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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
6489f455-d269-4069-bc79-413c4da0cc95 | turbo-training-with-token-dropout | 2210.04889 | null | https://arxiv.org/abs/2210.04889v1 | https://arxiv.org/pdf/2210.04889v1.pdf | Turbo Training with Token Dropout | The objective of this paper is an efficient training method for video tasks. We make three contributions: (1) We propose Turbo training, a simple and versatile training paradigm for Transformers on multiple video tasks. (2) We illustrate the advantages of Turbo training on action classification, video-language representation learning, and long-video activity classification, showing that Turbo training can largely maintain competitive performance while achieving almost 4X speed-up and significantly less memory consumption. (3) Turbo training enables long-schedule video-language training and end-to-end long-video training, delivering competitive or superior performance than previous works, which were infeasible to train under limited resources. | ['Andrew Zisserman', 'Weidi Xie', 'Tengda Han'] | 2022-10-10 | null | null | null | null | ['action-classification'] | ['computer-vision'] | [ 2.44122550e-01 -3.81143242e-01 -6.73971236e-01 -9.66749042e-02
-9.55960274e-01 -2.34767750e-01 3.19889992e-01 -6.30347848e-01
-5.23698509e-01 6.58195853e-01 2.94858456e-01 -4.91240948e-01
1.52312204e-01 -5.64646395e-03 -8.63332152e-01 -5.11023521e-01
-6.89881027e-01 -9.64052081e-02 3.06378633e-01 2.41135836e-01
1.56011330e-02 7.75204524e-02 -1.50087595e+00 8.62512171e-01
5.98437190e-01 9.60840464e-01 2.31375307e-01 1.24951947e+00
1.27413973e-01 1.63116407e+00 -3.96582872e-01 -2.31422931e-01
2.47389212e-01 -4.52746242e-01 -7.62322068e-01 3.09319854e-01
6.26462460e-01 -5.98184466e-01 -1.01506841e+00 4.96837676e-01
5.06025791e-01 1.57974049e-01 4.48682100e-01 -1.06038284e+00
-3.48501563e-01 3.50468606e-01 -5.32995760e-01 7.82241106e-01
5.00076115e-01 1.23906665e-01 7.05777049e-01 -8.86602700e-01
1.98249280e-01 1.01068068e+00 6.64662778e-01 7.07380593e-01
-6.27509713e-01 -5.88077486e-01 1.23616844e-01 4.31704313e-01
-1.26301360e+00 -9.80432987e-01 7.58253932e-02 -4.36982006e-01
1.39618194e+00 -6.64289072e-02 6.56104565e-01 1.47326040e+00
4.04301673e-01 1.38199174e+00 8.26931298e-01 -3.00171643e-01
-1.63802981e-01 -3.88970077e-01 -1.66635856e-01 9.88419235e-01
-1.66916549e-01 1.20405763e-01 -7.77916729e-01 -2.39862222e-02
8.36929202e-01 2.23941337e-02 -3.49787354e-01 -7.02926740e-02
-1.35031831e+00 5.12892962e-01 -4.64628860e-02 2.22920328e-01
-3.14890265e-01 8.27664375e-01 1.11579597e+00 7.54716814e-01
4.82522249e-01 -8.59106332e-02 -5.10452032e-01 -1.07694614e+00
-9.00000632e-01 -1.26964420e-01 6.27063692e-01 1.07977438e+00
2.05083057e-01 8.19266677e-01 -2.47548342e-01 7.78404713e-01
-1.89853851e-02 5.71957946e-01 8.18670988e-01 -1.13218844e+00
8.62338126e-01 -1.77901909e-01 -2.63012797e-01 -3.82479072e-01
-1.56398401e-01 -4.26714182e-01 -8.06436002e-01 -1.44216701e-01
1.17074057e-01 -4.27773654e-01 -9.07814682e-01 1.47066367e+00
-3.29163551e-01 8.61353278e-01 2.33374164e-01 5.56515098e-01
5.84436715e-01 6.94977939e-01 1.48267120e-01 -5.20761490e-01
1.11359239e+00 -1.64182413e+00 -6.34571970e-01 -2.15843037e-01
1.15936351e+00 -4.45804119e-01 9.20589924e-01 4.48524624e-01
-1.30472505e+00 -7.38901496e-01 -8.87367368e-01 2.03803584e-01
5.81871122e-02 3.13559860e-01 1.00652325e+00 7.58758843e-01
-1.22138751e+00 6.02397978e-01 -1.05424869e+00 -2.01377422e-01
6.72776163e-01 3.93094271e-01 -3.79789144e-01 -1.60242468e-01
-9.55569804e-01 4.89515156e-01 4.33098733e-01 -2.63066590e-01
-1.27901518e+00 -6.39537513e-01 -1.06785333e+00 2.41699293e-02
7.87804484e-01 -7.22827196e-01 1.55336761e+00 -1.29704857e+00
-1.66720629e+00 5.98390043e-01 -2.73252070e-01 -7.71172941e-01
4.12629664e-01 -6.05167627e-01 -7.00505733e-01 6.17985427e-01
-1.05819903e-01 6.14676356e-01 1.14494002e+00 -4.28097814e-01
-7.94390917e-01 -2.66682152e-02 -1.65571570e-02 4.47603613e-01
-6.59278393e-01 3.19996148e-01 -1.01125276e+00 -9.70382333e-01
-4.63091731e-01 -9.06250894e-01 -9.55017135e-02 -2.25738168e-01
3.19177479e-01 -7.52692446e-02 8.78295720e-01 -6.45318449e-01
1.38908887e+00 -2.26500559e+00 2.33103275e-01 -3.97284657e-01
2.95907080e-01 6.70063555e-01 -2.56156772e-01 1.58321425e-01
-2.56296486e-01 3.36282998e-02 1.82799697e-01 -1.76039457e-01
-3.71315986e-01 3.21520746e-01 -1.57368824e-01 4.52598572e-01
1.33448780e-01 1.18052971e+00 -9.71194148e-01 -5.01001358e-01
2.73303807e-01 2.50708282e-01 -5.34685433e-01 2.43908465e-01
4.38923836e-02 2.51610130e-01 -4.50151891e-01 8.92862797e-01
1.08669154e-01 -4.27418351e-01 2.30820447e-01 -7.53078163e-02
2.06110567e-01 1.36633947e-01 -7.13238597e-01 1.69909346e+00
-5.66728055e-01 9.35387015e-01 -4.57174182e-02 -1.26815045e+00
1.31901592e-01 6.41063809e-01 8.87130499e-01 -9.41850364e-01
-4.63084951e-02 3.23761292e-02 -2.26071864e-01 -8.92980874e-01
3.36883157e-01 1.29671291e-01 1.57783274e-04 3.41848224e-01
4.02646780e-01 5.04513502e-01 4.69959468e-01 2.24319771e-01
1.32808602e+00 3.59025627e-01 3.21269602e-01 -1.17790671e-02
5.61008394e-01 -5.96894801e-01 4.16664451e-01 6.12075448e-01
-2.87882060e-01 2.43540630e-01 4.97173250e-01 -4.27523255e-01
-9.59942698e-01 -7.28378713e-01 2.54734993e-01 1.78624868e+00
-3.11367154e-01 -8.80690813e-01 -5.02125978e-01 -9.49783981e-01
-7.14036748e-02 1.00422941e-01 -4.49472874e-01 1.61239021e-02
-8.48492324e-01 -5.13735771e-01 9.92749333e-01 9.45181012e-01
5.90492785e-01 -7.27135897e-01 -4.96681541e-01 1.03236087e-01
-1.89393207e-01 -1.51250327e+00 -8.75537217e-01 1.77587599e-01
-1.20347023e+00 -9.76903439e-01 -9.81114864e-01 -7.45805562e-01
3.28850031e-01 8.13980937e-01 1.06738162e+00 7.29572251e-02
-3.32465544e-02 7.95736849e-01 -4.37672019e-01 1.24641858e-01
-3.08047980e-01 -6.75903186e-02 3.42690140e-01 -6.60210401e-02
-4.89180237e-02 -5.41879654e-01 -3.56801718e-01 4.76711452e-01
-7.72455394e-01 -9.91489515e-02 8.64390850e-01 8.40308368e-01
4.05661523e-01 1.91072114e-02 5.43390691e-01 -8.23087752e-01
3.84815156e-01 -5.40400147e-01 -1.19366094e-01 4.10043508e-01
-4.82887596e-01 4.88222167e-02 5.76175869e-01 -5.27725518e-01
-8.69750500e-01 -3.55535410e-02 -2.83295125e-01 -1.06189895e+00
1.88822612e-01 6.36260986e-01 3.36443216e-01 -3.62134039e-01
6.52813137e-01 5.76411903e-01 1.26808003e-01 -3.68767917e-01
8.06831792e-02 6.09250724e-01 6.17332458e-01 -5.22121847e-01
4.21398580e-01 2.26264089e-01 -2.00055793e-01 -1.13158298e+00
-6.85077488e-01 -7.64479339e-01 -7.27334261e-01 -4.63408858e-01
7.46419549e-01 -1.57375717e+00 -6.81415617e-01 4.96838629e-01
-6.32800758e-01 -8.46954823e-01 1.27671063e-01 7.49832392e-01
-8.70708525e-01 1.02012086e+00 -8.00017655e-01 -7.06996620e-01
-3.57111305e-01 -1.01439738e+00 9.81034338e-01 -2.43146494e-02
1.39725193e-01 -1.09022653e+00 -1.65055841e-01 5.78704059e-01
2.86286652e-01 -2.37302650e-02 2.80051976e-01 -1.91468939e-01
-4.94896322e-01 -3.14608105e-02 -1.12849265e-01 5.77152193e-01
-4.33698408e-02 -1.39043823e-01 -8.33333731e-01 -8.36725831e-01
-2.66709208e-01 -1.01026678e+00 1.15420282e+00 1.95483536e-01
1.65032935e+00 -2.00557992e-01 -2.01627940e-01 1.03956544e+00
1.03375232e+00 1.93269327e-01 8.04663479e-01 5.51504865e-02
8.08180869e-01 -1.96278408e-01 5.89273453e-01 4.29669976e-01
2.47491017e-01 9.53933060e-01 -2.56699976e-02 5.95801622e-02
-4.12668616e-01 -1.52488247e-01 1.27284622e+00 9.49441254e-01
-4.56329346e-01 -1.48216620e-01 -5.29753745e-01 3.14784765e-01
-2.09342241e+00 -1.35995781e+00 7.62575492e-02 2.12374878e+00
3.89781892e-01 4.34784383e-01 6.11451089e-01 -2.66733095e-02
2.50917345e-01 3.85165930e-01 -4.84892309e-01 -2.84666777e-01
-2.55190004e-02 1.74170360e-01 7.49455035e-01 1.37423072e-02
-1.41085243e+00 9.17400181e-01 8.18296528e+00 1.05845022e+00
-1.05233014e+00 3.81917387e-01 3.20517629e-01 -4.11601067e-01
3.30951840e-01 -4.15126681e-01 -5.70457757e-01 5.17350852e-01
1.53258729e+00 -3.01691622e-01 4.32174355e-01 1.11075509e+00
1.06131606e-01 1.11866079e-01 -1.03970695e+00 1.44970715e+00
4.79263008e-01 -1.38643813e+00 4.74306457e-02 1.31064147e-01
6.05045080e-01 4.24808830e-01 -1.01540620e-02 8.71734142e-01
-7.34771043e-02 -9.54000354e-01 6.22689784e-01 1.55459687e-01
1.31399119e+00 -6.01093411e-01 4.69442844e-01 2.66410351e-01
-1.54911184e+00 -5.85938871e-01 -2.90712327e-01 -1.42575607e-01
1.59269080e-01 -3.12541355e-03 -4.97286677e-01 7.36305475e-01
6.60298586e-01 1.24989581e+00 -6.69573247e-01 1.17470515e+00
3.57778259e-02 1.00358236e+00 8.01790133e-02 1.69867486e-01
5.39580822e-01 2.06289068e-01 2.33982563e-01 1.80914950e+00
3.02500933e-01 6.10032156e-02 7.51689076e-01 -2.82587469e-01
-3.16097081e-01 -1.15278475e-01 -8.05673063e-01 -5.01671731e-01
-3.00163124e-02 8.77346098e-01 -4.27714288e-01 -4.95905340e-01
-8.53922725e-01 1.31256258e+00 2.79259235e-01 3.50395381e-01
-1.07166207e+00 -1.73001409e-01 6.26018941e-01 -1.49935976e-01
6.26309097e-01 -5.78413248e-01 2.38806024e-01 -1.46048069e+00
5.70378825e-02 -1.05633307e+00 8.29278350e-01 -6.74130023e-01
-7.46505737e-01 4.54951942e-01 2.17890739e-02 -1.50523591e+00
-6.20021522e-01 -8.27030122e-01 -4.05201584e-01 1.88023567e-01
-1.54629469e+00 -1.02617431e+00 -1.15947723e-01 8.74758005e-01
1.06178164e+00 -7.63396859e-01 7.06557572e-01 7.20952630e-01
-6.57703459e-01 1.02278399e+00 2.29851156e-01 1.73255637e-01
5.97198248e-01 -9.90021229e-01 3.76658082e-01 8.42513919e-01
3.64986390e-01 2.80759484e-01 1.36722893e-01 -3.43029141e-01
-1.80960369e+00 -1.12529647e+00 4.41222399e-01 -4.73578066e-01
8.11265290e-01 -3.07250530e-01 -7.03595757e-01 7.78041422e-01
1.91068411e-01 2.27803171e-01 8.46423030e-01 1.05151020e-01
-4.57937956e-01 2.21880730e-02 -5.54368675e-01 5.21956742e-01
1.33969343e+00 -8.81873429e-01 -3.12133223e-01 5.51412404e-01
7.09681630e-01 -6.93587363e-01 -9.72386360e-01 3.25882405e-01
7.13495553e-01 -7.47825384e-01 9.58576500e-01 -9.55122054e-01
3.89548808e-01 -7.56684244e-02 4.37796954e-03 -8.78624737e-01
-4.04401779e-01 -1.10930634e+00 -8.94507408e-01 4.55220103e-01
2.64203966e-01 -9.24575925e-02 8.55414331e-01 -1.30711943e-01
-5.77114046e-01 -9.73771751e-01 -9.91351426e-01 -9.73139644e-01
-2.98309654e-01 -8.72434676e-01 -1.20730586e-01 6.96338952e-01
1.66209713e-01 3.40053231e-01 -1.09311259e+00 -1.73300400e-01
3.13400567e-01 -2.82208055e-01 7.21418023e-01 -3.78375620e-01
-6.80846274e-01 -3.32014441e-01 -5.30921102e-01 -1.77930582e+00
3.27977031e-01 -7.72311091e-01 -2.59851098e-01 -1.13609171e+00
3.41022015e-01 1.03709012e-01 -5.25610387e-01 6.09266460e-01
-1.44448161e-01 4.27292943e-01 1.43929690e-01 2.76784360e-01
-1.34576035e+00 5.77110350e-01 1.21246135e+00 -7.20274867e-03
7.26577267e-02 3.52394849e-01 -5.13067424e-01 5.89921832e-01
4.27933842e-01 -1.12143971e-01 -7.52513051e-01 -7.00622261e-01
-5.21261469e-02 3.53358954e-01 7.26972595e-02 -1.35836017e+00
1.44200131e-01 -7.42270947e-02 3.67559284e-01 -2.93851703e-01
4.03036952e-01 -5.29184103e-01 -3.12184304e-01 5.42608559e-01
-3.18955958e-01 1.63319856e-01 2.89844513e-01 8.42010498e-01
-1.61318332e-01 2.68033873e-02 6.12242639e-01 -1.25989765e-01
-1.40687716e+00 4.99629050e-01 -9.19468105e-01 1.55271366e-01
1.22516322e+00 -4.45082366e-01 -4.64006625e-02 -4.96788770e-01
-6.27956927e-01 4.56425220e-01 1.48123011e-01 6.24072492e-01
8.31028640e-01 -1.44194627e+00 -6.83306038e-01 1.42943874e-01
1.62694648e-01 -6.66838825e-01 5.01185060e-01 9.23188865e-01
-5.07949471e-01 7.24980831e-01 -1.77243710e-01 -7.76387990e-01
-1.69673705e+00 5.95530510e-01 4.38601851e-01 -2.05555797e-01
-8.46274137e-01 7.00058579e-01 -2.41725016e-02 8.67758617e-02
5.66174746e-01 -1.50982380e-01 -1.27316892e-01 -1.65722668e-01
8.13529730e-01 3.88372958e-01 -4.86535905e-03 -4.57105935e-01
-2.37129897e-01 4.13388550e-01 -2.12325111e-01 1.26769871e-01
1.21196187e+00 5.07099740e-02 4.47998405e-01 3.33332032e-01
1.22175312e+00 -3.05668145e-01 -1.34005737e+00 -1.64372578e-01
-1.06937908e-01 -7.77287245e-01 2.40720436e-01 -3.93695056e-01
-1.10175467e+00 7.06306100e-01 6.91639721e-01 -1.27632290e-01
1.13948309e+00 -1.93852589e-01 8.99721265e-01 9.33233917e-01
4.89138067e-01 -1.23170757e+00 7.44817019e-01 7.48749018e-01
5.60448229e-01 -1.05033445e+00 1.01575077e-01 -2.26615340e-01
-1.02216804e+00 1.28113139e+00 6.72469318e-01 3.02076906e-01
4.88709241e-01 3.04925382e-01 -2.31815368e-01 -1.84422240e-01
-1.26328421e+00 -2.60820210e-01 2.86089629e-01 4.85129684e-01
6.51751637e-01 -2.92992890e-01 8.85112360e-02 9.75037217e-02
5.47849655e-01 4.59975690e-01 4.06555921e-01 1.11917341e+00
-3.90267998e-01 -8.58169615e-01 7.91105442e-03 5.91774642e-01
-8.18254948e-01 1.92904081e-02 2.30982788e-02 5.59225559e-01
-1.65061221e-01 7.91196346e-01 -5.61811477e-02 -7.22560227e-01
1.79230675e-01 2.19218180e-01 7.12614179e-01 -5.73926628e-01
-3.41475427e-01 1.16476208e-01 5.96206546e-01 -9.50596273e-01
-6.44054711e-01 -5.50370693e-01 -9.49228764e-01 -4.23476726e-01
-1.21906877e-01 9.36627090e-02 2.25809664e-01 1.09992599e+00
5.55088997e-01 7.78329194e-01 7.66482472e-01 -9.45197403e-01
-6.22043192e-01 -7.81151056e-01 -4.23463136e-01 2.17761561e-01
2.02964708e-01 -6.34269059e-01 -4.97677661e-02 3.47349495e-01] | [9.124082565307617, 0.756371259689331] |
709440d6-03a3-4226-81d8-c73bff27f615 | incentivizing-exploration-with-linear | 2306.01990 | null | https://arxiv.org/abs/2306.01990v1 | https://arxiv.org/pdf/2306.01990v1.pdf | Incentivizing Exploration with Linear Contexts and Combinatorial Actions | We advance the study of incentivized bandit exploration, in which arm choices are viewed as recommendations and are required to be Bayesian incentive compatible. Recent work has shown under certain independence assumptions that after collecting enough initial samples, the popular Thompson sampling algorithm becomes incentive compatible. We give an analog of this result for linear bandits, where the independence of the prior is replaced by a natural convexity condition. This opens up the possibility of efficient and regret-optimal incentivized exploration in high-dimensional action spaces. In the semibandit model, we also improve the sample complexity for the pre-Thompson sampling phase of initial data collection. | ['Mark Sellke'] | 2023-06-03 | null | null | null | null | ['thompson-sampling'] | ['methodology'] | [ 1.90034419e-01 5.55823326e-01 -1.24035907e+00 -4.51287985e-01
-7.28899837e-01 -7.75629997e-01 2.88819790e-01 -1.06257319e-01
-6.40988827e-01 1.11861515e+00 5.17933965e-01 -9.19279575e-01
-8.81310225e-01 -8.14324737e-01 -9.62932467e-01 -6.34389997e-01
1.76405348e-03 7.07545400e-01 -5.33306241e-01 1.96385920e-01
1.07257351e-01 2.05834061e-01 -1.01933229e+00 -2.88593164e-03
8.82628441e-01 1.04706264e+00 -1.69974923e-01 7.46818066e-01
2.70488113e-02 4.46104825e-01 -5.45107201e-02 -4.76706922e-01
8.17742825e-01 -4.72529292e-01 -9.93180335e-01 2.51649737e-01
1.72910579e-02 -8.38659048e-01 -2.55015999e-01 1.16294289e+00
1.25705272e-01 3.39040637e-01 2.56194532e-01 -1.13614750e+00
-4.53828186e-01 1.35924768e+00 -6.60004258e-01 1.74921483e-01
2.19654031e-02 -9.96785313e-02 1.51048660e+00 1.04677558e-01
4.48793948e-01 1.21616089e+00 1.67666167e-01 5.03885150e-01
-1.48418212e+00 -2.89257050e-01 4.67872441e-01 8.56644213e-02
-2.99502164e-01 -2.20721871e-01 7.00473964e-01 -1.79376841e-01
2.94982523e-01 7.82496810e-01 1.10003042e+00 1.04292989e+00
-7.27201581e-01 1.53472185e+00 1.27891159e+00 -6.59757018e-01
7.55470753e-01 4.89466563e-02 5.86761832e-01 9.69940573e-02
1.00239253e+00 6.69047117e-01 -4.61812437e-01 -5.92371225e-01
9.67787862e-01 2.38963857e-01 -2.14545667e-01 -9.31544483e-01
-9.02856529e-01 1.28198123e+00 1.65932506e-01 -1.93776846e-01
-8.12713742e-01 3.55500937e-01 1.17121145e-01 6.57723665e-01
4.09389794e-01 5.52438259e-01 -2.44942769e-01 -3.40235889e-01
-8.28174949e-01 4.30558741e-01 9.34329212e-01 8.72762442e-01
4.53559816e-01 -2.27488965e-01 -5.93303561e-01 3.46368134e-01
3.30346763e-01 3.93660724e-01 1.60899416e-01 -1.27208352e+00
7.40428567e-01 -5.78031503e-03 1.25498164e+00 -3.12278956e-01
-1.16460890e-01 -7.11028934e-01 -2.73735374e-01 3.29472572e-02
6.99623048e-01 -5.37213087e-01 -6.42844558e-01 1.86106133e+00
5.97227573e-01 -1.30308256e-01 -1.36696637e-01 8.44588220e-01
-2.32846916e-01 2.60466337e-01 -3.01622748e-01 -6.52204037e-01
9.23943877e-01 -8.89322817e-01 -8.74505639e-01 -2.77878344e-01
3.74069571e-01 -1.32871985e-01 9.31812346e-01 6.48016810e-01
-1.37219918e+00 1.43569112e-01 -8.85375142e-01 3.27067316e-01
2.74919331e-01 -1.28107101e-01 1.28800070e+00 1.34663820e+00
-7.15631604e-01 6.63012862e-01 -8.44716132e-01 -5.10294102e-02
9.40138519e-01 4.26112533e-01 2.13171959e-01 -1.28808960e-01
-7.62979627e-01 4.31634396e-01 3.50179523e-01 5.24733216e-03
-1.01364040e+00 -7.22425699e-01 -2.01501429e-01 1.69231176e-01
8.35501909e-01 -6.16495609e-01 1.78352928e+00 -1.06029987e+00
-1.81708515e+00 4.08699155e-01 4.38849851e-02 -8.73962700e-01
1.11711693e+00 -5.14408052e-01 3.08486313e-01 -8.57337844e-03
-1.58364177e-01 -8.61733705e-02 7.54584193e-01 -8.72256756e-01
-8.20730209e-01 -3.90298933e-01 7.55115807e-01 4.57057059e-01
-3.85044217e-01 -1.14182413e-01 1.46241970e-02 -4.71929014e-01
-1.18426196e-02 -1.05144095e+00 -6.07102454e-01 -2.85463452e-01
-3.91107619e-01 -5.15017249e-02 -7.48836398e-02 -3.73175561e-01
9.55693245e-01 -1.98257041e+00 -7.95475692e-02 5.37159681e-01
-5.52434251e-02 -2.91631281e-01 -6.50257431e-03 2.06596330e-01
2.96500046e-02 2.94567674e-01 2.22836286e-01 -1.66252285e-01
4.48776543e-01 2.21618131e-01 -5.95729172e-01 7.34419465e-01
-7.78733969e-01 8.37114930e-01 -8.58789921e-01 9.94360447e-02
-7.63916224e-02 -6.51111126e-01 -9.24916863e-01 1.27616778e-01
-5.05760133e-01 3.66214722e-01 -1.08329380e+00 2.24569663e-01
6.18160903e-01 -3.63222092e-01 3.97399843e-01 5.24589121e-01
-3.88885327e-02 5.75582743e-01 -1.32883620e+00 1.47781265e+00
-2.45680392e-01 2.06740409e-01 3.67721200e-01 -1.28562641e+00
2.25594163e-01 4.13266450e-01 6.40851974e-01 -9.60552245e-02
3.15365613e-01 4.11072895e-02 -5.44012450e-02 -2.43592530e-01
3.20843160e-01 -2.83837616e-01 1.47029430e-01 1.12358212e+00
-4.48585004e-01 2.71112174e-02 2.46705934e-02 1.72207177e-01
7.44485736e-01 3.32357921e-02 3.59171838e-01 -4.81335580e-01
-3.00631344e-01 1.59260839e-01 4.50633019e-01 1.53123987e+00
9.14472416e-02 4.41304855e-02 5.70736110e-01 -2.52257943e-01
-9.49426472e-01 -7.64649272e-01 -2.07496181e-01 1.19862926e+00
4.26024199e-02 2.27214128e-01 -5.55796027e-01 -7.17081428e-01
5.61024427e-01 9.12304163e-01 -1.01703858e+00 2.29439676e-01
-2.05373596e-02 -1.04879856e+00 -1.34651750e-01 4.53791142e-01
4.13184941e-01 -4.29694116e-01 -8.34567785e-01 3.32396269e-01
-1.43838689e-01 -4.64448810e-01 -5.48825264e-01 1.92249745e-01
-1.33688653e+00 -1.00396764e+00 -8.14775109e-01 1.02552280e-01
5.90999842e-01 3.81129265e-01 7.20975757e-01 -4.31145370e-01
2.19910577e-01 5.96682489e-01 -2.56211758e-01 -7.38125563e-01
7.14798719e-02 -1.95465803e-01 1.03530381e-03 2.61963785e-01
2.51281679e-01 -3.54813039e-01 -9.23662543e-01 3.04120481e-01
-6.47659063e-01 1.49206594e-01 2.99971461e-01 9.81145382e-01
5.58943152e-01 -1.08990490e-01 5.58493555e-01 -1.11861110e+00
7.09786117e-01 -5.74355245e-01 -1.16290176e+00 3.84792805e-01
-6.02929115e-01 2.05316424e-01 1.63304776e-01 -8.41806829e-01
-1.39256263e+00 -7.77081028e-02 4.89799201e-01 -1.09105362e-02
2.96249270e-01 6.26886845e-01 -2.24666744e-02 -9.79883596e-03
6.97174966e-01 -2.75752664e-01 -1.21572554e-01 -8.58171344e-01
6.28154874e-01 6.42287612e-01 8.83518234e-02 -9.61590767e-01
3.95634711e-01 8.01624119e-01 -3.33776101e-02 -3.21455508e-01
-1.18566728e+00 -2.04057112e-01 1.05754510e-01 7.40701854e-02
3.50088060e-01 -5.82671523e-01 -1.08837330e+00 -1.31921574e-01
-5.27627409e-01 -6.47781253e-01 -9.55406427e-01 1.01049733e+00
-1.18475628e+00 2.73126096e-01 -2.64736772e-01 -1.74696064e+00
8.56267661e-02 -6.64542198e-01 2.38963276e-01 2.36657292e-01
-4.20292914e-02 -7.10618019e-01 7.70692751e-02 6.11722171e-01
2.80652314e-01 -1.45191580e-01 6.22917116e-01 -6.90055847e-01
-7.63853312e-01 -3.24732304e-01 6.72022998e-02 -1.54128864e-01
4.27168980e-02 -4.18443799e-01 -6.61614656e-01 -5.03583312e-01
1.95411816e-01 -4.83442307e-01 8.86510849e-01 1.19722629e+00
1.26063299e+00 -9.13336694e-01 -4.38659579e-01 5.43628335e-01
1.12514341e+00 2.82384306e-01 2.43536294e-01 5.09264469e-01
-8.37094188e-02 4.82554018e-01 8.77971172e-01 9.89399791e-01
-2.49580927e-02 4.34976250e-01 5.48729420e-01 5.71112484e-02
8.68132234e-01 -5.02551615e-01 -2.43560094e-02 -3.13022673e-01
-1.93789363e-01 -2.02405989e-01 -2.34726414e-01 8.03591013e-01
-2.34149408e+00 -1.00583696e+00 2.15922475e-01 2.85999727e+00
1.06277692e+00 1.24900050e-01 8.33706975e-01 1.84768274e-01
6.45122170e-01 -1.32880121e-01 -1.10854077e+00 -4.12821323e-01
-3.38627398e-02 4.97182161e-02 1.29617929e+00 7.87866294e-01
-9.95115578e-01 3.50479245e-01 7.50439215e+00 6.77322328e-01
-4.52923685e-01 3.83018136e-01 1.00026679e+00 -7.85020113e-01
-7.32940435e-01 2.23203078e-01 -5.91440678e-01 3.45234185e-01
7.94906616e-01 -4.69509870e-01 8.48848701e-01 1.15310371e+00
3.09131414e-01 -4.55734134e-01 -1.28354716e+00 6.26907051e-01
-7.47644067e-01 -1.36289549e+00 -6.32544875e-01 5.42793095e-01
1.06260371e+00 -3.40962177e-03 1.66556507e-01 -1.40745133e-01
1.59170938e+00 -7.98943698e-01 8.19130421e-01 2.06889659e-01
5.52564740e-01 -7.08972633e-01 2.25323930e-01 4.24077868e-01
-2.31088966e-01 -6.72800362e-01 -4.81161267e-01 -3.56381893e-01
2.41573274e-01 6.28825188e-01 -6.42226815e-01 4.90127087e-01
4.08618808e-01 3.21417689e-01 4.24874544e-01 1.42451394e+00
-3.31050217e-01 1.01512587e+00 -8.13079596e-01 -1.75148115e-01
4.70917106e-01 -5.55707574e-01 5.05562603e-01 5.05061507e-01
2.64987051e-01 2.94249773e-01 1.01503581e-01 6.95322275e-01
-1.24892853e-02 1.05143897e-01 -1.55438587e-01 -2.80655980e-01
3.51016343e-01 7.20081031e-01 -4.16076124e-01 -3.82016063e-01
-2.73261011e-01 4.98227298e-01 6.00796416e-02 5.74218094e-01
-5.73771238e-01 3.56022000e-01 7.39282489e-01 -1.53295159e-01
5.92202067e-01 2.45801270e-01 -7.13256598e-01 -1.08636045e+00
3.90739366e-02 -6.57997072e-01 9.15378213e-01 -2.09345073e-01
-9.79777038e-01 -3.18412542e-01 2.70157993e-01 -6.71368718e-01
-3.02686989e-01 -2.42920458e-01 -5.18794775e-01 7.62290597e-01
-1.51039386e+00 -5.07748127e-01 4.24456686e-01 4.97023791e-01
2.65704185e-01 2.86777854e-01 5.54171681e-01 -2.04269707e-01
-5.42518973e-01 4.88858491e-01 8.52687418e-01 -4.92162079e-01
-1.26635795e-02 -1.37175202e+00 6.69160932e-02 7.23939240e-01
1.05078340e-01 6.45421684e-01 9.99678254e-01 -7.02279508e-01
-1.59402394e+00 -6.61858618e-01 9.87446904e-02 -6.74665570e-02
7.07410216e-01 -9.79606360e-02 -1.61059231e-01 9.87135231e-01
-4.03999761e-02 -2.67902195e-01 7.14179516e-01 6.46730304e-01
4.70772311e-02 -1.72363251e-01 -1.31844103e+00 5.27393460e-01
1.12842333e+00 -8.08259472e-02 -4.78393525e-01 7.04002500e-01
4.78215486e-01 -3.19696099e-01 -6.42550468e-01 6.03313558e-04
7.96590865e-01 -6.15362704e-01 8.54074478e-01 -1.23534286e+00
-7.98082501e-02 3.12456429e-01 -4.96915691e-02 -1.14788485e+00
-3.93346608e-01 -1.54851961e+00 -3.32090080e-01 6.13321841e-01
4.13651168e-01 -7.82531023e-01 1.23824739e+00 1.21482193e+00
4.20532882e-01 -3.90319556e-01 -1.14743161e+00 -1.08813071e+00
1.12332016e-01 -3.40684772e-01 8.89631510e-01 6.61331952e-01
5.29529750e-01 -3.31277907e-01 -8.25082362e-01 -1.42498225e-01
9.02435243e-01 5.68460822e-01 7.15310395e-01 -1.19958580e+00
-1.07448554e+00 -4.73731816e-01 5.55236578e-01 -1.63336229e+00
-3.58917773e-01 -4.42932636e-01 -1.12694234e-01 -1.31615162e+00
3.83513272e-01 -6.58149719e-01 -5.09278178e-01 3.88009012e-01
5.22449659e-03 -2.97707796e-01 9.06866267e-02 1.52801275e-01
-4.71553653e-01 3.47021520e-01 1.30980730e+00 -9.60362181e-02
-4.16602075e-01 8.94893646e-01 -1.31893229e+00 3.37000966e-01
7.56901026e-01 -4.56208378e-01 -7.28890300e-01 -2.94066548e-01
6.05612636e-01 5.60560822e-01 2.30167434e-03 1.19790044e-02
-1.49762020e-01 -9.26552892e-01 -9.67915356e-03 -5.19270718e-01
3.00815523e-01 -8.17569733e-01 3.69422764e-01 5.95626652e-01
-9.48021412e-01 -4.78927940e-01 -2.50123501e-01 9.60931301e-01
5.13729334e-01 -5.96625566e-01 5.77547908e-01 -2.21316248e-01
1.71833172e-01 4.88876760e-01 -3.20551813e-01 7.27547631e-02
9.24523830e-01 -2.36656412e-01 -2.79051978e-02 -1.01639366e+00
-7.84204125e-01 3.30727309e-01 3.42271537e-01 -2.20241383e-01
5.82624972e-03 -1.11593938e+00 -6.25281215e-01 -6.49081171e-02
-2.82624185e-01 -2.94310570e-01 2.91467428e-01 9.12035704e-01
2.12556779e-01 5.55809438e-01 4.14390415e-02 -1.16986409e-01
-9.38160479e-01 8.02551508e-01 2.26182640e-01 -3.67889524e-01
-2.94250280e-01 9.72228408e-01 9.14078504e-02 1.99854404e-01
4.72288787e-01 -4.90765154e-01 9.02653337e-02 -8.45033005e-02
5.59232712e-01 5.80677927e-01 -2.69617438e-01 4.35527712e-01
1.18974403e-01 -1.75461859e-01 -2.11681813e-01 -7.90206432e-01
1.59815991e+00 -3.69589269e-01 2.58170068e-01 1.22865908e-01
4.95775133e-01 1.11492649e-01 -1.62416756e+00 -5.59886336e-01
1.11486904e-01 -1.01609182e+00 3.34860325e-01 -7.25521326e-01
-9.52932000e-01 3.83846611e-01 2.68078685e-01 5.92988133e-01
7.75674164e-01 -1.26248747e-01 1.75763458e-01 4.99206483e-01
5.54776549e-01 -1.24184561e+00 -5.37233710e-01 -9.68331546e-02
6.09001994e-01 -9.95670259e-01 1.82205483e-01 -1.36671990e-01
-5.10644794e-01 6.32278979e-01 -9.92271900e-02 -1.91914260e-01
6.15374267e-01 -1.18749499e-01 -5.38921952e-01 1.60138175e-01
-5.91672182e-01 -1.60873160e-01 4.63031456e-02 3.57888311e-01
-5.14358059e-02 5.92238367e-01 -7.68193483e-01 8.42276692e-01
-2.36913919e-01 3.13434839e-01 6.14880562e-01 9.70767200e-01
-4.49620515e-01 -1.23364544e+00 -5.28340578e-01 9.75346684e-01
-7.90217459e-01 1.19472653e-01 -2.40721151e-01 3.01171273e-01
-8.71556401e-01 1.09987760e+00 -7.98089206e-02 2.93604672e-01
-8.65031630e-02 -8.64530951e-02 7.30849385e-01 -3.96168530e-01
-2.00294912e-01 4.16459739e-01 3.80131304e-01 -5.12580454e-01
-5.52785397e-01 -9.67236161e-01 -3.92494440e-01 -3.89808476e-01
-7.75492430e-01 7.24925876e-01 5.65153122e-01 9.22808051e-01
1.52762741e-01 1.37455195e-01 8.95885229e-01 -4.27572429e-01
-1.60161781e+00 -8.82727981e-01 -9.34826732e-01 1.84218049e-01
6.80407524e-01 -6.67967856e-01 -3.53084505e-01 -4.76020098e-01] | [4.490444660186768, 3.2543203830718994] |
ba940022-d4d5-435f-ab6e-20903e6a38ce | non-linear-pairwise-language-mappings-for-low | 2207.03391 | null | https://arxiv.org/abs/2207.03391v1 | https://arxiv.org/pdf/2207.03391v1.pdf | Non-Linear Pairwise Language Mappings for Low-Resource Multilingual Acoustic Model Fusion | Multilingual speech recognition has drawn significant attention as an effective way to compensate data scarcity for low-resource languages. End-to-end (e2e) modelling is preferred over conventional hybrid systems, mainly because of no lexicon requirement. However, hybrid DNN-HMMs still outperform e2e models in limited data scenarios. Furthermore, the problem of manual lexicon creation has been alleviated by publicly available trained models of grapheme-to-phoneme (G2P) and text to IPA transliteration for a lot of languages. In this paper, a novel approach of hybrid DNN-HMM acoustic models fusion is proposed in a multilingual setup for the low-resource languages. Posterior distributions from different monolingual acoustic models, against a target language speech signal, are fused together. A separate regression neural network is trained for each source-target language pair to transform posteriors from source acoustic model to the target language. These networks require very limited data as compared to the ASR training. Posterior fusion yields a relative gain of 14.65% and 6.5% when compared with multilingual and monolingual baselines respectively. Cross-lingual model fusion shows that the comparable results can be achieved without using posteriors from the language dependent ASR. | ['Thomas Hain', 'Darshan Adiga Haniya Narayana', 'Muhammad Umar Farooq'] | 2022-07-07 | null | null | null | null | ['transliteration'] | ['natural-language-processing'] | [-1.87285244e-01 1.67064667e-02 1.21826552e-01 -3.90524834e-01
-1.70456958e+00 -4.96170998e-01 6.36264205e-01 -1.07385851e-01
-7.64216900e-01 8.02883029e-01 2.60450393e-01 -5.35168231e-01
4.20749009e-01 -2.32646301e-01 -7.74571955e-01 -6.44885182e-01
5.45244336e-01 6.96952283e-01 6.09265231e-02 -1.99918285e-01
-3.90602678e-01 2.14249849e-01 -1.08516824e+00 3.45940306e-03
8.57085586e-01 5.74234784e-01 9.18090582e-01 7.82235444e-01
-2.63227582e-01 1.80916533e-01 -5.75470567e-01 -5.24967909e-01
2.05065086e-01 -2.11380705e-01 -1.48639604e-01 -2.70410329e-01
4.12639201e-01 7.46724755e-02 -2.58093804e-01 1.19580305e+00
9.31805074e-01 1.42832369e-01 5.58701932e-01 -7.83970892e-01
-4.56104428e-01 9.50856268e-01 -4.98695999e-01 9.56314951e-02
3.08507025e-01 -1.45563349e-01 7.17856944e-01 -1.15970159e+00
3.72289538e-01 1.60471714e+00 4.49648529e-01 4.14072514e-01
-1.16144884e+00 -7.84369409e-01 -2.40271632e-02 1.54185519e-01
-1.65462744e+00 -1.09475219e+00 4.71964568e-01 -2.16579303e-01
1.42333746e+00 1.83340371e-01 2.07704961e-01 1.20130730e+00
8.44482854e-02 7.54342973e-01 1.36171973e+00 -8.01426888e-01
-2.33673990e-01 6.32465303e-01 2.20643710e-02 3.51727247e-01
-2.12296203e-01 1.01399638e-01 -7.19347000e-01 2.00122356e-01
2.25314200e-01 -5.77097893e-01 -1.23651855e-01 4.10781980e-01
-1.10732281e+00 6.21316075e-01 -2.62073666e-01 5.14113247e-01
-4.27793622e-01 -1.35904834e-01 4.73732114e-01 1.93491086e-01
5.12462199e-01 -1.76368862e-01 -8.01127613e-01 -2.81727076e-01
-1.05202949e+00 -2.54544377e-01 6.63770795e-01 9.94255900e-01
4.99801338e-01 6.93882942e-01 1.07823767e-01 1.49130213e+00
8.86189163e-01 1.20604205e+00 7.99058020e-01 -3.50168258e-01
6.50283575e-01 -1.47171378e-01 -2.85146385e-01 -3.28553200e-01
-8.29083845e-02 -5.57784736e-01 -6.76231503e-01 -1.46201372e-01
2.04776540e-01 -2.89658368e-01 -1.13080192e+00 1.85864449e+00
1.64599434e-01 -1.17527135e-01 7.28937089e-01 4.45420057e-01
7.71093607e-01 1.24284112e+00 1.62027672e-01 -3.66853416e-01
1.39705086e+00 -1.11113691e+00 -1.15545475e+00 -4.35996562e-01
5.23648858e-01 -1.55236542e+00 1.06213343e+00 4.27595437e-01
-1.28283572e+00 -8.11391294e-01 -9.83702064e-01 -3.83257531e-02
-5.64856052e-01 5.03538728e-01 -1.81494057e-01 9.74340737e-01
-1.23438311e+00 -1.89785182e-01 -7.74594367e-01 -5.42833984e-01
-3.94097596e-01 4.91882890e-01 -4.30729032e-01 1.18227638e-02
-1.41040051e+00 1.24664998e+00 7.60650098e-01 1.53068617e-01
-9.43056345e-01 -3.00038904e-01 -9.35240269e-01 -1.39544234e-01
1.54503673e-01 -8.84813741e-02 1.31671214e+00 -5.58166206e-01
-2.01690483e+00 6.44526720e-01 -4.35575664e-01 -5.47549367e-01
3.77346724e-01 -2.46776402e-01 -8.63920450e-01 -4.83700216e-01
3.87311820e-03 7.09030092e-01 6.27873540e-01 -1.08605695e+00
-1.03435445e+00 -2.11664021e-01 -7.18596995e-01 5.97007990e-01
-1.43427089e-01 5.36174893e-01 -8.21614623e-01 -7.64692187e-01
2.86933482e-02 -1.02389479e+00 2.09133103e-02 -9.65685546e-01
-3.26785237e-01 -2.11059123e-01 7.89286256e-01 -1.48008430e+00
1.18342078e+00 -2.06957650e+00 5.38288504e-02 1.76018216e-02
-6.67352557e-01 4.96002227e-01 -3.24189290e-02 2.98830777e-01
1.88254982e-01 -9.81296897e-02 -1.80909276e-01 -8.53720963e-01
5.54415695e-02 3.29136699e-01 -1.30097628e-01 2.67189354e-01
-1.60466507e-01 6.12703323e-01 -4.74546373e-01 -5.30342579e-01
5.53229511e-01 9.07693207e-01 -6.85247332e-02 2.05783859e-01
7.62833357e-02 5.47500670e-01 2.95464665e-01 5.07123291e-01
6.99600279e-01 8.03283155e-01 3.24833661e-01 -1.00061417e-01
-3.45100135e-01 6.11697197e-01 -1.12654197e+00 1.70224679e+00
-1.06465113e+00 5.29971242e-01 3.63489479e-01 -7.63034582e-01
1.10695207e+00 9.36232686e-01 -7.79788271e-02 -6.57900929e-01
1.14810005e-01 8.46310616e-01 3.11465502e-01 -1.02365561e-01
4.81356144e-01 -3.64367187e-01 -3.84209961e-01 -6.52714595e-02
5.57749808e-01 -2.95836568e-01 -3.55439284e-03 -2.52467662e-01
3.38572860e-01 2.62753487e-01 4.58709985e-01 -2.19699875e-01
7.26298988e-01 -3.15370739e-01 6.57587707e-01 5.60355186e-01
-1.26095846e-01 4.02501196e-01 -3.56847614e-01 2.45115563e-01
-1.04736006e+00 -1.24614632e+00 -1.32817507e-01 1.14676225e+00
-4.81705189e-01 -2.43183896e-01 -8.23731303e-01 -3.70852530e-01
-4.65061516e-01 1.16283548e+00 2.84203976e-01 2.06161767e-01
-9.01129901e-01 -8.46287370e-01 9.04400051e-01 1.86289594e-01
4.34488028e-01 -9.54915524e-01 3.80067676e-01 5.56166887e-01
-3.24580401e-01 -1.62224340e+00 -6.00193381e-01 4.80274528e-01
-5.37557900e-01 -2.31093362e-01 -9.47344601e-01 -1.05931592e+00
7.02139288e-02 -7.63364509e-03 7.17597127e-01 -8.56468320e-01
4.03948545e-01 4.06908542e-02 -2.34757811e-01 -4.54294413e-01
-1.09735036e+00 1.09816164e-01 5.15505195e-01 3.73016633e-02
5.05028784e-01 -4.00851727e-01 2.53355831e-01 1.31747216e-01
-3.79426539e-01 1.66750606e-02 7.99952269e-01 6.96500897e-01
8.19446146e-01 -1.00384563e-01 7.26077259e-01 -5.47688663e-01
5.23308516e-01 -4.39883679e-01 -7.92063415e-01 3.66128862e-01
-4.03861374e-01 -1.17040537e-02 7.43716419e-01 -4.27763969e-01
-1.49905455e+00 2.99908072e-01 -7.43819356e-01 -3.20971310e-01
-3.47267598e-01 6.75942063e-01 -6.57736599e-01 3.89537662e-01
2.43100002e-01 4.67444271e-01 -3.10804963e-01 -8.43938053e-01
4.19716328e-01 1.19904232e+00 7.39579737e-01 -4.02544171e-01
4.54745650e-01 -2.79492944e-01 -5.17415822e-01 -1.21178055e+00
-2.74316758e-01 -5.06583273e-01 -5.39303243e-01 -4.49891575e-02
9.90904093e-01 -1.20671844e+00 -3.10659468e-01 6.94812536e-01
-1.43291712e+00 -1.07872680e-01 6.80314302e-02 1.15938294e+00
-4.10137743e-01 2.59595573e-01 -6.59058273e-01 -1.10955274e+00
-3.67073447e-01 -1.61164653e+00 7.92706549e-01 9.09504294e-02
2.76177377e-02 -1.08071947e+00 1.27295017e-01 5.70204675e-01
3.94032627e-01 -4.74557966e-01 6.79853737e-01 -1.07582366e+00
-1.49887562e-01 -2.69021779e-01 3.86719182e-02 9.85878646e-01
2.37122163e-01 -8.92103985e-02 -1.19474745e+00 -3.26867431e-01
1.44440949e-01 -3.17527466e-02 3.17185819e-01 5.78585923e-01
2.42466971e-01 -1.27374738e-01 -1.06715433e-01 3.80197078e-01
1.16984701e+00 6.64176702e-01 2.76333332e-01 -1.75834838e-02
8.36275220e-01 4.78444844e-01 3.96000654e-01 -1.53831854e-01
5.04034162e-01 9.38570201e-01 -2.03012377e-01 -1.28163531e-01
-6.43961489e-01 -1.94876850e-01 1.13341045e+00 2.06821775e+00
4.47380185e-01 -5.25020599e-01 -1.22959554e+00 6.70370102e-01
-1.59203935e+00 -3.98756385e-01 -2.56008357e-01 2.48845410e+00
8.91016603e-01 2.12567881e-01 -9.92598608e-02 -1.18068047e-02
9.75454748e-01 -2.43723523e-02 -3.29903848e-02 -6.52023673e-01
-6.39734745e-01 2.28789151e-01 7.34435201e-01 1.24109662e+00
-8.84515166e-01 1.49165106e+00 5.60019493e+00 1.52042961e+00
-1.14868760e+00 8.19950640e-01 2.84110337e-01 1.21718891e-01
3.10033821e-02 5.21510430e-02 -1.39440846e+00 4.64031070e-01
1.76871789e+00 3.77416536e-02 4.18354481e-01 5.48243284e-01
3.46022278e-01 -1.43276066e-01 -8.20878685e-01 1.20151460e+00
2.45939553e-01 -5.94987988e-01 -4.40613963e-02 3.11959479e-02
7.08748460e-01 5.95686197e-01 5.69795519e-02 5.21454692e-01
5.07179320e-01 -8.02350938e-01 9.51197803e-01 4.13465172e-01
8.18523943e-01 -8.74870002e-01 8.86602521e-01 5.29951572e-01
-1.31234324e+00 3.41206789e-01 -2.33628213e-01 3.53290588e-01
7.05571294e-01 2.82756954e-01 -1.27412951e+00 6.61796510e-01
4.35349047e-01 1.49872363e-01 -2.47758031e-01 6.82750821e-01
-1.40251160e-01 9.22868907e-01 -7.40874350e-01 2.08071381e-01
3.15306604e-01 -1.77434728e-01 7.86448777e-01 1.76788688e+00
7.73565650e-01 -6.45299554e-01 1.21716619e-01 2.49145657e-01
7.37468526e-02 7.06584632e-01 -6.26457572e-01 -7.05114305e-02
5.39976418e-01 9.59336042e-01 -4.46577221e-01 -4.51592267e-01
-4.98744309e-01 9.76405978e-01 7.61744455e-02 3.69259447e-01
-5.90161681e-01 -2.90229410e-01 3.04572612e-01 -2.71771610e-01
3.21951732e-02 -4.67991561e-01 4.49149171e-03 -9.08139706e-01
-1.63812995e-01 -1.08233595e+00 3.75450142e-02 -5.85959375e-01
-1.07092786e+00 1.02308416e+00 1.65178403e-02 -8.56377602e-01
-6.40189707e-01 -6.06146991e-01 -1.52491659e-01 1.58776450e+00
-1.44552588e+00 -1.47549200e+00 3.67610365e-01 3.91616732e-01
9.83282089e-01 -6.85758471e-01 1.03796887e+00 8.42721879e-01
-6.33234024e-01 7.98340082e-01 4.65740293e-01 -2.32315026e-02
9.64448154e-01 -1.09840012e+00 4.56756860e-01 1.19760501e+00
4.12984043e-01 3.27209353e-01 7.23215580e-01 -6.52046382e-01
-1.12227952e+00 -8.78219604e-01 1.34837413e+00 -3.83712649e-01
5.80196619e-01 -5.47601223e-01 -7.72585630e-01 6.83183849e-01
6.85140431e-01 -4.05602098e-01 7.10739195e-01 -1.49314580e-02
-3.43353562e-02 -1.96137369e-01 -8.13028753e-01 7.17586637e-01
3.40924144e-01 -8.76430929e-01 -5.14211476e-01 2.21893281e-01
8.17814052e-01 -2.68005401e-01 -7.78963029e-01 4.26263869e-01
3.81604493e-01 -6.30724967e-01 7.36235261e-01 2.57007722e-02
-4.71558273e-01 -4.97363627e-01 -9.92540181e-01 -1.60583401e+00
1.85118929e-01 -7.19661117e-01 2.22592533e-01 1.61689126e+00
8.00406098e-01 -6.75224781e-01 6.90707788e-02 -1.86667722e-02
-5.22791088e-01 -1.16514079e-01 -1.25573516e+00 -9.44863737e-01
4.49749231e-02 -8.05535376e-01 2.46496364e-01 6.74875736e-01
-3.75116885e-01 6.97281957e-01 -6.13530397e-01 4.46445107e-01
5.88244140e-01 -7.86278546e-01 6.08892441e-01 -8.42220068e-01
-3.76956820e-01 -3.79002184e-01 -1.86620459e-01 -1.01542604e+00
4.18875962e-01 -1.06539488e+00 4.10890609e-01 -1.38824570e+00
-2.23697379e-01 -4.12687957e-01 -3.79317582e-01 2.40238443e-01
-3.65749858e-02 3.03678006e-01 3.12421530e-01 2.41405144e-02
-1.56847443e-02 5.82495451e-01 7.49738038e-01 6.75840005e-02
-3.50335181e-01 4.30073403e-02 -1.10077471e-01 8.00036669e-01
8.81579757e-01 -6.08494163e-01 -2.90229946e-01 -5.44285893e-01
-1.70313552e-01 4.51941520e-01 -3.44157487e-01 -1.10925245e+00
3.02122086e-01 3.16115558e-01 -4.70701382e-02 -8.07444036e-01
7.90125847e-01 -7.18357682e-01 4.46552992e-01 1.40904874e-01
1.21309884e-01 1.30768910e-01 4.81389344e-01 2.14518219e-01
-6.66373551e-01 -3.53883743e-01 9.34860528e-01 2.55932286e-02
-3.97340715e-01 -1.02900483e-01 -7.36071825e-01 -2.39908710e-01
6.18494332e-01 -9.57402065e-02 1.18201435e-01 -3.81132394e-01
-9.47722137e-01 -1.87451690e-01 -3.53395566e-02 6.29067600e-01
1.80385605e-01 -1.35425067e+00 -1.00730491e+00 2.89033353e-01
-9.54399556e-02 -3.21616948e-01 2.98331589e-01 9.29839015e-01
-3.30456197e-01 9.09060836e-01 -2.58080196e-02 -5.57573974e-01
-1.30683994e+00 2.71546900e-01 3.38101566e-01 -3.00941110e-01
-1.15490658e-02 9.13940310e-01 2.45568231e-01 -9.53651309e-01
4.65185434e-01 -2.25999579e-02 -1.06286205e-01 -2.37392522e-02
2.57784184e-02 3.95950794e-01 2.80540258e-01 -1.47542489e+00
-4.94376749e-01 4.41328704e-01 -2.05461189e-01 -9.88666236e-01
8.85651946e-01 -5.21103203e-01 4.64505292e-02 8.53650391e-01
1.21289110e+00 6.10053539e-01 -7.52765954e-01 -3.87674302e-01
3.08654755e-02 5.93373924e-02 3.36636364e-01 -1.02606845e+00
-6.72426343e-01 1.18408763e+00 8.33272517e-01 -1.52316168e-01
8.90473843e-01 -2.23702088e-01 8.73513579e-01 2.71131128e-01
2.20339105e-01 -1.40173602e+00 -6.58185244e-01 9.00769293e-01
8.44292283e-01 -1.29428732e+00 -6.86238587e-01 -9.22101513e-02
-7.71284342e-01 8.51365626e-01 3.61316115e-01 2.51716048e-01
7.97197044e-01 5.54577529e-01 5.77517569e-01 4.52519208e-01
-4.49497014e-01 -2.66619444e-01 3.22302312e-01 5.94841242e-01
7.55242944e-01 4.97539848e-01 -2.63633579e-01 5.86784184e-01
-5.43103099e-01 -5.59445739e-01 1.47695437e-01 4.46054846e-01
-2.07570195e-01 -1.56764174e+00 -7.54872620e-01 -2.40515098e-02
-7.82550156e-01 -5.95227897e-01 1.20478325e-01 5.98515749e-01
1.93723559e-01 1.21281576e+00 -1.89515084e-01 -3.14746559e-01
1.92665100e-01 6.06005371e-01 2.75214463e-01 -6.09711587e-01
-5.57793617e-01 1.02842093e+00 2.89661050e-01 -2.25220099e-02
-2.48701125e-01 -7.09048271e-01 -1.05603409e+00 -1.57743730e-02
-5.54376006e-01 2.41979152e-01 1.26045334e+00 8.61065984e-01
3.85780185e-02 5.74400544e-01 2.44684920e-01 -8.00632477e-01
-2.11702108e-01 -1.44003522e+00 -4.19915259e-01 -2.85164684e-01
2.10033342e-01 -2.80635953e-01 -3.76273304e-01 1.43113285e-01] | [14.397988319396973, 6.8668999671936035] |
fb7fdb81-30f7-47eb-8537-ec33ee89e706 | linear-mode-connectivity-and-the-lottery | 1912.05671 | null | https://arxiv.org/abs/1912.05671v4 | https://arxiv.org/pdf/1912.05671v4.pdf | Linear Mode Connectivity and the Lottery Ticket Hypothesis | We study whether a neural network optimizes to the same, linearly connected minimum under different samples of SGD noise (e.g., random data order and augmentation). We find that standard vision models become stable to SGD noise in this way early in training. From then on, the outcome of optimization is determined to a linearly connected region. We use this technique to study iterative magnitude pruning (IMP), the procedure used by work on the lottery ticket hypothesis to identify subnetworks that could have trained in isolation to full accuracy. We find that these subnetworks only reach full accuracy when they are stable to SGD noise, which either occurs at initialization for small-scale settings (MNIST) or early in training for large-scale settings (ResNet-50 and Inception-v3 on ImageNet). | ['Daniel M. Roy', 'Michael Carbin', 'Gintare Karolina Dziugaite', 'Jonathan Frankle'] | 2019-12-11 | null | https://proceedings.icml.cc/static/paper_files/icml/2020/5787-Paper.pdf | https://proceedings.icml.cc/static/paper_files/icml/2020/5787-Paper.pdf | icml-2020-1 | ['linear-mode-connectivity'] | ['knowledge-base'] | [ 2.96438992e-01 3.49633515e-01 2.32284248e-01 -3.02500010e-01
-1.95094079e-01 -6.12498820e-01 6.07122421e-01 -1.98362514e-01
-7.91323960e-01 7.64302313e-01 -8.26775804e-02 -3.62492830e-01
-2.41227791e-01 -7.24322140e-01 -9.52325881e-01 -6.07950568e-01
-1.41179875e-01 5.03341973e-01 4.49915767e-01 -1.53505104e-02
1.10379159e-01 6.34196281e-01 -1.46919370e+00 1.85599163e-01
4.33222771e-01 5.90810418e-01 1.64578110e-01 9.57316875e-01
1.70358747e-01 5.91271460e-01 -8.13751101e-01 -3.10294062e-01
8.46466780e-01 -2.70851523e-01 -8.86166394e-01 1.24254048e-01
9.49702561e-01 -1.79255679e-01 -4.30470020e-01 1.39556575e+00
4.95395243e-01 8.71391445e-02 3.54105264e-01 -1.12699890e+00
-3.03038865e-01 8.57142270e-01 -5.99398255e-01 4.08134371e-01
-2.62690544e-01 6.08049154e-01 8.39784801e-01 -7.78142571e-01
7.53313601e-01 1.30109346e+00 1.04175866e+00 4.24017817e-01
-1.88203454e+00 -5.08994937e-01 3.03228796e-01 -2.56106317e-01
-1.29113483e+00 -7.44529128e-01 3.52725029e-01 -3.27953160e-01
1.24007297e+00 9.22013223e-02 8.45681131e-01 9.64725077e-01
1.36475533e-01 4.90321338e-01 1.09295964e+00 -2.90801048e-01
3.06740195e-01 -1.30279642e-02 4.02920455e-01 8.23787212e-01
7.08672404e-01 2.59514630e-01 -3.66489142e-01 -3.80846001e-02
1.04038799e+00 -4.97008950e-01 -3.32951158e-01 -3.54960039e-02
-9.26581085e-01 7.45830059e-01 7.35455811e-01 1.82670206e-01
-4.16290134e-01 3.07655543e-01 2.44444795e-02 7.08944380e-01
2.81854481e-01 1.07206082e+00 -8.07541907e-01 8.36898163e-02
-1.04255772e+00 7.69687295e-02 1.03827417e+00 7.22705960e-01
8.46124470e-01 1.83312446e-01 6.76559210e-02 8.80004227e-01
2.57224981e-02 -1.95009857e-02 3.26102674e-01 -1.26481843e+00
3.50133866e-01 5.16812563e-01 -3.00419003e-01 -6.42595112e-01
-6.80115581e-01 -8.70177507e-01 -9.33667719e-01 6.95070684e-01
7.89173424e-01 -5.95203340e-01 -1.31006742e+00 1.93350494e+00
9.67056379e-02 4.71156120e-01 1.76681280e-01 6.16061330e-01
7.42949903e-01 3.45718026e-01 -1.84431717e-01 -3.45908962e-02
9.82661724e-01 -7.86329985e-01 2.43731774e-02 -6.31272614e-01
5.82586825e-01 -4.55134660e-01 1.02192938e+00 3.88638794e-01
-1.19141328e+00 -5.06145775e-01 -1.21306813e+00 2.43692890e-01
-2.15001449e-01 -4.09898125e-02 5.49692571e-01 6.62278950e-01
-1.53529298e+00 1.22810149e+00 -9.49163556e-01 -4.20664608e-01
7.66860783e-01 6.75713241e-01 -2.97161400e-01 1.69755630e-02
-8.78043830e-01 9.60858941e-01 5.67348063e-01 5.97146265e-02
-1.01882839e+00 -6.26522183e-01 -5.25645733e-01 1.73930079e-02
2.47355714e-01 -9.08342242e-01 1.18226898e+00 -1.23193645e+00
-1.10752380e+00 1.00748706e+00 -1.22338003e-02 -9.99054670e-01
5.97914279e-01 -4.53813560e-02 2.04835966e-01 -3.05640213e-02
-2.75588371e-02 1.08934104e+00 1.05523658e+00 -1.36486971e+00
-3.99470866e-01 -5.13660550e-01 2.60576814e-01 3.91827524e-01
-1.58277556e-01 -1.07792124e-01 -4.19705927e-01 -4.39046174e-01
5.54519951e-01 -1.01447642e+00 -5.04718721e-01 -1.48352832e-01
-7.92436838e-01 -2.90628579e-02 3.59907925e-01 -2.83328205e-01
7.77171969e-01 -1.73696768e+00 -1.12640068e-01 6.40087128e-01
6.05380714e-01 2.68192023e-01 -2.92966455e-01 -1.91400141e-01
-2.64745116e-01 5.90791523e-01 -1.56782717e-01 -4.86367226e-01
-4.46973860e-01 4.08170223e-01 7.63022453e-02 6.00448012e-01
3.41188878e-01 8.54855955e-01 -7.02486396e-01 -4.14947510e-01
6.14506938e-03 -3.90066132e-02 -6.46284223e-01 -1.82659090e-01
-7.75256231e-02 1.58453166e-01 -1.42095730e-01 4.81580675e-01
5.14516652e-01 -4.55375046e-01 -9.69212502e-02 -1.37213320e-01
-9.60523486e-02 1.70055598e-01 -1.21932197e+00 1.19405425e+00
-1.52699307e-01 8.34563732e-01 3.54244500e-01 -9.04256403e-01
5.97848535e-01 -6.89065605e-02 1.30914718e-01 -3.06482285e-01
2.20707148e-01 1.36971414e-01 4.85446066e-01 -7.79432431e-02
2.16283917e-01 -8.34797621e-02 5.03005087e-01 2.34508649e-01
4.00219738e-01 -1.50071025e-01 3.10555130e-01 1.33961797e-01
1.66058099e+00 -4.20760602e-01 3.86977457e-02 -5.15039265e-01
1.33282170e-02 2.21459150e-01 3.40995222e-01 1.49282861e+00
-2.67274797e-01 8.62188637e-01 7.24363625e-01 -2.38872260e-01
-1.30062532e+00 -8.46994460e-01 -3.21348757e-01 9.52419698e-01
-1.22626359e-02 5.40434383e-02 -8.74756217e-01 -6.04477704e-01
-1.07500590e-01 6.92963481e-01 -6.22336447e-01 -4.13374960e-01
-5.92954576e-01 -8.78569126e-01 7.21376181e-01 3.93457651e-01
8.14124703e-01 -1.21344781e+00 -3.08201104e-01 1.85383961e-01
4.00420070e-01 -9.86901343e-01 -1.99514832e-02 8.37594807e-01
-1.24931169e+00 -1.13530922e+00 -6.41152918e-01 -8.96678329e-01
1.01766574e+00 -1.24068763e-02 1.38328874e+00 4.00477916e-01
-6.62266761e-02 2.23997191e-01 1.19191967e-01 -2.83288032e-01
-3.60619396e-01 3.24869096e-01 7.80392513e-02 -3.65399748e-01
7.87565783e-02 -7.06518888e-01 -5.65550566e-01 3.19025069e-01
-5.53158402e-01 -2.79007666e-02 6.85213149e-01 7.47591019e-01
5.96791863e-01 4.16042507e-01 2.82475322e-01 -9.85313237e-01
7.83013403e-01 -3.73148203e-01 -6.98348165e-01 7.23453164e-02
-7.00177252e-01 1.84459254e-01 7.09101081e-01 -6.95517600e-01
-5.15636742e-01 1.07915498e-01 -1.44768059e-01 -7.73839414e-01
-4.58997250e-01 4.42820460e-01 1.95096917e-02 -4.67255265e-01
1.19015157e+00 7.59673864e-02 6.83499277e-02 -4.54708338e-01
1.72828183e-01 4.01928611e-02 6.08289540e-01 -5.03645241e-01
1.02461421e+00 3.70866001e-01 4.11173999e-02 -8.49856734e-01
-8.04830372e-01 -7.98272863e-02 -5.67745805e-01 -5.33053800e-02
5.83233774e-01 -7.64516056e-01 -5.48159242e-01 6.32674456e-01
-1.03019810e+00 -7.27019072e-01 -8.10075939e-01 3.53590816e-01
-2.91997135e-01 -1.25017062e-01 -4.35791105e-01 -5.76075673e-01
-7.76282623e-02 -9.99544501e-01 6.91573799e-01 4.62011993e-01
-1.76234320e-01 -8.91182125e-01 -8.55059475e-02 -1.53659716e-01
1.50525287e-01 4.18348610e-02 7.21073270e-01 -9.21907842e-01
-5.55545330e-01 -1.62859317e-02 -3.40222001e-01 6.11766398e-01
-2.12157890e-01 5.05587645e-02 -8.67365777e-01 -4.47874844e-01
1.23827577e-01 -5.76946914e-01 1.27380478e+00 8.52741361e-01
1.07624006e+00 -4.60622013e-01 -2.77207524e-01 9.30203497e-01
1.50606585e+00 -2.82539427e-02 5.65006435e-01 5.43333769e-01
5.06428540e-01 2.97179043e-01 -1.30883187e-01 -1.05382137e-01
-2.13006943e-01 1.41861126e-01 5.57869136e-01 -2.88618416e-01
-3.38446259e-01 -1.42564550e-01 1.27242208e-01 1.92846134e-01
1.66522060e-03 -1.79927304e-01 -8.42630386e-01 6.21527195e-01
-1.48483074e+00 -8.76373649e-01 -1.55446112e-01 2.23174596e+00
1.04828548e+00 1.03767264e+00 -2.96280980e-02 -4.54670228e-02
7.92454541e-01 1.19978555e-01 -8.00115883e-01 -2.74038732e-01
-4.39328581e-01 2.01992035e-01 9.45827246e-01 6.05118632e-01
-1.18787909e+00 1.18619478e+00 7.33562374e+00 7.67053246e-01
-8.14416170e-01 -1.12240084e-01 9.95109439e-01 -3.60441595e-01
9.63870585e-02 1.52897105e-01 -1.06795979e+00 2.01531440e-01
7.32739508e-01 1.32444069e-01 7.75439918e-01 9.13542271e-01
1.25004604e-01 -2.76156217e-01 -1.11330116e+00 6.44347131e-01
-3.14458609e-01 -1.42344713e+00 -1.58517703e-01 8.63643065e-02
1.10631776e+00 4.98929054e-01 1.71810593e-02 1.66658893e-01
1.02349401e+00 -1.10942900e+00 5.53873122e-01 1.95188031e-01
6.60638452e-01 -5.05550683e-01 6.10707164e-01 5.96391499e-01
-8.19921494e-01 3.32277790e-02 -7.79434085e-01 -1.01343906e-02
-9.23905745e-02 7.20285773e-01 -1.09202850e+00 -1.94273159e-01
5.62421322e-01 4.07428890e-01 -8.94617260e-01 1.22080016e+00
-3.22307438e-01 9.32864845e-01 -9.54664052e-01 3.63333300e-02
3.21667492e-01 -1.31026536e-01 8.46441448e-01 8.62991512e-01
-3.06650866e-02 -2.54761074e-02 4.64077219e-02 9.32925582e-01
-3.99023652e-01 -4.47987378e-01 -8.55391264e-01 1.56516820e-01
3.87185723e-01 1.02186441e+00 -1.13429773e+00 -3.41416478e-01
-1.74871132e-01 7.17093587e-01 4.55360323e-01 5.87829590e-01
-4.77589160e-01 -2.27090567e-01 6.64343178e-01 1.61477387e-01
4.36366767e-01 -1.60496473e-01 -8.54795754e-01 -1.01061189e+00
-2.75677145e-01 -6.87513649e-01 1.17731430e-01 -7.53819644e-01
-1.34790862e+00 6.18124485e-01 -6.60784692e-02 -6.30681813e-01
2.03674883e-02 -6.31707847e-01 -9.35936630e-01 6.93414509e-01
-8.95062268e-01 -5.62667429e-01 -8.26975107e-02 6.58828855e-01
5.26828468e-01 -1.77346259e-01 3.30609798e-01 -4.04265858e-02
-6.83284044e-01 7.48824596e-01 2.82966763e-01 3.54034483e-01
1.51548237e-01 -1.43398738e+00 7.64815271e-01 1.04184520e+00
4.65119481e-01 8.91809821e-01 7.91174829e-01 -8.83979321e-01
-7.17289209e-01 -1.17246544e+00 5.56052148e-01 -3.74298871e-01
6.39534116e-01 -2.06161797e-01 -8.46580803e-01 7.72360742e-01
-3.44484001e-02 -1.01897225e-01 -3.64344306e-02 2.80699968e-01
3.34912748e-03 6.30896538e-02 -1.23180485e+00 9.88385797e-01
1.45718920e+00 -1.65833548e-01 -6.68006361e-01 4.39259022e-01
6.74705029e-01 -3.84373218e-01 -5.19347429e-01 3.74801517e-01
1.85493678e-01 -8.98643792e-01 1.09469402e+00 -9.20992672e-01
3.85358661e-01 -1.47200108e-01 8.79640356e-02 -1.46674967e+00
-3.14943641e-01 -7.82163024e-01 -4.60382029e-02 9.26778734e-01
7.58851171e-01 -8.58429849e-01 1.02299416e+00 5.86210191e-01
-8.36396590e-02 -9.14720237e-01 -1.04822648e+00 -8.75685394e-01
-4.07578014e-02 -3.63359511e-01 2.27814242e-01 7.82613873e-01
-7.62603045e-01 5.46974301e-01 3.16736251e-02 1.20253876e-01
8.19371462e-01 -1.91665381e-01 7.04020083e-01 -1.38595283e+00
-3.34916562e-01 -7.55510032e-01 -3.46301258e-01 -1.22501028e+00
-7.32566342e-02 -8.35284352e-01 6.01451993e-02 -1.43849218e+00
2.80878451e-02 -4.99802679e-01 -1.54570147e-01 6.11142576e-01
-1.70340061e-01 3.64640981e-01 1.11649781e-01 3.57715458e-01
-4.12436455e-01 9.31941122e-02 1.33981776e+00 3.03123146e-02
-4.70702052e-01 3.07786912e-01 -8.70557427e-01 1.15321445e+00
9.54985559e-01 -7.42900193e-01 -3.68255287e-01 -3.37683737e-01
1.51966065e-01 -4.02837723e-01 5.37297785e-01 -1.24884236e+00
2.53251314e-01 6.34893849e-02 7.62422800e-01 -4.05041009e-01
1.78275451e-01 -7.48916805e-01 -1.04964167e-01 5.15637815e-01
-5.81929922e-01 -6.52532578e-02 3.51738602e-01 4.43514168e-01
2.58808643e-01 -7.84639359e-01 1.20507705e+00 -5.40824175e-01
-3.93218488e-01 2.89003909e-01 -3.78871322e-01 5.17112851e-01
6.42351925e-01 -7.42516756e-01 -4.16132927e-01 -2.49165654e-01
-1.12924206e+00 3.80991668e-01 6.48491085e-01 -1.06741950e-01
4.92673278e-01 -8.83953154e-01 -6.95013702e-01 3.59081000e-01
-5.54509699e-01 4.79717970e-01 -2.73517966e-01 7.46389568e-01
-5.76574445e-01 -1.47822857e-01 4.88229515e-03 -7.20136642e-01
-1.23780000e+00 1.36024291e-02 8.51716816e-01 -4.97982800e-01
-7.32384205e-01 1.46572816e+00 1.41125679e-01 -2.10977077e-01
4.76514786e-01 -4.05335426e-01 -9.81310904e-02 -1.09908588e-01
1.49381638e-01 1.45861611e-01 1.66486815e-01 -2.70497978e-01
-1.45324856e-01 3.22342992e-01 -1.69630349e-01 -2.76501715e-01
1.41418791e+00 5.68889044e-02 7.55628198e-02 2.88517326e-01
1.31765997e+00 -3.88077676e-01 -1.55579638e+00 -1.91457748e-01
-1.06055126e-01 -3.07496250e-01 1.77026302e-01 -6.36066973e-01
-1.20854807e+00 6.11222148e-01 5.62505782e-01 4.70526934e-01
8.84254158e-01 7.07689151e-02 3.36584955e-01 8.16729724e-01
2.52218068e-01 -1.12980998e+00 8.41771886e-02 9.03356910e-01
8.68178606e-01 -1.14095652e+00 2.73152798e-01 9.63464156e-02
-3.12479824e-01 8.90100360e-01 1.00888443e+00 -5.74399292e-01
8.17162991e-01 4.41755861e-01 -2.89328128e-01 -4.28255677e-01
-8.31415534e-01 -2.85970211e-01 4.34874408e-02 6.47327304e-01
-2.97401726e-01 -3.50884229e-01 2.76072681e-01 1.05507270e-01
-5.34037352e-01 -2.35974565e-01 4.47743684e-01 8.20375681e-01
-7.85909474e-01 -3.87317985e-01 -3.35297972e-01 1.09744477e+00
-3.15260053e-01 -4.77749914e-01 -8.49966228e-01 8.91645193e-01
6.94117025e-02 7.72118807e-01 2.11308315e-01 -5.11428475e-01
3.33160162e-01 1.15648180e-01 5.45252860e-01 -7.50838637e-01
-8.49014103e-01 8.74432921e-02 2.06144676e-01 -5.19470692e-01
-5.36446236e-02 -5.52297950e-01 -1.10393500e+00 -3.66496623e-01
-7.55463719e-01 -2.05037579e-01 4.22662705e-01 8.96571279e-01
1.16746746e-01 5.41883826e-01 4.00596440e-01 -9.91411805e-01
-7.34407008e-01 -1.05839908e+00 -4.37691957e-01 1.59006014e-01
3.46529335e-01 -3.73341024e-01 -9.24267292e-01 -4.43544574e-02] | [8.5222806930542, 3.326758861541748] |
9919dfe6-c689-4e16-a0b7-4ea8bf82313a | unsupervised-concept-to-text-generation-with | null | null | https://aclanthology.org/N12-1093 | https://aclanthology.org/N12-1093.pdf | Unsupervised Concept-to-text Generation with Hypergraphs | null | ['Ioannis Konstas', 'Mirella Lapata'] | 2012-06-01 | null | null | null | naacl-2012-6 | ['concept-to-text-generation'] | ['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] | [-7.280180931091309, 3.7802298069000244] |
830d1f62-e509-494d-9b50-7bc50d79af16 | fully-connected-neural-network-with-advance | null | null | https://aclanthology.org/W18-4404 | https://aclanthology.org/W18-4404.pdf | Fully Connected Neural Network with Advance Preprocessor to Identify Aggression over Facebook and Twitter | Paper presents the different methodologies developed {\&} tested and discusses their results, with the goal of identifying the best possible method for the aggression identification problem in social media. | ['Teresa Gon{\\c{c}}alves', 'Vitor Beires Nogueira', 'Paulo Quaresma', 'Kashyap Raiyani'] | 2018-08-01 | null | null | null | coling-2018-8 | ['aggression-identification'] | ['natural-language-processing'] | [-3.80740047e-01 -4.08089757e-02 2.97317266e-01 -2.02551156e-01
1.39672965e-01 5.55699319e-02 3.04801345e-01 4.00977939e-01
-7.56942749e-01 6.14088476e-01 -4.87583205e-02 2.96353221e-01
-8.31297755e-01 -5.35080016e-01 5.46590447e-01 -5.15668213e-01
-4.99450266e-01 7.60798872e-01 3.00834119e-01 -5.25279522e-01
9.49349284e-01 7.38894761e-01 -1.82542884e+00 2.15911210e-01
2.71859437e-01 2.22745001e-01 7.69650042e-02 8.98907661e-01
-1.65958419e-01 1.13762498e+00 -1.07875299e+00 -1.07175875e+00
-6.27151579e-02 -6.03369534e-01 -9.53104436e-01 -1.23074025e-01
1.98674202e-01 -6.01525545e-01 -2.59542435e-01 1.20973611e+00
6.04268610e-01 4.54976350e-01 8.13094258e-01 -1.25799537e+00
-6.42236620e-02 8.42077792e-01 -5.11012852e-01 1.04319298e+00
1.04429567e+00 -2.90577233e-01 7.41948366e-01 -4.25816655e-01
6.32907867e-01 1.34304249e+00 3.89050603e-01 9.93562281e-01
-5.89159667e-01 -1.03555536e+00 -6.69123292e-01 3.93936962e-01
-1.54580569e+00 -2.79178977e-01 6.85598195e-01 -6.60710931e-01
1.22263026e+00 3.30479503e-01 8.28065574e-01 1.33017731e+00
-4.17025775e-01 4.33237433e-01 9.22680855e-01 -2.77093560e-01
-9.64150950e-03 3.82085800e-01 8.59333694e-01 7.33276188e-01
3.15823019e-01 -1.76201001e-01 -6.56028986e-01 -8.22115719e-01
3.37442398e-01 -6.60577059e-01 4.63969707e-01 5.99249244e-01
1.36627078e-01 1.02127075e+00 -6.06070101e-01 8.80069077e-01
-1.55873299e-01 -4.48385775e-01 8.59314919e-01 1.73734903e-01
3.89291286e-01 7.70535409e-01 2.01930568e-01 -1.14212561e+00
-5.46192348e-01 8.28666806e-01 6.89397693e-01 7.43716478e-01
-3.20407897e-01 3.16642404e-01 1.66012526e-01 1.30886412e+00
4.03057039e-01 2.86351770e-01 4.12302226e-01 -8.75500262e-01
4.12718505e-01 5.61570168e-01 -1.96626157e-01 -1.03250921e+00
-7.05295086e-01 1.11791901e-01 8.22286233e-02 1.98612273e-01
3.75270367e-01 -3.55915666e-01 2.17705995e-01 1.04289579e+00
2.32121602e-01 -6.31881058e-02 -3.58389199e-01 1.94051847e-01
1.13225377e+00 2.05548987e-01 5.58936238e-01 -6.54575467e-01
7.95068502e-01 -6.80927873e-01 -1.06296790e+00 2.35783547e-01
8.42448890e-01 -5.84004760e-01 4.09381956e-01 6.85530186e-01
-1.59397709e+00 -5.03118569e-03 -6.89991832e-01 3.23481947e-01
-4.82048780e-01 -5.43437123e-01 5.24856508e-01 1.37233293e+00
-8.76380622e-01 4.44599271e-01 -2.03134283e-01 -5.98009944e-01
7.13909566e-02 7.32963443e-01 -3.68255287e-01 7.35196173e-01
-7.73969650e-01 9.52149153e-01 2.87359774e-01 -2.58142769e-01
-2.46353492e-01 -8.02510604e-02 -3.36565614e-01 -1.89380899e-01
-1.41415879e-01 1.29553124e-01 9.64751244e-01 -6.21389747e-01
-1.52738297e+00 1.15304947e+00 4.19664443e-01 -1.02531776e-01
5.97927630e-01 -4.30930369e-02 -7.40033448e-01 2.96835870e-01
2.97302809e-02 -2.14868188e-01 5.67733765e-01 -6.58104062e-01
-1.00037682e+00 -7.12752879e-01 -1.41972676e-01 1.54844016e-01
-9.78828967e-01 1.36177361e+00 1.46205381e-01 -3.36690485e-01
-5.31419754e-01 -4.18932319e-01 1.08118849e-02 -6.96895838e-01
-6.32013306e-02 -7.16261148e-01 7.31596947e-01 -6.81724608e-01
1.84324920e+00 -2.28739929e+00 4.12934333e-01 6.08961344e-01
3.20808470e-01 8.97524536e-01 5.02729118e-01 1.13426161e+00
-2.37193272e-01 4.16994780e-01 6.80522025e-02 -5.27894497e-01
-7.71827996e-02 2.05241024e-01 2.25565121e-01 4.54618663e-01
-5.62339187e-01 -2.06317723e-01 -8.03371310e-01 -8.54242265e-01
4.09412503e-01 -1.70595031e-02 -3.01346660e-01 5.11073828e-01
6.00893140e-01 1.41283587e-01 -6.39124453e-01 2.25975722e-01
3.83844048e-01 9.34674263e-01 3.20415348e-02 9.51691747e-01
-4.83377308e-01 -5.75848408e-02 -1.28109741e+00 4.73360807e-01
1.73713505e-01 4.77703929e-01 3.17337036e-01 -6.85071945e-01
1.12155533e+00 5.92983842e-01 7.48122752e-01 -3.90527308e-01
7.77554870e-01 3.35557997e-01 3.54183495e-01 -1.27366531e+00
2.66660005e-01 6.52171001e-02 6.85651451e-02 3.28804761e-01
3.70553613e-01 9.02712159e-03 6.43348932e-01 -1.32402927e-01
1.24396491e+00 -6.48845971e-01 5.55808663e-01 -5.02275079e-02
9.04258490e-01 -3.44987601e-01 4.05880548e-02 8.95073771e-01
-6.94943011e-01 8.71131942e-03 7.17224360e-01 -1.88653529e-01
-6.69333756e-01 -6.27642691e-01 -4.63941842e-01 1.19964266e+00
-3.22453022e-01 -7.17955768e-01 -1.18320680e+00 -6.44471049e-01
-1.05921485e-01 7.68588603e-01 -7.21123576e-01 1.01077236e-01
-4.78097111e-01 -6.27504945e-01 1.11270154e+00 -1.58359874e-02
2.65004665e-01 -1.29378426e+00 -2.03664392e-01 9.59155187e-02
2.93136775e-01 -9.18676257e-01 4.73414063e-01 -4.62433696e-01
-6.15155458e-01 -1.18309283e+00 -2.38962591e-01 -5.17665982e-01
3.83529514e-01 7.51142874e-02 2.87164569e-01 6.65611744e-01
-3.82711202e-01 4.24190521e-01 -8.06749046e-01 -4.94388729e-01
-4.58146930e-01 -7.33312219e-02 1.98624462e-01 -2.58857787e-01
1.01726663e+00 -7.48341620e-01 1.64935261e-01 1.36698097e-01
-4.04351354e-01 -7.90208220e-01 -4.36173648e-01 2.95682847e-01
-6.94247842e-01 4.54126835e-01 4.84376162e-01 -9.77830052e-01
1.46262872e+00 -7.75520205e-01 -1.82991087e-01 -4.76580411e-01
-2.35074863e-01 -8.87967706e-01 4.16245610e-01 -1.10807613e-01
-6.70495749e-01 -7.83645868e-01 -1.10771751e+00 1.49207711e-01
-5.94045997e-01 -4.98777218e-02 4.00831819e-01 -5.69544673e-01
8.20941091e-01 -2.62755752e-01 -1.91350072e-03 -6.57408595e-01
-7.28704751e-01 1.12883794e+00 -2.94150710e-01 -1.82171553e-01
2.61454523e-01 -6.49751723e-02 1.18865728e-01 -1.64544630e+00
-2.85668612e-01 -9.99192119e-01 -4.89894927e-01 -9.37950671e-01
9.46581542e-01 -7.89217502e-02 -1.16115022e+00 8.35351884e-01
-8.17313790e-01 3.30660760e-01 5.34736961e-02 6.25194073e-01
-5.77851713e-01 5.59820414e-01 -8.15434515e-01 -1.64520836e+00
-3.19413543e-01 -8.39724660e-01 3.38944227e-01 5.38843237e-02
-7.87009954e-01 -7.95063734e-01 5.52274942e-01 8.15699697e-01
-1.22397698e-01 -9.32148769e-02 7.48552561e-01 -1.33627832e+00
9.52530563e-01 -5.87548077e-01 1.85060903e-01 5.75232148e-01
-3.52617115e-01 7.68790662e-01 -6.80930853e-01 1.85124353e-01
1.38750002e-01 5.91147598e-03 -1.08078711e-01 -1.79156303e-01
1.10752118e+00 -5.50554320e-02 -1.53586119e-01 -1.91256136e-01
6.85124993e-01 7.82328308e-01 5.62440693e-01 5.43762863e-01
3.08303028e-01 7.57013559e-01 4.73957062e-01 8.28979552e-01
-3.07837248e-01 6.54371679e-01 2.20429361e-01 6.29637837e-01
1.24091536e-01 2.36187667e-01 1.99312210e-01 7.69134343e-01
-9.27401781e-01 -3.71930331e-01 -1.11190844e+00 2.18169898e-01
-1.54977679e+00 -1.68464220e+00 -5.63186347e-01 1.66480112e+00
5.17759472e-02 6.22482970e-02 1.39231396e+00 7.64803112e-01
7.55648196e-01 5.03603593e-02 4.86356586e-01 -1.33008242e+00
2.06799343e-01 4.91569191e-01 -1.31610096e-01 6.86745644e-01
-6.95687532e-01 7.15585113e-01 8.23743820e+00 8.16929817e-01
-5.02112389e-01 1.34540439e-01 -2.45179713e-01 -3.12488914e-01
6.67655230e-01 -7.22013772e-01 -7.68750429e-01 6.40728474e-01
1.23028719e+00 -2.88482845e-01 4.94976103e-01 7.20106363e-01
3.13233644e-01 -2.48359039e-01 -4.55673218e-01 8.50875735e-01
4.39347416e-01 -3.58505398e-01 -3.59002203e-02 3.92389655e-01
-4.52912971e-02 -2.68671453e-01 7.85559509e-03 3.32912803e-01
-4.04697880e-02 -4.96663839e-01 3.61606598e-01 4.64820266e-01
-2.29503840e-01 -9.83688653e-01 9.36245620e-01 5.14022112e-01
-4.57673192e-01 -6.73687220e-01 -7.14402869e-02 -5.77425063e-01
-4.85570766e-02 -5.08152008e-01 -1.02820671e+00 9.76600349e-02
9.11206365e-01 5.31886518e-01 -7.28212655e-01 1.32406175e+00
8.82371143e-02 7.35918760e-01 -4.10236031e-01 -1.02574611e+00
2.11187825e-01 -5.84400892e-01 1.11984026e+00 1.46378481e+00
-7.90661648e-02 4.63866562e-01 -3.66195053e-01 9.55990851e-02
7.65847206e-01 9.05218542e-01 -8.62602234e-01 -3.44613910e-01
2.09486216e-01 9.89630282e-01 -4.83924985e-01 5.98360635e-02
-3.93094659e-01 2.38543853e-01 7.78448820e-01 -2.36934558e-01
-6.46386981e-01 -2.91258276e-01 6.53921664e-01 6.00656271e-01
-3.78701836e-01 -3.17586094e-01 -1.43958554e-01 -3.37206066e-01
-3.78818154e-01 -8.61678004e-01 9.47661161e-01 -4.01086867e-01
-1.39866674e+00 4.98568177e-01 8.67526293e-01 -1.01487029e+00
-3.26673478e-01 -6.73622251e-01 -6.35872602e-01 4.05836433e-01
3.50919180e-02 -6.74557447e-01 3.26969475e-01 4.67545509e-01
5.70972979e-01 -8.39998245e-01 8.01848114e-01 6.14960313e-01
-1.11942315e+00 5.07975876e-01 -1.62937656e-01 5.69608994e-02
1.19674765e-01 -9.52917933e-01 -3.60445321e-01 3.71361464e-01
-5.48263609e-01 5.14262855e-01 1.02688587e+00 -5.57086885e-01
-7.08156765e-01 -1.14459120e-01 9.33602035e-01 -4.19765353e-01
1.20191002e+00 -2.32496426e-01 -4.85892832e-01 3.14691663e-01
2.45243013e-01 -1.03378868e+00 1.34583592e+00 3.71397644e-01
7.44642019e-02 2.15905532e-01 -1.57160521e+00 6.93818212e-01
1.42221045e+00 -2.31080666e-01 -7.20827401e-01 4.97867584e-01
-3.75142992e-01 -1.92425907e-01 -8.31750035e-01 -6.63217083e-02
5.23408651e-01 -1.35364759e+00 9.42132652e-01 -1.01279604e+00
3.76631051e-01 7.40920782e-01 1.45066798e-01 -7.96308279e-01
-4.42723781e-01 -4.98002380e-01 2.36502036e-01 1.21506035e+00
3.18652272e-01 -7.81456828e-01 5.00300765e-01 8.13867271e-01
1.15492605e-02 -4.40438122e-01 -1.34899604e+00 -5.08929670e-01
-7.50386715e-03 -5.88621736e-01 2.89168954e-01 1.04736733e+00
9.70288992e-01 1.32852733e-01 -4.00444299e-01 -1.24198250e-01
4.90840584e-01 -1.17428482e+00 7.22823560e-01 -1.16510522e+00
-5.23743294e-02 -1.01693094e+00 -1.25269151e+00 1.26857609e-01
3.83139819e-01 -8.29289779e-02 -7.68577993e-01 -8.06601346e-01
1.68903425e-01 1.24111444e-01 1.13390774e-01 4.94301952e-02
6.95383549e-02 2.89947420e-01 2.20075413e-03 -4.21000540e-01
-3.36641937e-01 4.03866768e-02 7.62303591e-01 4.64681268e-01
-2.61965543e-01 1.69085532e-01 -4.31941152e-01 1.15725851e+00
8.48947108e-01 -6.48986936e-01 -4.04748321e-01 4.02940772e-02
4.50656056e-01 2.07210362e-01 2.02588215e-02 -1.28240073e+00
1.86352432e-01 -5.07190585e-01 -2.61883944e-01 -3.81382823e-01
6.80375636e-01 -7.28836358e-01 2.86929775e-02 -3.48848850e-02
-3.36066663e-01 3.62821370e-01 -8.94571170e-02 -3.64062972e-02
-1.21677980e-01 -1.40559185e+00 7.79314160e-01 -1.18041262e-01
-3.73898447e-01 6.48434684e-02 -1.18412566e+00 -9.75755509e-03
1.41346037e+00 -2.98788816e-01 -1.39386386e-01 -4.55645978e-01
-1.04102743e+00 2.66547669e-02 -6.56031966e-02 3.48095715e-01
4.55599636e-01 -5.13378143e-01 -6.77185297e-01 -7.94002041e-02
1.74963903e-02 -1.11794865e+00 4.88419384e-01 7.10458100e-01
-9.71941292e-01 -1.07612640e-01 -5.97293854e-01 2.47193128e-01
-2.25106382e+00 5.17981529e-01 5.03603637e-01 3.11871678e-01
-3.25759441e-01 8.91922832e-01 -7.02656031e-01 1.04481220e-01
8.34759399e-02 9.71016169e-01 -1.35991323e+00 -3.79876494e-02
9.32142019e-01 1.42853117e+00 -1.11897945e-01 -1.22954118e+00
7.85636902e-03 7.98712969e-02 -2.33996123e-01 -3.01089644e-01
1.17504311e+00 -5.47038168e-02 -7.32145846e-01 4.88806367e-01
1.02836347e+00 3.29852998e-01 3.66635650e-01 4.13143039e-01
9.33691561e-02 -9.77700770e-01 2.59829424e-02 -1.85914680e-01
-5.66982031e-01 4.75641608e-01 4.97611582e-01 1.24763858e+00
6.81310058e-01 -1.18416518e-01 2.12779105e-01 4.77515012e-01
3.11685890e-01 -1.67911839e+00 -2.25096062e-01 6.50254488e-01
9.70847785e-01 -7.95161963e-01 2.21708119e-01 -9.35345352e-01
-5.51286757e-01 1.33294761e+00 9.35560226e-01 -4.21985269e-01
1.02040148e+00 1.20903917e-01 -2.24879026e-01 -8.00648570e-01
-2.06710979e-01 -1.47760361e-01 -1.29920021e-01 9.44814563e-01
4.94622350e-01 -4.31730896e-02 -1.46983981e+00 6.09935701e-01
-4.92438436e-01 -3.43983680e-01 5.90368927e-01 1.05919743e+00
-4.24551308e-01 -1.15656924e+00 -4.50825155e-01 6.94394171e-01
-8.01836312e-01 4.91764337e-01 -1.41020334e+00 1.15984178e+00
2.05216408e-01 1.30081749e+00 -2.40982801e-01 -6.82493508e-01
7.86515772e-01 8.31580837e-04 3.69943142e-01 -8.10538828e-01
-1.29997194e+00 -2.26839855e-01 1.01917148e+00 -2.20033035e-01
-4.40814972e-01 -1.34867692e+00 -8.71386290e-01 -7.46262789e-01
-3.49541128e-01 6.59488797e-01 3.43555450e-01 1.00515711e+00
-3.71600986e-01 3.42258904e-03 7.99700677e-01 -4.89750385e-01
-2.15983465e-02 -1.20112860e+00 -1.13754201e+00 6.40643239e-01
-1.84923545e-01 -1.06410193e+00 -7.71804392e-01 -7.16875315e-01] | [8.76432991027832, 10.742283821105957] |
dac71238-a74b-4a35-a190-fa5eeda0018b | an-anchor-free-detector-for-continuous-speech | 2208.04622 | null | https://arxiv.org/abs/2208.04622v1 | https://arxiv.org/pdf/2208.04622v1.pdf | An Anchor-Free Detector for Continuous Speech Keyword Spotting | Continuous Speech Keyword Spotting (CSKWS) is a task to detect predefined keywords in a continuous speech. In this paper, we regard CSKWS as a one-dimensional object detection task and propose a novel anchor-free detector, named AF-KWS, to solve the problem. AF-KWS directly regresses the center locations and lengths of the keywords through a single-stage deep neural network. In particular, AF-KWS is tailored for this speech task as we introduce an auxiliary unknown class to exclude other words from non-speech or silent background. We have built two benchmark datasets named LibriTop-20 and continuous meeting analysis keywords (CMAK) dataset for CSKWS. Evaluations on these two datasets show that our proposed AF-KWS outperforms reference schemes by a large margin, and therefore provides a decent baseline for future research. | ['Chong Luo', 'Chengdong Yao', 'Chuanxin Tang', 'Zhiyuan Zhao'] | 2022-08-09 | null | null | null | null | ['keyword-spotting'] | ['speech'] | [ 1.28029823e-01 1.98394924e-01 -1.48288190e-01 -4.64644164e-01
-1.46287239e+00 -4.46722418e-01 5.58416665e-01 -1.20482333e-01
-5.39985240e-01 2.31177866e-01 3.60024154e-01 -5.71806729e-01
1.97674528e-01 9.12091285e-02 -7.39627600e-01 -6.76779389e-01
-2.12396495e-02 2.68150210e-01 5.07240534e-01 2.42990572e-02
4.99485098e-02 2.67601132e-01 -1.00403893e+00 3.03058177e-01
4.17931199e-01 9.94298220e-01 4.36278135e-01 9.90316212e-01
2.16780484e-01 3.04882616e-01 -1.03921783e+00 -3.33944224e-02
3.87157276e-02 -1.79125488e-01 -4.53713775e-01 -1.85087789e-02
3.94625306e-01 -1.78601503e-01 -8.62479806e-01 6.86006606e-01
8.71663094e-01 1.33787647e-01 5.41102171e-01 -1.47425473e+00
-5.78632891e-01 9.96383131e-01 -4.63605434e-01 4.89604741e-01
1.94953591e-01 -7.66378865e-02 1.08825421e+00 -1.46146393e+00
-5.14490940e-02 1.39283454e+00 7.67309785e-01 3.58363837e-01
-8.36493254e-01 -8.55652750e-01 7.10535765e-01 1.97592825e-01
-1.85266674e+00 -1.05389678e+00 6.69013083e-01 -1.70027986e-01
1.06808388e+00 6.14181936e-01 1.96115017e-01 1.38296402e+00
-1.31386533e-01 1.32505107e+00 4.07688051e-01 -5.13400257e-01
2.53561944e-01 6.56032842e-03 1.96023643e-01 5.22085369e-01
-2.35045254e-01 -4.73199397e-01 -8.96879613e-01 -5.18683612e-01
1.12701384e-02 -4.77175623e-01 -7.41760612e-01 6.15883730e-02
-1.50619996e+00 6.41298354e-01 -9.66381058e-02 1.11356370e-01
-6.76721334e-04 1.14637256e-01 4.11830992e-01 1.76836938e-01
6.46357298e-01 4.11084034e-02 -7.53655672e-01 -1.66766167e-01
-9.44963038e-01 4.04358178e-01 8.45017493e-01 1.29202747e+00
8.36008564e-02 3.95171791e-02 -5.60973883e-01 9.76428807e-01
3.80555421e-01 7.73785651e-01 6.46823287e-01 -5.29023826e-01
8.26688170e-01 1.16023973e-01 9.63558629e-02 -9.30248082e-01
-5.57769656e-01 -4.29644614e-01 -5.70669055e-01 -6.72638357e-01
-1.56389624e-02 -8.85125473e-02 -1.03093719e+00 1.65065527e+00
3.53151500e-01 3.84455770e-01 -3.24256904e-02 9.14503813e-01
1.10333014e+00 1.09081340e+00 -3.36285502e-01 -4.27655607e-01
1.28289688e+00 -1.22179782e+00 -9.76273596e-01 -4.22864318e-01
5.74343741e-01 -6.55481756e-01 1.32282686e+00 4.88374710e-01
-8.72828066e-01 -1.64817795e-01 -1.06932521e+00 -2.33933050e-03
-3.68704706e-01 7.21795022e-01 5.33169508e-02 4.72797126e-01
-1.19526982e+00 -2.77875394e-01 -7.50889361e-01 -6.14791550e-02
5.85750379e-02 3.36784720e-01 4.02871706e-02 3.98072064e-01
-1.34629965e+00 3.65155667e-01 1.61018953e-01 2.44108334e-01
-9.22924578e-01 -5.89572549e-01 -8.85231555e-01 1.28735796e-01
8.44851077e-01 -1.21145554e-01 1.93707895e+00 -5.74402750e-01
-1.53884327e+00 7.91527569e-01 -5.31528890e-01 -4.55346793e-01
4.15254951e-01 -8.91785771e-02 -8.62270117e-01 1.56754538e-01
2.21874431e-01 3.55111897e-01 1.30342782e+00 -1.26458311e+00
-6.98717296e-01 -1.57182395e-01 -3.40424180e-01 2.40007535e-01
-6.18331015e-01 2.10122153e-01 -1.10270345e+00 -1.25032341e+00
1.88171133e-01 -9.11886752e-01 2.34424412e-01 -2.34716326e-01
-1.03760564e+00 -6.65104508e-01 1.07275653e+00 -1.00957453e+00
1.84772730e+00 -2.38122916e+00 -9.02517512e-02 2.30884496e-02
3.49456310e-01 6.13877654e-01 -2.05795154e-01 3.95084888e-01
4.73849140e-02 -2.98861344e-03 -2.39576772e-01 -8.42718661e-01
1.79893568e-01 6.01214776e-03 -3.92442703e-01 7.26112366e-01
-1.11063812e-02 8.95738363e-01 -6.70875311e-01 -3.93897355e-01
-3.03568274e-01 3.40531617e-01 -1.27014250e-01 2.26671755e-01
-3.15572649e-01 -1.40126288e-01 -3.13086420e-01 6.32345140e-01
7.30485857e-01 -2.02451833e-02 -7.33025372e-03 2.45894976e-02
3.73692997e-02 6.49239004e-01 -1.16094923e+00 1.65560985e+00
-3.79199624e-01 9.16351855e-01 4.68716443e-01 -1.21026695e+00
6.59253955e-01 3.68581206e-01 6.42561615e-02 -6.70415699e-01
-1.36572182e-01 1.24054737e-01 -1.84473887e-01 -4.77695972e-01
6.66788697e-01 5.63865542e-01 -3.38683963e-01 1.71611965e-01
-6.23721220e-02 2.14003786e-01 -2.78846383e-01 6.21850908e-01
1.21231079e+00 -5.28389871e-01 9.49175134e-02 -3.03240567e-01
6.88315034e-01 -5.25090277e-01 8.43986154e-01 1.14511907e+00
-2.63915330e-01 8.37718785e-01 3.65694404e-01 -4.74763252e-02
-6.34957433e-01 -9.94198561e-01 1.34788051e-01 1.24539447e+00
1.66003898e-01 -6.81236327e-01 -7.43322551e-01 -8.68313253e-01
1.25475734e-01 7.23702848e-01 -3.70941520e-01 -2.11967170e-01
-6.93224907e-01 -3.80339324e-01 8.75682056e-01 3.80409151e-01
6.15895055e-02 -9.18981731e-01 -1.86976627e-01 1.22549295e-01
-4.55122113e-01 -1.72672307e+00 -1.48198354e+00 3.37373286e-01
1.05335973e-01 -7.01045930e-01 -1.05969667e+00 -1.15722108e+00
3.76232505e-01 6.56411707e-01 8.77773464e-01 -7.45993555e-02
-2.20617473e-01 3.95553082e-01 -4.66019452e-01 -5.91275990e-01
-4.72273827e-01 3.65451425e-01 3.54102194e-01 3.34458321e-01
2.83246636e-01 -2.00248390e-01 -6.98570251e-01 3.58231097e-01
-5.59818327e-01 -1.88681826e-01 4.49942142e-01 6.21852934e-01
2.95062780e-01 -1.01609208e-01 9.11288083e-01 -3.57871026e-01
9.39005613e-01 -3.22857320e-01 -5.88543117e-01 3.34962845e-01
-4.18922782e-01 -3.69956881e-01 4.50345963e-01 -6.58230543e-01
-6.64395869e-01 7.78680965e-02 -2.15487957e-01 -4.42956656e-01
1.05769308e-02 3.24646980e-01 -5.70553303e-01 2.01818332e-01
1.99009851e-01 6.92133665e-01 -4.34328109e-01 -6.18642092e-01
7.24292472e-02 1.44358981e+00 7.58261502e-01 -7.36440420e-02
6.95241809e-01 1.71878025e-01 -8.21639001e-01 -1.37498069e+00
-7.95498073e-01 -9.60411966e-01 -9.39030945e-02 3.84159498e-02
5.20060599e-01 -1.23526394e+00 -6.92877412e-01 7.32894957e-01
-1.46698225e+00 -3.41309726e-01 1.23173669e-01 3.40200216e-01
-1.07745953e-01 5.31031072e-01 -4.51830357e-01 -1.21358001e+00
-4.66951549e-01 -1.17188823e+00 1.74722207e+00 -4.55895774e-02
-1.73361495e-01 -4.75473434e-01 -1.81661218e-01 3.62300605e-01
-5.21427840e-02 -4.21377361e-01 5.19069791e-01 -1.27929723e+00
-2.21806228e-01 -2.35899732e-01 -8.59136879e-02 3.98044080e-01
1.21999115e-01 -2.20833123e-01 -1.12628412e+00 -3.89803022e-01
-9.92984772e-02 -2.44239807e-01 1.12894428e+00 3.59028280e-01
1.53396022e+00 -5.83953381e-01 -5.46579361e-01 5.17649710e-01
4.82714564e-01 5.28000236e-01 7.00865313e-02 2.01498255e-01
6.55395091e-01 7.57426918e-02 6.22089207e-01 4.92839724e-01
6.91932619e-01 1.06760609e+00 6.47830442e-02 -2.06137523e-01
-2.33696587e-02 -7.63468072e-02 5.65493643e-01 1.31862998e+00
9.39908266e-01 -9.90649819e-01 -9.05378878e-01 7.27543056e-01
-1.90987706e+00 -3.76151532e-01 -5.69080152e-02 1.84859192e+00
9.56189930e-01 3.92417163e-01 1.86523110e-01 2.65080273e-01
7.95502126e-01 3.58072966e-01 -6.53223574e-01 -1.71082795e-01
-3.12969863e-01 -1.51246086e-01 2.78462142e-01 6.13485456e-01
-1.43930578e+00 9.59525049e-01 5.73694420e+00 1.08409536e+00
-9.28579688e-01 2.45305300e-01 6.49432123e-01 -8.46071020e-02
-5.00322953e-02 -4.07747060e-01 -1.17710257e+00 7.43610442e-01
1.03537357e+00 -6.64246604e-02 6.17538691e-02 8.03178966e-01
5.69848299e-01 -1.72776226e-02 -8.86275828e-01 1.17284250e+00
4.09957021e-01 -1.16531932e+00 -1.21272914e-01 -1.32428274e-01
2.59091347e-01 1.97432995e-01 -2.19386593e-02 4.43535358e-01
-1.77625958e-02 -6.90857530e-01 9.47635829e-01 1.23010188e-01
7.41166055e-01 -6.84264541e-01 3.37325007e-01 6.13690376e-01
-1.16727543e+00 -3.65767889e-02 -2.68003102e-02 3.91230941e-01
2.26132408e-01 5.61006546e-01 -1.16601169e+00 3.89394283e-01
8.50506544e-01 2.64768809e-01 -3.76064271e-01 8.85338962e-01
-7.04751089e-02 1.10890949e+00 -4.08500761e-01 -1.24771595e-01
5.05246937e-01 4.31177318e-01 7.74243355e-01 1.74901271e+00
2.94925511e-01 -6.82331342e-03 4.61426467e-01 3.64192963e-01
-3.19055080e-01 9.94490460e-03 -8.56676027e-02 -2.74681091e-01
8.31523478e-01 1.16528714e+00 -8.23669612e-01 -3.73742461e-01
-4.63483334e-01 1.14815295e+00 4.96784896e-02 5.83595872e-01
-1.04573262e+00 -6.60060883e-01 7.63341844e-01 -4.68494138e-03
6.99559391e-01 -3.68167698e-01 2.88798243e-01 -1.25980985e+00
3.74126703e-01 -1.02129507e+00 1.88400745e-01 -7.05527723e-01
-7.84528673e-01 5.87550819e-01 -1.23276278e-01 -9.63381588e-01
1.36318639e-01 -3.80651981e-01 -5.60106993e-01 8.37828100e-01
-1.63361859e+00 -1.01228440e+00 -2.85938650e-01 5.52948773e-01
1.19268370e+00 -3.00272822e-01 5.88860631e-01 3.37986618e-01
-7.98773170e-01 1.16507161e+00 2.74413079e-01 3.69227499e-01
8.22455645e-01 -1.03680694e+00 1.08828270e+00 8.25371981e-01
4.58134085e-01 3.32054973e-01 6.08749449e-01 -5.85212946e-01
-1.43895566e+00 -1.12204254e+00 1.05454719e+00 -8.52393284e-02
6.97437167e-01 -1.19924879e+00 -8.88467014e-01 5.86955011e-01
1.20580502e-01 1.85408026e-01 2.97124028e-01 -2.74847120e-01
-2.17691615e-01 -1.49822742e-01 -5.26893198e-01 5.39691210e-01
9.96085167e-01 -5.56659400e-01 -5.00484526e-01 6.85305536e-01
1.57620704e+00 -5.02268195e-01 -1.89485878e-01 3.16031665e-01
3.35302889e-01 -4.43916321e-01 9.66482937e-01 -4.09960926e-01
-2.91144818e-01 -1.36117101e-01 -1.12063281e-01 -1.37702560e+00
2.19693426e-02 -1.15087008e+00 -3.63774031e-01 1.37913775e+00
6.71852291e-01 -4.85743135e-01 5.97888947e-01 2.64404900e-02
-6.34607434e-01 -6.88092828e-01 -1.31261599e+00 -8.89761209e-01
-3.92766595e-01 -6.17658556e-01 4.27460313e-01 8.36269021e-01
-2.16196537e-01 4.50621575e-01 -5.25441229e-01 5.89457393e-01
3.63122463e-01 -2.90428579e-01 6.67469203e-01 -8.49601567e-01
-1.74293295e-01 -1.63449273e-01 -1.80217922e-01 -1.62247074e+00
5.53781927e-01 -5.94466269e-01 5.98903835e-01 -1.22314632e+00
-1.06642507e-01 -2.48213679e-01 -1.68725505e-01 3.61051768e-01
-2.93357909e-01 -9.64785516e-02 -1.96204364e-01 3.08643766e-02
-8.20811331e-01 8.87302577e-01 7.08551168e-01 -4.60803390e-01
-3.24253082e-01 4.34723347e-01 -5.42333066e-01 7.92748392e-01
6.74121916e-01 -5.79617262e-01 -3.78706664e-01 -3.47243100e-01
-2.40030289e-02 1.61560215e-02 1.42237499e-01 -6.51395380e-01
5.18174589e-01 6.75135627e-02 -2.54465528e-02 -1.09168661e+00
6.63894415e-01 -4.97610778e-01 -6.42810106e-01 6.45445883e-02
-5.02037585e-01 -4.17052880e-02 1.52605146e-01 9.18650866e-01
-1.74682572e-01 -1.12258457e-01 3.75737756e-01 1.72404185e-01
-6.89847708e-01 1.84767291e-01 -6.63538039e-01 4.13009614e-01
8.48401010e-01 1.55866770e-02 -1.14877470e-01 -7.22702920e-01
-7.02687800e-01 5.61316073e-01 -3.62933695e-01 8.02051783e-01
9.05938148e-01 -1.11056292e+00 -6.85986936e-01 8.47543776e-02
3.63186330e-01 2.03320369e-01 -1.29529282e-01 8.12168479e-01
2.53182510e-03 6.31265223e-01 1.24227786e+00 -7.02002406e-01
-1.55936742e+00 4.10993218e-01 1.79321364e-01 1.73826694e-01
-8.70462656e-01 1.27019107e+00 2.31885135e-01 -6.15711629e-01
1.24295425e+00 -7.88739622e-01 4.04520966e-02 4.97133397e-02
6.15634680e-01 -7.35757779e-03 5.04485011e-01 -7.08256185e-01
-8.20456564e-01 -1.08411638e-02 -3.03032905e-01 -1.42758414e-01
1.12396693e+00 -5.09002268e-01 3.71468574e-01 6.68588102e-01
1.32182693e+00 1.40383422e-01 -1.05984926e+00 -6.72106802e-01
4.66686606e-01 -9.79367271e-02 2.67430305e-01 -6.42529786e-01
-8.09583604e-01 5.30206919e-01 2.93219090e-01 3.04997206e-01
9.32316720e-01 3.91999632e-01 9.47278798e-01 6.28791034e-01
-3.81562263e-01 -1.32232809e+00 3.15551549e-01 5.86309671e-01
1.27020800e+00 -1.25768507e+00 -2.56697148e-01 -4.48148638e-01
-6.02788568e-01 9.34886038e-01 5.26145399e-01 4.10886079e-01
6.73584759e-01 7.61513948e-01 1.35079846e-01 3.64621393e-02
-7.85324633e-01 1.21455364e-01 3.67109418e-01 2.47819036e-01
9.18091461e-02 -6.95445985e-02 -5.15175723e-02 9.88181174e-01
-3.39507498e-02 -4.42524821e-01 3.55929494e-01 9.49578464e-01
-4.05714035e-01 -6.44714117e-01 -4.00842369e-01 3.17342639e-01
-4.52661544e-01 -3.98915738e-01 -7.35221863e-01 5.42546868e-01
-3.92366976e-01 1.14217460e+00 -8.13787282e-02 -4.68804449e-01
5.71511924e-01 1.26817808e-01 -2.36664519e-01 -5.36226094e-01
-5.16619742e-01 6.04375303e-01 2.94285983e-01 -3.39819551e-01
-5.93073592e-02 -5.81991732e-01 -1.18089092e+00 2.98156589e-01
-7.44525731e-01 4.61188227e-01 7.77333617e-01 9.40372109e-01
5.20554781e-01 7.74218917e-01 8.75667930e-01 -7.12229371e-01
-9.54408407e-01 -1.13475835e+00 -6.48682833e-01 -1.48434401e-01
8.64765525e-01 -4.65269744e-01 -6.90365195e-01 -7.53443018e-02] | [14.287323951721191, 6.40095853805542] |
237593a5-e28c-4904-8637-0abdafc8ee96 | predicting-breast-tumor-proliferation-from | 1807.08284 | null | http://arxiv.org/abs/1807.08284v2 | http://arxiv.org/pdf/1807.08284v2.pdf | Predicting breast tumor proliferation from whole-slide images: the TUPAC16 challenge | Tumor proliferation is an important biomarker indicative of the prognosis of
breast cancer patients. Assessment of tumor proliferation in a clinical setting
is highly subjective and labor-intensive task. Previous efforts to automate
tumor proliferation assessment by image analysis only focused on mitosis
detection in predefined tumor regions. However, in a real-world scenario,
automatic mitosis detection should be performed in whole-slide images (WSIs)
and an automatic method should be able to produce a tumor proliferation score
given a WSI as input. To address this, we organized the TUmor Proliferation
Assessment Challenge 2016 (TUPAC16) on prediction of tumor proliferation scores
from WSIs. The challenge dataset consisted of 500 training and 321 testing
breast cancer histopathology WSIs. In order to ensure fair and independent
evaluation, only the ground truth for the training dataset was provided to the
challenge participants. The first task of the challenge was to predict mitotic
scores, i.e., to reproduce the manual method of assessing tumor proliferation
by a pathologist. The second task was to predict the gene expression based
PAM50 proliferation scores from the WSI. The best performing automatic method
for the first task achieved a quadratic-weighted Cohen's kappa score of
$\kappa$ = 0.567, 95% CI [0.464, 0.671] between the predicted scores and the
ground truth. For the second task, the predictions of the top method had a
Spearman's correlation coefficient of r = 0.617, 95% CI [0.581 0.651] with the
ground truth. This was the first study that investigated tumor proliferation
assessment from WSIs. The achieved results are promising given the difficulty
of the tasks and weakly-labelled nature of the ground truth. However, further
research is needed to improve the practical utility of image analysis methods
for this task. | ['Josien P. W. Pluim', 'Yan Xu', 'Eric I-Chao Chang', 'Kyunghyun Paeng', 'Erik Sjöblom', 'Hady Ahmady Phoulady', 'Francesco Ciompi', 'Thomas Wollmann', 'Babak Ehteshami Bejnordi', 'Simon Graham', 'Paul J. van Diest', 'Mitko Veta', 'Mikael Rousson', 'Erwan Zerhouni', 'Dayong Wang', 'Andrew H. Beck', 'Yujing J. Heng', 'Matheus Viana', 'Manan A. Shah', 'Karl Rohr', 'Francisco Beca', 'David Tellez', 'Zhipeng Jia', 'Vitali Liauchuk', 'Talha Qaiser', 'Sunggyun Park', 'Nikolas Stathonikos', 'Nasir Rajpoot', 'Martin Hedlund', 'Jesper Molin', 'Vassili Kovalev', 'Sangheum Hwang', 'David Lanyi'] | 2018-07-22 | null | null | null | null | ['mitosis-detection'] | ['medical'] | [ 4.75537807e-01 3.06644142e-01 -3.97062123e-01 -7.40879849e-02
-1.29927790e+00 -4.55263585e-01 4.54027146e-01 7.35918939e-01
-5.97983718e-01 8.26043844e-01 -3.80379260e-02 -5.19633472e-01
8.25577751e-02 -8.06799829e-01 -3.57682765e-01 -1.29312897e+00
2.19308957e-01 6.65635645e-01 3.43959421e-01 2.70444155e-01
3.08125973e-01 3.13502878e-01 -1.17439961e+00 4.32029009e-01
5.67412257e-01 9.44625795e-01 2.10124388e-01 1.10558105e+00
-7.98171479e-03 3.99930209e-01 -5.13457239e-01 -3.04868877e-01
-3.22210491e-02 -5.25541127e-01 -7.27875352e-01 6.71261689e-04
1.97489157e-01 3.47940363e-02 1.43694147e-01 9.54773962e-01
4.37170446e-01 -4.94563550e-01 1.20580089e+00 -1.11374497e+00
5.25217168e-02 3.23365003e-01 -6.10661209e-01 2.74953872e-01
5.52589074e-02 3.56028855e-01 9.34905767e-01 -5.94701588e-01
8.98563623e-01 2.95770824e-01 6.46184564e-01 4.03469384e-01
-1.39735317e+00 -6.81431293e-01 -6.25676990e-01 3.22782435e-02
-1.60333097e+00 -2.22053483e-01 4.47118245e-02 -7.18191981e-01
7.16812909e-01 4.62805718e-01 8.83299232e-01 5.57628036e-01
3.89934570e-01 4.75318372e-01 1.48036265e+00 -3.95010740e-01
3.34196299e-01 3.68478358e-01 4.24064249e-02 7.16553569e-01
4.94235009e-01 -8.09812322e-02 -3.06139618e-01 -2.27437690e-02
5.43470025e-01 -3.93269151e-01 -3.17060590e-01 4.92003597e-02
-1.47236133e+00 7.02326655e-01 3.66471201e-01 7.17418730e-01
-1.13442175e-01 -1.24119312e-01 6.17880702e-01 1.38010591e-01
4.43998754e-01 4.48951602e-01 -1.75334841e-01 3.53040211e-02
-1.27531147e+00 8.96973535e-02 6.52471066e-01 2.06336483e-01
4.03943032e-01 -5.94162107e-01 -4.11181062e-01 5.92220485e-01
3.11465681e-01 4.40194100e-01 7.20749795e-01 -6.47089064e-01
-8.22009146e-02 7.90521681e-01 6.44491985e-02 -7.34299719e-01
-7.24132895e-01 -5.41803956e-01 -1.00160336e+00 1.75379649e-01
1.21212721e+00 2.70936251e-01 -1.04305494e+00 1.36832631e+00
2.18101293e-01 -1.36702135e-02 1.56124234e-01 9.68160868e-01
8.80262017e-01 3.96402895e-01 3.64722282e-01 -4.80033576e-01
1.60115349e+00 -4.86939013e-01 -4.77803349e-01 -4.27906513e-02
1.25373995e+00 -7.85146475e-01 1.07677114e+00 1.96415707e-01
-8.39934647e-01 -1.40121371e-01 -1.12146711e+00 2.73417652e-01
-3.20067704e-01 4.90442038e-01 3.35653216e-01 6.38611913e-01
-1.15681791e+00 2.03427210e-01 -8.24835837e-01 -6.84833765e-01
4.93016273e-01 4.22911108e-01 -6.66511774e-01 -6.43245354e-02
-8.88216615e-01 9.15992081e-01 3.16484839e-01 3.06741465e-02
-6.30992413e-01 -8.04061055e-01 -4.60075885e-01 -2.15039924e-01
-9.55919698e-02 -5.73067069e-01 1.14349341e+00 -8.18181515e-01
-1.11863720e+00 1.65635860e+00 -2.76798338e-01 -2.91221052e-01
5.02472103e-01 9.26894128e-01 -8.75742733e-02 1.74023002e-01
3.06089163e-01 6.96484864e-01 3.50597464e-02 -1.18343663e+00
-9.05668855e-01 -5.30874312e-01 -5.13773024e-01 9.26459581e-02
-9.06548575e-02 -4.95162100e-01 -4.88537371e-01 -3.52683038e-01
9.05402526e-02 -1.20517242e+00 -1.88991070e-01 9.17545632e-02
-1.29031807e-01 -1.23654261e-01 2.11141750e-01 -8.43812048e-01
1.08778894e+00 -2.23905611e+00 -3.48174542e-01 5.48283100e-01
1.96408734e-01 3.84342581e-01 1.66993782e-01 1.27183925e-02
-2.37244368e-02 3.75028610e-01 -1.85670853e-01 -1.18511945e-01
-1.52343079e-01 -9.93941873e-02 4.67523903e-01 7.56940663e-01
1.02439530e-01 8.99586737e-01 -8.94290268e-01 -7.46440232e-01
-7.93389753e-02 2.26957679e-01 -1.16938874e-01 1.34536758e-01
8.35310481e-03 4.21445638e-01 1.69507086e-01 7.16103733e-01
3.29278290e-01 -3.98038864e-01 2.75048167e-01 -2.95615971e-01
1.17705286e-01 -1.37828484e-01 -7.75423467e-01 1.30412793e+00
-7.86899105e-02 8.85676503e-01 6.75177351e-02 -8.22230816e-01
9.32647526e-01 5.43597579e-01 6.18647754e-01 -5.65928578e-01
2.03233242e-01 5.86258590e-01 4.19829160e-01 -3.63557190e-01
2.21906230e-02 -4.93311256e-01 4.14592065e-02 1.56679168e-01
-1.93231046e-01 -3.78129631e-01 5.18999338e-01 -1.90230347e-02
1.40197062e+00 -5.10872424e-01 7.04579055e-01 -4.81559098e-01
6.63643479e-01 4.25579250e-01 4.80943918e-01 3.84901166e-01
-5.04991531e-01 5.79972625e-01 8.58126044e-01 -2.90732563e-01
-1.04338086e+00 -1.07959211e+00 -3.92063558e-01 4.22226071e-01
-1.52313188e-01 -1.70499966e-01 -6.96368754e-01 -5.94958246e-01
-1.72231823e-01 3.44410837e-01 -8.98513436e-01 2.11080108e-02
-3.26233469e-02 -1.04785848e+00 7.06474662e-01 3.88159364e-01
3.10640037e-01 -5.79533696e-01 -3.42561901e-01 8.34573433e-02
-4.70599264e-01 -9.16990757e-01 -2.70925969e-01 2.27789879e-01
-6.71605051e-01 -1.30351961e+00 -9.19241726e-01 -9.36337590e-01
1.02594578e+00 -1.03638165e-01 8.81514192e-01 2.88741499e-01
-5.20503461e-01 -1.95299275e-02 -1.68035522e-01 -5.69490492e-01
-7.29566097e-01 8.11050087e-02 -3.73334795e-01 -2.05414817e-01
6.32430851e-01 -4.89296131e-02 -7.65239358e-01 4.97243136e-01
-9.71448421e-01 2.02495426e-01 9.57289636e-01 1.00765920e+00
1.07818687e+00 9.44876671e-02 3.93577158e-01 -8.00136745e-01
1.91060156e-01 -2.83473641e-01 -3.67188424e-01 1.55853495e-01
-6.43299699e-01 -2.89167613e-01 2.93220490e-01 -2.58018166e-01
-7.46935248e-01 1.95911929e-01 -1.76362321e-01 2.46256560e-01
-2.01548263e-01 7.81922698e-01 3.93041432e-01 6.32774904e-02
8.34366083e-01 -4.32224311e-02 3.49779308e-01 5.09792745e-01
-5.20349145e-01 7.46577382e-01 6.00639045e-01 -1.01148315e-01
2.88215309e-01 3.49828154e-01 5.38242698e-01 -7.19649434e-01
-7.71306157e-01 -8.68445694e-01 -2.85664141e-01 -3.64528209e-01
8.49290907e-01 -8.21287751e-01 -7.39187837e-01 4.84881014e-01
-6.23192966e-01 -7.70330489e-01 -1.44016340e-01 6.95968151e-01
-6.68971658e-01 1.41545296e-01 -7.02822149e-01 -5.37728131e-01
-5.69896281e-01 -1.05921972e+00 1.20475197e+00 2.44824454e-01
-7.13274062e-01 -8.89444411e-01 3.40572029e-01 7.36213863e-01
2.07480803e-01 6.13002598e-01 1.03847814e+00 -6.21855021e-01
-1.31943062e-01 -6.64138675e-01 -3.38557273e-01 -3.80587988e-02
1.11308977e-01 4.66116995e-01 -9.72979903e-01 -9.76720378e-02
-3.80494982e-01 -1.83656693e-01 7.37736523e-01 6.33477569e-01
7.98587799e-01 -6.87540248e-02 -7.27051914e-01 1.44486114e-01
1.48069513e+00 1.88902438e-01 8.21236074e-01 4.16097879e-01
2.77926214e-03 7.07271099e-01 9.18849170e-01 7.15990886e-02
1.99616015e-01 4.86576319e-01 1.61978573e-01 -2.53658593e-01
-2.57684708e-01 -3.82087603e-02 -3.10951341e-02 3.58175904e-01
-8.10945630e-02 -3.10843885e-01 -1.46670735e+00 6.96521163e-01
-1.48543465e+00 -6.85009181e-01 -5.47228694e-01 2.32999158e+00
9.56392288e-01 3.56243372e-01 4.74684350e-02 6.97000861e-01
6.11725211e-01 -5.09642720e-01 -1.67518452e-01 -2.69193411e-01
3.57062407e-02 4.10121530e-02 4.06785250e-01 4.37685341e-01
-7.40181327e-01 3.95742089e-01 5.53569746e+00 9.30747926e-01
-1.16601717e+00 -3.14323418e-02 1.18062901e+00 6.58335686e-02
1.54368013e-01 -1.71489805e-01 -7.49454319e-01 5.44085443e-01
1.08529782e+00 -2.11527079e-01 -2.38131091e-01 2.43081182e-01
3.59315366e-01 -8.86925399e-01 -9.61951077e-01 7.54240215e-01
-1.22484662e-01 -1.23821807e+00 -4.96786833e-01 3.86294156e-01
7.11308658e-01 -1.88234955e-01 -7.71809369e-02 1.38006434e-01
3.87429260e-02 -1.12518191e+00 1.12956762e-02 6.58794999e-01
1.20599055e+00 -4.03649360e-01 1.55346453e+00 3.94395500e-01
-9.67045844e-01 3.38278770e-01 5.75277694e-02 1.31646991e-01
-1.90373048e-01 9.20608878e-01 -1.74140120e+00 3.78704779e-02
4.36899275e-01 1.74698040e-01 -6.46812916e-01 1.10013616e+00
9.62915644e-03 6.98859334e-01 -2.99773633e-01 -2.13221118e-01
5.21730408e-02 2.60981679e-01 7.96868727e-02 1.42968845e+00
5.76081574e-01 2.33157557e-02 -9.73843411e-02 3.39360446e-01
2.53620863e-01 2.94877142e-01 -2.41343856e-01 -3.02947849e-01
2.12294206e-01 1.66339314e+00 -1.47123468e+00 -2.82627225e-01
-1.86010480e-01 3.33044380e-01 5.94556704e-02 -5.44010364e-02
-5.98295033e-01 -3.10380049e-02 1.28790781e-01 6.48964107e-01
-2.59698421e-01 1.99628875e-01 -7.79780447e-01 -4.76362526e-01
-3.10501724e-01 -5.20107627e-01 6.30807877e-01 -6.36259854e-01
-1.17767870e+00 2.59535789e-01 -2.66779989e-01 -1.14645624e+00
-4.73614372e-02 -7.19227493e-01 -5.85710943e-01 8.86404634e-01
-1.22792840e+00 -9.95400310e-01 -6.34103775e-01 1.94379464e-02
1.99418902e-01 2.87371635e-01 1.04156363e+00 -4.22810614e-02
-3.45964223e-01 5.89985907e-01 -1.55838102e-01 8.42817128e-03
9.63758230e-01 -1.37045324e+00 -3.81068975e-01 2.76214361e-01
-4.00710315e-01 1.72051653e-01 8.48126113e-01 -6.11975491e-01
-7.89699197e-01 -1.05252457e+00 1.26557231e+00 -2.79227674e-01
6.46913707e-01 1.01391345e-01 -7.21100211e-01 2.24556088e-01
-1.65400997e-01 6.21431284e-02 1.31261408e+00 -2.53573030e-01
1.20707095e-01 -5.58426790e-02 -1.24498439e+00 4.77711439e-01
3.66522342e-01 -2.49951154e-01 1.47254631e-01 3.79840016e-01
-2.35347465e-01 -6.32225990e-01 -1.33317554e+00 6.48567915e-01
7.43735611e-01 -6.56331778e-01 5.37438691e-01 1.19080707e-01
4.90669817e-01 -6.33850336e-01 7.87372664e-02 -1.05119872e+00
-3.61485928e-01 3.05255115e-01 3.85972857e-01 1.03302300e+00
9.08095419e-01 -2.83833772e-01 1.23047733e+00 4.78471845e-01
-5.96411861e-02 -1.13166976e+00 -9.89505529e-01 -3.52025867e-01
3.63008678e-02 -3.81844081e-02 7.53226085e-03 6.56317592e-01
3.81052464e-01 1.06988385e-01 4.30019796e-01 1.22553766e-01
2.94176757e-01 -3.28713030e-01 7.62701631e-01 -1.08950818e+00
-1.51966944e-01 -6.91836178e-01 -9.46176589e-01 4.48083170e-02
-1.22132696e-01 -1.05272365e+00 1.43078119e-01 -1.63316250e+00
5.49040318e-01 -3.16169322e-01 -2.13725790e-01 4.02493864e-01
-3.47758889e-01 7.43318200e-01 -2.35201582e-01 3.68067980e-01
-2.61811495e-01 -2.54427582e-01 1.24405396e+00 -3.83004308e-01
-4.53670062e-02 -5.32878982e-03 -6.38065159e-01 4.67441946e-01
1.05944872e+00 -3.73596072e-01 -3.64823759e-01 4.03098434e-01
3.35722089e-01 1.88960612e-01 3.45729113e-01 -1.11269820e+00
3.65217775e-01 -2.66779035e-01 6.30584061e-01 -6.40458703e-01
1.96875408e-01 -5.44976354e-01 5.04601836e-01 8.35041225e-01
-2.47452408e-01 -3.41375113e-01 1.44767553e-01 3.11295211e-01
-2.38892019e-01 -3.32585037e-01 8.10081422e-01 -1.14611089e-01
-2.22183809e-01 7.47646689e-02 -6.78845286e-01 -3.66737455e-01
1.38444614e+00 -6.29357696e-01 -5.34186006e-01 -4.96997312e-02
-8.22938442e-01 1.11305319e-01 7.00112045e-01 -2.98109084e-01
1.71802372e-01 -1.01879430e+00 -8.07069719e-01 -1.90626517e-01
5.24557590e-01 1.09086163e-01 2.32499227e-01 1.45935655e+00
-7.08424628e-01 4.92392510e-01 -1.17919683e-01 -8.85266483e-01
-1.55518842e+00 2.68731087e-01 2.97139376e-01 -9.89410758e-01
7.67360302e-03 8.15223217e-01 1.99914396e-01 -3.40235122e-02
2.36197505e-02 -3.10699493e-01 -3.25322390e-01 1.23392582e-01
4.97531801e-01 3.03311110e-01 4.04165685e-01 -5.85269034e-01
-2.17354193e-01 3.43276799e-01 -1.01724632e-01 -9.34032053e-02
8.65975738e-01 7.27341771e-02 -2.50595391e-01 6.31625116e-01
1.06716537e+00 -3.12324941e-01 -8.84808719e-01 8.32223520e-02
2.95107905e-02 -2.59571999e-01 8.91132876e-02 -1.06594813e+00
-8.04262578e-01 5.01066625e-01 7.48422265e-01 1.98297277e-01
1.15547216e+00 -5.51276561e-03 3.07900965e-01 -6.35852367e-02
8.03058743e-02 -1.09186327e+00 -1.09293438e-01 2.51611918e-02
5.50196111e-01 -1.31310070e+00 -7.93157592e-02 -7.57974684e-01
-5.53885520e-01 1.15024447e+00 5.29358089e-01 2.94858068e-01
4.37657893e-01 4.43396956e-01 2.67285973e-01 -3.39172393e-01
-8.02970827e-01 -6.55744150e-02 1.69292405e-01 5.92864394e-01
8.07848036e-01 3.50157171e-01 -1.00395083e+00 6.63085520e-01
-3.36109847e-01 4.07827377e-01 4.64576542e-01 7.12187588e-01
-4.18424785e-01 -8.08904409e-01 -3.62883687e-01 8.70414317e-01
-6.41696453e-01 4.09532130e-01 -5.01767933e-01 8.09720099e-01
2.05543246e-02 9.05748546e-01 9.79539007e-02 -1.60507008e-01
8.14985186e-02 5.12630008e-02 2.48099327e-01 -5.05447209e-01
-4.20217901e-01 4.88815941e-02 1.56665817e-01 -5.15778363e-02
-5.67893803e-01 -6.98887944e-01 -1.39884031e+00 -2.49918446e-01
-3.79696816e-01 2.24108234e-01 7.87027180e-01 7.92077243e-01
-1.92184389e-01 5.42860031e-01 2.52194792e-01 -4.33452159e-01
9.28992108e-02 -9.76982892e-01 -7.65579879e-01 2.99524963e-01
5.86063340e-02 -4.24423665e-01 -4.76733536e-01 3.75759363e-01] | [15.063216209411621, -3.1614439487457275] |
902e2713-d4a7-4e4b-9cdf-d80f40548b7d | vihos-hate-speech-spans-detection-for | 2301.10186 | null | https://arxiv.org/abs/2301.10186v2 | https://arxiv.org/pdf/2301.10186v2.pdf | ViHOS: Hate Speech Spans Detection for Vietnamese | The rise in hateful and offensive language directed at other users is one of the adverse side effects of the increased use of social networking platforms. This could make it difficult for human moderators to review tagged comments filtered by classification systems. To help address this issue, we present the ViHOS (Vietnamese Hate and Offensive Spans) dataset, the first human-annotated corpus containing 26k spans on 11k comments. We also provide definitions of hateful and offensive spans in Vietnamese comments as well as detailed annotation guidelines. Besides, we conduct experiments with various state-of-the-art models. Specifically, XLM-R$_{Large}$ achieved the best F1-scores in Single span detection and All spans detection, while PhoBERT$_{Large}$ obtained the highest in Multiple spans detection. Finally, our error analysis demonstrates the difficulties in detecting specific types of spans in our data for future research. Disclaimer: This paper contains real comments that could be considered profane, offensive, or abusive. | ['Ngan Luu-Thuy Nguyen', 'Kiet Van Nguyen', 'Khanh Quoc Tran', 'Canh Duc Luu', 'Phu Gia Hoang'] | 2023-01-24 | null | null | null | null | ['xlm-r'] | ['natural-language-processing'] | [-3.81426841e-01 1.24331854e-01 -1.43074855e-01 -6.73097447e-02
-7.37861454e-01 -1.07528961e+00 4.76898015e-01 4.86372769e-01
-4.12674010e-01 8.01227331e-01 4.24680084e-01 -2.78612286e-01
1.74510404e-01 -2.39000693e-01 -3.55824642e-02 -3.25622827e-01
-9.31026507e-03 -1.73892960e-01 9.12343711e-02 -5.27863026e-01
8.24749768e-01 2.18880445e-01 -9.11202848e-01 3.08739811e-01
1.00749433e+00 4.71546590e-01 -2.94183910e-01 7.19228446e-01
4.12716158e-02 1.13398325e+00 -1.33378696e+00 -1.13949955e+00
-3.41626965e-02 -2.46680453e-01 -7.53898859e-01 -1.90551698e-01
8.17056835e-01 -2.73835570e-01 -3.90917838e-01 1.20004594e+00
5.55466831e-01 6.56699538e-02 7.17505097e-01 -1.23943806e+00
-9.93331075e-01 1.05394161e+00 -9.81150627e-01 6.61681056e-01
2.93107420e-01 1.03864327e-01 1.21964276e+00 -8.90822291e-01
7.40149081e-01 1.12148249e+00 4.68966216e-01 4.84602630e-01
-8.77079248e-01 -1.02744567e+00 -1.72222078e-01 -2.23319493e-02
-1.36068714e+00 -2.86165297e-01 7.43392587e-01 -9.62851822e-01
8.50832582e-01 4.35295969e-01 3.84696603e-01 1.46904397e+00
-2.67202053e-02 7.85169780e-01 8.35082471e-01 -2.07293421e-01
-2.31888413e-01 5.78374326e-01 4.62460101e-01 5.41182578e-01
4.81703788e-01 -4.62944359e-01 -6.19682610e-01 -6.70038819e-01
7.65645355e-02 -2.37115234e-01 -2.71147251e-01 5.32168627e-01
-7.84223795e-01 1.22644377e+00 3.22060436e-01 4.09242690e-01
6.37899488e-02 -2.37203822e-01 1.03571808e+00 1.92869961e-01
8.14769804e-01 1.06264567e+00 -1.92023948e-01 -4.80055720e-01
-9.34227824e-01 2.30651557e-01 7.62265682e-01 8.86841953e-01
4.83133137e-01 -5.97934462e-02 -1.52564377e-01 1.15874720e+00
-4.35531512e-02 4.22064364e-01 2.49581650e-01 -6.83952391e-01
8.33499312e-01 4.18151796e-01 2.41264969e-01 -1.48519993e+00
-3.39069009e-01 -3.71575415e-01 -5.28929353e-01 -2.52688915e-01
4.19872433e-01 -7.25513637e-01 -3.17302287e-01 1.28873801e+00
-4.10437472e-02 -5.55441141e-01 -4.75452542e-01 5.67716300e-01
7.48529136e-01 6.37365222e-01 3.78651708e-01 -4.31020319e-01
1.29714835e+00 -7.70525396e-01 -9.77545559e-01 -1.08000956e-01
1.03404880e+00 -1.12502563e+00 1.14580619e+00 4.68801469e-01
-8.31061840e-01 -1.80552363e-01 -8.81711781e-01 -2.13176563e-01
-6.07460260e-01 -1.30097672e-01 1.95542186e-01 8.03606153e-01
-3.88176501e-01 5.47798336e-01 -2.44165100e-02 -3.30345899e-01
3.70509654e-01 -2.17607871e-01 -3.09003621e-01 5.36287248e-01
-1.49711621e+00 1.13360691e+00 5.20075262e-02 -2.22073808e-01
-4.34322894e-01 -7.63966560e-01 -7.99372911e-01 -2.59477854e-01
4.08521950e-01 5.52925408e-01 1.25240397e+00 -7.68426597e-01
-5.70996642e-01 1.15781903e+00 2.19829932e-01 -2.05660433e-01
4.29469168e-01 -7.76245058e-01 -8.92749608e-01 1.34607837e-01
3.28958213e-01 1.50585607e-01 8.72523129e-01 -9.23164487e-01
-3.90666425e-01 -2.83409536e-01 2.92907171e-02 -3.42322499e-01
-9.51092303e-01 7.62396038e-01 3.02152157e-01 -8.96712005e-01
-7.51116633e-01 -8.99725556e-01 3.23193043e-01 -5.34738243e-01
-8.51500213e-01 -5.54915309e-01 1.06725299e+00 -1.13424265e+00
2.33781791e+00 -2.29193521e+00 -1.71982616e-01 1.36960015e-01
7.09883392e-01 6.83629632e-01 1.89899713e-01 1.02386475e+00
-1.29619380e-02 9.49779868e-01 1.47367343e-01 -1.93768620e-01
4.92237471e-02 -1.52496085e-01 -2.65660137e-01 7.19835401e-01
-7.87742510e-02 6.07943356e-01 -9.73981321e-01 -6.99276149e-01
-3.40057015e-01 1.36980295e-01 -2.64397383e-01 1.28758505e-01
1.82898179e-01 3.75869626e-04 -2.20422909e-01 6.18400335e-01
4.17078882e-01 -2.98770647e-02 -6.38314113e-02 1.19646005e-01
-5.07991791e-01 3.60191375e-01 -3.39018792e-01 8.83101165e-01
-7.95877725e-02 1.25110281e+00 1.01524070e-01 -9.72055346e-02
9.87954021e-01 2.97311604e-01 9.21667963e-02 -3.03337604e-01
3.98078948e-01 2.15051100e-01 4.42615859e-02 -8.97592425e-01
1.08873653e+00 5.35150357e-02 -6.02380514e-01 4.39580649e-01
-2.65299857e-01 1.32296756e-01 4.83279288e-01 8.02836835e-01
1.03907812e+00 -5.13280272e-01 5.73093593e-01 -3.08115393e-01
4.10251588e-01 -1.93520524e-02 3.97243440e-01 6.13432646e-01
-7.37575948e-01 3.90362144e-01 1.13038194e+00 -1.36201560e-01
-1.24321353e+00 -4.58719611e-01 -1.37421545e-02 1.46829128e+00
-3.38178933e-01 -8.23375344e-01 -7.50147521e-01 -1.11791515e+00
7.84555152e-02 1.00598204e+00 -8.02839994e-01 7.74297565e-02
-4.31949317e-01 -3.44559163e-01 1.12607574e+00 3.25250238e-01
3.43528032e-01 -9.75546122e-01 -4.93592620e-01 -8.40312764e-02
-5.34352660e-01 -8.79427433e-01 -7.92529225e-01 -3.01324390e-02
-2.57824198e-03 -1.16689920e+00 -6.72820330e-01 -4.46658909e-01
3.96802217e-01 3.21787089e-01 7.19249904e-01 4.28343117e-01
-2.20553070e-01 -1.76419407e-01 -8.40150535e-01 -7.21554935e-01
-4.85913426e-01 4.11107928e-01 -1.18711688e-01 -2.76335299e-01
6.20856524e-01 -8.93963575e-02 -1.71562195e-01 1.55399442e-01
-8.79551470e-01 -6.19052172e-01 -1.07885830e-01 6.36790395e-01
-4.93470281e-01 -2.64906317e-01 6.73019350e-01 -1.46205556e+00
1.08936226e+00 -8.89182866e-01 -7.50193000e-02 -1.34016722e-01
-3.10952038e-01 -6.24002993e-01 8.03353190e-01 -4.93732691e-01
-7.99137175e-01 -5.98437309e-01 -1.46565333e-01 -1.57748178e-01
-2.36402455e-04 4.25018370e-01 2.79606223e-01 2.70366043e-01
1.17030716e+00 -4.84533340e-01 -1.43447086e-01 -6.36371076e-01
-1.62385944e-02 1.23060858e+00 2.06989989e-01 -1.15101583e-01
9.64595735e-01 1.22657241e-02 -6.58478558e-01 -1.16861272e+00
-1.32223880e+00 -9.63658750e-01 -5.95156670e-01 -4.19935793e-01
8.60892177e-01 -7.00996757e-01 -6.35868609e-01 4.18710440e-01
-1.17588985e+00 -8.38830248e-02 3.56499761e-01 2.11666878e-02
1.94959924e-01 7.65680671e-01 -1.03386724e+00 -9.93103564e-01
-5.05882144e-01 -5.73175788e-01 4.94374990e-01 1.85304925e-01
-1.21739733e+00 -9.11476493e-01 1.44486532e-01 6.74537063e-01
2.82979131e-01 4.79215890e-01 7.96470284e-01 -1.08474922e+00
6.58205628e-01 -4.68426257e-01 -2.97975659e-01 4.52050626e-01
-7.57877007e-02 6.62914813e-01 -8.19352925e-01 -2.78401285e-01
-3.59001905e-01 -7.82732010e-01 4.77522492e-01 -2.92486399e-01
8.61595631e-01 -9.26085293e-01 -2.14058340e-01 -1.92708790e-01
1.14850199e+00 -1.02447374e-02 6.92096055e-01 4.30753142e-01
6.77966237e-01 6.81182384e-01 7.66166627e-01 9.71145570e-01
4.22754586e-02 3.66188228e-01 4.43563938e-01 1.69190109e-01
2.67709702e-01 -4.30908978e-01 4.80300546e-01 8.73329461e-01
1.41750127e-01 -5.97914815e-01 -1.01418865e+00 7.59160757e-01
-1.59330416e+00 -1.33162642e+00 -7.08526194e-01 1.79111922e+00
7.33983278e-01 1.66684255e-01 5.67936659e-01 1.96368128e-01
9.71703827e-01 6.27043903e-01 -2.15012711e-02 -1.07144237e+00
8.44803378e-02 -1.46115780e-01 4.98259574e-01 4.74078834e-01
-1.08284521e+00 8.88126552e-01 6.23000145e+00 8.58205259e-01
-9.78058398e-01 2.58400202e-01 5.20073116e-01 -2.44575903e-01
-5.90831526e-02 -3.42160076e-01 -9.58260536e-01 8.84304225e-01
1.04580283e+00 -1.98929355e-01 -5.26612112e-03 1.04237437e+00
3.73189688e-01 -2.68135425e-02 -5.45173526e-01 7.95091748e-01
4.64307666e-01 -1.06484187e+00 -4.06139821e-01 1.61234275e-01
7.19938099e-01 -1.33022889e-01 1.19654045e-01 5.03201127e-01
2.45082468e-01 -9.16484594e-01 7.39031971e-01 2.96395868e-02
7.71951437e-01 -8.52504790e-01 7.62128353e-01 4.64000642e-01
-5.61249018e-01 -2.48549938e-01 -4.13042694e-01 -2.39066854e-01
1.81137502e-01 5.44388056e-01 -7.19132364e-01 -6.20487668e-02
6.34387553e-01 7.81712711e-01 -9.23164189e-01 8.33640933e-01
-4.50832725e-01 9.96816099e-01 1.58222362e-01 -6.24005675e-01
4.08422589e-01 2.66496837e-01 6.69175804e-01 1.87189877e+00
-2.66550090e-02 3.48404795e-02 -3.59721035e-02 3.58573109e-01
-2.58355975e-01 3.17151636e-01 -7.69375682e-01 -5.19760191e-01
7.30067372e-01 1.55234170e+00 -4.20243561e-01 -3.40122193e-01
-4.31688517e-01 7.54905760e-01 7.46255875e-01 1.27472237e-01
-1.10494602e+00 -8.56510758e-01 6.18114829e-01 4.97373492e-01
-2.43119933e-02 -2.78898835e-01 -3.50200534e-01 -1.15116203e+00
-4.08190757e-01 -9.64643657e-01 6.04826152e-01 -6.03154182e-01
-1.60946190e+00 4.58012819e-01 -5.29457331e-02 -9.41510260e-01
5.25227487e-02 -3.28863233e-01 -5.58084667e-01 5.35009801e-01
-9.12357032e-01 -7.76503623e-01 -1.22730516e-01 1.92477211e-01
6.10099256e-01 9.59672853e-02 5.26242495e-01 4.94692206e-01
-9.00022984e-01 7.82396495e-01 -1.60503075e-01 8.30961287e-01
1.23226666e+00 -1.05479169e+00 1.14150189e-01 8.86449456e-01
-3.58370304e-01 7.39785850e-01 9.83756363e-01 -9.61016953e-01
-5.23873866e-01 -9.25377488e-01 1.40133262e+00 -7.93922961e-01
1.42178917e+00 -2.37765059e-01 -9.83281851e-01 7.16702223e-01
6.36220932e-01 -5.43387711e-01 1.29503381e+00 3.27994198e-01
-9.06309605e-01 5.03493488e-01 -1.06472015e+00 6.71522796e-01
1.00361359e+00 -5.71578622e-01 -7.05603063e-01 4.65980768e-01
5.56457520e-01 -1.13095947e-01 -8.52548063e-01 -1.90347239e-01
4.70750451e-01 -7.95496285e-01 2.44625792e-01 -9.89525437e-01
1.03008747e+00 2.06670418e-01 1.57203317e-01 -1.26698101e+00
-7.22773552e-01 -8.59137893e-01 -6.64079413e-02 1.71357238e+00
6.82462096e-01 -2.40735456e-01 4.77164298e-01 8.13949466e-01
-1.73982576e-01 -4.23441142e-01 -5.39713979e-01 -5.72153211e-01
4.13362682e-01 -1.18323088e-01 3.99798900e-02 1.57695663e+00
8.58955324e-01 4.96198207e-01 -8.99399996e-01 -2.10071668e-01
3.65943074e-01 -4.82174337e-01 7.60433793e-01 -1.05586505e+00
1.01275511e-01 -4.93036479e-01 -3.12575027e-02 -6.00581229e-01
3.10565442e-01 -5.48026025e-01 -2.38346890e-01 -1.03859818e+00
4.17481810e-01 -1.26514792e-01 1.41480684e-01 4.59038049e-01
-3.03804219e-01 6.19467854e-01 4.86462563e-01 3.50454420e-01
-7.66941488e-01 -8.96118954e-02 1.14829910e+00 -5.80085209e-03
-1.68189153e-01 -2.86588818e-01 -1.11896861e+00 7.89573967e-01
9.17192459e-01 -4.77764487e-01 4.59059794e-03 -9.57824811e-02
6.28033757e-01 -2.03106582e-01 -6.03967495e-02 -5.13125539e-01
-1.16013415e-01 -2.88478523e-01 1.52601525e-01 -5.91485739e-01
1.58767343e-01 -3.44262332e-01 -3.95967543e-01 3.78193319e-01
-7.18906224e-01 3.06227446e-01 -2.93850973e-02 2.77976871e-01
-2.36441083e-02 -6.25178933e-01 1.04223478e+00 -2.19416752e-01
-3.67037922e-01 -7.85537809e-02 -1.04635048e+00 4.88309950e-01
1.02352345e+00 -3.10412683e-02 -1.03930485e+00 -6.36237383e-01
-4.09255207e-01 1.66532144e-01 4.29993719e-01 5.27815521e-01
3.07238489e-01 -7.79520631e-01 -8.74868095e-01 -3.83712381e-01
4.24434185e-01 -7.88924694e-01 1.91837743e-01 7.33221948e-01
-6.44307852e-01 1.89956561e-01 -1.68681204e-01 1.81755841e-01
-1.69176960e+00 7.04015791e-01 -1.28617451e-01 -2.47132346e-01
-2.66104043e-01 8.41750443e-01 -3.53077680e-01 2.80111898e-02
1.64328292e-01 4.74144787e-01 -5.33908248e-01 7.48842180e-01
9.40724015e-01 9.40764666e-01 -3.14329654e-01 -1.14660406e+00
-3.01663011e-01 -1.27311036e-01 -5.87569952e-01 1.30791873e-01
7.90735602e-01 -3.02542895e-01 -3.09673250e-01 5.48790038e-01
1.58955717e+00 7.70991325e-01 -4.34578687e-01 8.71170908e-02
1.39198333e-01 -7.47171402e-01 -4.54470336e-01 -8.34619462e-01
-6.05446935e-01 7.43771136e-01 -3.29747379e-01 9.62272286e-01
4.42273229e-01 -2.38928329e-02 1.13325965e+00 1.92624759e-02
-7.59923225e-03 -1.43776619e+00 4.03998315e-01 7.47315168e-01
8.96051526e-01 -1.19481409e+00 1.91193581e-01 -4.54939395e-01
-1.05113864e+00 1.05499125e+00 1.02927232e+00 6.86242878e-02
6.07965708e-01 -6.91241771e-02 3.78578752e-01 -4.23627824e-01
-6.28006458e-01 2.09429234e-01 1.41820952e-01 3.65891486e-01
9.26825225e-01 9.31104124e-02 -1.04991269e+00 4.11103010e-01
-4.88790214e-01 -7.05411971e-01 1.12897503e+00 7.05540657e-01
-7.31940925e-01 -6.05482519e-01 -3.55285287e-01 6.87251151e-01
-1.15293443e+00 -9.45462361e-02 -1.12839842e+00 8.97355676e-01
1.77601841e-03 1.29363859e+00 -1.95214272e-01 -6.48059785e-01
8.42209235e-02 5.69839552e-02 -2.61859596e-01 -8.27200055e-01
-1.34990537e+00 -6.69873459e-03 7.50173688e-01 -1.16932794e-01
-1.64350137e-01 -5.94689429e-01 -9.62052405e-01 -9.80057836e-01
-6.29826844e-01 4.92763549e-01 3.02389234e-01 6.22318268e-01
2.39559889e-01 -5.10287210e-02 6.17056310e-01 -3.25971603e-01
-5.99090695e-01 -1.35340595e+00 -8.54099095e-01 8.29302192e-01
2.53691077e-01 -4.54763770e-01 -8.21594536e-01 -1.52821466e-01] | [8.793437004089355, 10.566301345825195] |
8bcc0cfb-d9e2-4e1c-9d51-b8faecf9c73c | fast-and-accurate-tensor-completion-with | 1804.06128 | null | http://arxiv.org/abs/1804.06128v3 | http://arxiv.org/pdf/1804.06128v3.pdf | Fast and Accurate Tensor Completion with Total Variation Regularized Tensor Trains | We propose a new tensor completion method based on tensor trains. The
to-be-completed tensor is modeled as a low-rank tensor train, where we use the
known tensor entries and their coordinates to update the tensor train. A novel
tensor train initialization procedure is proposed specifically for image and
video completion, which is demonstrated to ensure fast convergence of the
completion algorithm. The tensor train framework is also shown to easily
accommodate Total Variation and Tikhonov regularization due to their low-rank
tensor train representations. Image and video inpainting experiments verify the
superiority of the proposed scheme in terms of both speed and scalability,
where a speedup of up to 155X is observed compared to state-of-the-art tensor
completion methods at a similar accuracy. Moreover, we demonstrate the proposed
scheme is especially advantageous over existing algorithms when only tiny
portions (say, 1%) of the to-be-completed images/videos are known. | ['Ching-Yun Ko', 'Kim Batselier', 'Ngai Wong', 'Wenjian Yu'] | 2018-04-17 | null | null | null | null | ['video-inpainting'] | ['computer-vision'] | [-1.35616705e-01 -3.46631229e-01 -4.48891446e-02 -3.93820591e-02
-8.64802241e-01 -5.98984718e-01 4.52467293e-01 -2.18674317e-01
-5.03426135e-01 3.73211652e-01 3.59532982e-01 -1.55545130e-01
-3.13937098e-01 2.86281258e-02 -8.22855830e-01 -7.41455317e-01
-4.07641411e-01 3.57868284e-01 -1.64445892e-01 -1.79928139e-01
1.29068181e-01 3.59571457e-01 -1.19447732e+00 1.78718850e-01
7.75825024e-01 1.08837044e+00 1.40204176e-01 8.82473230e-01
2.60606240e-02 9.57421601e-01 -1.33108541e-01 -4.97453004e-01
5.31814635e-01 2.48146817e-01 -1.00862777e+00 6.32186174e-01
5.66690207e-01 -8.74844968e-01 -6.83770180e-01 7.72943497e-01
-1.04496861e-02 2.55641997e-01 3.54011178e-01 -1.00628626e+00
-4.98899788e-01 2.88623393e-01 -6.88751042e-01 1.16600707e-01
3.60601246e-01 -2.53445596e-01 1.22379243e+00 -1.57799613e+00
5.70166826e-01 9.96865988e-01 5.46509683e-01 4.27664146e-02
-1.29708457e+00 -3.88601184e-01 3.90603654e-02 3.07018310e-01
-1.41753781e+00 -4.06023353e-01 8.59664023e-01 -6.40725195e-01
5.31336427e-01 3.09475094e-01 6.10202551e-01 6.12656355e-01
1.25443995e-01 9.26590919e-01 9.66454804e-01 -1.58756465e-01
-1.35528855e-04 -3.61050189e-01 1.27614796e-01 7.82165229e-01
1.90797180e-01 -2.99891949e-01 -5.64535320e-01 -4.00204092e-01
1.00585496e+00 3.90049607e-01 -1.60369441e-01 -4.19601887e-01
-1.93660426e+00 6.57104552e-01 4.15365130e-01 1.73384964e-01
-6.66509807e-01 4.05787587e-01 6.85570240e-01 2.84529537e-01
5.55357158e-01 1.02620274e-01 -1.34790018e-01 -2.45722249e-01
-9.62316036e-01 2.97069728e-01 5.89189827e-01 1.20464838e+00
1.03338099e+00 3.74110162e-01 -1.32726133e-01 8.35886419e-01
2.73987986e-02 5.01466990e-01 1.46242753e-01 -1.17112458e+00
7.15347767e-01 3.57273608e-01 3.48450243e-01 -1.20537221e+00
-2.41955183e-02 -2.42581695e-01 -1.09340537e+00 -3.56819600e-01
2.81614512e-01 2.32988372e-01 -7.35484898e-01 1.26126969e+00
4.93367821e-01 5.69384754e-01 -1.34536266e-01 1.01819766e+00
4.49045449e-01 6.51732981e-01 -4.42947954e-01 -2.78176993e-01
1.53424728e+00 -8.90293479e-01 -8.17713916e-01 2.15621948e-01
6.69993520e-01 -1.06294990e+00 8.16924393e-01 4.94407505e-01
-1.09637892e+00 -4.60038662e-01 -9.32248473e-01 -3.74049634e-01
3.37490469e-01 4.22716528e-01 1.02905381e+00 3.48835856e-01
-1.00295424e+00 6.37760401e-01 -1.15024304e+00 1.10496944e-02
1.96526304e-01 4.25153673e-01 -7.72515297e-01 -5.65850675e-01
-5.70682466e-01 3.06224108e-01 8.46407712e-02 5.56813180e-01
-1.28247762e+00 -7.89504349e-01 -7.02980578e-01 -2.45801359e-01
2.21031651e-01 -6.90560997e-01 1.08844769e+00 -5.59594035e-01
-1.33118093e+00 3.52959305e-01 -4.17770386e-01 -1.03843965e-01
4.10104841e-01 -4.44344997e-01 -4.71604690e-02 6.38607562e-01
2.24880382e-01 2.68146008e-01 1.33663177e+00 -1.16131341e+00
-3.65954190e-01 -2.60934770e-01 2.44100764e-01 2.34245807e-01
-7.07077980e-01 2.77007706e-02 -8.65347624e-01 -9.88882244e-01
5.50751865e-01 -1.23107028e+00 -4.69329178e-01 1.19950749e-01
-3.91627759e-01 7.37103075e-02 6.93617821e-01 -1.05831790e+00
1.02166927e+00 -2.07515717e+00 4.62452739e-01 1.63876295e-01
7.39028215e-01 -3.16755660e-03 -2.91641027e-01 7.87799835e-01
-1.84118867e-01 -1.54572636e-01 -1.30567461e-01 -8.35925937e-01
-2.44427875e-01 6.44817472e-01 -4.62856323e-01 7.80114472e-01
1.80351496e-01 4.73652899e-01 -1.00367594e+00 -3.26246053e-01
2.47806311e-01 7.25470781e-01 -5.82177341e-01 3.29098344e-01
2.33373061e-01 7.52354681e-01 -6.59299910e-01 5.70954859e-01
8.90714288e-01 -2.18753427e-01 1.70442313e-01 -7.29375839e-01
-2.16603801e-01 1.44206479e-01 -1.21632540e+00 1.90829635e+00
-2.88233072e-01 4.35872942e-01 2.53645360e-01 -9.39449728e-01
5.39786756e-01 6.03919625e-01 9.34758008e-01 -2.54417330e-01
1.38713960e-02 3.74878824e-01 -2.29430109e-01 -3.18697244e-01
9.96832848e-01 9.23471525e-02 3.33481669e-01 5.49021363e-01
-1.28471088e-02 1.25941232e-01 6.62087560e-01 6.94984317e-01
1.18511415e+00 2.08903953e-01 -2.34546557e-01 -2.81248391e-01
4.22177613e-01 -1.50992423e-01 5.75133979e-01 4.61011559e-01
1.33931026e-01 4.86281902e-01 4.51837093e-01 -5.86393595e-01
-1.47142541e+00 -8.08375418e-01 -1.10840589e-01 9.76992846e-01
-1.39043555e-01 -7.44823754e-01 -6.28926516e-01 -3.13050389e-01
-2.24802375e-01 -3.08110807e-02 -6.15344942e-01 3.73165220e-01
-7.23017752e-01 -4.83733237e-01 3.65976512e-01 4.26318109e-01
4.07302052e-01 -1.41673535e-01 5.55026047e-02 2.02734023e-01
-4.76520181e-01 -1.78303075e+00 -8.93412769e-01 -4.51960266e-01
-1.44072294e+00 -1.07259965e+00 -9.94524717e-01 -6.46199584e-01
1.12664747e+00 9.41152632e-01 6.49612248e-01 1.02258272e-01
3.30439471e-02 6.35477424e-01 -5.70932508e-01 5.60479701e-01
-2.56189972e-01 -3.08097631e-01 2.95726538e-01 6.79617107e-01
-3.21518391e-01 -7.96387196e-01 -8.83137584e-01 4.24191147e-01
-1.33338010e+00 2.42642850e-01 5.04525661e-01 1.12178147e+00
6.76604927e-01 -1.42971531e-01 -2.47007355e-01 -6.91052139e-01
3.55176538e-01 -1.32888302e-01 -5.18301964e-01 1.83724687e-02
-4.30977494e-01 1.85103908e-01 6.40695035e-01 -5.98936856e-01
-7.70650864e-01 2.43963599e-01 3.05480570e-01 -1.05740106e+00
6.28179431e-01 7.16820478e-01 3.72093737e-01 -5.47773242e-01
2.24044025e-01 3.11746806e-01 1.59752041e-01 -7.79156208e-01
5.67696035e-01 6.82914034e-02 5.07981718e-01 -9.20620382e-01
1.32968938e+00 9.23552632e-01 1.41147688e-01 -9.57435489e-01
-6.94514453e-01 -8.47199082e-01 -8.14236403e-01 -2.03514770e-01
3.88488531e-01 -1.22705090e+00 -9.69672382e-01 4.22839314e-01
-1.18105829e+00 1.25280142e-01 -3.42180431e-02 9.75563586e-01
-4.57405865e-01 1.01933920e+00 -1.18469334e+00 -6.49633944e-01
-3.15610379e-01 -1.27255988e+00 1.05802584e+00 -5.80908775e-01
3.77359837e-01 -1.00381792e+00 -4.70194295e-02 6.02473021e-01
1.69525430e-01 1.80937394e-01 5.85344136e-01 -2.79311109e-02
-9.28060174e-01 -3.15728813e-01 -4.02938396e-01 6.32076502e-01
1.09868711e-02 -7.61637017e-02 -7.04586148e-01 -8.16395998e-01
1.22709321e-02 -1.77932411e-01 8.02503943e-01 9.91990417e-02
8.61088395e-01 -5.42343676e-01 6.86647221e-02 7.89953709e-01
1.31456721e+00 -4.88090396e-01 7.41638362e-01 9.25112050e-03
1.32913065e+00 2.96549737e-01 6.96605146e-01 8.33796382e-01
5.13747394e-01 7.35551655e-01 3.78616244e-01 -1.25641912e-01
-1.22227043e-01 -1.11286588e-01 5.04551053e-01 1.90469635e+00
-6.13559723e-01 5.95758557e-01 -6.12255633e-01 7.43223488e-01
-1.89929199e+00 -7.86173999e-01 -5.42403519e-01 2.50202227e+00
7.40408599e-01 -2.57108897e-01 1.20958135e-01 4.01851565e-01
4.05213326e-01 2.82211214e-01 -1.55656606e-01 -5.96344396e-02
9.09532160e-02 1.25525102e-01 7.07323551e-01 5.94115794e-01
-1.03872287e+00 8.36507857e-01 6.72420454e+00 8.06133330e-01
-9.33098674e-01 2.88187623e-01 1.08754121e-01 2.43261635e-01
-3.86786222e-01 4.54744458e-01 -3.20649385e-01 7.63377547e-03
5.40587902e-01 -4.45315689e-02 8.03264737e-01 6.95530176e-01
3.33036214e-01 1.92073554e-01 -9.90932941e-01 1.19611025e+00
1.62967846e-01 -1.36160409e+00 2.62481719e-01 3.33293706e-01
8.60937119e-01 -1.37955854e-02 9.58105922e-02 9.35355797e-02
-5.50674228e-03 -5.17376423e-01 6.65152252e-01 3.31360102e-01
8.88600886e-01 -4.88397002e-01 3.87145907e-01 -1.26802951e-01
-1.59239316e+00 5.94485775e-02 -6.59469426e-01 -2.76146173e-01
1.17049456e-01 9.22321498e-01 -1.00101054e+00 9.52811599e-01
5.83480954e-01 9.91686344e-01 -3.93158764e-01 9.82705176e-01
3.28107104e-02 7.86488354e-01 -3.99879903e-01 4.73904401e-01
4.43544060e-01 -6.50292695e-01 6.44592881e-01 9.51632500e-01
3.77895772e-01 1.20345101e-01 4.72818792e-01 3.68723422e-01
-6.58560768e-02 3.09068561e-01 -3.21747452e-01 -1.43963128e-01
-1.62078515e-02 1.48489428e+00 -4.75175917e-01 -3.42620045e-01
-4.69164610e-01 1.18837154e+00 2.44106516e-01 7.89996564e-01
-6.43125176e-01 -1.59121260e-01 7.76893318e-01 -9.83268991e-02
5.94760001e-01 -8.35087657e-01 -5.49008772e-02 -1.68784881e+00
5.05063534e-01 -7.15945244e-01 1.21557102e-01 -6.17146254e-01
-1.21624041e+00 6.44722402e-01 1.84361115e-02 -1.63259971e+00
-2.22764894e-01 -6.86802983e-01 -1.05091639e-01 4.98715848e-01
-1.46997261e+00 -1.64653218e+00 -3.43641549e-01 9.76853788e-01
2.75202155e-01 8.22644606e-02 7.53092647e-01 7.06835389e-01
-5.42420924e-01 5.26372313e-01 6.43381596e-01 2.47247785e-01
4.37172830e-01 -9.70626414e-01 3.21219176e-01 9.61364806e-01
9.00742710e-02 8.02594185e-01 6.58375442e-01 -5.04190266e-01
-2.31796122e+00 -1.01204717e+00 6.52608752e-01 -2.13688433e-01
1.11291480e+00 -4.02206272e-01 -7.62978137e-01 9.65095460e-01
-1.49936810e-01 2.65204400e-01 5.28896451e-01 2.58459508e-01
-9.98598814e-01 -5.04216909e-01 -6.25608265e-01 6.88678741e-01
7.57117629e-01 -7.56392300e-01 -2.40144551e-01 7.32743025e-01
7.06306994e-01 -5.75262249e-01 -1.28332794e+00 1.58668250e-01
5.71490824e-01 -6.83401346e-01 9.87806857e-01 -5.81240594e-01
3.97804230e-01 -6.46280944e-01 -4.47248966e-01 -8.85389507e-01
-4.43155020e-01 -1.21550441e+00 -2.88305461e-01 9.20538485e-01
3.60363275e-02 -4.73949015e-01 6.27928436e-01 5.24716318e-01
-3.35994482e-01 -4.99844134e-01 -1.13444459e+00 -8.08471203e-01
-4.63465333e-01 -7.39263356e-01 2.61924654e-01 1.05034089e+00
-1.16509058e-01 1.90724209e-01 -1.02853119e+00 3.01375479e-01
9.65202808e-01 2.81739496e-02 9.46160138e-01 -7.54419923e-01
-3.31409484e-01 1.84989035e-01 -5.84583282e-01 -1.59467471e+00
-3.03326230e-02 -7.83133626e-01 -2.51507342e-01 -1.31094885e+00
4.56893444e-01 -4.99089450e-01 -3.49372894e-01 5.14088929e-01
-4.71208431e-02 6.35861099e-01 3.79618347e-01 7.30941713e-01
-5.67642033e-01 8.14093292e-01 1.58094990e+00 -1.56486750e-01
9.13857073e-02 -2.07627073e-01 -3.32241476e-01 3.22850198e-01
5.79315424e-02 -3.77939612e-01 -3.47643524e-01 -7.53332376e-01
1.98295370e-01 3.67406666e-01 2.05615610e-01 -7.73611486e-01
1.69798493e-01 -1.53302560e-02 3.04463506e-02 -4.76390809e-01
8.26408684e-01 -7.33556569e-01 2.61950284e-01 2.22880304e-01
-6.50989711e-02 2.96426862e-01 1.10931350e-02 7.36544371e-01
-2.63558388e-01 5.33168092e-02 2.52537072e-01 3.13135356e-01
-3.44913423e-01 7.78979242e-01 -2.41764441e-01 -2.49366224e-01
6.32459402e-01 -2.06666142e-01 -7.06736147e-02 -6.22024119e-01
-5.94114363e-01 -9.32435784e-03 3.72753203e-01 1.36352643e-01
9.38154280e-01 -1.66908169e+00 -9.28665102e-01 6.99756369e-02
2.54300177e-01 -1.40429065e-01 6.18561685e-01 1.38454831e+00
-7.71339774e-01 3.12753469e-01 -1.43249277e-02 -8.13810945e-01
-1.09112406e+00 6.86263561e-01 -2.64643371e-01 -4.40131068e-01
-7.84421682e-01 4.46146846e-01 3.59387726e-01 -1.63520262e-01
6.72600791e-02 -4.46942300e-01 2.03970149e-01 -3.13403010e-01
7.22087681e-01 4.37296510e-01 1.93870708e-01 -9.69497263e-01
-1.51049003e-01 6.81886733e-01 -3.38178277e-01 -1.66365921e-01
1.37173510e+00 -1.06257526e-02 -5.66003323e-01 4.03914988e-01
1.46606708e+00 1.80110782e-01 -1.28217268e+00 -3.77918422e-01
-2.55316108e-01 -8.21434319e-01 2.14154124e-01 8.94523039e-02
-1.12045193e+00 7.32952774e-01 1.34068027e-01 -6.77028000e-02
1.11930120e+00 -4.01256621e-01 1.04361880e+00 5.93419611e-01
4.10460949e-01 -7.27344871e-01 3.47293168e-01 5.20547152e-01
1.02545643e+00 -9.41945314e-01 4.59311664e-01 -6.99408948e-01
-5.82866251e-01 1.13725197e+00 -1.66592151e-01 -4.61611897e-01
5.77740431e-01 -2.79704005e-01 -1.81647435e-01 -7.32755885e-02
-7.35106349e-01 -2.36930493e-02 5.63590705e-01 3.35123330e-01
4.09375727e-01 5.90064041e-02 -2.01160535e-01 -2.88598895e-01
8.61628652e-02 -1.64046571e-01 6.92753077e-01 7.60392368e-01
1.63312525e-01 -1.25040531e+00 -6.34327948e-01 2.99854428e-01
-4.83963162e-01 -1.61523029e-01 2.34156385e-01 4.97447461e-01
-5.49768627e-01 9.29718673e-01 -1.82079777e-01 -6.38054848e-01
8.31297189e-02 -4.20793742e-01 5.06927848e-01 -2.33324394e-01
-4.38099593e-01 5.01503289e-01 1.63780525e-01 -8.96624207e-01
-6.14463508e-01 -5.73820651e-01 -1.01486862e+00 -6.18614793e-01
-2.03602403e-01 2.43272826e-01 8.35705101e-01 8.59205782e-01
4.25420374e-01 7.82057047e-02 1.10160625e+00 -1.22350395e+00
-6.11777425e-01 -9.15943086e-01 -7.80842841e-01 6.63233221e-01
4.08565491e-01 -6.78532720e-01 -3.25312197e-01 3.55422646e-01] | [7.3702898025512695, 4.525243282318115] |
648fb307-3a04-43df-a01c-f5f253453a3f | bore-bayesian-optimization-by-density-ratio | 2102.09009 | null | https://arxiv.org/abs/2102.09009v1 | https://arxiv.org/pdf/2102.09009v1.pdf | BORE: Bayesian Optimization by Density-Ratio Estimation | Bayesian optimization (BO) is among the most effective and widely-used blackbox optimization methods. BO proposes solutions according to an explore-exploit trade-off criterion encoded in an acquisition function, many of which are computed from the posterior predictive of a probabilistic surrogate model. Prevalent among these is the expected improvement (EI) function. The need to ensure analytical tractability of the predictive often poses limitations that can hinder the efficiency and applicability of BO. In this paper, we cast the computation of EI as a binary classification problem, building on the link between class-probability estimation and density-ratio estimation, and the lesser-known link between density-ratios and EI. By circumventing the tractability constraints, this reformulation provides numerous advantages, not least in terms of expressiveness, versatility, and scalability. | ['Fabio Ramos', 'Cedric Archambeau', 'Edwin V. Bonilla', 'Matthias Seeger', 'Aaron Klein', 'Louis C. Tiao'] | 2021-02-17 | null | null | null | null | ['density-ratio-estimation'] | ['methodology'] | [ 2.24638298e-01 6.15392588e-02 -4.21720058e-01 -2.77089447e-01
-7.84509182e-01 -7.01044559e-01 7.28782475e-01 1.59164101e-01
-4.78069663e-01 1.01619303e+00 -1.41606972e-01 -4.26945984e-01
-8.30743313e-01 -7.83561647e-01 -4.77991968e-01 -8.08795571e-01
2.38625258e-02 3.74087125e-01 -1.99731752e-01 1.43487751e-01
3.83403778e-01 6.91668391e-01 -1.56659234e+00 -5.20913064e-01
1.13148999e+00 1.44061613e+00 -1.21362077e-03 3.46790463e-01
4.07256857e-02 4.25046593e-01 -4.39860255e-01 -7.83931792e-01
1.08354814e-01 -3.36503178e-01 -5.24485588e-01 -2.30898947e-01
-1.36042669e-01 -1.61701411e-01 1.13903277e-01 1.13730884e+00
1.95478275e-01 2.61627257e-01 9.38430667e-01 -1.43352973e+00
-1.43110573e-01 2.98141003e-01 -3.60670418e-01 8.66285712e-02
3.03681999e-01 6.06936663e-02 1.35640824e+00 -6.30036354e-01
3.89862329e-01 1.01474392e+00 5.95025361e-01 1.03606544e-02
-1.50270963e+00 -3.28372866e-01 -1.49236824e-02 2.25810353e-02
-1.65057826e+00 -4.73903865e-01 4.65857655e-01 -5.84545374e-01
5.76393664e-01 5.01104712e-01 5.85955918e-01 5.86487293e-01
3.01020890e-01 7.29742944e-01 1.13114536e+00 -3.81533056e-01
4.89743710e-01 4.09739405e-01 1.25242525e-03 4.49775845e-01
5.36120057e-01 5.47792077e-01 -5.17019987e-01 -2.71466881e-01
5.50868988e-01 -2.78916329e-01 -4.31818336e-01 -6.87561214e-01
-6.34262741e-01 9.36892807e-01 2.03367636e-01 1.16344109e-01
-2.03772962e-01 2.16618076e-01 1.03165992e-01 7.05494778e-03
4.45822835e-01 6.42868757e-01 -2.35152215e-01 -3.80695105e-01
-1.14781773e+00 4.85897034e-01 8.45453262e-01 7.08650649e-01
7.69182265e-01 1.24746729e-02 -5.51162772e-02 4.01983798e-01
7.09803879e-01 4.44771230e-01 1.98156342e-01 -7.64516830e-01
1.84798002e-01 3.51594180e-01 2.61973590e-01 -9.85547423e-01
-1.93885669e-01 -8.42211306e-01 -3.81281883e-01 3.45080227e-01
6.73413873e-01 7.87260830e-02 -4.89153296e-01 1.69571424e+00
4.52424914e-01 -3.22127789e-01 -2.24891424e-01 7.05785573e-01
1.43766582e-01 5.46091557e-01 -3.17853168e-02 -5.12219787e-01
9.42055941e-01 -3.83348525e-01 -5.76081157e-01 -7.97438845e-02
2.68285275e-01 -5.32979190e-01 9.47108626e-01 5.51811218e-01
-1.20654190e+00 -8.88444409e-02 -1.21188402e+00 2.49321714e-01
-1.71531320e-01 -1.90811660e-02 8.37328017e-01 1.28706479e+00
-7.70816386e-01 7.93936312e-01 -7.63501942e-01 8.97419378e-02
2.80788720e-01 4.78731036e-01 -1.36133417e-01 1.51546821e-01
-7.54633665e-01 1.08123279e+00 4.12019402e-01 2.26450115e-01
-6.94345415e-01 -7.90071368e-01 -8.45368922e-01 2.93875277e-01
5.85396945e-01 -5.86780190e-01 1.16793239e+00 -8.04295659e-01
-1.77150106e+00 4.93013054e-01 2.57352348e-02 -4.62406188e-01
8.32142770e-01 -2.80573368e-01 -3.13397422e-02 -2.95106098e-02
-1.07699573e-01 1.53297186e-01 8.27050328e-01 -1.01099229e+00
-5.50805032e-01 -3.67847770e-01 2.19494058e-03 7.28952810e-02
-1.43017963e-01 -2.00092942e-01 -3.55389088e-01 -4.87135023e-01
-7.27623627e-02 -6.12547219e-01 -1.81621194e-01 1.30691782e-01
-5.88929594e-01 6.40314296e-02 1.33188218e-01 -5.72779477e-01
1.56156647e+00 -2.07419562e+00 1.74443468e-01 6.59831882e-01
-5.00523811e-03 -1.76944479e-01 2.04636961e-01 3.54351342e-01
1.47024706e-01 1.56770676e-01 -6.00399017e-01 -1.52635157e-01
3.13747674e-01 3.63024510e-02 -3.76497805e-01 9.53187823e-01
4.56661999e-01 6.64969504e-01 -8.62470925e-01 -3.70158255e-01
2.96889663e-01 2.92178482e-01 -6.94836915e-01 2.00128689e-01
-3.04330558e-01 3.79450411e-01 -4.88830954e-01 5.90666115e-01
5.29815555e-01 -9.13900584e-02 2.53744185e-01 -1.03376843e-01
-2.07339466e-01 3.44877541e-01 -1.30943596e+00 1.25733924e+00
-3.78085971e-01 3.65600973e-01 1.81942776e-01 -1.13538659e+00
7.71312952e-01 -2.09451914e-01 5.33382118e-01 -3.75010431e-01
2.32809901e-01 3.12715858e-01 1.11314273e-02 -1.38928607e-01
4.87012058e-01 -4.06463534e-01 -1.46133348e-01 3.53191227e-01
1.30192488e-01 -4.18468028e-01 3.49003911e-01 -2.23326981e-01
6.84333503e-01 5.62537432e-01 7.66387403e-01 -6.20515049e-01
3.37577254e-01 -1.34656385e-01 3.99588645e-01 7.93141186e-01
1.59094390e-02 3.94547492e-01 8.10249925e-01 1.56307027e-01
-8.81980300e-01 -1.39042115e+00 -6.50280535e-01 6.55513346e-01
-4.78070835e-03 -3.47694665e-01 -4.48082387e-01 -5.98042607e-01
3.17572385e-01 1.02005243e+00 -4.79020298e-01 -1.68672457e-01
-2.52556562e-01 -9.96955752e-01 2.14982316e-01 3.01625103e-01
5.84681854e-02 -2.67949611e-01 -8.65696549e-01 1.45095676e-01
2.32926488e-01 -7.18814850e-01 -1.40834227e-01 3.59448612e-01
-8.75494003e-01 -1.01167047e+00 -7.35567510e-01 2.57758778e-02
3.63287628e-01 -3.15177798e-01 1.07561493e+00 -2.36915618e-01
-1.93333238e-01 5.20928204e-01 -5.27431965e-02 -3.60050410e-01
-3.76317084e-01 -9.46285650e-02 -6.49470314e-02 2.57669352e-02
-1.24642260e-01 -6.47707939e-01 -2.61272937e-01 2.62135267e-01
-8.08036029e-01 -1.16844118e-01 5.77709734e-01 9.68722284e-01
7.19561100e-01 1.21828750e-01 5.11135757e-01 -5.20255625e-01
4.58531350e-01 -4.62829202e-01 -1.31571186e+00 4.61261302e-01
-9.92229342e-01 4.38780367e-01 3.78904015e-01 -2.69661129e-01
-1.01876462e+00 -7.12593645e-03 -2.40472257e-02 -2.05688030e-01
1.88843220e-01 7.93737113e-01 -3.96892428e-01 -6.88902587e-02
3.12122405e-01 6.58997521e-02 5.77385686e-02 -4.78840023e-01
5.31592667e-01 4.30124491e-01 5.84549129e-01 -8.96915555e-01
6.89995110e-01 2.66234905e-01 4.22623187e-01 -7.44999230e-01
-8.60424578e-01 -2.45903313e-01 -2.25735530e-01 -3.26991677e-01
5.99257767e-01 -3.35030973e-01 -9.24448609e-01 -6.30669892e-02
-7.76640356e-01 -1.41397044e-01 -5.38077176e-01 6.28189087e-01
-7.57080197e-01 3.42501163e-01 4.03137989e-02 -1.38112593e+00
3.68031338e-02 -1.27254200e+00 7.42954075e-01 2.84504652e-01
-2.88764536e-01 -9.92290378e-01 -8.77670199e-02 1.57641247e-01
5.08956134e-01 4.10214424e-01 1.11120248e+00 -4.36015874e-01
-6.21959150e-01 -4.22675580e-01 -2.84203202e-01 3.01735312e-01
-1.39382392e-01 1.53143272e-01 -9.23068643e-01 -2.29375035e-01
3.02138329e-01 -7.91179389e-02 6.95818424e-01 6.54372633e-01
1.15664542e+00 -3.35719138e-01 -2.60708332e-01 7.09751308e-01
1.59261537e+00 2.26898476e-01 4.12432015e-01 3.33219886e-01
2.15095326e-01 6.40861928e-01 5.38466156e-01 7.09776700e-01
1.39631527e-02 8.12698483e-01 4.34607476e-01 4.68504727e-01
3.00838858e-01 -2.09638238e-01 3.62961888e-01 3.62671971e-01
7.10437000e-02 -2.54066378e-01 -9.21256721e-01 2.99781770e-01
-1.73901331e+00 -7.96061218e-01 2.32996210e-01 2.72432065e+00
8.38667214e-01 2.65032202e-01 2.74128616e-01 1.55101523e-01
5.80026329e-01 -3.17598879e-02 -4.81646895e-01 -2.64441848e-01
1.01327673e-01 2.65571296e-01 5.43434262e-01 5.76651335e-01
-9.40011799e-01 2.16763273e-01 6.91611433e+00 1.06537545e+00
-8.53855371e-01 -1.31340213e-02 8.06086242e-01 -1.05464466e-01
-5.37875235e-01 2.12522492e-01 -8.65197301e-01 5.48211753e-01
9.39608693e-01 -4.31511313e-01 6.67159677e-01 8.88577163e-01
-5.35710119e-02 -5.60868382e-01 -1.25735629e+00 7.84860313e-01
-7.13535622e-02 -1.05819833e+00 -2.98680693e-01 3.22070628e-01
5.06461143e-01 -3.98638010e-01 3.29578102e-01 1.91301435e-01
7.74507448e-02 -1.27212274e+00 1.15873766e+00 5.72202265e-01
8.72219741e-01 -8.34051847e-01 6.19858325e-01 1.71614975e-01
-7.67171741e-01 -1.84792623e-01 -2.46856511e-02 3.72246876e-02
3.90686631e-01 9.30069983e-01 -5.48959911e-01 7.36729980e-01
4.25052583e-01 3.11866224e-01 -5.51107749e-02 1.36914527e+00
-2.95532942e-01 4.85745758e-01 -6.85416341e-01 -1.88086569e-01
1.30747512e-01 -7.62899697e-01 8.27301562e-01 1.11044335e+00
3.82876277e-01 -4.04355288e-01 -1.50033101e-01 1.36950660e+00
3.22890818e-01 4.03583013e-02 -3.15543443e-01 -3.64609480e-01
4.67526942e-01 9.39743221e-01 -6.77505314e-01 -3.27818990e-02
-2.12120011e-01 2.08054110e-01 1.32624865e-01 2.25046217e-01
-8.64256501e-01 -2.36415833e-01 4.63954479e-01 -1.00805141e-01
3.75728339e-01 -1.89509049e-01 -7.69717455e-01 -8.16767752e-01
2.01754823e-01 -8.97523820e-01 4.02220935e-01 -3.67468953e-01
-1.26868629e+00 2.46207178e-01 6.30348444e-01 -1.01321900e+00
-4.40224171e-01 -8.08938324e-01 -3.37492466e-01 9.31565642e-01
-1.44340444e+00 -7.87881315e-01 5.78494444e-02 1.17130511e-01
7.98285753e-02 1.98072195e-02 5.76665103e-01 1.63420781e-01
-7.31126308e-01 6.49780273e-01 3.90679806e-01 -5.08381784e-01
6.89737424e-02 -1.30345654e+00 -2.16689363e-01 7.88199425e-01
-7.10741710e-03 6.34691775e-01 1.03727531e+00 -4.26905155e-01
-1.42803895e+00 -5.76685488e-01 6.48981869e-01 -3.22274178e-01
8.21898103e-01 -1.27521113e-01 -4.50376064e-01 2.02961460e-01
-4.32727516e-01 -3.47253472e-01 5.11746347e-01 3.74870121e-01
-1.46164969e-01 -1.75144762e-01 -1.11434126e+00 5.69925487e-01
5.55263579e-01 -4.25345868e-01 -4.73489493e-01 8.81707221e-02
1.85437858e-01 -2.82684982e-01 -1.05572569e+00 4.53909665e-01
7.92445838e-01 -1.01043808e+00 9.83033895e-01 -3.89324039e-01
4.15456116e-01 -1.46882296e-01 -4.80644971e-01 -9.42666650e-01
1.66577861e-01 -8.95504057e-01 -2.92203695e-01 1.24144399e+00
4.65432584e-01 -8.63781035e-01 6.76336706e-01 8.52870524e-01
2.88466434e-03 -9.69028056e-01 -1.26985860e+00 -1.20171309e+00
5.47221825e-02 -7.24227250e-01 7.44768500e-01 5.04888415e-01
3.02339662e-02 1.67325988e-01 -2.60971516e-01 -8.22386816e-02
6.46914601e-01 1.61374539e-01 6.07362151e-01 -1.22775137e+00
-7.98186123e-01 -8.61447394e-01 -3.86278719e-01 -1.11173928e+00
-6.27950430e-02 -7.29550660e-01 1.60414129e-01 -1.00452375e+00
6.57563731e-02 -6.14409924e-01 -2.91887850e-01 1.48690119e-01
-5.34257479e-02 -1.00280501e-01 2.04214126e-01 2.06757337e-01
-2.07016647e-01 8.22858930e-01 9.13591802e-01 1.78512007e-01
-2.44947195e-01 2.64656246e-01 -5.25478363e-01 7.15759695e-01
5.49804807e-01 -6.81546211e-01 -2.98736095e-01 6.43532649e-02
5.52650392e-01 1.87562138e-01 3.48278731e-01 -7.32387722e-01
-6.65295422e-02 -3.43277544e-01 1.47860616e-01 -4.67684507e-01
4.82002646e-01 -6.14822388e-01 3.77844423e-01 4.18062419e-01
-2.78445154e-01 3.63591425e-02 -1.62048005e-02 6.07483208e-01
-1.46192461e-01 -7.19662249e-01 8.10211658e-01 1.81331992e-01
-3.62887740e-01 -3.42083946e-02 -2.15797544e-01 -1.73039865e-02
9.03420866e-01 -2.79997885e-01 -1.59857154e-01 -3.55947942e-01
-4.30387855e-01 -5.77761754e-02 4.09485042e-01 -1.75188199e-01
1.95496947e-01 -9.90198612e-01 -4.13649619e-01 -1.25740264e-02
-9.71714482e-02 -1.45769402e-01 -1.55032650e-01 1.21863675e+00
-3.11762244e-01 4.37226504e-01 1.05952933e-01 -5.44341445e-01
-5.93323946e-01 4.60911810e-01 4.75911230e-01 -5.11677861e-01
-3.06365967e-01 7.05345035e-01 1.06056906e-01 -7.45682269e-02
3.09335291e-01 -2.09064066e-01 1.49474621e-01 -7.88110793e-02
3.05245280e-01 7.04095840e-01 -6.25633751e-04 -3.11563939e-01
-5.19018233e-01 1.85418069e-01 3.94142479e-01 -5.60325146e-01
1.15041745e+00 -5.74746393e-02 -7.83343390e-02 4.61574525e-01
8.73791099e-01 3.05308681e-02 -1.24264657e+00 7.80641362e-02
3.69133711e-01 -6.76893175e-01 3.88447672e-01 -9.05052722e-01
-6.37631893e-01 6.98457479e-01 4.33748394e-01 3.38445485e-01
1.04754794e+00 -2.42324173e-01 2.82370418e-01 1.96462661e-01
2.31091812e-01 -1.13609159e+00 -3.68977726e-01 1.80406764e-01
8.30527842e-01 -9.85986829e-01 4.80104923e-01 -4.16582644e-01
-2.95643061e-01 1.09672260e+00 1.16419926e-01 9.04280990e-02
6.17306828e-01 2.32793778e-01 -5.08193314e-01 -1.20953880e-02
-3.77269536e-01 -2.03906357e-01 6.97784901e-01 5.58398962e-01
2.99819857e-01 3.52488719e-02 -5.37487686e-01 7.54531980e-01
-2.31718615e-01 -2.70945340e-01 4.77782600e-02 7.30905831e-01
-1.86497226e-01 -1.02640295e+00 -3.47407132e-01 5.75231254e-01
-5.47467351e-01 -4.10147198e-02 4.12929319e-02 9.55100119e-01
-1.58245116e-01 9.03681636e-01 -3.40090357e-02 -6.04947805e-02
7.82915354e-02 -2.28147618e-02 4.81582135e-01 -2.70626932e-01
-3.33467036e-01 1.25266075e-01 3.48595351e-01 -5.52611351e-01
-2.58977473e-01 -7.86236882e-01 -5.53505480e-01 -2.55153865e-01
-8.93256843e-01 3.06136608e-01 9.90948558e-01 1.26713800e+00
6.14922531e-02 2.29856417e-01 5.64665318e-01 -8.71882498e-01
-1.32643282e+00 -5.75528204e-01 -6.37081504e-01 -5.19582368e-02
1.87241763e-01 -9.52913225e-01 -4.17262584e-01 -3.60288799e-01] | [6.401566982269287, 3.770442008972168] |
8e921f82-c418-4c7c-8315-43c6f829bf5e | deep-material-recognition-in-light-fields-via | null | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/4750_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123690647.pdf | Deep Material Recognition in Light-Fields via Disentanglement of Spatial and Angular Information | Light-field cameras capture sub-views from multiple perspectives simultaneously, with possibly reflectance variations that can be used to augment material recognition in remote sensing, autonomous driving, etc. Existing approaches for light-field based material recognition suffer from the entanglement between angular and spatial domains, leading to inefficient training which in turn limits their performances. In this paper, we propose an approach that achieves decoupling of angular and spatial information by establishing correspondences in the angular domain, then employs regularization to enforce a rotational invariance. As opposed to relying on the Lambertian surface assumption, we align the angular domain by estimating sub-pixel displacements using the Fourier transform. The network takes sparse inputs, i.e. sub-views along particular directions, to gain structural information about the angular domain. A novel regularization technique further improves generalization by weight sharing and max-pooling among different directions. The proposed approach outperforms any previously reported method on multiple datasets. The accuracy gain over 2D images is improved by a factor of 1.5. Extensive ablation studies are conducted to demonstrate the significance of each component. | ['Bichuan Guo', 'Yuxing Han', 'Jiangtao Wen'] | null | null | null | null | eccv-2020-8 | ['material-recognition'] | ['computer-vision'] | [ 7.52597809e-01 -2.29643732e-01 -1.08078003e-01 -2.94048965e-01
-5.14410853e-01 -6.49201572e-01 6.20724738e-01 -1.98056191e-01
-3.85612786e-01 7.80717611e-01 5.47366925e-02 7.57572204e-02
-3.88385475e-01 -8.64395916e-01 -7.91848421e-01 -1.00539112e+00
3.50255758e-01 4.00360161e-03 3.08712780e-01 5.08365501e-03
5.37997007e-01 9.56953049e-01 -1.67443132e+00 2.98510283e-01
5.82366586e-01 8.83262932e-01 3.53574932e-01 3.53144109e-01
8.62044562e-03 3.56523722e-01 -7.70860538e-02 9.98828411e-02
5.13705134e-01 3.56784910e-02 -5.78184903e-01 3.04868490e-01
1.03661788e+00 -4.58716810e-01 -2.86828727e-01 9.45702493e-01
3.15468311e-01 6.95766956e-02 5.86003602e-01 -7.36692548e-01
-6.86951101e-01 -3.59888375e-02 -7.20203161e-01 8.61837342e-03
4.64689314e-01 -2.33197436e-02 1.03334630e+00 -8.67067873e-01
8.25339854e-01 7.65665948e-01 5.65824270e-01 4.40604120e-01
-1.28004825e+00 -4.30061847e-01 2.55474955e-01 2.94171244e-01
-1.28350782e+00 -5.24055064e-01 1.26103997e+00 -4.95014697e-01
9.75531042e-01 2.35467970e-01 6.06579721e-01 8.61967146e-01
2.21568090e-03 5.54745734e-01 1.42876267e+00 -5.39403558e-01
4.80670929e-02 3.22171375e-02 -4.97657172e-02 5.57971597e-01
3.35006118e-01 1.61456600e-01 -7.75692999e-01 1.34064585e-01
9.25093412e-01 2.79456139e-01 -5.20387530e-01 -6.40311956e-01
-1.37363052e+00 4.68484640e-01 5.77567279e-01 2.99177498e-01
-4.38640535e-01 -1.63052492e-02 -2.40094721e-01 1.76642671e-01
5.64567804e-01 5.14159381e-01 -1.46333098e-01 3.70837718e-01
-7.51860023e-01 9.46564972e-03 3.30808550e-01 7.67614901e-01
1.06796050e+00 -4.31073904e-02 7.72150457e-02 7.85269439e-01
2.47525975e-01 8.44931960e-01 3.81569229e-02 -9.50124681e-01
5.00513017e-01 6.25353932e-01 3.03208768e-01 -9.51320529e-01
-5.96845865e-01 -2.79102921e-01 -8.12728643e-01 3.92625004e-01
6.43491387e-01 1.19552389e-01 -8.36630881e-01 1.61301279e+00
3.28956544e-01 1.32610396e-01 2.92434152e-02 1.03071642e+00
4.61651295e-01 3.46258402e-01 -4.89507943e-01 -1.22939095e-01
1.31931508e+00 -7.16500938e-01 -3.98750335e-01 -2.91095495e-01
1.72549412e-01 -8.54582310e-01 7.99889922e-01 4.61056203e-01
-9.79166508e-01 -2.98633039e-01 -1.09566522e+00 -3.75461802e-02
-3.90412688e-01 4.23213035e-01 5.59160829e-01 4.66492623e-01
-6.91121995e-01 3.78320903e-01 -8.49812686e-01 -3.15367639e-01
5.10277092e-01 4.29211110e-01 -5.24418771e-01 -3.43945146e-01
-6.33592367e-01 6.53403461e-01 -1.00117266e-01 1.26402497e-01
-3.25029552e-01 -7.54351258e-01 -6.04806244e-01 -2.07996562e-01
3.08466107e-01 -6.73704028e-01 3.62933010e-01 -5.94493508e-01
-1.60354972e+00 9.03586686e-01 -3.92109066e-01 -1.33026317e-01
4.00483787e-01 -2.12312460e-01 -3.08306545e-01 3.73731524e-01
-4.15519066e-02 5.33614516e-01 1.09689033e+00 -1.27737236e+00
-5.50228417e-01 -6.24750972e-01 3.89316529e-01 3.20653588e-01
-3.54428649e-01 -4.14567858e-01 -2.72533327e-01 -4.07636404e-01
7.45112717e-01 -1.06857574e+00 -6.72597662e-02 1.03588760e-01
-4.40834433e-01 1.62422001e-01 8.20894063e-01 -3.61714184e-01
6.64300382e-01 -2.01520371e+00 2.53315002e-01 2.40607515e-01
2.85851538e-01 3.14157344e-02 -1.08810738e-01 3.41570139e-01
7.79092833e-02 -3.91938478e-01 -3.67514461e-01 -7.32994005e-02
-4.29873109e-01 -5.51011153e-02 -1.87180787e-01 9.42346871e-01
3.31143737e-01 5.36023021e-01 -6.22625053e-01 -4.10558656e-02
4.74685729e-01 7.54661500e-01 -5.65261602e-01 -4.17014509e-02
-4.83221523e-02 8.56199980e-01 -4.59515721e-01 6.85585260e-01
1.08085442e+00 -2.01937854e-01 7.21780062e-02 -7.25268841e-01
-4.71042752e-01 1.52329832e-01 -1.30955744e+00 1.86271048e+00
-6.68478251e-01 6.71592474e-01 1.83536559e-01 -1.17569554e+00
9.95908201e-01 1.52728155e-01 7.13309765e-01 -7.56542444e-01
-1.93301275e-01 2.70123333e-01 -4.39548165e-01 -5.33113003e-01
3.81767243e-01 -2.29045361e-01 3.41158360e-01 3.36439461e-01
-1.14157446e-01 -1.20016232e-01 -4.13310081e-02 -4.70070064e-01
8.06528449e-01 2.82760918e-01 2.59877205e-01 -9.57016349e-02
7.93121934e-01 -2.23230466e-01 2.36616537e-01 5.51539600e-01
2.57125348e-01 7.10952640e-01 -3.38337272e-02 -6.91545427e-01
-8.93665135e-01 -9.70240772e-01 -3.75418127e-01 6.92259014e-01
5.18937171e-01 1.66989401e-01 -2.50335902e-01 -4.93702084e-01
5.54812364e-02 4.36045408e-01 -4.43390280e-01 9.41132978e-02
-7.11181939e-01 -7.74760664e-01 1.14079095e-01 4.18243825e-01
6.76710963e-01 -5.59878826e-01 -8.31894875e-01 -2.69819070e-02
-1.50327921e-01 -1.40764105e+00 -1.50635004e-01 6.71903566e-02
-9.78324354e-01 -1.00928855e+00 -7.68327117e-01 -5.70654094e-01
8.66064787e-01 8.07989180e-01 8.33450377e-01 -2.97943056e-01
-2.67178178e-01 5.28691828e-01 -1.17191173e-01 7.40667880e-02
2.10713491e-01 1.10281572e-01 5.52266203e-02 3.93956900e-01
4.34812099e-01 -1.02626371e+00 -6.63898885e-01 3.87378842e-01
-7.69324422e-01 2.33109146e-01 8.30279410e-01 9.43704903e-01
6.22160852e-01 -1.91282853e-01 2.95718998e-01 -9.00016308e-01
6.84215948e-02 -2.14383245e-01 -8.42223465e-01 2.05562353e-01
-6.20413482e-01 1.73255756e-01 4.93199021e-01 -2.51334339e-01
-1.24271536e+00 3.71967793e-01 1.35785908e-01 -2.74157077e-01
-5.66182435e-01 1.87664732e-01 -6.18144907e-02 -5.01861215e-01
6.33646488e-01 2.74155229e-01 -1.58353925e-01 -4.71169353e-01
3.28125119e-01 4.79597032e-01 2.95591503e-01 -4.25984621e-01
9.39801574e-01 1.07355583e+00 4.16364491e-01 -1.17052150e+00
-8.02063048e-01 -6.16304636e-01 -9.12156522e-01 -2.76327223e-01
7.76130378e-01 -9.81448650e-01 -7.14379728e-01 3.25306892e-01
-1.13779032e+00 6.71711043e-02 -1.97669595e-01 8.79873574e-01
-4.74240184e-01 3.19806814e-01 -2.52901107e-01 -7.90658593e-01
-4.81423317e-03 -9.02072787e-01 1.29693747e+00 1.10317126e-01
2.24065542e-01 -9.32405651e-01 -1.04354195e-01 5.97963810e-01
4.60548222e-01 2.05608774e-02 5.68096638e-01 -3.59795354e-02
-1.16078818e+00 -1.59745067e-01 -4.79505777e-01 2.03207061e-01
2.74205029e-01 -3.50431919e-01 -1.26084912e+00 -2.86463171e-01
1.51062787e-01 -1.28831834e-01 9.93158042e-01 5.07631600e-01
1.02825379e+00 -1.31866550e-02 -2.83629000e-01 8.83901834e-01
1.62870538e+00 3.07232179e-02 6.27759397e-01 4.26045030e-01
8.99671614e-01 7.56799817e-01 2.54468590e-01 4.17984009e-01
2.24073961e-01 7.94388473e-01 5.33950388e-01 -9.66780856e-02
-2.44397417e-01 8.10803175e-02 2.06139937e-01 4.73450661e-01
-7.18490541e-01 -6.28639534e-02 -6.67277753e-01 4.82591093e-01
-1.33848035e+00 -1.14012110e+00 -9.93517488e-02 2.38684845e+00
4.62032050e-01 -8.77951384e-02 -1.95714220e-01 7.40336850e-02
4.38134432e-01 4.94123131e-01 -6.65839851e-01 2.85151064e-01
-3.20644081e-01 1.13326199e-01 8.59175980e-01 8.34305465e-01
-9.78332996e-01 6.36813462e-01 5.79558086e+00 4.53401834e-01
-1.59485900e+00 -8.14796984e-02 1.57156326e-02 -1.59821123e-01
-5.96563816e-01 2.12131944e-02 -8.85495245e-01 5.09640723e-02
-3.20829707e-03 2.97674507e-01 7.56069183e-01 3.58110130e-01
5.59993424e-02 -1.06393017e-01 -9.57696915e-01 1.00838387e+00
2.30095729e-01 -1.35857773e+00 -3.34014259e-02 1.79128170e-01
8.30118001e-01 2.40978539e-01 5.46845682e-02 -5.83728909e-01
-1.22506119e-01 -6.68845713e-01 3.91475886e-01 7.59685993e-01
1.01670492e+00 -3.03407133e-01 4.81709719e-01 3.23980927e-01
-1.13409400e+00 -2.76769120e-02 -2.92685837e-01 -1.07328743e-01
1.87534541e-01 7.80854940e-01 -6.74764991e-01 7.08049297e-01
6.08927548e-01 1.04067338e+00 -5.00450075e-01 5.48063695e-01
-8.18304196e-02 2.85377562e-01 -6.47144616e-01 1.81055471e-01
6.72136666e-03 -6.52644098e-01 7.56267846e-01 8.37897599e-01
4.61554974e-01 -1.07311785e-01 6.74974173e-03 1.00052094e+00
8.16263258e-02 -2.53549248e-01 -8.72365475e-01 1.63364142e-01
2.71371990e-01 1.29923427e+00 -5.70088685e-01 -3.94333936e-02
-7.53329992e-01 9.02749240e-01 1.05596408e-01 5.82893074e-01
-4.45425391e-01 -3.29098433e-01 3.97058487e-01 3.67045403e-01
4.69726205e-01 -4.39657480e-01 -3.23370337e-01 -1.38737643e+00
4.60378915e-01 -3.97016138e-01 -1.22900575e-01 -6.30731940e-01
-1.25678754e+00 3.85975987e-01 4.48149927e-02 -1.58785450e+00
1.74889058e-01 -7.86193967e-01 -2.94517547e-01 9.53732610e-01
-2.04254794e+00 -1.38771439e+00 -6.12355769e-01 5.32008290e-01
2.41359770e-01 -5.87272272e-02 6.23706698e-01 3.10975969e-01
-2.08662555e-01 1.59498826e-01 1.82265267e-01 -2.51469880e-01
7.05326140e-01 -9.73742962e-01 -4.26559709e-02 7.94528306e-01
1.87595755e-01 5.53956211e-01 4.00061369e-01 -2.72040218e-01
-1.51616359e+00 -8.47393990e-01 6.44426823e-01 -3.32225591e-01
3.60959917e-01 -2.25907087e-01 -8.10280383e-01 5.30752957e-01
3.95334363e-02 2.16403544e-01 5.84904075e-01 3.62434387e-02
-5.18122733e-01 -4.96513188e-01 -1.12821150e+00 3.01400810e-01
1.27835274e+00 -8.38405192e-01 -3.28638703e-01 3.86549115e-01
1.59463733e-01 -2.84821689e-01 -8.25140238e-01 3.86657059e-01
8.03774834e-01 -1.30599201e+00 1.22141266e+00 -3.21566194e-01
3.05255234e-01 -5.41497529e-01 -5.20971239e-01 -1.05610144e+00
-2.88770676e-01 -3.84276271e-01 2.06299111e-01 1.02382445e+00
2.31593698e-01 -9.16179001e-01 9.62678015e-01 6.51058555e-02
-2.93977670e-02 -4.63124663e-01 -9.71598923e-01 -6.66746259e-01
-1.56509548e-01 -4.15600061e-01 3.98706585e-01 9.26814020e-01
-2.93333322e-01 4.89511341e-01 -4.50242937e-01 5.86552799e-01
8.20839047e-01 6.42250538e-01 5.51907480e-01 -1.31940496e+00
-2.74638653e-01 -3.62166584e-01 -3.38348955e-01 -1.45289803e+00
1.04820795e-01 -8.76171410e-01 -7.87399337e-02 -1.23711371e+00
-3.74291949e-02 -4.57219660e-01 -1.63218752e-01 2.37904310e-01
2.16396496e-01 6.46089375e-01 1.32158510e-02 4.09905970e-01
-2.69381344e-01 4.26243246e-01 1.38952184e+00 -1.97784066e-01
-1.76170826e-01 -4.06111963e-02 -4.63634372e-01 8.24853718e-01
8.53495717e-01 -1.32919937e-01 -3.17272425e-01 -7.90796697e-01
4.04428869e-01 -6.00418113e-02 5.95590174e-01 -1.24207628e+00
2.46136695e-01 -4.07576859e-01 3.99800330e-01 -4.73011583e-01
6.44530833e-01 -1.16331363e+00 1.62926927e-01 1.45034492e-01
-1.89829633e-01 -4.17680055e-01 -5.62005118e-02 8.05768073e-01
-1.84616283e-01 -1.06325060e-01 8.43360424e-01 -1.24503441e-01
-5.10014236e-01 2.21372858e-01 -2.34912694e-01 -3.47740889e-01
9.46437061e-01 -3.78540188e-01 -4.49809760e-01 -1.41383603e-01
-5.91510832e-01 -2.67552584e-01 7.78815210e-01 7.82767013e-02
5.53368270e-01 -1.34003425e+00 -4.89675790e-01 7.13962317e-01
2.59395778e-01 -1.15892487e-02 4.41456050e-01 1.18321526e+00
-5.20415187e-01 5.28939068e-01 -4.79794115e-01 -9.00878489e-01
-1.29297280e+00 2.32164696e-01 3.74632657e-01 -1.17421113e-01
-7.02352762e-01 6.81770623e-01 2.04070449e-01 -3.84822577e-01
-1.30315810e-01 -2.66080827e-01 -2.73967534e-01 -1.03773959e-02
3.17765921e-01 4.57469612e-01 2.09576532e-01 -7.34094024e-01
-5.24143040e-01 1.46854281e+00 1.17275164e-01 -8.72270167e-02
1.68073165e+00 -2.58875817e-01 -1.61442950e-01 3.79773915e-01
1.17747271e+00 4.85248119e-01 -1.45255876e+00 -5.77663302e-01
-2.61909038e-01 -6.51759386e-01 2.16173545e-01 -4.77871329e-01
-1.19091988e+00 1.11792898e+00 6.37028813e-01 2.56281763e-01
1.35750628e+00 8.45410600e-02 5.05935967e-01 6.05227590e-01
2.91488707e-01 -8.52366865e-01 -2.67121226e-01 2.83937842e-01
7.66109467e-01 -1.36031699e+00 3.21989894e-01 -6.22321486e-01
-1.58922508e-01 1.35338938e+00 3.05247068e-01 -1.12421267e-01
5.94316304e-01 1.28235325e-01 -5.64669035e-02 -3.22974056e-01
-3.50938648e-01 -1.87422276e-01 5.27866006e-01 6.67882502e-01
4.65217203e-01 -2.37733126e-02 -8.30800757e-02 -1.35686085e-01
1.86544746e-01 -1.54163111e-02 2.40759850e-01 8.66717100e-01
-3.71715873e-01 -1.15432775e+00 -4.48519081e-01 4.00213659e-01
-2.55181134e-01 1.07831962e-01 -1.56248644e-01 4.82801080e-01
3.09755534e-01 7.90994108e-01 2.32153848e-01 -2.62743413e-01
4.69100714e-01 -6.94622099e-02 8.46725464e-01 -3.51606786e-01
1.04703069e-01 1.68732718e-01 -3.63303628e-03 -6.50859356e-01
-1.14065671e+00 -8.56679857e-01 -8.07143807e-01 -5.84284514e-02
-4.00118470e-01 -2.84352988e-01 6.40452921e-01 8.97419810e-01
3.79301548e-01 2.13828832e-01 7.04021990e-01 -1.06419277e+00
-2.93300092e-01 -7.88173199e-01 -6.01193368e-01 4.53094870e-01
7.01006472e-01 -9.39560175e-01 -5.78215599e-01 2.42827833e-01] | [9.814162254333496, -2.7188754081726074] |
ec6b59e1-0d64-4d04-8ea5-a27e34233309 | an-exact-hypergraph-matching-algorithm-for | 2104.10003 | null | https://arxiv.org/abs/2104.10003v3 | https://arxiv.org/pdf/2104.10003v3.pdf | An Exact Hypergraph Matching Algorithm for Nuclear Identification in Embryonic Caenorhabditis elegans | Finding an optimal correspondence between point sets is a common task in computer vision. Existing techniques assume relatively simple relationships among points and do not guarantee an optimal match. We introduce an algorithm capable of exactly solving point set matching by modeling the task as hypergraph matching. The algorithm extends the classical branch and bound paradigm to select and aggregate vertices under a proposed decomposition of the multilinear objective function. The methodology is motivated by Caenorhabditis elegans, a model organism used frequently in developmental biology and neurobiology. The embryonic C. elegans contains seam cells that can act as fiducial markers allowing the identification of other nuclei during embryo development. The proposed algorithm identifies seam cells more accurately than established point-set matching methods, while providing a framework to approach other similarly complex point set matching tasks. | ['Radu Balan', 'Hari Shroff', 'Ryan Christensen', 'Andrew Lauziere'] | 2021-04-20 | null | null | null | null | ['set-matching', 'hypergraph-matching'] | ['computer-vision', 'graphs'] | [ 2.02074662e-01 1.14763431e-01 -2.66673505e-01 -1.32356778e-01
-4.25941765e-01 -8.33181858e-01 3.60881090e-01 6.69372141e-01
-5.25009692e-01 4.24901634e-01 -6.14262700e-01 -5.95333613e-02
-4.03704077e-01 -8.20008278e-01 -8.04377377e-01 -4.96414959e-01
-2.75206983e-01 9.05730605e-01 4.78342414e-01 -1.30110711e-01
6.01876676e-01 9.38061297e-01 -1.69072092e+00 -1.87314019e-01
7.70240128e-01 4.86334801e-01 3.88333082e-01 6.84938312e-01
-7.02484474e-02 -5.45724392e-01 -4.10273790e-01 -3.78824145e-01
6.02644145e-01 -4.28951204e-01 -7.56354332e-01 2.95494180e-02
7.69450724e-01 6.32129386e-02 2.45194107e-01 1.17171621e+00
2.81178921e-01 6.31998926e-02 7.71206379e-01 -1.68357098e+00
-3.39603424e-02 2.57773429e-01 -6.31803870e-01 3.91911790e-02
4.82934892e-01 -5.30094981e-01 1.00845957e+00 -7.20704556e-01
8.11678946e-01 1.14515686e+00 7.04529583e-01 5.39959967e-01
-1.38512862e+00 -3.43555897e-01 -3.83158296e-01 -3.22647542e-02
-1.45146453e+00 -1.79209158e-01 5.93596995e-01 -5.65608025e-01
4.40126538e-01 4.57027495e-01 9.65611398e-01 1.88443094e-01
2.70357937e-01 3.02899241e-01 6.61263406e-01 -6.63278043e-01
3.07645947e-01 -3.23733151e-01 2.11367399e-01 9.30116892e-01
6.98197842e-01 4.47243378e-02 -3.48489493e-01 -6.21137500e-01
1.26333094e+00 1.40857160e-01 -2.81119972e-01 -9.27454591e-01
-1.39265585e+00 7.05116153e-01 2.77378321e-01 4.63739067e-01
-1.12839742e-03 4.69897062e-01 1.55234769e-01 2.37775132e-01
-1.73518032e-01 4.67973918e-01 -5.20818941e-02 2.51497805e-01
-9.30452168e-01 4.18850392e-01 7.82736361e-01 9.54432905e-01
8.71906936e-01 -4.09056664e-01 3.02725971e-01 5.90676606e-01
3.93635899e-01 1.58331528e-01 -3.93304043e-02 -1.29203546e+00
1.29039913e-01 9.95825350e-01 -1.80890650e-01 -1.11865330e+00
-3.87517780e-01 -1.93021432e-01 -6.24997377e-01 7.20690668e-01
7.44012952e-01 3.00144494e-01 -5.83754718e-01 1.53875554e+00
7.79358864e-01 3.83114457e-01 -2.58065462e-01 8.29304457e-01
4.94219720e-01 3.22657257e-01 -3.18486422e-01 -2.10347936e-01
1.30220413e+00 -2.44586214e-01 -2.78379828e-01 2.66940624e-01
6.28506958e-01 -7.64621854e-01 4.33786452e-01 2.97231704e-01
-1.19836891e+00 -1.33389682e-01 -1.14488447e+00 -7.01473206e-02
-5.39358139e-01 -1.05724722e-01 5.86784720e-01 5.81025600e-01
-1.28885925e+00 8.81184280e-01 -6.70949042e-01 -6.17053926e-01
5.08739874e-02 8.91535580e-01 -5.28182268e-01 2.90336788e-01
-5.27482927e-01 9.51181054e-01 4.15623933e-01 3.67623346e-04
-3.68127078e-01 -7.33152628e-01 -8.27052832e-01 -1.47014186e-01
1.60139740e-01 -1.11669242e+00 8.88973475e-01 -2.59155959e-01
-1.12830937e+00 1.49396992e+00 -4.30647917e-02 -3.66596907e-01
3.98748130e-01 1.85200557e-01 2.07455307e-01 2.80913770e-01
6.87305704e-02 9.10790920e-01 4.49810863e-01 -1.39589536e+00
-6.47945881e-01 -6.29389703e-01 6.03325292e-02 9.33761597e-02
1.50225133e-01 2.40595620e-02 -5.44038713e-01 -4.75183606e-01
5.85183561e-01 -1.07314861e+00 -5.27794421e-01 5.95819235e-01
-3.67781907e-01 -1.45889670e-01 8.45207512e-01 -1.03853188e-01
8.78194988e-01 -1.61232781e+00 5.92210889e-01 6.82130933e-01
3.45667928e-01 -7.95537308e-02 5.15914261e-02 4.95438695e-01
-1.79261610e-01 2.60109715e-02 -3.83452892e-01 1.09391129e-02
-1.67850003e-01 1.81348398e-01 1.48275286e-01 8.43458533e-01
-2.53458858e-01 4.82546806e-01 -9.68282282e-01 -7.69050956e-01
3.26599985e-01 3.63109589e-01 -3.00058782e-01 -4.13759679e-01
-5.09074591e-02 1.26623645e-01 -3.75265658e-01 6.69149756e-01
5.58328688e-01 -1.08853178e-02 1.22033298e-01 -1.31681859e-01
-1.36272848e-01 -5.52969933e-01 -1.49031794e+00 1.89807820e+00
1.03031114e-01 5.01365244e-01 3.67603868e-01 -8.71951222e-01
1.12774003e+00 2.03015968e-01 1.01121831e+00 5.00675514e-02
2.66667575e-01 3.36008638e-01 -8.91365185e-02 4.38599326e-02
4.21962559e-01 -3.71032916e-02 -4.69700247e-02 1.41093671e-01
-9.45432782e-02 -5.82338154e-01 4.29383397e-01 1.16496027e-01
9.16421413e-01 1.20810717e-01 6.71261191e-01 -7.41192460e-01
5.14037192e-01 3.63515645e-01 7.02393055e-01 2.52004951e-01
1.50569845e-02 6.15745127e-01 3.32360715e-01 -5.14606237e-01
-1.23103154e+00 -1.14204991e+00 -4.89788592e-01 4.53101426e-01
6.44543767e-01 -2.42365927e-01 -9.11937654e-01 -2.24450633e-01
3.27296287e-01 1.90903440e-01 -6.33241236e-01 2.61192136e-02
-5.39680660e-01 -4.34804976e-01 4.93911177e-01 4.80015874e-02
6.87560141e-02 -3.87150645e-01 -8.81123245e-01 1.17538922e-01
1.57949135e-01 -7.44290054e-01 -4.08073127e-01 -2.96459310e-02
-1.18912315e+00 -1.48608649e+00 -8.93464029e-01 -9.56732929e-01
1.08657384e+00 2.64783174e-01 1.05999529e+00 6.28946900e-01
-6.77531540e-01 4.66089904e-01 -1.12490065e-01 -4.76430923e-01
-4.05362815e-01 -1.67090848e-01 1.28418431e-01 -4.81291085e-01
1.51006833e-01 -4.80792433e-01 -4.85269874e-01 6.56715631e-01
-9.35637772e-01 -1.20820746e-01 1.42660752e-01 5.54990411e-01
9.52013791e-01 -9.94501337e-02 3.27799708e-01 -7.45555639e-01
4.12262648e-01 -2.83910930e-01 -1.18971741e+00 3.67229521e-01
-3.59367698e-01 -6.66492283e-02 4.29082930e-01 -7.62568116e-02
-2.27917969e-01 6.13085985e-01 3.10654759e-01 -3.38361621e-01
-4.84427996e-02 1.32447913e-01 -1.09723113e-01 -7.75984347e-01
5.31819761e-01 -4.07099426e-02 1.42799646e-01 -2.64979631e-01
1.37612671e-01 2.35769436e-01 8.19693625e-01 -6.27823532e-01
9.58678544e-01 7.13641405e-01 8.80966783e-01 -8.21882010e-01
3.08496561e-02 -9.27049279e-01 -1.00926542e+00 -4.35994834e-01
5.91878831e-01 -3.77992213e-01 -1.05318320e+00 5.50975390e-02
-1.21683228e+00 1.61125332e-01 -1.04610540e-01 3.93813461e-01
-1.01612568e+00 5.20703316e-01 -4.24163759e-01 -5.06024718e-01
-2.71337926e-01 -1.00545073e+00 1.26544523e+00 2.91673064e-01
-2.37998351e-01 -1.13222861e+00 5.32483995e-01 3.21944803e-02
-2.62133121e-01 8.14649165e-01 9.93687868e-01 -3.38972777e-01
-7.07394838e-01 -3.31860393e-01 1.73161253e-01 -6.31547689e-01
-7.10441843e-02 6.35464728e-01 -5.26350617e-01 -2.55838335e-01
-3.29218000e-01 3.35578859e-01 5.26092708e-01 5.21753550e-01
7.11538792e-01 8.30370411e-02 -9.62063372e-01 6.79621100e-01
1.77074289e+00 3.44781160e-01 4.66678083e-01 5.82386374e-01
3.33473384e-01 8.62050891e-01 6.44908488e-01 2.45868772e-01
1.29535526e-01 8.98746371e-01 5.59436262e-01 -1.88781932e-01
2.25545526e-01 5.94016686e-02 -3.44230413e-01 4.94962364e-01
-2.85329878e-01 3.47385816e-02 -1.11983967e+00 5.45249403e-01
-1.93659532e+00 -8.73955667e-01 -4.56407189e-01 2.50653434e+00
6.22705579e-01 -2.74926066e-01 4.14792329e-01 4.17016715e-01
1.14053619e+00 -7.14250326e-01 -2.28181094e-01 -4.36731070e-01
5.78983277e-02 3.93060781e-02 7.07227588e-01 5.08549035e-01
-9.06670272e-01 4.76400346e-01 6.99107647e+00 3.58184099e-01
-6.05909884e-01 -5.16971231e-01 1.03947736e-01 2.50042349e-01
-2.95518041e-01 2.13402182e-01 -8.94641638e-01 2.57562220e-01
2.22585604e-01 -6.25436544e-01 2.30858967e-01 6.61570847e-01
1.09444084e-02 -1.58678025e-01 -1.48002088e+00 9.30758059e-01
-1.50031030e-01 -1.50542438e+00 -3.60832401e-02 3.21543276e-01
6.04950249e-01 -5.57953179e-01 -8.97837728e-02 -4.77275521e-01
1.57288939e-01 -8.18110287e-01 2.58242846e-01 5.73176026e-01
6.99264884e-01 -7.68923223e-01 4.16947097e-01 4.86235887e-01
-1.16174102e+00 3.14730257e-01 -6.29412234e-01 2.00595632e-01
1.30930886e-01 3.65269095e-01 -9.68895435e-01 4.65415418e-01
5.90821445e-01 3.01594168e-01 -2.80465990e-01 1.89096248e+00
3.70580733e-01 -1.58787444e-01 -7.80797958e-01 -4.06562276e-02
2.21016370e-02 -7.96502888e-01 8.74793351e-01 9.33474958e-01
5.36739707e-01 -2.06709746e-02 -1.32742822e-01 8.64849746e-01
1.47635490e-01 3.74623179e-01 -1.06426144e+00 3.89537990e-01
6.12010896e-01 1.24189115e+00 -1.45356858e+00 1.81347057e-02
-1.46912903e-01 6.83636010e-01 1.77990973e-01 6.21540584e-02
-5.61115026e-01 -3.92233461e-01 6.72454119e-01 1.77409545e-01
-1.13501213e-01 -3.46436352e-01 -3.89695942e-01 -6.93737268e-01
-3.98770243e-01 -5.03434122e-01 4.24605757e-01 -4.59799469e-01
-8.92665923e-01 7.34195858e-02 3.72729182e-01 -1.46944141e+00
-2.96311587e-01 -6.16988599e-01 -7.24106431e-01 4.19900388e-01
-8.98026466e-01 -9.62284207e-01 -2.56054789e-01 3.77788126e-01
1.63995713e-01 1.04457565e-01 7.55851150e-01 8.20926949e-02
-3.56580615e-01 2.71151632e-01 1.10907987e-01 -9.75551456e-02
4.11386698e-01 -1.43550146e+00 2.34876513e-01 8.18922400e-01
2.38293350e-01 9.84921515e-01 1.03766358e+00 -7.43374825e-01
-1.38079536e+00 -7.41054416e-01 5.84078968e-01 -3.47571313e-01
2.12359607e-01 -2.28409603e-01 -7.19696462e-01 5.36376953e-01
-7.52218217e-02 -1.40641302e-01 6.22683108e-01 -3.13064873e-01
8.02508146e-02 -5.17027825e-02 -1.48142827e+00 8.69019330e-01
9.60877061e-01 -2.88340896e-02 -5.91391861e-01 3.96873802e-01
2.86976427e-01 -5.79681754e-01 -1.05490494e+00 3.69279534e-01
5.09790659e-01 -7.83112645e-01 1.37048662e+00 -2.38107413e-01
5.30603379e-02 -6.45188749e-01 3.29268649e-02 -1.13595319e+00
-2.11905763e-01 -8.59475315e-01 2.44903713e-01 1.05954516e+00
1.59280628e-01 -3.28184783e-01 1.25171638e+00 5.49165130e-01
-1.30447656e-01 -5.67288220e-01 -1.17285395e+00 -9.52106357e-01
5.32088093e-02 1.90989316e-01 4.45652246e-01 1.00754452e+00
2.94267207e-01 2.94543188e-02 1.15142584e-01 2.44896561e-01
1.22660267e+00 2.88075179e-01 8.08635056e-01 -1.72371590e+00
9.15741473e-02 -7.81427443e-01 -1.02169764e+00 -5.26911497e-01
1.31178498e-01 -9.64830518e-01 -1.30024001e-01 -1.56679177e+00
-3.68877151e-03 -6.93587601e-01 2.47827813e-01 1.80458724e-01
1.77340984e-01 2.51166821e-01 1.91808164e-01 1.11008041e-01
-2.07837641e-01 -8.57133046e-02 1.15365481e+00 1.78667814e-01
-9.14946944e-02 2.62218803e-01 -2.08859101e-01 9.24041152e-01
8.18766177e-01 -6.09941304e-01 -3.54260415e-01 -2.37739474e-01
2.05608130e-01 2.69553065e-01 3.84722948e-01 -9.20584202e-01
6.21918142e-01 -2.43255705e-01 1.43814176e-01 -6.02689505e-01
4.59299684e-01 -1.43252635e+00 7.87778854e-01 7.02909052e-01
-1.23830721e-01 3.20267797e-01 -2.89704744e-02 5.96689701e-01
-5.25246374e-02 -7.91925848e-01 9.04385448e-01 -3.27776730e-01
-7.34799087e-01 3.91406149e-01 -2.01471061e-01 -2.92519689e-01
1.77741754e+00 -1.05457854e+00 -2.92186171e-01 4.32212017e-02
-5.85052669e-01 2.50736147e-01 1.22735000e+00 -4.47640195e-02
8.67097795e-01 -1.31849241e+00 -4.62245017e-01 -6.52387515e-02
8.38665739e-02 9.04886872e-02 -2.27348864e-01 6.91892803e-01
-1.13906121e+00 1.58560783e-01 -5.21088004e-01 -8.11789036e-01
-1.78386426e+00 5.06563723e-01 4.15030569e-01 3.40993166e-01
-3.85314584e-01 8.44215095e-01 -6.86472878e-02 -3.25682312e-01
-2.63109319e-02 -3.32142442e-01 -2.45741859e-01 -5.60575873e-02
5.77234067e-02 7.12057889e-01 -5.42756319e-02 -7.77420700e-01
-4.56004173e-01 1.24768293e+00 3.89630407e-01 3.74314934e-02
1.11497188e+00 1.60875116e-02 -4.05993879e-01 2.89325148e-01
1.06174135e+00 -2.30508111e-02 -5.90976059e-01 1.91941351e-01
2.39144742e-01 -6.16118670e-01 -1.60061508e-01 4.16662395e-02
-7.37842143e-01 5.81398666e-01 4.07043636e-01 3.29921722e-01
8.45581234e-01 -1.12353869e-01 5.79697073e-01 4.00744379e-01
9.17712271e-01 -1.03299224e+00 -3.20794642e-01 4.66183648e-02
7.79232502e-01 -8.86657238e-01 3.10086071e-01 -9.82286870e-01
8.63184631e-02 1.43734717e+00 5.50301850e-01 -4.41983759e-01
5.28329611e-01 6.60863042e-01 -1.44851997e-01 -2.27757365e-01
-5.36821127e-01 -2.05614477e-01 2.85457939e-01 7.77643800e-01
3.08534116e-01 -3.31768602e-01 -7.80821204e-01 -7.06639737e-02
-1.15574718e-01 -7.21623600e-02 5.70873380e-01 1.08642817e+00
-7.50432432e-01 -1.24864578e+00 -7.98814774e-01 3.89540613e-01
-4.02152479e-01 3.52412105e-01 -5.55580735e-01 9.13648784e-01
6.69144979e-03 5.36322236e-01 2.13445321e-01 -9.92747322e-02
3.63515526e-01 -3.70079100e-01 7.35475302e-01 -5.10719121e-01
-5.52971780e-01 9.92656425e-02 -3.16739976e-01 -4.55984294e-01
-5.12573063e-01 -6.26626968e-01 -1.66180444e+00 -2.56551892e-01
-5.90383530e-01 3.10179979e-01 8.90101373e-01 5.24104774e-01
1.42595142e-01 -3.63820307e-02 3.75515997e-01 -8.42897177e-01
-1.11347780e-01 -1.67495251e-01 -7.30326951e-01 3.82316411e-01
1.03893645e-01 -9.20838952e-01 -2.91011959e-01 2.64687896e-01] | [7.955733776092529, -2.2897374629974365] |
25aa3840-6501-4f81-9e19-95e0a4db6ee0 | beyond-top-grasps-through-scene-completion | 1909.12908 | null | https://arxiv.org/abs/1909.12908v2 | https://arxiv.org/pdf/1909.12908v2.pdf | Beyond Top-Grasps Through Scene Completion | Current end-to-end grasp planning methods propose grasps in the order of seconds that attain high grasp success rates on a diverse set of objects, but often by constraining the workspace to top-grasps. In this work, we present a method that allows end-to-end top-grasp planning methods to generate full six-degree-of-freedom grasps using a single RGB-D view as input. This is achieved by estimating the complete shape of the object to be grasped, then simulating different viewpoints of the object, passing the simulated viewpoints to an end-to-end grasp generation method, and finally executing the overall best grasp. The method was experimentally validated on a Franka Emika Panda by comparing 429 grasps generated by the state-of-the-art Fully Convolutional Grasp Quality CNN, both on simulated and real camera images. The results show statistically significant improvements in terms of grasp success rate when using simulated images over real camera images, especially when the real camera viewpoint is angled. Code and video are available at https://irobotics.aalto.fi/beyond-top-grasps-through-scene-completion/. | ['Ville Kyrki', 'Jens Lundell', 'Francesco Verdoja'] | 2019-09-15 | null | null | null | null | ['grasp-generation'] | ['computer-vision'] | [-2.29109764e-01 -7.59167224e-02 3.00646335e-01 -2.68784642e-01
-4.19567257e-01 -1.03563690e+00 8.44617411e-02 -1.03722900e-01
-1.44688085e-01 1.47575215e-01 -1.73727080e-01 -1.44136008e-02
-3.35673124e-01 -7.63715029e-01 -1.05295420e+00 -5.29989004e-01
-4.79267240e-01 1.04110920e+00 -5.31904697e-02 -3.98550212e-01
3.52400899e-01 8.83248627e-01 -1.41421378e+00 3.54754746e-01
5.60856521e-01 8.27353537e-01 1.10236084e+00 9.51445520e-01
1.60294369e-01 -1.17878737e-02 -3.12155992e-01 -2.06231281e-01
7.59928644e-01 3.15807730e-01 -6.65339947e-01 2.21998185e-01
2.70458132e-01 -1.05771732e+00 -3.08461875e-01 7.57013619e-01
4.86680984e-01 -4.91327792e-02 2.94825435e-01 -1.15256143e+00
-5.47892988e-01 5.62034190e-01 -3.58155966e-01 -7.08211660e-01
9.94549394e-01 6.46880269e-01 5.92433810e-01 -8.90386701e-01
8.18004966e-01 1.51738691e+00 2.92955697e-01 5.55576921e-01
-1.00996089e+00 -4.58709896e-01 -1.28024593e-01 -3.92919183e-01
-8.01678240e-01 3.96535583e-02 6.77097678e-01 -4.93356347e-01
1.05039549e+00 5.64870648e-02 8.23363841e-01 1.13059413e+00
4.72636014e-01 9.99328136e-01 8.70964706e-01 -3.13462377e-01
3.86983961e-01 -7.88443267e-01 -2.47177750e-01 6.53321326e-01
2.29747996e-01 3.83436739e-01 -2.58441448e-01 7.88512230e-02
1.36000633e+00 2.76425958e-01 -1.85752466e-01 -8.97827268e-01
-1.52596354e+00 3.93317372e-01 8.49280417e-01 -4.46855370e-03
-9.67673779e-01 4.02609289e-01 1.81233004e-01 9.22733694e-02
1.00305475e-01 6.20466530e-01 -6.61995769e-01 -9.86027569e-02
-6.14243031e-01 9.66196418e-01 8.82079780e-01 1.45220768e+00
4.14945245e-01 -3.42515886e-01 6.45661429e-02 1.60061076e-01
1.96412623e-01 7.61628091e-01 -3.06050420e-01 -1.33969891e+00
7.47942090e-01 5.60807407e-01 7.04545259e-01 -6.88318551e-01
-4.97067958e-01 2.37896770e-01 -3.48061442e-01 8.80104184e-01
6.14297390e-01 -1.20416082e-01 -1.52135766e+00 1.15078616e+00
3.79346669e-01 -7.59644210e-01 7.97380041e-03 1.41694868e+00
5.17922759e-01 3.92215014e-01 -1.61089584e-01 4.02510673e-01
1.28140497e+00 -1.03649557e+00 -5.25572777e-01 -2.44026095e-01
5.56810312e-02 -7.76721597e-01 1.12692416e+00 7.85725176e-01
-1.22039843e+00 -5.19591510e-01 -7.25766897e-01 3.00084371e-02
-1.97455034e-01 4.73166615e-01 8.93262565e-01 1.98667292e-02
-7.73347557e-01 9.11239982e-01 -1.35849118e+00 -2.63859510e-01
4.03593272e-01 5.32292366e-01 -6.25792623e-01 -5.34652293e-01
-3.94269228e-01 8.81861627e-01 5.88327646e-01 4.14047122e-01
-1.59569860e+00 -5.80402493e-01 -5.69924653e-01 8.14069062e-02
6.22074127e-01 -5.99204659e-01 1.33528996e+00 -2.64103681e-01
-1.60288095e+00 5.43649256e-01 3.23062450e-01 4.83283103e-02
9.55290496e-01 -8.27781260e-01 7.72166133e-01 2.56939560e-01
-1.20517882e-02 8.96917820e-01 6.95466697e-01 -1.79281020e+00
-1.57218993e-01 -6.32511973e-01 5.84107637e-01 2.91774422e-01
4.50759768e-01 -2.83482730e-01 -4.36679780e-01 -1.82009891e-01
4.85624045e-01 -1.19097257e+00 -3.94478768e-01 4.40865368e-01
-5.68239808e-01 -1.41296968e-01 8.68615746e-01 -6.69299603e-01
-7.90781230e-02 -1.69519246e+00 8.44152629e-01 -1.28169268e-01
-1.27660438e-01 2.87497818e-01 -2.56623894e-01 9.99100685e-01
1.54032201e-01 -3.58654320e-01 -5.34048602e-02 -4.01905954e-01
8.74471888e-02 1.47353321e-01 -3.09959382e-01 3.54136169e-01
-1.37375161e-01 9.33289945e-01 -1.17378211e+00 -3.58128175e-02
7.94985235e-01 3.95263672e-01 -3.06983829e-01 9.21414435e-01
-7.26668715e-01 7.25897551e-01 -5.63065231e-01 1.09263909e+00
9.84450340e-01 2.54494816e-01 8.42011720e-02 -3.43373418e-01
-2.45465606e-01 -2.63289034e-01 -1.12852311e+00 2.51513290e+00
-4.97484565e-01 -9.31863673e-03 6.19401813e-01 -3.35010767e-01
1.02372241e+00 2.71640450e-01 5.34632385e-01 -2.36911234e-03
1.70069605e-01 4.43600535e-01 -1.32334575e-01 -6.75055802e-01
3.70420128e-01 2.84913540e-01 1.48652509e-01 1.37951866e-01
1.06359087e-01 -9.09721851e-01 3.20309162e-01 -8.51789210e-03
7.53600299e-01 9.02686417e-01 -2.07565516e-01 -2.56601632e-01
-3.41950417e-01 2.12392360e-01 -2.71792620e-01 4.53596711e-01
2.50166476e-01 8.95201921e-01 1.94666341e-01 -7.00390160e-01
-1.35276139e+00 -1.33199346e+00 2.62137204e-01 6.18357956e-01
3.96524131e-01 2.67562307e-02 -9.24587548e-01 -4.05217916e-01
3.72803777e-01 6.35310709e-01 -5.65618873e-01 3.61256361e-01
-5.57414591e-01 1.23507231e-01 6.82580471e-02 7.37829268e-01
4.34554398e-01 -1.57681525e+00 -1.61906004e+00 2.90379941e-01
8.05231184e-02 -1.00516820e+00 -3.22230570e-02 3.41220200e-01
-9.45998192e-01 -1.32906151e+00 -1.23420858e+00 -4.98064667e-01
8.14265549e-01 2.64399320e-01 8.94134820e-01 -1.45244440e-02
-5.58630347e-01 4.80062157e-01 -6.37253463e-01 -4.22708511e-01
-2.68976808e-01 -3.46688479e-02 -2.67302170e-02 -9.19634640e-01
-2.13185295e-01 -3.69049877e-01 -7.56052613e-01 1.93640456e-01
-7.05248058e-01 3.61310631e-01 7.74275839e-01 4.50595587e-01
2.32427403e-01 -4.25399482e-01 -3.25781032e-02 2.05908030e-01
4.72040087e-01 -4.65042051e-03 -8.55366826e-01 2.83373624e-01
1.00654647e-01 -1.54709652e-01 3.52507889e-01 -3.91204000e-01
-8.60742211e-01 4.60619658e-01 -2.39090640e-02 -1.08723760e+00
-4.98335212e-01 3.04375172e-01 -2.80441102e-02 8.98496881e-02
3.33654881e-01 -6.94009736e-02 9.12354887e-02 -4.64295119e-01
4.94429231e-01 3.36092293e-01 4.52864975e-01 -9.41305161e-01
4.84463096e-01 4.35122401e-01 1.12788752e-01 -5.36615014e-01
-3.59181255e-01 -2.64406174e-01 -1.16074300e+00 -5.59993863e-01
9.11301613e-01 -4.30589736e-01 -1.27114296e+00 8.79548907e-01
-1.64099133e+00 -7.65297174e-01 -1.29105270e-01 4.93338615e-01
-1.05231607e+00 8.26574787e-02 -6.12213790e-01 -1.02970111e+00
-5.37730634e-01 -1.53689981e+00 1.65689838e+00 1.77276418e-01
4.66409884e-02 -2.89871246e-01 -3.17958921e-01 9.51377898e-02
2.74771988e-01 7.05533504e-01 4.32409436e-01 -4.96485382e-02
-8.36916208e-01 -3.35887700e-01 -1.25353992e-01 8.84937868e-02
3.08734596e-01 7.69363195e-02 -5.96949637e-01 -6.97958410e-01
-4.17993695e-01 -6.28945887e-01 4.88343656e-01 6.12167418e-01
9.13541198e-01 -7.59134442e-02 -4.86647189e-01 3.85709256e-01
1.60078061e+00 1.98975861e-01 5.41368365e-01 1.11514486e-01
7.83216238e-01 6.44787729e-01 1.08433974e+00 4.65757012e-01
-4.91591245e-02 8.69660556e-01 1.35797572e+00 2.63652384e-01
-1.23115204e-01 -2.38665283e-01 1.26841277e-01 1.74761131e-01
-6.43181622e-01 -4.17188019e-01 -1.11274087e+00 7.11631179e-01
-1.59058607e+00 -4.65699792e-01 -8.00358132e-02 2.20372629e+00
4.48592722e-01 -5.22092693e-02 3.33996974e-02 -2.51476943e-01
4.61750835e-01 -1.37856558e-01 -6.39776587e-01 -2.67305344e-01
5.89130223e-01 3.54350843e-02 6.64490521e-01 5.97893953e-01
-7.13687897e-01 1.24331093e+00 5.25664043e+00 2.51294822e-01
-9.20083046e-01 -4.03843105e-01 -2.47629449e-01 -3.49827111e-02
1.40350506e-01 2.74467878e-02 -3.68537307e-01 1.49184495e-01
9.19387788e-02 5.47585726e-01 8.42867792e-01 1.16763997e+00
4.46204096e-02 -3.95741671e-01 -1.41206634e+00 6.63099885e-01
-3.21651787e-01 -1.09319735e+00 -9.26842839e-02 -5.70991710e-02
4.27074820e-01 8.76364037e-02 -1.75688684e-01 4.39686701e-03
3.89425695e-01 -9.29823399e-01 1.10816312e+00 5.36517382e-01
7.66889513e-01 -5.09466767e-01 5.07572830e-01 5.95252931e-01
-1.18179524e+00 -1.40974075e-01 -4.43977326e-01 -1.11847065e-01
3.92392874e-01 2.85329551e-01 -9.04779077e-01 8.63639534e-01
1.03774846e+00 2.29520231e-01 2.39055920e-02 9.38839138e-01
-3.55205745e-01 -2.12687373e-01 -4.13295507e-01 -6.24230653e-02
4.24064577e-01 -1.77484125e-01 6.66094124e-01 9.63711321e-01
4.06300277e-01 4.28839564e-01 4.53678310e-01 1.22200334e+00
3.25052857e-01 -5.39134324e-01 -7.49468029e-01 -7.90354609e-02
3.13961357e-01 1.04834414e+00 -1.00779593e+00 -4.56762407e-03
4.42886591e-01 1.17170882e+00 2.57105172e-01 2.53751755e-01
-5.70302725e-01 -3.94822031e-01 4.24718499e-01 -2.79863793e-02
3.46459985e-01 -8.24917257e-01 -6.99384958e-02 -8.59697282e-01
3.93321514e-01 -6.23079777e-01 -5.11150062e-01 -1.37001133e+00
-9.11525786e-01 3.93620998e-01 4.90153491e-01 -9.91797507e-01
-1.16556220e-01 -1.07955313e+00 -3.38363647e-01 8.97978485e-01
-9.24349189e-01 -1.70295537e+00 -7.42622554e-01 1.73552901e-01
1.03931952e+00 2.98896700e-01 9.90458369e-01 -3.41976523e-01
3.48169476e-01 -6.86429739e-02 -2.44419023e-01 1.30591970e-02
4.43039328e-01 -1.26565313e+00 5.58760822e-01 4.55291301e-01
-4.16063994e-01 5.32766283e-01 9.25049841e-01 -9.85925615e-01
-2.03169346e+00 -4.30652678e-01 8.74061808e-02 -4.20287400e-01
1.53573573e-01 -5.26387274e-01 -6.26934409e-01 9.06569600e-01
2.55845010e-01 -8.84816647e-02 -5.97631276e-01 -2.03412145e-01
-5.42266713e-03 2.60648280e-01 -1.52274406e+00 5.13138235e-01
1.08141279e+00 1.68681264e-01 -5.98658979e-01 5.31294465e-01
6.87644899e-01 -1.38636601e+00 -8.30642462e-01 4.81995851e-01
1.15127981e+00 -9.50746357e-01 1.14022136e+00 -5.97892404e-01
9.70511794e-01 -1.60425663e-01 -3.54973286e-01 -1.58850443e+00
-5.35757467e-02 -5.74238062e-01 -1.26632944e-01 5.78061283e-01
-2.03292057e-01 -9.24271792e-02 9.03169096e-01 4.50143009e-01
-4.28037524e-01 -9.45819557e-01 -7.95870423e-01 -9.15634334e-01
1.37523003e-02 -2.06216216e-01 6.09122276e-01 3.81325454e-01
5.26151769e-02 -4.56261039e-01 -1.34566858e-01 5.25158584e-01
9.29741979e-01 4.82207000e-01 1.06810987e+00 -9.52655852e-01
-7.65268207e-02 -1.28092125e-01 -1.59845203e-01 -1.06474864e+00
-4.60713506e-02 -5.77259481e-01 6.21772468e-01 -2.15119195e+00
-4.36346419e-02 -6.23322845e-01 5.30454159e-01 6.84011996e-01
1.41175374e-01 -3.99235457e-01 7.01067448e-01 1.71829134e-01
-2.74983421e-02 3.50121766e-01 1.92881322e+00 2.59847641e-02
-1.61848843e-01 5.69281131e-02 3.39830667e-01 5.63181162e-01
7.62465417e-01 -2.92711288e-01 -2.48361692e-01 -9.70613480e-01
-1.46364763e-01 6.68431699e-01 7.85600066e-01 -1.09529281e+00
-6.37117773e-02 -4.60401922e-01 4.68705177e-01 -8.61176014e-01
7.76983619e-01 -1.12110555e+00 2.45956823e-01 1.02132452e+00
-3.18301499e-01 2.75741607e-01 4.57073241e-01 5.12590706e-01
4.05234367e-01 -3.90988946e-01 3.85107994e-01 -7.56792963e-01
-3.68378192e-01 2.82028377e-01 2.51879822e-02 -5.49683869e-01
1.26446629e+00 -1.98445037e-01 -1.57263041e-01 -2.88742602e-01
-9.20757473e-01 2.07602724e-01 7.37780869e-01 5.91160893e-01
9.73403335e-01 -9.50749457e-01 -8.38678777e-01 1.67198256e-01
-1.23749852e-01 6.60132945e-01 2.00725868e-01 3.95303816e-01
-1.11918879e+00 4.31747198e-01 -6.68786526e-01 -9.29227591e-01
-1.13465011e+00 7.45401621e-01 1.30983427e-01 4.23368216e-02
-8.25688779e-01 8.44935656e-01 -1.23087406e-01 -8.91570985e-01
3.25780362e-01 -7.40346611e-01 5.61302483e-01 -6.24771178e-01
7.51074925e-02 4.98106509e-01 1.05321273e-01 -8.00607279e-02
-3.23706150e-01 8.27576995e-01 1.85530901e-01 -2.03585908e-01
1.71685648e+00 4.43518847e-01 -8.36363658e-02 -4.47675139e-02
7.59535074e-01 -2.95692384e-01 -1.93765748e+00 5.75199723e-01
-5.36514103e-01 -9.31069136e-01 -4.39132124e-01 -1.25587535e+00
-1.09988213e+00 1.02719247e+00 6.01782262e-01 -9.56987813e-02
6.98276341e-01 1.07947685e-01 5.60832798e-01 6.96482241e-01
1.21278536e+00 -5.29631913e-01 3.13767374e-01 4.98878241e-01
1.73960459e+00 -1.11980557e+00 7.95313343e-02 -5.95094562e-01
-5.82379222e-01 1.52753091e+00 7.27008998e-01 -7.79732466e-01
2.80083239e-01 5.39816916e-01 -1.10968009e-01 -6.39322698e-01
-2.09531724e-01 3.22234452e-01 -3.18662710e-02 8.00962687e-01
-7.12943599e-02 3.70291144e-01 1.14397012e-01 -6.46683276e-02
-1.64408922e-01 2.91544825e-01 5.28258443e-01 1.37016523e+00
-2.68527538e-01 -8.12595844e-01 -3.90120029e-01 9.60531458e-02
-7.91244656e-02 3.51953655e-01 -1.91412240e-01 9.15264308e-01
-1.45589858e-01 8.18390369e-01 -2.08536512e-03 -2.34251440e-01
6.10856950e-01 -2.50620216e-01 1.27551186e+00 -6.46878839e-01
-6.37732983e-01 -1.35783851e-01 -1.71905845e-01 -1.02492225e+00
-1.81474268e-01 -5.60001612e-01 -1.50397301e+00 -1.41727820e-01
-4.50786740e-01 -2.79561996e-01 1.17006016e+00 7.99894154e-01
2.64945835e-01 4.20630753e-01 2.72085011e-01 -2.31300998e+00
-8.68775249e-01 -1.02356708e+00 -3.42712641e-01 3.34225804e-01
2.72200376e-01 -8.59211266e-01 -4.73906212e-02 -1.41182631e-01] | [5.736574649810791, -0.8199937343597412] |
cd24d02a-606b-4b8a-9815-e85ab74e7a9e | shared-space-transfer-learning-for-analyzing | 2010.15594 | null | https://arxiv.org/abs/2010.15594v1 | https://arxiv.org/pdf/2010.15594v1.pdf | Shared Space Transfer Learning for analyzing multi-site fMRI data | Multi-voxel pattern analysis (MVPA) learns predictive models from task-based functional magnetic resonance imaging (fMRI) data, for distinguishing when subjects are performing different cognitive tasks -- e.g., watching movies or making decisions. MVPA works best with a well-designed feature set and an adequate sample size. However, most fMRI datasets are noisy, high-dimensional, expensive to collect, and with small sample sizes. Further, training a robust, generalized predictive model that can analyze homogeneous cognitive tasks provided by multi-site fMRI datasets has additional challenges. This paper proposes the Shared Space Transfer Learning (SSTL) as a novel transfer learning (TL) approach that can functionally align homogeneous multi-site fMRI datasets, and so improve the prediction performance in every site. SSTL first extracts a set of common features for all subjects in each site. It then uses TL to map these site-specific features to a site-independent shared space in order to improve the performance of the MVPA. SSTL uses a scalable optimization procedure that works effectively for high-dimensional fMRI datasets. The optimization procedure extracts the common features for each site by using a single-iteration algorithm and maps these site-specific common features to the site-independent shared space. We evaluate the effectiveness of the proposed method for transferring between various cognitive tasks. Our comprehensive experiments validate that SSTL achieves superior performance to other state-of-the-art analysis techniques. | ['Russell Greiner', 'Andrew J. Greenshaw', 'Daoqiang Zhang', 'Alessandro Selvitella', 'Muhammad Yousefnezhad'] | 2020-10-24 | null | http://proceedings.neurips.cc/paper/2020/hash/b837305e43f7e535a1506fc263eee3ed-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/b837305e43f7e535a1506fc263eee3ed-Paper.pdf | neurips-2020-12 | ['art-analysis'] | ['computer-vision'] | [ 3.45098317e-01 -6.45351708e-01 -8.00597444e-02 -6.56829715e-01
-9.71013188e-01 -1.67730078e-01 2.35326603e-01 -1.78165346e-01
-5.40606499e-01 7.53201723e-01 9.46943760e-02 8.22168868e-03
-6.02094412e-01 -5.24536371e-01 -6.18067980e-01 -7.32217431e-01
-3.49214226e-01 4.50155616e-01 3.85556340e-01 2.90906757e-01
3.31893384e-01 3.67271066e-01 -1.32631171e+00 7.19101489e-01
1.03540885e+00 9.13909078e-01 8.57303560e-01 1.19092740e-01
2.10875720e-02 4.38133687e-01 -8.37315023e-02 2.64035851e-01
2.78632313e-01 -5.83362341e-01 -7.91819334e-01 -3.55689347e-01
4.09668684e-01 -9.05246139e-02 4.93845381e-02 1.04927361e+00
6.41046107e-01 4.24895376e-01 7.74688959e-01 -1.10339904e+00
-3.85234147e-01 3.64792228e-01 -7.82656908e-01 8.74271154e-01
4.64384519e-02 6.84823096e-02 6.61750615e-01 -1.02997565e+00
4.93974566e-01 1.04719567e+00 6.01275861e-01 2.63839483e-01
-1.38266098e+00 -1.13914192e+00 3.54223996e-02 5.58541059e-01
-1.37486207e+00 -2.47799337e-01 7.82076240e-01 -8.16996157e-01
7.84161985e-01 2.26898730e-01 5.38739324e-01 1.08764744e+00
8.15356970e-01 5.72274923e-01 1.75519156e+00 -3.89199285e-03
4.00184989e-01 -1.62615910e-01 3.32128167e-01 5.22569120e-01
-2.26290990e-03 -1.15222298e-01 -7.33423114e-01 -3.97482276e-01
9.47615862e-01 1.36896402e-01 -1.73931956e-01 -4.15036321e-01
-1.64049971e+00 7.53538311e-01 5.71326137e-01 7.15743303e-01
-7.80803382e-01 -3.52687657e-01 5.28639019e-01 1.90029442e-01
4.01444733e-01 2.23154843e-01 -5.90378761e-01 2.64201850e-01
-1.20584106e+00 2.56925106e-01 1.40351221e-01 4.36087608e-01
8.29398155e-01 -1.22424431e-01 -4.56441253e-01 9.83877420e-01
1.30734101e-01 3.24252754e-01 1.20533717e+00 -6.16788387e-01
4.61082101e-01 3.80917370e-01 -3.75794739e-01 -1.24020982e+00
-7.84207404e-01 -4.71064568e-01 -9.39452708e-01 7.87697546e-03
1.49250120e-01 -2.01318741e-01 -5.09578288e-01 1.87019181e+00
2.72257805e-01 2.68660188e-01 -4.21615779e-01 1.24428427e+00
4.60336566e-01 3.69928956e-01 3.75014186e-01 -3.84808213e-01
1.45458329e+00 -7.56626785e-01 -4.97433871e-01 -5.56665838e-01
6.87739432e-01 -2.31928110e-01 1.15450609e+00 2.16352463e-01
-7.48332679e-01 -8.06780875e-01 -8.68919611e-01 1.54082671e-01
-8.55478644e-02 2.65413880e-01 7.20783830e-01 4.46689188e-01
-7.79796600e-01 6.55258298e-01 -9.68696833e-01 -1.56551331e-01
6.71951354e-01 4.65435237e-01 -7.71040022e-01 -7.43627697e-02
-9.80934381e-01 8.00995827e-01 5.36074698e-01 -2.11278815e-02
-8.37270617e-01 -1.05673540e+00 -7.41103709e-01 6.56808913e-02
2.82169759e-01 -6.19857073e-01 7.83353567e-01 -1.07860625e+00
-1.23329818e+00 6.97677433e-01 -4.16110098e-01 -1.31362274e-01
1.93769872e-01 -9.07880962e-02 -4.49935019e-01 2.68114418e-01
5.96191406e-01 4.42023396e-01 7.98851967e-01 -7.01870084e-01
-2.36610919e-01 -9.01905298e-01 -6.21868312e-01 2.53933132e-01
-4.38726753e-01 2.85286427e-01 9.70747173e-02 -6.57287121e-01
5.41501045e-01 -7.19063759e-01 -1.56254217e-01 -7.72373527e-02
-5.18732220e-02 -1.37152493e-01 4.88967746e-01 -9.06311154e-01
9.27614868e-01 -2.13031054e+00 3.88238788e-01 3.41352880e-01
3.73619139e-01 -6.21610470e-02 -7.34398067e-02 -1.30995318e-01
-4.66125607e-01 -1.14144169e-01 -4.44957376e-01 4.93397750e-02
-2.08565295e-01 -5.16108572e-02 2.93932229e-01 6.21159434e-01
-3.14977579e-02 8.86814415e-01 -7.67386854e-01 -6.89742744e-01
2.08800063e-02 6.35212660e-02 -4.99062061e-01 1.26060545e-01
2.89768547e-01 9.87439692e-01 -6.78779066e-01 3.92026544e-01
6.40749931e-01 -3.68036211e-01 3.25911522e-01 -4.77744043e-01
-6.85783923e-02 7.55891018e-03 -1.02295315e+00 1.89941573e+00
-2.47576445e-01 3.65727454e-01 -4.52524088e-02 -1.38388824e+00
9.31949914e-01 3.06592315e-01 7.94815361e-01 -1.00534427e+00
2.44462386e-01 3.64114165e-01 4.89130646e-01 -5.90396464e-01
-3.77726227e-01 -3.67035210e-01 6.43572807e-02 7.32638359e-01
3.22797179e-01 6.13287747e-01 -2.30754927e-01 -1.76579475e-01
1.25191212e+00 -2.02690195e-02 4.71114844e-01 -7.13802755e-01
6.34270489e-01 -3.54945511e-01 8.99888754e-01 4.99596924e-01
-4.72023547e-01 3.27114314e-01 1.83296055e-01 -4.79378968e-01
-8.56211424e-01 -9.81854737e-01 -3.34869236e-01 1.26466393e+00
-3.21094245e-01 -3.90740000e-02 -7.02186346e-01 -6.47498190e-01
-4.40547019e-02 3.45379144e-01 -7.39242494e-01 -3.08900416e-01
-5.81252396e-01 -9.57250834e-01 1.92896783e-01 7.12233245e-01
8.39623749e-01 -1.06991148e+00 -5.27879059e-01 4.84816507e-02
-2.61875570e-01 -9.56819415e-01 -4.80338842e-01 1.20204955e-01
-1.11681092e+00 -1.06208599e+00 -6.55462503e-01 -8.62443507e-01
5.68505645e-01 4.14550364e-01 6.01432085e-01 -3.92975241e-01
-2.68149287e-01 -5.64633086e-02 -1.23320356e-01 9.20538679e-02
2.97040105e-01 -5.71500212e-02 2.54265368e-01 5.43756723e-01
3.20212305e-01 -8.18680882e-01 -7.20977366e-01 4.87826228e-01
-6.36252701e-01 2.04243556e-01 9.10817444e-01 8.40035021e-01
8.95920098e-01 -8.54933411e-02 8.66694868e-01 -8.25061798e-01
6.35565519e-01 -7.97158241e-01 -1.43984646e-01 2.83900082e-01
-3.36061537e-01 3.97758745e-02 6.61521912e-01 -5.89716434e-01
-1.17295837e+00 8.83344561e-02 2.86006123e-01 -5.26954412e-01
-2.52703071e-01 7.78034747e-01 -2.26361379e-01 -2.85759509e-01
7.53895879e-01 4.34691757e-01 5.05773649e-02 -7.19437361e-01
-6.11170791e-02 5.71396470e-01 4.41953003e-01 -5.81494033e-01
3.71249259e-01 2.22207963e-01 5.47471605e-02 -5.52788496e-01
-5.78919291e-01 -3.70339721e-01 -1.20884001e+00 -1.90469533e-01
8.69557440e-01 -8.30809057e-01 -3.84231180e-01 2.93562740e-01
-6.23963773e-01 -3.31615359e-01 2.88850099e-01 9.90615189e-01
-5.88671803e-01 4.09642994e-01 -5.06388783e-01 -3.86980563e-01
-4.69873011e-01 -1.20231473e+00 7.74587512e-01 -6.57780990e-02
-2.54740715e-01 -8.72774363e-01 2.07827628e-01 5.00026941e-01
3.68511379e-01 9.75784138e-02 1.15738738e+00 -9.08676565e-01
-8.15571845e-02 -7.97606166e-03 -2.88330108e-01 2.74066497e-02
1.28214344e-01 -7.64245510e-01 -7.72085845e-01 -3.18080306e-01
4.04595256e-01 -2.36536637e-01 6.53962672e-01 6.21309280e-01
1.65923679e+00 3.64314169e-02 -4.92901921e-01 5.69156349e-01
1.16634905e+00 1.74578980e-01 3.86734486e-01 3.05231839e-01
5.62248051e-01 5.58422089e-01 5.63528895e-01 1.54259041e-01
2.71000892e-01 6.95967019e-01 -1.74590349e-01 1.78498134e-01
4.53116074e-02 1.84249312e-01 2.94531792e-01 9.61326361e-01
-8.75210911e-02 6.20255530e-01 -9.92464781e-01 3.64935994e-01
-1.99941719e+00 -1.03478909e+00 -1.88542321e-01 2.25035954e+00
7.49586880e-01 -2.52787143e-01 8.95609111e-02 -1.20979205e-01
8.09668481e-01 -7.84287825e-02 -7.28733480e-01 -1.59985423e-02
-5.52212782e-02 4.11637634e-01 1.27121642e-01 -2.77880300e-02
-1.15751946e+00 5.36522210e-01 6.35716772e+00 8.92054975e-01
-1.17814529e+00 8.06520998e-01 5.85575402e-01 -2.97610968e-01
1.21107921e-01 -2.45059088e-01 -2.45796993e-01 5.09282947e-01
1.07412958e+00 -3.82654071e-01 3.95383686e-01 7.57050574e-01
4.12553012e-01 -1.73801124e-01 -1.04997993e+00 1.23265731e+00
2.29283303e-01 -1.00192881e+00 -2.09613502e-01 -1.39131635e-01
5.35619199e-01 1.35268420e-01 -5.75591587e-02 2.96201557e-01
-4.21452016e-01 -9.32830155e-01 3.87451023e-01 7.21579790e-01
6.19633079e-01 -3.89583528e-01 5.59793055e-01 5.51154673e-01
-1.16508937e+00 -2.29413718e-01 -5.71381807e-01 -9.21664909e-02
-1.33518293e-01 5.81748664e-01 -2.81431824e-01 4.54757065e-01
8.88908625e-01 7.31841505e-01 -6.36129498e-01 1.14120710e+00
1.36274397e-01 4.33579028e-01 -4.27570194e-02 2.21187726e-01
7.85995349e-02 -2.33814582e-01 3.41558516e-01 1.02829981e+00
4.61282104e-01 2.30917618e-01 5.61025381e-01 1.06134117e+00
2.29276121e-01 6.04901075e-01 -4.18686926e-01 3.12237173e-01
2.51440793e-01 1.55409575e+00 -6.52693033e-01 -2.52164721e-01
-4.60227430e-01 9.34898078e-01 5.69330454e-01 1.33775890e-01
-7.45323658e-01 -7.02082589e-02 2.58290052e-01 1.74081624e-02
4.09726687e-02 -2.52764910e-01 -3.77065986e-01 -1.14282525e+00
2.36984015e-01 -7.89862394e-01 5.42335927e-01 -7.93977141e-01
-1.66862357e+00 6.59415245e-01 2.90694147e-01 -1.12657487e+00
-6.27284050e-02 -5.61330378e-01 -6.03751600e-01 1.26024783e+00
-1.32396042e+00 -1.02003717e+00 -3.30479860e-01 1.07103693e+00
5.14958143e-01 -3.85587811e-01 8.80051196e-01 4.50703740e-01
-7.96748757e-01 2.79385686e-01 4.99164909e-02 5.37518673e-02
9.33514059e-01 -7.70627797e-01 -3.86850625e-01 6.52855337e-01
-1.85144126e-01 7.17015803e-01 5.53764887e-02 -8.23160708e-01
-1.27747130e+00 -1.00780356e+00 5.33229947e-01 3.19675021e-02
6.15784347e-01 -4.03406411e-01 -1.27454734e+00 7.50476241e-01
-2.01752260e-01 4.01094913e-01 1.11991537e+00 2.20755517e-01
-2.32884035e-01 -1.73178315e-01 -1.20122421e+00 2.74544507e-01
1.08606529e+00 -6.70339286e-01 -8.15917969e-01 6.56025469e-01
1.17843255e-01 -4.99766283e-02 -1.25054526e+00 2.16057375e-01
6.46587193e-01 -7.59262264e-01 8.54004622e-01 -7.50288367e-01
3.25785041e-01 -1.01860493e-01 -1.47911116e-01 -1.43289018e+00
-1.11023533e+00 5.27242497e-02 1.06385201e-01 6.90504730e-01
1.91192567e-01 -7.95807421e-01 4.82830167e-01 7.98152208e-01
-2.03381062e-01 -5.43463707e-01 -1.10260403e+00 -8.80595744e-01
1.46741092e-01 -2.38974124e-01 5.11955380e-01 1.19902265e+00
2.08034381e-01 3.29196304e-01 -2.89622754e-01 7.89747462e-02
6.98424459e-01 3.43925744e-01 1.86100900e-01 -1.36074364e+00
-2.35811785e-01 -3.41711402e-01 -4.16131645e-01 -3.44467610e-01
3.71809602e-01 -1.39398062e+00 -1.17277279e-01 -1.25691140e+00
7.49605954e-01 -4.56375122e-01 -7.54977345e-01 7.17375875e-01
-3.10919344e-01 1.12935677e-01 8.30728188e-03 4.54166830e-01
-5.19007564e-01 5.44546962e-01 1.30069637e+00 3.26278806e-02
-2.30262280e-01 -3.11297894e-01 -5.73923826e-01 5.12679875e-01
1.02058113e+00 -4.48199123e-01 -4.73521024e-01 -3.58241290e-01
-2.70929128e-01 5.92203550e-02 5.36420703e-01 -1.37343550e+00
2.09176138e-01 -2.47613296e-01 1.03033948e+00 -2.42991507e-01
1.50239348e-01 -6.59256220e-01 2.55252957e-01 5.10581136e-01
-2.80826837e-01 4.25196700e-02 2.61042476e-01 3.80906701e-01
-6.42393753e-02 -5.56753539e-02 7.44635582e-01 -3.23408782e-01
-8.25061142e-01 4.83526856e-01 -2.47153282e-01 -1.21574536e-01
1.15461886e+00 -2.37061560e-01 -7.59185404e-02 2.21790940e-01
-9.67051804e-01 7.19643012e-02 -1.83452666e-01 4.02691513e-01
6.55653119e-01 -1.51391220e+00 -7.17932999e-01 4.97220248e-01
-1.15042791e-01 -5.96271753e-01 8.96102667e-01 1.41753757e+00
4.19264585e-02 5.14776647e-01 -1.08567953e+00 -7.53990412e-01
-1.12316132e+00 5.13156950e-01 3.54205787e-01 -1.85379371e-01
-7.83825338e-01 4.56091553e-01 5.12937307e-01 -5.14592409e-01
-4.88640100e-01 -4.40136380e-02 -4.35897976e-01 9.76859629e-02
7.97920942e-01 2.90239692e-01 3.04324001e-01 -6.87732935e-01
-5.65694928e-01 4.28438723e-01 -1.12999544e-01 -3.71797346e-02
1.58691776e+00 1.35262549e-01 -3.61842364e-01 6.54360592e-01
1.35021865e+00 -4.24282759e-01 -9.97000933e-01 -5.23606420e-01
-9.87803638e-02 -6.72113657e-01 4.45979238e-01 -7.20488548e-01
-1.23576307e+00 8.85888398e-01 8.59716356e-01 -5.10090828e-01
1.14854705e+00 -2.05338940e-01 5.08013666e-01 1.11438617e-01
6.59454823e-01 -1.03677773e+00 1.03464723e-01 1.62084371e-01
1.00494313e+00 -1.13903534e+00 2.12266878e-03 -2.50047892e-01
-7.76468515e-01 9.43834722e-01 8.66054416e-01 -2.74541378e-01
8.20718229e-01 -1.07231244e-01 -4.82444286e-01 -4.30107385e-01
-4.49784070e-01 2.08572716e-01 6.79049253e-01 5.33574522e-01
4.38546181e-01 3.17968428e-01 -5.64851344e-01 1.24657476e+00
-1.42671866e-02 1.36968344e-01 -2.74269223e-01 8.81616235e-01
-5.47527850e-01 -1.03629351e+00 -4.75060999e-01 1.14933836e+00
-2.21181452e-01 -8.05541966e-03 -5.40821590e-02 5.46991587e-01
1.67023718e-01 5.06597638e-01 9.61571783e-02 -3.58484805e-01
1.13806695e-01 5.87998629e-01 5.57292521e-01 -6.04567409e-01
-5.86927652e-01 1.76551849e-01 -3.70824724e-01 -8.41999888e-01
-4.13899124e-01 -1.14975750e+00 -1.19154763e+00 -2.66226213e-02
3.61165628e-02 -8.07955638e-02 4.50815618e-01 1.20631742e+00
6.72544360e-01 5.99710584e-01 5.29209375e-01 -8.33342254e-01
-3.95335793e-01 -1.20833826e+00 -8.19865882e-01 4.43599701e-01
-2.11127698e-01 -1.10615945e+00 1.29046336e-01 -1.82502881e-01] | [12.637799263000488, 3.330491542816162] |
797695d3-f96a-4993-99ef-4ab570c0203b | revisiting-machine-translation-for-cross | 2305.14240 | null | https://arxiv.org/abs/2305.14240v1 | https://arxiv.org/pdf/2305.14240v1.pdf | Revisiting Machine Translation for Cross-lingual Classification | Machine Translation (MT) has been widely used for cross-lingual classification, either by translating the test set into English and running inference with a monolingual model (translate-test), or translating the training set into the target languages and finetuning a multilingual model (translate-train). However, most research in the area focuses on the multilingual models rather than the MT component. We show that, by using a stronger MT system and mitigating the mismatch between training on original text and running inference on machine translated text, translate-test can do substantially better than previously assumed. The optimal approach, however, is highly task dependent, as we identify various sources of cross-lingual transfer gap that affect different tasks and approaches differently. Our work calls into question the dominance of multilingual models for cross-lingual classification, and prompts to pay more attention to MT-based baselines. | ['Luke Zettlemoyer', 'Angela Fan', 'Shruti Bhosale', 'Vedanuj Goswami', 'Mikel Artetxe'] | 2023-05-23 | null | null | null | null | ['cross-lingual-transfer'] | ['natural-language-processing'] | [ 1.25478789e-01 -2.73932293e-02 -6.05643332e-01 -5.26387751e-01
-1.58071804e+00 -1.15190530e+00 9.10803318e-01 -9.44166258e-02
-5.94634652e-01 1.07891679e+00 1.31677657e-01 -1.19427943e+00
4.07669634e-01 -3.84716034e-01 -9.26068902e-01 -3.86525601e-01
5.83110631e-01 9.06605899e-01 -3.35305021e-03 -2.55850792e-01
-2.74672383e-03 6.02196762e-03 -9.94281411e-01 5.34602702e-01
1.05083609e+00 3.00219655e-01 5.79915568e-02 6.84933364e-01
-2.19071552e-01 6.16081834e-01 -5.17441392e-01 -9.59018350e-01
2.69556701e-01 -8.01472187e-01 -1.11785305e+00 -3.11403185e-01
8.13259959e-01 1.22421980e-01 1.01625673e-01 1.16639197e+00
4.37118530e-01 -4.86733735e-01 5.62659860e-01 -9.69543517e-01
-8.42894912e-01 1.01970923e+00 -2.81299680e-01 2.27485806e-01
2.96176463e-01 5.50575629e-02 9.71115589e-01 -9.52411473e-01
7.91704655e-01 1.32478261e+00 8.31583083e-01 4.03628469e-01
-1.42090976e+00 -6.78841114e-01 -5.34592047e-02 1.60446286e-01
-1.12294579e+00 -8.02717209e-01 3.34360033e-01 -4.71754849e-01
1.05493760e+00 3.03435594e-01 1.13168076e-01 1.49701297e+00
5.16844869e-01 6.95396841e-01 1.98109007e+00 -9.09884155e-01
-3.04180175e-01 5.63596964e-01 -1.62558988e-01 4.22210395e-01
8.39997157e-02 2.99063861e-01 -2.46120408e-01 3.66794467e-02
2.32219160e-01 -8.52562368e-01 -1.93601489e-01 3.99042889e-02
-1.45373487e+00 6.92180216e-01 -5.00558764e-02 6.75222218e-01
-1.11941315e-01 -1.72100320e-01 7.20358312e-01 8.51602376e-01
8.21806550e-01 5.16852200e-01 -8.51424992e-01 -2.16154039e-01
-1.10122621e+00 -2.25348528e-02 8.17896903e-01 8.77717495e-01
8.41033101e-01 -1.14660680e-01 -6.51124269e-02 1.00136113e+00
-2.12197192e-02 7.17971742e-01 6.02087736e-01 -5.57487845e-01
9.57464159e-01 1.51916832e-01 -3.05225998e-01 -3.35627526e-01
7.99291506e-02 -3.53689015e-01 -2.84016520e-01 -9.90562420e-03
6.65869415e-01 -2.85691559e-01 -5.78575134e-01 1.86988628e+00
9.87487063e-02 -2.87890017e-01 2.56289065e-01 6.03472412e-01
6.38796017e-02 5.82311809e-01 2.76635736e-01 -1.78962722e-01
1.23211205e+00 -1.11983633e+00 -4.04464275e-01 -5.47668517e-01
1.16469443e+00 -1.37030339e+00 1.46327162e+00 1.31972775e-01
-1.19419825e+00 -5.91707587e-01 -9.96079087e-01 -2.91358590e-01
-6.15863502e-01 2.69537002e-01 3.62975895e-01 7.82604635e-01
-1.27990150e+00 4.67345208e-01 -7.82813191e-01 -7.06866205e-01
-2.17681304e-01 1.30125970e-01 -4.74018335e-01 -4.64137197e-01
-1.41428626e+00 1.70655751e+00 2.64605880e-01 -6.36789054e-02
-7.21462429e-01 -3.51468205e-01 -8.47249150e-01 -5.67156136e-01
-1.16460823e-01 -5.86636364e-01 1.39113355e+00 -1.39681149e+00
-1.50026035e+00 1.32717001e+00 -2.96797723e-01 -2.31578544e-01
8.86694014e-01 -1.14350446e-01 -5.46185195e-01 -4.84101743e-01
5.45113623e-01 4.92076367e-01 5.27461231e-01 -1.10057902e+00
-8.46712053e-01 -4.29406285e-01 -8.15595239e-02 3.63208681e-01
-1.83192685e-01 4.96228218e-01 -2.79970080e-01 -5.25664926e-01
-2.05959365e-01 -1.18521202e+00 1.24886014e-01 -7.18015850e-01
-3.44895303e-01 -2.46609524e-01 4.48498696e-01 -9.57215250e-01
1.12173498e+00 -1.79057181e+00 1.81518614e-01 -2.34787732e-01
-2.97619253e-01 2.80331165e-01 -3.32387954e-01 5.41838288e-01
-9.57453176e-02 3.09317559e-01 -2.19464630e-01 -4.89192307e-01
4.31018174e-02 3.51965487e-01 -4.18317020e-01 4.65933502e-01
1.81924388e-01 1.20059633e+00 -8.48708212e-01 -5.97434521e-01
-5.12208529e-02 2.40933537e-01 -3.53589356e-01 -5.14326096e-02
-9.72234681e-02 7.52089202e-01 -2.16642216e-01 6.35537505e-01
5.74884355e-01 1.75405964e-01 6.29596531e-01 2.94338278e-02
-5.12879252e-01 1.01682413e+00 -4.73954439e-01 1.62851596e+00
-8.40177298e-01 8.48112762e-01 8.54669977e-03 -9.99849916e-01
5.04284322e-01 4.69988704e-01 4.50103655e-02 -9.89089727e-01
6.24328852e-02 9.29148495e-01 2.87077725e-01 -4.53433126e-01
3.52646798e-01 -4.63753402e-01 -2.96619445e-01 8.19355369e-01
1.62462756e-01 -2.07746252e-01 1.73959136e-01 -1.98125675e-01
7.57067800e-01 6.57837212e-01 1.73050135e-01 -5.76084793e-01
4.69709396e-01 4.92749244e-01 4.05527115e-01 6.51579976e-01
-1.92533657e-01 9.55923423e-02 9.84401181e-02 -2.62764096e-01
-1.13407719e+00 -1.08872652e+00 -5.59571505e-01 1.40658343e+00
-3.19181204e-01 -1.46512628e-01 -9.87226546e-01 -1.13884056e+00
-3.09885070e-02 9.78922665e-01 -4.08382446e-01 -9.55755711e-02
-1.00487053e+00 -9.15615499e-01 1.08938968e+00 2.23878399e-01
2.38218501e-01 -8.42726111e-01 1.12488225e-01 3.20354223e-01
-7.02367544e-01 -1.17116964e+00 -6.37415648e-01 4.30655986e-01
-8.97589505e-01 -8.27671707e-01 -5.22288382e-01 -1.09318364e+00
4.22022611e-01 4.44222055e-02 1.56501544e+00 -2.55177200e-01
5.17495871e-01 1.56446725e-01 -3.28367025e-01 -2.31311515e-01
-1.09973514e+00 6.72071874e-01 1.26860812e-01 -2.54591942e-01
6.47835553e-01 -5.38252532e-01 7.84798637e-02 4.25915688e-01
-5.37523448e-01 7.90831298e-02 8.33422959e-01 8.02328169e-01
4.11651105e-01 -3.79763156e-01 6.51062369e-01 -1.24949610e+00
7.29319394e-01 -5.18690109e-01 -4.38426763e-01 6.42456055e-01
-7.91342616e-01 1.60658807e-01 8.00034702e-01 -5.60072780e-01
-8.96536469e-01 -4.47791934e-01 -1.43175334e-01 -8.19455609e-02
-6.25672415e-02 6.49566889e-01 -2.41054624e-01 3.36444601e-02
7.48907685e-01 3.02964747e-01 -3.62843841e-01 -4.71859485e-01
3.80994797e-01 8.00621510e-01 5.73658645e-01 -1.04381204e+00
8.13118160e-01 -8.18964839e-02 -3.81668657e-01 -4.18607622e-01
-8.93857062e-01 1.58194918e-02 -1.04802227e+00 4.76429611e-02
7.99139380e-01 -1.06306076e+00 9.55035090e-02 1.39206290e-01
-1.14780223e+00 -6.41675830e-01 3.25908028e-02 6.56375766e-01
-4.00228620e-01 -6.21280037e-02 -8.57086420e-01 -3.03206980e-01
-2.33175814e-01 -1.45716977e+00 1.08774316e+00 -5.11392534e-01
-4.39415276e-01 -1.46612620e+00 4.96255845e-01 5.91147244e-01
4.62164015e-01 -1.87007740e-01 1.19943285e+00 -7.50356257e-01
-3.09076816e-01 -1.29265189e-01 -5.02427965e-02 4.42426503e-01
2.31407121e-01 -5.18789031e-02 -1.07245481e+00 -2.91221589e-01
8.03256035e-02 -7.22214341e-01 4.78318453e-01 3.96687575e-02
3.45858365e-01 -3.10874969e-01 -3.29513520e-01 6.58142030e-01
1.25164747e+00 -2.91661751e-02 4.24017191e-01 4.88096744e-01
6.45912170e-01 7.85460949e-01 4.54062909e-01 -6.37064576e-01
9.53091979e-01 6.68848217e-01 -3.85791779e-01 -1.86830878e-01
-1.72045335e-01 -4.51874852e-01 8.99672866e-01 1.59203732e+00
5.65408953e-02 -1.35717839e-01 -1.11646760e+00 6.05480373e-01
-1.58571935e+00 -7.22445011e-01 -2.44926333e-01 2.38467622e+00
1.44371128e+00 1.36651024e-01 -5.13728224e-02 -1.50350109e-01
6.57900870e-01 -1.98274598e-01 -2.31145874e-01 -8.04355443e-01
-3.17478031e-01 4.13463771e-01 4.90423024e-01 9.96202230e-01
-9.03105795e-01 1.53361499e+00 7.00423813e+00 8.19716394e-01
-1.58081400e+00 5.75763226e-01 6.00810707e-01 2.89519548e-01
-3.96176636e-01 4.30833578e-01 -9.28237319e-01 3.95960927e-01
1.35198855e+00 -1.89265102e-01 6.14594936e-01 5.70876658e-01
6.19881786e-02 2.96931323e-02 -1.57916355e+00 4.83054340e-01
2.51322776e-01 -7.12954819e-01 1.19088985e-01 1.48031920e-01
8.74780476e-01 5.59422374e-01 1.83468722e-02 8.07887375e-01
6.21493697e-01 -8.66510212e-01 9.61617112e-01 -8.41544196e-02
1.08543897e+00 -4.83922631e-01 6.09841883e-01 5.06983817e-01
-9.57616746e-01 5.67712605e-01 -2.34101623e-01 -5.43484576e-02
2.64402144e-02 3.08890820e-01 -1.00422323e+00 7.13788807e-01
3.54598701e-01 5.62842965e-01 -8.12629879e-01 2.26950452e-01
-3.92990679e-01 9.94278729e-01 1.20474221e-02 2.78326124e-01
3.35697383e-01 -3.95243883e-01 3.60055178e-01 1.54833186e+00
2.77359158e-01 -7.72353113e-01 4.86514866e-01 5.64062476e-01
-1.47601753e-01 4.68845367e-01 -8.41158986e-01 -1.48313507e-01
3.38158846e-01 1.05225742e+00 -2.96687573e-01 -5.72079778e-01
-6.91078365e-01 1.21480572e+00 7.06602991e-01 2.87085563e-01
-7.66055703e-01 -4.58824486e-02 3.50050598e-01 6.41607866e-03
-1.59495726e-01 -2.77782440e-01 -4.52737570e-01 -1.48354876e+00
2.50891149e-01 -1.37124443e+00 2.64779329e-01 -5.82139254e-01
-1.34080434e+00 9.38621044e-01 -5.73301464e-02 -1.16040051e+00
-6.44479156e-01 -6.12587631e-01 -3.74386907e-01 1.37833512e+00
-1.73528934e+00 -1.55961382e+00 5.62668264e-01 4.97964114e-01
5.35152078e-01 4.50053951e-03 9.31815326e-01 6.50920093e-01
-4.47988003e-01 1.00059867e+00 9.45155174e-02 2.07051396e-01
1.28808916e+00 -1.25078321e+00 5.97738206e-01 8.58094275e-01
3.65323275e-01 7.45648444e-01 5.58436573e-01 -5.91733992e-01
-1.32823348e+00 -1.29740572e+00 1.84584737e+00 -1.04922307e+00
9.49320853e-01 -5.73739529e-01 -8.35070908e-01 1.24000371e+00
7.20259964e-01 -3.76452982e-01 6.41206026e-01 3.52510124e-01
-5.25249422e-01 1.24972843e-01 -7.81151950e-01 7.23000705e-01
8.93582225e-01 -1.10041904e+00 -7.20963538e-01 4.50896084e-01
5.19680440e-01 -2.50663161e-01 -9.21391845e-01 3.11611563e-01
6.50863469e-01 -5.62287509e-01 5.39542317e-01 -7.72408426e-01
4.63686407e-01 -6.85977861e-02 -4.14560497e-01 -1.80182111e+00
-8.97802189e-02 -3.27195734e-01 6.16862237e-01 1.36373305e+00
1.00154722e+00 -1.07254398e+00 1.26224846e-01 7.01050535e-02
-2.62140065e-01 -4.68958527e-01 -8.26527655e-01 -9.41662192e-01
9.53506887e-01 -5.37361622e-01 3.15727741e-01 1.53536224e+00
-1.69918798e-02 7.95893729e-01 -2.02925161e-01 1.31791696e-01
3.37427765e-01 1.58898458e-01 9.78652954e-01 -8.79327953e-01
-3.92811537e-01 -4.54329908e-01 1.23271197e-01 -8.81995916e-01
5.95775008e-01 -1.49584293e+00 2.24267960e-01 -1.08948505e+00
1.25674486e-01 -6.15418255e-01 -2.30210759e-02 4.94784862e-01
-3.94320905e-01 4.31591153e-01 8.51594098e-03 3.75594109e-01
-2.06424832e-01 1.70604393e-01 1.25493360e+00 -5.30217253e-02
1.60850093e-01 -1.66831851e-01 -7.40916312e-01 5.62341928e-01
6.79028451e-01 -7.22714782e-01 -9.62012336e-02 -1.04001868e+00
2.00649932e-01 -6.94987178e-03 -1.08715966e-01 -5.46197295e-01
1.00299090e-01 -2.30250672e-01 1.53752103e-01 -1.36388671e-02
-9.82122421e-02 -5.12926936e-01 -4.09739316e-02 2.26869792e-01
-3.61908287e-01 6.92847848e-01 3.39962780e-01 -1.32044122e-01
-3.73065859e-01 -1.72581449e-01 6.34644449e-01 -1.03509329e-01
-2.01427460e-01 2.16951687e-02 -4.61922765e-01 3.42535764e-01
6.03146195e-01 6.99661598e-02 -4.22287554e-01 -9.42844804e-03
-6.45853877e-01 1.32339625e-02 6.81443810e-01 5.84550917e-01
-2.68536568e-01 -1.50338590e+00 -1.10426068e+00 2.55773097e-01
2.86208242e-01 -6.51144385e-01 -4.03345108e-01 1.11802816e+00
-3.45043182e-01 7.13436782e-01 1.31246448e-01 -5.59921682e-01
-8.25223029e-01 3.34909767e-01 4.54296499e-01 -5.60607195e-01
-2.10365988e-02 4.40917730e-01 8.60818028e-02 -1.20301592e+00
-2.59240121e-01 -9.53481123e-02 2.85576820e-01 -2.62677930e-02
1.04556382e-01 8.89059156e-02 4.35645819e-01 -9.14998949e-01
-3.67833525e-01 5.91346800e-01 -1.37560248e-01 -6.11140072e-01
8.50268245e-01 -2.79145896e-01 -3.78617674e-01 8.20719004e-01
1.30774689e+00 5.76098263e-01 -6.34703100e-01 -4.14932281e-01
3.18042517e-01 -2.05131844e-01 -1.70703202e-01 -1.02685487e+00
-5.14562607e-01 9.32134688e-01 1.58484206e-01 -1.08210772e-01
8.08253348e-01 -2.56458800e-02 6.69043005e-01 2.96417326e-01
7.69278705e-01 -1.06922221e+00 -4.82309163e-01 8.54241729e-01
7.50493944e-01 -1.39440656e+00 -3.58173460e-01 -1.11207247e-01
-5.10429144e-01 7.49001622e-01 5.88379800e-01 7.69336447e-02
3.77756685e-01 3.32725823e-01 7.42660880e-01 2.60894537e-01
-9.49441910e-01 -9.65324044e-02 3.13669592e-01 2.55159616e-01
1.06788003e+00 3.46131742e-01 -4.89827543e-01 1.51562601e-01
-6.70122385e-01 -2.27190346e-01 -2.13098261e-04 8.76599491e-01
1.12698205e-01 -1.80089068e+00 -5.39348543e-01 1.92359880e-01
-7.53557444e-01 -6.24794424e-01 -6.65745020e-01 1.00488687e+00
3.20580810e-01 1.01883328e+00 -4.60315719e-02 -4.55100834e-01
1.72283173e-01 6.38618827e-01 7.47933209e-01 -8.24459434e-01
-8.47793579e-01 4.10325304e-02 4.04951751e-01 -1.87677115e-01
-2.66039640e-01 -9.56613541e-01 -4.25149262e-01 -3.88387769e-01
-2.55123436e-01 3.63195509e-01 7.95681179e-01 1.32963490e+00
1.48744226e-01 1.38387427e-01 6.09106839e-01 -5.24279118e-01
-7.26240754e-01 -1.32015204e+00 1.51419044e-01 2.14519560e-01
1.68996438e-01 -2.15715632e-01 -4.03121501e-01 2.04371497e-01] | [11.231080055236816, 10.242451667785645] |
bdc52381-9ed8-451f-b94b-87268465202c | phasebook-and-friends-leveraging-discrete | 1810.01395 | null | http://arxiv.org/abs/1810.01395v2 | http://arxiv.org/pdf/1810.01395v2.pdf | Phasebook and Friends: Leveraging Discrete Representations for Source Separation | Deep learning based speech enhancement and source separation systems have
recently reached unprecedented levels of quality, to the point that performance
is reaching a new ceiling. Most systems rely on estimating the magnitude of a
target source by estimating a real-valued mask to be applied to a
time-frequency representation of the mixture signal. A limiting factor in such
approaches is a lack of phase estimation: the phase of the mixture is most
often used when reconstructing the estimated time-domain signal. Here, we
propose "magbook", "phasebook", and "combook", three new types of layers based
on discrete representations that can be used to estimate complex time-frequency
masks. Magbook layers extend classical sigmoidal units and a recently
introduced convex softmax activation for mask-based magnitude estimation.
Phasebook layers use a similar structure to give an estimate of the phase mask
without suffering from phase wrapping issues. Combook layers are an alternative
to the magbook-phasebook combination that directly estimate complex masks. We
present various training and inference schemes involving these representations,
and explain in particular how to include them in an end-to-end learning
framework. We also present an oracle study to assess upper bounds on
performance for various types of masks using discrete phase representations. We
evaluate the proposed methods on the wsj0-2mix dataset, a well-studied corpus
for single-channel speaker-independent speaker separation, matching the
performance of state-of-the-art mask-based approaches without requiring
additional phase reconstruction steps. | ['Shinji Watanabe', 'John R. Hershey', 'Jonathan Le Roux', 'Gordon Wichern', 'Andy Sarroff'] | 2018-10-02 | null | null | null | null | ['speaker-separation'] | ['speech'] | [ 2.52080977e-01 -1.20672747e-01 7.44813541e-03 -3.59795511e-01
-1.41686928e+00 -5.18188953e-01 6.21667504e-01 1.43789396e-01
-6.54877782e-01 5.40263951e-01 2.56520271e-01 -1.94813639e-01
-2.22174644e-01 -1.13631681e-01 -6.47003591e-01 -9.83594000e-01
-3.58877540e-01 1.50073588e-01 -1.07686815e-03 -1.96181372e-01
-3.14716324e-02 2.25063547e-01 -1.40423739e+00 3.48289192e-01
7.64209449e-01 1.10766363e+00 3.85705441e-01 9.67960536e-01
1.14867538e-01 2.44304180e-01 -1.27303195e+00 3.70259508e-02
1.42796114e-01 -4.10592914e-01 -1.97683975e-01 -2.56207049e-01
2.39832610e-01 -8.83856416e-02 -3.38906527e-01 9.36849952e-01
9.49526310e-01 1.68753237e-01 7.55930305e-01 -9.54091370e-01
-1.40347704e-01 9.16143775e-01 -5.20223379e-01 4.32273477e-01
3.18763375e-01 1.15678133e-02 6.78910017e-01 -7.98440754e-01
1.35447130e-01 9.42712545e-01 9.68731344e-01 2.65782386e-01
-1.27618313e+00 -6.94815159e-01 7.63344541e-02 4.07377511e-01
-1.42414880e+00 -1.11961615e+00 9.72053707e-01 -1.57525629e-01
1.32353663e+00 4.06056583e-01 2.68663138e-01 9.78787065e-01
-2.04616725e-01 5.61538696e-01 1.22280955e+00 -6.76951289e-01
2.39470437e-01 1.65009961e-01 3.69213074e-02 3.26488197e-01
-1.67304605e-01 2.95490921e-01 -8.66675794e-01 -8.34701210e-02
5.07008851e-01 -6.53496623e-01 -8.23339462e-01 2.01765344e-01
-1.10635972e+00 5.00625014e-01 3.29531044e-01 5.01455963e-01
-4.09688115e-01 3.24951738e-01 1.72867894e-01 3.25543225e-01
6.95676446e-01 3.44816595e-01 -4.10651922e-01 -1.72511786e-01
-1.57401335e+00 1.25823423e-01 7.90443242e-01 5.72715282e-01
4.16439116e-01 5.83777010e-01 -3.58219117e-01 1.03015590e+00
4.78659332e-01 5.66251993e-01 6.79154336e-01 -5.50912917e-01
5.63925922e-01 -3.42998028e-01 1.91139415e-01 -4.14475620e-01
-5.29890537e-01 -8.06776345e-01 -6.70225799e-01 4.30883408e-01
6.79136038e-01 -3.97142261e-01 -1.16097367e+00 1.92175055e+00
1.24888152e-01 4.73298311e-01 1.45729661e-01 8.20511162e-01
7.23011017e-01 9.03127849e-01 -2.34179378e-01 -4.71994638e-01
1.39979696e+00 -7.69103527e-01 -1.08582449e+00 -4.48543727e-01
-7.94101134e-02 -8.38669717e-01 4.65708464e-01 7.55055010e-01
-1.32268643e+00 -7.64684498e-01 -1.44627869e+00 2.50139862e-01
-2.71436036e-01 3.99236858e-01 2.77957052e-01 1.07278764e+00
-1.32666373e+00 8.94708991e-01 -8.35493326e-01 4.28812772e-01
1.13693848e-01 6.22586906e-01 -1.72086298e-01 4.11061317e-01
-1.17252409e+00 8.48125756e-01 2.38893494e-01 4.46797490e-01
-7.42264748e-01 -8.62978458e-01 -8.92727137e-01 2.72153854e-01
3.61894295e-02 -1.88235745e-01 1.41005862e+00 -1.01556003e+00
-1.83913004e+00 5.90812922e-01 -4.85984445e-01 -8.10968757e-01
2.46378154e-01 -1.98128015e-01 -8.89608860e-01 2.40144864e-01
-3.19348752e-01 3.81555349e-01 1.72767508e+00 -1.03650999e+00
-2.93035865e-01 -5.49187884e-02 -3.07494879e-01 1.30224526e-01
-6.67796284e-02 1.32813126e-01 -2.18299687e-01 -1.01281023e+00
1.85064241e-01 -4.67437685e-01 -1.51921570e-01 -2.69409269e-01
-3.36400509e-01 -1.91946793e-02 4.91162121e-01 -1.09210169e+00
1.20558965e+00 -2.32691097e+00 7.72671700e-02 2.10021734e-02
4.93151918e-02 4.83982772e-01 -1.46915346e-01 3.29256058e-01
-4.14569438e-01 -2.25958288e-01 -5.62575579e-01 -8.19550753e-01
1.61532447e-01 -4.09059972e-01 -3.39965612e-01 6.67272210e-01
4.12664473e-01 4.39699352e-01 -5.76965392e-01 -5.85198887e-02
3.32867533e-01 1.01378870e+00 -4.83525783e-01 2.45192155e-01
-7.69256428e-02 3.27981025e-01 4.60730970e-01 9.74314287e-02
9.93439019e-01 2.11735845e-01 2.18714640e-01 -4.51137155e-01
-2.02736020e-01 8.36415648e-01 -1.51430357e+00 1.73853731e+00
-7.13185310e-01 1.11697972e+00 6.26458585e-01 -9.59920466e-01
7.09611654e-01 9.45787728e-01 2.49205470e-01 -4.19061303e-01
1.46591723e-01 5.50524771e-01 2.35964328e-01 -2.49725297e-01
2.32083961e-01 -2.90209651e-01 2.55172044e-01 2.87562162e-01
4.02428210e-01 -2.52927959e-01 -3.55871138e-03 -2.86279321e-01
7.77491033e-01 9.57406983e-02 2.36805290e-01 -1.41720369e-01
5.90965211e-01 -6.06848598e-01 2.58836448e-01 4.59363699e-01
-2.02605098e-01 8.12992156e-01 2.50875682e-01 3.30615729e-01
-5.90567410e-01 -1.23613548e+00 -1.34688318e-01 8.94900441e-01
-2.45805934e-01 -2.09486514e-01 -1.00930882e+00 -1.12769082e-01
-3.01180273e-01 7.61342585e-01 -2.95105964e-01 -2.01447657e-03
-8.44844818e-01 -8.60099435e-01 7.26846337e-01 4.18136984e-01
3.00667107e-01 -8.81877601e-01 -3.99565190e-01 5.17466247e-01
-3.91662151e-01 -1.18456352e+00 -5.85233927e-01 8.87241721e-01
-5.77198029e-01 -5.70385277e-01 -1.08728898e+00 -7.42695749e-01
4.30422127e-01 1.27380714e-01 7.71263659e-01 -5.00012159e-01
4.63712439e-02 1.55369505e-01 -8.15297291e-02 -5.84201217e-01
-4.81905162e-01 -1.62822545e-01 2.63499945e-01 2.64173359e-01
9.48250145e-02 -8.83064628e-01 -5.91059566e-01 -1.07868370e-02
-7.84651041e-01 -1.84032366e-01 4.87467557e-01 6.49867535e-01
1.97953150e-01 2.60363132e-01 9.79124784e-01 -2.56357402e-01
7.02776015e-01 -2.81484514e-01 -5.87793112e-01 -2.10904345e-01
-3.50156486e-01 2.09692419e-01 6.99814916e-01 -8.31601083e-01
-1.10299015e+00 -2.22070515e-02 -7.46568441e-01 -4.48009521e-01
-6.33550063e-02 3.04859370e-01 -3.06779385e-01 9.12785307e-02
6.62242591e-01 2.37002656e-01 -9.29141641e-02 -6.93928719e-01
2.23592564e-01 9.79735732e-01 8.00699532e-01 -1.64563581e-01
8.63615036e-01 3.48848313e-01 -4.01787102e-01 -1.13175178e+00
-4.31471378e-01 -6.26815200e-01 -2.22377747e-01 1.57368809e-01
5.79926610e-01 -1.04848206e+00 -5.77465117e-01 7.78745234e-01
-1.22534204e+00 -4.80986565e-01 -2.95398802e-01 7.74270177e-01
-5.89044690e-01 3.07338148e-01 -5.64425886e-01 -1.13011730e+00
-4.61250246e-01 -1.08935356e+00 1.06691861e+00 3.25143903e-01
-8.17250162e-02 -8.25330555e-01 4.27390542e-03 1.70305565e-01
7.66743600e-01 -2.15429571e-02 4.77466762e-01 -5.25191247e-01
-9.55720916e-02 -8.90998840e-02 2.13288724e-01 5.67685187e-01
9.19695050e-02 -3.71144027e-01 -1.87144721e+00 -3.43637198e-01
5.53790033e-01 2.25520208e-01 9.45961177e-01 9.31862056e-01
8.38719904e-01 -4.04208809e-01 -2.99123049e-01 7.96975911e-01
1.20588100e+00 3.80943596e-01 6.31261468e-01 -7.02605769e-02
2.89500922e-01 2.90249527e-01 1.71521634e-01 3.46307278e-01
2.27319077e-02 8.14970315e-01 1.25207588e-01 -2.44849712e-01
-5.07884622e-01 7.64872655e-02 6.46970809e-01 9.05674756e-01
8.60590562e-02 -3.98991942e-01 -4.56897467e-01 5.29086649e-01
-1.21732008e+00 -1.01905978e+00 9.63430479e-03 2.50059271e+00
1.12155354e+00 3.55011582e-01 1.15861058e-01 8.35063696e-01
6.14207327e-01 4.67181891e-01 -2.17933238e-01 -4.10200059e-01
-5.50139248e-02 8.29199016e-01 3.61580193e-01 8.79929483e-01
-1.24823999e+00 4.73304510e-01 6.31497622e+00 1.06140578e+00
-1.49073207e+00 3.39200795e-01 3.46813112e-01 -1.46702752e-01
-4.51321118e-02 -3.82034540e-01 -6.02916837e-01 3.94980580e-01
1.40689051e+00 1.39528766e-01 7.53637135e-01 3.18379968e-01
2.45767012e-01 -1.53926760e-01 -1.18632960e+00 1.18245459e+00
-3.43989022e-02 -1.06602466e+00 -8.21365356e-01 -2.03109086e-01
3.78486782e-01 2.74896715e-02 2.40696356e-01 3.61525506e-01
-3.69155139e-01 -1.20038104e+00 9.89080667e-01 2.49631450e-01
1.09949028e+00 -6.19099140e-01 3.52266192e-01 3.19350213e-01
-1.43207908e+00 -1.05093628e-01 3.68610620e-02 4.60156752e-03
2.85389334e-01 7.04849362e-01 -1.13445032e+00 4.74395961e-01
4.52401102e-01 8.80176201e-02 -1.06566757e-01 1.16413069e+00
-3.83998841e-01 9.46312428e-01 -6.17082059e-01 2.10924566e-01
-2.81090592e-03 1.73439756e-01 8.33367467e-01 1.56915784e+00
2.36692086e-01 -2.55485624e-01 -3.37883413e-01 7.78382301e-01
1.21756466e-02 -3.79973471e-01 -2.00472325e-02 -1.38562784e-01
5.98713338e-01 1.18863678e+00 -6.09073937e-01 -1.86835974e-01
-1.81535840e-01 9.32695389e-01 -4.29638177e-02 6.98077917e-01
-9.69020844e-01 -6.70081019e-01 7.38896251e-01 1.27623364e-01
6.24456048e-01 -4.03590918e-01 -7.12393448e-02 -6.93274617e-01
1.45089567e-01 -9.95261312e-01 -1.37209460e-01 -5.66021085e-01
-8.11187565e-01 8.77908945e-01 5.41190468e-02 -1.19211888e+00
-6.30403101e-01 -6.29465938e-01 -6.89583957e-01 1.36405456e+00
-1.82705355e+00 -6.32392764e-01 4.17249203e-02 3.94992918e-01
5.72975934e-01 1.22782066e-02 9.13394094e-01 5.30743539e-01
-2.72763968e-01 7.12870121e-01 1.44180864e-01 -2.66174059e-02
7.41768658e-01 -1.32719541e+00 5.42969048e-01 9.16490912e-01
3.96975547e-01 4.75421935e-01 1.03571749e+00 -1.55462801e-01
-1.21123660e+00 -7.01970041e-01 8.10770988e-01 -6.05020449e-02
2.15219542e-01 -7.27708459e-01 -9.38221455e-01 3.35351378e-01
4.67041969e-01 -2.36293133e-02 4.94721264e-01 -5.20882383e-02
-3.81731182e-01 -3.52022022e-01 -1.07677329e+00 2.63523042e-01
5.27962387e-01 -6.89949214e-01 -5.98204017e-01 5.62878661e-02
7.34468341e-01 -5.62773108e-01 -4.89598900e-01 2.66019613e-01
4.30152535e-01 -1.14884007e+00 1.18385601e+00 1.31932184e-01
-2.29296371e-01 -5.13878465e-01 -8.75225663e-02 -1.73820424e+00
-1.50489867e-01 -1.24164557e+00 -4.93868649e-01 1.25400853e+00
4.32278395e-01 -8.50939810e-01 4.54868734e-01 -1.33249741e-02
-3.47111344e-01 -5.49193263e-01 -1.24082041e+00 -7.87818253e-01
-2.00575367e-01 -7.56465137e-01 4.84437972e-01 5.36541224e-01
-1.72927952e-03 2.15865493e-01 -2.18267560e-01 5.54797232e-01
6.06847048e-01 -2.89312810e-01 2.55250424e-01 -8.37728858e-01
-8.98502231e-01 -7.32429564e-01 -1.60639137e-01 -1.04746604e+00
-1.36558384e-01 -6.92642391e-01 4.47170883e-01 -1.38476384e+00
-5.68295717e-01 -2.34610483e-01 -4.82794821e-01 5.35778031e-02
-3.13949645e-01 3.11936736e-01 1.47779927e-01 -1.90157771e-01
9.45361033e-02 3.97783250e-01 6.80741549e-01 -3.45933616e-01
-4.20428902e-01 3.48214000e-01 -5.31394422e-01 7.36981571e-01
6.85421705e-01 -4.56219047e-01 -3.76836628e-01 -3.68313521e-01
-9.15261358e-02 3.43729347e-01 2.88050115e-01 -1.43650746e+00
2.45006889e-01 4.75634009e-01 5.33678949e-01 -4.99172121e-01
1.03482616e+00 -6.44947529e-01 1.80822432e-01 3.80650640e-01
-2.34481364e-01 -3.93393725e-01 4.87007201e-01 2.78685212e-01
-3.72920603e-01 -3.25631440e-01 8.91385555e-01 2.67593324e-01
-2.44345546e-01 -4.66453619e-02 -5.41101336e-01 -8.68130382e-03
4.46482629e-01 -1.59133986e-01 -6.73885792e-02 -6.69929445e-01
-6.43454194e-01 -2.88418829e-01 -2.71805286e-01 1.71942785e-01
5.45924842e-01 -1.00797665e+00 -9.01717961e-01 4.16942805e-01
-3.97116780e-01 -2.34546840e-01 3.95978004e-01 9.07241523e-01
-2.47392640e-01 4.27537620e-01 6.58746734e-02 -5.36927938e-01
-1.06531000e+00 4.51224238e-01 7.29063570e-01 -3.53960663e-01
-4.28356558e-01 1.11756432e+00 1.85603976e-01 -8.65519792e-02
5.58405638e-01 -7.00257659e-01 -2.14680642e-01 1.49588883e-01
7.64504433e-01 2.96290576e-01 3.72576088e-01 -4.86234784e-01
-5.45828223e-01 3.54922831e-01 3.28755796e-01 -7.60705292e-01
1.25210619e+00 4.42589149e-02 3.03904027e-01 5.38288713e-01
1.22722828e+00 5.36394835e-01 -1.33539295e+00 -1.50648490e-01
-1.13821916e-01 8.86716601e-03 3.16267282e-01 -9.73583758e-01
-9.00083482e-01 1.11896718e+00 9.96465564e-01 3.68216068e-01
1.57998323e+00 -3.18786442e-01 6.46635234e-01 -7.22192600e-02
-8.69503524e-03 -9.53373551e-01 -1.23653390e-01 1.66200757e-01
9.87309873e-01 -1.01689410e+00 -1.91061482e-01 -2.36902669e-01
-1.54615030e-01 1.05412626e+00 -2.49478873e-03 -6.75811544e-02
7.34237790e-01 7.66033888e-01 2.12909624e-01 2.78771788e-01
-3.95839453e-01 -2.81818151e-01 5.54049194e-01 8.48351657e-01
6.19025111e-01 1.10062994e-01 2.95498576e-02 7.69839942e-01
-3.95486116e-01 -3.88888478e-01 2.41882741e-01 5.27368426e-01
-4.60374475e-01 -1.08214962e+00 -8.95527959e-01 1.01200476e-01
-7.26456821e-01 -4.29248393e-01 1.19473241e-01 3.81631523e-01
3.67018320e-02 1.32193196e+00 9.96022373e-02 -1.33222148e-01
2.16965765e-01 2.93892503e-01 7.33143389e-01 -3.02214950e-01
-8.31339836e-01 6.55467749e-01 2.43576020e-01 -1.42309874e-01
-2.62187093e-01 -4.69302863e-01 -1.26587212e+00 2.09045365e-01
-5.07475793e-01 1.89927474e-01 1.09460211e+00 8.48432362e-01
5.57383448e-02 1.07139492e+00 5.49040616e-01 -1.37536478e+00
-6.54392600e-01 -1.42357481e+00 -4.96429354e-01 -2.54065450e-02
1.04343987e+00 -2.99473584e-01 -4.96860892e-01 2.43960638e-02] | [15.12677001953125, 5.782646179199219] |
a1a017ec-cc62-4474-80bc-275023d491df | dual-local-global-contextual-pathways-for | 1605.05462 | null | http://arxiv.org/abs/1605.05462v1 | http://arxiv.org/pdf/1605.05462v1.pdf | Dual Local-Global Contextual Pathways for Recognition in Aerial Imagery | Visual context is important in object recognition and it is still an open
problem in computer vision. Along with the advent of deep convolutional neural
networks (CNN), using contextual information with such systems starts to
receive attention in the literature. At the same time, aerial imagery is
gaining momentum. While advances in deep learning make good progress in aerial
image analysis, this problem still poses many great challenges. Aerial images
are often taken under poor lighting conditions and contain low resolution
objects, many times occluded by trees or taller buildings. In this domain, in
particular, visual context could be of great help, but there are still very few
papers that consider context in aerial image understanding. Here we introduce
context as a complementary way of recognizing objects. We propose a dual-stream
deep neural network model that processes information along two independent
pathways, one for local and another for global visual reasoning. The two are
later combined in the final layers of processing. Our model learns to combine
local object appearance as well as information from the larger scene at the
same time and in a complementary way, such that together they form a powerful
classifier. We test our dual-stream network on the task of segmentation of
buildings and roads in aerial images and obtain state-of-the-art results on the
Massachusetts Buildings Dataset. We also introduce two new datasets, for
buildings and road segmentation, respectively, and study the relative
importance of local appearance vs. the larger scene, as well as their
performance in combination. While our local-global model could also be useful
in general recognition tasks, we clearly demonstrate the effectiveness of
visual context in conjunction with deep nets for aerial image understanding. | ['Alina Marcu', 'Marius Leordeanu'] | 2016-05-18 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 5.29099703e-01 -2.77699113e-01 -2.09500524e-03 -5.25256753e-01
-2.77661085e-01 -8.60951781e-01 7.50653625e-01 4.75202680e-01
-3.79921615e-01 4.73203868e-01 1.08490214e-01 -4.45308745e-01
-2.01778010e-01 -1.08308852e+00 -7.82284617e-01 -7.44588673e-01
5.23789531e-05 1.75319329e-01 2.67834097e-01 -3.81628066e-01
6.42159767e-03 6.27892733e-01 -1.72539234e+00 5.44869006e-01
6.04753852e-01 9.74641025e-01 3.79797161e-01 7.59063363e-01
-5.12773581e-02 8.89348984e-01 -4.20865685e-01 -1.36334911e-01
4.82579410e-01 -2.76829571e-01 -1.08507705e+00 5.78887284e-01
9.61147189e-01 -2.91015178e-01 -9.89930779e-02 8.17232072e-01
1.62181884e-01 3.10610197e-02 3.96255016e-01 -8.98177266e-01
-7.13753819e-01 3.65616590e-01 -5.34021318e-01 2.96530813e-01
8.87030065e-02 3.96249890e-01 1.17569816e+00 -5.45614541e-01
4.76822793e-01 9.93562281e-01 5.70533633e-01 8.60760063e-02
-1.36580300e+00 -2.28928268e-01 8.14944386e-01 1.79222450e-01
-1.17177212e+00 -4.23598662e-02 7.31177330e-01 -4.77715701e-01
9.05402422e-01 3.06109518e-01 9.79063094e-01 8.88997853e-01
-4.08495516e-02 9.55188572e-01 1.16444921e+00 -2.97588825e-01
9.52796191e-02 -1.09603144e-01 1.29639030e-01 6.57462478e-01
3.78694609e-02 1.62012339e-01 -1.00094743e-01 3.82430047e-01
8.36409330e-01 3.66160423e-01 -2.68543452e-01 -2.17980936e-01
-1.15956664e+00 8.12041759e-01 1.11392236e+00 4.70938474e-01
-3.33724678e-01 3.42379093e-01 8.02446753e-02 2.09663436e-01
4.99057651e-01 4.99493182e-01 -5.08132637e-01 4.90577936e-01
-1.16209185e+00 4.17355120e-01 5.52165091e-01 7.05501080e-01
7.52815187e-01 -5.82381114e-02 -1.38305621e-02 7.44274139e-01
1.71338752e-01 4.24456835e-01 -1.67076945e-01 -7.34171510e-01
4.15166527e-01 8.91434669e-01 4.15153010e-03 -1.12599134e+00
-5.02686739e-01 -6.18461907e-01 -7.59038329e-01 4.01436269e-01
7.05688298e-01 -1.51561350e-02 -1.14151728e+00 1.67973506e+00
2.89302111e-01 3.16036493e-01 2.76602637e-02 1.05956733e+00
7.33291268e-01 7.82431483e-01 2.38124803e-01 3.73940438e-01
1.42251801e+00 -9.30979073e-01 -1.87796623e-01 -8.30688655e-01
3.11554432e-01 -7.94448376e-01 8.28445554e-01 2.68444747e-01
-8.44981372e-01 -6.70609832e-01 -1.11490369e+00 -2.79609829e-01
-8.54820549e-01 2.75486976e-01 8.59852076e-01 2.29464009e-01
-1.12290680e+00 6.63309395e-01 -5.52524149e-01 -7.13451385e-01
7.26041913e-01 3.50796342e-01 -2.73537219e-01 -3.83640140e-01
-9.90623236e-01 9.37289476e-01 1.67662829e-01 4.24077183e-01
-9.91070688e-01 -4.60656375e-01 -8.94674122e-01 4.03472111e-02
6.37100518e-01 -7.90056169e-01 8.98923695e-01 -1.47099161e+00
-7.74112284e-01 9.94021177e-01 -4.52015139e-02 -5.47194660e-01
3.53250325e-01 -2.59069979e-01 -1.58789083e-01 1.99984729e-01
6.59459010e-02 9.10295904e-01 9.05676901e-01 -1.42357457e+00
-1.05113900e+00 -6.06334269e-01 7.72910774e-01 3.34967434e-01
1.29388958e-01 -2.42708936e-01 -2.46153668e-01 -5.04449904e-01
1.65440336e-01 -9.60599303e-01 -4.10687268e-01 2.20254436e-01
-3.84022370e-02 -1.21787511e-01 9.97814715e-01 -5.50152361e-01
9.54672337e-01 -2.03744006e+00 3.52819443e-01 3.79921570e-02
1.41703829e-01 3.77684802e-01 -3.03585529e-01 3.17971408e-01
-2.79288441e-01 3.58821511e-01 -8.54606390e-01 -1.75402954e-01
-3.02041590e-01 4.82312292e-01 -4.36853617e-01 2.21021667e-01
7.88959444e-01 1.06854558e+00 -1.10353553e+00 -1.22467592e-01
4.60633218e-01 5.18110931e-01 -2.44367450e-01 -5.84263615e-02
-4.22410756e-01 4.35885787e-01 -1.59028217e-01 8.49033117e-01
4.08577472e-01 -9.79219452e-02 2.74930626e-01 -1.67499706e-01
-2.48373345e-01 -6.08709715e-02 -1.12557292e+00 1.58270812e+00
-5.73783875e-01 1.04575026e+00 2.53198087e-01 -1.29930496e+00
5.96547902e-01 1.24262773e-01 8.81026685e-02 -5.47209859e-01
1.33402035e-01 -1.50474951e-01 4.47439328e-02 -5.40874660e-01
5.59849143e-01 -2.26505205e-01 1.06333330e-01 1.23371273e-01
-1.34650752e-01 -4.39390302e-01 2.48428211e-01 6.50706291e-02
8.43817472e-01 2.81691581e-01 3.07922870e-01 -1.32134154e-01
3.81097138e-01 2.72189796e-01 1.86310962e-01 5.09983122e-01
-2.19157651e-01 7.46881545e-01 3.98426265e-01 -8.07737589e-01
-8.24712396e-01 -6.94816470e-01 -8.55967551e-02 9.89523947e-01
4.21562523e-01 -3.09489846e-01 -5.26845932e-01 -7.92815208e-01
1.46291047e-01 5.16761601e-01 -9.65200961e-01 9.35663730e-02
-5.72136045e-01 -7.12231338e-01 2.76714802e-01 9.74846423e-01
7.77046740e-01 -1.11134946e+00 -8.21534514e-01 8.38387534e-02
-6.40870556e-02 -1.27168596e+00 1.33518860e-01 4.82738286e-01
-1.02525079e+00 -1.20033550e+00 -5.17545342e-01 -6.58287108e-01
6.10663414e-01 7.34301627e-01 1.29623938e+00 5.61379492e-01
-5.25647819e-01 6.21209145e-01 -2.74555355e-01 -4.42451715e-01
1.45557866e-01 6.00906424e-02 -5.41017234e-01 1.93506591e-02
2.86050677e-01 -5.28472006e-01 -6.06529236e-01 3.53100747e-01
-1.22760987e+00 8.39405656e-02 6.03451669e-01 8.48333418e-01
5.18687129e-01 2.24797904e-01 1.16344191e-01 -7.86059618e-01
2.11574599e-01 -5.49752355e-01 -5.74900448e-01 3.00438702e-01
-2.16161132e-01 -1.78484097e-01 5.63858569e-01 4.92732264e-02
-8.86211574e-01 1.90299287e-01 -1.46753177e-01 -1.76056072e-01
-8.75816107e-01 7.28776813e-01 -8.88416991e-02 -8.08638483e-02
5.89244545e-01 -7.87890553e-02 -3.81242543e-01 -1.64902225e-01
4.78331476e-01 2.09661186e-01 3.03455532e-01 -4.70596969e-01
7.54722536e-01 7.10430324e-01 2.00498223e-01 -1.25814509e+00
-1.10943902e+00 -5.84247768e-01 -1.06421578e+00 -1.93473905e-01
1.16097391e+00 -9.48146522e-01 -2.97208250e-01 3.30596745e-01
-9.47037518e-01 -6.04613721e-01 -3.55306298e-01 8.57778117e-02
-2.00538307e-01 1.31347075e-01 -2.42138207e-01 -7.61377513e-01
1.71863422e-01 -1.30510521e+00 1.24723995e+00 3.51496220e-01
7.30035380e-02 -1.18144798e+00 -4.16472226e-01 5.10463357e-01
2.18582898e-01 5.59573412e-01 1.04314995e+00 -2.32475474e-01
-1.12060094e+00 -1.83653116e-01 -5.01571178e-01 3.99634421e-01
8.43101889e-02 1.68412536e-01 -1.24149871e+00 5.50261419e-03
-2.99650073e-01 -2.79190868e-01 1.29404795e+00 4.36931670e-01
9.93309319e-01 1.55320525e-01 -3.06048155e-01 4.76327300e-01
1.76096654e+00 8.40143561e-02 6.68742418e-01 2.10975036e-01
8.91460836e-01 9.31376994e-01 6.81064188e-01 -1.43685620e-02
3.25405389e-01 5.27724147e-01 9.97073710e-01 -5.08643150e-01
-2.78141052e-01 3.61255445e-02 -4.16605882e-02 2.13446632e-01
-4.64281321e-01 -1.32548392e-01 -1.18679380e+00 7.83268452e-01
-1.96336210e+00 -1.07688391e+00 -3.21344286e-01 2.03512478e+00
4.67026472e-01 2.38653701e-02 -3.66083942e-02 2.72330523e-01
4.34188187e-01 4.05174106e-01 -4.04597700e-01 -2.06584021e-01
-3.59011889e-01 2.69446820e-01 3.90187770e-01 4.87630397e-01
-1.60696959e+00 1.15682423e+00 5.67921400e+00 3.85235995e-01
-1.30010557e+00 -2.36131743e-01 7.60998964e-01 3.76648121e-02
-1.12647135e-02 2.55022168e-01 -6.51680768e-01 5.24035804e-02
4.99338061e-01 7.45381653e-01 3.80310595e-01 9.45409179e-01
-8.88921320e-02 -5.96029639e-01 -1.15317142e+00 8.37602139e-01
7.67132342e-02 -1.29474509e+00 1.70348883e-02 1.42984495e-01
7.93687701e-01 6.18173443e-02 8.53281170e-02 1.34405166e-01
2.12555453e-01 -1.11119151e+00 7.82580256e-01 2.62119859e-01
4.47075129e-01 -7.49479055e-01 8.12983811e-01 2.23740473e-01
-1.64570248e+00 -4.36876625e-01 -3.96314651e-01 -2.94701070e-01
-6.33755699e-02 4.45880949e-01 -5.55703640e-01 7.02224970e-01
7.36483216e-01 1.13718331e+00 -9.19421613e-01 9.95644391e-01
-5.79085469e-01 5.77795744e-01 -2.11355567e-01 3.07260215e-01
7.95625925e-01 -2.17273638e-01 4.20833319e-01 1.17249417e+00
-1.79906383e-01 4.17978495e-01 4.97527421e-01 9.30659592e-01
-1.13101797e-02 -2.28033736e-01 -9.41914916e-01 -1.12676211e-01
-5.24401702e-02 1.47308123e+00 -1.19249642e+00 -3.47364306e-01
-5.10721803e-01 9.18289006e-01 2.88204998e-01 4.28416729e-01
-6.35803580e-01 -1.15907192e-01 7.63196886e-01 3.49715948e-02
4.39530641e-01 -6.25774801e-01 -4.14993376e-01 -1.04175258e+00
-1.72430038e-01 -7.29022384e-01 2.71502674e-01 -9.01440203e-01
-1.14067364e+00 4.16793168e-01 1.25254676e-01 -9.79757130e-01
2.08356291e-01 -9.92869556e-01 -5.90188980e-01 8.15847218e-01
-2.00829530e+00 -1.61659849e+00 -7.16332674e-01 5.57447731e-01
9.39099669e-01 3.17267776e-01 7.83657074e-01 8.36478770e-02
-4.41031277e-01 -1.23376340e-01 -1.80436641e-01 2.81409323e-01
4.71557796e-01 -1.33680034e+00 4.37101692e-01 1.09797311e+00
4.86212045e-01 6.74079776e-01 3.29407483e-01 -5.22885740e-01
-1.06062162e+00 -1.19692492e+00 5.87076008e-01 -7.42467284e-01
4.66484964e-01 -3.28682333e-01 -8.03444982e-01 6.47923350e-01
3.89059514e-01 8.57046694e-02 5.54552197e-01 3.51380914e-01
-3.30397695e-01 -1.57866538e-01 -8.09739470e-01 7.21801102e-01
1.11418211e+00 -6.55775905e-01 -6.63832426e-01 2.45029375e-01
5.38121700e-01 -1.88612834e-01 -5.06920278e-01 4.16265756e-01
4.84900266e-01 -1.19472444e+00 1.15627325e+00 -5.85860908e-01
7.96501279e-01 -6.35732949e-01 -5.33217013e-01 -1.24475336e+00
-3.41799736e-01 4.88827042e-02 4.03374940e-01 1.11120093e+00
4.11159486e-01 -2.87629128e-01 5.35473764e-01 5.13073146e-01
-1.89341500e-01 -8.90716553e-01 -3.77588749e-01 -5.39240897e-01
-7.73270708e-03 -6.88512802e-01 4.39960510e-01 9.66642797e-01
-4.73118424e-01 5.97588301e-01 -1.99561790e-01 2.68914342e-01
2.45876566e-01 5.09099603e-01 6.74167573e-01 -1.37780571e+00
8.80354084e-03 -5.22821605e-01 -4.47548062e-01 -9.60706115e-01
-7.19462195e-03 -9.95776415e-01 7.30520114e-02 -2.01829791e+00
-5.27996235e-02 -4.84122902e-01 -6.59912080e-02 7.29657292e-01
-3.58869821e-01 3.50588679e-01 4.88923252e-01 -6.81912974e-02
-3.42136770e-01 2.32082412e-01 1.00316048e+00 -4.38973218e-01
-4.10788879e-02 -4.13972624e-02 -7.16973841e-01 8.68109941e-01
7.47570455e-01 -9.02145579e-02 -4.65745717e-01 -8.73690546e-01
1.96187481e-01 -3.39702338e-01 8.52841794e-01 -1.12252748e+00
3.01330071e-02 -1.04663171e-01 8.04303586e-01 -4.64428633e-01
3.16835254e-01 -1.29958785e+00 -2.81946719e-01 1.89010382e-01
-1.95730776e-01 2.24947669e-02 4.67669338e-01 7.60857880e-01
-3.03091496e-01 -1.29244685e-01 7.04927623e-01 -5.83604753e-01
-1.04727864e+00 2.43604466e-01 -3.75588924e-01 -4.42032218e-02
9.23195481e-01 -4.23909932e-01 -2.55464673e-01 -2.46542811e-01
-7.98786581e-01 3.20431352e-01 4.41413283e-01 4.97105420e-01
5.99763572e-01 -9.35153723e-01 -5.39771676e-01 1.64553508e-01
1.37909010e-01 2.98087418e-01 3.17213833e-01 7.75350273e-01
-5.15631020e-01 3.58182460e-01 -1.44499883e-01 -1.06402850e+00
-1.30850542e+00 5.48611224e-01 2.62942016e-01 -1.78122059e-01
-4.28914160e-01 9.38499391e-01 5.79554856e-01 -1.79765448e-01
1.47637635e-01 -7.97573626e-01 -5.27077973e-01 4.00625557e-01
5.12465894e-01 -4.52344641e-02 6.44119224e-03 -6.25796318e-01
-1.89718962e-01 9.53755677e-01 2.17553690e-01 1.36432648e-01
1.46642315e+00 -1.91113472e-01 -3.86783257e-02 4.23899561e-01
7.35801637e-01 -2.80923963e-01 -1.30803287e+00 -2.28910580e-01
8.77879262e-02 -4.75542217e-01 1.19294286e-01 -1.12942147e+00
-1.31634319e+00 1.40388036e+00 5.82893968e-01 3.98478419e-01
1.25482798e+00 -4.58184257e-02 4.29907948e-01 3.72803122e-01
1.97803527e-01 -9.02859747e-01 5.41582331e-02 6.84402168e-01
9.78828013e-01 -1.70151186e+00 2.44787157e-01 -4.16517675e-01
-6.90142930e-01 1.28885865e+00 5.25403440e-01 -1.62890941e-01
5.05017936e-01 1.15048416e-01 2.78894939e-02 -3.32896769e-01
-5.43826461e-01 -1.01623774e+00 4.41471964e-01 5.95247090e-01
5.00841737e-01 8.32379833e-02 1.82783410e-01 3.72024149e-01
2.17495322e-01 -2.70574510e-01 3.41017812e-01 1.13544416e+00
-4.31521028e-01 -1.10458601e+00 -4.16778803e-01 2.22763479e-01
-2.80761600e-01 -3.42807502e-01 -7.86685944e-01 1.06373572e+00
5.78912258e-01 1.05708289e+00 6.73436746e-02 -3.50315452e-01
3.77714694e-01 1.21774755e-01 5.81751823e-01 -7.97707200e-01
-9.55037475e-01 -3.83595601e-02 7.65428618e-02 -5.49133122e-01
-8.57829571e-01 -5.50466716e-01 -1.00188887e+00 -2.80595385e-02
-2.81627685e-01 -3.58001798e-01 7.83916831e-01 1.17723858e+00
1.94217581e-02 7.31026471e-01 2.97014683e-01 -1.10406208e+00
1.96194053e-01 -6.02049291e-01 -4.21504766e-01 3.40101272e-01
5.47797918e-01 -6.04728580e-01 -2.04772487e-01 3.00732940e-01] | [9.515789031982422, -0.40352290868759155] |
13a165d7-42ba-486a-bf06-43b87bc3dc25 | exploiting-long-term-dependencies-for | 2112.09828 | null | https://arxiv.org/abs/2112.09828v2 | https://arxiv.org/pdf/2112.09828v2.pdf | Exploiting Long-Term Dependencies for Generating Dynamic Scene Graphs | Dynamic scene graph generation from a video is challenging due to the temporal dynamics of the scene and the inherent temporal fluctuations of predictions. We hypothesize that capturing long-term temporal dependencies is the key to effective generation of dynamic scene graphs. We propose to learn the long-term dependencies in a video by capturing the object-level consistency and inter-object relationship dynamics over object-level long-term tracklets using transformers. Experimental results demonstrate that our Dynamic Scene Graph Detection Transformer (DSG-DETR) outperforms state-of-the-art methods by a significant margin on the benchmark dataset Action Genome. Our ablation studies validate the effectiveness of each component of the proposed approach. The source code is available at https://github.com/Shengyu-Feng/DSG-DETR. | ['Somdeb Majumdar', 'Marcel Nassar', 'Hesham Mostafa', 'Subarna Tripathi', 'Shengyu Feng'] | 2021-12-18 | null | null | null | null | ['scene-graph-generation'] | ['computer-vision'] | [ 1.64156109e-01 -1.77335575e-01 -2.21627772e-01 -2.01721862e-01
-3.50076586e-01 -4.44281965e-01 5.55746257e-01 -2.02424049e-01
1.51824504e-01 3.86039525e-01 4.02851641e-01 4.66167256e-02
-4.00984287e-01 -5.72489560e-01 -1.09965348e+00 -5.06620347e-01
-4.16462570e-01 2.96709090e-01 7.42811143e-01 -3.31659839e-02
-3.94108146e-03 2.26528332e-01 -1.33024132e+00 6.28445268e-01
6.79614246e-01 9.99000311e-01 5.21621525e-01 9.76489723e-01
1.43344551e-01 1.50425744e+00 -7.48941451e-02 -2.08443284e-01
2.42200628e-01 -5.91241717e-01 -5.36060512e-01 4.14760232e-01
7.88751662e-01 -4.03156638e-01 -9.68986392e-01 9.51786995e-01
2.24738181e-01 5.21762855e-02 1.37580723e-01 -1.41912401e+00
-4.16208029e-01 5.20997167e-01 -5.30897558e-01 5.98066449e-01
4.30995107e-01 5.14188647e-01 1.02076471e+00 -7.66672254e-01
9.49734211e-01 1.16218102e+00 5.89603543e-01 3.93737733e-01
-1.15966332e+00 -6.64295733e-01 7.86581635e-01 6.34287775e-01
-1.24943614e+00 -4.96005893e-01 8.49043369e-01 -7.32068360e-01
1.04992449e+00 -3.41708976e-04 9.92958963e-01 1.19386685e+00
5.58617234e-01 9.01674449e-01 8.34255636e-01 5.43053895e-02
-5.60987480e-02 -6.49707556e-01 9.84666348e-02 1.09496880e+00
2.32522622e-01 1.94509998e-01 -1.00649619e+00 3.10229715e-02
7.86309659e-01 2.74172183e-02 -2.24578768e-01 -4.75670904e-01
-1.09634447e+00 4.03218895e-01 4.65390354e-01 9.13408175e-02
-4.62509483e-01 6.98067725e-01 3.81398082e-01 1.55083075e-01
4.32908326e-01 6.06715642e-02 -3.60374629e-01 -4.32715893e-01
-6.54746890e-01 2.48198211e-01 6.44813776e-01 1.29291761e+00
2.91857272e-01 1.46517664e-01 -5.04006505e-01 2.98850209e-01
8.24422240e-02 3.34512830e-01 1.50509313e-01 -9.72332239e-01
4.39753503e-01 5.49995780e-01 -1.84711054e-01 -1.17040873e+00
-2.25799486e-01 -4.89221126e-01 -6.39267445e-01 -2.48827919e-01
3.38868797e-01 1.14227310e-02 -1.00851440e+00 1.79337966e+00
4.45678294e-01 9.48309779e-01 -4.15577799e-01 6.91543818e-01
7.91949630e-01 7.02883065e-01 1.11693271e-01 -3.38708460e-01
9.71436858e-01 -1.12992048e+00 -7.65250027e-01 -4.61795598e-01
3.42677653e-01 -4.55362260e-01 1.00783157e+00 5.87102287e-02
-8.95553231e-01 -5.94546497e-01 -7.49382257e-01 9.90436524e-02
5.99549562e-02 -9.99133885e-02 6.74707830e-01 -1.03845254e-01
-8.82634521e-01 4.20515418e-01 -1.21994138e+00 -4.88977194e-01
6.18221045e-01 8.83835331e-02 -2.21296296e-01 -3.19281965e-01
-8.01978052e-01 3.67945760e-01 1.76968917e-01 1.76661357e-01
-1.32446957e+00 -7.41140187e-01 -8.49887609e-01 -5.93368523e-02
7.67678142e-01 -8.38027835e-01 1.12143373e+00 -8.46637428e-01
-1.05630493e+00 4.62162286e-01 -3.28461111e-01 -4.38411027e-01
6.98549151e-01 -4.25987005e-01 -5.96986338e-02 3.13124776e-01
-3.12738195e-02 3.31574708e-01 7.07634211e-01 -9.91907179e-01
-6.11778140e-01 -3.19026530e-01 1.68739751e-01 3.09808105e-01
-1.80318534e-01 -1.80944309e-01 -9.84298229e-01 -7.40054846e-01
3.69425602e-02 -1.28237152e+00 -1.72223419e-01 -7.09581226e-02
-5.12710273e-01 -7.32496679e-02 1.05807650e+00 -6.05503678e-01
1.31291986e+00 -2.03904891e+00 1.79604352e-01 -1.51639909e-01
3.07010621e-01 2.48586349e-02 -3.17773342e-01 5.14464200e-01
-6.06643148e-02 -2.47009024e-01 1.61563277e-01 -1.51441723e-01
-2.80421704e-01 8.38577896e-02 -3.10125828e-01 4.37727660e-01
3.06927077e-02 1.21046782e+00 -1.13813174e+00 -4.16822314e-01
3.96936119e-01 4.19837147e-01 -4.72221434e-01 2.20595032e-01
-6.63557231e-01 5.29292643e-01 -7.08333910e-01 5.54400623e-01
3.39902163e-01 -7.47032285e-01 4.05865252e-01 -4.73942161e-01
1.02719255e-01 2.44878396e-01 -8.72453988e-01 1.83558905e+00
-6.03034627e-03 7.61972666e-01 -3.72635514e-01 -8.12115014e-01
2.96963096e-01 2.95445435e-02 8.30536664e-01 -7.28436232e-01
-6.69050068e-02 -2.36301333e-01 3.20242941e-02 -5.43006539e-01
2.72100091e-01 1.78177059e-01 -2.40516360e-03 1.76018924e-02
1.10959470e-01 1.06487960e-01 5.13694704e-01 7.78072953e-01
1.65362203e+00 4.15133923e-01 3.14827971e-02 -1.86664075e-01
5.22605069e-02 -2.91663334e-02 9.87274170e-01 8.73534977e-01
-2.99407572e-01 6.12087727e-01 6.58392608e-01 -4.05876160e-01
-8.31703722e-01 -9.20392692e-01 3.01762253e-01 8.31824839e-01
3.94897640e-01 -8.46848369e-01 -4.99712259e-01 -6.99842155e-01
4.58487049e-02 6.23783350e-01 -6.71037674e-01 -2.72980303e-01
-7.03202248e-01 -4.57919776e-01 7.13757128e-02 5.56187570e-01
4.14984673e-01 -9.66374457e-01 -5.93579471e-01 3.12650442e-01
-4.45619375e-01 -1.78881133e+00 -7.63927698e-01 -2.89737165e-01
-8.26695442e-01 -1.33740771e+00 -2.21496105e-01 -3.89711648e-01
5.75425684e-01 4.97542590e-01 1.26702940e+00 1.06832258e-01
-4.90044922e-01 7.08485305e-01 -3.82077843e-01 -9.56623480e-02
-2.28504732e-01 -2.57931858e-01 -1.24558315e-01 8.40565786e-02
-2.17679888e-02 -7.69917250e-01 -6.39457583e-01 2.70186424e-01
-7.28469729e-01 6.31103158e-01 3.33994538e-01 6.77187741e-01
8.49036574e-01 3.64972591e-01 8.63943547e-02 -1.02299476e+00
1.10440761e-01 -2.66426474e-01 -8.80828977e-01 3.65549952e-01
-4.86871332e-01 -1.07754119e-01 3.87974322e-01 -4.45051074e-01
-1.10066271e+00 2.99623847e-01 2.68517643e-01 -8.39787543e-01
1.36856169e-01 4.72804099e-01 -8.29857215e-02 4.34505045e-02
3.29445243e-01 3.27479392e-01 -3.16485703e-01 -7.15973303e-02
3.33905280e-01 -2.34800115e-01 4.86287475e-01 -4.78316039e-01
8.52640390e-01 5.99586189e-01 1.19637273e-01 -6.36651099e-01
-1.10989022e+00 -4.26738292e-01 -6.18814945e-01 -8.22839797e-01
9.40797806e-01 -1.21432626e+00 -4.35918123e-01 8.03321600e-01
-1.00297165e+00 -1.01683736e+00 -3.91871244e-01 3.39789569e-01
-7.29879797e-01 1.55109659e-01 -8.40466440e-01 -4.59769070e-01
-7.47502446e-02 -8.02007437e-01 1.01592350e+00 5.58630489e-02
1.13585584e-01 -9.47282851e-01 1.03966303e-01 3.44801635e-01
1.87964365e-02 4.72975224e-01 5.08564711e-01 -3.65051851e-02
-1.28030562e+00 -7.36607239e-02 -1.03771567e-01 9.04195011e-02
2.27243736e-01 3.05854797e-01 -6.68276131e-01 -3.08761537e-01
-2.47225508e-01 -1.60729811e-01 1.02433586e+00 7.87165821e-01
1.20479834e+00 -3.38832557e-01 -3.10283124e-01 7.89607346e-01
1.48209357e+00 -7.02288523e-02 6.62447572e-01 1.08449131e-01
1.30002546e+00 3.48024070e-01 9.02940452e-01 5.74713647e-01
6.22912288e-01 7.84096718e-01 4.67975438e-01 2.22492769e-01
-6.65083230e-01 -5.74922979e-01 6.89762115e-01 7.90270150e-01
-6.04956187e-02 -5.70202589e-01 -9.87047851e-01 6.96292877e-01
-2.43165255e+00 -1.28001153e+00 -3.24245989e-01 1.78748429e+00
4.06764060e-01 4.07328129e-01 -2.58192071e-03 -4.48282987e-01
7.17132807e-01 4.84118253e-01 -6.81332111e-01 3.39622438e-01
-1.73897073e-01 -3.85441154e-01 4.89687443e-01 5.08254945e-01
-1.01844740e+00 1.23870623e+00 5.59025574e+00 6.84229195e-01
-9.66809690e-01 1.22247659e-01 7.00780571e-01 -3.83896112e-01
-6.77800179e-02 1.40905514e-01 -8.05496871e-01 4.47541714e-01
8.21746230e-01 -5.24398923e-01 4.07421082e-01 6.65623128e-01
5.24640441e-01 -1.15186051e-01 -1.09262085e+00 8.93328726e-01
1.41610608e-01 -1.58285904e+00 4.01806384e-02 7.28883669e-02
9.23887432e-01 4.00901049e-01 -5.40896170e-02 2.97529660e-02
5.18728614e-01 -5.26127934e-01 8.93713534e-01 7.15839863e-01
6.16288781e-01 -3.38142902e-01 1.96890652e-01 2.93223768e-01
-1.72302890e+00 -1.37463734e-01 -8.90337080e-02 5.21869659e-02
4.38994467e-01 8.84729683e-01 -5.66689551e-01 4.94422257e-01
8.19839001e-01 1.37830603e+00 -8.05893421e-01 8.79006803e-01
-3.49567264e-01 9.16680396e-01 -2.98846304e-01 4.32299584e-01
7.25690126e-02 -2.25893483e-01 8.37829947e-01 1.16224229e+00
2.49870583e-01 2.28587061e-01 4.45759267e-01 6.27168238e-01
3.63026261e-02 -4.01807785e-01 -7.84166813e-01 -3.35541666e-01
1.72502145e-01 1.06999028e+00 -7.45716274e-01 -5.18098593e-01
-4.64226276e-01 9.49349344e-01 3.79523277e-01 4.43276256e-01
-1.21048963e+00 4.00971562e-01 6.12661779e-01 3.79384518e-01
5.47498524e-01 -5.05993366e-01 3.92070413e-02 -1.37369955e+00
3.83151919e-01 -8.38749886e-01 8.15211892e-01 -8.75572562e-01
-1.26727962e+00 4.66762424e-01 2.16488853e-01 -1.24093914e+00
-1.91047415e-01 -1.51546881e-01 -4.91139233e-01 1.45705566e-01
-1.28827798e+00 -1.36151004e+00 -8.53687942e-01 8.18478882e-01
9.01281416e-01 1.30014166e-01 2.69190401e-01 1.82733640e-01
-6.37964249e-01 1.40243575e-01 -2.10274637e-01 7.64412731e-02
3.30487400e-01 -9.82556522e-01 6.60926461e-01 1.39760518e+00
4.28122193e-01 1.11129418e-01 7.82597303e-01 -1.10561097e+00
-1.69963884e+00 -1.47732258e+00 5.31936944e-01 -6.26327217e-01
9.32222903e-01 -6.07110739e-01 -7.86969602e-01 9.77528274e-01
4.56214733e-02 5.50087035e-01 1.25758559e-01 -8.53895992e-02
-2.32369751e-01 -2.59049833e-01 -4.62107062e-01 5.56280673e-01
1.81183577e+00 -5.28269351e-01 5.83284982e-02 6.62503779e-01
8.15527320e-01 -5.83737373e-01 -5.79716265e-01 5.78111470e-01
5.71493089e-01 -1.11332214e+00 7.71830678e-01 -5.97120583e-01
5.19966662e-01 -3.49178344e-01 -3.42003889e-02 -9.75536644e-01
-6.94463372e-01 -8.12428534e-01 -6.14620447e-01 9.50552404e-01
1.13558918e-01 -3.19964617e-01 9.09304798e-01 5.51909924e-01
-1.79903984e-01 -6.24025047e-01 -6.45018101e-01 -1.04493403e+00
-5.85732996e-01 -5.92583776e-01 1.52202293e-01 8.46997440e-01
-2.86582857e-01 3.89315605e-01 -6.39628232e-01 3.51686358e-01
8.13938856e-01 4.38202083e-01 9.88981366e-01 -6.09980404e-01
-5.88338554e-01 -1.14370147e-02 -6.43472254e-01 -1.25607479e+00
1.46669239e-01 -5.61266720e-01 2.00201854e-01 -1.79422092e+00
6.22638881e-01 -1.79598600e-01 -3.75189602e-01 2.69604534e-01
-4.13539439e-01 2.77065616e-02 4.96941984e-01 8.37794095e-02
-1.17964292e+00 7.66561568e-01 1.41286397e+00 -5.68460040e-02
-1.52340233e-01 -1.31588399e-01 -3.20819110e-01 7.25831330e-01
6.75114870e-01 -6.05801344e-01 -8.34452927e-01 -4.81963962e-01
2.12871745e-01 6.80900738e-02 4.63892072e-01 -1.16579533e+00
4.13257718e-01 -4.86439288e-01 1.00222155e-01 -5.73558688e-01
4.06871825e-01 -6.25859141e-01 6.33711219e-01 4.78341371e-01
-2.98818469e-01 1.80131242e-01 2.57814765e-01 1.06279492e+00
-7.38848373e-02 6.01457953e-01 5.91674507e-01 -1.46564171e-01
-1.07032430e+00 9.50824618e-01 -1.21831909e-01 2.23727703e-01
1.19974005e+00 -1.08828649e-01 -5.84037542e-01 -6.53428316e-01
-6.41727388e-01 4.04084206e-01 5.21675527e-01 6.92958295e-01
6.93112969e-01 -1.17538571e+00 -7.06761181e-01 -1.74656808e-01
2.12780401e-01 -7.96438083e-02 4.75441635e-01 1.04319263e+00
-3.45677167e-01 2.25629151e-01 -9.20472592e-02 -8.02227914e-01
-1.63691092e+00 5.17716289e-01 3.77403289e-01 -4.09336060e-01
-1.05462253e+00 1.09746003e+00 8.66515338e-01 2.30960295e-01
1.10324644e-01 -3.25911373e-01 1.77154332e-01 -3.92769456e-01
3.39347094e-01 1.89747453e-01 -2.91253090e-01 -7.55089700e-01
-4.11330789e-01 4.75086242e-01 -2.80751288e-02 1.97024718e-01
1.39909565e+00 -1.71759948e-01 9.01370049e-02 5.55997372e-01
8.90729070e-01 -2.18310580e-01 -1.80306876e+00 -1.75165012e-01
-1.94552317e-01 -8.47576261e-01 2.66455114e-02 -5.14953434e-01
-1.26612997e+00 4.65016156e-01 4.83896941e-01 2.25798655e-02
1.21681249e+00 2.57170260e-01 8.69363070e-01 2.23461881e-01
5.17450392e-01 -8.34832072e-01 5.98601460e-01 5.02272546e-01
9.76970613e-01 -1.07850897e+00 3.25236738e-01 -9.25291061e-01
-7.09082544e-01 7.17803299e-01 8.98997426e-01 -2.10816234e-01
7.91500390e-01 4.45996761e-01 -2.11137146e-01 -5.79877257e-01
-1.48432481e+00 -2.95114756e-01 4.41474587e-01 3.49124253e-01
1.79303020e-01 -1.06713265e-01 7.32652843e-02 -1.97340604e-02
1.01586670e-01 1.40827939e-01 5.02448976e-01 9.98624027e-01
-4.65851948e-02 -8.38022351e-01 2.14166105e-01 5.99382699e-01
-1.81656316e-01 -7.59526044e-02 -4.18305963e-01 5.16231775e-01
-5.79739548e-02 8.40452373e-01 1.05503658e-02 -5.32173216e-01
3.74019891e-01 -3.04053158e-01 7.04322338e-01 -5.42207599e-01
-1.14823677e-01 3.21726143e-01 4.05194998e-01 -1.37254155e+00
-5.78950942e-01 -9.24010932e-01 -1.28380835e+00 -1.87540740e-01
-1.64078757e-01 -3.06864887e-01 1.84229165e-01 6.39403343e-01
8.15113127e-01 8.35955977e-01 4.07708883e-01 -5.63630939e-01
-5.79722375e-02 -7.60968924e-01 -3.85072500e-01 7.41663396e-01
3.05596441e-01 -7.34871268e-01 -2.38450423e-01 6.28632605e-01] | [9.20423412322998, 0.2541642487049103] |
e3d17559-9b61-4b3a-9140-860a242e6b20 | iterated-support-vector-machines-for-distance | 1502.00363 | null | http://arxiv.org/abs/1502.00363v1 | http://arxiv.org/pdf/1502.00363v1.pdf | Iterated Support Vector Machines for Distance Metric Learning | Distance metric learning aims to learn from the given training data a valid
distance metric, with which the similarity between data samples can be more
effectively evaluated for classification. Metric learning is often formulated
as a convex or nonconvex optimization problem, while many existing metric
learning algorithms become inefficient for large scale problems. In this paper,
we formulate metric learning as a kernel classification problem, and solve it
by iterated training of support vector machines (SVM). The new formulation is
easy to implement, efficient in training, and tractable for large-scale
problems. Two novel metric learning models, namely Positive-semidefinite
Constrained Metric Learning (PCML) and Nonnegative-coefficient Constrained
Metric Learning (NCML), are developed. Both PCML and NCML can guarantee the
global optimality of their solutions. Experimental results on UCI dataset
classification, handwritten digit recognition, face verification and person
re-identification demonstrate that the proposed metric learning methods achieve
higher classification accuracy than state-of-the-art methods and they are
significantly more efficient in training. | ['Liang Lin', 'Yuchi Huang', 'David Zhang', 'Faqiang Wang', 'Wangmeng Zuo', 'Lei Zhang', 'Deyu Meng'] | 2015-02-02 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [ 3.49490680e-02 -3.76918823e-01 -4.32333201e-01 -7.36134231e-01
-9.00513113e-01 -3.76918226e-01 1.96305722e-01 -1.76594719e-01
-3.63749236e-01 9.32766080e-01 -1.36733949e-01 -2.34050229e-01
-7.72047639e-01 -3.58969778e-01 -3.33504915e-01 -7.00864613e-01
-2.07512796e-01 6.00239336e-01 -2.56493211e-01 3.65890563e-01
3.73108119e-01 4.09991026e-01 -1.15226829e+00 -2.17424095e-01
1.26719737e+00 1.16748774e+00 -5.68892434e-02 3.69213462e-01
2.58674681e-01 8.10218573e-01 -2.90442497e-01 -3.12974900e-01
3.41256350e-01 -3.91035795e-01 -5.71814358e-01 3.62539262e-01
4.91265386e-01 -2.31892630e-01 -5.35427451e-01 1.19855702e+00
5.30827582e-01 1.67538390e-01 8.58040154e-01 -1.86503232e+00
-1.11286509e+00 1.49855554e-01 -6.43042266e-01 3.16378139e-02
4.41528976e-01 -4.59306061e-01 1.00165427e+00 -1.34467590e+00
2.43164420e-01 1.16939068e+00 7.33908474e-01 4.66189474e-01
-1.05155754e+00 -7.11901128e-01 -4.61902805e-02 7.41678655e-01
-1.78208876e+00 -1.35259792e-01 8.55495274e-01 -6.39407396e-01
3.59669507e-01 5.12676001e-01 3.80152076e-01 5.37974179e-01
-9.61525589e-02 9.09821451e-01 1.30791116e+00 -2.09274352e-01
3.81990075e-01 2.08989412e-01 2.00132057e-01 1.04712522e+00
3.06501031e-01 2.26796716e-01 -4.54203784e-01 -4.77013379e-01
4.96628851e-01 5.52738942e-02 -5.38315535e-01 -6.75592542e-01
-1.32931638e+00 1.01555932e+00 7.94043764e-02 5.91639848e-03
2.28940882e-02 -2.18128026e-01 2.60521650e-01 5.11696517e-01
5.83345652e-01 2.20294058e-01 -2.29469150e-01 -1.55251920e-01
-1.02887142e+00 1.66342467e-01 1.02819288e+00 8.14789116e-01
7.53061414e-01 -6.22635260e-02 -2.06920400e-01 1.09615779e+00
4.25698847e-01 8.60021591e-01 5.61839640e-01 -6.36172175e-01
5.40934265e-01 7.04527318e-01 2.53704667e-01 -1.31743658e+00
-3.77955616e-01 1.64538913e-03 -8.56359541e-01 1.29298925e-01
5.01895666e-01 -2.28479609e-01 -3.42233539e-01 1.48672843e+00
4.31783497e-01 7.11759448e-01 -1.44461066e-01 1.22496223e+00
3.43474984e-01 6.45765483e-01 -3.44982862e-01 -1.81709692e-01
7.62078881e-01 -8.03621888e-01 -6.81398869e-01 1.03050247e-02
9.75724459e-01 -7.56326377e-01 8.66843462e-01 4.61924851e-01
-5.69460213e-01 -2.45137691e-01 -1.49328256e+00 1.76093295e-01
-2.23267063e-01 5.11937022e-01 7.34357238e-01 8.87763977e-01
-7.90588796e-01 7.92511582e-01 -6.08910918e-01 -5.39097525e-02
4.34418619e-01 5.68120658e-01 -5.61621189e-01 -5.58523953e-01
-9.18669641e-01 9.18173909e-01 2.04289526e-01 1.85653389e-01
-6.66514575e-01 -9.16959465e-01 -1.09940934e+00 -4.17731941e-01
6.29565716e-02 -6.10687221e-06 8.47992539e-01 -5.11976242e-01
-1.42201042e+00 1.12165761e+00 -2.60054674e-02 -6.96474016e-02
4.98626351e-01 2.40098014e-01 -5.04395962e-01 -2.09121153e-01
5.76150380e-02 2.14808658e-02 8.10365915e-01 -7.52690315e-01
-6.55337155e-01 -6.82602048e-01 -2.52860010e-01 6.84986636e-02
-7.75655091e-01 2.72125900e-01 -5.50904684e-02 -7.19587743e-01
1.19605951e-01 -9.34489131e-01 -8.85001719e-02 5.24183869e-01
-3.26296091e-01 -5.12291074e-01 1.30263662e+00 -9.88378704e-01
1.32306528e+00 -2.04449081e+00 4.65371132e-01 5.03347099e-01
-5.68815805e-02 3.27130944e-01 -2.86661178e-01 -6.03202321e-02
-1.33068949e-01 -3.54835838e-01 -5.94295740e-01 -3.49811524e-01
3.20023239e-01 2.86912352e-01 6.10729754e-02 1.18039668e+00
5.00606596e-02 7.91095316e-01 -9.26868618e-01 -5.41923583e-01
2.61992395e-01 2.91930854e-01 -2.72423416e-01 2.56507456e-01
3.71054381e-01 1.87715247e-01 -3.83994758e-01 1.11362016e+00
8.30988407e-01 -9.58957523e-02 -1.76935736e-02 -2.28850812e-01
2.23622367e-01 -4.17485356e-01 -1.67293942e+00 1.72347116e+00
-2.69222558e-01 7.19263732e-01 6.53353781e-02 -1.57579112e+00
1.17758179e+00 5.30135967e-02 8.38511050e-01 -5.78307748e-01
2.11420469e-02 4.36316073e-01 -2.47370631e-01 -3.43437076e-01
-7.31259882e-02 6.75558448e-02 6.99066967e-02 7.20313489e-01
-1.80116564e-01 1.62743121e-01 3.62626970e-01 -9.49378312e-02
6.96500957e-01 -3.92935544e-01 1.33730888e-01 -5.71306705e-01
1.09900022e+00 -4.16492909e-01 8.75555098e-01 1.49909869e-01
-4.30969119e-01 4.11483616e-01 4.22784090e-01 -3.04521173e-01
-9.18046534e-01 -1.08162415e+00 -5.78731656e-01 8.40312123e-01
2.51223713e-01 -1.03382275e-01 -7.68158078e-01 -8.15701008e-01
5.35196424e-01 2.00704709e-01 -5.42517185e-01 -5.07925630e-01
-6.21601760e-01 -9.77487803e-01 3.24072689e-01 5.59592843e-01
5.71196079e-01 -4.41886842e-01 2.40904853e-01 2.10837717e-03
6.70564994e-02 -1.03619301e+00 -1.16959083e+00 -3.53040665e-01
-6.80454791e-01 -1.64909613e+00 -1.01263034e+00 -1.23015440e+00
1.03356624e+00 2.64757216e-01 6.20129764e-01 -2.13501647e-01
-7.14183927e-01 5.09713173e-01 -2.23089278e-01 -1.76489428e-01
5.60724176e-03 -3.32256109e-01 6.17858768e-01 5.42006433e-01
5.85322499e-01 -1.43329680e-01 -4.88373756e-01 8.62836063e-01
-5.54691494e-01 -5.76456726e-01 2.76525795e-01 1.15926778e+00
7.16407835e-01 -2.63965912e-02 7.57639110e-01 -5.12760162e-01
6.83829427e-01 -3.90165806e-01 -7.27524638e-01 6.67082548e-01
-1.05398381e+00 7.18662143e-02 4.69538629e-01 -6.27539992e-01
-5.41487157e-01 -7.60067776e-02 5.19924104e-01 -4.38608557e-01
5.83298445e-01 5.49520969e-01 -3.99394870e-01 -8.27207744e-01
6.53785825e-01 2.80977577e-01 2.47163638e-01 -4.34264779e-01
2.70951748e-01 1.04899871e+00 3.75937909e-01 -7.55076289e-01
1.07686973e+00 4.11411017e-01 1.49184942e-01 -6.13837481e-01
-9.58849132e-01 -5.92967033e-01 -7.62453556e-01 -2.39615649e-01
6.04066491e-01 -5.92362642e-01 -5.40738404e-01 6.62662089e-01
-6.43072605e-01 -2.39240572e-01 -9.27117616e-02 7.70503819e-01
-6.11563385e-01 7.07920253e-01 -3.57895046e-01 -6.89553738e-01
-3.31535548e-01 -8.79094243e-01 6.77135885e-01 3.02171648e-01
1.17498048e-01 -1.51646411e+00 1.35747403e-01 3.58245254e-01
1.70060053e-01 4.41585422e-01 6.87209666e-01 -5.36565006e-01
-8.02542120e-02 -5.85128069e-01 -6.39217317e-01 7.86215603e-01
5.52244008e-01 -1.92057639e-01 -5.51522791e-01 -8.21377933e-01
-5.48827015e-02 -4.70862567e-01 4.26607341e-01 1.61466926e-01
1.47319448e+00 -5.20607948e-01 -3.75731945e-01 9.47965741e-01
1.16574073e+00 8.77823681e-02 2.07472131e-01 7.85338581e-02
8.32995415e-01 4.09025908e-01 1.08317494e+00 9.00708079e-01
4.02344227e-01 7.32960522e-01 1.35710225e-01 -5.40772676e-02
2.61546344e-01 4.68631200e-02 2.76846558e-01 7.09331274e-01
2.98430532e-01 5.04265845e-01 -7.85038352e-01 3.53819907e-01
-2.20637536e+00 -8.08620811e-01 1.17321938e-01 2.50914836e+00
9.07607615e-01 -5.28304160e-01 2.05728963e-01 6.23264074e-01
9.30542529e-01 -1.51754528e-01 -8.69524181e-01 -3.50345671e-01
-2.45744064e-01 1.23267367e-01 4.47245538e-01 5.91133177e-01
-1.38522279e+00 6.23867810e-01 5.96005106e+00 7.99063027e-01
-1.03473854e+00 1.40471444e-01 6.08949780e-01 -1.06328771e-01
1.01177268e-01 -4.26055461e-01 -5.75998485e-01 7.28452504e-01
6.24195457e-01 -7.42485344e-01 6.78995550e-01 1.04306161e+00
2.26899937e-01 2.73129165e-01 -1.49242330e+00 1.84937418e+00
4.92622674e-01 -1.08544123e+00 -3.05980593e-01 -3.73867564e-02
9.87456977e-01 -3.85236174e-01 3.81984890e-01 3.11082363e-01
4.49203104e-02 -1.25195074e+00 3.29296589e-01 4.83714998e-01
1.04640484e+00 -1.04981875e+00 6.01615965e-01 1.72534704e-01
-1.21005273e+00 -2.94162691e-01 -8.90822113e-01 5.10298163e-02
-1.11349493e-01 7.51715481e-01 -3.22677821e-01 2.87400067e-01
3.55983555e-01 1.33544493e+00 -5.81575572e-01 1.46829462e+00
-1.09840967e-01 3.48030448e-01 -1.26562297e-01 -1.75230742e-01
1.36114821e-01 -7.67658174e-01 4.16914940e-01 1.07145000e+00
3.69801730e-01 5.18152432e-04 5.10443628e-01 6.31985247e-01
-1.05199683e-02 4.78046209e-01 -3.83762240e-01 5.44661470e-02
5.08695662e-01 1.12388313e+00 -1.07245989e-01 1.28780037e-01
-5.31766713e-01 1.30044734e+00 5.29488921e-01 3.45272958e-01
-9.94790316e-01 -7.21682549e-01 1.13929212e+00 -3.97696942e-01
1.07992359e-01 -4.04918879e-01 -2.68793494e-01 -1.37954688e+00
5.87871492e-01 -7.38827884e-01 7.53030896e-01 -1.60537615e-01
-1.67122579e+00 1.90409310e-02 -3.15979362e-01 -1.37164140e+00
-1.05042428e-01 -1.18760359e+00 -4.76932168e-01 7.67774463e-01
-1.59950721e+00 -9.71740067e-01 -3.07259411e-01 1.09188163e+00
3.36941238e-03 -4.26098794e-01 7.36248493e-01 7.28138328e-01
-7.62827396e-01 1.15786123e+00 7.40178227e-01 4.01642174e-01
8.96013439e-01 -1.53192759e+00 -4.48006898e-01 6.61773086e-01
2.41675824e-01 1.45320997e-01 4.83170301e-02 -2.71767437e-01
-1.67390037e+00 -1.33692861e+00 8.91818464e-01 -2.80382216e-01
7.64102101e-01 -2.72591591e-01 -6.30803108e-01 2.57230043e-01
-6.16350472e-01 4.50649440e-01 1.25467706e+00 -2.22364049e-02
-7.58982241e-01 -4.89223003e-01 -1.54145825e+00 2.17979565e-01
9.23392355e-01 -7.32542634e-01 -1.81735396e-01 1.05607867e+00
1.76687777e-01 -1.91906512e-01 -1.11681068e+00 2.83152103e-01
4.53904629e-01 -4.93577689e-01 1.06701362e+00 -8.77835333e-01
-1.70048982e-01 -4.64316279e-01 -5.61074436e-01 -1.29731917e+00
-3.53023559e-01 -6.40844703e-01 -7.40988731e-01 1.11278892e+00
2.41304606e-01 -6.65447354e-01 9.21136975e-01 8.32550466e-01
1.50691539e-01 -1.10550737e+00 -1.43978083e+00 -1.41337383e+00
2.02644169e-01 -2.15762839e-01 3.82843286e-01 1.32841539e+00
4.37086433e-01 -2.45047659e-01 -6.85570061e-01 1.28498286e-01
1.26479864e+00 1.10901594e-01 4.06450003e-01 -1.32675612e+00
-1.25583142e-01 -4.42332625e-01 -1.16487038e+00 -7.16097474e-01
6.39419019e-01 -1.34943068e+00 -3.52278709e-01 -1.23255992e+00
2.77286708e-01 -6.41895354e-01 -4.08165395e-01 4.21792656e-01
-2.91077524e-01 4.01116967e-01 -1.16868831e-01 1.13812365e-01
-5.22539616e-01 7.95058787e-01 1.10295248e+00 -5.13384581e-01
1.24549763e-02 1.38029799e-01 -5.28736770e-01 4.08883780e-01
5.62651396e-01 -1.99989796e-01 -5.25196254e-01 -2.56436855e-01
-2.79008955e-01 -3.02976221e-01 1.63206041e-01 -1.04839993e+00
3.23126942e-01 -2.75714844e-01 3.64937723e-01 -2.46435747e-01
-3.57993005e-04 -7.43298709e-01 -4.27483350e-01 4.78081703e-01
-2.85851229e-02 -1.84239939e-01 -2.93643117e-01 5.79404950e-01
-2.58211643e-01 -1.78085357e-01 1.15770817e+00 5.55882573e-01
-6.13043547e-01 1.05599284e+00 1.92406774e-01 3.45732272e-01
1.50430560e+00 -3.59894097e-01 1.46141201e-01 -1.61288306e-01
-6.31956220e-01 4.80601251e-01 1.01585001e-01 5.52115381e-01
1.03361320e+00 -1.97745728e+00 -9.93262351e-01 5.06899953e-02
2.39249721e-01 -3.65972489e-01 -1.07101396e-01 7.70847797e-01
-3.46746266e-01 7.07652807e-01 8.20848253e-03 -5.60976028e-01
-1.24634027e+00 6.66368365e-01 4.81038302e-01 -7.54413456e-02
-2.03468040e-01 9.83281493e-01 -3.70359242e-01 -9.20728743e-01
5.79006910e-01 -1.75739214e-01 7.33344406e-02 2.04970082e-03
7.83410728e-01 9.31801498e-01 -5.44385090e-02 -5.87699831e-01
-5.39905965e-01 8.77748728e-01 -4.70479392e-02 1.51318878e-01
1.01486695e+00 4.11227755e-02 -3.27247202e-01 4.22144197e-02
1.80085897e+00 -6.95130706e-01 -1.23501909e+00 -6.86380327e-01
3.35808963e-01 -8.52666438e-01 5.80903962e-02 -4.78285730e-01
-1.14141059e+00 4.01353598e-01 7.33541250e-01 -3.58128101e-01
9.61875200e-01 -1.65659532e-01 7.83299983e-01 6.51542127e-01
5.99041641e-01 -1.35970008e+00 3.04247200e-01 3.35659832e-01
8.99383605e-01 -1.58629632e+00 2.02056289e-01 -4.82011676e-01
-2.46940091e-01 1.17183840e+00 6.80241883e-01 -2.61706769e-01
1.20507956e+00 5.15348762e-02 -1.97057188e-01 3.05048406e-01
-8.53392407e-02 2.01779738e-01 7.66686976e-01 8.98830235e-01
1.71416864e-01 2.36483663e-01 -4.55323517e-01 6.80821002e-01
-2.43960932e-01 -2.45960489e-01 1.75021306e-01 8.42948079e-01
-2.11270705e-01 -1.15222538e+00 -3.08847308e-01 5.03523529e-01
6.45141900e-02 2.66575456e-01 -4.31868702e-01 3.38347495e-01
-9.74051431e-02 1.14226353e+00 -7.05622733e-02 -4.13526714e-01
1.86113596e-01 -1.88889772e-01 6.93616092e-01 -4.46490139e-01
2.82161623e-01 -8.57713580e-01 -3.24893832e-01 -7.66620219e-01
-1.44026592e-01 -9.53648746e-01 -1.02131033e+00 -3.05813402e-01
-4.40836191e-01 1.82286590e-01 7.49684751e-01 1.07497156e+00
1.04334831e-01 -1.88049346e-01 1.37236273e+00 -3.73065621e-01
-1.22871506e+00 -8.06680322e-01 -1.04145336e+00 3.98560971e-01
8.98324624e-02 -8.44887733e-01 -5.64166307e-01 -1.10239491e-01] | [9.212203025817871, 3.2988345623016357] |
d590cec3-0ad6-40cf-a937-590dbb4671ab | understanding-generalized-label-smoothing-1 | null | null | https://openreview.net/forum?id=UQQgMRq58O | https://openreview.net/pdf?id=UQQgMRq58O | Understanding Generalized Label Smoothing when Learning with Noisy Labels | Label smoothing (LS) is an arising learning paradigm that uses the positively weighted average of both the hard training labels and uniformly distributed soft labels. It was shown that LS serves as a regularizer for training data with hard labels and therefore improves the generalization of the model. Later it was reported LS even helps with improving robustness when learning with noisy labels. However, we observe that the advantage of LS vanishes when we operate in a high label noise regime. Puzzled by the observation, we proceeded to discover that several proposed learning-with-noisy-labels solutions in the literature instead relate more closely to $\textit{negative label smoothing}$ (NLS), which defines as using a negative weight to combine the hard and soft labels! We show that NLS differs substantially from LS in their achieved model confidence. To differentiate the two cases, we will call LS the positive label smoothing (PLS), and this paper unifies PLS and NLS into $\textit{generalized label smoothing}$ (GLS). We provide understandings for the properties of GLS when learning with noisy labels. Among other established properties, we theoretically show NLS is considered more beneficial when the label noise rates are high. We provide extensive experimental results on multiple benchmarks to support our findings too. | ['Yang Liu', 'Gang Niu', 'Tongliang Liu', 'Hangyu Liu', 'Jiaheng Wei'] | 2021-09-29 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [ 6.10465229e-01 5.64356267e-01 -2.31646523e-01 -7.15889037e-01
-1.02152789e+00 -6.85696661e-01 3.83495063e-01 2.33953685e-01
-4.37753320e-01 8.13532650e-01 -1.09782152e-01 -2.92990476e-01
-3.40991199e-01 -3.81699085e-01 -6.51298761e-01 -1.17062390e+00
1.08572975e-01 9.00468901e-02 1.43320441e-01 1.03673324e-01
1.54841244e-01 1.69161707e-01 -1.50366497e+00 3.11695375e-02
6.93431139e-01 1.03228629e+00 5.24590760e-02 3.98726642e-01
-9.80759859e-02 7.31872916e-01 -4.80472326e-01 -5.17078161e-01
6.11332357e-01 -8.35371912e-02 -8.42069805e-01 1.00192159e-01
5.25382876e-01 4.14010644e-01 3.08042020e-01 1.39582992e+00
4.92259175e-01 2.57147253e-01 8.14198911e-01 -1.18256736e+00
-7.04245389e-01 7.93434560e-01 -7.30719805e-01 -1.34316400e-01
4.97837774e-02 -1.18237771e-01 1.13149118e+00 -8.60493004e-01
4.36491698e-01 1.24644494e+00 1.12680817e+00 6.51092231e-01
-1.33823037e+00 -7.19450951e-01 3.75936866e-01 -1.36077598e-01
-1.18393373e+00 -2.33118683e-01 6.93830788e-01 -5.56387186e-01
4.02987719e-01 4.56969798e-01 -2.16133580e-01 1.04150772e+00
-2.47763187e-01 8.78498793e-01 1.75484037e+00 -8.94116521e-01
4.70037222e-01 5.14972448e-01 9.05188084e-01 6.48147643e-01
2.35773966e-01 3.37184042e-01 -3.14527303e-01 -1.72752321e-01
1.36052638e-01 -2.39827886e-01 -2.12074682e-01 -2.68588513e-02
-7.58024514e-01 7.53876865e-01 1.75455287e-01 2.35973045e-01
8.44633952e-02 4.41346355e-02 3.81940305e-01 5.57442188e-01
7.70655692e-01 5.16828358e-01 -6.11088753e-01 1.50971621e-01
-7.72566259e-01 -2.69791365e-01 8.44031692e-01 9.25129950e-01
8.21095049e-01 4.25255895e-02 -1.75217450e-01 1.06092966e+00
3.07072371e-01 4.52468812e-01 2.40677744e-01 -1.07802796e+00
1.31822437e-01 2.96200871e-01 5.14677390e-02 -6.55885160e-01
-7.10076451e-01 -6.92807019e-01 -9.56547856e-01 3.48327428e-01
6.61586225e-01 -2.20848009e-01 -9.96267498e-01 2.19817019e+00
1.12025708e-01 3.02658528e-01 -9.22927931e-02 4.37400609e-01
6.12019062e-01 3.98602784e-01 5.72961092e-01 -5.98975897e-01
1.04448664e+00 -8.30417812e-01 -9.73949909e-01 -1.93438247e-01
9.94833887e-01 -7.47514129e-01 1.37473691e+00 6.13994837e-01
-9.38683689e-01 -6.68303013e-01 -8.65735769e-01 5.87855885e-03
-6.21878862e-01 7.25136548e-02 5.90285718e-01 1.11232829e+00
-1.18721521e+00 7.87495136e-01 -4.90049869e-01 -3.09919238e-01
3.64189893e-01 4.41738278e-01 -1.01051129e-01 4.08176370e-02
-1.09327900e+00 8.94263387e-01 2.41217017e-01 -3.09330542e-02
-4.71722990e-01 -5.05403817e-01 -7.04109251e-01 -1.58021897e-01
6.81999564e-01 -2.31419593e-01 1.37112343e+00 -9.42805409e-01
-1.49199653e+00 1.09243321e+00 -3.11877578e-01 -1.52616605e-01
4.42894489e-01 2.43822951e-02 -2.95963794e-01 -2.75081009e-01
-2.75435727e-02 4.04826730e-01 8.94799232e-01 -1.81523752e+00
-6.84506118e-01 -2.73399413e-01 -4.06566709e-02 -2.34101191e-01
-2.47058421e-01 3.01517397e-02 4.80777808e-02 -7.07212985e-01
3.83690089e-01 -1.04163492e+00 -1.23519488e-01 -3.55853379e-01
-5.60531914e-01 -7.63122022e-01 6.92819774e-01 -3.82991761e-01
1.32495236e+00 -2.31208730e+00 -2.44999558e-01 5.62810063e-01
1.68131068e-01 4.19467896e-01 -5.08670211e-02 2.06874222e-01
-5.27480602e-01 5.35675406e-01 -3.69618058e-01 -7.71015763e-01
4.56903845e-01 4.65297133e-01 -4.11784917e-01 4.70223099e-01
8.17048922e-02 6.83095455e-01 -8.93788815e-01 -5.32093942e-01
-8.81575942e-02 1.95784643e-01 -1.45716473e-01 1.22533455e-01
3.19765396e-02 4.22598362e-01 -7.70819858e-02 7.16813147e-01
7.87387908e-01 -4.48390484e-01 1.63265333e-01 -1.90472499e-01
1.20177053e-01 -9.87078324e-02 -1.47059667e+00 1.07067287e+00
-1.66461989e-01 2.10802749e-01 2.09224403e-01 -1.22010040e+00
8.24136257e-01 2.70377100e-01 2.88542241e-01 -3.08038592e-01
2.13968545e-01 3.68821889e-01 -3.75914931e-01 -5.17522097e-01
2.17493340e-01 -6.38762712e-01 -8.72578025e-02 5.56329072e-01
8.06425884e-02 -2.33870745e-02 1.79906890e-01 1.82197094e-01
8.94272864e-01 -2.22143084e-02 3.36404920e-01 -4.63819057e-01
5.03353417e-01 -6.71728373e-01 5.69312513e-01 1.20373631e+00
-4.37065512e-01 3.92897159e-01 5.73684275e-01 4.38322127e-02
-6.64560497e-01 -1.16651487e+00 -4.68608409e-01 1.77050292e+00
7.90970623e-02 -3.46443534e-01 -7.49194980e-01 -1.11061299e+00
-3.83355133e-02 7.56254315e-01 -7.05662191e-01 -1.97174013e-01
-3.67832124e-01 -1.12401009e+00 4.95575190e-01 4.67923522e-01
1.14993230e-01 -8.58997285e-01 2.21618921e-01 -1.40833035e-01
5.26369177e-02 -8.20310354e-01 -2.72646606e-01 8.44474792e-01
-4.93907362e-01 -9.77077961e-01 -4.57379460e-01 -9.32106078e-01
7.61912942e-01 6.56509846e-02 1.10670233e+00 1.76369503e-01
2.50490040e-01 4.88069594e-01 -4.70662385e-01 -4.92034435e-01
-5.86754024e-01 2.77265534e-02 3.00016552e-01 8.89210217e-03
3.89946610e-01 -4.97011602e-01 -5.31730130e-02 4.94213343e-01
-1.19427121e+00 -2.89299399e-01 4.57280815e-01 9.35732126e-01
5.96391976e-01 2.06290200e-01 1.04169333e+00 -1.53444374e+00
4.85438496e-01 -5.71085155e-01 -3.83846283e-01 4.91129100e-01
-1.04868317e+00 2.87441373e-01 5.08119762e-01 -5.74094236e-01
-1.09271014e+00 8.12147111e-02 -3.92411679e-01 1.03142500e-01
-3.71545255e-01 2.23545492e-01 2.64558904e-02 -3.29236180e-01
6.32170439e-01 -2.56888121e-01 -3.65994871e-01 -7.43000805e-01
3.95229250e-01 8.82342696e-01 3.20372730e-01 -8.21787179e-01
7.12892234e-01 4.63995963e-01 2.20353693e-01 -4.36721146e-01
-1.62214017e+00 -5.82305372e-01 -5.22174120e-01 -1.84932828e-01
6.03251696e-01 -4.66510653e-01 -9.50970948e-01 3.94069105e-01
-6.21948600e-01 -4.62509900e-01 -7.09078193e-01 1.81110188e-01
-6.67396188e-01 6.19894445e-01 -6.18953288e-01 -1.21910834e+00
7.54566416e-02 -1.06757486e+00 9.82587814e-01 2.22660795e-01
-1.52294666e-01 -1.25561237e+00 -1.25289619e-01 1.86479121e-01
2.37736732e-01 1.37275383e-01 9.30510998e-01 -9.91575301e-01
9.92894322e-02 -3.32058579e-01 -3.89283806e-01 9.47239339e-01
-9.94107276e-02 -2.59976596e-01 -1.43155205e+00 -3.95621359e-01
3.60887855e-01 -5.47409117e-01 1.01650524e+00 3.63597095e-01
1.23491383e+00 -5.71093597e-02 -2.12486938e-01 3.74709725e-01
1.43181586e+00 -1.05753049e-01 3.02744448e-01 1.49392992e-01
5.98205268e-01 7.43565917e-01 5.16653538e-01 2.53677994e-01
1.31692171e-01 4.37993377e-01 2.95956016e-01 -2.27134645e-01
-2.16783613e-01 -9.14658010e-02 3.90699178e-01 9.81724083e-01
5.96113876e-02 -1.75442085e-01 -6.59059227e-01 8.79274979e-02
-1.67672944e+00 -6.25060320e-01 -3.72065037e-01 2.13865709e+00
1.36570418e+00 2.29939476e-01 -8.50829184e-02 5.06278217e-01
8.30299377e-01 -1.86078139e-02 -3.24541569e-01 -2.86120892e-01
-3.74709576e-01 4.50202197e-01 7.48089552e-01 9.29681718e-01
-1.34169221e+00 7.32457459e-01 6.72670650e+00 1.33368325e+00
-7.71251559e-01 4.83165652e-01 7.64806807e-01 1.58116426e-02
-7.80162066e-02 -1.52719757e-02 -1.06010115e+00 5.29478014e-01
8.63148451e-01 3.42721075e-01 2.30695784e-01 8.41051936e-01
8.09909478e-02 -3.05339038e-01 -1.22706878e+00 8.03663790e-01
-5.95028475e-02 -5.96192241e-01 -4.44782495e-01 -1.78189337e-01
1.17277014e+00 -2.43177697e-01 3.01366806e-01 6.34379327e-01
7.57348716e-01 -9.16077375e-01 6.64901555e-01 5.82993150e-01
7.74339318e-01 -4.36103851e-01 9.68246520e-01 5.88873804e-01
-8.61845493e-01 -2.97101825e-01 -3.12913477e-01 -7.12427869e-02
1.76125571e-01 1.04122400e+00 -6.38484836e-01 4.88196582e-01
4.27932709e-01 5.98491967e-01 -8.39983523e-01 8.71809244e-01
-2.56305695e-01 1.04973686e+00 -4.81612593e-01 3.63283992e-01
2.09200084e-01 -1.59439757e-01 3.78715664e-01 1.45526814e+00
1.28145441e-01 5.80012649e-02 5.43905139e-01 5.50600588e-01
-4.12614606e-02 3.01396940e-02 -2.93941468e-01 3.39530885e-01
2.15648144e-01 1.17652488e+00 -8.89245391e-01 -5.22870958e-01
-2.58166581e-01 3.99497986e-01 3.54052901e-01 4.94114488e-01
-5.63071847e-01 -1.37870058e-01 5.53957634e-02 -1.58851981e-01
-1.94432542e-01 2.79320814e-02 -7.74930596e-01 -8.78532350e-01
-1.31856158e-01 -5.04482567e-01 4.79001224e-01 -5.27711511e-01
-1.75696671e+00 3.43184978e-01 -4.58263829e-02 -1.00917661e+00
1.85564294e-01 -6.41261518e-01 -7.30269402e-02 9.15937304e-01
-1.71789801e+00 -9.13341522e-01 -3.20486836e-02 3.72083247e-01
3.83166492e-01 2.47225583e-01 6.44937456e-01 2.51709253e-01
-5.87560356e-01 8.45256507e-01 3.69941682e-01 -2.49591172e-01
1.15284860e+00 -1.64729273e+00 -1.73343375e-01 5.91167986e-01
-1.10403702e-01 5.84068418e-01 7.91867614e-01 -6.02801263e-01
-5.17168403e-01 -1.06708360e+00 1.07008672e+00 -6.51176810e-01
5.78888953e-01 -2.04527035e-01 -1.06605899e+00 7.31027126e-01
1.04638720e-02 8.87320470e-03 8.89116526e-01 2.13217318e-01
-5.99615455e-01 2.44166050e-02 -1.33030033e+00 3.58721852e-01
1.10830486e+00 -3.52407724e-01 -5.67731321e-01 7.43981481e-01
6.81585550e-01 -4.44620140e-02 -7.88560808e-01 6.47257507e-01
1.56573460e-01 -1.02790892e+00 7.79909074e-01 -5.04592597e-01
-8.39776918e-02 -2.12622166e-01 -3.68545473e-01 -1.22165847e+00
-4.24135864e-01 -5.57738662e-01 1.04183674e-01 1.40900743e+00
3.61259371e-01 -8.54304433e-01 6.45195603e-01 7.90357947e-01
-2.54375219e-01 -6.70063972e-01 -8.79776061e-01 -9.82392073e-01
9.32279602e-02 -6.27699137e-01 7.00311065e-02 1.23132765e+00
1.38436407e-01 1.51700631e-01 -3.84001166e-01 1.28914699e-01
7.27708995e-01 -5.26404679e-01 1.58813223e-01 -1.58259702e+00
-2.71562457e-01 -4.53935057e-01 5.34788854e-02 -1.01601112e+00
5.57979882e-01 -1.19996250e+00 2.86038697e-01 -9.44582760e-01
4.43356000e-02 -8.99973333e-01 -5.77396572e-01 7.21997380e-01
-4.56113338e-01 5.43591797e-01 1.97080374e-01 1.43099174e-01
-8.12938511e-01 3.88602614e-02 1.05028486e+00 9.36177894e-02
-2.27497295e-02 4.16522264e-01 -8.38238835e-01 1.13865650e+00
7.49936223e-01 -6.57870829e-01 -2.21727699e-01 -6.64637017e-04
5.79453528e-01 -4.77287292e-01 7.38850906e-02 -7.68044770e-01
9.21212584e-02 -1.35667488e-01 -1.14378527e-01 -2.03986406e-01
1.50491029e-01 -1.02612162e+00 -1.74945876e-01 1.58056006e-01
-9.97299731e-01 -6.02291584e-01 -2.00384527e-01 7.31476426e-01
1.20863579e-01 -8.91504526e-01 1.22575462e+00 -9.60884690e-02
-4.75059748e-01 -1.63257629e-01 -1.65826201e-01 1.66002154e-01
9.19489741e-01 -3.53430025e-02 -2.84062952e-01 -3.11330676e-01
-1.28451014e+00 3.14571336e-02 1.71705306e-01 -4.74754721e-02
-7.44197443e-02 -1.23664606e+00 -3.92826229e-01 1.88502327e-01
-1.05812699e-01 -8.20143446e-02 1.14401035e-01 1.14250219e+00
2.12931827e-01 1.42189667e-01 5.53178668e-01 -4.94247556e-01
-1.18090940e+00 6.53606892e-01 2.97705978e-02 -4.23874348e-01
-1.77845359e-01 1.18903446e+00 1.65065378e-01 -6.12004638e-01
8.00010622e-01 -3.59459996e-01 -1.04088604e-01 2.02996731e-01
4.29050267e-01 6.75162971e-01 1.46599308e-01 -5.62886000e-01
2.82335430e-02 8.05537581e-01 3.79472077e-02 9.65636447e-02
1.15213501e+00 -4.07077581e-01 -1.88872710e-01 8.21337044e-01
1.19189930e+00 8.45121220e-02 -1.28190291e+00 -7.61416495e-01
5.48965633e-01 -1.73717320e-01 -1.70528948e-01 -1.01080978e+00
-8.17457139e-01 7.63327897e-01 6.49847150e-01 9.06425953e-01
1.15385389e+00 2.30492279e-01 5.05966902e-01 2.75323838e-01
2.59174377e-01 -1.41327393e+00 5.36327213e-02 6.29268408e-01
5.54892480e-01 -1.47307324e+00 -1.83862358e-01 -7.70680249e-01
-6.64872766e-01 8.46821547e-01 2.81689644e-01 1.93859804e-02
9.65731800e-01 4.13816065e-01 1.48578599e-01 7.26458058e-02
-4.34058905e-01 -4.38908547e-01 1.55468762e-01 7.46115386e-01
2.74869889e-01 1.13472715e-01 -4.54073787e-01 9.80914235e-01
-3.80698144e-02 -6.70875758e-02 3.74403000e-01 8.37910414e-01
-7.87207603e-01 -1.28974390e+00 -6.08086467e-01 4.98799235e-01
-5.40508389e-01 -5.48251159e-02 -2.64470667e-01 4.47230756e-01
6.95456266e-01 1.24738204e+00 -4.14662927e-01 -3.96531254e-01
2.55586118e-01 7.23076522e-01 1.24117166e-01 -7.17475951e-01
-6.52079165e-01 5.72230369e-02 -6.77053705e-02 -2.72232175e-01
-7.02415049e-01 -5.52293420e-01 -1.05453110e+00 -1.27539888e-01
-5.66020608e-01 1.20626308e-01 6.66186988e-01 1.07191372e+00
-9.63458717e-02 3.90022308e-01 7.16943145e-01 -8.24342906e-01
-1.08811677e+00 -1.17356753e+00 -1.08053434e+00 7.13863909e-01
4.10221994e-01 -8.50486815e-01 -1.11234379e+00 3.33771855e-01] | [9.35429859161377, 4.033620834350586] |
a6b45a37-7768-4440-9c36-555f337dbf7a | logicsolver-towards-interpretable-math-word | 2205.08232 | null | https://arxiv.org/abs/2205.08232v3 | https://arxiv.org/pdf/2205.08232v3.pdf | LogicSolver: Towards Interpretable Math Word Problem Solving with Logical Prompt-enhanced Learning | Recently, deep learning models have made great progress in MWP solving on answer accuracy. However, they are uninterpretable since they mainly rely on shallow heuristics to achieve high performance without understanding and reasoning the grounded math logic. To address this issue and make a step towards interpretable MWP solving, we first construct a high-quality MWP dataset named InterMWP which consists of 11,495 MWPs and annotates interpretable logical formulas based on algebraic knowledge as the grounded linguistic logic of each solution equation. Different from existing MWP datasets, our InterMWP benchmark asks for a solver to not only output the solution expressions but also predict the corresponding logical formulas. We further propose a novel approach with logical prompt and interpretation generation, called LogicSolver. For each MWP, our LogicSolver first retrieves some highly-correlated algebraic knowledge and then passes them to the backbone model as prompts to improve the semantic representations of MWPs. With these improved semantic representations, our LogicSolver generates corresponding solution expressions and interpretable knowledge formulas in accord with the generated solution expressions, simultaneously. Experimental results show that our LogicSolver has stronger logical formula-based interpretability than baselines while achieving higher answer accuracy with the help of logical prompts, simultaneously. The source code and dataset is available at https://github.com/yangzhch6/InterMWP. | ['Xiaodan Liang', 'Liang Lin', 'Jiaqi Chen', 'Jinghui Qin', 'Zhicheng Yang'] | 2022-05-17 | null | null | null | null | ['math-word-problem-solving', 'math-word-problem-solving', 'math-word-problem-solving'] | ['knowledge-base', 'reasoning', 'time-series'] | [-9.55867618e-02 7.70626664e-01 -5.32230377e-01 -5.19273937e-01
-8.88892531e-01 -8.22064102e-01 5.39936088e-02 1.43726915e-01
3.09152991e-01 8.75297666e-01 2.83463657e-01 -6.94186270e-01
-4.24461871e-01 -1.50019813e+00 -1.11031592e+00 3.71323526e-02
2.88959503e-01 1.06989968e+00 -7.20674098e-02 -3.61967713e-01
-1.00109354e-03 9.32645351e-02 -1.21132803e+00 1.02044785e+00
1.31989431e+00 1.20504534e+00 -2.22727254e-01 3.94038320e-01
-7.00963736e-01 1.44035935e+00 -2.78749824e-01 -7.80019522e-01
2.23436862e-01 -3.23078662e-01 -1.50522852e+00 -4.38824534e-01
2.96466529e-01 -2.89577454e-01 -1.62513405e-01 1.00582337e+00
-1.52453467e-01 -7.33942613e-02 3.11531633e-01 -1.59076214e+00
-1.05303895e+00 1.37716556e+00 3.42291668e-02 -2.34805271e-01
8.56090009e-01 3.82138312e-01 1.68000102e+00 -8.28323781e-01
6.66318297e-01 1.61542404e+00 4.87050623e-01 4.68022168e-01
-1.23263001e+00 -6.15114450e-01 2.00510517e-01 9.29911673e-01
-1.23332214e+00 -2.13474423e-01 8.28034639e-01 1.24730840e-01
1.25822079e+00 4.45192546e-01 5.49406767e-01 8.11950862e-01
-1.02655172e-01 9.09165680e-01 8.46488237e-01 -3.89296412e-01
1.69727996e-01 2.13670153e-02 7.26010978e-01 1.23441768e+00
1.50605008e-01 -3.26390624e-01 -5.18842399e-01 -6.90782890e-02
4.08698231e-01 -3.71258914e-01 -3.45654070e-01 -2.58404459e-03
-1.10322773e+00 9.72711444e-01 8.02774787e-01 -1.50880339e-02
-2.73763329e-01 3.86502147e-01 2.79381752e-01 4.97989297e-01
6.09106990e-03 1.08175945e+00 -8.12708318e-01 4.69211526e-02
-3.69202793e-01 6.17545187e-01 1.30833590e+00 9.75202501e-01
8.52425218e-01 -3.84663612e-01 -2.03440413e-01 2.54171699e-01
2.44402960e-01 6.56287611e-01 -2.30397470e-02 -1.43465400e+00
9.78630364e-01 1.36199546e+00 1.65477972e-02 -1.36606359e+00
-4.38397974e-01 -4.49888073e-02 -4.05112028e-01 -3.57995719e-01
3.15063536e-01 -2.24451032e-02 -4.63531703e-01 1.72465491e+00
1.84346452e-01 2.90212005e-01 3.82474810e-01 9.46327865e-01
1.27744961e+00 1.00605881e+00 -1.65984139e-01 4.94252667e-02
1.19793439e+00 -1.15585363e+00 -5.87727129e-01 -2.03812078e-01
9.68866050e-01 -1.02724321e-01 1.36558270e+00 5.22029996e-01
-1.14134645e+00 -1.54433995e-01 -9.11456108e-01 -5.22001684e-01
-3.27907979e-01 -5.63060232e-02 1.04452395e+00 -1.44711465e-01
-8.72650802e-01 3.06428254e-01 -5.89888096e-01 1.77823395e-01
5.98283291e-01 4.41586673e-01 -1.46023020e-01 -4.21806365e-01
-1.53208816e+00 9.68922198e-01 8.27427328e-01 1.89692274e-01
-5.00048220e-01 -1.01241362e+00 -1.24614084e+00 3.00641149e-01
1.00713038e+00 -1.06560910e+00 1.35232031e+00 -8.94222081e-01
-1.48367476e+00 5.08372068e-01 -5.67626595e-01 -6.41110420e-01
1.23430640e-01 -2.71439612e-01 -3.84115726e-01 2.65340030e-01
6.59571663e-02 6.48761630e-01 1.63216054e-01 -1.06046903e+00
-2.41595447e-01 -5.82797714e-02 9.31577742e-01 -7.54800364e-02
8.72695819e-02 -2.41999969e-01 -2.78546274e-01 -1.00110453e-02
2.48972967e-01 -5.86256504e-01 1.90163981e-02 -5.37880920e-02
-6.83881104e-01 -6.12514555e-01 3.85129839e-01 -5.51602125e-01
1.21256506e+00 -1.64559710e+00 6.08871520e-01 3.45798850e-01
6.13617241e-01 7.82378167e-02 -4.18735892e-01 3.68990719e-01
-1.22456014e-01 2.34923989e-01 -1.87353373e-01 9.05877277e-02
5.96434176e-01 6.87025547e-01 -9.36744690e-01 -3.08845550e-01
6.26225293e-01 1.55325997e+00 -1.14133060e+00 -4.69815046e-01
1.61385387e-01 -2.55418420e-01 -9.46963370e-01 8.61413404e-02
-1.15301895e+00 2.35462245e-02 -5.45030713e-01 7.06559062e-01
6.21487319e-01 -5.93634427e-01 4.19350684e-01 -4.36259180e-01
4.75053996e-01 7.03888535e-01 -1.13342261e+00 1.61881649e+00
-7.51561165e-01 4.77680981e-01 -3.98901910e-01 -1.14268923e+00
8.39261413e-01 3.68936621e-02 -1.51776411e-02 -6.83204055e-01
-4.49089427e-03 3.54774982e-01 -3.85964923e-02 -6.93008661e-01
2.57407725e-01 -1.01821117e-01 -1.48462862e-01 3.43748599e-01
3.97032276e-02 -3.04432362e-01 3.29464972e-01 5.26937127e-01
1.11334777e+00 1.63687691e-01 1.71027645e-01 -9.05131176e-02
8.98741782e-01 4.73945975e-01 6.91056728e-01 6.16549969e-01
4.07618105e-01 7.99528509e-02 1.02286959e+00 -9.22990024e-01
-3.80477220e-01 -1.11228704e+00 2.18654960e-01 6.54783905e-01
3.16405445e-01 -9.70528841e-01 -5.83171666e-01 -8.07590783e-01
9.70491022e-02 1.10502911e+00 -3.03729862e-01 -3.90519321e-01
-8.11608076e-01 -1.84201866e-01 7.23800957e-01 7.42970347e-01
7.11748302e-01 -1.19125187e+00 -2.53556222e-01 1.50003150e-01
-9.15566385e-01 -1.17285752e+00 2.55613148e-01 -1.34875234e-02
-6.25675380e-01 -1.49550605e+00 4.68800992e-01 -5.64964890e-01
8.11463833e-01 -3.35699767e-01 1.56512678e+00 5.05342185e-01
1.26058444e-01 1.49905667e-01 -3.42364013e-01 -3.51945102e-01
-3.89056355e-01 2.11621955e-01 -1.38219222e-01 -2.76121467e-01
5.95960498e-01 -3.12863648e-01 -9.90172848e-02 -3.87854725e-02
-6.22704268e-01 5.85985720e-01 2.13752389e-01 6.67602241e-01
5.90996325e-01 2.08432764e-01 4.60293829e-01 -9.91508543e-01
5.09746432e-01 -5.72605073e-01 -7.08977461e-01 6.03296697e-01
-4.09864485e-01 5.06233454e-01 1.20756090e+00 4.27170321e-02
-9.08361673e-01 -3.51626039e-01 -6.54837191e-02 -3.24897230e-01
4.43189293e-02 1.00678086e+00 -3.22674990e-01 1.33908600e-01
7.36160517e-01 -3.28956507e-02 -3.44989032e-01 -1.05461247e-01
7.14464843e-01 1.60513490e-01 6.07534528e-01 -1.47528756e+00
1.08270323e+00 1.82228051e-02 1.14160582e-01 4.20243107e-02
-1.52221739e+00 2.76160836e-01 -2.70640492e-01 3.45857531e-01
4.56165105e-01 -6.35839820e-01 -1.07639217e+00 -4.39101048e-02
-1.50503075e+00 -6.57584786e-01 -3.16894740e-01 -2.47632321e-02
-5.57988048e-01 8.06858912e-02 -7.72378147e-01 -5.57767868e-01
-4.73294318e-01 -1.08906066e+00 9.11574602e-01 -1.53256627e-02
-7.24530041e-01 -1.11175692e+00 -3.12618613e-01 7.37247765e-01
9.44622979e-02 2.68928975e-01 1.73383760e+00 -5.79263985e-01
-9.22127962e-01 4.21533696e-02 -5.25741279e-01 1.59271270e-01
-1.22902051e-01 -1.33181244e-01 -6.79452717e-01 4.49742198e-01
-2.71702588e-01 -6.06765568e-01 5.50359547e-01 9.31841061e-02
1.69360542e+00 -9.85888779e-01 1.31484261e-02 7.26363122e-01
1.34254611e+00 -2.47633055e-01 6.18573725e-01 2.67787099e-01
5.74834645e-01 4.44091707e-01 5.82174003e-01 1.14190847e-01
9.17824268e-01 4.23895061e-01 4.98214006e-01 2.50841916e-01
2.51504958e-01 -5.79313219e-01 2.44261399e-01 5.30961752e-01
-4.48525622e-02 -1.05223477e-01 -1.23152637e+00 2.92199612e-01
-2.09974742e+00 -8.40775788e-01 -3.82861257e-01 1.47564507e+00
1.29147506e+00 4.85964231e-02 -3.43730867e-01 1.45544231e-01
1.37026589e-02 1.01224491e-02 -6.03979528e-01 -6.09199643e-01
-1.16328061e-01 7.14589596e-01 4.15976495e-02 1.02297115e+00
-6.30021572e-01 1.20683217e+00 5.02882290e+00 6.43738151e-01
-9.01119947e-01 -2.07794055e-01 4.00858760e-01 -9.07357857e-02
-7.71419823e-01 1.43096074e-01 -6.31929100e-01 2.53892362e-01
7.69604146e-01 -3.17082107e-01 1.01233423e+00 8.61932516e-01
-7.02715069e-02 5.40370122e-02 -1.53039336e+00 8.12616765e-01
-2.27926463e-01 -1.88429284e+00 3.77831221e-01 -5.31695247e-01
5.52880704e-01 -3.67561042e-01 -1.52219817e-01 7.85762429e-01
5.10523915e-01 -1.47075570e+00 6.69774652e-01 6.47812247e-01
4.04969096e-01 -8.17840040e-01 9.14246917e-01 3.95225108e-01
-1.16983771e+00 -3.30260634e-01 -3.18233848e-01 -7.32474685e-01
-7.44452327e-02 5.17664135e-01 -7.10212886e-01 8.93974602e-01
2.65354156e-01 8.94392610e-01 -5.40160298e-01 2.77239770e-01
-1.00389779e+00 5.42018175e-01 -3.25030982e-01 5.91459386e-02
3.04227680e-01 -1.61686584e-01 2.74502486e-01 7.26106882e-01
1.39400631e-01 6.13196194e-01 1.22511387e-01 1.81929433e+00
-2.89171726e-01 -2.60557115e-01 -3.28942627e-01 -2.20731318e-01
6.98435426e-01 1.33172703e+00 -2.32276022e-01 -7.89172709e-01
-1.41304940e-01 6.48232162e-01 6.86943412e-01 5.12246728e-01
-1.19303024e+00 -1.88599125e-01 6.79252446e-01 -1.23435527e-01
-2.43327186e-01 1.79557838e-02 -6.70845151e-01 -1.52852130e+00
4.01740611e-01 -1.11568165e+00 6.24220252e-01 -1.12948620e+00
-1.20707178e+00 3.07147563e-01 1.78267047e-01 -3.95815670e-01
-2.43491769e-01 -9.54799712e-01 -6.49582922e-01 9.09739673e-01
-1.56867599e+00 -1.22185433e+00 -5.37121236e-01 4.33269233e-01
3.68796140e-01 1.82065532e-01 9.96023655e-01 -1.77934840e-01
-6.12737298e-01 4.68351871e-01 -7.59239495e-01 2.41597205e-01
2.60029286e-01 -1.36880326e+00 1.88202575e-01 6.43746257e-01
1.49751380e-01 9.51019585e-01 5.51308274e-01 -3.78819555e-01
-2.01872683e+00 -1.24396372e+00 1.12887990e+00 -8.69479775e-01
8.67570400e-01 -5.41466288e-02 -9.48676646e-01 1.14765894e+00
5.46124056e-02 -6.40330166e-02 4.17968124e-01 3.53113115e-01
-6.08476639e-01 -8.29924196e-02 -1.05552602e+00 6.49287164e-01
1.24351156e+00 -4.76445496e-01 -1.19343996e+00 6.27006948e-01
1.17359602e+00 -9.50362325e-01 -7.85707653e-01 5.09632885e-01
8.79958719e-02 -6.65253162e-01 1.08765030e+00 -9.95291770e-01
1.23582172e+00 -5.15794754e-01 -1.93103388e-01 -1.16604018e+00
-2.13561639e-01 -2.18275815e-01 -7.07046151e-01 9.26053643e-01
6.64493740e-01 -8.78743410e-01 6.18099153e-01 1.14496291e+00
-2.40804538e-01 -1.27942967e+00 -4.71322745e-01 -4.92276937e-01
-2.06352267e-02 -8.25275958e-01 1.09402800e+00 1.01039577e+00
3.72834414e-01 3.90833437e-01 2.23862067e-01 4.13985789e-01
3.96581441e-01 9.46866870e-01 7.11247206e-01 -1.00257444e+00
-5.90241015e-01 -4.42244828e-01 1.52613232e-02 -7.81670451e-01
8.94717455e-01 -1.51879144e+00 -2.54301697e-01 -1.83089161e+00
3.25004645e-02 -4.12934422e-01 -1.27672344e-01 1.23855305e+00
-8.90403800e-03 -1.62728414e-01 2.58120745e-01 -3.22579503e-01
-8.69735897e-01 2.59476662e-01 1.26110387e+00 -3.95142496e-01
-9.95341572e-04 -5.82489431e-01 -1.09348953e+00 9.45623934e-01
7.31353223e-01 -3.24764736e-02 -6.54993832e-01 -8.65455985e-01
9.04290259e-01 1.60875455e-01 8.01733673e-01 -7.62308717e-01
2.56767601e-01 -6.63301945e-01 7.52191991e-02 -2.96853870e-01
1.22707114e-01 -6.70295537e-01 2.43650615e-01 3.66256058e-01
-5.08681834e-01 -7.36993477e-02 2.60015547e-01 -1.12283486e-03
-3.40339690e-01 -1.71577200e-01 2.94250816e-01 -2.61230677e-01
-9.22372878e-01 7.81822205e-02 1.10358424e-01 4.97515142e-01
7.23519981e-01 3.53199333e-01 -6.76140904e-01 -2.18637958e-01
-5.18641531e-01 7.34228790e-01 2.90514261e-01 1.53257474e-01
7.33684540e-01 -1.32540047e+00 -3.86593699e-01 6.15783595e-02
5.86521700e-02 6.83867216e-01 -3.76607589e-02 9.05055285e-01
-8.12262237e-01 7.37316728e-01 1.56976152e-02 -7.83991739e-02
-7.98951924e-01 4.48387414e-01 5.83333135e-01 -5.52080214e-01
-6.25483572e-01 9.28215742e-01 1.20323561e-01 -7.96793878e-01
7.71401376e-02 -1.09811711e+00 1.41144574e-01 -3.72726351e-01
6.47457957e-01 1.20349817e-01 -1.10948585e-01 4.79350761e-02
-5.26017606e-01 3.69096845e-01 2.34798312e-01 4.27397966e-01
1.32059264e+00 3.52973908e-01 -7.50873029e-01 9.34941694e-02
8.00666392e-01 -6.67838976e-02 -4.80382979e-01 -4.48321015e-01
1.54734418e-01 -3.54457617e-01 -3.52789342e-01 -1.09170926e+00
-1.08440661e+00 7.23409891e-01 -5.88999689e-01 -1.50267065e-01
1.09933484e+00 2.14628741e-01 1.13827360e+00 1.05611086e+00
5.83015501e-01 -6.46468520e-01 3.41754928e-02 9.59580362e-01
1.07038784e+00 -1.12438083e+00 -2.05819741e-01 -8.06911588e-01
-4.55090970e-01 1.30296278e+00 9.73925591e-01 -1.34791479e-01
-1.48937449e-01 2.85956711e-01 -1.50108367e-01 -3.35745066e-01
-1.06116569e+00 1.12420365e-01 3.66888136e-01 3.25320274e-01
2.14464203e-01 2.20451385e-01 -1.84183195e-02 1.36752975e+00
-7.03128517e-01 3.52402806e-01 1.94061324e-01 4.04153734e-01
-3.15969706e-01 -1.15131247e+00 -2.26808146e-01 3.96981388e-01
1.86555579e-01 -3.27155262e-01 -6.66716993e-01 5.54599285e-01
1.45012885e-01 7.97941148e-01 -2.84478217e-01 -3.20591480e-01
1.78095058e-01 2.86359072e-01 6.80209994e-01 -7.26015270e-01
-3.12771052e-01 -1.11832404e+00 6.37763441e-01 -1.01089442e+00
-2.07746942e-02 -4.04547481e-03 -2.07712817e+00 -7.72736013e-01
1.29146054e-01 3.28496158e-01 -1.37111410e-01 1.36642170e+00
4.89865363e-01 5.50868094e-01 1.07963011e-01 -3.02742988e-01
-5.30573666e-01 -3.92035514e-01 2.98784338e-02 4.94967669e-01
1.81174185e-02 -5.18984556e-01 -1.90475374e-01 -8.61338526e-02] | [9.472515106201172, 7.508758068084717] |
07020888-ad77-48d2-b5c3-feaf0851ce2a | a-conditional-adversarial-network-for-scene | 1904.11163 | null | http://arxiv.org/abs/1904.11163v1 | http://arxiv.org/pdf/1904.11163v1.pdf | A Conditional Adversarial Network for Scene Flow Estimation | The problem of Scene flow estimation in depth videos has been attracting
attention of researchers of robot vision, due to its potential application in
various areas of robotics. The conventional scene flow methods are difficult to
use in reallife applications due to their long computational overhead. We
propose a conditional adversarial network SceneFlowGAN for scene flow
estimation. The proposed SceneFlowGAN uses loss function at two ends: both
generator and descriptor ends. The proposed network is the first attempt to
estimate scene flow using generative adversarial networks, and is able to
estimate both the optical flow and disparity from the input stereo images
simultaneously. The proposed method is experimented on a large RGB-D benchmark
sceneflow dataset. | ['Snehasis Mukherjee', 'Ravi Kumar Thakur'] | 2019-04-25 | null | null | null | null | ['scene-flow-estimation'] | ['computer-vision'] | [ 3.33672851e-01 -1.20872492e-02 3.39816302e-01 -2.32212126e-01
-1.18806884e-01 -4.06605244e-01 6.21885955e-01 -4.84417886e-01
-5.73722124e-01 8.62910271e-01 -6.24527000e-02 -1.18324004e-01
3.15796971e-01 -8.25606048e-01 -6.97870433e-01 -6.67095184e-01
1.65208578e-01 1.19284891e-01 3.96145105e-01 -3.80814970e-02
4.83272761e-01 5.83087087e-01 -1.30344832e+00 -2.82868296e-01
7.73114204e-01 9.00223851e-01 3.01572502e-01 1.05523634e+00
-8.69755968e-02 1.65101016e+00 -5.03186584e-01 -3.05234134e-01
6.86107874e-01 -6.58860981e-01 -8.30494285e-01 1.35013342e-01
6.14758849e-01 -6.88372791e-01 -1.02178347e+00 1.28261900e+00
4.53248233e-01 3.81624937e-01 5.86383283e-01 -1.53131104e+00
-1.97672829e-01 1.54000282e-01 -5.98204076e-01 4.62218255e-01
3.34496200e-01 3.98465425e-01 4.65556502e-01 -4.54375565e-01
6.45502806e-01 1.28347063e+00 3.32479119e-01 6.03589237e-01
-8.90463591e-01 -6.82980418e-01 -8.07589218e-02 3.21187109e-01
-1.02248585e+00 -3.54404867e-01 1.00703704e+00 -6.67761207e-01
6.87587738e-01 -3.06051642e-01 5.82323372e-01 9.19761419e-01
5.08254826e-01 6.29349291e-01 1.10773909e+00 -9.86495018e-02
2.06724986e-01 -1.40677258e-01 -2.79656708e-01 7.30741441e-01
1.76496863e-01 3.00984621e-01 -2.38593549e-01 4.14357364e-01
1.29066467e+00 -2.44240277e-02 -4.66320843e-01 -4.33664680e-01
-1.00883925e+00 9.04945076e-01 8.39200914e-01 -3.64103280e-02
-2.57880449e-01 5.89618564e-01 3.02319884e-01 4.91897538e-02
2.33669609e-01 2.83307314e-01 1.20216146e-01 -2.95211852e-01
-6.59870386e-01 2.47342050e-01 7.95792937e-01 9.33767974e-01
6.81335628e-01 3.82860452e-01 2.88286805e-01 4.40912426e-01
4.06501442e-01 5.76338589e-01 6.83811128e-01 -1.24297380e+00
5.90092599e-01 3.91658694e-01 1.79215595e-02 -1.33246696e+00
-2.28384510e-01 6.69981539e-02 -1.05423832e+00 6.75462067e-01
5.69701314e-01 -4.09980983e-01 -9.54747081e-01 1.52205265e+00
3.22541058e-01 5.36279261e-01 2.77924418e-01 1.05993247e+00
9.74526703e-01 9.38778043e-01 -8.42302963e-02 1.03646107e-01
7.08674908e-01 -9.54536617e-01 -6.04494095e-01 -6.01323009e-01
2.96502709e-02 -9.14397776e-01 4.95008588e-01 3.02171201e-01
-9.86939490e-01 -6.80289388e-01 -1.10574973e+00 -3.00818682e-01
-9.79643539e-02 -1.52375072e-01 8.70946348e-01 5.93180537e-01
-8.82256746e-01 3.01610291e-01 -9.67673838e-01 -3.14842224e-01
5.15884936e-01 3.75152260e-01 -4.36387599e-01 -1.59902677e-01
-9.15630877e-01 7.79519141e-01 5.80549657e-01 4.44279134e-01
-1.21154237e+00 -2.74875224e-01 -1.18305421e+00 -1.39172256e-01
-4.59533669e-02 -8.76022696e-01 9.25245166e-01 -1.01920509e+00
-1.78490961e+00 7.06267059e-01 1.49546519e-01 -5.09842038e-01
9.60198045e-01 -1.72232583e-01 1.11177444e-01 1.26063049e-01
-4.02669422e-02 7.73583949e-01 8.96870494e-01 -9.21237588e-01
-6.19146168e-01 -3.15527201e-01 2.77282178e-01 3.92521113e-01
1.49523661e-01 -4.73327845e-01 -2.30075493e-01 -5.22143185e-01
8.62023532e-02 -8.93551230e-01 -4.44034040e-01 1.15041345e-01
-5.23703039e-01 2.84635991e-01 1.02827084e+00 -3.75248998e-01
4.44521368e-01 -1.78777969e+00 1.37754619e-01 -2.65329629e-01
6.55689761e-02 4.39151704e-01 1.39423281e-01 1.15069672e-01
-5.92500297e-03 -3.95303339e-01 -3.45439076e-01 -2.17314303e-01
-5.95222533e-01 2.53437191e-01 -3.37682962e-01 7.58014977e-01
1.98478177e-01 6.96827352e-01 -1.06805980e+00 -5.66781580e-01
9.12502587e-01 6.21811271e-01 -6.70959949e-01 6.18611455e-01
4.68662567e-02 8.06509078e-01 -4.02204871e-01 6.15585558e-02
9.02782261e-01 4.59296517e-02 -2.40102366e-01 -1.03540406e-01
-5.19250967e-02 -1.06949672e-01 -1.25859141e+00 1.74159050e+00
-5.08023441e-01 1.26434064e+00 -2.53309906e-01 -9.57066655e-01
9.60377932e-01 -1.07118107e-01 5.74509740e-01 -3.36353362e-01
5.22236168e-01 -2.38922387e-02 2.38087792e-02 -6.37518644e-01
5.87160587e-01 -2.02843800e-01 1.05758369e-01 1.39720857e-01
2.37929344e-01 -8.52602303e-01 2.24850386e-01 2.35720724e-01
1.10525894e+00 3.66156042e-01 3.28233927e-01 1.37595505e-01
9.79253590e-01 -8.57636407e-02 5.19696355e-01 3.27631980e-01
-5.25582671e-01 6.73476338e-01 4.19553608e-01 -5.53852260e-01
-1.13227475e+00 -9.51100588e-01 1.50106996e-01 2.14162201e-01
6.51529551e-01 3.05299789e-01 -6.49586082e-01 -6.00660145e-01
-1.96451336e-01 2.86957592e-01 -5.35851002e-01 -1.30654782e-01
-7.99716592e-01 -2.94199646e-01 5.93812764e-01 3.22315007e-01
1.11434782e+00 -1.26285470e+00 -9.58433807e-01 1.84129611e-01
-1.90008566e-01 -1.58471394e+00 -3.83903623e-01 -6.12186305e-02
-7.90321887e-01 -1.36088753e+00 -8.73602927e-01 -8.47268224e-01
6.61018431e-01 2.39432737e-01 7.50985920e-01 -3.09864789e-01
-4.38144594e-01 1.19902231e-01 -1.62475631e-01 -3.21253985e-01
-6.08754039e-01 -6.36999607e-02 -3.21931601e-01 1.17718883e-01
1.04070045e-01 -4.63847071e-01 -7.65356779e-01 3.47006112e-01
-1.05228245e+00 -2.99268048e-02 3.84951919e-01 8.34986866e-01
2.42103323e-01 7.60407373e-02 1.41508669e-01 -9.99181390e-01
2.46563241e-01 -3.67147356e-01 -1.05352330e+00 -2.92035639e-01
-1.45426288e-01 3.92551534e-02 7.46007621e-01 2.71390821e-03
-1.33913863e+00 3.17162186e-01 -2.21304715e-01 -5.38106978e-01
-1.59289062e-01 -1.46579137e-02 -1.61800515e-02 -3.41846794e-01
4.86939698e-01 1.08090468e-01 8.15890729e-02 1.75551534e-01
2.32737690e-01 4.37988311e-01 9.12224591e-01 -7.65278586e-04
8.36748004e-01 6.20508015e-01 4.59060311e-01 -9.41067338e-01
-5.00146449e-01 -5.03466070e-01 -4.85391200e-01 -3.84693146e-01
1.18986118e+00 -8.60709667e-01 -8.39536130e-01 9.81293380e-01
-1.49676442e+00 -2.45328337e-01 -1.65201068e-01 6.73297524e-01
-9.04769659e-01 4.56258237e-01 -5.16535938e-01 -5.81509113e-01
-1.86949372e-01 -1.52757990e+00 6.83127880e-01 5.32715201e-01
2.92656243e-01 -1.18779337e+00 1.72398314e-01 1.69030756e-01
2.89485306e-01 7.91518211e-01 2.64335960e-01 9.28771347e-02
-9.78277028e-01 -1.31424308e-01 -3.95687252e-01 6.70242548e-01
3.52538675e-01 5.58567718e-02 -9.77632940e-01 -6.18515573e-02
2.51605958e-01 -3.65429103e-01 7.62753010e-01 6.65418088e-01
9.59177077e-01 -7.66905993e-02 -4.52200994e-02 1.03279054e+00
1.90049708e+00 4.91181761e-01 1.02079546e+00 2.04518422e-01
1.14967263e+00 3.54847997e-01 4.29332823e-01 3.52230668e-01
1.49842724e-01 4.26257730e-01 8.57405543e-01 -3.78333777e-02
-3.26636165e-01 -4.19653773e-01 3.14843684e-01 4.56896544e-01
-7.06229582e-02 -5.40706098e-01 -8.23264599e-01 4.61659998e-01
-1.71328998e+00 -9.06439662e-01 -3.11971664e-01 2.00502896e+00
2.85645545e-01 2.02318907e-01 -2.49236420e-01 1.80404186e-01
7.04426110e-01 2.65362650e-01 -6.60331309e-01 -4.71284062e-01
2.48648264e-02 1.66262358e-01 8.51846278e-01 7.33839929e-01
-1.15163898e+00 1.08043444e+00 5.92828941e+00 1.82857305e-01
-1.27185190e+00 -1.49468914e-01 7.02947557e-01 2.60700256e-01
1.63120881e-01 4.40044254e-02 -5.36509216e-01 4.53814805e-01
4.33601111e-01 -8.25589076e-02 3.72452945e-01 7.81516969e-01
1.57260641e-01 -5.13876796e-01 -9.33362961e-01 1.30515766e+00
1.18811585e-01 -1.12416387e+00 1.81318093e-02 -1.19191967e-01
9.24395323e-01 6.57161046e-03 -7.62218959e-04 -2.49991536e-01
4.59322006e-01 -1.01331568e+00 4.39743280e-01 5.72400033e-01
6.19334400e-01 -7.51939952e-01 1.01464903e+00 2.17045292e-01
-8.78966093e-01 3.62772383e-02 -4.39505100e-01 -1.82271019e-01
5.37566960e-01 2.81287462e-01 -8.98435593e-01 1.77840054e-01
4.44003254e-01 1.03569305e+00 -2.34736592e-01 1.12474000e+00
-3.92800540e-01 2.80904651e-01 -1.61717057e-01 2.60809083e-02
4.26948994e-01 -3.69698673e-01 6.15768671e-01 8.80037844e-01
1.56847417e-01 -7.82954469e-02 -8.34662169e-02 9.42519367e-01
-7.44493529e-02 -1.52884454e-01 -9.88350511e-01 8.18727314e-02
8.15685019e-02 1.00576293e+00 -8.06849778e-01 -7.74189383e-02
-1.78443849e-01 1.08390176e+00 6.42560944e-02 2.32237652e-01
-7.73315072e-01 -7.75404692e-01 6.19233429e-01 -1.11640915e-01
4.11633477e-02 -4.10187542e-01 -1.43846646e-01 -1.26762033e+00
-3.87567937e-01 -4.00862515e-01 1.25028431e-01 -9.13598835e-01
-9.50072348e-01 6.50736511e-01 -1.15893863e-01 -1.41134310e+00
-5.79125464e-01 -7.41991341e-01 -5.56436777e-01 9.42766964e-01
-1.61349821e+00 -9.07588184e-01 -1.03610468e+00 9.11208451e-01
8.09492171e-01 -2.75358826e-01 3.26476038e-01 3.18871945e-01
-3.98174316e-01 2.40894318e-01 -1.38990968e-01 3.84324044e-01
5.50887287e-01 -1.28555560e+00 5.13463974e-01 1.37001717e+00
-1.72735110e-01 1.87138587e-01 9.38490927e-01 -3.40798706e-01
-1.32367134e+00 -1.13222158e+00 3.34819913e-01 -1.95791617e-01
3.43699485e-01 -9.51810703e-02 -3.77469391e-01 7.86687493e-01
4.62905526e-01 4.03069049e-01 -9.12804380e-02 -9.49723721e-01
2.90047228e-01 -3.11621457e-01 -1.37500715e+00 1.83694363e-01
6.99477613e-01 -1.34970054e-01 -2.33929604e-01 5.12644947e-02
5.15585899e-01 -6.93786740e-01 -5.89866638e-01 2.06039339e-01
4.29590106e-01 -1.32196820e+00 9.62410569e-01 -3.65241729e-02
8.92192841e-01 -4.49629068e-01 -3.17425653e-02 -1.43145323e+00
1.26801595e-01 -7.31506705e-01 2.07918271e-01 1.16699719e+00
-1.63134336e-01 -7.97978640e-01 1.00960457e+00 3.46293330e-01
-1.93238463e-02 -2.09077597e-01 -7.23354816e-01 -3.14124167e-01
-1.27892703e-01 -3.43381494e-01 2.61205971e-01 5.86543620e-01
-5.50403595e-01 4.47313726e-01 -5.97497344e-01 -6.63719475e-02
8.37676108e-01 -2.09469959e-01 1.33568406e+00 -8.36796463e-01
-3.10852498e-01 -3.44056696e-01 -1.20236826e+00 -1.27782035e+00
1.29218161e-01 -5.36487520e-01 4.87127483e-01 -1.50881672e+00
-2.34190896e-01 -4.42575157e-01 1.73080027e-01 -1.44155800e-01
-1.55208752e-01 4.80925173e-01 2.27720737e-01 -1.22960852e-02
-5.13432659e-02 5.83453238e-01 1.61821485e+00 -1.85078204e-01
-1.70998156e-01 6.29270375e-02 -4.47220877e-02 8.10313404e-01
6.09344065e-01 -2.20724002e-01 -7.54426718e-01 -4.58253950e-01
-4.74595018e-02 2.87383527e-01 4.48792130e-01 -1.35197043e+00
4.06865090e-01 -2.02832341e-01 4.70260233e-01 -5.67484260e-01
3.78612041e-01 -9.04369473e-01 2.02097088e-01 5.61050534e-01
-1.53856695e-01 3.44808809e-02 5.55590242e-02 5.84040761e-01
-4.93060976e-01 -2.85706192e-01 1.20376670e+00 -4.10638332e-01
-1.13598478e+00 5.10325909e-01 -1.77769721e-01 2.12221295e-01
1.33877909e+00 -2.53485799e-01 -4.10823107e-01 -5.47683299e-01
-2.28394628e-01 -9.72057953e-02 4.60797101e-01 5.16658127e-01
9.48484004e-01 -1.04007113e+00 -4.39637870e-01 3.13580304e-01
-4.21775281e-02 4.61754322e-01 6.99183643e-02 3.72470558e-01
-1.57531965e+00 2.62487859e-01 -7.41409898e-01 -6.94475174e-01
-8.43176126e-01 4.52473491e-01 5.51978469e-01 -9.54088420e-02
-5.75262308e-01 8.91848743e-01 3.62181067e-01 -8.99775848e-02
1.14465758e-01 -3.50849301e-01 -2.13876024e-01 -5.16103446e-01
2.97246724e-01 6.38327599e-01 -2.93244481e-01 -9.24944401e-01
-1.77248701e-01 7.67250717e-01 2.99214303e-01 -1.46666780e-01
1.07001364e+00 -3.10746878e-01 -1.31679684e-01 2.61383444e-01
1.50874138e+00 -3.48016381e-01 -1.52237499e+00 1.49574533e-01
-5.58868229e-01 -9.22756732e-01 2.24245027e-01 -6.31451532e-02
-1.58117223e+00 1.11092722e+00 7.46580243e-01 5.16832210e-02
1.13920796e+00 -6.02874339e-01 1.02658725e+00 1.21718556e-01
5.43974996e-01 -6.03851676e-01 1.53153405e-01 4.28711921e-01
4.69241798e-01 -1.37945998e+00 -1.28393332e-02 -3.92006069e-01
-5.45879841e-01 1.22780991e+00 8.96230876e-01 -7.58770585e-01
6.29757226e-01 1.52179196e-01 2.14211568e-01 1.34851381e-01
-2.80003160e-01 -2.83618346e-02 -6.85687512e-02 7.29039192e-01
2.42300376e-01 -2.51180202e-01 -2.29718924e-01 -4.94586974e-01
-2.20230684e-01 2.23080754e-01 1.08463657e+00 7.58155763e-01
-3.38963181e-01 -8.54690015e-01 -2.38891661e-01 1.70754269e-01
-6.35362506e-01 2.57432759e-01 9.18068181e-05 7.99471438e-01
8.26894268e-02 9.68095183e-01 1.63813174e-01 -1.54131874e-01
2.56855369e-01 -6.70324981e-01 6.32254601e-01 -1.96835041e-01
-1.94286585e-01 -3.57128382e-01 -3.57239872e-01 -6.83626473e-01
-8.53705108e-01 -5.29243886e-01 -1.14460659e+00 -2.89319187e-01
-3.84947732e-02 -2.14980263e-02 8.35996091e-01 6.73619211e-01
-1.40863612e-01 5.53806722e-01 9.21737313e-01 -1.00201166e+00
3.66848260e-02 -9.95626628e-01 -4.43156630e-01 5.81224918e-01
6.70234263e-01 -7.10394442e-01 -4.96915281e-01 3.74653786e-01] | [8.684539794921875, -2.071789026260376] |
ed2fff21-1378-466b-8bdb-73bd7f5fcb75 | retrieve-anyone-a-general-purpose-person-re | 2306.07520 | null | https://arxiv.org/abs/2306.07520v3 | https://arxiv.org/pdf/2306.07520v3.pdf | Retrieve Anyone: A General-purpose Person Re-identification Task with Instructions | Human intelligence can retrieve any person according to both visual and language descriptions. However, the current computer vision community studies specific person re-identification (ReID) tasks in different scenarios separately, which limits the applications in the real world. This paper strives to resolve this problem by proposing a new instruct-ReID task that requires the model to retrieve images according to the given image or language instructions.Our instruct-ReID is a more general ReID setting, where existing ReID tasks can be viewed as special cases by designing different instructions. We propose a large-scale OmniReID benchmark and an adaptive triplet loss as a baseline method to facilitate research in this new setting. Experimental results show that the baseline model trained on our OmniReID benchmark can improve +0.6%, +1.4%, 0.2% mAP on Market1501, CUHK03, MSMT17 for traditional ReID, +0.8%, +2.0%, +13.4% mAP on PRCC, VC-Clothes, LTCC for clothes-changing ReID, +11.7% mAP on COCAS+ real2 for clothestemplate based clothes-changing ReID when using only RGB images, +25.4% mAP on COCAS+ real2 for our newly defined language-instructed ReID. The dataset, model, and code will be available at https://github.com/hwz-zju/Instruct-ReID. | ['Yunfeng Yan', 'Donglian Qi', 'Wanli Ouyang', 'Rui Zhao', 'Feng Zhu', 'Lei Bai', 'Yizhou Wang', 'Qingsong Xie', 'Qihao Chen', 'Yiheng Deng', 'Shixiang Tang', 'Weizhen He'] | 2023-06-13 | null | null | null | null | ['person-re-identification'] | ['computer-vision'] | [-3.50259423e-01 -4.31799799e-01 1.19228020e-01 -4.18013483e-01
-5.48450351e-01 -5.84816992e-01 6.68288112e-01 -3.28144163e-01
-9.45173085e-01 6.74524188e-01 -3.08439974e-02 7.05540925e-02
1.49385661e-01 -5.24539292e-01 -7.28752673e-01 -6.31054878e-01
1.82978436e-01 5.80717683e-01 -1.26637697e-01 -2.68041700e-01
4.23139967e-02 2.94966131e-01 -1.48328996e+00 -1.30453527e-01
7.59460270e-01 7.07183957e-01 1.32688627e-01 6.31306767e-01
2.44419396e-01 1.81077674e-01 -6.85255170e-01 -1.22973490e+00
7.47784674e-01 -8.04382935e-02 -4.55729485e-01 -1.55269995e-01
1.04175794e+00 -4.50055957e-01 -5.44551790e-01 1.16619158e+00
9.36567426e-01 5.65431155e-02 6.28436863e-01 -1.52333450e+00
-1.17995334e+00 2.34432533e-01 -9.29830253e-01 5.65547235e-02
5.79509258e-01 3.20518404e-01 3.86275709e-01 -8.88322711e-01
5.96146405e-01 1.33225274e+00 7.44439602e-01 9.11495388e-01
-9.65335071e-01 -1.14098775e+00 2.26275474e-01 4.84248668e-01
-1.77105474e+00 -3.72107148e-01 4.14077580e-01 -2.57896900e-01
6.65350854e-01 4.36453521e-01 5.18575072e-01 1.33931315e+00
-2.27206364e-01 8.69502366e-01 1.28157914e+00 -1.77885085e-01
-2.26246506e-01 4.53030854e-01 4.92398590e-01 5.96674085e-01
4.28985804e-01 1.22649513e-01 -3.70104730e-01 2.19992951e-01
5.41880965e-01 1.68812964e-02 -4.93275881e-01 -1.11842439e-01
-1.21839619e+00 3.84279877e-01 6.17855668e-01 -1.49934202e-01
-7.53904600e-03 -2.00144723e-01 4.18771982e-01 4.02354866e-01
4.53996718e-01 2.98285306e-01 -3.26182485e-01 6.92515448e-02
-5.63120067e-01 5.63756764e-01 6.32340550e-01 1.41256464e+00
6.24897063e-01 -1.85062826e-01 -1.71601340e-01 1.17002606e+00
-2.92629711e-02 1.17176449e+00 3.63546878e-01 -7.56293297e-01
5.53323328e-01 1.84732690e-01 3.72344434e-01 -9.64735270e-01
-4.12804484e-01 -5.63973904e-01 -1.07407689e+00 -1.43895939e-01
3.74440253e-01 -3.72600332e-02 -9.05769408e-01 1.77173340e+00
3.81271303e-01 1.43050939e-01 -7.82742724e-02 1.20290506e+00
1.16607904e+00 4.55617368e-01 8.11637938e-02 1.64525524e-01
1.50241220e+00 -1.09248364e+00 -4.11022395e-01 -1.62152514e-01
3.58323514e-01 -7.94283569e-01 1.15137291e+00 3.30761045e-01
-9.45417345e-01 -8.67485762e-01 -9.37579989e-01 -2.36039981e-01
-6.51084423e-01 4.78319854e-01 2.57678360e-01 8.01257372e-01
-1.41582668e+00 -1.18460573e-01 -2.35207140e-01 -7.72308588e-01
5.94601221e-02 4.26897109e-01 -6.47549629e-01 -4.82647061e-01
-9.17363286e-01 9.20418859e-01 1.92834124e-01 1.05027087e-01
-4.21494544e-01 -6.83211446e-01 -8.60104680e-01 -4.00966048e-01
2.66420513e-01 -8.99619460e-01 1.07088041e+00 -6.57944500e-01
-1.09816790e+00 1.34095824e+00 -3.33667785e-01 -3.79051685e-01
1.10246646e+00 -2.74804950e-01 -6.40888214e-01 -4.00029570e-02
4.03211951e-01 1.05179346e+00 5.81186473e-01 -1.45868349e+00
-7.16689229e-01 -5.20305037e-01 -2.14402806e-02 3.03084046e-01
-2.17023596e-01 1.23360492e-01 -1.20814586e+00 -8.00969183e-01
-4.80836421e-01 -1.27401519e+00 2.01453939e-01 -2.64889076e-02
-5.55955946e-01 -3.10082853e-01 6.47199333e-01 -9.44781959e-01
5.65733731e-01 -2.01240253e+00 4.82017500e-03 4.97583784e-02
1.75516248e-01 3.73108268e-01 -3.97628367e-01 -7.39493640e-04
9.34648663e-02 2.82979403e-02 6.28452376e-02 -9.62672830e-01
1.02772035e-01 -9.53170285e-02 1.18732721e-01 5.93939900e-01
-2.78074235e-01 9.15756404e-01 -6.69564307e-01 -3.99180174e-01
2.31867000e-01 5.52677155e-01 -2.02975065e-01 2.37737715e-01
6.45333171e-01 4.99659240e-01 -3.54009792e-02 7.26726830e-01
1.16796148e+00 7.45403022e-02 -2.30086669e-01 -2.98009396e-01
-1.04990296e-01 -4.42763120e-01 -1.09460616e+00 1.54571676e+00
-3.16679984e-01 6.09122872e-01 1.29532978e-01 -6.04647100e-01
8.63133430e-01 -6.45466223e-02 1.43308923e-01 -1.06566536e+00
3.11741363e-02 -3.57722268e-02 -3.69214803e-01 -4.69005555e-01
7.58138955e-01 2.57408202e-01 -1.02988817e-01 1.48044959e-01
-1.19311728e-01 3.69029760e-01 4.86175805e-01 2.72484869e-01
7.00025916e-01 -1.20573923e-01 -1.60552546e-01 -3.06204468e-01
5.56007028e-01 -2.51639843e-01 5.75018585e-01 9.12441313e-01
-4.99005884e-01 9.68100309e-01 -2.10112020e-01 -4.25027549e-01
-1.00429976e+00 -1.34810352e+00 -2.14985773e-01 9.38367486e-01
7.09228694e-01 -2.28643984e-01 -7.97799528e-01 -6.74052894e-01
1.79113507e-01 5.61799288e-01 -5.95621943e-01 -4.14743200e-02
-5.72613358e-01 -9.11517382e-01 5.31309664e-01 3.52690846e-01
1.25201452e+00 -7.41483152e-01 -2.78816409e-02 -3.76676738e-01
-3.34383518e-01 -1.30445707e+00 -1.13508356e+00 -7.87425995e-01
-5.04085608e-02 -1.16958606e+00 -1.39887691e+00 -8.43139291e-01
9.36944187e-01 7.23375499e-01 1.09186351e+00 2.00776920e-01
-4.19656664e-01 7.72965252e-01 -3.21277440e-01 -3.89294744e-01
1.99215204e-01 1.04242079e-01 5.11595070e-01 3.61005008e-01
5.34110010e-01 -7.18582645e-02 -9.65884805e-01 7.30626285e-01
-4.28792000e-01 2.35682487e-01 3.96324992e-01 7.33773112e-01
3.90544593e-01 -3.55933189e-01 4.47876394e-01 -5.87220728e-01
4.71265018e-01 -4.27222937e-01 -6.24145389e-01 4.15920496e-01
-7.86970377e-01 -3.40019614e-01 3.87478501e-01 -6.79810822e-01
-1.04897761e+00 -2.12666839e-01 3.03770173e-02 -5.18400490e-01
-1.92011565e-01 -1.00543275e-01 -2.95083284e-01 -1.68812633e-01
5.06078362e-01 4.01813686e-01 -2.32734039e-01 -5.36125302e-01
1.76697209e-01 9.14351225e-01 1.05790806e+00 -5.75237691e-01
1.16577923e+00 4.12844419e-01 -4.74141687e-01 -6.89670324e-01
-3.86933804e-01 -6.17945552e-01 -4.76333171e-01 -2.38724977e-01
9.30093944e-01 -1.18275726e+00 -9.61790979e-01 8.94858956e-01
-1.03324163e+00 -1.64749220e-01 2.05525339e-01 4.60294306e-01
-3.88243496e-02 5.72894096e-01 -5.83472073e-01 -7.06972480e-01
-7.51021147e-01 -1.13900113e+00 9.86672044e-01 5.93846023e-01
-1.52685389e-01 -5.45487881e-01 -2.04126015e-01 8.46323371e-01
4.48828220e-01 7.18834857e-03 2.65372247e-01 -3.75409633e-01
-4.29161608e-01 -2.97948390e-01 -6.79314435e-01 2.62651891e-01
3.43840271e-02 -3.50517869e-01 -9.39727664e-01 -7.35092163e-01
-4.36972529e-01 -4.05855700e-02 8.07423055e-01 1.69130147e-01
1.05883014e+00 -2.90853471e-01 -3.31082761e-01 8.94614398e-01
1.17070556e+00 2.54283071e-01 6.16197467e-01 4.99899030e-01
9.56233919e-01 6.10298514e-01 4.14389908e-01 2.35401362e-01
1.05670393e+00 9.40674961e-01 5.90461008e-02 -1.46763384e-01
-4.17909890e-01 -1.47644699e-01 3.99000853e-01 5.81417263e-01
-3.11197728e-01 -1.51232496e-01 -9.47521508e-01 6.81026816e-01
-1.75655282e+00 -8.03313017e-01 -2.01394141e-01 2.40808225e+00
7.10619271e-01 -2.19332725e-01 4.09991205e-01 -2.22458616e-01
1.09865355e+00 -3.33809018e-01 -6.55289650e-01 -4.68336453e-04
-2.72645146e-01 -1.09114818e-01 7.40264595e-01 3.23055536e-01
-1.16684282e+00 9.39886272e-01 4.65010691e+00 6.62861884e-01
-1.06983781e+00 1.35031119e-01 6.23754740e-01 -3.03265899e-01
1.93710685e-01 -5.12237728e-01 -1.05157852e+00 8.67475629e-01
6.24871135e-01 -2.64008701e-01 6.84402406e-01 5.81107497e-01
1.58747169e-03 -8.88598785e-02 -1.10623491e+00 1.91009986e+00
4.84993160e-01 -8.14822674e-01 -1.71425007e-02 7.56814098e-03
6.45666361e-01 1.83283817e-02 5.01622677e-01 5.55580616e-01
1.61791325e-01 -8.70478213e-01 8.37184489e-01 5.18287301e-01
1.04219842e+00 -7.39599288e-01 8.28831017e-01 -8.54476541e-03
-1.14834523e+00 2.14876220e-01 -4.02121156e-01 2.80345261e-01
-9.83106419e-02 1.13239482e-01 -6.41244948e-01 7.35273719e-01
1.31648862e+00 7.58479893e-01 -1.08875668e+00 1.23049533e+00
-4.38588002e-04 1.50707513e-01 -2.53334939e-01 7.28980638e-03
-2.71133870e-01 -2.72382468e-01 4.45257962e-01 1.41251409e+00
4.28419858e-01 1.32149905e-02 1.27973408e-01 5.55863798e-01
-3.71358186e-01 1.71705317e-02 -4.27332580e-01 7.03916728e-01
3.88900310e-01 1.14573085e+00 -2.68831134e-01 -4.27027285e-01
-4.64700222e-01 1.37810493e+00 1.95053995e-01 7.99749017e-01
-1.01208329e+00 -5.14428437e-01 7.62438595e-01 1.29806623e-01
1.10718891e-01 -9.22651216e-02 -5.00355102e-02 -1.56507456e+00
2.99196124e-01 -8.62507582e-01 4.63060051e-01 -9.88049567e-01
-1.66131794e+00 6.00611269e-01 2.21322849e-01 -1.18344510e+00
7.14680627e-02 -6.31764770e-01 -3.79051656e-01 1.07554400e+00
-1.52313638e+00 -1.41368759e+00 -7.48967350e-01 9.02703464e-01
5.09554625e-01 -3.60609174e-01 5.50771892e-01 6.46024168e-01
-9.67265010e-01 1.38115072e+00 2.08724260e-01 5.25851607e-01
1.30502450e+00 -1.24801028e+00 5.26220381e-01 9.06585991e-01
-9.30637494e-02 6.83136404e-01 7.30934620e-01 -4.87193555e-01
-1.63220584e+00 -1.23463416e+00 7.65498042e-01 -6.32722557e-01
1.23319976e-01 -5.53195477e-01 -8.21878731e-01 6.89898133e-01
2.69796073e-01 1.74040310e-02 3.87503922e-01 1.64360732e-01
-8.11946332e-01 -5.74684203e-01 -1.52319360e+00 9.21093524e-01
1.46069646e+00 -4.31941688e-01 -2.71246493e-01 3.07733387e-01
3.95773172e-01 -5.65290391e-01 -8.67663205e-01 1.38058931e-01
7.15775609e-01 -7.11374760e-01 1.23586154e+00 -1.53895631e-01
-1.61576882e-01 -4.49283034e-01 5.25792269e-03 -1.24142766e+00
-2.56414115e-01 -5.07374287e-01 2.22828761e-01 1.44159317e+00
8.11361820e-02 -1.08838117e+00 5.69863260e-01 9.68836367e-01
8.91768113e-02 -1.38430402e-01 -8.91601741e-01 -1.03268540e+00
3.73193026e-02 -2.49966040e-01 8.04352641e-01 9.27227020e-01
-4.79123890e-01 1.00814715e-01 -5.53110719e-01 2.85748005e-01
1.04408252e+00 -1.38291046e-01 1.34752429e+00 -9.55394447e-01
-1.80253297e-01 -4.84372675e-01 -4.31451440e-01 -9.96583641e-01
1.32543653e-01 -8.95251691e-01 -3.22978258e-01 -1.32006097e+00
6.04291618e-01 -4.32009220e-01 -3.50390822e-01 5.67271471e-01
-3.89750987e-01 5.22274077e-01 6.46263242e-01 3.58885586e-01
-6.06025815e-01 4.02969450e-01 1.07327986e+00 -6.40259266e-01
7.52940327e-02 -1.10257804e-01 -7.71694899e-01 1.91568464e-01
9.77259517e-01 -8.08458924e-02 -3.74505371e-01 -6.92031205e-01
-1.32910341e-01 -5.54843009e-01 6.89388514e-01 -1.09045088e+00
3.79853964e-01 2.26033911e-01 6.82468891e-01 -4.48137909e-01
4.87472385e-01 -5.24644017e-01 2.90893495e-01 2.29276791e-01
-6.46806657e-02 4.61722195e-01 3.69907320e-01 3.58194053e-01
1.00061461e-01 2.15099543e-01 6.11148357e-01 -7.76881948e-02
-9.28416610e-01 6.13682747e-01 1.85061350e-01 -9.82141937e-04
9.56399024e-01 -2.88850188e-01 -6.05860889e-01 -5.45587003e-01
-4.28044319e-01 7.09453166e-01 6.99396193e-01 7.80941725e-01
7.26371109e-01 -1.37864542e+00 -1.12595487e+00 1.23995833e-01
4.01882589e-01 -1.65996924e-01 4.97605443e-01 7.09995389e-01
-3.99920046e-01 3.01531702e-01 -2.45692834e-01 -5.18578827e-01
-1.62460804e+00 7.35945165e-01 4.24155086e-01 5.41121513e-02
-4.98340547e-01 7.43086994e-01 4.22351509e-01 -5.93533754e-01
3.66011769e-01 9.99586955e-02 2.50968244e-02 -2.61570811e-01
7.56190598e-01 5.48304141e-01 -7.95982629e-02 -1.00677693e+00
-4.25420463e-01 9.01833355e-01 -4.23816711e-01 -7.50372410e-02
9.19077098e-01 -2.77820081e-01 2.36015506e-02 -1.24745116e-01
1.40748119e+00 -1.57192171e-01 -1.16496301e+00 -3.36209476e-01
-2.64294356e-01 -5.97247660e-01 -4.23326254e-01 -1.08462489e+00
-1.21792150e+00 5.63022137e-01 1.16109753e+00 -1.83690205e-01
1.22755635e+00 -3.82808596e-02 8.48418653e-01 3.80526870e-01
7.36079574e-01 -9.71281111e-01 -3.00063938e-01 2.58972794e-01
1.18424654e+00 -1.52082825e+00 -9.63478237e-02 -2.01907218e-01
-5.42287409e-01 6.09088659e-01 9.19066370e-01 7.72789717e-02
3.62384468e-01 -7.39126876e-02 1.99653909e-01 2.39159897e-01
-1.12544268e-01 -2.11385325e-01 4.13176835e-01 8.67969573e-01
1.24494001e-01 3.42262357e-01 -2.10139632e-01 4.25585628e-01
-5.45039177e-01 -3.16056103e-01 3.04383039e-01 5.81637859e-01
2.95907170e-01 -1.14268279e+00 -6.80261552e-01 2.57836819e-01
-2.25436419e-01 6.66688085e-02 -4.42635834e-01 9.36726809e-01
2.95492977e-01 1.17207968e+00 5.06090261e-02 -5.80619693e-01
5.55536807e-01 -2.68478900e-01 5.14698446e-01 -2.50797451e-01
-5.79694808e-01 -2.42788523e-01 -1.76136233e-02 -3.05870116e-01
-2.48642370e-01 -6.62422657e-01 -9.51391697e-01 -8.15395117e-01
1.25126287e-01 3.50483432e-02 6.81929767e-01 4.95978385e-01
5.39581716e-01 -1.42229795e-01 3.53409976e-01 -9.07271802e-01
-1.67623177e-01 -9.28658843e-01 -3.98657143e-01 7.82565475e-01
3.21621299e-01 -5.74249446e-01 -1.78382844e-01 -2.53842305e-03] | [14.643767356872559, 0.923568069934845] |
4cf4ae84-8820-4ced-8ab3-8aab98e5108c | pipe-overflow-smashing-voice-authentication | 2202.02751 | null | https://arxiv.org/abs/2202.02751v2 | https://arxiv.org/pdf/2202.02751v2.pdf | Tubes Among Us: Analog Attack on Automatic Speaker Identification | Recent years have seen a surge in the popularity of acoustics-enabled personal devices powered by machine learning. Yet, machine learning has proven to be vulnerable to adversarial examples. A large number of modern systems protect themselves against such attacks by targeting artificiality, i.e., they deploy mechanisms to detect the lack of human involvement in generating the adversarial examples. However, these defenses implicitly assume that humans are incapable of producing meaningful and targeted adversarial examples. In this paper, we show that this base assumption is wrong. In particular, we demonstrate that for tasks like speaker identification, a human is capable of producing analog adversarial examples directly with little cost and supervision: by simply speaking through a tube, an adversary reliably impersonates other speakers in eyes of ML models for speaker identification. Our findings extend to a range of other acoustic-biometric tasks such as liveness detection, bringing into question their use in security-critical settings in real life, such as phone banking. | ['Kassem Fawaz', 'Nicolas Papernot', 'Ilia Shumailov', 'Mohammad Yaghini', 'Ali Shahin Shamsabadi', 'Yash Wani', 'Shimaa Ahmed'] | 2022-02-06 | null | null | null | null | ['speaker-identification'] | ['speech'] | [ 4.91019666e-01 3.78902465e-01 2.07354963e-01 -2.54096121e-01
-9.16681290e-01 -1.19466662e+00 5.47554910e-01 -1.90857425e-01
-2.70847648e-01 5.21852672e-01 -2.84094602e-01 -7.57684767e-01
3.88987243e-01 -5.27309597e-01 -8.18121970e-01 -6.97377801e-01
6.24808446e-02 1.60147548e-01 -1.71950400e-01 -1.11299232e-01
6.59065768e-02 6.56833351e-01 -1.31438327e+00 1.10367322e-02
3.44301671e-01 7.51811385e-01 -7.70118535e-01 9.84685361e-01
4.25778419e-01 4.11961794e-01 -1.21138418e+00 -8.80776525e-01
2.97083735e-01 -3.27154756e-01 -4.22821552e-01 -4.58351940e-01
5.78878939e-01 -5.60553491e-01 -2.46122375e-01 1.22362280e+00
7.60595322e-01 -2.92635649e-01 3.74042600e-01 -1.69391346e+00
-3.21946114e-01 6.68209016e-01 -1.74215198e-01 -4.06838506e-02
6.84642196e-01 5.61662316e-01 7.15991676e-01 -3.68969560e-01
-1.83636591e-01 1.24105668e+00 6.87738955e-01 1.13507199e+00
-1.30816686e+00 -1.22903538e+00 -3.01193267e-01 -4.41156834e-01
-1.29549444e+00 -1.10449314e+00 8.77167046e-01 -2.64187664e-01
6.07329786e-01 6.75908148e-01 1.87541157e-01 1.67893863e+00
3.25780399e-02 3.93319160e-01 1.03399181e+00 -4.74966228e-01
2.47036576e-01 7.23681331e-01 -1.23680823e-01 4.40696478e-01
3.65764320e-01 4.38365489e-01 -6.60865605e-01 -8.34002554e-01
3.23580086e-01 -4.54308212e-01 -2.39432111e-01 1.69441581e-01
-9.95556116e-01 7.20753908e-01 -2.26300851e-01 2.56037831e-01
-3.33632715e-02 1.34741396e-01 4.52658206e-01 2.41267771e-01
9.26401913e-02 5.44068694e-01 -2.43156567e-01 -1.66215211e-01
-5.80426991e-01 1.39077716e-02 1.15327668e+00 7.39821255e-01
1.55072436e-01 3.76173943e-01 3.75250638e-01 3.84063721e-01
3.47392052e-01 9.29952383e-01 4.91432309e-01 -7.02937186e-01
3.15176189e-01 -8.78150463e-02 2.44246781e-01 -9.56904888e-01
4.01706025e-02 -9.98236015e-02 -5.15234709e-01 2.72898406e-01
6.10560298e-01 -4.67640460e-01 -4.55595851e-01 1.87501860e+00
4.04000372e-01 4.87046689e-01 6.74130470e-02 5.94125986e-01
2.19910622e-01 3.02654296e-01 1.85468465e-01 -9.86318961e-02
1.19325483e+00 -1.19913980e-01 -5.80478072e-01 -3.44930053e-01
3.70633155e-01 -7.93122053e-01 1.33089340e+00 6.25415027e-01
-9.20478225e-01 -4.12302196e-01 -1.35951471e+00 5.23173690e-01
-3.17727983e-01 -6.97206259e-01 6.77683234e-01 2.00138378e+00
-7.29019344e-01 2.53062665e-01 -5.28791130e-01 -1.58980653e-01
4.19733495e-01 6.15702868e-01 -3.80499482e-01 3.40629488e-01
-1.39909852e+00 5.98459482e-01 -2.01403484e-01 1.24890618e-02
-9.20066893e-01 -5.19128978e-01 -8.04861128e-01 -1.47394985e-01
-1.50959775e-01 -3.88110548e-01 1.35502338e+00 -1.17248988e+00
-1.56727612e+00 1.08483827e+00 4.54507433e-02 -6.52955651e-01
7.16702104e-01 -2.76855707e-01 -1.01188552e+00 2.29294207e-02
-1.36796206e-01 1.40063837e-01 1.49691403e+00 -1.22999251e+00
-2.27836818e-01 -4.66843635e-01 9.08945277e-02 -1.96513906e-01
-8.71417701e-01 4.08773392e-01 3.94549191e-01 -5.01940787e-01
-3.65132242e-01 -1.04225123e+00 1.79437026e-01 -1.80084497e-01
-9.27836776e-01 2.16369554e-01 1.15495968e+00 -3.73991162e-01
8.62665415e-01 -2.49469185e+00 -6.92817509e-01 5.42401135e-01
9.01834518e-02 5.62986553e-01 1.88901305e-01 7.82506689e-02
-1.72838569e-02 7.43107617e-01 -1.46533940e-02 -3.61511648e-01
2.69087493e-01 -4.62956466e-02 -8.59543204e-01 7.26428688e-01
-4.40566353e-02 7.34593213e-01 -7.36138701e-01 -3.20119858e-01
1.70340329e-01 6.18775487e-01 -2.97333807e-01 2.53232569e-01
2.16903731e-01 4.98926461e-01 -4.23239827e-01 7.54758596e-01
4.76194680e-01 3.19120497e-01 7.20250532e-02 2.00964183e-01
4.83999789e-01 3.83824706e-01 -9.43160117e-01 8.46098244e-01
-5.44489384e-01 5.85472167e-01 3.93388033e-01 -7.28036582e-01
5.45807898e-01 6.28625810e-01 6.67223632e-02 -1.63161576e-01
3.17278922e-01 2.58260310e-01 1.97659642e-01 -2.63245195e-01
1.27453849e-01 -3.67603391e-01 -3.20596993e-01 7.59899735e-01
-2.76328802e-01 -3.41824085e-01 -8.94505084e-01 1.40364796e-01
1.13977385e+00 -3.64626974e-01 1.04031235e-01 -2.51518805e-02
5.04484355e-01 -4.42234486e-01 1.90467432e-01 1.14997029e+00
-7.22194195e-01 2.36784443e-01 1.21593336e-02 4.11834866e-02
-8.94814610e-01 -1.36635280e+00 -8.90497789e-02 9.12637174e-01
4.20493856e-02 6.34353459e-02 -1.12135851e+00 -7.27929354e-01
2.91187942e-01 1.01838923e+00 -2.56404281e-01 -5.63137531e-01
-5.91928601e-01 -4.96978074e-01 1.56314242e+00 2.45236665e-01
2.24208176e-01 -9.69342113e-01 -4.34545666e-01 1.04211986e-01
9.24776495e-02 -1.30457735e+00 -5.50755739e-01 -8.05215240e-02
-5.27261138e-01 -8.72879505e-01 -2.26736143e-01 -4.64953035e-01
4.70354468e-01 1.30243171e-02 8.95867109e-01 2.35367581e-01
-4.78141487e-01 8.32950413e-01 -3.38342786e-02 -1.04717517e+00
-1.23429120e+00 -1.85128510e-01 6.75527275e-01 3.19827259e-01
6.13279164e-01 -6.81477427e-01 -4.15780425e-01 4.84571695e-01
-7.35318005e-01 -6.71293497e-01 8.97779465e-02 4.70746011e-01
-1.26148790e-01 1.68558449e-01 9.71502244e-01 -8.93199921e-01
8.59382629e-01 -3.67532283e-01 -3.07890415e-01 6.28759935e-02
-3.15545738e-01 -5.20628989e-01 8.03175867e-01 -9.05589581e-01
-6.54168785e-01 -8.58836398e-02 -3.66248846e-01 -1.56308219e-01
-5.94139934e-01 -3.07693094e-01 -6.85997605e-01 -4.90790009e-01
8.68152857e-01 2.44513229e-01 1.04881998e-03 -7.87903666e-02
1.76297054e-01 1.07032192e+00 6.85721576e-01 -5.01145720e-01
1.42051637e+00 4.70265657e-01 -2.53158510e-01 -1.15291703e+00
-4.01619315e-01 -6.26796186e-02 -9.98457745e-02 -3.93924415e-01
4.30176228e-01 -5.35454810e-01 -1.22639751e+00 8.96195650e-01
-9.56103683e-01 -1.32769182e-01 6.01679571e-02 2.01063931e-01
-3.31535935e-01 5.00892103e-01 -5.05572915e-01 -1.45631564e+00
-3.51915538e-01 -8.82043242e-01 9.04287577e-01 1.22528620e-01
-5.74160099e-01 -8.13546240e-01 -2.57747471e-01 7.63954461e-01
5.72885215e-01 2.08000645e-01 6.62341893e-01 -1.18567133e+00
-8.40316564e-02 -9.65985894e-01 6.59980237e-01 5.98008215e-01
4.16548222e-01 5.67876995e-02 -1.78843701e+00 -3.05312574e-01
3.72881174e-01 -4.28677529e-01 5.14498837e-02 -5.42341694e-02
1.10268819e+00 -7.14245081e-01 -2.66566634e-01 4.30280387e-01
8.26135874e-01 4.15752977e-01 6.25903666e-01 -3.33280265e-02
5.76968133e-01 7.48495638e-01 1.42063349e-01 5.63035384e-02
-2.92857885e-01 5.19659162e-01 4.06069905e-01 -1.40083898e-02
4.73263234e-01 -4.41600651e-01 6.71365023e-01 1.43720105e-01
3.44032794e-01 -3.70383173e-01 -7.35565484e-01 1.92975700e-01
-1.06836009e+00 -1.07402182e+00 2.44012445e-01 2.58868837e+00
8.78990352e-01 6.22417510e-01 3.62114757e-01 6.26073658e-01
1.03758740e+00 1.42689003e-02 -7.87126482e-01 -6.59531415e-01
-2.64670625e-02 2.89872140e-01 5.58562756e-01 5.47670364e-01
-1.14331913e+00 8.69863451e-01 6.65838861e+00 4.73018080e-01
-1.05361140e+00 2.20029712e-01 9.17138219e-01 -1.88850626e-01
-2.31158331e-01 -3.71143341e-01 -6.01013422e-01 6.60662532e-01
1.28682435e+00 -1.29356369e-01 4.73008543e-01 8.06250334e-01
1.10617921e-01 2.95750797e-01 -1.33516908e+00 8.55504811e-01
1.21045932e-01 -7.32216239e-01 -1.21462934e-01 3.00423235e-01
4.17803004e-02 -5.44447899e-01 5.56533754e-01 2.16078758e-01
4.33962375e-01 -1.31239641e+00 6.68039382e-01 5.02354354e-02
8.32835376e-01 -9.28144693e-01 5.08209050e-01 3.84929657e-01
-6.03012383e-01 3.14838476e-02 1.25249431e-01 3.78475115e-02
1.55495510e-01 4.65958953e-01 -9.59276974e-01 -1.42395407e-01
3.97307783e-01 -4.87720191e-01 -1.80447295e-01 4.04994011e-01
-2.18304679e-01 1.20685625e+00 -5.21408856e-01 -2.32230704e-02
-1.75235137e-01 2.69967526e-01 7.19503880e-01 1.05398631e+00
1.11874364e-01 -5.91364391e-02 -3.81481558e-01 7.46378303e-01
-1.73140466e-01 -2.34414175e-01 -1.05057859e+00 -3.22417438e-01
8.59429598e-01 9.93947268e-01 -1.82404026e-01 7.04877749e-02
-1.04446799e-01 1.11127031e+00 -4.67719465e-01 1.93849474e-01
-9.45014477e-01 -5.23678720e-01 8.86114597e-01 1.43918186e-01
-4.52785939e-01 -7.72093311e-02 -3.75115663e-01 -8.61384273e-01
4.96898703e-02 -1.35345423e+00 8.80219042e-02 -4.48721468e-01
-1.37506604e+00 3.17218333e-01 -4.87345874e-01 -8.85687888e-01
-4.99635398e-01 -4.01279002e-01 -8.48351300e-01 9.43243384e-01
-8.05887938e-01 -1.23722327e+00 1.20466799e-01 8.29775453e-01
6.84920847e-02 -3.41422886e-01 1.15080559e+00 2.11940214e-01
-4.26804036e-01 1.36173522e+00 -3.04690957e-01 5.64876318e-01
7.52758086e-01 -1.03496540e+00 8.83430481e-01 8.93799007e-01
3.38575482e-01 1.06924093e+00 1.01392460e+00 -4.42776918e-01
-1.56681478e+00 -7.60592520e-01 5.98288834e-01 -7.98405826e-01
7.65815020e-01 -9.05996561e-01 -7.62128830e-01 7.39278495e-01
-2.51209944e-01 -1.25840038e-01 1.06446719e+00 3.45706493e-02
-8.01084876e-01 -1.00979999e-01 -1.81267512e+00 6.36133194e-01
8.03090811e-01 -1.17762959e+00 -3.37849230e-01 3.94869298e-01
6.40821099e-01 -1.66734513e-02 -5.51940799e-01 8.63078088e-02
9.91280735e-01 -8.66852939e-01 1.18598127e+00 -7.06937730e-01
-2.69268394e-01 -3.78170200e-02 -2.59051323e-01 -8.28256905e-01
4.22505349e-01 -1.30665159e+00 -3.09258968e-01 1.61902010e+00
4.73274082e-01 -1.10107052e+00 8.41068864e-01 9.92970109e-01
3.72496039e-01 -1.66709825e-01 -1.08243895e+00 -9.05483723e-01
1.73855096e-01 -6.98895276e-01 7.53288865e-01 1.13479269e+00
1.90556943e-01 -9.35187470e-03 -5.39165020e-01 7.62318850e-01
8.99660170e-01 -6.51864529e-01 1.00938010e+00 -9.64660466e-01
-5.60634315e-01 -3.23046803e-01 -6.28521204e-01 -5.10507047e-01
2.58860230e-01 -2.58452564e-01 1.31747440e-01 -2.66316175e-01
-2.07771301e-01 -5.37942529e-01 -2.75928140e-01 1.77080035e-01
-1.67689577e-01 5.15693605e-01 4.47939858e-02 9.16234329e-02
7.50525594e-02 -1.59750417e-01 4.42408383e-01 -2.93733567e-01
-1.06239105e-02 5.30230105e-01 -1.04718876e+00 8.78528953e-01
1.09834397e+00 -4.94304955e-01 -3.91508132e-01 1.24983370e-01
2.89907623e-02 -1.85745358e-01 6.49350345e-01 -1.04217815e+00
1.43359020e-01 -4.49925959e-02 2.02841282e-01 2.47053564e-01
4.75459069e-01 -9.07659769e-01 3.20382114e-03 4.88873214e-01
-2.28739619e-01 -2.69593507e-01 2.53532052e-01 4.83598471e-01
2.01894149e-01 -1.70375526e-01 9.39085007e-01 9.77906585e-02
-3.68546098e-02 2.87433118e-02 -3.94090414e-01 2.08563302e-02
1.02653396e+00 -3.36392045e-01 -4.11161870e-01 -6.85808539e-01
-5.80116987e-01 -3.43919605e-01 4.77043539e-01 2.66444057e-01
4.54870999e-01 -8.26532423e-01 -4.51691747e-01 4.94651198e-01
-4.94305417e-02 -5.09141862e-01 -1.44463265e-02 3.73849571e-01
-1.22926503e-01 1.31404862e-01 3.15608203e-01 -4.46465194e-01
-1.69055927e+00 5.27823269e-01 6.67828202e-01 3.98108900e-01
-3.49621683e-01 1.05069661e+00 1.65154159e-01 -2.35752150e-01
4.85631555e-01 3.47640842e-01 3.78452539e-01 -4.29646075e-01
7.46860802e-01 6.55502006e-02 7.22547919e-02 -7.23734081e-01
-5.90126276e-01 2.11862773e-02 -3.07866279e-02 -4.40912336e-01
6.34156644e-01 1.56106189e-01 3.08934271e-01 3.33678424e-01
1.08424532e+00 5.90188444e-01 -8.89354765e-01 1.45416558e-01
-1.41579285e-01 -4.81412441e-01 -2.96484709e-01 -8.13115537e-01
-6.41449749e-01 8.26056242e-01 7.11708248e-01 7.49997199e-01
8.90917361e-01 -9.92190540e-02 9.37884152e-01 4.92022902e-01
6.30744755e-01 -8.17615211e-01 -1.71496924e-02 -6.26907125e-02
4.48052526e-01 -1.21100831e+00 -3.69367987e-01 -3.25396121e-01
-4.76894408e-01 6.52808607e-01 3.97611380e-01 1.76987350e-01
6.42608106e-01 5.85597992e-01 3.08776498e-01 7.89024681e-02
-1.90348819e-01 5.94374418e-01 -1.33091703e-01 1.10163021e+00
1.31058499e-01 3.63531619e-01 3.87309790e-01 6.42486215e-01
-6.97165668e-01 -5.63731015e-01 4.85678107e-01 9.27117169e-01
-3.02810907e-01 -1.03625143e+00 -8.47333610e-01 3.42670292e-01
-9.71618056e-01 -1.41667500e-01 -8.56479943e-01 5.96044362e-01
2.93996483e-02 1.48269498e+00 -1.39091775e-01 -6.29997730e-01
2.18136922e-01 3.88593853e-01 2.88181990e-01 -5.18689632e-01
-9.10040140e-01 -3.23380023e-01 1.11011855e-01 -2.73782343e-01
-2.17729911e-01 -7.57862628e-01 -7.95002282e-01 -6.20151997e-01
-3.74687493e-01 5.78153841e-02 8.87987971e-01 8.80090475e-01
7.96754062e-02 -2.06819177e-02 1.05388033e+00 -5.58578432e-01
-1.04172683e+00 -7.68449783e-01 -7.18847632e-01 4.29148406e-01
6.38422549e-01 -2.11219981e-01 -1.09413671e+00 -1.55511983e-02] | [13.896913528442383, 5.80698823928833] |
23a55474-b9d4-4b1a-a8ad-47c0bf5a98d0 | cmw-net-learning-a-class-aware-sample | 2202.05613 | null | https://arxiv.org/abs/2202.05613v3 | https://arxiv.org/pdf/2202.05613v3.pdf | CMW-Net: Learning a Class-Aware Sample Weighting Mapping for Robust Deep Learning | Modern deep neural networks can easily overfit to biased training data containing corrupted labels or class imbalance. Sample re-weighting methods are popularly used to alleviate this data bias issue. Most current methods, however, require to manually pre-specify the weighting schemes as well as their additional hyper-parameters relying on the characteristics of the investigated problem and training data. This makes them fairly hard to be generally applied in practical scenarios, due to their significant complexities and inter-class variations of data bias situations. To address this issue, we propose a meta-model capable of adaptively learning an explicit weighting scheme directly from data. Specifically, by seeing each training class as a separate learning task, our method aims to extract an explicit weighting function with sample loss and task/class feature as input, and sample weight as output, expecting to impose adaptively varying weighting schemes to different sample classes based on their own intrinsic bias characteristics. Synthetic and real data experiments substantiate the capability of our method on achieving proper weighting schemes in various data bias cases, like the class imbalance, feature-independent and dependent label noise scenarios, and more complicated bias scenarios beyond conventional cases. Besides, the task-transferability of the learned weighting scheme is also substantiated, by readily deploying the weighting function learned on relatively smaller-scale CIFAR-10 dataset on much larger-scale full WebVision dataset. A performance gain can be readily achieved compared with previous SOAT ones without additional hyper-parameter tuning and meta gradient descent step. The general availability of our method for multiple robust deep learning issues, including partial-label learning, semi-supervised learning and selective classification, has also been validated. | ['Zongben Xu', 'Deyu Meng', 'Xiang Yuan', 'Jun Shu'] | 2022-02-11 | null | null | null | null | ['partial-label-learning'] | ['methodology'] | [ 4.42020059e-01 4.68495488e-03 -3.79331797e-01 -1.00991499e+00
-7.13296950e-01 -3.61450166e-01 3.88406128e-01 2.54931487e-02
-7.88911581e-01 8.18643630e-01 -1.42810956e-01 -1.04953155e-01
-3.89764577e-01 -6.55955255e-01 -5.16120493e-01 -9.15236294e-01
1.54212952e-01 6.13901675e-01 1.12185568e-01 -2.01814592e-01
2.01142743e-01 5.19969821e-01 -1.65941787e+00 1.27635106e-01
8.78062308e-01 1.20424700e+00 -3.06089222e-02 2.43277416e-01
-1.87630337e-02 3.14145535e-01 -6.66880131e-01 -4.29447144e-01
2.90140361e-01 -1.40699357e-01 -6.65444553e-01 1.59360275e-01
7.07290590e-01 -3.11820775e-01 1.69374660e-01 1.07296813e+00
8.69744122e-01 5.06562106e-02 8.45791996e-01 -1.41855264e+00
-4.24453497e-01 6.39985263e-01 -6.33559883e-01 9.62209553e-02
-2.48845145e-01 2.91051984e-01 1.03295732e+00 -7.59415984e-01
1.76727414e-01 1.09573758e+00 7.73506582e-01 7.52669930e-01
-1.44038558e+00 -1.01233935e+00 4.26115453e-01 1.40114531e-01
-1.21515357e+00 -2.65191972e-01 1.04939830e+00 -3.55349809e-01
4.24459457e-01 1.06606409e-01 2.56531119e-01 1.45086896e+00
-1.82057381e-01 6.69585705e-01 1.21690381e+00 -3.45386773e-01
3.71422946e-01 5.37394524e-01 4.60744858e-01 3.15262616e-01
3.71612221e-01 1.78430647e-01 -2.33720183e-01 -2.33088493e-01
3.25215071e-01 -5.98635823e-02 -3.59596997e-01 -7.45168507e-01
-1.00265205e+00 7.40564704e-01 5.21898329e-01 2.00226992e-01
-1.56858519e-01 1.66155249e-01 6.45507693e-01 4.00739998e-01
6.26265466e-01 3.55231613e-01 -7.71494508e-01 2.90063471e-01
-7.88239360e-01 4.43045735e-01 4.38122094e-01 7.83658504e-01
8.31279933e-01 1.79591253e-01 -4.28450435e-01 1.26467097e+00
2.76171714e-01 4.54276472e-01 8.14297736e-01 -6.86371565e-01
6.15042329e-01 5.90032279e-01 2.10599080e-01 -8.77643168e-01
-8.27500761e-01 -9.15230691e-01 -1.05724835e+00 3.77188802e-01
7.11552083e-01 -2.49780416e-01 -9.75869656e-01 2.13216734e+00
6.36442780e-01 -2.87111551e-01 -8.48598182e-02 1.07793057e+00
6.41025901e-01 1.44976899e-01 1.09561108e-01 -1.01073757e-01
1.44408274e+00 -6.04553103e-01 -5.39224923e-01 -3.02230179e-01
6.45725131e-01 -3.26032996e-01 1.50671113e+00 5.04142523e-01
-7.50420332e-01 -6.65784538e-01 -1.14607942e+00 1.80220276e-01
-3.53501171e-01 2.72267342e-01 6.06567264e-01 8.75991285e-01
-6.27177596e-01 5.82684994e-01 -4.41234052e-01 -7.34189078e-02
8.28588665e-01 6.30472660e-01 -1.13644153e-01 5.62422946e-02
-1.34004390e+00 6.33777618e-01 5.19870758e-01 2.10299596e-01
-8.47305655e-01 -9.68093574e-01 -5.68769217e-01 1.30838752e-01
3.62162232e-01 -5.21198630e-01 1.17758179e+00 -1.50942278e+00
-1.51258123e+00 9.82159317e-01 3.93710136e-01 -3.55208695e-01
7.89002419e-01 -6.53351173e-02 -4.35160965e-01 -3.40391248e-01
-2.07417116e-01 6.80370092e-01 1.12329876e+00 -1.35085714e+00
-5.63657939e-01 -4.97035772e-01 2.15731725e-01 1.15511030e-01
-7.20030248e-01 -2.81375736e-01 5.25955670e-02 -8.33306134e-01
3.19604687e-02 -7.17015564e-01 -1.56779230e-01 1.08873539e-01
-2.72382319e-01 -2.38044232e-01 4.81962085e-01 -2.57247239e-02
1.08935034e+00 -2.11639690e+00 -1.51584163e-01 1.96534112e-01
1.17601313e-01 4.27553356e-01 -3.25984061e-01 -9.96336713e-02
-3.36688519e-01 -7.78568536e-02 -4.47367549e-01 4.51898240e-02
1.32554442e-01 1.07970662e-01 -2.13782936e-01 5.60079336e-01
4.06995565e-01 5.07029474e-01 -8.20434391e-01 -1.70335382e-01
-1.16236486e-01 4.03286457e-01 -7.53194809e-01 2.04669148e-01
-5.50832972e-02 3.00283253e-01 -4.22928542e-01 4.82339054e-01
8.42071235e-01 -1.69530928e-01 1.87818572e-01 -5.46411157e-01
3.68629783e-01 9.29665342e-02 -1.46358311e+00 1.21206975e+00
-6.60515726e-01 2.47657493e-01 8.98101255e-02 -1.45346153e+00
1.10079646e+00 8.25416297e-02 2.30345100e-01 -7.41147339e-01
3.20163190e-01 3.75744194e-01 2.35833675e-01 -4.88900304e-01
2.39212006e-01 -4.31076616e-01 -5.25154453e-03 4.70058173e-01
3.17421675e-01 8.74412358e-02 4.05965038e-02 -3.36227834e-01
5.67336321e-01 1.21887714e-01 1.56383470e-01 -4.58858103e-01
6.49188876e-01 -3.00265163e-01 5.78313172e-01 7.99943745e-01
-3.54528338e-01 7.07562447e-01 4.29332197e-01 -6.08518004e-01
-9.68307018e-01 -6.20278478e-01 -6.95720553e-01 1.41777229e+00
5.08231074e-02 2.83097088e-01 -7.10331380e-01 -1.09567881e+00
1.26247093e-01 6.77493036e-01 -7.27317929e-01 -4.51753825e-01
-5.03228664e-01 -1.47832096e+00 4.86089617e-01 5.61011136e-01
4.92691189e-01 -9.33339596e-01 -5.50190091e-01 2.54025787e-01
1.36524200e-01 -7.71648109e-01 -2.84575433e-01 7.39620686e-01
-9.07561779e-01 -9.76846337e-01 -7.45459199e-01 -5.69073737e-01
8.37346315e-01 9.54649597e-02 1.15654671e+00 1.13493092e-01
-2.65939325e-01 9.12971348e-02 -6.95710629e-02 -4.10818517e-01
-2.39144713e-01 3.64164174e-01 1.47645056e-01 3.02836984e-01
3.12703162e-01 -6.60033107e-01 -6.56649113e-01 6.51390195e-01
-1.15813327e+00 -2.13508606e-01 3.90450120e-01 1.24663925e+00
3.39241028e-01 6.92313211e-03 1.17836142e+00 -1.26353002e+00
5.72960496e-01 -5.06986976e-01 -7.09593773e-01 2.03300983e-01
-9.43360209e-01 2.20055416e-01 6.59184277e-01 -8.38869750e-01
-1.05690670e+00 -1.29503235e-01 -2.72304326e-01 -1.35595873e-02
-2.94000238e-01 1.76306963e-01 -5.58710992e-01 1.22356273e-01
1.00991762e+00 -1.34004593e-01 -6.23431168e-02 -4.43013549e-01
2.33198807e-01 8.24931085e-01 1.44888207e-01 -8.90654862e-01
7.38071859e-01 4.38798398e-01 -1.13030888e-01 -5.64155281e-02
-1.22104609e+00 -1.18112504e-01 -6.93516672e-01 -2.10451074e-02
4.49865252e-01 -8.35454285e-01 -5.66443324e-01 7.42443264e-01
-7.42949188e-01 -5.38242757e-01 -3.64917278e-01 3.54384542e-01
-3.39170665e-01 -1.99128501e-03 -2.14848280e-01 -6.64689064e-01
-2.94091225e-01 -1.13916421e+00 8.18454981e-01 -1.82442442e-02
-8.32923427e-02 -1.06548035e+00 -1.23196214e-01 3.65427136e-01
6.33983433e-01 1.05752058e-01 1.14620173e+00 -9.52283204e-01
4.70250547e-02 -1.74003348e-01 -4.09060031e-01 7.10981548e-01
2.31437728e-01 -1.91504851e-01 -1.36699057e+00 -5.54428637e-01
-7.90232345e-02 -7.93515503e-01 9.01476324e-01 2.14251101e-01
1.55882156e+00 -2.80814856e-01 -3.59051935e-02 7.01167822e-01
1.38229346e+00 -2.43846141e-02 1.88046902e-01 4.17177975e-01
5.89817882e-01 7.09555328e-01 5.54482102e-01 3.37545455e-01
7.81020448e-02 7.74712265e-01 6.75697446e-01 -1.38567060e-01
-1.20937720e-01 6.94128871e-02 4.49592173e-02 3.28028589e-01
1.88795909e-01 -2.45043859e-01 -7.90945530e-01 3.66602540e-01
-1.65085709e+00 -7.03023076e-01 2.12453809e-02 2.35486794e+00
1.18383002e+00 3.35674375e-01 1.45102710e-01 4.71699983e-01
7.02979624e-01 1.70900673e-01 -1.04080212e+00 -1.35494962e-01
2.82462202e-02 -4.76637762e-03 5.46924889e-01 2.79188305e-01
-1.22267294e+00 5.98725080e-01 5.92046213e+00 1.05975485e+00
-1.46101427e+00 1.23434417e-01 9.86895919e-01 -2.70902336e-01
-4.63683993e-01 -4.73793536e-01 -1.01834536e+00 4.15657312e-01
7.55940378e-01 2.05421448e-01 3.72689873e-01 9.64827478e-01
-9.54862759e-02 1.20596476e-01 -1.24530089e+00 8.94411743e-01
-3.16675007e-02 -8.38520288e-01 5.86286932e-02 -1.04558982e-01
5.54773390e-01 -6.50543943e-02 2.34584317e-01 6.52691364e-01
1.96212173e-01 -9.37989116e-01 8.34367037e-01 1.45635679e-01
8.18200588e-01 -7.34775186e-01 8.20379436e-01 3.76917571e-01
-3.53737354e-01 -4.29985285e-01 -5.89089513e-01 -3.99687700e-02
-2.34043375e-01 1.02562773e+00 -7.48030007e-01 2.81454474e-01
7.46899545e-01 3.39334995e-01 -6.32395267e-01 8.33382964e-01
-2.31134407e-02 8.27116072e-01 -2.57847190e-01 -1.68633536e-02
-2.74708141e-02 5.14756329e-02 1.59442157e-01 1.09666228e+00
1.95162266e-01 -2.68990397e-01 -5.32356277e-02 7.64508665e-01
-2.33697042e-01 1.65846318e-01 -2.41843998e-01 4.29670960e-01
3.40590030e-01 1.55538905e+00 -4.86755520e-01 -2.49923378e-01
-1.19997315e-01 5.25070071e-01 5.81433237e-01 4.57471848e-01
-1.01543057e+00 -3.72060686e-01 6.29935265e-01 1.34013593e-01
3.20685416e-01 3.29357922e-01 -4.31933910e-01 -1.19223213e+00
-5.81538565e-02 -1.15496719e+00 4.28623050e-01 -4.40027863e-01
-1.59367061e+00 6.33810043e-01 1.90722615e-01 -1.35604715e+00
1.61828957e-02 -8.53784204e-01 -3.63001913e-01 7.36914873e-01
-1.78400767e+00 -8.76696765e-01 -3.71073455e-01 4.98855412e-01
3.44020694e-01 -3.03853691e-01 5.78934252e-01 5.15716910e-01
-7.57678986e-01 9.76832271e-01 5.80976643e-02 2.53816899e-02
1.06341147e+00 -1.20662546e+00 -1.45322129e-01 3.52768391e-01
-6.26939237e-02 4.46960628e-01 6.27111375e-01 3.10799461e-02
-9.13903832e-01 -1.24056876e+00 4.52527404e-01 -3.14164221e-01
5.62511027e-01 -6.43741786e-01 -1.08709323e+00 3.12580556e-01
-1.27497450e-01 3.90647292e-01 7.53501892e-01 3.12007338e-01
-6.27126455e-01 -8.58110249e-01 -1.44536579e+00 3.92294765e-01
1.10441959e+00 -1.25411287e-01 -3.85135144e-01 3.76968533e-01
4.82916534e-01 -4.40838248e-01 -7.53673553e-01 8.40589285e-01
5.78424215e-01 -9.58460689e-01 9.77218807e-01 -8.13151240e-01
3.22142810e-01 -1.57176003e-01 -2.24220410e-01 -1.49257088e+00
-4.17956263e-01 4.33579534e-02 1.82702735e-01 1.52238476e+00
5.02294958e-01 -7.99507737e-01 7.19023347e-01 6.19844913e-01
1.24794610e-01 -8.56412113e-01 -9.57674146e-01 -5.70509851e-01
2.32462794e-01 -6.23356164e-01 8.88516068e-01 1.12584651e+00
-4.85641420e-01 3.46913099e-01 -3.99123549e-01 1.43714830e-01
6.60895586e-01 7.59227276e-02 8.13039601e-01 -1.43600047e+00
-3.33648264e-01 -5.43207884e-01 7.66714438e-05 -7.55652547e-01
3.09352964e-01 -8.82427216e-01 6.80252239e-02 -7.75608957e-01
-2.50666384e-02 -9.60398078e-01 -6.95220411e-01 6.39809668e-01
-3.58589441e-01 3.99252564e-01 -8.64798725e-02 7.71042854e-02
-3.62977147e-01 7.20468342e-01 1.23288274e+00 -3.67449969e-01
1.32802809e-02 3.00322384e-01 -9.04989898e-01 6.88207150e-01
8.30590606e-01 -6.93063974e-01 -6.90246165e-01 -4.16978031e-01
3.82722020e-01 -4.05650198e-01 4.03432488e-01 -9.02343273e-01
-2.06372678e-01 -3.18507761e-01 4.44008559e-01 -1.23635493e-01
3.85206938e-02 -1.09249258e+00 -1.83388606e-01 3.33046019e-01
-7.67091155e-01 -3.69392335e-01 1.52134508e-01 5.39934516e-01
-9.25709754e-02 -5.78029156e-01 1.01226461e+00 7.77106285e-02
-2.97832876e-01 2.62434453e-01 -9.76414084e-02 3.39993566e-01
7.22839892e-01 -1.41651660e-01 -1.78496733e-01 -3.10180783e-02
-4.70703810e-01 2.70362407e-01 1.67202786e-01 3.95282269e-01
1.00386292e-01 -1.49571049e+00 -6.92306042e-01 4.76337910e-01
4.13994431e-01 1.30515888e-01 3.20244342e-01 7.69529700e-01
9.18657035e-02 -1.19840843e-03 -2.15301484e-01 -7.50867009e-01
-9.04845476e-01 6.60100460e-01 5.58473885e-01 -3.14790517e-01
-1.88065961e-01 6.81552947e-01 3.49915802e-01 -8.79443944e-01
5.07231951e-01 -5.12876034e-01 -2.84812331e-01 3.21717143e-01
4.86618608e-01 2.47254133e-01 3.69247913e-01 -2.18264431e-01
-2.73803353e-01 5.33918083e-01 -1.52723908e-01 2.83976823e-01
1.31753993e+00 -1.39421329e-01 2.63898551e-01 4.12395954e-01
1.18732893e+00 -2.86895245e-01 -1.51045299e+00 -4.61364895e-01
8.31509829e-02 -2.79511452e-01 1.06015421e-01 -9.99708891e-01
-1.46012115e+00 1.01048183e+00 9.54449832e-01 1.29528537e-01
1.21167421e+00 -4.52263385e-01 3.86400074e-01 3.97103220e-01
3.85574162e-01 -1.28497505e+00 6.27876222e-02 2.75204152e-01
8.95000935e-01 -1.59399450e+00 -3.97259295e-02 1.26815494e-03
-5.40056646e-01 1.11410546e+00 7.84830153e-01 -1.14201576e-01
6.75727308e-01 1.49633646e-01 4.49716747e-01 2.01467834e-02
-7.26321101e-01 1.37359202e-01 4.29239988e-01 6.34412110e-01
5.71856022e-01 -1.74104378e-01 -4.38346446e-01 7.46483803e-01
1.02353066e-01 3.82757522e-02 2.20198110e-01 4.35068488e-01
-2.32915610e-01 -1.14898121e+00 -3.49091470e-01 4.46339011e-01
-4.55532283e-01 1.06089324e-01 1.06271364e-01 7.91528344e-01
3.36735159e-01 4.74130541e-01 5.71322106e-02 -1.86764404e-01
4.98123258e-01 1.68802798e-01 2.69423038e-01 -5.00565886e-01
-8.90066803e-01 -1.90914199e-01 -3.86265665e-02 -3.43633443e-01
-6.31296098e-01 -3.58590007e-01 -1.01879823e+00 1.90016150e-01
-4.75571036e-01 -3.33897993e-02 5.92975378e-01 8.51580083e-01
2.11008474e-01 5.64118147e-01 7.51389205e-01 -9.94927227e-01
-1.28320789e+00 -1.09006846e+00 -5.15205920e-01 7.50157654e-01
4.90346789e-01 -9.37649608e-01 -6.16397262e-01 -3.09442610e-01] | [9.230220794677734, 3.920546770095825] |
12f1e9d4-77b8-470b-87ac-31e7b5ea63c4 | unified-conversational-models-with-system | 2307.01664 | null | https://arxiv.org/abs/2307.01664v1 | https://arxiv.org/pdf/2307.01664v1.pdf | Unified Conversational Models with System-Initiated Transitions between Chit-Chat and Task-Oriented Dialogues | Spoken dialogue systems (SDSs) have been separately developed under two different categories, task-oriented and chit-chat. The former focuses on achieving functional goals and the latter aims at creating engaging social conversations without special goals. Creating a unified conversational model that can engage in both chit-chat and task-oriented dialogue is a promising research topic in recent years. However, the potential ``initiative'' that occurs when there is a change between dialogue modes in one dialogue has rarely been explored. In this work, we investigate two kinds of dialogue scenarios, one starts from chit-chat implicitly involving task-related topics and finally switching to task-oriented requests; the other starts from task-oriented interaction and eventually changes to casual chat after all requested information is provided. We contribute two efficient prompt models which can proactively generate a transition sentence to trigger system-initiated transitions in a unified dialogue model. One is a discrete prompt model trained with two discrete tokens, the other one is a continuous prompt model using continuous prompt embeddings automatically generated by a classifier. We furthermore show that the continuous prompt model can also be used to guide the proactive transitions between particular domains in a multi-domain task-oriented setting. | ['Wolfgang Maier', 'Wolfgang Minker', 'Stefan Ultes', 'Ye Liu'] | 2023-07-04 | null | null | null | null | ['spoken-dialogue-systems'] | ['speech'] | [ 4.50524181e-01 7.08878338e-01 2.28356183e-01 -5.87051630e-01
-6.76507592e-01 -5.98292351e-01 1.28547227e+00 1.03425607e-01
-2.21944690e-01 9.32791352e-01 6.51686192e-01 -3.39896530e-01
1.83827475e-01 -5.50015926e-01 -2.93794647e-02 -6.21491075e-01
1.82591200e-01 9.70013976e-01 4.27821368e-01 -8.25332224e-01
3.09539020e-01 -7.43414909e-02 -1.17900407e+00 3.83898169e-01
9.83204484e-01 5.57806790e-01 2.45247170e-01 9.52634871e-01
-7.83762336e-01 1.14727509e+00 -8.69925082e-01 -2.01958567e-01
-1.02048762e-01 -8.86044621e-01 -1.48644876e+00 2.48386428e-01
-2.92323440e-01 -3.31857800e-01 -5.06046489e-02 5.62723398e-01
7.59798944e-01 4.04525042e-01 7.19428658e-01 -1.50759220e+00
-3.35678041e-01 8.39871168e-01 1.20025657e-01 -7.16784745e-02
8.65512252e-01 5.10458827e-01 9.24921095e-01 -4.70998645e-01
6.04874909e-01 1.37641811e+00 3.35794657e-01 1.01667428e+00
-1.05456638e+00 -1.96775377e-01 9.32958424e-02 -2.07330827e-02
-6.89277470e-01 -5.65556943e-01 1.03223765e+00 -5.66690505e-01
9.47924495e-01 2.80752957e-01 5.56981683e-01 1.50590229e+00
-6.32611662e-02 9.28330183e-01 1.15904939e+00 -5.94807565e-01
3.11474353e-01 5.32972217e-01 3.46471518e-01 1.63894538e-02
-6.92855597e-01 -2.41049379e-01 -5.61445832e-01 -2.47808680e-01
3.35963011e-01 -3.23271096e-01 -2.36411616e-01 1.79493725e-01
-1.29189825e+00 1.11638975e+00 -2.61344761e-01 7.92991281e-01
-3.43646020e-01 -3.95363659e-01 8.64206433e-01 7.49746501e-01
6.31062746e-01 7.28571296e-01 -3.26686472e-01 -8.33132446e-01
-4.66731727e-01 4.55203295e-01 1.48707533e+00 1.04459262e+00
5.09980977e-01 -8.12639296e-02 -7.62195528e-01 1.18643010e+00
1.60456955e-01 -5.34200780e-02 5.24446726e-01 -8.98102045e-01
6.18938327e-01 6.41130030e-01 5.13882697e-01 -4.38851148e-01
-4.42389995e-01 4.15100843e-01 -7.37915933e-01 -2.09747165e-01
6.04901731e-01 -8.10887635e-01 -1.55890450e-01 1.64448869e+00
4.90527064e-01 -1.71698526e-01 3.68329138e-01 5.97651422e-01
1.13897085e+00 9.12134767e-01 8.15085135e-03 -4.18773472e-01
1.46892333e+00 -1.24746132e+00 -1.18224287e+00 -9.37676206e-02
7.77563989e-01 -7.88830757e-01 1.18586957e+00 1.28502235e-01
-1.10112250e+00 -4.38824207e-01 -6.18714094e-01 -7.14076459e-02
-2.88748622e-01 -2.60910302e-01 2.82640100e-01 3.95675600e-01
-1.26971936e+00 1.88327491e-01 -2.16846615e-01 -7.00842202e-01
-3.05786252e-01 1.15760691e-01 -1.28672555e-01 4.22603786e-01
-1.69948411e+00 1.08162105e+00 -1.57249887e-02 -3.65672298e-02
-6.10858083e-01 -3.79877865e-01 -7.79795527e-01 -5.71116048e-04
4.91679966e-01 -3.73684138e-01 2.05229497e+00 -8.85221660e-01
-2.61139369e+00 7.56181180e-01 -4.61921319e-02 -2.77824283e-01
8.11161518e-01 -8.84733573e-02 -1.13802783e-01 -1.98749243e-04
1.10784702e-01 6.16482496e-01 6.16555870e-01 -1.04523170e+00
-7.13941634e-01 -3.50957848e-02 2.89552778e-01 7.36628711e-01
-3.78956556e-01 2.90124834e-01 3.59629430e-02 -1.24856092e-01
-6.65470302e-01 -7.79062212e-01 -1.83533564e-01 -3.10878366e-01
-6.87712014e-01 -1.08118200e+00 8.90751064e-01 -2.73400158e-01
1.09873617e+00 -1.77756238e+00 2.01988846e-01 -5.43385029e-01
1.33289486e-01 4.88102525e-01 -1.89794391e-01 1.28313506e+00
2.95710504e-01 1.27378315e-01 -1.38980627e-01 -4.70351070e-01
2.31047288e-01 1.33899316e-01 -2.57774770e-01 -1.84967339e-01
4.11784500e-01 6.70273900e-01 -1.06013060e+00 -4.58147198e-01
2.08898976e-01 -1.80240646e-02 -4.53684896e-01 1.16320384e+00
-4.31051463e-01 9.89600897e-01 -5.98329484e-01 -6.85981363e-02
2.68818796e-01 -4.71043438e-02 3.27270895e-01 2.45351598e-01
-6.05559826e-01 8.28669965e-01 -7.22572207e-01 1.55602956e+00
-7.26880848e-01 5.84570289e-01 3.20785433e-01 -1.03097868e+00
1.22269964e+00 8.61661553e-01 2.70624608e-01 -4.45383817e-01
2.46890679e-01 -8.19223598e-02 1.87699303e-01 -8.01056147e-01
8.14608097e-01 -2.58274615e-01 -3.62555146e-01 9.57964659e-01
1.99954078e-01 -5.66357613e-01 1.16416439e-01 2.12469250e-01
1.11998892e+00 -1.15643345e-01 4.69910085e-01 -2.00275064e-01
5.99366844e-01 6.96087629e-02 1.61682814e-01 8.40211570e-01
-6.14343822e-01 3.79130006e-01 8.91209483e-01 -2.73738682e-01
-7.15317845e-01 -5.18222868e-01 1.62560597e-01 1.47609174e+00
1.51386419e-02 -1.37233272e-01 -7.81653464e-01 -9.36647654e-01
-3.48381042e-01 9.10652637e-01 -5.28826594e-01 -1.91858634e-01
-5.75861931e-01 -1.26679093e-01 4.98359084e-01 9.58106071e-02
7.97125101e-01 -1.62901556e+00 -5.63035846e-01 4.93660539e-01
-5.63758254e-01 -9.38612521e-01 -8.63092363e-01 3.26985031e-01
-3.04524481e-01 -8.68792951e-01 -9.07292962e-01 -8.79007578e-01
3.17711949e-01 9.30702016e-02 8.94988239e-01 -1.96342006e-01
2.54215956e-01 5.53380728e-01 -7.73452401e-01 -2.35568404e-01
-1.10000050e+00 3.85573268e-01 -1.07348122e-01 2.09513903e-01
2.74292082e-01 -3.80311310e-01 -4.09210354e-01 4.98496085e-01
-7.65907884e-01 4.05292958e-01 6.22868501e-02 1.04301286e+00
-4.63867605e-01 -5.32939374e-01 1.10616350e+00 -1.15112805e+00
1.47034013e+00 -7.11608112e-01 1.27774999e-01 1.77869737e-01
-1.81340948e-01 7.85461143e-02 9.51956689e-01 -6.39499903e-01
-1.54721367e+00 -1.89858198e-01 -6.11690521e-01 1.91317409e-01
-5.25843024e-01 3.01665187e-01 -4.93914038e-02 4.67796683e-01
6.08795524e-01 5.07451773e-01 2.64905900e-01 -3.65333706e-01
4.54876572e-01 1.39124656e+00 8.52525607e-02 -8.72141957e-01
4.70511198e-01 -1.52286470e-01 -8.71981740e-01 -1.11958206e+00
-4.85573858e-01 -5.50001919e-01 -6.13963902e-01 -5.66863120e-01
9.81339991e-01 -3.64762872e-01 -8.39149952e-01 6.21527493e-01
-1.56184900e+00 -9.60734785e-01 -4.82500583e-01 -6.40384667e-03
-6.83071494e-01 3.87205362e-01 -6.94343090e-01 -1.08015299e+00
-4.86585975e-01 -1.06372178e+00 9.68391716e-01 4.37095016e-01
-7.01956570e-01 -1.25518847e+00 4.75778192e-01 2.32421219e-01
6.10281289e-01 -1.43870056e-01 7.76292026e-01 -1.51358199e+00
5.29130623e-02 1.65537337e-03 8.24371651e-02 3.32510710e-01
5.31346679e-01 -8.40441510e-02 -7.93069959e-01 -5.80159947e-02
2.21359283e-01 -8.55159461e-01 -5.28870672e-02 -1.82885434e-02
3.22298586e-01 -7.30341673e-01 -4.94170897e-02 -2.62929499e-01
5.17356336e-01 7.94152737e-01 5.06674349e-01 3.34391966e-02
1.53500840e-01 1.12011266e+00 8.55956554e-01 5.42246938e-01
7.57646739e-01 9.49820578e-01 2.24357545e-02 3.06238160e-02
1.44896775e-01 -2.69039512e-01 4.96676326e-01 1.18071914e+00
1.37563020e-01 -3.68614554e-01 -8.64120245e-01 6.36666119e-01
-2.03387117e+00 -9.45344567e-01 -2.36145303e-01 1.81623852e+00
1.53577781e+00 3.18522118e-02 4.30162907e-01 -1.94675565e-01
8.17695737e-01 4.96131748e-01 -3.61489236e-01 -1.04852509e+00
4.31999654e-01 3.94893475e-02 -4.56085503e-01 5.89033008e-01
-8.42463136e-01 9.94054914e-01 5.89555883e+00 4.34399784e-01
-9.82399225e-01 2.43284211e-01 4.80938792e-01 3.96213144e-01
-2.23260477e-01 8.02151710e-02 -8.11393499e-01 5.88342071e-01
1.09636271e+00 -4.42670673e-01 8.71402770e-02 7.63064086e-01
5.05156338e-01 -3.25171873e-02 -1.35201788e+00 5.42783856e-01
-9.77127478e-02 -1.12324953e+00 -5.10030761e-02 -1.87587559e-01
3.20596427e-01 -5.74986458e-01 -3.68019611e-01 8.55884135e-01
7.82561898e-01 -6.39266074e-01 4.72760379e-01 3.42280000e-01
7.58482695e-01 -3.54300648e-01 6.54998481e-01 8.31662059e-01
-9.75948334e-01 2.08361968e-01 9.11822096e-02 -3.45736295e-01
4.74365622e-01 1.24516547e-01 -1.33036900e+00 5.23461580e-01
2.34823003e-01 3.74749273e-01 2.24348214e-02 5.74096799e-01
-1.27150208e-01 6.89291537e-01 -4.45449837e-02 -6.08549654e-01
3.01976740e-01 -2.94769734e-01 7.31342494e-01 1.28701687e+00
5.53184487e-02 2.53916264e-01 4.51267689e-01 7.54880071e-01
2.23237872e-01 9.47945714e-02 -8.31251085e-01 -1.15318771e-03
7.08994508e-01 1.30822098e+00 -3.43839854e-01 -3.72354299e-01
-2.38319308e-01 1.12596786e+00 2.38304541e-01 2.25323334e-01
-6.32369518e-01 -7.08554924e-01 4.30302203e-01 -2.42110044e-02
-1.58301845e-01 -3.67326848e-02 7.59230256e-02 -9.50838566e-01
-3.46231878e-01 -1.00603580e+00 -4.42861840e-02 -5.14707029e-01
-1.46607089e+00 8.54264319e-01 5.33278137e-02 -1.11969900e+00
-7.77523875e-01 -5.03052175e-02 -1.21971595e+00 9.01247263e-01
-1.23249519e+00 -8.80351961e-01 -1.87862515e-01 5.24439156e-01
1.28975737e+00 -9.15970728e-02 9.90335405e-01 6.62642866e-02
-6.44129753e-01 5.79992175e-01 -2.02085584e-01 9.89988744e-02
1.07398510e+00 -1.42290187e+00 2.41269365e-01 1.43864930e-01
-5.83285987e-01 4.35646117e-01 8.90856028e-01 -3.89863938e-01
-1.08731818e+00 -9.25549686e-01 1.37445140e+00 -2.88064957e-01
5.34672141e-01 -6.53310537e-01 -1.02381110e+00 5.06433964e-01
9.38712955e-01 -6.11358643e-01 7.05038965e-01 9.70634744e-02
1.91913053e-01 2.24401832e-01 -1.04443884e+00 7.37238169e-01
7.86982536e-01 -3.88232708e-01 -9.80699003e-01 4.81100440e-01
9.37808931e-01 -4.02392954e-01 -7.28464901e-01 -4.69028875e-02
1.30034313e-01 -9.14404333e-01 3.46095800e-01 -7.42301524e-01
5.27429819e-01 2.28599578e-01 3.45768124e-01 -1.64138949e+00
3.94653492e-02 -1.35791814e+00 2.76981592e-01 1.73335314e+00
1.34622350e-01 -8.75824034e-01 2.49592856e-01 7.72503674e-01
-4.64220822e-01 -6.36665523e-01 -9.52204108e-01 -5.32995999e-01
2.74383396e-01 -9.50514281e-04 5.52192092e-01 9.76959705e-01
7.09837139e-01 1.03166389e+00 -5.15633225e-01 -5.84480464e-01
-1.28502950e-01 -1.18137799e-01 1.23279083e+00 -1.27078795e+00
2.56614108e-02 -4.55028415e-01 2.73754746e-01 -1.70215929e+00
1.79744855e-01 -5.54852486e-01 6.40070438e-01 -1.59742963e+00
-1.50998384e-01 -4.51248169e-01 5.31436384e-01 4.21348810e-01
-1.76834300e-01 -6.17059052e-01 9.98260081e-02 1.25353411e-01
-6.68621540e-01 9.55065966e-01 1.33062243e+00 -2.76949853e-02
-7.63391852e-01 4.58678395e-01 -6.70283020e-01 3.43276292e-01
7.15737283e-01 -2.10395172e-01 -4.99443799e-01 2.18032941e-01
-2.07906723e-01 7.92175233e-01 -1.66047901e-01 -4.66503918e-01
4.21283096e-01 -3.66946936e-01 -7.92539954e-01 -1.80949673e-01
1.93824351e-01 -3.70619863e-01 -4.02765691e-01 2.05993444e-01
-1.01952982e+00 -3.31294656e-01 -1.39957666e-01 4.55368757e-01
-3.32594395e-01 -4.60581213e-01 7.68822074e-01 -1.61816180e-01
-3.16599727e-01 -7.73136038e-03 -1.14123464e+00 4.10454392e-01
1.14199281e+00 -1.61183164e-01 -4.28754061e-01 -9.80597675e-01
-7.79673994e-01 6.18515551e-01 -3.62379663e-02 6.69230223e-01
3.13392222e-01 -1.22975767e+00 -8.00400555e-01 -2.23077476e-01
7.20463917e-02 5.62383942e-02 3.14502716e-01 7.72312462e-01
-6.40126765e-02 5.79975009e-01 -1.09925270e-01 -6.81415379e-01
-1.18308437e+00 -3.53486426e-02 2.50200063e-01 -7.71527529e-01
-6.50319815e-01 6.98849738e-01 1.97670728e-01 -9.02571142e-01
4.32983041e-01 -1.59828961e-01 -7.29136586e-01 3.90326023e-01
3.88742268e-01 1.47074476e-01 -1.98570803e-01 -5.31050265e-01
4.44717035e-02 -1.20524675e-01 -1.85266271e-01 -4.04230505e-01
1.07552588e+00 -3.20749223e-01 -2.81720739e-02 9.82803106e-01
1.05527699e+00 -4.53704834e-01 -1.29661834e+00 -5.15590608e-01
1.80819735e-01 -9.87580046e-02 -5.99843383e-01 -8.40468585e-01
-3.50294799e-01 1.00312126e+00 1.34121239e-01 1.14583373e+00
6.42009616e-01 5.85840968e-03 9.11652744e-01 4.39376295e-01
2.74328142e-01 -1.32037127e+00 6.58281207e-01 1.10243797e+00
1.25150335e+00 -1.18697166e+00 -8.03538918e-01 -1.27022043e-01
-1.33686793e+00 1.16529894e+00 8.71331632e-01 1.74668863e-01
4.04778510e-01 1.21491507e-01 2.06558913e-01 -1.58979788e-01
-1.14269602e+00 -1.74272269e-01 -1.46508321e-01 9.00474548e-01
5.95042229e-01 -7.05829263e-02 -5.14532924e-01 8.28063667e-01
-2.13933408e-01 -2.33283546e-02 7.35746145e-01 1.01668108e+00
-5.25315940e-01 -1.34203804e+00 -1.80967435e-01 4.62254971e-01
1.74358170e-02 5.68810664e-02 -9.23824608e-01 5.91382861e-01
-4.79916245e-01 1.46040893e+00 8.51437822e-02 -3.31247807e-01
5.70320249e-01 8.01226258e-01 -3.81599739e-02 -1.23560596e+00
-1.15922391e+00 -1.19862966e-01 7.04276025e-01 8.41255486e-03
-3.89232844e-01 -6.45851552e-01 -1.07333493e+00 -1.89051211e-01
-4.01722878e-01 7.03426480e-01 2.67633677e-01 1.12898982e+00
2.61836112e-01 5.12912691e-01 1.18271720e+00 -9.95945036e-01
-8.72866035e-01 -1.49374568e+00 -3.86636049e-01 3.95099580e-01
2.88759798e-01 -5.38009524e-01 -4.62754130e-01 -9.10614654e-02] | [12.878854751586914, 7.988559246063232] |
072c9333-2caa-49f6-9642-e2b0ec7a2bf9 | federated-sufficient-dimension-reduction | 2301.09500 | null | https://arxiv.org/abs/2301.09500v1 | https://arxiv.org/pdf/2301.09500v1.pdf | Federated Sufficient Dimension Reduction Through High-Dimensional Sparse Sliced Inverse Regression | Federated learning has become a popular tool in the big data era nowadays. It trains a centralized model based on data from different clients while keeping data decentralized. In this paper, we propose a federated sparse sliced inverse regression algorithm for the first time. Our method can simultaneously estimate the central dimension reduction subspace and perform variable selection in a federated setting. We transform this federated high-dimensional sparse sliced inverse regression problem into a convex optimization problem by constructing the covariance matrix safely and losslessly. We then use a linearized alternating direction method of multipliers algorithm to estimate the central subspace. We also give approaches of Bayesian information criterion and hold-out validation to ascertain the dimension of the central subspace and the hyper-parameter of the algorithm. We establish an upper bound of the statistical error rate of our estimator under the heterogeneous setting. We demonstrate the effectiveness of our method through simulations and real world applications. | ['Haoyang Cheng', 'Jianjun Xu', 'Yue Zhao', 'Wenquan Cui'] | 2023-01-23 | null | null | null | null | ['variable-selection'] | ['methodology'] | [-3.64091754e-01 -2.13674963e-01 -4.80964422e-01 -3.82624149e-01
-1.31350720e+00 -4.24370259e-01 2.61243999e-01 -6.59987926e-01
-9.50550586e-02 8.11516225e-01 3.59252602e-01 -3.45670372e-01
-4.20605093e-01 -4.03930753e-01 -8.50195050e-01 -1.24003553e+00
-1.95082113e-01 6.11657619e-01 -5.26777387e-01 3.11590374e-01
-9.30374265e-02 2.26150706e-01 -9.94853854e-01 -3.89800705e-02
9.58735645e-01 1.05591941e+00 -5.91446161e-02 4.93111163e-01
1.60488456e-01 8.12162101e-01 -1.42860323e-01 -3.03342134e-01
8.51416886e-01 -3.94268066e-01 -5.60400128e-01 8.37861225e-02
3.13850433e-01 -6.29236579e-01 -3.67941976e-01 1.16472137e+00
6.35397077e-01 -2.34608864e-03 4.38079506e-01 -1.39238131e+00
-4.49589752e-02 8.22696149e-01 -8.17922235e-01 3.65539603e-02
5.40346950e-02 -4.10353057e-02 8.43166590e-01 -1.15475190e+00
8.16917360e-01 1.19688654e+00 6.85681343e-01 1.13024719e-01
-1.32028401e+00 -1.01738560e+00 9.98393223e-02 1.45241961e-01
-1.55183697e+00 -6.82681561e-01 5.44423342e-01 -3.76723915e-01
4.00075465e-01 2.00010568e-01 1.64546505e-01 8.17862391e-01
-3.04046832e-02 8.93956959e-01 9.65543628e-01 -2.65604407e-01
5.70648611e-01 2.33684659e-01 6.90286681e-02 7.99588561e-01
5.09949207e-01 1.81015626e-01 -9.25365567e-01 -8.29770625e-01
4.98779476e-01 2.54373759e-01 -3.41449916e-01 -1.02935517e+00
-1.11297989e+00 1.07215750e+00 6.31250367e-02 -2.61529386e-01
-5.30220389e-01 1.72826961e-01 3.00288469e-01 4.87940729e-01
7.38016903e-01 -4.97497231e-01 -7.38680184e-01 1.75245270e-01
-9.35070932e-01 4.98026498e-02 1.28919613e+00 1.12325370e+00
7.92023957e-01 -1.07932156e-02 -8.78222436e-02 6.78052187e-01
4.47077185e-01 9.99135017e-01 2.46721789e-01 -1.23719168e+00
8.93684864e-01 2.13734463e-01 1.14812858e-01 -7.88747609e-01
-1.15187176e-01 -5.12213171e-01 -1.12234902e+00 -8.79830718e-02
3.23939145e-01 -7.00528383e-01 -4.37320881e-02 1.61649096e+00
1.01592433e+00 4.53136027e-01 -7.84327835e-02 1.27581668e+00
-6.57378957e-02 4.22552764e-01 -4.65448499e-01 -6.46666348e-01
8.66080463e-01 -8.83733571e-01 -7.02867508e-01 2.41214871e-01
8.63347292e-01 -1.96507931e-01 3.95439327e-01 5.91483057e-01
-7.67483711e-01 1.87478244e-01 -9.16571796e-01 3.10189545e-01
3.38091493e-01 -7.94030726e-02 7.55642295e-01 7.80149996e-01
-7.22860456e-01 3.61082613e-01 -8.63029003e-01 1.12723798e-01
5.27370751e-01 5.09108841e-01 -5.64369857e-01 -3.67063791e-01
-7.83791661e-01 1.83941439e-01 -1.85219333e-01 -2.52484024e-01
-9.66022253e-01 -9.76645172e-01 -5.57954550e-01 1.45864636e-01
5.01300871e-01 -1.20815063e+00 1.12741411e+00 -6.92670703e-01
-1.28141642e+00 4.84940082e-01 -2.02477545e-01 -5.86543143e-01
5.99664569e-01 -2.68494457e-01 -2.95103826e-02 7.16329413e-03
2.03621700e-01 -5.08748353e-01 1.07567155e+00 -8.79628778e-01
-8.73430967e-01 -1.09404969e+00 -5.71609080e-01 6.61524907e-02
-4.45825100e-01 -1.80228144e-01 -4.19521302e-01 -4.75300729e-01
3.26661855e-01 -9.45484102e-01 -4.34522301e-01 1.23914279e-01
-2.22048789e-01 8.25555623e-02 8.56869042e-01 -5.97317040e-01
1.23920476e+00 -2.28504252e+00 2.30012074e-01 6.64525867e-01
5.88993847e-01 -4.49180692e-01 1.57732844e-01 2.37496346e-01
2.33562917e-01 -1.30196586e-01 -2.16682062e-01 -3.37804079e-01
-4.93762273e-05 -2.17870817e-01 -4.22298551e-01 1.37427819e+00
-6.93517983e-01 2.02168137e-01 -7.37065077e-01 -5.34339726e-01
-8.87555778e-02 5.44001997e-01 -8.32954466e-01 3.41534495e-01
2.33939901e-01 4.03157324e-01 -7.74678171e-01 5.49658835e-01
9.44799125e-01 -5.34671366e-01 6.19415283e-01 -2.05284566e-01
1.03828661e-01 -1.79352894e-01 -1.86174500e+00 1.83226621e+00
-6.99380457e-01 7.89466277e-02 1.01262736e+00 -1.08656394e+00
6.29128516e-01 3.49597424e-01 9.47226048e-01 -4.01020288e-01
1.16130568e-01 3.75318676e-01 -6.39267206e-01 -4.38359112e-01
-3.98026258e-02 -2.19402507e-01 -8.87469947e-02 9.17040467e-01
1.59484461e-01 3.93144667e-01 -2.67129213e-01 3.71648103e-01
1.07800865e+00 -5.13821654e-02 3.54528248e-01 -4.35269296e-01
5.15407622e-01 -2.71498322e-01 8.20989311e-01 7.85314679e-01
-6.33528605e-02 4.48953807e-01 2.71288306e-01 -4.68980134e-01
-9.45884943e-01 -8.35852921e-01 -1.61208197e-01 1.16532409e+00
-1.49108600e-02 -4.08193409e-01 -4.56474692e-01 -7.61095822e-01
5.02176225e-01 4.75267053e-01 -2.95292228e-01 1.93419620e-01
-3.17305624e-01 -8.00974250e-01 1.14169478e-01 1.05097219e-02
2.91001558e-01 8.80851448e-02 -1.48601651e-01 6.31824285e-02
-1.88376710e-01 -8.39487433e-01 -6.09226942e-01 2.12851092e-01
-1.00410414e+00 -1.12396872e+00 -6.71573102e-01 -4.58966345e-01
5.97675264e-01 5.66448331e-01 8.16805303e-01 -3.40833485e-01
-7.33655170e-02 3.43934596e-01 9.85332429e-02 2.71159485e-02
-5.29892482e-02 -5.31143285e-02 5.45541346e-01 5.17996132e-01
9.57347527e-02 -5.81990600e-01 -7.19563901e-01 4.82187301e-01
-5.43068528e-01 -2.39370748e-01 3.36205214e-01 9.03672636e-01
8.11469376e-01 -1.79787934e-01 5.73495805e-01 -1.26657152e+00
5.48020065e-01 -1.06137037e+00 -9.85685408e-01 2.23427430e-01
-9.55499589e-01 4.75527495e-01 7.06186295e-01 -1.38952687e-01
-1.05444694e+00 4.65642333e-01 4.39605296e-01 -8.57786715e-01
4.62684870e-01 4.16857779e-01 -2.82001466e-01 -5.83037063e-02
5.43254137e-01 6.75021708e-02 2.31882751e-01 -7.89216280e-01
6.33713007e-01 1.03992903e+00 3.77554417e-01 -8.39850128e-01
8.62631083e-01 9.14548576e-01 8.34807828e-02 -4.43230987e-01
-6.71404064e-01 -7.78438509e-01 -2.42864877e-01 -9.00104176e-03
2.62945686e-02 -1.46361196e+00 -1.02715063e+00 -2.38650087e-02
-5.76752543e-01 -3.12367638e-06 -3.52906168e-01 7.55938828e-01
-6.90812767e-01 4.46095616e-01 -3.03084910e-01 -8.47681046e-01
-8.05750430e-01 -7.91214049e-01 9.20175910e-01 -2.84482211e-01
2.36452684e-01 -8.18151116e-01 6.01958990e-01 4.04201478e-01
3.25189590e-01 1.29391328e-01 4.70632195e-01 -6.36030257e-01
-5.94393492e-01 -3.70892793e-01 -1.53813079e-01 -4.15023081e-02
-1.82948589e-01 -4.92109120e-01 -7.18462169e-01 -6.42492712e-01
5.45155048e-01 -1.22595064e-01 6.84570312e-01 5.13818264e-01
1.15806699e+00 -9.23760831e-01 -4.45948869e-01 1.28045952e+00
1.61530519e+00 -6.39257133e-01 5.51688224e-02 1.87995255e-01
5.88945746e-01 2.86298603e-01 5.43410182e-01 1.31236899e+00
2.38278091e-01 4.42350954e-01 1.54612482e-01 2.16162413e-01
1.82757974e-01 -2.39439726e-01 4.68994230e-01 8.80481601e-01
3.55088830e-01 2.51454841e-02 -7.07181990e-01 3.95157099e-01
-2.06141472e+00 -9.54734743e-01 -1.17171444e-01 2.54356408e+00
7.75563657e-01 -7.04410851e-01 2.12299719e-01 -9.84868482e-02
8.24406862e-01 -1.27399966e-01 -9.15996492e-01 -2.11273134e-02
-1.88458368e-01 -1.44954666e-01 1.04920626e+00 3.18416715e-01
-8.21029425e-01 4.58565444e-01 6.07727718e+00 6.09730542e-01
-8.23727131e-01 4.51915711e-01 5.65204144e-01 -4.79890257e-01
-2.85507709e-01 4.18799669e-02 -7.38594651e-01 4.62214530e-01
1.04792583e+00 -5.44630051e-01 8.40539098e-01 1.23760891e+00
1.91886559e-01 1.12627782e-01 -1.01719511e+00 1.32124889e+00
-3.88645902e-02 -1.37541020e+00 -3.21792126e-01 4.31960970e-01
1.06741166e+00 5.87182581e-01 -1.48135275e-01 1.27482608e-01
6.02381468e-01 -6.56294167e-01 3.83097649e-01 4.73264664e-01
1.00377524e+00 -7.88110733e-01 4.37691480e-01 5.64260542e-01
-9.52432632e-01 -4.72104430e-01 -3.40589523e-01 1.52880803e-01
-1.41151240e-02 9.52923894e-01 -6.24026120e-01 3.98413688e-01
6.37651026e-01 5.22826076e-01 -7.43746534e-02 1.13785446e+00
2.53631264e-01 7.70884275e-01 -7.04065025e-01 2.95886546e-01
-3.22257698e-01 -6.34799600e-01 6.52552903e-01 8.17807853e-01
4.94400948e-01 6.65485561e-02 5.86241744e-02 5.04366755e-01
-2.75849491e-01 4.13625568e-01 -6.62714183e-01 3.75388354e-01
7.50272810e-01 1.17159367e+00 -6.14594854e-02 -1.17007405e-01
-5.61293960e-01 6.91867411e-01 3.41709256e-01 5.00841439e-01
-4.99108940e-01 1.61710605e-02 8.28966737e-01 -1.55295536e-01
6.69374824e-01 -1.15268007e-02 -4.19276804e-01 -1.55005789e+00
6.32637693e-03 -1.22087348e+00 8.40733469e-01 -1.34190675e-02
-1.56390512e+00 3.87429520e-02 -2.99548835e-01 -1.27809584e+00
-5.30475259e-01 -9.22260284e-02 -1.84461609e-01 7.52848387e-01
-1.01015854e+00 -7.59879708e-01 -4.02283855e-02 1.22313452e+00
1.48099050e-01 -5.12523592e-01 7.08749115e-01 4.04674977e-01
-8.25223088e-01 7.47860730e-01 1.10266781e+00 -9.70933959e-02
7.87575543e-01 -7.82722056e-01 -2.56998211e-01 7.87424624e-01
-6.69731647e-02 6.91119254e-01 7.18514562e-01 -5.11095583e-01
-2.24726224e+00 -9.58498597e-01 3.19508731e-01 2.55752658e-03
7.79864371e-01 -3.33478689e-01 -4.43844050e-01 6.88447952e-01
-2.58195430e-01 5.74181676e-01 9.16503549e-01 5.56311786e-01
-7.10408747e-01 -6.70705438e-01 -1.48212385e+00 9.15463716e-02
8.55729163e-01 -4.78809327e-01 1.68855786e-01 7.83432186e-01
7.56567299e-01 -2.16846153e-01 -9.42740977e-01 1.01080589e-01
6.41965449e-01 -8.42166603e-01 1.01136398e+00 -7.91949928e-01
-5.93282729e-02 -7.62301311e-02 -6.31753206e-01 -1.05658448e+00
-2.90623426e-01 -1.11090469e+00 -8.79796982e-01 1.00999737e+00
1.89175218e-01 -7.62329280e-01 1.08611155e+00 7.57362783e-01
4.40161735e-01 -5.00265002e-01 -1.32590878e+00 -7.66880929e-01
-5.63425757e-02 -5.05278945e-01 5.51431537e-01 8.55390787e-01
8.69830176e-02 2.36084998e-01 -6.87214494e-01 3.65644693e-01
1.45674145e+00 3.95565480e-01 9.97473240e-01 -1.05179656e+00
-5.64008117e-01 1.60505518e-01 -1.36559814e-01 -8.75928283e-01
3.88687044e-01 -1.00433230e+00 -1.24139071e-01 -9.20371592e-01
6.84790194e-01 -6.56697810e-01 -4.63912219e-01 4.92442325e-02
5.70282713e-02 -2.39882171e-01 1.23687815e-02 5.14004230e-01
-8.17481995e-01 6.15268171e-01 6.50163531e-01 -4.79912274e-02
-1.32927239e-01 2.58235157e-01 -7.91883528e-01 3.11367661e-01
6.34201229e-01 -8.69240046e-01 -3.23520899e-01 -3.73968571e-01
1.91958249e-01 5.84663749e-01 1.01036578e-01 -6.69144928e-01
5.31103253e-01 -3.27153832e-01 2.41472498e-01 -4.23106521e-01
-3.01799141e-02 -1.11866570e+00 2.90148497e-01 4.58156288e-01
-1.63249388e-01 -3.44631702e-01 -4.22430277e-01 9.90511775e-01
1.96257755e-01 9.61672589e-02 6.44279599e-01 2.51641572e-01
-1.27641231e-01 6.00405514e-01 4.69830371e-02 2.31209680e-01
9.52087581e-01 3.87216747e-01 -5.45970574e-02 -6.11324072e-01
-3.20319086e-01 5.13706267e-01 3.68404984e-01 -2.77477726e-02
5.69026768e-01 -1.18543220e+00 -9.98784304e-01 5.43375611e-01
-2.79625565e-01 -1.93359688e-01 2.55636394e-01 1.18893647e+00
-1.31453007e-01 4.34976012e-01 4.05830771e-01 -7.27004647e-01
-1.11831284e+00 7.52948642e-01 2.46975660e-01 -1.35459760e-02
-7.39588916e-01 6.85471237e-01 6.25131130e-02 -6.61100924e-01
4.68537211e-01 4.61169094e-01 5.43471634e-01 -1.31609946e-01
7.75803804e-01 7.22978354e-01 1.37976363e-01 -3.36970091e-01
-3.78161877e-01 2.18725368e-01 1.08433485e-01 -1.84136048e-01
1.70108783e+00 -7.04495609e-01 -2.29159385e-01 1.21905170e-01
1.71710873e+00 1.49672598e-01 -1.31071651e+00 -5.96079886e-01
-1.56786054e-01 -8.53686750e-01 5.90157926e-01 -3.16290855e-01
-1.50611806e+00 3.37304115e-01 7.90407717e-01 -2.61269897e-01
1.05747247e+00 1.62082873e-02 6.99400365e-01 5.42787015e-01
8.04365396e-01 -1.00923157e+00 -5.71096897e-01 1.08721964e-01
6.05872095e-01 -1.20087564e+00 4.94624853e-01 -1.47348344e-01
-4.89661455e-01 8.28544199e-01 1.08080536e-01 -1.74331501e-01
8.95865321e-01 4.97842133e-01 -2.40326375e-01 -7.31195584e-02
-1.10284460e+00 2.92735130e-01 -2.09520444e-01 4.75076199e-01
5.34924045e-02 1.05261266e-01 -2.53360540e-01 8.28328729e-01
1.93382353e-02 2.74162203e-01 2.49015927e-01 6.37977540e-01
-2.21277177e-01 -8.61503243e-01 -8.34461093e-01 5.90993941e-01
-5.25847614e-01 1.31488845e-01 4.54197871e-03 2.40410432e-01
-4.84503031e-01 8.89829278e-01 -2.48500869e-01 -3.32505792e-01
-2.02175766e-01 1.13963671e-01 2.86192328e-01 -1.87535062e-01
-3.31143320e-01 2.69986808e-01 -1.17561735e-01 -8.69775295e-01
-1.85878538e-02 -7.76819348e-01 -1.10766232e+00 -6.02658689e-01
-4.55739945e-01 3.51286322e-01 1.03060448e+00 6.47309542e-01
7.59785235e-01 -2.67734677e-01 1.46349013e+00 -5.12096047e-01
-1.54375243e+00 -5.63657641e-01 -9.56541777e-01 1.96080387e-01
4.32819754e-01 -3.23173136e-01 -7.97403455e-01 -1.04357153e-01] | [5.94724178314209, 6.063390254974365] |
72b67489-044d-4a84-a116-85ee44bbb557 | handling-inter-annotator-agreement-for | 1906.02415 | null | https://arxiv.org/abs/1906.02415v1 | https://arxiv.org/pdf/1906.02415v1.pdf | Handling Inter-Annotator Agreement for Automated Skin Lesion Segmentation | In this work, we explore the issue of the inter-annotator agreement for training and evaluating automated segmentation of skin lesions. We explore what different degrees of agreement represent, and how they affect different use cases for segmentation. We also evaluate how conditioning the ground truths using different (but very simple) algorithms may help to enhance agreement and may be appropriate for some use cases. The segmentation of skin lesions is a cornerstone task for automated skin lesion analysis, useful both as an end-result to locate/detect the lesions and as an ancillary task for lesion classification. Lesion segmentation, however, is a very challenging task, due not only to the challenge of image segmentation itself but also to the difficulty in obtaining properly annotated data. Detecting accurately the borders of lesions is challenging even for trained humans, since, for many lesions, those borders are fuzzy and ill-defined. Using lesions and annotations from the ISIC Archive, we estimate inter-annotator agreement for skin-lesion segmentation and propose several simple procedures that may help to improve inter-annotator agreement if used to condition the ground truths. | ['Sandra Avila', 'Eduardo Valle', 'Vinicius Ribeiro'] | 2019-06-06 | null | null | null | null | ['skin-lesion-segmentation'] | ['medical'] | [ 4.01249439e-01 1.33483618e-01 -6.50351495e-02 -3.83979857e-01
-9.23279226e-01 -1.00627220e+00 2.78210014e-01 7.44461238e-01
-5.81898034e-01 5.17959058e-01 -6.94738925e-02 -2.71286488e-01
3.15702483e-02 -4.51719105e-01 -1.93377063e-01 -7.39806592e-01
2.38123953e-01 5.84076285e-01 6.09724879e-01 1.46653458e-01
2.20519990e-01 5.60925722e-01 -1.19580662e+00 4.37234372e-01
1.23186350e+00 6.78672910e-01 1.63814664e-01 8.29383612e-01
-3.17567021e-01 3.10148627e-01 -4.90250587e-01 -5.26091933e-01
1.78918809e-01 -4.68873650e-01 -1.17187667e+00 3.68172824e-01
4.02124614e-01 3.15113813e-02 5.96433878e-01 1.12458682e+00
2.55464941e-01 -3.09656173e-01 1.06001437e+00 -8.14665914e-01
-7.63550459e-04 2.91820645e-01 -6.04879498e-01 8.03974792e-02
1.98348686e-01 1.94653332e-01 5.94749570e-01 -2.90559173e-01
6.73309982e-01 6.82067215e-01 8.54636312e-01 6.10400498e-01
-1.25199854e+00 -1.36700839e-01 -3.51267338e-01 1.73979595e-01
-1.48019254e+00 -2.19424322e-01 2.86692232e-01 -7.94897735e-01
2.49466106e-01 5.86590707e-01 5.16262114e-01 5.17586827e-01
-2.02115178e-01 5.98838449e-01 1.50026226e+00 -7.62913167e-01
3.98535430e-01 4.84138995e-01 2.18230069e-01 6.86485410e-01
3.48238587e-01 -4.55713540e-01 2.22672731e-01 -2.79283196e-01
8.01457167e-01 -4.21119422e-01 -2.98651874e-01 -1.13163851e-01
-7.91408658e-01 5.92666805e-01 3.70303690e-01 8.13643515e-01
-4.80471283e-01 -2.00354069e-01 5.57406962e-01 4.18581069e-02
4.23925549e-01 5.88735700e-01 -8.28671828e-02 4.00121964e-04
-1.20563674e+00 -1.90909505e-01 8.00524950e-01 1.34464234e-01
4.49331820e-01 -6.27272010e-01 -1.36221930e-01 1.01141322e+00
-2.41789035e-02 1.33264899e-01 2.16214344e-01 -8.72875810e-01
-5.86916730e-02 7.93295503e-01 1.21098556e-01 -7.00224161e-01
-3.22546065e-01 -3.94729488e-02 -7.35999942e-01 4.86700982e-01
1.16744101e+00 -8.30139890e-02 -1.28131449e+00 1.15781879e+00
4.32875961e-01 -1.83125481e-01 -3.21813971e-01 1.02261078e+00
5.16328871e-01 2.07321510e-01 6.03622496e-01 -1.24626584e-01
1.57687485e+00 -6.02269888e-01 -5.73471487e-01 -2.13403434e-01
8.74987066e-01 -1.03875828e+00 9.07038093e-01 3.11820656e-01
-1.06817520e+00 -2.57745415e-01 -6.49698138e-01 2.13697225e-01
-3.76006752e-01 5.30896842e-01 3.97747248e-01 8.49528730e-01
-1.04794943e+00 5.32174349e-01 -8.37799728e-01 -8.40744555e-01
3.71149451e-01 2.47341305e-01 -7.06155002e-01 -2.72230133e-02
-8.15650702e-01 1.39418638e+00 4.08603489e-01 9.48967561e-02
-1.13991551e-01 -4.02246118e-01 -5.27924597e-01 -3.71150821e-01
4.66460705e-01 -2.89855480e-01 9.84832168e-01 -1.22954035e+00
-9.33117628e-01 1.47191513e+00 -2.07512274e-01 -2.83305883e-01
7.36114919e-01 3.67950529e-01 -2.70984679e-01 3.68947208e-01
1.03956200e-01 6.71450377e-01 5.91709852e-01 -1.47181547e+00
-6.07916296e-01 -3.14648926e-01 -8.31959099e-02 1.19645342e-01
9.72399414e-02 1.97808102e-01 -4.64999139e-01 -2.80733585e-01
-2.51675514e-03 -1.01417303e+00 -5.65839112e-01 2.70741314e-01
-4.06600326e-01 -2.27108851e-01 4.19659227e-01 -1.17684948e+00
1.17941177e+00 -1.98775685e+00 -1.78525701e-01 6.53702974e-01
1.67763203e-01 6.14407361e-01 6.31190687e-02 1.05582401e-01
-2.03692317e-02 5.75600564e-01 -5.27417064e-01 -9.57425535e-02
-2.93903083e-01 1.36920899e-01 4.53827590e-01 4.79510128e-01
3.17322642e-01 5.08513510e-01 -7.44836092e-01 -9.90409017e-01
5.41385353e-01 4.87162977e-01 1.77889466e-01 3.94039564e-02
2.23493185e-02 6.41053081e-01 -2.78080970e-01 5.95196486e-01
5.30342281e-01 -1.12385772e-01 2.21212924e-01 -3.68141770e-01
1.43767204e-02 -2.73005545e-01 -1.34562862e+00 1.20735991e+00
-3.86133999e-01 5.32528460e-01 4.85722810e-01 -7.87404954e-01
7.91510463e-01 4.47764456e-01 5.27869582e-01 -3.07189107e-01
2.81080872e-01 5.57517648e-01 2.87367940e-01 -8.23259473e-01
2.43947804e-01 -3.73177946e-01 1.42228588e-01 6.13735259e-01
-1.95414543e-01 -5.09092033e-01 5.95351756e-01 -4.81926128e-02
9.68633771e-01 -3.23754072e-01 6.22559011e-01 -4.13423747e-01
8.07305455e-01 3.11053008e-01 3.70732188e-01 4.45528507e-01
-5.67218542e-01 7.98307598e-01 6.76935077e-01 -3.29764336e-01
-1.03772318e+00 -7.95901716e-01 -4.30648565e-01 5.42207837e-01
2.70379614e-03 -7.75787607e-02 -1.30409491e+00 -8.99169385e-01
-2.99041897e-01 5.36692381e-01 -8.28249574e-01 2.82181084e-01
-2.85210133e-01 -7.05498397e-01 5.56011677e-01 4.78575140e-01
2.07572624e-01 -8.55333447e-01 -6.19981349e-01 7.24805966e-02
-3.60990793e-01 -1.00705016e+00 -2.86188126e-01 1.41069502e-01
-4.83488142e-01 -1.46934736e+00 -1.02876854e+00 -7.98675656e-01
1.24351192e+00 -7.65966401e-02 1.01422894e+00 6.88365459e-01
-6.71623051e-01 3.56767833e-01 -5.40968001e-01 -3.02284390e-01
-9.76094902e-01 1.32505432e-01 -4.51327622e-01 2.87619419e-02
2.65937775e-01 8.99217799e-02 -5.02851009e-01 6.50967777e-01
-1.17175639e+00 -1.26568988e-01 5.71265340e-01 4.06224728e-01
6.47028804e-01 4.35931720e-02 2.33770639e-01 -1.28361857e+00
7.46517003e-01 -5.09714037e-02 -2.93874145e-01 6.89366996e-01
-2.03482658e-01 -1.68939367e-01 3.26235175e-01 -1.31434307e-01
-9.92260635e-01 3.32298994e-01 -1.89457729e-01 1.42122462e-01
-7.08536804e-01 2.53704041e-01 3.81531924e-01 -4.54826683e-01
1.03212857e+00 -3.79285008e-01 3.83887321e-01 -8.53374824e-02
5.85264228e-02 7.70689428e-01 4.41636324e-01 -3.08097571e-01
5.83806336e-01 4.64629024e-01 5.75223900e-02 -8.44810545e-01
-7.51994014e-01 -9.49887395e-01 -1.00874615e+00 -4.46957707e-01
1.15227282e+00 -2.21703231e-01 -1.85263053e-01 2.85600483e-01
-1.14749265e+00 -5.76030374e-01 -1.79816961e-01 1.68050751e-01
-2.28362560e-01 5.59294701e-01 -4.77853954e-01 -7.46772826e-01
-3.21211308e-01 -1.33611536e+00 1.00616276e+00 3.85246366e-01
-6.70477092e-01 -1.46875393e+00 -3.02144215e-02 6.88511431e-01
3.50244284e-01 5.02409041e-01 9.76447940e-01 -5.75379908e-01
6.51097968e-02 -6.03942513e-01 -2.70800918e-01 6.36144996e-01
2.61623949e-01 5.15023291e-01 -8.46457660e-01 5.61834276e-02
-3.04697216e-01 -2.68906653e-01 7.93431044e-01 5.19885063e-01
1.05220079e+00 1.71420753e-01 -4.40397531e-01 -7.45419338e-02
1.48020113e+00 -1.61669686e-01 6.40179396e-01 9.86727402e-02
3.25792760e-01 1.22482312e+00 8.32111716e-01 -2.03783922e-02
5.02775647e-02 3.65996987e-01 3.66253078e-01 -6.46961689e-01
-2.08464727e-01 3.72600585e-01 -1.88272610e-01 5.08681238e-01
-4.15031672e-01 4.22362871e-02 -1.30912364e+00 7.21445799e-01
-1.53325546e+00 -6.00594759e-01 -8.59578431e-01 2.37264371e+00
9.57075834e-01 -6.49269968e-02 4.90912169e-01 3.41363221e-01
1.17895842e+00 -3.53969395e-01 -1.06132425e-01 -4.59906399e-01
1.94289237e-01 3.34658206e-01 5.31167746e-01 7.15452433e-01
-1.07210433e+00 6.86367452e-01 6.47628450e+00 8.24832857e-01
-1.28063357e+00 3.79695860e-03 7.93809474e-01 5.16692758e-01
1.21804275e-01 -5.64397797e-02 -3.73467952e-01 4.38939214e-01
3.83118004e-01 2.44247451e-01 6.76508397e-02 3.86141002e-01
1.33825839e-01 -8.23652625e-01 -9.57834721e-01 6.07735515e-01
1.18249588e-01 -9.69080806e-01 -6.04750931e-01 -6.70336410e-02
6.99800432e-01 -3.96275789e-01 -2.54042923e-01 -2.73255616e-01
7.61788040e-02 -1.10809731e+00 2.59784669e-01 3.68160605e-01
8.58440340e-01 -4.05116290e-01 1.26275384e+00 1.56965539e-01
-9.76801157e-01 4.47483867e-01 -5.24297208e-02 3.68792206e-01
2.51560956e-01 8.33135247e-01 -1.39496040e+00 2.73630083e-01
3.96917582e-01 -1.20291278e-01 -8.79198313e-01 1.56204510e+00
-4.36226487e-01 6.44036829e-01 -4.78969157e-01 -4.19170829e-03
2.14598447e-01 -6.46161288e-02 2.89244264e-01 1.38562393e+00
1.29101826e-02 -9.50955451e-02 1.90355822e-01 7.57565618e-01
5.28220117e-01 4.76392418e-01 -1.11447230e-01 7.55724777e-03
3.54436398e-01 1.68842757e+00 -1.47823441e+00 -1.95288107e-01
6.05858397e-03 9.40486729e-01 1.00485362e-01 2.57149130e-01
-6.68684423e-01 -2.32307404e-01 3.04176733e-02 4.41828579e-01
-2.21221507e-01 -1.69612076e-02 -5.70216358e-01 -5.73004961e-01
7.52054974e-02 -7.75777936e-01 6.03469074e-01 -5.25603294e-01
-1.37528241e+00 4.46885496e-01 -1.11277647e-01 -1.05109739e+00
-2.00925499e-01 -6.53938890e-01 -1.00544548e+00 9.02559400e-01
-1.12558329e+00 -1.21698129e+00 -6.32513285e-01 2.51910359e-01
4.66205478e-02 5.36798179e-01 9.43881214e-01 6.89957738e-02
-3.99088949e-01 3.54443550e-01 -2.18695626e-01 2.49471873e-01
9.48105693e-01 -1.62681234e+00 -1.09801285e-01 6.20126247e-01
-4.48574796e-02 2.21270844e-01 8.26319695e-01 -6.08889580e-01
-5.29677331e-01 -8.49265516e-01 6.94227934e-01 -3.91149998e-01
5.83995044e-01 4.86331061e-02 -1.14169586e+00 2.15821072e-01
1.62355870e-01 3.12137809e-02 1.06328547e+00 2.07096085e-01
5.59671689e-03 1.64923102e-01 -1.61654115e+00 5.21180987e-01
2.98444033e-01 -3.26698273e-01 -2.57197469e-01 4.01240915e-01
-2.36714005e-01 -3.94497514e-01 -1.07453561e+00 3.19986492e-01
3.72775167e-01 -1.11502016e+00 4.65632021e-01 -1.42873615e-01
1.69129640e-01 -3.95499915e-01 4.01494086e-01 -1.41372371e+00
-6.08328395e-02 -1.49171278e-01 8.26730847e-01 1.33504891e+00
6.60549283e-01 -3.77289355e-01 9.93316233e-01 9.46061671e-01
1.51035398e-01 -5.95228255e-01 -7.67898738e-01 -4.19009835e-01
1.47818506e-01 -4.08096164e-01 6.12479076e-02 1.09717333e+00
-2.35862825e-02 -1.54381886e-01 2.20992267e-01 7.12833479e-02
4.86384600e-01 -3.05092365e-01 5.56162000e-01 -1.21095729e+00
-3.10154166e-02 -5.73190153e-01 -6.37901962e-01 1.33269303e-03
-1.67424843e-01 -8.62595797e-01 2.38101944e-01 -1.85001004e+00
1.15101777e-01 -8.09826076e-01 1.09499559e-01 6.26044929e-01
-4.17607933e-01 5.59741914e-01 1.80467814e-01 8.04387778e-02
-4.97183979e-01 -6.29818738e-01 1.23312128e+00 2.77602356e-02
-6.83414191e-02 5.39262034e-02 -4.98611391e-01 9.62616980e-01
8.20757687e-01 -2.93572307e-01 7.10240081e-02 -1.77525744e-01
-1.81513224e-02 -1.51543617e-01 5.55792212e-01 -1.02131832e+00
2.91659206e-01 -2.91879505e-01 2.75498986e-01 -2.37400964e-01
3.87274027e-02 -9.10126746e-01 1.22732431e-01 5.54918826e-01
-2.88640946e-01 -4.48044300e-01 1.11199580e-01 1.04580663e-01
-2.92989403e-01 -1.06559718e+00 1.12969911e+00 -4.41168129e-01
-6.16580546e-01 -1.54535368e-01 -5.08128226e-01 -2.00167462e-01
1.41650033e+00 -4.81445938e-01 -1.61234856e-01 -3.08005840e-01
-1.16628969e+00 2.85609543e-01 8.89020920e-01 -2.22889915e-01
-5.26791885e-02 -9.03348088e-01 -8.34233582e-01 -3.09806436e-01
1.31654441e-01 1.62167951e-01 3.43344927e-01 1.24209142e+00
-9.92135108e-01 -1.04174009e-02 -2.73590744e-01 -8.43778491e-01
-1.72161853e+00 1.98044535e-02 4.68855739e-01 -5.37422240e-01
9.18388516e-02 8.35286140e-01 -1.36588499e-01 -2.01559499e-01
1.53065279e-01 -3.17182869e-01 -2.53437817e-01 2.30362728e-01
3.53477508e-01 4.83941883e-01 2.85822183e-01 -5.72866678e-01
-2.30735198e-01 7.50976682e-01 -2.05780745e-01 6.72058091e-02
7.89250672e-01 2.35684682e-02 -3.51851791e-01 2.55836278e-01
6.90088928e-01 3.77180688e-02 -6.68081522e-01 1.44779220e-01
1.66676372e-01 -4.43844497e-01 1.33980354e-02 -1.11560321e+00
-8.23811889e-01 9.46740806e-01 6.91490829e-01 7.09159493e-01
1.02699161e+00 5.96469752e-02 4.87664700e-01 -1.20946690e-01
2.79917628e-01 -1.41329324e+00 -2.66049743e-01 -2.61796862e-02
6.44383132e-01 -1.43385649e+00 3.26867253e-02 -9.81087744e-01
-8.83386731e-01 1.20542324e+00 5.22695065e-01 2.26045754e-02
3.49980593e-01 2.77913213e-01 4.29631740e-01 -8.42674822e-02
-8.08963999e-02 -5.34354925e-01 5.08359551e-01 8.35644484e-01
7.43781745e-01 2.60459036e-01 -7.76580215e-01 2.07455289e-02
2.13199869e-01 1.30628916e-02 4.94079083e-01 7.48876691e-01
-4.47840214e-01 -1.29274845e+00 -6.70676887e-01 7.97955573e-01
-7.67118096e-01 3.06026191e-01 -9.71378207e-01 8.07331324e-01
5.27749538e-01 8.48018467e-01 -3.82854231e-02 7.20343366e-03
9.97395217e-02 -4.33726981e-02 5.89264810e-01 -6.72741652e-01
-9.46701944e-01 -5.07438034e-02 4.04289335e-01 -7.07669556e-02
-4.73079175e-01 -6.75545931e-01 -1.25350451e+00 -1.03046484e-01
-6.31463289e-01 2.01680481e-01 9.17810440e-01 1.19223213e+00
-1.20988943e-01 1.48947611e-01 2.69263893e-01 -4.28856432e-01
-3.37514013e-01 -8.83607745e-01 -5.10398567e-01 7.63220608e-01
-1.22605994e-01 -3.05991888e-01 -3.63987476e-01 3.27547252e-01] | [15.555456161499023, -2.9780287742614746] |
a87472e0-40cd-413f-a868-b9612414314c | travelbert-pre-training-language-model | 2109.01048 | null | https://arxiv.org/abs/2109.01048v2 | https://arxiv.org/pdf/2109.01048v2.pdf | TravelBERT: Pre-training Language Model Incorporating Domain-specific Heterogeneous Knowledge into A Unified Representation | Existing technologies expand BERT from different perspectives, e.g. designing different pre-training tasks, different semantic granularities and different model architectures. Few models consider expanding BERT from different text formats. In this paper, we propose a heterogeneous knowledge language model (HKLM), a unified pre-trained language model (PLM) for all forms of text, including unstructured text, semi-structured text and well-structured text. To capture the corresponding relations among these multi-format knowledge, our approach uses masked language model objective to learn word knowledge, uses triple classification objective and title matching objective to learn entity knowledge and topic knowledge respectively. To obtain the aforementioned multi-format text, we construct a corpus in the tourism domain and conduct experiments on 5 tourism NLP datasets. The results show that our approach outperforms the pre-training of plain text using only 1/4 of the data. The code, datasets, corpus and knowledge graph will be released. | ['Zhiheng Lyu', 'Jinghui Xiao', 'Juanzi Li', 'Lei Hou', 'Hao Peng', 'Hongyin Zhu'] | 2021-09-02 | null | null | null | null | ['triple-classification'] | ['graphs'] | [-1.95747301e-01 3.48236531e-01 -5.33377409e-01 -1.99872717e-01
-4.28900927e-01 -6.94453239e-01 6.74654126e-01 6.36453182e-02
-8.13190162e-01 1.04972947e+00 5.83005011e-01 -3.90979350e-01
-4.58201289e-01 -1.12461746e+00 -5.35828412e-01 -1.06235221e-01
-2.44923178e-02 8.71269107e-01 6.60487413e-01 -6.44845843e-01
2.49203965e-01 -1.45187259e-01 -1.23144913e+00 4.85274822e-01
1.31050539e+00 9.01396930e-01 5.65347135e-01 2.05571741e-01
-1.15488923e+00 1.02304137e+00 -4.73149002e-01 -8.40930879e-01
2.42885575e-01 1.22372515e-01 -1.21873391e+00 -1.47807598e-01
9.26204920e-02 1.78136423e-01 -4.60072994e-01 8.51104259e-01
2.87169844e-01 2.34657362e-01 6.53639317e-01 -1.02587295e+00
-1.09650981e+00 1.24517488e+00 -3.23630452e-01 1.04471654e-01
3.10720563e-01 -5.35919249e-01 9.49900150e-01 -8.28514040e-01
8.10709059e-01 1.26869512e+00 6.14523470e-01 2.84569472e-01
-5.95527589e-01 -5.45177579e-01 2.70956874e-01 5.07347882e-01
-1.75413918e+00 2.55329814e-02 4.91738200e-01 -2.60431707e-01
1.14989758e+00 1.43942624e-01 3.51923555e-01 8.13748360e-01
-3.72474343e-02 1.08017755e+00 9.49034750e-01 -6.55698955e-01
-2.06850484e-01 7.32954264e-01 4.12568748e-01 6.25180960e-01
4.44198459e-01 -2.78230220e-01 -5.94691753e-01 2.18108505e-01
3.48889261e-01 -1.65766016e-01 -9.69670117e-02 7.08013773e-02
-8.31648290e-01 5.53339601e-01 5.10021895e-02 4.65354592e-01
-8.50956813e-02 -3.15972209e-01 3.95940095e-01 4.13027793e-01
3.42551827e-01 3.39493603e-01 -1.12529850e+00 1.22347705e-01
-7.25813448e-01 6.30745590e-02 1.15785134e+00 1.54675615e+00
1.03575385e+00 -2.16240987e-01 -1.06192261e-01 1.02846336e+00
3.74543756e-01 5.23497403e-01 1.09513259e+00 -2.16379061e-01
8.27322364e-01 1.04493237e+00 4.87993732e-02 -8.86465669e-01
-5.84121883e-01 -4.51730847e-01 -5.07275403e-01 -7.43217528e-01
5.22559835e-03 -4.62830663e-01 -1.11089921e+00 1.42866266e+00
7.91626275e-02 2.37790853e-01 7.89984345e-01 4.34225261e-01
1.53430927e+00 7.60922074e-01 3.87696862e-01 1.83193982e-01
1.49336791e+00 -1.17698550e+00 -6.82386279e-01 -4.63806838e-01
8.68703544e-01 -8.88998091e-01 1.09999073e+00 2.14540586e-01
-7.30068505e-01 -3.72527391e-01 -7.24004269e-01 -3.40480387e-01
-1.52889132e+00 6.25607729e-01 4.51250881e-01 5.49387753e-01
-7.93549836e-01 -2.64210850e-02 -1.61212444e-01 -5.93653679e-01
2.97557041e-02 3.94831419e-01 -3.28331918e-01 -2.89216489e-01
-1.74284434e+00 9.17409837e-01 1.35690260e+00 -1.49745077e-01
-5.05198598e-01 -5.75325429e-01 -8.56263578e-01 2.20463365e-01
8.24161232e-01 -7.21520066e-01 8.00786853e-01 -8.71054709e-01
-1.26748347e+00 8.62074375e-01 7.46727884e-02 -3.84602219e-01
1.05588846e-01 -2.69963920e-01 -9.77979243e-01 -2.62637228e-01
1.83611929e-01 4.02087867e-01 5.18781722e-01 -1.31009519e+00
-9.99029815e-01 -1.89553425e-01 3.87168378e-01 4.65541214e-01
-5.80115259e-01 -5.02397791e-02 -8.65728140e-01 -6.06839478e-01
-5.29948235e-01 -5.27159452e-01 4.48893085e-02 -1.00960255e+00
-4.19194400e-01 -3.07953209e-01 7.96886802e-01 -7.95196354e-01
1.80206394e+00 -1.97261786e+00 -7.04910532e-02 1.99341178e-01
-2.36511603e-03 2.96098590e-01 -1.86506897e-01 8.66296589e-01
6.71433151e-01 3.55524361e-01 2.17867140e-02 -7.68095553e-02
1.95683867e-01 6.90580010e-01 -4.53823060e-01 -5.21939456e-01
-2.26331696e-01 1.19675946e+00 -7.95436263e-01 -9.22751069e-01
-7.89331496e-02 -1.66529382e-03 -1.44881040e-01 -1.76739916e-01
-3.26462984e-01 -1.53790593e-01 -7.73858845e-01 8.08185697e-01
5.62825739e-01 -2.64778376e-01 3.16883147e-01 -2.95924485e-01
-1.60021484e-01 -6.87565804e-02 -1.16514432e+00 1.66833794e+00
-8.95243526e-01 3.95170599e-01 -1.31182745e-01 -9.51845646e-01
8.98640037e-01 2.54316717e-01 3.92067164e-01 -5.32387435e-01
1.88924193e-01 4.04496253e-01 -2.95170039e-01 -1.06846988e+00
9.45550561e-01 -2.03731790e-01 -2.80140489e-01 -6.71454072e-02
4.06144142e-01 1.52144775e-01 6.16027474e-01 3.31404269e-01
9.65899944e-01 7.75151998e-02 1.53560132e-01 -3.43724668e-01
6.80890799e-01 4.36888933e-01 3.04431409e-01 7.24210382e-01
3.31504256e-01 -6.77824542e-02 1.21212378e-01 -7.82717392e-02
-4.58072841e-01 -7.21343994e-01 -2.28073955e-01 1.34652364e+00
5.39806962e-01 -7.04204917e-01 -2.93203354e-01 -9.95144129e-01
2.49873802e-01 7.99968600e-01 -3.16729426e-01 -1.51365086e-01
-3.67130339e-01 -7.45378256e-01 6.48432970e-01 4.35529143e-01
8.67346227e-01 -9.47430015e-01 -8.55435356e-02 2.44721651e-01
-3.91382277e-01 -1.46109283e+00 -2.82016665e-01 1.78926419e-02
-4.65089083e-01 -1.03645325e+00 -2.94251114e-01 -1.22610056e+00
3.74168754e-01 1.21938333e-01 1.10512698e+00 -1.25956088e-01
-2.04706997e-01 5.81751943e-01 -7.97003329e-01 -6.70505881e-01
1.95808172e-01 5.37252307e-01 -2.81131208e-01 -1.88784916e-02
8.15888405e-01 -3.47669631e-01 -1.44990548e-01 2.93928802e-01
-1.32085323e+00 1.22592196e-01 9.26714420e-01 8.33989382e-01
1.81661040e-01 3.83129448e-01 8.53100061e-01 -1.05466950e+00
8.81253600e-01 -9.54825461e-01 -3.06185454e-01 9.28703010e-01
-7.94099510e-01 3.06231380e-01 5.65712631e-01 -4.77537543e-01
-1.47989321e+00 -2.90558785e-01 3.59025560e-02 1.61701724e-01
-4.80641872e-02 1.48395777e+00 -2.51191199e-01 -2.73388419e-02
6.44655704e-01 5.22210240e-01 -6.77206159e-01 -8.00007463e-01
6.95301473e-01 1.09284854e+00 4.20434445e-01 -8.64755988e-01
6.21421576e-01 3.99556577e-01 -3.13326776e-01 -9.83340621e-01
-1.03287947e+00 -1.04575634e+00 -7.63518751e-01 -1.07062876e-01
6.39161110e-01 -1.02445805e+00 -2.42086664e-01 1.16627641e-01
-9.02729928e-01 -2.68825710e-01 -1.58385962e-01 5.64167976e-01
-1.60917446e-01 4.13152874e-01 -2.89393574e-01 -5.42068183e-01
-4.74836648e-01 -4.47017610e-01 1.10495985e+00 2.36775085e-01
3.25065613e-01 -1.49578011e+00 -2.44650934e-02 4.96968061e-01
4.20142740e-01 -1.61813293e-02 1.17649758e+00 -1.09722483e+00
-3.36449623e-01 -3.20402503e-01 -2.89260358e-01 2.26493031e-01
1.76064461e-01 -4.54759121e-01 -4.89418536e-01 6.07838407e-02
-5.95224857e-01 -6.31714165e-01 1.05720603e+00 -8.28323588e-02
9.82502878e-01 -5.29002309e-01 -5.05480886e-01 4.62428153e-01
1.77056801e+00 1.96592525e-01 5.98985374e-01 7.79029727e-01
5.91280639e-01 6.96031332e-01 6.76894605e-01 3.94338399e-01
9.22940552e-01 2.91617572e-01 -1.03762195e-01 2.07834288e-01
-5.59153445e-02 -3.52578759e-01 3.18534821e-01 9.68234658e-01
-1.82121634e-01 -6.50825918e-01 -1.09094477e+00 7.42395520e-01
-1.96304047e+00 -9.40683484e-01 -3.53564024e-02 1.80370712e+00
1.09618294e+00 7.82344788e-02 -1.26116931e-01 -4.74271119e-01
3.84049892e-01 -1.97286472e-01 -4.02177036e-01 4.50148173e-02
-5.79851866e-01 9.51028466e-02 8.94532382e-01 6.68139815e-01
-8.45260561e-01 1.84549606e+00 5.41087723e+00 1.31760395e+00
-8.72041821e-01 1.11731447e-01 -1.15059555e-01 1.06103048e-01
-5.92747152e-01 2.33343691e-01 -1.28897738e+00 3.25508386e-01
7.40603805e-01 -8.69477749e-01 2.71891415e-01 7.86496460e-01
-2.32604116e-01 4.87887785e-02 -3.64535093e-01 9.54191208e-01
4.08888489e-01 -1.35200131e+00 4.76676494e-01 -1.72365353e-01
7.24657476e-01 2.69280493e-01 -3.89399678e-01 1.18238282e+00
7.34513044e-01 -8.58347774e-01 6.00811243e-01 6.81304455e-01
7.57409990e-01 -3.55382085e-01 9.14039314e-01 6.28798962e-01
-1.49348760e+00 -1.52021483e-01 -4.61365014e-01 2.93434739e-01
1.49597034e-01 3.37035745e-01 -7.19425559e-01 1.22942698e+00
6.92389190e-01 8.23357880e-01 -7.48971641e-01 7.98309386e-01
-1.38756052e-01 5.63811064e-01 -4.73247588e-01 -2.53956139e-01
4.95124310e-01 -5.73903210e-02 1.99903816e-01 1.43117523e+00
3.41734529e-01 1.01624444e-01 4.86274511e-01 2.96937615e-01
-3.10105920e-01 6.88976824e-01 -4.47532862e-01 -2.23116234e-01
4.28214163e-01 1.32063437e+00 -4.46363926e-01 -6.13580406e-01
-9.01190817e-01 8.49181950e-01 4.61608142e-01 7.87397623e-01
-5.21847725e-01 -9.28501308e-01 -6.73389584e-02 -1.15692541e-01
5.21445461e-02 -3.36677939e-01 -2.73390338e-02 -1.46090281e+00
2.16145873e-01 -3.81504238e-01 8.01708460e-01 -6.75893903e-01
-1.36230516e+00 7.53769815e-01 4.12579924e-01 -1.01374781e+00
6.74383575e-03 -9.36334193e-01 -5.77911176e-02 8.36644650e-01
-2.15636182e+00 -1.68070626e+00 -5.23258984e-01 8.69744182e-01
6.79720283e-01 -4.17023748e-01 4.14298773e-01 5.76477945e-01
-4.08457160e-01 4.18245047e-01 3.98099095e-01 2.33432591e-01
7.93897033e-01 -1.10103071e+00 -1.69504941e-01 4.00431126e-01
3.13214689e-01 5.79345584e-01 2.89499402e-01 -1.05399907e+00
-1.36987758e+00 -1.20525193e+00 1.22692990e+00 -6.31917596e-01
1.04284954e+00 -1.37984276e-01 -8.36712837e-01 1.10937726e+00
3.77810746e-01 -5.40398240e-01 1.05523300e+00 2.38084406e-01
-1.52579501e-01 -1.08492427e-01 -8.09969306e-01 4.58076119e-01
1.22705674e+00 -2.10937962e-01 -1.12279630e+00 4.82225239e-01
8.88123095e-01 -3.19789678e-01 -1.07195449e+00 3.79949689e-01
3.46820772e-01 -2.32746422e-01 9.58440959e-01 -9.34776366e-01
1.17746830e-01 -1.84878886e-01 -3.04549545e-01 -1.21063530e+00
-1.24961130e-01 -2.91441053e-01 -8.77517611e-02 1.47293806e+00
8.61000419e-01 -7.41827071e-01 5.69920719e-01 5.07331192e-01
-4.37822789e-01 -6.32870793e-01 -8.36654842e-01 -1.11442888e+00
1.71152741e-01 -4.87712055e-01 7.50351846e-01 1.14431560e+00
5.12796760e-01 6.35405600e-01 -4.66227680e-01 1.40203103e-01
4.30812649e-02 1.19668514e-01 5.75542092e-01 -1.22467470e+00
2.50454843e-02 -2.97715545e-01 -1.91942900e-01 -1.26346469e+00
5.57972610e-01 -1.25526261e+00 -2.77212977e-01 -1.99811435e+00
-4.20218259e-02 -6.75015032e-01 -1.62151009e-01 7.92436719e-01
-5.30808046e-02 -3.51916760e-01 -1.46902516e-01 2.81240225e-01
-8.22879195e-01 7.27891207e-01 9.95511770e-01 -2.57759422e-01
-3.34167212e-01 -2.58569300e-01 -7.88522422e-01 6.50210023e-01
8.22939813e-01 -3.11347276e-01 -6.70092463e-01 -8.18370342e-01
5.72629750e-01 -3.12154055e-01 6.24688268e-02 -6.82616055e-01
5.06895542e-01 -4.39942122e-01 -6.34127632e-02 -4.69070822e-01
1.49084136e-01 -1.18192816e+00 -1.66930839e-01 9.54948273e-03
-3.05170357e-01 -1.11502640e-01 5.06465912e-01 5.68707526e-01
-5.72893798e-01 -6.29197478e-01 5.88991754e-02 -4.49501246e-01
-1.38981140e+00 3.68350446e-01 -2.09815457e-01 4.50765759e-01
7.55023479e-01 -5.59736416e-02 -4.91996497e-01 -4.76864353e-02
-7.98305631e-01 7.20106661e-01 -9.76034403e-02 7.68058717e-01
4.13077563e-01 -1.18169689e+00 -6.83673739e-01 -1.87997356e-01
4.21964645e-01 -1.40742272e-01 2.95284003e-01 6.60194218e-01
-4.47943449e-01 7.96153426e-01 8.65977928e-02 5.89464456e-02
-9.15151656e-01 5.91572702e-01 2.44066808e-02 -6.27062917e-01
-3.00391197e-01 5.80427527e-01 -1.94093779e-01 -7.73767412e-01
1.78646266e-01 -3.19380641e-01 -8.28663766e-01 1.23069920e-01
3.25239658e-01 2.78762847e-01 3.72964144e-02 -6.65505588e-01
-1.20178923e-01 7.59195387e-01 4.55934368e-02 -1.59388155e-01
1.06661737e+00 -4.58843142e-01 -2.37371117e-01 3.93536240e-01
1.07850099e+00 2.79744089e-01 -3.15108836e-01 -7.93739676e-01
5.76711059e-01 -2.17684284e-01 6.24117330e-02 -1.12005341e+00
-7.38854229e-01 4.30746347e-01 1.51332900e-01 1.69406697e-01
1.22844100e+00 1.90915957e-01 8.43035340e-01 7.84284413e-01
6.08358979e-01 -1.45774484e+00 -2.57055342e-01 9.51993525e-01
8.06715727e-01 -1.19857001e+00 -1.63331449e-01 -6.08472109e-01
-8.71891975e-01 1.11778343e+00 8.52388203e-01 4.58160907e-01
1.05243587e+00 1.63073629e-01 1.62243813e-01 -3.81034762e-01
-6.56446934e-01 -8.79580498e-01 5.43864369e-01 5.21083474e-01
2.76976168e-01 -1.27805799e-01 -6.79597139e-01 1.02901983e+00
-5.82142472e-01 3.01133424e-01 2.37791970e-01 9.68139172e-01
-6.33133590e-01 -1.19076240e+00 6.37465268e-02 6.87499940e-01
-3.05343777e-01 -5.15677869e-01 -5.55297554e-01 1.06284571e+00
3.89036089e-01 1.04198647e+00 -2.55105078e-01 -6.51179731e-01
5.80871522e-01 3.46696675e-01 4.56397124e-02 -8.02053571e-01
-5.21394193e-01 -2.07288891e-01 4.74250048e-01 3.39235477e-02
-5.06802917e-01 -2.70693064e-01 -1.45017421e+00 -6.07468709e-02
-4.45494235e-01 5.33701658e-01 4.39602882e-01 1.12057781e+00
5.21621764e-01 2.64978558e-01 6.40722364e-02 6.15558140e-02
-1.72852635e-01 -1.18180478e+00 -7.37641275e-01 3.41047108e-01
-1.51018798e-01 -6.44945860e-01 -1.79335549e-01 2.99843431e-01] | [9.374539375305176, 8.417804718017578] |
2ea8c5f9-d448-46ba-aa30-b3aa8fcb5b07 | from-parse-execute-to-parse-execute-refine | 2305.03356 | null | https://arxiv.org/abs/2305.03356v1 | https://arxiv.org/pdf/2305.03356v1.pdf | From Parse-Execute to Parse-Execute-Refine: Improving Semantic Parser for Complex Question Answering over Knowledge Base | Parsing questions into executable logical forms has showed impressive results for knowledge-base question answering (KBQA). However, complex KBQA is a more challenging task that requires to perform complex multi-step reasoning. Recently, a new semantic parser called KoPL has been proposed to explicitly model the reasoning processes, which achieved the state-of-the-art on complex KBQA. In this paper, we further explore how to unlock the reasoning ability of semantic parsers by a simple proposed parse-execute-refine paradigm. We refine and improve the KoPL parser by demonstrating the executed intermediate reasoning steps to the KBQA model. We show that such simple strategy can significantly improve the ability of complex reasoning. Specifically, we propose three components: a parsing stage, an execution stage and a refinement stage, to enhance the ability of complex reasoning. The parser uses the KoPL to generate the transparent logical forms. Then, the execution stage aligns and executes the logical forms over knowledge base to obtain intermediate reasoning processes. Finally, the intermediate step-by-step reasoning processes are demonstrated to the KBQA model in the refinement stage. With the explicit reasoning processes, it is much easier to answer the complex questions. Experiments on benchmark dataset shows that the proposed PER-KBQA performs significantly better than the stage-of-the-art baselines on the complex KBQA. | ['Jian Yin', 'Hanjiang Lai', 'Linyin Luo', 'Wangzhen Guo'] | 2023-05-05 | null | null | null | null | ['knowledge-base-question-answering'] | ['natural-language-processing'] | [-1.14029229e-01 5.93043983e-01 2.29483396e-01 -5.86304963e-01
-1.23326433e+00 -7.66556025e-01 3.82558793e-01 -8.63127131e-03
-2.37571657e-01 5.90950012e-01 3.34025890e-01 -7.49710143e-01
3.68423760e-02 -1.20146608e+00 -7.76644170e-01 -8.28838050e-02
5.52889645e-01 8.98372114e-01 9.86359060e-01 -5.87456167e-01
-6.96134344e-02 1.39735350e-02 -1.22897923e+00 9.76453781e-01
1.18605733e+00 8.92121792e-01 1.13973424e-01 8.77643108e-01
-1.01030886e+00 1.50905120e+00 -5.79145014e-01 -9.93752003e-01
-1.72777865e-02 -4.62180823e-01 -1.54891527e+00 -6.57142043e-01
2.27482960e-01 -4.28305656e-01 5.01053371e-02 9.78284180e-01
9.71395969e-02 -6.94524497e-02 1.62349075e-01 -1.24166620e+00
-7.33850062e-01 8.32364738e-01 -1.19098714e-02 4.91060615e-02
8.68756950e-01 1.53795451e-01 1.37636554e+00 -5.34393668e-01
3.32974583e-01 1.63952398e+00 5.29854834e-01 7.78564394e-01
-5.47117651e-01 -6.60950184e-01 1.57026455e-01 6.05222464e-01
-9.18073833e-01 -9.95547771e-02 5.80293179e-01 -4.04168628e-02
1.28777146e+00 2.51900941e-01 4.48436856e-01 4.60037678e-01
-1.04713537e-01 1.00208008e+00 1.01146972e+00 -4.71511066e-01
1.54496402e-01 -7.64447823e-02 9.10051465e-01 9.90761220e-01
2.88390011e-01 -5.07176101e-01 -2.90516287e-01 -1.30179405e-01
4.54952002e-01 -4.06980515e-01 1.13560623e-02 6.62916899e-02
-7.61393189e-01 9.30435896e-01 4.39078391e-01 3.47302854e-02
-3.30992311e-01 1.70171037e-01 3.36287856e-01 3.04033130e-01
-1.80681884e-01 6.53908610e-01 -7.63976872e-01 -2.33767956e-01
-5.45714915e-01 5.02717972e-01 1.29655445e+00 9.17202890e-01
8.03014398e-01 -5.17933309e-01 -3.36639076e-01 5.45067966e-01
5.43594360e-01 6.57087862e-01 2.08367333e-01 -1.19248593e+00
8.76379967e-01 1.23410678e+00 9.62766819e-03 -5.50192535e-01
-3.60932469e-01 -4.30962779e-02 -3.63206297e-01 -2.45698467e-01
4.93797362e-01 1.48224253e-02 -7.83796251e-01 1.53848863e+00
7.48210073e-01 1.54868200e-01 7.43906736e-01 8.19332719e-01
1.30050194e+00 8.67642105e-01 5.98655283e-01 1.61436394e-01
2.13367081e+00 -1.55795455e+00 -9.48539257e-01 -3.12659651e-01
9.62934732e-01 -4.88616616e-01 1.46651554e+00 2.45312378e-01
-1.34536004e+00 -5.23003578e-01 -8.60344768e-01 -6.31983101e-01
-3.80298197e-01 -1.74028084e-01 9.29072618e-01 5.45664966e-01
-1.05724335e+00 -4.60551679e-02 -7.21183717e-01 -1.21018708e-01
4.26586568e-01 8.24129730e-02 -2.48529896e-01 -3.84412736e-01
-1.70403981e+00 8.83948565e-01 7.16831982e-01 9.68715921e-02
-6.93430007e-01 -8.27004850e-01 -1.13690174e+00 3.05549800e-01
9.70576167e-01 -1.30651605e+00 1.93684292e+00 -3.73561949e-01
-1.80489099e+00 6.03118658e-01 -3.32951784e-01 -3.96283686e-01
1.78942025e-01 -6.78260982e-01 -4.26558107e-01 4.25119460e-01
3.44129324e-01 4.64356482e-01 3.57992530e-01 -9.61295009e-01
-8.31234813e-01 -3.16580236e-01 1.07280731e+00 2.45828897e-01
2.90564209e-01 6.65981025e-02 -8.92202318e-01 -1.78364404e-02
1.41441420e-01 -6.40893519e-01 -9.15514380e-02 -4.52256799e-01
2.25313916e-03 -8.12057078e-01 3.90465796e-01 -8.53195608e-01
1.34127641e+00 -1.90010369e+00 -2.19850503e-02 -1.27155840e-01
2.62739658e-01 4.40761924e-01 -1.63425773e-01 3.02033633e-01
2.02894375e-01 1.78664625e-01 -1.66958824e-01 -3.52786481e-02
3.04097623e-01 5.91909826e-01 -6.54593110e-01 -5.71727633e-01
7.07759798e-01 1.57546306e+00 -9.74359453e-01 -7.51961410e-01
-2.34836608e-01 -2.17987612e-01 -8.88922513e-01 6.14854455e-01
-8.42357635e-01 3.91974635e-02 -8.89295340e-01 7.16210425e-01
7.50263512e-01 -4.69989866e-01 1.61315709e-01 -2.99925655e-01
3.68838727e-01 8.36835623e-01 -1.10454249e+00 1.81778395e+00
-5.46191096e-01 -6.10173643e-02 -4.42753248e-02 -7.48629272e-01
6.35737062e-01 2.68131137e-01 -4.53013331e-01 -8.34056556e-01
-2.00052798e-01 2.16740146e-01 3.89265753e-02 -8.72403324e-01
4.09314245e-01 -4.39930052e-01 -4.44492549e-01 4.00869489e-01
9.59976912e-02 -5.51266074e-01 3.29662740e-01 5.74697793e-01
1.18047380e+00 4.64736462e-01 3.70521516e-01 1.27136111e-01
1.04431653e+00 5.31078815e-01 4.74703699e-01 7.87563205e-01
-5.57600372e-02 -5.36029711e-02 6.87879980e-01 -4.67331499e-01
-4.55334574e-01 -1.24681234e+00 5.06195903e-01 1.17402709e+00
2.49610394e-01 -7.14645028e-01 -1.00943375e+00 -1.15468597e+00
-2.41596058e-01 8.49404395e-01 -2.39420757e-01 -1.41325653e-01
-8.79050374e-01 -3.42774183e-01 1.08283901e+00 7.52720714e-01
1.19363761e+00 -1.35176361e+00 -5.81898928e-01 2.92176843e-01
-5.63152909e-01 -1.50899732e+00 2.37609446e-01 -2.25167170e-01
-6.72563851e-01 -1.30167425e+00 1.32891327e-01 -6.89797282e-01
3.01897973e-01 1.02423476e-02 1.32901609e+00 3.82594377e-01
2.82352418e-01 3.39938730e-01 -8.12857211e-01 -4.28556412e-01
-5.68611205e-01 9.74755827e-03 -6.13103926e-01 -5.14201164e-01
4.87259239e-01 -1.49806231e-01 -5.41301727e-01 4.40873317e-02
-1.03484857e+00 4.60379869e-01 8.05317640e-01 4.22045529e-01
4.23912704e-01 1.92210481e-01 4.41155940e-01 -1.23300135e+00
7.80654907e-01 -2.79773802e-01 -4.11548167e-01 7.04028845e-01
-2.59606779e-01 5.45310259e-01 9.99812126e-01 6.26548156e-02
-1.66290641e+00 -4.18513626e-01 -6.10526919e-01 2.50754595e-01
-2.51095325e-01 7.22168505e-01 -5.00163138e-01 3.16385001e-01
4.05722558e-01 8.40468109e-02 -4.56666529e-01 -3.86045039e-01
7.65491009e-01 4.99414027e-01 6.71688378e-01 -1.24743187e+00
1.04191244e+00 2.96121001e-01 -1.04708746e-01 -1.06753334e-01
-1.38316715e+00 -4.99377012e-01 -2.77818024e-01 2.08665282e-01
1.18664813e+00 -8.40588033e-01 -9.50260162e-01 3.32863688e-01
-1.41644347e+00 -6.06819689e-01 -3.73036712e-01 6.41567409e-02
-5.34547329e-01 4.86761719e-01 -9.20720994e-01 -5.95624268e-01
-7.01071084e-01 -1.07761312e+00 1.11691070e+00 5.78385472e-01
-7.93690905e-02 -7.85712004e-01 1.33967608e-01 1.17624640e+00
2.85662979e-01 -3.24686587e-01 1.48794329e+00 -8.81707430e-01
-7.50907063e-01 6.44633546e-02 -4.61052090e-01 3.92761290e-01
-1.17643453e-01 -4.18287992e-01 -9.24023807e-01 4.20210958e-01
6.73169792e-02 -4.94971842e-01 5.02610266e-01 -3.71082753e-01
9.66252685e-01 -2.75506139e-01 -2.86409967e-02 3.64003599e-01
1.13847899e+00 2.22897112e-01 9.31044281e-01 5.49534857e-01
5.72528720e-01 5.33883095e-01 8.73615444e-01 -4.18853432e-01
1.17380476e+00 2.29430348e-01 8.52600783e-02 6.74414411e-02
-3.09402674e-01 -4.93727177e-01 2.86274016e-01 1.09453070e+00
1.09913701e-03 1.28945753e-01 -1.21369922e+00 3.49510401e-01
-1.99727571e+00 -7.89513111e-01 -4.07022417e-01 1.38754535e+00
1.14960384e+00 1.64873183e-01 -3.08026284e-01 4.07089889e-02
1.60782754e-01 -7.83403125e-03 -3.04198384e-01 -7.36600637e-01
1.47060946e-01 7.43528128e-01 -2.12580964e-01 8.04994464e-01
-6.87162578e-01 1.64107776e+00 5.63066673e+00 6.73425853e-01
-4.81699824e-01 1.49913311e-01 -2.86513064e-02 5.10426700e-01
-4.56805497e-01 3.62006962e-01 -1.13770711e+00 -3.03895008e-02
1.07931209e+00 1.11418188e-01 4.75681990e-01 8.41885448e-01
-1.81281045e-01 -2.33630523e-01 -1.01054633e+00 6.36587858e-01
-1.28791019e-01 -1.31539869e+00 7.96844602e-01 -6.89974129e-01
2.12710530e-01 -4.67832744e-01 -4.57810909e-01 1.15297949e+00
6.36988759e-01 -8.63154709e-01 5.37184298e-01 5.07658362e-01
2.49811545e-01 -4.94248539e-01 1.01899445e+00 5.52882075e-01
-1.48845017e+00 -2.50137478e-01 -3.55024189e-01 -3.28406662e-01
3.99164021e-01 2.89208412e-01 -9.11018968e-01 1.14571059e+00
8.39513540e-01 7.56822973e-02 -7.22189128e-01 4.81304467e-01
-1.03151035e+00 7.25053191e-01 -6.50085658e-02 -1.62265241e-01
4.59572971e-01 -2.26609007e-01 3.19274899e-04 1.11174047e+00
-1.08000405e-01 8.57877374e-01 3.29125524e-02 8.49967122e-01
-9.44690034e-02 1.05999298e-01 5.82897924e-02 -9.81804654e-02
2.82740116e-01 1.27474749e+00 -3.23593229e-01 -8.33949566e-01
-7.01310635e-01 1.00876784e+00 8.69368255e-01 3.56110483e-01
-8.85193884e-01 -4.06300187e-01 4.31132644e-01 -1.74388140e-01
2.02524945e-01 -5.57929464e-02 -8.06153864e-02 -1.13744628e+00
3.84556383e-01 -1.18302059e+00 1.02636409e+00 -1.39326394e+00
-1.12645996e+00 5.52325189e-01 1.79613084e-01 -1.59556687e-01
-1.80999205e-01 -7.01400936e-01 -5.83778679e-01 1.02606010e+00
-2.08121347e+00 -1.34931755e+00 -6.02381825e-01 7.73354769e-01
6.65791452e-01 3.69886279e-01 9.88726735e-01 3.14652361e-02
-2.94648945e-01 2.93688625e-01 -7.95703530e-01 3.38079900e-01
3.89240742e-01 -1.42090511e+00 5.20948052e-01 8.42341065e-01
-2.14903504e-02 8.35912466e-01 2.64302343e-01 -7.12620497e-01
-1.56349766e+00 -9.55322266e-01 8.80982637e-01 -7.92537987e-01
8.72758865e-01 -1.40099943e-01 -1.29674900e+00 9.42550778e-01
8.51030126e-02 -2.78358907e-01 7.15661168e-01 8.95381570e-02
-7.24211752e-01 -1.51758656e-01 -9.79781687e-01 6.35517120e-01
9.42357063e-01 -8.06648672e-01 -1.74270809e+00 3.25678918e-03
1.36411297e+00 -5.19415975e-01 -7.39429653e-01 5.61209559e-01
4.17663068e-01 -7.14355469e-01 9.46005046e-01 -1.03520191e+00
5.48955321e-01 -8.03498566e-01 -2.43296102e-01 -8.48918498e-01
-2.98955441e-01 -3.93028170e-01 -3.42453331e-01 1.23675168e+00
5.25614500e-01 -7.09734201e-01 6.13293231e-01 8.40315878e-01
-2.99610257e-01 -7.97917962e-01 -7.31377244e-01 -3.56020033e-01
1.19345963e-01 -7.72341490e-01 1.00543356e+00 5.93395531e-01
1.03840403e-01 6.99002683e-01 5.19356728e-01 6.51906788e-01
2.11981386e-01 3.57584804e-01 1.02407813e+00 -1.10964239e+00
-4.93021607e-01 -1.27203315e-01 6.04212694e-02 -1.39303207e+00
3.37868720e-01 -8.33472133e-01 3.60021591e-02 -2.22314954e+00
2.29091123e-01 -4.43854213e-01 -1.15868188e-01 6.83702290e-01
-8.26380253e-01 -3.36633265e-01 2.31848300e-01 5.32457940e-02
-1.10222030e+00 1.91645607e-01 1.50767839e+00 -1.42852694e-01
-9.96448919e-02 -2.37241000e-01 -9.62704301e-01 7.40727842e-01
7.14736760e-01 -2.77295649e-01 -7.32679665e-01 -5.86756766e-01
7.86303222e-01 1.73594877e-01 2.75290459e-01 -9.10830855e-01
4.80830401e-01 -2.22373515e-01 -2.23747641e-01 -5.16425669e-01
2.08964109e-01 -6.86552763e-01 -1.48530856e-01 4.66285408e-01
-1.64409623e-01 1.43336803e-01 3.82799685e-01 2.75014728e-01
-4.37774003e-01 -4.67962265e-01 3.42373431e-01 -4.09379601e-01
-9.78181481e-01 -1.10291652e-02 -5.61491586e-02 5.61417103e-01
9.73304033e-01 1.13966890e-01 -7.54190266e-01 -1.42209902e-01
-5.59669077e-01 7.12472379e-01 -1.04738303e-01 2.37655416e-01
4.34378952e-01 -9.15365458e-01 -4.68144774e-01 -7.35763228e-03
1.29157305e-01 7.23805904e-01 2.67626852e-01 6.40312970e-01
-1.03373075e+00 6.53212488e-01 5.55521622e-02 -2.75950760e-01
-1.19323742e+00 5.72786987e-01 4.32253897e-01 -1.01316810e+00
-4.25301373e-01 9.02502596e-01 3.38044494e-01 -7.74717212e-01
-1.00760803e-01 -7.53676474e-01 -4.55662757e-01 -3.06039572e-01
6.95376039e-01 3.81134115e-02 7.96655640e-02 -1.22771941e-01
-4.25862044e-01 6.14895344e-01 -1.61607251e-01 -7.32882395e-02
1.06572592e+00 2.79232264e-02 -4.57805932e-01 2.65069511e-02
5.50683439e-01 3.04990932e-02 -6.47228360e-01 -2.91981608e-01
2.43112296e-01 1.22394627e-02 -4.47546393e-01 -1.24638414e+00
-4.14778233e-01 9.43316400e-01 -2.17976406e-01 2.84926385e-01
1.15111506e+00 4.36607957e-01 1.18688345e+00 8.53245676e-01
4.52761233e-01 -7.53289461e-01 1.20806962e-01 9.08611119e-01
7.17922151e-01 -9.60603952e-01 -3.09428930e-01 -9.62905347e-01
-6.41838193e-01 1.05902088e+00 8.92942071e-01 1.62769213e-01
2.16031745e-01 2.00810626e-01 4.29710120e-01 -4.18784767e-01
-8.92003059e-01 -2.69004703e-01 2.00855315e-01 4.33432668e-01
8.03197771e-02 7.12164491e-02 -3.07058930e-01 1.30918825e+00
-4.84815001e-01 1.04666643e-01 1.83911011e-01 1.05169117e+00
-6.01405144e-01 -1.15283203e+00 -4.30382729e-01 1.29500613e-01
-4.80706275e-01 -4.11132246e-01 -5.25128007e-01 1.02569056e+00
-4.45099175e-02 1.23886287e+00 -2.03883126e-01 -1.30691864e-02
7.75945783e-01 6.42433465e-01 5.34181416e-01 -7.74474680e-01
-7.78236985e-01 -7.22855806e-01 5.58394253e-01 -8.73523772e-01
-3.57999176e-01 3.18804421e-02 -2.02357125e+00 -1.36175275e-01
-2.13217050e-01 6.91337109e-01 4.01831180e-01 1.41958702e+00
3.40162247e-01 7.10239589e-01 -1.00664236e-01 2.78174400e-01
-5.41065693e-01 -7.30918407e-01 3.69392931e-02 5.55187345e-01
1.04777012e-02 -2.43206531e-01 -1.40702888e-01 7.41873682e-02] | [10.456854820251465, 7.887436389923096] |
838321c3-9437-40fe-b949-1f651ac14146 | exploiting-cross-lingual-subword-similarities | 1812.09617 | null | https://arxiv.org/abs/1812.09617v4 | https://arxiv.org/pdf/1812.09617v4.pdf | Exploiting Cross-Lingual Subword Similarities in Low-Resource Document Classification | Text classification must sometimes be applied in a low-resource language with no labeled training data. However, training data may be available in a related language. We investigate whether character-level knowledge transfer from a related language helps text classification. We present a cross-lingual document classification framework (CACO) that exploits cross-lingual subword similarity by jointly training a character-based embedder and a word-based classifier. The embedder derives vector representations for input words from their written forms, and the classifier makes predictions based on the word vectors. We use a joint character representation for both the source language and the target language, which allows the embedder to generalize knowledge about source language words to target language words with similar forms. We propose a multi-task objective that can further improve the model if additional cross-lingual or monolingual resources are available. Experiments confirm that character-level knowledge transfer is more data-efficient than word-level transfer between related languages. | ['Jordan Boyd-Graber', 'Mozhi Zhang', 'Yoshinari Fujinuma'] | 2018-12-22 | null | https://openreview.net/forum?id=S1e_H3AqYQ | https://openreview.net/pdf?id=S1e_H3AqYQ | null | ['cross-lingual-document-classification'] | ['natural-language-processing'] | [ 9.14343446e-02 -4.23752218e-01 -7.21359193e-01 -5.99660337e-01
-9.24848437e-01 -7.11483240e-01 5.33160925e-01 3.63280207e-01
-8.97564232e-01 7.55261242e-01 2.91227788e-01 -4.39875901e-01
4.26846087e-01 -8.95983160e-01 -5.87640047e-01 -2.44852796e-01
3.34879845e-01 4.16109800e-01 -8.75856280e-02 -3.66818994e-01
2.57661492e-01 1.13188758e-01 -1.06656778e+00 8.18060994e-01
1.23291874e+00 5.91009498e-01 6.89820588e-01 4.46346104e-01
-7.93266892e-01 4.66198027e-01 -4.24729973e-01 -4.95743692e-01
1.27549142e-01 -5.07242620e-01 -1.09203315e+00 1.29678249e-02
3.29631746e-01 2.03367844e-02 1.28274262e-01 9.51162696e-01
1.40183479e-01 1.45506829e-01 1.04251981e+00 -8.62849414e-01
-1.17166150e+00 9.34523344e-01 -4.31474268e-01 1.36371568e-01
2.64512241e-01 -2.79781312e-01 1.07235527e+00 -1.17182636e+00
6.75301671e-01 1.49086666e+00 6.75091684e-01 5.54049134e-01
-1.23990607e+00 -6.69812500e-01 4.62462515e-01 3.12670767e-01
-1.46133721e+00 -1.79985911e-01 8.26230109e-01 -5.23862302e-01
1.16486228e+00 -1.97324976e-01 2.78057218e-01 1.10157847e+00
2.37654433e-01 9.63420630e-01 1.19134033e+00 -1.10572815e+00
-3.74801680e-02 7.07228363e-01 5.36586285e-01 7.23566294e-01
2.15966597e-01 -2.37307236e-01 -6.34240746e-01 1.28343757e-02
2.96637505e-01 -2.06568405e-01 -2.54225641e-01 -2.32211947e-01
-1.06227827e+00 1.19689047e+00 2.74382114e-01 7.34175801e-01
-1.40191242e-01 -2.11148605e-01 7.07713366e-01 5.89597225e-01
8.11609805e-01 6.10805154e-01 -8.11645329e-01 2.03949153e-01
-7.23521411e-01 -3.49496394e-01 9.19729710e-01 1.13228583e+00
1.13739181e+00 6.54208362e-02 -9.61592868e-02 1.25917137e+00
3.56666327e-01 6.89072430e-01 1.22575366e+00 -1.04098670e-01
7.51098454e-01 4.52663988e-01 -4.27987039e-01 -6.15427911e-01
-5.61905690e-02 -2.25417763e-01 -5.35977662e-01 -3.08832973e-01
2.10327581e-01 -3.37218374e-01 -5.40080130e-01 1.79557014e+00
5.26860356e-02 -3.56783748e-01 6.22978508e-01 4.11098391e-01
5.21721423e-01 9.79480624e-01 2.39607126e-01 -1.64628446e-01
1.58505535e+00 -1.14380896e+00 -4.90774423e-01 -5.78763783e-01
1.41956675e+00 -8.37683260e-01 1.32180882e+00 6.17909245e-02
-6.07038558e-01 -7.79599726e-01 -9.89117444e-01 -3.84718060e-01
-9.70879793e-01 3.33644509e-01 1.92702040e-01 7.90129364e-01
-7.52506018e-01 2.66920149e-01 -3.63085806e-01 -6.49440169e-01
7.33096078e-02 -1.03478357e-01 -5.20917714e-01 -4.53641444e-01
-1.46867895e+00 1.23309982e+00 8.32841933e-01 -4.02939230e-01
-4.05382723e-01 -7.23429203e-01 -1.16739368e+00 -1.71706259e-01
-4.36194986e-02 -4.70592976e-01 1.05683720e+00 -1.31968498e+00
-1.40059412e+00 1.04528701e+00 -2.13508800e-01 -1.14472903e-01
1.77227467e-01 -7.44677410e-02 -4.60932732e-01 -1.46027192e-01
3.75833064e-01 5.20247579e-01 7.74600208e-01 -1.12196541e+00
-7.81095624e-01 -4.37520802e-01 -3.23254675e-01 3.99921000e-01
-1.00208676e+00 4.77370881e-02 -3.66100580e-01 -9.63479698e-01
-4.35166031e-01 -7.96022594e-01 2.87441939e-01 4.12554294e-02
-1.03655435e-01 -7.13892877e-01 4.34660375e-01 -7.38128841e-01
1.10400355e+00 -1.96199107e+00 1.68098256e-01 9.38151404e-02
-3.48706603e-01 3.59139144e-01 -6.46358967e-01 6.63509667e-01
1.13411635e-01 1.62693769e-01 1.03134743e-03 -2.65465945e-01
-1.62785560e-01 3.47260475e-01 -3.31616521e-01 1.27797589e-01
3.49067032e-01 9.12437379e-01 -8.40792835e-01 -4.32932526e-01
-1.25994965e-01 4.14409995e-01 -3.73622626e-01 1.23018488e-01
-7.78611079e-02 -3.47192176e-02 -4.08092409e-01 2.14837372e-01
3.63668054e-01 6.74869716e-02 4.69035834e-01 -1.46124750e-01
-9.82120931e-02 5.92822373e-01 -6.73366129e-01 1.79133177e+00
-1.32342696e+00 5.65749288e-01 -2.93507397e-01 -1.22787952e+00
1.12639892e+00 2.59633899e-01 -1.90496519e-01 -6.06594622e-01
9.16324481e-02 3.68016064e-01 9.00664181e-02 -4.66212630e-01
5.56882203e-01 -3.81677210e-01 -3.25417489e-01 7.03587592e-01
2.79874921e-01 -1.77069724e-01 2.80759603e-01 -9.72430408e-02
5.81405938e-01 1.38989374e-01 5.22200584e-01 -5.89381397e-01
7.73424506e-01 1.05712585e-01 2.35157251e-01 4.81905490e-01
1.89015254e-01 -1.56366080e-01 8.49971548e-02 -2.27950305e-01
-9.47735846e-01 -6.66378558e-01 -4.29416865e-01 1.84227479e+00
-1.54418558e-01 -5.33828259e-01 -5.22156477e-01 -9.20458853e-01
3.01677912e-01 1.06208968e+00 -6.97189867e-01 -3.52821141e-01
-4.79577005e-01 -3.79494339e-01 5.88225305e-01 7.84446836e-01
2.22081274e-01 -9.63180959e-01 1.00332171e-01 3.45782250e-01
-2.93648848e-03 -1.08958924e+00 -8.52391779e-01 3.74039829e-01
-7.76875556e-01 -5.77601254e-01 -1.03072202e+00 -1.38310122e+00
7.95103550e-01 1.94627240e-01 1.06619108e+00 -1.01935521e-01
-9.01153162e-02 3.65773171e-01 -6.71365261e-01 -3.68027061e-01
-7.33263195e-01 5.90752482e-01 2.72957057e-01 1.12146817e-01
8.06692362e-01 1.96444225e-02 2.54183561e-01 1.96355224e-01
-8.03995430e-01 4.67303023e-02 4.48619753e-01 8.22180748e-01
3.63188058e-01 -2.60853589e-01 7.50035286e-01 -1.06586468e+00
9.28490758e-01 -5.40118098e-01 -2.71004468e-01 6.69755816e-01
-5.52530348e-01 5.52038610e-01 9.52022910e-01 -7.76279926e-01
-1.03274095e+00 -6.17359988e-02 3.46485190e-02 -3.44992727e-02
-1.14761360e-01 1.00886583e+00 -1.58086911e-01 2.60861278e-01
7.82632589e-01 4.93494421e-01 -1.91943437e-01 -7.12453544e-01
6.54878139e-01 1.08220756e+00 1.22889884e-01 -1.03473842e+00
6.31223679e-01 -1.19797930e-01 -6.35574639e-01 -1.16366339e+00
-9.77733016e-01 -5.21274567e-01 -1.14531779e+00 1.86792031e-01
8.65585268e-01 -1.01459575e+00 3.31601948e-02 4.14320499e-01
-1.44416535e+00 -4.04339105e-01 -1.91754341e-01 7.48690963e-01
-1.63496435e-01 3.42636317e-01 -5.51520646e-01 -3.84691060e-01
-3.17175657e-01 -8.53362322e-01 7.85174251e-01 -2.36787841e-01
-3.63871336e-01 -1.63113415e+00 2.67024517e-01 5.34515269e-02
3.17725629e-01 -6.61236584e-01 1.51926959e+00 -1.23368621e+00
1.53934821e-01 -2.65528560e-01 -2.24509686e-01 7.08426058e-01
4.94253278e-01 -3.07638288e-01 -7.97670364e-01 -4.56591874e-01
-1.31145522e-01 -7.44383454e-01 7.91886151e-01 -1.15998656e-01
8.97823215e-01 -3.81226540e-01 -3.27707857e-01 4.80882734e-01
1.33484352e+00 -4.42300215e-02 1.75636858e-02 2.61945426e-01
8.81630599e-01 9.82001185e-01 2.83722162e-01 2.12084260e-02
6.51003540e-01 5.07707238e-01 -5.87436080e-01 6.23343438e-02
-3.52993369e-01 -3.84003162e-01 7.20847428e-01 1.61571074e+00
3.52201730e-01 -3.76634926e-01 -1.14006460e+00 6.52629614e-01
-1.47219706e+00 -4.34013367e-01 3.76536064e-02 2.04619360e+00
1.42500579e+00 -2.59446144e-01 -1.14461526e-01 -1.34783566e-01
7.52076268e-01 -1.72915995e-01 -5.81640482e-01 -5.78029931e-01
-1.28184348e-01 1.87106013e-01 4.07124609e-01 7.32633829e-01
-8.42104852e-01 1.40082633e+00 5.81063223e+00 1.15183902e+00
-1.34699428e+00 2.35780329e-01 2.05094114e-01 4.00835097e-01
-4.44523305e-01 -2.47708976e-01 -1.26571250e+00 3.73688161e-01
1.02110004e+00 -7.87663281e-01 7.00527504e-02 9.45464551e-01
-3.95537704e-01 2.74693340e-01 -1.52039540e+00 8.80193472e-01
5.97614169e-01 -1.01092887e+00 6.29127443e-01 -5.12504466e-02
6.74458444e-01 7.67065212e-02 -5.59209920e-02 6.77236617e-01
4.12982464e-01 -7.86183119e-01 5.50899267e-01 2.38890424e-01
9.53189611e-01 -7.96999872e-01 6.80014908e-01 5.33185124e-01
-1.35700417e+00 4.11532581e-01 -6.30112946e-01 1.47566170e-01
-1.54063106e-01 1.23769514e-01 -8.09007823e-01 5.12613833e-01
1.73176378e-01 1.04056954e+00 -7.75398791e-01 3.67808729e-01
-4.26903129e-01 4.57597941e-01 -6.63694460e-03 -4.50825691e-01
2.45772332e-01 -6.09612986e-02 3.33418250e-02 1.73907125e+00
3.25119227e-01 -1.67342469e-01 6.91816330e-01 5.35117805e-01
-3.44821960e-01 1.06598592e+00 -8.03492844e-01 -1.99760810e-01
2.93593287e-01 1.00057900e+00 -4.32841927e-01 -6.62462890e-01
-8.23818862e-01 1.43103719e+00 8.53670657e-01 2.92224348e-01
5.30643202e-03 -6.87762082e-01 4.38954592e-01 -2.16611996e-01
2.02053964e-01 -3.04056138e-01 -1.99497908e-01 -1.50843978e+00
6.32637227e-03 -7.68204451e-01 3.83839428e-01 -6.71052754e-01
-2.11892223e+00 9.33268070e-01 -3.62112112e-02 -1.09453702e+00
-5.15728891e-01 -1.16481483e+00 -2.14071184e-01 1.25839567e+00
-1.81917083e+00 -1.34933507e+00 2.45930418e-01 8.09034526e-01
8.94160092e-01 -7.39687324e-01 1.26883864e+00 6.84916675e-02
-3.75827581e-01 9.92149413e-01 5.23276925e-01 6.18317187e-01
1.06313241e+00 -1.11812091e+00 2.71953791e-01 5.26869655e-01
3.96382898e-01 7.02062607e-01 6.68775663e-02 -6.67235792e-01
-1.27287877e+00 -1.30965590e+00 1.42651677e+00 -3.94307911e-01
1.02820015e+00 -6.34973347e-01 -1.41883862e+00 7.06683338e-01
4.04790789e-01 -1.16758339e-01 1.31307876e+00 4.61276323e-01
-9.52982128e-01 1.08583197e-01 -6.80379570e-01 5.73196590e-01
5.75016856e-01 -1.28036046e+00 -9.80014145e-01 5.58265865e-01
5.93585193e-01 2.71472454e-01 -8.26432943e-01 -2.73706108e-01
4.19069797e-01 -3.54047194e-02 5.92739642e-01 -9.82245684e-01
4.43244815e-01 5.58523312e-02 -4.46036935e-01 -1.87634027e+00
-2.42204756e-01 1.09353714e-01 4.37566072e-01 1.33892763e+00
7.88645744e-01 -7.08814800e-01 1.41878694e-01 4.79808390e-01
-3.77660468e-02 -3.50455880e-01 -7.71187961e-01 -1.36805260e+00
1.10910177e+00 -6.63884461e-01 2.84316599e-01 1.42847395e+00
3.70922923e-01 1.00672197e+00 -1.63483560e-01 -5.25447391e-02
2.66472697e-01 1.16267294e-01 4.11102444e-01 -1.40129077e+00
-1.33895189e-01 -4.20502961e-01 -2.10654229e-01 -1.13869214e+00
1.07097638e+00 -1.88114989e+00 1.85454309e-01 -1.22406793e+00
1.39908671e-01 -4.50252682e-01 -2.39383757e-01 8.63498807e-01
-2.73784429e-01 1.61747336e-01 1.47945404e-01 2.41290241e-01
-2.94269234e-01 7.77885139e-01 1.02409673e+00 -2.19277620e-01
-1.61504894e-01 -2.78049767e-01 -6.10289037e-01 8.13847780e-01
7.43433714e-01 -7.54609585e-01 -2.24292919e-01 -8.69619071e-01
7.39890849e-03 -3.15902472e-01 -2.68403918e-01 -5.60938537e-01
1.80924132e-01 -1.50650829e-01 3.92263949e-01 -1.10753223e-01
6.08191714e-02 -7.90804207e-01 -7.90634394e-01 4.29877281e-01
-8.30469310e-01 5.93134575e-02 3.65963817e-01 3.82146508e-01
-1.60890698e-01 -7.33304977e-01 8.23757052e-01 -5.38992435e-02
-7.19918728e-01 1.14327170e-01 -8.51140440e-01 3.06959301e-01
7.85775542e-01 -1.13711335e-01 -1.89001545e-01 -9.48488563e-02
-4.13033754e-01 8.32359716e-02 4.17449713e-01 9.28068101e-01
3.42490256e-01 -1.62065935e+00 -1.09679210e+00 4.93509024e-01
7.60848045e-01 -8.39502215e-01 -2.69199669e-01 1.21958308e-01
-1.87153697e-01 6.07918024e-01 -2.83978432e-01 -3.95185649e-01
-1.13306582e+00 6.72181249e-01 2.00835884e-01 -1.30474418e-01
-2.85078913e-01 7.69921184e-01 2.63395309e-01 -9.67924953e-01
3.73334810e-02 -3.08854789e-01 -2.44057715e-01 4.85921741e-01
6.48787260e-01 1.84190797e-03 1.76269710e-01 -9.12639737e-01
-2.86437958e-01 8.47636044e-01 -5.96592963e-01 -3.29428375e-01
1.07179499e+00 -2.26233929e-01 -3.86927612e-02 9.64787543e-01
1.59959042e+00 8.75576809e-02 -5.86930394e-01 -1.01270914e+00
1.84245273e-01 -2.38375321e-01 2.30034947e-01 -7.22972929e-01
-7.53616273e-01 1.21240067e+00 2.91702479e-01 -9.40264836e-02
6.52806818e-01 6.08853064e-02 6.55445218e-01 9.31106806e-01
4.25063998e-01 -1.31122816e+00 8.25571716e-02 1.13137949e+00
9.49238837e-01 -1.24778831e+00 -2.35163957e-01 -3.25464189e-01
-8.02144468e-01 1.25550520e+00 6.41886473e-01 1.46942614e-02
9.10656273e-01 -3.82550508e-02 3.84130478e-01 1.98781013e-01
-7.45201528e-01 -2.54383773e-01 6.65300012e-01 7.99646080e-01
1.00906706e+00 3.86298820e-02 -3.00897092e-01 7.47191608e-01
-2.95972824e-01 -3.78984421e-01 2.51168340e-01 9.98834312e-01
-4.57695037e-01 -1.63040805e+00 -2.91058421e-01 2.78163612e-01
-1.93111852e-01 -7.11212397e-01 -4.70818788e-01 5.68584442e-01
1.00671425e-01 8.73600900e-01 2.66590595e-01 -1.15176179e-01
1.64329886e-01 7.68354833e-01 4.16977555e-01 -1.28772521e+00
-6.11065328e-01 -8.77314657e-02 4.14250530e-02 4.79437225e-02
-2.89728731e-01 -5.30488849e-01 -1.05768239e+00 -1.00720890e-01
-3.77480924e-01 3.97379249e-01 7.53613114e-01 1.21307409e+00
1.31719932e-01 2.97747225e-01 6.23556972e-01 -5.91493130e-01
-4.29277867e-01 -1.11122382e+00 -7.74802864e-01 2.20592946e-01
1.60051316e-01 -2.14640766e-01 -9.51188877e-02 4.29849744e-01] | [10.949043273925781, 9.936883926391602] |
31d69d60-6560-484e-8ef9-d74ce2a26091 | feature-engineering-for-predictive-modeling | 1709.07150 | null | http://arxiv.org/abs/1709.07150v1 | http://arxiv.org/pdf/1709.07150v1.pdf | Feature Engineering for Predictive Modeling using Reinforcement Learning | Feature engineering is a crucial step in the process of predictive modeling.
It involves the transformation of given feature space, typically using
mathematical functions, with the objective of reducing the modeling error for a
given target. However, there is no well-defined basis for performing effective
feature engineering. It involves domain knowledge, intuition, and most of all,
a lengthy process of trial and error. The human attention involved in
overseeing this process significantly influences the cost of model generation.
We present a new framework to automate feature engineering. It is based on
performance driven exploration of a transformation graph, which systematically
and compactly enumerates the space of given options. A highly efficient
exploration strategy is derived through reinforcement learning on past
examples. | ['Horst Samulowitz', 'Deepak Turaga', 'Udayan Khurana'] | 2017-09-21 | null | null | null | null | ['automated-feature-engineering'] | ['methodology'] | [ 3.86476755e-01 6.06251806e-02 -1.53634414e-01 -2.45696291e-01
-5.35864770e-01 -5.38583100e-01 7.71333456e-01 1.83778375e-01
-1.83542833e-01 8.85473669e-01 -2.66057640e-01 -5.43146849e-01
-6.25521719e-01 -8.85393023e-01 -5.94562709e-01 -3.01948339e-01
1.91363364e-01 5.12075305e-01 -8.81836563e-02 -2.24090427e-01
7.73846507e-01 6.57257378e-01 -1.57527435e+00 -3.84699047e-01
1.18436384e+00 1.06972086e+00 2.41285756e-01 4.45803076e-01
-1.47107348e-01 3.97612303e-01 -3.91513854e-01 -2.08466113e-01
1.97975129e-01 -6.35586917e-01 -8.79978597e-01 2.39257812e-01
-3.04343551e-01 8.56417865e-02 1.07103065e-01 9.92267728e-01
-3.17353159e-02 3.10689092e-01 6.82201028e-01 -1.29439068e+00
-3.59831959e-01 5.28166175e-01 9.95178893e-02 7.86356926e-02
4.27876234e-01 4.35691357e-01 1.08378446e+00 -9.47998464e-01
3.90564173e-01 9.46404934e-01 3.26950073e-01 2.27422699e-01
-1.39086580e+00 -4.52320844e-01 -3.42275761e-02 5.44915237e-02
-1.46989822e+00 -2.09690854e-01 8.36148262e-01 -7.32662976e-01
1.07850289e+00 4.33460057e-01 1.35586035e+00 5.04544616e-01
5.08606017e-01 5.34803629e-01 8.52627397e-01 -7.78492153e-01
5.37662029e-01 3.47416043e-01 -1.11803733e-01 5.55940151e-01
3.87872607e-01 7.30892301e-01 -4.70317751e-01 -3.52626354e-01
6.74616337e-01 4.01517861e-02 -7.91799054e-02 -9.47336435e-01
-6.27215803e-01 1.00048721e+00 1.93330958e-01 1.08305052e-01
-3.99581224e-01 1.83415830e-01 -9.42345634e-02 4.73602235e-01
-7.67500699e-02 1.44333971e+00 -4.92643714e-01 -3.36489737e-01
-9.80604231e-01 3.59476864e-01 8.63923967e-01 7.08700657e-01
8.98636818e-01 2.30983183e-01 1.02207690e-01 3.64613831e-01
4.29595411e-01 2.42955521e-01 1.98219970e-01 -9.38520789e-01
7.91201517e-02 1.01250899e+00 4.11469936e-01 -6.61357999e-01
-2.36506656e-01 -6.14031971e-01 -1.19327314e-01 5.46285510e-01
2.17806965e-01 -3.38341504e-01 -9.12305355e-01 1.49870527e+00
2.81049430e-01 -3.51004571e-01 -2.15407982e-01 4.24424767e-01
-6.70300797e-02 5.92406631e-01 1.76058426e-01 -4.37369198e-01
7.47987270e-01 -7.55673766e-01 -6.13169014e-01 -4.02600467e-01
4.79769558e-01 -4.05667275e-01 1.10779464e+00 4.27082717e-01
-1.19023824e+00 -3.50002468e-01 -1.24740911e+00 5.07378817e-01
-4.09965515e-01 -3.14941145e-02 9.27787423e-01 6.77974641e-01
-7.42137492e-01 1.15336132e+00 -8.52813184e-01 -1.79171264e-01
2.77326673e-01 7.52031744e-01 -1.67902321e-01 1.03401899e-01
-1.21676075e+00 1.31311142e+00 7.14554846e-01 2.32277066e-01
-7.88224459e-01 -7.14580357e-01 -9.88259852e-01 1.55193895e-01
6.56116784e-01 -7.46757984e-01 1.30979264e+00 -1.01774883e+00
-1.65248966e+00 1.52472258e-01 7.40785897e-02 -4.73982722e-01
5.31199574e-01 -3.37709248e-01 -3.87451440e-01 -3.20427448e-01
-3.08538944e-01 4.90856141e-01 1.21737778e+00 -8.75533879e-01
-5.76828599e-01 -1.15486674e-01 -2.26174697e-01 3.15456688e-01
-2.55545735e-01 -8.96461159e-02 -8.87271464e-02 -3.11669856e-01
-7.50214001e-03 -1.01068640e+00 -6.77988112e-01 -9.96552482e-02
-2.58728415e-01 -1.16346799e-01 4.09543663e-01 -4.44682330e-01
1.74966323e+00 -1.88688898e+00 1.26601413e-01 8.00803661e-01
1.88146774e-02 6.65417165e-02 1.37954861e-01 6.06226623e-01
-9.38565750e-03 2.79365063e-01 -1.48077130e-01 4.51974869e-01
4.46958207e-02 -1.61312282e-01 -4.13844198e-01 2.24377632e-01
5.48326850e-01 1.01822472e+00 -9.77312565e-01 -1.58467397e-01
2.68174618e-01 1.53524905e-01 -8.27248573e-01 3.99621934e-01
-5.07258534e-01 4.10494655e-01 -7.58506238e-01 5.95370352e-01
1.98016036e-02 -3.57161254e-01 1.10172078e-01 1.05631292e-01
-2.47284979e-01 3.17727655e-01 -1.16528499e+00 1.32076418e+00
-4.01568562e-01 3.28203201e-01 -6.71608210e-01 -8.39647353e-01
1.23224688e+00 -3.18763182e-02 3.96494240e-01 -2.86445051e-01
2.06111461e-01 4.17606026e-01 2.14233354e-01 -2.31267869e-01
3.99439722e-01 -2.45100111e-01 -1.50900602e-01 5.80785394e-01
8.51069987e-02 -6.77031577e-01 2.88100451e-01 -2.72034377e-01
9.02690470e-01 4.62734133e-01 8.66744637e-01 -2.95269370e-01
3.97250831e-01 4.63737637e-01 4.92635876e-01 8.46037209e-01
1.27968699e-01 1.26855880e-01 5.00141919e-01 -6.06520593e-01
-9.77419496e-01 -8.55157912e-01 -2.78775804e-02 6.85484469e-01
-1.71890795e-01 -5.10713458e-01 -4.22377139e-01 -4.94360983e-01
4.25057784e-02 1.30296957e+00 -8.02483201e-01 -6.88512743e-01
-4.84155655e-01 -3.51006180e-01 -1.87342092e-01 4.38232064e-01
4.35314178e-02 -1.22763622e+00 -1.19096160e+00 5.13736248e-01
5.00012398e-01 -4.19758648e-01 -4.49841857e-01 6.52450383e-01
-1.29799652e+00 -1.04327643e+00 -2.19235942e-01 -4.24905688e-01
8.54746282e-01 -2.78962225e-01 1.18081427e+00 1.93982497e-01
-5.60822129e-01 2.14545280e-01 -8.37379843e-02 -6.18553519e-01
-4.90456730e-01 -7.24261850e-02 -1.63653150e-01 -3.63808513e-01
2.31788591e-01 -3.66597742e-01 -4.25599575e-01 2.42397025e-01
-4.34952289e-01 1.87904403e-01 9.26877201e-01 1.13638616e+00
7.22221196e-01 4.12601918e-01 4.36263531e-01 -7.02128410e-01
9.35279369e-01 -1.02274433e-01 -9.75837529e-01 6.55713737e-01
-1.22128439e+00 5.86878240e-01 5.03225088e-01 -4.29561287e-01
-7.63445616e-01 5.16689062e-01 2.40401551e-01 -3.84601057e-01
3.45752463e-02 6.11874521e-01 -2.32557386e-01 -2.93689340e-01
6.86548769e-01 3.30129623e-01 2.32239619e-01 -2.53997952e-01
4.34818983e-01 1.59989581e-01 1.81760445e-01 -5.91167331e-01
8.91137958e-01 -2.07163647e-01 1.82872042e-01 -6.22849464e-01
-4.56846595e-01 1.32188037e-01 -8.00319254e-01 -1.16854854e-01
2.85774648e-01 -2.31775999e-01 -4.80374187e-01 -4.01093177e-02
-6.75203919e-01 -2.66494244e-01 -7.36497045e-01 4.09903854e-01
-8.79262924e-01 -2.52621293e-01 2.30498597e-01 -8.71444583e-01
-3.26179445e-01 -1.24648428e+00 5.23100555e-01 3.96863818e-01
-8.12567353e-01 -8.30829024e-01 2.72191856e-02 -1.23018876e-01
4.16473448e-01 2.29284555e-01 1.19389057e+00 -8.50178599e-01
-8.37489247e-01 -5.07188857e-01 4.39238161e-01 1.74668491e-01
2.53907353e-01 3.23557168e-01 -4.04519796e-01 -2.02181071e-01
-7.13918135e-02 -2.97503144e-01 2.91754872e-01 3.99507344e-01
1.01503766e+00 -3.09220642e-01 -5.21882176e-01 3.74578565e-01
1.17329693e+00 8.78551662e-01 3.95623505e-01 6.02199554e-01
2.34783098e-01 5.57750046e-01 1.10165203e+00 5.81523061e-01
-1.00325778e-01 5.70661485e-01 2.32781515e-01 3.01591516e-01
4.57642645e-01 -4.36583132e-01 -1.11416601e-01 -4.27664593e-02
8.91520060e-04 1.60763457e-01 -1.07195294e+00 4.10890847e-01
-1.69223976e+00 -7.38061428e-01 7.44857609e-01 2.30532265e+00
8.22800636e-01 4.35366511e-01 1.50752157e-01 2.74316430e-01
5.30239344e-01 -4.44999427e-01 -8.94845128e-01 -5.94523191e-01
4.64451373e-01 2.00897634e-01 1.57321200e-01 5.59903204e-01
-7.93138266e-01 8.90305102e-01 7.30805731e+00 6.27173603e-01
-9.65264559e-01 -8.16168606e-01 5.82777381e-01 -3.65693197e-02
-5.83152175e-01 3.50364685e-01 -7.11341083e-01 8.65412429e-02
8.96101773e-01 -9.19904113e-01 6.99910760e-01 1.13867795e+00
2.31855899e-01 -1.68258309e-01 -1.35213518e+00 6.33115828e-01
-4.04075354e-01 -1.46178091e+00 1.61593348e-01 -1.41852470e-02
5.29045880e-01 -7.33788371e-01 9.84774455e-02 3.70361298e-01
3.18226308e-01 -1.37067258e+00 8.34527373e-01 5.92735767e-01
5.60920775e-01 -1.16806948e+00 2.28335574e-01 4.21356857e-01
-9.07004952e-01 -4.21808898e-01 -4.13034810e-03 5.94838001e-02
1.15895279e-01 1.59742758e-01 -1.02992511e+00 1.02505237e-01
1.85458228e-01 3.54865611e-01 -5.06565213e-01 1.33383095e+00
-2.95984477e-01 3.26517791e-01 -2.55434692e-01 -5.57853580e-01
1.93571731e-01 -2.28759795e-01 4.77031201e-01 7.48791516e-01
6.00446761e-01 -2.58658919e-02 8.06115195e-02 1.11603987e+00
4.25506234e-01 -3.22312452e-02 -7.95256436e-01 -4.11655515e-01
6.43702626e-01 9.41937506e-01 -5.88840067e-01 4.34641130e-02
-2.94313245e-02 3.39156061e-01 2.29034498e-01 1.13365136e-01
-6.02342784e-01 -4.75317299e-01 3.81511509e-01 2.58463383e-01
3.18415016e-01 -1.77197039e-01 -5.20503938e-01 -5.30954182e-01
-3.22465092e-01 -1.03510594e+00 2.81922430e-01 -4.33548033e-01
-7.32729852e-01 7.10644782e-01 1.39041692e-01 -1.21010256e+00
-8.54184330e-01 -1.91250399e-01 -7.71309614e-01 1.00943279e+00
-1.20138574e+00 -4.74995732e-01 -1.46840110e-01 4.48846817e-01
5.27547300e-01 -2.51046449e-01 8.07893634e-01 -3.75333995e-01
-3.72327685e-01 4.83305335e-01 -1.91857740e-01 -6.24669373e-01
-2.30470411e-02 -1.18879378e+00 4.34177220e-01 6.01817787e-01
-3.65359783e-02 7.84212828e-01 1.07579470e+00 -8.08914363e-01
-1.69486880e+00 -7.99636126e-01 7.62342036e-01 -2.18504503e-01
9.69429910e-01 -9.83441770e-02 -7.12784350e-01 2.84129560e-01
-1.80126682e-01 -3.10934335e-01 5.91553807e-01 8.88098329e-02
2.89339244e-01 -8.48705918e-02 -1.08886492e+00 8.59277844e-01
7.81405091e-01 -2.46527955e-01 -6.32029653e-01 8.49541873e-02
6.41565979e-01 -3.45646918e-01 -7.95611084e-01 5.83829999e-01
4.44081217e-01 -5.49427986e-01 8.57501566e-01 -8.52678716e-01
7.69277140e-02 -2.14462087e-01 8.62108842e-02 -1.55347490e+00
-5.42709827e-01 -1.15146005e+00 -3.98162752e-01 7.47445583e-01
7.89320886e-01 -4.45929915e-01 8.42175663e-01 1.20864022e+00
1.14061579e-01 -1.08501852e+00 -5.30865610e-01 -9.85498726e-01
-1.10808454e-01 -3.05867523e-01 7.04151809e-01 3.55762452e-01
1.68508306e-01 3.56252760e-01 -1.06935360e-01 -1.54338494e-01
6.21422291e-01 4.62929875e-01 6.01784885e-01 -1.35651553e+00
-3.22871625e-01 -5.47091246e-01 -2.42050484e-01 -6.52446806e-01
-1.83818713e-01 -4.59363043e-01 4.37282771e-02 -1.25023448e+00
-2.12166250e-01 -4.32078093e-01 -1.02987066e-01 2.25872740e-01
-1.28648356e-01 -5.76538324e-01 6.49381578e-02 2.13790506e-01
-3.22541267e-01 7.08174825e-01 1.37890887e+00 6.85121864e-02
-7.31090009e-01 3.86008710e-01 -9.64824438e-01 7.24394083e-01
9.41101670e-01 -4.24864829e-01 -6.86668873e-01 2.49003351e-01
3.83092821e-01 4.26978379e-01 3.06398552e-02 -8.33308160e-01
1.30866751e-01 -6.88989341e-01 5.53061962e-01 -5.38012743e-01
2.95053661e-01 -9.52709675e-01 5.23162007e-01 9.07490611e-01
-4.74135429e-01 2.51347840e-01 1.82094067e-01 6.62062585e-01
-2.06518501e-01 -5.00869989e-01 7.09702790e-01 -7.73207992e-02
-8.99887264e-01 1.78578481e-01 -1.69581816e-01 -1.23155802e-01
1.30544531e+00 -3.78944755e-01 3.09451669e-01 -2.41991997e-01
-9.29891467e-01 2.33790100e-01 2.62700856e-01 3.46888304e-01
6.72206104e-01 -1.17377126e+00 -2.13249609e-01 4.89153534e-01
2.88069304e-02 -2.41543412e-01 -3.69461626e-01 3.31813067e-01
-2.68413395e-01 4.96644199e-01 -2.78940707e-01 -3.82400036e-01
-8.88113439e-01 5.35418510e-01 4.84641612e-01 -4.06871855e-01
-4.52935249e-01 6.06050968e-01 -1.95598483e-01 -9.84448120e-02
1.46437109e-01 -1.28345266e-01 -3.72880280e-01 -2.02743441e-01
4.11890835e-01 4.08029735e-01 3.59652960e-03 6.03602938e-02
-2.95230538e-01 3.51936638e-01 -1.41000926e-01 -2.00639144e-01
1.47573864e+00 1.06937990e-01 5.67826331e-02 2.33583927e-01
7.17292309e-01 -6.28815472e-01 -1.47765028e+00 6.47558421e-02
4.89982337e-01 -7.48327792e-01 4.49972078e-02 -9.31344986e-01
-4.04053837e-01 4.09949183e-01 3.11168730e-01 2.56104022e-01
1.13299203e+00 -2.99903363e-01 9.76339206e-02 8.00000429e-01
5.15967429e-01 -1.35506439e+00 2.49106899e-01 4.19502735e-01
1.17784965e+00 -1.01478004e+00 8.33947808e-02 -2.78436303e-01
-6.94568634e-01 1.30550814e+00 7.85270512e-01 -5.90879619e-02
8.52051020e-01 1.96920410e-01 -3.96845847e-01 -2.75735557e-01
-9.02925789e-01 8.60742182e-02 5.64691305e-01 4.73748565e-01
1.57295018e-01 -1.54027089e-01 -3.25797737e-01 5.47423363e-01
-3.64462882e-01 1.75300598e-01 1.89213917e-01 1.24605465e+00
-7.99983323e-01 -1.15995598e+00 -2.83685088e-01 7.16000378e-01
-5.88271283e-02 -6.87448531e-02 -2.68915921e-01 9.31267798e-01
-2.38560006e-01 8.02665889e-01 -2.88386136e-01 -3.70126069e-01
3.41714025e-01 1.63500682e-01 5.24502933e-01 -6.85935080e-01
-4.48851913e-01 2.62270812e-02 2.09145948e-01 -5.41676223e-01
3.09538931e-01 -8.99437547e-01 -1.00852334e+00 -5.62768094e-02
-5.27231336e-01 4.35638219e-01 6.72199547e-01 1.02797198e+00
2.42021680e-01 4.76620793e-01 9.62594807e-01 -6.71375930e-01
-9.50255752e-01 -4.36828405e-01 -4.47809547e-01 1.05723582e-01
-2.48214249e-02 -9.64367330e-01 -2.22168952e-01 -3.46737623e-01] | [6.410566329956055, 3.7960588932037354] |
f0ae6e63-bb30-4d42-8b8d-91b8531e1df7 | libcity-an-open-library-for-traffic | null | null | https://dl.acm.org/doi/10.1145/3474717.3483923 | https://dl.acm.org/doi/10.1145/3474717.3483923 | LibCity: An Open Library for Traffic Prediction | With the increase of traffic prediction models, there has become an urgent need to develop a standardized framework to implement and evaluate these methods. This paper presents LibCity, a unified, comprehensive, and extensible library for traffic prediction, which provides researchers with a credible experimental tool and a convenient development framework. In this library, we reproduce 42 traffic prediction models and collect 29 spatial-temporal datasets, which allows researchers to conduct comprehensive experiments in a convenient way. To accelerate the development of new models, we design unified model interfaces based on unified data formats, which effectively encapsulate the details of the implementation. To verify the effectiveness of our implementations, we also report the reproducibility comparison results of LibCity, and set up a performance leaderboard for the four kinds of traffic prediction tasks. Our library will contribute to the standardization and reproducibility in the field of traffic prediction. The open source link of LibCity is https://github.com/LibCity/Bigscity-LibCity. | ['Wayne Xin Zhao', 'Chao Li', 'Wenjun Jiang', 'Jiawei Jiang', 'Jingyuan Wang'] | 2021-11-04 | null | null | null | international-conference-on-advances-in-2 | ['time-series-prediction', 'spatio-temporal-forecasting'] | ['time-series', 'time-series'] | [-7.39538193e-01 -9.24089730e-01 -4.70704198e-01 -5.28545439e-01
-2.75691628e-01 -1.06585786e-01 2.28526935e-01 -4.83865172e-01
-3.09337657e-02 6.88791931e-01 -3.81692462e-02 -6.96270645e-01
3.05718966e-02 -1.22009492e+00 -2.49422178e-01 -5.58771968e-01
2.73332715e-01 5.26124239e-01 8.05935502e-01 -4.10980135e-01
1.12081520e-01 4.60116208e-01 -1.94752264e+00 2.81850874e-01
9.89836633e-01 9.13744032e-01 4.59887862e-01 1.47394523e-01
-3.96865994e-01 4.20999795e-01 -3.11826229e-01 -4.89510745e-01
8.48934054e-02 -5.57373762e-02 -5.68870425e-01 -4.17347133e-01
-1.25255249e-02 -2.41847515e-01 -5.75164616e-01 8.12200963e-01
4.10822779e-01 1.02027126e-01 2.65870243e-01 -1.68740487e+00
-5.02410531e-01 3.00407708e-01 -2.67590523e-01 3.33652437e-01
9.69235376e-02 3.75334918e-01 7.73515403e-01 -5.09325862e-01
2.89736897e-01 1.10502863e+00 6.27174973e-01 2.70178288e-01
-7.77149856e-01 -1.12352347e+00 1.54741779e-01 6.66231692e-01
-1.45066333e+00 -5.67964852e-01 6.29165292e-01 -4.72592294e-01
6.56038165e-01 5.72688758e-01 4.97843444e-01 1.14427030e+00
7.69336820e-02 7.45458066e-01 6.27168536e-01 -4.64153960e-02
-2.00648978e-01 2.78107256e-01 4.87884521e-01 4.84119356e-01
3.36546153e-01 1.67050272e-01 -5.23731783e-02 4.47372347e-02
6.05800629e-01 3.16434383e-01 1.06075652e-01 1.21775819e-02
-1.09494090e+00 4.67193693e-01 2.14843094e-01 2.13381290e-01
-1.61654711e-01 7.66102374e-02 5.52212119e-01 -9.55601316e-03
5.74341714e-01 -3.01174462e-01 -3.73476356e-01 -6.53091669e-01
-4.54112589e-01 2.57032543e-01 6.37646973e-01 1.21266007e+00
7.50902951e-01 1.97288003e-02 -2.97568709e-01 1.02366769e+00
3.21416885e-01 6.94450676e-01 2.93590963e-01 -8.67479324e-01
6.00419343e-01 6.22030973e-01 3.44766192e-02 -9.82956231e-01
-3.18062305e-01 -2.14507133e-01 -7.21428514e-01 -2.35017627e-01
3.32504570e-01 -5.94899356e-02 -2.48365119e-01 1.53569639e+00
3.65788221e-01 5.94879329e-01 -2.64808893e-01 9.43208039e-01
8.63157094e-01 8.04415882e-01 4.33559269e-01 6.95322230e-02
1.43679738e+00 -1.08244956e+00 -6.18747175e-01 4.50985819e-01
8.88542593e-01 -8.37545931e-01 1.47882569e+00 4.26517092e-02
-6.45650566e-01 -9.35922921e-01 -6.28193855e-01 -2.19441038e-02
-6.98380828e-01 1.87506571e-01 6.97839499e-01 6.74816370e-01
-6.80547655e-01 3.01158696e-01 -8.12391102e-01 -6.13809049e-01
3.52056772e-01 -9.40224901e-03 -8.93649682e-02 4.96296361e-02
-1.54001069e+00 9.95732903e-01 3.34672540e-01 -8.30884799e-02
-4.70952630e-01 -7.38145888e-01 -4.38073933e-01 4.82298583e-02
2.96884954e-01 -5.08885980e-01 1.12352622e+00 -3.67535084e-01
-1.27263165e+00 4.55696076e-01 -3.05623204e-01 -5.95938042e-02
3.74362975e-01 2.06197076e-03 -1.12839937e+00 -5.07440209e-01
2.36014694e-01 1.96675569e-01 5.06268404e-02 -7.50665545e-01
-7.82625616e-01 -1.35619212e-02 -2.23689377e-01 -2.71406859e-01
-4.63352054e-01 4.31347728e-01 -1.11073124e+00 -5.03457904e-01
-5.56150794e-01 -7.25843966e-01 -5.17133735e-02 -1.74283102e-01
-1.79809257e-01 -4.70399261e-01 8.98049116e-01 -4.66351002e-01
1.90182388e+00 -2.19966531e+00 -6.79757416e-01 2.32527256e-01
1.07997358e-01 6.09093010e-01 -5.57770506e-02 5.19102991e-01
2.39780713e-02 1.69302210e-01 1.87523842e-01 -1.68092310e-01
2.07078785e-01 2.93112814e-01 -4.43616539e-01 9.00053456e-02
-1.26735628e-01 7.92383611e-01 -7.28131294e-01 -5.55591762e-01
5.94199777e-01 3.29582602e-01 -3.44119191e-01 2.61617862e-02
-1.66429803e-01 4.84415025e-01 -1.00363946e+00 4.52092767e-01
1.00032687e+00 -3.04468811e-01 -6.45827055e-02 -3.05503532e-02
-5.19219995e-01 4.96729970e-01 -1.02619278e+00 1.27227795e+00
-6.23260617e-01 5.10410786e-01 -4.36885118e-01 -7.60702908e-01
1.15824926e+00 1.89914793e-01 5.21224737e-01 -9.83620107e-01
8.60906020e-02 1.94658786e-01 -3.64881426e-01 -7.73453295e-01
3.15573275e-01 2.81610310e-01 -2.22853702e-02 5.33879519e-01
-5.23271263e-01 4.74278182e-01 6.18789375e-01 -1.13123935e-02
7.67843783e-01 1.09314144e-01 -7.48946443e-02 -1.86089590e-01
9.20868814e-01 -1.01079568e-02 7.24628866e-01 2.90688515e-01
-4.18815345e-01 1.67755947e-01 4.21557635e-01 -9.05521691e-01
-8.18293452e-01 -1.05744827e+00 -4.96499479e-01 1.09201050e+00
3.56897175e-01 -7.54614234e-01 -3.12430501e-01 -4.34320420e-01
6.83091134e-02 7.59480238e-01 -1.25437319e-01 5.14109209e-02
-5.23097157e-01 -7.97456682e-01 6.14902139e-01 4.88966048e-01
9.78784621e-01 -9.08117950e-01 2.77146727e-01 9.46202129e-02
-7.07995474e-01 -1.17303562e+00 -3.23593765e-01 -6.49110973e-01
-6.64639056e-01 -8.99062395e-01 -1.53052270e-01 -6.86546624e-01
2.43726760e-01 7.29214668e-01 1.09378850e+00 4.94530439e-01
1.51929185e-02 -2.63748080e-01 -2.29897201e-01 -2.57453799e-01
-1.41655058e-01 4.27662134e-01 1.48530200e-01 -1.40095517e-01
8.10688436e-01 -7.45071888e-01 -4.78952140e-01 1.08562827e+00
-5.70150316e-01 4.33344513e-01 2.51720935e-01 2.36019999e-01
3.26140553e-01 2.37056598e-01 6.76577747e-01 -6.51439428e-01
6.52695596e-01 -7.32704520e-01 -9.54650581e-01 2.94323832e-01
-7.96085238e-01 -2.46357411e-01 5.94128609e-01 -2.33373418e-02
-1.07132828e+00 -2.88227797e-01 -7.11937845e-01 -5.69618233e-02
-4.87459123e-01 3.88536155e-01 -3.89892042e-01 2.72329330e-01
3.82868081e-01 3.03742170e-01 3.58201563e-02 -8.95304322e-01
2.49992833e-01 1.12396908e+00 2.08109826e-01 -8.49110842e-01
6.46411300e-01 2.67190516e-01 -2.65015155e-01 -7.31677592e-01
-4.64256197e-01 -4.68747348e-01 -4.97499943e-01 -4.98616964e-01
6.49054289e-01 -9.06906903e-01 -9.30608690e-01 3.86472076e-01
-1.02389133e+00 -3.71481895e-01 2.66534150e-01 4.63421702e-01
-2.62381524e-01 5.00531614e-01 -5.17620444e-01 -3.89829695e-01
-6.56820685e-02 -1.35178757e+00 7.21485913e-01 1.22249134e-01
2.19732765e-02 -9.80460942e-01 1.85394436e-01 6.32305264e-01
6.94938004e-01 -3.37280631e-01 7.38563955e-01 -4.56499785e-01
-7.89448202e-01 -2.22126484e-01 -4.38247770e-01 1.98661223e-01
6.07656576e-02 5.90157807e-01 -7.53135204e-01 2.48307928e-01
-7.31328249e-01 2.79004246e-01 6.74802780e-01 6.13070838e-02
1.73879158e+00 3.99551913e-02 -7.00755239e-01 6.52425587e-01
9.72668052e-01 2.21636131e-01 1.07986331e+00 6.19305313e-01
3.50182831e-01 5.03299356e-01 6.50123954e-01 3.27394605e-01
8.62667203e-01 1.10663736e+00 -2.14572344e-02 -4.53822650e-02
-1.06819823e-01 -1.77254111e-01 1.94585234e-01 1.04267657e+00
-4.48864996e-01 -3.24688584e-01 -1.12502897e+00 2.49688804e-01
-1.97883892e+00 -1.35563850e+00 -8.62233400e-01 2.07859707e+00
4.86294121e-01 8.49819109e-02 5.57343245e-01 4.47481535e-02
8.83249164e-01 5.44943660e-03 -7.88469911e-02 -1.95065945e-01
2.92678565e-01 -9.39486101e-02 2.61102140e-01 4.53663856e-01
-1.03856504e+00 1.16377139e+00 6.40907526e+00 1.23167622e+00
-1.25343764e+00 2.02458754e-01 4.36436146e-01 3.10957246e-03
-1.10306300e-01 -9.43844691e-02 -1.05961001e+00 1.22576463e+00
1.36509156e+00 -4.40932065e-01 3.45753670e-01 1.00692964e+00
8.37675154e-01 1.89366654e-01 -5.26175022e-01 8.33003700e-01
-6.09449029e-01 -1.64379418e+00 -2.53686123e-02 1.01141870e-01
2.32758030e-01 3.80368382e-01 6.84722736e-02 6.79479539e-01
2.13841468e-01 -4.65033412e-01 1.45232067e-01 7.29058146e-01
6.41361892e-01 -5.76624632e-01 7.47934520e-01 3.10477942e-01
-1.49306130e+00 1.00453094e-01 -5.41178823e-01 -3.19318116e-01
4.81668204e-01 5.58148801e-01 -2.81865209e-01 7.85625994e-01
7.18632221e-01 9.83388782e-01 -6.74730599e-01 1.32065630e+00
-3.69180855e-03 7.19444931e-01 -1.41326904e-01 1.87203307e-02
-2.33828977e-01 -4.39365089e-01 1.66838896e-02 1.25784767e+00
4.21482295e-01 -1.10201858e-01 2.80433565e-01 6.38209105e-01
1.52159229e-01 3.07510138e-01 -3.75345498e-01 2.48446211e-01
8.82300735e-01 1.29230309e+00 -3.92431378e-01 -3.52860808e-01
-7.94980466e-01 2.31348127e-01 1.27415612e-01 2.28080601e-01
-1.42903733e+00 -4.83455867e-01 1.03858817e+00 4.50326979e-01
-8.29647183e-02 -3.26280832e-01 -4.70431983e-01 -1.28137505e+00
1.06463045e-01 -5.71464896e-01 3.59102786e-01 -9.19101357e-01
-1.29065681e+00 6.23508155e-01 2.86291957e-01 -1.63167489e+00
2.47301176e-01 -5.99691093e-01 -9.82892811e-01 9.45224345e-01
-1.55523396e+00 -1.02786672e+00 -7.06968069e-01 5.26220977e-01
2.69726753e-01 -4.77835715e-01 6.00945950e-01 1.03467309e+00
-1.42470086e+00 4.77507055e-01 1.74378380e-01 1.05522096e-01
7.28305578e-01 -4.59166586e-01 7.48678207e-01 6.34396136e-01
-3.01865608e-01 7.15094864e-01 3.50182682e-01 -5.07437527e-01
-9.26956952e-01 -1.35704017e+00 1.06147730e+00 -6.04640305e-01
1.12426925e+00 -3.30879927e-01 -8.56207848e-01 7.68158972e-01
-1.88267082e-01 -8.47046748e-02 8.55650485e-01 3.57627898e-01
-1.52982488e-01 -8.16078067e-01 -6.35078788e-01 5.86919665e-01
1.07857406e+00 -3.23491663e-01 -2.32925147e-01 4.06174481e-01
6.99810565e-01 9.32675302e-02 -1.00976181e+00 1.91819578e-01
6.58864379e-01 -8.38132977e-01 9.01379228e-01 -5.83405137e-01
5.45959957e-02 -6.21112406e-01 -4.42313962e-02 -7.16328382e-01
-5.49682438e-01 -1.79214075e-01 -1.28872439e-01 1.41279054e+00
4.84344572e-01 -1.12451351e+00 6.88601673e-01 6.93451822e-01
-3.44966829e-01 -6.21966898e-01 -6.31012022e-01 -9.44167256e-01
4.12191749e-02 -1.05763328e+00 1.38342929e+00 5.47868073e-01
5.79818450e-02 1.38685152e-01 -5.11063397e-01 2.79355538e-03
2.11789832e-01 1.53491989e-01 1.20803738e+00 -1.29052460e+00
1.85725182e-01 -6.66749299e-01 -6.42162859e-01 -1.42264366e+00
1.69899449e-01 -9.33788538e-01 -5.79290092e-01 -1.37655497e+00
1.60725236e-01 -9.39041317e-01 -4.37384665e-01 6.60521865e-01
1.10800599e-03 1.46794245e-01 1.16090411e-02 6.92780972e-01
-9.47566211e-01 7.15588152e-01 1.30010092e+00 1.63950488e-01
8.24645907e-02 4.12698478e-01 -8.42649519e-01 3.68758619e-01
1.20627069e+00 -2.63750941e-01 -4.41338420e-01 -4.99167442e-01
-1.20425135e-01 -3.81464958e-01 3.44096839e-01 -1.11213553e+00
1.87311217e-01 -5.96695542e-01 4.93481606e-02 -7.63954103e-01
2.70333976e-01 -8.59877527e-01 3.59137386e-01 3.10630441e-01
1.06249094e-01 1.14170074e-01 2.34414533e-01 1.49976909e-01
-7.13869631e-02 2.59106666e-01 3.78064245e-01 3.25480789e-01
-1.04810739e+00 7.46085465e-01 -4.88106281e-01 -2.31583670e-01
1.27321231e+00 -4.80180308e-02 -8.16778064e-01 -1.74720004e-01
-6.19354546e-01 5.84064364e-01 4.95810151e-01 7.41025686e-01
3.14271122e-01 -1.69942522e+00 -5.94528615e-01 5.33792675e-01
3.90380204e-01 -5.71055830e-01 3.94037127e-01 1.01432455e+00
-6.51007891e-01 7.86820471e-01 -4.82781410e-01 -4.00449365e-01
-1.07301939e+00 6.54381394e-01 3.17556709e-01 7.27244886e-03
-5.75620592e-01 9.64378864e-02 2.05610935e-02 -6.97483242e-01
6.29268661e-02 -1.34032831e-01 -2.19950154e-01 -4.62518513e-01
6.91114962e-01 7.41058290e-01 1.39220044e-01 -8.29645216e-01
-4.46261764e-01 4.18197244e-01 1.73996627e-01 4.69390512e-01
1.13851917e+00 -3.20217043e-01 -9.82539579e-02 3.60438049e-01
9.18126762e-01 -1.25181586e-01 -8.16791534e-01 3.42518128e-02
-1.62182257e-01 -7.79749095e-01 -1.25973493e-01 -5.31102479e-01
-1.25251997e+00 7.91716635e-01 6.49083018e-01 3.32855523e-01
1.02138710e+00 -1.51888743e-01 1.12044048e+00 3.10928822e-01
5.74058652e-01 -8.52955580e-01 -5.64074099e-01 6.91515565e-01
5.86504936e-01 -1.17048013e+00 -1.38532653e-01 -7.07878053e-01
-7.03503251e-01 9.74135041e-01 8.38623285e-01 3.23944658e-01
1.12655985e+00 1.09987363e-01 2.11871549e-01 -2.32239179e-02
-1.01274133e+00 -5.00749350e-01 1.53403744e-01 6.68127835e-01
7.65345037e-01 2.86999822e-01 -6.00837350e-01 7.89072335e-01
-1.70029148e-01 3.63196373e-01 2.08441485e-02 4.71285909e-01
-4.23588932e-01 -1.62177956e+00 -2.30871156e-01 4.21520740e-01
3.55114453e-02 9.61917490e-02 2.96708524e-01 8.66168618e-01
1.99964225e-01 1.18904853e+00 1.48889899e-01 -8.65393221e-01
5.16888976e-01 -6.39558807e-02 -2.46940762e-01 -2.07104340e-01
-6.23253658e-02 -3.48347276e-01 3.89696896e-01 -6.94952905e-01
-4.15934883e-02 -4.40419078e-01 -1.00517452e+00 -1.32489324e+00
-1.08644590e-01 4.56786841e-01 5.65692484e-01 7.00060368e-01
8.37886572e-01 4.73664254e-01 6.81922317e-01 -4.36630756e-01
1.27046192e-02 -8.50928962e-01 -3.89516950e-01 3.21301371e-01
-4.36018646e-01 -1.10321844e+00 -1.33930019e-03 -1.62678510e-01] | [6.385086536407471, 1.9210418462753296] |
3ed580c1-b99f-4269-97dd-f208438e154e | semi-supervised-multi-task-learning-for-lung | 1802.06181 | null | http://arxiv.org/abs/1802.06181v2 | http://arxiv.org/pdf/1802.06181v2.pdf | Semi-supervised multi-task learning for lung cancer diagnosis | Early detection of lung nodules is of great importance in lung cancer
screening. Existing research recognizes the critical role played by CAD systems
in early detection and diagnosis of lung nodules. However, many CAD systems,
which are used as cancer detection tools, produce a lot of false positives (FP)
and require a further FP reduction step. Furthermore, guidelines for early
diagnosis and treatment of lung cancer are consist of different shape and
volume measurements of abnormalities. Segmentation is at the heart of our
understanding of nodules morphology making it a major area of interest within
the field of computer aided diagnosis systems. This study set out to test the
hypothesis that joint learning of false positive (FP) nodule reduction and
nodule segmentation can improve the computer aided diagnosis (CAD) systems'
performance on both tasks. To support this hypothesis we propose a 3D deep
multi-task CNN to tackle these two problems jointly. We tested our system on
LUNA16 dataset and achieved an average dice similarity coefficient (DSC) of 91%
as segmentation accuracy and a score of nearly 92% for FP reduction. As a proof
of our hypothesis, we showed improvements of segmentation and FP reduction
tasks over two baselines. Our results support that joint training of these two
tasks through a multi-task learning approach improves system performance on
both. We also showed that a semi-supervised approach can be used to overcome
the limitation of lack of labeled data for the 3D segmentation task. | ['Naji Khosravan', 'Ulas Bagci'] | 2018-02-17 | null | null | null | null | ['lung-cancer-diagnosis'] | ['medical'] | [ 1.57964323e-02 1.80701673e-01 -3.86760652e-01 -2.19537169e-01
-9.83280241e-01 -5.63038886e-01 4.67437804e-01 1.26797304e-01
-4.56243724e-01 2.79304028e-01 -6.47362247e-02 -6.82731807e-01
1.58186723e-02 -8.15317512e-01 -4.88394469e-01 -6.18310809e-01
1.52146488e-01 8.27781022e-01 8.34689498e-01 1.73107952e-01
-3.56646590e-02 6.42865598e-01 -1.03964639e+00 5.74182510e-01
7.50282109e-01 8.00464094e-01 3.49061042e-01 1.00518286e+00
-8.79180208e-02 5.91165543e-01 -2.45943099e-01 -2.29374290e-01
4.25477087e-01 -5.82301915e-01 -1.05314410e+00 1.10320613e-01
4.78421956e-01 -3.80734026e-01 -9.73713025e-02 6.86262190e-01
5.32024384e-01 -3.86275709e-01 8.64542186e-01 -7.70833075e-01
-1.31479383e-01 3.25588614e-01 -6.43644810e-01 4.18515652e-01
-2.14981765e-01 2.54959643e-01 9.02698457e-01 -7.28791833e-01
1.82029232e-01 7.88675249e-01 9.73315775e-01 4.58489776e-01
-9.11245525e-01 -4.97917622e-01 -5.63349605e-01 -2.14677781e-01
-1.15908706e+00 -7.95361176e-02 4.86980490e-02 -5.86099565e-01
7.28343964e-01 4.47746307e-01 7.08438277e-01 4.25486356e-01
5.05110562e-01 6.20027781e-01 7.60793626e-01 -3.87367100e-01
-9.72474664e-02 3.44878674e-01 -2.32748568e-01 1.08223069e+00
7.49538720e-01 4.85343002e-02 4.01351839e-01 -9.66178440e-03
1.08495581e+00 2.57733345e-01 7.42459372e-02 -5.21685898e-01
-1.42760932e+00 9.33872163e-01 8.57734084e-01 6.61513925e-01
-1.82676435e-01 2.36367732e-01 5.56632340e-01 -6.92051426e-02
3.41324389e-01 5.85947514e-01 -1.22951359e-01 2.37191737e-01
-1.17090130e+00 -6.90203682e-02 9.25343990e-01 3.34210336e-01
1.53388545e-01 -4.14844006e-01 -8.41530204e-01 7.67494440e-01
3.49690259e-01 6.64438128e-01 7.55690694e-01 -7.87559509e-01
2.64826044e-02 7.62025654e-01 -1.27965942e-01 -7.68980622e-01
-5.47906101e-01 -6.77034974e-01 -9.20803547e-01 8.98435041e-02
6.43045723e-01 3.56941372e-02 -1.17403030e+00 1.12742245e+00
2.49567688e-01 6.00865372e-02 -4.21178043e-01 8.21152568e-01
7.98432469e-01 6.20770678e-02 7.21925050e-02 -2.96169929e-02
1.52193189e+00 -1.09629011e+00 -3.31175297e-01 -7.49357194e-02
1.21719193e+00 -8.00078273e-01 8.18669617e-01 -9.63731259e-02
-9.11503315e-01 -5.04447818e-01 -7.73187160e-01 3.48487422e-02
-1.80667825e-02 5.58405817e-01 6.25066757e-01 9.06450033e-01
-9.56278741e-01 5.09184599e-01 -1.29375374e+00 -5.01610935e-01
9.10037220e-01 6.57689273e-01 -1.76259711e-01 -8.09138417e-02
-6.33105457e-01 1.10169780e+00 2.78159410e-01 -1.25803024e-01
-8.15576553e-01 -8.84519756e-01 -2.90290207e-01 4.60877009e-02
5.60469389e-01 -7.71917045e-01 1.60012209e+00 -5.70233047e-01
-9.52545583e-01 1.14031160e+00 3.04887258e-02 -6.87829435e-01
7.71417677e-01 -4.89421263e-02 1.87241003e-01 2.31318370e-01
2.09230155e-01 7.68354535e-01 5.09148240e-01 -6.88536763e-01
-8.75884950e-01 -1.72797382e-01 -3.93944502e-01 1.36504054e-01
-4.36016731e-02 -2.91693866e-01 -5.50893664e-01 -1.05519220e-01
4.85283136e-02 -1.13539684e+00 -5.30411124e-01 3.17520022e-01
-2.40800649e-01 -1.55219823e-01 9.59277570e-01 -7.02404201e-01
9.67854857e-01 -1.74606776e+00 -3.77820164e-01 1.84277147e-01
5.46576858e-01 7.43944883e-01 2.01629788e-01 -4.41660583e-01
6.02984354e-02 5.95215678e-01 -1.88895255e-01 7.93907568e-02
-4.80046183e-01 1.65184319e-01 4.48843539e-01 3.68034214e-01
4.13393259e-01 1.30117679e+00 -7.90590942e-01 -8.95140290e-01
5.65122902e-01 3.86044472e-01 -5.30677974e-01 7.60207549e-02
-1.08697653e-01 3.57030809e-01 -5.15527189e-01 4.33270067e-01
4.23550665e-01 -6.43786132e-01 1.57680824e-01 -2.67844081e-01
6.76062182e-02 1.21184833e-01 -8.03056121e-01 1.11569679e+00
-3.90705973e-01 4.59354579e-01 -2.02080011e-01 -8.62725258e-01
5.34201443e-01 5.79028010e-01 9.13711250e-01 -5.31732798e-01
2.82158941e-01 5.30273914e-01 7.17871666e-01 -4.31579590e-01
-2.03636378e-01 -3.38877261e-01 3.85852993e-01 3.60577047e-01
-2.41051182e-01 -5.52584350e-01 2.81321913e-01 4.85026315e-02
1.48744845e+00 -3.76533955e-01 7.18712509e-01 -3.34821731e-01
5.95405698e-01 4.00570393e-01 2.26299062e-01 7.37818420e-01
-4.83607262e-01 8.58944833e-01 5.04583776e-01 -3.62209260e-01
-1.14493048e+00 -8.32103193e-01 -2.95744568e-01 5.23170948e-01
-2.11695001e-01 1.34703100e-01 -5.34249902e-01 -1.25311673e+00
8.35938230e-02 2.45245397e-01 -7.25148022e-01 -3.97366025e-02
-6.59371316e-01 -9.66397882e-01 5.62184691e-01 6.47878587e-01
5.75935185e-01 -8.38244140e-01 -8.88795912e-01 -2.35105152e-04
-1.59519851e-01 -1.00588250e+00 -3.41226488e-01 3.62240940e-01
-1.16775227e+00 -1.64547217e+00 -1.11503160e+00 -8.99334073e-01
6.73024654e-01 6.01624727e-01 1.20589101e+00 5.71659148e-01
-6.20302975e-01 2.91586250e-01 -1.95991978e-01 -4.26210493e-01
-8.33242595e-01 3.54986340e-01 -5.12731373e-01 -6.36662781e-01
1.14066482e-01 -6.45058556e-03 -7.34799862e-01 5.93678236e-01
-9.09048855e-01 6.44148290e-02 1.29517901e+00 6.69356465e-01
8.69135320e-01 2.66747829e-02 1.85685068e-01 -1.22658122e+00
2.66597509e-01 -2.89262325e-01 -3.83671731e-01 2.75381953e-01
-4.73730624e-01 8.42015594e-02 2.46248171e-01 -2.95862526e-01
-8.15864265e-01 3.63030046e-01 -1.56360656e-01 -2.84111738e-01
-4.06220444e-02 2.74341315e-01 4.02720511e-01 -3.18989128e-01
8.67260933e-01 -2.00609952e-01 3.81359994e-01 -2.01097205e-02
-1.02086462e-01 4.48564023e-01 2.12888718e-01 1.17877141e-01
8.19957972e-01 4.95961368e-01 5.24538279e-01 -7.97067821e-01
-1.11630630e+00 -9.58763957e-01 -8.37644458e-01 1.30479792e-02
1.11075711e+00 -9.02814090e-01 -2.55674720e-01 8.54123905e-02
-7.39070058e-01 -4.44862515e-01 -3.65759790e-01 6.26636207e-01
-2.58160651e-01 4.00539696e-01 -4.18680638e-01 -3.01566958e-01
-5.27653635e-01 -1.20875490e+00 9.44044232e-01 1.00593388e-01
-2.56154358e-01 -1.23847294e+00 1.03914618e-01 4.25592005e-01
6.61130250e-01 1.85707808e-01 7.86153316e-01 -1.08579767e+00
-5.06177187e-01 -5.65705836e-01 -6.02822661e-01 3.44015658e-01
4.43682462e-01 6.59789667e-02 -6.45436525e-01 -1.32753730e-01
1.91052869e-01 -1.83170006e-01 1.12496376e+00 9.48487937e-01
1.22238111e+00 1.86328769e-01 -7.64514565e-01 3.27087164e-01
1.37493813e+00 -4.67580967e-02 4.82608765e-01 3.58565971e-02
8.53795469e-01 1.71548009e-01 5.00157297e-01 -2.56167315e-02
-2.80623175e-02 5.41623890e-01 5.27450264e-01 -4.67698157e-01
-6.72384381e-01 1.47979110e-01 -1.38232902e-01 4.17150199e-01
-6.27957284e-02 -1.77217305e-01 -1.35215652e+00 4.98137683e-01
-1.43189394e+00 -7.15937197e-01 -7.92186677e-01 2.26263022e+00
6.20517015e-01 2.10755631e-01 2.53125191e-01 6.38707653e-02
6.12995028e-01 -3.14212650e-01 -3.80686700e-01 -8.54947791e-03
4.28424805e-01 2.30928451e-01 8.05694640e-01 2.75423557e-01
-1.41884649e+00 5.22481203e-01 6.26365185e+00 7.23611236e-01
-1.21314359e+00 8.26895088e-02 9.54376996e-01 9.32280347e-02
1.36013225e-01 -3.42791557e-01 -7.52587914e-01 1.37744993e-01
7.80789554e-01 -2.93169077e-03 -2.47603998e-01 7.90440023e-01
2.16700479e-01 -4.64664429e-01 -1.26775825e+00 6.04731143e-01
-1.19278587e-01 -1.24038100e+00 -8.24090093e-02 2.80490786e-01
7.81897545e-01 3.51938248e-01 -2.49877665e-02 3.46782804e-01
8.67299736e-02 -1.21994150e+00 -1.24418899e-01 2.45791107e-01
7.82781541e-01 -2.29405195e-01 1.35482132e+00 3.92241120e-01
-1.19334614e+00 2.64365762e-01 -2.17734411e-01 3.89232099e-01
-1.97998926e-01 7.87328839e-01 -1.76633656e+00 2.89648086e-01
4.96578366e-01 4.84952718e-01 -1.12435389e+00 1.51084661e+00
1.05535120e-01 7.19541609e-01 -4.34397459e-01 -7.78056309e-02
1.88313305e-01 1.79461256e-01 2.58116007e-01 1.26427639e+00
4.48405713e-01 -5.05999513e-02 2.47898489e-01 8.08043361e-01
-2.23191269e-02 5.15159778e-02 -4.35738117e-01 -1.69610307e-01
1.09083131e-01 1.39883626e+00 -1.40776277e+00 -3.57746363e-01
-4.63221192e-01 7.43419290e-01 -2.49948025e-01 -4.19689864e-01
-1.01756370e+00 1.80974826e-01 -1.05850354e-01 5.28050780e-01
3.98913503e-01 1.47958711e-01 -6.20606899e-01 -6.89337969e-01
-3.27261925e-01 -6.03952169e-01 4.15554881e-01 -1.66716725e-01
-9.77827609e-01 2.18089268e-01 -3.74746382e-01 -1.22331405e+00
-1.79351002e-01 -7.20658243e-01 -8.32108676e-01 6.06834531e-01
-1.41814554e+00 -1.06009877e+00 -5.93222678e-01 1.26457572e-01
5.63979268e-01 1.42296165e-01 6.64670885e-01 1.84005603e-01
-3.30091774e-01 3.58257443e-01 9.11264420e-02 4.40397233e-01
8.29535604e-01 -1.52579939e+00 1.93498746e-01 6.56984925e-01
4.05150205e-02 -9.71079990e-02 2.07153827e-01 -8.08515310e-01
-1.04822338e+00 -1.48256338e+00 6.51804328e-01 -6.27942622e-01
1.71514928e-01 3.52792680e-01 -8.63366306e-01 4.80942577e-01
-3.00525039e-01 2.81053841e-01 6.81392372e-01 -2.73662955e-01
2.99606532e-01 1.75980508e-01 -1.24124026e+00 2.65937507e-01
4.98037159e-01 4.27257754e-02 -3.06455433e-01 6.34628534e-01
4.93049264e-01 -5.18482804e-01 -7.01602221e-01 7.25198925e-01
4.90674347e-01 -1.21487582e+00 1.08459425e+00 -3.79989929e-02
4.89803344e-01 -2.37860084e-02 2.85099030e-01 -8.16545904e-01
-5.01260996e-01 4.07460839e-01 7.36904815e-02 5.08641243e-01
6.06234372e-01 -1.44451216e-01 1.22435784e+00 4.05255735e-01
-1.64379686e-01 -8.18391621e-01 -6.03574336e-01 -5.85393071e-01
3.46650690e-01 -1.92692295e-01 -2.42142364e-01 5.52569091e-01
-6.53915346e-01 2.02116042e-01 3.60181838e-01 4.64857416e-03
3.49647075e-01 -1.73510492e-01 5.40857375e-01 -1.40364969e+00
-1.50836304e-01 -7.16952026e-01 -2.18777612e-01 -4.34189051e-01
-4.55537289e-01 -1.25351989e+00 6.03197441e-02 -1.99470747e+00
5.75464904e-01 -2.12386146e-01 -8.51550475e-02 4.81971979e-01
-2.67092645e-01 4.68731612e-01 -4.37331796e-02 4.39599782e-01
-5.88205934e-01 -6.45382106e-02 1.60596323e+00 2.36154664e-02
-2.36065060e-01 6.30928159e-01 -5.90102017e-01 8.24142754e-01
8.11448038e-01 -5.95126092e-01 -2.31302083e-01 -2.92680383e-01
-5.98125458e-02 8.04842710e-02 4.87514764e-01 -1.32620335e+00
8.53865147e-02 -3.88371013e-02 6.89430237e-01 -6.96014881e-01
-4.36699949e-02 -8.99562955e-01 -1.73310578e-01 1.42404938e+00
-1.28070250e-01 -5.30706286e-01 2.57486850e-01 4.51315641e-01
-2.31210403e-02 -2.28097111e-01 1.20761716e+00 -4.91710424e-01
-2.97340810e-01 3.05093497e-01 -3.71860057e-01 4.54817973e-02
1.15196443e+00 -3.58648837e-01 1.99789330e-01 -5.06984144e-02
-5.34422636e-01 3.87227207e-01 1.51997030e-01 -2.28259359e-02
3.81848335e-01 -1.15368009e+00 -9.39545631e-01 6.25420660e-02
-1.77750826e-01 4.47782964e-01 -8.58395249e-02 1.39858282e+00
-8.78349304e-01 9.07402992e-01 5.74355125e-02 -1.01750863e+00
-1.47777009e+00 5.40081821e-02 8.89410138e-01 -8.88928831e-01
-3.71253669e-01 9.09396887e-01 1.59648716e-01 -3.47192496e-01
1.51373625e-01 -6.93052530e-01 -1.82386070e-01 -9.79173481e-02
2.15253666e-01 4.82576996e-01 3.69601786e-01 -1.32981151e-01
-3.87824386e-01 3.76367688e-01 -3.03926170e-01 2.50426531e-01
1.11995471e+00 2.34223679e-01 2.36007664e-02 1.66819051e-01
1.05054617e+00 -2.31409460e-01 -7.88844824e-01 -9.66701061e-02
1.48546115e-01 -2.75244951e-01 2.48271093e-01 -1.08230293e+00
-1.01669335e+00 9.26854968e-01 9.31884110e-01 1.96417794e-01
7.74513781e-01 2.23720029e-01 6.37434244e-01 4.96292531e-01
-3.13789159e-01 -5.45368612e-01 3.64158988e-01 3.77873600e-01
6.09041750e-01 -1.88840353e+00 1.66975901e-01 -6.86743200e-01
-6.56944573e-01 1.18220699e+00 7.17360556e-01 -2.12262124e-01
6.10349476e-01 2.04163998e-01 -3.72724310e-02 -2.76653945e-01
-5.35217941e-01 -3.69124144e-01 7.70241320e-01 4.37736541e-01
9.40560520e-01 9.70576257e-02 -3.65379095e-01 3.68622810e-01
1.02859117e-01 1.47084206e-01 2.19651997e-01 8.09158385e-01
-9.14510667e-01 -8.85923743e-01 -4.39557791e-01 1.13626885e+00
-6.96252465e-01 1.92739472e-01 -5.30671775e-01 1.09182036e+00
1.45364895e-01 5.68535149e-01 1.00624040e-01 -5.57337217e-02
8.51551220e-02 8.34049005e-03 3.98845941e-01 -1.11020446e+00
-6.84780240e-01 1.97497413e-01 -2.36021921e-01 -3.50063503e-01
-3.12172830e-01 -5.28507829e-01 -1.37374723e+00 -5.84393963e-02
-6.01218283e-01 7.57338572e-03 4.33188707e-01 1.06305337e+00
-7.89033398e-02 9.30580914e-01 5.05238354e-01 -5.50990224e-01
-7.49938488e-01 -9.19899881e-01 -3.14113438e-01 2.83560306e-01
4.95768487e-02 -2.84042120e-01 -2.49956131e-01 2.40211058e-02] | [15.357747077941895, -2.1584713459014893] |
9c635475-7451-4998-9397-8574bf558b4d | is-independent-learning-all-you-need-in-the | 2011.09533 | null | https://arxiv.org/abs/2011.09533v1 | https://arxiv.org/pdf/2011.09533v1.pdf | Is Independent Learning All You Need in the StarCraft Multi-Agent Challenge? | Most recently developed approaches to cooperative multi-agent reinforcement learning in the \emph{centralized training with decentralized execution} setting involve estimating a centralized, joint value function. In this paper, we demonstrate that, despite its various theoretical shortcomings, Independent PPO (IPPO), a form of independent learning in which each agent simply estimates its local value function, can perform just as well as or better than state-of-the-art joint learning approaches on popular multi-agent benchmark suite SMAC with little hyperparameter tuning. We also compare IPPO to several variants; the results suggest that IPPO's strong performance may be due to its robustness to some forms of environment non-stationarity. | ['Shimon Whiteson', 'Mingfei Sun', 'Philip H. S. Torr', 'Viktor Makoviychuk', 'Denys Makoviichuk', 'Tarun Gupta', 'Christian Schroeder de Witt'] | 2020-11-18 | null | null | null | null | ['smac-1', 'smac'] | ['playing-games', 'playing-games'] | [-6.15534544e-01 -1.07975163e-01 -6.07398391e-01 -7.65453354e-02
-1.06717873e+00 -4.51164275e-01 7.24178016e-01 1.92530230e-01
-7.28104532e-01 1.54281306e+00 1.11259408e-01 -1.30696326e-01
-3.98240358e-01 -2.89746046e-01 -8.17516983e-01 -1.11210108e+00
-8.27861190e-01 9.19024765e-01 1.43361673e-01 -3.52611482e-01
3.58927436e-02 -9.94042233e-02 -1.12374473e+00 -3.34391832e-01
6.86334193e-01 6.43138826e-01 -5.89933805e-02 1.06592512e+00
4.59829777e-01 1.33559310e+00 -1.19324636e+00 9.56234187e-02
2.89034188e-01 -1.60182372e-01 -6.53910220e-01 5.40863425e-02
-2.17162549e-01 -6.04731560e-01 -8.36930349e-02 6.70803487e-01
6.87485099e-01 8.02456215e-02 5.40091932e-01 -1.69637167e+00
-2.96009690e-01 1.04074359e+00 -9.45080817e-01 1.15400329e-01
7.58588091e-02 5.69498301e-01 1.10467827e+00 -2.08952334e-02
3.31465691e-01 1.15418911e+00 8.37070525e-01 2.69104540e-01
-1.52965450e+00 -5.00985980e-01 2.40205541e-01 3.30263823e-02
-1.02243030e+00 -4.20430928e-01 5.22530913e-01 -2.99033783e-02
1.31931031e+00 -1.60505682e-01 5.18361509e-01 1.17315888e+00
4.49645281e-01 9.93554294e-01 1.43372214e+00 -5.34370363e-01
7.15192199e-01 -7.98507482e-02 -3.53447229e-01 7.35608459e-01
4.37853605e-01 4.81841922e-01 -4.82538760e-01 -7.74771512e-01
7.44786620e-01 -6.34876013e-01 1.12118490e-01 -6.76128983e-01
-1.28534496e+00 9.89237487e-01 3.06986943e-02 -8.34527537e-02
-6.28576756e-01 9.46912408e-01 5.11777163e-01 7.58254588e-01
3.85720640e-01 5.26096404e-01 -8.67119849e-01 -4.40080762e-01
-2.73828924e-01 3.78764480e-01 1.35211325e+00 8.36666763e-01
7.03710198e-01 4.84609902e-01 2.19402879e-01 5.87730110e-01
5.04736841e-01 5.79180777e-01 5.76837182e-01 -1.51668227e+00
5.55507064e-01 7.75203258e-02 6.43646240e-01 -4.45056073e-02
-6.46535039e-01 -3.62471521e-01 -3.90762180e-01 7.63149917e-01
5.45712590e-01 -1.11222160e+00 -2.83039629e-01 1.90198183e+00
2.88247615e-01 2.51854956e-01 7.96536684e-01 4.03251916e-01
2.99025685e-01 4.39097613e-01 3.24911535e-01 -4.19042885e-01
8.94403219e-01 -1.20658624e+00 -3.76258701e-01 -3.95168781e-01
5.45557320e-01 -5.40699720e-01 5.30402422e-01 3.17852050e-01
-1.07820153e+00 -2.33704641e-01 -1.11390567e+00 1.03876626e+00
-1.32813724e-02 -2.00021222e-01 9.11647916e-01 5.72886884e-01
-1.13481307e+00 4.27110821e-01 -9.94771779e-01 -6.50566518e-02
2.06329450e-01 5.71106374e-01 -8.70922878e-02 2.57490605e-01
-1.04598987e+00 1.06966496e+00 5.75065672e-01 -5.77469528e-01
-1.37747645e+00 -6.40758514e-01 -6.53405726e-01 3.13242972e-02
8.04733038e-01 -7.08844244e-01 1.84449494e+00 -9.34353769e-01
-1.96598399e+00 -3.54083925e-02 3.68266702e-01 -6.96075797e-01
4.17475343e-01 -2.94215024e-01 -9.57044065e-02 1.38695344e-01
2.20099911e-01 3.78313541e-01 8.86410177e-01 -1.62965119e+00
-7.89953887e-01 -1.48289919e-01 1.41211003e-01 5.92255056e-01
-1.92864984e-01 -6.14721403e-02 1.26715347e-01 -2.70301819e-01
-7.61129737e-01 -9.49903131e-01 -6.95893288e-01 -5.90253115e-01
-3.42360474e-02 -6.53606832e-01 6.23045921e-01 -1.69661522e-01
8.42402577e-01 -1.56983471e+00 2.38565519e-01 1.78802297e-01
1.14834972e-01 6.10826425e-02 -4.56471592e-01 8.86760771e-01
2.36578137e-01 -3.03436331e-02 2.30604425e-01 -5.37529886e-01
4.78449225e-01 4.75197554e-01 1.92319259e-01 6.69219732e-01
-2.78947200e-03 8.02735627e-01 -1.19323325e+00 -6.58604741e-01
1.25458524e-01 -1.29682451e-01 -2.85593241e-01 1.18646584e-01
-6.11044824e-01 3.35205227e-01 -6.28777862e-01 5.72215617e-01
1.83272794e-01 -4.63606924e-01 7.13894367e-01 5.14148355e-01
3.08391382e-03 -4.11148936e-01 -1.23767149e+00 1.53936791e+00
-4.18406695e-01 3.91935050e-01 5.39472222e-01 -1.08121550e+00
3.95407051e-01 7.26124227e-01 1.06775594e+00 -5.42157292e-01
8.65095183e-02 2.36829251e-01 1.87473461e-01 -4.33771551e-01
1.67392269e-01 4.59850617e-02 -2.54567683e-01 9.01088417e-01
1.26738399e-01 -1.71377257e-01 3.74418855e-01 5.87119125e-02
1.46950126e+00 2.70895660e-01 5.76991141e-01 -2.15469539e-01
2.29996979e-01 2.15937167e-01 6.70474172e-01 1.22492492e+00
-6.84356153e-01 -2.50173420e-01 7.14744091e-01 -1.89170599e-01
-9.97627556e-01 -1.05572009e+00 2.80095309e-01 1.49026740e+00
5.57476878e-02 -2.32130066e-01 -3.92246604e-01 -7.28170812e-01
4.34486330e-01 5.08652806e-01 -5.92743158e-01 2.32137471e-01
-6.39816046e-01 -1.26934397e+00 6.24688983e-01 5.63086152e-01
6.39373004e-01 -9.61421013e-01 -7.72274077e-01 7.08246350e-01
2.14316666e-01 -8.78793716e-01 -6.29254729e-02 6.50477290e-01
-4.96759683e-01 -1.12014210e+00 -6.77770615e-01 -6.13035619e-01
2.37535059e-01 1.71230793e-01 1.54298449e+00 -5.89641556e-02
2.68619508e-01 9.72503901e-01 -4.22183245e-01 -6.37315333e-01
-7.88694441e-01 1.62254095e-01 2.19753578e-01 -4.49000090e-01
1.13161817e-01 -6.74239695e-01 -5.28583884e-01 2.96762168e-01
-6.83183372e-01 -5.10949969e-01 6.95302129e-01 1.13748050e+00
1.76385388e-01 3.26863259e-01 1.11455023e+00 -6.48217797e-01
1.12047780e+00 -6.74454749e-01 -9.38712239e-01 4.56694752e-01
-8.07608962e-01 2.01171771e-01 7.50959992e-01 -4.64298278e-01
-1.12320614e+00 -1.60149336e-02 3.69479090e-01 -3.01506042e-01
-1.66186914e-01 6.03638589e-01 4.42124546e-01 -1.97876245e-01
6.59533262e-01 1.62636891e-01 7.26858377e-01 3.14541161e-03
1.96654007e-01 2.64793932e-01 1.21081278e-01 -9.36946273e-01
8.79400790e-01 1.57289892e-01 6.90948516e-02 -5.27263999e-01
-5.86879790e-01 -4.63119298e-01 -1.59956560e-01 -1.34080872e-01
6.55883670e-01 -1.36814189e+00 -1.16229022e+00 4.68681931e-01
-7.74158955e-01 -1.25602782e+00 -2.54013360e-01 7.05567300e-01
-1.08424270e+00 2.45535612e-01 -6.92068517e-01 -7.75521517e-01
-1.66931421e-01 -1.08134234e+00 6.84291661e-01 3.48544776e-01
1.44343972e-01 -1.57886589e+00 7.95026660e-01 2.61218488e-01
5.48773408e-01 1.35551050e-01 7.22694278e-01 -8.28745186e-01
-3.38562071e-01 -9.60155427e-02 3.05348456e-01 1.79448590e-01
1.13821343e-01 3.00541259e-02 -6.06690645e-01 -9.05758917e-01
-1.96395591e-01 -1.04686308e+00 3.90291810e-01 4.41633701e-01
4.33176339e-01 -3.20872486e-01 -1.85617417e-01 4.17382121e-02
1.68336868e+00 1.30400062e-01 9.12327021e-02 8.38078916e-01
1.04476042e-01 7.86992460e-02 7.34172702e-01 1.14054430e+00
6.11029387e-01 5.49012542e-01 6.79940760e-01 -7.73901939e-02
1.45255774e-01 2.55707860e-01 7.70247877e-01 5.51399708e-01
-1.43095747e-01 -3.00470710e-01 -7.41040111e-01 2.64993966e-01
-2.46378231e+00 -1.19880080e+00 3.96917522e-01 1.94069719e+00
9.13340986e-01 9.36188400e-02 6.56127036e-01 -3.89479667e-01
4.30562019e-01 2.65542835e-01 -9.03089464e-01 -1.99264556e-01
-2.22556815e-01 7.42148980e-02 8.19350898e-01 4.67696607e-01
-1.22451401e+00 7.62814760e-01 7.72145796e+00 6.57527864e-01
-5.94458222e-01 3.93614471e-01 4.57671374e-01 -9.39044580e-02
2.14438170e-01 -1.63294151e-01 -6.03145242e-01 2.51059234e-01
1.10341287e+00 -2.22302511e-01 5.65667331e-01 1.00980783e+00
2.74470985e-01 -5.70481360e-01 -9.76642787e-01 5.40263951e-01
-1.53609961e-01 -1.23363483e+00 -7.71630824e-01 1.22512788e-01
1.28465784e+00 7.86317587e-01 -7.99660906e-02 8.19127798e-01
1.61627817e+00 -8.58328104e-01 4.98431116e-01 5.56849800e-02
2.20687687e-02 -8.77693653e-01 8.92294049e-01 5.72534382e-01
-1.19112647e+00 -5.17202497e-01 -2.21829727e-01 -2.63611108e-01
-4.07485440e-02 -1.76191404e-01 -9.56679344e-01 5.94232321e-01
5.86010754e-01 3.19406658e-01 -5.64042866e-01 1.16570938e+00
-2.29503363e-01 8.18096757e-01 -3.65758866e-01 -3.07563305e-01
7.24901140e-01 6.46815002e-02 4.70527261e-01 8.30928504e-01
-2.05533773e-01 -3.37174654e-01 7.20057845e-01 2.61243761e-01
1.95880949e-01 -9.68983844e-02 -5.25219321e-01 2.41802320e-01
7.71131396e-01 1.05297792e+00 -6.13844514e-01 -4.06726956e-01
-5.80963671e-01 4.91663784e-01 7.43408680e-01 5.64171314e-01
-9.37263668e-01 -9.77681875e-02 5.95219612e-01 -7.96559095e-01
6.33417010e-01 -3.44442278e-01 1.22538306e-01 -1.00752807e+00
-1.67479485e-01 -1.14739430e+00 4.54441905e-01 -1.70743302e-01
-1.52268100e+00 1.31065696e-01 1.69916600e-01 -1.14985323e+00
-7.53512025e-01 -3.25657994e-01 -9.50425625e-01 2.32529417e-01
-1.68923616e+00 -1.06546974e+00 1.49980515e-01 6.83875501e-01
5.04534006e-01 -8.73610973e-01 1.02043426e+00 -2.29738280e-01
-8.46046388e-01 5.08094609e-01 9.83311951e-01 -1.21953435e-01
8.15738678e-01 -1.68896818e+00 -1.01180039e-01 3.47857296e-01
-6.48597181e-02 6.29740804e-02 9.16817546e-01 -3.85840893e-01
-1.78356111e+00 -7.39047647e-01 -1.78784192e-01 -3.24541986e-01
1.16640198e+00 2.60833744e-02 -2.17407718e-01 8.70855093e-01
8.91732752e-01 9.50925648e-02 7.03197420e-01 3.11110914e-01
-2.10152552e-01 -3.08818370e-01 -9.99826610e-01 5.48529327e-01
1.49795011e-01 1.19810358e-01 -3.50920498e-01 6.25196636e-01
5.46205759e-01 -1.45434607e-02 -1.37191164e+00 -2.61516264e-03
2.40771279e-01 -9.74766731e-01 8.17566991e-01 -7.35158503e-01
2.35229488e-02 -1.24634787e-01 -2.03404397e-01 -1.92789841e+00
-3.30135077e-01 -1.12154782e+00 -2.76535034e-01 1.03633666e+00
3.74045908e-01 -9.54727471e-01 9.85136449e-01 3.05640787e-01
-1.38863437e-02 -5.65707266e-01 -1.22257817e+00 -1.10517132e+00
4.63504970e-01 -3.25206183e-02 4.37559754e-01 6.78714931e-01
2.91755348e-01 3.92341167e-01 -6.36962235e-01 3.19780260e-01
1.15316010e+00 2.23088078e-02 1.05748498e+00 -9.14316356e-01
-9.69313383e-01 -5.14428854e-01 1.61361232e-01 -5.95704317e-01
7.02257276e-01 -2.30638951e-01 1.65351897e-01 -1.24565530e+00
5.87732047e-02 -8.82431924e-01 -4.58917707e-01 6.47726595e-01
-3.83028425e-02 -1.96091048e-02 2.37554371e-01 3.52687061e-01
-1.42887437e+00 6.79668903e-01 1.06827819e+00 -1.32679328e-01
-3.60682130e-01 2.23180637e-01 -3.49102825e-01 6.52087450e-01
1.18776214e+00 -4.39129114e-01 -5.23762763e-01 -3.90474290e-01
5.01473807e-02 5.75737298e-01 2.52574563e-01 -1.09666240e+00
3.32349002e-01 -6.86433256e-01 2.31354386e-01 -8.31634551e-02
4.26640779e-01 -7.03858137e-01 -1.41126722e-01 8.68991435e-01
-1.95589021e-01 3.56756598e-01 1.72113791e-01 7.97217190e-01
-3.49297076e-02 -5.02596617e-01 5.65916002e-01 -5.38922369e-01
-7.85969079e-01 -1.28181681e-01 -9.19683576e-01 1.68236911e-01
1.39711022e+00 2.53882021e-01 -7.66891897e-01 -6.54978454e-01
-3.64656508e-01 8.73118222e-01 2.73143679e-01 -4.94278222e-03
1.56261668e-01 -1.00812256e+00 -9.07802522e-01 -3.89206141e-01
-5.47980331e-02 -3.11790168e-01 -1.12662129e-01 8.43850672e-01
-2.38230139e-01 3.25831562e-01 -1.91062465e-01 -4.28791791e-01
-1.01184201e+00 3.57769042e-01 4.34403181e-01 -7.09278524e-01
-3.00605029e-01 3.59066933e-01 -4.41244185e-01 -5.28933823e-01
4.17428523e-01 4.39557135e-01 1.07317969e-01 -6.23719022e-02
1.71027780e-01 4.51597184e-01 -4.17174697e-01 -1.58796534e-01
-1.12555176e-01 9.31922123e-02 -1.48163661e-01 -3.38762045e-01
1.61624086e+00 -1.43949986e-01 1.22281864e-01 3.94916505e-01
6.49928093e-01 -3.06495875e-01 -1.77276480e+00 -3.26549619e-01
1.02623925e-01 1.13637507e-01 1.13668256e-01 -9.96958613e-01
-9.40219641e-01 -2.81095430e-02 4.31911319e-01 3.00146401e-01
5.88033617e-01 5.60235903e-02 2.82003850e-01 9.65855896e-01
7.56822169e-01 -1.64367914e+00 5.43392301e-01 5.06731153e-01
4.59506541e-01 -1.81166327e+00 4.16883528e-01 4.42849249e-01
-9.90544319e-01 1.12904978e+00 8.89727712e-01 -3.41621667e-01
6.84330583e-01 5.81402183e-01 1.89732268e-01 -9.79756564e-02
-1.47225964e+00 -2.86700487e-01 -6.68820381e-01 9.33317661e-01
-4.27030167e-03 2.39432633e-01 1.22848593e-01 1.68275908e-02
1.10516250e-01 -1.96352914e-01 5.93759418e-01 1.48179412e+00
-5.89057326e-01 -1.15007901e+00 -3.84662002e-01 3.04206699e-01
-2.93213218e-01 4.29879904e-01 2.37230435e-02 1.26019537e+00
-2.61103839e-01 1.15529275e+00 -8.65750015e-02 1.59892589e-02
-1.90496266e-01 -3.30218494e-01 5.26603162e-01 -2.62150109e-01
-9.60122168e-01 3.91189635e-01 3.56500149e-01 -5.10912657e-01
-7.05272853e-01 -7.29628086e-01 -1.22756577e+00 -3.28051001e-01
-2.15287030e-01 4.21552718e-01 4.48178619e-01 8.53391469e-01
1.06865697e-01 3.28546464e-01 9.76636946e-01 -1.05173981e+00
-1.50791717e+00 -1.00110710e+00 -7.41222858e-01 1.01777583e-01
4.00612086e-01 -7.58234859e-01 -4.30799931e-01 -5.34124374e-01] | [3.794020891189575, 2.0233681201934814] |
73d305df-94ec-4b49-8918-1b621d63189c | nonparametric-quantile-regression-non | 2210.10161 | null | https://arxiv.org/abs/2210.10161v1 | https://arxiv.org/pdf/2210.10161v1.pdf | Nonparametric Quantile Regression: Non-Crossing Constraints and Conformal Prediction | We propose a nonparametric quantile regression method using deep neural networks with a rectified linear unit penalty function to avoid quantile crossing. This penalty function is computationally feasible for enforcing non-crossing constraints in multi-dimensional nonparametric quantile regression. We establish non-asymptotic upper bounds for the excess risk of the proposed nonparametric quantile regression function estimators. Our error bounds achieve optimal minimax rate of convergence for the Holder class, and the prefactors of the error bounds depend polynomially on the dimension of the predictor, instead of exponentially. Based on the proposed non-crossing penalized deep quantile regression, we construct conformal prediction intervals that are fully adaptive to heterogeneity. The proposed prediction interval is shown to have good properties in terms of validity and accuracy under reasonable conditions. We also derive non-asymptotic upper bounds for the difference of the lengths between the proposed non-crossing conformal prediction interval and the theoretically oracle prediction interval. Numerical experiments including simulation studies and a real data example are conducted to demonstrate the effectiveness of the proposed method. | ['Jian Huang', 'Yuanyuan Lin', 'Guohao Shen', 'Wenlu Tang'] | 2022-10-18 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [-1.54184341e-01 2.40516201e-01 -4.27402496e-01 -8.39253783e-01
-9.97421741e-01 -4.42359924e-01 -5.92096383e-03 2.86389053e-01
-3.19426209e-01 1.04718709e+00 2.20178179e-02 -4.73353028e-01
-7.25714922e-01 -8.94224226e-01 -9.11971450e-01 -8.23132992e-01
-1.60002172e-01 3.43717694e-01 -2.52682388e-01 1.28207773e-01
3.06607723e-01 3.96286577e-01 -9.82108295e-01 -2.92127430e-01
1.35060894e+00 1.66154063e+00 -7.50886619e-01 3.14624488e-01
3.15037012e-01 1.28152847e-01 -4.41195130e-01 -3.99371028e-01
3.10598850e-01 -1.17932469e-01 -1.43982664e-01 -4.26753163e-01
4.17123795e-01 -5.90021968e-01 8.02278072e-02 9.57423806e-01
5.84619999e-01 4.15520251e-01 1.23421252e+00 -1.18225324e+00
-8.29927206e-01 5.54761052e-01 -9.51954186e-01 1.09137513e-01
-3.66134495e-01 -7.11940587e-01 9.21588540e-01 -6.25674963e-01
-1.62454903e-01 1.04478371e+00 1.15030658e+00 2.16603875e-01
-1.07275438e+00 -1.04539394e+00 -1.82487965e-01 -1.97136998e-01
-1.50614369e+00 -2.59607702e-01 5.48304737e-01 -5.56917608e-01
4.63704556e-01 -6.92660734e-02 9.67686474e-02 8.17759693e-01
6.51706159e-01 6.57396391e-02 6.33236766e-01 -2.71293789e-01
3.50583524e-01 1.32293895e-01 3.10881913e-01 4.86978024e-01
5.49260855e-01 3.54611754e-01 -5.97994663e-02 -5.46159565e-01
1.21407902e+00 6.87393695e-02 -8.45163912e-02 -3.52350354e-01
-4.01572376e-01 1.17417669e+00 2.86795259e-01 -2.93073565e-01
-2.69432634e-01 2.44124636e-01 4.83970255e-01 -6.17101304e-02
9.80064571e-01 -2.17750624e-01 -4.71955866e-01 -1.07693730e-03
-1.00507748e+00 2.93208033e-01 6.26460910e-01 1.04058719e+00
-3.56314424e-03 1.61386192e-01 -5.81155300e-01 7.67579854e-01
5.67574017e-02 5.94094932e-01 3.37419659e-02 -1.05202639e+00
7.04064488e-01 1.76987797e-01 6.35145426e-01 -9.61669624e-01
-6.14515245e-01 -5.49064696e-01 -1.24012256e+00 -3.59203704e-02
7.85285771e-01 -6.00012958e-01 -4.52559441e-01 1.94950604e+00
3.51726770e-01 7.03397244e-02 -1.54144749e-01 7.96386719e-01
1.64301038e-01 6.24539673e-01 2.04147115e-01 -3.58724028e-01
9.92708802e-01 -2.54900753e-01 -5.51717401e-01 6.02103412e-01
4.33503866e-01 -6.79783523e-02 1.06321490e+00 1.46657541e-01
-1.00280368e+00 -4.34806049e-01 -1.07197416e+00 -1.17537146e-03
1.14812717e-01 2.40818515e-01 2.82834291e-01 8.26917052e-01
-2.66859889e-01 7.34422982e-01 -6.61237359e-01 4.03375715e-01
5.71954370e-01 4.06420112e-01 -1.48575112e-01 4.82357472e-01
-1.25256801e+00 2.72851884e-01 1.61039263e-01 3.91781688e-01
-2.36804277e-01 -1.08583152e+00 -7.88873613e-01 7.67703712e-01
7.97765628e-02 -4.78016049e-01 1.07144237e+00 -9.10739720e-01
-1.37494397e+00 2.65089124e-01 8.75480622e-02 -5.26269317e-01
9.55676019e-01 -4.55074757e-01 -2.18251720e-01 -8.94449055e-02
2.20389560e-01 -1.23040311e-01 5.25802016e-01 -6.16151989e-01
-6.40482485e-01 -4.65439260e-01 -2.98335671e-01 -1.05221994e-01
-4.25181538e-01 -3.62346232e-01 2.64060527e-01 -6.41984403e-01
-2.77053893e-01 -4.57014501e-01 -1.51888847e-01 1.92153584e-02
-3.19802999e-01 -4.55734015e-01 1.36557221e-01 -5.53153634e-01
1.18766809e+00 -2.09382606e+00 -5.65491974e-01 6.14389300e-01
-9.60495919e-02 -2.98429877e-01 1.85666084e-01 1.00619964e-01
-6.94816858e-02 2.15722382e-01 -4.54462886e-01 1.13471327e-02
3.16904873e-01 -2.94785470e-01 -6.48082912e-01 9.73289967e-01
9.54792202e-02 3.59531164e-01 -4.95099157e-01 -2.79474288e-01
-1.25476345e-01 5.59674442e-01 -6.07234180e-01 8.76749679e-02
1.57192260e-01 2.09635437e-01 -5.91484129e-01 2.63040185e-01
1.00921643e+00 -6.88963607e-02 -5.21922484e-02 3.78449857e-02
-1.63237989e-01 -7.38688409e-02 -9.50619340e-01 8.22577059e-01
-4.88423198e-01 3.23279560e-01 -1.46564871e-01 -1.45803165e+00
1.10800529e+00 1.58560514e-01 2.36651227e-01 -5.95618606e-01
2.98352122e-01 2.02003360e-01 -3.45163435e-01 -1.61074296e-01
3.56880277e-01 -7.43855000e-01 -4.22475785e-01 8.72901455e-02
-2.28876725e-01 5.29472470e-01 -5.49183376e-02 -5.21550953e-01
2.80644506e-01 1.41530216e-01 3.73803258e-01 -7.64506817e-01
5.94530880e-01 -5.86014628e-01 7.57535398e-01 9.55123723e-01
-2.83893198e-01 4.86092031e-01 1.27679551e+00 -4.16144401e-01
-1.16844094e+00 -1.24620998e+00 -8.79766107e-01 9.99746382e-01
6.05757982e-02 5.64815462e-01 -6.76748633e-01 -5.92433214e-01
5.21595895e-01 1.05738032e+00 -9.61693645e-01 -1.32910252e-01
-2.77368873e-01 -8.46474051e-01 8.09835076e-01 1.01891112e+00
4.53824043e-01 -4.59982634e-01 -3.33545506e-01 9.11516026e-02
2.37161517e-01 -8.20379615e-01 -6.38602972e-01 1.72848001e-01
-6.54000521e-01 -9.36401606e-01 -1.01320422e+00 -5.49906969e-01
4.79835361e-01 -4.69854891e-01 7.33698249e-01 -4.32827234e-01
3.67825121e-01 1.70523494e-01 2.80000389e-01 -3.68474722e-01
-3.61363441e-02 1.13321856e-01 4.89624254e-02 1.16012534e-02
1.10999562e-01 -6.29144371e-01 -9.28808212e-01 4.64020818e-01
-7.31815338e-01 -5.19833386e-01 2.42564127e-01 9.76499021e-01
8.41033220e-01 -5.01527674e-02 1.59574783e+00 -6.82773829e-01
9.26256359e-01 -8.37192833e-01 -1.36032808e+00 2.73107529e-01
-8.42416286e-01 1.31630838e-01 1.01566076e+00 -4.51631248e-01
-1.20071268e+00 -4.25258100e-01 1.09242588e-01 -1.71640784e-01
5.09155653e-02 5.88796377e-01 6.40733540e-02 3.35285187e-01
3.92821133e-01 -1.61898538e-01 -2.77332187e-01 -3.74506831e-01
2.27073744e-01 5.98186672e-01 7.33438253e-01 -9.91082847e-01
3.39824647e-01 4.00296926e-01 5.14578640e-01 -5.29267490e-01
-1.09031153e+00 9.21912119e-03 -3.78755689e-01 6.19895346e-02
6.07971668e-01 -6.37706578e-01 -1.45523489e+00 -5.02246898e-03
-8.27544689e-01 -1.71857506e-01 -2.07083911e-01 5.62816381e-01
-7.88505554e-01 3.42259288e-01 -6.12630367e-01 -1.26787531e+00
-7.24327087e-01 -5.53578317e-01 7.64665067e-01 4.31839406e-01
1.11597098e-01 -1.26104534e+00 5.71109764e-02 -1.44979224e-01
3.34807962e-01 7.73706198e-01 1.33072317e+00 -7.42190957e-01
3.40721220e-01 -6.91978157e-01 -6.95449293e-01 4.19559181e-01
-1.90669790e-01 3.56820859e-02 -5.50707698e-01 -1.22913249e-01
-1.00276574e-01 -5.68200648e-02 8.34516585e-01 1.14788258e+00
1.61026216e+00 -5.10307729e-01 -1.26574576e-01 8.58926773e-01
1.26972651e+00 1.10904023e-01 3.33020478e-01 1.35465622e-01
2.28529096e-01 6.31069899e-01 5.56118906e-01 1.04459369e+00
2.47099221e-01 1.81905627e-01 2.04806194e-01 2.30793878e-01
8.41639042e-01 -1.35489523e-01 1.63767010e-01 3.94823581e-01
-1.10892974e-01 -4.40461040e-01 -7.63888717e-01 3.93246204e-01
-1.77806747e+00 -8.76624584e-01 -2.07615241e-01 2.72494435e+00
7.41697550e-01 3.23014796e-01 2.11536899e-01 -2.26332515e-01
9.59178627e-01 -4.26737070e-01 -9.11593199e-01 -1.00054383e+00
-1.31055601e-02 3.61602694e-01 8.81286442e-01 4.50814903e-01
-1.23070800e+00 2.55895168e-01 5.78612661e+00 8.59449029e-01
-8.19599926e-01 -1.15639336e-01 1.03862011e+00 -5.53758666e-02
-1.04096979e-01 -2.75545627e-01 -7.02646315e-01 5.01839161e-01
1.15259874e+00 -5.70511401e-01 -6.37205988e-02 1.04627705e+00
2.93089628e-01 9.73716751e-02 -8.58262181e-01 3.98716271e-01
-5.66062391e-01 -1.26365674e+00 -2.94893384e-01 -7.03470632e-02
7.77801514e-01 -3.93915206e-01 4.58676606e-01 4.66447055e-01
-6.92075416e-02 -1.22236598e+00 7.08929896e-01 8.59521329e-01
1.02177715e+00 -1.63075352e+00 1.02685821e+00 2.76108205e-01
-8.68470848e-01 -3.18924874e-01 -6.85663521e-01 -7.95833543e-02
-6.18940480e-02 7.84378231e-01 -3.49641830e-01 3.91050309e-01
5.25897920e-01 2.29902074e-01 -1.19019365e-02 1.06109834e+00
1.71160027e-01 7.03639925e-01 -4.20010418e-01 -5.72126023e-02
1.84240937e-01 -4.93515849e-01 4.97545749e-02 1.09354854e+00
7.97620535e-01 1.14775345e-01 -2.56944597e-01 1.04245412e+00
-2.52152592e-01 4.15435523e-01 -2.50479788e-01 1.96568847e-01
5.93060553e-01 7.95129180e-01 -2.79724509e-01 -1.80219159e-01
-5.26793301e-01 3.53503495e-01 2.90623397e-01 5.96979618e-01
-1.24747586e+00 -1.04779172e+00 4.74068671e-01 1.87677950e-01
5.13593733e-01 6.45779297e-02 -8.28886271e-01 -6.63307786e-01
2.03347087e-01 -8.09132159e-02 5.73828101e-01 -2.06213757e-01
-1.70749998e+00 1.35836929e-01 1.81898162e-01 -9.52373266e-01
-2.74608999e-01 -4.94319737e-01 -6.52627110e-01 1.00005591e+00
-1.22649908e+00 -6.93177879e-01 -2.34022141e-02 3.99854481e-01
-1.79860577e-01 -5.52446544e-02 6.80324852e-01 3.88375103e-01
-7.00213492e-01 1.17168355e+00 9.35961664e-01 1.30351812e-01
6.24914348e-01 -1.11340678e+00 -8.61062407e-02 2.50448823e-01
-8.16141963e-01 5.22198498e-01 6.28312647e-01 -6.64028764e-01
-5.96058071e-01 -1.02360523e+00 4.92894828e-01 1.03489988e-01
7.22257853e-01 -2.16798082e-01 -9.23375368e-01 6.23391211e-01
-1.40251651e-01 1.75404504e-01 9.29014921e-01 2.90533483e-01
-4.35543209e-01 -4.01939839e-01 -1.37946963e+00 2.88190603e-01
3.71096790e-01 -9.51778889e-02 -1.39958009e-01 2.24544466e-01
5.96684873e-01 -2.60814130e-01 -1.17700231e+00 7.26917028e-01
1.27242148e+00 -8.94028187e-01 7.78023660e-01 -8.69190693e-01
6.88504696e-01 2.84201503e-01 -3.14006388e-01 -8.05884540e-01
-1.97172239e-01 -4.53647077e-01 -3.73993181e-02 1.24180377e+00
3.65405887e-01 -9.14778888e-01 6.36703849e-01 8.22979212e-01
6.51257054e-04 -1.16929531e+00 -1.49293888e+00 -1.00793588e+00
9.13971186e-01 -2.62762249e-01 6.88880861e-01 7.01988041e-01
-1.40385151e-01 -1.00977160e-01 -4.32436228e-01 3.52640927e-01
9.46337461e-01 2.62861460e-01 3.80736083e-01 -1.30532789e+00
-1.32554010e-01 -5.42468429e-01 -1.76228032e-01 -7.61891544e-01
4.84152973e-01 -5.52952707e-01 -7.34226629e-02 -9.19090450e-01
2.18144163e-01 -3.39895338e-01 -7.68155992e-01 1.41848683e-01
5.43198474e-02 -2.73217767e-01 -5.60494781e-01 -3.05416048e-01
-1.49102837e-01 1.09841108e+00 9.21848118e-01 2.10477278e-01
-3.20912004e-01 4.13509935e-01 -6.22401357e-01 8.69431138e-01
9.71092522e-01 -6.89070761e-01 -3.84435743e-01 3.68360840e-02
2.69173533e-01 6.44745171e-01 2.39134833e-01 -6.14202201e-01
-2.64875650e-01 -4.19701546e-01 8.13470662e-01 -6.30223930e-01
-1.32341161e-01 -3.92818063e-01 -2.94334233e-01 2.28548478e-02
-7.79883862e-01 -2.72965766e-02 2.29689270e-01 8.52533519e-01
9.41817462e-02 -7.35947788e-02 1.02542782e+00 7.19077587e-01
3.39625478e-01 2.31054008e-01 -1.47605732e-01 3.86958539e-01
7.73820341e-01 8.61279592e-02 -4.03840363e-01 -6.89221144e-01
-5.63493669e-01 3.70816082e-01 -1.61801606e-01 4.10274714e-02
3.96770477e-01 -1.44819641e+00 -7.04763889e-01 1.31603435e-01
-2.28881374e-01 -1.67059153e-01 4.79819179e-01 8.60338569e-01
-3.77949059e-01 5.57962477e-01 -1.27953753e-01 -2.54669905e-01
-4.13169175e-01 3.09285313e-01 5.90707600e-01 -1.34742588e-01
-3.83430630e-01 6.27125800e-01 6.90686703e-01 -3.97348285e-01
3.68737191e-01 -6.19049609e-01 9.56197232e-02 -1.44844294e-01
3.00378650e-01 6.89660311e-01 -1.94325835e-01 -1.67237356e-01
-2.61278242e-01 3.75711262e-01 1.72315165e-01 9.95039381e-03
1.38615513e+00 -9.47263166e-02 6.16566949e-02 4.75130945e-01
1.30906832e+00 2.68436149e-02 -1.55009222e+00 4.81286719e-02
8.58313069e-02 -1.37247980e-01 -1.09722957e-01 -6.32707238e-01
-8.83376777e-01 8.82164001e-01 7.28379428e-01 2.31958658e-01
1.08355999e+00 -3.77945900e-01 7.44647682e-01 2.19647601e-01
-1.89611733e-01 -1.25186908e+00 -4.64913189e-01 3.78208041e-01
8.55790257e-01 -1.00914860e+00 1.91596031e-01 1.21638095e-02
-3.41237217e-01 1.43513691e+00 4.03777272e-01 -6.33459449e-01
8.68168950e-01 -4.67586815e-02 -3.01249981e-01 3.09642673e-01
-5.15866220e-01 5.34141779e-01 6.07264340e-01 3.85221899e-01
6.64344549e-01 2.88401216e-01 -7.02324212e-01 1.57255781e+00
-3.05439532e-01 -4.69859317e-02 4.45913911e-01 1.59854978e-01
-2.86042720e-01 -3.64478916e-01 -1.85340494e-01 4.63578910e-01
-8.81165922e-01 9.67660174e-02 2.44275287e-01 1.14536107e+00
-2.55579680e-01 5.91142714e-01 3.65202308e-01 1.94802850e-01
3.24327469e-01 -3.52133065e-02 3.94760519e-01 5.92095703e-02
-3.74178169e-03 8.86873230e-02 -2.04437852e-01 -2.22128257e-01
2.45304808e-01 -3.17129254e-01 -1.34487462e+00 -5.34266353e-01
-4.29338187e-01 2.27979854e-01 3.34493190e-01 8.93324673e-01
3.28638852e-01 1.35199428e-01 7.26225257e-01 -4.21028435e-01
-1.28866887e+00 -8.64210129e-01 -8.87396693e-01 4.25764844e-02
5.52401602e-01 -8.76886606e-01 -4.78200465e-01 -5.14820337e-01] | [7.291483402252197, 3.9114229679107666] |
05530041-cc60-4645-870d-e2b14a59c0d5 | deep-bidirectional-language-knowledge-graph | 2210.09338 | null | https://arxiv.org/abs/2210.09338v2 | https://arxiv.org/pdf/2210.09338v2.pdf | Deep Bidirectional Language-Knowledge Graph Pretraining | Pretraining a language model (LM) on text has been shown to help various downstream NLP tasks. Recent works show that a knowledge graph (KG) can complement text data, offering structured background knowledge that provides a useful scaffold for reasoning. However, these works are not pretrained to learn a deep fusion of the two modalities at scale, limiting the potential to acquire fully joint representations of text and KG. Here we propose DRAGON (Deep Bidirectional Language-Knowledge Graph Pretraining), a self-supervised approach to pretraining a deeply joint language-knowledge foundation model from text and KG at scale. Specifically, our model takes pairs of text segments and relevant KG subgraphs as input and bidirectionally fuses information from both modalities. We pretrain this model by unifying two self-supervised reasoning tasks, masked language modeling and KG link prediction. DRAGON outperforms existing LM and LM+KG models on diverse downstream tasks including question answering across general and biomedical domains, with +5% absolute gain on average. In particular, DRAGON achieves notable performance on complex reasoning about language and knowledge (+10% on questions involving long contexts or multi-step reasoning) and low-resource QA (+8% on OBQA and RiddleSense), and new state-of-the-art results on various BioNLP tasks. Our code and trained models are available at https://github.com/michiyasunaga/dragon. | ['Jure Leskovec', 'Percy Liang', 'Christopher D Manning', 'Xikun Zhang', 'Hongyu Ren', 'Antoine Bosselut', 'Michihiro Yasunaga'] | 2022-10-17 | null | null | null | null | ['riddle-sense', 'common-sense-reasoning'] | ['natural-language-processing', 'reasoning'] | [-2.93371174e-02 8.66069138e-01 -6.44488037e-01 -5.26588798e-01
-1.18410182e+00 -6.04928374e-01 5.17130196e-01 5.00576019e-01
-4.12402809e-01 8.00439358e-01 6.07467592e-01 -6.58420622e-01
-1.87242150e-01 -9.06038940e-01 -1.13566411e+00 -1.15166031e-01
2.93980271e-01 9.33135569e-01 2.71110564e-01 -2.17341974e-01
-4.05595541e-01 9.05130710e-03 -6.33032203e-01 8.72217417e-01
1.07666814e+00 7.50199318e-01 9.58682969e-02 7.77686596e-01
-5.47971249e-01 1.44868374e+00 -9.89872143e-02 -1.05696297e+00
-1.50600016e-01 -4.05810386e-01 -1.61163473e+00 -6.08892798e-01
5.72105050e-01 -1.98813468e-01 -6.82782710e-01 6.07187331e-01
5.23436308e-01 -5.63307665e-02 4.98778462e-01 -1.02966106e+00
-9.01175380e-01 1.10618842e+00 -2.45187014e-01 2.51188517e-01
4.28807467e-01 7.71870986e-02 1.28894150e+00 -6.18542612e-01
9.32821691e-01 1.33488619e+00 7.70598292e-01 6.36807501e-01
-1.15910208e+00 -4.07485068e-01 7.67280161e-02 4.92157191e-01
-1.03665543e+00 -3.20863634e-01 4.48555529e-01 -1.03208691e-01
1.55736828e+00 -3.53241935e-02 4.12885547e-01 1.17359841e+00
1.22476816e-01 1.27785122e+00 8.16868901e-01 -2.72580743e-01
-3.23583215e-01 -1.17494613e-01 3.33925903e-01 1.35289717e+00
1.45836532e-01 -5.52564621e-01 -8.21618438e-01 -2.99622267e-01
1.95367813e-01 -2.28491232e-01 -3.28550220e-01 -3.38944383e-02
-1.22750008e+00 6.32361948e-01 7.51873374e-01 1.47831470e-01
-2.23635599e-01 4.31525648e-01 3.39396268e-01 4.05730337e-01
4.03432101e-01 2.62331963e-01 -9.60591793e-01 5.83058922e-03
-5.91905057e-01 -7.44043812e-02 1.10413623e+00 8.33572567e-01
6.90419912e-01 -5.30529082e-01 -3.50253046e-01 7.45896637e-01
4.31951582e-01 5.70768297e-01 1.93378702e-01 -7.36270905e-01
9.11017358e-01 8.91377985e-01 -4.47491407e-01 -6.83339953e-01
-6.72435403e-01 -4.62954253e-01 -6.47877753e-01 -8.13211083e-01
6.38059437e-01 -1.14078999e-01 -9.45714176e-01 1.95274591e+00
4.52635318e-01 7.45851025e-02 4.24496144e-01 5.09405315e-01
1.61458397e+00 5.11794388e-01 4.58902031e-01 3.56329203e-01
1.58545589e+00 -1.30156136e+00 -7.00957298e-01 -3.99012566e-01
1.20826960e+00 -4.99342054e-01 1.00114906e+00 2.20226794e-01
-1.24644804e+00 -1.67007148e-01 -5.11197507e-01 -9.84573603e-01
-6.25111282e-01 2.17346817e-01 8.50372434e-01 7.14823082e-02
-1.19144189e+00 3.34717184e-01 -9.16411519e-01 -4.57886428e-01
6.04244053e-01 2.14614376e-01 -4.70090091e-01 -6.44861042e-01
-1.62558174e+00 1.09560466e+00 4.79028225e-01 4.46401574e-02
-8.00305426e-01 -1.19286990e+00 -9.47425783e-01 4.51353639e-02
5.21711469e-01 -1.47991681e+00 1.10002875e+00 -2.46493459e-01
-1.35747194e+00 1.16005838e+00 -1.85204074e-01 -5.50197363e-01
5.41849673e-01 -4.29395139e-01 -3.57434243e-01 4.71415877e-01
9.21672136e-02 8.86783123e-01 1.89312875e-01 -7.39418745e-01
-1.67194813e-01 -3.24035555e-01 2.69754797e-01 2.39991739e-01
-1.59410372e-01 -3.77993315e-01 -9.23375249e-01 -2.62429953e-01
-5.90616651e-02 -5.87707639e-01 9.67847779e-02 -4.62969877e-02
-7.05101311e-01 -5.84463120e-01 3.85510296e-01 -1.11656010e+00
8.29428077e-01 -1.67151821e+00 2.54505128e-01 -1.10425726e-01
3.78703445e-01 1.78599924e-01 -2.34965459e-01 7.30801404e-01
1.19967833e-01 5.97233921e-02 -1.06384709e-01 -3.66998851e-01
8.73378068e-02 5.45139432e-01 -4.04570073e-01 1.93434611e-01
3.37102950e-01 1.78228867e+00 -1.20201504e+00 -6.65902734e-01
-3.04084420e-01 4.31285113e-01 -5.46086848e-01 1.27063990e-01
-8.24685097e-01 3.90883833e-01 -4.85345870e-01 7.70028591e-01
3.74813110e-01 -9.28495646e-01 5.42442441e-01 -4.88343209e-01
7.46608675e-01 8.29336643e-01 -3.51202250e-01 2.02610803e+00
-4.98651892e-01 4.06566143e-01 -6.96454197e-02 -1.08015883e+00
5.37255466e-01 2.46351600e-01 1.85790971e-01 -6.91095829e-01
1.03827007e-03 1.29819572e-01 -1.45964071e-01 -7.41466641e-01
-3.21651697e-02 -1.86127737e-01 1.17815717e-03 4.63677675e-01
5.35349190e-01 -2.68916875e-01 1.83136761e-01 8.38477790e-01
1.45895863e+00 1.99960202e-01 1.31886840e-01 2.11672541e-02
4.38922852e-01 1.71545282e-01 3.54282856e-01 6.49473608e-01
1.54420257e-01 1.55502781e-01 8.78224730e-01 1.54812457e-02
-4.11195189e-01 -1.22297180e+00 3.85527611e-02 1.36948299e+00
-8.61398205e-02 -5.65465331e-01 -4.13128614e-01 -1.12625444e+00
3.63508463e-01 7.75158048e-01 -6.27483130e-01 -3.77744228e-01
-4.59472358e-01 -7.83635020e-01 1.21936047e+00 7.40473390e-01
6.69772685e-01 -8.70846927e-01 2.89605975e-01 -7.09248707e-03
-5.77998936e-01 -1.54539800e+00 -1.68215290e-01 6.98472112e-02
-9.25408423e-01 -1.35265243e+00 -4.03997630e-01 -7.23565698e-01
6.93940341e-01 -1.13952018e-01 1.54553890e+00 1.87846795e-01
-2.76238441e-01 8.33152652e-01 -2.53568798e-01 -2.14786142e-01
-4.52550739e-01 1.87526822e-01 -4.29971308e-01 -3.36374015e-01
2.56703883e-01 -4.08496648e-01 -5.86303651e-01 -8.50915313e-02
-7.40984142e-01 4.45080370e-01 7.43600309e-01 8.48666370e-01
6.62758648e-01 -4.23559934e-01 7.92270184e-01 -1.19695461e+00
5.87582052e-01 -7.97794223e-01 -2.42051005e-01 7.44091451e-01
-4.06634748e-01 3.12579811e-01 5.38187325e-01 3.70855555e-02
-1.16480672e+00 -4.56925690e-01 -4.54521984e-01 -7.50815943e-02
1.24275930e-01 1.01170743e+00 -1.59236625e-01 1.57329142e-01
5.47702789e-01 1.31107897e-01 -7.96262175e-02 -5.41542709e-01
9.00589645e-01 2.60041118e-01 6.09573185e-01 -7.95379639e-01
5.24150133e-01 5.21038175e-01 1.28001971e-02 -3.91692102e-01
-1.35143936e+00 -4.96935934e-01 -5.73469579e-01 3.24204385e-01
9.70230162e-01 -1.15997851e+00 -8.61935377e-01 2.01897234e-01
-1.10506213e+00 -9.02479887e-01 -1.52852908e-01 3.60954285e-01
-4.54560518e-01 4.60897893e-01 -1.09281313e+00 -1.65987521e-01
-6.65592074e-01 -5.99566579e-01 1.13844216e+00 -7.65786916e-02
-8.12461078e-02 -1.47549403e+00 2.48182081e-02 1.27628517e+00
5.43474630e-02 -9.47837308e-02 1.39281833e+00 -8.67729425e-01
-6.04986131e-01 1.18112788e-02 -4.64302599e-01 9.61918831e-02
3.12735960e-02 -5.33516347e-01 -8.59999359e-01 -8.48475993e-02
-6.28795385e-01 -1.02963185e+00 1.39470029e+00 -1.06139183e-01
1.17707586e+00 -2.77032793e-01 -6.31224751e-01 6.04119658e-01
1.12468612e+00 -5.01690507e-01 3.29992682e-01 -3.01105846e-02
9.59178627e-01 4.69960451e-01 2.33543813e-01 -2.38029554e-01
1.13681924e+00 2.76755065e-01 2.47576833e-01 -2.05613241e-01
-5.45845151e-01 -6.12493992e-01 3.83291721e-01 9.57713008e-01
1.81930527e-01 -4.92531985e-01 -1.40417814e+00 7.09164500e-01
-1.86175060e+00 -6.55188859e-01 -1.39965847e-01 1.48135793e+00
1.35979474e+00 -1.03894621e-02 -1.97665170e-01 -4.67238814e-01
1.32871240e-01 -7.20017031e-02 -8.34272683e-01 -5.92412390e-02
-3.25580001e-01 2.38748476e-01 2.70109475e-01 7.65494883e-01
-8.89666617e-01 1.27481377e+00 5.40988779e+00 9.23155487e-01
-7.65794337e-01 4.00694132e-01 6.20077133e-01 -1.23717450e-02
-7.13963091e-01 -6.91950694e-02 -9.13860798e-01 -1.48057742e-02
1.04098558e+00 -1.18092939e-01 1.99919507e-01 4.90263551e-01
-2.57787764e-01 -1.29956052e-01 -1.08812606e+00 6.68473363e-01
1.52151048e-01 -1.80083048e+00 3.52755576e-01 -2.44819000e-01
6.23448610e-01 7.05274224e-01 -9.91621986e-02 6.22759402e-01
9.03711975e-01 -1.07382393e+00 2.12929383e-01 6.22308969e-01
5.38677931e-01 -3.17481071e-01 8.20033908e-01 5.59081197e-01
-9.67178285e-01 1.39023945e-01 -2.23937139e-01 2.46051595e-01
2.40636587e-01 7.70438910e-01 -1.17064250e+00 1.13218415e+00
5.50337553e-01 8.37158024e-01 -5.29017031e-01 4.66807127e-01
-8.10319722e-01 9.02980983e-01 -3.98572177e-01 1.23562865e-01
3.08441877e-01 2.40286455e-01 1.17242523e-01 1.51458180e+00
-1.50150284e-01 2.09868982e-01 1.74136385e-01 9.76368010e-01
-6.78600669e-01 2.49538571e-01 -4.63935375e-01 -5.31251669e-01
1.76426619e-01 1.19693589e+00 -4.56652254e-01 -6.60691202e-01
-6.01253390e-01 9.08629775e-01 9.76574898e-01 4.52224463e-01
-8.09427261e-01 -1.29240185e-01 2.67335534e-01 -1.96834221e-01
1.91433564e-01 -6.51969761e-02 2.21471619e-02 -1.45423269e+00
-1.63077176e-01 -5.97540259e-01 1.12493527e+00 -9.00642216e-01
-1.62936640e+00 3.29424232e-01 -1.36331052e-01 -3.83100152e-01
-4.91588749e-02 -7.97124922e-01 -2.65835017e-01 9.21833813e-01
-1.91774261e+00 -1.68897629e+00 -1.49795488e-01 8.07288468e-01
8.64519924e-02 4.86031882e-02 7.75666893e-01 2.61557847e-01
-4.82450336e-01 6.56045198e-01 -9.07434374e-02 4.80823070e-01
8.36953282e-01 -1.35910094e+00 3.19558084e-01 3.07900369e-01
2.41546482e-01 8.83987248e-01 8.72592181e-02 -1.01396012e+00
-1.85972047e+00 -1.34677339e+00 1.28295612e+00 -9.75150228e-01
9.61221874e-01 -3.90906006e-01 -1.09888625e+00 1.14982736e+00
2.13229299e-01 2.22194344e-01 1.01008379e+00 4.98562723e-01
-6.79413557e-01 8.50734785e-02 -9.27994728e-01 4.89100635e-01
1.29846084e+00 -9.41298664e-01 -9.72772717e-01 9.35001791e-01
1.02435577e+00 -6.32813215e-01 -1.22282004e+00 5.62144458e-01
2.31825337e-01 -4.90724266e-01 1.06540322e+00 -1.10454333e+00
5.99012077e-01 -3.21101286e-02 4.80203703e-02 -9.49359000e-01
-9.37927738e-02 -3.92731428e-01 -5.21927834e-01 1.01438022e+00
8.45396757e-01 -7.49728262e-01 8.47610950e-01 3.44927877e-01
-2.93327987e-01 -1.21745765e+00 -7.22071052e-01 -5.47382534e-01
2.58388698e-01 -5.55191040e-01 1.61590979e-01 1.22766805e+00
4.51273680e-01 8.63478243e-01 4.00021188e-02 2.09282562e-01
3.50503474e-01 3.17362249e-01 6.24475539e-01 -8.00008774e-01
-5.85149825e-01 -2.36868903e-01 2.67645880e-03 -1.28313363e+00
6.27262115e-01 -1.89339697e+00 -3.54949862e-01 -2.33640790e+00
5.02627075e-01 -1.90752760e-01 -1.42293796e-01 1.16732275e+00
-4.61178839e-01 -7.34496536e-03 -8.00672919e-02 -1.79850571e-02
-8.79820645e-01 6.06911719e-01 1.32143080e+00 -3.68569911e-01
1.58762604e-01 -4.61547345e-01 -7.65266359e-01 7.19228029e-01
6.85734630e-01 -1.92849621e-01 -8.24113190e-01 -8.97244751e-01
8.71612012e-01 1.52555168e-01 5.53288162e-01 -4.49385464e-01
5.14955521e-01 1.00874558e-01 3.36698771e-01 -5.80438554e-01
4.04852659e-01 -3.28331739e-01 -1.47543490e-01 5.05613446e-01
-5.37695944e-01 -2.59463817e-01 5.83467484e-01 7.67386734e-01
-1.04794793e-01 6.85076043e-02 3.05561185e-01 -3.15759122e-01
-6.93093419e-01 3.10887992e-01 5.09234667e-02 5.37628233e-01
3.82636487e-01 4.59538966e-01 -9.97515380e-01 -4.39967483e-01
-1.06996405e+00 8.72631609e-01 -1.86664358e-01 1.69264719e-01
5.54138720e-01 -9.47298110e-01 -7.12563396e-01 -1.57984808e-01
8.36458728e-02 1.38356894e-01 4.95205373e-01 1.28974593e+00
-5.56927204e-01 8.32529426e-01 3.01165134e-01 -3.34987164e-01
-1.01299775e+00 5.23331583e-01 3.38731706e-01 -8.18981409e-01
-6.00336730e-01 1.31952488e+00 2.45004997e-01 -9.95926380e-01
9.21620801e-02 -5.72622001e-01 -5.58449961e-02 -2.56618802e-02
2.10111201e-01 9.97384936e-02 1.60756573e-01 -1.05149902e-01
-4.45327908e-01 2.94988960e-01 -2.30206609e-01 1.96501762e-01
1.18803442e+00 3.70096751e-02 -5.05944610e-01 3.13515931e-01
1.14847779e+00 -2.91305669e-02 -6.37845814e-01 -5.85966110e-01
1.43351585e-01 2.59633690e-01 -1.19667053e-01 -1.41803873e+00
-8.16144943e-01 1.06932104e+00 -2.62906820e-01 -3.63470912e-01
7.86197364e-01 7.12647796e-01 1.06303596e+00 1.02201855e+00
1.82769209e-01 -6.87666714e-01 3.01939875e-01 7.63298631e-01
9.63538349e-01 -1.28893089e+00 5.98737272e-03 -7.41165757e-01
-5.63310444e-01 8.66532326e-01 6.20763421e-01 4.44824487e-01
6.35330737e-01 9.19975266e-02 3.67736444e-02 -5.32431304e-01
-1.16077554e+00 -3.28011543e-01 6.10214710e-01 2.63863057e-01
6.45560801e-01 -3.30832861e-02 -9.75564942e-02 8.13537419e-01
-2.36408070e-01 2.25432217e-01 -1.78286880e-01 7.84416676e-01
-1.06832877e-01 -1.07539725e+00 1.32556602e-01 6.40853643e-01
-4.57580477e-01 -5.59035063e-01 -5.83721817e-01 7.46968508e-01
3.74917500e-02 7.75602460e-01 -3.24694306e-01 -1.25351116e-01
1.41418487e-01 4.35132861e-01 6.86169565e-01 -7.05770075e-01
-4.67681170e-01 -2.84827769e-01 6.65256500e-01 -5.90819836e-01
-3.03778380e-01 -3.14702153e-01 -1.65674639e+00 -2.61443019e-01
-4.23207246e-02 1.94314301e-01 1.01971075e-01 1.15652061e+00
6.93260372e-01 6.82963610e-01 -4.96463388e-01 1.12556383e-01
-3.19803566e-01 -9.02962506e-01 -1.72202960e-01 3.58394653e-01
4.20220755e-02 -3.74814451e-01 5.88644017e-03 9.89995599e-02] | [10.421198844909668, 7.986384868621826] |
e93b6b9a-1dc9-4a10-ad88-93b666aaeb70 | robust-face-recognition-with-deeply | 1805.00406 | null | http://arxiv.org/abs/1805.00406v1 | http://arxiv.org/pdf/1805.00406v1.pdf | Robust Face Recognition with Deeply Normalized Depth Images | Depth information has been proven useful for face recognition. However,
existing depth-image-based face recognition methods still suffer from noisy
depth values and varying poses and expressions. In this paper, we propose a
novel method for normalizing facial depth images to frontal pose and neutral
expression and extracting robust features from the normalized depth images. The
method is implemented via two deep convolutional neural networks (DCNN),
normalization network ($Net_{N}$) and feature extraction network ($Net_{F}$).
Given a facial depth image, $Net_{N}$ first converts it to an HHA image, from
which the 3D face is reconstructed via a DCNN. $Net_{N}$ then generates a
pose-and-expression normalized (PEN) depth image from the reconstructed 3D
face. The PEN depth image is finally passed to $Net_{F}$, which extracts a
robust feature representation via another DCNN for face recognition. Our
preliminary evaluation results demonstrate the superiority of the proposed
method in recognizing faces of arbitrary poses and expressions with depth
images. | ['Qijun Zhao', 'Ziqing Feng'] | 2018-05-01 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [ 3.63763809e-01 9.75337997e-02 4.15401213e-04 -1.06769931e+00
-4.53753442e-01 -2.94604361e-01 3.47058475e-01 -7.37735927e-01
-3.06641787e-01 4.83816922e-01 -7.33721107e-02 1.29529431e-01
1.67411845e-02 -1.04737496e+00 -4.61840898e-01 -8.41026485e-01
-4.99807149e-02 4.92455401e-02 -5.81273139e-01 -1.21790379e-01
2.11142451e-01 1.38979900e+00 -2.17043138e+00 1.75233483e-01
-2.37381235e-02 1.68524003e+00 -5.01386225e-01 6.38409376e-01
-2.65658855e-01 5.00094771e-01 -7.65643001e-01 -5.23502290e-01
5.67102015e-01 -3.15344363e-01 -6.08151078e-01 1.83736444e-01
6.85760200e-01 -8.95060420e-01 -5.71425855e-01 1.27173209e+00
8.71741176e-01 2.39581555e-01 6.23383999e-01 -1.36345828e+00
-5.33825994e-01 -2.20961258e-01 -5.88710189e-01 -1.49530396e-01
6.15325809e-01 -8.35420489e-02 2.21586615e-01 -1.17428827e+00
7.61880338e-01 1.74390328e+00 3.86171639e-01 1.16274464e+00
-7.55393207e-01 -1.08686686e+00 -1.77243605e-01 -3.91933471e-02
-1.57327938e+00 -7.34912097e-01 9.53941882e-01 -1.56275913e-01
1.04187763e+00 7.90908411e-02 5.41108370e-01 9.19959545e-01
-1.14071622e-01 5.50230920e-01 9.69720781e-01 -3.97924364e-01
1.33682325e-01 -3.19545656e-01 -3.45060140e-01 9.56591785e-01
-8.00446123e-02 2.41784588e-01 -6.47942424e-01 2.14456469e-01
1.11177135e+00 3.82852182e-02 -1.67251036e-01 6.84759319e-02
-5.10015190e-01 6.83635473e-01 4.23941344e-01 1.88144539e-02
-4.03780609e-01 5.97977713e-02 1.55421332e-01 3.49193245e-01
3.17349315e-01 -1.16984166e-01 -1.71555415e-01 -1.73121199e-01
-9.29424584e-01 1.59214124e-01 6.69972539e-01 8.14807892e-01
1.22562194e+00 1.89928308e-01 1.89989060e-02 1.06643713e+00
3.54498684e-01 7.62571752e-01 2.04528496e-01 -1.22158217e+00
1.16979800e-01 7.44539261e-01 -3.31996322e-01 -1.12349594e+00
-4.68991160e-01 1.41226083e-01 -9.32153404e-01 7.32504725e-01
2.76286483e-01 -3.51053089e-01 -1.05909884e+00 1.86585414e+00
3.55731249e-01 -3.39187562e-01 4.36596349e-02 8.86729956e-01
1.31436658e+00 6.16448045e-01 -1.61177158e-01 -3.39628816e-01
1.04184508e+00 -3.98060173e-01 -6.77970827e-01 -1.32288173e-01
1.89402729e-01 -7.71773756e-01 6.83703899e-01 4.73924100e-01
-1.22167361e+00 -4.98160154e-01 -1.17500377e+00 -3.14394861e-01
-4.64676291e-01 1.79381773e-01 4.86632288e-01 1.02563751e+00
-1.38123918e+00 4.32234675e-01 -6.08365297e-01 -3.25183064e-01
6.56295002e-01 1.03598928e+00 -7.71679163e-01 -3.15224290e-01
-8.31626654e-01 5.90158522e-01 -9.88488868e-02 4.95743215e-01
-9.65469778e-01 -2.08915249e-01 -1.16715038e+00 -2.75316268e-01
-1.88017339e-01 -1.40056148e-01 1.08395076e+00 -1.23516035e+00
-1.93726444e+00 1.26734185e+00 -5.23184597e-01 4.70274061e-01
1.19245678e-01 2.94353336e-01 -5.10654032e-01 3.56398970e-01
-2.35462151e-02 1.10183287e+00 1.03278220e+00 -1.05271327e+00
-3.44438851e-01 -1.13050389e+00 1.90607935e-01 1.11127980e-01
-3.68732631e-01 2.81311333e-01 -7.16773152e-01 -3.11042815e-01
6.81046009e-01 -5.91948628e-01 4.46112037e-01 5.55860639e-01
-1.71240523e-01 -1.01299658e-01 9.74869668e-01 -5.29577315e-01
7.58850634e-01 -2.10898328e+00 9.75644588e-02 5.86318970e-01
1.25850245e-01 1.34443641e-01 -3.52953315e-01 -1.03243403e-01
-3.61380041e-01 -1.66869462e-01 2.79278643e-02 -3.90500039e-01
-1.40044823e-01 3.28814477e-01 3.50494951e-01 6.27730429e-01
2.25069940e-01 7.17647195e-01 -4.15529698e-01 -3.40294182e-01
2.90702790e-01 1.05252504e+00 -6.41630471e-01 4.42097247e-01
9.18269064e-03 1.93053469e-01 -2.11193159e-01 1.29637682e+00
1.18339896e+00 4.62123185e-01 6.51582927e-02 -4.12234008e-01
-1.12052910e-01 -2.88455546e-01 -1.02930033e+00 1.28668559e+00
-2.42613450e-01 6.46586657e-01 4.28622037e-01 -8.15851390e-01
1.54886985e+00 2.02061236e-01 4.88893688e-01 -1.05187798e+00
7.65265405e-01 1.57571927e-01 -3.34831744e-01 -5.25131464e-01
2.05204755e-01 -4.56265181e-01 1.16941921e-01 3.84188354e-01
2.93019265e-01 -3.75283629e-01 -4.79078256e-02 -4.42416579e-01
6.69725776e-01 4.46745418e-02 -1.25408620e-01 4.74950001e-02
8.96335542e-01 -7.12462246e-01 4.02936697e-01 1.44055486e-01
-4.22400504e-01 5.31087339e-01 8.38739932e-01 -6.22706890e-01
-8.28609109e-01 -9.29097712e-01 -1.26215488e-01 9.07374203e-01
-1.40039667e-01 -3.09626032e-02 -1.20873249e+00 -4.26826060e-01
-1.00600556e-01 7.04885349e-02 -9.65245843e-01 -2.82872856e-01
-4.98025119e-01 -4.86930102e-01 9.39803898e-01 4.74697798e-01
9.13225770e-01 -1.11728191e+00 -3.82769972e-01 -1.45148754e-01
3.96335311e-02 -8.38539600e-01 -1.88424274e-01 9.80024487e-02
-7.65294909e-01 -8.99628222e-01 -6.94835842e-01 -9.45879519e-01
1.02078533e+00 -4.91369702e-02 5.44545233e-01 -1.09959040e-02
-4.07476932e-01 2.65395314e-01 -6.54972298e-03 -4.54372674e-01
-1.03797773e-02 -2.70410210e-01 2.62917906e-01 1.86897129e-01
8.58284056e-01 -7.07662702e-01 -6.96706653e-01 1.89328358e-01
-1.02912772e+00 -4.88269389e-01 4.16623741e-01 7.29775906e-01
6.04890406e-01 -1.01861738e-01 4.07107890e-01 -5.26202321e-01
4.22346920e-01 2.55929053e-01 -6.16326094e-01 -3.34286153e-01
-1.28606647e-01 -1.21446632e-01 3.79575968e-01 -1.18468143e-01
-1.08807385e+00 2.65062839e-01 -8.12358320e-01 -5.77480793e-01
-3.52637142e-01 2.32649580e-01 -8.71102035e-01 -4.44648981e-01
7.68220663e-01 8.60495046e-02 5.28402865e-01 -1.75656676e-01
1.45816386e-01 9.31586504e-01 8.94380093e-01 -4.50924158e-01
7.34385073e-01 6.44558966e-01 1.83927402e-01 -1.09996617e+00
-5.80864191e-01 5.66512421e-02 -8.83301258e-01 -4.37506765e-01
8.10395777e-01 -7.02367842e-01 -1.22421384e+00 8.33695829e-01
-1.07095575e+00 2.43952796e-02 6.21048808e-02 3.33700180e-01
-4.29152191e-01 1.27058908e-01 -6.57596827e-01 -8.64804685e-01
-1.93937793e-01 -1.13824093e+00 1.08266091e+00 6.03056848e-01
3.07032727e-02 -5.50774872e-01 -2.42422357e-01 2.65413821e-01
3.83211493e-01 3.87010157e-01 9.31807876e-01 5.86611107e-02
-3.45154673e-01 -6.81634426e-01 -5.58822989e-01 6.91496372e-01
3.73622090e-01 3.04469585e-01 -1.44890606e+00 -1.99115336e-01
5.74249774e-02 -5.43109715e-01 4.66697544e-01 3.90905261e-01
1.39615536e+00 -2.97202110e-01 9.43085998e-02 1.13081837e+00
1.24266458e+00 6.27149463e-01 1.12136149e+00 2.34444112e-01
4.66603905e-01 7.18038559e-01 2.31764778e-01 5.33188522e-01
-2.81206034e-02 3.41999859e-01 6.44424319e-01 -2.13292375e-01
-2.55294163e-02 2.59003006e-02 4.25213337e-01 2.21789807e-01
-3.35471183e-01 1.74515337e-01 -5.34761608e-01 3.46823893e-02
-9.40316081e-01 -1.01930737e+00 7.90825427e-01 2.01087093e+00
7.05887198e-01 -2.45501608e-01 -7.27521479e-02 3.59794676e-01
6.21358752e-01 1.24701723e-01 -6.43596888e-01 -6.96609855e-01
-3.14346820e-01 9.03590798e-01 -1.85264405e-02 4.43844259e-01
-9.57496107e-01 8.48114967e-01 5.65126276e+00 4.65622902e-01
-1.70098710e+00 -3.97194296e-01 8.45058918e-01 -4.26432341e-01
5.70113249e-02 -6.95572913e-01 -8.47643435e-01 -1.10152975e-01
6.52862370e-01 2.91947499e-02 4.24593270e-01 1.05006266e+00
9.56782326e-02 2.70345714e-02 -1.22834539e+00 1.58703279e+00
4.93108332e-01 -9.62035298e-01 2.58901387e-01 1.23723432e-01
6.41457140e-01 -4.59264517e-01 3.83599252e-01 2.49030307e-01
-3.55962425e-01 -1.54786325e+00 4.25922662e-01 3.98192495e-01
1.46954191e+00 -1.34002423e+00 7.66523123e-01 -1.21875063e-01
-1.06480920e+00 -7.21569080e-03 -5.54001212e-01 -1.48549691e-01
-2.89043963e-01 4.34725314e-01 -4.94054168e-01 1.74915209e-01
8.10850918e-01 4.03341115e-01 -1.48910448e-01 1.82246551e-01
-2.17272341e-01 -1.78731069e-01 -2.64773428e-01 -2.86465865e-02
5.56753501e-02 -1.53602391e-01 4.27042209e-02 1.13030338e+00
5.89047134e-01 5.34784079e-01 -3.94242644e-01 7.55945265e-01
-6.58169866e-01 6.29102364e-02 -7.17413008e-01 -1.39964938e-01
2.91787595e-01 1.42923570e+00 -6.26051664e-01 -1.94581151e-01
-2.61072606e-01 1.13238513e+00 2.63382066e-02 3.53992522e-01
-4.75444078e-01 -7.79232919e-01 1.12010527e+00 -3.25189680e-02
1.12894379e-01 -1.08250313e-01 -4.81202528e-02 -9.07901406e-01
5.34962006e-02 -8.92744660e-01 1.84018582e-01 -8.10107768e-01
-9.18890357e-01 9.01613414e-01 -1.61660627e-01 -9.27630246e-01
-3.08027625e-01 -1.20669723e+00 -2.04803661e-01 1.03321052e+00
-1.49044394e+00 -9.87917304e-01 -8.19157004e-01 1.00865281e+00
7.96803460e-02 -2.82561898e-01 1.00497890e+00 4.77891117e-01
-6.52841985e-01 1.02808177e+00 -2.20435575e-01 4.64332283e-01
5.36383212e-01 -6.60500526e-01 -2.96783261e-03 4.30869132e-01
-5.61698735e-01 5.64761639e-01 9.97338444e-02 -3.33933204e-01
-1.72249138e+00 -1.16521168e+00 9.27110910e-01 -1.09095918e-03
-2.33834803e-01 -4.15021837e-01 -4.42298979e-01 6.06825054e-01
-2.44855940e-01 2.13544697e-01 6.70664489e-01 -4.18974042e-01
-4.83074039e-01 -5.40021658e-01 -1.81870604e+00 5.03402591e-01
1.04170358e+00 -9.76322770e-01 -1.40199780e-01 1.50484620e-02
5.12253195e-02 -5.14522076e-01 -9.71649647e-01 5.34283936e-01
9.86234069e-01 -1.24105322e+00 9.63371575e-01 -1.39420748e-01
5.68866909e-01 -1.28423333e-01 -5.52444398e-01 -8.79695415e-01
-2.18932666e-02 -2.98793107e-01 4.34561759e-01 1.10070908e+00
2.35924736e-01 -4.42017943e-01 1.32596421e+00 6.54978156e-01
1.69426277e-01 -6.85488462e-01 -1.14777803e+00 -3.59544307e-01
5.66154048e-02 -5.88755846e-01 7.26209402e-01 7.34719455e-01
-1.67388722e-01 -1.21180661e-01 5.47835827e-02 4.33559902e-02
6.53848708e-01 -2.18061373e-01 6.49577975e-01 -1.10034740e+00
4.49960023e-01 -3.02080035e-01 -8.26132238e-01 -9.92550552e-01
4.88286555e-01 -7.80349433e-01 2.26617847e-02 -1.20871508e+00
2.47353707e-02 5.42023592e-02 1.30060792e-01 6.59902871e-01
5.47271430e-01 8.19149911e-01 -2.18017772e-01 -3.23103487e-01
1.62989587e-01 5.19560158e-01 1.26887357e+00 -1.52208209e-01
-8.47540647e-02 -1.97826371e-01 -4.97786373e-01 7.73746729e-01
6.29404962e-01 7.77294636e-02 -2.67805248e-01 -3.68262589e-01
-1.63320959e-01 1.52068987e-01 2.25944072e-01 -8.21682036e-01
1.34650439e-01 -4.72208485e-02 1.24830794e+00 -6.63535297e-01
8.68365347e-01 -5.89051485e-01 -4.24519144e-02 2.75108099e-01
5.56746349e-02 -3.36946035e-03 3.70348483e-01 -3.39999825e-01
-3.95545334e-01 3.12146135e-02 1.23087442e+00 -1.82282910e-01
-4.73446190e-01 6.54451370e-01 -2.66023397e-01 -5.02647281e-01
8.07814837e-01 -8.50269139e-01 -1.46963745e-01 -6.47789061e-01
-8.60588849e-01 -2.69566089e-01 4.84207541e-01 2.37135738e-01
1.20392895e+00 -1.62669933e+00 -4.24692214e-01 1.07850683e+00
-9.19488072e-02 1.76585585e-01 3.54976594e-01 3.10870647e-01
-8.38516176e-01 2.81171113e-01 -7.70646453e-01 -4.70478207e-01
-1.36769760e+00 1.46697015e-01 7.27091372e-01 5.55215955e-01
3.48141901e-02 1.35476649e+00 -2.95179803e-02 -6.80998087e-01
4.24678832e-01 -9.81155783e-02 -1.88450187e-01 2.82574087e-01
7.93594003e-01 3.76735002e-01 3.07125181e-01 -1.07472277e+00
-5.03499389e-01 1.01219344e+00 1.22326419e-01 -3.96763831e-01
1.42693901e+00 4.89210652e-04 -6.35246992e-01 -2.01164916e-01
1.92901254e+00 -3.78154457e-01 -1.13121414e+00 8.89663324e-02
-5.02526402e-01 -7.43458033e-01 -5.49221300e-02 -4.98457193e-01
-1.67340589e+00 1.11347973e+00 9.38294291e-01 -6.07694447e-01
1.55940962e+00 -1.48325443e-01 4.69434232e-01 5.67455053e-01
2.29227930e-01 -1.11859477e+00 2.17442304e-01 6.49507046e-01
9.51165438e-01 -8.98424506e-01 -2.47943178e-01 -3.38673472e-01
4.26565930e-02 1.63177216e+00 8.32642496e-01 1.98066846e-01
7.95689762e-01 4.56817627e-01 5.34306943e-01 -3.56462955e-01
-1.95030734e-01 5.47010936e-02 -7.38369152e-02 9.26326811e-01
4.71661955e-01 -2.35840946e-01 3.05906832e-01 5.05825460e-01
-5.75628161e-01 2.71750420e-01 2.91691571e-01 1.06628346e+00
-4.41728354e-01 -1.09297693e+00 -4.86756086e-01 2.00132206e-01
-3.88969183e-01 3.44957441e-01 -6.84774399e-01 7.13430405e-01
4.35821891e-01 7.32474148e-01 2.78780103e-01 -5.94864428e-01
5.78695536e-01 2.86099762e-01 9.48949873e-01 -3.47463697e-01
-3.34773272e-01 1.31180823e-01 -2.15457499e-01 -8.20586383e-01
-4.78154182e-01 -4.58393365e-01 -1.40289581e+00 -6.34164929e-01
2.18868151e-01 -2.49171957e-01 8.71406078e-01 6.02019727e-01
-3.66923362e-02 4.73366976e-02 1.03609216e+00 -1.32038951e+00
-7.84347355e-02 -7.89429843e-01 -8.21006417e-01 3.97702038e-01
2.80353427e-01 -6.55133963e-01 -4.72103655e-01 -1.82355829e-02] | [13.438067436218262, 0.9264023900032043] |
ce4dce16-0db2-4701-9aaf-0bd126ec4c5e | dg-labeler-and-dgl-mots-dataset-boost-the | 2110.07790 | null | https://arxiv.org/abs/2110.07790v1 | https://arxiv.org/pdf/2110.07790v1.pdf | DG-Labeler and DGL-MOTS Dataset: Boost the Autonomous Driving Perception | Multi-object tracking and segmentation (MOTS) is a critical task for autonomous driving applications. The existing MOTS studies face two critical challenges: 1) the published datasets inadequately capture the real-world complexity for network training to address various driving settings; 2) the working pipeline annotation tool is under-studied in the literature to improve the quality of MOTS learning examples. In this work, we introduce the DG-Labeler and DGL-MOTS dataset to facilitate the training data annotation for the MOTS task and accordingly improve network training accuracy and efficiency. DG-Labeler uses the novel Depth-Granularity Module to depict the instance spatial relations and produce fine-grained instance masks. Annotated by DG-Labeler, our DGL-MOTS dataset exceeds the prior effort (i.e., KITTI MOTS and BDD100K) in data diversity, annotation quality, and temporal representations. Results on extensive cross-dataset evaluations indicate significant performance improvements for several state-of-the-art methods trained on our DGL-MOTS dataset. We believe our DGL-MOTS Dataset and DG-Labeler hold the valuable potential to boost the visual perception of future transportation. | ['Dongfang Liu', 'Lin Li', 'Yingjie Chen', 'Feng Tao', 'Xingyu Jiang', 'Yixin Xie', 'Zhiwen Cao', 'Yiming Cui'] | 2021-10-15 | null | null | null | null | ['multi-object-tracking-and-segmentation'] | ['computer-vision'] | [ 2.09410533e-01 -6.88619763e-02 -3.57875079e-01 -5.46724200e-01
-6.74448967e-01 -5.59713244e-01 5.72063625e-01 -6.48747385e-02
-4.13445264e-01 7.04129159e-01 -2.76198655e-01 -3.64223242e-01
-2.29274660e-01 -5.97684026e-01 -7.58755565e-01 -5.11616349e-01
-4.85222088e-03 7.14439392e-01 1.15750790e+00 -4.41804260e-01
8.24409574e-02 3.89742136e-01 -2.12220931e+00 2.75946170e-01
8.16800535e-01 1.14714074e+00 3.64716709e-01 6.62947476e-01
-9.19052400e-03 6.57011867e-01 -6.01558685e-01 -1.30480900e-01
4.60620821e-01 2.52580456e-02 -7.35221386e-01 -4.13595676e-01
1.11141419e+00 -2.33841211e-01 -3.14315498e-01 6.92954123e-01
4.57089663e-01 2.73195714e-01 6.03066325e-01 -1.89307821e+00
-1.05771415e-01 2.38827974e-01 -6.00970507e-01 6.37160778e-01
-3.56642485e-01 7.06397116e-01 8.31977725e-01 -1.77949443e-01
6.88168287e-01 1.24516809e+00 8.89440298e-01 6.02342546e-01
-9.87392187e-01 -1.02338326e+00 2.41668358e-01 4.39009398e-01
-1.26913846e+00 -4.68137443e-01 4.60325062e-01 -6.04909122e-01
1.13030815e+00 3.00353765e-01 8.36033940e-01 1.16518068e+00
2.78296858e-01 8.90235424e-01 1.04296708e+00 6.17923699e-02
2.15222538e-01 1.45934271e-02 3.07810664e-01 6.70638084e-01
3.08403730e-01 4.57949787e-01 -6.94812238e-01 5.63520074e-01
3.78636509e-01 -3.21431667e-01 4.77342784e-01 -2.65042156e-01
-1.14443564e+00 6.10212386e-01 4.74992514e-01 1.52102008e-01
7.97588564e-03 6.68976068e-01 7.48446405e-01 -1.11088686e-01
5.39072156e-01 3.44747990e-01 -5.93978047e-01 -4.91758347e-01
-1.00705528e+00 2.98554063e-01 3.50423604e-01 1.33865166e+00
8.48316193e-01 6.36431873e-02 -3.48414809e-01 6.92925572e-01
1.62975103e-01 6.00889683e-01 1.11041568e-01 -1.25287294e+00
6.61316156e-01 7.10210323e-01 3.21311615e-02 -7.04104543e-01
-8.98752511e-01 -3.50321531e-01 -3.61799955e-01 4.25995499e-01
6.57914758e-01 -9.00687724e-02 -1.02050197e+00 1.44898081e+00
3.98637712e-01 4.34584409e-01 -1.39943473e-02 7.04731822e-01
1.20616543e+00 5.32954335e-01 3.42034400e-01 4.58605170e-01
1.35647333e+00 -1.21778727e+00 -7.50647187e-01 -6.77305520e-01
1.16623604e+00 -2.58988589e-01 1.10032368e+00 3.12393457e-02
-7.90784061e-01 -8.94877315e-01 -1.18940341e+00 1.33968079e-02
-9.29374933e-01 2.45386958e-01 7.40684986e-01 6.51558697e-01
-1.07331824e+00 1.92861021e-01 -9.52589571e-01 -5.94066262e-01
9.42279935e-01 5.78968465e-01 -7.53168762e-02 -1.27689585e-01
-1.04521847e+00 1.12029052e+00 6.72658980e-01 -8.57139006e-02
-1.14300132e+00 -8.72820973e-01 -8.51167262e-01 -4.45656717e-01
4.45505381e-01 -5.47745049e-01 1.19802785e+00 -3.66841674e-01
-9.61773694e-01 9.41774607e-01 -1.21723572e-02 -6.91658080e-01
4.98103648e-01 1.89224243e-01 -6.21189058e-01 -9.30387080e-02
5.41148543e-01 1.62204027e+00 5.28125584e-01 -1.20300364e+00
-1.36879194e+00 3.37947793e-02 2.47266799e-01 1.27813548e-01
-7.29304180e-03 -4.33283210e-01 -5.06621599e-01 -2.57031143e-01
-2.77837783e-01 -1.16961336e+00 -1.02533609e-01 -1.14811458e-01
-2.55602866e-01 -5.01788259e-01 1.60429549e+00 -1.59138262e-01
1.05288863e+00 -2.25439787e+00 -4.01922643e-01 -2.82045722e-01
2.18342572e-01 5.70794225e-01 -2.09158078e-01 6.34851158e-02
3.35879534e-01 6.82266057e-02 -1.65701002e-01 -4.56567913e-01
1.83526456e-01 5.93606651e-01 7.86430314e-02 4.48699206e-01
3.87065649e-01 1.22745156e+00 -1.02510190e+00 -8.27734768e-01
6.71253502e-01 2.31903911e-01 -5.43660045e-01 -8.44307244e-02
-5.57039976e-01 7.54968882e-01 -2.86181718e-01 8.50790799e-01
5.84896564e-01 -3.71593200e-02 -4.37820941e-01 -3.91703159e-01
-4.44543242e-01 3.03895354e-01 -1.15609658e+00 1.58655131e+00
-3.03239048e-01 1.35151303e+00 -3.26834232e-01 -6.69844568e-01
7.54439414e-01 -1.93199694e-01 6.83164477e-01 -1.18582726e+00
2.17906460e-02 2.52579123e-01 1.35850925e-02 -7.35493124e-01
1.03412020e+00 3.17602724e-01 -2.31391624e-01 -7.50930384e-02
-6.47386536e-02 -1.12014964e-01 6.11854315e-01 1.33193985e-01
1.04514158e+00 4.71044332e-01 -3.26638490e-01 -3.67027730e-01
1.68106049e-01 6.37728930e-01 5.39884329e-01 7.12906003e-01
-7.44852006e-01 1.82236567e-01 2.95449048e-01 -5.47920346e-01
-8.94561291e-01 -5.97838998e-01 -4.15548921e-01 1.41422105e+00
6.52047396e-01 -3.41859192e-01 -6.99634612e-01 -7.72881031e-01
1.11122914e-01 7.94439733e-01 -6.71911418e-01 -4.30883877e-02
-5.94719291e-01 -6.62302554e-01 1.08170986e+00 8.94564450e-01
8.03054392e-01 -1.10706866e+00 -1.12256932e+00 2.04846576e-01
-2.43888795e-01 -1.55647123e+00 -1.88310593e-01 4.01505321e-01
-5.13872325e-01 -1.18954563e+00 -1.84108272e-01 -6.86265469e-01
2.98340946e-01 3.60265255e-01 1.05907929e+00 -6.65409192e-02
-2.84019977e-01 1.42542586e-01 -2.20041007e-01 -6.21393442e-01
-2.34711125e-01 3.95785272e-01 -1.75169647e-01 -4.09380317e-01
5.00959337e-01 -2.23782390e-01 -5.74332714e-01 7.77402043e-01
-6.75250888e-01 4.33634400e-01 4.84888464e-01 4.63585019e-01
6.97609186e-01 2.05406129e-01 4.74729896e-01 -6.88546002e-01
1.95147877e-03 -5.25756896e-01 -8.82696450e-01 -9.44431797e-02
-6.51668966e-01 -1.44319534e-01 1.64106801e-01 -4.13640946e-01
-8.35814536e-01 1.46107182e-01 -1.17314860e-01 -2.13190928e-01
-5.61505556e-01 9.87962186e-02 1.68957561e-02 -2.91892439e-01
5.85301280e-01 -1.19232126e-01 -3.29013765e-01 -1.07896708e-01
3.98406178e-01 6.17429376e-01 7.92782903e-01 -4.62148428e-01
5.64506888e-01 5.62827468e-01 4.47191089e-01 -7.05654681e-01
-7.94574857e-01 -6.82251096e-01 -8.66752863e-01 -6.31373405e-01
1.18505490e+00 -1.04551792e+00 -8.70707631e-01 3.59816641e-01
-9.75472748e-01 -8.60707402e-01 -4.58799779e-01 2.54424155e-01
-4.93126154e-01 -1.27340585e-01 -6.38305172e-02 -7.69021928e-01
1.27294078e-01 -1.49484980e+00 1.52761900e+00 3.06199759e-01
-1.41030345e-02 -1.06020796e+00 -1.58779293e-01 6.14150107e-01
4.92308646e-01 6.09433293e-01 4.41245526e-01 -4.17860448e-01
-9.35866833e-01 -1.17636370e-02 -5.42150140e-01 -8.65230616e-03
-1.54831961e-01 2.41806936e-02 -1.08817732e+00 -1.25309825e-01
-8.64272296e-01 -3.73658091e-01 8.96899462e-01 5.47751606e-01
1.02348208e+00 3.36786211e-02 -6.89546049e-01 7.29338825e-01
1.06933355e+00 3.00662994e-01 4.96317804e-01 6.01449192e-01
9.35989499e-01 7.24449456e-01 1.14868176e+00 6.68943822e-02
8.34998608e-01 9.38978493e-01 9.01632011e-01 -1.95836633e-01
-5.79077542e-01 -1.70558736e-01 1.49450392e-01 3.20270628e-01
1.90181762e-01 -6.01520360e-01 -1.25424707e+00 9.46306646e-01
-2.08635235e+00 -6.50777876e-01 -8.07932556e-01 1.56965888e+00
3.41572285e-01 5.52160501e-01 4.70873952e-01 -7.25311367e-03
4.83521134e-01 2.76117384e-01 -7.51425207e-01 -7.57476389e-02
-4.07702588e-02 -2.82200664e-01 1.21910799e+00 2.68294930e-01
-1.29186845e+00 1.24931788e+00 5.99981642e+00 1.00124311e+00
-9.64736938e-01 2.18127921e-01 2.33511746e-01 -1.69720724e-01
1.27358928e-01 -9.26597342e-02 -1.55885196e+00 5.58369040e-01
1.35000861e+00 4.80358601e-01 2.72164106e-01 7.60136366e-01
2.91974306e-01 -2.98386663e-01 -1.08041954e+00 9.40332532e-01
-2.65067905e-01 -1.58728802e+00 -4.72153574e-01 1.54758453e-01
8.22728515e-01 6.04344845e-01 1.12715192e-01 5.91168284e-01
4.28128630e-01 -9.20810580e-01 1.02050710e+00 2.78686762e-01
1.15678358e+00 -4.83259916e-01 8.41181099e-01 3.90270174e-01
-1.74510169e+00 -6.44336715e-02 -1.77596554e-01 1.11011833e-01
3.45607281e-01 6.11524517e-03 -9.48583841e-01 5.91827035e-01
1.05938125e+00 1.20474923e+00 -1.23709822e+00 1.03923118e+00
2.94995219e-01 7.96262085e-01 -6.05972588e-01 -3.79732661e-02
7.53814101e-01 1.60721630e-01 5.05288720e-01 1.62159038e+00
-1.17706828e-01 -3.48554760e-01 3.39012116e-01 7.19524205e-01
1.68077677e-01 -4.70660478e-01 -6.85165346e-01 -1.48741540e-03
5.53495467e-01 1.32670772e+00 -1.02537298e+00 -3.88818651e-01
-2.97029316e-01 2.70302683e-01 -5.96224479e-02 1.60260454e-01
-1.25435925e+00 -5.93227558e-02 8.72247398e-01 4.05597299e-01
3.98388922e-01 -2.24671125e-01 -5.47363639e-01 -2.36821592e-01
-2.88649440e-01 -3.98371726e-01 3.54242414e-01 -9.56367135e-01
-8.23408008e-01 4.88888174e-01 5.33836126e-01 -1.39328980e+00
-7.33847469e-02 -6.99718058e-01 -2.55965322e-01 5.35053551e-01
-1.98673928e+00 -1.58578515e+00 -8.73351753e-01 4.33327496e-01
7.63827860e-01 -1.75724849e-01 3.18927824e-01 7.39506006e-01
-7.75557339e-01 6.20086074e-01 -1.73352569e-01 -1.62075117e-01
4.28177387e-01 -1.11005759e+00 6.43581092e-01 6.38950408e-01
-1.79804280e-01 8.87328759e-03 5.82243085e-01 -6.55642927e-01
-1.11958814e+00 -1.77379155e+00 4.08422381e-01 -9.41610634e-01
5.02366483e-01 -3.79691839e-01 -5.99823415e-01 8.16881537e-01
2.14523464e-01 3.93298864e-01 2.33594298e-01 -2.75798738e-01
-1.14714347e-01 -3.28253269e-01 -1.26439309e+00 4.37193573e-01
1.39863741e+00 -2.66175508e-01 -8.68609995e-02 4.28916812e-01
8.26921046e-01 -8.41005147e-01 -7.37721205e-01 5.10837495e-01
4.39136028e-01 -8.52743447e-01 8.03111613e-01 -3.09664756e-01
6.33709729e-02 -8.01145017e-01 -2.55065143e-01 -8.68160009e-01
-1.56422600e-01 -3.05996180e-01 -8.64197612e-02 1.33597291e+00
3.71328264e-01 -5.33449888e-01 8.94706726e-01 3.54880601e-01
-8.20623219e-01 -8.15387845e-01 -1.18549991e+00 -9.22130525e-01
-3.41941893e-01 -8.53602171e-01 7.88671255e-01 7.52430737e-01
-6.78873777e-01 1.66359439e-01 -1.74965769e-01 2.47242540e-01
5.60092330e-01 -4.39483583e-01 1.08533013e+00 -1.30750692e+00
2.26159334e-01 -4.53500539e-01 -6.58916950e-01 -1.24247789e+00
4.53006215e-02 -9.43012774e-01 2.20375672e-01 -1.81185961e+00
-3.33473027e-01 -9.67330396e-01 2.23326124e-02 8.19125593e-01
1.67934433e-01 6.25200093e-01 1.33476220e-02 2.48383328e-01
-1.17737222e+00 3.01265419e-01 1.28500044e+00 -3.43704104e-01
-1.62447408e-01 -1.55993477e-01 -4.01083350e-01 5.17714739e-01
7.30537713e-01 -7.13070989e-01 -5.05036116e-01 -4.23543394e-01
4.49302942e-02 -4.04638439e-01 7.39771187e-01 -1.42520130e+00
3.46750408e-01 -2.77107865e-01 8.01153108e-02 -1.32305312e+00
5.71118474e-01 -9.81455207e-01 3.26848179e-01 3.29206467e-01
-1.28914759e-01 4.94005717e-02 8.72334242e-01 5.08125186e-01
2.01377735e-01 1.12333959e-02 8.04319561e-01 1.51695371e-01
-1.49290836e+00 2.43512601e-01 -3.16468924e-01 3.72462660e-01
1.18373227e+00 -7.69899189e-01 -7.57284224e-01 1.14605933e-01
-7.23962486e-02 6.36504889e-01 2.68393695e-01 7.47549117e-01
2.15840399e-01 -1.27097011e+00 -3.67827654e-01 2.22622007e-01
5.49762726e-01 4.94333506e-01 2.18496978e-01 1.01722121e+00
-5.40205300e-01 6.56506002e-01 -3.70322436e-01 -1.32343507e+00
-1.20683968e+00 1.23413354e-01 3.99104208e-01 -1.65279165e-01
-7.84236073e-01 8.34995270e-01 5.53751215e-02 -3.73362631e-01
3.75060171e-01 -6.23361707e-01 -2.42278919e-01 1.21617608e-01
1.23456195e-01 6.90070331e-01 1.16559535e-01 -9.39520061e-01
-5.71847677e-01 4.50277537e-01 1.81583688e-01 8.24496448e-02
1.11060894e+00 -2.31730759e-01 5.09723425e-01 4.81315643e-01
9.89508033e-01 -4.85601693e-01 -1.81678069e+00 1.95891336e-01
1.72072962e-01 -3.02686095e-01 1.43001303e-01 -1.03848076e+00
-1.02525342e+00 8.32842827e-01 1.11294317e+00 9.15900245e-02
7.84436584e-01 1.76422551e-01 8.63228619e-01 1.98932618e-01
4.11683679e-01 -1.27279484e+00 -1.42216133e-02 5.93395412e-01
4.86450762e-01 -1.48057795e+00 -3.22538674e-01 -2.90957242e-01
-7.83008099e-01 6.89065754e-01 1.08711743e+00 3.34521830e-01
3.47730041e-01 3.44656795e-01 1.53837979e-01 -6.02065682e-01
-5.10622740e-01 -5.62050879e-01 4.75081563e-01 9.12797034e-01
4.73292992e-02 -4.39668596e-02 3.13374698e-01 1.83569774e-01
-2.63505667e-01 -1.08080050e-02 1.74713507e-01 9.21694756e-01
-5.13969004e-01 -6.73380315e-01 -1.69506192e-01 5.25900960e-01
9.94109958e-02 2.11795166e-01 -1.32179543e-01 1.09153175e+00
8.88128579e-01 1.07397377e+00 1.40763834e-01 -6.90398753e-01
6.61472559e-01 -1.25900835e-01 3.16464275e-01 -4.23609346e-01
-4.88645285e-01 -3.13579559e-01 3.37729186e-01 -6.55317843e-01
-6.68751657e-01 -7.10019588e-01 -1.49128103e+00 -3.94557595e-01
-4.41489041e-01 -2.32293710e-01 9.83726084e-01 1.09342408e+00
4.49264020e-01 1.30368125e+00 -1.09086260e-02 -1.16271281e+00
2.56719470e-01 -9.39033628e-01 -2.72960842e-01 2.18150750e-01
3.87573451e-01 -1.17957056e+00 5.90610802e-02 3.01943403e-02] | [7.864168167114258, -1.5466086864471436] |
7422889b-1478-4b71-a7e2-ebce8fb98c8f | automatic-liver-and-tumor-segmentation-of-ct | 1702.05970 | null | http://arxiv.org/abs/1702.05970v2 | http://arxiv.org/pdf/1702.05970v2.pdf | Automatic Liver and Tumor Segmentation of CT and MRI Volumes using Cascaded Fully Convolutional Neural Networks | Automatic segmentation of the liver and hepatic lesions is an important step
towards deriving quantitative biomarkers for accurate clinical diagnosis and
computer-aided decision support systems. This paper presents a method to
automatically segment liver and lesions in CT and MRI abdomen images using
cascaded fully convolutional neural networks (CFCNs) enabling the segmentation
of a large-scale medical trial or quantitative image analysis. We train and
cascade two FCNs for a combined segmentation of the liver and its lesions. In
the first step, we train a FCN to segment the liver as ROI input for a second
FCN. The second FCN solely segments lesions within the predicted liver ROIs of
step 1. CFCN models were trained on an abdominal CT dataset comprising 100
hepatic tumor volumes. Validations on further datasets show that CFCN-based
semantic liver and lesion segmentation achieves Dice scores over 94% for liver
with computation times below 100s per volume. We further experimentally
demonstrate the robustness of the proposed method on an 38 MRI liver tumor
volumes and the public 3DIRCAD dataset. | ['Seyed-Ahmad Ahmadi', 'Freba Ahmaddy', 'Sebastian Schlecht', 'Jana Lipkova', 'Felix Grün', 'Wieland Sommer', 'Rickmer Braren', 'Mohamed Ezzeldin A. Elshaera', 'Melvin D Anastasi', 'Markus Rempfler', 'Marc Bickel', 'Bjoern Menze', 'Volker Heinemann', 'Julian Holch', 'Sunil Tatavarty', 'Patrick Ferdinand Christ', 'Patrick Bilic', 'Georgios Kaissis', 'Florian Ettlinger', 'Felix Hofmann'] | 2017-02-20 | null | null | null | null | ['automatic-liver-and-tumor-segmentation'] | ['medical'] | [ 4.36137021e-02 1.91921443e-01 -8.67283642e-02 -6.04829729e-01
-6.62060559e-01 -4.87988800e-01 3.90466988e-01 2.35573247e-01
-3.38011891e-01 4.01987731e-01 2.65294373e-01 -5.55667162e-01
7.31232166e-02 -6.56141043e-01 -2.42261454e-01 -7.75346577e-01
-6.36498868e-01 8.72350574e-01 1.41951263e-01 5.84929228e-01
-1.36504292e-01 1.06688309e+00 -3.57277185e-01 3.30769002e-01
8.90548646e-01 1.15819430e+00 1.03562959e-01 9.03464317e-01
2.53598560e-02 9.12255645e-01 -2.64020473e-01 -7.50826597e-02
5.98839939e-01 -7.05643952e-01 -9.34496641e-01 3.65703821e-01
1.41569108e-01 -5.39347410e-01 -2.41589576e-01 1.07984376e+00
6.01339221e-01 -3.97890300e-01 7.48815119e-01 -7.81853557e-01
-2.30806172e-01 1.08032930e+00 -2.93669671e-01 3.92463714e-01
-3.35138321e-01 4.88921553e-01 1.69380337e-01 -7.99814820e-01
3.80603999e-01 5.87407768e-01 1.02294123e+00 3.99349034e-01
-9.29119408e-01 -5.15101612e-01 -8.56281102e-01 -2.77613282e-01
-1.29407227e+00 -1.38664544e-01 1.84081674e-01 -7.78930783e-01
5.88194489e-01 3.46506923e-01 9.66600597e-01 1.28201604e-01
3.51230413e-01 5.94989777e-01 1.10509956e+00 -1.83122620e-01
1.07034691e-01 -2.35610142e-01 4.14253883e-02 1.17671430e+00
3.48363936e-01 2.46716022e-01 3.50847334e-01 -1.22626178e-01
1.19252360e+00 6.48441017e-02 -3.31381977e-01 -5.03149033e-01
-1.84041834e+00 9.78190362e-01 1.11158919e+00 5.37720978e-01
-8.05741787e-01 3.46768111e-01 5.78523219e-01 -1.32865667e-01
3.31979036e-01 3.92934054e-01 -1.51550472e-01 5.95339060e-01
-1.28971791e+00 -4.65291649e-01 9.24174190e-01 6.50580943e-01
1.16488852e-01 1.77266710e-02 -6.07594550e-01 4.28188354e-01
3.66228938e-01 2.58316606e-01 7.57233679e-01 -6.89238310e-01
-2.69774139e-01 5.96922278e-01 -2.61984825e-01 -3.73186350e-01
-8.23852837e-01 -6.13044024e-01 -1.17321539e+00 1.80527329e-01
5.98009765e-01 -2.98952013e-01 -1.41332889e+00 9.39819455e-01
5.29699206e-01 8.35275799e-02 -2.00997621e-01 1.39175272e+00
1.21082938e+00 1.60744954e-02 4.86052811e-01 -1.64083347e-01
1.45834160e+00 -1.15428734e+00 -3.90809149e-01 3.36013198e-01
9.26013768e-01 -6.53917253e-01 2.49879718e-01 -6.28873259e-02
-1.12130582e+00 -4.45851199e-02 -7.85840154e-01 1.92808345e-01
-1.27082197e-02 6.95215225e-01 7.27947414e-01 5.69841683e-01
-1.33245909e+00 6.52620792e-01 -1.34015608e+00 -3.70681316e-01
9.23228323e-01 7.58887589e-01 -2.39968091e-01 -1.54603615e-01
-7.91421056e-01 1.06996107e+00 5.93410492e-01 2.43235141e-01
-1.45724714e+00 -1.18819261e+00 -9.18177485e-01 1.25530109e-01
-8.50624293e-02 -9.13020611e-01 1.22938502e+00 -9.22644198e-01
-1.28556383e+00 1.05537689e+00 3.39064211e-01 -8.63027811e-01
8.59912455e-01 5.50390303e-01 6.53674603e-02 5.86954176e-01
3.30552310e-02 9.80047822e-01 4.64350849e-01 -8.70215118e-01
-2.35009730e-01 -2.19691649e-01 -5.86996436e-01 1.51979387e-01
4.16860253e-01 7.06402883e-02 -6.61958009e-02 -5.97916126e-01
1.78105950e-01 -8.52087855e-01 -5.64966679e-01 3.70829523e-01
-5.89718044e-01 6.99100792e-02 6.54631853e-01 -1.16626358e+00
5.45335233e-01 -1.68747652e+00 -1.18747316e-02 4.19279069e-01
5.34786642e-01 2.28597537e-01 1.02599505e-02 -5.93425393e-01
-3.23677242e-01 1.26382709e-01 -4.21878636e-01 7.59231225e-02
-3.56558084e-01 -7.04228133e-02 5.77091992e-01 1.04331899e+00
3.49650905e-02 1.47217476e+00 -8.99709702e-01 -7.39397466e-01
4.61987227e-01 6.28604472e-01 -2.98172951e-01 2.48075128e-01
9.45238322e-02 8.55400741e-01 -3.26142102e-01 8.41824710e-01
5.78511953e-01 -6.12286687e-01 2.76449412e-01 -4.72101450e-01
-4.26945947e-02 -1.29401654e-01 -3.23171616e-01 1.77245378e+00
-2.85219938e-01 4.32709485e-01 3.08749914e-01 -9.38701630e-01
4.63084161e-01 6.00534379e-01 1.12134194e+00 -4.31183606e-01
5.83138883e-01 4.44086254e-01 5.20981431e-01 -4.56012666e-01
-4.56532091e-01 -5.35215497e-01 2.16256037e-01 5.91566682e-01
2.91644812e-01 -3.25207531e-01 2.74126261e-01 5.94466329e-02
9.99065101e-01 -2.69256830e-01 3.73691678e-01 -8.74208629e-01
7.43929207e-01 4.49945152e-01 1.66990519e-01 3.15564334e-01
-7.59980738e-01 6.74209893e-01 3.46638590e-01 -8.63350153e-01
-1.08310258e+00 -9.17614818e-01 -3.16110671e-01 5.18409193e-01
-1.91651151e-01 3.17044318e-01 -9.36836421e-01 -1.15669918e+00
-6.59761727e-02 3.55709195e-01 -7.87599981e-01 3.52401555e-01
-6.70046985e-01 -1.15721035e+00 6.09448195e-01 4.56786156e-01
5.49163103e-01 -9.37923551e-01 -9.67820168e-01 2.53166974e-01
-1.82774767e-01 -8.67723703e-01 -4.43204522e-01 2.16618419e-01
-1.03835809e+00 -1.67811072e+00 -1.23527634e+00 -1.12084413e+00
1.12577844e+00 -1.39393508e-01 1.27791929e+00 5.35671175e-01
-9.42567647e-01 8.14596042e-02 1.61964193e-01 -1.74385846e-01
-8.24239910e-01 -1.23871520e-01 -3.20237160e-01 -5.04924953e-01
8.69024545e-02 -8.45225453e-02 -1.13355649e+00 1.41502514e-01
-6.52720094e-01 2.21406743e-01 8.45941782e-01 9.62793410e-01
6.02894783e-01 -2.37300724e-01 3.61742407e-01 -6.51454568e-01
3.17513168e-01 -6.56119347e-01 -7.29742885e-01 3.76476765e-01
-3.79158348e-01 -1.49515867e-01 3.37820947e-01 -2.21765250e-01
-7.30090559e-01 5.71349978e-01 -9.57618058e-02 -3.39135081e-01
-3.16952288e-01 5.96320271e-01 6.28225625e-01 -4.97664392e-01
6.28462911e-01 1.43979400e-01 3.71033520e-01 -7.15738684e-02
2.18799606e-01 2.27219701e-01 5.68199039e-01 -1.04957230e-01
4.44341362e-01 4.61651504e-01 3.56601179e-01 -3.62721205e-01
-4.37418461e-01 -4.85825479e-01 -1.14214599e+00 -1.99823022e-01
1.23636818e+00 -8.23326468e-01 -5.17935574e-01 2.15797201e-01
-8.43515098e-01 -7.31151998e-01 -3.98389578e-01 7.55167723e-01
-4.34962273e-01 2.83062637e-01 -1.21011090e+00 1.00089395e-02
-9.32539701e-01 -1.71539044e+00 6.89137697e-01 2.55378872e-01
1.04112953e-01 -1.32667518e+00 -1.67908490e-01 1.08723335e-01
8.54909003e-01 6.65505290e-01 9.68183696e-01 -1.03749645e+00
-4.86036569e-01 -2.20698729e-01 -6.33857965e-01 1.84580341e-01
9.54304785e-02 -2.55420476e-01 -4.25842583e-01 -2.86588848e-01
-1.85354412e-01 -2.20014095e-01 8.06222320e-01 9.67754781e-01
1.14850879e+00 -1.95805579e-01 -2.57255495e-01 9.11947250e-01
1.59110320e+00 2.20416725e-01 1.52625024e-01 -1.41275451e-01
5.94826996e-01 1.92291111e-01 -1.57326683e-02 1.91642016e-01
4.99482965e-03 -8.03843364e-02 6.31055534e-01 -7.83680737e-01
-7.76901186e-01 3.72656524e-01 -2.62352288e-01 7.87679255e-01
1.24781102e-01 1.64611802e-01 -1.36573982e+00 7.87521124e-01
-1.10144091e+00 -4.01130050e-01 -4.47867721e-01 1.81900585e+00
8.53102803e-01 -5.31781554e-01 -4.37609591e-02 -3.98759156e-01
7.47845888e-01 -5.50046027e-01 -5.24625063e-01 1.00940257e-01
3.25577348e-01 3.17106366e-01 6.87810242e-01 5.24001122e-01
-1.40211403e+00 5.34289539e-01 6.39497662e+00 3.83524150e-01
-1.38284135e+00 3.42441291e-01 1.24037337e+00 1.52266100e-01
1.84012666e-01 -3.22799683e-01 -3.45842764e-02 1.47533610e-01
8.20159495e-01 -9.11025479e-02 3.72385561e-01 6.19973481e-01
2.77698785e-01 -2.86369503e-01 -1.11040223e+00 5.09070694e-01
-1.50808185e-01 -1.52402973e+00 -4.65071239e-02 7.95565695e-02
9.45222139e-01 5.57410538e-01 -1.58659846e-01 4.81587090e-02
4.67489988e-01 -1.51513255e+00 1.86277792e-01 3.08215052e-01
1.37325561e+00 -4.82752353e-01 1.29444981e+00 1.34676665e-01
-8.88404429e-01 2.45096862e-01 -1.12227172e-01 5.26128232e-01
-1.10353008e-01 6.58424020e-01 -1.72694862e+00 4.18650180e-01
3.23209405e-01 4.32795107e-01 -6.95057392e-01 1.60960519e+00
-3.42071131e-02 6.41128898e-01 -2.89187282e-01 3.87717187e-01
4.30318326e-01 -1.41893417e-01 2.99862564e-01 1.61383033e+00
4.31527704e-01 4.54296976e-01 3.54249686e-01 1.08008933e+00
-3.66918057e-01 1.58474952e-01 -7.10870177e-02 3.60928699e-02
-4.09028819e-03 1.89512730e+00 -1.54502726e+00 -7.93182790e-01
-1.31390005e-01 7.54744351e-01 -1.51047826e-01 9.25832987e-02
-7.93528020e-01 7.65626132e-02 -9.01392326e-02 3.96037428e-03
-1.69839506e-04 1.68844789e-01 -4.77059662e-01 -1.18012655e+00
-8.38120759e-01 -4.92698789e-01 4.84443337e-01 -3.71371537e-01
-1.19827437e+00 6.05121553e-01 -3.47646177e-01 -7.98159242e-01
-5.09006418e-02 -4.41844553e-01 -7.60327578e-01 1.04110301e+00
-1.68034232e+00 -1.21358907e+00 -7.85837710e-01 4.59923118e-01
3.82593304e-01 1.44638792e-01 8.62924039e-01 2.11031958e-01
-2.89520741e-01 6.86603459e-03 -2.01394469e-01 9.24571574e-01
3.24497610e-01 -1.58029842e+00 -1.11318663e-01 8.27960074e-01
-4.55882072e-01 2.78441370e-01 2.38980219e-01 -7.89634466e-01
-1.06141329e+00 -1.50513899e+00 5.24249375e-01 -1.00634154e-02
4.21042442e-01 3.06790322e-01 -6.07255399e-01 7.13680089e-01
4.74750489e-01 8.55595410e-01 8.32772434e-01 -9.24020171e-01
2.38944426e-01 4.23715621e-01 -1.81365240e+00 5.46682766e-03
3.19536388e-01 1.68678071e-02 -4.41338360e-01 7.37723291e-01
2.92380482e-01 -8.00451815e-01 -1.46888542e+00 7.54445016e-01
3.76996517e-01 -8.18283141e-01 1.08658791e+00 -4.30775940e-01
5.02047658e-01 -3.70886683e-01 2.65598863e-01 -1.37381935e+00
-2.83329129e-01 -1.02857053e-01 -7.18664564e-03 1.19907364e-01
2.38658547e-01 -1.19605713e-01 8.22914183e-01 4.98978287e-01
-4.09342796e-01 -8.05845022e-01 -7.66476214e-01 -1.95914537e-01
5.72921634e-01 1.53517529e-01 3.98342937e-01 1.09631217e+00
-2.83865869e-01 -3.37345958e-01 5.65747797e-01 1.28367573e-01
1.03934741e+00 1.01903714e-01 -2.88059283e-02 -1.17509091e+00
1.77564800e-01 -8.32153201e-01 -2.99438447e-01 -3.98797750e-01
-6.23382740e-02 -1.51573849e+00 8.41982514e-02 -1.87276149e+00
7.47544467e-01 -6.83125854e-01 -3.74700397e-01 6.17714822e-01
2.43619597e-03 7.88790882e-01 4.22764048e-02 2.40947440e-01
-2.97327369e-01 7.72814220e-03 1.63606668e+00 -3.66948158e-01
8.43904242e-02 -8.77184421e-02 -6.99172914e-02 6.65441096e-01
7.89429128e-01 -3.28750968e-01 -2.17481479e-02 -1.66833088e-01
-6.16459191e-01 4.86561239e-01 7.11438298e-01 -1.04429674e+00
2.62531072e-01 1.51151851e-01 1.29229057e+00 -5.78365266e-01
-4.92326468e-01 -8.64418447e-01 2.60355443e-01 1.48423183e+00
-3.89569193e-01 -2.07189664e-01 6.01457767e-02 6.95014521e-02
-4.75944020e-02 -1.75300147e-02 1.28355956e+00 -7.76731789e-01
-2.61323273e-01 5.96086025e-01 -5.03313303e-01 -2.49187827e-01
1.32949233e+00 2.56797522e-01 1.73783898e-02 -7.30528086e-02
-1.19945729e+00 2.98920095e-01 1.89402431e-01 -2.60241747e-01
6.50822639e-01 -1.20862782e+00 -1.00673103e+00 2.33099028e-01
-4.07518327e-01 3.10906708e-01 2.03872360e-02 1.73363554e+00
-1.49677968e+00 8.10389400e-01 -2.55502731e-01 -9.44390059e-01
-1.09879184e+00 4.51782376e-01 1.26046526e+00 -6.47638619e-01
-6.85507298e-01 8.35410058e-01 1.61967933e-01 -5.24433613e-01
3.48507576e-02 -7.60560036e-01 -6.28147796e-02 -2.65676290e-01
3.75836432e-01 6.11431338e-02 2.41561279e-01 -8.12277496e-01
-3.14945906e-01 8.67732689e-02 3.13407719e-01 2.52045631e-01
1.37801766e+00 7.35333711e-02 -5.40923119e-01 -3.48787636e-01
1.20594335e+00 -5.21620154e-01 -1.09906709e+00 -2.11609557e-01
1.84985831e-01 -3.86978090e-02 6.44661665e-01 -1.56738329e+00
-1.64890885e+00 8.74186695e-01 8.51408899e-01 -7.72964284e-02
1.10106695e+00 -1.14962511e-01 6.82559371e-01 -3.21238756e-01
1.36338681e-01 -1.28870443e-01 -1.61165372e-01 1.60836026e-01
6.97151721e-01 -1.38164926e+00 2.22390264e-01 -4.51086700e-01
-5.44711053e-01 1.59713316e+00 4.75961447e-01 -1.82151765e-01
6.72937155e-01 4.16253239e-01 3.09552789e-01 -4.74238932e-01
-1.70145795e-01 -8.88494700e-02 5.24878323e-01 3.38087350e-01
6.69288516e-01 4.59209681e-01 -2.86233097e-01 2.93336183e-01
2.14815542e-01 2.81690717e-01 4.53401953e-01 7.41530418e-01
-3.84208858e-01 -5.12239456e-01 -2.55880147e-01 6.60086572e-01
-7.87283242e-01 -3.18710685e-01 -2.22600609e-01 8.61301482e-01
1.67676732e-01 2.98834175e-01 4.62537259e-02 3.41546863e-01
-3.49087745e-01 1.32446080e-01 5.42545021e-01 -5.23742557e-01
-1.25088620e+00 3.02121431e-01 -3.00334990e-01 -5.76521218e-01
-4.01091933e-01 -4.75396305e-01 -1.60584593e+00 -1.30723277e-03
-2.07127392e-01 1.88859671e-01 1.00626767e+00 8.26862633e-01
5.57895936e-02 6.69235587e-01 4.84279275e-01 -9.18016732e-01
-5.55407643e-01 -1.16660619e+00 -3.50172728e-01 3.24838877e-01
4.06993002e-01 -2.24926963e-01 -1.65433854e-01 1.95170701e-01] | [14.503683090209961, -2.6657230854034424] |
b33551f8-e81f-4f30-9aa4-81b1ca3b2c03 | sign-pose-based-transformer-for-word-level | null | null | https://openaccess.thecvf.com/content/WACV2022W/HADCV/html/Bohacek_Sign_Pose-Based_Transformer_for_Word-Level_Sign_Language_Recognition_WACVW_2022_paper.html | https://openaccess.thecvf.com/content/WACV2022W/HADCV/papers/Bohacek_Sign_Pose-Based_Transformer_for_Word-Level_Sign_Language_Recognition_WACVW_2022_paper.pdf | Sign Pose-Based Transformer for Word-Level Sign Language Recognition | In this paper we present a system for word-level sign language recognition based on the Transformer model. We aim at a solution with low computational cost, since we see great potential in the usage of such recognition system on handheld devices. We base the recognition on the estimation of the pose of the human body in the form of 2D landmark locations. We introduce a robust pose normalization scheme which takes the signing space in considerationand processes the hand poses in a separate local coordinate system, independent on the body pose. We show experimentally the significant impact of this normalization on the accuracy of our proposed system. We introduce several augmentations of the body pose that further improve the accuracy, including a novel sequential joint rotation augmentation. With all the systems in place, we achieve state of theart top-1 results on the WLASL and LSA64 datasets. For WLASL, we are able to successfully recognize 63.18% of sign recordings in the 100-gloss subset, which is a relative improvement of 5% from the prior state of the art. For the 300-gloss subset, we achieve recognition rate of 43.78% which is a relative improvement of 3.8%. With the LSA64 dataset, we report test recognition accuracy of 100%. | ['Marek Hrúz', 'Matyáš Boháček'] | 2022-01-04 | null | null | null | wacv-2022-1 | ['sign-language-recognition'] | ['computer-vision'] | [ 3.17272097e-01 -1.00729279e-01 1.41019039e-02 -2.91076154e-01
-1.17699087e+00 -4.65079159e-01 6.62742555e-01 -6.18955493e-01
-8.55224252e-01 2.49923587e-01 2.51287222e-01 -9.97453742e-03
2.64926068e-02 -2.13003352e-01 -5.52216470e-01 -8.63011777e-01
2.78004110e-01 4.87453520e-01 2.46911153e-01 -2.88111418e-01
1.33905202e-01 7.57441282e-01 -1.67715073e+00 7.45220929e-02
3.81788224e-01 1.06752264e+00 -1.43218964e-01 9.03937817e-01
3.55329394e-01 2.61937648e-01 -8.70031476e-01 -3.77032369e-01
4.38536167e-01 -2.27632493e-01 -4.87472832e-01 1.68310702e-01
1.15408552e+00 -4.00299907e-01 -1.96291372e-01 5.81841230e-01
9.64487076e-01 -2.26287588e-01 5.97902060e-01 -9.66997445e-01
1.13934778e-01 1.25617921e-01 -4.55101073e-01 -3.40688705e-01
4.39051479e-01 1.25222713e-01 9.68200207e-01 -1.02909076e+00
7.80576468e-01 8.67848873e-01 7.64296710e-01 4.99991566e-01
-9.52527285e-01 -5.82234323e-01 7.67585039e-02 1.40052676e-01
-1.61143804e+00 -7.69599915e-01 5.83565533e-01 -3.51804316e-01
9.57813442e-01 2.78210282e-01 7.40395844e-01 8.89554262e-01
-2.47329026e-01 8.70511889e-01 1.28570747e+00 -9.85007286e-01
-7.85519555e-02 -4.23232853e-01 2.98126757e-01 7.24474370e-01
4.92758572e-01 -1.36149094e-01 -8.85606587e-01 1.91144332e-01
7.93998837e-01 -4.05448943e-01 -3.03547293e-01 -5.55249155e-01
-1.22811067e+00 2.71056741e-01 2.09226355e-01 4.45097357e-01
-2.86743343e-01 2.83025682e-01 1.47117853e-01 7.42259175e-02
-7.81475008e-02 1.47456512e-01 -2.23025128e-01 -4.68600243e-01
-8.90023589e-01 2.53917247e-01 7.44721174e-01 8.75112176e-01
6.93709031e-02 1.69863388e-01 -1.29084125e-01 7.57920682e-01
5.00263333e-01 1.16325784e+00 5.03755748e-01 -5.46314716e-01
5.83866239e-01 3.96732032e-01 1.89835131e-01 -5.32255590e-01
-4.15908039e-01 -3.91704530e-01 -3.46108168e-01 5.08981764e-01
9.65458691e-01 1.04296431e-02 -1.54387832e+00 1.59710157e+00
9.15952027e-02 -1.66191295e-01 4.87361252e-02 1.00731111e+00
5.16967237e-01 1.10810794e-01 -1.01857863e-01 1.06320888e-01
1.48080504e+00 -6.77272737e-01 -6.57659233e-01 -5.87485731e-02
6.47794664e-01 -9.96037245e-01 9.95113432e-01 7.99156547e-01
-8.57911348e-01 -3.62975359e-01 -1.28516996e+00 -2.49011312e-02
-1.48704007e-01 8.53856802e-01 5.03558815e-01 1.00065231e+00
-1.07035649e+00 2.44522661e-01 -1.04907346e+00 -5.99859357e-01
-5.66175394e-02 6.42078161e-01 -5.75557351e-01 1.28306910e-01
-5.25149226e-01 1.08691525e+00 -2.86492649e-02 3.67253214e-01
-1.17621079e-01 -6.03106096e-02 -7.73647726e-01 -4.59147006e-01
2.28082761e-01 -2.60077327e-01 1.18733823e+00 -3.11468095e-01
-1.84289157e+00 1.15247393e+00 -4.08960193e-01 -2.64863193e-01
8.45886409e-01 -5.41846097e-01 -5.02857804e-01 6.24940172e-02
-2.24488050e-01 2.85545617e-01 9.39039409e-01 -1.12255871e+00
-6.18907392e-01 -4.75791126e-01 -4.00323302e-01 2.10976511e-01
-1.58002913e-01 1.51388705e-01 -9.51837540e-01 -7.21795738e-01
5.07014275e-01 -1.36723053e+00 2.58148819e-01 1.39749795e-01
-2.49805078e-01 -3.86416875e-02 5.40053248e-01 -1.09084761e+00
9.00475562e-01 -2.08566618e+00 1.29187658e-01 6.73757255e-01
-1.66586086e-01 5.60572505e-01 -6.41522855e-02 2.12135971e-01
1.92814365e-01 -2.64900416e-01 -1.41720414e-01 -6.35310113e-01
9.82045829e-02 2.09405243e-01 -1.47817969e-01 7.03360379e-01
-6.29297420e-02 7.97029734e-01 -3.70232075e-01 -1.74438775e-01
3.51971626e-01 7.09850490e-01 -4.96456116e-01 5.56776673e-02
2.87044376e-01 4.13297594e-01 -2.18337327e-01 8.18196416e-01
4.92591679e-01 1.64711922e-01 2.58215755e-01 -4.09794122e-01
-1.32110626e-01 3.03892523e-01 -1.47722685e+00 1.70766735e+00
-3.99793625e-01 7.35965967e-01 1.45454422e-01 -4.19066727e-01
9.54516947e-01 4.03694659e-01 3.30854058e-01 -4.27513450e-01
2.52898335e-01 7.10794866e-01 2.39285782e-01 -2.26849422e-01
3.56622070e-01 -1.12078078e-01 5.27533293e-02 2.09483147e-01
1.12903163e-01 -2.01546386e-01 1.60768673e-01 -3.10293585e-01
8.25798988e-01 3.44509274e-01 3.64346117e-01 -2.30985619e-02
5.64821064e-01 -4.51739728e-01 5.95292673e-02 6.85118675e-01
-6.55312166e-02 9.61578369e-01 3.78480628e-02 -1.27565622e-01
-8.16658974e-01 -9.48235095e-01 -1.77009523e-01 6.17584109e-01
-1.52764499e-01 -5.33700049e-01 -6.11718357e-01 -4.32096213e-01
5.16932942e-02 2.66972840e-01 -3.73728573e-01 2.26180702e-01
-1.07564414e+00 -6.53647125e-01 8.81450534e-01 8.11927199e-01
6.02325797e-01 -8.40971649e-01 -6.58845663e-01 -1.84934735e-01
5.44471070e-02 -1.34379137e+00 -4.74167585e-01 -1.08363859e-01
-6.55883133e-01 -1.01671243e+00 -1.07681847e+00 -9.17376101e-01
6.03326857e-01 -2.75576293e-01 4.09370422e-01 -5.67345172e-02
-2.43923053e-01 4.87702191e-01 -3.04967672e-01 -1.59882635e-01
-1.26605704e-01 1.88392028e-02 3.10531557e-01 1.88420370e-01
1.46440253e-01 -3.31506401e-01 -4.74818259e-01 4.98693317e-01
-3.54898155e-01 -1.03654392e-01 7.20179439e-01 8.21470618e-01
3.99569690e-01 -7.04361141e-01 -3.69914323e-02 -3.14554423e-01
1.97401807e-01 6.58597291e-01 -5.87379515e-01 2.55947471e-01
-5.57386577e-01 3.07025760e-01 1.53742865e-01 -4.55432057e-01
-8.94742489e-01 4.29055035e-01 -5.62994957e-01 -1.98500212e-02
-1.85112566e-01 1.87378347e-01 -4.76365417e-01 -6.16229951e-01
4.84578907e-01 1.91894263e-01 9.66238230e-02 -8.77680540e-01
4.32812929e-01 9.83399153e-01 7.62946725e-01 -4.69316006e-01
7.08433926e-01 4.28237766e-01 2.98147410e-01 -1.04947400e+00
-3.37188572e-01 -7.08695471e-01 -9.43123400e-01 -2.50254959e-01
5.26640892e-01 -7.86820590e-01 -8.81401062e-01 1.13955677e+00
-9.53686714e-01 -1.81006223e-01 -2.41379529e-01 8.03003907e-01
-6.02703750e-01 4.32188749e-01 -2.13629961e-01 -9.48707759e-01
-3.17305237e-01 -1.12323165e+00 1.40369630e+00 -8.70741680e-02
-4.06104386e-01 -4.42962289e-01 7.42311496e-03 5.04958630e-01
2.36497134e-01 2.05307201e-01 3.05689931e-01 -5.32447994e-01
-4.21669245e-01 -6.99126124e-01 1.27835022e-02 5.27136564e-01
1.86291263e-01 -1.72421798e-01 -1.17647946e+00 -5.37285030e-01
-3.52464020e-01 -1.34509400e-01 7.84938216e-01 1.23528048e-01
2.99465925e-01 1.10770419e-01 -9.89439934e-02 5.64441800e-01
1.10111547e+00 1.91209838e-01 5.76036274e-01 2.31073990e-01
9.09082174e-01 3.30674648e-01 4.82731581e-01 3.23532611e-01
2.98099548e-01 1.39719939e+00 -3.84146757e-02 -3.96449380e-02
-7.65836716e-01 -2.60053605e-01 4.41712976e-01 9.32088137e-01
-7.02739179e-01 3.42307985e-02 -1.08162630e+00 5.22527456e-01
-1.66305447e+00 -5.72360992e-01 -1.25753134e-01 2.57454658e+00
6.13845408e-01 1.76770147e-02 2.34726220e-01 6.41855776e-01
4.42765146e-01 1.36570051e-01 -1.87538147e-01 -1.97986826e-01
-1.84296399e-01 5.39973259e-01 7.49275863e-01 9.37777758e-01
-1.03108168e+00 1.00510490e+00 6.30586243e+00 4.90244120e-01
-1.50133717e+00 -2.41188228e-01 -2.85597384e-01 -2.12555766e-01
4.94641602e-01 -3.42501432e-01 -1.18144369e+00 9.52016413e-02
7.07453012e-01 3.58367026e-01 2.27159724e-01 6.14543796e-01
-4.49609868e-02 7.13313883e-03 -9.16198015e-01 1.33514619e+00
6.79072380e-01 -5.63080072e-01 5.57170808e-02 3.16282928e-01
4.25921798e-01 -4.40273732e-02 -4.77205850e-02 -1.92184635e-02
-2.96805322e-01 -6.35368764e-01 9.50637877e-01 5.19313693e-01
1.17436159e+00 -4.33581293e-01 7.69777238e-01 9.18263420e-02
-1.15400696e+00 3.07587892e-01 2.53579557e-01 -9.57839340e-02
3.51535887e-01 -8.05633236e-03 -1.04392254e+00 3.05203319e-01
3.59317183e-01 4.61457461e-01 -5.96021473e-01 1.27110183e+00
-7.33129978e-01 6.84832335e-01 -8.12780499e-01 -4.18098308e-02
-3.01103890e-01 1.00834891e-01 6.87024176e-01 1.25170171e+00
4.69516456e-01 -3.99268121e-02 -7.21910223e-02 -6.77321032e-02
4.19344977e-02 3.14771950e-01 -2.53388405e-01 1.96311593e-01
-5.54376207e-02 7.71737218e-01 -4.97374594e-01 -2.95455277e-01
-1.05678961e-01 1.15810382e+00 -5.67771941e-02 2.42613181e-01
-4.28690702e-01 -5.34639299e-01 6.99597895e-01 4.07316126e-02
6.32901609e-01 -6.75539851e-01 -2.94198006e-01 -1.39540768e+00
7.76874721e-01 -1.04696834e+00 1.87400624e-01 -5.67245603e-01
-6.40129209e-01 5.27807236e-01 -1.51300952e-01 -1.33735955e+00
-7.23721087e-01 -1.26650918e+00 -1.68237329e-01 9.93482947e-01
-1.20163417e+00 -1.60665596e+00 -3.23943079e-01 4.72410262e-01
4.58672382e-02 -1.46331683e-01 1.15356457e+00 3.43678981e-01
-9.00322124e-02 1.09319139e+00 8.97461101e-02 3.40711117e-01
9.50815916e-01 -1.08415401e+00 6.79457307e-01 1.07210600e+00
5.78086555e-01 8.14684570e-01 6.31685972e-01 -5.22454143e-01
-1.43617249e+00 -3.24414670e-01 1.38754833e+00 -6.47281647e-01
4.50273663e-01 -6.73328519e-01 -4.42833751e-01 5.77381849e-01
-3.83744240e-01 -1.03618884e-02 4.10579860e-01 2.43043944e-01
-5.23857474e-01 -1.74113676e-01 -1.11823404e+00 6.52612150e-01
1.30203760e+00 -6.67284906e-01 -7.72494674e-01 1.00538790e-01
-5.61268404e-02 -6.54331744e-01 -7.56306350e-01 5.19857883e-01
1.42717433e+00 -5.49889028e-01 8.53766561e-01 -3.04726928e-01
-4.79347289e-01 -5.63642085e-01 -2.86695480e-01 -9.26806748e-01
1.36972278e-01 -5.74262440e-01 -1.14565335e-01 1.15556157e+00
2.72603780e-01 -8.12601507e-01 1.05350959e+00 7.10166097e-01
1.56718180e-01 -4.60683465e-01 -1.31871653e+00 -1.02992058e+00
-2.89622217e-01 -6.44205511e-01 1.62921771e-01 1.65444523e-01
3.64865037e-03 4.53679152e-02 -6.70991004e-01 1.85919344e-01
6.21100366e-01 1.61665216e-01 1.11943185e+00 -1.03222585e+00
-4.33271825e-01 -3.19586515e-01 -9.58641768e-01 -1.38711250e+00
-2.11961210e-01 -4.53223228e-01 1.65479913e-01 -1.38939333e+00
-2.39878342e-01 2.55858134e-02 -2.38959849e-01 7.29760051e-01
1.63233623e-01 7.36701608e-01 5.33159375e-01 2.61537492e-01
-2.45572492e-01 2.67119169e-01 1.00985920e+00 2.67794058e-02
-1.49619162e-01 1.68479666e-01 -1.97481766e-01 8.80850852e-01
6.84622109e-01 -3.23891938e-02 1.81855768e-01 -4.01229203e-01
-3.31454098e-01 -3.48587275e-01 2.83926338e-01 -1.20256793e+00
2.20323011e-01 3.97892982e-01 2.46594876e-01 -5.80596149e-01
8.58926594e-01 -8.35935950e-01 -9.36691314e-02 6.15083992e-01
-5.54335192e-02 -1.87517747e-01 2.83634454e-01 8.15521702e-02
-1.19908020e-01 3.80141251e-02 6.50541782e-01 4.33073640e-01
-6.47185326e-01 -2.15375736e-01 -2.47723252e-01 -2.16669634e-01
5.35017431e-01 -4.18149978e-01 -8.03358182e-02 -4.87399697e-01
-7.02081978e-01 -2.97318339e-01 4.76817697e-01 4.36814845e-01
3.79054099e-01 -1.21702182e+00 -6.01281822e-01 6.55087888e-01
2.76603013e-01 -4.73804206e-01 -2.47518659e-01 8.15092683e-01
-6.79075837e-01 7.61737287e-01 -1.94949299e-01 -4.49068516e-01
-1.92223060e+00 -2.47562960e-01 2.84609556e-01 -8.96639004e-02
-5.81057668e-01 7.55979300e-01 -3.74350280e-01 -3.30801755e-01
5.11682630e-01 -7.23083973e-01 -1.14382252e-01 -1.21330973e-02
5.29961288e-01 3.75974715e-01 3.25007230e-01 -1.20914924e+00
-7.64916658e-01 1.29591501e+00 3.62919420e-02 -6.63343012e-01
1.16091263e+00 3.07726979e-01 3.00615340e-01 2.83453554e-01
1.05532110e+00 6.26645923e-01 -9.91979599e-01 -1.35108069e-01
-2.68050190e-02 -6.12225115e-01 -1.69031337e-01 -1.20211327e+00
-6.75556660e-01 8.25617909e-01 9.23799276e-01 -4.43481803e-01
9.67376649e-01 -1.33770257e-01 4.51706380e-01 5.47689974e-01
6.47512555e-01 -9.09897864e-01 -4.17208880e-01 6.22287571e-01
1.20856500e+00 -1.10504413e+00 1.00878529e-01 -4.04448301e-01
-2.86825150e-01 9.21433151e-01 1.69013008e-01 -1.52008504e-01
3.46963793e-01 3.13570619e-01 5.72214901e-01 2.38354169e-02
1.22447900e-01 -5.06614387e-01 8.70838583e-01 5.08173347e-01
6.08579040e-01 1.84760243e-01 -6.57062888e-01 3.68422925e-01
-4.51430500e-01 -6.71008676e-02 1.78577751e-02 1.20359790e+00
-2.27364153e-01 -1.49240375e+00 -5.89908242e-01 5.60580939e-02
-2.66543925e-01 -4.02508117e-02 -6.72119856e-01 1.03234100e+00
9.48495120e-02 7.18768418e-01 -3.36508304e-01 -4.21944767e-01
9.51270401e-01 4.97451007e-01 9.98458922e-01 -2.63319969e-01
-4.47252929e-01 2.47742563e-01 3.27907234e-01 -5.53138077e-01
-2.97084451e-01 -9.94050145e-01 -1.25574720e+00 2.09032208e-01
-3.72538954e-01 -3.63166213e-01 1.00326622e+00 9.26275790e-01
6.78590387e-02 7.31823072e-02 -1.31407902e-01 -1.03736913e+00
-6.34249985e-01 -1.13163328e+00 -6.75615907e-01 4.71010089e-01
3.80239695e-01 -6.84904099e-01 -3.43448430e-01 8.95452499e-02] | [9.143054008483887, -6.460640907287598] |
faea83aa-a950-4d1d-9c44-5f8ce06bb7cd | cross-modal-global-interaction-and-local | 2305.09212 | null | https://arxiv.org/abs/2305.09212v1 | https://arxiv.org/pdf/2305.09212v1.pdf | Cross-Modal Global Interaction and Local Alignment for Audio-Visual Speech Recognition | Audio-visual speech recognition (AVSR) research has gained a great success recently by improving the noise-robustness of audio-only automatic speech recognition (ASR) with noise-invariant visual information. However, most existing AVSR approaches simply fuse the audio and visual features by concatenation, without explicit interactions to capture the deep correlations between them, which results in sub-optimal multimodal representations for downstream speech recognition task. In this paper, we propose a cross-modal global interaction and local alignment (GILA) approach for AVSR, which captures the deep audio-visual (A-V) correlations from both global and local perspectives. Specifically, we design a global interaction model to capture the A-V complementary relationship on modality level, as well as a local alignment approach to model the A-V temporal consistency on frame level. Such a holistic view of cross-modal correlations enable better multimodal representations for AVSR. Experiments on public benchmarks LRS3 and LRS2 show that our GILA outperforms the supervised learning state-of-the-art. | ['Eng Siong Chng', 'Qiushi Zhu', 'Heqing Zou', 'Chen Chen', 'Ruizhe Li', 'Yuchen Hu'] | 2023-05-16 | null | null | null | null | ['visual-speech-recognition', 'audio-visual-speech-recognition'] | ['speech', 'speech'] | [-1.49286002e-01 -4.50732559e-01 -1.97422475e-01 -3.38745117e-01
-1.29673362e+00 -4.27509248e-01 9.61644709e-01 5.21936044e-02
-2.33343407e-03 1.92868449e-02 7.26051211e-01 -1.93260953e-01
-4.21942733e-02 -2.70565689e-01 -4.81019795e-01 -9.72115159e-01
1.76617593e-01 -1.65280059e-01 9.42783505e-02 -3.13857615e-01
-1.09149642e-01 5.04568756e-01 -1.72205997e+00 6.87453628e-01
3.50097775e-01 1.28088295e+00 2.00652674e-01 7.84150422e-01
-3.43606830e-01 9.10157382e-01 -4.17929053e-01 -1.69184595e-01
-1.26000747e-01 -4.10979927e-01 -6.51169538e-01 2.37516865e-01
5.37421525e-01 9.71867368e-02 -7.48346567e-01 8.83598268e-01
7.47239470e-01 1.57001644e-01 5.26559055e-01 -1.39791119e+00
-7.52140820e-01 5.22715747e-01 -7.56743729e-01 9.50695723e-02
4.80570585e-01 3.52600038e-01 1.31909549e+00 -1.10832095e+00
4.02331799e-01 1.72234535e+00 3.17988724e-01 4.10734057e-01
-1.27466881e+00 -4.57168072e-01 5.02396226e-01 7.70225763e-01
-1.55108380e+00 -8.65689695e-01 9.48867142e-01 -3.35222453e-01
1.05136645e+00 4.48737025e-01 6.38742745e-01 1.40031600e+00
-2.40717277e-01 1.14187896e+00 9.74981189e-01 -3.23697180e-01
-4.37743291e-02 -2.61541575e-01 1.64662264e-02 2.61749208e-01
-5.90401351e-01 1.92992300e-01 -1.08037233e+00 4.28488739e-02
3.75289351e-01 -8.96961540e-02 -3.56764734e-01 -4.07624573e-01
-1.24970222e+00 4.77023005e-01 3.55607092e-01 5.27824819e-01
-1.73175097e-01 1.06731370e-01 7.33902991e-01 4.56166595e-01
3.19262892e-01 -3.42828453e-01 -2.87501305e-01 -2.18581960e-01
-9.54709768e-01 -6.56410828e-02 1.25257179e-01 7.33837962e-01
4.91084903e-01 4.07590508e-01 -5.80334961e-01 1.30388021e+00
7.87794709e-01 5.83734274e-01 4.60164785e-01 -5.76410115e-01
6.63752437e-01 3.31919938e-01 -4.16997790e-01 -8.78813088e-01
-1.90763026e-01 -4.64009017e-01 -9.92202520e-01 9.83809903e-02
8.03293288e-02 4.71704602e-01 -1.00503671e+00 1.87942529e+00
1.27152771e-01 4.05293673e-01 3.30437243e-01 9.44912076e-01
1.35777879e+00 8.19817245e-01 9.05792564e-02 -2.47127682e-01
1.56226397e+00 -1.00870192e+00 -1.01673055e+00 -1.41656756e-01
4.44966435e-01 -9.55919802e-01 1.13968563e+00 2.46577352e-01
-1.10813367e+00 -7.69071341e-01 -1.05221188e+00 -1.84461176e-01
-2.91299760e-01 3.77005994e-01 3.50352302e-02 4.44955587e-01
-1.17209399e+00 -2.66669206e-02 -7.40008831e-01 -1.85753405e-01
1.65754721e-01 6.18041214e-03 -6.54740632e-01 -2.06560567e-01
-1.22681868e+00 6.34881735e-01 5.91328777e-02 4.04516906e-01
-1.14563954e+00 -5.57838440e-01 -1.16246080e+00 4.81147580e-02
3.38306308e-01 -4.97448474e-01 1.04599273e+00 -8.24499607e-01
-1.61767089e+00 7.09726274e-01 -5.78966737e-01 -2.15428010e-01
2.42131412e-01 1.62570607e-02 -7.13962257e-01 3.13802749e-01
-3.53181005e-01 7.09363759e-01 1.00764036e+00 -1.59895527e+00
-3.42133999e-01 -3.83231491e-01 -1.65216386e-01 3.89494866e-01
-3.32615435e-01 1.84030473e-01 -6.75855458e-01 -8.09337199e-01
4.06362653e-01 -5.48197389e-01 2.29597494e-01 -1.65373996e-01
-2.96396285e-01 -2.30270714e-01 1.24551260e+00 -7.43781447e-01
1.14861691e+00 -2.48836136e+00 4.76663738e-01 5.25622293e-02
9.18131769e-02 4.49631780e-01 -6.63821518e-01 5.69484591e-01
-3.35475117e-01 -8.85521844e-02 8.39277133e-02 -9.07910168e-01
-2.24537756e-02 2.00744301e-01 -5.73888958e-01 4.07786041e-01
3.00095230e-01 9.31857646e-01 -7.42626548e-01 -5.42497754e-01
4.31912750e-01 1.00221038e+00 -1.70326471e-01 3.00562650e-01
1.32933006e-01 5.99630415e-01 -3.31725106e-02 9.77196217e-01
7.24340856e-01 7.69956335e-02 6.56895190e-02 -5.23729324e-01
2.59484351e-02 3.37130964e-01 -9.86191928e-01 1.83313477e+00
-6.16825223e-01 8.64517093e-01 3.29711407e-01 -9.86389577e-01
1.10278273e+00 6.20173335e-01 4.56031978e-01 -9.41138387e-01
-7.86719397e-02 1.61209121e-01 -2.45919287e-01 -3.41131836e-01
3.03268999e-01 -7.44120851e-02 1.79696009e-01 -9.96396020e-02
2.86726803e-01 1.51187867e-01 -2.77635932e-01 3.17450851e-01
9.55421329e-01 -1.15125440e-01 1.91241249e-01 2.88338572e-01
1.10825145e+00 -5.65001369e-01 3.85042191e-01 5.36147952e-01
-3.08248580e-01 1.05194116e+00 3.76140565e-01 -1.01561688e-01
-7.52542913e-01 -1.28664684e+00 3.59460041e-02 9.36676264e-01
2.33365640e-01 -7.69029140e-01 -4.08273578e-01 -5.26069880e-01
-2.57950157e-01 4.14467275e-01 -3.45482856e-01 -2.39811093e-01
-3.86281252e-01 -1.27964452e-01 7.95873106e-01 6.05890751e-01
4.50416028e-01 -8.60121667e-01 1.20587647e-01 -1.11743294e-01
-5.63369513e-01 -1.48197532e+00 -5.76097608e-01 -4.35397029e-02
-4.71660674e-01 -5.94674587e-01 -8.35558593e-01 -6.74177408e-01
1.32189289e-01 6.96703613e-01 8.23813319e-01 -7.13918880e-02
-2.05828071e-01 7.69028008e-01 -6.53661311e-01 1.37802884e-01
-3.47493201e-01 -3.61619800e-01 7.67030716e-02 4.73736227e-01
1.57437578e-01 -6.74073994e-01 -5.16745508e-01 4.84067678e-01
-7.77183294e-01 5.92518039e-02 6.07064366e-01 1.08359170e+00
7.51309097e-01 -3.34171146e-01 5.80035329e-01 2.37686485e-01
4.23297018e-01 -2.83897042e-01 -3.15545052e-01 5.39371252e-01
-1.47306710e-01 -8.60804915e-02 4.76867437e-01 -6.21161699e-01
-1.12437427e+00 -5.17195351e-02 -2.63198942e-01 -1.02887058e+00
-2.80608833e-01 5.94639540e-01 -7.39446223e-01 1.15921922e-01
1.61864698e-01 5.98660111e-01 -2.54366314e-03 -6.34592295e-01
5.81786871e-01 8.27125251e-01 7.52018750e-01 -4.67688233e-01
7.08441675e-01 4.82110590e-01 1.59901418e-02 -1.21309757e+00
-4.31081146e-01 -8.05080354e-01 -5.25067866e-01 -4.18436617e-01
7.44116783e-01 -1.31298161e+00 -6.83978975e-01 6.15334809e-01
-1.22554302e+00 -5.01032136e-02 -1.60852715e-01 5.57480693e-01
-5.52861452e-01 6.54789925e-01 -4.01803136e-01 -1.01220274e+00
-9.52728018e-02 -1.50758338e+00 1.59735250e+00 -1.28822640e-01
1.31954730e-01 -6.54808640e-01 1.30931467e-01 6.40926242e-01
2.73679882e-01 -2.91928709e-01 6.46144509e-01 -5.44306159e-01
-5.99021316e-01 2.17416435e-02 -3.84865165e-01 4.75370079e-01
8.41910020e-03 8.81069079e-02 -1.33239114e+00 -2.89742261e-01
-2.34909341e-01 -2.56012172e-01 1.23662484e+00 3.24365586e-01
9.98506308e-01 -1.91811681e-01 -9.14512500e-02 5.30402660e-01
9.07159865e-01 1.02680810e-01 8.51669252e-01 8.86656195e-02
9.40729916e-01 6.24083877e-01 8.02005708e-01 3.42364550e-01
4.08952177e-01 1.21988928e+00 5.31894565e-01 -1.58638433e-01
-6.20587409e-01 -3.17734063e-01 9.21492100e-01 1.22096944e+00
1.42254889e-01 -2.34989628e-01 -8.16403806e-01 6.09446466e-01
-2.06616640e+00 -1.06370735e+00 2.83720400e-02 2.05725574e+00
6.57800615e-01 -3.73775899e-01 1.03244200e-01 3.87907535e-01
7.03475296e-01 6.84353173e-01 -8.39359015e-02 -2.73822904e-01
-6.00266099e-01 -1.19910546e-01 -7.79749677e-02 5.30636132e-01
-1.04895937e+00 8.24554443e-01 5.47147465e+00 1.47337425e+00
-1.21266437e+00 2.26155207e-01 4.28038508e-01 -2.34086633e-01
-5.26331127e-01 -1.41431037e-02 -5.99442959e-01 1.69760853e-01
8.61413777e-01 3.44641656e-01 3.93861473e-01 5.39959848e-01
3.68322343e-01 3.70116979e-01 -9.56766963e-01 1.50554037e+00
7.13793486e-02 -1.17374647e+00 3.27827036e-01 1.34694293e-01
3.32345754e-01 -4.81312573e-02 3.43622744e-01 1.60358608e-01
-1.58443451e-01 -1.16654611e+00 1.03446710e+00 6.01798534e-01
7.17256069e-01 -8.00310612e-01 5.55359125e-01 8.39248151e-02
-1.95135474e+00 -7.65692368e-02 6.46117106e-02 5.25227666e-01
3.40000719e-01 2.71601975e-01 -3.30496222e-01 1.00816798e+00
9.11078036e-01 9.28508341e-01 -4.83765364e-01 7.14873850e-01
-1.26947433e-01 4.28227156e-01 2.96861269e-02 4.83164728e-01
2.85931826e-01 -1.42049938e-01 9.27367866e-01 1.42206478e+00
2.28161797e-01 -1.76654488e-01 4.36646082e-02 6.32980347e-01
1.90459877e-01 2.34149575e-01 -6.26615107e-01 -3.47771615e-01
5.13353825e-01 1.16010702e+00 -2.65854567e-01 -1.35923713e-01
-6.37758791e-01 6.21167600e-01 5.92119992e-02 4.51139420e-01
-7.26952910e-01 -3.68062109e-02 1.09562182e+00 -2.65784889e-01
4.61449742e-01 -5.00381827e-01 -1.17436484e-01 -1.24328125e+00
2.22666070e-01 -1.22410882e+00 3.76147211e-01 -9.19908404e-01
-1.24275815e+00 6.42998159e-01 -1.25782877e-01 -1.60309184e+00
-1.55423194e-01 -4.87653106e-01 -5.09179413e-01 7.92831719e-01
-1.52751327e+00 -1.73330140e+00 -1.86269060e-01 8.74608457e-01
6.89161003e-01 -4.65496004e-01 6.03555083e-01 4.64544415e-01
-4.67424631e-01 9.13141489e-01 -1.74276933e-01 -2.67498586e-02
6.41519606e-01 -8.31733346e-01 2.24656984e-01 9.33600664e-01
6.81811452e-01 4.21337336e-01 5.51225722e-01 -3.48134875e-01
-1.61400580e+00 -8.90450180e-01 6.75385833e-01 -4.25736941e-02
7.42179990e-01 -3.06501687e-01 -1.07584321e+00 3.23560148e-01
2.98702657e-01 2.65719146e-01 5.04613996e-01 5.63193299e-02
-1.04618764e+00 -3.95855159e-01 -5.93213797e-01 6.44549251e-01
1.02850199e+00 -1.38632643e+00 -3.71322513e-01 -8.21237192e-02
9.24975097e-01 -1.53617710e-01 -8.00366104e-01 5.61489761e-01
5.86815298e-01 -9.25506175e-01 1.44835341e+00 -3.44044924e-01
2.24286199e-01 -6.73813343e-01 -7.44757116e-01 -1.14768970e+00
1.04286075e-01 -6.22510076e-01 -2.16570660e-01 1.71048295e+00
9.88129154e-02 -3.37542653e-01 -2.99452506e-02 -3.39284927e-01
-3.29175264e-01 -5.99820197e-01 -1.43147945e+00 -9.09389019e-01
-3.60437155e-01 -7.78855622e-01 4.61430162e-01 8.53168249e-01
6.70491979e-02 2.66691625e-01 -6.94336414e-01 4.67768699e-01
3.55401158e-01 -4.59254012e-02 6.88792467e-01 -8.77262175e-01
-2.89186865e-01 -6.55118883e-01 -8.48500907e-01 -1.16729224e+00
3.24051827e-01 -7.94462383e-01 3.34343649e-02 -1.41613889e+00
1.96754485e-01 1.89171806e-01 -4.33077037e-01 4.84922945e-01
5.90668730e-02 1.45845801e-01 3.09588522e-01 2.83680528e-01
-7.43617773e-01 1.12248695e+00 1.00465000e+00 -4.12389606e-01
-1.96179047e-01 -3.44700336e-01 -3.19297165e-01 4.51729745e-01
3.29156399e-01 -4.83962148e-02 -5.08205414e-01 -3.70961428e-01
-1.65080994e-01 3.69884908e-01 6.95240736e-01 -7.63719678e-01
3.98701370e-01 -5.16248867e-02 5.17358854e-02 -9.03583050e-01
9.08995152e-01 -6.87949300e-01 1.08419366e-01 -7.99044147e-02
-4.61690873e-01 -1.15693621e-01 1.43293485e-01 8.59974623e-01
-9.90979671e-01 3.00603300e-01 7.77758300e-01 2.72600174e-01
-6.49133384e-01 2.44638115e-01 -2.82472461e-01 -2.30030850e-01
6.05703950e-01 -5.32215238e-02 -5.12088656e-01 -7.57241964e-01
-8.64728749e-01 8.18431973e-02 9.98590738e-02 8.31009328e-01
1.05550683e+00 -1.68767190e+00 -6.36264861e-01 2.13651076e-01
5.16954541e-01 -4.35010940e-01 8.23224604e-01 1.21171927e+00
1.43485337e-01 4.31868345e-01 8.02039802e-02 -1.02987337e+00
-1.84157491e+00 4.64848250e-01 3.47606093e-01 -2.26951614e-02
-7.58496702e-01 7.98774660e-01 5.03340662e-01 -2.61365771e-01
7.15219438e-01 -4.16722558e-02 -4.34938103e-01 3.09793890e-01
6.66128814e-01 1.90896273e-01 1.56329602e-01 -1.40717256e+00
-5.83173513e-01 7.44362175e-01 8.03661272e-02 -5.38073182e-01
1.06249464e+00 -4.78146464e-01 4.38458882e-02 7.92481601e-01
1.38933218e+00 6.95675164e-02 -9.40907776e-01 -6.39492035e-01
-2.38436624e-01 -5.29219151e-01 2.95809090e-01 -6.18949890e-01
-1.32323658e+00 1.46337223e+00 6.79974258e-01 1.02074519e-01
1.30257916e+00 2.95527667e-01 6.17588997e-01 1.10233568e-01
-1.98490173e-02 -7.92999566e-01 4.55456138e-01 4.77389306e-01
1.30722618e+00 -1.06874132e+00 -2.62381762e-01 -3.21821839e-01
-9.33978796e-01 1.18000925e+00 2.98977971e-01 4.93684947e-01
5.75213492e-01 8.28065500e-02 2.68339962e-01 5.17238937e-02
-1.04910600e+00 -6.38466179e-01 7.58078158e-01 6.86457634e-01
3.14758092e-01 -1.31585421e-02 1.84053257e-01 6.50674045e-01
1.44555554e-01 -6.84755981e-01 -8.18719864e-02 7.01226950e-01
-2.20409751e-01 -1.27265501e+00 -5.37519991e-01 -3.13990861e-01
-1.56817704e-01 -1.80346176e-01 -6.09043956e-01 5.75352848e-01
-2.42673293e-01 1.28736556e+00 -2.32121795e-02 -7.77913392e-01
2.98154324e-01 6.95887506e-02 3.40174526e-01 -8.38823169e-02
-4.29548025e-01 7.43195176e-01 4.17246744e-02 -6.88580036e-01
-4.06985253e-01 -6.80087328e-01 -1.06106400e+00 -1.61352262e-01
-8.35802928e-02 -1.59191683e-01 7.67826736e-01 9.94789541e-01
4.35937911e-01 6.01659000e-01 8.02779317e-01 -9.50205803e-01
-4.33357209e-01 -8.64670634e-01 -5.02441287e-01 3.91440660e-01
7.34530628e-01 -8.04336965e-01 -6.92085207e-01 -2.21977517e-01] | [14.217751502990723, 5.126428127288818] |
c3d964fd-6f70-4022-ab5d-39d0a1a0e294 | what-can-i-cook-with-these-ingredients | 2112.04788 | null | https://arxiv.org/abs/2112.04788v2 | https://arxiv.org/pdf/2112.04788v2.pdf | "What can I cook with these ingredients?" -- Understanding cooking-related information needs in conversational search | As conversational search becomes more pervasive, it becomes increasingly important to understand the user's underlying information needs when they converse with such systems in diverse domains. We conduct an in-situ study to understand information needs arising in a home cooking context as well as how they are verbally communicated to an assistant. A human experimenter plays this role in our study. Based on the transcriptions of utterances, we derive a detailed hierarchical taxonomy of diverse information needs occurring in this context, which require different levels of assistance to be solved. The taxonomy shows that needs can be communicated through different linguistic means and require different amounts of context to be understood. In a second contribution we perform classification experiments to determine the feasibility of predicting the type of information need a user has during a dialogue using the turn provided. For this multi-label classification problem, we achieve average F1 measures of 40% using BERT-based models. We demonstrate with examples, which types of need are difficult to predict and show why, concluding that models need to include more context information in order to improve both information need classification and assistance to make such systems usable. | ['Bernd Ludwig', 'David Elsweiler', 'Alexander Frummet'] | 2021-12-09 | null | null | null | null | ['conversational-search'] | ['natural-language-processing'] | [ 3.85770530e-01 2.18287647e-01 1.24242812e-04 -1.02401447e+00
-7.59683967e-01 -8.69688988e-01 4.02203739e-01 4.91651624e-01
-2.66371727e-01 5.93205571e-01 9.17704105e-01 -5.00728846e-01
-1.80816814e-01 -1.43227518e-01 4.71587898e-03 -2.62315989e-01
3.79191458e-01 6.92808807e-01 -1.55933574e-01 -5.89398861e-01
3.82297128e-01 5.42523302e-02 -1.67344904e+00 9.48016942e-01
6.76805556e-01 6.07230902e-01 7.57314503e-01 8.49319518e-01
-4.40665901e-01 1.20785904e+00 -9.01523948e-01 1.38803408e-01
-1.58847108e-01 -5.41696668e-01 -1.61450064e+00 3.80850136e-01
1.98896125e-01 -4.65529919e-01 2.57812172e-01 4.17460352e-01
5.73946893e-01 3.96348238e-01 7.54507840e-01 -1.20012903e+00
-2.19562486e-01 6.02842867e-01 3.35482508e-01 2.27088317e-01
1.36255598e+00 -5.81182130e-02 1.10629988e+00 -2.26293936e-01
5.22494674e-01 1.19990849e+00 5.46310961e-01 7.41804481e-01
-1.13879478e+00 -2.18568027e-01 1.81773826e-01 3.15526366e-01
-7.51586854e-01 -7.72748351e-01 7.32984662e-01 -5.75349391e-01
1.11362338e+00 8.36883128e-01 2.22677439e-01 1.02477396e+00
-3.30140859e-01 8.39249194e-01 1.08759665e+00 -8.69803131e-01
1.71948627e-01 7.18892872e-01 6.25070751e-01 4.36867476e-01
-4.35520053e-01 -5.87211728e-01 -5.55260301e-01 -4.15709317e-01
-9.39175300e-03 -2.03509897e-01 -4.05413866e-01 2.36438766e-01
-1.01608860e+00 7.89386570e-01 7.00617284e-02 8.69177699e-01
-3.27059150e-01 -4.67226386e-01 5.62959552e-01 6.83298767e-01
3.66032690e-01 8.45133662e-01 -7.72272110e-01 -5.75991511e-01
-3.38710487e-01 1.97139814e-01 1.69731379e+00 1.24951327e+00
5.92931509e-01 -7.58611441e-01 9.09251720e-03 1.24479961e+00
2.10645273e-01 -2.66234819e-02 4.29925144e-01 -1.19774020e+00
5.51437378e-01 6.87175274e-01 5.38033068e-01 -1.04336023e+00
-7.10047305e-01 2.76892513e-01 -1.79412290e-01 -3.01005125e-01
8.19380760e-01 -3.78646016e-01 -1.20271809e-01 1.49288034e+00
2.68679589e-01 -6.42636538e-01 1.95373058e-01 7.94194341e-01
9.18717921e-01 4.80900526e-01 3.12517703e-01 -5.65633714e-01
1.69165802e+00 -7.24694908e-01 -1.03877091e+00 -3.66673738e-01
1.31441402e+00 -9.96076643e-01 1.42563570e+00 2.53490746e-01
-7.48521805e-01 -3.50396067e-01 -7.02920437e-01 -2.68096477e-01
-1.96854666e-01 3.05646425e-03 5.44657290e-01 6.80647790e-01
-9.81336296e-01 2.61572927e-01 -2.82433450e-01 -9.12429512e-01
-4.70267266e-01 2.74426728e-01 -2.19349414e-01 -4.45735753e-02
-1.29589963e+00 1.14957368e+00 -3.73876840e-02 -1.03227913e-01
-1.47180974e-01 -2.77044863e-01 -9.05199111e-01 7.87036680e-03
3.83805186e-01 -2.55316794e-01 1.88898170e+00 -8.00574243e-01
-1.33425999e+00 8.00764024e-01 -5.35812497e-01 3.21055532e-01
7.49224350e-02 1.28956661e-01 -2.81109989e-01 1.24012157e-01
3.76346678e-01 2.03530073e-01 1.78848118e-01 -1.34421861e+00
-1.13826895e+00 -2.83497721e-01 6.22711837e-01 5.65226197e-01
-3.15187484e-01 4.95018512e-01 1.64733678e-01 1.20314490e-02
-2.01657135e-02 -9.20571506e-01 4.22955453e-02 -4.55273211e-01
-1.04698174e-01 -6.58775926e-01 7.35499203e-01 -8.79640877e-01
1.39876115e+00 -1.87094486e+00 1.18254267e-01 -1.35339841e-01
2.01985240e-01 -8.18298683e-02 1.14791334e-01 8.83358896e-01
2.32546031e-01 3.25075328e-01 -5.98285673e-03 -2.11824089e-01
2.17848569e-01 2.81946719e-01 2.24355951e-01 2.59593129e-02
-1.98823676e-01 3.47264856e-01 -1.03163242e+00 -5.29115021e-01
2.07469583e-01 1.66094080e-01 -4.79691684e-01 6.71514213e-01
-2.14576527e-01 4.31421250e-01 -4.67601329e-01 3.93499523e-01
-1.18194371e-01 -2.46512443e-01 4.93080527e-01 -1.71783894e-01
-1.76855996e-01 8.41369212e-01 -8.07295978e-01 1.39147949e+00
-1.09097278e+00 6.54200137e-01 6.76275253e-01 -1.01286352e+00
6.84427917e-01 7.55874813e-01 3.88936967e-01 -4.40091610e-01
1.94648057e-01 6.91391304e-02 1.32752404e-01 -1.19829297e+00
4.33889419e-01 -1.39495850e-01 -4.18602914e-01 1.03423905e+00
-2.14121774e-01 -2.21276671e-01 2.56983399e-01 8.23842883e-02
1.18178976e+00 -5.06423056e-01 4.12633568e-01 -5.20885408e-01
6.37549043e-01 2.54142344e-01 1.60466418e-01 6.06307149e-01
-4.44499463e-01 1.03133015e-01 3.24029684e-01 -5.05474627e-01
-5.74903846e-01 8.11031312e-02 1.25161290e-01 1.70470369e+00
1.86428316e-02 -4.03280467e-01 -8.70852649e-01 -7.58163929e-01
-3.80472273e-01 1.06627595e+00 -1.28662825e-01 2.13174805e-01
-5.61644554e-01 -4.99602497e-01 3.40642750e-01 1.41901821e-01
2.91950643e-01 -1.16915500e+00 -9.56219435e-01 4.33273941e-01
-1.16958380e+00 -1.26135719e+00 -5.28163850e-01 4.77810204e-01
-3.37163538e-01 -1.17838454e+00 -1.67815447e-01 -1.00921512e+00
4.98911083e-01 4.20468867e-01 1.17668045e+00 4.62978274e-01
-1.79582775e-01 9.39773679e-01 -9.14955199e-01 -2.53531754e-01
-8.66214752e-01 2.49971434e-01 -5.92431650e-02 -2.40144417e-01
7.38388121e-01 -5.02480924e-01 -2.72958189e-01 8.43640804e-01
-5.45436919e-01 1.07864425e-01 8.15179646e-02 7.25315094e-01
-4.27658975e-01 2.07152367e-01 6.06194854e-01 -8.92109394e-01
1.43030369e+00 -7.21376300e-01 3.46603990e-01 4.98875022e-01
-4.01359648e-01 2.12957077e-02 5.34407616e-01 -3.91664058e-01
-1.18950379e+00 -2.30103228e-02 -3.30789477e-01 6.56145453e-01
-8.71997654e-01 4.76204365e-01 -1.85811952e-01 1.85021371e-01
8.19769919e-01 -8.54919925e-02 -4.59494814e-02 -4.41783458e-01
5.96778765e-02 1.54008031e+00 -2.55261883e-02 -9.70099151e-01
-7.24928677e-02 -2.59561241e-01 -6.95791721e-01 -1.17139304e+00
-8.61176491e-01 -1.04775012e+00 -6.57473385e-01 -5.35378397e-01
8.91047478e-01 -5.39550662e-01 -1.04670179e+00 4.58634570e-02
-1.29250288e+00 -7.36453474e-01 2.69522548e-01 2.16042832e-01
-6.88963115e-01 2.68165082e-01 -6.79745495e-01 -1.31261408e+00
-3.95390205e-02 -1.16566789e+00 8.45400453e-01 -3.32698859e-02
-1.25912404e+00 -1.08609569e+00 -4.49629813e-01 1.13742173e+00
3.88332039e-01 -2.20552430e-01 1.17245221e+00 -1.01674592e+00
2.32091755e-01 3.58115025e-02 -3.89174186e-03 1.35344235e-04
7.93222129e-01 -5.80030203e-01 -9.87488806e-01 -8.32362920e-02
5.07210970e-01 -6.10862434e-01 -1.62856370e-01 -8.82012919e-02
6.98930979e-01 -8.07563663e-01 -3.24778080e-01 -4.11578417e-01
8.81742895e-01 7.07911909e-01 1.07095093e-01 6.51676729e-02
3.73188317e-01 1.51053691e+00 6.14191175e-01 1.72424644e-01
9.09026682e-01 8.23479593e-01 -1.01735666e-01 2.70404547e-01
1.65094450e-01 4.28430707e-04 2.45936632e-01 9.71098602e-01
1.78784102e-01 -3.24097693e-01 -1.03778076e+00 5.04681349e-01
-1.85487986e+00 -9.45607364e-01 -1.58322290e-01 1.64668024e+00
1.09268129e+00 -2.44304568e-01 3.68632287e-01 4.42619711e-01
6.66782618e-01 -4.22289558e-02 -2.22803265e-01 -7.32544422e-01
6.76452637e-01 -4.78157729e-01 -8.86739343e-02 1.05770326e+00
-7.28934526e-01 5.84904075e-01 6.80648947e+00 6.31385669e-02
-7.63909578e-01 1.00074753e-01 8.05875540e-01 2.35730037e-01
-2.96678334e-01 -2.62122929e-01 -6.71728492e-01 4.96062040e-01
1.07794213e+00 4.02480848e-02 7.30028093e-01 6.46164000e-01
3.91909003e-01 -4.66750950e-01 -1.58865476e+00 7.85530686e-01
1.49534270e-01 -6.11413121e-01 -5.27157664e-01 -1.92889839e-01
-4.93276790e-02 -5.80899656e-01 -7.01889098e-01 4.18789953e-01
4.27459508e-01 -6.61724806e-01 2.93724924e-01 1.81454241e-01
3.89687210e-01 -3.22141200e-01 6.46079421e-01 1.13809216e+00
-9.88149762e-01 -3.70470464e-01 8.06674957e-02 -7.37790108e-01
3.34106684e-01 9.35017392e-02 -1.21772313e+00 1.37807712e-01
5.42661190e-01 1.81508526e-01 -1.15926854e-01 3.70956481e-01
3.59169468e-02 3.15362573e-01 -3.45938861e-01 -6.68689966e-01
1.19427098e-02 -2.73019131e-02 1.58910677e-01 1.40937746e+00
1.06790505e-01 7.91703641e-01 5.00726700e-01 4.22004461e-01
2.89618522e-01 3.58752668e-01 -7.22253501e-01 -5.10351770e-02
5.71543157e-01 1.18181551e+00 -6.63963974e-01 -2.00415716e-01
-4.30468440e-01 1.02696943e+00 1.50030196e-01 2.89143145e-01
-1.84233353e-01 -3.05745006e-01 6.24508321e-01 9.01775956e-02
-3.74856114e-01 -8.85110870e-02 -8.78912956e-02 -9.29011047e-01
5.72095439e-02 -1.24714887e+00 2.17270374e-01 -7.66866386e-01
-1.20307660e+00 5.42542100e-01 1.44653097e-01 -8.21543396e-01
-7.99141645e-01 -3.09201986e-01 -3.59920621e-01 9.10428166e-01
-1.06396282e+00 -8.23456109e-01 -4.21114326e-01 2.93242782e-01
1.06942499e+00 1.48120686e-01 1.32889843e+00 4.67092127e-01
-1.37308270e-01 1.75171673e-01 -5.41708231e-01 -1.18849576e-01
7.55425692e-01 -1.20546317e+00 -1.92745868e-02 4.58233692e-02
-1.68015555e-01 6.23565137e-01 9.96229887e-01 -4.48566258e-01
-1.29161799e+00 -3.06085348e-01 1.61336160e+00 -7.21618354e-01
4.37228978e-01 -3.78872186e-01 -7.03900397e-01 5.87334096e-01
3.22415650e-01 -7.84733653e-01 1.30398858e+00 6.20283306e-01
-1.08670704e-01 2.84918338e-01 -1.50095212e+00 2.78185785e-01
1.14253402e+00 -9.22681987e-01 -9.95724499e-01 6.49667084e-01
7.35460758e-01 -2.14376435e-01 -7.33660519e-01 -4.42303084e-02
5.73375225e-01 -9.51928079e-01 4.47610885e-01 -6.54377043e-01
3.10060471e-01 7.46494457e-02 -3.98202360e-01 -1.40522015e+00
-2.02989310e-01 -6.07126296e-01 3.72044504e-01 1.24359250e+00
6.25997365e-01 -4.09948647e-01 4.21538472e-01 1.58702266e+00
-8.99923891e-02 -5.15704870e-01 -5.60283661e-01 -1.32972062e-01
-3.86019886e-01 -4.56801057e-01 4.53764975e-01 1.29458439e+00
1.10588455e+00 8.55634868e-01 -2.73247540e-01 -5.40088415e-02
2.08128039e-02 -1.04903989e-01 6.09088123e-01 -1.44874930e+00
-3.18589667e-03 -2.79372722e-01 1.00897819e-01 -1.38049591e+00
2.60864079e-01 -7.23417819e-01 4.21481818e-01 -1.75091195e+00
-1.88985065e-01 -6.92121148e-01 2.82499760e-01 4.94732618e-01
-2.29720473e-02 -4.37333584e-01 2.82037973e-01 2.13770896e-01
-4.97008622e-01 -2.69959718e-02 7.57036805e-01 -1.94514185e-01
-5.48092842e-01 2.72233993e-01 -1.13816202e+00 7.22322285e-01
8.02541554e-01 -1.82533473e-01 -6.48645163e-01 -4.50458944e-01
2.63931006e-01 5.73206067e-01 -2.62449741e-01 -7.65411675e-01
4.14433330e-01 -4.40367013e-01 -1.05648197e-01 -9.62934569e-02
3.40724885e-01 -1.37146854e+00 -4.95321266e-02 1.40854985e-01
-9.71035957e-01 7.73786334e-03 7.64501989e-02 1.30164251e-01
-5.95302507e-02 -6.89371765e-01 1.37563422e-01 -3.75062704e-01
-5.08993566e-01 -4.70193774e-01 -1.04238129e+00 -1.64499536e-01
9.20411468e-01 4.00057584e-02 -2.98206732e-02 -1.04550779e+00
-1.05033088e+00 5.29931009e-01 2.42382243e-01 4.87398744e-01
2.97052920e-01 -9.71847594e-01 -3.63570005e-01 7.46464431e-02
2.33155549e-01 -2.48919651e-01 -4.97211441e-02 5.06801844e-01
-2.07404062e-01 5.83292544e-01 -2.96865497e-02 -4.38591719e-01
-1.76819575e+00 2.25522518e-01 3.06933075e-01 -3.76053974e-02
-2.40880311e-01 6.08888566e-01 -1.12296537e-01 -7.23077774e-01
5.79318881e-01 -5.22483647e-01 -6.69574261e-01 6.01886809e-01
7.50627160e-01 2.93966323e-01 6.65981099e-02 -5.90741873e-01
-1.26680389e-01 1.85640365e-01 -7.10361227e-02 -1.96847558e-01
1.04493177e+00 -6.06249869e-01 -1.04751088e-01 8.79334331e-01
1.19373739e+00 -2.58467555e-01 -5.79774499e-01 -1.58897266e-01
4.14511800e-01 -3.92218322e-01 -2.73346394e-01 -1.08827567e+00
-2.80337214e-01 5.72862387e-01 2.02537194e-01 1.13653016e+00
9.63608325e-01 1.86232969e-01 7.73329437e-01 8.90111685e-01
6.19621813e-01 -1.30978060e+00 -1.52627425e-02 5.99704266e-01
1.09318793e+00 -1.49328697e+00 -2.59728819e-01 -6.22505486e-01
-7.49695599e-01 1.06939673e+00 5.02107084e-01 6.93173826e-01
5.82757950e-01 3.14218760e-01 5.11126041e-01 -3.36884856e-01
-9.61004019e-01 -1.04605891e-01 -1.47931635e-01 6.10564113e-01
9.07241344e-01 1.37720734e-01 -4.77211654e-01 4.15828139e-01
-3.52310389e-01 -2.26260304e-01 5.82897961e-01 1.18388212e+00
-5.10035098e-01 -1.24953139e+00 -2.53032476e-01 6.86461449e-01
-2.63691455e-01 1.48662180e-01 -7.72346914e-01 3.72721344e-01
-1.93564668e-01 1.94583857e+00 -3.74322385e-01 -5.84808469e-01
5.79254448e-01 5.89618087e-01 5.84723204e-02 -1.10543776e+00
-9.59362149e-01 -2.33401582e-01 9.78943527e-01 -3.64758581e-01
-7.86052823e-01 -7.18518198e-01 -1.02386045e+00 -1.18690915e-01
-5.05726755e-01 6.42854929e-01 8.56255233e-01 1.31408715e+00
1.04596727e-01 2.76601315e-01 6.47389114e-01 -7.77588844e-01
-4.72336441e-01 -1.30756354e+00 -1.04350068e-01 6.13469720e-01
4.73484337e-01 -3.29420209e-01 -7.77295053e-01 2.83307314e-01] | [12.54422664642334, 7.851447582244873] |
2e177aba-a050-4181-93a1-2783f4fcb9ab | a-blackbox-approach-to-best-of-both-worlds-in | 2302.09739 | null | https://arxiv.org/abs/2302.09739v1 | https://arxiv.org/pdf/2302.09739v1.pdf | A Blackbox Approach to Best of Both Worlds in Bandits and Beyond | Best-of-both-worlds algorithms for online learning which achieve near-optimal regret in both the adversarial and the stochastic regimes have received growing attention recently. Existing techniques often require careful adaptation to every new problem setup, including specialised potentials and careful tuning of algorithm parameters. Yet, in domains such as linear bandits, it is still unknown if there exists an algorithm that can simultaneously obtain $O(\log(T))$ regret in the stochastic regime and $\tilde{O}(\sqrt{T})$ regret in the adversarial regime. In this work, we resolve this question positively and present a general reduction from best of both worlds to a wide family of follow-the-regularized-leader (FTRL) and online-mirror-descent (OMD) algorithms. We showcase the capability of this reduction by transforming existing algorithms that are only known to achieve worst-case guarantees into new algorithms with best-of-both-worlds guarantees in contextual bandits, graph bandits and tabular Markov decision processes. | ['Julian Zimmert', 'Chen-Yu Wei', 'Christoph Dann'] | 2023-02-20 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [ 0.27764204 0.5518552 -0.3580169 -0.47301787 -1.3344393 -1.0476863
0.17842028 0.31583345 -0.47964102 1.1115861 -0.24282932 -0.74908704
-0.8397168 -0.921372 -1.218093 -0.7928211 -0.4472296 1.0497701
0.2628674 -0.25427088 -0.0578295 0.29509252 -0.9403496 0.02373636
0.8776661 1.230896 -0.29719234 0.8771117 -0.27170599 0.685992
-0.12909606 -0.9328133 0.88036966 -0.6197062 -1.0354904 0.08108162
0.45435265 -0.41791627 -0.50044 1.1525198 0.3808882 0.17866983
0.21027991 -1.1658475 -0.4408553 1.1046348 -0.77332604 0.27594513
0.35550797 0.02494201 1.3277602 0.28577936 0.6506951 1.2721512
0.39580157 0.51147074 -1.6125379 -0.5140594 0.65911645 0.06161914
-0.68347406 -0.29048792 0.3762035 -0.19088212 0.7377307 0.6929101
0.56143785 0.75474894 -0.26492098 1.0829393 1.3648772 -0.63302463
0.37987846 0.06536478 0.11315431 0.65578234 0.45449278 0.1146138
-0.57099116 -0.4146693 0.6657983 0.03519852 -0.20064218 -0.58351797
-0.8690742 0.7811072 0.24981403 -0.24243517 -0.24509373 0.6950569
0.23385112 1.0131764 0.67968243 0.16365157 -0.44906518 -0.1516364
-0.9879276 0.2874016 1.3752998 1.1246486 0.81060946 -0.11846891
-0.13789484 0.38548988 -0.1255702 0.70030254 -0.06364542 -0.82941896
0.9449487 0.08662413 0.7013493 -0.59717256 -0.15712157 -0.6268021
-0.5526602 0.17611349 0.82591766 -0.3833921 -0.6345913 1.8000522
0.69824165 -0.30150425 -0.02095365 0.988128 -0.19772542 0.47179854
-0.45935655 -0.64922416 0.81134033 -0.9078496 -0.46490407 -0.51045066
0.6803238 -0.47046468 0.74665844 0.4919828 -1.3733345 0.27807644
-0.85285485 0.25702012 0.12727487 -0.83392197 0.9289649 1.1913927
-0.9948574 0.8875514 -0.9144065 -0.20810011 0.45487574 0.51509273
-0.30172932 -0.14742672 -0.8650332 0.60549 0.23088367 -0.04776225
-1.1815755 -0.4451892 -0.23144616 -0.04472009 1.2871399 -0.791037
1.5001397 -1.3600314 -1.6206523 0.8638764 -0.13256806 -0.97703296
1.3092484 -0.1256646 0.26993567 -0.02324456 -0.20662692 -0.2896811
0.7343325 -0.9365795 -0.9167179 -0.8290935 0.8206314 0.33779812
-0.21845967 -0.20274714 -0.11903953 -0.13157497 0.2525831 -1.1367645
-0.50355077 0.02490634 -0.49123424 -0.07773437 0.14586446 -0.37327978
0.8192349 -1.7338686 0.13269787 0.23953082 -0.10695023 -0.10443419
0.10501459 0.8520477 0.2646539 0.27162632 0.12569304 -0.3562844
0.31619135 0.40417677 -0.42071941 0.83498013 -0.53819084 0.5956706
-0.9855232 -0.26676533 -0.16334203 -0.49290726 -0.539669 0.04481984
-0.5283444 0.16441119 -0.75569254 0.34477943 0.6020039 -0.33054882
0.41888785 0.63288856 0.08895501 0.1241744 -1.4405164 1.4927199
-0.55792737 0.23985876 0.67568517 -1.1505803 0.32503837 0.1098313
0.27524236 -0.54800713 -0.07719632 0.33977434 -0.16396491 -0.4361212
0.07623366 -0.62155455 0.14804766 0.8103413 -0.24936435 0.09138747
0.12078667 0.06819057 1.2851924 0.0218729 0.04451782 -0.2911492
0.10473093 -0.02977143 0.43121016 1.4404954 -0.2153098 0.19106771
1.0038198 -0.51845086 -0.9175092 -0.9480853 0.48971006 1.6314594
0.22314848 0.1978866 -0.54983795 -0.820436 0.4230106 0.99798876
-0.8155899 0.09375273 -0.1794975 -0.71186614 0.29323375 0.05205738
0.14116889 -0.3967312 -0.42366832 0.40305412 0.08867276 -0.83702004
-0.44177437 0.49140236 -0.9839602 -0.926708 -0.3371781 -0.15004562
0.49880317 0.23855665 0.7603018 -0.2901733 0.14655913 0.33031383
-0.3211905 -0.6068269 -0.38403764 0.02604779 -0.0843925 0.23917954
-0.09818651 -0.64254874 -0.6413021 0.21546692 -0.823601 -0.01506793
0.37171885 0.6811855 0.6335566 -0.01899318 0.24750975 -1.2594274
0.47584137 -0.44482705 -1.1746734 0.5028248 -0.79604954 0.37232375
0.82782674 -0.32179418 -0.88246983 -0.06986316 0.21887645 -0.29059651
0.14256452 0.47012368 -0.01438764 -0.25722906 0.67608064 0.472067
-0.01606057 -0.28141207 0.5457389 0.4840232 0.12019417 -0.907155
0.9233474 0.9496568 0.36426708 -0.39545166 -1.410945 -0.365109
-0.1925917 -0.17818539 0.24093868 -0.47473603 -1.1208167 0.16103232
-0.67144215 -0.5333887 -0.5910273 0.3060717 -1.0760419 0.3050449
-0.26009205 -1.6371274 -0.39233375 -0.7695872 0.63259465 0.11626038
0.4183931 -0.97754365 -0.03972518 0.68226504 0.35907865 0.41749692
0.6604108 -0.9910287 -0.9465144 -0.50672174 -0.06486257 0.21397525
-0.27492997 -0.57636595 -0.7541665 -0.6706973 0.0412062 -0.5106787
0.96891105 0.36792073 0.9807775 -1.0247109 -0.14980945 0.7661267
1.6095968 0.06438772 0.3103758 0.60204047 0.25755292 0.44691646
0.50944144 0.6308086 0.16817923 0.4067579 0.886558 0.2518722
0.5265974 -0.26570246 0.28266358 0.0929842 -0.14051326 -0.38385484
-0.71215904 0.71812594 -2.1369016 -1.1386225 0.13327204 2.6748016
1.0415905 0.47136414 0.26749018 0.03363182 0.78644127 0.2990804
-0.8827926 -0.71637577 0.08775515 0.01163438 1.1968336 0.6192372
-0.88984346 0.6758366 5.4930577 1.1088074 -0.68942827 0.3954017
0.8414119 -0.65925986 -0.37108102 0.459872 -0.70714724 0.12650439
0.97828335 -0.44670358 1.1363865 1.1718903 0.00821061 -0.10525926
-1.3802812 0.8460791 -0.19123304 -1.2105731 -0.32736015 0.29385373
0.921099 0.00593601 -0.0490105 0.11618023 0.92090297 -0.7987946
0.9005723 0.28170916 0.6811394 -0.93147737 0.46844858 0.81878465
-0.6497199 -0.32093307 -0.37451026 -0.16585834 0.07900988 0.68599135
-0.6835648 0.88019454 0.48346746 0.02814245 0.19960266 1.0715898
0.12313786 0.6462188 -0.8773379 -0.3086483 0.6712005 -0.47283384
0.74507946 0.87275136 0.13144252 0.08067592 0.49421582 0.33551562
-0.18882206 0.20029749 -0.26645643 -0.15279528 0.24090417 0.85928446
-0.6367642 -0.01950129 -0.04946015 0.96855885 0.48896652 0.19460939
-0.747727 -0.16552107 0.62109524 0.2968769 0.64238685 -0.08057129
-0.07445747 -1.0172648 0.3010071 -0.7038067 0.7966546 -0.1308296
-1.5885373 0.02075916 -0.1071603 -0.6415572 -0.06656986 -0.42764878
-0.21538742 0.55563307 -1.5849847 -1.0974134 0.2777997 0.68888205
0.26531303 0.05844832 0.56481606 -0.13088895 -0.28691757 0.7137048
1.0382422 -0.34609026 0.5267103 -1.5612699 0.04801617 0.80694574
0.06202995 0.22023317 1.1100922 -0.36908928 -2.047091 -0.9042787
0.39486775 -0.14595392 1.2474303 -0.3692152 -0.21870412 0.9524234
-0.16685353 0.16600464 0.4587548 0.42893013 -0.48822346 -0.6845011
-1.2306033 0.5316784 1.3416282 -0.44416347 -0.046007 0.91532123
0.6576509 -0.70312417 -0.69519615 -0.31024063 0.6561553 -1.1650283
0.68721175 -0.97940874 0.05102574 0.3272349 -0.19895138 -1.0964904
0.115697 -1.6193548 -0.3325356 0.8408984 0.56570625 -1.0137035
1.0420344 0.651026 0.17358325 -0.7841192 -1.5648905 -1.1487068
0.33041972 -0.56425524 0.53984326 0.6498902 -0.12847401 -0.14328234
-0.6593617 0.2106964 0.9734123 0.5385179 0.90693945 -1.0060979
-0.9599868 -0.33223802 -0.20020248 -1.1991616 -0.0628289 -0.88169295
-0.08469793 -1.5972075 0.39664212 -0.55365163 -0.418416 0.37683353
0.10822316 -0.483141 0.33610883 0.01292735 -1.1146624 0.15379006
1.2070985 -0.09605505 -0.07733338 0.5739122 -0.7482339 0.41089243
0.5907389 -0.80833745 -0.42137054 -0.6315013 0.72569275 0.6609767
0.23387966 -0.6702789 0.39591157 -0.6694897 -0.2554159 -0.22540626
0.22377004 -0.79227227 0.18265735 0.64353406 -0.6857896 -0.35235536
-0.17541648 1.3811536 0.43010888 -0.56673664 0.717925 -0.35598084
0.09361921 0.51920646 -0.032436 0.4243279 1.2644709 -0.05845928
-0.41453242 -1.0057391 -0.7906503 0.47884476 0.37522012 -0.25438228
-0.00901964 -0.75062686 -0.8182234 -0.41985723 -0.13606472 0.03175102
0.35585594 0.9633824 -0.44495243 0.4067718 0.3142899 -0.26291233
-1.0561264 0.72853357 0.4587804 -0.79939675 -0.2864639 1.0986896
-0.19659403 -0.24269946 0.40585712 0.15953194 0.7881952 0.03038249
0.3076489 0.60556996 -0.04203315 0.11717862 -0.04062483 -0.0587906
-0.24813989 -0.39197147 1.5444068 -0.3491975 -0.08520912 0.50855744
0.83038473 -0.03551702 -1.380165 -0.448531 0.0463203 -0.6281559
0.03078765 -0.7715981 -0.95421636 0.7270676 0.44383237 0.97727674
1.0093638 -0.10039609 0.8545928 0.83294 0.9895366 -1.2214128
-0.42186606 0.321849 0.5856067 -1.1447477 0.0791893 -0.10677533
-0.48406738 1.1361369 -0.0345448 -0.24627264 0.5263058 0.03732052
-0.34899583 0.1876005 -1.0087092 -0.34937185 -0.3225989 0.3321587
-0.13704385 0.24867998 -0.23581375 0.38314968 -0.22782011 -0.15702699
0.4842383 1.0057002 -0.645354 -1.132658 -0.3564075 0.6610143
-0.9010518 0.02756188 -0.25376907 0.80070794 -0.4295533 1.060521
-0.2889695 0.0898316 0.27150425 0.13541286 0.6964008 -0.3111328
-0.59259224 0.11837485 0.48324358 -0.5878481 -0.2763291 -0.6431683
-0.825092 -0.6626052 -0.5824179 0.19004749 0.68990934 1.0640944
0.08952998 -0.01214175 0.7534416 -0.40135464 -1.581839 -0.6056637
-0.8239474 0.19774112 0.5154714 -0.09337284 -0.5704364 -0.26398593] | [4.609299182891846, 3.400695562362671] |
4f2660de-2467-43a4-9a2d-48c5337301b6 | tkdp-threefold-knowledge-enriched-deep-prompt | 2306.03974 | null | https://arxiv.org/abs/2306.03974v2 | https://arxiv.org/pdf/2306.03974v2.pdf | TKDP: Threefold Knowledge-enriched Deep Prompt Tuning for Few-shot Named Entity Recognition | Few-shot named entity recognition (NER) exploits limited annotated instances to identify named mentions. Effectively transferring the internal or external resources thus becomes the key to few-shot NER. While the existing prompt tuning methods have shown remarkable few-shot performances, they still fail to make full use of knowledge. In this work, we investigate the integration of rich knowledge to prompt tuning for stronger few-shot NER. We propose incorporating the deep prompt tuning framework with threefold knowledge (namely TKDP), including the internal 1) context knowledge and the external 2) label knowledge & 3) sememe knowledge. TKDP encodes the three feature sources and incorporates them into the soft prompt embeddings, which are further injected into an existing pre-trained language model to facilitate predictions. On five benchmark datasets, our knowledge-enriched model boosts by at most 11.53% F1 over the raw deep prompt method, and significantly outperforms 8 strong-performing baseline systems in 5-/10-/20-shot settings, showing great potential in few-shot NER. Our TKDP can be broadly adapted to other few-shot tasks without effort. | ['Donghong Ji', 'Chong Teng', 'Liang Zhao', 'Bobo Li', 'Jingye Li', 'Fei Li', 'Hao Fei', 'Jiang Liu'] | 2023-06-06 | null | null | null | null | ['few-shot-ner', 'named-entity-recognition-ner'] | ['natural-language-processing', 'natural-language-processing'] | [-1.83865309e-01 2.95389384e-01 -3.47065419e-01 -3.69458705e-01
-1.12204301e+00 -5.92805564e-01 6.78273737e-01 1.07486390e-01
-1.00129795e+00 8.89074922e-01 7.50176668e-01 -9.25076380e-02
2.41510332e-01 -8.84584129e-01 -5.65600276e-01 -4.20628458e-01
4.03146327e-01 3.74118000e-01 4.47370768e-01 -3.58054399e-01
-4.55786251e-02 1.61413148e-01 -1.10190535e+00 5.59307709e-02
9.06964839e-01 6.98617220e-01 1.12772465e-01 5.99357665e-01
-7.03880668e-01 1.09917665e+00 -4.42266375e-01 -8.61745834e-01
4.65221591e-02 -1.64995924e-01 -1.06461620e+00 -6.56406283e-01
2.93110371e-01 -3.40075523e-01 -4.08061802e-01 7.00650811e-01
9.68940377e-01 6.94141209e-01 4.42818582e-01 -6.77910328e-01
-1.17944527e+00 1.03752053e+00 -6.20047711e-02 5.24017930e-01
1.70934916e-01 2.84114510e-01 1.11186612e+00 -1.04521489e+00
7.62276649e-01 8.10125649e-01 7.90333152e-01 9.62602496e-01
-9.74247575e-01 -5.23267031e-01 -5.32560796e-02 1.52421355e-01
-1.10385168e+00 -6.96185708e-01 3.43567044e-01 -1.56700641e-01
1.64460456e+00 -1.68878436e-01 1.50341019e-01 1.33724117e+00
-3.21468860e-01 8.37015033e-01 8.53431523e-01 -5.72542489e-01
2.55596012e-01 1.56415254e-01 7.10129797e-01 4.74622965e-01
-1.32276705e-02 6.66540787e-02 -4.98360008e-01 -1.28625229e-01
3.15160632e-01 1.24573037e-02 -1.57863095e-01 5.06062925e-01
-1.03892457e+00 6.77887082e-01 5.34473419e-01 5.38170755e-01
-6.20213866e-01 1.41420932e-02 4.59431171e-01 1.14218287e-01
5.56505740e-01 8.53936255e-01 -9.65974510e-01 -5.07874370e-01
-7.94710755e-01 -2.98325539e-01 1.16374910e+00 1.12004662e+00
9.04426455e-01 6.63080141e-02 -8.33592594e-01 1.03624129e+00
-1.11488737e-01 3.22825372e-01 6.28439367e-01 -5.14804125e-01
6.10408008e-01 5.68849146e-01 2.28824303e-01 -2.47651219e-01
-5.02228439e-01 -2.70142823e-01 -4.14376825e-01 -6.95095778e-01
1.35338858e-01 -7.12964714e-01 -1.09805083e+00 1.74278140e+00
4.68870401e-01 6.83622897e-01 2.61399209e-01 4.82431322e-01
1.22608459e+00 9.17962134e-01 8.72531950e-01 -1.24014489e-01
1.50581515e+00 -1.25225139e+00 -7.93446839e-01 -3.85734499e-01
1.05903900e+00 -6.37681067e-01 1.11169136e+00 -4.36118066e-01
-8.14851284e-01 -3.32629770e-01 -6.34031773e-01 -4.55434859e-01
-8.32951903e-01 3.18005979e-01 4.64359492e-01 5.65107584e-01
-8.89077604e-01 7.52285480e-01 -7.12889493e-01 -6.12851977e-01
2.87793666e-01 2.44983986e-01 -3.54248852e-01 -2.51323432e-01
-1.77431440e+00 1.15637720e+00 6.73618853e-01 -2.67382890e-01
-5.80234885e-01 -1.32119048e+00 -9.63213801e-01 3.88003558e-01
4.93714482e-01 -7.42518783e-01 1.45493591e+00 1.14845401e-02
-1.66453147e+00 6.65261149e-01 -2.65066534e-01 -3.94642085e-01
3.26228179e-02 -5.65342367e-01 -6.28105521e-01 -1.91179559e-01
2.71428257e-01 7.37853706e-01 1.72613561e-01 -6.56462967e-01
-4.62551147e-01 -4.16186042e-02 2.75492370e-01 7.47425556e-02
-7.05626428e-01 4.13743287e-01 -4.10566032e-01 -4.38787729e-01
-7.38529980e-01 -5.74433744e-01 -2.96907336e-01 -6.21628344e-01
-4.57888216e-01 -6.16116345e-01 3.59917104e-01 -6.11783326e-01
1.49304795e+00 -1.98479044e+00 -2.02739134e-01 -5.65025091e-01
-6.60795793e-02 8.41075897e-01 -3.72338712e-01 6.70126379e-01
5.56487925e-02 2.70985126e-01 9.21840146e-02 -4.48837191e-01
2.97767043e-01 1.86663255e-01 -3.44180077e-01 -1.91708371e-01
5.65457940e-01 1.44371927e+00 -1.25057769e+00 -3.78173351e-01
4.63708714e-02 5.84667206e-01 -3.26572716e-01 5.08409202e-01
-6.49466068e-02 5.28847706e-03 -4.60239798e-01 4.98386472e-01
3.74305099e-01 -3.73573333e-01 -1.69594124e-01 -2.65136212e-01
-2.29488283e-01 7.42174327e-01 -8.01906347e-01 1.71091187e+00
-5.93705893e-01 9.75516364e-02 -5.18701017e-01 -3.87830228e-01
9.06616330e-01 6.92798972e-01 1.08570894e-02 -8.07183206e-01
1.36773422e-01 9.57177877e-02 -3.97126973e-01 -3.50034803e-01
7.27560341e-01 -4.94599193e-01 -2.96091080e-01 4.37932968e-01
8.55847597e-01 4.16749418e-01 2.93461531e-01 4.22177374e-01
1.53121436e+00 4.38755080e-02 6.95292711e-01 2.76543852e-02
8.73441994e-02 -1.97799709e-02 8.56698930e-01 7.44035363e-01
-4.50904131e-01 2.64228493e-01 1.87898144e-01 -1.73058182e-01
-9.05029893e-01 -9.16877866e-01 1.23963989e-01 1.75989020e+00
-1.70651481e-01 -4.63217169e-01 -6.56142056e-01 -1.03354025e+00
-2.34197974e-01 1.25697279e+00 -7.78300881e-01 -2.98364371e-01
-5.60461700e-01 -6.83327615e-01 9.41902876e-01 1.08976519e+00
3.72059971e-01 -1.28699696e+00 -2.14560285e-01 5.39844692e-01
-9.39839408e-02 -1.36573756e+00 -7.01114357e-01 6.26275241e-01
-6.89523220e-01 -5.71068585e-01 -9.37894464e-01 -6.59065545e-01
2.24350110e-01 6.96723685e-02 1.14492559e+00 -3.55943859e-01
-1.69131160e-01 4.32312548e-01 -7.45310783e-01 -3.40987653e-01
-2.98653822e-02 4.35217440e-01 1.17890768e-01 -4.20522124e-01
9.23650265e-01 -4.90857899e-01 -4.00659084e-01 7.61347115e-02
-5.22276461e-01 -2.21402019e-01 7.81811535e-01 1.06878996e+00
3.98464531e-01 -5.75505018e-01 9.08182144e-01 -1.18495166e+00
3.73229206e-01 -7.82067537e-01 -2.11880601e-04 5.70532918e-01
-4.31846678e-01 2.29713544e-01 6.91452980e-01 -5.26124179e-01
-1.66141379e+00 -6.31263107e-02 -4.30535257e-01 -3.04716885e-01
-3.60106409e-01 4.82126653e-01 -2.51002699e-01 2.75848806e-01
9.39001560e-01 3.74401286e-02 -1.02033508e+00 -8.82768154e-01
9.50946629e-01 5.86443782e-01 5.10385394e-01 -6.78196788e-01
5.88758707e-01 5.57042025e-02 -7.27461755e-01 -7.20782161e-01
-1.60198700e+00 -1.00469053e+00 -7.93028891e-01 1.95839286e-01
9.77722347e-01 -1.04026520e+00 -1.17058106e-01 1.18547834e-01
-1.41251945e+00 -4.30339187e-01 -6.90415502e-01 4.18990523e-01
-1.80387557e-01 8.74886736e-02 -1.09429789e+00 -6.94975078e-01
-7.34926581e-01 -5.78497529e-01 9.74901080e-01 6.62784278e-01
2.29162266e-04 -1.07210100e+00 4.62781936e-01 2.14628935e-01
6.51797950e-01 -2.95956343e-01 7.08459914e-01 -1.42406428e+00
-1.01228185e-01 -2.28200465e-01 -4.18367952e-01 1.65506765e-01
-7.28731602e-02 -4.53648448e-01 -1.37756324e+00 1.72587439e-01
-3.70342642e-01 -5.95887959e-01 1.07723022e+00 -4.99982424e-02
4.09902573e-01 -2.59695023e-01 -3.98677081e-01 3.61247092e-01
1.37030804e+00 -3.13643068e-02 5.50453842e-01 3.07471186e-01
8.54798734e-01 3.06367487e-01 6.30280733e-01 5.05008459e-01
5.77635467e-01 4.70368594e-01 -1.08562134e-01 8.68543833e-02
-3.86103213e-01 -5.42037368e-01 4.79801208e-01 1.32889366e+00
-2.02816829e-01 -2.42411941e-01 -1.04543507e+00 8.86577725e-01
-1.62521565e+00 -1.01666713e+00 2.02746987e-01 1.68146574e+00
1.28957498e+00 -4.11458723e-02 -1.39126047e-01 -5.26600778e-01
8.60400021e-01 2.70645618e-01 -6.26406074e-01 -3.72095585e-01
4.22904342e-02 4.74609524e-01 4.85252619e-01 1.01969689e-01
-1.21191084e+00 1.44693875e+00 5.28904963e+00 9.07577097e-01
-6.93763316e-01 6.40314996e-01 1.95624501e-01 3.42701264e-02
-1.80875331e-01 1.72101080e-01 -1.56892538e+00 3.22708398e-01
1.64803064e+00 -5.00796139e-01 5.21089733e-02 1.10988355e+00
-3.13772798e-01 3.67652833e-01 -1.02108777e+00 4.51534480e-01
1.35746505e-03 -1.29089963e+00 -6.41043559e-02 -1.32516950e-01
9.78028953e-01 5.12361050e-01 -4.44754064e-01 1.19922757e+00
7.25941718e-01 -6.77046120e-01 3.46816182e-01 6.38971925e-01
8.78314197e-01 -5.80123246e-01 1.03209949e+00 5.40704846e-01
-1.27492738e+00 -9.66864377e-02 -6.63223743e-01 1.18624583e-01
5.95993519e-01 6.35370851e-01 -9.18172717e-01 6.27447903e-01
4.28125054e-01 6.35267377e-01 -2.61949033e-01 1.08523691e+00
-6.22905672e-01 9.31663811e-01 -1.86544910e-01 -2.55120933e-01
4.38280642e-01 4.75436181e-01 3.39342088e-01 1.75500810e+00
2.35260651e-01 5.19302011e-01 6.02306761e-02 7.23320365e-01
-6.63676262e-01 3.71943980e-01 -3.51161271e-01 -6.13705754e-01
8.23815823e-01 1.38893163e+00 -5.02672255e-01 -6.31228685e-01
-7.06476271e-01 1.15338004e+00 9.15728331e-01 2.55103230e-01
-8.27962041e-01 -8.13270092e-01 5.18620133e-01 -3.91916752e-01
7.12584376e-01 1.25720963e-01 -2.78755389e-02 -1.41941261e+00
-4.07936901e-01 -1.77303031e-01 5.72406471e-01 -5.84759414e-01
-1.83159137e+00 7.27166951e-01 -3.88425320e-01 -7.40885139e-01
-7.48995245e-02 -5.90918124e-01 -9.47036803e-01 7.70144224e-01
-1.78466809e+00 -1.32668138e+00 -3.80947739e-02 2.89535582e-01
6.25017583e-01 6.41489848e-02 1.33574712e+00 5.60513198e-01
-8.43127966e-01 8.40797842e-01 2.19534580e-02 4.94488984e-01
1.12953460e+00 -1.48518860e+00 7.18596399e-01 8.72608185e-01
2.30671048e-01 8.31645429e-01 2.96575129e-01 -6.84157431e-01
-1.07632685e+00 -1.41587293e+00 1.56626606e+00 -7.03666806e-01
8.60457838e-01 -1.35904729e-01 -1.05055249e+00 8.71319652e-01
2.45799199e-01 3.07408541e-01 1.15018237e+00 6.04432285e-01
-7.82166421e-01 2.09890693e-01 -8.96377206e-01 4.05548185e-01
1.05981112e+00 -6.42421663e-01 -1.36491382e+00 7.38184750e-02
1.44731390e+00 -1.31508172e-01 -1.25312948e+00 2.49058768e-01
1.47489905e-01 -4.36016977e-01 8.15089047e-01 -1.06278479e+00
4.25421894e-01 7.35081658e-02 -1.39914721e-01 -1.46587384e+00
-5.46610653e-01 -4.58733857e-01 -4.89660084e-01 1.91068411e+00
7.50503600e-01 -2.67157733e-01 4.26204920e-01 9.42131341e-01
-3.79345030e-01 -6.32728219e-01 -8.01540911e-01 -9.20059144e-01
1.41492888e-01 -3.21171016e-01 5.84830940e-01 1.27367783e+00
3.88678163e-01 8.21658909e-01 -4.34506208e-01 -1.41323023e-02
2.52786309e-01 -3.35071057e-01 4.05875653e-01 -1.07235289e+00
-2.52090037e-01 -1.67228997e-01 -1.82044581e-01 -9.27478254e-01
3.51457298e-01 -1.04947686e+00 2.70585626e-01 -1.70121622e+00
5.67591488e-01 -3.97215366e-01 -8.33953977e-01 1.18114960e+00
-7.88159072e-01 2.94168200e-03 2.38024920e-01 -1.49100780e-01
-1.15520322e+00 7.56228864e-01 7.08539546e-01 1.90183952e-01
-1.57613471e-01 -4.44760352e-01 -8.71966779e-01 4.92363721e-01
5.79674184e-01 -6.66534126e-01 6.97649689e-03 -5.22055447e-01
2.13437080e-01 -8.21002349e-02 -1.25188664e-01 -7.73263633e-01
7.83190727e-01 3.10035106e-02 1.19410731e-01 -2.22571135e-01
2.85777152e-01 -2.45894685e-01 -4.07732844e-01 -7.36276731e-02
-4.67984229e-01 -3.29364419e-01 3.12924296e-01 6.71000659e-01
-1.07115738e-01 -4.18424040e-01 6.91036403e-01 -2.07268685e-01
-1.28204405e+00 4.00708914e-01 9.35923010e-02 6.88522756e-01
8.15468848e-01 2.22115651e-01 -7.54770696e-01 9.27714407e-02
-8.33783746e-01 1.90227211e-01 7.70373717e-02 4.42724347e-01
2.12235421e-01 -1.27973187e+00 -5.98893464e-01 -3.31694216e-01
4.07197535e-01 -4.31730390e-01 6.13793612e-01 8.57379377e-01
1.97121635e-01 4.84651536e-01 2.34778896e-02 9.66213718e-02
-7.11793482e-01 6.46846890e-01 -4.00214680e-02 -6.11835897e-01
-6.73555553e-01 1.20960522e+00 -7.02602789e-02 -7.92540073e-01
1.57402202e-01 9.97468736e-03 -3.10332239e-01 2.92062014e-01
9.82114017e-01 6.40019596e-01 -5.61278453e-03 -3.82309020e-01
-3.51378232e-01 2.49859303e-01 -3.68110806e-01 4.80501801e-02
1.68497205e+00 5.79658849e-03 2.02425256e-01 6.31944478e-01
1.17460251e+00 -3.52026783e-02 -1.05157089e+00 -6.75087452e-01
4.90041316e-01 4.65386845e-02 4.55441549e-02 -1.23038971e+00
-6.81575060e-01 1.05225265e+00 1.94431879e-02 -2.12932810e-01
7.46032894e-01 2.82770425e-01 1.33585978e+00 6.42600417e-01
4.62152004e-01 -1.09871387e+00 -1.19574271e-01 1.11306846e+00
1.63364410e-01 -1.27806461e+00 -4.13988084e-01 -1.62671268e-01
-9.42902565e-01 9.18129742e-01 7.95463800e-01 5.34537155e-03
6.09393597e-01 3.72759789e-01 7.70356432e-02 -8.71877298e-02
-1.12455511e+00 -6.74539030e-01 3.38016450e-01 4.18336511e-01
6.52356803e-01 -1.62501838e-02 -3.02366853e-01 1.43763852e+00
1.57569960e-01 2.25142151e-01 2.27792650e-01 7.10156560e-01
-7.25439370e-01 -9.83087897e-01 3.15454483e-01 4.69481826e-01
-5.96598566e-01 -6.22731328e-01 -2.30518758e-01 4.93215472e-01
1.80791184e-01 7.89775074e-01 -3.08622539e-01 -3.39307696e-01
6.63675606e-01 7.00531304e-01 -3.28893736e-02 -1.08136237e+00
-9.42400694e-01 -3.85628194e-01 4.29751158e-01 -5.57136595e-01
-2.08263427e-01 -3.83004308e-01 -1.30477285e+00 -4.35588770e-02
-6.75062120e-01 3.29842865e-01 4.35951471e-01 1.21106946e+00
6.30827129e-01 6.01369023e-01 2.23929122e-01 -5.89937866e-01
-7.26199806e-01 -1.21301460e+00 -4.56209958e-01 2.93625891e-01
-2.14308366e-01 -6.61275446e-01 -4.11251515e-01 -5.77570796e-02] | [9.699664115905762, 9.342737197875977] |
cfe0e21c-d7da-4dc6-81fb-d00aa230dcc9 | motion-detection-using-csi-from-raspberry-pi | 2111.09091 | null | https://arxiv.org/abs/2111.09091v1 | https://arxiv.org/pdf/2111.09091v1.pdf | Motion Detection using CSI from Raspberry Pi 4 | Monitoring behaviour in smart homes using sensors can offer insights into changes in the independent ability and long-term health of residents. Passive Infrared motion sensors (PIRs) are standard, however may not accurately track the full duration of movement. They also require line-of-sight to detect motion which can restrict performance and ensures they must be visible to residents. Channel State Information (CSI) is a low cost, unintrusive form of radio sensing which can monitor movement but also offers opportunities to generate rich data. We have developed a novel, self-calibrating motion detection system which uses CSI data collected and processed on a stock Raspberry Pi 4. This system exploits the correlation between CSI frames, on which we perform variance analysis using our algorithm to accurately measure the full period of a resident's movement. We demonstrate the effectiveness of this approach in several real-world environments. Experiments conducted demonstrate that activity start and end time can be accurately detected for motion examples of different intensities at different locations. | ['Christopher Clare', 'Susan Craw', 'Stewart Massie', 'Glenn Forbes'] | 2021-11-17 | null | null | null | null | ['motion-detection'] | ['computer-vision'] | [ 3.89761209e-01 -3.18674564e-01 -8.54106620e-02 -1.28003150e-01
-7.60632694e-01 -2.92316556e-01 7.48161077e-02 -1.00266241e-01
-7.24608243e-01 9.82425928e-01 4.81758505e-01 -2.52872795e-01
-1.05923191e-01 -8.73731554e-01 -2.81659335e-01 -8.09140384e-01
-4.80694741e-01 -2.80530006e-01 5.48529863e-01 -1.50348172e-01
-2.88529187e-01 4.48806375e-01 -1.41136432e+00 1.93223525e-02
1.73464417e-01 7.07129121e-01 3.68749201e-01 1.17945194e+00
6.07415617e-01 1.10641885e+00 -8.85022521e-01 8.59424174e-01
1.37389123e-01 -1.97037041e-01 -1.61852181e-01 -3.75381619e-01
-2.00972900e-01 -9.17110145e-01 -6.95294201e-01 2.96933979e-01
9.53056991e-01 -3.03494860e-03 2.23779067e-01 -1.36111236e+00
9.71965566e-02 3.09634835e-01 -3.97534162e-01 6.74579084e-01
1.27155280e+00 6.20339848e-02 1.01954177e-01 3.17226440e-01
-1.19739681e-01 8.37193429e-01 1.05320406e+00 4.04191971e-01
-1.01160216e+00 -7.40458012e-01 -3.40516835e-01 2.26850986e-01
-1.67927539e+00 -7.46364594e-01 5.21592617e-01 -2.48808935e-01
1.05517972e+00 6.97867036e-01 6.30332291e-01 1.21681786e+00
2.56439030e-01 5.55054188e-01 9.19143736e-01 -6.73690736e-01
4.92768168e-01 -4.78539392e-02 2.34450307e-02 3.22134018e-01
3.74272048e-01 3.58105421e-01 -4.43303317e-01 -3.66757333e-01
7.12097347e-01 4.82193828e-01 -5.61086655e-01 3.16884369e-02
-1.30842829e+00 1.42840013e-01 1.80169791e-01 6.32607281e-01
-5.38763762e-01 5.72476387e-01 -1.75571352e-01 1.39728367e-01
-2.08411917e-01 -4.30929571e-01 -3.42467010e-01 -6.44179225e-01
-7.31925726e-01 1.03382252e-01 7.93968976e-01 1.05223668e+00
3.89204800e-01 -2.96554297e-01 -5.05292296e-01 2.37923726e-01
8.22334528e-01 1.22081506e+00 6.91750169e-01 -9.63165641e-01
5.00807524e-01 3.98708463e-01 8.18385780e-01 -6.95644021e-01
-1.10374761e+00 1.50891021e-01 -3.66876334e-01 2.60061711e-01
5.28535247e-01 -6.24406874e-01 -4.24533665e-01 1.60260046e+00
3.17774773e-01 3.08973551e-01 -3.09715383e-02 3.96013707e-01
4.17884856e-01 4.74071145e-01 1.98313341e-01 -4.25744653e-01
1.32856572e+00 -1.29352063e-01 -8.49672675e-01 -2.50365734e-01
7.15859056e-01 -3.09868068e-01 5.30064285e-01 1.56270843e-02
-7.02766359e-01 -3.48845929e-01 -1.37228572e+00 7.11741269e-01
-8.21769238e-02 -3.82333784e-03 2.59822011e-01 1.57115924e+00
-1.07333994e+00 1.70850664e-01 -1.53003716e+00 -4.78614897e-01
2.48611078e-01 6.31246209e-01 -2.13444382e-01 -1.22278146e-01
-1.00642025e+00 7.06369400e-01 -1.93615913e-01 -1.74377516e-01
-1.17960975e-01 -4.20055181e-01 -7.95722187e-01 -1.98066160e-01
-2.28374213e-01 -3.86827081e-01 1.55219567e+00 -3.79706174e-01
-1.45256245e+00 2.45030776e-01 -4.21514630e-01 -4.41590071e-01
5.69548309e-01 -3.05549167e-02 -1.07850873e+00 8.30353498e-02
4.31299090e-01 1.42978117e-01 2.50529647e-01 -7.14957178e-01
-8.47963393e-01 -5.31510711e-01 -2.96269029e-01 -2.08308734e-02
-1.64890155e-01 7.42973620e-03 1.20938629e-01 -1.71149984e-01
1.36105344e-01 -1.03998411e+00 -2.51360744e-01 -1.76232472e-01
-3.04724090e-02 3.98020744e-01 7.46701360e-01 -5.99703014e-01
1.53994012e+00 -1.96306086e+00 -1.28454769e+00 4.55429226e-01
-2.45781973e-01 1.16649531e-01 4.58229482e-01 3.99301857e-01
3.96605790e-01 -4.37982976e-01 9.03474167e-02 -2.76577082e-02
2.15263516e-02 2.23721504e-01 3.68773431e-01 8.94168019e-01
-7.23578513e-01 8.23516548e-01 -9.86892939e-01 -2.42698550e-01
7.60292649e-01 7.51382351e-01 -5.69504127e-02 3.47101390e-02
6.74424291e-01 4.88283068e-01 -6.21674776e-01 4.66730267e-01
4.80720967e-01 -4.35617827e-02 2.39149965e-02 3.01243246e-01
-3.53568643e-01 2.42873877e-01 -1.45770669e+00 1.26135981e+00
-4.40318525e-01 8.25363696e-01 -1.88634291e-01 -6.65877283e-01
5.66966355e-01 7.33842850e-01 8.45973849e-01 -1.16018128e+00
2.01899320e-01 -1.76685810e-01 -5.29890478e-01 -9.19589579e-01
1.63043961e-01 -5.44537138e-03 -2.99348980e-01 6.52532995e-01
-8.48309636e-01 8.51193845e-01 -2.31873512e-01 -3.43395412e-01
2.15716457e+00 1.71543825e-02 7.34355211e-01 -1.00717098e-01
2.62064517e-01 -1.57825425e-01 2.12114707e-01 9.93236303e-01
-7.72729874e-01 2.04557404e-01 -5.87548077e-01 -2.07438111e-01
-4.71191496e-01 -1.32085145e+00 2.09492110e-02 1.05247569e+00
2.08365336e-01 1.19430572e-01 -5.66722512e-01 1.13278449e-01
-2.50317063e-02 6.94547594e-01 -4.09673750e-01 -9.58060101e-02
-5.81976235e-01 -9.17369187e-01 7.35453904e-01 8.57752860e-01
7.63079762e-01 -8.17229748e-01 -1.70484030e+00 7.51569986e-01
-5.27759969e-01 -1.05227697e+00 -1.80024251e-01 1.67334139e-01
-6.33297682e-01 -9.80714083e-01 -7.73261547e-01 -5.20720363e-01
4.82956171e-01 9.58997488e-01 6.91769123e-01 -1.71259910e-01
-4.08914596e-01 1.14570689e+00 -3.45638871e-01 -5.99001765e-01
-9.98210534e-02 -4.31912392e-01 5.42311966e-02 -2.44329304e-01
1.00675726e+00 -7.82278180e-01 -1.01419032e+00 6.28768325e-01
-3.17666858e-01 -5.88891625e-01 3.23722690e-01 -4.98609692e-02
1.30029067e-01 3.68297756e-01 4.15529698e-01 -2.33434036e-01
4.00306374e-01 -4.96663868e-01 -5.51748157e-01 1.05020441e-01
-3.64607811e-01 2.01689452e-02 1.95022412e-02 -3.19881678e-01
-1.08123350e+00 5.13813972e-01 3.19387242e-02 5.96775413e-01
-5.88460207e-01 -1.52736068e-01 -2.36668870e-01 -1.28589749e-01
9.40399706e-01 8.29845853e-03 -1.38668939e-01 -1.50919139e-01
-1.54063538e-01 1.10131371e+00 9.13667202e-01 1.68688416e-01
5.59250712e-01 1.16225731e+00 -1.17082588e-01 -1.33257854e+00
-1.36073053e-01 -1.33349037e+00 -5.07992804e-01 -4.49392200e-01
8.61420572e-01 -1.29313505e+00 -1.32737684e+00 4.37527925e-01
-7.24191070e-01 -6.33606493e-01 1.07991122e-01 7.00638056e-01
-5.03995836e-01 3.37751120e-01 -2.19379142e-01 -1.44210279e+00
-2.00049594e-01 -4.42907006e-01 1.01175189e+00 3.90490711e-01
-8.18310201e-01 -9.91410077e-01 4.46495742e-01 3.91188055e-01
5.30455887e-01 6.94891036e-01 -3.57576430e-01 2.22722858e-01
-4.30690438e-01 -7.80157804e-01 3.58886003e-01 -4.17511314e-01
5.50710618e-01 -6.91126585e-01 -1.20053554e+00 -3.20843190e-01
1.08909741e-01 5.68536937e-01 2.42480621e-01 1.07528722e+00
3.37821126e-01 -3.94110769e-01 -1.07514536e+00 2.19481081e-01
1.42076862e+00 5.41299403e-01 1.29674387e+00 8.23410273e-01
3.17750752e-01 1.41378185e-02 5.29754400e-01 9.44115460e-01
5.32437921e-01 1.05360675e+00 -3.65465553e-03 4.70501445e-02
1.93189308e-01 -1.98719949e-02 7.65072584e-01 -1.77419335e-02
-3.54474574e-01 -3.54956090e-01 -5.80401897e-01 3.95092338e-01
-1.83490121e+00 -1.60081911e+00 -5.78437746e-01 2.42226529e+00
1.78015634e-01 -2.10519433e-02 5.87869406e-01 8.88819993e-01
6.16433501e-01 -3.46390337e-01 -1.96623281e-01 3.03958982e-01
3.81542414e-01 1.21896945e-01 1.66620362e+00 5.16066730e-01
-1.01197779e+00 1.75212743e-03 7.16500950e+00 -2.58596718e-01
-5.24497092e-01 2.38165230e-01 -2.12961257e-01 -2.17165291e-01
1.90077975e-01 -5.88223040e-01 -7.08034635e-01 7.57373929e-01
1.37343848e+00 1.74454868e-01 2.06362858e-01 8.57520401e-01
9.69707131e-01 -9.04991806e-01 -8.73953462e-01 1.14363778e+00
-1.95170581e-01 -6.24118090e-01 -1.19603252e+00 2.95043766e-01
3.47907275e-01 2.05534101e-01 -3.93691093e-01 -7.66947269e-02
4.74283338e-01 -7.10138142e-01 6.12718284e-01 5.94888628e-01
3.97852629e-01 -7.68182397e-01 5.48980534e-01 5.55765033e-01
-1.64659131e+00 -3.87865901e-01 -1.83006331e-01 -5.93843043e-01
5.41300058e-01 2.69503891e-01 -8.74012947e-01 -9.25918221e-02
8.73101354e-01 2.88960803e-02 -3.09130162e-01 1.29367006e+00
-1.03538759e-01 5.70020258e-01 -7.44748235e-01 -3.35477293e-01
-2.20082596e-01 3.76493156e-01 1.81896225e-01 1.07471967e+00
6.87421739e-01 5.01188040e-01 -8.92993286e-02 2.62981892e-01
8.60247195e-01 -6.65848434e-01 -8.14841628e-01 7.66484380e-01
8.19133997e-01 6.75616920e-01 -6.18263781e-01 -6.69514388e-02
-6.59709036e-01 9.07188296e-01 -7.47197628e-01 3.63504708e-01
-7.49835491e-01 -2.07729816e-01 5.06077886e-01 5.04359663e-01
1.99889302e-01 -4.91405666e-01 -6.65424392e-02 -4.10319537e-01
8.43142252e-03 -1.43456116e-01 3.78558785e-01 -6.61639631e-01
-4.16246831e-01 -1.98102593e-01 1.48288861e-01 -1.62600148e+00
-5.03400147e-01 -3.43862385e-01 -5.58476865e-01 5.93206048e-01
-1.16410577e+00 -6.16734266e-01 -9.32448030e-01 1.10618794e+00
-3.62572558e-02 2.66194910e-01 1.01965344e+00 6.42861426e-01
-2.08093479e-01 7.49890089e-01 4.33950096e-01 1.19257592e-01
2.11453646e-01 -9.56413627e-01 3.71648669e-01 7.70239234e-01
-1.91960424e-01 2.49515593e-01 9.31852818e-01 -5.95835984e-01
-1.36809957e+00 -9.54292476e-01 7.23046780e-01 -8.76143754e-01
2.37664521e-01 8.12959895e-02 -3.26494962e-01 8.27179253e-01
-4.52914327e-01 -9.32547376e-02 1.04472697e+00 -4.96475607e-01
5.08309722e-01 -1.80134192e-01 -1.44327927e+00 4.06613410e-01
1.15854263e+00 -3.02363455e-01 -4.36719745e-01 4.04217504e-02
-3.99951963e-03 -8.56955498e-02 -5.60113072e-01 -1.24926805e-01
1.12607884e+00 -1.12792933e+00 1.29052067e+00 3.77403557e-01
-8.21871340e-01 -4.51675713e-01 -3.19588274e-01 -7.42382169e-01
-4.93693173e-01 -6.64508522e-01 -7.36529380e-02 9.56918120e-01
1.47462681e-01 -8.32999766e-01 9.18649673e-01 1.06753027e+00
2.73815274e-01 3.89099032e-01 -1.19254768e+00 -1.11381185e+00
-9.71746743e-01 -8.35446179e-01 6.76625192e-01 3.38253498e-01
7.23758519e-01 3.89849022e-02 -4.47140813e-01 5.73995590e-01
7.58650601e-01 -8.17724943e-01 7.50535548e-01 -1.18725443e+00
-2.62581497e-01 1.52814716e-01 -1.04313290e+00 -1.04442573e+00
-5.00593364e-01 -7.10171983e-02 3.69573027e-01 -1.83001924e+00
-2.04390720e-01 -1.97845250e-01 -2.17269406e-01 3.72647792e-01
1.16383895e-01 2.60302704e-02 -6.38252497e-01 2.94324066e-02
-4.96474624e-01 -1.24367207e-01 4.08126742e-01 -1.49992287e-01
-6.99376643e-01 8.10246408e-01 -3.99539471e-01 5.01856387e-01
9.76759553e-01 -4.87613082e-01 -6.24488413e-01 4.03128564e-02
8.00893977e-02 2.33507350e-01 6.68698013e-01 -1.88149917e+00
2.67203778e-01 -1.05882652e-01 7.89104581e-01 -6.09975874e-01
3.34900022e-01 -1.41120172e+00 6.93961203e-01 1.01114416e+00
1.24091595e-01 1.51810378e-01 1.63100496e-01 8.87120843e-01
6.90116704e-01 2.78570373e-02 5.43831289e-01 -2.80173779e-01
-7.53617704e-01 -1.11157238e-01 -1.01433980e+00 -4.74504083e-01
1.22170985e+00 -7.65919268e-01 -1.91067785e-01 -6.77878857e-01
-2.72919625e-01 -2.63273828e-02 4.43991095e-01 1.84976101e-01
1.61799133e-01 -1.63950682e+00 -1.19800381e-01 2.20610797e-01
1.88088909e-01 -4.45966870e-01 1.87121511e-01 7.72391260e-01
-6.08492911e-01 6.52195573e-01 -9.69953388e-02 -7.09655285e-01
-1.57398307e+00 4.06776249e-01 3.16794664e-01 1.20564297e-01
-9.22723949e-01 3.69563252e-01 -4.84592766e-01 2.95515925e-01
3.66347313e-01 -6.00525677e-01 -3.74244958e-01 -3.61557275e-01
1.34067774e+00 9.78916287e-01 -1.17189571e-01 -7.79912412e-01
-7.82051027e-01 4.34757113e-01 7.27162778e-01 -5.77344716e-01
1.01666474e+00 -8.65556479e-01 6.00840449e-01 4.64730978e-01
1.06063712e+00 -1.86369941e-02 -1.38315260e+00 3.06075253e-02
2.73383588e-01 -3.02309722e-01 2.80173663e-02 -4.04618979e-01
-3.14956933e-01 1.35901004e-01 1.67504764e+00 4.38823730e-01
1.01767337e+00 -1.56317204e-01 8.77885222e-01 4.93344843e-01
8.81684244e-01 -1.31485760e+00 -3.04431528e-01 -3.56805444e-01
-9.48334229e-04 -1.14168370e+00 1.35375336e-02 -1.29292920e-01
-7.66495168e-02 7.15078771e-01 -5.61846346e-02 3.74395251e-01
7.93164551e-01 7.19513535e-01 3.63796294e-01 1.57379918e-02
-1.38726965e-01 -4.57675457e-01 -4.21492994e-01 1.53029168e+00
3.13943714e-01 6.29859924e-01 1.12402730e-01 2.83951670e-01
-1.96641698e-01 3.77836168e-01 5.23214757e-01 1.68120575e+00
-9.63206351e-01 -7.37410486e-01 -1.06549072e+00 2.60301441e-01
-4.34780210e-01 6.01173699e-01 2.28850558e-01 6.60284102e-01
-1.82753593e-01 1.78492105e+00 -1.97252944e-01 -3.40092063e-01
6.85124576e-01 -1.04059227e-01 5.31316280e-01 -6.16508424e-02
9.33704004e-02 -1.57968223e-01 2.51195997e-01 -7.88430870e-01
-8.91461551e-01 -8.74902248e-01 -1.33840227e+00 -4.27308470e-01
-6.48536608e-02 -2.50779301e-01 6.14797652e-01 1.10911405e+00
3.53375822e-02 7.07678974e-01 6.51274860e-01 -8.34519565e-01
-2.10999221e-01 -7.73243308e-01 -8.61278653e-01 1.54009849e-01
6.00497842e-01 -5.71802497e-01 -2.55321205e-01 3.01578194e-01] | [6.897494792938232, 0.6319673657417297] |
dac838f2-8657-4acc-9ea1-81cb9d2cc757 | the-effectiveness-of-simple-hybrid-systems | null | null | https://aclanthology.org/P19-1327 | https://aclanthology.org/P19-1327.pdf | The Effectiveness of Simple Hybrid Systems for Hypernym Discovery | Hypernymy modeling has largely been separated according to two paradigms, pattern-based methods and distributional methods. However, recent works utilizing a mix of these strategies have yielded state-of-the-art results. This paper evaluates the contribution of both paradigms to hybrid success by evaluating the benefits of hybrid treatment of baseline models from each paradigm. Even with a simple methodology for each individual system, utilizing a hybrid approach establishes new state-of-the-art results on two domain-specific English hypernym discovery tasks and outperforms all non-hybrid approaches in a general English hypernym discovery task. | ['Nizar Habash', 'William Held'] | 2019-07-01 | null | null | null | acl-2019-7 | ['hypernym-discovery'] | ['natural-language-processing'] | [-6.11061603e-02 1.92390695e-01 -5.91899633e-01 -1.82150587e-01
-3.59673232e-01 -5.24561584e-01 1.20569909e+00 4.65993077e-01
-9.42458630e-01 6.91712797e-01 2.12587819e-01 -3.66735399e-01
-6.38479769e-01 -8.90495479e-01 -8.28354619e-03 -4.60273802e-01
2.02037588e-01 1.16926658e+00 3.29413742e-01 -6.61398172e-01
2.51605123e-01 2.90556192e-01 -2.18444729e+00 3.12460065e-01
6.25882149e-01 4.88478422e-01 -2.35796198e-02 -7.98485056e-03
-9.23622429e-01 5.16886771e-01 -3.56253326e-01 -6.38219595e-01
3.68964076e-01 1.88452542e-01 -1.01627994e+00 -6.59689724e-01
5.78064859e-01 3.29841107e-01 -3.79584521e-01 1.12728024e+00
6.62007332e-01 7.24483803e-02 6.32601857e-01 -1.31476271e+00
-7.19404995e-01 1.11328197e+00 -3.30573589e-01 1.48786120e-02
6.38409436e-01 -2.83554077e-01 1.37875867e+00 -6.39575303e-01
1.10094988e+00 1.12126625e+00 7.40760863e-01 4.80940878e-01
-1.27257359e+00 -9.62051988e-01 -3.04506093e-01 3.49848002e-01
-1.42015386e+00 -2.07876027e-01 3.85464191e-01 -2.14232594e-01
1.70634341e+00 2.34324839e-02 4.20767665e-01 8.73911083e-01
-5.51959693e-01 4.73320365e-01 1.29935169e+00 -1.03461957e+00
-2.37484761e-02 4.89446282e-01 5.47730267e-01 4.08055961e-01
6.91932380e-01 2.39279307e-02 -3.31084281e-01 -6.55975640e-01
8.30184892e-02 -2.35124961e-01 2.42919493e-02 -4.41774249e-01
-9.96482432e-01 7.75021911e-01 -1.61935985e-01 9.05400276e-01
-3.08326900e-01 -3.33303541e-01 7.08680332e-01 2.79369831e-01
4.18942988e-01 9.42437887e-01 -6.60873055e-01 -4.90921199e-01
-1.29091668e+00 6.76296651e-01 1.33887708e+00 1.12237060e+00
5.16629457e-01 -9.58809927e-02 -5.54487556e-02 1.24483192e+00
2.90288061e-01 1.84725687e-01 8.86336029e-01 -4.70107317e-01
3.69362831e-01 9.71806705e-01 -2.54400373e-01 -6.07896924e-01
-5.34724236e-01 -1.89714670e-01 -1.87069729e-01 -5.62583916e-02
2.26087764e-01 2.16297492e-01 -1.14166975e+00 1.82564199e+00
3.35280687e-01 1.91245787e-02 1.55309409e-01 2.24323422e-01
1.24283135e+00 2.49034390e-01 5.88388383e-01 -3.49442624e-02
1.47063637e+00 -7.18963385e-01 -7.55544364e-01 2.23573953e-01
9.20776129e-01 -7.32189417e-01 1.09199834e+00 2.92267054e-01
-9.26959574e-01 -1.40214592e-01 -1.08236313e+00 -8.51054043e-02
-1.32050598e+00 -1.53984785e-01 6.13212526e-01 9.11682606e-01
-8.30749393e-01 5.22499442e-01 -1.99875221e-01 -1.01281738e+00
8.47039819e-02 2.80065686e-01 -4.69453663e-01 -2.68243011e-02
-1.78679204e+00 1.65017021e+00 8.90515387e-01 -7.34429598e-01
-3.62081617e-01 -1.03688180e+00 -9.03851449e-01 -1.03228167e-01
4.63585436e-01 -7.62797832e-01 1.13750315e+00 -1.05174802e-01
-1.09575629e+00 1.34666348e+00 -6.89259917e-02 -7.27058232e-01
1.13287121e-01 -3.12507421e-01 -8.49310279e-01 -2.10930765e-01
1.76878422e-01 5.94196081e-01 1.77226231e-01 -1.20419443e+00
-8.55405211e-01 -1.26970336e-01 7.26360828e-03 1.93775818e-01
-7.37844646e-01 3.73654962e-01 -4.35162149e-02 -5.12033999e-01
-2.54457206e-01 -8.40452671e-01 5.42992316e-02 -5.95883369e-01
-1.40516743e-01 -8.05780828e-01 6.82855785e-01 -1.47689879e-01
1.89982319e+00 -1.81190848e+00 3.60203460e-02 5.90929016e-02
2.68601626e-01 7.16026127e-01 -5.27412482e-02 1.02559853e+00
-4.85875845e-01 5.76304793e-01 -2.86225647e-01 -3.35327715e-01
3.25015664e-01 5.61926842e-01 -3.51719081e-01 4.02310975e-02
-2.79106528e-01 7.31809556e-01 -1.05851936e+00 -5.79173326e-01
5.10421634e-01 2.87955225e-01 -3.63976173e-02 1.99349690e-02
-4.57955822e-02 -6.43154681e-01 1.18278433e-02 6.98635399e-01
3.34061295e-01 1.29938379e-01 5.34169495e-01 -1.00545347e-01
-1.04877144e-01 5.53019881e-01 -1.15475643e+00 1.76897681e+00
-4.51666921e-01 4.37268347e-01 -4.66570377e-01 -8.79231751e-01
7.31333971e-01 6.58844650e-01 6.51309192e-01 -4.64973897e-01
-8.17899704e-02 7.91963041e-01 3.73320401e-01 -6.05773687e-01
8.89109910e-01 -3.80956471e-01 -2.21068170e-02 3.25549543e-01
5.26840091e-01 -1.39040291e-01 6.60207808e-01 2.13081613e-01
1.34026289e+00 3.08214545e-01 1.05848479e+00 -5.08923769e-01
3.11750889e-01 2.28093803e-01 2.50401855e-01 8.86749506e-01
-9.65645090e-02 3.82920712e-01 -2.04605684e-02 -2.63814002e-01
-1.23649037e+00 -1.03228223e+00 -6.81825101e-01 1.13702297e+00
-6.40034303e-02 -1.02775395e+00 -4.26073015e-01 -7.95879126e-01
2.89844036e-01 1.23135495e+00 -6.35662973e-01 6.92897364e-02
-4.65698034e-01 -1.09512055e+00 1.32837343e+00 1.67678714e-01
1.31297961e-01 -1.39073110e+00 -6.56979620e-01 3.10422480e-01
9.37931463e-02 -1.31771362e+00 4.45541471e-01 3.94257277e-01
-6.15825415e-01 -1.07428563e+00 -3.59478474e-01 -6.46946907e-01
-2.86439117e-02 1.16044529e-01 1.60902393e+00 7.63275698e-02
-6.44682467e-01 3.57003152e-01 -7.99071908e-01 -4.75661129e-01
-2.68424004e-01 5.65856159e-01 2.06827313e-01 -6.21968985e-01
1.31345415e+00 -8.30992877e-01 1.32800549e-01 -1.92200407e-01
-1.21399128e+00 -6.10847235e-01 4.67194885e-01 5.66735983e-01
1.77753642e-01 7.80570582e-02 5.04479289e-01 -1.49318802e+00
1.00420475e+00 -9.73862708e-01 -1.21248394e-01 5.16045988e-01
-1.37083447e+00 1.05161145e-01 2.88948566e-01 -4.85463798e-01
-9.20669138e-01 -1.93460986e-01 -1.03274807e-01 -2.36548647e-01
-4.91091341e-01 7.02269495e-01 -6.13990501e-02 1.01451464e-01
7.44191825e-01 9.45172608e-02 -3.85609388e-01 -6.95289254e-01
8.76028776e-01 7.51093447e-01 4.92654532e-01 -9.20486093e-01
8.45062137e-01 2.63838112e-01 -1.16480969e-01 -9.92357671e-01
-6.45815611e-01 -1.07451332e+00 -8.32958817e-01 3.21259052e-01
6.49790108e-01 -6.72359109e-01 -1.32455975e-01 3.95294577e-01
-1.06191325e+00 2.11787120e-01 -6.26998007e-01 1.85921699e-01
-1.06328301e-01 6.08456969e-01 -2.10395321e-01 -7.19260991e-01
-3.41687799e-01 -6.27763748e-01 7.76574135e-01 4.67277020e-02
-7.54533827e-01 -1.18282151e+00 6.88107848e-01 9.97772366e-02
5.04267693e-01 1.30195618e-01 1.37731349e+00 -1.61064661e+00
3.60405117e-01 -5.17617702e-01 -8.63463953e-02 1.91317238e-02
9.57195312e-02 1.26488969e-01 -9.94616807e-01 1.87005848e-01
-2.48433068e-01 -3.69890064e-01 8.21924031e-01 6.24082126e-02
6.89289212e-01 3.81024256e-02 -5.92497528e-01 3.83358449e-01
1.73457301e+00 7.50305951e-02 7.61438310e-01 1.02430999e+00
5.75982988e-01 7.50014007e-01 5.97010076e-01 3.35965306e-01
6.79455817e-01 8.13930333e-01 2.59593129e-01 1.24259144e-01
-2.43445531e-01 -1.22236215e-01 -2.37308014e-02 6.19136214e-01
-2.68251121e-01 -2.23302066e-01 -1.24194002e+00 9.79194641e-01
-2.07563686e+00 -1.25592697e+00 -2.52309382e-01 2.15002894e+00
1.16962206e+00 1.05023291e-03 2.29627579e-01 5.18020056e-02
6.10733926e-01 2.22242251e-01 -2.22411845e-02 -4.95394170e-01
-5.07484853e-01 7.74861991e-01 4.11517441e-01 1.71692327e-01
-9.90109026e-01 1.19280589e+00 7.10008860e+00 1.26402318e+00
-4.69210953e-01 2.56906956e-01 -4.66261595e-01 -1.55247927e-01
-6.10775471e-01 4.51711684e-01 -1.03876543e+00 3.16179186e-01
8.95691514e-01 -3.27907532e-01 2.18348578e-01 8.82599473e-01
-5.74657738e-01 -6.37480840e-02 -9.66478467e-01 9.56675470e-01
2.38215640e-01 -1.04641461e+00 4.07221258e-01 2.89424986e-01
6.23440683e-01 3.68316770e-01 -3.61130953e-01 6.63913608e-01
6.23851657e-01 -1.10790181e+00 3.82758081e-01 2.53526896e-01
6.43015385e-01 -4.33449924e-01 9.37087119e-01 1.59738660e-01
-1.19580162e+00 -4.15643938e-02 -3.70205909e-01 -8.46308842e-02
3.90042990e-01 7.14890480e-01 -4.30756658e-01 8.89705896e-01
7.23112285e-01 4.77007270e-01 -6.54924393e-01 1.13384295e+00
-2.85498410e-01 5.66715717e-01 -5.50751388e-01 3.21751870e-02
3.15953910e-01 -4.42613624e-02 7.75548518e-01 1.75741374e+00
6.67922758e-03 -9.95448157e-02 1.00158796e-01 8.26128125e-01
4.51678261e-02 4.47026342e-01 -8.80978525e-01 -3.23550016e-01
1.08373451e+00 1.50310123e+00 -3.35118949e-01 -5.10574698e-01
-5.62419772e-01 3.93142700e-01 6.16391122e-01 8.95415097e-02
-6.49521112e-01 -8.36399674e-01 6.25425577e-01 7.99419284e-02
7.09840730e-02 -1.01636499e-01 -4.78108585e-01 -1.05431008e+00
9.75532737e-03 -9.79498863e-01 1.02076924e+00 -4.40731704e-01
-1.65831435e+00 7.10615993e-01 7.05389261e-01 -1.08564913e+00
-4.08746839e-01 -8.03439856e-01 -3.14061373e-01 7.53040433e-01
-1.47024298e+00 -1.38181496e+00 -8.60870779e-02 2.04949901e-01
2.62515604e-01 -5.52055180e-01 1.34589458e+00 5.75885177e-01
-1.63771987e-01 6.02782428e-01 -6.02575205e-02 -1.15290888e-01
9.78462040e-01 -1.57681298e+00 3.69538516e-01 5.88461578e-01
6.31518900e-01 1.00266504e+00 5.89272559e-01 -7.70938396e-01
-7.57097363e-01 -4.86610562e-01 1.51013601e+00 -8.28917742e-01
9.32419479e-01 -1.77446641e-02 -9.90756571e-01 4.09953028e-01
6.49711907e-01 -2.49348700e-01 8.36839437e-01 8.90196979e-01
-9.15249646e-01 3.92707378e-01 -1.13459480e+00 4.97200251e-01
1.20077503e+00 -6.06491804e-01 -1.46411657e+00 7.39492774e-02
5.66625953e-01 4.22051586e-02 -1.11823916e+00 8.52721095e-01
6.86580300e-01 -6.97588205e-01 9.21712935e-01 -1.02039051e+00
4.56017405e-01 -3.41247171e-01 -4.03217614e-01 -1.20425487e+00
-1.72075212e-01 -4.43559766e-01 -4.07817632e-01 1.57696533e+00
7.41024315e-01 -6.64179146e-01 3.45477670e-01 4.38241273e-01
7.15353061e-03 -7.21550703e-01 -8.76440883e-01 -1.07635415e+00
3.63595933e-01 -4.40575749e-01 6.49220645e-01 1.44858611e+00
5.57551444e-01 5.23013055e-01 -1.05467856e-01 -4.62538481e-01
3.94576609e-01 -1.15498535e-01 4.95781690e-01 -1.36380529e+00
2.86424849e-02 -8.64494383e-01 -6.42644465e-01 -2.03519180e-01
6.25944912e-01 -1.32322550e+00 -2.39301190e-01 -1.66567111e+00
5.26992142e-01 -3.74439538e-01 -3.74414474e-01 8.33739519e-01
-2.29338124e-01 4.43472788e-02 1.06071442e-01 3.34035724e-01
-3.86106580e-01 2.89409161e-01 2.54835457e-01 8.44629705e-02
2.64951382e-02 -5.01549602e-01 -5.51203787e-01 6.28798306e-01
6.62222266e-01 -6.98158383e-01 -5.37340164e-01 -2.90501982e-01
3.93682033e-01 -8.48253369e-01 6.62127063e-02 -9.49759841e-01
2.87377477e-01 8.88086483e-02 -3.97626013e-01 -2.94892669e-01
1.97077826e-01 -8.12722266e-01 7.83745497e-02 3.24866593e-01
-3.66608292e-01 3.06910127e-01 3.05067867e-01 3.46483678e-01
-3.40080023e-01 -7.57225394e-01 6.48794115e-01 -3.21628928e-01
-9.89145398e-01 3.12692709e-02 -5.21571562e-02 6.36527121e-01
8.82715821e-01 -4.96427804e-01 -4.44938958e-01 2.99165696e-02
-5.29313684e-01 -7.95248672e-02 3.73854965e-01 8.27723205e-01
1.21604495e-01 -1.19207239e+00 -7.53950894e-01 -2.61570305e-01
6.50556326e-01 -6.82385147e-01 -2.75043279e-01 6.28606737e-01
-2.89684623e-01 5.70960402e-01 -1.28454909e-01 -1.36362150e-01
-1.25962710e+00 6.66453481e-01 2.39053577e-01 -6.49121106e-01
-4.80747759e-01 4.22554344e-01 -3.91328603e-01 -1.10373044e+00
2.92236447e-01 2.85283059e-01 -5.16697586e-01 2.99105316e-01
2.14530185e-01 6.58347011e-01 2.67266065e-01 -6.86793029e-01
-4.99434143e-01 4.96132106e-01 -1.88228920e-01 -4.43317056e-01
1.26406550e+00 1.68291599e-01 -6.10284626e-01 6.91532671e-01
9.38914120e-01 -6.70085177e-02 2.46826246e-01 -4.02724385e-01
6.62084639e-01 -3.25844467e-01 -1.52082801e-01 -1.19780564e+00
-5.13750315e-01 6.52389288e-01 4.95092273e-01 4.56310689e-01
8.05609465e-01 -4.19737995e-02 6.82615936e-01 5.27631044e-01
4.46447700e-01 -1.16571188e+00 -8.40108454e-01 7.20306993e-01
3.82092714e-01 -9.59271312e-01 3.78090233e-01 -5.32921851e-01
-4.99944091e-01 1.12953937e+00 6.61266387e-01 5.69803528e-02
1.09128737e+00 4.02761906e-01 4.65709604e-02 -6.58263266e-01
-6.53700531e-01 -8.78271759e-01 4.64749932e-01 6.42017543e-01
9.61517334e-01 1.45064965e-01 -9.88375068e-01 4.58568782e-01
-3.13283354e-01 -2.60456085e-01 8.57258141e-02 9.98684525e-01
-3.04510891e-01 -1.84691417e+00 3.00295800e-01 5.64155877e-01
-6.15983009e-01 -8.76272261e-01 -8.47178996e-01 1.45530009e+00
3.14275295e-01 7.26425529e-01 -3.30711782e-01 -3.37517351e-01
3.69549155e-01 6.01127565e-01 4.64757502e-01 -1.07053685e+00
-1.12187111e+00 -5.73510647e-01 6.09622002e-01 -1.99136749e-01
-8.63575697e-01 -7.50914454e-01 -1.21642816e+00 -2.19244942e-01
-5.54952860e-01 2.57709086e-01 4.71125841e-01 1.09011173e+00
3.29474777e-01 1.58296507e-02 -1.49603501e-01 -2.29113340e-01
-6.77080691e-01 -1.29015815e+00 -7.57647276e-01 8.66249025e-01
-5.66042244e-01 -8.75313222e-01 -2.26637870e-01 -3.62338811e-01] | [9.868945121765137, 8.720837593078613] |
627633a5-c184-42d8-a97a-70b1d54f7a5d | well-calibrated-probabilistic-predictive | 2306.06642 | null | https://arxiv.org/abs/2306.06642v1 | https://arxiv.org/pdf/2306.06642v1.pdf | Well-Calibrated Probabilistic Predictive Maintenance using Venn-Abers | When using machine learning for fault detection, a common problem is the fact that most data sets are very unbalanced, with the minority class (a fault) being the interesting one. In this paper, we investigate the usage of Venn-Abers predictors, looking specifically at the effect on the minority class predictions. A key property of Venn-Abers predictors is that they output well-calibrated probability intervals. In the experiments, we apply Venn-Abers calibration to decision trees, random forests and XGBoost models, showing how both overconfident and underconfident models are corrected. In addition, the benefit of using the valid probability intervals produced by Venn-Abers for decision support is demonstrated. When using techniques producing opaque underlying models, e.g., random forest and XGBoost, each prediction will consist of not only the label, but also a valid probability interval, where the width is an indication of the confidence in the estimate. Adding Venn-Abers on top of a decision tree allows inspection and analysis of the model, to understand both the underlying relationship, and finding out in which parts of feature space that the model is accurate and/or confident. | ['Cecilia Sönströd', 'Tuwe Löfström', 'Ulf Johansson'] | 2023-06-11 | null | null | null | null | ['fault-detection'] | ['miscellaneous'] | [-1.11947790e-01 4.90423024e-01 -4.86623645e-01 -7.26535201e-01
-2.35327661e-01 -1.82928354e-01 3.59233260e-01 5.21765709e-01
9.64668319e-02 1.23366451e+00 -6.88662902e-02 -6.05604947e-01
-7.86893368e-01 -9.73654747e-01 -5.99682748e-01 -6.24285460e-01
-5.05537316e-02 9.21377182e-01 3.82232100e-01 2.07454666e-01
4.76842225e-01 3.13350707e-01 -1.71574998e+00 5.11241317e-01
1.00914931e+00 1.22065449e+00 -2.58207440e-01 3.34511012e-01
-2.71506369e-01 7.64445543e-01 -9.78675365e-01 -5.05382240e-01
1.16038613e-01 -4.02113557e-01 -7.19725549e-01 -2.56716818e-01
1.80624157e-01 -1.30802065e-01 6.49790347e-01 7.83073485e-01
-2.32583769e-02 -5.47687769e-01 1.00345492e+00 -1.77111387e+00
7.11075515e-02 1.03894937e+00 -5.63940644e-01 4.46183905e-02
4.12641644e-01 -1.54082522e-01 9.45009351e-01 -6.33989692e-01
5.22108138e-01 1.16119385e+00 1.08134913e+00 -1.60690367e-01
-1.26698220e+00 -7.71953464e-01 2.94685699e-02 3.20755243e-01
-1.19308174e+00 -1.00135200e-01 2.39531994e-01 -7.37964272e-01
6.26238585e-01 3.90365571e-01 6.94738150e-01 6.18063450e-01
6.20962203e-01 2.58602917e-01 1.41277719e+00 -7.66526341e-01
6.60741985e-01 6.32609308e-01 4.37041909e-01 3.13568801e-01
7.32854128e-01 3.20917308e-01 -5.82788944e-01 -5.51627159e-01
4.64405537e-01 1.71487734e-01 -1.93664193e-01 -4.02520746e-01
-7.78351724e-01 8.52443993e-01 4.59371328e-01 2.33393699e-01
-4.40923810e-01 -2.71617267e-02 1.89193249e-01 2.48546749e-01
6.23787820e-01 4.37495768e-01 -6.75740063e-01 1.25142947e-01
-1.32165945e+00 1.12114340e-01 8.01369429e-01 6.26780093e-01
8.00686061e-01 -3.42573732e-01 -6.91525042e-02 7.13834763e-01
4.31421191e-01 -4.06336822e-02 4.32327896e-01 -6.73380792e-01
-5.17050847e-02 7.82247782e-01 2.24486887e-01 -6.73049986e-01
-4.41700071e-01 -6.96161568e-01 -5.72493851e-01 7.17001021e-01
7.30905354e-01 -1.02932015e-02 -9.53706980e-01 1.37459552e+00
3.74704927e-01 -1.76712856e-01 -1.73409179e-01 4.56697315e-01
2.52583623e-01 2.66042203e-01 1.20180823e-01 -2.82491624e-01
1.19577730e+00 -6.27599359e-01 -5.32521248e-01 -1.87926143e-01
5.75896859e-01 -4.91845697e-01 6.92510366e-01 7.04833388e-01
-4.35035199e-01 -5.66072941e-01 -1.06279707e+00 6.78575993e-01
-2.57000595e-01 -1.85892612e-01 5.40678620e-01 6.88037753e-01
-6.34036303e-01 1.11354792e+00 -6.90515459e-01 -3.11161995e-01
1.40395418e-01 1.13124639e-01 -3.77688587e-01 -2.28874519e-01
-1.16288292e+00 1.23912227e+00 5.35747051e-01 -1.54402435e-01
-4.96706247e-01 -6.08194768e-01 -4.51578766e-01 1.70308962e-01
2.74390519e-01 -5.19326270e-01 9.83798862e-01 -8.73313487e-01
-6.35186911e-01 4.27043468e-01 2.60529947e-02 -6.54994845e-01
1.02197194e+00 -7.90028647e-02 -4.33713675e-01 -2.78305233e-01
3.93104613e-01 2.95635074e-01 5.31654537e-01 -1.31888247e+00
-9.34428930e-01 -6.06418967e-01 -2.85328776e-01 -3.21961790e-01
3.11598092e-01 -1.56385243e-01 3.69745106e-01 -3.25795710e-01
5.36949992e-01 -4.74349588e-01 -2.49715969e-01 -2.36857355e-01
-6.76690102e-01 -2.79437423e-01 4.33070302e-01 -6.95846260e-01
1.38431048e+00 -1.94021988e+00 -7.11491466e-01 7.63925970e-01
1.26022965e-01 -3.46809953e-01 7.68599451e-01 5.16200066e-01
-3.30432415e-01 2.93084353e-01 -1.86769471e-01 2.39132494e-01
-1.82177022e-01 3.78051460e-01 -2.00634480e-01 4.51810628e-01
2.22623780e-01 6.02344535e-02 -5.48994720e-01 -4.91800904e-01
1.02468520e-01 1.42356262e-01 -1.88577965e-01 7.58272409e-02
-3.59088331e-02 9.50637013e-02 -4.44647670e-02 6.58742487e-01
6.70005381e-01 -1.94869146e-01 2.17670470e-01 1.55250693e-03
-1.71653911e-01 5.28375149e-01 -1.38123441e+00 4.92044121e-01
-2.23757401e-01 4.42644924e-01 -4.74238873e-01 -8.79480600e-01
1.54344213e+00 1.74208328e-01 3.11583485e-02 -2.83840716e-01
-1.20232888e-01 5.11832714e-01 6.16224343e-03 -1.31427720e-01
1.97200149e-01 -4.47677940e-01 1.70848683e-01 4.65750039e-01
7.79225081e-02 1.34114427e-02 7.53135905e-02 -1.88572243e-01
1.01618016e+00 6.36953413e-02 8.52302611e-01 -5.28876960e-01
2.42822268e-03 2.43168324e-01 6.89457536e-01 8.16150844e-01
1.75441831e-01 6.38466895e-01 1.05027127e+00 -5.14876068e-01
-1.01300025e+00 -9.41238284e-01 -8.05976808e-01 7.80067623e-01
-1.82832733e-01 -2.87319005e-01 -4.02052552e-01 -9.28390980e-01
4.54149723e-01 1.27335382e+00 -8.19420457e-01 -1.57342374e-01
4.72438484e-02 -7.05905497e-01 2.16879457e-01 5.35153568e-01
1.96232125e-02 -7.20593452e-01 -8.43837142e-01 2.09784836e-01
1.38625979e-01 -2.53674150e-01 4.54665542e-01 9.13701355e-01
-1.25706017e+00 -1.28791404e+00 -3.61701518e-01 -1.72813535e-01
5.15197694e-01 -1.80027708e-01 1.54429829e+00 2.89863259e-01
1.48845047e-01 -4.09030765e-01 -5.24986684e-01 -6.71170413e-01
-6.30185306e-01 -2.60622203e-01 -1.76497072e-01 -2.64348269e-01
4.55518365e-01 -5.60282767e-01 -2.18768448e-01 7.74515331e-01
-4.13339019e-01 -1.21427916e-01 4.97645855e-01 9.83726561e-01
6.65146232e-01 4.16352451e-01 7.01286137e-01 -1.41020250e+00
2.99537301e-01 -6.92590654e-01 -4.27033633e-01 4.26917344e-01
-1.42289877e+00 1.15659080e-01 3.68377090e-01 -1.42869860e-01
-6.33764386e-01 -3.31328899e-01 6.34675194e-03 -1.89625934e-01
-3.32580328e-01 7.34008908e-01 -4.27898541e-02 4.15742666e-01
1.04441607e+00 -4.09294546e-01 -7.32061081e-03 -4.81691122e-01
-2.15221867e-01 8.58621001e-01 2.95460463e-01 -3.20939153e-01
3.23009610e-01 1.04763851e-01 6.14956371e-04 -3.11856538e-01
-7.98618257e-01 -3.24950129e-01 -6.59499168e-01 -3.67518693e-01
2.84356743e-01 -4.85477030e-01 -2.21104860e-01 8.75152647e-02
-7.59727299e-01 -1.74814329e-01 -5.14264166e-01 6.38633668e-01
-4.16632861e-01 -1.06710389e-01 -1.40671283e-01 -1.22383177e+00
-1.88779496e-02 -8.85290086e-01 5.44012427e-01 1.37493774e-01
-7.68848240e-01 -7.97300458e-01 -9.79522020e-02 1.71725944e-01
1.29506618e-01 4.32421297e-01 1.20100164e+00 -1.29964650e+00
-2.36353427e-02 -6.83808863e-01 -4.21922691e-02 3.31775814e-01
-8.02983865e-02 4.87190902e-01 -1.09633720e+00 2.26821769e-02
3.35403904e-02 2.23019924e-02 9.00469601e-01 5.07309914e-01
1.00310493e+00 -2.93610692e-01 -5.69246531e-01 1.46909565e-01
1.25078011e+00 2.98520982e-01 6.34704471e-01 7.18277156e-01
-2.46966556e-02 1.03327465e+00 1.06734598e+00 4.76311713e-01
1.79648846e-01 5.79869986e-01 6.53601646e-01 -4.40348834e-02
1.88787431e-01 -3.33748698e-01 8.56053680e-02 -3.25701088e-02
3.69328745e-02 -3.62597629e-02 -1.17475629e+00 3.64456177e-01
-1.72089708e+00 -7.49429524e-01 -5.21312594e-01 2.54128289e+00
5.79071701e-01 6.92864776e-01 2.65649378e-01 8.53332400e-01
9.76410985e-01 -3.76991570e-01 -3.86000186e-01 -5.48292875e-01
1.05842110e-02 -6.45396635e-02 4.53711212e-01 4.21612144e-01
-7.56026387e-01 1.03809595e-01 6.36286592e+00 5.75397730e-01
-7.81142294e-01 -2.66712427e-01 9.38247204e-01 2.60872215e-01
-3.19540739e-01 3.82432163e-01 -8.04372549e-01 6.06130242e-01
9.53349411e-01 -8.49970505e-02 -1.67930752e-01 1.28920078e+00
7.62786344e-03 -7.97354519e-01 -1.28168380e+00 3.06300342e-01
-3.09176296e-01 -1.01367521e+00 -2.66798645e-01 1.83018386e-01
4.73241001e-01 -3.83856148e-01 -5.03601968e-01 -5.82022499e-03
5.48525453e-01 -1.13072288e+00 1.08108699e+00 6.22506618e-01
7.91671395e-01 -8.71663153e-01 1.52528167e+00 4.72365767e-01
-3.79805952e-01 -2.10255861e-01 -3.88095558e-01 -3.15796584e-01
-1.04800403e-01 1.54294550e+00 -1.41656220e+00 6.69912279e-01
1.03096652e+00 4.04970974e-01 -6.96663916e-01 1.41581285e+00
-5.60148120e-01 8.73086989e-01 -3.93989444e-01 7.40685090e-02
-2.70324230e-01 2.19527874e-02 8.20209756e-02 1.08719623e+00
5.91512859e-01 -4.41594481e-01 6.81885034e-02 7.88615048e-01
5.61100602e-01 -7.73780122e-02 -3.90367061e-01 4.61992204e-01
7.00062573e-01 8.35630655e-01 -8.55052590e-01 -3.34581584e-01
-1.08592339e-01 2.18821123e-01 6.27154186e-02 -3.19712944e-02
-4.40731347e-01 -2.42869258e-01 3.23534817e-01 4.82177496e-01
1.49796546e-01 5.17865956e-01 -9.56480682e-01 -4.34071809e-01
-8.74590799e-02 -7.60808408e-01 6.61089957e-01 -9.40811038e-01
-1.48615396e+00 6.87887251e-01 1.55738384e-01 -1.20104957e+00
-5.85449040e-01 -5.76642096e-01 -7.42833912e-01 1.10759234e+00
-1.04248333e+00 -5.76198816e-01 -4.71501678e-01 1.59413472e-01
-4.31530513e-02 6.79221377e-02 7.39520192e-01 -1.85877204e-01
-2.90821284e-01 4.30385888e-01 6.59605563e-02 -1.74992442e-01
8.76034796e-01 -1.43479049e+00 5.54540083e-02 5.32443702e-01
3.97599190e-02 2.80971497e-01 1.23825896e+00 -9.20374095e-01
-1.01670131e-01 -8.04393291e-01 9.24268126e-01 -3.54410112e-01
4.84523207e-01 4.53523956e-02 -1.19885945e+00 6.81372643e-01
-3.09421957e-01 -1.18881620e-01 8.65692556e-01 6.75611556e-01
-2.95417815e-01 -2.22143129e-01 -1.62501943e+00 -1.42952055e-01
4.77696806e-01 -4.70835716e-02 -7.40766168e-01 3.10771972e-01
3.44847023e-01 -1.21909812e-01 -9.31833088e-01 6.70455158e-01
8.39420915e-01 -1.66346455e+00 5.65876782e-01 -6.09828293e-01
6.25131011e-01 -3.15334737e-01 -2.08900571e-01 -1.44831693e+00
-3.65763098e-01 2.50561684e-01 1.97746783e-01 1.33249962e+00
8.55536640e-01 -7.36198604e-01 8.43155026e-01 4.27863926e-01
3.28592211e-02 -9.14198041e-01 -8.74795914e-01 -7.47073233e-01
-5.17591052e-02 -6.11722529e-01 8.09483349e-01 9.52662289e-01
4.97014336e-02 1.19965166e-01 -2.29766995e-01 1.54940650e-01
5.46227813e-01 2.46975422e-01 7.71743357e-01 -1.86654019e+00
-3.30392450e-01 -1.22876383e-01 -6.79635644e-01 -1.73419639e-01
-3.60849947e-01 -5.08565784e-01 1.18144982e-01 -1.61659527e+00
1.21188067e-01 -8.79589260e-01 -2.70816594e-01 4.76106137e-01
-2.66198248e-01 -5.93413375e-02 -7.56117105e-02 4.20739800e-01
-3.34825106e-02 -1.19139522e-01 5.12274504e-01 2.32053980e-01
-1.49030760e-01 5.98880827e-01 -6.39360964e-01 9.80701566e-01
8.13517272e-01 -7.66817153e-01 -5.88203929e-02 1.41471788e-01
2.91530699e-01 2.38716394e-01 4.45694357e-01 -1.08625555e+00
-2.89988276e-02 -1.32931516e-01 9.51122344e-01 -8.22989464e-01
-1.28266007e-01 -9.40857768e-01 7.86792576e-01 7.67333210e-01
-4.00401860e-01 8.60718787e-02 -8.88488218e-02 4.77594227e-01
-2.44657636e-01 -7.47562110e-01 7.78817475e-01 -1.43658474e-01
-3.45203042e-01 -4.61375415e-01 -1.42605767e-01 -3.97403419e-01
9.68459964e-01 -5.70995569e-01 -4.23156202e-01 -3.00728977e-01
-7.75135279e-01 2.14111581e-01 6.46043360e-01 -1.11971751e-01
3.17720503e-01 -9.49527979e-01 -7.83135116e-01 2.87053555e-01
3.24351490e-01 -3.58417369e-02 -1.25135794e-01 7.91866362e-01
-3.74549478e-01 1.57470062e-01 -4.38623369e-01 -7.47217715e-01
-1.20603526e+00 4.99996781e-01 2.95415252e-01 -3.82944554e-01
-3.05795938e-01 8.46092820e-01 -1.06643677e-01 -3.36252272e-01
1.32940367e-01 -4.05700684e-01 -3.88263792e-01 3.29426020e-01
4.09777075e-01 7.53006339e-01 3.82796168e-01 -1.71539143e-01
-5.69305182e-01 1.55483365e-01 1.55411243e-01 3.59911397e-02
1.31411564e+00 1.31923303e-01 -3.65320027e-01 9.05797362e-01
3.65814298e-01 7.24771544e-02 -1.05575418e+00 1.07299432e-01
5.36191225e-01 -6.74528122e-01 -7.70466179e-02 -1.37619293e+00
-6.54091179e-01 8.25736523e-01 5.13087034e-01 7.63487697e-01
9.40077305e-01 6.68134987e-02 -3.54722589e-01 -1.91242337e-01
6.56186759e-01 -8.48274708e-01 -5.28811634e-01 3.60447057e-02
9.96088445e-01 -1.10191512e+00 3.43012780e-01 -5.42400658e-01
-7.15996146e-01 1.37562346e+00 5.14689445e-01 -9.99022275e-02
6.11153364e-01 2.19255775e-01 -5.04273325e-02 -1.21856861e-01
-9.17135239e-01 2.00838760e-01 7.10823759e-02 6.94737256e-01
4.84798104e-01 4.03279126e-01 -3.58461767e-01 8.33814323e-01
-6.67825997e-01 1.73541322e-01 4.66854125e-01 5.77828348e-01
-7.02740908e-01 -7.74252713e-01 -8.60794485e-01 1.16951251e+00
-4.34309840e-01 1.72035873e-01 -5.35547674e-01 9.59281445e-01
3.97831321e-01 1.17415202e+00 2.49062777e-01 -5.82451165e-01
3.94423485e-01 3.80649358e-01 -5.72672561e-02 -5.13560057e-01
-4.60378498e-01 -6.14063367e-02 4.68203515e-01 -3.53962421e-01
2.57495437e-02 -7.08486140e-01 -1.00937045e+00 -2.90698916e-01
-9.05276418e-01 6.15359664e-01 8.25141311e-01 8.83941650e-01
-2.28877813e-02 5.19721925e-01 6.04528904e-01 -3.76466393e-01
-7.84739614e-01 -1.33221078e+00 -8.58125865e-01 5.01941405e-02
4.26211730e-02 -1.02445483e+00 -7.48308778e-01 -3.81581336e-01] | [8.43179988861084, 4.5119099617004395] |
e3944391-467b-497a-8428-88c840035be8 | weakly-supervised-person-search-with-region | 2109.06109 | null | https://arxiv.org/abs/2109.06109v1 | https://arxiv.org/pdf/2109.06109v1.pdf | Weakly Supervised Person Search with Region Siamese Networks | Supervised learning is dominant in person search, but it requires elaborate labeling of bounding boxes and identities. Large-scale labeled training data is often difficult to collect, especially for person identities. A natural question is whether a good person search model can be trained without the need of identity supervision. In this paper, we present a weakly supervised setting where only bounding box annotations are available. Based on this new setting, we provide an effective baseline model termed Region Siamese Networks (R-SiamNets). Towards learning useful representations for recognition in the absence of identity labels, we supervise the R-SiamNet with instance-level consistency loss and cluster-level contrastive loss. For instance-level consistency learning, the R-SiamNet is constrained to extract consistent features from each person region with or without out-of-region context. For cluster-level contrastive learning, we enforce the aggregation of closest instances and the separation of dissimilar ones in feature space. Extensive experiments validate the utility of our weakly supervised method. Our model achieves the rank-1 of 87.1% and mAP of 86.0% on CUHK-SYSU benchmark, which surpasses several fully supervised methods, such as OIM and MGTS, by a clear margin. More promising performance can be reached by incorporating extra training data. We hope this work could encourage the future research in this field. | ['Changhu Wang', 'Yi Yang', 'Nong Sang', 'Changxin Gao', 'Zehuan Yuan', 'Dongdong Yu', 'Kai Su', 'Chuchu Han'] | 2021-09-13 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Han_Weakly_Supervised_Person_Search_With_Region_Siamese_Networks_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Han_Weakly_Supervised_Person_Search_With_Region_Siamese_Networks_ICCV_2021_paper.pdf | iccv-2021-1 | ['person-search'] | ['computer-vision'] | [-1.07101858e-01 -4.08612676e-02 -3.79632086e-01 -7.40638256e-01
-8.13275158e-01 -6.29173100e-01 7.80973911e-01 3.18643972e-02
-6.29495025e-01 7.80328631e-01 6.54806048e-02 3.17385852e-01
-5.67509085e-02 -5.30369043e-01 -7.52272487e-01 -7.89223611e-01
-1.32908002e-01 7.00429738e-01 3.74049395e-02 -3.60396393e-02
-9.63344648e-02 3.65660399e-01 -1.43504512e+00 1.05460718e-01
9.32653666e-01 1.09722793e+00 -1.96597919e-01 1.22932546e-01
1.85717031e-01 5.57884037e-01 -3.96421552e-01 -7.25127995e-01
6.89949214e-01 -3.29254359e-01 -8.52563083e-01 1.90267742e-01
9.52172697e-01 -1.86230138e-01 -3.89665544e-01 1.09551787e+00
5.05898237e-01 3.40068609e-01 6.65532112e-01 -1.23175204e+00
-8.22565019e-01 4.46035117e-01 -7.42940605e-01 1.26724511e-01
3.46869946e-01 6.44122958e-02 1.24944127e+00 -1.02823615e+00
5.59502125e-01 1.09572864e+00 6.37672067e-01 7.27810621e-01
-1.18476677e+00 -8.20729733e-01 4.81279105e-01 2.15540364e-01
-1.91836274e+00 -4.68368083e-01 6.59696758e-01 -2.25523815e-01
5.99827945e-01 3.22387159e-01 2.08613560e-01 1.13224709e+00
-6.74076140e-01 1.03725481e+00 1.16681886e+00 -4.06454176e-01
1.38174951e-01 3.71940792e-01 3.83326381e-01 7.74657130e-01
2.53023982e-01 1.79518938e-01 -7.27330387e-01 -2.73478419e-01
7.66575217e-01 8.28397945e-02 -1.98572531e-01 -6.60793245e-01
-1.07098675e+00 7.38671482e-01 9.15681899e-01 1.60632521e-01
-7.38470852e-02 -4.08056565e-02 3.47041726e-01 2.21767977e-01
5.16189575e-01 3.31338763e-01 -2.36279711e-01 2.82520622e-01
-1.17954409e+00 4.08575177e-01 5.72614074e-01 1.22987604e+00
7.01122820e-01 -2.64099330e-01 -2.66943425e-01 1.08925974e+00
8.40356573e-02 4.26657736e-01 2.99885839e-01 -7.44746149e-01
4.94270951e-01 6.37904108e-01 2.29148999e-01 -7.41466999e-01
-2.05906302e-01 -8.09509635e-01 -9.47275877e-01 -4.73038703e-02
8.46655250e-01 6.15209434e-03 -9.23628032e-01 1.76041496e+00
3.48661393e-01 4.17802066e-01 -8.28536823e-02 1.31259620e+00
7.89989471e-01 1.91452980e-01 6.32836996e-03 1.07408203e-01
1.50275469e+00 -1.25035501e+00 -3.22317302e-01 -1.26442939e-01
5.77523768e-01 -4.23438102e-01 8.60213757e-01 1.06339939e-01
-9.99374449e-01 -5.81417859e-01 -9.18175638e-01 -6.64894879e-02
-4.28334773e-01 4.11359906e-01 7.43358850e-01 7.63518810e-01
-9.04472053e-01 4.29686546e-01 -5.80327868e-01 -4.43691790e-01
7.66179025e-01 5.23399353e-01 -7.39367902e-01 -2.45566279e-01
-1.15249455e+00 6.18466616e-01 1.54378504e-01 2.31782928e-01
-6.65792704e-01 -6.34207129e-01 -8.95300269e-01 -4.35297377e-02
5.35199165e-01 -5.11273921e-01 8.52437019e-01 -9.40422177e-01
-8.63896132e-01 1.38292086e+00 -4.67993230e-01 -5.64827502e-01
8.70311320e-01 -2.30868712e-01 -3.07384729e-01 8.09432343e-02
4.73116159e-01 5.65346301e-01 6.72614336e-01 -1.39403141e+00
-7.09060252e-01 -6.61107600e-01 1.10961691e-01 3.12129706e-01
-4.18463290e-01 2.19591767e-01 -8.66390705e-01 -6.57801867e-01
1.46634296e-01 -1.04832935e+00 -2.41168082e-01 1.83981478e-01
-4.35581237e-01 -5.76530635e-01 3.62102628e-01 -6.01849735e-01
9.61944044e-01 -1.99868715e+00 -5.09700105e-02 5.54469883e-01
1.60416424e-01 1.30854577e-01 -1.50188105e-02 1.01675987e-02
1.00794166e-01 -3.08780670e-02 -2.46445477e-01 -6.92902088e-01
1.11541070e-01 1.90495588e-02 -1.94189906e-01 7.10800231e-01
2.16384932e-01 9.72283185e-01 -8.08452189e-01 -5.51928699e-01
-1.45542949e-01 2.30556831e-01 -4.52114224e-01 9.82378870e-02
1.17129080e-01 4.48551089e-01 -3.02261859e-01 1.01915383e+00
7.18605757e-01 -4.47771072e-01 -2.97930948e-02 5.72518120e-03
1.80777401e-01 9.30246245e-03 -1.34690630e+00 1.68667090e+00
-1.85144976e-01 2.20510915e-01 6.38297424e-02 -1.05265749e+00
8.52346599e-01 -7.40086194e-03 3.20858657e-01 -7.45383143e-01
-2.43388489e-01 2.68237829e-01 -1.30898908e-01 -1.26464382e-01
3.35543633e-01 1.25457034e-01 -2.71231323e-01 2.35157996e-01
3.95118892e-02 6.97504580e-01 1.96326122e-01 2.50458419e-01
6.84279323e-01 1.77025333e-01 1.64263695e-01 -4.07290190e-01
7.86504865e-01 -1.79717883e-01 6.11335635e-01 1.00334311e+00
-4.52658892e-01 8.40542912e-01 1.54298857e-01 -5.78701735e-01
-9.71491933e-01 -1.05185187e+00 -4.48285878e-01 1.46209252e+00
3.89931381e-01 -1.56753659e-01 -6.76960230e-01 -1.10297179e+00
2.30030254e-01 2.81889662e-02 -7.72249579e-01 1.20778121e-01
-6.86686635e-01 -7.51628697e-01 6.89582050e-01 7.01894581e-01
6.60237312e-01 -7.38742292e-01 2.20611155e-01 -1.11811511e-01
-6.81131408e-02 -1.25319564e+00 -9.17561114e-01 4.61429209e-02
-4.22432840e-01 -1.04689837e+00 -1.18230665e+00 -1.14382422e+00
1.11893153e+00 2.09667325e-01 9.19038653e-01 2.72126466e-01
-2.10561231e-01 -4.39746603e-02 -2.65315920e-01 -4.56260182e-02
3.37842971e-01 2.66972542e-01 3.54235560e-01 2.98305094e-01
6.53831720e-01 -3.11005294e-01 -7.16367424e-01 7.06176579e-01
-3.21892560e-01 -3.49213481e-01 4.88449365e-01 1.06548119e+00
6.56198323e-01 -4.05697748e-02 5.78140080e-01 -9.48273718e-01
1.98034734e-01 -3.59605879e-01 -4.99727666e-01 4.63756680e-01
-5.09954691e-01 2.03588922e-02 6.53131604e-01 -4.38765109e-01
-9.56182063e-01 2.59937584e-01 1.68601140e-01 -4.63012695e-01
-2.29559943e-01 4.80074435e-02 -3.88502717e-01 -2.69989341e-01
7.47265458e-01 3.54319394e-01 -2.64114588e-01 -5.36162376e-01
2.47043684e-01 7.71416128e-01 8.46480072e-01 -9.03482735e-01
1.06447423e+00 7.63442516e-01 -2.68672734e-01 -5.91794193e-01
-1.10919428e+00 -9.08323526e-01 -7.67333210e-01 6.34446740e-03
6.88705742e-01 -1.21883833e+00 -6.82031393e-01 3.03203434e-01
-7.37847567e-01 -1.54751584e-01 -1.06429741e-01 3.68237972e-01
-1.46086544e-01 4.62078601e-01 -5.29619098e-01 -9.28449392e-01
-2.43811071e-01 -8.48199368e-01 1.10891783e+00 4.39260542e-01
-2.30628580e-01 -8.46287072e-01 -2.43658975e-01 7.96057403e-01
2.74374541e-02 6.68042302e-02 2.76241541e-01 -1.15175104e+00
-6.08437419e-01 -4.48755860e-01 -5.16921043e-01 2.37646490e-01
-6.92893863e-02 -5.95593870e-01 -1.16893435e+00 -4.26046342e-01
-4.55077678e-01 -6.63183868e-01 1.21305382e+00 1.40943304e-01
1.30845988e+00 -2.68113434e-01 -3.81832600e-01 8.54413867e-01
1.18817186e+00 -2.59264588e-01 3.36155683e-01 2.99220145e-01
9.05577302e-01 7.67542422e-01 6.11227036e-01 3.25152665e-01
5.27476788e-01 9.69549954e-01 7.46206567e-02 -2.87251174e-01
-9.20963883e-02 -4.49573904e-01 3.04701459e-02 7.52704963e-02
-4.43568766e-01 1.05498925e-01 -8.37543905e-01 6.37006700e-01
-2.04252458e+00 -1.08958364e+00 2.11753957e-02 2.36926579e+00
8.66580665e-01 1.93751883e-02 4.72293556e-01 -1.19723655e-01
8.88941467e-01 6.55389801e-02 -5.48888028e-01 2.33136997e-01
-3.55373472e-01 7.95472339e-02 5.58544636e-01 5.74414730e-01
-1.59292531e+00 1.09854782e+00 5.27311945e+00 9.55938816e-01
-6.47666693e-01 1.17478572e-01 7.81474233e-01 -3.01174849e-01
1.58756331e-01 -1.82610005e-01 -1.31751966e+00 6.12615705e-01
4.23247188e-01 1.73395038e-01 3.65975648e-01 1.04883027e+00
-3.69472951e-02 2.62870174e-03 -1.40695167e+00 1.27065921e+00
3.50902945e-01 -1.06225944e+00 -1.82843193e-01 2.76033103e-01
9.46489692e-01 -2.91520923e-01 1.06519260e-01 5.03892422e-01
2.37728134e-01 -1.25192463e+00 5.16178727e-01 3.49357069e-01
6.85986042e-01 -8.67625296e-01 9.16302741e-01 3.40468377e-01
-1.35391557e+00 -8.12333543e-03 -5.00851154e-01 5.33163827e-03
-8.92278180e-02 3.22451383e-01 -5.65139234e-01 5.80377877e-01
8.20074439e-01 6.75214946e-01 -7.54942775e-01 1.16996527e+00
-1.37015939e-01 4.55820799e-01 -4.90761548e-01 6.06804863e-02
2.97418237e-01 -1.76953524e-01 1.41565830e-01 1.25672448e+00
-1.45931333e-01 -2.38465723e-02 5.54447234e-01 8.04524362e-01
-4.04812872e-01 2.13195443e-01 -3.37503374e-01 4.57001716e-01
5.37902594e-01 1.13349068e+00 -5.14154077e-01 -4.41703588e-01
-3.67474377e-01 1.30269849e+00 6.73351467e-01 4.53093410e-01
-8.51027668e-01 -1.75627042e-02 5.83857834e-01 2.00823292e-01
2.41129473e-01 1.25320777e-01 -4.36743826e-01 -1.29160058e+00
4.89853442e-01 -7.42628813e-01 8.20498466e-01 -2.41735086e-01
-1.88187230e+00 4.82737750e-01 -4.65428717e-02 -1.08098710e+00
-8.43967646e-02 -4.15880769e-01 -4.59177881e-01 1.04353440e+00
-1.65761483e+00 -1.57644248e+00 -2.76276529e-01 7.37919927e-01
3.59102935e-01 -5.16041875e-01 6.51699781e-01 3.83259624e-01
-6.88164771e-01 1.31925273e+00 -6.39478239e-05 7.97924042e-01
9.65279579e-01 -1.29852867e+00 2.28332713e-01 8.06951284e-01
3.87778997e-01 9.49811578e-01 4.01555330e-01 -5.69121718e-01
-8.96800578e-01 -1.13373399e+00 1.15016317e+00 -6.84658647e-01
4.02073562e-01 -7.26415873e-01 -9.41712976e-01 6.00724757e-01
-2.05095395e-01 5.84421992e-01 8.48061144e-01 5.51821470e-01
-7.33408988e-01 -2.50640899e-01 -1.27441287e+00 5.43571711e-01
1.39618909e+00 -6.48486555e-01 -4.30071414e-01 6.26653254e-01
3.42724532e-01 -2.54204541e-01 -7.77541459e-01 2.70167202e-01
5.06932557e-01 -6.94503605e-01 1.24369550e+00 -7.47535348e-01
1.02805272e-01 -3.44342917e-01 -1.87373767e-03 -8.22355032e-01
-4.06908184e-01 -3.98327112e-01 -1.35426268e-01 1.45700562e+00
4.68418330e-01 -5.32746375e-01 1.26459980e+00 1.01487350e+00
2.88591087e-01 -7.84211218e-01 -1.02740967e+00 -1.24160910e+00
2.06058577e-01 -4.91420180e-02 5.34773231e-01 1.11991644e+00
-2.44784132e-02 1.66330472e-01 -5.87430418e-01 4.92730200e-01
1.08916223e+00 1.68729633e-01 8.02405953e-01 -1.16449142e+00
-2.67197043e-01 -4.41929579e-01 -4.05065000e-01 -1.13041449e+00
5.09605229e-01 -1.08121204e+00 -1.39179751e-01 -1.07287073e+00
5.43234348e-01 -7.67886579e-01 -4.03104246e-01 5.15321255e-01
-4.45205867e-01 6.24801278e-01 1.45520955e-01 5.04115701e-01
-1.04011881e+00 4.57281619e-01 9.23464775e-01 -2.05367595e-01
-9.35790092e-02 1.39508322e-01 -6.78832412e-01 6.35673761e-01
8.07640791e-01 -2.63549536e-01 -1.93563819e-01 -1.05997503e-01
-2.65672922e-01 -4.40316051e-01 4.87972617e-01 -8.87291431e-01
5.50347447e-01 5.66214003e-05 7.79638946e-01 -4.33654964e-01
4.10712749e-01 -7.37967789e-01 -2.31230095e-01 2.10838363e-01
-5.40429831e-01 -2.67992854e-01 -3.38417321e-01 5.68335354e-01
-2.29225531e-01 -1.48577720e-01 8.06058586e-01 -2.61132270e-01
-8.51437569e-01 6.19543612e-01 4.65098411e-01 2.45082423e-01
1.00128686e+00 -3.96016121e-01 -1.85022041e-01 -2.05484360e-01
-7.58449316e-01 6.07297122e-01 5.88471115e-01 3.25848818e-01
4.26191479e-01 -1.50669229e+00 -9.49704587e-01 2.34793440e-01
5.41479349e-01 2.40742788e-02 1.95033833e-01 5.77382088e-01
-1.73352405e-01 5.08861005e-01 1.25148073e-01 -6.10714197e-01
-1.38085544e+00 5.18828273e-01 4.36234623e-01 -3.03397268e-01
-4.23575759e-01 1.11878538e+00 4.54084903e-01 -6.14971161e-01
5.89694679e-01 2.69398600e-01 -1.11642823e-01 -3.69093493e-02
8.05350006e-01 2.15663731e-01 -1.23398326e-01 -9.17743742e-01
-6.94105923e-01 5.33819854e-01 -4.73252386e-01 -3.60790677e-02
1.02403617e+00 7.48231029e-03 9.26986560e-02 -2.52562445e-02
1.17879581e+00 -7.01657459e-02 -1.37242329e+00 -5.40988982e-01
1.54796645e-01 -6.15633726e-01 -3.33883643e-01 -7.86468148e-01
-9.98921394e-01 5.55298388e-01 5.22339225e-01 -1.85335889e-01
7.15079606e-01 3.22413415e-01 6.59518182e-01 3.75889182e-01
5.11270165e-01 -1.29430139e+00 -1.05963461e-03 2.73601949e-01
7.98215210e-01 -1.75431252e+00 6.49638101e-02 -4.91475999e-01
-6.79004610e-01 5.23001194e-01 9.39975917e-01 -2.72248983e-01
3.57114106e-01 -6.04415312e-02 -3.59242819e-02 9.31344088e-03
-4.04651374e-01 -6.36891663e-01 6.75760627e-01 7.07903504e-01
3.32258701e-01 8.06697607e-02 -1.69809058e-01 7.88572788e-01
-1.87637135e-01 -3.33237648e-01 -1.12356961e-01 5.96433163e-01
-2.75940746e-01 -1.08146846e+00 -6.02980793e-01 4.03820604e-01
-4.32851344e-01 -1.01207197e-01 -6.55257225e-01 6.74175620e-01
3.36702108e-01 8.35109890e-01 -1.74579658e-02 -5.35867959e-02
2.73685098e-01 1.95298061e-01 4.72157627e-01 -5.57914495e-01
-7.20839798e-01 -8.06021467e-02 1.50376379e-01 -3.48470688e-01
-3.04332644e-01 -8.23635161e-01 -1.15276504e+00 -2.19035238e-01
-3.16519678e-01 1.23351663e-01 2.49014705e-01 7.98848867e-01
1.16411850e-01 -1.16262026e-01 5.57010055e-01 -5.45913160e-01
-6.66897893e-01 -8.42276156e-01 -6.61751747e-01 8.27532053e-01
2.51575500e-01 -6.86891615e-01 -3.53412539e-01 1.31862136e-02] | [14.76862621307373, 0.9778971076011658] |
c0463176-fdc5-4002-bff3-678ee46ab522 | can-very-large-pretrained-language-models | 2301.09790 | null | https://arxiv.org/abs/2301.09790v2 | https://arxiv.org/pdf/2301.09790v2.pdf | Can Very Large Pretrained Language Models Learn Storytelling With A Few Examples? | While pre-trained language models can generate individually fluent sentences for automatic story generation, they struggle to generate stories that are coherent, sensible and interesting. Current state-of-the-art (SOTA) story generation models explore using higher-level features such as plots or commonsense knowledge to improve the quality of generated stories. Prompt-based learning using very large pre-trained language models (VLPLMs) such as GPT3 has demonstrated impressive performance even across various NLP tasks. In this paper, we present an extensive study using automatic and human evaluation to compare the story generation capability of VLPLMs to those SOTA models in three different datasets where stories differ in style, register and length. Our results show that VLPLMs generate much higher quality stories than other story generation models, and to a certain extent rival human authors, although preliminary investigation also reveals that they tend to ``plagiarise'' real stories in scenarios that involve world knowledge. | ['Jey Han Lau', 'Trevor Cohn', 'Zhuohan Xie'] | 2023-01-24 | null | null | null | null | ['story-generation'] | ['natural-language-processing'] | [ 3.55144203e-01 5.94293475e-01 8.14460497e-03 1.18411072e-01
-1.03724504e+00 -6.98151588e-01 1.45021963e+00 1.23483703e-01
2.41662621e-01 1.26474059e+00 1.17067957e+00 -9.43629295e-02
1.30065605e-01 -1.15079582e+00 -8.02182555e-01 -5.58126122e-02
3.22289944e-01 6.32815063e-01 -4.19654474e-02 -7.37054825e-01
8.33624542e-01 -6.13622703e-02 -1.31437445e+00 1.06174278e+00
9.04853463e-01 -2.55719144e-02 8.03170279e-02 9.66994643e-01
-6.36400104e-01 1.51990950e+00 -1.35359716e+00 -7.88043559e-01
-3.65771145e-01 -9.32131469e-01 -1.11182511e+00 -4.45924252e-02
3.84637177e-01 -2.01084465e-01 -1.54299796e-01 4.68118846e-01
6.97848380e-01 -1.50234565e-01 1.07567978e+00 -1.24931359e+00
-1.30655932e+00 1.46968758e+00 -3.31287056e-01 1.38072506e-01
1.21285975e+00 4.56930339e-01 1.07188559e+00 -7.79593170e-01
1.28319156e+00 1.53943658e+00 8.54134798e-01 4.50299412e-01
-1.37762153e+00 -4.77934867e-01 -3.64560544e-01 8.84161741e-02
-1.10865343e+00 -4.04031664e-01 9.36043620e-01 -6.97323978e-01
1.32492757e+00 3.24394584e-01 8.36974263e-01 1.84334671e+00
4.49273348e-01 9.07835960e-01 1.20873368e+00 -5.39036095e-01
3.72807570e-02 4.32804585e-01 -3.76375586e-01 3.48882854e-01
2.66476691e-01 -1.44160405e-01 -9.44011629e-01 -3.45838130e-01
7.79625893e-01 -8.48406613e-01 -1.59711674e-01 2.61606872e-01
-1.73946071e+00 1.24084854e+00 9.38521251e-02 7.09509611e-01
-3.74993712e-01 1.56936646e-01 6.32370353e-01 2.38265544e-01
7.18938112e-01 1.36268795e+00 1.65651441e-01 -4.81574386e-01
-1.10069001e+00 8.97486508e-01 1.01728237e+00 1.13964152e+00
1.41648322e-01 2.49662176e-01 -8.87372732e-01 7.79614270e-01
-1.83186293e-01 3.67008269e-01 6.12183571e-01 -7.55075812e-01
7.96527505e-01 5.01579225e-01 1.93285063e-01 -1.45643520e+00
3.50276218e-03 -3.53294492e-01 -6.13774359e-01 6.85608163e-02
2.41243273e-01 -3.13392997e-01 -4.35956687e-01 1.47663498e+00
-3.74058247e-01 -1.52769297e-01 3.45382154e-01 3.98191065e-01
1.34566855e+00 1.17453671e+00 -6.22049533e-02 -2.91384697e-01
1.16140842e+00 -8.42879713e-01 -8.52384090e-01 -4.19226050e-01
8.40771675e-01 -1.22199488e+00 1.61098051e+00 2.83594251e-01
-1.44119990e+00 -4.26411271e-01 -1.08908510e+00 2.07766760e-02
-3.99120659e-01 1.38370916e-01 4.63104784e-01 4.81160015e-01
-1.09852076e+00 7.02589989e-01 1.11051649e-01 -6.62960768e-01
6.07304037e-01 -6.05269253e-01 -1.96175575e-01 5.73267378e-02
-1.35040140e+00 1.20246601e+00 7.20356584e-01 -7.30637908e-01
-7.99328029e-01 -8.41392219e-01 -9.76938963e-01 -3.51222247e-01
3.08526754e-01 -8.98313820e-01 1.36986637e+00 -6.02110744e-01
-1.38865697e+00 1.03473818e+00 -3.53437811e-02 -5.08402050e-01
7.19531059e-01 -4.34089154e-01 -2.17695206e-01 7.76576027e-02
4.79953200e-01 8.68460715e-01 6.36559963e-01 -1.27982688e+00
-1.20028071e-01 6.09993041e-01 1.44777179e-01 1.61996335e-01
-2.55324513e-01 2.70991683e-01 5.54218650e-01 -1.24101055e+00
-5.42973101e-01 -4.18813974e-01 -8.35673884e-02 -4.61192101e-01
-7.08227575e-01 -3.97710115e-01 5.80414593e-01 -8.01905930e-01
1.31765389e+00 -1.43539774e+00 -1.04217476e-03 -3.83738160e-01
2.03312133e-02 -3.02006397e-03 -2.16670275e-01 1.07345510e+00
1.11186534e-01 6.41434550e-01 -8.30531567e-02 -7.68308863e-02
1.39825577e-02 -8.92873928e-02 -6.73024595e-01 -2.03013286e-01
5.92564464e-01 1.25704944e+00 -1.17052150e+00 -8.39097679e-01
6.38979003e-02 2.50131518e-01 -4.33327079e-01 1.16621979e-01
-5.78196824e-01 1.76993608e-01 -2.51455575e-01 2.10770711e-01
-3.35575901e-02 -3.35272223e-01 -4.08305287e-01 3.35768729e-01
-8.91628936e-02 7.19055414e-01 -7.00216353e-01 1.70347881e+00
-4.13245976e-01 1.36584592e+00 -1.03520834e+00 -2.18883425e-01
1.13477635e+00 6.23840511e-01 -3.07543069e-01 -3.82746696e-01
1.55835375e-01 3.23287994e-02 -2.42892683e-01 -7.04110563e-01
1.06631315e+00 -4.59646940e-01 -4.67474848e-01 9.86686587e-01
5.72221838e-02 -9.91371095e-01 5.09845376e-01 5.90617716e-01
1.13466501e+00 2.28087991e-01 5.07687271e-01 -8.57062042e-02
3.24478924e-01 7.00539052e-01 -2.05668762e-01 1.09592485e+00
3.40267003e-01 9.55776632e-01 7.22882271e-01 -1.79266334e-01
-1.55259025e+00 -9.55580950e-01 3.01128000e-01 7.10629284e-01
-2.71860659e-01 -6.13651812e-01 -8.77445996e-01 -3.11150521e-01
-4.46500659e-01 1.94138873e+00 -5.67644477e-01 -3.71104553e-02
-5.85011721e-01 -6.17300749e-01 9.40494299e-01 3.52265716e-01
2.75732368e-01 -1.86036038e+00 -6.57359958e-01 6.36302233e-01
-6.53350472e-01 -9.55284774e-01 -1.56727791e-01 -5.64605892e-01
-3.84020120e-01 -6.31177068e-01 -8.81117642e-01 -4.89087015e-01
2.36609593e-01 5.91453910e-02 1.59477866e+00 -1.26174912e-01
-2.79720545e-01 -6.86922818e-02 -7.75861561e-01 -7.37069011e-01
-1.21644115e+00 8.62724781e-02 -4.55423087e-01 -6.26094759e-01
1.16535854e-02 -5.51070690e-01 2.16288611e-01 -2.44984582e-01
-1.03679514e+00 7.86673009e-01 4.14119661e-01 7.51449227e-01
-8.67869779e-02 9.08080563e-02 7.96357810e-01 -9.96981382e-01
1.41490269e+00 -5.54620683e-01 3.74621391e-01 2.54044503e-01
-2.44080350e-01 -7.97353778e-03 5.45545280e-01 -6.46582663e-01
-1.32580829e+00 -7.18725502e-01 -4.76357378e-02 2.03011483e-01
-1.85416967e-01 6.84035480e-01 1.32494137e-01 6.03338420e-01
1.20259857e+00 2.90785789e-01 -3.36948842e-01 -1.98921338e-02
6.60380900e-01 3.29582483e-01 5.91043413e-01 -4.42866296e-01
1.08404613e+00 2.27872044e-01 -4.91913319e-01 -6.27272427e-01
-9.33881760e-01 1.98783144e-01 -3.41194034e-01 -6.64494336e-01
6.39094114e-01 -8.61095190e-01 1.62682936e-01 3.48015726e-01
-1.76773214e+00 -3.63439918e-01 -7.43266761e-01 2.29796350e-01
-1.02033985e+00 -1.21914662e-01 -6.99569345e-01 -6.54995441e-01
-5.50535619e-01 -4.63330269e-01 8.93314183e-01 1.89981803e-01
-1.41450334e+00 -1.07387412e+00 2.84715295e-01 4.54695582e-01
4.07142788e-01 1.00293493e+00 1.04889524e+00 -6.53940737e-01
-3.35591495e-01 -2.70203054e-01 -2.23534450e-01 -3.32157850e-01
5.05644120e-02 2.50455767e-01 -7.62152016e-01 2.16651261e-01
-1.70090839e-01 -7.29066670e-01 4.98292178e-01 2.34579578e-01
2.58573741e-01 -8.82121623e-01 -1.94070697e-01 -1.27929837e-01
1.15700066e+00 -5.81737645e-02 9.71812785e-01 5.43200016e-01
6.16822243e-01 8.32287610e-01 6.94272220e-01 6.70405686e-01
2.90150732e-01 2.77950674e-01 -1.20717317e-01 8.81372988e-02
-5.08882701e-01 -9.10292983e-01 7.74437308e-01 6.93784595e-01
-3.89597118e-02 -7.58160293e-01 -8.86250675e-01 9.90107059e-01
-1.83614171e+00 -1.65697396e+00 -6.36318862e-01 1.40164745e+00
1.16183257e+00 5.67440212e-01 -1.00517096e-02 1.92927718e-01
6.08886421e-01 6.45876527e-01 7.95773491e-02 -8.63756537e-01
-7.59710848e-01 1.06379859e-01 -1.41486213e-01 2.33801812e-01
-5.28525114e-01 1.14131701e+00 6.91660786e+00 1.24148309e+00
-6.51676536e-01 8.98816288e-02 5.86135983e-01 -2.96274841e-01
-8.48894060e-01 1.01973556e-01 -4.03312862e-01 1.75992683e-01
8.65840077e-01 -9.04325545e-01 4.89551052e-02 7.96230674e-01
4.50458735e-01 -5.31839244e-02 -1.08216572e+00 7.13111162e-01
6.22468352e-01 -2.13262415e+00 6.47300899e-01 -2.13451594e-01
1.15485716e+00 -6.76206350e-01 -2.30835974e-01 4.02507216e-01
6.45616233e-01 -1.42488551e+00 1.27760506e+00 5.86733341e-01
5.59939802e-01 -8.62351477e-01 7.89122462e-01 5.14720500e-01
-6.33190632e-01 4.75775391e-01 -3.70297581e-01 -5.76522648e-01
7.38814294e-01 5.33371985e-01 -1.26137054e+00 3.37125659e-01
2.79724926e-01 7.84930408e-01 -8.83674145e-01 8.13024402e-01
-7.56605208e-01 7.14083910e-01 2.09239453e-01 -5.73726296e-01
3.31259042e-01 3.42963547e-01 9.57974374e-01 1.56287241e+00
6.48908794e-01 1.03272513e-01 -1.60637215e-01 1.44946635e+00
-5.66117615e-02 4.14126784e-01 -1.16337097e+00 -6.50199950e-01
2.38679141e-01 8.68125975e-01 -7.81479955e-01 -6.14245117e-01
1.35623202e-01 8.54256749e-01 1.56174988e-01 -4.40241173e-02
-7.76344121e-01 -3.56710821e-01 -1.02788964e-02 4.98480409e-01
-7.83625692e-02 1.04739636e-01 -7.60009050e-01 -8.25646400e-01
-1.46190971e-01 -9.40689623e-01 3.75970080e-02 -1.60433924e+00
-1.50419557e+00 8.87637675e-01 2.02516288e-01 -1.18986940e+00
-7.75387347e-01 3.67907435e-02 -1.18682027e+00 6.27662718e-01
-8.96216452e-01 -1.49303460e+00 -7.55111948e-02 1.33424014e-01
1.06696689e+00 -4.29446220e-01 8.13174665e-01 -5.01920223e-01
6.06837869e-02 4.17184114e-01 -4.37009901e-01 1.24745183e-01
7.77381539e-01 -1.07646477e+00 8.51045847e-01 7.75825024e-01
3.20535898e-01 2.11035177e-01 1.34913230e+00 -1.04120636e+00
-6.57658458e-01 -1.06733048e+00 1.55402267e+00 -7.68175244e-01
8.65098298e-01 -2.80863345e-01 -8.10064554e-01 3.93176943e-01
9.98289287e-01 -1.13784218e+00 7.72251904e-01 -1.46163508e-01
-3.89591038e-01 9.43492889e-01 -1.09205282e+00 1.05374718e+00
1.08073366e+00 -4.07415241e-01 -1.22018862e+00 6.77866101e-01
1.10596597e+00 -2.46823266e-01 -5.83844244e-01 -3.03063076e-02
2.85146654e-01 -9.25803304e-01 8.29623461e-01 -6.48156345e-01
1.72908199e+00 -7.52055412e-03 3.28705758e-01 -1.54360175e+00
-3.72925043e-01 -1.07171857e+00 1.47220567e-01 1.83529091e+00
7.26113737e-01 -1.28310263e-01 4.58188236e-01 3.38208437e-01
-1.17722668e-01 -3.16045582e-01 -3.90910804e-01 -7.09773362e-01
3.35467309e-01 -5.29015541e-01 6.15554273e-01 1.18996227e+00
4.45730418e-01 8.85454834e-01 -5.39818764e-01 -5.64649403e-01
3.72333497e-01 -1.48499552e-02 9.35356855e-01 -9.57147300e-01
-3.20400387e-01 -7.66378820e-01 -7.66806304e-02 -2.82930940e-01
1.60998538e-01 -8.19145083e-01 1.84244424e-01 -2.03668904e+00
5.84177196e-01 1.16203055e-01 6.10474586e-01 2.52291203e-01
-3.22312474e-01 2.30570182e-01 2.68530846e-01 6.68776855e-02
-2.95372516e-01 3.98877054e-01 1.50722861e+00 -2.48332053e-01
-4.86394346e-01 -3.03161621e-01 -1.03851593e+00 7.21124768e-01
8.13291848e-01 -6.38002396e-01 -4.69971240e-01 -2.84050494e-01
6.60077691e-01 1.37325630e-01 4.85997707e-01 -8.90682340e-01
-1.90008983e-01 -4.86686796e-01 4.28264409e-01 -6.97691023e-01
1.64144203e-01 2.80490339e-01 4.78854150e-01 2.56574631e-01
-8.90902281e-01 7.05872998e-02 3.66417736e-01 1.00116558e-01
-1.55544192e-01 -5.26745558e-01 4.98154491e-01 -4.83205885e-01
-2.82628536e-01 -4.61765379e-01 -9.32137907e-01 4.09040630e-01
9.69263673e-01 -4.79170054e-01 -7.62447894e-01 -1.04570127e+00
-1.76428303e-01 -1.61220238e-01 6.15711093e-01 6.79874301e-01
8.08038712e-01 -1.58660638e+00 -1.65438485e+00 -5.76590002e-01
3.98539752e-01 -1.67291030e-01 -4.74753976e-03 2.15141013e-01
-7.77876198e-01 4.31215078e-01 -3.40302467e-01 -1.89636931e-01
-1.02614701e+00 4.76336807e-01 -4.13254350e-01 -6.80240393e-01
-7.83520401e-01 1.00095785e+00 -4.11809236e-02 -9.66461673e-02
-3.70337695e-01 9.42286476e-02 -4.20859158e-01 2.86267847e-01
8.31564486e-01 3.32549006e-01 -5.46494246e-01 -7.84971058e-01
1.31968200e-01 -2.87876874e-02 -6.41265363e-02 -6.00556552e-01
1.38704610e+00 1.44510776e-01 3.29352133e-02 6.53264701e-01
8.02905560e-01 1.53479621e-01 -7.43045092e-01 -5.65564781e-02
2.97153562e-01 -4.44513977e-01 -3.81676525e-01 -1.02284873e+00
-1.85654193e-01 7.54013777e-01 -4.45250541e-01 3.81888330e-01
5.82208455e-01 3.11387450e-01 1.02465868e+00 1.94542125e-01
3.29552859e-01 -1.02366209e+00 6.87106371e-01 5.63035429e-01
1.78404880e+00 -8.62371504e-01 1.87967584e-01 -2.93378055e-01
-1.15927017e+00 9.93377686e-01 5.52401721e-01 -1.68079749e-01
-1.14677817e-01 3.05479318e-01 -1.12897880e-01 -2.00523496e-01
-1.08716297e+00 1.48441911e-01 2.17548460e-01 7.74468958e-01
6.58288240e-01 5.11238649e-02 -4.46649939e-01 5.95912516e-01
-1.23777354e+00 -3.25211771e-02 1.29083693e+00 8.19497883e-01
-4.32619661e-01 -7.99564600e-01 -6.81419432e-01 3.73526394e-01
-3.42323810e-01 -2.73631960e-01 -1.16250777e+00 8.75053883e-01
-2.75513120e-02 1.16559577e+00 -9.17654410e-02 -2.62886018e-01
1.21092647e-01 1.13681637e-01 4.92966563e-01 -9.94542837e-01
-8.74766409e-01 -1.60314798e-01 8.40659976e-01 1.32702878e-02
-3.88496310e-01 -8.34558427e-01 -1.04833531e+00 -7.84357131e-01
-2.12494329e-01 -1.49012106e-02 1.31439656e-01 9.88101244e-01
6.50780946e-02 6.80650413e-01 1.93656251e-01 -7.62556493e-01
8.94131735e-02 -1.31476557e+00 -2.58403748e-01 5.28178096e-01
-1.84205651e-01 -2.45257348e-01 -1.23044088e-01 4.96485263e-01] | [11.775208473205566, 8.88914680480957] |
9923c1ad-cde7-4a74-bad6-063ab33841a4 | a-personalized-zero-shot-ecg-arrhythmia | 2207.07089 | null | https://arxiv.org/abs/2207.07089v1 | https://arxiv.org/pdf/2207.07089v1.pdf | A Personalized Zero-Shot ECG Arrhythmia Monitoring System: From Sparse Representation Based Domain Adaption to Energy Efficient Abnormal Beat Detection for Practical ECG Surveillance | This paper proposes a low-cost and highly accurate ECG-monitoring system intended for personalized early arrhythmia detection for wearable mobile sensors. Earlier supervised approaches for personalized ECG monitoring require both abnormal and normal heartbeats for the training of the dedicated classifier. However, in a real-world scenario where the personalized algorithm is embedded in a wearable device, such training data is not available for healthy people with no cardiac disorder history. In this study, (i) we propose a null space analysis on the healthy signal space obtained via sparse dictionary learning, and investigate how a simple null space projection or alternatively regularized least squares-based classification methods can reduce the computational complexity, without sacrificing the detection accuracy, when compared to sparse representation-based classification. (ii) Then we introduce a sparse representation-based domain adaptation technique in order to project other existing users' abnormal and normal signals onto the new user's signal space, enabling us to train the dedicated classifier without having any abnormal heartbeat of the new user. Therefore, zero-shot learning can be achieved without the need for synthetic abnormal heartbeat generation. An extensive set of experiments performed on the benchmark MIT-BIH ECG dataset shows that when this domain adaptation-based training data generator is used with a simple 1-D CNN classifier, the method outperforms the prior work by a significant margin. (iii) Then, by combining (i) and (ii), we propose an ensemble classifier that further improves the performance. This approach for zero-shot arrhythmia detection achieves an average accuracy level of 98.2% and an F1-Score of 92.8%. Finally, a personalized energy-efficient ECG monitoring scheme is proposed using the above-mentioned innovations. | ['Moncef Gabbouj', 'Serkan Kiranyaz', 'İlke Adalıoğlu', 'Mert Duman', 'Mehmet Yamaç'] | 2022-07-14 | null | null | null | null | ['sparse-representation-based-classification', 'arrhythmia-detection', 'ecg-classification'] | ['computer-vision', 'medical', 'medical'] | [ 7.52192974e-01 6.46392033e-02 -7.51636326e-02 -1.17074057e-01
-6.60507977e-01 -1.47208139e-01 -1.05165079e-01 2.32790709e-01
-1.05558909e-01 7.90790439e-01 -2.26256181e-03 -1.00165829e-01
-2.18559995e-01 -6.78350449e-01 -3.97635460e-01 -8.39633346e-01
-1.38011247e-01 -4.74030189e-02 -1.92144990e-01 1.12143962e-03
-1.72819197e-01 2.75534570e-01 -1.52120984e+00 4.90551032e-02
1.08220720e+00 1.14834511e+00 4.26851250e-02 5.48807621e-01
5.15705466e-01 2.40477964e-01 -7.91117489e-01 1.07316431e-02
2.40307912e-01 -1.03016794e+00 -2.39851967e-01 5.84979169e-02
-2.84959730e-02 -2.10421592e-01 -1.29786879e-01 7.33044744e-01
1.18882871e+00 -3.33946049e-02 4.33661997e-01 -7.57679760e-01
-1.36699066e-01 4.66313809e-01 -3.12553257e-01 2.60697842e-01
3.69743556e-01 -1.05079405e-01 4.55801815e-01 -7.21729577e-01
3.96451533e-01 1.01509027e-01 1.16430759e+00 5.47540843e-01
-1.24608684e+00 -6.13139331e-01 -3.44622254e-01 1.81178704e-01
-1.61356032e+00 -3.41757357e-01 1.29241884e+00 -2.38228023e-01
6.12682819e-01 6.19216323e-01 1.12831259e+00 1.21651173e+00
1.72677532e-01 3.21069688e-01 8.56468439e-01 -6.41034961e-01
5.09381950e-01 8.75007287e-02 1.66122727e-02 5.72340071e-01
5.76254189e-01 7.64562413e-02 -4.35809433e-01 -3.93475235e-01
5.17127991e-01 3.21781307e-01 -5.02556384e-01 -3.78042847e-01
-1.26689732e+00 4.87016022e-01 1.39942497e-01 8.53798091e-01
-6.28889680e-01 -1.68208092e-01 4.95197982e-01 1.28153995e-01
5.44217646e-01 4.53464419e-01 -2.98601687e-01 -1.85499549e-01
-1.23180294e+00 -1.17161363e-01 8.31599891e-01 5.47582388e-01
5.03406882e-01 5.99283695e-01 -4.61951792e-01 8.27016413e-01
1.84257384e-02 4.21269894e-01 9.05148506e-01 -6.43298745e-01
-2.44186353e-03 5.11340320e-01 9.38601047e-02 -1.18743682e+00
-6.42603159e-01 -1.13271606e+00 -1.27348518e+00 -3.22256207e-01
9.22497809e-02 -5.33195376e-01 -6.39673233e-01 1.57088792e+00
4.08160299e-01 8.54740143e-01 3.19801986e-01 7.99736440e-01
7.67943382e-01 3.52755070e-01 -1.53168768e-01 -8.81839871e-01
1.35875988e+00 -3.58135879e-01 -8.04217696e-01 2.97961272e-02
5.81526935e-01 -2.90950388e-01 8.96562576e-01 5.05159974e-01
-7.29972184e-01 -5.74461579e-01 -1.37914789e+00 5.91957748e-01
3.44762504e-01 4.55942929e-01 4.46760446e-01 1.13115847e+00
-6.74883783e-01 6.87349498e-01 -7.71161675e-01 -4.35451359e-01
4.55813259e-01 2.43962616e-01 -3.95970419e-02 1.30287334e-01
-1.25129163e+00 4.22281951e-01 2.07521394e-01 -6.12853356e-02
-6.70785964e-01 -7.72607565e-01 -7.21466720e-01 1.01359867e-01
7.38535949e-04 -8.16924155e-01 7.81207263e-01 -1.02255166e+00
-1.66830957e+00 5.75030029e-01 -2.12389350e-01 -6.58600092e-01
3.43638003e-01 -9.88793448e-02 -7.01252401e-01 2.48551175e-01
-1.83591507e-02 -1.56174004e-01 1.16167402e+00 -1.02237463e+00
-3.10368478e-01 -3.52602214e-01 -3.82009804e-01 1.80493593e-01
-8.25792134e-01 -4.70015734e-01 -1.61186233e-01 -1.06694841e+00
3.54641318e-01 -9.05360341e-01 -2.57714480e-01 -3.11056346e-01
-2.35113293e-01 3.97264004e-01 8.14254463e-01 -8.13525736e-01
1.49065351e+00 -2.33619618e+00 -2.79260557e-02 5.13798594e-01
2.17864498e-01 2.81848729e-01 3.05950046e-01 5.89728169e-02
-2.84354389e-01 -2.73918122e-01 -5.93543649e-01 -1.54706165e-01
-5.01454890e-01 9.50546265e-02 8.38347822e-02 5.43891370e-01
-1.84815869e-01 7.50343263e-01 -7.64079094e-01 -1.87684581e-01
1.82479352e-01 5.55076063e-01 -4.79159862e-01 1.58672825e-01
4.85843927e-01 7.88135052e-01 -1.93608254e-01 5.71996927e-01
2.91029155e-01 -2.42637232e-01 6.14450514e-01 -5.00386775e-01
3.05790365e-01 -1.32978097e-01 -1.32022810e+00 2.03233552e+00
-4.93524760e-01 2.59976298e-01 -2.06406370e-01 -1.54646325e+00
1.25588202e+00 7.26085007e-01 9.48650599e-01 -5.92647910e-01
1.02191932e-01 2.92890251e-01 5.92571571e-02 -5.39380312e-01
-1.57147691e-01 -3.14434826e-01 -4.23757266e-03 3.98983926e-01
1.01334766e-01 4.79982972e-01 -1.99443221e-01 -1.53150275e-01
1.22849703e+00 -8.53866246e-03 5.42170346e-01 -3.25842053e-01
5.77968538e-01 -4.11329716e-01 8.58906627e-01 6.18073642e-01
-9.15628448e-02 6.20775223e-01 -7.82311335e-02 -5.78433752e-01
-5.68178356e-01 -6.87823951e-01 -2.60869026e-01 5.51240325e-01
8.45958814e-02 -4.03813034e-01 -7.95012474e-01 -4.66712207e-01
-2.34163329e-01 5.27233839e-01 -3.86753052e-01 -5.38026989e-01
-6.25395119e-01 -1.09692633e+00 7.13949382e-01 4.80257809e-01
5.27565897e-01 -6.92681253e-01 -1.18618274e+00 5.00373602e-01
-4.82932448e-01 -8.49246740e-01 -2.12206826e-01 1.71737716e-01
-1.18142641e+00 -7.64092982e-01 -1.02595592e+00 -6.18344605e-01
6.00621343e-01 -3.40357572e-02 6.62923992e-01 7.62886852e-02
-5.43908775e-01 6.11527681e-01 -4.74678606e-01 -4.24955815e-01
-5.58337234e-02 -1.56918485e-02 4.52928543e-01 6.43695056e-01
1.88376710e-01 -9.14591551e-01 -9.10054982e-01 2.89454143e-02
-2.80903816e-01 -2.22742110e-01 5.05888283e-01 9.12360966e-01
7.20684946e-01 1.17121257e-01 1.13654280e+00 -6.60773516e-01
5.71461022e-01 -4.80369002e-01 -1.35024572e-02 -8.80270451e-03
-9.02556956e-01 -3.75071347e-01 6.48390949e-01 -6.06765985e-01
-8.75691295e-01 3.68672132e-01 -2.45266482e-01 -4.39559311e-01
-1.16295749e-02 3.80392015e-01 -2.26714820e-01 -7.53734484e-02
1.11827493e+00 6.63544297e-01 3.29632200e-02 -4.01172608e-01
-4.93255481e-02 7.24601090e-01 6.37180924e-01 -3.46849799e-01
6.61044002e-01 4.18945104e-01 5.76676689e-02 -9.57230389e-01
-4.17580038e-01 -4.90611404e-01 -5.05687058e-01 -2.71388769e-01
7.02472985e-01 -1.01936579e+00 -4.99209911e-01 2.79501081e-01
-7.61297226e-01 -1.70862526e-02 -7.63927579e-01 6.79208755e-01
-5.18917680e-01 7.01585531e-01 -2.02709273e-01 -9.27705884e-01
-9.02429700e-01 -7.68788695e-01 7.38552928e-01 2.26437114e-02
-5.86871326e-01 -7.97829747e-01 3.91101688e-02 1.13433346e-01
4.48518813e-01 6.76275849e-01 6.04425550e-01 -7.45846212e-01
3.78649756e-02 -4.25769448e-01 4.33566481e-01 4.69998956e-01
4.37602937e-01 -6.77354336e-01 -1.05046248e+00 -4.39681649e-01
5.95438957e-01 2.30340451e-01 4.37748313e-01 4.89922673e-01
1.22047079e+00 -4.09123182e-01 -4.37405348e-01 6.97126031e-01
1.29587841e+00 3.47727031e-01 6.27175391e-01 -9.42577124e-02
6.05585933e-01 1.05252057e-01 4.78335053e-01 8.59774768e-01
1.99191540e-01 6.80190921e-01 -1.25793330e-02 -2.73999989e-01
-5.86325265e-02 -1.97665356e-02 2.63978153e-01 9.93846655e-01
-3.83925110e-01 1.07944891e-01 -7.92332590e-01 5.57585537e-01
-1.80251658e+00 -1.04991055e+00 -8.31406936e-02 2.46927929e+00
8.08819711e-01 -1.34399474e-01 1.75649509e-01 8.37747395e-01
6.27827048e-01 -1.98569745e-01 -7.38497615e-01 -1.83092073e-01
-1.92121565e-02 6.09435439e-01 1.39778048e-01 -3.28313746e-02
-1.06540287e+00 1.79653224e-02 5.55601120e+00 5.78690767e-01
-1.43406117e+00 4.95538324e-01 4.95076388e-01 -1.01065986e-01
-3.98028120e-02 -2.82130241e-01 -2.95548022e-01 6.02763474e-01
1.18288302e+00 -3.43497962e-01 1.71668023e-01 8.63418460e-01
2.19870225e-01 2.65355617e-01 -9.32208776e-01 1.51639581e+00
4.49864835e-01 -1.18200469e+00 -3.11103851e-01 -1.59944460e-01
5.66871822e-01 -4.06694502e-01 -2.48271689e-01 1.78219914e-01
-8.90991211e-01 -6.43144786e-01 4.02116954e-01 8.54417622e-01
1.28376949e+00 -8.00249040e-01 7.64686704e-01 6.56988800e-01
-1.16268766e+00 -3.76157612e-01 -5.33035770e-03 -7.81515166e-02
-2.45443191e-02 1.20504212e+00 -6.91991985e-01 7.43765593e-01
6.93003774e-01 8.54387045e-01 -2.37535954e-01 9.37256455e-01
2.08647028e-02 1.01623833e+00 -3.67807508e-01 2.38735497e-01
-6.19780898e-01 3.83343571e-03 8.25321853e-01 1.18621778e+00
7.83487856e-01 4.31925595e-01 1.18006825e-01 5.42235434e-01
1.01823211e-01 2.59970427e-01 -7.32927501e-01 4.10242021e-01
2.93457925e-01 1.24436677e+00 -6.31744981e-01 -4.19545799e-01
-2.19310120e-01 1.13596666e+00 -4.19205874e-01 2.37264767e-01
-8.81982565e-01 -6.88577592e-01 2.62422740e-01 3.20511132e-01
1.02336094e-01 8.83691758e-02 -5.56092143e-01 -1.28133452e+00
2.01813325e-01 -9.16249812e-01 2.51764327e-01 -2.63125479e-01
-8.86887610e-01 5.47690094e-01 -3.10707748e-01 -1.73920548e+00
-4.02648389e-01 9.54570621e-02 -5.88871837e-01 7.78955758e-01
-1.01138258e+00 -8.03842783e-01 -5.56506038e-01 7.48613656e-01
2.98380107e-01 -3.97810161e-01 1.34298980e+00 7.03032672e-01
-5.38286686e-01 8.38243127e-01 -2.33917627e-02 -6.16255440e-02
4.80079085e-01 -9.63507712e-01 -1.66727021e-01 9.20053542e-01
1.14893407e-01 5.66531897e-01 4.95988280e-01 -5.92416763e-01
-1.32959342e+00 -1.34124339e+00 8.02759886e-01 -2.01608792e-01
-1.01535596e-01 -2.04880580e-01 -8.90161276e-01 1.05430670e-01
-5.88234663e-02 1.90173268e-01 1.12985826e+00 3.56456377e-02
2.45717436e-01 -6.52160048e-01 -1.37814772e+00 2.29117483e-01
9.83836174e-01 -3.97951722e-01 -5.29383242e-01 3.70999515e-01
2.82284617e-01 -3.27406585e-01 -1.26600206e+00 7.91735053e-01
8.70756626e-01 -7.47504473e-01 9.22733247e-01 -1.41610578e-01
-7.54607618e-02 -3.10538977e-01 -1.46539301e-01 -1.15339255e+00
-3.73314321e-01 -8.24778914e-01 -6.35445833e-01 1.00036144e+00
1.59960285e-01 -6.93136692e-01 7.54471600e-01 2.43940622e-01
-1.59303337e-01 -1.03361928e+00 -1.13357854e+00 -8.40699494e-01
-6.03165090e-01 -5.69016635e-01 3.61517936e-01 1.02045131e+00
1.56220630e-01 3.51672709e-01 -7.28753686e-01 2.70149466e-02
7.37147391e-01 7.13418424e-02 4.91329730e-01 -1.45955241e+00
-5.56562543e-01 2.01720819e-01 -5.30357897e-01 -5.61134875e-01
-3.14023912e-01 -9.71496642e-01 -2.09570259e-01 -1.35606527e+00
5.02408743e-02 -5.32224119e-01 -7.32101381e-01 4.83801574e-01
-1.26778767e-01 6.58727229e-01 -1.05217725e-01 1.94345236e-01
-2.15009838e-01 3.77735853e-01 7.47480452e-01 -8.69614929e-02
-6.51300907e-01 3.34390730e-01 -8.45012188e-01 6.91263855e-01
7.48217344e-01 -4.77884769e-01 -6.70432925e-01 2.05535024e-01
-7.97987804e-02 2.88674712e-01 3.97139698e-01 -1.58947039e+00
5.08798324e-02 4.21968520e-01 6.04201138e-01 -1.22244231e-01
3.75758708e-01 -8.48483860e-01 5.14785349e-01 8.78909707e-01
1.50136992e-01 -4.17683154e-01 8.79706889e-02 7.28538930e-01
-2.16457341e-02 9.86140408e-03 7.37350583e-01 1.11232311e-01
-2.71437645e-01 1.45530656e-01 -3.51864398e-01 -2.45770156e-01
1.21814334e+00 -5.50876796e-01 3.51556778e-01 -2.69296855e-01
-1.16281831e+00 -4.34287965e-01 2.99419649e-02 1.29750982e-01
7.91967213e-01 -1.42346704e+00 -6.71166956e-01 5.12576282e-01
1.82038531e-01 -3.52746755e-01 5.79027295e-01 1.24537706e+00
-1.44090503e-01 8.67746025e-02 -2.19622776e-01 -8.13316405e-01
-1.15940475e+00 2.61170179e-01 3.56315374e-01 -6.31919280e-02
-1.01195097e+00 4.46053475e-01 -3.00254077e-01 5.71255572e-02
1.01211272e-01 -2.67327547e-01 -2.97569782e-01 2.32091784e-01
4.46326196e-01 3.88508081e-01 4.24193054e-01 -4.34902012e-01
-5.24759471e-01 5.90457618e-01 4.78934318e-01 2.45464399e-01
1.24123502e+00 3.51211987e-02 2.48714328e-01 5.75743258e-01
9.23570275e-01 6.27049208e-02 -7.34402955e-01 -9.21142325e-02
-2.58395612e-01 -1.22524418e-01 2.90632527e-02 -7.04594910e-01
-1.17036569e+00 7.84936905e-01 1.28946257e+00 7.27716312e-02
1.74779689e+00 -4.85302061e-01 9.92154479e-01 2.62464702e-01
5.74184060e-01 -8.99284542e-01 1.59973721e-03 -6.90837353e-02
6.84895098e-01 -7.88686514e-01 8.71651322e-02 -4.76089567e-02
-6.79634869e-01 9.48078275e-01 1.59671634e-01 -9.63265747e-02
8.38062525e-01 1.83979087e-02 -1.26178399e-01 1.96362101e-02
-2.61097133e-01 2.80917529e-02 3.11852604e-01 8.37591588e-01
2.66164720e-01 8.56519118e-02 -7.43139684e-01 1.28595793e+00
2.95563079e-02 4.35517609e-01 3.61555904e-01 8.88226986e-01
-2.78293520e-01 -9.51023400e-01 -2.67818451e-01 7.28863239e-01
-5.77176809e-01 4.14449498e-02 1.09547205e-01 8.07691664e-02
4.65486586e-01 8.82799983e-01 -1.67088374e-01 -5.01162410e-01
3.72204512e-01 3.74744147e-01 3.76416624e-01 -7.32760131e-01
-5.47114670e-01 2.18015060e-01 -2.70181019e-02 -4.65104610e-01
-4.64440912e-01 -6.53993487e-01 -1.10276878e+00 3.88418436e-01
-2.87993103e-01 6.50721639e-02 5.73837936e-01 7.60944426e-01
7.63761759e-01 8.43504071e-01 7.37164438e-01 -7.49939859e-01
-4.35552388e-01 -8.84612441e-01 -7.63597310e-01 3.02650511e-01
3.17975938e-01 -5.48338413e-01 -1.02340117e-01 3.55316281e-01] | [14.248369216918945, 3.251767873764038] |
913cdabd-1d33-46a1-aaaf-1e4d1afb8221 | tracking-and-reconstructing-hand-object | 2209.12009 | null | https://arxiv.org/abs/2209.12009v1 | https://arxiv.org/pdf/2209.12009v1.pdf | Tracking and Reconstructing Hand Object Interactions from Point Cloud Sequences in the Wild | In this work, we tackle the challenging task of jointly tracking hand object pose and reconstructing their shapes from depth point cloud sequences in the wild, given the initial poses at frame 0. We for the first time propose a point cloud based hand joint tracking network, HandTrackNet, to estimate the inter-frame hand joint motion. Our HandTrackNet proposes a novel hand pose canonicalization module to ease the tracking task, yielding accurate and robust hand joint tracking. Our pipeline then reconstructs the full hand via converting the predicted hand joints into a template-based parametric hand model MANO. For object tracking, we devise a simple yet effective module that estimates the object SDF from the first frame and performs optimization-based tracking. Finally, a joint optimization step is adopted to perform joint hand and object reasoning, which alleviates the occlusion-induced ambiguity and further refines the hand pose. During training, the whole pipeline only sees purely synthetic data, which are synthesized with sufficient variations and by depth simulation for the ease of generalization. The whole pipeline is pertinent to the generalization gaps and thus directly transferable to real in-the-wild data. We evaluate our method on two real hand object interaction datasets, e.g. HO3D and DexYCB, without any finetuning. Our experiments demonstrate that the proposed method significantly outperforms the previous state-of-the-art depth-based hand and object pose estimation and tracking methods, running at a frame rate of 9 FPS. | ['He Wang', 'Shuran Song', 'Li Yi', 'Yijia Weng', 'Xiaolong Li', 'Yinzhen Xu', 'Jiazhao Zhang', 'Mi Yan', 'Jiayi Chen'] | 2022-09-24 | null | null | null | null | ['hand-object-pose'] | ['computer-vision'] | [-3.31515849e-01 -1.63950503e-01 2.57630926e-02 1.73456222e-01
-5.98223388e-01 -7.08723009e-01 3.93154502e-01 -4.10267353e-01
-4.79920506e-01 5.00556231e-01 -3.03764462e-01 1.83522366e-02
2.98335347e-02 -3.74088407e-01 -6.46987736e-01 -6.02698982e-01
2.52975523e-01 1.26682365e+00 6.58071697e-01 3.35081406e-02
8.81713182e-02 1.08840752e+00 -1.63585627e+00 -2.05991507e-01
5.63148320e-01 8.05176914e-01 5.27924120e-01 7.96937466e-01
9.71861705e-02 4.52456564e-01 -6.04801834e-01 -2.71457225e-01
3.92708451e-01 1.23906918e-01 -6.40843630e-01 3.32202673e-01
6.05481446e-01 -8.74281645e-01 -5.26078343e-01 7.01193273e-01
9.26636279e-01 -2.90501993e-02 3.31102550e-01 -1.20884097e+00
1.31391346e-01 9.50644985e-02 -6.13589406e-01 -3.43858659e-01
5.86738110e-01 6.43138289e-01 6.54393315e-01 -1.16496491e+00
1.19039249e+00 1.39798307e+00 6.38715267e-01 7.43122041e-01
-1.06796181e+00 -6.24980509e-01 3.39390516e-01 -3.03336438e-02
-1.66304719e+00 -1.71176553e-01 8.14466417e-01 -6.76468372e-01
8.93973172e-01 1.92205861e-01 1.07725048e+00 1.25230777e+00
-4.75125536e-02 9.18786287e-01 7.05798328e-01 -4.32706177e-01
1.40935868e-01 -3.37670088e-01 -1.01650529e-01 7.81874061e-01
7.07306340e-02 1.46169245e-01 -5.43511808e-01 -3.89111005e-02
1.41454947e+00 1.65839940e-01 -4.36024040e-01 -7.96461105e-01
-1.48927236e+00 1.32885113e-01 4.03204262e-01 6.70061558e-02
-5.32045782e-01 3.41273278e-01 1.39626771e-01 -2.92644858e-01
7.12648705e-02 -1.96952596e-01 -7.40810812e-01 -2.17203304e-01
-8.80179703e-01 9.29295242e-01 8.16133738e-01 1.37228537e+00
3.80830973e-01 -2.47949019e-01 -4.33108538e-01 2.61460900e-01
5.78282714e-01 6.78188384e-01 -2.60659069e-01 -1.10998499e+00
5.41304410e-01 4.38267440e-01 5.58491707e-01 -5.61707795e-01
-5.37508547e-01 -4.60558385e-01 -3.00428599e-01 6.61601603e-01
9.87947822e-01 -3.37951072e-02 -1.07397175e+00 1.40566254e+00
8.29364717e-01 -7.22414181e-02 -4.76466864e-01 1.33156538e+00
6.10853612e-01 7.78566226e-02 1.33511335e-01 -3.79828401e-02
1.50967789e+00 -9.82655823e-01 -7.29045928e-01 3.45218927e-02
1.83717072e-01 -7.88280129e-01 1.04984546e+00 5.68541467e-01
-1.17738533e+00 -6.40133083e-01 -6.22451425e-01 -1.52837455e-01
1.55951619e-01 4.66319412e-01 5.46428740e-01 3.74723613e-01
-6.22813761e-01 7.60183275e-01 -1.40552759e+00 -2.22142428e-01
5.44641793e-01 4.93960083e-01 -3.09066683e-01 1.02551416e-01
-4.44107503e-01 7.70177603e-01 3.76940489e-01 4.10372198e-01
-8.20333779e-01 -8.92319083e-01 -5.16215980e-01 -1.10949948e-01
6.36794269e-01 -9.88716364e-01 1.37433076e+00 -2.81382531e-01
-1.85778129e+00 7.54495800e-01 -1.97455466e-01 8.22801515e-02
1.15950429e+00 -5.26661515e-01 2.90714234e-01 2.43715327e-02
-2.49997199e-01 7.45934069e-01 1.03090119e+00 -1.46419501e+00
-4.63739902e-01 -7.13355124e-01 1.82685326e-03 3.43747400e-02
8.66341963e-02 1.96519308e-02 -1.21030366e+00 -7.85690069e-01
1.68583646e-01 -1.19861948e+00 7.12962123e-03 6.82877541e-01
-4.61164027e-01 -3.27051997e-01 8.47596049e-01 -8.67230177e-01
8.88083458e-01 -1.75197721e+00 6.60865605e-01 2.12733492e-01
3.13532710e-01 2.51894027e-01 7.39848241e-02 8.45493749e-02
2.49801636e-01 -5.77197611e-01 -1.30409986e-01 -8.63674998e-01
1.43955827e-01 1.57808736e-01 -4.25200872e-02 4.73585874e-01
-6.06317297e-02 1.06191838e+00 -8.18057299e-01 -7.22391427e-01
3.80270600e-01 9.69836771e-01 -5.93612134e-01 4.37298268e-01
-7.37757385e-01 1.00112021e+00 -3.81467342e-01 9.29361701e-01
7.99533725e-01 -8.41253549e-02 2.44592264e-01 -5.17735183e-01
-2.54754066e-01 -7.82577470e-02 -1.49270940e+00 2.36027408e+00
-2.98283517e-01 1.73847616e-01 3.77375364e-01 -6.55922815e-02
5.28251708e-01 4.25789058e-01 5.23913145e-01 -3.47798467e-02
5.04623473e-01 2.46568665e-01 1.69560581e-03 -3.79016489e-01
1.51657417e-01 1.62303910e-01 5.06454945e-01 4.15948570e-01
1.00405425e-01 -2.89868087e-01 -8.78539234e-02 -2.05649406e-01
7.00540721e-01 9.86462235e-01 -5.63157834e-02 1.02759525e-02
4.03062075e-01 7.77276754e-02 2.78804570e-01 2.86331922e-01
-1.61604863e-02 8.12565029e-01 1.44266665e-01 -3.32763940e-01
-1.07287133e+00 -1.06758809e+00 -3.45149264e-02 7.80225575e-01
6.40510693e-02 -3.29017550e-01 -9.48603511e-01 -6.21463537e-01
3.40957552e-01 2.27016941e-01 -3.94673824e-01 6.28041446e-01
-9.32093680e-01 -6.72678500e-02 1.53634772e-01 9.39258695e-01
2.01967016e-01 -1.17021704e+00 -9.05441821e-01 2.85313159e-01
-4.23347251e-03 -1.12869155e+00 -7.47083783e-01 -1.75778717e-01
-8.44771922e-01 -1.13570273e+00 -1.15943623e+00 -5.73728681e-01
6.14958525e-01 -1.39276743e-01 7.25166678e-01 1.89009830e-01
-5.48279703e-01 4.51558501e-01 -7.00650960e-02 -3.21917057e-01
-2.50978589e-01 2.49839380e-01 1.12749308e-01 -2.03921899e-01
-1.61454722e-01 -7.36590028e-01 -8.42315853e-01 4.43451643e-01
-3.87077063e-01 1.28572294e-02 4.16728914e-01 5.26544094e-01
8.06299746e-01 -4.27725852e-01 -5.45963794e-02 -2.15694740e-01
1.70833111e-01 3.69593740e-01 -1.01365912e+00 3.05126548e-01
-2.10733324e-01 8.32219198e-02 1.68126687e-01 -8.11307251e-01
-1.12381649e+00 7.42480457e-01 -3.32491308e-01 -1.06084692e+00
-1.74906030e-01 -1.68208554e-01 -4.44507092e-01 -3.29104364e-01
4.76366162e-01 -1.02442533e-01 1.25908643e-01 -9.66521680e-01
4.60657418e-01 4.29205656e-01 7.93671727e-01 -8.24096680e-01
9.49764073e-01 6.64313912e-01 1.09667547e-01 -5.10936737e-01
-4.09908473e-01 -3.10109615e-01 -1.30193925e+00 -4.37130123e-01
9.55536902e-01 -7.08978176e-01 -1.47232413e+00 8.81994843e-01
-1.79885101e+00 -5.36257207e-01 -2.44876489e-01 5.24878085e-01
-7.24583924e-01 4.15245384e-01 -6.19034886e-01 -9.27035451e-01
-4.21824098e-01 -1.32792437e+00 1.63294697e+00 -1.50681213e-01
-2.37154752e-01 -5.48635304e-01 8.20446992e-04 1.20852441e-01
-1.54841274e-01 3.23593289e-01 3.00884157e-01 -7.70283211e-03
-1.03001893e+00 -1.54403821e-01 -1.11524716e-01 -1.33660465e-01
2.87290271e-02 1.25107110e-01 -9.87586617e-01 -5.61095715e-01
-3.45045507e-01 1.73306670e-02 3.80083382e-01 2.94919521e-01
1.10710812e+00 7.64760077e-02 -5.16408861e-01 6.65063024e-01
1.14720929e+00 -4.93316911e-02 4.43977475e-01 2.33163029e-01
1.21492779e+00 5.68068564e-01 7.19980538e-01 6.22844279e-01
3.70982289e-01 1.21991968e+00 5.81235647e-01 1.86816692e-01
-5.09888411e-01 -1.63228050e-01 -9.09639448e-02 4.72138882e-01
-7.39498854e-01 5.31553179e-02 -9.38396513e-01 2.38214135e-01
-1.79978466e+00 -5.25358379e-01 -1.61140651e-01 2.16074061e+00
7.96683133e-01 7.89608806e-02 5.52694440e-01 2.03886211e-01
4.85332221e-01 -2.87224799e-01 -6.95056558e-01 6.25330389e-01
3.64631683e-01 5.02431214e-01 2.51013190e-01 7.75928438e-01
-7.96096861e-01 1.30202723e+00 5.13819551e+00 4.55540121e-01
-9.73549128e-01 2.62444288e-01 -4.48943079e-01 -2.50116289e-01
1.90594658e-01 -8.59508887e-02 -9.74223256e-01 9.11812782e-02
4.72110957e-02 4.51561451e-01 4.47959363e-01 9.27235484e-01
1.00988232e-01 1.27223551e-01 -1.30410874e+00 1.17083538e+00
-1.80696249e-01 -1.02317667e+00 -1.26521975e-01 9.85313859e-03
3.02186936e-01 -1.79567203e-01 -2.91345298e-01 -1.86691463e-01
-1.21909484e-01 -6.12505376e-01 1.31589580e+00 7.26868510e-01
8.53023052e-01 -5.30953407e-01 2.79376030e-01 5.85918605e-01
-1.43491554e+00 1.92734227e-01 9.62378383e-02 3.92660983e-02
4.46413994e-01 2.46375963e-01 -8.00065339e-01 4.61579889e-01
7.53415942e-01 3.48995119e-01 -3.84741127e-01 9.71829593e-01
-3.12267542e-01 -8.74212161e-02 -4.80976313e-01 2.81133741e-01
-3.66055191e-01 1.13194406e-01 7.98089266e-01 8.96238804e-01
1.50537238e-01 1.91140905e-01 3.18966091e-01 1.02916861e+00
2.35615090e-01 -3.30978453e-01 7.73864016e-02 2.98020691e-01
4.77052122e-01 1.20702243e+00 -8.90919983e-01 -1.53913885e-01
9.51341242e-02 1.30421817e+00 3.15136850e-01 2.69354612e-01
-7.79082716e-01 -1.67272970e-01 7.52465963e-01 4.07169402e-01
5.41865230e-01 -7.05095649e-01 -1.66352823e-01 -1.23159969e+00
5.25192976e-01 -6.76093340e-01 -9.50453728e-02 -7.73748159e-01
-9.02750075e-01 6.55875444e-01 9.82086137e-02 -1.16179991e+00
-4.40988809e-01 -7.86550045e-01 -1.55075610e-01 1.05965853e+00
-1.32117069e+00 -1.55038297e+00 -7.96683252e-01 7.57577062e-01
5.67306578e-01 2.25124761e-01 6.64832354e-01 2.15711460e-01
-4.41052437e-01 6.14810407e-01 -6.42836690e-01 1.18035853e-01
5.85695744e-01 -1.11864769e+00 5.69422066e-01 4.73004043e-01
-1.34950336e-02 6.35670304e-01 6.74005270e-01 -8.86646509e-01
-1.75603437e+00 -7.99339294e-01 4.32268530e-01 -1.01116371e+00
3.12539220e-01 -6.18614435e-01 -9.04436529e-01 7.71277368e-01
-4.16978925e-01 3.60983968e-01 -3.25735986e-01 -2.67713130e-01
-2.98332095e-01 1.65656194e-01 -1.15852511e+00 4.90977317e-01
1.65950286e+00 -4.25539821e-01 -4.64605331e-01 5.31727076e-01
6.42799139e-01 -1.25412738e+00 -9.76035655e-01 2.08903208e-01
1.15122271e+00 -7.21335053e-01 1.23763824e+00 -4.04403418e-01
-1.93569496e-01 -6.74607694e-01 1.62585199e-01 -6.93124712e-01
-1.32874191e-01 -8.58662367e-01 -6.55908644e-01 1.09357381e+00
-2.46011078e-01 -3.28325152e-01 1.22743905e+00 5.04323483e-01
1.38981044e-01 -7.87215233e-01 -1.01966965e+00 -9.09506321e-01
-2.50004709e-01 -5.16654253e-01 7.04828143e-01 3.27821046e-01
-4.54154879e-01 -6.93470389e-02 -8.52286518e-02 4.25965786e-01
1.11594868e+00 1.42136917e-01 1.21738255e+00 -1.46915245e+00
-5.13445139e-01 -2.94904232e-01 -2.58365721e-01 -1.39221585e+00
3.09117526e-01 -5.45863628e-01 2.60048777e-01 -1.29083359e+00
9.03741177e-03 -4.69167858e-01 4.96980608e-01 5.37469864e-01
-9.90105271e-02 8.33449960e-02 4.80716407e-01 4.26992446e-01
-9.92071480e-02 3.46668929e-01 1.72371614e+00 1.13464691e-01
-4.53477889e-01 2.30435789e-01 2.39272460e-01 8.20301473e-01
2.56335706e-01 -3.88274431e-01 -7.88928866e-02 -5.14103532e-01
-3.25523615e-01 2.91742951e-01 9.69092548e-01 -9.36914563e-01
3.51734012e-01 -1.59553643e-02 3.68015170e-01 -1.08059049e+00
5.55749536e-01 -1.15057945e+00 4.92859602e-01 6.64950490e-01
1.97086021e-01 -3.57658826e-02 2.26916239e-01 2.70729542e-01
3.92698824e-01 1.53307216e-02 6.20871127e-01 -7.34213460e-03
-4.08251613e-01 7.43074119e-01 2.38890931e-01 -3.22250873e-01
9.24496770e-01 -2.90574640e-01 1.16229795e-01 2.38814987e-02
-1.12981856e+00 9.17870030e-02 6.57837033e-01 3.87965590e-01
5.55826843e-01 -1.22602415e+00 -6.48033679e-01 2.36462593e-01
-4.20019031e-02 6.69214964e-01 1.73719242e-01 8.59260678e-01
-6.23687983e-01 3.45720768e-01 -1.16438955e-01 -1.07682037e+00
-1.38832653e+00 6.03275359e-01 4.07649308e-01 -6.58320785e-02
-1.05472898e+00 7.24182725e-01 -1.77154876e-02 -5.47168553e-01
6.03735805e-01 -5.75279355e-01 2.68267334e-01 -1.97746500e-01
4.27615881e-01 7.14822590e-01 1.81517988e-01 -7.09237933e-01
-5.09048700e-01 1.08652115e+00 2.36758694e-01 -3.07393819e-01
1.27901542e+00 2.79318660e-01 -7.38091618e-02 5.61108929e-04
8.53101194e-01 -6.14891425e-02 -1.89585972e+00 -1.66036990e-02
-3.06761622e-01 -7.18838155e-01 -1.39368504e-01 -7.99992919e-01
-1.11486661e+00 9.70829666e-01 7.44604945e-01 -5.72552621e-01
7.36267090e-01 2.24281356e-01 7.81319320e-01 4.63606089e-01
8.22399557e-01 -6.39503837e-01 4.06000800e-02 3.77130330e-01
1.30042470e+00 -7.81715214e-01 5.28870001e-02 -8.63343298e-01
-1.07373931e-01 1.26464355e+00 7.43729413e-01 -6.02337113e-03
5.48111737e-01 6.49454713e-01 -1.15181938e-01 -2.18156978e-01
-1.05748445e-01 -1.54515699e-01 4.80756968e-01 7.42623985e-01
6.70345277e-02 -1.05961710e-01 -4.24967110e-02 4.37090009e-01
-1.20194785e-01 4.05025870e-01 -2.47359216e-01 1.11683011e+00
-4.40531112e-02 -1.37085247e+00 -8.55001986e-01 -1.47037283e-01
-7.65604228e-02 4.01321679e-01 -1.89595118e-01 1.14696455e+00
2.39021912e-01 1.74512833e-01 5.80293797e-02 -2.29815453e-01
8.06409001e-01 6.89824373e-02 1.33181810e+00 -5.09586811e-01
-6.88768148e-01 3.88819814e-01 -3.49825144e-01 -8.09840679e-01
-3.27495158e-01 -8.72578084e-01 -1.55314839e+00 -2.26501107e-01
-5.62283695e-01 -4.13664699e-01 7.76634514e-01 9.43896115e-01
3.44474822e-01 5.94475389e-01 -8.99811834e-03 -1.97270977e+00
-6.75834596e-01 -8.69831800e-01 -4.43914056e-01 2.38398805e-01
3.74765813e-01 -1.21289504e+00 6.45757141e-03 -1.86687075e-02] | [6.583222389221191, -0.9390112161636353] |
b34daa5c-5ddd-4c99-949b-2bf7b13f2f68 | rethinking-pose-in-3d-multi-stage-refinement | 1808.01525 | null | http://arxiv.org/abs/1808.01525v1 | http://arxiv.org/pdf/1808.01525v1.pdf | Rethinking Pose in 3D: Multi-stage Refinement and Recovery for Markerless Motion Capture | We propose a CNN-based approach for multi-camera markerless motion capture of
the human body. Unlike existing methods that first perform pose estimation on
individual cameras and generate 3D models as post-processing, our approach
makes use of 3D reasoning throughout a multi-stage approach. This novelty
allows us to use provisional 3D models of human pose to rethink where the
joints should be located in the image and to recover from past mistakes. Our
principled refinement of 3D human poses lets us make use of image cues, even
from images where we previously misdetected joints, to refine our estimates as
part of an end-to-end approach. Finally, we demonstrate how the high-quality
output of our multi-camera setup can be used as an additional training source
to improve the accuracy of existing single camera models. | ['Denis Tome', 'Chris Russell', 'Lourdes Agapito', 'Matteo Toso'] | 2018-08-04 | null | null | null | null | ['markerless-motion-capture'] | ['computer-vision'] | [-3.06602065e-02 2.78497100e-01 1.81948289e-01 -1.21071286e-01
-7.39349902e-01 -6.36291265e-01 4.22306508e-01 -2.34285280e-01
-6.21517420e-01 4.15854871e-01 3.43251050e-01 2.97689270e-02
3.70563835e-01 -3.04265648e-01 -8.20241511e-01 -6.60411417e-02
1.84689865e-01 6.74320817e-01 5.83003044e-01 -3.06132048e-01
6.33697882e-02 6.75647259e-01 -1.13076043e+00 -2.28306409e-02
1.60978675e-01 4.36446637e-01 1.07754767e-02 1.04010212e+00
8.17376554e-01 3.69730055e-01 -3.24278027e-01 -5.74308038e-01
5.40682375e-01 -2.68747211e-01 -5.61823785e-01 6.24822915e-01
8.02628815e-01 -7.56356359e-01 -2.33603537e-01 6.07895195e-01
4.36272025e-01 1.16752781e-01 2.87364334e-01 -9.61633146e-01
1.60740256e-01 -1.33053526e-01 -5.87443650e-01 -1.69927493e-01
8.24976504e-01 3.74062359e-01 6.53147161e-01 -9.25508142e-01
8.44548106e-01 1.13544762e+00 1.04948807e+00 5.65444410e-01
-1.06961930e+00 -1.78440839e-01 7.44421706e-02 -2.18886346e-01
-1.31804240e+00 -6.64686382e-01 7.94732928e-01 -5.10063052e-01
9.03744340e-01 2.05800049e-02 1.22800922e+00 1.10123742e+00
1.76694527e-01 6.73713803e-01 6.27852559e-01 -6.42259240e-01
-1.69942841e-01 -2.75488049e-01 -3.77695471e-01 1.12810469e+00
3.24113697e-01 1.68016434e-01 -4.53772604e-01 -1.49254978e-01
1.49488950e+00 1.55935794e-01 -1.61385670e-01 -7.76229501e-01
-1.39247990e+00 4.43891764e-01 4.25438970e-01 -5.04866540e-02
-5.35122335e-01 4.28407550e-01 -2.17548490e-01 -3.35570425e-01
2.69517571e-01 3.62841278e-01 -4.87394482e-01 -1.92221135e-01
-1.13086545e+00 4.84573692e-01 5.16340077e-01 8.30055892e-01
7.51198530e-01 -2.03407124e-01 2.50732124e-01 5.02785742e-01
5.44941366e-01 3.14874202e-01 2.45070025e-01 -1.59226024e+00
3.32184374e-01 6.10467434e-01 4.22254145e-01 -8.98220837e-01
-5.01234055e-01 -2.57562757e-01 -1.97663113e-01 3.84048104e-01
6.05723023e-01 -2.93305188e-01 -1.09720409e+00 1.21798253e+00
6.77129865e-01 1.42470658e-01 -2.31230229e-01 1.27131784e+00
2.57877801e-02 -9.67236236e-02 -2.49696821e-01 4.08835828e-01
1.29299378e+00 -1.12080252e+00 -1.95308208e-01 -5.69952190e-01
5.28235018e-01 -8.39060903e-01 5.46452820e-01 6.01025164e-01
-1.31860948e+00 -6.14169478e-01 -1.07110739e+00 -2.98332185e-01
8.06428939e-02 2.79232353e-01 4.31432664e-01 4.33875769e-01
-1.10084331e+00 8.91193688e-01 -1.40033460e+00 -4.07096684e-01
5.67782223e-02 5.67734957e-01 -7.88293183e-01 7.71825686e-02
-7.74261475e-01 1.32612586e+00 1.80269763e-01 4.24571902e-01
-9.12831903e-01 -4.07467753e-01 -1.06770563e+00 -4.79523778e-01
6.60637558e-01 -1.21513188e+00 1.48963225e+00 -6.19766116e-01
-1.55095196e+00 8.92611861e-01 -1.07554376e-01 -1.75887048e-01
9.00785804e-01 -7.70704210e-01 4.08303797e-01 5.70280075e-01
-1.60213448e-02 8.16003323e-01 8.27758491e-01 -1.22500634e+00
-5.61275184e-01 -4.66551840e-01 1.22810654e-01 2.76010543e-01
2.79995203e-01 -2.98820913e-01 -9.44969654e-01 -4.41742450e-01
2.47395322e-01 -1.49052262e+00 -6.76051617e-01 2.98405290e-01
-4.65399176e-01 5.55824995e-01 3.50835919e-01 -9.81499076e-01
5.33905029e-01 -1.64820039e+00 4.82944041e-01 2.31786057e-01
3.83629650e-01 1.51527017e-01 2.18560696e-01 3.44656557e-01
6.36204630e-02 -7.84215033e-02 -1.68289170e-02 -8.97842765e-01
-2.34130949e-01 6.53882772e-02 2.98467547e-01 6.82974815e-01
3.44335705e-01 7.90949047e-01 -8.30254614e-01 -6.48421526e-01
6.57724082e-01 6.67628765e-01 -6.45547688e-01 2.90279061e-01
-3.92951518e-02 7.60497570e-01 -2.94249803e-01 6.34002328e-01
1.69083282e-01 -2.10912451e-01 1.46681547e-01 -4.19205040e-01
8.82525072e-02 -1.07190706e-01 -1.39522564e+00 2.33204770e+00
-3.50510806e-01 1.42621800e-01 9.25972313e-02 -3.75001311e-01
3.31480056e-01 5.08404553e-01 6.35547876e-01 -6.63817897e-02
3.02936465e-01 2.33358331e-02 -2.30873287e-01 -3.03860664e-01
6.27107322e-01 -2.30040342e-01 -9.73209813e-02 2.50682116e-01
1.62880108e-01 -1.05046585e-01 -1.98340341e-01 1.56782076e-01
9.58436489e-01 9.47928905e-01 3.44782054e-01 3.11881512e-01
2.80026466e-01 3.96241754e-01 4.90255833e-01 3.45636398e-01
-1.29341453e-01 1.50447309e+00 1.02700666e-02 -4.42236245e-01
-1.51005423e+00 -9.11637366e-01 3.93136948e-01 5.57353556e-01
-2.91324537e-02 -4.22781289e-01 -6.27862215e-01 -5.46161473e-01
7.23665804e-02 1.43851712e-01 -5.87479353e-01 8.44906494e-02
-9.29780066e-01 -4.40596670e-01 5.77823937e-01 7.87052214e-01
2.14829877e-01 -4.31852102e-01 -1.16598165e+00 2.86280483e-01
-1.63738251e-01 -1.31303275e+00 -4.59092796e-01 5.32328226e-02
-1.05289388e+00 -1.39986658e+00 -1.10100126e+00 -4.31118876e-01
7.62734115e-01 8.87256861e-02 9.19672191e-01 3.64684463e-01
-2.82970190e-01 7.93461323e-01 -2.24776998e-01 -4.13212590e-02
-3.13980341e-01 -2.08715141e-01 1.78091675e-01 -2.90997088e-01
-2.76518792e-01 -5.52102208e-01 -8.42266917e-01 1.91930011e-01
-4.62618202e-01 3.42574000e-01 5.25567114e-01 5.74770927e-01
4.38314110e-01 -6.15479350e-01 -2.87074715e-01 -6.38500154e-01
1.65069327e-01 -1.00362170e-02 -5.63906014e-01 9.37822983e-02
-3.08628410e-01 -1.94663461e-02 1.89729482e-01 -4.64969486e-01
-9.85496521e-01 8.47871125e-01 -3.63227457e-01 -9.21430707e-01
-2.30340302e-01 2.40419686e-01 5.53759336e-02 -1.76062599e-01
5.14958262e-01 -2.74226457e-01 2.70877123e-01 -6.71323001e-01
5.05635440e-01 2.17462018e-01 9.75437820e-01 -4.72166747e-01
8.21679890e-01 6.66448534e-01 7.00874776e-02 -5.24107635e-01
-5.58608830e-01 -5.65448225e-01 -1.42918479e+00 -3.97657424e-01
9.16124403e-01 -1.33249938e+00 -6.57947421e-01 4.05264586e-01
-1.38775945e+00 -2.81114697e-01 5.07763177e-02 6.80081785e-01
-6.96627557e-01 6.37875080e-01 -7.71113157e-01 -8.94163430e-01
-1.39819026e-01 -1.16897964e+00 1.71356308e+00 1.47413567e-01
-7.08113849e-01 -9.95768309e-01 1.35881990e-01 6.49750710e-01
-2.94775754e-01 5.98749220e-01 2.81027228e-01 -3.11229616e-01
-5.89933038e-01 -6.80589914e-01 5.41836023e-01 4.26933579e-02
-1.32452279e-01 1.36104792e-01 -9.05666590e-01 -2.36913949e-01
-2.37621233e-01 -1.97169065e-01 4.05922830e-01 2.95584142e-01
2.40444675e-01 -6.13814257e-02 -4.77471977e-01 6.13880098e-01
1.27286363e+00 -3.39530259e-01 6.85498834e-01 6.20123863e-01
1.16758633e+00 5.43922901e-01 5.87044597e-01 3.37687820e-01
5.81066847e-01 9.84546781e-01 4.84677702e-01 -1.71312034e-01
-7.27526173e-02 -4.44759876e-01 3.42873037e-01 6.36256754e-01
-4.26307172e-01 1.24519929e-01 -9.19139743e-01 6.10755205e-01
-1.96380186e+00 -6.24399543e-01 -1.94148481e-01 2.36570406e+00
6.00690186e-01 2.77014017e-01 4.98021215e-01 3.94625543e-03
5.65816522e-01 -2.63355166e-01 -3.00530076e-01 9.17638019e-02
4.00085628e-01 8.87339413e-02 6.51359439e-01 6.79261804e-01
-8.78065288e-01 1.04997492e+00 6.50716639e+00 -6.34855255e-02
-8.70465279e-01 -1.24027543e-01 1.94845423e-02 -2.47335091e-01
1.59934849e-01 2.24989697e-01 -6.98758960e-01 4.73426543e-02
7.25193441e-01 4.43360984e-01 9.13263485e-03 7.60209382e-01
2.96939582e-01 -5.59009016e-01 -1.35579681e+00 8.80834758e-01
1.03507884e-01 -1.14188457e+00 -3.16300422e-01 2.73146629e-01
4.91288096e-01 -6.92945570e-02 -4.11681890e-01 -2.29033709e-01
5.23088634e-01 -5.72795272e-01 1.05321133e+00 7.12891340e-01
4.39381361e-01 -4.84467715e-01 5.05113363e-01 6.50032401e-01
-9.66473043e-01 2.62992978e-01 -6.62132353e-02 -3.99065167e-01
4.84005779e-01 2.97487170e-01 -1.10561073e+00 7.24249601e-01
3.49102139e-01 5.47071218e-01 -8.56065035e-01 1.18721604e+00
-3.73123318e-01 1.19247660e-01 -4.43046480e-01 4.73893553e-01
-6.80426732e-02 1.99014008e-01 4.83747780e-01 8.41318667e-01
2.83141226e-01 -1.26414998e-02 3.72840315e-01 4.92166817e-01
2.27868378e-01 -4.34485704e-01 -5.63540459e-01 4.15764540e-01
5.60875498e-02 1.34779227e+00 -8.38364661e-01 -3.30053955e-01
-1.54574350e-01 1.45853686e+00 3.52180541e-01 7.14626461e-02
-6.86508000e-01 1.07699133e-01 5.07107973e-01 2.46677354e-01
4.61561084e-01 -8.28332245e-01 -1.80650830e-01 -1.51674664e+00
1.71476096e-01 -7.56806433e-01 1.66888833e-01 -1.20778096e+00
-6.72230899e-01 2.05471575e-01 1.39697447e-01 -1.34809947e+00
-8.91737998e-01 -4.73515064e-01 -3.37513924e-01 8.41303945e-01
-9.95266378e-01 -1.43611741e+00 -2.29287937e-01 4.04329181e-01
4.38907474e-01 3.92987788e-01 6.74901962e-01 2.55718548e-02
-3.60454440e-01 2.25160420e-01 -5.34743369e-01 4.55460012e-01
8.05756092e-01 -1.12838566e+00 6.02556229e-01 1.16761088e+00
2.59473115e-01 7.74010420e-01 6.66090429e-01 -1.06294858e+00
-1.49584222e+00 -7.27424920e-01 5.59604585e-01 -1.22676110e+00
3.27977687e-01 -3.28487754e-01 -4.89642322e-01 1.03794622e+00
-1.49522483e-01 4.68527973e-02 3.05320680e-01 7.95721263e-02
7.86964130e-03 1.97857037e-01 -9.04790699e-01 6.44763529e-01
1.00861967e+00 -4.35916275e-01 -9.14774656e-01 4.95162643e-02
4.98583615e-01 -1.00784409e+00 -8.13115716e-01 2.55520731e-01
1.03594148e+00 -8.72142673e-01 1.12338889e+00 -3.95120144e-01
4.04073745e-01 -5.08578420e-01 2.31748745e-01 -1.21948969e+00
-1.56605244e-01 -6.37500465e-01 2.98619606e-02 7.81709671e-01
2.87237883e-01 3.74027006e-02 1.14536417e+00 9.45698023e-01
-6.82526678e-02 -3.08985084e-01 -8.56774807e-01 -3.46090883e-01
-2.61410445e-01 -5.83494008e-01 9.68684703e-02 4.16299075e-01
-1.21114969e-01 1.20827615e-01 -8.00138235e-01 4.05506283e-01
7.09934294e-01 -1.96186468e-01 1.29063141e+00 -1.19975090e+00
-4.67177093e-01 8.17580372e-02 -5.72212040e-01 -1.16410899e+00
-2.40053833e-01 -2.82607853e-01 3.75431538e-01 -1.36640573e+00
2.79495176e-02 -5.61432615e-02 2.25120142e-01 6.21952534e-01
-2.29474813e-01 7.04165995e-01 5.35982370e-01 4.48318839e-01
-6.59169257e-01 2.08379865e-01 1.22509897e+00 3.01290929e-01
-7.00912848e-02 8.21877737e-03 -4.34484512e-01 1.12098861e+00
3.21889549e-01 -5.57086110e-01 -7.27305263e-02 -6.74706936e-01
6.99607283e-02 3.32877040e-01 1.14571118e+00 -1.33375084e+00
3.20315540e-01 7.83653408e-02 1.03491175e+00 -7.29490995e-01
7.81373084e-01 -1.04010952e+00 4.48916525e-01 5.34198999e-01
1.68295540e-02 3.52520764e-01 4.81926985e-02 6.01740420e-01
1.65426224e-01 -8.13443139e-02 4.36337233e-01 -7.35022485e-01
-6.06492102e-01 1.10738920e-02 -3.67547274e-01 -2.68709958e-01
8.42494130e-01 -6.86423182e-01 2.06727281e-01 -3.85659575e-01
-1.12937343e+00 1.70812920e-01 1.10763311e+00 2.74885297e-01
6.90977216e-01 -1.11149848e+00 -3.45680684e-01 5.12156188e-02
-2.24323303e-01 4.28320050e-01 1.57760847e-02 9.15518820e-01
-8.94379675e-01 3.10291886e-01 -3.24125141e-01 -8.62210691e-01
-1.32749331e+00 3.07947427e-01 4.87790734e-01 -1.39124751e-01
-8.99055004e-01 4.49831337e-01 -2.51555145e-01 -3.11481327e-01
1.39349297e-01 -2.62979895e-01 2.57421434e-01 -3.09326589e-01
2.18248755e-01 2.15728849e-01 2.07010567e-01 -9.17519331e-01
-4.70759124e-01 8.62385631e-01 1.42096639e-01 -7.71023810e-01
1.43064356e+00 -4.01823014e-01 3.72712493e-01 3.26707900e-01
1.10105717e+00 2.77286798e-01 -1.79210961e+00 1.12203911e-01
-2.37570871e-02 -4.89364177e-01 -1.66767955e-01 -7.48121738e-01
-7.69342244e-01 8.75902236e-01 3.36324453e-01 -6.09328568e-01
6.99961364e-01 -1.25489831e-01 8.70248973e-01 3.81296486e-01
6.47759557e-01 -1.08087409e+00 2.26852551e-01 2.89027005e-01
6.95678294e-01 -1.18854034e+00 4.80326027e-01 -4.36206579e-01
-6.64923429e-01 1.36976266e+00 6.13094807e-01 -3.93906415e-01
3.16808671e-01 2.85711497e-01 2.49441981e-01 -1.84724182e-01
-3.65251392e-01 -3.64062101e-01 4.11458999e-01 7.24370360e-01
3.66463035e-01 -2.80427635e-01 1.72256261e-01 2.40377769e-01
-2.43413270e-01 2.56834537e-01 6.57920480e-01 1.39119053e+00
-1.61137238e-01 -1.33247554e+00 -6.90808177e-01 -1.46745935e-01
-5.11588931e-01 2.33348355e-01 -5.10548532e-01 1.01035321e+00
2.09557652e-01 6.43812299e-01 -1.71594054e-01 -6.32652938e-01
5.88601828e-01 2.25434422e-01 9.98336494e-01 -7.85627306e-01
-5.51304162e-01 5.08231223e-01 2.85628319e-01 -7.77876437e-01
-6.44466221e-01 -6.53660417e-01 -1.11055827e+00 1.44847959e-01
-3.00760061e-01 -2.70397961e-01 6.12263918e-01 1.13488531e+00
3.08381528e-01 2.93482602e-01 1.12919934e-01 -1.61809945e+00
-2.64807612e-01 -7.69280374e-01 -1.55156866e-01 5.04496872e-01
5.83453357e-01 -8.38388979e-01 8.71373042e-02 4.95101303e-01] | [7.052670955657959, -0.9975826144218445] |
07130297-a684-46a7-a15e-fd38026d6cff | guiding-the-guidance-a-comparative-analysis | 2303.06942 | null | https://arxiv.org/abs/2303.06942v1 | https://arxiv.org/pdf/2303.06942v1.pdf | Guiding the Guidance: A Comparative Analysis of User Guidance Signals for Interactive Segmentation of Volumetric Images | Interactive segmentation reduces the annotation time of medical images and allows annotators to iteratively refine labels with corrective interactions, such as clicks. While existing interactive models transform clicks into user guidance signals, which are combined with images to form (image, guidance) pairs, the question of how to best represent the guidance has not been fully explored. To address this, we conduct a comparative study of existing guidance signals by training interactive models with different signals and parameter settings to identify crucial parameters for the model's design. Based on our findings, we design a guidance signal that retains the benefits of other signals while addressing their limitations. We propose an adaptive Gaussian heatmaps guidance signal that utilizes the geodesic distance transform to dynamically adapt the radius of each heatmap when encoding clicks. We conduct our study on the MSD Spleen and the AutoPET datasets to explore the segmentation of both anatomy (spleen) and pathology (tumor lesions). Our results show that choosing the guidance signal is crucial for interactive segmentation as we improve the performance by 14% Dice with our adaptive heatmaps on the challenging AutoPET dataset when compared to non-interactive models. This brings interactive models one step closer to deployment on clinical workflows. We will make our code publically available. | ['Jens Kleesiek', 'Rainer Stiefelhagen', 'Zdravko Marinov'] | 2023-03-13 | null | null | null | null | ['interactive-segmentation', 'anatomy'] | ['computer-vision', 'miscellaneous'] | [ 2.91040570e-01 2.98978925e-01 8.48098174e-02 -7.11429477e-01
-7.50167370e-01 -8.91723692e-01 1.83687210e-01 4.35509831e-01
-5.20179033e-01 1.03821054e-01 2.06560344e-01 -7.34851718e-01
-1.47687584e-01 -2.90685266e-01 -2.69420266e-01 -4.35232580e-01
-2.74147004e-01 5.82649887e-01 6.57534301e-01 1.24627903e-01
4.70841557e-01 2.90813088e-01 -9.01704669e-01 3.77613842e-01
1.11022222e+00 5.51050961e-01 3.68187428e-01 8.03113341e-01
-1.88023135e-01 -4.41593770e-03 -4.34776604e-01 -3.10444176e-01
4.44775820e-01 -4.18704063e-01 -7.83878624e-01 -1.04751043e-01
4.54319358e-01 -4.24097806e-01 3.43386263e-01 7.28676796e-01
7.71454453e-01 -1.02281436e-01 6.27283156e-01 -1.07030511e+00
-4.83627886e-01 8.29320252e-01 -6.98554158e-01 2.96396047e-01
1.91550955e-01 4.74237889e-01 6.92969501e-01 -4.57540125e-01
9.00811613e-01 9.32565272e-01 9.80202496e-01 4.93719071e-01
-1.66380322e+00 -5.84003031e-01 -5.02869822e-02 -1.45268217e-01
-1.27768230e+00 -1.78537071e-02 2.83173531e-01 -7.05539405e-01
5.61775148e-01 5.32216609e-01 7.87197888e-01 5.61247826e-01
1.87800035e-01 7.25986898e-01 1.21856201e+00 -4.08037603e-01
1.42288238e-01 1.98204771e-01 1.10429898e-01 8.50994527e-01
-6.38108104e-02 -2.56331980e-01 -1.86110035e-01 -3.21312219e-01
9.79006350e-01 -3.04287314e-01 -3.33088249e-01 -5.45150638e-01
-1.45351994e+00 8.57947350e-01 4.32609618e-01 2.55678266e-01
-2.28678465e-01 2.54512757e-01 2.57616073e-01 4.04395647e-02
3.02737951e-01 7.50830531e-01 -1.81743950e-01 -3.20633769e-01
-1.24566233e+00 4.54995148e-02 6.37395799e-01 7.04512119e-01
4.13340986e-01 -5.60154021e-01 -5.44463158e-01 8.49279642e-01
4.02026653e-01 1.30401120e-01 1.55560672e-01 -1.06310260e+00
2.76524462e-02 5.83710194e-01 -7.55457878e-02 -1.01434326e+00
-9.02334690e-01 -4.11825836e-01 -1.85854048e-01 3.82790804e-01
7.30790615e-01 -3.43368828e-01 -1.06884444e+00 1.34669936e+00
5.52132010e-01 6.61562681e-02 -8.69082332e-01 1.13746834e+00
7.37672865e-01 5.26888520e-02 5.18696845e-01 1.37726858e-01
1.43483126e+00 -7.14231849e-01 -6.47721887e-01 1.84512809e-01
1.14780068e+00 -9.07441616e-01 1.64376998e+00 2.82735318e-01
-1.06383550e+00 -2.14045376e-01 -8.12106133e-01 4.12485115e-02
-1.67550981e-01 1.70397416e-01 6.75890028e-01 9.05436099e-01
-1.41489232e+00 6.30839884e-01 -1.30414832e+00 -4.89419818e-01
3.86245996e-01 4.78331745e-01 -4.65443768e-02 2.13679686e-01
-8.52564991e-01 8.60289931e-01 2.24730000e-02 -2.24206537e-01
-3.06987435e-01 -1.15710533e+00 -4.53720748e-01 -2.49690905e-01
3.60258281e-01 -7.58359492e-01 1.26111674e+00 -6.66505456e-01
-1.18248498e+00 8.56198370e-01 1.91530511e-01 -2.94395447e-01
9.23702776e-01 1.46843627e-01 8.73817131e-02 5.12355983e-01
9.69240740e-02 1.22998261e+00 4.62477267e-01 -1.18704903e+00
-3.90116990e-01 -2.72134990e-01 4.86227423e-02 2.90260673e-01
-1.87811568e-01 1.16400898e-01 -8.17255914e-01 -6.47308826e-01
2.80954093e-01 -1.32787156e+00 -5.15892327e-01 4.29738104e-01
-4.94279623e-01 1.73332244e-01 6.62272155e-01 -7.48457611e-01
1.43093896e+00 -2.06110644e+00 -2.78465956e-01 5.16542614e-01
2.12155655e-01 -7.60267153e-02 9.78894159e-02 6.70140311e-02
-3.55973840e-02 5.54100394e-01 -4.12015826e-01 -2.18594104e-01
-1.06965683e-01 -1.72302295e-02 3.05893213e-01 4.62940395e-01
-2.53928363e-01 9.54238713e-01 -8.75281930e-01 -9.61476147e-01
1.58344328e-01 5.92409015e-01 -9.52108502e-01 -6.32451400e-02
9.89660248e-02 7.80441642e-01 -3.82565886e-01 4.36879247e-01
7.39878654e-01 -5.54100275e-01 2.15681374e-01 -3.56976867e-01
-1.54023871e-01 1.30972520e-01 -1.03050816e+00 1.87572372e+00
-2.52173185e-01 4.53711241e-01 2.13806912e-01 -3.17164034e-01
6.67084157e-01 9.22262520e-02 8.70198905e-01 -3.08988184e-01
-2.96109468e-02 7.82090053e-02 3.15290093e-01 -6.58250690e-01
2.26118729e-01 -5.63791208e-03 3.18414241e-01 9.44544852e-01
-3.58787000e-01 -3.61875385e-01 2.04400346e-01 5.59726179e-01
1.15794396e+00 2.03283578e-01 -1.41317800e-01 -5.01486301e-01
4.32562754e-02 2.30308786e-01 6.44440576e-02 9.64787364e-01
-3.79082710e-01 1.01630461e+00 6.37761414e-01 -2.11920232e-01
-7.92596340e-01 -9.55344975e-01 -3.81883323e-01 1.25726867e+00
9.97957215e-02 -5.06168664e-01 -1.12467301e+00 -1.24236286e+00
2.53907908e-02 8.27937245e-01 -8.39134932e-01 3.55470866e-01
-5.66523254e-01 -7.98910975e-01 5.30148387e-01 4.84363198e-01
9.21629667e-02 -8.70894492e-01 -1.19145858e+00 8.14699680e-02
-1.88353077e-01 -7.45829523e-01 -9.99305129e-01 9.30056944e-02
-1.06197572e+00 -1.23885357e+00 -7.71050453e-01 -4.00225312e-01
1.09980559e+00 -3.47094722e-02 9.52687085e-01 3.86657089e-01
-6.85208142e-01 7.19367743e-01 -4.68184710e-01 -1.48996204e-01
-4.72836435e-01 2.24899366e-01 -5.93495786e-01 -4.83123183e-01
1.11113325e-01 -3.01598936e-01 -1.19791305e+00 7.09088981e-01
-1.05741060e+00 4.96541113e-01 4.04314280e-01 5.75605631e-01
6.26375616e-01 -5.83884656e-01 2.69854039e-01 -1.38298750e+00
8.19335163e-01 -3.88746172e-01 -4.49347973e-01 2.46985927e-01
-7.31551051e-01 9.17735174e-02 4.67255153e-02 -3.57107848e-01
-8.90163422e-01 2.42558435e-01 -8.90914500e-02 -9.83236805e-02
-1.36421204e-01 4.31974232e-01 7.01448917e-01 -2.81713247e-01
8.29803944e-01 -2.51644284e-01 1.65179163e-01 -4.11661416e-01
5.23854434e-01 4.40985799e-01 3.60107392e-01 -4.53187704e-01
3.15409422e-01 5.64476728e-01 -1.85974643e-01 -2.47147664e-01
-1.12197235e-01 -4.45187956e-01 -1.03378570e+00 -4.99669969e-01
9.74377275e-01 -1.63306206e-01 -4.81058896e-01 -2.30665147e-01
-7.96048224e-01 -5.15328228e-01 -8.56062099e-02 5.65151930e-01
-4.07709420e-01 5.69673181e-01 -6.43517137e-01 -4.11452770e-01
-4.22511786e-01 -1.69792914e+00 9.99281883e-01 9.91164148e-02
-7.58418918e-01 -1.22832739e+00 -1.10050015e-01 3.15329850e-01
6.72344983e-01 4.47515041e-01 9.58306730e-01 -7.70598888e-01
-5.43599844e-01 2.27255048e-03 -2.35623598e-01 -1.86849207e-01
1.98157579e-01 7.40585774e-02 -5.33496499e-01 -5.61588220e-02
-2.86455870e-01 1.19447589e-01 5.71184218e-01 8.14354300e-01
1.44816446e+00 1.04784869e-01 -4.86220092e-01 7.87457526e-01
1.00514913e+00 3.09305459e-01 4.35432971e-01 4.26683277e-01
4.66447681e-01 6.50515974e-01 7.06226885e-01 3.42907786e-01
2.50153393e-01 6.94128215e-01 3.02042335e-01 -4.01800483e-01
-1.83413312e-01 -1.78666245e-02 -2.78524309e-01 5.20153284e-01
-7.12844171e-03 2.08025113e-01 -1.30925226e+00 3.29116762e-01
-1.53576875e+00 -3.05453897e-01 -3.83452713e-01 2.16123939e+00
9.99606133e-01 1.36790857e-01 1.94558993e-01 -3.19406748e-01
5.93593538e-01 -1.76230833e-01 -4.50667292e-01 -3.21706504e-01
4.10932451e-01 6.95571825e-02 6.74055398e-01 6.35412872e-01
-1.13639128e+00 6.62105262e-01 7.17920208e+00 5.57864964e-01
-1.27174413e+00 1.46924809e-01 8.65856230e-01 -1.64228499e-01
-5.43468833e-01 1.38326257e-01 -3.76162350e-01 5.59408545e-01
5.53390145e-01 2.34595925e-01 2.23296165e-01 6.64823771e-01
6.91925764e-01 -5.02416253e-01 -1.17646515e+00 6.94451690e-01
-1.73836306e-01 -1.28385401e+00 -2.30313674e-01 -4.80719060e-02
4.27008182e-01 -1.39145255e-01 1.05133660e-01 -1.09729022e-01
4.53605354e-01 -1.06133819e+00 2.82009661e-01 5.11256397e-01
9.29969013e-01 -1.65987760e-01 4.53368276e-01 -1.12262130e-01
-7.24107504e-01 3.76253843e-01 2.47619823e-01 6.85801566e-01
3.27782273e-01 3.53932202e-01 -1.71073818e+00 1.52183801e-01
6.03034854e-01 3.28067720e-01 -1.03033543e+00 1.56218469e+00
-3.13747674e-02 7.33581424e-01 -4.64395732e-01 1.06863491e-01
2.09041536e-01 -2.56711274e-01 5.40243804e-01 1.67688906e+00
2.77756333e-01 6.22596368e-02 1.07773222e-01 1.02002823e+00
4.69230592e-01 5.34839571e-01 -2.99028810e-02 1.98986873e-01
4.51700538e-01 1.43309867e+00 -1.54759896e+00 -2.83050418e-01
-1.02158710e-01 7.04088748e-01 -2.60004491e-01 3.33142847e-01
-9.46693957e-01 -3.27946246e-01 -4.44731899e-02 6.97213888e-01
3.16964574e-02 -2.63013989e-01 -8.39543164e-01 -5.74756503e-01
-3.26292902e-01 -6.41433895e-01 7.30369747e-01 -8.54350984e-01
-8.64061415e-01 3.65933567e-01 3.03019613e-01 -1.01638687e+00
-9.04776379e-02 -2.45245010e-01 -6.05146348e-01 8.67470026e-01
-9.39955473e-01 -1.03475130e+00 -5.81233084e-01 3.98145497e-01
3.25936466e-01 6.38174653e-01 5.98053277e-01 3.39866519e-01
-1.49306923e-01 6.23240292e-01 -1.75722942e-01 8.57337192e-02
1.10544884e+00 -1.64608872e+00 2.53195584e-01 3.77242178e-01
-1.16652824e-01 9.90211129e-01 8.12839746e-01 -8.99656713e-01
-9.31418419e-01 -5.73911786e-01 2.38660052e-01 -4.51642066e-01
5.36135077e-01 -1.53770804e-01 -8.73778701e-01 6.35749936e-01
2.44326636e-01 -1.34691428e-02 1.06897044e+00 -1.00652941e-01
1.55463606e-01 3.55074942e-01 -1.34827673e+00 7.32422829e-01
9.35475886e-01 -1.21823866e-02 -2.35535577e-01 2.94140220e-01
4.85978872e-01 -6.14601433e-01 -1.08375990e+00 3.42511535e-01
6.44549549e-01 -1.03299844e+00 8.05205822e-01 -2.68913865e-01
8.52117538e-02 -3.83636236e-01 4.23195243e-01 -1.50632882e+00
-1.77377969e-01 -5.88762045e-01 3.52034509e-01 7.70939171e-01
6.99965954e-01 -4.64560598e-01 8.11914682e-01 1.07255781e+00
-3.42070490e-01 -9.41066325e-01 -6.36620522e-01 3.76245333e-03
-4.70148027e-02 -4.51723069e-01 3.59798133e-01 9.86291289e-01
3.32245469e-01 -3.00128698e-01 1.59220487e-01 -1.40389904e-01
4.94273454e-01 -4.81560789e-02 6.33273542e-01 -9.58743036e-01
-3.16111416e-01 -6.11771941e-01 -5.80530986e-02 -9.58525538e-01
-6.68945551e-01 -1.07746792e+00 -9.28473026e-02 -1.77641118e+00
2.23064721e-01 -1.05988419e+00 -6.87068934e-03 7.62919009e-01
-3.63973290e-01 4.39432651e-01 2.86606759e-01 2.07771063e-01
-6.03938282e-01 -1.10983349e-01 1.56887555e+00 2.08911017e-01
-4.43469852e-01 -1.38739154e-01 -7.78178453e-01 5.75959742e-01
5.79800546e-01 -4.58903283e-01 -4.58516985e-01 -3.87709498e-01
5.44934124e-02 -2.50339732e-02 2.19194412e-01 -6.31625593e-01
3.83091271e-01 9.47328284e-02 4.14783180e-01 -6.50591433e-01
-7.96194971e-02 -6.44874394e-01 1.67940170e-01 6.52487397e-01
-7.42503583e-01 7.79973492e-02 3.60726506e-01 2.35424370e-01
3.17210555e-01 -3.30754727e-01 7.55463421e-01 -1.64050937e-01
-1.27371550e-01 6.45187348e-02 -3.46781164e-01 8.56865942e-02
1.09234154e+00 -4.44018632e-01 -6.63060397e-02 -5.13207734e-01
-1.36738265e+00 5.57919800e-01 5.22525012e-01 1.82724819e-01
4.02755380e-01 -9.03488398e-01 -5.91388702e-01 1.18249804e-01
-1.54144734e-01 -3.18655483e-02 3.69780391e-01 1.58526170e+00
-9.82588649e-01 2.65201211e-01 -4.23057377e-02 -1.03365600e+00
-1.52531481e+00 1.08987227e-01 4.06713426e-01 -1.94878042e-01
-8.07545364e-01 7.59291291e-01 2.07233101e-01 -5.24433315e-01
2.58684278e-01 -5.16045392e-01 -2.49724716e-01 3.42716984e-02
3.22762877e-01 3.27347636e-01 3.03174611e-02 -2.08978623e-01
-3.75235021e-01 4.01164532e-01 -2.27931246e-01 -2.81411082e-01
1.15128076e+00 -2.92615384e-01 1.36433065e-01 1.31452978e-01
9.62921619e-01 8.27814713e-02 -1.40288746e+00 2.03338698e-01
1.74057305e-01 -7.07983792e-01 -7.08573684e-03 -1.29233205e+00
-1.05827534e+00 7.52106607e-01 1.07890010e+00 3.71831089e-01
9.16842878e-01 3.49646881e-02 4.65909839e-01 -3.62492174e-01
1.11825868e-01 -1.06125736e+00 1.77255258e-01 -2.38185562e-02
6.95799470e-01 -1.04713929e+00 6.47062212e-02 -5.70105135e-01
-1.00323343e+00 1.13515747e+00 5.70078552e-01 1.37705445e-01
6.72545016e-01 6.37591720e-01 4.75853622e-01 -4.17935401e-01
-3.05274576e-01 1.41398102e-01 3.62779796e-01 5.32135367e-01
7.57499814e-01 8.76711830e-02 -7.79590428e-01 2.81408817e-01
-2.93000072e-01 7.38624716e-03 4.63786840e-01 9.81556714e-01
-3.76051158e-01 -1.05159640e+00 -6.26625299e-01 6.76832855e-01
-6.93021953e-01 -1.93710357e-01 -1.55662864e-01 7.20987439e-01
-3.27694986e-04 6.39707506e-01 -1.00554340e-02 -1.54427126e-01
1.99439660e-01 1.23746268e-01 2.58837610e-01 -7.49054074e-01
-9.49668884e-01 3.25858057e-01 -3.32222669e-03 -6.71311855e-01
-1.56534806e-01 -7.53584743e-01 -1.56379735e+00 1.27618555e-02
-4.05199498e-01 9.41432044e-02 1.02918887e+00 5.74452460e-01
4.59633410e-01 5.45624256e-01 1.19848162e-01 -6.90269589e-01
-3.11096251e-01 -9.08736110e-01 -1.96239159e-01 6.27016664e-01
-1.86262190e-01 -6.35069132e-01 -2.06724674e-01 1.66564599e-01] | [14.709507942199707, -2.305349826812744] |
233c4236-88db-4691-9b3c-0cf33545a754 | is-end-to-end-learning-enough-for-fitness | 2305.08191 | null | https://arxiv.org/abs/2305.08191v1 | https://arxiv.org/pdf/2305.08191v1.pdf | Is end-to-end learning enough for fitness activity recognition? | End-to-end learning has taken hold of many computer vision tasks, in particular, related to still images, with task-specific optimization yielding very strong performance. Nevertheless, human-centric action recognition is still largely dominated by hand-crafted pipelines, and only individual components are replaced by neural networks that typically operate on individual frames. As a testbed to study the relevance of such pipelines, we present a new fully annotated video dataset of fitness activities. Any recognition capabilities in this domain are almost exclusively a function of human poses and their temporal dynamics, so pose-based solutions should perform well. We show that, with this labelled data, end-to-end learning on raw pixels can compete with state-of-the-art action recognition pipelines based on pose estimation. We also show that end-to-end learning can support temporally fine-grained tasks such as real-time repetition counting. | ['Roland Memisevic', 'Ingo Bax', 'Nahua Kang', 'Cornelius Boehm', 'Florian Letsch', 'Sunny Panchal', 'Guillaume Berger', 'Antoine Mercier'] | 2023-05-14 | null | null | null | null | ['action-recognition-in-videos'] | ['computer-vision'] | [ 3.37723643e-01 -1.08389460e-01 -9.66517702e-02 -2.92438865e-01
-7.75743306e-01 -4.15560395e-01 7.15075314e-01 -1.72666341e-01
-8.13533604e-01 4.97166365e-01 2.85131782e-01 1.79855451e-01
1.14349529e-01 -2.69360900e-01 -8.92876267e-01 -5.69268107e-01
-4.24557418e-01 7.31987298e-01 8.96166444e-01 -2.51445860e-01
1.56970680e-01 5.12945354e-01 -1.75006914e+00 5.57575285e-01
-1.91509366e-01 1.03569353e+00 8.38160440e-02 1.12756455e+00
1.81177065e-01 1.09165406e+00 -5.53717315e-01 -1.37525931e-01
2.21831292e-01 -3.20127964e-01 -8.01409304e-01 3.75593215e-01
6.84098363e-01 -4.04315859e-01 -5.33613384e-01 4.69704092e-01
5.45387089e-01 1.78867400e-01 3.46320987e-01 -1.19750786e+00
1.97596177e-01 1.06097810e-01 -4.46720481e-01 3.64837110e-01
5.15015900e-01 6.89512551e-01 7.77579427e-01 -7.81050920e-01
7.74681211e-01 1.14868259e+00 6.38685644e-01 4.92226422e-01
-1.07606781e+00 -8.12836587e-02 3.90971005e-01 4.82887059e-01
-8.88493836e-01 -6.62894785e-01 5.26526928e-01 -5.10264516e-01
1.38245690e+00 9.11303461e-02 9.50343847e-01 1.38676476e+00
1.99804336e-01 1.21229136e+00 8.39000821e-01 -3.76248032e-01
2.79999554e-01 -6.73484743e-01 -3.90017390e-01 8.10272455e-01
-8.96160752e-02 2.28534028e-01 -8.85249197e-01 3.85390341e-01
8.71832848e-01 -7.56784063e-03 -4.51327786e-02 -8.23520660e-01
-1.62080550e+00 4.56720978e-01 1.74835607e-01 2.45797217e-01
-3.46800297e-01 6.85506463e-01 6.58819139e-01 -3.66887040e-02
2.58618534e-01 3.02828282e-01 -8.61045003e-01 -7.28885889e-01
-1.22073019e+00 3.06063563e-01 6.27138555e-01 6.43328726e-01
4.66254056e-01 -5.83915450e-02 -4.30049866e-01 2.75979012e-01
8.86696577e-02 2.14591816e-01 4.19621319e-01 -1.30391157e+00
2.10149258e-01 3.93510342e-01 1.74270630e-01 -4.98952448e-01
-8.23927045e-01 -4.43698645e-01 -2.96235502e-01 5.21461666e-01
9.72189844e-01 -6.24264590e-02 -9.51153457e-01 1.51104474e+00
3.47057581e-01 2.89399415e-01 -2.91046441e-01 1.09398055e+00
2.10188344e-01 1.51654109e-01 8.29564333e-02 -7.46360198e-02
1.25989306e+00 -1.44394588e+00 -3.53541732e-01 -5.54080069e-01
6.66077971e-01 -4.13634151e-01 8.24951470e-01 6.94724202e-01
-1.22610903e+00 -5.39817870e-01 -7.44480133e-01 -8.88097957e-02
-1.85378641e-01 1.85280830e-01 7.12986767e-01 4.79610503e-01
-8.58447492e-01 8.18914592e-01 -1.41845751e+00 -6.87124789e-01
5.82406580e-01 2.62718201e-01 -6.19566619e-01 1.32355496e-01
-6.14857912e-01 9.94198382e-01 3.82924765e-01 1.29438415e-01
-1.24680889e+00 -3.52022439e-01 -8.83784711e-01 -2.62655854e-01
7.24932969e-01 -6.63021207e-01 1.50919080e+00 -1.11090004e+00
-1.60211635e+00 1.04460871e+00 2.31076665e-02 -8.45501959e-01
9.95515466e-01 -5.96022964e-01 -3.22953947e-02 3.56679618e-01
6.11265860e-02 9.30625200e-01 8.46399307e-01 -7.39431798e-01
-8.21624935e-01 -5.01225293e-01 5.74655384e-02 9.12448466e-02
-1.44086316e-01 2.53335565e-01 -7.44064808e-01 -4.23967570e-01
-1.07460529e-01 -9.97795403e-01 -3.77632171e-01 4.48625237e-01
3.93337943e-02 -8.89955610e-02 9.37988102e-01 -6.55317545e-01
7.41369247e-01 -1.88629270e+00 2.83276141e-01 -4.26611423e-01
-1.69121340e-01 2.33658671e-01 -2.05132544e-01 2.74603695e-01
-8.00948068e-02 -5.00204921e-01 -1.46245822e-01 -5.20047665e-01
9.86204520e-02 2.76061535e-01 7.16080293e-02 9.00491953e-01
3.30109566e-01 1.15400589e+00 -1.06003988e+00 -5.63360691e-01
7.43970931e-01 3.33968252e-01 -4.20770645e-01 1.85153782e-01
-7.12590337e-01 8.04731965e-01 -2.28300661e-01 6.90698385e-01
-8.84297416e-02 -2.65589744e-01 7.67901540e-02 -7.39966705e-02
-1.18172735e-01 -1.32406309e-01 -1.02660179e+00 2.32552552e+00
-1.79844677e-01 6.62384450e-01 1.21204242e-01 -1.16773033e+00
3.95769298e-01 1.92813203e-01 1.00015199e+00 -6.22938097e-01
1.02379888e-01 5.85981719e-02 7.55733848e-02 -5.47135770e-01
3.31442088e-01 1.59792170e-01 -7.02873468e-02 2.64535636e-01
3.47623199e-01 9.89265665e-02 4.72579867e-01 -6.31531850e-02
1.60671234e+00 1.14717948e+00 3.16293210e-01 1.85483590e-01
4.33520854e-01 3.97138596e-01 4.14321244e-01 5.81995249e-01
-6.11910760e-01 9.37884092e-01 3.59429270e-01 -7.87195921e-01
-1.17803645e+00 -7.06342876e-01 2.65741851e-02 1.38983786e+00
-4.73359153e-02 -3.36306989e-01 -8.26170385e-01 -7.74895489e-01
-3.18473786e-01 2.59190351e-01 -6.25836134e-01 1.26777768e-01
-8.27507794e-01 -4.29020196e-01 5.33584654e-01 8.77970815e-01
4.76379484e-01 -1.43207836e+00 -1.43987608e+00 4.70810592e-01
1.87472790e-01 -1.43489099e+00 -2.24057958e-01 4.32590336e-01
-8.05900872e-01 -1.39784312e+00 -9.93943751e-01 -2.82515705e-01
1.54841617e-01 -3.76676023e-02 1.31671524e+00 -2.25014567e-01
-6.20095909e-01 8.30799460e-01 -4.87495720e-01 -1.84839457e-01
-8.32036138e-02 -9.81475264e-02 8.91112350e-03 1.49460986e-01
2.33729064e-01 -6.39071286e-01 -8.77844155e-01 4.85603869e-01
-6.51450038e-01 -1.24865659e-01 8.70751858e-01 6.64628625e-01
6.03464603e-01 -2.28637159e-01 -4.62195314e-02 -4.54814732e-01
-8.50561932e-02 1.64890826e-01 -6.37664855e-01 2.82497555e-01
-4.03946592e-03 8.47543702e-02 3.81683975e-01 -3.84780794e-01
-7.50757694e-01 1.01261079e+00 2.10709292e-02 -7.10027695e-01
-5.81451476e-01 1.51133671e-01 -1.07244797e-01 -4.21930104e-02
8.44991446e-01 2.65436530e-01 5.55498749e-02 -3.83773565e-01
4.79440778e-01 1.40440807e-01 9.73801970e-01 -5.02716005e-01
3.52134794e-01 7.06968606e-01 1.88109145e-01 -8.15996885e-01
-8.57482314e-01 -5.87727606e-01 -1.08396041e+00 -5.76976955e-01
1.28609610e+00 -1.05627024e+00 -7.65558481e-01 6.41134620e-01
-1.11507177e+00 -7.62952685e-01 -5.19095004e-01 4.97879654e-01
-1.25457442e+00 3.51678848e-01 -3.84732425e-01 -9.07608449e-01
1.46245301e-01 -9.61978853e-01 1.68556929e+00 -7.56992027e-02
-3.66217345e-01 -7.75099993e-01 2.49540761e-01 6.54781401e-01
1.00531645e-01 5.06001770e-01 1.21929534e-01 -5.12743890e-01
-8.32830846e-01 -4.38535720e-01 9.20890495e-02 2.29164332e-01
-4.55003351e-01 -7.81947225e-02 -1.05470729e+00 -2.39973217e-01
-2.69367486e-01 -7.45446742e-01 1.04291832e+00 3.72177929e-01
1.15541840e+00 1.70201227e-01 -2.41274238e-01 5.41267991e-01
1.01941454e+00 -2.68951327e-01 7.78376877e-01 4.80931461e-01
7.73855209e-01 5.71982026e-01 7.45856762e-01 5.43881893e-01
1.58473462e-01 1.08383334e+00 7.74215043e-01 8.08004364e-02
-2.71993965e-01 -6.93905577e-02 5.64278364e-01 3.48513350e-02
-5.27477801e-01 -8.09220728e-05 -8.88533175e-01 4.69575226e-01
-2.29972148e+00 -1.25399303e+00 -1.74067855e-01 2.12215185e+00
4.84990209e-01 3.83800238e-01 6.51568472e-01 1.24750055e-01
3.94671470e-01 4.24456179e-01 -6.52589798e-01 2.09164768e-01
-1.00009507e-02 3.15312713e-01 7.88768589e-01 6.33619651e-02
-1.55190992e+00 9.66720402e-01 6.80110264e+00 6.75371528e-01
-9.64718640e-01 1.31828725e-01 4.64639276e-01 -4.28375393e-01
6.17602170e-01 -5.07089607e-02 -4.65397298e-01 2.65917599e-01
9.03069794e-01 5.71598709e-01 3.01103711e-01 1.09075534e+00
4.22662169e-01 -4.01903898e-01 -1.54241288e+00 1.11277413e+00
-7.02224718e-03 -1.17833138e+00 -5.32989562e-01 6.66016713e-02
5.79192221e-01 2.78686970e-01 -3.23921114e-01 3.36637735e-01
2.47889593e-01 -1.10261989e+00 9.25295830e-01 5.06758451e-01
5.03578246e-01 -5.06318688e-01 4.58411276e-01 5.20565569e-01
-1.12028849e+00 -1.49329379e-01 -1.93880111e-01 -3.31109017e-01
4.60952282e-01 3.70790869e-01 -5.15191138e-01 2.62499511e-01
6.71126842e-01 9.93349314e-01 -8.21057320e-01 1.14809203e+00
-3.21809828e-01 2.68525839e-01 -3.21871758e-01 2.23534480e-01
2.89313197e-01 1.81895211e-01 3.24713141e-01 1.37460601e+00
1.50134578e-01 -8.13302621e-02 4.78851557e-01 2.04775050e-01
3.28021765e-01 -4.01351392e-01 -3.97965461e-01 -1.66896433e-01
-2.08017856e-01 1.30566239e+00 -9.81723070e-01 -2.33175710e-01
-3.62046272e-01 1.30587816e+00 3.63873780e-01 5.53157367e-02
-9.73073244e-01 2.03548327e-01 6.79101288e-01 3.12006444e-01
7.73814380e-01 -5.89984179e-01 -5.80753088e-02 -1.19513834e+00
1.42222300e-01 -9.86080885e-01 3.29619735e-01 -7.82552302e-01
-8.37138474e-01 2.32461080e-01 -1.42121419e-01 -1.28281271e+00
-6.40138984e-01 -9.53846812e-01 -4.56977814e-01 2.61800766e-01
-1.03525257e+00 -1.27426779e+00 -3.93576592e-01 8.60988677e-01
8.47824395e-01 -9.22075808e-02 8.51075351e-01 2.05247402e-01
-4.41225469e-01 1.80360720e-01 -3.38557750e-01 3.10685754e-01
5.85561514e-01 -1.27906191e+00 5.02822101e-01 9.68357027e-01
7.33412802e-01 -1.50389194e-01 7.86665499e-01 -4.54917610e-01
-1.57383204e+00 -9.47317004e-01 4.45247591e-01 -9.60146964e-01
6.43655598e-01 -4.06880945e-01 -4.37912971e-01 7.28948832e-01
4.30488996e-02 6.97363734e-01 9.78585705e-02 4.15690094e-02
-1.24602884e-01 1.41220624e-02 -6.07033074e-01 4.10051972e-01
1.59999847e+00 -4.85175848e-01 -3.03829491e-01 7.97679067e-01
2.71529377e-01 -6.54750705e-01 -5.19733012e-01 4.04505104e-01
6.69061303e-01 -1.33749843e+00 1.13958836e+00 -9.42759693e-01
5.41249752e-01 -4.22590762e-01 -4.61295694e-02 -8.38552713e-01
-2.41029918e-01 -6.52302206e-01 -4.62118626e-01 5.86268961e-01
6.77511618e-02 1.06259085e-01 1.26704431e+00 5.23636758e-01
-9.96449217e-02 -6.40000820e-01 -1.00568509e+00 -9.96318936e-01
-3.80092204e-01 -7.66037226e-01 -1.43796951e-01 3.24393839e-01
-2.63234437e-01 3.28379393e-01 -5.52851975e-01 -9.09653306e-02
5.24711430e-01 -6.32415786e-02 1.13260877e+00 -8.93304825e-01
-7.99273312e-01 -4.67861891e-01 -1.01877081e+00 -1.03190053e+00
3.95472869e-02 -2.82054931e-01 3.26542050e-01 -1.46715212e+00
-1.86399799e-02 1.64524198e-01 -3.89097445e-02 4.93408620e-01
1.19413592e-01 5.00531018e-01 2.35262766e-01 9.94346663e-02
-1.25373852e+00 3.10820788e-01 1.13079345e+00 -1.28677543e-02
1.77276924e-01 7.94222504e-02 9.32698995e-02 9.61666346e-01
4.64524120e-01 -2.84152180e-01 -2.83733994e-01 -3.31363350e-01
4.03421521e-02 2.24546835e-01 7.07467318e-01 -1.55298531e+00
3.99782628e-01 -2.90326715e-01 6.37464523e-01 -5.35146117e-01
7.16324568e-01 -8.94767225e-01 5.30771762e-02 5.55321336e-01
-3.63022208e-01 -1.78108498e-01 -7.10156411e-02 7.59105742e-01
-2.23084450e-01 2.55373996e-02 6.64679646e-01 -5.93160033e-01
-1.26272178e+00 3.66333902e-01 -2.62090147e-01 2.63234317e-01
1.39250910e+00 -5.34304976e-01 2.78054308e-02 -3.58218104e-01
-1.05521870e+00 2.39341352e-02 4.25432324e-01 3.26439112e-01
3.32194000e-01 -1.04032207e+00 -7.15688884e-01 -1.52108803e-01
2.04001367e-01 1.49670362e-01 1.70175761e-01 1.04519796e+00
-6.06127322e-01 5.41934252e-01 -3.87771696e-01 -9.46876466e-01
-1.27223766e+00 6.43275976e-01 4.31689024e-01 -5.19615769e-01
-7.17891753e-01 9.82447743e-01 1.02906153e-01 -5.45743480e-02
5.90112388e-01 -1.53131947e-01 8.31392780e-02 7.34397322e-02
5.48105121e-01 1.91166520e-01 2.23804623e-01 -5.77062368e-01
-6.69281602e-01 4.91708040e-01 2.60252535e-01 -3.81145865e-01
1.43706334e+00 2.64392614e-01 3.38056594e-01 3.26497138e-01
1.03306174e+00 -5.20121098e-01 -2.19150233e+00 1.37668818e-01
9.63122025e-02 -4.83286560e-01 -6.64057434e-02 -7.89974749e-01
-1.14623272e+00 9.88148570e-01 5.96426666e-01 -1.95335060e-01
9.96544957e-01 -4.94191796e-03 5.88508666e-01 5.45226753e-01
7.78677821e-01 -1.36973298e+00 6.28508627e-01 6.00892246e-01
7.49945402e-01 -1.31226039e+00 1.28941461e-01 -1.09799214e-01
-6.13290370e-01 1.26836550e+00 6.08955026e-01 -2.27674246e-01
1.86582983e-01 3.10986996e-01 -1.37149185e-01 -1.63772315e-01
-7.60525882e-01 -6.31929159e-01 1.53326899e-01 7.31961191e-01
2.87983865e-01 -2.22982317e-01 3.96934226e-02 1.56985059e-01
1.16707243e-01 3.43764514e-01 2.66995907e-01 1.24587429e+00
-5.32684326e-01 -1.13355541e+00 -3.54827225e-01 1.41377002e-01
-3.50443631e-01 2.58808106e-01 -4.35859650e-01 7.58615613e-01
1.87593833e-01 6.69882059e-01 4.67132665e-02 -2.60288745e-01
3.87684047e-01 2.45154887e-01 1.03959143e+00 -6.02243662e-01
-5.64645290e-01 6.43405691e-02 3.26077968e-01 -1.36947370e+00
-7.58756757e-01 -1.06086016e+00 -1.06259775e+00 2.44164951e-02
1.56546891e-01 -5.93768954e-01 5.88221610e-01 1.22072577e+00
1.92336872e-01 6.86160624e-01 2.48677153e-02 -1.43617976e+00
-5.52630067e-01 -8.20678234e-01 -4.54726517e-01 4.93423760e-01
2.53326893e-01 -7.12716103e-01 2.67977025e-02 2.91461408e-01] | [8.046395301818848, 0.3584420084953308] |
e75bbd80-d7e6-49bc-a36a-83f71f4d71b1 | multi-task-image-based-dietary-assessment-for | 2004.13188 | null | https://arxiv.org/abs/2004.13188v1 | https://arxiv.org/pdf/2004.13188v1.pdf | Multi-Task Image-Based Dietary Assessment for Food Recognition and Portion Size Estimation | Deep learning based methods have achieved impressive results in many applications for image-based diet assessment such as food classification and food portion size estimation. However, existing methods only focus on one task at a time, making it difficult to apply in real life when multiple tasks need to be processed together. In this work, we propose an end-to-end multi-task framework that can achieve both food classification and food portion size estimation. We introduce a food image dataset collected from a nutrition study where the groundtruth food portion is provided by registered dietitians. The multi-task learning uses L2-norm based soft parameter sharing to train the classification and regression tasks simultaneously. We also propose the use of cross-domain feature adaptation together with normalization to further improve the performance of food portion size estimation. Our results outperforms the baseline methods for both classification accuracy and mean absolute error for portion estimation, which shows great potential for advancing the field of image-based dietary assessment. | ['Fengqing Zhu', 'Carol Boushey', 'Zeman Shao', 'Janine Wright', 'Jiangpeng He', 'Deborah Kerr'] | 2020-04-27 | null | null | null | null | ['food-recognition'] | ['computer-vision'] | [ 2.18715653e-01 -4.83564168e-01 -5.57360053e-01 -7.65369773e-01
-9.13788855e-01 -4.14980322e-01 -6.07362054e-02 1.03176141e+00
-4.96603936e-01 9.30296108e-02 3.32186073e-01 1.94102883e-01
2.13373244e-01 -8.36131454e-01 -9.51654255e-01 -6.88841879e-01
-8.63517523e-02 2.01011777e-01 -1.63059637e-01 3.57942767e-02
-3.58983845e-01 -4.24226284e-01 -1.29818225e+00 3.86542171e-01
7.76848495e-01 1.03388977e+00 4.43437636e-01 4.92785454e-01
1.06338121e-01 5.59830666e-01 -8.31583440e-02 -1.08150125e-01
4.83021855e-01 -4.46083844e-01 -1.49047509e-01 1.73965335e-01
6.54044986e-01 -6.83925986e-01 1.05482526e-01 1.26071429e+00
7.65762150e-01 1.88088998e-01 5.93033791e-01 -7.93159008e-01
-9.89482939e-01 7.29228616e-01 -9.47988033e-01 1.05370253e-01
6.60735741e-02 1.47071570e-01 6.97803438e-01 -4.89159286e-01
-1.44482478e-01 1.15136683e+00 9.06207979e-01 1.86858073e-01
-1.36229146e+00 -9.17409837e-01 1.37575746e-01 2.34162673e-01
-1.18093741e+00 -3.75496686e-01 6.47549093e-01 -6.34531379e-01
7.63975024e-01 -2.24735424e-01 8.65422428e-01 7.01508582e-01
2.50349939e-01 8.74404907e-01 1.07329440e+00 -3.19106400e-01
5.04793338e-02 -1.05536252e-01 2.28241220e-01 8.44284296e-01
4.41395193e-01 1.43356293e-01 -1.66996986e-01 3.94724347e-02
8.35368752e-01 2.04498172e-01 2.66148210e-01 -4.50888872e-01
-1.40518379e+00 1.23433769e+00 6.16945207e-01 2.81441491e-02
-6.44779027e-01 2.96860058e-02 6.10251367e-01 2.43390620e-01
9.04808462e-01 -5.91840148e-02 -6.17452860e-01 6.62508786e-01
-1.05634940e+00 3.50943178e-01 4.29728329e-01 4.54639554e-01
3.68777394e-01 -1.46420196e-01 -2.90317833e-01 1.17810607e+00
5.10763824e-01 9.79987204e-01 5.64305127e-01 -5.38039267e-01
1.17936477e-01 5.17515361e-01 2.47600868e-01 -1.04582155e+00
-9.25928891e-01 -4.65522856e-01 -9.01153743e-01 9.73772928e-02
6.14834666e-01 -8.82437304e-02 -7.90112197e-01 1.78320277e+00
8.05453956e-01 -1.88309103e-01 -3.82973962e-02 1.34600449e+00
1.09657502e+00 4.77368474e-01 7.50463486e-01 -7.20952824e-02
1.89127660e+00 -1.20274210e+00 -7.67422855e-01 -2.12290734e-01
6.51478469e-01 -9.23553765e-01 1.02843750e+00 3.09704602e-01
-9.27852094e-01 -8.59212339e-01 -1.26947200e+00 -2.37712175e-01
-2.11309165e-01 5.14198482e-01 6.48761630e-01 5.74476957e-01
-2.82354772e-01 2.27275193e-01 -1.15405965e+00 -1.29668698e-01
2.88589746e-01 2.25026354e-01 -2.12941855e-01 -3.01568210e-01
-9.57742274e-01 6.54111505e-01 4.59010869e-01 1.37693423e-03
-7.22977877e-01 -1.05895495e+00 -1.31546056e+00 1.98273268e-02
4.10745054e-01 -9.80174959e-01 1.41112709e+00 -1.08878803e+00
-1.40111327e+00 1.12285924e+00 2.54532337e-01 -4.71823037e-01
2.87065893e-01 -2.98130423e-01 -3.89772803e-01 5.96734472e-02
3.79928350e-01 7.83817112e-01 5.54187477e-01 -5.76564729e-01
-5.65365016e-01 -6.51480854e-01 -1.54540777e-01 2.59131461e-01
-1.94426075e-01 1.44651949e-01 1.55431286e-01 -7.96769500e-01
-6.22572638e-02 -6.33644998e-01 -2.85136521e-01 5.11994421e-01
-1.70881942e-01 -2.39364013e-01 1.99349865e-01 -9.13441777e-01
8.68590057e-01 -2.15916800e+00 -1.47083580e-01 -2.55000114e-01
1.23828337e-01 1.97868034e-01 -3.02408904e-01 -1.39616936e-01
-9.23336670e-02 -2.60875940e-01 -1.02628700e-01 -9.32203680e-02
1.76902905e-01 -2.40054265e-01 5.56798577e-01 9.24637794e-01
-1.12229818e-02 9.31134045e-01 -9.95646060e-01 -4.83523548e-01
6.30161464e-01 5.24244905e-01 -5.46211064e-01 1.50277868e-01
-4.75558013e-01 3.53831142e-01 -4.05718774e-01 5.31363368e-01
8.19824696e-01 -4.47542787e-01 1.58555105e-01 -8.30721676e-01
-9.23034251e-02 -1.39446616e-01 -1.14055812e+00 2.13186717e+00
-5.48437357e-01 -1.95747569e-01 2.06649899e-01 -1.17774463e+00
5.50277650e-01 2.96643734e-01 8.25994909e-01 -1.06105042e+00
2.60391593e-01 7.91425332e-02 1.78957522e-01 -7.97992587e-01
8.30979180e-03 -3.56136948e-01 -8.43664706e-02 3.53496820e-01
-5.53966910e-02 3.69043678e-01 4.20463562e-01 -5.40290654e-01
5.74965954e-01 4.05999750e-01 8.01519513e-01 -6.01655126e-01
4.42806453e-01 3.12702060e-01 5.49348354e-01 1.89813912e-01
-4.79681730e-01 2.27078915e-01 -3.80202144e-01 -6.55005634e-01
-1.25691235e+00 -1.10716760e+00 -1.76316187e-01 1.52305841e+00
-1.26678701e-02 1.33264335e-02 -5.63557088e-01 -6.51463747e-01
3.86303961e-01 3.14707875e-01 -5.89458168e-01 -2.68095583e-01
-6.37565434e-01 -1.12673700e+00 3.73297408e-02 6.49542511e-01
7.26442635e-01 -6.54340506e-01 -7.58771122e-01 6.50186718e-01
-4.66219097e-01 -9.77688789e-01 -1.08138573e+00 1.56222746e-01
-6.47349238e-01 -1.30410779e+00 -1.07015169e+00 -1.00397897e+00
3.84393901e-01 5.68287194e-01 1.16178215e+00 -2.76422411e-01
-4.86304134e-01 2.51515955e-01 -3.88874173e-01 -6.35037243e-01
-4.52308685e-01 -4.17890660e-02 -7.18719438e-02 -2.35007927e-01
8.32165122e-01 -2.12985083e-01 -1.36285222e+00 2.55967379e-01
-7.92267382e-01 1.74441904e-01 5.10485530e-01 6.70929432e-01
1.09361660e+00 -7.95663800e-03 6.74583375e-01 -5.61321616e-01
1.36015445e-01 -6.20639384e-01 -4.93360788e-01 2.20737636e-01
-4.84633625e-01 1.35590537e-02 5.02387106e-01 -7.65886188e-01
-6.66636229e-01 3.36537391e-01 -2.94808447e-01 2.77086068e-02
-9.77671891e-02 5.57867944e-01 1.50591284e-01 1.37683421e-01
7.85507679e-01 -1.36289718e-02 2.25978255e-01 -6.58341467e-01
2.26409867e-01 4.88615990e-01 3.55288595e-01 -2.39160597e-01
2.79058754e-01 1.35341093e-01 -3.83968391e-02 -6.52395606e-01
-1.27046609e+00 -7.14967310e-01 -4.17618543e-01 -2.08112039e-02
1.18497753e+00 -1.61306584e+00 -7.16435552e-01 4.61084664e-01
-4.17860508e-01 -6.03659630e-01 -2.64624916e-02 1.01075292e+00
-4.75394100e-01 3.04043412e-01 -5.83180368e-01 -2.55966395e-01
-1.03719115e+00 -1.13458848e+00 1.18186986e+00 8.63065645e-02
1.05255488e-02 -9.72642899e-01 2.03920126e-01 4.87290621e-01
4.49774921e-01 6.44539416e-01 1.00152898e+00 -3.71413082e-01
2.14007348e-01 1.80364330e-03 -5.14835894e-01 2.38984838e-01
2.77276397e-01 -6.59785569e-01 -5.62823951e-01 -4.28715944e-01
-1.20084558e-03 -5.92800915e-01 9.55167949e-01 1.21609235e+00
9.10472572e-01 -2.34166812e-02 -9.30879414e-02 6.62171960e-01
1.69831848e+00 -4.25072908e-02 -1.88307613e-02 2.61756897e-01
5.99040926e-01 3.77917618e-01 4.87059593e-01 4.78197753e-01
9.11480308e-01 7.90219128e-01 1.72211409e-01 -6.34752154e-01
-6.64737165e-01 -1.04651205e-01 3.68389726e-01 5.84702849e-01
5.07027090e-01 6.17980547e-02 -4.56009895e-01 4.31760937e-01
-1.73090041e+00 -6.68925345e-01 -2.79249579e-01 2.19278622e+00
1.02110219e+00 -4.18096721e-01 7.51223683e-01 -1.58910424e-01
6.97713792e-01 -1.22041188e-01 -9.66264069e-01 3.87319066e-02
1.38728008e-01 9.14114788e-02 7.13332653e-01 3.51374336e-02
-1.68479931e+00 3.68631184e-01 6.11289549e+00 4.22928393e-01
-1.00450742e+00 6.06726706e-01 6.38189912e-01 -3.25248986e-01
2.00871870e-01 -9.35628176e-01 -5.99203408e-01 4.19497371e-01
9.71627474e-01 2.27318928e-01 3.97445470e-01 6.82892919e-01
4.19012159e-01 -3.33390921e-01 -9.77495193e-01 1.03192770e+00
6.12277538e-02 -7.32679069e-01 -5.04297554e-01 -2.61620104e-01
6.41585648e-01 1.89707920e-01 3.26934531e-02 3.77273709e-01
3.49576503e-01 -7.16004491e-01 8.70124280e-01 1.12113394e-01
8.59833837e-01 -6.18363857e-01 4.58679110e-01 2.32776761e-01
-1.58160269e+00 -1.06978565e-01 -5.57079136e-01 3.35296355e-02
4.10676934e-02 1.02107680e+00 -2.58227348e-01 3.29989880e-01
5.86722374e-01 6.77663565e-01 -3.05100590e-01 1.18578863e+00
1.23072997e-01 4.27784801e-01 -5.98521113e-01 8.86089578e-02
1.53462812e-01 -2.75149763e-01 8.46079923e-03 1.24009550e+00
3.84784728e-01 6.83962628e-02 1.05728006e+00 7.86387980e-01
-2.08704011e-03 7.42915511e-01 -1.69513300e-02 1.74747333e-01
-1.55432567e-01 1.38523293e+00 -9.25080419e-01 -2.75835812e-01
-8.02082181e-01 8.64798069e-01 9.78347510e-02 -1.35699987e-01
-9.32694554e-01 7.76685327e-02 6.56611562e-01 6.29668161e-02
4.41889822e-01 -2.51837850e-01 -2.14473680e-02 -1.19037938e+00
-1.50401101e-01 -8.47431064e-01 7.05448568e-01 -2.11694762e-01
-1.47594523e+00 -1.00305013e-01 8.55326429e-02 -8.72493327e-01
1.23206545e-02 -4.96538460e-01 -1.35231704e-01 6.70846820e-01
-1.80360115e+00 -1.43823433e+00 -3.49290431e-01 4.20122921e-01
7.11760044e-01 2.32592791e-01 9.86677527e-01 8.41292560e-01
-7.84645796e-01 8.31987977e-01 2.46461153e-01 1.48760960e-01
9.78358150e-01 -1.25991762e+00 8.80380645e-02 4.52919245e-01
-1.34921119e-01 1.90860584e-01 7.31475413e-01 -7.87340164e-01
-1.47107780e+00 -1.46424520e+00 3.50389957e-01 1.22750133e-01
2.55483478e-01 -2.05512390e-01 -6.11621082e-01 3.46855432e-01
-3.26451249e-02 1.89677894e-01 1.18512619e+00 7.24850073e-02
-4.49416399e-01 -3.35632712e-01 -1.54374778e+00 4.07979898e-02
7.13702261e-01 -5.18527292e-02 -8.87301117e-02 7.58503079e-01
5.37109613e-01 -5.12739062e-01 -1.34168994e+00 3.15476477e-01
9.65923011e-01 -4.96968687e-01 1.30729783e+00 -2.95847028e-01
4.48411286e-01 -2.39076257e-01 -3.52485329e-01 -1.40512490e+00
-7.16527462e-01 3.18675458e-01 -1.52547643e-01 8.45696032e-01
2.04901740e-01 -2.95322031e-01 5.61512470e-01 2.17240632e-01
-6.83313087e-02 -5.23131847e-01 -4.53890920e-01 -6.85832858e-01
2.46458501e-01 1.19440749e-01 6.72529578e-01 8.32596600e-01
-1.81779414e-01 3.37288648e-01 -3.77662033e-01 8.30260515e-02
1.03897202e+00 5.93634665e-01 4.28694576e-01 -1.05101228e+00
-3.55311990e-01 -3.51322293e-01 -2.54556239e-02 -8.20342064e-01
-9.32672843e-02 -1.22845280e+00 3.01971823e-01 -1.55411422e+00
7.68228590e-01 -1.84531838e-01 -4.87000644e-01 6.33586824e-01
-4.39037323e-01 5.47221839e-01 4.58064079e-02 5.51953027e-03
-4.85255688e-01 2.50018686e-01 1.37035787e+00 -4.65459615e-01
-3.50175053e-01 -1.72863945e-01 -8.33465636e-01 5.29370785e-01
1.04080355e+00 -5.86667061e-01 -3.75727296e-01 -6.73283219e-01
7.12511316e-02 -4.48051915e-02 3.18517148e-01 -1.06384385e+00
-3.20897996e-01 -1.06962457e-01 9.10778582e-01 -4.09013540e-01
-1.65302753e-01 -8.28453124e-01 3.01404625e-01 9.00559247e-01
-5.22143483e-01 -1.43049896e-01 2.22129226e-01 4.83151734e-01
2.45876566e-01 -6.15406269e-03 1.02191985e+00 -3.65426064e-01
-5.20457625e-01 5.08034050e-01 -1.14122726e-01 -2.18130559e-01
9.75994408e-01 2.53897130e-01 -7.63244703e-02 2.34682024e-01
-8.12468410e-01 3.72003764e-01 1.46284327e-01 3.15194190e-01
3.00722450e-01 -1.63725388e+00 -1.20947742e+00 3.22757810e-01
4.09243315e-01 -3.71862710e-01 5.73383272e-01 1.03941429e+00
-3.39692175e-01 2.08135337e-01 -3.55430365e-01 -6.16513073e-01
-1.20830286e+00 9.85308468e-01 3.18145484e-01 -5.23597479e-01
-9.16614115e-01 4.09962475e-01 4.77122128e-01 -4.38983709e-01
1.16498262e-01 -7.94750392e-01 -2.45457247e-01 2.86851786e-02
1.03225517e+00 2.95659423e-01 -5.26400171e-02 -4.75302994e-01
-7.14662001e-02 7.07647920e-01 -2.77760774e-02 6.04016066e-01
1.47874451e+00 -5.26058972e-01 3.99881899e-01 4.02503252e-01
1.27803671e+00 -5.44473767e-01 -1.25060165e+00 -6.10045254e-01
-3.99707139e-01 -1.59578428e-01 6.25257432e-01 -1.26673388e+00
-1.28488588e+00 7.09977627e-01 1.50716162e+00 -7.68570155e-02
1.37688422e+00 -1.77033573e-01 1.35134792e+00 -3.09917443e-02
1.89468428e-01 -8.54783356e-01 -1.47284165e-01 -1.91565171e-01
6.59680307e-01 -1.87105989e+00 1.56505778e-01 -3.86978894e-01
-2.84124583e-01 1.08662844e+00 2.26335600e-01 -2.90268034e-01
8.14221501e-01 1.90131903e-01 2.23523065e-01 -1.44591004e-01
-1.33377612e-01 -2.88495809e-01 4.55422550e-01 6.93928838e-01
7.03583539e-01 4.84752297e-01 -4.65144664e-01 7.18649626e-01
1.76179066e-01 4.08154041e-01 -2.03961328e-01 5.72018683e-01
-6.62160039e-01 -9.56871748e-01 -4.20245141e-01 7.60191619e-01
-8.06936026e-01 -6.71449527e-02 5.60856402e-01 4.41494226e-01
4.11576033e-01 9.71902966e-01 -8.00840110e-02 1.40906587e-01
3.75750780e-01 1.67735517e-02 8.85059953e-01 -5.79917252e-01
-8.00753117e-01 5.31095982e-01 5.45368996e-04 -3.62918407e-01
-7.80584931e-01 -7.44977176e-01 -1.12275612e+00 -3.83441776e-01
-1.64243266e-01 -3.76653403e-01 7.84029961e-01 8.26768994e-01
4.85610366e-02 5.93255579e-01 4.64942783e-01 -7.41235495e-01
-9.15953040e-01 -1.12817085e+00 -7.08418548e-01 6.45558357e-01
3.27396721e-01 -7.53675461e-01 3.01890820e-01 3.34430099e-01] | [11.560769081115723, 4.3974504470825195] |
42d8d5cd-e9bc-466c-979b-a737a56b7d5c | graph-embedded-multi-layer-kernel-extreme | 1904.06491 | null | http://arxiv.org/abs/1904.06491v1 | http://arxiv.org/pdf/1904.06491v1.pdf | Graph-Embedded Multi-layer Kernel Extreme Learning Machine for One-class Classification or (Graph-Embedded Multi-layer Kernel Ridge Regression for One-class Classification) | A brain can detect outlier just by using only normal samples. Similarly,
one-class classification (OCC) also uses only normal samples to train the model
and trained model can be used for outlier detection. In this paper, a
multi-layer architecture for OCC is proposed by stacking various Graph-Embedded
Kernel Ridge Regression (KRR) based Auto-Encoders in a hierarchical fashion.
These Auto-Encoders are formulated under two types of Graph-Embedding, namely,
local and global variance-based embedding. This Graph-Embedding explores the
relationship between samples and multi-layers of Auto-Encoder project the input
features into new feature space. The last layer of this proposed architecture
is Graph-Embedded regression-based one-class classifier. The Auto-Encoders use
an unsupervised approach of learning and the final layer uses semi-supervised
(trained by only positive samples and obtained closed-form solution) approach
to learning. The proposed method is experimentally evaluated on 21 publicly
available benchmark datasets. Experimental results verify the effectiveness of
the proposed one-class classifiers over 11 existing state-of-the-art
kernel-based one-class classifiers. Friedman test is also performed to verify
the statistical significance of the claim of the superiority of the proposed
one-class classifiers over the existing state-of-the-art methods. By using two
types of Graph-Embedding, 4 variants of Graph-Embedded multi-layer KRR-based
one-class classifier has been presented in this paper. All 4 variants performed
better than the existing one-class classifiers in terms of various discussed
criteria in this paper. Hence, it can be a viable alternative for OCC task. In
the future, various other types of Auto-Encoders can be explored within
proposed architecture. | ['Chandan Gautam', 'M. Tanveer', 'Aruna Tiwari'] | 2019-04-13 | null | null | null | null | ['one-class-classifier'] | ['methodology'] | [-2.08960980e-01 9.77933854e-02 -4.40741070e-02 -4.05499607e-01
-2.67676383e-01 1.21173643e-01 4.43004131e-01 4.12815392e-01
-3.71523887e-01 6.62473321e-01 -1.30181655e-01 -1.95269361e-01
-3.27123523e-01 -7.86227882e-01 -7.59543478e-01 -6.44571006e-01
-7.04324067e-01 4.44136858e-02 2.15932786e-01 2.55262218e-02
3.23270559e-01 6.30581260e-01 -1.67045331e+00 3.90361041e-01
8.17947030e-01 1.06497049e+00 -3.52570236e-01 7.91486979e-01
4.79097031e-02 8.13084543e-01 -6.30934954e-01 -1.72822759e-01
3.68010342e-01 -3.98219615e-01 -1.92334071e-01 -5.01744263e-02
5.07459700e-01 -1.35203674e-01 -5.27908206e-01 1.07829428e+00
3.27849478e-01 4.48271856e-02 1.11234784e+00 -1.66955793e+00
-1.11424291e+00 3.24513793e-01 -5.91745198e-01 4.27371681e-01
3.19445312e-01 -9.75413918e-02 9.35430944e-01 -1.03465915e+00
3.73851955e-02 1.14573157e+00 7.07305670e-01 1.45164624e-01
-1.13083076e+00 -7.22429931e-01 -3.88839096e-02 5.11124909e-01
-1.50919628e+00 1.55974803e-02 9.08123851e-01 -6.34496808e-01
1.37310362e+00 2.15089805e-02 5.82875848e-01 8.53197813e-01
8.49137783e-01 6.74406886e-01 1.51720154e+00 -4.10516888e-01
2.12132841e-01 2.84783900e-01 5.24057746e-01 1.08153665e+00
8.18373501e-01 2.09834114e-01 -6.46743238e-01 -2.84390539e-01
3.95667583e-01 1.08516410e-01 -2.84887403e-02 -3.01709920e-01
-9.06760752e-01 1.03961706e+00 6.26493096e-01 4.55993444e-01
-2.76240557e-01 5.77127971e-02 7.17392862e-01 5.77582002e-01
3.61321270e-01 2.23228410e-01 -3.10727865e-01 3.47450763e-01
-1.03080857e+00 -1.03057183e-01 5.81085205e-01 7.58379340e-01
6.82980061e-01 6.90965772e-01 -6.54236674e-02 5.99334776e-01
6.82885528e-01 7.98464492e-02 1.14285183e+00 1.50675654e-01
1.95399582e-01 1.01538670e+00 -6.88840508e-01 -1.02816713e+00
-6.90734684e-01 -9.11594629e-01 -9.80450690e-01 6.26330137e-01
5.38563244e-02 3.30860727e-02 -1.21560550e+00 1.05302584e+00
1.25819549e-01 6.49122596e-01 3.22704941e-01 3.80490094e-01
9.08983231e-01 5.21949530e-01 -2.75607347e-01 -8.78602341e-02
1.27984774e+00 -8.51130605e-01 -6.68992341e-01 1.30467162e-01
6.96257651e-01 -5.18460393e-01 7.98592746e-01 6.39919221e-01
-4.16914284e-01 -7.26938248e-01 -1.72375274e+00 3.02003413e-01
-9.75115895e-01 4.58442718e-01 4.59210366e-01 7.78961420e-01
-1.03261030e+00 5.65655529e-01 -6.92203939e-01 -1.76134437e-01
4.62371200e-01 6.57930851e-01 -7.18412220e-01 7.84175470e-02
-1.04515481e+00 8.82757187e-01 7.06849694e-01 1.70377597e-01
-8.30562711e-01 -2.14444086e-01 -9.64494705e-01 -2.09392291e-02
-8.11630487e-02 -3.03806961e-01 4.04734612e-01 -7.64925182e-01
-1.20204282e+00 5.81325173e-01 8.22330415e-02 -6.37434065e-01
4.52466428e-01 -1.23130210e-01 -7.38610983e-01 4.43520221e-05
4.42908183e-02 1.21266559e-01 1.24178708e+00 -9.81312573e-01
-4.30329859e-01 -5.85147262e-01 -4.65043783e-01 -3.73662375e-02
-3.94798994e-01 -2.09772959e-01 2.48786122e-01 -7.00326443e-01
2.48165488e-01 -7.80521274e-01 2.27203980e-01 -4.63133901e-01
-5.19299030e-01 -5.34128904e-01 1.15836418e+00 -5.93808889e-01
1.29203963e+00 -2.11966610e+00 -1.83649376e-01 4.33443606e-01
2.82307267e-01 1.42553106e-01 3.38053964e-02 3.95177394e-01
-5.92479765e-01 1.30990729e-01 -1.85015664e-01 -1.99722603e-01
-1.28189728e-01 5.76663241e-02 1.33748323e-01 9.11485016e-01
4.65824693e-01 6.49357617e-01 -4.90519762e-01 -3.44042361e-01
2.64091730e-01 4.84166205e-01 -2.25693867e-01 2.43491396e-01
2.75852323e-01 1.40309289e-01 -1.55414179e-01 8.42429161e-01
5.28320074e-01 2.85792023e-01 -3.84192109e-01 -2.84369230e-01
1.60317257e-01 -4.57081228e-01 -1.25439739e+00 1.16424906e+00
-2.01748624e-01 6.98750615e-01 -8.61004889e-01 -1.33303428e+00
1.29146636e+00 2.20724255e-01 1.46770298e-01 -3.86994898e-01
1.91846699e-01 4.12771642e-01 4.98578638e-01 -5.34895957e-01
1.26291409e-01 1.05340444e-01 8.34102407e-02 8.24545622e-02
5.68037033e-01 4.60561544e-01 1.42485261e-01 -7.28083849e-02
1.20612419e+00 7.07698762e-02 6.87412322e-01 -1.94247186e-01
9.76646602e-01 -5.00744045e-01 4.02071208e-01 6.96725845e-01
-3.99418920e-01 2.06208825e-01 6.60093188e-01 -5.13521433e-01
-9.29170191e-01 -1.04591215e+00 -4.03040886e-01 5.94277978e-01
-4.76027876e-01 -2.29297996e-01 -4.96074975e-01 -8.55202436e-01
3.28906983e-01 8.84106159e-01 -9.22718465e-01 -4.01774347e-01
-1.19344495e-01 -9.33291554e-01 8.89192522e-01 5.51991165e-01
6.05230987e-01 -1.02029598e+00 -4.23233330e-01 -3.78721133e-02
5.87111235e-01 -9.63819861e-01 1.75299585e-01 5.93123972e-01
-1.13629258e+00 -1.21002960e+00 -3.94737691e-01 -7.60215998e-01
7.76729882e-01 -1.95717126e-01 5.47460258e-01 1.41897157e-01
-4.61127520e-01 3.17160428e-01 -6.03441596e-01 -5.98970056e-01
-2.01281846e-01 -1.22962274e-01 3.91718209e-01 5.13249159e-01
6.59410834e-01 -5.72927713e-01 -4.23469365e-01 -1.07050076e-01
-8.16114068e-01 -6.51688874e-01 1.00986719e+00 1.04393315e+00
5.04076660e-01 3.74309152e-01 9.19121504e-01 -9.47721899e-01
9.26171124e-01 -7.13809848e-01 -6.74728036e-01 1.61313444e-01
-8.40060711e-01 3.42008144e-01 9.86145198e-01 -2.69414037e-01
-5.29089212e-01 -9.40328650e-03 1.88551545e-01 -6.45006359e-01
-2.08568841e-01 5.81765354e-01 8.09831545e-02 -2.82865137e-01
7.36877441e-01 4.28975940e-01 -1.02193430e-01 -1.72890663e-01
2.32367814e-01 9.64037836e-01 4.04536098e-01 -2.84174591e-01
8.73959780e-01 1.11325402e-02 3.30752462e-01 -1.16169655e+00
-3.70986581e-01 -6.18924022e-01 -8.34830642e-01 -1.87980801e-01
1.03287637e+00 -9.43689823e-01 -3.96917045e-01 5.77465355e-01
-8.23459506e-01 1.65954575e-01 8.67934059e-03 7.77235389e-01
-4.02371109e-01 3.34065497e-01 -3.84554952e-01 -1.27601707e+00
-4.27824527e-01 -1.15056896e+00 7.01576471e-01 1.79732814e-01
3.80174033e-02 -9.33049679e-01 2.79175602e-02 2.65181754e-02
1.42864034e-01 6.61394954e-01 9.71239984e-01 -1.50310159e+00
-3.19035292e-01 -7.72593975e-01 -1.49711624e-01 7.31586695e-01
7.55141452e-02 9.51045007e-02 -1.23497391e+00 -5.01672924e-01
-4.17574011e-02 -2.33864978e-01 8.56091201e-01 2.32969135e-01
1.04318404e+00 -2.84388453e-01 -5.81382886e-02 5.49513578e-01
1.97549260e+00 4.06498879e-01 6.48990095e-01 7.08513260e-01
7.45359600e-01 3.76277566e-02 2.03909904e-01 3.20948988e-01
6.89476281e-02 1.89843565e-01 4.80290413e-01 8.37540701e-02
1.37513369e-01 -1.05706081e-01 7.01315761e-01 1.14156389e+00
-1.33427605e-02 2.30911840e-02 -7.38831639e-01 3.34166616e-01
-1.71875978e+00 -8.74574184e-01 -3.87430489e-01 2.25522685e+00
2.93858945e-01 4.49211746e-01 1.58853531e-01 7.36029625e-01
4.62644368e-01 -5.97118028e-02 -5.77375233e-01 -9.30445373e-01
-6.82129562e-02 3.44032407e-01 6.00886881e-01 2.26910532e-01
-1.29118192e+00 6.08612001e-01 5.19909859e+00 4.92959827e-01
-1.09785008e+00 -4.65953983e-02 5.11238098e-01 2.76582867e-01
3.43856186e-01 -1.18207358e-01 -6.97139680e-01 4.85549480e-01
1.21163523e+00 -3.39029566e-03 2.79205292e-01 1.12545860e+00
1.21094438e-03 -1.70580626e-01 -1.34277344e+00 1.06205225e+00
5.54041266e-01 -7.98641503e-01 5.41654117e-02 1.35009244e-01
6.26670837e-01 8.09548795e-02 -6.27818406e-02 5.77530563e-01
-9.55256894e-02 -1.20012641e+00 3.69447857e-01 8.12886059e-01
6.13636851e-01 -9.97683644e-01 1.28833592e+00 1.52385592e-01
-1.26263344e+00 -3.39943945e-01 -6.01975560e-01 -1.14678912e-01
-5.85503757e-01 5.97671866e-01 -7.21851528e-01 9.76788402e-01
8.08552802e-01 6.98318720e-01 -1.23346913e+00 1.26587951e+00
-1.17478490e-01 6.47990108e-01 -1.29778802e-01 -2.17902407e-01
1.77383602e-01 -1.60785854e-01 4.75532174e-01 1.28422093e+00
3.02708477e-01 -2.79643804e-01 1.52538911e-01 6.25372350e-01
1.71307683e-01 4.36738551e-01 -1.14408910e+00 -3.66763696e-02
1.60529345e-01 1.25288332e+00 -7.66867101e-01 -5.14564872e-01
-8.07936192e-01 8.72722566e-01 5.75753212e-01 2.94921875e-01
-7.59623110e-01 -9.87553716e-01 1.75629348e-01 -7.91778602e-03
2.37070307e-01 -2.07284287e-01 -3.41628075e-01 -1.18935943e+00
-8.12092945e-02 -9.11049068e-01 5.55172205e-01 -5.09173095e-01
-1.51750791e+00 7.70849586e-01 8.52132961e-02 -1.28080690e+00
-8.88383687e-02 -1.12656820e+00 -7.69673288e-01 5.85830748e-01
-1.45420575e+00 -1.05947685e+00 -3.87565255e-01 7.77427256e-01
3.12567860e-01 -8.46010208e-01 8.17180216e-01 8.59081745e-02
-8.03845823e-01 9.74568844e-01 2.66071767e-01 3.56607556e-01
5.99528670e-01 -1.48434103e+00 -2.01019332e-01 1.17237079e+00
4.15436059e-01 7.09455013e-01 3.79796088e-01 -6.78093016e-01
-1.21265137e+00 -1.16232705e+00 4.57586348e-01 -6.61998168e-02
6.69955075e-01 -2.62366861e-01 -1.01255476e+00 8.58417392e-01
2.53356308e-01 4.02999967e-01 9.83379960e-01 -5.21791130e-02
-3.82965177e-01 -3.32812220e-01 -1.19784987e+00 2.80457288e-01
3.89909774e-01 -4.22505140e-01 -9.08792078e-01 2.35671163e-01
3.99315238e-01 -8.32986236e-02 -1.00923169e+00 3.97035927e-01
4.53042537e-01 -1.21430862e+00 5.76325178e-01 -6.77418053e-01
1.72933757e-01 -5.67310572e-01 -3.42791706e-01 -1.40738380e+00
-3.45495045e-01 -9.51241329e-02 -3.93127352e-01 9.40189958e-01
4.43195641e-01 -1.23175132e+00 5.49589396e-01 -7.41361231e-02
-3.73401165e-01 -1.20166564e+00 -7.90453494e-01 -1.07437229e+00
-3.20044123e-02 -4.45655555e-01 1.90108508e-01 9.32410061e-01
8.78982525e-03 3.10232490e-01 -1.83476090e-01 3.61811131e-01
8.65447760e-01 -6.19831383e-01 7.33861625e-01 -1.41630292e+00
-1.31790534e-01 -2.28230402e-01 -1.52535605e+00 1.09556548e-01
2.11026683e-01 -1.28149509e+00 -3.76986355e-01 -1.37640262e+00
-2.81601120e-03 3.62350866e-02 -9.03148413e-01 5.47076285e-01
-4.63117585e-02 1.21591777e-01 -1.24115504e-01 4.96563781e-03
-1.41314745e-01 5.42389452e-01 6.23259723e-01 2.05916236e-03
-1.18950419e-01 -1.57130182e-01 -2.26836562e-01 7.13457644e-01
1.07583225e+00 -5.24351478e-01 -6.44157588e-01 3.24480534e-01
-1.14421256e-01 -3.90515745e-01 4.29501772e-01 -1.51449001e+00
2.20927969e-01 3.37030888e-01 1.09294319e+00 -6.03163421e-01
-4.39856425e-02 -7.66012847e-01 -1.89995602e-01 7.19681203e-01
-1.06645606e-01 6.25695825e-01 2.09918454e-01 8.84323478e-01
-3.06396037e-01 -4.33262527e-01 8.64712179e-01 6.17696233e-02
-5.60918808e-01 2.46091783e-01 -2.85880297e-01 -5.11926189e-02
1.43078732e+00 -6.08080328e-01 -8.12529773e-02 9.68384650e-03
-8.49907696e-01 -1.52316287e-01 1.45732574e-02 2.96231776e-01
1.07134163e+00 -1.52550340e+00 -6.45646036e-01 5.94483674e-01
4.18104142e-01 -4.14194077e-01 -1.09747171e-01 8.92203689e-01
-6.48974717e-01 4.57567066e-01 -6.21119678e-01 -5.92935205e-01
-1.24146450e+00 5.86669981e-01 3.21369797e-01 -2.76013136e-01
-5.44959903e-01 5.63886583e-01 -2.35007390e-01 -2.83530474e-01
2.66364932e-01 -4.85360324e-01 -4.29036140e-01 4.17996012e-02
1.73300847e-01 5.10290861e-01 2.88946182e-01 -7.65469670e-01
-5.10830045e-01 5.02017558e-01 -2.06554323e-01 1.29414961e-01
1.29243135e+00 4.73240733e-01 -2.86155999e-01 9.10529375e-01
1.54890585e+00 -2.06041381e-01 -6.96576357e-01 7.81177133e-02
3.11769605e-01 -2.32612833e-01 3.31417061e-02 -5.13538599e-01
-9.48325157e-01 8.85099232e-01 9.51152027e-01 8.31953585e-02
1.08130538e+00 -6.79650486e-01 2.37131342e-01 4.95057732e-01
1.39311388e-01 -1.17645550e+00 3.80250156e-01 2.02758089e-01
8.65017354e-01 -1.32840860e+00 2.63289392e-01 -4.15968196e-03
-5.02027988e-01 1.64768493e+00 8.18571150e-01 -8.11737835e-01
9.66926992e-01 3.58483829e-02 -1.53091863e-01 -2.59228438e-01
-7.64829814e-01 -8.12244937e-02 7.21362412e-01 6.65613770e-01
7.15633690e-01 1.27873361e-01 -4.41391438e-01 5.98346829e-01
-3.32783610e-01 -2.12963864e-01 6.54558778e-01 7.37701833e-01
-4.39061612e-01 -7.30281770e-01 -2.92544276e-01 1.00905228e+00
-3.54377151e-01 9.22397971e-02 -2.10617334e-01 1.11094618e+00
3.25843483e-01 8.94175410e-01 -2.13670418e-01 -7.24905014e-01
2.79120684e-01 4.81586933e-01 3.16607982e-01 -4.56463039e-01
-4.63746250e-01 -2.74327576e-01 -9.79536623e-02 -5.28810203e-01
9.39799566e-03 -5.17191589e-01 -1.12569332e+00 2.30224937e-01
-5.43721020e-01 1.63880140e-02 6.74063444e-01 7.72677004e-01
8.39747041e-02 9.61287677e-01 5.30073166e-01 -4.42681074e-01
-6.45324647e-01 -1.12091362e+00 -7.32518137e-01 1.29380181e-01
3.72709692e-01 -9.67530429e-01 -7.56570220e-01 -2.36984208e-01] | [8.210896492004395, 3.732754707336426] |
976e6e4b-865f-44e1-8828-1996c73e96b1 | enhancement-encoding-a-new-imbalanced | 2208.11056 | null | https://arxiv.org/abs/2208.11056v2 | https://arxiv.org/pdf/2208.11056v2.pdf | Enhancement Encoding: A Novel Imbalanced Classification Approach via Encoding the Training Labels | Class imbalance, which is also called long-tailed distribution, is a common problem in classification tasks based on machine learning. If it happens, the minority data will be overwhelmed by the majority, which presents quite a challenge for data science. To address the class imbalance problem, researchers have proposed lots of methods: some people make the data set balanced (SMOTE), some others refine the loss function (Focal Loss), and even someone has noticed the value of labels influences class-imbalanced learning (Yang and Xu. Rethinking the value of labels for improving class-imbalanced learning. In NeurIPS 2020), but no one changes the way to encode the labels of data yet. Nowadays, the most prevailing technique to encode labels is the one-hot encoding due to its nice performance in the general situation. However, it is not a good choice for imbalanced data, because the classifier will treat majority and minority samples equally. In this paper, we innovatively propose the enhancement encoding technique, which is specially designed for the imbalanced classification. The enhancement encoding combines re-weighting and cost-sensitiveness, which can reflect the difference between hard and easy (or minority and majority) classes. To reduce the number of validation samples and the computation cost, we also replace the confusion matrix with a novel soft-confusion matrix which works better with a small validation set. In the experiments, we evaluate the enhancement encoding with three different types of loss. And the results show that enhancement encoding is very effective to improve the performance of the network trained with imbalanced data. Particularly, the performance on minority classes is much better. | ['Jia-Chen Zhao'] | 2022-08-23 | null | null | null | null | ['imbalanced-classification'] | ['miscellaneous'] | [-2.99002789e-02 -3.55389833e-01 -2.25675553e-01 -6.74353123e-01
-1.01822741e-01 -8.82022679e-02 6.50083944e-02 2.81616479e-01
-5.05412221e-01 8.69154513e-01 -3.95277292e-02 3.98540491e-04
-2.37809986e-01 -1.07492197e+00 -2.37174243e-01 -9.27562714e-01
3.85750085e-01 3.10479701e-01 1.86318353e-01 -4.52510953e-01
3.73180449e-01 1.59667119e-01 -1.93647921e+00 6.48881972e-01
1.13555932e+00 1.13568377e+00 -1.55232057e-01 3.02710244e-03
-5.37281394e-01 8.29128563e-01 -9.31447148e-01 -5.02723277e-01
2.28230685e-01 -5.77241480e-01 -4.89937693e-01 -1.62544012e-01
-1.50347114e-01 3.07519045e-02 8.87969881e-02 1.25213337e+00
8.52491736e-01 -2.83943266e-01 6.51914895e-01 -1.67413473e+00
-1.14616625e-01 3.87301266e-01 -8.26384366e-01 2.91255176e-01
-1.39363274e-01 -1.00252129e-01 4.48947042e-01 -4.55046505e-01
2.06711069e-01 1.26573467e+00 8.52495134e-01 4.03378844e-01
-6.13273501e-01 -1.14852834e+00 7.86403567e-02 7.20242262e-01
-1.27518415e+00 -9.94754657e-02 8.29423368e-01 -4.27580684e-01
2.04700634e-01 4.88200784e-01 6.31935298e-01 6.85961604e-01
3.13188553e-01 6.06332123e-01 1.18631041e+00 -3.84731263e-01
1.21179208e-01 3.36160660e-01 3.14960420e-01 1.57398656e-01
5.56075752e-01 -1.61420882e-01 -1.76485091e-01 2.32146472e-01
-1.84536517e-01 3.07319671e-01 -3.65353703e-01 -2.41780549e-01
-8.72473359e-01 6.96972013e-01 4.56942618e-01 4.35143381e-01
3.65300030e-02 -5.32781363e-01 8.41678321e-01 4.99821365e-01
6.02146685e-01 2.20552459e-01 -5.02796590e-01 -2.33697399e-01
-6.04586303e-01 2.69930333e-01 4.87829417e-01 2.54772097e-01
6.95807099e-01 -2.83264518e-01 -1.95820704e-01 1.07949984e+00
1.73407972e-01 3.05046856e-01 1.09117806e+00 -2.43213013e-01
5.66112697e-01 1.24572384e+00 -2.02302545e-01 -1.48903108e+00
-6.08950555e-01 -8.72869432e-01 -1.30036342e+00 3.51891994e-01
3.17944020e-01 5.05127646e-02 -7.76480496e-01 1.60725355e+00
3.80841106e-01 -2.60117024e-01 -8.71234313e-02 9.18810189e-01
8.81061554e-01 7.60458887e-01 -2.26730667e-02 -5.35771966e-01
1.11253834e+00 -8.26453567e-01 -1.05967450e+00 -3.89142931e-02
8.40736687e-01 -7.32544482e-01 9.38166022e-01 7.61034131e-01
-6.55686319e-01 -7.54801095e-01 -1.15336406e+00 1.84319854e-01
-5.75067043e-01 5.84506169e-02 5.56278765e-01 9.12889540e-01
-3.20086658e-01 5.35678208e-01 -2.09737644e-01 2.56663244e-02
5.66528618e-01 3.19319010e-01 -4.92490828e-01 -3.41416597e-01
-1.57283747e+00 1.04848969e+00 6.74064875e-01 1.41064703e-01
-1.60467759e-01 -5.65038323e-01 -5.56051195e-01 8.36459771e-02
-1.78609602e-02 -1.41440943e-01 7.63840258e-01 -1.38333714e+00
-1.08683467e+00 8.36257756e-01 2.52956659e-01 -2.24795099e-02
6.70852602e-01 1.76165730e-01 -6.54724836e-01 -4.98371094e-01
-6.74911290e-02 2.70735770e-01 4.90605265e-01 -8.75009239e-01
-7.28866696e-01 -6.78048313e-01 -2.57774830e-01 4.04496521e-01
-6.67556942e-01 -1.57849669e-01 1.01286620e-01 -4.19256598e-01
3.13267827e-01 -4.84396458e-01 4.45216671e-02 -1.28926843e-01
-2.05003545e-01 -3.03957939e-01 8.79472554e-01 -5.78751147e-01
1.57210469e+00 -2.30947876e+00 -1.79766297e-01 1.86116472e-01
6.13374524e-02 6.69565320e-01 1.53038025e-01 1.88826174e-02
-4.89113271e-01 1.96418226e-01 -4.29614961e-01 8.36481154e-02
-1.62412196e-01 1.99668512e-01 3.46819721e-02 3.57490063e-01
1.16006196e-01 3.45322639e-01 -6.31841540e-01 -4.10003483e-01
2.32249051e-02 3.21387887e-01 -4.84564722e-01 3.07170421e-01
2.12717339e-01 2.46451899e-01 -1.50224820e-01 4.57493693e-01
1.30773199e+00 1.01793669e-01 -7.34414682e-02 -4.06306595e-01
-6.59089386e-02 1.61820725e-02 -1.57464862e+00 1.09801376e+00
-2.08529502e-01 2.79063910e-01 -2.78896987e-01 -1.38528049e+00
1.38524592e+00 1.00185201e-01 1.79007471e-01 -1.20445895e+00
3.12392771e-01 5.88121116e-01 3.13500851e-01 -8.37113082e-01
4.71592918e-02 -5.21513760e-01 8.78559723e-02 1.66153386e-01
-4.81758803e-01 -8.84358212e-03 2.49149024e-01 -3.80259275e-01
7.46985853e-01 -1.18982069e-01 2.66958803e-01 -3.33433673e-02
7.11222529e-01 -1.08485058e-01 1.06097972e+00 1.02433581e-02
-2.76155412e-01 4.77266222e-01 7.21607089e-01 -6.44547343e-01
-8.33635926e-01 -4.42220509e-01 -6.00638211e-01 7.40934491e-01
3.60961169e-01 -1.18707784e-01 -6.46234810e-01 -7.38967359e-01
1.15102984e-01 3.15960169e-01 -4.72379655e-01 -8.04624140e-01
-3.12010914e-01 -1.54435611e+00 4.62556750e-01 1.23550788e-01
1.05538404e+00 -9.94592071e-01 -2.28614271e-01 -1.55511033e-02
-2.91660994e-01 -2.67280430e-01 1.69215769e-01 2.22732201e-01
-9.14807498e-01 -1.15137470e+00 -7.51423419e-01 -7.11091936e-01
5.77132702e-01 1.24600381e-01 9.68251467e-01 3.77668172e-01
-1.67610839e-01 -7.88150191e-01 -5.62225163e-01 -6.33077085e-01
-2.32520312e-01 2.32231006e-01 1.56783648e-02 1.65536806e-01
5.91694713e-01 -3.49445432e-01 -4.03684944e-01 6.48310542e-01
-1.05140018e+00 -1.09783024e-01 5.24024427e-01 8.22391570e-01
3.05948198e-01 6.81914449e-01 9.32515264e-01 -1.00778806e+00
6.06992960e-01 -6.02190375e-01 -1.53627127e-01 6.08467460e-02
-8.97654772e-01 -1.75130099e-01 6.57072783e-01 -2.03622669e-01
-9.21550751e-01 -5.90590060e-01 -5.47835171e-01 1.43482551e-01
1.91629995e-02 3.43612909e-01 -6.41615987e-01 -6.87713251e-02
5.20351708e-01 -2.69558653e-02 2.56782651e-01 -5.55052698e-01
-3.18383545e-01 1.24251759e+00 -1.42058924e-01 -1.98002249e-01
3.68186027e-01 9.26449522e-02 -4.69364785e-02 -1.96513459e-01
-8.31603885e-01 -2.92867452e-01 -1.27795577e-01 -1.06433034e-01
4.88493592e-01 -7.18335211e-01 -6.93871856e-01 1.05242920e+00
-1.04832542e+00 8.50183666e-02 -2.38516137e-01 6.11847162e-01
-2.42782589e-02 2.25353837e-01 -2.88164824e-01 -6.20056033e-01
-1.05760656e-01 -1.11649895e+00 5.76349497e-01 5.53990602e-01
8.57843310e-02 -5.70941150e-01 1.91028163e-01 3.40868443e-01
6.63752615e-01 2.54237354e-01 1.18927896e+00 -7.28121459e-01
6.64728344e-04 -2.84068525e-01 -4.09487873e-01 8.21620107e-01
1.91564381e-01 3.62135582e-02 -9.94202375e-01 -1.53071612e-01
2.51125187e-01 -2.67152846e-01 8.47480357e-01 8.01998973e-02
1.60041022e+00 -2.30049267e-01 -2.21148536e-01 5.86618304e-01
1.33370459e+00 6.34516358e-01 1.15111089e+00 7.82146633e-01
4.64196116e-01 8.75286698e-01 7.89111912e-01 2.90975809e-01
3.76334667e-01 6.22194767e-01 7.11105227e-01 -2.32916161e-01
-2.26919390e-02 1.58911347e-02 -2.33094208e-02 1.22057009e+00
3.88813429e-02 -2.39809066e-01 -7.49150157e-01 2.49471173e-01
-1.55312538e+00 -1.00970232e+00 -3.96318853e-01 2.33921242e+00
1.21653688e+00 3.46896112e-01 -8.14565644e-02 1.09575605e+00
9.78329957e-01 5.43056726e-02 -4.33933616e-01 -4.86880690e-01
-2.54460692e-01 6.33956268e-02 1.01296954e-01 2.13713795e-01
-9.95687008e-01 1.26560718e-01 5.26269293e+00 1.22987020e+00
-1.42498970e+00 1.52548984e-01 1.19708586e+00 1.57122388e-02
-3.79612565e-01 -3.43722314e-01 -8.44531596e-01 1.23221588e+00
4.99846399e-01 2.87984796e-02 -5.22464737e-02 6.82152927e-01
1.47695036e-03 -7.12459311e-02 -5.74511170e-01 1.25995064e+00
2.26885691e-01 -7.50673532e-01 -1.55040830e-01 -1.59155279e-01
5.64111292e-01 -4.99186099e-01 -1.46423280e-02 6.50786221e-01
-4.71410573e-01 -1.05062795e+00 3.57738346e-01 4.97870713e-01
5.81196129e-01 -1.04771912e+00 1.54874218e+00 5.79978704e-01
-6.74662590e-01 -2.39366263e-01 -6.34387314e-01 -5.29009461e-01
-2.83084303e-01 1.42318416e+00 -2.31393635e-01 6.36678994e-01
8.77268672e-01 7.78899312e-01 -7.69048810e-01 1.43380427e+00
-3.40928696e-02 3.65072340e-01 -2.18375817e-01 -5.62479496e-02
-2.26471275e-01 -1.84322864e-01 1.60965428e-01 8.81976783e-01
3.78851026e-01 -8.35386664e-02 1.89032070e-02 1.87874943e-01
-1.47996247e-01 5.57347894e-01 -4.71946776e-01 4.89154875e-01
1.89604208e-01 1.14758527e+00 -5.38858950e-01 -2.66658902e-01
9.12065133e-02 5.02741575e-01 1.10040449e-01 -9.65741575e-02
-7.09297717e-01 -8.77053201e-01 2.65495867e-01 1.13968864e-01
-2.94119269e-01 4.98870492e-01 -5.77768683e-01 -9.82859850e-01
3.49299759e-01 -1.13050985e+00 4.68402594e-01 -6.55633748e-01
-1.33246922e+00 5.57396889e-01 -2.85138518e-01 -1.49722683e+00
3.30837846e-01 -5.37114739e-01 -4.48401332e-01 8.08022261e-01
-1.75767207e+00 -4.37437147e-01 -7.40299821e-01 1.88603908e-01
1.99682504e-01 -8.78256038e-02 5.00711799e-01 1.18187177e+00
-8.08661997e-01 6.44791961e-01 2.10974902e-01 -3.47331055e-02
1.07882619e+00 -8.59101832e-01 -4.46491867e-01 4.45191503e-01
-4.95260656e-01 1.84088141e-01 4.55199003e-01 -3.55621308e-01
-4.85079318e-01 -9.02438641e-01 1.08913267e+00 -3.41625288e-02
-2.33858764e-01 -1.40565529e-03 -1.13856792e+00 2.24345922e-03
-8.34830180e-02 1.59142241e-01 8.02503347e-01 8.68036002e-02
-2.04260394e-01 -8.17179918e-01 -1.60526621e+00 9.11624134e-02
6.98591828e-01 8.17669407e-02 -5.88013649e-01 3.00443918e-01
5.45155764e-01 -3.26516360e-01 -5.64024568e-01 9.44809616e-01
5.88151455e-01 -1.43906593e+00 5.99972546e-01 -5.80798686e-01
6.61318064e-01 -7.09121168e-01 -7.28009790e-02 -1.57291257e+00
-2.28409693e-01 3.31831515e-01 3.61775905e-01 1.41433072e+00
1.05283491e-01 -8.76823545e-01 7.68365622e-01 1.05364606e-01
-5.81507124e-02 -1.10998595e+00 -8.62789989e-01 -5.88285983e-01
3.76928300e-02 -9.75046679e-02 1.01566172e+00 1.22258711e+00
-9.01668295e-02 1.97159052e-01 -1.62381321e-01 -3.51827860e-01
2.41329446e-01 3.01651787e-02 5.95797539e-01 -1.54207039e+00
-2.34123282e-02 -5.05281508e-01 -8.67324710e-01 -4.07401443e-01
-1.94196016e-01 -8.73920262e-01 -1.69081599e-01 -1.41240382e+00
2.79048473e-01 -9.59872365e-01 -5.02485812e-01 4.66691345e-01
-4.27591950e-01 5.50367296e-01 -3.12439539e-02 1.95012048e-01
-4.44329083e-01 7.45771766e-01 1.41113746e+00 -3.52565169e-01
-3.65496650e-02 2.21468255e-01 -7.80211449e-01 6.45482540e-01
8.06833982e-01 -6.83154881e-01 -4.20104146e-01 -2.62886047e-01
6.27861619e-01 -4.05750066e-01 -4.22619767e-02 -1.31091869e+00
4.36672680e-02 -6.25015274e-02 5.83977282e-01 -4.55608010e-01
-1.32264689e-01 -8.88397396e-01 2.00285822e-01 8.32430482e-01
-1.95069745e-01 1.09330714e-01 6.76696524e-02 1.55628234e-01
-6.11724377e-01 -4.83746469e-01 1.05784595e+00 9.56636202e-03
-5.53978562e-01 2.59878725e-01 8.56268555e-02 2.80620456e-01
1.18229842e+00 -3.11792880e-01 -5.08543372e-01 -5.34794778e-02
-2.84482002e-01 2.63669580e-01 3.80076706e-01 3.58990133e-01
3.63149375e-01 -1.62909579e+00 -7.38662958e-01 3.92188013e-01
1.96508378e-01 1.35871500e-01 5.91119170e-01 8.98202598e-01
-6.33068323e-01 -9.16551426e-02 -7.36381292e-01 -4.81452078e-01
-1.19269562e+00 3.97340268e-01 4.83481199e-01 -5.02594709e-01
-1.57882236e-02 8.37460220e-01 -7.00697303e-02 -8.60550225e-01
2.12420195e-01 -1.29711643e-01 -8.61484408e-01 5.73994935e-01
8.18232894e-01 6.86779499e-01 6.18530273e-01 -3.88710529e-01
-4.25512403e-01 6.56223118e-01 -9.47130397e-02 5.31823337e-01
1.14852762e+00 2.01688871e-01 -6.32400155e-01 5.57363808e-01
1.31231213e+00 -7.35158026e-02 -6.21217132e-01 1.77837163e-01
-3.04009676e-01 -6.51100039e-01 -1.29348010e-01 -9.85060990e-01
-1.30635905e+00 1.22384536e+00 1.09775805e+00 4.95310694e-01
1.31560636e+00 -6.93719745e-01 7.84243882e-01 1.76626127e-02
4.16676015e-01 -1.03431225e+00 -8.06136131e-02 4.31473255e-01
6.02795124e-01 -1.27369559e+00 6.51399791e-02 -4.83831793e-01
-4.66111898e-01 1.09374893e+00 7.75915623e-01 1.16371997e-01
8.12666833e-01 1.37751728e-01 1.52162269e-01 4.82895896e-02
-4.07870859e-01 1.86793610e-01 -1.22187883e-01 4.81643677e-01
4.17986602e-01 3.57893668e-02 -1.07323146e+00 8.74370217e-01
-1.86466038e-01 2.08577976e-01 4.12405759e-01 6.30758762e-01
-5.69445908e-01 -1.46180165e+00 -5.20098627e-01 7.76595891e-01
-4.11471754e-01 2.34615922e-01 1.20223850e-01 4.84754056e-01
1.12749815e+00 8.87931764e-01 2.51718313e-01 -6.92979932e-01
5.79806626e-01 9.89120454e-02 9.36009586e-02 -4.53443229e-01
-6.30819738e-01 -7.52978027e-01 -2.17468470e-01 -2.94451386e-01
-5.80021799e-01 -1.39219850e-01 -1.12080336e+00 -4.90857035e-01
-4.57517326e-01 4.78901207e-01 7.24476039e-01 8.56503785e-01
2.97786832e-01 8.09567332e-01 1.01578736e+00 -2.75048167e-01
-6.23984337e-01 -9.86911356e-01 -6.43934608e-01 5.97688437e-01
1.54498577e-01 -9.78518188e-01 -7.27060735e-01 -4.57052618e-01] | [9.048050880432129, 3.976097345352173] |
2f08f94e-8f9e-44e1-ad2f-13a6cff15490 | class-wise-and-reduced-calibration-methods | 2210.03702 | null | https://arxiv.org/abs/2210.03702v1 | https://arxiv.org/pdf/2210.03702v1.pdf | Class-wise and reduced calibration methods | For many applications of probabilistic classifiers it is important that the predicted confidence vectors reflect true probabilities (one says that the classifier is calibrated). It has been shown that common models fail to satisfy this property, making reliable methods for measuring and improving calibration important tools. Unfortunately, obtaining these is far from trivial for problems with many classes. We propose two techniques that can be used in tandem. First, a reduced calibration method transforms the original problem into a simpler one. We prove for several notions of calibration that solving the reduced problem minimizes the corresponding notion of miscalibration in the full problem, allowing the use of non-parametric recalibration methods that fail in higher dimensions. Second, we propose class-wise calibration methods, based on intuition building on a phenomenon called neural collapse and the observation that most of the accurate classifiers found in practice can be thought of as a union of K different functions which can be recalibrated separately, one for each class. These typically out-perform their non class-wise counterparts, especially for classifiers trained on imbalanced data sets. Applying the two methods together results in class-wise reduced calibration algorithms, which are powerful tools for reducing the prediction and per-class calibration errors. We demonstrate our methods on real and synthetic datasets and release all code as open source at https://github.com/appliedAI-Initiative | ['Miguel de Benito Delgado', 'Anes Benmerzoug', 'Michael Panchenko'] | 2022-10-07 | null | null | null | null | ['classifier-calibration', 'classifier-calibration'] | ['computer-vision', 'miscellaneous'] | [ 1.87425479e-01 2.17333883e-01 -2.24581689e-01 -6.19236231e-01
-9.54690158e-01 -8.19563687e-01 3.45680058e-01 4.07038361e-01
-2.48009399e-01 7.99647331e-01 -2.12332696e-01 -4.08988237e-01
-3.28822076e-01 -7.14145243e-01 -8.28962266e-01 -1.01342905e+00
2.30557650e-01 8.31451952e-01 2.77867794e-01 -1.46263475e-02
2.90172756e-01 4.91979182e-01 -1.75582159e+00 1.87957883e-01
9.58370686e-01 8.52486193e-01 -4.61144328e-01 8.06247354e-01
8.23969617e-02 4.85723406e-01 -5.70369661e-01 -7.94677913e-01
2.76576817e-01 -2.86049932e-01 -8.03059161e-01 -2.42032200e-01
5.66236973e-01 1.75023615e-01 2.20734388e-01 1.19417942e+00
2.81750083e-01 -1.03410669e-01 8.31443131e-01 -1.64709318e+00
-3.19207609e-01 7.01951087e-01 -6.04357541e-01 -6.76272139e-02
1.20584540e-01 -3.87555510e-01 9.86116827e-01 -6.11049652e-01
1.80374738e-02 1.11447787e+00 1.18731868e+00 4.04392540e-01
-1.55888891e+00 -8.44261408e-01 -2.76815802e-01 5.19730933e-02
-1.33092999e+00 -2.44822517e-01 5.06791949e-01 -6.24262869e-01
5.12298644e-01 4.28060234e-01 3.14588577e-01 8.67229998e-01
1.02853879e-01 3.67691725e-01 1.15184629e+00 -6.03718758e-01
3.16935182e-01 5.89469731e-01 6.81592584e-01 2.56919771e-01
9.13142204e-01 1.80598795e-01 -3.00063878e-01 -4.54853356e-01
3.14904451e-01 -6.55631572e-02 -4.24193740e-01 -7.75861084e-01
-9.50855851e-01 1.00221646e+00 1.16528578e-01 3.20719182e-01
2.83076707e-02 -5.65229058e-02 2.92473137e-01 2.41911098e-01
3.99105012e-01 4.89999413e-01 -6.52935982e-01 -5.31482138e-02
-8.89770925e-01 3.37792248e-01 1.09457624e+00 7.67280459e-01
7.72550285e-01 -5.31193733e-01 1.87793016e-01 8.09485674e-01
-2.57508736e-02 3.67797732e-01 5.46716213e-01 -8.32859814e-01
1.45515740e-01 6.06514335e-01 2.91027099e-01 -9.30027187e-01
-4.42506939e-01 -5.12061059e-01 -1.00573766e+00 3.37093547e-02
8.45124543e-01 2.25785136e-01 -6.66050434e-01 1.94563055e+00
3.73781949e-01 1.86032727e-01 -1.62955541e-02 4.83437032e-01
4.96357344e-02 3.20211321e-01 4.82634865e-02 -2.53678262e-01
1.09836209e+00 -5.16536891e-01 -4.12694126e-01 -4.37446609e-02
8.34390640e-01 -6.81118071e-01 1.14424312e+00 7.86614776e-01
-6.57064080e-01 -5.16005337e-01 -1.32796848e+00 1.77741632e-01
-4.29337323e-01 -1.15003608e-01 6.26810074e-01 1.24707675e+00
-8.49200845e-01 9.39426720e-01 -7.25452542e-01 -2.17484400e-01
4.46703702e-01 4.69747573e-01 -5.23396373e-01 -1.40909463e-01
-9.39276874e-01 1.02276134e+00 4.49758172e-01 -4.97686900e-02
-2.38957554e-01 -7.52276182e-01 -5.59023976e-01 -2.37789564e-02
1.82361543e-01 -5.25112152e-01 1.37049878e+00 -1.08362901e+00
-1.13579547e+00 9.32278752e-01 4.15980816e-04 -2.55524725e-01
6.85031235e-01 -2.14659408e-01 -1.53540626e-01 -3.18749607e-01
-8.13611150e-02 1.64686993e-01 6.12809479e-01 -1.28809500e+00
-5.91996133e-01 -5.94178140e-01 -7.16545135e-02 -2.69302547e-01
-3.45900774e-01 -2.95094132e-01 -1.43150330e-01 -4.99996185e-01
3.00844431e-01 -9.63309705e-01 -1.59104273e-01 -1.97007388e-01
-4.99278843e-01 -1.93638071e-01 3.96480501e-01 -4.40536112e-01
9.91784036e-01 -1.90454328e+00 5.61339483e-02 4.60697591e-01
2.17906237e-01 1.72209248e-01 1.07548177e-01 1.05143420e-01
-6.04819953e-01 3.19414109e-01 -7.02602684e-01 -2.90748328e-01
-2.84864176e-02 3.48386794e-01 -4.23423439e-01 7.50500441e-01
1.79220945e-01 5.06384373e-01 -7.64767647e-01 -3.07413459e-01
1.10022433e-01 5.15233219e-01 -5.01285613e-01 1.12599768e-01
1.97684899e-01 2.34111562e-01 -2.45208833e-02 3.26131970e-01
1.00050688e+00 -1.69318214e-01 2.72938401e-01 -1.49746448e-01
2.90990472e-01 1.49918646e-01 -1.51756334e+00 1.02065837e+00
-2.07778305e-01 3.90472919e-01 -3.23679000e-01 -1.48168564e+00
9.13102508e-01 7.78309032e-02 3.74922991e-01 -2.06794456e-01
2.55555242e-01 2.39083886e-01 -1.74569145e-01 -7.07298666e-02
2.36896619e-01 -5.20154655e-01 -2.70212531e-01 4.37515050e-01
1.39874041e-01 -2.79611170e-01 1.32553533e-01 3.98199446e-02
8.64171982e-01 -1.38982445e-01 5.20323515e-01 -4.50835049e-01
3.76179427e-01 -1.04068555e-01 6.23214960e-01 1.01646912e+00
-1.47753596e-01 7.15224624e-01 8.37666690e-01 -2.66699016e-01
-1.29273593e+00 -1.09582222e+00 -6.18190527e-01 8.40963602e-01
-6.69837072e-02 -2.08416730e-01 -9.57791984e-01 -5.71545899e-01
2.28137091e-01 9.09059584e-01 -8.02133203e-01 -4.81970578e-01
-2.76376367e-01 -1.22867906e+00 6.60425603e-01 5.75894415e-01
-3.71679999e-02 -3.06094587e-01 -4.44972277e-01 -1.63054299e-02
-2.07029283e-02 -8.11593413e-01 3.95979173e-02 5.42488992e-01
-1.03458035e+00 -1.46892929e+00 -4.82354194e-01 -3.26760411e-01
5.28828025e-01 8.96836519e-02 1.38621449e+00 2.25023851e-01
-6.99780956e-02 2.12037414e-01 -2.38429829e-01 -5.82424223e-01
-6.74677312e-01 8.39511827e-02 3.36599827e-01 -1.87831521e-01
6.43664360e-01 -6.70071125e-01 -9.97731835e-02 6.06999040e-01
-9.36683118e-01 -1.38772458e-01 5.21044433e-01 1.05956101e+00
6.96971953e-01 -1.40814576e-02 5.74559987e-01 -1.28917837e+00
4.64379579e-01 -5.14558673e-01 -8.50858152e-01 5.34097552e-01
-9.07054186e-01 3.96600515e-01 4.58710641e-01 -5.16828775e-01
-5.39224565e-01 1.93231300e-01 -6.64578900e-02 -3.02558064e-01
-1.70898348e-01 2.24893764e-01 -3.27594757e-01 -2.19033025e-02
1.00044405e+00 -1.57020673e-01 -4.78966534e-02 -3.64172399e-01
2.92662799e-01 7.76134610e-01 7.16189086e-01 -7.62227893e-01
8.33687961e-01 2.59318590e-01 8.01802352e-02 -4.48665261e-01
-1.25968158e+00 -4.57929045e-01 -8.48473907e-01 6.14686962e-03
4.18003321e-01 -5.21853685e-01 -7.90732682e-01 4.32570845e-01
-8.55483294e-01 -1.74394310e-01 -4.89912987e-01 3.64813328e-01
-5.50743461e-01 3.99363637e-01 -8.96735936e-02 -8.74280751e-01
-4.50046873e-03 -1.07280457e+00 7.95436561e-01 2.29439065e-01
-1.95997149e-01 -9.11460996e-01 4.77450281e-01 1.43402800e-01
2.64818698e-01 2.24853799e-01 9.56048965e-01 -1.09220779e+00
8.51118565e-02 -6.64587021e-01 -1.56649187e-01 7.11847723e-01
-2.55186111e-02 2.54954010e-01 -1.30472589e+00 -2.15331420e-01
2.90917873e-01 -1.29366204e-01 8.97659063e-01 3.20137918e-01
1.38308251e+00 -1.89182028e-01 -3.15298826e-01 5.17986000e-01
1.42276776e+00 -2.00528949e-01 6.11955583e-01 1.66841596e-01
6.13850117e-01 6.36376441e-01 4.37392294e-01 2.14970231e-01
2.10160270e-01 7.23406434e-01 2.70155191e-01 2.03277498e-01
3.42742711e-01 -6.13396689e-02 9.52465385e-02 8.37958097e-01
4.19638939e-02 1.03793889e-02 -1.10177922e+00 2.67113417e-01
-1.84486830e+00 -8.03049386e-01 -3.87468904e-01 2.91951489e+00
1.11194551e+00 2.50387818e-01 1.51215434e-01 6.74567938e-01
9.39823747e-01 -5.26499093e-01 -5.67435563e-01 -3.07037294e-01
-2.46957973e-01 4.19732183e-01 6.80969596e-01 6.38322949e-01
-1.09119666e+00 3.84226263e-01 6.69061470e+00 7.08548486e-01
-7.81024754e-01 2.30623901e-01 8.89820099e-01 2.29333997e-01
-1.57605350e-01 2.57225037e-01 -7.80499101e-01 3.29809695e-01
1.30921078e+00 -3.04910749e-01 1.29335269e-01 1.10909283e+00
-4.27477002e-01 -3.94925833e-01 -1.44399810e+00 1.05520582e+00
-3.59839313e-02 -1.01612115e+00 -1.62472442e-01 -1.42879233e-01
6.66307986e-01 -2.79712111e-01 -8.46527219e-02 2.15623707e-01
4.72985417e-01 -1.06352544e+00 6.00702167e-01 5.43997526e-01
6.71696186e-01 -9.06971335e-01 1.07588112e+00 3.41505200e-01
-6.43060327e-01 -1.11245029e-01 -4.50456411e-01 -3.69208865e-02
-2.89508700e-01 1.23981881e+00 -7.23683357e-01 5.28186083e-01
6.00593805e-01 4.54936206e-01 -7.54143357e-01 1.15509701e+00
-8.39165822e-02 7.21054435e-01 -6.15798056e-01 2.41027832e-01
-4.92711842e-01 -1.59594998e-01 2.17028007e-01 1.12091887e+00
2.50763148e-01 -9.87907797e-02 -3.01334083e-01 6.36155963e-01
4.25833911e-02 -1.10671490e-01 -4.20389801e-01 2.29897961e-01
4.95269001e-01 1.01738131e+00 -7.76367009e-01 -4.22223061e-01
-1.15369961e-01 5.64033210e-01 3.23373020e-01 9.35272221e-03
-9.42613900e-01 -3.65404814e-01 6.51527762e-01 -1.35368019e-01
7.11202174e-02 6.46866783e-02 -6.41332865e-01 -1.26447594e+00
4.72073592e-02 -8.88254225e-01 6.55793548e-01 -5.72156131e-01
-1.62082493e+00 4.47912991e-01 1.36026025e-01 -1.08168101e+00
-2.24277273e-01 -7.86571681e-01 -3.06164294e-01 6.85266614e-01
-1.29072309e+00 -6.50681853e-01 -4.27922308e-01 4.22968477e-01
-6.37632981e-02 3.39360237e-01 1.07824564e+00 3.20522875e-01
-6.71253860e-01 9.48868990e-01 2.45893732e-01 -1.34341747e-01
1.08814692e+00 -1.39314508e+00 7.75609165e-03 8.39350581e-01
1.90401450e-01 4.62687075e-01 9.69652951e-01 -3.11690837e-01
-9.97181654e-01 -8.74501765e-01 8.39037597e-01 -8.24896753e-01
6.20815396e-01 -3.42081428e-01 -1.18662488e+00 5.90063214e-01
-4.20061082e-01 2.83906772e-03 9.18925762e-01 3.89569730e-01
-8.36241066e-01 -3.43713790e-01 -1.29852486e+00 5.59675880e-02
6.89054310e-01 -4.58303303e-01 -6.70474768e-01 5.47594845e-01
4.68976438e-01 -2.64880925e-01 -8.72319639e-01 5.05830348e-01
6.74527943e-01 -1.30751836e+00 9.20009315e-01 -7.01714933e-01
2.56671190e-01 -2.73766071e-01 -3.55062902e-01 -1.20289540e+00
-2.01927759e-02 -3.05481404e-01 1.21521298e-02 1.25763810e+00
3.84557664e-01 -9.67105150e-01 6.32901847e-01 8.28327894e-01
1.85467899e-01 -6.27780855e-01 -1.02845597e+00 -8.39149117e-01
4.92871046e-01 -6.83191359e-01 5.77059209e-01 1.13913774e+00
-6.50763065e-02 1.82854414e-01 -1.16211005e-01 1.92790613e-01
6.97873652e-01 -7.95419514e-02 8.37933481e-01 -1.77983928e+00
-3.69241863e-01 -5.57741582e-01 -6.76179171e-01 -4.72069681e-01
8.06674808e-02 -7.76469886e-01 1.73526987e-01 -8.47421825e-01
4.29103166e-01 -8.66649210e-01 -2.92959243e-01 5.69973469e-01
-2.88759530e-01 3.08634669e-01 -1.53679669e-01 3.73364776e-01
-1.88538402e-01 1.40116960e-01 4.70742702e-01 8.20410773e-02
-7.95133561e-02 2.91096181e-01 -1.01364350e+00 8.13631892e-01
9.69224155e-01 -9.49369788e-01 -8.22617859e-02 1.12304082e-02
5.42878509e-01 -1.76394746e-01 4.65309143e-01 -1.26969182e+00
1.81718856e-01 -9.98058319e-02 2.68010944e-01 -3.65202159e-01
1.34128094e-01 -7.51339436e-01 3.57909828e-01 4.46542352e-01
-3.66596788e-01 1.03902742e-01 1.73139274e-01 6.27821028e-01
-9.50422436e-02 -4.50181484e-01 1.15886426e+00 2.16987759e-01
-9.14409459e-02 -6.20911159e-02 1.39653251e-01 9.40750688e-02
9.40758646e-01 -1.50359526e-01 -3.65067631e-01 -2.70257264e-01
-5.85130692e-01 -7.08941743e-02 6.35903955e-01 1.13291107e-01
1.23182133e-01 -1.09693563e+00 -6.25989258e-01 6.48841262e-02
1.44235581e-01 9.46608037e-02 1.83820203e-02 1.03717279e+00
-3.88263792e-01 3.09973717e-01 -1.51450276e-01 -8.50089192e-01
-1.11588550e+00 7.03448951e-01 5.89256525e-01 -2.80043781e-01
-1.51791334e-01 8.86114717e-01 1.87731177e-01 -7.21890032e-01
1.78718954e-01 -4.66447681e-01 -6.77375942e-02 1.83888689e-01
5.32816887e-01 2.21728414e-01 4.57529485e-01 -4.92849499e-01
-3.85898322e-01 6.34495258e-01 -3.08641139e-02 3.72105390e-02
1.41789246e+00 1.18806124e-01 -2.27260113e-01 6.87261164e-01
1.14259112e+00 -4.81682159e-02 -9.14348423e-01 8.12974665e-03
1.89987555e-01 -4.84155416e-01 -2.34657839e-01 -7.53980815e-01
-9.13240612e-01 9.90352035e-01 6.87371969e-01 4.04816657e-01
1.25760150e+00 -4.56488691e-02 8.08902532e-02 5.09986162e-01
2.51099199e-01 -8.61890793e-01 -3.44283730e-01 2.24707603e-01
6.84050083e-01 -1.08838868e+00 1.43169984e-01 -4.33777243e-01
-3.78869653e-01 1.21462584e+00 3.16984415e-01 -2.37795547e-01
7.58489549e-01 3.69584173e-01 -1.33198693e-01 2.05844209e-01
-5.98992407e-01 3.14982474e-01 1.43129632e-01 6.09306455e-01
3.68492126e-01 3.00708503e-01 -2.33454272e-01 1.06993639e+00
-5.31302273e-01 -4.08548824e-02 7.38096297e-01 5.40834188e-01
-3.09340388e-01 -1.19675827e+00 -6.19874239e-01 5.60395718e-01
-3.04361999e-01 9.93956402e-02 -3.05621326e-01 8.11429322e-01
7.10670948e-02 8.39444280e-01 1.44252568e-01 -5.79340398e-01
3.25155914e-01 3.83714348e-01 4.90405530e-01 -3.63737494e-01
-3.22839797e-01 -5.37217855e-01 -1.12337530e-01 -5.90468347e-01
-6.07207894e-01 -7.82647967e-01 -7.54555106e-01 -5.83776414e-01
-6.62845969e-01 3.38863254e-01 7.90727913e-01 9.78112817e-01
1.74105570e-01 1.59428358e-01 4.73895371e-01 -8.71360183e-01
-9.99725223e-01 -9.01897907e-01 -5.07494807e-01 4.85766143e-01
2.51191944e-01 -9.72529531e-01 -7.92884707e-01 1.78928711e-02] | [8.410452842712402, 4.2953901290893555] |
40d7130f-1d33-41ad-867c-44f012e1d95b | learning-key-value-store-design | 1907.05443 | null | https://arxiv.org/abs/1907.05443v1 | https://arxiv.org/pdf/1907.05443v1.pdf | Learning Key-Value Store Design | We introduce the concept of design continuums for the data layout of key-value stores. A design continuum unifies major distinct data structure designs under the same model. The critical insight and potential long-term impact is that such unifying models 1) render what we consider up to now as fundamentally different data structures to be seen as views of the very same overall design space, and 2) allow seeing new data structure designs with performance properties that are not feasible by existing designs. The core intuition behind the construction of design continuums is that all data structures arise from the very same set of fundamental design principles, i.e., a small set of data layout design concepts out of which we can synthesize any design that exists in the literature as well as new ones. We show how to construct, evaluate, and expand, design continuums and we also present the first continuum that unifies major data structure designs, i.e., B+tree, B-epsilon-tree, LSM-tree, and LSH-table. The practical benefit of a design continuum is that it creates a fast inference engine for the design of data structures. For example, we can predict near instantly how a specific design change in the underlying storage of a data system would affect performance, or reversely what would be the optimal data structure (from a given set of designs) given workload characteristics and a memory budget. In turn, these properties allow us to envision a new class of self-designing key-value stores with a substantially improved ability to adapt to workload and hardware changes by transitioning between drastically different data structure designs to assume a diverse set of performance properties at will. | ['James Lennon', 'Andrew Ross', 'Sophie Hilgard', 'Mali Akmanalp', 'Harshita Gupta', 'David Li', 'Wilson Qin', 'Stratos Idreos', 'Niv Dayan', 'Zichen Zhu', 'Varun Jain'] | 2019-07-11 | null | null | null | null | ['layout-design'] | ['computer-vision'] | [-3.77886564e-01 -1.16404511e-01 -4.12821740e-01 -3.09195459e-01
-1.46037027e-01 -7.86886334e-01 3.15102518e-01 4.83698368e-01
5.10154888e-02 3.72123301e-01 4.88703072e-01 -8.79125237e-01
-4.14025635e-01 -9.87760007e-01 -6.10522449e-01 -3.18400532e-01
-2.57354200e-01 4.30295706e-01 5.20843029e-01 -4.35779840e-01
4.64555144e-01 8.22036862e-01 -1.96683264e+00 5.26551843e-01
2.41655365e-01 1.12476826e+00 -3.19685265e-02 3.95953536e-01
-3.74226451e-01 6.19441390e-01 -3.86717677e-01 -6.63116053e-02
3.57310474e-01 -3.07972450e-02 -1.05452657e+00 -2.67526627e-01
3.35743636e-01 -2.58412659e-01 -3.70144218e-01 4.03704107e-01
1.88890323e-01 -2.90875286e-01 3.82451087e-01 -1.18348610e+00
-7.43101597e-01 7.29088247e-01 -4.95670885e-01 2.71991551e-01
1.19543597e-01 2.12409258e-01 1.07808876e+00 -6.85616434e-01
4.39391583e-01 1.01505995e+00 6.62625015e-01 1.80459991e-01
-1.85874319e+00 -1.61264345e-01 6.62922859e-02 -2.51310050e-01
-1.49491298e+00 -7.58746803e-01 3.34193885e-01 -2.35795856e-01
1.08894134e+00 7.01776385e-01 5.46092629e-01 1.17372662e-01
5.80722034e-01 5.22323549e-01 9.96526182e-01 -4.54382360e-01
4.21480834e-01 -7.08996952e-02 5.36172330e-01 4.15396899e-01
9.67348993e-01 5.41554838e-02 -5.13948500e-01 -5.03981650e-01
5.07516384e-01 1.85802707e-03 -2.84337122e-02 -9.44473743e-01
-1.12857223e+00 3.00920963e-01 2.76814878e-01 3.81317228e-01
4.17434908e-02 4.68957275e-01 8.25154960e-01 4.85208273e-01
-3.18144888e-01 5.69261968e-01 -5.24859786e-01 -2.34252121e-03
-7.98409462e-01 5.21875560e-01 1.02060914e+00 1.13503182e+00
7.65640199e-01 1.48579016e-01 -1.80593237e-01 1.69533119e-01
3.91437739e-01 4.71532822e-01 1.95933282e-01 -6.62225187e-01
3.22963446e-01 6.33074999e-01 1.49169937e-01 -7.16830432e-01
-6.50911689e-01 -2.44244263e-01 -5.20027041e-01 3.45110968e-02
2.95832127e-01 2.15552807e-01 -7.10575998e-01 1.72526693e+00
1.19658731e-01 -6.03197753e-01 -1.21641543e-03 3.53860468e-01
4.21202660e-01 6.06597602e-01 -2.49120802e-01 -1.66576639e-01
1.37921166e+00 -2.40803465e-01 -5.07890224e-01 3.57537717e-02
9.81553257e-01 -6.36950016e-01 1.14610279e+00 2.28016242e-01
-1.39909065e+00 -6.82938814e-01 -1.31105983e+00 -1.57138646e-01
-4.01303858e-01 -5.14397740e-01 9.63970304e-01 9.45322871e-01
-1.41667354e+00 2.40814030e-01 -9.44180131e-01 -2.48580098e-01
-1.26599953e-01 5.98720491e-01 3.27286422e-02 3.80683780e-01
-7.35346377e-01 8.11403096e-01 4.89081204e-01 -1.87754616e-01
-5.58590055e-01 -9.77988303e-01 -2.68363893e-01 4.40997183e-01
4.14661407e-01 -7.90281415e-01 1.13346326e+00 -3.71804059e-01
-9.25184071e-01 7.25021064e-01 -1.55001402e-01 -5.11629105e-01
-2.05979273e-01 1.24804407e-01 -5.18714368e-01 -3.98212910e-01
-2.97636181e-01 1.47813931e-01 2.48644918e-01 -1.44248736e+00
-5.72171748e-01 -5.00574946e-01 -5.32744303e-02 -1.59150407e-01
-4.56838578e-01 -1.01956949e-01 -3.28054637e-01 -3.93860102e-01
1.91415980e-01 -7.97647357e-01 -1.69843525e-01 -7.95889199e-02
-3.35740179e-01 9.30601731e-02 6.74821734e-01 -3.85640413e-02
1.99586356e+00 -2.29772329e+00 -9.22798365e-02 5.98676443e-01
3.85442376e-01 1.79639217e-02 1.52553633e-01 9.78830636e-01
-4.45689278e-04 3.93676937e-01 2.23415539e-01 2.86165088e-01
2.04243928e-01 1.97650760e-01 -7.06764877e-01 2.53721923e-01
-1.62424758e-01 8.83970737e-01 -4.97363448e-01 -1.74506485e-01
-1.39472350e-01 -6.35790452e-02 -9.27987218e-01 -3.35478298e-02
-3.56032878e-01 -2.88857311e-01 -4.31562304e-01 5.15581131e-01
7.12176383e-01 -2.81176895e-01 6.92534566e-01 -2.87144065e-01
-6.08034909e-01 4.82917368e-01 -1.37125170e+00 1.12945521e+00
-2.51925260e-01 1.37968808e-01 1.41288295e-01 -8.45484138e-01
1.17207575e+00 -1.30763754e-01 6.38954401e-01 -9.21490431e-01
-6.40549809e-02 3.78943831e-01 1.28369629e-01 -2.19698012e-01
1.12769973e+00 -6.43965229e-02 -5.44900656e-01 8.71506393e-01
-5.30698597e-01 -5.77865541e-02 1.22107051e-01 1.94484234e-01
1.44673848e+00 -3.59068304e-01 5.26477061e-02 -1.00830352e+00
1.79635569e-01 8.18024483e-03 5.79781771e-01 8.60634744e-01
-6.88476339e-02 2.04592034e-01 6.54899359e-01 -5.85743666e-01
-1.45981920e+00 -1.42423844e+00 -4.17010367e-01 1.09304130e+00
1.19891793e-01 -9.82357144e-01 -2.98128605e-01 4.12636809e-02
5.42655051e-01 4.81354564e-01 -2.96727151e-01 -1.34124309e-01
-8.24165583e-01 -5.25304794e-01 5.08671403e-01 6.32173181e-01
1.71550319e-01 -4.00856972e-01 -1.14507639e+00 2.15068474e-01
4.60669696e-01 -5.50267220e-01 -3.24239641e-01 6.77802324e-01
-1.15830886e+00 -6.43837929e-01 1.40098363e-01 -7.35184550e-01
3.80592465e-01 4.49791014e-01 1.54357934e+00 5.11644721e-01
-1.71132371e-01 3.80110294e-01 -1.25343710e-01 -1.81636646e-01
-6.20033264e-01 1.49709880e-01 3.95728320e-01 -4.23897564e-01
2.79972523e-01 -6.87373638e-01 -6.67603910e-01 3.83896023e-01
-1.22339177e+00 -9.02590342e-03 4.88402009e-01 7.42773831e-01
6.78796589e-01 3.37424815e-01 5.97602308e-01 -8.56955707e-01
6.01797938e-01 -5.80786645e-01 -6.31638765e-01 5.06656349e-01
-1.08620691e+00 6.39715195e-01 8.32531691e-01 5.47332130e-03
-8.33472669e-01 -1.54438838e-01 1.36193223e-02 -3.07449363e-02
1.91215038e-01 5.46305537e-01 -5.23372531e-01 4.17208403e-01
6.52169883e-01 1.88066661e-01 2.31821746e-01 -5.55794537e-01
7.66957581e-01 5.86884618e-01 6.29036129e-01 -1.37101221e+00
6.02873743e-01 6.00409150e-01 1.63466707e-01 -5.72507083e-01
6.51511103e-02 -4.29205522e-02 -6.06903255e-01 2.68949479e-01
3.41319472e-01 -6.65815055e-01 -1.04671896e+00 2.07532868e-01
-7.53220558e-01 -2.52528250e-01 -5.44502795e-01 -3.07613999e-01
-6.25801146e-01 1.69469818e-01 -4.80998605e-01 -5.20643234e-01
-1.45145476e-01 -1.17616951e+00 8.31594706e-01 -6.69716299e-02
-5.11555672e-01 -7.91631341e-01 -1.18307166e-01 -1.21764757e-01
7.27475047e-01 1.23037435e-01 1.77503443e+00 -3.34411711e-01
-9.66097236e-01 2.51108617e-01 -1.87860012e-01 -6.53932765e-02
8.83387476e-02 3.42348158e-01 -4.59502310e-01 -6.61027908e-01
-8.07139277e-03 -1.07540026e-01 4.58948046e-01 1.08164668e-01
1.16145003e+00 -3.07585716e-01 -6.57526731e-01 4.90001410e-01
1.71285594e+00 4.81769919e-01 7.47455776e-01 7.74387196e-02
2.96312302e-01 3.88954133e-01 2.72072226e-01 6.18708968e-01
4.41068858e-01 9.31592405e-01 2.07189918e-01 2.96473689e-02
9.74084213e-02 -3.92175138e-01 7.69573227e-02 9.71261024e-01
5.45661807e-01 -5.79103492e-02 -1.21807301e+00 7.06410885e-01
-1.71673727e+00 -8.31929028e-01 -2.65385240e-01 2.63524270e+00
1.00154817e+00 6.07937217e-01 4.86295104e-01 2.10171893e-01
4.18785036e-01 -1.11382722e-03 -7.82162726e-01 -9.82088625e-01
5.07830493e-02 4.27546918e-01 8.58872950e-01 6.48159161e-02
-3.74729484e-01 4.85800833e-01 7.75194168e+00 4.81831938e-01
-1.09699297e+00 -1.75801203e-01 6.54077590e-01 -2.20880553e-01
-8.96164000e-01 3.77476543e-01 -1.20310974e+00 5.55579305e-01
1.50802124e+00 -7.07503319e-01 2.49044999e-01 6.56693041e-01
8.34508687e-02 -2.07145348e-01 -1.58637750e+00 6.70287967e-01
-4.90094751e-01 -1.89164877e+00 1.23152144e-01 4.81100321e-01
4.53746051e-01 -2.11037830e-01 2.03880891e-01 1.32949889e-01
4.94226992e-01 -9.08958912e-01 8.96807969e-01 4.87786442e-01
8.17395508e-01 -7.16228664e-01 1.65908813e-01 -6.51372829e-03
-1.32819974e+00 -4.06133652e-01 -1.20954283e-01 -2.00976953e-01
5.23884334e-02 5.64540088e-01 -2.04860345e-01 3.67999703e-01
6.88161194e-01 -2.18027085e-01 -5.89407742e-01 8.53560030e-01
7.53243089e-01 3.35032374e-01 -5.53892970e-01 -3.73439118e-02
-1.03068858e-01 2.31285796e-01 -5.23285121e-02 9.76864278e-01
2.72282898e-01 5.61748557e-02 -2.57882457e-02 8.40290129e-01
5.59594668e-02 -2.38145083e-01 -6.52251780e-01 1.29566938e-01
1.15538275e+00 5.23783684e-01 -8.23656797e-01 -1.52265489e-01
-6.42876387e-01 1.12663001e-01 2.51313329e-01 3.48390304e-02
-5.85321069e-01 -2.34042436e-01 9.98657227e-01 7.92786300e-01
3.76562476e-01 -4.49988097e-01 -9.99790967e-01 -7.74091959e-01
1.25060484e-01 -9.28783715e-01 3.56847793e-01 -4.08965439e-01
-1.09886348e+00 2.71636069e-01 4.12336230e-01 -8.42701495e-01
-4.78927686e-04 -5.02330005e-01 -4.19174522e-01 7.47161567e-01
-9.53717887e-01 -5.64434230e-01 1.12052590e-01 2.33049005e-01
-1.95907116e-01 1.64678935e-02 7.24004090e-01 2.66894966e-01
-3.19165051e-01 8.96301091e-01 6.92051232e-01 -3.04660082e-01
3.90361428e-01 -1.09512496e+00 7.94297099e-01 6.14712238e-01
-1.23550430e-01 1.19512558e+00 7.73058295e-01 -4.05935079e-01
-2.36049223e+00 -6.23975813e-01 5.69121242e-01 -7.73146808e-01
8.60526443e-01 -4.97506231e-01 -1.14235723e+00 6.77911103e-01
-5.08920522e-03 -1.29480064e-01 6.97582185e-01 4.71859396e-01
-5.31688988e-01 -6.37732804e-01 -9.44696426e-01 6.91078246e-01
1.14196527e+00 -5.42958915e-01 -5.22946417e-01 -1.51941627e-01
7.23777831e-01 -3.56973737e-01 -1.23457229e+00 3.86707842e-01
8.84345829e-01 -1.32184732e+00 1.03429830e+00 -7.31046617e-01
1.95515826e-01 -3.71378720e-01 -5.96423805e-01 -7.89313734e-01
-5.57474554e-01 -7.92762578e-01 -4.13187504e-01 1.21244633e+00
3.06782603e-01 -6.72325313e-01 7.45165706e-01 1.10629916e+00
-2.67574936e-01 -1.10997117e+00 -6.63075089e-01 -1.18149555e+00
5.13626039e-01 -2.21263766e-01 1.24176443e+00 5.43646276e-01
3.08519900e-01 4.29347605e-01 1.05902709e-01 -1.74081162e-01
4.12040651e-01 5.73021829e-01 1.08141398e+00 -1.12681675e+00
-5.77016294e-01 -5.46055794e-01 -3.88413697e-01 -1.52162111e+00
-5.08697808e-01 -9.73664582e-01 -3.97985250e-01 -1.20896113e+00
2.71512240e-01 -1.29171872e+00 -3.67927998e-01 4.93151277e-01
3.44570786e-01 -3.61962438e-01 3.19899887e-01 5.86536407e-01
-1.32670045e-01 2.41117686e-01 8.32868576e-01 2.37835318e-01
-4.15947706e-01 -1.96873441e-01 -1.31851530e+00 1.53482899e-01
5.55479109e-01 -9.79422256e-02 -7.67519593e-01 -5.33734381e-01
5.89123666e-01 3.11458498e-01 -4.81381416e-02 -8.75381231e-01
1.62582994e-01 -3.79857033e-01 6.50083125e-02 -8.18652928e-01
-1.06606983e-01 -8.56190145e-01 7.52248824e-01 5.93777359e-01
-1.47965953e-01 7.58994818e-01 5.07950485e-01 2.83517510e-01
1.08411491e-01 -3.28080729e-02 6.09149098e-01 8.65441412e-02
-7.64398754e-01 -3.80209088e-02 -2.71176368e-01 1.57466158e-01
1.01266932e+00 -6.73093081e-01 -8.34381104e-01 9.45643634e-02
-4.81930465e-01 1.80405766e-01 9.37527299e-01 4.14768070e-01
3.27412128e-01 -1.48173118e+00 -2.79432356e-01 5.58973670e-01
2.17168450e-01 -7.80383348e-02 4.91393991e-02 4.35744613e-01
-6.32825196e-01 6.55621290e-01 -4.21681881e-01 -4.96281952e-01
-1.02738774e+00 9.71186101e-01 7.97748193e-02 -2.47121051e-01
-5.31522572e-01 2.63667166e-01 2.78196990e-01 8.18698108e-02
-1.73337373e-03 -6.83488548e-01 7.47215986e-01 -1.30805388e-01
5.31722188e-01 1.40427828e-01 2.76733339e-01 -2.52244353e-01
-5.08115947e-01 2.23040879e-01 -3.25712264e-01 2.14449391e-01
1.29110777e+00 -1.82176575e-01 -3.79952431e-01 5.13400376e-01
1.03768909e+00 -6.92952424e-02 -8.27422619e-01 -2.71399230e-01
3.94680619e-01 -5.81151903e-01 -4.39464599e-02 -5.50600648e-01
-7.67901480e-01 5.06885767e-01 5.70819676e-01 6.94325984e-01
1.27322173e+00 1.96495373e-02 8.41527700e-01 3.72503340e-01
8.11774552e-01 -1.00986373e+00 2.29447465e-02 4.26500618e-01
5.23051143e-01 -4.14528042e-01 3.55919957e-01 -1.07699297e-01
-2.40007546e-02 1.06558359e+00 5.80309033e-01 2.25744724e-01
1.02957952e+00 1.08428454e+00 -4.66921419e-01 -4.77291793e-01
-1.11878431e+00 4.94985841e-03 -3.17536354e-01 4.74304140e-01
5.29563844e-01 1.88379243e-01 -5.60275495e-01 3.58325571e-01
-1.99236453e-01 9.86419395e-02 6.37209177e-01 1.43619478e+00
-7.43636727e-01 -1.57835627e+00 -4.39244032e-01 6.86500609e-01
-7.14315772e-02 3.27357613e-02 -1.24808237e-01 9.00171995e-01
-1.25102073e-01 6.36138797e-01 4.85503107e-01 -6.56867683e-01
5.20784736e-01 6.61200434e-02 4.37689781e-01 -6.20283842e-01
-7.22135305e-01 -3.60206544e-01 6.62757084e-02 -3.02527606e-01
2.03560397e-01 -4.52685684e-01 -1.50334597e+00 -1.27216351e+00
-1.01226315e-01 1.81571379e-01 4.28213388e-01 4.05778646e-01
9.97910857e-01 4.15899962e-01 6.62748039e-01 -2.49974713e-01
-7.86136687e-01 -1.86916515e-01 -7.67996252e-01 2.38067418e-01
1.30089611e-01 -5.38397908e-01 2.53485590e-01 -6.91741928e-02] | [8.050891876220703, 4.707298755645752] |
f9d22de0-8caa-4f6c-89e9-b6b861c12f05 | distributional-inclusion-vector-embedding-for | 1710.00880 | null | http://arxiv.org/abs/1710.00880v3 | http://arxiv.org/pdf/1710.00880v3.pdf | Distributional Inclusion Vector Embedding for Unsupervised Hypernymy Detection | Modeling hypernymy, such as poodle is-a dog, is an important generalization
aid to many NLP tasks, such as entailment, coreference, relation extraction,
and question answering. Supervised learning from labeled hypernym sources, such
as WordNet, limits the coverage of these models, which can be addressed by
learning hypernyms from unlabeled text. Existing unsupervised methods either do
not scale to large vocabularies or yield unacceptably poor accuracy. This paper
introduces distributional inclusion vector embedding (DIVE), a
simple-to-implement unsupervised method of hypernym discovery via per-word
non-negative vector embeddings which preserve the inclusion property of word
contexts in a low-dimensional and interpretable space. In experimental
evaluations more comprehensive than any previous literature of which we are
aware-evaluating on 11 datasets using multiple existing as well as newly
proposed scoring functions-we find that our method provides up to double the
precision of previous unsupervised embeddings, and the highest average
performance, using a much more compact word representation, and yielding many
new state-of-the-art results. | ['Haw-Shiuan Chang', 'Andrew McCallum', 'ZiYun Wang', 'Luke Vilnis'] | 2017-10-02 | distributional-inclusion-vector-embedding-for-1 | https://aclanthology.org/N18-1045 | https://aclanthology.org/N18-1045.pdf | naacl-2018-6 | ['hypernym-discovery'] | ['natural-language-processing'] | [-2.74126953e-03 4.46353227e-01 -8.46260011e-01 -4.48226452e-01
-4.09985900e-01 -5.69472134e-01 6.22339010e-01 7.08513856e-01
-8.91407311e-01 7.61109531e-01 7.13763714e-01 -4.90539342e-01
-5.17979622e-01 -7.92847455e-01 5.87990917e-02 -4.07343209e-01
-2.47187130e-02 1.12718892e+00 2.16048270e-01 -5.42514145e-01
7.51410704e-03 2.62395799e-01 -1.78428614e+00 -1.06618673e-01
6.24215007e-01 4.81904268e-01 -3.06885839e-01 1.75306007e-01
-5.76671541e-01 5.64803898e-01 -2.10564792e-01 -7.13416338e-01
5.35049103e-02 1.90011352e-01 -1.12191737e+00 -5.07744312e-01
4.82109129e-01 -1.51565254e-01 -5.87641418e-01 9.55079734e-01
4.11361843e-01 2.30810553e-01 7.15464175e-01 -1.36265337e+00
-9.16943789e-01 6.85228288e-01 -4.82327752e-02 1.46211505e-01
6.37405753e-01 -3.65367413e-01 2.03328681e+00 -8.37907016e-01
1.09331715e+00 1.12949610e+00 6.45096362e-01 6.16494477e-01
-1.24877203e+00 -5.77641010e-01 -3.34532857e-01 4.77011085e-01
-1.25289500e+00 -1.88375086e-01 3.53008479e-01 -2.20258057e-01
1.65946531e+00 2.89287984e-01 5.15165806e-01 8.59090507e-01
-2.07375169e-01 4.78884459e-01 6.55705214e-01 -8.56725752e-01
8.64678994e-02 1.78645730e-01 5.73361278e-01 6.17139757e-01
1.04750144e+00 -2.48776115e-02 -3.90984237e-01 -7.16752887e-01
2.92399883e-01 -3.21682915e-03 -1.03409924e-01 -8.13990235e-01
-1.35002983e+00 1.22335315e+00 9.00460482e-02 6.42677724e-01
-1.69842601e-01 8.99479985e-02 5.23381829e-01 3.63608479e-01
3.77933502e-01 1.10013998e+00 -7.01007366e-01 -2.25402653e-01
-6.34679317e-01 4.87681657e-01 1.11002743e+00 9.62530673e-01
7.81493425e-01 -2.23539397e-01 2.55056769e-01 9.43523049e-01
3.75257134e-01 2.15668485e-01 7.64392495e-01 -6.97510898e-01
3.18325400e-01 1.03820014e+00 2.48170812e-02 -8.41504455e-01
-6.00685954e-01 3.28061432e-01 -2.33793259e-01 -9.45086405e-02
6.16991855e-02 1.85532793e-01 -8.26022148e-01 1.66516674e+00
5.36849856e-01 -9.23743173e-02 3.57743621e-01 6.34525895e-01
1.22039831e+00 3.13097894e-01 2.48235419e-01 -1.90439001e-01
1.65235019e+00 -6.28554285e-01 -9.59118366e-01 -4.46595699e-02
1.17447567e+00 -6.37117207e-01 9.90681291e-01 1.69437416e-02
-5.86765409e-01 2.61294514e-01 -1.16470253e+00 -6.31830513e-01
-1.04121494e+00 -4.58854616e-01 1.18505836e+00 7.17890799e-01
-6.43816948e-01 3.83479178e-01 -4.47464615e-01 -9.36675131e-01
1.34773538e-01 4.49126422e-01 -9.39230323e-01 -1.62176669e-01
-1.66566133e+00 1.45502102e+00 8.32061231e-01 -6.31827593e-01
-1.88688308e-01 -8.28656137e-01 -1.41937196e+00 -2.38889866e-02
4.08591449e-01 -4.96613324e-01 9.49760437e-01 -1.03133515e-01
-8.64249825e-01 1.17214060e+00 -2.63692707e-01 -6.08450592e-01
-3.56063634e-01 -2.08439723e-01 -8.98657024e-01 -1.38791848e-03
2.08124995e-01 6.13565087e-01 3.15600485e-01 -9.60051537e-01
-2.35173449e-01 -2.76428044e-01 1.66549787e-01 2.57821858e-01
-8.91719997e-01 3.06708485e-01 8.18836875e-03 -4.87396091e-01
1.13473564e-01 -7.23328114e-01 -2.00444266e-01 3.05443108e-02
-1.36834070e-01 -7.62821317e-01 7.32064664e-01 -1.46076947e-01
1.66551852e+00 -1.95671284e+00 -1.38008595e-02 2.06373632e-01
5.89114666e-01 6.47018075e-01 -9.24928635e-02 8.54383290e-01
-5.25329888e-01 2.48571262e-01 -3.51312786e-01 1.30766466e-01
4.58582997e-01 9.18658018e-01 -2.80373305e-01 3.58956575e-01
1.17303997e-01 1.02488208e+00 -1.24539077e+00 -7.10047305e-01
4.01047051e-01 7.84832835e-02 -4.45570081e-01 -1.71391629e-02
-1.62998959e-01 -6.64215028e-01 -2.76465476e-01 5.06638765e-01
2.80645311e-01 -2.02825680e-01 6.26673639e-01 -2.17677116e-01
1.65228829e-01 6.59453094e-01 -1.24046326e+00 1.55307949e+00
-3.58136833e-01 7.12456346e-01 -5.37296176e-01 -9.18508708e-01
7.65647411e-01 6.07430935e-01 6.00901365e-01 -3.81618530e-01
3.54461074e-02 3.50393534e-01 1.42559856e-01 -8.98818791e-01
8.51894736e-01 -3.93734753e-01 -1.71339169e-01 5.38550854e-01
6.50381804e-01 -1.73161328e-01 5.09975433e-01 4.58330870e-01
1.29861200e+00 -3.93231332e-01 1.00091481e+00 -4.41016376e-01
8.35095495e-02 1.42743364e-01 4.02229995e-01 2.62468010e-01
-9.86563936e-02 2.66370386e-01 4.32840526e-01 -5.31954944e-01
-1.20010793e+00 -1.20145524e+00 -6.06927574e-01 1.27984810e+00
1.65468842e-01 -1.00420833e+00 -7.50081912e-02 -6.05708718e-01
4.42754269e-01 1.09323943e+00 -6.13929868e-01 -3.97138298e-02
-2.88680226e-01 -7.51788318e-01 8.44464481e-01 5.53396046e-01
-3.43780160e-01 -1.15433908e+00 -5.28854012e-01 2.60589838e-01
1.23632558e-01 -1.21847296e+00 -6.78392649e-02 5.30435443e-01
-8.30851495e-01 -1.47063792e+00 -9.31135863e-02 -1.13065338e+00
4.27347243e-01 -4.05200347e-02 1.40991223e+00 5.88755794e-02
-6.71270072e-01 3.70381504e-01 -5.63907981e-01 -4.41709429e-01
-9.63838305e-03 -7.09215505e-03 5.18036008e-01 -5.53102434e-01
1.46427560e+00 -6.31104887e-01 -6.30583614e-02 -1.74704596e-01
-1.31117415e+00 -7.18922257e-01 2.20950782e-01 1.01418233e+00
2.73501635e-01 -3.49217653e-01 5.52877724e-01 -1.41569543e+00
9.33737516e-01 -6.27257347e-01 -2.00223103e-01 3.39783281e-01
-1.13694775e+00 4.57376212e-01 1.25424564e-01 -4.08500582e-01
-7.58141816e-01 -1.57071620e-01 -1.83258444e-01 -1.79870829e-01
-1.47933573e-01 5.38862705e-01 -1.40512899e-01 9.01957229e-02
8.90044212e-01 -2.25609571e-01 -1.71378776e-01 -3.36291373e-01
1.20198810e+00 6.90290272e-01 4.05473024e-01 -5.07086277e-01
9.09506440e-01 4.01424885e-01 3.41649875e-02 -1.05743885e+00
-6.96449637e-01 -1.28708386e+00 -7.38077521e-01 5.16251981e-01
7.88911760e-01 -6.98166251e-01 -4.74062651e-01 -3.89464498e-01
-1.19825137e+00 4.97716099e-01 -6.27839267e-01 8.19607735e-01
-3.01133722e-01 6.88249230e-01 -4.07699496e-01 -6.13235891e-01
-3.51908416e-01 -5.23167133e-01 7.63071835e-01 -1.74075097e-01
-9.81354535e-01 -1.32778013e+00 7.07301199e-01 2.03795850e-01
1.48860797e-01 8.31668079e-02 1.56827772e+00 -1.48658860e+00
8.57654214e-02 -4.73305404e-01 -2.51329869e-01 8.50944519e-02
1.26836672e-01 -1.66908354e-01 -8.14582109e-01 9.67372283e-02
-3.98784459e-01 -6.92048728e-01 9.20654118e-01 -1.71285883e-01
6.16937935e-01 -3.27925384e-01 -3.88955355e-01 3.25709879e-01
1.58884919e+00 -2.96549141e-01 7.47094870e-01 4.78926092e-01
5.69516420e-01 6.62614644e-01 5.26243329e-01 3.83342892e-01
4.66678143e-01 3.62334371e-01 4.06526744e-01 2.46761560e-01
1.71056448e-03 -3.64670455e-01 -2.36389078e-02 1.08472383e+00
1.60292685e-02 -1.04277872e-01 -9.31341290e-01 1.24242914e+00
-1.71826208e+00 -1.16105795e+00 -2.38145724e-01 2.14481592e+00
1.15485871e+00 -1.97849229e-01 -1.87480494e-01 1.94098592e-01
4.18926299e-01 4.60836172e-01 -2.02984035e-01 -6.44569099e-01
-1.26403704e-01 9.01401341e-01 5.70672035e-01 6.76259816e-01
-9.66720283e-01 1.21060538e+00 6.84304333e+00 6.39999568e-01
-2.81956851e-01 3.51247072e-01 -5.27360380e-01 -1.35102078e-01
-9.42371547e-01 3.70559454e-01 -7.04039752e-01 -1.65402368e-02
6.94192410e-01 -3.68481725e-01 1.24760196e-01 9.26967978e-01
-3.56351346e-01 1.80200219e-01 -1.29367459e+00 1.05420876e+00
4.72037911e-01 -1.19176066e+00 2.40689591e-01 2.14352719e-02
6.80469453e-01 1.17002770e-01 -4.65654463e-01 3.93404454e-01
4.41018343e-01 -1.16216469e+00 -1.76280797e-01 -5.16146757e-02
9.30907190e-01 -5.24049044e-01 9.56983745e-01 -9.73271020e-03
-1.17166913e+00 -2.86365091e-03 -6.37836158e-01 -1.12517774e-01
3.09903800e-01 6.74446702e-01 -7.00353563e-01 3.91931146e-01
3.08488160e-01 6.05282962e-01 -1.48879185e-01 9.41605866e-01
-5.72647154e-01 4.96307641e-01 -4.78911310e-01 -4.95884538e-01
2.62316257e-01 6.69766739e-02 6.02585435e-01 1.28958213e+00
-1.36627018e-01 2.54818857e-01 -7.56119145e-03 5.06233215e-01
-3.99560481e-01 4.56476152e-01 -1.05904198e+00 -4.19284046e-01
7.69719064e-01 1.20794106e+00 -8.34703222e-02 -4.00613368e-01
-5.81661344e-01 5.46537459e-01 4.53677744e-01 1.35227233e-01
-5.81722736e-01 -9.99740601e-01 1.06649780e+00 -9.57616791e-02
7.97423646e-02 -1.84962526e-01 -1.13979019e-01 -1.34579444e+00
1.14836529e-01 -6.40882432e-01 7.17362285e-01 -3.87574285e-01
-1.63959730e+00 3.95143747e-01 3.33901078e-01 -1.05218101e+00
-4.42180216e-01 -9.20236707e-01 -4.27871794e-01 5.07886529e-01
-1.60567653e+00 -1.09648395e+00 2.30261505e-01 4.15778905e-01
1.57806575e-01 -2.92882711e-01 1.52360511e+00 4.32020575e-01
2.21518837e-02 7.86954343e-01 1.93094447e-01 1.03360541e-01
8.46115470e-01 -1.37567890e+00 1.61846995e-01 3.57109129e-01
7.36319661e-01 9.32210445e-01 6.12406492e-01 -5.38099229e-01
-1.34578502e+00 -7.48835742e-01 1.94680524e+00 -8.36069286e-01
1.21980953e+00 -2.41007924e-01 -9.92606997e-01 6.78139448e-01
1.68083116e-01 1.07484907e-01 1.22756314e+00 7.87982047e-01
-1.09982181e+00 2.23923802e-01 -1.18369198e+00 5.48669755e-01
1.06960070e+00 -7.08464324e-01 -1.65920222e+00 5.00368357e-01
1.00641644e+00 7.80930594e-02 -1.13962066e+00 4.43658382e-01
6.27214193e-01 -4.36126977e-01 9.29019570e-01 -1.40374911e+00
2.63075858e-01 -3.11105084e-02 -4.00158286e-01 -1.07698476e+00
-2.75206506e-01 -4.29998845e-01 -4.45059538e-01 1.05392981e+00
7.30187297e-01 -8.28411698e-01 7.09865093e-01 7.58838832e-01
1.94060475e-01 -7.08660126e-01 -1.05969739e+00 -9.06494141e-01
9.14478898e-02 -4.87706125e-01 6.71126127e-01 1.41114521e+00
9.59375858e-01 6.16282642e-01 -6.49784207e-02 -1.06972344e-01
6.56182468e-01 -2.49795392e-02 4.05651718e-01 -1.51055408e+00
2.35534668e-01 -3.42896819e-01 -1.02286720e+00 -6.81990325e-01
5.22202075e-01 -1.20354664e+00 -2.05029562e-01 -1.64524734e+00
2.52825022e-01 -1.96268499e-01 -3.34773123e-01 7.03581035e-01
-9.69399735e-02 9.59940106e-02 -1.55185372e-01 3.17498818e-02
-6.68120980e-01 6.07096612e-01 6.09885395e-01 -1.44626066e-01
1.27496466e-01 -7.94135213e-01 -5.74750006e-01 8.95603478e-01
5.77337384e-01 -7.76722312e-01 -5.05248249e-01 -4.34393346e-01
4.63413805e-01 -5.82702339e-01 9.45026577e-02 -4.31876987e-01
2.20018610e-01 -8.66446272e-02 -1.82311043e-01 -1.94268405e-01
3.42747360e-01 -9.37769651e-01 -4.77869183e-01 1.37603357e-01
-4.96516883e-01 9.28561464e-02 -9.43766311e-02 5.60480416e-01
-3.54747534e-01 -7.44536221e-01 3.64426911e-01 7.05081318e-03
-1.16597629e+00 2.27794141e-01 -2.04597831e-01 6.59769177e-01
9.26786602e-01 -8.76899362e-02 -3.13802898e-01 -1.29317835e-01
-4.28008765e-01 2.67568171e-01 3.61556321e-01 6.46030605e-01
7.62872636e-01 -1.51904905e+00 -5.00278473e-01 -4.47225384e-02
1.00546253e+00 -3.37167084e-01 -2.74137706e-01 3.07110310e-01
-4.76702660e-01 5.87688923e-01 8.73688906e-02 8.83555338e-02
-1.25899935e+00 7.13425040e-01 -1.95166305e-01 -3.38108331e-01
-5.43536544e-01 6.46035135e-01 -3.54894340e-01 -9.93316293e-01
3.35409075e-01 -3.47053260e-01 -4.21519607e-01 3.95240217e-01
3.77035648e-01 3.21203887e-01 -2.04590708e-02 -6.51396573e-01
-6.12860799e-01 4.26017046e-01 -1.22987092e-01 -2.40512118e-02
1.42451584e+00 3.67809922e-01 -5.55644453e-01 6.07595563e-01
1.46758258e+00 -1.01666205e-01 1.52581543e-01 -3.68759722e-01
4.14855093e-01 -5.03108561e-01 -1.92788973e-01 -3.65505785e-01
-2.93296129e-01 7.55950809e-01 3.05729270e-01 1.82869896e-01
4.35151905e-01 3.28685194e-01 1.03680146e+00 1.01917374e+00
3.59801471e-01 -1.07190049e+00 -2.64690995e-01 6.11349642e-01
3.93873423e-01 -1.20394254e+00 3.77673119e-01 -3.78380418e-01
-5.09254754e-01 9.91750002e-01 4.29737359e-01 -1.36791140e-01
9.38945770e-01 9.97004434e-02 1.83164384e-02 -6.33507073e-01
-8.35888207e-01 -7.18444705e-01 2.45532855e-01 7.33125746e-01
6.09726489e-01 2.24366724e-01 -7.18352318e-01 4.34363216e-01
-1.72586247e-01 -2.63251394e-01 2.78561980e-01 8.81644607e-01
-5.93435645e-01 -1.50387132e+00 1.81247056e-01 8.61730278e-01
-3.44761878e-01 -6.29054427e-01 -4.59865838e-01 1.00636816e+00
2.52848685e-01 8.11302304e-01 -9.58232731e-02 -1.84049219e-01
2.60555267e-01 4.03096139e-01 4.04809475e-01 -1.06109810e+00
-3.49148422e-01 -8.57783258e-01 5.85069478e-01 -4.14059043e-01
-6.18372440e-01 -2.54409939e-01 -1.39556813e+00 -2.00626299e-01
-6.63670897e-01 4.01127666e-01 4.19126958e-01 1.05537426e+00
1.06323302e-01 -1.25682428e-01 1.70746893e-01 -1.48477450e-01
-7.70431876e-01 -1.00845075e+00 -8.80636811e-01 9.81760800e-01
-2.04507634e-02 -9.00892854e-01 -4.26897436e-01 -2.39072934e-01] | [9.897062301635742, 8.744596481323242] |
c713a904-f6c1-4ebc-8eda-40fd10999061 | bringing-generalization-to-deep-multi-view | 2109.12227 | null | https://arxiv.org/abs/2109.12227v4 | https://arxiv.org/pdf/2109.12227v4.pdf | Bringing Generalization to Deep Multi-View Pedestrian Detection | Multi-view Detection (MVD) is highly effective for occlusion reasoning in a crowded environment. While recent works using deep learning have made significant advances in the field, they have overlooked the generalization aspect, which makes them impractical for real-world deployment. The key novelty of our work is to formalize three critical forms of generalization and propose experiments to evaluate them: generalization with i) a varying number of cameras, ii) varying camera positions, and finally, iii) to new scenes. We find that existing state-of-the-art models show poor generalization by overfitting to a single scene and camera configuration. To address the concerns: (a) we propose a novel Generalized MVD (GMVD) dataset, assimilating diverse scenes with changing daytime, camera configurations, varying number of cameras, and (b) we discuss the properties essential to bring generalization to MVD and propose a barebones model to incorporate them. We perform a comprehensive set of experiments on the WildTrack, MultiViewX, and the GMVD datasets to motivate the necessity to evaluate the generalization abilities of MVD methods and to demonstrate the efficacy of the proposed approach. The code and the proposed dataset can be found at https://github.com/jeetv/GMVD | ['Vineet Gandhi', 'Shyamgopal Karthik', 'Kanishk Jain', 'Swetanjal Dutta', 'Jeet Vora'] | 2021-09-24 | null | null | null | null | ['multiview-detection'] | ['computer-vision'] | [-3.86975199e-01 -6.27907634e-01 2.84981310e-01 -4.08844233e-01
-3.05964708e-01 -7.14334249e-01 6.41470730e-01 -1.66663215e-01
-1.77389666e-01 4.73088443e-01 -7.28256330e-02 -3.07923257e-01
-8.34811032e-02 -5.32619298e-01 -7.38413215e-01 -5.90998352e-01
3.23202610e-02 2.44205400e-01 6.43849254e-01 -2.81584054e-01
6.48100153e-02 8.52099895e-01 -2.02411199e+00 1.55267045e-01
7.99389839e-01 8.52567434e-01 4.25310105e-01 6.98191226e-01
2.41186336e-01 5.11134982e-01 -7.94426799e-01 -5.26462257e-01
6.78281605e-01 8.57942775e-02 -4.61255580e-01 5.54660678e-01
8.80206227e-01 -6.42010033e-01 -4.27127630e-01 6.88529968e-01
8.62361848e-01 3.60449880e-01 3.59807014e-01 -1.61094236e+00
-6.33276284e-01 -2.14536935e-01 -6.59133792e-01 4.25251454e-01
3.70053440e-01 4.70690250e-01 6.06585264e-01 -9.10993278e-01
5.79339027e-01 1.32753813e+00 8.47295523e-01 4.74241078e-01
-7.40940571e-01 -3.23807597e-01 5.54476261e-01 2.72261739e-01
-1.37144053e+00 -2.90398687e-01 7.65016317e-01 -5.56432188e-01
1.04758239e+00 1.66720167e-01 7.07144856e-01 1.29982018e+00
1.58824190e-01 7.41780698e-01 1.07543695e+00 -2.02587336e-01
3.24955940e-01 1.15031458e-01 6.09400030e-03 5.15726507e-01
4.78647292e-01 5.24713062e-02 -2.36202255e-01 -5.38871437e-02
4.12279695e-01 2.92631179e-01 -3.31139386e-01 -6.63782179e-01
-7.94765711e-01 7.73173928e-01 3.58015239e-01 3.53162326e-02
-2.14533150e-01 -2.10006356e-01 4.16734248e-01 1.11608349e-01
4.53027099e-01 9.92210731e-02 -4.37698781e-01 1.61826849e-01
-5.66475987e-01 5.32521248e-01 5.81609428e-01 1.23360538e+00
7.37032473e-01 1.21873364e-01 1.64529189e-01 8.11191857e-01
2.57811230e-02 5.84562600e-01 3.01243007e-01 -1.16930473e+00
6.65145040e-01 5.25505841e-01 2.12541550e-01 -1.10356820e+00
-6.21486843e-01 -4.57761556e-01 -7.00021505e-01 3.11039507e-01
2.36801267e-01 -3.13176930e-01 -7.33170152e-01 1.53062439e+00
5.36462605e-01 3.43860127e-02 6.36190400e-02 9.72059608e-01
1.02117598e+00 3.66304368e-01 -3.13556045e-01 -5.25521152e-02
9.39072073e-01 -1.13791418e+00 -3.65030289e-01 -4.26970899e-01
5.92604876e-01 -7.20348537e-01 1.20178187e+00 4.65402812e-01
-8.81772101e-01 -7.70866275e-01 -1.12379575e+00 -3.41630429e-02
-5.61877728e-01 1.05558641e-01 5.70990384e-01 6.22902334e-01
-1.20385039e+00 2.81637251e-01 -7.05472469e-01 -8.17474365e-01
2.73772001e-01 7.73842558e-02 -4.17752653e-01 -4.12345201e-01
-8.99496138e-01 8.20373535e-01 1.09876834e-01 2.12078661e-01
-1.19342017e+00 -3.98895353e-01 -7.63189733e-01 -1.98419616e-01
4.22670335e-01 -1.00499070e+00 1.09093714e+00 -7.66838133e-01
-1.00084162e+00 7.50670910e-01 -2.35692531e-01 -1.75579667e-01
8.54877472e-01 -4.62315917e-01 -4.67916995e-01 9.95709598e-02
1.49164021e-01 5.73594749e-01 6.84261739e-01 -1.68864250e+00
-7.82359719e-01 -5.23496926e-01 5.86661220e-01 3.41557503e-01
-2.47174650e-01 -2.46694386e-01 -4.68157679e-01 -3.50305796e-01
2.93622781e-02 -1.04827034e+00 -8.27924237e-02 1.53492577e-02
-3.25438380e-01 -3.76574248e-02 1.06920826e+00 -4.66756761e-01
9.40312088e-01 -2.30880189e+00 8.30620751e-02 -4.09575284e-01
2.38643155e-01 4.56382483e-01 -1.17515616e-01 6.28114998e-01
5.44131324e-02 -4.77749892e-02 -3.94467227e-02 -6.12965047e-01
-1.08733945e-01 4.84364331e-01 -8.25639069e-02 5.92727721e-01
-5.54013252e-02 4.14751917e-01 -6.88100576e-01 -2.57227540e-01
5.63749075e-01 3.91800165e-01 -6.40221655e-01 2.42447853e-01
-1.09893158e-02 4.83783782e-01 -1.20932817e-01 7.11654603e-01
1.08329487e+00 -1.16350532e-01 -1.20997325e-01 -2.38482550e-01
-2.33793497e-01 -3.58958900e-01 -1.49704921e+00 1.51444638e+00
-2.48303801e-01 6.45205557e-01 -1.87725369e-02 -8.62444580e-01
8.44007790e-01 6.14779256e-02 3.62940609e-01 -3.82647663e-01
1.31277576e-01 1.01310723e-01 -2.80946702e-01 -9.19970751e-01
6.26702011e-01 -4.34613451e-02 1.73724785e-01 5.17752096e-02
1.27066448e-01 1.35140531e-02 1.67346269e-01 2.00022399e-01
8.12474906e-01 3.11274398e-02 2.99540102e-01 -1.07692398e-01
4.89104033e-01 -1.69624209e-01 5.81717014e-01 9.16645050e-01
-4.55136806e-01 7.63913751e-01 2.24603891e-01 -1.04503393e+00
-9.84421790e-01 -1.08421016e+00 -1.08011551e-02 1.01236391e+00
4.35793191e-01 -1.03059292e-01 -6.06383801e-01 -5.67827940e-01
9.28045288e-02 6.55151963e-01 -5.90880752e-01 1.57323346e-01
-4.22388971e-01 -9.43348110e-01 4.31634158e-01 7.49934137e-01
8.53422701e-01 -3.77910703e-01 -8.55465591e-01 -2.52314568e-01
-3.54704976e-01 -1.47779930e+00 -1.52871296e-01 -9.98336449e-02
-7.82903433e-01 -1.19468451e+00 -5.62157929e-01 -4.97151852e-01
4.24155951e-01 8.77524495e-01 1.07540226e+00 5.06723225e-02
-2.12845549e-01 8.37380052e-01 -4.93194044e-01 -5.14163554e-01
-1.73079267e-01 -2.61823889e-02 4.47293788e-01 -7.55352676e-02
4.01129037e-01 -7.01630533e-01 -7.01792896e-01 5.27429879e-01
-9.94580567e-01 -1.46725044e-01 4.13959265e-01 4.61235166e-01
4.49570596e-01 -9.51660331e-03 2.18772054e-01 -6.03393793e-01
4.91350889e-01 -4.22420710e-01 -6.97066724e-01 2.41419435e-01
-4.03273523e-01 -4.57753956e-01 5.56282878e-01 -3.96119148e-01
-1.10172272e+00 8.93889442e-02 -2.02073827e-01 -7.69050062e-01
-4.93966341e-01 -2.82320362e-02 -3.59672457e-01 -1.64332047e-01
8.17625046e-01 1.10187799e-01 -2.80524164e-01 -4.89978760e-01
4.72198039e-01 7.31537700e-01 5.10636330e-01 -5.83003461e-01
6.46840930e-01 9.01719749e-01 6.27883524e-02 -1.15889537e+00
-7.72323549e-01 -4.97667104e-01 -9.25881207e-01 -4.58122820e-01
8.00496101e-01 -1.14770460e+00 -5.68153799e-01 6.01675272e-01
-1.14422178e+00 -2.32926488e-01 9.07385908e-03 4.11375850e-01
-4.86444473e-01 7.51766980e-01 -2.77515948e-01 -7.79613793e-01
3.83763723e-02 -1.24825013e+00 1.18825877e+00 1.58278182e-01
2.67789751e-01 -1.21545064e+00 5.32703772e-02 5.61659873e-01
2.26705745e-01 5.87192774e-01 6.06128097e-01 -4.63786125e-01
-5.72900832e-01 -1.69066951e-01 -1.36888310e-01 5.03235042e-01
9.75076556e-02 7.86837861e-02 -1.28075647e+00 -6.98018909e-01
2.40741208e-01 -2.31910259e-01 7.08505750e-01 4.37379122e-01
9.69713926e-01 5.65071292e-02 -3.23928207e-01 9.68637586e-01
1.53597677e+00 1.21099040e-01 4.91063356e-01 5.27696669e-01
7.05967307e-01 3.32640260e-01 3.48476559e-01 7.04922736e-01
7.62416899e-01 7.28244603e-01 8.79423261e-01 -1.63437381e-01
-1.57608911e-01 -1.63250174e-02 1.51404455e-01 7.66043246e-01
-3.67705584e-01 -6.15947127e-01 -8.41334939e-01 6.19267166e-01
-1.76279783e+00 -8.48392308e-01 -2.41124123e-01 2.05900049e+00
-2.17900962e-01 -7.44182020e-02 2.67163694e-01 4.58225720e-02
6.54871881e-01 2.40603372e-01 -5.80212831e-01 -2.09023416e-01
-4.77978975e-01 -4.94307935e-01 3.07635665e-01 4.04752761e-01
-1.09704852e+00 7.15695322e-01 6.46131277e+00 3.21224421e-01
-1.11370265e+00 3.47675234e-01 3.44694406e-01 -2.23340273e-01
1.70979556e-02 -9.99550521e-02 -9.40232813e-01 2.48340681e-01
1.93483353e-01 2.48721436e-01 4.06955034e-01 9.77365017e-01
9.97846574e-02 -1.35521144e-01 -1.07515037e+00 1.03740299e+00
4.95366514e-01 -1.01502764e+00 2.26010323e-01 -8.20809901e-02
8.20477366e-01 4.10139352e-01 1.57316867e-02 2.74069697e-01
1.75557718e-01 -4.66853887e-01 6.24171436e-01 3.90639454e-01
2.90653944e-01 -3.78987908e-01 8.63045335e-01 4.39625919e-01
-1.14901376e+00 -4.41034675e-01 -6.44113243e-01 -2.96595156e-01
8.39126632e-02 2.58490384e-01 -5.35708845e-01 8.71633649e-01
1.15584660e+00 5.85999966e-01 -1.01261961e+00 1.14317369e+00
-1.06614307e-01 1.27950922e-01 -3.34330082e-01 1.82629272e-01
3.60416412e-01 7.60697946e-02 7.70777285e-01 1.04494953e+00
4.83059198e-01 -7.41388500e-02 3.23496550e-01 5.35988331e-01
2.80509919e-01 -1.43003404e-01 -8.52854848e-01 6.92462564e-01
3.65286887e-01 1.00467217e+00 -4.88343358e-01 -7.80086666e-02
-8.81382287e-01 9.28243577e-01 4.90523040e-01 8.05039883e-01
-8.54610741e-01 -2.64280196e-02 6.49394870e-01 2.73829460e-01
5.04663765e-01 -3.70466828e-01 3.55533138e-02 -1.33098745e+00
3.86904746e-01 -8.36225808e-01 4.54318196e-01 -1.06331110e+00
-1.38655150e+00 7.26005137e-01 4.20287937e-01 -1.43482482e+00
9.31856781e-03 -7.08596945e-01 -4.34369057e-01 6.07790351e-01
-1.43779981e+00 -1.29135787e+00 -9.28647339e-01 7.39097834e-01
6.65267169e-01 -3.40452284e-01 4.55451280e-01 4.90041554e-01
-6.20296419e-01 4.45407271e-01 2.53604382e-01 -1.91825777e-01
6.96579456e-01 -9.41534758e-01 4.02673721e-01 1.03154290e+00
-1.40555277e-01 3.90448809e-01 7.88047850e-01 -3.42925996e-01
-1.24478090e+00 -1.17080951e+00 3.83301854e-01 -7.85777569e-01
2.76392341e-01 -4.38250273e-01 -8.24341416e-01 9.18636501e-01
1.69434100e-01 2.35944748e-01 5.33878446e-01 1.15168802e-02
-1.80233315e-01 -3.05289954e-01 -1.07261288e+00 5.06778061e-01
1.24044883e+00 -9.29093957e-02 -4.59660172e-01 4.36625749e-01
7.08002329e-01 -6.11133397e-01 -6.44673824e-01 5.23102045e-01
4.62111592e-01 -1.61896563e+00 1.01555145e+00 -5.69005728e-01
2.47030929e-01 -4.76925492e-01 -7.22309470e-01 -1.36619496e+00
-3.87569308e-01 -2.30139062e-01 -6.80413470e-02 1.02371168e+00
2.48256065e-02 -7.71092117e-01 4.74618942e-01 3.43836844e-01
-3.76304895e-01 -9.06935513e-01 -9.97183204e-01 -1.12632251e+00
1.91350758e-01 -4.41580743e-01 7.19018757e-01 8.12502563e-01
-5.76843560e-01 2.50930786e-01 -6.83387101e-01 6.62356555e-01
4.61949319e-01 1.63275614e-01 1.30127025e+00 -1.25604451e+00
-3.83177757e-01 -1.72756419e-01 -6.59055710e-01 -1.13221252e+00
-1.81177348e-01 -3.94019872e-01 -3.68700594e-01 -1.73138666e+00
3.08407426e-01 -2.27244496e-01 2.16267034e-01 1.16268829e-01
-2.49962524e-01 1.30615812e-02 3.37718189e-01 3.13415855e-01
-8.09447050e-01 5.73809803e-01 1.14734828e+00 2.43347064e-02
-1.41147226e-01 2.44565263e-01 -5.71176112e-01 8.94989967e-01
7.11831450e-01 -7.41819590e-02 -6.82222605e-01 -1.02133000e+00
2.04561263e-01 -2.02057958e-01 5.97519755e-01 -1.37294865e+00
1.19196199e-01 -5.41202873e-02 5.15691042e-01 -6.22625053e-01
6.12502515e-01 -7.84322143e-01 1.38147056e-01 3.07599664e-01
2.37665445e-01 5.59302628e-01 4.84504372e-01 5.81207752e-01
-3.62600647e-02 -1.91339806e-01 6.97106063e-01 -2.87833035e-01
-1.03584480e+00 2.60753959e-01 -2.83951700e-01 2.25684762e-01
1.17468870e+00 -3.19405973e-01 -6.43769860e-01 -5.60770154e-01
-6.72191083e-01 2.27285817e-01 7.15062618e-01 5.95247030e-01
5.57255685e-01 -1.14188313e+00 -5.65070450e-01 1.85506463e-01
3.23724717e-01 6.80842325e-02 5.82570314e-01 7.02586234e-01
-7.45949268e-01 2.11027026e-01 -2.65006512e-01 -8.94952774e-01
-1.22974312e+00 9.91628408e-01 5.79688728e-01 2.13595212e-01
-6.33103073e-01 7.45987058e-01 4.55413103e-01 -6.71978891e-01
3.21779311e-01 -2.92603135e-01 -4.47034426e-02 -2.13425025e-01
3.53297651e-01 7.09438026e-01 1.71006098e-01 -7.54294932e-01
-4.01592135e-01 7.57834315e-01 1.90360263e-01 2.99950212e-01
1.16175592e+00 -4.26441252e-01 2.35322207e-01 4.56074476e-01
8.81404459e-01 -4.73766364e-02 -1.44014025e+00 -1.64773330e-01
-6.24624372e-01 -7.00918019e-01 -2.43680656e-01 -4.46541101e-01
-1.11063790e+00 1.06398010e+00 9.47424650e-01 1.50755376e-01
1.25409341e+00 -4.08673808e-02 5.12912869e-01 5.68019569e-01
4.41725165e-01 -8.52988482e-01 1.74962506e-01 2.97660351e-01
7.52678931e-01 -1.44308555e+00 2.30067849e-01 -2.89075613e-01
-7.28177547e-01 9.65080380e-01 9.04947281e-01 1.70617163e-01
6.76447570e-01 1.43282533e-01 2.87138015e-01 -3.35596204e-01
-5.65872252e-01 -2.21654400e-01 -1.01594567e-01 9.03153718e-01
5.65375620e-03 -1.35645956e-01 1.48183197e-01 3.29848588e-01
-5.64124510e-02 -1.47982001e-01 6.96449935e-01 9.06936824e-01
-3.88558596e-01 -6.48084819e-01 -4.90372121e-01 1.52684629e-01
-4.65281494e-02 3.34768713e-01 -3.62881988e-01 1.15245926e+00
6.12415731e-01 1.20619512e+00 -8.83020684e-02 -5.79526782e-01
7.33420670e-01 -1.95653319e-01 4.13681388e-01 -3.48810375e-01
-2.52231896e-01 -1.91049665e-01 -1.64467216e-01 -4.77861196e-01
-6.26246274e-01 -8.45089197e-01 -4.21436995e-01 -4.83740866e-01
-2.57886201e-01 -1.65039629e-01 4.71258432e-01 9.63621616e-01
4.51715708e-01 4.97876018e-01 5.28855324e-01 -1.07509148e+00
-4.49897498e-01 -9.55484867e-01 -4.81491864e-01 6.27976418e-01
4.38916713e-01 -1.00627935e+00 -5.90841651e-01 -3.99627760e-02] | [8.046614646911621, -2.2086572647094727] |
35fa6153-0dda-47f6-9028-1ac771eeba11 | cross-modal-fusion-techniques-for-utterance | 2302.02447 | null | https://arxiv.org/abs/2302.02447v1 | https://arxiv.org/pdf/2302.02447v1.pdf | cross-modal fusion techniques for utterance-level emotion recognition from text and speech | Multimodal emotion recognition (MER) is a fundamental complex research problem due to the uncertainty of human emotional expression and the heterogeneity gap between different modalities. Audio and text modalities are particularly important for a human participant in understanding emotions. Although many successful attempts have been designed multimodal representations for MER, there still exist multiple challenges to be addressed: 1) bridging the heterogeneity gap between multimodal features and model inter- and intra-modal interactions of multiple modalities; 2) effectively and efficiently modelling the contextual dynamics in the conversation sequence. In this paper, we propose Cross-Modal RoBERTa (CM-RoBERTa) model for emotion detection from spoken audio and corresponding transcripts. As the core unit of the CM-RoBERTa, parallel self- and cross- attention is designed to dynamically capture inter- and intra-modal interactions of audio and text. Specially, the mid-level fusion and residual module are employed to model long-term contextual dependencies and learn modality-specific patterns. We evaluate the approach on the MELD dataset and the experimental results show the proposed approach achieves the state-of-art performance on the dataset. | ['Joshua Reiss', 'Huy Phan', 'Jiachen Luo'] | 2023-02-05 | null | null | null | null | ['multimodal-emotion-recognition', 'multimodal-emotion-recognition'] | ['computer-vision', 'speech'] | [ 1.01138549e-02 -5.11064470e-01 2.61394799e-01 -5.71108758e-01
-1.11682487e+00 -3.80188555e-01 6.41658306e-01 -5.79398237e-02
-3.15912843e-01 3.45723510e-01 6.39582872e-01 4.28232998e-01
-4.93873507e-02 -1.21094962e-03 -2.13396281e-01 -6.99606061e-01
-2.56569624e-01 4.20256779e-02 -3.16762209e-01 -4.37930554e-01
8.87044054e-03 2.46450976e-01 -1.93383622e+00 9.98270273e-01
5.44377267e-01 1.27748191e+00 8.90906826e-02 1.16103709e+00
-4.27621543e-01 1.16002977e+00 -4.34729546e-01 -1.06836200e-01
-5.33943117e-01 -7.22231865e-01 -9.10010695e-01 2.47702166e-01
-2.11714655e-01 2.43742406e-01 -2.03721151e-01 7.13525951e-01
8.73156130e-01 4.73919004e-01 7.15373814e-01 -1.30058229e+00
-8.82820711e-02 6.31847441e-01 -7.21006095e-01 3.90271433e-02
7.17095852e-01 -2.40343720e-01 8.70884120e-01 -1.12257349e+00
3.61728758e-01 1.52302516e+00 5.09164095e-01 6.15062654e-01
-8.64209533e-01 -4.09586459e-01 4.45711136e-01 6.03342116e-01
-1.26430333e+00 -7.23173738e-01 1.08715224e+00 -3.98927093e-01
1.00211811e+00 5.06033659e-01 4.20326501e-01 1.36100829e+00
1.06461808e-01 1.14496171e+00 1.06897295e+00 -5.56092203e-01
-8.83470327e-02 2.06946895e-01 2.20303729e-01 2.77209729e-01
-1.35992348e+00 -2.65634090e-01 -1.04591691e+00 -2.15185091e-01
8.98094177e-02 -2.31193230e-01 -2.94060320e-01 1.52059332e-01
-1.07851970e+00 5.90689063e-01 -1.45045146e-01 6.14986598e-01
-5.86247802e-01 -9.97292101e-02 8.96693707e-01 4.13220346e-01
2.76374012e-01 -1.36675075e-01 -4.06392545e-01 -7.58952320e-01
-7.32268333e-01 -1.12237431e-01 6.67587221e-01 5.86310089e-01
4.08303350e-01 2.39333455e-02 -3.05914227e-02 1.39843917e+00
3.82708311e-01 3.21917802e-01 6.77201629e-01 -5.86737633e-01
4.33535039e-01 4.17504460e-01 6.73218910e-03 -1.15123177e+00
-5.72507083e-01 1.34338126e-01 -1.11691511e+00 -2.59195149e-01
1.44818211e-02 -3.49569231e-01 -4.96990830e-01 1.92185700e+00
3.82842422e-01 2.19745219e-01 3.88478845e-01 8.81543815e-01
1.10697687e+00 1.15534186e+00 4.09460276e-01 -4.96918827e-01
1.52316451e+00 -8.72931838e-01 -1.16745603e+00 -1.34370506e-01
3.66511494e-01 -1.00221372e+00 8.70803058e-01 4.47627097e-01
-1.15132451e+00 -7.41295993e-01 -6.69627428e-01 -5.95416594e-03
-4.40840989e-01 2.84791142e-01 2.98680663e-01 2.23152548e-01
-5.46638787e-01 -2.31495854e-02 -5.64693093e-01 -2.62489498e-01
-1.73966736e-01 2.74276048e-01 -5.80635488e-01 1.14621870e-01
-1.60273433e+00 7.04111874e-01 3.00616384e-01 6.96610630e-01
-6.90087914e-01 -2.88974792e-01 -8.64047825e-01 6.33044690e-02
1.43476337e-01 -1.67371884e-01 1.14918530e+00 -1.29245019e+00
-1.74252653e+00 6.24273956e-01 -5.33539295e-01 6.64551482e-02
1.16949894e-01 -2.81738847e-01 -1.00397301e+00 1.30773365e-01
-6.35754406e-01 3.73725653e-01 7.99019158e-01 -1.41950750e+00
-7.88915575e-01 -4.55642700e-01 -4.01469946e-01 6.68199062e-01
-3.11380565e-01 5.44366896e-01 -5.18030226e-01 -4.73989785e-01
-3.81119736e-02 -7.83701658e-01 1.71113938e-01 -7.23591745e-01
-1.74752489e-01 -2.90238529e-01 9.38633084e-01 -6.51215434e-01
1.50448215e+00 -2.47837520e+00 6.05444729e-01 1.30928010e-01
-3.64132494e-01 -5.35709970e-02 -4.16511387e-01 8.81983459e-01
-2.61308104e-01 -2.98396170e-01 7.10339099e-02 -6.38228238e-01
2.75935233e-01 2.90180147e-02 -4.87758011e-01 1.68975204e-01
3.31103921e-01 6.37259722e-01 -5.24198174e-01 -5.58903992e-01
2.00565860e-01 9.43579435e-01 -3.41998506e-03 6.97347403e-01
1.96112134e-02 7.29906738e-01 -2.28723496e-01 7.35854685e-01
4.41957861e-01 1.80655777e-01 2.23521248e-01 -5.24201691e-01
-1.24041088e-01 -1.33352950e-01 -1.48952460e+00 1.74080813e+00
-6.62892997e-01 6.49772048e-01 5.00753403e-01 -9.03601706e-01
9.31259990e-01 8.17333400e-01 4.99020606e-01 -6.86062217e-01
4.74095851e-01 -2.24103648e-02 -1.69064984e-01 -9.99717891e-01
6.56731188e-01 -3.16641092e-01 -5.40882945e-01 2.77232319e-01
2.53278136e-01 1.45849481e-01 -9.56698880e-02 3.06278025e-03
6.55178964e-01 -4.04755883e-02 2.47297078e-01 2.69974470e-01
9.79281068e-01 -4.80448782e-01 6.48346841e-01 3.21815312e-01
-2.50260144e-01 5.10144711e-01 3.97019267e-01 -2.68759668e-01
-3.62682283e-01 -5.23285508e-01 1.10746972e-01 1.68395174e+00
9.00316387e-02 -3.21642101e-01 -4.19383854e-01 -3.50944728e-01
-4.58262086e-01 4.41877782e-01 -8.25812340e-01 -9.99402031e-02
-2.91617334e-01 -6.88719988e-01 7.10954309e-01 4.35893953e-01
3.11978877e-01 -1.27554166e+00 -4.68303233e-01 3.71617973e-01
-8.25329542e-01 -1.22937262e+00 -4.57918823e-01 3.08700770e-01
-2.25189939e-01 -8.04577351e-01 -4.54701155e-01 -7.33859658e-01
7.05940067e-04 -3.86016667e-02 9.63060141e-01 -5.07080138e-01
-4.28907394e-01 9.35601950e-01 -6.00150943e-01 -3.29320490e-01
-4.01238978e-01 -1.69696942e-01 -1.89495325e-01 6.91920102e-01
4.00392562e-01 -4.81815517e-01 -2.81331301e-01 3.53125691e-01
-9.89794135e-01 -1.41293302e-01 4.20436293e-01 1.02397859e+00
4.84314591e-01 1.11103266e-01 8.19399595e-01 -4.70120728e-01
8.45601439e-01 -6.72914267e-01 7.79221430e-02 4.20373559e-01
1.29173607e-01 -2.37721935e-01 3.59701395e-01 -8.66036713e-01
-1.53531706e+00 2.00480998e-01 -1.95536822e-01 -3.49360794e-01
-5.18347919e-01 8.36567283e-01 -5.72649539e-01 3.39560062e-01
2.02411264e-02 2.34023705e-01 -2.63550788e-01 -3.12932044e-01
4.52027559e-01 1.15380883e+00 6.44244671e-01 -7.09785104e-01
-2.70920582e-02 2.82484740e-01 -3.14738750e-01 -1.20466447e+00
-5.56364417e-01 -6.97874546e-01 -4.52112228e-01 -7.14625001e-01
9.71917868e-01 -1.17972410e+00 -1.04016817e+00 7.19581306e-01
-1.19060159e+00 3.73611078e-02 9.38634723e-02 5.45712948e-01
-5.25946200e-01 3.05960625e-01 -7.57370532e-01 -1.39246166e+00
-3.13902110e-01 -1.07997715e+00 1.21034825e+00 3.73605132e-01
-4.71169621e-01 -1.00388932e+00 2.69648314e-01 4.01928306e-01
2.62388170e-01 3.08564901e-01 8.55840504e-01 -4.98869002e-01
1.48127690e-01 -1.47010252e-01 4.39316481e-02 3.39671940e-01
3.82352173e-02 1.48602381e-01 -1.31555736e+00 4.56435792e-02
-9.95787457e-02 -8.42572868e-01 5.53244174e-01 5.97522594e-02
8.97021770e-01 1.45807043e-01 1.37783855e-01 1.30469531e-01
8.54641974e-01 4.60289299e-01 4.93410826e-01 -5.76061122e-02
5.13692975e-01 1.00566816e+00 8.40489447e-01 8.77377033e-01
6.34949505e-01 5.10486066e-01 3.07220817e-01 9.08830669e-03
4.08918381e-01 -5.18118106e-02 6.88391685e-01 1.51092243e+00
5.01958542e-02 -3.36522907e-01 -7.31028736e-01 5.69280803e-01
-2.21710515e+00 -1.29325807e+00 -3.90604287e-02 1.73502529e+00
8.38822603e-01 -5.31302989e-01 1.36245340e-01 1.97630569e-01
6.00805163e-01 2.82179952e-01 -3.27495337e-01 -7.11212039e-01
-4.75395471e-01 -2.22296283e-01 -6.50961459e-01 5.91233969e-01
-1.32320809e+00 6.49637818e-01 5.55034304e+00 8.23876619e-01
-1.18729317e+00 -1.87481511e-02 4.38733667e-01 -1.09155081e-01
-4.75473739e-02 -3.90543014e-01 -5.25009274e-01 3.55877727e-01
1.09025753e+00 1.37198195e-01 4.28482831e-01 3.59167278e-01
2.02740759e-01 -1.61077827e-01 -1.12463450e+00 1.47941256e+00
3.19796503e-01 -5.77379525e-01 -2.87602693e-01 -4.06028330e-01
5.47969341e-01 -2.84095764e-01 7.33757466e-02 5.94497979e-01
-3.02814215e-01 -9.87715900e-01 6.20273590e-01 9.63634074e-01
4.59836394e-01 -1.23213506e+00 8.62078369e-01 3.61482918e-01
-1.52326715e+00 -9.58895832e-02 1.39696136e-01 2.47923210e-01
3.05106968e-01 1.90797523e-01 -2.47974545e-01 6.30715966e-01
8.50319445e-01 5.18922508e-01 -1.00594170e-01 3.93357545e-01
1.74394190e-01 4.24559832e-01 -1.27271146e-01 5.12631461e-02
1.15585983e-01 -1.50954992e-01 5.22538543e-01 1.84246266e+00
3.25928152e-01 1.34783864e-01 6.53825924e-02 2.41695523e-01
-3.07331583e-03 3.69616270e-01 -2.96061277e-01 -3.95703614e-01
2.72839040e-01 1.36534142e+00 -9.86218676e-02 -8.95719901e-02
-4.20781016e-01 1.19206464e+00 2.13600278e-01 5.34363747e-01
-9.47737694e-01 -3.65729839e-01 7.48094618e-01 -7.01643527e-01
2.23104298e-01 -7.50770345e-02 2.48766467e-01 -1.19923043e+00
-9.77949575e-02 -1.29378915e+00 7.18743145e-01 -8.96716475e-01
-1.45306659e+00 7.90907562e-01 -2.79273868e-01 -1.10517085e+00
-4.85721678e-01 -2.30934873e-01 -4.46783274e-01 8.47971678e-01
-1.38117158e+00 -1.38861477e+00 -3.01058114e-01 1.03758633e+00
7.08966017e-01 -1.06321990e-01 1.16354001e+00 5.38626790e-01
-5.12953579e-01 5.66588998e-01 -7.12415297e-03 -1.33322701e-02
9.51433241e-01 -8.88514459e-01 -6.41595542e-01 4.19079632e-01
1.04041405e-01 3.92110676e-01 5.62142968e-01 -1.48199320e-01
-1.67804945e+00 -5.97122788e-01 9.57717597e-01 -8.58128294e-02
7.07000732e-01 -4.02117193e-01 -1.06523526e+00 4.77477610e-01
7.20468342e-01 -2.42426679e-01 1.13560510e+00 3.86230111e-01
-4.22914416e-01 -7.75599405e-02 -7.96630263e-01 4.12482917e-01
4.40532684e-01 -1.01860642e+00 -6.02806151e-01 -3.28561693e-01
3.12666446e-01 -2.95550823e-01 -1.04638398e+00 5.64295948e-01
8.17436278e-01 -9.89046156e-01 6.25163853e-01 -7.34927535e-01
3.37531835e-01 -3.15527439e-01 -5.56087077e-01 -1.30143547e+00
1.57037422e-01 -7.76111484e-01 -1.55397341e-01 1.85338688e+00
1.40681103e-01 -8.23303908e-02 1.13302112e-01 5.94135642e-01
1.05123945e-01 -6.71846032e-01 -1.08400738e+00 3.24943382e-03
-3.36486846e-01 -8.07318687e-01 3.97732049e-01 1.11931789e+00
6.26455367e-01 6.66808546e-01 -1.00672257e+00 3.69671136e-01
1.32101402e-01 3.17581564e-01 6.10211194e-01 -9.02376950e-01
-2.81147182e-01 -4.06717122e-01 -3.64929199e-01 -8.61902714e-01
4.46417153e-01 -3.82314831e-01 3.41670126e-01 -1.02678192e+00
2.74587214e-01 1.45569980e-01 -5.88389218e-01 2.27181524e-01
-1.25472397e-01 -1.20516628e-01 2.39798933e-01 -1.53535753e-01
-1.02755511e+00 1.10394061e+00 6.86600685e-01 -1.42448097e-01
-4.04972225e-01 -2.46216729e-01 -3.32657754e-01 7.27079749e-01
5.09180009e-01 -1.37280583e-01 -3.41814488e-01 -2.18767270e-01
2.97937989e-01 5.77454865e-01 6.21878281e-02 -7.18933225e-01
4.09104794e-01 -1.32503763e-01 2.87620276e-01 -9.01886880e-01
8.81466627e-01 -1.06968367e+00 1.64830118e-01 -3.66074443e-01
-6.50403857e-01 1.51916206e-01 4.05449748e-01 5.96188068e-01
-9.75398242e-01 1.45331249e-01 7.49481142e-01 2.17299595e-01
-8.21216702e-01 -7.76829245e-03 -7.62490451e-01 6.81788325e-02
7.84578800e-01 2.51691818e-01 -7.18514025e-02 -8.20632637e-01
-1.11568546e+00 4.60918486e-01 -2.10904896e-01 7.66316891e-01
7.69750118e-01 -1.53769457e+00 -5.99778354e-01 9.26353484e-02
3.69740844e-01 -4.49203700e-01 1.11906397e+00 8.73806179e-01
4.00509655e-01 2.45506205e-02 3.58240269e-02 -5.50195813e-01
-1.77192402e+00 1.78182289e-01 4.23230678e-01 -2.90915012e-01
1.06110103e-01 8.88551176e-01 1.03175513e-01 -5.64178109e-01
6.94550335e-01 1.11447528e-01 -5.46474457e-01 6.57794058e-01
7.58503318e-01 3.49919677e-01 -2.32088342e-01 -1.27137184e+00
-4.26291525e-01 6.27761841e-01 5.99739663e-02 -4.07099277e-01
1.39434421e+00 -7.45225132e-01 -2.96243846e-01 1.36151266e+00
1.26538241e+00 -5.34868836e-02 -9.17941213e-01 -3.31514537e-01
-1.05772071e-01 -5.09178117e-02 4.09551412e-02 -9.36065793e-01
-6.36329532e-01 1.27047062e+00 7.80030668e-01 1.93533137e-01
1.48736537e+00 -3.58150117e-02 6.48526251e-01 2.68479407e-01
-8.44105110e-02 -1.52500248e+00 2.30312958e-01 7.93261170e-01
1.12784636e+00 -1.26242852e+00 -5.47258675e-01 -1.92835964e-02
-1.29989517e+00 1.20007491e+00 4.47126687e-01 5.00209689e-01
6.61301076e-01 3.73500049e-01 6.08595550e-01 -6.90545663e-02
-1.14712250e+00 -1.85531870e-01 4.88107651e-01 3.23552817e-01
8.23411703e-01 -1.17527023e-01 8.38278159e-02 1.08986390e+00
3.08237255e-01 -1.64018109e-01 -2.57953443e-02 9.07079816e-01
-1.40779689e-01 -1.00187445e+00 -5.63040257e-01 -2.28678182e-01
-5.16413331e-01 1.87842473e-01 -5.06336749e-01 6.18872762e-01
1.59321893e-02 1.36101544e+00 -1.67793095e-01 -7.05064535e-01
5.87740242e-01 7.07767546e-01 2.02136546e-01 9.38938484e-02
-7.61722863e-01 6.94402874e-01 2.59145468e-01 -5.54004073e-01
-8.68937016e-01 -7.96093345e-01 -1.35614896e+00 1.54147819e-01
-1.37494028e-01 3.09790373e-01 7.44382560e-01 9.07921612e-01
5.89494824e-01 7.03937531e-01 9.70214486e-01 -9.75147665e-01
-3.35322946e-01 -1.18430746e+00 -7.55165100e-01 6.36015594e-01
4.55977350e-01 -3.93714428e-01 -4.78889436e-01 2.22873971e-01] | [13.232353210449219, 5.43088960647583] |
f9466f93-0db4-4649-bd66-e8b0e24354e1 | zero-shot-learning-by-convex-combination-of | 1312.5650 | null | http://arxiv.org/abs/1312.5650v3 | http://arxiv.org/pdf/1312.5650v3.pdf | Zero-Shot Learning by Convex Combination of Semantic Embeddings | Several recent publications have proposed methods for mapping images into
continuous semantic embedding spaces. In some cases the embedding space is
trained jointly with the image transformation. In other cases the semantic
embedding space is established by an independent natural language processing
task, and then the image transformation into that space is learned in a second
stage. Proponents of these image embedding systems have stressed their
advantages over the traditional \nway{} classification framing of image
understanding, particularly in terms of the promise for zero-shot learning --
the ability to correctly annotate images of previously unseen object
categories. In this paper, we propose a simple method for constructing an image
embedding system from any existing \nway{} image classifier and a semantic word
embedding model, which contains the $\n$ class labels in its vocabulary. Our
method maps images into the semantic embedding space via convex combination of
the class label embedding vectors, and requires no additional training. We show
that this simple and direct method confers many of the advantages associated
with more complex image embedding schemes, and indeed outperforms state of the
art methods on the ImageNet zero-shot learning task. | ['Samy Bengio', 'Mohammad Norouzi', 'Jonathon Shlens', 'Greg S. Corrado', 'Yoram Singer', 'Andrea Frome', 'Tomas Mikolov', 'Jeffrey Dean'] | 2013-12-19 | null | null | null | null | ['multi-label-zero-shot-learning'] | ['computer-vision'] | [ 5.18065393e-01 4.13967282e-01 -3.14561635e-01 -5.88585973e-01
-5.10041952e-01 -3.35826427e-01 8.53569746e-01 1.81974828e-01
-7.28874862e-01 3.37264121e-01 4.58105505e-02 -4.43728939e-02
-8.51229131e-02 -1.00357044e+00 -6.17891431e-01 -7.49905050e-01
2.86004782e-01 3.97386312e-01 7.93789849e-02 -1.47521988e-01
1.09507196e-01 2.05473781e-01 -1.68034399e+00 -1.92401692e-01
2.77139992e-01 1.03059149e+00 1.96874037e-01 7.13206828e-01
-5.03337741e-01 7.24322259e-01 -2.81633675e-01 -3.53549272e-01
3.41486841e-01 -3.18127096e-01 -9.19888735e-01 2.40225628e-01
4.34599698e-01 -2.35486969e-01 -5.57555676e-01 1.27530086e+00
1.01337291e-01 2.60723829e-01 8.61976743e-01 -1.61462045e+00
-1.28046227e+00 4.95421700e-02 -1.48900628e-01 -2.31603291e-02
6.35963380e-02 -1.40845075e-01 1.26754177e+00 -9.40390885e-01
7.97293067e-01 1.20558870e+00 4.53017354e-01 7.65416741e-01
-1.47861469e+00 -4.35119838e-01 -2.89290119e-02 3.43041390e-01
-1.36732364e+00 -3.36853802e-01 6.93106294e-01 -7.01750875e-01
1.01206529e+00 5.04788719e-02 4.20915931e-01 8.53254080e-01
-1.88628376e-01 4.36849028e-01 7.87482321e-01 -8.36731076e-01
3.88999552e-01 6.94326878e-01 4.06282783e-01 8.29194486e-01
9.42496285e-02 -9.54906195e-02 -2.44947746e-01 -5.88123389e-02
6.69318855e-01 3.88748765e-01 -3.64154913e-02 -9.60586071e-01
-1.10836422e+00 1.23373902e+00 7.69553781e-01 5.78647852e-01
4.75810245e-02 4.55329448e-01 4.29956734e-01 3.21825057e-01
7.24905849e-01 4.67773855e-01 -1.18317395e-01 1.83891535e-01
-5.14041662e-01 -9.93090346e-02 6.69586062e-01 1.08357954e+00
1.13893855e+00 -1.45481363e-01 1.08481944e-01 9.09591377e-01
2.34413475e-01 1.85317278e-01 5.67577839e-01 -9.87247288e-01
-3.08470782e-02 5.56227207e-01 5.81556857e-02 -9.83734965e-01
8.54260102e-03 2.24373549e-01 -5.09716868e-01 3.55589062e-01
-4.77343462e-02 4.25666034e-01 -1.07347345e+00 1.81802547e+00
1.93024129e-02 3.35069686e-01 2.44315311e-01 6.00249827e-01
6.64391518e-01 7.51537859e-01 4.19675440e-01 2.60641456e-01
1.61093771e+00 -8.92617464e-01 -7.14566529e-01 -2.73277074e-01
6.80605412e-01 -3.57740402e-01 1.13213551e+00 -1.61903724e-01
-6.17521405e-01 -6.89012289e-01 -1.33284068e+00 -4.12551820e-01
-1.10302305e+00 -1.53591171e-01 5.83041847e-01 6.37465239e-01
-1.23027515e+00 3.83334428e-01 -5.65513909e-01 -8.61317635e-01
5.49822569e-01 2.51108319e-01 -6.32781446e-01 -2.14303911e-01
-1.19802105e+00 1.10170519e+00 6.14947796e-01 -4.74447489e-01
-7.85579443e-01 -5.80743134e-01 -1.61280537e+00 2.10313246e-01
1.28149629e-01 -5.08540869e-01 9.81141090e-01 -1.01641488e+00
-1.11055493e+00 1.34245944e+00 -8.57280120e-02 -4.18227702e-01
-9.37972143e-02 7.42775872e-02 -3.24809700e-01 4.09659684e-01
3.22710097e-01 1.22244000e+00 1.08615029e+00 -1.18745112e+00
-4.18081135e-01 -3.50712746e-01 3.50132704e-01 7.76768178e-02
-8.55900228e-01 -1.49250731e-01 -1.13248222e-01 -5.86052239e-01
1.10634621e-02 -9.49579239e-01 -2.01837704e-01 6.25168443e-01
1.55688733e-01 -4.16057408e-01 1.10870349e+00 -2.76279151e-01
8.26359093e-01 -2.45278406e+00 2.35455215e-01 -2.27256306e-02
3.80521685e-01 2.59228498e-01 -4.14023131e-01 3.25689018e-01
-3.55828404e-01 2.79195905e-01 -3.86595726e-01 -2.70686746e-01
2.19244733e-01 7.07216263e-01 -4.13333595e-01 5.40559232e-01
3.61072838e-01 1.16022098e+00 -1.14626896e+00 -4.57772493e-01
5.91368973e-01 6.13628268e-01 -4.32740211e-01 1.34802639e-01
2.31961068e-02 -8.63285735e-02 -1.83299735e-01 1.12980209e-01
1.93294123e-01 -4.42126870e-01 2.86852941e-03 -2.03610912e-01
1.52155221e-01 -1.60549387e-01 -8.17601860e-01 1.97207236e+00
-5.31506538e-01 9.15031314e-01 -3.35437119e-01 -1.51837182e+00
8.74824107e-01 6.40200138e-01 6.06354296e-01 -4.88713175e-01
1.21364474e-01 -3.40630896e-02 -4.13949996e-01 -6.22108042e-01
3.57780010e-01 -5.71705878e-01 -2.63567656e-01 5.85695684e-01
8.08175862e-01 -3.44801426e-01 -1.20418988e-01 3.68919879e-01
9.04559433e-01 5.31788310e-03 4.89573717e-01 -1.92059278e-01
4.56026286e-01 -8.17761049e-02 1.57847360e-01 5.01542687e-01
-3.36407870e-01 5.18643200e-01 2.18219608e-01 -5.74753642e-01
-1.52754962e+00 -1.16730189e+00 -2.78950512e-01 1.18702555e+00
2.98419654e-01 -4.24660712e-01 -7.95810759e-01 -5.22324920e-01
-2.43072007e-02 8.13465714e-01 -9.32267725e-01 -4.28258866e-01
-1.67475909e-01 -4.15991724e-01 3.28457683e-01 6.03176594e-01
3.05698454e-01 -1.07046354e+00 -5.61423242e-01 2.69113984e-02
2.49591656e-02 -1.23530853e+00 -2.92759389e-01 1.74493983e-01
-5.93846023e-01 -1.03010654e+00 -6.88340783e-01 -1.32425916e+00
1.01065922e+00 4.35622096e-01 7.87337184e-01 -1.52280256e-02
-8.94297779e-01 8.86099219e-01 -4.15806264e-01 -3.48743379e-01
-3.18289459e-01 -3.10748488e-01 1.08489227e-02 2.14212850e-01
9.26257193e-01 -3.13872099e-01 -3.93987536e-01 5.69880269e-02
-1.27342606e+00 -1.45027235e-01 2.05462933e-01 9.19499576e-01
4.14058506e-01 -4.18344438e-02 5.32630324e-01 -8.92820001e-01
3.19562376e-01 -3.50394368e-01 -2.78884709e-01 2.22179636e-01
-6.41374588e-01 2.32851416e-01 1.84949785e-01 -3.40105265e-01
-8.02580833e-01 2.45473906e-01 1.19494885e-01 -6.15653038e-01
-2.36595020e-01 1.13800578e-01 7.71958753e-03 -2.63554782e-01
5.94360948e-01 2.11318061e-01 3.34734172e-01 -3.15526992e-01
9.91544485e-01 8.17951441e-01 4.26442504e-01 -2.48918802e-01
9.48778868e-01 8.08288693e-01 -2.67684549e-01 -9.96453702e-01
-9.06599104e-01 -9.32694495e-01 -1.03767514e+00 1.33257106e-01
1.56738389e+00 -7.99869955e-01 -3.70755345e-01 6.88716248e-02
-1.18524039e+00 1.66314900e-01 -7.76308656e-01 4.40786213e-01
-9.42286432e-01 3.96619052e-01 -5.59073567e-01 -4.97191548e-01
-8.86371881e-02 -1.06157696e+00 1.17967463e+00 -1.02576092e-01
-4.21611935e-01 -1.27091527e+00 5.95537387e-02 1.72986090e-01
2.63796955e-01 9.48805511e-02 1.30169833e+00 -6.44899070e-01
-2.56882101e-01 -5.02207756e-01 -5.10564446e-01 8.36854637e-01
2.55447000e-01 -4.36055034e-01 -1.18333912e+00 -3.11695874e-01
2.14380905e-01 -5.20650268e-01 7.92738438e-01 1.53234869e-01
1.14270484e+00 -2.30167344e-01 -3.32087994e-01 3.73232186e-01
1.91611958e+00 1.20824359e-01 6.23130262e-01 1.72734484e-01
6.52495503e-01 5.79780817e-01 3.13195199e-01 5.99747337e-02
2.61460006e-01 5.87152481e-01 3.93761396e-01 -2.28184238e-01
-5.53006902e-02 -2.47567207e-01 1.18461475e-01 8.07316899e-01
2.47376278e-01 4.69434448e-02 -6.88812137e-01 8.07792962e-01
-1.80776393e+00 -9.65034306e-01 5.59137762e-01 2.04500508e+00
6.22192681e-01 -9.33102593e-02 -3.08877349e-01 6.93838596e-02
7.07997561e-01 4.27573025e-01 -3.78836691e-01 -4.81440157e-01
1.44348383e-01 4.55378294e-01 5.26073158e-01 6.42478049e-01
-1.24587536e+00 1.04720497e+00 6.84780073e+00 5.59043467e-01
-7.63011634e-01 5.29428184e-01 3.49091142e-01 5.63490838e-02
-2.40012720e-01 1.78757235e-01 -4.75540489e-01 1.27775952e-01
8.45687032e-01 -3.71114224e-01 3.09186488e-01 1.01223779e+00
-3.73395443e-01 2.61509478e-01 -1.43791366e+00 1.21431053e+00
6.42281234e-01 -1.30846941e+00 1.92619860e-01 1.55462310e-01
5.33634484e-01 -2.30772689e-01 -2.54966561e-02 3.71102035e-01
2.00052425e-01 -1.13643301e+00 4.80322003e-01 3.93422604e-01
1.07210553e+00 -4.31763709e-01 5.67670584e-01 1.48822948e-01
-1.24341583e+00 -9.27962512e-02 -7.65752077e-01 -6.48442805e-02
1.07619420e-01 -1.12616853e-03 -6.12674356e-01 1.02801390e-01
5.06178677e-01 1.00715303e+00 -4.98525560e-01 5.21149039e-01
-1.42247260e-01 2.55261213e-01 8.13728347e-02 8.29891339e-02
4.68895048e-01 -1.90432936e-01 2.99400359e-01 1.04579067e+00
2.46471390e-01 3.56010765e-01 2.25159824e-01 1.01612747e+00
-1.63384050e-01 -8.54697451e-03 -1.13034689e+00 -2.89604247e-01
2.62058944e-01 1.29132414e+00 -6.94777846e-01 -7.41217196e-01
-6.79245472e-01 1.18815553e+00 2.65590072e-01 4.14317906e-01
-7.52508223e-01 -7.85978317e-01 8.40581715e-01 -3.37961167e-02
3.02675337e-01 -2.20766589e-01 -2.33497415e-02 -1.32346857e+00
-1.37899682e-01 -1.86299026e-01 3.24558556e-01 -9.12167251e-01
-1.35527158e+00 3.94659281e-01 1.74996946e-02 -1.22163761e+00
-1.36520520e-01 -1.05522084e+00 -4.67745215e-01 6.58948958e-01
-1.47557247e+00 -1.16004717e+00 -2.33095005e-01 6.69221103e-01
7.72083521e-01 -1.83752760e-01 1.41800117e+00 2.08797872e-01
-2.70521231e-02 3.25511873e-01 -4.89453599e-02 3.82204801e-01
4.92601693e-01 -1.23278499e+00 2.71621972e-01 4.61417317e-01
4.68067050e-01 4.70612288e-01 6.26233935e-01 -1.19444408e-01
-1.07495761e+00 -1.07552969e+00 1.14821184e+00 -6.51194632e-01
8.62935662e-01 -6.98663890e-01 -9.49102104e-01 9.41920042e-01
2.74522394e-01 3.52991849e-01 9.85403359e-01 -1.54991314e-01
-7.54789352e-01 -7.50569925e-02 -1.14108884e+00 4.43843991e-01
9.09963965e-01 -1.08674514e+00 -9.21475172e-01 5.51412880e-01
1.12583482e+00 3.65413874e-01 -1.05359900e+00 1.38738509e-02
4.40907240e-01 -2.70002991e-01 1.30208004e+00 -9.03506696e-01
2.89434910e-01 -9.87226665e-02 -5.27266085e-01 -1.16363502e+00
-4.24264163e-01 -3.65764201e-02 2.42193937e-01 9.31353629e-01
1.72006682e-01 -6.81312263e-01 6.11207664e-01 7.64133632e-01
8.15090612e-02 -3.13690126e-01 -9.58093643e-01 -8.65375996e-01
4.77604344e-02 -3.50321352e-01 2.79982775e-01 1.13065207e+00
2.74811476e-01 4.52882916e-01 -3.07402074e-01 -1.10914104e-01
8.71279359e-01 -2.21871018e-01 4.50957865e-01 -1.34324622e+00
-3.48650776e-02 -2.17873365e-01 -1.21413898e+00 -6.09849572e-01
5.39230347e-01 -1.24666858e+00 1.68411911e-01 -1.47465074e+00
3.23260486e-01 -2.47505322e-01 -6.36941433e-01 7.36145675e-01
1.18935838e-01 5.25585473e-01 2.58537531e-01 2.25688100e-01
-6.10242128e-01 6.52044058e-01 8.50151896e-01 -4.37735021e-01
3.36862415e-01 -6.09588563e-01 -5.94801962e-01 7.83965230e-01
5.55300057e-01 -6.20835245e-01 -6.93909347e-01 -4.40051943e-01
-5.16394898e-02 -3.14840496e-01 6.82137072e-01 -7.52315700e-01
1.10855088e-01 1.21473417e-01 1.36873350e-01 6.15637489e-02
5.96108675e-01 -1.07381785e+00 -1.99101523e-01 4.26543653e-01
-6.53247774e-01 -2.04997897e-01 -5.79851419e-02 8.34188938e-01
-4.13349599e-01 -5.30493438e-01 1.04210365e+00 -2.95319974e-01
-1.31249022e+00 3.71794850e-01 -2.06208944e-01 -1.84921309e-01
1.39462650e+00 -5.52820861e-01 -1.75572991e-01 -2.48234257e-01
-1.17179823e+00 -1.14090234e-01 5.25277972e-01 8.46278489e-01
7.36386418e-01 -1.59818292e+00 -3.35394740e-01 4.28095222e-01
6.86425686e-01 -4.63378668e-01 -2.60125324e-02 4.40718412e-01
-3.83386374e-01 4.57079172e-01 -4.12456423e-01 -5.86508512e-01
-1.05517328e+00 9.90584373e-01 2.01081142e-01 2.77893513e-01
-8.31998289e-01 7.84763873e-01 5.39142430e-01 -5.92740059e-01
1.11996718e-01 -1.82207450e-02 -9.27737653e-02 5.02916845e-03
5.29496193e-01 1.13386743e-01 -2.88834631e-01 -9.11768556e-01
-2.56224304e-01 7.29024529e-01 1.29218213e-03 -3.05385768e-01
1.42254901e+00 -2.37554088e-01 -3.53705108e-01 6.98113620e-01
1.86353302e+00 -8.35807979e-01 -9.64393914e-01 -6.87524140e-01
7.55882785e-02 -5.72457910e-01 1.09216338e-02 -1.92332834e-01
-7.75427938e-01 1.14637649e+00 8.40308428e-01 3.27596784e-01
8.20920885e-01 3.93399447e-01 6.36742055e-01 4.14916396e-01
4.62263972e-01 -1.17742443e+00 3.52135837e-01 2.43901521e-01
6.12348199e-01 -1.44333398e+00 -2.21088529e-01 -3.00939739e-01
-4.74215508e-01 1.05655324e+00 3.25788587e-01 -4.69060302e-01
1.04488873e+00 -1.99873015e-01 2.78555229e-03 -4.82340932e-01
-5.43299735e-01 -2.01740056e-01 1.13084242e-01 6.58742070e-01
2.18831852e-01 3.45361717e-02 -9.83280241e-02 1.48714781e-01
4.24378663e-02 -8.91479924e-02 2.17108354e-01 9.96463954e-01
-7.76079118e-01 -1.12949681e+00 2.55636089e-02 3.45789194e-01
-4.26367596e-02 -1.17779493e-01 -1.16988301e-01 7.11421311e-01
3.40826750e-01 8.26165617e-01 3.92751217e-01 -1.65497094e-01
7.67328963e-02 4.71113563e-01 5.27539194e-01 -1.15388620e+00
1.09684683e-01 -3.70098263e-01 -3.76400590e-01 -5.38096309e-01
-5.12662292e-01 -3.60921621e-01 -1.22009110e+00 9.97413248e-02
-2.39479885e-01 8.89378563e-02 9.16279733e-01 9.17149246e-01
1.97936907e-01 2.80360281e-01 4.97901112e-01 -8.79002154e-01
-4.72790778e-01 -7.37441421e-01 -7.50538349e-01 9.61090088e-01
2.50857979e-01 -7.76710451e-01 -3.31219405e-01 5.79263389e-01] | [10.053356170654297, 2.3336997032165527] |
5909ec70-d2c3-49c0-a061-b5b381f3561b | uamd-net-a-unified-adaptive-multimodal-neural | 2204.07791 | null | https://arxiv.org/abs/2204.07791v1 | https://arxiv.org/pdf/2204.07791v1.pdf | UAMD-Net: A Unified Adaptive Multimodal Neural Network for Dense Depth Completion | Depth prediction is a critical problem in robotics applications especially autonomous driving. Generally, depth prediction based on binocular stereo matching and fusion of monocular image and laser point cloud are two mainstream methods. However, the former usually suffers from overfitting while building cost volume, and the latter has a limited generalization due to the lack of geometric constraint. To solve these problems, we propose a novel multimodal neural network, namely UAMD-Net, for dense depth completion based on fusion of binocular stereo matching and the weak constrain from the sparse point clouds. Specifically, the sparse point clouds are converted to sparse depth map and sent to the multimodal feature encoder (MFE) with binocular image, constructing a cross-modal cost volume. Then, it will be further processed by the multimodal feature aggregator (MFA) and the depth regression layer. Furthermore, the existing multimodal methods ignore the problem of modal dependence, that is, the network will not work when a certain modal input has a problem. Therefore, we propose a new training strategy called Modal-dropout which enables the network to be adaptively trained with multiple modal inputs and inference with specific modal inputs. Benefiting from the flexible network structure and adaptive training method, our proposed network can realize unified training under various modal input conditions. Comprehensive experiments conducted on KITTI depth completion benchmark demonstrate that our method produces robust results and outperforms other state-of-the-art methods. | ['Huabiao Qin', 'Junli Lin', 'Guancheng Chen'] | 2022-04-16 | null | null | null | null | ['depth-completion', 'stereo-matching-1'] | ['computer-vision', 'computer-vision'] | [ 7.75109157e-02 -2.40002304e-01 -2.18253270e-01 -5.65003753e-01
-4.88654435e-01 6.32976964e-02 3.62754107e-01 -5.15364051e-01
-3.59008402e-01 5.40035129e-01 1.10094054e-02 -4.72817086e-02
-1.57754108e-01 -9.22932327e-01 -9.36198950e-01 -8.87273192e-01
6.64907932e-01 3.55596334e-01 3.21723014e-01 -1.41180918e-01
3.03190947e-01 1.62204310e-01 -1.74725306e+00 9.73000452e-02
1.12238252e+00 1.29094994e+00 8.13320398e-01 4.99560311e-02
-3.72825950e-01 7.92593718e-01 -1.45661980e-01 -1.34258822e-01
2.80069500e-01 -1.44009595e-03 -3.71976912e-01 1.60333231e-01
3.10660660e-01 -8.92619669e-01 -9.12931740e-01 1.23298454e+00
5.26932955e-01 1.59404829e-01 3.80214363e-01 -1.15828931e+00
-6.65119171e-01 2.02992499e-01 -5.73006570e-01 -1.73350379e-01
1.26955658e-01 2.95734644e-01 6.31657302e-01 -1.09860873e+00
3.68995458e-01 1.47966194e+00 3.19705814e-01 6.98941708e-01
-8.42833698e-01 -8.21447551e-01 2.44633049e-01 3.90220314e-01
-1.42294359e+00 -3.25068653e-01 1.04299033e+00 -2.97345757e-01
8.70604217e-01 -2.61674196e-01 5.62818348e-01 8.07184696e-01
1.40809387e-01 9.24388111e-01 8.65218818e-01 1.75386831e-01
-2.11826935e-02 -1.66625865e-02 -1.78417504e-01 7.46242285e-01
-8.38052332e-02 4.06544209e-01 -4.33428466e-01 3.61621857e-01
9.07151997e-01 4.58194077e-01 -4.30663556e-01 -4.18327928e-01
-1.24151433e+00 8.52225065e-01 8.24656010e-01 -1.35463595e-01
-2.17554033e-01 1.16447896e-01 3.20702404e-01 2.65888602e-01
2.86557287e-01 -1.55054659e-01 -2.84517080e-01 -1.20135853e-02
-6.47476017e-01 1.70319483e-01 2.38395289e-01 1.04886389e+00
1.28030229e+00 3.02504818e-03 2.16101870e-01 9.03455675e-01
6.58221781e-01 6.84730828e-01 5.42695045e-01 -1.08524215e+00
9.42547143e-01 9.06180382e-01 -2.39764780e-01 -1.04389954e+00
-3.84884685e-01 -1.98978826e-01 -1.11382556e+00 2.73211807e-01
4.61289473e-02 -8.65299776e-02 -9.63550985e-01 1.69334018e+00
1.10470891e-01 3.83740842e-01 2.53014982e-01 1.22324431e+00
1.29089594e+00 7.79161990e-01 -3.15852642e-01 -1.18444376e-01
7.73072302e-01 -7.62829363e-01 -5.84895670e-01 -4.76693332e-01
3.94039810e-01 -5.43252587e-01 9.30067658e-01 3.13222080e-01
-9.24052298e-01 -7.13606000e-01 -1.28745222e+00 -5.77699721e-01
-1.41109854e-01 -3.03793605e-02 7.80916572e-01 -1.04655242e-02
-7.89293170e-01 2.82692105e-01 -7.87145376e-01 1.45690851e-02
4.43221450e-01 6.79830074e-01 -3.98645461e-01 -6.25399411e-01
-1.06127107e+00 6.60480797e-01 4.94674563e-01 3.51805329e-01
-9.17979956e-01 -5.13519585e-01 -1.23542404e+00 -1.41363248e-01
9.74673703e-02 -9.57872868e-01 1.04694283e+00 -6.46658778e-01
-1.59867597e+00 5.34245193e-01 -3.66240323e-01 -1.15676515e-01
2.02642575e-01 -1.14454366e-01 -1.78583547e-01 1.06523246e-01
1.24042496e-01 1.23897505e+00 6.75035954e-01 -1.23517299e+00
-9.83654141e-01 -5.50225139e-01 2.55227268e-01 5.68583429e-01
-3.08178455e-01 -6.25387728e-01 -6.31826460e-01 2.97971386e-02
7.50878692e-01 -5.87474048e-01 -1.12650573e-01 -1.01793751e-01
-2.92128116e-01 -2.55228490e-01 9.49415684e-01 -3.41485322e-01
7.83569634e-01 -2.34230399e+00 6.22909665e-01 8.86830594e-03
2.12943256e-01 -2.79973179e-01 -9.01334062e-02 -1.53213674e-02
1.61677673e-01 -4.36760187e-01 -2.97369093e-01 -6.36559904e-01
-5.40019087e-02 3.89048159e-01 -3.00986648e-01 4.07746881e-01
1.47523165e-01 7.44442642e-01 -8.45263481e-01 -5.93577206e-01
7.15792000e-01 5.46898663e-01 -6.95480049e-01 3.06738257e-01
-2.52008766e-01 6.50264680e-01 -4.17393386e-01 8.02338541e-01
1.03363395e+00 -6.57962784e-02 -5.23175895e-01 -4.15915817e-01
-3.64311993e-01 2.31846124e-01 -9.62272644e-01 2.39617348e+00
-5.77548683e-01 4.11298424e-01 2.45004371e-02 -9.73087072e-01
1.11700916e+00 3.05196899e-03 2.77446568e-01 -9.12244618e-01
3.14817578e-01 5.11366367e-01 -5.17531596e-02 -6.83133900e-01
4.76242721e-01 -3.43039602e-01 1.59757048e-01 -8.61530080e-02
1.27979308e-01 -5.04568398e-01 -1.32222727e-01 3.22406814e-02
7.30716646e-01 2.55068779e-01 -4.75372851e-01 2.03504071e-01
7.31154859e-01 -1.49533436e-01 8.64956021e-01 1.06265740e-02
-7.21438304e-02 7.08867729e-01 2.67183214e-01 -2.83333421e-01
-8.81580055e-01 -9.82677579e-01 -3.05887878e-01 5.54347873e-01
9.22765672e-01 1.33623198e-01 -2.60984212e-01 -3.54618609e-01
8.21793154e-02 2.84400731e-01 -3.48085731e-01 -5.40212035e-01
-4.37223703e-01 -4.68274474e-01 9.55344737e-02 5.47788143e-01
9.80523050e-01 -1.01104677e+00 -2.32572615e-01 6.62390841e-03
-2.96893209e-01 -1.08921015e+00 -3.55353832e-01 4.98608164e-02
-1.15690553e+00 -8.48905504e-01 -7.39949286e-01 -1.13281548e+00
6.50378585e-01 6.29316509e-01 6.38163984e-01 -5.34682460e-02
2.70906657e-01 -3.83657664e-02 9.36128106e-03 -8.73872191e-02
2.06431195e-01 1.15882993e-01 3.84964934e-03 3.90105732e-02
6.77571297e-01 -9.55696523e-01 -8.58627617e-01 1.96287066e-01
-9.77201164e-01 3.35995913e-01 9.36695874e-01 9.99311149e-01
7.41707802e-01 -3.75598073e-02 4.79198247e-01 -4.20721203e-01
1.91678241e-01 -5.80083072e-01 -7.19067395e-01 -3.13982427e-01
-3.34461421e-01 -6.91993581e-03 4.35070395e-01 -2.13927731e-01
-1.11500359e+00 3.38975549e-01 -3.89771432e-01 -1.05040705e+00
4.05750051e-02 7.64000893e-01 -6.37335658e-01 -3.18253279e-01
2.57873476e-01 5.46923101e-01 2.66117752e-01 -3.35453808e-01
6.36853799e-02 7.49018371e-01 7.72494376e-01 -2.87725538e-01
8.77068222e-01 5.76494157e-01 1.28765464e-01 -4.52996582e-01
-6.43129826e-01 -3.55467111e-01 -4.53674972e-01 -1.61699235e-01
8.42045009e-01 -1.33405280e+00 -9.98518586e-01 7.64231622e-01
-1.28567231e+00 -2.01922040e-02 1.63841113e-01 8.23059261e-01
-5.36721408e-01 4.83140141e-01 -6.64486766e-01 -5.78759491e-01
-7.42674246e-02 -1.42818105e+00 1.22831786e+00 5.72331667e-01
6.39184415e-01 -7.57692039e-01 -1.97928935e-01 3.77546519e-01
1.13840006e-01 5.69743887e-02 8.30502927e-01 1.09959893e-01
-1.14842808e+00 -8.17474946e-02 -5.65222859e-01 4.28453922e-01
1.62452966e-01 -2.32821524e-01 -1.08996868e+00 -1.26632392e-01
1.05879389e-01 -4.88197327e-01 1.12271953e+00 4.48574930e-01
1.14348972e+00 2.00735554e-01 -2.75711209e-01 1.13528848e+00
1.44876099e+00 2.49483630e-01 7.11373031e-01 4.57005292e-01
1.12541175e+00 6.84676826e-01 8.16349804e-01 2.94826239e-01
8.42043340e-01 3.62948775e-01 1.15068245e+00 3.42309065e-02
1.37139425e-01 -4.90420818e-01 3.66897970e-01 1.09075356e+00
1.91360205e-01 3.63556594e-01 -6.61173582e-01 4.92057025e-01
-1.92285371e+00 -7.70240009e-01 1.22516483e-01 2.14686108e+00
6.00258410e-01 3.63269597e-01 -3.00614774e-01 9.10684988e-02
5.82576871e-01 1.44794241e-01 -9.81109619e-01 9.69803482e-02
-3.17456603e-01 -2.36056849e-01 4.17592198e-01 5.04533231e-01
-9.09300566e-01 7.98829198e-01 4.51324177e+00 8.01832438e-01
-1.37536716e+00 -3.06778941e-02 3.95293474e-01 -2.17644095e-01
-5.95432758e-01 -5.97023554e-02 -8.21046531e-01 6.85044169e-01
3.92929941e-01 8.65194723e-02 5.74150622e-01 8.68305624e-01
1.70220613e-01 -9.76155624e-02 -1.06001604e+00 1.47683632e+00
6.30794689e-02 -1.20590031e+00 1.26276836e-01 1.09808691e-01
6.16481245e-01 3.81088763e-01 2.02897429e-01 4.93027776e-01
-9.53322873e-02 -9.93805707e-01 6.22577965e-01 6.42671525e-01
8.83965492e-01 -9.52354550e-01 9.65336800e-01 6.11507952e-01
-1.25023854e+00 -2.93213993e-01 -9.00129616e-01 -2.40634322e-01
2.72517949e-01 6.28789425e-01 -5.53356372e-02 7.24157631e-01
6.72745705e-01 1.39457166e+00 -2.17525557e-01 1.04647386e+00
-3.84371467e-02 -9.88532603e-02 -4.32440341e-01 2.04248294e-01
3.27405989e-01 -4.45323378e-01 3.28320503e-01 4.92971152e-01
5.87163091e-01 7.06958398e-02 2.10032120e-01 1.02723217e+00
2.19677575e-02 -7.95286819e-02 -8.18813980e-01 3.19808960e-01
5.27353764e-01 1.11441350e+00 7.44090453e-02 -9.25965980e-02
-8.86142433e-01 7.51009524e-01 4.23499882e-01 3.65097761e-01
-5.12358010e-01 -5.02461851e-01 6.04149163e-01 -2.30955198e-01
1.63670659e-01 -2.63544142e-01 -5.18888533e-01 -1.32515740e+00
3.13819081e-01 -5.29094398e-01 3.11347488e-02 -1.03944075e+00
-1.22467721e+00 5.45760989e-01 -1.79683268e-01 -1.70340657e+00
-5.05557517e-03 -6.17476881e-01 -6.10644579e-01 1.17132223e+00
-1.96787834e+00 -1.16983461e+00 -8.26037049e-01 8.87899756e-01
4.75639462e-01 -1.29808292e-01 2.68108040e-01 7.43129551e-01
-6.60736382e-01 3.83385420e-01 -1.06199555e-01 -2.34983832e-01
4.89762455e-01 -9.85305130e-01 -1.32878259e-01 5.09429634e-01
-6.12597525e-01 4.73600835e-01 3.28908354e-01 -4.84268904e-01
-1.56739938e+00 -1.17908037e+00 6.43569648e-01 -1.17308736e-01
4.36149120e-01 -1.15891404e-01 -9.10278559e-01 5.83016336e-01
-6.90726042e-02 -4.60088179e-02 1.11187473e-01 -1.90511137e-01
-6.54473454e-02 -4.22832727e-01 -8.84021282e-01 5.17873824e-01
1.11799049e+00 -6.91387475e-01 -5.86836994e-01 3.77500057e-01
1.08947206e+00 -8.18423033e-01 -7.48579919e-01 8.21937382e-01
3.12832981e-01 -1.12274802e+00 7.82313764e-01 7.43623823e-02
8.80671024e-01 -5.43170214e-01 -3.60550284e-01 -1.09014130e+00
-1.09829307e-01 -6.55192956e-02 8.41282532e-02 1.15679944e+00
2.21482068e-01 -7.93243706e-01 9.28593576e-01 4.27790791e-01
-8.12032521e-01 -8.78326714e-01 -1.20182741e+00 -4.33497787e-01
1.17046319e-01 -4.01262999e-01 7.20067680e-01 6.30849600e-01
-3.15812714e-02 6.02404058e-01 -4.05801713e-01 4.16011184e-01
4.48944420e-01 2.37536818e-01 8.24079394e-01 -1.06804705e+00
-5.80509454e-02 -3.56612712e-01 -4.93935406e-01 -1.87564826e+00
3.30553204e-01 -8.37720990e-01 3.63163441e-01 -1.57973969e+00
1.33869812e-01 -4.64048535e-01 -1.28416091e-01 2.30961114e-01
-9.14279968e-02 4.03172746e-02 -4.89215441e-02 4.72301811e-01
-3.59148234e-01 1.19538641e+00 1.70944428e+00 -3.87931317e-01
-2.40367293e-01 -9.26705226e-02 -3.44303459e-01 5.77687919e-01
4.85606045e-01 -1.06048726e-01 -6.55458391e-01 -8.50248694e-01
2.60529846e-01 3.30672503e-01 4.91570950e-01 -1.13385868e+00
5.10602891e-01 -1.70027331e-01 3.28577667e-01 -1.07082689e+00
8.21521461e-01 -9.30776298e-01 -1.70454189e-01 3.38288009e-01
1.60008833e-01 -9.55749303e-02 7.53799006e-02 5.80133617e-01
-6.54730141e-01 -8.57755318e-02 5.75699627e-01 -2.38899756e-02
-1.02653253e+00 9.05324161e-01 1.47613958e-01 -2.58684754e-01
6.82649910e-01 -4.61376995e-01 -5.28167665e-01 -2.32937619e-01
-3.29079449e-01 7.18459666e-01 5.68261504e-01 4.85900015e-01
1.01798213e+00 -1.67641556e+00 -4.62220550e-01 4.53979969e-01
1.97382286e-01 9.32069480e-01 5.87972701e-01 1.07268369e+00
-5.59540391e-01 3.75543535e-01 -2.44335070e-01 -1.04543841e+00
-6.39878392e-01 5.40848672e-01 5.27514696e-01 2.62806386e-01
-6.91825330e-01 8.82284880e-01 6.04615569e-01 -6.70788705e-01
2.65758961e-01 -3.98721844e-01 -3.23963284e-01 -2.85831541e-01
2.17919573e-01 -2.23694909e-02 -9.27084759e-02 -7.69078195e-01
-2.04075426e-01 9.49732006e-01 2.88059376e-02 -7.02222586e-02
1.15661967e+00 -3.62327367e-01 -2.18691692e-01 5.75832665e-01
1.41561687e+00 -4.59545910e-01 -1.65257382e+00 -4.20056105e-01
-6.26493871e-01 -5.98249733e-01 2.31453210e-01 -1.70280263e-01
-1.46930385e+00 1.29507113e+00 6.46933675e-01 -2.27300271e-01
1.15283704e+00 -2.56243289e-01 1.19031024e+00 4.76560980e-01
3.75654221e-01 -8.15361977e-01 -3.17213759e-02 7.91811526e-01
6.51651859e-01 -1.64398050e+00 -3.01683217e-01 -3.78463835e-01
-3.96135867e-01 9.84846830e-01 1.09630525e+00 -2.77922481e-01
7.26341724e-01 -2.23625258e-01 1.95972383e-01 -1.90634936e-01
-7.13857353e-01 -1.04260854e-01 1.35715887e-01 5.06018877e-01
-1.36382990e-02 -2.75920123e-01 -4.51103635e-02 6.25192940e-01
-2.12012365e-01 1.82761088e-01 1.56832904e-01 7.30271995e-01
-6.32640898e-01 -7.12322652e-01 -7.84721375e-02 5.23551226e-01
1.57536432e-01 -2.69053966e-01 1.00068403e-02 6.40489578e-01
4.01556730e-01 9.21093524e-01 1.90272272e-01 -8.98977220e-01
4.11200017e-01 -2.75478363e-01 3.90492469e-01 -6.07755542e-01
-6.81519285e-02 7.85055291e-03 -3.71179819e-01 -5.54514647e-01
-5.42333245e-01 -5.92041314e-01 -1.49846601e+00 -4.74115431e-01
-3.06150496e-01 -9.50233787e-02 5.40104866e-01 1.27921247e+00
2.78451383e-01 4.88329738e-01 7.86142170e-01 -1.22601008e+00
-3.01216066e-01 -1.00440156e+00 -6.22897446e-01 3.36223602e-01
4.73664641e-01 -1.01467490e+00 -5.76269686e-01 -1.95718214e-01] | [8.899906158447266, -2.570754289627075] |
a9f88254-2412-4c22-93df-0b6f01f0ad83 | multi-instance-partial-label-learning-towards | 2212.08997 | null | https://arxiv.org/abs/2212.08997v1 | https://arxiv.org/pdf/2212.08997v1.pdf | Multi-Instance Partial-Label Learning: Towards Exploiting Dual Inexact Supervision | Weakly supervised machine learning algorithms are able to learn from ambiguous samples or labels, e.g., multi-instance learning or partial-label learning. However, in some real-world tasks, each training sample is associated with not only multiple instances but also a candidate label set that contains one ground-truth label and some false positive labels. Specifically, at least one instance pertains to the ground-truth label while no instance belongs to the false positive labels. In this paper, we formalize such problems as multi-instance partial-label learning (MIPL). Existing multi-instance learning algorithms and partial-label learning algorithms are suboptimal for solving MIPL problems since the former fail to disambiguate a candidate label set, and the latter cannot handle a multi-instance bag. To address these issues, a tailored algorithm named MIPLGP, i.e., Multi-Instance Partial-Label learning with Gaussian Processes, is proposed. MIPLGP first assigns each instance with a candidate label set in an augmented label space, then transforms the candidate label set into a logarithmic space to yield the disambiguated and continuous labels via an exclusive disambiguation strategy, and last induces a model based on the Gaussian processes. Experimental results on various datasets validate that MIPLGP is superior to well-established multi-instance learning and partial-label learning algorithms for solving MIPL problems. Our code and datasets will be made publicly available. | ['Min-Ling Zhang', 'Weijia Zhang', 'Wei Tang'] | 2022-12-18 | null | null | null | null | ['partial-label-learning'] | ['methodology'] | [ 5.60769141e-01 2.91859776e-01 -5.25048196e-01 -4.49966311e-01
-1.39572108e+00 -7.37129986e-01 4.32141751e-01 3.47062290e-01
-1.75579071e-01 1.03475440e+00 -6.94946945e-01 -1.64920747e-01
-4.05768394e-01 -9.26732183e-01 -8.52801740e-01 -1.08394706e+00
4.59227152e-02 1.23091519e+00 1.37574136e-01 6.69233978e-01
1.04699396e-02 9.97071043e-02 -1.59113514e+00 3.21400791e-01
9.59504545e-01 9.77135718e-01 4.44804225e-03 2.42245436e-01
-5.87216258e-01 6.25154674e-01 -4.53728527e-01 -3.59581202e-01
2.04304725e-01 -1.99588865e-01 -9.83898044e-01 3.02879602e-01
4.61095691e-01 4.03751075e-01 5.02829611e-01 1.37102222e+00
4.95243147e-02 3.69292162e-02 9.98873830e-01 -1.73893321e+00
-4.75120276e-01 4.10837442e-01 -7.87187099e-01 -2.66389817e-01
1.42813176e-01 -1.73730269e-01 9.96866822e-01 -1.02787960e+00
2.02064961e-01 1.55838490e+00 6.03144288e-01 2.77208209e-01
-1.49542713e+00 -8.63750279e-01 5.14434159e-01 -1.19050965e-01
-1.40729058e+00 7.98198655e-02 5.40780842e-01 -4.81618613e-01
3.09401065e-01 2.34080255e-01 3.74760255e-02 7.72955000e-01
1.02719590e-01 8.69080961e-01 1.70946026e+00 -4.24410969e-01
6.09359026e-01 2.08591089e-01 5.57219028e-01 7.13582516e-01
3.66353631e-01 7.64772519e-02 -2.42589906e-01 -5.32150924e-01
4.97520298e-01 8.13164487e-02 -1.38061093e-02 -3.72661144e-01
-1.20630515e+00 7.03125119e-01 -5.46655655e-02 1.40329376e-01
-3.18399400e-01 -2.67736632e-02 1.81843042e-01 7.55956024e-02
8.04504991e-01 3.55341882e-01 -8.35899293e-01 2.95339525e-01
-8.66025269e-01 1.13076590e-01 9.88259077e-01 1.11284268e+00
1.31601191e+00 -6.24387085e-01 -3.11380386e-01 8.59495819e-01
3.70754272e-01 6.67827368e-01 3.28281283e-01 -8.75420868e-01
4.40438777e-01 6.21927202e-01 3.46140683e-01 -6.35846257e-01
-6.00591302e-01 -4.78273839e-01 -7.13383257e-01 -4.84590717e-02
6.21069074e-01 -1.15287542e-01 -9.62908864e-01 1.79961646e+00
5.35246193e-01 7.90928483e-01 1.06252864e-01 6.08969390e-01
5.46468854e-01 8.77151072e-01 5.36769032e-01 -7.15584576e-01
1.34106970e+00 -1.05050278e+00 -7.68346310e-01 -4.34000015e-01
5.68071663e-01 -5.74000001e-01 9.61197197e-01 6.36460125e-01
-7.06497967e-01 -6.34235024e-01 -7.26734281e-01 4.75703806e-01
-4.08948392e-01 1.79800540e-01 6.37503505e-01 5.96254766e-01
-5.70694447e-01 3.18420321e-01 -6.95275903e-01 2.35921275e-02
2.77938306e-01 4.23542231e-01 -3.98490995e-01 -3.19251746e-01
-1.17425048e+00 4.59286571e-01 7.63195872e-01 -1.62729651e-01
-8.87917519e-01 -5.83314598e-01 -8.61941993e-01 -4.99067456e-02
8.64184618e-01 -2.74851441e-01 1.27661693e+00 -9.21261668e-01
-1.03083026e+00 1.01544499e+00 -4.67525989e-01 -1.52858317e-01
3.45949858e-01 9.71786305e-02 -4.38344628e-01 -8.25694762e-03
5.45914233e-01 3.56523216e-01 1.02632916e+00 -1.78457916e+00
-1.18214059e+00 -5.42159736e-01 3.98588367e-02 1.20263129e-01
1.05494067e-01 -2.17676356e-01 -3.50238979e-01 -4.68584508e-01
6.83217108e-01 -1.07753325e+00 -1.81462184e-01 -6.51704192e-01
-6.14888608e-01 -6.81127787e-01 8.78744602e-01 -6.61556125e-02
1.08196378e+00 -1.89438510e+00 -1.63442805e-01 2.27297783e-01
3.71760838e-02 2.10320622e-01 -1.75264001e-01 4.52143475e-02
-2.43419811e-01 3.57463717e-01 -5.56464374e-01 -5.01716495e-01
5.36284298e-02 6.65844917e-01 -2.95210361e-01 4.62188393e-01
2.22407416e-01 7.04839647e-01 -1.39281547e+00 -7.31800854e-01
3.04294601e-02 -5.41423932e-02 -3.73932309e-02 1.18469641e-01
-4.75974232e-01 4.45258200e-01 -4.89428967e-01 9.36923802e-01
8.72386634e-01 -5.08839190e-01 1.31881550e-01 -4.91076522e-02
3.38006504e-02 -1.98847562e-01 -1.67656767e+00 1.05219460e+00
-5.74137628e-01 -2.15121806e-01 -3.31167616e-02 -1.07848561e+00
9.27089453e-01 4.83289391e-01 5.28333306e-01 -7.79367834e-02
-1.82177752e-01 4.55191165e-01 -6.25549555e-01 -4.04029548e-01
7.76288584e-02 -8.71487379e-01 -2.80467957e-01 5.84259450e-01
2.84135550e-01 -1.87379003e-01 3.16440493e-01 -9.89592597e-02
6.92678154e-01 4.39325012e-02 3.52249682e-01 -1.80965990e-01
7.11240888e-01 -3.75578612e-01 9.78567064e-01 1.06615448e+00
-3.67391616e-01 4.81524110e-01 5.44689357e-01 -2.89828539e-01
-3.27095479e-01 -1.32942057e+00 -4.60672647e-01 1.30258739e+00
4.54649866e-01 -1.54816285e-01 -5.45510173e-01 -1.25430083e+00
-9.06403363e-03 7.38737762e-01 -4.66153324e-01 1.47353234e-02
-1.95097521e-01 -1.16459405e+00 2.19389379e-01 3.14011037e-01
4.14168030e-01 -1.03716338e+00 1.84450269e-01 3.97571117e-01
-3.62071812e-01 -9.27479327e-01 -3.04026783e-01 6.41402423e-01
-7.10887969e-01 -1.33275807e+00 -1.99317694e-01 -1.10095835e+00
8.27642560e-01 -1.32375164e-02 1.03362501e+00 -2.65967429e-01
1.62209690e-01 2.85322756e-01 -1.88287199e-01 -4.31693137e-01
-4.00321096e-01 -1.08986720e-03 2.80135304e-01 3.22695553e-01
6.68681085e-01 -2.73127675e-01 1.28637999e-01 5.08946598e-01
-1.00660431e+00 -2.64691800e-01 6.12477958e-01 1.13887382e+00
1.36282778e+00 6.75193310e-01 1.10324800e+00 -1.36843467e+00
3.62959355e-01 -6.43310189e-01 -7.07139075e-01 6.37942970e-01
-7.24239111e-01 1.20483048e-01 6.03991628e-01 -6.44651055e-01
-1.01173222e+00 3.30366313e-01 2.71642894e-01 -2.62521863e-01
-4.92356241e-01 5.88586450e-01 -6.45833790e-01 1.33067861e-01
3.95394355e-01 -7.00836033e-02 -3.60614359e-01 -2.67949104e-01
3.78289968e-01 6.27414584e-01 4.64115918e-01 -1.02884960e+00
7.14199781e-01 4.48330283e-01 3.18040550e-01 -2.72931278e-01
-1.67571855e+00 -7.48759568e-01 -8.54357600e-01 -1.71833754e-01
7.69825459e-01 -7.51179636e-01 -4.36983347e-01 4.64275002e-01
-8.88107419e-01 -2.33397931e-01 -2.52378345e-01 4.50455189e-01
-7.03582764e-01 2.83041090e-01 -5.80590427e-01 -9.79870021e-01
2.96838898e-02 -1.25810289e+00 1.24960792e+00 2.89871871e-01
7.53480569e-02 -1.26890004e+00 -8.72443523e-03 4.17996854e-01
-2.58243710e-01 1.58697814e-01 1.00685465e+00 -1.12016034e+00
-5.13832867e-01 -4.67984825e-01 -1.84312299e-01 4.36732471e-01
3.32605630e-01 -3.40833008e-01 -9.28841770e-01 -2.90818840e-01
2.08713397e-01 -5.87848604e-01 6.93932354e-01 4.23483312e-01
1.13531387e+00 -2.58671463e-01 -5.10112524e-01 2.24982470e-01
1.44511163e+00 2.04568416e-01 5.90703823e-02 2.73769677e-01
6.36981666e-01 5.84361911e-01 1.05071247e+00 3.79267275e-01
4.30724144e-01 3.00147474e-01 2.71204472e-01 5.52068055e-02
3.04239243e-01 -2.98542202e-01 -3.76791134e-02 5.06692648e-01
3.62897366e-01 -2.10109279e-01 -1.02408087e+00 1.96957782e-01
-2.02793217e+00 -8.13530207e-01 -3.58887106e-01 2.39955783e+00
1.05890095e+00 1.75733209e-01 -8.16484317e-02 1.94544062e-01
1.14612424e+00 -1.58749446e-01 -5.13090014e-01 9.22917053e-02
-2.05798149e-01 1.62174672e-01 4.43946093e-01 7.64369845e-01
-1.73895431e+00 9.97844160e-01 5.80255461e+00 1.08341622e+00
-6.99005246e-01 3.66528183e-01 8.88777316e-01 3.21768492e-01
-4.69726287e-02 2.13211268e-01 -1.18440831e+00 4.34728444e-01
7.13643551e-01 -3.80286500e-02 1.12756744e-01 8.96063507e-01
-6.53921813e-02 -2.41077363e-01 -1.22408295e+00 8.51569533e-01
8.97074640e-02 -6.69183671e-01 3.08229998e-02 -3.49682309e-02
1.19063509e+00 -5.14815629e-01 5.90450652e-02 7.67634094e-01
4.35299844e-01 -8.80562603e-01 5.05671084e-01 5.42090595e-01
7.35536814e-01 -9.59923565e-01 8.91234636e-01 7.82841563e-01
-1.18691552e+00 -1.44431174e-01 -3.68308157e-01 1.98673550e-02
1.72019020e-01 8.93306494e-01 -6.64441049e-01 5.92829585e-01
2.87343115e-01 4.26123232e-01 -4.01641101e-01 1.04869533e+00
-5.36939263e-01 8.09501588e-01 -2.82202780e-01 1.98949620e-01
4.12287503e-01 -3.59356552e-01 3.55808824e-01 1.06073701e+00
1.55008167e-01 1.99756369e-01 1.07214165e+00 7.27015138e-01
-5.37729561e-02 1.24711044e-01 -2.42378742e-01 2.80357242e-01
6.24790192e-01 1.36463022e+00 -9.50011790e-01 -5.79688787e-01
-4.97662723e-01 6.76588297e-01 3.70456874e-01 6.19591355e-01
-7.59566426e-01 -2.42997613e-02 2.66834527e-01 -7.35730827e-02
-6.35256618e-02 2.92666644e-01 -3.63668710e-01 -9.81652319e-01
-1.28086835e-01 -6.67245388e-01 6.60736024e-01 -4.72232252e-01
-1.82738876e+00 2.58257091e-01 1.11166351e-01 -1.07341886e+00
-1.85994104e-01 -6.57993197e-01 -2.36192212e-01 9.89933252e-01
-1.37840521e+00 -1.33575678e+00 -1.29398759e-04 4.17098612e-01
4.62370127e-01 1.34371191e-01 8.90273213e-01 1.63044035e-02
-5.61921477e-01 4.05268043e-01 4.54862043e-02 -2.00027227e-01
9.38166916e-01 -1.64930737e+00 -3.38555992e-01 5.35020292e-01
2.48096019e-01 3.93037319e-01 6.04316711e-01 -7.56245255e-01
-7.94387579e-01 -1.60242426e+00 9.42954957e-01 -6.11535728e-01
6.20399356e-01 -2.54979074e-01 -1.18205440e+00 9.95118678e-01
-3.24687064e-01 3.40488493e-01 1.01344407e+00 3.85319263e-01
-3.40468198e-01 -1.04240939e-01 -1.34780729e+00 1.49135381e-01
6.15142286e-01 -5.30945480e-01 -6.90577626e-01 8.71467948e-01
6.64650798e-01 -3.65579903e-01 -7.89959133e-01 6.74520969e-01
-8.57463330e-02 -4.69516248e-01 8.89257789e-01 -5.49699783e-01
1.46880776e-01 -6.30988836e-01 -1.11708678e-01 -1.31616747e+00
-4.98770773e-01 -9.39969197e-02 -2.14851171e-01 1.44635010e+00
5.91969252e-01 -6.52121723e-01 6.77794814e-01 8.20283413e-01
4.53998847e-03 -8.85808110e-01 -1.04060495e+00 -1.05189908e+00
1.97751373e-01 -6.30498528e-01 6.14286959e-01 1.12527108e+00
-9.44152400e-02 3.30034763e-01 -2.22770497e-01 6.71891153e-01
1.09413600e+00 5.82175314e-01 3.16107541e-01 -1.57120407e+00
-4.28972363e-01 -1.68312773e-01 -1.80090487e-01 -7.04487860e-01
7.21487641e-01 -1.11216640e+00 3.71986598e-01 -1.54253554e+00
4.15420532e-01 -1.02178502e+00 -4.39479589e-01 7.60481596e-01
-6.23695135e-01 2.39943922e-01 -4.16683182e-02 2.74630070e-01
-1.11504173e+00 2.43604854e-01 1.06779146e+00 -2.23840937e-01
-1.70558408e-01 4.62453991e-01 -5.83209991e-01 9.39095676e-01
7.67522752e-01 -7.02975810e-01 -3.20676535e-01 2.05553874e-01
1.85646676e-02 1.11933917e-01 1.79721519e-01 -9.37872827e-01
3.02045345e-01 -3.63726258e-01 1.73236132e-01 -7.10398972e-01
7.35464469e-02 -7.41162241e-01 7.30014220e-02 2.09380910e-01
-4.56013381e-01 -5.81832707e-01 -8.42684135e-02 8.92850518e-01
-3.43032748e-01 -7.63867259e-01 9.26159620e-01 -1.97767079e-01
-6.49998963e-01 4.77411985e-01 -1.68132007e-01 2.49486834e-01
1.27964449e+00 2.39105359e-01 -1.67369261e-01 -1.53275933e-02
-1.17234945e+00 5.40056467e-01 2.18979567e-01 1.72608480e-01
2.34893098e-01 -1.43289483e+00 -5.70070505e-01 5.60438037e-02
3.49908799e-01 3.84919643e-01 9.76980403e-02 5.57115555e-01
2.73149908e-01 4.49840456e-01 5.83095074e-01 -7.70135820e-01
-1.02881801e+00 1.02912033e+00 1.61179632e-01 -6.77877605e-01
8.31587054e-03 8.94598722e-01 5.34887016e-01 -9.66203809e-01
1.97717980e-01 -1.63993433e-01 -8.30869749e-02 2.48910218e-01
5.44526637e-01 3.36276412e-01 -1.28505109e-02 -6.89467609e-01
-1.51018098e-01 5.95609128e-01 4.27883044e-02 7.40037635e-02
7.74532378e-01 -1.09666228e-01 -5.49942017e-01 9.59726632e-01
1.15148461e+00 -2.40913764e-01 -1.23043597e+00 -6.36774421e-01
4.20694709e-01 -2.92739183e-01 -1.80514783e-01 -9.55863237e-01
-7.98148036e-01 5.35593569e-01 3.64584625e-01 2.96547949e-01
1.06701922e+00 3.16043228e-01 2.19286829e-01 3.27073514e-01
8.35904062e-01 -1.05022085e+00 1.82915300e-01 6.71388924e-01
4.78945643e-01 -1.45625150e+00 -3.30296338e-01 -7.41696596e-01
-4.98352796e-01 9.43359613e-01 8.98958206e-01 1.31565899e-01
7.55242407e-01 2.95682251e-01 1.40178017e-02 -4.86326031e-02
-6.47702277e-01 -1.19207956e-01 1.98683590e-01 4.29442048e-01
2.83336341e-01 4.92146164e-01 -2.74843574e-01 1.15165246e+00
3.07031870e-01 -6.79727346e-02 1.03502452e-01 6.53995752e-01
-5.59580147e-01 -1.21493578e+00 -8.93472791e-01 6.40708745e-01
-3.47435594e-01 2.78599616e-02 -3.47932763e-02 5.86026490e-01
6.86559677e-01 1.17094982e+00 -3.57003324e-02 -1.23138480e-01
1.84767827e-01 5.36021948e-01 3.35398585e-01 -1.05306506e+00
-2.30525494e-01 2.16902062e-01 -1.90348357e-01 -1.84310436e-01
-5.12168765e-01 -7.49494672e-01 -1.50712299e+00 2.72984892e-01
-6.09244525e-01 3.40914160e-01 2.75932997e-01 1.27252841e+00
-2.16321081e-01 3.16250622e-01 5.45292258e-01 -6.43282533e-01
-6.25770807e-01 -9.31822121e-01 -9.31469381e-01 3.68076563e-01
8.43804628e-02 -8.71944726e-01 -7.73625314e-01 1.18065238e-01] | [9.461615562438965, 4.050572395324707] |
db6fe230-b20f-4ee6-a190-63bdc33c6523 | long-form-video-language-pre-training-with | 2210.06031 | null | https://arxiv.org/abs/2210.06031v2 | https://arxiv.org/pdf/2210.06031v2.pdf | Long-Form Video-Language Pre-Training with Multimodal Temporal Contrastive Learning | Large-scale video-language pre-training has shown significant improvement in video-language understanding tasks. Previous studies of video-language pretraining mainly focus on short-form videos (i.e., within 30 seconds) and sentences, leaving long-form video-language pre-training rarely explored. Directly learning representation from long-form videos and language may benefit many long-form video-language understanding tasks. However, it is challenging due to the difficulty of modeling long-range relationships and the heavy computational burden caused by more frames. In this paper, we introduce a Long-Form VIdeo-LAnguage pre-training model (LF-VILA) and train it on a large-scale long-form video and paragraph dataset constructed from an existing public dataset. To effectively capture the rich temporal dynamics and to better align video and language in an efficient end-to-end manner, we introduce two novel designs in our LF-VILA model. We first propose a Multimodal Temporal Contrastive (MTC) loss to learn the temporal relation across different modalities by encouraging fine-grained alignment between long-form videos and paragraphs. Second, we propose a Hierarchical Temporal Window Attention (HTWA) mechanism to effectively capture long-range dependency while reducing computational cost in Transformer. We fine-tune the pre-trained LF-VILA model on seven downstream long-form video-language understanding tasks of paragraph-to-video retrieval and long-form video question-answering, and achieve new state-of-the-art performances. Specifically, our model achieves 16.1% relative improvement on ActivityNet paragraph-to-video retrieval task and 2.4% on How2QA task, respectively. We release our code, dataset, and pre-trained models at https://github.com/microsoft/XPretrain. | ['Jianlong Fu', 'Huan Yang', 'Bei Liu', 'Ruihua Song', 'Hongwei Xue', 'Yuchong Sun'] | 2022-10-12 | null | null | null | null | ['video-question-answering'] | ['computer-vision'] | [ 8.92177001e-02 -5.00693798e-01 -5.58105052e-01 -5.27772844e-01
-1.26748383e+00 -6.05897069e-01 5.54596484e-01 -2.78623939e-01
-6.06664419e-01 3.93176675e-01 5.58831632e-01 -4.13635343e-01
2.41415367e-01 -2.65400767e-01 -1.20187366e+00 -2.77775764e-01
4.54740822e-02 3.21469605e-01 2.71782628e-03 6.36971816e-02
-3.26657556e-02 9.38433781e-03 -1.09332585e+00 8.95751536e-01
6.25312209e-01 7.76107609e-01 6.17901564e-01 1.01117146e+00
-1.77643746e-01 1.00403559e+00 -2.54802287e-01 -2.99455941e-01
8.83972794e-02 -5.35203218e-01 -9.04896259e-01 5.00229262e-02
8.52948725e-01 -8.09280634e-01 -8.05288196e-01 6.59702897e-01
3.68547022e-01 1.73926383e-01 3.46153349e-01 -1.14490020e+00
-7.33202219e-01 5.35512745e-01 -6.82598114e-01 5.05655825e-01
6.32974744e-01 4.52273965e-01 1.05200267e+00 -1.04174542e+00
7.39484072e-01 1.52586973e+00 4.52756584e-01 7.53527105e-01
-8.58784318e-01 -9.28987741e-01 5.32769322e-01 4.97691333e-01
-1.28996956e+00 -6.86164021e-01 4.96921331e-01 -4.01867777e-01
1.29555893e+00 4.00985368e-02 5.63059092e-01 1.48496008e+00
7.82557297e-03 1.24937618e+00 6.30238175e-01 -2.44074896e-01
-4.87950683e-01 -3.85198444e-01 3.02344505e-02 7.40698695e-01
-2.95158684e-01 -1.21078268e-01 -7.15436041e-01 3.29469740e-01
7.60750532e-01 2.33924285e-01 -3.35290194e-01 -1.58213153e-02
-1.45060122e+00 6.91245139e-01 2.29222447e-01 3.80485952e-01
-2.10328773e-01 6.12590730e-01 7.22509384e-01 4.35222954e-01
4.51091886e-01 1.70936614e-01 -5.21362960e-01 -5.75222492e-01
-9.80685055e-01 -8.18920955e-02 4.76156980e-01 8.96382511e-01
6.40797675e-01 -1.56014040e-02 -3.57958078e-01 9.06103790e-01
3.95007908e-01 7.51091719e-01 4.44333553e-01 -9.18365657e-01
9.06906486e-01 9.33787301e-02 -3.37548941e-01 -7.27850318e-01
5.08715622e-02 -1.29577545e-02 -6.88501179e-01 -6.96250916e-01
2.89144367e-01 -2.23512799e-02 -1.18029284e+00 1.92924273e+00
-1.12859093e-01 7.23127306e-01 1.01427205e-01 9.24542964e-01
9.21420693e-01 1.32225418e+00 2.20388070e-01 -4.14505899e-01
1.32769239e+00 -1.55401576e+00 -8.35319459e-01 -3.13558161e-01
8.16340387e-01 -8.41316938e-01 1.50387192e+00 2.02146359e-02
-1.16223872e+00 -8.02500844e-01 -6.02247894e-01 -5.70389688e-01
-6.27911687e-02 7.27422014e-02 3.55163336e-01 -1.00845799e-01
-9.77402210e-01 1.76703811e-01 -1.07072532e+00 -4.51528698e-01
3.46315295e-01 2.80211568e-01 -4.54094797e-01 -5.84125161e-01
-1.30403471e+00 3.74327272e-01 2.62359560e-01 9.55049545e-02
-1.12883401e+00 -1.00912046e+00 -1.09418023e+00 2.54685041e-02
5.67818940e-01 -9.31790352e-01 1.36060238e+00 -1.37241745e+00
-1.43217027e+00 9.35827136e-01 -6.65504515e-01 -5.54555893e-01
4.19139683e-01 -6.53657019e-01 -3.05138260e-01 6.10451102e-01
1.43412411e-01 1.16753137e+00 9.29742157e-01 -8.62139463e-01
-4.57732290e-01 -8.56308267e-03 3.61428976e-01 3.86365920e-01
-4.45265830e-01 9.21657830e-02 -1.28940737e+00 -9.02846634e-01
-4.91820335e-01 -8.88145685e-01 1.59395069e-01 -8.80495608e-02
6.94137812e-02 -2.48499095e-01 1.03150797e+00 -9.95478451e-01
1.43122947e+00 -2.12777996e+00 2.34987900e-01 -3.84927452e-01
6.02949373e-02 4.29104149e-01 -8.33205342e-01 3.39235544e-01
-1.75754339e-01 2.53040910e-01 2.19544792e-03 -5.61858952e-01
-1.95286542e-01 3.48759800e-01 -4.17727262e-01 1.97060227e-01
2.09470898e-01 1.22024882e+00 -9.18206751e-01 -4.80008245e-01
2.24731892e-01 5.49270988e-01 -8.42933416e-01 5.86004853e-01
-4.79795843e-01 4.36476767e-01 -5.08969128e-01 6.39419556e-01
3.46199036e-01 -5.17851651e-01 -4.61948030e-02 -5.75563252e-01
1.81918561e-01 2.80535460e-01 -4.04093742e-01 2.33134842e+00
-9.28104997e-01 9.47497308e-01 -1.46429567e-02 -9.73193586e-01
2.40705207e-01 3.89508069e-01 6.24042630e-01 -1.07767475e+00
-1.47186801e-01 8.53336528e-02 -1.88935071e-01 -8.52264583e-01
3.29596996e-01 6.39902875e-02 -9.10430104e-02 3.08552474e-01
2.54051864e-01 6.30170256e-02 4.78719562e-01 5.74256837e-01
8.98253381e-01 3.55407894e-01 7.82308578e-02 1.18736885e-01
7.30503559e-01 -2.44994134e-01 2.35242799e-01 5.14711916e-01
-1.47903070e-01 7.52427816e-01 3.45380694e-01 -2.60445178e-01
-1.04718137e+00 -8.57708693e-01 3.53346944e-01 1.42280626e+00
1.44953385e-01 -7.03524172e-01 -4.75813180e-01 -7.64168441e-01
-1.67404622e-01 4.72264200e-01 -3.63240719e-01 -1.01426579e-01
-9.75813985e-01 -1.44764856e-01 5.77600062e-01 5.26179373e-01
5.17342627e-01 -1.01566744e+00 2.07880102e-02 7.73615092e-02
-7.91350484e-01 -1.87331367e+00 -1.35785639e+00 -5.00543773e-01
-7.00983286e-01 -9.26500559e-01 -7.93139696e-01 -9.22123075e-01
4.66321528e-01 5.89654624e-01 1.31678307e+00 2.31746018e-01
3.14770937e-02 9.06060398e-01 -5.07163703e-01 1.11101590e-01
-1.81981593e-01 6.34855106e-02 -1.53365061e-01 8.03203732e-02
3.54390830e-01 -4.02057409e-01 -7.14203477e-01 3.37681711e-01
-9.43379581e-01 3.07673842e-01 6.95057392e-01 1.02610636e+00
7.10627556e-01 -6.26570582e-01 4.32816565e-01 -4.67960238e-01
3.46160561e-01 -5.69322467e-01 -4.57021475e-01 4.26735789e-01
-1.07179374e-01 -1.09681204e-01 6.97099984e-01 -6.86174512e-01
-9.55448270e-01 -1.21161029e-01 -3.21502864e-01 -1.16969657e+00
8.54313448e-02 6.16947472e-01 -1.26912326e-01 2.78269380e-01
-6.51075765e-02 2.87682235e-01 -3.77781354e-02 -2.65657932e-01
4.81409490e-01 5.22873282e-01 6.80919468e-01 -6.60110652e-01
6.00743949e-01 4.18661863e-01 -4.57814366e-01 -9.63233054e-01
-1.02658510e+00 -9.46056902e-01 -4.00757521e-01 -1.91385865e-01
1.31303680e+00 -1.51744735e+00 -6.21673346e-01 2.68914491e-01
-1.29382169e+00 -7.92088151e-01 6.06086850e-02 6.50765061e-01
-5.85193098e-01 6.29735649e-01 -8.39611173e-01 -3.20414513e-01
-4.95763481e-01 -1.20009899e+00 1.23001444e+00 3.92019190e-02
1.04943611e-01 -9.33808446e-01 -2.93013873e-03 9.21029866e-01
2.08602890e-01 -3.16827953e-01 4.87712204e-01 -3.59472126e-01
-9.43965614e-01 2.01897696e-01 -4.21881229e-01 3.95642996e-01
-1.22832268e-01 -1.19178548e-01 -5.80757797e-01 -6.91464901e-01
-2.81322479e-01 -6.46495044e-01 1.15684438e+00 4.39810634e-01
1.46507478e+00 -3.05480599e-01 -1.26714319e-01 9.15644228e-01
1.14318633e+00 1.58573702e-01 6.38052225e-01 1.32451579e-01
1.07218552e+00 2.24901587e-01 8.61810684e-01 1.11412592e-01
5.50019622e-01 8.16071689e-01 1.21529244e-01 -1.21195681e-01
-4.29881096e-01 -5.85125804e-01 9.66086030e-01 1.45757151e+00
1.17719263e-01 -5.70083976e-01 -7.82020926e-01 7.28850305e-01
-1.98670220e+00 -1.21347213e+00 2.77989298e-01 1.80277717e+00
7.89631724e-01 6.16691168e-03 8.28009695e-02 -5.79417646e-01
4.83842373e-01 4.35795933e-01 -4.79102343e-01 -9.10712034e-02
-1.38863415e-01 -8.88153017e-02 1.70045480e-01 7.86056936e-01
-1.11497617e+00 1.19314098e+00 4.72547531e+00 1.07378089e+00
-1.43541288e+00 2.58146107e-01 7.75865793e-01 -4.17965889e-01
-3.84902596e-01 -6.64561614e-02 -7.71034479e-01 4.71136570e-01
1.18606234e+00 -1.23846769e-01 4.56487298e-01 4.17655438e-01
6.81882322e-01 1.29417792e-01 -1.26765847e+00 1.39792597e+00
3.84879529e-01 -1.53652883e+00 3.84915709e-01 -2.66658902e-01
7.52629459e-01 3.45969498e-01 -9.40456837e-02 7.08652377e-01
-3.24096471e-01 -1.05306315e+00 7.13777304e-01 2.90609121e-01
1.27594376e+00 -3.90599072e-01 5.86196303e-01 1.19013898e-01
-1.59942067e+00 3.83049389e-03 -1.03336364e-01 1.09493211e-01
5.68608701e-01 1.84568226e-01 -4.58796591e-01 5.41758955e-01
7.20617056e-01 1.25625098e+00 -4.01229650e-01 6.29674137e-01
2.54412461e-02 8.75415266e-01 -3.94104660e-01 3.12539458e-01
6.66386843e-01 -1.19674392e-01 4.49480116e-01 1.60718548e+00
3.02423775e-01 2.33286336e-01 4.43150342e-01 3.58154207e-01
-5.35954356e-01 1.38754612e-02 -5.63138485e-01 -5.08606970e-01
2.35667095e-01 9.37319994e-01 -3.02951127e-01 -5.90122163e-01
-9.73115325e-01 1.13228750e+00 2.00337037e-01 6.25574589e-01
-1.17005432e+00 9.49235931e-02 7.37719297e-01 1.98914349e-01
4.77684915e-01 -4.25989628e-01 5.02873600e-01 -1.69542837e+00
2.12369919e-01 -1.15933955e+00 6.39982879e-01 -9.81176972e-01
-1.22667801e+00 4.95754063e-01 2.11158216e-01 -1.03974354e+00
-4.08448905e-01 -4.07148600e-01 -3.56299281e-01 4.56067413e-01
-1.68234086e+00 -1.55424094e+00 -3.53410691e-01 9.41091061e-01
1.42965949e+00 -1.05792053e-01 3.36740255e-01 7.77873039e-01
-5.32656610e-01 7.40876138e-01 5.32801040e-02 2.86846936e-01
1.07390404e+00 -6.97566509e-01 2.23064333e-01 1.03246439e+00
3.44812810e-01 5.50189197e-01 2.63434321e-01 -4.67365056e-01
-1.79467511e+00 -1.24827707e+00 8.26666772e-01 -3.68831009e-01
8.75219345e-01 -5.37176073e-01 -9.66018438e-01 9.44157124e-01
5.93821466e-01 1.53765902e-01 3.87109905e-01 -1.85437396e-01
-5.59716165e-01 -2.40272895e-01 -4.05380815e-01 6.54101968e-01
1.14084506e+00 -1.12160957e+00 -4.97225434e-01 5.00656486e-01
1.18026388e+00 -3.91175807e-01 -7.68252134e-01 4.43005919e-01
6.34828568e-01 -5.96107423e-01 1.15152574e+00 -6.82824552e-01
7.39300966e-01 -1.55301332e-01 -2.31367782e-01 -7.74916947e-01
-5.88370003e-02 -9.81254578e-01 -3.13187271e-01 1.10902023e+00
3.30992669e-01 -1.40238956e-01 6.99444294e-01 5.50160781e-02
-2.28878766e-01 -8.86864424e-01 -7.58057952e-01 -6.67985737e-01
-1.93875730e-02 -6.35731339e-01 -6.93155304e-02 9.01362658e-01
-2.88983881e-01 6.92670584e-01 -7.73531675e-01 4.61584739e-02
2.12799236e-01 1.58997655e-01 8.83179307e-01 -2.49792576e-01
-4.62099791e-01 -3.46097052e-01 3.88598740e-02 -1.93535018e+00
5.53151250e-01 -8.25009227e-01 1.39477216e-02 -1.39993751e+00
5.88260233e-01 1.55139998e-01 -2.89750785e-01 3.85459721e-01
-3.07265013e-01 2.84677535e-01 5.30485213e-01 2.98254430e-01
-1.17862201e+00 6.26294672e-01 1.36759436e+00 -2.95920461e-01
-4.05431949e-02 -4.25198108e-01 -8.78479779e-02 5.54388285e-01
4.14740533e-01 -2.73148268e-01 -6.22136056e-01 -1.04321074e+00
1.98821835e-02 3.79135668e-01 3.56081963e-01 -5.86484790e-01
9.62729529e-02 -1.31660402e-01 8.12311769e-02 -8.45246673e-01
5.53612888e-01 -6.51850641e-01 -1.26533240e-01 4.01581347e-01
-4.49219376e-01 4.64409769e-01 4.15702611e-01 7.13075697e-01
-7.11579859e-01 1.78979650e-01 6.45331204e-01 -6.41982853e-02
-1.08064497e+00 6.19511664e-01 -2.05339119e-01 3.76243025e-01
8.52599919e-01 2.03822795e-02 -1.78119406e-01 -8.68487298e-01
-6.08211696e-01 6.85539186e-01 2.68420964e-01 7.87363350e-01
7.22748160e-01 -1.26309133e+00 -7.74984241e-01 1.20615009e-02
8.71024355e-02 -1.23466246e-01 6.86907828e-01 1.02091420e+00
-5.30632377e-01 7.88166881e-01 4.17522937e-02 -9.36997831e-01
-1.58396387e+00 6.17908955e-01 1.98468283e-01 -5.63944280e-01
-6.15655005e-01 9.70469654e-01 8.69336665e-01 -1.32290006e-01
4.14426953e-01 -4.61752772e-01 7.02596307e-02 -9.71320346e-02
5.35727799e-01 -1.82651088e-01 -3.89145732e-01 -7.02629089e-01
-3.19406301e-01 9.64652300e-01 -3.16012919e-01 3.63578234e-04
1.14032900e+00 -4.88118321e-01 1.32932946e-01 3.14196438e-01
1.77956629e+00 -5.23649454e-02 -1.41207135e+00 -2.81851381e-01
-3.73143852e-01 -4.19402242e-01 -1.49886549e-01 -6.32935286e-01
-1.19790030e+00 1.08937740e+00 3.79152805e-01 -5.31538546e-01
1.35568309e+00 1.68918222e-01 1.30234051e+00 5.23429930e-01
-1.59702078e-01 -8.34237039e-01 7.35978484e-01 7.79006124e-01
1.06751537e+00 -1.45223999e+00 -6.30937889e-02 -2.34126717e-01
-5.72812796e-01 1.15871596e+00 7.61289716e-01 7.51210824e-02
4.25781578e-01 -2.79439669e-02 1.39573887e-01 9.76376906e-02
-1.22494054e+00 -1.27151549e-01 5.47225356e-01 2.34279469e-01
4.85304624e-01 -3.18217546e-01 -6.98343739e-02 4.52976465e-01
2.66316831e-01 1.66995749e-01 2.27776393e-01 6.55761003e-01
8.01076442e-02 -9.46910560e-01 -1.17619969e-01 2.12173879e-01
-6.57024205e-01 -3.58363897e-01 6.38968199e-02 8.62654269e-01
-1.13747589e-01 7.13173270e-01 2.60232449e-01 -1.96261168e-01
1.52655125e-01 -1.19871214e-01 4.92485464e-01 -3.42810243e-01
-3.56121361e-01 6.39254034e-01 2.80241042e-01 -1.01729119e+00
-6.90616310e-01 -4.62314576e-01 -1.16755819e+00 -3.00721496e-01
1.42903486e-02 7.96845406e-02 3.35579097e-01 1.02009940e+00
5.79489470e-01 5.15732944e-01 4.01823729e-01 -8.63359332e-01
-2.06715375e-01 -8.91459882e-01 2.08677769e-01 7.59060264e-01
5.18220127e-01 -3.13872606e-01 -2.06000015e-01 4.92188156e-01] | [10.245963096618652, 0.8877231478691101] |
20651a5d-a8c1-4c15-85f9-403447c9048a | self-supervision-with-superpixels-training | 2007.09886 | null | https://arxiv.org/abs/2007.09886v2 | https://arxiv.org/pdf/2007.09886v2.pdf | Self-Supervision with Superpixels: Training Few-shot Medical Image Segmentation without Annotation | Few-shot semantic segmentation (FSS) has great potential for medical imaging applications. Most of the existing FSS techniques require abundant annotated semantic classes for training. However, these methods may not be applicable for medical images due to the lack of annotations. To address this problem we make several contributions: (1) A novel self-supervised FSS framework for medical images in order to eliminate the requirement for annotations during training. Additionally, superpixel-based pseudo-labels are generated to provide supervision; (2) An adaptive local prototype pooling module plugged into prototypical networks, to solve the common challenging foreground-background imbalance problem in medical image segmentation; (3) We demonstrate the general applicability of the proposed approach for medical images using three different tasks: abdominal organ segmentation for CT and MRI, as well as cardiac segmentation for MRI. Our results show that, for medical image segmentation, the proposed method outperforms conventional FSS methods which require manual annotations for training. | ['Turkay Kart', 'Cheng Ouyang', 'Daniel Rueckert', 'Chen Chen', 'Carlo Biffi', 'Huaqi Qiu'] | 2020-07-20 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/6977_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123740749.pdf | eccv-2020-8 | ['cardiac-segmentation'] | ['medical'] | [ 6.29447520e-01 4.11464661e-01 -1.92729324e-01 -6.12019002e-01
-7.16559470e-01 -2.44680464e-01 3.16029608e-01 4.19475824e-01
-5.23594856e-01 6.04015648e-01 -2.37707943e-01 -1.93399534e-01
3.51915602e-03 -5.25316238e-01 -5.84028840e-01 -7.62840331e-01
1.12003513e-01 4.08487618e-01 8.15315485e-01 4.73303795e-02
1.55193061e-01 4.67144847e-01 -1.39960814e+00 3.34137589e-01
1.19884658e+00 1.06709921e+00 4.00567114e-01 3.82519215e-01
-4.16205913e-01 7.70958185e-01 -5.44591963e-01 2.34507889e-01
2.85045862e-01 -6.21091962e-01 -1.01674867e+00 5.22366643e-01
3.80730294e-02 -1.84512556e-01 3.41961980e-01 1.28346789e+00
3.97107631e-01 4.59146678e-01 6.14620090e-01 -9.38213289e-01
-1.64196387e-01 5.34286916e-01 -5.91372371e-01 3.45873773e-01
-1.45858988e-01 -6.88171983e-02 6.16821349e-01 -5.55806994e-01
6.97041750e-01 9.13796782e-01 4.66228276e-01 6.60916030e-01
-1.04611313e+00 -3.47321033e-01 1.57989681e-01 5.73236542e-03
-1.20601428e+00 -2.35847160e-01 6.14935935e-01 -4.20861155e-01
5.58007479e-01 2.54866391e-01 6.57395959e-01 6.92411423e-01
-1.72854941e-02 1.02508998e+00 1.26512241e+00 -5.48806071e-01
6.24503255e-01 3.99524420e-01 5.16517282e-01 7.50607133e-01
1.58114016e-01 -4.17607129e-01 -1.16419904e-01 -2.56056990e-02
6.77819788e-01 -1.90202408e-02 -1.41377524e-01 -4.05288845e-01
-1.06889403e+00 7.72854388e-01 4.88713890e-01 8.59389544e-01
-3.25282097e-01 3.03977425e-03 4.97363687e-01 -2.34328896e-01
6.24246180e-01 4.54509944e-01 -3.35412174e-01 3.30853730e-01
-1.31137443e+00 -1.10798314e-01 6.55989885e-01 7.17856824e-01
6.43383324e-01 -1.53262615e-02 -4.09480542e-01 8.54389489e-01
1.91140220e-01 -1.43815875e-01 8.33557606e-01 -6.91189408e-01
1.47156447e-01 6.36336088e-01 2.95806322e-02 -6.82120860e-01
-7.07477868e-01 -3.86032224e-01 -6.79349303e-01 1.14630610e-01
3.73786837e-01 -1.70114599e-02 -1.41576111e+00 1.40559304e+00
5.72131217e-01 3.16542238e-01 9.42518860e-02 9.86851692e-01
1.22197115e+00 5.23909211e-01 4.03861910e-01 -3.04343909e-01
1.32263422e+00 -1.35256422e+00 -8.92927527e-01 -1.78750426e-01
5.53800642e-01 -4.62815344e-01 1.10047579e+00 6.02347776e-02
-1.02203858e+00 -5.10163009e-01 -1.14759886e+00 1.84719712e-01
-3.73591244e-01 2.82085091e-01 5.36358774e-01 8.33469033e-01
-1.00692213e+00 5.08289993e-01 -1.33193338e+00 -4.71264869e-01
6.67303681e-01 4.81173635e-01 -6.45665601e-02 8.20301101e-02
-9.97232020e-01 7.45928884e-01 7.86293328e-01 1.37075901e-01
-7.46372581e-01 -6.60384178e-01 -1.00479460e+00 1.51168630e-01
6.51718915e-01 -5.93172312e-01 1.12596631e+00 -1.48954976e+00
-1.55164647e+00 1.13846457e+00 6.22258484e-02 -4.37351018e-01
5.04498720e-01 1.33935750e-01 7.39511028e-02 5.02645254e-01
1.69797406e-01 7.40502894e-01 7.06145525e-01 -1.24110496e+00
-4.04083967e-01 -3.20890486e-01 2.67697629e-02 3.73456120e-01
-2.43684664e-01 7.20375180e-02 -6.06607616e-01 -8.97608936e-01
1.71141535e-01 -7.70266533e-01 -7.57490635e-01 -6.16759658e-02
-5.56985676e-01 -5.90995997e-02 6.41399980e-01 -6.71131551e-01
1.04247308e+00 -2.29935837e+00 7.62426704e-02 2.73527682e-01
-6.28869757e-02 4.43873286e-01 1.29106387e-01 -2.28584528e-01
3.69158834e-02 -3.19070220e-02 -7.49342382e-01 -2.99672067e-01
-2.87285447e-01 3.39905232e-01 3.12342465e-01 2.69007862e-01
1.34754792e-01 8.29038501e-01 -8.36923420e-01 -9.94191587e-01
5.00912011e-01 2.32063130e-01 -2.62978941e-01 1.94086835e-01
-2.75967151e-01 8.81269157e-01 -5.65160692e-01 6.35339618e-01
5.46039164e-01 -4.96405840e-01 1.78138927e-01 -4.75248247e-02
5.14401458e-02 -1.70747995e-01 -1.10538113e+00 2.00074291e+00
-1.32813215e-01 -8.28981772e-02 2.54138112e-01 -1.37523091e+00
5.16650021e-01 4.63580817e-01 9.03880239e-01 -4.00529593e-01
3.66437465e-01 3.07991773e-01 -1.13735266e-01 -6.83029473e-01
1.41484946e-01 -3.77765775e-01 2.40489878e-02 1.87730968e-01
3.16418022e-01 -1.61131576e-01 4.17075992e-01 9.73927826e-02
8.53505969e-01 2.44894013e-01 2.97737092e-01 -6.05101109e-01
8.65947425e-01 2.56073117e-01 7.16033638e-01 6.36484921e-01
-4.69725221e-01 7.26005733e-01 3.21204066e-01 -4.85561907e-01
-6.27583623e-01 -8.02766323e-01 -1.51227206e-01 1.06858981e+00
4.34627444e-01 -1.08506801e-02 -1.27597964e+00 -8.60484183e-01
-4.06171411e-01 5.27800500e-01 -5.27967513e-01 2.35639408e-01
-4.60992396e-01 -1.25513101e+00 2.35746324e-01 5.83112895e-01
4.37112629e-01 -1.21446884e+00 -1.07202840e+00 1.95196763e-01
-3.49828988e-01 -1.29228616e+00 -3.47830206e-01 2.44902536e-01
-1.12442541e+00 -1.24221551e+00 -1.16204953e+00 -9.31772470e-01
1.12781858e+00 9.06722099e-02 7.28583515e-01 1.22893773e-01
-8.33147526e-01 3.77158016e-01 -3.39717895e-01 -3.82151574e-01
-5.29849768e-01 1.68297112e-01 -5.47610283e-01 -8.52467120e-03
2.33437624e-02 -2.48591557e-01 -7.36419499e-01 1.18552215e-01
-1.10258687e+00 2.81257480e-01 4.72247422e-01 8.95055056e-01
7.71343887e-01 -1.61654666e-01 5.66168666e-01 -1.48313057e+00
2.58289516e-01 -2.65014261e-01 -5.82512081e-01 4.83735859e-01
-5.28864741e-01 -1.98780179e-01 3.97623390e-01 -1.38408631e-01
-1.39515996e+00 3.20002794e-01 -2.48953626e-01 -1.72793314e-01
-4.47438180e-01 3.29466134e-01 2.96445817e-01 -1.71336859e-01
7.42127240e-01 -3.49154370e-03 1.01233907e-01 -3.73434275e-01
7.34355301e-02 4.34082031e-01 5.08651674e-01 -3.32051039e-01
1.16279781e-01 5.74664533e-01 -3.98716740e-02 -7.56518364e-01
-9.06286359e-01 -7.83682466e-01 -8.75032902e-01 -1.36621639e-01
1.34277380e+00 -7.55582511e-01 -7.63452351e-02 3.59802246e-01
-9.21332777e-01 -4.65284050e-01 -4.94831830e-01 5.95671654e-01
-6.34471476e-01 6.08388126e-01 -9.55942035e-01 -6.84886813e-01
-5.52870810e-01 -1.39400494e+00 9.51194227e-01 5.05492210e-01
-1.57780130e-03 -1.04370308e+00 -2.89680153e-01 6.61680579e-01
2.91866899e-01 5.93584597e-01 9.24833655e-01 -8.41269493e-01
-4.53347713e-01 4.04947102e-02 -3.25321585e-01 5.14162540e-01
3.29922080e-01 -3.47686738e-01 -8.96149933e-01 -2.17409000e-01
5.02704829e-02 -4.01742369e-01 8.54519844e-01 8.03282976e-01
1.39804673e+00 1.67254105e-01 -4.07810479e-01 5.27855814e-01
1.41639030e+00 4.81756777e-01 2.68432319e-01 1.92864954e-01
6.40329838e-01 7.39619136e-01 7.23997474e-01 2.65485972e-01
1.01199098e-01 4.48879272e-01 2.75729984e-01 -7.42318392e-01
-2.65862703e-01 3.51028085e-01 -8.60415474e-02 7.49460757e-01
-3.00907139e-02 -1.36956170e-01 -8.74624491e-01 6.56470060e-01
-1.98096585e+00 -3.70442450e-01 -8.47342983e-02 2.04318810e+00
7.90351152e-01 -6.17639069e-03 1.14688449e-01 -1.79370232e-02
8.48237872e-01 -1.04257025e-01 -5.48399985e-01 -1.65592935e-02
2.44245902e-01 4.62773442e-01 4.50550199e-01 3.49737614e-01
-1.45192146e+00 8.95563960e-01 6.13038254e+00 8.83664727e-01
-1.03303897e+00 5.75857222e-01 8.71771336e-01 2.31309831e-01
8.53308663e-02 -2.45949611e-01 -4.06713068e-01 4.16671127e-01
4.27256525e-01 2.00957924e-01 1.84157956e-02 8.23297858e-01
6.91020489e-02 -4.35888439e-01 -7.91200638e-01 7.40320265e-01
2.05950528e-01 -1.13384902e+00 -1.91701218e-01 -5.75413764e-01
9.04116929e-01 -1.73575059e-01 -2.66480267e-01 -7.82157332e-02
2.99555548e-02 -8.00923288e-01 4.32110906e-01 2.01156542e-01
6.20227575e-01 -4.19207215e-01 9.49052155e-01 3.50491703e-01
-8.98536742e-01 8.12061876e-02 -1.94029272e-01 4.70474839e-01
2.04783469e-01 4.90751088e-01 -9.56055939e-01 7.83760130e-01
5.96778691e-01 4.68400806e-01 -5.18588066e-01 1.15723991e+00
-1.89358264e-01 5.44721067e-01 -2.50925452e-01 2.16530874e-01
5.75162590e-01 -9.28716436e-02 2.49719575e-01 1.33530927e+00
1.28925815e-01 8.17190558e-02 5.63836396e-01 8.38297427e-01
2.60206014e-01 4.59485471e-01 -9.72999483e-02 2.77521126e-02
-9.19398293e-02 1.33677030e+00 -1.82087815e+00 -6.92853808e-01
-2.93129921e-01 1.13851917e+00 -2.29444370e-01 3.26908410e-01
-7.09815919e-01 -2.84564495e-01 -2.26091325e-01 7.66114369e-02
1.03985935e-01 2.91530102e-01 -3.62265438e-01 -1.05451047e+00
-2.77039021e-01 -3.65556389e-01 7.46752739e-01 -5.74811280e-01
-9.84349668e-01 5.69309771e-01 2.42881104e-01 -9.39777076e-01
-1.21785708e-01 -4.62344527e-01 -4.63470429e-01 4.89529550e-01
-1.66294467e+00 -1.18783450e+00 -4.92333591e-01 5.13053596e-01
7.73905396e-01 2.88719069e-02 7.91695356e-01 5.10019839e-01
-5.29161811e-01 2.63693064e-01 -5.64107001e-02 -1.22543909e-02
6.00503802e-01 -1.34279859e+00 -4.18101391e-03 8.47782552e-01
-1.53263569e-01 2.93508232e-01 4.84153032e-01 -5.94355464e-01
-6.17773294e-01 -1.12539518e+00 4.26547647e-01 2.88422376e-01
2.07800210e-01 -1.51825398e-02 -9.21403050e-01 5.13044357e-01
3.49526852e-02 4.63656306e-01 7.30790436e-01 -3.32739353e-01
3.07347804e-01 4.39390875e-02 -1.54223454e+00 2.51517564e-01
7.17530191e-01 3.49513218e-02 -5.63288212e-01 6.76464915e-01
6.10471010e-01 -6.85702145e-01 -7.25812137e-01 5.28888106e-01
2.69712359e-01 -9.52147365e-01 9.03587103e-01 -2.36766219e-01
2.99485385e-01 -3.68565679e-01 2.90412217e-01 -8.98087323e-01
5.16160578e-02 -2.51566499e-01 3.15686285e-01 8.90575707e-01
2.50433058e-01 -4.80775356e-01 8.80278826e-01 5.85637152e-01
-4.76509809e-01 -7.61373878e-01 -9.58970070e-01 -4.87001151e-01
-2.47571126e-01 -1.06074601e-01 1.14333495e-01 1.09006178e+00
-2.22391799e-01 -3.60810533e-02 -6.57859072e-02 4.88854535e-02
5.40159047e-01 9.10090283e-02 2.68275142e-01 -1.12592638e+00
-3.79945844e-01 -3.72947186e-01 -3.65835398e-01 -5.87256074e-01
3.16432491e-02 -1.00074339e+00 4.08039212e-01 -1.68862092e+00
4.31801379e-01 -6.47781134e-01 -5.23084402e-01 5.95485568e-01
-3.21515113e-01 6.65920675e-01 1.15461141e-01 9.00213718e-02
-8.04197013e-01 1.96594656e-01 1.35174131e+00 -1.34787783e-01
-2.50009984e-01 1.68829083e-01 -3.89466465e-01 8.90876710e-01
6.98830962e-01 -3.84719610e-01 -6.44010007e-01 -5.08569926e-02
-3.65914494e-01 1.51167572e-01 2.19793141e-01 -1.01104927e+00
8.91624019e-02 -1.21798359e-01 2.79882878e-01 -3.50168258e-01
-1.87957305e-02 -7.14524746e-01 -5.38114943e-02 7.57213175e-01
-1.92818046e-01 -4.18414295e-01 1.38781682e-01 4.98133630e-01
-3.56957436e-01 -7.31974483e-01 1.21473193e+00 -5.99121749e-01
-8.34993839e-01 1.30405366e-01 -4.25437570e-01 -2.04082355e-02
1.45044160e+00 -3.69170874e-01 1.51835531e-01 9.44218859e-02
-1.11630952e+00 3.36832702e-01 2.56400615e-01 1.78869933e-01
3.36339235e-01 -8.26601028e-01 -2.22046703e-01 1.17016762e-01
1.53313428e-02 5.72394371e-01 6.01772606e-01 1.07377756e+00
-9.13118660e-01 2.11664304e-01 -2.99741387e-01 -7.82006204e-01
-1.26857722e+00 5.77811420e-01 3.09935033e-01 -4.80475008e-01
-8.35952282e-01 8.54419351e-01 5.33342957e-01 -3.59531432e-01
3.46887887e-01 -6.27945900e-01 -3.42108935e-01 -3.41358371e-02
3.82318228e-01 1.60860658e-01 1.71379566e-01 -4.91188943e-01
-3.21991593e-01 4.29001004e-01 1.96300577e-02 -4.50087078e-02
1.24440277e+00 -2.11769491e-01 -1.16241790e-01 5.79147041e-01
7.08544075e-01 -5.07272542e-01 -1.00472486e+00 -1.52933925e-01
2.39317670e-01 -1.58517271e-01 4.48106974e-03 -8.95420015e-01
-1.09285498e+00 9.34198976e-01 8.61863971e-01 1.29947007e-01
1.26942790e+00 -3.03236581e-02 8.13252270e-01 4.72538359e-02
2.72270977e-01 -1.38180196e+00 2.57586297e-02 1.07739586e-02
2.74501175e-01 -1.31258512e+00 -1.15334252e-02 -8.22953224e-01
-7.42051601e-01 1.09840238e+00 4.70300496e-01 -2.79620532e-02
5.09881020e-01 1.46042466e-01 1.64568663e-01 -1.93409011e-01
-8.75512958e-02 -2.51864165e-01 3.11283469e-01 2.87738204e-01
4.72638756e-01 -1.80270717e-01 -7.58037150e-01 6.36026204e-01
5.18940091e-01 2.34606519e-01 4.02448118e-01 1.16128719e+00
-4.30705786e-01 -1.01134384e+00 -3.56205493e-01 4.25996453e-01
-8.81561756e-01 1.20645389e-01 2.94052204e-03 5.52608311e-01
4.27728087e-01 6.29787862e-01 -3.37239821e-03 2.35070333e-01
2.18054187e-02 2.62443542e-01 5.66606581e-01 -1.04623461e+00
-7.59314477e-01 3.64253730e-01 -2.86169171e-01 -4.09212798e-01
-9.35537577e-01 -3.46080422e-01 -1.57407737e+00 5.87351322e-01
-4.62584078e-01 1.95295602e-01 7.09006369e-01 1.07891846e+00
-6.79728924e-04 9.34132218e-01 3.21789622e-01 -8.39789927e-01
-2.35712811e-01 -9.05744731e-01 -6.51313722e-01 6.27771258e-01
6.77358583e-02 -6.58341527e-01 -1.29798412e-01 3.72375757e-01] | [14.687374114990234, -2.1445515155792236] |
ad58b656-2cc4-4d5f-a98e-82cf9fa50c71 | many-hands-make-light-work-using-essay-traits | 2102.00781 | null | https://arxiv.org/abs/2102.00781v1 | https://arxiv.org/pdf/2102.00781v1.pdf | Many Hands Make Light Work: Using Essay Traits to Automatically Score Essays | Most research in the area of automatic essay grading (AEG) is geared towards scoring the essay holistically while there has also been some work done on scoring individual essay traits. In this paper, we describe a way to score essays holistically using a multi-task learning (MTL) approach, where scoring the essay holistically is the primary task, and scoring the essay traits is the auxiliary task. We compare our results with a single-task learning (STL) approach, using both LSTMs and BiLSTMs. We also compare our results of the auxiliary task with such tasks done in other AEG systems. To find out which traits work best for different types of essays, we conduct ablation tests for each of the essay traits. We also report the runtime and number of training parameters for each system. We find that MTL-based BiLSTM system gives the best results for scoring the essay holistically, as well as performing well on scoring the essay traits. | ['Pushpak Bhattacharyya', 'Sriparna Saha', 'Sandeep Mathias', 'Rahul Kumar'] | 2021-02-01 | null | https://aclanthology.org/2022.naacl-main.106 | https://aclanthology.org/2022.naacl-main.106.pdf | naacl-2022-7 | ['automated-essay-scoring'] | ['natural-language-processing'] | [-9.93793458e-02 -1.84536576e-01 7.36246035e-02 -5.35932660e-01
-1.20181060e+00 -6.15175724e-01 4.36543345e-01 1.27612323e-01
-4.81136531e-01 5.71441114e-01 3.04563582e-01 -3.47756475e-01
-5.67289479e-02 -4.99276787e-01 -5.23444377e-02 -6.27742589e-01
5.90876520e-01 4.57283378e-01 2.78073490e-01 -3.43968600e-01
7.82165468e-01 -1.13532908e-01 -1.12647080e+00 4.93412584e-01
8.80262196e-01 8.03175867e-01 -5.49097732e-02 1.19965100e+00
-2.59162277e-01 1.17638981e+00 -1.25077558e+00 -9.44661558e-01
-1.27593011e-01 -6.07606590e-01 -9.04039860e-01 -3.60396266e-01
1.22093189e+00 -1.44395083e-01 3.82740915e-01 8.60485315e-01
5.50507247e-01 2.05815032e-01 9.04059887e-01 -9.93771017e-01
-8.05549204e-01 5.81012964e-01 -6.19711995e-01 4.57338057e-02
2.85632908e-01 -3.41368109e-01 1.42893386e+00 -8.70474100e-01
-1.39685757e-02 9.82950032e-01 1.02874577e+00 5.60390949e-01
-8.45463574e-01 -4.36757654e-01 4.60630730e-02 2.43401021e-01
-6.39476478e-01 -1.39749691e-01 5.45829415e-01 -4.64302570e-01
9.65125561e-01 4.60015684e-02 3.06185812e-01 8.71063411e-01
5.96363127e-01 1.03477228e+00 1.44157803e+00 -6.17795706e-01
-1.07047811e-01 1.05096236e-01 8.41850221e-01 1.01803279e+00
-2.55069416e-02 -4.12420928e-01 -8.44377279e-01 -7.44793266e-02
2.79282570e-01 -2.26605177e-01 4.13195610e-01 2.83356577e-01
-1.05899549e+00 7.81211972e-01 -7.27369189e-02 4.36302572e-01
-8.89924318e-02 2.68501937e-01 6.17168665e-01 9.20053780e-01
7.66086221e-01 6.34974837e-01 -6.92810953e-01 -3.36456597e-01
-1.60495496e+00 3.63007814e-01 1.00280082e+00 4.03084427e-01
5.83362758e-01 2.56707489e-01 -9.23423409e-01 1.27610946e+00
-2.87533421e-02 1.05939828e-01 7.62199879e-01 -9.49580848e-01
4.15613145e-01 6.75294399e-01 -2.75750551e-02 -5.28149426e-01
-5.46308160e-01 -1.77189738e-01 -4.42538261e-01 4.45950389e-01
6.67321146e-01 -5.38184285e-01 -7.75695086e-01 1.60903215e+00
-3.61343145e-01 -4.61681873e-01 -1.94712564e-01 5.06457865e-01
1.08775079e+00 4.50995445e-01 1.92376539e-01 1.93503916e-01
1.58331013e+00 -1.43507552e+00 -8.47707152e-01 -3.81873176e-02
1.11021757e+00 -1.04060102e+00 1.45004690e+00 6.62925422e-01
-1.29567015e+00 -6.34649515e-01 -1.22060251e+00 -4.08715248e-01
-3.73773992e-01 7.54847229e-01 3.30799311e-01 8.99500191e-01
-1.46051192e+00 5.74842691e-01 -5.01845479e-01 -1.52508974e-01
1.33313596e-01 1.11297064e-01 6.48437291e-02 4.77354109e-01
-1.27940488e+00 1.69026458e+00 -1.59538746e-01 -5.29640436e-01
-6.80779397e-01 -5.92556596e-01 -5.75258434e-01 2.98524171e-01
-9.01880860e-02 -4.65256661e-01 1.87524140e+00 -6.55633390e-01
-1.78793752e+00 9.43292975e-01 -2.86780894e-01 -2.71760345e-01
2.09234953e-01 -2.55586773e-01 4.00146749e-03 -2.53805429e-01
1.23590477e-01 3.79302979e-01 5.98035753e-01 -7.42749929e-01
-8.01596582e-01 -3.50998312e-01 2.42320672e-01 3.68633837e-01
-1.11790180e+00 3.16637754e-01 1.82083115e-01 -5.67633986e-01
-4.53175187e-01 -7.53268600e-01 1.26676783e-01 -3.51843446e-01
-3.13194036e-01 -9.46162999e-01 7.26260722e-01 -8.03780556e-01
1.52528548e+00 -1.48975217e+00 3.34012508e-02 -3.96396071e-01
1.23190187e-01 3.21553171e-01 -2.05880702e-01 2.24921256e-01
2.48761624e-01 5.39336838e-02 4.69708584e-06 -8.94274533e-01
2.52359807e-01 -9.84372497e-02 -5.53511567e-02 -1.16157666e-01
-5.08251712e-02 1.12098050e+00 -5.88234663e-01 -7.43500233e-01
-3.83920133e-01 2.30366159e-02 -1.06246337e-01 1.24853373e-01
-1.30442455e-01 -4.39959019e-02 -2.82132685e-01 6.96115851e-01
1.67686537e-01 -1.63097993e-01 -2.49919444e-01 2.30438352e-01
-1.46128491e-01 6.81704402e-01 -4.34698910e-01 1.57769120e+00
-8.40424657e-01 7.53776491e-01 -2.70621687e-01 -5.89128435e-01
1.23679197e+00 5.88503420e-01 -9.76941828e-03 -6.64138436e-01
-1.79005831e-01 3.82484436e-01 3.30545127e-01 -4.10347909e-01
9.06614661e-01 -4.72580403e-01 -4.55944031e-01 1.47438943e+00
4.17634845e-01 -4.27061915e-01 5.27839601e-01 2.54323632e-01
1.21536839e+00 4.31617558e-01 3.32785487e-01 -4.38635826e-01
5.88144183e-01 4.18499559e-02 4.50508669e-02 8.07926297e-01
-2.92107791e-01 3.94773632e-01 6.98104382e-01 -5.18163919e-01
-1.25611198e+00 -7.40190506e-01 1.03697158e-01 1.94923961e+00
-5.15246451e-01 -2.14824751e-01 -8.50921571e-01 -8.18264306e-01
-2.08539307e-01 9.16981936e-01 -6.99080825e-01 9.38053243e-03
-4.71285909e-01 -9.34557140e-01 1.03053737e+00 5.01530826e-01
5.17826021e-01 -1.44267511e+00 -8.96486223e-01 2.72904903e-01
-2.61839479e-01 -6.63877726e-01 -6.33101046e-01 3.10726851e-01
-7.98938334e-01 -7.99933255e-01 -9.48081791e-01 -8.00351381e-01
1.49841502e-01 1.45353600e-01 1.35034275e+00 1.53110012e-01
4.47644182e-02 8.18799213e-02 -2.70310402e-01 -8.90555382e-01
-3.51119220e-01 7.52451241e-01 -3.32060605e-01 -1.92524165e-01
6.36170626e-01 -2.41718203e-01 -7.52582923e-02 6.22683801e-02
-4.79053140e-01 7.80430203e-03 5.27419567e-01 8.24741840e-01
-1.73015386e-01 -5.01953065e-01 8.94644737e-01 -1.07847238e+00
1.61431122e+00 -4.62716371e-02 -3.71177793e-02 5.90445459e-01
-7.40919709e-01 6.23638295e-02 8.33921134e-01 -4.09463018e-01
-1.20974445e+00 -3.79966468e-01 -1.28514335e-01 2.34255314e-01
-4.81980061e-03 6.44108653e-01 4.69798893e-01 -2.42188036e-01
7.11494327e-01 9.15695652e-02 -9.93421897e-02 -5.11877835e-01
-8.70787054e-02 7.18742192e-01 2.23459944e-01 -8.10970724e-01
3.30487818e-01 -4.29391079e-02 -8.50949436e-02 -3.41081619e-01
-1.92485034e+00 -4.54210281e-01 -8.08088660e-01 -4.57248539e-01
5.98463476e-01 -5.51885724e-01 -7.79751122e-01 7.07294643e-01
-1.19128561e+00 -5.84248304e-01 3.78959754e-04 8.69801790e-02
-5.38721621e-01 3.28779012e-01 -1.06608772e+00 -7.00151443e-01
-9.57475662e-01 -1.11156023e+00 9.82911408e-01 3.54960144e-01
-7.98002243e-01 -1.49404192e+00 6.15178645e-01 5.17753959e-01
7.60014355e-01 -1.47555739e-01 1.18277097e+00 -9.72968400e-01
2.47049108e-01 -3.40183020e-01 -2.57453203e-01 4.03971523e-01
-5.78292869e-02 -5.64165227e-02 -1.12592077e+00 -4.33508158e-02
3.15175265e-01 -9.55927491e-01 1.30280888e+00 3.65750253e-01
1.08409297e+00 -1.34180516e-01 3.72440010e-01 2.79724747e-01
1.15022027e+00 -1.90148372e-02 4.94960994e-01 6.05482578e-01
7.25192070e-01 6.20837927e-01 6.28021717e-01 2.16930941e-01
8.17627311e-01 6.07519627e-01 -6.47473037e-02 -2.75902063e-01
-1.99355453e-01 2.22180769e-01 8.19615066e-01 1.20990169e+00
-1.46951318e-01 -1.23068452e-01 -9.34360504e-01 6.67181730e-01
-1.92372441e+00 -9.83555853e-01 -6.22140169e-01 1.98067534e+00
1.12193549e+00 7.38301650e-02 3.58957559e-01 3.29708934e-01
4.79196012e-01 3.40063930e-01 -2.68287897e-01 -1.40003550e+00
-5.51715530e-02 5.74842811e-01 1.86513186e-01 8.06202173e-01
-1.17766678e+00 9.44920838e-01 7.11648512e+00 8.21805418e-01
-9.21683550e-01 6.12278521e-01 5.66223383e-01 -3.49524230e-01
-3.52342315e-02 -1.31369114e-01 -9.28213120e-01 3.27737361e-01
1.10728145e+00 -2.80300915e-01 -1.09709121e-01 5.97559333e-01
3.35231572e-02 -4.84106630e-01 -8.69953811e-01 2.72615045e-01
5.11603713e-01 -7.28328526e-01 3.61488163e-02 -2.07389146e-01
9.23486590e-01 -2.32153162e-01 1.66556045e-01 6.67330205e-01
5.92873514e-01 -9.76828098e-01 6.88731134e-01 5.68678200e-01
6.96368217e-01 -5.76758146e-01 1.00328779e+00 3.84898275e-01
-8.12192917e-01 4.56990376e-02 -4.64915127e-01 -5.72695792e-01
-9.70320851e-02 6.47898912e-01 -5.49002886e-01 1.91041529e-01
5.86584330e-01 7.09506392e-01 -9.76541162e-01 9.47781563e-01
-6.74769402e-01 7.53481686e-01 1.27382025e-01 -6.17834330e-01
5.42807639e-01 -2.52027631e-01 1.48111358e-01 1.42355239e+00
3.90972316e-01 -3.86937350e-01 1.22540727e-01 6.00555718e-01
-1.81352779e-01 1.44882202e-01 -1.69794053e-01 2.27477267e-01
6.43359199e-02 1.74002492e+00 -3.17502439e-01 -7.17430949e-01
-3.47324610e-01 8.25495064e-01 6.05189741e-01 5.95470741e-02
-6.96706474e-01 -6.88468099e-01 1.28512457e-01 -4.33154762e-01
-6.20690174e-02 -8.24210420e-02 -1.09449756e+00 -1.03822076e+00
5.00051351e-03 -6.78522646e-01 5.86337388e-01 -7.83156335e-01
-1.29053593e+00 5.34288824e-01 -3.93370897e-01 -8.25709224e-01
-1.38520941e-01 -6.09990120e-01 -1.30778170e+00 1.15537202e+00
-1.76380372e+00 -1.16656935e+00 -4.05557483e-01 2.28266597e-01
8.20904911e-01 -4.81173187e-01 1.18549442e+00 2.90605165e-02
-5.31046212e-01 9.34911191e-01 8.96593034e-02 6.67996109e-02
1.39100575e+00 -1.79070282e+00 4.01819825e-01 5.49837351e-01
1.03041030e-01 3.60455066e-01 3.90047312e-01 -3.79400820e-01
-4.38690454e-01 -7.47399271e-01 1.54560113e+00 -7.59262562e-01
9.05829906e-01 -1.81550458e-01 -9.25372720e-01 6.44274890e-01
7.33998895e-01 -9.08217669e-01 9.29282248e-01 7.45870292e-01
-5.12098670e-01 1.34960800e-01 -8.55513394e-01 5.29785216e-01
2.89975643e-01 -4.66451705e-01 -7.92924583e-01 4.28353250e-01
1.38547972e-01 -2.63469934e-01 -1.10808575e+00 1.55999079e-01
8.36847365e-01 -1.04160309e+00 4.95485455e-01 -5.41458905e-01
1.23402536e+00 1.18743636e-01 4.56647098e-01 -1.74625695e+00
-4.60542917e-01 -1.34407938e-01 -1.44679084e-01 1.09005690e+00
7.31652319e-01 -3.52897823e-01 8.95137787e-01 3.66763294e-01
-3.31072330e-01 -9.74759340e-01 -5.67965388e-01 -6.54987216e-01
6.39778614e-01 9.10408944e-02 2.87639052e-01 8.95142734e-01
2.75792927e-01 8.34354579e-01 -4.38091606e-01 -6.95657372e-01
6.10330403e-01 3.93024296e-01 4.65512753e-01 -1.44116783e+00
5.72622009e-02 -1.07919216e+00 2.32271269e-01 -5.64901233e-01
2.16656864e-01 -8.62686753e-01 1.19381756e-01 -1.92289186e+00
5.70624173e-01 -2.44339332e-01 -4.55909014e-01 1.11768436e+00
-6.73168838e-01 5.94719410e-01 3.15753609e-01 2.43559107e-01
-9.33516145e-01 1.36448190e-01 1.33627176e+00 -2.25253731e-01
9.55061242e-02 1.52550325e-01 -9.49068129e-01 6.46501541e-01
1.13294268e+00 -6.76719368e-01 -1.02504387e-01 -5.37282407e-01
3.65542322e-01 -1.32781155e-02 -9.01561975e-02 -8.62022996e-01
5.48376501e-01 6.73545972e-02 3.78448069e-01 -6.41121268e-01
3.40580612e-01 1.21514425e-01 -8.33291233e-01 4.96711314e-01
-7.60520458e-01 3.68497699e-01 -5.10434993e-02 -4.50968295e-01
-3.42208028e-01 -8.77459109e-01 9.50302839e-01 -4.65385199e-01
-3.13243330e-01 -9.93079543e-02 -5.92251122e-01 1.41598538e-01
6.30526483e-01 -3.12137425e-01 -4.93927985e-01 -5.45004427e-01
-6.73212469e-01 4.32325512e-01 3.09173763e-01 3.73356849e-01
4.59077835e-01 -1.08225334e+00 -1.18719089e+00 -2.34685078e-01
-1.23539949e-02 -4.88615066e-01 -6.60753474e-02 9.78169322e-01
-3.52677613e-01 7.62909234e-01 -5.09672523e-01 -9.21926051e-02
-1.57889175e+00 -1.80743724e-01 4.14420396e-01 -1.05052567e+00
-1.17207088e-01 9.24285412e-01 -1.46512747e-01 -7.78386533e-01
2.83871830e-01 1.87617242e-01 -7.63774693e-01 6.71414793e-01
6.20655298e-01 5.30532837e-01 5.14760911e-01 -2.25172818e-01
1.49476051e-01 7.61793077e-01 -4.85413522e-01 -2.45128110e-01
1.28478134e+00 2.69371867e-01 -4.02087867e-01 9.36716676e-01
7.97644019e-01 -1.40851825e-01 -7.92938292e-01 -1.80204079e-01
5.22254825e-01 1.54864416e-02 5.15293181e-02 -1.30310881e+00
-7.05461740e-01 1.36794686e+00 -9.40988436e-02 4.73295599e-01
7.48803079e-01 -6.16346359e-01 7.78175533e-01 4.71381456e-01
3.65900785e-01 -1.40765202e+00 5.17846227e-01 1.22833741e+00
5.77572107e-01 -1.05823243e+00 2.53201395e-01 3.33346605e-01
-1.04804230e+00 1.41686380e+00 9.27103043e-01 -3.81703675e-01
1.52516514e-01 3.74582946e-01 4.92666274e-01 -3.64596039e-01
-1.14202952e+00 7.78686032e-02 3.45768690e-01 2.59638075e-02
1.10224521e+00 3.54467146e-02 -5.40785372e-01 5.98842263e-01
-5.40454447e-01 -1.53177515e-01 1.03603804e+00 6.48490429e-01
-6.51503623e-01 -1.27048576e+00 -3.85639906e-01 1.05148065e+00
-8.88071954e-01 -3.52959991e-01 -8.49703670e-01 3.88037920e-01
-3.41345787e-01 9.05425727e-01 -2.46388689e-01 -3.09460372e-01
2.56727755e-01 6.72191083e-01 6.00700557e-01 -1.00043011e+00
-1.50247574e+00 -3.60122114e-01 2.29375213e-01 -6.22890405e-02
-2.76053488e-01 -7.34241724e-01 -9.19891000e-01 -3.80804926e-01
-5.05743504e-01 2.03832999e-01 7.47558057e-01 1.29799390e+00
-2.40660056e-01 8.55665207e-01 4.49862748e-01 -7.04521060e-01
-9.31625307e-01 -1.50390160e+00 -5.11978388e-01 1.23573415e-01
1.84954301e-01 -4.20960546e-01 -3.23981404e-01 -1.61057234e-01] | [11.29997444152832, 9.366113662719727] |
0a230356-2f28-449d-812e-f8a64ae3450b | 2d-medical-image-synthesis-using-transformer | null | null | https://iopscience.iop.org/article/10.1088/1361-6560/acca5c/meta | https://iopscience.iop.org/article/10.1088/1361-6560/acca5c/pdf | 2D medical image synthesis using transformer-based denoising diffusion probabilistic model | Objective. Artificial intelligence (AI) methods have gained popularity in medical imaging research. The size and scope of the training image datasets needed for successful AI model deployment does not always have the desired scale. In this paper, we introduce a medical image synthesis framework aimed at addressing the challenge of limited training datasets for AI models. Approach. The proposed 2D image synthesis framework is based on a diffusion model using a Swin-transformer-based network. This model consists of a forward Gaussian noise process and a reverse process using the transformer-based diffusion model for denoising. Training data includes four image datasets: chest x-rays, heart MRI, pelvic CT, and abdomen CT. We evaluated the authenticity, quality, and diversity of the synthetic images using visual Turing assessments conducted by three medical physicists, and four quantitative evaluations: the Inception score (IS), Fréchet Inception Distance score (FID), feature similarity and diversity score (DS, indicating diversity similarity) between the synthetic and true images. To leverage the framework value for training AI models, we conducted COVID-19 classification tasks using real images, synthetic images, and mixtures of both images. Main results. Visual Turing assessments showed an average accuracy of 0.64 (accuracy converging to $50 \% $ indicates a better realistic visual appearance of the synthetic images), sensitivity of 0.79, and specificity of 0.50. Average quantitative accuracy obtained from all datasets were IS = 2.28, FID = 37.27, FDS = 0.20, and DS = 0.86. For the COVID-19 classification task, the baseline network obtained an accuracy of 0.88 using a pure real dataset, 0.89 using a pure synthetic dataset, and 0.93 using a dataset mixed of real and synthetic data. Significance. A image synthesis framework was demonstrated for medical image synthesis, which can generate high-quality medical images of different imaging modalities with the purpose of supplementing existing training sets for AI model deployment. This method has potential applications in many data-driven medical imaging research. | ['Justin Roper', 'Sagar A Patel', 'Joseph Shelton', 'Ashish B Patel', 'Junbo Peng', 'Chih-Wei Chang', 'Marian Axente', 'Richard L J Qiu', 'Tonghe Wang', 'Shaoyan Pan'] | 2023-05-05 | null | null | null | physics-in-medicine-biology-2023-5 | ['image-generation', 'specificity'] | ['computer-vision', 'natural-language-processing'] | [ 4.62462366e-01 2.86375850e-01 3.04238617e-01 -2.49318793e-01
-9.36165273e-01 -4.51652825e-01 5.22565186e-01 -7.25480765e-02
-5.18805027e-01 4.66020495e-01 -6.31013364e-02 -2.65079767e-01
-2.68396378e-01 -5.77381611e-01 -5.25272071e-01 -9.35137570e-01
-1.68489709e-01 6.20602012e-01 -1.14257134e-01 -1.73329860e-02
2.92074438e-02 5.26069105e-01 -1.26439106e+00 5.33501387e-01
9.32913005e-01 1.34336364e+00 1.55483335e-01 9.54909623e-01
4.57812250e-01 1.03599691e+00 -9.96781051e-01 -4.56273049e-01
4.62289006e-01 -6.67815745e-01 -5.72910011e-01 3.39078397e-01
3.14269572e-01 -3.47568333e-01 -2.69968927e-01 9.64840472e-01
9.97430980e-01 -3.03942680e-01 7.96312451e-01 -1.19156647e+00
-7.79751837e-01 3.01937550e-01 -5.35613537e-01 2.93487668e-01
2.10939541e-01 7.46304870e-01 5.24812937e-01 -5.68222940e-01
9.40460980e-01 9.57514405e-01 6.94962084e-01 5.39926648e-01
-1.24357915e+00 -6.33772135e-01 -7.09295511e-01 -5.37524894e-02
-1.25620461e+00 -9.56344977e-02 6.18131578e-01 -6.35313511e-01
5.92479944e-01 4.03013408e-01 7.25649357e-01 1.18045831e+00
5.93239009e-01 6.57070458e-01 1.50263011e+00 -2.94690192e-01
4.35760021e-01 4.00372386e-01 -2.13581249e-01 6.41485095e-01
2.78782547e-01 2.75481880e-01 -1.59085795e-01 -8.23361576e-02
7.97879875e-01 -3.38911563e-01 -1.91994429e-01 -7.97984675e-02
-1.33819580e+00 7.99080729e-01 6.55596018e-01 5.52869380e-01
-7.22089708e-01 -1.45675227e-01 2.69430876e-01 3.88223469e-01
3.04048806e-01 9.26473379e-01 1.14074253e-01 3.02514117e-02
-9.40934777e-01 1.70544311e-01 5.39866745e-01 5.29537976e-01
1.20385908e-01 2.71347255e-01 -4.15221512e-01 8.21559608e-01
-1.07179843e-01 7.51167059e-01 8.33332419e-01 -1.24264967e+00
-3.79695930e-02 4.87873644e-01 -1.64443076e-01 -1.16525066e+00
-6.00507259e-01 -7.75466979e-01 -1.21186268e+00 3.42180103e-01
2.79183120e-01 -2.64997780e-02 -1.18898726e+00 1.69229150e+00
1.41817451e-01 1.31879449e-01 2.64161736e-01 9.07834411e-01
1.04683435e+00 4.87051070e-01 -1.01480834e-01 -2.06720412e-01
1.37638485e+00 -6.98358655e-01 -6.39528275e-01 -6.57656491e-02
6.78251982e-01 -6.98964119e-01 1.14053226e+00 6.66389585e-01
-1.43453062e+00 -6.62148058e-01 -1.20251858e+00 4.83437061e-01
1.05622143e-01 2.41543293e-01 1.31119266e-01 9.41730022e-01
-1.23971057e+00 3.68078798e-01 -8.08878243e-01 -5.75110838e-02
6.09070897e-01 2.46849567e-01 -4.03181821e-01 -3.82937342e-01
-9.40463543e-01 9.55364525e-01 2.53878921e-01 -6.43161684e-02
-1.04878473e+00 -7.84194171e-01 -6.27805591e-01 -2.41161689e-01
-2.89269358e-01 -8.70367706e-01 8.95010769e-01 -1.28332210e+00
-1.11412752e+00 1.16126585e+00 5.21217704e-01 -6.14749014e-01
8.67649615e-01 4.10265207e-01 -4.13606167e-01 5.35078049e-01
2.62162447e-01 9.05266583e-01 7.07077265e-01 -1.51104581e+00
-3.67233336e-01 -3.44340086e-01 -3.05509299e-01 2.40867391e-01
-2.09339276e-01 -2.17055365e-01 -2.03456342e-01 -8.92313004e-01
6.23125508e-02 -1.17341757e+00 -2.07070306e-01 1.29765317e-01
-2.19288796e-01 3.53890389e-01 4.00533050e-01 -8.79783511e-01
1.04233384e+00 -2.24726367e+00 -1.41538352e-01 4.42021847e-01
4.99792874e-01 3.29374045e-01 -3.80824834e-01 -4.31455038e-02
-1.89592063e-01 9.21754539e-02 -5.58466494e-01 -6.20218106e-02
-3.09521168e-01 1.32899642e-01 1.32088661e-01 3.60809386e-01
3.45373422e-01 9.91775572e-01 -8.45710397e-01 -5.73129535e-01
1.21515021e-01 5.10650933e-01 -5.08866966e-01 2.35724539e-01
2.04502851e-01 5.82805276e-01 4.15515825e-02 6.37304187e-01
6.91380978e-01 -5.24732649e-01 6.35496005e-02 -3.97568554e-01
4.13502306e-01 -5.04834950e-01 -1.13651466e+00 1.48998690e+00
-1.87055796e-01 6.05692744e-01 -4.48661074e-02 -7.85977006e-01
9.57602918e-01 3.61219257e-01 6.89567029e-01 -1.09001946e+00
1.72212735e-01 3.10729384e-01 6.12852275e-01 -6.53830886e-01
8.03412646e-02 -4.18207854e-01 5.29395007e-02 5.53658783e-01
7.89457336e-02 -7.54084706e-01 2.46206507e-01 2.51155645e-01
1.27005744e+00 -4.74619448e-01 -1.60178859e-02 -1.82086751e-01
3.19124132e-01 2.93514043e-01 2.07921401e-01 9.89919364e-01
-3.65115523e-01 1.04498112e+00 4.77263957e-01 -4.54579026e-01
-1.30257547e+00 -1.28128946e+00 -3.26741278e-01 2.14290529e-01
6.05514124e-02 4.32500429e-02 -9.86060500e-01 -5.03576279e-01
-3.46609175e-01 7.95497596e-01 -9.08311903e-01 -5.15256524e-01
-3.72057289e-01 -9.15528238e-01 9.85736728e-01 3.20519358e-01
5.55301011e-01 -1.30756009e+00 -8.52782011e-01 4.64762701e-03
-3.93652022e-01 -1.06198835e+00 -2.02085957e-01 -2.56224930e-01
-8.17032456e-01 -8.48607361e-01 -9.73538280e-01 -7.81645119e-01
7.71319211e-01 -1.92451805e-01 1.14486992e+00 1.64073944e-01
-6.41368985e-01 3.63849998e-01 -8.05153176e-02 -3.06844085e-01
-1.03106761e+00 -4.62239176e-01 3.25949490e-02 -1.60012528e-01
-2.79641002e-01 -2.32983857e-01 -9.48855877e-01 5.13019025e-01
-1.45111263e+00 2.37816721e-01 8.73807847e-01 1.38457680e+00
5.50513089e-01 2.43178830e-02 3.31305832e-01 -6.02849305e-01
8.46942782e-01 -3.31621736e-01 -6.82947561e-02 2.76977867e-01
-6.92194521e-01 -2.14495420e-01 5.20521760e-01 -7.54085660e-01
-8.31847727e-01 -2.44647264e-01 -7.14869127e-02 -5.96040010e-01
2.55954862e-02 5.89699388e-01 4.64197636e-01 -5.67646511e-02
1.26822269e+00 2.67484307e-01 5.31838357e-01 2.36349985e-01
4.41308990e-02 6.25988543e-01 6.95677161e-01 -3.21932107e-01
4.24419701e-01 4.78274286e-01 -1.14262670e-01 -7.74819553e-01
-3.79630655e-01 1.92640781e-01 -2.50145406e-01 -4.01126981e-01
7.47565210e-01 -7.10646629e-01 -4.91583765e-01 5.86327910e-01
-8.24132621e-01 -3.24086636e-01 -4.82655078e-01 7.24545836e-01
-4.74655092e-01 2.97816038e-01 -7.15653121e-01 -4.40435559e-01
-5.45553327e-01 -1.67367601e+00 7.89174199e-01 5.37768491e-02
-3.64001364e-01 -9.43036973e-01 -6.94190934e-02 5.96201062e-01
6.20961905e-01 7.08204150e-01 9.48930860e-01 -6.60120785e-01
-1.39742538e-01 -4.26053137e-01 -2.11958051e-01 6.44968569e-01
2.35632390e-01 -1.71710983e-01 -8.83007526e-01 -3.04559737e-01
4.16541874e-01 -5.14123857e-01 4.23257053e-01 7.25683570e-01
9.27966893e-01 -2.39927188e-01 -2.10069362e-02 4.99529362e-01
1.12104774e+00 6.19436443e-01 8.25519025e-01 2.22225621e-01
3.40041101e-01 5.13701618e-01 4.59859610e-01 3.72193545e-01
4.87189516e-02 3.11959952e-01 2.01766700e-01 -4.69853222e-01
-5.49309254e-01 1.85365498e-01 5.01137115e-02 7.35605061e-01
1.88488752e-01 -2.55667478e-01 -1.36457038e+00 5.26655555e-01
-1.22721910e+00 -8.79049838e-01 8.29654485e-02 1.96232903e+00
7.83417523e-01 3.39611411e-01 -7.81231076e-02 2.32565641e-01
5.68423390e-01 -3.86869013e-01 -6.80836558e-01 -2.41872534e-01
-3.35873693e-01 2.41487294e-01 5.10013513e-02 2.06093565e-01
-8.33752930e-01 3.02781910e-01 5.99298716e+00 8.39773715e-01
-1.44596982e+00 -6.36116713e-02 1.40410936e+00 2.79359091e-02
-2.08585486e-01 -6.45158112e-01 1.01778857e-01 5.25822580e-01
8.93318415e-01 -1.34727836e-01 2.39476070e-01 5.65202713e-01
2.42236294e-02 -2.07188219e-01 -8.13453197e-01 1.14690983e+00
3.28531533e-01 -1.44900358e+00 1.89836547e-01 -2.90531777e-02
9.83030796e-01 -1.80935323e-01 6.87866211e-01 3.44526656e-02
8.93549547e-02 -1.43925095e+00 5.69044769e-01 5.23559988e-01
1.13122809e+00 -4.49325830e-01 1.10699868e+00 3.42193902e-01
-4.30008233e-01 -4.00133803e-02 5.49081862e-02 3.59637380e-01
-9.88090932e-02 5.11881053e-01 -1.27413046e+00 4.88283813e-01
8.18837881e-01 2.58685559e-01 -5.90743005e-01 8.21521938e-01
2.60550052e-01 5.09083092e-01 -3.02017897e-01 1.14258632e-01
2.08646804e-01 6.86642826e-02 3.75812143e-01 1.10843563e+00
4.27618593e-01 2.24769264e-01 -1.71889693e-01 7.81486332e-01
-1.11831157e-02 -2.76474860e-02 -5.68280041e-01 5.54262921e-02
3.01925808e-01 1.25265765e+00 -1.01137006e+00 -5.94861031e-01
8.70754197e-02 8.55227590e-01 -3.31920236e-01 9.61595774e-02
-9.87888455e-01 -2.98531830e-01 3.07532580e-04 3.12337698e-03
-1.26975432e-01 3.12410563e-01 -7.96016693e-01 -7.45496273e-01
-8.44700858e-02 -1.51372778e+00 4.42992061e-01 -1.11401284e+00
-1.40235388e+00 1.13448143e+00 3.72193009e-02 -1.46011567e+00
-3.20670158e-01 -5.20448327e-01 -3.18492115e-01 7.76889503e-01
-8.47914577e-01 -1.07054746e+00 -6.80465639e-01 3.95563215e-01
3.82832497e-01 -3.85951072e-01 7.68151999e-01 2.46481970e-01
-2.13834777e-01 6.18094146e-01 3.56659926e-02 1.99721739e-01
6.31644070e-01 -1.04439366e+00 1.64794788e-01 5.71929038e-01
-2.08853915e-01 4.54139531e-01 6.35969102e-01 -6.33496225e-01
-1.13182092e+00 -8.54645669e-01 3.80780905e-01 -3.01020682e-01
2.89037317e-01 2.37228140e-01 -7.82810748e-01 1.27152249e-01
1.00836739e-01 9.43662375e-02 5.86179912e-01 -8.83404672e-01
-2.65485905e-02 -7.66768605e-02 -1.71358907e+00 6.76014602e-01
6.01207972e-01 -1.36902109e-01 -2.94043273e-01 2.83825278e-01
4.09076989e-01 -6.16616189e-01 -1.20331442e+00 6.66982710e-01
6.93008721e-01 -1.04464591e+00 1.18106341e+00 -1.88724384e-01
8.00524116e-01 -1.44431859e-01 -1.82251394e-01 -1.24701774e+00
-2.60574609e-01 -1.70771748e-01 3.17468375e-01 5.87418139e-01
4.75095183e-01 -4.08127129e-01 5.31583071e-01 7.44596660e-01
-9.36447755e-02 -8.35413098e-01 -9.27121520e-01 -5.54137170e-01
1.31450623e-01 -4.15286988e-01 2.41218418e-01 1.07632303e+00
-3.05565417e-01 1.57883987e-01 -9.16780159e-02 -7.90305138e-02
6.68857694e-01 -3.53516221e-01 5.27260065e-01 -9.26138639e-01
-3.93438160e-01 -6.31234229e-01 -6.54148161e-01 -2.81983316e-01
-4.21299517e-01 -9.86264050e-01 -1.85762614e-01 -1.54015708e+00
2.31078461e-01 -6.09326839e-01 -1.98499471e-01 2.94455081e-01
-9.24345031e-02 6.50931597e-01 2.35035002e-01 4.04927075e-01
3.99833769e-02 1.27855122e-01 1.51845944e+00 -4.69322830e-01
-5.15630469e-02 -1.95154890e-01 -5.87626278e-01 3.98059815e-01
9.21940267e-01 -4.58850592e-01 -5.53194344e-01 -3.94203871e-01
-8.92461911e-02 3.67397845e-01 3.65123540e-01 -1.36351109e+00
1.11807592e-01 2.04068661e-01 7.00036764e-01 -1.73551098e-01
3.91851068e-01 -7.13799000e-01 6.95951045e-01 1.24857175e+00
-3.92920911e-01 2.43816063e-01 2.64159590e-01 1.05851278e-01
-1.77243650e-01 -4.46538515e-02 1.09408975e+00 -2.42690906e-01
-3.04000586e-01 -5.97704835e-02 -1.73192203e-01 8.10237676e-02
1.19854999e+00 -5.31407118e-01 -1.96015641e-01 -5.40522695e-01
-7.84663975e-01 -1.19513407e-01 3.69723082e-01 2.49918163e-01
1.06828070e+00 -1.26574790e+00 -1.07497776e+00 3.80049407e-01
9.85623598e-02 -2.08122879e-01 4.92742509e-01 1.11437380e+00
-9.77283478e-01 -1.28949866e-01 -5.36186993e-01 -1.05604720e+00
-1.25924695e+00 4.20731127e-01 6.01448774e-01 -4.10303354e-01
-4.40581888e-01 7.13722348e-01 3.21201622e-01 -4.54489738e-01
3.30530256e-02 -2.41561338e-01 1.11758649e-01 -1.24050207e-01
4.67873573e-01 3.14161062e-01 1.90582573e-01 -5.67195892e-01
-2.91814744e-01 4.93893087e-01 4.46308404e-02 -5.14126837e-01
1.27639437e+00 2.69545823e-01 8.94640088e-02 2.42811754e-01
1.10277295e+00 -4.45881426e-01 -8.56175303e-01 7.58055307e-04
-3.38702708e-01 -2.89653152e-01 -4.60610492e-03 -1.34762824e+00
-1.06956148e+00 8.60407293e-01 1.15790308e+00 2.23510250e-01
1.47784865e+00 -2.10108787e-01 5.08566380e-01 5.19854110e-03
2.76079495e-02 -8.23533177e-01 5.54341495e-01 5.92959449e-02
1.10161459e+00 -1.17272079e+00 1.44760758e-02 8.76213238e-02
-1.31096542e+00 9.48316336e-01 4.86013710e-01 5.26682287e-03
4.45985705e-01 4.60091829e-01 4.83439893e-01 -3.66129726e-01
-6.06232524e-01 3.27975690e-01 3.83404016e-01 6.41500533e-01
3.22342604e-01 -7.65366554e-02 -1.19343445e-01 4.87317771e-01
-2.41707683e-01 1.45172313e-01 4.71106142e-01 9.67665553e-01
-9.17705894e-02 -5.08604527e-01 -5.43593943e-01 5.81878006e-01
-4.82792407e-01 -3.22043337e-02 -1.51480719e-01 7.21908450e-01
1.09030597e-01 9.17343915e-01 9.68274027e-02 -6.26586020e-01
3.00736636e-01 -3.68658483e-01 4.72735256e-01 -1.37795582e-02
-9.34768140e-01 1.68364011e-02 -9.66039747e-02 -2.25314304e-01
-3.43222558e-01 -3.80853027e-01 -1.00913966e+00 -2.38071382e-01
-1.99149102e-01 5.70143461e-02 7.38979518e-01 4.56673205e-01
3.84738564e-01 6.60773873e-01 7.18344092e-01 -5.74333251e-01
-4.87547755e-01 -1.01737916e+00 -3.61479312e-01 8.17409992e-01
2.18771324e-01 -3.08487743e-01 -2.63281345e-01 2.00989887e-01] | [14.2398099899292, -1.933763861656189] |
df951591-f105-449c-9e15-e41399da974c | generation-of-anonymous-chest-radiographs | 2211.01323 | null | https://arxiv.org/abs/2211.01323v2 | https://arxiv.org/pdf/2211.01323v2.pdf | Generation of Anonymous Chest Radiographs Using Latent Diffusion Models for Training Thoracic Abnormality Classification Systems | The availability of large-scale chest X-ray datasets is a requirement for developing well-performing deep learning-based algorithms in thoracic abnormality detection and classification. However, biometric identifiers in chest radiographs hinder the public sharing of such data for research purposes due to the risk of patient re-identification. To counteract this issue, synthetic data generation offers a solution for anonymizing medical images. This work employs a latent diffusion model to synthesize an anonymous chest X-ray dataset of high-quality class-conditional images. We propose a privacy-enhancing sampling strategy to ensure the non-transference of biometric information during the image generation process. The quality of the generated images and the feasibility of serving as exclusive training data are evaluated on a thoracic abnormality classification task. Compared to a real classifier, we achieve competitive results with a performance gap of only 3.5% in the area under the receiver operating characteristic curve. | ['Andreas Maier', 'Florian Thamm', 'Lukas Folle', 'Kai Packhäuser'] | 2022-11-02 | null | null | null | null | ['synthetic-data-generation', 'synthetic-data-generation'] | ['medical', 'miscellaneous'] | [ 5.54771364e-01 4.14497852e-01 -1.02502517e-01 -6.25383735e-01
-1.11801314e+00 -3.35484385e-01 2.92237669e-01 3.21857989e-01
-6.40866339e-01 8.49166453e-01 -4.16830704e-02 -4.04957801e-01
-4.17709686e-02 -8.98782790e-01 -7.24320531e-01 -8.29850852e-01
3.20780665e-01 1.79108784e-01 -3.87726724e-01 4.99164343e-01
-1.00968517e-01 4.85913903e-01 -1.12052059e+00 4.88964945e-01
1.02306664e+00 1.11690485e+00 -4.12982255e-01 4.39435422e-01
1.18837073e-01 4.05669987e-01 -8.19966197e-01 -9.13823009e-01
7.36313820e-01 -4.78642255e-01 -6.52662396e-01 1.64114073e-01
5.00395954e-01 -7.83836246e-01 -4.80974138e-01 1.07632887e+00
7.92357445e-01 -4.17604029e-01 5.84585786e-01 -1.41726160e+00
-7.52628744e-01 5.48036397e-01 -3.86607945e-01 -7.55657554e-02
1.82106018e-01 -1.62240528e-02 6.12192214e-01 -4.27536428e-01
6.23969913e-01 4.01167989e-01 8.19784582e-01 7.43228376e-01
-1.08518636e+00 -7.76936412e-01 -5.65765023e-01 -2.42658809e-01
-1.45145643e+00 -2.36296400e-01 6.26716375e-01 -4.16198343e-01
1.06293689e-02 5.26365995e-01 5.46531022e-01 1.37249064e+00
5.77347457e-01 5.67528546e-01 1.17683411e+00 -3.07030201e-01
1.82663217e-01 3.76163155e-01 -2.17638642e-01 7.70371079e-01
7.45354056e-01 8.84167403e-02 -5.92818022e-01 -5.44707477e-01
8.30505431e-01 3.19561690e-01 -1.89611107e-01 -6.75468683e-01
-1.09086037e+00 8.35442543e-01 4.80695814e-01 2.92948097e-01
-4.97060031e-01 2.32737442e-03 2.75980711e-01 1.83161423e-01
5.33151627e-01 3.22574139e-01 4.56200577e-02 1.77515835e-01
-9.93364692e-01 1.00186393e-01 4.73661184e-01 7.30761945e-01
1.29984245e-01 -8.03782642e-02 -2.33438939e-01 6.43619299e-01
-8.72076079e-02 5.04945338e-01 7.75017440e-01 -7.72958994e-01
6.12301409e-01 6.10286176e-01 -1.07350135e-02 -9.67558146e-01
-7.58815706e-02 -4.22144860e-01 -1.10689282e+00 6.09826669e-02
6.31448209e-01 -1.88794851e-01 -8.08518112e-01 1.57178664e+00
3.19368273e-01 -2.57689297e-01 1.46997437e-01 7.39674270e-01
5.09707570e-01 1.26496464e-01 4.29759584e-02 -1.50681660e-01
1.37350309e+00 -4.10834253e-01 -7.17658043e-01 3.93952310e-01
5.42663991e-01 -4.41095889e-01 1.00195932e+00 1.38497457e-01
-8.97305846e-01 -4.30017382e-01 -1.09923005e+00 6.72301278e-02
-1.33262932e-01 3.86121839e-01 4.25897926e-01 1.40216386e+00
-6.92182064e-01 4.23134983e-01 -8.88044834e-01 -1.51951779e-02
1.20174181e+00 5.68024397e-01 -7.66426563e-01 -1.69177771e-01
-1.04912055e+00 2.84093440e-01 9.06576291e-02 -2.06888944e-01
-3.76957417e-01 -8.02870929e-01 -5.78280270e-01 -3.47919837e-02
1.74230397e-01 -8.75945568e-01 8.42813909e-01 -1.17718494e+00
-1.21919954e+00 1.22224605e+00 3.27307731e-01 -7.64965415e-01
1.29971218e+00 -1.12989470e-01 -3.34728628e-01 3.46693546e-01
2.59939760e-01 5.84886551e-01 1.19433951e+00 -9.99551535e-01
-5.21487772e-01 -5.63613474e-01 -4.09006357e-01 -2.50155516e-02
-6.51837349e-01 -3.67014170e-01 9.68298595e-03 -9.39338624e-01
2.57182479e-01 -1.13376832e+00 -2.48394072e-01 3.17426145e-01
-4.14008409e-01 3.21113616e-01 6.94418609e-01 -8.52428615e-01
8.10986876e-01 -2.23192263e+00 -4.35636461e-01 3.69515091e-01
5.15432715e-01 2.31998935e-01 2.15536550e-01 -4.57158051e-02
-1.38380185e-01 3.08040798e-01 -7.07157791e-01 -3.95602137e-01
-2.36737624e-01 2.63063516e-02 -2.12712616e-01 7.04069674e-01
3.83519530e-02 9.48436677e-01 -5.86937189e-01 -5.28396308e-01
-1.06657147e-01 3.20630521e-01 -3.19179058e-01 2.85613835e-01
2.71108091e-01 5.94853699e-01 -4.18342084e-01 4.65494961e-01
8.05292428e-01 -2.30372161e-01 4.32740487e-02 4.95882668e-02
4.80329216e-01 -6.56903395e-03 -8.20878148e-01 1.64485121e+00
-2.49281108e-01 4.40807790e-01 -8.72979462e-02 -5.74550509e-01
8.11275780e-01 4.15720642e-01 9.85588014e-01 -5.77284336e-01
2.94465452e-01 2.28882194e-01 -3.03169843e-02 -4.46230233e-01
3.29507887e-01 -2.02523440e-01 -1.88767150e-01 9.44405913e-01
-4.09118176e-01 3.60487774e-02 -4.72365648e-01 1.51200384e-01
1.04962718e+00 -3.05011123e-01 -1.65774059e-02 -9.66353994e-03
3.16845059e-01 -2.67956972e-01 4.34380561e-01 8.52047920e-01
-2.77991116e-01 9.96344447e-01 2.53513008e-01 -5.06329358e-01
-1.14621234e+00 -1.11418116e+00 -3.98117989e-01 2.72378892e-01
-2.47172654e-01 3.78671065e-02 -1.01489353e+00 -1.13746631e+00
2.17362374e-01 6.49226904e-01 -6.40655696e-01 -3.59754413e-01
-3.97808194e-01 -8.06408465e-01 1.07854533e+00 4.74689722e-01
7.67928302e-01 -7.04777598e-01 -9.36038554e-01 -8.52918476e-02
-3.22800875e-01 -1.01639557e+00 -5.11421621e-01 -2.98953354e-01
-7.10846961e-01 -1.09918714e+00 -1.10517395e+00 -3.15214157e-01
1.10876858e+00 -2.39284247e-01 5.58082461e-01 8.01595598e-02
-6.88750565e-01 4.65312988e-01 -1.01741940e-01 -4.79287714e-01
-9.26693141e-01 2.20060021e-01 1.80697590e-02 3.49983245e-01
2.47941285e-01 -1.31163806e-01 -8.89025629e-01 1.79801166e-01
-1.09066606e+00 -1.49022818e-01 4.53126490e-01 1.03461552e+00
6.18301570e-01 6.35014102e-02 7.20952809e-01 -1.15087211e+00
8.03902626e-01 -3.70224595e-01 -5.21893680e-01 2.16158748e-01
-9.74749148e-01 -2.89555788e-01 5.53709924e-01 -3.91397476e-01
-1.04546678e+00 2.91835427e-01 2.02041864e-02 -4.28335398e-01
-2.03764513e-01 1.25693396e-01 -1.09636029e-02 -1.85100958e-01
6.63905263e-01 2.47705147e-01 3.63443017e-01 -1.04434311e-01
2.66916901e-01 9.05150712e-01 5.02505124e-01 -2.54664183e-01
7.68166363e-01 5.69930017e-01 1.40475199e-01 -5.36614716e-01
-5.07467687e-01 -9.39530879e-02 -6.63657665e-01 -1.06198862e-01
9.94790852e-01 -8.59557629e-01 -3.76413852e-01 5.96884668e-01
-6.54064178e-01 3.23146462e-01 -6.10204518e-01 6.25253439e-01
-4.80302334e-01 5.85238039e-01 -5.20392716e-01 -6.36788070e-01
-5.75209260e-01 -1.13157785e+00 9.67796683e-01 -1.14710227e-01
-2.22208053e-01 -5.15726030e-01 -1.06585741e-01 7.74506986e-01
3.02605659e-01 5.45088351e-01 9.84409988e-01 -6.93267822e-01
-6.67838097e-01 -8.61166120e-01 -1.59414306e-01 5.40360689e-01
4.64635193e-01 -2.82430738e-01 -1.02899075e+00 -4.79732543e-01
4.80963975e-01 -3.52006108e-01 5.13312399e-01 3.54559749e-01
1.60500979e+00 -4.26968604e-01 -3.58083159e-01 7.40536690e-01
1.05976856e+00 1.14497460e-01 6.21440232e-01 -9.63194221e-02
6.99244678e-01 6.72436595e-01 3.78659546e-01 5.07620156e-01
9.89570618e-02 2.87705213e-01 1.13834202e-01 -3.33132714e-01
6.85864538e-02 -3.85242850e-01 -2.88544536e-01 4.68246281e-01
3.00357223e-01 -2.90635884e-01 -1.02345288e+00 4.65923250e-01
-1.31772101e+00 -7.52934217e-01 -7.14857578e-02 2.42021918e+00
7.83632457e-01 -1.51277602e-01 6.39170259e-02 3.27161431e-01
7.54799128e-01 -1.42123148e-01 -6.82981491e-01 -7.13389888e-02
-1.47512525e-01 2.62296826e-01 8.92452657e-01 -9.50220674e-02
-1.13855100e+00 3.44964743e-01 6.03691721e+00 5.75301766e-01
-1.29003501e+00 1.53576598e-01 1.14997876e+00 -1.06329910e-01
-2.87197679e-01 -4.81384963e-01 -1.55213699e-01 5.35030067e-01
8.09154093e-01 -2.86036104e-01 -1.17812373e-01 7.81083405e-01
-2.06393749e-01 2.34574894e-03 -9.06310201e-01 1.13263273e+00
1.31529734e-01 -1.46304131e+00 6.54158443e-02 4.10350949e-01
7.43757725e-01 -2.71544188e-01 5.99066019e-01 -4.33729470e-01
-1.00884169e-01 -1.02777290e+00 5.49803793e-01 4.85445827e-01
1.38712883e+00 -7.99840033e-01 8.82184684e-01 2.28429377e-01
-5.20500720e-01 -1.57206520e-01 -2.25398913e-01 5.77073693e-01
-8.79551023e-02 4.73068267e-01 -9.86931384e-01 7.14164972e-01
6.62858546e-01 1.22009955e-01 -6.95611775e-01 7.60636151e-01
1.30839124e-01 5.91675282e-01 -3.27636689e-01 2.77159512e-01
-1.71924740e-01 -7.92168230e-02 3.11814457e-01 5.01017749e-01
5.13765931e-01 6.70941696e-02 -2.28917032e-01 8.97197068e-01
-6.35869741e-01 2.85405189e-01 -8.97544920e-01 -2.03198507e-01
3.09315652e-01 8.88906479e-01 -6.88384295e-01 -2.42753476e-02
-1.83004364e-01 1.13203526e+00 -1.70475379e-01 -1.35465577e-01
-8.56856227e-01 -1.81297094e-01 6.87478632e-02 5.39067209e-01
1.11130871e-01 1.45941198e-01 -7.54165590e-01 -1.11450982e+00
2.55288839e-01 -1.07138956e+00 4.36919063e-01 -4.76706058e-01
-1.39999092e+00 6.56439602e-01 -1.88042030e-01 -1.33958650e+00
-2.55744398e-01 -1.33533716e-01 -1.62192196e-01 9.30309057e-01
-9.96070206e-01 -1.37547970e+00 -3.37957084e-01 6.17812753e-01
-4.67744991e-02 -6.44180775e-01 1.07399344e+00 3.05935860e-01
-3.62714052e-01 1.29926038e+00 3.20870459e-01 4.18474138e-01
5.19964397e-01 -1.01590705e+00 4.42881644e-01 8.52784634e-01
2.56698221e-01 5.34094095e-01 1.30976528e-01 -8.10300708e-01
-1.11367357e+00 -1.15721810e+00 5.60522437e-01 -5.55085361e-01
1.37557117e-02 -4.21606749e-01 -9.87672687e-01 4.29813236e-01
-1.04232833e-01 2.43698712e-02 1.17013752e+00 -4.71003383e-01
-2.30061650e-01 -2.06030563e-01 -1.76846325e+00 4.76580113e-01
6.98736310e-01 -7.23880470e-01 -3.10131699e-01 2.57614464e-01
4.75938916e-01 -3.59247327e-01 -1.01966989e+00 2.82759994e-01
6.31126881e-01 -9.09836411e-01 9.56680238e-01 -4.61552233e-01
4.78522569e-01 1.45128310e-01 1.77712813e-02 -7.88990319e-01
1.53109372e-01 -5.87481737e-01 4.04918432e-01 1.07959521e+00
2.95815200e-01 -7.91356206e-01 1.45546222e+00 1.12208343e+00
5.05913138e-01 -6.49262190e-01 -1.32517123e+00 -5.48652768e-01
2.11052388e-01 -1.82317391e-01 8.51707518e-01 1.08275771e+00
-4.64418352e-01 -2.15843976e-01 -3.36303413e-01 7.93218538e-02
8.69479179e-01 4.89622913e-02 8.84034693e-01 -9.67313290e-01
-2.39777006e-02 -2.93554813e-02 -4.22947198e-01 -3.68033767e-01
7.90866911e-02 -1.02847862e+00 -4.21478659e-01 -9.41560507e-01
8.55262950e-02 -7.07568944e-01 -3.99916559e-01 2.85479844e-01
-8.49720091e-02 3.21235180e-01 9.30478722e-02 3.48702848e-01
5.88520393e-02 5.21636307e-01 1.23061347e+00 -1.88272800e-02
-7.33262254e-03 4.83335972e-01 -6.07423067e-01 3.90338004e-01
9.17424262e-01 -1.04119837e+00 -5.27765632e-01 -3.94050777e-01
-2.23246157e-01 2.90105820e-01 4.02585268e-01 -1.12837040e+00
3.19543341e-03 1.29066676e-01 5.94872653e-01 -6.67655766e-01
3.26501906e-01 -1.16689742e+00 3.38781148e-01 8.31119537e-01
-7.13428497e-01 1.53307408e-01 -1.11336194e-01 8.17562819e-01
-1.34916559e-01 -1.14916988e-01 7.74932802e-01 -9.82216448e-02
2.07839340e-01 5.11339188e-01 -3.67847979e-01 -1.07486293e-01
1.27002811e+00 -4.17598158e-01 -1.19041756e-01 -3.86217743e-01
-3.58823985e-01 -2.78793246e-01 5.86267412e-01 3.39917481e-01
8.24816823e-01 -1.16795790e+00 -7.40432143e-01 7.31265604e-01
2.41315410e-01 -7.24568442e-02 3.16653639e-01 4.28189009e-01
-7.39743412e-01 1.02771789e-01 -3.53205800e-01 -6.24595881e-01
-1.43710041e+00 3.42986703e-01 4.36282933e-01 -2.25864813e-01
-7.32755363e-01 7.80250847e-01 -6.75776927e-03 -3.12575191e-01
1.23317473e-01 -2.17845902e-01 4.04899091e-01 -1.96686670e-01
2.71057248e-01 2.29841664e-01 1.57105461e-01 -6.22549415e-01
-1.38710693e-01 -5.01443744e-02 -2.04125196e-01 -1.31028771e-01
1.07485080e+00 -1.17205761e-01 -1.68221388e-02 9.32573155e-02
1.33181632e+00 1.62346914e-01 -1.07380652e+00 -6.32562712e-02
-1.69465244e-01 -8.25896382e-01 -5.64483404e-02 -8.18535089e-01
-1.22551298e+00 8.94229770e-01 1.13398480e+00 2.23422050e-01
1.00481582e+00 -4.13586766e-01 1.03655326e+00 2.44570404e-01
1.81450129e-01 -8.93349648e-01 3.77907157e-02 -4.04003084e-01
7.11702704e-01 -1.26757324e+00 3.23149651e-01 -3.15578729e-01
-7.61653781e-01 7.72765696e-01 3.68209749e-01 1.93937719e-01
6.96009815e-01 -1.37061568e-03 4.54571515e-01 -4.85944152e-02
-1.93819121e-01 5.78339636e-01 1.22782521e-01 7.57579803e-01
9.92476344e-02 3.47321987e-01 -3.46883714e-01 6.09622359e-01
-4.02207255e-01 1.66288957e-01 5.05243540e-01 1.06196821e+00
3.41929525e-01 -1.28907347e+00 -3.79360139e-01 8.02591741e-01
-1.02327645e+00 1.93067729e-01 -3.75441164e-01 5.78661442e-01
1.28875703e-01 6.31173134e-01 -2.55082957e-02 -2.14189604e-01
1.77173451e-01 1.29041478e-01 2.96219885e-01 -3.78847241e-01
-7.91598976e-01 -2.72126079e-01 -1.55677691e-01 -2.73999006e-01
-2.95532674e-01 -5.74688733e-01 -8.04610372e-01 -3.98657709e-01
-3.51393223e-01 -2.36633923e-02 8.71596694e-01 5.52783370e-01
4.62034374e-01 4.22121167e-01 7.69541979e-01 4.58732322e-02
-8.13157141e-01 -2.70083010e-01 -5.26780367e-01 9.06443238e-01
1.97441399e-01 -2.19159707e-01 2.26047244e-02 1.55965820e-01] | [6.160682201385498, 6.856271743774414] |
f0dfd360-ccb4-4f41-971f-3e162d589804 | recurrent-calibration-network-for-irregular | 1812.07145 | null | http://arxiv.org/abs/1812.07145v1 | http://arxiv.org/pdf/1812.07145v1.pdf | Recurrent Calibration Network for Irregular Text Recognition | Scene text recognition has received increased attention in the research
community. Text in the wild often possesses irregular arrangements, typically
including perspective text, curved text, oriented text. Most existing methods
are hard to work well for irregular text, especially for severely distorted
text. In this paper, we propose a novel Recurrent Calibration Network (RCN) for
irregular scene text recognition. The RCN progressively calibrates the
irregular text to boost the recognition performance. By decomposing the
calibration process into multiple steps, the irregular text can be calibrated
to normal one step by step. Besides, in order to avoid the accumulation of lost
information caused by inaccurate transformation, we further design a
fiducial-point refinement structure to keep the integrity of text during the
recurrent process. Instead of the calibrated images, the coordinates of
fiducial points are tracked and refined, which implicitly models the
transformation information. Based on the refined fiducial points, we estimate
the transformation parameters and sample from the original image at each step.
In this way, the original character information is preserved until the final
transformation. Such designs lead to optimal calibration results to boost the
performance of succeeding recognition. Extensive experiments on challenging
datasets demonstrate the superiority of our method, especially on irregular
benchmarks. | ['Hanqing Lu', 'Xiao-Yu Zhang', 'Yunze Gao', 'Yingying Chen', 'Zhen Lei', 'Jinqiao Wang'] | 2018-12-18 | null | null | null | null | ['irregular-text-recognition'] | ['computer-vision'] | [ 3.25181842e-01 -4.78086650e-01 -8.82158056e-02 -2.34272599e-01
-4.52431887e-01 -5.49526572e-01 5.62658131e-01 -1.00608610e-01
-1.79676428e-01 2.71277696e-01 1.73893228e-01 2.88812593e-02
8.96059647e-02 -5.42170584e-01 -6.74248338e-01 -9.84664857e-01
9.55225170e-01 4.42772895e-01 1.49937049e-01 9.31273848e-02
5.95511734e-01 5.35598040e-01 -7.99108207e-01 -1.09928891e-01
9.99299824e-01 8.52887392e-01 1.45572275e-01 3.34610105e-01
-2.22000495e-01 4.88582194e-01 -5.87565720e-01 -4.48327839e-01
2.82494009e-01 -1.05136238e-01 -1.54040009e-01 6.95314884e-01
5.13295114e-01 -4.87505555e-01 -5.61843157e-01 1.20690894e+00
3.62049311e-01 4.96401824e-02 6.88418508e-01 -6.38507485e-01
-5.00272930e-01 5.50450861e-01 -1.13889837e+00 -3.21021944e-01
1.36812344e-01 -7.80893043e-02 6.68435693e-01 -9.80992019e-01
4.02181953e-01 1.09285188e+00 8.17713797e-01 1.71019331e-01
-9.62782323e-01 -6.30861878e-01 2.45336801e-01 -8.94515440e-02
-1.54849279e+00 -4.64826822e-01 9.90952253e-01 -2.66882390e-01
9.17477906e-02 2.25638360e-01 4.82232392e-01 9.14750040e-01
5.37381321e-02 8.13299000e-01 4.11897808e-01 -4.81455863e-01
-2.30429515e-01 -5.60518131e-02 1.76847592e-01 6.67204440e-01
5.73008955e-01 -4.53913003e-01 -2.58826345e-01 1.09406799e-01
1.13819647e+00 3.23637664e-01 -6.27195477e-01 -4.77374822e-01
-1.51536870e+00 4.68812466e-01 3.63207310e-01 3.61673206e-01
-3.64705771e-02 -5.87060340e-02 3.75553876e-01 -1.23925894e-01
1.63022548e-01 1.56111687e-01 1.00669831e-01 1.32368105e-02
-9.39191222e-01 -2.57200301e-01 4.62010890e-01 9.98590589e-01
6.34327948e-01 1.86451837e-01 -9.88509357e-02 1.17117786e+00
2.64709890e-01 7.12780893e-01 5.88444412e-01 -3.10497850e-01
1.05925691e+00 8.28053951e-01 -5.47809945e-03 -1.43201864e+00
-1.14637144e-01 -4.99832064e-01 -1.44627666e+00 -4.68817681e-01
4.66483265e-01 -3.35001163e-02 -8.22431386e-01 9.93993998e-01
3.08885664e-01 2.53578067e-01 -1.34706050e-01 9.22418356e-01
4.29755807e-01 8.28605235e-01 -4.74772930e-01 -1.23435445e-01
1.10131645e+00 -9.63501275e-01 -7.56649375e-01 -1.38063475e-01
5.81900895e-01 -1.14945340e+00 9.18372810e-01 5.06239831e-01
-5.42511344e-01 -5.04636884e-01 -1.20677495e+00 -5.88563047e-02
2.12166518e-01 7.73844838e-01 2.05793992e-01 5.44276953e-01
-4.88033891e-01 2.50473708e-01 -8.07673216e-01 -1.34569213e-01
2.48961657e-01 3.27703148e-01 -1.29850283e-01 -7.87624717e-02
-7.23041892e-01 4.25796270e-01 2.63034284e-01 6.17348552e-01
-1.99849173e-01 -3.22936773e-01 -7.00447619e-01 -8.81486759e-02
4.63605374e-01 -2.91956693e-01 9.70114589e-01 -9.34873879e-01
-1.58747160e+00 3.81568909e-01 -2.77506053e-01 -1.42765850e-01
1.03195822e+00 -1.68948531e-01 -3.11693400e-01 -6.67694509e-02
3.14694047e-02 1.12654626e-01 1.31933296e+00 -1.34918523e+00
-4.66240436e-01 -2.94714987e-01 -5.20467877e-01 4.49927270e-01
-4.44756418e-01 -4.34591860e-01 -1.08919859e+00 -1.03358972e+00
7.29120672e-01 -8.95533860e-01 -1.14917681e-02 1.32889763e-01
-6.58634305e-01 5.88058904e-02 1.23558390e+00 -7.48929799e-01
1.26958406e+00 -2.40071225e+00 8.26767087e-02 5.04153788e-01
1.10673673e-01 2.65619159e-01 1.27124280e-01 1.16361715e-02
-3.62302223e-03 -1.06525473e-01 -2.47033015e-01 -4.28307205e-01
-2.14082673e-01 9.26584974e-02 -5.70226908e-01 7.10394621e-01
-1.83065429e-01 7.47565329e-01 -3.00372928e-01 -6.02233827e-01
4.76315320e-01 5.15943766e-01 -2.12251216e-01 -2.82846624e-03
-8.43258351e-02 3.82617474e-01 -7.68791676e-01 6.12304389e-01
9.33099389e-01 -2.67556727e-01 4.72937785e-02 -6.22972190e-01
-1.29299954e-01 -2.81190932e-01 -1.47987020e+00 1.34595609e+00
-2.50368923e-01 6.49749577e-01 -1.09311398e-02 -8.74316335e-01
1.34826493e+00 8.35585222e-02 4.47992444e-01 -5.02111614e-01
3.67406011e-01 1.58706605e-01 -3.81623417e-01 -3.12277764e-01
7.54586697e-01 4.08960909e-01 1.10731646e-01 3.31335664e-01
-6.80302322e-01 -2.42859259e-01 -1.61643013e-01 1.24928309e-02
2.35096976e-01 1.88607331e-02 1.42188191e-01 8.50746259e-02
9.30671513e-01 -1.51797026e-01 4.78036910e-01 4.84261066e-01
1.30717188e-01 9.44789469e-01 3.28111827e-01 -5.76373935e-01
-1.22105300e+00 -4.94534522e-01 -3.67564082e-01 4.77093667e-01
6.66303217e-01 -3.14559132e-01 -9.29200470e-01 -4.38107878e-01
-2.08855733e-01 2.66034365e-01 -4.67694402e-01 -6.32456839e-02
-8.80043864e-01 -6.88255668e-01 6.89273417e-01 3.61317515e-01
1.02505815e+00 -5.59613705e-01 -2.29206551e-02 1.68166056e-01
-4.28436935e-01 -1.09840155e+00 -1.20498717e+00 -2.40944251e-01
-9.15877104e-01 -8.84173095e-01 -1.01059067e+00 -9.26963508e-01
1.12369835e+00 5.74248254e-01 3.02717000e-01 3.32126766e-01
1.63831234e-01 2.62857266e-02 -2.42536172e-01 1.66458070e-01
-2.50610799e-01 3.71452682e-02 -7.92864040e-02 5.94022036e-01
-7.91815445e-02 -9.56826434e-02 -3.70981902e-01 7.84853280e-01
-9.66972888e-01 2.94303954e-01 5.17944515e-01 9.83632028e-01
7.94844091e-01 1.89812630e-01 -1.91040710e-01 -6.93310380e-01
3.77140075e-01 5.37067242e-02 -8.62937689e-01 5.18342018e-01
-3.12638283e-01 1.63919285e-01 8.57160330e-01 -9.01034355e-01
-1.08122540e+00 2.70717025e-01 9.87498015e-02 -6.59750998e-01
1.60830989e-01 3.73173386e-01 -2.80957907e-01 -4.72000360e-01
4.10259664e-01 7.54833460e-01 -1.82946920e-01 -4.04070586e-01
-2.64446251e-02 8.68737698e-01 6.59569979e-01 -6.36985838e-01
1.01570499e+00 4.15718883e-01 -5.63412905e-02 -1.11174667e+00
-6.54096544e-01 -2.50758260e-01 -7.14665592e-01 -1.77442595e-01
4.81465727e-01 -6.70007050e-01 -7.33804822e-01 9.44639206e-01
-1.14148355e+00 -1.18973620e-01 1.41126484e-01 4.07085359e-01
-1.59039915e-01 1.11638272e+00 -5.00520945e-01 -5.30011892e-01
-3.97565961e-01 -1.30870640e+00 1.13281476e+00 5.13920367e-01
2.31013864e-01 -8.33490670e-01 -1.36659950e-01 3.33899379e-01
1.77489236e-01 -5.72895035e-02 7.00184882e-01 -1.97433844e-01
-7.80890584e-01 -4.09488320e-01 -3.87453586e-01 5.87126613e-02
3.34601730e-01 4.03885871e-01 -5.48864722e-01 -2.02588215e-01
2.02995583e-01 4.80675362e-02 6.73658729e-01 2.45772257e-01
1.26641345e+00 -2.64791310e-01 -3.91236454e-01 1.02167571e+00
1.10606003e+00 2.96112388e-01 6.94741249e-01 4.35814202e-01
1.22251391e+00 1.64380908e-01 5.70459783e-01 4.69050437e-01
2.64901698e-01 7.14989007e-01 6.95786253e-02 -3.35641787e-03
-1.19601432e-02 -4.26657349e-01 1.89070985e-01 1.08111429e+00
8.01446438e-02 -2.25384191e-01 -8.39632928e-01 1.09774448e-01
-1.74206066e+00 -5.80097973e-01 -3.11141759e-01 2.29672766e+00
8.02760482e-01 7.07219318e-02 -3.86792690e-01 2.07877979e-01
1.26511276e+00 2.99408257e-01 -6.55249178e-01 7.76478872e-02
-3.69412750e-01 -4.96412992e-01 5.74749053e-01 4.72690731e-01
-9.11686361e-01 1.02029669e+00 4.93900919e+00 1.14707935e+00
-1.57248104e+00 -4.40219969e-01 6.80169642e-01 4.81326252e-01
-1.38785720e-01 -7.74813220e-02 -1.06607103e+00 4.86780316e-01
-3.13536339e-02 -1.47318549e-03 6.27881408e-01 6.04577541e-01
9.25976336e-02 3.19598280e-02 -7.36953259e-01 1.32313800e+00
2.86920190e-01 -1.19458354e+00 4.83308166e-01 -1.14472397e-01
7.74213493e-01 -2.75007844e-01 1.09404214e-01 -1.27866775e-01
-6.31826371e-02 -6.05071127e-01 7.64087021e-01 5.12738168e-01
1.03422904e+00 -7.22814023e-01 5.24149001e-01 4.32918459e-01
-1.46024430e+00 2.85607308e-01 -6.46195054e-01 6.26454771e-01
-8.12724903e-02 6.42774403e-01 -7.34697580e-01 6.94506347e-01
2.39686668e-01 1.06827748e+00 -7.16381431e-01 9.72718894e-01
-1.11447930e-01 4.07929629e-01 -6.15032673e-01 -1.46778733e-01
2.17411414e-01 -7.54017532e-01 4.64356720e-01 1.07613516e+00
4.12472427e-01 -2.15385687e-02 9.38053727e-02 5.49457014e-01
-2.47715399e-01 3.81568819e-01 -3.85121882e-01 1.43013343e-01
4.22887295e-01 1.11270332e+00 -8.90524089e-01 -2.04427004e-01
-2.32261404e-01 1.07007086e+00 2.35664085e-01 4.23573375e-01
-8.39872479e-01 -4.44802672e-01 -1.34662941e-01 -1.24112569e-01
3.13261360e-01 -4.03152019e-01 -5.87088346e-01 -1.63306928e+00
3.98899198e-01 -1.01375961e+00 2.38518625e-01 -6.60766125e-01
-9.90752816e-01 4.50264961e-01 -4.59843278e-01 -1.51399708e+00
2.83999324e-01 -3.15325081e-01 -6.28881514e-01 7.23056674e-01
-1.15767097e+00 -1.18553591e+00 -7.52836764e-01 7.04785228e-01
6.83057547e-01 -4.02228348e-02 2.97198385e-01 1.88370496e-01
-9.49862361e-01 8.96784842e-01 4.89231437e-01 4.84738082e-01
1.03804183e+00 -6.90333128e-01 3.70680839e-01 9.75393772e-01
1.04563318e-01 5.86127281e-01 2.92685211e-01 -9.38835859e-01
-1.50761354e+00 -1.21542537e+00 3.25215846e-01 -1.61146790e-01
5.50432622e-01 -3.36160302e-01 -1.26454246e+00 6.63681507e-01
-2.87834883e-01 -9.68469605e-02 3.35457809e-02 -2.91530907e-01
-4.45389479e-01 -4.05855387e-01 -7.42318392e-01 9.54256773e-01
5.30505955e-01 -3.44426304e-01 -2.89517105e-01 3.56486440e-01
3.87055695e-01 -8.38757932e-01 -6.25515401e-01 2.63367444e-01
5.40018797e-01 -6.47151470e-01 7.72278070e-01 3.36028308e-01
2.20182002e-01 -6.07741833e-01 -5.40006859e-03 -9.59385097e-01
-1.76694542e-01 -6.80772364e-01 1.52949527e-01 1.39261687e+00
1.60642520e-01 -7.72758365e-01 8.57111275e-01 3.88512999e-01
3.87838632e-02 -4.99251276e-01 -8.16219032e-01 -3.93404365e-01
-2.36891493e-01 -1.39616370e-01 7.07961321e-01 1.03937352e+00
-3.82723421e-01 2.40519553e-01 -5.97632825e-01 3.25936079e-01
6.53812945e-01 1.32350922e-01 9.90701854e-01 -9.54310775e-01
-1.87132042e-02 -4.47352350e-01 -2.54672050e-01 -1.68667185e+00
-6.41557574e-02 -4.80634868e-01 6.65887445e-02 -1.12125361e+00
1.25725910e-01 -8.03489089e-01 2.95632482e-01 2.40942299e-01
-3.37219894e-01 1.04739323e-01 3.19186263e-02 6.95016265e-01
-4.14121211e-01 6.85804904e-01 1.62908912e+00 -3.63443941e-01
-2.75830269e-01 1.24645293e-01 -4.35168326e-01 7.14702070e-01
7.36073196e-01 -2.86878139e-01 -2.06384286e-01 -7.99018979e-01
-7.76199326e-02 1.59362867e-01 -1.96182579e-02 -8.66372883e-01
6.39164686e-01 -1.40185535e-01 6.90267146e-01 -8.81468892e-01
1.45491347e-01 -1.05173945e+00 2.66456842e-01 2.37222508e-01
-3.77354249e-02 -1.53647855e-01 -4.96562235e-02 6.24169827e-01
-2.15070933e-01 -4.25093949e-01 9.56546664e-01 4.07990217e-01
-4.27428633e-02 5.08993626e-01 2.86394674e-02 -2.74748266e-01
9.38713670e-01 -5.66563547e-01 -3.07639629e-01 -3.42643738e-01
-1.40799180e-01 3.04868191e-01 7.57107973e-01 2.44563147e-01
6.41661108e-01 -1.20331466e+00 -5.61393023e-01 4.58193332e-01
-5.21662943e-02 6.01815462e-01 1.64183781e-01 8.58965099e-01
-8.51204455e-01 4.09124911e-01 2.29941353e-01 -9.40414190e-01
-1.14850199e+00 3.79053801e-01 6.21008515e-01 -2.10115388e-01
-9.22313035e-01 2.43000776e-01 3.05704623e-01 -2.88959652e-01
4.44795132e-01 -3.54893357e-01 -2.87927389e-01 -7.00574443e-02
3.84276241e-01 1.91230610e-01 1.06848419e-01 -7.54030466e-01
1.49885463e-02 1.41607463e+00 -4.82495248e-01 4.96086627e-02
1.04312611e+00 -2.72367507e-01 -1.41717866e-01 2.83785760e-01
9.56458569e-01 4.25531298e-01 -1.41774690e+00 -4.57420349e-01
-3.12852949e-01 -6.56784654e-01 -8.98803771e-02 -2.49244466e-01
-1.24731457e+00 7.87320256e-01 2.76881695e-01 -7.63681754e-02
9.13313389e-01 -7.07320631e-01 7.50642478e-01 7.45606661e-01
1.05359100e-01 -8.74406219e-01 9.56994295e-02 6.47316158e-01
7.09425211e-01 -8.44028234e-01 1.72125876e-01 -4.92131293e-01
-6.48305595e-01 1.57434797e+00 6.06954455e-01 -1.35285854e-01
2.00488836e-01 2.52718538e-01 9.85808298e-02 2.53024906e-01
2.20591697e-04 6.08096838e-01 2.70311296e-01 1.90071866e-01
9.68049169e-02 -9.25195292e-02 3.76895294e-02 4.63411212e-01
-7.75240958e-02 -3.70514482e-01 5.84500790e-01 6.66086853e-01
-3.45970899e-01 -9.93600845e-01 -9.12148178e-01 3.86160523e-01
-1.97611988e-01 -5.43767959e-02 -3.58789653e-01 5.25455713e-01
-2.78514713e-01 5.56358635e-01 3.46729606e-02 -2.79903263e-01
4.04784799e-01 -4.34462577e-01 2.47112826e-01 -1.74511641e-01
-4.10131693e-01 5.25291383e-01 -3.56289774e-01 1.74758979e-03
-8.88104364e-02 -6.90537870e-01 -1.27436996e+00 -4.82372433e-01
-8.64225328e-01 1.78878531e-01 5.84172666e-01 8.04786384e-01
-2.76165903e-02 1.56672120e-01 1.05336308e+00 -8.38143826e-01
-4.99794602e-01 -8.20838511e-01 -4.22317386e-01 2.12795377e-01
3.61568540e-01 -4.06544149e-01 -4.30960804e-01 2.88641393e-01] | [12.030805587768555, 2.2143514156341553] |
3e340ada-35a6-4455-80a5-6dde2b5279a9 | sinsy-a-deep-neural-network-based-singing | 2108.02776 | null | https://arxiv.org/abs/2108.02776v1 | https://arxiv.org/pdf/2108.02776v1.pdf | Sinsy: A Deep Neural Network-Based Singing Voice Synthesis System | This paper presents Sinsy, a deep neural network (DNN)-based singing voice synthesis (SVS) system. In recent years, DNNs have been utilized in statistical parametric SVS systems, and DNN-based SVS systems have demonstrated better performance than conventional hidden Markov model-based ones. SVS systems are required to synthesize a singing voice with pitch and timing that strictly follow a given musical score. Additionally, singing expressions that are not described on the musical score, such as vibrato and timing fluctuations, should be reproduced. The proposed system is composed of four modules: a time-lag model, a duration model, an acoustic model, and a vocoder, and singing voices can be synthesized taking these characteristics of singing voices into account. To better model a singing voice, the proposed system incorporates improved approaches to modeling pitch and vibrato and better training criteria into the acoustic model. In addition, we incorporated PeriodNet, a non-autoregressive neural vocoder with robustness for the pitch, into our systems to generate a high-fidelity singing voice waveform. Moreover, we propose automatic pitch correction techniques for DNN-based SVS to synthesize singing voices with correct pitch even if the training data has out-of-tune phrases. Experimental results show our system can synthesize a singing voice with better timing, more natural vibrato, and correct pitch, and it can achieve better mean opinion scores in subjective evaluation tests. | ['Keiichi Tokuda', 'Yoshihiko Nankaku', 'Keiichiro Oura', 'Kei Hashimoto', 'Yukiya Hono'] | 2021-08-05 | null | null | null | null | ['singing-voice-synthesis'] | ['speech'] | [-4.60831255e-01 -4.11529273e-01 1.51367448e-02 7.00012371e-02
-5.05437672e-01 -4.57282066e-01 -1.12738997e-01 -7.13053882e-01
1.33016542e-01 4.45539594e-01 2.09992245e-01 9.78602655e-03
1.92979813e-01 -6.51990294e-01 -3.42412710e-01 -7.35381544e-01
2.56137967e-01 4.87503037e-02 2.45652780e-01 -3.16788018e-01
-2.65106112e-01 3.86160761e-01 -1.75092149e+00 5.53059876e-02
6.88207448e-01 8.14767659e-01 4.17523772e-01 1.14719307e+00
1.31091669e-01 4.91043031e-01 -1.13233495e+00 1.72411084e-01
1.72789529e-01 -8.29024374e-01 -3.95181701e-02 -1.85037196e-01
9.32703018e-02 -4.06735390e-01 -3.42628598e-01 8.96450281e-01
9.93235886e-01 4.75612640e-01 6.80420101e-01 -8.39347541e-01
-7.80055106e-01 7.84269571e-01 1.41398266e-01 -1.05425119e-01
1.51310816e-01 5.35191238e-01 1.04724574e+00 -8.27456057e-01
2.00441763e-01 1.15142322e+00 7.00889707e-01 9.76606905e-01
-9.00159240e-01 -1.01789212e+00 -5.90843260e-01 3.50111425e-01
-1.37253296e+00 -4.32424486e-01 1.09796369e+00 -3.36659163e-01
9.22900796e-01 4.87679482e-01 1.03349793e+00 1.01212215e+00
1.90214776e-02 6.68311715e-01 6.14176452e-01 -4.67292130e-01
6.59482479e-02 -8.07904303e-02 -3.43844771e-01 8.27269107e-02
-5.10989726e-01 5.61165631e-01 -5.91753304e-01 1.10452518e-01
1.05505776e+00 -4.36349899e-01 -2.82971323e-01 4.37435716e-01
-9.44058597e-01 7.12685645e-01 2.82973256e-02 7.00712323e-01
-5.82254708e-01 4.70552742e-02 4.79034841e-01 4.56861518e-02
7.93461502e-03 5.26219547e-01 -2.43932471e-01 -3.58171314e-01
-1.38270903e+00 5.38490295e-01 8.85467231e-01 6.17185354e-01
6.10597916e-02 1.25424838e+00 -4.56845254e-01 1.15600598e+00
2.61826336e-01 8.49372864e-01 8.58141780e-01 -1.25254524e+00
6.37788232e-03 -1.16798989e-01 5.18359207e-02 -8.41381431e-01
-2.25415930e-01 -5.51564932e-01 -8.98995638e-01 1.31590128e-01
2.92700119e-02 -2.56811082e-01 -8.92907321e-01 1.84802783e+00
1.78317860e-01 3.41957211e-01 2.43835703e-01 1.18592608e+00
1.19886339e+00 1.21771157e+00 -3.66966635e-01 -7.53351271e-01
1.08634746e+00 -1.04582214e+00 -1.52652061e+00 3.21806312e-01
-3.85693237e-02 -1.16389978e+00 1.38291109e+00 6.37180924e-01
-1.28164566e+00 -1.22350299e+00 -1.09683526e+00 4.12456878e-02
2.78088868e-01 2.83496678e-01 -5.76984137e-02 7.98903644e-01
-9.54151690e-01 1.06812608e+00 -5.50243676e-01 2.63931483e-01
-5.31214833e-01 1.09027565e-01 3.13731194e-01 9.01197672e-01
-1.65566599e+00 5.35255134e-01 3.48280579e-01 2.41726905e-01
-1.09483421e+00 -6.60213649e-01 -6.82604373e-01 2.49476522e-01
3.56290787e-02 -6.60191536e-01 1.84071636e+00 -7.20388412e-01
-2.31658721e+00 8.52570683e-02 -2.16171414e-01 -2.33711571e-01
3.35353650e-02 -1.03823910e-03 -1.10844529e+00 3.00246924e-02
-2.63748318e-01 1.94745019e-01 1.14858198e+00 -9.86171961e-01
-2.01477140e-01 2.72415698e-01 -6.12721026e-01 1.86396718e-01
-2.71411210e-01 1.40840188e-01 -6.18322864e-02 -9.07746315e-01
3.06253135e-02 -7.87385046e-01 4.97820638e-02 -5.60123563e-01
-5.74670792e-01 -3.16337109e-01 7.87703156e-01 -1.02287579e+00
1.73422110e+00 -2.27359724e+00 1.07588321e-01 6.14674501e-02
-2.12255850e-01 9.02123332e-01 -8.57248977e-02 3.52249891e-01
-1.21669516e-01 1.70866754e-02 -4.51220758e-02 -2.99121618e-01
7.04978630e-02 2.31791228e-01 -5.69283426e-01 -1.78563893e-01
-8.52254219e-03 5.76241016e-01 -6.37205303e-01 -5.12036562e-01
2.93331712e-01 6.50369048e-01 -7.42273390e-01 7.33060956e-01
-3.25594544e-01 6.60142660e-01 1.34392202e-01 4.52482730e-01
2.99700767e-01 6.80818617e-01 -2.04195067e-01 -2.20631838e-01
-3.76098961e-01 4.29164499e-01 -1.53846669e+00 1.21626651e+00
-5.92414677e-01 3.99082005e-01 1.89508915e-01 -3.72068197e-01
1.48114574e+00 1.00281668e+00 2.09100440e-01 -2.28553593e-01
2.29195982e-01 4.75219727e-01 4.59033787e-01 -6.93530798e-01
6.64547503e-01 -6.12543344e-01 3.51931185e-01 1.20633096e-01
2.76670843e-01 -7.54192472e-01 3.91514823e-02 -5.63800693e-01
4.60148424e-01 7.68282488e-02 1.00403182e-01 1.51287362e-01
5.52940547e-01 -3.76042992e-01 1.03651583e+00 1.99344128e-01
-3.91843021e-02 9.55813408e-01 3.47179845e-02 -7.31958970e-02
-1.18601644e+00 -9.56051409e-01 -4.96410467e-02 8.79948378e-01
-3.01527232e-01 -4.18341160e-01 -9.53051031e-01 4.23941404e-01
-3.66414279e-01 9.38334227e-01 -2.01095846e-02 -2.32655898e-01
-8.19524705e-01 -1.14446953e-01 9.93096054e-01 5.52209496e-01
1.90741703e-01 -1.81823170e+00 1.28301263e-01 6.66163921e-01
-3.14095289e-01 -6.58899307e-01 -1.05231893e+00 3.73728536e-02
-7.15852499e-01 -5.05886197e-01 -1.06951654e+00 -9.47622359e-01
-2.06811622e-01 -2.34535709e-01 6.04732037e-01 -1.57679245e-01
-3.22739892e-02 -1.67147189e-01 -2.56678164e-01 -6.10997081e-01
-9.63313758e-01 -1.49271935e-01 7.83979654e-01 -3.41876969e-02
-8.11711550e-02 -9.52830732e-01 -4.91550326e-01 4.06505585e-01
-8.89078796e-01 -2.02372208e-01 2.28460059e-01 8.11347604e-01
8.57329071e-01 1.46049261e-01 8.57616663e-01 -1.79560736e-01
1.12635946e+00 -2.04282850e-02 -4.87573475e-01 -1.70696229e-01
-4.37702775e-01 -4.28148210e-01 1.33468509e+00 -1.03099787e+00
-9.27716792e-01 -2.14120448e-01 -8.30628693e-01 -1.10081530e+00
-2.88219135e-02 4.36085969e-01 -3.74927640e-01 3.90660405e-01
7.52365887e-01 3.95717055e-01 1.44160897e-01 -6.27174020e-01
1.54044405e-01 1.11799908e+00 9.01867986e-01 -3.00665081e-01
9.94906843e-01 -3.44111830e-01 -2.96811461e-01 -1.23825109e+00
-4.27473485e-01 -3.40785682e-01 -4.50492173e-01 -3.10951710e-01
7.47822464e-01 -7.08173573e-01 -9.60272372e-01 7.39087164e-01
-1.45119905e+00 -2.51192898e-01 -6.27800524e-01 1.02177536e+00
-8.96605611e-01 3.76514584e-01 -9.74117696e-01 -1.22023475e+00
-7.85946906e-01 -9.87182200e-01 4.33741301e-01 6.71993077e-01
-4.98086959e-01 -6.32324457e-01 4.17575866e-01 2.36414671e-01
6.11375332e-01 9.93396901e-03 6.67963743e-01 -4.84984994e-01
-8.23459029e-02 1.56204160e-02 6.06093884e-01 1.09679830e+00
2.41923124e-01 6.15607679e-01 -1.18558455e+00 2.00820807e-02
3.34141999e-01 4.62178327e-02 4.51383024e-01 8.11759830e-01
8.99292290e-01 -5.06756842e-01 4.80652422e-01 7.62000799e-01
7.65621841e-01 7.42261291e-01 6.45246804e-01 -3.02556098e-01
6.21164978e-01 4.63694781e-01 6.34386241e-01 4.78538066e-01
-2.90547740e-02 6.72598600e-01 1.91986933e-01 -7.76808113e-02
-6.17726028e-01 -5.35177052e-01 6.37440205e-01 1.98290277e+00
-3.77762049e-01 -2.60593653e-01 -3.13743114e-01 6.64633214e-01
-1.37998438e+00 -1.24765575e+00 -3.53206635e-01 2.14519143e+00
1.19517875e+00 -1.34446606e-01 4.20412928e-01 7.38064110e-01
9.90154862e-01 3.25743288e-01 -7.13044345e-01 -1.03435528e+00
-9.61783603e-02 5.29396117e-01 -9.10590589e-02 4.67078626e-01
-7.03447342e-01 1.06797159e+00 6.39107466e+00 1.25209022e+00
-1.49479389e+00 -1.91515014e-02 -1.14889026e-01 -1.88867167e-01
-4.55906838e-01 -3.51013333e-01 -8.65822971e-01 4.45260704e-01
1.08176887e+00 -3.20498675e-01 9.04108763e-01 9.27442551e-01
8.76727045e-01 6.26002967e-01 -7.65153289e-01 9.98455107e-01
-9.84315872e-02 -1.04878747e+00 4.18134481e-02 -3.54384035e-01
7.31535137e-01 -5.87286115e-01 3.02629620e-01 5.77009201e-01
-3.48031759e-01 -1.21085715e+00 9.66183424e-01 6.58877671e-01
1.04102027e+00 -9.51255500e-01 5.86871505e-01 5.59927285e-01
-1.36707771e+00 2.23593131e-01 -3.33800554e-01 -8.22279155e-02
4.02191162e-01 4.26929414e-01 -1.01476657e+00 4.13250208e-01
4.36735302e-01 3.65679950e-01 3.21405202e-01 1.21840727e+00
-5.35857022e-01 1.21025312e+00 -2.14855567e-01 -3.62030029e-01
-7.62472153e-02 -2.41110027e-02 9.26976383e-01 9.83106911e-01
6.83835030e-01 1.26122057e-01 -1.40312776e-01 1.20321894e+00
2.14896768e-01 5.24508297e-01 -2.79934078e-01 -4.76168841e-01
6.85750186e-01 1.04883575e+00 1.08128786e-01 -1.94352835e-01
1.48917437e-01 5.25744736e-01 -6.47545278e-01 4.48794484e-01
-8.76432776e-01 -7.76153445e-01 6.38036907e-01 1.26509950e-01
2.59819299e-01 -2.69026726e-01 -1.39756843e-01 -7.19336033e-01
-1.36748910e-01 -1.00405943e+00 -1.88913614e-01 -1.01974690e+00
-9.76379395e-01 9.07263160e-01 -4.36006904e-01 -1.54566193e+00
-8.84799302e-01 -2.97662586e-01 -1.07356727e+00 1.33952129e+00
-1.25455379e+00 -8.09341252e-01 1.40571013e-01 4.99736995e-01
6.68408215e-01 -3.66555512e-01 1.10413861e+00 4.20299292e-01
-3.91735852e-01 4.70401138e-01 -4.93114255e-02 -1.32038938e-02
7.75904059e-01 -1.08552134e+00 3.62968922e-01 7.11842954e-01
1.62247509e-01 4.59831357e-01 9.77583230e-01 -5.11492193e-01
-8.92264843e-01 -1.05012524e+00 1.00625920e+00 1.46243468e-01
4.07756060e-01 5.67007028e-02 -1.21366692e+00 8.84783491e-02
1.18314862e-01 -2.40680575e-01 7.10301757e-01 -2.14263409e-01
5.63954189e-02 -2.60488451e-01 -8.33698988e-01 5.26274025e-01
5.14303982e-01 -6.62691772e-01 -8.58911753e-01 5.05354144e-02
1.10472500e+00 -5.91606736e-01 -8.22644532e-01 5.32512546e-01
6.73460603e-01 -9.60265577e-01 7.31071591e-01 -2.73555338e-01
2.79555619e-01 -6.25691712e-01 1.50228972e-02 -1.63633180e+00
-3.31966221e-01 -1.15356100e+00 -1.81886166e-01 1.44931293e+00
3.40247631e-01 -9.19963717e-02 4.34502035e-01 -3.62516660e-03
-5.33479095e-01 -4.13520247e-01 -7.90909708e-01 -1.26885021e+00
7.44581670e-02 -6.31534278e-01 7.53197849e-01 6.69146299e-01
-1.72673538e-01 4.11931455e-01 -9.48726773e-01 2.99791187e-01
2.37927169e-01 2.22030766e-02 6.83651209e-01 -1.09828532e+00
-5.72918236e-01 -4.02343869e-01 9.45796221e-02 -9.27492499e-01
-4.34633382e-02 -5.06224096e-01 4.45036650e-01 -1.39328671e+00
-3.90364826e-01 -2.34827842e-03 -2.79936910e-01 2.07006648e-01
-1.70781121e-01 9.74649638e-02 4.73145247e-01 1.06633596e-01
1.95011243e-01 9.43305433e-01 1.70822871e+00 -4.28825915e-02
-8.63472342e-01 6.65457487e-01 7.83325434e-02 8.75668406e-01
8.14118624e-01 -4.50038403e-01 -3.18465620e-01 5.65491952e-02
-3.35477054e-01 7.65858650e-01 -7.54411221e-02 -1.30447745e+00
3.23182374e-01 -1.24598481e-01 1.73046011e-02 -9.18288112e-01
8.09214354e-01 -4.99990135e-01 4.70515400e-01 5.05271554e-01
-2.27735966e-01 -3.63668948e-01 2.62219727e-01 -3.33476290e-02
-5.71888685e-01 -3.88153464e-01 9.09459710e-01 1.09005839e-01
-1.50450990e-01 2.53359884e-01 -5.58421373e-01 -1.97141141e-01
5.89412808e-01 -1.84963182e-01 1.88395098e-01 -7.00837731e-01
-1.01218653e+00 -2.82632023e-01 -1.30867958e-01 4.41534787e-01
7.57491112e-01 -1.68686152e+00 -7.61584401e-01 3.46598297e-01
-3.31756294e-01 -6.40883371e-02 7.28901982e-01 4.61908102e-01
-4.72255200e-01 5.04064083e-01 -5.67119494e-02 -3.98186147e-01
-1.39744246e+00 3.76787990e-01 5.05220115e-01 7.52320588e-02
-2.93965340e-01 7.83800662e-01 8.15012604e-02 -6.47044301e-01
4.98440474e-01 -5.12588978e-01 -6.01744175e-01 -2.73029916e-02
4.54211086e-01 4.68158871e-01 -9.35736373e-02 -6.79260969e-01
1.20772775e-02 6.62883282e-01 6.36146903e-01 -3.58121663e-01
1.02895939e+00 1.87558919e-01 9.90203544e-02 9.02229905e-01
7.76307344e-01 5.16270399e-01 -8.97070050e-01 -5.63969277e-02
-5.55969656e-01 -1.31756037e-01 2.46169090e-01 -8.02330434e-01
-9.63878274e-01 1.03893733e+00 3.21363032e-01 3.57005954e-01
1.09626293e+00 -5.55295646e-01 1.57034934e+00 2.66585685e-02
-1.22726513e-02 -1.28067148e+00 7.64783993e-02 7.87038088e-01
1.15110004e+00 -6.35009229e-01 -6.30342126e-01 -1.09471202e-01
-8.03063095e-01 1.35923159e+00 5.71507812e-01 -2.32506856e-01
6.13955796e-01 2.44427308e-01 4.51144636e-01 5.34334600e-01
-7.36914575e-01 -1.16846666e-01 5.37350118e-01 5.87544262e-01
3.48173141e-01 3.06905359e-01 -3.80466223e-01 1.33687425e+00
-9.38558578e-01 9.06017236e-03 4.45393533e-01 -8.84694159e-02
-6.06680334e-01 -1.06553471e+00 -7.86269784e-01 -1.00544631e-01
-5.91433942e-01 -3.02239865e-01 -1.79491639e-01 3.69812936e-01
3.21837693e-01 1.09755838e+00 -2.26011276e-02 -9.83493507e-01
7.17312217e-01 2.52859056e-01 1.42965481e-01 -7.37200916e-01
-7.96752632e-01 7.54369974e-01 -8.92457962e-02 -1.15743697e-01
2.33725701e-02 -3.58636349e-01 -1.47716665e+00 -1.95789948e-01
-5.51290154e-01 4.37894374e-01 8.91061902e-01 6.07548952e-01
-5.52229322e-02 1.05439079e+00 1.13327360e+00 -7.92650819e-01
-7.40948856e-01 -1.29881871e+00 -1.20521748e+00 -8.74505788e-02
3.68101925e-01 -1.24979220e-01 -7.24174857e-01 6.84387535e-02] | [15.509612083435059, 6.225228786468506] |
a40d6181-0e4b-495d-b863-1268fb53a3c3 | modeling-multimodal-dynamic-spatiotemporal | 1810.05993 | null | https://arxiv.org/abs/1810.05993v3 | https://arxiv.org/pdf/1810.05993v3.pdf | The Trajectron: Probabilistic Multi-Agent Trajectory Modeling With Dynamic Spatiotemporal Graphs | Developing safe human-robot interaction systems is a necessary step towards the widespread integration of autonomous agents in society. A key component of such systems is the ability to reason about the many potential futures (e.g. trajectories) of other agents in the scene. Towards this end, we present the Trajectron, a graph-structured model that predicts many potential future trajectories of multiple agents simultaneously in both highly dynamic and multimodal scenarios (i.e. where the number of agents in the scene is time-varying and there are many possible highly-distinct futures for each agent). It combines tools from recurrent sequence modeling and variational deep generative modeling to produce a distribution of future trajectories for each agent in a scene. We demonstrate the performance of our model on several datasets, obtaining state-of-the-art results on standard trajectory prediction metrics as well as introducing a new metric for comparing models that output distributions. | ['Marco Pavone', 'Boris Ivanovic'] | 2018-10-14 | the-trajectron-probabilistic-multi-agent | http://openaccess.thecvf.com/content_ICCV_2019/html/Ivanovic_The_Trajectron_Probabilistic_Multi-Agent_Trajectory_Modeling_With_Dynamic_Spatiotemporal_Graphs_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Ivanovic_The_Trajectron_Probabilistic_Multi-Agent_Trajectory_Modeling_With_Dynamic_Spatiotemporal_Graphs_ICCV_2019_paper.pdf | iccv-2019-10 | ['trajectory-modeling'] | ['time-series'] | [-2.96518475e-01 1.41943857e-01 1.68368012e-01 -1.00626044e-01
-4.20116305e-01 -7.81433105e-01 1.49149334e+00 6.01547072e-03
-1.84690595e-01 6.33103967e-01 6.93068385e-01 -1.74560308e-01
-5.92162535e-02 -8.35678577e-01 -5.39074659e-01 -6.60069048e-01
-4.48312819e-01 1.22039247e+00 3.07472199e-01 -4.36358631e-01
-2.27527827e-01 6.30486310e-01 -1.43384361e+00 -2.93524191e-02
3.03590208e-01 2.03847602e-01 3.77612710e-01 1.16150832e+00
3.43145072e-01 1.23245943e+00 -3.68511319e-01 -6.58833683e-01
-3.28734182e-02 -4.53905404e-01 -6.67035997e-01 1.81461021e-01
-1.71934754e-01 -3.15320909e-01 -1.01114511e+00 8.73941481e-01
-6.93796277e-02 5.37922919e-01 1.08301628e+00 -1.77552152e+00
-4.99098390e-01 6.75173163e-01 -3.47717315e-01 -4.30819578e-02
4.90085989e-01 6.17184043e-01 7.90688097e-01 -5.74655116e-01
1.04835272e+00 1.53171635e+00 3.53070080e-01 4.71595883e-01
-1.29469860e+00 5.21053486e-02 2.35363960e-01 4.88978475e-01
-1.09431899e+00 -4.15628672e-01 4.34196591e-01 -7.49262154e-01
1.19899571e+00 1.08240999e-01 8.27332914e-01 1.47094405e+00
6.67632222e-01 8.31617236e-01 3.33899647e-01 1.17773235e-01
3.53070408e-01 -2.96903759e-01 3.15890647e-03 6.30980492e-01
-1.90890908e-01 1.39956549e-01 -4.38222170e-01 -3.15831989e-01
5.00641942e-01 2.10419506e-01 2.46512905e-01 -3.82773817e-01
-1.66730952e+00 6.96955502e-01 2.54194647e-01 4.41744328e-02
-6.63475275e-01 6.02804959e-01 3.36735137e-02 -2.55954806e-02
1.44625917e-01 1.67451590e-01 -9.11342949e-02 -6.51298761e-01
-5.37314892e-01 9.83270824e-01 9.71635520e-01 1.10867488e+00
5.91127872e-01 -1.82067633e-01 -3.19837332e-01 2.87595749e-01
5.71767390e-01 6.24281287e-01 -1.83887985e-02 -1.41140187e+00
3.68243635e-01 5.44867516e-01 6.09769523e-01 -8.34671974e-01
-6.23147070e-01 -6.77521676e-02 -6.02061033e-01 3.32501292e-01
5.97433746e-01 -6.81654513e-01 -8.27513993e-01 2.03958678e+00
3.84455979e-01 4.00186539e-01 2.63851643e-01 5.92559874e-01
4.20163363e-01 1.17410874e+00 1.83591262e-01 -2.73025393e-01
8.91709447e-01 -1.09374440e+00 -5.25994956e-01 -2.22971246e-01
4.71258849e-01 -4.58716542e-01 6.21331513e-01 8.62478688e-02
-1.12808645e+00 -2.22652450e-01 -4.72682416e-01 7.63549749e-03
-2.75107384e-01 -2.81131506e-01 6.53223455e-01 9.89341736e-02
-1.26941407e+00 3.78003269e-01 -1.40555656e+00 -6.05775714e-01
3.43792178e-02 1.15912810e-01 -2.00920209e-01 6.86979890e-02
-7.47090220e-01 1.11054599e+00 2.35912472e-01 -9.62175503e-02
-1.37908888e+00 -4.09192413e-01 -7.66491652e-01 2.15916652e-02
2.73970604e-01 -8.86625171e-01 1.58840525e+00 -2.89068222e-01
-1.34500468e+00 3.35968822e-01 -4.94707555e-01 -6.08090222e-01
8.39292407e-01 -1.84628535e-02 -4.77456540e-01 -2.66412884e-01
6.11794516e-02 9.24370944e-01 5.43141484e-01 -1.31370068e+00
-9.70691502e-01 -2.48008355e-01 2.11048983e-02 5.41395128e-01
3.22599620e-01 -7.37404004e-02 -3.22881401e-01 -2.63485491e-01
-4.68303710e-01 -1.45132840e+00 -6.38822377e-01 -2.11800575e-01
-5.91832876e-01 -5.93439043e-01 8.09267759e-01 -4.02077317e-01
7.61594176e-01 -1.70552325e+00 5.93155682e-01 -9.00812540e-03
3.64767015e-01 -1.13162056e-01 -2.22877041e-01 1.13174582e+00
4.25487041e-01 -1.57132506e-01 -4.88713197e-02 -7.66989529e-01
4.58745033e-01 2.42767841e-01 -4.34397548e-01 5.21531522e-01
1.96287539e-02 1.16580653e+00 -1.16436148e+00 -1.15489706e-01
4.18738395e-01 5.45383871e-01 -3.09804410e-01 1.92311555e-01
-8.25132787e-01 7.10330129e-01 -3.52637470e-01 -1.79087669e-02
2.32228532e-01 -3.89798582e-01 1.66628107e-01 5.64117610e-01
-1.79459527e-01 5.62503561e-02 -7.84628510e-01 1.62828410e+00
-2.09386256e-02 8.26193631e-01 -3.24497461e-01 -1.03791498e-01
4.82291788e-01 3.53225797e-01 5.34435213e-01 -2.41212428e-01
7.17942864e-02 -1.31696507e-01 1.37670040e-01 -4.62959588e-01
8.90396595e-01 -4.23478037e-02 -2.94346094e-01 6.94081068e-01
-1.81204364e-01 5.08723184e-02 5.13576627e-01 5.23367941e-01
1.26546264e+00 2.01672092e-01 1.11406989e-01 5.79834394e-02
6.00514710e-02 3.25042278e-01 3.44877630e-01 8.45355988e-01
-3.62525851e-01 3.36616784e-01 6.55817807e-01 -5.58151841e-01
-1.28482151e+00 -1.34777260e+00 4.30535048e-01 9.06489432e-01
4.40737128e-01 -4.45312381e-01 -5.98430037e-01 -2.31920317e-01
-1.54290989e-01 1.19606173e+00 -6.44016743e-01 -2.12769080e-02
-5.84445834e-01 -5.90427816e-01 1.61599994e-01 3.27078760e-01
8.99355859e-03 -1.28937471e+00 -9.58147585e-01 3.94377649e-01
-3.73375326e-01 -1.13919914e+00 -3.30424577e-01 -4.65338171e-01
-3.34480613e-01 -8.40207994e-01 -6.15379870e-01 -6.10577464e-01
2.44631544e-01 1.86620772e-01 1.13416982e+00 -4.13769841e-01
-2.66064014e-02 7.02180743e-01 -1.52664527e-01 -2.65956610e-01
-8.86308551e-01 -6.11382574e-02 2.08516911e-01 -1.85729004e-02
1.70449138e-01 -5.06684124e-01 -5.22062719e-01 -9.78614576e-03
-5.19088387e-01 4.21111226e-01 1.02208003e-01 3.67845058e-01
2.78075695e-01 3.12624782e-01 4.05151039e-01 -3.72563869e-01
7.84346104e-01 -1.16647935e+00 -5.49251676e-01 3.73488128e-01
-9.72887948e-02 -2.66715158e-02 5.86246610e-01 -5.32184184e-01
-1.26845407e+00 8.74524117e-02 3.77302840e-02 -2.67968416e-01
-6.28032029e-01 4.82324600e-01 1.37423307e-01 7.27715611e-01
4.68763143e-01 2.91271687e-01 1.07381240e-01 6.42449036e-02
9.58157718e-01 1.85803413e-01 6.78576946e-01 -3.13531607e-01
7.58027792e-01 6.98922276e-01 2.65841186e-01 -1.06302273e+00
-5.87785207e-02 -1.61484718e-01 -5.94821215e-01 -6.82127714e-01
9.81125236e-01 -8.04880619e-01 -1.13448858e+00 7.41791010e-01
-1.46824992e+00 -8.73875797e-01 -1.91990361e-01 4.76623744e-01
-8.76124740e-01 -7.74886236e-02 -6.20572031e-01 -1.20676219e+00
2.70400107e-01 -1.27937031e+00 1.03862286e+00 4.29927558e-01
-6.36080384e-01 -1.32945693e+00 6.04618371e-01 -1.19026177e-01
1.56920180e-01 4.75542516e-01 7.22939610e-01 -6.40061498e-01
-9.31929886e-01 -3.10824122e-02 2.79176891e-01 -6.87606931e-01
-1.00207493e-01 3.22882265e-01 -5.49880862e-01 -3.08691174e-01
-2.72597104e-01 -8.14422145e-02 5.59149027e-01 6.59350216e-01
2.19978318e-01 -4.40273970e-01 -8.52348030e-01 9.25034136e-02
1.00674200e+00 5.31214714e-01 6.10754192e-01 1.35410586e-02
6.93914235e-01 9.38004136e-01 4.04390454e-01 6.46381378e-01
1.17672527e+00 7.19314516e-01 5.43670058e-01 4.23925698e-01
7.09339902e-02 -4.62444693e-01 5.74500144e-01 2.99690902e-01
6.69847131e-02 -9.51626956e-01 -1.17352569e+00 9.62812364e-01
-2.50168920e+00 -1.43760753e+00 -3.83962601e-01 1.79753590e+00
5.82472757e-02 -1.84387162e-01 6.91763878e-01 -7.25106239e-01
6.03194952e-01 2.40739018e-01 -9.35163736e-01 -1.40695021e-01
6.49065301e-02 -6.95246875e-01 1.34974301e-01 8.28719676e-01
-9.77288604e-01 1.04385626e+00 6.69053602e+00 4.97865736e-01
-5.49408019e-01 -9.22690555e-02 6.99402809e-01 -3.29832971e-01
-4.73316848e-01 -5.61659485e-02 -7.13273466e-01 4.59310949e-01
1.20316160e+00 -4.29651201e-01 8.01245570e-01 5.49674213e-01
4.44484144e-01 -8.87951180e-02 -1.34102535e+00 8.59655380e-01
-1.31902188e-01 -1.58825672e+00 -7.14364424e-02 3.83176684e-01
8.56640279e-01 5.16167462e-01 2.71842837e-01 1.08907819e-01
1.24553597e+00 -1.11503875e+00 1.13877356e+00 9.33108330e-01
9.28455070e-02 -8.33365440e-01 2.58303553e-01 8.44219208e-01
-1.12763631e+00 -1.88583210e-01 1.90069377e-02 -1.64823368e-01
8.51448417e-01 -2.86251009e-02 -1.00005102e+00 3.17937225e-01
4.21422452e-01 9.87237513e-01 -2.28329986e-01 9.55911219e-01
-9.62025672e-02 1.88497916e-01 -3.54028940e-01 -3.34818423e-01
3.86031300e-01 -5.34039736e-01 1.09048748e+00 1.01529300e+00
5.59359074e-01 2.00017869e-01 1.13687083e-01 1.20056152e+00
1.70740187e-01 -5.98057806e-01 -1.17838454e+00 -1.58900946e-01
6.72311664e-01 1.06014895e+00 -7.54225910e-01 -1.53693497e-01
-2.73799747e-01 8.94752383e-01 3.95666122e-01 6.46480560e-01
-1.01546264e+00 1.74879819e-01 1.09280229e+00 -1.90006763e-01
1.34821147e-01 -7.95873821e-01 3.62215526e-02 -1.04209292e+00
-2.41043746e-01 -6.12975597e-01 2.96884120e-01 -8.43409836e-01
-1.26677036e+00 6.18901730e-01 1.28557846e-01 -1.10287547e+00
-1.01424980e+00 -2.92665988e-01 -9.14999187e-01 6.47299647e-01
-8.30211222e-01 -1.66764808e+00 1.50176585e-01 6.01681530e-01
6.72159672e-01 -4.21423316e-01 6.40490055e-01 -4.22225922e-01
-4.69020456e-01 -1.73022449e-01 1.60555407e-01 -2.66695321e-01
-9.04034153e-02 -1.30730832e+00 1.12726521e+00 1.13477528e+00
4.95926201e-01 5.22490025e-01 1.01767910e+00 -8.99047673e-01
-1.56802356e+00 -1.06932747e+00 9.81508732e-01 -1.02232516e+00
9.38035667e-01 -3.94826829e-01 -4.98725295e-01 1.31622767e+00
4.83960748e-01 -6.61818504e-01 3.91462028e-01 -9.64722782e-02
2.24133745e-01 5.49784362e-01 -7.28803933e-01 1.43953848e+00
1.12246168e+00 -3.31904501e-01 -4.28864837e-01 2.99533218e-01
8.31481934e-01 -4.04691845e-01 -4.74235058e-01 -4.96229827e-02
6.27784073e-01 -1.20747423e+00 9.85429645e-01 -5.95180333e-01
4.53651041e-01 -3.55345577e-01 -1.84097588e-01 -1.53791225e+00
-5.09476662e-01 -1.24299490e+00 -4.83480722e-01 9.31266010e-01
6.07765436e-01 -4.91668671e-01 8.19289744e-01 1.23773038e+00
-9.29256901e-02 -5.06277025e-01 -7.82836139e-01 -6.23471022e-01
1.24119163e-01 -5.93818545e-01 8.34867537e-01 2.91755199e-01
2.08658025e-01 4.23328519e-01 -6.06973410e-01 3.74527782e-01
6.75139129e-01 -5.80143277e-03 1.06098461e+00 -1.08013284e+00
-1.60004362e-01 -6.54090464e-01 -3.98258090e-01 -1.23622680e+00
3.91039938e-01 -6.49938047e-01 1.93630159e-01 -1.97480357e+00
2.78308839e-01 -1.79908052e-01 3.84495676e-01 4.07890454e-02
-1.44734057e-02 -3.00049931e-01 4.96771783e-01 4.26449269e-01
-8.56205881e-01 9.82235253e-01 1.14720023e+00 1.24210480e-03
-4.33522016e-01 1.43931031e-01 -2.76091129e-01 9.35865641e-01
6.76869571e-01 -3.49164426e-01 -7.98147440e-01 -4.00228500e-01
3.19117218e-01 5.65020978e-01 5.77309787e-01 -9.57780123e-01
5.89716852e-01 -6.46310985e-01 1.24621853e-01 -8.55733275e-01
7.98996031e-01 -7.35369265e-01 7.76422918e-01 4.96769190e-01
-4.28128034e-01 4.27503794e-01 -6.75270781e-02 1.04495692e+00
2.70587653e-01 2.44785562e-01 3.39468002e-01 -1.33832648e-01
-9.51357484e-01 6.43691599e-01 -8.89971495e-01 -6.75809532e-02
1.50762808e+00 1.43410275e-02 -6.42378390e-01 -1.04437649e+00
-8.53188038e-01 6.83684826e-01 7.13896036e-01 7.08666384e-01
7.02187717e-01 -1.28593171e+00 -8.17555487e-01 -8.06781501e-02
8.29371810e-02 -1.41516373e-01 3.57064813e-01 5.15914500e-01
-5.39766371e-01 2.58281469e-01 -1.85180455e-01 -4.05482501e-01
-9.50651824e-01 6.34387493e-01 3.21050137e-01 -3.01319242e-01
-8.51188719e-01 6.19697154e-01 2.92364269e-01 -4.90444839e-01
1.51406109e-01 -1.03248872e-01 -3.55597854e-01 -1.94318190e-01
5.97605944e-01 7.96128571e-01 -8.27573657e-01 -1.18436873e+00
-2.64751762e-01 1.02577463e-01 1.01420134e-01 -8.58756244e-01
1.40421486e+00 -4.40705597e-01 -2.23155454e-01 8.96468103e-01
7.74400413e-01 -4.00941551e-01 -1.77032924e+00 1.62765067e-02
-2.13289540e-02 -1.93694696e-01 -4.88104731e-01 -6.78452492e-01
-4.35326338e-01 6.36416793e-01 1.05347998e-01 6.18399024e-01
3.82359028e-01 4.32276845e-01 9.62353945e-01 3.77174854e-01
8.14277291e-01 -6.07165694e-01 3.09648048e-02 7.78393805e-01
9.18243051e-01 -1.03651035e+00 -4.58431184e-01 1.78661719e-02
-1.22679925e+00 9.26592767e-01 3.08975697e-01 -4.80923578e-02
6.65404081e-01 6.71591237e-02 -2.69193798e-02 -4.15438354e-01
-1.28289115e+00 -2.18417913e-01 1.03425905e-01 9.62619483e-01
-1.72694679e-02 4.95775461e-01 4.35777009e-01 1.26210928e-01
-4.05473500e-01 -2.80665100e-01 7.96573222e-01 5.38422406e-01
-2.53544301e-01 -9.21892524e-01 -1.82676706e-02 2.44522035e-01
7.67837614e-02 3.78402203e-01 -3.64665002e-01 5.21888793e-01
-2.15759769e-01 1.43218136e+00 2.17760712e-01 -4.19374764e-01
-5.08828461e-03 -1.11418590e-01 3.93903404e-01 -5.02104521e-01
-1.72951803e-01 -8.44864324e-02 1.35432571e-01 -5.47774017e-01
-2.72158682e-01 -1.22954714e+00 -1.43109655e+00 -7.62306452e-01
3.55116904e-01 -2.30756074e-01 3.05868864e-01 9.04787540e-01
5.51555216e-01 3.52389842e-01 3.06840539e-01 -1.20832753e+00
-1.58717036e-01 -7.31392801e-01 -5.29413104e-01 4.10551220e-01
5.93701720e-01 -5.09465992e-01 5.82847595e-02 3.56313705e-01] | [5.866784572601318, 0.7998443841934204] |
2c879843-373f-40c5-a8dc-5fc15c7275f5 | generative-table-pre-training-empowers-models | 2305.09696 | null | https://arxiv.org/abs/2305.09696v1 | https://arxiv.org/pdf/2305.09696v1.pdf | Generative Table Pre-training Empowers Models for Tabular Prediction | Recently, the topic of table pre-training has attracted considerable research interest. However, how to employ table pre-training to boost the performance of tabular prediction remains an open challenge. In this paper, we propose TapTap, the first attempt that leverages table pre-training to empower models for tabular prediction. After pre-training on a large corpus of real-world tabular data, TapTap can generate high-quality synthetic tables to support various applications on tabular data, including privacy protection, low resource regime, missing value imputation, and imbalanced classification. Extensive experiments on 12 datasets demonstrate that TapTap outperforms a total of 16 baselines in different scenarios. Meanwhile, it can be easily combined with various backbone models, including LightGBM, Multilayer Perceptron (MLP) and Transformer. Moreover, with the aid of table pre-training, models trained using synthetic data generated by TapTap can even compete with models using the original dataset on half of the experimental datasets, marking a milestone in the development of synthetic tabular data generation. The codes are available at https://github.com/ZhangTP1996/TapTap. | ['Qian Liu', 'Jian Li', 'Shuicheng Yan', 'Shaowen Wang', 'Tianping Zhang'] | 2023-05-16 | null | null | null | null | ['imputation', 'imbalanced-classification', 'imputation', 'imputation'] | ['computer-vision', 'miscellaneous', 'miscellaneous', 'time-series'] | [ 1.85733791e-02 4.60736305e-01 -7.03360140e-01 -4.53855038e-01
-1.16873634e+00 -6.01194620e-01 3.49120528e-01 4.40783679e-01
1.53512537e-01 1.31320369e+00 2.00049862e-01 -5.44038534e-01
2.75265843e-01 -9.75357890e-01 -1.19110811e+00 -4.12891269e-01
2.72536725e-01 9.83119130e-01 -3.01373899e-01 -2.51485020e-01
1.25056386e-01 9.13628340e-02 -1.29417384e+00 1.05364501e+00
9.22135115e-01 1.28253460e+00 -5.50989747e-01 8.44498649e-02
-1.66104421e-01 1.16133738e+00 -5.53050637e-01 -1.13420999e+00
5.22882879e-01 -1.33727059e-01 -5.38340747e-01 -2.96428710e-01
5.16211152e-01 -3.30007195e-01 -3.66286457e-01 6.80569589e-01
5.29993415e-01 -2.74731994e-01 2.94717044e-01 -1.62503684e+00
-5.56951582e-01 1.17172027e+00 -3.90899360e-01 -4.51846719e-01
1.26753375e-01 1.16877317e-01 1.31684279e+00 -9.33557510e-01
7.44520843e-01 1.21970284e+00 8.64480555e-01 3.59405160e-01
-1.48257744e+00 -1.02014291e+00 -1.82939842e-01 3.52933675e-01
-9.26339865e-01 -6.48048699e-01 5.87077796e-01 -2.59228110e-01
6.83041930e-01 5.62736392e-01 2.74032265e-01 1.20394206e+00
1.76716194e-01 1.26456964e+00 1.05822325e+00 -1.30796611e-01
1.16636291e-01 5.88268578e-01 -1.12167493e-01 2.98570007e-01
6.45878911e-01 -2.09624112e-01 -8.70219886e-01 -4.42099273e-01
4.88920249e-02 -2.03009561e-01 -1.24070823e-01 -7.84180462e-01
-1.34367681e+00 7.50285089e-01 2.39796653e-01 -3.63019556e-01
-2.13152722e-01 -1.45069525e-01 8.31220269e-01 4.21534419e-01
5.18133223e-01 7.94519484e-01 -7.80313849e-01 -8.51608142e-02
-8.38302553e-01 4.81119394e-01 9.31871355e-01 1.17370212e+00
6.22100472e-01 -1.73206046e-01 -5.61075330e-01 7.32056499e-01
-6.16954714e-02 7.07323670e-01 1.40705898e-01 -8.22331429e-01
1.26869071e+00 8.40925515e-01 1.10427275e-01 -7.94378102e-01
-2.02115372e-01 -2.25863755e-01 -1.21406305e+00 -2.30013877e-01
6.67658150e-01 -9.63902026e-02 -5.89904130e-01 1.29657137e+00
4.25083697e-01 -5.05719006e-01 1.26724198e-01 3.86726797e-01
6.79919422e-01 6.88605726e-01 -3.35519373e-01 -7.42167383e-02
1.23001003e+00 -1.01145267e+00 -7.25646853e-01 -7.62387738e-02
9.00805116e-01 -5.35408735e-01 1.02826679e+00 5.81908703e-01
-1.06359780e+00 -2.06956252e-01 -9.01084661e-01 -1.74211934e-01
-5.96826375e-01 -2.82876361e-02 6.70506477e-01 9.19131458e-01
-7.16763377e-01 3.02865922e-01 -6.10017180e-01 5.60716167e-02
7.54179358e-01 1.68189168e-01 -5.92167616e-01 -2.03174129e-01
-1.44051659e+00 5.55576980e-01 6.58072710e-01 9.93469208e-02
-4.11329210e-01 -1.03183079e+00 -8.08941424e-01 1.43152162e-01
6.41002178e-01 -6.28719032e-01 9.94370461e-01 -2.39309743e-01
-1.03395927e+00 6.45555198e-01 -1.45950347e-01 -8.35136712e-01
8.65478635e-01 -1.55417264e-01 -1.08820394e-01 -3.59160691e-01
1.18807927e-01 6.35607243e-01 4.12893116e-01 -1.28095376e+00
-5.24581015e-01 -2.85669118e-01 -2.58494765e-01 -3.43951397e-02
-3.04250687e-01 -3.23418558e-01 -4.04014498e-01 -5.71281493e-01
-1.36488929e-01 -8.66585493e-01 -2.09303364e-01 -1.69717923e-01
-1.11247420e+00 1.35587761e-03 3.43536556e-01 -7.37330317e-01
1.22109485e+00 -1.84271419e+00 -2.16329098e-01 3.49029243e-01
2.19886214e-01 1.75445363e-01 1.06132016e-01 8.32051396e-01
-9.69372690e-02 3.96786004e-01 -2.87272006e-01 -5.23412347e-01
3.67264181e-01 -9.94448140e-02 -6.85691535e-01 1.69505730e-01
1.20582934e-02 1.12135363e+00 -3.63018125e-01 -5.67750037e-01
-6.41532019e-02 8.58065560e-02 -8.59576106e-01 9.90793407e-02
-6.44533694e-01 2.93725342e-01 -1.30836368e-01 1.00701153e+00
9.03345048e-01 -4.45635438e-01 3.87643427e-01 3.95286418e-02
3.17248583e-01 6.35542512e-01 -1.11380816e+00 1.17780054e+00
-4.26109992e-02 4.31198627e-01 -3.28373253e-01 -7.71183729e-01
1.10036254e+00 2.34560087e-01 4.75850672e-01 -1.11993992e+00
-1.61039472e-01 2.26915479e-01 -3.93714488e-01 -8.78526345e-02
8.12663794e-01 2.81558543e-01 -3.41932654e-01 6.62302732e-01
-4.81314570e-01 2.34379128e-01 4.08977926e-01 1.68355390e-01
9.70156372e-01 -8.03270116e-02 4.90446016e-02 1.29496947e-01
5.72724879e-01 1.91775039e-01 7.30272174e-01 9.22920942e-01
4.09487486e-02 6.78135753e-01 1.02089310e+00 -7.71693766e-01
-1.27550173e+00 -8.74660730e-01 -3.46214384e-01 9.01412845e-01
-1.92927510e-01 -7.04909980e-01 -7.08700061e-01 -9.30734992e-01
4.52384561e-01 8.32200170e-01 -7.44844198e-01 -6.86912192e-03
-6.28987730e-01 -1.01221001e+00 6.93508446e-01 3.70951533e-01
5.18203437e-01 -1.42633474e+00 -7.73154944e-02 1.57187521e-01
-6.23941660e-01 -1.16264451e+00 -2.62511909e-01 3.89585346e-01
-7.49824882e-01 -9.19705093e-01 1.02796231e-03 -3.38431060e-01
5.82304120e-01 -1.03300862e-01 1.54754150e+00 -4.13391627e-02
4.04135659e-02 -4.87401903e-01 -1.51682571e-01 -3.91756207e-01
-5.51064610e-01 8.33274186e-01 -1.69780552e-01 1.34335933e-02
2.82459199e-01 -5.15448451e-01 -3.24962497e-01 4.81162935e-01
-7.22009182e-01 4.85490322e-01 6.32246554e-01 1.06688523e+00
6.68562472e-01 -1.26834542e-01 6.23614728e-01 -1.60860229e+00
3.39394808e-01 -6.16791129e-01 -8.49701703e-01 3.86124223e-01
-9.11134303e-01 3.93169880e-01 9.61347699e-01 3.71782482e-02
-8.36172342e-01 -2.36903325e-01 -8.16249177e-02 -1.87173516e-01
1.35958984e-01 7.28509784e-01 -6.58107281e-01 6.08402431e-01
5.79218149e-01 3.51922721e-01 9.09626186e-02 -6.15491033e-01
3.43428671e-01 6.27368033e-01 4.01704162e-01 -6.18759990e-01
1.15737557e+00 4.28679347e-01 -3.27842645e-02 2.20197424e-01
-8.73236060e-01 3.50851081e-02 -6.11365199e-01 2.75545478e-01
2.45485812e-01 -1.06014156e+00 -8.86785567e-01 4.12377685e-01
-7.07163095e-01 -4.67638463e-01 -1.24471381e-01 -9.86369029e-02
-5.17319024e-01 -7.44806528e-02 -4.44470823e-01 -5.71185768e-01
-7.95880139e-01 -9.91209805e-01 9.55730021e-01 -3.16217363e-01
-1.21993579e-01 -7.33496785e-01 -7.22868368e-02 1.02781904e+00
3.06699544e-01 3.24818611e-01 1.19667280e+00 -8.94957423e-01
-1.06455040e+00 -4.22978461e-01 -2.77926415e-01 1.04634754e-01
3.96178663e-03 -2.21778840e-01 -9.11790907e-01 -3.82381648e-01
-5.27613401e-01 -8.63148212e-01 7.80445158e-01 4.21118969e-03
1.62919188e+00 -7.69593537e-01 -2.23790139e-01 1.01739192e+00
1.04349017e+00 2.20171839e-01 9.41481590e-01 8.41473699e-01
8.10570955e-01 5.39659798e-01 1.05920482e+00 6.45022810e-01
8.27929795e-01 5.38978100e-01 5.22439957e-01 -6.65302202e-02
4.00559902e-01 -7.57833600e-01 2.38953844e-01 5.37705839e-01
6.54207885e-01 -3.15531552e-01 -1.02566874e+00 2.97540694e-01
-1.95133197e+00 -1.02381110e+00 -8.79938900e-02 2.35184884e+00
1.22832489e+00 5.81925690e-01 8.52801129e-02 3.34430099e-01
3.77342463e-01 8.88983458e-02 -8.64867270e-01 -4.56858695e-01
-3.21068168e-01 -2.79658973e-01 8.28480124e-01 8.59041885e-02
-1.20061028e+00 8.96422207e-01 5.77152777e+00 9.14690733e-01
-8.85089457e-01 -2.94508636e-01 1.50634801e+00 -1.40064970e-01
-5.69708943e-01 -9.45009515e-02 -1.19485772e+00 5.79227746e-01
1.18175256e+00 -3.58670145e-01 4.82854098e-01 1.13881528e+00
-6.82091415e-02 2.52681643e-01 -1.22328591e+00 8.18838239e-01
-1.51076078e-01 -1.79305375e+00 3.26169401e-01 1.89189732e-01
6.35070264e-01 -8.82478878e-02 3.36085707e-01 7.48352766e-01
4.34674382e-01 -1.31025839e+00 7.70004511e-01 2.14980796e-01
9.55846965e-01 -8.50416839e-01 9.22762513e-01 3.59073102e-01
-7.66706705e-01 4.99943160e-02 -4.00232047e-01 3.54798511e-02
-2.77738333e-01 6.66576207e-01 -1.42156506e+00 7.42443383e-01
7.81084418e-01 7.00832725e-01 -1.02915394e+00 8.21399689e-01
1.00556388e-01 8.06263924e-01 -2.05769122e-01 6.70975074e-02
-1.99001595e-01 -1.52256270e-03 1.08918004e-01 8.41316521e-01
-9.86576639e-03 -3.81217629e-01 -9.14830044e-02 6.33043647e-01
-6.89065397e-01 2.54416466e-01 -5.10498285e-01 4.15833555e-02
8.80571187e-01 1.12913370e+00 -2.11862445e-01 -3.20379108e-01
-2.42501691e-01 2.08776310e-01 4.54154134e-01 -2.97890995e-02
-8.56731951e-01 -1.88905090e-01 5.50832868e-01 3.93370628e-01
3.00755113e-01 3.59223694e-01 -1.12404299e+00 -1.25354517e+00
5.26494920e-01 -1.74088216e+00 6.72923923e-01 -7.01743186e-01
-1.14227355e+00 7.32048690e-01 -1.03127146e-02 -1.24612558e+00
-3.65639865e-01 -3.24894130e-01 -9.33964178e-02 7.32915461e-01
-1.34851325e+00 -1.22385001e+00 -2.39867270e-01 4.51549083e-01
1.15031116e-01 -4.03674036e-01 7.68599749e-01 3.74151111e-01
-8.42866302e-01 1.36295211e+00 6.05062068e-01 3.77265513e-01
1.03012311e+00 -1.48400843e+00 1.02814233e+00 5.97608984e-01
-1.18900547e-02 4.58028466e-01 6.64103925e-01 -6.78697348e-01
-1.66651487e+00 -1.47094190e+00 1.00145078e+00 -9.11593318e-01
5.39248645e-01 -1.15041673e+00 -1.02359414e+00 8.98550272e-01
9.16647315e-02 -6.40324429e-02 7.78443694e-01 3.37440044e-01
-6.97269738e-01 -8.11757863e-01 -1.23193824e+00 5.45162201e-01
7.16425776e-01 -2.18423679e-01 1.39574967e-02 4.09742475e-01
8.20146263e-01 -6.77022040e-01 -1.12775278e+00 4.00422782e-01
6.87199473e-01 -9.27139640e-01 9.71046805e-01 -5.68103313e-01
7.90850997e-01 7.86276162e-02 -1.77371830e-01 -1.15958428e+00
-3.16227823e-02 -8.32339704e-01 -2.09308147e-01 1.51941919e+00
8.48829687e-01 -7.90528119e-01 1.34911656e+00 7.64383554e-01
1.43644944e-01 -9.03708696e-01 -8.67015839e-01 -4.51550364e-01
3.43028009e-01 -1.48969308e-01 1.15839577e+00 8.94373119e-01
6.14184998e-02 1.33580282e-01 -8.65647793e-01 -1.88129246e-01
8.53654802e-01 3.61528754e-01 1.54599845e+00 -1.12061083e+00
-1.34597406e-01 -3.43948379e-02 1.24393590e-01 -7.10104167e-01
1.54284164e-01 -1.09214926e+00 -2.59953260e-01 -1.34601116e+00
2.91228473e-01 -8.29445779e-01 -3.52499664e-01 1.03504121e+00
-3.17309767e-01 2.04655707e-01 3.31151873e-01 2.19292209e-01
-7.05748200e-01 6.82967424e-01 1.15864217e+00 -3.31439435e-01
-3.88585851e-02 3.19459401e-02 -1.14693117e+00 -5.13310172e-02
1.12252021e+00 -8.10014188e-01 -2.62120962e-01 -3.34027670e-02
6.00921035e-01 5.03237903e-01 -1.17415242e-01 -9.02744234e-01
2.12310404e-01 -2.99191564e-01 4.45759267e-01 -1.26559091e+00
2.00621724e-01 -7.28066146e-01 4.46550250e-01 2.94538230e-01
-5.62746823e-01 2.86594510e-01 5.46531856e-01 2.29244411e-01
-1.90987855e-01 3.57403666e-01 4.30765092e-01 9.31606889e-02
1.47409644e-02 3.85585725e-01 -1.83715997e-03 3.61600876e-01
7.45376289e-01 -7.95243122e-03 -7.82765448e-01 -3.36509734e-01
-1.48889303e-01 6.72009408e-01 5.08306384e-01 4.10256892e-01
3.28687310e-01 -1.35059416e+00 -9.90907073e-01 4.05692667e-01
3.38414997e-01 2.10908622e-01 -6.70388117e-02 6.53954566e-01
-4.30795640e-01 7.76794076e-01 -3.40721369e-01 -4.69492972e-01
-1.08507216e+00 7.60887861e-01 5.17921671e-02 -8.87093842e-01
-4.58013982e-01 5.39041877e-01 1.79066490e-02 -1.00336587e+00
4.02479857e-01 -1.84034362e-01 3.16544324e-01 1.20818324e-01
4.37649399e-01 1.36053354e-01 4.63960737e-01 -9.45986286e-02
-2.14823142e-01 -3.61624479e-01 -4.82044488e-01 1.63669407e-01
1.40730679e+00 9.61807966e-02 -3.38224441e-01 5.01291573e-01
8.93180251e-01 1.77794755e-01 -1.08200061e+00 -2.94751436e-01
1.41118154e-01 -6.64868057e-01 -5.58352828e-01 -1.20721734e+00
-1.09416461e+00 6.55824006e-01 -1.02258138e-01 4.07011695e-02
1.04582572e+00 -5.82928479e-01 9.66965675e-01 6.36127055e-01
5.53831995e-01 -6.50976300e-01 -1.77970812e-01 3.14493030e-01
8.39003563e-01 -1.47864926e+00 5.56978807e-02 -3.75338346e-01
-8.22837949e-01 7.93354928e-01 7.70550430e-01 3.95274609e-01
3.87705356e-01 3.86179805e-01 2.17500508e-01 2.38943711e-01
-1.50393176e+00 5.46715617e-01 9.40595791e-02 3.88089806e-01
2.82291591e-01 1.30530789e-01 1.68254688e-01 5.94350874e-01
-7.71992266e-01 5.53467637e-03 5.95011950e-01 7.36995637e-01
1.23363771e-01 -1.48941875e+00 -6.39348209e-01 9.25976098e-01
-4.96697992e-01 -4.79664415e-01 -3.91147703e-01 7.28239357e-01
-2.56102562e-01 6.71445608e-01 -1.06710635e-01 -5.83184600e-01
2.49539152e-01 1.91441521e-01 -1.72383472e-01 -2.24509299e-01
-8.93671989e-01 -4.73047942e-01 4.02375400e-01 -5.54438055e-01
5.61969876e-01 -9.13440406e-01 -6.47257507e-01 -9.86062586e-01
2.64356583e-02 4.59359020e-01 3.52143049e-01 3.55540872e-01
5.75940311e-01 2.07268178e-01 6.61379397e-01 -2.07383454e-01
-6.96410477e-01 -9.32076514e-01 -2.29897767e-01 4.27022874e-01
2.47321337e-01 -4.04536843e-01 -1.21651784e-01 -3.89220193e-02] | [9.684686660766602, 7.891162872314453] |
8d3c79b3-32c8-4cd2-b543-d05d4e7f6b76 | lys-acoruna-at-semeval-2022-task-10 | 2204.12820 | null | https://arxiv.org/abs/2204.12820v1 | https://arxiv.org/pdf/2204.12820v1.pdf | LyS_ACoruña at SemEval-2022 Task 10: Repurposing Off-the-Shelf Tools for Sentiment Analysis as Semantic Dependency Parsing | This paper addressed the problem of structured sentiment analysis using a bi-affine semantic dependency parser, large pre-trained language models, and publicly available translation models. For the monolingual setup, we considered: (i) training on a single treebank, and (ii) relaxing the setup by training on treebanks coming from different languages that can be adequately processed by cross-lingual language models. For the zero-shot setup and a given target treebank, we relied on: (i) a word-level translation of available treebanks in other languages to get noisy, unlikely-grammatical, but annotated data (we release as much of it as licenses allow), and (ii) merging those translated treebanks to obtain training data. In the post-evaluation phase, we also trained cross-lingual models that simply merged all the English treebanks and did not use word-level translations, and yet obtained better results. According to the official results, we ranked 8th and 9th in the monolingual and cross-lingual setups. | ['Carlos Gómez-Rodríguez', 'David Vilares', 'Iago Alonso-Alonso'] | 2022-04-27 | null | https://aclanthology.org/2022.semeval-1.193 | https://aclanthology.org/2022.semeval-1.193.pdf | semeval-naacl-2022-7 | ['semantic-dependency-parsing'] | ['natural-language-processing'] | [-2.00011730e-01 3.14387679e-01 -8.97838101e-02 -5.44491887e-01
-1.53844619e+00 -9.45687056e-01 2.64618874e-01 4.02189553e-01
-8.63439858e-01 9.89324272e-01 1.99003458e-01 -7.24700212e-01
4.20563728e-01 -5.80708802e-01 -7.91336000e-01 -3.19708198e-01
1.00840069e-01 7.38270104e-01 2.19972953e-01 -5.04967332e-01
-8.52415264e-02 -2.20393483e-03 -9.33812559e-01 5.21410704e-01
1.11913955e+00 5.70123255e-01 3.03913623e-01 6.21990442e-01
-4.86690015e-01 6.19789422e-01 -5.26133716e-01 -1.24130154e+00
3.33959281e-01 -3.73861015e-01 -1.04274166e+00 8.14437196e-02
1.64139628e-01 2.79156327e-01 5.33141017e-01 1.07884693e+00
3.97818200e-02 -3.43055487e-01 2.36348331e-01 -5.48971593e-01
-6.00826383e-01 1.03650939e+00 -3.95353705e-01 -1.86533630e-01
3.77779156e-01 -1.60515428e-01 1.24477911e+00 -8.91154408e-01
9.86235023e-01 1.32685423e+00 5.72007716e-01 5.44623792e-01
-1.12369788e+00 -5.64306796e-01 3.76657635e-01 -2.64665604e-01
-1.12879360e+00 -5.96979022e-01 3.20712596e-01 -2.64645576e-01
1.30065584e+00 -5.32455295e-02 3.55488151e-01 1.12348342e+00
2.44241968e-01 4.21993494e-01 1.36503756e+00 -1.01085246e+00
1.48244679e-01 6.21228039e-01 1.97905466e-01 6.15960956e-01
1.98862463e-01 -4.58693177e-01 -2.71873146e-01 -7.35476390e-02
-6.36287630e-02 -7.75962293e-01 7.26638734e-03 1.73576251e-01
-1.10767519e+00 1.00125897e+00 -4.83824573e-02 6.92911565e-01
-2.56152093e-01 -5.28636158e-01 7.91801274e-01 6.68894708e-01
8.20052862e-01 4.56705511e-01 -1.16957498e+00 1.27133802e-01
-9.50035572e-01 -1.91798970e-01 1.09481406e+00 9.16647553e-01
9.51406419e-01 -2.59571165e-01 4.54817504e-01 1.02809751e+00
8.17759410e-02 6.03464901e-01 6.55508161e-01 -6.34677827e-01
1.16209829e+00 3.82481188e-01 9.64520052e-02 -2.80368656e-01
-2.94662476e-01 -2.17288807e-01 -3.29334289e-01 -7.71940425e-02
6.85524583e-01 -4.28096592e-01 -1.04995954e+00 1.87088585e+00
1.19402565e-01 -5.63097715e-01 7.50606656e-01 4.68024194e-01
4.43105400e-01 6.99892104e-01 3.43879372e-01 -3.88281554e-01
1.59929740e+00 -1.06918478e+00 -6.43196881e-01 -6.17933095e-01
1.35266960e+00 -1.23485684e+00 1.27126086e+00 4.78227586e-01
-1.20803428e+00 -5.36175370e-01 -1.02739215e+00 -1.76033959e-01
-6.78581178e-01 3.30278099e-01 2.55025059e-01 7.97345221e-01
-1.15766752e+00 5.66292882e-01 -1.01643813e+00 -5.37708879e-01
-1.12042263e-01 3.62233743e-02 -7.04840422e-01 -3.08081895e-01
-1.43705106e+00 1.15455222e+00 4.44349080e-01 -7.23895431e-03
-3.79937917e-01 -1.23749532e-01 -1.19962156e+00 -3.29868466e-01
3.97038013e-01 -2.52192080e-01 1.14738441e+00 -1.33682942e+00
-1.55878437e+00 1.33493268e+00 -2.07226485e-01 -2.27187902e-01
3.27733278e-01 -3.86530697e-01 -5.13886154e-01 -2.43373141e-01
4.91992474e-01 2.56191522e-01 4.79593009e-01 -9.49525535e-01
-7.61469424e-01 -6.12605035e-01 4.92416508e-02 3.07473660e-01
-2.77265191e-01 5.92159927e-01 -6.08065069e-01 -5.05847156e-01
1.13907024e-01 -1.03173316e+00 -5.30352890e-01 -7.93881655e-01
-3.42870623e-01 -1.39925480e-01 2.42649972e-01 -1.38610983e+00
9.99594331e-01 -2.00716758e+00 1.12807341e-01 -5.96743962e-03
-5.79818428e-01 2.30070248e-01 -5.13500631e-01 4.64219391e-01
-4.84350324e-02 3.19811046e-01 -4.10922110e-01 -9.80411470e-01
-2.28368193e-01 5.89107692e-01 -2.37252578e-01 1.65183544e-01
3.28491002e-01 7.44610786e-01 -9.50750649e-01 -6.10082507e-01
-1.13448068e-01 2.01548770e-01 -5.10894954e-01 6.41907677e-02
-2.73253173e-01 4.62983906e-01 -2.92202830e-01 6.26304269e-01
5.51128983e-01 3.17204446e-01 7.33748138e-01 -1.08513139e-01
-3.65562111e-01 1.03269517e+00 -1.12432015e+00 2.01101041e+00
-1.09573758e+00 1.49884894e-01 2.28865057e-01 -9.64873731e-01
8.38397443e-01 4.45453256e-01 1.37995243e-01 -8.06529582e-01
6.60388619e-02 8.87186766e-01 -1.79344676e-02 -3.71950120e-01
5.05528510e-01 -2.63562173e-01 -6.51078165e-01 4.84411329e-01
4.78506565e-01 -4.36245680e-01 4.64888543e-01 2.18769498e-02
6.76630557e-01 4.48436916e-01 3.68171275e-01 -4.29527938e-01
6.89201951e-01 1.41052917e-01 6.06633723e-01 2.93998122e-01
5.09561896e-02 4.63324875e-01 5.39263129e-01 -3.12249124e-01
-1.06029344e+00 -7.42328405e-01 -1.61365524e-01 1.37018716e+00
-4.55229461e-01 -6.86437547e-01 -1.09184754e+00 -1.00427258e+00
-5.43904006e-01 9.95007992e-01 -3.68435264e-01 4.40937638e-01
-8.59395146e-01 -8.89388978e-01 5.80745339e-01 1.43926337e-01
7.96108395e-02 -9.67702091e-01 -5.88422902e-02 4.98453736e-01
-4.37219143e-01 -1.53340876e+00 -4.15358424e-01 8.13159347e-01
-6.42982364e-01 -9.12330508e-01 -4.18174922e-01 -1.04122782e+00
4.66290385e-01 -3.91031951e-01 1.32323062e+00 -1.08442351e-01
2.61615127e-01 -1.42254993e-01 -5.61784208e-01 -4.43392307e-01
-9.32679772e-01 4.93661493e-01 -4.22967710e-02 -1.54975414e-01
3.59954387e-01 -2.07386240e-01 2.54835397e-01 4.46742065e-02
-7.80261993e-01 -1.80839449e-01 4.77647483e-01 6.36781335e-01
4.49171871e-01 -1.74992919e-01 6.31815851e-01 -1.35861528e+00
3.87875825e-01 -1.87205210e-01 -6.92569911e-01 4.88643914e-01
-4.08499718e-01 3.04938316e-01 9.17901278e-01 -9.90923718e-02
-1.28866446e+00 -1.09948963e-03 -7.79632628e-01 2.37206012e-01
-1.12406917e-01 6.36617422e-01 -5.77242255e-01 2.90513605e-01
5.75425744e-01 -6.47739321e-02 -3.37053031e-01 -6.53961241e-01
6.95201993e-01 8.26631963e-01 2.89391071e-01 -7.82306552e-01
5.73127031e-01 5.97788915e-02 -5.53947210e-01 -7.70629346e-01
-1.21928799e+00 -2.57916868e-01 -1.14999747e+00 4.46191281e-01
1.33487117e+00 -1.16581059e+00 1.89428478e-01 4.40305799e-01
-1.50026071e+00 -4.16489482e-01 -2.43826389e-01 5.86178720e-01
-8.88560042e-02 4.29736495e-01 -9.40190077e-01 -6.31144345e-01
-3.40658486e-01 -1.16897571e+00 1.11517119e+00 -2.25182801e-01
-2.50574708e-01 -1.31205237e+00 3.71415436e-01 3.44564915e-01
1.61097944e-01 -1.03843473e-01 9.11262155e-01 -9.51457679e-01
3.47949006e-02 -2.82490760e-01 1.14328444e-01 8.79233181e-01
1.72929466e-01 3.02684307e-03 -9.85733807e-01 -2.49412224e-01
3.45031053e-01 -8.04673731e-01 5.78171670e-01 5.79170175e-02
2.24968165e-01 -1.75730392e-01 -2.10397169e-02 4.07766640e-01
1.36204898e+00 5.10876440e-02 4.33759391e-01 5.12033999e-01
5.26338220e-01 9.46605086e-01 6.49318933e-01 -2.11899400e-01
9.22127843e-01 4.59736675e-01 -8.79178941e-02 -1.97170123e-01
2.36021027e-01 -3.35118830e-01 8.16527784e-01 1.63204825e+00
4.61410284e-02 -1.89915314e-01 -6.54291868e-01 7.68194616e-01
-1.74785388e+00 -3.74119580e-01 -2.69921511e-01 2.26095033e+00
1.11612999e+00 4.38401937e-01 -6.30688220e-02 -2.96057701e-01
5.84573627e-01 1.11374274e-01 -8.71522576e-02 -1.01144469e+00
-4.60612655e-01 7.44712174e-01 7.76017845e-01 8.73422861e-01
-1.02099013e+00 1.72093868e+00 5.69780731e+00 7.43037939e-01
-1.11411345e+00 6.61117077e-01 5.93354106e-01 2.36240581e-01
-3.79487664e-01 5.09033501e-01 -1.09877491e+00 3.01296175e-01
1.31288242e+00 7.95537978e-02 2.20921904e-01 7.48858631e-01
-3.51358615e-02 -2.23149449e-01 -9.09174025e-01 4.21439081e-01
5.53123280e-02 -7.81591952e-01 -1.35860771e-01 -6.45203963e-02
7.45531082e-01 6.16278470e-01 -5.50543964e-01 6.82683110e-01
7.16916978e-01 -6.75403953e-01 8.13289464e-01 -1.66946515e-01
7.29630411e-01 -5.71873605e-01 1.09742296e+00 5.90310216e-01
-1.24936593e+00 3.86009544e-01 -5.88373065e-01 -3.97654399e-02
4.44726825e-01 7.70623207e-01 -2.68274695e-01 1.23213816e+00
5.64833879e-01 6.34778142e-01 -6.35483444e-01 1.54104218e-01
-6.33091390e-01 7.61243701e-01 -4.72790509e-01 2.64043272e-01
3.94075304e-01 -5.50246477e-01 2.09099710e-01 1.47721195e+00
3.42066824e-01 -3.92861217e-01 5.75861037e-01 2.15504944e-01
-4.56899963e-02 7.24418700e-01 -3.71543825e-01 1.47241771e-01
8.16178024e-02 1.41089928e+00 -7.65091658e-01 -5.26849687e-01
-8.86227071e-01 1.16065371e+00 5.61060250e-01 1.31092861e-01
-5.56168675e-01 -3.70396137e-01 4.29524064e-01 -3.46548967e-02
2.55731583e-01 -3.21142673e-01 -2.04817429e-01 -1.65799439e+00
3.01255047e-01 -9.06787038e-01 6.68707907e-01 -5.68972647e-01
-1.33907068e+00 1.32264555e+00 -2.78374135e-01 -7.29757488e-01
-3.64014685e-01 -7.50823140e-01 -2.83144563e-01 1.38778639e+00
-1.88718212e+00 -1.40757799e+00 5.47917187e-01 4.21992302e-01
6.11804903e-01 -9.67707932e-02 1.21225023e+00 4.01189238e-01
-4.60684925e-01 5.75963497e-01 -1.25969574e-01 2.62791246e-01
1.09085095e+00 -1.21026337e+00 7.25087106e-01 9.57219362e-01
3.60380769e-01 5.77291489e-01 4.47252363e-01 -6.77416325e-01
-8.85620952e-01 -1.16815126e+00 1.78292382e+00 -6.54591024e-01
1.14239180e+00 -7.91259229e-01 -1.24537253e+00 1.03304410e+00
5.26248336e-01 -2.43016317e-01 6.83675528e-01 3.29874665e-01
-2.12790117e-01 1.59828529e-01 -9.00127828e-01 3.04683000e-01
6.84420347e-01 -6.47059500e-01 -9.21434820e-01 4.75640208e-01
7.42365181e-01 -4.53211963e-01 -7.51239359e-01 6.97899098e-03
4.18263286e-01 -6.43579483e-01 2.92184353e-01 -7.19644666e-01
2.97073841e-01 -3.94153893e-02 -3.17400932e-01 -1.51505554e+00
1.18058003e-01 -4.63614792e-01 8.25455010e-01 1.42161930e+00
9.55640793e-01 -7.98183382e-01 3.41367394e-01 4.39335048e-01
-3.10864717e-01 -5.18021047e-01 -9.54234362e-01 -9.04121220e-01
6.64590120e-01 -7.58849025e-01 2.38985345e-01 9.91752863e-01
1.61754519e-01 7.79430389e-01 -2.05360994e-01 1.65087923e-01
4.26467001e-01 -1.02910720e-01 6.42404497e-01 -1.10992420e+00
-3.33957076e-01 -8.51636976e-02 1.67145312e-01 -7.80162394e-01
6.69058621e-01 -1.16526747e+00 2.00006202e-01 -1.38964236e+00
-1.20688520e-01 -5.75646102e-01 -5.63043840e-02 7.24825025e-01
-2.99751729e-01 2.98690587e-01 8.88423100e-02 1.34502441e-01
-3.23237032e-01 2.21247628e-01 7.70803928e-01 2.52150476e-01
-2.04659253e-01 -1.51877388e-01 -8.35056782e-01 8.77267003e-01
6.16097391e-01 -8.78335655e-01 2.63676364e-02 -6.80611372e-01
3.65350455e-01 4.17335294e-02 -3.06826741e-01 -4.63833243e-01
-1.26949608e-01 -2.87508592e-02 -3.84791591e-03 -2.15243965e-01
-2.26446707e-02 -6.80355310e-01 -2.60031164e-01 3.67289931e-01
-1.37361124e-01 2.25114673e-01 3.79218459e-01 -1.48503557e-01
-4.02716517e-01 -6.61674023e-01 8.57649982e-01 -4.39221919e-01
-3.68823826e-01 -6.11929819e-02 -4.18385953e-01 3.09208423e-01
7.50673175e-01 7.60205835e-02 -8.92009120e-03 4.22203988e-02
-9.39547598e-01 1.05357036e-01 7.41288960e-01 4.88179594e-01
-2.60995716e-01 -9.35316861e-01 -8.69373500e-01 3.71192694e-01
1.21769942e-01 -4.88474742e-02 -1.71678871e-01 6.06006145e-01
-3.31855655e-01 4.66130584e-01 9.83151272e-02 -3.06772202e-01
-1.01124454e+00 4.16819245e-01 1.72161907e-01 -9.96115625e-01
-1.74127132e-01 6.89588010e-01 9.53715518e-02 -1.15919459e+00
-1.76060930e-01 -4.93646413e-01 -1.88168153e-01 1.65150926e-01
2.19807804e-01 -3.37373644e-01 7.12402463e-01 -8.50464284e-01
-5.67850471e-01 6.71144664e-01 -6.09578490e-02 -5.50977468e-01
1.31175983e+00 -3.51369828e-01 -3.48338574e-01 6.07873440e-01
1.13701761e+00 7.19906092e-01 -6.08549416e-01 -3.04387063e-01
4.89723533e-01 2.07131550e-01 -2.67327607e-01 -7.17939675e-01
-7.66632378e-01 6.93419278e-01 -7.79538080e-02 1.43951938e-01
9.19925153e-01 1.46573797e-01 9.21395123e-01 3.46253633e-01
6.91980243e-01 -1.38916421e+00 -4.69012052e-01 1.04455185e+00
4.90888894e-01 -1.31050420e+00 -1.70544624e-01 -4.30685848e-01
-8.51071477e-01 1.15069318e+00 3.21898878e-01 -3.24742645e-02
7.32831001e-01 5.03789067e-01 6.78765714e-01 8.70228112e-02
-7.63233840e-01 -2.18529075e-01 1.14656456e-01 3.06194663e-01
8.50674987e-01 3.43645290e-02 -6.33240640e-01 9.39525425e-01
-5.50988317e-01 -7.08881244e-02 4.95054513e-01 9.66582894e-01
-2.70183921e-01 -1.78745019e+00 -2.59802073e-01 1.81087144e-02
-8.15898240e-01 -6.28511190e-01 -2.65936702e-01 8.67675245e-01
2.36147359e-01 1.09586024e+00 -5.38621917e-02 -4.09543468e-03
6.16401017e-01 4.78223562e-01 3.76320720e-01 -1.08551908e+00
-7.18601644e-01 2.71705031e-01 5.30339777e-01 -3.76320153e-01
-3.47954124e-01 -7.03761399e-01 -1.03811884e+00 2.12459683e-01
-3.66145819e-01 6.98728681e-01 1.09926558e+00 1.36343789e+00
-6.75286278e-02 2.55028367e-01 5.92001438e-01 -6.45246923e-01
-4.77343827e-01 -1.24282098e+00 -3.95283431e-01 3.35170865e-01
-9.80514139e-02 2.40867846e-02 -5.02070963e-01 3.28216404e-01] | [10.437235832214355, 9.78281021118164] |
c6aeb2ba-ea69-46c9-b36d-2e436c36d834 | nade-a-benchmark-for-robust-adverse-drug | 2109.10080 | null | https://arxiv.org/abs/2109.10080v2 | https://arxiv.org/pdf/2109.10080v2.pdf | NADE: A Benchmark for Robust Adverse Drug Events Extraction in Face of Negations | Adverse Drug Event (ADE) extraction models can rapidly examine large collections of social media texts, detecting mentions of drug-related adverse reactions and trigger medical investigations. However, despite the recent advances in NLP, it is currently unknown if such models are robust in face of negation, which is pervasive across language varieties. In this paper we evaluate three state-of-the-art systems, showing their fragility against negation, and then we introduce two possible strategies to increase the robustness of these models: a pipeline approach, relying on a specific component for negation detection; an augmentation of an ADE extraction dataset to artificially create negated samples and further train the models. We show that both strategies bring significant increases in performance, lowering the number of spurious entities predicted by the models. Our dataset and code will be publicly released to encourage research on the topic. | ['Giuseppe Serra', 'Enrico Santus', 'Emmanuele Chersoni', 'Beatrice Portelli', 'Simone Scaboro'] | 2021-09-21 | null | https://aclanthology.org/2021.wnut-1.26 | https://aclanthology.org/2021.wnut-1.26.pdf | wnut-acl-2021-11 | ['negation-detection'] | ['natural-language-processing'] | [ 3.50693643e-01 4.44722474e-01 -3.07737947e-01 -2.61750191e-01
-8.67798209e-01 -8.47964942e-01 8.23850751e-01 9.28905845e-01
-4.66793656e-01 9.79172885e-01 3.14415634e-01 -6.82663023e-01
-1.02850787e-01 -7.56131470e-01 -7.05322146e-01 -2.05812067e-01
-9.31699574e-02 5.39889932e-01 3.45002949e-01 -1.95577055e-01
7.27856159e-02 7.30652988e-01 -9.97983575e-01 5.81479132e-01
7.64061570e-01 4.76447433e-01 -4.04338419e-01 4.22362983e-01
-8.97106677e-02 1.11849165e+00 -6.89974070e-01 -7.31851041e-01
1.35596553e-02 -4.32378262e-01 -6.97740734e-01 -3.07193518e-01
1.57515392e-01 -1.25899553e-01 1.02269091e-01 1.08489001e+00
5.38701773e-01 -4.37448710e-01 5.82529485e-01 -9.59383070e-01
-5.47725439e-01 7.46890128e-01 -2.51556188e-01 3.39885831e-01
7.14107811e-01 2.34951481e-01 1.02138686e+00 -7.48735905e-01
1.13549984e+00 9.87828076e-01 8.13752174e-01 6.29183292e-01
-1.08595490e+00 -6.48818731e-01 -1.34633616e-01 -1.20366886e-01
-1.27613306e+00 -5.37660480e-01 3.31655562e-01 -4.13256615e-01
1.38184750e+00 1.92044169e-01 5.52121222e-01 1.19587338e+00
2.96247095e-01 5.76110601e-01 1.17209291e+00 -3.77929807e-01
3.13014299e-01 4.52055991e-01 5.26964143e-02 5.72263002e-01
5.85222483e-01 -8.65414962e-02 -4.88014191e-01 -7.34482408e-01
1.91618994e-01 -1.71176195e-01 -2.63476372e-01 5.44050895e-02
-1.00672340e+00 7.62985826e-01 -2.69830618e-02 5.11424005e-01
-8.06750596e-01 -4.67474610e-01 5.06420612e-01 3.06693465e-01
6.09816432e-01 6.83346093e-01 -8.56137693e-01 -4.52385321e-02
-1.00495553e+00 4.51741904e-01 1.12704957e+00 6.74650609e-01
2.01353922e-01 -6.02184474e-01 -1.86960027e-01 4.75131214e-01
3.41791570e-01 2.26027697e-01 2.56474078e-01 -2.14520648e-01
4.77762789e-01 7.49977946e-01 6.05784170e-02 -7.92827427e-01
-6.39304936e-01 -3.06403399e-01 -3.79415333e-01 -3.29122335e-01
4.32221204e-01 -4.57373112e-01 -9.34872389e-01 1.55695283e+00
4.58236486e-01 1.63846672e-01 3.13678265e-01 3.91019791e-01
9.17078614e-01 2.14414239e-01 5.58272779e-01 -3.15392882e-01
1.44781041e+00 -2.89872646e-01 -9.62183058e-01 -2.27614954e-01
9.64568734e-01 -1.00813305e+00 4.57202047e-01 4.13218737e-01
-1.02700281e+00 2.38977686e-01 -9.79050040e-01 1.28429517e-01
-7.47322679e-01 -1.41349137e-01 5.92588067e-01 7.85349131e-01
-7.08711684e-01 6.84978604e-01 -9.05478477e-01 -4.26500976e-01
6.56876147e-01 6.37221098e-01 -4.37873960e-01 -9.03887451e-02
-1.50457942e+00 1.27122545e+00 2.43670464e-01 -1.55873597e-01
-4.19306576e-01 -8.91859651e-01 -8.54106486e-01 -1.39117599e-01
3.77658546e-01 -6.69645488e-01 1.11758304e+00 -6.37944937e-01
-8.97189081e-01 1.03641009e+00 -1.29734099e-01 -5.81174135e-01
6.31637394e-01 -1.34164214e-01 -7.85402179e-01 4.71818913e-03
2.13118404e-01 4.30275947e-01 2.69963056e-01 -5.09073317e-01
-5.01847625e-01 -3.21029395e-01 -7.31651206e-03 -2.59128183e-01
-3.38581204e-01 5.06167173e-01 -9.95570943e-02 -5.52458763e-01
-3.98745894e-01 -8.85939777e-01 -4.81728077e-01 -3.13135475e-01
-5.93266606e-01 -4.71039087e-01 4.41779792e-01 -5.72528601e-01
1.33824003e+00 -1.89278615e+00 -4.14944232e-01 3.08878243e-01
2.44648576e-01 4.98871386e-01 -9.58210081e-02 6.17187560e-01
-4.06560630e-01 6.85494363e-01 -2.39155456e-01 1.39122391e-02
-2.10985705e-01 8.76842961e-02 -2.20546022e-01 5.34673810e-01
9.51827943e-01 9.06521142e-01 -1.09641755e+00 -4.09336776e-01
-1.42653033e-01 5.91825724e-01 -5.81182122e-01 -1.68347746e-01
-4.42712426e-01 1.80017084e-01 -6.13216341e-01 7.35478282e-01
5.25470197e-01 -2.79927164e-01 4.53515917e-01 -6.84412150e-03
-7.47083724e-02 9.55709457e-01 -1.20621610e+00 9.34753954e-01
1.91124916e-01 2.42087618e-01 -1.39775530e-01 -5.58242261e-01
6.39431655e-01 6.65927231e-01 4.68656361e-01 -3.94412220e-01
1.79236472e-01 4.64686483e-01 1.49768203e-01 -6.11314297e-01
3.37889969e-01 -3.61008495e-01 5.73425181e-02 3.31853896e-01
-8.32092166e-02 1.87263042e-01 3.51317763e-01 3.15470397e-01
1.40242398e+00 -2.12950736e-01 5.59874892e-01 -2.11236686e-01
4.69018579e-01 2.90004015e-01 4.90479141e-01 6.79880738e-01
-1.33580849e-01 3.34329844e-01 8.32611442e-01 -3.09491277e-01
-8.39609146e-01 -6.26194894e-01 -4.89448518e-01 4.62250412e-01
-5.49857318e-01 -6.63502395e-01 -6.27342463e-01 -1.08471799e+00
1.34090438e-01 7.00501263e-01 -6.23021543e-01 1.16106242e-01
-5.40338457e-01 -1.38220358e+00 8.26960206e-01 3.22368085e-01
-1.13171376e-01 -1.06847489e+00 -3.21945459e-01 3.41654986e-01
-2.70438436e-02 -1.22432923e+00 3.00266352e-02 3.41467261e-01
-6.17880046e-01 -1.54605687e+00 -5.52594602e-01 -4.65053886e-01
5.17997801e-01 -5.62982917e-01 1.26818120e+00 1.35272011e-01
-2.15300351e-01 6.71117678e-02 -2.89678812e-01 -9.56227005e-01
-9.43548918e-01 2.57359110e-02 1.32453647e-02 -3.91953737e-01
1.09641969e+00 -2.78951287e-01 -3.62260222e-01 -1.99479476e-01
-1.18330562e+00 -4.56708908e-01 5.20842075e-01 3.60596418e-01
3.49142760e-01 -3.42553645e-01 9.09820974e-01 -1.58761263e+00
9.36749399e-01 -9.02963698e-01 -4.29767013e-01 1.40257999e-01
-6.72035456e-01 1.96026843e-02 5.43302774e-01 -3.10846567e-01
-5.79660475e-01 2.84000278e-01 -6.72961891e-01 1.74741298e-01
-3.70637208e-01 8.31277311e-01 5.23831025e-02 1.94279224e-01
8.08476627e-01 -1.81506217e-01 -2.13883609e-01 -4.13460344e-01
2.51462817e-01 8.05516779e-01 7.88648278e-02 -3.21169458e-02
4.99218941e-01 2.93787628e-01 -1.11744255e-01 -7.53779233e-01
-9.96583641e-01 -4.42302853e-01 -3.24842960e-01 3.07214558e-01
8.08288574e-01 -9.91421580e-01 -4.82593596e-01 2.12252930e-01
-1.21325433e+00 8.91936868e-02 -1.10802449e-01 5.26774287e-01
7.65559226e-02 2.90011674e-01 -8.82744789e-01 -5.19179821e-01
-3.82341146e-01 -9.02562022e-01 8.41188669e-01 6.70410097e-02
-7.04088628e-01 -9.33222175e-01 3.91527086e-01 7.32158124e-02
1.59828097e-01 4.45541412e-01 1.05892289e+00 -1.52632058e+00
-2.44183555e-01 -5.07297039e-01 5.90908900e-02 7.84972161e-02
1.56950250e-01 1.20626360e-01 -9.30541635e-01 -9.00460407e-02
-5.95530048e-02 -1.21048249e-01 6.95481896e-01 1.66477904e-01
5.24697363e-01 -4.08364743e-01 -6.25470221e-01 1.56842321e-02
1.21185255e+00 1.31447390e-01 6.87026024e-01 3.55717272e-01
2.48933434e-01 5.30028224e-01 3.26403946e-01 2.66962081e-01
2.13414580e-01 2.89198190e-01 9.63826627e-02 -3.95891398e-01
1.24541773e-02 -2.01967895e-01 2.16097221e-01 2.72860527e-01
2.34398857e-01 -4.36535418e-01 -1.08345354e+00 8.18532884e-01
-1.52291358e+00 -8.13065469e-01 -4.64536339e-01 1.90512919e+00
1.17357337e+00 4.37570304e-01 1.50469273e-01 9.25523937e-02
4.53143954e-01 -1.42635494e-01 -3.16314906e-01 -4.48475659e-01
-3.78502607e-01 6.10638201e-01 5.35891473e-01 3.34774494e-01
-1.15518999e+00 7.44887888e-01 7.07910442e+00 3.28389943e-01
-1.05470169e+00 1.37820989e-02 6.08070731e-01 1.84085350e-02
-3.14960361e-01 -1.55770093e-01 -1.01253426e+00 3.23289156e-01
1.21035826e+00 -1.47382900e-01 -1.96308360e-01 5.73561013e-01
1.73699066e-01 1.24926306e-01 -1.15787494e+00 3.15373451e-01
1.08380169e-01 -1.50010121e+00 -4.11785915e-02 1.04190938e-01
5.20893931e-01 4.28351521e-01 -2.62697726e-01 9.51393023e-02
3.68293941e-01 -8.84186864e-01 2.59822875e-01 4.31339592e-01
3.67514431e-01 -6.45105064e-01 9.61345315e-01 2.17184961e-01
-6.16440952e-01 1.74719021e-01 -2.82301521e-03 3.85325290e-02
2.56607473e-01 9.28111672e-01 -1.33693588e+00 4.11856413e-01
5.00913680e-01 7.51045167e-01 -9.05900717e-01 1.08698869e+00
-4.28246170e-01 7.39126325e-01 -4.75167900e-01 -2.61739910e-01
1.94017112e-01 3.49469483e-01 6.24512315e-01 1.52599287e+00
-4.63018082e-02 2.28365764e-01 9.44651812e-02 8.58833075e-01
-2.23190859e-01 3.87556255e-01 -7.56849885e-01 -5.13529599e-01
3.17469418e-01 1.13503981e+00 -8.72654796e-01 -5.15449166e-01
-5.94893157e-01 6.74331546e-01 1.41560391e-01 -5.84761240e-03
-7.12928832e-01 -4.15769637e-01 4.00753558e-01 3.71396363e-01
3.64113927e-01 2.42162511e-01 -1.45044684e-01 -1.08192694e+00
1.62054092e-01 -1.21352291e+00 6.80271029e-01 -4.34765339e-01
-1.57922471e+00 5.84889233e-01 -3.55271816e-01 -8.50111485e-01
-3.58425617e-01 -6.78115666e-01 -2.78392553e-01 8.22255194e-01
-1.52992702e+00 -8.57847154e-01 5.04739583e-01 4.34505075e-01
2.58414764e-02 1.60507411e-01 1.07260418e+00 6.29669368e-01
-4.60023642e-01 4.56415653e-01 -2.87827611e-01 4.63625520e-01
9.82280910e-01 -1.10898626e+00 5.11048615e-01 6.64401174e-01
6.68170378e-02 1.04646504e+00 8.26411188e-01 -1.10877740e+00
-1.10987937e+00 -1.05814838e+00 1.78792465e+00 -1.04968846e+00
9.24082696e-01 -3.49054992e-01 -1.03779876e+00 8.01619351e-01
1.56462073e-01 -2.29770944e-01 1.13424158e+00 3.08695108e-01
-3.64995778e-01 5.22188962e-01 -1.37712419e+00 4.76432323e-01
6.05008841e-01 -4.60745186e-01 -9.46096957e-01 8.61640334e-01
5.51987231e-01 -3.29841077e-01 -1.00947857e+00 5.36926031e-01
3.33870560e-01 -6.98283255e-01 7.73141146e-01 -1.08956456e+00
5.13849080e-01 -2.67540902e-01 2.64046460e-01 -8.76005411e-01
4.52334732e-02 -5.47025561e-01 3.78978476e-02 1.23205352e+00
1.15865886e+00 -7.97000527e-01 7.14155078e-01 7.73091018e-01
1.69704542e-01 -8.59480143e-01 -6.44525051e-01 -3.98199886e-01
5.60612939e-02 -3.60195220e-01 4.29493189e-01 1.37838769e+00
4.41274881e-01 6.37552619e-01 5.17322682e-02 3.56057137e-01
9.36037600e-02 -3.88680786e-01 2.78421819e-01 -1.28234947e+00
-1.76381335e-01 -2.70486385e-01 -4.02393818e-01 -4.29275036e-01
-1.29817575e-02 -7.94992805e-01 -2.99961627e-01 -1.49272883e+00
3.63109171e-01 -2.51791507e-01 -1.86716944e-01 9.15779591e-01
-3.48210543e-01 3.46960574e-01 -1.67973518e-01 -1.10266693e-01
-5.26545882e-01 -8.49814638e-02 6.26264334e-01 -2.43185069e-02
-3.55199158e-01 -1.45702448e-03 -9.20528352e-01 8.79332781e-01
6.27419114e-01 -1.11706173e+00 -4.35301401e-02 2.80551035e-02
6.74742579e-01 -3.01061332e-01 2.43865564e-01 -5.34945011e-01
2.58969486e-01 1.21462718e-01 4.67549771e-01 -5.70811391e-01
-1.96643695e-01 -5.04674494e-01 5.19646443e-02 6.60124540e-01
-3.35870147e-01 3.01287532e-01 5.87323427e-01 5.48350096e-01
-1.54299781e-01 -2.94504464e-01 5.63973308e-01 -2.10137010e-01
2.66993195e-02 -9.62579548e-02 -6.37476325e-01 1.08087637e-01
8.73228490e-01 1.94011405e-01 -3.80620360e-01 -1.14744946e-01
-7.06368148e-01 8.63396451e-02 3.55744630e-01 4.30807114e-01
2.11293265e-01 -7.11349726e-01 -8.84656012e-01 5.99611476e-02
1.31345212e-01 -4.10992891e-01 -6.49837106e-02 9.85113621e-01
-4.47026014e-01 5.53111196e-01 2.59507716e-01 -2.47244030e-01
-1.36379313e+00 7.39289939e-01 2.91853189e-01 -6.28337979e-01
-4.14627433e-01 6.64470434e-01 -3.46216112e-01 -4.68082041e-01
1.46931559e-01 -1.67461976e-01 -3.71432573e-01 1.84339494e-01
6.98135614e-01 -5.10613210e-02 5.75216889e-01 -3.77469540e-01
-8.94580364e-01 -2.88035274e-02 -4.15840894e-01 1.34064466e-01
1.63744092e+00 4.24418449e-01 -3.63552451e-01 7.04888180e-02
9.20132160e-01 4.80198145e-01 -3.13139975e-01 5.45144603e-02
5.01256108e-01 6.52395636e-02 -6.80026487e-02 -1.30525219e+00
-5.86948454e-01 4.37999099e-01 5.15343964e-01 3.95453483e-01
9.52453375e-01 6.04190156e-02 6.70145988e-01 2.95078486e-01
-1.74737826e-01 -7.51774967e-01 -5.67738593e-01 4.98839796e-01
4.47086394e-01 -1.07611537e+00 1.87115759e-01 -6.99707448e-01
-3.92424732e-01 9.38528538e-01 1.70994978e-02 7.55328611e-02
7.04172015e-01 5.93855143e-01 1.35538384e-01 -6.50870621e-01
-8.12964559e-01 -1.14935383e-01 2.46856540e-01 3.81549686e-01
9.85527992e-01 -7.84000382e-02 -8.07838857e-01 7.56909668e-01
5.86217176e-03 3.62060070e-01 7.10283458e-01 1.00160682e+00
-2.87051443e-02 -1.58827126e+00 -1.70316890e-01 5.90922117e-01
-1.28476882e+00 -6.76206887e-01 -9.31322217e-01 8.93397152e-01
3.00575048e-01 9.95303929e-01 -3.54967386e-01 5.71304150e-02
7.33631313e-01 4.25803483e-01 2.40617394e-01 -8.32745135e-01
-1.15073800e+00 5.88482879e-02 5.30317485e-01 -3.53397459e-01
-6.32774174e-01 -7.98412681e-01 -1.11689162e+00 -1.13269784e-01
-4.57375705e-01 2.54015356e-01 4.39061701e-01 1.05687547e+00
6.70643806e-01 3.71061653e-01 -3.96214202e-02 6.43941090e-02
-3.76873434e-01 -8.38381171e-01 -2.00592607e-01 4.46983725e-01
3.66000146e-01 -1.65281340e-01 -2.77234763e-01 6.11948557e-02] | [8.468567848205566, 8.840911865234375] |
1e9197ce-c779-4329-9e70-19e7751deedf | benchmarking-robustness-of-deep-reinforcement | 2306.10950 | null | https://arxiv.org/abs/2306.10950v1 | https://arxiv.org/pdf/2306.10950v1.pdf | Benchmarking Robustness of Deep Reinforcement Learning approaches to Online Portfolio Management | Deep Reinforcement Learning approaches to Online Portfolio Selection have grown in popularity in recent years. The sensitive nature of training Reinforcement Learning agents implies a need for extensive efforts in market representation, behavior objectives, and training processes, which have often been lacking in previous works. We propose a training and evaluation process to assess the performance of classical DRL algorithms for portfolio management. We found that most Deep Reinforcement Learning algorithms were not robust, with strategies generalizing poorly and degrading quickly during backtesting. | ['Fabrice Daniel', 'Fabrice Popineau', 'Arpad Rimmel', 'Bich-Liên Doan', 'Marc Velay'] | 2023-06-19 | null | null | null | null | ['management', 'benchmarking', 'benchmarking'] | ['miscellaneous', 'miscellaneous', 'robots'] | [-3.64318371e-01 -3.41963381e-01 -1.63468286e-01 -1.75158054e-01
-5.97046673e-01 -8.23600709e-01 6.00157917e-01 8.95781666e-02
-5.16143799e-01 1.09881997e+00 1.22948652e-02 -5.69780171e-01
-6.02415502e-01 -8.23658943e-01 -5.22464514e-01 -4.07001853e-01
-3.65684897e-01 6.89870298e-01 -1.16829574e-01 -6.49225771e-01
5.71136475e-01 4.40754831e-01 -1.44157302e+00 3.63234617e-02
6.30311966e-01 1.12837529e+00 -2.75983840e-01 6.51441574e-01
-9.17896852e-02 1.26683819e+00 -1.04960728e+00 -6.60072565e-01
5.69507718e-01 -6.73699498e-01 -5.92390537e-01 1.03818186e-01
-9.66160372e-02 -6.83697045e-01 -5.74249446e-01 8.51772189e-01
7.77202904e-01 2.60195434e-01 3.60990793e-01 -1.18049073e+00
-7.52564669e-01 8.86300147e-01 -5.00302911e-01 5.84596276e-01
2.64409989e-01 5.45725107e-01 1.18064392e+00 -4.78086501e-01
4.40463752e-01 1.10323250e+00 5.82632780e-01 5.37045538e-01
-1.02063179e+00 -5.92586100e-01 1.40102684e-01 1.06295593e-01
-6.73102140e-01 -1.99073955e-01 6.11100733e-01 -3.47583503e-01
1.20492923e+00 -1.06515110e-01 1.16920340e+00 1.10184193e+00
4.19480443e-01 9.87185538e-01 1.32508898e+00 -3.01092952e-01
5.36864758e-01 -1.81329846e-02 -3.20035070e-01 3.85763645e-01
4.18786347e-01 1.00994408e+00 -5.00934243e-01 -2.78655469e-01
9.46520448e-01 -2.53269464e-01 3.78692091e-01 -5.51353216e-01
-6.80935740e-01 1.18272984e+00 8.58963281e-02 4.23256010e-02
-6.29574716e-01 3.71874630e-01 5.37911534e-01 1.17959654e+00
1.96511298e-01 1.05275202e+00 -5.69554925e-01 -7.37883568e-01
-7.52760112e-01 8.91885519e-01 9.06551957e-01 6.34664178e-01
1.90759882e-01 6.20540917e-01 -2.03484401e-01 6.51723802e-01
8.36638808e-02 3.43279272e-01 5.62759519e-01 -1.42272377e+00
3.95013869e-01 4.09126610e-01 3.56792569e-01 -6.16279542e-01
-2.58758724e-01 -6.41246438e-01 -1.50009409e-01 7.19458759e-01
4.39762086e-01 -8.08637977e-01 -3.92950088e-01 1.24429727e+00
2.87002506e-04 -3.69649172e-01 2.86262125e-01 5.17510295e-01
9.19890329e-02 2.16819495e-01 -4.33609858e-02 -1.79997802e-01
3.53777885e-01 -7.75262952e-01 -5.87278068e-01 -1.45688176e-01
4.22955394e-01 -6.01577282e-01 8.77907336e-01 6.54149711e-01
-1.67081738e+00 -2.82831401e-01 -1.15537548e+00 9.36370850e-01
-1.34023085e-01 -7.30123460e-01 1.03555810e+00 9.27993119e-01
-1.18213439e+00 1.12425387e+00 -5.67374885e-01 1.73912734e-01
6.06311500e-01 5.30159771e-01 3.41610909e-01 3.08282465e-01
-1.34401202e+00 1.16578460e+00 4.21687633e-01 -4.03161235e-02
-1.19856524e+00 -2.74125218e-01 -3.73695046e-01 6.06196783e-02
5.86653888e-01 -4.64187473e-01 1.87950683e+00 -1.37215638e+00
-1.95359933e+00 2.28681982e-01 9.05822277e-01 -8.83926392e-01
9.77175593e-01 -8.79775360e-02 -5.01202583e-01 -1.72048554e-01
-3.92096937e-01 3.05085003e-01 9.26237106e-01 -9.34167445e-01
-9.09389555e-01 -8.17707852e-02 3.29607755e-01 6.12351239e-01
-2.41486832e-01 2.16347218e-01 3.93225014e-01 -8.47076595e-01
-6.83904111e-01 -6.13844156e-01 -3.53533000e-01 -5.70093393e-01
4.15519267e-01 -1.46422207e-01 3.40852141e-01 -2.01324582e-01
1.11491776e+00 -1.75303984e+00 -2.26799492e-02 3.40492398e-01
-6.68171374e-03 7.95407966e-02 -3.23989421e-01 7.23968089e-01
-7.28664175e-02 1.38116390e-01 2.21755549e-01 2.98018396e-01
5.88488042e-01 9.21428800e-02 -4.61867392e-01 3.23305577e-01
1.55497715e-01 9.90513682e-01 -1.06527352e+00 -5.19033819e-02
2.22365689e-02 -2.34299928e-01 -6.75406218e-01 4.41218853e-01
-6.52010679e-01 -2.28636749e-02 -4.69188571e-01 1.02470613e+00
6.51241913e-02 -1.44446537e-01 1.71485975e-01 7.62123525e-01
-1.70929447e-01 4.57042664e-01 -1.11861002e+00 9.98767555e-01
-1.66681185e-01 5.35205483e-01 -1.38326317e-01 -1.07245350e+00
9.31908071e-01 3.78323466e-01 8.01658154e-01 -1.38783967e+00
1.68493643e-01 3.89263690e-01 4.71641779e-01 -2.17879221e-01
4.75618899e-01 -3.46971124e-01 2.85977069e-02 8.34668517e-01
1.02862867e-03 -2.04555213e-01 4.35570091e-01 -4.62698191e-01
1.21882248e+00 3.35972100e-01 9.94904712e-02 -1.86713025e-01
-4.21261154e-02 2.61625051e-01 5.48285186e-01 1.16430557e+00
-4.29138184e-01 1.46156386e-01 7.06977189e-01 -5.51860034e-01
-9.54288781e-01 -9.13983822e-01 1.24706782e-01 1.30095339e+00
-2.86867350e-01 9.53275114e-02 -6.29946470e-01 -7.61340439e-01
4.99070793e-01 7.16767073e-01 -5.74045539e-01 -2.73036689e-01
-6.11019671e-01 -7.24988937e-01 3.18012625e-01 6.18698120e-01
2.74880081e-01 -1.75552177e+00 -9.77096736e-01 8.58640373e-01
8.26303542e-01 -3.48514676e-01 -1.59114704e-01 6.06238544e-01
-9.74337220e-01 -1.08669412e+00 -8.03607404e-01 -4.24196064e-01
8.21183994e-02 -2.89966673e-01 1.55190802e+00 1.27276897e-01
-6.63322769e-03 5.51830590e-01 -3.26729208e-01 -6.74422741e-01
-4.40567732e-01 1.54748624e-02 1.37513191e-01 -5.77097952e-01
2.49381751e-01 -2.46762469e-01 -7.54329085e-01 2.37612069e-01
-7.31733799e-01 -6.55901968e-01 6.17395282e-01 1.22633994e+00
1.64309084e-01 3.19817215e-01 1.10670817e+00 -8.95435274e-01
1.43162107e+00 -4.75738853e-01 -1.06975543e+00 2.28583142e-01
-1.18712997e+00 5.46320379e-02 5.51986754e-01 -4.53810692e-01
-9.14612770e-01 -6.10741317e-01 -2.32385933e-01 -1.75446033e-01
2.31973663e-01 6.75736248e-01 3.61103535e-01 -3.19115937e-01
5.84651709e-01 4.94099269e-03 2.43561536e-01 -8.34775344e-02
-4.32146303e-02 2.57318556e-01 -1.95250854e-01 -7.92438626e-01
6.58772349e-01 -2.20399290e-01 -3.98838073e-01 -2.65642762e-01
-5.06848335e-01 3.85547996e-01 4.41512046e-03 -3.44614297e-01
1.73689112e-01 -6.79441869e-01 -7.86247909e-01 4.86866593e-01
-3.72173339e-01 -8.60606253e-01 -8.43181551e-01 3.74033004e-01
-9.85852242e-01 -1.41394615e-01 -6.74287558e-01 -8.77728164e-01
-3.57210100e-01 -1.19040811e+00 1.99283123e-01 4.26072776e-01
-2.38440841e-01 -1.19052851e+00 6.48213744e-01 2.06946265e-02
7.66090631e-01 1.82575732e-01 7.65387774e-01 -7.59442568e-01
-4.21636522e-01 -1.18573375e-01 2.53623515e-01 2.47204453e-01
6.62549287e-02 6.86512515e-02 -5.90919435e-01 -5.61810076e-01
8.25354271e-03 -9.22278345e-01 4.62364793e-01 3.93879920e-01
9.32532072e-01 -3.07469994e-01 3.85999382e-01 4.27394122e-01
1.32274699e+00 9.33580518e-01 2.52763718e-01 1.11676562e+00
-6.63249344e-02 5.16429722e-01 9.36438799e-01 9.24501419e-01
-1.57716781e-01 2.58881241e-01 5.20001531e-01 1.08130738e-01
4.76889461e-01 -1.16403833e-01 6.63978517e-01 4.68961179e-01
5.68661243e-02 -2.76185721e-01 -9.35181618e-01 3.35919321e-01
-1.94006920e+00 -1.26563382e+00 9.31465387e-01 1.89079726e+00
8.94056022e-01 6.70649707e-01 6.87882245e-01 -1.78238183e-01
2.79431909e-01 1.15218103e-01 -1.17040491e+00 -6.80403352e-01
-2.51500845e-01 5.03025472e-01 6.13329291e-01 1.96211666e-01
-8.12433779e-01 7.82946885e-01 8.21381760e+00 4.28344637e-01
-8.62867534e-01 -3.21691126e-01 7.91310310e-01 -2.71618843e-01
-6.10911250e-01 -1.01756968e-01 -5.37169635e-01 3.63514394e-01
1.04552674e+00 -5.68919718e-01 7.47888088e-01 9.08004105e-01
1.03676785e-02 1.20531201e-01 -1.06880534e+00 7.32547581e-01
-3.81060302e-01 -1.46860695e+00 -2.33224601e-01 1.30020648e-01
9.60945308e-01 5.99322729e-02 3.62035364e-01 9.72494066e-01
1.08911872e+00 -1.20545185e+00 7.82732844e-01 3.52168292e-01
1.30829304e-01 -1.28581691e+00 6.36703610e-01 -4.13472392e-02
-6.24313116e-01 -5.96104920e-01 -3.69457006e-01 -3.42980385e-01
-2.55540639e-01 -2.03140512e-01 -5.11719763e-01 2.17576727e-01
6.45658433e-01 4.66849536e-01 -4.00847733e-01 1.26861119e+00
-8.98233801e-02 7.82734334e-01 -1.37464598e-01 -6.70763135e-01
6.14637256e-01 -2.73000151e-01 2.72104591e-01 7.48876512e-01
2.29404569e-01 -3.01940441e-01 3.39649290e-01 6.36947215e-01
-9.48583260e-02 -3.26081850e-02 -6.68606460e-01 -6.03660285e-01
3.14773768e-01 7.86353052e-01 -5.65747499e-01 9.13952757e-03
-5.67492008e-01 4.54049021e-01 2.87926853e-01 4.36136395e-01
-7.98132300e-01 -2.24492580e-01 8.94858003e-01 6.71247765e-02
5.08100390e-01 -2.41064876e-01 -2.33482316e-01 -9.31215823e-01
-1.64773375e-01 -1.68021131e+00 5.93911171e-01 -1.51990771e-01
-1.64835918e+00 3.65415514e-01 -2.23284587e-01 -8.92928004e-01
-7.18726099e-01 -6.15713000e-01 -5.90902150e-01 5.36400080e-01
-1.50638890e+00 -1.68221995e-01 2.44786546e-01 4.50986564e-01
5.57474971e-01 -9.78131294e-01 6.50743127e-01 7.82224387e-02
-5.82070470e-01 7.12752521e-01 6.44240320e-01 -1.57273829e-01
5.04092991e-01 -1.49173415e+00 5.27080894e-01 3.41921240e-01
3.15301158e-02 3.42393398e-01 7.19111025e-01 -6.43049240e-01
-1.57352757e+00 -5.14725506e-01 3.65734361e-02 4.01005708e-02
1.28101897e+00 -1.19799022e-02 -4.56211209e-01 4.41999286e-01
7.36266017e-01 -6.20286107e-01 6.10742807e-01 1.40211463e-01
1.72704160e-01 -2.50957280e-01 -9.81974363e-01 8.06810975e-01
7.10036933e-01 -3.50749731e-01 -5.86251199e-01 2.22723663e-01
3.91897857e-01 -4.73857731e-01 -9.01610315e-01 1.65426970e-01
3.83345217e-01 -9.79640007e-01 8.85581017e-01 -8.47210467e-01
2.98499614e-01 2.16159344e-01 8.42055902e-02 -1.40245712e+00
-4.13552582e-01 -1.00548744e+00 -4.41095889e-01 9.93493855e-01
4.56404269e-01 -7.28162289e-01 1.17474234e+00 6.56837165e-01
1.69914559e-01 -9.93373275e-01 -6.34517252e-01 -9.60843265e-01
5.08882642e-01 4.11793515e-02 7.35642016e-01 8.35939229e-01
-6.10248484e-02 -6.67748228e-02 -2.74743497e-01 -5.42395711e-01
6.40233457e-01 3.86465132e-01 3.67267579e-01 -9.34816599e-01
-7.67044842e-01 -1.22146738e+00 6.72096387e-02 -4.85260725e-01
-6.12216555e-02 -4.47798252e-01 -3.08907241e-01 -1.29691648e+00
-2.43457288e-01 -3.33088189e-01 -9.06558633e-01 3.12944710e-01
-2.56922375e-02 -2.88067311e-01 2.18959019e-01 1.16468528e-02
-7.45714784e-01 8.61300349e-01 1.21082592e+00 -2.46155381e-01
-3.45651150e-01 3.03018034e-01 -9.21375871e-01 4.19113636e-01
1.27170706e+00 -4.22814041e-01 -5.60133576e-01 -3.26287627e-01
6.92763329e-01 3.99795502e-01 -1.54425949e-01 -9.88067627e-01
-2.15046108e-02 -5.04381180e-01 7.11788535e-01 -2.94767410e-01
-4.76500243e-02 -5.96953571e-01 5.21093458e-02 5.73067546e-01
-5.35117626e-01 9.72663462e-01 2.69806534e-01 3.15779418e-01
-3.55471909e-01 -4.68826383e-01 6.24445796e-01 -4.34638739e-01
-7.14587808e-01 3.64859998e-01 -6.60157323e-01 5.21780491e-01
1.17162824e+00 -3.26711833e-01 1.20963021e-04 -4.70578849e-01
-4.90951657e-01 5.57642102e-01 5.24505198e-01 4.41976696e-01
8.10267091e-01 -1.18099725e+00 -6.39796734e-01 -8.23215097e-02
-3.39086324e-01 -4.94624645e-01 -2.05616742e-01 1.05347194e-01
-7.53532350e-01 1.31755590e-01 -8.82891059e-01 1.57683149e-01
-5.70578277e-01 5.10157883e-01 9.58691239e-01 -5.50895631e-01
-4.69331890e-01 6.88687980e-01 -3.91458124e-01 -1.78983271e-01
6.35254145e-01 2.38581464e-01 -2.40086421e-01 3.15948606e-01
2.59556651e-01 4.30047810e-01 -3.09341662e-02 3.71126771e-01
1.30089065e-02 2.83299033e-02 -3.46466541e-01 -4.41803515e-01
1.71196568e+00 3.37210119e-01 3.02116871e-01 4.33878273e-01
4.88787383e-01 -5.37864506e-01 -1.62327862e+00 9.92686972e-02
6.77914739e-01 -3.96933913e-01 1.16359413e-01 -9.65230942e-01
-9.82235134e-01 5.39837778e-01 7.55042970e-01 5.68957150e-01
9.37720239e-01 -6.94687366e-01 6.61006331e-01 5.90884149e-01
5.44297397e-01 -1.83456457e+00 3.30873698e-01 8.93373489e-01
7.66280472e-01 -1.05025721e+00 2.91835871e-02 8.21049690e-01
-7.78852820e-01 1.26104033e+00 7.36135721e-01 -5.63999534e-01
6.51074886e-01 5.10558426e-01 2.17893526e-01 -2.98205137e-01
-1.23406029e+00 -8.17185566e-02 -2.64004320e-01 5.45293093e-01
4.20940667e-01 -1.22746751e-01 -4.18327115e-02 1.79294705e-01
-1.12667203e-01 -6.41010553e-02 4.04925823e-01 1.33939135e+00
-4.57873851e-01 -1.44503999e+00 -6.25509620e-02 5.68035066e-01
-7.27982521e-01 1.17472149e-02 -5.03722906e-01 9.98088062e-01
-3.99420798e-01 6.75790071e-01 1.08713759e-02 -2.87563473e-01
4.91515785e-01 -1.14269041e-01 5.97120225e-01 -3.27425092e-01
-1.42902708e+00 -1.23207504e-02 1.96064964e-01 -5.51155150e-01
-8.92273635e-02 -1.02157569e+00 -9.73994792e-01 -3.89895022e-01
-9.99372229e-02 2.24857166e-01 3.29064697e-01 7.12550342e-01
3.80390510e-02 6.56219125e-01 1.02370489e+00 -4.69655216e-01
-1.57464349e+00 -4.67139989e-01 -7.67593205e-01 2.28162527e-01
2.83173203e-01 -8.08922529e-01 -2.47017503e-01 -6.73549771e-01] | [4.370500087738037, 3.797865867614746] |
e622a9bc-7d74-4a92-aa04-8aa2661ac547 | a-novel-multi-scale-dilated-3d-cnn-for | 2105.02823 | null | https://arxiv.org/abs/2105.02823v1 | https://arxiv.org/pdf/2105.02823v1.pdf | A Novel Multi-scale Dilated 3D CNN for Epileptic Seizure Prediction | Accurate prediction of epileptic seizures allows patients to take preventive measures in advance to avoid possible injuries. In this work, a novel convolutional neural network (CNN) is proposed to analyze time, frequency, and channel information of electroencephalography (EEG) signals. The model uses three-dimensional (3D) kernels to facilitate the feature extraction over the three dimensions. The application of multiscale dilated convolution enables the 3D kernel to have more flexible receptive fields. The proposed CNN model is evaluated with the CHB-MIT EEG database, the experimental results indicate that our model outperforms the existing state-of-the-art, achieves 80.5% accuracy, 85.8% sensitivity and 75.1% specificity. | ['Mohamad Sawan', 'Jie Yang', 'Ziyu Wang'] | 2021-05-05 | null | null | null | null | ['seizure-prediction'] | ['medical'] | [-4.39420640e-01 -5.04804075e-01 4.36714500e-01 -1.84660986e-01
-1.45057350e-01 6.06657229e-02 1.12188756e-01 -1.89467028e-01
-6.19843900e-01 8.19637001e-01 9.44752768e-02 -1.94057673e-01
-4.53785002e-01 -4.19130474e-01 -2.63364255e-01 -6.17555737e-01
-9.49621737e-01 -5.11049628e-01 2.11366508e-02 7.04653412e-02
2.38190398e-01 8.93963993e-01 -1.10451448e+00 4.32822108e-01
8.06372344e-01 1.38704801e+00 2.02058583e-01 3.24969679e-01
3.16428810e-01 5.33909023e-01 -8.84709120e-01 2.55717844e-01
5.32670952e-02 1.18760042e-01 -5.44855118e-01 -2.20916823e-01
-5.45030236e-01 -3.52682292e-01 -7.79578090e-01 1.01625443e+00
7.26926625e-01 -5.79320192e-02 6.96597219e-01 -1.01070023e+00
-7.55329907e-01 4.51351047e-01 -4.42827731e-01 9.15112138e-01
1.42792076e-01 -5.70337884e-02 6.30491227e-02 -7.51788616e-01
-8.34594816e-02 5.92160642e-01 7.14425087e-01 4.51489627e-01
-7.66995370e-01 -1.13207519e+00 -1.88146323e-01 5.61281502e-01
-1.84832978e+00 1.33725554e-01 7.40270078e-01 -5.19259751e-01
1.50793540e+00 -7.73784742e-02 1.13443255e+00 1.18084443e+00
1.01103616e+00 4.91218507e-01 1.33310866e+00 2.83618877e-03
2.44231641e-01 -2.12228298e-01 3.58572811e-01 2.90189922e-01
-2.41618380e-02 1.89460397e-01 -6.78002238e-01 -2.08653450e-01
1.00304139e+00 2.96160460e-01 -6.93550050e-01 1.36633128e-01
-1.15013921e+00 5.08666754e-01 6.48271143e-01 8.27526271e-01
-9.36304331e-01 -4.63875793e-02 5.28414547e-01 1.78282663e-01
5.36769390e-01 4.17548269e-01 -5.89087903e-01 -4.26923156e-01
-7.58789480e-01 8.95503238e-02 4.90264356e-01 7.30069876e-01
1.43780902e-01 3.68543446e-01 -3.97635857e-03 6.07972324e-01
5.68381101e-02 2.35064492e-01 9.77976859e-01 -2.31178790e-01
2.49091372e-01 6.75517201e-01 -2.36663356e-01 -7.67349839e-01
-9.95676279e-01 -5.76280355e-01 -1.26268578e+00 -7.69757554e-02
-3.72281492e-01 -3.11002702e-01 -8.81949604e-01 1.16965842e+00
-3.62906665e-01 4.95766222e-01 1.32253215e-01 8.66194904e-01
8.44687521e-01 5.27516544e-01 1.00763150e-01 -8.26009586e-02
1.28565371e+00 -1.91177711e-01 -9.23892796e-01 9.62665677e-02
3.87836725e-01 -4.77684438e-01 5.49223483e-01 7.74902463e-01
-8.89445364e-01 -5.17436683e-01 -1.18854964e+00 5.04709721e-01
-4.69236791e-01 4.52887952e-01 6.39354825e-01 3.63814920e-01
-1.05896163e+00 5.70174396e-01 -1.05328679e+00 -1.67693824e-01
6.69064164e-01 8.50727260e-01 -5.31707466e-01 3.83589476e-01
-1.41227376e+00 1.00142467e+00 5.72984993e-01 1.63527235e-01
-6.88901186e-01 -6.84528410e-01 -5.17302096e-01 2.09291652e-01
-6.18441164e-01 -7.31209517e-02 8.63612115e-01 -4.74624783e-01
-1.20632935e+00 1.55333161e-01 6.23381585e-02 -6.55569077e-01
1.04826495e-01 -2.10723892e-01 -8.00997317e-01 3.26660305e-01
-2.40105703e-01 3.40918988e-01 4.76235300e-01 -2.79647976e-01
-4.81507540e-01 -5.28372645e-01 -1.59589991e-01 -8.48709140e-03
-6.10586941e-01 1.85919836e-01 4.55853678e-02 -8.80909681e-01
1.49847016e-01 -5.17223895e-01 6.51947930e-02 -5.49129248e-01
-1.63182780e-01 -3.10150415e-01 8.43693674e-01 -8.98060620e-01
1.24734020e+00 -2.15075350e+00 -3.74615705e-03 1.77043930e-01
3.96600068e-01 2.62828320e-01 3.09516758e-01 2.04370767e-01
-4.73270267e-01 -2.87404388e-01 -1.27401963e-01 7.79767036e-02
-2.76927233e-01 -3.00086886e-01 2.19387398e-03 6.59995496e-01
2.12183475e-01 7.86514521e-01 -4.25354779e-01 1.48724630e-01
4.09021616e-01 9.61454988e-01 -1.76281929e-01 2.18327463e-01
7.36741841e-01 5.34343779e-01 -5.73799789e-01 5.79161525e-01
7.82378614e-01 -1.75453290e-01 -2.95937389e-01 -2.52709359e-01
-1.22161545e-01 1.70200646e-01 -9.01436388e-01 1.57876325e+00
-3.32516462e-01 1.02983522e+00 -4.38673586e-01 -1.13676560e+00
1.03152633e+00 6.22292638e-01 6.67824268e-01 -8.19819510e-01
5.85992277e-01 2.33777225e-01 3.41133088e-01 -8.74583960e-01
-2.36902699e-01 3.44605774e-01 1.52293056e-01 3.23983908e-01
2.44822383e-01 3.82383585e-01 -3.14317435e-01 -1.39929950e-01
1.29324281e+00 -3.25882375e-01 3.02842557e-01 -7.56668627e-01
4.66424972e-01 -6.60212755e-01 5.42385638e-01 2.97982961e-01
-1.63248003e-01 1.27495602e-01 1.74300611e-01 -9.74237978e-01
-5.43052912e-01 -9.28911686e-01 -6.32763267e-01 2.41956905e-01
-3.64214182e-02 -8.69814232e-02 -1.05707085e+00 -4.02284652e-01
-4.67224978e-02 3.75465184e-01 -7.28061974e-01 -4.44998831e-01
-3.47584337e-01 -1.04255247e+00 5.38160741e-01 9.69295025e-01
9.89567697e-01 -1.26549804e+00 -1.06790400e+00 3.38810325e-01
7.33872131e-02 -1.02866697e+00 -7.43433088e-02 3.99447054e-01
-9.03216302e-01 -1.09448290e+00 -7.51119852e-01 -8.77192676e-01
5.05917728e-01 -2.66664237e-01 5.71894586e-01 -1.80228382e-01
-5.73233068e-01 5.55239916e-02 -5.42205453e-01 -6.31097198e-01
5.28485298e-01 1.64219085e-02 4.32647288e-01 4.77494784e-02
9.76378202e-01 -1.09161937e+00 -8.69405806e-01 3.45171541e-02
-7.57636428e-01 -1.59230441e-01 7.33804643e-01 7.48300850e-01
3.80019069e-01 1.61838070e-01 8.93495739e-01 -1.62297159e-01
1.16776800e+00 -4.74817127e-01 -4.81166154e-01 1.06127992e-01
-4.36092407e-01 -2.38382399e-01 7.96965837e-01 -7.09251404e-01
-5.71045756e-01 -2.42566429e-02 -1.45670205e-01 -4.44469184e-01
-3.72548342e-01 3.22174489e-01 1.08969249e-01 -5.52704930e-01
4.90087539e-01 6.17011726e-01 -6.29450917e-01 -5.25860250e-01
-4.97143179e-01 1.10686040e+00 4.22874033e-01 -1.01938397e-01
-1.18479936e-03 1.79081574e-01 -1.60166696e-01 -8.04111183e-01
-9.62365642e-02 -2.82396734e-01 -7.49348938e-01 -1.18053958e-01
9.83468831e-01 -1.01587903e+00 -8.85449469e-01 6.90637827e-01
-1.34776616e+00 -9.71097425e-02 3.23997349e-01 1.24361265e+00
-3.81110340e-01 -1.29148979e-02 -8.14693987e-01 -6.92946374e-01
-8.91040862e-01 -1.36946213e+00 5.15025616e-01 3.45394969e-01
-1.78162977e-01 -6.85072303e-01 -3.12419474e-01 -5.10601938e-01
6.37816012e-01 1.89332157e-01 6.87343121e-01 -7.83933163e-01
-3.59106362e-01 -4.40067559e-01 -2.66005814e-01 4.73653883e-01
2.09628001e-01 -4.13245857e-01 -1.07974660e+00 -2.66477764e-01
2.03685760e-01 -1.02814306e-02 5.59809625e-01 6.35368347e-01
1.69411719e+00 2.59310547e-02 -2.33913124e-01 1.06387568e+00
1.23713899e+00 8.54200780e-01 7.18951285e-01 3.97582173e-01
2.03393459e-01 4.10068035e-03 -6.97579384e-02 8.09055686e-01
3.17871749e-01 7.73568749e-02 2.87793696e-01 1.23362131e-02
2.95026273e-01 4.46665704e-01 -1.17156468e-01 1.07732511e+00
-3.06220859e-01 2.07309231e-01 -1.00216782e+00 6.14053607e-01
-1.50100362e+00 -5.83729923e-01 1.85684130e-01 1.68860769e+00
4.49640423e-01 1.24959223e-01 -2.07914978e-01 4.34789717e-01
5.16317546e-01 -3.18565279e-01 -5.43120682e-01 -2.38930121e-01
-2.78970562e-02 7.74950624e-01 4.42645967e-01 -8.22639018e-02
-1.22846317e+00 3.95329744e-01 6.52234936e+00 5.44318318e-01
-1.40916932e+00 1.64857849e-01 4.01769817e-01 -8.06495100e-02
4.32559639e-01 -6.36718750e-01 -5.00111103e-01 7.93071806e-01
1.00155294e+00 -2.38231659e-01 7.36863375e-01 5.58975637e-01
1.86195448e-01 5.37270755e-02 -8.64913464e-01 1.56723154e+00
-1.30385652e-01 -1.28815663e+00 -2.73668498e-01 -1.04932869e-02
4.16263163e-01 3.97458881e-01 -1.33372843e-02 1.56434014e-01
-3.92528713e-01 -1.21171367e+00 3.39823723e-01 1.00888133e+00
8.10885608e-01 -1.26137853e+00 1.01644850e+00 3.45641613e-01
-1.33087564e+00 -4.10100013e-01 -4.07248676e-01 -2.77743429e-01
-2.05398679e-01 4.59228724e-01 -5.92994928e-01 3.38340282e-01
1.32525551e+00 8.59361529e-01 -4.14063513e-01 1.40056026e+00
-5.10363244e-02 4.41189229e-01 -1.41237140e-01 -4.05491948e-01
2.16572970e-01 1.26823783e-01 2.11000443e-01 1.29948103e+00
5.82900047e-01 5.94879150e-01 -2.15260759e-01 7.40880430e-01
-8.59279260e-02 1.24570066e-02 -4.70258355e-01 7.81032294e-02
5.08092463e-01 1.09384930e+00 -5.55871308e-01 -8.86679739e-02
-4.69463617e-01 8.72864068e-01 2.20683157e-01 4.71974880e-01
-7.24546731e-01 -1.23048782e+00 4.71449256e-01 -4.12417442e-01
-1.43226516e-02 -1.87443346e-01 -4.02631342e-01 -1.15680397e+00
1.76905915e-01 -4.99889851e-01 2.34869439e-02 -9.55992162e-01
-1.18937361e+00 1.17019165e+00 -7.37201720e-02 -1.27744520e+00
-9.19774696e-02 -8.07057798e-01 -6.98745489e-01 1.23764074e+00
-1.46043587e+00 -6.50698423e-01 -2.95489937e-01 1.05629027e+00
4.09558028e-01 -5.15428662e-01 1.11468220e+00 3.88288349e-01
-4.65230852e-01 4.83314097e-01 1.14659347e-01 4.32365894e-01
2.55203992e-01 -8.81477058e-01 2.63862759e-01 6.66241169e-01
-3.56767625e-01 7.55227923e-01 2.64984787e-01 -4.24590021e-01
-1.19999266e+00 -1.02693808e+00 6.66306555e-01 2.16382444e-01
6.51507199e-01 -3.68830562e-01 -9.41614270e-01 5.90715230e-01
4.14651722e-01 1.31111041e-01 7.75675297e-01 -2.09399983e-01
3.49037200e-02 -3.28109443e-01 -1.29621243e+00 1.56780049e-01
6.03025854e-01 -6.00473285e-01 -7.07865179e-01 1.32380322e-01
3.38128269e-01 -3.16834927e-01 -1.32610703e+00 7.31142521e-01
7.01447248e-01 -7.83099234e-01 7.43538022e-01 -2.36055881e-01
-8.33546370e-02 9.50671956e-02 -6.09088019e-02 -1.61711681e+00
-4.78565425e-01 -4.09608662e-01 -2.36016124e-01 3.31764668e-01
1.50632754e-01 -8.50319862e-01 2.51133829e-01 4.11838561e-01
-4.00884867e-01 -1.37202942e+00 -1.09544289e+00 -6.27904654e-01
-1.15852274e-01 -6.17753386e-01 9.80637133e-01 6.44920647e-01
6.11410558e-01 -1.64512470e-01 -1.25588432e-01 3.95662874e-01
2.70670772e-01 -2.21793503e-01 -1.95357740e-01 -1.30236447e+00
2.39686087e-01 -3.21138233e-01 -9.28380489e-01 -6.00084245e-01
6.02025054e-02 -5.62980115e-01 -4.40106332e-01 -1.35759163e+00
2.00166225e-01 -3.96442153e-02 -9.62559760e-01 5.96340179e-01
2.37530366e-01 2.98693538e-01 -2.08802253e-01 -1.20142572e-01
-3.19047958e-01 5.83844662e-01 1.05707610e+00 -7.96260163e-02
-3.39970857e-01 -3.52405384e-02 -2.55846918e-01 7.44910777e-01
1.35499179e+00 -2.21150249e-01 -3.02782208e-01 -4.88381892e-01
-3.54663104e-01 5.80713227e-02 2.62841195e-01 -1.53837061e+00
3.89507890e-01 3.46383333e-01 1.25995624e+00 -7.18023181e-01
4.20887321e-01 -7.70971119e-01 -2.44277865e-02 6.54244244e-01
-1.21297583e-01 3.82613450e-01 5.50670207e-01 1.87624812e-01
-3.05256635e-01 2.74063200e-01 6.18573010e-01 3.57936621e-02
-6.96626067e-01 5.99485397e-01 -6.11306071e-01 -4.33424920e-01
1.22769964e+00 -1.08034618e-01 -6.29348457e-02 9.10952464e-02
-9.03893709e-01 5.09423204e-03 -2.78365821e-01 4.09511179e-01
1.23534334e+00 -1.51573396e+00 -4.19371039e-01 8.38431239e-01
1.48181856e-01 -4.71142620e-01 4.71921355e-01 1.04848659e+00
-5.58673739e-01 6.89667940e-01 -8.07221830e-01 -3.64408404e-01
-1.06566286e+00 1.58597067e-01 6.44548535e-01 1.74337715e-01
-9.96717453e-01 9.14520383e-01 5.72172143e-02 9.02301893e-02
6.35646701e-01 -5.19569755e-01 -6.65370226e-01 -4.19554114e-01
9.25720870e-01 2.81019956e-01 5.55962205e-01 -5.09558916e-01
-6.30289197e-01 3.55194241e-01 4.46129683e-03 1.37342528e-01
1.62176621e+00 3.25469464e-01 -2.43106872e-01 1.75987571e-01
1.30097413e+00 -6.86298728e-01 -1.07406867e+00 1.39470354e-01
-2.64514983e-01 -2.38124609e-01 3.77026498e-01 -1.03683770e+00
-1.19312215e+00 1.29209900e+00 1.10917020e+00 2.01111585e-01
1.68219912e+00 -2.79690534e-01 8.62345934e-01 5.58697999e-01
6.39115572e-01 -9.99690413e-01 -3.19582969e-01 5.77475011e-01
8.38211000e-01 -6.27351880e-01 -2.37228826e-01 2.04405859e-01
-3.93821239e-01 1.61595368e+00 5.28338671e-01 -6.20567322e-01
1.12271035e+00 5.72576940e-01 -1.87546119e-01 -2.93466240e-01
-5.53542018e-01 1.92956164e-01 5.75086594e-01 6.32670343e-01
3.76766741e-01 3.92596066e-01 -3.90079916e-01 1.40556872e+00
-2.33928576e-01 2.84095168e-01 1.98589548e-01 9.95383024e-01
-4.27371502e-01 -5.01943111e-01 -1.80512652e-01 7.55455315e-01
-8.73557746e-01 -1.90916941e-01 -4.13576327e-02 7.93589413e-01
3.60613585e-01 8.10756445e-01 2.35050410e-01 -8.74023080e-01
4.19494241e-01 9.68370363e-02 4.71791983e-01 -2.02357635e-01
-5.88911653e-01 2.40426347e-01 -5.17524421e-01 -6.18463933e-01
-2.41407022e-01 -3.39797676e-01 -1.41567707e+00 4.94888686e-02
-2.03028902e-01 1.75699949e-01 7.83315718e-01 8.88263524e-01
6.67035699e-01 1.03543890e+00 7.48710573e-01 -7.63714492e-01
-2.18821004e-01 -1.47346687e+00 -1.04809403e+00 -6.64493516e-02
5.28316796e-01 -8.35945010e-01 -2.24507451e-01 -7.64168948e-02] | [13.189807891845703, 3.4912312030792236] |
254d9393-f686-4183-b65e-9e6dd4d62dc3 | solving-visual-madlibs-with-multiple-cues | 1608.03410 | null | http://arxiv.org/abs/1608.03410v1 | http://arxiv.org/pdf/1608.03410v1.pdf | Solving Visual Madlibs with Multiple Cues | This paper focuses on answering fill-in-the-blank style multiple choice
questions from the Visual Madlibs dataset. Previous approaches to Visual
Question Answering (VQA) have mainly used generic image features from networks
trained on the ImageNet dataset, despite the wide scope of questions. In
contrast, our approach employs features derived from networks trained for
specialized tasks of scene classification, person activity prediction, and
person and object attribute prediction. We also present a method for selecting
sub-regions of an image that are relevant for evaluating the appropriateness of
a putative answer. Visual features are computed both from the whole image and
from local regions, while sentences are mapped to a common space using a simple
normalized canonical correlation analysis (CCA) model. Our results show a
significant improvement over the previous state of the art, and indicate that
answering different question types benefits from examining a variety of image
cues and carefully choosing informative image sub-regions. | ['Tatiana Tommasi', 'Bryan Plummer', 'Tamara L. Berg', 'Svetlana Lazebnik', 'Arun Mallya', 'Alexander C. Berg'] | 2016-08-11 | null | null | null | null | ['activity-prediction', 'activity-prediction'] | ['computer-vision', 'time-series'] | [ 2.96184540e-01 -3.36938798e-02 2.41488274e-02 -7.81894267e-01
-9.36505616e-01 -6.77965403e-01 9.41544414e-01 4.08561289e-01
-6.70598388e-01 5.30746400e-01 5.27926624e-01 -1.35458618e-01
-1.49191186e-01 -6.14497662e-01 -5.43560207e-01 -3.82545650e-01
4.03556824e-01 5.40232480e-01 3.28268856e-01 -6.07979409e-02
5.11958599e-01 5.15448272e-01 -1.76978266e+00 6.80257201e-01
3.75961542e-01 1.22287083e+00 1.74094081e-01 7.69523621e-01
-3.05521995e-01 1.12500083e+00 -6.44003510e-01 -3.31153303e-01
1.63351730e-01 -7.26155341e-01 -1.28726697e+00 2.55062342e-01
1.18625820e+00 -1.90854087e-01 -4.29845482e-01 6.98387980e-01
3.50109637e-01 3.59362572e-01 8.78486872e-01 -1.23427379e+00
-7.89945185e-01 -4.07787234e-01 -1.72336265e-01 6.82540298e-01
9.03871298e-01 4.40191150e-01 1.21321654e+00 -9.85146642e-01
9.11873221e-01 1.29667842e+00 2.06982240e-01 4.31398690e-01
-1.38382375e+00 -8.88975412e-02 2.28694916e-01 6.19228959e-01
-1.31466734e+00 -4.47845280e-01 6.42108619e-01 -6.39670014e-01
8.96418214e-01 5.07749975e-01 6.21003091e-01 9.69049871e-01
-1.08732972e-02 1.00934827e+00 1.41502476e+00 -4.45530474e-01
1.36916116e-01 3.45051974e-01 1.27165422e-01 8.85786712e-01
-1.75497115e-01 -1.82571903e-01 -6.53143704e-01 -2.27515534e-01
6.54063940e-01 -1.69542432e-01 -1.62269682e-01 -6.13178432e-01
-1.33823133e+00 9.22937036e-01 6.69892132e-01 2.19393939e-01
-3.78414482e-01 1.40908554e-01 1.35859594e-01 1.94184735e-01
2.23558754e-01 6.30987167e-01 -3.44936758e-01 1.35578409e-01
-9.45631623e-01 5.46291828e-01 6.51731551e-01 5.56509912e-01
1.03170514e+00 -3.40684801e-01 -8.95099163e-01 7.03600168e-01
3.05126250e-01 3.13194066e-01 7.95771852e-02 -1.28024423e+00
3.89686406e-01 8.15587342e-01 -3.00220605e-02 -1.13785553e+00
-5.19592762e-01 -6.72964975e-02 -2.92463183e-01 2.88986206e-01
1.02052355e+00 3.01783741e-01 -9.89226043e-01 1.52281916e+00
4.16111529e-01 -5.10125041e-01 -2.41436705e-01 1.12133229e+00
1.16068006e+00 4.26638842e-01 3.22580725e-01 2.62820721e-01
1.79598105e+00 -9.46337640e-01 -4.01379824e-01 -5.79059601e-01
2.63642430e-01 -6.24742448e-01 1.43048167e+00 4.88038994e-02
-1.07252705e+00 -7.20857561e-01 -7.18147039e-01 -7.29549587e-01
-7.06010044e-01 1.49638161e-01 4.11379635e-01 5.31630695e-01
-1.17955327e+00 1.83739010e-02 1.04143247e-02 -7.22052455e-01
5.73146403e-01 1.37882546e-01 -6.23336852e-01 -3.12700957e-01
-8.20233524e-01 1.07863677e+00 1.43630905e-02 -1.60515353e-01
-8.59426081e-01 -4.28081334e-01 -8.75814855e-01 -1.33106653e-02
3.66183937e-01 -9.56760228e-01 1.06718755e+00 -1.25174487e+00
-1.02095068e+00 1.38573241e+00 -4.28407580e-01 -3.35435033e-01
3.18783581e-01 6.38066754e-02 -6.36636540e-02 8.17017257e-01
4.08287227e-01 1.09434879e+00 8.77736032e-01 -1.26247334e+00
-4.98868287e-01 -5.75721264e-01 4.06406939e-01 3.25802267e-01
-8.84372219e-02 2.79856473e-01 -7.32245743e-01 -4.77071851e-01
-6.51474968e-02 -6.92882478e-01 -1.73867419e-01 4.19317573e-01
-2.79695958e-01 -3.76954943e-01 7.49190509e-01 -9.53856111e-01
7.76261210e-01 -1.74270427e+00 4.94056158e-02 3.24970752e-01
2.65381157e-01 -1.58774540e-01 -3.65580946e-01 1.65179655e-01
9.95052829e-02 -1.94985606e-02 5.89660667e-02 -8.61581117e-02
-1.34808049e-01 6.88797161e-02 1.24649610e-02 5.11558771e-01
3.68754387e-01 1.26605821e+00 -7.20595956e-01 -9.34846044e-01
3.14869523e-01 2.28716016e-01 -2.86879778e-01 2.45143786e-01
-4.70014393e-01 5.45686960e-01 -4.29854423e-01 8.68024349e-01
1.75674066e-01 -5.42213917e-01 -3.28587145e-02 -2.39109680e-01
2.49783993e-01 1.52179152e-01 -7.38390565e-01 1.59547377e+00
-2.35781133e-01 1.05745018e+00 -1.18533902e-01 -9.23928559e-01
8.45785081e-01 -5.16514555e-02 3.01674813e-01 -1.29157388e+00
1.04542010e-01 -3.16495538e-01 -2.68594772e-01 -9.22807753e-01
5.01790106e-01 3.44233960e-01 -4.66591120e-02 3.64429384e-01
3.19211453e-01 -2.36288663e-02 3.03857774e-01 3.42025012e-01
9.63651717e-01 2.63892829e-01 3.73809069e-01 -2.67996520e-01
7.75731623e-01 2.13573396e-01 1.14520499e-02 1.09531057e+00
-4.80033487e-01 9.65665758e-01 6.06125414e-01 -6.16546810e-01
-1.13958919e+00 -1.22088563e+00 1.79824308e-02 1.40905750e+00
1.73208237e-01 -4.08885717e-01 -5.99795282e-01 -7.84204245e-01
-6.83457702e-02 6.52389169e-01 -1.01854992e+00 2.78820902e-01
-4.73564804e-01 -3.74282390e-01 3.48132551e-01 5.07982910e-01
5.82891524e-01 -1.29710913e+00 -7.98539698e-01 -1.91520885e-01
-4.99648631e-01 -1.15599430e+00 -3.97336870e-01 -1.45112455e-01
-4.89723772e-01 -1.34365809e+00 -6.23351038e-01 -7.95819163e-01
6.44104898e-01 4.07343358e-01 1.52083206e+00 1.55319437e-01
-5.73978424e-01 1.16657484e+00 -1.22507758e-01 -1.82829499e-01
1.14656001e-01 -1.10733159e-01 -5.06255686e-01 2.84739912e-01
6.59851849e-01 5.79616316e-02 -8.35984588e-01 3.10910493e-01
-7.08044291e-01 -1.45648628e-01 4.56189871e-01 6.68190062e-01
7.84695923e-01 -7.04394162e-01 2.80087858e-01 -5.39815843e-01
6.47470891e-01 -4.09482747e-01 -3.40127468e-01 6.08897686e-01
-2.96827763e-01 1.08553968e-01 1.33222327e-01 -1.14248045e-01
-1.03122866e+00 1.41743302e-01 1.14342153e-01 -2.50096500e-01
-5.95509827e-01 5.43549769e-02 -6.83762878e-02 -1.69219926e-01
9.30994868e-01 1.70660153e-01 3.39714065e-02 -1.84469670e-02
5.81525028e-01 1.78850964e-01 6.29070461e-01 -3.87008280e-01
5.29074967e-01 6.69697583e-01 7.49406368e-02 -9.30093169e-01
-1.04943979e+00 -8.67349684e-01 -8.28148246e-01 -7.40705729e-01
1.45414865e+00 -7.53883004e-01 -1.00870514e+00 -4.89969924e-02
-1.09140658e+00 -1.96736045e-02 -2.41383299e-01 1.44974992e-01
-6.11150563e-01 3.38400364e-01 -2.43208557e-01 -9.05319035e-01
-8.66427645e-02 -9.64549959e-01 1.19711995e+00 3.50864321e-01
-4.63724166e-01 -8.53214920e-01 -1.20469168e-01 1.04628909e+00
3.28075111e-01 3.02373499e-01 1.01591015e+00 -6.40082061e-01
-9.72325385e-01 -1.03651315e-01 -5.33420026e-01 -4.12522852e-02
-3.87238085e-01 -4.04956609e-01 -1.07950401e+00 3.00709195e-02
-7.45779499e-02 -6.34346306e-01 1.13075125e+00 4.03949857e-01
1.40073609e+00 -2.20209748e-01 -2.54518211e-01 2.41726786e-01
1.38283074e+00 -1.03718713e-01 7.51629531e-01 5.46991706e-01
5.95560968e-01 9.74002659e-01 5.81888199e-01 5.43309152e-02
7.63302684e-01 6.81560755e-01 6.01212025e-01 -2.51410186e-01
-3.39516878e-01 -1.45248011e-01 6.58730194e-02 -2.94053435e-01
4.38265577e-02 -1.64311215e-01 -1.05644071e+00 7.95281768e-01
-1.70070851e+00 -1.11324418e+00 -3.24036628e-01 2.08246207e+00
5.03189981e-01 -2.20026985e-01 3.71570587e-01 -1.46523789e-01
4.64575797e-01 2.99787372e-01 -3.55048895e-01 -1.99058577e-01
-3.93733233e-01 1.32176384e-01 2.73642510e-01 2.83760011e-01
-1.18555367e+00 7.37994134e-01 7.29682112e+00 5.77430487e-01
-5.83160758e-01 6.73520751e-03 1.04111516e+00 -1.77651212e-01
-2.63379663e-01 8.17527175e-02 -5.53733289e-01 1.99491456e-02
6.88503742e-01 5.08121431e-01 3.63880575e-01 7.21832097e-01
1.66155212e-02 -6.94784760e-01 -1.06774139e+00 1.12417424e+00
7.69325614e-01 -1.44515097e+00 1.28241345e-01 -1.35775790e-01
5.07908762e-01 -1.52084783e-01 2.19059676e-01 -6.98360279e-02
4.71099317e-02 -1.30099440e+00 8.69540393e-01 1.05704463e+00
4.16359574e-01 -4.95877087e-01 4.74336535e-01 2.88480055e-02
-9.87978160e-01 -2.35168681e-01 -1.63861051e-01 1.10951560e-02
3.18107978e-02 8.05314109e-02 -8.99227917e-01 1.77076563e-01
1.08286250e+00 3.71062994e-01 -1.52276385e+00 9.90140557e-01
-1.32267565e-01 3.88116658e-01 -1.27337454e-02 -2.69836515e-01
1.57268316e-01 -5.42294979e-02 3.50047678e-01 1.05716348e+00
-8.49404112e-02 -8.42707325e-03 9.36053917e-02 9.36189592e-01
1.51017725e-01 3.74044657e-01 -5.86043477e-01 1.77917957e-01
7.78737804e-03 1.26267898e+00 -7.84374237e-01 -4.18314278e-01
-7.05119491e-01 8.87277186e-01 4.25306410e-01 5.87982714e-01
-4.58903760e-01 -1.96611688e-01 3.33742172e-01 2.79133260e-01
4.34362620e-01 -1.23635046e-01 -3.69531661e-01 -1.06484735e+00
2.20914464e-02 -9.43004370e-01 7.06697881e-01 -1.30017281e+00
-1.47905958e+00 5.35158932e-01 8.62509944e-03 -1.00869834e+00
-2.41953805e-01 -7.68401444e-01 -4.59496707e-01 7.98113108e-01
-1.35975957e+00 -1.46375048e+00 -6.21972442e-01 8.59033167e-01
5.16295612e-01 -2.31879324e-01 7.60139048e-01 -1.24121271e-01
-6.89955726e-02 3.26316655e-01 -2.08641008e-01 1.94748521e-01
6.90311670e-01 -1.27860081e+00 7.40258768e-02 7.17115998e-01
5.45358658e-01 4.10079777e-01 6.53633237e-01 -2.93005168e-01
-1.03421056e+00 -7.40737557e-01 1.07529664e+00 -1.01761079e+00
3.27746779e-01 -4.14946020e-01 -7.35066593e-01 4.35587853e-01
6.57113433e-01 -8.60706717e-02 7.10302711e-01 1.77424207e-01
-5.51705837e-01 6.88474476e-02 -1.14189744e+00 6.29121661e-01
7.55154431e-01 -9.63543952e-01 -6.45192981e-01 4.01554435e-01
1.60209134e-01 -2.81108152e-02 -6.45189345e-01 4.32381220e-02
5.06107152e-01 -1.05064046e+00 1.34722257e+00 -8.38726282e-01
4.56819296e-01 -3.63867998e-01 -3.44611406e-01 -7.15775013e-01
-4.49073792e-01 1.50654744e-02 1.15493238e-01 8.94174397e-01
3.61187011e-01 -3.01057305e-02 9.31686878e-01 6.53371811e-01
4.14090484e-01 -5.97545266e-01 -8.65301609e-01 -1.02937333e-01
-2.13310510e-01 -2.49923542e-01 1.46736637e-01 7.01873422e-01
-3.67034793e-01 8.08204710e-01 -4.09956723e-01 -1.03684217e-01
5.26151061e-01 1.87954634e-01 9.72427845e-01 -1.19748795e+00
-1.58023253e-01 -3.96658510e-01 -4.96050537e-01 -1.00701475e+00
1.22195348e-01 -7.92967916e-01 -3.90353054e-02 -1.94903767e+00
4.01484311e-01 1.00908488e-01 -4.55396995e-02 3.35208654e-01
-4.08045024e-01 5.68390429e-01 5.54540634e-01 8.51922035e-02
-1.07618248e+00 3.02408129e-01 1.23712981e+00 -3.63313854e-01
1.81799307e-01 -2.14570001e-01 -6.99530900e-01 6.61686182e-01
7.17621326e-01 -1.40792564e-01 -2.82468259e-01 -3.78480136e-01
2.50673532e-01 1.06512845e-01 1.04746151e+00 -9.87273812e-01
2.09770456e-01 -2.14825481e-01 1.06878817e+00 -7.82160580e-01
8.05479467e-01 -6.72976971e-01 -3.95035535e-01 1.01113126e-01
-6.34508967e-01 2.22530246e-01 4.72694635e-02 7.64027953e-01
-2.71238148e-01 -2.23482117e-01 6.57789409e-01 -4.38049078e-01
-1.14312768e+00 1.64739013e-01 -5.57796180e-01 2.20963761e-01
9.23120737e-01 -3.85938317e-01 -4.30177659e-01 -7.78385222e-01
-8.56966019e-01 2.71356255e-01 3.49872172e-01 6.05260074e-01
8.50256205e-01 -1.34845090e+00 -5.26584685e-01 -5.83557673e-02
5.89424551e-01 -5.42628169e-01 3.97526413e-01 6.16252124e-01
-4.76240993e-01 6.28707707e-01 -3.15848172e-01 -7.80516088e-01
-1.50286436e+00 6.42002940e-01 3.81217718e-01 -1.19993217e-01
-3.04515630e-01 7.88707495e-01 3.60174924e-01 -3.13782215e-01
3.43904532e-02 7.57033676e-02 -6.56792819e-01 2.41097286e-01
4.86448050e-01 1.88753754e-01 -2.00563654e-01 -1.05514693e+00
-4.57058787e-01 4.81287628e-01 7.55848065e-02 -3.94247144e-01
1.02931643e+00 -4.13482368e-01 -1.21959932e-01 3.61565799e-01
1.17762589e+00 -1.92806408e-01 -1.18861020e+00 -4.53963727e-01
5.41555360e-02 -7.31174231e-01 -1.78188205e-01 -9.61073935e-01
-7.89036214e-01 9.45713460e-01 7.76605606e-01 1.94459587e-01
1.08683789e+00 5.77190399e-01 1.17226936e-01 5.10539949e-01
4.11987044e-02 -1.22580242e+00 7.99638689e-01 3.53451282e-01
9.65592682e-01 -1.43066335e+00 9.08158123e-02 -1.42225221e-01
-1.05136776e+00 1.00324619e+00 8.89809310e-01 -3.90278511e-02
3.26764882e-01 -6.17408991e-01 2.49811903e-01 -5.67564905e-01
-6.86838627e-01 -5.61618984e-01 7.99188852e-01 6.97375119e-01
2.65415400e-01 -2.07982391e-01 -9.18099191e-03 8.74523073e-02
4.83214073e-02 -5.00345588e-01 3.26967031e-01 8.03398728e-01
-5.50456822e-01 -7.91947663e-01 -6.30636573e-01 8.12258840e-01
-4.18462723e-01 -1.38184549e-02 -7.08892882e-01 8.35560858e-01
1.78147987e-01 1.14783061e+00 2.30318457e-01 -1.40869785e-02
1.41016364e-01 2.19540849e-01 7.62978911e-01 -3.86616945e-01
-6.86694443e-01 -2.73872197e-01 2.08690599e-01 -7.91158438e-01
-7.81321168e-01 -9.24131334e-01 -7.00124323e-01 1.69498622e-01
3.28894332e-02 -2.68130928e-01 5.55793285e-01 1.09477198e+00
2.00681016e-01 2.40753323e-01 1.08337931e-01 -5.87292850e-01
8.68832618e-02 -7.57896602e-01 -2.31886774e-01 7.22170770e-01
3.84371132e-01 -5.90835750e-01 -4.40114737e-03 2.52268195e-01] | [10.858418464660645, 1.7601937055587769] |
dae54178-fd7c-4c85-aef8-6c8410756fb1 | visual-speech-enhancement-without-a-real | 2012.10852 | null | https://arxiv.org/abs/2012.10852v1 | https://arxiv.org/pdf/2012.10852v1.pdf | Visual Speech Enhancement Without A Real Visual Stream | In this work, we re-think the task of speech enhancement in unconstrained real-world environments. Current state-of-the-art methods use only the audio stream and are limited in their performance in a wide range of real-world noises. Recent works using lip movements as additional cues improve the quality of generated speech over "audio-only" methods. But, these methods cannot be used for several applications where the visual stream is unreliable or completely absent. We propose a new paradigm for speech enhancement by exploiting recent breakthroughs in speech-driven lip synthesis. Using one such model as a teacher network, we train a robust student network to produce accurate lip movements that mask away the noise, thus acting as a "visual noise filter". The intelligibility of the speech enhanced by our pseudo-lip approach is comparable (< 3% difference) to the case of using real lips. This implies that we can exploit the advantages of using lip movements even in the absence of a real video stream. We rigorously evaluate our model using quantitative metrics as well as human evaluations. Additional ablation studies and a demo video on our website containing qualitative comparisons and results clearly illustrate the effectiveness of our approach. We provide a demo video which clearly illustrates the effectiveness of our proposed approach on our website: \url{http://cvit.iiit.ac.in/research/projects/cvit-projects/visual-speech-enhancement-without-a-real-visual-stream}. The code and models are also released for future research: \url{https://github.com/Sindhu-Hegde/pseudo-visual-speech-denoising}. | ['C. V. Jawahar', 'Vinay Namboodiri', 'Rudrabha Mukhopadhyay', 'K R Prajwal', 'Sindhu B Hegde'] | 2020-12-20 | null | null | null | null | ['speech-denoising'] | ['speech'] | [ 1.25323132e-01 5.14757484e-02 1.31690567e-02 -1.60832092e-01
-1.08251715e+00 -3.47055078e-01 4.09847856e-01 -3.01331997e-01
-3.65737140e-01 6.77695811e-01 5.00621021e-01 -2.96534866e-01
2.03832611e-01 -3.37002754e-01 -7.71247089e-01 -7.94360757e-01
2.36320227e-01 -2.28452921e-01 3.26693714e-01 -3.83298516e-01
1.20502293e-01 3.16414028e-01 -1.97559428e+00 4.98942316e-01
7.14091778e-01 7.04372704e-01 6.27052248e-01 1.02095735e+00
8.39143395e-02 4.26031649e-01 -7.78668582e-01 -1.36205226e-01
3.45009387e-01 -5.22523642e-01 -3.07556808e-01 1.14406683e-01
3.07125926e-01 -4.33805317e-01 -5.16202629e-01 1.31455624e+00
1.16860414e+00 9.15109962e-02 2.38003016e-01 -1.40274727e+00
-4.69017029e-01 3.71497899e-01 -5.03345609e-01 3.29912896e-03
6.34787500e-01 4.87018466e-01 6.43766642e-01 -8.03014576e-01
5.41564167e-01 1.09044552e+00 5.82596242e-01 8.17734361e-01
-1.15705919e+00 -9.02702987e-01 -9.15633440e-02 2.74704188e-01
-1.18684685e+00 -1.29732764e+00 9.28982139e-01 -1.76940605e-01
6.63712323e-01 5.25109649e-01 5.10236740e-01 1.43809235e+00
-3.67462188e-01 9.58204567e-01 1.17957842e+00 -6.21334314e-01
7.61872344e-03 3.95765573e-01 -3.90875936e-01 2.58124381e-01
-2.14896664e-01 5.39363861e-01 -7.86950886e-01 2.87249625e-01
6.76783144e-01 -4.73284632e-01 -7.19376266e-01 -8.12246092e-03
-1.09380674e+00 4.73851383e-01 8.60027447e-02 5.38517177e-01
-2.89620459e-01 2.33632609e-01 3.46238762e-01 4.07946050e-01
6.90143228e-01 6.75604492e-02 -2.99740493e-01 -3.90924603e-01
-1.31781387e+00 3.72153968e-02 4.37394112e-01 9.16669905e-01
3.34495395e-01 4.19711769e-01 -4.07706089e-02 1.19673312e+00
4.08009946e-01 5.58903098e-01 4.51680630e-01 -1.19586229e+00
3.46416414e-01 -3.41660589e-01 1.15587793e-01 -6.22095585e-01
-9.16758999e-02 -2.55824417e-01 -6.66902602e-01 7.45236039e-01
4.63404387e-01 -4.41243798e-01 -9.20213461e-01 1.91049552e+00
2.85071105e-01 3.00828815e-01 9.12696943e-02 9.15226638e-01
9.36322689e-01 7.14855731e-01 -1.05798043e-01 -5.27671337e-01
1.09068918e+00 -1.03861558e+00 -1.13031268e+00 2.44275648e-02
1.90169349e-01 -1.22605038e+00 1.27690458e+00 6.09988570e-01
-1.44082117e+00 -7.38581181e-01 -1.02004099e+00 7.61122480e-02
-1.85288236e-01 1.10249862e-01 2.06681862e-01 9.34620619e-01
-1.57266855e+00 4.10421640e-01 -5.34065187e-01 -4.15394247e-01
2.27820411e-01 2.94044346e-01 -3.77406240e-01 1.74318001e-01
-1.14941001e+00 5.56879699e-01 -9.77357477e-02 -7.98277184e-02
-9.92634714e-01 -7.14047194e-01 -8.92537236e-01 -1.66604910e-02
3.20415735e-01 -3.44411612e-01 1.53591299e+00 -1.17868447e+00
-1.80952156e+00 6.59594059e-01 -4.02018934e-01 -2.77075559e-01
7.36298501e-01 -2.42117494e-01 -4.72764313e-01 3.53743970e-01
-1.71795830e-01 8.59589219e-01 1.13644123e+00 -1.67035234e+00
-6.40065074e-01 6.17295802e-02 -1.01227060e-01 2.03734621e-01
-2.14979097e-01 3.20127428e-01 -4.21751231e-01 -9.47695076e-01
-1.45417511e-01 -7.33722389e-01 8.76289606e-02 2.19356045e-01
-3.19628447e-01 7.21490607e-02 1.00513518e+00 -8.97053063e-01
1.06908417e+00 -2.30876946e+00 -3.74205470e-01 -2.28027031e-01
-1.82070434e-02 5.93453646e-01 -3.17057699e-01 4.92396474e-01
-3.83099109e-01 3.40091407e-01 -1.32821247e-01 -7.16055930e-01
-1.86738968e-02 -2.44733468e-01 -1.14921480e-01 4.32121754e-01
-1.34052828e-01 5.33689559e-01 -8.43510926e-01 -4.75792229e-01
4.39542711e-01 1.06328845e+00 -4.03359294e-01 3.09284896e-01
8.11219662e-02 4.88238513e-01 1.31688073e-01 5.20773411e-01
9.27348018e-01 4.58440959e-01 -1.61545888e-01 -2.55662024e-01
-2.12459281e-01 1.85259104e-01 -1.41763413e+00 1.71881497e+00
-5.74502349e-01 1.04673851e+00 6.43069208e-01 -6.70594394e-01
6.97293758e-01 9.31084871e-01 3.32932860e-01 -7.17754960e-01
-8.37163441e-03 3.16629767e-01 -7.32523575e-02 -7.97140718e-01
2.35118046e-01 -1.18019357e-01 6.51133239e-01 2.24004537e-01
1.49968207e-01 -3.30990046e-01 1.43192843e-01 1.16883419e-01
8.50460410e-01 2.17069343e-01 2.64421068e-02 -2.65188399e-04
5.39754033e-01 -7.29976118e-01 3.10625345e-01 5.76782167e-01
-6.38937235e-01 8.69657755e-01 3.52164567e-01 3.55134428e-01
-1.10422719e+00 -1.12625611e+00 -7.75010437e-02 1.19468510e+00
2.78698988e-02 -3.81978810e-01 -1.06869769e+00 -4.50307935e-01
-4.08324301e-01 7.68922865e-01 -3.20750684e-01 1.60801739e-01
-3.45147550e-01 -1.46984503e-01 7.31054664e-01 2.17871770e-01
3.50984305e-01 -1.38511229e+00 -3.23187888e-01 -6.16381690e-02
-4.89659369e-01 -1.15037942e+00 -5.01608014e-01 -9.88534391e-02
-4.48769182e-01 -6.35369062e-01 -1.17738032e+00 -7.38542736e-01
3.72606993e-01 3.59226197e-01 6.60989344e-01 1.22875482e-01
-8.90234038e-02 4.62876916e-01 -2.57756650e-01 -5.72898805e-01
-8.14792752e-01 -4.69092965e-01 1.85514480e-01 7.67229572e-02
-9.75847170e-02 -7.84155846e-01 -7.85010636e-01 2.17327803e-01
-8.61948550e-01 1.19749002e-01 3.34954500e-01 7.87040412e-01
2.00092733e-01 1.09190099e-01 9.09453273e-01 -2.50253290e-01
8.40823233e-01 -1.79989412e-01 -5.21865666e-01 -1.97049975e-01
-2.14722484e-01 -3.19235921e-01 3.69606525e-01 -6.71615779e-01
-1.28326595e+00 -1.69882178e-01 -7.92757809e-01 -5.17395198e-01
-6.57800019e-01 -2.65576765e-02 -4.06036496e-01 6.81834891e-02
5.34759521e-01 2.12029353e-01 2.54533410e-01 -5.77295065e-01
2.76903659e-01 1.16083944e+00 5.56856096e-01 -2.96458274e-01
6.96843624e-01 4.66705590e-01 -2.70407140e-01 -1.26960456e+00
-2.68954486e-01 -5.46240985e-01 -1.09831601e-01 -4.64514494e-01
6.13794446e-01 -8.77379119e-01 -7.83859193e-01 7.89997280e-01
-1.26415002e+00 -6.78470731e-01 -3.82833302e-01 6.05867505e-01
-8.79807949e-01 5.38088322e-01 -6.59837425e-01 -1.18852341e+00
-1.65074885e-01 -1.47619855e+00 9.13172603e-01 2.65704334e-01
3.78756449e-02 -8.29979777e-01 -1.08094200e-01 5.51655352e-01
5.92179656e-01 -1.93286911e-01 3.15799683e-01 -4.21319336e-01
-2.36520469e-01 1.28402606e-01 -9.65575203e-02 8.37559938e-01
1.60698578e-01 1.45049155e-01 -1.72006917e+00 -3.34716916e-01
1.15673833e-01 -1.78507894e-01 7.08743751e-01 6.76684797e-01
9.82172132e-01 -1.91232875e-01 -6.02384023e-02 4.87328500e-01
1.21897829e+00 2.97993660e-01 9.23534453e-01 9.70990807e-02
1.20734692e-01 8.20611894e-01 5.37531734e-01 2.99311250e-01
-9.82131213e-02 7.36347437e-01 4.89727259e-01 -5.11730909e-01
-9.19932067e-01 -3.34560841e-01 6.96030259e-01 6.86625242e-01
-6.01654388e-02 -6.01436317e-01 -5.32910347e-01 8.66651237e-01
-1.46957576e+00 -1.09179008e+00 -1.51551059e-02 2.30866480e+00
1.00308108e+00 1.02340519e-01 2.68927962e-01 5.53373218e-01
1.00104523e+00 9.55740735e-02 -8.07125345e-02 -3.10705632e-01
-9.36115310e-02 3.15617025e-01 6.84046298e-02 9.70984638e-01
-9.89728808e-01 9.33427036e-01 5.43612909e+00 1.17150915e+00
-1.34736538e+00 3.79765421e-01 4.64001805e-01 -1.53085947e-01
-1.13485590e-01 -3.13066870e-01 -4.63263363e-01 5.36565125e-01
9.89570856e-01 8.02310482e-02 3.43596309e-01 4.82802361e-01
1.01237547e+00 -1.88388884e-01 -7.55708575e-01 1.03642797e+00
-1.52053526e-02 -1.03924716e+00 -3.52597654e-01 6.58208951e-02
4.19894397e-01 -1.25828022e-02 3.59495163e-01 3.27738412e-02
-1.24230973e-01 -9.95534778e-01 8.12403738e-01 3.43969077e-01
1.22399628e+00 -5.71745872e-01 4.69939560e-01 2.92620242e-01
-1.09907079e+00 1.05189763e-01 2.20502331e-03 2.58437067e-01
3.72182310e-01 4.34150219e-01 -7.52097845e-01 3.65084171e-01
9.20462430e-01 2.56110400e-01 -2.36599967e-01 1.27241969e+00
-5.31914353e-01 8.74090612e-01 -3.69243473e-01 2.64487028e-01
-1.59456909e-01 2.17306927e-01 1.01962543e+00 1.46442425e+00
4.01402384e-01 -1.31592065e-01 -4.22821850e-01 6.91627204e-01
1.29071593e-01 1.44873962e-01 -7.52017379e-01 1.88995272e-01
3.22634995e-01 1.13460255e+00 -4.14539546e-01 -1.54607862e-01
-3.87216657e-01 7.71115124e-01 -3.51066202e-01 7.20190823e-01
-8.79620135e-01 -6.70528114e-01 7.38867521e-01 3.97541702e-01
1.90868065e-01 -7.44058788e-02 -4.42611426e-02 -8.13625216e-01
1.11774124e-01 -1.19363451e+00 -2.40949661e-01 -1.29064429e+00
-8.09121549e-01 7.64872491e-01 -7.42084831e-02 -1.08226430e+00
-2.58926421e-01 -4.16123509e-01 -5.28114617e-01 9.34600472e-01
-1.57033658e+00 -1.10622144e+00 -2.49953285e-01 7.61700988e-01
9.89482343e-01 -7.87911937e-02 6.12375021e-01 5.82128704e-01
-2.22479120e-01 8.32260191e-01 2.15970259e-02 -1.01090176e-04
1.20715559e+00 -9.68054891e-01 1.49754882e-01 1.08434057e+00
3.08971349e-02 1.60141319e-01 1.07921529e+00 -5.39515674e-01
-7.92724550e-01 -6.64137006e-01 8.10340106e-01 4.09555919e-02
5.56353986e-01 -5.47810733e-01 -8.25469792e-01 3.69535923e-01
8.29293609e-01 -4.06877622e-02 4.15121108e-01 -3.30274969e-01
-1.38423473e-01 -2.21516147e-01 -1.33073723e+00 7.03599155e-01
9.03446138e-01 -4.51554060e-01 -2.94989079e-01 1.96870968e-01
9.96166945e-01 -1.44534409e-01 -3.89774024e-01 3.84023577e-01
5.76955795e-01 -1.32784343e+00 8.94074738e-01 -1.54121697e-01
1.84175387e-01 -3.11734170e-01 -2.09449947e-01 -1.43335152e+00
3.58695328e-01 -1.26546407e+00 -2.68341638e-02 1.59182787e+00
3.69208366e-01 -6.36563599e-01 6.05081201e-01 -3.11864335e-02
-1.40606388e-01 -4.57936794e-01 -9.52512503e-01 -7.61304617e-01
-1.11528121e-01 -9.54205155e-01 2.86706477e-01 6.19712949e-01
6.69214036e-03 1.49208019e-02 -5.16924679e-01 1.83995649e-01
6.26948357e-01 -6.74624801e-01 7.13704884e-01 -7.44579673e-01
-3.03547174e-01 -3.60691011e-01 -2.65988618e-01 -9.68320906e-01
1.46492019e-01 -4.68938738e-01 3.20914239e-01 -1.48553216e+00
-1.53069243e-01 -1.87164620e-01 -3.43530178e-02 2.71800637e-01
4.85214218e-02 3.80698889e-01 5.09462953e-01 -1.44912869e-01
-8.11489597e-02 4.82445687e-01 1.27765012e+00 -7.77363107e-02
-2.93188095e-01 3.25813681e-01 -5.49392760e-01 7.50757396e-01
9.89617348e-01 -4.51913446e-01 -5.62090456e-01 -2.96252251e-01
-2.26672307e-01 2.38691449e-01 4.54168499e-01 -1.12729549e+00
3.95009011e-01 8.96521732e-02 3.69756632e-02 -3.47747266e-01
8.31609190e-01 -7.59116888e-01 -7.40979984e-02 2.16815129e-01
-2.75557876e-01 -2.88634866e-01 4.60133433e-01 2.53463626e-01
-2.77601272e-01 -3.04640532e-01 1.04618311e+00 3.00322920e-02
-4.52860415e-01 9.13687609e-03 -6.30150855e-01 -1.70450602e-02
7.80920744e-01 -2.72533655e-01 -2.91377097e-01 -1.10867572e+00
-9.73442733e-01 -2.47108176e-01 3.24231684e-01 3.72865677e-01
7.49862909e-01 -1.05930197e+00 -9.17170405e-01 3.36381733e-01
-3.29502672e-01 -4.46572900e-01 3.69253337e-01 9.72261488e-01
-3.50717992e-01 8.18671212e-02 -1.81379601e-01 -6.57796144e-01
-1.80557728e+00 5.67364514e-01 4.49079812e-01 1.72873482e-01
-4.75799561e-01 7.48180449e-01 2.85938829e-01 -1.34105161e-01
6.57516539e-01 -7.64851319e-03 5.47503605e-02 -4.50496040e-02
5.77684700e-01 3.62985879e-01 4.45505045e-03 -6.23835981e-01
-9.21895355e-02 3.17498147e-01 4.27935421e-01 -7.44213343e-01
1.28517377e+00 -4.23484683e-01 2.11057052e-01 3.39585096e-01
1.09081149e+00 6.35832965e-01 -1.17943895e+00 1.71016470e-01
-4.80943829e-01 -5.58720708e-01 1.28546789e-01 -9.08212066e-01
-1.11886454e+00 1.29855347e+00 9.74037766e-01 2.01619193e-01
1.55938709e+00 -1.78275421e-01 4.72222298e-01 -1.33606061e-01
-1.25231624e-01 -9.72572923e-01 3.26548159e-01 1.92246646e-01
1.30108106e+00 -1.23125017e+00 -4.43681151e-01 -5.63288867e-01
-6.43847525e-01 9.87806916e-01 3.60630155e-01 3.83601129e-01
7.49169230e-01 7.90708125e-01 6.30281448e-01 4.19794410e-01
-6.36981547e-01 -6.89314544e-01 3.31056155e-02 1.20564616e+00
6.11860573e-01 -1.82883039e-01 -1.15798183e-01 3.31085324e-01
-3.48234981e-01 8.56410712e-02 6.52484357e-01 7.04001427e-01
-3.48932534e-01 -1.10604382e+00 -6.33598924e-01 -2.08538383e-01
-7.01154947e-01 -3.35206866e-01 -8.67889747e-02 8.63575876e-01
1.95802748e-01 1.49778056e+00 -3.03251535e-01 -2.53246993e-01
2.90320933e-01 1.97932735e-01 3.36222917e-01 -2.58936793e-01
-4.68006700e-01 7.69982696e-01 1.05843127e-01 -4.67442840e-01
-4.15319800e-01 -5.41241646e-01 -9.53010976e-01 -2.98805475e-01
-3.00227106e-01 8.07031393e-02 9.86535966e-01 3.62658799e-01
2.85310835e-01 5.85057378e-01 4.90796685e-01 -1.11867678e+00
-2.85417080e-01 -1.08861935e+00 -4.25286680e-01 3.23419124e-01
7.40020335e-01 -5.22447288e-01 -7.68527389e-01 2.04924911e-01] | [14.576987266540527, 5.377188682556152] |
05c4cf9b-e85d-49b3-b5eb-569a37883789 | in-memory-hyperdimensional-computing | 1906.01548 | null | https://arxiv.org/abs/1906.01548v2 | https://arxiv.org/pdf/1906.01548v2.pdf | In-memory hyperdimensional computing | Hyperdimensional computing (HDC) is an emerging computational framework that takes inspiration from attributes of neuronal circuits such as hyperdimensionality, fully distributed holographic representation, and (pseudo)randomness. When employed for machine learning tasks such as learning and classification, HDC involves manipulation and comparison of large patterns within memory. Moreover, a key attribute of HDC is its robustness to the imperfections associated with the computational substrates on which it is implemented. It is therefore particularly amenable to emerging non-von Neumann paradigms such as in-memory computing, where the physical attributes of nanoscale memristive devices are exploited to perform computation in place. Here, we present a complete in-memory HDC system that achieves a near optimum trade-off between design complexity and classification accuracy based on three prototypical HDC related learning tasks, namely, language classification, news classification, and hand gesture recognition from electromyography signals. Comparable accuracies to software implementations are demonstrated, experimentally, using 760,000 phase-change memory devices performing analog in-memory computing. | ['Giovanni Cherubini', 'Abbas Rahimi', 'Geethan Karunaratne', 'Manuel Le Gallo', 'Luca Benini', 'Abu Sebastian'] | 2019-06-04 | null | null | null | null | ['news-classification'] | ['natural-language-processing'] | [ 5.07793605e-01 -2.54967034e-01 4.33060974e-02 7.92072564e-02
-2.44017139e-01 -4.80170846e-01 6.77015662e-01 1.02653824e-01
-8.53763282e-01 6.91785276e-01 -1.00903645e-01 -8.68161246e-02
-4.78007227e-01 -7.97896266e-01 -8.11856151e-01 -1.01834667e+00
-1.26475781e-01 4.26174760e-01 3.76548976e-01 -7.61448313e-03
5.72110593e-01 3.63712281e-01 -2.00524545e+00 3.65857214e-01
3.35259795e-01 1.19168746e+00 3.09884131e-01 6.46912336e-01
1.55393675e-01 4.63325799e-01 -5.47800601e-01 1.55494213e-02
-1.01209357e-01 -3.48663986e-01 -3.88909072e-01 -7.36510038e-01
-2.08320506e-02 2.05285456e-02 -6.96922243e-01 8.82981062e-01
7.17783749e-01 -2.54554744e-03 8.07375491e-01 -8.27026427e-01
-6.79705679e-01 6.06390178e-01 1.80397883e-01 2.98483640e-01
3.14437181e-01 4.45983946e-01 4.39827114e-01 -5.61343372e-01
8.16164553e-01 8.63421798e-01 6.34640992e-01 1.04910588e+00
-1.31455219e+00 -6.34835660e-01 -6.94391787e-01 1.06216103e-01
-1.21600497e+00 -4.18455601e-01 5.35138965e-01 -2.79412270e-01
1.36955154e+00 2.47495130e-01 1.09756184e+00 1.09217048e+00
9.09843087e-01 4.49919492e-01 1.50717103e+00 -3.58942181e-01
1.15744662e+00 -1.71669617e-01 1.99052095e-01 4.93014395e-01
4.74962384e-01 1.11728519e-01 -1.20895088e+00 -1.48139810e-02
6.40430868e-01 -1.06894197e-02 -1.17988929e-01 3.32253464e-02
-1.13090110e+00 -1.72285344e-02 3.33297879e-01 3.43734384e-01
-4.38160717e-01 6.80960238e-01 3.02997649e-01 4.07353133e-01
-3.01261008e-01 8.94567013e-01 1.80487245e-01 -7.59107471e-01
-6.64139748e-01 1.53700486e-01 9.00042236e-01 7.94994533e-01
2.08308294e-01 2.03296766e-01 9.98580381e-02 2.69053727e-01
5.54532558e-02 1.03486753e+00 1.07652092e+00 -9.13921595e-01
6.40172213e-02 6.80531085e-01 -1.90355822e-01 -6.05476499e-01
-6.55841768e-01 -1.60726026e-01 -9.84216988e-01 3.28827262e-01
5.33225834e-02 3.48198444e-01 -8.80428433e-01 1.74514019e+00
-1.67416453e-01 -7.06714466e-02 2.26313382e-01 9.70422626e-01
6.58366740e-01 3.47969830e-01 -4.10761833e-02 -1.76191896e-01
1.07350171e+00 2.07512453e-01 -6.70503914e-01 -4.42446657e-02
3.73256952e-01 -1.11131817e-01 1.02709317e+00 4.06274348e-01
-1.25798082e+00 -1.35969386e-01 -1.40084314e+00 -5.18152975e-02
-6.13890171e-01 -5.42235672e-01 5.19102812e-01 8.38850498e-01
-1.09522903e+00 9.22938585e-01 -1.17996836e+00 -2.74438053e-01
2.95112222e-01 8.48447382e-01 -1.54080167e-01 3.94591272e-01
-1.03163743e+00 8.65146041e-01 3.24683934e-01 -4.66907360e-02
-3.67574781e-01 -4.02076006e-01 -1.51774168e-01 -1.03194229e-01
-5.28857410e-01 -6.81543171e-01 7.01784194e-01 -1.69842154e-01
-1.84107268e+00 1.03432500e+00 4.53757346e-02 -6.38432384e-01
-7.25051910e-02 1.71431527e-01 -4.64564174e-01 1.04275495e-01
-4.33441877e-01 5.92300653e-01 8.46188962e-01 -5.53096175e-01
1.47904783e-01 -8.26678038e-01 -5.25201917e-01 -2.94923075e-02
-7.45829940e-01 -6.10540509e-01 1.40966341e-01 -3.14210296e-01
7.66715944e-01 -1.31204140e+00 3.49302769e-01 6.27330542e-02
-1.15621351e-01 1.31805241e-01 8.60180080e-01 6.97942823e-02
1.11005640e+00 -2.28253961e+00 3.76106977e-01 5.20043015e-01
3.97987545e-01 2.34466329e-01 1.42926723e-01 5.10950267e-01
6.64952874e-01 -1.88568428e-01 -5.69193959e-02 2.28275269e-01
1.11353785e-01 1.95970848e-01 -3.13695848e-01 6.45771742e-01
7.91948214e-02 1.28112090e+00 -5.83333015e-01 -1.37228638e-01
4.24007066e-02 4.13831502e-01 -4.40833181e-01 -2.45450631e-01
-9.68596563e-02 3.82228971e-01 -3.38757098e-01 6.37605846e-01
2.14239791e-01 -5.56306183e-01 6.04743004e-01 3.83522660e-02
-1.25867471e-01 5.03881991e-01 -9.98915434e-01 1.82914162e+00
-2.82602251e-01 8.18784535e-01 -1.95343420e-01 -7.44538963e-01
9.32698727e-01 2.21548736e-01 2.96406806e-01 -1.40148687e+00
2.07396701e-01 7.09405601e-01 2.20478058e-01 -3.85633111e-01
5.53481996e-01 -3.25388499e-02 -8.48139077e-02 9.91329432e-01
1.89434275e-01 -4.04915839e-01 -1.52701020e-01 -1.04675613e-01
1.51187754e+00 -3.34953099e-01 4.22390103e-02 -7.97975242e-01
-1.05257764e-01 -7.94482827e-02 8.32061917e-02 8.99801791e-01
-1.75842285e-01 4.62885469e-01 4.20683157e-03 -3.98125142e-01
-1.00938702e+00 -1.45243227e+00 -4.64044124e-01 4.39167857e-01
4.46549833e-01 -1.40622690e-01 -7.49942899e-01 7.36872733e-01
1.46948352e-01 1.19976863e-01 -3.78834873e-01 -6.18992686e-01
-6.55453265e-01 -6.16091430e-01 9.01790202e-01 5.11714041e-01
5.38521647e-01 -1.06972623e+00 -1.53130627e+00 4.38080192e-01
3.92904311e-01 -9.89412487e-01 3.19561988e-01 6.80514336e-01
-1.45464849e+00 -6.88217461e-01 -3.90567243e-01 -7.83017635e-01
2.29021996e-01 -2.26685017e-01 8.18979561e-01 -7.81248659e-02
-7.66684413e-01 5.21624327e-01 1.36567160e-01 -2.01115012e-01
-2.92387813e-01 3.01760346e-01 7.02286661e-01 -6.01557195e-01
4.69148546e-01 -1.22518444e+00 -9.19847608e-01 -7.56049380e-02
-9.45154369e-01 5.96399745e-03 6.65490627e-01 8.18242371e-01
6.59122765e-01 -3.07475869e-02 5.58218420e-01 -6.06225252e-01
7.23339498e-01 -1.37358651e-01 -1.79743052e-01 1.28246650e-01
-7.02004135e-01 4.57103997e-01 5.21524668e-01 -6.61652625e-01
-3.69179934e-01 -5.83066940e-02 1.34224862e-01 3.89099456e-02
1.53835535e-01 2.21913770e-01 2.18514830e-01 -3.03034604e-01
9.00779068e-01 7.26671934e-01 3.98807496e-01 6.90737441e-02
-4.46132151e-03 9.64670479e-01 7.09311485e-01 -6.75301671e-01
2.53207356e-01 5.34693241e-01 4.87180084e-01 -1.14628577e+00
3.64421517e-01 1.58221677e-01 -4.90694016e-01 -2.63012499e-01
5.22051036e-01 -6.62215889e-01 -1.36129427e+00 9.36683416e-01
-9.04877365e-01 -3.66444290e-01 -3.02334875e-01 5.53788185e-01
-8.96846294e-01 -5.31978250e-01 -1.07127166e+00 -9.81595159e-01
-6.27415836e-01 -9.58411455e-01 9.43415046e-01 4.17738557e-01
-4.41742539e-01 -6.84484959e-01 5.73826022e-02 -1.63815290e-01
8.53295147e-01 3.34849022e-02 1.50263858e+00 -1.66178390e-01
-7.54308641e-01 -2.13636920e-01 3.83136272e-01 -3.23783994e-01
-1.50495633e-01 -5.47028303e-01 -1.14639866e+00 -4.18724418e-01
1.61735918e-02 -6.37062073e-01 6.90728307e-01 1.27804846e-01
8.70487750e-01 3.65997441e-02 -3.81895453e-01 3.12159717e-01
1.40537143e+00 2.50295430e-01 9.39892113e-01 1.98297441e-01
1.72417596e-01 1.84848025e-01 -2.58777082e-01 3.55954200e-01
4.96565513e-02 5.09317279e-01 9.76417363e-02 6.58749223e-01
-7.60774612e-02 -2.90845722e-01 3.43375653e-03 1.43516529e+00
-2.55566835e-01 -2.56169084e-02 -9.06466067e-01 3.21813882e-03
-1.40460026e+00 -9.19533014e-01 1.13861345e-01 2.43725610e+00
9.02684093e-01 3.55105072e-01 -2.64838070e-01 5.39678574e-01
6.28080904e-01 -1.12058364e-01 -1.05188560e+00 -3.52275729e-01
-5.07252932e-01 7.66244650e-01 3.77833694e-01 -3.06990087e-01
-5.09100258e-01 5.48865199e-01 6.62594032e+00 4.77799743e-01
-1.50251448e+00 1.71671629e-01 2.04693571e-01 -5.36282182e-01
-3.21420372e-01 -4.91321355e-01 -5.01383901e-01 7.89259136e-01
1.29174936e+00 1.93837926e-01 8.46387863e-01 3.21908742e-01
-3.78812671e-01 -3.33918929e-01 -1.22319698e+00 1.46265757e+00
-3.44264805e-01 -1.75699580e+00 -1.00258403e-01 1.06386401e-01
7.33962297e-01 4.07102630e-02 6.65270686e-01 -7.61154853e-03
-5.47756076e-01 -9.44178581e-01 8.01485896e-01 6.90044105e-01
1.14860475e+00 -1.90178558e-01 2.80203551e-01 2.72887737e-01
-8.58073652e-01 -4.29883569e-01 -3.11340630e-01 -3.93514752e-01
-1.79523811e-01 5.23249745e-01 -1.28133193e-01 -4.44829166e-01
7.05344617e-01 2.41800368e-01 -1.66884944e-01 6.59471989e-01
4.34277892e-01 3.72700185e-01 -5.69170296e-01 -9.02996898e-01
-1.41227007e-01 -6.85305241e-03 4.85715955e-01 8.01923931e-01
3.52112025e-01 3.96812320e-01 -7.87161469e-01 1.14717507e+00
-1.72712088e-01 -4.72246379e-01 -8.12109530e-01 -5.77674508e-01
1.02736485e+00 7.88659215e-01 -9.71534193e-01 -6.93455711e-02
2.45441705e-01 9.95017588e-01 1.19921692e-01 1.75582334e-01
-3.41774046e-01 -6.27900243e-01 5.52058756e-01 3.41281384e-01
-5.55230305e-02 -6.88065767e-01 -7.97875822e-01 -6.95671737e-01
4.06097472e-01 -4.92830694e-01 -2.83945143e-01 -5.07977426e-01
-8.05252612e-01 3.59479040e-01 -6.49553120e-01 -1.21839166e+00
-4.47999656e-01 -9.15189445e-01 -3.92413914e-01 5.23769081e-01
-6.74284875e-01 -4.68071520e-01 -7.99947679e-02 4.18838710e-01
-1.33753985e-01 -2.44713843e-01 1.27021778e+00 2.03508824e-01
-1.71866357e-01 5.14454305e-01 6.22891068e-01 -2.78353512e-01
1.62174582e-01 -9.39590216e-01 5.00665128e-01 2.71559238e-01
3.97263048e-03 1.02613807e+00 4.11449254e-01 -6.14228249e-01
-2.72117805e+00 -6.34682119e-01 5.52492201e-01 -3.16884577e-01
5.77705443e-01 -7.32319832e-01 -8.08342755e-01 -5.46144657e-02
-4.70807254e-01 2.63697864e-03 5.22232294e-01 -5.17330244e-02
-3.76557350e-01 -1.43825233e-01 -1.15072417e+00 4.70834017e-01
1.39741004e+00 -1.26488662e+00 -3.42984021e-01 1.00746065e-01
2.63719618e-01 -5.46095550e-01 -9.78472054e-01 2.11366877e-01
1.31204641e+00 -7.19786942e-01 7.03372002e-01 -3.65880847e-01
2.95681477e-01 3.54248136e-02 -4.54497397e-01 -9.36176538e-01
-8.05816352e-02 -7.01186776e-01 -4.90017116e-01 6.43788636e-01
3.03699136e-01 -1.00647449e+00 8.80074084e-01 8.51445436e-01
2.58434378e-02 -7.81510592e-01 -1.41262436e+00 -1.06769931e+00
1.29072741e-01 -2.10033849e-01 4.88613725e-01 4.28664863e-01
6.31419659e-01 7.41204843e-02 2.13323280e-01 -1.35995716e-01
6.26023769e-01 2.22297117e-01 2.43172869e-02 -1.16496229e+00
-4.09054637e-01 -4.13683087e-01 -1.23325920e+00 -8.13074589e-01
8.58128220e-02 -1.04370964e+00 1.24269411e-01 -8.65058661e-01
2.14638442e-01 -6.48257136e-01 -5.05773365e-01 2.06736382e-02
4.28118318e-01 5.87801635e-01 4.46024686e-02 7.92153299e-01
-6.67268157e-01 4.21947956e-01 8.33005428e-01 -1.17469944e-01
-5.01074910e-01 -3.04314703e-01 -4.88339774e-02 6.86051399e-02
6.52570605e-01 -4.12985802e-01 -2.11637571e-01 -6.14156544e-01
4.25818533e-01 8.70208591e-02 5.09843886e-01 -1.93262804e+00
1.00467181e+00 4.16075051e-01 5.45049965e-01 -2.79530790e-02
6.96388960e-01 -3.90230745e-01 4.68398511e-01 1.03329384e+00
-5.59652746e-01 1.89084589e-01 1.13742695e-04 5.69946826e-01
9.92948562e-02 2.34226529e-02 6.22630000e-01 1.13960676e-04
-8.09298992e-01 -2.50537157e-01 -8.19095016e-01 1.42074659e-01
7.46288478e-01 -6.55784905e-01 -8.72651577e-01 8.84547085e-02
-5.09753585e-01 -3.20820481e-01 5.00092089e-01 1.65984049e-01
8.56163979e-01 -1.34222329e+00 9.88582894e-02 6.39281034e-01
-9.02117938e-02 -5.46084106e-01 2.32605636e-01 6.92008018e-01
-4.63655740e-01 7.77398229e-01 -7.64439583e-01 -8.64162564e-01
-7.49719501e-01 1.61801770e-01 1.57006770e-01 4.17258173e-01
-8.18521500e-01 4.63401943e-01 -6.20861292e-01 -2.17039257e-01
1.69847935e-01 -1.70407489e-01 5.69886923e-01 -4.32060957e-01
5.62376022e-01 4.92953777e-01 4.74809855e-01 3.01126484e-02
-4.58110332e-01 6.23002350e-01 2.48655766e-01 -4.85441238e-01
1.13982928e+00 1.91175625e-01 -2.09866911e-01 9.24985588e-01
1.05200279e+00 -9.23671901e-01 -9.39825714e-01 -2.31004372e-01
1.98836312e-01 8.22971761e-02 4.60904883e-03 -7.03122437e-01
-6.26870155e-01 1.00450575e+00 1.21357715e+00 5.01901135e-02
1.05206358e+00 -9.71064270e-02 9.68250096e-01 1.08196819e+00
1.11981416e+00 -1.47290921e+00 1.84997529e-01 4.62673128e-01
3.45725149e-01 -6.37868106e-01 -2.07543507e-01 2.09246248e-01
5.64344041e-02 1.17191565e+00 3.49493772e-01 -2.61512727e-01
7.42489159e-01 7.80901670e-01 -3.93747598e-01 -4.23058361e-01
-8.74394178e-01 1.52553082e-01 -1.19717434e-01 5.48205197e-01
3.60253274e-01 3.44261944e-01 -2.84005046e-01 3.82884204e-01
-2.89644808e-01 2.64680505e-01 3.41294050e-01 1.52854395e+00
-6.89492941e-01 -5.66899180e-01 -8.80061369e-03 8.23741078e-01
-2.18161102e-02 -5.22903986e-02 -2.40527675e-01 4.49163020e-01
-3.40487152e-01 4.46094692e-01 4.83038753e-01 -9.62989569e-01
2.96162311e-02 2.95281768e-01 9.41245317e-01 -1.88087493e-01
-4.59467649e-01 -7.14187682e-01 -4.60350990e-01 -6.20929837e-01
-2.31764719e-01 -3.25203389e-01 -1.77220976e+00 -5.73339820e-01
-1.32739916e-01 -2.30586842e-01 1.19981301e+00 9.93598998e-01
7.90631592e-01 3.37034762e-01 1.01006761e-01 -8.06212664e-01
-7.33054101e-01 -6.42490923e-01 -1.04712367e+00 3.24895144e-01
5.86269004e-03 -6.95800126e-01 -3.52564789e-02 -3.32627773e-01] | [8.219341278076172, 2.478900909423828] |
7df031ac-d05d-4871-97da-afb1a13dea45 | combining-chest-x-rays-and-ehr-data-using | 2108.12530 | null | https://arxiv.org/abs/2108.12530v2 | https://arxiv.org/pdf/2108.12530v2.pdf | Combining chest X-rays and electronic health record (EHR) data using machine learning to diagnose acute respiratory failure | Objective: When patients develop acute respiratory failure, accurately identifying the underlying etiology is essential for determining the best treatment. However, differentiating between common medical diagnoses can be challenging in clinical practice. Machine learning models could improve medical diagnosis by aiding in the diagnostic evaluation of these patients. Materials and Methods: Machine learning models were trained to predict the common causes of acute respiratory failure (pneumonia, heart failure, and/or COPD). Models were trained using chest radiographs and clinical data from the electronic health record (EHR) and applied to an internal and external cohort. Results: The internal cohort of 1,618 patients included 508 (31%) with pneumonia, 363 (22%) with heart failure, and 137 (8%) with COPD based on physician chart review. A model combining chest radiographs and EHR data outperformed models based on each modality alone. Models had similar or better performance compared to a randomly selected physician reviewer. For pneumonia, the combined model area under the receiver operating characteristic curve (AUROC) was 0.79 (0.77-0.79), image model AUROC was 0.74 (0.72-0.75), and EHR model AUROC was 0.74 (0.70-0.76). For heart failure, combined: 0.83 (0.77-0.84), image: 0.80 (0.71-0.81), and EHR: 0.79 (0.75-0.82). For COPD, combined: AUROC = 0.88 (0.83-0.91), image: 0.83 (0.77-0.89), and EHR: 0.80 (0.76-0.84). In the external cohort, performance was consistent for heart failure and increased for COPD, but declined slightly for pneumonia. Conclusions: Machine learning models combining chest radiographs and EHR data can accurately differentiate between common causes of acute respiratory failure. Further work is needed to determine how these models could act as a diagnostic aid to clinicians in clinical settings. | ['Michael W Sjoding', 'Jenna Wiens', 'Ella Kazerooni', 'David Fouhey', 'Sarah Jabbour'] | 2021-08-27 | null | null | null | null | ['respiratory-failure'] | ['medical'] | [-1.46326929e-01 -9.19543356e-02 -4.13235992e-01 -1.91449508e-01
-9.57601905e-01 -7.55760550e-01 -6.96047544e-02 6.52289629e-01
-2.98636854e-01 4.75386024e-01 3.36774170e-01 -8.23534012e-01
-5.91033816e-01 -5.52911639e-01 -2.29082063e-01 -3.79698545e-01
-8.22175667e-02 9.76760983e-01 -1.02228217e-01 8.26729715e-01
-1.01222932e-01 3.48483503e-01 -7.69777536e-01 2.50484705e-01
7.27822721e-01 7.51836538e-01 2.06403464e-01 1.12270105e+00
1.34319022e-01 1.21595204e+00 -2.96439648e-01 1.30213991e-01
2.10794821e-01 -7.83225715e-01 -5.77667892e-01 1.39997452e-01
2.30408847e-01 -8.19188476e-01 -2.24701002e-01 6.33873120e-02
6.66939676e-01 -3.91342193e-01 1.00828636e+00 -6.02896094e-01
-3.38747859e-01 1.16965696e-01 -1.99323535e-01 3.02331418e-01
3.75263214e-01 4.74008173e-01 6.78589761e-01 -4.76461321e-01
4.20918107e-01 5.96577585e-01 1.00502455e+00 6.11077487e-01
-1.31542706e+00 -6.12165034e-01 -4.38454121e-01 -1.00815721e-01
-1.21540153e+00 6.32279217e-02 -4.24613178e-01 -8.52356851e-01
7.94284165e-01 4.69786167e-01 9.49783444e-01 2.64979333e-01
4.10853386e-01 -5.85626140e-02 1.05573857e+00 -1.64084852e-01
-1.26694724e-01 2.08147019e-01 2.30583429e-01 7.40332663e-01
5.31648040e-01 -5.44537827e-02 3.33213061e-01 -8.83498311e-01
8.39874923e-01 7.37728775e-01 -4.89428520e-01 3.07977647e-02
-1.35161328e+00 6.22355819e-01 3.79636317e-01 6.93277791e-02
-5.43444633e-01 -2.37815693e-01 2.20145807e-01 2.68984735e-01
-3.60204458e-01 5.52343428e-01 -5.27257979e-01 -1.10062301e-01
-7.66595781e-01 -7.75435418e-02 8.71148586e-01 3.09769839e-01
1.50694713e-01 -3.03699285e-01 -3.58287334e-01 1.15288150e+00
4.77176905e-01 8.61575902e-01 5.28392613e-01 -1.22390532e+00
2.38148704e-01 6.86692238e-01 3.52484852e-01 -4.43287522e-01
-9.63320494e-01 -5.97196400e-01 -9.01682675e-01 -2.20360339e-01
3.46497804e-01 -2.88276136e-01 -8.77138674e-01 1.24971867e+00
-2.78337654e-02 -1.90819934e-01 2.25168183e-01 7.01966226e-01
7.35020280e-01 4.53809649e-01 4.15586472e-01 -3.93203020e-01
1.83392477e+00 -6.08820081e-01 -4.53583270e-01 -7.33369589e-03
1.01412976e+00 -7.63675928e-01 8.86362195e-01 1.75880924e-01
-1.21002758e+00 -2.42356613e-01 -4.41704631e-01 6.51747406e-01
2.63998955e-01 3.97511572e-01 -2.92154521e-01 5.52158237e-01
-1.16244805e+00 5.27801454e-01 -8.77080619e-01 -7.25482464e-01
4.46409255e-01 3.88829768e-01 -3.41524810e-01 -5.95091164e-01
-6.94635510e-01 9.10786390e-01 -1.20373787e-02 -4.94002372e-01
-5.11031032e-01 -7.56708980e-01 -4.61309016e-01 7.84150288e-02
2.78256703e-02 -1.57337928e+00 1.47675371e+00 -4.35275704e-01
-4.33810413e-01 9.23439622e-01 -2.82554150e-01 -1.95162043e-01
2.54398823e-01 -1.95248321e-01 -3.54318142e-01 5.76846719e-01
3.82954329e-01 3.15000594e-01 2.59950250e-01 -1.20086968e+00
-9.95859861e-01 -6.25741541e-01 -3.81763786e-01 4.18816686e-01
-1.30356789e-01 1.73992574e-01 -1.12071231e-01 -4.91204739e-01
8.98614451e-02 -1.18657899e+00 -3.84373873e-01 -5.84538244e-02
2.32527629e-02 -1.74633384e-01 6.70307696e-01 -1.02573943e+00
1.48952639e+00 -2.01359177e+00 -6.75418019e-01 1.07539676e-01
8.54598463e-01 5.19896567e-01 5.88934302e-01 3.89514983e-01
-1.03423215e-01 5.29041350e-01 -3.41173083e-01 1.45009279e-01
-6.31528795e-01 1.14379115e-01 4.74274218e-01 2.84523219e-01
2.14143470e-01 8.70023310e-01 -7.58076131e-01 -5.97918451e-01
6.28330946e-01 5.59826434e-01 -2.91157037e-01 5.50969660e-01
4.27989781e-01 4.83770072e-01 -2.77020097e-01 5.74427485e-01
9.50366631e-02 -1.33623254e+00 3.90269309e-01 2.48651415e-01
6.73080087e-02 1.76247418e-01 -7.82977104e-01 4.94055986e-01
-2.77679563e-01 3.45931888e-01 8.34881142e-02 -5.15326262e-01
8.83111060e-01 9.50424612e-01 9.21820819e-01 -1.67656705e-01
-5.90907931e-02 3.26266408e-01 5.30959487e-01 -8.51523399e-01
-6.27115011e-01 -6.37079954e-01 5.43502152e-01 7.99711525e-01
-6.59088790e-01 -1.86474398e-01 -2.49234423e-01 2.08429292e-01
1.56666529e+00 -7.04046428e-01 5.16336441e-01 1.37880221e-01
3.04251611e-01 2.07909733e-01 1.94699764e-01 9.73297179e-01
-2.73975611e-01 1.04990304e+00 2.02856869e-01 -3.26235503e-01
-8.22820067e-01 -1.29332054e+00 -5.99075198e-01 2.11089805e-01
-3.92174065e-01 -4.23810601e-01 -3.56202543e-01 -6.55882180e-01
1.49235263e-01 6.04286432e-01 -1.81681708e-01 -1.44932508e-01
-3.13083410e-01 -8.02196443e-01 4.46974158e-01 9.04693127e-01
1.82971075e-01 -7.95048535e-01 -7.88045526e-01 4.84717548e-01
-4.33214545e-01 -7.48812199e-01 -3.56984377e-01 -1.87787283e-02
-1.25242221e+00 -1.31784892e+00 -1.29700470e+00 -6.40403032e-01
3.61165553e-01 3.72085094e-01 1.24869502e+00 7.33748317e-01
-5.21412909e-01 1.09839058e+00 -1.79121837e-01 -3.65902722e-01
-5.32219589e-01 -3.23205173e-01 -5.63452989e-02 -4.65502948e-01
2.47906551e-01 -1.31414965e-01 -1.02146745e+00 5.18496275e-01
-6.92846477e-01 -2.03214437e-01 8.92272890e-01 7.84653783e-01
6.53508902e-01 -2.70592660e-01 5.55627108e-01 -6.68833911e-01
4.16245401e-01 -8.24289858e-01 7.05781430e-02 1.83816090e-01
-1.39584744e+00 -5.22532880e-01 5.28985739e-01 -1.98559657e-01
-5.48519135e-01 -7.23035038e-02 2.05002546e-01 -2.75035799e-01
-4.03949291e-01 5.03830314e-01 8.11541736e-01 5.12684643e-01
6.26563609e-01 -1.24283336e-01 2.62413234e-01 -4.41152364e-01
-5.72229028e-01 1.05456614e+00 2.59706676e-01 -6.22447021e-02
6.42635465e-01 1.26590386e-01 1.03856280e-01 -6.02927804e-01
-7.20824301e-01 -1.07265770e+00 -2.23927677e-01 -2.50319135e-03
1.26102626e+00 -1.20370746e+00 -3.56538236e-01 2.03780849e-02
-5.26967168e-01 -3.40938836e-01 -4.20030028e-01 1.31091857e+00
-1.85888454e-01 3.07510257e-01 -6.93663955e-01 -6.83609605e-01
-6.74783051e-01 -1.05084491e+00 6.64495349e-01 2.17961997e-01
-1.05586112e+00 -1.07382631e+00 4.09554780e-01 6.53286815e-01
4.85109329e-01 3.79594028e-01 1.16601562e+00 -8.45896602e-01
-2.76474178e-01 -5.47455251e-01 -6.41231477e-01 2.72812039e-01
8.27868283e-01 2.39747211e-01 -4.33795542e-01 -7.52477348e-02
5.12623973e-02 -6.73113689e-02 5.21731257e-01 8.60158086e-01
9.37816739e-01 -2.34011680e-01 -4.78480250e-01 1.91558331e-01
1.29940534e+00 8.17296684e-01 4.06265497e-01 1.83864459e-01
6.23064637e-01 5.59301734e-01 2.82622844e-01 3.58380407e-01
6.18709564e-01 3.26588184e-01 2.33808756e-02 -2.70210534e-01
-9.70145389e-02 1.19735502e-01 9.52423178e-03 2.78506339e-01
-1.39681250e-01 -4.20533232e-02 -1.63898206e+00 6.37465596e-01
-1.48432970e+00 -7.00956464e-01 -8.08675945e-01 2.17497492e+00
4.14808691e-01 -4.97416919e-03 4.42148268e-01 1.53248049e-02
9.49251771e-01 -4.55996692e-01 -4.96277213e-01 -4.83628333e-01
1.82627827e-01 3.10691595e-01 2.56982863e-01 1.08405411e-01
-5.88670313e-01 -1.87682271e-01 5.87450266e+00 -2.90250152e-01
-7.33999133e-01 -1.04986273e-01 7.72573054e-01 3.52318399e-04
1.08874559e-01 -1.08114831e-01 -4.38999921e-01 4.82892394e-01
1.29587770e+00 -1.39480166e-03 1.76367927e-02 5.79411924e-01
4.07123178e-01 -3.06122899e-01 -9.59771931e-01 8.88488650e-01
-1.24385916e-01 -1.16063511e+00 -3.40572774e-01 2.33307138e-01
7.22654939e-01 3.22068959e-01 -1.40072480e-01 1.05667621e-01
4.03599031e-02 -1.26012456e+00 -9.33257490e-02 4.96841222e-01
1.29306328e+00 -8.76133367e-02 1.05796218e+00 7.93802589e-02
-1.05641711e+00 -1.51650533e-01 1.51375473e-01 -1.13500915e-01
7.56860673e-02 6.92116320e-01 -1.58174860e+00 9.95966569e-02
9.09716725e-01 4.92864817e-01 -3.22568774e-01 1.15023625e+00
-2.37029772e-02 1.04433799e+00 -4.10439074e-01 2.48079479e-01
1.39936447e-01 1.82349026e-01 2.46693745e-01 1.12982333e+00
3.35049808e-01 6.81103408e-01 -1.88459251e-02 3.95872295e-01
-9.97165591e-02 3.80098283e-01 -7.97043443e-01 -9.70312059e-02
4.07159895e-01 1.04124427e+00 -6.77186906e-01 -7.77891695e-01
-7.85578728e-01 2.80521661e-01 -2.17108473e-01 3.42037618e-01
-5.14743090e-01 -1.45474583e-01 2.76321601e-02 8.26592386e-01
-2.17698544e-01 5.32226562e-01 -7.57097125e-01 -8.38253438e-01
-1.90383628e-01 -8.19977820e-01 1.01548767e+00 -8.65384340e-01
-1.06592000e+00 1.89146861e-01 -3.38218272e-01 -1.19714093e+00
-3.76256645e-01 -4.00061995e-01 -4.07833099e-01 1.14414573e+00
-1.02992928e+00 -3.27288479e-01 -5.49031794e-01 1.52152240e-01
1.89390868e-01 -5.38653843e-02 1.10882199e+00 1.87345315e-02
-2.98457503e-01 1.66532516e-01 2.70556957e-01 2.41788283e-01
1.09702122e+00 -1.25244951e+00 -1.70394838e-01 -1.55950338e-01
-6.85759723e-01 7.48613715e-01 -1.90118194e-01 -8.47238839e-01
-8.02478790e-01 -1.11822236e+00 1.14359033e+00 -7.89356411e-01
1.05555132e-01 1.06749856e+00 -1.24431491e+00 5.95897436e-01
-2.03987181e-01 -3.08250189e-01 1.46678543e+00 -1.22300282e-01
-1.21720746e-01 1.40429869e-01 -1.45567119e+00 2.84063399e-01
4.51772302e-01 -4.30568725e-01 -6.30009949e-01 3.65453154e-01
2.03703150e-01 -9.50938165e-02 -1.53801811e+00 8.26757789e-01
1.10345960e+00 -7.49201536e-01 1.08644199e+00 -6.91727579e-01
5.97483099e-01 -2.33725831e-01 -9.29922387e-02 -7.09104717e-01
-6.77896261e-01 -4.79662567e-02 -6.83082044e-02 5.15889943e-01
5.87747931e-01 -9.06262398e-01 5.28728902e-01 1.00270522e+00
-1.08743817e-01 -1.02891123e+00 -3.26732099e-01 -2.93488950e-01
-5.34749143e-02 4.51352373e-02 1.46254018e-01 1.01814485e+00
-1.60224348e-01 3.93742591e-01 2.16330811e-01 2.25887507e-01
1.85185432e-01 -1.24383695e-01 4.18223530e-01 -1.44224823e+00
-3.93657535e-01 -3.18483472e-01 3.66069637e-02 -3.38055670e-01
-7.07144380e-01 -7.98211157e-01 -6.01672709e-01 -2.59785104e+00
8.29969883e-01 -6.38828099e-01 -6.79155052e-01 4.56675857e-01
-6.31754994e-01 7.36581311e-02 5.02537340e-02 9.93618965e-01
2.91301729e-03 -2.73409754e-01 8.91583025e-01 2.05154598e-01
-4.12339956e-01 3.22089881e-01 -8.57854247e-01 5.98265529e-01
1.61512470e+00 -7.70251274e-01 -2.54332989e-01 -2.13327065e-01
-7.48786852e-02 6.86060965e-01 6.05961382e-01 -1.02121401e+00
-1.05190337e-01 -2.07586974e-01 8.46798003e-01 -5.79010665e-01
5.75599745e-02 -7.27752626e-01 5.33650696e-01 1.45301795e+00
-1.11751728e-01 5.28997540e-01 6.45461082e-02 3.72374415e-01
2.48681055e-03 -2.01272354e-01 8.18206489e-01 -4.41618651e-01
2.02731296e-01 -4.87181433e-02 -9.19805706e-01 3.22087854e-01
8.15634429e-01 -2.09181264e-01 -3.25668365e-01 -6.91535652e-01
-9.41707551e-01 2.32935160e-01 6.62334621e-01 -5.23473509e-02
5.57212532e-01 -9.49241042e-01 -7.88641036e-01 -2.38069251e-01
1.27656817e-01 -1.25978189e-02 1.79302782e-01 1.45913255e+00
-8.15364838e-01 5.51823616e-01 3.77533972e-01 -8.66378665e-01
-1.47885323e+00 2.90910363e-01 4.40473974e-01 -4.11814451e-01
-5.94121218e-01 3.57356280e-01 3.37979883e-01 -1.49571016e-01
1.99465364e-01 -2.36139178e-01 2.54193544e-01 -1.63641602e-01
5.85385799e-01 7.78326809e-01 -9.74785611e-02 -3.50845456e-01
-5.38817823e-01 9.04838562e-01 -1.03515379e-01 -3.99616687e-03
1.01255858e+00 -3.04905057e-01 -8.81112665e-02 4.76961493e-01
1.03242934e+00 -2.44124100e-01 -3.21140647e-01 2.27461178e-02
-2.34186232e-01 -1.15422040e-01 -2.62591213e-01 -1.13122034e+00
-5.03386736e-01 8.10946584e-01 1.05283678e+00 4.64269012e-01
1.07214653e+00 1.59380093e-01 5.90590537e-01 2.83046812e-01
-2.11802706e-01 -6.74951732e-01 7.26708770e-02 6.97307363e-02
3.34648341e-01 -1.33017981e+00 -9.55353230e-02 1.17786983e-02
-8.32773745e-01 1.17426205e+00 2.37406448e-01 1.76915959e-01
7.92220652e-01 -1.64383560e-01 3.73569757e-01 -3.19144964e-01
-8.55879247e-01 1.85558483e-01 1.05319627e-01 4.77823049e-01
7.03310966e-01 5.00374317e-01 -4.81804341e-01 5.28063118e-01
3.18755805e-01 2.86116719e-01 4.77811903e-01 1.07711422e+00
-5.90990782e-01 -9.19393182e-01 -8.44093859e-01 1.38084269e+00
-8.24751675e-01 -1.38345407e-02 -6.71619892e-01 8.24013233e-01
5.35844779e-03 1.25571537e+00 3.59813645e-02 -1.02403142e-01
4.56960648e-01 5.81985295e-01 7.61969835e-02 -9.08675075e-01
-6.95629239e-01 4.83135104e-01 2.42735341e-01 -1.65198967e-02
-5.63009083e-01 -7.56323338e-01 -1.49661410e+00 1.23235881e-01
-3.12043607e-01 2.54623085e-01 2.29012206e-01 5.29684424e-01
4.25686955e-01 7.35202372e-01 5.43660998e-01 3.03483069e-01
-5.08076131e-01 -6.45588875e-01 -4.91466194e-01 1.08703654e-02
6.29113138e-01 -4.29131389e-01 -6.00810647e-01 1.10923171e-01] | [15.454805374145508, -1.9620283842086792] |
d5f84da8-1505-4aab-a9d3-92aeed3a6d69 | nocola-the-norwegian-corpus-of-linguistic | 2306.07790 | null | https://arxiv.org/abs/2306.07790v1 | https://arxiv.org/pdf/2306.07790v1.pdf | NoCoLA: The Norwegian Corpus of Linguistic Acceptability | While there has been a surge of large language models for Norwegian in recent years, we lack any tool to evaluate their understanding of grammaticality. We present two new Norwegian datasets for this task. NoCoLA_class is a supervised binary classification task where the goal is to discriminate between acceptable and non-acceptable sentences. On the other hand, NoCoLA_zero is a purely diagnostic task for evaluating the grammatical judgement of a language model in a completely zero-shot manner, i.e. without any further training. In this paper, we describe both datasets in detail, show how to use them for different flavors of language models, and conduct a comparative study of the existing Norwegian language models. | ['David Samuel', 'Matias Jentoft'] | 2023-06-13 | null | null | null | null | ['linguistic-acceptability'] | ['natural-language-processing'] | [-2.89826870e-01 3.81193846e-01 7.81885535e-02 -8.16857457e-01
-7.20463037e-01 -6.42751753e-01 4.39001560e-01 4.05410111e-01
-5.06405652e-01 7.08765507e-01 6.29908890e-02 -7.87478387e-01
8.40309262e-02 -6.30665541e-01 -1.01735435e-01 -4.25876111e-01
2.02416897e-01 6.16973579e-01 -8.96210521e-02 -5.34771621e-01
-2.13036984e-02 3.16952467e-02 -1.59769988e+00 3.87398422e-01
9.82236147e-01 4.95591402e-01 -9.54098031e-02 7.98218429e-01
-1.36702344e-01 8.02988887e-01 -6.74533784e-01 -7.16620445e-01
-5.28914258e-02 -3.62459511e-01 -9.85027850e-01 -1.10702991e-01
2.17304468e-01 8.75398740e-02 1.29345670e-01 1.01911604e+00
5.00868201e-01 6.61634430e-02 7.90351212e-01 -1.19171500e+00
-6.14723206e-01 9.72097456e-01 2.37181276e-01 1.54061869e-01
6.87027872e-01 1.68681502e-01 1.26488924e+00 -6.29437149e-01
5.62537491e-01 1.80530071e+00 6.13372147e-01 7.37821102e-01
-1.29321444e+00 -5.61712027e-01 2.47509167e-01 -9.48626101e-02
-1.09363294e+00 -5.25677502e-01 6.48544729e-01 -6.40633523e-01
1.32206774e+00 2.42737576e-01 4.44776207e-01 1.15584397e+00
5.43393075e-01 6.61159635e-01 1.37935770e+00 -7.32404530e-01
1.65845603e-01 2.60448426e-01 6.23330057e-01 4.56865489e-01
8.51115957e-02 3.09196681e-01 -2.10150599e-01 8.21180455e-03
-8.58231038e-02 -7.47663319e-01 1.19079985e-01 1.70997113e-01
-7.88279474e-01 9.06423211e-01 -1.29617646e-01 5.09208858e-01
1.38440102e-01 -2.78590292e-01 5.77016652e-01 7.14895546e-01
7.87047982e-01 4.25079823e-01 -7.18934953e-01 -2.04641417e-01
-8.85024607e-01 4.08360451e-01 1.07709169e+00 5.51188290e-01
5.22978127e-01 1.63705856e-01 -1.98167972e-02 8.81286919e-01
6.55160546e-01 2.48826221e-01 4.98923153e-01 -7.48439193e-01
5.08119643e-01 6.00398302e-01 -3.80106181e-01 -6.82214499e-01
-6.70704424e-01 -2.15737447e-01 -4.76737261e-01 2.67909855e-01
4.60593522e-01 8.09953511e-02 -8.00962090e-01 1.86798358e+00
-2.53264941e-02 -3.40540946e-01 4.88773555e-01 4.38251793e-01
1.34418511e+00 6.68831944e-01 4.77519989e-01 -8.39558020e-02
1.06669939e+00 -6.66305840e-01 -6.59922779e-01 -4.52707291e-01
1.30309522e+00 -8.03206325e-01 1.39208698e+00 5.47004163e-01
-1.10615277e+00 -4.86086667e-01 -9.37813222e-01 -3.76773864e-01
-5.87043464e-01 1.47170443e-02 7.81447411e-01 1.21245944e+00
-1.18224871e+00 3.74502659e-01 -5.88239849e-01 -2.37882391e-01
-2.56401032e-01 2.64628470e-01 -5.14953613e-01 -2.85737813e-02
-1.58252847e+00 1.11176455e+00 5.62136412e-01 9.79931727e-02
-4.04342413e-01 -1.54419139e-01 -1.57480729e+00 -1.31374940e-01
-1.01899937e-01 -2.54619151e-01 1.36732888e+00 -9.26599801e-01
-1.16975343e+00 1.55439055e+00 -3.15552145e-01 -2.76443601e-01
5.05869210e-01 1.75951019e-01 -6.53669655e-01 -6.81218505e-01
3.45975757e-01 7.23445237e-01 1.01334192e-01 -1.07340944e+00
-4.30007100e-01 -3.59035969e-01 3.06818277e-01 1.71698257e-02
-8.49370658e-02 5.87751985e-01 -4.41697240e-02 -7.27847099e-01
2.61746168e-01 -7.86541283e-01 -1.45667642e-01 -5.86490035e-01
-4.51127350e-01 -7.38073051e-01 2.66176760e-01 -6.16948783e-01
1.69527912e+00 -2.14603257e+00 -1.38652116e-01 1.63693503e-02
5.73207438e-02 2.97137499e-01 -2.31228933e-01 3.43020678e-01
-2.72588998e-01 7.46024549e-01 -6.25330150e-01 -6.05598748e-01
3.17502260e-01 5.44121563e-01 -3.74298617e-02 1.94458738e-01
3.98666173e-01 8.81152391e-01 -1.00750780e+00 -6.38920009e-01
7.28561059e-02 8.17277953e-02 -6.09801471e-01 5.43048754e-02
-1.65247023e-01 2.40589306e-01 6.09014928e-02 7.60984123e-01
6.97519422e-01 4.78351146e-01 3.17814469e-01 4.86633241e-01
-1.28418848e-01 8.56637657e-01 -1.03862178e+00 1.25283825e+00
-5.41672826e-01 6.05924785e-01 4.88101924e-03 -8.89266849e-01
9.34067011e-01 3.72894734e-01 -2.49042049e-01 -8.53886247e-01
2.25572437e-01 6.28238142e-01 6.34320676e-01 -4.84992266e-01
6.33246601e-01 -5.08394182e-01 -7.58356512e-01 3.33657086e-01
5.16944826e-02 -2.96701282e-01 7.51398861e-01 1.80352423e-02
7.85918891e-01 1.53873309e-01 4.80688244e-01 -5.19761741e-01
7.65278161e-01 -1.65205643e-01 7.62520075e-01 6.14454627e-01
-3.24193686e-01 5.06770670e-01 8.34383667e-01 -3.23236525e-01
-6.87856913e-01 -1.06553149e+00 -3.76984894e-01 1.18638527e+00
-3.12635541e-01 -6.55891180e-01 -8.00803721e-01 -7.10483849e-01
-1.61162525e-01 1.31194663e+00 -4.16006565e-01 -8.29091370e-02
-6.09701514e-01 -8.14935088e-01 8.79464805e-01 4.11794156e-01
8.38923454e-02 -1.42288685e+00 -2.64117211e-01 2.86230594e-01
-1.46825507e-01 -1.04189694e+00 -1.64072216e-02 4.47766453e-01
-6.15129948e-01 -9.25187469e-01 7.65126944e-02 -1.23983574e+00
4.55511898e-01 -4.91178513e-01 1.83129728e+00 4.44582850e-01
5.93118854e-02 4.18946482e-02 -4.77572143e-01 -5.97519219e-01
-1.13259149e+00 5.66487294e-03 1.32505178e-01 -5.95539451e-01
7.27060258e-01 -1.18283056e-01 2.22194403e-01 1.41133159e-01
-7.83064723e-01 -1.48187965e-01 2.04428628e-01 7.53773570e-01
1.12383060e-01 -8.36334825e-02 5.52690327e-01 -1.12800837e+00
1.07668877e+00 -4.18298811e-01 -2.79538721e-01 2.51066506e-01
-4.04760480e-01 3.97717692e-02 6.13541067e-01 6.24104924e-02
-6.47222996e-01 -8.05469453e-02 -9.01382685e-01 4.31235164e-01
-3.86399955e-01 8.89685214e-01 -4.06511992e-01 3.78975868e-01
4.66493756e-01 -9.68704075e-02 -1.78815767e-01 -5.84629059e-01
1.04132473e-01 9.52474892e-01 3.02174807e-01 -7.08584785e-01
4.41297323e-01 -1.86422840e-01 -3.70053858e-01 -9.53855217e-01
-8.59908760e-01 -3.62576872e-01 -9.74550843e-01 -1.23744637e-01
7.22960949e-01 -7.49197721e-01 -5.14391661e-01 5.19846559e-01
-1.17827058e+00 -4.95384932e-01 -6.63107112e-02 3.13332886e-01
-4.18144524e-01 4.48216915e-01 -6.11854136e-01 -1.09664261e+00
-9.63567495e-02 -1.15775800e+00 7.86167026e-01 -1.95152149e-01
-5.74394524e-01 -1.36767054e+00 4.64034751e-02 2.18979374e-01
5.71795888e-02 -2.70147622e-02 1.29466116e+00 -8.51781011e-01
1.76861867e-01 -1.90271124e-01 3.74060348e-02 5.68936467e-01
3.16365920e-02 1.66600749e-01 -1.08585167e+00 -1.96405962e-01
1.28603756e-01 -6.35872126e-01 8.92820060e-01 2.14259550e-01
8.37643921e-01 -1.61369815e-01 1.32252127e-01 2.76048541e-01
9.01154101e-01 2.10418388e-01 5.59202611e-01 3.77092242e-01
3.27275604e-01 8.92422259e-01 7.50582874e-01 -7.29252920e-02
9.27348137e-01 6.58390820e-01 2.42932081e-01 -3.71544547e-02
-9.40899476e-02 -1.17398903e-01 6.83079302e-01 1.11822748e+00
3.16805214e-01 -6.75392091e-01 -1.33351326e+00 5.88744223e-01
-1.60266805e+00 -5.91682494e-01 -3.29655886e-01 2.08759785e+00
9.10867631e-01 4.81460690e-01 -2.10681692e-01 5.90697229e-01
5.75963438e-01 1.53374374e-01 1.19465172e-01 -1.46370339e+00
-5.07661343e-01 2.32310593e-01 -2.93416798e-01 9.82657313e-01
-1.31927931e+00 1.31331086e+00 7.35263157e+00 7.36646950e-01
-1.01338542e+00 -1.02726363e-01 9.72973406e-01 3.95613432e-01
-5.77663839e-01 1.53044134e-01 -9.19643521e-01 3.71273726e-01
1.33320725e+00 -5.79731958e-03 -6.96631968e-02 5.91582775e-01
4.99424189e-01 -3.56145650e-01 -1.32218611e+00 5.89688480e-01
1.57685041e-01 -6.91182613e-01 1.06468461e-01 -2.26115450e-01
2.76171178e-01 -1.45219505e-01 -1.56147048e-01 9.29530680e-01
3.78055692e-01 -1.43822062e+00 1.05565667e+00 2.02080578e-01
7.05680370e-01 -7.33746886e-01 1.06707954e+00 7.28827298e-01
-9.80049908e-01 -2.94785779e-02 -3.51540208e-01 -7.78424859e-01
2.39850447e-01 4.41654742e-01 -5.62308669e-01 5.25902212e-01
5.88818371e-01 5.33598423e-01 -8.58135879e-01 6.98138237e-01
-3.91009867e-01 9.41926360e-01 -1.98967829e-01 -1.89810857e-01
4.73304749e-01 -1.69631243e-01 4.17017788e-01 1.40819812e+00
1.89100072e-01 -9.22340378e-02 4.91173446e-01 7.45540679e-01
7.26779774e-02 4.15307730e-01 -5.40035129e-01 4.11557332e-02
2.21635431e-01 9.41647768e-01 -6.54055297e-01 -2.22951636e-01
-5.81571698e-01 5.41034281e-01 3.78734708e-01 -1.48958683e-01
-5.16550779e-01 -3.69394958e-01 4.89232421e-01 1.49199247e-01
-3.75069976e-01 -3.47460896e-01 -5.42716920e-01 -1.07887685e+00
1.21211104e-01 -8.58891964e-01 4.13027257e-01 -7.05693603e-01
-1.42164373e+00 7.12963998e-01 -5.40291406e-02 -1.03508842e+00
-6.73026443e-01 -1.00060594e+00 -5.99758565e-01 1.07943118e+00
-1.67273140e+00 -1.20381653e+00 1.61861002e-01 1.04145877e-01
6.37047946e-01 -1.16336644e-01 1.17821741e+00 2.14394256e-01
-5.77496111e-01 7.41317570e-01 -4.78493124e-02 5.13070047e-01
6.80725574e-01 -1.43668449e+00 7.08326578e-01 7.67275035e-01
1.74793184e-01 6.89384758e-01 7.44943976e-01 -6.19691014e-01
-6.82145119e-01 -8.35989237e-01 1.95153368e+00 -5.26970327e-01
6.55834317e-01 -6.63714230e-01 -8.93389881e-01 5.46211421e-01
1.33618116e-01 -2.51235098e-01 8.13961864e-01 3.54247719e-01
-2.01858342e-01 2.42669791e-01 -1.14277124e+00 4.17321235e-01
8.47123325e-01 -4.07673955e-01 -9.91625965e-01 1.85121223e-01
5.04483044e-01 -3.37213576e-01 -5.54036736e-01 4.92624551e-01
3.39706808e-01 -1.26371145e+00 5.76302230e-01 -9.23759341e-01
4.47532326e-01 -6.18574806e-02 -2.30921179e-01 -1.35556233e+00
-1.34366840e-01 -3.12457979e-01 5.94259560e-01 1.40470183e+00
9.06040728e-01 -8.21549416e-01 5.37432671e-01 6.94958925e-01
-5.21188557e-01 -7.81276762e-01 -1.18293738e+00 -9.65829551e-01
7.65323102e-01 -1.04386818e+00 3.73346388e-01 8.61645699e-01
8.35283250e-02 2.94202536e-01 1.11749686e-01 -1.72339126e-01
1.58107355e-01 -3.34379077e-03 3.62696677e-01 -1.55175209e+00
-6.02072030e-02 -5.93851268e-01 -6.36670291e-01 -7.32326627e-01
8.06032300e-01 -1.36239183e+00 2.60239840e-01 -1.53302920e+00
1.34427667e-01 -4.75581676e-01 -8.61403495e-02 6.37343407e-01
-1.70870677e-01 3.31312597e-01 1.47546142e-01 -1.75782308e-01
-4.22314167e-01 2.79995233e-01 7.06126988e-01 -1.32050484e-01
-5.93960471e-02 1.53359741e-01 -8.24652195e-01 1.01303780e+00
7.73696601e-01 -4.97925401e-01 -1.75282538e-01 -5.67926049e-01
5.03076613e-01 -2.24429592e-01 2.45460570e-01 -8.08200300e-01
-2.23751307e-01 1.32490909e-02 4.89688404e-02 -4.23443317e-01
1.12437740e-01 -1.67567134e-01 -3.69418263e-01 3.32716703e-01
-3.27228636e-01 4.07424182e-01 3.94149542e-01 9.94913355e-02
-6.15103066e-01 -6.24004066e-01 7.00045824e-01 -3.19182366e-01
-7.97599554e-01 -2.04234973e-01 -8.35107327e-01 3.37549359e-01
7.37835586e-01 -2.94615388e-01 -2.21767444e-02 -2.32178301e-01
-1.18405712e+00 3.62421691e-01 4.16549146e-01 6.60927474e-01
4.00449276e-01 -1.03807509e+00 -1.01473522e+00 3.56313825e-01
2.63004959e-01 -1.75372705e-01 4.25133668e-02 5.07506788e-01
-5.62813878e-01 5.59344471e-01 1.42879277e-01 -5.01449049e-01
-1.42192018e+00 2.05525830e-01 3.54466349e-01 -5.30960381e-01
-1.46574542e-01 8.52729380e-01 9.37380269e-02 -1.19171667e+00
-7.74767026e-02 -4.33401227e-01 -6.22762382e-01 2.09461555e-01
1.37042552e-01 -6.53077960e-02 5.24257362e-01 -1.06385136e+00
-5.14844120e-01 2.86080688e-01 1.08680457e-01 -1.61949903e-01
1.09063005e+00 -1.35570124e-01 -5.60470819e-01 9.57004130e-01
9.55628753e-01 -1.26550794e-02 -4.64279503e-01 1.24709047e-01
6.25696898e-01 -2.19103768e-01 -1.06952466e-01 -1.02332544e+00
-4.89823431e-01 9.97539818e-01 2.43051469e-01 4.82066125e-01
8.15218329e-01 1.33585408e-01 4.57176745e-01 4.86663938e-01
2.77364880e-01 -1.14597774e+00 -5.34545004e-01 1.36033833e+00
8.92675221e-01 -1.36339366e+00 -3.21482480e-01 -6.45239055e-01
-6.80970371e-01 1.14687538e+00 6.07618332e-01 2.06533298e-01
7.88370192e-01 3.01480114e-01 6.07356250e-01 2.38421671e-02
-9.43916738e-01 -3.08681667e-01 1.37478754e-01 5.03730655e-01
1.26238799e+00 2.45936632e-01 -7.77071476e-01 1.04275572e+00
-9.38226640e-01 -4.20987755e-01 7.14491844e-01 8.83048058e-01
-4.13569033e-01 -1.51146996e+00 -2.67203987e-01 4.75866050e-01
-4.68101710e-01 -4.26942021e-01 -5.48666120e-01 7.44588733e-01
1.31400570e-01 1.35697699e+00 8.61734822e-02 -2.85837591e-01
3.91605169e-01 4.40669209e-01 2.12871090e-01 -1.18882549e+00
-8.26146305e-01 -1.11745141e-01 6.57801747e-01 -1.69261634e-01
-2.42356658e-01 -7.97717392e-01 -1.24837041e+00 -3.40290636e-01
-2.60975003e-01 3.21620971e-01 5.12473106e-01 1.09099185e+00
-1.85400546e-01 2.84049988e-01 4.17663038e-01 -7.08725870e-01
-5.24964213e-01 -1.04968250e+00 -6.27161503e-01 3.04883540e-01
8.95078387e-03 -5.44345558e-01 -6.61434829e-01 -2.66918033e-01] | [10.726436614990234, 9.726648330688477] |
1afd1347-7826-4193-9437-b508f47b006e | selftalk-a-self-supervised-commutative | 2306.10799 | null | https://arxiv.org/abs/2306.10799v1 | https://arxiv.org/pdf/2306.10799v1.pdf | SelfTalk: A Self-Supervised Commutative Training Diagram to Comprehend 3D Talking Faces | Speech-driven 3D face animation technique, extending its applications to various multimedia fields. Previous research has generated promising realistic lip movements and facial expressions from audio signals. However, traditional regression models solely driven by data face several essential problems, such as difficulties in accessing precise labels and domain gaps between different modalities, leading to unsatisfactory results lacking precision and coherence. To enhance the visual accuracy of generated lip movement while reducing the dependence on labeled data, we propose a novel framework SelfTalk, by involving self-supervision in a cross-modals network system to learn 3D talking faces. The framework constructs a network system consisting of three modules: facial animator, speech recognizer, and lip-reading interpreter. The core of SelfTalk is a commutative training diagram that facilitates compatible features exchange among audio, text, and lip shape, enabling our models to learn the intricate connection between these factors. The proposed framework leverages the knowledge learned from the lip-reading interpreter to generate more plausible lip shapes. Extensive experiments and user studies demonstrate that our proposed approach achieves state-of-the-art performance both qualitatively and quantitatively. We recommend watching the supplementary video. | ['Zhaoxin Fan', 'Jun He', 'Hongyan Liu', 'Xiangyu Zhu', 'Hao Xu', 'Yue Shi', 'Yihao Luo', 'Ziqiao Peng'] | 2023-06-19 | null | null | null | null | ['3d-face-animation'] | ['computer-vision'] | [ 1.55787870e-01 3.11719060e-01 -4.33848262e-01 -5.50456464e-01
-8.68623257e-01 -2.23414466e-01 5.13801634e-01 -6.46942973e-01
1.30188406e-01 4.54806268e-01 4.18141097e-01 -2.79661100e-02
3.28987509e-01 -3.29839855e-01 -6.53238356e-01 -4.35460746e-01
1.19430460e-01 1.90079868e-01 -5.89005686e-02 -1.15270838e-01
-2.93100806e-04 3.41747761e-01 -1.85544765e+00 4.09637690e-01
7.43786454e-01 1.19224596e+00 -1.06256463e-01 5.24848163e-01
-4.02642697e-01 7.54186988e-01 -2.23347604e-01 -3.75368178e-01
1.72262639e-01 -4.25674647e-01 -6.34241760e-01 3.61698329e-01
5.20133138e-01 -5.30823171e-01 -1.15287304e-01 9.41819608e-01
7.53144920e-01 1.78716090e-02 7.24328101e-01 -1.83614039e+00
-6.67141199e-01 5.10103226e-01 -7.53280818e-01 -4.97906178e-01
7.35431015e-01 4.08211291e-01 7.87285328e-01 -1.02148032e+00
5.80509424e-01 1.66923094e+00 6.33274913e-01 1.08044529e+00
-1.08203089e+00 -1.05171752e+00 1.72395125e-01 9.37346891e-02
-1.48472297e+00 -1.22337723e+00 1.22199500e+00 -4.09383208e-01
5.00550091e-01 1.14891611e-01 5.44407189e-01 1.45129204e+00
-5.28241038e-01 9.01016295e-01 1.07132471e+00 -5.03956258e-01
-5.29334173e-02 6.18200779e-01 -3.90556961e-01 7.24775016e-01
-5.52766502e-01 1.75651550e-01 -9.59264636e-01 2.16395557e-02
7.97432065e-01 -4.39508706e-01 -1.69200540e-01 -5.89406073e-01
-7.95964360e-01 5.36315978e-01 -1.87575724e-02 -1.44821659e-01
-6.49076477e-02 -6.06887043e-02 3.90434474e-01 1.17887281e-01
4.78671640e-01 -3.37314844e-01 -3.22494447e-01 -8.97516012e-02
-9.13111210e-01 1.00862347e-01 6.35960102e-01 1.14321887e+00
5.69855392e-01 2.47786298e-01 8.40084553e-02 1.05917239e+00
9.38281000e-01 6.35378182e-01 2.94465244e-01 -1.12198412e+00
4.19110447e-01 5.17418206e-01 -1.78443968e-01 -1.02027452e+00
-1.15403771e-01 1.95322901e-01 -7.17695355e-01 3.45742762e-01
2.53197104e-01 -2.80919373e-01 -6.76017106e-01 2.06620526e+00
6.83639288e-01 5.41410565e-01 7.53303096e-02 9.12098646e-01
1.26841867e+00 4.17628616e-01 4.33931857e-01 -2.08734378e-01
1.12160015e+00 -8.66228819e-01 -7.74916470e-01 -3.04161385e-02
2.18159273e-01 -8.60799193e-01 1.27944303e+00 1.49288297e-01
-1.39010084e+00 -7.98689306e-01 -6.83510840e-01 -1.65320590e-01
-1.13035575e-01 2.46392220e-01 5.10684788e-01 5.11737525e-01
-1.20139360e+00 2.25891054e-01 -4.13701445e-01 -4.25160468e-01
5.69358110e-01 5.02676666e-01 -4.74405259e-01 4.68424290e-01
-1.01682460e+00 5.35962880e-01 -3.16966549e-02 4.28138897e-02
-8.10932875e-01 -7.46203959e-01 -1.08405507e+00 -1.54849946e-01
3.80451143e-01 -6.55913889e-01 1.42781079e+00 -1.44730151e+00
-2.18534851e+00 1.12326050e+00 -2.46017963e-01 -1.07997656e-01
6.26758516e-01 -7.48956501e-02 -5.62806904e-01 1.13702931e-01
-1.85744882e-01 1.36270368e+00 1.24134147e+00 -1.55343783e+00
-5.16930759e-01 -3.64121616e-01 -1.52894542e-01 4.19669747e-01
-2.53859609e-01 5.01829088e-02 -6.27408385e-01 -4.71018910e-01
-1.24457829e-01 -7.80836642e-01 3.46694559e-01 5.91994822e-01
-3.93349916e-01 -3.51025015e-01 1.08943498e+00 -5.22597551e-01
8.31879735e-01 -2.33971238e+00 -6.73386678e-02 -1.33388024e-02
1.93254594e-02 1.97739601e-01 -3.89348835e-01 1.49365544e-01
-2.88260579e-01 -9.22721159e-03 1.63641319e-01 -7.24635780e-01
2.87063271e-02 -1.37968287e-01 -3.31365317e-01 3.38726878e-01
3.33743155e-01 7.04290152e-01 -5.86026013e-01 -9.53227401e-01
3.92303854e-01 8.97193491e-01 -6.36176646e-01 5.45220137e-01
-4.06094253e-01 6.52493834e-01 -2.28802308e-01 8.33638251e-01
8.31223786e-01 -8.28416422e-02 -5.93343861e-02 -4.85393912e-01
6.10792227e-02 1.96222097e-01 -1.32726216e+00 1.94267559e+00
-6.34862483e-01 6.10097528e-01 5.00152826e-01 -6.64251804e-01
1.04558420e+00 6.21884227e-01 4.39338267e-01 -5.08292079e-01
3.78979892e-01 -1.91707850e-01 -4.52227741e-01 -9.72713053e-01
1.74133435e-01 -1.34974346e-01 2.25350037e-01 3.72774124e-01
2.31503293e-01 -1.72218814e-01 -3.60039175e-01 3.69511060e-02
1.70060068e-01 7.02110469e-01 7.46366456e-02 1.30145416e-01
8.81531179e-01 -4.56679493e-01 2.93005645e-01 -9.15004686e-03
-3.61714154e-01 3.73461515e-01 3.41654748e-01 -1.02703320e-02
-7.20104754e-01 -8.88292074e-01 4.42578197e-02 1.42425692e+00
1.97669953e-01 -1.93569228e-01 -9.45327938e-01 -4.68855977e-01
-1.24125510e-01 3.95657659e-01 -5.79564333e-01 -2.47612037e-03
-2.47527421e-01 5.24891876e-02 7.52035677e-01 3.75814706e-01
5.32015324e-01 -1.22401035e+00 -1.86503336e-01 -2.84568042e-01
-2.94236749e-01 -1.27002931e+00 -6.41968429e-01 -4.57418889e-01
-6.19698882e-01 -8.84537339e-01 -5.09606540e-01 -1.03505635e+00
6.12032831e-01 1.98782787e-01 8.41877460e-01 -3.81483957e-02
-1.35326490e-01 5.62243998e-01 2.75489222e-02 -3.43635440e-01
-8.38180482e-01 -1.25227451e-01 2.41324067e-01 3.80870044e-01
2.27470770e-01 -7.56247759e-01 -4.62189287e-01 3.38197529e-01
-5.59556544e-01 4.64901865e-01 3.78610641e-01 5.20866036e-01
3.41904163e-01 -6.11373186e-01 7.15623200e-01 -6.23350024e-01
5.47284484e-01 -4.35202152e-01 -4.31377143e-01 8.61315355e-02
-3.97365749e-01 -1.03024259e-01 3.78332436e-01 -7.23137379e-01
-1.39561188e+00 2.59328157e-01 -3.31131905e-01 -7.86221743e-01
-5.84013164e-01 -1.45545438e-01 -5.77569187e-01 -4.08360362e-02
3.79534513e-01 7.14587569e-02 5.08143544e-01 -4.03008163e-01
6.63058817e-01 1.19720006e+00 7.92190194e-01 -5.61907768e-01
7.30932057e-01 4.27977592e-01 -1.38363957e-01 -9.95257497e-01
-5.43346643e-01 -2.26213977e-01 -5.60301363e-01 -6.20846808e-01
7.36779630e-01 -1.12057614e+00 -1.30519331e+00 5.45239627e-01
-1.19513512e+00 -2.53365934e-01 -1.49194628e-01 1.71020016e-01
-8.18674982e-01 2.66058892e-01 -4.15767431e-01 -9.96828973e-01
-2.37556756e-01 -1.05965483e+00 1.24539375e+00 4.48285639e-01
-3.26296061e-01 -9.42608297e-01 -5.77815101e-02 6.31847799e-01
3.10675651e-01 1.28153890e-01 7.55482197e-01 -4.20852184e-01
-4.47643995e-01 1.69453517e-01 -2.60402292e-01 2.82236636e-01
4.20481682e-01 3.09404641e-01 -1.47813284e+00 4.35324796e-02
-3.71992439e-01 -7.89337695e-01 2.14867488e-01 3.16318333e-01
1.15345943e+00 -4.06713277e-01 -2.69285709e-01 6.55568480e-01
8.89636278e-01 -5.46518853e-03 2.24982396e-01 -2.49087840e-01
6.24848068e-01 1.04760098e+00 4.00383621e-01 4.09680039e-01
6.00670993e-01 7.85273314e-01 3.98917764e-01 -3.83488894e-01
-4.52049971e-01 -7.64969587e-01 4.73563641e-01 8.31176043e-01
1.40908465e-01 8.01469907e-02 -5.92873037e-01 3.68400216e-01
-1.61544156e+00 -9.23995912e-01 3.84650022e-01 1.90329945e+00
1.03615665e+00 -1.33503959e-01 3.81332785e-01 4.74727675e-02
8.19043159e-01 1.11010596e-01 -6.71634734e-01 -2.59704798e-01
1.41146794e-01 -9.99074429e-02 -1.83578134e-01 7.47567058e-01
-9.69495058e-01 1.22331715e+00 6.07465410e+00 7.31579423e-01
-1.42370391e+00 -4.97273207e-02 6.12529993e-01 3.23614962e-02
-3.62490147e-01 -3.09992909e-01 -7.61584401e-01 2.61026412e-01
7.32015371e-01 -6.36413023e-02 3.87563109e-01 8.12053740e-01
6.69826269e-01 2.40303740e-01 -1.34168589e+00 1.25766921e+00
3.40987027e-01 -1.22047675e+00 2.74417579e-01 -2.47977421e-01
2.67592430e-01 -3.73927116e-01 3.39375556e-01 2.56778508e-01
1.29847527e-01 -1.11915731e+00 7.31706381e-01 5.19604266e-01
1.32675731e+00 -5.29711962e-01 9.04844552e-02 2.88126767e-01
-1.24228919e+00 1.78175449e-01 2.84260154e-01 2.20579535e-01
1.31797269e-01 -2.98936456e-01 -1.11447906e+00 1.28507867e-01
6.14542484e-01 7.06567526e-01 -8.51048529e-02 4.93360400e-01
-6.65769354e-02 3.54427308e-01 -1.88891292e-01 2.25658268e-01
-2.34646618e-01 -3.54881934e-03 5.08194268e-01 1.15878010e+00
3.77421454e-02 -5.45774847e-02 4.88395318e-02 9.31809843e-01
-2.79216200e-01 3.84662867e-01 -8.65764558e-01 1.34739354e-01
6.91677749e-01 1.23832858e+00 -2.19817922e-01 -1.19501501e-01
-5.77648461e-01 6.80040181e-01 1.66963547e-01 3.90562743e-01
-7.99857914e-01 8.91283751e-02 7.83485115e-01 2.60664999e-01
-3.94723080e-02 1.18866108e-01 -1.34111211e-01 -8.40304315e-01
-1.34380504e-01 -1.22285020e+00 1.41287491e-01 -1.00140166e+00
-1.21844089e+00 6.18116915e-01 -3.39108966e-02 -1.31922638e+00
-5.05154073e-01 -4.29583669e-01 -4.24019724e-01 7.83360839e-01
-1.55828071e+00 -1.54319441e+00 -5.28729916e-01 1.05808914e+00
9.47347403e-01 -3.81922603e-01 8.34083736e-01 2.48327732e-01
-5.43490767e-01 9.39007461e-01 -6.46039009e-01 2.17374265e-01
9.73330736e-01 -5.79631209e-01 5.81304953e-02 2.83819586e-01
1.31533325e-01 3.41131091e-01 5.25029838e-01 -3.99819136e-01
-1.37466586e+00 -8.08008492e-01 6.29077494e-01 -1.01921849e-01
4.56613243e-01 -4.38706845e-01 -7.28256285e-01 4.30240899e-01
3.57068568e-01 5.36205433e-02 8.04899514e-01 -7.98695683e-02
-5.24616957e-01 -3.45239878e-01 -1.18742073e+00 8.05149496e-01
1.14967060e+00 -7.61979997e-01 -3.59471381e-01 1.57008693e-02
6.00651443e-01 -5.00407279e-01 -6.21141851e-01 4.01219487e-01
8.05735469e-01 -1.01314747e+00 8.62648726e-01 -6.23587728e-01
4.28219944e-01 -9.30635035e-02 -5.60556315e-02 -8.69437754e-01
3.80119890e-01 -1.09241247e+00 -2.07606986e-01 1.62659907e+00
3.48759234e-01 -1.20859481e-01 9.72252846e-01 6.65096283e-01
1.61256462e-01 -6.35446012e-01 -9.14285421e-01 -2.63219804e-01
-1.20393172e-01 -7.09737003e-01 7.40771413e-01 1.00013518e+00
1.69315353e-01 6.23218775e-01 -5.75016439e-01 2.15909168e-01
5.68174422e-01 -1.31734416e-01 1.15318847e+00 -1.23515868e+00
-1.50218561e-01 -4.34503675e-01 -1.36728972e-01 -1.34077752e+00
6.71036482e-01 -7.23893821e-01 1.76052645e-01 -9.34759259e-01
1.23879269e-01 -5.50502837e-01 2.40377203e-01 5.79779863e-01
1.32097080e-01 2.97691256e-01 2.50331044e-01 1.71547607e-01
-4.76467848e-01 7.04478204e-01 1.36179531e+00 -6.11383989e-02
-4.23527181e-01 1.16926648e-01 -6.64426684e-01 1.16072536e+00
5.52708268e-01 -1.55343518e-01 -7.08574176e-01 -5.07427990e-01
-2.08668098e-01 4.38730478e-01 4.72660482e-01 -7.98335969e-01
3.16888988e-01 -2.23129511e-01 2.22816169e-01 -3.63769740e-01
7.34535754e-01 -9.95973587e-01 -1.95398573e-02 -2.34492406e-01
-6.61030173e-01 -1.98480353e-01 4.16480064e-01 2.74799049e-01
-1.89821079e-01 1.76211402e-01 8.49042833e-01 1.56175449e-01
-4.31964546e-01 4.77827847e-01 -6.23316914e-02 1.03970811e-01
9.25889552e-01 -4.50655729e-01 1.77055553e-01 -9.13755417e-01
-9.18768287e-01 1.29670352e-01 3.06170911e-01 8.31727445e-01
6.82538986e-01 -1.43486607e+00 -6.12231255e-01 6.84921086e-01
-2.21984331e-02 2.47020982e-02 2.86625415e-01 6.09628022e-01
-4.33817208e-02 1.30351037e-01 -2.67402202e-01 -9.64138567e-01
-1.67927873e+00 2.49087811e-01 4.96108145e-01 4.00114417e-01
-3.89715672e-01 8.52380276e-01 1.74596742e-01 -4.89649236e-01
8.80087972e-01 -1.23662420e-01 -2.83473521e-01 8.74210820e-02
3.89161289e-01 2.38558933e-01 -3.32884401e-01 -1.09997106e+00
-1.87162295e-01 8.46879542e-01 1.64649859e-01 -3.24631572e-01
1.08440864e+00 -5.07049739e-01 1.69334710e-01 4.50661927e-01
1.14115334e+00 1.43812135e-01 -1.58127248e+00 -2.26415917e-01
-4.38212931e-01 -2.59751409e-01 -9.10062715e-02 -7.10955679e-01
-1.10209882e+00 1.02256787e+00 7.75151610e-01 -1.73314109e-01
1.21679270e+00 2.68015526e-02 6.79335415e-01 2.72983611e-02
6.80286139e-02 -9.96632576e-01 2.03315318e-01 1.14833407e-01
8.35655689e-01 -1.48716021e+00 -3.93233746e-01 -6.60688996e-01
-8.51997018e-01 9.46281791e-01 9.10356283e-01 3.32666725e-01
7.24606991e-01 5.55798411e-01 6.07847691e-01 1.68609262e-01
-8.49407256e-01 -6.13682978e-02 3.78808677e-01 9.24631536e-01
4.91177112e-01 -2.97145814e-01 2.93294281e-01 6.10480130e-01
-1.90396488e-01 4.46063936e-01 1.41479420e-02 4.15721357e-01
-1.29942954e-01 -1.15851939e+00 -2.44748995e-01 -1.45104840e-01
-2.53077775e-01 7.51388669e-02 -3.14979434e-01 9.29775000e-01
2.57633120e-01 9.66809034e-01 1.88486576e-01 -2.98775315e-01
3.09647799e-01 3.47737819e-01 4.18711603e-01 -3.46376419e-01
-2.65921682e-01 2.35796049e-01 6.01567142e-03 -5.35378456e-01
-7.33406603e-01 -5.91818631e-01 -1.17672956e+00 -3.37603360e-01
-1.54585898e-01 -1.23020172e-01 6.65451527e-01 8.05869758e-01
6.07349992e-01 1.20910950e-01 8.97925258e-01 -9.58292902e-01
-4.49873239e-01 -1.01652157e+00 -2.42121056e-01 5.24148047e-01
5.99774659e-01 -7.93331504e-01 -2.33267859e-01 4.70628321e-01] | [13.235984802246094, -0.4164407253265381] |
08a60798-78ec-49b1-a557-e2c82b9cfd49 | enhancing-entity-boundary-detection-for | null | null | https://aclanthology.org/2021.acl-short.4 | https://aclanthology.org/2021.acl-short.4.pdf | Enhancing Entity Boundary Detection for Better Chinese Named Entity Recognition | In comparison with English, due to the lack of explicit word boundary and tenses information, Chinese Named Entity Recognition (NER) is much more challenging. In this paper, we propose a boundary enhanced approach for better Chinese NER. In particular, our approach enhances the boundary information from two perspectives. On one hand, we enhance the representation of the internal dependency of phrases by an additional Graph Attention Network(GAT) layer. On the other hand, taking the entity head-tail prediction (i.e., boundaries) as an auxiliary task, we propose an unified framework to learn the boundary information and recognize the NE jointly. Experiments on both the OntoNotes and the Weibo corpora show the effectiveness of our approach. | ['Fang Kong', 'Chun Chen'] | 2021-08-01 | null | null | null | acl-2021-5 | ['boundary-detection', 'chinese-named-entity-recognition'] | ['computer-vision', 'natural-language-processing'] | [-3.01303715e-01 6.88305199e-02 -7.96100721e-02 -5.42918444e-01
-6.28932357e-01 -6.03239238e-01 1.59088925e-01 1.55788243e-01
-7.53146768e-01 8.11064005e-01 5.50202549e-01 -3.55653018e-01
2.61463284e-01 -7.59786904e-01 -4.70557958e-01 -4.64536369e-01
2.79878616e-01 1.90655977e-01 2.82564729e-01 -1.51832297e-01
-5.62931085e-03 3.98690909e-01 -5.05753219e-01 -5.29867709e-02
1.40759265e+00 7.70791113e-01 3.58221054e-01 7.32340515e-02
-4.70339507e-01 4.85912085e-01 -4.34273332e-01 -5.63920140e-01
-2.57817730e-02 -2.22220317e-01 -8.68738294e-01 -1.92540549e-02
-5.59661724e-02 -1.97049808e-02 -2.49509871e-01 1.30751073e+00
4.72706914e-01 1.82928696e-01 2.90381134e-01 -6.20000482e-01
-8.16939652e-01 9.79047239e-01 -7.06418872e-01 3.60079736e-01
-3.05044726e-02 -2.64704764e-01 1.40956795e+00 -1.06437540e+00
7.29990900e-01 8.54931533e-01 5.68729520e-01 5.59570014e-01
-7.65105069e-01 -6.28024936e-01 6.90573037e-01 1.56936541e-01
-1.52453756e+00 -2.21336886e-01 8.54041815e-01 3.15029873e-03
7.68776000e-01 -1.64271235e-01 1.75583065e-01 7.34249473e-01
-1.81971013e-03 9.03124392e-01 7.83883393e-01 -3.52260470e-01
6.67696074e-02 -9.57179666e-02 5.79806089e-01 3.74757439e-01
3.72956365e-01 -3.63030940e-01 1.68279007e-01 1.38627380e-01
5.27308881e-01 -1.00407012e-01 -4.18019682e-01 1.56656861e-01
-9.24935699e-01 6.43600166e-01 4.29377854e-01 6.71678543e-01
-5.53855240e-01 -1.09503292e-01 3.18653405e-01 -1.35992780e-01
6.99467421e-01 3.18310469e-01 -6.59711719e-01 -1.44691765e-02
-7.25035608e-01 -3.29232782e-01 9.92923379e-01 9.97210085e-01
8.55628908e-01 -1.48021141e-02 -1.60583124e-01 8.48049581e-01
3.82332057e-01 1.70870990e-01 3.37507188e-01 -9.37932804e-02
9.76309299e-01 4.03763860e-01 2.89969176e-01 -8.66258383e-01
-5.68735182e-01 -6.25702620e-01 -7.18185246e-01 -6.14103675e-01
2.86821455e-01 -8.44358683e-01 -9.95128632e-01 1.81765974e+00
3.81791979e-01 3.37080449e-01 1.69060469e-01 9.08365369e-01
8.24346364e-01 8.86021674e-01 4.39466387e-01 -9.24838856e-02
1.54867828e+00 -1.24212670e+00 -1.15044224e+00 -3.41869384e-01
7.90487647e-01 -6.23097837e-01 7.41055429e-01 -1.01400964e-01
-7.53987730e-01 -4.53926772e-01 -7.88348317e-01 -1.26480713e-01
-4.06779826e-01 4.82358307e-01 5.23628116e-01 4.71199393e-01
-6.69933558e-01 4.31914747e-01 -1.01023030e+00 -1.49062380e-01
1.36684254e-01 4.78700161e-01 -2.82759845e-01 -1.53988272e-01
-1.69374478e+00 6.70583308e-01 8.08785141e-01 5.22695899e-01
-2.85165697e-01 -4.26381588e-01 -1.07103503e+00 4.60731655e-01
5.15859723e-01 -1.65608183e-01 9.95288312e-01 -8.16717029e-01
-1.36696529e+00 3.20198894e-01 -1.88467905e-01 -2.04424322e-01
1.63255155e-01 -4.93198991e-01 -6.68617904e-01 -3.06475926e-02
-1.80466678e-02 4.17947769e-01 1.45349249e-01 -1.12251425e+00
-6.25802457e-01 -3.31579983e-01 9.09416229e-02 3.30380499e-01
-6.22569323e-01 2.75593281e-01 -8.11588705e-01 -8.51403594e-01
-2.68185046e-02 -7.84528196e-01 -6.08516157e-01 -7.26100922e-01
-5.90464413e-01 -5.97010136e-01 5.79246998e-01 -1.24959290e+00
1.83088279e+00 -2.25826335e+00 1.01935536e-01 1.73852816e-01
7.62687325e-02 5.22888899e-01 -1.90803453e-01 4.14402187e-01
-1.05717838e-01 6.43424749e-01 -2.30052844e-01 -3.89685959e-01
5.19484282e-02 2.86081046e-01 -5.25961705e-02 1.11029409e-01
5.64191103e-01 9.73036766e-01 -8.57055426e-01 -6.63200736e-01
-3.98285478e-01 4.45293099e-01 -4.02023822e-01 1.76738277e-01
-1.82824489e-02 5.02797067e-01 -9.58214283e-01 4.30492729e-01
9.09104943e-01 -2.28672996e-01 4.61629510e-01 -4.61162739e-02
-4.26182866e-01 7.51513958e-01 -1.27259791e+00 1.54277968e+00
-4.34231222e-01 1.58058986e-01 1.88555837e-01 -6.91541255e-01
9.85430717e-01 6.54848754e-01 3.08454949e-02 -5.57132602e-01
8.88249800e-02 2.48253658e-01 1.77861661e-01 -2.52065659e-01
5.79553545e-01 -2.58547753e-01 -2.58189499e-01 2.95286238e-01
2.61190552e-02 5.67426562e-01 9.40196440e-02 3.07133608e-03
9.53439415e-01 1.84760332e-01 4.11984950e-01 -3.11990410e-01
6.73163235e-01 -4.81009372e-02 1.34998345e+00 2.36777306e-01
-2.69337714e-01 5.15144825e-01 5.70711017e-01 -2.01663747e-01
-7.82570064e-01 -7.16411710e-01 2.46672034e-02 9.17123199e-01
2.94665694e-01 -5.27234852e-01 -9.93362963e-01 -1.26116240e+00
-3.49478722e-01 6.80677831e-01 -4.02928948e-01 2.67107069e-01
-1.26651955e+00 -6.80406451e-01 5.85799694e-01 1.08228922e+00
5.83456814e-01 -1.19204223e+00 1.68657944e-01 4.29421782e-01
-4.03759509e-01 -1.43685448e+00 -1.03608143e+00 2.53363788e-01
-7.67205417e-01 -5.32537818e-01 -8.45066905e-01 -1.17696881e+00
6.44332230e-01 2.48339158e-02 7.77070582e-01 1.82266414e-01
4.07574058e-01 -1.07825011e-01 -6.30761743e-01 -2.99030453e-01
8.13854784e-02 5.68385005e-01 -1.77606717e-01 -2.29076389e-02
5.59419215e-01 -3.88243467e-01 -4.00099486e-01 1.75699040e-01
-8.65784109e-01 -1.25980601e-01 7.91523814e-01 8.67035806e-01
5.43464005e-01 1.51849732e-01 7.30032742e-01 -1.14450610e+00
6.05933011e-01 -5.15237153e-01 -5.81766605e-01 6.13458514e-01
-3.54877084e-01 1.38430253e-01 7.93638468e-01 -3.17778498e-01
-1.65234828e+00 1.42436713e-01 -4.53637481e-01 -5.08203618e-02
-1.68960884e-01 9.82254148e-01 -8.30587268e-01 2.91554987e-01
-3.00779939e-02 6.36588261e-02 -7.70479321e-01 -1.01859140e+00
2.07469881e-01 7.02205241e-01 2.77028590e-01 -6.02254152e-01
8.05121481e-01 9.83950943e-02 -4.65018958e-01 -8.37246716e-01
-1.01867807e+00 -5.65091252e-01 -9.60651040e-01 4.06858802e-01
1.28527009e+00 -8.79828274e-01 -3.72422695e-01 3.05516154e-01
-1.46501791e+00 -1.16905674e-01 1.50165901e-01 8.25722933e-01
1.95025280e-01 7.13959754e-01 -1.07703185e+00 -6.93088531e-01
-5.39230585e-01 -8.08776200e-01 6.64697349e-01 6.69863284e-01
3.10361832e-01 -1.18434823e+00 1.01838358e-01 1.05144642e-01
1.54656976e-01 -4.82771173e-02 9.51267004e-01 -1.22822106e+00
-3.91969442e-01 -1.26040354e-01 -5.90214193e-01 3.15208882e-01
1.41414180e-01 -2.63560027e-01 -6.15136921e-01 5.40782660e-02
-7.82185495e-02 1.50600255e-01 8.80792439e-01 -2.20640842e-02
6.81400657e-01 -2.47491136e-01 -4.44518626e-01 4.53166664e-01
1.46667969e+00 3.18620712e-01 6.03079975e-01 2.23587304e-01
9.86486077e-01 6.59274697e-01 8.28773320e-01 4.58659738e-01
5.89722812e-01 4.48370248e-01 6.07831813e-02 -2.09256858e-01
-4.51479070e-02 -3.64793122e-01 2.97371864e-01 1.35810840e+00
-1.47299379e-01 -4.70349342e-01 -1.17317569e+00 5.94895303e-01
-1.67106402e+00 -4.58144456e-01 -1.16473995e-01 1.85382521e+00
8.85342121e-01 1.62401035e-01 -2.26557449e-01 -4.52954412e-01
1.29419816e+00 1.88965946e-01 -2.34614938e-01 -2.40581900e-01
-1.85472891e-02 2.77825773e-01 4.79964197e-01 4.11656886e-01
-1.22620726e+00 1.38975656e+00 5.01574469e+00 7.68909335e-01
-9.51164722e-01 5.45349866e-02 4.62435693e-01 7.87702560e-01
-4.32902038e-01 3.52441192e-01 -1.39015126e+00 4.37472731e-01
6.44576371e-01 -5.33455610e-02 1.50955811e-01 6.60032094e-01
-4.05579470e-02 3.77488762e-01 -7.72658050e-01 4.62576449e-01
-1.32577389e-01 -8.88314843e-01 -1.47757038e-01 2.41598830e-01
6.44113183e-01 1.07764527e-01 -5.06503403e-01 4.40051407e-01
3.62922609e-01 -6.46624446e-01 5.95490873e-01 3.81811827e-01
5.23530424e-01 -8.97961199e-01 1.17858863e+00 4.02913004e-01
-1.58292747e+00 2.35672072e-01 -5.13713241e-01 1.69877615e-02
4.74058151e-01 5.33781409e-01 -4.28042591e-01 1.08394575e+00
2.71487057e-01 6.07650101e-01 -4.08068329e-01 1.16075897e+00
-9.20735836e-01 9.68643308e-01 -2.44646221e-01 -1.63276330e-01
5.40850639e-01 -3.73722702e-01 3.22001636e-01 1.49528110e+00
1.85197502e-01 5.06681979e-01 3.68971765e-01 8.33626926e-01
-4.76817995e-01 4.71094579e-01 -8.90927985e-02 -2.87866801e-01
6.36337042e-01 1.47875953e+00 -7.63148248e-01 -1.69415280e-01
-7.27237046e-01 1.00396228e+00 8.44254553e-01 4.87707406e-01
-8.39706659e-01 -9.33219969e-01 3.65415335e-01 -4.85564172e-01
8.50758076e-01 -4.91851062e-01 -1.56203747e-01 -1.57073271e+00
1.15703613e-01 -4.30299312e-01 6.11836076e-01 -2.50105500e-01
-1.47204280e+00 7.84117162e-01 -4.76191968e-01 -8.07577312e-01
2.25015149e-01 -5.39171815e-01 -1.02474427e+00 1.04439402e+00
-1.97799730e+00 -1.22157049e+00 2.34776475e-02 2.13160172e-01
2.16483235e-01 5.47392309e-01 6.76440835e-01 5.97395182e-01
-1.12274933e+00 6.74461007e-01 3.79704162e-02 1.03897047e+00
6.07491374e-01 -1.32399297e+00 7.21579850e-01 1.04632378e+00
2.12866217e-01 7.84039199e-01 1.75989270e-01 -9.86802638e-01
-9.05508101e-01 -1.04088974e+00 1.41864192e+00 8.72280151e-02
8.22174728e-01 -5.46356320e-01 -1.13022506e+00 9.61448312e-01
3.19574803e-01 -8.65903124e-02 8.38999748e-01 3.78622383e-01
-1.77243382e-01 1.27054617e-01 -6.05434060e-01 6.14915252e-01
9.52656925e-01 -3.44878674e-01 -1.07011569e+00 -1.80499718e-01
1.01696825e+00 -1.89307362e-01 -1.04911649e+00 3.57372999e-01
1.96296453e-01 -2.82820821e-01 4.35973525e-01 -7.41786659e-01
9.66663212e-02 -2.84438908e-01 -4.43104617e-02 -1.28630733e+00
-5.10140479e-01 -5.23779809e-01 2.22464114e-01 1.97587931e+00
6.94994807e-01 -6.06008172e-01 7.00836420e-01 7.32484937e-01
-5.21418214e-01 -7.03902543e-01 -8.15851331e-01 -8.50554764e-01
2.02024370e-01 -2.28606567e-01 7.05850244e-01 1.04591060e+00
1.70255363e-01 7.02130139e-01 -4.15520728e-01 3.89658749e-01
1.50816113e-01 2.02797383e-01 2.02300340e-01 -1.22002125e+00
-9.79955643e-02 -5.60552478e-02 3.02989241e-02 -1.53454125e+00
4.25043672e-01 -7.70772159e-01 2.73605019e-01 -1.79064536e+00
1.16242610e-01 -6.79850280e-01 -7.58526385e-01 5.35710871e-01
-6.63532972e-01 -2.30706781e-01 3.46097201e-01 1.45121127e-01
-8.18169653e-01 6.95451200e-01 1.20077515e+00 2.48582140e-01
-2.95207620e-01 -6.79396316e-02 -6.59180343e-01 8.05523992e-01
8.02805960e-01 -6.73362255e-01 2.55621433e-01 -8.42774749e-01
2.57435858e-01 5.57865873e-02 -3.42431039e-01 -5.86000085e-01
4.26282912e-01 -7.74825290e-02 2.64908463e-01 -6.66845798e-01
-6.47764141e-03 -6.58231318e-01 -5.29357970e-01 2.48288855e-01
-1.33097529e-01 2.16656402e-01 1.78112850e-01 7.55260468e-01
-4.36554074e-01 -5.69006443e-01 4.10808861e-01 -1.94004372e-01
-9.01159883e-01 4.35126841e-01 -1.06432185e-01 4.91435289e-01
7.26045072e-01 1.77742139e-01 -3.25196326e-01 -1.01213709e-01
-7.41780698e-01 5.98898292e-01 2.84081727e-01 2.93346375e-01
4.13401842e-01 -1.26916087e+00 -6.35951698e-01 -7.04242140e-02
-5.41285761e-02 9.92016643e-02 3.91871840e-01 8.65961850e-01
-2.84292787e-01 4.13782448e-01 1.11499891e-01 1.47695035e-01
-9.79860902e-01 6.71001017e-01 4.86081876e-02 -9.39935327e-01
-4.62275326e-01 7.13469923e-01 5.45557559e-01 -3.86940897e-01
8.62146914e-02 -1.51466653e-01 -8.57349575e-01 1.18965566e-01
4.54446673e-01 1.44601762e-01 -1.40201733e-01 -7.12926626e-01
-5.14059782e-01 4.25836533e-01 -3.26228738e-01 -5.18544856e-03
1.54619336e+00 -2.44230568e-01 -2.28910565e-01 1.09058812e-01
9.39750493e-01 6.09439373e-01 -1.08469951e+00 -3.40606570e-01
8.70157123e-01 2.14747787e-02 -1.39767870e-01 -6.28962874e-01
-1.02562177e+00 1.08905172e+00 -2.11008802e-01 -4.93310504e-02
9.76809978e-01 -4.79164608e-02 1.21147525e+00 3.72059524e-01
9.82103050e-02 -1.25088549e+00 -3.69915336e-01 9.50620234e-01
3.34344745e-01 -1.03721869e+00 -3.21477413e-01 -8.14720392e-01
-7.96662450e-01 1.13421118e+00 8.34348798e-01 -3.05376779e-02
7.21141398e-01 2.04245165e-01 4.43977490e-02 6.72212839e-02
-4.93514806e-01 -4.74599808e-01 4.32551771e-01 2.87879378e-01
6.17892563e-01 1.19680567e-02 -9.46656406e-01 1.23814261e+00
4.63522762e-01 -2.33111367e-01 3.74182284e-01 8.05427492e-01
-3.51273596e-01 -1.38740408e+00 -2.86200456e-03 2.17682458e-02
-1.02568758e+00 -6.72553599e-01 -3.00446719e-01 8.58337522e-01
-3.72938113e-03 9.69388008e-01 -7.94049203e-02 -2.92200387e-01
3.53570759e-01 2.31489092e-01 -1.17528982e-01 -1.02734649e+00
-8.09271812e-01 2.96333879e-01 4.08841193e-01 -1.61531940e-01
-1.08448319e-01 -5.05909145e-01 -1.74658751e+00 2.29546383e-01
-8.81107211e-01 6.60790980e-01 5.38776040e-01 1.06303978e+00
4.79716599e-01 4.40143615e-01 6.17736042e-01 -2.36563474e-01
-5.73583782e-01 -1.05471396e+00 -8.51138830e-01 4.14399296e-01
-2.51496017e-01 -3.57765436e-01 -9.86239463e-02 -3.23383510e-01] | [9.75252628326416, 9.741922378540039] |
3c29e238-a1e4-4dee-b2f0-47397ee57c6e | end-to-end-3d-point-cloud-instance | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Jiang_End-to-End_3D_Point_Cloud_Instance_Segmentation_Without_Detection_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Jiang_End-to-End_3D_Point_Cloud_Instance_Segmentation_Without_Detection_CVPR_2020_paper.pdf | End-to-End 3D Point Cloud Instance Segmentation Without Detection | 3D instance segmentation plays a predominant role in environment perception of robotics and augmented reality. Many deep learning based methods have been presented recently for this task. These methods rely on either a detection branch to propose objects or a grouping step to assemble same-instance points. However, detection based methods do not ensure a consistent instance label for each point, while the grouping step requires parameter-tuning and is computationally expensive. In this paper, we introduce a novel framework to enable end-to-end instance segmentation without detection and a separate step of grouping. The core idea is to convert instance segmentation to a candidate assignment problem. At first, a set of instance candidates is sampled. Then we propose an assignment module for candidate assignment and a suppression module to eliminate redundant candidates. A mapping between instance labels and instance candidates is further sought to construct an instance grouping loss for the network training. Experimental results demonstrate that our method is more effective and efficient than previous approaches.
| [' Jun Xiao', ' Jianmin Zheng', ' Jianfei Cai', ' Feilong Yan', 'Haiyong Jiang'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['3d-instance-segmentation-1'] | ['computer-vision'] | [ 3.68882596e-01 9.02170315e-02 -2.56458133e-01 -6.86576605e-01
-6.25841916e-01 -2.34501511e-01 4.68878448e-01 2.89988697e-01
-3.95579845e-01 4.21762019e-01 -3.10883701e-01 -5.27531356e-02
-1.97743490e-01 -8.88442814e-01 -7.26061702e-01 -5.73505104e-01
5.89474589e-02 6.34761393e-01 6.74194157e-01 7.77675435e-02
5.24417758e-01 5.50673366e-01 -1.61118400e+00 1.54378086e-01
1.03404939e+00 1.03251064e+00 7.07564652e-01 1.84538111e-01
-5.50835550e-01 3.03647488e-01 -6.06996477e-01 1.76318303e-01
5.46344042e-01 -5.19238293e-01 -8.15452039e-01 6.54716253e-01
2.93319672e-01 -2.11108744e-01 2.60785043e-01 1.17509282e+00
6.02819920e-01 3.32203180e-01 7.49974310e-01 -1.13413537e+00
-5.88275827e-02 4.30745244e-01 -7.88456500e-01 -4.39211801e-02
1.74152210e-01 -9.29791629e-02 8.32190394e-01 -1.00301123e+00
4.38460022e-01 1.00532055e+00 5.07999361e-01 4.40878302e-01
-9.27634656e-01 -5.28382778e-01 5.31641006e-01 1.43702149e-01
-1.36047781e+00 -1.29316598e-01 1.08796227e+00 -3.10527891e-01
6.31265283e-01 1.79027110e-01 9.99901056e-01 3.79573375e-01
-3.21124583e-01 9.99441385e-01 8.54407787e-01 -5.22421479e-01
4.56321239e-01 2.18635574e-01 4.51296359e-01 8.14369261e-01
9.42777619e-02 -3.03337425e-01 -5.59682250e-02 1.21461920e-01
1.01700187e+00 1.95554167e-01 -1.49238527e-01 -7.15245962e-01
-1.12647653e+00 7.10282981e-01 8.49568844e-01 2.34594405e-01
-4.46058422e-01 -4.43999209e-02 2.89743334e-01 -3.92353944e-02
2.76343316e-01 4.53904718e-01 -5.41290224e-01 4.82520401e-01
-9.60542679e-01 2.91934311e-01 6.31470263e-01 9.49775934e-01
9.76853192e-01 -3.47439408e-01 -3.16492841e-02 1.05090475e+00
4.24997598e-01 -9.49661732e-02 3.94614607e-01 -6.58625901e-01
5.12253702e-01 1.00074828e+00 1.56109616e-01 -1.01249981e+00
-5.78993440e-01 -7.51243830e-01 -5.83227277e-01 3.51170301e-01
2.14233041e-01 8.88577923e-02 -1.22217584e+00 1.42291069e+00
7.80141234e-01 4.16952044e-01 -1.49128824e-01 1.22531092e+00
8.13826859e-01 6.84040487e-01 -1.34853542e-01 -4.31418382e-02
1.27962744e+00 -1.27773511e+00 -5.62539101e-01 -3.58416200e-01
5.50813138e-01 -6.04353189e-01 1.06639254e+00 4.00786072e-01
-1.02396572e+00 -7.95772254e-01 -1.15049469e+00 -9.66201946e-02
-4.20390248e-01 3.45510066e-01 5.81217289e-01 4.12204295e-01
-6.95706487e-01 4.00631875e-01 -7.54753351e-01 -2.29893327e-01
4.88020927e-01 6.89299583e-01 -2.02383831e-01 1.82264552e-01
-6.97923303e-01 6.38940632e-01 8.36267829e-01 3.48086536e-01
-7.45053887e-01 -2.24680081e-01 -8.99316907e-01 -4.99431454e-02
4.90056336e-01 -6.75548851e-01 1.18145788e+00 -1.01047671e+00
-1.49094677e+00 8.72637928e-01 -4.08012979e-03 -4.76701438e-01
3.98702919e-01 -2.48225972e-01 9.64275375e-02 -3.16283144e-02
1.26460925e-01 7.91100740e-01 8.20122957e-01 -1.73135066e+00
-1.16395628e+00 -4.56495345e-01 1.96774796e-01 5.90529919e-01
-1.08232647e-01 -2.52043843e-01 -8.59329522e-01 -2.88346946e-01
9.37963963e-01 -7.18446910e-01 -6.19581521e-01 -1.58716738e-01
-6.71650767e-01 -3.43609780e-01 8.03638995e-01 -4.21613097e-01
1.00012267e+00 -1.88520181e+00 1.40240714e-01 3.97708356e-01
1.81814477e-01 3.42718899e-01 -8.44701938e-03 2.07709037e-02
1.45753343e-02 1.87566402e-04 -4.87817317e-01 -6.30608737e-01
-7.65668303e-02 2.03902461e-02 -1.43353313e-01 3.64523381e-01
2.71344841e-01 6.15211546e-01 -8.05505276e-01 -4.91654396e-01
6.36857748e-01 2.34553531e-01 -7.75585771e-01 1.87493861e-01
-4.83031094e-01 4.85097378e-01 -5.60082436e-01 6.67082608e-01
9.08007264e-01 -1.12499058e-01 1.54560665e-02 -2.33783767e-01
-2.10592851e-01 4.00488973e-01 -1.56764638e+00 1.88840783e+00
-2.00888947e-01 2.28124484e-03 3.32606435e-02 -1.35892606e+00
1.28779697e+00 -1.00536816e-01 4.97398347e-01 -4.37820703e-01
3.25046897e-01 3.87043685e-01 -3.42300572e-02 -5.35715699e-01
5.69741964e-01 -4.74743284e-02 -8.46975204e-03 1.94856942e-01
-1.58551380e-01 -2.33622283e-01 1.78879529e-01 -1.89325407e-01
6.97684526e-01 4.59330022e-01 2.17264459e-01 -2.95570772e-02
7.10985541e-01 2.79652447e-01 7.89144397e-01 6.03636026e-01
-2.09783658e-01 8.07498395e-01 5.91951273e-02 -6.39796793e-01
-7.70417333e-01 -7.44227827e-01 -1.65137425e-01 5.84072649e-01
8.46605659e-01 -7.01992512e-02 -8.28746498e-01 -9.14267063e-01
-2.65709877e-01 4.19296384e-01 -2.74827689e-01 -6.19190708e-02
-7.14782894e-01 -8.33971500e-01 -1.37074903e-01 4.55355793e-01
7.66566455e-01 -1.14094126e+00 -7.47279346e-01 3.39672863e-01
-6.39612377e-02 -8.20459843e-01 -2.83508867e-01 5.39203346e-01
-1.11010909e+00 -8.94531429e-01 -5.00783205e-01 -1.22851527e+00
1.16915941e+00 3.82117659e-01 8.31328630e-01 3.60052943e-01
-1.60938844e-01 -1.81948796e-01 -4.26933616e-01 -4.44939464e-01
1.11419119e-01 3.23774874e-01 -1.05237179e-01 -3.63702811e-02
3.63054663e-01 -4.45552677e-01 -6.99726701e-01 4.19520497e-01
-7.71715999e-01 6.06332682e-02 8.43381166e-01 7.27010965e-01
1.06170774e+00 7.04227239e-02 6.42723858e-01 -7.85789430e-01
3.20663184e-01 -2.26563305e-01 -7.53030658e-01 4.96954471e-02
-1.65475026e-01 -1.11978397e-01 5.22409856e-01 -2.78039277e-01
-9.24599886e-01 6.91842079e-01 -2.40192756e-01 -4.38069105e-01
-5.04199326e-01 5.35448074e-01 -4.51822311e-01 2.25035120e-02
4.37994629e-01 1.91540360e-01 -1.50110513e-01 -6.05344355e-01
1.91333115e-01 5.24402499e-01 4.06398088e-01 -4.78471160e-01
6.90916359e-01 4.04137850e-01 -1.69223323e-01 -5.48490286e-01
-7.89846659e-01 -7.54573941e-01 -8.58307540e-01 -1.12768933e-01
9.00263906e-01 -6.62631452e-01 -3.81502330e-01 3.58349055e-01
-1.22059047e+00 -2.90438324e-01 -3.04012448e-01 5.15547395e-01
-5.52813709e-01 1.92037970e-01 -2.70412594e-01 -8.41481030e-01
-1.50523901e-01 -1.27643120e+00 1.26965630e+00 3.63121420e-01
2.29646355e-01 -6.38482809e-01 -1.31987393e-01 3.06044579e-01
-1.90453500e-01 2.19885617e-01 6.01284087e-01 -7.08541393e-01
-9.00937021e-01 -3.91624212e-01 -2.84842730e-01 1.53278530e-01
2.49696225e-01 -1.68634921e-01 -6.86594367e-01 -1.24467447e-01
1.06760405e-01 -1.17184646e-01 8.09901834e-01 5.63009977e-01
1.40293062e+00 -3.21587250e-02 -7.62619197e-01 6.43402874e-01
1.49029732e+00 5.26509643e-01 5.91678500e-01 4.98522997e-01
7.50788450e-01 6.56867087e-01 1.01153433e+00 2.18672916e-01
2.38326281e-01 6.30418897e-01 5.59241235e-01 -3.31622630e-01
1.05775163e-01 -1.70560449e-01 -6.15357347e-02 6.58838511e-01
1.09361224e-01 -2.23266900e-01 -9.10537779e-01 6.21849954e-01
-2.02335024e+00 -7.58396029e-01 -1.86816022e-01 2.29106355e+00
5.58392823e-01 4.10171360e-01 2.55381733e-01 6.01742804e-01
1.00039852e+00 -1.62336543e-01 -4.21271622e-01 -3.20215970e-02
3.74170631e-01 8.31078142e-02 1.74110219e-01 5.36018312e-01
-1.31042695e+00 1.13882387e+00 5.38177872e+00 6.28567636e-01
-1.08205426e+00 -1.92972943e-01 5.71911931e-01 1.56840563e-01
-1.56419184e-02 2.03587767e-02 -9.77450430e-01 3.40160340e-01
-1.06969841e-01 4.33093578e-01 -1.19533107e-01 9.70924675e-01
2.23279029e-01 -2.09383696e-01 -1.00274682e+00 8.87600601e-01
-4.11330014e-02 -1.05098557e+00 6.22542948e-02 -1.16590537e-01
6.01896524e-01 -3.52912754e-01 -2.13277429e-01 2.40179241e-01
-2.69196592e-02 -7.19147801e-01 7.53454685e-01 3.64087343e-01
6.00904562e-02 -9.27097440e-01 5.93724489e-01 7.27698565e-01
-1.21540964e+00 -1.04928888e-01 -5.73455274e-01 -2.11744785e-01
1.68443486e-01 7.27099478e-01 -1.11032045e+00 7.75858760e-01
4.66328055e-01 4.85747725e-01 -2.65167028e-01 1.59820867e+00
-3.24141681e-01 1.87941283e-01 -4.60545748e-01 -1.54211670e-01
4.68011111e-01 -3.98095310e-01 5.71725965e-01 9.04066443e-01
2.61519194e-01 2.87294742e-02 7.86317050e-01 9.09489334e-01
1.88497216e-01 1.68849498e-01 -3.96448314e-01 4.87878770e-01
6.48628950e-01 1.34609628e+00 -1.38136947e+00 -3.95572692e-01
-1.83174923e-01 1.07161832e+00 3.85408372e-01 1.30876392e-01
-9.39687490e-01 -6.06469095e-01 2.32602373e-01 1.35026410e-01
4.14696783e-01 -3.35254520e-01 -4.96396184e-01 -8.14643681e-01
1.27851844e-01 -6.59940004e-01 1.58021942e-01 -5.21393299e-01
-8.44605207e-01 5.03332615e-01 -9.50926095e-02 -1.48605812e+00
-1.56568959e-01 -4.89082307e-01 -6.10352337e-01 7.58816659e-01
-1.63197136e+00 -1.01454914e+00 -5.74158192e-01 4.70048517e-01
9.00331438e-01 1.23689704e-01 4.53329533e-01 3.79846185e-01
-7.87754953e-01 3.22710574e-01 -3.81773084e-01 1.02887966e-01
2.13882402e-01 -1.36334324e+00 1.20506518e-01 9.71040130e-01
2.21409023e-01 7.70819426e-01 5.76611400e-01 -7.23168492e-01
-1.12243772e+00 -1.05182576e+00 4.69646662e-01 -1.94485471e-01
-3.41656320e-02 -3.50079596e-01 -8.22316408e-01 5.85413694e-01
-1.13020889e-01 9.35743824e-02 3.50371361e-01 3.84313278e-02
1.75929427e-01 -1.32163242e-01 -1.31599438e+00 5.21070123e-01
1.15589273e+00 4.48596068e-02 -6.72383547e-01 3.33493769e-01
6.75723135e-01 -6.28488779e-01 -3.70137453e-01 7.24421144e-01
3.27938974e-01 -9.19514537e-01 8.80481303e-01 -2.01740533e-01
3.05166960e-01 -1.03530300e+00 1.51850432e-01 -1.05490494e+00
-3.27656984e-01 -2.88989663e-01 5.08171692e-02 1.18215418e+00
4.24758822e-01 -3.60721946e-01 1.03408718e+00 3.18729311e-01
-6.90573573e-01 -1.02636433e+00 -5.72329164e-01 -6.82546914e-01
-3.54090273e-01 -3.56170058e-01 5.76268554e-01 9.20019388e-01
-2.92577922e-01 3.74207497e-01 6.22327104e-02 5.22047579e-01
6.81568503e-01 4.63752359e-01 9.77876782e-01 -1.39814758e+00
-1.86760277e-01 -4.63380992e-01 -4.48931813e-01 -1.35997438e+00
-1.51222512e-01 -7.78477132e-01 4.41892833e-01 -2.11410499e+00
-5.79777360e-02 -9.84329700e-01 -2.99328983e-01 4.48169172e-01
-3.11183780e-01 1.37764767e-01 -5.49693555e-02 2.61666656e-01
-5.92237830e-01 4.72571671e-01 1.33677232e+00 -3.90640553e-03
-8.45392168e-01 2.27640286e-01 -4.13810819e-01 8.41677547e-01
9.44364667e-01 -3.31944108e-01 -6.29898310e-01 -3.69440109e-01
-1.73308983e-01 -2.75710732e-01 4.10950273e-01 -1.36832345e+00
3.10856104e-01 -1.03529200e-01 4.78781402e-01 -1.12884212e+00
3.99381369e-01 -9.83915329e-01 -5.82465567e-02 4.62934792e-01
-1.39815122e-01 -2.68260062e-01 2.97203772e-02 5.36311626e-01
-2.08626032e-01 -5.42670250e-01 8.36351275e-01 -4.56372708e-01
-1.05375433e+00 3.76404524e-01 7.17177019e-02 -4.30843800e-01
1.24414515e+00 -6.81116700e-01 1.87892318e-01 2.51824021e-01
-8.21353853e-01 3.67549062e-01 3.56682718e-01 3.57383192e-01
7.06907809e-01 -1.18273664e+00 -3.42086196e-01 1.75733984e-01
1.65210832e-02 8.35423052e-01 -1.20213017e-01 7.34848797e-01
-5.58965206e-01 3.36239375e-02 -4.22675386e-02 -8.29340100e-01
-1.14681458e+00 5.57270825e-01 3.88013989e-01 -6.29573315e-02
-6.77837729e-01 1.12260699e+00 2.59842813e-01 -8.38978648e-01
6.24167860e-01 -5.28618038e-01 -3.64261389e-01 5.26147932e-02
2.97788382e-01 9.43472981e-02 -5.52335661e-03 -4.44763541e-01
-2.73551911e-01 7.56057322e-01 -1.00760177e-01 -7.50741959e-02
1.25008869e+00 -1.08604781e-01 -7.98462480e-02 4.72793907e-01
9.66087103e-01 -3.09923589e-01 -1.14934373e+00 -3.94585021e-02
1.76913619e-01 -4.55419749e-01 1.59499713e-03 -6.17425323e-01
-1.09022653e+00 7.99847305e-01 6.21776700e-01 4.40937847e-01
1.22846770e+00 -1.42064065e-01 8.40638161e-01 3.45323414e-01
4.94211376e-01 -1.16510022e+00 8.97071660e-02 5.71057379e-01
7.88136363e-01 -1.19021189e+00 -1.29882962e-01 -8.28235209e-01
-1.61078081e-01 8.92083585e-01 1.09952474e+00 -3.05217236e-01
5.19549966e-01 1.30369008e-01 9.86358337e-03 -2.93023020e-01
-8.86768326e-02 -4.44070041e-01 3.67882729e-01 5.02135158e-01
2.96226472e-01 -2.12905422e-01 -7.56364703e-01 5.60500324e-01
-6.05840012e-02 -6.12634607e-03 9.03684720e-02 1.09572518e+00
-9.87057626e-01 -1.13551927e+00 -4.50992495e-01 4.71526712e-01
-1.97844580e-01 2.39366114e-01 -3.40694815e-01 7.46720135e-01
6.29626691e-01 7.32072532e-01 1.35052204e-01 -4.97873157e-01
3.96939635e-01 5.34256138e-02 4.23272103e-01 -9.90260243e-01
-6.50253236e-01 2.69497246e-01 -1.33420289e-01 -3.32917839e-01
-6.41267717e-01 -4.64925677e-01 -1.76455534e+00 3.91931862e-01
-8.53956342e-01 1.13133423e-01 7.98231125e-01 9.36848342e-01
2.66872197e-01 6.38154924e-01 7.33872116e-01 -1.14420140e+00
-1.77108809e-01 -8.00233245e-01 -3.95280451e-01 1.88865468e-01
1.70237243e-01 -7.63395667e-01 -2.06317917e-01 -2.06302144e-02] | [8.089330673217773, -2.9907619953155518] |
777a1cd3-9926-4d42-8a8a-3131ff92687f | crop-rotation-modeling-for-deep-learning | 2110.08187 | null | https://arxiv.org/abs/2110.08187v2 | https://arxiv.org/pdf/2110.08187v2.pdf | Crop Rotation Modeling for Deep Learning-Based Parcel Classification from Satellite Time Series | While annual crop rotations play a crucial role for agricultural optimization, they have been largely ignored for automated crop type mapping. In this paper, we take advantage of the increasing quantity of annotated satellite data to propose the first deep learning approach modeling simultaneously the inter- and intra-annual agricultural dynamics of parcel classification. Along with simple training adjustments, our model provides an improvement of over 6.3 mIoU points over the current state-of-the-art of crop classification. Furthermore, we release the first large-scale multi-year agricultural dataset with over 300,000 annotated parcels. | ['Loic Landrieu', 'Félix Quinton'] | 2021-10-15 | null | null | null | null | ['crop-classification'] | ['miscellaneous'] | [-1.66729480e-01 -2.55795032e-01 -5.40553391e-01 -3.30218643e-01
-3.23203117e-01 -9.19307172e-01 4.59823489e-01 7.11210251e-01
-2.54882455e-01 8.61707270e-01 -6.45155385e-02 -5.56566119e-01
7.54036009e-02 -1.30715132e+00 -9.91660297e-01 -6.51414156e-01
-5.80649197e-01 1.70442536e-01 6.46157861e-02 -8.01900864e-01
-4.47101891e-01 6.39699936e-01 -1.68559468e+00 7.69737065e-02
1.16072738e+00 9.41225469e-01 4.18696910e-01 6.63941622e-01
-1.43508956e-01 2.66259044e-01 -2.29866371e-01 3.94223444e-02
2.38792449e-01 5.97089529e-02 -4.11007285e-01 -9.07100737e-02
3.78856301e-01 -6.46484256e-01 -3.91412199e-01 1.02127922e+00
3.28059137e-01 -2.88677335e-01 4.51012492e-01 -9.12412405e-01
-6.07050717e-01 8.07590604e-01 -8.18789601e-01 3.31957713e-02
-2.41843030e-01 -1.33411661e-01 1.04977643e+00 -5.37964702e-01
5.15039206e-01 9.05543268e-01 8.44288886e-01 -2.81023026e-01
-1.18957925e+00 -5.79478443e-01 5.26452303e-01 -1.10621668e-01
-1.32444632e+00 -3.34519655e-01 3.28203827e-01 -5.82390606e-01
1.08995867e+00 1.47546247e-01 1.27293587e+00 4.79860991e-01
4.47969228e-01 1.02531898e+00 8.90438914e-01 -3.07455868e-01
-1.07886061e-01 -3.86166185e-01 1.84707597e-01 6.19387209e-01
4.33771819e-01 3.69076908e-01 -1.66295722e-01 -5.71864136e-02
9.36944425e-01 -5.96956396e-03 1.40189976e-01 -4.85592693e-01
-1.22861814e+00 8.64020050e-01 9.50685799e-01 2.97908306e-01
-7.57088900e-01 2.02338517e-01 2.21254155e-01 3.01786602e-01
1.06079352e+00 5.70826113e-01 -1.04938030e+00 3.48100625e-02
-1.18341589e+00 7.00041592e-01 5.36436439e-01 1.05870152e+00
7.60959744e-01 3.06315720e-01 2.06947356e-01 7.88399696e-01
1.22639969e-01 9.06655252e-01 -2.46403903e-01 -6.31969035e-01
2.03289449e-01 6.93647504e-01 4.61723655e-01 -1.05696392e+00
-6.43499911e-01 -5.66012979e-01 -1.02530706e+00 1.06640197e-01
3.60199243e-01 -2.82967269e-01 -1.05976689e+00 1.42535019e+00
-8.26098304e-03 -1.11249067e-01 -5.99578172e-02 6.94254458e-01
5.28706670e-01 7.13791132e-01 3.88086021e-01 -5.88766672e-02
1.39206111e+00 -5.96798182e-01 -6.97592020e-01 -3.04534316e-01
6.61335230e-01 -6.37180090e-01 5.95140338e-01 1.38104513e-01
-6.00062251e-01 -3.15261513e-01 -1.25980318e+00 2.97460288e-01
-1.08479261e+00 4.45481569e-01 1.42859471e+00 3.75394821e-01
-1.02771223e+00 4.66335803e-01 -9.54722524e-01 -6.07126474e-01
7.09966362e-01 1.43576771e-01 -4.97848392e-01 1.92283001e-02
-1.16839004e+00 9.48539734e-01 6.67589188e-01 5.38433969e-01
-9.10168886e-01 -1.03173351e+00 -8.33018243e-01 -1.11289020e-03
2.82767385e-01 -1.55952543e-01 9.49706614e-01 -8.73330176e-01
-1.23041022e+00 1.08241487e+00 -1.63966909e-01 -6.14127338e-01
1.85308322e-01 -5.16541004e-01 -3.45382482e-01 -3.83676708e-01
1.24794364e-01 1.01364076e+00 3.60933810e-01 -1.35120666e+00
-1.02323520e+00 -4.09186274e-01 2.97898620e-01 6.70872182e-02
-3.54893982e-01 -2.30980031e-02 6.17512651e-02 -9.41422760e-01
2.61179209e-01 -1.06976843e+00 -4.92511064e-01 1.73750408e-02
3.11353449e-02 2.51826763e-01 5.67173779e-01 -1.11614585e+00
1.12708533e+00 -1.85828531e+00 1.15709573e-01 -2.19311267e-01
2.51141898e-02 3.62518489e-01 -4.47983742e-01 2.47733116e-01
2.20175483e-03 2.45296612e-01 -7.07598388e-01 2.99752742e-01
-1.41195282e-01 3.39189410e-01 -4.47191298e-01 4.31907862e-01
7.08869696e-01 1.22560930e+00 -1.06896889e+00 -4.75261407e-03
2.74154335e-01 1.64385244e-01 -3.00239444e-01 2.44051382e-01
-5.98493397e-01 2.53298819e-01 -4.76296157e-01 1.35111225e+00
1.21057403e+00 2.02930152e-01 2.23970979e-01 -7.07719103e-02
-5.82215011e-01 -2.29775965e-01 -7.08977520e-01 1.59275973e+00
-2.48158932e-01 7.47793913e-01 2.56724864e-01 -1.17726493e+00
8.40861022e-01 1.45728171e-01 7.15879798e-01 -4.28762555e-01
-1.05094925e-01 3.31691056e-01 1.41625598e-01 -4.74845953e-02
9.28093076e-01 4.31228340e-01 -3.08515668e-01 8.87441337e-02
1.59000739e-01 -2.77345091e-01 2.21299991e-01 -2.81247020e-01
5.15914142e-01 5.93958795e-01 6.09156489e-01 -8.02192688e-01
3.63279171e-02 4.64686990e-01 4.85868007e-01 6.70887530e-01
-2.89552957e-01 4.75193411e-01 5.63677669e-01 -9.93701875e-01
-1.15239608e+00 -7.92939961e-01 -4.15904194e-01 1.42825162e+00
-4.62731458e-02 -8.85212421e-03 -5.34399152e-01 -5.82170606e-01
6.00619078e-01 4.47469145e-01 -6.65781915e-01 3.76279354e-01
-5.67288220e-01 -1.71889246e+00 7.83745050e-01 6.17796957e-01
5.62677503e-01 -1.11871445e+00 -4.52567220e-01 5.66909969e-01
-1.22623660e-01 -1.04960573e+00 2.86801457e-01 5.95094442e-01
-8.68420780e-01 -8.89337659e-01 -8.32537830e-01 -6.43258750e-01
4.16242778e-01 3.06776851e-01 1.51320314e+00 -1.40739158e-01
-2.99765915e-01 -3.99049819e-01 -4.48396474e-01 -9.08781588e-01
-2.35767394e-01 9.65999544e-01 -4.27351221e-02 -4.50998366e-01
4.76466089e-01 -5.84704220e-01 -2.94143766e-01 -7.47969449e-02
-6.25415206e-01 3.62915434e-02 6.76588178e-01 8.08079123e-01
5.34160078e-01 -3.97041626e-02 7.65642285e-01 -6.22777700e-01
1.45051315e-01 -7.55094171e-01 -1.20191908e+00 3.49162251e-01
-4.83984798e-01 -1.51073754e-01 1.66589662e-01 -1.21437430e-01
-8.21142614e-01 5.14697373e-01 -9.83582810e-02 2.02563733e-01
-4.60075766e-01 1.15488517e+00 -1.75063666e-02 -1.23053581e-01
2.23632619e-01 -1.32887531e-02 -3.24822366e-01 -4.18095499e-01
6.26983762e-01 5.96914470e-01 4.58890110e-01 -3.11800867e-01
6.89527333e-01 2.98113793e-01 -8.91841277e-02 -9.50854719e-01
-8.44396830e-01 -3.66477013e-01 -1.29524827e+00 -4.43471819e-02
7.34211504e-01 -1.40154016e+00 -3.03623319e-01 1.12968946e+00
-1.18375802e+00 -6.67944968e-01 2.65544094e-02 2.84847438e-01
-2.23655432e-01 -6.57830155e-03 -2.89280564e-01 -7.07749069e-01
-3.88634682e-01 -7.95848310e-01 1.08682549e+00 1.68811947e-01
2.09936202e-01 -8.00236762e-01 -9.43304747e-02 -3.11667293e-01
6.50706589e-01 5.42332232e-01 8.85219514e-01 -4.58910801e-02
-5.52891970e-01 -4.39788818e-01 -5.91172636e-01 5.34654735e-03
4.93866891e-01 2.78291762e-01 -9.94005561e-01 -2.15853363e-01
-7.63700783e-01 -1.93956792e-01 1.17955410e+00 9.12478149e-01
1.18254244e+00 2.80742347e-01 -3.88869792e-01 9.02377188e-01
1.46693909e+00 8.11126158e-02 4.49035466e-01 6.09113872e-01
7.03515887e-01 5.90108097e-01 1.26927567e+00 3.19453925e-01
4.64880347e-01 4.00528967e-01 8.73840868e-01 -6.92468882e-01
3.69075656e-01 -1.39819130e-01 -1.66961309e-02 6.52083457e-01
-2.56097287e-01 -3.05885285e-01 -1.22370613e+00 1.06239283e+00
-2.02295470e+00 -7.63231933e-01 -3.48363891e-02 2.01818180e+00
7.81794429e-01 -5.43074757e-02 -1.45749658e-01 -2.98274577e-01
3.81365210e-01 6.68103158e-01 -6.48122251e-01 -1.46551520e-01
-7.95453906e-01 2.35332489e-01 1.56727469e+00 3.86443704e-01
-2.04689860e+00 1.76150954e+00 7.65164900e+00 6.43539429e-01
-1.05976748e+00 -2.79333889e-01 6.45812094e-01 3.16359580e-01
-1.31231174e-01 -1.67774353e-02 -9.17254627e-01 7.06478730e-02
7.29068756e-01 1.80588946e-01 4.48781490e-01 6.43594205e-01
2.17174202e-01 -3.70434910e-01 -5.10259926e-01 3.31364721e-01
-4.06484544e-01 -1.31715059e+00 -9.32418332e-02 1.36901870e-01
9.74963367e-01 5.34493864e-01 -2.68116176e-01 3.66757691e-01
7.87208438e-01 -1.10233998e+00 6.00434124e-01 4.30263191e-01
8.12710762e-01 -6.96950853e-01 9.77993131e-01 8.66169110e-02
-1.69063389e+00 -2.16574389e-02 -5.37319839e-01 -2.06579298e-01
9.65912640e-03 7.47436464e-01 -1.76709145e-01 9.31318700e-01
9.80483472e-01 1.24895930e+00 -5.82724035e-01 7.55585432e-01
-2.11063400e-02 8.21379066e-01 -5.32603502e-01 3.04550856e-01
4.74661380e-01 -2.20728070e-01 3.00025135e-01 1.21600473e+00
5.33368170e-01 1.25368670e-01 4.15480316e-01 5.62588096e-01
-8.04654509e-02 -1.54249668e-02 -8.05428922e-01 -5.24304152e-01
5.04699469e-01 1.30428362e+00 -5.75375855e-01 -7.65679181e-02
-2.53632724e-01 7.18032300e-01 2.43660301e-01 3.14148247e-01
-6.15341187e-01 -4.21526223e-01 1.21801627e+00 -2.68996358e-01
4.20721412e-01 -6.31155014e-01 -3.69765848e-01 -1.10844088e+00
-5.05167246e-02 -7.73606062e-01 1.42758433e-02 -6.31120503e-01
-1.03798366e+00 2.18728393e-01 4.34817653e-03 -9.96418536e-01
-1.44989327e-01 -9.38894570e-01 -2.34281003e-01 1.10230672e+00
-2.20675445e+00 -1.76229501e+00 -5.10890782e-01 -3.46342742e-01
4.80461776e-01 -4.64268774e-01 1.42026699e+00 1.18602753e-01
-4.36434597e-01 1.52592853e-01 4.32173342e-01 1.13149904e-01
5.49994409e-01 -1.27732730e+00 1.00174463e+00 8.64597201e-01
-2.77654260e-01 8.13306421e-02 4.97616827e-01 -5.77622294e-01
-1.34668422e+00 -1.40867388e+00 9.59381104e-01 -1.42247245e-01
9.35306311e-01 -4.07107234e-01 -8.97372067e-01 8.40884030e-01
1.23757109e-01 6.96187317e-02 2.46644020e-01 3.18324417e-01
-6.27172142e-02 -5.37357688e-01 -8.73071611e-01 3.05791825e-01
8.63330185e-01 -2.66213596e-01 3.52342940e-05 3.28925073e-01
9.24691319e-01 -3.80236417e-01 -1.24849975e+00 8.40062737e-01
1.09633279e+00 -2.63294458e-01 9.32000339e-01 -5.86621702e-01
4.61313188e-01 -1.61809132e-01 -5.16596317e-01 -1.78455162e+00
-5.54619908e-01 -1.09885760e-01 1.28452882e-01 1.06985819e+00
3.46495599e-01 -5.59561789e-01 7.56980777e-01 -1.80906743e-01
-1.36621132e-01 -4.87640440e-01 -5.64600825e-01 -5.21622181e-01
4.84638393e-01 -1.01665966e-01 1.10811532e+00 1.23058701e+00
-3.76143456e-01 -4.61325705e-01 -4.37417030e-01 5.88070035e-01
3.62861514e-01 2.20658839e-01 7.53441632e-01 -1.58967173e+00
2.69404322e-01 -6.01153314e-01 -2.69188285e-01 -7.94675708e-01
3.54520917e-01 -6.33624256e-01 2.23114818e-01 -1.44393134e+00
1.29775805e-02 -7.53303349e-01 -4.88067329e-01 9.14668798e-01
-4.87142235e-01 8.76265615e-02 -3.50214802e-02 -1.06656522e-01
1.67408511e-01 6.28549933e-01 6.81748331e-01 -3.49888355e-01
-2.92092443e-01 -1.19677141e-01 -4.87124771e-01 4.80330408e-01
1.07285964e+00 -3.71695966e-01 1.56239793e-01 -9.91851449e-01
2.36177579e-01 -2.47538731e-01 2.06194058e-01 -7.28451312e-01
-5.86229324e-01 -6.08057857e-01 4.42143947e-01 -1.06155217e+00
-8.33546668e-02 -6.56311214e-01 -1.17366448e-01 4.46261406e-01
-1.33313164e-01 1.85814708e-01 9.66180265e-01 2.39083707e-01
-2.98971236e-01 2.56394833e-01 4.64403361e-01 -1.56628922e-01
-1.03989828e+00 5.92085779e-01 -7.87602186e-01 -5.67140937e-01
7.05528080e-01 2.94625372e-01 -3.35751265e-01 8.79867822e-02
-5.54495990e-01 5.75802982e-01 5.67603171e-01 8.69012296e-01
-2.49982551e-02 -1.13012445e+00 -1.29204786e+00 3.12246263e-01
4.17097360e-01 -1.33921234e-02 -3.15863034e-03 3.41814548e-01
-1.01730788e+00 8.10183227e-01 -5.84038317e-01 -6.84889495e-01
-9.42022920e-01 1.61341339e-01 4.66042638e-01 -4.99075830e-01
-2.82445133e-01 6.17553592e-01 -8.06912705e-02 -1.07040548e+00
-1.06526680e-01 -4.90065724e-01 -4.17196721e-01 5.66433370e-01
8.62533003e-02 -1.48160085e-01 1.87659606e-01 -6.87680364e-01
-2.13482156e-01 3.45994502e-01 2.75681406e-01 1.01178661e-01
1.82507980e+00 1.44606024e-01 -4.37175304e-01 7.79129684e-01
6.34260952e-01 -4.59785521e-01 -1.28353345e+00 -3.32949050e-02
3.01684767e-01 -4.08098221e-01 4.40924555e-01 -9.53962207e-01
-1.16347969e+00 7.44873345e-01 1.02558434e+00 1.06091373e-01
1.14750290e+00 -4.41820234e-01 5.75135648e-01 5.43003738e-01
5.49706578e-01 -8.06392729e-01 -8.21283221e-01 6.76206470e-01
8.34453344e-01 -1.82300746e+00 2.48656839e-01 -4.03932244e-01
-2.14296445e-01 9.50630367e-01 4.01490688e-01 -2.06898302e-01
8.78848791e-01 5.68965137e-01 2.07342789e-01 1.02992706e-01
-4.36167806e-01 -6.59501135e-01 5.98584749e-02 7.69360960e-01
8.89779329e-01 8.86156678e-01 -9.60877091e-02 4.49745119e-01
-4.45231535e-02 3.26952428e-01 1.23803720e-01 1.08757985e+00
-5.90357542e-01 -1.08653820e+00 -2.34866813e-01 6.75947070e-01
-2.90961832e-01 -3.44534039e-01 -9.65359062e-02 8.60056579e-01
-6.40725121e-02 5.91236889e-01 2.16077536e-01 -1.15740329e-01
3.24834555e-01 -1.17483549e-01 4.55257803e-01 -4.11447167e-01
-5.89667976e-01 -2.05027461e-02 -4.98642505e-04 -3.40525776e-01
-5.37614286e-01 -8.75180662e-01 -7.09305823e-01 -5.55381954e-01
-3.07606250e-01 -3.28010440e-01 8.24664176e-01 7.51957476e-01
3.02352875e-01 4.89874214e-01 5.70160449e-01 -1.28764105e+00
-2.20799133e-01 -1.20372570e+00 -8.76190543e-01 -2.88373441e-01
4.66884226e-01 -6.32806540e-01 1.21885456e-01 -6.16452210e-02] | [9.403570175170898, -1.5687239170074463] |
6d758273-4ddd-4c98-afaf-ee53d3f322c3 | towards-language-free-training-for-text-to | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Zhou_Towards_Language-Free_Training_for_Text-to-Image_Generation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Zhou_Towards_Language-Free_Training_for_Text-to-Image_Generation_CVPR_2022_paper.pdf | Towards Language-Free Training for Text-to-Image Generation | One of the major challenges in training text-to-image generation models is the need of a large number of high-quality text-image pairs. While image samples are often easily accessible, the associated text description typically requires careful human captioning, which is particularly time- and cost-consuming. In this paper, we propose the first work to train text-to-image generation models without any text data. It intelligently leverages the well-aligned cross-modal semantic space of the powerful pre-trained CLIP model: the requirement of text-conditioning is alleviated via generating text features from image features. Extensive experiments are conducted to illustrate the effectiveness of the proposed method. We obtain state-of-the-art results in the standard text-to-image generation tasks. Importantly, the proposed language-free model outperforms most existing models trained with full text-image pairs. Furthermore, our method can be applied in fine-tuning pre-trained models, which saves both training time and cost in training text-to-image generation models. Our pre-trained model obtains competitive results in zero-shot text-to-image generation on MS-COCO dataset, yet with around only 1% of the model size compared to the recently proposed large DALL-E model. | ['Tong Sun', 'Jinhui Xu', 'Jiuxiang Gu', 'Tong Yu', 'Chris Tensmeyer', 'Chunyuan Li', 'Changyou Chen', 'Ruiyi Zhang', 'Yufan Zhou'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['zero-shot-text-to-image-generation'] | ['natural-language-processing'] | [ 7.62479126e-01 3.19416076e-02 -8.45572874e-02 -2.58818865e-01
-1.41491771e+00 -2.52171338e-01 9.38878477e-01 -2.91443288e-01
-4.09391016e-01 6.53602123e-01 3.72084975e-02 -1.07319467e-01
3.75653625e-01 -7.31613934e-01 -8.88569355e-01 -6.67550385e-01
6.49545848e-01 5.43347776e-01 -7.96560869e-02 -2.57363617e-01
5.27641624e-02 -1.09583147e-01 -1.59920096e+00 3.53159726e-01
1.21842289e+00 1.15337014e+00 6.95289969e-01 6.80843771e-01
-2.10744530e-01 7.12463915e-01 -4.57510412e-01 -6.48029268e-01
2.31402084e-01 -8.31355453e-01 -5.15057802e-01 4.43550974e-01
6.58888876e-01 -5.56865990e-01 -3.12468588e-01 9.35545206e-01
7.30398834e-01 -1.94758456e-02 8.06152165e-01 -1.32828975e+00
-1.01773715e+00 4.65265065e-01 -6.58409655e-01 -2.57404745e-01
2.11158246e-01 2.60543644e-01 9.35094655e-01 -1.25522077e+00
8.23468804e-01 1.06219935e+00 2.78667152e-01 6.72731876e-01
-1.14502621e+00 -8.29348564e-01 -1.95220113e-01 1.97934546e-02
-1.37631834e+00 -7.01382697e-01 7.11580932e-01 -4.33245748e-01
7.53016293e-01 1.55871548e-02 4.51851428e-01 1.35705996e+00
-3.28021646e-02 1.03860593e+00 9.99509573e-01 -7.14590967e-01
-4.49102111e-02 2.59115130e-01 -6.03317976e-01 6.85253501e-01
-1.44688748e-02 -5.03968894e-02 -6.01507485e-01 9.53108519e-02
7.62650728e-01 -3.14715016e-03 -2.14882061e-01 -1.52369902e-01
-1.39281547e+00 9.19197381e-01 2.90087432e-01 2.21605122e-01
-1.09608524e-01 2.73411095e-01 3.90301466e-01 8.06897134e-02
6.62204802e-01 3.87520760e-01 -1.95989553e-02 -1.59727946e-01
-1.38063037e+00 2.91620195e-01 3.05653214e-01 1.39924157e+00
6.65457368e-01 2.16128096e-01 -6.20971859e-01 1.03953099e+00
5.91200665e-02 9.07098949e-01 6.64583623e-01 -5.91094613e-01
9.17013049e-01 3.08807641e-01 1.10232130e-01 -6.89186215e-01
2.07692876e-01 -2.76757568e-01 -1.13710785e+00 -2.32960969e-01
1.18802376e-01 -1.93332016e-01 -1.22651684e+00 1.58048439e+00
1.37421876e-01 1.72472268e-01 1.29704386e-01 6.93845510e-01
7.58929133e-01 9.78720963e-01 3.08851395e-02 -1.68150350e-01
1.32521987e+00 -1.49788010e+00 -7.64500856e-01 -4.05800849e-01
5.22621155e-01 -1.14899671e+00 1.25654685e+00 -8.27923119e-02
-1.18492544e+00 -6.92017257e-01 -8.63971472e-01 -2.98912048e-01
-2.87209064e-01 5.98778009e-01 4.18746114e-01 4.36436057e-01
-9.84604776e-01 2.78189242e-01 -3.36881310e-01 -3.89709413e-01
5.56882918e-01 -4.46134061e-02 -3.58274549e-01 -4.48599190e-01
-1.15146613e+00 5.82414269e-01 4.00676429e-01 3.75007503e-02
-1.01189256e+00 -6.56421483e-01 -1.03960431e+00 -2.43288986e-02
3.54187965e-01 -9.23875034e-01 1.30623698e+00 -1.03082609e+00
-1.51135039e+00 8.39343429e-01 -4.14326370e-01 -2.44497508e-01
8.18978548e-01 -1.07692972e-01 -2.34773129e-01 3.42513472e-01
4.86154854e-01 1.25498116e+00 1.31330037e+00 -1.28845799e+00
-5.50094306e-01 8.19838867e-02 -1.63124815e-01 2.84383744e-01
-6.14054322e-01 -9.55517888e-02 -9.27202046e-01 -1.04052973e+00
-3.57244015e-01 -9.95735168e-01 -2.79879600e-01 9.41119343e-02
-4.71337229e-01 -4.41241227e-02 5.35601377e-01 -5.22437751e-01
1.04278588e+00 -1.97777474e+00 -1.74800903e-01 -2.76889086e-01
-4.12755013e-02 2.64782578e-01 -5.71744025e-01 6.16954267e-01
-4.46572118e-02 1.92778602e-01 -3.17925960e-01 -7.51349151e-01
1.50824651e-01 -1.05265073e-01 -6.55344486e-01 5.57204448e-02
4.47728038e-01 1.30980313e+00 -8.71053576e-01 -9.56497431e-01
3.94748539e-01 4.46146160e-01 -4.65417981e-01 4.14099157e-01
-3.44559401e-01 3.65231603e-01 -4.34442252e-01 4.38065290e-01
5.91167927e-01 -5.48715293e-01 -2.19507098e-01 -1.86765984e-01
2.13557571e-01 9.87070575e-02 -7.28195310e-01 2.14875460e+00
-7.66088247e-01 6.58143997e-01 -3.66931468e-01 -9.28285122e-01
6.80385709e-01 4.58825469e-01 3.03907692e-01 -8.97951782e-01
1.97280675e-01 2.08592206e-01 -5.02144694e-01 -4.10422623e-01
7.25149870e-01 -2.54693508e-01 -1.69140324e-01 5.89769363e-01
2.90200651e-01 -5.31740308e-01 5.16962290e-01 3.50813925e-01
5.45800686e-01 3.23091239e-01 9.85907763e-02 1.05180234e-01
4.18444484e-01 -3.23119871e-02 1.31846547e-01 7.85769522e-01
1.26162991e-01 9.79829073e-01 1.37941599e-01 -6.55109109e-03
-1.38265860e+00 -8.85275781e-01 -1.71228781e-01 9.82410550e-01
7.33223408e-02 -4.29222316e-01 -9.87238050e-01 -6.64677680e-01
-4.38673645e-01 7.52039969e-01 -6.04325652e-01 1.33633260e-02
-2.58085161e-01 -6.95768654e-01 6.06540501e-01 3.46728116e-01
5.66191971e-01 -1.01147497e+00 -1.09806143e-01 1.12709403e-01
-5.83461463e-01 -1.43610048e+00 -1.02113962e+00 -2.66611755e-01
-8.34720373e-01 -5.38805068e-01 -1.32635629e+00 -1.01385736e+00
1.13474321e+00 4.91240263e-01 1.12427652e+00 7.34702870e-02
-4.76995140e-01 1.73520461e-01 -5.71744680e-01 -3.44360948e-01
-5.36636889e-01 1.47505566e-01 -3.89781594e-01 2.19629034e-01
-1.20850079e-01 -3.30450356e-01 -6.98928595e-01 2.33272091e-01
-1.25803196e+00 7.89362371e-01 9.29587424e-01 1.26349211e+00
7.04903066e-01 -5.17968424e-02 7.28114903e-01 -8.51338923e-01
4.98098254e-01 -2.97821283e-01 -5.22494614e-01 4.44509864e-01
-6.60513043e-01 1.36772782e-01 8.47696006e-01 -5.13717234e-01
-1.22529638e+00 2.77975142e-01 -4.60331999e-02 -5.27010083e-01
-1.02437509e-03 3.52495700e-01 -5.26846722e-02 1.09429993e-01
4.62634653e-01 6.85707211e-01 -2.74421304e-01 -1.65662318e-01
7.71592081e-01 7.96183705e-01 5.57493567e-01 -5.88442326e-01
1.14532340e+00 3.84635180e-01 -2.01522738e-01 -6.59339964e-01
-9.79730070e-01 -3.42206270e-01 -4.75814462e-01 -1.48458928e-01
8.00280452e-01 -1.24979091e+00 -1.04754148e-02 5.99591732e-01
-1.16032243e+00 -3.43619496e-01 -1.92567810e-01 1.27231076e-01
-7.22575247e-01 4.62873667e-01 -5.85485101e-01 -6.08198345e-01
-8.14414918e-01 -1.17684162e+00 1.67889178e+00 5.63279353e-02
2.27750674e-01 -8.79654944e-01 -9.22299102e-02 7.75116622e-01
4.62473482e-01 -1.95907839e-02 6.50347769e-01 -2.48773664e-01
-7.69927263e-01 -4.05256599e-01 -5.83267510e-01 2.89160579e-01
7.79084191e-02 -2.77665466e-01 -9.62412059e-01 -3.12136889e-01
-2.23597169e-01 -8.96159172e-01 8.82326901e-01 2.43944839e-01
1.30557835e+00 -1.72171757e-01 -2.08101690e-01 4.81547952e-01
1.40943873e+00 -4.97914962e-02 7.48736203e-01 -1.28314523e-02
8.03368628e-01 4.27309632e-01 8.97217035e-01 4.65487033e-01
4.10328686e-01 7.57107139e-01 2.53092293e-02 -5.15911877e-01
-3.62454981e-01 -7.19121337e-01 3.46294999e-01 9.51752245e-01
3.08262825e-01 -5.83588958e-01 -5.10787904e-01 7.43559837e-01
-1.85406780e+00 -9.27556336e-01 3.49555202e-02 2.00314903e+00
1.26261163e+00 -1.15295954e-01 -2.62816697e-01 -1.99958608e-01
7.14697421e-01 2.34741598e-01 -3.82746279e-01 1.81706876e-01
-8.33614469e-02 3.28633070e-01 3.41434389e-01 4.15435612e-01
-1.09803474e+00 1.17815900e+00 6.17509604e+00 1.57500136e+00
-1.11760223e+00 2.13364422e-01 8.99424613e-01 -3.07558328e-01
-2.85744727e-01 -9.93070304e-02 -7.31650233e-01 5.06235838e-01
8.43974650e-01 -3.65983486e-01 3.96370023e-01 9.39119339e-01
2.77652740e-01 -7.62233064e-02 -1.07584214e+00 1.37063920e+00
5.34674942e-01 -1.46211815e+00 5.07100403e-01 -7.63364136e-02
1.18309247e+00 -1.54358983e-01 1.29037738e-01 3.63677293e-01
1.22204654e-01 -1.11828852e+00 9.20274913e-01 2.12039053e-01
1.41266334e+00 -5.75064003e-01 5.54172695e-01 2.06108093e-01
-1.22376657e+00 2.64664114e-01 -4.53991055e-01 3.70162815e-01
4.09103900e-01 8.16859484e-01 -8.07849467e-01 7.11112142e-01
2.95424521e-01 6.92536831e-01 -6.59439385e-01 7.48484731e-01
-1.61599502e-01 4.55697119e-01 -1.11001194e-01 3.23944762e-02
2.90284663e-01 -6.58643693e-02 1.52814880e-01 1.17546785e+00
7.46519923e-01 -1.60834298e-01 2.69157559e-01 1.13371921e+00
-4.34734613e-01 2.60012090e-01 -7.13845134e-01 -3.19769025e-01
2.30038449e-01 1.45716393e+00 -6.24558389e-01 -7.01915264e-01
-4.39627439e-01 1.46900904e+00 1.27643585e-01 3.88659984e-01
-9.63986874e-01 -5.96626163e-01 9.30283442e-02 5.58770634e-02
4.11744118e-01 9.37367454e-02 -1.45575330e-01 -1.35341823e+00
2.14364067e-01 -8.51093411e-01 -2.57226285e-02 -1.21879065e+00
-1.43719268e+00 7.24593282e-01 -1.22359835e-01 -1.47768772e+00
-5.89429557e-01 -3.40549082e-01 -4.47461396e-01 9.93266940e-01
-1.71101296e+00 -1.66695905e+00 -4.84933645e-01 7.17469454e-01
9.61776793e-01 -1.97219297e-01 7.16882527e-01 4.17977601e-01
-4.46226567e-01 9.26129639e-01 2.14974090e-01 2.24734768e-01
9.99760330e-01 -9.12425160e-01 7.24069476e-01 1.00355792e+00
2.85998672e-01 3.52495104e-01 4.61046934e-01 -6.25680566e-01
-1.36838841e+00 -1.43072760e+00 9.23918068e-01 -3.38980943e-01
5.68746448e-01 -6.43288612e-01 -4.26623434e-01 4.45765615e-01
4.82321769e-01 -3.11059766e-02 5.42731345e-01 -4.65687007e-01
-3.43536198e-01 -7.52495900e-02 -7.22054780e-01 9.05797780e-01
9.68433142e-01 -6.38327479e-01 -2.98524797e-01 5.26973307e-01
8.93962443e-01 -3.96343499e-01 -6.61008835e-01 1.62218988e-01
3.53921354e-01 -6.77250266e-01 9.23143327e-01 -1.91152483e-01
9.79342163e-01 -1.97959095e-01 5.69874905e-02 -1.19578075e+00
-7.80250644e-03 -8.44525456e-01 2.71691710e-01 1.44282568e+00
5.80627263e-01 -2.15623006e-01 6.83424056e-01 4.15614069e-01
-8.62082466e-02 -6.74162269e-01 -7.80830622e-01 -7.74330139e-01
1.46909971e-02 -4.77018893e-01 4.48565543e-01 8.13299954e-01
-1.60850450e-01 6.98921144e-01 -8.52346957e-01 -4.14244324e-01
6.40940666e-01 2.26549238e-01 1.02591062e+00 -6.39997780e-01
-2.89856791e-01 -3.23537171e-01 4.98778224e-02 -1.35344148e+00
1.76221848e-01 -1.04785728e+00 3.83081973e-01 -1.56543779e+00
6.22597933e-01 -4.22955930e-01 1.19257897e-01 4.12730604e-01
-6.18544102e-01 5.04205287e-01 3.62271428e-01 1.64876938e-01
-7.18697906e-01 1.05939174e+00 1.69732463e+00 -2.37529770e-01
1.76341310e-01 -4.16669399e-01 -7.43257582e-01 2.82093734e-01
5.06142616e-01 -5.72893620e-01 -7.14589715e-01 -5.81190526e-01
1.01032302e-01 1.60282999e-01 3.03583443e-01 -7.95496166e-01
1.71652362e-01 -2.60554403e-01 3.70626271e-01 -6.60666823e-01
4.83922362e-01 -6.50683522e-01 7.14405719e-03 1.14960805e-01
-4.87767637e-01 -1.46335080e-01 1.05129875e-01 5.82123280e-01
-3.52069199e-01 -4.09777939e-01 7.62636840e-01 -6.08147755e-02
-3.40879112e-01 5.94663620e-01 -1.56372473e-01 2.28121400e-01
9.73265111e-01 -9.24556851e-02 -3.50567788e-01 -6.91095293e-01
-2.54773721e-02 2.04348445e-01 5.22221029e-01 6.70190930e-01
7.16909349e-01 -1.63138092e+00 -8.65204990e-01 2.06287965e-01
5.26510298e-01 8.86964425e-02 4.86066937e-01 6.79123998e-01
-2.71091431e-01 6.42108560e-01 7.88616464e-02 -6.16432905e-01
-1.09035456e+00 5.68406045e-01 -2.22576875e-02 -5.14514446e-01
-5.88286459e-01 6.59741044e-01 5.75962245e-01 -3.09193969e-01
-1.84584707e-02 -6.32189661e-02 2.46532783e-01 -1.21678740e-01
7.38794565e-01 -4.36344296e-02 -1.86365589e-01 -6.05693102e-01
1.77707449e-01 7.12122500e-01 -2.44024277e-01 -5.54065347e-01
1.12070346e+00 -2.13311404e-01 5.13644218e-02 1.69517890e-01
1.18342948e+00 -2.78191060e-01 -1.13673782e+00 -2.36896813e-01
-4.41652030e-01 -5.74836731e-01 1.05791144e-01 -7.37826526e-01
-1.14638519e+00 1.16673553e+00 4.15519983e-01 -3.30150425e-01
1.31388891e+00 -1.16966531e-01 1.25356531e+00 3.91237646e-01
3.23516160e-01 -1.16457999e+00 5.20588994e-01 2.49438196e-01
9.83853400e-01 -1.51861370e+00 -1.27046093e-01 -4.04939532e-01
-8.66266549e-01 7.97151089e-01 5.58881700e-01 1.19706742e-01
2.40512371e-01 -1.72669515e-02 1.89680159e-02 1.03752092e-01
-8.37123156e-01 -2.18104154e-01 4.68796968e-01 6.45687759e-01
4.50044751e-01 -1.02588139e-01 -9.60538462e-02 5.06576359e-01
-1.97666675e-01 2.57733792e-01 4.53835458e-01 7.05772400e-01
-1.07515737e-01 -1.31250238e+00 -2.31784463e-01 4.82870638e-01
-3.88244212e-01 -6.75767601e-01 -2.32927948e-01 5.55847645e-01
-1.84694827e-01 1.01085842e+00 -8.13206956e-02 -1.83037043e-01
-7.17488071e-03 7.07839429e-02 4.37466502e-01 -6.45996392e-01
-1.90340832e-01 2.85099506e-01 2.31665117e-03 -3.86355072e-01
-4.18599993e-01 -3.45053881e-01 -9.45011139e-01 -2.03432783e-01
-6.40262067e-01 -6.47689924e-02 6.58501327e-01 8.74913871e-01
5.68495274e-01 3.87718141e-01 7.40100503e-01 -1.06750345e+00
-4.31859314e-01 -1.17096710e+00 -3.83545697e-01 7.36250699e-01
6.83471560e-02 -4.15810674e-01 -1.20011255e-01 6.19945347e-01] | [11.215081214904785, 0.5328801870346069] |
cb67536d-2158-4fb5-9ec6-f6e7ab655839 | mrcn-a-novel-modality-restitution-and | 2303.14626 | null | https://arxiv.org/abs/2303.14626v1 | https://arxiv.org/pdf/2303.14626v1.pdf | MRCN: A Novel Modality Restitution and Compensation Network for Visible-Infrared Person Re-identification | Visible-infrared person re-identification (VI-ReID), which aims to search identities across different spectra, is a challenging task due to large cross-modality discrepancy between visible and infrared images. The key to reduce the discrepancy is to filter out identity-irrelevant interference and effectively learn modality-invariant person representations. In this paper, we propose a novel Modality Restitution and Compensation Network (MRCN) to narrow the gap between the two modalities. Specifically, we first reduce the modality discrepancy by using two Instance Normalization (IN) layers. Next, to reduce the influence of IN layers on removing discriminative information and to reduce modality differences, we propose a Modality Restitution Module (MRM) and a Modality Compensation Module (MCM) to respectively distill modality-irrelevant and modality-relevant features from the removed information. Then, the modality-irrelevant features are used to restitute to the normalized visible and infrared features, while the modality-relevant features are used to compensate for the features of the other modality. Furthermore, to better disentangle the modality-relevant features and the modality-irrelevant features, we propose a novel Center-Quadruplet Causal (CQC) loss to encourage the network to effectively learn the modality-relevant features and the modality-irrelevant features. Extensive experiments are conducted to validate the superiority of our method on the challenging SYSU-MM01 and RegDB datasets. More remarkably, our method achieves 95.1% in terms of Rank-1 and 89.2% in terms of mAP on the RegDB dataset. | ['Hanzi Wang', 'Jie Li', 'Yan Yan', 'Yukang Zhang'] | 2023-03-26 | null | null | null | null | ['person-re-identification'] | ['computer-vision'] | [ 3.22573692e-01 -4.08902735e-01 2.01834813e-02 -5.32790720e-01
-5.50901294e-01 -4.21970069e-01 6.68335319e-01 -2.52407968e-01
-6.44702733e-01 5.21196187e-01 3.85025114e-01 8.74670222e-02
-3.34034175e-01 -5.88919222e-01 -5.61206043e-01 -8.50777447e-01
4.05379742e-01 -1.42994538e-01 -3.14282894e-01 -3.42946678e-01
-5.32191098e-02 4.51947331e-01 -1.62095428e+00 2.29431421e-01
1.07789433e+00 1.00182569e+00 -1.03889242e-01 -5.16793653e-02
1.01469785e-01 5.97999752e-01 -2.47994691e-01 -4.19167250e-01
5.38167417e-01 -4.91557747e-01 -4.56644803e-01 -2.33178437e-01
5.40275574e-01 -1.83007672e-01 -7.56771386e-01 1.12521327e+00
7.38958240e-01 4.40787882e-01 6.37838185e-01 -1.39640629e+00
-7.83328176e-01 1.90037966e-01 -9.17082548e-01 1.38912052e-01
5.31487763e-01 1.38599575e-01 5.65145493e-01 -8.47548544e-01
3.90753061e-01 1.43380249e+00 4.72449660e-01 7.29723215e-01
-1.30835092e+00 -1.10116982e+00 3.51127744e-01 4.83733147e-01
-1.60944605e+00 -5.12484908e-01 9.84268367e-01 -2.98823357e-01
3.40585917e-01 5.99387944e-01 3.23192447e-01 1.14953494e+00
-3.86091918e-01 6.25335515e-01 1.02617967e+00 -4.49499369e-01
-3.22082847e-01 3.60775262e-01 1.29775032e-01 4.34686601e-01
1.46015361e-01 1.93420187e-01 -6.03535831e-01 4.15382646e-02
5.30302465e-01 3.44003975e-01 -3.08325946e-01 -2.08052799e-01
-1.09053075e+00 4.24329221e-01 7.39550292e-01 9.87541378e-02
-2.77540445e-01 -3.99563372e-01 2.59332269e-01 3.11192900e-01
3.06500286e-01 1.21344686e-01 -1.85327828e-01 3.76658469e-01
-4.73642588e-01 5.41151129e-02 1.62064105e-01 7.28091002e-01
7.59244978e-01 -2.87826151e-01 -6.56806767e-01 1.20590568e+00
2.68816620e-01 5.88705003e-01 2.73981303e-01 -6.43201470e-01
6.82291269e-01 8.75253022e-01 1.40079791e-02 -9.21505630e-01
-4.80561584e-01 -4.24576402e-01 -1.22914183e+00 -1.21224917e-01
3.74429077e-01 -6.77887201e-02 -8.72936130e-01 2.08806705e+00
3.45814347e-01 7.43246526e-02 -1.05212770e-01 1.31301200e+00
1.17466784e+00 2.87993759e-01 4.06408638e-01 -2.39240646e-01
1.55271697e+00 -6.76730216e-01 -3.61360729e-01 -2.13436961e-01
2.79851228e-01 -7.06778049e-01 9.18196678e-01 -2.38665894e-01
-6.76526368e-01 -7.67251253e-01 -8.43949735e-01 -2.45025441e-01
-3.78815025e-01 2.48867080e-01 4.24996495e-01 3.98726195e-01
-5.89858592e-01 2.56697893e-01 -2.32730761e-01 -3.15399647e-01
4.24832523e-01 5.48108637e-01 -6.75294638e-01 -3.98572654e-01
-1.50011051e+00 5.51169872e-01 4.18964088e-01 3.46736968e-01
-1.63467109e-01 -8.32138956e-01 -8.36252451e-01 -1.61603168e-02
2.67352581e-01 -7.35878408e-01 3.70503366e-01 -1.03295267e+00
-1.18708861e+00 9.17044520e-01 -2.09248126e-01 1.27995268e-01
5.42857707e-01 7.80390203e-02 -7.56119370e-01 5.71048073e-02
1.41496435e-01 6.70266867e-01 7.04118431e-01 -1.21519005e+00
-6.48392677e-01 -5.47176838e-01 4.85381819e-02 6.20394409e-01
-5.96115768e-01 3.44591290e-02 -6.79718673e-01 -7.27608979e-01
3.26562494e-01 -9.50690031e-01 3.30725640e-01 1.14485426e-02
-5.66187561e-01 -3.05070579e-01 3.82774353e-01 -9.02543664e-01
7.46021271e-01 -2.45769119e+00 2.08353192e-01 3.98545563e-01
1.53962746e-01 6.67260885e-02 -6.04847312e-01 6.89940304e-02
-4.62385923e-01 -1.26437128e-01 7.39355534e-02 -4.21507895e-01
-9.89224985e-02 -1.17770486e-01 -1.00571282e-01 5.87756991e-01
1.20022997e-01 8.13892961e-01 -6.67046547e-01 -3.05848479e-01
3.82378638e-01 7.07285047e-01 -1.91044644e-01 -4.40384075e-02
3.82674605e-01 7.90373802e-01 -3.08934242e-01 6.84649527e-01
1.09530437e+00 9.49589983e-02 -7.76311532e-02 -6.71265543e-01
-3.47918458e-02 -2.06696615e-02 -1.15104675e+00 1.58935523e+00
-2.78088689e-01 2.74262071e-01 1.47942395e-03 -9.06147122e-01
7.34397590e-01 4.98172976e-02 6.38655961e-01 -1.38713920e+00
1.93678483e-01 -9.01072547e-02 -1.80682927e-01 -3.80150706e-01
3.12802255e-01 -3.87409255e-02 -7.38136470e-02 2.24785924e-01
-9.15161595e-02 8.76934826e-01 -6.41713962e-02 8.65583420e-02
5.55349290e-01 1.42646506e-02 -2.54796207e-01 -5.72749898e-02
1.00168753e+00 -4.48935837e-01 7.71404386e-01 5.82891345e-01
-3.03861141e-01 7.12610602e-01 1.55660123e-01 -3.59725893e-01
-5.37371337e-01 -1.09466100e+00 -6.37820140e-02 1.31694400e+00
7.91495860e-01 -5.28775975e-02 -4.03794974e-01 -6.24553323e-01
1.61076680e-01 4.84886020e-01 -7.49313772e-01 -5.93022168e-01
-3.89019996e-01 -9.86053050e-01 3.64905566e-01 4.27132785e-01
9.38269377e-01 -7.10195899e-01 2.71335661e-01 -3.69408727e-01
-6.28564596e-01 -9.80763972e-01 -7.00206757e-01 -4.59782809e-01
-3.10296804e-01 -1.02999163e+00 -7.76326299e-01 -6.66836381e-01
9.17295992e-01 7.55563974e-01 4.83502746e-01 -5.01898192e-02
-2.44627565e-01 2.98704207e-01 -1.56749189e-01 -1.62190035e-01
2.36789316e-01 7.25679845e-02 1.56303406e-01 5.34194529e-01
5.86057603e-01 -3.51759166e-01 -8.46377373e-01 4.02581722e-01
-7.71066606e-01 6.60208017e-02 5.68028808e-01 9.45959747e-01
5.10845602e-01 4.88470774e-03 4.43394631e-01 -4.86007392e-01
5.03942728e-01 -3.74672264e-01 -3.60202491e-01 4.44967955e-01
-4.99874502e-01 -8.46574679e-02 5.73635995e-01 -6.59164846e-01
-1.23687339e+00 1.77611023e-01 2.06663296e-01 -4.11363542e-01
-7.00977147e-02 2.41786227e-01 -5.65951109e-01 -1.91400632e-01
5.37092328e-01 3.00467879e-01 -9.98260453e-02 -4.84620810e-01
2.52386719e-01 5.49076617e-01 7.66922474e-01 -3.91556650e-01
1.14310551e+00 5.76509833e-01 3.47822085e-02 -5.05519390e-01
-7.77372360e-01 -6.12911105e-01 -4.06107247e-01 -1.46146253e-01
8.00566196e-01 -1.38586509e+00 -1.06120813e+00 5.32877386e-01
-8.27035964e-01 4.17890787e-01 -2.42667764e-01 6.25485241e-01
6.94917813e-02 4.41954970e-01 -4.53080237e-01 -7.39325702e-01
-4.02369529e-01 -8.62090111e-01 8.99658799e-01 9.96591032e-01
2.10151911e-01 -5.16543269e-01 -1.80193365e-01 7.02648461e-01
3.18217009e-01 6.91635534e-02 7.16533780e-01 -4.79882807e-01
-2.60765463e-01 -4.34214622e-01 -9.27228689e-01 1.21269956e-01
2.86903858e-01 -6.09450400e-01 -1.29978812e+00 -4.44278121e-01
-3.54129046e-01 -6.19291998e-02 1.27005279e+00 1.83462471e-01
1.06605721e+00 3.33004594e-02 -3.15121502e-01 9.02125716e-01
1.11499012e+00 -6.94254264e-02 6.51841283e-01 4.49513257e-01
1.08541501e+00 7.27937758e-01 3.73264551e-01 2.27312312e-01
6.58422828e-01 7.73788035e-01 9.45843533e-02 -2.77745783e-01
-2.09816769e-01 -3.94560069e-01 1.84256777e-01 2.48179361e-01
-3.57373595e-01 4.10805047e-02 -3.44070584e-01 2.43766144e-01
-1.90648520e+00 -1.13668680e+00 1.63599625e-02 2.53497815e+00
6.50689363e-01 -2.82762885e-01 3.23186278e-01 -8.45352039e-02
1.03059125e+00 -3.61987315e-02 -6.82380855e-01 3.56016845e-01
-4.89507347e-01 -2.30325162e-01 4.44034189e-01 2.34869584e-01
-1.34673011e+00 6.36462867e-01 4.81197596e+00 8.60803843e-01
-9.79392707e-01 7.96663240e-02 4.99159932e-01 -3.39386970e-01
-3.22362423e-01 -1.86047345e-01 -5.80010831e-01 6.14167929e-01
3.27802449e-01 -8.15839395e-02 7.73488879e-01 4.72161561e-01
1.03332274e-01 -1.19949169e-02 -9.61724699e-01 1.45155919e+00
1.93722978e-01 -7.76789665e-01 -2.55235396e-02 6.99944273e-02
5.31622231e-01 -3.14261317e-01 1.98066905e-01 4.33811486e-01
-1.62102908e-01 -9.77048457e-01 5.77501237e-01 9.00379300e-01
9.31061387e-01 -9.75182772e-01 7.62456059e-01 8.19908753e-02
-1.33029902e+00 -2.38663554e-01 -4.43699658e-01 1.04643770e-01
-2.21736178e-01 4.37778473e-01 -4.90649790e-02 9.41942871e-01
7.37331808e-01 6.18571758e-01 -7.14107990e-01 9.69534338e-01
-7.27610365e-02 -1.41017456e-02 -2.80809075e-01 4.49978948e-01
-4.93726194e-01 -3.27782124e-01 4.92606521e-01 8.41047227e-01
-8.74710828e-03 6.78312629e-02 1.21244885e-01 8.86004508e-01
-3.28145444e-01 -2.64307833e-03 -2.20353201e-01 2.41876945e-01
6.16249561e-01 1.34787214e+00 -1.73641160e-01 -3.62450741e-02
-4.67342287e-01 1.31687629e+00 1.61604494e-01 7.73738265e-01
-8.13716769e-01 -4.99961942e-01 7.30846405e-01 -1.44463494e-01
-1.67304486e-01 2.50051975e-01 -1.77871287e-01 -1.44572365e+00
3.90105009e-01 -5.28005183e-01 5.95091105e-01 -5.62734723e-01
-1.73698485e+00 3.62444043e-01 -4.43656743e-02 -1.43156290e+00
1.50675014e-01 -2.45038792e-01 -4.52604502e-01 1.55755496e+00
-1.81907892e+00 -1.64149392e+00 -6.90958321e-01 9.44973111e-01
8.11875053e-03 -3.33674490e-01 5.59463203e-01 7.48522162e-01
-1.01451063e+00 1.23477471e+00 1.15403123e-01 3.76195312e-01
1.07171798e+00 -7.12893784e-01 -1.18415780e-01 8.16240907e-01
-3.30179155e-01 9.04730201e-01 3.79814893e-01 -5.04147828e-01
-1.40908897e+00 -1.11346793e+00 7.25064278e-01 -1.56695202e-01
1.41927212e-01 -3.94723028e-01 -8.62607896e-01 3.03988189e-01
-2.95393378e-01 2.83456653e-01 7.33863413e-01 2.92091072e-01
-8.89149010e-01 -6.35277987e-01 -1.16611207e+00 6.15706146e-01
1.23976874e+00 -9.27964747e-01 -3.05770725e-01 2.07694665e-01
4.86009508e-01 -1.21200785e-01 -6.99185073e-01 4.80682671e-01
9.10862625e-01 -7.38758862e-01 1.31473029e+00 -4.68806922e-01
7.44817927e-02 -3.49231482e-01 -4.00897190e-02 -1.07597923e+00
-6.31911278e-01 -2.65968800e-01 1.39995486e-01 1.67890131e+00
4.49950285e-02 -8.34206581e-01 5.07952273e-01 9.48210120e-01
3.68486553e-01 -1.28333107e-01 -9.91451383e-01 -7.09518731e-01
-1.41269803e-01 -4.13552783e-02 7.90046155e-01 1.11297381e+00
-2.67325222e-01 4.33395773e-01 -5.36362290e-01 2.06431955e-01
8.54053378e-01 2.79857039e-01 4.64098513e-01 -1.37482917e+00
6.12280294e-02 -3.92792135e-01 -2.28084177e-01 -6.51705623e-01
1.87143251e-01 -8.91497910e-01 -1.35679364e-01 -1.30449760e+00
7.31354415e-01 -4.31031257e-01 -8.82886827e-01 5.82678139e-01
-5.86852431e-01 4.09571856e-01 3.55075240e-01 3.98796171e-01
-5.80631375e-01 7.13905036e-01 1.17959380e+00 -3.95799726e-01
-3.55657578e-01 -4.71437350e-02 -1.11376226e+00 4.41974431e-01
6.05750382e-01 -2.84668803e-01 -4.03323650e-01 -3.34899664e-01
-4.69571948e-02 -3.37778419e-01 7.56270409e-01 -8.12327683e-01
3.08948725e-01 -3.31579864e-01 1.05807984e+00 -4.81034547e-01
4.30692226e-01 -6.25260293e-01 7.94589594e-02 8.71588364e-02
-4.06834960e-01 -4.10107344e-01 1.15739800e-01 5.81243992e-01
-2.15211306e-02 3.18012238e-01 7.29557455e-01 1.69202194e-01
-7.96711981e-01 3.69700581e-01 2.25262135e-01 -1.20071426e-01
8.08518946e-01 -1.51677236e-01 -6.76093996e-01 -2.75386512e-01
-4.41131741e-01 4.54299986e-01 4.06564832e-01 8.10725152e-01
5.85495174e-01 -1.74554718e+00 -8.49488676e-01 5.19219816e-01
4.87099588e-01 -3.76239449e-01 1.03536713e+00 8.28199029e-01
3.01923286e-02 2.86527753e-01 -3.46755564e-01 -3.55935127e-01
-1.30017054e+00 5.73640764e-01 4.56011295e-01 5.13742529e-02
-2.99520254e-01 9.24535513e-01 6.42177880e-01 -6.44623399e-01
2.39103153e-01 5.40577650e-01 -5.15168488e-01 1.23354129e-01
7.99660861e-01 4.64668572e-01 -1.08086757e-01 -1.00246608e+00
-7.80741453e-01 6.73411369e-01 -1.77688763e-01 5.24890423e-02
1.09263039e+00 -3.16847891e-01 -1.64726973e-01 -8.20921287e-02
1.20909977e+00 -9.17537883e-02 -1.13157678e+00 -4.85535115e-01
-4.68327343e-01 -5.27605772e-01 -1.62591543e-02 -8.96508038e-01
-1.22596657e+00 6.40339911e-01 1.15454376e+00 -1.42021328e-01
1.58465898e+00 -8.88048410e-02 5.91410398e-01 1.16653465e-01
7.27974623e-03 -1.13485169e+00 -1.72190160e-01 2.80804724e-01
7.75258422e-01 -1.36827576e+00 9.93745402e-02 -5.92734814e-01
-4.29017007e-01 7.97906220e-01 8.12766552e-01 2.29304001e-01
3.40507746e-01 -3.78952175e-01 -5.49975559e-02 1.31251842e-01
-1.48567513e-01 -4.55073863e-01 7.94998050e-01 7.90421247e-01
2.10472301e-01 8.01236555e-02 -2.06744000e-01 8.92911255e-01
1.98927641e-01 -1.53768495e-01 -2.14462057e-01 5.06188631e-01
1.00488991e-01 -9.60823178e-01 -5.14004111e-01 3.22084963e-01
1.89840022e-04 -8.65854695e-02 -5.47185302e-01 5.29884279e-01
5.74165106e-01 1.14407420e+00 -2.17896327e-02 -8.65708470e-01
5.25675297e-01 -1.08159922e-01 4.44445789e-01 -6.46345392e-02
-4.68791038e-01 5.41973598e-02 -1.53932944e-01 -4.38402414e-01
-5.39976597e-01 -6.12390876e-01 -9.88276958e-01 -5.15213788e-01
-3.18892449e-01 -2.06194997e-01 4.19046938e-01 9.42145109e-01
6.66044474e-01 4.90410149e-01 8.77355933e-01 -7.02335715e-01
-3.88371527e-01 -8.53836656e-01 -5.12227297e-01 8.56940925e-01
2.40442857e-01 -7.45165288e-01 -2.46701881e-01 -2.45931774e-01] | [14.7119722366333, 0.9526773691177368] |
4e8abb5c-dfe9-4065-af60-7d00e4bf57e5 | demodulation-of-sparse-ppm-signals-with-low | 1309.5854 | null | http://arxiv.org/abs/1309.5854v1 | http://arxiv.org/pdf/1309.5854v1.pdf | Demodulation of Sparse PPM Signals with Low Samples Using Trained RIP Matrix | Compressed sensing (CS) theory considers the restricted isometry property
(RIP) as a sufficient condition for measurement matrix which guarantees the
recovery of any sparse signal from its compressed measurements. The RIP
condition also preserves enough information for classification of sparse
symbols, even with fewer measurements. In this work, we utilize RIP bound as
the cost function for training a simple neural network in order to exploit the
near optimal measurements or equivalently near optimal features for
classification of a known set of sparse symbols. As an example, we consider
demodulation of pulse position modulation (PPM) signals. The results indicate
that the proposed method has much better performance than the random
measurements and requires less samples than the optimum matched filter
demodulator, at the expense of some performance loss. Further, the proposed
approach does not need equalizer for multipath channels in contrast to the
conventional receiver. | ['Mehdi Chehel Amirani', 'Seyed Hossein Hosseini', 'Mahrokh G. Shayesteh'] | 2013-09-01 | null | null | null | null | ['human-dynamics'] | ['computer-vision'] | [ 9.06222224e-01 -3.55754122e-02 -1.97491512e-01 -3.81564289e-01
-6.78704500e-01 -2.38195583e-01 2.91281372e-01 -3.03376038e-02
-1.49820611e-01 8.55290651e-01 -1.19531266e-02 -4.38260943e-01
-3.85415226e-01 -4.86606538e-01 -6.49636269e-01 -1.02994990e+00
-5.82645297e-01 1.22463591e-02 -3.46826255e-01 2.97807306e-02
5.41904628e-01 6.58533275e-01 -1.20889652e+00 -1.83295846e-01
8.56911898e-01 1.31839430e+00 5.68538308e-01 5.82892299e-01
1.11267023e-01 6.34092331e-01 -5.60343862e-01 2.03407839e-01
5.30134499e-01 -6.55974865e-01 -3.72579508e-02 -6.62837401e-02
3.65227699e-01 -3.32032055e-01 -8.43044281e-01 1.37526989e+00
4.63740110e-01 1.98345957e-03 5.14588058e-01 -8.54950309e-01
-4.89417344e-01 9.23560202e-01 -5.40310264e-01 2.92365968e-01
2.22314611e-01 -5.33079624e-01 5.91467083e-01 -7.80606925e-01
-8.75720847e-03 8.45924199e-01 6.68660641e-01 -7.34964982e-02
-5.86995959e-01 -7.62444973e-01 -6.13285661e-01 4.96150464e-01
-1.70438600e+00 -9.94023561e-01 8.91203403e-01 1.54792517e-01
4.48963523e-01 3.98848027e-01 2.66688764e-01 4.81134087e-01
4.61874008e-01 4.26382571e-01 1.03223777e+00 -7.54653275e-01
2.36330688e-01 -5.72306775e-02 2.14353293e-01 5.02227068e-01
8.76906812e-01 2.33454674e-01 -1.64536044e-01 -1.55766606e-01
5.72742641e-01 5.95579334e-02 -6.36487842e-01 -1.14066057e-01
-1.12632120e+00 7.20084190e-01 1.75644338e-01 7.39926398e-01
-6.33948267e-01 2.12251663e-01 -1.00783885e-01 7.53353596e-01
-1.89905554e-01 4.38934684e-01 1.06567487e-01 2.42926240e-01
-1.23501337e+00 -3.84559125e-01 9.66317773e-01 8.85309398e-01
7.08217919e-01 7.25953579e-01 1.08963512e-01 5.79391360e-01
2.25097224e-01 1.26017439e+00 4.19465750e-01 -1.02644980e+00
6.71205223e-01 -3.18283528e-01 1.79011479e-01 -1.48770380e+00
-3.01893651e-01 -1.10420763e+00 -1.30673432e+00 -2.35902101e-01
1.82542298e-02 -3.53308648e-01 -5.78387678e-01 1.52653575e+00
-1.16916262e-01 7.97724605e-01 4.22387809e-01 8.17747295e-01
2.57772624e-01 7.84811199e-01 -7.72413909e-01 -6.17585659e-01
7.80191541e-01 -4.19924319e-01 -1.02380788e+00 -6.49675071e-01
3.19880635e-01 -8.76741052e-01 1.31193727e-01 6.19036794e-01
-8.44726026e-01 -4.00775194e-01 -1.65553260e+00 7.82624662e-01
1.82968646e-01 1.70883298e-01 5.96492469e-01 9.20228779e-01
-7.27365017e-01 7.06707537e-01 -4.08963472e-01 -1.09863780e-01
1.39313608e-01 3.47347677e-01 -4.42085803e-01 -5.53364813e-01
-1.21593356e+00 7.73603737e-01 3.95146877e-01 2.95813411e-01
-7.36306787e-01 -2.40049541e-01 -9.05059636e-01 3.38659048e-01
-4.23081033e-02 6.73432723e-02 9.43789542e-01 -8.29052627e-01
-1.31718349e+00 7.30243996e-02 -2.42476970e-01 -8.43312144e-01
-2.76960433e-01 -1.39584597e-02 -8.48232150e-01 7.47616649e-01
-2.62840897e-01 -3.45503300e-01 1.28101110e+00 -1.01724625e+00
-3.46116871e-01 -1.37142688e-01 -4.71367151e-01 2.37173103e-02
-1.82714686e-01 -3.14359218e-01 3.33395973e-02 -3.82360905e-01
9.32026207e-01 -6.74605072e-01 -3.08209836e-01 -5.08412659e-01
-2.87435591e-01 5.26819885e-01 8.58515680e-01 -7.86736250e-01
1.19234943e+00 -2.35717058e+00 -2.21634939e-01 8.22329462e-01
-1.82341874e-01 3.25826883e-01 -2.84417868e-01 6.55888796e-01
-1.07055835e-01 -4.36418116e-01 -4.82299984e-01 -1.46628380e-01
-3.07383418e-01 1.71710789e-01 -3.86661172e-01 1.08070731e+00
-2.83396214e-01 2.82992691e-01 -6.45379364e-01 -1.66705132e-01
2.60380208e-01 2.71076709e-01 -1.46297410e-01 -7.92862773e-02
4.34110492e-01 5.50645173e-01 -5.38679302e-01 5.99016011e-01
1.04365897e+00 -1.53652191e-01 2.91155994e-01 -2.67368376e-01
2.01435741e-02 1.17888726e-01 -1.53675175e+00 1.49610317e+00
-4.51487690e-01 7.65988290e-01 5.95821857e-01 -1.70595038e+00
1.29350758e+00 6.49333119e-01 6.34577513e-01 -8.47490668e-01
5.32157719e-01 4.83470738e-01 1.75496891e-01 -4.96424228e-01
2.66100049e-01 -1.93214729e-01 -7.27710053e-02 5.44147372e-01
-2.71055818e-01 2.40269616e-01 -1.61382213e-01 6.46023601e-02
1.24104273e+00 -6.76238894e-01 6.89523876e-01 -1.43273905e-01
7.39532292e-01 -5.84765494e-01 5.60947776e-01 1.07672179e+00
1.59534305e-01 3.05519730e-01 -3.47414881e-01 1.16443589e-01
-9.17233229e-01 -6.56420469e-01 -2.78863519e-01 1.69044793e-01
4.27572429e-01 2.48584747e-01 -2.33423144e-01 4.93031852e-02
1.48412287e-02 7.34890342e-01 -2.11182591e-02 -1.71504557e-01
-5.37266970e-01 -4.04479295e-01 7.44343162e-01 -7.85914734e-02
6.29551530e-01 -6.56760097e-01 -5.44894874e-01 2.47355327e-01
-8.05798694e-02 -1.31213975e+00 -1.40492603e-01 3.10802877e-01
-8.31416190e-01 -9.76765215e-01 -6.86854959e-01 -7.96800137e-01
8.08489144e-01 8.47840428e-01 4.02900100e-01 -1.20198727e-02
1.91255778e-01 2.27724895e-01 -7.06750751e-01 -1.52925909e-01
-4.08665895e-01 -3.93211812e-01 3.88095766e-01 2.69621074e-01
2.28069916e-01 -8.98644328e-01 -4.14755553e-01 1.68888345e-01
-8.63718748e-01 -4.57297236e-01 1.09217477e+00 7.33493924e-01
1.98619068e-01 4.76179302e-01 8.38742733e-01 -3.88692766e-01
5.36008120e-01 -6.75470173e-01 -4.91876453e-01 1.40569776e-01
-5.35883605e-01 -1.73703209e-02 8.88773441e-01 -1.48843497e-01
-7.47617424e-01 6.01839721e-02 1.09109832e-02 -3.25084746e-01
1.00510359e-01 8.89182508e-01 -4.75588143e-02 -7.79103041e-01
6.70752227e-01 7.71200001e-01 2.41924942e-01 -5.86550474e-01
6.20882101e-02 1.22555482e+00 7.65458047e-01 -2.99075872e-01
1.15399957e+00 3.77791226e-01 4.49692905e-01 -1.29635024e+00
-5.73019922e-01 -6.59151435e-01 -2.71972746e-01 1.92547485e-01
-7.93990567e-02 -8.26155543e-01 -5.42665780e-01 3.11810989e-03
-8.69427741e-01 3.50058287e-01 8.47432464e-02 1.16033006e+00
-4.67599928e-01 1.09593034e+00 -3.34946573e-01 -1.17154717e+00
-4.89008218e-01 -6.27342045e-01 3.13598633e-01 2.00958088e-01
2.13457972e-01 -6.20824099e-01 -1.41236350e-01 -3.72741371e-02
7.34302759e-01 2.26232752e-01 3.49364996e-01 -6.81132197e-01
-5.75326800e-01 -8.72199476e-01 -1.91340238e-01 6.01493835e-01
3.14352840e-01 -7.93471813e-01 -6.86478496e-01 -7.62455106e-01
6.12617850e-01 8.32152516e-02 6.79454923e-01 5.99660695e-01
9.11040723e-01 -5.69039047e-01 -1.60302624e-01 8.69143724e-01
1.72675252e+00 5.81371903e-01 8.23687255e-01 7.01998174e-02
2.37129524e-01 1.78951055e-01 6.46356285e-01 7.00994134e-01
-3.21305335e-01 2.65256554e-01 3.53312969e-01 1.17693037e-01
1.69009566e-01 7.24112988e-02 3.68207514e-01 1.14103520e+00
4.81554389e-01 -3.24530482e-01 -5.07808030e-01 3.68903786e-01
-1.42455924e+00 -1.17973351e+00 -9.69355926e-02 2.66012359e+00
3.98963690e-01 -2.36833394e-01 -6.02868855e-01 6.37992740e-01
8.05394292e-01 3.00932139e-01 -3.10392022e-01 -3.56425434e-01
-2.32835904e-01 3.40619028e-01 1.00174868e+00 7.08314061e-01
-8.58688891e-01 3.02703172e-01 6.63701010e+00 7.70145953e-01
-1.22933710e+00 -8.71203318e-02 -2.23928094e-01 3.15063924e-01
-7.22213387e-02 2.02966601e-01 -4.43046480e-01 5.15556991e-01
1.05330348e+00 -1.79805934e-01 5.97050846e-01 5.11898994e-01
3.36274743e-01 -7.70943612e-02 -8.15855324e-01 1.28216159e+00
3.33438009e-01 -1.20874786e+00 -2.67674595e-01 -6.97647547e-03
6.04283392e-01 5.85620627e-02 -8.74629840e-02 -1.21527060e-03
-3.02043974e-01 -9.65748310e-01 2.90481836e-01 6.44737363e-01
8.66165698e-01 -5.82157016e-01 1.02214551e+00 6.64864182e-01
-1.05712783e+00 -3.38824958e-01 -7.27780461e-01 -3.24905753e-01
2.94728994e-01 1.00927138e+00 -9.13082838e-01 8.61371338e-01
-4.59265336e-02 6.66828334e-01 6.99678734e-02 1.54245794e+00
6.09650835e-02 8.32988262e-01 -5.44214249e-01 1.04180269e-01
3.25001419e-01 -2.96122491e-01 8.72432351e-01 1.24868047e+00
1.07722390e+00 2.19463229e-01 1.68509990e-01 8.40592682e-02
1.34657636e-01 -7.18038995e-03 -8.94591570e-01 -1.70073494e-01
1.10373700e+00 9.09046710e-01 -2.14743406e-01 -2.76444644e-01
-2.04320028e-01 7.37291157e-01 -4.23019588e-01 4.83951896e-01
-2.72355437e-01 -6.91400409e-01 6.31808937e-02 -2.10626960e-01
5.53127944e-01 -3.92670572e-01 -1.84079900e-01 -9.56031919e-01
-9.56348255e-02 -1.09523773e+00 -5.14628701e-02 -3.56015712e-01
-1.07190907e+00 2.08094642e-01 -3.29391688e-01 -1.70883346e+00
-3.34815294e-01 -1.19172849e-01 -5.61564386e-01 8.85955453e-01
-1.90156198e+00 -5.23214042e-01 -3.03291231e-01 5.87745547e-01
3.52909267e-02 -5.28302372e-01 7.93820500e-01 3.38419855e-01
-3.36793363e-01 5.68360090e-01 7.89008081e-01 -1.17453873e-01
4.74911004e-01 -6.58483028e-01 -2.28805721e-01 1.15230787e+00
1.40366957e-01 7.64573038e-01 1.14901328e+00 -5.44652224e-01
-1.75743592e+00 -6.70865357e-01 8.22328091e-01 5.02604187e-01
3.79186809e-01 3.10496092e-01 -7.58874476e-01 5.10019004e-01
1.87759846e-01 -1.91435024e-01 8.48896146e-01 -3.47134769e-01
-1.24273553e-01 -2.80455023e-01 -1.37169802e+00 -3.09068952e-02
4.32994068e-01 -3.45408589e-01 -6.17621303e-01 3.05205643e-01
3.49637389e-01 -2.34232351e-01 -8.00271630e-01 3.92491668e-01
6.42074406e-01 -8.42533767e-01 1.03573358e+00 2.18512222e-01
-5.34052402e-02 -5.73258460e-01 -9.49937463e-01 -1.08298206e+00
-5.39403737e-01 -8.76998544e-01 -3.53631437e-01 6.67415082e-01
1.98219866e-01 -8.25294137e-01 6.91199958e-01 -2.91661710e-01
-2.18948320e-01 -4.76061761e-01 -1.14817107e+00 -1.06472456e+00
-5.18828630e-01 -4.24574852e-01 5.76416314e-01 1.09965575e+00
9.47341397e-02 3.13757919e-02 -1.02544439e+00 7.89043605e-01
1.16154170e+00 1.05845064e-01 5.97336113e-01 -1.18194878e+00
-7.64616311e-01 8.56606141e-02 -5.39272666e-01 -1.44280338e+00
1.60963461e-02 -9.36602890e-01 1.87872946e-01 -1.40720344e+00
-2.17065454e-01 -8.07729840e-01 -7.01420486e-01 4.74276729e-02
3.49303961e-01 3.11128914e-01 1.29888013e-01 4.32485312e-01
-2.70812452e-01 4.61384445e-01 1.07465720e+00 -1.18972108e-01
5.49785532e-02 4.57442701e-01 -8.48348141e-01 4.12647516e-01
1.02720416e+00 -5.87083817e-01 -4.30383533e-01 -3.15059513e-01
6.01144359e-02 7.09692180e-01 1.02099076e-01 -1.52131462e+00
3.22313666e-01 -4.14666114e-03 2.90498197e-01 -6.11021876e-01
4.00745153e-01 -1.22216630e+00 2.38576367e-01 7.35019624e-01
-2.04300478e-01 -5.30080140e-01 -2.10414380e-01 8.07411551e-01
-3.02025467e-01 -8.38709712e-01 7.63748169e-01 2.07711250e-01
-5.94458997e-01 1.15613759e-01 -2.61041105e-01 -6.06426179e-01
6.46102309e-01 -5.62570691e-01 -1.32197335e-01 -7.96262920e-01
-3.02960068e-01 -1.43077625e-02 1.07051983e-01 -9.71835554e-02
8.81991804e-01 -1.20181227e+00 -8.46105576e-01 4.70132738e-01
-2.54481673e-01 -6.04786932e-01 2.53226846e-01 9.86523390e-01
-5.47587872e-01 8.77167404e-01 -1.07554957e-01 -4.10480589e-01
-1.04962087e+00 3.29869777e-01 8.76182839e-02 9.05437618e-02
-6.99839532e-01 3.55088651e-01 -5.88303268e-01 -4.48439308e-02
3.15527231e-01 1.62432373e-01 -2.57616848e-01 -4.13855314e-01
8.22993875e-01 3.40774357e-01 -3.02748173e-03 -5.47271669e-01
-2.97350198e-01 7.21565962e-01 2.70475298e-01 -1.15164913e-01
1.13096774e+00 -2.78600335e-01 -1.08543195e-01 1.24040082e-01
1.49735570e+00 4.11259651e-01 -8.26065242e-01 -6.63620710e-01
-1.13072373e-01 -7.22846508e-01 2.60537058e-01 -4.55592483e-01
-8.73610377e-01 6.75580680e-01 7.35411227e-01 3.47358853e-01
1.15487671e+00 -4.75118816e-01 9.88260388e-01 9.88447726e-01
7.59953976e-01 -7.23423123e-01 -3.50006908e-01 4.98361230e-01
6.49110138e-01 -9.21136975e-01 2.20170051e-01 -1.33743823e-01
-1.48145586e-01 1.24309766e+00 -2.66157389e-01 -4.34326172e-01
6.94787621e-01 2.19377443e-01 -1.59544677e-01 1.77352095e-03
-1.31836697e-01 6.61027282e-02 1.14846118e-01 7.39685535e-01
6.01735637e-02 1.33322880e-01 -5.58685780e-01 1.65290594e-01
-2.90892214e-01 1.84234213e-02 7.03055620e-01 9.58804786e-01
-1.27938426e+00 -8.03419113e-01 -8.19975436e-01 7.67979324e-01
-4.43579614e-01 -1.83720410e-01 3.11751723e-01 2.83504337e-01
-2.27345824e-01 1.35483003e+00 -2.39770114e-01 -5.20944595e-01
-1.31208897e-01 -2.22471640e-01 4.92409945e-01 -3.82899731e-01
2.26191834e-01 8.17494541e-02 2.53011193e-02 -4.25215364e-01
-3.83637100e-01 -5.72361887e-01 -1.23589742e+00 -2.17855766e-01
-4.83060211e-01 5.54677665e-01 8.88529301e-01 1.21677959e+00
2.02229857e-01 1.90359190e-01 1.30231619e+00 -5.95025718e-01
-1.05516374e+00 -9.55605447e-01 -1.04452991e+00 -2.38281727e-01
7.21305966e-01 -2.53833354e-01 -7.69406796e-01 -2.44149566e-01] | [6.484221935272217, 1.3733774423599243] |
c17cb597-3e41-4fa9-bfe1-c3005d305caa | concurrent-ischemic-lesion-age-estimation-and | 2306.12242 | null | https://arxiv.org/abs/2306.12242v1 | https://arxiv.org/pdf/2306.12242v1.pdf | Concurrent ischemic lesion age estimation and segmentation of CT brain using a Transformer-based network | The cornerstone of stroke care is expedient management that varies depending on the time since stroke onset. Consequently, clinical decision making is centered on accurate knowledge of timing and often requires a radiologist to interpret Computed Tomography (CT) of the brain to confirm the occurrence and age of an event. These tasks are particularly challenging due to the subtle expression of acute ischemic lesions and the dynamic nature of their appearance. Automation efforts have not yet applied deep learning to estimate lesion age and treated these two tasks independently, so, have overlooked their inherent complementary relationship. To leverage this, we propose a novel end-to-end multi-task transformer-based network optimized for concurrent segmentation and age estimation of cerebral ischemic lesions. By utilizing gated positional self-attention and CT-specific data augmentation, the proposed method can capture long-range spatial dependencies while maintaining its ability to be trained from scratch under low-data regimes commonly found in medical imaging. Furthermore, to better combine multiple predictions, we incorporate uncertainty by utilizing quantile loss to facilitate estimating a probability density function of lesion age. The effectiveness of our model is then extensively evaluated on a clinical dataset consisting of 776 CT images from two medical centers. Experimental results demonstrate that our method obtains promising performance, with an area under the curve (AUC) of 0.933 for classifying lesion ages <=4.5 hours compared to 0.858 using a conventional approach, and outperforms task-specific state-of-the-art algorithms. | ['Daniel Rueckert', 'Paul Bentley', 'Adam Marcus'] | 2023-06-21 | null | null | null | null | ['age-estimation', 'computed-tomography-ct', 'age-estimation'] | ['computer-vision', 'methodology', 'miscellaneous'] | [-0.05933563 -0.34482217 -0.20581767 -0.45170715 -1.2312866 -0.43060175
0.34701705 0.41176784 -0.69575125 0.76922315 0.34726864 -0.5318789
-0.22337495 -0.5041366 -0.5776902 -0.6142305 -0.3718525 0.8078494
0.3669524 0.4579186 0.08517648 0.6181612 -0.9720506 0.24040438
1.219663 1.0538172 0.29477957 0.67305213 0.16870834 0.7689444
-0.38583657 -0.09819058 0.14765543 -0.14985187 -0.65963006 -0.10437658
0.14429356 -0.78924155 -0.65039897 0.614843 0.8292344 0.03923589
0.989523 -0.9765292 -0.26440328 0.40624276 -0.6531104 0.8149884
-0.23540947 0.15950882 0.5955427 -0.62196785 0.16020486 0.6024251
0.5805512 0.41004002 -1.0053164 -0.53690624 0.18953311 0.67672473
-1.1889136 -0.18091604 0.33764392 -0.60239196 0.64582705 -0.1207123
0.7469138 1.1353099 0.47853535 0.66844827 1.1065158 -0.05205856
0.22478127 -0.4324308 0.27183893 0.6272998 0.08242144 0.09734394
-0.25530502 -0.15015635 0.7590719 0.4144113 -0.31194198 -0.27834395
-1.1173311 0.66180027 0.57932585 -0.04818225 -0.75031257 0.13850878
0.5458004 -0.19429904 0.42585188 -0.16632904 -0.3795895 -0.36361519
-1.1315334 0.10179998 0.29578823 0.52366316 -0.01548082 -0.19777243
-0.49120256 0.70610917 -0.24466076 0.49410343 0.57464045 -0.5835325
0.36244267 0.43113175 0.03360655 -0.18127413 -0.8019412 -0.7489533
-0.8275643 0.2761752 0.78263515 -0.3649228 -1.3554609 1.7771933
-0.02698992 0.35454977 -0.3713193 0.9471147 0.32274136 0.2326503
0.47131234 -0.06845347 1.543016 -0.7404197 -0.23258027 -0.37725747
0.5876323 -0.2580316 1.07553 0.19907574 -1.0135038 0.02329697
-0.74225914 0.03014295 -0.02126872 0.21420002 0.46774745 0.5497236
-0.7080892 0.44034153 -1.4319618 -0.1468177 1.0024434 0.27427348
-0.15836129 -0.20234853 -0.85951066 1.1993092 0.09947804 -0.08412155
-0.9037693 -1.3396472 -0.27654818 0.15446933 0.09634692 -1.094975
1.1863617 -0.39771596 -1.0915948 0.60671 -0.4230816 -0.83970445
1.000403 -0.5141176 -0.14497162 0.47365013 0.24951747 0.43616402
0.5561786 -0.8096697 -0.7249582 -0.6692313 -0.19905977 0.17693749
-0.2757162 0.10144641 -0.15709694 -0.6620881 0.02923003 -0.76342285
-0.41397938 0.26944658 0.01787952 0.1312663 0.61633134 -0.936596
1.0171647 -1.6980038 -0.13396426 0.10432985 0.55884814 0.24854368
0.31106046 -0.0912045 -0.10046082 -0.08610063 -0.47084683 -0.1260569
-0.39413658 -0.04809055 -0.17460093 0.62281686 0.05922947 1.088917
-0.77312416 -0.62549126 0.2717254 0.64200014 -0.31859833 0.3320836
0.14375135 0.79698443 -0.67348325 0.45158917 0.32521087 -0.4649605
-0.18879654 -0.26250455 -0.05477648 0.05437033 -0.48467267 1.677387
-0.5859626 0.35816258 -0.3105325 -1.1838931 0.41451216 0.45530525
0.98411053 -0.6428944 0.38784733 0.18543388 0.3229062 -0.54545015
-0.35673475 -0.10862664 0.15052547 0.6665996 -0.41497207 0.2554691
0.05269263 0.21699108 1.5419953 -0.16359305 0.12354852 -0.145772
0.2182372 -0.09982491 0.5326454 0.90285856 -0.6070982 0.9128831
0.42608884 -0.45707616 -1.137989 -1.4762465 -0.5422788 0.72190857
-0.18471432 0.08132611 -0.73737365 -0.6467488 -0.10800126 0.59087557
-0.7273603 -0.259095 -0.8062227 -1.1450037 0.38387522 1.1316255
0.49421954 -0.6448928 -1.148866 0.43333554 -0.42321086 -1.2391063
-0.5084227 0.02591233 -1.0216917 -1.074616 -1.3036445 -0.54709315
0.49055383 0.10435622 0.9539717 -0.03542773 -0.5557739 0.4437037
-0.33179975 -0.41997573 0.00920337 0.36656564 -0.17480892 -0.08399136
0.13410813 -1.1152422 -1.3000091 0.1933966 -0.5957433 0.01131268
0.96107763 0.94113344 0.47208372 -0.4013224 1.0141733 -0.6926219
0.4373731 -0.588427 -0.2996763 0.37387732 -0.7479161 0.03262919
0.56177425 -0.595831 -0.98148084 0.03405246 0.09183881 -0.3764043
-0.15155965 0.5493028 0.10265858 0.13441119 0.54858226 0.2667592
0.08283649 -0.21337228 0.23473309 0.55744797 0.94064754 -0.7306913
0.3842687 0.59289026 0.11138538 -0.49998927 -0.81230706 -0.43569207
-0.6553949 -0.29632965 0.81953037 -0.93598163 -0.5542616 0.60281575
-0.956089 -0.36704236 -0.0056591 0.67204446 -0.5543629 0.34714207
-0.5942686 -0.53063846 -0.78135043 -1.3068192 0.8153108 0.12214531
-0.00878307 -0.8839881 -0.20762298 0.3666106 0.7242454 0.43747348
1.3348948 -0.45610443 -0.49918804 -0.46266815 -0.68565017 -0.02104747
0.21058625 -0.4387021 -0.6602809 -0.22572203 -0.31832096 -0.11994343
0.8833825 1.0423009 1.4559672 0.23956639 -0.35261992 0.55038506
1.0337226 0.18086296 0.45774826 0.42235166 0.6132844 0.22231999
0.2272661 0.7392307 0.71182483 0.510808 0.44953316 -0.04579541
-0.32225996 0.23843966 0.01048801 0.21029194 -0.18297945 -0.04648272
-1.2331942 0.71843654 -1.7728229 -0.8104421 -0.08137041 2.2443194
0.7612504 0.18717106 0.35320616 -0.05255939 0.7108257 -0.07465534
-0.99448353 0.15292019 0.07247157 0.44280848 0.71620464 0.2106232
-1.1492418 0.5320064 5.653282 0.47292545 -1.3324329 0.36796603
0.9119115 -0.3621838 0.20468916 -0.16718638 -0.27263653 0.6231391
1.0088826 -0.4065307 0.16836584 0.54713035 0.53069144 -0.21830903
-1.1110398 0.6967749 -0.2243062 -1.0785259 -0.20739537 -0.14193228
0.44729877 0.4251487 0.22167845 0.07429057 0.12577118 -1.0182941
0.7241749 0.67014885 0.9292992 -0.72979885 0.64736384 0.32715738
-1.0876894 -0.14453539 0.13625523 0.14698943 0.5939992 0.68773764
-0.75903773 0.33532584 0.81622 0.59540725 -0.5183067 1.5427246
-0.1389273 0.73311335 -0.41029927 0.36744383 0.19724874 0.08401983
0.4117817 0.92726624 0.44055176 0.42646095 0.17109922 0.63186646
0.12759665 0.05343751 -0.07874113 0.4365071 0.547771 0.98544794
-0.79733694 -0.39759082 -0.4342366 0.6592692 0.4574735 0.2550677
-0.97764224 -0.06279297 0.39618194 0.29260665 0.13828877 -0.29166168
-0.8586406 -1.0870824 0.28097254 -0.45260492 0.47433203 -0.5177132
-1.3295723 0.7083854 -0.01369066 -1.0803187 -0.25620562 -0.37903616
-0.98091924 0.9927083 -1.5194128 -1.1795679 -0.5022675 0.560561
0.47530872 -0.03060283 0.6688763 0.5297649 -0.73015743 0.6515748
0.05473875 0.20350435 0.7129313 -1.2148852 0.19191672 1.0513309
-0.4077103 0.28358865 0.4221217 -0.6415415 -0.8176348 -1.2185949
0.5110855 -0.49116006 0.73915875 0.12333058 -1.0447905 0.489328
-0.3504592 0.3505281 0.5271069 -0.17654619 -0.3886745 -0.0636671
-1.01532 0.5318118 0.9864962 -0.40043133 -0.565939 0.38552162
0.26308152 -0.50493026 -0.98840433 0.62587434 0.84247255 -0.66961795
1.1123027 -0.8044194 0.48853186 0.13032417 0.19008073 -1.1795572
-0.22569644 -0.2005779 -0.08633307 0.6740503 0.41575482 -0.65121746
1.1310673 1.0461138 -0.4538529 -1.1517463 -0.99926925 -0.64705104
0.55475837 -0.4876168 0.42941424 0.4180721 -0.12728956 -0.03203232
-0.1822197 0.51806957 0.7324548 0.12908496 0.19127183 -1.222176
-0.0729671 -0.7182247 -0.3343088 -0.7135607 0.12426589 -0.9544185
-0.09835562 -1.7099085 0.5675862 -0.6522364 -0.6277151 0.4361304
-0.5001367 -0.06483224 -0.11863898 0.23925114 -0.23137406 0.54368114
1.0765526 0.01491631 -0.17090635 0.25444785 -0.5306324 0.67465264
0.91919 -0.5806978 -0.4421332 -0.6345756 -0.4206462 0.43549263
0.5906681 -1.1691326 0.34432322 -0.0190246 0.53331006 -0.6579862
0.22262307 -0.6913672 -0.28434178 0.6275907 -0.2866342 0.32524258
-0.01534456 0.5382943 0.12682661 0.10545542 0.9134943 0.09185522
-0.44598758 0.9284735 -0.678193 0.42808244 1.0106678 0.03546246
-0.31858385 -0.29449457 -0.852903 0.39252624 0.06620834 0.17722623
0.6805344 -0.9387177 -0.94134307 -0.2049524 0.01643252 -0.01215376
0.4972371 1.2881716 -0.6064858 0.24612342 -0.3728776 -0.7595381
-0.9660964 0.09356246 0.4960516 -0.42658657 -1.2181394 0.50699204
0.17031363 0.2171155 0.38064405 -0.3827226 -0.01024403 -0.19030403
0.6627069 0.3337007 0.3115636 -0.29876643 -0.40312418 0.4142081
-0.50428325 -0.15809335 1.4664147 0.01278636 0.21259002 0.01496723
1.0128094 -0.6369499 -1.5354844 -0.30817246 -0.07717279 -0.24632967
0.42178056 -1.1481701 -1.2862824 1.1510661 1.036626 -0.54037774
1.1449127 0.05166116 1.2324415 0.17782757 0.28854427 -0.6615841
-0.02405442 0.26140293 0.58687896 -1.2245938 -0.08506106 -0.19677326
-0.6208243 1.0686735 0.38362578 -0.02977219 0.82523054 0.28454733
0.17472471 0.08276877 -0.57151663 -0.15221128 0.1784801 0.62159795
0.23216383 0.17805207 -0.41254622 0.92645806 -0.09641166 0.22928558
0.43302393 0.9384307 -0.3283022 -1.0155662 0.0089687 0.9968822
-0.78990525 -0.21490403 0.33888263 0.47882408 -0.175132 0.6767645
0.00674923 0.04113727 0.35068962 0.13166377 0.54610085 -0.34235764
-0.31034413 -0.20381448 -0.16820909 -0.31873608 -0.03225497 -1.0092185
-1.4620763 -0.03439216 0.06490191 -0.06153273 0.61987364 1.2187438
0.45477706 0.7589462 0.4847307 -0.7138889 -0.6751754 -0.77399576
-0.29966775 0.2994272 0.30606365 -0.89892834 -0.08990701 0.08033718] | [14.367881774902344, -2.0885422229766846] |
79ba0ddb-8608-4954-ac41-cadc5b16a29c | target-aware-tracking-with-long-term-context | 2302.13840 | null | https://arxiv.org/abs/2302.13840v1 | https://arxiv.org/pdf/2302.13840v1.pdf | Target-Aware Tracking with Long-term Context Attention | Most deep trackers still follow the guidance of the siamese paradigms and use a template that contains only the target without any contextual information, which makes it difficult for the tracker to cope with large appearance changes, rapid target movement, and attraction from similar objects. To alleviate the above problem, we propose a long-term context attention (LCA) module that can perform extensive information fusion on the target and its context from long-term frames, and calculate the target correlation while enhancing target features. The complete contextual information contains the location of the target as well as the state around the target. LCA uses the target state from the previous frame to exclude the interference of similar objects and complex backgrounds, thus accurately locating the target and enabling the tracker to obtain higher robustness and regression accuracy. By embedding the LCA module in Transformer, we build a powerful online tracker with a target-aware backbone, termed as TATrack. In addition, we propose a dynamic online update algorithm based on the classification confidence of historical information without additional calculation burden. Our tracker achieves state-of-the-art performance on multiple benchmarks, with 71.1\% AUC, 89.3\% NP, and 73.0\% AO on LaSOT, TrackingNet, and GOT-10k. The code and trained models are available on https://github.com/hekaijie123/TATrack. | ['Zhiwen Wang', 'Zhixin Li', 'Sheng Xie', 'Canlong Zhang', 'Kaijie He'] | 2023-02-27 | null | null | null | null | ['visual-tracking', 'video-object-tracking', 'visual-object-tracking'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-4.43948209e-01 -5.76269627e-01 -2.56312937e-01 -2.56982058e-01
-5.05417407e-01 -5.04041433e-01 5.01215994e-01 -1.08921677e-01
-4.62503701e-01 4.40070570e-01 2.67328080e-02 2.40848646e-01
2.55952418e-01 -4.22029227e-01 -6.49207771e-01 -8.76986623e-01
-5.37500717e-02 7.00176433e-02 7.59296775e-01 1.02606609e-01
-1.06575117e-01 3.74771416e-01 -1.33714092e+00 -1.92948744e-01
7.66632140e-01 1.39902341e+00 2.88675576e-01 2.97665834e-01
1.80409700e-01 4.37412292e-01 -7.00096905e-01 -3.33579987e-01
2.58915901e-01 -1.14631720e-01 1.98914990e-01 -2.81306088e-01
6.71124995e-01 -2.98389345e-01 -6.49518013e-01 1.18113661e+00
6.12170458e-01 3.37860323e-02 1.82642221e-01 -1.19977856e+00
-4.07657743e-01 1.45628065e-01 -8.20560217e-01 5.69949865e-01
6.38122261e-02 6.16034210e-01 6.08222783e-01 -9.07468975e-01
3.35941195e-01 1.19215798e+00 7.45046258e-01 4.45052415e-01
-8.32152545e-01 -1.18266058e+00 6.23140633e-01 3.76905739e-01
-1.41602576e+00 -3.92368406e-01 6.00032687e-01 -3.45462352e-01
3.05127680e-01 8.68589655e-02 9.61926043e-01 1.08965099e+00
3.21267247e-01 7.72026122e-01 6.35380149e-01 1.49381846e-01
6.56407773e-02 -1.97149336e-01 2.14361385e-01 5.02883554e-01
4.57783461e-01 4.48472887e-01 -4.13716465e-01 -2.71379948e-03
4.77888942e-01 2.69314051e-01 -2.37412035e-01 -3.16727519e-01
-1.39731598e+00 4.76530701e-01 7.91231811e-01 1.37928754e-01
-3.16389382e-01 2.62053698e-01 2.76530147e-01 -1.56276852e-01
2.79519707e-01 -9.51507539e-02 -2.95240074e-01 -1.67691335e-01
-7.41263986e-01 -1.27127748e-02 3.21279734e-01 8.91953468e-01
5.60141742e-01 2.27457002e-01 -5.62664032e-01 4.41516221e-01
6.02538645e-01 1.17617118e+00 4.00676966e-01 -6.26891434e-01
3.39963645e-01 6.04765594e-01 3.65092635e-01 -1.06961071e+00
-3.34146202e-01 -9.68180180e-01 -5.31215847e-01 1.51869416e-01
4.54681277e-01 -2.22155869e-01 -9.31342900e-01 2.04399276e+00
9.20535207e-01 7.02407360e-01 -2.16482267e-01 1.11000824e+00
7.01202154e-01 6.64999485e-01 1.10658422e-01 -3.56042624e-01
1.39529502e+00 -9.83258009e-01 -7.72203207e-01 -3.19375217e-01
3.22545737e-01 -8.01143587e-01 7.51679182e-01 1.37729153e-01
-4.72927809e-01 -8.03706348e-01 -1.02507162e+00 2.81790286e-01
-9.95531008e-02 5.84764659e-01 5.36378384e-01 3.27309072e-01
-6.29477382e-01 2.49541864e-01 -1.10688043e+00 -3.17064345e-01
4.52770323e-01 2.84185410e-01 2.37083845e-02 -1.53917829e-02
-1.08292651e+00 7.43319690e-01 2.65781462e-01 4.23926413e-01
-1.08441687e+00 -7.71676898e-01 -6.95351660e-01 -1.24931321e-01
6.72818363e-01 -2.95508981e-01 1.18740988e+00 -8.00546587e-01
-1.31464136e+00 2.55997032e-01 -2.75665700e-01 -4.70062882e-01
5.54066598e-01 -5.72378278e-01 -6.80658340e-01 -2.55847573e-01
5.13431840e-02 5.08950830e-01 8.45998704e-01 -7.95276642e-01
-8.58629763e-01 -3.99939895e-01 -2.13806957e-01 2.33228430e-01
-2.53600329e-01 -3.11644468e-02 -1.04577482e+00 -7.04894185e-01
-1.36544451e-01 -1.11137938e+00 1.02719348e-02 3.38052064e-01
-1.23186886e-01 -3.66419643e-01 1.16175628e+00 -5.82550049e-01
1.36379397e+00 -2.40677023e+00 -3.01567376e-01 -7.34381080e-02
2.85719901e-01 7.41874874e-01 -2.17521295e-01 -2.40536388e-02
1.52353331e-01 -5.60270965e-01 2.86471397e-01 -1.91056281e-01
-1.09936245e-01 -7.48017207e-02 -1.34524658e-01 6.19062304e-01
-2.90633012e-02 9.77156818e-01 -1.02709174e+00 -5.18047929e-01
3.26857835e-01 6.03305817e-01 -1.76104113e-01 7.04573244e-02
-2.66309947e-01 4.77814674e-01 -6.90364838e-01 6.90937817e-01
7.06010938e-01 -2.76866823e-01 -1.25099912e-01 -4.88197535e-01
-3.54524672e-01 2.10545845e-02 -1.29253018e+00 1.54612553e+00
-4.22555320e-02 6.32838130e-01 -4.80149984e-02 -3.88121158e-01
9.26120162e-01 3.97993624e-02 6.83914423e-01 -9.41811502e-01
2.68855751e-01 2.02570986e-02 2.92175680e-01 -4.32687402e-02
1.78207710e-01 5.09473085e-01 7.26030692e-02 6.56893328e-02
-1.69654027e-01 7.54722178e-01 2.07084820e-01 3.08534205e-01
1.02280867e+00 2.59644568e-01 7.84138590e-02 -1.17919467e-01
5.65212131e-01 -7.14864954e-02 1.19623542e+00 5.89550734e-01
-5.77754319e-01 1.42528608e-01 2.89003570e-02 -5.38748562e-01
-5.23020566e-01 -1.07454479e+00 -8.93877968e-02 8.42762053e-01
7.52217114e-01 -4.22028512e-01 -4.96900976e-01 -6.53812408e-01
2.33047292e-01 3.69646132e-01 -6.55603826e-01 -4.86042678e-01
-5.69984913e-01 -6.63106978e-01 2.15204924e-01 4.71544057e-01
5.42540848e-01 -7.97709703e-01 -5.39018989e-01 2.87419766e-01
-2.41761610e-01 -1.09504747e+00 -1.04459071e+00 -4.20971602e-01
-6.02316201e-01 -1.14139593e+00 -5.39639235e-01 -3.63029599e-01
4.60771829e-01 4.75877255e-01 7.47329831e-01 5.85346632e-02
-2.42475763e-01 2.24687696e-01 -1.36822551e-01 -4.83537793e-01
1.36733562e-01 -2.10160136e-01 3.18332672e-01 3.06774855e-01
3.13455582e-01 -2.51922220e-01 -8.75377595e-01 7.30789840e-01
-3.29773009e-01 -7.22118020e-02 6.22660279e-01 5.61212182e-01
8.75404418e-01 -8.80189613e-02 4.50395167e-01 -1.40971635e-02
-4.30398807e-02 -4.30160135e-01 -1.12970185e+00 2.43281066e-01
-5.19180536e-01 -1.88330054e-01 4.39990282e-01 -9.85439599e-01
-9.11147296e-01 2.71152198e-01 1.94993988e-01 -9.09012616e-01
3.11582714e-01 3.62261944e-02 -4.74368542e-01 -1.72494188e-01
4.64450806e-01 2.54730612e-01 -4.77508940e-02 -4.74065721e-01
4.36059684e-02 2.61400491e-01 6.77640140e-01 -2.38254488e-01
1.07312274e+00 5.22756100e-01 -1.23431809e-01 -4.04463708e-01
-1.11161971e+00 -5.30044734e-01 -2.21086547e-01 -5.92856109e-01
6.98028862e-01 -1.22999418e+00 -9.24442112e-01 6.80173099e-01
-9.09251392e-01 -1.47238553e-01 -2.15982169e-01 8.43453526e-01
1.33859515e-01 4.02132481e-01 -4.06990170e-01 -7.30903685e-01
-5.74342787e-01 -1.01546788e+00 8.28561425e-01 9.07707155e-01
3.73750687e-01 -5.23089349e-01 1.03228785e-01 7.54529843e-03
5.42454422e-01 2.47709215e-01 -1.73000649e-01 -4.78171974e-01
-9.40715015e-01 -3.82310480e-01 -2.73553431e-01 6.38330504e-02
2.23782152e-01 6.80951774e-02 -7.27987409e-01 -6.53595746e-01
-1.64175197e-01 7.69340768e-02 8.25887263e-01 4.96099979e-01
7.11863697e-01 -1.04065724e-01 -7.82900929e-01 6.13843799e-01
1.11968017e+00 3.76965821e-01 3.31344962e-01 8.31707790e-02
8.10340106e-01 -1.71421599e-02 1.26992559e+00 4.35745865e-01
4.97171909e-01 1.08677685e+00 6.57954454e-01 1.84934199e-01
-3.58927786e-01 -2.52808869e-01 7.59796321e-01 5.54579616e-01
2.46698499e-01 -7.07170293e-02 -6.82525218e-01 5.85860550e-01
-1.96320736e+00 -9.98077810e-01 -1.91099107e-01 2.40379071e+00
6.32567704e-01 3.81951809e-01 1.99480623e-01 -6.23315454e-01
1.06400907e+00 1.43456757e-01 -1.06442606e+00 5.63165665e-01
4.12403978e-02 -3.66884142e-01 5.52244902e-01 3.10711294e-01
-1.24752510e+00 9.63901758e-01 4.76448441e+00 1.00314629e+00
-1.33071589e+00 1.52375847e-01 2.67858088e-01 -5.13491571e-01
3.81631970e-01 2.51970068e-02 -1.25740445e+00 1.10268295e+00
7.70158589e-01 -2.47037098e-01 2.00097039e-01 9.90058124e-01
3.18765551e-01 -2.98015356e-01 -7.82898724e-01 1.02623487e+00
-5.59610836e-02 -1.00856912e+00 -3.75833780e-01 1.65783212e-01
3.88860703e-01 5.05150914e-01 6.04219772e-02 4.49077904e-01
2.40895778e-01 -3.71605068e-01 7.79568911e-01 6.79035544e-01
4.01097268e-01 -5.17329514e-01 6.56563580e-01 2.51134217e-01
-1.83061671e+00 -8.99010673e-02 -3.00394058e-01 2.97238708e-01
4.85357275e-04 7.54184306e-01 -3.13233554e-01 6.40054762e-01
9.24447894e-01 1.05906582e+00 -8.32985997e-01 1.66287875e+00
-1.98631570e-01 5.61757684e-01 -7.73288667e-01 7.15080127e-02
4.29011919e-02 1.47742825e-02 8.84351850e-01 9.13604736e-01
3.25901151e-01 -9.56399925e-03 5.75860858e-01 3.91246885e-01
1.68139398e-01 -3.18957306e-02 -6.85197860e-02 2.38600776e-01
7.60375738e-01 1.50737035e+00 -5.60858130e-01 -4.49704140e-01
-3.93066466e-01 4.76346970e-01 1.26735002e-01 1.51163459e-01
-1.52110982e+00 -1.47158936e-01 8.37559104e-01 -1.04823910e-01
6.39656842e-01 -1.86749786e-01 2.79920310e-01 -1.40694964e+00
3.83838505e-01 -6.95208132e-01 5.44957101e-01 -6.57333136e-01
-1.06585801e+00 5.83201051e-01 -2.04599187e-01 -1.72859013e+00
2.70257741e-01 -2.58699387e-01 -8.00180793e-01 6.29987001e-01
-1.34534323e+00 -1.14788604e+00 -6.63693547e-01 5.70017576e-01
2.16039240e-01 -2.71599535e-02 4.94837910e-02 6.04256690e-01
-9.56534207e-01 7.86619186e-01 4.65506539e-02 2.45161116e-01
1.06634474e+00 -8.23191404e-01 2.46659845e-01 1.07154322e+00
-5.08022215e-03 4.48504031e-01 5.98785102e-01 -9.52489495e-01
-1.51545668e+00 -1.28813875e+00 3.95903796e-01 -5.23317754e-01
7.66211033e-01 -3.34391028e-01 -9.12131250e-01 4.94737625e-01
-1.74326256e-01 5.53871870e-01 1.74899355e-01 2.64072791e-03
-3.25703770e-01 -6.39172316e-01 -7.34955609e-01 4.96571153e-01
1.07584035e+00 -3.26075330e-02 -3.74367535e-01 2.09150970e-01
7.66076982e-01 -7.21370339e-01 -7.62657821e-01 4.57920045e-01
7.48612821e-01 -5.99611104e-01 8.90136302e-01 -1.22510307e-01
-4.42969978e-01 -9.78037179e-01 1.66851953e-01 -8.61184835e-01
-4.61607188e-01 -4.46869314e-01 -6.44473970e-01 1.26573217e+00
6.18719086e-02 -6.67884529e-01 6.54296100e-01 3.38034689e-01
-2.08663523e-01 -8.09298277e-01 -1.08782327e+00 -9.67056811e-01
-5.51627815e-01 -3.92832249e-01 5.24867475e-01 6.73948944e-01
-6.40851378e-01 2.60903776e-01 -5.95066011e-01 5.16294658e-01
8.76356781e-01 3.70870680e-01 9.11671221e-01 -1.28982699e+00
-5.32146655e-02 -1.96264580e-01 -4.94262487e-01 -1.11314273e+00
-1.06322289e-01 -5.44253290e-01 1.13289334e-01 -1.17842579e+00
2.17853159e-01 -4.92723793e-01 -7.02400208e-01 6.32011652e-01
-3.57647926e-01 2.18281627e-01 3.64804596e-01 5.40796638e-01
-1.19770527e+00 8.63599658e-01 1.28090835e+00 -9.95432064e-02
-1.36959776e-01 1.56310230e-01 -5.40054083e-01 5.60710669e-01
7.40064263e-01 -7.65018165e-01 -1.54266551e-01 -4.08498377e-01
-3.00086945e-01 -2.17251465e-01 5.99579334e-01 -1.47033286e+00
5.16090214e-01 -1.51918799e-01 7.75741160e-01 -9.41255152e-01
4.83525991e-01 -9.19086576e-01 3.74552786e-01 7.25917339e-01
1.67688817e-01 4.17152680e-02 3.66703123e-01 9.46746051e-01
-3.83225791e-02 2.94926614e-01 8.67640436e-01 4.36664969e-01
-8.34095538e-01 8.10798585e-01 1.75850958e-01 3.80556062e-02
1.34632432e+00 -2.88542011e-03 -4.27347630e-01 -1.28076971e-01
-6.36567652e-01 8.72184038e-01 5.51518500e-01 7.71674395e-01
2.34388411e-01 -1.60854447e+00 -4.76966590e-01 1.76132806e-02
2.23719418e-01 -2.11855009e-01 5.44869781e-01 1.17530859e+00
9.92952064e-02 2.52141029e-01 -1.23106986e-02 -9.38952446e-01
-1.31376278e+00 6.35655642e-01 6.27798140e-01 -1.69440135e-01
-7.50889182e-01 4.75418150e-01 5.08474410e-01 8.08920190e-02
3.77539158e-01 -2.92310864e-01 -1.46506965e-01 -8.40217993e-03
8.81865501e-01 1.44111097e-01 -2.66589344e-01 -7.67218173e-01
-7.58087993e-01 6.32459402e-01 -2.40369812e-01 3.60135764e-01
8.41854930e-01 -1.16925858e-01 2.69666135e-01 1.51024565e-01
6.87245965e-01 6.71466514e-02 -1.93258095e+00 -5.40215552e-01
-1.42034590e-01 -6.14149392e-01 6.41901121e-02 -9.52611983e-01
-1.43281221e+00 5.56344211e-01 1.13286543e+00 -2.93114334e-01
1.06868315e+00 -3.35737392e-02 8.80254567e-01 2.09567860e-01
2.97219485e-01 -9.34355199e-01 1.06940493e-01 5.68838179e-01
6.39915526e-01 -1.27562213e+00 1.35791674e-01 -1.41556859e-01
-5.34482718e-01 6.50980592e-01 1.06164968e+00 -5.20190522e-02
6.81717515e-01 1.37892455e-01 8.34693387e-02 2.22837590e-02
-7.81984031e-01 -4.41900253e-01 4.55026001e-01 4.18396950e-01
-1.70912668e-01 -1.05296619e-01 -1.51399389e-01 5.77836990e-01
2.79015571e-01 -1.10170484e-01 -3.32149416e-02 8.32792163e-01
-4.00063962e-01 -8.29213679e-01 -4.32559162e-01 3.01927418e-01
-4.40611988e-01 1.28731623e-01 -1.57605901e-01 7.66226947e-01
2.87819028e-01 7.78445065e-01 -1.01294182e-02 -4.99666750e-01
4.04951364e-01 -3.38868350e-01 1.46399230e-01 -2.23096177e-01
-5.21477520e-01 3.69470567e-01 -1.90519109e-01 -9.35014844e-01
-3.45308930e-01 -8.81495774e-01 -1.22174132e+00 -2.32241675e-01
-6.44963086e-01 2.01543480e-01 5.43387890e-01 7.65908837e-01
7.65771091e-01 5.27407289e-01 6.32876396e-01 -8.27217221e-01
-3.28497261e-01 -9.19668972e-01 -1.59194425e-01 2.03452796e-01
4.17350084e-01 -1.14493561e+00 -1.56511500e-01 -2.21815810e-01] | [6.270144939422607, -2.1452155113220215] |
b270e3df-49f3-4827-917b-31b9f1a9c8a4 | hrtf-upsampling-with-a-generative-adversarial | 2306.05812 | null | https://arxiv.org/abs/2306.05812v1 | https://arxiv.org/pdf/2306.05812v1.pdf | HRTF upsampling with a generative adversarial network using a gnomonic equiangular projection | An individualised head-related transfer function (HRTF) is essential for creating realistic virtual reality (VR) and augmented reality (AR) environments. However, acoustically measuring high-quality HRTFs requires expensive equipment and an acoustic lab setting. To overcome these limitations and to make this measurement more efficient HRTF upsampling has been exploited in the past where a high-resolution HRTF is created from a low-resolution one. This paper demonstrates how generative adversarial networks (GANs) can be applied to HRTF upsampling. We propose a novel approach that transforms the HRTF data for convenient use with a convolutional super-resolution generative adversarial network (SRGAN). This new approach is benchmarked against two baselines: barycentric upsampling and a HRTF selection approach. Experimental results show that the proposed method outperforms both baselines in terms of log-spectral distortion (LSD) and localisation performance using perceptual models when the input HRTF is sparse. | ['Lorenzo Picinali', 'Samuel J. Cooper', 'Isaac Squires', 'He Liu', 'Mads Jenkins', 'Aidan O. T. Hogg'] | 2023-06-09 | null | null | null | null | ['super-resolution'] | ['computer-vision'] | [ 4.63378519e-01 2.30589569e-01 4.73817945e-01 -3.19295555e-01
-1.57824266e+00 -3.67614836e-01 4.76750016e-01 -6.45584881e-01
-2.11290747e-01 8.78154516e-01 6.15892828e-01 8.40139315e-02
8.74230340e-02 -5.65588057e-01 -7.94229388e-01 -8.77817273e-01
1.37494057e-01 -1.83118917e-02 -5.12391888e-02 -2.85201937e-01
-2.66461641e-01 4.52987194e-01 -1.98466647e+00 3.48556526e-02
7.25680530e-01 7.78799713e-01 3.86914074e-01 1.01891494e+00
4.05080199e-01 4.24931943e-01 -8.93664837e-01 -2.25525200e-01
5.55495977e-01 -3.89997065e-01 -2.55494535e-01 -4.23258245e-01
8.03515077e-01 -2.12513164e-01 -3.84227455e-01 9.09109175e-01
1.21966994e+00 5.17275989e-01 4.21531200e-01 -8.79602134e-01
-5.63583374e-01 4.66027498e-01 -3.09548587e-01 2.40673702e-02
7.96433091e-01 -3.35078016e-02 5.96263587e-01 -7.50438154e-01
5.45273542e-01 1.12864876e+00 8.49681079e-01 5.67890346e-01
-1.35144913e+00 -9.47851360e-01 -4.35546964e-01 5.01065589e-02
-1.47566283e+00 -7.67205358e-01 8.86526227e-01 -1.89725533e-02
7.18167901e-01 5.37614226e-01 4.57765847e-01 1.53910458e+00
-6.27887994e-02 1.77926943e-01 1.42080319e+00 -4.50204313e-01
2.36241549e-01 2.78999031e-01 -5.50716579e-01 4.70309928e-02
-1.48336008e-01 6.87648475e-01 -5.25954366e-01 -6.43291995e-02
9.17625248e-01 -5.93995571e-01 -7.10128188e-01 3.30129638e-02
-8.63738298e-01 7.11849034e-01 3.56224597e-01 5.05924642e-01
-4.45811510e-01 3.21126074e-01 2.93201596e-01 4.92974877e-01
3.92735630e-01 5.02492785e-01 -2.15209886e-01 -1.85695335e-01
-1.14107966e+00 7.74535090e-02 4.55731481e-01 8.29441547e-01
4.04479504e-01 7.90399075e-01 -7.73835182e-02 1.15078211e+00
3.45371991e-01 9.23051417e-01 5.90867281e-01 -8.82298291e-01
2.19872817e-02 -6.14228785e-01 -1.56913269e-02 -5.92436492e-01
-2.43987679e-01 -6.33842945e-01 -6.34996891e-01 1.46955207e-01
5.87961376e-02 -1.30828321e-01 -1.32298601e+00 1.84423423e+00
5.32044709e-01 6.97083116e-01 2.07611412e-01 1.09270716e+00
1.19925237e+00 6.76675856e-01 -4.89048846e-02 -2.40396306e-01
1.07780504e+00 -4.09075886e-01 -1.02438998e+00 1.01283938e-02
-2.95730270e-02 -1.22011161e+00 1.02916050e+00 5.90129435e-01
-1.09265351e+00 -1.03977394e+00 -1.11042535e+00 -1.09209567e-01
-8.37401152e-02 -2.12568238e-01 3.35783273e-01 1.46584845e+00
-1.43126535e+00 2.65210152e-01 -3.56472313e-01 -2.18451291e-01
-2.61197060e-01 2.87759900e-01 -4.28136170e-01 4.24581729e-02
-1.40076458e+00 8.49171102e-01 -2.33826097e-02 2.88159754e-02
-9.61633265e-01 -8.58074725e-01 -1.08941758e+00 -2.31842175e-01
-4.12026607e-02 -6.51419163e-01 1.25925469e+00 -8.43692303e-01
-2.26291132e+00 5.11453688e-01 1.70864500e-02 -6.04893923e-01
4.18028772e-01 -3.37721884e-01 -8.40562105e-01 1.06045596e-01
-3.57986927e-01 3.71268988e-01 1.45888615e+00 -1.50963140e+00
-3.29678684e-01 -1.77311748e-01 -3.27092350e-01 3.50965858e-01
2.79460102e-01 3.83723900e-02 2.25851871e-02 -7.88738787e-01
9.93979126e-02 -8.84607494e-01 2.90100705e-02 -6.93374455e-01
-1.56405047e-01 4.74263787e-01 7.95043170e-01 -1.06181264e+00
8.04050207e-01 -2.03138208e+00 -2.24181917e-02 2.27523416e-01
-1.00869998e-01 3.58946621e-01 2.19302494e-02 2.12266490e-01
-4.20710027e-01 -1.56624138e-01 3.14364098e-02 -6.10524714e-01
-1.12410970e-02 9.40520614e-02 -3.98660034e-01 7.36644030e-01
-3.69008422e-01 5.29788554e-01 -7.21904159e-01 -2.69527406e-01
6.62477612e-01 1.36950839e+00 -5.23636043e-01 5.14426351e-01
3.67719233e-01 9.62280095e-01 1.70306444e-01 3.78692627e-01
9.25743103e-01 8.74835670e-01 -2.08087519e-01 -5.42963862e-01
-1.61087830e-02 9.83269811e-02 -1.20501721e+00 1.77062118e+00
-1.08430040e+00 7.88278520e-01 1.23399265e-01 -3.87820154e-01
1.29160893e+00 6.57192409e-01 3.70750159e-01 -7.92113721e-01
1.12267844e-02 4.22101349e-01 -1.65417418e-01 -2.53919512e-01
6.14692688e-01 -4.71092135e-01 5.68566471e-02 7.23753944e-02
3.87854815e-01 -5.98941624e-01 -7.95328736e-01 -1.51248619e-01
8.13188076e-01 4.53205705e-01 1.88614726e-01 2.05252498e-01
5.42138755e-01 -8.18069041e-01 3.68104190e-01 5.24439573e-01
-1.76991910e-01 1.12259173e+00 -3.71671468e-01 -1.22295525e-02
-1.26939631e+00 -1.08474004e+00 -1.70138687e-01 9.17506337e-01
-2.57366925e-01 4.97400761e-04 -9.11183000e-01 -1.72084033e-01
-3.24924886e-01 1.11552310e+00 -5.28158844e-01 -2.12414503e-01
-7.41938770e-01 -1.77023336e-01 7.20575452e-01 2.89398223e-01
2.91639924e-01 -9.76218224e-01 -3.53376269e-01 6.07928559e-02
-3.77589524e-01 -1.26912022e+00 -3.90783966e-01 9.35734734e-02
-4.09695446e-01 -2.81820476e-01 -1.00752759e+00 -3.75614792e-01
1.06029995e-01 2.09669009e-01 9.09395814e-01 -7.26174951e-01
-2.49570847e-01 7.35865295e-01 -5.80766201e-01 -6.50935709e-01
-5.20336032e-01 -2.33917981e-01 2.46608049e-01 2.28403673e-01
-1.58181041e-01 -9.05968964e-01 -6.50035083e-01 1.18955687e-01
-8.84804606e-01 -3.75143975e-01 4.84258235e-01 6.25743091e-01
6.76632226e-01 1.80078093e-02 8.60889852e-01 -5.26619196e-01
5.77724695e-01 -2.02073500e-01 -4.49576586e-01 -2.73283660e-01
-2.68698215e-01 -2.28658319e-01 6.53008282e-01 -5.33789694e-01
-1.57599545e+00 2.26458814e-02 -7.01717973e-01 -7.31376410e-01
-4.40006286e-01 -9.85420942e-02 -3.27443719e-01 -4.16782916e-01
8.93303990e-01 1.97031409e-01 -2.41778567e-01 -4.24066573e-01
5.51547766e-01 9.52077270e-01 8.88571143e-01 -2.16817424e-01
1.13301194e+00 4.14551497e-01 -5.04468568e-02 -1.23244476e+00
-3.17107379e-01 -4.87645596e-01 -2.80636370e-01 -4.01326150e-01
7.17999041e-01 -1.07118690e+00 -3.04527074e-01 1.81903139e-01
-8.57905626e-01 -2.82337606e-01 -6.53794169e-01 8.82577240e-01
-8.14280927e-01 2.85834908e-01 -3.18363398e-01 -1.24205458e+00
-4.25806969e-01 -1.00783432e+00 1.30604744e+00 3.99426639e-01
3.58321778e-02 -6.27700269e-01 4.66333419e-01 5.00482440e-01
8.12289774e-01 4.98399854e-01 1.23741679e-01 -2.22600549e-01
-2.04116389e-01 -2.14006662e-01 2.35248491e-01 4.05933291e-01
3.64087597e-02 -1.39438316e-01 -1.84083152e+00 -2.87577927e-01
3.25198352e-01 -1.98318332e-01 4.76668566e-01 8.15821528e-01
7.26087511e-01 -7.67004639e-02 1.41350254e-01 9.49100256e-01
1.43289280e+00 2.87277848e-01 1.34882438e+00 1.53896898e-01
6.68615997e-01 4.15385157e-01 6.74252272e-01 1.53098449e-01
1.51668921e-01 1.02661014e+00 4.49800253e-01 -4.13974762e-01
-6.46805286e-01 -3.67356449e-01 4.76833969e-01 6.22333586e-01
-3.61820698e-01 -1.69345528e-01 -2.68889606e-01 4.32403564e-01
-1.21037114e+00 -1.06526196e+00 8.82234350e-02 2.60863733e+00
6.89099193e-01 -2.60148466e-01 1.06560513e-01 4.33174223e-01
7.34195411e-01 1.58392891e-01 -2.39767849e-01 -8.92769635e-01
-8.67270455e-02 9.63828206e-01 6.06434464e-01 7.25340068e-01
-8.45905006e-01 7.95191705e-01 6.01552534e+00 9.04322565e-01
-1.36388612e+00 5.37452519e-01 1.23199120e-01 8.12991560e-02
-4.51394141e-01 -5.11845589e-01 -5.10431111e-01 6.35132790e-02
1.43535554e+00 1.28678039e-01 6.16812944e-01 7.03088284e-01
3.13924313e-01 1.14915341e-01 -6.28221154e-01 1.16907561e+00
1.90065831e-01 -5.03253758e-01 -4.58430439e-01 2.08852947e-01
4.94139999e-01 -4.71360087e-02 6.26743138e-01 3.78029168e-01
2.45867178e-01 -1.34837306e+00 6.39320910e-01 6.35325670e-01
1.35908914e+00 -1.11555469e+00 7.64520168e-01 -1.41221061e-01
-1.38879454e+00 2.03012973e-01 -3.35901469e-01 6.11250639e-01
4.13310707e-01 3.08536589e-01 -1.19820499e+00 7.40784466e-01
7.86263227e-01 -8.84740427e-02 7.33277053e-02 1.03099203e+00
-1.30694628e-01 6.01008654e-01 -4.13443476e-01 5.71307063e-01
-2.22150847e-01 -3.35281491e-02 9.09765601e-01 1.27155781e+00
6.34074092e-01 3.36295925e-02 -3.65203321e-01 6.69155121e-01
2.29606003e-01 1.94638938e-01 -8.48320782e-01 2.98967659e-01
4.30524588e-01 1.34989834e+00 -3.01849544e-01 2.86591023e-01
-5.34142666e-02 1.04853690e+00 -4.64512259e-01 4.92961615e-01
-1.20252383e+00 -5.58081567e-01 7.67871320e-01 1.33321136e-01
3.20647478e-01 -1.25744700e-01 1.80513978e-01 -8.41821611e-01
-3.38205308e-01 -9.56584871e-01 -5.05405739e-02 -1.13661838e+00
-7.89735913e-01 1.16704082e+00 -1.48279190e-01 -1.29467988e+00
-6.30893767e-01 -2.07465520e-04 -1.51583284e-01 1.18262732e+00
-1.65668678e+00 -1.46925664e+00 -4.99437243e-01 7.46234596e-01
5.59811831e-01 9.75950807e-03 1.03726554e+00 6.78254724e-01
1.87724270e-02 9.37763751e-01 -1.86361119e-01 -3.94494861e-01
1.05309033e+00 -1.27679813e+00 3.78857136e-01 8.60746562e-01
1.84813052e-01 4.64246869e-01 1.29889774e+00 -4.68774021e-01
-1.31330872e+00 -1.08166456e+00 3.43487978e-01 -1.90076590e-01
1.03862278e-01 -4.72233295e-01 -8.71608138e-01 5.73617816e-01
3.60668391e-01 1.43827677e-01 7.52688885e-01 -3.08228821e-01
-3.97244036e-01 -3.01292211e-01 -1.70098531e+00 3.06362897e-01
7.96612978e-01 -7.39182115e-01 -4.53874320e-01 1.04411077e-02
9.35364008e-01 -6.98719859e-01 -1.17769837e+00 5.23796618e-01
7.19365239e-01 -1.23214459e+00 1.27371907e+00 1.66935489e-01
-2.18471482e-01 -4.63974714e-01 -4.61343884e-01 -1.67566228e+00
1.77431162e-02 -1.19798398e+00 -2.08313800e-02 1.27896214e+00
1.43415749e-01 -8.02258492e-01 3.51552367e-01 1.76130846e-01
-4.17552799e-01 5.17849252e-02 -1.02816272e+00 -7.40159094e-01
-2.98473597e-01 -6.77814722e-01 7.28571415e-01 8.32199693e-01
-5.85111260e-01 2.76634067e-01 -7.79556870e-01 4.91670072e-01
7.66348064e-01 -4.74300146e-01 9.97855842e-01 -1.06490457e+00
-3.61024886e-01 1.31002665e-01 -6.46941841e-01 -5.13443828e-01
-7.10418075e-02 -2.62922525e-01 1.99460715e-01 -1.21052206e+00
-5.68515241e-01 -3.66545260e-01 -1.15355454e-01 -2.06273630e-01
2.10593209e-01 5.48424602e-01 5.78231812e-02 -2.45845333e-01
1.83168054e-01 5.67278087e-01 1.18401217e+00 4.38237488e-01
-3.81654322e-01 2.06371188e-01 -4.33083177e-01 5.32563090e-01
5.65740883e-01 -2.70625025e-01 -6.71593785e-01 8.12441409e-02
3.22393775e-02 3.12628061e-01 3.03377450e-01 -1.35966289e+00
4.95693423e-02 3.61119241e-01 3.78460646e-01 -4.29330945e-01
9.59523797e-01 -9.19365108e-01 8.85874510e-01 -2.09741518e-02
-2.41122514e-01 -3.61057043e-01 2.50653505e-01 4.97398406e-01
-2.00048447e-01 6.70222268e-02 1.05028677e+00 4.23007682e-02
-2.51224697e-01 4.96509373e-02 -3.62750702e-02 -2.99601316e-01
8.00756276e-01 -4.15767968e-01 -1.16101101e-01 -1.09255970e+00
-7.31192470e-01 -8.46069872e-01 3.16857874e-01 3.33792716e-01
8.69432092e-01 -1.24181223e+00 -6.97219849e-01 3.04876894e-01
-1.45849109e-01 -3.06982666e-01 6.14008129e-01 6.66502297e-01
-4.51699674e-01 3.80061865e-01 -2.95256197e-01 -4.53097820e-01
-1.38133442e+00 3.59293729e-01 6.89298689e-01 4.38741222e-02
-7.46608794e-01 1.01463294e+00 1.91750631e-01 -4.88403797e-01
3.17013681e-01 -2.00460985e-01 -6.12334728e-01 -1.69750988e-01
5.92264473e-01 3.20414633e-01 2.73242563e-01 -1.24724698e+00
3.28692645e-02 7.46759534e-01 3.96867007e-01 -6.84784234e-01
1.30355430e+00 -3.91272783e-01 4.05254334e-01 2.85507858e-01
1.25464642e+00 6.48514569e-01 -9.63648736e-01 -1.12770215e-01
-5.96634388e-01 -7.27760613e-01 5.87762713e-01 -5.58899999e-01
-1.24237430e+00 7.05269456e-01 1.10483861e+00 2.82465786e-01
1.46966207e+00 -3.73815626e-01 8.74025822e-01 -3.34431469e-01
6.94570959e-01 -8.60774577e-01 -7.52316415e-02 -3.73948440e-02
1.05426669e+00 -7.46717930e-01 -3.04344594e-01 -4.67418313e-01
-7.75705934e-01 8.53290796e-01 2.51602471e-01 -3.09726018e-02
4.97311622e-01 4.17496353e-01 3.29430014e-01 8.05663317e-02
-2.19673678e-01 -5.05530834e-01 4.26669955e-01 1.20359135e+00
4.68433678e-01 -5.82302595e-03 6.22164868e-02 1.92275047e-01
-7.91594386e-01 -2.44457513e-01 5.06124198e-01 3.71354640e-01
-1.86743185e-01 -8.91109169e-01 -8.27651978e-01 -1.34111226e-01
-9.21635509e-01 -1.62561163e-01 -2.73105204e-02 6.03351653e-01
1.30225956e-01 1.11629224e+00 -3.28930885e-01 -6.52326882e-01
4.39408481e-01 -4.77054156e-02 7.40359545e-01 -3.35794032e-01
-7.78935969e-01 6.98484302e-01 7.07072318e-02 -7.61061192e-01
-5.47255933e-01 -5.23117900e-01 -8.23849916e-01 -1.38297811e-01
-5.10563612e-01 1.70666203e-02 1.25382888e+00 4.00875717e-01
1.22939616e-01 9.67019081e-01 7.59796619e-01 -1.06718886e+00
-1.70249283e-01 -1.06640863e+00 -7.87315965e-01 8.92110169e-02
6.82482719e-01 -5.56584179e-01 -7.10372448e-01 5.22760078e-02] | [15.34433364868164, 6.036479949951172] |
b2df7c36-6c7f-4658-8874-253b22f0d77c | openhls-high-level-synthesis-for-low-latency | 2302.06751 | null | https://arxiv.org/abs/2302.06751v4 | https://arxiv.org/pdf/2302.06751v4.pdf | OpenHLS: High-Level Synthesis for Low-Latency Deep Neural Networks for Experimental Science | In many experiment-driven scientific domains, such as high-energy physics, material science, and cosmology, high data rate experiments impose hard constraints on data acquisition systems: collected data must either be indiscriminately stored for post-processing and analysis, thereby necessitating large storage capacity, or accurately filtered in real-time, thereby necessitating low-latency processing. Deep neural networks, effective in other filtering tasks, have not been widely employed in such data acquisition systems, due to design and deployment difficulties. We present an open source, lightweight, compiler framework, without any proprietary dependencies, OpenHLS, based on high-level synthesis techniques, for translating high-level representations of deep neural networks to low-level representations, suitable for deployment to near-sensor devices such as field-programmable gate arrays. We evaluate OpenHLS on various workloads and present a case-study implementation of a deep neural network for Bragg peak detection in the context of high-energy diffraction microscopy. We show OpenHLS is able to produce an implementation of the network with a throughput 4.8 $\mu$s/sample, which is approximately a 4$\times$ improvement over the existing implementation | ['Kazutomo Yoshii', 'Ryan Chard', 'Ian Foster', 'Kyle Chard', 'Arham Khan', 'Maksim Levental'] | 2023-02-13 | null | null | null | null | ['low-latency-processing'] | ['robots'] | [ 2.76776105e-01 3.49637344e-02 1.79238409e-01 -5.56851804e-01
-3.65384281e-01 1.29525606e-02 2.05607206e-01 7.01714873e-01
-9.46983337e-01 5.73018610e-01 -2.28320032e-01 -5.69551587e-01
2.95624137e-02 -1.13260627e+00 -8.76723707e-01 -6.83735907e-01
-1.67870864e-01 7.06701696e-01 5.06575286e-01 7.78463297e-03
1.94335654e-01 9.38217044e-01 -1.84549022e+00 3.40571582e-01
-1.27125412e-01 1.27806532e+00 2.32068628e-01 6.67072952e-01
-8.27187323e-04 5.93658149e-01 -4.34706956e-01 3.22509855e-01
1.74251154e-01 1.55334529e-02 -6.16633594e-01 -6.84998333e-01
3.31268489e-01 -5.82580864e-01 -3.63470137e-01 9.01093066e-01
7.91137457e-01 -1.83134526e-01 1.89078525e-01 -6.64519727e-01
-9.65514854e-02 6.25467300e-01 -3.04287791e-01 4.58085835e-01
-1.69738263e-01 5.32470584e-01 5.34284055e-01 -6.65418983e-01
7.99708605e-01 1.04673147e+00 5.62776208e-01 6.13496542e-01
-1.48509753e+00 -5.90809882e-01 -7.02584565e-01 2.01737043e-02
-1.07128096e+00 -9.79050577e-01 3.74193698e-01 -1.74261495e-01
1.73811305e+00 1.04717046e-01 4.84751880e-01 7.82950938e-01
6.82402253e-01 1.87060684e-01 1.29241586e+00 -4.23589706e-01
6.75668061e-01 -4.43769783e-01 4.22264934e-01 6.90357029e-01
6.19212389e-01 1.16566643e-01 -8.85914207e-01 -6.02033809e-02
9.91126299e-01 -1.59963235e-01 -8.35989490e-02 3.97196785e-02
-1.25270736e+00 4.31542993e-01 3.71721476e-01 2.95476824e-01
-5.19041181e-01 6.07161403e-01 7.10407615e-01 3.44111562e-01
3.89923275e-01 5.11855483e-01 -5.27304828e-01 -1.45559788e-01
-1.11243713e+00 3.21754396e-01 7.83903003e-01 8.09464514e-01
8.60637546e-01 6.21889941e-02 -1.71923056e-01 2.31205598e-01
2.44690150e-01 4.87825692e-01 3.88890237e-01 -8.79262745e-01
-2.38258764e-01 5.73236704e-01 -1.65726528e-01 -4.47960287e-01
-1.01033974e+00 -1.09513387e-01 -1.19550216e+00 4.86765414e-01
2.55857170e-01 -1.66767649e-02 -1.01042402e+00 1.53630316e+00
2.53154695e-01 -1.62669867e-01 1.01944573e-01 9.47692335e-01
9.41937864e-01 7.45322645e-01 -1.98012195e-03 -3.72546285e-01
1.53876770e+00 -3.92142922e-01 -4.96683300e-01 -8.96179602e-02
4.28467304e-01 -6.66371107e-01 9.17810380e-01 5.30158877e-01
-1.45374990e+00 -4.87518191e-01 -1.42006457e+00 -5.64177811e-01
-5.39869845e-01 -2.60341316e-01 7.12252438e-01 4.77181226e-01
-1.27980196e+00 1.03586090e+00 -1.40799916e+00 -3.27189714e-01
3.29773575e-01 9.63072777e-01 -7.71375895e-02 3.45702142e-01
-6.57373607e-01 2.76404887e-01 3.55371386e-01 -9.64695141e-02
-6.85217559e-01 -1.02348125e+00 -4.88323301e-01 3.05920392e-01
-1.14257477e-01 -4.87062037e-01 1.30505335e+00 -6.55553713e-02
-1.55645680e+00 9.27773476e-01 3.59135345e-02 -8.03813279e-01
-2.08324239e-01 2.11932048e-01 -4.17098254e-01 -3.27762566e-03
-3.79892498e-01 6.54240847e-01 4.01592612e-01 -4.99299258e-01
-3.69546056e-01 -5.05009592e-01 -1.75131261e-01 -4.74492759e-01
-4.14967030e-01 1.41974419e-01 -1.75465524e-01 -7.68254921e-02
4.39602472e-02 -6.41749799e-01 -3.01386356e-01 1.65565476e-01
-9.15136412e-02 -2.49602161e-02 8.86539996e-01 -1.11738913e-01
7.84472108e-01 -2.03473854e+00 -2.51502335e-01 1.68715537e-01
5.20854056e-01 3.08315665e-01 3.11476141e-01 1.12918846e-01
8.51752460e-02 -2.88054943e-01 -7.94629604e-02 -1.75302088e-01
-1.42736465e-01 2.09701508e-01 9.24206674e-02 6.17845356e-01
1.42922208e-01 5.79734445e-01 -6.43345237e-01 -1.18053094e-01
2.58552134e-01 4.66837943e-01 -6.85942411e-01 2.69365579e-01
-5.27404249e-01 4.70099645e-03 -1.79224178e-01 6.57501400e-01
6.09086633e-01 -6.65632904e-01 3.01331848e-01 -5.73675513e-01
-6.67237699e-01 5.03422678e-01 -8.69440973e-01 1.87920177e+00
-3.37860674e-01 8.87281537e-01 5.93328714e-01 -9.20874178e-01
8.70946169e-01 -2.79194321e-02 5.94708741e-01 -1.15056396e+00
4.57420230e-01 4.14010733e-01 3.21997926e-02 -3.70628625e-01
8.86429012e-01 -5.94589151e-02 9.97867882e-02 6.60796523e-01
-7.23440340e-03 -8.19060653e-02 4.27211821e-01 -1.07436441e-01
1.54831660e+00 2.48283613e-02 -8.86055529e-02 -8.91406357e-01
1.64969057e-01 9.80856940e-02 2.05352485e-01 6.68377340e-01
-8.18993524e-02 3.34566742e-01 3.70246410e-01 -7.87758827e-01
-1.61444068e+00 -7.75165796e-01 -4.65802193e-01 1.05196393e+00
-3.73078763e-01 -4.05801415e-01 -9.18417335e-01 1.37504071e-01
-1.71864823e-01 1.21207997e-01 -2.82788966e-02 -3.16487104e-02
-6.41637921e-01 -1.12358367e+00 4.96152997e-01 4.12513286e-01
4.48772281e-01 -1.11461592e+00 -1.56497002e+00 7.65911222e-01
8.08934927e-01 -1.11669195e+00 1.78845719e-01 9.98404682e-01
-8.76658082e-01 -7.73904026e-01 -6.23172941e-03 -4.89370346e-01
6.19635880e-01 -8.52942765e-02 1.24467528e+00 7.37820659e-03
-8.44805956e-01 -2.58999318e-01 2.33228907e-01 -3.87821376e-01
-5.59368491e-01 1.64476410e-01 3.15017790e-01 -6.36153400e-01
8.62753570e-01 -8.06923270e-01 -9.67545450e-01 -1.72725543e-01
-1.04842341e+00 1.24223612e-01 6.41051233e-01 9.41534460e-01
8.42156589e-01 -2.96824500e-02 1.72530875e-01 -8.68994772e-01
4.06815022e-01 -2.22871944e-01 -1.09176600e+00 -2.92133719e-01
-8.49921167e-01 5.20141363e-01 9.44853187e-01 -1.56782735e-02
-5.39480567e-01 -1.84794385e-02 -7.21475303e-01 -3.77880991e-01
-2.42919207e-01 3.72990727e-01 1.65930554e-01 -2.77030945e-01
8.87649298e-01 -1.93115491e-02 3.64298642e-01 -3.63802940e-01
-1.92027256e-01 7.91168571e-01 7.49125719e-01 -8.30250025e-01
-6.43307492e-02 4.33739662e-01 5.42989373e-01 -9.87208664e-01
-1.88967675e-01 -2.87069738e-01 -2.46940747e-01 -9.07721445e-02
7.52659619e-01 -8.81696463e-01 -1.24311864e+00 5.99840105e-01
-1.10336530e+00 -4.48772490e-01 -5.29388487e-01 2.47082800e-01
-4.74156052e-01 -1.05992146e-01 -9.84087884e-01 -4.23810929e-01
-8.68023455e-01 -1.56548703e+00 1.18244886e+00 3.90104592e-01
-1.23249069e-01 -4.85168397e-01 -2.18497172e-01 -1.72688156e-01
9.93815005e-01 2.74164200e-01 1.16137254e+00 -5.18513381e-01
-9.15002048e-01 1.26677260e-01 -4.42715973e-01 1.89840659e-01
-3.70736063e-01 -1.53055657e-02 -1.23990738e+00 -4.73818839e-01
1.59222469e-01 -6.52947366e-01 7.52952814e-01 3.88561159e-01
1.58502233e+00 3.33973877e-02 -3.31175715e-01 7.08250284e-01
1.59803748e+00 -1.00863703e-01 7.40944266e-01 -1.58836052e-03
4.25241649e-01 8.53054747e-02 -1.83931664e-01 5.98357737e-01
-1.36631519e-01 4.96647418e-01 5.26294231e-01 -4.72919531e-02
-8.58548656e-02 3.66127700e-01 2.08287895e-01 1.02740443e+00
7.70105049e-02 -1.77854955e-01 -9.65557992e-01 2.57547408e-01
-1.43341601e+00 -5.84078610e-01 -3.90424460e-01 2.17866683e+00
9.22023237e-01 4.76530731e-01 -1.66064277e-01 1.30205944e-01
2.54908204e-01 1.11533448e-01 -6.83104336e-01 -8.06262314e-01
2.92821705e-01 1.15673423e+00 1.07208657e+00 7.27776065e-02
-6.20232761e-01 5.80629230e-01 6.30894566e+00 7.68954575e-01
-1.58520091e+00 2.37980708e-01 7.29920805e-01 -5.68366706e-01
2.24601524e-03 -3.92840892e-01 -1.00180304e+00 5.08387446e-01
1.90250444e+00 6.57504722e-02 3.40413034e-01 9.34452415e-01
1.10334657e-01 -1.30582482e-01 -1.54593968e+00 9.27336991e-01
-6.01543069e-01 -2.04106736e+00 -3.12062949e-01 2.13536963e-01
1.41332835e-01 4.84988421e-01 -2.89936423e-01 3.50576974e-02
2.57949263e-01 -9.16424036e-01 4.80444074e-01 2.63184428e-01
1.21854007e+00 -8.37282956e-01 6.37458205e-01 7.90229887e-02
-6.70427680e-01 4.86638606e-01 -6.88260734e-01 -3.49250019e-01
-7.27603585e-02 1.18751276e+00 -8.11318219e-01 6.62834346e-02
9.30264890e-01 1.50433034e-01 -1.30515844e-01 3.28289092e-01
6.98511660e-01 4.68015581e-01 -8.07246506e-01 -3.83881927e-01
5.76602444e-02 1.77935034e-01 8.97948295e-02 1.40189254e+00
3.79613549e-01 1.86220929e-01 -5.27748503e-02 9.34905291e-01
-4.37031180e-01 -2.63017684e-01 -2.83981860e-01 -1.37072176e-01
3.76166016e-01 1.41661608e+00 -1.02191913e+00 -3.10855746e-01
-4.55845147e-01 4.73669201e-01 3.23009551e-01 -2.93896705e-01
-4.11546588e-01 -4.61874783e-01 8.84379148e-01 5.54129004e-01
3.67458202e-02 -4.12563741e-01 -3.65140736e-01 -7.19122171e-01
-1.70491919e-01 -6.64505303e-01 -5.72927073e-02 -4.78885442e-01
-1.00529480e+00 4.94165838e-01 -2.19864085e-01 -4.35614854e-01
-2.04841241e-01 -1.28674185e+00 -6.28501475e-02 8.51563871e-01
-1.23194909e+00 -5.04254878e-01 -3.07723343e-01 3.20818871e-01
1.81672275e-01 -1.63777098e-01 1.02159667e+00 6.39291048e-01
-5.76959252e-01 2.11209729e-01 2.05202252e-01 -2.04795510e-01
4.19391423e-01 -1.09867167e+00 8.85359645e-01 6.92854941e-01
-4.48471427e-01 7.18551278e-01 8.55647326e-01 -3.75549316e-01
-2.28482819e+00 -9.52753603e-01 4.98078734e-01 2.80500144e-01
7.37338126e-01 -7.38596261e-01 -1.01469254e+00 3.46879393e-01
3.35904092e-01 3.69170815e-01 6.04560494e-01 -7.42466003e-02
4.78502475e-02 -3.25652570e-01 -1.38029754e+00 2.57413238e-01
9.47390497e-01 -5.24996161e-01 -9.01570544e-03 5.46045184e-01
6.31419897e-01 -8.23639512e-01 -1.18686378e+00 5.11612058e-01
4.23081279e-01 -1.15830791e+00 8.52513909e-01 -1.66847676e-01
6.50019586e-01 -9.17470306e-02 -8.66650268e-02 -6.39817834e-01
-3.43220145e-01 -5.18913031e-01 -3.15938830e-01 6.42630637e-01
8.94001648e-02 -5.45986831e-01 1.07388306e+00 5.49313664e-01
-4.40404624e-01 -8.18435729e-01 -1.16135919e+00 -4.80127960e-01
-8.39016140e-02 -4.03852314e-01 7.10616291e-01 5.33795476e-01
-2.07357824e-01 3.21724087e-01 9.41124111e-02 1.43521547e-01
7.17413604e-01 4.60073678e-03 4.73294646e-01 -1.11129534e+00
-4.18714523e-01 -4.81189251e-01 -6.00161850e-01 -1.00727665e+00
-1.40357181e-01 -6.29825592e-01 3.76056492e-01 -1.09625018e+00
-4.21080813e-02 -5.28536499e-01 -3.00115556e-01 4.28497255e-01
6.80603564e-01 4.36107308e-01 -3.04360777e-01 6.77656755e-02
-3.10969621e-01 7.64028227e-04 6.43716276e-01 2.80216169e-02
2.03862280e-01 -8.09329450e-01 -2.45254815e-01 5.46994448e-01
3.94392341e-01 -4.08145070e-01 -1.20725550e-01 -5.08529246e-01
4.76546019e-01 -9.23970528e-03 4.19320792e-01 -1.75157297e+00
5.92336774e-01 -6.79002106e-02 5.08404016e-01 -6.23931468e-01
2.52894670e-01 -7.14103937e-01 3.64141077e-01 6.24950290e-01
-1.72497272e-01 1.36041671e-01 4.74780053e-01 -8.18607956e-02
7.02742562e-02 -1.79800645e-01 1.06483889e+00 -2.86968887e-01
-7.49249399e-01 3.42123210e-01 -3.39136571e-01 -2.76960969e-01
7.33049989e-01 1.23611994e-01 -6.66617453e-01 5.31370521e-01
-4.91855949e-01 -1.24764062e-01 7.54085004e-01 -1.89676791e-01
4.91071552e-01 -9.04638171e-01 -4.82961923e-01 7.85557449e-01
-9.82516780e-02 5.67725480e-01 2.96285242e-01 3.78988892e-01
-1.08750331e+00 4.85177308e-01 -5.30838072e-01 -7.77568638e-01
-8.17890406e-01 5.32641470e-01 2.69107878e-01 -2.82063872e-01
-7.72998273e-01 7.68590152e-01 -2.20418274e-01 4.21537608e-02
-1.49110571e-01 -8.65560174e-01 4.81431872e-01 -2.29885280e-01
7.11762965e-01 1.49477676e-01 8.99496496e-01 1.41354557e-02
-3.20316374e-01 8.92976224e-02 -2.40960538e-01 1.79574430e-01
1.61610723e+00 3.40511590e-01 -5.33284903e-01 3.29496562e-01
1.17117298e+00 -5.08105755e-01 -8.94248307e-01 -1.03949502e-01
-2.02431846e-02 1.76787406e-01 7.15265632e-01 -3.81327152e-01
-1.13253474e+00 8.33456218e-01 9.27582026e-01 3.21796477e-01
1.28055620e+00 -2.30784833e-01 1.03767741e+00 8.18907142e-01
7.54492939e-01 -1.24631298e+00 -3.48523617e-01 6.08019173e-01
2.79035181e-01 -8.65449667e-01 2.66464561e-01 -2.21853494e-03
5.97675443e-01 1.38243103e+00 5.55871844e-01 -1.49135053e-01
6.69454455e-01 1.32408333e+00 -3.62346917e-01 -7.02060580e-01
-9.38519716e-01 2.06029251e-01 -4.05054063e-01 2.24514470e-01
6.81702733e-01 1.04464591e-02 -1.74957052e-01 4.97838520e-02
-1.03036754e-01 4.67668384e-01 7.06679642e-01 1.34253085e+00
-7.67230213e-01 -1.10047722e+00 -1.97166383e-01 7.87582278e-01
-5.92379034e-01 -2.04658523e-01 2.06004381e-01 4.54839259e-01
1.06734738e-01 2.54033953e-01 8.30684602e-01 -1.73234463e-01
2.57694870e-01 8.08502510e-02 5.15193641e-01 -5.50175428e-01
-6.74507499e-01 -7.34395459e-02 9.75011215e-02 -7.88624763e-01
-5.70902288e-01 -3.09386551e-01 -1.62275112e+00 -9.15301383e-01
1.78318441e-01 -1.95482433e-01 1.16627729e+00 7.24959552e-01
7.44671702e-01 8.68251920e-01 9.83390138e-02 -1.12344325e+00
-2.63351083e-01 -7.17919886e-01 -4.28153366e-01 7.67785460e-02
3.67059439e-01 -3.37613285e-01 7.83982202e-02 -2.48664189e-02] | [8.334893226623535, 2.9283699989318848] |
9c87021d-64e8-493c-b2c3-e25027d381e4 | a-factor-based-framework-for-decision-making | 2203.11981 | null | https://arxiv.org/abs/2203.11981v1 | https://arxiv.org/pdf/2203.11981v1.pdf | A Factor-Based Framework for Decision-Making Competency Self-Assessment | We summarize our efforts to date in developing a framework for generating succinct human-understandable competency self-assessments in terms of machine self confidence, i.e. a robot's self-trust in its functional abilities to accomplish assigned tasks. Whereas early work explored machine self-confidence in ad hoc ways for niche applications, our Factorized Machine Self-Confidence framework introduces and combines several aspects of probabilistic meta reasoning for algorithmic planning and decision-making under uncertainty to arrive at a novel set of generalizable self-confidence factors, which can support competency assessment for a wide variety of problems. | ['Nisar Ahmed', 'Brett W. Israelsen'] | 2022-03-22 | null | null | null | null | ['decision-making-under-uncertainty', 'decision-making-under-uncertainty'] | ['medical', 'reasoning'] | [ 6.15918748e-02 1.21327531e+00 -3.31245661e-01 -7.64029086e-01
-7.28538454e-01 -4.34294462e-01 6.62974358e-01 5.64261913e-01
-3.63250703e-01 9.20418859e-01 3.58107150e-01 -3.63235146e-01
-6.78079963e-01 -6.29465580e-01 -7.56290615e-01 1.44461319e-02
9.33833420e-02 9.53752279e-01 1.67094484e-01 -9.04617179e-03
5.13423383e-01 7.10312873e-02 -1.71864843e+00 -1.59687743e-01
1.46838355e+00 8.83485794e-01 9.11852270e-02 7.52022743e-01
5.75331986e-01 1.28083229e+00 -2.31774762e-01 -6.58836424e-01
-1.37027085e-01 5.55002131e-02 -1.18863559e+00 -2.68928021e-01
3.88664678e-02 -2.90787131e-01 -1.60550222e-01 1.28748393e+00
-1.41140059e-01 1.34283304e-01 1.15967858e+00 -1.59655380e+00
-3.98064673e-01 1.12530792e+00 2.08539665e-01 4.26329821e-02
9.05019641e-01 2.78476298e-01 9.13986325e-01 -2.80542016e-01
1.94623604e-01 1.37980747e+00 7.14484632e-01 5.69269240e-01
-9.40215945e-01 -4.61079538e-01 1.66949153e-01 -5.42303659e-02
-1.12694025e+00 -2.68860400e-01 1.27071425e-01 -5.85950792e-01
8.97424757e-01 -4.51012626e-02 5.86376786e-01 1.11963809e+00
7.54232466e-01 6.44083440e-01 1.57820511e+00 -4.00668353e-01
7.97001243e-01 2.99886584e-01 1.59553602e-01 8.88313293e-01
7.21088290e-01 6.20019078e-01 -7.95971513e-01 -1.80424571e-01
8.99230182e-01 -5.26962757e-01 5.32640815e-01 -5.30290544e-01
-1.17106414e+00 6.42186224e-01 1.41060203e-01 -7.28046298e-02
-4.30025578e-01 5.66072941e-01 -8.17578882e-02 2.18974873e-01
2.92907566e-01 1.08215380e+00 -6.53488636e-01 -7.07781196e-01
-4.38082248e-01 4.57182050e-01 1.01303506e+00 1.10925317e+00
7.12520063e-01 -1.99462444e-01 -6.47381470e-02 2.58399583e-02
6.13038182e-01 2.20246196e-01 4.00117368e-01 -1.87081027e+00
1.52850077e-02 3.48651528e-01 5.39564073e-01 -9.05490279e-01
-5.06216705e-01 -4.76312548e-01 -1.04383919e-02 5.32183707e-01
-5.33424653e-02 -3.81142616e-01 -8.55840504e-01 1.97882318e+00
1.16572805e-01 2.82548778e-02 6.26709282e-01 7.59205520e-01
3.63253951e-01 1.07329376e-01 4.32291180e-01 1.21095173e-01
1.19520807e+00 -9.14728880e-01 -3.86556357e-01 -1.05204833e+00
3.96586001e-01 3.49598266e-02 9.05797482e-01 6.45794809e-01
-1.35834408e+00 -5.75460017e-01 -1.44156992e+00 3.15628499e-01
-2.75676418e-02 -5.36982536e-01 1.51400304e+00 9.54498172e-01
-1.26372933e+00 7.54861295e-01 -1.06182766e+00 -5.19010574e-02
3.51916850e-01 3.10412765e-01 -5.56602299e-01 5.27727325e-03
-1.22654593e+00 1.70563984e+00 7.10622668e-01 -1.82460636e-01
-1.35221207e+00 -2.66907454e-01 -1.28126824e+00 -2.33655497e-02
5.61812520e-01 -1.04730499e+00 1.91106308e+00 -8.93679798e-01
-1.58398783e+00 6.57628715e-01 1.78614989e-01 -6.44998610e-01
6.01770759e-01 -5.04210413e-01 -2.44834200e-01 -7.50637352e-02
3.25114816e-01 8.68240952e-01 6.30688906e-01 -1.09882641e+00
-9.30820107e-01 -3.55754375e-01 3.45774472e-01 7.70845532e-01
-1.13124065e-01 -2.77804792e-01 3.54536064e-02 -2.09755972e-01
4.39218909e-01 -1.09455764e+00 -7.03490198e-01 -4.07407045e-01
-6.19262084e-02 -3.80843937e-01 -4.02916938e-01 -6.77351607e-03
7.96365142e-01 -1.50745273e+00 2.42292613e-01 3.45531493e-01
4.76870924e-01 -6.23699009e-01 1.68303996e-01 1.51893467e-01
6.84100240e-02 1.36651814e-01 -1.83301661e-02 -1.57641783e-01
3.55504781e-01 2.78256148e-01 -7.15966448e-02 3.06428254e-01
4.33421105e-01 7.01845646e-01 -1.37101948e+00 -7.68957496e-01
2.50179559e-01 -2.55779713e-01 -6.30256176e-01 3.42080176e-01
-1.81001186e-01 1.37398541e-01 -8.27102184e-01 5.71494162e-01
7.20269606e-02 -1.04839034e-01 1.97016969e-01 5.51669657e-01
1.36253074e-01 3.84588897e-01 -1.13980591e+00 1.82958865e+00
-3.56638163e-01 2.05322921e-01 -2.45863140e-01 -3.09662133e-01
7.89746344e-01 2.21073851e-01 1.32811189e-01 -2.75563031e-01
1.61476642e-01 1.52902812e-01 -7.05915093e-02 -2.76367456e-01
9.53328371e-01 -4.53168809e-01 -7.09190071e-01 5.06346643e-01
3.58445615e-01 -9.83032525e-01 -1.55195624e-01 2.01830938e-01
1.35120475e+00 3.51229370e-01 4.67055768e-01 -6.94322050e-01
7.74830282e-02 4.21175301e-01 5.82841396e-01 1.18159163e+00
-5.90264738e-01 2.44244844e-01 2.34075382e-01 -2.95105994e-01
-8.41480792e-01 -1.35530710e+00 2.22378392e-02 1.05817425e+00
2.21856728e-01 -2.51365781e-01 -8.26698065e-01 -5.71746707e-01
8.81599784e-02 1.15520680e+00 -8.35250199e-01 -5.11369228e-01
2.39304423e-01 -2.77973302e-02 5.12454927e-01 9.99961019e-01
3.48968431e-02 -9.92254436e-01 -8.73535931e-01 1.44656554e-01
8.86439160e-02 -8.51950407e-01 5.61057106e-02 6.32878304e-01
-9.05836344e-01 -1.04598534e+00 -5.74574843e-02 -6.25812232e-01
6.49118483e-01 1.75364800e-02 1.48033869e+00 -2.54586995e-01
2.98047036e-01 9.63294208e-01 -2.23311111e-01 -8.43363583e-01
-6.48233831e-01 -2.34044537e-01 8.52208614e-01 -1.24264121e+00
2.83912152e-01 -3.44555944e-01 -3.32743764e-01 3.80039662e-01
-4.47636068e-01 -7.59100020e-02 7.90828168e-01 6.55349076e-01
2.91022450e-01 7.43445039e-01 4.50927138e-01 -8.89758885e-01
1.28038681e+00 -6.43636703e-01 -3.78564954e-01 2.92932987e-01
-1.38506854e+00 4.45134103e-01 -1.70133650e-01 -3.72152656e-01
-1.40344167e+00 1.22049503e-01 3.26552689e-01 -3.38010639e-02
-9.24830586e-02 6.83810174e-01 9.24294516e-02 -3.69384303e-03
8.96215856e-01 -1.20477453e-01 -8.72414373e-03 3.27855140e-01
6.82619989e-01 3.88325185e-01 1.06009567e+00 -1.42878866e+00
5.51569104e-01 -3.03017348e-02 -9.88935754e-02 1.14748314e-01
-1.12577760e+00 -6.14952035e-02 -2.86641419e-01 -3.21943372e-01
7.13238835e-01 -1.25835562e+00 -1.13166809e+00 1.13132775e-01
-5.34662783e-01 -5.21090508e-01 -3.87342989e-01 6.02522254e-01
-1.46560776e+00 2.16953471e-01 -5.70175111e-01 -1.34804511e+00
-6.48161992e-02 -1.16104019e+00 9.48979557e-01 4.98329103e-01
-9.25522625e-01 -8.76830578e-01 3.36985767e-01 3.97913009e-01
5.72540089e-02 -5.24477139e-02 6.60976887e-01 -6.39099479e-01
-1.55376658e-01 -1.66136160e-01 2.13619888e-01 2.89966583e-01
-4.90207702e-01 -1.55570090e-01 -8.87980223e-01 -9.71187837e-03
2.23457754e-01 -1.10141635e+00 3.08343500e-01 3.29818666e-01
8.80847216e-01 -4.02108401e-01 -2.90257096e-01 -5.24174459e-02
9.54969764e-01 2.29187328e-02 6.75959527e-01 4.35005337e-01
-2.03375742e-01 8.20003986e-01 1.44276059e+00 5.18611014e-01
1.04518485e+00 -1.60437614e-01 5.33216596e-01 3.76031369e-01
8.68004799e-01 -8.19422185e-01 1.98636144e-01 2.36274272e-01
-1.83879554e-01 -1.34266913e-01 -1.22409356e+00 4.82231975e-01
-2.17401052e+00 -7.08290279e-01 4.96335864e-01 1.88388622e+00
9.08849835e-01 8.71110559e-01 -4.91564646e-02 -2.12266371e-02
5.72680295e-01 -1.70648247e-01 -7.16542482e-01 -3.67342025e-01
3.63769025e-01 -1.30276503e-02 6.51579916e-01 6.64017260e-01
-9.75345910e-01 1.16273594e+00 8.62040806e+00 5.48578620e-01
1.83261670e-02 -5.03956601e-02 8.97304714e-01 3.93202811e-01
-8.49497676e-01 3.37871790e-01 -3.21940184e-01 -4.56675924e-02
1.11769795e+00 -2.44607106e-01 2.81073749e-01 1.40886724e+00
-5.36564350e-01 -5.43453693e-01 -1.32549870e+00 5.51870406e-01
-3.32141519e-01 -8.56783867e-01 -5.13180196e-01 -2.76916832e-01
7.33244658e-01 -1.29933879e-01 1.57433063e-01 7.98025608e-01
1.68357217e+00 -1.46537912e+00 1.34994173e+00 7.72147775e-01
7.14013457e-01 -9.61268842e-01 8.58940721e-01 6.63297653e-01
-4.76455301e-01 -3.70022506e-01 -3.50095063e-01 -5.65655112e-01
-3.24765682e-01 4.51950103e-01 -1.15613770e+00 1.61494061e-01
5.70510983e-01 1.05474867e-01 -3.80433887e-01 4.11731809e-01
-4.75454450e-01 1.46550760e-01 -7.65390322e-02 -2.98464537e-01
2.63833553e-01 1.58272207e-01 2.37425715e-01 7.88451076e-01
1.92091748e-01 3.72138053e-01 4.79994714e-02 9.35631275e-01
6.37841463e-01 -6.98313475e-01 -4.72865373e-01 -2.67442197e-01
8.55798900e-01 1.15444434e+00 -5.33220410e-01 -1.22702926e-01
1.86690897e-01 7.26740479e-01 5.94855189e-01 -7.29978383e-02
-6.33062780e-01 -2.24711606e-03 8.18206787e-01 -3.76224816e-01
-4.23385024e-01 -4.60870504e-01 -6.52724862e-01 -8.44869673e-01
-2.68837214e-01 -8.06161582e-01 4.62194622e-01 -1.29338253e+00
-1.11409390e+00 6.25652790e-01 1.66891441e-01 -9.22066092e-01
-8.64762485e-01 -7.79265225e-01 -4.68007892e-01 5.74026048e-01
-9.16589558e-01 -9.66872633e-01 -2.54687369e-01 1.18345380e-01
1.88299745e-01 -1.93020895e-01 1.06613135e+00 -1.00438380e+00
-5.08637540e-02 4.48462158e-01 -5.96909881e-01 -5.87216735e-01
5.97793818e-01 -1.53871679e+00 6.29242778e-01 7.05054402e-01
-3.16765994e-01 9.30586457e-01 1.32640600e+00 -1.16536832e+00
-1.29889691e+00 -3.73954028e-01 3.85663539e-01 -1.14520264e+00
6.95267737e-01 -2.37969272e-02 -1.69234887e-01 9.52563226e-01
-2.18263581e-01 -4.84351069e-01 7.04230189e-01 7.06572533e-01
-2.90419638e-01 2.84664541e-01 -1.52004540e+00 7.32265830e-01
1.19219196e+00 -5.94409227e-01 -1.13582337e+00 6.19895793e-02
9.49711382e-01 -5.49200773e-01 -1.27472329e+00 6.31416798e-01
6.23527825e-01 -9.64134812e-01 6.37811840e-01 -7.45050550e-01
5.64686775e-01 -2.15404123e-01 -2.72463351e-01 -1.35776103e+00
-8.75358760e-01 -7.33635247e-01 -1.82736352e-01 5.80005705e-01
7.03034878e-01 -4.33548152e-01 1.14285457e+00 1.56172371e+00
-4.70951408e-01 -6.93625689e-01 -7.93771505e-01 -5.31364143e-01
5.08180857e-02 -1.01268733e+00 6.47610307e-01 7.03544438e-01
8.17793489e-01 4.21941727e-02 5.37372120e-02 2.80024141e-01
6.62200928e-01 -3.94844443e-01 5.07566512e-01 -1.57104003e+00
-2.53708601e-01 -2.49971643e-01 -5.58415890e-01 -6.65974319e-01
3.40680987e-01 -4.14614320e-01 7.64185250e-01 -1.56937468e+00
4.43781704e-01 -6.44160092e-01 -3.17787588e-01 3.41138363e-01
-3.23536515e-01 -4.56494242e-01 -2.68194795e-01 -7.19171837e-02
-1.20160890e+00 4.23347801e-01 9.05986011e-01 1.18590243e-01
1.84865937e-01 1.35589182e-01 -1.55281878e+00 1.18623674e+00
6.73795581e-01 -4.44946289e-01 -7.67343462e-01 -2.17003077e-01
7.64640689e-01 4.52004731e-01 2.39900406e-02 -1.42366087e+00
5.67801595e-01 -8.33864331e-01 6.07537270e-01 -1.70419991e-01
3.75519484e-01 -7.53321409e-01 1.46834567e-01 6.59739792e-01
-5.44897497e-01 3.96133244e-01 1.34209335e-01 8.96879554e-01
4.72613573e-02 -4.53180820e-01 2.66138822e-01 -3.66088271e-01
-1.00976455e+00 -1.08876608e-01 -7.26047456e-01 -6.78751618e-02
1.06489587e+00 -4.17757899e-01 -3.69060159e-01 -8.87970090e-01
-7.00084627e-01 5.56727469e-01 7.68942833e-01 5.60689747e-01
7.99572110e-01 -9.92234051e-01 -5.93041420e-01 -7.23618716e-02
4.63619828e-01 3.55107002e-02 1.85225219e-01 5.59631251e-02
-3.82301986e-01 5.98872185e-01 -5.60307801e-01 -3.16220433e-01
-4.57986563e-01 3.65185291e-01 1.59013107e-01 -2.95144498e-01
4.21388373e-02 1.09601176e+00 -7.30729708e-03 -5.82068205e-01
2.66947091e-01 -5.12261868e-01 1.09407799e-02 -7.83878803e-01
3.94173145e-01 4.41883773e-01 -3.08889061e-01 -1.09015368e-01
-3.54124159e-01 8.60746950e-03 1.27588764e-01 -8.66996050e-01
1.04580545e+00 -1.72642365e-01 2.24327087e-01 4.93326813e-01
-3.35080689e-03 -1.71461478e-01 -1.64365935e+00 1.24511749e-01
3.86499316e-01 -1.89910859e-01 1.18148237e-01 -1.18387473e+00
3.82244512e-02 1.36519149e-01 3.81315023e-01 -3.27688128e-01
6.59868777e-01 2.70455360e-01 -1.13062896e-01 8.86134505e-01
1.15288413e+00 -1.52045512e+00 2.32166290e-01 6.43620074e-01
6.92976117e-01 -1.45028126e+00 3.70801598e-01 -3.11846852e-01
-1.37313449e+00 8.80603373e-01 1.18081713e+00 5.96746318e-02
4.59688663e-01 3.24686199e-01 -2.42499128e-01 -2.39751294e-01
-9.30966675e-01 -1.27238650e-02 2.70851016e-01 9.74894166e-01
2.79454529e-01 6.31001055e-01 1.52613558e-02 1.20386767e+00
-7.20887721e-01 -6.01944551e-02 6.04014218e-01 1.11341202e+00
-9.95295405e-01 -5.52003741e-01 -8.64190012e-02 5.77514052e-01
-1.03491053e-01 2.23648623e-01 -5.20697653e-01 4.80880857e-01
-1.52517319e-01 1.23271728e+00 1.11061707e-01 -9.79484618e-01
6.24285592e-03 -1.83259591e-01 7.74597406e-01 -9.54116762e-01
-2.89799899e-01 -6.80504143e-01 5.66561222e-01 -8.88973296e-01
-2.60715306e-01 -7.32127249e-01 -1.31032979e+00 -1.35126635e-01
-1.77576572e-01 4.07953054e-01 7.31059372e-01 9.06921268e-01
1.15961120e-01 3.60571653e-01 1.93253905e-01 -6.34747326e-01
-9.41908002e-01 -1.14131021e+00 -4.18463260e-01 -3.41257602e-02
-1.36953071e-01 -9.47756708e-01 -2.13676259e-01 -6.24566138e-01] | [4.191713809967041, 1.1556198596954346] |
5f535c89-768c-4fc5-bcd5-d0a53ff17fa2 | pegg-net-background-agnostic-pixel-wise | 2203.16301 | null | https://arxiv.org/abs/2203.16301v2 | https://arxiv.org/pdf/2203.16301v2.pdf | PEGG-Net: Background Agnostic Pixel-Wise Efficient Grasp Generation Under Closed-Loop Conditions | Performing closed-loop grasping at close proximity to an object requires a large field of view. However, such images will inevitably bring large amounts of unnecessary background information, especially when the camera is far away from the target object at the initial stage, resulting in performance degradation of the grasping network. To address this problem, we design a novel PEGG-Net, a real-time, pixel-wise, robotic grasp generation network. The proposed lightweight network is inherently able to learn to remove background noise that can reduce grasping accuracy. Our proposed PEGG-Net achieves improved state-of-the-art performance on both Cornell dataset (98.9%) and Jacquard dataset (93.8%). In the real-world tests, PEGG-Net can support closed-loop grasping at up to 50Hz using an image size of 480x480 in dynamic environments. The trained model also generalizes to previously unseen objects with complex geometrical shapes, household objects and workshop tools and achieved an overall grasp success rate of 91.2% in our real-world grasping experiments. | ['Marcelo H Ang Jr', 'Huan Yin', 'Lei Zhou', 'Haozhe Wang', 'Zhiyang Liu'] | 2022-03-30 | null | null | null | null | ['grasp-generation'] | ['computer-vision'] | [ 2.16757491e-01 -2.21561864e-01 4.18157279e-01 -1.76794454e-01
-3.73226315e-01 -5.40159583e-01 -6.06277101e-02 -2.56368846e-01
-3.93673331e-01 3.12840551e-01 -4.57308859e-01 -8.70197453e-03
-6.50248751e-02 -8.52658153e-01 -1.30557084e+00 -7.76417375e-01
-5.27163804e-01 2.20538020e-01 3.60942602e-01 -8.76094960e-03
2.88258851e-01 8.18226397e-01 -1.72896588e+00 4.66787100e-01
6.12519741e-01 1.14106345e+00 1.10324514e+00 6.69991672e-01
1.66092306e-01 4.29836631e-01 -8.03138554e-01 -3.95656377e-02
7.59356618e-01 5.98595798e-01 -4.70902294e-01 -1.09197304e-01
5.89473784e-01 -1.01600087e+00 -5.46984613e-01 1.05675137e+00
6.86930597e-01 -1.44028872e-01 2.47645125e-01 -1.22233427e+00
-7.81266928e-01 7.12172627e-01 -7.03165829e-01 -1.82717726e-01
7.57502839e-02 6.16924226e-01 3.97168189e-01 -9.76424575e-01
5.56541264e-01 1.74575913e+00 4.00047332e-01 7.16969013e-01
-1.01026630e+00 -8.54456425e-01 1.64690673e-01 -1.36074424e-01
-7.03525186e-01 -5.74288797e-03 7.56582379e-01 -9.69370902e-02
7.52909184e-01 4.50789602e-03 5.36501110e-01 1.48743284e+00
4.01129901e-01 1.11301434e+00 9.86588657e-01 -3.05389971e-01
7.26546645e-02 -4.95271713e-01 -1.29163668e-01 6.27906919e-01
4.80390936e-01 6.65995255e-02 -3.58048946e-01 8.08389857e-02
1.37510991e+00 4.03554529e-01 -2.44158879e-01 -5.59021294e-01
-1.71648979e+00 3.08518440e-01 8.80077481e-01 -1.12036932e-02
-5.55054247e-01 4.87294644e-01 3.46432984e-01 1.82277441e-01
-2.86142789e-02 4.64490712e-01 -4.35675383e-01 4.78537679e-02
-8.83065462e-02 2.03661203e-01 9.28268552e-01 1.64037728e+00
3.41547638e-01 -7.80606791e-02 -1.48674533e-01 6.85470521e-01
2.16449395e-01 8.73626232e-01 -1.82121143e-01 -1.16141379e+00
7.78254986e-01 5.30003905e-01 5.84927082e-01 -1.04908264e+00
-3.19753945e-01 -2.33113810e-01 -8.40696692e-01 7.13569045e-01
8.18048775e-01 -2.02347428e-01 -1.08168745e+00 1.41483414e+00
2.78294265e-01 -3.67580622e-01 -2.82060746e-02 1.18178999e+00
6.90048754e-01 6.17787123e-01 -3.97109007e-03 2.42308229e-01
1.05996084e+00 -1.26213074e+00 -5.12483776e-01 -4.22636628e-01
-1.22827895e-01 -9.28300679e-01 1.39368832e+00 7.31262803e-01
-8.94872785e-01 -6.41790926e-01 -1.02457118e+00 2.23994121e-01
-4.90253456e-02 7.09525645e-01 9.81084228e-01 -1.02058358e-01
-7.35931933e-01 8.24616611e-01 -1.06761765e+00 -4.69353586e-01
6.45333707e-01 4.19727296e-01 -3.63903314e-01 -5.88329375e-01
-3.52982849e-01 6.63545489e-01 5.16103625e-01 6.10782743e-01
-1.06872368e+00 -5.49794018e-01 -3.43969345e-01 1.45922095e-01
5.97653866e-01 -2.23647416e-01 9.62590873e-01 -4.71744359e-01
-1.39804411e+00 6.05626822e-01 5.04121900e-01 2.67518442e-02
8.62446249e-01 -7.76195467e-01 2.53173798e-01 3.68694365e-01
2.94793285e-02 8.26650262e-01 1.10052490e+00 -1.58760130e+00
-3.74045968e-01 -6.83938742e-01 2.48790652e-01 -7.27007585e-03
-1.57310620e-01 -1.77716359e-01 -3.42000932e-01 -8.29298556e-01
3.72654140e-01 -1.05299556e+00 3.31155173e-02 7.04693198e-01
-2.76495218e-01 -4.31702524e-01 1.40858519e+00 -3.65370005e-01
9.34095457e-02 -2.24647284e+00 -8.94110277e-02 -2.60066748e-01
-1.17654145e-01 4.53054398e-01 -5.28163195e-01 5.18125176e-01
2.57434696e-01 -5.95094323e-01 1.91134572e-01 9.92386937e-02
-5.83489053e-02 2.10441872e-01 -5.26678801e-01 4.42131251e-01
1.32854789e-01 1.01954770e+00 -1.09298801e+00 -1.24696955e-01
5.20201683e-01 2.87439108e-01 -3.02383214e-01 5.67608476e-01
-4.57882494e-01 2.89597154e-01 -4.98433441e-01 1.24503422e+00
1.01676178e+00 -1.57412440e-01 -7.07764551e-02 -5.31959474e-01
4.72465269e-02 -7.20066905e-01 -1.10659754e+00 2.09890342e+00
-2.50822872e-01 4.26893920e-01 5.71371019e-01 -7.12610126e-01
1.28796995e+00 -2.35495001e-01 4.30329591e-01 -1.47048458e-01
3.72021496e-01 3.43551993e-01 4.91866991e-02 -8.06027055e-01
1.22410402e-01 5.72419822e-01 3.31757635e-01 1.49253011e-01
1.05077781e-01 -6.77224845e-02 -1.41905621e-02 -1.16091110e-01
1.22104561e+00 4.82301414e-01 -4.45734739e-01 -5.18607318e-01
-1.64300010e-01 -9.70114991e-02 1.03279635e-01 1.04665959e+00
-2.04126224e-01 6.13841891e-01 1.31827682e-01 -7.13652611e-01
-1.07047999e+00 -1.28212857e+00 8.02465528e-02 9.85271811e-01
7.19455242e-01 2.75766999e-01 -5.72477221e-01 -3.78643513e-01
4.45547163e-01 1.25046849e-01 -4.27982688e-01 -9.90316421e-02
-7.14983106e-01 -3.58381480e-01 1.60267428e-01 8.50997508e-01
9.39299822e-01 -1.70432973e+00 -1.33082092e+00 4.41074967e-01
4.18107808e-02 -1.34313452e+00 -1.55210599e-01 1.98545322e-01
-1.05767524e+00 -1.37594175e+00 -1.09883177e+00 -1.25696158e+00
9.74466503e-01 5.91101885e-01 6.51416421e-01 5.76849058e-02
-8.58935952e-01 3.28538030e-01 -6.04485810e-01 -6.57444596e-01
9.10670089e-04 -2.20189486e-02 6.20342791e-02 -3.12933147e-01
4.12976533e-01 -3.09055299e-01 -8.62486243e-01 4.03324276e-01
-7.39642143e-01 -6.46919161e-02 9.15143490e-01 1.03421009e+00
1.98142350e-01 -1.97269753e-01 5.58158517e-01 5.28624952e-02
5.31093597e-01 1.45175308e-01 -7.96561062e-01 4.91463363e-01
9.33074579e-02 -2.59145230e-01 5.60190916e-01 -9.90266562e-01
-8.09816062e-01 3.25070739e-01 4.39059794e-01 -8.52400362e-01
-1.58491090e-01 -1.81794971e-01 6.02928996e-02 -3.74255031e-01
5.61722338e-01 -7.99931288e-02 2.75162041e-01 -5.15877068e-01
1.32795304e-01 8.52664769e-01 8.38757038e-01 -9.55991149e-01
6.03716552e-01 5.17580867e-01 -1.20645329e-01 -6.94447339e-01
-5.11963785e-01 -2.11993501e-01 -6.44540846e-01 -4.96166319e-01
5.25589347e-01 -8.79987955e-01 -1.34543967e+00 1.19596195e+00
-1.51641405e+00 -6.50510252e-01 -6.15201071e-02 5.16837299e-01
-5.07656813e-01 2.68153667e-01 -8.25516105e-01 -8.74009192e-01
-8.06989789e-01 -1.22921503e+00 1.37171495e+00 3.42604399e-01
3.95971537e-01 8.62707198e-03 -9.83991146e-01 -1.01597200e-03
3.86431187e-01 6.55411303e-01 6.06352985e-01 -4.66172211e-02
-9.70719695e-01 -4.72890347e-01 -7.12119281e-01 4.91510600e-01
4.57954019e-01 -4.75902744e-02 -8.60741258e-01 -8.98441494e-01
-8.98389444e-02 -6.34632051e-01 8.17975938e-01 2.89456606e-01
1.38289154e+00 -1.04229927e-01 -4.50045019e-01 3.13817471e-01
1.45776176e+00 4.12211806e-01 5.12148380e-01 1.56635508e-01
6.36739671e-01 4.61456388e-01 1.00324190e+00 6.28294766e-01
-2.86106884e-01 3.02599967e-01 1.12352943e+00 -5.05593466e-03
-3.91488150e-03 -1.06378451e-01 1.27221778e-01 3.14109564e-01
-2.57657737e-01 -2.39458486e-01 -8.40926707e-01 5.37673116e-01
-1.90159190e+00 -5.42055428e-01 3.78143974e-02 1.98955798e+00
5.13964474e-01 1.12214051e-01 -2.54206419e-01 -2.62634635e-01
9.09794271e-01 -1.74407721e-01 -9.94825661e-01 3.80802721e-01
-7.09099621e-02 -2.96104215e-02 7.25830615e-01 -1.10269273e-02
-1.03860152e+00 1.03624535e+00 5.40033770e+00 5.31386018e-01
-1.20762777e+00 -3.59555990e-01 1.28582940e-01 -5.51788323e-02
5.65649331e-01 -5.28759480e-01 -4.31737721e-01 4.35057163e-01
-3.25519085e-01 4.36543107e-01 7.33050764e-01 1.42719471e+00
-2.32609391e-01 -3.49181831e-01 -1.23299980e+00 1.11191630e+00
-4.65377346e-02 -9.69822109e-01 -1.07563883e-01 -2.29971930e-01
4.67039138e-01 1.49662718e-01 -2.53291428e-01 1.48093835e-01
2.87291348e-01 -8.04489613e-01 7.21344888e-01 2.45751843e-01
8.01137030e-01 -4.47836816e-01 7.00621545e-01 4.27016735e-01
-8.10062289e-01 -4.62061942e-01 -8.85069668e-01 -1.22291714e-01
-7.78261125e-02 4.09305751e-01 -5.61309218e-01 8.78656209e-02
1.40240276e+00 4.38652813e-01 -5.36273867e-02 9.93859947e-01
-7.35616013e-02 -6.37176633e-02 -5.28384805e-01 -4.14781481e-01
3.61674160e-01 1.44115150e-01 5.38207829e-01 9.22793806e-01
2.69073188e-01 2.38720402e-01 3.86129409e-01 9.84689057e-01
-2.49342099e-02 -3.79094988e-01 -6.99463427e-01 1.13987830e-02
4.39088374e-01 1.24923313e+00 -9.34506714e-01 -1.34933874e-01
-6.08465113e-02 1.04625976e+00 3.60581219e-01 4.48807269e-01
-5.69503307e-01 -8.52053285e-01 4.12139952e-01 -1.61033839e-01
3.89169484e-01 -5.95731437e-01 -5.74228391e-02 -1.09106505e+00
6.55774057e-01 -7.50797987e-01 -3.02182555e-01 -1.06595409e+00
-1.55877030e+00 4.90069866e-01 -1.39381230e-01 -1.04514503e+00
5.83406210e-01 -1.38148355e+00 -3.10729176e-01 5.44891894e-01
-1.27492285e+00 -1.13169575e+00 -1.28973532e+00 5.05992949e-01
7.62027144e-01 -1.64990410e-01 7.04109728e-01 8.77742916e-02
-5.95233068e-02 2.36926302e-01 1.05869994e-01 3.38367462e-01
9.05850470e-01 -9.76439893e-01 2.72308171e-01 5.10407507e-01
-5.29543579e-01 6.52020097e-01 4.43866730e-01 -7.14298069e-01
-2.13507080e+00 -1.06656921e+00 -1.94700330e-01 -2.12809071e-01
4.42618847e-01 -7.32190192e-01 -7.84236133e-01 5.11838377e-01
3.56393084e-02 3.57162744e-01 -3.86450827e-01 -4.09766942e-01
-2.32706666e-01 -3.72371912e-01 -1.40360761e+00 5.40298641e-01
1.35156238e+00 1.21431552e-01 -4.27163780e-01 8.31970215e-01
6.93725109e-01 -9.59077656e-01 -8.42183888e-01 7.80055523e-01
9.82842863e-01 -6.59228146e-01 1.17316854e+00 -5.09544492e-01
6.34853721e-01 -5.40970936e-02 -3.20124388e-01 -1.05488729e+00
-2.50984997e-01 -6.41442478e-01 -1.49668232e-01 7.42155194e-01
-2.49508113e-01 -5.61108232e-01 8.36535811e-01 4.21130747e-01
-7.52305388e-02 -8.74582469e-01 -6.45364940e-01 -1.12113583e+00
-1.09564692e-01 2.90111266e-02 4.81854677e-01 4.15186644e-01
1.88476518e-02 -3.35191220e-01 -2.75084078e-01 4.03091460e-01
9.96820688e-01 4.47142631e-01 9.51594710e-01 -1.24165070e+00
5.88761456e-02 6.91518709e-02 -3.93512487e-01 -1.33355188e+00
-1.08510502e-01 -4.05442089e-01 6.10136271e-01 -1.67461741e+00
1.06382050e-01 -8.84875357e-01 8.84986147e-02 6.66813016e-01
-4.53648455e-02 6.48202300e-02 5.11654019e-01 2.08594561e-01
-2.07091451e-01 4.60171014e-01 1.85450161e+00 -4.16358173e-01
-8.77399871e-04 -9.53060463e-02 -1.71182770e-02 5.32377958e-01
8.08139503e-01 -1.95615366e-01 -1.72030434e-01 -9.30183232e-01
-5.39626002e-01 8.53774250e-02 7.76150465e-01 -1.16381896e+00
1.75814092e-01 -3.41621011e-01 6.15372717e-01 -6.18822813e-01
3.66025835e-01 -1.17717469e+00 -1.39675125e-01 8.88124108e-01
-1.15166441e-01 -6.38814643e-02 3.41438890e-01 6.83031142e-01
2.68328398e-01 -2.56333739e-01 7.16648042e-01 -5.03087580e-01
-7.08510220e-01 2.73481011e-01 2.58841962e-01 -3.26129943e-01
1.33437550e+00 -1.74957171e-01 -6.59609079e-01 6.12557754e-02
-5.86399376e-01 2.71530807e-01 3.30513179e-01 9.41501141e-01
8.14745605e-01 -1.19548476e+00 -5.47763646e-01 2.83296883e-01
4.94235195e-04 5.49888909e-01 2.69986451e-01 4.00883496e-01
-9.65912044e-01 1.38371006e-01 -5.84690869e-01 -1.04891145e+00
-1.08642387e+00 6.92889273e-01 -8.80394131e-02 3.85338455e-01
-1.10080194e+00 1.02195990e+00 -9.18876156e-02 -3.12791228e-01
9.94243681e-01 -7.30018079e-01 4.64243710e-01 -5.62584698e-01
3.50771010e-01 4.86242533e-01 -1.67477190e-01 1.91582233e-01
-2.47390658e-01 5.69763005e-01 -1.41553104e-01 5.95479012e-01
1.66839182e+00 4.34872508e-01 -2.02080294e-01 6.61614910e-02
1.01763678e+00 -6.15433991e-01 -1.90410388e+00 -9.80991572e-02
-4.98765171e-01 -8.28598559e-01 -3.93520921e-01 -1.00057602e+00
-1.16270220e+00 8.52961481e-01 8.01760674e-01 -1.17667161e-01
7.92902529e-01 1.41278747e-02 7.58491874e-01 1.09629333e+00
1.20023966e+00 -1.05024719e+00 5.64699292e-01 4.74742264e-01
1.47735238e+00 -1.55668414e+00 -1.56918764e-01 -6.22540832e-01
-1.81630880e-01 1.62657523e+00 1.27474618e+00 -7.48508036e-01
3.58384371e-01 5.22972524e-01 8.77382010e-02 -2.23408505e-01
-2.29106724e-01 4.85570133e-01 -1.87179908e-01 6.50494814e-01
-3.71356815e-01 6.21751249e-02 3.13359529e-01 1.46186650e-01
1.12854175e-01 2.42533490e-01 4.24882919e-01 1.53065097e+00
-5.37283301e-01 -2.92208433e-01 -3.06172103e-01 4.47532266e-01
-1.20721400e-01 3.07447195e-01 -3.75894271e-02 8.23024392e-01
-1.18831113e-01 8.25260520e-01 -1.95442196e-02 -2.74056286e-01
5.48105717e-01 -5.27187109e-01 9.01289105e-01 -3.74706060e-01
-2.65538394e-01 -5.88115864e-02 -5.09163737e-01 -9.30984497e-01
-5.21735251e-01 -3.41549188e-01 -1.10584295e+00 -1.26494646e-01
-5.80621541e-01 -4.15600419e-01 8.40329766e-01 4.06040102e-01
4.05486286e-01 5.94042957e-01 3.57418269e-01 -1.54411244e+00
-1.10339355e+00 -1.18647015e+00 -5.92793226e-01 2.90380657e-01
2.92437375e-01 -7.80674934e-01 -2.19118923e-01 -7.03083128e-02] | [5.79473876953125, -0.899118959903717] |
6e358af9-a460-42d9-80a1-3fadb1b83143 | dexvip-learning-dexterous-grasping-with-human | 2202.00164 | null | https://arxiv.org/abs/2202.00164v1 | https://arxiv.org/pdf/2202.00164v1.pdf | DexVIP: Learning Dexterous Grasping with Human Hand Pose Priors from Video | Dexterous multi-fingered robotic hands have a formidable action space, yet their morphological similarity to the human hand holds immense potential to accelerate robot learning. We propose DexVIP, an approach to learn dexterous robotic grasping from human-object interactions present in in-the-wild YouTube videos. We do this by curating grasp images from human-object interaction videos and imposing a prior over the agent's hand pose when learning to grasp with deep reinforcement learning. A key advantage of our method is that the learned policy is able to leverage free-form in-the-wild visual data. As a result, it can easily scale to new objects, and it sidesteps the standard practice of collecting human demonstrations in a lab -- a much more expensive and indirect way to capture human expertise. Through experiments on 27 objects with a 30-DoF simulated robot hand, we demonstrate that DexVIP compares favorably to existing approaches that lack a hand pose prior or rely on specialized tele-operation equipment to obtain human demonstrations, while also being faster to train. Project page: https://vision.cs.utexas.edu/projects/dexvip-dexterous-grasp-pose-prior | ['Kristen Grauman', 'Priyanka Mandikal'] | 2022-02-01 | null | null | null | null | ['robotic-grasping'] | ['robots'] | [-4.80371743e-01 -1.42012402e-01 -3.20574045e-02 -5.68118915e-02
-4.57961679e-01 -9.58487689e-01 1.86932027e-01 -6.85955524e-01
-5.03307521e-01 6.50604904e-01 -1.66634068e-01 -3.97349983e-01
-1.51669979e-01 -9.01098773e-02 -1.13476026e+00 -4.39527333e-01
-3.37104023e-01 8.07054758e-01 1.61729530e-01 -1.80493325e-01
3.08204293e-01 8.53255689e-01 -1.47797036e+00 2.16065839e-01
3.23609293e-01 6.84038639e-01 8.38865638e-01 9.29718256e-01
5.69610834e-01 8.62234771e-01 -3.86450648e-01 1.64599404e-01
8.66308987e-01 7.68701881e-02 -8.62135768e-01 1.47455871e-01
5.72380126e-01 -1.11576164e+00 -8.66006553e-01 6.73350394e-01
5.40909410e-01 8.29604194e-02 3.55349541e-01 -1.59525514e+00
-4.18901265e-01 3.11604023e-01 -4.84882087e-01 -1.22160919e-01
6.66500926e-01 9.75728273e-01 7.30677187e-01 -7.30771720e-01
9.26579893e-01 1.56111741e+00 5.64057648e-01 7.90825546e-01
-9.35113966e-01 -5.76656699e-01 7.75626376e-02 3.69808152e-02
-8.49821210e-01 -1.14698023e-01 3.92677486e-01 -6.97676539e-01
1.05844057e+00 -7.55965710e-02 8.58171761e-01 1.56266844e+00
1.33933336e-01 1.19886589e+00 1.03888547e+00 -3.47509116e-01
4.00792584e-02 -3.57973963e-01 -3.10898781e-01 9.31463599e-01
2.39081904e-01 4.29156989e-01 -3.26639503e-01 1.57016739e-02
1.67317104e+00 4.73059803e-01 -5.02041042e-01 -1.26978123e+00
-1.58486962e+00 3.02956462e-01 5.88226974e-01 -4.53085639e-02
-6.83102608e-01 6.39479220e-01 2.33787268e-01 5.89831710e-01
-5.96819878e-01 6.59143269e-01 -7.43428171e-01 -5.39173186e-01
-9.98428315e-02 8.18265796e-01 1.08609831e+00 1.44085002e+00
4.55098361e-01 -1.74685985e-01 1.77553028e-01 3.15464228e-01
2.39387751e-01 5.76481283e-01 9.22620818e-02 -1.63277447e+00
5.75169981e-01 2.96221495e-01 7.83701837e-01 -3.71974736e-01
-3.12748104e-01 4.27742660e-01 4.46795933e-02 1.21205616e+00
8.50081265e-01 -3.34652573e-01 -1.06182992e+00 1.27405286e+00
3.76840919e-01 -5.16578376e-01 -1.46907404e-01 1.36930692e+00
2.23005280e-01 2.76230007e-01 8.48722607e-02 4.27239060e-01
1.08704269e+00 -1.19483149e+00 -2.30992228e-01 -2.60689795e-01
3.38243067e-01 -5.44268966e-01 1.59653652e+00 8.10764074e-01
-9.23516810e-01 -4.27323490e-01 -8.52830887e-01 -7.00047240e-02
-1.23458348e-01 3.40053946e-01 1.00416398e+00 6.66090148e-03
-7.20755696e-01 9.99581873e-01 -1.29239559e+00 -4.69295353e-01
3.58941853e-01 4.99945402e-01 -6.53770983e-01 -3.00597101e-01
-3.43611836e-01 1.22158575e+00 5.57430863e-01 -3.93470638e-02
-1.48790610e+00 -5.93476176e-01 -7.11895049e-01 -1.60226837e-01
7.92367995e-01 -4.77678269e-01 1.74125302e+00 -6.40225053e-01
-1.55870581e+00 5.95602512e-01 5.30486822e-01 5.33364192e-02
1.01051497e+00 -7.84855783e-01 4.43630010e-01 5.20658135e-01
-1.05288677e-01 8.49980891e-01 9.77515221e-01 -1.58754086e+00
-4.40607160e-01 -3.64522010e-01 5.32632232e-01 2.93023765e-01
3.40183340e-02 -2.58227944e-01 -4.00384873e-01 -6.31345630e-01
-1.42845005e-01 -1.39409387e+00 -6.41713217e-02 8.26874375e-01
-8.66702572e-02 -3.39189202e-01 1.19204807e+00 -6.53346956e-01
4.64971811e-02 -1.95987058e+00 6.52813077e-01 1.14245266e-02
1.65050894e-01 3.04997385e-01 -1.64798245e-01 7.09471107e-01
1.88023731e-01 -4.96613622e-01 3.15919816e-01 -1.40235238e-02
2.39612609e-01 5.00633419e-01 -1.92254901e-01 3.39221150e-01
-8.24956521e-02 9.46783781e-01 -1.23941433e+00 -3.05352390e-01
3.89777243e-01 4.34920579e-01 -5.55111170e-01 5.76757967e-01
-4.79556620e-01 7.87327528e-01 -4.70883757e-01 1.04467046e+00
3.07050824e-01 -7.86807314e-02 4.14916575e-01 -1.97531208e-01
-2.84551270e-02 -1.45550892e-01 -1.09440243e+00 2.16371608e+00
-4.27454263e-01 4.18127030e-01 6.87852740e-01 -6.36218667e-01
4.86770570e-01 2.80920982e-01 5.37309766e-01 -1.14105180e-01
2.34559119e-01 4.17252481e-01 7.85100013e-02 -9.55776632e-01
1.88534498e-01 2.03880727e-01 2.94532180e-01 4.04466510e-01
4.46829349e-01 -6.46304250e-01 2.02769369e-01 -4.93716970e-02
1.29999530e+00 9.43495512e-01 -1.29140005e-01 -6.70593530e-02
-3.26988757e-01 2.69696683e-01 -2.52999482e-03 7.13328183e-01
-1.94632187e-01 2.78071970e-01 2.73037404e-01 -4.59310353e-01
-1.48227024e+00 -1.34848070e+00 4.28454489e-01 1.08242059e+00
4.50230946e-05 1.09168805e-01 -5.53699255e-01 -7.42339969e-01
6.47871375e-01 3.90880406e-01 -3.99258614e-01 3.10871392e-01
-7.60142863e-01 4.60337698e-01 3.04545105e-01 1.06345153e+00
3.03054512e-01 -1.55465376e+00 -1.43093216e+00 2.38434095e-02
2.94407874e-01 -1.01458144e+00 -3.18191826e-01 2.32668459e-01
-8.21476817e-01 -1.38238597e+00 -1.28142631e+00 -1.02961850e+00
8.93677056e-01 3.10612470e-01 6.14192843e-01 -1.73334554e-01
-8.48171115e-01 1.23490834e+00 -5.66163361e-01 -4.75553930e-01
-2.75110066e-01 -3.21132272e-01 2.46124789e-01 -9.64866996e-01
1.67129621e-01 -6.74380064e-01 -5.85525572e-01 2.82348573e-01
-4.12395567e-01 -1.22668110e-01 9.59248841e-01 1.00049388e+00
1.33736759e-01 -4.64570493e-01 4.11585085e-02 1.09558348e-02
5.96393466e-01 2.01713741e-02 -8.29477847e-01 2.84760803e-01
-1.76005453e-01 -9.96247157e-02 3.29125702e-01 -9.27170753e-01
-8.41058433e-01 5.18664718e-01 3.44398171e-01 -1.22736692e+00
-2.46061429e-01 1.05630226e-01 1.48517579e-01 -2.96178192e-01
6.19841635e-01 -2.13776171e-01 4.58796620e-01 -5.50194800e-01
4.11980450e-01 6.28544092e-01 7.98948050e-01 -1.15126336e+00
7.01381028e-01 3.39240581e-01 -2.07158506e-01 -6.20467305e-01
-1.01607949e-01 -4.32500303e-01 -9.65246320e-01 -3.16347837e-01
5.89481533e-01 -8.05500090e-01 -1.55147743e+00 5.96294880e-01
-1.30775583e+00 -1.07538891e+00 -1.34240374e-01 8.71350408e-01
-1.18489230e+00 4.00892675e-01 -8.81505847e-01 -7.55816579e-01
-1.39232904e-01 -1.05757594e+00 1.05118573e+00 1.02330774e-01
-2.17524350e-01 -4.37001377e-01 -2.05694526e-01 1.89892024e-01
1.56086281e-01 2.97406226e-01 5.06154299e-01 -1.65777400e-01
-7.61578798e-01 -3.42966765e-01 -2.03848094e-01 4.50824201e-01
5.59981391e-02 -3.26693147e-01 -5.30047894e-01 -9.51039016e-01
-2.37436965e-01 -1.09704900e+00 3.18895668e-01 2.81671435e-02
1.11382043e+00 -2.51870066e-01 -2.30207816e-01 2.33523518e-01
1.21860778e+00 2.59060413e-01 3.94050241e-01 3.07817847e-01
8.04001570e-01 3.81890774e-01 1.00641096e+00 5.16336620e-01
9.62009095e-03 5.34185708e-01 5.70962429e-01 2.74315834e-01
-6.01191185e-02 -3.55710149e-01 4.23506379e-01 1.96814373e-01
-7.36534119e-01 7.77147785e-02 -9.65681374e-01 4.55750227e-01
-1.90688539e+00 -7.71538258e-01 4.05947119e-01 1.94548666e+00
7.86188424e-01 -6.29529655e-02 3.64943266e-01 -1.71157449e-01
2.52372324e-01 -5.24279594e-01 -8.82819831e-01 -3.79773579e-03
5.86711168e-01 1.53000563e-01 6.58857405e-01 3.86171967e-01
-7.04063952e-01 9.62995529e-01 5.37579679e+00 1.02407023e-01
-1.04321563e+00 -1.71415478e-01 -5.66968799e-01 -3.68314594e-01
5.28746068e-01 -7.34856352e-02 -2.85373956e-01 2.11191490e-01
-1.53508037e-02 2.86567152e-01 9.66210425e-01 1.29041171e+00
-1.08388462e-03 -3.09451848e-01 -1.78237522e+00 1.14572322e+00
-1.78727582e-01 -7.93039560e-01 -2.60503501e-01 9.35373157e-02
3.04742843e-01 2.11657792e-01 -1.75016478e-01 4.24637586e-01
6.03893638e-01 -9.81707215e-01 8.14211488e-01 4.26053852e-01
8.58151197e-01 -2.96751082e-01 2.90531248e-01 6.35851204e-01
-7.83894479e-01 -5.82012236e-01 -3.53468239e-01 -3.85625064e-01
1.51800975e-01 -4.25121993e-01 -1.14982355e+00 5.71265481e-02
1.12670851e+00 3.65406245e-01 6.30427077e-02 1.03104246e+00
-3.86597455e-01 -1.08017117e-01 -3.56606603e-01 -3.29738438e-01
2.95166910e-01 1.26550347e-01 5.19983828e-01 8.91227722e-01
2.43457988e-01 3.32068175e-01 6.43595874e-01 7.94584990e-01
1.75186381e-01 -5.13091624e-01 -8.22530448e-01 -3.78513575e-01
3.94802004e-01 9.05350387e-01 -5.20806134e-01 -1.67920217e-01
-1.10438056e-01 1.29905689e+00 4.26555336e-01 4.59857017e-01
-6.34606183e-01 -6.90823555e-01 6.34228826e-01 -1.01692274e-01
7.35211253e-01 -9.58404541e-01 2.22275704e-01 -1.08323109e+00
4.92656559e-01 -1.20742023e+00 -1.65477931e-01 -1.17638707e+00
-1.18531251e+00 8.89855921e-02 1.54914916e-01 -1.25347257e+00
-3.77688825e-01 -1.50651324e+00 -2.12874740e-01 7.01943338e-01
-8.95688653e-01 -1.33802235e+00 -7.69966543e-01 7.81003416e-01
7.94805586e-01 -8.94612446e-02 8.51531982e-01 -7.64083415e-02
1.83167592e-01 2.28473514e-01 -1.67246625e-01 1.28542706e-01
7.73143947e-01 -1.21338713e+00 2.63937682e-01 1.37324616e-01
-2.91567951e-01 8.75135958e-01 7.42830336e-01 -6.78950846e-01
-2.16486001e+00 -2.26137236e-01 -2.45055482e-01 -9.29886937e-01
7.40431666e-01 -4.19685125e-01 -5.82796812e-01 1.23747778e+00
1.81541443e-01 2.48476770e-02 -2.45017320e-01 -1.09497271e-01
-4.54717338e-01 1.86524808e-01 -1.22225559e+00 6.48623347e-01
1.38508093e+00 -3.58013839e-01 -1.10764515e+00 6.54097319e-01
5.33605278e-01 -8.00875306e-01 -8.65804672e-01 2.54558593e-01
1.22088075e+00 -4.56615537e-01 1.05027211e+00 -7.73672819e-01
5.31798005e-01 -2.90194571e-01 -1.61719710e-01 -1.31753063e+00
-2.69604534e-01 -7.68983722e-01 -4.39192355e-01 4.62615967e-01
-1.25094011e-01 -4.20513600e-01 8.15471113e-01 6.22271299e-01
-5.95282353e-02 -6.65868878e-01 -5.02004206e-01 -1.32808733e+00
8.98467973e-02 -5.38548164e-04 1.16348535e-01 7.07089603e-01
5.32972693e-01 -1.17641650e-01 -3.54851663e-01 1.89436138e-01
8.05487096e-01 -6.82977289e-02 1.35127389e+00 -1.14341247e+00
-7.72081733e-01 -2.11899400e-01 -2.03253865e-01 -1.20636439e+00
1.13521382e-01 -4.31485981e-01 4.12596166e-01 -1.56014395e+00
2.81891488e-02 -5.14830649e-01 1.92705974e-01 9.17672813e-01
2.92139262e-01 -2.87287563e-01 6.13033473e-01 4.34830010e-01
-4.53751564e-01 1.90756842e-01 1.68845439e+00 6.27926141e-02
-2.06193998e-01 -6.12076223e-02 1.81489512e-01 7.17618763e-01
7.46395946e-01 -1.69774264e-01 -4.13667887e-01 -6.53991699e-01
-3.02701622e-01 2.43601143e-01 1.01227617e+00 -1.08195770e+00
2.36007527e-01 -2.88702905e-01 4.93701011e-01 -4.11844760e-01
4.91405338e-01 -1.00715017e+00 -6.06410727e-02 9.40945923e-01
-1.20493785e-01 2.07681432e-01 4.07248467e-01 5.20168543e-01
4.48259652e-01 -2.81098187e-01 3.84742677e-01 -7.74443865e-01
-1.03759849e+00 5.86201325e-02 -3.04754108e-01 -6.74999878e-02
1.32933509e+00 -2.03514025e-01 -2.47059777e-01 -2.22747058e-01
-7.44054854e-01 4.35833752e-01 8.45784605e-01 6.78529024e-01
6.22033715e-01 -8.65513027e-01 -2.73392558e-01 4.12725881e-02
-1.51135977e-02 7.72856735e-03 -3.04545183e-02 6.11375213e-01
-9.74695504e-01 1.62037373e-01 -8.60819757e-01 -6.88062549e-01
-1.31556094e+00 8.19606423e-01 1.06191993e-01 3.86124551e-01
-1.08392453e+00 7.67407119e-01 -1.00607157e-01 -8.36819232e-01
8.63340139e-01 -5.25112391e-01 6.70966446e-01 -7.23428369e-01
1.91245094e-01 5.30785859e-01 -3.71914774e-01 2.54087180e-01
-1.22717872e-01 3.07850927e-01 -1.12631239e-01 -2.24772930e-01
1.56093323e+00 3.95288974e-01 9.21904892e-02 2.69975424e-01
8.28788459e-01 -3.04420084e-01 -2.14693403e+00 1.73454314e-01
-2.76095957e-01 -8.14541459e-01 -4.81191546e-01 -1.22950351e+00
-6.26922131e-01 7.97743976e-01 5.38435459e-01 -3.90171438e-01
4.47449356e-01 2.62156665e-01 5.36499739e-01 1.53527522e+00
1.13778257e+00 -1.31707597e+00 7.00718343e-01 5.86394250e-01
1.55884874e+00 -1.35941875e+00 6.17813058e-02 -1.97782114e-01
-7.22706676e-01 1.35718822e+00 9.52425718e-01 -3.96102816e-01
3.97501975e-01 5.19293904e-01 1.86435491e-01 -3.49704504e-01
-3.03777426e-01 1.31999865e-01 -1.11568302e-01 9.54988480e-01
-1.04793876e-01 9.94002223e-02 2.20785916e-01 1.47969499e-02
-3.30362171e-02 4.65362757e-01 4.60752010e-01 1.81873500e+00
-4.07423049e-01 -8.47436249e-01 -3.20882171e-01 1.94149077e-01
3.56835760e-02 4.66880143e-01 -3.03360164e-01 1.19064343e+00
-2.46373251e-01 4.80756879e-01 -3.14641595e-01 -1.45785570e-01
6.16525352e-01 9.52335075e-03 1.39958751e+00 -6.44224405e-01
-2.56722301e-01 -1.92436993e-01 -2.14219064e-01 -1.03843951e+00
-4.89673987e-02 -7.38487661e-01 -1.38108146e+00 -2.33601764e-01
-8.78566951e-02 -3.54400635e-01 8.81161451e-01 6.40196681e-01
3.18203360e-01 2.71964222e-01 7.05672055e-02 -2.04140830e+00
-1.31305242e+00 -1.00050354e+00 -5.68492293e-01 4.56181884e-01
4.26163852e-01 -1.17656314e+00 -1.48885593e-01 -1.19298734e-01] | [4.756779670715332, 0.5417726635932922] |
4e40568f-1f12-4f3c-b10c-94783f5b2781 | personalized-federated-learning-on-long | 2303.15168 | null | https://arxiv.org/abs/2303.15168v1 | https://arxiv.org/pdf/2303.15168v1.pdf | Personalized Federated Learning on Long-Tailed Data via Adversarial Feature Augmentation | Personalized Federated Learning (PFL) aims to learn personalized models for each client based on the knowledge across all clients in a privacy-preserving manner. Existing PFL methods generally assume that the underlying global data across all clients are uniformly distributed without considering the long-tail distribution. The joint problem of data heterogeneity and long-tail distribution in the FL environment is more challenging and severely affects the performance of personalized models. In this paper, we propose a PFL method called Federated Learning with Adversarial Feature Augmentation (FedAFA) to address this joint problem in PFL. FedAFA optimizes the personalized model for each client by producing a balanced feature set to enhance the local minority classes. The local minority class features are generated by transferring the knowledge from the local majority class features extracted by the global model in an adversarial example learning manner. The experimental results on benchmarks under different settings of data heterogeneity and long-tail distribution demonstrate that FedAFA significantly improves the personalized performance of each client compared with the state-of-the-art PFL algorithm. The code is available at https://github.com/pxqian/FedAFA. | ['Hanzi Wang', 'Gang Huang', 'Pinxin Qian', 'Yang Lu'] | 2023-03-27 | null | null | null | null | ['personalized-federated-learning'] | ['methodology'] | [-3.97412002e-01 -1.50880799e-01 -5.17215312e-01 -7.00622380e-01
-1.10361385e+00 -6.90534234e-01 3.41618925e-01 -2.19351962e-01
-1.63403079e-02 7.87403941e-01 3.92241180e-01 -1.56655684e-02
-8.15676246e-03 -9.41494763e-01 -9.04966414e-01 -1.09944379e+00
2.45514400e-02 3.52371126e-01 -1.60544828e-01 9.77154300e-02
-9.05435830e-02 4.13111389e-01 -1.33395147e+00 8.49934697e-01
8.24686706e-01 1.14201438e+00 -3.02175581e-01 3.85046452e-01
-2.99811959e-01 9.50072825e-01 -3.83538067e-01 -8.61220777e-01
7.60562062e-01 -2.65698522e-01 -7.27446437e-01 -1.54597029e-01
5.77726543e-01 -7.24232137e-01 -7.68816173e-01 1.00102544e+00
5.60719311e-01 -2.20555370e-03 2.50049472e-01 -1.81608689e+00
-9.76317465e-01 7.89833844e-01 -2.54069269e-01 -9.59277675e-02
-1.30594850e-01 3.82914305e-01 9.55656409e-01 -8.02799463e-01
4.59663182e-01 1.07554972e+00 6.37924314e-01 6.33910716e-01
-1.20892358e+00 -9.71960902e-01 3.79554242e-01 4.45045203e-01
-1.34082198e+00 -3.59826773e-01 6.46930754e-01 -2.70600408e-01
3.18881005e-01 5.96834362e-01 8.67101178e-02 8.97508562e-01
9.64177251e-02 1.03834510e+00 9.71186936e-01 5.23513258e-02
2.88267672e-01 4.22010034e-01 8.79450515e-02 3.92623782e-01
-1.82387338e-03 1.65241629e-01 -5.43279588e-01 -9.03789759e-01
2.77539849e-01 6.69407308e-01 -4.21374887e-01 -6.36477888e-01
-6.08657897e-01 9.85849619e-01 5.34862161e-01 -9.99219641e-02
-3.84698302e-01 -2.36980952e-02 5.62282383e-01 5.36821246e-01
4.88053918e-01 -1.89307988e-01 -9.06026125e-01 3.50301087e-01
-4.57467347e-01 6.29113674e-01 9.80145276e-01 1.15272260e+00
1.25709307e+00 -2.51759589e-01 -7.05958962e-01 5.78103125e-01
1.64349914e-01 4.41990942e-01 7.28223324e-01 -1.10977900e+00
5.92239141e-01 6.89844847e-01 2.17360303e-01 -7.15570092e-01
2.48249084e-01 -5.04582345e-01 -6.55291319e-01 2.36604333e-01
3.40266734e-01 -4.36831772e-01 -3.85078222e-01 1.91967058e+00
7.72183061e-01 1.77568734e-01 1.67634964e-01 8.03248286e-01
5.63768744e-01 6.00116849e-01 1.08308397e-01 -1.05859227e-01
8.23871076e-01 -1.30871820e+00 -6.24514520e-01 1.73784032e-01
5.17027736e-01 -5.40193796e-01 1.03375006e+00 1.38606444e-01
-6.85305536e-01 -1.21940887e-02 -4.58252013e-01 5.14862388e-02
-3.46708953e-01 -3.36974174e-01 5.29567301e-01 6.29160106e-01
-7.87963986e-01 5.96559346e-01 -5.60413599e-01 1.10882218e-03
1.09554791e+00 5.04150569e-01 -6.29775763e-01 -4.27248538e-01
-1.06255162e+00 2.03954756e-01 1.45570030e-02 -4.17191684e-01
-1.03473198e+00 -1.22558200e+00 -6.08598769e-01 3.41028333e-01
3.80015075e-01 -8.31102788e-01 1.47376585e+00 -1.12570727e+00
-1.22046030e+00 5.80659628e-01 -1.10996492e-01 -3.95289809e-01
9.75389540e-01 -1.65913388e-01 -1.64404392e-01 -3.46495450e-01
-3.49326842e-02 -1.69381797e-01 7.42887020e-01 -1.35693395e+00
-9.03326392e-01 -7.67279923e-01 -1.11134127e-01 1.33593101e-02
-6.01940274e-01 -1.57464415e-01 -3.96351755e-01 -6.79477096e-01
-5.25768697e-01 -6.08964503e-01 -2.52899885e-01 3.85796934e-01
-3.75700921e-01 -1.76769778e-01 1.29625916e+00 -5.40932536e-01
1.06279922e+00 -2.24757624e+00 -2.85898536e-01 2.20398441e-01
2.08824381e-01 2.52451926e-01 -2.20919341e-01 7.07906187e-01
1.74877301e-01 -3.72117050e-02 -1.57067418e-01 -5.20432651e-01
2.91849703e-01 3.32645476e-01 -4.84248251e-01 6.85760736e-01
-3.01618904e-01 8.62637222e-01 -6.75357580e-01 -4.01407063e-01
-8.87957662e-02 5.86173594e-01 -8.64536285e-01 7.43763387e-01
-3.80727977e-01 4.77279991e-01 -8.64031315e-01 6.80591345e-01
1.13168740e+00 -2.20619634e-01 1.68774232e-01 -2.71542855e-02
1.44534603e-01 -1.04570091e-01 -1.05981755e+00 1.37884092e+00
-3.61099273e-01 -6.01769164e-02 3.36352766e-01 -6.56635702e-01
7.77585864e-01 3.41266692e-01 6.34906769e-01 -5.92237830e-01
9.74654183e-02 1.07093468e-01 -5.56156039e-01 -5.33252418e-01
-2.87067965e-02 -1.50594175e-01 -7.68884793e-02 6.91718817e-01
-4.31066528e-02 7.19133198e-01 -4.74368989e-01 1.44619420e-01
1.10674584e+00 -2.00342968e-01 -5.41385077e-02 -1.43013522e-01
5.83954275e-01 -3.09324831e-01 1.05351162e+00 8.34514380e-01
-3.83256286e-01 5.17651081e-01 2.49439359e-01 -7.23416686e-01
-8.17310154e-01 -9.83932316e-01 1.15744881e-01 1.59181690e+00
6.24046810e-02 -2.25150287e-01 -7.46493518e-01 -1.19803905e+00
7.54350603e-01 7.14819968e-01 -7.23240316e-01 -2.73079157e-01
-3.11900586e-01 -6.10016763e-01 2.66497016e-01 3.37972969e-01
5.63428819e-01 -9.24550951e-01 1.57237574e-01 9.68809724e-02
-6.96996367e-03 -6.66240931e-01 -1.07899356e+00 -1.60292596e-01
-4.82123911e-01 -1.11650193e+00 -5.58485508e-01 -6.14250958e-01
5.82583487e-01 2.35489190e-01 6.70636714e-01 -2.34133840e-01
-1.99748039e-01 3.39528292e-01 -3.04208308e-01 -2.73913413e-01
-2.26981416e-01 3.57611738e-02 -1.13265976e-01 8.12382340e-01
3.74891371e-01 -6.64405406e-01 -8.89402747e-01 4.48218763e-01
-1.08603573e+00 -3.83001119e-01 3.47047865e-01 7.93182254e-01
7.36146510e-01 2.36105956e-02 8.39891791e-01 -1.48640859e+00
3.85695279e-01 -9.98841643e-01 -3.81313950e-01 4.04957503e-01
-8.42131972e-01 -1.14495143e-01 1.14700556e+00 -4.32038516e-01
-1.20853257e+00 4.14511785e-02 1.38415471e-01 -7.20680654e-01
-6.06069416e-02 -1.08868368e-01 -9.80213821e-01 -1.83840752e-01
4.81345803e-01 3.44362676e-01 2.69392490e-01 -8.50610554e-01
5.31718969e-01 9.50560868e-01 5.36761582e-01 -7.03223705e-01
8.73623371e-01 5.33291817e-01 -3.07089537e-01 7.11483806e-02
-7.35718489e-01 -2.84783691e-01 -2.08792184e-03 1.93460181e-01
1.87131628e-01 -1.02298260e+00 -8.06467414e-01 8.45541596e-01
-6.48993909e-01 -4.63715732e-01 -7.10315704e-01 1.09828688e-01
-7.08766222e-01 2.33908966e-01 -5.10951221e-01 -5.75483322e-01
-8.61946762e-01 -9.68152821e-01 5.70819438e-01 4.03004289e-01
3.12776625e-01 -8.60398889e-01 1.54625624e-01 6.33547962e-01
7.32149541e-01 2.53896266e-01 7.51170576e-01 -1.06051922e+00
-6.91615820e-01 -4.90495741e-01 -2.77822401e-04 5.72795928e-01
4.21701431e-01 -3.19686532e-01 -1.16955590e+00 -5.33496916e-01
1.89344026e-02 -3.61782372e-01 4.59640682e-01 -5.68483286e-02
1.73814774e+00 -1.22513103e+00 -1.41658679e-01 9.90884960e-01
1.61633825e+00 -1.90760046e-01 3.02960694e-01 3.24561447e-01
7.10270345e-01 4.36891288e-01 6.53405607e-01 1.10390425e+00
6.48872375e-01 3.99969637e-01 8.32153201e-01 2.49524370e-01
1.59501135e-01 -4.29786623e-01 2.66188234e-01 6.73549026e-02
6.26844227e-01 -2.75903404e-01 -5.46599209e-01 6.85972273e-01
-2.14122128e+00 -9.64928985e-01 1.93457335e-01 2.23290229e+00
9.48309302e-01 -7.29747713e-01 2.09292307e-01 -3.35054696e-01
9.40061986e-01 1.20254681e-01 -1.12135136e+00 -3.31495941e-01
-1.94287106e-01 -1.30302712e-01 7.46665061e-01 2.76840121e-01
-1.02186418e+00 5.93414128e-01 4.97127867e+00 9.12019849e-01
-1.08126354e+00 4.48556036e-01 8.64605844e-01 -3.61794174e-01
-6.39214575e-01 -1.19424559e-01 -8.16455305e-01 8.06310654e-01
6.51260197e-01 -8.92864883e-01 7.43453145e-01 1.26929212e+00
-6.76286519e-02 7.06775725e-01 -1.08619595e+00 7.79725850e-01
-2.21432656e-01 -1.34940493e+00 1.53319404e-01 1.81069329e-01
1.14667034e+00 3.81335348e-01 4.42377955e-01 3.91578257e-01
7.09370852e-01 -8.14061940e-01 4.39717561e-01 7.17048109e-01
6.82158709e-01 -1.05400479e+00 6.75861895e-01 5.15324593e-01
-8.32506657e-01 -6.05552077e-01 -2.39381313e-01 3.88114482e-01
-3.06710660e-01 4.50929523e-01 -4.42469835e-01 6.34163082e-01
1.00793242e+00 3.11988056e-01 -3.33277076e-01 1.04355848e+00
2.57568240e-01 5.61086833e-01 -2.19234884e-01 4.38889265e-01
-1.44250868e-02 -1.18156765e-02 3.93968284e-01 9.38304126e-01
1.10451765e-01 -6.08982816e-02 1.74893349e-01 6.37478411e-01
-6.90432072e-01 5.02660036e-01 -3.38640809e-01 1.79065451e-01
6.69334054e-01 1.21252108e+00 5.85100055e-01 -1.58049479e-01
-6.05905175e-01 9.12758648e-01 5.58320940e-01 3.57727945e-01
-6.53524220e-01 -1.87269211e-01 1.22340691e+00 2.31909081e-01
5.39295614e-01 6.10642612e-01 -7.99364150e-02 -1.23741078e+00
2.47026667e-01 -1.09142315e+00 1.04408371e+00 -1.54314399e-01
-2.15869617e+00 5.06463170e-01 -5.47583282e-01 -1.14910305e+00
-5.45246936e-02 -2.96534374e-02 -1.00083995e+00 1.13478613e+00
-1.55871725e+00 -1.47119033e+00 -4.67866302e-01 1.36082673e+00
1.28238589e-01 -3.31808656e-01 9.09813821e-01 4.58476663e-01
-6.34421051e-01 1.39382672e+00 8.89645636e-01 3.32281962e-02
9.70032096e-01 -9.45707917e-01 7.95293823e-02 4.21762437e-01
-4.81945038e-01 5.01092374e-01 4.55075502e-01 -3.86030406e-01
-1.67723894e+00 -1.84103477e+00 6.44609809e-01 -1.59054399e-01
4.55761164e-01 -1.73875585e-01 -1.18750727e+00 1.01754117e+00
5.64755201e-02 8.65625918e-01 1.12307870e+00 -3.05563360e-01
-8.53556216e-01 -7.21853018e-01 -2.02462268e+00 2.15530127e-01
8.05061281e-01 -4.33486432e-01 8.28911364e-02 5.23318350e-01
8.84961486e-01 -1.46586224e-01 -1.14028287e+00 2.89133996e-01
4.90792841e-01 -9.39053655e-01 5.71928501e-01 -1.11286056e+00
6.57321652e-03 -1.26007885e-01 -5.96413434e-01 -1.08049417e+00
-3.49342078e-01 -8.85743737e-01 -4.97961372e-01 1.64055812e+00
4.34392802e-02 -1.26276672e+00 1.10090590e+00 1.08911717e+00
2.43693382e-01 -9.78202045e-01 -9.79552031e-01 -6.58116579e-01
3.03962588e-01 9.84256268e-02 1.47810698e+00 1.07531512e+00
-3.24975371e-01 -4.81426537e-01 -3.87952238e-01 3.73252600e-01
1.01053882e+00 5.80218911e-01 1.16913402e+00 -6.96365893e-01
-5.35956681e-01 -1.08922593e-01 -2.76617169e-01 -4.89493638e-01
3.76525223e-01 -1.08071113e+00 -3.57398719e-01 -1.04453790e+00
5.41223407e-01 -6.74636662e-01 -6.27379298e-01 7.81341970e-01
-3.22284162e-01 -8.56561437e-02 2.90621340e-01 3.16213965e-01
-5.96406996e-01 9.13732469e-01 1.08187425e+00 -5.79541884e-02
8.47304836e-02 3.79012883e-01 -1.22141719e+00 2.44236410e-01
1.02272201e+00 -5.65671325e-01 -3.81064385e-01 -3.99485081e-01
-5.55967808e-01 -8.92612115e-02 3.45297456e-01 -5.04592836e-01
2.92758733e-01 -5.15055597e-01 2.67206252e-01 -2.92349994e-01
-8.08433667e-02 -1.16795111e+00 4.08575565e-01 3.44745040e-01
-3.67122114e-01 -3.14641565e-01 -1.40786488e-02 7.07349598e-01
-9.22824889e-02 2.68345118e-01 8.05830836e-01 1.65121220e-02
-2.65911549e-01 1.17501545e+00 3.25160533e-01 1.41453907e-01
1.33272660e+00 3.50077569e-01 -7.40744114e-01 -6.00475311e-01
-6.10235751e-01 6.94040418e-01 7.21321881e-01 3.54331315e-01
3.12299371e-01 -1.48638511e+00 -8.62005770e-01 4.78960782e-01
1.81172878e-01 1.34935185e-01 7.19870329e-01 3.52545738e-01
-1.58481464e-01 3.23122256e-02 -8.85642543e-02 -1.29216835e-01
-1.21878421e+00 9.47180927e-01 6.15413845e-01 -2.40789309e-01
-4.17708129e-01 8.38520050e-01 5.84507108e-01 -9.22301412e-01
4.58630174e-01 5.90533137e-01 3.90004247e-01 -3.65365207e-01
7.97067046e-01 5.79370439e-01 -2.92781219e-02 -6.94487512e-01
-3.02989274e-01 6.58420175e-02 -4.60864812e-01 3.75090629e-01
1.56708610e+00 -1.28903061e-01 -1.07169628e-01 -2.03615557e-02
1.70617771e+00 2.20191464e-01 -1.55810618e+00 -8.19900751e-01
-6.37979507e-01 -9.80319321e-01 -2.01585874e-01 -8.22971582e-01
-1.71206641e+00 4.05573547e-01 7.42255509e-01 -1.44147873e-01
1.26665032e+00 -3.61461565e-02 1.00638127e+00 -6.29163384e-02
5.18569052e-01 -5.63310087e-01 -3.73801172e-01 -3.09862271e-02
9.61080968e-01 -1.15856123e+00 -2.30845451e-01 -2.53694087e-01
-8.09545457e-01 6.13234282e-01 8.01982760e-01 -1.63881108e-01
8.79780710e-01 2.12485373e-01 2.81996101e-01 2.75401980e-01
-1.02816212e+00 3.95917773e-01 -1.61406159e-01 5.40872514e-01
-2.05096036e-01 2.08171219e-01 -7.49212652e-02 1.26590753e+00
1.16004879e-02 1.25128686e-01 2.29770155e-03 1.11292219e+00
3.25019099e-03 -1.51615858e+00 -4.24438089e-01 5.34176528e-01
-9.11001742e-01 3.53514373e-01 -3.13020378e-01 3.33627790e-01
2.52088010e-01 7.27348208e-01 -1.11854970e-01 -4.58987534e-01
2.96458423e-01 1.35377273e-01 1.06633734e-02 -3.11581194e-01
-9.58485484e-01 -2.52200186e-01 -3.75607103e-01 -9.33556914e-01
2.56769270e-01 -7.31278241e-01 -1.18695211e+00 -7.70174921e-01
-7.51083791e-02 3.47821534e-01 5.42206645e-01 4.64189619e-01
9.08613741e-01 4.60090069e-03 1.52601075e+00 -3.66402656e-01
-1.24927258e+00 -6.38334394e-01 -8.34246218e-01 6.62986100e-01
5.31033993e-01 -1.09458767e-01 -6.02118790e-01 -1.21453203e-01] | [5.837131977081299, 6.360169410705566] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.