854e-01 -2.91315734e-01
-7.65311420e-01 6.24016166e-01 4.89204943e-01 -1.29269719e-01
-2.26544559e-01 -1.67864114e-01 6.57258749e-01 -3.86460006e-01
5.46590127e-02 -1.24487802e-01 -5.44383191e-02 -8.04160774e-01
-3.49133223e-01 -1.01228178e+00 7.66745269e-01 -2.19787970e-01
-1.49646044e+00 9.96492624e-01 7.14217797e-02 -8.72069061e-01
-9.24113512e-01 -8.88187230e-01 -2.61334509e-01 1.02086306e+00
-1.25195694e+00 -1.32587755e+00 1.77571028e-02 2.24049956e-01
7.48460114e-01 -3.91278356e-01 1.56334603e+00 2.33839184e-01
-2.18255907e-01 4.07751262e-01 -5.91558702e-02 7.32194901e-01
1.05614543e+00 -1.66139972e+00 4.51801687e-01 4.62250896e-02
1.16838580e-02 4.23014820e-01 6.35099411e-01 -4.09373194e-01
-4.61294562e-01 -8.13720882e-01 1.39668095e+00 -7.67303646e-01
9.67476487e-01 -4.03628290e-01 -9.72187459e-01 7.47293532e-01
8.20068002e-01 -3.96167487e-01 1.21918368e+00 7.77678192e-01
-3.75870943e-01 3.83085877e-01 -1.12218857e+00 1.17161952e-01
7.48870492e-01 -6.37912631e-01 -1.07942200e+00 2.93218672e-01
5.16709805e-01 -6.60581708e-01 -1.37729192e+00 1.18827283e-01
4.13644493e-01 -9.28055406e-01 5.52530468e-01 -7.42639005e-01
2.08549514e-01 2.29349375e-01 -1.08591080e-01 -1.55935192e+00
-8.38867426e-02 -8.39476883e-01 3.28854233e-01 1.62151182e+00
6.11146569e-01 -5.13554931e-01 5.61705112e-01 6.78817272e-01
-4.19749916e-01 -3.14536631e-01 -1.16657448e+00 -5.02072573e-01
3.51872951e-01 -4.87006530e-02 4.45038408e-01 1.23981452e+00
7.14671791e-01 1.15075123e+00 -4.46590692e-01 -5.87052822e-01
5.81563003e-02 -1.66640252e-01 6.60778403e-01 -1.86321199e+00
-3.03898036e-01 -5.02160251e-01 -2.64237672e-01 -9.64190006e-01
7.50561297e-01 -7.03702450e-01 2.77494371e-01 -1.51972127e+00
-2.78476357e-01 -6.87794328e-01 4.84371819e-02 8.43291104e-01
-9.67865810e-03 8.81371200e-02 1.31367788e-01 1.73684895e-01
-9.59424973e-01 3.38077605e-01 7.31322348e-01 1.89416502e-02
-2.17960685e-01 -1.26875520e-01 -4.77970570e-01 7.52852082e-01
9.08723414e-01 -3.98633093e-01 -1.91850252e-02 -1.24426134e-01
6.18868805e-02 2.48002708e-01 -3.13296884e-01 -7.50758886e-01
-1.60294771e-01 -4.76501659e-02 -4.27906103e-02 -1.12772718e-01
6.12049341e-01 -8.12687397e-01 -2.55148530e-01 1.58457488e-01
-8.12679529e-01 -1.56410858e-01 3.43766659e-01 1.96220145e-01
-2.60488331e-01 -5.96251190e-01 6.91484094e-01 -5.06354809e-01
-7.65221834e-01 -3.65887374e-01 -7.00309455e-01 4.76446271e-01
6.44181132e-01 1.15784362e-01 -5.70105135e-01 -2.62951493e-01
-6.96343362e-01 1.66634038e-01 4.37852144e-01 6.16044760e-01
-4.86947715e-01 -1.12706435e+00 -6.20776057e-01 2.64221150e-03
2.71708459e-01 -3.46078694e-01 -1.07783683e-01 4.36575472e-01
-2.25048006e-01 9.00676668e-01 -4.80598658e-01 -3.89128059e-01
-1.36982763e+00 5.53837642e-02 4.19012129e-01 -1.08732963e+00
-3.49779248e-01 6.42296910e-01 -1.29346013e-01 -9.64674830e-01
2.15607107e-01 -9.70370397e-02 -8.82241726e-01 4.38648283e-01
4.11196738e-01 -7.53185973e-02 1.34314969e-01 -1.11114240e+00
-2.74256438e-01 7.33132381e-03 -1.49056002e-01 -4.31217253e-01
1.18485832e+00 -3.07381392e-01 -2.69964695e-01 7.64037073e-01
1.10815012e+00 -6.05786312e-03 -1.03947914e+00 -4.22386169e-01
6.28980935e-01 -8.70335251e-02 -1.80772468e-01 -1.18326962e+00
-3.79997373e-01 6.90887451e-01 2.11497247e-01 9.05735016e-01
4.97662842e-01 1.64253473e-01 6.80305958e-01 3.93186182e-01
5.29900968e-01 -1.32177806e+00 -9.79086459e-02 1.30894327e+00
6.77770555e-01 -1.32848907e+00 -3.76818776e-01 -3.16296726e-01
-9.22471166e-01 1.07821572e+00 6.50892019e-01 2.00015426e-01
4.05422240e-01 3.36642861e-01 6.71581030e-01 -3.37109298e-01
-8.07796180e-01 -4.25031513e-01 8.46555531e-02 6.19390965e-01
1.17451036e+00 6.55958951e-02 -5.62075496e-01 4.82836664e-01
-2.57908881e-01 -3.87499750e-01 5.14952838e-01 1.02510190e+00
-4.39795285e-01 -1.81695771e+00 -7.17459619e-02 3.11048299e-01
-1.00717902e+00 -6.08113110e-02 -8.16152155e-01 9.60260332e-01
-7.51792395e-04 1.36561704e+00 1.06127538e-01 -2.21458018e-01
2.73123413e-01 6.09001517e-01 2.42466241e-01 -1.22084963e+00
-1.49987698e+00 9.24915299e-02 1.29361045e+00 -2.89309740e-01
-9.80965018e-01 -7.12344170e-01 -1.04457796e+00 -4.57797162e-02
-2.80030966e-01 7.26920724e-01 6.19530022e-01 1.36020935e+00
-2.94253435e-02 4.96881157e-01 2.84746438e-01 -8.05603445e-01
-6.28771424e-01 -1.66720486e+00 -4.56141055e-01 6.94011927e-01
1.88053489e-01 -7.38791645e-01 -2.21075669e-01 -4.56150174e-02] | [12.782687187194824, 7.909097671508789] |
c0b6595f-0227-4560-9e88-9f0c49e635f6 | where-are-we-in-named-entity-recognition-from | null | null | https://aclanthology.org/2020.lrec-1.556 | https://aclanthology.org/2020.lrec-1.556.pdf | Where are we in Named Entity Recognition from Speech? | Named entity recognition (NER) from speech is usually made through a pipeline process that consists in (i) processing audio using an automatic speech recognition system (ASR) and (ii) applying a NER to the ASR outputs. The latest data available for named entity extraction from speech in French were produced during the ETAPE evaluation campaign in 2012. Since the publication of ETAPE{'}s campaign results, major improvements were done on NER and ASR systems, especially with the development of neural approaches for both of these components. In addition, recent studies have shown the capability of End-to-End (E2E) approach for NER / SLU tasks. In this paper, we propose a study of the improvements made in speech recognition and named entity recognition for pipeline approaches. For this type of systems, we propose an original 3-pass approach. We also explore the capability of an E2E system to do structured NER. Finally, we compare the performances of ETAPE{'}s systems (state-of-the-art systems in 2012) with the performances obtained using current technologies. The results show the interest of the E2E approach, which however remains below an updated pipeline approach. | ['Yannick Est{\\`e}ve', 'Antoine Caubri{\\`e}re', 'Antoine Laurent', 'Sophie Rosset', 'Emmanuel Morin'] | 2020-05-01 | null | null | null | lrec-2020-5 | ['entity-extraction'] | ['natural-language-processing'] | [ 4.42030840e-02 2.86687613e-01 6.27002180e-01 -5.65734506e-01
-1.19304276e+00 -6.81927383e-01 6.08747721e-01 8.50036889e-02
-1.08722603e+00 4.51885700e-01 7.33642638e-01 -2.87532061e-01
1.47078529e-01 -4.20626700e-01 -5.14030099e-01 -6.51711971e-02
-6.29625237e-03 3.89488429e-01 3.39167893e-01 -4.41326737e-01
1.04086743e-04 9.43382740e-01 -1.44278514e+00 7.18708634e-01
2.65967637e-01 7.67511845e-01 1.09270766e-01 1.18038702e+00
-4.46781069e-01 9.03069317e-01 -1.07887685e+00 -4.00650620e-01
-2.91598924e-02 -2.31380835e-01 -1.05217779e+00 -2.55436540e-01
1.90391511e-01 -6.64114654e-02 -4.68312055e-01 7.58601069e-01
1.03452885e+00 4.39945579e-01 2.06663609e-01 -7.28744149e-01
-3.31908584e-01 1.06519496e+00 2.51420528e-01 3.19512933e-01
5.20554900e-01 -7.14662895e-02 7.72789061e-01 -1.14542913e+00
7.80824065e-01 9.51248109e-01 7.96860993e-01 9.03332651e-01
-8.83696496e-01 -5.09506762e-01 -1.23228684e-01 7.58365020e-02
-1.56765985e+00 -1.26632619e+00 1.85582712e-01 -2.16494933e-01
1.78947675e+00 2.86668360e-01 2.03484863e-01 1.11227369e+00
-4.41431791e-01 9.00654793e-01 1.03905094e+00 -6.21416032e-01
5.15959680e-01 2.82122314e-01 2.65838057e-01 1.52913436e-01
-2.86675215e-01 1.91097379e-01 -7.47463763e-01 1.81536388e-03
2.17261001e-01 -7.15727448e-01 -2.82298714e-01 6.38680398e-01
-1.22543132e+00 3.35462093e-01 -1.60551593e-02 8.64156783e-01
-8.59195232e-01 -2.59224862e-01 6.78119540e-01 2.90162653e-01
4.58680421e-01 5.23653030e-01 -7.86571562e-01 -6.35768354e-01
-1.50412571e+00 -3.51768702e-01 1.36182475e+00 1.04829621e+00
1.70131952e-01 2.19475895e-01 -4.19098169e-01 1.00495970e+00
3.17057967e-01 1.75289854e-01 7.12864101e-01 -4.55743134e-01
6.76954865e-01 2.17100024e-01 8.36504176e-02 -3.01701188e-01
-5.47404170e-01 -2.51268864e-01 -5.91581523e-01 1.83617156e-02
3.97377849e-01 -6.08203411e-01 -1.22770190e+00 1.37980306e+00
1.98251262e-01 7.94972107e-02 6.03181005e-01 3.92634094e-01
1.06949580e+00 9.65779841e-01 3.71790349e-01 2.98681622e-03
1.40614235e+00 -9.88306880e-01 -1.04127836e+00 -7.24444687e-02
6.69641614e-01 -9.73859429e-01 5.07484257e-01 4.47028965e-01
-1.16027725e+00 -6.10558450e-01 -6.75150216e-01 -3.48087177e-02
-7.89844930e-01 6.39439225e-01 -6.27900288e-02 1.15774274e+00
-1.48410702e+00 6.14627898e-01 -8.99339080e-01 -7.58596182e-01
-4.27169696e-04 4.43172812e-01 -6.02809608e-01 4.97734040e-01
-1.40444481e+00 8.96889448e-01 7.14592993e-01 4.40047026e-01
-7.87910044e-01 -5.35561562e-01 -6.48035765e-01 1.74620301e-01
2.08378047e-01 -1.78794831e-01 1.67400837e+00 -6.15041256e-01
-1.85816741e+00 6.94161475e-01 -3.53762090e-01 -7.30432391e-01
3.31493288e-01 -4.09744233e-01 -9.82739449e-01 2.20048845e-01
-2.82552838e-01 6.59850955e-01 2.11844563e-01 -8.20042491e-01
-9.31498766e-01 -1.80573821e-01 -2.63220638e-01 -3.88985910e-02
-2.73887157e-01 8.32677484e-01 -2.39692569e-01 -6.53369784e-01
-2.54894346e-01 -7.71978080e-01 -1.11600906e-01 -7.05641747e-01
-5.08317769e-01 -5.04250169e-01 4.50710446e-01 -1.31716537e+00
1.55315363e+00 -2.37121987e+00 -2.15866074e-01 -1.06868893e-01
-1.54918030e-01 8.58506143e-01 -1.87061518e-01 9.32633579e-01
-4.77022707e-01 5.51018298e-01 -1.74743876e-01 -6.25985265e-01
1.25584722e-01 -6.63641766e-02 -2.61446148e-01 -7.02967197e-02
5.10396481e-01 5.90212464e-01 -6.23843372e-01 -1.82021663e-01
1.31279364e-01 7.87971139e-01 -1.29902095e-01 5.35689354e-01
2.54380673e-01 2.94468731e-01 -8.01883936e-02 1.60235539e-01
3.80179763e-01 4.51665401e-01 -1.16032086e-01 -1.64039910e-01
-8.42940509e-01 1.09134424e+00 -1.38393950e+00 1.54511893e+00
-5.63691080e-01 8.09354305e-01 3.47615093e-01 -4.35950696e-01
1.04991627e+00 1.20013928e+00 8.40547532e-02 -2.33648330e-01
1.07290246e-01 5.21116257e-01 -1.26185700e-01 -4.24631238e-01
6.89751327e-01 -8.75174999e-02 1.16035037e-01 1.43816143e-01
6.20272994e-01 -5.68976393e-03 2.74138212e-01 -4.48256824e-03
1.45116615e+00 -8.39321017e-02 4.90245521e-01 1.17182598e-01
6.96626544e-01 -5.53392991e-03 3.59099001e-01 6.99572623e-01
-4.78531450e-01 7.75662124e-01 1.71994135e-01 1.14993863e-01
-1.10988665e+00 -7.39293218e-01 2.44187191e-02 9.55443978e-01
-9.73120570e-01 -5.41315317e-01 -1.15003788e+00 -7.33941913e-01
-7.88408041e-01 1.23870504e+00 -3.33763361e-01 2.95748204e-01
-8.43602777e-01 -3.52777779e-01 1.56476986e+00 5.38654029e-01
5.75893462e-01 -1.45870817e+00 -2.59219438e-01 6.21412516e-01
-1.62244216e-01 -1.49207354e+00 -3.92171592e-01 5.08808792e-01
-6.39817476e-01 -5.08702099e-01 -1.02390039e+00 -6.42054021e-01
-2.29517445e-02 -2.41962165e-01 9.31519568e-01 -5.01650333e-01
3.13724875e-01 5.93303323e-01 -8.24067712e-01 -6.05753362e-01
-8.92115474e-01 5.66675484e-01 2.54332870e-02 8.91447514e-02
4.99755591e-01 -3.13730508e-01 -2.10903704e-01 1.74684659e-01
-9.01632011e-01 -4.33858931e-01 8.14453185e-01 4.75913554e-01
4.34438080e-01 1.07950009e-01 6.89439297e-01 -7.24495530e-01
5.86922765e-01 -3.28612983e-01 -3.23829800e-01 2.77506500e-01
-3.70427847e-01 5.78856394e-02 5.77330768e-01 -1.45379856e-01
-1.41835093e+00 3.82908493e-01 -1.25551903e+00 -2.73876399e-01
-9.02843654e-01 4.65385973e-01 -4.01424855e-01 3.18433911e-01
5.36388814e-01 2.78410673e-01 -5.44193208e-01 -9.44682837e-01
2.38682434e-01 1.55180466e+00 5.31630039e-01 -7.95470029e-02
3.85246992e-01 1.12812743e-02 -6.46010518e-01 -1.51573145e+00
-6.08737290e-01 -8.27105939e-01 -7.74830401e-01 -6.88845068e-02
1.11429822e+00 -9.74669814e-01 -3.22177082e-01 7.36000121e-01
-1.63405395e+00 -1.57727614e-01 -6.59016728e-01 7.47195721e-01
-2.94243127e-01 3.05812329e-01 -7.60774434e-01 -1.29461348e+00
-5.29299021e-01 -9.10385966e-01 1.11057496e+00 3.37958425e-01
-2.48428032e-01 -8.53212595e-01 3.79707485e-01 3.40019941e-01
4.53770906e-01 -4.69289482e-01 1.73089221e-01 -1.70772183e+00
5.23964241e-02 -1.12158358e-01 4.85333577e-02 7.92659342e-01
-2.80028224e-01 -9.64184571e-03 -1.50830972e+00 -3.15347090e-02
-7.30726272e-02 1.54074878e-01 6.42785430e-01 -5.95504753e-02
4.14380223e-01 -2.32625231e-01 1.73506904e-02 1.61658108e-01
9.92424250e-01 6.14236236e-01 9.63606238e-01 3.67709637e-01
2.63555914e-01 9.71809089e-01 3.13411951e-01 1.54190123e-01
2.56367922e-01 6.01927161e-01 -2.01636940e-01 -4.14125323e-02
-5.29834986e-01 -3.46143126e-01 9.70353842e-01 1.28227210e+00
1.11504018e-01 -4.51375335e-01 -1.17026436e+00 7.40420461e-01
-1.20767176e+00 -1.00555944e+00 -2.97626346e-01 2.09814572e+00
8.43431830e-01 9.55047905e-02 2.22344249e-01 1.59971610e-01
1.03505814e+00 2.37720273e-02 1.84055492e-02 -7.37862289e-01
-1.74306065e-01 6.35238886e-01 3.64227474e-01 2.15870872e-01
-1.17445469e+00 1.03606462e+00 6.39820910e+00 9.60676610e-01
-1.16337991e+00 4.35315490e-01 1.96304291e-01 2.42267817e-01
2.43904710e-01 -8.32082927e-02 -1.35752976e+00 1.43533945e-01
2.17958260e+00 2.11815029e-01 2.63600618e-01 7.55909503e-01
3.64398837e-01 1.60580426e-01 -9.07212973e-01 7.31183231e-01
-1.33666527e-02 -8.78932238e-01 -2.74386734e-01 -1.98047295e-01
2.41052598e-01 6.62178218e-01 -6.14429474e-01 5.93778193e-01
1.42962426e-01 -6.24820769e-01 9.43652689e-01 5.75388670e-01
7.07423091e-01 -6.77280128e-01 9.38126922e-01 1.80227384e-01
-1.29158413e+00 8.88806134e-02 -1.26787126e-01 4.42334145e-01
6.19755864e-01 6.06537759e-01 -1.25817788e+00 7.32442677e-01
7.09440708e-01 6.42881989e-02 -4.84761775e-01 1.33649659e+00
-4.51844692e-01 1.13633132e+00 -5.33809364e-01 -1.65696695e-01
2.49632850e-01 2.88045049e-01 9.04136181e-01 2.00578237e+00
3.67803782e-01 2.36882955e-01 -3.47111732e-01 4.49700236e-01
-1.76079109e-01 4.00407016e-01 -3.00942838e-01 -6.50591433e-01
5.54175854e-01 1.23850906e+00 -7.38134861e-01 -4.18653369e-01
-3.24991465e-01 1.10186779e+00 2.84389436e-01 1.37258261e-01
-3.80234420e-01 -1.01153493e+00 4.03389156e-01 -8.67097825e-02
6.23274803e-01 -4.40026224e-01 2.17624381e-03 -9.56413031e-01
-1.55231118e-01 -8.32088768e-01 3.17082316e-01 -9.89247859e-01
-1.08277798e+00 1.34519160e+00 -2.41603300e-01 -7.68494785e-01
-2.57838190e-01 -7.29498088e-01 -4.76617545e-01 9.28005636e-01
-1.27268839e+00 -8.91800284e-01 2.98964053e-01 3.27266157e-01
7.35085726e-01 -3.10703158e-01 1.07690704e+00 6.45986140e-01
-7.45126545e-01 5.99028528e-01 1.07188232e-01 6.23627126e-01
7.13091373e-01 -1.31409216e+00 9.13848519e-01 1.11930728e+00
5.23052275e-01 4.95581061e-01 5.19381881e-01 -6.82583451e-01
-9.22842085e-01 -1.16046298e+00 1.69702113e+00 -5.37003756e-01
8.33886981e-01 -3.78671378e-01 -7.55228579e-01 7.02920496e-01
5.71232498e-01 -4.23792481e-01 7.95532525e-01 -2.20793709e-02
-1.79633468e-01 1.50295734e-01 -1.03552830e+00 2.89140791e-01
9.12974775e-01 -7.85589039e-01 -9.72282887e-01 -1.27041817e-01
8.43082130e-01 -2.27030039e-01 -1.11215270e+00 2.28336096e-01
3.34735930e-01 -8.13098371e-01 4.52076405e-01 -6.50158644e-01
-1.53493404e-01 -3.04600686e-01 -3.80575567e-01 -1.44601107e+00
1.88754573e-01 -9.65934038e-01 1.60784319e-01 1.99081433e+00
8.61816406e-01 -5.47485471e-01 2.83174902e-01 4.50536877e-01
-5.27935803e-01 -1.39645770e-01 -1.31531215e+00 -7.54828572e-01
-1.71978727e-01 -8.43326986e-01 4.57587510e-01 6.19341910e-01
-2.21381769e-01 4.01828438e-01 2.72888038e-02 5.06974399e-01
-1.19589604e-01 -9.38579440e-01 3.70968938e-01 -8.49088967e-01
-1.17411681e-01 -2.34957770e-01 -5.11787593e-01 -8.56274903e-01
1.00791216e-01 -8.04477632e-01 3.56559843e-01 -1.66768146e+00
-3.94401461e-01 -1.17738560e-01 -3.18152666e-01 7.59874642e-01
1.16169266e-01 -8.50223601e-02 4.47546303e-01 1.52278230e-01
-3.66727799e-01 3.53015840e-01 2.62659997e-01 2.20140889e-01
-4.44272399e-01 2.61215251e-02 -2.89381057e-01 4.49745327e-01
7.64346004e-01 -7.60914683e-01 3.37397248e-01 -3.42208683e-01
1.80171639e-01 8.18162709e-02 5.35925888e-02 -1.33210111e+00
5.19504130e-01 5.70936322e-01 9.75644514e-02 -8.56439471e-01
2.04961851e-01 -7.33938873e-01 1.51703387e-01 1.74767435e-01
-2.81875104e-01 6.26608506e-02 4.01594073e-01 1.14505894e-01
-6.09996796e-01 -6.26034856e-01 7.60927975e-01 8.05456340e-02
-8.43841076e-01 -2.82265335e-01 -1.04295421e+00 -5.59219643e-02
6.12701535e-01 3.93993817e-02 -1.29341871e-01 -1.80686668e-01
-1.26840591e+00 -1.87877968e-01 -2.14800343e-01 3.74957412e-01
3.32917482e-01 -7.71492243e-01 -9.64923203e-01 1.87271293e-02
1.55970883e-02 -4.16585565e-01 2.30444372e-01 8.57516348e-01
-4.27563280e-01 6.48977578e-01 1.15207464e-01 -2.77629234e-02
-1.20592391e+00 2.67717451e-01 5.07169306e-01 -4.91368771e-01
-3.62469435e-01 8.29986811e-01 -3.90708953e-01 -8.63704264e-01
2.41006345e-01 -8.52500424e-02 -6.89178228e-01 5.24983764e-01
8.23078752e-01 8.09140563e-01 7.63924003e-01 -9.13524151e-01
-5.11273623e-01 -4.22770083e-02 -4.15418744e-02 -7.99397171e-01
1.44368696e+00 -1.23091489e-01 1.02439396e-01 5.96901953e-01
1.05031836e+00 5.13775468e-01 -5.22751987e-01 2.48240709e-01
5.41414022e-01 4.32593346e-01 2.77144760e-01 -1.21744323e+00
-8.04529309e-01 1.00009167e+00 7.97613025e-01 3.26171726e-01
1.23426855e+00 -1.23033494e-01 8.28273714e-01 6.53085470e-01
1.21950366e-01 -1.21953058e+00 -8.48308682e-01 1.00787663e+00
7.82821715e-01 -7.07252979e-01 -7.09802449e-01 -3.57094586e-01
-5.59318423e-01 1.37532914e+00 4.31432761e-02 2.62249142e-01
7.35517502e-01 5.09014130e-01 2.72002876e-01 -5.79463504e-03
-5.56791067e-01 -5.38016081e-01 3.02591801e-01 4.65886146e-01
7.84815192e-01 -9.39524621e-02 -5.07350862e-01 9.03289378e-01
-2.47080699e-01 1.01738662e-01 5.69285750e-01 9.53715026e-01
-3.84568721e-01 -1.24106038e+00 -4.54851002e-01 -3.64207178e-02
-8.20018709e-01 -3.74303222e-01 -7.87029088e-01 7.51015604e-01
-4.64368761e-02 1.54664481e+00 -2.49021634e-01 -5.19239187e-01
1.16535091e+00 6.35085404e-01 -1.33472547e-01 -9.30601716e-01
-1.39892316e+00 2.67936271e-02 7.53513873e-01 -3.94397199e-01
-3.21954727e-01 -7.90215790e-01 -1.29857445e+00 1.76783010e-01
-4.36613798e-01 6.50098741e-01 1.27534151e+00 7.89746225e-01
6.40914798e-01 6.99228466e-01 6.51876032e-01 -7.49471962e-01
-5.53843319e-01 -1.28084338e+00 -5.40525496e-01 -1.36828467e-01
-1.72348600e-02 3.45842056e-02 -4.19353217e-01 1.43966213e-01] | [14.210550308227539, 7.002226829528809] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.