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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0a1e6660-42a8-4563-b1f4-9deae33fa8db | clip-forge-towards-zero-shot-text-to-shape | 2110.02624 | null | https://arxiv.org/abs/2110.02624v2 | https://arxiv.org/pdf/2110.02624v2.pdf | CLIP-Forge: Towards Zero-Shot Text-to-Shape Generation | Generating shapes using natural language can enable new ways of imagining and creating the things around us. While significant recent progress has been made in text-to-image generation, text-to-shape generation remains a challenging problem due to the unavailability of paired text and shape data at a large scale. We present a simple yet effective method for zero-shot text-to-shape generation that circumvents such data scarcity. Our proposed method, named CLIP-Forge, is based on a two-stage training process, which only depends on an unlabelled shape dataset and a pre-trained image-text network such as CLIP. Our method has the benefits of avoiding expensive inference time optimization, as well as the ability to generate multiple shapes for a given text. We not only demonstrate promising zero-shot generalization of the CLIP-Forge model qualitatively and quantitatively, but also provide extensive comparative evaluations to better understand its behavior. | ['Kamal Rahimi Malekshan', 'Marco Fumero', 'Chin-Yi Cheng', 'Ye Wang', 'Joseph G. Lambourne', 'Hang Chu', 'Aditya Sanghi'] | 2021-10-06 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Sanghi_CLIP-Forge_Towards_Zero-Shot_Text-To-Shape_Generation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Sanghi_CLIP-Forge_Towards_Zero-Shot_Text-To-Shape_Generation_CVPR_2022_paper.pdf | cvpr-2022-1 | ['text-to-shape-generation'] | ['computer-vision'] | [ 6.26729131e-01 1.44330427e-01 2.80935287e-01 -2.44778648e-01
-1.00912488e+00 -7.59210646e-01 9.80732441e-01 -1.63747072e-01
6.08126745e-02 7.31242418e-01 1.51592836e-01 -2.56935477e-01
1.88163489e-01 -9.83549654e-01 -7.02833593e-01 -6.28688574e-01
5.75719237e-01 6.23299241e-01 5.03058545e-02 -3.45800996e-01
1.85107157e-01 4.94477451e-01 -1.48926413e+00 2.18610615e-01
9.71737981e-01 9.72888350e-01 2.93097675e-01 6.29491389e-01
-3.73030305e-01 4.86867428e-01 -4.64491755e-01 -7.90686905e-01
4.54516500e-01 -5.68000376e-01 -4.16074783e-01 4.98063356e-01
4.80178952e-01 -2.62860268e-01 -3.99860069e-02 7.59783566e-01
7.53915668e-01 1.69059440e-01 9.82221961e-01 -1.17158806e+00
-9.90939677e-01 3.48201215e-01 -6.08468950e-01 -5.69098592e-01
1.98149979e-01 3.72134268e-01 8.46111000e-01 -1.06667066e+00
8.38154554e-01 1.18091762e+00 4.66279209e-01 7.88247347e-01
-1.26229417e+00 -4.35654432e-01 -4.95409042e-01 -3.56929630e-01
-1.29783845e+00 -5.30695677e-01 9.23182964e-01 -3.88800532e-01
5.72821319e-01 2.02961296e-01 7.35754311e-01 9.19233263e-01
-1.14106908e-01 8.79579604e-01 9.25432980e-01 -5.69473207e-01
3.35150450e-01 1.71299666e-01 -7.51878619e-01 6.91759169e-01
1.40397996e-01 -7.42150098e-02 -2.67584234e-01 -9.56094190e-02
1.24750555e+00 -1.23534463e-01 9.24316049e-03 -5.12942672e-01
-1.20738399e+00 7.06234396e-01 2.46773794e-01 2.53408313e-01
-2.27506429e-01 2.89310306e-01 1.98303938e-01 7.62054324e-02
5.45935094e-01 5.69439054e-01 1.72261912e-02 -2.34848976e-01
-1.12450647e+00 4.05611515e-01 7.50852168e-01 1.13206446e+00
5.95249951e-01 4.42064136e-01 -4.02376205e-01 9.28369701e-01
-9.94020253e-02 7.47081995e-01 3.96086991e-01 -6.30622625e-01
5.43484628e-01 5.51636696e-01 2.67229676e-01 -8.90595198e-01
2.32360307e-02 -9.55779478e-02 -9.07464147e-01 2.68864363e-01
3.98238391e-01 -3.64708036e-01 -1.14677882e+00 1.34493828e+00
2.72141933e-01 -1.16642550e-01 -2.42170431e-02 7.34212518e-01
7.53176451e-01 7.31308043e-01 -2.12722257e-01 1.02880679e-01
1.12219334e+00 -1.00497890e+00 -5.39357722e-01 -1.61633998e-01
4.90383923e-01 -1.01650441e+00 1.28678131e+00 1.48024382e-02
-1.39506364e+00 -3.84751409e-01 -9.54178989e-01 -2.47625187e-01
-3.74494195e-01 2.92413622e-01 6.45457387e-01 7.52253175e-01
-1.06274176e+00 4.22965407e-01 -3.09292138e-01 -3.67882937e-01
8.26011896e-01 6.75657466e-02 -2.55926549e-01 1.05476789e-02
-6.54249787e-01 5.48308730e-01 2.55737901e-01 -3.08621258e-01
-6.71462595e-01 -5.84474623e-01 -8.38307142e-01 1.00083232e-01
4.68701720e-01 -8.08045566e-01 1.23986793e+00 -9.32727993e-01
-1.66291976e+00 8.41276705e-01 -2.17143357e-01 -3.00345093e-01
6.57351971e-01 1.53320774e-01 -3.10144778e-02 8.93642977e-02
-5.17973192e-02 1.20406437e+00 1.11357605e+00 -1.55059540e+00
-2.18670800e-01 -7.81546384e-02 -1.69040915e-02 2.25742415e-01
-3.74135077e-01 -1.76571056e-01 -4.59885299e-01 -1.07821381e+00
2.36659460e-02 -8.01076531e-01 -4.05695170e-01 4.37244266e-01
-3.98177445e-01 -1.04399353e-01 8.43443274e-01 -2.88269043e-01
8.35739136e-01 -1.85824192e+00 -2.13750213e-01 3.80514637e-02
1.00614078e-01 3.36643934e-01 -4.14706826e-01 7.90930390e-01
1.57298997e-01 3.42178911e-01 -5.02220988e-01 -5.87144673e-01
-4.06365804e-02 4.85981256e-02 -5.96968293e-01 5.12935556e-02
4.85605478e-01 1.42156565e+00 -9.10270512e-01 -7.22164392e-01
2.79690951e-01 4.38753337e-01 -5.11140108e-01 1.14681631e-01
-6.24054849e-01 4.23359394e-01 -3.61827612e-01 7.31229424e-01
7.15808928e-01 -3.40970427e-01 -1.27035126e-01 -1.16897926e-01
-1.55259028e-01 -2.01655969e-01 -8.71025383e-01 1.75599027e+00
-4.87436593e-01 5.11479616e-01 -2.44864300e-01 -8.15592170e-01
1.19698393e+00 3.91570389e-01 3.68848771e-01 -6.82153761e-01
2.81904995e-01 3.58113885e-01 -2.55955845e-01 -5.21043658e-01
7.27210701e-01 -5.26497483e-01 -3.95510793e-02 8.60468984e-01
-2.21321844e-02 -8.38719785e-01 3.95844996e-01 2.74180114e-01
7.42360771e-01 4.40494627e-01 2.37569794e-01 -9.59906355e-02
2.35248953e-01 -7.73997381e-02 1.83077440e-01 6.00451767e-01
2.65090823e-01 1.03315532e+00 2.76333034e-01 -3.22418302e-01
-1.64012206e+00 -1.07442391e+00 5.72101548e-02 7.99092889e-01
-8.75227526e-03 -1.31293356e-01 -7.74117410e-01 -4.78257984e-01
-2.22950384e-01 9.04868364e-01 -4.25172776e-01 2.81082928e-01
-4.83675033e-01 -5.02276778e-01 5.16798615e-01 5.30389786e-01
3.83830607e-01 -1.28588879e+00 -5.14116168e-01 8.15960616e-02
-1.99034795e-01 -1.12308311e+00 -7.37344146e-01 -3.41944307e-01
-9.87448692e-01 -5.37811875e-01 -1.24830306e+00 -9.22676682e-01
1.16694474e+00 5.95977366e-01 1.01156425e+00 8.76270086e-02
-5.93153954e-01 2.66128212e-01 -3.36384803e-01 -6.76645875e-01
-5.81774414e-01 -2.23457620e-01 -4.41947281e-01 1.45893991e-01
-1.39792621e-01 -7.32111692e-01 -5.75993717e-01 2.65742272e-01
-1.23230934e+00 6.08384132e-01 7.84414828e-01 9.32449639e-01
5.42431414e-01 -1.22359760e-01 8.19652796e-01 -7.53946185e-01
7.42223859e-01 -1.40875429e-01 -5.57727754e-01 3.05784762e-01
-2.57894993e-01 1.30693063e-01 7.44732976e-01 -3.92341018e-01
-1.31999087e+00 3.29571009e-01 -7.73887411e-02 -4.45473731e-01
-8.78898725e-02 1.76062480e-01 -1.23840451e-01 -1.20570935e-01
7.42318749e-01 5.47288716e-01 1.68004736e-01 -2.78277695e-01
9.17182744e-01 6.54245198e-01 5.71158707e-01 -7.34638453e-01
1.14694023e+00 7.04394877e-01 5.34553006e-02 -9.92190421e-01
-6.09304249e-01 -1.89257771e-01 -7.55467057e-01 -2.51010537e-01
7.73323119e-01 -6.97181225e-01 -4.62707043e-01 5.04334450e-01
-1.29301870e+00 -3.72307420e-01 -5.85426033e-01 -2.39467561e-01
-8.70215774e-01 3.83783549e-01 -2.75802314e-01 -1.02019310e+00
-6.16090775e-01 -7.83436060e-01 1.38372076e+00 1.88855976e-01
2.06375178e-02 -1.03714001e+00 -1.70758590e-01 4.34186608e-01
5.43245256e-01 3.97962928e-01 9.63468194e-01 -2.31027111e-01
-9.06379998e-01 -3.15507978e-01 -6.07024848e-01 1.81098044e-01
2.75830179e-01 -1.73601255e-01 -8.11934054e-01 -1.70845240e-01
-1.74148113e-01 -7.65651286e-01 5.37715614e-01 8.05837661e-02
1.14289737e+00 -4.98824388e-01 -1.94576308e-01 3.14316660e-01
1.46006358e+00 1.78846285e-01 8.05967271e-01 -2.17476577e-01
7.45674431e-01 6.20220244e-01 3.24904114e-01 5.14087498e-01
3.23103696e-01 6.30236447e-01 2.11667627e-01 -3.86830360e-01
-3.95578980e-01 -5.92434406e-01 -6.70378134e-02 8.64329696e-01
-1.18197992e-01 -4.61559713e-01 -6.52030349e-01 8.12906861e-01
-1.69670236e+00 -1.03250110e+00 1.37336021e-02 2.35582852e+00
9.75970149e-01 -2.92504821e-02 -6.83171675e-02 5.24725467e-02
6.68244362e-01 2.01142684e-01 -4.25871670e-01 -2.68115848e-01
-1.59174696e-01 3.82320851e-01 1.36138678e-01 1.89090893e-01
-7.85926521e-01 1.15436518e+00 6.22162771e+00 1.24079609e+00
-1.15626681e+00 -1.35986045e-01 6.78753853e-01 -8.65303795e-04
-6.45093203e-01 5.82365552e-03 -4.77206618e-01 2.28585973e-01
3.64640504e-01 -5.19821227e-01 5.03551066e-01 7.75250554e-01
2.97852218e-01 -7.76564479e-02 -8.88838112e-01 1.14651668e+00
4.26985323e-01 -1.62033653e+00 4.91139799e-01 -1.08798228e-01
1.15859747e+00 -5.32305777e-01 -6.04843274e-02 5.13826171e-03
1.63383886e-01 -1.02886665e+00 7.78512180e-01 5.13044238e-01
1.31150126e+00 -6.36471689e-01 2.24812880e-01 4.88272130e-01
-1.27585995e+00 1.64712772e-01 -4.82140720e-01 1.67012349e-01
3.68786186e-01 7.27754474e-01 -1.00441980e+00 5.21374464e-01
-1.74753536e-02 4.24253494e-01 -4.00616169e-01 1.07945573e+00
-1.74458116e-01 2.33473092e-01 -1.48819849e-01 -2.09321737e-01
1.28148630e-01 -2.10276991e-01 4.39348429e-01 9.78751481e-01
7.39907146e-01 1.77469939e-01 1.94117486e-01 1.33561325e+00
-2.53394276e-01 3.63950372e-01 -9.65365887e-01 -4.43425268e-01
4.41689968e-01 1.49673557e+00 -9.55660403e-01 -5.27622223e-01
-2.97387242e-01 1.11891842e+00 2.06156909e-01 2.23475888e-01
-6.16043687e-01 -6.08358264e-01 1.48397812e-04 3.32926095e-01
2.77302146e-01 -3.69268537e-01 -6.09103620e-01 -1.04684329e+00
1.09499238e-01 -7.45991409e-01 -2.05996558e-01 -1.21889710e+00
-1.36072874e+00 4.51980591e-01 -2.56450534e-01 -1.44149101e+00
-2.20463082e-01 -4.06083018e-01 -9.84200001e-01 7.18529820e-01
-1.11546504e+00 -1.56524444e+00 -3.90252203e-01 3.64475846e-01
8.91332567e-01 -2.25740984e-01 8.08618665e-01 3.24766785e-02
-1.86628133e-01 6.97066128e-01 9.26411301e-02 1.19503103e-01
5.41476130e-01 -1.06980228e+00 8.80963743e-01 6.56623900e-01
1.73291788e-01 3.83930445e-01 6.17107987e-01 -7.50694335e-01
-1.45096326e+00 -1.23081243e+00 7.79097319e-01 -2.49699324e-01
4.43394512e-01 -6.10507905e-01 -5.56320012e-01 3.43702763e-01
2.19075769e-01 -1.77364856e-01 2.94943064e-01 -3.88616920e-01
-1.15260661e-01 1.41409382e-01 -1.18286037e+00 1.07978272e+00
1.17902160e+00 -2.82353312e-01 -4.77853477e-01 3.38843614e-01
7.31107354e-01 -4.49632555e-01 -5.15839577e-01 2.09250242e-01
4.99803692e-01 -8.72951269e-01 9.48502064e-01 -2.22637221e-01
9.81397092e-01 -4.89673018e-02 1.25972033e-01 -1.29109502e+00
-4.03165184e-02 -9.19955552e-01 1.43327326e-01 1.26171875e+00
4.48378652e-01 -4.58687067e-01 1.01576209e+00 6.26220942e-01
-1.17618077e-01 -9.99411881e-01 -6.74540639e-01 -1.02448153e+00
6.36729449e-02 -2.48170525e-01 7.11189747e-01 7.19757974e-01
-4.11909483e-02 5.43758452e-01 -5.98374367e-01 -5.28739035e-01
6.90979958e-01 4.67401952e-01 1.06286883e+00 -1.06588578e+00
-2.14487195e-01 -3.94441724e-01 -1.27075851e-01 -1.23867226e+00
-2.55027562e-01 -1.04818034e+00 3.08040261e-01 -1.85071254e+00
3.64641219e-01 -5.29080272e-01 3.77313256e-01 4.96832252e-01
-5.11309542e-02 5.22826850e-01 3.72498453e-01 2.27945894e-01
-4.25813913e-01 9.37973619e-01 2.03486943e+00 -1.18640192e-01
-1.69278860e-01 -1.44685537e-01 -7.61952043e-01 5.39819121e-01
7.12265849e-01 -3.05840194e-01 -6.49448216e-01 -3.60406190e-01
2.31330186e-01 2.52877712e-01 3.99823934e-01 -1.04855525e+00
5.82830869e-02 -2.73964733e-01 3.80438060e-01 -8.09374392e-01
5.76779246e-01 -4.92515475e-01 -1.92692429e-02 1.60088241e-01
-1.79228634e-01 -8.75355974e-02 1.66031212e-01 4.62630898e-01
5.73063307e-02 -4.80429351e-01 7.98375130e-01 -3.47557068e-01
-2.58070529e-01 3.20271224e-01 -1.44150808e-01 1.65185258e-01
9.90074754e-01 -3.56469572e-01 -4.03627038e-01 -7.25000203e-01
-2.73168653e-01 1.74161773e-02 6.06807649e-01 5.14627576e-01
7.30296016e-01 -1.63387537e+00 -7.92556822e-01 1.94653556e-01
1.51828662e-01 9.71544534e-02 2.43108675e-01 3.73417348e-01
-3.62159044e-01 4.96167719e-01 -1.82887793e-01 -3.47980082e-01
-9.91530955e-01 5.32060206e-01 -4.14240919e-03 -1.51408523e-01
-7.27144957e-01 6.47247314e-01 4.91056472e-01 -4.45709378e-01
-1.17325641e-01 7.51597062e-02 1.16648085e-01 -2.08056629e-01
5.19207597e-01 2.58032382e-01 -2.35512838e-01 -5.10223031e-01
3.08261544e-01 6.54544830e-01 1.12085141e-01 -4.40675944e-01
1.17652321e+00 2.94402204e-02 -1.07440159e-01 2.83458084e-01
8.48169625e-01 4.85048182e-02 -1.17905366e+00 -1.02812953e-01
-3.83163244e-01 -6.19585335e-01 -2.67911822e-01 -7.61232734e-01
-1.03518391e+00 9.58726346e-01 2.12360263e-01 2.13495985e-01
8.40299249e-01 -1.95816770e-01 1.30139387e+00 4.60108906e-01
5.48602819e-01 -1.13501501e+00 7.12146461e-01 3.98493111e-01
1.30013824e+00 -1.23388135e+00 -2.00801753e-02 -5.95280647e-01
-6.84828103e-01 1.12860656e+00 5.55012882e-01 3.11956629e-02
3.43353301e-01 2.41012469e-01 1.18836887e-01 -1.14304319e-01
-6.97737813e-01 -2.11471856e-01 2.85222977e-01 6.22371614e-01
3.80555421e-01 2.73444559e-02 -4.45118517e-01 4.04768586e-01
-2.73217231e-01 3.06013435e-01 6.61033452e-01 9.27898467e-01
-4.93205547e-01 -1.27791429e+00 -3.87742221e-01 5.95887959e-01
-1.17366023e-01 -2.42872119e-01 -5.14010429e-01 3.72296274e-01
-2.73205750e-02 8.08926344e-01 -2.90609449e-02 -8.60814750e-02
7.39376694e-02 1.09809004e-01 5.79789042e-01 -6.61164165e-01
-2.27891877e-01 1.29663169e-01 2.73669623e-02 -1.10157780e-01
-1.55546084e-01 -3.51002485e-01 -1.22745883e+00 -2.31616169e-01
-3.71240199e-01 -2.85123348e-01 7.68918633e-01 7.98245549e-01
6.34088159e-01 2.71997809e-01 7.21142054e-01 -1.33758855e+00
-3.86317044e-01 -8.33038330e-01 -4.67203796e-01 6.15058064e-01
-1.60332948e-01 -4.77503121e-01 -6.25366867e-02 4.93106246e-01] | [11.346632957458496, -0.2715771794319153] |
780c0c08-4b3d-4de4-be8c-32fecfd01bd7 | jpeg-xt-image-compression-with-hue | 1904.11315 | null | http://arxiv.org/abs/1904.11315v1 | http://arxiv.org/pdf/1904.11315v1.pdf | JPEG XT Image Compression with Hue Compensation for Two-Layer HDR Coding | We propose a novel JPEG XT image compression with hue compensation for
two-layer HDR coding. LDR images produced from JPEG XT bitstreams have some
distortion in hue due to tone mapping operations. In order to suppress the
color distortion, we apply a novel hue compensation method based on the
maximally saturated colors. Moreover, the bitstreams generated by using the
proposed method are fully compatible with the JPEG XT standard. In an
experiment, the proposed method is demonstrated not only to produce images with
small hue degradation but also to maintain well-mapped luminance, in terms of
three kinds of criterion: TMQI, hue value in CIEDE2000, and the maximally
saturated color on the constant-hue plane. | ['Hitoshi Kiya', 'Hiroyuki Kobayashi'] | 2019-04-25 | null | null | null | null | ['tone-mapping'] | ['computer-vision'] | [ 4.31878716e-01 -4.37837631e-01 -1.05117261e-01 -1.24291539e-01
-1.36328146e-01 -2.22306296e-01 1.43057734e-01 -2.41918996e-01
-3.67739439e-01 8.64864528e-01 -9.01165083e-02 -1.33556142e-01
2.10401058e-01 -9.54065859e-01 -2.43296176e-01 -7.28527129e-01
1.41148552e-01 -3.23321909e-01 3.77888501e-01 -3.07668477e-01
3.97702694e-01 3.28819811e-01 -1.77915430e+00 -1.25414215e-03
1.07617021e+00 1.30184281e+00 2.60324150e-01 7.80505836e-01
1.69098794e-01 9.82127428e-01 -6.69598818e-01 -3.97529870e-01
3.81745994e-01 -7.79470563e-01 -9.53921899e-02 2.17040241e-01
-2.73996815e-02 -7.52734363e-01 -5.04626811e-01 1.43127728e+00
2.74276137e-01 -2.77710348e-01 5.52124143e-01 -1.05796063e+00
-8.53715599e-01 6.35031164e-02 -8.11083853e-01 -1.85062438e-01
2.23639533e-01 1.29558235e-01 6.22300565e-01 -2.87678897e-01
5.15794814e-01 1.00423229e+00 1.03814207e-01 2.46491119e-01
-1.09062088e+00 -5.69941580e-01 -5.77246666e-01 5.99326432e-01
-1.72799885e+00 -3.63374531e-01 9.43136215e-01 6.24737404e-02
5.70082426e-01 6.52387798e-01 7.47420907e-01 2.50475824e-01
6.29643321e-01 1.00755319e-01 1.66418076e+00 -5.96148968e-01
2.09862396e-01 2.15240091e-01 -4.30118293e-01 7.33314395e-01
3.58053446e-01 -8.50605890e-02 -2.52710015e-01 2.55257070e-01
1.16460299e+00 -3.59920382e-01 -6.63091660e-01 -2.24543586e-01
-1.10920870e+00 5.53998113e-01 3.60129774e-01 3.19616139e-01
-1.46871030e-01 -1.17406361e-01 1.79871842e-01 3.27970028e-01
1.35349974e-01 1.10518418e-01 2.17464224e-01 -1.66848332e-01
-8.96458507e-01 -3.97168487e-01 6.25080705e-01 1.12190735e+00
8.22261095e-01 2.95307517e-01 9.36040059e-02 8.92899334e-01
5.97688079e-01 9.19939518e-01 3.89855295e-01 -1.03696930e+00
2.75520384e-01 3.01718414e-01 1.73090830e-01 -8.93160939e-01
3.55536081e-02 2.35529020e-01 -1.17127943e+00 7.80397117e-01
-1.20135076e-01 2.55395591e-01 -1.11671555e+00 1.33802366e+00
-2.56224543e-01 -3.15382689e-01 5.05732119e-01 1.04059911e+00
5.26777804e-01 1.05818272e+00 -9.96435657e-02 -7.38554776e-01
1.17883670e+00 -5.98771095e-01 -1.30736017e+00 2.07268700e-01
-1.08749360e-01 -1.05107582e+00 1.25468731e+00 8.52289200e-01
-1.39837921e+00 -7.41165459e-01 -1.76480937e+00 -3.57456356e-01
-1.43446997e-01 1.78218111e-01 6.04027286e-02 1.04111922e+00
-1.01983702e+00 2.04037547e-01 -1.78225592e-01 -6.90152645e-02
-2.17838705e-01 2.78635453e-02 -1.02514856e-01 -2.55602539e-01
-1.42692995e+00 8.13272893e-01 6.07845485e-01 7.91127086e-02
-2.85311520e-01 -1.33611813e-01 -6.42453611e-01 2.93286920e-01
-7.45716393e-02 -3.39290984e-02 6.96278632e-01 -7.72538602e-01
-1.85307372e+00 7.26843596e-01 1.40603915e-01 -2.04475656e-01
7.01652408e-01 8.17785934e-02 -9.47961092e-01 6.84856176e-01
-4.27324176e-01 4.88495469e-01 7.39582956e-01 -1.35824442e+00
-5.22387981e-01 -1.17695332e-02 -2.05327064e-01 3.49759072e-01
-3.26202005e-01 -3.38512063e-01 -7.65466750e-01 -4.83167410e-01
1.88375577e-01 -5.60544610e-01 1.70721024e-01 1.60979494e-01
-3.99723172e-01 4.43006307e-01 1.13154626e+00 -8.10597122e-01
1.31174791e+00 -2.42923021e+00 -3.08865160e-01 3.93959850e-01
-2.96431091e-02 2.60141730e-01 2.18095168e-01 1.36395365e-01
1.92577913e-01 1.05944365e-01 -4.09834415e-01 2.90083140e-01
3.75891253e-02 6.37163743e-02 -9.92311761e-02 4.11306173e-01
-2.65157044e-01 4.59013954e-02 -5.03154159e-01 -7.93377519e-01
5.80367804e-01 6.53878033e-01 -2.25434691e-01 3.82425457e-01
-2.90277228e-03 -1.32397460e-02 1.58855811e-01 6.93599999e-01
1.31938100e+00 1.90281376e-01 3.89464468e-01 -6.00530922e-01
-4.94122773e-01 -3.52579147e-01 -1.04066157e+00 1.12937307e+00
-2.38287672e-01 5.78344941e-01 -6.73236027e-02 -2.31208801e-01
1.44952118e+00 2.71909148e-01 3.56828392e-01 -1.27159679e+00
2.00187660e-05 4.18972373e-01 -1.55475318e-01 -4.00842041e-01
9.06924605e-01 -2.18557686e-01 2.46178180e-01 -1.43796662e-02
-3.33732486e-01 -4.51446533e-01 5.27679443e-01 -2.66290307e-02
4.23365861e-01 -7.35234842e-02 4.52721328e-01 -3.15435648e-01
7.13906765e-01 -3.02908808e-01 4.80155259e-01 3.62765729e-01
-1.49282426e-01 6.74677968e-01 5.45065165e-01 2.27840673e-02
-1.79282618e+00 -1.25078249e+00 -4.15154457e-01 3.00846845e-01
7.96807528e-01 -1.07025787e-01 -4.24820930e-01 1.77299529e-01
-3.61613274e-01 7.25409627e-01 1.79058395e-03 -2.91166127e-01
-4.55266774e-01 -5.90756893e-01 4.88254666e-01 7.52343759e-02
1.43757474e+00 -1.03157318e+00 -8.21283221e-01 -6.96868123e-03
-2.09672764e-01 -1.02952254e+00 -3.02016228e-01 -4.79623638e-02
-5.89207530e-01 -1.04347324e+00 -1.04811394e+00 -8.01056206e-01
4.33306366e-01 3.31392199e-01 8.33127081e-01 1.86060384e-01
-3.65744323e-01 4.71671298e-02 -5.77469945e-01 2.42608413e-01
-6.14756167e-01 -4.07936424e-01 -2.97496200e-01 -1.48387745e-01
2.08224982e-01 -1.59110650e-01 -8.81174922e-01 3.91582727e-01
-1.26867151e+00 4.30586696e-01 7.40898430e-01 4.45025384e-01
6.22964859e-01 6.83065593e-01 1.32116750e-01 -5.20179093e-01
5.40240526e-01 9.40221027e-02 -9.02862668e-01 5.10723829e-01
-9.37617123e-01 -1.91238612e-01 8.72767150e-01 -4.86120507e-02
-1.45283449e+00 -3.19303840e-01 -2.46513709e-01 -1.76916286e-01
9.15509686e-02 1.12245999e-01 -3.52933735e-01 -2.70763129e-01
2.04869330e-01 4.98099595e-01 9.92302690e-03 -2.11175442e-01
2.99778610e-01 9.41042602e-01 9.11446571e-01 -1.79758534e-01
8.86603475e-01 4.21155423e-01 1.84722900e-01 -1.08263302e+00
4.05286700e-01 3.28757986e-02 -1.53760776e-01 -4.25545007e-01
1.08280361e+00 -8.57163012e-01 -9.17432308e-01 8.82564723e-01
-7.51115203e-01 -1.63825393e-01 -2.56784465e-02 6.65091634e-01
-6.58699751e-01 6.98342443e-01 -1.08658779e+00 -7.21467733e-01
-3.34592551e-01 -1.24224401e+00 5.43549061e-01 5.90083599e-01
5.17975152e-01 -6.88351512e-01 -5.29412664e-02 -1.54881522e-01
5.82652152e-01 2.11762637e-01 1.31199622e+00 5.97731471e-01
-7.61384308e-01 2.09556341e-01 -5.83902359e-01 5.16120493e-01
4.85163957e-01 3.95134389e-01 -6.60291433e-01 -3.77601743e-01
1.67695522e-01 -2.97159344e-01 8.04128826e-01 2.90038317e-01
1.06747496e+00 -1.17478654e-01 2.51697779e-01 9.44055259e-01
2.08183742e+00 9.47057188e-01 1.63835764e+00 6.60215378e-01
3.90542388e-01 4.53743227e-02 7.87642181e-01 6.20895922e-01
5.10012619e-02 5.99034727e-01 3.17435741e-01 -5.87214887e-01
-1.47965774e-01 -2.34274372e-01 2.49994114e-01 9.44006503e-01
-1.76616624e-01 -4.85011399e-01 -3.09157908e-01 -1.60580978e-01
-1.17912996e+00 -8.63803446e-01 -1.48070097e-01 2.35341811e+00
1.01978827e+00 2.75692046e-01 -2.33517423e-01 5.08078873e-01
9.48123634e-01 3.00871581e-01 -3.10404032e-01 -6.20426357e-01
-4.57506210e-01 1.44819468e-01 8.87985349e-01 5.36519468e-01
-7.37233639e-01 5.69426894e-01 6.85481215e+00 7.36655831e-01
-1.40262568e+00 -5.23778379e-01 5.46958387e-01 3.55442792e-01
-4.69050348e-01 -7.43825436e-02 -2.32699484e-01 7.43260264e-01
6.08281970e-01 -3.12297761e-01 5.65781295e-01 5.55192471e-01
1.95283875e-01 -5.05361021e-01 -2.85536140e-01 1.11641526e+00
1.74014375e-01 -8.14998150e-01 1.81239665e-01 -2.69978400e-03
3.96908402e-01 -8.79855931e-01 5.75104833e-01 -1.79341927e-01
-1.12040855e-01 -6.70422316e-01 6.59429669e-01 4.77578819e-01
1.50694978e+00 -9.21439886e-01 6.65686607e-01 -4.18759793e-01
-1.42054653e+00 1.75299823e-01 -8.26862693e-01 4.27348673e-01
2.38833785e-01 4.48310584e-01 -4.91011083e-01 5.18851638e-01
6.07647777e-01 4.70429659e-01 -6.57408774e-01 1.30523169e+00
-2.77285457e-01 1.59126416e-01 -3.63843031e-02 2.18757018e-01
-2.05467250e-02 -7.34141469e-01 1.13035068e-01 1.05672133e+00
7.36710370e-01 1.67296797e-01 -3.38509202e-01 7.09416270e-01
6.41240599e-03 1.39330074e-01 -3.09860885e-01 -5.62349632e-02
4.78560477e-01 8.98877263e-01 -7.29522169e-01 -5.35006344e-01
-4.57793981e-01 1.27192664e+00 -5.38048804e-01 6.70191467e-01
-9.89904702e-01 -9.28996146e-01 2.29163796e-01 -7.58242980e-02
2.61104435e-01 -2.98149377e-01 -1.21374197e-01 -9.55438197e-01
-2.72721082e-01 -9.10456777e-01 -8.49678889e-02 -1.15188134e+00
-6.95716202e-01 6.27301991e-01 -8.32497105e-02 -1.55627143e+00
1.14276238e-01 -6.05054080e-01 -3.30144912e-01 8.32454443e-01
-1.81978750e+00 -6.46364033e-01 -6.58943832e-01 8.79844368e-01
1.45212471e-01 -4.71221879e-02 7.16894031e-01 5.48408747e-01
-2.95875549e-01 4.59401309e-01 4.94551867e-01 -2.72416651e-01
9.51856673e-01 -1.24430358e+00 -1.51726246e-01 9.19162393e-01
-5.87565243e-01 2.94969618e-01 8.84535372e-01 -6.13788545e-01
-1.30980718e+00 -6.53685153e-01 1.82127669e-01 1.00351679e+00
5.82851879e-02 -1.21118948e-01 -8.59606504e-01 2.93281823e-01
6.53288960e-01 -5.13425708e-01 4.18197513e-01 -7.44146883e-01
-2.91894764e-01 -5.08985221e-01 -1.63049233e+00 5.22350848e-01
1.74916282e-01 -5.29452503e-01 -1.27188191e-01 -8.95051286e-02
5.55515289e-01 -2.73276210e-01 -1.19266236e+00 4.22093093e-01
7.69343436e-01 -1.33333218e+00 7.89951861e-01 5.99235296e-01
4.80042458e-01 -7.29836881e-01 -4.32183534e-01 -1.02701974e+00
-4.34784830e-01 -3.66091996e-01 4.83214796e-01 1.21736681e+00
6.55535683e-02 -5.85216999e-01 4.69991058e-01 1.69171408e-01
2.41209753e-02 -5.95905585e-03 -7.56038308e-01 -6.88773811e-01
-4.92873609e-01 3.72224092e-01 5.43431699e-01 5.31359971e-01
1.68396369e-01 -7.94408396e-02 -9.96086419e-01 3.58673395e-03
9.01184738e-01 -1.32491559e-01 4.05554324e-01 -8.04610014e-01
-2.04360932e-01 -1.62638173e-01 -4.70323712e-01 -9.11012769e-01
-5.65237582e-01 -1.10016726e-01 1.87803239e-01 -1.56242287e+00
3.38425756e-01 -3.90631765e-01 -5.27971387e-01 -1.48222046e-02
-4.97470573e-02 8.14413667e-01 4.70030546e-01 4.63690430e-01
-3.18716079e-01 4.93992716e-01 1.29718447e+00 -2.54862070e-01
-1.30232483e-01 -4.82804477e-01 -1.84660226e-01 3.58030945e-01
8.50679934e-01 8.34027380e-02 -6.29534185e-01 -1.71163529e-01
2.30106294e-01 3.87819797e-01 3.39395106e-02 -1.42080975e+00
-1.16978630e-01 -2.50077635e-01 5.66093147e-01 -6.39127016e-01
4.89232302e-01 -1.24309385e+00 4.96352732e-01 7.96824038e-01
-2.35477984e-01 9.33133885e-02 -1.45491749e-01 2.28562802e-01
-4.27206427e-01 -2.01161832e-01 1.29787815e+00 1.46267144e-02
-1.01320231e+00 -2.13703752e-01 -5.37201166e-01 -4.98269409e-01
1.15422583e+00 -5.61040938e-01 -8.85319531e-01 -5.66783726e-01
-2.43801981e-01 -3.67266357e-01 1.22758281e+00 -2.62914207e-02
9.14227426e-01 -1.51920843e+00 -3.29748601e-01 5.22217751e-01
6.40051514e-02 -8.61592472e-01 4.43427861e-01 4.64677662e-01
-1.51493168e+00 3.29528600e-01 -9.41605031e-01 -2.63389796e-01
-1.32031226e+00 5.89726567e-01 1.73778564e-01 -2.67080832e-02
-5.30794084e-01 1.16428673e-01 1.34436652e-01 5.95088661e-01
-4.24168585e-03 -2.37714261e-01 -2.91828871e-01 -5.48953414e-01
5.76439321e-01 5.20405948e-01 -1.22468360e-01 -4.82517838e-01
-1.32856458e-01 7.85724521e-01 1.12136886e-01 -3.45411479e-01
8.55061054e-01 -4.75234032e-01 -2.96086907e-01 2.85875410e-01
1.21679521e+00 3.00251871e-01 -1.30075800e+00 4.10600513e-01
-5.53680837e-01 -9.77151334e-01 -6.17896877e-02 -9.28676128e-01
-1.14226305e+00 6.88066006e-01 1.24695313e+00 6.77158892e-01
1.72446060e+00 -7.26282477e-01 9.46310461e-01 -4.16853987e-02
3.39136094e-01 -1.33058441e+00 -1.15817906e-02 -1.17692895e-01
5.34901321e-01 -8.29823375e-01 1.88980237e-01 -4.58974659e-01
-6.90668344e-01 1.54711103e+00 6.02728069e-01 1.82357598e-02
1.95195943e-01 3.98680359e-01 3.06579709e-01 3.40329498e-01
-3.71246338e-01 -9.07989442e-02 -1.52410999e-01 6.36905074e-01
5.38761437e-01 5.59438094e-02 -8.09335947e-01 -3.84730756e-01
-5.84849790e-02 1.77226692e-01 9.81133997e-01 7.98146069e-01
-9.26599324e-01 -7.87878215e-01 -5.86457372e-01 1.79802582e-01
-3.31438482e-01 -3.43969129e-02 6.76337630e-02 8.76283109e-01
1.23960160e-01 1.03616822e+00 1.71346679e-01 -6.16558135e-01
1.63987339e-01 -3.19242567e-01 5.62877595e-01 3.10364157e-01
3.38070899e-01 4.85722661e-01 -1.36723265e-01 -3.32881778e-01
-4.24711585e-01 1.58757627e-01 -1.17967558e+00 -7.12281764e-01
5.05073648e-03 2.08928987e-01 6.17778182e-01 4.91596282e-01
-4.43116188e-01 5.66762269e-01 8.99271011e-01 -1.53469294e-01
-1.34765590e-02 -7.40553200e-01 -1.36323845e+00 4.41263497e-01
2.61488587e-01 -2.86978394e-01 -4.68914181e-01 2.39399895e-01] | [10.92265796661377, -2.421475410461426] |
ff509f8c-1a71-45c7-afbc-61767f32181e | wr-one2set-towards-well-calibrated-keyphrase | 2211.06862 | null | https://arxiv.org/abs/2211.06862v2 | https://arxiv.org/pdf/2211.06862v2.pdf | WR-ONE2SET: Towards Well-Calibrated Keyphrase Generation | Keyphrase generation aims to automatically generate short phrases summarizing an input document. The recently emerged ONE2SET paradigm (Ye et al., 2021) generates keyphrases as a set and has achieved competitive performance. Nevertheless, we observe serious calibration errors outputted by ONE2SET, especially in the over-estimation of $\varnothing$ token (means "no corresponding keyphrase"). In this paper, we deeply analyze this limitation and identify two main reasons behind: 1) the parallel generation has to introduce excessive $\varnothing$ as padding tokens into training instances; and 2) the training mechanism assigning target to each slot is unstable and further aggravates the $\varnothing$ token over-estimation. To make the model well-calibrated, we propose WR-ONE2SET which extends ONE2SET with an adaptive instance-level cost Weighting strategy and a target Re-assignment mechanism. The former dynamically penalizes the over-estimated slots for different instances thus smoothing the uneven training distribution. The latter refines the original inappropriate assignment and reduces the supervisory signals of over-estimated slots. Experimental results on commonly-used datasets demonstrate the effectiveness and generality of our proposed paradigm. | ['Jinsong Su', 'Min Zhang', 'Xiaoli Wang', 'Jun Xie', 'Huan Lin', 'Baosong Yang', 'Xiangpeng Wei', 'Binbin Xie'] | 2022-11-13 | null | null | null | null | ['keyphrase-generation'] | ['natural-language-processing'] | [ 3.79698336e-01 3.08716059e-01 -4.50946122e-01 1.78247038e-02
-1.13736796e+00 -6.13777578e-01 6.47263288e-01 3.45623136e-01
-4.41436529e-01 1.03731966e+00 1.76200435e-01 -1.13104329e-01
-2.70237803e-01 -7.78103709e-01 -7.30375290e-01 -7.37622917e-01
1.22866817e-01 5.46600103e-01 3.05510014e-01 -1.66785896e-01
6.47787631e-01 -1.13924639e-02 -1.72200322e+00 3.45507264e-01
1.18057561e+00 8.64590526e-01 4.69879895e-01 5.15293837e-01
-6.32518291e-01 5.80506802e-01 -1.22275829e+00 -5.12968242e-01
3.26243281e-01 -4.38594937e-01 -5.44550836e-01 -9.40143596e-04
3.47770333e-01 -1.00478306e-01 -3.15691368e-03 1.20583773e+00
3.57625186e-01 1.12500884e-01 5.88694632e-01 -1.36966836e+00
-4.00813729e-01 1.30266845e+00 -7.57242680e-01 3.19678992e-01
9.27820727e-02 -1.69502228e-01 1.17431474e+00 -1.04129934e+00
4.53900188e-01 1.00893140e+00 4.22001451e-01 5.05736470e-01
-1.10891247e+00 -8.82805586e-01 3.45261157e-01 -6.53955201e-03
-1.67127216e+00 -2.69689709e-01 9.22120631e-01 -1.11893244e-01
5.26656032e-01 4.97585446e-01 3.82701784e-01 1.03333461e+00
8.09371620e-02 1.01144612e+00 9.43508685e-01 -6.22701764e-01
1.12883762e-01 4.90599871e-01 1.88385382e-01 3.76795053e-01
3.55106294e-01 -2.99670659e-02 -6.54664278e-01 -3.97189438e-01
6.19880080e-01 -1.06689684e-01 -1.97166994e-01 1.84648275e-01
-1.31795359e+00 7.21557498e-01 -2.19541658e-02 3.29969734e-01
-4.25677806e-01 1.56726971e-01 3.77266467e-01 1.73814908e-01
5.27351916e-01 6.88544571e-01 -5.53743303e-01 -3.34064454e-01
-1.24658811e+00 6.01238430e-01 5.24373770e-01 1.28919613e+00
8.05718660e-01 1.34125307e-01 -6.23406589e-01 1.00003934e+00
-2.77968585e-01 4.84260589e-01 8.06664526e-01 -5.91190040e-01
8.89308751e-01 5.79820216e-01 3.75146061e-01 -9.74295437e-01
-8.66065547e-02 -5.62128246e-01 -9.43981111e-01 -4.73586202e-01
3.40346992e-01 -4.02662516e-01 -9.33860004e-01 1.73766184e+00
-4.10610884e-02 1.42508477e-01 -7.80511200e-02 5.12560606e-01
6.16576791e-01 9.03564572e-01 1.66172788e-01 -5.17547488e-01
1.38493240e+00 -9.57622647e-01 -1.03612947e+00 -1.78869992e-01
3.17372620e-01 -9.16056395e-01 1.24567854e+00 4.50292975e-01
-1.10552013e+00 -6.48021460e-01 -9.41173494e-01 3.09010267e-01
-4.29311603e-01 4.86707479e-01 6.37176871e-01 5.61145604e-01
-5.25792301e-01 5.37118196e-01 -2.26635218e-01 1.80352688e-01
1.19099841e-01 2.27923512e-01 1.08625881e-01 3.82122368e-01
-1.55589998e+00 4.32849020e-01 7.99075961e-01 -1.58527374e-01
-3.82669687e-01 -1.09175467e+00 -6.14752173e-01 2.20824316e-01
8.23433459e-01 -3.95683289e-01 1.24977088e+00 -7.12784767e-01
-1.34742892e+00 4.02950436e-01 -1.97769538e-01 -4.03055072e-01
5.27835011e-01 -2.21338436e-01 -5.68286061e-01 1.35872550e-02
3.49651009e-01 6.45162761e-01 1.03471899e+00 -1.41015255e+00
-1.09314573e+00 -4.06689569e-02 -1.68471619e-01 4.18556511e-01
-6.06758237e-01 -2.32875586e-01 -4.58437592e-01 -1.26776266e+00
1.92205101e-01 -6.92091227e-01 -2.63367534e-01 -7.92811573e-01
-7.71260262e-01 -4.10493523e-01 6.04209840e-01 -2.27097169e-01
2.04050159e+00 -2.00411534e+00 -2.26889491e-01 3.60641211e-01
2.52814949e-01 1.82877228e-01 -6.63306797e-03 6.04835451e-01
-5.85828945e-02 3.16029787e-01 -1.53977007e-01 -1.67843193e-01
-1.34232556e-02 1.04572721e-01 -9.38892841e-01 -2.29728460e-01
2.52583921e-02 5.81364751e-01 -1.06503415e+00 -5.91954470e-01
-2.24135257e-02 -7.28777349e-02 -3.53329271e-01 2.68965334e-01
-5.61814189e-01 -1.53569460e-01 -4.71547335e-01 5.27223170e-01
6.84959531e-01 -1.50989443e-01 -6.89129159e-02 -2.72321761e-01
-2.33065620e-01 4.22844231e-01 -1.69229174e+00 1.40336788e+00
-2.25630254e-01 1.86934054e-01 -4.25457418e-01 -8.85503709e-01
1.13970780e+00 2.52051592e-01 6.12756371e-01 -2.65659541e-01
1.74442958e-03 2.32156768e-01 -3.20635736e-01 -2.26597041e-01
1.29871261e+00 3.06173991e-02 -4.49727654e-01 4.78168070e-01
-5.45680448e-02 -3.21830213e-01 5.26431024e-01 4.90154535e-01
7.10548997e-01 -1.75876379e-01 1.55132145e-01 -5.06548345e-01
6.00486696e-01 -2.64638215e-02 7.45874405e-01 1.07664597e+00
5.45923747e-02 5.88541567e-01 5.79247475e-01 -2.57093370e-01
-8.88465583e-01 -9.37883794e-01 5.71818762e-02 1.10566044e+00
2.41289362e-01 -7.86832035e-01 -6.98734760e-01 -8.46583188e-01
7.93347210e-02 9.38624084e-01 -5.95880508e-01 -1.47396341e-01
-5.47280371e-01 -9.87292647e-01 8.19373429e-01 4.06687617e-01
2.91002691e-01 -1.14398801e+00 -4.28321958e-01 4.94999766e-01
-5.85625112e-01 -9.73680675e-01 -7.79777467e-01 3.93313169e-01
-5.01004398e-01 -7.33402193e-01 -8.77976954e-01 -6.23529136e-01
8.44420016e-01 2.43416816e-01 1.19596410e+00 -7.73203820e-02
-9.15437341e-02 -5.58190234e-02 -4.69507724e-01 -8.28994691e-01
-4.42444503e-01 4.00538296e-01 1.18754499e-01 -3.13944258e-02
4.68307078e-01 -1.82935566e-01 -5.11470258e-01 9.40135568e-02
-9.74824071e-01 1.46228060e-01 6.03494644e-01 9.92217898e-01
6.05862677e-01 5.43830156e-01 9.47269917e-01 -1.32424629e+00
1.12549186e+00 -4.13446099e-01 -5.69847345e-01 3.07453662e-01
-1.13770461e+00 3.04340512e-01 8.70146513e-01 -6.23340309e-01
-1.03012574e+00 -4.98788953e-01 1.38992608e-01 -3.65412831e-01
1.71419770e-01 5.10377944e-01 1.80345569e-02 5.09303153e-01
5.05877554e-01 6.44860685e-01 -3.78687710e-01 -3.00398082e-01
3.58203739e-01 5.60863554e-01 5.76804161e-01 -9.54155266e-01
1.13056540e+00 4.51479554e-02 -3.40077102e-01 -6.47555053e-01
-8.21352303e-01 -4.38983202e-01 -6.58893585e-02 -3.73784974e-02
2.38150150e-01 -8.28103244e-01 -4.58853692e-01 3.65840644e-01
-1.05161273e+00 -1.81153300e-04 -5.66600442e-01 1.44220144e-01
-2.95459390e-01 5.24495959e-01 -3.34005028e-01 -9.18919325e-01
-8.01214457e-01 -9.97221947e-01 9.45689559e-01 5.67916751e-01
-4.47284549e-01 -5.79672873e-01 -6.78728521e-02 -1.78743880e-02
2.71190971e-01 -3.79174389e-02 9.19842541e-01 -7.15246141e-01
-3.64726037e-01 -1.57403439e-01 -1.53589085e-01 1.73893154e-01
3.22992891e-01 2.67302990e-01 -7.47580707e-01 -9.23746675e-02
-3.29583675e-01 -2.18123347e-01 7.37169087e-01 2.89106637e-01
1.54526091e+00 -5.72000921e-01 -3.09379578e-01 1.68841377e-01
1.28135622e+00 2.90291518e-01 4.96779799e-01 4.47868705e-01
4.32842523e-01 5.68858922e-01 9.05920744e-01 7.24468827e-01
3.12891036e-01 4.72875983e-01 5.67648076e-02 -4.65775430e-02
1.51338369e-01 -7.03830957e-01 2.46358097e-01 7.07459688e-01
1.24401152e-01 -3.66113126e-01 -4.87933785e-01 7.47615516e-01
-1.77335823e+00 -1.09215367e+00 7.42390938e-03 2.27281690e+00
1.40449345e+00 5.80854595e-01 8.97638220e-03 5.45275867e-01
8.15205336e-01 3.24392349e-01 -2.46947333e-01 -3.04840326e-01
-1.99572086e-01 4.61151600e-01 5.58563948e-01 3.70851934e-01
-8.72263432e-01 1.16246581e+00 5.63667774e+00 1.44857848e+00
-8.72955918e-01 -4.14319724e-01 5.99159539e-01 1.86951369e-01
-6.46731198e-01 -1.11640040e-02 -1.26393247e+00 8.74332249e-01
6.84793353e-01 -6.23940051e-01 1.82995632e-01 8.08159709e-01
8.49617794e-02 -1.44048691e-01 -8.59114230e-01 9.05124366e-01
-4.54494245e-02 -1.50383186e+00 5.02338767e-01 -2.53564507e-01
7.99163282e-01 -5.78509271e-01 2.01068491e-01 4.29635584e-01
3.09528738e-01 -5.10509253e-01 9.40222502e-01 1.82350054e-01
8.15666020e-01 -9.86676574e-01 5.31381369e-01 4.69220251e-01
-1.30781472e+00 1.22393593e-02 -3.70010227e-01 1.85866028e-01
-4.66201380e-02 8.47988307e-01 -1.11227584e+00 6.98981881e-01
5.07913888e-01 1.90267920e-01 -4.82628226e-01 6.38008058e-01
-3.38636428e-01 5.55799067e-01 -2.21412361e-01 -2.25511387e-01
3.23139161e-01 -2.41304994e-01 6.73754871e-01 1.50108469e+00
3.25278878e-01 4.50744620e-03 2.56988823e-01 9.32210565e-01
-4.24290672e-02 1.04876697e-01 -2.70510763e-01 -8.73830393e-02
1.00721407e+00 1.29405570e+00 -7.39259362e-01 -7.14627206e-01
8.25569034e-02 6.84617281e-01 -7.33412281e-02 4.54295963e-01
-8.25056195e-01 -6.88901305e-01 4.79002208e-01 5.79930432e-02
2.61601985e-01 1.63545161e-01 -5.81328869e-01 -1.13812256e+00
4.42275824e-03 -8.65400434e-01 5.32202244e-01 -4.41500753e-01
-1.14160037e+00 6.33948803e-01 3.00536245e-01 -1.40435600e+00
-3.62394363e-01 -9.40662101e-02 -5.72034895e-01 8.64606440e-01
-1.48594928e+00 -7.33060360e-01 -1.52246550e-01 4.21281368e-01
1.00047171e+00 -1.43884078e-01 8.04808795e-01 1.94276080e-01
-6.89397991e-01 1.14747417e+00 -1.33727238e-01 -7.27596059e-02
9.84100819e-01 -1.51799655e+00 3.26346248e-01 9.29156780e-01
2.69729886e-02 7.21202374e-01 9.93041873e-01 -6.11462772e-01
-1.03770185e+00 -1.00710404e+00 1.14452112e+00 -3.74936104e-01
6.27035618e-01 -3.72031927e-01 -8.23947489e-01 3.92412812e-01
9.74411890e-02 -4.22714472e-01 6.26863182e-01 -9.31098014e-02
-2.77345151e-01 -2.95795918e-01 -1.00441897e+00 8.37060928e-01
5.87370992e-01 -2.00238198e-01 -6.52257919e-01 2.26224363e-01
8.15257192e-01 -5.64329207e-01 -6.62474990e-01 5.65421879e-01
3.43171746e-01 -8.54018092e-01 8.69130075e-01 -3.35534245e-01
4.70052421e-01 -2.73589849e-01 3.93920355e-02 -1.33148646e+00
-1.44659787e-01 -1.03383589e+00 -3.06091994e-01 1.60131705e+00
5.86358666e-01 -3.75026882e-01 9.26728666e-01 5.55917621e-01
-6.17741086e-02 -7.96992302e-01 -7.13120937e-01 -6.70446634e-01
-8.51273313e-02 -3.43885630e-01 8.55689585e-01 9.62460577e-01
1.42273307e-01 2.77220696e-01 -5.13724744e-01 5.52459769e-02
6.34594023e-01 1.57411188e-01 8.95852327e-01 -9.22201514e-01
-1.78676188e-01 -5.81398189e-01 3.48641247e-01 -1.26634717e+00
-1.11830853e-01 -5.44477999e-01 1.63500637e-01 -1.02600431e+00
7.91607574e-02 -8.74437153e-01 -5.53093612e-01 4.18854177e-01
-8.57911527e-01 -1.39738873e-01 1.76547587e-01 2.40230262e-01
-3.22314024e-01 4.98663217e-01 1.16364408e+00 -2.10623175e-01
-4.30461735e-01 3.13716412e-01 -1.08582366e+00 5.05373776e-01
7.97750413e-01 -5.33084989e-01 -6.66874409e-01 -1.54821977e-01
1.77733988e-01 -7.77963642e-03 1.25808716e-02 -6.74738526e-01
1.67899892e-01 -3.89161676e-01 3.33371848e-01 -1.08712673e+00
3.39989141e-02 -7.92905748e-01 -3.18278819e-02 3.36258531e-01
-5.46968162e-01 2.51068413e-01 3.22152138e-01 5.05195796e-01
-2.32632846e-01 -3.41966152e-01 5.42091548e-01 -1.79655239e-01
-5.98080218e-01 1.83816567e-01 -2.84622341e-01 2.37506270e-01
8.26510131e-01 -2.72772640e-01 -3.68796706e-01 -1.88499540e-01
-9.32812318e-02 2.75502831e-01 6.26952499e-02 6.13310397e-01
5.45555234e-01 -1.24870360e+00 -6.61026061e-01 3.77803087e-01
3.30631405e-01 2.23447844e-01 2.21776426e-01 2.39977479e-01
-4.35592420e-02 3.56193900e-01 3.43156964e-01 -3.35492015e-01
-8.90861571e-01 3.81354183e-01 -1.20129898e-01 -7.20220208e-01
-4.79516268e-01 1.00801671e+00 3.21261473e-02 -8.25322419e-02
3.80752444e-01 -5.12532353e-01 -2.46420190e-01 3.29969019e-01
6.02471173e-01 3.76086354e-01 1.21807225e-01 -1.15659676e-01
-1.06396496e-01 2.34064057e-01 -6.35108531e-01 -1.79185495e-01
8.87998760e-01 -4.86232117e-02 -1.15871923e-02 4.22431618e-01
6.64511502e-01 2.20567495e-01 -1.03616261e+00 -4.15626585e-01
1.34238631e-01 -4.49755311e-01 -1.60467684e-01 -7.46985197e-01
-6.86268926e-01 3.15863758e-01 3.52221914e-02 4.75025773e-01
1.09226620e+00 -3.40541244e-01 1.09833813e+00 1.94795713e-01
3.10647190e-01 -1.62072539e+00 1.16249785e-01 4.75272954e-01
6.42155230e-01 -8.81902993e-01 2.30750889e-02 -4.81595337e-01
-8.89878213e-01 1.07831955e+00 8.10624599e-01 6.23597996e-03
3.76322448e-01 3.37639630e-01 2.67253183e-02 2.80614514e-02
-8.68218780e-01 2.48769462e-01 2.48139501e-01 2.61084467e-01
2.17122495e-01 1.50344998e-01 -7.58973897e-01 1.01473618e+00
-5.50738513e-01 -1.66093424e-01 7.24395752e-01 8.13473880e-01
-5.06856203e-01 -1.27949846e+00 -4.96467233e-01 6.35264754e-01
-4.77939337e-01 -3.33156258e-01 -2.33502284e-01 6.31028175e-01
2.73166984e-01 8.65805805e-01 -2.03724671e-02 -4.46580380e-01
3.86390030e-01 1.73928037e-01 7.59727657e-02 -6.79680288e-01
-7.30200648e-01 3.21214974e-01 -7.39148110e-02 -9.43398178e-02
-2.42026985e-01 -3.86563361e-01 -1.25557494e+00 -1.86422348e-01
-4.64383423e-01 7.56475329e-01 3.01468343e-01 6.81016326e-01
3.19748878e-01 6.82949662e-01 9.36419964e-01 -4.63426501e-01
-1.03195417e+00 -1.12320673e+00 -6.90388799e-01 3.06362867e-01
1.32549956e-01 -6.25334978e-01 -4.57782686e-01 2.20019415e-01] | [12.1439208984375, 8.883356094360352] |
1be5c804-fdb7-4559-83c4-a4a0c7f1b463 | neural-activation-constellations-unsupervised | 1504.08289 | null | http://arxiv.org/abs/1504.08289v3 | http://arxiv.org/pdf/1504.08289v3.pdf | Neural Activation Constellations: Unsupervised Part Model Discovery with Convolutional Networks | Part models of object categories are essential for challenging recognition
tasks, where differences in categories are subtle and only reflected in
appearances of small parts of the object. We present an approach that is able
to learn part models in a completely unsupervised manner, without part
annotations and even without given bounding boxes during learning. The key idea
is to find constellations of neural activation patterns computed using
convolutional neural networks. In our experiments, we outperform existing
approaches for fine-grained recognition on the CUB200-2011, NA birds, Oxford
PETS, and Oxford Flowers dataset in case no part or bounding box annotations
are available and achieve state-of-the-art performance for the Stanford Dog
dataset. We also show the benefits of neural constellation models as a data
augmentation technique for fine-tuning. Furthermore, our paper unites the areas
of generic and fine-grained classification, since our approach is suitable for
both scenarios. The source code of our method is available online at
http://www.inf-cv.uni-jena.de/part_discovery | ['Marcel Simon', 'Erik Rodner'] | 2015-04-30 | neural-activation-constellations-unsupervised-1 | http://openaccess.thecvf.com/content_iccv_2015/html/Simon_Neural_Activation_Constellations_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Simon_Neural_Activation_Constellations_ICCV_2015_paper.pdf | iccv-2015-12 | ['model-discovery'] | ['miscellaneous'] | [ 2.05572292e-01 6.84753284e-02 -1.46623537e-01 -6.45958185e-01
-3.04181010e-01 -9.69920099e-01 7.18993485e-01 8.88269171e-02
-1.63565889e-01 4.74052370e-01 -8.05652812e-02 2.76364125e-02
-3.51953268e-01 -8.11189055e-01 -1.05720258e+00 -4.57009822e-01
-4.84116264e-02 7.05182731e-01 2.81891733e-01 -2.21356243e-01
4.20153402e-02 8.88673604e-01 -1.97675049e+00 7.55384922e-01
4.43101972e-01 1.39696670e+00 2.90616080e-02 5.61108887e-01
-1.59500226e-01 6.19612813e-01 -4.33973491e-01 -5.39768040e-01
6.43546999e-01 -3.44557501e-02 -1.00230539e+00 3.36823732e-01
1.08047771e+00 -1.91798694e-02 -2.60205328e-01 9.70423996e-01
-2.22534556e-02 6.27544820e-02 8.47154558e-01 -1.25025153e+00
-6.37269497e-01 8.60569775e-01 -3.31820935e-01 2.76117306e-02
-2.53959715e-01 4.83064502e-02 1.12862480e+00 -7.47162938e-01
4.52719778e-01 1.36759412e+00 8.07731807e-01 6.66660011e-01
-1.59699678e+00 -4.65503842e-01 3.68796378e-01 3.61396044e-01
-1.45948887e+00 -4.74897504e-01 4.92720991e-01 -5.99120677e-01
9.81586277e-01 2.53977358e-01 4.89132434e-01 9.45682585e-01
-2.75819153e-01 9.05493498e-01 1.16414285e+00 -4.04899925e-01
1.46871597e-01 1.46148473e-01 4.61248219e-01 7.21344113e-01
3.58678192e-01 2.07968667e-01 -1.19373910e-01 -4.04135510e-02
8.67756546e-01 2.46296629e-01 -6.80051968e-02 -8.21192920e-01
-1.30408967e+00 7.03465998e-01 8.63020539e-01 4.51985508e-01
-2.96254545e-01 3.11851323e-01 4.24476564e-01 3.16967160e-01
3.27170134e-01 5.31114995e-01 -8.36713135e-01 6.61941618e-02
-9.82344031e-01 2.42702544e-01 9.39281106e-01 1.06111228e+00
8.73941183e-01 -7.69346803e-02 -3.27551335e-01 1.08940244e+00
2.03968454e-02 2.18508057e-02 3.38558137e-01 -9.03613329e-01
8.37309659e-02 7.39751697e-01 9.10426751e-02 -3.45266432e-01
-3.76952112e-01 -4.35568810e-01 -8.54676247e-01 3.79675150e-01
6.81455076e-01 2.70487994e-01 -1.39778233e+00 1.59493053e+00
3.63170356e-01 -6.99181259e-02 -2.77690113e-01 6.43741488e-01
8.78754795e-01 1.05210342e-01 5.28747290e-02 3.52140218e-01
1.60083520e+00 -1.27297533e+00 -2.67821461e-01 -1.73808813e-01
3.87377352e-01 -4.71460700e-01 9.00215566e-01 4.94012237e-01
-8.22116494e-01 -9.71084535e-01 -9.97852862e-01 -6.97648060e-03
-8.37120116e-01 4.85916346e-01 9.52973902e-01 6.55766726e-01
-9.57012355e-01 9.81240273e-01 -7.69680142e-01 -4.79910851e-01
9.29033875e-01 4.83808994e-01 -5.55854261e-01 -3.92247103e-02
-5.92253149e-01 7.47481227e-01 6.86131895e-01 1.55511215e-01
-9.74342048e-01 -6.78233564e-01 -8.01874816e-01 2.28203684e-01
2.60141701e-01 -5.46587050e-01 1.43457830e+00 -1.27996778e+00
-1.09661996e+00 1.15785563e+00 3.14670265e-01 -6.75060451e-01
5.32656491e-01 1.19203396e-01 -1.81484386e-01 -7.87017569e-02
-1.26800165e-01 1.14664519e+00 8.97366881e-01 -1.13444173e+00
-6.60838544e-01 -4.01719779e-01 3.90328914e-01 -3.35173130e-01
4.61725406e-02 -1.15685858e-01 -2.49164641e-01 -7.51111388e-01
-4.89430688e-02 -8.96266520e-01 -2.46189281e-01 3.68873537e-01
-2.34620765e-01 -3.87283355e-01 6.75525069e-01 -2.56819576e-01
6.48921549e-01 -2.07682610e+00 2.84559187e-02 9.39355791e-03
1.50907710e-01 2.76772857e-01 -4.38996434e-01 2.95948058e-01
-2.37338647e-01 6.93718344e-02 -3.66488248e-01 -1.05976962e-01
3.78620118e-01 4.56382424e-01 -1.30807623e-01 4.42262113e-01
4.57088053e-01 9.52164888e-01 -3.18720490e-01 -1.38946891e-01
3.13122541e-01 2.95315981e-01 -4.22328532e-01 1.09371975e-01
-4.67346966e-01 8.46623555e-02 -1.67529106e-01 1.02867103e+00
8.22580278e-01 -4.76439595e-01 2.78352574e-03 -4.35074151e-01
-4.47629802e-02 1.43742740e-01 -1.34684443e+00 1.51392031e+00
-2.84238368e-01 3.35747898e-01 2.36048222e-01 -1.22825658e+00
8.19827497e-01 -1.10655181e-01 1.04788639e-01 -5.91048658e-01
1.42824501e-01 2.71656662e-01 2.27088392e-01 1.05168179e-01
4.06652242e-01 -7.05249310e-02 -8.81481767e-02 1.08412325e-01
5.08341193e-01 -1.36947393e-01 7.18578219e-01 -2.31354833e-01
1.01618898e+00 2.94103384e-01 6.48193777e-01 -4.02717561e-01
2.91557729e-01 2.64176816e-01 3.18156272e-01 9.44012880e-01
-8.95799920e-02 6.46527946e-01 4.41744536e-01 -6.88007414e-01
-1.13830185e+00 -8.46367180e-01 -6.21407747e-01 1.31771576e+00
-6.32133633e-02 -2.59044021e-01 -6.83556736e-01 -1.00405097e+00
4.04708624e-01 4.59370077e-01 -1.35713148e+00 7.10140839e-02
-4.44081873e-01 -4.69204456e-01 5.46419144e-01 9.33800340e-01
3.19378287e-01 -1.38367033e+00 -4.15375352e-01 5.47496714e-02
2.17839599e-01 -1.01056933e+00 -1.01790302e-01 7.75117755e-01
-9.26145911e-01 -1.33100188e+00 -7.18988895e-01 -7.46843457e-01
6.88734233e-01 1.60311952e-01 1.35671592e+00 1.37349010e-01
-6.68215454e-01 2.38829359e-01 -4.69705075e-01 -4.18615490e-01
-2.98988402e-01 4.71106544e-02 -1.87285155e-01 1.35127949e-02
3.96886617e-01 -4.41077620e-01 -4.09945965e-01 6.91001117e-01
-8.81185949e-01 -6.65201172e-02 9.39367235e-01 1.05909777e+00
8.28742087e-01 -2.36289799e-01 1.78260371e-01 -1.15886593e+00
-4.66294475e-02 -3.17300081e-01 -9.50481534e-01 3.47689420e-01
-3.66539896e-01 1.38820246e-01 5.55938423e-01 -5.45197606e-01
-6.74180806e-01 3.50887954e-01 -2.57211536e-01 -6.87406540e-01
-8.62330019e-01 -4.49731238e-02 -2.33582616e-01 -3.36253822e-01
7.91711569e-01 1.02998942e-01 -3.95006180e-01 -9.41377521e-01
4.17096853e-01 5.09389222e-01 4.38400596e-01 -6.08485460e-01
6.49165988e-01 4.95496422e-01 1.13889305e-02 -6.37766898e-01
-9.22844946e-01 -7.49176025e-01 -1.18615413e+00 2.14962989e-01
5.84834635e-01 -8.86383355e-01 -6.49694026e-01 4.35182154e-01
-8.62190664e-01 -5.88721812e-01 -7.33865678e-01 5.11909612e-02
-7.51822591e-01 1.09827600e-01 -4.36685979e-01 -4.09393460e-01
-2.00042099e-01 -9.11146283e-01 1.26441050e+00 3.38463157e-01
5.69566227e-02 -6.54706120e-01 -1.46105513e-01 2.16333255e-01
3.01611811e-01 2.00567648e-01 7.87416816e-01 -9.18302894e-01
-6.38461053e-01 -1.90748870e-01 -5.44510484e-01 4.56953704e-01
6.34587416e-03 -3.13533768e-02 -1.16007376e+00 -1.89916015e-01
-6.39753520e-01 -5.71662128e-01 1.14152110e+00 2.98180938e-01
1.62038279e+00 -3.58623207e-01 -3.89199197e-01 7.37300754e-01
1.45610583e+00 -2.32940316e-01 4.50135231e-01 3.20259839e-01
6.76620960e-01 6.69352949e-01 6.59801781e-01 4.55945224e-01
2.94085257e-02 8.57434928e-01 7.06647754e-01 6.90397471e-02
-4.36487406e-01 -6.12892359e-02 -1.29918993e-01 5.42706996e-02
-1.24185622e-01 -1.11722954e-01 -7.01324046e-01 8.70596945e-01
-1.80062509e+00 -1.05701745e+00 -1.58646986e-01 1.91949892e+00
6.34951591e-01 4.81555201e-02 3.37637901e-01 1.83998030e-02
6.57745302e-01 -1.15652539e-01 -6.45731151e-01 -4.20122534e-01
-6.75881952e-02 5.38391590e-01 6.05423331e-01 3.00487340e-03
-1.47823799e+00 9.74131644e-01 6.32122135e+00 9.94105875e-01
-8.83170903e-01 3.10581587e-02 5.48026443e-01 7.11323768e-02
2.25530580e-01 -1.79113761e-01 -1.00525427e+00 1.83900282e-01
7.76303589e-01 5.09493351e-01 4.84869659e-01 1.25787473e+00
-4.49001521e-01 5.81552833e-02 -1.41075194e+00 8.10001731e-01
-1.40897617e-01 -1.59589744e+00 -1.30357593e-01 9.85142365e-02
8.45016241e-01 2.61241525e-01 -2.42852390e-01 5.22714317e-01
3.29678446e-01 -1.10545957e+00 1.08572936e+00 4.64882046e-01
8.35370302e-01 -4.67492163e-01 5.89742899e-01 3.68977994e-01
-1.27210522e+00 -1.01871029e-01 -7.05300927e-01 1.02472134e-01
-4.07403439e-01 3.67996067e-01 -8.60541701e-01 3.50032747e-01
1.04135931e+00 5.99933088e-01 -9.84053910e-01 1.25639427e+00
-1.85370147e-01 5.18164039e-01 -4.91040438e-01 -4.61531281e-02
2.54900843e-01 1.38086200e-01 9.34625715e-02 1.56777048e+00
-4.57258224e-02 -1.85365438e-01 2.93826729e-01 9.69296753e-01
-3.21461201e-01 -1.89560220e-01 -3.51519048e-01 -2.49627739e-01
1.39093474e-01 1.64326727e+00 -9.55145299e-01 -4.01887119e-01
-2.72196203e-01 8.21901143e-01 6.39055490e-01 -5.59425652e-02
-5.92871010e-01 -4.87882465e-01 7.66098797e-01 7.02892542e-02
1.16847956e+00 9.01025832e-02 -2.11030409e-01 -1.14265108e+00
-1.41382664e-01 -9.75063920e-01 5.98270237e-01 -6.32192254e-01
-1.61119306e+00 6.45450473e-01 9.56230462e-02 -1.12395048e+00
-2.36266807e-01 -1.23791051e+00 -4.33637947e-01 5.99993587e-01
-1.35934973e+00 -1.64723885e+00 -5.43849587e-01 5.17786384e-01
6.22752964e-01 5.89265227e-02 9.52359498e-01 1.61272824e-01
-1.77702740e-01 7.02367783e-01 1.98109269e-01 2.98224181e-01
4.75220978e-01 -1.50952339e+00 6.09899402e-01 6.78418934e-01
5.66592157e-01 4.16490734e-01 5.50111353e-01 -3.18619698e-01
-1.10175812e+00 -1.19753802e+00 4.88067627e-01 -5.13125598e-01
7.17219114e-01 -7.09640622e-01 -8.19559991e-01 7.62527943e-01
-6.80599874e-03 6.19913936e-01 6.01221502e-01 4.38696444e-01
-8.17041516e-01 -6.75240010e-02 -1.36844540e+00 1.00869276e-01
1.26111269e+00 -2.57079750e-01 -5.90873063e-01 3.85146171e-01
4.31882620e-01 -3.85153413e-01 -1.02129865e+00 6.77826047e-01
8.18613410e-01 -9.89931047e-01 1.01258731e+00 -8.99326384e-01
2.67162234e-01 -4.19579297e-01 -5.35622954e-01 -1.16086590e+00
-6.89419150e-01 -6.37700409e-02 -3.13138515e-01 1.06726742e+00
3.60775799e-01 -4.56291705e-01 8.18045378e-01 4.23029326e-02
-2.51546770e-01 -5.62421978e-01 -7.34048784e-01 -8.99122298e-01
3.14247459e-02 -3.40036273e-01 6.99738979e-01 5.98620474e-01
-4.98732209e-01 9.94420499e-02 -2.77593974e-02 8.12167302e-02
6.13552332e-01 7.64558196e-01 8.88004124e-01 -1.44308472e+00
-5.78441262e-01 -7.15329528e-01 -7.09663451e-01 -8.18924308e-01
1.10802092e-01 -8.74972224e-01 1.90187469e-01 -1.45933843e+00
2.77869433e-01 -5.79875827e-01 -4.03345883e-01 1.03701305e+00
2.66262770e-01 7.54016161e-01 3.34262818e-01 2.29375675e-01
-7.20288396e-01 3.25643480e-01 1.13864148e+00 -3.87621880e-01
3.29338580e-01 3.12456757e-01 -7.43148863e-01 6.68433547e-01
6.90068305e-01 -4.49936330e-01 1.28316581e-01 -1.41988546e-01
-3.62468690e-01 -5.18640995e-01 7.48201907e-01 -1.10217118e+00
-1.01378158e-01 -2.02868078e-02 8.25651646e-01 -7.90478587e-01
5.09029686e-01 -9.85357165e-01 2.08589181e-01 4.70620960e-01
-2.61931896e-01 -3.03286403e-01 5.49738824e-01 4.18571860e-01
-1.31394356e-01 -4.54980522e-01 1.05145776e+00 -4.72398639e-01
-9.69006777e-01 4.49509948e-01 -9.75538269e-02 -2.84824595e-02
8.95536184e-01 -3.74533921e-01 -3.52740109e-01 2.99137920e-01
-9.64311719e-01 -6.28085583e-02 5.96520841e-01 5.12972355e-01
3.77318189e-02 -1.28575003e+00 -7.16764987e-01 3.34389240e-01
6.50067091e-01 -1.74625561e-01 3.96593988e-01 6.45506382e-01
-3.96023303e-01 6.45474017e-01 -6.96064532e-01 -7.71646082e-01
-1.32829666e+00 7.57637322e-01 4.46456403e-01 -1.93074957e-01
-3.80735755e-01 9.60559964e-01 5.57243049e-01 -8.48477483e-01
4.19990003e-01 -6.43015802e-01 -1.57537609e-01 8.66237506e-02
5.65812349e-01 9.70376134e-02 4.16004092e-01 -5.82492292e-01
-4.45968240e-01 4.49680030e-01 -2.48243302e-01 4.73049074e-01
1.59098589e+00 2.53989071e-01 -4.01024185e-02 2.83799320e-01
1.06544578e+00 -2.12734848e-01 -1.35831249e+00 -2.46064126e-01
-3.82918715e-02 -3.64635766e-01 -1.21633440e-01 -1.25052893e+00
-8.51416767e-01 8.27904761e-01 7.88942993e-01 2.85842508e-01
1.08556032e+00 4.89292353e-01 8.48462209e-02 6.65497899e-01
3.04767877e-01 -8.20435941e-01 -3.16933334e-01 6.14799798e-01
1.08142185e+00 -1.26834035e+00 -1.14715897e-01 -3.67853940e-01
-4.20731723e-01 1.25060809e+00 7.74314046e-01 -2.55498320e-01
5.64154923e-01 3.06456089e-01 -1.36057571e-01 -2.37125218e-01
-8.08840871e-01 -7.24855244e-01 7.00006068e-01 8.05953503e-01
3.52808982e-01 3.02482277e-01 9.00393575e-02 5.94657779e-01
-1.94549203e-01 -1.73335552e-01 2.61624515e-01 7.89922953e-01
-4.56871331e-01 -1.20749593e+00 -3.70402515e-01 6.55399203e-01
-3.98145139e-01 -4.34744284e-02 -6.81194127e-01 9.81810510e-01
5.31055808e-01 4.06662226e-01 1.98953643e-01 -1.72698513e-01
5.06337285e-01 1.37518778e-01 9.43509102e-01 -8.10489476e-01
-8.70863676e-01 -5.27840555e-02 1.70994624e-01 -6.90714538e-01
-3.91080409e-01 -7.75287688e-01 -9.15555954e-01 -1.95347920e-01
-3.39604467e-01 1.26919039e-02 7.09188521e-01 7.85787642e-01
2.48531967e-01 5.30608714e-01 3.50589693e-01 -1.15631044e+00
-6.73184395e-01 -1.21172750e+00 -7.08994210e-01 2.83214241e-01
2.93554842e-01 -8.29326093e-01 -1.12796366e-01 1.20967217e-01] | [9.627416610717773, 1.94867742061615] |
a73bff52-58b8-4e63-8e60-b4a8428177dd | pcr4all-a-comprehensive-evaluation-benchmark | null | null | https://aclanthology.org/2022.lrec-1.641 | https://aclanthology.org/2022.lrec-1.641.pdf | PCR4ALL: A Comprehensive Evaluation Benchmark for Pronoun Coreference Resolution in English | Pronoun Coreference Resolution (PCR) is the task of resolving pronominal expressions to all mentions they refer to. The correct resolution of pronouns typically involves the complex inference over both linguistic knowledge and general world knowledge. Recently, with the help of pre-trained language representation models, the community has made significant progress on various PCR tasks. However, as most existing works focus on developing PCR models for specific datasets and measuring the accuracy or F1 alone, it is still unclear whether current PCR systems are reliable in real applications. Motivated by this, we propose PCR4ALL, a new benchmark and a toolbox that evaluates and analyzes the performance of PCR systems from different perspectives (i.e., knowledge source, domain, data size, frequency, relevance, and polarity). Experiments demonstrate notable performance differences when the models are examined from different angles. We hope that PCR4ALL can motivate the community to pay more attention to solving the overall PCR problem and understand the performance comprehensively. All data and codes are available at: https://github.com/HKUST-KnowComp/PCR4ALL. | ['Yangqiu Song', 'Hongming Zhang', 'Xinran Zhao'] | null | null | null | null | lrec-2022-6 | ['coreference-resolution'] | ['natural-language-processing'] | [-5.02692424e-02 1.10715926e-01 -6.85129225e-01 -2.74373680e-01
-1.19756508e+00 -8.16910207e-01 7.28268266e-01 3.67200106e-01
-4.51236159e-01 8.64553928e-01 6.09025657e-01 -1.20969810e-01
-9.16954800e-02 -5.66060305e-01 -4.26439732e-01 -4.32785630e-01
1.73385724e-01 7.04458177e-01 1.90196589e-01 -4.88158107e-01
3.38233471e-01 5.82493067e-01 -1.43261206e+00 2.32033432e-01
8.01499367e-01 3.53498876e-01 9.96672362e-02 1.16755612e-01
-2.85895944e-01 8.98697317e-01 -6.21649325e-01 -8.21686745e-01
-2.40764350e-01 7.24521140e-03 -1.24494040e+00 -6.11435771e-01
2.20639065e-01 2.21836641e-01 -2.35955805e-01 1.12486887e+00
5.27316093e-01 1.67212218e-01 3.77436608e-01 -9.80961204e-01
-6.47530973e-01 8.84962320e-01 -4.78691638e-01 5.08921266e-01
7.46617019e-01 -1.70424789e-01 1.37845576e+00 -8.23029220e-01
9.54944134e-01 1.44108605e+00 4.91394222e-01 6.30077243e-01
-1.09654403e+00 -7.29301870e-01 3.09270948e-01 7.09474027e-01
-1.51529741e+00 -6.27382517e-01 5.92845440e-01 -2.12323457e-01
1.06347525e+00 5.58300376e-01 -2.69934279e-03 1.37860346e+00
-1.41763404e-01 8.61873806e-01 1.10112357e+00 -3.94730806e-01
-1.19833417e-01 7.35131055e-02 5.82508326e-01 2.58924454e-01
4.50612366e-01 -2.48002961e-01 -8.55036497e-01 -2.03870997e-01
3.32105994e-01 -5.25323749e-01 -6.44096732e-01 -7.87221454e-03
-8.38350534e-01 7.37844288e-01 2.22691044e-01 6.33618891e-01
-2.94272184e-01 -3.45685899e-01 2.94919521e-01 3.16654630e-02
2.64292657e-01 7.22050369e-01 -4.85136658e-01 -2.83349544e-01
-6.04901075e-01 5.76231420e-01 1.11469495e+00 9.49776173e-01
5.69279790e-01 -5.45739055e-01 -2.19955683e-01 1.07316506e+00
-1.91504764e-03 4.94794309e-01 2.68544316e-01 -1.12863362e+00
6.12523913e-01 6.99600101e-01 3.50446910e-01 -1.12928867e+00
-4.99278814e-01 -1.98965564e-01 -4.26341087e-01 -4.91017461e-01
5.67129374e-01 -4.14530523e-02 -1.77407861e-01 1.91331911e+00
1.13293856e-01 -1.06647670e-01 2.99987823e-01 1.06471980e+00
1.23020446e+00 5.01137376e-01 2.77089238e-01 -4.69762415e-01
1.69232774e+00 -6.26088858e-01 -1.05585241e+00 -5.39225042e-01
4.48090196e-01 -9.11129415e-01 9.49400127e-01 2.20091879e-01
-1.09042168e+00 6.41429722e-02 -7.79930830e-01 -3.62026215e-01
-2.59488106e-01 -1.90491769e-02 5.91352344e-01 3.94205190e-02
-7.10386574e-01 4.01917428e-01 -6.94645286e-01 -6.16007447e-01
4.52559395e-03 2.31184334e-01 -4.61072981e-01 -2.42348567e-01
-1.65709877e+00 1.33654654e+00 2.43401945e-01 1.39979213e-01
-2.47625470e-01 -5.47580063e-01 -6.26051128e-01 -6.39625192e-02
6.10100627e-01 -3.94929945e-01 1.40179336e+00 -5.49631417e-01
-1.19596171e+00 1.01900911e+00 -6.80318773e-01 -2.89933056e-01
2.95746863e-01 -3.42308134e-01 -4.95886415e-01 -1.10058293e-01
3.43857795e-01 3.02430153e-01 1.03107607e-02 -1.13786030e+00
-6.96241558e-01 -3.58104944e-01 2.54963905e-01 3.02371055e-01
-1.79135390e-02 6.55778646e-01 -7.94212520e-01 -3.12050283e-01
-3.49929221e-02 -8.04395676e-01 1.09477535e-01 -5.62104642e-01
-4.85948801e-01 -8.50813627e-01 1.58460438e-01 -7.15695322e-01
1.37886310e+00 -2.12850046e+00 4.16570783e-01 -8.54658559e-02
2.91977096e-02 3.37234169e-01 -3.94727401e-02 5.76996148e-01
-2.18381673e-01 3.24777931e-01 -1.62155017e-01 -1.95678696e-01
-1.08617600e-02 2.69751221e-01 -4.49507862e-01 3.54430616e-01
3.51885676e-01 6.58554912e-01 -1.03402555e+00 -5.77981174e-01
-2.15426013e-01 5.71910083e-01 -3.84446800e-01 2.50973165e-01
-2.14927435e-01 2.03927875e-01 -5.07070065e-01 6.85489714e-01
5.29393375e-01 -2.63033450e-01 7.83988118e-01 -2.29840994e-01
-2.74009556e-01 6.83950424e-01 -1.11155057e+00 1.38393247e+00
-9.74860340e-02 6.61595464e-01 1.20400161e-01 -8.13120604e-01
8.69158566e-01 3.46088111e-01 2.11853623e-01 -8.10929418e-01
-3.32704037e-02 3.18861097e-01 -5.59139848e-02 -5.18343389e-01
7.04656005e-01 -6.74614981e-02 -2.22092479e-01 2.30872750e-01
-1.61460727e-01 3.52786303e-01 4.65982974e-01 1.96496308e-01
1.05011404e+00 -4.53657657e-02 4.85082984e-01 -2.67293304e-01
5.40650725e-01 2.85248488e-01 8.78836453e-01 4.34854746e-01
-2.06181511e-01 4.27202165e-01 6.60209596e-01 -2.51993477e-01
-3.08312684e-01 -9.22288179e-01 -3.54782522e-01 1.32452297e+00
3.45437258e-01 -6.36981785e-01 -6.63266897e-01 -4.53484625e-01
9.78280157e-02 1.00733709e+00 -4.85989213e-01 9.09619778e-02
-9.13233936e-01 -8.42657804e-01 7.15321064e-01 4.83483434e-01
1.57305285e-01 -1.23491704e+00 -3.08841258e-01 1.29213423e-01
-8.50093603e-01 -1.35005057e+00 -3.04737329e-01 9.61584784e-03
-4.50943410e-01 -1.32027459e+00 -1.03202462e-01 -6.29718661e-01
3.05550307e-01 3.42327088e-01 1.24806023e+00 8.52970481e-02
9.76507664e-02 2.66383529e-01 -3.43185365e-01 -4.32916492e-01
-1.74058661e-01 2.24758387e-01 5.31016104e-02 -3.31827223e-01
9.95044529e-01 -5.26901603e-01 -2.48184040e-01 1.13715738e-01
-5.17282069e-01 -1.57826915e-01 4.06924278e-01 4.48166162e-01
5.28494358e-01 -3.18087667e-01 3.68334234e-01 -1.14146626e+00
1.04371023e+00 -6.99342310e-01 -3.96279305e-01 3.77250224e-01
-6.51245236e-01 6.25660345e-02 2.09360212e-01 -2.21734792e-01
-1.17385757e+00 -3.98019344e-01 -8.09266791e-02 -1.74167436e-02
-2.43037567e-02 7.94194043e-01 -1.86834455e-01 3.52354407e-01
7.00683594e-01 8.84247385e-03 -2.24412337e-01 -7.05624044e-01
3.33362341e-01 6.31101310e-01 8.56193483e-01 -9.95931685e-01
4.28728998e-01 2.26110548e-01 -4.83266711e-01 -6.81021810e-01
-1.08105278e+00 -5.34227014e-01 -3.94375890e-01 1.45013314e-02
5.81807256e-01 -9.12543535e-01 -7.78519630e-01 3.34976315e-02
-1.52592099e+00 5.32811061e-02 1.83816150e-01 1.37688920e-01
-2.04115391e-01 4.61310327e-01 -6.66110933e-01 -6.82366729e-01
-4.85033423e-01 -1.07812679e+00 8.49308312e-01 3.96302640e-01
-7.96909451e-01 -8.88737500e-01 9.80638564e-02 6.21364117e-01
1.24320485e-01 1.16120651e-01 9.42892313e-01 -8.27752233e-01
-4.22899097e-01 1.03413910e-02 -3.41145664e-01 -1.04335479e-01
3.49496119e-02 4.98837680e-02 -8.37186158e-01 -1.67125076e-01
-2.93541908e-01 -1.60031989e-01 7.34402418e-01 6.18544072e-02
9.47893679e-01 -3.65814865e-01 -4.71023023e-01 3.64159822e-01
1.25072634e+00 -2.69473176e-02 5.98356664e-01 7.19617069e-01
4.11893994e-01 7.92351127e-01 8.31750691e-01 1.96153715e-01
8.99019897e-01 8.95170510e-01 2.02974781e-01 3.33122581e-01
-1.26685679e-01 -1.53959364e-01 3.07040185e-01 7.95328617e-01
-5.03393352e-01 -2.12414831e-01 -1.38957930e+00 6.85037732e-01
-1.95169389e+00 -1.14498305e+00 -4.07215692e-02 1.94497979e+00
1.40689838e+00 2.62658428e-02 -1.65986329e-01 -2.82276012e-02
8.34359348e-01 1.63768470e-01 -3.27966124e-01 -3.38415533e-01
-4.58644569e-01 1.47541180e-01 2.40201354e-01 7.89871514e-01
-9.21979368e-01 1.16590357e+00 6.04676437e+00 5.81044555e-01
-1.14954734e+00 -1.87139083e-02 1.63579926e-01 -2.81568225e-02
-4.37274665e-01 5.02039902e-02 -1.19591761e+00 1.81467324e-01
9.03699219e-01 -6.82893574e-01 5.64829648e-01 5.91206014e-01
-2.50074118e-02 -3.78854051e-02 -1.21578646e+00 8.70321691e-01
2.01393157e-01 -1.27083814e+00 8.52501318e-02 -2.04728141e-01
1.71070978e-01 2.17269018e-01 -2.82662243e-01 5.02475679e-01
4.16698247e-01 -1.11050153e+00 5.84312320e-01 5.83149612e-01
4.94977802e-01 -6.43388927e-01 9.10689116e-01 3.60497236e-01
-9.20711577e-01 6.09386005e-02 -9.10592377e-02 -1.13149382e-01
6.56915968e-03 2.00029850e-01 -7.81226695e-01 7.28531003e-01
7.96655297e-01 6.67646170e-01 -4.12900269e-01 6.41320646e-01
-5.49671948e-01 5.80424964e-01 -1.20238408e-01 -9.70344543e-02
-1.77659184e-01 1.45177273e-02 7.41438687e-01 1.56745875e+00
-5.03447130e-02 4.24464107e-01 -2.51929294e-02 8.39620650e-01
-1.87870145e-01 3.48143697e-01 -2.15932682e-01 -2.50180721e-01
1.17000639e+00 1.12953043e+00 -2.17487603e-01 5.15207015e-02
-4.50829983e-01 4.02994782e-01 9.24237251e-01 3.46935987e-01
-5.32241523e-01 -1.52168006e-01 1.07698357e+00 6.93705603e-02
-6.91000670e-02 -1.02728523e-01 -1.23017952e-01 -1.08729541e+00
8.75841901e-02 -1.08752239e+00 8.15777659e-01 -6.49605155e-01
-1.40682662e+00 4.65903282e-01 1.39622808e-01 -7.31159151e-01
-2.42413178e-01 -5.32872975e-01 -3.70052993e-01 8.79311025e-01
-1.86365128e+00 -7.40086496e-01 -2.05989838e-01 4.00027514e-01
3.26175392e-01 2.90788654e-02 1.18170440e+00 3.08310866e-01
-1.02395892e+00 6.95971549e-01 -2.90448070e-01 3.11973929e-01
1.14322197e+00 -1.04959035e+00 1.39830187e-01 6.59672141e-01
-8.75548646e-03 1.10129058e+00 1.00953174e+00 -3.78328949e-01
-1.59175634e+00 -8.32142770e-01 1.59561908e+00 -8.16909671e-01
8.08967412e-01 6.58863112e-02 -1.19447768e+00 9.10434306e-01
2.56274819e-01 -2.50478059e-01 7.95484424e-01 7.08123803e-01
-5.46884358e-01 -1.07390173e-01 -9.06428516e-01 6.01720989e-01
1.02310681e+00 -4.96855706e-01 -1.08193600e+00 2.44032487e-01
4.99834925e-01 -7.41759300e-01 -1.09253061e+00 4.64613527e-01
2.53580481e-01 -6.91618145e-01 9.48620260e-01 -6.79666042e-01
4.81814295e-01 -2.17411146e-01 -3.72653246e-01 -9.20682669e-01
-4.75780398e-01 -4.35594201e-01 -8.35113525e-02 1.50567722e+00
5.69569767e-01 -5.73154807e-01 3.03594232e-01 7.56275058e-01
8.33827183e-02 -6.66652858e-01 -1.12487304e+00 -3.80722731e-01
2.23615259e-01 -4.59430516e-01 7.56439090e-01 1.35282803e+00
5.56546450e-01 6.65815353e-01 -1.50034711e-01 4.95108277e-01
4.40008819e-01 4.16903406e-01 5.36054671e-01 -1.44644570e+00
1.10536270e-01 -5.87356925e-01 -4.75999750e-02 -6.48310423e-01
6.02361679e-01 -1.07450128e+00 -1.04342185e-01 -1.55317605e+00
4.65359181e-01 -3.37311715e-01 -4.29630697e-01 8.22624624e-01
-5.96062720e-01 -1.63244367e-01 1.82408348e-01 4.83179212e-01
-8.08111548e-01 2.05398664e-01 1.02844429e+00 -7.66936168e-02
-5.12944572e-02 -3.75811696e-01 -1.30848217e+00 7.16171682e-01
8.45602334e-01 -3.63147914e-01 2.84498576e-02 -6.65971220e-01
3.76096874e-01 5.58900610e-02 7.50603527e-02 -4.05587196e-01
4.75464702e-01 -5.95218241e-01 -1.61081403e-01 -1.48320675e-01
3.50190431e-01 -3.34343165e-01 4.80816215e-02 1.07819987e-02
-3.01947564e-01 1.82933167e-01 3.44786644e-01 2.33845517e-01
-3.10248822e-01 -4.65386324e-02 4.66196537e-01 -1.15129031e-01
-1.03335428e+00 -5.95707670e-02 -1.96062312e-01 5.44405937e-01
6.58679783e-01 3.64192128e-01 -7.12511241e-01 -2.09472388e-01
-5.56260824e-01 4.87999678e-01 2.77047753e-01 7.64960110e-01
3.36865336e-01 -1.13602877e+00 -9.72657084e-01 -2.66667128e-01
3.20609182e-01 1.21636167e-02 -4.54846434e-02 7.25679874e-01
-2.27617413e-01 6.78541183e-01 -1.04710591e-04 -3.48817468e-01
-1.45201695e+00 2.50910401e-01 3.52442861e-01 -2.57802606e-01
-4.17339027e-01 7.95057833e-01 -1.22791879e-01 -4.85888392e-01
3.69098961e-01 8.10404047e-02 -7.42997944e-01 3.33367914e-01
7.71057606e-01 2.03853980e-01 3.64148058e-02 -8.74417245e-01
-7.45682955e-01 3.27261031e-01 -3.12017232e-01 2.37827405e-01
1.43569243e+00 1.03098318e-01 -4.62121040e-01 3.95778894e-01
6.67569339e-01 3.08039635e-01 -5.20511389e-01 -5.41294515e-01
6.03110790e-01 -2.19308928e-01 -1.63549066e-01 -8.91818047e-01
-9.45606947e-01 4.96977448e-01 2.81698424e-02 -9.96411741e-02
8.63801658e-01 2.91233152e-01 3.62433374e-01 4.05985236e-01
4.65010822e-01 -1.01693106e+00 -3.89580429e-01 7.43008912e-01
1.27743065e+00 -1.09400737e+00 3.40315066e-02 -6.91324413e-01
-6.81568861e-01 8.62807035e-01 7.32928634e-01 1.14800595e-01
2.79943347e-01 4.45333034e-01 2.82369763e-01 -1.95805132e-01
-1.01409185e+00 -1.97466627e-01 1.72244623e-01 4.48365748e-01
1.07449579e+00 3.33187401e-01 -9.04677927e-01 1.25840688e+00
-5.35710812e-01 -1.14601053e-01 5.05426228e-01 6.98053956e-01
-2.46107444e-01 -1.43123794e+00 -3.66489589e-01 1.27559289e-01
-6.25363588e-01 -9.48986784e-02 -7.63366044e-01 8.67259085e-01
-1.52645454e-01 1.08861113e+00 -1.92480713e-01 -1.54775470e-01
7.84883201e-01 1.30565688e-01 3.59374791e-01 -5.92115998e-01
-3.86367112e-01 -3.67171645e-01 7.71493912e-01 -6.38347149e-01
-5.05405724e-01 -1.04982793e+00 -1.52236366e+00 -5.86908638e-01
-7.30772465e-02 4.97382849e-01 2.75058776e-01 1.01177239e+00
3.42263073e-01 2.98016101e-01 1.01051599e-01 -3.23130846e-01
-5.89751124e-01 -1.00463367e+00 -2.94239074e-01 5.23092330e-01
4.15094346e-02 -8.37713838e-01 -3.13217759e-01 -3.08005124e-01] | [9.353658676147461, 9.480691909790039] |
a4be58b1-5d45-4286-a1cc-2d3b5e48031b | selfpromer-self-prompt-dehazing-transformers | 2303.07033 | null | https://arxiv.org/abs/2303.07033v2 | https://arxiv.org/pdf/2303.07033v2.pdf | SelfPromer: Self-Prompt Dehazing Transformers with Depth-Consistency | This work presents an effective depth-consistency self-prompt Transformer for image dehazing. It is motivated by an observation that the estimated depths of an image with haze residuals and its clear counterpart vary. Enforcing the depth consistency of dehazed images with clear ones, therefore, is essential for dehazing. For this purpose, we develop a prompt based on the features of depth differences between the hazy input images and corresponding clear counterparts that can guide dehazing models for better restoration. Specifically, we first apply deep features extracted from the input images to the depth difference features for generating the prompt that contains the haze residual information in the input. Then we propose a prompt embedding module that is designed to perceive the haze residuals, by linearly adding the prompt to the deep features. Further, we develop an effective prompt attention module to pay more attention to haze residuals for better removal. By incorporating the prompt, prompt embedding, and prompt attention into an encoder-decoder network based on VQGAN, we can achieve better perception quality. As the depths of clear images are not available at inference, and the dehazed images with one-time feed-forward execution may still contain a portion of haze residuals, we propose a new continuous self-prompt inference that can iteratively correct the dehazing model towards better haze-free image generation. Extensive experiments show that our method performs favorably against the state-of-the-art approaches on both synthetic and real-world datasets in terms of perception metrics including NIQE, PI, and PIQE. | ['Xiao-Ming Wu', 'Jiangxin Dong', 'WanYu Lin', 'Jinshan Pan', 'Cong Wang'] | 2023-03-13 | null | null | null | null | ['image-dehazing'] | ['computer-vision'] | [ 2.32131436e-01 -6.18852489e-02 4.62357342e-01 -4.62589741e-01
-3.25603127e-01 -1.49059430e-01 4.87993300e-01 -3.45019400e-01
-1.35891110e-04 4.68850642e-01 3.79463971e-01 -7.12379143e-02
-1.95477530e-02 -1.15911829e+00 -8.56410027e-01 -1.08820915e+00
1.70294330e-01 -3.89276356e-01 4.12647218e-01 -5.73988318e-01
3.89143735e-01 4.49319035e-02 -1.88993979e+00 5.56868792e-01
1.33041048e+00 1.11557364e+00 5.87654889e-01 8.83472860e-01
1.07275784e-01 1.24222040e+00 -8.16033900e-01 -1.46095559e-01
4.18487787e-01 -5.51579356e-01 -3.05021286e-01 2.70456553e-01
7.37711549e-01 -1.11218190e+00 -6.06279433e-01 1.23273742e+00
4.63343710e-01 1.65062696e-01 5.92469335e-01 -1.26456237e+00
-1.42197144e+00 1.16576262e-01 -4.40576524e-01 2.93449074e-01
1.12871975e-01 6.61816180e-01 5.82841516e-01 -9.67845380e-01
-4.54377569e-02 1.26818287e+00 1.67383507e-01 5.37424624e-01
-6.77883446e-01 -7.95883060e-01 3.20419788e-01 5.85335016e-01
-1.29447889e+00 -4.82432246e-01 7.58023739e-01 -3.83632958e-01
7.05815792e-01 2.65971363e-01 5.50189137e-01 6.83800995e-01
5.54165900e-01 4.94921058e-01 1.02984405e+00 -2.02582449e-01
1.80485949e-01 1.30579114e-01 -2.13077083e-01 5.30367613e-01
2.29879022e-01 3.93374711e-01 -5.78574359e-01 5.34586549e-01
5.67489386e-01 1.64449751e-01 -6.27052307e-01 2.48621970e-01
-8.82219732e-01 5.55551171e-01 7.39877462e-01 -1.26338750e-01
-4.90304887e-01 2.23752618e-01 -2.78597742e-01 4.11930621e-01
5.89303732e-01 3.43305051e-01 -1.11338004e-01 3.73662561e-01
-7.15413213e-01 2.36402676e-01 2.12305680e-01 9.51424420e-01
1.18103337e+00 3.23797077e-01 -2.28622913e-01 6.39164269e-01
5.06526351e-01 7.14153409e-01 1.44504011e-01 -1.11888266e+00
4.16718692e-01 4.05653030e-01 3.21448326e-01 -1.13698208e+00
2.80368954e-01 -2.58886456e-01 -1.12854660e+00 7.96089888e-01
-9.72803757e-02 1.33981153e-01 -1.23631525e+00 1.40035343e+00
1.72488496e-01 3.82551938e-01 2.36577764e-01 1.23419547e+00
8.26797545e-01 1.34957910e+00 -3.89488310e-01 -1.16269104e-01
1.13763726e+00 -1.20231378e+00 -1.04898143e+00 -4.95706409e-01
4.14246805e-02 -7.15271592e-01 1.00793123e+00 5.62895000e-01
-1.09169972e+00 -9.09283400e-01 -1.34485757e+00 -3.38482052e-01
-3.55043083e-01 -1.52105451e-01 8.86845067e-02 3.52850020e-01
-1.35102928e+00 3.44391525e-01 -5.00390887e-01 8.09331909e-02
3.00614864e-01 1.78772826e-02 -1.10749044e-01 -6.19538784e-01
-1.39297509e+00 9.82631028e-01 3.48954350e-01 6.16457641e-01
-1.31009591e+00 -7.60152698e-01 -1.02174377e+00 1.48447037e-01
2.29066089e-01 -8.14076602e-01 9.43884015e-01 -1.10212302e+00
-1.54463506e+00 3.31296593e-01 -1.34547114e-01 -2.19947383e-01
1.48806274e-01 -3.41112524e-01 -5.32671928e-01 2.97031045e-01
-1.01723902e-01 9.06809151e-01 1.47977960e+00 -1.77712107e+00
-9.05231714e-01 -7.81076849e-02 5.03295898e-01 4.52243924e-01
-3.76853198e-01 -3.65972489e-01 -3.48361015e-01 -5.35416245e-01
-2.89138146e-02 -3.78107876e-01 -1.11421771e-01 3.95155907e-01
-3.59810919e-01 1.60457283e-01 9.14913177e-01 -9.06319141e-01
1.09280133e+00 -2.44298053e+00 4.20006327e-02 -8.91010985e-02
6.11184895e-01 2.72866637e-01 -2.31949523e-01 2.51649916e-01
4.90563810e-02 8.23043287e-03 -2.75007159e-01 -2.35637531e-01
-1.04536600e-01 4.97166365e-01 -6.78101361e-01 4.69045520e-01
5.68868577e-01 6.30219162e-01 -9.10867035e-01 -2.30560631e-01
5.80352902e-01 6.12687290e-01 -5.62756360e-01 9.89110112e-01
-2.40689009e-01 3.60097557e-01 -4.73970510e-02 3.88264090e-01
1.24606597e+00 4.44089398e-02 -5.77357829e-01 -3.30786258e-01
-3.59979481e-01 2.11661413e-01 -8.39552760e-01 1.14497554e+00
-5.01592398e-01 7.18495786e-01 1.19729191e-01 -5.73002517e-01
8.27736020e-01 1.62015960e-01 -2.19521627e-01 -9.83949482e-01
-3.93703058e-02 2.19556354e-02 -2.11074665e-01 -8.01376760e-01
7.10413635e-01 -1.32769495e-01 5.71613967e-01 4.49939817e-03
-2.60652035e-01 -5.14943242e-01 -1.90353394e-01 2.01467097e-01
7.95427978e-01 -1.52156502e-01 -5.30211218e-02 -4.27810429e-03
6.54018700e-01 -3.87426019e-01 5.16015291e-01 6.40655935e-01
-1.36876509e-01 1.04084682e+00 2.47459680e-01 -3.75032872e-01
-9.82082427e-01 -1.22361171e+00 8.53342861e-02 8.01115990e-01
6.73389196e-01 -1.25199273e-01 -6.98733091e-01 -9.58667621e-02
-3.84964079e-01 8.62610817e-01 -8.37785959e-01 -6.04024053e-01
-3.08724672e-01 -5.58187723e-01 1.11550272e-01 1.39963329e-01
1.00838625e+00 -8.67610037e-01 -4.66176867e-01 8.27989578e-02
-3.25302750e-01 -1.02204871e+00 -5.56899607e-01 5.41122747e-04
-5.01938939e-01 -7.72478640e-01 -6.58507943e-01 -7.84923494e-01
7.84446716e-01 8.93646061e-01 7.84167945e-01 4.14764494e-01
5.55405170e-02 7.35185072e-02 -4.97983932e-01 -4.35307890e-01
-5.06098688e-01 -4.97469962e-01 -2.66993016e-01 2.27830946e-01
-1.97463214e-01 -6.59747899e-01 -1.13197470e+00 2.04722300e-01
-1.55037141e+00 4.55331892e-01 5.86836696e-01 7.73123980e-01
3.25602263e-01 6.59633219e-01 1.79621637e-01 -5.14836729e-01
3.30523431e-01 -7.03648508e-01 -5.59934139e-01 -6.72533736e-02
-8.05150211e-01 -1.95698012e-02 7.77433872e-01 -2.75428236e-01
-1.38603055e+00 -4.88114178e-01 -1.08128726e-01 -6.61106706e-01
-9.45193097e-02 2.32145309e-01 -4.75311100e-01 -9.12873596e-02
5.75749695e-01 6.27027810e-01 -4.62283902e-02 -2.28856653e-01
4.74328399e-01 8.40826809e-01 8.86715293e-01 -2.01068506e-01
1.36124957e+00 6.72592282e-01 -3.68078321e-01 -7.09560752e-01
-9.41127837e-01 -2.68148452e-01 -1.60033390e-01 -2.99864471e-01
1.05242634e+00 -1.27607906e+00 -4.63515073e-01 1.03770638e+00
-1.44416034e+00 -4.34046924e-01 1.72702260e-02 1.57744125e-01
-1.90525234e-01 5.69959223e-01 -7.32370913e-01 -8.53821754e-01
-2.56533474e-01 -1.11228359e+00 1.10387909e+00 4.45343196e-01
5.78121662e-01 -8.14325631e-01 -1.72942832e-01 3.20425153e-01
6.99641228e-01 -3.90240923e-02 1.06214643e+00 2.61843711e-01
-1.22922039e+00 3.36843461e-01 -4.94416386e-01 9.81741905e-01
4.72849250e-01 1.46151438e-01 -1.25371516e+00 -2.44166836e-01
3.23475033e-01 1.39080398e-02 1.14377117e+00 2.68796057e-01
1.18068647e+00 -6.33086026e-01 2.39991814e-01 9.95009661e-01
1.49777925e+00 2.33493000e-01 1.28141284e+00 4.60274875e-01
8.74202132e-01 7.07736909e-01 7.80402780e-01 3.64169955e-01
6.94201231e-01 2.32000515e-01 1.11336434e+00 -5.00206351e-01
-4.52485532e-01 -7.86718056e-02 5.95267653e-01 8.30160201e-01
5.22024930e-02 -7.67302871e-01 -3.54149550e-01 7.25017726e-01
-1.59274650e+00 -8.64137769e-01 -1.11709617e-01 1.91033876e+00
9.44476545e-01 9.49784890e-02 -6.54249489e-01 7.92696998e-02
5.75447798e-01 5.81640184e-01 -3.78515035e-01 -3.51554811e-01
-3.12549591e-01 4.73833159e-02 3.36500674e-01 7.68088996e-01
-8.33816409e-01 9.07751679e-01 5.58633232e+00 5.35559654e-01
-1.19942904e+00 -1.15939006e-01 6.16722882e-01 5.06868400e-02
-7.54298270e-01 -9.99879651e-03 -4.02992845e-01 8.19052756e-01
7.83356369e-01 -1.02951229e-01 8.12293172e-01 4.70067143e-01
5.33874989e-01 -3.50904197e-01 -9.76811230e-01 1.00393093e+00
2.16268599e-01 -1.24891651e+00 3.16915423e-01 -1.98436044e-02
9.87647176e-01 -3.18175375e-01 4.83195037e-01 1.31016463e-01
1.14931159e-01 -1.07268620e+00 9.68121588e-01 6.27741158e-01
7.31513560e-01 -5.06897569e-01 9.22625303e-01 2.86469996e-01
-9.52417135e-01 -1.49614036e-01 -7.27997959e-01 -2.44543493e-01
-1.14924395e-02 8.45150948e-01 -5.90640724e-01 6.97682738e-01
9.51187670e-01 8.36828530e-01 -5.52762806e-01 8.52686584e-01
-6.26017749e-01 5.58354795e-01 1.86294407e-01 6.27866685e-01
1.42398447e-01 -1.66381806e-01 2.68554866e-01 8.86210680e-01
4.11073416e-01 4.51687127e-01 -3.12483281e-01 1.03143442e+00
1.90406129e-01 -5.32533705e-01 -5.25993884e-01 3.46823573e-01
4.09512699e-01 9.58613932e-01 -1.88698117e-02 -4.39187616e-01
-4.02386844e-01 1.31901991e+00 -6.59819469e-02 6.76301360e-01
-1.00992036e+00 -5.02316475e-01 1.15699005e+00 -4.48786281e-02
3.86550248e-01 -5.41380718e-02 -1.68368906e-01 -9.75443482e-01
1.06956065e-01 -8.87953997e-01 -3.62621024e-02 -1.53928852e+00
-1.33540690e+00 6.65705681e-01 -9.84178111e-02 -1.23677218e+00
7.62459934e-02 -5.59000731e-01 -8.61018658e-01 1.19119477e+00
-2.39114189e+00 -8.76164377e-01 -1.13323188e+00 6.61364019e-01
6.71439052e-01 3.45038444e-01 3.79662335e-01 4.00282592e-01
-5.24846435e-01 3.32986563e-01 -3.92867215e-02 -2.01181874e-01
8.87387276e-01 -1.16229033e+00 3.93848240e-01 1.43997431e+00
-4.49378431e-01 4.21384513e-01 9.47070658e-01 -4.36336696e-01
-1.36765122e+00 -1.45896637e+00 3.39781076e-01 -2.71617711e-01
2.77201027e-01 -1.87286898e-01 -1.35558832e+00 5.04168272e-01
5.68350017e-01 1.64846480e-01 4.90423292e-02 -8.28911543e-01
-4.40416425e-01 -5.73778808e-01 -8.85114610e-01 6.77491307e-01
8.32004249e-01 -6.11570299e-01 -6.36619270e-01 8.33169818e-02
1.24057126e+00 -5.81511617e-01 -4.74669069e-01 3.31265479e-01
2.30698749e-01 -1.38873899e+00 9.67301607e-01 4.84452993e-02
9.52478588e-01 -7.88455725e-01 -2.78516620e-01 -1.56444025e+00
-5.32736182e-01 -5.63689232e-01 -1.39269382e-01 1.15190172e+00
6.73827678e-02 -6.63905859e-01 4.28517729e-01 2.59211272e-01
-5.27667284e-01 -5.81682861e-01 -5.66761434e-01 -5.36522925e-01
-6.03132807e-02 -2.18695402e-01 9.10481691e-01 7.72468865e-01
-6.72834635e-01 2.40643471e-02 -7.92351246e-01 1.01853418e+00
8.09950888e-01 3.75104183e-03 7.64044225e-01 -6.42300904e-01
-1.17238015e-01 -5.57489917e-02 -4.20795262e-01 -1.34615827e+00
-1.08096153e-01 -2.81836182e-01 6.18772805e-01 -1.82941782e+00
2.58310046e-02 -1.42612725e-01 -3.79391074e-01 4.86764282e-01
-7.34726846e-01 2.05180123e-01 1.25198245e-01 1.95954695e-01
-2.14248568e-01 9.65645075e-01 1.56925881e+00 -4.89899844e-01
2.78732702e-02 -3.44409198e-01 -9.10164833e-01 2.92500407e-01
6.43419802e-01 -3.89536291e-01 -7.33149886e-01 -9.33395743e-01
1.57976374e-01 -4.04547043e-02 5.40195286e-01 -1.06972420e+00
2.75395840e-01 -5.02100766e-01 4.32055235e-01 -5.61104119e-01
4.76615310e-01 -7.36753702e-01 6.39463123e-03 3.06457102e-01
-6.59608990e-02 -1.84104890e-01 -9.75743402e-03 5.54399788e-01
-5.27184248e-01 7.24698231e-02 8.38827848e-01 -4.78752814e-02
-9.70559359e-01 4.77678627e-01 -5.13305128e-01 -2.15600595e-01
7.52268851e-01 -5.69903791e-01 -8.57683063e-01 -7.48326898e-01
-5.88952415e-02 3.30974936e-01 6.01679862e-01 4.70937461e-01
1.27105498e+00 -1.07768321e+00 -8.50516081e-01 4.71986860e-01
2.25036114e-01 5.63375413e-01 5.82323432e-01 4.66468751e-01
-7.39861846e-01 -9.21183378e-02 -2.17062160e-01 -4.80755329e-01
-9.01059210e-01 6.80188417e-01 4.93492395e-01 3.07939798e-01
-6.26851737e-01 9.13722992e-01 1.06621802e+00 -2.95731192e-03
2.65355036e-02 -6.00031793e-01 -5.65680005e-02 -3.99619997e-01
8.94304037e-01 2.22961694e-01 -2.56746430e-02 -2.42824778e-01
-3.70885730e-02 4.15546894e-01 9.74666048e-03 3.49009596e-02
1.25747859e+00 -6.54393375e-01 -4.14009601e-01 2.01689437e-01
1.05545163e+00 1.04205407e-01 -1.81356347e+00 -1.65879592e-01
-7.33801126e-01 -9.98847723e-01 3.07068646e-01 -7.31249988e-01
-1.36535966e+00 1.11597538e+00 7.17006505e-01 1.04285143e-01
1.67219877e+00 -2.42494822e-01 9.83643353e-01 1.04828596e-01
2.56421864e-02 -5.57992816e-01 4.31304365e-01 4.56108928e-01
9.94646788e-01 -1.20558536e+00 -8.09282288e-02 -5.33121228e-01
-4.65742528e-01 9.96446252e-01 1.08382463e+00 -5.88048948e-03
3.49586368e-01 5.31027615e-02 4.55297112e-01 -1.72564700e-01
-1.00227451e+00 -4.60764356e-02 2.05536216e-01 8.07323694e-01
-1.62995443e-01 -2.50224918e-01 3.97648335e-01 2.96645969e-01
-1.49702489e-01 -2.24090099e-01 9.35910702e-01 5.39839327e-01
-8.63490939e-01 -4.69520867e-01 -6.22422159e-01 2.84662507e-02
1.17812574e-01 -4.84722465e-01 -3.26221772e-02 3.62598568e-01
4.78832722e-01 1.38103783e+00 1.82688788e-01 -6.64529562e-01
1.55196384e-01 -5.53065598e-01 4.04123992e-01 -6.33948445e-01
-1.76321357e-01 -1.93107203e-01 -2.98472822e-01 -5.06397128e-01
-2.63611078e-01 1.44381359e-01 -1.12372768e+00 -5.57408750e-01
-3.14608276e-01 -9.92631614e-02 3.91054362e-01 8.52720737e-01
3.19213450e-01 9.15022254e-01 1.15331471e+00 -8.78053963e-01
-3.09267014e-01 -8.71268988e-01 -6.89488769e-01 3.13449740e-01
1.08293569e+00 -4.58587080e-01 -8.15394700e-01 2.71363199e-01] | [10.977899551391602, -3.231940746307373] |
164f6988-2a74-43d0-ae9b-a36462e7bc71 | flexibly-mining-better-subgroups | 1510.08382 | null | http://arxiv.org/abs/1510.08382v1 | http://arxiv.org/pdf/1510.08382v1.pdf | Flexibly Mining Better Subgroups | In subgroup discovery, also known as supervised pattern mining, discovering
high quality one-dimensional subgroups and refinements of these is a crucial
task. For nominal attributes, this is relatively straightforward, as we can
consider individual attribute values as binary features. For numerical
attributes, the task is more challenging as individual numeric values are not
reliable statistics. Instead, we can consider combinations of adjacent values,
i.e. bins. Existing binning strategies, however, are not tailored for subgroup
discovery. That is, they do not directly optimize for the quality of subgroups,
therewith potentially degrading the mining result.
To address this issue, we propose FLEXI. In short, with FLEXI we propose to
use optimal binning to find high quality binary features for both numeric and
ordinal attributes. We instantiate FLEXI with various quality measures and show
how to achieve efficiency accordingly. Experiments on both synthetic and
real-world data sets show that FLEXI outperforms state of the art with up to 25
times improvement in subgroup quality. | ['Hoang-Vu Nguyen', 'Jilles Vreeken'] | 2015-10-28 | null | null | null | null | ['subgroup-discovery'] | ['methodology'] | [ 1.98651284e-01 2.06020679e-02 -5.88549018e-01 -5.38606167e-01
-4.98306155e-01 -5.80766976e-01 1.08071193e-01 5.91347933e-01
-3.73168468e-01 1.03962493e+00 1.00222766e-01 -3.45612019e-01
-8.03213716e-01 -1.07769418e+00 -4.44212645e-01 -5.95679760e-01
-4.06552941e-01 9.01304364e-01 1.91141486e-01 1.99850708e-01
2.15534583e-01 3.53206545e-01 -1.84473658e+00 3.15204710e-01
9.34085786e-01 1.14157283e+00 -4.51052666e-01 8.71614143e-02
-3.87513228e-02 -9.84070674e-02 -5.97584724e-01 -2.96011925e-01
2.92265356e-01 -2.57571071e-01 -7.35339403e-01 2.01617301e-01
-6.67266622e-02 -1.50742903e-01 3.27600926e-01 8.76384914e-01
4.52783704e-02 4.41657379e-02 7.32993126e-01 -1.56822276e+00
5.34468181e-02 9.34337497e-01 -6.89604223e-01 -1.75695829e-02
2.48464227e-01 -3.26201975e-01 1.42147017e+00 -6.62033856e-01
2.79650509e-01 1.08219361e+00 4.86358106e-01 2.15693235e-01
-1.62574387e+00 -7.36031890e-01 1.63973585e-01 1.48486480e-01
-1.67537248e+00 -4.50227678e-01 3.03991705e-01 -2.08685189e-01
6.09608352e-01 8.38707030e-01 4.97806519e-01 3.88599008e-01
-1.38114274e-01 4.50497419e-01 1.12802827e+00 -2.47098446e-01
4.32986826e-01 8.51250663e-02 4.24774081e-01 2.89299071e-01
9.27245796e-01 -9.10978094e-02 -4.92527485e-01 -5.83608925e-01
2.59624273e-01 1.59028485e-01 -2.17031568e-01 -4.82251763e-01
-1.43440306e+00 7.90678203e-01 -4.23798151e-02 -3.23559041e-03
-4.76609409e-01 -2.18340337e-01 1.86397687e-01 3.65423650e-01
3.09831172e-01 5.75098753e-01 -6.48995399e-01 -2.41851017e-01
-9.54534233e-01 5.13340056e-01 9.28141654e-01 7.91841567e-01
9.38851058e-01 -6.76582098e-01 -2.34846815e-01 8.91864121e-01
1.09564357e-01 5.73406108e-02 2.23507300e-01 -8.27932894e-01
4.27769363e-01 9.02152777e-01 2.94242561e-01 -8.47924590e-01
-7.03146696e-01 -2.33829379e-01 -1.05801606e+00 5.76328300e-03
7.20264256e-01 -2.26526670e-02 -9.35249329e-01 1.62936985e+00
4.10870403e-01 -2.30249632e-02 -4.83610742e-02 7.08130121e-01
4.37171936e-01 3.74176204e-01 2.41630003e-02 -6.16258442e-01
1.51256144e+00 -3.18033576e-01 -6.31331682e-01 -3.13638411e-02
5.25422335e-01 -3.48274827e-01 8.52407753e-01 6.01788580e-01
-9.52177048e-01 -1.35243356e-01 -1.06777346e+00 5.75965226e-01
-3.90559047e-01 -2.01789334e-01 8.67648721e-01 6.74395144e-01
-4.88239467e-01 7.78421104e-01 -7.23881483e-01 -1.70405194e-01
3.30463946e-01 7.09756136e-01 -4.72392708e-01 -7.08480328e-02
-1.05793834e+00 2.35160097e-01 5.20859122e-01 -2.36834615e-01
3.55974846e-02 -7.81294048e-01 -7.67278075e-01 3.20830613e-01
7.86895275e-01 -4.03429091e-01 9.31848705e-01 -3.08254153e-01
-8.67170572e-01 4.74372327e-01 -2.43791297e-01 -4.77282345e-01
1.05586559e-01 6.36973605e-02 -3.35707158e-01 -2.97697932e-01
2.64725119e-01 4.23180103e-01 3.90865892e-01 -1.09848225e+00
-1.01578820e+00 -4.61273223e-01 -2.15157866e-02 -5.82901947e-02
-6.47544622e-01 -1.14528805e-01 -3.23927969e-01 -4.73793715e-01
5.18472970e-01 -8.51914048e-01 -4.89630908e-01 -2.97403246e-01
-5.19498110e-01 -4.48537469e-01 4.15086776e-01 -1.26028717e-01
1.68383038e+00 -2.16542864e+00 -1.01131923e-01 9.04065132e-01
3.66211385e-01 -1.99397430e-01 3.10654819e-01 1.28464088e-01
1.31764844e-01 3.22481483e-01 -3.11538905e-01 -1.75821576e-02
1.20376281e-01 6.24267817e-01 -1.42113529e-02 4.12302256e-01
4.47379574e-02 6.02879941e-01 -7.43013382e-01 -5.78180313e-01
-1.54063836e-01 -8.12724456e-02 -8.88137043e-01 -7.70849064e-02
-1.79796696e-01 4.85212952e-02 -4.16476488e-01 7.70265877e-01
8.00481677e-01 -4.23457563e-01 3.41034979e-01 -1.52996823e-01
-9.17228460e-02 2.56249040e-01 -1.67233288e+00 9.27929640e-01
-2.36479118e-02 -9.22842473e-02 -3.18081528e-02 -1.25229371e+00
8.81114364e-01 1.00457057e-01 9.07115877e-01 -4.24214959e-01
-1.70065656e-01 4.45300192e-01 2.86649674e-01 -4.14092764e-02
4.92205292e-01 -1.62407681e-01 -3.75157863e-01 4.47156399e-01
-4.68069881e-01 2.94337928e-01 5.37338018e-01 -1.90339357e-01
1.17067897e+00 -4.46356267e-01 6.97580457e-01 -3.19594324e-01
2.32347921e-01 2.92821974e-02 1.13266945e+00 6.68751478e-01
2.54948616e-01 6.57652438e-01 8.94658089e-01 -2.03990564e-01
-7.57967830e-01 -9.45962071e-01 -4.42896426e-01 8.37405980e-01
1.02824628e-01 -7.62100875e-01 -3.83857191e-01 -7.02913642e-01
4.97122914e-01 4.57664937e-01 -6.17684960e-01 6.19629771e-02
-4.43476617e-01 -1.24461269e+00 -6.70749098e-02 4.62289840e-01
7.27137551e-02 -7.80231833e-01 -7.99398601e-01 5.35247743e-01
-1.97628755e-02 -1.02853858e+00 -2.47641683e-01 5.57914913e-01
-7.79477775e-01 -1.17228234e+00 -1.79641679e-01 -2.95775741e-01
6.03661954e-01 1.36956543e-01 1.14610732e+00 1.92966267e-01
-1.90648511e-01 -2.46942803e-01 -5.26818693e-01 -3.14494491e-01
-7.05151409e-02 2.83069193e-01 8.10146704e-02 1.76638931e-01
5.24626911e-01 -8.97297025e-01 -5.47769964e-01 6.45305514e-01
-9.21835542e-01 -9.45394337e-02 7.12551475e-01 8.24328899e-01
8.81764412e-01 5.99813282e-01 7.77099550e-01 -1.03094614e+00
4.93796885e-01 -6.30499065e-01 -6.35382712e-01 2.16525957e-01
-1.06578672e+00 1.24218263e-01 6.34687245e-01 -3.58953118e-01
-4.12325293e-01 6.40247539e-02 1.78831257e-02 6.75116703e-02
-3.62496287e-01 7.25594521e-01 -4.84530598e-01 2.60917395e-01
3.86589289e-01 -3.69791463e-02 -2.16312781e-02 -6.11655533e-01
-1.27798632e-01 8.56623292e-01 2.59899855e-01 -7.21503854e-01
5.56592584e-01 4.36460286e-01 3.92907917e-01 -4.41939086e-01
-6.99906826e-01 -6.02897584e-01 -4.80449140e-01 2.23055869e-01
3.98054898e-01 -3.08322579e-01 -9.82208610e-01 -7.54973814e-02
-5.31103373e-01 -9.69566330e-02 -2.46219680e-01 3.93975735e-01
-4.89412904e-01 2.82081008e-01 -9.63262618e-02 -6.83650076e-01
-1.38972536e-01 -1.24453747e+00 9.18936312e-01 1.22462407e-01
-5.21908581e-01 -4.36577708e-01 -2.72868127e-01 4.69445325e-02
1.12648547e-01 4.79960442e-01 1.34244108e+00 -1.07020950e+00
-3.03350121e-01 -2.58166015e-01 -1.84369266e-01 -1.48554608e-01
4.51545030e-01 -1.83265001e-01 -4.68347520e-01 -1.70404673e-01
-5.63046455e-01 -3.44515704e-02 6.32561386e-01 1.73649445e-01
1.67122781e+00 -5.94085515e-01 -6.19168222e-01 6.11542284e-01
1.08986223e+00 1.95856228e-01 4.20251191e-01 4.90636349e-01
3.42089176e-01 7.20043004e-01 1.17836964e+00 9.92015243e-01
4.10498351e-01 9.45542574e-01 4.18846518e-01 7.60006383e-02
3.85020196e-01 1.94604740e-01 -1.95468768e-01 -5.92132146e-03
-1.36006966e-01 -1.27859682e-01 -9.69854534e-01 5.63692629e-01
-1.98432279e+00 -7.97967374e-01 -1.09470308e-01 2.61147404e+00
1.13497937e+00 4.01761264e-01 6.40801132e-01 8.50621819e-01
4.74718899e-01 -3.90810519e-01 -4.83460367e-01 -1.34245425e-01
1.10599965e-01 2.61404365e-01 5.52907109e-01 2.78269202e-01
-9.95024443e-01 3.68790478e-01 5.54042578e+00 6.54369533e-01
-6.70477211e-01 -3.16311777e-01 7.96437919e-01 -3.18140835e-01
-3.22783977e-01 -1.21379793e-02 -1.07440257e+00 5.92790425e-01
7.87396550e-01 -3.20226312e-01 2.22340211e-01 7.37349391e-01
1.55261755e-01 -1.59193560e-01 -1.32049382e+00 7.83471465e-01
-5.35001338e-01 -1.02192473e+00 -5.22151813e-02 6.37958467e-01
5.33490181e-01 -6.61884248e-01 -8.99299309e-02 1.89699724e-01
1.59712687e-01 -1.11320579e+00 3.93394798e-01 2.67201692e-01
5.60982645e-01 -1.32859790e+00 6.50011241e-01 3.43015999e-01
-9.71597672e-01 -1.79285854e-01 -4.49867785e-01 1.03323616e-01
4.82601561e-02 9.74060535e-01 -6.47519112e-01 5.45600414e-01
9.13935721e-01 4.83764350e-01 -5.69807231e-01 1.37844431e+00
3.23103189e-01 7.23295331e-01 -7.25448906e-01 4.73536402e-02
4.64604460e-02 -1.66526496e-01 2.75520504e-01 1.03019226e+00
4.74399000e-01 1.38599545e-01 4.53706354e-01 4.19694513e-01
1.60698622e-01 2.29160726e-01 -2.54380018e-01 2.01900139e-01
7.21404970e-01 1.19965041e+00 -9.07274604e-01 -2.24045560e-01
-4.20064241e-01 4.52466160e-01 3.09316237e-02 2.33166050e-02
-6.83335841e-01 -5.07433772e-01 1.13697553e+00 4.95380938e-01
4.63431716e-01 -6.54446241e-03 -6.20030522e-01 -8.14077556e-01
1.31760329e-01 -1.14106452e+00 8.65000188e-01 1.44012004e-01
-1.12354445e+00 1.86301827e-01 4.71438795e-01 -1.23400223e+00
-4.17331219e-01 -5.56615531e-01 -2.31408268e-01 4.84811187e-01
-1.36319184e+00 -6.95712984e-01 -2.42759705e-01 4.99521554e-01
6.94317520e-02 1.70617491e-01 7.70220339e-01 3.44032586e-01
-4.98263448e-01 1.01026559e+00 1.33768022e-01 -1.81211948e-01
6.52851522e-01 -1.17908716e+00 1.51918799e-01 4.37863916e-01
6.50273412e-02 5.71088135e-01 9.39085186e-01 -5.33434331e-01
-1.26857388e+00 -1.07669365e+00 7.96554565e-01 2.07677800e-02
4.99332249e-01 -4.04639959e-01 -1.09288573e+00 3.01561803e-01
-3.70507538e-01 1.90061241e-01 1.00182927e+00 4.87615377e-01
-1.28201455e-01 -3.60078126e-01 -1.35009718e+00 5.63783884e-01
1.16190636e+00 3.65153462e-01 -1.88064009e-01 1.85467556e-01
4.89117920e-01 -5.11337332e-02 -1.35180223e+00 9.26098228e-01
5.83700657e-01 -1.02146113e+00 9.99179900e-01 -4.92802978e-01
-2.46264134e-03 -5.23966730e-01 -3.02985400e-01 -1.10577035e+00
-3.87700230e-01 -4.50902194e-01 -2.74779886e-01 1.34541833e+00
7.50189900e-01 -7.68139660e-01 1.04028881e+00 8.52315784e-01
1.90811381e-01 -1.06358993e+00 -9.64755774e-01 -9.50139701e-01
-2.72352397e-01 -4.19041544e-01 1.40428483e+00 6.66048110e-01
3.24502051e-01 2.51935780e-01 -2.77629316e-01 1.12072781e-01
8.69884610e-01 7.22335041e-01 6.70008123e-01 -1.74156988e+00
-4.17060286e-01 -5.85958123e-01 -4.61573184e-01 -6.61828697e-01
-3.81032303e-02 -6.36971354e-01 3.13254483e-02 -1.32155037e+00
2.93813676e-01 -1.07649291e+00 -4.39444453e-01 8.39184165e-01
-3.15988988e-01 3.64088535e-01 -1.68136820e-01 1.24067016e-01
-5.53676248e-01 5.09527288e-02 8.53148758e-01 2.26547085e-02
-4.40777123e-01 4.26258236e-01 -1.00895238e+00 4.99846965e-01
1.08695769e+00 -6.04837716e-01 -2.54349023e-01 3.40723932e-01
3.72752309e-01 9.19636264e-02 1.28406927e-01 -1.03994632e+00
2.53519583e-02 -6.70218885e-01 3.49513471e-01 -8.43071222e-01
1.97860867e-01 -9.16945040e-01 2.76351184e-01 3.08943063e-01
-1.30550086e-01 -2.56553739e-02 -4.36797552e-02 4.77882296e-01
-1.97642952e-01 -2.00504929e-01 6.31160676e-01 1.91270635e-01
-4.28360224e-01 3.74290049e-01 -1.84104696e-01 -1.08364508e-01
1.08803320e+00 -3.60755831e-01 -6.00593910e-02 -1.88474074e-01
-6.91208422e-01 4.27276373e-01 5.71204007e-01 1.00683272e-01
4.78909940e-01 -1.28805053e+00 -6.74365103e-01 3.16304564e-01
5.17850220e-01 3.21249545e-01 -2.78422445e-01 9.36825395e-01
2.06532657e-01 4.04606313e-01 3.46735269e-02 -5.06105602e-01
-1.48539889e+00 7.57084370e-01 -1.23202994e-01 -3.62611681e-01
-4.53041106e-01 4.48133171e-01 6.14204183e-02 -1.30591050e-01
3.31406325e-01 -4.03286278e-01 -3.42654318e-01 4.29671794e-01
5.31955957e-01 3.93603563e-01 2.22464576e-01 -2.43292853e-01
-5.89502931e-01 3.81611526e-01 -1.50115117e-01 3.61493565e-02
1.50788260e+00 -4.89704544e-03 -3.27214539e-01 1.75048277e-01
9.73038912e-01 9.30903181e-02 -7.78982997e-01 -2.68335462e-01
5.12794733e-01 -5.61094463e-01 -3.84069085e-01 -5.29937506e-01
-9.26223457e-01 1.99836105e-01 1.20443605e-01 5.81852138e-01
1.31506574e+00 1.13231629e-01 5.77582300e-01 3.87756228e-01
3.76999795e-01 -8.77496839e-01 -3.78435016e-01 2.19048381e-01
4.98345733e-01 -1.16039097e+00 2.33412415e-01 -5.71714044e-01
-4.09666836e-01 9.71937776e-01 5.64602494e-01 1.99316025e-01
8.12065601e-01 3.41547102e-01 -3.05333853e-01 -1.59618016e-02
-8.39081109e-01 -3.89811724e-01 3.19243848e-01 2.51135975e-01
4.19913352e-01 3.38317305e-01 -8.08142304e-01 8.90954733e-01
-4.54483747e-01 -2.22798318e-01 3.62380505e-01 9.82274890e-01
-5.12246966e-01 -1.48519266e+00 -7.11348832e-01 1.27418745e+00
-6.36234701e-01 1.35994434e-01 -4.46257368e-02 6.94571018e-01
2.05781534e-01 1.13978994e+00 1.54872030e-01 -4.55152750e-01
4.50248241e-01 -1.06966840e-02 2.09040985e-01 -6.15055799e-01
-3.54136944e-01 -2.33082622e-02 2.87202358e-01 -5.13574600e-01
-1.27828300e-01 -1.02453625e+00 -1.13933158e+00 -3.45576763e-01
-1.65647179e-01 5.44577062e-01 4.25437629e-01 7.27212429e-01
3.16918820e-01 2.56119043e-01 7.67146170e-01 -4.15064931e-01
-7.10884452e-01 -6.88869059e-01 -7.00223088e-01 3.53158325e-01
3.67700189e-01 -9.12758052e-01 -3.05092752e-01 -3.59138340e-01] | [7.779691219329834, 4.870410442352295] |
f1bb5c48-89f2-4d86-b1fe-0f3f8d56c277 | a-probabilistic-framework-for-dynamic-object | 2201.11608 | null | https://arxiv.org/abs/2201.11608v1 | https://arxiv.org/pdf/2201.11608v1.pdf | A Probabilistic Framework for Dynamic Object Recognition in 3D Environment With A Novel Continuous Ground Estimation Method | In this thesis a probabilistic framework is developed and proposed for Dynamic Object Recognition in 3D Environments. A software package is developed using C++ and Python in ROS that performs the detection and tracking task. Furthermore, a novel Gaussian Process Regression (GPR) based method is developed to detect ground points in different urban scenarios of regular, sloped and rough. The ground surface behavior is assumed to only demonstrate local input-dependent smoothness. kernel's length-scales are obtained. Bayesian inference is implemented sing \textit{Maximum a Posteriori} criterion. The log-marginal likelihood function is assumed to be a multi-task objective function, to represent a whole-frame unbiased view of the ground at each frame because adjacent segments may not have similar ground structure in an uneven scene while having shared hyper-parameter values. Simulation results shows the effectiveness of the proposed method in uneven and rough scenes which outperforms similar Gaussian process based ground segmentation methods. | ['Pouria Mehrabi'] | 2022-01-27 | null | null | null | null | ['gpr', 'gpr'] | ['computer-vision', 'miscellaneous'] | [ 1.05570853e-01 7.93961659e-02 5.24489939e-01 -3.94650698e-01
-6.97834909e-01 -2.07693353e-01 2.69241124e-01 3.25491935e-01
-2.17107907e-01 7.26355433e-01 -4.78633940e-01 -2.51862761e-02
-1.62794679e-01 -1.08735001e+00 -4.44142669e-01 -9.46289718e-01
-2.08205864e-01 8.04746747e-01 8.10711205e-01 1.47920310e-01
3.46738368e-01 7.28816092e-01 -1.71984231e+00 -3.31195980e-01
8.44921112e-01 3.93714696e-01 7.24094331e-01 1.06570327e+00
1.14099368e-01 1.82865128e-01 -3.27083051e-01 1.04270436e-01
2.60913163e-01 1.23515882e-01 -4.54971164e-01 3.20509851e-01
1.25542492e-01 -1.53044358e-01 3.80024239e-02 1.09120953e+00
6.20019138e-01 5.30118287e-01 1.23273396e+00 -8.70595813e-01
1.84773251e-01 -1.06258772e-01 -1.04243994e+00 2.85333425e-01
4.40057367e-01 -1.85553767e-02 4.10466731e-01 -8.21171880e-01
1.72804788e-01 1.56573057e+00 8.14363539e-01 -1.60100922e-01
-1.23583090e+00 -3.84653062e-01 2.84572225e-02 -2.47393832e-01
-1.71983266e+00 2.64462918e-01 3.47876072e-01 -6.82285249e-01
7.33506143e-01 2.04986691e-01 4.98473227e-01 4.51242447e-01
7.30629563e-01 4.57321823e-01 1.38231277e+00 -4.00474846e-01
4.30206150e-01 -8.52416903e-02 5.36545515e-01 5.59158266e-01
7.54929841e-01 -1.31221145e-01 -3.62529069e-01 -2.36107722e-01
8.91912282e-01 -3.48878175e-01 3.90781835e-02 -2.60749012e-01
-8.39275658e-01 4.62646246e-01 -1.17291152e-01 -4.24882310e-04
-7.34775960e-01 3.09526294e-01 -1.85458675e-01 -2.25016758e-01
4.23564732e-01 -5.49363434e-01 -1.97506547e-01 -4.04204875e-02
-9.96772945e-01 5.37333310e-01 8.16160142e-01 1.20879471e+00
9.04619813e-01 3.02673310e-01 -8.00162628e-02 4.62273747e-01
1.25954676e+00 1.19648528e+00 -5.56331985e-02 -8.20856571e-01
1.91001847e-01 2.81057417e-01 4.54150021e-01 -1.08243084e+00
-4.06964630e-01 -2.00509906e-01 -3.84348363e-01 6.33192420e-01
3.88234556e-01 -2.11432144e-01 -1.27390778e+00 8.09131324e-01
8.99684966e-01 3.88027251e-01 9.89367068e-02 4.08803046e-01
7.95983613e-01 8.27052236e-01 2.02173814e-01 8.40882733e-02
1.26485336e+00 -1.96814731e-01 -7.01204240e-01 -2.39705339e-01
6.75250143e-02 -9.03277695e-01 5.08430183e-01 3.70019197e-01
-9.53487456e-01 -6.58681273e-01 -1.03897035e+00 4.84188169e-01
-2.48430427e-02 1.82708815e-01 3.36981773e-01 9.83359754e-01
-9.49182212e-01 5.38507044e-01 -1.35437155e+00 -6.26460075e-01
3.03309351e-01 2.61474848e-01 3.07465307e-02 7.78088719e-02
-5.12306333e-01 7.05676615e-01 4.72995579e-01 3.72330040e-01
-1.05935454e+00 -2.21437410e-01 -6.48030519e-01 -3.88272822e-01
-1.53468549e-01 -6.81770384e-01 9.00814772e-01 -2.64275372e-01
-1.57827508e+00 1.02783394e+00 -3.17457408e-01 -2.79675722e-01
7.22378492e-01 -3.95611554e-01 -1.03271849e-01 1.48705021e-01
4.02138889e-01 3.25266458e-02 8.50987077e-01 -1.67440486e+00
-7.68840790e-01 -4.74225163e-01 -4.67287660e-01 5.50132334e-01
9.48217154e-01 1.30470723e-01 -2.99402565e-01 -2.61680484e-01
7.86908507e-01 -9.69016671e-01 -5.58117330e-01 -2.50458241e-01
-5.60138881e-01 9.20405891e-03 6.87964618e-01 -6.56850338e-01
7.36528754e-01 -1.99205446e+00 -5.42497218e-01 6.50285780e-01
-2.29096025e-01 -2.65102416e-01 7.05028594e-01 3.87447536e-01
3.21139783e-01 -2.81949397e-02 -5.30946314e-01 -3.05954605e-01
-1.39845371e-01 1.65305778e-01 1.54089168e-01 9.25196528e-01
-1.60062730e-01 1.00359116e-02 -7.76101172e-01 -8.28103960e-01
5.72654903e-01 5.87904751e-01 1.97868362e-01 -1.91474348e-04
1.33128658e-01 5.12178242e-01 -9.95787084e-01 6.32121205e-01
1.33920217e+00 2.97444254e-01 -2.56656826e-01 1.07117116e-01
-5.16470194e-01 -4.51022595e-01 -2.03884292e+00 1.21382940e+00
-1.08731970e-01 4.38113540e-01 2.84843147e-01 -5.53936541e-01
1.40979481e+00 2.63138831e-01 3.66665125e-01 6.73017576e-02
3.26796949e-01 5.38135432e-02 -3.41312170e-01 -6.92880750e-01
6.10643744e-01 -1.54040098e-01 4.00801629e-01 -6.44374192e-02
-3.02285820e-01 -7.19676077e-01 -1.51534930e-01 -2.20190734e-01
8.46725225e-01 7.90339410e-01 1.53616488e-01 -6.39375985e-01
4.24475998e-01 2.48642504e-01 5.59214532e-01 1.11886680e+00
-2.40266174e-01 8.66073787e-01 -2.08785892e-01 7.55692646e-02
-7.34444439e-01 -1.46036220e+00 -5.34040868e-01 4.99433011e-01
5.94614506e-01 3.14023197e-01 -5.47126472e-01 1.28447667e-01
-1.02598548e-01 9.52757835e-01 -3.66601616e-01 4.99202877e-01
-4.04287338e-01 -1.26307356e+00 3.44066262e-01 1.97249591e-01
7.06551373e-01 -6.04115844e-01 -9.12877202e-01 5.09217262e-01
1.65781870e-01 -1.06517756e+00 4.83389229e-01 1.09941095e-01
-1.22918749e+00 -9.45933223e-01 -6.37318730e-01 -5.41538119e-01
5.60178161e-01 3.49198639e-01 9.61926818e-01 -3.46023083e-01
-4.71522868e-01 8.35901022e-01 -2.52240449e-01 -9.32682157e-01
-1.86441839e-01 -5.55583477e-01 -3.34088951e-01 4.66445014e-02
2.21817896e-01 -4.77439106e-01 -6.02194011e-01 6.09038115e-01
-4.20830697e-01 -1.97212771e-01 9.47389528e-02 1.39344186e-01
9.66518223e-01 4.96470898e-01 -7.14026690e-02 -3.95233303e-01
2.69512206e-01 -5.64426422e-01 -7.88552701e-01 2.55354866e-02
-6.32969737e-02 -5.32943249e-01 -4.42233175e-01 -1.64289046e-02
-1.46100998e+00 1.72209918e-01 1.62087753e-01 -8.21868479e-02
-7.77005851e-01 3.92698318e-01 -2.34645694e-01 2.32680086e-02
6.84228420e-01 2.86961962e-02 -1.33377180e-01 -1.81680322e-01
-9.34555158e-02 3.86812061e-01 3.76394331e-01 -7.94487178e-01
9.78456199e-01 1.15195382e+00 4.82583672e-01 -1.56780684e+00
-4.08619851e-01 -9.33825493e-01 -7.78013527e-01 -7.35445142e-01
1.20756185e+00 -1.16598296e+00 -4.72053170e-01 7.48415887e-01
-1.03613353e+00 -4.60651696e-01 -8.18242803e-02 9.00679469e-01
-6.45519495e-01 4.65592533e-01 -1.99031070e-01 -1.76630139e+00
-1.26153648e-01 -8.24341953e-01 1.23119485e+00 6.68289125e-01
4.36694920e-02 -1.05020511e+00 1.88924745e-01 2.13617593e-01
-2.92472858e-02 7.77138114e-01 1.79672033e-01 -3.10656488e-01
-7.49857306e-01 -4.40474331e-01 -4.55703586e-03 1.88316628e-01
2.59888954e-02 5.78011930e-01 -1.08957279e+00 5.60988374e-02
2.79005736e-01 5.75524330e-01 3.02462578e-01 1.21058357e+00
3.53941798e-01 4.79292125e-01 -6.84197247e-01 2.57186443e-01
1.77910507e+00 3.29357773e-01 8.12068701e-01 5.76017141e-01
6.47216201e-01 8.12652886e-01 1.33703172e+00 5.24765491e-01
4.10591245e-01 3.75612974e-01 6.37179136e-01 5.58062568e-02
9.79264379e-02 2.83299029e-01 3.22250158e-01 1.05142249e-02
-5.77475131e-01 -4.28227633e-01 -1.16345489e+00 5.56967020e-01
-1.88741815e+00 -1.02798092e+00 -1.47095203e+00 2.52814960e+00
3.58950607e-02 1.90750614e-01 -1.27948061e-01 3.92201543e-02
1.18318105e+00 -2.83685625e-01 -3.03156134e-02 -1.70419887e-01
-2.19404131e-01 3.30406338e-01 1.08331835e+00 7.44664669e-01
-1.22394311e+00 7.61172354e-01 5.90385962e+00 7.82553673e-01
-6.61127806e-01 1.48150576e-02 7.51584098e-02 6.15511179e-01
1.04369201e-01 2.29566395e-01 -1.16631997e+00 3.78944427e-01
8.49660158e-01 -2.06358731e-02 -4.71903145e-01 8.94177437e-01
9.64554429e-01 -9.81068730e-01 -1.99969411e-01 8.10783088e-01
-3.90471846e-01 -6.52585030e-01 -3.96363109e-01 2.48199582e-01
8.96464109e-01 3.30293000e-01 -1.93315923e-01 -1.91867366e-01
8.14766467e-01 -7.61444569e-01 9.09952998e-01 9.55660343e-01
7.57682770e-02 -6.94715858e-01 9.34874356e-01 3.54811877e-01
-1.47633922e+00 4.26415265e-01 -4.72962946e-01 1.22589931e-01
5.81495404e-01 7.80237198e-01 -1.10864913e+00 9.04609799e-01
1.03590417e+00 4.13997471e-01 -3.14106226e-01 1.67678785e+00
-2.39581123e-01 8.38512421e-01 -9.97100592e-01 1.23211600e-01
1.50443152e-01 -7.04240620e-01 1.11408901e+00 1.36542070e+00
7.02701271e-01 -2.66668145e-02 4.50598329e-01 7.35604167e-01
1.03837514e+00 2.52329767e-01 -6.50491416e-01 7.44048595e-01
4.28850681e-01 1.07353234e+00 -1.33271956e+00 -6.39655590e-02
-3.12919915e-02 6.57857478e-01 -7.32577085e-01 3.43707174e-01
-8.53192091e-01 -5.49573041e-02 3.94554473e-02 5.44183016e-01
3.97594750e-01 -6.60895765e-01 -4.73702729e-01 -3.79517734e-01
-3.62574160e-01 -7.61552528e-02 2.34908059e-01 -1.08377588e+00
-9.66291785e-01 1.49637595e-01 6.51301682e-01 -1.18163145e+00
3.14375848e-01 -4.94819522e-01 -1.01696301e+00 1.12109101e+00
-1.18652594e+00 -1.47020555e+00 -6.67655110e-01 4.61170673e-01
7.17879176e-01 1.67358935e-01 4.46818560e-01 -6.38141111e-02
-4.79322493e-01 -3.93709064e-01 4.87545073e-01 -3.61459672e-01
2.12477744e-01 -1.47215652e+00 2.97522675e-02 1.08447361e+00
-4.91110563e-01 3.36580038e-01 1.41802788e+00 -1.26774299e+00
-1.06807578e+00 -1.16353440e+00 1.25990123e-01 -3.21298152e-01
5.49846947e-01 1.23052992e-01 -9.24332380e-01 5.95148981e-01
-2.91051805e-01 -1.47475421e-01 4.61460710e-01 -3.17754149e-01
6.87542081e-01 2.33614773e-01 -1.45794928e+00 3.30578595e-01
4.28571850e-01 9.02965367e-02 -5.58429182e-01 1.75887406e-01
-7.41281286e-02 -5.06422222e-01 -8.24236631e-01 6.20148778e-01
4.15769130e-01 -9.79481578e-01 1.04139376e+00 3.92586112e-01
-5.08149385e-01 -9.02157843e-01 -4.18628216e-01 -6.91174209e-01
-4.25684378e-02 -7.64965117e-01 3.25600445e-01 1.32950270e+00
3.10904145e-01 -6.37903690e-01 1.05023456e+00 4.98738170e-01
-4.38104570e-01 -1.22467570e-01 -1.16340470e+00 -8.85508299e-01
-1.98443323e-01 -7.50152409e-01 8.68756324e-02 4.28826809e-01
-1.02579105e+00 -8.98573548e-02 1.85153574e-01 1.15091038e+00
1.22450721e+00 -2.79675126e-01 1.13918316e+00 -1.66340744e+00
1.71345085e-01 1.45065874e-01 -7.06978023e-01 -7.65754879e-01
-2.18423650e-01 -3.46068144e-01 6.12613678e-01 -2.00520325e+00
-1.40013546e-01 -7.36730099e-01 2.59808332e-01 -2.95101702e-01
-1.01139843e-01 -2.63011068e-01 -4.82432306e-01 3.10247898e-01
6.35617897e-02 4.17544931e-01 9.64249909e-01 1.43499568e-01
-4.07510608e-01 5.99080324e-01 2.39502490e-01 1.22808838e+00
8.81709278e-01 -8.51155639e-01 -3.63181770e-01 -1.15413563e-02
1.52583838e-01 -7.24969432e-02 6.13691390e-01 -1.46098971e+00
7.18252510e-02 -3.22993279e-01 1.50718436e-01 -1.34900558e+00
5.14626265e-01 -9.79223251e-01 5.50534487e-01 2.35781074e-01
5.66776156e-01 -2.56136190e-02 2.52660334e-01 1.10555017e+00
1.22438021e-01 -7.48179495e-01 9.50008810e-01 -2.21417591e-01
-6.23055577e-01 2.11741805e-01 -7.88585126e-01 -3.94576371e-01
1.43856823e+00 -1.24508345e+00 2.43943244e-01 -2.32619509e-01
-1.00246000e+00 8.97822231e-02 5.29372752e-01 -3.56321990e-01
5.30254245e-01 -6.28164470e-01 -9.35345650e-01 -4.07731563e-01
-1.73057154e-01 5.10838270e-01 5.89134157e-01 8.78015339e-01
-1.15950775e+00 -3.13970685e-01 1.52871788e-01 -1.08960593e+00
-1.49455643e+00 -1.95755795e-01 4.80846494e-01 -7.72094280e-02
-7.39908397e-01 8.56738925e-01 1.31559893e-01 -1.71513975e-01
-1.28727227e-01 -4.51325715e-01 -4.07465726e-01 -5.81545308e-02
5.34278527e-02 8.76181066e-01 -1.01150043e-01 -9.66466427e-01
-1.36458293e-01 1.00289512e+00 4.99208242e-01 -4.09272462e-01
1.21690273e+00 -5.85061789e-01 1.47212908e-01 8.57602060e-01
2.14441538e-01 1.17122576e-01 -1.46744847e+00 1.91511139e-01
8.92572943e-03 -4.55934167e-01 2.23418474e-01 -1.75803572e-01
-2.40360811e-01 6.34098411e-01 1.09617317e+00 2.34426469e-01
4.04250205e-01 -2.20026985e-01 -1.36557251e-01 1.36789501e-01
5.22779882e-01 -1.30139875e+00 -3.36710095e-01 4.47860062e-01
7.26351023e-01 -1.05105412e+00 5.07604182e-01 -1.06511509e+00
-6.47913098e-01 1.19934666e+00 3.20067376e-01 -4.54831958e-01
1.10332859e+00 2.80936748e-01 -3.01038414e-01 -6.15900457e-01
1.17717661e-01 -5.48384905e-01 -7.24391192e-02 1.16598117e+00
1.49787977e-01 1.66082457e-01 -1.10030800e-01 5.96693717e-02
-2.14737952e-01 -2.75841564e-01 6.66328251e-01 1.39432704e+00
-9.71635401e-01 -4.69642937e-01 -1.24418044e+00 1.63169101e-01
-6.19473517e-01 2.26131424e-01 4.71090049e-01 1.08426094e+00
7.85686746e-02 1.22255039e+00 1.15270369e-01 3.37994277e-01
3.39441627e-01 -2.70494670e-01 4.43433225e-01 -8.49663973e-01
-1.52323529e-01 3.93863618e-01 1.84434980e-01 -1.87399819e-01
-6.02843285e-01 -1.25972879e+00 -1.64048243e+00 1.78368822e-01
-6.09149039e-01 -4.86473143e-02 1.16326976e+00 8.22844326e-01
-2.38754839e-01 3.64978373e-01 2.69550323e-01 -1.13878512e+00
9.52945873e-02 -1.00062621e+00 -9.99355733e-01 -2.61241943e-01
-2.97693193e-01 -1.02063930e+00 -4.03831035e-01 4.13451403e-01] | [7.391127109527588, -2.1387243270874023] |
2d372063-3776-4386-b8c8-2af210d2ea31 | diffusion-model-based-low-light-image | 2306.14227 | null | https://arxiv.org/abs/2306.14227v1 | https://arxiv.org/pdf/2306.14227v1.pdf | Diffusion Model Based Low-Light Image Enhancement for Space Satellite | Space-based visible camera is an important sensor for space situation awareness during proximity operations. However, visible camera can be easily affected by the low illumination in the space environment. Recently, deep learning approaches have achieved remarkable success in image enhancement of natural images datasets, but seldom applied in space due to the data bottleneck. In this article, we propose a data-driven method for low-light image enhancement (LLIE) of spin targets in space environment based on diffusion model. Firstly, a dataset collection scheme is devised. To reduce the domain gap and improve the diversity and quality of the dataset, we collect the data with the camera on a ground-test system imitating the low lighting conditions and relative attitude change of satellite in space. The satellite motion is controlled by a 6-DoF robot. To generate different poses, a advanced sampling method is combined with collision detection in physical simulation. The entire process is automated. Based on our dataset, a novel diffusion model is proposed. The diffusion and denoising process are directly conducted on the grayscale channel to save computational resources. To take advantage of the inner information of RGB channels, we rescale the RGB feature maps and insert them into the downsampling layers to help feature extraction. The enhanced results with our method have been verified to be better in image light enhancement and competitive in image quality compared with previous methods. To the best of our knowledge, this is the first work of LLIE using diffusion model. | ['Yu Guo', 'Jingyi Yuan', 'Lu Wang', 'Yiman Zhu'] | 2023-06-25 | null | null | null | null | ['image-enhancement', 'low-light-image-enhancement'] | ['computer-vision', 'computer-vision'] | [ 2.11165071e-01 -5.01200736e-01 2.59528965e-01 -5.66944927e-02
-1.59770682e-01 -3.00119281e-01 3.79754692e-01 -5.14330208e-01
-6.16385281e-01 7.21343040e-01 -1.15802921e-01 -1.52604848e-01
-1.46511048e-01 -7.54174829e-01 -4.62586075e-01 -1.20340395e+00
1.27431363e-01 -1.24928966e-01 2.74627894e-01 -5.28634012e-01
2.09302127e-01 8.22241306e-01 -1.71896172e+00 -1.77424386e-01
9.22456920e-01 9.50541735e-01 6.76670492e-01 4.98374730e-01
4.68954630e-02 6.55711412e-01 -3.77957344e-01 7.44464099e-02
5.47176301e-01 -4.38186675e-01 -3.20047766e-01 3.90538871e-01
6.16107695e-02 -4.54154909e-01 -5.71190298e-01 1.19898474e+00
8.17725658e-01 3.09754312e-01 2.71972358e-01 -1.04046535e+00
-4.02929008e-01 -8.94047990e-02 -7.72961855e-01 3.56482923e-01
2.56013181e-02 4.12558824e-01 2.43171409e-01 -7.70259082e-01
5.90091228e-01 8.97734046e-01 6.04506075e-01 2.41338640e-01
-8.06492329e-01 -3.26717287e-01 -3.34214777e-01 6.79311872e-01
-1.18527448e+00 -1.94616824e-01 1.19375539e+00 -7.35361949e-02
7.17948258e-01 3.94836307e-01 9.18866098e-01 7.37054467e-01
3.35753500e-01 2.94419616e-01 1.54043579e+00 -5.99387944e-01
2.79639453e-01 -6.07607886e-02 -2.25278974e-01 4.86372948e-01
2.32707217e-01 3.94347131e-01 -3.48302543e-01 2.68967450e-01
7.86447763e-01 2.44680762e-01 -4.89171982e-01 -1.78065717e-01
-1.16152692e+00 6.18179440e-01 7.63064206e-01 3.85156304e-01
-4.21703607e-01 -7.35608563e-02 -5.59556298e-02 3.80828589e-01
2.48773247e-01 1.80587217e-01 -2.24424750e-01 -5.97206764e-02
-6.71610355e-01 -1.54388592e-01 5.13927221e-01 4.83142525e-01
7.14612722e-01 2.68272877e-01 2.49492407e-01 6.16995156e-01
2.12239444e-01 9.31741536e-01 5.04316568e-01 -7.27373838e-01
2.42285952e-01 5.56677818e-01 1.81302205e-01 -1.19841146e+00
-6.22894585e-01 -6.36112034e-01 -9.23461497e-01 8.23123038e-01
8.12622979e-02 -1.38055414e-01 -7.94013321e-01 1.12947941e+00
6.33818328e-01 -2.99648810e-02 2.54049689e-01 1.45788217e+00
6.47100627e-01 8.73349845e-01 -4.59429204e-01 -4.21931565e-01
1.14487028e+00 -7.35460222e-01 -8.31404984e-01 -2.49193862e-01
5.53246379e-01 -8.14421773e-01 1.02506006e+00 4.59969789e-01
-7.21258342e-01 -6.49938941e-01 -1.27762938e+00 -8.47138539e-02
-2.20650449e-01 4.35874164e-01 6.76886082e-01 4.89984125e-01
-9.66197133e-01 5.45750678e-01 -8.81622851e-01 -4.13327187e-01
9.47777405e-02 1.34708121e-01 -2.80111939e-01 -1.79709010e-02
-1.17503607e+00 1.02093720e+00 1.80374205e-01 4.67085421e-01
-7.45271683e-01 -1.77133203e-01 -4.38768357e-01 -2.91061193e-01
3.81772548e-01 -5.06784499e-01 8.13819170e-01 -1.01297486e+00
-1.83832920e+00 5.97089767e-01 1.43011600e-01 -3.52003694e-01
4.00720239e-01 -5.71197905e-02 -4.80575025e-01 2.23741472e-01
-3.53432745e-01 1.88387141e-01 8.83508325e-01 -1.22415555e+00
-5.30105531e-01 -4.64892536e-01 1.09042645e-01 7.04179943e-01
-3.04533660e-01 5.01862653e-02 -2.15246931e-01 -3.42101723e-01
4.15332526e-01 -9.92878377e-01 -3.04200590e-01 3.35675091e-01
2.97854487e-02 4.17917460e-01 1.06631517e+00 -8.71127725e-01
7.73642719e-01 -2.36381197e+00 1.15771748e-01 -8.43204558e-02
-2.56715808e-03 2.93005168e-01 2.53677487e-01 2.47840703e-01
2.95579452e-02 -5.18195510e-01 -3.08588266e-01 -2.00257897e-01
-3.98491234e-01 2.26587906e-01 -1.19831003e-01 8.22750986e-01
-1.68539569e-01 3.97774816e-01 -5.06033480e-01 -6.16219044e-01
4.17675674e-01 5.56821525e-01 -2.28383318e-01 8.28688368e-02
1.24748483e-01 7.55648971e-01 -3.28137517e-01 6.94097161e-01
1.12832642e+00 2.84925073e-01 -2.80824661e-01 -6.17537796e-01
-6.20029509e-01 -1.61476970e-01 -1.42583072e+00 1.58375442e+00
-4.98163640e-01 5.78532755e-01 5.49248397e-01 -9.08932507e-01
1.20012295e+00 -1.51425421e-01 3.28738242e-01 -1.11728513e+00
3.24180812e-01 1.90101683e-01 -1.90813452e-01 -8.12055409e-01
6.40657008e-01 -4.46773201e-01 3.41016769e-01 2.51491606e-01
-3.94115955e-01 -4.65698808e-01 -1.87581345e-01 -5.23113906e-02
5.69942653e-01 3.01597267e-01 -2.30957642e-02 -1.10110873e-02
5.00208974e-01 2.21616283e-01 6.16101742e-01 4.27180946e-01
-7.14691579e-02 4.56372648e-01 -1.44034401e-01 -5.42586088e-01
-1.10128629e+00 -5.95679224e-01 -6.76676780e-02 5.28847516e-01
6.51186109e-01 1.67935789e-01 -5.33795476e-01 -2.13102445e-01
-4.02990699e-01 5.65453172e-01 -9.34895277e-02 -2.97310024e-01
-5.26379943e-01 -1.14451528e+00 8.75707418e-02 2.12821782e-01
1.09695160e+00 -8.28384817e-01 -9.93537009e-01 1.69172913e-01
-1.53100416e-01 -9.94466543e-01 1.34659946e-01 1.14060648e-01
-1.08609772e+00 -9.25851107e-01 -5.15218616e-01 -6.56431973e-01
5.17844796e-01 7.73975909e-01 5.41221261e-01 3.46580558e-02
-4.93138164e-01 1.26005456e-01 -5.16576707e-01 -2.01795414e-01
-4.89313602e-02 -4.53261584e-01 1.98412046e-01 2.78789327e-02
1.08261049e-01 -4.24055159e-01 -6.61844611e-01 3.16986173e-01
-1.00491548e+00 3.52928340e-01 6.91349506e-01 8.02182615e-01
4.74335968e-01 7.59721279e-01 -2.03255489e-01 7.66870305e-02
1.88187942e-01 -2.50965524e-02 -8.96038592e-01 -7.02163251e-03
-4.04580802e-01 -4.22322676e-02 4.78280544e-01 -4.27895665e-01
-1.48930085e+00 2.38773182e-01 5.24427788e-03 -1.45402759e-01
-6.39037266e-02 4.15768951e-01 -2.70450473e-01 -6.40747249e-01
4.77931887e-01 4.42241699e-01 2.52953261e-01 -4.84760374e-01
4.96668927e-02 9.68806207e-01 4.46509212e-01 -4.57922742e-02
9.84906018e-01 9.29297924e-01 2.67186433e-01 -1.22467840e+00
-4.45956409e-01 -1.64336607e-01 -5.00836611e-01 -5.00322104e-01
8.56795192e-01 -8.73489678e-01 -5.15801013e-01 7.96154380e-01
-9.13061202e-01 -2.29325265e-01 -1.17126018e-01 9.85791564e-01
-2.89764404e-01 7.02462196e-01 -6.18701637e-01 -8.62561345e-01
-4.23341364e-01 -1.14782393e+00 8.18169057e-01 5.91273248e-01
6.83369398e-01 -7.55992293e-01 -3.01084425e-02 4.05929118e-01
6.28257513e-01 2.57247597e-01 2.29490325e-01 3.27622473e-01
-7.74720132e-01 3.07533387e-02 -1.56632841e-01 3.96295279e-01
-9.37670562e-03 -1.81239005e-02 -8.24437201e-01 -3.84916842e-01
7.88794219e-01 -1.51475415e-01 7.66444683e-01 4.01767641e-01
9.75358307e-01 -2.33929008e-02 -2.38175601e-01 1.07723153e+00
1.73434305e+00 3.61379623e-01 8.49284232e-01 9.81882632e-01
6.44459128e-01 4.56558704e-01 1.02629185e+00 5.88336051e-01
-1.34783881e-02 7.22280502e-01 7.72414625e-01 -6.01984799e-01
-1.54035062e-01 2.65657574e-01 5.40862799e-01 7.27841854e-01
-3.77957791e-01 -6.01792447e-02 -7.14684963e-01 2.29626834e-01
-1.30792320e+00 -8.90000522e-01 -5.53599060e-01 2.16434360e+00
5.73285878e-01 4.84346040e-03 -3.38607639e-01 2.64609128e-01
7.14551508e-01 1.55419707e-01 -3.36840987e-01 -3.73445936e-02
-5.52891433e-01 -1.67553142e-01 8.16404879e-01 5.16836226e-01
-8.78454089e-01 6.63309038e-01 5.29057932e+00 6.93207741e-01
-1.60796535e+00 1.40490770e-01 2.41766676e-01 -7.95979127e-02
-1.10294215e-01 2.35110268e-01 -7.01948941e-01 6.55374765e-01
4.56390172e-01 2.48013571e-01 6.21341884e-01 5.98307967e-01
3.40456247e-01 -5.09497583e-01 -2.16103688e-01 1.27433169e+00
2.47646064e-01 -9.04572248e-01 -4.22190458e-01 8.24975520e-02
6.03477418e-01 1.67908385e-01 8.99428800e-02 -2.39404455e-01
-2.35077724e-01 -5.25230348e-01 6.08396351e-01 8.79221976e-01
6.96448147e-01 -8.08840096e-01 8.94883990e-01 5.10598779e-01
-9.07152891e-01 -2.40488902e-01 -6.38395309e-01 -3.91490370e-01
4.11773622e-01 8.93813670e-01 -5.51004529e-01 7.78758287e-01
8.58972788e-01 5.30145168e-01 -6.05673432e-01 1.13230205e+00
-1.48310989e-01 4.69001353e-01 -6.04241848e-01 -5.70824780e-02
1.96192950e-01 -7.39072561e-01 7.16875911e-01 7.44854569e-01
7.31797576e-01 1.96090534e-01 -1.30110458e-01 7.23693907e-01
4.97271419e-01 -2.02438340e-01 -6.30984187e-01 1.50139332e-01
5.21948934e-03 1.55388272e+00 -6.13608301e-01 -1.94321290e-01
-3.65352869e-01 1.09618473e+00 -2.57656008e-01 3.30142945e-01
-1.09035742e+00 -5.26080549e-01 3.24746929e-02 6.20066002e-02
1.49920553e-01 -6.44111574e-01 -2.70788252e-01 -1.07048929e+00
7.51881972e-02 -6.29706383e-01 2.67066434e-02 -1.35400140e+00
-8.74119103e-01 6.58821583e-01 -1.25476509e-01 -1.63069093e+00
4.53980118e-01 -6.90328956e-01 -4.65381444e-01 8.26285720e-01
-1.55818212e+00 -1.02798390e+00 -8.43970895e-01 5.33786118e-01
2.78255254e-01 -1.62608981e-01 3.96419823e-01 3.09882760e-01
-5.16806424e-01 -1.85572997e-01 5.92548668e-01 -4.96810585e-01
6.12178087e-01 -7.34654367e-01 -9.35278162e-02 1.14659739e+00
-3.58022928e-01 2.09261477e-01 8.36867630e-01 -7.08737910e-01
-1.82225347e+00 -6.98418498e-01 4.56725806e-01 4.90165912e-02
6.21217847e-01 -1.28555939e-01 -7.94977665e-01 2.75805533e-01
1.79599747e-01 4.60933484e-02 1.28687546e-01 -5.54422200e-01
5.19416034e-01 -4.15286571e-01 -1.26164114e+00 4.38443303e-01
9.00393367e-01 -3.57529461e-01 -5.89785337e-01 3.97251219e-01
5.26819766e-01 -5.98745704e-01 -5.43754280e-01 3.92098546e-01
1.77809700e-01 -9.99019325e-01 9.68843997e-01 1.08225308e-01
2.12064803e-01 -8.68003368e-01 -5.24118021e-02 -1.26451766e+00
-4.40532982e-01 -4.96376097e-01 2.34348610e-01 1.13858414e+00
-1.13952979e-01 -6.14739656e-01 5.64050734e-01 2.15878934e-01
-3.13290864e-01 -2.79455990e-01 -9.42918956e-01 -7.35140800e-01
-5.90047598e-01 -1.77341074e-01 5.29893219e-01 8.40699553e-01
-4.42719966e-01 2.09087282e-01 -4.74689186e-01 6.39180005e-01
8.41172636e-01 2.95993298e-01 7.44372368e-01 -1.10014307e+00
-3.33178401e-01 1.14810914e-01 -3.86970252e-01 -7.69061685e-01
-3.42054814e-01 -5.13999045e-01 -8.82509202e-02 -1.58327341e+00
1.47063375e-01 -4.88063425e-01 1.08793512e-01 3.90524715e-02
-6.48589060e-02 2.92024821e-01 6.22917302e-02 3.69234651e-01
-2.69444555e-01 8.90949845e-01 1.46430588e+00 -4.70980220e-02
-1.89075112e-01 -1.89823806e-01 -1.42240748e-01 7.14268208e-01
1.05225575e+00 -3.08438480e-01 -1.63246199e-01 -7.53110886e-01
3.82192045e-01 1.56447053e-01 5.61690748e-01 -1.43884611e+00
3.91353965e-01 -2.50979722e-01 4.92033005e-01 -7.80457795e-01
5.48406720e-01 -1.16705179e+00 3.87426615e-01 7.33217776e-01
3.32047731e-01 -7.46936426e-02 7.34136403e-02 3.91036242e-01
-2.94498593e-01 -2.74112731e-01 9.49322939e-01 -6.73727617e-02
-1.08976340e+00 1.97515115e-02 -2.43888035e-01 -6.05302870e-01
1.06795931e+00 -4.50146943e-01 -3.26314688e-01 -2.84437627e-01
-4.43938047e-01 -1.57662868e-01 8.24802876e-01 -1.22144490e-01
6.56133592e-01 -1.28137708e+00 -5.06167948e-01 3.95020843e-01
-6.62169158e-02 1.05322137e-01 6.75085783e-01 1.22319543e+00
-1.10754824e+00 -9.27925389e-03 -4.67668682e-01 -5.03487110e-01
-1.27094150e+00 6.98245525e-01 4.54212964e-01 1.74879268e-01
-7.58973777e-01 6.18253112e-01 -1.62300706e-01 -2.74799585e-01
-9.91762504e-02 -2.70541042e-01 -3.07050794e-01 -2.02651471e-01
6.67658627e-01 5.55626750e-01 1.17148146e-01 -6.95730984e-01
-2.26173505e-01 8.91740024e-01 3.97818297e-01 -2.27125496e-01
1.59930074e+00 -5.27934551e-01 -2.10984722e-01 1.59212396e-01
9.42082822e-01 1.17471613e-01 -1.58848214e+00 -3.81575897e-02
-5.54764450e-01 -6.75597727e-01 6.71739221e-01 -6.13617778e-01
-1.17598581e+00 9.37506855e-01 1.15395212e+00 -2.78237499e-02
1.59762716e+00 -3.13833177e-01 7.68998325e-01 3.77053261e-01
4.42739964e-01 -1.40535975e+00 -1.22837581e-01 4.48308647e-01
8.69617522e-01 -1.30826378e+00 2.18079016e-01 -1.88582718e-01
-8.16355765e-01 1.23726022e+00 4.99207795e-01 9.86297876e-02
1.00424260e-01 4.12903994e-01 8.67283493e-02 -1.21173397e-01
-1.92835242e-01 -3.52833241e-01 -2.08434612e-01 6.49662912e-01
-8.66665468e-02 -2.10181028e-01 -4.13178444e-01 1.67837083e-01
-6.95900107e-03 1.67082116e-01 7.03925490e-01 1.23628771e+00
-7.87858427e-01 -7.83641815e-01 -1.01359296e+00 -1.38385803e-01
-1.39319628e-01 3.27731408e-02 -1.86764792e-01 5.73053002e-01
2.95840651e-01 9.05811369e-01 -7.12134242e-02 -4.56070721e-01
2.27329642e-01 -4.29856002e-01 4.71570343e-01 1.61027133e-01
-3.06773912e-02 5.17900288e-02 2.17539109e-02 -4.95049298e-01
-6.60647571e-01 -5.01606882e-01 -1.36437404e+00 -2.46129423e-01
-4.51185971e-01 1.12519324e-01 1.16516411e+00 6.53126597e-01
2.23829627e-01 4.86952513e-01 9.78416681e-01 -1.04703629e+00
-4.83555764e-01 -8.84813070e-01 -8.05371404e-01 2.94227719e-01
3.39346081e-01 -8.72012019e-01 -8.22129965e-01 -7.65757039e-02] | [10.575260162353516, -2.4342379570007324] |
bdb68339-d125-446d-87ff-1190ae8fe68b | master-multi-task-pre-trained-bottlenecked | 2212.07841 | null | https://arxiv.org/abs/2212.07841v2 | https://arxiv.org/pdf/2212.07841v2.pdf | MASTER: Multi-task Pre-trained Bottlenecked Masked Autoencoders are Better Dense Retrievers | Pre-trained Transformers (\eg BERT) have been commonly used in existing dense retrieval methods for parameter initialization, and recent studies are exploring more effective pre-training tasks for further improving the quality of dense vectors. Although various novel and effective tasks have been proposed, their different input formats and learning objectives make them hard to be integrated for jointly improving the model performance. In this work, we aim to unify a variety of pre-training tasks into the bottlenecked masked autoencoder manner, and integrate them into a multi-task pre-trained model, namely MASTER. Concretely, MASTER utilizes a shared-encoder multi-decoder architecture that can construct a representation bottleneck to compress the abundant semantic information across tasks into dense vectors. Based on it, we integrate three types of representative pre-training tasks: corrupted passages recovering, related passages recovering and PLMs outputs recovering, to characterize the inner-passage information, inter-passage relations and PLMs knowledge. Extensive experiments have shown that our approach outperforms competitive dense retrieval methods. Our code and data are publicly released in \url{https://github.com/microsoft/SimXNS}. | ['Ji-Rong Wen', 'Nan Duan', 'Daxin Jiang', 'Wayne Xin Zhao', 'Yeyun Gong', 'Xiao Liu', 'Kun Zhou'] | 2022-12-15 | null | null | null | null | ['passage-retrieval'] | ['natural-language-processing'] | [-1.02396213e-01 -4.32722420e-01 -1.51459754e-01 -2.14697614e-01
-1.26514030e+00 -2.71817237e-01 6.58623457e-01 3.52847837e-02
-4.87941891e-01 6.07582510e-01 7.06971765e-01 9.25252810e-02
-1.43434823e-01 -6.58631027e-01 -7.82858551e-01 -5.77362180e-01
1.83455214e-01 7.00270712e-01 1.52194083e-01 -4.04365510e-01
1.24892712e-01 -2.80819349e-02 -1.45652401e+00 5.50174832e-01
8.67220402e-01 6.59150720e-01 5.34044385e-01 5.85830569e-01
-1.60203308e-01 7.95043170e-01 -7.82227755e-01 -5.03329337e-01
-5.95270507e-02 -4.22457457e-01 -9.26446795e-01 -2.62322187e-01
-3.60319391e-02 -6.75617158e-01 -8.44313204e-01 8.15057755e-01
8.74376655e-01 4.90252703e-01 7.61502266e-01 -7.64744163e-01
-1.10969174e+00 9.99988258e-01 -2.89507419e-01 3.27250183e-01
3.89864296e-01 -1.33647010e-01 1.28748500e+00 -1.19223630e+00
5.73648632e-01 1.17024291e+00 4.66200739e-01 3.99263382e-01
-7.18688548e-01 -6.63114905e-01 7.11087063e-02 5.71359932e-01
-1.79994535e+00 -6.45771742e-01 5.33615768e-01 -5.79308495e-02
1.22259498e+00 1.11806296e-01 3.43064338e-01 1.50003409e+00
-2.90517598e-01 1.36378062e+00 5.12089729e-01 -4.78557348e-01
-2.51912475e-01 1.54652730e-01 3.22293967e-01 4.95301932e-01
8.24302658e-02 -6.70204014e-02 -4.82344210e-01 -5.03115691e-02
6.16666675e-01 2.12154865e-01 -6.06062472e-01 2.68464863e-01
-1.17782104e+00 7.61865795e-01 4.95065510e-01 5.38059831e-01
-2.84494609e-01 -6.25195056e-02 4.64880496e-01 4.16207016e-01
5.46354711e-01 4.17087823e-01 -3.36222649e-01 -1.50442034e-01
-9.75942492e-01 2.13059098e-01 7.56848752e-01 1.28450787e+00
7.76113153e-01 -1.43007845e-01 -6.17923021e-01 1.37379861e+00
4.51793075e-01 4.39726174e-01 6.79636180e-01 -5.33808053e-01
7.70003676e-01 4.51370388e-01 -4.06628102e-02 -8.60851943e-01
-1.42774954e-01 -7.94672310e-01 -1.06087351e+00 -8.43882024e-01
-8.45911056e-02 -3.74953938e-03 -7.91325569e-01 1.54114401e+00
-5.29845022e-02 4.05633599e-01 1.60416424e-01 1.14869726e+00
1.08812022e+00 1.13187993e+00 3.48376967e-02 1.84197485e-01
1.30622184e+00 -1.55602193e+00 -8.82226050e-01 -1.12205304e-01
6.60429358e-01 -1.02617645e+00 1.21991575e+00 1.44878864e-01
-1.28407252e+00 -7.57812321e-01 -1.05792511e+00 -6.55859947e-01
-4.57478285e-01 4.02220428e-01 3.56615543e-01 -9.49232206e-02
-1.04272497e+00 5.41206956e-01 -7.57285714e-01 -1.60287499e-01
3.22162747e-01 2.97074411e-02 -1.84441924e-01 -5.59546471e-01
-1.70546269e+00 9.83497977e-01 5.00255466e-01 3.36148947e-01
-1.05153096e+00 -7.35276759e-01 -8.38940203e-01 3.52895975e-01
2.28439674e-01 -1.09415817e+00 1.27726829e+00 -4.11861449e-01
-1.39520884e+00 7.04119265e-01 -2.46913016e-01 -2.26646304e-01
4.35608141e-02 -6.50723815e-01 -3.87538821e-01 2.01713368e-01
1.79663859e-02 5.79205692e-01 7.12457299e-01 -1.21048141e+00
-2.35313877e-01 -1.45848140e-01 1.59497093e-02 5.28268039e-01
-5.67324698e-01 2.35531956e-01 -1.29860866e+00 -8.86217356e-01
-3.61998916e-01 -5.66764355e-01 -1.75577313e-01 -6.04263723e-01
-6.28742814e-01 -3.60461175e-01 3.11674803e-01 -8.04413915e-01
1.59155178e+00 -2.10281897e+00 5.01271248e-01 -3.12426109e-02
5.58810756e-02 4.92997587e-01 -6.94565117e-01 8.84039104e-01
3.73902649e-01 1.59591995e-02 -1.41246259e-01 -9.06057358e-01
2.25692227e-01 2.57624745e-01 -4.84575778e-01 1.77825123e-01
1.42721623e-01 1.12804699e+00 -8.21316123e-01 -5.47676802e-01
5.41662984e-02 1.01880515e+00 -5.19223332e-01 7.09905684e-01
-1.59215987e-01 4.11445409e-01 -6.43021286e-01 6.66126490e-01
5.42754352e-01 -6.21539772e-01 -2.47590408e-01 -1.95540234e-01
2.88238764e-01 7.28542626e-01 -9.87694085e-01 2.25428462e+00
-5.39123654e-01 4.97174770e-01 -7.95475692e-02 -1.01105249e+00
6.88349485e-01 3.85935128e-01 2.39543676e-01 -9.67830539e-01
2.62883455e-01 1.20931275e-01 -4.28631961e-01 -5.56905508e-01
9.75070179e-01 1.00288197e-01 -2.69345455e-02 6.59235299e-01
5.38511634e-01 1.82712704e-01 3.47155303e-01 5.10875702e-01
1.03456855e+00 -2.62045185e-03 3.39743048e-02 1.85489416e-01
5.09921908e-01 -1.92936257e-01 1.97516426e-01 8.83655667e-01
2.61360228e-01 1.04333413e+00 1.21496059e-02 8.43886212e-02
-1.00434709e+00 -9.26167488e-01 -1.33302897e-01 1.33340669e+00
3.35375726e-01 -7.26340055e-01 -5.08938491e-01 -2.80424297e-01
-1.44005328e-01 4.34055179e-01 -4.33406711e-01 -5.14168561e-01
-6.36962771e-01 -7.07133412e-01 8.19854319e-01 5.89475989e-01
3.88383895e-01 -1.17905962e+00 2.33218580e-01 2.53068715e-01
-6.17673576e-01 -1.19981468e+00 -4.25151646e-01 5.82732819e-02
-6.44637883e-01 -9.01616871e-01 -1.09479606e+00 -8.30767393e-01
4.90230590e-01 5.62991023e-01 1.52776492e+00 5.58128953e-01
-2.21286997e-01 3.87175322e-01 -9.07589257e-01 -7.35894591e-02
-5.41255213e-02 5.75690687e-01 -2.35745698e-01 -2.67377973e-01
4.73568141e-01 -5.15079498e-01 -7.52765715e-01 2.42215663e-01
-1.09879613e+00 2.95143779e-02 9.21684742e-01 9.81637776e-01
7.32122064e-01 -1.46809131e-01 5.11374652e-01 -8.53076875e-01
9.95480001e-01 -9.06377435e-01 -2.44241022e-02 4.33441460e-01
-4.10021931e-01 9.16877389e-02 5.38642168e-01 -3.33016664e-01
-1.02903140e+00 -6.37667358e-01 -5.30105531e-01 -6.71514809e-01
6.77151084e-02 7.55949974e-01 -1.48237601e-01 3.50670248e-01
3.45650464e-01 5.09613454e-01 -2.25238666e-01 -9.48952198e-01
6.64186299e-01 8.12193930e-01 2.68820167e-01 -7.55723417e-01
7.88482070e-01 4.24656309e-02 -7.69275427e-01 -5.21505833e-01
-1.13883483e+00 -9.00655746e-01 -5.15934348e-01 1.72413543e-01
5.01208007e-01 -1.40915418e+00 -1.59143075e-01 3.57994527e-01
-1.26730883e+00 -3.80702078e-01 -1.94989204e-01 4.97469246e-01
-1.77974656e-01 3.50998640e-01 -9.83345449e-01 -2.42934883e-01
-7.52553642e-01 -1.19120955e+00 1.31843269e+00 1.96151301e-01
5.62199242e-02 -1.07708561e+00 3.87318283e-01 6.14162862e-01
7.94257402e-01 -6.21942639e-01 6.77585185e-01 -8.45702708e-01
-7.11777031e-01 -2.47357190e-01 -3.66119087e-01 5.54006994e-01
-2.58245915e-01 -1.80870205e-01 -9.69749212e-01 -4.61453557e-01
-1.64697185e-01 -6.99482918e-01 1.29796529e+00 1.40718386e-01
1.39297390e+00 -3.32592547e-01 -4.82459694e-01 7.16969490e-01
1.32597756e+00 -2.59863377e-01 7.76842475e-01 2.60273874e-01
6.94020092e-01 2.93654293e-01 4.25674558e-01 4.28932309e-01
6.60896182e-01 6.30866885e-01 2.98433900e-02 -7.69663602e-02
-4.78530228e-01 -3.64866912e-01 3.46419394e-01 1.78643107e+00
-2.84732487e-02 -5.53822160e-01 -6.06948674e-01 5.16685247e-01
-1.81177330e+00 -9.07729685e-01 2.47873291e-01 1.77282274e+00
1.27432895e+00 -3.28561068e-01 -2.45864242e-01 -1.53725818e-01
4.87412542e-01 3.63020003e-01 -3.21828455e-01 1.49165466e-01
-1.49476364e-01 4.92466092e-01 1.68415383e-02 5.90477705e-01
-1.00468659e+00 1.07026315e+00 5.72927046e+00 1.00087357e+00
-7.19630420e-01 3.42993408e-01 2.01885596e-01 -3.05333793e-01
-6.86889648e-01 -1.71700701e-01 -1.19183457e+00 4.46994513e-01
9.91588295e-01 -1.49923429e-01 2.77150631e-01 5.59362829e-01
-1.68956473e-01 2.44614035e-01 -1.04508805e+00 9.69189465e-01
4.67743576e-01 -1.22031522e+00 4.42949802e-01 -4.36307549e-01
6.03677094e-01 3.47552896e-01 1.13268696e-01 1.08643377e+00
2.06212521e-01 -1.04957855e+00 4.24520522e-01 7.38666534e-01
5.48727393e-01 -4.31610316e-01 9.47660208e-01 3.44458789e-01
-1.14602256e+00 1.29342631e-01 -7.33181953e-01 2.10529208e-01
3.26886833e-01 6.90983057e-01 -5.00909686e-01 8.35019290e-01
6.51518583e-01 1.01728344e+00 -5.84649384e-01 1.26093853e+00
-4.27114874e-01 6.09308898e-01 -1.46636575e-01 1.08345561e-02
2.13585064e-01 -1.65055409e-01 3.71002734e-01 1.66085732e+00
2.41287231e-01 -3.35030910e-03 -4.43202667e-02 9.01000023e-01
-4.51915264e-01 1.60574853e-01 -3.59487444e-01 -1.52962819e-01
6.86592996e-01 1.10373211e+00 -2.61704624e-02 -4.32761997e-01
-6.46790087e-01 1.07740378e+00 5.61238050e-01 7.88107812e-01
-7.94055462e-01 -5.92124999e-01 7.77700424e-01 -1.32429257e-01
2.98104256e-01 -2.21728086e-01 1.73864931e-01 -1.68350446e+00
5.57614677e-02 -1.10646033e+00 5.10536611e-01 -7.48735428e-01
-1.53850687e+00 7.21450329e-01 -2.05228422e-02 -1.19765830e+00
-2.82347590e-01 -2.32169300e-01 -3.51667374e-01 9.88226473e-01
-2.12787199e+00 -1.13415742e+00 -2.81794220e-01 9.15782750e-01
8.88935626e-01 -2.20217213e-01 1.08630872e+00 9.61268842e-01
-8.51521254e-01 8.98448348e-01 3.13115001e-01 3.80323291e-01
9.34242845e-01 -9.22581255e-01 1.16513759e-01 8.05861890e-01
4.99010533e-01 9.56407249e-01 1.08644769e-01 -2.50215352e-01
-1.63903856e+00 -1.10835683e+00 8.12687993e-01 -3.19421530e-01
6.19608343e-01 -3.87404025e-01 -1.26995504e+00 7.78845906e-01
6.02329135e-01 -9.92625579e-02 9.06648159e-01 4.11724538e-01
-4.52392966e-01 5.61439916e-02 -3.60978216e-01 5.29959798e-01
9.45246577e-01 -8.97266686e-01 -7.80259430e-01 4.90403533e-01
9.53400314e-01 -5.07935584e-01 -1.02518940e+00 3.87455672e-01
1.75274625e-01 -5.99758029e-01 1.30858171e+00 -6.10338748e-01
6.25592589e-01 -8.53402317e-02 -1.21941574e-01 -1.42361903e+00
-5.06760895e-01 -3.92075241e-01 -5.41855395e-01 1.34983122e+00
4.74817544e-01 -2.82956809e-01 3.24783534e-01 1.72592521e-01
-4.89610344e-01 -9.80384290e-01 -5.07322371e-01 -4.77168322e-01
1.80773854e-01 -3.81200433e-01 5.14414251e-01 8.41167808e-01
-9.18741822e-02 7.30469167e-01 -5.26218534e-01 1.62533656e-01
3.44176948e-01 1.42062768e-01 7.59117424e-01 -8.03675711e-01
-4.34957415e-01 -5.35021961e-01 1.31258622e-01 -1.77123678e+00
3.06028515e-01 -1.18919253e+00 6.93077967e-02 -1.81887865e+00
5.62770963e-01 -6.13513649e-01 -5.48659623e-01 5.51231444e-01
-5.80565870e-01 1.41379118e-01 3.01631726e-02 5.99642277e-01
-1.11982405e+00 1.12621462e+00 1.39854944e+00 -2.73655236e-01
7.97330495e-03 -1.42185152e-01 -9.26617801e-01 1.01711422e-01
6.80475712e-01 -4.86542642e-01 -5.66111028e-01 -1.25030804e+00
2.82999337e-01 -7.09676230e-03 1.25926644e-01 -8.12707424e-01
4.93334532e-01 3.99079680e-01 3.00642848e-01 -6.77181005e-01
5.82613707e-01 -5.46559513e-01 -2.16094211e-01 -1.93532318e-01
-5.73162019e-01 -9.81565192e-02 2.72593588e-01 4.80920672e-01
-8.01454723e-01 -5.01812756e-01 3.20805103e-01 -2.17634454e-01
-6.79090381e-01 6.31984532e-01 -1.51737168e-01 3.33564132e-01
4.42404151e-01 3.16395193e-01 -5.43281078e-01 -4.59098846e-01
-5.08457422e-01 4.57127571e-01 8.53334740e-02 6.40254021e-01
8.81620049e-01 -1.31771696e+00 -9.70885277e-01 7.79329017e-02
1.82769001e-01 2.60428727e-01 5.15092731e-01 7.93190420e-01
-2.25039661e-01 7.01237261e-01 1.17101081e-01 -3.55379224e-01
-9.09744084e-01 4.37208384e-01 1.57022566e-01 -6.24868155e-01
-8.65305662e-01 1.07834554e+00 -2.16588769e-02 -4.23867166e-01
5.57576180e-01 -1.09700292e-01 -3.84447694e-01 1.26968861e-01
8.85815144e-01 5.01650386e-02 1.15487047e-01 -3.86692286e-01
-1.84201915e-02 4.94630277e-01 -5.49938679e-01 1.86064728e-02
1.56542468e+00 -3.93958807e-01 -2.03285143e-01 2.91514635e-01
1.77753723e+00 -2.42400855e-01 -7.84802020e-01 -7.94870913e-01
-1.95186242e-01 -2.95717835e-01 1.45888805e-01 -6.59640491e-01
-1.13194573e+00 1.09801018e+00 3.51167247e-02 -1.00123711e-01
1.22789323e+00 2.07216874e-01 1.25030625e+00 6.73779666e-01
1.67371824e-01 -8.30132604e-01 2.75748283e-01 8.57641757e-01
1.01501906e+00 -1.20596290e+00 -1.02253919e-02 -1.56770691e-01
-6.65515542e-01 8.45825672e-01 5.18282473e-01 -1.92468762e-01
7.41258323e-01 2.93133259e-01 -1.64358560e-02 -2.57303715e-01
-9.95989740e-01 -4.51693475e-01 5.46109438e-01 3.28034252e-01
5.53948283e-01 -2.83579648e-01 -1.37489811e-01 9.01940286e-01
-1.02270938e-01 -1.41553842e-02 4.92639020e-02 7.55752087e-01
-2.95937210e-01 -1.23740339e+00 5.54753020e-02 3.52455437e-01
-4.53346342e-01 -5.55671275e-01 -9.56039801e-02 6.72351420e-01
-2.62811631e-01 7.33194590e-01 1.69533920e-02 -4.05963093e-01
3.45677376e-01 1.15641646e-01 3.65364105e-01 -7.48000264e-01
-6.35592520e-01 9.00611747e-03 1.07163757e-01 -5.16870499e-01
-3.82521987e-01 -1.97311953e-01 -9.55056787e-01 -1.86631471e-01
-4.86406863e-01 4.06607896e-01 3.70378822e-01 9.39692020e-01
5.68192124e-01 6.91875100e-01 4.90112990e-01 -7.11953342e-01
-5.89990556e-01 -1.44503725e+00 -2.57204831e-01 5.30057609e-01
1.70150980e-01 -5.95814526e-01 -3.77882928e-01 -4.62808982e-02] | [11.392664909362793, 7.705741882324219] |
1fe30a37-22a5-46ba-875d-ae802744deca | leandojo-theorem-proving-with-retrieval | 2306.15626 | null | https://arxiv.org/abs/2306.15626v1 | https://arxiv.org/pdf/2306.15626v1.pdf | LeanDojo: Theorem Proving with Retrieval-Augmented Language Models | Large language models (LLMs) have shown promise in proving formal theorems using proof assistants such as Lean. However, existing methods are difficult to reproduce or build on, due to private code, data, and large compute requirements. This has created substantial barriers to research on machine learning methods for theorem proving. This paper removes these barriers by introducing LeanDojo: an open-source Lean playground consisting of toolkits, data, models, and benchmarks. LeanDojo extracts data from Lean and enables interaction with the proof environment programmatically. It contains fine-grained annotations of premises in proofs, providing valuable data for premise selection: a key bottleneck in theorem proving. Using this data, we develop ReProver (Retrieval-Augmented Prover): the first LLM-based prover that is augmented with retrieval for selecting premises from a vast math library. It is inexpensive and needs only one GPU week of training. Our retriever leverages LeanDojo's program analysis capability to identify accessible premises and hard negative examples, which makes retrieval much more effective. Furthermore, we construct a new benchmark consisting of 96,962 theorems and proofs extracted from Lean's math library. It features challenging data split requiring the prover to generalize to theorems relying on novel premises that are never used in training. We use this benchmark for training and evaluation, and experimental results demonstrate the effectiveness of ReProver over non-retrieval baselines and GPT-4. We thus provide the first set of open-source LLM-based theorem provers without any proprietary datasets and release it under a permissive MIT license to facilitate further research. | ['Anima Anandkumar', 'Ryan Prenger', 'Saad Godil', 'Shixing Yu', 'Peiyang Song', 'Rahul Chalamala', 'Alex Gu', 'Aidan M. Swope', 'Kaiyu Yang'] | 2023-06-27 | null | null | null | null | ['retrieval', 'automated-theorem-proving', 'automated-theorem-proving'] | ['methodology', 'miscellaneous', 'reasoning'] | [-1.16045117e-01 2.89202537e-02 -5.29852867e-01 -8.39410126e-02
-1.23951030e+00 -1.09520733e+00 5.17725587e-01 3.46607745e-01
6.95791468e-02 7.72678256e-01 -2.55226225e-01 -1.43765092e+00
-2.81568229e-01 -9.51837540e-01 -1.31308997e+00 7.25355148e-02
-3.77363145e-01 4.23162013e-01 2.28195325e-01 -5.81749231e-02
2.64635861e-01 7.30358064e-02 -1.49585998e+00 6.73737109e-01
9.62445080e-01 2.95296669e-01 -1.37919905e-02 1.24835849e+00
3.61732244e-02 1.30471194e+00 -3.45063001e-01 -7.35837400e-01
2.45390102e-01 -2.28351671e-02 -1.17614508e+00 -1.01860178e+00
8.69218051e-01 -8.02843988e-01 -2.18694553e-01 9.46015179e-01
1.64419591e-01 -2.69273400e-01 2.27528989e-01 -1.92810035e+00
-3.24179858e-01 1.23596978e+00 -9.17770714e-03 9.94220078e-02
8.02861631e-01 3.57321471e-01 1.43515491e+00 -5.37923276e-01
8.25691044e-01 1.23054051e+00 9.18613970e-01 6.21565104e-01
-1.46981704e+00 -8.09856534e-01 -4.36163068e-01 3.91097993e-01
-1.28697586e+00 -7.13345587e-01 5.57331800e-01 -4.05714333e-01
1.37221444e+00 7.60687709e-01 3.67654830e-01 9.47498858e-01
1.59190491e-01 9.30261493e-01 7.92846680e-01 -6.16444230e-01
4.85763848e-02 2.83638835e-01 3.19144398e-01 1.32529080e+00
2.72672743e-01 -3.02687157e-02 -3.11821043e-01 -6.44760609e-01
1.71640068e-01 -2.07684785e-01 -1.24262467e-01 -5.81310511e-01
-1.40340769e+00 7.24104285e-01 -2.52259579e-02 -1.04758181e-01
3.61943185e-01 4.92627591e-01 7.96115637e-01 7.66614079e-01
-1.87919080e-01 7.27079868e-01 -9.29199994e-01 -6.27952278e-01
-8.19995284e-01 1.03035891e+00 1.33730292e+00 1.27901900e+00
5.77546477e-01 -6.42632604e-01 -1.25308737e-01 3.32156301e-01
1.27695739e-01 7.14200079e-01 -3.00487280e-01 -1.38381672e+00
8.67869079e-01 6.95903420e-01 2.02754047e-02 -9.02214527e-01
-6.51766434e-02 6.00540303e-02 -1.81803107e-01 2.01320443e-02
3.43886971e-01 -4.74647544e-02 -2.62103766e-01 1.70316100e+00
2.46138394e-01 -4.16877382e-02 2.94302493e-01 5.46309233e-01
1.07347012e+00 6.23281419e-01 -3.39782029e-01 9.05686393e-02
8.83625388e-01 -9.58757579e-01 -1.69176489e-01 1.72301546e-01
1.43977034e+00 -5.06743133e-01 1.24959850e+00 7.18074024e-01
-1.31607568e+00 1.86493114e-01 -1.00555599e+00 -3.29083204e-01
-5.84258676e-01 -1.39918074e-01 1.51276255e+00 2.51335144e-01
-1.14875698e+00 5.85628152e-01 -6.66695178e-01 -7.80173391e-02
4.65758443e-01 3.13780546e-01 -2.89051712e-01 -4.39157456e-01
-1.05614519e+00 6.11815095e-01 3.91615629e-01 -2.21148968e-01
-1.01154637e+00 -1.37374854e+00 -1.15624893e+00 6.53056651e-02
7.53909647e-01 -7.20293701e-01 1.77351606e+00 -1.98885411e-01
-9.38722312e-01 6.46589279e-01 -2.82454696e-02 -4.70372230e-01
5.56249499e-01 2.65199170e-02 -6.45818561e-02 6.02150783e-02
3.78903508e-01 4.81725723e-01 4.67019826e-01 -8.73990774e-01
-6.05180144e-01 2.75648922e-01 8.62175465e-01 -5.35348952e-01
5.50589487e-02 2.24334836e-01 -2.00437233e-01 1.06883146e-01
-3.00964534e-01 -4.97861594e-01 4.52126443e-01 -1.96345523e-01
-4.23629194e-01 -6.52128339e-01 7.98160911e-01 -4.11025196e-01
1.32988107e+00 -2.04414582e+00 -3.58560905e-02 2.94325560e-01
6.69196010e-01 6.55566081e-02 -1.45936444e-01 6.95680380e-01
8.09426382e-02 5.75761914e-01 1.84919029e-01 1.07591301e-01
8.23802710e-01 2.64407648e-03 -6.12687886e-01 1.58706859e-01
2.64950871e-01 1.03760898e+00 -1.34244990e+00 -1.03820336e+00
1.26706302e-01 -2.35305741e-01 -1.25184321e+00 -9.12690461e-02
-1.02853930e+00 -5.74685752e-01 -4.05755877e-01 1.00384414e+00
5.96772909e-01 -4.09532517e-01 2.18149021e-01 2.25587726e-01
-6.20785244e-02 1.03937602e+00 -1.00622904e+00 1.89808917e+00
-4.13445711e-01 5.88634729e-01 3.90290380e-01 -6.22208774e-01
1.25733793e-01 2.28786558e-01 -7.37212673e-02 -4.42179739e-01
-8.73846561e-02 3.61461818e-01 5.60470521e-02 -5.27466178e-01
3.14151376e-01 3.25179249e-01 -1.24247611e-01 7.36886263e-01
-5.47177996e-03 -5.41536093e-01 8.47905457e-01 8.55123043e-01
1.90091848e+00 3.38869750e-01 -2.40596801e-01 -9.72833782e-02
3.85015666e-01 5.14530122e-01 5.31650662e-01 1.15230930e+00
1.36288390e-01 -2.47233450e-01 9.41489220e-01 -4.46495295e-01
-1.02670121e+00 -1.03220606e+00 -1.02939844e-01 1.12017906e+00
-2.92651981e-01 -1.34895909e+00 -5.24754941e-01 -9.50691342e-01
4.64371920e-01 7.32721031e-01 -3.36933769e-02 -5.75990342e-02
-3.11115682e-01 1.71401769e-01 1.01175475e+00 3.68668646e-01
3.76675606e-01 -8.93348396e-01 -2.88830489e-01 -1.63906217e-01
-2.40681872e-01 -7.60827422e-01 -5.79226241e-02 1.08740509e-01
-6.19844258e-01 -1.73100793e+00 3.15038949e-01 -7.00664997e-01
6.28275394e-01 1.57248303e-01 1.37689674e+00 7.20276058e-01
-4.05168921e-01 5.69864333e-01 -1.41833663e-01 -4.93530214e-01
-9.52144086e-01 1.64054021e-01 -2.32428964e-03 -1.28220224e+00
3.29765648e-01 -6.68278754e-01 2.10615411e-01 -2.16884926e-01
-6.24667287e-01 4.64406103e-01 4.94396538e-01 8.96825314e-01
-9.02859941e-02 4.07918543e-01 1.83501437e-01 -1.00777590e+00
6.31204665e-01 -2.93150336e-01 -1.33165467e+00 3.51161718e-01
-6.55584812e-01 1.28710523e-01 9.30574000e-01 -1.21693037e-01
-4.11806285e-01 -5.89913845e-01 2.12771103e-01 -5.12561560e-01
1.26910463e-01 6.49893343e-01 1.08608447e-01 -3.14474329e-02
7.15876818e-01 -6.32505715e-02 -1.48349246e-02 -4.23182845e-02
3.57456118e-01 5.71457088e-01 6.46021903e-01 -1.57296228e+00
1.36481357e+00 -3.27078700e-01 1.26123741e-01 -9.26691815e-02
-7.34069943e-01 -2.38704935e-01 -9.84821171e-02 2.72114992e-01
-1.61444575e-01 -7.19272852e-01 -1.65507615e+00 1.11969998e-02
-1.25741506e+00 -1.00321710e+00 -6.26989920e-03 3.39603513e-01
-5.42057395e-01 2.42110565e-01 -8.40531886e-01 -8.83253515e-01
-6.43389463e-01 -1.04987264e+00 8.73198032e-01 -2.89188355e-01
-5.34538448e-01 -9.14047360e-01 -3.22548598e-02 5.11691988e-01
1.49300560e-01 2.00518295e-02 1.54453027e+00 -5.44246793e-01
-1.09316754e+00 -2.50310540e-01 -4.77692217e-01 3.80773753e-01
-4.01780009e-01 3.90188277e-01 -8.37188840e-01 -2.34930560e-01
-6.46155119e-01 -9.18853402e-01 4.97482210e-01 -2.10227489e-01
1.20566940e+00 -7.19898820e-01 -5.00227213e-01 6.20322704e-01
1.16259921e+00 -5.61059296e-01 2.73227066e-01 6.11687720e-01
3.15957189e-01 -1.91289008e-01 6.24431670e-01 4.94891256e-02
5.45724511e-01 -8.03007856e-02 3.88043851e-01 2.06758991e-01
3.43458802e-01 -5.58385134e-01 5.19429028e-01 3.79804641e-01
1.14018090e-01 3.70192349e-01 -1.24911976e+00 5.14726877e-01
-1.89965081e+00 -1.32404244e+00 -5.38257062e-02 1.95586348e+00
1.64038873e+00 4.91449982e-01 -1.05846055e-01 2.75350034e-01
-6.59219474e-02 -3.28725338e-01 -1.27770036e-01 -7.72978961e-01
3.46738100e-01 5.88526249e-01 3.58837694e-01 7.80891776e-01
-1.05312264e+00 9.37563300e-01 6.57028913e+00 8.84853959e-01
-8.75856459e-01 -2.04112202e-01 -2.81267762e-01 -2.43696824e-01
-6.71954095e-01 4.20252919e-01 -6.08172357e-01 6.29943311e-02
1.01695287e+00 -4.28829700e-01 1.05387330e+00 1.15576673e+00
-8.39845613e-02 -2.66347706e-01 -1.92633569e+00 6.70518100e-01
-1.90846398e-01 -1.86111748e+00 -2.66240358e-01 -1.48749873e-01
4.64652538e-01 6.76442981e-02 -1.23400465e-01 1.11571968e+00
5.75677872e-01 -1.07014787e+00 7.26229370e-01 3.24896067e-01
8.02495718e-01 -7.59075344e-01 7.72651136e-01 6.24922037e-01
-7.40787089e-01 -1.61377758e-01 -7.70025775e-02 -7.16647089e-01
-8.61377835e-01 2.86423206e-01 -1.47429109e+00 4.78737622e-01
6.05310202e-01 8.20462942e-01 -7.37036645e-01 6.82898760e-01
-3.23561788e-01 8.70509863e-01 -6.23130977e-01 -4.07883346e-01
1.05174877e-01 1.99874967e-01 5.76643467e-01 1.35753477e+00
-3.42680901e-01 -1.70558333e-01 3.01875830e-01 1.77029777e+00
-4.11198944e-01 -3.42032969e-01 -8.29457104e-01 -3.73284459e-01
5.94668627e-01 1.46857750e+00 -6.70386255e-02 -5.16046524e-01
-3.90615672e-01 3.31974119e-01 5.15123546e-01 2.85165399e-01
-1.14879525e+00 -6.77688897e-01 4.82605040e-01 7.31944665e-02
-3.13576832e-02 -1.94040060e-01 8.77946336e-03 -1.19118786e+00
4.74674940e-01 -1.39651728e+00 2.46654704e-01 -9.16229606e-01
-1.04726350e+00 -4.10857677e-01 4.84271526e-01 -4.06036109e-01
-4.29413259e-01 -7.58046925e-01 -5.13986170e-01 9.63410914e-01
-1.56107378e+00 -1.30487680e+00 1.28307760e-01 4.04196948e-01
-2.87151579e-02 2.91248900e-04 1.07171500e+00 3.11470091e-01
-3.98398697e-01 9.27906156e-01 -7.76981860e-02 4.56741661e-01
7.50603020e-01 -1.62266409e+00 2.47826323e-01 7.24877298e-01
-1.37348875e-01 1.45709193e+00 5.96998394e-01 -4.76079315e-01
-2.47205496e+00 -9.63810682e-01 1.06267214e+00 -9.83280897e-01
1.29235423e+00 -7.33292580e-01 -4.13602650e-01 1.07926369e+00
-2.96796829e-01 -1.69970002e-02 3.19224060e-01 8.17459226e-01
-7.93280303e-01 -2.93358594e-01 -1.05753887e+00 7.79877067e-01
1.06042886e+00 -9.52702999e-01 -7.43481755e-01 7.30082452e-01
7.76498318e-01 -8.40542257e-01 -9.21720982e-01 2.67002970e-01
7.12033868e-01 -7.17117310e-01 7.76636660e-01 -8.64271760e-01
9.36087728e-01 -3.07604730e-01 -1.63666144e-01 -6.85988784e-01
5.03848977e-02 -1.11857247e+00 -6.59165740e-01 1.45976758e+00
7.34886646e-01 -7.51943469e-01 4.24109012e-01 8.68118644e-01
-1.07433021e-01 -9.36007857e-01 -4.81060088e-01 -6.45997107e-01
2.76948750e-01 -9.97070849e-01 6.57542348e-01 9.80172634e-01
8.20288599e-01 3.55998933e-01 5.69505870e-01 1.97181568e-01
4.55277234e-01 5.51838875e-01 1.43468726e+00 -9.70187545e-01
-5.33348382e-01 -5.64505279e-01 -1.73158109e-01 -6.00629330e-01
6.51483297e-01 -1.61374784e+00 8.19975883e-03 -1.47764993e+00
7.07548797e-01 -8.27849507e-01 1.72479913e-01 1.19532287e+00
8.19383860e-02 -2.30313107e-01 -3.24508292e-03 -8.49776268e-02
-1.13264167e+00 -6.03410453e-02 8.42458487e-01 -4.84716445e-01
2.11765282e-02 -2.41004691e-01 -6.36913240e-01 7.39634037e-01
4.47534442e-01 -1.71565041e-01 -3.29628378e-01 -3.15222740e-01
9.68370438e-01 -3.52799632e-02 1.07299602e+00 -1.04109597e+00
3.13455850e-01 -2.95912147e-01 -4.82917130e-02 -6.04654670e-01
-2.01280192e-01 -5.63849747e-01 -2.65781671e-01 1.29085571e-01
-4.01167512e-01 1.02581449e-01 6.81973219e-01 1.19342431e-02
2.57482558e-01 -2.36731663e-01 -4.08647396e-02 -7.07758144e-02
-5.34107089e-01 -4.85707261e-02 -2.10308172e-02 4.29157019e-01
7.73122370e-01 3.41581881e-01 -7.62642503e-01 -1.03276029e-01
8.95544887e-02 7.07082868e-01 8.13849747e-01 8.05249661e-02
4.19604510e-01 -9.21246350e-01 -5.07793784e-01 1.55873656e-01
1.59101978e-01 3.67002606e-01 -3.47308666e-01 1.06930101e+00
-6.57731235e-01 6.71264887e-01 4.48691666e-01 -4.76432055e-01
-1.52919233e+00 8.08822632e-01 2.85386473e-01 -3.39964986e-01
-8.94194424e-01 6.96800888e-01 -4.20300603e-01 -1.02790165e+00
3.34739745e-01 -1.00504911e+00 8.23234677e-01 -9.07019317e-01
5.82383513e-01 1.87369823e-01 1.13729149e-01 6.06236100e-01
-4.98239219e-01 -1.84744701e-01 -1.65043533e-01 1.37174577e-01
1.32475257e+00 7.00208783e-01 -7.61448979e-01 4.20352608e-01
9.03041780e-01 4.28696811e-01 -3.70910525e-01 9.76025388e-02
-1.78132161e-01 -2.20018461e-01 -2.77835261e-02 -1.17106891e+00
-1.67518750e-01 5.97568154e-01 -4.36506778e-01 1.76106781e-01
6.69981301e-01 -5.96072851e-03 7.28509665e-01 1.18261909e+00
6.30757868e-01 -7.75450706e-01 -5.21792352e-01 6.82315946e-01
5.70010781e-01 -1.20916021e+00 2.51343280e-01 -2.88343787e-01
3.84690702e-01 1.08573461e+00 5.14472127e-01 7.00544715e-02
4.75585610e-02 8.16602349e-01 -3.45101684e-01 -1.68197095e-01
-1.47717750e+00 4.48972315e-01 -5.71218170e-02 4.72555369e-01
4.30238038e-01 -1.06527340e-02 1.49617597e-01 7.37616181e-01
-6.26735151e-01 5.32965183e-01 3.83849084e-01 1.33675420e+00
7.11173937e-02 -1.17730176e+00 -5.44264913e-01 5.23207664e-01
-2.97650784e-01 -5.64699829e-01 -3.86809528e-01 1.28149927e+00
8.29217285e-02 8.89246225e-01 -4.63623703e-01 -2.15342343e-01
-9.56360474e-02 3.21353257e-01 1.02036238e+00 -7.29458034e-01
-3.72615486e-01 -7.69952774e-01 5.88382661e-01 -6.99145794e-01
1.36014089e-01 -4.70834702e-01 -1.51740026e+00 -1.28099489e+00
-4.51061964e-01 5.39547980e-01 4.43624526e-01 9.33738708e-01
4.12671268e-01 4.72749680e-01 1.48368433e-01 -4.33302760e-01
-9.90166068e-01 -6.95736051e-01 -2.35738099e-01 6.13460205e-02
5.87948263e-01 -4.35997277e-01 -4.54553574e-01 5.46548143e-02] | [8.937458038330078, 7.052908897399902] |
5e66eba3-08d2-42f9-bcad-f65a3ba1c62d | on-the-eigenvalues-of-global-covariance | 2205.13282 | null | https://arxiv.org/abs/2205.13282v1 | https://arxiv.org/pdf/2205.13282v1.pdf | On the Eigenvalues of Global Covariance Pooling for Fine-grained Visual Recognition | The Fine-Grained Visual Categorization (FGVC) is challenging because the subtle inter-class variations are difficult to be captured. One notable research line uses the Global Covariance Pooling (GCP) layer to learn powerful representations with second-order statistics, which can effectively model inter-class differences. In our previous conference paper, we show that truncating small eigenvalues of the GCP covariance can attain smoother gradient and improve the performance on large-scale benchmarks. However, on fine-grained datasets, truncating the small eigenvalues would make the model fail to converge. This observation contradicts the common assumption that the small eigenvalues merely correspond to the noisy and unimportant information. Consequently, ignoring them should have little influence on the performance. To diagnose this peculiar behavior, we propose two attribution methods whose visualizations demonstrate that the seemingly unimportant small eigenvalues are crucial as they are in charge of extracting the discriminative class-specific features. Inspired by this observation, we propose a network branch dedicated to magnifying the importance of small eigenvalues. Without introducing any additional parameters, this branch simply amplifies the small eigenvalues and achieves state-of-the-art performances of GCP methods on three fine-grained benchmarks. Furthermore, the performance is also competitive against other FGVC approaches on larger datasets. Code is available at \href{https://github.com/KingJamesSong/DifferentiableSVD}{https://github.com/KingJamesSong/DifferentiableSVD}. | ['Wei Wang', 'Nicu Sebe', 'Yue Song'] | 2022-05-26 | null | null | null | null | ['fine-grained-visual-recognition', 'fine-grained-image-classification', 'fine-grained-visual-categorization'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-3.12214017e-01 -2.55648553e-01 -2.26949498e-01 -3.01379651e-01
-4.39915299e-01 -5.56997001e-01 6.01254284e-01 8.51390511e-02
-2.84556389e-01 6.05106831e-01 1.92031235e-01 -2.46907189e-01
-3.07352424e-01 -6.88914239e-01 -5.51954031e-01 -9.40126717e-01
-1.77841634e-01 -1.68467090e-01 3.49046677e-01 -2.25394100e-01
3.41134995e-01 4.46555108e-01 -1.67028069e+00 4.47861671e-01
9.67264116e-01 1.10369754e+00 8.72597024e-02 3.07708532e-01
-2.24337816e-01 3.16971213e-01 -5.48893154e-01 -1.73026219e-01
3.11578870e-01 -2.48725161e-01 -5.02927959e-01 -2.68049836e-01
5.13222814e-01 -1.52617604e-01 -2.73205549e-01 1.35784566e+00
3.75779957e-01 6.29060119e-02 8.31477284e-01 -1.37139297e+00
-9.44611311e-01 5.25161326e-01 -8.67413819e-01 3.69391531e-01
-1.03216894e-01 1.73171639e-01 1.18718851e+00 -9.48427618e-01
3.81392241e-01 1.25509715e+00 5.44761240e-01 3.53921980e-01
-1.19599915e+00 -8.99810016e-01 6.28884435e-01 2.96330124e-01
-1.67166638e+00 -1.38406292e-01 9.21891093e-01 -4.26594108e-01
3.70913208e-01 5.44297338e-01 3.19590390e-01 1.14935970e+00
9.88641605e-02 6.89750850e-01 1.32149911e+00 -9.69541743e-02
7.43064880e-02 1.80755541e-01 5.75833261e-01 6.55725658e-01
3.42450440e-01 2.91924588e-02 -2.59255946e-01 -6.66249171e-02
6.87951982e-01 2.87257999e-01 -5.02971888e-01 -3.31650496e-01
-1.25106144e+00 7.51309931e-01 8.13335896e-01 5.55739999e-01
-1.03951618e-01 8.73735175e-02 2.79821187e-01 3.60071212e-01
4.32409227e-01 3.96355063e-01 -3.57662171e-01 -8.49040598e-02
-6.14364982e-01 1.16071070e-03 4.37995374e-01 6.63528204e-01
9.05668676e-01 -1.27556249e-02 -5.62022328e-01 1.02356446e+00
1.18950300e-01 2.87035942e-01 4.55946237e-01 -7.56359577e-01
2.70627409e-01 7.84092367e-01 -6.04855269e-02 -1.23959124e+00
-5.70932031e-01 -7.98239887e-01 -1.27112055e+00 3.29791814e-01
8.03136587e-01 1.12149552e-01 -8.21939051e-01 1.67835879e+00
1.05141275e-01 -1.10525385e-01 -3.38652849e-01 1.19392073e+00
1.03559351e+00 5.41822910e-01 1.02255471e-01 4.66511957e-02
1.50575936e+00 -8.16017687e-01 -4.14594382e-01 -4.88201715e-02
2.66647995e-01 -6.45174623e-01 1.62995207e+00 1.51652023e-01
-4.96913612e-01 -6.44841313e-01 -1.03328240e+00 7.19188852e-03
-6.03460193e-01 5.13023019e-01 6.21711075e-01 4.57502633e-01
-8.67602348e-01 8.73551786e-01 -7.25286961e-01 -1.58298254e-01
6.09649241e-01 8.70447010e-02 -3.50162506e-01 1.67784933e-02
-1.06971312e+00 3.47627431e-01 1.99538797e-01 3.15808892e-01
-6.29605353e-01 -7.59568095e-01 -4.26908553e-01 3.22125494e-01
3.60063910e-01 -3.41742009e-01 7.42620826e-01 -7.77339339e-01
-1.20152926e+00 6.24484718e-01 -8.21870789e-02 1.86505169e-02
7.40874171e-01 -6.69841096e-02 -1.41232774e-01 7.34306499e-02
-3.09197996e-02 4.31086004e-01 1.04098797e+00 -1.30725932e+00
-2.44856611e-01 -4.43617135e-01 1.80426970e-01 -1.61496028e-01
-7.89010942e-01 -2.53433049e-01 -4.10422951e-01 -8.95577550e-01
1.83078974e-01 -7.83129156e-01 -4.69120312e-03 1.03544392e-01
-6.92229450e-01 -3.21130484e-01 7.43338466e-01 -3.02196860e-01
1.33768249e+00 -2.49152231e+00 -8.65972191e-02 2.27154061e-01
4.47726488e-01 1.89911455e-01 -2.19730496e-01 4.41563934e-01
-5.09159975e-02 4.73340452e-01 1.15354536e-02 -7.44199427e-03
2.94544280e-01 -1.84845969e-01 -2.72024035e-01 4.23195422e-01
2.18886122e-01 9.28723454e-01 -6.75421655e-01 -1.60786778e-01
5.65673821e-02 5.58006942e-01 -3.71315181e-01 -1.13565527e-01
1.11106262e-01 2.85353363e-01 -7.32880712e-01 5.25229692e-01
9.66378033e-01 -6.11014545e-01 1.38638780e-01 -5.64622641e-01
-1.95722684e-01 -7.85697550e-02 -1.25820708e+00 1.22213221e+00
-7.50972554e-02 4.72331673e-01 -9.98211745e-03 -9.27650928e-01
9.58713233e-01 -1.98175520e-01 1.11672550e-01 -6.65088594e-01
3.57074797e-01 6.67402893e-02 8.27953145e-02 -9.51131508e-02
3.24526966e-01 1.77910291e-02 -3.11267544e-02 3.16403091e-01
-2.13581726e-01 1.11068763e-01 2.02194095e-01 2.68840164e-01
8.42564106e-01 -2.13093489e-01 2.16782495e-01 -5.12336493e-01
4.99475747e-01 -3.83522809e-01 5.94238281e-01 7.69913733e-01
-4.14133638e-01 7.02210248e-01 8.58661294e-01 -3.95313978e-01
-7.53971100e-01 -1.11025405e+00 -1.82266608e-01 1.31022966e+00
3.62879366e-01 -5.71552098e-01 -6.08119488e-01 -7.34294176e-01
2.12278515e-01 2.74209559e-01 -9.59274769e-01 -3.84804070e-01
-1.66048005e-01 -9.79582667e-01 4.59327579e-01 4.84789193e-01
6.85432792e-01 -8.33343387e-01 -3.28176230e-01 -3.25375199e-01
-1.23148955e-01 -7.98869967e-01 -4.75731671e-01 1.17332898e-01
-6.53789282e-01 -1.03372359e+00 -8.62911046e-01 -3.80919874e-01
7.57725120e-01 4.43589300e-01 7.77336776e-01 3.15194100e-01
-2.50861108e-01 -5.43385819e-02 -3.76930445e-01 -2.50642419e-01
6.07293621e-02 1.91833064e-01 -1.34466082e-01 1.92707792e-01
3.49540591e-01 -7.05654144e-01 -8.28430057e-01 5.17299056e-01
-7.20993876e-01 -9.05268302e-04 6.05489850e-01 9.65789199e-01
5.83128572e-01 -5.04325628e-02 5.77840567e-01 -7.23765075e-01
7.58204997e-01 -2.23314404e-01 -4.97465640e-01 7.78547972e-02
-4.77220267e-01 6.68880939e-02 1.08685505e+00 -6.27950072e-01
-7.37986863e-01 -3.79534423e-01 9.34104249e-03 -6.08541727e-01
-2.43089139e-01 1.90089792e-01 1.61459409e-02 -1.12925181e-02
6.73978806e-01 2.52067953e-01 -1.97234929e-01 -8.04109871e-01
2.84889340e-01 5.36853254e-01 1.13325015e-01 -5.56394458e-01
8.81467700e-01 5.61155200e-01 -1.83053806e-01 -6.66051447e-01
-7.11657643e-01 -4.07855749e-01 -5.14809251e-01 2.32588835e-02
5.88691771e-01 -7.71741629e-01 -8.89789104e-01 4.58560526e-01
-6.64340556e-01 -3.74403775e-01 -1.52242616e-01 3.31377417e-01
-1.14779197e-01 4.25185293e-01 -6.40771806e-01 -5.53500891e-01
-2.65058964e-01 -1.13185036e+00 8.70059967e-01 3.75930190e-01
-9.01938230e-02 -6.69220209e-01 -3.18065107e-01 6.44816607e-02
5.77102125e-01 7.23670945e-02 1.10317016e+00 -5.31602144e-01
-5.48709929e-01 -1.15312792e-01 -7.54995048e-01 4.55834210e-01
1.65802121e-01 2.37132087e-01 -1.14788973e+00 -3.70105088e-01
-4.31284606e-01 -1.19486123e-01 1.11645365e+00 3.44615757e-01
1.68986571e+00 -7.71845728e-02 -2.04946324e-01 7.60768235e-01
1.26815760e+00 -9.58194733e-02 4.76958722e-01 3.80915314e-01
8.21081817e-01 4.71334428e-01 4.16426778e-01 5.77745080e-01
3.31928730e-01 6.41616881e-01 4.25643623e-01 -1.16063781e-01
-2.95521170e-01 -1.33695483e-01 2.04756871e-01 6.75734401e-01
-3.92637134e-01 7.12744147e-03 -7.34155357e-01 3.45211685e-01
-1.68003774e+00 -8.54736626e-01 -2.19535500e-01 2.14992356e+00
7.00932145e-01 1.58543631e-01 -1.72824394e-02 1.33079082e-01
5.98522902e-01 5.87476432e-01 -5.31432450e-01 -1.68151781e-01
-1.93402976e-01 -5.51378019e-02 3.60972017e-01 2.26837754e-01
-1.18089104e+00 8.79257679e-01 4.83977318e+00 1.22230327e+00
-1.33384299e+00 1.14153214e-01 8.19741964e-01 -1.42735735e-01
-3.83299768e-01 -3.34019840e-01 -6.96856558e-01 6.59061909e-01
4.86758113e-01 -1.97169513e-01 3.49330932e-01 7.16762185e-01
2.94193149e-01 5.33747720e-03 -8.40264916e-01 9.03145909e-01
-2.55267590e-01 -1.30519938e+00 5.67378178e-02 -1.34122572e-04
5.88211417e-01 2.44465806e-02 2.66765088e-01 3.18997204e-01
5.20111956e-02 -9.66826677e-01 6.49311781e-01 4.42363590e-01
8.01387191e-01 -6.99840188e-01 7.37828016e-01 2.10083410e-01
-1.26951790e+00 -2.78280079e-01 -6.19165123e-01 3.61525267e-03
-3.42059374e-01 9.21291649e-01 -2.36361325e-01 5.17975152e-01
1.06811595e+00 6.08614564e-01 -9.09085512e-01 9.68134761e-01
-3.44148874e-01 5.49146473e-01 -1.46497428e-01 -1.19241521e-01
1.64961040e-01 -2.35484794e-01 4.97585922e-01 1.16188061e+00
1.95630342e-01 -1.05662569e-02 4.09841500e-02 9.78277087e-01
-2.61732459e-01 2.00404033e-01 -3.16461146e-01 -1.00067943e-01
3.16031367e-01 1.53376222e+00 -1.03028190e+00 -3.73627424e-01
-2.63319641e-01 8.76327872e-01 5.31185985e-01 6.67219341e-01
-7.40085304e-01 -6.24510527e-01 9.10167038e-01 5.27816415e-02
5.71056068e-01 -9.90644842e-02 -4.12608594e-01 -1.39278710e+00
3.02755386e-01 -8.65596592e-01 3.99623990e-01 -4.29219544e-01
-1.48756742e+00 7.28365898e-01 -2.11140871e-01 -1.27372527e+00
2.30932415e-01 -5.73892891e-01 -6.28724873e-01 8.12064171e-01
-1.54898429e+00 -9.48166966e-01 -7.30506301e-01 6.82425380e-01
1.52721286e-01 8.73533264e-02 6.22374177e-01 3.68953407e-01
-8.20199072e-01 9.23688054e-01 3.01132202e-01 2.56890595e-01
8.32943082e-01 -1.12755561e+00 1.24461874e-01 6.94373667e-01
-9.86997597e-03 7.98030138e-01 5.32916844e-01 -3.82953286e-01
-1.18287921e+00 -1.07932532e+00 4.54333961e-01 -3.01985502e-01
7.51292706e-01 -6.69107199e-01 -1.19491029e+00 3.05349469e-01
-1.33240432e-01 4.18953061e-01 4.39560592e-01 1.13467209e-01
-8.25675011e-01 -3.47964168e-01 -9.98557568e-01 5.90878546e-01
1.10983050e+00 -5.71042418e-01 -3.04882526e-01 1.83232993e-01
5.32505810e-01 6.83557391e-02 -8.00370812e-01 3.65364671e-01
7.16953456e-01 -1.12536907e+00 9.60362554e-01 -5.32313526e-01
4.39256966e-01 -3.87659729e-01 -1.33700907e-01 -1.37614560e+00
-5.42705297e-01 -2.66738027e-01 1.62500501e-01 1.19584298e+00
3.14955354e-01 -1.05208397e+00 6.71172917e-01 2.57963896e-01
-1.22377872e-02 -9.23875332e-01 -7.62076139e-01 -9.12608445e-01
2.82532632e-01 -2.20629483e-01 6.28261924e-01 9.80301082e-01
-9.18935090e-02 1.33750569e-02 -2.80316025e-01 4.03710119e-02
5.28893232e-01 5.53251863e-01 5.91589928e-01 -1.32503068e+00
-1.33975908e-01 -7.57534266e-01 -4.09462363e-01 -8.90754700e-01
-8.21449012e-02 -8.43001187e-01 -1.56879202e-01 -1.33864021e+00
3.55106950e-01 -5.56784093e-01 -6.40784264e-01 6.59772694e-01
-3.57406378e-01 5.94890535e-01 5.46737015e-01 3.56860608e-01
-4.86613393e-01 6.57914102e-01 1.43682969e+00 -2.08275944e-01
-5.98117523e-02 -3.36630605e-02 -1.09904444e+00 6.59310699e-01
9.01442647e-01 -3.15817773e-01 -3.56444299e-01 -1.96606919e-01
-4.02240604e-02 -5.51290631e-01 5.64039886e-01 -7.26478755e-01
6.86698779e-02 -1.19387515e-01 5.80194771e-01 -3.21069121e-01
1.82758912e-01 -5.28917611e-01 -4.06903364e-02 3.30697924e-01
-1.51608974e-01 -1.03108443e-01 3.00186515e-01 4.41290945e-01
-3.84658277e-01 7.43531138e-02 7.93880880e-01 -5.97742759e-03
-7.59303331e-01 4.18438822e-01 -2.53992438e-01 1.10752255e-01
9.00733232e-01 -7.83542097e-02 -7.46920347e-01 -2.77653992e-01
-6.24208152e-01 6.91065788e-02 4.56246406e-01 5.50821662e-01
3.71037453e-01 -1.29607213e+00 -5.47716379e-01 3.52504075e-01
1.85631171e-01 -3.83061141e-01 6.31094694e-01 1.12516367e+00
-1.12369649e-01 3.44943911e-01 -4.03904557e-01 -5.64358592e-01
-1.15413857e+00 5.31376779e-01 2.28224814e-01 -2.26461142e-01
-4.78450686e-01 9.12692308e-01 6.17502451e-01 -4.31684792e-01
1.72554761e-01 -4.90323424e-01 -3.47340107e-01 4.28769261e-01
7.84763813e-01 3.98698062e-01 4.77085896e-02 -3.84678572e-01
-5.50707102e-01 7.65725851e-01 -2.09703594e-01 5.32325625e-01
1.35321367e+00 -1.45887569e-01 2.26082914e-02 4.21448559e-01
1.43060899e+00 1.79425552e-01 -1.37403560e+00 -1.34872392e-01
-2.61715233e-01 -7.06195951e-01 -1.64438225e-02 -6.82167828e-01
-1.35184550e+00 1.17901087e+00 7.30515301e-01 4.15668726e-01
1.28180432e+00 1.32935151e-01 3.21626782e-01 1.72879726e-01
3.07229668e-01 -8.48014951e-01 1.36468858e-01 4.68010664e-01
1.04382956e+00 -1.36705041e+00 8.21743980e-02 -6.04097545e-01
-6.39183402e-01 1.17217219e+00 5.59446454e-01 -2.54001617e-01
7.37590373e-01 4.90484908e-02 1.69174165e-01 -2.14149803e-01
-5.39692461e-01 -1.70871198e-01 5.73981047e-01 3.47409815e-01
3.60287726e-01 2.95712978e-01 -4.76312310e-01 7.30101347e-01
-1.82894409e-01 -2.51566738e-01 3.16559613e-01 5.01929224e-01
-2.57568806e-01 -9.03010309e-01 -2.27143481e-01 6.49065137e-01
-3.54969054e-01 -2.01995656e-01 -4.25999790e-01 8.46828759e-01
-7.13261869e-03 7.12697387e-01 3.76986563e-02 -4.90599245e-01
4.46289450e-01 -1.55215234e-01 6.68975338e-02 -2.68839866e-01
-5.59105992e-01 -2.90443413e-02 -1.88089535e-01 -9.09474432e-01
-3.14769633e-02 -4.28444892e-01 -1.11828649e+00 -5.34951925e-01
-7.16140717e-02 1.02048010e-01 3.17242086e-01 5.88699698e-01
5.21088123e-01 6.79316282e-01 6.71256423e-01 -8.11544359e-01
-7.40190387e-01 -1.04616427e+00 -7.65016913e-01 4.58749533e-01
4.26702529e-01 -9.18629706e-01 -7.15765357e-01 -4.21047032e-01] | [9.597843170166016, 2.1108856201171875] |
f1078a22-c41b-4878-a08a-165553afe03e | cross-camera-deep-colorization | 2209.01211 | null | https://arxiv.org/abs/2209.01211v2 | https://arxiv.org/pdf/2209.01211v2.pdf | Cross-Camera Deep Colorization | In this paper, we consider the color-plus-mono dual-camera system and propose an end-to-end convolutional neural network to align and fuse images from it in an efficient and cost-effective way. Our method takes cross-domain and cross-scale images as input, and consequently synthesizes HR colorization results to facilitate the trade-off between spatial-temporal resolution and color depth in the single-camera imaging system. In contrast to the previous colorization methods, ours can adapt to color and monochrome cameras with distinctive spatial-temporal resolutions, rendering the flexibility and robustness in practical applications. The key ingredient of our method is a cross-camera alignment module that generates multi-scale correspondences for cross-domain image alignment. Through extensive experiments on various datasets and multiple settings, we validate the flexibility and effectiveness of our approach. Remarkably, our method consistently achieves substantial improvements, i.e., around 10dB PSNR gain, upon the state-of-the-art methods. Code is at: https://github.com/IndigoPurple/CCDC | ['Ruqi Huang', 'Mengqi Ji', 'Haitian Zheng', 'Yaping Zhao'] | 2022-08-26 | null | null | null | null | ['colorization'] | ['computer-vision'] | [ 1.37175545e-01 -5.73919356e-01 -3.42185274e-02 -1.15390122e-01
-9.48278189e-01 -9.01642561e-01 2.52990991e-01 -6.59122169e-01
-3.02621454e-01 3.23851585e-01 -7.17487261e-02 -3.90913606e-01
2.12483585e-01 -5.34983873e-01 -6.75979078e-01 -5.93090296e-01
4.53938663e-01 -3.36892933e-01 1.41454548e-01 -1.68374449e-01
2.06892997e-01 5.37910163e-01 -1.34046614e+00 6.72462136e-02
6.59805715e-01 1.02544367e+00 1.96008876e-01 9.17302132e-01
4.44764018e-01 6.58516943e-01 -7.89066777e-02 -5.29170215e-01
8.41991365e-01 -4.21301961e-01 -6.41445816e-01 5.09405196e-01
8.71014357e-01 -7.88772106e-01 -6.37438715e-01 1.14497757e+00
4.58242595e-01 -1.14507861e-01 3.58704887e-02 -1.08542538e+00
-1.02372241e+00 -7.45537505e-02 -1.10689306e+00 5.36103472e-02
4.38375384e-01 3.33475828e-01 8.04211199e-01 -9.79054689e-01
3.14035833e-01 9.82160151e-01 5.00466466e-01 3.91748697e-01
-1.30818188e+00 -7.87071168e-01 7.83277601e-02 -1.29271140e-02
-1.46798849e+00 -7.71733999e-01 7.24920928e-01 -1.34880617e-01
4.80624855e-01 3.67119052e-02 4.57842231e-01 9.01458025e-01
8.51736814e-02 4.48240697e-01 1.28343129e+00 -4.01169449e-01
-2.03922644e-01 -2.77120531e-01 -4.18222129e-01 6.15121782e-01
3.51510406e-01 3.15508455e-01 -5.03687322e-01 1.48674130e-01
1.31351149e+00 2.78300554e-01 -6.24747455e-01 -3.41689199e-01
-1.37800169e+00 3.16568196e-01 5.42949557e-01 1.09120056e-01
-1.50308967e-01 4.03936088e-01 1.55091612e-02 1.91772982e-01
2.84837246e-01 3.19559246e-01 -3.13248515e-01 3.23360302e-02
-7.90058970e-01 -1.69621885e-01 2.51306146e-01 1.32046366e+00
7.02504933e-01 -4.87149432e-02 1.73781529e-01 8.69352341e-01
6.52399510e-02 7.13282347e-01 1.84297234e-01 -1.35786200e+00
5.01622260e-01 2.06758305e-01 2.36794278e-01 -8.25018823e-01
-1.08589843e-01 -2.50542998e-01 -9.74011958e-01 4.76681918e-01
2.40925461e-01 -2.38485545e-01 -7.91195869e-01 1.64128041e+00
2.90283471e-01 2.12813169e-01 -2.49987375e-02 1.22662103e+00
4.39248502e-01 4.80763286e-01 -4.32947457e-01 -1.26542404e-01
1.40158081e+00 -1.14263654e+00 -5.99550545e-01 -3.14523637e-01
-8.36479887e-02 -1.31045091e+00 1.06608462e+00 4.46580023e-01
-1.44549751e+00 -6.04036331e-01 -1.15175843e+00 -4.48918670e-01
-2.71421286e-05 4.32408184e-01 4.27426070e-01 4.73193288e-01
-1.26988721e+00 3.46477956e-01 -7.38266528e-01 -2.98229635e-01
4.48844396e-02 1.38570830e-01 -4.75920051e-01 -3.76374930e-01
-7.14530051e-01 6.18731916e-01 1.67509764e-01 6.24798909e-02
-5.88805795e-01 -7.00756848e-01 -5.90682089e-01 -4.81201001e-02
4.87329215e-01 -8.84322107e-01 1.38196170e+00 -1.14797425e+00
-1.80482662e+00 9.57725167e-01 -1.49205908e-01 6.23835139e-02
5.83745480e-01 -3.66710722e-01 -4.34281707e-01 3.93697262e-01
7.72677734e-02 6.38755500e-01 8.37261319e-01 -1.41404784e+00
-9.47592616e-01 -1.04772881e-01 2.46491298e-01 3.70316446e-01
-2.02132627e-01 1.55738294e-01 -1.37383127e+00 -6.15628362e-01
1.92242041e-01 -1.05900693e+00 -1.13750145e-01 3.81410241e-01
-3.70185673e-01 6.28897607e-01 6.79576457e-01 -5.43741286e-01
9.65547144e-01 -2.32148790e+00 5.48334457e-02 -1.34761930e-01
4.25514013e-01 1.02165155e-01 -2.34231800e-01 2.96371609e-01
-2.40329310e-01 -3.21061552e-01 -1.93070769e-01 -3.95190835e-01
-3.00443232e-01 -1.49918601e-01 -2.64804959e-01 8.46756399e-01
1.51884675e-01 7.05271602e-01 -7.27841735e-01 -2.54365325e-01
5.32463193e-01 5.54928601e-01 -4.58687037e-01 3.99888873e-01
2.30753198e-01 5.28129399e-01 -1.08718067e-01 9.37122643e-01
1.07393122e+00 -4.73600358e-01 3.43137473e-01 -5.39408624e-01
-3.09986830e-01 -1.09776072e-01 -1.14005768e+00 1.75045216e+00
-6.61472738e-01 8.55698109e-01 2.85476029e-01 -3.90106946e-01
6.91130877e-01 4.51366082e-02 4.27677155e-01 -1.10001290e+00
1.23596862e-01 3.16994518e-01 -3.31042379e-01 7.70279532e-03
8.67830455e-01 1.65985823e-01 2.13312116e-02 3.86352450e-01
-7.33129084e-02 3.61095145e-02 1.07742965e-01 1.03048682e-01
7.16722846e-01 1.78374276e-01 3.51067454e-01 5.91116324e-02
5.24283886e-01 -4.61087853e-01 5.85520208e-01 3.27603012e-01
-1.69630051e-01 1.14577961e+00 2.86521822e-01 -3.18672329e-01
-1.46351612e+00 -1.17318690e+00 5.17500788e-02 8.44639003e-01
6.82284653e-01 -2.03702703e-01 -5.04583418e-01 -1.44181341e-01
-2.60710865e-01 1.82041794e-01 -3.00642520e-01 1.29002213e-01
-6.49019241e-01 -3.31066608e-01 3.77369851e-01 5.91092765e-01
8.64401579e-01 -2.65599638e-01 -7.10834920e-01 -9.34291035e-02
-2.62323529e-01 -1.58488190e+00 -9.32198107e-01 -1.02016903e-01
-6.40766323e-01 -1.30394495e+00 -9.06297028e-01 -6.94847167e-01
7.02336609e-01 1.11161709e+00 1.10752261e+00 -2.21681762e-02
-2.88437933e-01 5.00099719e-01 -3.60325456e-01 1.74718708e-01
-1.57483444e-02 -1.78465977e-01 -2.27206200e-02 1.43632129e-01
-5.70753925e-02 -6.73178613e-01 -1.21069312e+00 5.59599817e-01
-1.12151194e+00 5.31039536e-01 7.58032978e-01 8.48074019e-01
6.79098845e-01 -3.39274019e-01 1.09808752e-02 -4.70369160e-01
3.21349412e-01 -6.14834353e-02 -1.17316198e+00 2.75945067e-01
-6.65257514e-01 -2.55416036e-01 6.99398696e-01 -2.05871344e-01
-9.84137177e-01 3.25221062e-01 7.76435211e-02 -7.44970500e-01
-1.22443270e-02 -7.32636778e-03 -1.00934327e-01 -3.93115968e-01
3.85152787e-01 3.27843070e-01 2.28837710e-02 -3.08246911e-01
7.44112313e-01 5.55679023e-01 1.00914562e+00 -4.52179402e-01
9.89165008e-01 8.37783873e-01 -1.47563979e-01 -4.82961118e-01
-7.64501214e-01 -5.26412487e-01 -6.82902753e-01 -3.07232976e-01
8.22153509e-01 -1.45238352e+00 -8.17365766e-01 6.70645475e-01
-1.04198170e+00 -4.42750871e-01 2.40718439e-01 3.55097413e-01
-6.23476505e-01 5.64855635e-01 -7.01329291e-01 -3.72566015e-01
-3.65599543e-01 -1.24905860e+00 1.32756126e+00 5.27433455e-01
4.17352766e-01 -7.14740098e-01 -5.47630042e-02 2.45329440e-01
5.42216241e-01 9.18474793e-02 3.35733294e-01 4.23845708e-01
-1.08377326e+00 1.36172935e-01 -7.74970949e-01 3.71402055e-01
2.51971692e-01 1.83510378e-01 -1.03867579e+00 -5.40387511e-01
-3.36088568e-01 -2.30010405e-01 6.17261231e-01 3.18823069e-01
1.28355289e+00 -1.09772861e-01 1.18459119e-02 1.29954898e+00
1.87118721e+00 1.95907384e-01 8.13457787e-01 4.95445460e-01
8.11906099e-01 1.08014531e-01 6.10222340e-01 4.06549066e-01
4.81640249e-01 1.03422129e+00 4.85908717e-01 -7.61234105e-01
-1.91392705e-01 5.44573553e-02 3.78058910e-01 5.40884078e-01
-3.90474051e-02 -9.89905670e-02 -6.27328396e-01 4.27203298e-01
-1.71348727e+00 -9.11263525e-01 7.65464008e-02 2.22897482e+00
8.32779050e-01 -3.81206214e-01 9.26581174e-02 -2.06612557e-01
8.75281513e-01 2.08202720e-01 -6.38526976e-01 2.97449883e-02
-3.37848067e-01 9.75667872e-03 9.84335542e-01 5.02127230e-01
-1.12656987e+00 7.82305658e-01 6.04359770e+00 5.56213021e-01
-1.45520127e+00 1.37820118e-03 8.91852021e-01 -3.57628614e-01
-6.59292191e-02 -9.04959068e-02 -4.54440475e-01 4.39032197e-01
5.82633197e-01 -1.72424644e-01 9.27086413e-01 6.30954504e-01
2.75725704e-02 -1.15427710e-01 -1.08940673e+00 1.46044672e+00
1.86154574e-01 -1.40498281e+00 -4.52885419e-01 1.09406695e-01
9.42441106e-01 1.85838506e-01 5.30989051e-01 -4.09506470e-01
3.59763622e-01 -7.01647103e-01 9.94158387e-01 3.01364094e-01
1.36334801e+00 -7.72146821e-01 4.65934396e-01 -2.45449930e-01
-1.43786597e+00 -1.04242414e-01 -2.67719299e-01 2.99887657e-01
2.52456516e-01 4.07344609e-01 -1.20194651e-01 8.58994305e-01
1.00076222e+00 8.57643068e-01 -6.42094910e-01 9.29714262e-01
-1.61928266e-01 -9.13019013e-03 -8.12558085e-02 6.31606340e-01
2.14149266e-01 -5.16075075e-01 1.18222967e-01 1.12152576e+00
4.68702793e-01 3.34953249e-01 2.24695671e-02 8.46531630e-01
-2.19839841e-01 -4.00937468e-01 -5.23454189e-01 2.05070764e-01
6.48118019e-01 1.50023603e+00 -5.41196167e-01 -1.10514961e-01
-7.97124386e-01 1.42036092e+00 1.81622490e-01 5.26756406e-01
-1.16055870e+00 -5.20590425e-01 9.55611646e-01 -2.12528855e-01
5.15329719e-01 -4.04308915e-01 -2.79355317e-01 -1.30223739e+00
3.22662979e-01 -1.01609004e+00 1.79602996e-01 -1.08779526e+00
-1.30361521e+00 5.80348194e-01 -2.35953122e-01 -1.79062915e+00
7.59400427e-02 -7.77144551e-01 -3.99511248e-01 9.24530566e-01
-1.91104686e+00 -1.33167350e+00 -7.77362168e-01 8.23737085e-01
2.85866469e-01 3.18619646e-02 3.96402240e-01 5.77656090e-01
-5.73875606e-01 7.58666813e-01 5.09081721e-01 3.11169177e-01
1.12230837e+00 -9.48354900e-01 6.22771680e-01 1.48745108e+00
-1.88508078e-01 6.36916578e-01 3.51966500e-01 -5.44975288e-02
-1.78494155e+00 -1.15117717e+00 1.77614421e-01 -2.30035633e-01
5.89902878e-01 -2.09737763e-01 -6.31714880e-01 5.84055901e-01
6.59215987e-01 3.19184661e-01 5.33266842e-01 -3.35812092e-01
-6.93962991e-01 -4.27339166e-01 -7.87351727e-01 7.82878995e-01
9.88522649e-01 -7.42675364e-01 1.09729997e-03 1.19730420e-01
7.50599086e-01 -8.99408817e-01 -7.13191807e-01 8.51604193e-02
7.96886027e-01 -1.40284634e+00 1.13948119e+00 7.70097300e-02
7.30490148e-01 -8.45522761e-01 -3.21018189e-01 -1.11626804e+00
-5.08563876e-01 -7.88044333e-01 1.84452325e-01 1.15166998e+00
2.29112640e-01 -7.62970507e-01 3.65202904e-01 4.66637194e-01
-1.92062199e-01 -5.70637941e-01 -7.87218988e-01 -6.81031168e-01
-1.71497121e-01 -2.36120015e-01 5.20951986e-01 8.67291749e-01
-3.51800323e-01 3.56274806e-02 -6.91959441e-01 5.04238307e-01
8.38519394e-01 4.74214137e-01 9.55373466e-01 -6.03601456e-01
-4.68369514e-01 -4.64149535e-01 -1.66314140e-01 -1.36118567e+00
-1.89501107e-01 -2.57638514e-01 -3.60442623e-02 -1.21816921e+00
2.94633836e-01 -2.51748919e-01 -1.70539334e-01 3.51562768e-01
-2.20337152e-01 7.08598793e-01 5.25324821e-01 3.12693179e-01
-7.34564006e-01 3.54442209e-01 1.26084280e+00 1.44713670e-01
-4.68171388e-02 -4.31929827e-01 -1.07366478e+00 5.41557729e-01
6.63545370e-01 -3.00698029e-03 -2.43239671e-01 -9.31382895e-01
1.07832700e-01 2.50708699e-01 5.28089345e-01 -9.35435116e-01
4.11007732e-01 -2.58837551e-01 6.41711473e-01 -4.48203057e-01
4.44207311e-01 -9.66681004e-01 3.45852166e-01 1.80571005e-01
-1.29824355e-01 4.71633524e-01 1.57026932e-01 4.99184906e-01
-3.16781223e-01 4.75879759e-01 1.22766685e+00 3.49982977e-02
-7.46753573e-01 5.37242770e-01 1.53045803e-02 -9.63732004e-02
1.10116231e+00 -2.95983195e-01 -8.16544533e-01 -5.65657616e-01
3.00657935e-02 8.60440806e-02 1.09867108e+00 3.65387231e-01
7.58290350e-01 -1.47441757e+00 -4.66515332e-01 3.16817820e-01
3.31614614e-01 -9.80631784e-02 5.29291391e-01 9.78290200e-01
-9.83777702e-01 2.91132927e-01 -4.75916833e-01 -6.71862960e-01
-1.28752339e+00 4.86961365e-01 5.89596331e-01 1.53075248e-01
-7.20086992e-01 6.24136567e-01 3.72230113e-01 3.67068383e-03
1.44382557e-02 -2.31839463e-01 3.76949817e-01 -5.25141895e-01
6.69091702e-01 2.12764174e-01 -8.37104991e-02 -6.35393798e-01
-2.83318222e-01 9.45990324e-01 -2.91670598e-02 -3.21368247e-01
1.23444855e+00 -6.28227115e-01 -1.45831272e-01 8.19369406e-02
1.32185781e+00 2.02539563e-01 -1.69307339e+00 -3.54977489e-01
-6.67832136e-01 -1.10286975e+00 1.47562116e-01 -6.00078523e-01
-1.55507827e+00 6.21359944e-01 7.44314909e-01 -5.31689227e-02
1.73910666e+00 -1.83393776e-01 7.95461416e-01 1.26900151e-01
3.05757046e-01 -9.70102787e-01 2.56552309e-01 3.42762291e-01
6.78752482e-01 -1.39116168e+00 1.65071264e-01 -3.97354543e-01
-7.91470170e-01 1.37044787e+00 6.98611915e-01 -9.07074362e-02
1.98713765e-01 3.10517460e-01 3.95479769e-01 5.08498959e-02
-7.16072917e-01 -1.38231233e-01 2.15589538e-01 4.22077864e-01
5.21969616e-01 8.79882718e-05 1.40223742e-01 -1.10747768e-02
8.01539943e-02 -1.49268731e-01 8.51530135e-01 7.57328153e-01
-1.08604260e-01 -1.01414287e+00 -5.23742497e-01 -2.30120063e-01
-3.56973112e-01 -1.89057603e-01 -1.47838354e-01 7.97909439e-01
-1.07378475e-01 1.04206979e+00 7.88866282e-02 -5.25540650e-01
3.23165864e-01 -6.61221743e-01 4.78685945e-01 -8.98083672e-02
-2.49750808e-01 3.61027837e-01 -1.88400865e-01 -1.06792140e+00
-5.65111399e-01 -4.08169240e-01 -7.71316111e-01 -6.32635355e-01
-1.54107153e-01 -5.03566504e-01 5.65489352e-01 4.13788021e-01
6.27951384e-01 4.01891530e-01 1.12991679e+00 -1.11521816e+00
-1.66004717e-01 -4.24542069e-01 -4.83631581e-01 3.82147372e-01
7.30835915e-01 -3.99943590e-01 -1.60419941e-01 3.02676469e-01] | [10.740056991577148, -2.0167064666748047] |
80e939ed-c33d-404a-9db3-89c66baae4a5 | cs-um6p-at-semeval-2021-task-7-deep-multi | null | null | https://aclanthology.org/2021.semeval-1.159 | https://aclanthology.org/2021.semeval-1.159.pdf | CS-UM6P at SemEval-2021 Task 7: Deep Multi-Task Learning Model for Detecting and Rating Humor and Offense | Humor detection has become a topic of interest for several research teams, especially those involved in socio-psychological studies, with the aim to detect the humor and the temper of a targeted population (e.g. a community, a city, a country, the employees of a given company). Most of the existing studies have formulated the humor detection problem as a binary classification task, whereas it revolves around learning the sense of humor by evaluating its different degrees. In this paper, we propose an end-to-end deep Multi-Task Learning (MTL) model to detect and rate humor and offense. It consists of a pre-trained transformer encoder and task-specific attention layers. The model is trained using MTL uncertainty loss weighting to adaptively combine all sub-tasks objective functions. Our MTL model tackles all sub-tasks of the SemEval-2021 Task-7 in one end-to-end deep learning system and shows very promising results. | ['Ismail Berrada', 'Nabil El Mamoun', 'Abdelkader El Mahdaouy', 'Abdellah El Mekki', 'Kabil Essefar'] | 2021-08-01 | null | null | null | semeval-2021 | ['humor-detection'] | ['natural-language-processing'] | [-2.57793009e-01 3.19892466e-02 9.75932106e-02 -2.50272751e-01
-4.09065098e-01 1.64699391e-01 6.65288329e-01 2.80734122e-01
-3.04708451e-01 6.59230053e-01 5.88123143e-01 2.91609764e-02
4.32832539e-01 -6.49638474e-01 -3.33118886e-01 -3.85941595e-01
2.55177766e-01 5.07496715e-01 -9.04947817e-02 -3.71049404e-01
4.00746107e-01 -9.50834900e-02 -1.04788375e+00 6.15916550e-01
1.03177083e+00 8.38535070e-01 -1.63934156e-01 6.27311587e-01
5.33585489e-01 1.80836046e+00 -4.49667841e-01 -9.39603746e-01
-4.28622097e-01 -6.24452293e-01 -9.17979777e-01 2.72498131e-02
5.37184477e-01 -1.99740112e-01 -4.54582602e-01 1.12856281e+00
8.26219916e-01 3.79618704e-02 6.17322564e-01 -1.20514417e+00
-7.26818800e-01 8.36064100e-01 -5.20599842e-01 5.99435009e-02
3.00462037e-01 2.67165959e-01 1.27773440e+00 -1.02752686e+00
1.19378269e-01 1.42007864e+00 8.18353474e-01 4.22701925e-01
-1.10388756e+00 -5.21479964e-01 -5.79764426e-01 7.90080011e-01
-1.06201756e+00 -3.49695802e-01 1.29498744e+00 -7.14168489e-01
6.99419618e-01 -8.70804042e-02 2.13146016e-01 1.50498617e+00
3.11753750e-01 1.10213745e+00 1.12424481e+00 -1.72094256e-01
1.31233305e-01 3.89817417e-01 2.12569565e-01 6.79881632e-01
-2.01402754e-02 -3.18557978e-01 -6.35767162e-01 -1.19964883e-01
1.93983570e-01 -1.84348688e-01 -7.06832781e-02 1.57928001e-02
-7.04259992e-01 1.30616498e+00 6.92377687e-01 5.48387527e-01
-4.19149607e-01 1.08992323e-01 8.85175526e-01 1.90357000e-01
5.47280848e-01 4.04894650e-01 5.95841818e-02 -1.60204336e-01
-1.25802422e+00 4.41095024e-01 8.95143092e-01 4.46845859e-01
2.04676956e-01 9.40108579e-03 -5.63584566e-01 1.01649356e+00
2.32091919e-01 2.11039782e-01 5.62813938e-01 -4.90029037e-01
4.65008020e-01 6.38271570e-01 7.75671825e-02 -1.22000837e+00
-9.21807647e-01 -7.38309979e-01 -9.65509534e-01 2.88105696e-01
2.47729957e-01 -2.18731776e-01 -2.36316130e-01 1.84081280e+00
2.88820677e-02 -5.77963255e-02 -3.29225212e-01 1.11890614e+00
1.09798670e+00 5.20061910e-01 -3.73741835e-02 -5.07828742e-02
1.52392852e+00 -1.34212589e+00 -7.06491172e-01 -4.42131609e-01
3.54416192e-01 -6.86162651e-01 1.10224199e+00 5.35465717e-01
-1.33064139e+00 -4.41900611e-01 -1.08817792e+00 -6.01029515e-01
-2.73846626e-01 1.94118574e-01 9.26728398e-02 5.41407049e-01
-5.18761575e-01 5.32599688e-01 -3.21245700e-01 -2.30051965e-01
6.20763600e-01 -1.70232654e-01 -1.99150983e-02 1.09031558e-01
-1.63302827e+00 1.36673212e+00 1.72104523e-01 7.11460710e-02
-1.11321783e+00 -5.24357200e-01 -8.78741741e-01 3.56909007e-01
-9.17169750e-02 -8.17496598e-01 1.15515351e+00 -1.14488971e+00
-1.25984192e+00 1.19573998e+00 1.37548149e-01 -7.89770365e-01
7.46996760e-01 -4.03950006e-01 -3.58155131e-01 -1.34882241e-01
8.55913982e-02 -2.38851998e-02 1.16421163e+00 -1.06340635e+00
-3.28819036e-01 -3.84725213e-01 -1.10528775e-01 -4.72356193e-02
-7.76973128e-01 3.86992604e-01 2.86531717e-01 -4.56919938e-01
-7.51434565e-01 -5.22119939e-01 1.88238993e-01 -6.09931409e-01
-5.62666059e-01 -3.29450667e-01 7.67392933e-01 -1.04511762e+00
1.47357416e+00 -1.85963142e+00 3.13939214e-01 -4.13032442e-01
6.44060194e-01 4.12369937e-01 9.24764499e-02 3.46997023e-01
1.31335035e-01 -3.33446831e-01 -2.76930392e-01 -6.20392978e-01
4.15932149e-01 -5.72591722e-01 -2.16949984e-01 6.45582736e-01
1.49370193e-01 9.24520671e-01 -9.71696913e-01 -5.16181052e-01
1.85412839e-01 5.05778372e-01 -4.81257647e-01 4.91032302e-01
-1.07400894e-01 5.52701093e-02 -6.09586723e-02 4.72808152e-01
5.50426245e-01 -2.86119580e-01 -4.98677045e-03 7.74343237e-02
2.87490226e-02 5.50810695e-01 -4.71707940e-01 1.08927333e+00
-7.78836310e-01 9.96362507e-01 -6.90450668e-02 -6.71740770e-01
1.07696438e+00 3.65269721e-01 3.12868208e-01 -7.10041702e-01
5.23285508e-01 3.04866254e-01 1.89411446e-01 -6.51401103e-01
6.47282839e-01 -7.02072620e-01 -4.53986973e-01 3.09795171e-01
4.06164303e-02 2.55883951e-02 -1.44357041e-01 -3.53550538e-02
1.22017634e+00 -4.05533284e-01 5.08772850e-01 -1.74647301e-01
8.47750366e-01 -5.49531639e-01 3.95152479e-01 3.74103248e-01
-5.24215281e-01 6.06839895e-01 7.65844226e-01 -4.42312896e-01
-1.22266710e+00 -7.11930692e-01 -2.73412794e-01 1.23907268e+00
-2.75272399e-01 -1.22954249e-01 -6.48462117e-01 -5.37778139e-01
1.59331977e-01 1.20192051e+00 -8.32433343e-01 -4.89612311e-01
-5.00539482e-01 -7.17380404e-01 5.80609143e-01 2.47505620e-01
6.79340959e-01 -1.48495030e+00 -1.21970809e+00 3.14656347e-01
-5.03999412e-01 -1.12287283e+00 -5.08906126e-01 1.33895814e-01
-3.65485311e-01 -8.36565673e-01 -6.08111858e-01 -5.90803623e-01
-8.56041014e-02 -1.33669198e-01 1.45400262e+00 2.66827829e-02
-1.88022614e-01 -1.99228227e-02 -4.08972323e-01 -3.24536324e-01
-4.40005124e-01 3.14682484e-01 -1.65826067e-01 2.47939721e-01
7.15703964e-01 -6.83198690e-01 -4.42368209e-01 7.42048919e-02
-4.79357660e-01 1.18477577e-02 1.93528131e-01 1.20255768e+00
-3.98792922e-01 2.31017219e-03 9.34071779e-01 -8.01896453e-01
1.02703738e+00 -7.31086791e-01 -4.76879962e-02 -5.98358065e-02
-3.55724752e-01 -2.57952601e-01 9.16535795e-01 -2.35146552e-01
-8.14705729e-01 -2.93172032e-01 -1.41616836e-02 -4.18951273e-01
5.09295426e-02 5.60093403e-01 -2.71067709e-01 3.16642284e-01
8.68326485e-01 -4.74766456e-02 -3.06383520e-01 -2.59047985e-01
2.95578569e-01 1.06196499e+00 7.78920949e-01 -2.72175878e-01
6.25225544e-01 -6.54073507e-02 -8.47901553e-02 -7.98179328e-01
-1.25527561e+00 -5.08527339e-01 -6.17508411e-01 -3.66358697e-01
8.26488912e-01 -9.41726923e-01 -9.53581095e-01 7.18027532e-01
-1.50303876e+00 -4.18510824e-01 1.57606736e-01 1.13041796e-01
-7.43749440e-01 1.92504197e-01 -6.63662374e-01 -1.08706796e+00
-9.09099102e-01 -7.14658022e-01 9.01068985e-01 2.13396162e-01
-5.12694955e-01 -1.12967408e+00 3.51527363e-01 9.77966905e-01
2.79070884e-01 3.97907585e-01 9.92423058e-01 -6.74155593e-01
1.47055745e-01 -2.05434427e-01 -4.34333950e-01 4.75991517e-01
-3.02249014e-01 -4.84967560e-01 -1.33326793e+00 -3.16606343e-01
2.95071334e-01 -1.21622968e+00 1.31970608e+00 2.03027949e-01
9.90262628e-01 -6.35635853e-01 3.06848168e-01 2.78972536e-01
1.14297771e+00 -3.13371748e-01 7.80719817e-01 2.88372576e-01
8.70713949e-01 6.25192046e-01 1.90190539e-01 9.80895519e-01
7.35335827e-01 6.56587183e-01 6.70925379e-01 7.42435316e-03
-1.63190998e-02 -2.74461955e-01 7.35447347e-01 5.95593750e-01
1.35646909e-01 -1.71386525e-01 -8.80042076e-01 7.36090302e-01
-2.04910660e+00 -1.55915701e+00 -4.16447282e-01 2.11388278e+00
9.50956762e-01 5.09918891e-02 7.08532810e-01 2.86599308e-01
5.92143774e-01 5.38774312e-01 -4.94088888e-01 -8.72781336e-01
-1.04076713e-01 -9.71670598e-02 -1.37789428e-01 7.24740803e-01
-1.30775070e+00 9.49523568e-01 5.05756569e+00 5.79099119e-01
-1.06462312e+00 3.91726434e-01 6.81903362e-01 -3.25345248e-01
-2.57268101e-02 -3.77444357e-01 -4.15757895e-01 6.21239305e-01
9.37150955e-01 -1.60195604e-01 5.40983677e-01 8.26491833e-01
4.63787794e-01 7.28109032e-02 -1.04737866e+00 1.00916445e+00
5.57753205e-01 -5.72760642e-01 -4.28282261e-01 -1.94466636e-01
7.12580323e-01 -5.14164492e-02 3.19033563e-01 7.56523192e-01
-5.58406636e-02 -1.49984050e+00 1.06897640e+00 5.42288959e-01
3.83317858e-01 -8.80264640e-01 1.00299692e+00 4.93758470e-01
-5.53639829e-01 -4.49307770e-01 -3.70056391e-01 -4.06827122e-01
2.06026435e-01 9.37415004e-01 -9.50531304e-01 -4.80218157e-02
3.55708510e-01 5.81272721e-01 -6.71566367e-01 9.36325610e-01
-4.12203133e-01 7.79782414e-01 2.60918707e-01 -4.00927961e-01
1.85378909e-01 1.60649896e-01 5.39510727e-01 1.55727422e+00
4.67499830e-02 -2.53929228e-01 7.18020229e-03 1.30489790e+00
-5.67905545e-01 1.40403047e-01 -3.23783308e-01 -4.87037078e-02
3.00942380e-02 1.60724878e+00 1.43114030e-02 -5.23668714e-02
-2.61071950e-01 1.20848489e+00 6.80391192e-01 -2.96602339e-01
-1.06036305e+00 -5.69788277e-01 3.06674451e-01 2.75624335e-01
1.57679707e-01 1.17238849e-01 -7.96157360e-01 -1.12593937e+00
-2.16405779e-01 -1.02145815e+00 3.57609063e-01 -6.65501535e-01
-1.47788894e+00 4.23833370e-01 -5.16329348e-01 -7.20358849e-01
-1.19650662e-01 -6.00959301e-01 -1.08536482e+00 8.55211020e-01
-1.31606364e+00 -1.50864601e+00 -3.91805351e-01 4.04469550e-01
5.82009256e-01 -2.14251667e-01 2.10209414e-01 3.89016509e-01
-7.10352361e-01 7.39367664e-01 -1.81046709e-01 2.71359354e-01
6.94979548e-01 -1.51476693e+00 3.21410447e-02 6.47205472e-01
-3.93165469e-01 2.18766093e-01 1.12706578e+00 -4.14820075e-01
-7.61031747e-01 -1.09930003e+00 1.73413873e+00 -6.48397148e-01
8.34793985e-01 -4.86533374e-01 -9.68841314e-01 2.35939503e-01
6.49219394e-01 -1.99202925e-01 4.65050489e-01 4.97131795e-01
-4.92081106e-01 9.29662660e-02 -1.23986888e+00 2.45897248e-01
4.96306986e-01 -6.78714633e-01 -5.87826252e-01 3.54931474e-01
3.21555942e-01 -1.85462579e-01 -6.56777382e-01 4.46005426e-02
5.64820290e-01 -1.38455987e+00 6.71369195e-01 -8.29859495e-01
1.49613118e+00 7.64628872e-02 -1.58920698e-02 -1.37728429e+00
-8.21023941e-01 -4.31823313e-01 -3.93392116e-01 1.19976223e+00
8.82509500e-02 3.12349591e-02 4.72965091e-01 3.39020669e-01
4.29497622e-02 -6.88040257e-01 -9.70988512e-01 -2.59808540e-01
5.08982956e-01 -1.00238591e-01 2.09504530e-01 1.09081173e+00
6.35069370e-01 1.26272666e+00 -1.12089813e+00 -1.80954739e-01
8.11596990e-01 2.42037237e-01 6.43374622e-01 -1.15350378e+00
-3.43030751e-01 -9.71850634e-01 -4.21129525e-01 -5.00394940e-01
4.99391079e-01 -1.09792542e+00 1.61802813e-01 -1.36316466e+00
8.77562165e-01 3.68275404e-01 -4.42304552e-01 4.69044328e-01
-4.06197906e-01 1.81024611e-01 4.41534400e-01 -2.85847813e-01
-7.67145753e-01 1.04473817e+00 1.01004744e+00 -3.75378102e-01
-5.18041737e-02 -1.41289895e-02 -7.28285789e-01 6.46752357e-01
8.18757534e-01 -3.82015407e-01 -1.32132679e-01 -3.80147509e-02
5.13389945e-01 1.46459937e-01 9.06265199e-01 -1.00311434e+00
1.31345227e-01 -1.15740612e-01 3.09614152e-01 -5.70545018e-01
2.80613899e-01 -4.61796343e-01 -5.10161817e-01 5.26859462e-01
-5.33385873e-01 -3.66526008e-01 -2.18122005e-01 -3.65459621e-02
-7.28968307e-02 -4.41072792e-01 1.46422756e+00 -2.86840126e-02
1.25095155e-03 -7.53459260e-02 -3.00749987e-01 3.14346552e-01
6.90232396e-01 4.81577098e-01 -6.30242169e-01 -6.27519488e-01
-2.93979317e-01 3.17172617e-01 2.76159585e-01 4.07794476e-01
6.73555970e-01 -1.46529579e+00 -1.41754878e+00 -2.45453298e-01
9.19402391e-02 -4.92497802e-01 2.42260903e-01 1.06956267e+00
-1.44336343e-01 2.58595735e-01 -4.67019349e-01 -3.23105566e-02
-1.18459344e+00 7.36536324e-01 5.30022800e-01 -5.32721877e-01
-4.33046460e-01 5.80667317e-01 -8.24475214e-02 -3.54713827e-01
4.51994799e-02 2.82036096e-01 -4.85619813e-01 3.12362134e-01
6.33242369e-01 9.55627382e-01 -1.65527053e-02 -8.70283067e-01
-2.10638121e-01 -1.22028351e-01 2.54655117e-03 9.51045677e-02
1.28979313e+00 1.12000547e-01 -3.70732456e-01 7.16099024e-01
1.31153464e+00 -1.81795210e-01 -7.36810446e-01 -3.20602953e-01
1.96465179e-02 -2.37760708e-01 2.04569027e-01 -9.39095378e-01
-7.69130349e-01 1.33976054e+00 2.20594496e-01 1.54741898e-01
1.01027131e+00 -1.28284395e-01 1.01495063e+00 2.51830786e-01
2.36823022e-01 -1.29012382e+00 6.11828744e-01 1.00680101e+00
1.23320723e+00 -1.43935132e+00 -1.34818733e-01 3.99915099e-01
-8.72536063e-01 1.16000021e+00 7.55572379e-01 -1.73686936e-01
2.43421465e-01 -1.70681909e-01 -3.24539185e-01 -2.76708901e-01
-9.41576958e-01 -1.13350421e-01 4.32503879e-01 1.85719475e-01
9.28190768e-01 3.02369148e-01 -5.71161926e-01 1.15475380e+00
-4.43822533e-01 3.74856554e-02 6.55062377e-01 7.00757504e-02
-8.20372999e-01 -3.52881163e-01 -3.70226711e-01 3.56465101e-01
-3.77268940e-01 -1.44623026e-01 -7.67334461e-01 1.55589983e-01
2.60000616e-01 1.13440824e+00 -3.26706260e-01 -7.20757365e-01
1.54271647e-01 1.88907146e-01 4.26661491e-01 -3.56799066e-01
-9.96803045e-01 -3.58599454e-01 3.47841769e-01 -1.26008466e-01
-3.59250535e-03 -7.62367427e-01 -8.74327838e-01 -6.48631811e-01
-7.90638849e-02 -8.73515978e-02 5.29663451e-02 9.14099991e-01
-1.98706180e-01 5.51645815e-01 1.12371445e+00 -6.88258886e-01
-8.16992283e-01 -1.45995080e+00 -6.99275076e-01 7.07565069e-01
5.04925132e-01 -4.64267254e-01 -2.32019231e-01 -1.20409414e-01] | [8.873779296875, 11.04246997833252] |
314dfe7d-3e64-4666-8bc4-f148d6964560 | hologen-an-open-source-toolbox-for-high-speed | 2008.12214 | null | https://arxiv.org/abs/2008.12214v2 | https://arxiv.org/pdf/2008.12214v2.pdf | HoloGen: An open source toolbox for high-speed hologram generation | The rise of mixed reality systems such as Microsoft HoloLens has prompted an increase in interest in the fields of 2D and 3D holography. Already applied in fields including telecommunications, imaging, projection, lithography, beam shaping and optical tweezing, Computer Generated Holography (CGH) offers an exciting approach to a wide range of light shaping problems. The numerical processing required to generate a hologram is high and requires significant domain expertise. This has historically slowed the adoption of holographic techniques in emerging fields. In this paper we present HoloGen, an open-source Cuda C and C ++ framework for computer generated holography. HoloGen unites, for the first time, a wide array of existing hologram generation algorithms with state of the art performance while attempting to remain intuitive and easy to use. This is enabled by a C # and Windows Presentation Framework (WPF) graphical user interface (GUI). A novel reflection based parameter hierarchy is used to ensure ease of modification. Extensive use of C ++ templates based on the Standard Template Library (STL), compile time flexibility is preserved while maintaining runtime performance. The current release of HoloGen unites implementations of well known generation algorithms including Gerchberg-Saxton (GS), Liu-Taghizadeh (LT), Direct Search (DS), Simulated Annealing (SA) and One-Step Phase-Retrieval (OSPR) with less known specialist variants including Weighted GS and Adaptive OSPR. Benchmarking results are presented for several key algorithms. The software is freely available under an MIT license. | ['Timothy D. Wilkinson', 'George S. D. Gordon', 'Andrew Kadis', 'Peter J. Christopher'] | 2020-08-24 | null | null | null | null | ['3d-holography'] | ['computer-vision'] | [ 3.71115297e-01 -1.47298992e-01 7.57095218e-01 -2.50375301e-01
-4.73811239e-01 -3.61331999e-01 7.70539701e-01 -2.89088398e-01
-4.46918011e-01 9.80142236e-01 2.90285498e-01 -2.81989247e-01
-4.61066812e-01 -1.21360338e+00 -3.17601025e-01 -7.89946020e-01
2.04737470e-01 9.61543798e-01 6.10571921e-01 -2.35006511e-01
6.37867749e-01 7.58992851e-01 -1.96248364e+00 1.50352344e-02
4.18823898e-01 7.23091245e-01 6.16202354e-01 9.41062868e-01
5.08513534e-03 3.88402075e-01 -3.63219172e-01 -1.85987160e-01
3.79755676e-01 -4.71574366e-01 -6.69333398e-01 -1.76501811e-01
1.72635671e-02 -2.63420254e-01 -2.64378399e-01 6.61126137e-01
1.11330581e+00 1.36344925e-01 3.82732600e-02 -6.26096308e-01
-2.83524901e-01 8.43621492e-02 -1.99948493e-02 -1.03437260e-01
8.74535561e-01 2.43234321e-01 3.87321472e-01 -8.87326837e-01
1.10589051e+00 9.22147989e-01 5.85734010e-01 4.78017032e-01
-7.56759942e-01 -6.31523013e-01 -1.19333303e+00 -1.17388442e-01
-1.14060390e+00 -3.97345901e-01 3.41241747e-01 -2.51310080e-01
1.64811957e+00 6.71354473e-01 8.99152040e-01 2.39603236e-01
5.29465079e-01 5.79581708e-02 1.62669480e+00 -8.47455919e-01
4.55306828e-01 -5.87249249e-02 1.42206937e-01 5.85635662e-01
1.56296849e-01 4.99796718e-01 -9.15327370e-01 -2.18622923e-01
7.43611395e-01 -2.21489533e-03 -2.29368314e-01 -3.59906912e-01
-1.03448009e+00 4.96297628e-01 -2.46938188e-02 9.86487269e-02
-1.58922344e-01 1.17825709e-01 -6.93894178e-02 2.37946585e-01
1.07699819e-01 8.11849892e-01 1.66453585e-01 -6.45856142e-01
-8.32007110e-01 7.11005747e-01 9.13571239e-01 9.98693526e-01
1.01872492e+00 -1.30033687e-01 7.21364394e-02 6.12842143e-01
2.52342701e-01 5.65491080e-01 5.75707138e-01 -7.62589335e-01
-5.67158032e-03 4.67373580e-01 2.52907813e-01 -4.12214875e-01
-3.48862290e-01 -2.02716082e-01 -4.71905768e-01 4.63450849e-01
-2.87839882e-02 8.00160989e-02 -9.37247992e-01 8.59424889e-01
6.70047522e-01 1.90749601e-01 -1.49375675e-02 7.64680028e-01
9.04817164e-01 6.62532568e-01 -6.53089643e-01 -3.21145147e-01
1.00509989e+00 -5.08409083e-01 -5.64444125e-01 7.30672702e-02
2.63893843e-01 -1.49953079e+00 6.84075654e-01 5.91265857e-01
-1.46789420e+00 -2.05197688e-02 -1.17606568e+00 -2.07131326e-01
-4.81013626e-01 -4.38378751e-01 7.74305403e-01 1.07328749e+00
-1.10821283e+00 7.14519262e-01 -8.79215896e-01 -5.95651567e-01
1.22844286e-01 1.00008583e+00 -2.43164256e-01 -9.04568955e-02
-5.38075089e-01 6.87176406e-01 1.84770733e-01 -3.11423838e-01
-2.25600451e-01 -8.19282770e-01 -3.46686423e-01 -3.07603598e-01
5.11333533e-02 -8.43948662e-01 1.06085789e+00 2.70809054e-01
-2.17087340e+00 1.15426075e+00 9.92238224e-02 -5.03995717e-01
3.12266290e-01 -1.88715216e-02 -1.65525675e-01 9.43823904e-02
-1.40304551e-01 3.16726446e-01 4.31032985e-01 -8.28869224e-01
-3.24677110e-01 -4.24651444e-01 -4.37809378e-01 1.54167548e-01
1.84502706e-01 3.31336051e-01 -1.80440828e-01 -7.43867010e-02
4.36430484e-01 -7.95122743e-01 3.20612565e-02 -4.33384061e-01
-1.37932688e-01 2.65153259e-01 1.02292490e+00 -1.44206226e-01
1.20151472e+00 -1.62812090e+00 -1.76554278e-01 1.50285542e-01
1.75154172e-02 4.22658145e-01 5.08163989e-01 1.23776674e+00
2.71082789e-01 -5.13872683e-01 -9.67507958e-02 -2.23112747e-01
1.25053704e-01 2.95432508e-01 -3.75270456e-01 6.48545086e-01
-6.67337358e-01 7.25678146e-01 -7.44613945e-01 -1.30320653e-01
7.80400634e-01 3.74129504e-01 -8.48048270e-01 2.98579633e-01
1.43943885e-02 2.96777457e-01 -1.50264695e-01 6.09734654e-01
8.56169164e-01 -2.34536573e-01 -2.93092318e-02 9.04702619e-02
-9.85804439e-01 6.01541162e-01 -1.51137149e+00 1.50729072e+00
-4.78242874e-01 3.31785500e-01 -2.72928309e-02 -2.92524219e-01
1.23709273e+00 1.72591880e-01 2.49406919e-01 -8.88352871e-01
-3.36412713e-03 6.46792948e-01 -2.67324120e-01 -4.13721234e-01
7.87733972e-01 -4.10817325e-01 3.08830500e-01 8.32573831e-01
7.43225217e-02 -8.80439222e-01 9.45580658e-03 2.27524322e-02
1.36237848e+00 3.50063056e-01 5.56726575e-01 -3.65546286e-01
1.89089507e-01 1.81625411e-01 1.26331104e-02 8.32681656e-01
3.17832470e-01 1.00436985e+00 -4.59828794e-01 -5.02606332e-01
-1.32036936e+00 -1.00563061e+00 -3.63849938e-01 2.92884737e-01
3.99205923e-01 -6.64149821e-01 -3.71854782e-01 6.53328180e-01
-3.19445580e-01 5.78972340e-01 3.37661654e-02 2.49391824e-01
-8.08951139e-01 -9.44804907e-01 1.11008920e-01 -1.75260663e-01
6.91337287e-01 -1.34749389e+00 -1.27520549e+00 5.23332298e-01
7.57897973e-01 -9.37733293e-01 3.25341105e-01 1.01166125e-02
-8.53834391e-01 -6.85378253e-01 -4.20813620e-01 -3.97384882e-01
2.49915883e-01 2.73018897e-01 1.14907634e+00 1.87448077e-02
-9.10756290e-01 6.21860445e-01 -3.69445622e-01 -3.86157721e-01
-2.43747249e-01 -2.77883083e-01 -1.02132574e-01 -5.59714377e-01
3.65042537e-01 -1.01275170e+00 -7.46490180e-01 1.91516146e-01
-8.40636075e-01 3.88548076e-01 3.41196179e-01 7.90691078e-01
5.80510259e-01 -1.90425575e-01 -5.84067963e-02 -9.17975783e-01
6.07224047e-01 -1.38065796e-02 -1.07531643e+00 -3.03181950e-02
-7.89423585e-01 -1.97071612e-01 3.64003450e-01 2.36970708e-01
-1.10723829e+00 6.97173402e-02 -4.11185145e-01 1.20184191e-01
-1.65140867e-01 2.43971795e-01 2.01560318e-01 -6.95513785e-01
5.12465596e-01 3.34278554e-01 1.13901021e-02 -5.69177151e-01
-7.95935690e-02 9.01696444e-01 6.47955775e-01 -3.50653082e-01
6.56620264e-01 6.17705107e-01 2.61677563e-01 -1.17542982e+00
-1.32908329e-01 -7.15519369e-01 -6.84884116e-02 -1.78236291e-01
5.96370637e-01 -4.04491633e-01 -8.46892536e-01 7.21475899e-01
-9.65976954e-01 -1.83952034e-01 -3.50776345e-01 5.02533138e-01
-5.51651657e-01 2.44085863e-01 -2.83305824e-01 -9.95274365e-01
-5.74128866e-01 -1.15739977e+00 1.28851128e+00 4.88111377e-01
-2.81483799e-01 -6.14134073e-01 5.91355801e-01 3.75642836e-01
7.04011559e-01 2.01024860e-01 6.46844387e-01 -3.38363200e-02
-1.32304549e+00 -2.51064479e-01 1.20736971e-01 -3.50910932e-01
-6.39307797e-02 -8.65471810e-02 -1.01343966e+00 -7.47435838e-02
2.12272614e-01 -1.63958624e-01 2.93748915e-01 3.24330091e-01
4.98795688e-01 -1.08087830e-01 -2.65223205e-01 9.13011611e-01
1.91276872e+00 3.44291538e-01 1.21776879e+00 7.10129082e-01
2.31532395e-01 1.42958820e-01 3.36279839e-01 8.94739985e-01
-2.37357407e-03 9.49792147e-01 2.04485774e-01 4.48692828e-01
-1.73002630e-01 -6.92540258e-02 -6.50860784e-06 8.78571332e-01
-3.20460260e-01 -1.67977840e-01 -1.06533480e+00 5.72038032e-02
-1.57273519e+00 -1.05019784e+00 -4.35038716e-01 2.69272137e+00
6.51420891e-01 -1.90744221e-01 -1.39353514e-01 2.94745266e-01
4.04173970e-01 -1.22733369e-01 8.25392008e-02 -4.52097237e-01
-3.29987071e-02 1.17195809e+00 7.09037900e-01 8.80657375e-01
-4.74978030e-01 9.49024618e-01 5.83348703e+00 7.04147696e-01
-1.17654121e+00 1.49462536e-01 -2.74107844e-01 -1.92622408e-01
-3.05160671e-01 4.14405257e-01 -1.09254646e+00 4.60395068e-01
7.05440938e-01 -4.84739900e-01 8.25759590e-01 4.38573599e-01
1.93765625e-01 -5.51638067e-01 -5.22456765e-01 1.37851048e+00
-1.21861570e-01 -2.05092454e+00 -2.41264299e-01 1.19361453e-01
7.27320075e-01 2.69974381e-01 -2.54694402e-01 -2.51070887e-01
1.56884909e-01 -7.80264258e-01 3.65981162e-01 3.22306335e-01
7.42485344e-01 -3.76445711e-01 4.00441349e-01 1.37240916e-01
-7.46171594e-01 3.18507522e-01 -5.46202719e-01 -4.69359219e-01
5.76603234e-01 6.41376674e-01 -1.12321293e+00 5.51924706e-01
4.90492463e-01 1.30447745e-01 -1.41959906e-01 1.53041303e+00
2.17896909e-01 4.25515294e-01 -6.49383426e-01 -2.56241918e-01
2.56404728e-01 -5.83155930e-01 7.30338812e-01 8.78193676e-01
7.17737138e-01 4.11156833e-01 -3.50329310e-01 6.16231799e-01
3.84908766e-01 -8.96733776e-02 -5.26588500e-01 1.44597575e-01
4.02055323e-01 1.14433944e+00 -8.85303319e-01 -1.21726103e-01
-6.47652075e-02 6.32851481e-01 -3.30519050e-01 2.79796831e-02
-2.90093392e-01 -5.70400059e-01 2.66619865e-02 6.72552824e-01
1.18578017e-01 -5.31672120e-01 -4.08354312e-01 -7.68893659e-01
-3.45495403e-01 -6.04599714e-01 8.05057436e-02 -1.08991754e+00
-6.85596824e-01 4.93604839e-01 1.06378272e-01 -1.01762092e+00
-4.49248880e-01 -5.96441984e-01 -5.02872229e-01 9.15868104e-01
-1.16068494e+00 -7.92322636e-01 -5.96759796e-01 4.02390569e-01
2.93069780e-01 -2.79805779e-01 9.12221730e-01 3.86788785e-01
9.68422294e-02 -9.63415578e-02 4.18607086e-01 -7.27980793e-01
6.06562972e-01 -1.16659141e+00 2.85175920e-01 7.29619384e-01
-1.77145928e-01 8.71497810e-01 1.15859389e+00 -7.62916267e-01
-2.16101384e+00 -5.82255065e-01 1.06346154e+00 -9.13754776e-02
4.33379531e-01 -3.69169891e-01 -4.91176993e-01 3.62690985e-01
3.58468145e-01 -1.80162564e-01 4.68016267e-01 -4.24102783e-01
3.30758542e-01 -1.88738480e-01 -1.28165126e+00 2.72025585e-01
1.01572740e+00 -4.08936113e-01 -3.31539840e-01 6.51175499e-01
1.08023524e-01 -9.44950104e-01 -5.12131274e-01 1.61270827e-01
6.35730267e-01 -1.87471426e+00 9.24143016e-01 2.92246342e-01
2.43997216e-01 -2.92457074e-01 -2.08106130e-01 -6.79513514e-01
4.36556004e-02 -1.71223128e+00 2.10213527e-01 9.20556724e-01
-1.41205817e-01 -1.07205224e+00 9.67954576e-01 7.73145616e-01
-6.00923955e-01 -6.18580937e-01 -1.07709432e+00 -7.34899282e-01
-5.85167646e-01 -2.25574419e-01 8.81378651e-01 3.97567809e-01
9.08264294e-02 3.53100181e-01 -3.71639282e-01 9.74937156e-02
6.78301871e-01 3.72533917e-01 1.01862884e+00 -1.12464142e+00
-7.75300741e-01 -3.25741284e-02 -6.81325138e-01 -8.89022529e-01
-6.31294310e-01 -7.05789804e-01 -2.45532826e-01 -1.47414875e+00
5.58363199e-02 -7.15851963e-01 5.49656153e-01 -6.20925426e-02
5.43444395e-01 5.57339311e-01 -3.75938573e-04 2.98637927e-01
-3.35528851e-01 3.42321545e-01 1.07403278e+00 6.13958180e-01
-4.34167087e-01 6.96617588e-02 6.83582276e-02 3.48218352e-01
6.00865662e-01 -4.70598102e-01 -3.39442521e-01 -3.21010441e-01
7.71674812e-01 7.01861084e-02 4.51124787e-01 -1.34692776e+00
5.78425109e-01 2.22642511e-01 -4.68141250e-02 -8.30241382e-01
6.27582312e-01 -3.74900192e-01 9.28848624e-01 5.47399104e-01
4.71422255e-01 6.92641586e-02 -1.13337398e-01 6.77082762e-02
-1.08947240e-01 -6.17187023e-01 9.82050419e-01 -5.38306296e-01
-5.99574447e-01 1.55740589e-01 -1.88677743e-01 -3.29353094e-01
9.66310680e-01 -7.23464608e-01 -6.57355905e-01 -3.33129734e-01
-2.64170140e-01 -3.22715431e-01 9.63598073e-01 -3.72969598e-01
8.20005715e-01 -9.70896363e-01 -1.76429868e-01 5.24157465e-01
-2.20898360e-01 -8.13739523e-02 1.66288018e-01 4.78658795e-01
-1.39811862e+00 5.71369588e-01 -2.68207759e-01 -4.47980821e-01
-1.39818621e+00 5.37220649e-02 1.63678303e-01 -1.72061801e-01
-8.57968092e-01 9.06660974e-01 -3.03489834e-01 -3.09442520e-01
-3.11795533e-01 3.78546529e-02 2.92978197e-01 -4.71615374e-01
6.50398910e-01 8.20059776e-01 5.12924254e-01 -2.05241010e-01
-3.35220277e-01 8.09009016e-01 1.61177084e-01 -5.29656589e-01
1.91810799e+00 5.41199706e-02 -4.01503712e-01 -5.10074235e-02
8.72000098e-01 1.43734336e-01 -6.59273982e-01 3.47752750e-01
-2.50128865e-01 -7.19596386e-01 1.85503200e-01 -7.89762080e-01
-3.46133918e-01 8.81410837e-01 6.99759364e-01 1.09504431e-01
9.80693996e-01 -1.68560892e-01 8.75829577e-01 4.14591879e-01
9.47132409e-01 -9.81121778e-01 -3.94080043e-01 5.32791615e-01
5.64285576e-01 -5.29899418e-01 4.56493407e-01 -4.14719045e-01
6.98072836e-02 1.24999905e+00 9.60032791e-02 -2.56376803e-01
5.66303134e-01 8.51823866e-01 -3.26454639e-01 -5.47442734e-01
-4.95515138e-01 -1.07683659e-01 -3.15421402e-01 4.99446183e-01
5.46176553e-01 -2.15074733e-01 -6.48945391e-01 -9.84006096e-04
-6.89201236e-01 4.92077887e-01 7.50261486e-01 1.47457254e+00
-6.82225883e-01 -1.64763844e+00 -6.08368278e-01 3.62858862e-01
-3.58291358e-01 -1.55551791e-01 -1.83731660e-01 7.08636224e-01
-8.52739364e-02 4.55640584e-01 -8.47657695e-02 -1.56846344e-01
-4.15800475e-02 -1.95056841e-01 9.98778343e-01 -8.58111441e-01
-8.34103286e-01 2.56163359e-01 1.55270517e-01 -5.67119181e-01
-5.50477862e-01 -5.26148915e-01 -1.14897132e+00 -4.22837943e-01
-3.11027914e-01 1.63651586e-01 8.70455861e-01 8.11422348e-01
5.83250284e-01 -1.67951763e-01 1.92202285e-01 -1.06226850e+00
-3.39440286e-01 -5.05451322e-01 -7.36747622e-01 -1.04115292e-01
-1.44940570e-01 -5.72293878e-01 -7.31699094e-02 -1.28429577e-01] | [9.614103317260742, -2.629831075668335] |
c9ac32d3-b599-4942-923e-03b2d965737e | meta-embedding-sentence-representation-for | null | null | https://aclanthology.org/R19-1055 | https://aclanthology.org/R19-1055.pdf | Meta-Embedding Sentence Representation for Textual Similarity | Word embedding models are now widely used in most NLP applications. Despite their effectiveness, there is no clear evidence about the choice of the most appropriate model. It often depends on the nature of the task and on the quality and size of the used data sets. This remains true for bottom-up sentence embedding models. However, no straightforward investigation has been conducted so far. In this paper, we propose a systematic study of the impact of the main word embedding models on sentence representation. By contrasting in-domain and pre-trained embedding models, we show under which conditions they can be jointly used for bottom-up sentence embeddings. Finally, we propose the first bottom-up meta-embedding representation at the sentence level for textual similarity. Significant improvements are observed in several tasks including question-to-question similarity, paraphrasing and next utterance ranking. | ['Hern', 'Amir Hazem', 'Nicolas ez'] | 2019-09-01 | null | null | null | ranlp-2019-9 | ['question-similarity'] | ['natural-language-processing'] | [ 1.25403047e-01 1.73937287e-02 -2.87633836e-01 -2.63341784e-01
-7.46819377e-01 -4.47401613e-01 8.13836992e-01 9.00836229e-01
-8.39513421e-01 4.03472543e-01 7.31167138e-01 -5.01595199e-01
-2.72229254e-01 -6.48563921e-01 -3.01040858e-01 -3.29179555e-01
2.54117250e-01 3.19651246e-01 3.33723962e-01 -5.36879778e-01
5.10913968e-01 2.55737692e-01 -1.57880890e+00 2.32648611e-01
8.94795716e-01 6.30990744e-01 4.02848542e-01 8.18384290e-01
-4.21807170e-01 4.30103421e-01 -5.40689290e-01 -5.79451263e-01
-1.77426025e-01 -5.67944646e-01 -8.91712725e-01 -9.76813138e-02
4.54416066e-01 -1.13728037e-03 -4.05550599e-01 9.34388101e-01
7.42683649e-01 2.54698783e-01 7.34981179e-01 -6.69032037e-01
-8.77450764e-01 5.61920226e-01 5.08358777e-02 6.01449192e-01
5.99886835e-01 -1.65883765e-01 1.59908223e+00 -1.14145112e+00
8.41055155e-01 1.09090960e+00 5.10648906e-01 4.49840605e-01
-1.22563267e+00 -2.83961538e-02 6.43160418e-02 4.96403068e-01
-9.50588822e-01 -4.52645272e-01 6.84568465e-01 -2.26685867e-01
1.35626388e+00 3.30756426e-01 2.85832137e-01 1.08796716e+00
3.04140925e-01 3.64639848e-01 1.10416472e+00 -9.23467457e-01
1.06368482e-01 6.25707626e-01 5.93893409e-01 2.70293891e-01
3.35871458e-01 -2.83998221e-01 -5.88115692e-01 -2.76390523e-01
1.28657341e-01 -2.78242141e-01 -3.84642571e-01 -4.51305687e-01
-8.84987891e-01 1.17204916e+00 1.46668464e-01 9.93127644e-01
-2.92002320e-01 -1.25133678e-01 7.79306948e-01 5.58975458e-01
7.61751771e-01 7.83283591e-01 -5.47322392e-01 -3.67575139e-01
-7.60367513e-01 4.74812388e-01 7.12198794e-01 4.95304972e-01
4.13291812e-01 -3.39346230e-01 -4.12821233e-01 1.23058498e+00
1.94601715e-01 3.88097428e-02 8.24773252e-01 -5.08165598e-01
4.74371254e-01 3.81791770e-01 1.48019031e-01 -1.17467332e+00
-3.16238195e-01 -2.61555046e-01 -2.13724691e-02 -2.36484542e-01
2.71999061e-01 -2.65358146e-02 -2.83484906e-01 1.59219468e+00
2.14371353e-01 -1.83921680e-01 2.19974369e-01 6.72138751e-01
8.30429077e-01 6.53738022e-01 1.21288383e-02 -9.25275981e-02
1.82072353e+00 -1.05920088e+00 -9.37621534e-01 -4.52777117e-01
8.70451927e-01 -9.92612422e-01 1.30563593e+00 -8.25208277e-02
-1.13477075e+00 -6.64172590e-01 -1.02119970e+00 -3.25314134e-01
-6.86577260e-01 -1.58301324e-01 3.35670680e-01 6.76566899e-01
-1.02128339e+00 7.15449393e-01 -4.78267729e-01 -7.90178061e-01
-1.62055939e-01 1.14609614e-01 -4.15267140e-01 -2.48448238e-01
-1.44177568e+00 1.53645658e+00 3.33640814e-01 -1.82497501e-01
-1.80905879e-01 -6.81729972e-01 -9.25405741e-01 2.33061269e-01
5.85623272e-02 -7.03396440e-01 1.16661072e+00 -5.01714110e-01
-1.39812124e+00 8.42898250e-01 -4.04133350e-01 -5.27712286e-01
-2.03660019e-02 -2.83260852e-01 -4.82070476e-01 1.37297630e-01
2.33916398e-02 4.49316204e-01 6.66005969e-01 -9.05216753e-01
-3.65122169e-01 -3.40479761e-01 3.23604882e-01 2.87157506e-01
-8.98729086e-01 4.18423623e-01 -1.61824763e-01 -7.29755402e-01
-9.96664613e-02 -7.28710532e-01 -6.61127567e-02 -7.01482520e-02
2.09732965e-01 -7.10438550e-01 5.23198783e-01 -7.10999787e-01
1.53781319e+00 -2.29148984e+00 1.32453278e-01 -4.31663036e-01
-2.02657685e-01 3.40857208e-01 -4.20714259e-01 1.24972820e+00
-1.54742628e-01 1.01077698e-01 -9.11600441e-02 -5.13747573e-01
3.96431148e-01 2.37337217e-01 -5.29135644e-01 2.71631747e-01
3.46292317e-01 9.27028179e-01 -1.01630545e+00 -5.98312020e-01
2.38018230e-01 4.75699186e-01 -5.40172756e-01 3.33981097e-01
1.05807766e-01 -1.46648318e-01 -4.14889544e-01 7.27739409e-02
3.37542415e-01 2.42922232e-02 2.23947853e-01 -6.37119636e-02
-1.58040717e-01 9.99850810e-01 -7.36327350e-01 1.70340621e+00
-8.58243823e-01 8.27201426e-01 -2.69095063e-01 -9.97940302e-01
7.67767906e-01 5.37348807e-01 1.06887683e-01 -7.51555681e-01
4.23079636e-03 2.88402617e-01 3.27731669e-01 -7.73254633e-01
1.03770530e+00 -4.98748153e-01 -1.00679301e-01 6.19226813e-01
3.24723423e-01 -6.60154894e-02 5.02381623e-01 1.28110602e-01
1.08450687e+00 -1.80074632e-01 5.32878578e-01 -3.95526379e-01
5.92330515e-01 5.40077910e-02 -6.61407188e-02 6.30123019e-01
-2.33360931e-01 8.07284176e-01 4.36249405e-01 -8.33211541e-02
-9.49509263e-01 -8.69637012e-01 -5.83146572e-01 1.22945857e+00
-3.29675786e-02 -5.49927652e-01 -5.99308610e-01 -6.19422913e-01
9.18746293e-02 9.16362643e-01 -7.07754314e-01 -3.36259425e-01
-5.14953136e-01 -5.75945914e-01 3.26123118e-01 3.93112659e-01
-2.99004912e-01 -1.17845178e+00 -4.58216250e-01 4.02146637e-01
-1.44249722e-01 -1.18810952e+00 -4.50234622e-01 2.11424395e-01
-1.20418549e+00 -7.81855345e-01 -5.97543001e-01 -8.09887111e-01
3.92062932e-01 4.13832277e-01 1.18126142e+00 9.35591161e-02
1.10006474e-01 4.29596603e-01 -8.54336321e-01 -2.95552671e-01
-5.37783146e-01 2.60846436e-01 5.85101955e-02 -2.63229936e-01
6.47532403e-01 -4.76244658e-01 -4.18818921e-01 -2.82808810e-01
-9.52822149e-01 -5.97922683e-01 5.94021916e-01 9.18041766e-01
-2.18301099e-02 -4.01222527e-01 7.62200236e-01 -8.79452288e-01
1.44078326e+00 -3.30989808e-01 1.13288648e-01 3.78923416e-01
-7.54559815e-01 2.17774242e-01 6.05042338e-01 -2.90758938e-01
-5.85632443e-01 -6.80148184e-01 -6.83166325e-01 4.20243628e-02
-3.18046451e-01 6.76458001e-01 2.56354362e-01 1.89207047e-01
4.82439190e-01 1.54913366e-01 -3.52767063e-03 -5.79498947e-01
4.70398933e-01 8.71360540e-01 -1.16263404e-02 -2.87165612e-01
5.97563028e-01 7.07505122e-02 -5.29326081e-01 -1.15862703e+00
-8.08661938e-01 -6.64700329e-01 -6.15327418e-01 1.42427281e-01
9.67969835e-01 -5.52806735e-01 -3.12584564e-02 -2.50836909e-01
-1.49592042e+00 1.60044149e-01 -4.11445826e-01 4.53995198e-01
-3.11830074e-01 7.57822752e-01 -4.99067664e-01 -5.82357466e-01
-3.46371889e-01 -1.01433897e+00 8.30050170e-01 -1.97487786e-01
-7.22625315e-01 -1.35344195e+00 4.74774361e-01 4.11151201e-01
6.31400108e-01 -1.78489566e-01 1.34688652e+00 -9.83616173e-01
-1.52407000e-02 -4.58632976e-01 -6.51468486e-02 4.81926888e-01
2.26949543e-01 -1.26203552e-01 -9.86751914e-01 -1.08369395e-01
2.13401288e-01 -2.11670250e-01 9.01803792e-01 2.51799613e-01
7.31345534e-01 -1.65299445e-01 -1.28900275e-01 -1.40589541e-02
1.43161881e+00 -2.19301656e-01 5.43279052e-01 5.00751853e-01
1.80604830e-01 1.00003314e+00 5.81869602e-01 2.10255936e-01
2.42782459e-01 8.91430259e-01 1.58307150e-01 3.94616276e-01
-1.55021891e-01 -1.48200229e-01 4.81596649e-01 1.29345620e+00
3.83561432e-01 -3.61825109e-01 -5.77237546e-01 8.57857227e-01
-1.53323627e+00 -9.82511401e-01 -1.49992242e-01 2.17303610e+00
7.88546443e-01 2.53829658e-01 5.90638444e-03 3.63771677e-01
3.43888104e-01 6.98443651e-01 1.80885643e-01 -1.12195647e+00
6.66312054e-02 4.20215189e-01 4.92638275e-02 7.18414366e-01
-7.85486400e-01 6.47273302e-01 6.36578274e+00 8.58336091e-01
-8.28632176e-01 3.99935454e-01 2.71455534e-02 2.16303781e-01
-5.75426698e-01 2.55666189e-02 -6.66144967e-01 3.50242347e-01
1.08335495e+00 -2.29163885e-01 -1.01466306e-01 6.49921775e-01
2.60185421e-01 -3.42424144e-03 -1.19265795e+00 5.80352068e-01
3.47154886e-01 -1.14805186e+00 5.87801412e-02 -1.74609274e-01
4.06625748e-01 -9.10937116e-02 -2.36231815e-02 5.64121783e-01
-5.13378382e-01 -7.99839854e-01 3.93744677e-01 1.52749196e-01
1.53001368e-01 -6.27331674e-01 9.44976270e-01 3.75877172e-01
-7.71208644e-01 -4.09615003e-02 -6.51961565e-01 -1.82128966e-01
3.61010224e-01 4.80443418e-01 -7.50785410e-01 6.61010683e-01
2.71014035e-01 3.92989844e-01 -7.00909317e-01 6.41379356e-01
-2.87501186e-01 5.79542160e-01 6.96130618e-02 -5.38703501e-01
4.19106334e-01 -1.09052658e-01 5.42259514e-01 1.46588480e+00
2.46337935e-01 -2.17693433e-01 -3.01166862e-01 5.99210620e-01
7.74028972e-02 5.18881261e-01 -7.07112193e-01 -2.78292507e-01
4.41256970e-01 1.25524175e+00 -3.84865731e-01 -1.26685143e-01
-7.41680384e-01 1.03808916e+00 5.52155077e-01 1.27843767e-01
-5.41620731e-01 -5.25348186e-01 8.90519857e-01 1.75835729e-01
5.32236218e-01 -3.18387121e-01 -1.51615202e-01 -1.05947232e+00
2.10916668e-01 -7.60546565e-01 2.03002691e-01 -4.78853911e-01
-1.61718929e+00 7.23574817e-01 9.32884663e-02 -9.88071203e-01
-3.13493282e-01 -8.48569632e-01 -9.56247807e-01 8.10207605e-01
-1.74574304e+00 -6.57583475e-01 2.02718258e-01 -7.64947087e-02
7.29692221e-01 1.11950502e-01 1.18451285e+00 3.72530878e-01
-4.46889967e-01 4.77177709e-01 3.33905488e-01 -5.13907969e-02
8.00973654e-01 -1.29136348e+00 3.65315199e-01 8.15518558e-01
5.95480323e-01 9.11069989e-01 1.10511220e+00 -1.76290005e-01
-1.35336149e+00 -6.00085437e-01 1.88055122e+00 -5.82637608e-01
9.86767769e-01 -4.13202167e-01 -1.15605116e+00 1.79092735e-01
7.45948851e-01 -2.21798763e-01 8.73580575e-01 5.31817317e-01
-2.64242947e-01 4.42091003e-02 -7.34599829e-01 6.99640214e-01
6.25304818e-01 -8.37698042e-01 -1.22518706e+00 3.83783698e-01
1.06064916e+00 -3.53698023e-02 -9.87811148e-01 2.01676950e-01
3.66213560e-01 -9.36125100e-01 1.00655591e+00 -7.71015704e-01
7.80657470e-01 2.55754203e-01 -2.41256312e-01 -1.34538090e+00
-3.92431974e-01 -5.37922047e-02 -1.52520210e-01 1.35384333e+00
5.44752002e-01 -6.75539613e-01 5.73207557e-01 4.14586872e-01
-2.22372085e-01 -1.11417925e+00 -1.01209533e+00 -8.88963819e-01
3.83229405e-01 -4.45191979e-01 3.38853389e-01 7.98646629e-01
3.28722298e-01 6.94721282e-01 3.96964774e-02 -2.60550734e-02
5.01261130e-02 2.09538132e-01 5.06725907e-01 -8.51250648e-01
-3.12052727e-01 -6.47218108e-01 -3.56892496e-01 -1.03426325e+00
3.31778288e-01 -9.93274152e-01 -1.12380624e-01 -1.84518135e+00
9.09673572e-02 4.00082767e-02 -4.40747350e-01 -2.19715759e-01
-5.64704657e-01 1.50081264e-02 1.49190143e-01 -1.62471622e-01
-3.54294628e-01 7.26894081e-01 8.18444133e-01 6.16758354e-02
1.35517478e-01 -3.34947497e-01 -7.80587614e-01 4.31741416e-01
1.01415932e+00 -7.28683949e-01 -3.28071058e-01 -6.64514482e-01
2.08372444e-01 -2.14191020e-01 9.92526934e-02 -6.03778958e-01
3.59278508e-02 7.12570474e-02 -2.51578897e-01 -3.27609628e-01
4.32723641e-01 -6.75640106e-01 -6.59589291e-01 4.43899542e-01
-6.71439350e-01 5.63459516e-01 1.44836292e-01 6.18454039e-01
-5.14029801e-01 -1.14036119e+00 4.22434628e-01 1.58657506e-01
-5.78835368e-01 -6.52678311e-02 -5.16270041e-01 3.15398097e-01
6.63845479e-01 -2.65429050e-01 8.23848769e-02 -3.23625267e-01
-3.76339883e-01 -4.89823632e-02 3.00599575e-01 8.06166530e-01
6.53793931e-01 -1.24904418e+00 -8.03460181e-01 -8.21574777e-02
5.76531470e-01 -7.77413070e-01 2.14628860e-01 7.34061956e-01
-2.91434675e-01 6.77943647e-01 1.07183769e-01 -1.74702495e-01
-1.42658615e+00 6.78999126e-01 -7.80199766e-02 -5.33908427e-01
-3.03338349e-01 6.35235965e-01 -1.65901646e-01 -3.90184402e-01
7.77187571e-02 -4.19005275e-01 -4.43680793e-01 3.13353419e-01
5.08950591e-01 2.62741476e-01 2.33695671e-01 -6.28497720e-01
-3.71520102e-01 4.34834093e-01 -1.42020196e-01 -2.01840788e-01
1.43200469e+00 -3.01494092e-01 -3.32124650e-01 8.31854761e-01
1.55046785e+00 -1.67222377e-02 -4.65002775e-01 -1.01373807e-01
3.86983275e-01 -5.22429943e-01 -1.43236980e-01 -1.65684417e-01
-3.25072616e-01 1.31469941e+00 3.02358091e-01 5.55542588e-01
6.76318765e-01 -3.38679291e-02 8.78941834e-01 4.41471219e-01
2.19102874e-02 -1.20484078e+00 1.34910226e-01 6.49764776e-01
1.00754166e+00 -1.11990273e+00 1.24981023e-01 -1.83014467e-01
-6.00517869e-01 1.25395691e+00 2.06431925e-01 -4.10592467e-01
6.87332571e-01 -1.48411751e-01 -3.10210381e-02 -1.74808979e-01
-9.90257680e-01 -2.02242538e-01 4.53677714e-01 4.69125122e-01
9.41513300e-01 -2.64974445e-01 -9.95598495e-01 4.66735661e-01
-2.54550725e-01 -4.36365545e-01 4.79885608e-01 8.93365860e-01
-5.99095643e-01 -1.61424673e+00 -7.80922994e-02 3.77921313e-01
-6.67707026e-01 -3.19683671e-01 -5.05646288e-01 7.17207313e-01
-6.40366003e-02 1.01188421e+00 -8.76423791e-02 -1.91793829e-01
5.20590425e-01 4.80437666e-01 5.77487111e-01 -1.09978056e+00
-9.59033251e-01 -5.27450323e-01 4.39130336e-01 -2.63547758e-03
-4.06645209e-01 -6.91503763e-01 -7.53880978e-01 -8.54184926e-02
-6.56245708e-01 3.33710074e-01 6.03407204e-01 1.03132999e+00
4.46520358e-01 4.41167474e-01 5.16825199e-01 -4.96613294e-01
-1.04820776e+00 -1.29622626e+00 -3.51890147e-01 5.88359833e-01
1.65445164e-01 -4.55619335e-01 -5.08017898e-01 -4.27047104e-01] | [10.692692756652832, 8.737533569335938] |
53878d26-6ee4-4215-bbbd-11b0ef151377 | distinguishing-rule-and-exemplar-based-1 | 2110.04328 | null | https://arxiv.org/abs/2110.04328v2 | https://arxiv.org/pdf/2110.04328v2.pdf | Distinguishing rule- and exemplar-based generalization in learning systems | Machine learning systems often do not share the same inductive biases as humans and, as a result, extrapolate or generalize in ways that are inconsistent with our expectations. The trade-off between exemplar- and rule-based generalization has been studied extensively in cognitive psychology; in this work, we present a protocol inspired by these experimental approaches to probe the inductive biases that control this tradeoff in category-learning systems. We isolate two such inductive biases: feature-level bias (differences in which features are more readily learned) and exemplar or rule bias (differences in how these learned features are used for generalization). We find that standard neural network models are feature-biased and exemplar-based, and discuss the implications of these findings for machine learning research on systematic generalization, fairness, and data augmentation. | ['Thomas L. Griffiths', 'Erin Grant', 'Ishita Dasgupta'] | 2021-10-08 | distinguishing-rule-and-exemplar-based | https://openreview.net/forum?id=ljCoTzUsdS | https://openreview.net/pdf?id=ljCoTzUsdS | null | ['systematic-generalization'] | ['reasoning'] | [ 3.83990407e-01 -2.19142474e-02 -4.69475210e-01 -9.44618642e-01
3.72975498e-01 -6.53523862e-01 8.17755282e-01 6.22639179e-01
-9.04677510e-01 7.99933195e-01 1.79374903e-01 -4.90425438e-01
-4.89156485e-01 -9.14660811e-01 -5.47716141e-01 -4.13464099e-01
8.29871371e-02 3.26737970e-01 -3.64088535e-01 -2.21353516e-01
9.49070632e-01 6.92084193e-01 -1.72976243e+00 2.97869474e-01
1.06419206e+00 7.39951253e-01 -4.47906077e-01 2.50815481e-01
-6.87246397e-02 5.26490033e-01 -3.55190635e-01 -3.56276900e-01
2.33588040e-01 -6.43252671e-01 -8.81031275e-01 -1.92538291e-01
7.38307476e-01 -2.77852833e-01 -1.89194828e-01 1.06998026e+00
3.57358068e-01 6.34111762e-01 1.01563537e+00 -1.32060039e+00
-1.55047047e+00 6.29405558e-01 -2.02334434e-01 4.54242766e-01
5.48070818e-02 3.69609475e-01 8.48924160e-01 -7.62814820e-01
5.35200536e-01 1.47769105e+00 5.37359715e-01 8.23857307e-01
-1.77155352e+00 -1.05476356e+00 3.79081726e-01 6.59072399e-02
-1.29876864e+00 -6.88137531e-01 5.92730582e-01 -6.28686965e-01
4.98664230e-01 1.68804720e-01 8.16915393e-01 1.12824428e+00
3.44647706e-01 3.99883240e-02 1.94120061e+00 -4.68988746e-01
5.92097342e-01 6.74215078e-01 3.80425185e-01 9.58511308e-02
8.84573042e-01 8.90492141e-01 -5.30129731e-01 -4.70770672e-02
9.93391335e-01 1.62188560e-01 -1.84259042e-01 -3.52035582e-01
-1.00327301e+00 1.04239261e+00 8.94451618e-01 3.42558652e-01
-3.62812817e-01 6.22255355e-02 2.92813480e-01 6.04276359e-01
1.57424659e-01 1.42094815e+00 -3.80147606e-01 4.85094249e-01
-7.05665231e-01 5.58278501e-01 4.76510555e-01 6.51269317e-01
9.81937587e-01 3.78914446e-01 -4.03661847e-01 8.79067361e-01
9.64889079e-02 3.96695793e-01 8.80344570e-01 -1.04482055e+00
-2.90839195e-01 3.86893809e-01 3.70670594e-02 -9.94885206e-01
-3.74388367e-01 -6.68265879e-01 -4.79849577e-01 6.52384758e-01
4.03317481e-01 -9.41471457e-02 -8.79841685e-01 2.13266230e+00
5.00050820e-02 -4.27052647e-01 -1.61348760e-01 9.80357468e-01
4.08518106e-01 4.72757854e-02 8.59730124e-01 -2.72312582e-01
9.95442629e-01 -3.51023078e-01 -3.48398685e-01 -3.57796222e-01
6.84219003e-01 -3.37702155e-01 1.38203979e+00 8.77248421e-02
-9.02688563e-01 -6.92571580e-01 -9.56819594e-01 7.77758285e-02
-7.11749315e-01 -5.52592099e-01 9.90262449e-01 9.64825809e-01
-9.26641405e-01 9.43472147e-01 7.59689584e-02 -6.07022703e-01
6.63895309e-01 1.13490842e-01 -2.57207870e-01 1.20324709e-01
-1.08308935e+00 1.26597846e+00 6.63889706e-01 -1.01596586e-01
-7.52116501e-01 -8.45363557e-01 -7.94484973e-01 8.92356038e-02
1.06456317e-01 -6.89177990e-01 1.17257667e+00 -1.56432855e+00
-1.05995905e+00 1.28193569e+00 7.10596936e-03 -2.28042603e-01
-3.62591259e-02 8.65872204e-02 -5.28113008e-01 -3.28941584e-01
1.21073775e-01 8.95374298e-01 7.35381961e-01 -1.58145916e+00
-4.22194004e-01 -5.71287930e-01 8.22088197e-02 2.72579432e-01
-2.66042948e-01 -1.70411289e-01 1.07743621e+00 -7.46330559e-01
3.67379338e-02 -9.77799118e-01 -2.42842302e-01 2.19912753e-01
5.29609248e-02 -3.27971965e-01 2.99910694e-01 1.45433769e-01
9.93989766e-01 -2.04284501e+00 -2.77062207e-01 4.22764659e-01
3.39625567e-01 1.53561488e-01 -2.10738584e-01 2.29779661e-01
-3.02340180e-01 6.51934981e-01 -1.78418562e-01 4.11499828e-01
2.54121304e-01 2.39637434e-01 -3.86199415e-01 3.71960729e-01
1.46112576e-01 8.09080064e-01 -9.69090343e-01 -3.53080660e-01
5.84305683e-03 6.34474028e-03 -7.92210639e-01 1.09362729e-01
1.03628263e-01 4.61505294e-01 -1.28901854e-01 4.28007364e-01
6.60955727e-01 1.21593639e-01 1.71052605e-01 5.08112041e-03
-2.94307917e-01 6.11645222e-01 -6.83932364e-01 1.14016604e+00
-3.51776183e-01 7.60161579e-01 -6.49804652e-01 -1.02823639e+00
1.03047609e+00 -6.69066310e-02 -3.88762146e-01 -8.63550305e-01
2.98453242e-01 3.76266569e-01 9.02060390e-01 -2.29630366e-01
5.41520000e-01 -9.37745392e-01 5.87648302e-02 7.18508422e-01
1.93034917e-01 -5.91188967e-01 3.46886590e-02 -5.06915934e-02
2.14958027e-01 -2.13914341e-03 5.98852217e-01 -8.47976863e-01
1.86318278e-01 -3.50936526e-03 8.40070069e-01 1.28123784e+00
-4.30359185e-01 6.97860569e-02 3.17065567e-01 -4.49372679e-01
-1.05817294e+00 -1.22542942e+00 -6.67340934e-01 1.68095088e+00
3.71404388e-03 3.38668257e-01 -4.51245785e-01 -5.00158191e-01
3.77575755e-01 1.75680339e+00 -1.12830424e+00 -8.40902567e-01
-1.28920078e-01 -4.93981034e-01 3.62754077e-01 7.05469072e-01
6.93165436e-02 -1.33105695e+00 -5.75157762e-01 -1.60398498e-01
5.54664791e-01 -3.84418368e-01 -2.20224038e-01 3.24237674e-01
-1.11856771e+00 -8.24283957e-01 -7.28759989e-02 -3.84225398e-01
9.61888671e-01 3.02805305e-01 1.44295895e+00 5.75305581e-01
-3.88662368e-02 4.52968895e-01 -3.60011160e-01 -9.51399624e-01
-3.89883429e-01 -3.61226261e-01 3.53518605e-01 -3.40575516e-01
7.36404300e-01 -7.02355444e-01 -7.04315007e-01 2.60760427e-01
-9.76037443e-01 -2.97092438e-01 6.90427244e-01 8.32801223e-01
3.11351448e-01 -3.50981861e-01 1.07954609e+00 -1.14372373e+00
1.03619039e+00 -6.36901975e-01 -3.21195662e-01 5.27765974e-02
-1.36861086e+00 -1.09546617e-01 4.63568658e-01 -6.89677179e-01
-1.25757861e+00 -5.45478404e-01 3.78706664e-01 8.56597424e-02
-4.65092301e-01 4.41168636e-01 2.25140452e-01 -4.64322150e-01
1.30612993e+00 8.34901705e-02 -9.69886556e-02 -4.77658734e-02
4.46052641e-01 5.71070790e-01 1.85653016e-01 -1.11703944e+00
8.09560835e-01 8.91143009e-02 -5.10275960e-02 -5.20229459e-01
-1.14990997e+00 3.42126369e-01 -6.50809050e-01 -2.56578207e-01
5.64475298e-01 -6.47578835e-01 -5.53066909e-01 2.57808082e-02
-5.75728416e-01 -3.52972329e-01 -9.04337049e-01 8.74873161e-01
-6.61859095e-01 -2.96715438e-01 -2.33866841e-01 -8.94486725e-01
-1.01406835e-01 -8.28523219e-01 2.28440419e-01 6.10318303e-01
-7.33124912e-01 -9.03779089e-01 -9.18993801e-02 -1.03451133e-01
8.47851098e-01 3.16835158e-02 1.34092104e+00 -1.26550639e+00
-2.63823211e-01 8.32801163e-02 -3.09533924e-01 2.29101524e-01
5.57680652e-02 7.11846212e-03 -9.55421031e-01 -1.28939673e-01
1.27789617e-01 -7.29199231e-01 9.51112926e-01 3.76087099e-01
1.27180612e+00 -2.90297240e-01 -4.89105098e-02 5.05114615e-01
1.26979458e+00 3.34776282e-01 3.74383003e-01 3.62890542e-01
1.52672738e-01 8.67417634e-01 5.03925025e-01 7.46680945e-02
1.43660426e-01 1.04870290e-01 -6.54851049e-02 2.38395691e-01
1.88106105e-01 -4.03199762e-01 -1.36589304e-01 1.30270377e-01
-2.12416649e-01 1.62829906e-01 -6.97596788e-01 6.30142391e-01
-1.37136674e+00 -1.03790689e+00 1.85439125e-01 2.64100742e+00
1.09335160e+00 1.64439946e-01 7.25730136e-02 -1.55714937e-02
6.95621252e-01 -1.29549548e-01 -7.80407250e-01 -1.04364324e+00
-1.34931192e-01 2.59301096e-01 3.84158552e-01 2.20723644e-01
-6.27466321e-01 8.92943859e-01 7.87771988e+00 3.57686520e-01
-1.08968306e+00 -2.86839277e-01 8.78369272e-01 -1.49852261e-01
-6.69387221e-01 2.37001687e-01 -4.59379405e-01 1.17777929e-01
8.14176381e-01 -5.39062142e-01 5.87411523e-01 8.02859008e-01
-2.76976041e-02 -1.32490292e-01 -1.63256168e+00 3.30847323e-01
-2.30406910e-01 -9.85872090e-01 4.30155873e-01 -4.21917848e-02
7.72708893e-01 -5.89840114e-01 4.95547831e-01 5.54768801e-01
5.23476005e-01 -1.35559905e+00 7.32994139e-01 4.30204451e-01
7.04053938e-01 -4.68781471e-01 4.02755320e-01 -3.03930622e-02
-2.28960499e-01 -3.66057456e-01 -6.93619132e-01 -6.75584137e-01
-4.01322365e-01 4.21140879e-01 -5.34659147e-01 -3.40902567e-01
6.08041108e-01 1.74215943e-01 -6.07227087e-01 8.68096769e-01
-3.62341881e-01 5.67954898e-01 2.30036572e-01 -1.83167055e-01
6.76389560e-02 -1.42574161e-01 2.49680966e-01 1.03362262e+00
6.30201995e-02 6.73646256e-02 -2.83775806e-01 1.40595520e+00
-1.60794392e-01 1.78065687e-01 -7.64669836e-01 -1.88454449e-01
9.56128538e-01 1.02833021e+00 -6.97194755e-01 -4.46808726e-01
-2.95392066e-01 3.63413543e-01 5.10405660e-01 5.34121454e-01
-3.74485314e-01 -1.18510157e-01 8.88314784e-01 1.47044852e-01
-1.59234703e-01 -7.57207349e-02 -6.99260831e-01 -9.04431760e-01
-5.50984740e-01 -8.56261253e-01 2.03403398e-01 -6.26682580e-01
-1.84091437e+00 6.95866272e-02 1.83775768e-01 -6.37558639e-01
4.48468328e-02 -6.31854057e-01 -5.29908955e-01 1.07508266e+00
-1.35322928e+00 -7.39289403e-01 -1.01787604e-01 4.38516706e-01
1.52363271e-01 -2.89254785e-02 9.37412918e-01 -2.89734304e-01
-2.23205894e-01 7.11253166e-01 -9.12926793e-02 1.23469085e-01
1.05454302e+00 -1.12122643e+00 1.62333637e-01 2.82862425e-01
-4.31636721e-02 1.32410800e+00 5.95090032e-01 -5.10158539e-01
-7.53948212e-01 -8.45442891e-01 6.58226371e-01 -3.49306583e-01
3.65913063e-01 -2.15283539e-02 -1.14688277e+00 8.46760035e-01
9.84486267e-02 -8.11242014e-02 1.34318376e+00 6.96399748e-01
-8.95507634e-01 -7.85652772e-02 -1.44425762e+00 7.93455720e-01
1.35324800e+00 -5.53499162e-01 -9.95493829e-01 3.92686985e-02
5.97339451e-01 2.06493974e-01 -9.30664420e-01 2.64409631e-01
8.63873661e-01 -1.25819218e+00 9.73628044e-01 -1.33526480e+00
5.55539966e-01 1.17415935e-01 -2.61586338e-01 -1.81909955e+00
-1.12936664e+00 1.14055417e-01 5.59318602e-01 1.07680857e+00
4.88806695e-01 -1.17817020e+00 2.47705549e-01 9.99518514e-01
-1.94023773e-02 -4.36169356e-01 -6.43368661e-01 -7.88441658e-01
8.15208972e-01 6.75847381e-02 8.46820116e-01 1.41574073e+00
1.18739679e-01 2.84224063e-01 1.46364599e-01 -3.55364054e-01
6.48598969e-01 1.61136433e-01 4.14449334e-01 -1.59052229e+00
1.21813416e-01 -9.26355600e-01 -3.33476365e-01 -2.95343041e-01
3.13423812e-01 -9.69971061e-01 -6.95205331e-02 -7.26329565e-01
3.30099553e-01 -7.70774007e-01 -6.32771254e-01 2.19228730e-01
-4.03451860e-01 -8.75003263e-02 4.56517190e-01 1.30431652e-01
-5.60902432e-02 4.45757657e-01 1.13690746e+00 1.92508265e-01
-2.67645091e-01 -3.27317476e-01 -1.74207091e+00 9.11478102e-01
1.05168629e+00 -5.17275274e-01 -4.38385636e-01 -2.77344555e-01
1.37932390e-01 -4.78383899e-01 5.06915033e-01 -8.50033104e-01
-2.79743131e-02 -8.79498780e-01 9.18570936e-01 2.85223186e-01
3.18362229e-02 -8.82027030e-01 -2.21591339e-01 4.77265835e-01
-9.86332655e-01 -6.98844120e-02 2.88614392e-01 4.08131748e-01
3.11272174e-01 -5.27539253e-01 1.15364563e+00 -3.24881226e-01
-6.08134568e-01 1.14042200e-01 -6.23485208e-01 2.62681842e-01
9.37301934e-01 -3.88622582e-01 -6.37104988e-01 -2.87329882e-01
-6.24127865e-01 -1.28806457e-01 6.72091305e-01 4.54215676e-01
4.15141940e-01 -1.45733154e+00 -8.43606114e-01 5.31191304e-02
3.93447936e-01 -6.60673618e-01 1.58532709e-02 6.42808855e-01
-2.02253298e-03 3.29206139e-01 -6.84597313e-01 -1.09517433e-01
-5.77077925e-01 9.07899439e-01 4.75975364e-01 4.42598045e-01
1.04398221e-01 8.31394196e-01 6.04310036e-01 -5.40455818e-01
-3.30823421e-01 2.89462004e-02 -1.30440205e-01 3.98091041e-02
4.39060181e-01 3.97049129e-01 -4.18933004e-01 -4.75388616e-01
-2.03514829e-01 2.63648599e-01 -2.96544224e-01 3.02541330e-02
9.92511511e-01 1.08596366e-02 -2.27091104e-01 8.09937537e-01
6.64880812e-01 -2.47138053e-01 -7.34615743e-01 -3.27174217e-01
-1.92444220e-01 -9.81282651e-01 -3.25049423e-02 -9.24481809e-01
-5.18514872e-01 6.75568759e-01 6.04533494e-01 2.35820279e-01
1.05714536e+00 -2.79964864e-01 -1.85244218e-01 5.51257789e-01
2.74636000e-01 -1.31310415e+00 -2.04850256e-01 1.05038233e-01
1.18444920e+00 -1.32647300e+00 2.68529028e-01 -1.38935640e-01
-6.11548305e-01 8.21894646e-01 1.02480173e+00 -3.61343563e-01
4.93498534e-01 -2.88476288e-01 -9.10348743e-02 -1.21100672e-01
-8.29132736e-01 -2.57866800e-01 2.69258648e-01 1.09896421e+00
8.91073287e-01 3.65687221e-01 -8.37751091e-01 5.08074760e-01
-4.36326593e-01 6.98923320e-02 6.11328721e-01 8.89285266e-01
-7.08351851e-01 -8.15564632e-01 -5.55588543e-01 9.47450638e-01
-1.38533279e-01 -4.22680408e-01 -6.50505304e-01 8.15546274e-01
1.93738788e-01 1.04880810e+00 4.16429877e-01 -3.04222226e-01
3.04858416e-01 2.91487306e-01 6.71213448e-01 -8.00033689e-01
-7.69260764e-01 -6.00449622e-01 -1.64785907e-02 -3.92437130e-01
-4.08474386e-01 -8.42760742e-01 -1.00765264e+00 -5.72707832e-01
-3.03388089e-01 1.61694869e-01 4.66059953e-01 8.00068378e-01
2.32846200e-01 2.95403898e-01 4.75535035e-01 -6.78242743e-01
-8.15025210e-01 -1.03206301e+00 -7.75876999e-01 7.63879895e-01
1.42003536e-01 -9.18489158e-01 -6.00673616e-01 -1.43779606e-01] | [9.415956497192383, 6.26626443862915] |
d504ab34-59df-491a-8f3a-f90aa2ba4f80 | tokenflow-rethinking-fine-grained-cross-modal | 2209.13822 | null | https://arxiv.org/abs/2209.13822v2 | https://arxiv.org/pdf/2209.13822v2.pdf | TokenFlow: Rethinking Fine-grained Cross-modal Alignment in Vision-Language Retrieval | Most existing methods in vision-language retrieval match two modalities by either comparing their global feature vectors which misses sufficient information and lacks interpretability, detecting objects in images or videos and aligning the text with fine-grained features which relies on complicated model designs, or modeling fine-grained interaction via cross-attention upon visual and textual tokens which suffers from inferior efficiency. To address these limitations, some recent works simply aggregate the token-wise similarities to achieve fine-grained alignment, but they lack intuitive explanations as well as neglect the relationships between token-level features and global representations with high-level semantics. In this work, we rethink fine-grained cross-modal alignment and devise a new model-agnostic formulation for it. We additionally demystify the recent popular works and subsume them into our scheme. Furthermore, inspired by optimal transport theory, we introduce TokenFlow, an instantiation of the proposed scheme. By modifying only the similarity function, the performance of our method is comparable to the SoTA algorithms with heavy model designs on major video-text retrieval benchmarks. The visualization further indicates that TokenFlow successfully leverages the fine-grained information and achieves better interpretability. | ['Zhongyuan Wang', 'Lele Cheng', 'Changqiao Wu', 'Xiaohan Zou'] | 2022-09-28 | null | null | null | null | ['video-text-retrieval'] | ['computer-vision'] | [-6.13662452e-02 -5.15951037e-01 -4.08841610e-01 -3.08545500e-01
-5.51558971e-01 -6.64247930e-01 1.15813649e+00 3.14895868e-01
-3.75062138e-01 1.55834034e-01 5.38353741e-01 -6.11709803e-02
-5.03651083e-01 -5.57855666e-01 -5.07324815e-01 -5.41429877e-01
1.26646489e-01 1.51873291e-01 3.74336869e-01 -1.26890332e-01
4.82460976e-01 4.12586570e-01 -1.82095134e+00 6.46083593e-01
6.14199460e-01 1.20671523e+00 1.07488327e-01 6.40184581e-01
-3.87601614e-01 7.53150940e-01 -2.60193795e-01 -3.71798486e-01
3.88008118e-01 -1.38389647e-01 -9.80273604e-01 -4.77441866e-03
1.04813445e+00 -4.82328862e-01 -6.07809901e-01 1.06556165e+00
2.50690669e-01 -7.77899697e-02 7.42903769e-01 -1.46995604e+00
-1.03321385e+00 3.36908698e-01 -8.10872972e-01 1.67498082e-01
4.24791515e-01 -8.54343548e-03 1.29471433e+00 -9.46609676e-01
4.22272652e-01 1.63463449e+00 4.91575569e-01 3.31088513e-01
-9.55048382e-01 -3.16502929e-01 5.93852401e-01 5.54389834e-01
-1.43209708e+00 -4.01212633e-01 5.49446166e-01 -5.12101710e-01
9.16211426e-01 5.54401219e-01 5.94051301e-01 9.61540699e-01
1.95302725e-01 8.90936077e-01 1.12118697e+00 -3.96742016e-01
-5.82329817e-02 -1.82572544e-01 3.93020242e-01 7.96361268e-01
2.96774507e-01 -7.66477957e-02 -7.76981652e-01 -1.84839725e-01
7.19117343e-01 5.52049994e-01 -1.86908558e-01 -5.19323349e-01
-1.71793067e+00 5.77766001e-01 5.71348190e-01 4.23354268e-01
-2.94227928e-01 4.17137027e-01 5.47005653e-01 2.33750463e-01
2.72072732e-01 1.35455385e-01 -1.04118161e-01 -1.31793339e-02
-1.12027156e+00 2.10171744e-01 4.44317192e-01 1.09993291e+00
9.26452577e-01 -1.73714876e-01 -6.08186007e-01 6.75209105e-01
6.81686580e-01 5.67379355e-01 7.21064866e-01 -9.03276563e-01
3.40204269e-01 6.47984922e-01 1.96599849e-02 -1.34717607e+00
-2.86691785e-01 -3.19738775e-01 -7.53756166e-01 -1.71422288e-01
2.36014873e-01 5.91106713e-01 -1.05608320e+00 1.56679022e+00
1.17536657e-01 1.57413170e-01 -3.28933269e-01 1.13854909e+00
9.23327863e-01 3.38064283e-01 1.89899340e-01 9.15067866e-02
1.61172581e+00 -1.46777403e+00 -7.87434459e-01 4.43870388e-02
5.03854990e-01 -8.05586100e-01 1.29244566e+00 1.15352519e-01
-1.12379396e+00 -5.01327991e-01 -1.01117754e+00 -4.64906037e-01
-6.83748603e-01 -9.81798247e-02 7.15334058e-01 4.08816069e-01
-1.26072681e+00 5.00371873e-01 -8.20112467e-01 -7.36878216e-01
3.41714561e-01 2.21875936e-01 -4.06965524e-01 -1.09951407e-01
-1.08229899e+00 7.83041537e-01 1.86970294e-01 1.26667768e-01
-5.39335251e-01 -6.23723805e-01 -7.74008036e-01 2.02381685e-01
3.14592093e-01 -1.13583136e+00 9.41712737e-01 -7.81040370e-01
-1.36176622e+00 8.80476475e-01 -2.90021628e-01 -3.65453869e-01
5.36724448e-01 -3.78439724e-01 -1.60442680e-01 3.20692956e-01
8.32863450e-02 8.24710965e-01 1.10753667e+00 -1.22930157e+00
-5.89846134e-01 -4.13294256e-01 5.69002748e-01 2.66519248e-01
-7.71546066e-01 4.52106781e-02 -9.28172588e-01 -8.32806110e-01
1.37161374e-01 -6.79876804e-01 -4.43374217e-02 2.62800604e-01
-4.77044672e-01 -2.19698459e-01 9.15663660e-01 -1.90447658e-01
1.52519894e+00 -1.90923643e+00 1.68705776e-01 1.22147411e-01
8.54021072e-01 1.24969013e-01 -3.85937572e-01 6.32352591e-01
1.20933525e-01 2.71677881e-01 1.03949517e-01 -4.98518705e-01
4.57878262e-01 2.86543250e-01 -4.00115073e-01 5.82863867e-01
1.88907403e-02 1.22181463e+00 -9.94006932e-01 -8.95047009e-01
4.59152520e-01 3.94663215e-01 -6.38526022e-01 -3.85220796e-02
-4.56912704e-02 -7.83520639e-02 -7.57252395e-01 7.87939668e-01
7.32371330e-01 -5.29228032e-01 -1.16902940e-01 -8.55247438e-01
-7.01041669e-02 -5.24879210e-02 -1.15674508e+00 1.97846162e+00
-2.96420097e-01 5.26354611e-01 -9.77666602e-02 -1.07936609e+00
3.30496728e-01 -7.73212500e-03 6.53190970e-01 -7.68279910e-01
2.17819847e-02 -3.67927663e-02 -3.66769284e-01 -7.07558453e-01
7.06937194e-01 1.01511084e-01 -9.22599807e-02 4.03674603e-01
8.17092974e-03 1.73718885e-01 -8.46470073e-02 4.86067653e-01
9.93134618e-01 2.96574861e-01 3.21559697e-01 -2.01745391e-01
3.86421710e-01 -1.86854824e-01 1.03929136e-02 1.32696295e+00
-2.61281550e-01 7.90348828e-01 1.05289429e-01 -4.06550139e-01
-7.99702823e-01 -1.09458530e+00 -1.20158635e-01 1.33792520e+00
5.84575415e-01 -1.04760444e+00 -6.54399693e-01 -7.55608737e-01
3.76569331e-02 9.11833271e-02 -8.02600801e-01 -7.01740682e-02
-2.48559326e-01 -6.18632436e-01 6.12776697e-01 4.56043959e-01
5.45005798e-01 -6.14972055e-01 -5.23085237e-01 -2.67902881e-01
-3.19206566e-01 -1.23091960e+00 -7.14062750e-01 -1.69519648e-01
-5.23125291e-01 -8.71663988e-01 -6.94215238e-01 -5.03111601e-01
4.74417329e-01 7.29293168e-01 1.05112612e+00 3.44458342e-01
-3.82865071e-01 8.41682017e-01 -6.04323745e-01 -1.52172029e-01
2.33995259e-01 -1.49249062e-01 1.63430184e-01 2.61665344e-01
3.91613007e-01 -2.37873659e-01 -8.40713918e-01 2.66809314e-01
-1.28846622e+00 1.88716352e-01 5.31948626e-01 9.46420729e-01
4.64584529e-01 -3.40150714e-01 2.02947706e-01 -5.59637904e-01
5.14299035e-01 -4.38339949e-01 -1.65109619e-01 5.70507586e-01
-5.67477167e-01 3.60335380e-01 4.19328213e-01 -4.55649495e-01
-8.07490349e-01 -2.76201606e-01 2.30577797e-01 -8.03324044e-01
-3.94733157e-03 4.16496664e-01 5.06613869e-04 -3.14609349e-01
3.49171609e-01 1.73723757e-01 -2.65597641e-01 -4.86113310e-01
7.81448364e-01 5.38996994e-01 2.97891259e-01 -8.16954494e-01
8.89509201e-01 8.92179012e-01 -1.92475859e-02 -6.97228074e-01
-9.78751719e-01 -6.58507049e-01 -6.48520827e-01 1.85179804e-02
8.79277647e-01 -7.47198820e-01 -1.05914617e+00 4.17317420e-01
-1.09608853e+00 1.82995439e-01 -2.72695184e-01 4.86899346e-01
-6.08591259e-01 9.02602196e-01 -5.37605226e-01 -4.71524388e-01
-1.81221128e-01 -1.09724367e+00 1.65166104e+00 -3.58517207e-02
-6.10045195e-02 -1.05959737e+00 -8.29668418e-02 3.05170804e-01
6.44585073e-01 -4.57435846e-02 8.37048113e-01 -5.79160154e-01
-7.14087307e-01 -5.56765646e-02 -7.00587749e-01 -1.78902447e-02
1.14026375e-01 1.43724963e-01 -1.03764057e+00 -2.98240453e-01
-1.70209140e-01 -2.54875064e-01 1.18143737e+00 3.35185647e-01
1.37857389e+00 -3.43256474e-01 -4.33452904e-01 6.84321463e-01
1.36450851e+00 -4.39554185e-01 3.93123716e-01 6.41999066e-01
1.00566101e+00 6.12349510e-01 5.20605624e-01 5.16028523e-01
7.65562475e-01 9.55427051e-01 5.57270288e-01 -2.46828631e-01
-3.22647095e-01 -2.07126915e-01 2.57182479e-01 9.17240679e-01
-1.40817821e-01 -3.20285022e-01 -6.25761569e-01 5.84536731e-01
-2.35163760e+00 -1.09768236e+00 -1.72383096e-02 1.86467695e+00
5.32318294e-01 -2.19069332e-01 -2.95155104e-02 -2.05153331e-01
5.82752109e-01 5.14138699e-01 -3.87064427e-01 -1.12439185e-01
-3.08191717e-01 -3.67107019e-02 5.37975669e-01 4.61300641e-01
-1.08864605e+00 1.07894623e+00 6.48357058e+00 1.22226477e+00
-1.14491010e+00 1.75531566e-01 1.37075186e-01 -2.74473518e-01
-5.19788325e-01 1.06541544e-01 -6.09746039e-01 2.75295317e-01
5.32899082e-01 -3.10496651e-02 3.33553642e-01 4.33721870e-01
2.49126941e-01 1.25393063e-01 -1.39150167e+00 1.31040967e+00
3.06695282e-01 -1.54454172e+00 6.52857125e-01 -1.33144911e-02
4.65351760e-01 -1.46514131e-03 1.23681337e-01 7.75072873e-02
7.02294558e-02 -1.06315351e+00 1.06382465e+00 9.10352707e-01
5.11376262e-01 -1.25348233e-02 5.38312376e-01 -8.04015156e-03
-1.51879108e+00 6.19154908e-02 -2.39087507e-01 -7.04605803e-02
3.01218033e-03 1.70598328e-01 -2.99786329e-01 8.24338913e-01
8.93715084e-01 1.05262971e+00 -8.90493095e-01 1.03940165e+00
3.56123269e-01 3.21262270e-01 -3.44585866e-01 1.10455461e-01
4.87010747e-01 6.22484135e-03 4.52371895e-01 1.58080983e+00
1.68656275e-01 -2.42676124e-01 4.29396838e-01 6.92067683e-01
1.24174086e-02 1.07743882e-01 -4.65311110e-01 3.80790643e-02
4.40407157e-01 1.29809344e+00 -5.13630867e-01 -5.67662895e-01
-7.18337595e-01 1.01963568e+00 2.53489494e-01 6.87281549e-01
-8.07327271e-01 -2.82807350e-01 8.10729325e-01 9.17236581e-02
2.33995408e-01 -2.26158336e-01 -2.08011433e-01 -1.63236833e+00
1.38128012e-01 -7.40610182e-01 4.90823328e-01 -8.60947907e-01
-1.56373811e+00 5.17087638e-01 2.34557360e-01 -1.28566265e+00
-1.53436899e-01 -6.17911994e-01 -2.44316593e-01 6.87599838e-01
-1.86240506e+00 -1.63056910e+00 -5.49045980e-01 1.09711826e+00
5.98184347e-01 2.20338479e-01 6.35363281e-01 3.45824271e-01
-3.67156953e-01 7.91641533e-01 1.04794130e-01 -6.05603866e-03
9.67356503e-01 -1.07422817e+00 1.52939394e-01 5.42104840e-01
2.82099575e-01 1.02848637e+00 5.56750834e-01 -3.14036101e-01
-1.74675679e+00 -9.69741344e-01 7.75308728e-01 -5.82943261e-01
1.06741822e+00 -3.41288924e-01 -7.99121320e-01 6.30167544e-01
3.90640736e-01 3.32913190e-01 3.74447256e-01 -7.15963244e-02
-7.95021951e-01 -1.56559557e-01 -9.56836402e-01 7.70903766e-01
1.22694981e+00 -8.82892728e-01 -6.66897118e-01 2.30286300e-01
7.22503901e-01 -1.81973055e-01 -9.02743518e-01 2.98248738e-01
1.00203478e+00 -1.03762102e+00 1.25304067e+00 -8.79230797e-01
2.70565182e-01 -4.71693486e-01 -5.83912551e-01 -7.83293545e-01
-5.19072115e-01 -4.46119875e-01 -2.03057870e-01 1.18104815e+00
-1.85546249e-01 -5.20388961e-01 3.78340483e-01 3.60991687e-01
-3.12085971e-02 -6.48951054e-01 -6.62290633e-01 -7.98624814e-01
-1.02240264e-01 -5.93735337e-01 6.78497016e-01 9.84611809e-01
3.27706784e-02 7.56988674e-02 -5.05392253e-01 1.56408191e-01
7.42563069e-01 4.06505495e-01 6.46896243e-01 -1.03564584e+00
-2.28831738e-01 -7.55747020e-01 -7.17664659e-01 -1.41877961e+00
1.44853696e-01 -8.70724499e-01 -1.47974342e-01 -1.63197267e+00
7.18575239e-01 -2.78497934e-01 -6.58551037e-01 6.36889994e-01
-3.26573104e-02 5.89434266e-01 5.22118270e-01 5.12835383e-01
-1.08099949e+00 4.78141546e-01 1.23667717e+00 -3.99540365e-01
2.19323680e-01 -5.17424881e-01 -7.07339704e-01 8.13085318e-01
2.41734713e-01 -2.21896425e-01 -3.26890975e-01 -6.83847845e-01
2.44499370e-01 -2.23183438e-01 8.57608318e-01 -5.92865586e-01
4.48860347e-01 -3.92713696e-01 2.32799560e-01 -5.52630067e-01
4.46101099e-01 -1.02017868e+00 -9.64944810e-02 1.65799975e-01
-5.18894136e-01 1.14081599e-01 2.71972194e-02 7.57159412e-01
-4.16411012e-01 -1.87386870e-02 3.11242044e-01 3.90425883e-03
-8.05752456e-01 5.52172720e-01 -2.72752941e-01 -1.17005050e-01
7.36774921e-01 -4.91797060e-01 -6.28545105e-01 -2.89527982e-01
-6.02368057e-01 2.82326579e-01 6.56714797e-01 6.92515075e-01
5.49800277e-01 -1.55737555e+00 -4.39060152e-01 4.42315899e-02
5.34959674e-01 -3.93943280e-01 4.05058235e-01 1.09917831e+00
-2.60210156e-01 7.68621385e-01 -1.25894248e-01 -9.68174636e-01
-1.19924188e+00 6.82004929e-01 1.56593412e-01 -1.94887206e-01
-5.81748247e-01 5.31400800e-01 8.35994542e-01 2.20394330e-04
2.25604251e-01 -5.46446085e-01 -2.28848830e-01 1.17109597e-01
4.13193136e-01 2.30445862e-01 -4.11630869e-02 -7.13362098e-01
-6.51244342e-01 1.20795691e+00 -2.16251373e-01 1.29344165e-01
1.07834494e+00 -6.62954807e-01 -1.89782426e-01 4.22144771e-01
1.30875492e+00 6.38050139e-02 -1.12541115e+00 -6.30032003e-01
-8.25815573e-02 -7.05131710e-01 1.78609043e-01 -5.35446167e-01
-9.17133987e-01 1.13774359e+00 5.55388391e-01 2.84295529e-01
1.20594561e+00 2.61993378e-01 7.49732137e-01 5.08108377e-01
2.25477189e-01 -9.06342506e-01 2.58401036e-01 4.93344605e-01
7.57497251e-01 -1.25817263e+00 5.30951470e-02 -2.61425257e-01
-4.75354016e-01 1.12159741e+00 3.92807454e-01 -1.20231330e-01
6.61133409e-01 4.83096093e-02 -9.79770795e-02 -3.63645464e-01
-7.34291494e-01 -3.39700133e-01 7.01453745e-01 3.83799076e-01
3.08093339e-01 -9.91018042e-02 -2.16367617e-01 5.19446135e-01
1.42320424e-01 -1.23250365e-01 9.80662927e-02 8.79448056e-01
-3.69616598e-01 -1.00951397e+00 -3.74209195e-01 4.16774660e-01
-3.79085690e-01 -5.42429507e-01 -2.56300211e-01 7.50423968e-01
6.46968931e-03 8.48780274e-01 8.80350471e-02 -3.55319440e-01
9.33024511e-02 -2.42264599e-01 6.25919938e-01 -2.98941255e-01
-4.92247403e-01 1.82271853e-01 -2.61976331e-01 -9.58580613e-01
-8.27618241e-01 -3.79990876e-01 -1.11638880e+00 -3.51836085e-01
-2.31473565e-01 -1.29825696e-01 5.84978223e-01 1.29578340e+00
5.99033535e-01 3.72125238e-01 3.59109849e-01 -1.00182462e+00
-5.33121884e-01 -6.47035956e-01 -4.26370770e-01 5.75950205e-01
6.27087235e-01 -8.22765768e-01 -2.10517153e-01 2.70047784e-01] | [10.512856483459473, 1.098780870437622] |
586805db-257b-4d01-a2f0-4a77c1e00141 | spam-detection-using-bert | 2206.02443 | null | https://arxiv.org/abs/2206.02443v2 | https://arxiv.org/pdf/2206.02443v2.pdf | Spam Detection Using BERT | Emails and SMSs are the most popular tools in today communications, and as the increase of emails and SMSs users are increase, the number of spams is also increases. Spam is any kind of unwanted, unsolicited digital communication that gets sent out in bulk, spam emails and SMSs are causing major resource wastage by unnecessarily flooding the network links. Although most spam mail originate with advertisers looking to push their products, some are much more malicious in their intent like phishing emails that aims to trick victims into giving up sensitive information like website logins or credit card information this type of cybercrime is known as phishing. To countermeasure spams, many researches and efforts are done to build spam detectors that are able to filter out messages and emails as spam or ham. In this research we build a spam detector using BERT pre-trained model that classifies emails and messages by understanding to their context, and we trained our spam detector model using multiple corpuses like SMS collection corpus, Enron corpus, SpamAssassin corpus, Ling-Spam corpus and SMS spam collection corpus, our spam detector performance was 98.62%, 97.83%, 99.13% and 99.28% respectively. Keywords: Spam Detector, BERT, Machine learning, NLP, Transformer, Enron Corpus, SpamAssassin Corpus, SMS Spam Detection Corpus, Ling-Spam Corpus. | ['Dr. Mohammad Mikki', 'Thaer Sahmoud'] | 2022-06-06 | null | null | null | null | ['spam-detection'] | ['natural-language-processing'] | [ 2.79617961e-02 -1.16740309e-01 -3.97399738e-02 -3.07003975e-01
3.45233679e-02 -7.08851695e-01 1.09713566e+00 1.50205027e-02
-2.64072299e-01 8.48475456e-01 1.81061029e-01 -7.22682953e-01
2.26483867e-01 -1.13389719e+00 -8.56050104e-02 -1.54453471e-01
1.95499152e-01 6.17864430e-01 9.05463219e-01 -6.29042327e-01
9.94967103e-01 4.46279585e-01 -1.05474567e+00 5.12454510e-01
7.68380105e-01 7.00909495e-01 -2.19798144e-02 7.30725884e-01
-7.87510872e-01 7.10657001e-01 -9.98476505e-01 -5.76899529e-01
1.91440329e-01 -4.61992502e-01 -1.03889620e+00 5.39968675e-03
-2.95644905e-03 -3.01574022e-01 -4.39508736e-01 1.27526546e+00
1.53130308e-01 -1.51661500e-01 5.67186058e-01 -1.38226879e+00
-4.98905063e-01 4.20694828e-01 -5.49766541e-01 4.28352118e-01
3.79307389e-01 2.76068989e-02 1.84287474e-01 -3.31859291e-01
5.98643422e-01 1.72503555e+00 6.31573021e-01 7.16508150e-01
-9.12226260e-01 -9.72690403e-01 -3.75123799e-01 -5.73354065e-02
-6.10503316e-01 -2.41773114e-01 3.13663542e-01 -3.69251192e-01
6.08419895e-01 5.65570474e-01 2.82722920e-01 1.31563985e+00
3.19262087e-01 6.60700858e-01 1.28397346e+00 -1.25570402e-01
9.29668993e-02 8.40980589e-01 8.07517290e-01 1.37400866e-01
3.42627138e-01 1.36742577e-01 -1.99875489e-01 -6.05828285e-01
3.22135687e-01 2.41792053e-01 1.54960707e-01 5.72124779e-01
-4.78109837e-01 1.28035367e+00 1.37006745e-01 5.50742745e-01
-3.78635302e-02 -2.76039153e-01 9.39333320e-01 7.52699614e-01
4.25441027e-01 1.34637728e-01 -6.14670932e-01 -2.23758638e-01
-5.10565758e-01 1.80850729e-01 1.31888080e+00 9.68804657e-01
4.75372314e-01 -1.81102544e-01 3.69694561e-01 8.28322351e-01
4.57573891e-01 6.43967390e-01 1.01659739e+00 -2.49360532e-01
4.68049973e-01 1.15251064e+00 2.58500367e-01 -1.30913937e+00
-3.49554837e-01 -1.82885788e-02 -5.84267139e-01 -1.69975117e-01
2.98448652e-01 -1.84187993e-01 -8.61797988e-01 1.02425992e+00
1.40673667e-01 8.59959722e-02 -4.00326848e-01 7.18957901e-01
9.11316216e-01 7.98099220e-01 4.17901456e-01 -2.30546027e-01
1.55479836e+00 -6.18693173e-01 -5.38819313e-01 -3.81552786e-01
4.70221758e-01 -1.30910122e+00 8.67034614e-01 2.78357506e-01
-6.05909348e-01 -1.11185744e-01 -4.90674704e-01 5.48938870e-01
-1.00844765e+00 -4.08827186e-01 6.35067165e-01 1.29504204e+00
-3.35012227e-01 6.68460250e-01 -3.15912127e-01 -8.78522098e-01
6.06311977e-01 4.82053846e-01 -1.60637975e-01 2.87148446e-01
-1.38925791e+00 1.02231622e+00 4.24617320e-01 -4.49675620e-01
-4.48429942e-01 -1.90682366e-01 -1.25691935e-01 -7.60901123e-02
2.66931742e-01 -1.83525473e-01 1.23920012e+00 -1.30839753e+00
-9.91572559e-01 9.67881560e-01 -1.87778801e-01 -6.54747784e-01
5.17937601e-01 2.51467258e-01 -1.05379343e+00 3.61330099e-02
1.88998058e-01 -1.97621018e-01 1.05345571e+00 -1.02974224e+00
-7.86243856e-01 -5.29710889e-01 -5.18734872e-01 -4.86468166e-01
-2.69592404e-01 7.70451367e-01 5.12893438e-01 -3.51482540e-01
3.27031612e-01 -5.86990416e-01 3.91896032e-02 -1.06103766e+00
-4.79282796e-01 -1.93228737e-01 1.76257122e+00 -8.79538894e-01
1.45548332e+00 -1.73646879e+00 -8.87032688e-01 5.60937762e-01
-8.55462532e-03 9.90564525e-01 2.31297776e-01 9.33266521e-01
2.10164323e-01 5.57812572e-01 -8.49756524e-02 4.86507803e-01
-1.55346230e-01 8.34906921e-02 -4.89970535e-01 2.70062178e-01
-1.44210532e-01 6.86986983e-01 -9.97076988e-01 -4.29019302e-01
3.36645275e-01 -9.05464068e-02 -1.02562252e-02 -8.48676264e-02
4.58699837e-02 4.57191646e-01 -8.74662101e-01 8.04604411e-01
9.87512290e-01 -1.36863455e-01 -1.88575424e-02 1.74421206e-01
1.17657699e-01 5.66135049e-01 -9.97193217e-01 3.79554272e-01
-4.62435514e-01 5.44970810e-01 2.83842236e-01 -1.17939770e+00
1.07258105e+00 1.63369715e-01 7.69172329e-03 -3.42610806e-01
7.85655797e-01 6.48584127e-01 -1.15290567e-01 -7.36353517e-01
3.18942636e-01 -1.76185489e-01 5.77160567e-02 5.21108389e-01
-2.42593557e-01 3.02129507e-01 2.86343873e-01 6.61873162e-01
1.48320615e+00 -5.55174530e-01 3.61346394e-01 -1.49406463e-01
1.30522656e+00 9.53102112e-02 2.01653559e-02 6.47011697e-01
-4.29494381e-01 8.11052471e-02 5.60680687e-01 -2.64957637e-01
-1.01722038e+00 -1.03843713e+00 -3.63644123e-01 8.96163344e-01
2.95789868e-01 -1.86244488e-01 -7.59289443e-01 -1.29636455e+00
4.00893182e-01 5.84033847e-01 1.61568061e-01 -2.74009317e-01
-7.10434556e-01 -8.91985714e-01 6.07933044e-01 -1.98265031e-01
1.02522933e+00 -1.33649695e+00 2.54266083e-01 4.96097088e-01
1.43970437e-02 -8.56181145e-01 -2.84309119e-01 -4.05184954e-01
-1.07088804e+00 -1.37792444e+00 -2.14859620e-01 -9.21488166e-01
7.79227495e-01 4.38024729e-01 9.27212477e-01 3.78049046e-01
-4.20586675e-01 -1.39401808e-01 -6.43903971e-01 -3.35376769e-01
-1.19311810e+00 4.79666665e-02 -5.14207361e-03 -3.52406385e-03
1.24320173e+00 -6.65240526e-01 -5.22249818e-01 8.01029205e-01
-7.87940741e-01 -5.43512464e-01 5.20742297e-01 8.90808523e-01
-9.03932631e-01 1.23272814e-01 1.40128756e+00 -1.47579086e+00
1.04687381e+00 -9.87140656e-01 -4.53387588e-01 -2.66507000e-01
-6.33202791e-01 -4.43722367e-01 9.99835849e-01 -4.21211302e-01
-1.20527470e+00 -2.69970417e-01 -2.62707412e-01 7.44942963e-01
-6.65372670e-01 -1.76775619e-01 -6.20053010e-03 -2.00368136e-01
1.00250411e+00 4.85036701e-01 3.44996125e-01 -8.22527409e-01
1.02890640e-01 1.81155765e+00 -1.77156031e-02 1.10125676e-01
8.91651332e-01 4.30566430e-01 -3.38704437e-01 -1.08919418e+00
-2.86307782e-01 -1.21724021e+00 -1.49854898e-01 -1.31353796e-01
3.26295316e-01 -1.08993821e-01 -1.29049146e+00 7.61159182e-01
-1.39536488e+00 7.59370029e-01 4.52461421e-01 6.31398261e-02
-3.74219120e-02 8.84160459e-01 -1.00015986e+00 -1.27942502e+00
-5.38435400e-01 -5.12547076e-01 5.72943687e-01 2.83111423e-01
-5.53327441e-01 -9.10725355e-01 -4.15252477e-01 6.67547047e-01
7.88907051e-01 -2.68726349e-01 7.02534676e-01 -1.47987235e+00
-5.85463703e-01 -9.20811772e-01 -7.52091467e-01 6.60706282e-01
3.86892110e-01 -4.46612895e-01 -5.73415577e-01 1.30813405e-01
3.39832515e-01 3.49407084e-02 6.28901362e-01 -8.80750865e-02
7.65251517e-01 -8.48462284e-01 -6.27504945e-01 -9.44937393e-02
1.26965952e+00 6.20785236e-01 9.44149613e-01 5.92962980e-01
-1.27389468e-02 7.35503972e-01 4.52242643e-01 2.52641767e-01
-3.38786036e-01 -2.00520176e-02 3.95113587e-01 6.06679142e-01
3.06117445e-01 -3.96247417e-01 4.18344885e-01 5.89963913e-01
3.24713618e-01 -1.01993255e-01 -5.96289158e-01 2.41252705e-01
-1.77690172e+00 -1.20593035e+00 -9.99078035e-01 2.03542256e+00
7.20516801e-01 3.25354099e-01 4.40787464e-01 1.75837591e-01
1.22116888e+00 -1.36664093e-01 -5.39236665e-02 -6.38918817e-01
1.31256446e-01 2.88466066e-01 1.11601305e+00 6.03954911e-01
-1.15219879e+00 1.36101139e+00 5.57469940e+00 1.11940634e+00
-8.04399371e-01 2.91269451e-01 4.49498653e-01 4.21210110e-01
1.19720578e-01 3.83371621e-01 -1.20929277e+00 1.43452847e+00
1.07510352e+00 -1.70028195e-01 3.94412190e-01 9.02626872e-01
7.25879729e-01 -5.52378595e-01 -3.11192453e-01 7.44338572e-01
-1.09613597e-01 -1.29667175e+00 7.32094795e-02 -7.24762678e-02
4.24397498e-01 -2.26407543e-01 -5.56948841e-01 3.38231564e-01
2.73992687e-01 -7.62986720e-01 -1.81641355e-01 1.54625386e-01
-3.07401627e-01 -6.16850913e-01 1.27165973e+00 7.21135795e-01
-4.99765337e-01 -2.63709009e-01 -4.43644494e-01 -1.98162913e-01
6.38484180e-01 7.67314255e-01 -1.15346670e+00 1.38117552e-01
3.67649436e-01 2.98127055e-01 -6.60081029e-01 1.12519038e+00
4.94017079e-02 9.64587092e-01 -4.10853475e-01 -8.81585002e-01
3.89534295e-01 -5.56787610e-01 7.60961413e-01 1.39553416e+00
-4.13128734e-02 -1.89970329e-01 -1.26019835e-01 8.63943160e-01
1.86631650e-01 2.65922010e-01 -7.10756183e-01 -4.13531572e-01
5.37186325e-01 1.37653553e+00 -8.70010614e-01 -5.74895442e-01
-1.46511927e-01 8.35700452e-01 -5.94297230e-01 9.17587355e-02
-7.34404504e-01 -8.22872221e-01 3.15155387e-01 8.18444371e-01
-5.18848419e-01 -3.29937227e-02 -5.46146035e-01 -7.19602525e-01
-2.18223721e-01 -9.47653890e-01 1.14643514e-01 -2.69320965e-01
-1.80734217e+00 1.51567161e-01 -3.57477188e-01 -1.04159594e+00
7.15882331e-02 -8.98854613e-01 -1.00781476e+00 1.01178432e+00
-8.10081482e-01 -9.52747643e-01 -9.30479765e-02 3.21278095e-01
7.01757252e-01 -9.54757452e-01 2.78681844e-01 5.84755540e-01
-1.28437415e-01 5.23193553e-02 2.06238344e-01 2.01558948e-01
6.21571600e-01 -7.25459158e-01 6.40889883e-01 2.72008657e-01
-5.95203817e-01 1.11789322e+00 7.69110382e-01 -1.14815378e+00
-9.99542415e-01 -7.54810035e-01 1.21118784e+00 -6.94517493e-01
1.12906551e+00 -3.79560143e-01 -7.94948041e-01 3.68177444e-01
-1.65939346e-01 -8.52038383e-01 1.77963465e-01 -1.23668320e-01
-1.47011578e-02 8.11776593e-02 -1.62016344e+00 4.64850098e-01
7.72646010e-01 -2.02335626e-01 -6.90390468e-01 9.84358966e-01
3.81208271e-01 3.04364562e-01 -1.02137104e-01 -2.01167479e-01
3.31059396e-01 -1.25866532e+00 8.08838069e-01 -1.09499419e+00
1.45532534e-01 -4.16856818e-02 2.71147579e-01 -7.87517726e-01
8.67647082e-02 -7.39544034e-01 9.70243663e-02 1.48701596e+00
3.04996997e-01 -1.21130145e+00 1.36239254e+00 3.62792671e-01
2.90755153e-01 -8.12686458e-02 -7.11765766e-01 -9.92514312e-01
-8.88978690e-02 -1.88162148e-01 4.34513211e-01 1.03548849e+00
2.89993227e-01 6.23194575e-01 -4.39844280e-01 -3.44432145e-01
5.56981921e-01 -6.04660153e-01 8.34073067e-01 -1.37353241e+00
9.91616249e-02 -5.73150814e-01 -5.84651411e-01 -1.11822903e+00
-5.52842282e-02 -6.85537994e-01 -5.51732421e-01 -1.17721236e+00
-1.32694185e-01 -3.84520024e-01 6.74912155e-01 -1.18485227e-01
2.87350506e-01 -9.69766751e-02 -8.47513601e-02 2.46547446e-01
-1.26544803e-01 -9.58030373e-02 9.38831627e-01 -1.50056124e-01
-2.12879270e-01 8.56027961e-01 -3.92934501e-01 9.77896929e-01
1.09442496e+00 -5.63267648e-01 -5.00367098e-02 6.60058558e-01
2.13236004e-01 -2.12131247e-01 3.92863393e-01 -3.31124991e-01
1.98347583e-01 -3.58575940e-01 -2.08230317e-02 -5.43595254e-01
-1.35438163e-02 -9.20555770e-01 -2.89239734e-01 7.87235796e-01
1.17235139e-01 -5.10836780e-01 -3.24006945e-01 6.73727632e-01
-1.32542431e-01 -1.00601375e+00 1.09730983e+00 -5.63157618e-01
-6.09545588e-01 -5.21851629e-02 -9.42304671e-01 5.46907680e-03
1.08920169e+00 -4.41408128e-01 -9.20957804e-01 -5.46836436e-01
-6.09750211e-01 2.91962147e-01 1.18486192e-02 7.34586775e-01
4.10440773e-01 -1.02174878e+00 -3.18049580e-01 3.56270999e-01
-3.54010999e-01 -7.51379013e-01 -7.46281743e-02 8.27198982e-01
-9.06707883e-01 6.37092948e-01 -1.21863328e-01 -6.35028854e-02
-1.53981864e+00 2.56587356e-01 8.59177485e-02 7.99478665e-02
-3.33100915e-01 5.68250954e-01 -3.89683545e-01 -8.01381469e-01
1.91873848e-01 2.95026779e-01 -3.01936448e-01 -1.03940569e-01
6.62245989e-01 1.18580008e+00 -1.65437996e-01 -7.01091468e-01
-3.10472518e-01 -5.69571927e-02 -2.55410761e-01 2.04714686e-01
6.70365572e-01 -7.05625787e-02 -7.62915313e-01 -1.23631500e-01
1.45481896e+00 3.68302539e-02 1.54537842e-01 1.83100492e-01
7.21546173e-01 -1.00270510e+00 -5.51451683e-01 -9.69866991e-01
-4.46895242e-01 5.58876336e-01 4.51499283e-01 1.19310355e+00
3.62145901e-01 -3.04487217e-02 1.49581468e+00 2.98929185e-01
4.15748924e-01 -1.58733249e+00 -1.09854244e-01 7.69446254e-01
5.33756554e-01 -1.71093631e+00 -2.34342039e-01 -8.61297846e-01
-4.52996999e-01 1.10465372e+00 6.89448953e-01 -4.28815156e-01
1.02599871e+00 -2.35635955e-02 -1.61854014e-01 -1.53439388e-01
-3.01536471e-01 3.19579393e-02 -2.82664567e-01 7.21332133e-01
4.14047986e-01 -5.67103326e-02 -1.38100815e+00 7.84153223e-01
-3.00461233e-01 -6.99617416e-02 6.85845792e-01 9.29735184e-01
-1.45097828e+00 -1.39191210e+00 -6.69308841e-01 1.18669784e+00
-6.78673625e-01 -4.25959639e-02 -1.02429926e+00 7.34086096e-01
2.32821740e-02 1.41878092e+00 -1.58051535e-01 -5.29736578e-01
2.93077648e-01 3.84012997e-01 -2.42134273e-01 -6.57144129e-01
-9.56804514e-01 -4.72244680e-01 6.18337393e-01 -7.27180466e-02
5.01601920e-02 -3.11426640e-01 -1.19803083e+00 -1.18500590e+00
-7.74304807e-01 4.82691437e-01 1.05242920e+00 9.90825534e-01
-4.30538468e-02 -1.56930730e-01 8.97662461e-01 -8.62689465e-02
-1.07634127e+00 -1.31100094e+00 -7.63219655e-01 7.82552660e-01
-6.89864457e-02 -3.46354097e-01 -9.09039199e-01 -2.55749017e-01] | [7.853854179382324, 10.03609561920166] |
1838ea6e-d8ac-4e1f-b9cc-6feefe355709 | 3d-mpa-multi-proposal-aggregation-for-3d | 2003.13867 | null | https://arxiv.org/abs/2003.13867v1 | https://arxiv.org/pdf/2003.13867v1.pdf | 3D-MPA: Multi Proposal Aggregation for 3D Semantic Instance Segmentation | We present 3D-MPA, a method for instance segmentation on 3D point clouds. Given an input point cloud, we propose an object-centric approach where each point votes for its object center. We sample object proposals from the predicted object centers. Then, we learn proposal features from grouped point features that voted for the same object center. A graph convolutional network introduces inter-proposal relations, providing higher-level feature learning in addition to the lower-level point features. Each proposal comprises a semantic label, a set of associated points over which we define a foreground-background mask, an objectness score and aggregation features. Previous works usually perform non-maximum-suppression (NMS) over proposals to obtain the final object detections or semantic instances. However, NMS can discard potentially correct predictions. Instead, our approach keeps all proposals and groups them together based on the learned aggregation features. We show that grouping proposals improves over NMS and outperforms previous state-of-the-art methods on the tasks of 3D object detection and semantic instance segmentation on the ScanNetV2 benchmark and the S3DIS dataset. | ['Matthias Nießner', 'Martin Bokeloh', 'Francis Engelmann', 'Bastian Leibe', 'Alireza Fathi'] | 2020-03-30 | null | null | null | null | ['3d-instance-segmentation-1', '3d-semantic-instance-segmentation'] | ['computer-vision', 'computer-vision'] | [ 1.88857272e-01 4.07940894e-01 -2.60827363e-01 -7.25606322e-01
-7.75549412e-01 -4.26545978e-01 5.97003222e-01 6.68813407e-01
-2.37756789e-01 -7.85462931e-03 -4.63886738e-01 -8.02155361e-02
4.68856618e-02 -8.94256115e-01 -1.11287546e+00 -4.95372295e-01
-3.11474085e-01 1.22415757e+00 1.27179015e+00 3.69877875e-01
4.79155451e-01 1.04764414e+00 -1.75688303e+00 4.98345107e-01
6.23117805e-01 1.40141284e+00 4.48267668e-01 3.64753008e-01
-6.86401248e-01 1.75264731e-01 -5.53881764e-01 -7.66383931e-02
4.89620000e-01 2.60055035e-01 -8.20599139e-01 5.28668523e-01
8.02152932e-01 -1.46977022e-01 2.76688844e-01 9.91203785e-01
-4.62797992e-02 1.13547347e-01 8.30532134e-01 -1.33341992e+00
-1.28031284e-01 3.14923525e-01 -8.59535992e-01 4.76812273e-02
-2.03089491e-02 2.38754779e-01 1.21614110e+00 -1.35538840e+00
7.57389665e-01 1.50655329e+00 5.00827610e-01 3.45940977e-01
-1.34142447e+00 -4.99529809e-01 5.15969336e-01 -7.20246285e-02
-1.19846833e+00 5.63530624e-02 7.22531080e-01 -4.84585434e-01
1.03548765e+00 3.40354204e-01 9.68380332e-01 2.40819454e-01
-2.33463109e-01 1.21477997e+00 7.43392229e-01 -9.51637179e-02
4.59968984e-01 -3.33319977e-02 5.77338934e-01 8.26409161e-01
3.46913636e-01 -2.84320265e-01 -3.43072832e-01 -3.39408427e-01
6.46749914e-01 1.75007939e-01 1.45005599e-01 -1.02238941e+00
-1.23013377e+00 7.09769487e-01 9.71738994e-01 -2.66902987e-03
-5.39914787e-01 2.33296603e-01 4.64529730e-02 -2.05914259e-01
7.34363019e-01 1.79666027e-01 -7.59373307e-01 6.62580729e-01
-1.21436465e+00 5.38964868e-01 6.27125800e-01 1.06611753e+00
1.25636578e+00 -4.96029496e-01 -3.97121817e-01 5.50472379e-01
5.95049739e-01 3.24407846e-01 -3.80761236e-01 -1.06101632e+00
3.28876317e-01 1.23441374e+00 1.88821673e-01 -6.82039499e-01
-5.28011680e-01 -5.42836785e-01 -2.01112181e-01 5.37539661e-01
2.10854754e-01 5.28343797e-01 -1.59508121e+00 1.12577605e+00
9.16635513e-01 4.50192302e-01 -4.11664814e-01 1.16836762e+00
1.09712446e+00 6.24391139e-01 2.19606627e-02 1.47885323e-01
1.22382975e+00 -1.01839852e+00 9.93623585e-02 -3.22620630e-01
4.66267139e-01 -6.51448905e-01 6.74795389e-01 2.29694381e-01
-1.21416616e+00 -6.77660525e-01 -7.35004842e-01 -1.04862768e-02
-2.70472050e-01 1.28672525e-01 5.40078759e-01 2.15722993e-01
-9.88392711e-01 6.82827294e-01 -1.11893320e+00 -2.54423261e-01
1.05544877e+00 6.57397985e-01 -1.39630869e-01 5.55344149e-02
-2.22891748e-01 7.24769473e-01 6.45620763e-01 -1.54595792e-01
-8.70837390e-01 -8.49055529e-01 -7.04528928e-01 -1.19286999e-02
4.78206575e-01 -7.89143026e-01 1.17985630e+00 -7.20045269e-01
-1.03074884e+00 1.49249351e+00 -2.56472737e-01 -5.59431970e-01
4.24121946e-01 -1.01101264e-01 1.33378759e-01 9.42729264e-02
4.14594501e-01 1.36155534e+00 9.12015140e-01 -1.68080091e+00
-1.16994667e+00 -5.95387638e-01 -1.46882474e-01 8.14038590e-02
5.86902559e-01 -1.63252711e-01 -9.28653657e-01 3.45179997e-02
1.05330694e+00 -9.05593812e-01 -4.66438353e-01 1.98051810e-01
-8.29996526e-01 -8.47329140e-01 1.01202428e+00 1.85131337e-02
4.90758657e-01 -2.10911036e+00 -1.22825000e-02 5.10952652e-01
3.83809835e-01 6.35994822e-02 -9.04930830e-02 -2.24614412e-01
2.47006968e-01 8.16754028e-02 -3.20907831e-01 -6.92564845e-01
9.24115628e-02 2.96681046e-01 -1.48608938e-01 5.34772992e-01
6.27835989e-01 8.60750854e-01 -8.82297695e-01 -7.01391101e-01
6.68885887e-01 2.22203270e-01 -7.17951953e-01 9.44962054e-02
-7.83690274e-01 3.05156887e-01 -7.05584526e-01 8.25707138e-01
1.06018996e+00 -5.00769973e-01 -2.19371021e-01 -2.24082693e-01
-1.88559815e-01 5.11098504e-01 -1.29858673e+00 1.71010411e+00
1.51536269e-02 3.65792841e-01 9.00636017e-02 -1.02887046e+00
1.14005744e+00 -3.02998722e-01 6.35332704e-01 -1.12345114e-01
-2.64587104e-02 4.97558743e-01 -2.29686037e-01 -1.31718844e-01
5.07103801e-01 2.51614481e-01 1.14683196e-01 9.06431824e-02
2.26614028e-01 -5.17593980e-01 -1.98605247e-02 2.54365116e-01
1.04959750e+00 3.29271495e-01 1.31852135e-01 -2.28475735e-01
3.38030875e-01 4.88200039e-01 4.66560990e-01 1.02249765e+00
-3.33781689e-02 8.85870993e-01 5.07184625e-01 -5.73099375e-01
-8.58458519e-01 -1.16205597e+00 -2.93862492e-01 9.32387352e-01
7.37565339e-01 -3.17214131e-01 -6.54422820e-01 -1.17451441e+00
5.03418386e-01 8.39503467e-01 -5.23553193e-01 2.08419263e-01
-6.25644207e-01 -2.21005723e-01 -9.99680609e-02 3.99016589e-01
7.69033879e-02 -1.11105597e+00 -8.55586052e-01 1.94957986e-01
3.50747466e-01 -9.96470273e-01 -1.12336919e-01 5.29390216e-01
-1.21391582e+00 -1.05517721e+00 -2.97737569e-01 -8.04896414e-01
9.65918720e-01 4.56395596e-01 1.36052966e+00 2.21833169e-01
-3.37980539e-01 1.47570401e-01 -1.38779953e-01 -5.62171102e-01
-2.48674303e-01 1.60187736e-01 -2.90916771e-01 -1.32754520e-01
6.38843298e-01 -2.25463122e-01 -6.05402410e-01 3.41906786e-01
-5.44241846e-01 6.21690042e-02 6.79895103e-01 4.28327978e-01
1.36801541e+00 -3.93250316e-01 1.38805479e-01 -8.70112240e-01
-2.64741242e-01 -3.10782343e-01 -8.77488852e-01 7.24596009e-02
-1.83648527e-01 -1.73166059e-02 -2.61948228e-01 -1.82258323e-01
-5.18746138e-01 5.53814590e-01 1.02692723e-01 -8.35069120e-01
-6.57376587e-01 4.83799167e-02 -1.55629694e-01 5.74559085e-02
5.03608167e-01 -1.82043072e-02 -3.85377735e-01 -6.57941043e-01
5.85125148e-01 3.85898411e-01 3.17646235e-01 -4.69535202e-01
8.37188125e-01 7.53324151e-01 1.51375577e-01 -6.89054906e-01
-1.09285879e+00 -9.12165463e-01 -1.10587096e+00 -2.59005755e-01
1.07073867e+00 -8.29758942e-01 -6.06089413e-01 1.95489377e-02
-1.54704309e+00 -9.08954814e-02 -6.95480406e-01 2.97952920e-01
-5.05243003e-01 -1.05642058e-01 -1.85054734e-01 -8.59165013e-01
-1.28122136e-01 -1.29878724e+00 1.78579080e+00 9.33814719e-02
-1.08313330e-01 -4.12448764e-01 -3.84310156e-01 3.01193297e-01
-2.19628319e-01 1.98245287e-01 8.18466663e-01 -1.03244519e+00
-1.28963041e+00 -2.47348398e-01 -4.47282434e-01 9.99412015e-02
-2.30280772e-01 1.25344798e-01 -8.05979311e-01 -1.72305964e-02
-2.26258501e-01 -3.62385847e-02 1.37729311e+00 6.29936337e-01
1.47225606e+00 4.94316407e-02 -1.00198162e+00 4.86407787e-01
1.21931088e+00 -1.60134330e-01 1.36075526e-01 1.67841643e-01
7.87644565e-01 5.71005523e-01 7.78375328e-01 2.66263247e-01
2.05691159e-01 6.48656011e-01 1.09728980e+00 -2.84492135e-01
-1.17875643e-01 -1.78437278e-01 -1.69214934e-01 6.91187102e-03
4.55513708e-02 -2.07771569e-01 -1.06708467e+00 6.70783639e-01
-1.87816024e+00 -5.14638603e-01 -6.98603570e-01 1.98669040e+00
3.78766626e-01 6.00835204e-01 2.52159834e-01 -7.69545585e-02
9.46448028e-01 9.57206711e-02 -7.74237752e-01 1.33276716e-01
7.94819146e-02 3.25674593e-01 6.92072809e-01 2.71309227e-01
-1.42614233e+00 1.23846161e+00 4.85005522e+00 6.68294489e-01
-8.27789307e-01 1.85244620e-01 5.20932794e-01 -6.86578378e-02
-1.95114911e-01 1.78850114e-01 -1.21584177e+00 9.75091904e-02
1.37029335e-01 4.75896269e-01 -2.62516230e-01 1.24944758e+00
-1.15293212e-01 -2.26412952e-01 -1.42272532e+00 7.01570213e-01
-1.33642361e-01 -1.57655907e+00 2.66300559e-01 1.13333985e-01
8.05392385e-01 5.98146200e-01 -1.18485466e-01 4.26099934e-02
2.42325351e-01 -6.96760952e-01 1.10758197e+00 3.60945702e-01
2.00881019e-01 -6.61288500e-01 4.52295154e-01 5.65241098e-01
-1.20240235e+00 1.04298912e-01 -6.58245027e-01 3.11156154e-01
9.67803970e-02 8.32641423e-01 -1.28642213e+00 4.03252751e-01
7.42043793e-01 7.32778251e-01 -4.57815588e-01 1.41185701e+00
-2.89026558e-01 6.22203946e-01 -7.52577364e-01 -6.38566464e-02
5.41865170e-01 -1.37261927e-01 8.58974814e-01 1.12127280e+00
2.08050177e-01 6.77355453e-02 7.88199127e-01 1.37349641e+00
-2.30177008e-02 2.22279299e-02 -2.61895776e-01 5.23684382e-01
5.56554139e-01 1.35679543e+00 -1.47714996e+00 -5.80567777e-01
-3.02036405e-01 7.66745269e-01 3.32450598e-01 -9.56898332e-02
-5.84686875e-01 4.61879410e-02 6.88660741e-01 3.57925832e-01
7.96614587e-01 -2.45260671e-01 -5.93048513e-01 -7.70796895e-01
4.39068265e-02 -3.30573358e-02 2.94719845e-01 -6.57140911e-01
-1.22994578e+00 2.32334524e-01 -5.15471734e-02 -1.33173537e+00
2.51570672e-01 -5.75278938e-01 -6.23840868e-01 7.40274549e-01
-1.44389391e+00 -1.29696631e+00 -3.23423803e-01 8.48562419e-02
8.56630325e-01 2.84417987e-01 2.81158805e-01 -1.57574624e-01
-1.52906716e-01 -7.45295361e-02 -3.15336913e-01 3.23940217e-02
1.38118416e-01 -1.43843842e+00 7.38006592e-01 4.28321749e-01
4.65180188e-01 2.19451636e-01 4.84671295e-01 -8.68065357e-01
-9.89354491e-01 -1.48234403e+00 7.17745781e-01 -7.01639712e-01
3.05909067e-01 -5.22005618e-01 -1.15300608e+00 5.90642154e-01
-3.75430137e-01 5.89014113e-01 8.78013074e-02 -8.00273344e-02
-2.59084441e-02 9.65600982e-02 -1.11939263e+00 2.08987296e-01
1.31014621e+00 7.64138922e-02 -7.66133606e-01 6.74465358e-01
1.04735672e+00 -6.71767056e-01 -5.01770854e-01 5.53670347e-01
8.29576179e-02 -1.00244832e+00 1.21011412e+00 -6.51472211e-01
1.96693823e-01 -7.40892470e-01 -1.26587644e-01 -8.70642841e-01
-3.62524897e-01 3.92460525e-02 -5.76125309e-02 8.84283245e-01
5.62561572e-01 -7.67395943e-02 1.22897553e+00 3.74962956e-01
-6.61443889e-01 -8.24033916e-01 -1.08372462e+00 -8.78546476e-01
-2.31781214e-01 -8.04054677e-01 7.97537446e-01 4.90934104e-01
-8.11201870e-01 1.76912874e-01 3.79092723e-01 7.23666370e-01
1.01543844e+00 6.28243744e-01 9.66272891e-01 -1.81986964e+00
1.70069829e-01 -5.63938916e-01 -7.26857722e-01 -1.30148542e+00
2.55338132e-01 -1.25954258e+00 2.76128054e-01 -1.91081583e+00
9.09285247e-02 -8.61078322e-01 -2.23236054e-01 6.54224277e-01
-6.85179234e-02 2.44174927e-01 2.58855909e-01 4.55525875e-01
-1.07334757e+00 2.49310449e-01 1.12514544e+00 -2.73185045e-01
-6.25555754e-01 4.89979953e-01 7.70301446e-02 9.58701789e-01
5.61872959e-01 -8.36073816e-01 7.16980770e-02 -3.21420401e-01
-1.36550128e-01 -1.60833076e-01 8.07948172e-01 -1.09051394e+00
2.17153713e-01 -2.49440581e-01 6.33464813e-01 -1.60290039e+00
5.14759362e-01 -8.63967299e-01 2.46677315e-03 5.82053959e-01
-2.19523057e-01 -6.33040249e-01 -7.43631870e-02 6.14824176e-01
1.08810320e-01 -2.04040259e-01 8.48359883e-01 -3.19150954e-01
-8.58621001e-01 5.02921283e-01 1.76157311e-01 -3.13677728e-01
1.23430002e+00 -5.56406736e-01 -1.51116243e-02 4.76226866e-01
-9.81406987e-01 4.34274614e-01 6.16187811e-01 4.66375887e-01
7.90123761e-01 -1.01975942e+00 -5.88522136e-01 3.47757608e-01
3.04872006e-01 9.90734875e-01 -1.11385249e-02 6.67293489e-01
-2.90101230e-01 2.96558857e-01 2.62510270e-01 -1.53980327e+00
-1.28204906e+00 4.06541646e-01 2.87579238e-01 1.80850536e-01
-9.10330236e-01 1.19109833e+00 5.27823806e-01 -6.15637004e-01
1.64759681e-01 -7.94042826e-01 -1.12268843e-01 -1.38815213e-02
-1.41377673e-02 1.71301931e-01 2.54604787e-01 -6.07491970e-01
-7.52734900e-01 6.87003493e-01 -1.77426562e-02 8.99643600e-02
1.49731052e+00 2.49059930e-01 -1.15022711e-01 4.64104325e-01
8.63705218e-01 -4.00060773e-01 -1.41796041e+00 -5.12450635e-01
3.72543424e-01 -6.12833023e-01 1.06594555e-01 -6.11758351e-01
-1.14291978e+00 7.74410009e-01 5.85885584e-01 3.33572298e-01
6.09781921e-01 1.03805113e+00 3.61076713e-01 3.51707965e-01
4.36276972e-01 -8.11033607e-01 1.84620153e-02 5.18656433e-01
6.85627818e-01 -1.28381765e+00 8.30961242e-02 -8.28831613e-01
-2.37296179e-01 1.05668998e+00 8.41509640e-01 -4.05984282e-01
6.91212654e-01 -9.95145515e-02 -2.90958881e-01 -6.54971123e-01
-5.76392770e-01 -6.26546860e-01 6.87346697e-01 5.23899317e-01
-4.46134470e-02 2.38578498e-01 4.68798615e-02 4.26637560e-01
-1.25800231e-02 -3.82966876e-01 7.94128925e-02 7.87225544e-01
-1.16773391e+00 -7.81360030e-01 -5.60084522e-01 8.36499751e-01
-7.11773932e-02 3.29766005e-01 -5.44015706e-01 7.77408004e-01
5.25054574e-01 4.71129119e-01 7.71456361e-01 -2.03875139e-01
5.48099458e-01 6.37702690e-03 2.58124471e-01 -1.20019150e+00
-4.92538780e-01 1.92866325e-01 -1.77154124e-01 -7.00039744e-01
-5.19678235e-01 -9.06618416e-01 -1.73495889e+00 2.68267155e-01
-6.74947202e-01 -4.75146156e-03 1.02072287e+00 9.17143047e-01
4.91314620e-01 3.66228521e-01 5.36918819e-01 -1.46482778e+00
-1.20402083e-01 -7.09072828e-01 -4.68071818e-01 3.83273214e-01
1.68365881e-01 -7.80361354e-01 -2.74077654e-01 -1.82793766e-01] | [7.966889381408691, -3.1372084617614746] |
3e530dc1-419a-41e6-a781-dd838ae86762 | a-non-asymptotic-analysis-of-oversmoothing-in | 2212.10701 | null | https://arxiv.org/abs/2212.10701v2 | https://arxiv.org/pdf/2212.10701v2.pdf | A Non-Asymptotic Analysis of Oversmoothing in Graph Neural Networks | Oversmoothing is a central challenge of building more powerful Graph Neural Networks (GNNs). While previous works have only demonstrated that oversmoothing is inevitable when the number of graph convolutions tends to infinity, in this paper, we precisely characterize the mechanism behind the phenomenon via a non-asymptotic analysis. Specifically, we distinguish between two different effects when applying graph convolutions -- an undesirable mixing effect that homogenizes node representations in different classes, and a desirable denoising effect that homogenizes node representations in the same class. By quantifying these two effects on random graphs sampled from the Contextual Stochastic Block Model (CSBM), we show that oversmoothing happens once the mixing effect starts to dominate the denoising effect, and the number of layers required for this transition is $O(\log N/\log (\log N))$ for sufficiently dense graphs with $N$ nodes. We also extend our analysis to study the effects of Personalized PageRank (PPR), or equivalently, the effects of initial residual connections on oversmoothing. Our results suggest that while PPR mitigates oversmoothing at deeper layers, PPR-based architectures still achieve their best performance at a shallow depth and are outperformed by the graph convolution approach on certain graphs. Finally, we support our theoretical results with numerical experiments, which further suggest that the oversmoothing phenomenon observed in practice can be magnified by the difficulty of optimizing deep GNN models. | ['Ali Jadbabaie', 'William Wang', 'Zhengdao Chen', 'Xinyi Wu'] | 2022-12-21 | null | null | null | null | ['stochastic-block-model'] | ['graphs'] | [ 2.76804745e-01 4.61215019e-01 1.76257581e-01 -9.30642709e-02
-2.46948272e-01 -3.82369220e-01 4.64240670e-01 3.12508643e-01
-4.12849545e-01 3.51518214e-01 2.29468092e-01 -5.04636407e-01
-1.11686312e-01 -9.38246191e-01 -1.02929556e+00 -7.35848248e-01
-5.72304845e-01 4.46791053e-02 2.02352583e-01 -3.14631790e-01
-3.14818881e-02 4.97548044e-01 -1.16437924e+00 -6.51607811e-02
9.91195381e-01 7.32499123e-01 -1.58564392e-02 9.36860681e-01
-2.01161295e-01 9.30275679e-01 -3.94187629e-01 -5.64929307e-01
4.65797991e-01 -3.08454424e-01 -6.87616408e-01 7.39206597e-02
7.45702267e-01 -2.40570590e-01 -6.48158193e-01 1.56853151e+00
1.89608306e-01 2.43532836e-01 5.18550932e-01 -1.06700385e+00
-7.42463171e-01 1.25033915e+00 -6.52648926e-01 3.74867320e-01
-3.66191089e-01 2.53809690e-01 1.20287859e+00 -4.66635495e-01
5.43433785e-01 1.27663434e+00 1.09189451e+00 3.74385715e-01
-1.50305557e+00 -4.57179666e-01 5.54489613e-01 -2.15073973e-01
-1.30617011e+00 -2.38214627e-01 8.55295658e-01 -3.26384962e-01
7.31928468e-01 1.67762265e-01 5.43374598e-01 8.48987520e-01
1.07210241e-01 6.36275828e-01 7.63995230e-01 -1.93837151e-01
5.30377924e-02 -1.17183685e-01 5.85299969e-01 8.91442597e-01
7.31055558e-01 1.15253581e-02 -1.90492049e-01 3.16795036e-02
9.16378498e-01 -3.76139097e-02 -4.06168640e-01 -2.86636144e-01
-6.58025503e-01 8.22037280e-01 9.65624332e-01 2.97458321e-01
-5.42063892e-01 7.10780442e-01 2.56708741e-01 5.13023138e-01
6.94149673e-01 3.28605086e-01 -3.08554858e-01 1.40324652e-01
-1.06459832e+00 8.89229700e-02 8.49966168e-01 8.60443294e-01
9.79738235e-01 1.95671603e-01 3.55240796e-03 6.64820492e-01
6.01382963e-02 1.91146195e-01 2.43729241e-02 -8.66830349e-01
4.72513527e-01 6.15025520e-01 -1.27547413e-01 -1.26547194e+00
-4.53083158e-01 -8.93699884e-01 -1.32097304e+00 7.98760131e-02
9.65816736e-01 -3.45655590e-01 -9.99912441e-01 2.14804578e+00
-1.99374542e-01 -5.88020571e-02 -2.33646110e-01 7.13909566e-01
6.40270591e-01 5.21490037e-01 2.53716350e-01 4.67896797e-02
1.11412489e+00 -9.56829965e-01 -4.59405661e-01 -5.40760458e-01
7.60105371e-01 -3.18250954e-01 1.22388101e+00 1.91237569e-01
-1.35367239e+00 -2.09501296e-01 -1.06569171e+00 3.71115915e-02
-2.69234061e-01 -2.63779312e-01 8.41466904e-01 6.82371795e-01
-1.60340714e+00 1.27343965e+00 -8.58294308e-01 -4.38596636e-01
5.61661839e-01 2.98232824e-01 -8.00324753e-02 -2.73191839e-01
-1.07043934e+00 5.83755374e-01 -1.18025295e-01 3.21146965e-01
-7.01817334e-01 -8.13023508e-01 -6.27790213e-01 6.26323521e-01
2.30200574e-01 -4.89209533e-01 8.45369995e-01 -1.30198252e+00
-9.81755495e-01 5.91424584e-01 6.46495074e-02 -7.34454930e-01
5.29754996e-01 -4.28810976e-02 -9.59309191e-02 1.45955250e-01
-2.92800009e-01 5.03043592e-01 6.92998827e-01 -1.17403281e+00
-7.92298913e-02 -3.82934809e-01 3.20515871e-01 1.76778480e-01
-5.60235560e-01 -2.65513510e-01 -4.46236640e-01 -6.46699071e-01
8.99345726e-02 -8.90359461e-01 -6.73884034e-01 -1.58371001e-01
-4.61838514e-01 6.23014849e-03 2.39662796e-01 -5.95305622e-01
1.13371170e+00 -2.16939211e+00 -9.49171837e-03 4.96084034e-01
7.76494026e-01 1.03935376e-01 -4.02413756e-01 4.96272832e-01
-2.48424053e-01 3.61514062e-01 -1.53088659e-01 -2.64356166e-01
-2.95896791e-02 1.08767428e-01 -9.61903185e-02 4.38677281e-01
9.29301903e-02 8.75327826e-01 -8.16507459e-01 -4.02273089e-02
-1.93472281e-01 6.32023752e-01 -7.16049850e-01 -3.24142873e-01
-1.06136478e-01 -1.65208846e-01 -2.15526640e-01 6.34459555e-02
9.54169154e-01 -7.76359499e-01 4.38207895e-01 -1.81993619e-01
1.48676485e-01 3.50435048e-01 -1.20193982e+00 1.19146013e+00
-2.15541601e-01 6.67321086e-01 6.85613215e-01 -1.06309319e+00
5.38067937e-01 -5.65762706e-02 2.44156882e-01 -3.18228424e-01
1.41477168e-01 1.18831687e-01 2.73588002e-01 5.53488825e-03
5.48859775e-01 -2.62609482e-01 2.71674424e-01 4.42340910e-01
6.15773769e-03 3.78983945e-01 3.15116554e-01 6.44798934e-01
1.67888415e+00 -5.30536592e-01 -2.59597987e-01 -7.55565822e-01
-3.60780954e-02 -9.58178565e-02 2.16999710e-01 1.15003967e+00
-1.37182698e-01 2.61388451e-01 1.06960630e+00 -3.80526185e-01
-8.83809328e-01 -9.15711284e-01 2.87909389e-01 1.15536404e+00
1.50416851e-01 -3.51562858e-01 -9.20219123e-01 -7.08079100e-01
-7.48324841e-02 3.87041777e-01 -7.71057546e-01 -3.31687689e-01
-5.75874448e-01 -1.10484612e+00 4.94422376e-01 5.22382677e-01
4.58713382e-01 -8.53994846e-01 -6.33697510e-02 1.09268770e-01
1.60590157e-01 -9.41541493e-01 -6.92833066e-01 4.45887744e-01
-1.12593853e+00 -9.10620511e-01 -8.07995439e-01 -7.32063830e-01
1.10089839e+00 5.22682071e-01 1.34204423e+00 5.85884154e-01
6.52572587e-02 2.29952902e-01 -2.46077199e-02 -4.92410436e-02
-5.16634464e-01 4.38954979e-01 -2.89087683e-01 -1.80951610e-01
9.42405313e-02 -1.05428421e+00 -8.21722388e-01 -2.06595752e-02
-1.11057436e+00 -2.69573592e-02 8.22911799e-01 5.78749061e-01
-9.11687538e-02 3.90121043e-01 2.18843073e-01 -1.17043185e+00
8.54373336e-01 -2.99935669e-01 -6.39113307e-01 -7.67889321e-02
-7.76417017e-01 4.86469954e-01 7.98292935e-01 -5.50032437e-01
-6.29183650e-01 -2.23126188e-01 -1.04620263e-01 -3.93011332e-01
2.25627631e-01 7.93386459e-01 1.21050455e-01 -2.40706608e-01
7.05335140e-01 -3.69271589e-03 1.41461670e-01 -4.00750995e-01
4.43644375e-01 2.97944676e-02 2.97530800e-01 -3.79984021e-01
9.47456360e-01 5.17073154e-01 6.54096752e-02 -9.13371027e-01
-7.54214466e-01 6.49400521e-03 -5.10636084e-02 -1.37436226e-01
5.33935785e-01 -8.08864355e-01 -5.94423890e-01 4.06964451e-01
-9.31277990e-01 -7.98148930e-01 -3.13445866e-01 1.69949979e-01
3.52374315e-02 4.62626636e-01 -1.07730818e+00 -6.71342850e-01
-3.34681958e-01 -1.03009927e+00 4.37918723e-01 1.15750886e-01
2.22187806e-02 -1.06830132e+00 -2.33744785e-01 -1.59362495e-01
6.57918751e-01 -4.36280593e-02 1.20386827e+00 -6.51839614e-01
-7.84945786e-01 -2.17388228e-01 -6.60291016e-01 3.65923464e-01
-1.16682068e-01 4.11179801e-03 -5.96270621e-01 -4.75805789e-01
-2.47804806e-01 1.10391773e-01 1.23451102e+00 7.24751472e-01
9.88599479e-01 -6.31313860e-01 -1.94429889e-01 6.18545413e-01
1.59467268e+00 -4.29593682e-01 7.13936329e-01 -3.05341892e-02
9.57845807e-01 4.81444627e-01 -3.47041279e-01 3.30963619e-02
3.10884178e-01 1.29520789e-01 6.25901282e-01 -4.74424064e-01
-4.41904038e-01 -2.73783803e-01 2.67946064e-01 8.41581523e-01
-9.14310589e-02 -3.88491541e-01 -8.46918404e-01 5.89435458e-01
-1.57214308e+00 -6.78608179e-01 -2.72453994e-01 2.23777986e+00
7.27567375e-01 4.59011644e-01 2.00967506e-01 -4.93760556e-02
9.28493321e-01 4.87492800e-01 -4.11203086e-01 -2.92317390e-01
-1.11085854e-01 2.79844314e-01 1.08662200e+00 8.32967281e-01
-7.93858349e-01 9.22546923e-01 6.63458967e+00 8.53581369e-01
-1.09079313e+00 -8.83739293e-02 8.68933797e-01 1.59529634e-02
-6.08180761e-01 7.50850365e-02 -4.02364969e-01 3.28020066e-01
8.73043835e-01 1.33207440e-01 8.06916058e-01 6.93378270e-01
2.63715181e-02 1.87539175e-01 -1.00309587e+00 5.44228554e-01
-3.32787097e-01 -1.51286423e+00 8.69896337e-02 3.45191419e-01
6.35309517e-01 4.43882555e-01 1.09206803e-01 2.59228289e-01
9.83517170e-01 -1.07430887e+00 4.94092435e-01 1.29069641e-01
4.37550753e-01 -6.24752283e-01 3.46041590e-01 2.21724182e-01
-1.08007002e+00 -6.67085405e-03 -4.56110716e-01 -1.99340954e-01
-1.81344852e-01 1.07582843e+00 -5.17740309e-01 2.35306695e-01
5.27559638e-01 3.82466227e-01 -6.38967812e-01 7.55497217e-01
-7.59931579e-02 7.17076004e-01 -5.02788842e-01 -7.20230937e-02
4.70508367e-01 -3.91748667e-01 6.23040736e-01 1.00013781e+00
1.69111431e-01 -1.22800909e-01 -2.38320649e-01 1.00809383e+00
-4.67090130e-01 -1.25559971e-01 -4.02181745e-01 -3.27985883e-01
2.43443295e-01 1.26613462e+00 -1.05262458e+00 -2.65620738e-01
-4.59551573e-01 8.62859547e-01 6.99123621e-01 6.70593977e-01
-6.93512619e-01 -4.54632431e-01 5.96150100e-01 5.00042439e-01
5.77165365e-01 -3.50818753e-01 -3.28646660e-01 -1.12930763e+00
7.32959583e-02 -6.72148407e-01 2.17717364e-01 -3.96200418e-01
-1.21099460e+00 6.12080932e-01 -4.46495384e-01 -5.10507166e-01
3.23809534e-01 -5.35360396e-01 -7.50123620e-01 8.05814743e-01
-1.49418581e+00 -8.48218501e-01 -2.72163808e-01 2.63987899e-01
-1.45743191e-02 5.20526111e-01 3.35175365e-01 5.31692505e-01
-6.96896613e-01 7.68431962e-01 1.14561655e-01 2.94638485e-01
2.55557328e-01 -1.29065096e+00 8.04514825e-01 1.17544639e+00
8.97807851e-02 7.93910503e-01 9.76866007e-01 -6.85952723e-01
-1.33126402e+00 -1.08664382e+00 7.66966641e-01 -8.13664570e-02
8.26677918e-01 -4.55566466e-01 -1.03638935e+00 6.77253723e-01
1.34548396e-01 1.66066527e-01 1.84582785e-01 4.18662548e-01
-5.64717352e-01 -3.63549329e-02 -9.23183680e-01 1.03460777e+00
1.31200278e+00 -3.14461589e-01 1.93049341e-01 2.07979336e-01
8.70006502e-01 -1.07142359e-01 -7.70789862e-01 1.98422804e-01
2.80397147e-01 -1.06390536e+00 8.12289834e-01 -5.30341268e-01
5.27504265e-01 8.71843994e-02 2.71122493e-02 -1.39985108e+00
-4.89727914e-01 -9.43068743e-01 -7.02067018e-02 9.50803339e-01
6.52569413e-01 -7.16135025e-01 1.16335094e+00 6.46424294e-01
-7.12388661e-03 -5.75484872e-01 -5.76704979e-01 -7.37963855e-01
4.13754374e-01 -2.58606255e-01 5.34378469e-01 8.44389081e-01
-2.57260680e-01 2.97191024e-01 -1.62987262e-01 3.52496892e-01
6.78616941e-01 -2.07645103e-01 5.07637441e-01 -1.12196374e+00
-3.08480978e-01 -8.38706613e-01 -1.44816801e-01 -1.40089190e+00
-1.19617134e-01 -9.24617529e-01 5.28472140e-02 -1.43194926e+00
1.57843113e-01 -6.07578874e-01 -3.23332548e-01 1.91313684e-01
-3.38941455e-01 9.37548205e-02 2.10649699e-01 1.01831913e-01
-4.37791526e-01 2.15069562e-01 1.22946095e+00 -5.33784218e-02
-2.40404829e-01 -8.48596096e-02 -1.13048208e+00 7.46776640e-01
5.72656214e-01 -4.10643727e-01 -3.82571787e-01 -6.97083354e-01
6.41879201e-01 -5.09305969e-02 5.19657791e-01 -9.28902507e-01
1.81295872e-01 2.24818632e-01 -1.82381328e-02 -2.43205965e-01
-6.60892054e-02 -7.26820588e-01 -6.53801486e-02 4.84424114e-01
-5.18139064e-01 9.04757082e-02 4.72264588e-02 7.55242705e-01
2.17910707e-01 -4.27676961e-02 9.19533491e-01 -2.62500495e-01
-1.35710344e-01 4.43892121e-01 -5.78552485e-01 1.39289826e-01
3.13958317e-01 -1.42151415e-01 -4.47315425e-01 -7.49861360e-01
-7.26882815e-01 1.18634783e-01 5.56687117e-01 -8.32230151e-02
2.54057318e-01 -1.13172650e+00 -6.01187170e-01 1.34734675e-01
-4.29644436e-01 1.12510314e-02 4.57111210e-01 1.00527036e+00
-6.43074453e-01 -1.47516146e-01 3.17837954e-01 -3.57446760e-01
-9.67533290e-01 4.76534694e-01 4.64372665e-01 -7.27124929e-01
-7.79719651e-01 1.17245698e+00 5.02963305e-01 -2.45767102e-01
4.72438961e-01 -5.27625382e-01 2.18751132e-01 -1.69702113e-01
1.85994416e-01 3.48064899e-01 2.25750864e-01 -2.11221129e-01
-1.27907380e-01 1.02250285e-01 -3.47028017e-01 1.81732818e-01
1.41103065e+00 -1.32530987e-01 -2.04687178e-01 8.87639672e-02
1.19107664e+00 2.48945672e-02 -1.46693718e+00 -3.41270387e-01
-4.85290624e-02 3.71092372e-02 1.79727182e-01 -4.23394948e-01
-1.50567436e+00 7.63386250e-01 2.26535589e-01 8.76767695e-01
1.03372931e+00 4.41525914e-02 8.27798009e-01 2.61213809e-01
-5.46603389e-02 -8.54914904e-01 -6.31977171e-02 5.08878708e-01
6.10626876e-01 -8.48991156e-01 8.51432458e-02 -6.70085132e-01
-3.21327209e-01 7.10660458e-01 3.72150928e-01 -7.07769752e-01
9.06126857e-01 2.34334245e-01 -2.56859720e-01 -6.20377958e-01
-7.00322270e-01 -9.93758887e-02 1.74205452e-02 3.40447545e-01
3.16662550e-01 1.46601526e-02 3.18054147e-02 5.00685275e-01
-2.36288413e-01 -3.40451241e-01 6.61866665e-01 5.81195474e-01
-4.13172215e-01 -8.04690301e-01 8.00110251e-02 7.56380081e-01
-6.14358604e-01 -6.78586781e-01 -4.30200338e-01 9.31009710e-01
-3.72186840e-01 7.41494179e-01 2.39269182e-01 -3.39512616e-01
1.33227602e-01 -1.22449040e-01 5.60313940e-01 -3.69259417e-01
-7.53891706e-01 2.38833204e-03 1.51542678e-01 -5.26456058e-01
-1.07799217e-01 -3.84142339e-01 -8.90064299e-01 -8.21836710e-01
-3.80468071e-01 9.20291021e-02 3.67442906e-01 7.28668094e-01
4.13321882e-01 7.31819630e-01 3.26952785e-01 -7.65082657e-01
-6.06844723e-01 -9.87875760e-01 -8.67289186e-01 3.42056841e-01
4.45005924e-01 -1.91377014e-01 -9.38298047e-01 -2.94362903e-01] | [6.856550693511963, 6.054480075836182] |
8770f1d7-049c-4592-864f-14a4f0f7e005 | face-morphing-attacks-and-face-image-quality | 2208.05864 | null | https://arxiv.org/abs/2208.05864v2 | https://arxiv.org/pdf/2208.05864v2.pdf | Face Morphing Attacks and Face Image Quality: The Effect of Morphing and the Unsupervised Attack Detection by Quality | Morphing attacks are a form of presentation attacks that gathered increasing attention in recent years. A morphed image can be successfully verified to multiple identities. This operation, therefore, poses serious security issues related to the ability of a travel or identity document to be verified to belong to multiple persons. Previous works touched on the issue of the quality of morphing attack images, however, with the main goal of quantitatively proofing the realistic appearance of the produced morphing attacks. We theorize that the morphing processes might have an effect on both, the perceptual image quality and the image utility in face recognition (FR) when compared to bona fide samples. Towards investigating this theory, this work provides an extensive analysis of the effect of morphing on face image quality, including both general image quality measures and face image utility measures. This analysis is not limited to a single morphing technique, but rather looks at six different morphing techniques and five different data sources using ten different quality measures. This analysis reveals consistent separability between the quality scores of morphing attack and bona fide samples measured by certain quality measures. Our study goes further to build on this effect and investigate the possibility of performing unsupervised morphing attack detection (MAD) based on quality scores. Our study looks intointra and inter-dataset detectability to evaluate the generalizability of such a detection concept on different morphing techniques and bona fide sources. Our final results point out that a set of quality measures, such as MagFace and CNNNIQA, can be used to perform unsupervised and generalized MAD with a correct classification accuracy of over 70%. | ['Naser Damer', 'Biying Fu'] | 2022-08-11 | null | null | null | null | ['face-image-quality'] | ['computer-vision'] | [ 2.12053835e-01 5.27739525e-02 9.69896689e-02 -2.95801371e-01
-5.73470771e-01 -7.46415734e-01 8.14473867e-01 1.23070359e-01
-3.32933426e-01 3.26300442e-01 -2.23364294e-01 -3.22842687e-01
-3.15939397e-01 -9.03960049e-01 -5.57822108e-01 -6.18567348e-01
-3.03452045e-01 2.64165998e-01 -6.50826618e-02 -2.33079404e-01
3.19823116e-01 1.02661479e+00 -1.62321639e+00 2.50232816e-01
4.89081740e-01 1.01460266e+00 -5.71691215e-01 6.52961373e-01
1.67016879e-01 2.52454132e-01 -1.14420712e+00 -1.15095210e+00
6.10359907e-01 -3.73627514e-01 -7.49335527e-01 3.83783013e-01
9.01079834e-01 -3.47462177e-01 1.33913606e-01 1.42916036e+00
4.60418969e-01 -3.47696364e-01 6.15916789e-01 -1.65061462e+00
-6.48457408e-01 3.10614824e-01 -3.92432034e-01 1.48902297e-01
6.58891797e-01 4.62972194e-01 6.40458405e-01 -7.29368925e-01
5.70056438e-01 1.49168777e+00 4.82431591e-01 4.19602185e-01
-1.30105603e+00 -1.00915849e+00 -4.99742985e-01 1.15624800e-01
-1.37670243e+00 -5.67042828e-01 5.02113223e-01 -5.97135842e-01
2.89266378e-01 4.15906250e-01 3.44699591e-01 9.87603784e-01
1.17498182e-01 -6.74024271e-03 1.55800939e+00 -6.44312203e-01
2.55300011e-02 5.55850446e-01 -8.15433636e-02 7.92809367e-01
5.89218676e-01 2.14311436e-01 -5.50905049e-01 -1.97577685e-01
5.82829595e-01 -5.18980742e-01 -1.65603667e-01 -6.84106573e-02
-5.30682504e-01 8.85992646e-01 8.04986656e-02 5.70443213e-01
-1.28147721e-01 -3.16609621e-01 3.09673905e-01 6.43691838e-01
2.54793882e-01 7.90473223e-01 -6.12983517e-02 -6.89895153e-02
-1.17611647e+00 -9.62612554e-02 7.02270389e-01 4.32419419e-01
9.54807878e-01 5.29112406e-02 6.97603822e-02 5.56547761e-01
2.59798408e-01 5.60333014e-01 2.88674623e-01 -6.89129949e-01
3.11294675e-01 6.99154794e-01 -8.25705901e-02 -1.44724512e+00
-1.28704414e-01 1.40110195e-01 -5.98601460e-01 8.96074831e-01
7.35021710e-01 -2.17175290e-01 -7.14050472e-01 1.67251205e+00
2.45634794e-01 -2.58703262e-01 -1.09922402e-01 7.03277290e-01
5.01377642e-01 3.21944892e-01 2.11776514e-02 -8.58654976e-02
1.60082746e+00 -2.32501790e-01 -6.19325578e-01 2.88890451e-01
3.61179233e-01 -1.25215089e+00 1.10063195e+00 6.41157866e-01
-1.14763665e+00 -6.34871542e-01 -1.31019127e+00 4.89107609e-01
-6.70870543e-01 4.61402833e-02 4.65762913e-01 1.76607370e+00
-1.25376356e+00 7.59079754e-01 -2.55813003e-01 -5.97078204e-01
5.57220757e-01 4.95745182e-01 -8.89310598e-01 3.61789577e-02
-1.03762472e+00 1.05639458e+00 2.02291787e-01 -2.06584901e-01
-7.36466110e-01 -5.06818891e-01 -5.03722489e-01 -1.87094495e-01
2.13149190e-02 -1.11820981e-01 4.26804394e-01 -1.45189524e+00
-1.29416609e+00 1.35786772e+00 1.78011596e-01 -4.27837580e-01
5.57572365e-01 1.68396726e-01 -9.96661723e-01 4.20545250e-01
-5.16544767e-02 4.90455359e-01 1.23533309e+00 -1.35613132e+00
-2.29428753e-01 -5.61453521e-01 1.90019667e-01 -3.21841359e-01
-5.16134083e-01 6.03160143e-01 -7.91946612e-03 -6.74902380e-01
-2.12961361e-01 -8.11836243e-01 3.89762551e-01 3.72479185e-02
-3.79031152e-01 1.94807068e-01 7.95377314e-01 -6.62824750e-01
9.88828540e-01 -2.18918848e+00 -1.44285828e-01 5.47404408e-01
-2.88449787e-02 7.47643113e-01 -2.26970062e-01 4.30210084e-01
-4.55097109e-01 5.76004803e-01 -8.03406537e-02 -1.38698101e-01
1.35147482e-01 -7.13792071e-02 -1.43534511e-01 7.81088352e-01
4.77245271e-01 6.47450387e-01 -4.01509613e-01 -4.58313316e-01
1.13610618e-01 5.15670359e-01 -3.36320400e-01 1.18469939e-01
3.77720326e-01 1.25720575e-01 1.11215994e-01 8.74462724e-01
1.04908466e+00 2.96745330e-01 1.88951150e-01 -3.14478695e-01
2.05589041e-01 -1.63380891e-01 -1.32658553e+00 8.64774823e-01
-2.74276048e-01 6.87255383e-01 1.49190634e-01 -6.89368784e-01
9.41682696e-01 3.35623145e-01 2.60844707e-01 -7.14163959e-01
3.50632161e-01 2.29220420e-01 2.12901458e-01 -3.81895751e-01
6.48429215e-01 -3.99443239e-01 6.80847913e-02 6.04840457e-01
3.44487816e-01 6.20018914e-02 2.27488056e-01 1.73739696e-04
8.38591158e-01 -2.91137516e-01 9.06713158e-02 -4.26446676e-01
6.73296630e-01 -4.28015471e-01 6.68566534e-03 4.65895712e-01
-4.03452009e-01 4.04129922e-01 7.01575100e-01 -1.42022252e-01
-9.46572840e-01 -1.04488564e+00 -3.43541324e-01 5.51194787e-01
5.04817031e-02 -3.46697807e-01 -1.15622246e+00 -8.91316175e-01
5.33840768e-02 2.67534077e-01 -8.25190365e-01 -2.89905280e-01
-1.95171759e-01 -9.20096397e-01 1.25795364e+00 -7.60779455e-02
5.10526955e-01 -9.93019700e-01 -6.12602115e-01 -3.47409397e-01
8.32383409e-02 -9.96772408e-01 -1.66112944e-01 -2.88537741e-01
-6.08717322e-01 -1.40633178e+00 -3.77814651e-01 -2.63785034e-01
7.61618793e-01 -4.26521786e-02 9.36467528e-01 4.52588975e-01
-4.70657080e-01 5.99671900e-01 -4.40373689e-01 -3.02982330e-01
-9.11287367e-01 -4.50198174e-01 2.58619905e-01 5.95512927e-01
4.01447713e-01 -4.19183850e-01 -3.04821014e-01 6.56212687e-01
-1.28212059e+00 -7.37821400e-01 5.79003394e-01 2.90878713e-01
-3.27720307e-02 3.45088720e-01 4.03294653e-01 -7.22486198e-01
6.75609231e-01 -1.45183414e-01 -5.11018515e-01 3.52131873e-01
-7.17985272e-01 -1.33065701e-01 3.01032096e-01 -5.24890125e-01
-7.53018081e-01 -3.31329793e-01 -1.85229361e-01 -3.51157784e-01
-4.15281236e-01 2.86837816e-02 -6.14954114e-01 -7.36319900e-01
8.29515636e-01 -8.90718959e-03 3.65273952e-01 -8.71973112e-02
2.38084331e-01 6.15678310e-01 3.25923502e-01 -4.23602730e-01
1.36765254e+00 6.10009730e-01 2.47066572e-01 -1.01249325e+00
-5.49189486e-02 -1.38778210e-01 -5.05816162e-01 -6.44784093e-01
7.13685572e-01 -3.97184819e-01 -8.78959656e-01 7.41873384e-01
-1.00910950e+00 -3.13606262e-02 -5.16446531e-02 2.59525031e-01
-2.14120850e-01 8.73135448e-01 -5.45298517e-01 -8.80604625e-01
9.72555578e-03 -1.17951441e+00 8.41089070e-01 3.49369533e-02
-3.40026110e-01 -9.62960601e-01 -1.95921108e-01 5.85073948e-01
3.59648347e-01 4.87726837e-01 8.15685332e-01 -5.88950455e-01
-2.37024516e-01 -5.29426277e-01 -3.02505732e-01 5.91872513e-01
2.90494859e-01 3.34550023e-01 -1.14507174e+00 -5.08055389e-01
2.46573672e-01 -2.31569603e-01 4.32025552e-01 -4.91461046e-02
5.50062478e-01 -4.42563504e-01 1.73193619e-01 3.31364155e-01
1.56667185e+00 1.52886927e-01 1.15974450e+00 2.85842717e-01
3.77396286e-01 1.17784786e+00 6.70153677e-01 2.90163815e-01
-3.92479956e-01 8.85062933e-01 5.70545316e-01 -2.51361400e-01
-2.20056564e-01 3.25838812e-02 6.25045478e-01 2.07117274e-01
-1.60347924e-01 -1.28077671e-01 -6.69435740e-01 9.55182090e-02
-9.67087865e-01 -1.12808347e+00 -1.05269484e-01 2.50359154e+00
3.72426838e-01 9.38839763e-02 4.21897411e-01 5.52538693e-01
1.00584793e+00 3.42848711e-02 3.81359421e-02 -7.42554367e-01
-2.93720633e-01 3.92454654e-01 5.56355596e-01 5.08350611e-01
-9.55680907e-01 6.26407921e-01 6.48017311e+00 9.98678327e-01
-1.34913099e+00 1.01487778e-01 7.40501642e-01 2.64930248e-01
-6.58691972e-02 -1.31350085e-02 -7.14238882e-01 5.79404056e-01
1.01217198e+00 1.76096018e-02 2.90153056e-01 5.32518625e-01
3.36678103e-02 -1.92738682e-01 -8.96400869e-01 9.29997802e-01
5.58887184e-01 -9.72650349e-01 3.98146003e-01 6.34022772e-01
4.06673223e-01 -7.45962977e-01 4.08745944e-01 -9.66149569e-02
-2.48508248e-02 -1.22583151e+00 7.32928097e-01 2.21783906e-01
9.17800963e-01 -9.70503986e-01 6.79485142e-01 -1.62759855e-01
-8.70512307e-01 7.38996789e-02 -2.07857862e-01 1.03455894e-01
-4.36611511e-02 4.20879394e-01 -7.91579008e-01 6.65193081e-01
5.14620900e-01 -1.37934112e-03 -9.45586145e-01 6.52313113e-01
-1.17302291e-01 4.07009065e-01 -1.21426105e-01 3.84420782e-01
7.89172724e-02 -1.80697054e-01 5.13383985e-01 1.16297162e+00
2.94198364e-01 -3.05882543e-01 -4.42355692e-01 8.55729401e-01
5.06244041e-02 3.04171652e-01 -6.32015526e-01 -2.22280070e-01
2.93517470e-01 1.45358002e+00 -9.90989864e-01 -7.80695304e-02
-2.73447871e-01 9.40358222e-01 -1.08060285e-01 4.82655596e-03
-9.29212451e-01 -3.30832422e-01 7.02289343e-01 4.75533009e-01
1.25262693e-01 -6.29617497e-02 -1.09160349e-01 -8.69988382e-01
4.61193323e-02 -1.30039060e+00 3.72302026e-01 -3.50709230e-01
-1.20718753e+00 6.61019564e-01 -4.86501344e-02 -1.20208287e+00
2.48144008e-02 -8.97962213e-01 -4.96891081e-01 9.37833369e-01
-1.23254764e+00 -1.09319615e+00 -1.84902251e-01 6.49163127e-01
-2.44434904e-02 -4.09462273e-01 8.05555403e-01 4.89646554e-01
-5.18615007e-01 1.04441702e+00 -5.68410158e-01 4.00106996e-01
7.96914220e-01 -9.33196723e-01 2.32676402e-01 1.26281035e+00
3.54280323e-01 7.33530223e-01 6.88351214e-01 -4.51077342e-01
-1.15615594e+00 -7.57786095e-01 7.37206280e-01 -6.11446857e-01
4.89428401e-01 -2.85894483e-01 -7.03405797e-01 1.94785818e-01
2.00168952e-01 -1.80635899e-01 9.03191805e-01 -9.83935967e-02
-6.21027410e-01 4.29686904e-02 -1.67076588e+00 4.03338611e-01
6.34518147e-01 -7.77672887e-01 -2.78773636e-01 1.91047974e-03
2.16522496e-02 1.86962113e-01 -1.10880578e+00 3.11137259e-01
6.08796000e-01 -1.60008514e+00 1.00076604e+00 -4.25820172e-01
2.55497187e-01 -2.22397104e-01 -1.72707751e-01 -7.96560884e-01
-2.77740378e-02 -5.34437835e-01 3.02462786e-01 1.69200909e+00
3.45018774e-01 -7.39764452e-01 7.52737463e-01 6.57549798e-01
4.27907497e-01 -2.80493975e-01 -9.53646719e-01 -1.14686024e+00
9.61473882e-02 -2.98234344e-01 7.73879945e-01 1.26224327e+00
-6.17967770e-02 -8.44630301e-02 -3.22052449e-01 4.92280930e-01
5.91579318e-01 -3.11498195e-01 9.80129957e-01 -1.23564625e+00
-3.30093086e-01 -5.34942746e-01 -1.12974823e+00 -9.80992466e-02
1.00600123e-01 -7.50656545e-01 -4.87801224e-01 -6.57685816e-01
2.64943782e-02 -2.01651335e-01 -3.66331980e-04 2.43523642e-01
1.06940873e-01 7.95086622e-01 4.90130126e-01 1.38037235e-01
-2.64340006e-02 -9.43063051e-02 9.43733156e-01 -6.16259277e-02
2.11358190e-01 -8.30003172e-02 -5.27871132e-01 5.64981282e-01
7.25241482e-01 -5.06288052e-01 -2.14791343e-01 1.39777005e-01
1.41055927e-01 -1.75443232e-01 6.26027644e-01 -1.27383018e+00
-9.85008031e-02 -5.78026958e-02 3.85743111e-01 1.03866905e-01
3.47317427e-01 -9.36217487e-01 3.12263191e-01 6.08626068e-01
1.34574305e-02 1.85532033e-01 2.87299514e-01 1.17435478e-01
-2.87385494e-01 -5.02046645e-01 1.11595774e+00 -7.12688640e-02
-4.74531621e-01 1.41589344e-01 -4.25495684e-01 -2.09507450e-01
1.18956137e+00 -7.57090151e-01 -2.21142337e-01 -3.80097657e-01
-7.18344033e-01 -6.02959216e-01 8.74764442e-01 3.29080284e-01
4.63138819e-01 -1.26315844e+00 -6.96739852e-01 4.74363089e-01
2.11603418e-02 -1.09682035e+00 6.51196465e-02 6.36225224e-01
-5.73669076e-01 7.88217485e-02 -7.47707069e-01 -3.02825212e-01
-1.80060732e+00 8.58656943e-01 4.16341335e-01 -7.52280429e-02
1.00231640e-01 6.68366194e-01 -8.00197423e-02 -1.63163215e-01
-1.97756346e-02 2.06387088e-01 -1.46012112e-01 5.00635147e-01
6.42367661e-01 4.58241314e-01 2.40068719e-01 -1.20008028e+00
-4.24131989e-01 6.85414195e-01 1.17360763e-01 -3.47600758e-01
7.22490430e-01 -2.28973348e-02 -3.03205639e-01 -9.69065726e-02
1.32166910e+00 4.51369345e-01 -6.90313399e-01 3.74199599e-01
6.37912899e-02 -9.04088199e-01 -3.96404684e-01 -6.69169128e-01
-1.11287761e+00 7.47003675e-01 9.40737307e-01 6.49960220e-01
1.12033665e+00 -2.15257227e-01 3.12806189e-01 -1.49997175e-01
3.85886014e-01 -8.62339973e-01 4.15758640e-01 -1.64432019e-01
7.30526626e-01 -1.17999792e+00 3.19731347e-02 -5.83161592e-01
-3.98907483e-01 1.13872528e+00 4.20956314e-01 -2.47930433e-03
4.30851460e-01 1.37604937e-01 2.34785870e-01 -4.13010240e-01
2.55032871e-02 -1.15473479e-01 3.16262990e-01 8.35286856e-01
2.65017539e-01 1.10049639e-02 -2.93198436e-01 1.48772463e-01
-4.00055617e-01 -3.29963446e-01 6.12221181e-01 7.12236702e-01
-2.02331305e-01 -1.54054666e+00 -8.68322551e-01 2.24273890e-01
-7.60824502e-01 1.29762799e-01 -9.00578558e-01 9.93634641e-01
4.64170963e-01 1.18630278e+00 -8.81695896e-02 -6.28076375e-01
2.46041507e-01 1.25933155e-01 6.69700682e-01 -3.90970767e-01
-8.85149539e-01 -5.22276103e-01 1.47509634e-01 -4.55340922e-01
-7.04137325e-01 -6.23266220e-01 -5.27888119e-01 -7.95308590e-01
-5.28372943e-01 1.50029108e-01 7.35282302e-01 8.22216749e-01
1.26829609e-01 2.32596118e-02 7.54586756e-01 -7.37329066e-01
-2.94718236e-01 -7.98933208e-01 -7.78711617e-01 8.36894095e-01
5.07039428e-02 -5.75425565e-01 -4.66533005e-01 2.53796438e-03] | [12.954313278198242, 0.994962215423584] |
491c5bc6-d56a-4e9a-8939-88a0aae740e4 | momentum-contrastive-pre-training-for | 2212.05762 | null | https://arxiv.org/abs/2212.05762v2 | https://arxiv.org/pdf/2212.05762v2.pdf | Momentum Contrastive Pre-training for Question Answering | Existing pre-training methods for extractive Question Answering (QA) generate cloze-like queries different from natural questions in syntax structure, which could overfit pre-trained models to simple keyword matching. In order to address this problem, we propose a novel Momentum Contrastive pRe-training fOr queStion anSwering (MCROSS) method for extractive QA. Specifically, MCROSS introduces a momentum contrastive learning framework to align the answer probability between cloze-like and natural query-passage sample pairs. Hence, the pre-trained models can better transfer the knowledge learned in cloze-like samples to answering natural questions. Experimental results on three benchmarking QA datasets show that our method achieves noticeable improvement compared with all baselines in both supervised and zero-shot scenarios. | ['Irwin King', 'Yasheng Wang', 'Muzhi Li', 'Minda Hu'] | 2022-12-12 | null | null | null | null | ['natural-questions'] | ['miscellaneous'] | [ 2.19623119e-01 1.79814875e-01 6.10003136e-02 -4.47784960e-01
-1.85261440e+00 -5.54851174e-01 5.05030394e-01 1.67484775e-01
-5.19551814e-01 4.63044405e-01 5.25748730e-01 -3.74448627e-01
4.03904803e-02 -8.50770354e-01 -7.20539987e-01 9.64694917e-02
4.60987806e-01 7.82745481e-01 5.43248594e-01 -5.42153358e-01
1.72734201e-01 -3.50847274e-01 -1.08006167e+00 9.77000237e-01
1.45167148e+00 7.59790361e-01 3.18936616e-01 9.98700738e-01
-6.25120163e-01 1.12716818e+00 -6.80641770e-01 -7.89728343e-01
-6.62616789e-02 -7.38155961e-01 -1.33887696e+00 -3.88335645e-01
8.69583309e-01 -4.21860725e-01 -3.67618442e-01 1.00758731e+00
4.36213762e-01 2.39329070e-01 5.74172258e-01 -5.40424943e-01
-1.09626222e+00 6.78245664e-01 5.08450754e-02 5.42093396e-01
8.43836784e-01 4.14258301e-01 1.64572418e+00 -1.32629883e+00
4.04865205e-01 1.37023187e+00 3.31377298e-01 7.68509507e-01
-1.04849076e+00 -1.61896855e-01 -8.95202756e-02 4.37600374e-01
-9.29819822e-01 -4.46121752e-01 6.19511306e-01 -9.94816944e-02
1.08594048e+00 6.24066710e-01 2.34883472e-01 1.08854890e+00
-3.39648813e-01 1.49541032e+00 7.82512903e-01 -5.52792788e-01
-1.64476410e-02 7.66965598e-02 4.91466403e-01 6.54439270e-01
-2.13182136e-01 -1.97009876e-01 -4.45990115e-01 -4.82054353e-01
1.82859465e-01 -3.22493047e-01 -5.25374591e-01 -4.10171449e-02
-9.60311413e-01 1.10258472e+00 3.20433438e-01 1.82994991e-03
-2.66087860e-01 -1.06562361e-01 3.56447130e-01 6.27406776e-01
3.67828816e-01 1.05634344e+00 -5.85540175e-01 -3.17010611e-01
-7.90833890e-01 6.93770349e-01 9.60015595e-01 9.73997533e-01
6.53440535e-01 -3.27011824e-01 -1.10843658e+00 9.18699563e-01
3.17631185e-01 6.92136168e-01 6.29188955e-01 -8.65324438e-01
1.17548203e+00 7.41202354e-01 1.88858449e-01 -6.28385305e-01
6.83718920e-02 -2.34383032e-01 -3.44266355e-01 -7.56582737e-01
5.45972347e-01 -7.50027671e-02 -7.25328922e-01 1.33566117e+00
2.29064643e-01 -2.47894615e-01 4.71002311e-01 8.76561165e-01
1.21811867e+00 9.75241065e-01 7.65679032e-02 9.91989970e-02
1.68387258e+00 -1.43994963e+00 -7.92771101e-01 -6.66354120e-01
6.82632625e-01 -8.13197792e-01 2.32676768e+00 5.14421519e-03
-1.14656556e+00 -4.43082064e-01 -7.31917918e-01 -5.16267180e-01
8.35169256e-02 2.23334298e-01 7.49791712e-02 5.86302578e-01
-4.87349361e-01 2.41406158e-01 -3.95703435e-01 -7.01663941e-02
3.64005297e-01 -3.32123756e-01 2.01373234e-01 -4.64381456e-01
-1.56390774e+00 6.66905940e-01 3.33416492e-01 -1.97076291e-01
-9.53587174e-01 -1.02735174e+00 -8.27830553e-01 3.21417153e-01
6.16590261e-01 -8.00499737e-01 1.81164026e+00 -5.88499367e-01
-1.72485793e+00 8.19907963e-01 -4.46060985e-01 -4.71702427e-01
3.41210872e-01 -8.13819706e-01 -2.70210505e-01 4.41086203e-01
1.07535154e-01 6.06344759e-01 7.30179429e-01 -9.28400278e-01
-4.10248458e-01 5.12340255e-02 4.00391459e-01 3.37654650e-01
-2.98520565e-01 9.53771546e-02 -3.71965080e-01 -5.12323737e-01
6.33960473e-04 -2.91171521e-01 -3.50547805e-02 -4.58108634e-01
-4.85973775e-01 -8.44092071e-01 4.11230803e-01 -1.02679729e+00
1.44967413e+00 -1.85988307e+00 -6.22352846e-02 -2.24969059e-01
8.71097073e-02 5.50535202e-01 -6.64418817e-01 5.48458397e-01
5.20370066e-01 5.73767498e-02 -2.87671983e-01 -1.79165155e-01
2.23217934e-01 9.05697346e-02 -8.61868262e-01 -2.63126135e-01
7.11234629e-01 1.35356951e+00 -1.38097191e+00 -6.12033844e-01
-3.50219220e-01 -1.63989574e-01 -8.97078454e-01 9.10989285e-01
-9.37227845e-01 2.87416488e-01 -6.21188581e-01 5.76471806e-01
4.74655837e-01 -3.77939761e-01 -1.71744421e-01 6.32621646e-02
5.86586356e-01 1.28848231e+00 -4.27014112e-01 1.74387050e+00
-5.95469832e-01 4.27370250e-01 -4.47930634e-01 -5.85190117e-01
8.29801679e-01 4.11137938e-01 -3.17079842e-01 -1.07349718e+00
-9.28950831e-02 2.28074819e-01 -1.94040209e-01 -9.40904796e-01
8.42742503e-01 -1.13977797e-01 -1.39302418e-01 4.96289521e-01
2.21907839e-01 -5.75412154e-01 2.65397251e-01 5.64678729e-01
1.29856777e+00 -1.15885949e-02 -1.11278594e-01 3.68984696e-03
6.59178972e-01 1.43383250e-01 2.46454626e-01 9.83204186e-01
-2.65722543e-01 6.18382275e-01 2.83006728e-01 -6.04145080e-02
-7.33422697e-01 -1.37518072e+00 1.53164566e-01 1.37554574e+00
8.48747417e-02 -6.85843527e-01 -9.79230702e-01 -1.16568387e+00
-3.35692823e-01 1.14919949e+00 -1.66827530e-01 -3.98214102e-01
-8.83141637e-01 -3.43252748e-01 8.76214743e-01 3.40436786e-01
4.48428780e-01 -1.34791374e+00 -4.36003327e-01 3.79920036e-01
-8.46574545e-01 -1.18983090e+00 -6.57476246e-01 -1.84972569e-01
-1.00547564e+00 -9.96331394e-01 -7.63322890e-01 -8.51687074e-01
3.67521912e-01 2.46317551e-01 1.77313554e+00 1.93298519e-01
8.54759067e-02 5.03969312e-01 -8.14790606e-01 -2.40157276e-01
-5.47091842e-01 3.84944528e-01 -4.96672392e-01 -3.24796975e-01
8.78277957e-01 -1.80229366e-01 -6.34557962e-01 -7.64838234e-02
-9.16004062e-01 -2.45426923e-01 5.42781711e-01 9.92424250e-01
5.53321183e-01 -7.17628658e-01 6.97500169e-01 -8.39882433e-01
1.32301760e+00 -6.08932614e-01 -3.62324178e-01 8.98612201e-01
-2.93332517e-01 3.79715025e-01 7.83532321e-01 -4.88107026e-01
-1.20750237e+00 -4.36667800e-01 -4.56924051e-01 -5.66509217e-02
-4.52156961e-02 5.86902678e-01 -2.67866015e-01 4.48895335e-01
9.43556666e-01 4.34877276e-01 -4.43560630e-01 -7.03208745e-01
7.68596947e-01 6.92738354e-01 5.56136191e-01 -8.84202719e-01
8.86166632e-01 1.26054674e-01 -9.36355114e-01 -6.56926870e-01
-1.51890850e+00 -7.21690834e-01 -1.78854555e-01 2.39731312e-01
8.46201599e-01 -6.23845696e-01 -3.17371130e-01 5.99875972e-02
-1.56025767e+00 -3.43715042e-01 -4.74693298e-01 2.86659688e-01
-4.02752161e-01 6.34745359e-01 -8.79506648e-01 -6.63414538e-01
-7.55150020e-01 -8.37967813e-01 1.07755458e+00 4.12849635e-01
-2.89020956e-01 -8.75700414e-01 6.21780217e-01 9.51945782e-01
4.72674847e-01 -6.31151915e-01 1.25582802e+00 -1.02182925e+00
-8.00764084e-01 -2.21143529e-01 -2.98664331e-01 4.54877675e-01
-1.12650506e-01 -4.73645389e-01 -1.01395547e+00 -1.08986750e-01
3.49175483e-01 -1.14310920e+00 9.19627070e-01 -3.14068824e-01
1.02589929e+00 -7.03304470e-01 1.27747327e-01 2.67594039e-01
1.08670187e+00 -3.43646914e-01 8.01590085e-01 2.00494647e-01
5.80309927e-01 6.29440427e-01 8.14184844e-01 -4.76647504e-02
5.82321584e-01 5.28519750e-01 2.11736694e-01 2.67233014e-01
-1.30868226e-01 -8.17710996e-01 5.08482516e-01 1.26095009e+00
6.04760766e-01 -2.62393236e-01 -8.20017815e-01 7.29753852e-01
-1.54292846e+00 -8.90323222e-01 -1.39765278e-01 1.97236466e+00
1.37159228e+00 1.87313941e-03 -9.31373704e-03 -3.39101642e-01
2.45469436e-01 1.91082299e-01 -4.19930339e-01 -2.91904688e-01
-1.52660133e-02 5.67585886e-01 -2.17453733e-01 6.54697895e-01
-8.29014361e-01 1.20789051e+00 5.80205631e+00 1.06961298e+00
-6.33634984e-01 3.50787401e-01 2.88499922e-01 5.60825206e-02
-8.58118713e-01 1.51973620e-01 -9.37211990e-01 2.97307938e-01
9.78225589e-01 -2.49158502e-01 2.51424104e-01 8.57408822e-01
-1.88747063e-01 1.40046135e-01 -1.15747213e+00 6.76820815e-01
2.57475972e-01 -1.19333768e+00 4.42543477e-01 -6.01293147e-01
6.31939530e-01 1.29442126e-01 1.16657808e-01 1.04337060e+00
4.03120250e-01 -1.01743507e+00 5.01215518e-01 5.10128140e-01
4.74042118e-01 -4.54451203e-01 7.50909746e-01 7.11370051e-01
-8.64140034e-01 1.13219067e-01 -7.06884921e-01 8.92979652e-02
4.53546017e-01 3.46649319e-01 -1.04181564e+00 5.00844002e-01
5.86420298e-01 -2.73999162e-02 -8.40985239e-01 1.11049736e+00
-7.14306295e-01 1.18279481e+00 -9.06873494e-02 -6.48723125e-01
5.06958902e-01 9.22078118e-02 6.75173163e-01 1.17302394e+00
5.60771339e-02 7.57575706e-02 -1.17638156e-01 1.26576400e+00
-3.84446025e-01 4.06745553e-01 -2.56624259e-03 -3.39509666e-01
6.11933649e-01 1.02338636e+00 1.08057246e-01 -5.80014288e-01
-4.92055148e-01 1.36210680e+00 8.80768716e-01 4.48732376e-01
-5.39008439e-01 -8.12487721e-01 3.04538429e-01 -5.05669937e-02
3.38046879e-01 1.18951537e-01 7.49756396e-02 -1.54139721e+00
5.53071380e-01 -1.49302530e+00 6.07208967e-01 -6.26955390e-01
-1.74886262e+00 6.32218182e-01 -1.75079346e-01 -9.69073832e-01
-4.29037064e-01 -4.03064758e-01 -7.97142684e-01 9.75174546e-01
-1.86006629e+00 -8.96097839e-01 -1.75643072e-01 4.33070302e-01
9.67240095e-01 2.20278511e-03 8.83659422e-01 1.79665551e-01
-1.11752257e-01 9.68961656e-01 -4.88764383e-02 3.48861396e-01
6.56577110e-01 -1.50679684e+00 7.49260306e-01 8.99711549e-01
7.56131113e-01 8.22895050e-01 5.04557908e-01 -4.73563105e-01
-1.25187588e+00 -1.04212415e+00 1.25942659e+00 -1.13826716e+00
6.72159433e-01 -3.43260080e-01 -1.39188564e+00 3.79030526e-01
3.94419432e-01 -2.82781869e-01 8.28766704e-01 2.32240260e-01
-6.67464733e-01 1.37240991e-01 -8.24706137e-01 6.32450283e-01
6.26761258e-01 -1.17813921e+00 -1.54768956e+00 6.42948210e-01
1.35672951e+00 -4.97386754e-01 -5.03728151e-01 3.38308901e-01
-2.70437598e-02 -6.96067452e-01 1.00693953e+00 -1.04655063e+00
5.34977257e-01 -9.61142629e-02 -1.09002054e-01 -1.09469211e+00
8.10065493e-02 -6.63878322e-01 -6.24426126e-01 1.12245643e+00
6.49621665e-01 -1.38335690e-01 7.41968930e-01 4.22415078e-01
-1.76840410e-01 -9.29018915e-01 -8.22922170e-01 -8.65408480e-01
5.60896993e-01 -2.65428871e-01 6.83395743e-01 6.18528068e-01
2.07308292e-01 1.01919019e+00 -7.71820098e-02 3.25888284e-02
3.20400864e-01 2.58740969e-02 8.26106489e-01 -7.20269620e-01
-6.50595009e-01 -1.50550425e-01 3.06880683e-01 -2.07448816e+00
3.06890216e-02 -1.06754792e+00 3.26572955e-01 -1.53940499e+00
3.38753790e-01 -3.64348799e-01 1.29526854e-01 6.23538271e-02
-1.15587282e+00 -1.46353498e-01 1.01506442e-01 1.79826647e-01
-1.10903478e+00 9.33489561e-01 1.40286446e+00 -2.36515075e-01
-2.59645820e-01 2.62917727e-02 -5.98151207e-01 3.98177058e-01
6.34698987e-01 -6.19363248e-01 -5.38135231e-01 -8.80079091e-01
5.40708780e-01 9.38488245e-02 3.88944298e-01 -7.65896976e-01
1.42016247e-01 4.26698327e-02 -2.97539413e-01 -6.94621384e-01
1.15392596e-01 -2.34838650e-01 -1.13614690e+00 2.76136249e-01
-7.76574314e-01 1.25484187e-02 6.39346913e-02 6.81210220e-01
-5.34354091e-01 -8.40387285e-01 3.27810049e-01 -2.71848470e-01
-4.24575508e-01 1.35582238e-01 -2.25154996e-01 1.27107811e+00
5.93351796e-02 1.51360497e-01 -6.57977462e-01 -7.40083218e-01
-2.16871276e-01 6.41002476e-01 -3.13701332e-02 4.08290952e-01
9.30959642e-01 -1.28483653e+00 -1.01126134e+00 -2.02314079e-01
5.46708941e-01 1.27815291e-01 2.48481706e-01 5.06926179e-01
-4.52602655e-01 7.30632126e-01 5.01219213e-01 -3.46369833e-01
-1.00466037e+00 3.11989486e-01 3.82649779e-01 -7.54316270e-01
-3.18974763e-01 1.42864335e+00 -6.31965697e-02 -1.10937011e+00
2.83270478e-01 -5.40513456e-01 -1.88791692e-01 -2.89716184e-01
8.28414738e-01 1.69938385e-01 4.95218448e-02 -7.80598670e-02
8.17592591e-02 1.09507866e-01 -4.85354543e-01 -2.24759489e-01
8.36449921e-01 -1.87142510e-02 8.86784196e-02 3.57499629e-01
1.26774693e+00 3.05113852e-01 -9.27623987e-01 -6.23779953e-01
5.69819808e-01 -4.19788927e-01 -4.69360262e-01 -8.64127040e-01
-2.38972440e-01 1.35401762e+00 2.38567322e-01 1.22915603e-01
7.83719420e-01 3.51923525e-01 1.51195014e+00 1.16856992e+00
-7.23939762e-03 -1.02479529e+00 9.13003743e-01 1.09204733e+00
9.15708840e-01 -1.31602037e+00 -5.38975775e-01 -3.75335425e-01
-6.17231309e-01 8.77838254e-01 9.68826771e-01 -1.01785116e-01
1.21830441e-02 -6.23304069e-01 3.43284756e-01 -4.24695879e-01
-8.44730556e-01 -2.74072140e-01 6.53534949e-01 4.84872043e-01
5.16711473e-01 -2.97224313e-01 -1.64508179e-01 1.00306809e+00
-4.92972791e-01 -1.71951443e-01 1.21078327e-01 8.17446530e-01
-7.46726096e-01 -1.08527625e+00 -9.24888328e-02 5.46227694e-01
-5.41038871e-01 -7.49058723e-01 -6.34361625e-01 4.01135027e-01
-5.16072512e-01 1.04861236e+00 -3.06383576e-02 -1.78714588e-01
4.58327562e-01 4.22491372e-01 5.56073427e-01 -9.63080585e-01
-7.18759298e-01 -4.61372435e-01 3.12162936e-01 -4.46774751e-01
7.15670586e-02 -1.70721009e-01 -1.13711131e+00 3.39443624e-01
-7.08445847e-01 6.92232370e-01 -9.53053311e-03 1.13861108e+00
4.07987326e-01 3.22717756e-01 3.89301836e-01 4.14933152e-02
-1.36056852e+00 -1.34162176e+00 1.00616291e-01 6.82371974e-01
3.00341249e-01 2.51212548e-02 -5.18435717e-01 -2.45521426e-01] | [11.37887954711914, 8.033238410949707] |
2a4ba4ab-f13c-47b3-8eb2-2cc17f4e520e | nilc-at-webnlg-pretrained-sequence-to | null | null | https://aclanthology.org/2020.webnlg-1.14 | https://aclanthology.org/2020.webnlg-1.14.pdf | NILC at WebNLG+: Pretrained Sequence-to-Sequence Models on RDF-to-Text Generation | This paper describes the submission by the NILC Computational Linguistics research group of the University of São Paulo/Brazil to the RDF-to-Text task for English at the WebNLG+ challenge. The success of the current pretrained models like BERT or GPT-2 in text-to-text generation tasks is well-known, however, its application/success on data-totext generation has not been well-studied and proven. This way, we explore how good a pretrained model, in particular BART, performs on the data-to-text generation task. The results obtained were worse than the baseline and other systems in almost all automatic measures. However, the human evaluation shows better results for our system. Besides, results suggest that BART may generate paraphrases of reference texts. | ['Thiago A. S. Pardo', 'Marco Antonio Sobrevilla Cabezudo'] | null | null | null | null | acl-webnlg-inlg-2020-12 | ['data-to-text-generation'] | ['natural-language-processing'] | [-1.55516453e-02 7.49715567e-01 -1.15776531e-01 -1.65938690e-01
-8.98865700e-01 -3.86327356e-01 1.21674204e+00 1.00754023e-01
-6.42515481e-01 1.30510473e+00 7.50609100e-01 -3.53154629e-01
1.83641478e-01 -7.31471241e-01 -6.85322583e-01 -3.51561964e-01
4.53385800e-01 1.21508098e+00 1.45118739e-02 -7.62273550e-01
3.37824106e-01 8.58828351e-02 -1.19462430e+00 5.73498845e-01
1.16937506e+00 2.55361587e-01 3.90875995e-01 5.64380586e-01
-4.00791258e-01 7.16173887e-01 -8.94840479e-01 -8.45893741e-01
-8.30284134e-02 -6.48400486e-01 -1.31913650e+00 -4.05744195e-01
3.59007388e-01 -1.19985558e-01 7.06354156e-02 7.61077046e-01
8.22868347e-01 9.14368480e-02 7.56621003e-01 -1.01915812e+00
-1.03210723e+00 1.58159566e+00 -4.63332944e-02 2.00380802e-01
6.51396573e-01 -8.26223404e-05 1.02682471e+00 -1.03247464e+00
1.19734979e+00 1.40410006e+00 3.38395625e-01 9.39070344e-01
-1.07927811e+00 -4.81849581e-01 -4.68809158e-01 2.18674600e-01
-1.19446218e+00 -6.30058408e-01 4.78324413e-01 -1.76803663e-01
1.45751393e+00 9.20136720e-02 5.41401684e-01 1.41463089e+00
-4.75650206e-02 1.06330359e+00 1.20713067e+00 -8.24983776e-01
-2.30058879e-01 3.03024530e-01 -1.66713998e-01 3.15013617e-01
2.69945860e-01 -3.56266089e-03 -7.98837364e-01 1.54820666e-01
2.91272789e-01 -1.05023801e+00 -3.37712169e-01 4.93824750e-01
-1.22004235e+00 1.03351676e+00 1.91041589e-01 8.25843334e-01
-2.86899328e-01 4.88323718e-03 6.94884956e-01 3.14460814e-01
8.45179439e-01 7.34950304e-01 -4.46815819e-01 -4.74749058e-01
-1.15333354e+00 7.37853646e-01 1.04471540e+00 1.27034485e+00
3.32144558e-01 1.95098564e-01 -4.84246194e-01 9.57488894e-01
3.08861434e-01 4.46035117e-01 1.00575578e+00 -4.88697320e-01
1.03216767e+00 4.32957351e-01 7.15294778e-02 -6.58548713e-01
-3.17193747e-01 -1.28912181e-01 -6.82784677e-01 -4.21500832e-01
5.05745947e-01 -3.68395954e-01 -6.57097220e-01 1.50768983e+00
-2.39046142e-02 -4.57617968e-01 5.10428846e-01 5.01209855e-01
1.18788767e+00 9.35378850e-01 1.65734023e-01 -3.76046509e-01
1.05930042e+00 -9.27059591e-01 -1.15547919e+00 -3.65702689e-01
1.02329659e+00 -1.11631405e+00 1.18691540e+00 1.30959377e-01
-1.54566932e+00 -8.35180223e-01 -7.76762784e-01 -4.64708716e-01
-7.58816838e-01 4.86152321e-01 3.15101177e-01 8.16308022e-01
-1.19080722e+00 6.03631020e-01 -5.47313750e-01 -7.18940616e-01
6.50988370e-02 -6.06817715e-02 -3.11588198e-01 1.68016069e-02
-1.59172606e+00 1.38760746e+00 1.00447690e+00 -2.09481753e-02
-5.29909492e-01 -5.72572291e-01 -8.20017874e-01 -3.73824537e-01
6.02451302e-02 -6.32378757e-01 1.35628057e+00 -4.37442392e-01
-1.47085726e+00 1.12492239e+00 -2.92266011e-01 -9.79590595e-01
7.84183323e-01 -2.70598769e-01 -4.62861896e-01 -1.88743353e-01
3.12303513e-01 1.08471262e+00 3.90466064e-01 -1.00852060e+00
-5.44023871e-01 -1.08812906e-01 -3.53173316e-02 2.77312368e-01
-9.86873806e-02 3.01168978e-01 1.54718235e-01 -7.97882617e-01
-4.87686992e-01 -8.12327087e-01 1.16601922e-02 -7.36250281e-01
-4.67976987e-01 -9.33259368e-01 5.57184994e-01 -9.81441379e-01
1.44250846e+00 -1.57499862e+00 2.34911479e-02 -2.82629699e-01
-5.06640136e-01 4.49689150e-01 -1.39489144e-01 1.12976360e+00
-1.29927956e-02 7.73278058e-01 2.69762408e-02 -5.32563508e-01
1.29944444e-01 3.29162866e-01 -5.63233674e-01 -1.90615579e-01
1.92076698e-01 1.26740408e+00 -1.10061622e+00 -7.56513596e-01
1.09582737e-01 2.53190279e-01 -2.34202161e-01 2.14702189e-01
-3.93816531e-01 2.04435751e-01 -2.08395541e-01 -1.23942439e-02
3.13672274e-01 2.67094165e-01 -7.23502785e-02 5.87015972e-02
-3.50127578e-01 9.07174289e-01 -5.69597960e-01 1.68067157e+00
-6.25530303e-01 8.20217073e-01 -5.98481238e-01 -6.29272580e-01
8.36055398e-01 7.92427897e-01 -3.82090434e-02 -6.01489604e-01
-8.77760202e-02 7.80073643e-01 2.97161341e-01 -5.23849130e-01
1.22583818e+00 -3.17462265e-01 2.55439319e-02 5.77137887e-01
2.74349630e-01 -8.39225113e-01 1.02738297e+00 4.07177687e-01
5.54212511e-01 6.03189290e-01 4.40899700e-01 -3.73206913e-01
4.94534045e-01 3.53686154e-01 -5.93355782e-02 6.22087777e-01
3.41089040e-01 7.74964094e-01 2.95071363e-01 5.42185977e-02
-1.10280728e+00 -5.57972491e-01 -1.33417815e-01 9.06220615e-01
-4.57245708e-01 -6.41147733e-01 -1.08515322e+00 -8.57129872e-01
-4.08759892e-01 1.61777639e+00 -5.25339365e-01 1.94996297e-01
-8.42801154e-01 -7.17121482e-01 9.42603111e-01 3.15479100e-01
3.50160003e-01 -1.79130197e+00 -2.04838127e-01 5.30608118e-01
-7.35352397e-01 -1.17076087e+00 -4.36415672e-01 -1.58437639e-01
-8.60179305e-01 -6.86074555e-01 -8.17513466e-01 -7.87928820e-01
2.77360320e-01 -1.92864805e-01 1.44122422e+00 1.05283901e-01
1.32418990e-01 -1.63382873e-01 -8.12719464e-01 -9.16389167e-01
-1.24845672e+00 5.48308432e-01 -4.08649772e-01 -5.66664040e-01
2.85717189e-01 -1.98206440e-01 -8.60612094e-02 -2.49993652e-02
-9.80081260e-01 2.71072477e-01 4.01777655e-01 7.95512795e-01
2.23507211e-01 -4.06840384e-01 9.46344435e-01 -1.19122827e+00
1.21655393e+00 -2.67671049e-01 -1.00344591e-01 2.48699099e-01
-6.43831015e-01 1.22573808e-01 7.82628596e-01 -1.87718812e-02
-1.28804970e+00 -4.04585004e-01 -6.11623049e-01 3.09725225e-01
-1.09142326e-01 7.85892069e-01 -8.76927078e-02 5.21206498e-01
9.13767517e-01 1.91307664e-01 -3.31483573e-01 -4.62212443e-01
6.53173864e-01 8.16967249e-01 2.76389033e-01 -7.04353034e-01
6.35442138e-01 -1.54604092e-02 -4.33954686e-01 -9.09574330e-01
-9.17691648e-01 -2.19508365e-01 -5.96415460e-01 -1.33605272e-01
8.88929844e-01 -7.60104418e-01 1.85624763e-01 3.84743929e-01
-1.73720300e+00 -5.00180960e-01 -6.14578068e-01 1.87540933e-01
-5.86560190e-01 2.65698314e-01 -4.43008989e-01 -6.07527077e-01
-7.86733389e-01 -1.01840687e+00 1.19546819e+00 7.93630928e-02
-6.75357699e-01 -1.45878041e+00 2.26098374e-01 5.71530223e-01
4.80356753e-01 -1.12266997e-02 9.38886881e-01 -9.46083844e-01
-3.07920128e-02 -1.18139409e-01 -2.44656056e-02 3.69167507e-01
-8.79103318e-03 1.72586918e-01 -8.57311368e-01 -9.42632481e-02
-1.31149545e-01 -5.57368517e-01 6.25983894e-01 8.53899345e-02
6.87525213e-01 -5.86302578e-01 -1.89708099e-01 -9.74670351e-02
1.17297208e+00 -1.95593566e-01 9.00628924e-01 2.14757025e-01
4.83929962e-01 8.40085447e-01 7.56484509e-01 1.14217825e-01
6.76473737e-01 7.67426848e-01 7.57369548e-02 1.84474453e-01
-6.91839457e-01 -6.10702217e-01 6.65529847e-01 1.22638357e+00
-1.89806461e-01 -8.58526528e-01 -7.94660330e-01 9.07030284e-01
-1.78013277e+00 -1.03205299e+00 -7.82389402e-01 1.90903008e+00
1.36437654e+00 6.27094582e-02 4.68399562e-02 6.59140944e-02
5.31164587e-01 2.93673217e-01 3.77342463e-01 -7.48173296e-01
-5.01145422e-01 5.78774214e-01 1.85905080e-02 5.55023074e-01
-6.33976340e-01 1.31118572e+00 6.27442551e+00 1.16625547e+00
-9.62092280e-01 3.40199411e-01 3.60482365e-01 1.38263777e-01
-5.29526591e-01 1.86741445e-02 -1.13616168e+00 4.53314424e-01
1.33153689e+00 -7.47229159e-01 2.29191363e-01 6.15935624e-01
5.29220402e-01 -5.16218022e-02 -1.15520465e+00 5.03420115e-01
4.27837223e-01 -1.26552761e+00 4.98205394e-01 -4.12708595e-02
1.06239605e+00 2.44115308e-01 -4.13224310e-01 6.00625515e-01
5.84516823e-01 -1.01774371e+00 9.40184057e-01 3.36210698e-01
5.78362823e-01 -5.66988230e-01 1.04145050e+00 6.07332468e-01
-6.55340135e-01 5.36247492e-01 -7.50573218e-01 1.97839692e-01
5.42440236e-01 6.78734601e-01 -1.49326706e+00 8.29793453e-01
1.51419580e-01 4.76501644e-01 -8.85544956e-01 8.20496738e-01
-8.06635141e-01 8.79990160e-01 -7.56491572e-02 -4.59706485e-01
3.21105093e-01 -4.14763987e-02 6.20228231e-01 1.57387698e+00
5.72753549e-01 -3.82847190e-01 -9.41743404e-02 8.89625192e-01
-3.78663450e-01 7.30764866e-01 -7.38985538e-01 -2.43903548e-01
3.19946885e-01 1.19921815e+00 -3.41197044e-01 -5.06718338e-01
-9.03370976e-02 8.24364722e-01 5.54620922e-01 9.06255245e-02
-4.76110846e-01 -3.49808395e-01 -2.18357936e-01 1.60627216e-01
-7.41509199e-02 -6.38792515e-02 -2.46988654e-01 -9.01349485e-01
1.05373621e-01 -9.35761631e-01 3.45438451e-01 -1.16143668e+00
-1.45230055e+00 8.60162616e-01 2.72215903e-01 -8.53133202e-01
-1.04366314e+00 -4.78821635e-01 -5.56242943e-01 1.21218801e+00
-1.48726225e+00 -1.36741805e+00 1.02646582e-01 3.26088578e-01
8.50239336e-01 -2.68538356e-01 1.00731194e+00 -4.64353748e-02
-1.69614434e-01 5.51653624e-01 -1.34803131e-01 1.66556001e-01
8.19808960e-01 -1.50986195e+00 8.98585558e-01 9.10654724e-01
4.59130555e-01 5.53022027e-01 9.21211600e-01 -8.14057171e-01
-8.26862931e-01 -1.05646300e+00 1.97927403e+00 -4.45449740e-01
9.11065161e-01 -8.70317444e-02 -7.65556455e-01 5.25293231e-01
1.08734453e+00 -6.55932367e-01 3.33425015e-01 -5.66760711e-02
-2.37841085e-02 2.71750689e-01 -1.06875491e+00 8.42686772e-01
8.89338076e-01 -4.57604975e-01 -1.09226596e+00 7.68230736e-01
7.79776394e-01 -5.57758749e-01 -9.70656514e-01 9.75348130e-02
1.38795435e-01 -6.83049798e-01 4.98047054e-01 -6.25231922e-01
9.24420178e-01 1.40683874e-02 2.27450594e-01 -1.83374643e+00
2.57662565e-01 -8.86149287e-01 2.08246261e-01 1.67566705e+00
1.01474607e+00 -5.32068491e-01 4.58020985e-01 2.67819345e-01
-4.38147455e-01 -3.75871062e-01 -1.05614424e+00 -7.70668209e-01
6.93085313e-01 -5.05201340e-01 6.24763966e-01 9.52685237e-01
2.01591283e-01 7.54855514e-01 -2.05143258e-01 -5.88977277e-01
1.98536336e-01 -2.53807694e-01 8.18071246e-01 -8.82656813e-01
-1.26453310e-01 -5.14879227e-01 2.08428741e-01 -6.38457716e-01
4.44530159e-01 -1.39675295e+00 1.65883854e-01 -2.10060620e+00
-1.82360917e-01 -2.91473955e-01 6.59450412e-01 4.96116221e-01
-4.90056366e-01 2.32382134e-01 3.98957908e-01 1.18496507e-01
-2.30245981e-02 5.31903148e-01 1.51506484e+00 5.18385097e-02
-2.00395912e-01 -9.87744629e-02 -5.66561282e-01 2.37580329e-01
9.81502652e-01 -6.02840304e-01 -3.02082002e-01 -7.78024435e-01
3.55858952e-01 2.81785410e-02 -6.20123893e-02 -7.85737336e-01
-1.04951955e-01 6.14523962e-02 -9.86096561e-02 -6.79473102e-01
8.60891193e-02 -2.18015313e-01 7.29797408e-02 2.39297464e-01
-5.25178134e-01 4.28472549e-01 3.04700732e-01 -2.36946344e-01
-3.54385853e-01 -1.01805961e+00 4.33146238e-01 -3.64335299e-01
-1.28151983e-01 -9.78769585e-02 -5.02084076e-01 6.71603382e-01
4.34178680e-01 -1.13628261e-01 -5.58329463e-01 -5.87394297e-01
-3.27750146e-01 1.02615923e-01 1.52178764e-01 5.83519340e-01
2.87054330e-01 -1.19853318e+00 -1.48658597e+00 -2.20580921e-01
2.08635539e-01 -2.12550551e-01 -9.69103202e-02 6.53685391e-01
-6.21755600e-01 8.49816203e-01 -8.33488256e-03 -2.00953022e-01
-9.76623535e-01 1.07802093e-01 1.56457260e-01 -8.40707481e-01
-1.81073934e-01 6.68382704e-01 -5.34064412e-01 -6.48705006e-01
-2.50484645e-01 -2.13940084e-01 -5.04180908e-01 3.47599685e-01
4.41943318e-01 5.12055874e-01 4.37734336e-01 -7.64705896e-01
1.36520639e-01 5.70472777e-02 -2.03564093e-01 -6.34309173e-01
1.11043692e+00 3.03678215e-02 -3.13818097e-01 3.96067917e-01
8.58509302e-01 7.67518133e-02 -1.97646484e-01 7.00229630e-02
2.76833594e-01 2.20368244e-03 -1.38273880e-01 -1.04932940e+00
-6.59116328e-01 1.09061503e+00 -1.47880450e-01 5.05703330e-01
5.87091029e-01 -8.99140835e-02 7.12964714e-01 4.92247581e-01
2.42865533e-01 -1.37480593e+00 -2.29908198e-01 1.09791899e+00
1.47640812e+00 -1.09651446e+00 -1.09315366e-01 -2.75188953e-01
-8.65019798e-01 1.18275344e+00 5.75429738e-01 1.42860249e-01
2.56782442e-01 1.34631246e-01 6.64957762e-02 4.98000495e-02
-9.24874306e-01 -3.09368521e-01 4.10143614e-01 7.69634604e-01
1.19115841e+00 -9.06173082e-04 -1.04671526e+00 1.93380952e-01
-1.14258146e+00 1.25847176e-01 9.21013653e-01 4.73512620e-01
-1.63956523e-01 -1.54945207e+00 -8.53478014e-02 3.07413369e-01
-6.31817639e-01 -6.50060356e-01 -7.94689119e-01 1.20537186e+00
-9.60143209e-02 1.11603582e+00 -1.25137463e-01 7.10131675e-02
4.55389887e-01 4.64562178e-01 4.48862761e-01 -1.12280881e+00
-1.26110744e+00 -2.23867986e-02 7.87309647e-01 2.81394850e-02
-6.66599333e-01 -7.90345073e-01 -1.21500373e+00 -2.99566388e-01
-3.89288545e-01 6.26678944e-01 7.54839599e-01 1.03346407e+00
3.84977716e-03 3.14517498e-01 1.60285190e-01 -7.33467102e-01
-3.80065352e-01 -1.74519408e+00 -2.55127430e-01 4.44604278e-01
-3.88633013e-01 5.89665882e-02 -2.27702320e-01 1.98120192e-01] | [11.422876358032227, 9.267518997192383] |
e8c2c32d-e66f-46ec-8785-70b0da18af8d | evil-from-within-machine-learning-backdoors | 2304.08411 | null | https://arxiv.org/abs/2304.08411v2 | https://arxiv.org/pdf/2304.08411v2.pdf | Evil from Within: Machine Learning Backdoors through Hardware Trojans | Backdoors pose a serious threat to machine learning, as they can compromise the integrity of security-critical systems, such as self-driving cars. While different defenses have been proposed to address this threat, they all rely on the assumption that the hardware on which the learning models are executed during inference is trusted. In this paper, we challenge this assumption and introduce a backdoor attack that completely resides within a common hardware accelerator for machine learning. Outside of the accelerator, neither the learning model nor the software is manipulated, so that current defenses fail. To make this attack practical, we overcome two challenges: First, as memory on a hardware accelerator is severely limited, we introduce the concept of a minimal backdoor that deviates as little as possible from the original model and is activated by replacing a few model parameters only. Second, we develop a configurable hardware trojan that can be provisioned with the backdoor and performs a replacement only when the specific target model is processed. We demonstrate the practical feasibility of our attack by implanting our hardware trojan into the Xilinx Vitis AI DPU, a commercial machine-learning accelerator. We configure the trojan with a minimal backdoor for a traffic-sign recognition system. The backdoor replaces only 30 (0.069%) model parameters, yet it reliably manipulates the recognition once the input contains a backdoor trigger. Our attack expands the hardware circuit of the accelerator by 0.24% and induces no run-time overhead, rendering a detection hardly possible. Given the complex and highly distributed manufacturing process of current hardware, our work points to a new threat in machine learning that is inaccessible to current security mechanisms and calls for hardware to be manufactured only in fully trusted environments. | ['Christof Paar', 'Konrad Rieck', 'Jan-Niklas Möller', 'Julian Speith', 'Alexander Warnecke'] | 2023-04-17 | null | null | null | null | ['backdoor-attack', 'traffic-sign-recognition', 'self-driving-cars'] | ['adversarial', 'computer-vision', 'computer-vision'] | [ 2.59156823e-01 8.86259973e-02 -6.49231791e-01 -1.63949206e-01
-4.31451678e-01 -1.00397158e+00 4.53464866e-01 9.28109810e-02
-4.22663569e-01 1.45164251e-01 -6.90127671e-01 -1.44285631e+00
5.37726700e-01 -8.07566822e-01 -1.10880148e+00 -4.21800137e-01
-3.99347171e-02 -7.36963972e-02 7.50365198e-01 -1.51888564e-01
5.96441068e-02 7.13369668e-01 -1.67823732e+00 3.50872278e-01
2.88719326e-01 1.02555728e+00 -6.98921144e-01 7.59830356e-01
3.64338964e-01 4.34980839e-01 -1.02166522e+00 -1.14342205e-01
5.35631716e-01 5.52460318e-03 -4.57726359e-01 -7.43719995e-01
6.58263028e-01 -6.72930062e-01 -2.73554832e-01 9.82074082e-01
-1.62174210e-01 -5.17591476e-01 1.81279019e-01 -1.84645271e+00
4.55355912e-01 5.17425299e-01 -4.51487809e-01 -2.89243571e-02
4.38381284e-02 4.29527700e-01 5.83386302e-01 -1.57811139e-02
3.34636599e-01 8.31997454e-01 4.48626250e-01 7.00978816e-01
-1.10113108e+00 -1.26029325e+00 -1.91929266e-01 -2.77797014e-01
-1.29891670e+00 -4.44267213e-01 5.80029488e-01 -3.48442644e-01
1.21231353e+00 6.18673563e-01 2.09172726e-01 1.28742051e+00
1.12658536e+00 7.15891197e-02 1.05380177e+00 -5.96725404e-01
5.42605519e-01 3.30135852e-01 7.24120677e-01 7.91540861e-01
8.71640146e-01 6.31006658e-01 -2.40169793e-01 -7.62763381e-01
3.31566393e-01 -1.63552582e-01 9.72294584e-02 -1.80016518e-01
-7.61670113e-01 8.08898389e-01 5.14707863e-02 7.74853230e-02
2.63926417e-01 6.00967526e-01 7.36180604e-01 2.64033288e-01
-5.15299082e-01 5.63791692e-01 -5.48756361e-01 -2.30379235e-02
-6.88648820e-01 -4.15476877e-03 9.88756120e-01 5.59932768e-01
6.17285907e-01 3.71347517e-01 6.58767462e-01 -4.37596321e-01
5.34754694e-01 5.88777542e-01 2.95954049e-01 -3.71337086e-01
2.25023508e-01 2.62015879e-01 -1.30860239e-01 -7.14461148e-01
-3.58000427e-01 -4.01135802e-01 -1.07129015e-01 9.53733027e-01
3.72274727e-01 -2.04042971e-01 -8.06708813e-01 1.56046534e+00
6.09971583e-01 3.60237360e-01 2.67877847e-01 6.90054595e-01
2.18650520e-01 4.86378253e-01 9.16601792e-02 2.19162807e-01
1.70157731e+00 -4.73418921e-01 -2.43365094e-01 -2.55479842e-01
9.68284249e-01 -7.47447014e-01 8.76732528e-01 6.76228285e-01
-4.27415907e-01 -5.05536258e-01 -2.32809067e+00 3.24103355e-01
-5.66207051e-01 -1.97552085e-01 7.62972116e-01 1.60811925e+00
-6.58365011e-01 3.51631403e-01 -1.17352021e+00 1.95496812e-01
-1.63856566e-01 9.10133064e-01 -3.71714979e-01 3.57052684e-01
-1.19191599e+00 8.79502177e-01 4.52846289e-01 -1.24500088e-01
-1.02676690e+00 -7.42832065e-01 -9.33051765e-01 -1.54872373e-01
2.72715211e-01 -4.52906489e-01 1.15982711e+00 -5.81030846e-01
-1.62759888e+00 5.19389153e-01 3.66238952e-01 -9.14127648e-01
2.31934234e-01 -1.26859277e-01 -9.44876552e-01 3.86701114e-02
-3.96550834e-01 2.77482122e-01 1.21675336e+00 -1.04392111e+00
-5.42349160e-01 4.96820323e-02 3.65452886e-01 -8.48125279e-01
-4.65975821e-01 1.41857237e-01 2.24312276e-01 -7.26464093e-02
-4.13017243e-01 -1.45673275e+00 -1.09103881e-01 -3.05662841e-01
-3.97151500e-01 4.73259896e-01 1.36871815e+00 -3.09366789e-02
9.59494293e-01 -2.41527796e+00 -7.51283824e-01 6.47763610e-01
1.35357156e-01 6.01188481e-01 2.79229522e-01 3.97588499e-02
-3.69371139e-02 2.49578238e-01 2.47198660e-02 -2.23720055e-02
3.41017805e-02 3.20566446e-01 -1.20408690e+00 8.72476697e-01
-8.87705833e-02 5.51700532e-01 -5.36311030e-01 -2.16396363e-03
1.29236177e-01 3.64900261e-01 -6.41689181e-01 -2.60920465e-01
-2.19819099e-01 -2.23451126e-02 -5.75068235e-01 5.65194070e-01
8.12066913e-01 2.68173009e-01 5.15428960e-01 -1.46484375e-01
-3.67941201e-01 6.31486893e-01 -1.11838567e+00 9.16726887e-01
-5.67474306e-01 4.71605331e-01 1.75985456e-01 -1.58459187e-01
7.87727833e-01 -7.29657570e-03 -1.57339782e-01 -4.02393520e-01
5.76159418e-01 2.97008991e-01 1.33946165e-01 2.71433949e-01
6.09311998e-01 -4.68351804e-02 -7.20449865e-01 7.36552060e-01
-3.59648466e-01 -1.48929968e-01 -5.71069062e-01 4.72993143e-02
1.42631578e+00 -1.17657043e-01 6.48226589e-02 -3.85869771e-01
5.99086463e-01 2.09655508e-01 5.87318182e-01 7.08278954e-01
-1.76547721e-01 -1.82437003e-01 5.16719103e-01 -3.49142104e-01
-7.12453723e-01 -1.10247624e+00 -1.53598562e-01 6.32049203e-01
1.41958505e-01 -9.21146035e-01 -8.53373945e-01 -1.06262362e+00
2.79008001e-01 1.08415306e+00 -2.02686921e-01 -8.74874473e-01
-6.57801330e-01 -2.00953603e-01 1.51484787e+00 5.68547428e-01
2.55097479e-01 -1.77501068e-01 -1.15607440e+00 -9.53817964e-02
8.91298354e-01 -1.33806336e+00 -3.29100519e-01 4.22127962e-01
-7.46893466e-01 -1.13810325e+00 9.65282500e-01 -2.27967486e-01
6.44423842e-01 3.17391694e-01 6.51517391e-01 5.65621495e-01
-4.14654195e-01 2.71154583e-01 -2.60539516e-03 -5.69586992e-01
-1.12265003e+00 -7.09481388e-02 4.53652054e-01 -1.95862502e-01
4.24651057e-01 -2.62294501e-01 -9.40841064e-02 4.05259073e-01
-8.88691962e-01 -3.40824753e-01 3.01385105e-01 7.66103804e-01
2.02426776e-01 4.30729568e-01 1.74111232e-01 -7.52954781e-01
-5.56874648e-02 -3.13518167e-01 -1.46612430e+00 -2.04667106e-01
-7.64293730e-01 2.09537640e-01 1.07874739e+00 -5.21043777e-01
-6.34025037e-01 3.02614242e-01 -1.93702996e-01 -5.52609742e-01
-9.82845277e-02 -1.68209914e-02 -5.52093506e-01 -6.99415445e-01
8.11989367e-01 -2.81082302e-01 2.19881490e-01 -3.36801000e-02
8.32139924e-02 4.89072740e-01 6.35712206e-01 -7.14632452e-01
1.30279183e+00 5.98547399e-01 4.09472913e-01 -7.48936176e-01
-1.04881071e-01 1.41149670e-01 -1.71664953e-01 -5.18527403e-02
5.00536144e-01 -7.54102051e-01 -1.53609705e+00 4.17214930e-01
-9.02141631e-01 -3.33462983e-01 1.96518973e-01 4.34965402e-01
-3.54683287e-02 3.95441562e-01 -6.14172459e-01 -6.77346110e-01
-4.44323659e-01 -1.49587989e+00 7.70205021e-01 8.20525661e-02
-5.19654572e-01 -5.25027156e-01 -3.44268411e-01 3.58909488e-01
2.94805646e-01 4.25832093e-01 1.01373482e+00 -7.33503997e-01
-6.48614228e-01 -6.03658557e-01 5.15773177e-01 4.00411248e-01
-2.69129872e-01 2.75947452e-01 -1.13793790e+00 -6.57752991e-01
3.68443280e-01 -1.24593534e-01 5.33132792e-01 -5.23586929e-01
9.10530865e-01 -1.94241673e-01 -5.63469768e-01 6.20106459e-01
1.18410194e+00 4.32243854e-01 4.37818497e-01 4.51815933e-01
5.37676513e-01 1.32990420e-01 6.57652080e-01 -2.84527261e-02
-2.60137558e-01 7.26490796e-01 7.31453717e-01 1.50862977e-01
4.08604234e-01 -3.83865654e-01 9.96222496e-01 1.83008119e-01
7.83941150e-01 2.42645398e-01 -8.90268266e-01 -1.75674766e-01
-1.10330379e+00 -5.69949031e-01 -2.61995077e-01 2.66540599e+00
7.45903015e-01 1.05459082e+00 -2.24490270e-01 3.89209151e-01
1.97801664e-01 -8.90296698e-02 -3.75299662e-01 -1.08103883e+00
4.79462445e-01 6.60412848e-01 1.30272102e+00 7.98447192e-01
-9.96416867e-01 9.66212809e-01 6.26009417e+00 8.15304458e-01
-1.63062799e+00 1.85140133e-01 -2.92828563e-03 -1.90707400e-01
-2.49026492e-01 5.90694904e-01 -1.50739849e+00 3.99809629e-01
1.61618173e+00 3.32052745e-02 -1.11033469e-01 1.17330360e+00
-1.69693530e-01 -2.59355605e-01 -1.38791466e+00 3.19557726e-01
-5.16410992e-02 -1.10257041e+00 -3.30193073e-01 5.20460248e-01
-4.77776341e-02 -2.57896990e-01 3.34127218e-01 4.84605819e-01
3.18858683e-01 -8.87822449e-01 8.63976955e-01 -2.76145101e-01
7.46818006e-01 -1.29051757e+00 4.81013179e-01 3.18705529e-01
-8.71880770e-01 -1.11883998e-01 -2.57347804e-02 -2.67232865e-01
-1.33508384e-01 4.73782897e-01 -1.26125526e+00 2.86688298e-01
2.43434995e-01 -4.56328243e-01 -5.25308847e-01 2.25251615e-01
-4.64961767e-01 1.11177921e+00 -6.22671187e-01 1.74833804e-01
9.91000757e-02 3.49033266e-01 4.41576213e-01 9.62799013e-01
-5.64537244e-03 -3.24363649e-01 2.22388253e-01 5.41733384e-01
2.18862727e-01 -6.52053952e-01 -7.44755805e-01 8.80650133e-02
6.10992491e-01 1.19057035e+00 -7.47845888e-01 -6.88149333e-02
-3.58623207e-01 5.19916475e-01 -2.13698134e-01 -2.24378780e-01
-1.31652296e+00 -3.91952515e-01 1.21575689e+00 1.17406778e-01
8.63149092e-02 -6.80811763e-01 -3.91979933e-01 -8.01974535e-01
-4.29128073e-02 -1.24256897e+00 3.44494820e-01 2.78065428e-02
-6.36186123e-01 4.84524548e-01 -2.65655946e-02 -9.96035457e-01
-3.96448970e-01 -7.00966656e-01 -4.05154884e-01 5.65377593e-01
-1.08320546e+00 -1.25560987e+00 2.18636438e-01 3.85400474e-01
-2.68477350e-01 -3.46842259e-01 9.47251618e-01 2.27715552e-01
-6.24617040e-01 1.31809485e+00 -3.36258084e-01 -1.60874531e-01
6.98924005e-01 -5.85609257e-01 8.94778550e-01 1.20453537e+00
-1.36889786e-01 1.21823668e+00 9.85634208e-01 -9.07010615e-01
-2.25120687e+00 -9.98027205e-01 3.40004116e-01 -5.33303618e-01
8.33931565e-01 -1.01873124e+00 -7.92094827e-01 9.34748173e-01
-7.42312148e-02 1.70302436e-01 8.39576066e-01 -7.97652453e-02
-1.02160192e+00 -3.10143292e-01 -1.29427683e+00 6.90464258e-01
1.80935025e-01 -8.31860423e-01 -2.63562024e-01 7.82592148e-02
9.41780388e-01 -6.17369473e-01 -6.72168612e-01 2.61872977e-01
7.16008425e-01 -5.82082808e-01 8.48883927e-01 -5.03143549e-01
-3.07846546e-01 -8.35766673e-01 -2.17687488e-01 -5.30389130e-01
2.97900528e-01 -9.69319284e-01 -4.86850798e-01 1.16461933e+00
2.99066722e-01 -1.26525867e+00 8.53011072e-01 7.03802645e-01
-1.70003414e-01 -3.68800282e-01 -1.16387784e+00 -1.03463650e+00
1.27152905e-01 -7.62543559e-01 8.69866252e-01 9.26490366e-01
2.26629257e-01 2.05566689e-01 -2.09269911e-01 8.40778708e-01
6.37149334e-01 -3.23084086e-01 9.94094908e-01 -7.57938683e-01
-7.28466034e-01 -9.04373303e-02 -7.01465607e-01 -6.41451836e-01
5.38601458e-01 -8.40406835e-01 -1.95217162e-01 5.58041316e-03
-3.85102987e-01 -5.16046762e-01 -2.78828293e-01 9.18053508e-01
5.27785957e-01 1.03830278e-01 2.38354132e-01 -1.03017651e-01
2.02149034e-01 -1.00197956e-01 1.90377593e-01 -2.64678776e-01
-1.26131341e-01 -2.25529317e-02 -3.40498835e-01 6.88009679e-01
8.08684886e-01 -7.10160613e-01 -5.84499896e-01 1.28210753e-01
1.12331949e-01 -2.20906094e-01 5.47952771e-01 -1.28493857e+00
4.37788844e-01 -5.93464710e-02 2.36050859e-01 -3.11844975e-01
4.15611416e-01 -1.20535254e+00 3.73337656e-01 1.02887404e+00
1.84336364e-01 1.46793485e-01 8.41721416e-01 2.57613569e-01
2.66635597e-01 -3.08595896e-01 7.59132922e-01 5.51324487e-01
-4.94352609e-01 -2.14942545e-01 -7.30569303e-01 -6.78605855e-01
1.48866534e+00 -9.47549492e-02 -8.09098721e-01 2.82191992e-01
-2.48067193e-02 -8.37562792e-03 1.02446401e+00 5.73001206e-01
4.18896705e-01 -9.40781951e-01 1.30943075e-01 8.62256825e-01
-7.26041645e-02 -6.31962419e-01 6.28778934e-02 3.45808864e-01
-5.86382210e-01 6.11457646e-01 -1.68920055e-01 -4.72668678e-01
-1.64966679e+00 1.06427789e+00 3.23725879e-01 1.68467104e-03
-6.43912673e-01 1.84945077e-01 -3.08149640e-04 -1.58849597e-01
-1.43742338e-01 -3.94834369e-01 3.40608656e-01 -4.49906051e-01
8.11973631e-01 -2.27330830e-02 5.89181483e-01 -3.10279280e-01
-6.45996153e-01 1.68400645e-01 -4.50055778e-01 -2.68785328e-01
4.26546156e-01 7.07695305e-01 -1.22853369e-01 1.10291041e-01
1.21052265e+00 5.49059749e-01 -7.67225325e-01 3.56972635e-01
-1.51015401e-01 -4.71484303e-01 2.95232952e-01 -4.54144031e-01
-6.88962758e-01 5.31636298e-01 4.89130855e-01 -1.36546880e-01
7.98642457e-01 -5.66284120e-01 1.14337671e+00 4.79132295e-01
8.85085702e-01 -7.64772654e-01 -2.27207363e-01 5.71050227e-01
4.85015917e-04 -5.22923231e-01 1.57759055e-01 -5.87337971e-01
-4.47882004e-02 1.11808991e+00 8.41079414e-01 -4.11914103e-02
6.53463364e-01 1.19918239e+00 3.70600373e-02 6.71521872e-02
-7.64654994e-01 6.40571058e-01 -1.98374957e-01 5.62306881e-01
-1.39773726e-01 1.79014549e-01 -2.09317729e-01 6.91691935e-01
-4.60137069e-01 -1.05254523e-01 9.89858031e-01 1.56703210e+00
-3.96332979e-01 -1.51390970e+00 -8.80706012e-01 4.28211987e-02
-5.17892122e-01 2.22310245e-01 -2.77907461e-01 9.18843687e-01
1.26632780e-01 1.04355383e+00 8.26257840e-03 -1.01389396e+00
1.70195192e-01 1.69083968e-01 2.66729772e-01 -5.30691028e-01
-9.76440072e-01 -3.03076595e-01 3.51424575e-01 -8.44801068e-01
6.65436447e-01 -4.06893462e-01 -1.58114409e+00 -5.74284196e-01
-3.53661418e-01 1.06182009e-01 1.05219460e+00 7.89045930e-01
4.80071843e-01 4.87011045e-01 7.58854926e-01 -4.07138795e-01
-8.73277187e-01 7.45429322e-02 -3.98754656e-01 -4.74244058e-01
4.48238671e-01 -8.14935923e-01 -5.80649793e-01 -4.20152873e-01] | [5.658563137054443, 7.450559139251709] |
7b51745a-2e43-474a-b48d-d193305a8def | rethinking-pseudo-labeled-sample-mining-for | null | null | https://openreview.net/forum?id=60GDNLY-m3M | https://openreview.net/pdf?id=60GDNLY-m3M | Rethinking Pseudo-labeled Sample Mining for Semi-Supervised Object Detection | Consistency-based method has been proved effective for semi-supervised learning (SSL). However, the impact of the pseudo-labeled samples' quality as well as the mining strategies for high quality training sample have rarely been studied in SSL. An intuitive idea is to select pseudo-labeled training samples by threshold. We find it essential the selection of these thresholds to the final result of SSL. Following this discovery, we propose SEAT (Score Ensemble with Adaptive Threshold), a simple and efficient semi-supervised learning object detection method, in which the high confidence pseudo-labels are selected for self-training. Apart from confidence score as the indicator of the sample's quality, we also introduce the scores of temporal consistency and augmentation consistency. The scores provide a more comprehensive description to the quality of each sample. To cope with the data distribution difference among categories, the adaptive threshold strategy is used to automatically determine the sample mining threshold for each category. We conduct experiments on PASCAL-VOC and MSCOCO, extensive results show that our method is competitive and can be easily combined with consistency-based methods. | ['Xiaokang Yang', 'Fei Wu', 'Wenming Tan', 'Yi Niu', 'ShiLiang Pu', 'Zhanzhan Cheng', 'Sanli Tang', 'Duo Li'] | 2021-01-01 | null | null | null | null | ['semi-supervised-object-detection'] | ['computer-vision'] | [ 7.18358159e-02 -1.62460700e-01 -4.75415558e-01 -7.94756353e-01
-4.74912018e-01 -2.91177630e-01 3.92677367e-01 1.62537754e-01
-4.98754829e-01 8.12869072e-01 -3.26185852e-01 1.58690482e-01
-3.64665687e-01 -5.87028503e-01 -4.31766242e-01 -8.20802689e-01
1.78681081e-03 4.86028641e-01 6.50570452e-01 1.31008580e-01
3.24624538e-01 2.95626909e-01 -2.15418601e+00 4.46636587e-01
1.17926133e+00 1.14860392e+00 3.63505691e-01 2.69605577e-01
-3.74179184e-01 7.04439223e-01 -6.85614824e-01 3.38247307e-02
6.99140057e-02 -6.37351096e-01 -8.51079524e-01 4.12550896e-01
2.67111570e-01 2.35609695e-01 6.61847115e-01 1.09790063e+00
1.19296528e-01 -7.32093304e-02 7.09984004e-01 -1.42927539e+00
1.39540091e-01 7.02037930e-01 -4.92040753e-01 7.49420971e-02
8.33410695e-02 -1.96978524e-01 8.30381095e-01 -1.05740440e+00
6.73599362e-01 9.00396526e-01 5.50755203e-01 3.54189217e-01
-8.68669093e-01 -6.67394817e-01 9.83609036e-02 5.62798858e-01
-1.33805192e+00 -3.46438557e-01 8.69226933e-01 -4.92911667e-01
1.44262522e-01 3.26929808e-01 7.22782195e-01 6.56947076e-01
-2.05320809e-02 9.82131660e-01 1.52507257e+00 -7.70106137e-01
4.07739252e-01 7.23058283e-01 5.31921506e-01 6.91695750e-01
2.86106527e-01 -4.56909835e-03 -6.11998796e-01 7.14596361e-03
1.04241222e-01 -2.07966402e-01 -3.65282744e-02 -4.87249076e-01
-9.14814293e-01 6.05085194e-01 6.70155510e-02 3.45010459e-01
-8.97182226e-02 -3.99739176e-01 4.56336290e-01 3.51070106e-01
3.68450880e-01 1.23966374e-01 -6.88509047e-01 1.58264786e-01
-1.01942253e+00 1.20359408e-02 5.82207918e-01 8.77088666e-01
7.81834602e-01 -1.87219396e-01 -3.15461487e-01 1.04165912e+00
3.10243219e-01 2.72459865e-01 6.03424132e-01 -7.49031782e-01
1.03635103e-01 1.07167959e+00 1.17629275e-01 -6.73701882e-01
-2.88025081e-01 -6.69427395e-01 -6.79794431e-01 3.29536051e-01
3.01067263e-01 1.30132556e-01 -7.16147840e-01 1.51035058e+00
5.97824395e-01 -5.12764696e-03 -4.47681174e-02 6.58019125e-01
6.52217686e-01 4.20216024e-01 1.99762899e-02 -9.41859782e-01
1.14687777e+00 -8.65310609e-01 -9.55008745e-01 1.03224359e-01
5.99590838e-01 -6.19635224e-01 1.07212782e+00 7.15975523e-01
-6.48143470e-01 -8.88144016e-01 -1.21764779e+00 6.84778690e-01
-4.23939943e-01 6.91467822e-01 4.72595483e-01 4.83502477e-01
-4.61358428e-01 8.76387656e-01 -5.47027230e-01 -2.97576517e-01
2.95671076e-01 2.39436761e-01 -1.17063761e-01 2.83274233e-01
-1.04872811e+00 7.25966871e-01 8.26304615e-01 -2.01652348e-02
-4.24255371e-01 -2.51449436e-01 -4.86526757e-01 -4.02770936e-01
5.83850682e-01 -3.49297747e-02 1.15232670e+00 -1.22847629e+00
-1.17900074e+00 9.98347878e-01 -6.85402974e-02 -4.36173111e-01
5.65652251e-01 5.32219782e-02 -7.73878157e-01 1.32439181e-01
4.62955922e-01 4.20275450e-01 7.77693689e-01 -1.32201791e+00
-1.23898757e+00 -3.97878289e-01 -6.86675966e-01 2.24209130e-01
-3.34539682e-01 5.02369069e-02 -6.13612831e-01 -2.57602006e-01
3.05404246e-01 -8.52851927e-01 -1.03470363e-01 -6.24216795e-02
-2.31315315e-01 -7.21049964e-01 9.01007235e-01 -2.48347834e-01
1.64379692e+00 -2.27169824e+00 -3.30617428e-01 5.04121006e-01
-6.47540540e-02 3.51559371e-01 2.87573218e-01 -5.39915413e-02
1.68997776e-02 -8.79931897e-02 -3.55225772e-01 -9.87513140e-02
-2.56498158e-01 1.25004783e-01 1.38690323e-01 2.37734064e-01
2.25213602e-01 3.34762663e-01 -9.46864665e-01 -1.21389651e+00
1.84370741e-01 -1.57965854e-01 -8.87011811e-02 3.69813800e-01
-1.53267354e-01 1.22784398e-01 -4.48793471e-01 6.17228627e-01
5.85350215e-01 -1.18095726e-01 2.13975921e-01 -4.01410073e-01
-1.70277551e-01 -3.90669797e-03 -1.75297081e+00 1.21655536e+00
-5.47544435e-02 1.86132029e-01 -1.25703231e-01 -7.47597992e-01
1.43584931e+00 7.56935552e-02 4.83596444e-01 -5.27715266e-01
1.88550502e-02 4.55614358e-01 -7.75163844e-02 -5.94689727e-01
4.91364598e-01 4.99929748e-02 1.97882149e-02 3.78183395e-01
-6.13621995e-02 -1.31598013e-02 7.24676728e-01 1.31957769e-01
6.13500595e-01 4.38232958e-01 4.56865579e-01 -4.38509464e-01
7.89887369e-01 3.36794406e-01 9.02384460e-01 5.32129288e-01
-4.29006100e-01 3.31120253e-01 2.51489788e-01 -3.24857146e-01
-7.77941406e-01 -8.29838037e-01 -4.65921760e-01 9.17617083e-01
2.96550721e-01 -3.33962917e-01 -7.07794905e-01 -1.09863651e+00
-2.53760014e-02 4.97564793e-01 -5.62774956e-01 -2.44429544e-01
-3.69746804e-01 -7.27465510e-01 6.10502698e-02 3.73782128e-01
7.84516454e-01 -1.17567742e+00 -5.52135646e-01 3.46695572e-01
-2.08812088e-01 -8.73108864e-01 -3.08994293e-01 5.39944708e-01
-9.83485341e-01 -1.41329241e+00 -2.50536148e-02 -8.38401437e-01
8.22053254e-01 1.16000846e-01 1.10739553e+00 2.10991666e-01
-2.23308355e-01 7.83977956e-02 -7.04423487e-01 -5.10953605e-01
-7.36145318e-01 -9.46269371e-04 1.78034842e-01 3.01462591e-01
4.38924968e-01 -1.77720949e-01 -3.41584414e-01 8.82984996e-01
-6.47320867e-01 3.72629948e-02 6.09480560e-01 7.95387864e-01
9.86110091e-01 4.66478914e-01 8.16922009e-01 -1.29908741e+00
4.53561723e-01 -2.54432708e-01 -5.42066336e-01 4.64107335e-01
-1.17665386e+00 2.39912391e-01 7.51727700e-01 -4.02785301e-01
-1.14234221e+00 2.89375156e-01 1.32716492e-01 -2.15104491e-01
-2.98033506e-01 3.68349016e-01 -2.13403448e-01 2.54677117e-01
6.84766769e-01 2.25361198e-01 9.29806679e-02 -5.34357667e-01
-3.58085297e-02 9.09706950e-01 5.47171056e-01 -5.21608591e-01
6.19153976e-01 4.08337921e-01 -3.36975791e-02 -4.96811330e-01
-1.21185458e+00 -7.66114116e-01 -9.36368763e-01 -5.67707777e-01
5.78553498e-01 -6.62091851e-01 -4.08939391e-01 3.11928928e-01
-6.47043407e-01 -2.11087223e-02 -3.45666587e-01 5.11466384e-01
-3.73620480e-01 3.83435428e-01 -1.27714574e-01 -1.14381552e+00
-3.91283095e-01 -7.43958890e-01 6.69460952e-01 2.43629560e-01
-2.11964548e-01 -6.36072516e-01 -5.87594919e-02 3.64779443e-01
-7.08232895e-02 2.59130836e-01 5.24653077e-01 -7.58288026e-01
-3.00443083e-01 -2.32339457e-01 -3.70771438e-02 6.83125973e-01
9.98456478e-02 2.61391133e-01 -8.18576097e-01 -2.82211930e-01
-1.71789934e-03 -4.37907189e-01 8.59816551e-01 1.99357659e-01
1.28871453e+00 -9.47733372e-02 -4.52673286e-01 1.98323503e-01
1.47377491e+00 4.40553188e-01 3.39130640e-01 4.09448087e-01
3.71043861e-01 6.67313933e-01 1.66330612e+00 5.83661318e-01
8.75390507e-03 6.10816658e-01 2.75742292e-01 4.16193791e-02
-8.60049129e-02 -1.61774993e-01 3.24476779e-01 1.04570460e+00
7.20962510e-02 1.72990710e-01 -5.30189335e-01 3.82315665e-01
-1.84129655e+00 -9.61047351e-01 -4.38346058e-01 2.27464962e+00
1.06366062e+00 6.25070393e-01 3.01307678e-01 7.98271000e-01
8.70051265e-01 -2.69649237e-01 -3.98023546e-01 -1.68914810e-01
-2.25269869e-01 -6.09469116e-02 2.86818832e-01 1.84857652e-01
-1.24236941e+00 6.68224871e-01 6.10776711e+00 1.23449290e+00
-7.93033361e-01 3.73710841e-02 6.66211367e-01 1.73924044e-01
2.59282882e-04 6.19673636e-03 -1.22138870e+00 6.84291661e-01
6.36097789e-01 6.49757609e-02 -1.29814461e-01 1.07065916e+00
3.31103146e-01 -5.73204994e-01 -1.10554981e+00 8.36277544e-01
2.66758408e-02 -9.71557915e-01 -1.42234311e-01 -2.01014310e-01
7.97662258e-01 -4.04229254e-01 -3.01268011e-01 2.58574158e-01
2.67391831e-01 -4.77781773e-01 1.00073588e+00 4.40147668e-01
8.21323276e-01 -9.06512678e-01 8.52913380e-01 6.53171659e-01
-1.39799869e+00 -2.13718861e-01 -3.88084471e-01 2.25158587e-01
-3.24134856e-01 9.27374423e-01 -9.59217012e-01 7.11290717e-01
7.42094040e-01 6.47370756e-01 -8.64496410e-01 1.07939625e+00
-3.17564070e-01 8.71615529e-01 -2.57213235e-01 -3.41587156e-01
-8.08155760e-02 -2.44660005e-01 2.38426894e-01 1.31495929e+00
1.01204433e-01 -1.19994804e-01 3.49586129e-01 7.30282426e-01
3.67788434e-01 4.23770547e-01 -3.11160415e-01 3.89825344e-01
8.37280095e-01 1.16131175e+00 -1.00576675e+00 -4.94787246e-01
-1.64681468e-02 6.66893363e-01 3.08864832e-01 -8.98236185e-02
-7.65192688e-01 -3.68683815e-01 -8.15016478e-02 1.17709443e-01
1.20901570e-01 8.22661966e-02 -4.89965171e-01 -7.31246173e-01
1.86189160e-01 -6.63703859e-01 6.44251347e-01 -5.39774299e-01
-1.07121837e+00 5.71075201e-01 7.31695518e-02 -1.67714214e+00
-7.13074133e-02 -3.14404219e-01 -5.45016170e-01 3.27078938e-01
-1.29137981e+00 -7.88625658e-01 -4.92758811e-01 5.12466490e-01
7.09915578e-01 -4.10642475e-01 4.27821636e-01 1.39597356e-01
-7.56063163e-01 4.34426874e-01 1.08115591e-01 9.82678775e-03
7.41008341e-01 -1.33305562e+00 -4.49572831e-01 9.78627682e-01
1.51566133e-01 2.52799183e-01 7.24001825e-01 -7.26830721e-01
-8.42311203e-01 -1.18302679e+00 8.92959535e-01 -2.89694935e-01
2.07755908e-01 1.94485653e-02 -8.97877812e-01 5.62968701e-02
-2.57825404e-01 -7.33607709e-02 6.40667498e-01 4.23875712e-02
-1.04692101e-01 -7.05749035e-01 -1.16623342e+00 2.50823826e-01
9.81487334e-01 -6.71924874e-02 -6.05061889e-01 1.43264115e-01
4.48792934e-01 -9.75188017e-02 -7.49506593e-01 7.82697797e-01
3.77457350e-01 -1.20679176e+00 4.45793420e-01 -5.52873388e-02
1.31842598e-01 -7.71915853e-01 1.96632817e-01 -1.03677392e+00
-3.74997944e-01 2.84689907e-02 1.73155054e-01 1.52709448e+00
4.96147692e-01 -2.02594608e-01 8.41371238e-01 2.43724644e-01
-1.10276818e-01 -9.15495336e-01 -7.69915879e-01 -9.74280536e-01
-6.72128022e-01 -4.61148351e-01 4.36586082e-01 1.00524247e+00
-8.69903415e-02 3.25036496e-01 -2.98805505e-01 -7.70525187e-02
8.76073241e-01 3.67379069e-01 7.34703422e-01 -1.64492452e+00
-2.02404916e-01 -4.01828766e-01 -1.47982344e-01 -6.14490211e-01
-1.02418855e-01 -7.44532406e-01 1.55275971e-01 -1.34689355e+00
3.25884283e-01 -8.49095404e-01 -4.65646684e-01 4.50740427e-01
-2.43679821e-01 1.10028714e-01 -1.16108395e-01 4.88550395e-01
-1.03321004e+00 2.51108199e-01 1.22665823e+00 7.24643916e-02
-4.63758588e-01 4.10717309e-01 -2.91691244e-01 6.69410527e-01
7.95346797e-01 -6.41718805e-01 -4.82374460e-01 3.66083562e-01
5.03553599e-02 -1.80594906e-01 -1.33151665e-01 -1.19207251e+00
2.17672572e-01 -4.46806163e-01 4.39383924e-01 -1.07555079e+00
-2.25707412e-01 -9.10699546e-01 1.56639576e-01 6.95257127e-01
-4.26935405e-01 -3.40303987e-01 -1.99209258e-01 5.21788836e-01
-3.75551730e-01 -5.89391947e-01 1.18941700e+00 -9.15941000e-02
-9.68827963e-01 2.02233583e-01 -9.84791592e-02 -1.07690938e-01
1.19770551e+00 -5.68189859e-01 2.82050848e-01 2.18343303e-01
-7.60193646e-01 4.15816367e-01 3.01198155e-01 2.77270615e-01
5.41738927e-01 -1.38777173e+00 -6.19769514e-01 4.23813075e-01
4.74737078e-01 1.61096320e-01 -1.30209044e-01 6.47270501e-01
-2.04283252e-01 1.75828990e-02 -1.36333212e-01 -8.62959027e-01
-1.65099585e+00 4.54339176e-01 8.02143663e-02 -3.81801456e-01
-1.29007816e-01 7.25189626e-01 -4.97448266e-01 -2.19064012e-01
5.86390316e-01 -2.39071444e-01 -6.66047573e-01 3.28282475e-01
3.39094430e-01 4.54369992e-01 1.27151266e-01 -2.74052620e-01
-4.06184644e-01 5.64782023e-01 7.64200464e-02 9.90556926e-02
1.01052439e+00 -1.39440358e-01 -1.52392894e-01 9.28777397e-01
9.84152555e-01 -1.33626014e-01 -1.10985899e+00 -4.11144435e-01
4.64382410e-01 -2.65552551e-01 -2.89855450e-01 -8.18874300e-01
-8.99630249e-01 3.76845568e-01 7.87054479e-01 1.75201789e-01
1.18165112e+00 -6.64950758e-02 3.27254802e-01 2.28704154e-01
5.20696282e-01 -1.75806057e+00 3.25732052e-01 2.36658409e-01
5.58527052e-01 -1.40936136e+00 1.65735051e-01 -5.80423594e-01
-7.41674125e-01 1.09609890e+00 9.54918921e-01 1.34624958e-01
7.03691900e-01 2.46310100e-01 -2.98096729e-03 1.70956377e-03
-9.02437508e-01 -2.83336073e-01 3.84531945e-01 3.78297418e-01
2.52153993e-01 1.72269732e-01 -7.67296255e-01 5.93895793e-01
-1.31898791e-01 1.93524957e-01 2.32820183e-01 9.89869833e-01
-9.40409124e-01 -1.17478871e+00 -5.19058526e-01 6.60277069e-01
-1.41501367e-01 4.55939114e-01 -3.28139842e-01 5.99431396e-01
6.37102902e-01 9.99804318e-01 1.74600892e-02 -5.92211246e-01
3.05869520e-01 2.72969753e-01 2.30790898e-01 -6.46610141e-01
-4.18570787e-01 7.21689761e-02 1.91637784e-01 -2.89381504e-01
-7.70814240e-01 -6.83863878e-01 -1.39103281e+00 7.56881163e-02
-8.06671441e-01 6.25627816e-01 7.15191424e-01 8.99322927e-01
-4.30581234e-02 2.59942800e-01 1.13703191e+00 -1.71886891e-01
-6.84364140e-01 -9.11792517e-01 -8.32940340e-01 6.82104111e-01
-1.63092896e-01 -6.66211247e-01 -5.41123867e-01 1.65103883e-01] | [9.204261779785156, 3.9933085441589355] |
22ed904b-e5a5-4d34-beef-6b4d4d4f3b39 | applicability-and-interpretation-of-the | 1803.03104 | null | http://arxiv.org/abs/1803.03104v1 | http://arxiv.org/pdf/1803.03104v1.pdf | Applicability and interpretation of the deterministic weighted cepstral distance | Quantifying similarity between data objects is an important part of modern
data science. Deciding what similarity measure to use is very application
dependent. In this paper, we combine insights from systems theory and machine
learning, and investigate the weighted cepstral distance, which was previously
defined for signals coming from ARMA models. We provide an extension of this
distance to invertible deterministic linear time invariant single input single
output models, and assess its applicability. We show that it can always be
interpreted in terms of the poles and zeros of the underlying model, and that,
in the case of stable, minimum-phase, or unstable, maximum-phase models, a
geometrical interpretation in terms of subspace angles can be given. We then
devise a method to assess stability and phase-type of the generating models,
using only input/output signal information. In this way, we prove a connection
between the extended weighted cepstral distance and a weighted cepstral model
norm. In this way, we provide a purely data-driven way to assess different
underlying dynamics of input/output signal pairs, without the need for any
system identification step. This can be useful in machine learning tasks such
as time series clustering. An iPython tutorial is published complementary to
this paper, containing implementations of the various methods and algorithms
presented here, as well as some numerical illustrations of the equivalences
proven here. | ['Oliver Lauwers', 'Bart De Moor'] | 2018-03-08 | null | null | null | null | ['time-series-clustering'] | ['time-series'] | [ 3.38049978e-01 -1.05683059e-01 2.06121117e-01 -2.81389579e-02
-1.72834173e-01 -9.68515575e-01 6.35119677e-01 2.15237439e-01
-2.55675465e-01 3.65357429e-01 -1.23671949e-01 -4.36516166e-01
-7.80317724e-01 -5.09272933e-01 -1.90072909e-01 -1.09432507e+00
-3.00461113e-01 1.62410870e-01 1.34612257e-02 -4.45713848e-01
1.99253753e-01 6.03039980e-01 -1.59082007e+00 -6.66241571e-02
4.61613059e-01 1.02733171e+00 -8.94214436e-02 8.81923914e-01
2.97942281e-01 5.47555014e-02 -7.38839805e-01 5.97031340e-02
2.78134018e-01 -7.86072016e-01 -6.77359879e-01 -5.99560924e-02
-1.25483602e-01 4.84021902e-01 -1.73677988e-02 8.57947767e-01
4.82315302e-01 1.11883126e-01 1.03639591e+00 -1.29015040e+00
5.59137166e-02 3.84689003e-01 -1.04973368e-01 3.42579573e-01
3.89680326e-01 -1.68883935e-01 7.17266083e-01 -6.62744164e-01
2.87607193e-01 9.06641424e-01 7.35130966e-01 4.37911227e-02
-1.54834461e+00 -1.81680784e-01 -2.19041467e-01 2.75448889e-01
-1.33053386e+00 -2.91945457e-01 9.71892059e-01 -7.01171875e-01
5.37411928e-01 5.60626447e-01 6.87493324e-01 6.50697589e-01
1.11016624e-01 3.62119734e-01 1.09949565e+00 -8.93856227e-01
2.68762946e-01 1.04638308e-01 5.16619325e-01 3.84889215e-01
8.68398175e-02 2.90691871e-02 5.72287594e-04 -2.19932348e-01
5.92076063e-01 -2.14789331e-01 -5.33915997e-01 -6.91576600e-01
-1.26590514e+00 7.95387685e-01 4.63728756e-02 9.40907001e-01
-6.14847466e-02 -2.22702533e-01 2.51745939e-01 7.02802718e-01
2.39585996e-01 4.37815458e-01 -2.51525789e-01 -1.00751042e-01
-6.38905048e-01 6.77039400e-02 1.02728009e+00 4.71537322e-01
4.99440968e-01 1.17186859e-01 3.67548794e-01 5.43419838e-01
2.03620285e-01 6.26849234e-01 6.72035694e-01 -7.32292950e-01
8.35132897e-02 3.50085706e-01 -1.74465090e-01 -1.27248430e+00
-7.60938287e-01 -3.99945587e-01 -8.86968434e-01 2.01397821e-01
7.65947461e-01 -2.66063102e-02 -1.78716689e-01 1.68691027e+00
1.98598474e-01 1.27096131e-01 1.76350892e-01 6.45572662e-01
1.30432531e-01 4.77728993e-01 -5.16576350e-01 -7.73910522e-01
1.23754907e+00 -1.18087031e-01 -7.51692116e-01 3.43167901e-01
5.38652301e-01 -8.27906728e-01 8.12593162e-01 6.50953650e-01
-9.56745863e-01 -5.29226363e-01 -1.10977733e+00 3.98915887e-01
-5.43242037e-01 1.53409600e-01 -1.32881040e-02 6.70716584e-01
-8.66347373e-01 8.97413135e-01 -7.44910598e-01 -4.13202167e-01
-5.31550050e-01 2.38920793e-01 -2.33817711e-01 6.81014121e-01
-1.11268675e+00 9.26908016e-01 3.37283492e-01 3.54368180e-01
-3.96669656e-02 -2.81501323e-01 -6.11274183e-01 -5.73231503e-02
-1.01771411e-02 -2.61535704e-01 1.18894303e+00 -9.89075720e-01
-1.42983210e+00 5.45395613e-01 -1.56162232e-01 -4.78193372e-01
3.88844520e-01 3.31499845e-01 -8.01304102e-01 2.41419926e-01
-3.98600578e-01 -2.50949144e-01 1.11977065e+00 -9.61371779e-01
-2.60749370e-01 -3.96433860e-01 -1.44752592e-01 -9.51454416e-02
-2.52502322e-01 -1.73470348e-01 2.99449831e-01 -5.42321920e-01
4.82362866e-01 -9.61007237e-01 -3.12745273e-02 -2.62895137e-01
-7.74172619e-02 -1.83629096e-01 7.39671469e-01 -4.35144842e-01
1.51038253e+00 -2.27562308e+00 3.73766065e-01 7.18363583e-01
-1.43553251e-02 2.58166224e-01 1.59447670e-01 8.49718332e-01
-4.76816297e-01 -1.91008732e-01 -4.93698120e-01 2.54328907e-01
2.81463787e-02 -5.69056757e-02 -6.05568707e-01 7.85484493e-01
2.82925993e-01 3.73841673e-01 -8.45687687e-01 -1.89166486e-01
3.95629436e-01 5.10242045e-01 -1.48152381e-01 -2.74738163e-01
3.37495208e-01 5.56913078e-01 -1.21013671e-01 -3.35001573e-02
3.15935612e-01 1.76039442e-01 2.31087863e-01 -3.83681238e-01
-2.90526718e-01 1.35816649e-01 -1.71123815e+00 1.07792425e+00
-5.41735947e-01 9.42689598e-01 1.23533040e-01 -1.56407630e+00
1.04283428e+00 5.38385153e-01 6.05394363e-01 -3.35700452e-01
3.23257327e-01 4.67416257e-01 2.95249283e-01 -4.03052270e-01
8.86521786e-02 -1.90985680e-01 -9.95372236e-02 4.38453883e-01
1.03579707e-01 -2.74531335e-01 3.88352334e-01 -2.48459280e-01
7.86120355e-01 -8.19795206e-02 7.68364489e-01 -5.47255337e-01
8.89235258e-01 -3.61372828e-01 1.00793816e-01 2.79836744e-01
2.61423010e-02 5.76796114e-01 6.84337795e-01 -2.47018173e-01
-9.13732469e-01 -1.02224374e+00 -4.42887723e-01 4.72170413e-01
3.26253660e-03 -4.43403244e-01 -6.85367703e-01 -2.40106676e-02
-1.43774793e-01 3.67865562e-01 -4.38896447e-01 -2.83305705e-01
-7.25352168e-01 -6.47809088e-01 3.82265240e-01 1.63298711e-01
-1.55035511e-01 -6.64322555e-01 -8.37420464e-01 2.53625989e-01
-1.34569123e-01 -7.62250662e-01 -9.15172473e-02 4.57120925e-01
-1.10394692e+00 -1.16875088e+00 -7.64542162e-01 -6.11277938e-01
3.68600726e-01 1.04794830e-01 8.31019878e-01 -2.56825089e-01
-1.26312822e-01 6.07956469e-01 -2.75579542e-01 -4.59770322e-01
-5.27164817e-01 -2.00254351e-01 4.71003264e-01 3.56096268e-01
7.35242069e-02 -8.65605116e-01 -4.30070877e-01 6.84397936e-01
-1.06120288e+00 -3.48054469e-01 2.41392970e-01 6.42045140e-01
3.18228424e-01 2.15503737e-01 5.45535028e-01 -2.75892466e-01
7.40128219e-01 -3.17221165e-01 -6.13825798e-01 1.74276829e-01
-5.03676653e-01 2.40696892e-01 9.23234642e-01 -5.74637830e-01
-4.49831545e-01 2.29265913e-01 -3.32143568e-02 -2.20537305e-01
-9.65377539e-02 4.81162578e-01 -8.97668451e-02 -6.41565993e-02
7.86575973e-01 1.78961754e-01 1.25559986e-01 -5.19113779e-01
4.72466111e-01 8.65921021e-01 5.96683741e-01 -3.56847197e-01
8.13074946e-01 4.26366508e-01 2.45174378e-01 -1.32807577e+00
-1.52870446e-01 -7.16697514e-01 -9.09584701e-01 -2.25654721e-01
4.18068051e-01 -2.87469596e-01 -8.26784492e-01 5.00230491e-01
-1.20578134e+00 4.34871428e-02 -5.39036572e-01 7.64078319e-01
-9.53706086e-01 4.61050421e-01 -2.22606122e-01 -1.24931812e+00
2.44944915e-02 -7.97099292e-01 9.42367077e-01 6.09836821e-03
-4.62319762e-01 -1.20030224e+00 3.89971316e-01 -3.05679142e-01
1.57678455e-01 2.88805991e-01 7.69035101e-01 -7.32290983e-01
9.93694514e-02 -3.59061629e-01 4.57328945e-01 3.91360462e-01
7.30488151e-02 2.44198501e-01 -9.24704194e-01 -2.96984404e-01
5.21684408e-01 4.33439702e-01 5.86598992e-01 4.12584722e-01
6.87762022e-01 -3.24211389e-01 -2.10288554e-01 2.79936999e-01
1.36498642e+00 4.44614947e-01 3.77656549e-01 -8.27686861e-02
1.90406024e-01 9.39455390e-01 3.41533214e-01 3.06265533e-01
-1.08762883e-01 9.00342584e-01 1.73724443e-01 1.28296599e-01
3.61401767e-01 1.15106240e-01 4.87480909e-01 1.17056048e+00
-2.13642627e-01 8.83400366e-02 -7.77306259e-01 4.37430888e-01
-1.64683771e+00 -1.16468585e+00 -6.02997482e-01 2.54551935e+00
5.61290920e-01 1.11449718e-01 5.28205574e-01 1.07997727e+00
8.02417755e-01 -1.68103620e-01 -5.14125079e-02 -5.71378410e-01
-2.97940791e-01 4.58393574e-01 2.27610275e-01 6.94679558e-01
-1.00265634e+00 2.46992093e-02 6.21366167e+00 7.55718291e-01
-1.25342059e+00 -2.44042262e-01 1.03369867e-02 1.18418492e-01
-5.68547212e-02 4.57166368e-03 -2.90012330e-01 5.93200922e-01
1.25939667e+00 -5.80224514e-01 2.74579138e-01 4.30973500e-01
3.87581319e-01 -1.01713136e-01 -1.09804761e+00 1.09770167e+00
-4.32000048e-02 -7.03627110e-01 -3.40168208e-01 8.73455480e-02
2.88169742e-01 -5.54093659e-01 -1.12344623e-02 -7.89776966e-02
-5.03455579e-01 -7.58005500e-01 7.07378507e-01 7.61950791e-01
3.58971953e-01 -5.59054255e-01 5.90562701e-01 4.67876583e-01
-1.52909529e+00 -2.16708675e-01 -1.73750482e-02 -3.42580467e-01
2.57080823e-01 7.97376275e-01 -5.12153208e-01 8.59762669e-01
2.36114740e-01 6.28439546e-01 -2.58787930e-01 1.14302373e+00
1.21435702e-01 5.08264244e-01 -6.33569300e-01 -1.48426354e-01
-1.15782902e-01 -5.82657516e-01 7.22525835e-01 1.12371385e+00
6.14829659e-01 1.13856383e-01 -3.14069927e-01 5.38226664e-01
7.89855599e-01 1.53551385e-01 -8.31273615e-01 -3.88338640e-02
2.42082492e-01 1.27128768e+00 -9.47758138e-01 -5.17693441e-03
-2.38589704e-01 6.62790120e-01 -4.91023034e-01 2.84805208e-01
-5.31544685e-01 -8.23393345e-01 4.15834785e-01 2.49369200e-02
1.02159739e-01 -4.13541377e-01 -6.46323115e-02 -1.01792026e+00
2.43486986e-01 -7.44452119e-01 2.72683740e-01 -3.90785038e-01
-1.01465774e+00 6.46485031e-01 3.39687616e-01 -1.83188593e+00
-9.58479047e-01 -8.04887652e-01 -6.70110404e-01 8.81016076e-01
-8.87805402e-01 -4.40145284e-01 1.17414862e-01 6.22837067e-01
-5.27474703e-03 1.09610818e-01 8.88275027e-01 1.81345478e-01
-3.42652172e-01 1.67872861e-01 4.24106240e-01 -5.38301840e-02
4.15187716e-01 -1.33891833e+00 1.38422221e-01 6.74728930e-01
5.80719531e-01 6.83186710e-01 1.16965759e+00 4.86294217e-02
-1.13214660e+00 -4.55857873e-01 1.00911522e+00 -4.19131219e-01
9.16713238e-01 -2.29847938e-01 -8.30213308e-01 2.17061266e-01
8.16612095e-02 -8.82460847e-02 5.44778347e-01 -1.77523538e-01
-7.95589313e-02 -3.56627733e-01 -8.21444988e-01 5.10039747e-01
6.82215929e-01 -6.79710329e-01 -8.14504743e-01 8.37100521e-02
2.96147972e-01 -2.70064995e-02 -9.80655253e-01 3.72109711e-01
7.20892608e-01 -1.20331252e+00 8.72033119e-01 -3.70772928e-01
-2.09949613e-01 -6.64774776e-01 -1.44499496e-01 -1.43533158e+00
-1.16609856e-01 -1.00411761e+00 -1.31163016e-01 9.33407485e-01
3.32384467e-01 -9.21004474e-01 1.74219236e-01 -9.53773260e-02
7.24844933e-02 -6.72744393e-01 -1.07688415e+00 -1.12984502e+00
1.49102673e-01 -4.60779965e-01 1.81065902e-01 8.16238880e-01
6.45120263e-01 4.74465668e-01 2.06655543e-02 9.99828503e-02
3.01516831e-01 1.81766793e-01 5.10403872e-01 -1.56788599e+00
-4.83001202e-01 -7.92883515e-01 -8.86673808e-01 -7.26949155e-01
-1.87661573e-01 -7.99743772e-01 -1.13764018e-01 -9.56908703e-01
-6.24390006e-01 -2.07610175e-01 -3.85362118e-01 -8.94278809e-02
3.18282038e-01 2.53527194e-01 2.96482444e-01 2.69000649e-01
4.70202826e-02 2.93585926e-01 5.23893654e-01 1.93796232e-01
-3.01296204e-01 6.67804480e-01 -1.82856560e-01 9.98188376e-01
8.04571331e-01 -2.28790700e-01 -5.15587151e-01 1.94201961e-01
3.03885221e-01 6.96328878e-02 4.82174128e-01 -1.32575774e+00
2.06538022e-01 2.30823860e-01 7.05477297e-02 -4.79453206e-01
3.21601033e-01 -1.09192216e+00 3.87462020e-01 7.14650571e-01
-1.74024403e-01 2.68822670e-01 -6.07247613e-02 5.03537953e-01
-3.91939193e-01 -5.10944843e-01 7.33201683e-01 3.08375567e-01
-2.43449032e-01 -3.14283520e-01 -5.29648662e-01 -1.48892894e-01
1.09677625e+00 -3.98000717e-01 4.03238460e-02 -5.71464479e-01
-8.95600855e-01 -9.85463038e-02 1.52058214e-01 1.02115422e-01
2.32622787e-01 -1.39408565e+00 -4.18421328e-01 4.81488943e-01
5.23537546e-02 -5.76006413e-01 1.41619876e-01 1.39428508e+00
-3.05190682e-01 6.59387827e-01 -1.33117735e-01 -8.88374865e-01
-1.37285256e+00 5.67780077e-01 4.69206333e-01 1.39785670e-02
-1.55134514e-01 5.73164038e-02 -6.46447688e-02 -2.37567812e-01
3.71954329e-02 -7.02076316e-01 -3.57688069e-01 5.71246326e-01
5.07706523e-01 5.92238665e-01 3.66977632e-01 -7.24100053e-01
-4.55938220e-01 1.10437346e+00 7.31762826e-01 -5.41296601e-01
9.11018372e-01 -2.64677525e-01 -8.21383893e-02 1.11040819e+00
1.48546576e+00 1.21373184e-01 -8.02180409e-01 -1.31289199e-01
2.98071921e-01 7.31311366e-02 -3.69850427e-01 -2.59579420e-01
-6.40062928e-01 1.07789683e+00 7.38510966e-01 1.15902257e+00
1.48271263e+00 -1.29549265e-01 2.18230620e-01 4.40460712e-01
3.10018301e-01 -9.19629216e-01 -2.70003438e-01 4.14409518e-01
1.00319755e+00 -7.68586695e-01 -3.03223342e-01 -2.46719137e-01
-1.99525028e-01 1.61294889e+00 -2.95256823e-01 -3.78953218e-01
9.71114218e-01 2.70891905e-01 1.69146154e-02 2.72667468e-01
-5.76257527e-01 -4.48669821e-01 5.03713489e-01 7.08704948e-01
5.50568402e-01 1.60441734e-02 -7.49043465e-01 3.86033416e-01
-4.95109320e-01 -2.85165459e-01 6.50651693e-01 7.66801178e-01
-4.63472903e-01 -1.24629211e+00 -5.43582499e-01 1.95618376e-01
-2.42427036e-01 2.42713511e-01 -4.65434998e-01 8.29730093e-01
-1.49201632e-01 1.03909194e+00 2.00409457e-01 -4.75253642e-01
6.73681438e-01 3.35739404e-01 5.59488595e-01 -2.09769934e-01
-4.16742176e-01 1.82119042e-01 -2.51924008e-01 -2.36227483e-01
-6.07339144e-01 -6.52636945e-01 -9.99576509e-01 -2.14078322e-01
-4.63140935e-01 3.89517277e-01 8.21325600e-01 8.85914326e-01
3.21701467e-02 3.69773865e-01 9.39430296e-01 -8.86612833e-01
-8.20707321e-01 -8.79191458e-01 -7.61990309e-01 1.67199820e-01
7.48711169e-01 -4.01661962e-01 -6.88736320e-01 1.10313602e-01] | [7.439803600311279, 3.496417284011841] |
26c3a415-27e8-4161-86ed-35e75001cb16 | distance-preserving-machine-learning-for | 2307.02367 | null | https://arxiv.org/abs/2307.02367v1 | https://arxiv.org/pdf/2307.02367v1.pdf | Distance Preserving Machine Learning for Uncertainty Aware Accelerator Capacitance Predictions | Providing accurate uncertainty estimations is essential for producing reliable machine learning models, especially in safety-critical applications such as accelerator systems. Gaussian process models are generally regarded as the gold standard method for this task, but they can struggle with large, high-dimensional datasets. Combining deep neural networks with Gaussian process approximation techniques have shown promising results, but dimensionality reduction through standard deep neural network layers is not guaranteed to maintain the distance information necessary for Gaussian process models. We build on previous work by comparing the use of the singular value decomposition against a spectral-normalized dense layer as a feature extractor for a deep neural Gaussian process approximation model and apply it to a capacitance prediction problem for the High Voltage Converter Modulators in the Oak Ridge Spallation Neutron Source. Our model shows improved distance preservation and predicts in-distribution capacitance values with less than 1% error. | ['Sudarshan Harave', 'Sarah Cousineau', 'Majdi I. Radaideh', 'Jared Walden', 'Dan Lu', 'Chris Pappas', 'Thomas Britton', 'Kishansingh Rajput', 'Malachi Schram', 'Steven Goldenberg'] | 2023-07-05 | null | null | null | null | ['dimensionality-reduction'] | ['methodology'] | [-1.17559157e-01 5.04515171e-02 3.63842219e-01 -4.75293308e-01
-7.35119939e-01 -1.41985729e-01 5.73657990e-01 6.17300510e-01
-5.36102355e-01 5.95177174e-01 7.36668259e-02 -6.33622766e-01
-4.53388810e-01 -7.88291395e-01 -4.24230844e-01 -1.05566657e+00
-1.81953721e-02 1.26416469e+00 2.72245735e-01 2.64543384e-01
1.01536646e-01 1.09101427e+00 -8.02602112e-01 3.26818705e-01
3.77719462e-01 1.25499547e+00 -2.57746786e-01 3.53439093e-01
-1.33234337e-01 6.40951157e-01 -5.24001718e-01 -1.91834405e-01
8.57663080e-02 -1.32419765e-02 -4.57868993e-01 -7.39710510e-01
-1.48887619e-01 -1.39088467e-01 -7.17252433e-01 1.17508197e+00
5.89535475e-01 2.68441588e-01 1.18159318e+00 -1.00046432e+00
-1.01374805e-01 7.98119664e-01 -5.13372660e-01 1.85886174e-01
-2.77357340e-01 2.38222212e-01 5.81133604e-01 -5.47908902e-01
3.93554598e-01 1.13198721e+00 1.17788291e+00 4.46569890e-01
-1.74508417e+00 -5.86134970e-01 -5.51240921e-01 1.02626547e-01
-1.37417293e+00 -3.03108811e-01 5.63949883e-01 -5.18143773e-01
1.53108561e+00 -1.55989721e-01 2.94326782e-01 1.04257369e+00
9.73305285e-01 4.89016593e-01 7.25605667e-01 -1.76838860e-01
6.90553784e-01 -1.05737902e-01 2.69354403e-01 1.00148678e-01
3.80325109e-01 3.74824345e-01 -3.38764191e-01 -5.07350028e-01
7.68924177e-01 -9.65331271e-02 -1.79422781e-01 -6.34165645e-01
-8.75198722e-01 9.82962310e-01 4.67955172e-01 4.14481789e-01
-5.15852630e-01 5.89760303e-01 6.15539610e-01 2.28838474e-01
5.28091848e-01 7.32198834e-01 -3.90002310e-01 -3.58716875e-01
-1.27542174e+00 5.51985800e-01 9.88118649e-01 7.52233982e-01
1.60425022e-01 2.39735380e-01 -4.67548817e-01 3.50199640e-01
4.47775394e-01 3.44588697e-01 1.87636882e-01 -7.31149018e-01
2.14672402e-01 7.78643563e-02 -3.52724865e-02 -4.46447104e-01
-9.15745854e-01 -4.54503149e-01 -9.78647113e-01 6.67357028e-01
4.61043358e-01 -3.82644385e-01 -9.10604358e-01 1.11405945e+00
-1.41468784e-02 -3.10704172e-01 1.40324235e-01 5.83131254e-01
3.04497361e-01 5.10537565e-01 2.01749757e-01 -5.32119162e-02
1.16610575e+00 -4.70437855e-01 -6.54953480e-01 -1.64880544e-01
4.56442982e-01 -6.58365130e-01 2.40159899e-01 6.79077923e-01
-1.03812277e+00 -2.69759238e-01 -1.43029726e+00 -4.84757088e-02
9.35287103e-02 -3.30802768e-01 8.03446352e-01 9.58563566e-01
-9.15943146e-01 1.52732527e+00 -1.48253393e+00 8.84677395e-02
7.61266410e-01 6.22006595e-01 -2.10909113e-01 2.73560788e-02
-9.09305453e-01 1.06375897e+00 5.05019426e-01 1.31590813e-01
-9.50584888e-01 -9.44741488e-01 -5.84083259e-01 2.75467426e-01
-8.19530636e-02 -5.26970387e-01 1.56757832e+00 -8.49035606e-02
-1.23825014e+00 1.22980364e-02 -5.27021028e-02 -1.11068201e+00
6.15807235e-01 -1.20806947e-01 -4.34201151e-01 8.15498456e-02
-3.66348505e-01 2.77374089e-01 9.24981594e-01 -7.24817514e-01
-4.22059298e-01 -3.55952173e-01 -6.18752122e-01 -2.89214551e-01
1.40023023e-01 1.51857689e-01 8.71845558e-02 -1.37695342e-01
6.14088595e-01 -7.07822800e-01 -8.30256164e-01 -1.79694176e-01
-2.43227229e-01 -2.11879253e-01 7.73436964e-01 -6.79131627e-01
7.60882616e-01 -1.98183513e+00 -8.12248960e-02 6.21936142e-01
4.90118027e-01 -9.68512371e-02 4.03287083e-01 3.66367191e-01
-3.65267396e-01 -2.94297457e-01 -5.11020839e-01 -5.33286870e-01
9.75779146e-02 5.34668416e-02 -4.43609431e-02 5.96833527e-01
2.76842535e-01 6.80067182e-01 -6.42697573e-01 1.04799710e-01
1.96762651e-01 7.28114188e-01 -2.52290636e-01 -3.86375666e-01
-3.21065575e-01 3.20137113e-01 -1.88502803e-01 3.00573885e-01
7.86092877e-01 1.00058682e-01 -6.12504929e-02 -4.73417819e-01
1.49239272e-01 1.48045108e-01 -1.07284498e+00 1.70766556e+00
-3.24259043e-01 7.87507713e-01 1.73583522e-01 -7.29816794e-01
9.84107316e-01 2.20068663e-01 6.79235935e-01 -5.71595967e-01
3.53491187e-01 2.89747238e-01 4.13500369e-01 -6.33453112e-03
5.99185407e-01 -1.02055156e+00 -3.47833112e-02 2.77404487e-01
3.64548206e-01 -5.49219131e-01 1.37641886e-02 5.47631457e-02
1.35858834e+00 2.86434665e-02 -1.80432931e-01 -7.47690380e-01
6.16922192e-02 -5.12720580e-05 4.25313234e-01 5.20466566e-01
-1.29670396e-01 7.58914411e-01 9.46179152e-01 -5.37140667e-01
-1.57049835e+00 -1.16792440e+00 -7.14296043e-01 3.73749658e-02
-4.69055951e-01 -4.07801688e-01 -5.74371278e-01 -3.89294386e-01
1.74074441e-01 1.30763280e+00 -1.01235166e-01 -3.32039684e-01
-2.93223321e-01 -1.35902643e+00 6.49573982e-01 1.17159688e+00
-2.16991305e-02 -6.48226619e-01 -1.81257769e-01 5.51389337e-01
6.77182555e-01 -1.08040464e+00 3.57974529e-01 1.06452036e+00
-9.04433668e-01 -8.06251228e-01 -2.67141879e-01 -1.86049595e-01
4.66832787e-01 -8.16583335e-01 9.96805310e-01 -6.87727153e-01
-4.46067959e-01 -8.58769864e-02 3.07657957e-01 -7.11250126e-01
-9.09367859e-01 -5.63676357e-02 4.99662042e-01 -5.15120387e-01
8.84865522e-01 -6.50172532e-01 -4.17671114e-01 -1.62681267e-01
-7.01231658e-01 -3.87217253e-01 5.28966606e-01 7.81477511e-01
5.96793592e-01 7.30154097e-01 2.15728536e-01 -5.29984355e-01
9.71907854e-01 -1.49979562e-01 -8.21132839e-01 -4.35296208e-01
-7.38088906e-01 6.09513044e-01 5.47244310e-01 1.96062699e-02
-9.43639874e-01 1.04073748e-01 -7.39508748e-01 -4.91414934e-01
-2.10044742e-01 2.25694209e-01 -1.90790042e-01 -1.92187905e-01
1.03365517e+00 -2.29513049e-01 1.84826568e-01 -4.15107280e-01
5.45325726e-02 2.95619607e-01 4.42515701e-01 -6.17620349e-01
7.66021729e-01 5.40713131e-01 6.98191047e-01 -7.24416852e-01
-2.40594462e-01 -3.96126151e-01 -7.91374743e-01 2.35273629e-01
1.07990396e+00 -7.62298048e-01 -8.99578929e-01 3.21403474e-01
-1.08356953e+00 -1.99073434e-01 -6.98475003e-01 8.00335467e-01
-6.62259340e-01 4.24092382e-01 -9.24222112e-01 -8.67809415e-01
-6.20258570e-01 -1.28393078e+00 1.05072343e+00 2.41100099e-02
-4.65306520e-01 -9.63469028e-01 -1.21256329e-01 -2.83872873e-01
7.30573952e-01 2.66259342e-01 1.33905053e+00 -1.12242532e+00
-4.01859283e-01 -5.98471045e-01 -1.03125967e-01 7.67490268e-01
-2.49942988e-01 -3.11791182e-01 -1.09303176e+00 -1.44527838e-01
7.61216104e-01 8.50509554e-02 8.76220524e-01 6.45983815e-01
1.12650061e+00 4.88361120e-01 -6.51947320e-01 4.32130098e-01
1.52589321e+00 2.84237713e-01 8.06241512e-01 2.18071714e-01
7.15881526e-01 2.30851352e-01 8.63377657e-03 1.00223586e-01
-3.64417762e-01 2.57388026e-01 2.01542765e-01 2.29488149e-01
1.93532765e-01 1.02632023e-01 5.45405559e-02 3.24376911e-01
6.97745159e-02 3.38779837e-01 -1.40964103e+00 2.25861773e-01
-1.61885691e+00 -9.81001914e-01 -7.21145749e-01 2.24207115e+00
5.54765046e-01 7.60030150e-01 -4.13582861e-01 3.18525940e-01
1.29525974e-01 -3.00422996e-01 -4.60296541e-01 -6.05936587e-01
2.33068973e-01 5.41299284e-01 9.11526799e-01 3.70921135e-01
-1.09477437e+00 2.37966359e-01 6.35367823e+00 1.09733105e+00
-8.15353930e-01 2.33909249e-01 6.38578892e-01 -3.94051462e-01
-5.40634617e-02 -1.55317470e-01 -1.06844234e+00 3.46670121e-01
1.59048605e+00 -2.15666085e-01 9.40805152e-02 9.45012391e-01
1.51430294e-01 -3.61374438e-01 -1.52313292e+00 1.14516580e+00
-2.74558634e-01 -1.29625416e+00 -3.65140110e-01 8.13424960e-02
6.54905260e-01 3.24169487e-01 -7.45957866e-02 5.49679458e-01
3.47682476e-01 -1.41806936e+00 4.38003004e-01 7.64831901e-01
2.31150270e-01 -1.26543212e+00 1.07696176e+00 2.66020596e-01
-6.38220370e-01 -1.63614631e-01 -6.90516770e-01 2.83417881e-01
6.53106093e-01 1.15254653e+00 -1.04495764e+00 5.35841525e-01
6.36600256e-01 6.34949356e-02 -2.01696172e-01 1.37254298e+00
8.01680535e-02 5.86696506e-01 -8.43012154e-01 8.50920454e-02
3.89338076e-01 -2.16044113e-01 6.24319971e-01 8.62971902e-01
5.19740760e-01 -4.96598572e-01 -2.34840736e-01 1.12426543e+00
1.36595383e-01 -3.10099661e-01 -4.23880249e-01 -1.69837683e-01
-6.39678836e-02 1.13281739e+00 -9.74680781e-01 5.40141426e-02
-2.32380986e-01 6.52360022e-01 -2.44986147e-01 1.01441607e-01
-7.42097557e-01 -3.15935969e-01 8.38131964e-01 3.96511853e-01
2.63846636e-01 -6.19455576e-01 -6.35572314e-01 -4.67538208e-01
-5.62899448e-02 -5.01450837e-01 -1.15031809e-01 -5.34919798e-01
-1.46179473e+00 6.98521197e-01 2.61890471e-01 -7.28443861e-01
-4.09484446e-01 -1.01510549e+00 -6.64551377e-01 1.55324769e+00
-1.10466456e+00 -9.31768954e-01 1.87226251e-01 2.89984117e-03
2.01297894e-01 -2.91417927e-01 9.27716732e-01 1.97769523e-01
-1.85231447e-01 -4.95128520e-02 7.15138316e-01 -7.40525573e-02
4.71132964e-01 -1.48695469e+00 9.73123372e-01 5.71403742e-01
-5.21232076e-02 4.48992997e-01 1.33268940e+00 -8.19286823e-01
-1.42829001e+00 -1.04385436e+00 3.71016622e-01 -6.65246248e-01
1.10085642e+00 -5.26379824e-01 -1.05235565e+00 4.96409088e-01
1.74157098e-01 -9.10071563e-03 5.87962985e-01 2.73815155e-01
2.27959499e-01 5.02993949e-02 -1.34378982e+00 2.31710836e-01
5.09999037e-01 -5.27842760e-01 -6.03193939e-01 4.25388604e-01
5.66479206e-01 -5.74549198e-01 -1.27693343e+00 5.85906863e-01
1.10045925e-01 -6.14226282e-01 8.57651591e-01 -4.98137802e-01
1.91447824e-01 -3.28189701e-01 -5.74566536e-02 -1.48410833e+00
-3.75878811e-01 -4.10263687e-01 -3.56228799e-01 9.51737523e-01
6.11935139e-01 -4.50730562e-01 1.24202132e+00 1.16864073e+00
-3.84912223e-01 -7.80655801e-01 -1.29637086e+00 -1.06589246e+00
5.79955339e-01 -9.60171402e-01 5.22235990e-01 3.25929314e-01
-1.38930678e-01 2.48653188e-01 2.04331219e-01 3.63452196e-01
1.00392199e+00 -5.40418088e-01 1.67458788e-01 -1.59437668e+00
-4.54187244e-01 -6.28522038e-01 -9.09105957e-01 -3.62692237e-01
2.23955382e-02 -8.43011737e-01 5.24361655e-02 -1.80897927e+00
-1.60695121e-01 -5.36103368e-01 -1.24604769e-01 -8.46757460e-03
3.40749443e-01 -1.50364906e-01 -1.91813558e-01 -8.94088745e-02
6.06590919e-02 7.00892150e-01 4.92344916e-01 -2.78336465e-01
6.54097646e-02 3.43162596e-01 -3.15186441e-01 8.12197804e-01
8.60848725e-01 -6.87994063e-01 -4.58768398e-01 -8.51507038e-02
5.05632281e-01 -5.95837645e-02 2.79890954e-01 -1.79929388e+00
5.75114131e-01 5.79630435e-01 1.10753465e+00 -6.81798220e-01
4.50145781e-01 -1.15208602e+00 4.37032104e-01 3.53138357e-01
-4.58831228e-02 1.34905875e-02 7.88795292e-01 5.72952986e-01
-2.72180974e-01 -7.65342593e-01 6.57981217e-01 -4.45255637e-02
-4.51156616e-01 3.92463326e-01 -4.77165043e-01 -5.03399551e-01
9.21244144e-01 1.14088967e-01 1.01568671e-02 4.23902310e-02
-1.13612390e+00 8.74196887e-02 2.49146163e-01 1.67290330e-01
3.59049767e-01 -1.31749976e+00 -4.47132051e-01 2.48758793e-01
-1.57058969e-01 2.50624806e-01 4.12753552e-01 8.85711372e-01
-7.50881076e-01 4.81266379e-01 -1.09868392e-01 -8.11209917e-01
-9.99561310e-01 5.93722105e-01 6.69278443e-01 -5.10006666e-01
-8.66390944e-01 8.36225212e-01 -1.87943205e-01 -1.66895732e-01
5.12487767e-03 -7.37303674e-01 2.03205019e-01 -7.03735724e-02
6.89640582e-01 1.64991215e-01 7.79757202e-01 -1.58994317e-01
-7.25965276e-02 -1.52023524e-01 -1.20742165e-01 -2.49045193e-01
1.55157673e+00 3.88048947e-01 -7.80387968e-02 5.01125932e-01
1.09356415e+00 -4.64079231e-01 -1.25401580e+00 1.15354732e-01
4.15815115e-01 1.30106062e-01 5.27669430e-01 -4.33511019e-01
-7.39298761e-01 1.28072631e+00 8.79994452e-01 1.44960329e-01
7.14585543e-01 -1.14552647e-01 5.52478671e-01 6.24778032e-01
4.08704937e-01 -1.32244444e+00 -5.65103412e-01 5.25729477e-01
6.94040895e-01 -7.43378162e-01 1.99256420e-01 7.48412982e-02
-6.23674929e-01 1.39589727e+00 2.62657106e-01 -7.33395442e-02
1.00322783e+00 9.06597853e-01 -5.20574331e-01 -4.90519375e-01
-6.26501799e-01 3.86804521e-01 1.13433404e-02 7.70183682e-01
2.68141955e-01 6.34426847e-02 1.03600889e-01 6.89594746e-01
-1.18753560e-01 -1.88329905e-01 4.57674742e-01 9.18860674e-01
-4.51717108e-01 -1.22287941e+00 -5.94602227e-01 9.88617241e-01
-6.34279847e-01 -3.15984190e-01 4.31577057e-01 5.24143398e-01
-3.00101191e-01 4.01029646e-01 3.09262186e-01 -1.52166942e-02
5.19262195e-01 6.99154556e-01 6.38356328e-01 -4.77007449e-01
-6.15712464e-01 5.50914817e-02 3.21763754e-01 -5.72133660e-01
1.97580427e-01 -9.46618736e-01 -1.44192040e+00 -2.43796557e-01
-7.52837881e-02 1.74985766e-01 1.31006336e+00 8.70430529e-01
-8.15573335e-02 9.86718655e-01 -1.13991715e-01 -9.91837561e-01
-1.00401473e+00 -1.04613841e+00 -8.11165571e-01 -1.10869966e-01
1.06096677e-01 -5.61694443e-01 -2.00289950e-01 -4.96821344e-01] | [7.225008964538574, 3.771913528442383] |
347395b5-c01b-402d-bb38-ac8217d4eaff | identification-of-binary-neutron-star-mergers | 2207.00591 | null | https://arxiv.org/abs/2207.00591v1 | https://arxiv.org/pdf/2207.00591v1.pdf | Identification of Binary Neutron Star Mergers in Gravitational-Wave Data Using YOLO One-Shot Object Detection | We demonstrate the application of the YOLOv5 model, a general purpose convolution-based single-shot object detection model, in the task of detecting binary neutron star (BNS) coalescence events from gravitational-wave data of current generation interferometer detectors. We also present a thorough explanation of the synthetic data generation and preparation tasks based on approximant waveform models used for the model training, validation and testing steps. Using this approach, we achieve mean average precision ($\text{mAP}_{[0.50]}$) values of 0.945 for a single class validation dataset and as high as 0.978 for test datasets. Moreover, the trained model is successful in identifying the GW170817 event in the LIGO H1 detector data. The identification of this event is also possible for the LIGO L1 detector data with an additional pre-processing step, without the need of removing the large glitch in the final stages of the inspiral. The detection of the GW190425 event is less successful, which attests to performance degradation with the signal-to-noise ratio. Our study indicates that the YOLOv5 model is an interesting approach for first-stage detection alarm pipelines and, when integrated in more complex pipelines, for real-time inference of physical source parameters. | ['José A. Font', 'Gonçalo Gonçalves', 'Constança Providência', 'Antonio Onofre', 'Márcio Ferreira', 'Felipe F. Freitas', 'João Aveiro'] | 2022-07-01 | null | null | null | null | ['one-shot-object-detection'] | ['computer-vision'] | [-2.42822379e-01 1.11067288e-01 6.47306919e-01 -2.61629000e-02
-7.43046522e-01 -3.64312202e-01 1.05714238e+00 2.62657762e-01
-7.54868150e-01 4.38768446e-01 -4.28688258e-01 -5.02874553e-01
-9.27942544e-02 -6.89287126e-01 -3.57909918e-01 -9.95588899e-01
-4.10233065e-02 9.40516949e-01 7.32559025e-01 1.10243849e-01
1.47801206e-01 7.19516754e-01 -1.14148772e+00 -1.16167637e-02
1.38038918e-01 1.05606699e+00 1.72670648e-01 9.21704829e-01
3.70367825e-01 6.87448025e-01 -9.34502363e-01 -2.50191063e-01
2.48409688e-01 -6.85804188e-01 -3.09645116e-01 -6.08021319e-01
-1.53824791e-01 3.06694321e-02 -5.42377114e-01 1.00657618e+00
8.20875287e-01 -1.65757239e-01 5.54620802e-01 -8.99259090e-01
3.60193402e-01 5.36820650e-01 -1.65190294e-01 8.47642243e-01
-1.36274487e-01 8.09163809e-01 5.30827641e-01 -9.58765447e-01
4.61320102e-01 6.10884905e-01 6.10816300e-01 7.85678029e-02
-9.95858908e-01 -6.24896228e-01 -1.15818059e+00 2.93259829e-01
-1.26253915e+00 -3.64223093e-01 2.98664808e-01 -6.87088311e-01
1.43383121e+00 1.19888961e-01 5.46897829e-01 8.08929086e-01
-2.83661410e-02 1.36317015e-01 8.29930663e-01 -5.73927045e-01
4.54042137e-01 -4.42107022e-02 6.68687299e-02 4.23893690e-01
6.55166090e-01 7.38422692e-01 -5.74778736e-01 -2.80605108e-01
6.70485318e-01 -6.88792765e-01 -4.33309861e-02 2.58634120e-01
-1.15852702e+00 9.95779037e-01 5.42523675e-02 6.00625217e-01
-4.93151248e-01 1.08439311e-01 4.98796880e-01 1.31753400e-01
4.19664979e-01 7.09008873e-01 -3.08977306e-01 -2.67384231e-01
-1.28340960e+00 5.08877575e-01 6.61234081e-01 6.65532172e-01
2.71996558e-01 5.34594893e-01 -3.19372952e-01 1.55457959e-01
2.84137577e-01 7.20126867e-01 3.22273552e-01 -1.79458395e-01
6.11232221e-02 1.84556961e-01 8.74285474e-02 -2.07031474e-01
-8.56999457e-01 -6.99849069e-01 -4.30717438e-01 6.93904400e-01
6.59913659e-01 -3.94418269e-01 -8.79534423e-01 1.19483531e+00
3.38644981e-01 3.23040187e-01 3.03653181e-01 8.55448544e-01
9.75786150e-01 4.84945655e-01 7.20357671e-02 -1.37385234e-01
1.61938751e+00 -3.61807495e-01 -2.35383973e-01 -5.07793963e-01
4.31561261e-01 -9.45507646e-01 1.25832483e-01 4.16233927e-01
-1.01439905e+00 -5.01053035e-01 -1.32591903e+00 4.09724116e-01
1.17277309e-01 3.25346738e-01 6.93584800e-01 6.67987585e-01
-4.74377841e-01 8.14917922e-01 -1.09105730e+00 -2.31410727e-01
8.09651762e-02 1.01955742e-01 1.42553762e-01 4.93766248e-01
-9.65432763e-01 9.36183989e-01 7.96966791e-01 -7.62374997e-02
-1.03478718e+00 -6.79331601e-01 -6.47419393e-01 2.12057114e-01
3.06606572e-02 -3.80991668e-01 1.43001485e+00 -9.50668529e-02
-1.09125674e+00 9.69021142e-01 5.89820623e-01 -1.30711687e+00
5.15838861e-01 2.24988118e-01 -7.05294371e-01 4.22448069e-01
-1.36159614e-01 -1.42852189e-02 7.46819496e-01 -3.94352347e-01
-6.89716756e-01 3.86908874e-02 -5.41832268e-01 -4.75060582e-01
6.77168429e-01 7.70174146e-01 -1.39849156e-01 -5.56823611e-01
9.53066498e-02 -5.06273806e-01 -1.64734691e-01 -6.42025590e-01
-3.02765727e-01 -9.25089121e-02 5.79244733e-01 -7.98100829e-01
7.21251845e-01 -2.12871885e+00 -2.79184014e-01 1.76956207e-01
5.55985905e-02 7.42370129e-01 5.12434363e-01 3.84452790e-01
-3.56163859e-01 -3.57281923e-01 -4.01732683e-01 -1.42241806e-01
-1.20535143e-01 -3.81012291e-01 -1.51386410e-01 5.83761573e-01
2.98851877e-01 7.95021832e-01 -7.13474274e-01 7.68688172e-02
4.14363891e-01 3.35091412e-01 -1.82595998e-01 3.57480973e-01
-2.48292148e-01 7.41477013e-01 -2.44635511e-02 3.26952070e-01
7.17676342e-01 1.09338015e-01 -2.23234788e-01 -1.74798220e-01
-6.05494916e-01 5.52107573e-01 -1.04248691e+00 1.42683303e+00
7.19849914e-02 9.09140468e-01 2.40030736e-01 -1.05599391e+00
1.17459309e+00 5.65466642e-01 9.45933461e-02 -8.16290557e-01
4.56815749e-01 4.08068419e-01 4.60625410e-01 -6.25465393e-01
3.99289221e-01 -6.47950590e-01 -5.86839281e-02 3.07797730e-01
4.79398370e-01 -3.92417341e-01 4.18455601e-01 2.12159231e-01
1.25395894e+00 4.37279828e-02 1.57090023e-01 -4.45108682e-01
1.67174399e-01 1.07584484e-01 4.16985661e-01 9.75058734e-01
9.77887213e-02 8.66249859e-01 5.26711464e-01 -5.23344219e-01
-1.42074001e+00 -8.38898182e-01 -3.72013867e-01 9.46072862e-03
-3.37086737e-01 -4.77470547e-01 -4.90042984e-01 -2.28272393e-01
-3.21652085e-01 1.16754937e+00 -1.80266082e-01 -4.00021046e-01
-4.93193775e-01 -1.60934794e+00 8.78907621e-01 2.63048798e-01
1.55026451e-01 -1.42394066e+00 -1.28713298e+00 2.90626138e-01
2.34874666e-01 -1.05418074e+00 5.82038164e-01 4.70572531e-01
-4.64178979e-01 -1.26365924e+00 -1.27263740e-01 -1.41828671e-01
1.58280775e-01 -3.83965015e-01 1.04723513e+00 -1.33276521e-03
-9.17336106e-01 -1.31056890e-01 -3.55438769e-01 -5.94599247e-01
-7.82635629e-01 -6.51708364e-01 2.00729027e-01 -2.32974663e-01
7.09759831e-01 -2.70960599e-01 -3.34681213e-01 9.84712616e-02
-6.84103608e-01 -1.04646377e-01 4.14948851e-01 7.26473153e-01
7.52059519e-02 2.15138525e-01 4.45078999e-01 -1.50411099e-01
5.89516526e-03 -5.15594244e-01 -1.40989780e+00 -7.48231336e-02
-3.06762815e-01 1.26643047e-01 4.32408959e-01 -1.70063689e-01
-9.06699479e-01 -8.87981877e-02 -4.89268869e-01 -4.19587076e-01
-2.36434639e-01 1.84612602e-01 1.42623365e-01 -5.85934110e-02
8.77225399e-01 1.09139979e-01 -1.01615585e-01 -6.47989631e-01
-1.53144807e-01 3.05786997e-01 1.01652431e+00 -1.14490710e-01
1.08786821e+00 3.24455708e-01 5.09109572e-02 -1.12942171e+00
-4.98759508e-01 -6.19168639e-01 -2.65512943e-01 -7.15766847e-02
8.66450787e-01 -8.73076737e-01 -8.06757033e-01 6.53892756e-01
-1.27979422e+00 -1.15094148e-01 -4.94811147e-01 1.10351932e+00
-1.69758707e-01 3.23354661e-01 -5.06801248e-01 -1.06119299e+00
-5.17049134e-01 -1.03512883e+00 6.61871374e-01 4.87731695e-01
-3.81610021e-02 -5.31846464e-01 1.71883777e-01 1.18745141e-01
5.21231115e-01 2.37444371e-01 4.79748011e-01 -1.18905318e+00
-3.72852355e-01 -5.15742779e-01 -2.91722417e-01 1.38675585e-01
-4.88182098e-01 1.27367049e-01 -1.07628536e+00 -3.30924869e-01
8.31276357e-01 -1.10769346e-01 1.15071845e+00 5.16524434e-01
5.10384679e-01 3.09066832e-01 -1.94282666e-01 5.65958023e-01
1.25008893e+00 4.06588137e-01 5.78289390e-01 2.16613486e-01
1.76966384e-01 1.91273034e-01 5.76448381e-01 7.22720742e-01
-4.44432944e-01 7.44211853e-01 4.28833783e-01 -3.53871398e-02
-2.42006525e-01 3.06378514e-01 2.18701333e-01 9.33998451e-02
-7.60015771e-02 2.35512620e-03 -1.06306660e+00 4.11078662e-01
-1.51017809e+00 -1.29053104e+00 -8.41458738e-01 2.60303950e+00
2.11940840e-01 7.31647611e-01 9.26976576e-02 1.94906831e-01
6.36075974e-01 -1.05035052e-01 -3.42933051e-02 -1.23173654e-01
-2.01393664e-01 9.13474500e-01 8.36938620e-01 5.17904043e-01
-1.06560981e+00 3.93821090e-01 5.63590574e+00 7.89745092e-01
-1.02997673e+00 2.66839117e-01 1.81829140e-01 -4.10712421e-01
1.80273056e-01 2.94900566e-01 -8.96690071e-01 7.54935026e-01
1.31639433e+00 -1.58891864e-02 1.37417853e-01 6.71990931e-01
1.05951518e-01 -6.24497056e-01 -7.99566567e-01 1.07601357e+00
1.27123706e-02 -1.54563713e+00 -8.74487996e-01 -2.23530993e-01
5.61521687e-02 4.76267755e-01 -7.84784019e-01 4.34326351e-01
4.06417847e-02 -6.92263126e-01 1.06710625e+00 3.74156088e-01
6.22025967e-01 -9.04565394e-01 9.51296866e-01 4.31362629e-01
-8.64583910e-01 1.90319002e-01 -5.10777831e-01 -3.50227281e-02
6.84829652e-01 1.04543829e+00 -1.42467761e+00 7.89483964e-01
7.32462525e-01 -2.29350537e-01 -6.25601709e-01 1.46072221e+00
-4.03870165e-01 9.58817124e-01 -7.99780965e-01 7.86607787e-02
1.17634252e-01 -6.51401728e-02 1.06411719e+00 1.45160186e+00
4.34666336e-01 1.42707136e-02 -2.67247260e-01 1.06725729e+00
3.02845061e-01 -5.62407851e-01 -4.22377735e-01 -3.43627296e-02
2.47900873e-01 1.35808194e+00 -1.06834626e+00 -4.07088906e-01
-1.68432623e-01 4.75586236e-01 -1.25736058e-01 -6.40185326e-02
-9.30465519e-01 -6.04427993e-01 3.17631334e-01 1.82321653e-01
6.38543367e-01 -3.43764722e-01 -2.68211275e-01 -8.73301923e-01
-1.37616932e-01 -3.56141269e-01 3.38801742e-01 -5.52072287e-01
-7.83065975e-01 6.81511164e-01 9.10327584e-02 -1.03376770e+00
-4.91855651e-01 -6.75761521e-01 -1.29803932e+00 9.89861250e-01
-8.52618873e-01 -7.90932000e-01 -2.31417939e-02 -1.22539304e-01
2.22237393e-01 -6.00456715e-01 6.02359831e-01 3.24299693e-01
-4.54418749e-01 -3.59563306e-02 4.26275842e-02 1.77344322e-01
2.64837444e-01 -1.04239500e+00 9.12275434e-01 1.48896015e+00
9.13027078e-02 9.92271025e-03 1.44845843e+00 -9.28893805e-01
-1.02544188e+00 -7.25262403e-01 9.56326962e-01 -2.36612260e-01
7.73849249e-01 -5.66650689e-01 -8.21261346e-01 4.08634067e-01
2.74683625e-01 1.00691237e-01 2.62993515e-01 -3.18773568e-01
5.81019325e-03 4.24456775e-01 -1.35269785e+00 -1.07145555e-01
3.90092283e-01 -3.06539714e-01 -7.56134510e-01 4.86737460e-01
3.26116443e-01 -2.08125263e-01 -3.64344627e-01 6.62861109e-01
8.65127072e-02 -1.22780275e+00 8.57093036e-01 -3.38507682e-01
-1.28212988e-01 -5.72180748e-01 3.49406600e-01 -9.75244462e-01
-1.61530733e-01 -6.24593616e-01 6.78227767e-02 1.06219709e+00
3.24912280e-01 -3.82700920e-01 6.36164963e-01 9.36984569e-02
-2.13414893e-01 1.14489853e-01 -1.48071361e+00 -8.31382692e-01
-2.84814209e-01 -7.01792121e-01 8.30645561e-02 6.49718761e-01
-1.51966184e-01 2.38575652e-01 -3.17555934e-01 6.17945552e-01
8.88878167e-01 -1.29547166e-02 4.73740548e-01 -1.25081766e+00
-9.42717910e-01 -3.20433944e-01 -5.56904495e-01 -2.00347945e-01
-6.87245190e-01 -8.42342496e-01 6.21780194e-02 -1.03060305e+00
-2.39023313e-01 -4.16618139e-01 -2.62406040e-02 3.56776007e-02
1.09497719e-01 4.63968277e-01 -8.64569396e-02 -6.85647056e-02
-1.93694904e-01 2.88514823e-01 -3.30487685e-03 1.81207001e-01
1.92471743e-01 2.11533308e-01 2.17499003e-01 7.60413170e-01
8.31316054e-01 -9.08906400e-01 3.07681084e-01 -3.50905061e-02
3.14496428e-01 3.13576937e-01 7.55387783e-01 -1.53721690e+00
3.00444365e-01 4.06141758e-01 4.39156771e-01 -9.25582469e-01
3.14849555e-01 -2.60064274e-01 5.92249751e-01 9.85503554e-01
5.11894703e-01 -1.32197186e-01 3.67285132e-01 6.54150546e-02
-1.07326312e-02 -1.04687369e+00 1.21004808e+00 -2.12952271e-01
-3.05835038e-01 -7.06819743e-02 -6.26740575e-01 2.81281397e-02
1.00207567e+00 3.40371728e-01 -2.09046125e-01 2.74457455e-01
-7.99220443e-01 2.86973044e-02 2.79209912e-01 1.09430037e-01
1.28747359e-01 -8.35011423e-01 -1.09555197e+00 4.81534719e-01
-1.09775513e-02 4.18570712e-02 2.33165354e-01 9.52607930e-01
-9.72764015e-01 4.45477962e-01 -1.54481620e-01 -5.28681993e-01
-1.11360073e+00 5.97467065e-01 5.33071101e-01 -2.82371163e-01
-7.19148576e-01 1.08495045e+00 -1.60855889e-01 -2.24201102e-02
-1.12942584e-01 -9.15160701e-02 5.15996851e-02 1.36518747e-01
9.46235836e-01 3.72704983e-01 5.60750544e-01 -4.18952018e-01
-3.95546108e-01 -1.65825188e-01 1.55819148e-01 -2.63751358e-01
1.37242568e+00 5.96544087e-01 7.38818124e-02 3.58350314e-02
4.78258163e-01 8.48665237e-02 -9.21048522e-01 -2.47804657e-01
4.00232404e-01 -7.38548785e-02 1.79166645e-02 -6.67540014e-01
-7.72421241e-01 7.64934242e-01 7.57943511e-01 4.17491257e-01
5.65344036e-01 5.11113763e-01 1.55013129e-01 5.18468134e-02
2.79250860e-01 -8.54989588e-01 -4.22270209e-01 5.04736245e-01
7.79774725e-01 -1.00084698e+00 6.32342622e-02 1.07344978e-01
-3.12517881e-01 1.40805137e+00 3.07875633e-01 -1.42846003e-01
3.86618197e-01 5.85529804e-01 -2.90058404e-01 -7.75392115e-01
-5.59709311e-01 -2.09016651e-01 -7.05799088e-02 2.53391534e-01
-1.51970303e-02 9.48988944e-02 -5.80391288e-01 5.35009265e-01
-2.47059137e-01 -4.52990741e-01 6.38171017e-01 8.20654213e-01
-7.47678518e-01 -9.45618510e-01 -8.00683916e-01 2.80760139e-01
-5.32950282e-01 -3.42777580e-01 -5.87010495e-02 6.40287280e-01
1.30847260e-01 8.54463458e-01 3.09259623e-01 -3.96586545e-02
1.10496044e-01 4.63695228e-01 4.14267957e-01 -6.51289582e-01
-9.50150549e-01 1.96608663e-01 4.13111746e-01 -9.34582055e-02
-7.55746663e-02 -8.18097770e-01 -1.34576547e+00 4.54712771e-02
-4.05268103e-01 4.75539744e-01 1.08015692e+00 1.19201982e+00
-2.03780398e-01 6.22003973e-01 2.75287151e-01 -1.17800772e+00
-4.66819674e-01 -1.41808271e+00 -4.10810858e-01 -1.28055960e-02
9.36487615e-02 -6.00980759e-01 -6.87246442e-01 -2.39150032e-01] | [7.547070503234863, 3.137256622314453] |
cccc332e-64d5-41db-ac14-7ef74f53e4ba | fetsmcs-feature-based-ets-model-component | 2206.12882 | null | https://arxiv.org/abs/2206.12882v1 | https://arxiv.org/pdf/2206.12882v1.pdf | fETSmcs: Feature-based ETS model component selection | The well-developed ETS (ExponenTial Smoothing or Error, Trend, Seasonality) method incorporating a family of exponential smoothing models in state space representation has been widely used for automatic forecasting. The existing ETS method uses information criteria for model selection by choosing an optimal model with the smallest information criterion among all models fitted to a given time series. The ETS method under such a model selection scheme suffers from computational complexity when applied to large-scale time series data. To tackle this issue, we propose an efficient approach for ETS model selection by training classifiers on simulated data to predict appropriate model component forms for a given time series. We provide a simulation study to show the model selection ability of the proposed approach on simulated data. We evaluate our approach on the widely used forecasting competition data set M4, in terms of both point forecasts and prediction intervals. To demonstrate the practical value of our method, we showcase the performance improvements from our approach on a monthly hospital data set. | ['Suling Jia', 'Qiang Wang', 'Xixi Li', 'Lingzhi Qi'] | 2022-06-26 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [ 1.27681091e-01 -9.72468928e-02 -5.28852195e-02 -3.32628399e-01
-7.05075264e-01 -2.42585495e-01 5.92775941e-01 3.97581428e-01
-4.18373078e-01 7.64620125e-01 4.80906181e-02 -6.70420706e-01
-4.88391012e-01 -5.36114931e-01 -7.78012797e-02 -6.51690304e-01
-5.03376424e-01 4.46592957e-01 1.44299731e-01 -4.25227433e-01
4.01009232e-01 6.27166569e-01 -1.30579484e+00 -1.03198431e-01
1.01798153e+00 1.02641881e+00 1.09329678e-01 5.56480289e-01
-1.88830551e-02 3.85216355e-01 -5.91956437e-01 7.66437873e-02
2.95300126e-01 -4.17264223e-01 -4.50012296e-01 -7.36189401e-03
-4.87244457e-01 2.06968069e-01 3.38929921e-01 5.31753838e-01
4.46692616e-01 2.18017563e-01 9.26458597e-01 -1.09826303e+00
3.75732183e-01 3.68162364e-01 -7.82788917e-02 4.46608782e-01
4.11476195e-02 -1.88661143e-01 3.45703214e-01 -5.84528863e-01
3.43487173e-01 1.02512944e+00 1.04317796e+00 1.41447365e-01
-1.58888531e+00 -5.35020232e-01 -1.28265664e-01 -1.96664333e-01
-1.42028201e+00 -2.25673512e-01 7.91835248e-01 -6.27289116e-01
9.44109440e-01 3.97141069e-01 5.87942898e-01 4.48200941e-01
8.57373536e-01 3.11667502e-01 1.41371644e+00 -5.13155997e-01
3.59524697e-01 3.21399480e-01 4.45791572e-01 2.38662049e-01
-1.13843463e-01 4.32725638e-01 1.55267134e-01 -7.00022697e-01
6.25630736e-01 9.32518020e-02 2.77123228e-02 1.95136115e-01
-1.11717165e+00 8.16174328e-01 1.29615786e-02 5.59084237e-01
-8.72191429e-01 -9.12911370e-02 5.29764116e-01 5.64968765e-01
8.86523306e-01 2.45169520e-01 -7.24290311e-01 -5.28366975e-02
-1.17532456e+00 2.26042345e-01 9.56019580e-01 6.39067888e-01
3.12813997e-01 3.34782213e-01 -2.98031479e-01 6.20472491e-01
2.16499045e-01 4.90051001e-01 6.92202091e-01 -5.02465427e-01
8.29573572e-02 3.07670504e-01 4.49798107e-01 -8.92874897e-01
-8.91790390e-01 -4.75639313e-01 -1.03513443e+00 8.22869018e-02
3.23085696e-01 -5.05562544e-01 -8.43421638e-01 1.53228474e+00
3.17066759e-01 5.24520338e-01 3.23029459e-01 1.81879953e-01
2.71738321e-01 9.86383438e-01 2.26535752e-01 -9.48821306e-01
1.13395929e+00 -4.11945313e-01 -9.32779729e-01 4.18989420e-01
8.60622883e-01 -8.98994982e-01 2.62722731e-01 5.10188997e-01
-7.24568665e-01 -5.24032712e-01 -5.23797035e-01 7.91265666e-01
-2.43673503e-01 3.40887219e-01 3.24340105e-01 6.33870244e-01
-9.97110546e-01 7.90567398e-01 -1.08846688e+00 -4.40645456e-01
-3.81576926e-01 4.89528328e-01 1.46077825e-02 7.76664734e-01
-1.40898752e+00 1.01975310e+00 5.26155889e-01 6.45572767e-02
-4.20250922e-01 -5.66591740e-01 -6.53579056e-01 2.19536856e-01
-1.16739608e-01 -4.01215404e-01 1.45298707e+00 -9.59972203e-01
-1.60878503e+00 1.62880391e-01 -3.81585985e-01 -8.70189011e-01
5.25460064e-01 2.24686548e-01 -1.00067818e+00 -1.96512893e-01
-4.25432682e-01 -2.05274582e-01 8.40911925e-01 -8.30490589e-01
-5.40788174e-01 -3.99134904e-02 -6.03230298e-01 -3.06404859e-01
-2.22723763e-02 3.16735774e-01 4.56627458e-01 -8.92086327e-01
1.64005280e-01 -1.03100288e+00 -7.82893717e-01 -8.35602701e-01
-8.57053623e-02 -4.32397336e-01 3.91643763e-01 -8.85474443e-01
1.95471132e+00 -1.80826247e+00 -2.89943010e-01 8.13451707e-01
-4.52070743e-01 9.93251577e-02 2.98398465e-01 1.05160558e+00
-4.71413106e-01 -4.59638052e-02 -3.51774693e-01 -2.41891861e-01
-2.33556792e-01 1.34192020e-01 -4.73554045e-01 3.66245419e-01
7.16784522e-02 2.85116404e-01 -4.53492880e-01 -4.82502937e-01
2.49727532e-01 3.72249037e-01 -2.20956162e-01 2.68457592e-01
1.30310282e-01 5.38566887e-01 -5.69216788e-01 1.92604020e-01
4.85896111e-01 -1.63736969e-01 1.32148698e-01 1.59776166e-01
-5.44430435e-01 1.57271102e-02 -1.39377940e+00 9.22975600e-01
-6.59769952e-01 2.69157797e-01 -3.37850213e-01 -1.29159915e+00
1.48914683e+00 9.29857314e-01 8.70133877e-01 -8.87109637e-02
2.19690681e-01 5.72706938e-01 1.15904808e-01 -5.57333589e-01
2.81909913e-01 -4.92275834e-01 -9.66894149e-04 1.94803312e-01
-2.46521235e-01 -7.43453717e-03 8.82851854e-02 -2.48327017e-01
8.43068898e-01 -1.18309945e-01 7.39798963e-01 -7.70877659e-01
6.72182739e-01 2.80983418e-01 5.18600166e-01 4.45925713e-01
-2.10465733e-02 1.14174522e-01 4.59911257e-01 -5.57782710e-01
-1.07441342e+00 -2.14640334e-01 -5.24484634e-01 6.07117712e-01
-4.72922087e-01 -2.47378603e-01 -4.73454148e-01 -2.81462163e-01
-1.17296726e-03 9.87473667e-01 -7.45832860e-01 1.76506117e-01
-5.73610544e-01 -9.09772336e-01 2.33125404e-01 2.00735867e-01
-8.21682587e-02 -1.00709426e+00 -9.08662319e-01 9.07712460e-01
1.19422168e-01 -4.80275482e-01 -2.44595453e-01 3.15323144e-01
-1.22966540e+00 -7.54597902e-01 -8.15227807e-01 -5.61519623e-01
5.58034658e-01 -2.75903553e-01 1.02864838e+00 1.86475273e-02
1.72854587e-01 2.59903193e-01 -1.74652457e-01 -6.39170170e-01
-9.34855521e-01 -2.34814286e-02 8.91917720e-02 1.59494236e-01
1.07225321e-01 -2.64443368e-01 -4.87561226e-01 4.91530895e-01
-8.84430647e-01 5.44013456e-04 2.16891050e-01 1.05280316e+00
3.96549106e-01 1.79765508e-01 1.25862157e+00 -8.89140785e-01
9.14749622e-01 -7.73460090e-01 -1.05196822e+00 3.94042879e-01
-1.08176506e+00 1.04101680e-01 6.76720440e-01 -3.39171857e-01
-1.03180563e+00 3.11476499e-01 -2.97325552e-01 -2.06588760e-01
-1.61198035e-01 1.09273446e+00 8.48152459e-01 8.05048570e-02
5.32222927e-01 4.11688656e-01 2.13030279e-01 -6.66153431e-01
-3.12425882e-01 6.71927035e-01 2.81543106e-01 -2.56275922e-01
4.12985235e-01 2.27554440e-01 3.80620509e-01 -6.53514385e-01
-1.53664991e-01 -7.46696770e-01 -6.02529228e-01 -2.68915027e-01
5.65404475e-01 -7.96280384e-01 -7.65652359e-01 2.29362056e-01
-1.11051762e+00 -6.11764975e-02 4.11779284e-02 8.55145454e-01
-7.69703627e-01 1.10975474e-01 -4.11698610e-01 -1.56152773e+00
-4.90550846e-01 -8.85749161e-01 8.13133717e-01 -8.74585137e-02
-3.04451227e-01 -1.47359145e+00 5.43462992e-01 -6.32668793e-01
6.51169538e-01 6.69549286e-01 8.02461863e-01 -1.08796728e+00
3.14620763e-01 -5.93512058e-01 3.64308029e-01 2.21948564e-01
6.87597170e-02 2.00370505e-01 -5.84054887e-01 -3.55850369e-01
2.90792018e-01 4.00957018e-01 6.52275741e-01 7.57380784e-01
1.01665485e+00 -3.29056650e-01 -5.06638348e-01 1.95910648e-01
1.63112581e+00 7.68490553e-01 4.07484680e-01 2.29658544e-01
-1.85259301e-02 5.71604311e-01 9.46520507e-01 7.93149054e-01
6.16928041e-02 6.03620112e-01 -8.39764103e-02 -1.48557946e-01
7.44011581e-01 2.24223837e-01 1.25254929e-01 1.03346622e+00
-3.43285471e-01 -1.19665474e-01 -1.26769388e+00 5.50556362e-01
-1.91780436e+00 -1.00331402e+00 -1.75945386e-01 2.54786754e+00
6.47799551e-01 2.73858100e-01 4.81080651e-01 3.81537408e-01
6.79030657e-01 -4.19531822e-01 -7.03885183e-02 -7.96688259e-01
3.19228262e-01 1.20732784e-01 6.75626516e-01 4.38641578e-01
-1.14275348e+00 3.12240362e-01 6.99436617e+00 7.91293383e-01
-1.34657252e+00 -1.13893986e-01 7.32974470e-01 5.90678036e-01
-1.49600627e-02 -1.03068195e-01 -7.48794377e-01 6.18167460e-01
1.64174712e+00 -6.62249148e-01 2.35939622e-02 7.63183117e-01
1.00338292e+00 -4.60917363e-03 -7.27647603e-01 7.22933471e-01
-5.05252421e-01 -1.27289152e+00 -4.69932795e-01 -4.27122526e-02
8.51727962e-01 -1.48503900e-01 -2.33708069e-01 4.24226671e-01
1.21940255e-01 -9.45900261e-01 3.30977708e-01 8.33022058e-01
4.41857338e-01 -7.68350303e-01 1.02114058e+00 6.06546462e-01
-1.30274868e+00 -2.31323913e-01 -7.27157146e-02 -9.90299061e-02
3.63495320e-01 3.94817442e-01 -1.07037032e+00 9.18180943e-01
1.25235334e-01 6.22510612e-01 -1.41488060e-01 1.49587488e+00
6.29105151e-01 1.09570646e+00 -6.89310133e-01 -2.11132895e-02
3.49713475e-01 -2.64014482e-01 5.08498371e-01 1.33562577e+00
8.76093924e-01 2.31109455e-01 3.16819996e-01 2.70728976e-01
8.88943255e-01 5.41619658e-01 -6.23622656e-01 2.21468940e-01
3.27485591e-01 9.46043432e-01 -7.80639648e-01 -7.15223789e-01
-3.13034981e-01 1.94876492e-01 -3.92398328e-01 4.71753806e-01
-6.66707516e-01 -3.80477041e-01 -2.51416769e-02 1.43927678e-01
9.22875628e-02 -2.53047496e-01 -3.18301529e-01 -7.79739201e-01
-3.94815415e-01 -6.58744991e-01 7.48758972e-01 -4.02199894e-01
-1.05078518e+00 1.08636057e+00 4.82858270e-01 -1.89482021e+00
-9.23725307e-01 -2.44542077e-01 -8.71643722e-01 1.21052158e+00
-1.17190993e+00 -6.78739250e-01 3.46158035e-02 4.67393875e-01
7.10317671e-01 -2.02827573e-01 9.58573580e-01 2.59824932e-01
-3.99086148e-01 6.25029877e-02 7.29394436e-01 -4.69084382e-01
3.62381011e-01 -1.19363439e+00 3.07833403e-01 5.47617733e-01
-5.70849478e-01 5.20657957e-01 1.17247856e+00 -7.17517912e-01
-9.00007546e-01 -9.40961897e-01 1.21637905e+00 -2.14848831e-01
7.70491183e-01 1.06826313e-01 -1.15190983e+00 5.90524733e-01
2.43275911e-02 -3.71328920e-01 4.94483739e-01 -6.32300153e-02
4.38795328e-01 -3.02845594e-02 -1.22407115e+00 2.51647353e-01
1.22110724e-01 1.97491154e-01 -6.72147930e-01 3.38411570e-01
3.29014748e-01 -2.00127244e-01 -1.47461128e+00 6.19987667e-01
5.57008743e-01 -7.79727697e-01 4.98548627e-01 -5.53189814e-01
-6.97883591e-02 -7.53272772e-02 1.61405444e-01 -1.70962715e+00
-3.36802363e-01 -1.00730693e+00 2.06006646e-01 9.34154212e-01
5.50017655e-01 -1.06546915e+00 1.57366753e-01 6.97539628e-01
-2.53655273e-03 -8.89729023e-01 -1.06642711e+00 -1.08736610e+00
6.91948012e-02 -3.89046460e-01 8.39418828e-01 8.79735291e-01
1.20560639e-01 -1.16141461e-01 -4.94550139e-01 2.43940037e-02
4.01088387e-01 2.52362847e-01 5.95328093e-01 -1.56125927e+00
-1.02657773e-01 -3.04215044e-01 -3.16215008e-01 -3.37833047e-01
-3.19518335e-02 -4.88976479e-01 7.19578117e-02 -1.24590969e+00
-3.36486012e-01 -6.35296404e-01 -7.05352485e-01 2.12191537e-01
-5.17277531e-02 -2.70255774e-01 7.07170740e-02 3.35379034e-01
1.77449062e-01 2.08864406e-01 8.52296591e-01 3.58640790e-01
-5.14961779e-01 7.40364850e-01 6.04726523e-02 5.60262144e-01
8.84718418e-01 -5.73542416e-01 -4.46664572e-01 4.06390756e-01
-2.59802509e-02 8.31626415e-01 1.82733491e-01 -8.58818233e-01
-6.47994364e-03 -4.00718182e-01 1.39870927e-01 -8.49468410e-01
-9.14968271e-03 -1.31402957e+00 9.80202675e-01 9.21551406e-01
-4.50982958e-01 5.53950250e-01 4.11699086e-01 5.59254229e-01
-3.39112759e-01 -1.93355039e-01 7.27237642e-01 3.85873526e-01
-5.68167925e-01 1.11777008e-01 -6.36680603e-01 -6.79872215e-01
1.07042909e+00 -8.50639865e-02 1.96672395e-01 -4.89700198e-01
-1.17260993e+00 2.76077867e-01 7.90321920e-03 1.71119049e-01
1.06580824e-01 -1.18310797e+00 -8.89441967e-01 3.00214291e-01
4.09455113e-02 -6.55113757e-01 1.58509910e-01 1.27819550e+00
-4.39319223e-01 6.98829949e-01 -1.23762168e-01 -5.89305639e-01
-1.31183493e+00 6.71263695e-01 4.41659361e-01 -5.74387074e-01
-5.29033422e-01 -4.92126262e-03 -1.36495516e-01 -7.11495057e-02
-2.13009045e-01 -8.42214108e-01 -5.45585454e-01 1.05661443e-02
3.44864160e-01 5.83576024e-01 2.85162240e-01 -6.64739132e-01
-3.31844419e-01 4.55316037e-01 4.39478248e-01 -2.16339231e-01
1.28212821e+00 -7.01684952e-02 2.14924067e-02 7.87604570e-01
1.06833184e+00 -3.62672806e-01 -1.00824165e+00 -8.74497369e-02
6.31621778e-01 -7.15363026e-02 -4.82147932e-02 -5.59500992e-01
-6.08734787e-01 3.66878837e-01 8.70096922e-01 9.93132114e-01
1.35890865e+00 -5.79102099e-01 4.71924454e-01 2.14842215e-01
2.19861463e-01 -9.69794452e-01 -8.68358910e-01 4.29930836e-01
9.86112893e-01 -1.21704304e+00 -1.79148272e-01 -3.40114355e-01
-7.44624436e-01 1.33478773e+00 -1.32320017e-01 -4.30986136e-01
1.37135732e+00 9.96630117e-02 2.44786739e-01 2.97797564e-02
-1.23950326e+00 -8.07958469e-02 4.06919628e-01 1.50460288e-01
6.46773160e-01 3.78337145e-01 -1.19614744e+00 5.00142336e-01
-1.39298990e-01 4.51663882e-01 4.28327948e-01 7.66925812e-01
-3.24946582e-01 -8.52684259e-01 -5.73513687e-01 7.23263979e-01
-6.92878723e-01 8.62542093e-02 4.41397160e-01 7.87106574e-01
-4.34541672e-01 1.01931715e+00 5.74320145e-02 -1.10384591e-01
4.51398253e-01 4.63846087e-01 -1.68625489e-01 -2.60100752e-01
-1.03705609e+00 6.90728903e-01 1.88166961e-01 -1.52279332e-01
-4.58067566e-01 -7.34036446e-01 -1.12651455e+00 -1.63287014e-01
-3.88276577e-01 5.40682971e-01 9.29230928e-01 9.00677979e-01
3.44995707e-01 5.56288421e-01 1.14803743e+00 -8.49589050e-01
-1.10062277e+00 -1.21873200e+00 -8.18681538e-01 3.43172044e-01
3.23162705e-01 -6.60769641e-01 -4.06915516e-01 2.67137766e-01] | [6.850902557373047, 3.1843698024749756] |
e6df1c15-2c0f-408d-947a-e4a3aa5c17b1 | detection-of-paroxysmal-atrial-fibrillation | 1805.09133 | null | http://arxiv.org/abs/1805.09133v1 | http://arxiv.org/pdf/1805.09133v1.pdf | Detection of Paroxysmal Atrial Fibrillation using Attention-based Bidirectional Recurrent Neural Networks | Detection of atrial fibrillation (AF), a type of cardiac arrhythmia, is
difficult since many cases of AF are usually clinically silent and undiagnosed.
In particular paroxysmal AF is a form of AF that occurs occasionally, and has a
higher probability of being undetected. In this work, we present an attention
based deep learning framework for detection of paroxysmal AF episodes from a
sequence of windows. Time-frequency representation of 30 seconds recording
windows, over a 10 minute data segment, are fed sequentially into a deep
convolutional neural network for image-based feature extraction, which are then
presented to a bidirectional recurrent neural network with an attention layer
for AF detection. To demonstrate the effectiveness of the proposed framework
for transient AF detection, we use a database of 24 hour Holter
Electrocardiogram (ECG) recordings acquired from 2850 patients at the
University of Virginia heart station. The algorithm achieves an AUC of 0.94 on
the testing set, which exceeds the performance of baseline models. We also
demonstrate the cross-domain generalizablity of the approach by adapting the
learned model parameters from one recording modality (ECG) to another
(photoplethysmogram) with improved AF detection performance. The proposed high
accuracy, low false alarm algorithm for detecting paroxysmal AF has potential
applications in long-term monitoring using wearable sensors. | ['Gari. D. Clifford', 'Amit J. Shah', 'Supreeth P. Shashikumar', 'Shamim Nemati'] | 2018-05-07 | null | null | null | null | ['atrial-fibrillation-detection', 'electrocardiography-ecg'] | ['medical', 'methodology'] | [ 3.92864585e-01 -4.65936869e-01 1.39904797e-01 -2.55730867e-01
-9.73057210e-01 -7.35355437e-01 -7.59847388e-02 6.61214218e-02
-2.21470445e-01 8.94819438e-01 1.28422752e-01 -7.02888310e-01
-1.67293176e-01 -5.06520987e-01 -3.80948484e-01 -6.33274078e-01
-7.34680057e-01 1.02139108e-01 -5.75806916e-01 4.34048057e-01
-2.15731174e-01 7.15076089e-01 -9.10852849e-01 4.17816341e-01
7.82393038e-01 1.35886061e+00 -2.72400528e-01 1.10466969e+00
7.20736504e-01 2.96594054e-01 -1.01294243e+00 2.42102772e-01
1.67946726e-01 -8.04269850e-01 -3.18885446e-01 -1.55293807e-01
4.58181381e-01 -2.65013546e-01 -1.95796400e-01 5.27027428e-01
1.18103850e+00 -4.06581163e-01 5.14568150e-01 -6.19304478e-01
-3.43751162e-01 2.85419673e-01 -3.09695780e-01 1.24079216e+00
2.23929048e-01 1.35500416e-01 6.18541420e-01 -1.07435882e+00
6.84815198e-02 6.36955261e-01 9.42388177e-01 1.64607659e-01
-1.06103909e+00 -8.23516369e-01 -2.13402435e-01 6.91367239e-02
-1.43225646e+00 -3.81642401e-01 8.29509020e-01 -5.27453661e-01
1.02300847e+00 3.32531989e-01 1.24518979e+00 1.11893153e+00
7.88488150e-01 3.95779639e-01 8.33617449e-01 -3.68715823e-01
-7.11602941e-02 -2.88474739e-01 4.21719760e-01 4.88745838e-01
3.33816528e-01 3.99904221e-01 -2.30724603e-01 -6.95242047e-01
8.67137015e-01 3.25706303e-01 -6.18809164e-01 1.69841647e-01
-1.36614871e+00 6.14163458e-01 3.98742318e-01 4.82694924e-01
-7.51454949e-01 -4.51861061e-02 5.35167038e-01 6.39168084e-01
3.99291754e-01 4.75200802e-01 -4.79246140e-01 -1.88361168e-01
-1.05517459e+00 -6.56714588e-02 5.14026523e-01 2.09012613e-01
5.56880906e-02 3.94608587e-01 -6.73073828e-01 6.49255991e-01
1.05051689e-01 6.36556923e-01 8.80153239e-01 -4.49073732e-01
2.93106157e-02 5.18045008e-01 2.19442502e-01 -8.14083934e-01
-6.30272627e-01 -1.12929153e+00 -1.24633098e+00 -5.13092689e-02
-5.52457571e-02 -6.42332792e-01 -1.04223609e+00 1.47760510e+00
-1.37261927e-01 5.57852387e-01 2.04512477e-01 8.31037939e-01
7.38150239e-01 4.95433390e-01 1.45928517e-01 -9.09300506e-01
1.36521256e+00 -4.15077418e-01 -7.62469172e-01 -1.68420643e-01
2.31393248e-01 -4.02108610e-01 7.97770560e-01 5.19734442e-01
-8.98170531e-01 -7.08950281e-01 -1.15564728e+00 4.09514993e-01
6.74991757e-02 6.16200507e-01 4.15590674e-01 8.13151181e-01
-9.13107038e-01 4.57420200e-01 -9.75459039e-01 -2.53848433e-01
8.22781801e-01 4.17314291e-01 -6.75276816e-02 3.58431190e-01
-1.32888675e+00 5.13957322e-01 1.33628890e-01 4.27923918e-01
-1.20380068e+00 -6.60450101e-01 -6.45772815e-01 1.21943451e-01
-2.93822110e-01 -9.31590378e-01 6.78571641e-01 -1.00096452e+00
-9.98495221e-01 7.96750188e-01 -4.19725001e-01 -8.88277352e-01
3.37462246e-01 -5.63600183e-01 -8.73480082e-01 2.90504061e-02
5.77740036e-02 -1.47877797e-01 1.22406030e+00 -4.00362700e-01
-4.93428171e-01 -5.72283566e-01 -3.58546793e-01 2.99385898e-02
-2.25366831e-01 -3.27595510e-02 2.28084877e-01 -1.05018306e+00
-3.36209722e-02 -7.49430239e-01 -2.19543025e-01 -3.95260572e-01
-1.55998006e-01 -1.27910152e-02 7.98030436e-01 -7.36459553e-01
1.65943122e+00 -2.53937769e+00 -1.37714326e-01 2.91465819e-01
4.50793833e-01 5.34760892e-01 2.11302444e-01 -8.61835331e-02
-4.14947271e-01 3.63313630e-02 -5.98710954e-01 3.35302353e-01
-8.74432027e-01 -3.20035890e-02 -4.29584563e-01 6.20327115e-01
5.61517626e-02 1.13825834e+00 -7.10647345e-01 2.37401854e-02
1.56018600e-01 7.10487664e-01 8.27928632e-02 4.42650944e-01
3.95755768e-01 6.64022088e-01 -1.63261250e-01 5.19406378e-01
3.24128300e-01 -4.53140646e-01 -1.09217362e-02 -2.76808441e-01
1.60394505e-01 1.42929643e-01 -6.61900342e-01 1.67915440e+00
-2.27421090e-01 6.26124322e-01 -4.73000884e-01 -1.07491374e+00
1.04954207e+00 9.11126256e-01 5.15017271e-01 -2.74101079e-01
1.63873479e-01 1.56544849e-01 5.16538501e-01 -4.24691677e-01
-6.92507923e-01 -1.94544688e-01 2.74087101e-01 6.77280366e-01
-1.75834566e-01 4.62906271e-01 -1.62257433e-01 -2.72458166e-01
1.24035001e+00 -4.79356349e-01 6.07530296e-01 -4.51918870e-01
5.50547421e-01 -3.36954772e-01 6.43280387e-01 1.24940073e+00
-4.06305373e-01 6.52401447e-01 2.44569946e-02 -1.37966168e+00
-5.35134137e-01 -1.27087557e+00 -4.75409836e-01 2.70597249e-01
-3.96068871e-01 -4.39989835e-01 -3.53380084e-01 -7.05243587e-01
-6.53679436e-03 -3.78938537e-04 -6.34598553e-01 -4.64391291e-01
-6.20665371e-01 -1.19776618e+00 8.45572412e-01 8.18830252e-01
6.48059309e-01 -1.39324176e+00 -1.41934693e+00 6.19745195e-01
-2.35239044e-01 -6.33737803e-01 -2.90407419e-01 1.26546368e-01
-1.08181643e+00 -1.13238072e+00 -1.03054988e+00 -7.40483820e-01
2.45318592e-01 -2.41196305e-01 1.16186047e+00 7.87213072e-02
-9.45847929e-01 3.72831136e-01 1.86748393e-02 -8.57040286e-01
-1.73685640e-01 -1.28434584e-01 2.88226068e-01 4.74473834e-01
5.82855880e-01 -7.35368252e-01 -1.28872824e+00 -9.75274518e-02
-2.18498945e-01 -4.51339096e-01 7.44465530e-01 9.50505376e-01
6.94739640e-01 -4.25843954e-01 1.18034208e+00 -8.46742213e-01
1.20270741e+00 -2.82331824e-01 -2.90821940e-01 1.43710852e-01
-8.64179671e-01 -6.93498015e-01 4.24877554e-01 -4.69619215e-01
-4.82109070e-01 2.37973437e-01 6.99844211e-02 -6.17816925e-01
-2.51175821e-01 5.94650805e-01 1.79793626e-01 2.61976242e-01
1.07691371e+00 3.58691782e-01 -2.03826651e-02 -8.79614949e-02
-1.32919431e-01 9.28720832e-01 7.59220064e-01 -9.80592892e-02
3.18255872e-01 4.22813535e-01 -1.74420297e-01 -8.73687446e-01
-6.93892777e-01 -4.76389468e-01 -3.94818485e-01 -1.57192219e-02
6.72830403e-01 -1.20401978e+00 -3.55343938e-01 3.94297421e-01
-9.80799735e-01 6.34720325e-02 -3.23748261e-01 5.68079174e-01
-3.37337643e-01 3.22924107e-01 -6.09211028e-01 -1.00074661e+00
-1.34768188e+00 -6.29660487e-01 7.73413539e-01 -5.86671494e-02
-5.24431944e-01 -7.32205987e-01 2.42277279e-01 -2.78715611e-01
5.82241952e-01 7.27891743e-01 7.42053747e-01 -5.88779747e-01
7.01657832e-02 -5.35244644e-01 3.61930467e-02 5.35694897e-01
9.62902546e-01 -3.61840099e-01 -1.19189334e+00 -7.44750261e-01
3.71847242e-01 2.36767963e-01 8.79195392e-01 9.64279532e-01
1.15382755e+00 -1.06365852e-01 -4.99609977e-01 7.57741570e-01
9.50093567e-01 9.06300664e-01 5.73884189e-01 5.29902941e-03
5.75917423e-01 -5.53101480e-01 2.61170059e-01 4.67096150e-01
-2.59873122e-01 2.21548587e-01 -7.26875365e-02 -7.09039450e-01
9.00965407e-02 4.41952199e-01 2.66157627e-01 2.77266383e-01
-4.02288288e-01 -3.84299606e-02 -9.75490928e-01 6.34526849e-01
-1.65883946e+00 -8.95288229e-01 8.62247944e-02 2.45576358e+00
6.08481348e-01 -8.94042030e-02 3.12626481e-01 5.02124786e-01
6.72174990e-01 -1.79150179e-01 -7.96402156e-01 -4.81429189e-01
-2.67104298e-01 6.24995470e-01 -2.57024109e-01 2.77764201e-01
-1.41817677e+00 2.54757643e-01 6.70636654e+00 -1.55455619e-01
-1.63841510e+00 2.26535007e-01 1.05493963e+00 -5.54668382e-02
3.27551037e-01 -4.53323603e-01 -1.81126162e-01 4.54990178e-01
1.34070587e+00 -3.70303184e-01 1.23661846e-01 4.36661005e-01
3.99085850e-01 4.31432694e-01 -1.27505076e+00 1.47256315e+00
9.07562152e-02 -1.41601431e+00 1.13322631e-01 -2.21602768e-01
3.28023523e-01 1.88004285e-01 -1.12469634e-02 6.22852296e-02
-5.91188788e-01 -1.32176960e+00 -2.11528726e-02 5.42425692e-01
1.36259151e+00 -7.56072402e-01 8.45818162e-01 -7.99365435e-03
-1.06483185e+00 -4.41202253e-01 -8.72502327e-02 -1.92784086e-01
1.38873294e-01 9.91376281e-01 -8.84922743e-01 2.87456095e-01
1.02851701e+00 1.16950285e+00 -3.65668923e-01 1.14974546e+00
3.12192459e-02 1.09232366e+00 -3.48692536e-01 5.30103207e-01
-3.46419245e-01 1.30650848e-01 7.78563380e-01 1.39388037e+00
4.79869813e-01 3.04147929e-01 1.54414356e-01 7.92920351e-01
9.30677056e-02 -1.91319492e-02 -1.04253030e+00 4.04646285e-02
2.80366898e-01 1.08521974e+00 -4.91532475e-01 -5.94463050e-01
-3.24183077e-01 8.92809212e-01 -1.45225927e-01 5.24364054e-01
-6.94379866e-01 -5.22365034e-01 2.50826597e-01 1.64350897e-01
-3.85016464e-02 2.45454252e-01 -4.41799611e-01 -1.14547551e+00
2.51610458e-01 -9.68797982e-01 9.84561563e-01 -2.99501687e-01
-1.20896459e+00 1.20426750e+00 -5.17866135e-01 -1.43275511e+00
-3.90679628e-01 -2.17678472e-01 -9.81635869e-01 1.21328819e+00
-1.15188599e+00 -7.64727056e-01 -4.39404666e-01 8.31084967e-01
5.01316845e-01 -3.27278495e-01 1.44803631e+00 5.08251369e-01
-4.70045090e-01 6.19076610e-01 -6.00406528e-01 2.93154508e-01
6.05617344e-01 -1.20267093e+00 4.95655209e-01 1.11633623e+00
4.37277615e-01 8.58973503e-01 2.52496988e-01 -6.44812465e-01
-8.80555570e-01 -1.45249152e+00 8.14564645e-01 -5.84625542e-01
-5.52439652e-02 -2.88426131e-02 -8.93851459e-01 7.46156752e-01
2.71577295e-02 4.19942707e-01 8.74940097e-01 -4.76151183e-02
1.63526267e-01 -2.79942572e-01 -7.92174578e-01 1.27614468e-01
6.99935794e-01 -5.27391076e-01 -7.71417856e-01 4.03570592e-01
5.31508029e-02 -4.04822856e-01 -1.01461685e+00 9.01425362e-01
9.44166422e-01 -7.98518479e-01 1.01628780e+00 -6.74908459e-01
-1.54770657e-01 -1.83063224e-01 3.26498300e-01 -1.22162533e+00
-5.16152740e-01 -1.03599381e+00 -4.90957677e-01 4.96612668e-01
5.47303617e-01 -8.09709847e-01 5.35546243e-01 9.27851349e-03
-1.56968281e-01 -8.46482396e-01 -9.20800447e-01 -5.26233971e-01
-2.30126843e-01 -2.12410808e-01 2.04976201e-01 8.05565596e-01
-5.02695851e-02 6.04999602e-01 -3.56916726e-01 3.75981778e-01
3.87741536e-01 4.62917000e-01 1.66973084e-01 -1.58863103e+00
-1.78843245e-01 2.94880290e-02 -5.81943572e-01 -3.99357736e-01
-3.41930985e-01 -9.23866093e-01 -2.70889461e-01 -1.30891383e+00
1.02125585e-01 -1.88028604e-01 -1.13803744e+00 4.75017399e-01
-3.30167890e-01 5.22432685e-01 -4.58006382e-01 2.26305425e-01
1.01839639e-01 1.80143118e-02 6.51784480e-01 -3.30423087e-01
-6.96507692e-01 4.71420795e-01 -5.54377735e-01 6.42092645e-01
8.38216245e-01 -3.61538023e-01 -5.21096349e-01 -1.62246704e-01
-2.07318053e-01 3.30914080e-01 3.16954285e-01 -1.24124265e+00
-1.47105977e-01 6.49035156e-01 1.06726789e+00 -3.01932603e-01
8.07050169e-02 -6.56640291e-01 3.25206369e-01 9.19417977e-01
-1.54268831e-01 4.79409516e-01 4.86444056e-01 6.61455750e-01
-2.32515156e-01 6.41687751e-01 7.01521814e-01 -2.36278415e-01
2.01689936e-02 6.71024263e-01 -6.60562158e-01 5.47920018e-02
8.44354630e-01 -2.83478081e-01 1.44040614e-01 -2.48384669e-01
-1.37529922e+00 -3.00878078e-01 -4.68065888e-01 4.92407084e-01
1.07337189e+00 -1.22755051e+00 -1.10717833e+00 8.03471029e-01
1.07471481e-01 -1.73373431e-01 8.97409394e-02 1.14524972e+00
-4.57594663e-01 3.33832175e-01 -2.87112117e-01 -1.09372079e+00
-1.34487545e+00 3.33491921e-01 8.32041204e-01 9.63953808e-02
-1.44322753e+00 7.33248830e-01 1.97024435e-01 6.47012174e-01
3.74705791e-01 -5.67715824e-01 -3.60602140e-01 -2.21285313e-01
8.80972981e-01 2.12616809e-02 4.21704352e-01 -2.06371307e-01
-6.32141948e-01 5.21073103e-01 -7.74451941e-02 2.37660959e-01
1.00828409e+00 2.19536036e-01 -6.53979108e-02 5.77932775e-01
6.79433286e-01 -2.91846931e-01 -7.59260476e-01 1.77081749e-01
-2.64486313e-01 -1.85871199e-01 1.37711450e-01 -1.11091375e+00
-1.25097907e+00 1.05806625e+00 1.70102978e+00 1.42421797e-01
1.38740349e+00 -2.42499992e-01 7.28808403e-01 2.47548416e-01
-2.56267618e-02 -3.31999034e-01 -3.40944529e-01 -1.15159750e-01
9.93296206e-01 -9.27205324e-01 -2.32646331e-01 1.03129618e-01
-3.27266246e-01 1.30736327e+00 1.48640633e-01 -3.55472177e-01
1.05541468e+00 1.42806083e-01 6.60949171e-01 -4.62917358e-01
-5.78407884e-01 4.18748617e-01 2.50720382e-01 6.00408554e-01
5.96282780e-01 2.31211051e-01 -3.38579118e-01 8.46205771e-01
1.62624747e-01 3.77904505e-01 3.75505507e-01 8.73384058e-01
-1.70579329e-01 -5.98001897e-01 4.59298007e-02 1.00487101e+00
-9.60633159e-01 -2.88065106e-01 -2.78454900e-01 2.02108830e-01
1.11829434e-02 9.91787314e-01 -1.48546658e-02 5.11068404e-02
2.76543915e-01 3.95797908e-01 2.46407822e-01 -5.93203068e-01
-7.18544066e-01 2.86706865e-01 -1.33314118e-01 -3.56866241e-01
-3.23649287e-01 -3.94127280e-01 -9.16759312e-01 7.28428662e-01
-1.28693312e-01 3.04897279e-01 8.24322477e-02 8.11943293e-01
9.01700377e-01 6.85268402e-01 5.86586058e-01 -3.25266957e-01
-1.09716259e-01 -1.14488685e+00 -6.02084041e-01 3.25793564e-01
1.14112461e+00 -4.21994269e-01 -5.49364388e-01 3.20803523e-01] | [14.269021034240723, 3.2455484867095947] |
20c82f36-2b31-4c76-acd4-7705d821765c | a-deep-learning-approach-for-multimodal | 1803.00344 | null | http://arxiv.org/abs/1803.00344v1 | http://arxiv.org/pdf/1803.00344v1.pdf | A Deep Learning Approach for Multimodal Deception Detection | Automatic deception detection is an important task that has gained momentum
in computational linguistics due to its potential applications. In this paper,
we propose a simple yet tough to beat multi-modal neural model for deception
detection. By combining features from different modalities such as video,
audio, and text along with Micro-Expression features, we show that detecting
deception in real life videos can be more accurate. Experimental results on a
dataset of real-life deception videos show that our model outperforms existing
techniques for deception detection with an accuracy of 96.14% and ROC-AUC of
0.9799. | ['Erik Cambria', 'Soujanya Poria', 'Gangeshwar Krishnamurthy', 'Navonil Majumder'] | 2018-03-01 | null | null | null | null | ['deception-detection'] | ['miscellaneous'] | [-1.65367246e-01 -3.68523508e-01 -2.11027399e-01 -6.11511230e-01
-1.02379394e+00 -5.15704930e-01 6.29315794e-01 -1.08558767e-01
-4.45479035e-01 4.91428673e-01 3.26933384e-01 5.73454462e-02
2.93536067e-01 -3.80364470e-02 -1.74730837e-01 -4.12190288e-01
7.83058256e-02 -4.02124822e-01 -2.27149472e-01 -1.31720811e-01
7.56166220e-01 2.86713570e-01 -1.49206042e+00 7.29851007e-01
9.27831531e-01 1.20089877e+00 -9.03943896e-01 7.03060448e-01
5.49336016e-01 8.96448970e-01 -1.12083411e+00 -1.00509274e+00
-1.98595896e-01 -5.54286540e-01 -5.00514686e-01 -3.45799956e-03
6.96463645e-01 -7.66527534e-01 -6.37342513e-01 1.16510165e+00
4.28573221e-01 -1.94950178e-02 6.34354770e-01 -1.44799578e+00
-9.61577177e-01 8.34839419e-02 -6.39234245e-01 4.60184783e-01
9.15317237e-01 -7.17365891e-02 8.38091254e-01 -9.38666821e-01
2.73985505e-01 1.29824007e+00 5.88578880e-01 8.49960446e-01
-5.90020120e-01 -8.15392613e-01 -4.75936204e-01 8.82291615e-01
-1.46442401e+00 -9.05401409e-01 8.58571053e-01 -1.42727613e-01
7.25126326e-01 4.33571130e-01 4.74195659e-01 1.65141547e+00
3.50350112e-01 1.23262775e+00 1.17986763e+00 -2.14914709e-01
-2.49528345e-02 7.77088180e-02 2.62792856e-01 8.56314719e-01
1.04026437e-01 1.21619172e-01 -1.09006584e+00 -5.58109462e-01
-4.32510152e-02 -1.59543499e-01 -3.80543500e-01 3.51448447e-01
-6.18359745e-01 1.13191652e+00 8.88481215e-02 2.97061116e-01
-8.23415518e-02 8.17843080e-02 9.22026873e-01 5.19134998e-01
7.68076420e-01 2.73887575e-01 1.99831605e-01 -8.32974792e-01
-1.06939459e+00 6.11297190e-02 8.67720425e-01 2.00443834e-01
-1.96462318e-01 3.89749438e-01 1.50322020e-01 1.11251998e+00
2.99548179e-01 4.07360643e-01 9.62759793e-01 -1.04923666e+00
2.03260541e-01 4.36593503e-01 1.71290040e-01 -1.57987607e+00
-1.71378732e-01 3.11997943e-02 -7.22470641e-01 -1.07985370e-01
3.73921067e-01 -6.77663311e-02 -6.95677340e-01 1.28697658e+00
8.48877653e-02 2.30303466e-01 -1.43398987e-02 1.20331466e+00
9.63463008e-01 6.46489561e-01 -6.25092536e-02 -2.67343551e-01
1.26281166e+00 -7.40044653e-01 -9.24896002e-01 -2.62613803e-01
6.94430947e-01 -5.32006323e-01 7.71488011e-01 9.82782662e-01
-9.09596920e-01 4.65518758e-02 -1.17598510e+00 -1.09289363e-01
-2.53821462e-01 9.86059606e-02 6.02726936e-01 1.25197935e+00
-6.66700840e-01 2.72889733e-01 -3.73974711e-01 -2.37969577e-01
7.44362772e-01 -1.45641178e-01 -6.27259552e-01 -3.60220104e-01
-1.26622963e+00 1.09761572e+00 1.00187421e-01 4.53078359e-01
-8.78214538e-01 -3.26098688e-02 -1.01342976e+00 -3.00603248e-02
1.22536682e-01 -2.67866731e-01 1.05450690e+00 -1.49561858e+00
-1.58748341e+00 1.17263913e+00 -2.46775419e-01 -3.91670138e-01
3.63410711e-01 -4.54382807e-01 -9.88510489e-01 6.32769644e-01
-3.29254717e-01 3.56072634e-02 1.50235999e+00 -9.72862542e-01
-2.53508598e-01 -6.37904286e-01 -2.76228171e-02 1.61238253e-01
-8.68651867e-01 7.66476810e-01 2.76927978e-01 -2.52784997e-01
-1.91750243e-01 -5.97867668e-01 7.27473378e-01 -6.91294894e-02
-2.60626078e-01 -3.76590699e-01 1.18123984e+00 -1.15253854e+00
1.40015066e+00 -2.29318666e+00 -9.93124247e-02 -1.92437246e-02
4.74281281e-01 6.15332186e-01 4.32557054e-02 3.36252064e-01
2.91699380e-01 1.71822309e-01 2.70171966e-02 -4.60159600e-01
1.87124819e-01 1.19693065e-02 -3.70949917e-02 9.45485711e-01
-1.53601751e-01 1.00082135e+00 -8.54646444e-01 -4.30424482e-01
5.44669060e-03 5.82789779e-01 -1.00332528e-01 2.32110862e-02
5.80814660e-01 -2.51376480e-01 -1.63093030e-01 1.12994862e+00
4.97492403e-01 2.09848300e-01 -7.32369795e-02 1.08837970e-01
3.33519548e-01 9.80490223e-02 -3.26690614e-01 1.22181547e+00
-4.54093292e-02 1.19908929e+00 4.39357877e-01 -1.10706365e+00
8.04029644e-01 5.42656541e-01 -1.67229727e-01 -4.69947726e-01
4.73928869e-01 2.10601091e-01 7.02279210e-02 -9.26697373e-01
5.69553018e-01 -4.04711813e-01 -4.58536297e-01 3.73972148e-01
8.13722908e-02 -2.53931344e-01 -6.96479008e-02 4.18755770e-01
1.10270190e+00 -3.75182301e-01 4.06736016e-01 2.10408106e-01
6.18746877e-01 -3.38252723e-01 3.83859754e-01 6.98435426e-01
-1.09377551e+00 5.06241739e-01 6.17027581e-01 -3.22559595e-01
-5.07379591e-01 -6.46126986e-01 7.43928552e-02 1.00968647e+00
-3.70133743e-02 -4.33938950e-01 -7.13432074e-01 -8.58382106e-01
3.14402506e-02 7.70883739e-01 -6.93520188e-01 -7.45429516e-01
-2.68051684e-01 -8.13431740e-01 1.18404913e+00 7.31118768e-02
6.01620972e-01 -7.66166091e-01 -5.10145366e-01 -3.34257334e-01
-4.65434730e-01 -1.23881829e+00 -5.08187294e-01 -5.61303616e-01
-5.75794697e-01 -9.14617956e-01 -3.67399246e-01 -3.96203369e-01
3.15297723e-01 3.79775971e-01 7.75851309e-01 4.12330389e-01
-2.15553880e-01 5.18375933e-01 -6.61658883e-01 -3.76430720e-01
-5.41847825e-01 -4.60943282e-01 1.91946402e-01 4.97993857e-01
6.58508122e-01 -2.41863742e-01 -3.51722717e-01 1.48426339e-01
-1.05444789e+00 -5.25802851e-01 2.61207432e-01 1.06195581e+00
-4.80540484e-01 -3.66771132e-01 9.34354842e-01 -3.37048560e-01
1.22437525e+00 -7.46847093e-01 -4.71889116e-02 1.25855757e-02
-4.21145469e-01 -7.69354522e-01 4.41813439e-01 -5.05961180e-01
-9.69159842e-01 -4.57289398e-01 -1.48798168e-01 -3.85256290e-01
-3.46621573e-01 8.50650251e-01 3.52440514e-02 -3.38907838e-01
5.98652184e-01 4.24199700e-01 2.38855451e-01 -1.09383531e-01
5.69691882e-02 1.17382443e+00 5.06738901e-01 2.93846838e-02
4.23452586e-01 2.72929847e-01 -3.82467240e-01 -1.17352283e+00
-1.07947397e+00 -5.01740396e-01 -2.62619764e-01 -6.85279548e-01
2.38978073e-01 -6.77027225e-01 -1.04538155e+00 1.04127300e+00
-1.14593077e+00 3.60088468e-01 6.21406019e-01 4.41669405e-01
-2.14283109e-01 1.00112200e+00 -1.05653417e+00 -1.26593697e+00
-5.04058063e-01 -4.96350855e-01 8.55808198e-01 1.15831263e-01
-3.53465229e-01 -8.36618543e-01 -1.45700157e-01 1.23728824e+00
3.76775384e-01 3.70686382e-01 2.79918045e-01 -9.42331910e-01
1.75712898e-01 -8.65820050e-01 -2.85883486e-01 6.92608297e-01
-6.09132536e-02 8.97962525e-02 -1.05700564e+00 -1.69582486e-01
3.52835566e-01 -1.02829671e+00 9.10034239e-01 7.64054731e-02
1.16904092e+00 -6.09190881e-01 -5.04941754e-02 2.06145614e-01
9.78032529e-01 8.60873461e-02 8.27230036e-01 8.03698506e-03
4.93564755e-01 4.01873231e-01 5.57025194e-01 5.19208431e-01
4.53467727e-01 4.52310175e-01 7.05710649e-01 5.77437758e-01
2.67822742e-01 -5.10613695e-02 8.34286809e-01 7.88903534e-01
-3.53350081e-02 -6.05364323e-01 -8.51607740e-01 6.37742221e-01
-1.74714291e+00 -1.52673113e+00 -3.14283580e-01 1.92587280e+00
6.02980435e-01 -1.87928572e-01 3.89142841e-01 4.07065272e-01
5.81950009e-01 2.33019754e-01 -2.69271374e-01 -1.13509011e+00
-3.36296290e-01 -2.71589726e-01 -3.26551914e-01 4.89156097e-01
-9.78633046e-01 7.41982520e-01 6.88148165e+00 9.03827727e-01
-1.00924313e+00 2.70008087e-01 6.40018284e-01 -4.90235090e-01
1.67048052e-01 -6.59337044e-01 -2.40275204e-01 7.91998327e-01
1.32186389e+00 -1.57070383e-01 3.11359793e-01 5.74850976e-01
4.17300463e-01 -4.58893031e-01 -8.89640212e-01 1.31430459e+00
1.32474124e+00 -7.88244903e-01 -1.60591990e-01 -7.14608952e-02
2.69101262e-01 -2.47873262e-01 1.56160697e-01 3.23163956e-01
-2.27606013e-01 -1.31239629e+00 6.87349796e-01 3.75758260e-01
4.68097836e-01 -8.31658542e-01 9.87123132e-01 6.74983442e-01
-4.95066680e-02 -2.38073230e-01 -8.61132294e-02 -3.90933126e-01
2.36963719e-01 5.95710337e-01 -6.64972126e-01 2.69196630e-01
5.89166999e-01 8.51335227e-01 -3.34382355e-01 9.42905366e-01
-3.74171972e-01 9.19167519e-01 -7.33812973e-02 -4.76106912e-01
1.90119699e-01 3.39237675e-02 5.79646230e-01 1.36566281e+00
2.92307913e-01 2.91358501e-01 -3.73372316e-01 6.13358915e-01
-2.53530681e-01 -3.71054113e-02 -7.14138150e-01 -2.76160955e-01
3.06983292e-01 1.37697816e+00 3.04791871e-02 -3.49761903e-01
-3.87079090e-01 1.50589180e+00 3.70845854e-01 1.11500241e-01
-9.66584742e-01 -5.66194654e-01 5.53377390e-01 -3.88729930e-01
-8.30994546e-02 -3.44672650e-01 -1.51906768e-02 -1.51727283e+00
3.11203077e-02 -1.16634846e+00 5.45765758e-01 -8.67453635e-01
-1.44280803e+00 3.61332506e-01 -1.82649374e-01 -9.08574402e-01
-3.10637295e-01 -5.42198181e-01 -5.44822216e-01 4.29061204e-01
-1.28747272e+00 -1.41701341e+00 -3.91883671e-01 5.49854636e-01
4.55803990e-01 -1.38342336e-01 8.81901205e-01 2.45751202e-01
-7.42214739e-01 9.46348250e-01 4.28919196e-02 3.72187823e-01
6.87583447e-01 -7.15618610e-01 -2.64048398e-01 8.02787483e-01
1.11363582e-01 3.58382821e-01 7.20179439e-01 -3.75966877e-01
-1.43604517e+00 -3.24469835e-01 1.18188035e+00 -5.20319164e-01
9.58398521e-01 -3.49140644e-01 -9.87771332e-01 4.33814436e-01
1.94418907e-01 -8.67797211e-02 1.04041433e+00 -6.64838850e-02
-8.63795817e-01 4.64842208e-02 -1.68489528e+00 1.90853208e-01
7.42556393e-01 -8.25701356e-01 -8.03702533e-01 2.10555151e-01
-2.24682912e-01 -3.06409866e-01 -7.98558474e-01 -5.45965098e-02
1.21472764e+00 -1.25152099e+00 7.20942676e-01 -6.35258675e-01
7.76238501e-01 3.49585265e-01 -8.28819424e-02 -1.52103627e+00
1.18682384e-02 -5.75944006e-01 -6.02369428e-01 1.06441820e+00
6.25329763e-02 -7.36043274e-01 5.61197400e-01 6.06902897e-01
-4.13443893e-02 -5.36655843e-01 -1.45804679e+00 -7.31302500e-01
-7.62287015e-03 -4.22090948e-01 -3.39114480e-02 1.25955319e+00
8.91731024e-01 2.82676160e-01 -9.82486188e-01 2.33928836e-03
4.99653369e-01 -2.47236192e-02 3.30214679e-01 -1.07727706e+00
2.35372394e-01 -3.38380575e-01 -8.76997411e-01 -8.79669070e-01
6.95902407e-01 -5.33279896e-01 -1.04991496e-01 -9.67657626e-01
6.30571604e-01 5.03775060e-01 -1.43249959e-01 4.74568933e-01
-1.03998952e-01 9.32833552e-01 -4.46504615e-02 4.80431728e-02
-8.05246294e-01 7.62077928e-01 8.93425465e-01 -2.33125329e-01
3.29432309e-01 -2.70886958e-01 -7.38349915e-01 8.62859130e-01
9.91115689e-01 -3.53078187e-01 1.43734202e-01 -3.04503411e-01
1.81522667e-01 2.29192823e-01 7.36380219e-01 -4.50931072e-01
-3.52555923e-02 -4.23563682e-02 3.41467112e-01 -1.48316428e-01
9.94434297e-01 -3.96452606e-01 -6.42058015e-01 3.53277236e-01
-1.45400345e-01 -2.82514952e-02 1.17202923e-01 5.75585842e-01
-4.80712444e-01 -4.91234392e-01 7.08789885e-01 1.22696329e-02
-5.00823498e-01 -2.23972708e-01 -8.69838536e-01 -7.24355951e-02
9.70509827e-01 -4.93299842e-01 -6.85302734e-01 -1.20331156e+00
-5.02765477e-01 -2.45079119e-02 1.89362898e-01 7.51745164e-01
1.25872076e+00 -1.27028680e+00 -9.85349298e-01 -7.61611946e-03
2.02495992e-01 -1.31228912e+00 1.15501039e-01 1.06201851e+00
-7.48264864e-02 3.60757113e-01 -2.34913155e-01 -2.60948926e-01
-1.97533214e+00 9.25759748e-02 4.05066669e-01 4.14927125e-01
-9.01953056e-02 7.67467141e-01 -5.77873707e-01 6.52042180e-02
1.64417606e-02 2.47879371e-01 -1.09964445e-01 7.10749626e-02
8.48887205e-01 6.99502587e-01 -2.95147449e-02 -1.15505135e+00
-5.04845679e-01 -5.67904487e-02 -6.15991689e-02 -1.04624718e-01
1.15870154e+00 -2.18282178e-01 -2.61724025e-01 6.26108766e-01
1.25133848e+00 1.04823701e-01 -4.67774451e-01 1.27400294e-01
-1.06861383e-01 -1.06687057e+00 2.17866525e-01 -1.16754723e+00
-8.45747054e-01 7.12607801e-01 2.36403838e-01 5.89156210e-01
9.55762327e-01 -9.15436968e-02 1.01039934e+00 2.20537037e-01
2.03392684e-01 -1.19198501e+00 4.65753585e-01 4.47275132e-01
1.18751824e+00 -1.51781166e+00 1.29479542e-01 -1.83993772e-01
-6.94493294e-01 1.06476426e+00 4.82999772e-01 -1.47600278e-01
3.98402691e-01 -3.27388465e-01 2.02182084e-01 -3.04690838e-01
-6.38552725e-01 3.92247796e-01 4.94553715e-01 5.23767948e-01
4.03673828e-01 3.89667489e-02 -7.74485111e-01 6.93728626e-01
-1.75994277e-01 -1.04872651e-01 8.95096242e-01 6.79235280e-01
-4.04442996e-01 -7.11059809e-01 -4.84203815e-01 6.85863554e-01
-1.02815306e+00 -4.91153523e-02 -1.15340960e+00 6.95807040e-01
-3.21985722e-01 1.51349735e+00 -3.33632678e-01 -4.95784760e-01
-7.77212530e-02 4.02603328e-01 6.28075540e-01 -2.31150240e-01
-7.08267093e-01 -3.86080265e-01 6.47350132e-01 -6.56580389e-01
-4.95774537e-01 -9.12908852e-01 -6.34412706e-01 -7.94247389e-01
-5.54370105e-01 1.21818250e-02 6.69490874e-01 1.15295339e+00
4.64572042e-01 -3.29174668e-01 5.82125545e-01 -7.66576827e-01
-7.39610374e-01 -1.22263098e+00 -7.37391889e-01 6.37156188e-01
7.03263044e-01 -5.35290301e-01 -1.01143718e+00 -2.25961939e-01] | [13.266925811767578, 2.029111623764038] |
7872640e-20a2-4eaf-a9f0-ecf6bb39b6c5 | power-efficient-analog-features-for-audio | 2110.03715 | null | https://arxiv.org/abs/2110.03715v2 | https://arxiv.org/pdf/2110.03715v2.pdf | PEAF: Learnable Power Efficient Analog Acoustic Features for Audio Recognition | At the end of Moore's law, new computing paradigms are required to prolong the battery life of wearable and IoT smart audio devices. Theoretical analysis and physical validation have shown that analog signal processing (ASP) can be more power-efficient than its digital counterpart in the realm of low-to-medium signal-to-noise ratio applications. In addition, ASP allows a direct interface with an analog microphone without a power-hungry analog-to-digital converter. Here, we present power-efficient analog acoustic features (PEAF) that are validated by fabricated CMOS chips for running audio recognition. Linear, non-linear, and learnable PEAF variants are evaluated on two speech processing tasks that are demanded in many battery-operated devices: wake word detection (WWD) and keyword spotting (KWS). Compared to digital acoustic features, higher power efficiency with competitive classification accuracy can be obtained. A novel theoretical framework based on information theory is established to analyze the information flow in each individual stage of the feature extraction pipeline. The analysis identifies the information bottleneck and helps improve the KWS accuracy by up to 7%. This work may pave the way to building more power-efficient smart audio devices with best-in-class inference performance. | ['Milos Cernak', 'Minhao Yang', 'Boris Bergsma'] | 2021-10-07 | null | null | null | null | ['speaker-identification'] | ['speech'] | [ 5.15369594e-01 -3.84405345e-01 2.10694019e-02 -3.61450136e-01
-9.80861843e-01 -5.87178767e-01 -6.06960505e-02 4.88520533e-01
-3.76856834e-01 4.57702935e-01 -1.46022916e-01 -3.83740008e-01
-3.67888927e-01 -6.20415926e-01 -2.89937139e-01 -5.53229332e-01
-1.82486758e-01 1.71970751e-04 5.14051080e-01 5.80552965e-02
5.21701935e-04 5.17377734e-01 -1.92643106e+00 2.13074908e-01
5.20948231e-01 1.72240365e+00 2.99691290e-01 9.01684284e-01
2.06536084e-01 1.39719918e-01 -7.83547878e-01 -1.80123210e-01
-1.98913980e-02 -6.36834651e-02 -4.08003181e-02 -7.77801692e-01
1.20430335e-01 -5.01628637e-01 -3.87673557e-01 8.42486799e-01
1.05164242e+00 -1.77128345e-01 4.11014706e-01 -1.30494559e+00
1.46040633e-01 7.46391714e-01 -5.44463843e-03 3.24356437e-01
7.23987758e-01 1.52487069e-01 8.13553512e-01 -8.86921704e-01
-2.02967450e-01 8.28807056e-01 8.06856334e-01 5.92431091e-02
-1.07490647e+00 -9.01621163e-01 -4.39068884e-01 5.45580506e-01
-1.52494895e+00 -7.36196458e-01 8.36131692e-01 -5.39276898e-02
1.48959708e+00 5.52302241e-01 1.05165827e+00 7.03726172e-01
4.36252654e-01 4.96923238e-01 9.35714602e-01 -4.30828363e-01
6.64614379e-01 -7.49513656e-02 1.12599581e-01 4.15621102e-01
4.33428943e-01 -8.41395035e-02 -1.21658099e+00 -2.20842287e-01
9.38584954e-02 -7.67541006e-02 -2.16479614e-01 2.90689260e-01
-7.35448778e-01 2.45975509e-01 3.99775803e-02 2.69025475e-01
-3.36631238e-01 4.17469323e-01 4.00686175e-01 2.32957318e-01
-3.57377902e-02 4.05152470e-01 -4.35980946e-01 -8.30274284e-01
-1.14341295e+00 -1.63948849e-01 9.16015804e-01 7.81849682e-01
3.33412528e-01 3.73994857e-01 2.41807150e-03 4.64404255e-01
5.68403006e-01 1.01397967e+00 5.63023269e-01 -7.03535259e-01
1.92037746e-01 3.96219224e-01 -1.54725313e-02 -6.02465928e-01
-7.74037302e-01 -4.06539142e-01 -7.16534793e-01 -6.95756003e-02
1.74932063e-01 1.29504740e-01 -4.31848496e-01 1.29372418e+00
9.02180076e-02 -1.31524857e-02 4.57552727e-03 6.27308905e-01
5.31333804e-01 8.02772045e-01 -1.88123286e-01 -3.75848323e-01
1.73423553e+00 -8.47580954e-02 -1.12987363e+00 -3.05956036e-01
-1.22594759e-01 -6.12240732e-01 1.17975473e+00 7.33179986e-01
-1.10322595e+00 -6.40246153e-01 -1.76212037e+00 -1.06713712e-01
-2.43083000e-01 -2.30637435e-02 5.95283806e-01 1.41061389e+00
-7.88856864e-01 5.20269334e-01 -1.10338914e+00 2.49576107e-01
4.59111780e-01 8.47039938e-01 2.00564310e-01 3.30388963e-01
-1.16783953e+00 4.28149343e-01 -2.56443792e-03 -1.35560636e-03
-6.82304800e-01 -1.05501437e+00 -2.98630148e-01 3.97568375e-01
-2.86521995e-03 -5.60876250e-01 1.34628463e+00 -5.98070085e-01
-1.89114916e+00 3.18254441e-01 -7.02991635e-02 -7.10615754e-01
-5.84773235e-02 -3.41631800e-01 -9.00697291e-01 4.13734466e-01
-3.20971757e-01 1.42218992e-01 1.19820058e+00 -3.54719549e-01
-5.26399195e-01 -4.87447530e-01 -3.69946122e-01 -2.85042882e-01
-7.93516934e-01 -8.08753967e-02 2.44040904e-03 -3.68376970e-01
1.83302298e-01 -7.02108443e-01 2.72269547e-01 7.32978210e-02
-6.40749186e-02 -1.86613873e-01 7.97998726e-01 -5.14739275e-01
1.54809594e+00 -2.51235390e+00 -3.91806960e-01 3.44975650e-01
-9.64684319e-03 5.08346379e-01 3.98229867e-01 4.08717513e-01
3.46454382e-01 -2.22120076e-01 2.35088933e-02 -3.52693498e-02
1.57642856e-01 -1.39651150e-01 -3.53143871e-01 4.96916711e-01
6.12889081e-02 5.83542049e-01 -5.31694949e-01 -1.61117777e-01
2.55748510e-01 5.87168813e-01 -5.49996972e-01 1.61566019e-01
1.15456641e-01 6.40085489e-02 -2.67004043e-01 7.43122160e-01
3.37766379e-01 1.30658522e-01 2.42188126e-01 -7.40029693e-01
-1.80592909e-01 7.27658808e-01 -1.16873467e+00 1.72568965e+00
-6.40789211e-01 7.57305443e-01 3.22664678e-01 -6.49348617e-01
9.44381714e-01 2.96359450e-01 3.10181201e-01 -1.04842544e+00
2.99960047e-01 5.08894801e-01 5.86889572e-02 -5.62382877e-01
4.06864822e-01 -1.72103107e-01 -2.51169384e-01 1.38520107e-01
1.85784087e-01 -4.11336690e-01 -1.64696172e-01 -4.75588813e-02
1.21065092e+00 -2.84455240e-01 4.02420342e-01 -5.05614281e-01
3.82667959e-01 -4.54700232e-01 3.56976837e-01 6.27886772e-01
-1.83616042e-01 -8.68500248e-02 -5.07092252e-02 5.95854931e-02
-7.60681033e-01 -1.42212176e+00 -3.05395275e-01 1.02912867e+00
1.26344217e-02 -6.29899859e-01 -6.79148734e-01 3.26207966e-01
-2.49847416e-02 5.33469796e-01 4.26158011e-01 -3.14474225e-01
-3.45363379e-01 -5.50377727e-01 1.05802214e+00 5.74466646e-01
3.77193451e-01 -2.21376613e-01 -1.33435392e+00 4.13931787e-01
1.48812458e-01 -1.22180641e+00 -1.60037994e-01 7.60442257e-01
-8.46552193e-01 -3.97601038e-01 -6.40097558e-02 -4.60921407e-01
1.89240590e-01 -2.32725382e-01 6.11737728e-01 -4.25062627e-01
-4.90525544e-01 8.35222006e-01 -1.40495121e-01 -9.10537839e-01
-2.52461016e-01 -5.76165989e-02 7.46087909e-01 -5.72361350e-02
3.85562778e-01 -1.09590328e+00 -9.17016029e-01 1.28270417e-01
-5.38041413e-01 -2.94330001e-01 6.92092955e-01 4.05802101e-01
5.15075862e-01 2.99122363e-01 9.39716756e-01 1.64225832e-01
5.61522663e-01 -9.75400805e-02 -7.00016439e-01 -6.15383796e-02
-6.65855050e-01 1.09038353e-01 9.86181140e-01 -6.08525693e-01
-5.92074811e-01 3.48021716e-01 -6.64185405e-01 -9.10621062e-02
2.72801548e-01 -7.68633513e-03 -4.87885922e-01 -1.00036070e-01
4.51478809e-01 2.89168119e-01 -1.12993337e-01 -4.51173097e-01
5.34569882e-02 1.32880437e+00 8.11347127e-01 -2.08553314e-01
5.07031560e-01 2.88057774e-01 2.88983494e-01 -1.25781250e+00
-1.26922458e-01 -3.10187578e-01 -2.13132620e-01 -3.45141739e-01
5.55568874e-01 -1.14349842e+00 -1.30305374e+00 3.60583395e-01
-7.30730116e-01 2.28220206e-02 -3.62124562e-01 6.93634272e-01
-3.55715603e-01 1.33700907e-01 -4.58053112e-01 -1.39229119e+00
-9.61360514e-01 -9.18927789e-01 1.02827919e+00 3.51897418e-01
-4.60403830e-01 -3.94262463e-01 -3.55537593e-01 2.12450445e-01
5.83700538e-01 -3.41229558e-01 9.33572412e-01 -4.90044653e-01
-3.39925855e-01 -4.26135123e-01 2.92018294e-01 3.84567946e-01
-1.70405552e-01 -3.07414353e-01 -1.52634656e+00 -3.79809499e-01
3.97055298e-01 -6.18913881e-02 4.28492188e-01 2.61326462e-01
1.06790876e+00 -2.61484116e-01 -4.06821877e-01 3.47232163e-01
1.26237178e+00 5.15002728e-01 6.07282996e-01 -4.64608818e-01
1.07343838e-01 -6.55203164e-02 4.02968705e-01 6.70431554e-01
-4.65786718e-02 8.37465465e-01 2.16280341e-01 3.53524476e-01
-2.28086248e-01 -3.29172283e-01 8.51662219e-01 1.31429183e+00
4.42411751e-01 -1.26040429e-01 -7.12444067e-01 2.55829632e-01
-1.02551603e+00 -5.74111283e-01 -1.27045363e-01 2.46077824e+00
7.52214611e-01 4.60770130e-01 7.29883462e-02 1.17246473e+00
3.95389289e-01 -3.93554538e-01 -5.80954194e-01 -5.95218003e-01
1.48262411e-01 8.53797913e-01 5.10322690e-01 3.12706918e-01
-6.47458136e-01 1.25796109e-01 5.94227409e+00 9.87205088e-01
-1.34404981e+00 4.70730007e-01 9.06974077e-02 -4.55285221e-01
-1.04107045e-01 -4.27430004e-01 -8.25643003e-01 5.82050800e-01
1.76859927e+00 -9.34322625e-02 4.75378841e-01 8.55473876e-01
1.80795610e-01 -3.50570679e-01 -1.39273894e+00 1.52514160e+00
-1.81525633e-01 -1.01679778e+00 -2.74721831e-01 -2.92780787e-01
-8.86937827e-02 -2.58485049e-01 1.32778719e-01 8.03742483e-02
-8.04755569e-01 -7.41263866e-01 8.93359721e-01 4.73357439e-01
1.38645124e+00 -7.38021910e-01 4.43591654e-01 1.02320008e-01
-1.57475150e+00 -6.14908576e-01 -1.56834066e-01 -6.20318949e-01
1.03096694e-01 1.18951416e+00 -1.02111483e+00 3.03687006e-01
8.21769238e-01 1.14299983e-01 -4.15753752e-01 8.80816042e-01
4.58348244e-02 9.52137291e-01 -8.58717263e-01 -8.03312778e-01
-3.27964634e-01 2.04787076e-01 5.06923556e-01 1.20190895e+00
6.61275983e-01 3.32598954e-01 -3.45815837e-01 5.49313664e-01
-8.35364312e-02 -1.98296711e-01 -2.68317252e-01 -5.82233779e-02
9.77773011e-01 1.22132790e+00 -7.03045070e-01 -1.92091122e-01
-2.18117356e-01 7.00209737e-01 -6.68867230e-01 -2.74185598e-01
-7.55598903e-01 -8.91082108e-01 6.32278323e-01 4.29300368e-01
6.62128776e-02 -4.19825673e-01 -3.50531250e-01 -4.67598051e-01
1.84640780e-01 -5.60422242e-01 -8.84732604e-02 -6.29024446e-01
-9.26541865e-01 1.62733316e-01 -2.33389169e-01 -1.28360999e+00
-6.57213554e-02 -6.67416513e-01 -2.85316795e-01 4.63224560e-01
-1.15302145e+00 -5.44910431e-01 -1.61159098e-01 5.72381854e-01
4.48487312e-01 -9.16090515e-03 1.04929006e+00 8.28510940e-01
-7.67553151e-02 8.41587245e-01 2.14820564e-01 -1.98166162e-01
2.64637470e-01 -7.38683820e-01 2.74597645e-01 5.83349347e-01
-1.58714708e-02 3.71168047e-01 6.64051294e-01 -3.04925472e-01
-2.56571603e+00 -6.68181479e-01 4.99478042e-01 -6.16694801e-03
7.04363525e-01 -9.08223093e-01 -5.34832060e-01 -2.20573753e-01
-1.00842185e-01 -1.48265451e-01 9.43378270e-01 -3.05417866e-01
-4.98643294e-02 -1.09070361e+00 -1.20756674e+00 3.25500548e-01
9.40535843e-01 -8.82921696e-01 -3.76054347e-01 4.77137119e-02
7.24086344e-01 -4.36197668e-02 -9.29891229e-01 4.28967178e-01
1.10779226e+00 -6.05124056e-01 1.03048420e+00 1.31054625e-01
-3.00563008e-01 -4.19493169e-01 -5.52518785e-01 -7.55621910e-01
4.29896191e-02 -1.10295451e+00 -3.57903928e-01 1.29367721e+00
3.31496835e-01 -6.98423088e-01 4.31894749e-01 2.79742539e-01
-2.33808368e-01 -7.17615783e-01 -1.76181114e+00 -1.11310375e+00
-5.14411390e-01 -9.39691424e-01 3.96496028e-01 1.67310052e-03
7.19415069e-01 5.91993034e-01 -5.30103259e-02 2.59553105e-01
4.30955857e-01 -2.36802951e-01 1.70987353e-01 -1.30149138e+00
-4.84755933e-01 -3.60759437e-01 -8.43880534e-01 -1.17001331e+00
-3.96550894e-01 -6.98825657e-01 9.37815532e-02 -9.99535322e-01
-2.28413865e-01 -3.39726269e-01 -3.56556445e-01 2.62586087e-01
3.01722884e-01 4.95537102e-01 1.06387138e-01 -3.27498674e-01
-5.49114168e-01 1.97282836e-01 3.44536364e-01 -3.31278980e-01
-2.18204930e-01 2.14090556e-01 -2.59435445e-01 6.90343261e-01
7.66713977e-01 -5.32720685e-01 -3.95595193e-01 -4.18988317e-02
6.48220539e-01 2.20732152e-01 1.55780911e-01 -1.81154847e+00
6.35990918e-01 5.09546816e-01 3.78487170e-01 -5.60268044e-01
8.40632617e-01 -1.13877475e+00 2.85154641e-01 9.20845151e-01
7.05513433e-02 -2.72616278e-02 3.21170628e-01 3.58696401e-01
1.24069519e-01 -5.25645614e-02 5.96522033e-01 5.35736442e-01
-3.15761924e-01 -4.28994864e-01 -8.31959009e-01 -2.29142144e-01
7.32000589e-01 -2.67443717e-01 -1.98399812e-01 -3.16741049e-01
-4.27439302e-01 -2.82916486e-01 -2.40307972e-01 1.82630152e-01
9.14502144e-01 -1.01770544e+00 -1.38087079e-01 6.08802319e-01
-2.54071683e-01 -4.17250067e-01 2.59685874e-01 8.47178400e-01
-2.04598203e-01 6.53852046e-01 -8.92760530e-02 -8.44120502e-01
-1.45693934e+00 2.72932142e-01 1.97020009e-01 1.50039911e-01
-3.46807450e-01 6.55577719e-01 -8.42296779e-01 5.95250189e-01
4.66428787e-01 -7.04319179e-01 3.92109543e-01 1.67373314e-01
8.66493106e-01 9.46866155e-01 4.78892267e-01 7.80963451e-02
-7.75101185e-01 6.92924619e-01 5.35647750e-01 -3.04999053e-01
1.09900260e+00 -7.76222348e-02 1.40944287e-01 8.54212344e-01
1.19448984e+00 2.11911023e-01 -8.29424322e-01 1.91657439e-01
-6.16403855e-02 -2.35099539e-01 3.72110933e-01 -9.09334838e-01
-6.49230599e-01 1.12892294e+00 1.44818294e+00 4.00311857e-01
1.72384286e+00 -8.63722190e-02 1.11253071e+00 5.50616562e-01
6.90024257e-01 -1.32664156e+00 2.77469885e-02 4.73962501e-02
5.05541384e-01 -5.42568266e-01 3.04560810e-01 -2.46374711e-01
7.65138641e-02 1.24268937e+00 -1.08802490e-01 3.18537265e-01
9.09440577e-01 1.19205499e+00 -4.61950630e-01 1.46563068e-01
-5.51364303e-01 9.24703758e-03 1.57120824e-01 6.13295078e-01
1.47454560e-01 1.86310276e-01 -1.98027894e-01 1.34771311e+00
-3.88522416e-01 1.15170702e-01 1.99395761e-01 7.69143879e-01
-6.35398328e-01 -8.38281274e-01 -4.89539415e-01 7.60913551e-01
-6.93736255e-01 -2.08567828e-01 -1.57513648e-01 -6.12587258e-02
9.26889032e-02 1.51254964e+00 3.32181185e-01 -7.66875029e-01
4.53647494e-01 3.38285178e-01 6.28970325e-01 -9.85259637e-02
-8.16727400e-01 1.43526062e-01 1.04252480e-01 -6.51207864e-01
-9.17959437e-02 -4.34397608e-01 -1.56944513e+00 -1.93802923e-01
-4.06888753e-01 -4.81615290e-02 1.29224432e+00 7.35573053e-01
6.29963219e-01 8.93166065e-01 5.95592082e-01 -8.32959950e-01
-4.47231263e-01 -7.29762137e-01 -5.71795523e-01 -4.06409144e-01
3.91387820e-01 -3.81737441e-01 -3.93097997e-01 -1.83590665e-01] | [14.523079872131348, 5.574769496917725] |
cb6fb17c-860e-4e1b-986d-9369c412a81f | you-can-generate-it-again-data-to-text | 2306.15933 | null | https://arxiv.org/abs/2306.15933v1 | https://arxiv.org/pdf/2306.15933v1.pdf | You Can Generate It Again: Data-to-text Generation with Verification and Correction Prompting | Despite significant advancements in existing models, generating text descriptions from structured data input, known as data-to-text generation, remains a challenging task. In this paper, we propose a novel approach that goes beyond traditional one-shot generation methods by introducing a multi-step process consisting of generation, verification, and correction stages. Our approach, VCP(Verification and Correction Prompting), begins with the model generating an initial output. We then proceed to verify the correctness of different aspects of the generated text. The observations from the verification step are converted into a specialized error-indication prompt, which instructs the model to regenerate the output while considering the identified errors. To enhance the model's correction ability, we have developed a carefully designed training procedure. This procedure enables the model to incorporate feedback from the error-indication prompt, resulting in improved output generation. Through experimental results, we demonstrate that our approach effectively reduces slot error rates while maintaining the overall quality of the generated text. | ['Lingqiao Liu', 'Xuan Ren'] | 2023-06-28 | null | null | null | null | ['text-generation', 'data-to-text-generation'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.63211226e-01 3.96442860e-01 -7.90432915e-02 -2.88315922e-01
-1.06082499e+00 -5.64268589e-01 6.27403617e-01 5.53586125e-01
3.31403613e-02 8.74171257e-01 3.38210374e-01 -3.66703242e-01
2.64804482e-01 -7.55474746e-01 -5.47010243e-01 3.79695115e-03
6.36988401e-01 5.99258900e-01 1.59866408e-01 -1.90168902e-01
6.13729358e-01 -1.12016231e-01 -1.64288664e+00 4.80546564e-01
1.42636204e+00 5.35218716e-01 4.19131279e-01 8.89862120e-01
-5.35052299e-01 8.43402982e-01 -9.07495379e-01 -3.60951245e-01
-1.59157261e-01 -8.55512440e-01 -9.45293009e-01 1.95195258e-01
-2.20521778e-01 -2.95426756e-01 3.04809809e-01 7.54895985e-01
3.18524092e-01 2.69678801e-01 5.01130581e-01 -1.23858142e+00
-5.52046835e-01 9.52647984e-01 9.15915743e-02 -3.08436722e-01
7.48670876e-01 7.35921264e-02 9.57610011e-01 -1.09850395e+00
6.32914007e-01 1.01442933e+00 4.49153543e-01 1.02775693e+00
-1.23028588e+00 -4.79329735e-01 1.50531223e-02 -2.48470664e-01
-1.06513023e+00 -5.58417082e-01 6.72495186e-01 -4.14648771e-01
9.61819291e-01 3.14984977e-01 4.35852468e-01 8.45813990e-01
3.32348719e-02 7.09987164e-01 6.74497724e-01 -8.95318747e-01
2.60452390e-01 4.20724720e-01 1.48138911e-01 6.38503551e-01
2.05871463e-01 -1.76048581e-03 -5.78765154e-01 -1.17468372e-01
3.42977464e-01 -2.67870009e-01 -1.81313440e-01 -1.23201422e-02
-9.36099589e-01 5.91503918e-01 -1.01537362e-01 2.07802624e-01
-4.59512830e-01 -9.23204049e-02 3.25205117e-01 1.22190677e-01
4.98037547e-01 5.89387834e-01 -2.95787990e-01 -3.18090260e-01
-1.31127703e+00 5.68279684e-01 9.64830875e-01 1.24255228e+00
6.47449613e-01 -1.67057309e-02 -6.79272652e-01 6.92033470e-01
3.36127430e-01 2.56641537e-01 5.03862381e-01 -4.83132392e-01
8.29796255e-01 1.03439796e+00 5.27924478e-01 -6.33729160e-01
-1.01860575e-01 -2.20121622e-01 -5.21146595e-01 6.21280111e-02
2.13490784e-01 -4.34310615e-01 -9.53675866e-01 1.58330917e+00
3.00874233e-01 -2.20688790e-01 3.33793521e-01 5.04839420e-01
5.95660746e-01 7.14724422e-01 1.06764540e-01 -2.52554327e-01
9.88357484e-01 -9.96585429e-01 -9.48142648e-01 -1.10856295e-01
8.97676945e-01 -8.82153213e-01 1.02974355e+00 1.20335564e-01
-1.24597180e+00 -7.07748175e-01 -9.65126395e-01 -1.63980380e-01
-1.82398558e-01 4.76488888e-01 1.51058570e-01 4.40658689e-01
-9.25707042e-01 7.07369328e-01 -6.04570806e-01 -2.05307081e-01
5.03571145e-02 9.18712318e-02 1.10062761e-02 8.57509226e-02
-1.13405609e+00 7.00842261e-01 6.30343735e-01 -2.59746194e-01
-7.33741939e-01 -8.45115244e-01 -1.12272847e+00 2.01737136e-01
5.21765172e-01 -6.96224868e-01 1.94363844e+00 -8.61968756e-01
-1.60462713e+00 1.44781739e-01 -6.17311358e-01 -2.32403696e-01
4.45777923e-01 -1.95341468e-01 -3.55100572e-01 -2.01100200e-01
1.89937949e-01 5.23329973e-01 6.51427090e-01 -1.55469179e+00
-8.13492894e-01 3.59954610e-02 -8.81297514e-02 1.85998425e-01
-6.05955161e-02 1.33308768e-01 -4.60673422e-01 -7.25716233e-01
-1.59429297e-01 -6.33763313e-01 -2.35371307e-01 -3.35512787e-01
-6.90821230e-01 -2.65119493e-01 5.47620475e-01 -7.61963248e-01
1.78319514e+00 -1.91937041e+00 5.49027398e-02 2.77600110e-01
-7.17001921e-03 4.34344023e-01 -2.08307013e-01 8.67595494e-01
-3.93854603e-02 2.60119021e-01 -2.77928233e-01 -8.27656507e-01
-1.76230865e-03 -4.55119833e-02 -5.72754741e-01 -4.67668086e-01
6.55242026e-01 8.30153048e-01 -1.08823013e+00 -4.22432333e-01
2.52484195e-02 4.99507003e-02 -6.63331330e-01 6.11794531e-01
-7.34882772e-01 3.18452626e-01 -3.40338409e-01 4.23361242e-01
3.75296503e-01 -3.63243371e-01 6.77607134e-02 1.30895898e-01
-1.88437134e-01 5.65002322e-01 -1.18963826e+00 1.42424071e+00
-6.07122600e-01 3.74411196e-01 -3.41636151e-01 -5.53779244e-01
1.12664866e+00 5.25289118e-01 2.30232403e-02 -4.59114194e-01
9.01293755e-02 1.71362564e-01 -1.30849361e-01 -5.55210948e-01
1.20928109e+00 4.15457562e-02 -1.48164645e-01 9.49261248e-01
2.52269357e-02 -2.25791171e-01 6.34563029e-01 6.72848940e-01
8.28063488e-01 3.91390264e-01 2.93578684e-01 4.14751530e-01
5.63066423e-01 3.84039402e-01 2.83542007e-01 7.55636632e-01
2.90249705e-01 6.71072483e-01 4.19944733e-01 5.12183458e-02
-1.30605483e+00 -6.24957085e-01 3.98879766e-01 8.02075028e-01
-4.39229310e-02 -9.63973403e-01 -9.90374446e-01 -7.86708534e-01
-1.79356948e-01 1.27994072e+00 -4.54413056e-01 -2.90288031e-01
-3.51690054e-01 -2.56371319e-01 4.70903665e-01 7.65868545e-01
2.42252558e-01 -1.20427179e+00 -5.63431501e-01 4.49966580e-01
-5.08332551e-01 -7.19956219e-01 -6.22955918e-01 8.20361748e-02
-8.40610981e-01 -9.15864706e-01 -3.77154052e-01 -6.64305091e-01
1.10593462e+00 9.72708315e-02 9.10237014e-01 4.71723139e-01
-1.01353657e-02 -7.56935636e-03 -6.83007121e-01 -5.39833128e-01
-1.24423218e+00 6.99933693e-02 -1.95260257e-01 -2.14664593e-01
2.36059185e-02 -7.02536851e-02 -5.35066798e-02 7.37940893e-02
-1.05862367e+00 6.44005954e-01 5.80035150e-01 9.39104676e-01
5.01971066e-01 4.13359068e-02 7.47924328e-01 -1.23040247e+00
1.07821071e+00 -3.92553508e-01 -5.14811218e-01 4.50343728e-01
-9.47922111e-01 4.51618224e-01 9.88077462e-01 -4.45670009e-01
-1.33738625e+00 8.21822882e-02 -1.46076500e-01 -1.38506144e-01
-1.21769063e-01 7.80721068e-01 -4.83237170e-02 4.52358395e-01
6.62304163e-01 4.60820794e-01 -7.06723630e-02 -4.92015600e-01
3.65986496e-01 8.35539997e-01 5.36621809e-01 -4.36003298e-01
9.84781325e-01 -2.17223316e-01 -6.23665571e-01 -2.16705725e-01
-7.75435090e-01 -2.64245838e-01 -5.35899460e-01 -2.38875225e-01
4.58169252e-01 -6.16942942e-01 -2.76705056e-01 2.57981718e-01
-1.40717483e+00 -3.72038007e-01 -4.04124439e-01 4.68957648e-02
-3.90582889e-01 1.46698117e-01 -3.30224991e-01 -1.19247985e+00
-6.40257359e-01 -9.21011627e-01 1.18068230e+00 4.32573646e-01
-7.71614909e-01 -8.37191463e-01 1.12139240e-01 2.71024764e-01
3.75898004e-01 -1.66195780e-01 1.15917182e+00 -7.91076362e-01
-4.92564440e-01 -3.79486173e-01 -8.28793794e-02 1.91437036e-01
3.24521959e-01 2.83083826e-01 -7.24602580e-01 1.10406302e-01
-3.81731957e-01 -3.42741877e-01 4.20289427e-01 -2.61958838e-01
1.17066252e+00 -4.64121282e-01 -3.44498366e-01 1.80253491e-01
1.09225857e+00 3.79179865e-01 4.43222880e-01 8.57342221e-03
4.53690946e-01 4.60082352e-01 9.24794257e-01 7.01446533e-01
4.80918527e-01 5.01701653e-01 1.43440112e-01 -8.00072476e-02
-9.74349976e-02 -8.80438864e-01 2.27451667e-01 6.90318406e-01
2.16678798e-01 -7.04613924e-01 -6.58492565e-01 7.19909966e-01
-1.91711903e+00 -1.08719373e+00 -1.08661745e-02 2.15523195e+00
1.16753531e+00 1.89170256e-01 -1.40758842e-01 5.25020301e-01
5.90184450e-01 -1.20002784e-01 -3.69231880e-01 -6.16523921e-01
4.74045426e-01 3.25824916e-01 -1.85402166e-02 6.09643042e-01
-5.68665564e-01 1.00463378e+00 6.64173508e+00 4.84778285e-01
-8.70453477e-01 -2.66370833e-01 2.57123470e-01 1.15203485e-01
-7.45576620e-01 1.93800732e-01 -1.02782619e+00 5.42347729e-01
9.55036700e-01 -7.58479536e-01 3.00510794e-01 7.37792671e-01
5.51896691e-01 -2.72416741e-01 -1.28424084e+00 3.97033304e-01
2.65386164e-01 -1.32125282e+00 4.67272043e-01 -3.01932514e-01
7.08340108e-01 -9.02159870e-01 -3.35559547e-01 4.99183685e-01
5.13820350e-01 -7.52407432e-01 9.04667377e-01 5.54654360e-01
8.53496432e-01 -7.90894151e-01 5.43034673e-01 6.68845713e-01
-1.06073606e+00 1.40604414e-02 2.59973109e-02 -1.29033893e-01
4.35802519e-01 4.82884169e-01 -1.43142009e+00 8.95007551e-01
-3.06269643e-03 4.16847587e-01 -5.08063018e-01 8.87401640e-01
-5.44942141e-01 5.16512752e-01 2.04078063e-01 -2.80033678e-01
-2.22912997e-01 1.27572298e-01 4.18417722e-01 1.12863863e+00
4.97069806e-01 1.18492067e-01 3.06302100e-01 1.17300010e+00
-2.22670764e-01 1.81178719e-01 -4.78213280e-01 -3.79893839e-01
6.89382970e-01 1.20946527e+00 -4.03670877e-01 -7.18462825e-01
-4.48685139e-02 1.08742785e+00 3.84059608e-01 2.90031135e-01
-6.47221923e-01 -6.71655953e-01 2.43592858e-01 6.48562163e-02
2.02922419e-01 2.66317315e-02 -4.44272846e-01 -9.96620774e-01
2.98859358e-01 -9.48865116e-01 1.02012523e-01 -9.92476165e-01
-6.97871089e-01 6.79917216e-01 -7.98783004e-02 -1.17970991e+00
-8.32976818e-01 1.12256631e-02 -9.13160443e-01 1.21532023e+00
-1.36286807e+00 -8.31151307e-01 -4.54219639e-01 1.30466238e-01
9.00711656e-01 1.80796921e-01 9.33347166e-01 8.33893046e-02
-6.27623618e-01 7.00176656e-01 -2.75260121e-01 8.76037925e-02
6.21796608e-01 -1.34379506e+00 7.22058713e-01 1.13124728e+00
-1.63456023e-01 7.84583330e-01 8.68766427e-01 -1.11780679e+00
-1.04600585e+00 -1.45186830e+00 1.51182747e+00 -2.62734085e-01
4.74583030e-01 -3.89737576e-01 -9.63208079e-01 5.17414749e-01
1.40680954e-01 -5.74015260e-01 7.02120125e-01 -1.35918096e-01
-2.18446478e-01 2.81683385e-01 -9.77557838e-01 7.09181905e-01
6.98636532e-01 -4.12695259e-01 -7.80055225e-01 2.20866814e-01
9.10177469e-01 -7.26857483e-01 -3.61998737e-01 8.94348398e-02
3.69042367e-01 -3.88677806e-01 2.82039613e-01 -7.52863884e-01
8.58525097e-01 -3.50158155e-01 3.06610167e-01 -1.66951454e+00
-8.92945677e-02 -8.03651154e-01 -2.03960344e-01 1.56392097e+00
8.51139247e-01 -1.68267146e-01 4.97826338e-01 8.78025770e-01
-3.59213978e-01 -5.09142160e-01 -3.24075133e-01 -5.31181514e-01
-3.12623680e-01 -3.79013836e-01 7.30115950e-01 5.35026073e-01
5.80715775e-01 3.29496235e-01 -4.75935429e-01 6.28840253e-02
2.73728967e-01 2.13362262e-01 9.88416731e-01 -1.02830160e+00
-2.96755046e-01 -5.15742563e-02 3.52164865e-01 -1.04913354e+00
-1.08780272e-01 -8.44101012e-01 6.93559825e-01 -1.88249445e+00
2.16754526e-01 -4.82396662e-01 2.12478518e-01 6.86072409e-01
-7.19450533e-01 -1.41106546e-01 1.57090917e-01 8.07878673e-02
-4.77704674e-01 4.45609242e-01 1.10021555e+00 9.10102427e-02
-6.39069855e-01 2.13768423e-01 -9.36258197e-01 2.86177427e-01
8.45321953e-01 -5.38091898e-01 -5.25906324e-01 -3.22166026e-01
9.59276035e-03 2.79491633e-01 1.10220715e-01 -1.11778927e+00
3.26352119e-01 -2.22295523e-01 2.73429602e-01 -7.61726499e-01
2.88592316e-02 -5.27898073e-01 1.54924735e-01 3.90629441e-01
-8.64025950e-01 3.08848768e-01 2.34962270e-01 3.28750074e-01
-1.45322949e-01 -5.16718566e-01 4.63795125e-01 5.16189896e-02
-2.88192332e-01 4.51692343e-02 -5.89544058e-01 -3.78793776e-02
9.15669799e-01 -1.20028034e-01 -5.06877661e-01 -3.71555805e-01
-5.25595784e-01 4.06662434e-01 4.44327861e-01 5.44783056e-01
6.47681355e-01 -1.14300334e+00 -6.24445260e-01 4.27693397e-01
3.63725662e-01 6.80557340e-02 -6.38291091e-02 3.38734388e-01
-1.36672825e-01 4.33212459e-01 1.83111742e-01 -2.17120528e-01
-1.41088963e+00 4.74588245e-01 3.59218195e-02 -5.36014438e-01
-4.56620574e-01 3.79877061e-01 -3.52026314e-01 -3.36122125e-01
2.84606993e-01 -5.05453050e-01 -2.31941000e-01 -6.86396807e-02
8.16686571e-01 1.66008011e-01 2.74978876e-01 -2.72443503e-01
1.12773418e-01 4.05406430e-02 -3.61400038e-01 -5.31380415e-01
9.57851112e-01 -1.18260793e-01 3.27659786e-01 3.82039219e-01
5.74232459e-01 8.25466290e-02 -9.40711677e-01 -2.58992076e-01
1.25246838e-01 -3.74772012e-01 -3.13834757e-01 -1.26149023e+00
-5.07530272e-01 6.48703277e-01 -4.10533808e-02 2.81155109e-01
9.96821821e-01 -1.82180628e-01 9.03257132e-01 2.55822659e-01
2.12880611e-01 -1.13614714e+00 1.95598751e-01 7.60307789e-01
7.37512350e-01 -1.11469305e+00 -3.59577656e-01 -5.18013716e-01
-7.31416404e-01 9.91975665e-01 9.22019303e-01 3.96567881e-01
-5.60869649e-02 4.33869451e-01 1.45700455e-01 1.25688806e-01
-1.24956524e+00 -1.21065408e-01 2.10632786e-01 6.04698300e-01
5.45336366e-01 -1.36947975e-01 -5.54133415e-01 7.71251440e-01
-3.81075919e-01 4.98602122e-01 7.40281940e-01 1.17136085e+00
-7.06582248e-01 -1.49143696e+00 -3.10058922e-01 4.82886881e-01
-1.67926714e-01 -2.80154943e-01 -7.36639142e-01 4.53763634e-01
-6.00117333e-02 1.33767426e+00 9.14283469e-02 -5.32948017e-01
5.30771911e-01 5.12970865e-01 1.16495229e-01 -1.21000004e+00
-8.31242681e-01 -1.24827839e-01 4.05078590e-01 -2.91384071e-01
1.73385069e-02 -5.54977953e-01 -1.58351195e+00 -1.06014319e-01
-5.98537862e-01 5.54995179e-01 8.33252192e-01 1.12224185e+00
5.42215765e-01 7.76499748e-01 7.49831140e-01 -5.37422240e-01
-6.74479485e-01 -1.13872528e+00 -4.27210927e-02 4.93652314e-01
2.80375272e-01 -2.81112283e-01 -8.21745470e-02 4.40584421e-01] | [11.777220726013184, 8.894551277160645] |
83d55c29-39e7-44b7-a2b0-a9ad2a7c5cfc | illumination-aware-image-quality-assessment | null | null | https://link.springer.com/chapter/10.1007/978-3-030-88010-1_19 | https://link.springer.com/chapter/10.1007/978-3-030-88010-1_19 | Illumination-Aware Image Quality Assessment for Enhanced Low-light Image | Images captured in a dark environment may suffer from low visibility, which may degrade the visual aesthetics of images and the performance of vision-based systems. Extensive studies have focused on the low-light image enhancement (LIE) problem. However, we observe that even though the state-of-the-art LIE methods may oversharpen the low-light image and introduce visual artifacts. To reduce the overshoot effects of LIE, this paper proposes an illumination-aware image quality assessment, called LIE-IQA, for the enhanced low-light images. Since directly using the IQA of degraded image may fail to perform well, the proposed LIE-IQA is an illumination-aware and learnable metric. At first, the reflectance and shading components of both the enhanced low-light image and reference image are extracted by intrinsic image decomposition. Then, we use the weighted similarity between the VGG-based feature of enhanced low-light image and reference image to obtain LIE-IQA, where the weight of the measurement can be learned from pairs of data on benchmark dataset. Qualitative and quantitative experiments illustrate the superiority of the LIE-IQA to measure the image quality of LIE on different datasets, including a new IQA dataset built for LIE. We also use the LIE-IQA as a regularization of a loss function to optimize an end-to-end LIE method, and the results indicate the potential of the optimization framework with LIE-IQA to reduce the overshoot effects of low-light enhancement. | ['Sigan Yao，Yiqin Zhu，Lingyu Liang，Tao Wang'] | 2021-10-22 | null | null | null | prcv-2021-2021-10 | ['intrinsic-image-decomposition'] | ['computer-vision'] | [ 3.12473089e-01 -4.45112139e-01 3.56823057e-01 -4.63564068e-01
-6.81597888e-01 -2.64501154e-01 3.12352210e-01 -3.55946720e-01
-1.51048735e-01 4.54329997e-01 5.05940728e-02 2.01892197e-01
-2.27754444e-01 -8.77041399e-01 -6.13414049e-01 -1.04213989e+00
3.91168326e-01 -6.82201207e-01 3.63399796e-02 -2.48508260e-01
3.16067606e-01 3.08232933e-01 -1.75586939e+00 2.44102508e-01
1.18515003e+00 1.28383374e+00 2.86038905e-01 3.43125880e-01
-3.93615514e-02 3.68139863e-01 -3.96252543e-01 -3.16381246e-01
7.13755906e-01 -4.86822218e-01 -2.28296831e-01 3.47557932e-01
7.04029858e-01 -2.46729270e-01 -7.22188205e-02 1.28732336e+00
6.99543893e-01 1.95229471e-01 3.83791953e-01 -1.33809197e+00
-8.58136415e-01 -3.53497118e-01 -8.06524754e-01 7.16573969e-02
3.31187606e-01 5.00112951e-01 8.10829520e-01 -1.22409093e+00
2.91867256e-01 1.22363997e+00 5.23558199e-01 1.39527351e-01
-1.19917738e+00 -5.90692759e-01 -1.26004636e-01 2.60742098e-01
-1.31141293e+00 -3.34957987e-01 1.00161135e+00 -2.91814625e-01
3.89880836e-01 1.76867127e-01 5.97219110e-01 4.46041614e-01
4.49470103e-01 3.29966605e-01 1.77983701e+00 -4.84745413e-01
2.92457212e-02 1.50853664e-01 -2.80833077e-02 9.46798146e-01
4.21331525e-01 6.66261733e-01 -6.00040734e-01 1.13631301e-01
3.66892129e-01 9.80986166e-04 -5.20838618e-01 -3.17492783e-01
-9.06356633e-01 3.30573380e-01 7.21851170e-01 -1.22357443e-01
-2.76817352e-01 -2.64250457e-01 -6.22423403e-02 1.35982439e-01
6.52599216e-01 3.40092927e-01 -1.94677673e-02 2.21450061e-01
-5.66262484e-01 -5.62086999e-02 2.61455953e-01 3.04117471e-01
1.09704936e+00 -2.29261927e-02 -3.75924945e-01 1.01490128e+00
4.28790033e-01 6.93014264e-01 -8.69051367e-03 -9.96700287e-01
3.06319326e-01 8.59995186e-01 2.66260117e-01 -1.29384792e+00
-2.54973829e-01 -4.59121764e-01 -7.19014347e-01 9.11212087e-01
3.30441564e-01 1.74884871e-01 -6.47096515e-01 1.65725362e+00
3.96485686e-01 4.62541617e-02 1.57960132e-01 1.31434405e+00
9.44610894e-01 7.80364096e-01 -8.65483880e-02 -4.42000687e-01
9.98390675e-01 -1.05278718e+00 -6.96439564e-01 -2.75656372e-01
8.87631178e-02 -9.75044727e-01 1.65565515e+00 4.31216091e-01
-1.36650956e+00 -7.35682666e-01 -1.30290842e+00 -1.36583805e-01
-9.46033895e-02 2.52669305e-01 3.39401901e-01 7.26188183e-01
-9.00652766e-01 4.62252349e-01 -2.81180710e-01 -1.58067778e-01
2.45978728e-01 -1.47758514e-01 -2.91102886e-01 -5.40418386e-01
-9.71528113e-01 7.79878020e-01 1.57766804e-01 3.99052829e-01
-7.55323946e-01 -8.51930737e-01 -7.46854305e-01 -6.27196804e-02
3.20240587e-01 -5.71619630e-01 3.32336247e-01 -9.84540701e-01
-1.52959180e+00 1.02447462e+00 -7.80458525e-02 1.61752000e-01
2.96876013e-01 -6.50438368e-02 -5.03598273e-01 6.18229657e-02
1.62930377e-02 2.62903512e-01 1.00962698e+00 -1.79058135e+00
-4.58493799e-01 -4.25055742e-01 2.68357813e-01 4.53174740e-01
-2.00985700e-01 -1.35757491e-01 -4.90945786e-01 -3.70397687e-01
9.99906436e-02 -6.96217597e-01 2.76137441e-01 4.23827618e-01
-1.10452935e-01 1.45744830e-01 8.44119012e-01 -6.45312011e-01
1.03780651e+00 -2.40632772e+00 -2.79278070e-01 -5.29192574e-02
4.35906909e-02 4.44200665e-01 -5.02007842e-01 4.14748974e-02
5.02292765e-03 -1.40736386e-01 -4.30261075e-01 5.04680630e-03
-3.60292405e-01 1.06259868e-01 1.04577497e-01 6.72557473e-01
3.46745521e-01 7.82899857e-01 -9.98030663e-01 -3.92189294e-01
5.51866055e-01 7.11602211e-01 -1.99602872e-01 5.50893486e-01
8.02286640e-02 4.60148335e-01 -3.13146710e-02 6.57343745e-01
1.09806168e+00 1.62597179e-01 -4.73901361e-01 -8.25381637e-01
-1.89647257e-01 -2.66818225e-01 -1.09520352e+00 1.48648930e+00
-7.24245369e-01 6.15461409e-01 1.64969012e-01 -3.70641679e-01
1.25579524e+00 -2.41085514e-01 4.60767508e-01 -1.28431237e+00
1.43306494e-01 1.05393432e-01 -2.48572409e-01 -7.76513875e-01
1.55758917e-01 -2.50606000e-01 4.80614722e-01 2.81835377e-01
-4.67938632e-01 -1.97086528e-01 1.06992155e-01 -2.82549746e-02
4.89529371e-01 3.22028607e-01 1.15858734e-01 -1.06342606e-01
1.01218057e+00 -4.46591079e-01 5.69551289e-01 3.49204540e-01
-2.97914207e-01 8.16227973e-01 1.35722950e-01 -2.68742889e-01
-9.14025784e-01 -1.33969522e+00 -2.72067130e-01 9.79846179e-01
6.87091827e-01 -7.52757117e-02 -8.22581112e-01 -4.28541034e-01
-1.41263098e-01 7.49085903e-01 -2.60000169e-01 -4.85202879e-01
-2.03138307e-01 -1.00691593e+00 4.71049324e-02 4.64629605e-02
1.09066224e+00 -8.22293520e-01 -5.65879464e-01 -1.71927989e-01
-4.20796335e-01 -1.20737171e+00 -5.53045332e-01 -3.25069636e-01
-4.72544342e-01 -1.17437339e+00 -4.59858865e-01 -4.42301661e-01
7.80193925e-01 8.22879374e-01 1.11672556e+00 1.02592684e-01
-5.67152739e-01 3.74608487e-01 -2.27897570e-01 -2.85967380e-01
-1.97478458e-01 -7.27276504e-01 -2.51743138e-01 6.55615151e-01
-2.71805096e-02 -3.17722201e-01 -1.12585509e+00 6.05825841e-01
-9.90939379e-01 1.75085291e-01 7.08609164e-01 9.47889686e-01
7.24650562e-01 2.96854556e-01 2.27656081e-01 -4.04445976e-01
6.56057775e-01 7.52384365e-02 -6.68142974e-01 3.41779590e-01
-1.11243260e+00 -1.12080134e-01 3.10040295e-01 -2.01147884e-01
-1.58167124e+00 -1.44850984e-01 1.79413468e-01 -3.54213983e-01
1.44431040e-01 2.43075117e-01 -6.14434719e-01 -6.89308345e-01
6.71933591e-01 1.69783771e-01 3.11177466e-02 -2.80288905e-01
3.25362265e-01 5.81926405e-01 5.62698960e-01 -4.18654889e-01
7.53330946e-01 7.37082183e-01 4.23821867e-01 -7.46788323e-01
-1.22924197e+00 -4.82061863e-01 -1.47733003e-01 -5.99870443e-01
7.65640795e-01 -9.47033286e-01 -8.08075488e-01 6.30434215e-01
-9.14774120e-01 -1.57623485e-01 -2.30734050e-01 2.21353754e-01
-4.45183486e-01 5.76638162e-01 -3.20133597e-01 -8.69638324e-01
-4.89970237e-01 -1.19626641e+00 1.05758488e+00 5.35430014e-01
5.86097121e-01 -7.26088643e-01 -2.90418994e-02 7.03445792e-01
2.81371623e-01 4.48267192e-01 9.56168354e-01 5.60424745e-01
-6.16141975e-01 6.51129410e-02 -8.41500998e-01 9.90317643e-01
1.88868672e-01 4.80116531e-02 -1.22574449e+00 -4.67926890e-01
1.75212339e-01 -2.35402361e-01 6.80968761e-01 3.98547202e-01
1.12791300e+00 -9.90144610e-02 2.65362799e-01 1.04996252e+00
1.90292764e+00 9.97982249e-02 9.56545174e-01 4.37171221e-01
7.17273772e-01 6.67466640e-01 1.04214728e+00 2.66227365e-01
2.97651142e-01 8.03851843e-01 8.28413188e-01 -6.82197273e-01
-6.26843035e-01 -9.69957933e-02 5.60495138e-01 5.59493184e-01
-1.68221161e-01 -5.67332469e-02 -5.72630048e-01 3.23181450e-01
-1.52368844e+00 -6.75042391e-01 -4.38443333e-01 2.44100213e+00
7.21191227e-01 -1.62479207e-01 -2.75699496e-01 6.74845949e-02
5.96400917e-01 2.09153458e-01 -5.51131725e-01 -2.38423675e-01
-5.50010383e-01 -2.02283449e-02 3.64193529e-01 5.28405964e-01
-7.91454017e-01 6.00991905e-01 5.64258480e+00 7.25940764e-01
-1.32789528e+00 1.42808035e-01 7.31794834e-01 -7.45032355e-02
-2.08733410e-01 5.58650903e-02 -5.07259369e-01 5.68971515e-01
5.15250325e-01 -1.99500620e-01 7.25610673e-01 5.78338444e-01
6.83876812e-01 -4.77397621e-01 -7.87715375e-01 1.47039032e+00
4.38361883e-01 -6.66293800e-01 -1.90571137e-02 -3.96592952e-02
9.31717992e-01 -2.99922019e-01 3.87566924e-01 -1.77160919e-01
-2.04537168e-01 -8.22444856e-01 4.23306227e-01 7.53176033e-01
8.62126172e-01 -6.81219876e-01 5.78292072e-01 1.10595971e-02
-1.14608693e+00 -7.08335117e-02 -5.09280324e-01 1.70760557e-01
1.41560957e-01 8.26261401e-01 -3.76328558e-01 5.72496533e-01
7.77434289e-01 7.80837834e-01 -9.47009504e-01 1.18836308e+00
-3.99837047e-01 2.31432810e-01 1.00061245e-01 3.61726135e-01
1.33568734e-01 -7.50106573e-01 6.33568764e-01 9.58372593e-01
2.24045902e-01 2.15652615e-01 2.02248588e-01 1.13183713e+00
8.67131725e-02 1.19111434e-01 -4.75621164e-01 2.57221758e-01
-1.40341371e-01 1.54612899e+00 -2.40053251e-01 1.02216147e-01
-4.55212504e-01 9.73542154e-01 -1.38726383e-01 6.48076773e-01
-7.39148378e-01 -3.48229200e-01 5.78920364e-01 3.16814274e-01
-3.29241842e-01 -5.49214669e-02 -5.08120477e-01 -9.98053253e-01
3.60683441e-01 -7.31246352e-01 1.68910861e-01 -1.59498167e+00
-1.25190222e+00 5.66489697e-01 -2.07773656e-01 -1.41558182e+00
4.01068479e-01 -5.88701665e-01 -6.26357079e-01 9.80475962e-01
-1.93336284e+00 -1.09530461e+00 -1.01994824e+00 8.26656520e-01
3.82613629e-01 1.32690951e-01 3.38491559e-01 3.37022632e-01
-5.14419913e-01 2.89861768e-01 1.38788104e-01 -2.81233221e-01
9.15148139e-01 -9.66418266e-01 -2.67291009e-01 1.10473347e+00
-1.53406382e-01 2.49215052e-01 5.88169813e-01 -3.27701569e-01
-1.38827765e+00 -1.16551495e+00 3.58313888e-01 -1.52995884e-01
1.44825369e-01 2.78551765e-02 -9.70137000e-01 -9.35556889e-02
1.98465884e-01 3.90283465e-01 3.31708759e-01 -3.78190428e-01
-4.07173425e-01 -8.22777927e-01 -1.35710728e+00 5.76467276e-01
9.64813113e-01 -5.27101219e-01 -2.66689211e-01 1.56203449e-01
5.11551440e-01 -9.91164893e-02 -7.81662107e-01 6.58609688e-01
5.55098236e-01 -1.42088020e+00 1.21805942e+00 3.96654047e-02
5.28021514e-01 -6.06045604e-01 -1.70592338e-01 -1.46681213e+00
-2.86067963e-01 -2.98891544e-01 3.40186507e-01 1.35089433e+00
1.27905354e-01 -5.79672813e-01 3.10357511e-01 4.94304031e-01
6.53938204e-02 -7.33221233e-01 -4.84337270e-01 -6.98556840e-01
-3.07123423e-01 -2.33018771e-01 3.90236914e-01 6.48013711e-01
-5.98773420e-01 2.56556481e-01 -3.53388488e-01 3.33974302e-01
1.22375453e+00 3.48942578e-01 7.51295567e-01 -1.05061591e+00
-7.72047713e-02 -2.19250739e-01 -2.50349581e-01 -4.16403592e-01
-1.04792707e-01 -7.84684062e-01 2.28990376e-01 -1.69364738e+00
3.95829260e-01 -3.39802921e-01 -5.30411184e-01 1.84403211e-01
-4.17783648e-01 5.33510923e-01 3.53136212e-01 1.45443648e-01
-5.26738107e-01 7.59600580e-01 1.63788295e+00 -3.44878763e-01
-2.61092424e-01 -2.36728281e-01 -4.98001456e-01 7.78428197e-01
5.38169384e-01 -3.01812291e-01 -4.71734405e-01 -3.35380733e-01
3.95632267e-01 -1.75798386e-01 6.40319705e-01 -9.78536427e-01
1.64753217e-02 -3.90460968e-01 3.98758143e-01 -3.94444853e-01
4.46135581e-01 -8.91803682e-01 7.91979283e-02 1.89950183e-01
-1.72014937e-01 -4.45795178e-01 -4.27456014e-02 5.40152013e-01
-2.90166438e-01 -2.98313461e-02 1.39740145e+00 -9.57059935e-02
-7.73175120e-01 3.02026421e-01 3.77362698e-01 6.95250854e-02
1.04662991e+00 -4.16672260e-01 -6.15338027e-01 -2.24031821e-01
-2.01265216e-01 -1.68939888e-01 6.73067868e-01 3.20360720e-01
9.87159014e-01 -1.37578213e+00 -7.40098953e-01 3.74616742e-01
5.69145739e-01 -3.16453874e-01 4.21329975e-01 8.54806244e-01
-4.29313242e-01 -1.91160157e-01 -5.51138818e-01 -7.23440945e-01
-1.41409457e+00 5.71126997e-01 5.31031966e-01 -2.77817156e-02
-4.02330101e-01 4.79780316e-01 6.16806030e-01 -3.76155302e-02
3.82841453e-02 -1.64348837e-02 -9.40063223e-02 -3.91043603e-01
6.01314068e-01 6.09417498e-01 2.53966212e-01 -7.47114539e-01
-2.29650900e-01 9.85113323e-01 3.91865700e-01 3.85582400e-03
1.35782874e+00 -7.77360976e-01 -2.31421247e-01 3.22277635e-01
1.28782237e+00 -5.55809448e-03 -1.47069049e+00 -2.69697160e-01
-4.96729404e-01 -1.19832468e+00 5.93351066e-01 -1.06702626e+00
-1.46453273e+00 9.59090650e-01 1.19926178e+00 -1.17379762e-01
1.76872635e+00 -4.45861012e-01 7.43886948e-01 2.86404800e-04
3.91979784e-01 -9.99519706e-01 4.96988654e-01 2.95107197e-02
1.05123806e+00 -1.53906906e+00 1.21031791e-01 -5.97570121e-01
-7.57685006e-01 9.55422699e-01 6.52189970e-01 6.71163499e-02
4.75754321e-01 4.82592098e-02 3.46015841e-01 -5.52845187e-02
-3.65292579e-01 -4.73143518e-01 6.21672511e-01 7.23064065e-01
2.40760922e-01 -2.11056113e-01 -4.70086366e-01 1.97616041e-01
1.90469101e-01 -5.72744990e-03 4.12228972e-01 4.12444860e-01
-4.38307017e-01 -7.68303931e-01 -5.64891100e-01 9.52357054e-02
-3.31296384e-01 -6.51734248e-02 -1.89610735e-01 3.19742233e-01
5.21543682e-01 1.27051687e+00 -3.53344291e-01 -3.97622019e-01
5.42093992e-01 -3.95502746e-01 5.31712115e-01 -2.23932192e-01
-4.12400097e-01 1.37258187e-01 -2.59179294e-01 -9.44205165e-01
-6.44838095e-01 -1.57481790e-01 -9.05099332e-01 -2.26586118e-01
-3.14099878e-01 -3.75425965e-01 8.13178241e-01 5.85610688e-01
2.51081318e-01 2.82948077e-01 1.09100151e+00 -7.47941792e-01
-2.32128084e-01 -5.78841090e-01 -8.27714860e-01 9.42808032e-01
3.32320482e-01 -7.82849610e-01 -6.50658906e-01 4.31649424e-02] | [10.727858543395996, -2.5181689262390137] |
ae8bc451-bd7a-4936-ac7e-d28bf0e3b273 | learning-an-efficient-multimodal-depth | 2208.10771 | null | https://arxiv.org/abs/2208.10771v1 | https://arxiv.org/pdf/2208.10771v1.pdf | Learning an Efficient Multimodal Depth Completion Model | With the wide application of sparse ToF sensors in mobile devices, RGB image-guided sparse depth completion has attracted extensive attention recently, but still faces some problems. First, the fusion of multimodal information requires more network modules to process different modalities. But the application scenarios of sparse ToF measurements usually demand lightweight structure and low computational cost. Second, fusing sparse and noisy depth data with dense pixel-wise RGB data may introduce artifacts. In this paper, a light but efficient depth completion network is proposed, which consists of a two-branch global and local depth prediction module and a funnel convolutional spatial propagation network. The two-branch structure extracts and fuses cross-modal features with lightweight backbones. The improved spatial propagation module can refine the completed depth map gradually. Furthermore, corrected gradient loss is presented for the depth completion problem. Experimental results demonstrate the proposed method can outperform some state-of-the-art methods with a lightweight architecture. The proposed method also wins the championship in the MIPI2022 RGB+TOF depth completion challenge. | ['Yang Zhao', 'Kai Zhao', 'Yuanyuan Du', 'Dewang Hou'] | 2022-08-23 | null | null | null | null | ['depth-completion'] | ['computer-vision'] | [ 6.73478067e-01 1.96465105e-01 -4.43213806e-02 -5.42509675e-01
-8.98851156e-01 1.54628515e-01 9.37766358e-02 -2.86015451e-01
-4.12656516e-01 5.52010477e-01 3.43996316e-01 2.39987090e-01
-6.25490099e-02 -8.14303696e-01 -7.23101318e-01 -7.51386404e-01
3.12670618e-01 1.01969294e-01 4.26835179e-01 -6.93389326e-02
2.36315027e-01 1.01253033e-01 -1.66646862e+00 4.87983167e-01
1.08642864e+00 1.68030620e+00 5.65490663e-01 4.76484805e-01
-1.85202658e-01 1.11601293e+00 7.62205571e-02 -1.77788839e-01
4.18637961e-01 -5.82567230e-02 -4.85476047e-01 -1.81551068e-03
7.68624783e-01 -8.16896200e-01 -9.05779004e-01 1.03306067e+00
7.29704618e-01 6.75890073e-02 -2.41438579e-02 -1.15406668e+00
-3.50268856e-02 4.53745633e-01 -9.16657090e-01 -2.19767317e-01
4.99320596e-01 1.84593379e-01 5.37266195e-01 -1.03700078e+00
4.89852875e-01 1.22018421e+00 7.47034907e-01 6.02491617e-01
-6.99314535e-01 -6.33639455e-01 2.24137276e-01 2.46269122e-01
-1.10548794e+00 -4.28732038e-01 9.69941497e-01 -3.02557964e-02
8.91537547e-01 -5.46440994e-03 8.45346093e-01 9.06039357e-01
6.79280329e-03 1.05090737e+00 1.06437349e+00 3.17071192e-02
9.83462781e-02 -2.64149815e-01 -2.91136175e-01 9.07106042e-01
2.57414252e-01 6.98400140e-02 -1.15554011e+00 2.77800232e-01
9.22009110e-01 5.33030689e-01 -5.23020923e-01 -3.56863678e-01
-1.46893573e+00 3.96909833e-01 8.95393312e-01 7.99011067e-02
-3.97033840e-01 4.77670819e-01 6.31183162e-02 -6.53690025e-02
4.64337945e-01 -2.10729569e-01 -3.76957297e-01 -2.51301765e-01
-1.14406395e+00 9.18646455e-02 3.38998228e-01 9.17389274e-01
1.20526183e+00 -2.03893766e-01 1.56625405e-01 5.47621071e-01
5.89568079e-01 6.78410888e-01 3.36556554e-01 -1.19371271e+00
1.06416678e+00 7.69555986e-01 -1.16993882e-01 -9.32016134e-01
-6.96038008e-01 -1.67194381e-01 -1.18956947e+00 5.67429140e-02
4.17097002e-01 -1.04985945e-01 -9.12760437e-01 1.35649478e+00
4.71004874e-01 4.23321068e-01 -1.42114416e-01 1.22927213e+00
1.24377120e+00 5.78689277e-01 -3.94087285e-01 -1.58038780e-01
8.41286838e-01 -1.11549950e+00 -7.67698467e-01 -5.77436388e-01
2.60124385e-01 -5.93687713e-01 8.04142833e-01 6.60321951e-01
-1.31502247e+00 -4.94334579e-01 -1.20818615e+00 -8.18911195e-01
1.60764143e-01 -1.27602648e-02 1.13217485e+00 3.86260986e-01
-1.09572530e+00 6.63193405e-01 -9.35593069e-01 -1.05973259e-01
5.51642537e-01 4.73104894e-01 -4.46475685e-01 -9.48059857e-01
-8.46742392e-01 2.47399837e-01 2.15121686e-01 5.08486688e-01
-8.69681299e-01 -7.59538591e-01 -1.11593282e+00 -3.26634049e-01
2.43468642e-01 -1.06962717e+00 1.23282444e+00 -6.57014489e-01
-1.60857284e+00 4.95597720e-01 -4.41417873e-01 2.41509397e-02
6.00147307e-01 -2.70131826e-01 -3.85788875e-03 4.93390054e-01
1.45731553e-01 8.91072035e-01 9.61317301e-01 -1.10707796e+00
-1.05323660e+00 -8.95505011e-01 5.25546307e-03 5.51018238e-01
-3.34223360e-01 -7.64523447e-01 -6.47822559e-01 -2.69404292e-01
1.17182744e+00 -5.31651676e-01 -3.82588029e-01 2.78374553e-01
-2.23261714e-01 1.74734786e-01 7.32571602e-01 -4.88191187e-01
1.00725424e+00 -1.97615004e+00 3.11312139e-01 1.90239504e-01
7.22266793e-01 -5.55303752e-01 8.92669708e-02 7.72746429e-02
1.31568566e-01 -3.80625516e-01 -4.14214224e-01 -1.08598030e+00
-3.01779181e-01 1.63844198e-01 -9.62041467e-02 6.72328234e-01
-2.68998593e-01 7.97407627e-01 -1.06448472e+00 -4.20943558e-01
4.89604115e-01 8.21294785e-01 -6.44768238e-01 2.11146280e-01
3.09657827e-02 7.64475465e-01 -4.22267795e-01 1.11502850e+00
1.12385809e+00 -2.40014106e-01 -1.62793681e-01 -7.71687627e-01
-2.46770248e-01 2.84113824e-01 -1.19719601e+00 2.85162854e+00
-3.28068525e-01 3.41843039e-01 3.39662611e-01 -6.90590441e-01
6.18905544e-01 -1.53008355e-02 7.94287503e-01 -1.01986122e+00
2.61739284e-01 5.61388254e-01 -4.73585188e-01 -5.34102023e-01
7.48647451e-01 -1.87049434e-01 1.22834794e-01 1.18129373e-01
8.32390338e-02 -5.24364769e-01 -1.98038295e-01 3.25419664e-01
1.17949855e+00 3.44794840e-01 -4.12106663e-01 3.35414588e-01
4.30706918e-01 -2.42967129e-01 7.37674117e-01 4.35125172e-01
-1.62111089e-01 9.21085835e-01 9.76160765e-02 -2.74670392e-01
-7.20231950e-01 -9.66541469e-01 8.18386450e-02 4.62895006e-01
8.12977314e-01 -4.68931139e-01 -3.19091141e-01 -4.83775735e-01
-1.65708527e-01 -1.06194176e-01 -4.97891545e-01 -5.05389906e-02
-5.44373453e-01 -4.45902854e-01 2.26646647e-01 5.53343654e-01
1.17885661e+00 -6.02654696e-01 -6.02123260e-01 2.69507226e-02
-6.75915360e-01 -1.29596674e+00 -2.97069848e-01 1.39806405e-01
-1.44398928e+00 -1.01707447e+00 -9.35665011e-01 -7.29961336e-01
6.60933793e-01 7.80927896e-01 7.65696526e-01 -1.10497698e-01
-3.95515524e-02 5.24010360e-01 -3.42023134e-01 -1.10087328e-01
5.79776585e-01 1.94793288e-02 -1.90615639e-01 1.92660958e-01
4.99091595e-02 -8.14895630e-01 -1.23720849e+00 1.08450495e-01
-1.04426491e+00 4.95743871e-01 7.66374528e-01 6.91606283e-01
9.45798278e-01 -2.18810588e-02 -2.09822715e-03 -5.06486595e-01
-4.57799509e-02 -3.36753815e-01 -3.83383572e-01 -1.58528909e-01
-5.12695253e-01 -1.88621357e-02 2.00835820e-02 1.64453383e-03
-1.36281300e+00 4.62572098e-01 -2.56082803e-01 -5.41751504e-01
2.02070177e-01 5.12790322e-01 -5.14375687e-01 -4.86908913e-01
2.85602868e-01 2.99065858e-01 -4.50452380e-02 -4.78633374e-01
3.63993585e-01 4.19739485e-01 7.17973232e-01 -3.89001071e-01
7.37345397e-01 1.04383612e+00 2.90083200e-01 -6.03494048e-01
-8.78525496e-01 -6.33802176e-01 -2.97268480e-01 -3.95638168e-01
7.14983940e-01 -1.46886194e+00 -8.27543736e-01 8.80680859e-01
-1.19670904e+00 -2.10704774e-01 -3.01619828e-01 6.65670872e-01
-5.22155523e-01 6.53122306e-01 -7.07372487e-01 -5.79661489e-01
-3.57936233e-01 -1.18135023e+00 1.51968431e+00 4.97990727e-01
5.82364440e-01 -5.26856899e-01 -1.25715747e-01 7.25567997e-01
3.25469196e-01 2.59823054e-01 1.11418873e-01 6.77943468e-01
-1.26553679e+00 -9.30807889e-02 -3.92239541e-01 2.51016974e-01
3.66046652e-02 -5.41964173e-01 -1.39220548e+00 -1.85875639e-01
2.49447271e-01 -4.58729565e-01 1.22616208e+00 5.17547965e-01
1.34222221e+00 1.18881129e-01 -2.25558236e-01 1.34043086e+00
1.42576039e+00 -1.74637586e-01 9.16824639e-01 2.26682246e-01
1.40424407e+00 5.94750881e-01 7.21978724e-01 6.46109402e-01
8.23948324e-01 2.13492543e-01 9.64951277e-01 -1.13131225e-01
-1.08288556e-01 -4.44845259e-01 3.09599996e-01 8.19537520e-01
-2.56584764e-01 1.57815054e-01 -5.92959046e-01 2.91151494e-01
-1.96329856e+00 -6.82232261e-01 -3.28914881e-01 2.13572907e+00
5.23939848e-01 -4.52299342e-02 -2.54062206e-01 4.50029433e-01
1.47557467e-01 3.62252265e-01 -7.93081462e-01 4.08963442e-01
-4.41254050e-01 3.21609050e-01 7.55462527e-01 4.40095246e-01
-8.42704892e-01 5.32617092e-01 4.60730219e+00 7.41213620e-01
-1.07712197e+00 2.21793145e-01 5.26086688e-01 -3.89252067e-01
-5.83421707e-01 -5.70297055e-02 -5.91896415e-01 2.37320423e-01
3.78440589e-01 3.03156257e-01 4.08675134e-01 6.31182194e-01
1.00480348e-01 -7.32713938e-01 -1.02515912e+00 1.79612041e+00
1.61380753e-01 -1.28663981e+00 -2.21473873e-01 -2.95637306e-02
9.23302472e-01 3.09657186e-01 7.18331262e-02 -1.02698356e-01
-3.52958292e-01 -8.87987494e-01 8.94003808e-01 6.85760200e-01
1.05665088e+00 -7.34045804e-01 7.28946209e-01 3.22004050e-01
-1.48874879e+00 -1.99823186e-01 -3.30761015e-01 -2.65992582e-01
3.47504526e-01 1.18649411e+00 -8.97062570e-02 7.68593550e-01
1.08418965e+00 1.36592245e+00 -3.82071644e-01 1.08288097e+00
-3.58544320e-01 -1.87310986e-02 -4.91904527e-01 3.66723835e-01
1.59475043e-01 -1.25914946e-01 2.58496493e-01 6.52820945e-01
6.27184272e-01 2.10183978e-01 1.05573796e-03 6.08809888e-01
-6.38486296e-02 -2.75648892e-01 -4.69223082e-01 2.24833444e-01
1.31287292e-01 1.38865912e+00 -3.83208871e-01 -1.73031598e-01
-6.16820395e-01 1.21743560e+00 1.30418435e-01 3.55950594e-01
-6.40118301e-01 -2.04744041e-01 5.91890037e-01 3.53397727e-01
-2.55162083e-02 -3.18969876e-01 -7.53581941e-01 -1.47972322e+00
2.89222419e-01 -4.89151835e-01 2.11157963e-01 -1.07921696e+00
-1.03271568e+00 4.63381320e-01 -4.15602922e-01 -1.39176631e+00
2.09104761e-01 -5.15465140e-01 -2.08776250e-01 9.61906612e-01
-1.92741370e+00 -1.26794577e+00 -1.26740468e+00 1.16868317e+00
4.03183639e-01 3.66398215e-01 1.75783932e-01 6.24225318e-01
-4.90073651e-01 3.80843431e-01 -2.72475004e-01 -2.83795357e-01
6.21236265e-01 -9.51975822e-01 9.13301017e-03 8.16462100e-01
-4.56744760e-01 3.66727024e-01 2.05451861e-01 -7.81541288e-01
-2.00141072e+00 -9.78986084e-01 6.63094759e-01 1.52060473e-02
3.25583756e-01 -1.86035335e-01 -4.30933028e-01 4.53905374e-01
-1.59331694e-01 3.26431125e-01 6.99621961e-02 -3.31144840e-01
-2.12855816e-01 -5.20791769e-01 -1.26584888e+00 2.70228237e-01
1.41531098e+00 -6.47053540e-01 -3.75142619e-02 3.51466119e-01
8.74572873e-01 -9.47603524e-01 -7.21559227e-01 6.81605279e-01
6.60341084e-01 -1.50041604e+00 1.09420109e+00 4.81869638e-01
7.97002196e-01 -6.04031563e-01 -4.39679593e-01 -8.32409024e-01
8.05216357e-02 -6.39941335e-01 -4.22731936e-01 7.92901218e-01
-1.06418645e-03 -4.79993492e-01 1.24308264e+00 7.18712330e-01
-4.66298848e-01 -8.04184914e-01 -1.15659809e+00 -1.66295946e-01
-5.64020336e-01 -8.16603065e-01 4.36607748e-01 5.33730268e-01
7.69614475e-03 2.41494849e-01 -5.22265851e-01 2.02747881e-01
1.08007419e+00 -7.47217610e-02 7.88240731e-01 -1.03557241e+00
-2.79092789e-01 -8.71008709e-02 -4.38348264e-01 -1.92892849e+00
-3.69896084e-01 -5.92940509e-01 2.09882140e-01 -1.78491592e+00
2.93881036e-02 -4.41652507e-01 -8.33319724e-02 2.60349959e-01
-1.13617741e-01 4.55750018e-01 2.70769019e-02 9.55166072e-02
-7.09412694e-01 9.24701750e-01 1.52669048e+00 -3.03851128e-01
-1.00507051e-01 -1.92754686e-01 -6.07743740e-01 7.15592086e-01
4.47623819e-01 -2.27398738e-01 -6.72509432e-01 -7.26318419e-01
6.47742391e-01 3.26499730e-01 4.06373680e-01 -1.30929935e+00
5.83969057e-01 9.59686041e-02 5.62498450e-01 -9.73603249e-01
8.32163692e-01 -1.03461921e+00 1.79913230e-02 3.49880099e-01
2.89657325e-01 -1.15191981e-01 -2.33300757e-02 6.94070756e-01
-4.67198759e-01 3.27150047e-01 5.57848215e-01 -2.83368915e-01
-9.16902363e-01 9.23038244e-01 3.63792270e-01 -2.56262809e-01
6.62188232e-01 -5.89457273e-01 -3.00314367e-01 -4.28939104e-01
-4.27553505e-01 3.49743843e-01 3.84427637e-01 1.70052215e-01
1.34778070e+00 -1.25183105e+00 -3.92006457e-01 4.09973949e-01
1.33801103e-01 8.31817329e-01 8.68134856e-01 1.27548146e+00
-6.25094414e-01 6.26455992e-02 -9.20524225e-02 -9.35541034e-01
-8.43262076e-01 1.09477580e-01 3.71805012e-01 -1.03793971e-01
-7.66901791e-01 1.35540974e+00 4.10614014e-01 -3.05114448e-01
4.41403627e-01 -6.83305383e-01 1.28145158e-01 -7.64958709e-02
7.28546679e-01 5.34543276e-01 1.61543921e-01 -4.59861100e-01
-3.26152295e-01 8.72493446e-01 2.30477259e-01 -1.92484975e-01
1.47316766e+00 -5.90676248e-01 -2.42302820e-01 4.25627053e-01
1.15731454e+00 -2.86369950e-01 -1.72684407e+00 -3.53493661e-01
-6.62442207e-01 -8.61235678e-01 4.68410432e-01 -5.97515106e-01
-1.56782842e+00 1.06008077e+00 6.65840864e-01 -4.24041450e-01
1.64539564e+00 -2.59504884e-01 1.25548279e+00 1.47777379e-01
8.18014920e-01 -8.79568517e-01 1.69153348e-01 5.48480928e-01
7.20018506e-01 -1.55987024e+00 8.71569812e-02 -7.20838785e-01
-2.47296497e-01 1.08260405e+00 8.06797743e-01 9.47972164e-02
5.81862688e-01 1.75021261e-01 -2.11404607e-01 -2.87642658e-01
-2.61218101e-01 -1.71997905e-01 2.02196062e-01 6.82240605e-01
1.32140294e-01 -1.51391685e-01 -1.12917513e-01 5.96657813e-01
-5.38683729e-03 2.73260742e-01 3.39821994e-01 9.85560119e-01
-4.23757017e-01 -8.36340368e-01 -4.22399998e-01 4.11162376e-01
-2.96164136e-02 -3.90328795e-01 -1.84561126e-02 4.40799922e-01
5.10503590e-01 1.03669786e+00 -1.86792478e-01 -7.33231962e-01
3.32583696e-01 -5.31561136e-01 8.31835866e-01 -3.52616638e-01
-2.98524916e-01 2.30414316e-01 -1.84030265e-01 -1.44586742e+00
-7.45464683e-01 -5.63100636e-01 -1.34081507e+00 -5.68401515e-01
3.05812638e-02 -3.52490813e-01 9.62031186e-01 8.40483010e-01
2.20709145e-01 4.65704143e-01 4.76863921e-01 -1.47673965e+00
9.93390679e-02 -8.69252920e-01 -8.06811094e-01 7.70226195e-02
5.69896281e-01 -6.18263066e-01 -4.33242738e-01 -2.54155695e-01] | [8.927861213684082, -2.55668306350708] |
ba49e594-e032-47aa-a78f-2813f897dc14 | discrete-time-nonlinear-feedback | 2303.08884 | null | https://arxiv.org/abs/2303.08884v1 | https://arxiv.org/pdf/2303.08884v1.pdf | Discrete-Time Nonlinear Feedback Linearization via Physics-Informed Machine Learning | We present a physics-informed machine learning (PIML) scheme for the feedback linearization of nonlinear discrete-time dynamical systems. The PIML finds the nonlinear transformation law, thus ensuring stability via pole placement, in one step. In order to facilitate convergence in the presence of steep gradients in the nonlinear transformation law, we address a greedy-wise training procedure. We assess the performance of the proposed PIML approach via a benchmark nonlinear discrete map for which the feedback linearization transformation law can be derived analytically; the example is characterized by steep gradients, due to the presence of singularities, in the domain of interest. We show that the proposed PIML outperforms, in terms of numerical approximation accuracy, the traditional numerical implementation, which involves the construction--and the solution in terms of the coefficients of a power-series expansion--of a system of homological equations as well as the implementation of the PIML in the entire domain, thus highlighting the importance of continuation techniques in the training procedure of PIML. | ['Ioannis G. Kevrekidis', 'Constantinos Siettos', 'Nikolaos Kazantzis', 'Gianluca Fabiani', 'Hector Vargas Alvarez'] | 2023-03-15 | null | null | null | null | ['physics-informed-machine-learning'] | ['graphs'] | [-8.64476413e-02 2.61619002e-01 1.77718893e-01 3.90088081e-01
-4.64426041e-01 -6.84894800e-01 7.94932485e-01 5.83692253e-01
-1.43792018e-01 6.96285665e-01 -3.38105768e-01 -7.15459704e-01
-3.59646082e-01 -5.53250015e-01 -1.15096068e+00 -6.86779797e-01
-2.11665183e-01 6.24309123e-01 9.78350863e-02 -6.87508285e-01
2.81300336e-01 6.51436448e-01 -1.01114798e+00 -4.45844293e-01
9.63654697e-01 8.34879637e-01 -2.73368418e-01 6.96015894e-01
1.99517131e-01 3.20553839e-01 -8.88196081e-02 -1.02621634e-02
5.41792095e-01 -5.04915416e-01 -8.26737583e-01 -1.49909124e-01
4.11354572e-01 -6.72307163e-02 -1.11667834e-01 8.96293521e-01
2.03831270e-01 9.60900262e-02 1.04224956e+00 -1.00329375e+00
-7.42847547e-02 8.12840164e-02 -2.43409723e-01 6.14536880e-03
1.88018844e-01 6.45649880e-02 8.26742291e-01 -9.57042813e-01
7.46479154e-01 7.82839417e-01 9.42241430e-01 -1.38804555e-01
-1.48691308e+00 -1.55347968e-02 -4.31005597e-01 4.98561189e-02
-1.26760662e+00 -1.94333076e-01 9.50171769e-01 -8.46854746e-01
6.45784914e-01 -1.25746965e-01 6.11277997e-01 2.53822654e-01
5.83677828e-01 1.39919519e-01 8.99105489e-01 -8.85335267e-01
5.73987961e-01 2.12486729e-01 1.45805459e-02 8.92211378e-01
1.59988940e-01 3.33424807e-01 -7.40261525e-02 -4.18583214e-01
8.56311858e-01 -6.18839800e-01 -1.28591627e-01 -8.26600194e-01
-9.75307941e-01 7.16628194e-01 6.49624944e-01 2.74866134e-01
-3.05294842e-01 -8.61154646e-02 2.54741371e-01 5.49340069e-01
4.72892791e-01 7.82238543e-01 -1.33927062e-01 1.06269568e-01
-1.05789304e+00 3.26798469e-01 1.12661970e+00 6.18797421e-01
9.70490694e-01 6.37106448e-02 4.60509032e-01 1.56099245e-01
-1.41705334e-01 9.23103988e-02 4.37051117e-01 -8.56760979e-01
1.44705296e-01 7.75552928e-01 3.11613530e-01 -1.00225830e+00
-5.09942055e-01 -3.93245578e-01 -9.62707877e-01 5.01169205e-01
6.91440880e-01 -1.23623058e-01 -1.85544819e-01 1.47285271e+00
7.58206666e-01 5.77457920e-02 2.45273501e-01 6.60982668e-01
1.13202065e-01 9.26635146e-01 -2.28830710e-01 -4.88651723e-01
5.33117354e-01 -4.33584273e-01 -1.88395917e-01 3.12630504e-01
8.78822684e-01 -5.23708344e-01 1.03564990e+00 3.04989666e-01
-1.19262183e+00 -5.15297115e-01 -1.39119768e+00 8.66525527e-03
-2.36171648e-01 3.83069456e-01 1.62703156e-01 -2.30146572e-02
-1.06905448e+00 1.24857903e+00 -6.89640343e-01 -1.52944371e-01
-2.68028766e-01 3.82621169e-01 -4.09090906e-01 7.25803137e-01
-1.03149772e+00 1.06571865e+00 1.50409371e-01 2.30672777e-01
-3.94664973e-01 -1.20042944e+00 -5.26944160e-01 1.96628287e-01
1.80305079e-01 -4.03709710e-01 9.89638805e-01 -8.50820720e-01
-1.52474248e+00 6.69933498e-01 -9.31378230e-02 -5.28543591e-01
1.02923894e+00 1.59147370e-03 2.18709096e-01 2.64423281e-01
-2.18965337e-01 3.32828403e-01 9.96752739e-01 -9.00708258e-01
-8.71893670e-03 9.56175029e-02 2.10886106e-01 -2.13648807e-02
-1.91630706e-01 -4.73850369e-01 2.77713150e-01 -3.51494163e-01
9.39771384e-02 -9.85306680e-01 -2.57992864e-01 1.75375938e-01
-7.74918348e-02 -1.67535037e-01 6.07316971e-01 -7.57773936e-01
7.34834135e-01 -1.84097731e+00 6.01816356e-01 4.65341926e-01
5.80244623e-02 2.63279200e-01 5.40806770e-01 7.91741192e-01
-3.76564026e-01 -1.50742874e-01 -3.91114771e-01 -3.22357193e-02
-1.60265416e-01 -1.93337709e-01 -6.15262926e-01 8.55266988e-01
3.84403199e-01 5.49531579e-01 -6.88635886e-01 -4.61890012e-01
3.43738467e-01 3.79165322e-01 -6.92975760e-01 -4.56590615e-02
-4.50045377e-01 7.58163512e-01 -1.90966994e-01 -9.04724523e-02
4.67096210e-01 -1.39707446e-01 1.60828978e-01 -3.02030981e-01
-5.00049174e-01 1.64289162e-01 -1.13041210e+00 1.23636770e+00
-6.30253911e-01 4.56505328e-01 1.79264620e-01 -1.41514492e+00
8.98123920e-01 3.19166571e-01 6.24738693e-01 -3.04841667e-01
2.18861416e-01 5.60214698e-01 -1.24810547e-01 -1.87225536e-01
-1.59515887e-02 2.70632803e-02 2.68539581e-02 4.58797291e-02
6.22796640e-02 -1.60076052e-01 2.31514066e-01 2.08228350e-01
7.29330778e-01 2.26567328e-01 5.45690298e-01 -8.41164589e-01
1.09132969e+00 3.70412350e-01 7.52971768e-02 2.70816624e-01
2.26681724e-01 9.74332020e-02 8.50884259e-01 -3.75888109e-01
-1.30387580e+00 -7.61928678e-01 -2.85093188e-01 4.94295478e-01
5.95547594e-02 -1.73666537e-01 -8.83397937e-01 1.23504110e-01
-1.84838977e-02 4.44678605e-01 -4.05384183e-01 -2.53562748e-01
-1.12461805e+00 -3.26122761e-01 1.38973668e-02 -4.86336686e-02
3.63261968e-01 -8.21928263e-01 -6.14815354e-01 1.41247749e-01
2.89254695e-01 -9.55701709e-01 -1.23621412e-01 1.26807481e-01
-1.02984214e+00 -1.11263609e+00 -5.64125657e-01 -1.04770827e+00
6.65851653e-01 -5.45441091e-01 5.63005984e-01 7.79411942e-02
-3.07271063e-01 1.16720721e-01 2.25795120e-01 1.10494345e-01
-1.05057609e+00 2.74375737e-01 8.15228894e-02 -1.50136784e-01
-5.04652023e-01 -9.07365620e-01 -3.61987174e-01 1.99316338e-01
-5.95669568e-01 1.96132749e-01 2.32411250e-01 7.05142319e-01
3.97492200e-01 -8.53012726e-02 6.89793706e-01 -2.06433609e-01
5.54449797e-01 -3.66998792e-01 -1.40372944e+00 1.93902045e-01
-5.17407835e-01 3.34193379e-01 1.31286108e+00 -5.71406126e-01
-4.62602466e-01 4.19630110e-01 5.01140393e-02 -2.35149533e-01
2.02941060e-01 5.46026111e-01 2.89482147e-01 -9.34252858e-01
9.07367527e-01 2.24774808e-01 3.04555148e-01 -5.29186785e-01
3.13668579e-01 2.10829929e-01 9.26937640e-01 -8.13731194e-01
8.51202965e-01 7.79832155e-02 9.00125206e-01 -9.90420520e-01
-2.82350510e-01 -3.06739926e-01 -7.50370443e-01 -3.10985386e-01
1.37023658e-01 -4.95732009e-01 -9.23370719e-01 4.15176094e-01
-1.04577184e+00 -7.64967129e-02 -3.30148041e-01 2.17072025e-01
-9.20643687e-01 4.31512862e-01 -6.92993164e-01 -8.24182153e-01
-2.58930653e-01 -1.08102572e+00 7.01701462e-01 -9.95126590e-02
-1.04599759e-01 -1.05383146e+00 2.47521579e-01 -3.47505569e-01
2.66305029e-01 4.36346710e-01 1.37862325e+00 -3.82770747e-01
-3.84705186e-01 -4.65247989e-01 2.59163171e-01 7.07369968e-02
-3.49087954e-01 1.52646169e-01 -6.15293264e-01 -4.40305561e-01
3.09431851e-01 -8.46565217e-02 2.66858399e-01 3.23675901e-01
4.25788999e-01 -4.29995120e-01 -1.14464082e-01 5.12338758e-01
1.60638440e+00 -8.86797085e-02 -1.06332906e-01 1.09836750e-01
3.15257013e-01 5.65909922e-01 1.28069580e-01 7.33828619e-02
8.52599740e-02 8.29768419e-01 1.82856202e-01 3.36599462e-02
5.73723689e-02 -3.45596641e-01 2.95602024e-01 8.08440626e-01
7.79891238e-02 6.53903484e-01 -9.38008487e-01 4.69150096e-01
-1.86646485e+00 -3.92718405e-01 -7.42809698e-02 2.40004301e+00
7.71954656e-01 1.00080259e-01 5.32997012e-01 4.84286636e-01
4.97266293e-01 -3.51626635e-01 -4.10614699e-01 -6.37508392e-01
2.33352467e-01 7.95393065e-02 4.51431662e-01 1.12360716e+00
-8.24937046e-01 5.52260697e-01 6.28542042e+00 6.19004607e-01
-1.47603750e+00 -1.79345623e-01 1.63463578e-01 4.94272262e-01
2.28088304e-01 2.47440740e-01 -6.78152919e-01 3.36205900e-01
9.00511563e-01 -4.24901903e-01 4.65618372e-01 8.25881779e-01
4.79512602e-01 -6.81053549e-02 -1.00910962e+00 4.25124019e-01
-3.23684663e-01 -1.55646586e+00 -2.08861530e-01 -3.12405135e-02
6.33562744e-01 -4.97199714e-01 -1.63310811e-01 -4.37519774e-02
-3.79440874e-01 -4.05446678e-01 7.79089630e-01 4.41728950e-01
2.71762669e-01 -6.25952065e-01 2.71619678e-01 7.42818475e-01
-9.47555542e-01 -3.41829769e-02 -1.33360639e-01 -2.69355088e-01
4.63614762e-02 5.20504773e-01 -8.33259404e-01 5.91226935e-01
5.63637353e-02 3.48110735e-01 -4.45683807e-01 1.06074774e+00
1.59638971e-01 5.92456222e-01 -6.58101916e-01 -2.59902954e-01
1.57126218e-01 -8.28835487e-01 7.20119119e-01 8.79218996e-01
2.43644178e-01 -3.91587056e-02 1.20846555e-01 9.22029436e-01
1.97376996e-01 4.34749782e-01 -3.17848146e-01 1.56164914e-01
-6.38579056e-02 1.05487597e+00 -5.80942690e-01 -8.46275985e-02
1.75155103e-01 3.96121800e-01 3.60414088e-01 3.17782581e-01
-4.34599072e-01 -3.35676730e-01 1.04778640e-01 2.84727246e-01
3.77096683e-01 -5.98068476e-01 -3.99512291e-01 -8.51734042e-01
3.89138132e-01 -5.01825750e-01 3.77934165e-02 -5.38518369e-01
-6.50103867e-01 5.11174619e-01 3.44626963e-01 -1.14395630e+00
-9.53457534e-01 -6.69494331e-01 -7.22393751e-01 9.10901427e-01
-1.18015182e+00 -9.12316024e-01 1.97573349e-01 3.51633191e-01
-1.08324893e-01 -2.32493673e-02 6.53745592e-01 1.58557415e-01
-1.48990095e-01 2.69800186e-01 3.54245156e-01 -3.69640470e-01
1.90625325e-01 -1.09939969e+00 1.25041738e-01 7.78170466e-01
-3.53766590e-01 4.62421656e-01 1.17401111e+00 -5.67934155e-01
-1.34492064e+00 -5.44153452e-01 1.09002280e+00 8.64150077e-02
9.91440654e-01 -2.02559084e-01 -1.01237118e+00 3.76367629e-01
-1.87213004e-01 -2.32976750e-01 -2.01949984e-01 -2.16921911e-01
2.07307756e-01 -2.73264408e-01 -1.22731137e+00 6.79382384e-01
4.05330539e-01 -4.06509101e-01 -4.66086477e-01 6.13765597e-01
5.40041506e-01 -4.51605588e-01 -8.20027351e-01 3.20349097e-01
3.89528126e-01 -4.71530378e-01 1.05752349e+00 -4.53063369e-01
4.98255491e-01 -5.00346303e-01 4.26879644e-01 -1.14523017e+00
-1.69844389e-01 -1.16788900e+00 -2.20258072e-01 8.17081451e-01
2.05895498e-01 -7.18374074e-01 4.89266127e-01 2.02474654e-01
-1.35012409e-02 -8.95609617e-01 -1.08536100e+00 -6.41600668e-01
4.78690147e-01 1.02752507e-01 -7.28438348e-02 5.54011166e-01
4.52026695e-01 2.07016483e-01 -3.17514479e-01 2.75302798e-01
5.25057077e-01 2.39161268e-01 6.15379691e-01 -1.02918446e+00
-5.09400725e-01 -3.76861751e-01 -5.75529158e-01 -7.56059587e-01
2.09875539e-01 -7.15781569e-01 -8.79821405e-02 -7.69595802e-01
-6.68079436e-01 -3.30108792e-01 1.60589382e-01 1.28873393e-01
1.95626706e-01 -1.39317170e-01 3.92022394e-02 4.75030333e-01
1.58913940e-01 3.56794506e-01 8.60564411e-01 4.23002392e-01
-5.16911864e-01 1.03617638e-01 -1.01586975e-01 8.20982158e-01
6.18688166e-01 -1.41907707e-01 -2.72325516e-01 3.03420901e-01
5.30684233e-01 3.26131642e-01 6.50677741e-01 -1.26596725e+00
5.81705511e-01 2.06008866e-01 1.75864905e-01 -4.89804178e-01
2.92465597e-01 -5.32342553e-01 2.45521709e-01 7.37829387e-01
-4.38386351e-01 2.35945553e-01 2.50953376e-01 2.28783995e-01
-8.73297378e-02 -5.08804917e-01 9.17449117e-01 2.25370094e-01
-1.50780618e-01 -2.50386745e-01 -4.88814354e-01 -1.70154929e-01
6.23137534e-01 7.39731863e-02 1.67238135e-02 -4.21316832e-01
-6.63712084e-01 5.11424094e-02 4.78747159e-01 -2.99268633e-01
-6.69964850e-02 -9.57869053e-01 -2.84231156e-01 3.74630630e-01
-2.99880147e-01 -1.35361761e-01 -1.58183098e-01 1.05470228e+00
-1.03164625e+00 4.18593973e-01 -2.48930216e-01 -5.95917940e-01
-7.45596588e-01 6.27377808e-01 8.24156523e-01 -2.55129337e-01
-6.77044332e-01 9.52251330e-02 -1.31554067e-01 -3.64827812e-01
1.34032562e-01 -5.02006233e-01 1.84761837e-01 -2.89340496e-01
9.43247322e-03 4.49509382e-01 3.85896206e-01 -5.61455965e-01
2.39361357e-02 9.34577048e-01 4.80623543e-01 -2.11945847e-01
1.01319575e+00 1.88592896e-01 -3.65170896e-01 3.94026995e-01
1.22589767e+00 -6.28993511e-02 -1.06628621e+00 -2.12198585e-01
-1.49142861e-01 1.44770637e-01 -1.17408611e-01 -3.32415611e-01
-3.80834401e-01 6.74928963e-01 3.62379164e-01 3.93671900e-01
8.88827622e-01 -3.99032861e-01 4.25895035e-01 6.77130580e-01
7.62592480e-02 -8.99890900e-01 -4.49351072e-01 4.75559741e-01
9.74724174e-01 -8.36329520e-01 1.03873827e-01 -5.80850661e-01
1.14266962e-01 1.51068938e+00 1.48073733e-01 -9.33447719e-01
9.93046045e-01 5.24495058e-02 -2.88337111e-01 2.88675576e-01
-3.79445404e-01 4.16119844e-01 5.61097801e-01 5.36070392e-02
1.32349446e-01 -3.34118426e-01 -7.12169766e-01 -1.22260988e-01
-5.13572633e-01 1.66371793e-01 4.42250192e-01 7.07993209e-01
-3.65405947e-01 -9.11785841e-01 -2.84227252e-01 -4.77808155e-02
6.49212450e-02 2.43870109e-01 -5.08657038e-01 9.44424331e-01
-1.19604729e-01 4.60134476e-01 -1.78556636e-01 8.15542042e-02
2.13357449e-01 3.48061562e-01 6.51296198e-01 1.94355145e-01
-5.83432794e-01 2.54820939e-02 -1.44004405e-01 -1.86336607e-01
-1.30723059e-01 -3.32738519e-01 -1.16379356e+00 -5.04683852e-01
-8.45703781e-02 4.15550828e-01 7.72098958e-01 1.25251615e+00
2.49125049e-01 6.99178576e-02 6.28103554e-01 -1.16457176e+00
-1.14738381e+00 -6.52800918e-01 -2.14931861e-01 2.09097981e-01
6.14550292e-01 -5.90553045e-01 -5.87316334e-01 -2.40552053e-02] | [6.511131286621094, 3.471754312515259] |
a9fe2031-f636-4528-942c-5ec9315f0268 | domain-adaptation-in-multi-channel | 1907.04048 | null | https://arxiv.org/abs/1907.04048v1 | https://arxiv.org/pdf/1907.04048v1.pdf | Domain Adaptation in Multi-Channel Autoencoder based Features for Robust Face Anti-Spoofing | While the performance of face recognition systems has improved significantly in the last decade, they are proved to be highly vulnerable to presentation attacks (spoofing). Most of the research in the field of face presentation attack detection (PAD), was focused on boosting the performance of the systems within a single database. Face PAD datasets are usually captured with RGB cameras, and have very limited number of both bona-fide samples and presentation attack instruments. Training face PAD systems on such data leads to poor performance, even in the closed-set scenario, especially when sophisticated attacks are involved. We explore two paths to boost the performance of the face PAD system against challenging attacks. First, by using multi-channel (RGB, Depth and NIR) data, which is still easily accessible in a number of mass production devices. Second, we develop a novel Autoencoders + MLP based face PAD algorithm. Moreover, instead of collecting more data for training of the proposed deep architecture, the domain adaptation technique is proposed, transferring the knowledge of facial appearance from RGB to multi-channel domain. We also demonstrate, that learning the features of individual facial regions, is more discriminative than the features learned from an entire face. The proposed system is tested on a very recent publicly available multi-channel PAD database with a wide variety of presentation attacks. | ['Olegs Nikisins', 'Anjith George', 'Sebastien Marcel'] | 2019-07-09 | null | null | null | null | ['face-presentation-attack-detection'] | ['computer-vision'] | [ 1.92186147e-01 -1.83968306e-01 1.72216073e-01 -1.99156970e-01
-5.36747754e-01 -7.23832846e-01 5.75774252e-01 -3.54166240e-01
-2.79810011e-01 3.52209359e-01 -6.29483238e-02 -3.58021520e-02
2.24612262e-02 -7.43811965e-01 -7.15532720e-01 -9.88378346e-01
-1.59024298e-01 1.29245192e-01 7.17377290e-02 -3.14221352e-01
-9.95768830e-02 1.09841359e+00 -1.89029706e+00 4.72538143e-01
4.52766158e-02 1.56528556e+00 -3.17483127e-01 6.68956935e-01
1.40296161e-01 3.40029538e-01 -8.60153139e-01 -8.57947409e-01
5.66632748e-01 -1.84924841e-01 -4.07384634e-01 -1.14402203e-02
6.51703596e-01 -7.82137990e-01 -5.42086065e-01 9.50013220e-01
8.71158123e-01 -2.99601406e-01 5.57429016e-01 -1.47706389e+00
-4.96091574e-01 4.12559696e-02 -6.03847146e-01 1.44170284e-01
4.25989658e-01 1.58440828e-01 2.40108162e-01 -8.53454709e-01
4.05161321e-01 1.31787896e+00 5.29883802e-01 7.76752532e-01
-8.33797812e-01 -1.23709345e+00 -3.98924172e-01 3.05403084e-01
-1.35395408e+00 -5.81907153e-01 1.06831694e+00 -2.72509128e-01
5.55735350e-01 -1.15354974e-02 3.39406371e-01 1.74420071e+00
8.91020298e-02 3.23761791e-01 1.31889319e+00 -3.40071768e-01
1.16516687e-01 5.19436240e-01 -3.03075939e-01 5.91542125e-01
2.55168021e-01 2.99109876e-01 -4.91662323e-01 -3.07451189e-01
6.84621513e-01 9.81152952e-02 -2.46135384e-01 -1.07867949e-01
-2.23923430e-01 9.23555255e-01 2.42374063e-01 4.01796252e-01
-3.98451477e-01 -2.30210528e-01 4.58137810e-01 3.72011065e-01
2.11219206e-01 -6.45274669e-02 -3.27388674e-01 4.91897985e-02
-8.25309753e-01 4.26320545e-02 8.20586801e-01 5.50367355e-01
6.74618900e-01 5.85009828e-02 2.27114931e-01 6.55470014e-01
3.74231577e-01 8.26951504e-01 4.58122551e-01 -3.27505141e-01
4.89396572e-01 2.26056203e-01 -1.52836904e-01 -1.46349156e+00
-3.17852855e-01 8.63226578e-02 -8.70102167e-01 4.44623679e-01
5.28109610e-01 -3.57162803e-01 -7.48533905e-01 1.54730916e+00
5.28570473e-01 2.77730346e-01 7.42504373e-02 7.46125102e-01
7.11514771e-01 6.88105822e-01 1.56373959e-02 -2.15347409e-01
1.46155787e+00 -3.26176047e-01 -7.31263638e-01 1.81535199e-01
2.69888133e-01 -9.30378914e-01 7.58836925e-01 8.64287734e-01
-6.99679494e-01 -7.60561407e-01 -1.19997048e+00 2.69065887e-01
-7.30667412e-01 1.29777640e-01 4.13258165e-01 1.43984699e+00
-1.03388643e+00 4.47708428e-01 -5.16384184e-01 -3.21352184e-01
8.73930633e-01 8.08961213e-01 -9.43454444e-01 -2.45458841e-01
-1.18622768e+00 7.34562695e-01 5.38461059e-02 1.70046523e-01
-1.16957819e+00 -4.84560877e-01 -6.27105355e-01 5.74726649e-02
8.99954513e-02 1.73603281e-01 7.65125453e-01 -1.18013728e+00
-1.88792562e+00 7.82864094e-01 3.03084463e-01 -2.76129872e-01
3.58789325e-01 -3.65889192e-01 -9.04084265e-01 5.12329996e-01
-7.08527207e-01 3.74440759e-01 1.59759271e+00 -1.11811244e+00
-2.11810246e-01 -7.61152983e-01 -3.67712439e-03 -4.96884495e-01
-1.06167793e+00 5.00913501e-01 -1.02755502e-01 -3.56330454e-01
-5.79307914e-01 -9.72968102e-01 3.36344898e-01 2.81936049e-01
-7.90182576e-02 -9.99297947e-03 1.55118179e+00 -9.46996033e-01
7.65643775e-01 -2.35024047e+00 -1.82833984e-01 3.32088828e-01
-2.48142108e-01 1.02464855e+00 -1.68313146e-01 4.18998510e-01
-3.81341189e-01 -1.92563012e-02 2.28925161e-02 -2.47060925e-01
-2.02543318e-01 2.20134899e-01 -4.79545385e-01 7.31310546e-01
4.96793687e-01 2.02068627e-01 -2.43024781e-01 -3.47171128e-01
3.03711057e-01 9.60110068e-01 -5.41999459e-01 4.48390156e-01
2.16625884e-01 4.87272561e-01 -3.59080672e-01 7.65593588e-01
1.39283240e+00 2.17227370e-01 -6.21297956e-02 -2.65140176e-01
2.84603745e-01 -2.68975258e-01 -1.26348281e+00 1.31881368e+00
-4.24244076e-01 6.01910174e-01 2.88699806e-01 -9.36913431e-01
9.97106493e-01 7.15609908e-01 5.00275075e-01 -6.47569060e-01
5.31482816e-01 1.66208800e-02 -1.22239769e-01 -6.54837430e-01
-4.32042032e-02 1.90066714e-02 1.94346324e-01 5.09955883e-01
4.83813018e-01 3.27091664e-01 -3.18100393e-01 -2.46667475e-01
9.93035555e-01 -9.93788987e-02 2.36656163e-02 1.57817170e-01
9.26268280e-01 -5.40505290e-01 1.93652049e-01 2.81330794e-01
-4.41640645e-01 4.24752533e-01 3.69698286e-01 -3.74273181e-01
-8.64927411e-01 -1.01069272e+00 -3.33061963e-01 9.13168252e-01
-5.27388901e-02 -2.34293669e-01 -1.06398225e+00 -8.67419600e-01
1.16417207e-01 -1.63633674e-01 -6.55749142e-01 -2.40911439e-01
-6.57114983e-01 -8.22168648e-01 1.20046604e+00 4.64845419e-01
7.45688856e-01 -1.01411319e+00 -6.87476754e-01 -3.29102650e-02
2.90605426e-01 -1.38037288e+00 1.98907062e-01 -8.60771984e-02
-4.45656091e-01 -1.17323184e+00 -7.50323355e-01 -6.34654582e-01
4.11006242e-01 -8.42913538e-02 4.76940155e-01 -7.05085602e-03
-7.05736220e-01 5.71430206e-01 -3.85732591e-01 -7.17737913e-01
-2.33414158e-01 -4.27524328e-01 4.45783675e-01 5.95541954e-01
5.58132529e-01 -5.30183613e-01 -5.84477365e-01 3.30714047e-01
-9.53519046e-01 -8.74893665e-01 7.37002134e-01 7.70218432e-01
2.80809589e-02 2.71643251e-01 4.31874871e-01 -6.06326938e-01
4.08280909e-01 -4.17304575e-01 -5.65931022e-01 6.54464886e-02
-7.24246278e-02 -3.45766097e-01 7.06745625e-01 -7.78014839e-01
-1.15657794e+00 1.18039429e-01 -5.52283406e-01 -6.40075564e-01
-6.44504905e-01 -1.50232658e-01 -5.37797213e-01 -6.96531355e-01
8.31820369e-01 7.23206177e-02 4.05131519e-01 -4.88229305e-01
9.58586484e-02 9.78902578e-01 3.77154469e-01 -4.51611608e-01
1.08143699e+00 6.90954149e-01 1.28874063e-01 -1.15486133e+00
3.39532197e-02 -2.61129528e-01 -5.88027179e-01 -3.42863768e-01
6.30009353e-01 -7.91719139e-01 -1.07780802e+00 8.91513824e-01
-1.04035091e+00 9.41919014e-02 3.81439418e-01 3.43755394e-01
-2.47744754e-01 5.95663011e-01 -6.51892304e-01 -1.01151717e+00
-2.80912220e-01 -1.24781108e+00 1.08665431e+00 3.73592824e-01
2.63665140e-01 -6.12036347e-01 -2.66647022e-02 1.39650419e-01
5.59735775e-01 3.64670783e-01 5.85408092e-01 -7.83886373e-01
-2.74380565e-01 -6.49443746e-01 -1.86522663e-01 7.34951377e-01
2.75177181e-01 1.71531975e-01 -1.58503795e+00 -4.92869198e-01
3.36597592e-01 -5.59888661e-01 4.13643956e-01 -9.02128220e-02
1.47079182e+00 -3.83741081e-01 -2.70991474e-01 5.85778594e-01
1.43355274e+00 2.92256832e-01 8.29463720e-01 1.81245301e-02
4.59145784e-01 8.75859022e-01 3.28716397e-01 6.66910887e-01
-2.29637757e-01 8.13570678e-01 5.57052791e-01 -3.94390011e-03
-2.79636849e-02 1.16743855e-02 7.60516942e-01 1.60653755e-01
-8.44101515e-03 -2.28720441e-01 -6.26267314e-01 1.05220810e-01
-1.22235382e+00 -9.12966490e-01 4.70517606e-01 2.10920596e+00
4.06529963e-01 -2.23844126e-01 4.38891053e-01 6.56500697e-01
6.69159710e-01 -1.08871721e-01 -1.49238974e-01 -5.11145473e-01
-1.55168921e-01 7.60408700e-01 3.91837060e-01 -2.16012448e-03
-1.25997198e+00 7.13973343e-01 5.62705183e+00 7.05873311e-01
-1.70992374e+00 1.64873809e-01 4.37790126e-01 6.37582317e-02
5.26169658e-01 -5.56215346e-01 -9.57946658e-01 6.46894038e-01
1.24353862e+00 4.88641173e-01 3.72142851e-01 1.07969487e+00
-3.07658881e-01 1.36669859e-01 -9.88704562e-01 1.46056914e+00
4.65113848e-01 -9.13854599e-01 1.46867894e-02 3.51489067e-01
1.97088704e-01 -4.43421394e-01 6.87867403e-01 5.39182723e-02
-1.38617069e-01 -1.19339979e+00 1.87252805e-01 1.01018205e-01
7.49427795e-01 -1.01773798e+00 8.37759137e-01 1.10425197e-01
-8.04267049e-01 -4.45686519e-01 -6.55070543e-01 2.65896082e-01
-2.69513965e-01 2.21518919e-01 -7.95565724e-01 4.05675054e-01
1.02249396e+00 1.73887715e-01 -6.10126376e-01 6.49966478e-01
-8.09546746e-03 5.71856678e-01 -5.63003361e-01 1.00818150e-01
2.36903206e-02 1.79646909e-01 2.35858321e-01 1.23031807e+00
4.46097821e-01 9.85025540e-02 -1.79588139e-01 3.08470428e-01
-3.46065611e-02 1.52732328e-01 -1.02332044e+00 -7.36018643e-02
1.90576687e-01 1.48819733e+00 -3.21505696e-01 5.62938191e-02
-5.12817621e-01 9.02623355e-01 8.25139806e-02 1.36975735e-01
-7.69441426e-01 -2.21890122e-01 8.08577180e-01 2.02670433e-02
5.91295958e-01 -1.37001216e-01 2.50348538e-01 -8.18576336e-01
7.30956718e-02 -1.08734727e+00 5.37807286e-01 -3.73366147e-01
-1.36620617e+00 8.75900209e-01 -1.25234008e-01 -1.05138767e+00
-3.02464485e-01 -1.41244733e+00 -3.98148388e-01 8.02304566e-01
-1.51521504e+00 -1.26517391e+00 -3.72383177e-01 1.30104458e+00
3.29587758e-02 -5.94935954e-01 1.15221035e+00 7.48944581e-01
-6.97164655e-01 1.18309033e+00 -7.98232630e-02 4.58304375e-01
8.49070191e-01 -5.32763302e-01 -1.01227500e-01 6.27593160e-01
1.46303296e-01 4.83647764e-01 3.96806419e-01 -2.59418935e-01
-1.82327807e+00 -8.82743776e-01 1.09198622e-01 -3.55828613e-01
2.33076066e-01 -5.58017790e-01 -8.78283679e-01 4.00473952e-01
1.15363188e-01 3.73809874e-01 1.00845015e+00 -2.34564692e-01
-7.29946733e-01 -4.89574432e-01 -1.70290625e+00 2.56418318e-01
5.51350117e-01 -7.80604899e-01 -1.44506112e-01 2.99260080e-01
-5.77257620e-03 -1.80356428e-01 -9.63509023e-01 3.00735444e-01
7.91326523e-01 -9.81363654e-01 1.19679320e+00 -5.48416018e-01
1.67056531e-01 -9.77847874e-02 -3.99739325e-01 -1.00925851e+00
8.14663693e-02 -5.20202875e-01 -2.67313629e-01 1.52166772e+00
-2.63835210e-02 -6.84158146e-01 1.05388713e+00 3.74506295e-01
4.39121395e-01 -3.03896576e-01 -1.10275722e+00 -7.14931071e-01
-2.82223579e-02 -2.23125130e-01 7.98376024e-01 7.32847035e-01
-1.66911691e-01 -3.40468615e-01 -7.03466237e-01 6.46730900e-01
6.30960822e-01 -3.91481906e-01 1.01308668e+00 -1.21181881e+00
-3.99824709e-01 -7.29308799e-02 -9.44078267e-01 -4.00345773e-01
2.28104085e-01 -4.28678423e-01 -3.64324719e-01 -3.60634595e-01
-2.80661821e-01 -1.29328921e-01 -3.03539097e-01 4.97956365e-01
3.38238984e-01 7.09272027e-01 2.29420990e-01 -1.48480132e-01
1.14269286e-01 3.72690529e-01 7.02568173e-01 -7.23577961e-02
1.00867160e-01 -6.24888167e-02 -3.43391687e-01 6.04911685e-01
5.80301404e-01 -4.20794904e-01 -2.35508189e-01 -1.73490450e-01
-3.22343200e-01 -1.45311747e-02 3.93873066e-01 -1.33140349e+00
1.80554658e-01 2.54166365e-01 1.01596570e+00 -3.15846711e-01
7.57660449e-01 -1.28835058e+00 1.87597238e-02 4.15407479e-01
1.29179731e-01 1.00013195e-02 5.63707530e-01 4.01661754e-01
-2.37858459e-01 -3.67100760e-02 1.05865717e+00 3.00302934e-02
-7.00110197e-01 4.97232169e-01 -1.09050803e-01 -5.59291899e-01
1.27450752e+00 -3.04327786e-01 -2.77155459e-01 -1.71215534e-01
-5.71112633e-01 -6.24115705e-01 2.71516263e-01 5.51057160e-01
6.36634111e-01 -1.23623824e+00 -4.43473727e-01 7.99395859e-01
-4.77069467e-02 -4.54528302e-01 4.32127982e-01 3.56866330e-01
-5.14230847e-01 2.76123047e-01 -8.62113357e-01 -5.68340302e-01
-1.58756518e+00 9.85583484e-01 1.95320591e-01 2.35735357e-01
-5.00102580e-01 8.72059703e-01 1.11154236e-01 -9.60245670e-04
3.53560299e-01 3.81800473e-01 -3.70533317e-01 2.05554858e-01
1.16373444e+00 2.21770287e-01 4.41120982e-01 -9.31843519e-01
-4.22435939e-01 7.27894664e-01 -2.26794332e-01 -8.80092010e-02
1.22047079e+00 2.35214487e-01 -1.87847205e-02 -2.50686407e-01
1.50802183e+00 -5.05485609e-02 -1.08314097e+00 -5.91087043e-02
-2.85531014e-01 -8.04441929e-01 -6.25308380e-02 -5.12683868e-01
-1.27183425e+00 1.27459824e+00 1.26371980e+00 1.24660820e-01
1.36451149e+00 -3.72363329e-01 7.25980759e-01 2.79440373e-01
4.39885378e-01 -8.75510871e-01 4.86762971e-01 5.33877760e-02
8.02329957e-01 -1.17547941e+00 -3.44897062e-01 -4.89078045e-01
-4.11712646e-01 1.30519283e+00 6.29491031e-01 -2.40619868e-01
1.09568691e+00 4.98438954e-01 1.50044918e-01 -6.54502437e-02
-2.29046509e-01 1.38337806e-01 -5.34684733e-02 1.20999134e+00
5.67635298e-02 -3.27631056e-01 3.05107951e-01 5.01637638e-01
-1.95383012e-01 -1.20001853e-01 3.83716911e-01 8.36682022e-01
-1.16547182e-01 -1.26681209e+00 -8.77397001e-01 4.66143191e-02
-9.65000808e-01 4.19021606e-01 -4.05143768e-01 8.63869607e-01
2.59012043e-01 1.02902400e+00 4.16769199e-02 -4.68779832e-01
1.03719495e-01 1.66463941e-01 6.71249688e-01 -1.34473428e-01
-6.38433635e-01 -2.76217222e-01 -3.08042198e-01 -5.11556447e-01
-3.43456388e-01 -6.18981719e-01 -5.15290558e-01 -5.32608092e-01
-1.72328815e-01 -1.92633644e-01 1.04903316e+00 8.49453330e-01
3.38729113e-01 8.37665200e-02 1.07262194e+00 -9.63519275e-01
-6.65266752e-01 -8.93491447e-01 -6.91708148e-01 4.68674123e-01
5.24025977e-01 -9.18191016e-01 -2.28913710e-01 -3.51557061e-02] | [13.11751651763916, 1.1109609603881836] |
6dbe5b31-09ca-45d4-b7b4-bace6fd7de95 | 2305-14826 | 2305.14826 | null | https://arxiv.org/abs/2305.14826v1 | https://arxiv.org/pdf/2305.14826v1.pdf | Building Transportation Foundation Model via Generative Graph Transformer | Efficient traffic management is crucial for maintaining urban mobility, especially in densely populated areas where congestion, accidents, and delays can lead to frustrating and expensive commutes. However, existing prediction methods face challenges in terms of optimizing a single objective and understanding the complex composition of the transportation system. Moreover, they lack the ability to understand the macroscopic system and cannot efficiently utilize big data. In this paper, we propose a novel approach, Transportation Foundation Model (TFM), which integrates the principles of traffic simulation into traffic prediction. TFM uses graph structures and dynamic graph generation algorithms to capture the participatory behavior and interaction of transportation system actors. This data-driven and model-free simulation method addresses the challenges faced by traditional systems in terms of structural complexity and model accuracy and provides a foundation for solving complex transportation problems with real data. The proposed approach shows promising results in accurately predicting traffic outcomes in an urban transportation setting. | ['Yilun Lin', 'Liang Chen', 'Ding Wang', 'Xuhong Wang'] | 2023-05-24 | null | null | null | null | ['traffic-prediction'] | ['time-series'] | [-2.16050386e-01 1.58762038e-02 -2.62992412e-01 2.41442118e-02
-1.89400434e-01 -1.70098394e-01 5.15014589e-01 3.05528581e-01
6.21551834e-02 9.80593503e-01 1.20701678e-01 -8.58132482e-01
-5.36606610e-01 -1.35796583e+00 -3.81205618e-01 -3.16657633e-01
-1.81489214e-01 7.07080483e-01 4.36757714e-01 -6.97955728e-01
-1.21815063e-01 7.00871944e-01 -1.71315694e+00 -8.52744058e-02
1.35740507e+00 5.91275752e-01 7.06805736e-02 4.91492182e-01
-3.05105060e-01 6.88251436e-01 -9.35689881e-02 -4.45288569e-01
1.21217497e-01 -2.98274606e-01 -5.97115397e-01 -4.15173210e-02
-1.27938867e-01 -2.82417685e-02 -6.39297247e-01 5.15057147e-01
2.39255399e-01 4.76761371e-01 4.89033967e-01 -1.88115585e+00
1.39608130e-01 3.81285101e-01 -1.72028840e-01 1.16508231e-01
3.37539911e-01 4.10272360e-01 6.60665452e-01 -4.16092843e-01
3.55724514e-01 1.25666714e+00 5.25281370e-01 1.91675812e-01
-1.44926703e+00 -4.54553932e-01 2.01260954e-01 4.83736724e-01
-1.47345579e+00 -4.10139978e-01 5.89951754e-01 -5.65437257e-01
9.02618825e-01 6.00688398e-01 1.01415384e+00 5.85444629e-01
2.51976699e-01 3.64233494e-01 6.86571419e-01 -6.99977055e-02
3.01535636e-01 -5.59166409e-02 3.30972433e-01 6.51884437e-01
5.78156650e-01 8.11737403e-02 -1.96406946e-01 6.84466287e-02
1.87433392e-01 2.02428728e-01 2.62101948e-01 -3.63660417e-02
-9.11420286e-01 4.65942472e-01 3.86216730e-01 1.21129729e-01
-7.30732679e-01 2.44782627e-01 4.04881909e-02 1.81399018e-01
1.86365277e-01 5.49618863e-02 -1.70701236e-01 -4.34962511e-01
-8.00335228e-01 5.01559734e-01 6.90372407e-01 9.93867159e-01
1.11865485e+00 1.33376688e-01 -1.12681001e-01 2.47787222e-01
3.42223555e-01 7.96952307e-01 -4.66989040e-01 -1.17716205e+00
7.14348495e-01 1.15946102e+00 9.07642245e-02 -1.50717163e+00
-7.34971702e-01 -1.87902898e-01 -8.43475103e-01 -5.90632521e-02
5.12696922e-01 -2.04710633e-01 -5.84782004e-01 1.50770950e+00
5.22880197e-01 1.35204852e-01 -2.97570646e-01 5.48786819e-01
4.43706900e-01 8.52812469e-01 3.26739430e-01 -5.32535054e-02
8.65825295e-01 -8.76297235e-01 -7.38803744e-01 2.88122352e-02
9.22495008e-01 -4.50143009e-01 1.01549995e+00 -2.17328504e-01
-1.34821451e+00 -3.98504645e-01 -3.85087222e-01 2.69698471e-01
-7.11010337e-01 -3.45645785e-01 6.62414491e-01 8.17398071e-01
-9.92870986e-01 1.92379788e-01 -8.28980684e-01 -5.71559191e-01
4.26107645e-01 5.42639017e-01 1.31154833e-02 -3.60506445e-01
-1.15384448e+00 1.10440016e+00 3.67143601e-02 2.37300292e-01
-5.54103494e-01 -9.05959904e-01 -5.82025707e-01 4.80422795e-01
5.27952433e-01 -8.62163484e-01 8.50548446e-01 -3.47199380e-01
-1.15848017e+00 -1.45306751e-01 -3.58469129e-01 -1.61611870e-01
5.67132294e-01 4.39737290e-01 -7.51718879e-01 -2.56712317e-01
5.67017458e-02 2.04177514e-01 -8.87050331e-02 -1.30268455e+00
-6.56438529e-01 -9.56221744e-02 2.41370633e-01 -5.36558554e-02
1.22070145e-02 -4.08391595e-01 -2.11646155e-01 -3.01197097e-02
-2.31482536e-01 -9.15917516e-01 -8.49682510e-01 -2.72744775e-01
-4.78378206e-01 5.39040193e-02 8.17081153e-01 -2.57980853e-01
1.87777781e+00 -1.63821745e+00 -3.06279927e-01 9.20414507e-01
3.50470692e-01 3.98262560e-01 -9.11592469e-02 1.12853694e+00
3.45792651e-01 3.43592018e-01 4.15261984e-02 -6.18062653e-02
2.08217114e-01 3.35242599e-01 1.04121827e-01 4.32105362e-02
-9.78810266e-02 1.07408524e+00 -9.59962308e-01 -6.48304343e-01
6.99389756e-01 2.46723294e-01 -7.69591928e-01 -6.33343239e-04
-2.11378008e-01 5.57790875e-01 -7.79840350e-01 4.77126896e-01
6.23877645e-01 -1.92461520e-01 3.72301072e-01 6.22620024e-02
-5.50984561e-01 1.68986514e-01 -1.21912038e+00 8.81271422e-01
-6.29047275e-01 5.73775351e-01 2.40137316e-02 -1.02325034e+00
6.46842778e-01 1.32614151e-01 8.44355166e-01 -1.09346831e+00
1.88722208e-01 1.73716038e-01 1.31425232e-01 -6.47705078e-01
6.06148124e-01 2.06731841e-01 -1.86602294e-01 6.47006154e-01
-6.45494044e-01 9.61395726e-02 6.33323431e-01 3.13486308e-01
1.17384923e+00 -4.29876983e-01 6.62361681e-02 -2.16673255e-01
5.14823437e-01 4.76494998e-01 4.31403160e-01 5.06043077e-01
-2.27128789e-01 -4.37818505e-02 6.59463048e-01 -5.75597167e-01
-8.74411762e-01 -1.00835633e+00 3.79782736e-01 6.88791454e-01
3.25774193e-01 -6.30425155e-01 -7.35255301e-01 -1.93975434e-01
1.13803804e-01 8.33398700e-01 -3.61539900e-01 -1.79654390e-01
-6.84236467e-01 -6.46879077e-01 1.37853816e-01 3.42656225e-02
3.41255099e-01 -5.57712913e-01 -2.29704514e-01 4.24192309e-01
-5.63505292e-01 -1.19599545e+00 -2.06547618e-01 -7.46342063e-01
-6.17861211e-01 -1.17238045e+00 -9.28403623e-03 -2.83405989e-01
8.55036795e-01 8.11205924e-01 8.96170676e-01 5.46814561e-01
-1.26269788e-01 3.12145472e-01 5.57798594e-02 -2.08363086e-01
-4.57395107e-01 3.22780728e-01 -1.45847574e-01 2.42768377e-01
1.74833715e-01 -7.61249006e-01 -7.04376042e-01 8.10793877e-01
-5.97822964e-01 4.37715054e-01 2.42738158e-01 1.65004745e-01
3.13365906e-01 4.37454194e-01 7.53935933e-01 -5.84945500e-01
7.91183949e-01 -9.66713846e-01 -5.94830453e-01 3.94180059e-01
-9.30523992e-01 -2.01007858e-01 7.50446856e-01 9.14765224e-02
-9.24020648e-01 -1.14422776e-01 -6.14115875e-03 1.90658525e-01
-2.25874394e-01 5.86082220e-01 -3.59956354e-01 -4.84316759e-02
1.60784736e-01 -4.55790460e-02 4.91014421e-02 -2.31556505e-01
2.61472225e-01 3.82133573e-01 -8.22081268e-02 -6.38041615e-01
8.11151266e-01 5.08626163e-01 7.07621872e-01 -8.67605746e-01
-6.17946424e-02 -2.15939805e-01 -6.88565254e-01 -9.17980134e-01
4.68959481e-01 -5.32299399e-01 -1.38425529e+00 1.75449923e-01
-8.61856639e-01 -4.31100696e-01 -3.39251935e-01 4.32682008e-01
-4.88550663e-01 1.76817358e-01 -5.47067299e-02 -9.58592832e-01
2.60682583e-01 -1.20909476e+00 6.08464658e-01 1.91087529e-01
-2.55337179e-01 -1.22009003e+00 1.70616716e-01 4.79123533e-01
9.72637653e-01 5.50109386e-01 1.23944211e+00 -1.99300773e-03
-1.18531358e+00 -3.02953184e-01 -2.97108531e-01 -4.35355961e-01
7.27575719e-02 4.33274001e-01 -3.50398839e-01 4.05142121e-02
-9.51458931e-01 3.68059903e-01 2.06895679e-01 4.06765103e-01
9.72241938e-01 -4.69894499e-01 -6.44301295e-01 9.68012735e-02
1.32936823e+00 1.15469724e-01 8.02678525e-01 1.13485575e-01
7.25057065e-01 1.02413321e+00 4.58298236e-01 4.13783401e-01
1.22698474e+00 7.91958988e-01 3.70337933e-01 -2.69449741e-01
-1.65439159e-01 -5.36104739e-01 4.55203541e-02 8.71631503e-01
-4.49427336e-01 -5.81590593e-01 -1.53184128e+00 5.64923763e-01
-2.26098514e+00 -1.33275616e+00 -9.22057033e-01 2.00722766e+00
-6.14216886e-02 1.99434847e-01 5.18752038e-01 2.59298146e-01
6.56765759e-01 -1.25089541e-01 -1.73060864e-01 -4.81241971e-01
1.07040048e-01 -6.31151274e-02 5.68743289e-01 6.94929063e-01
-1.81986332e-01 9.94911194e-01 6.70419025e+00 7.12578714e-01
-8.16964865e-01 1.36536300e-01 3.72132778e-01 -2.28874246e-03
-5.94036281e-01 3.12797129e-01 -5.44001222e-01 6.29417002e-01
1.65638447e+00 -5.35302639e-01 6.69419706e-01 3.18279594e-01
1.38560653e+00 -3.11799109e-01 -6.15733445e-01 4.64583933e-01
-7.57016361e-01 -1.55748343e+00 2.20012367e-01 6.15194738e-01
7.61767149e-01 -5.88615891e-03 -1.99182466e-01 2.38435060e-01
4.18951929e-01 -1.05744278e+00 4.39091563e-01 1.05513000e+00
3.80814433e-01 -8.18922579e-01 4.23826784e-01 5.08726239e-01
-1.61131692e+00 -1.66864663e-01 1.64038956e-01 -4.21143264e-01
8.42438936e-01 4.36587900e-01 -7.08665967e-01 5.66765547e-01
2.55797565e-01 5.10474741e-01 -4.01610583e-01 1.23910701e+00
3.53206009e-01 6.55060589e-01 -4.81445998e-01 -1.64488137e-01
3.37321818e-01 -5.23302615e-01 3.53231072e-01 8.99920285e-01
4.80944961e-01 1.54638648e-01 2.57700533e-01 7.61175215e-01
1.88274294e-01 1.21790938e-01 -9.31779146e-01 2.10638922e-02
5.49849629e-01 1.04417825e+00 -7.15825260e-01 -1.21903181e-01
-5.33790231e-01 -2.09628135e-01 3.46546508e-02 5.69432497e-01
-9.70820904e-01 -9.95696560e-02 8.85547996e-01 8.20702016e-01
-1.20327123e-01 -5.91567636e-01 -5.15358746e-01 -8.82696986e-01
-2.45238468e-01 -5.07394612e-01 1.21194869e-02 -4.94941533e-01
-6.90221488e-01 1.74344808e-01 1.56950444e-01 -1.20108020e+00
-4.13876362e-02 -2.54843026e-01 -1.05775487e+00 5.99904299e-01
-1.51663244e+00 -1.31660724e+00 -5.40648580e-01 8.29752684e-01
2.10974932e-01 1.12898335e-01 4.42110509e-01 9.13221359e-01
-1.03792822e+00 2.57915884e-01 2.31063396e-01 -4.36474890e-01
-1.46200836e-01 -6.86479270e-01 4.93635118e-01 6.91078901e-01
-5.71553648e-01 3.08339357e-01 6.33774698e-01 -6.86353505e-01
-1.65877545e+00 -1.24092400e+00 1.34044290e+00 -5.05279183e-01
7.47312427e-01 -4.35440391e-01 -6.57400250e-01 3.13359678e-01
-1.51011065e-01 -1.98886201e-01 6.64196849e-01 6.94051310e-02
4.62785333e-01 -5.78897893e-01 -1.02444613e+00 9.41741288e-01
1.19547915e+00 -1.95601627e-01 1.49084017e-01 3.13528121e-01
4.11131293e-01 2.92205453e-01 -7.90916026e-01 5.51256090e-02
6.08214140e-01 -8.42796803e-01 8.76261771e-01 -8.00998390e-01
-1.06731698e-01 -2.93325484e-01 -1.44559562e-01 -1.19698358e+00
-3.80039215e-01 -8.42984140e-01 -1.85435519e-01 1.13118696e+00
6.86085522e-01 -8.15547287e-01 8.10924768e-01 1.32338345e+00
-1.31893620e-01 -6.85690224e-01 -9.07185972e-01 -7.54002154e-01
-9.22963247e-02 -8.44002843e-01 1.29616904e+00 5.91349363e-01
1.54573179e-04 1.71468601e-01 -3.35267872e-01 1.86737835e-01
6.87960386e-01 -3.13116945e-02 1.19116914e+00 -1.31803191e+00
4.98770446e-01 -6.39594674e-01 -4.91220295e-01 -6.44661009e-01
1.33445829e-01 -6.38225615e-01 -4.32689369e-01 -2.02894497e+00
-1.13666050e-01 -9.01050091e-01 2.26565897e-02 4.54270393e-02
2.43235812e-01 -1.85144469e-01 1.01953462e-01 1.57292560e-01
-6.95273697e-01 6.72264040e-01 1.31718242e+00 -1.12578280e-01
-2.51243323e-01 3.31980705e-01 -5.83058536e-01 4.57924813e-01
9.46146607e-01 -4.63488817e-01 -6.73834503e-01 -3.58490318e-01
3.83955181e-01 3.22889209e-01 3.98300856e-01 -9.59914148e-01
5.26507318e-01 -9.35761333e-01 -3.87353003e-01 -8.05119216e-01
3.05404484e-01 -1.24705434e+00 8.09323251e-01 6.00128353e-01
6.10617958e-02 4.78822470e-01 1.25445411e-01 5.73819757e-01
-7.01916590e-02 5.45510232e-01 3.75123173e-01 1.67735279e-01
-5.70839942e-01 6.37533903e-01 -9.95093048e-01 5.77333160e-02
1.45028841e+00 -4.70207900e-01 -5.94684899e-01 -6.17778599e-01
-6.94014847e-01 8.75235081e-01 5.10012865e-01 2.57327974e-01
4.40283060e-01 -1.35221100e+00 -4.65052426e-01 1.58224151e-01
-1.43401891e-01 -3.35648149e-01 5.05256236e-01 1.17644918e+00
-6.92397833e-01 6.56046689e-01 -2.26662293e-01 -2.41855159e-01
-7.50052750e-01 3.88489097e-01 3.72357219e-01 -3.02152097e-01
-2.55043656e-01 -1.71578035e-01 -1.98745310e-01 -4.60378826e-01
-2.09361464e-01 -2.86610484e-01 -3.04440353e-02 -3.95293757e-02
8.03934783e-02 1.24832404e+00 -5.68764023e-02 -1.02240241e+00
-3.37709367e-01 5.14288366e-01 5.85157394e-01 6.34466261e-02
1.20124829e+00 -8.37074518e-01 -7.93073177e-02 1.64364934e-01
8.86570275e-01 -1.22070774e-01 -8.52634311e-01 8.17926899e-02
1.15923367e-01 -6.35992229e-01 7.02864751e-02 -5.95271170e-01
-1.11684346e+00 9.60111439e-01 1.59384117e-01 6.70516729e-01
8.96566868e-01 -2.70586908e-01 1.21820509e+00 5.21476805e-01
4.10374373e-01 -1.21672750e+00 -4.03420419e-01 5.21896005e-01
3.25941563e-01 -1.02463937e+00 -2.85625249e-01 -5.60514390e-01
-3.18848103e-01 8.50951254e-01 7.81130373e-01 3.38744611e-01
1.14597332e+00 -3.73463668e-02 -2.58021474e-01 -3.51809531e-01
-9.80678916e-01 -5.60564637e-01 2.91178562e-02 5.60791671e-01
-9.49165225e-02 3.51540685e-01 -2.89331883e-01 2.65896082e-01
-4.86162528e-02 4.83459756e-02 6.30354762e-01 7.99057424e-01
-5.47930896e-01 -1.16483259e+00 -1.36080086e-01 5.03930807e-01
2.56094903e-01 2.77542740e-01 -1.45584866e-01 9.43342268e-01
2.26657495e-01 1.31883478e+00 1.27151404e-02 -5.49872577e-01
8.05904150e-01 -2.25015134e-01 -3.33772376e-02 -2.55793780e-01
-3.83768737e-01 -4.93693680e-01 4.45096403e-01 -7.50483096e-01
-2.57380754e-01 -6.99277341e-01 -1.19954896e+00 -1.40594614e+00
-1.37030929e-01 4.85304862e-01 7.27157354e-01 9.11635518e-01
8.76151443e-01 4.73609537e-01 9.88520861e-01 -6.90884769e-01
3.10661554e-01 -2.76654959e-01 -4.60530162e-01 2.19145283e-01
2.40387678e-01 -8.79570782e-01 -2.64945310e-02 -3.62338126e-01] | [6.265746593475342, 1.8514360189437866] |
923e0699-75cf-41e7-889c-1bcecd84a45e | github-copilot-ai-pair-programmer-asset-or | 2206.15331 | null | https://arxiv.org/abs/2206.15331v2 | https://arxiv.org/pdf/2206.15331v2.pdf | GitHub Copilot AI pair programmer: Asset or Liability? | Automatic program synthesis is a long-lasting dream in software engineering. Recently, a promising Deep Learning (DL) based solution, called Copilot, has been proposed by OpenAI and Microsoft as an industrial product. Although some studies evaluate the correctness of Copilot solutions and report its issues, more empirical evaluations are necessary to understand how developers can benefit from it effectively. In this paper, we study the capabilities of Copilot in two different programming tasks: (i) generating (and reproducing) correct and efficient solutions for fundamental algorithmic problems, and (ii) comparing Copilot's proposed solutions with those of human programmers on a set of programming tasks. For the former, we assess the performance and functionality of Copilot in solving selected fundamental problems in computer science, like sorting and implementing data structures. In the latter, a dataset of programming problems with human-provided solutions is used. The results show that Copilot is capable of providing solutions for almost all fundamental algorithmic problems, however, some solutions are buggy and non-reproducible. Moreover, Copilot has some difficulties in combining multiple methods to generate a solution. Comparing Copilot to humans, our results show that the correct ratio of humans' solutions is greater than Copilot's suggestions, while the buggy solutions generated by Copilot require less effort to be repaired. | ['Jiang', 'Zhen Ming', 'Michel C. Desmarais', 'Foutse khomh', 'Amin Nikanjam', 'Vahid Majdinasab', 'Arghavan Moradi Dakhel'] | 2022-06-30 | null | null | null | null | ['program-synthesis'] | ['computer-code'] | [-5.72146416e-01 -1.48384282e-02 2.24952828e-02 -2.05540717e-01
-3.34010273e-01 -5.22767007e-01 1.83565229e-01 3.98476183e-01
-1.65609121e-01 5.24836183e-01 -3.13337952e-01 -4.46175486e-01
-1.33047163e-01 -9.76243317e-01 -7.38687515e-01 -3.64070892e-01
1.59788221e-01 3.29405338e-01 1.67288378e-01 -2.94332415e-01
7.06349969e-01 1.99137717e-01 -1.87238109e+00 6.18146479e-01
1.16835320e+00 3.47073287e-01 4.89559710e-01 6.68062985e-01
-6.17008507e-01 9.46034133e-01 -8.42083514e-01 -5.48130989e-01
2.13962570e-01 -2.12753505e-01 -8.71045351e-01 -3.34762394e-01
6.61480427e-01 -1.28942981e-01 2.41697520e-01 1.18958831e+00
4.60728794e-01 -4.56478328e-01 2.65812486e-01 -1.73737371e+00
-7.56661654e-01 9.45389867e-01 -4.98141438e-01 -1.48149371e-01
5.04425943e-01 3.12985450e-01 1.06870961e+00 -7.15527356e-01
5.22309959e-01 1.08709335e+00 1.00069427e+00 6.04017675e-01
-1.16111147e+00 -6.76013649e-01 -2.26148576e-01 3.71459313e-02
-1.34098017e+00 -7.82187581e-02 6.92925274e-01 -8.23656917e-01
1.23327124e+00 2.35393569e-01 6.50032401e-01 6.43074930e-01
6.77194059e-01 9.58400428e-01 7.90800989e-01 -6.42630577e-01
3.45581651e-01 4.56521541e-01 6.22268319e-01 1.06072617e+00
4.99186456e-01 -2.86090840e-03 -1.96750790e-01 -3.32233161e-01
4.71919894e-01 -1.91695556e-01 -1.12288713e-01 -3.90464276e-01
-1.26863086e+00 8.21118653e-01 2.14910612e-01 7.03443170e-01
-6.87620267e-02 3.73045862e-01 6.11404002e-01 5.55091739e-01
1.68249384e-01 1.13540673e+00 -6.22009337e-01 -2.04106256e-01
-9.20945644e-01 7.32248247e-01 1.10175133e+00 1.19823027e+00
8.38718593e-01 2.40255594e-01 4.28578630e-02 5.75925410e-01
6.83263987e-02 8.80172625e-02 7.19716489e-01 -1.05964303e+00
3.57567012e-01 1.00523698e+00 1.02036618e-01 -1.55296361e+00
-3.83303493e-01 -3.58699948e-01 -6.13969564e-01 3.44558328e-01
4.40249205e-01 -3.03539544e-01 -2.76240289e-01 1.39934742e+00
-1.94259062e-01 -3.56674701e-01 -1.45061076e-01 7.71783769e-01
1.14579594e+00 1.00252867e+00 -3.60244542e-01 6.40262365e-02
8.58016014e-01 -1.34224176e+00 -4.84840542e-01 -2.87402213e-01
1.11208224e+00 -7.41010368e-01 1.27284527e+00 9.10755873e-01
-1.11937082e+00 -9.36512589e-01 -1.11415458e+00 7.92707726e-02
-1.92320496e-01 4.76964831e-01 9.09065485e-01 7.56179988e-01
-1.26761079e+00 7.23719299e-01 -6.43862188e-01 -1.64385632e-01
9.61682498e-02 2.00547978e-01 -5.53080402e-02 -1.68467443e-02
-6.84799135e-01 9.39912975e-01 4.79551792e-01 -3.06008041e-01
-8.24048877e-01 -9.44229603e-01 -7.29682863e-01 2.73326546e-01
3.54541212e-01 -5.68535030e-01 1.52662122e+00 -9.52031255e-01
-1.03592205e+00 7.83588886e-01 1.17940813e-01 -2.70905882e-01
4.70018148e-01 -1.93201616e-01 -4.51304615e-02 -4.43691254e-01
1.23600326e-01 5.41093409e-01 5.15113294e-01 -1.14540458e+00
-6.37325704e-01 4.72987294e-02 4.21953142e-01 -4.03413832e-01
-4.11914706e-01 1.12113785e-02 -1.63142860e-01 -5.24361610e-01
-2.28232592e-01 -8.53235722e-01 -3.79879445e-01 1.71508595e-01
-2.07022816e-01 -6.76537275e-01 3.34732622e-01 -5.86841166e-01
1.37215352e+00 -2.04644966e+00 3.43939871e-01 -2.07091585e-01
7.27841616e-01 3.75486314e-01 -3.71949971e-01 7.14258134e-01
-5.48009612e-02 3.13617945e-01 -2.34006539e-01 6.80605620e-02
3.81423742e-01 9.48674008e-02 -1.62509143e-01 2.10798666e-01
-1.43203512e-01 9.40495729e-01 -9.82797980e-01 -3.57639313e-01
-1.03735486e-02 -1.23939224e-01 -1.05533934e+00 2.20306262e-01
-7.34118342e-01 -1.10959888e-01 -1.16077848e-01 6.00683987e-01
5.92401922e-01 -8.10524151e-02 1.74273297e-01 1.21637687e-01
-4.14526135e-01 1.64249629e-01 -1.26733172e+00 1.59763157e+00
-7.52958655e-01 9.85163152e-01 -1.62213415e-01 -9.78794277e-01
1.22787535e+00 1.58824489e-01 1.15504771e-01 -5.90789258e-01
6.09173737e-02 3.28540772e-01 2.72013962e-01 -9.70431805e-01
4.89237159e-01 3.76317412e-01 -1.06504112e-01 1.16368067e+00
-6.83979038e-03 -5.01611650e-01 5.94657421e-01 7.20163658e-02
1.13180625e+00 1.12786114e-01 4.49914843e-01 -5.49230933e-01
4.34075743e-01 1.25571802e-01 5.62125027e-01 9.69080448e-01
7.13590011e-02 4.97686088e-01 8.72959435e-01 -1.07705605e+00
-1.10395348e+00 -5.91360092e-01 3.60950083e-01 1.06691134e+00
-1.87062144e-01 -8.06503356e-01 -1.02655876e+00 -6.88504755e-01
-1.27632797e-01 8.57266843e-01 -2.87080348e-01 -3.37880291e-02
-5.95588386e-01 -5.56217730e-01 5.17723918e-01 4.53889489e-01
4.95741904e-01 -1.32106662e+00 -1.11211479e+00 2.36045361e-01
-8.02117810e-02 -6.07089281e-01 -3.34156543e-01 -8.67658779e-02
-7.10604012e-01 -1.15867519e+00 -4.56787080e-01 -1.02940583e+00
7.68675983e-01 3.83737385e-01 1.51193166e+00 8.88431370e-01
-5.90789497e-01 1.66978985e-01 -4.36514944e-01 -3.92430425e-01
-8.86468530e-01 1.70343071e-01 -2.29675159e-01 -6.84074283e-01
4.17377591e-01 -4.69959229e-01 1.00473687e-01 1.51205882e-01
-7.33464062e-01 2.42568389e-01 6.33849323e-01 7.62301445e-01
-9.97650921e-02 3.84149283e-01 4.99078035e-01 -9.67430770e-01
1.04019499e+00 -3.93609107e-01 -9.43613529e-01 2.98056930e-01
-6.72489107e-01 1.16512954e-01 6.88784301e-01 -4.00809020e-01
-7.39177287e-01 -8.08964372e-02 -2.61556089e-01 4.16727439e-02
-2.78167557e-02 8.46211374e-01 3.61620188e-02 -2.27799326e-01
1.11122346e+00 2.15419888e-01 -1.90243199e-01 -3.43271106e-01
9.01037902e-02 6.61004186e-01 2.44039506e-01 -1.00545263e+00
6.51067019e-01 -3.87920380e-01 -5.88081598e-01 -6.55822396e-01
-5.51282704e-01 1.50075793e-01 -4.27857161e-01 -6.53594211e-02
3.22980583e-01 -4.38185126e-01 -6.34852767e-01 5.47931910e-01
-1.76255953e+00 -3.11288506e-01 3.25059704e-02 6.64902776e-02
-3.96342009e-01 3.82323325e-01 -6.13751888e-01 -4.15441245e-01
-4.61187452e-01 -1.58785629e+00 7.68195629e-01 1.02654956e-01
-7.12091804e-01 -8.32375407e-01 1.58475146e-01 3.24379116e-01
4.15646791e-01 1.77179873e-01 1.46799088e+00 -4.93820786e-01
-7.60957837e-01 -2.53402233e-01 -2.35237926e-01 5.31159103e-01
-6.64359108e-02 5.21665096e-01 -7.64836967e-01 -2.85162479e-01
1.40119553e-01 -2.58708715e-01 3.06220472e-01 1.46435499e-01
1.25198734e+00 -3.91647935e-01 -1.57031894e-01 3.44648004e-01
1.41609812e+00 3.49464327e-01 6.52702987e-01 3.55621219e-01
5.28476775e-01 7.81329930e-01 5.64219296e-01 4.72818702e-01
3.03944081e-01 3.87717068e-01 4.80384201e-01 2.46581092e-01
8.36149752e-02 -1.75452888e-01 4.11552846e-01 9.45062459e-01
1.52550846e-01 -6.56998083e-02 -1.39673734e+00 6.36333883e-01
-1.85973012e+00 -6.53879642e-01 -6.62589550e-01 1.75406289e+00
8.67226660e-01 2.22009718e-01 7.46337883e-03 3.20146680e-01
5.95529616e-01 -1.88616857e-01 -2.07429603e-01 -8.61439168e-01
4.87098604e-01 2.84866065e-01 -2.96049684e-01 2.58410484e-01
-6.49157584e-01 8.09020221e-01 6.58115721e+00 6.66173518e-01
-1.20410931e+00 1.46700097e-02 2.24845916e-01 2.16675222e-01
-4.09878284e-01 -6.25650166e-03 -5.99185765e-01 4.70727414e-01
6.90136492e-01 -6.55126870e-01 3.98882568e-01 1.37661529e+00
-9.02948529e-02 -1.53128639e-01 -1.56245875e+00 8.14441860e-01
1.67153910e-01 -1.72687137e+00 -1.64480641e-01 -2.49599829e-01
7.63441384e-01 -3.95523310e-01 -1.58955738e-01 6.77563131e-01
2.28170276e-01 -1.02378047e+00 8.24743688e-01 3.43815237e-01
1.82847425e-01 -9.96479511e-01 1.03889680e+00 7.83200800e-01
-6.36451602e-01 -1.81379363e-01 -5.01572728e-01 -6.17559254e-01
-5.46661556e-01 8.66468132e-01 -8.90697002e-01 3.53737593e-01
8.46157193e-01 6.12470925e-01 -9.35496628e-01 1.30053151e+00
-4.81865138e-01 2.51475215e-01 2.31713250e-01 -5.42637885e-01
5.84638976e-02 1.20896608e-01 2.09778324e-01 1.24185872e+00
4.70977455e-01 -3.00912142e-01 2.79801726e-01 1.68414354e+00
1.44537076e-01 9.39546674e-02 -7.00234175e-01 -3.37520897e-01
5.13061106e-01 1.23542285e+00 -6.28096819e-01 -2.29138330e-01
-3.88318598e-01 3.72693360e-01 3.37996095e-01 -5.85786588e-02
-9.50552642e-01 -8.71119440e-01 3.40171516e-01 8.65085870e-02
-1.17079288e-01 -3.53012085e-01 -5.66922247e-01 -9.28321958e-01
1.34275094e-01 -1.38099766e+00 -1.53188363e-01 -1.03699219e+00
-9.91181076e-01 7.15421200e-01 -8.70665535e-02 -1.07559180e+00
-1.45502016e-01 -6.19869173e-01 -8.08972955e-01 6.73532426e-01
-9.52420533e-01 -7.26120949e-01 -5.42094052e-01 -6.31391332e-02
6.54701233e-01 -4.70891118e-01 6.78023040e-01 4.54026520e-01
-5.27952194e-01 5.87929249e-01 -3.80575627e-01 -5.84210083e-02
3.95922005e-01 -1.19914091e+00 6.57255411e-01 7.96768904e-01
4.16172110e-02 9.46316600e-01 9.07406688e-01 -4.54767376e-01
-1.56510401e+00 -9.33878303e-01 1.08495939e+00 -4.54517692e-01
5.68151236e-01 -3.05571347e-01 -9.95777011e-01 6.71640337e-01
3.71889889e-01 -3.64762187e-01 4.44350034e-01 -6.33503869e-02
-3.02126437e-01 3.63456011e-02 -1.04135585e+00 5.98054230e-01
6.87971532e-01 -1.45774588e-01 -8.82226527e-01 4.35070962e-01
8.51875901e-01 -3.66283536e-01 -5.45744896e-01 1.54811680e-01
4.44709748e-01 -1.52385008e+00 5.66393614e-01 -3.76200825e-01
1.13712978e+00 -2.65688479e-01 -8.43388960e-02 -1.34702873e+00
-1.32993743e-01 -4.78649795e-01 2.49027714e-01 1.24414885e+00
2.66760379e-01 -4.09738541e-01 6.79862201e-01 5.48913240e-01
-5.18494546e-01 -7.84908593e-01 -3.01431507e-01 -7.37333000e-01
1.40758097e-01 -5.79401314e-01 6.27356350e-01 1.18344486e+00
2.93905437e-02 1.91617101e-01 -2.71108657e-01 -7.25362897e-02
2.23759189e-01 4.67747867e-01 1.22800040e+00 -1.38294590e+00
-5.82853079e-01 -5.16200840e-01 -6.00915030e-02 -6.91037178e-01
3.65380973e-01 -1.06696582e+00 1.22900389e-01 -1.45908034e+00
1.35747105e-01 -3.84446502e-01 3.44022512e-01 6.66783214e-01
1.83382183e-01 -2.56409705e-01 2.88222820e-01 -3.62046212e-02
-2.70080000e-01 2.08980858e-01 1.06716514e+00 -4.52298433e-01
-3.52975249e-01 5.00323884e-02 -9.62073445e-01 1.06215870e+00
6.16223872e-01 -6.54273033e-01 -2.24536955e-01 -7.82497287e-01
8.47736478e-01 1.22021347e-01 2.34957859e-01 -1.21902120e+00
3.33837897e-01 -3.38837713e-01 -1.75016806e-01 -4.67346966e-01
-4.66954857e-01 -5.23732662e-01 1.56936362e-01 9.55341578e-01
-3.95986468e-01 4.03388947e-01 3.50516677e-01 -1.74189627e-01
-1.46613687e-01 -1.11960399e+00 7.87016273e-01 -5.09760082e-01
-9.56996620e-01 -1.25534117e-01 -6.22533381e-01 -4.37115617e-02
1.04789078e+00 -7.30640069e-02 -5.93331873e-01 -2.02186704e-02
-2.72961468e-01 2.30005637e-01 4.47973132e-01 5.42039692e-01
7.04097688e-01 -9.97083724e-01 -5.87009192e-01 2.93532103e-01
6.42793030e-02 -3.96918841e-02 -3.92903537e-02 5.72695732e-01
-1.22863328e+00 4.52134222e-01 -5.08505881e-01 -6.00104511e-01
-1.26641202e+00 6.44255161e-01 3.45791221e-01 -1.83919370e-01
-5.75671852e-01 8.62667561e-01 1.28038764e-01 -1.04656541e+00
4.14885014e-01 -5.37435293e-01 4.29341830e-02 3.14191841e-02
6.19325042e-01 4.11674470e-01 2.08360702e-01 3.85570139e-01
-2.00782329e-01 4.18602765e-01 -1.62523285e-01 3.42145324e-01
1.64923334e+00 5.28308511e-01 -7.61289418e-01 2.53905833e-01
7.72215724e-01 9.67783504e-04 -5.36366165e-01 3.58592421e-02
3.08756202e-01 -4.10013109e-01 -9.46577564e-02 -7.72881925e-01
-1.10961068e+00 1.13332582e+00 3.00209671e-01 3.60539377e-01
8.22936356e-01 -3.78878623e-01 6.83689058e-01 8.29295874e-01
8.56948376e-01 -7.89719105e-01 4.89090800e-01 6.75547719e-01
1.18247664e+00 -1.04067159e+00 -3.33683006e-02 -2.58414686e-01
-4.06882316e-01 1.39902794e+00 1.18012428e+00 -2.46655658e-01
2.47110292e-01 4.85306054e-01 -1.56783774e-01 -1.90347925e-01
-9.79764640e-01 4.71751779e-01 -6.50034249e-02 6.01798236e-01
7.42400706e-01 -2.05397725e-01 -5.32426894e-01 8.21055233e-01
-5.17888308e-01 2.87530720e-01 1.10618579e+00 1.19643462e+00
-6.04002595e-01 -1.04945958e+00 -4.19370055e-01 3.33755493e-01
3.90006602e-02 -3.58923301e-02 -4.11184907e-01 8.80441010e-01
4.03163731e-01 6.76301718e-01 -1.97795093e-01 -4.90417689e-01
1.76883906e-01 -1.18131161e-01 5.45671821e-01 -1.12612939e+00
-8.54826629e-01 -3.59153897e-01 7.11701289e-02 -5.26010036e-01
-5.14205359e-02 -1.90117911e-01 -1.10573637e+00 -5.12400389e-01
-1.19785853e-01 3.24252874e-01 6.52343392e-01 9.35810089e-01
2.81569809e-01 7.28036106e-01 2.82556444e-01 -8.25797796e-01
-5.93788266e-01 -6.33260250e-01 -2.45455101e-01 -1.52046725e-01
9.28530023e-02 -3.77649039e-01 -3.78226101e-01 -7.53903463e-02] | [7.927910804748535, 7.6548848152160645] |
fa7404fa-e854-4701-a3d0-89d06d47a402 | fairness-in-multi-task-learning-via | 2306.10155 | null | https://arxiv.org/abs/2306.10155v2 | https://arxiv.org/pdf/2306.10155v2.pdf | Fairness in Multi-Task Learning via Wasserstein Barycenters | Algorithmic Fairness is an established field in machine learning that aims to reduce biases in data. Recent advances have proposed various methods to ensure fairness in a univariate environment, where the goal is to de-bias a single task. However, extending fairness to a multi-task setting, where more than one objective is optimised using a shared representation, remains underexplored. To bridge this gap, we develop a method that extends the definition of Strong Demographic Parity to multi-task learning using multi-marginal Wasserstein barycenters. Our approach provides a closed form solution for the optimal fair multi-task predictor including both regression and binary classification tasks. We develop a data-driven estimation procedure for the solution and run numerical experiments on both synthetic and real datasets. The empirical results highlight the practical value of our post-processing methodology in promoting fair decision-making. | ['Arthur Charpentier', 'Philipp Ratz', 'François Hu'] | 2023-06-16 | null | null | null | null | ['multi-task-learning'] | ['methodology'] | [ 2.43197367e-01 1.33104801e-01 -4.56597894e-01 -8.83712053e-01
-8.08529615e-01 -1.41670451e-01 4.99073952e-01 3.78593177e-01
-8.16328824e-01 1.19448233e+00 1.70173004e-01 -4.53138292e-01
-3.84064555e-01 -5.17749906e-01 -4.26896662e-01 -8.18193495e-01
1.38956693e-03 3.83142799e-01 -4.59369242e-01 -6.91837296e-02
6.14747047e-01 4.93085198e-02 -1.12099421e+00 -2.56139070e-01
1.30924129e+00 8.53318930e-01 -4.39339072e-01 5.99399507e-01
3.27627510e-01 6.07472718e-01 -4.80953336e-01 -1.01638985e+00
5.51665068e-01 -3.29609096e-01 -7.61318922e-01 -2.83062637e-01
4.19352084e-01 -2.57582337e-01 3.56778830e-01 1.22126746e+00
8.74961376e-01 1.28133968e-01 9.83325601e-01 -1.60704422e+00
-6.63262904e-01 6.13688529e-01 -1.05810845e+00 1.58643261e-01
-2.88017362e-01 -1.75404906e-01 1.29804528e+00 -3.36050034e-01
-7.99252614e-02 1.60040796e+00 6.93220794e-01 6.04234815e-01
-1.48165810e+00 -8.73793244e-01 7.86048621e-02 -1.08725607e-01
-1.08790362e+00 -4.63852823e-01 4.95465904e-01 -4.34239447e-01
2.97698587e-01 3.26484919e-01 1.28736734e-01 9.45421696e-01
5.02965808e-01 4.20963854e-01 1.62770057e+00 -5.17802715e-01
4.17054832e-01 9.65953395e-02 3.68558913e-01 3.66342634e-01
8.14094424e-01 1.61628947e-01 -6.11571193e-01 -6.29964709e-01
4.73129451e-01 -5.44559490e-03 2.05144539e-01 -4.15524721e-01
-7.58209407e-01 1.27832294e+00 1.13173895e-01 -2.47474089e-01
-3.21419060e-01 2.98093706e-01 6.08806670e-01 4.52060133e-01
1.03913867e+00 3.60592723e-01 -2.47355789e-01 3.03231347e-02
-1.05526745e+00 5.59423685e-01 8.91628623e-01 4.69205618e-01
6.70498967e-01 -1.79835126e-01 -6.38909161e-01 8.45827937e-01
2.59030044e-01 4.55121547e-01 2.80922681e-01 -1.32952654e+00
4.93416309e-01 1.42197564e-01 4.06718373e-01 -9.79660094e-01
-1.84736133e-01 -5.40285587e-01 -1.11062467e+00 3.58562768e-01
7.99479425e-01 -4.02670383e-01 -3.99122238e-01 2.15420580e+00
4.97645319e-01 -3.40503901e-01 -2.63612837e-01 9.24732029e-01
-6.54206872e-02 1.97029933e-01 4.66451645e-01 -4.01155800e-01
1.25532687e+00 -8.08116734e-01 -6.62049174e-01 -1.52596071e-01
4.75663871e-01 -4.88406718e-01 9.57387745e-01 4.17151749e-01
-1.02883291e+00 -1.87321663e-01 -7.21347511e-01 -1.88425571e-01
5.82078956e-02 -2.36617550e-01 6.82678998e-01 1.20143354e+00
-9.30190623e-01 5.12576222e-01 -2.76424080e-01 -1.50990516e-01
7.73806036e-01 3.66015643e-01 4.24739867e-02 7.25844726e-02
-1.26975703e+00 1.07010603e+00 -3.92939604e-04 -1.02747418e-01
-7.72820532e-01 -9.55499470e-01 -5.75542748e-01 1.86210647e-01
4.69857186e-01 -1.12734210e+00 1.25108993e+00 -1.30662632e+00
-1.31032634e+00 1.02729452e+00 -1.31403655e-01 -5.09990871e-01
9.88209307e-01 -1.66818097e-01 2.34640688e-01 -4.79669660e-01
3.18359941e-01 2.98943192e-01 8.48675728e-01 -1.13604963e+00
-6.78772628e-01 -5.03798842e-01 3.36787328e-02 3.85614008e-01
-6.08185947e-01 4.48445827e-01 4.23624635e-01 -7.55322576e-01
-6.15160286e-01 -8.18935633e-01 -6.20415747e-01 1.43236667e-01
-3.94745409e-01 -2.23892167e-01 4.25995477e-02 -5.26938021e-01
1.25475931e+00 -1.83273602e+00 1.28686912e-02 2.65349597e-01
3.41471702e-01 -1.66759312e-01 1.20810337e-01 -2.28029992e-02
1.52227074e-01 3.23851436e-01 -4.18432951e-01 -7.66220689e-01
2.95484930e-01 -1.34617379e-02 -1.62381902e-01 7.78412938e-01
-1.15955852e-01 5.84385753e-01 -8.19403112e-01 -6.88657582e-01
-2.21377283e-01 5.14905602e-02 -7.50920534e-01 8.14689472e-02
2.40314990e-01 1.57869935e-01 -4.48452681e-01 4.12788957e-01
8.09380710e-01 -1.89593226e-01 2.50699908e-01 5.09829760e-01
-6.67764619e-02 1.63024887e-01 -1.16790164e+00 1.15773261e+00
-4.07392800e-01 2.24599764e-01 4.42329109e-01 -1.26389945e+00
7.26570606e-01 -9.67038125e-02 4.11489129e-01 -4.77049083e-01
2.13171974e-01 2.24897370e-01 2.42735408e-02 -4.34100488e-03
6.58278525e-01 -8.08521807e-01 -5.27146816e-01 8.80312383e-01
-3.68580967e-01 -2.46049110e-02 1.10670337e-02 -3.06340009e-02
8.31405222e-01 -1.50372684e-01 7.06889212e-01 -9.76789176e-01
2.14748427e-01 -2.79536009e-01 7.66343832e-01 1.02724552e+00
-7.08570242e-01 3.24776173e-01 9.39138055e-01 -3.17759871e-01
-1.10610306e+00 -7.70253420e-01 -3.91385704e-01 1.57873821e+00
6.74714800e-03 8.68890211e-02 -6.11559212e-01 -6.47466004e-01
6.51560307e-01 6.64078116e-01 -1.03958118e+00 -6.70664385e-02
-1.77559718e-01 -1.44735205e+00 5.55143833e-01 5.65121174e-02
2.49294266e-01 -5.28569639e-01 -9.82874632e-01 -1.44352913e-01
-8.84082094e-02 -5.59069514e-01 -6.20505154e-01 2.17424497e-01
-8.18314016e-01 -8.97265673e-01 -9.87049460e-01 -2.54079878e-01
6.31908834e-01 6.00407235e-02 1.07346594e+00 -5.44873960e-02
-1.01157114e-01 1.22893229e-01 2.00706080e-01 -6.64397895e-01
-1.75470904e-01 1.64378267e-02 1.78472176e-01 1.08373635e-01
3.02779615e-01 -2.60661781e-01 -6.62503839e-01 3.56928647e-01
-7.49930561e-01 1.37779519e-01 3.87629867e-01 1.01874757e+00
2.38393366e-01 -3.07410538e-01 8.66321325e-01 -1.09921372e+00
1.15546131e+00 -7.34512746e-01 -5.00117719e-01 3.70597541e-01
-9.90075827e-01 2.61741161e-01 4.72834140e-01 -3.00830245e-01
-1.27741170e+00 -4.17310297e-01 4.45127964e-01 -2.73355711e-02
3.32698941e-01 3.54847252e-01 -8.12903978e-03 -3.02768610e-02
5.84971130e-01 -4.42346931e-01 2.04960287e-01 -1.77060768e-01
3.00620407e-01 6.83303237e-01 9.76646617e-02 -1.15054119e+00
4.52530354e-01 4.15648788e-01 4.51248646e-01 -2.90420890e-01
-1.05655515e+00 -2.59963926e-02 -3.30002844e-01 -2.27755725e-01
5.73899209e-01 -8.05407703e-01 -1.09666681e+00 3.50730002e-01
-9.12122071e-01 -5.31917930e-01 -1.51316017e-01 2.61539787e-01
-5.81636190e-01 3.70590836e-01 -3.24557781e-01 -1.38311338e+00
-5.97048938e-01 -9.15165961e-01 9.33631778e-01 1.03610277e-01
-3.97018522e-01 -1.15500093e+00 4.48469996e-01 4.93791550e-01
4.17046428e-01 2.43735194e-01 9.14435863e-01 -6.45777464e-01
-2.89861243e-02 1.74015090e-01 -3.63968253e-01 2.71436065e-01
1.23538543e-02 -7.83155188e-02 -9.72339630e-01 -3.80137295e-01
6.71933591e-02 -6.34353399e-01 9.66174006e-01 7.93241203e-01
1.48359990e+00 -4.95726705e-01 -1.03810363e-01 5.24166882e-01
1.33135724e+00 -2.58963257e-01 1.74529999e-01 3.73290896e-01
4.22716737e-01 1.10753596e+00 7.65640676e-01 9.56610322e-01
7.50281990e-01 5.73680282e-01 4.17321354e-01 -1.08613990e-01
5.06098211e-01 3.89616936e-02 1.04594581e-01 2.89699256e-01
-1.73620790e-01 -6.59989864e-02 -1.02252471e+00 4.51109231e-01
-2.12410307e+00 -1.04453647e+00 -1.31172478e-01 2.60162306e+00
1.12109697e+00 2.98806019e-02 4.25665021e-01 5.98936044e-02
9.84865606e-01 -8.03694651e-02 -6.42346740e-01 -7.88995206e-01
5.52459620e-02 1.22377850e-01 8.47672105e-01 6.86794221e-01
-1.12007689e+00 6.26015127e-01 6.87265587e+00 8.51000130e-01
-7.91456699e-01 2.93645173e-01 1.49794233e+00 -3.05809468e-01
-4.59875762e-01 1.18602929e-03 -6.21823668e-01 5.80075443e-01
8.44847023e-01 -7.48646498e-01 3.32646132e-01 7.34341204e-01
5.28992236e-01 -5.05377948e-01 -1.08081472e+00 7.84839630e-01
-4.84832190e-02 -5.61106026e-01 -2.39778772e-01 3.06629539e-01
8.89539480e-01 -4.65240210e-01 3.81279290e-01 2.02523351e-01
8.94971907e-01 -1.24672604e+00 8.13211203e-01 4.37134951e-01
8.14501584e-01 -9.72483933e-01 5.72863221e-01 4.46045190e-01
-4.33946878e-01 -2.40613922e-01 -5.16486883e-01 -6.87706053e-01
-1.33509217e-02 1.06483734e+00 -4.34765339e-01 5.02440453e-01
7.10369647e-01 4.65543598e-01 -4.32367116e-01 9.07862246e-01
9.27128047e-02 3.93579572e-01 -1.07869200e-01 -7.66068697e-02
-1.23264538e-02 -3.74075770e-01 2.35132277e-01 1.09574378e+00
2.48475060e-01 -5.62854856e-02 -1.35060742e-01 8.08090985e-01
-2.25470021e-01 3.25144678e-01 -3.55919808e-01 3.74410629e-01
3.50649476e-01 1.30789423e+00 -4.90496933e-01 -9.68498513e-02
-1.20981708e-01 6.09786928e-01 6.30225658e-01 3.14010501e-01
-8.51615787e-01 -2.47522593e-01 1.11978161e+00 -2.00049549e-01
-2.07084253e-01 7.59706879e-03 -1.05917752e+00 -1.09467411e+00
-1.62697956e-01 -8.87345076e-01 7.00468123e-01 -1.96737841e-01
-1.60908413e+00 -1.66438162e-01 2.30728522e-01 -6.10819161e-01
-1.23762079e-01 -4.56122875e-01 -6.12008095e-01 1.21035695e+00
-1.76017022e+00 -7.64285803e-01 -2.11161021e-02 3.02864075e-01
4.01010364e-02 2.89216023e-02 6.07457817e-01 3.24709833e-01
-7.69003451e-01 7.83194482e-01 3.03682148e-01 -3.18415105e-01
1.26260889e+00 -1.30691206e+00 -6.85257688e-02 8.23743403e-01
-6.80164456e-01 5.76754272e-01 9.57975149e-01 -6.20280683e-01
-9.24092591e-01 -9.17817235e-01 1.12283218e+00 -3.76478583e-01
5.87529421e-01 -4.55915183e-01 -4.93462890e-01 4.32900667e-01
1.69561177e-01 -8.39054212e-02 1.05428231e+00 5.45299351e-01
-2.65145183e-01 -3.90289575e-01 -1.44506168e+00 5.66033006e-01
9.66542900e-01 -2.54529506e-01 -1.82304382e-01 7.02438280e-02
1.70978263e-01 -1.61446035e-01 -7.86597192e-01 2.45866612e-01
7.96391070e-01 -1.08386922e+00 8.25093389e-01 -8.89655232e-01
8.44920576e-01 2.82710284e-01 -1.01150513e-01 -1.61222184e+00
-2.36909837e-01 -9.03156996e-01 2.29025185e-01 1.17423058e+00
2.26404443e-01 -8.51824284e-01 5.91638863e-01 1.12044275e+00
3.43821913e-01 -7.50905156e-01 -1.20187807e+00 -4.45484549e-01
7.75256574e-01 -2.28232175e-01 5.19095540e-01 9.81168807e-01
-5.87235950e-02 1.05665632e-01 -1.01355577e+00 -2.65034378e-01
1.20237958e+00 9.68339741e-02 7.40627110e-01 -1.41198111e+00
-2.24973932e-01 -6.96074009e-01 1.67099506e-01 -5.52161157e-01
6.04698539e-01 -8.17260385e-01 -6.85276184e-03 -1.05034316e+00
8.27565312e-01 -7.56414056e-01 -3.67085516e-01 3.60722274e-01
-7.11858988e-01 7.89854378e-02 1.72783881e-01 1.11696891e-01
-6.22556031e-01 7.08889186e-01 1.09540963e+00 5.16929515e-02
2.00960889e-01 2.27506220e-01 -1.36402297e+00 4.12292570e-01
9.12680209e-01 -9.47218895e-01 -2.90261269e-01 -2.79053509e-01
4.77414578e-01 -1.12933487e-01 4.41994548e-01 -3.78244579e-01
2.71637402e-02 -7.41980195e-01 3.24719489e-01 2.03667626e-01
1.36344892e-03 -6.17013156e-01 -1.23434804e-01 5.28561175e-01
-8.45028162e-01 3.27951163e-02 -1.57295808e-01 5.17052770e-01
2.57303864e-01 -2.37687737e-01 9.71052825e-01 5.21971062e-02
1.36759013e-01 2.63500959e-01 -1.01942964e-01 4.61089432e-01
1.13164496e+00 2.18686104e-01 -5.09475529e-01 -4.77272391e-01
-1.75583571e-01 5.91135025e-01 4.29604858e-01 5.97130843e-02
9.46124867e-02 -1.32057691e+00 -1.08956981e+00 -1.62492082e-01
-1.33613005e-01 -3.96498829e-01 1.86552107e-02 8.59437644e-01
-1.55900985e-01 3.74075562e-01 -4.10160452e-01 -1.86535388e-01
-1.24694824e+00 3.12509954e-01 4.27860200e-01 -4.55066353e-01
2.08972290e-01 6.27310038e-01 3.48392993e-01 -4.04211849e-01
1.97705626e-01 -5.09336926e-02 8.46870691e-02 2.37054437e-01
3.03539842e-01 8.25595438e-01 -1.19197838e-01 -4.46802408e-01
-2.77090609e-01 1.93428993e-01 1.49040818e-01 -4.05425221e-01
1.29400158e+00 -3.58197212e-01 -3.85705471e-01 5.34626007e-01
8.72431636e-01 -3.76969688e-02 -1.25204408e+00 -1.76540360e-01
1.19780324e-01 -8.22130144e-01 6.32416457e-02 -6.01938963e-01
-7.49138892e-01 7.98359632e-01 2.37948239e-01 3.21589410e-01
9.85467017e-01 -6.06479645e-01 8.65135640e-02 1.73304304e-01
3.02573651e-01 -1.30122471e+00 -5.80212288e-02 3.47386539e-01
6.73282266e-01 -1.51440310e+00 3.07643056e-01 -4.72432077e-02
-8.74356031e-01 7.68998682e-01 5.86906254e-01 -5.43688498e-02
7.39934385e-01 8.80274922e-02 -2.92433575e-02 2.44878337e-01
-7.44874001e-01 1.46118656e-01 2.40662426e-01 2.44099885e-01
6.58031464e-01 3.67221743e-01 -8.95046413e-01 7.35444963e-01
-2.07489774e-01 -2.12512650e-02 5.31645238e-01 7.41818905e-01
-1.99199483e-01 -1.27271974e+00 -4.70925301e-01 7.90964782e-01
-9.83239532e-01 -9.89050418e-02 -2.62927324e-01 3.82604271e-01
-1.08536825e-01 9.35794950e-01 1.42173460e-02 1.94651484e-01
-1.04017265e-01 -3.69678624e-02 3.96969259e-01 -4.89679158e-01
-8.62537026e-01 -2.09673926e-01 6.98700026e-02 -3.10730547e-01
-6.21383607e-01 -9.10385907e-01 -5.13070881e-01 -9.03213322e-01
-2.18746476e-02 2.81075001e-01 5.01562417e-01 8.15306664e-01
1.11675493e-01 2.44261444e-01 7.90970445e-01 -7.37643957e-01
-1.28118742e+00 -7.24753320e-01 -7.09794939e-01 3.37897420e-01
5.31884551e-01 -7.62160301e-01 -5.03691673e-01 -3.40102226e-01] | [8.871633529663086, 5.277050971984863] |
95beb2d7-c12a-4397-aca0-ea6a8c5ef7de | what-do-language-models-know-about-word | 2302.03353 | null | https://arxiv.org/abs/2302.03353v1 | https://arxiv.org/pdf/2302.03353v1.pdf | What do Language Models know about word senses? Zero-Shot WSD with Language Models and Domain Inventories | Language Models are the core for almost any Natural Language Processing system nowadays. One of their particularities is their contextualized representations, a game changer feature when a disambiguation between word senses is necessary. In this paper we aim to explore to what extent language models are capable of discerning among senses at inference time. We performed this analysis by prompting commonly used Languages Models such as BERT or RoBERTa to perform the task of Word Sense Disambiguation (WSD). We leverage the relation between word senses and domains, and cast WSD as a textual entailment problem, where the different hypothesis refer to the domains of the word senses. Our results show that this approach is indeed effective, close to supervised systems. | ['German Rigau', 'Eneko Agirre', 'Oier Lopez de Lacalle', 'Oscar Sainz'] | 2023-02-07 | null | null | null | null | ['word-sense-disambiguation'] | ['natural-language-processing'] | [ 1.09680936e-01 1.66544795e-01 -3.13183427e-01 -9.65767130e-02
-2.05843702e-01 -1.03940237e+00 1.16637254e+00 6.66856289e-01
-7.83553481e-01 6.58262491e-01 5.60413182e-01 -8.55323970e-01
-2.82700688e-01 -1.01862121e+00 -2.11353466e-01 -1.20521754e-01
1.34066314e-01 4.04556692e-01 5.11818349e-01 -1.00343442e+00
4.35236216e-01 2.49418482e-01 -1.59075797e+00 4.17635471e-01
7.51019418e-01 5.33514202e-01 2.68115014e-01 4.78861541e-01
-5.87825358e-01 8.57606888e-01 -5.38762569e-01 -5.92663050e-01
6.67483434e-02 -9.92906466e-02 -1.26192749e+00 -6.05241433e-02
-1.35408655e-01 3.09348285e-01 2.06243452e-02 1.11622751e+00
1.39125541e-01 2.75446266e-01 5.39957166e-01 -9.61534798e-01
-2.71486998e-01 9.06497300e-01 -2.14410260e-01 4.99626309e-01
1.00264585e+00 -1.91964619e-02 1.52656949e+00 -6.22385085e-01
7.10986137e-01 1.34256124e+00 3.57552320e-01 3.48692209e-01
-1.14559019e+00 -2.32397728e-02 3.74573499e-01 4.81120676e-01
-1.16123784e+00 -2.40172505e-01 5.69577217e-01 -6.09759748e-01
1.30794680e+00 3.33012104e-01 5.26217699e-01 8.91251743e-01
2.85376698e-01 6.68894053e-01 1.37934113e+00 -8.83991659e-01
4.01646912e-01 8.60170275e-02 4.77874756e-01 2.82221466e-01
4.36441034e-01 9.07696411e-03 -5.07775307e-01 -9.50319320e-02
4.31993484e-01 -2.23152503e-01 9.27125663e-02 -1.20221265e-01
-1.05686998e+00 8.44889224e-01 2.94066779e-03 9.77460861e-01
-4.71005499e-01 -2.03972995e-01 5.09434342e-01 4.40184712e-01
4.72332388e-01 1.07167590e+00 -6.26400948e-01 -2.88843840e-01
-6.39163494e-01 5.10110795e-01 8.65653992e-01 4.54854280e-01
6.53693497e-01 -4.09962118e-01 8.73330534e-02 6.36520624e-01
2.93548465e-01 4.33664843e-02 1.00010908e+00 -6.09659255e-01
-9.97030884e-02 8.59383702e-01 1.61649585e-01 -1.27190006e+00
-3.36412281e-01 -2.24458680e-01 -1.34751886e-01 -1.10966705e-01
3.25965703e-01 -4.57614474e-03 -5.17478466e-01 1.65585184e+00
2.01289296e-01 2.48066559e-01 2.98854768e-01 6.52748704e-01
4.46607083e-01 1.64536074e-01 4.82063174e-01 -1.27719954e-01
1.71156085e+00 -8.98101777e-02 -6.50836945e-01 -9.71227586e-01
8.25712502e-01 -8.75034213e-01 1.19511259e+00 3.07703227e-01
-7.34713733e-01 -3.54836464e-01 -9.65067029e-01 -3.25061530e-02
-8.40475261e-01 -3.80200833e-01 8.24872136e-01 2.88113862e-01
-9.65887666e-01 4.45875615e-01 -2.84417659e-01 -6.84961855e-01
-2.39898413e-01 -2.79014390e-02 -2.85856515e-01 3.36918123e-02
-1.85167527e+00 1.31877291e+00 8.75847995e-01 -2.64445364e-01
-2.09076405e-01 -2.70434201e-01 -1.12570858e+00 -1.70428246e-01
8.43905568e-01 -4.61184114e-01 1.33385491e+00 -9.87784982e-01
-1.01112711e+00 1.47279012e+00 -2.88871288e-01 -7.94798732e-01
1.26830682e-01 -1.09012827e-01 -7.17751980e-01 -2.97727346e-01
5.16564190e-01 1.54810622e-01 3.98046046e-01 -1.01282895e+00
-8.07923496e-01 -3.03197145e-01 7.87725806e-01 3.95768359e-02
4.69684117e-02 9.28260386e-02 1.06730394e-01 -5.24304807e-01
-3.90629806e-02 -6.64774418e-01 -4.43260133e-01 -6.36757076e-01
-3.73951554e-01 -6.66088641e-01 1.12460487e-01 -3.20547849e-01
1.45083761e+00 -1.97261882e+00 -3.14234234e-02 3.11204463e-01
2.97284186e-01 4.35472280e-01 -5.85154444e-02 7.08637893e-01
-4.02497143e-01 1.65520549e-01 -5.53189823e-03 7.69158900e-02
4.82455760e-01 7.64024973e-01 -6.73204601e-01 -4.36412990e-02
3.15507144e-01 8.41583490e-01 -1.14558232e+00 -4.01103586e-01
3.40108484e-01 -4.52936962e-02 -3.94596308e-01 8.28702897e-02
-5.72689354e-01 -7.05863535e-02 -5.78925610e-01 3.86562645e-02
6.62851706e-02 1.05921462e-01 7.23376155e-01 1.19806275e-01
-1.65975317e-01 9.88101304e-01 -1.34648979e+00 1.61145294e+00
-4.85910565e-01 4.50299919e-01 -4.47713405e-01 -1.11143064e+00
8.79644275e-01 1.41753942e-01 1.63277760e-01 -7.39613235e-01
1.13885462e-01 1.32067263e-01 1.19554006e-01 -6.91471994e-01
8.48772407e-01 -7.45476842e-01 -5.37907124e-01 2.53526986e-01
1.37731982e-02 -1.83496639e-01 5.48712909e-01 3.23351264e-01
1.05822086e+00 -4.27112207e-02 9.90239203e-01 -5.88654280e-01
5.19322932e-01 3.37804437e-01 4.03315723e-01 7.01471508e-01
3.78766768e-02 -1.32184541e-02 7.57513404e-01 -4.62322891e-01
-5.87858021e-01 -8.65157127e-01 3.70770507e-02 1.30793285e+00
2.03964069e-01 -9.39755261e-01 -4.11469191e-01 -5.56000769e-01
-2.18522459e-01 1.06171203e+00 -3.08053195e-01 -2.79438674e-01
-2.78910458e-01 -2.34522566e-01 4.82331067e-01 3.11161071e-01
1.76180050e-01 -9.96512175e-01 -8.51352215e-01 2.91145742e-01
-2.15618640e-01 -1.21655798e+00 1.20502040e-01 4.16821718e-01
-4.07268703e-01 -1.22510898e+00 1.56805709e-01 -4.98366058e-01
-1.11520901e-01 -2.11713109e-02 1.66381335e+00 1.77784607e-01
-5.26268594e-02 4.11412835e-01 -6.37357116e-01 -5.36082506e-01
-3.91116709e-01 3.50158513e-02 2.81106710e-01 -2.02956930e-01
9.07604337e-01 -5.90328038e-01 -1.13482125e-01 -1.53146405e-03
-1.49503827e+00 -3.41647536e-01 3.37963760e-01 3.87556940e-01
4.74952847e-01 1.41464889e-01 2.45197982e-01 -1.16230321e+00
1.18392110e+00 -6.29825532e-01 -2.70975858e-01 2.19386443e-01
-6.55163527e-01 4.59326804e-01 4.04968411e-01 -2.84831703e-01
-7.53902376e-01 -2.08176196e-01 -3.86265874e-01 2.79493123e-01
-5.66913605e-01 1.17240071e+00 -3.44850361e-01 4.89930242e-01
9.00423884e-01 8.29942003e-02 -2.90763736e-01 -3.08474988e-01
7.23201334e-01 5.17287970e-01 3.82224143e-01 -9.11776006e-01
6.96763992e-01 1.88096151e-01 -1.45320863e-01 -8.67292345e-01
-1.11969292e+00 -7.69964516e-01 -7.69465029e-01 -1.00942284e-01
8.71831954e-01 -5.97626030e-01 -4.38253105e-01 -1.17450252e-01
-1.15978646e+00 -1.95916623e-01 -4.45873171e-01 2.64046460e-01
-3.41033936e-01 5.51298916e-01 -1.78121567e-01 -8.36073935e-01
1.49762526e-01 -7.22446322e-01 8.00057769e-01 1.63931802e-01
-9.25596237e-01 -1.41965783e+00 3.16673428e-01 1.11210970e-02
2.93701857e-01 2.45211005e-01 1.21173644e+00 -1.28630018e+00
3.04292649e-01 -1.85343504e-01 2.23307639e-01 -1.15882466e-02
2.81023085e-01 -3.20130259e-01 -8.13975811e-01 1.45459771e-01
-3.18102054e-02 -1.11917786e-01 6.05371475e-01 8.15368816e-02
5.75757027e-01 -3.46820801e-01 -2.43093789e-01 -3.16451609e-01
1.51436937e+00 9.02327448e-02 5.91812670e-01 6.58161044e-01
1.52167067e-01 7.79484272e-01 8.94741714e-01 3.51738930e-01
3.85755330e-01 7.04844356e-01 2.28718251e-01 1.39642522e-01
1.85718894e-01 -5.19659519e-01 2.52519488e-01 5.12170076e-01
8.18049386e-02 -8.19399729e-02 -1.25209773e+00 7.40296245e-01
-1.82479835e+00 -9.25101340e-01 -4.54902500e-02 1.90299702e+00
9.73327756e-01 5.46294928e-01 2.43218422e-01 2.31697768e-01
4.77212161e-01 5.10953784e-01 -9.33444053e-02 -6.68178439e-01
-9.47500840e-02 6.30069852e-01 1.97264194e-01 9.10856664e-01
-8.75411689e-01 1.36480856e+00 6.44101143e+00 8.63989651e-01
-8.79005015e-01 -1.37875050e-01 1.06735349e-01 2.86528498e-01
-5.93625128e-01 2.76068270e-01 -3.09428036e-01 2.58041352e-01
6.42395496e-01 -4.00038719e-01 3.12247127e-01 5.95245004e-01
8.87232646e-02 -2.61161238e-01 -1.16958261e+00 6.63499415e-01
7.27759898e-02 -9.94253874e-01 3.91983598e-01 -7.05285743e-02
1.23000532e-01 -2.41520852e-01 -3.48906338e-01 3.48513961e-01
7.61943340e-01 -1.05063200e+00 6.48889482e-01 5.48074245e-01
3.71401012e-01 -5.51490009e-01 8.08239877e-01 4.82167006e-01
-1.27905703e+00 1.33710697e-01 -4.19109523e-01 -7.94865668e-01
-1.79091264e-02 6.51707292e-01 -9.53903496e-01 7.12583601e-01
1.42906427e-01 7.18345881e-01 -5.04565895e-01 5.12285173e-01
-6.28702879e-01 4.10293430e-01 -1.80814818e-01 -2.28460476e-01
2.44364321e-01 -6.62087575e-02 8.10515642e-01 1.28136659e+00
-5.10120876e-02 2.73467690e-01 3.90174717e-01 8.35306287e-01
3.79203618e-01 3.85866836e-02 -8.52023244e-01 -2.72193998e-01
5.77523291e-01 9.03071105e-01 -7.70310998e-01 -3.49135578e-01
-2.13572413e-01 6.62347198e-01 1.60672918e-01 1.34167716e-01
-3.94195080e-01 -1.91057906e-01 1.16857016e+00 3.05875272e-01
-3.78905199e-02 -4.09551173e-01 -8.66480172e-02 -1.17430019e+00
-3.44386734e-02 -8.72971773e-01 6.64388716e-01 -8.18800092e-01
-1.42087805e+00 5.72865188e-01 1.24418296e-01 -9.67580318e-01
-6.94330037e-01 -1.11504161e+00 -6.62502229e-01 6.24536991e-01
-1.56933951e+00 -8.07059348e-01 4.32354063e-01 6.24541700e-01
5.24858236e-01 1.08868219e-01 1.07509017e+00 -1.88379571e-01
-3.92062142e-02 1.03500865e-01 -5.27732193e-01 2.15477586e-01
5.49938798e-01 -1.52272737e+00 2.94222116e-01 1.07878375e+00
7.54014730e-01 8.58743906e-01 1.33063781e+00 -5.24495721e-01
-1.16125321e+00 -6.11137331e-01 1.57636762e+00 -5.33789277e-01
1.24475050e+00 -5.72458133e-02 -7.21085191e-01 6.69640779e-01
3.67308736e-01 -3.75925362e-01 9.45204198e-01 5.29654026e-01
-5.32350600e-01 2.69087166e-01 -7.33986139e-01 6.87010348e-01
1.04210472e+00 -9.21514034e-01 -1.50502467e+00 1.20413460e-01
6.75651789e-01 -8.83715302e-02 -5.57797432e-01 -2.10088938e-02
1.00513041e-01 -8.42201471e-01 9.42716837e-01 -1.23135746e+00
6.12062216e-01 -3.44149709e-01 -3.74183029e-01 -1.60356784e+00
-1.16195992e-01 -3.65321130e-01 3.86896640e-01 1.20759881e+00
5.55166304e-01 -6.65532947e-01 2.49679938e-01 7.51230657e-01
2.60112882e-01 -3.95225644e-01 -9.10435975e-01 -7.02372670e-01
2.32334748e-01 -1.08415544e+00 6.51515901e-01 1.37189102e+00
7.82796383e-01 8.05317342e-01 1.76772311e-01 6.82324991e-02
-7.57283196e-05 3.60505968e-01 2.51917183e-01 -1.54490232e+00
-4.94324982e-01 -6.21512711e-01 -8.19782257e-01 -9.84407604e-01
6.08633220e-01 -1.08539212e+00 -2.11265404e-02 -1.51800847e+00
-1.66547686e-01 -1.93337575e-01 -3.39518070e-01 6.93955302e-01
-2.38332972e-01 -1.02516927e-01 1.60368457e-01 -2.29526699e-01
-7.50142276e-01 2.87847221e-02 7.21313715e-01 -1.33394986e-01
-6.16364852e-02 -2.40134478e-01 -1.04431105e+00 9.99562442e-01
8.11059356e-01 -3.16060275e-01 -3.63753855e-01 -5.27260602e-01
1.02448869e+00 -2.64251858e-01 3.57587576e-01 -6.05812430e-01
3.03286672e-01 -4.83110845e-01 -3.32950354e-01 2.60745227e-01
-4.77213562e-02 -8.24102581e-01 -2.02995613e-01 2.89374501e-01
-6.40235066e-01 2.96273500e-01 3.09977263e-01 3.53006870e-01
-4.29503173e-01 -4.74772811e-01 2.35653535e-01 -3.01512718e-01
-1.39129996e+00 -2.75111765e-01 -8.81514847e-01 4.22283530e-01
7.65337646e-01 -8.60719904e-02 1.74932659e-01 -3.50277662e-01
-8.09464037e-01 -2.80919503e-02 2.56963044e-01 5.96354306e-01
4.44168836e-01 -1.12340629e+00 -5.04503429e-01 -3.80235352e-02
4.42890614e-01 -2.93093979e-01 -3.07782441e-01 3.45269173e-01
1.48230484e-02 3.80579829e-01 -9.43532884e-02 -2.34110922e-01
-1.30858135e+00 6.47860527e-01 4.77815092e-01 -5.40451646e-01
-1.09948188e-01 6.74552917e-01 -1.36241689e-01 -4.37366456e-01
-2.22132772e-01 -5.26198387e-01 -6.84944928e-01 2.11673945e-01
6.19057298e-01 -1.61304861e-01 2.04152033e-01 -6.99190080e-01
-5.29261291e-01 3.78286928e-01 1.54303983e-01 -1.70942694e-01
1.17752337e+00 -1.02628671e-01 -3.96857649e-01 7.93053806e-01
5.36802828e-01 1.87677860e-01 -3.20714653e-01 -5.77472389e-01
7.03497648e-01 -3.32576841e-01 -2.16821730e-01 -8.47524643e-01
-3.88951540e-01 5.34568548e-01 2.34201148e-01 8.27751219e-01
8.88381124e-01 4.78560656e-01 5.00109315e-01 4.03315067e-01
7.13185251e-01 -9.13982630e-01 -4.05088484e-01 1.03253484e+00
5.24383426e-01 -9.25469458e-01 -2.35270053e-01 -3.55348229e-01
-6.97972178e-01 9.44475770e-01 3.64655524e-01 -2.51095206e-01
4.74003643e-01 3.70329708e-01 1.15516178e-01 -4.49305624e-01
-7.23605454e-01 -9.67747152e-01 3.29862297e-01 5.80388665e-01
6.16917253e-01 3.74546379e-01 -8.90375912e-01 1.07442927e+00
-6.01300895e-01 -1.95610106e-01 3.21849883e-01 8.92398894e-01
-5.22194982e-01 -1.50862849e+00 -1.22806445e-01 1.46344066e-01
-5.75335205e-01 -3.60773444e-01 -1.05009842e+00 6.97922170e-01
3.02120924e-01 1.38388956e+00 -1.50769472e-01 -7.14507401e-01
5.21041811e-01 3.00185680e-01 2.52565861e-01 -9.05887127e-01
-5.65716445e-01 -4.40116167e-01 3.52239192e-01 -6.26920164e-01
-4.88747954e-01 -3.59759957e-01 -1.34185159e+00 -2.74023801e-01
-1.54004589e-01 4.85031873e-01 7.61208609e-02 1.78971493e+00
1.37865871e-01 4.18784708e-01 1.99918926e-01 -2.29641780e-01
-3.84362429e-01 -7.93089569e-01 -6.30170643e-01 7.32883990e-01
5.89757646e-03 -6.59189045e-01 -1.44909397e-01 -4.93877903e-02] | [10.222016334533691, 8.962084770202637] |
d207de9b-4b1b-4b6e-b925-eacb1411a817 | you-only-segment-once-towards-real-time | 2303.14651 | null | https://arxiv.org/abs/2303.14651v1 | https://arxiv.org/pdf/2303.14651v1.pdf | You Only Segment Once: Towards Real-Time Panoptic Segmentation | In this paper, we propose YOSO, a real-time panoptic segmentation framework. YOSO predicts masks via dynamic convolutions between panoptic kernels and image feature maps, in which you only need to segment once for both instance and semantic segmentation tasks. To reduce the computational overhead, we design a feature pyramid aggregator for the feature map extraction, and a separable dynamic decoder for the panoptic kernel generation. The aggregator re-parameterizes interpolation-first modules in a convolution-first way, which significantly speeds up the pipeline without any additional costs. The decoder performs multi-head cross-attention via separable dynamic convolution for better efficiency and accuracy. To the best of our knowledge, YOSO is the first real-time panoptic segmentation framework that delivers competitive performance compared to state-of-the-art models. Specifically, YOSO achieves 46.4 PQ, 45.6 FPS on COCO; 52.5 PQ, 22.6 FPS on Cityscapes; 38.0 PQ, 35.4 FPS on ADE20K; and 34.1 PQ, 7.1 FPS on Mapillary Vistas. Code is available at https://github.com/hujiecpp/YOSO. | ['Liujuan Cao', 'Rongrong Ji', 'Shengchuan Zhang', 'Tianhe Ren', 'Linyan Huang', 'Jie Hu'] | 2023-03-26 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Hu_You_Only_Segment_Once_Towards_Real-Time_Panoptic_Segmentation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Hu_You_Only_Segment_Once_Towards_Real-Time_Panoptic_Segmentation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['panoptic-segmentation'] | ['computer-vision'] | [-2.51129150e-01 -3.01523566e-01 2.41968706e-01 -3.54332089e-01
-7.21205533e-01 -6.90072894e-01 2.97521710e-01 -1.49548769e-01
-3.88004661e-01 -2.07306221e-02 -2.71122307e-01 -5.08728862e-01
2.25532472e-01 -1.11191654e+00 -7.57626295e-01 -5.85256994e-01
-4.86119874e-02 1.94422573e-01 5.23391843e-01 6.16139024e-02
6.65818602e-02 4.81403619e-01 -1.44688928e+00 3.30263227e-01
9.93871212e-01 1.36500633e+00 3.86072606e-01 1.16540718e+00
-5.95660694e-02 4.47829336e-01 -3.70321393e-01 -3.82876277e-01
6.38024211e-01 5.54198623e-02 -5.60042322e-01 1.15652107e-01
6.78644896e-01 -7.23968744e-01 -4.15859401e-01 1.03401685e+00
3.27029526e-01 -4.79009263e-02 2.51460612e-01 -1.00596118e+00
-7.34285235e-01 3.31987411e-01 -7.88068712e-01 4.54284996e-01
-3.32582086e-01 4.82879072e-01 1.14175797e+00 -8.50625813e-01
1.38595030e-01 1.14668179e+00 4.15766120e-01 4.62658107e-02
-1.05663288e+00 -8.41788888e-01 -2.91981065e-04 -1.72252998e-01
-1.35885656e+00 -3.74941647e-01 -2.07546614e-02 -4.29616153e-01
1.01431012e+00 3.41023892e-01 6.84932470e-01 1.40509993e-01
1.09917961e-01 8.30617845e-01 8.66480708e-01 1.04956090e-01
-3.06786038e-02 -2.19018962e-02 -1.12233525e-02 7.75950730e-01
-5.98741286e-02 -1.17700525e-01 -1.19814456e-01 1.79402858e-01
1.01613104e+00 8.45263526e-02 -2.57784784e-01 3.82711828e-01
-1.02612066e+00 9.07844424e-01 7.91303158e-01 -1.44951707e-02
-3.67240280e-01 3.51111472e-01 1.84959546e-01 1.93123654e-01
8.37854862e-01 2.95783311e-01 -6.97568715e-01 -1.64131999e-01
-1.26685643e+00 9.29332450e-02 6.55365646e-01 1.09384513e+00
9.65315044e-01 7.34004900e-02 -2.40561888e-02 8.18581402e-01
2.59109795e-01 1.02076757e+00 8.32746252e-02 -1.02747059e+00
6.86520159e-01 5.27288675e-01 5.85102737e-02 -8.46282721e-01
-5.06030917e-01 -3.66829664e-01 -4.85999584e-01 -4.25205790e-02
3.76813412e-01 -5.32907724e-01 -1.23305750e+00 9.95174348e-01
4.26419139e-01 3.95749718e-01 -1.55585542e-01 1.12740064e+00
8.72132421e-01 1.45797598e+00 -1.01414286e-01 3.75800401e-01
1.61122799e+00 -1.49272668e+00 -1.82422996e-01 -2.69706279e-01
3.76539975e-01 -1.11239636e+00 1.25757909e+00 1.93071216e-01
-1.01952720e+00 -6.41277075e-01 -7.64650643e-01 -3.16304862e-01
-3.80170971e-01 6.08568490e-01 6.43032014e-01 4.32886094e-01
-1.17959189e+00 3.43159825e-01 -1.17492330e+00 -2.47654185e-01
3.60389024e-01 3.98975194e-01 1.93887845e-01 1.77972481e-01
-9.70094204e-01 5.21317184e-01 2.36355051e-01 2.30002448e-01
-5.36645889e-01 -1.00508511e+00 -7.92092562e-01 4.61314529e-01
3.87739778e-01 -2.92633921e-01 1.57995260e+00 -8.03944468e-01
-1.55654681e+00 6.11310065e-01 -1.42063275e-01 -6.37522399e-01
3.28337997e-01 -5.34089625e-01 -5.34512222e-01 5.32861888e-01
1.75573885e-01 1.14055443e+00 7.73761749e-01 -6.35866344e-01
-1.19433820e+00 -7.33452141e-02 1.73038706e-01 2.22523808e-01
-5.27969711e-02 4.36449759e-02 -1.22656238e+00 -4.91852790e-01
7.24463612e-02 -9.91978884e-01 -4.07800347e-01 1.24853671e-01
-4.43807155e-01 9.04866457e-02 7.47177124e-01 -8.30043733e-01
1.23217726e+00 -2.29808378e+00 -4.75464404e-01 1.09039985e-01
1.92110583e-01 5.91139317e-01 -2.00688951e-02 3.83988097e-02
3.01851541e-01 -2.72839628e-02 -5.15560508e-01 -3.99090648e-01
2.73680091e-02 4.27408293e-02 -4.19918120e-01 3.57948869e-01
2.92393357e-01 9.51551676e-01 -5.14102757e-01 -3.32013488e-01
4.79311556e-01 5.95544815e-01 -5.69915175e-01 1.95903182e-01
-2.60957032e-01 2.28138983e-01 -3.78195971e-01 8.87304485e-01
1.15293396e+00 -4.33657885e-01 -4.95071709e-01 -4.17822190e-02
-8.26815367e-01 4.02228683e-01 -1.02890003e+00 1.51101649e+00
-7.81395316e-01 9.63040769e-01 1.12311050e-01 -4.22883749e-01
8.26462269e-01 8.09117109e-02 2.26264656e-01 -7.70608366e-01
1.39886737e-01 3.57662410e-01 -2.66455293e-01 -2.72286564e-01
8.42272460e-01 3.10174286e-01 -7.35040009e-03 2.14586154e-01
-8.59811977e-02 -4.02818412e-01 4.32384551e-01 6.12859391e-02
6.51047945e-01 -9.52010080e-02 -1.52948424e-01 -2.73391604e-01
3.49509686e-01 2.00913072e-01 5.09215713e-01 5.28744102e-01
-7.09646121e-02 9.42163467e-01 6.15925729e-01 -4.23345298e-01
-1.05430543e+00 -1.08457291e+00 -4.07370150e-01 1.04216731e+00
1.82893664e-01 -3.95283699e-01 -9.75513995e-01 -2.28808105e-01
3.64337489e-03 5.47583938e-01 -4.07284200e-01 5.12260377e-01
-5.99887550e-01 -8.86693478e-01 5.44129848e-01 4.76595283e-01
9.99292314e-01 -9.38180983e-01 -9.15389717e-01 2.45772436e-01
1.47361815e-01 -1.19299591e+00 -8.78705800e-01 -1.08967498e-01
-8.44541192e-01 -8.47841084e-01 -8.43838871e-01 -5.25165319e-01
4.48100775e-01 5.13097167e-01 8.83513272e-01 -2.07569897e-01
-3.63534749e-01 -1.26617178e-01 -2.22901300e-01 -8.34246799e-02
2.07760021e-01 3.86049390e-01 -6.44147992e-01 6.01005852e-02
3.67284745e-01 -3.62865657e-01 -1.08361566e+00 3.46140653e-01
-9.20012712e-01 4.56941068e-01 5.90389907e-01 3.70824307e-01
8.13184738e-01 -1.75595313e-01 -1.21128321e-01 -8.23511541e-01
-3.00505292e-02 -6.61270261e-01 -1.40808558e+00 9.84399319e-02
-3.70953977e-01 -2.39132166e-01 8.55719209e-01 1.43434005e-02
-1.09789801e+00 2.51800716e-01 -3.25617135e-01 -3.81532639e-01
-1.64999232e-01 2.56408125e-01 1.87993139e-01 1.08098097e-01
3.07655901e-01 2.83650488e-01 -4.48946804e-01 -6.94997668e-01
6.80061638e-01 9.39700425e-01 6.03206456e-01 -2.31490973e-02
7.40356207e-01 7.97293425e-01 -4.82834667e-01 -9.15817678e-01
-9.12443638e-01 -7.30819404e-01 -5.32127202e-01 1.56401828e-01
1.15578675e+00 -1.20204091e+00 -5.81893861e-01 6.78191960e-01
-9.76637125e-01 -7.58862555e-01 -5.95618002e-02 4.19345587e-01
-2.51959264e-01 -1.41561413e-02 -8.54002476e-01 -2.71094382e-01
-8.63031805e-01 -1.31451809e+00 1.13543415e+00 7.31666386e-01
3.26972514e-01 -6.14574790e-01 -1.53602704e-01 1.96559951e-01
5.98988056e-01 -1.41387999e-01 2.27532778e-02 -2.05148831e-02
-9.63076591e-01 4.71729413e-02 -1.03910053e+00 3.52743030e-01
-7.57082477e-02 4.31823552e-01 -9.97947454e-01 -8.94269813e-03
-2.83860087e-01 8.98383185e-02 1.08425915e+00 6.25854373e-01
1.21740258e+00 -2.93925434e-01 -1.96553007e-01 1.36197722e+00
1.48575068e+00 5.12869179e-01 5.26073456e-01 2.65166372e-01
6.95013642e-01 2.66067773e-01 5.26565731e-01 4.72476542e-01
6.72577202e-01 5.39602935e-01 3.21446389e-01 -4.82280791e-01
-6.49881586e-02 -8.12761486e-02 2.29428962e-01 6.89126492e-01
1.82349071e-01 -3.79380733e-01 -1.12879384e+00 7.26108015e-01
-1.66209602e+00 -5.76332748e-01 -4.94092613e-01 1.91719592e+00
6.15766108e-01 -5.43236956e-02 -4.39059213e-02 -4.87381399e-01
6.20322168e-01 4.31983352e-01 -5.05803525e-01 -4.88719374e-01
1.13758087e-01 5.46634495e-01 1.10924757e+00 7.78028369e-01
-1.59151924e+00 1.42166317e+00 4.72689104e+00 8.48953426e-01
-1.40243077e+00 2.24101484e-01 9.33206320e-01 -3.16952467e-01
1.17124230e-01 1.18276179e-01 -1.07223952e+00 6.96939588e-01
1.20537031e+00 1.50324047e-01 4.04389054e-01 9.82762575e-01
3.00461918e-01 -3.25259358e-01 -3.50758046e-01 9.11165774e-01
-3.33623260e-01 -1.45262218e+00 -1.36842251e-01 8.87402445e-02
8.40617657e-01 7.48807371e-01 1.33055151e-01 1.34372255e-02
3.89550447e-01 -8.19176495e-01 7.32536972e-01 3.14091414e-01
7.23731875e-01 -8.47326219e-01 5.88111997e-01 -2.80401260e-02
-1.63985789e+00 1.12143651e-01 -6.27107143e-01 1.95392802e-01
4.15137172e-01 7.82284319e-01 -4.52887952e-01 2.77998328e-01
1.02137494e+00 6.51104391e-01 -5.77828228e-01 1.31005037e+00
-3.41341376e-01 7.74309814e-01 -7.65217960e-01 2.77761489e-01
9.37037110e-01 -2.57072121e-01 2.13880241e-01 1.61252701e+00
5.61022401e-01 2.73554862e-01 1.22209541e-01 8.13673556e-01
1.95599515e-02 8.41241181e-02 -4.65375744e-02 9.12093818e-02
4.21026051e-01 1.49617672e+00 -9.82506335e-01 -6.04770720e-01
-6.55508101e-01 1.19166458e+00 2.27954239e-02 3.03059489e-01
-1.30370748e+00 -7.04949379e-01 8.16096723e-01 -7.90108889e-02
8.56858194e-01 -2.65639812e-01 -5.09251833e-01 -1.40266430e+00
-9.32072103e-02 -3.32728624e-01 3.57832223e-01 -8.02824676e-01
-7.84969985e-01 7.86481559e-01 -2.09185258e-01 -1.07847536e+00
3.59655082e-01 -5.18702924e-01 -6.30595922e-01 9.17765737e-01
-1.69264519e+00 -1.07720602e+00 -5.25209904e-01 3.28428715e-01
7.90973723e-01 3.05172116e-01 3.22695494e-01 5.34896910e-01
-7.19585359e-01 3.63203377e-01 2.67646641e-01 1.48898795e-01
3.77088487e-01 -1.27140307e+00 1.22775066e+00 1.08004308e+00
8.78926308e-04 2.15279698e-01 2.78829575e-01 -3.03554863e-01
-1.11752939e+00 -1.40848660e+00 9.20758128e-01 -3.92511636e-02
8.42197776e-01 -3.44393194e-01 -1.00101793e+00 6.11359000e-01
1.29127160e-01 2.88526863e-01 2.34573618e-01 -5.19258738e-01
-1.54600397e-01 -4.13544863e-01 -8.37603569e-01 6.01952910e-01
5.56964040e-01 -3.55539590e-01 -5.18991090e-02 2.75224835e-01
1.01837432e+00 -8.35223794e-01 -7.44754016e-01 6.99458495e-02
4.44572181e-01 -1.03187776e+00 7.97846794e-01 3.33238631e-01
5.17949164e-01 -6.03625894e-01 -1.63704157e-01 -9.96085584e-01
-2.49543101e-01 -6.38912678e-01 2.12460626e-02 9.99240220e-01
6.89409316e-01 -8.41760755e-01 5.50444365e-01 4.50832158e-01
-4.32207674e-01 -6.84435129e-01 -7.94519842e-01 -4.92458463e-01
-5.66168465e-02 -5.58797538e-01 7.96321154e-01 5.22588611e-01
-5.84077120e-01 3.69333033e-03 -2.85818309e-01 6.54557526e-01
2.86991149e-01 6.30479157e-01 8.15515876e-01 -7.04343438e-01
-4.30832863e-01 -5.54408610e-01 7.31466860e-02 -1.70112348e+00
-4.17061031e-01 -5.51902175e-01 1.05657503e-02 -1.39121830e+00
-1.66188553e-01 -4.92012322e-01 -1.24382218e-02 5.03690124e-01
-1.13170236e-01 5.99135339e-01 4.37088609e-01 3.56413037e-01
-3.94346476e-01 4.25171405e-01 1.28906620e+00 1.43386528e-01
-5.42769372e-01 1.68511316e-01 -5.24860620e-01 6.82308614e-01
1.47671163e+00 -4.05128449e-01 -2.93013722e-01 -9.80667353e-01
-1.22867100e-01 -1.72484517e-02 3.51455688e-01 -9.98723626e-01
2.63297945e-01 -6.68089017e-02 6.59221038e-02 -8.72305810e-01
4.16440517e-01 -5.24774611e-01 5.10089509e-02 5.07520616e-01
1.83531523e-01 2.73331046e-01 4.96267706e-01 1.39793187e-01
-2.93978989e-01 2.67254841e-02 8.93750310e-01 -1.99621886e-01
-8.35694730e-01 6.79636776e-01 -3.51931453e-01 -3.94929834e-02
8.92217815e-01 -2.76269540e-02 -5.51725328e-01 -8.55309665e-02
-5.36901593e-01 5.47210693e-01 2.57383674e-01 2.36627132e-01
3.71026486e-01 -7.60664523e-01 -6.19655311e-01 3.02116126e-01
-2.62760639e-01 4.06568378e-01 5.96166849e-01 1.01808703e+00
-1.41955268e+00 6.81224704e-01 3.86741385e-02 -6.40462637e-01
-9.60725307e-01 5.26327863e-02 3.86851788e-01 -6.37912080e-02
-9.03950632e-01 1.01144755e+00 6.03964388e-01 -2.73155719e-01
-1.14243373e-01 -8.50679994e-01 1.07536361e-01 1.43167287e-01
6.97854817e-01 4.71520901e-01 -6.96329176e-02 -4.91566896e-01
-1.35962576e-01 5.89451790e-01 -8.53098556e-02 -2.87547857e-01
1.30356765e+00 -2.57749945e-01 -9.86513272e-02 1.51082411e-01
1.10579002e+00 -1.72786549e-01 -1.92573059e+00 -2.04924718e-01
-3.04260880e-01 -6.14216745e-01 3.12278539e-01 -8.35514963e-01
-1.56659222e+00 1.19169509e+00 5.57418644e-01 1.11026898e-01
1.41501307e+00 -3.04852482e-02 1.40177691e+00 3.90630141e-02
-1.91426948e-01 -9.79232848e-01 -4.71657544e-01 6.79504216e-01
6.23249054e-01 -1.12534797e+00 -1.97806448e-01 -6.00333750e-01
-6.22013211e-01 1.15045702e+00 7.15830743e-01 -3.35504889e-01
7.86054909e-01 3.59766573e-01 2.36145988e-01 -2.77891248e-01
-7.05751836e-01 -4.48809028e-01 2.68330723e-01 -1.66600253e-02
2.33019918e-01 4.36057687e-01 3.50007452e-02 2.37982810e-01
-4.15235788e-01 -1.88938066e-01 3.67957324e-01 5.86947739e-01
-6.85473680e-01 -5.07970452e-01 -2.99328506e-01 3.29827607e-01
-5.25788665e-01 -7.52613783e-01 1.82105094e-01 8.19597960e-01
1.61649927e-01 8.09012651e-01 7.47971237e-01 -1.35174915e-01
2.71848619e-01 -2.73657262e-01 -1.95881397e-01 -5.18480539e-01
-7.34312534e-01 5.48192620e-01 -6.18979633e-02 -5.87646365e-01
-3.97533104e-02 -4.20817196e-01 -1.38593018e+00 -6.91332936e-01
-1.86665639e-01 8.99056420e-02 6.52669728e-01 4.38552707e-01
6.43457174e-01 2.59275287e-01 7.16364920e-01 -8.56024384e-01
-7.87842926e-03 -9.01429057e-01 -4.55469638e-01 -2.80583650e-01
3.24281305e-01 -1.46677881e-01 -2.78990895e-01 1.74142092e-01] | [9.366968154907227, -0.024011297151446342] |
116b88e0-db68-462f-9a5e-844d29fbe029 | visual-keyword-spotting-with-attention | 2110.15957 | null | https://arxiv.org/abs/2110.15957v1 | https://arxiv.org/pdf/2110.15957v1.pdf | Visual Keyword Spotting with Attention | In this paper, we consider the task of spotting spoken keywords in silent video sequences -- also known as visual keyword spotting. To this end, we investigate Transformer-based models that ingest two streams, a visual encoding of the video and a phonetic encoding of the keyword, and output the temporal location of the keyword if present. Our contributions are as follows: (1) We propose a novel architecture, the Transpotter, that uses full cross-modal attention between the visual and phonetic streams; (2) We show through extensive evaluations that our model outperforms the prior state-of-the-art visual keyword spotting and lip reading methods on the challenging LRW, LRS2, LRS3 datasets by a large margin; (3) We demonstrate the ability of our model to spot words under the extreme conditions of isolated mouthings in sign language videos. | ['Andrew Zisserman', 'Triantafyllos Afouras', 'Liliane Momeni', 'K R Prajwal'] | 2021-10-29 | null | null | null | null | ['visual-keyword-spotting'] | ['computer-vision'] | [ 4.16880369e-01 -2.81953011e-02 -3.77191693e-01 -3.00333172e-01
-1.16587937e+00 -5.56148291e-01 7.65661359e-01 -6.05817139e-01
-4.14499402e-01 2.62989461e-01 5.59022665e-01 -3.43130618e-01
2.67676443e-01 2.16811836e-01 -1.16389024e+00 -5.39904416e-01
1.18603343e-02 2.63676107e-01 3.98355514e-01 1.31682411e-01
5.32570601e-01 2.74464995e-01 -2.03751683e+00 7.44765401e-01
4.58325982e-01 1.29172158e+00 5.89292705e-01 1.02341449e+00
-2.86281765e-01 9.10715699e-01 -4.97386336e-01 -5.76962233e-02
1.38544263e-02 -5.02512753e-01 -7.82758236e-01 3.28019828e-01
8.85399282e-01 -5.09720206e-01 -4.56152409e-01 7.13732243e-01
6.71443105e-01 5.04353866e-02 6.15457535e-01 -1.39940977e+00
-7.87310123e-01 6.46327496e-01 -4.44031924e-01 7.64381960e-02
6.28638864e-01 5.54336011e-01 1.11265492e+00 -1.30656171e+00
8.54748070e-01 1.45517111e+00 4.30799603e-01 7.16746688e-01
-9.65637922e-01 -5.88724792e-01 4.30837661e-01 6.77687466e-01
-1.48737359e+00 -1.13432634e+00 5.36335170e-01 -2.95451075e-01
1.23681319e+00 1.96957961e-01 5.01475573e-01 1.50358498e+00
-2.36466154e-01 1.31955862e+00 1.09182596e+00 -6.02137208e-01
-9.30515584e-04 -7.21492097e-02 -1.52964458e-01 5.03277063e-01
-6.77767575e-01 -3.12211774e-02 -1.30649436e+00 7.37789646e-02
2.65186697e-01 -5.23336053e-01 -7.49847770e-01 -2.95401186e-01
-1.36563921e+00 5.07892549e-01 -1.31276459e-01 5.81601374e-02
-3.50024968e-01 3.30752969e-01 4.49937016e-01 2.12829590e-01
2.86732495e-01 -1.25505567e-01 -2.36053765e-01 -3.84622902e-01
-1.31560028e+00 2.45852202e-01 6.82341576e-01 1.03984666e+00
7.71127194e-02 8.36665258e-02 -4.29659992e-01 1.04915607e+00
6.07114911e-01 8.05690706e-01 7.24449933e-01 -9.47130859e-01
5.84630549e-01 -3.15029025e-01 2.26042420e-01 -3.27081859e-01
5.45647182e-02 1.68498158e-01 -5.72887212e-02 4.11591679e-02
1.27053052e-01 2.30489969e-01 -1.41710961e+00 1.70616198e+00
7.11431578e-02 5.97709894e-01 2.49922816e-02 9.89853442e-01
1.18142223e+00 8.02809119e-01 8.89745820e-03 -1.87811300e-01
1.55297196e+00 -1.12424982e+00 -1.03294969e+00 -2.54342467e-01
-1.49484709e-01 -8.74737501e-01 1.41647446e+00 3.20680201e-01
-1.28460336e+00 -5.25236070e-01 -8.94595385e-01 -2.33525902e-01
-1.84861258e-01 3.51597548e-01 -7.60049373e-02 2.78979450e-01
-1.57602036e+00 4.83428724e-02 -6.15765750e-01 -4.94393945e-01
2.20561743e-01 1.29609555e-01 -2.06318095e-01 7.19617084e-02
-1.04162741e+00 7.65772700e-01 -1.10145152e-01 7.59174153e-02
-1.31699657e+00 -5.77920914e-01 -1.09937727e+00 6.33863285e-02
5.82277179e-01 -4.14541364e-01 1.75594163e+00 -9.71878290e-01
-1.79612291e+00 1.16577995e+00 -9.51118529e-01 -3.95259291e-01
5.67454994e-01 -3.57379258e-01 -4.07173067e-01 6.51169062e-01
2.98145250e-03 9.85413730e-01 1.44011831e+00 -1.57622182e+00
-6.82689548e-01 -8.31789896e-02 -4.48466003e-01 4.03908372e-01
5.22016622e-02 3.53237689e-01 -1.01820266e+00 -8.80123436e-01
-5.28307557e-02 -4.87843931e-01 5.47478735e-01 1.30045891e-01
-6.22948647e-01 -3.38115901e-01 1.17571115e+00 -9.10072088e-01
9.44006562e-01 -2.39964056e+00 3.07583451e-01 -1.09320350e-01
-7.68757612e-02 4.36941266e-01 -3.37603182e-01 3.94467294e-01
2.83610951e-02 8.44643340e-02 -4.87204902e-02 -8.03876460e-01
1.76450953e-01 7.08303899e-02 -9.67190802e-01 5.11197627e-01
9.64266658e-02 1.16013563e+00 -5.79379976e-01 -7.28199899e-01
2.10694253e-01 6.19024336e-01 -1.44871891e-01 4.28701341e-01
-4.64156747e-01 5.85351959e-02 9.82708856e-02 8.69157970e-01
3.72398674e-01 6.48757862e-03 -3.58593315e-02 -2.31057003e-01
-1.35507688e-01 5.25837481e-01 -9.34954464e-01 1.60185230e+00
-1.92480370e-01 9.74126220e-01 4.71493065e-01 -6.98328912e-01
4.36937690e-01 6.58386230e-01 1.40082583e-01 -8.05130541e-01
-9.95642021e-02 2.19323188e-01 -5.66147268e-01 -1.08414042e+00
4.20294911e-01 8.04494172e-02 2.36817777e-01 4.73127902e-01
2.57100284e-01 -2.41597891e-01 8.29662904e-02 3.17616582e-01
5.98273039e-01 3.47291410e-01 6.39543636e-03 1.92038253e-01
5.69512129e-01 -4.45514709e-01 -3.64836007e-02 8.84061158e-01
-1.59129679e-01 8.24404776e-01 5.81044018e-01 2.56489158e-01
-9.96265173e-01 -1.19706953e+00 -1.32095320e-02 1.26464128e+00
6.38445541e-02 -3.82809490e-01 -7.34244108e-01 -5.30717254e-01
1.66674748e-01 7.84134448e-01 -4.22800779e-01 1.71919763e-01
-5.45361876e-01 2.66832054e-01 7.49214351e-01 6.19162738e-01
2.27135092e-01 -1.42622983e+00 -6.21650159e-01 -2.99311340e-01
-5.39626360e-01 -1.38817894e+00 -1.05460107e+00 -9.07675475e-02
-2.12045088e-01 -7.98258722e-01 -1.33817768e+00 -1.11985672e+00
2.66919613e-01 2.05787957e-01 6.94821119e-01 -1.00630641e-01
-2.24138319e-01 8.36044490e-01 -3.06748480e-01 -1.68083817e-01
-4.15577799e-01 -2.10118994e-01 6.00044020e-02 3.30580592e-01
8.69330093e-02 -2.21506074e-01 -3.74999732e-01 3.23635131e-01
-6.44805193e-01 5.05855121e-02 3.27529520e-01 8.42300951e-01
5.95884264e-01 -6.31048381e-01 3.20827454e-01 -1.54344231e-01
6.64260626e-01 -1.38858050e-01 -4.94304150e-01 5.53038657e-01
-3.11687291e-01 6.44765273e-02 2.35566169e-01 -6.75484121e-01
-8.35909188e-01 -8.53917077e-02 -2.77791798e-01 -8.45902324e-01
-1.79225430e-01 1.25976637e-01 -2.39074573e-01 -1.93644743e-02
7.04034939e-02 7.93516695e-01 2.53613919e-01 -7.19526291e-01
6.14806652e-01 1.28219759e+00 1.10441399e+00 -2.28324980e-01
3.90310347e-01 4.75933254e-01 -4.56926286e-01 -1.19692254e+00
-3.24863464e-01 -6.69773042e-01 -4.80173647e-01 -5.14970005e-01
8.63577485e-01 -9.41028893e-01 -1.02282572e+00 8.73133302e-01
-1.23752344e+00 -6.31470680e-01 -6.87270090e-02 2.89294362e-01
-1.03002334e+00 7.84025013e-01 -4.87320811e-01 -1.11320114e+00
-1.66785210e-01 -1.47529364e+00 1.67731810e+00 4.05847281e-02
-2.27025926e-01 -4.03969169e-01 -2.40763843e-01 4.65441495e-01
3.70510072e-01 -4.61951017e-01 6.52935982e-01 -5.50659597e-01
-6.06138408e-01 2.11360946e-01 -2.75299847e-01 2.94463933e-01
-1.23089179e-01 1.33436481e-02 -1.38811326e+00 -1.99542955e-01
-3.68516058e-01 -7.27732003e-01 1.35080969e+00 4.85406041e-01
1.17806578e+00 -3.87455076e-01 -4.49960291e-01 7.32639611e-01
8.00530910e-01 3.65689665e-01 5.50906599e-01 -4.57084961e-02
5.53822577e-01 6.60813868e-01 4.90556866e-01 1.91795886e-01
4.79288369e-01 1.11570585e+00 3.36548239e-01 -9.82991979e-02
-7.00837016e-01 -5.28850317e-01 7.08176613e-01 5.06397963e-01
3.83429140e-01 -6.73126042e-01 -7.40354717e-01 8.70659053e-01
-1.76182163e+00 -9.84211504e-01 4.78625268e-01 2.10099339e+00
9.09048200e-01 -1.22810900e-01 2.33580083e-01 7.77411833e-02
7.09761143e-01 5.73416531e-01 -5.92533350e-01 -3.14422965e-01
-2.45810434e-01 1.56564847e-01 3.25848490e-01 8.60241950e-01
-1.13130772e+00 1.26060379e+00 6.95121431e+00 7.52859414e-01
-1.39007807e+00 1.51188537e-01 3.32812309e-01 -3.88159841e-01
-3.08219701e-01 -4.55060989e-01 -9.23788548e-01 5.87905049e-01
8.94007981e-01 1.93301827e-01 6.00369513e-01 4.99389380e-01
3.58359575e-01 -4.43613194e-02 -1.06368625e+00 1.13802612e+00
8.02644908e-01 -1.17928183e+00 1.32161006e-01 -1.09690219e-01
2.18003646e-01 1.43040568e-01 3.63810539e-01 -4.22527492e-02
-9.15735960e-02 -1.32152438e+00 1.45075953e+00 5.00372946e-01
1.38734639e+00 -2.76222974e-01 1.18000545e-01 -2.55092350e-03
-1.25098372e+00 -1.19590826e-01 2.27982640e-01 5.47377825e-01
6.16880596e-01 -2.32511804e-01 -9.50273514e-01 9.81860831e-02
9.02806461e-01 6.72817826e-01 -2.83494830e-01 1.15807486e+00
-4.62522745e-01 7.97615767e-01 -2.75578916e-01 8.82328302e-02
2.44769529e-01 4.77246404e-01 7.56785333e-01 1.38152862e+00
2.49326646e-01 -1.27528951e-01 -2.17807107e-02 7.25648105e-01
-1.12384021e-01 4.88243140e-02 -5.88355660e-01 -2.14969501e-01
5.21571696e-01 6.28178060e-01 -4.04602319e-01 -3.55121702e-01
-3.31293821e-01 1.30040181e+00 -1.00870669e-01 8.01606238e-01
-7.25973904e-01 -3.40150684e-01 7.53746808e-01 1.36925876e-02
9.97942805e-01 7.16262916e-03 8.55774656e-02 -9.80447650e-01
3.06394607e-01 -1.06608427e+00 1.20895579e-01 -1.34829080e+00
-8.20167124e-01 4.20482934e-01 -1.83595158e-02 -9.55421984e-01
-5.55324078e-01 -7.46177495e-01 -3.43751907e-01 8.93951595e-01
-1.76340580e+00 -1.27199960e+00 -1.52159676e-01 6.56405866e-01
9.16408896e-01 -1.98158219e-01 4.75757718e-01 8.61330479e-02
-2.85297539e-02 7.38899469e-01 1.10904872e-02 7.31090084e-02
8.66062462e-01 -9.35893118e-01 7.92324543e-01 7.30342567e-01
5.38779259e-01 2.73249716e-01 7.17662573e-01 -6.68165803e-01
-1.41825247e+00 -5.74368358e-01 1.30463123e+00 -2.86957055e-01
5.78154743e-01 -8.18243563e-01 -7.82882690e-01 7.52002478e-01
4.61211562e-01 -9.93679017e-02 1.28680155e-01 -5.38645387e-01
-5.83445072e-01 1.65633321e-01 -8.34548116e-01 6.07993245e-01
1.02695537e+00 -1.02839613e+00 -7.97335863e-01 2.88962603e-01
1.01345301e+00 -5.62100828e-01 -1.62290722e-01 1.86033830e-01
9.87111390e-01 -7.26075232e-01 1.23789728e+00 -5.98545015e-01
2.38189265e-01 -1.31389931e-01 -1.59608155e-01 -9.87323523e-01
3.66300732e-01 -8.62094641e-01 -1.10769205e-01 1.19243336e+00
3.95852089e-01 -3.62578183e-01 3.33130270e-01 2.92767026e-02
-1.57533392e-01 -6.54344916e-01 -1.39195526e+00 -6.35935545e-01
-4.15990114e-01 -6.64518178e-01 2.33504161e-01 5.04912317e-01
6.65003881e-02 2.45620161e-02 -8.76463115e-01 7.11615710e-03
3.97743702e-01 -3.93515825e-02 5.06185532e-01 -5.60803056e-01
-2.85943270e-01 -5.39004445e-01 -1.78921029e-01 -1.65510213e+00
4.76956487e-01 -5.76040268e-01 6.62727714e-01 -1.45383704e+00
4.60997149e-02 4.16238904e-01 -5.93426153e-02 6.76734507e-01
5.53865805e-02 2.16725633e-01 3.85205686e-01 2.80433536e-01
-6.22888029e-01 6.27315164e-01 1.03564525e+00 -3.00240904e-01
5.09858020e-02 -1.38124987e-01 -1.86466083e-01 3.59208435e-01
7.89163783e-02 -1.00216724e-01 -3.17743182e-01 -4.30838674e-01
-3.55737865e-01 3.23395580e-01 6.74651146e-01 -5.61376393e-01
3.59571129e-01 1.19443476e-01 5.17495088e-02 -8.52903545e-01
8.14886808e-01 -4.83141690e-01 -5.10684848e-01 1.23241715e-01
-7.96498001e-01 1.43809570e-02 2.73952365e-01 6.32435501e-01
-1.76684648e-01 -8.80723447e-02 6.24751151e-01 3.50755043e-02
-9.54692543e-01 -1.08242668e-01 -6.79547250e-01 2.50763357e-01
8.13735008e-01 -2.14588061e-01 -3.23236197e-01 -7.61071563e-01
-6.44578755e-01 3.32866132e-01 3.54414850e-01 7.86816895e-01
1.00404882e+00 -1.08445621e+00 -5.29249370e-01 4.29066062e-01
1.23266689e-01 -4.51518208e-01 4.36036251e-02 9.79820192e-01
-2.47687995e-01 7.95528650e-01 1.83466971e-01 -8.49162221e-01
-1.70179188e+00 4.77203578e-01 4.23848599e-01 6.15163326e-01
-8.98878813e-01 1.28273284e+00 3.20436299e-01 2.72427518e-02
1.09117413e+00 -4.09486949e-01 -1.54996902e-01 1.03847280e-01
7.00902164e-01 -4.83463854e-02 -1.36182562e-01 -1.10968137e+00
-6.13614738e-01 8.09951603e-01 3.28417569e-02 -7.51718581e-01
8.98159683e-01 -3.63696575e-01 1.74732029e-01 7.21585929e-01
1.14427531e+00 2.60890126e-02 -1.32912576e+00 -3.30268294e-01
-3.33090797e-02 -4.25293922e-01 1.01587540e-02 -9.73949194e-01
-7.97193348e-01 9.64635074e-01 6.22163057e-01 -2.67551273e-01
1.00313091e+00 4.75534260e-01 8.98172796e-01 9.54615474e-02
-1.21184282e-01 -1.14963031e+00 1.27053782e-01 3.59345794e-01
1.08433914e+00 -1.03313458e+00 -5.04999757e-01 -1.97258472e-01
-1.02519500e+00 9.00508046e-01 1.22013561e-01 4.62457120e-01
3.61487836e-01 5.04649878e-01 4.45759982e-01 -1.79044664e-01
-9.31710720e-01 -3.79336417e-01 6.34890914e-01 5.86372316e-01
6.19696230e-02 -3.07108343e-01 1.37615159e-01 2.94255674e-01
-3.41473520e-02 1.69539511e-01 1.71246529e-01 7.27535605e-01
-4.42695975e-01 -7.77801871e-01 -5.17435551e-01 1.24990400e-02
-4.23956722e-01 -4.43958789e-01 -7.02426434e-01 5.83943427e-01
-2.39864007e-01 9.62451398e-01 1.33784980e-01 -3.39818686e-01
3.17252696e-01 7.09981322e-01 3.24958324e-01 -3.84490073e-01
-2.32744917e-01 4.83835965e-01 1.07835896e-01 -8.51643324e-01
-2.41709501e-01 -8.26574862e-01 -1.04740191e+00 2.49313891e-01
-1.35176629e-01 -8.39804634e-02 7.79534936e-01 1.08762431e+00
3.39753330e-01 3.04291129e-01 1.89378038e-01 -1.09792936e+00
-8.39058161e-01 -9.55316424e-01 -4.47425276e-01 3.36721152e-01
8.68261635e-01 -6.84032500e-01 -7.22676337e-01 3.81530762e-01] | [14.300654411315918, 5.008805751800537] |
6fcb0cec-0fb2-4dfb-84b8-c80366ec9dbc | optimised-convolutional-neural-networks-for | 2004.00505 | null | https://arxiv.org/abs/2004.00505v1 | https://arxiv.org/pdf/2004.00505v1.pdf | Optimised Convolutional Neural Networks for Heart Rate Estimation and Human Activity Recognition in Wrist Worn Sensing Applications | Wrist-worn smart devices are providing increased insights into human health, behaviour and performance through sophisticated analytics. However, battery life, device cost and sensor performance in the face of movement-related artefact present challenges which must be further addressed to see effective applications and wider adoption through commoditisation of the technology. We address these challenges by demonstrating, through using a simple optical measurement, photoplethysmography (PPG) used conventionally for heart rate detection in wrist-worn sensors, that we can provide improved heart rate and human activity recognition (HAR) simultaneously at low sample rates, without an inertial measurement unit. This simplifies hardware design and reduces costs and power budgets. We apply two deep learning pipelines, one for human activity recognition and one for heart rate estimation. HAR is achieved through the application of a visual classification approach, capable of robust performance at low sample rates. Here, transfer learning is leveraged to retrain a convolutional neural network (CNN) to distinguish characteristics of the PPG during different human activities. For heart rate estimation we use a CNN adopted for regression which maps noisy optical signals to heart rate estimates. In both cases, comparisons are made with leading conventional approaches. Our results demonstrate a low sampling frequency can achieve good performance without significant degradation of accuracy. 5 Hz and 10 Hz were shown to have 80.2% and 83.0% classification accuracy for HAR respectively. These same sampling frequencies also yielded a robust heart rate estimation which was comparative with that achieved at the more energy-intensive rate of 256 Hz. | ['Alan F. Smeaton', 'Eoin Brophy', 'Willie Muehlhausen', 'Tomas E. Ward'] | 2020-03-30 | null | null | null | null | ['photoplethysmography-ppg', 'heart-rate-estimation'] | ['medical', 'medical'] | [ 4.76349920e-01 7.25229830e-02 -1.14168683e-02 2.72482466e-02
-5.21053314e-01 -3.89115125e-01 2.63992310e-01 5.02113327e-02
-4.50947791e-01 6.96182370e-01 2.45332241e-01 -1.55888587e-01
1.96268454e-01 -4.24092412e-01 -2.70405591e-01 -4.61218566e-01
-3.23309571e-01 -3.47151428e-01 -4.26995188e-01 2.64854014e-01
-9.87907574e-02 3.28609824e-01 -1.37168670e+00 -6.35794848e-02
2.56386071e-01 1.55249941e+00 -4.32779759e-01 1.30838096e+00
7.33800709e-01 4.72623944e-01 -9.60804582e-01 3.37635368e-01
4.71485555e-01 -4.39729422e-01 -3.48332047e-01 -1.16906449e-01
4.86757249e-01 -6.37207925e-01 -2.30485737e-01 1.55182898e-01
1.07280755e+00 -2.03472584e-01 1.58459216e-01 -1.15946138e+00
-2.19277173e-01 -9.46104452e-02 -2.42142960e-01 6.35875225e-01
8.46508920e-01 5.82774580e-01 2.33934969e-01 -2.11170182e-01
7.84538388e-02 6.92286193e-01 1.21644711e+00 3.82560194e-01
-1.33182585e+00 -5.82831204e-01 -8.32885146e-01 -2.16897443e-01
-1.20981228e+00 -6.26432478e-01 7.33592927e-01 -3.87391329e-01
1.51468909e+00 4.84088391e-01 1.11565435e+00 1.37665808e+00
6.70646548e-01 -2.03069393e-02 1.35841537e+00 -1.69644803e-01
2.86496401e-01 1.12052396e-01 -2.49239132e-01 5.07753134e-01
3.90166849e-01 6.13118261e-02 -6.54041886e-01 5.24314232e-02
9.18146253e-01 2.39599109e-01 -3.48999888e-01 1.73995107e-01
-1.35617864e+00 2.51629233e-01 2.74287492e-01 2.75668681e-01
-4.66627896e-01 6.88341379e-01 5.80319762e-01 3.15816849e-01
1.26575574e-01 4.71017092e-01 -4.95422751e-01 -8.91096771e-01
-9.91931975e-01 -1.70205712e-01 1.06180573e+00 5.22978663e-01
-2.12365706e-02 2.44539157e-01 -3.81419152e-01 3.56773019e-01
4.34695184e-01 6.73788607e-01 5.89028895e-01 -9.71797824e-01
1.79447964e-01 5.05252600e-01 3.49232256e-01 -9.13568199e-01
-9.62219417e-01 -3.39188933e-01 -8.23152184e-01 1.85336664e-01
4.88040060e-01 -3.39073837e-01 -7.69806027e-01 1.37087750e+00
3.81702572e-01 3.18472624e-01 -2.35323161e-01 9.67998624e-01
5.22999585e-01 2.47075826e-01 3.92738223e-01 -3.16027343e-01
1.80257714e+00 -3.38556677e-01 -7.74342954e-01 -1.85208932e-01
2.97564447e-01 -4.79140580e-01 1.14002502e+00 4.37373489e-01
-9.62051332e-01 -7.41672873e-01 -1.63080740e+00 -1.87987313e-01
-2.02610523e-01 1.26397237e-01 5.40553451e-01 1.44003403e+00
-9.10214126e-01 8.40189517e-01 -1.21686339e+00 -6.96136475e-01
5.63335478e-01 4.96485442e-01 -1.64338738e-01 4.56004024e-01
-9.77060080e-01 7.98229992e-01 -1.13135464e-01 2.49724552e-01
-6.46520793e-01 -8.31366777e-01 -6.82416618e-01 6.32652268e-02
-2.89223731e-01 -7.49289811e-01 1.02723694e+00 -5.95721185e-01
-2.00651741e+00 7.11719036e-01 2.93171018e-01 -6.92177653e-01
7.79351056e-01 -5.28559983e-01 -6.48352444e-01 3.39412153e-01
-3.42535019e-01 2.70644605e-01 9.75074053e-01 -1.98427752e-01
-1.30734956e-02 -2.56497711e-01 -3.84777784e-01 5.54525945e-03
-3.58577579e-01 -2.97363043e-01 1.43920556e-01 -3.82576466e-01
-7.03101158e-02 -1.02877223e+00 4.65146840e-01 2.73855746e-01
1.42961349e-02 2.10483387e-01 7.66889334e-01 -8.79752636e-01
1.10622120e+00 -1.95650029e+00 -4.51630533e-01 3.14445384e-02
3.43176007e-01 3.94559085e-01 4.49772209e-01 2.89178878e-01
3.41312704e-03 6.88627660e-02 1.58993036e-01 -2.94827253e-01
-1.24694161e-01 -5.02984896e-02 2.75891632e-01 9.54988420e-01
1.90659702e-01 1.13322914e+00 -5.08391798e-01 -3.47192176e-02
7.21897662e-01 9.76013958e-01 -1.31296664e-01 2.76828796e-01
4.87833947e-01 6.19693458e-01 1.24138027e-01 9.77531135e-01
1.03521697e-01 -6.99293986e-02 3.21974337e-01 -7.70157576e-01
-1.64101496e-02 3.55698854e-01 -1.04832363e+00 1.79193687e+00
-6.69430733e-01 7.50542045e-01 -1.37001947e-01 -9.20421004e-01
1.07920563e+00 5.51370859e-01 7.02730417e-01 -1.12624359e+00
5.09001076e-01 8.96718204e-02 -4.89270464e-02 -9.32896495e-01
-6.00168966e-02 -3.21694493e-01 -1.19495638e-01 4.46810931e-01
-4.25433274e-03 3.41187775e-01 -3.77466410e-01 -3.36271018e-01
1.41217613e+00 3.96085411e-01 5.41799545e-01 -3.78769577e-01
1.46403864e-01 -4.98394430e-01 2.37449393e-01 7.29823947e-01
-8.56303394e-01 5.65994918e-01 7.63673782e-02 -7.89696753e-01
-1.03457093e+00 -1.06346905e+00 -8.87203515e-02 6.67154729e-01
-1.82715654e-01 -2.96685785e-01 -5.55271089e-01 -9.22923163e-02
1.06009290e-01 -5.27379178e-02 -7.86438882e-01 -3.97012621e-01
-6.85783863e-01 -7.60649562e-01 9.09769773e-01 7.88667798e-01
8.07654023e-01 -8.90816867e-01 -1.87562680e+00 3.81574392e-01
1.36493891e-01 -1.13645601e+00 -1.76065952e-01 2.41240174e-01
-9.23874855e-01 -1.04749978e+00 -6.91999137e-01 -1.05869032e-01
-6.05602935e-02 -2.53751278e-01 9.77504909e-01 -1.22825548e-01
-1.10858047e+00 9.12072062e-01 -6.98695034e-02 -6.28308713e-01
1.12626307e-01 -1.31735895e-02 3.74108762e-01 -1.14465840e-01
4.47505623e-01 -8.89762998e-01 -1.54604757e+00 -5.36490371e-03
-2.59882808e-01 -3.29308778e-01 5.93312263e-01 2.87731886e-01
1.28187388e-01 -6.21581137e-01 6.92832351e-01 -2.53435522e-01
7.20605552e-01 -3.76997530e-01 -1.54918596e-01 -2.66247153e-01
-9.25765336e-01 -1.26658991e-01 4.91349190e-01 -5.37236214e-01
-4.02041435e-01 1.40035465e-01 2.15431541e-01 -1.13177523e-01
-2.24735051e-01 -2.76645795e-02 3.28950912e-01 -4.91035655e-02
9.83978391e-01 -2.93487497e-02 5.08420527e-01 -2.93236136e-01
-7.40615055e-02 8.89029980e-01 7.76212037e-01 -9.15189236e-02
2.21212745e-01 4.77556437e-01 3.61885428e-01 -9.09614444e-01
-3.08518797e-01 -4.45205331e-01 -3.56444091e-01 -5.04089355e-01
9.56888914e-01 -1.21076393e+00 -1.23967874e+00 4.85217690e-01
-4.99354899e-01 -3.49796176e-01 -2.41218030e-01 4.95403290e-01
-5.64649224e-01 1.25011474e-01 -5.61364889e-01 -1.31404269e+00
-1.10989034e+00 -4.39556330e-01 1.13451779e+00 4.06925082e-01
-8.90585303e-01 -8.76934946e-01 1.49330478e-02 5.47191620e-01
9.84091938e-01 1.09973776e+00 1.27245620e-01 -2.96410490e-02
-1.34699279e-02 -5.93567908e-01 2.93052942e-02 1.09130554e-01
3.97358626e-01 -3.69143337e-01 -1.54364038e+00 -5.57596743e-01
1.34308845e-01 -2.90916562e-01 1.53850108e-01 5.15109897e-01
7.66580820e-01 -3.67779791e-01 -1.78197503e-01 6.50661170e-01
1.46072757e+00 1.01324067e-01 1.02022767e+00 3.31032157e-01
5.91130555e-01 -6.36819378e-02 3.20871919e-02 6.63258374e-01
-3.39302309e-02 6.25568330e-01 1.76517949e-01 -3.22673619e-01
-3.78178328e-01 1.47635695e-02 3.96228075e-01 2.75974989e-01
-4.18504149e-01 1.70481741e-01 -7.16479182e-01 1.53882101e-01
-1.34717977e+00 -7.09645808e-01 1.73148829e-02 2.59394789e+00
6.11387849e-01 1.12312883e-01 7.13576853e-01 6.35558307e-01
2.89459080e-01 -8.96436796e-02 -7.60725260e-01 -5.26943445e-01
4.47559923e-01 6.66163981e-01 7.93952227e-01 9.48440060e-02
-1.07588017e+00 -1.53610572e-01 6.72897720e+00 -3.38360876e-01
-1.37461448e+00 2.26657361e-01 6.44403279e-01 -7.53017604e-01
5.55753112e-01 -5.08906424e-01 -3.00649762e-01 7.74320126e-01
1.71298075e+00 1.45548731e-01 4.64950025e-01 6.86986029e-01
5.80325902e-01 -3.19039643e-01 -1.10192096e+00 1.58912528e+00
7.43982866e-02 -1.03992987e+00 -8.13431203e-01 9.14708227e-02
4.93684895e-02 -7.15797255e-03 -2.37372309e-01 1.99702799e-01
-6.81022465e-01 -1.18988419e+00 2.20593259e-01 7.55928218e-01
1.12512887e+00 -6.11768901e-01 5.71150005e-01 -1.41048580e-01
-1.26394939e+00 -1.96692452e-01 1.20179065e-01 -6.27533972e-01
-4.44902703e-02 6.61004484e-01 -9.05625343e-01 6.81816861e-02
9.54802275e-01 5.52704394e-01 -4.19225007e-01 7.48603582e-01
3.12741041e-01 5.23069680e-01 -6.26968145e-01 -1.96402743e-01
-3.70623350e-01 3.09312075e-01 1.52773663e-01 1.31290865e+00
4.66872662e-01 -7.80507475e-02 -3.20341252e-02 7.48136580e-01
1.45279333e-01 -3.20191085e-01 -4.96041894e-01 3.07019390e-02
3.51650894e-01 1.50194824e+00 -6.71478689e-01 -2.23393962e-01
-3.02823424e-01 1.03584993e+00 -1.66750744e-01 -1.43623417e-02
-8.96071732e-01 -6.13541722e-01 5.82079947e-01 5.80365002e-01
-2.12152004e-01 -3.92942518e-01 -4.83313590e-01 -8.71958613e-01
2.76936501e-01 -5.69543958e-01 3.36174458e-01 -6.07766628e-01
-7.82312989e-01 1.34955272e-01 -2.62575269e-01 -1.22354209e+00
-3.50406438e-01 -5.67616284e-01 -6.31415427e-01 8.99248719e-01
-1.08725357e+00 -7.84544945e-01 -1.00383103e+00 4.45254982e-01
3.64390701e-01 3.39109451e-01 9.57610667e-01 4.09388065e-01
-6.31529748e-01 5.18097222e-01 -5.93912959e-01 5.50555391e-03
5.00483036e-01 -1.31970584e+00 2.89858043e-01 7.47922838e-01
-4.02521431e-01 6.07265592e-01 7.02105880e-01 -3.76927227e-01
-1.99489081e+00 -8.36722076e-01 5.15096247e-01 -6.68232560e-01
2.89833486e-01 -5.49727976e-01 -4.66449738e-01 2.19215691e-01
1.56727672e-01 3.26864988e-01 9.64223325e-01 -1.71326354e-01
-2.71601509e-02 -4.27700788e-01 -1.50693333e+00 1.93063021e-01
7.38172829e-01 -6.36442304e-01 -3.76635253e-01 1.92669094e-01
-6.55787811e-03 -3.95537734e-01 -1.42251086e+00 2.57646620e-01
1.36020935e+00 -7.84972608e-01 8.50387156e-01 -1.42149463e-01
-1.05430193e-01 -1.87795922e-01 1.89067811e-01 -8.09852660e-01
-2.07827523e-01 -1.08791804e+00 -7.40226984e-01 1.03696322e+00
-8.73564556e-02 -6.97868288e-01 7.00183928e-01 1.05452061e+00
2.04785571e-01 -5.67158759e-01 -1.17234802e+00 -7.75149226e-01
-6.74139380e-01 -3.51199627e-01 -2.74377409e-02 6.39709473e-01
4.30021405e-01 3.35464329e-01 -6.13930881e-01 -1.02658637e-01
5.12443185e-01 -3.46244335e-01 6.65119350e-01 -1.13488436e+00
-3.73705953e-01 8.02790150e-02 -8.10388982e-01 -3.72919858e-01
-7.86149502e-01 -3.19302797e-01 -2.43410617e-01 -1.32654500e+00
-1.90965414e-01 2.03861427e-02 -5.38304985e-01 5.77024519e-01
9.73165184e-02 1.01999271e+00 -2.82183904e-02 6.46314994e-02
-1.73948005e-01 -4.45717610e-02 6.36906087e-01 8.94620568e-02
-6.02033317e-01 -2.20631897e-01 -4.62896049e-01 2.45695561e-01
9.87361848e-01 -1.70160294e-01 -3.73404086e-01 8.38694498e-02
1.74628168e-01 2.21737660e-02 7.24801838e-01 -1.83310950e+00
-8.55724961e-02 5.56216240e-01 1.12847352e+00 2.74603128e-01
3.57740819e-01 -8.16726804e-01 5.25224268e-01 1.14717221e+00
-7.01269507e-02 2.06608325e-01 3.78472865e-01 5.34237444e-01
5.21447837e-01 6.05356157e-01 8.82268727e-01 -1.47424638e-01
-2.48229727e-01 -1.59187496e-01 -3.89875174e-01 -3.18758041e-01
8.76060724e-01 -7.42305338e-01 -3.25699836e-01 -3.57148290e-01
-8.60182106e-01 -2.96418369e-01 7.29263350e-02 4.88702267e-01
2.78359383e-01 -1.33179128e+00 -3.41952831e-01 4.58333433e-01
1.26215862e-02 -6.05494022e-01 2.43506342e-01 1.30735290e+00
-7.69212544e-01 3.52836221e-01 -6.30270898e-01 -7.36932576e-01
-1.13793910e+00 2.17212245e-01 8.01180720e-01 3.00470978e-01
-8.56663883e-01 1.86993524e-01 -9.58956897e-01 4.27952141e-01
1.36394933e-01 -6.55501485e-01 -2.38224072e-03 1.63141694e-02
6.70756221e-01 7.65623868e-01 3.63027483e-01 1.67850382e-03
-5.50286770e-01 5.40728152e-01 4.10131991e-01 8.00874829e-02
1.05809760e+00 -2.50057936e-01 5.18347859e-01 5.88283479e-01
1.21858156e+00 -3.96771073e-01 -1.54580760e+00 5.38868248e-01
-1.72227442e-01 -2.94598371e-01 2.68516362e-01 -1.14136422e+00
-9.22678530e-01 7.21788645e-01 1.72546506e+00 4.42011356e-01
1.32656276e+00 -5.00851929e-01 7.73768842e-01 1.09778471e-01
3.45814079e-02 -1.22248781e+00 1.21132098e-01 -3.23144019e-01
4.56765890e-01 -1.00249636e+00 2.24909544e-01 1.17248194e-02
-8.81098658e-02 1.24771047e+00 3.72701734e-01 -2.54472941e-01
4.50172842e-01 4.27193344e-01 1.86248675e-01 -1.66063860e-01
-3.25554371e-01 6.74959794e-02 1.28867507e-01 8.51343811e-01
7.69484520e-01 8.47223774e-02 -3.74513805e-01 1.46379396e-01
-7.07661137e-02 6.80179834e-01 3.52373183e-01 1.23570299e+00
-1.62584871e-01 -4.14277196e-01 -1.72335297e-01 7.15823591e-01
-8.60130727e-01 2.54382193e-01 -2.84460653e-02 6.40901923e-01
4.20124792e-02 1.06939626e+00 2.73419529e-01 -3.32950383e-01
4.50220585e-01 3.78407061e-01 5.47378361e-01 -3.10094178e-01
-9.87879097e-01 3.87037881e-02 1.21184066e-01 -1.02443659e+00
-5.84118724e-01 -3.92297387e-01 -7.87832201e-01 -1.64242148e-01
1.67468250e-01 -6.11133635e-01 9.96000707e-01 8.97756934e-01
8.78436744e-01 7.36671090e-01 3.98813516e-01 -9.52043951e-01
-3.01712304e-01 -1.14711976e+00 -6.64972901e-01 4.50049251e-01
5.72546124e-01 -4.87228334e-01 -4.45306093e-01 2.58185536e-01] | [13.82914924621582, 3.0124168395996094] |
e17910f0-5f97-4cb0-88e4-1e25beaca298 | clustering-induced-generative-incomplete | 2209.13763 | null | https://arxiv.org/abs/2209.13763v2 | https://arxiv.org/pdf/2209.13763v2.pdf | Clustering-Induced Generative Incomplete Image-Text Clustering (CIGIT-C) | The target of image-text clustering (ITC) is to find correct clusters by integrating complementary and consistent information of multi-modalities for these heterogeneous samples. However, the majority of current studies analyse ITC on the ideal premise that the samples in every modality are complete. This presumption, however, is not always valid in real-world situations. The missing data issue degenerates the image-text feature learning performance and will finally affect the generalization abilities in ITC tasks. Although a series of methods have been proposed to address this incomplete image text clustering issue (IITC), the following problems still exist: 1) most existing methods hardly consider the distinct gap between heterogeneous feature domains. 2) For missing data, the representations generated by existing methods are rarely guaranteed to suit clustering tasks. 3) Existing methods do not tap into the latent connections both inter and intra modalities. In this paper, we propose a Clustering-Induced Generative Incomplete Image-Text Clustering(CIGIT-C) network to address the challenges above. More specifically, we first use modality-specific encoders to map original features to more distinctive subspaces. The latent connections between intra and inter-modalities are thoroughly explored by using the adversarial generating network to produce one modality conditional on the other modality. Finally, we update the corresponding modalityspecific encoders using two KL divergence losses. Experiment results on public image-text datasets demonstrated that the suggested method outperforms and is more effective in the IITC job. | ['Zhiwei Xu', 'Zhiyong Pei', 'Limin Liu', 'Jiatai Wang', 'Xiaoming Su', 'Dongjin Guo'] | 2022-09-28 | null | null | null | null | ['text-clustering'] | ['natural-language-processing'] | [ 4.42057908e-01 -1.78192914e-01 -2.57132977e-01 -2.94161052e-01
-9.50121224e-01 -4.49380070e-01 6.30051672e-01 -4.60463077e-01
-1.18292674e-01 6.34992898e-01 6.76114857e-02 7.63334706e-02
-2.00106531e-01 -3.24303508e-01 -6.68013573e-01 -1.11332417e+00
4.62847352e-01 3.50036919e-01 -1.22540183e-01 2.28903040e-01
4.52195592e-02 -7.32954368e-02 -1.62954819e+00 5.57850122e-01
1.14317441e+00 9.00372148e-01 3.21054041e-01 3.05396914e-01
-1.47283360e-01 6.68880820e-01 -3.77134144e-01 -4.24688905e-01
3.35112870e-01 -7.78007209e-01 -6.86298490e-01 5.10801315e-01
2.18662187e-01 2.28350516e-03 -2.23007053e-01 1.37169003e+00
3.73422891e-01 -1.88088641e-02 9.29146051e-01 -1.76045549e+00
-9.13203239e-01 6.19080544e-01 -8.76664996e-01 -3.62963885e-01
1.40967205e-01 -1.13088101e-01 7.81447589e-01 -8.97620916e-01
6.98063910e-01 1.28869712e+00 5.06506801e-01 7.67367542e-01
-1.07163000e+00 -6.94766462e-01 1.64324149e-01 1.95402190e-01
-1.60695648e+00 -3.21870983e-01 1.00435400e+00 -4.04356599e-01
1.96833134e-01 2.18669340e-01 2.82723814e-01 1.34511769e+00
7.70910382e-02 9.95181262e-01 1.48373818e+00 -4.03487802e-01
1.33443028e-01 4.91751850e-01 -2.12212458e-01 3.55853438e-01
-2.04858650e-03 -1.64115936e-01 -4.33665395e-01 6.01477772e-02
4.33206707e-01 1.58128828e-01 -4.85144794e-01 -4.06168610e-01
-1.28386724e+00 8.74607742e-01 1.95474580e-01 5.41091919e-01
-1.37393028e-01 -2.71724105e-01 4.37280864e-01 3.40914369e-01
3.34275782e-01 -8.00439119e-02 -3.53233069e-02 7.78072774e-02
-9.29415226e-01 -1.26221046e-01 5.22301316e-01 1.18840289e+00
8.34862769e-01 3.51200067e-02 8.27819705e-02 8.55908334e-01
2.85886854e-01 6.34392321e-01 6.41565740e-01 -8.97931278e-01
7.48465359e-01 5.04738152e-01 -3.15272629e-01 -1.04822648e+00
-1.26907349e-01 -3.33544672e-01 -1.32946241e+00 -7.86882937e-02
2.74935722e-01 -6.15085177e-02 -8.81066382e-01 1.83171237e+00
1.99807465e-01 2.40244344e-01 3.27500880e-01 9.36897099e-01
7.64393628e-01 6.27029121e-01 4.61037345e-02 -3.58197004e-01
1.04697013e+00 -8.31709981e-01 -9.96719480e-01 -2.73794066e-02
2.68382579e-01 -8.88674557e-01 9.60878968e-01 2.35496134e-01
-8.07291269e-01 -7.00550735e-01 -1.05514157e+00 8.63600373e-02
-4.57531989e-01 2.57602185e-01 3.40960145e-01 8.16362679e-01
-1.05192971e+00 7.79817477e-02 -6.52299941e-01 -4.56393093e-01
2.97180802e-01 3.26883733e-01 -4.97198701e-01 -4.92548525e-01
-1.15461922e+00 4.56400752e-01 6.16173506e-01 1.80353150e-01
-7.56489038e-01 -3.99853259e-01 -8.19501102e-01 -1.94464207e-01
4.36418325e-01 -6.94036782e-01 6.24794424e-01 -1.55633986e+00
-1.32117045e+00 7.31904447e-01 -1.48735955e-01 -1.53406903e-01
5.40629268e-01 2.28535056e-01 -6.25057220e-01 3.88120681e-01
3.46738875e-01 9.43320751e-01 1.16522217e+00 -1.97337866e+00
-7.26658642e-01 -4.08890277e-01 -3.96027446e-01 5.73640287e-01
-3.74785393e-01 -2.55418122e-01 -9.23612416e-01 -7.11765826e-01
3.61687720e-01 -1.07695377e+00 1.56708181e-01 -1.92280307e-01
-7.33286202e-01 -8.70862901e-02 1.25301290e+00 -5.51071167e-01
8.91773045e-01 -2.34424829e+00 2.82709122e-01 2.63253391e-01
-3.12239658e-02 -1.22892290e-01 -1.57739311e-01 4.64627206e-01
-1.23274773e-01 1.00064673e-01 -3.52474004e-01 -6.40014112e-01
7.75579587e-02 3.49888742e-01 -3.21097434e-01 6.40452325e-01
8.21955875e-02 8.34625781e-01 -4.87995893e-01 -9.18565810e-01
3.58634055e-01 5.99804819e-01 -4.14760262e-01 -5.07576503e-02
2.20304243e-02 6.96337938e-01 -4.19774383e-01 7.75736392e-01
9.15650547e-01 -3.12569350e-01 2.89847761e-01 -3.77419025e-01
2.11647868e-01 -6.19496584e-01 -1.25700092e+00 1.77530181e+00
-1.84377581e-01 4.42494094e-01 3.77592295e-02 -1.38750756e+00
5.88248134e-01 4.66783702e-01 7.74810016e-01 -6.03717625e-01
6.47062808e-02 2.72625476e-01 -1.40743747e-01 -5.68028450e-01
4.01138723e-01 -4.53281641e-01 -2.16482401e-01 2.72356093e-01
1.40081912e-01 9.37044434e-03 9.54204518e-03 2.79717594e-01
4.58475649e-01 1.11796945e-01 1.56476162e-02 2.44617630e-02
6.25794172e-01 -5.59282638e-02 7.43245840e-01 6.88992500e-01
-3.65196794e-01 1.07337439e+00 2.90058643e-01 2.00947165e-01
-9.44182932e-01 -1.16757143e+00 -2.15550169e-01 7.29689598e-01
4.39508349e-01 -1.53795123e-01 -6.79452479e-01 -8.01583648e-01
-2.32112989e-01 4.78985012e-01 -6.74778163e-01 -1.67004883e-01
-9.07160491e-02 -7.16479063e-01 5.86464107e-01 2.97729909e-01
7.67132640e-01 -8.15446913e-01 -3.35729159e-02 -6.77278712e-02
-7.37021387e-01 -1.33666098e+00 -5.99340856e-01 -3.91199403e-02
-6.65993512e-01 -1.02136564e+00 -9.11523223e-01 -1.08557928e+00
8.95842493e-01 4.43106413e-01 6.08850479e-01 -2.00429946e-01
-3.92982289e-02 7.45476663e-01 -6.50121391e-01 -1.45445028e-02
-5.13746202e-01 -3.31639103e-03 -8.64991918e-02 5.71815670e-01
3.69980425e-01 -3.68018389e-01 -4.57294792e-01 3.53941679e-01
-1.31758523e+00 3.12541574e-01 7.54352152e-01 1.00563407e+00
7.26795375e-01 6.74976349e-01 8.35475028e-01 -9.72183108e-01
3.78832132e-01 -5.78485191e-01 -7.37124216e-03 5.93454540e-01
-5.12512624e-01 -1.47731364e-01 8.42852354e-01 -4.24962372e-01
-1.39757216e+00 3.96399945e-01 1.32777661e-01 -8.02074790e-01
-4.26173270e-01 5.92988908e-01 -6.24474883e-01 1.70405075e-01
1.72582075e-01 7.62818217e-01 2.09104329e-01 -1.25282317e-01
4.06979114e-01 9.12277937e-01 6.30485773e-01 -7.41831899e-01
9.35648799e-01 8.22545886e-01 -2.20869824e-01 -8.41895819e-01
-6.72496796e-01 -6.18258953e-01 -6.93589687e-01 -2.91845828e-01
1.19003725e+00 -1.16476882e+00 -3.86780500e-01 5.36297262e-01
-8.18156362e-01 6.27349094e-02 -1.95542835e-02 5.51856160e-01
-6.47046685e-01 7.96074212e-01 -3.58483613e-01 -6.79165959e-01
-9.79374722e-02 -1.39574826e+00 1.16362298e+00 2.01970115e-01
2.72362560e-01 -1.12116480e+00 -2.79572546e-01 6.54360473e-01
1.33151546e-01 2.25229561e-01 9.29235756e-01 -4.36009467e-01
-4.83452618e-01 -8.31944942e-02 -1.81907207e-01 4.89676654e-01
3.21850389e-01 -9.59113091e-02 -1.01140463e+00 -4.57314044e-01
1.31031364e-01 -5.22905231e-01 8.55136871e-01 4.32463080e-01
1.22831988e+00 -1.53254449e-01 -2.77196258e-01 5.97709715e-01
1.65051627e+00 2.95868278e-01 7.80703366e-01 9.99132171e-02
8.12495768e-01 8.00047040e-01 4.85450178e-01 2.62721181e-01
5.39356053e-01 4.64597523e-01 2.96396375e-01 4.50529670e-03
-1.12140305e-01 -3.89109164e-01 5.10667741e-01 1.11494923e+00
8.34518000e-02 -5.11528909e-01 -6.61482453e-01 5.93088031e-01
-1.92724812e+00 -1.17259336e+00 -4.29033861e-02 2.00428700e+00
7.21833050e-01 -2.20120832e-01 -1.40210345e-01 1.43581942e-01
1.16686773e+00 -2.19634566e-02 -6.57830238e-01 2.70963341e-01
-5.72129905e-01 -3.50147039e-01 1.97758049e-01 1.37969285e-01
-1.32915163e+00 6.38812602e-01 5.24752522e+00 1.06359518e+00
-9.62615967e-01 2.20076084e-01 7.51607180e-01 8.75023976e-02
-4.68894541e-01 1.49938434e-01 -5.38574696e-01 7.06364334e-01
5.42362154e-01 4.11235876e-02 3.78805608e-01 5.98881960e-01
-2.43563298e-02 -5.15215471e-02 -9.46780264e-01 1.29874837e+00
4.34525192e-01 -8.85190904e-01 3.13006967e-01 1.32963628e-01
1.16916358e+00 -3.30107480e-01 4.71641481e-01 3.58896852e-01
-2.35733278e-02 -9.46030259e-01 7.05829859e-01 5.27921438e-01
1.00052321e+00 -8.32393050e-01 7.43657887e-01 4.09129560e-01
-1.10886741e+00 -1.34494781e-01 -3.25376064e-01 5.93137801e-01
9.54456721e-03 4.69000876e-01 -4.79690552e-01 1.06746614e+00
6.75894856e-01 9.79018211e-01 -8.07455063e-01 7.08919108e-01
1.37991697e-01 5.29788077e-01 -8.44378024e-02 3.50296736e-01
2.44470149e-01 -4.62709635e-01 3.91019195e-01 8.63828599e-01
4.32366818e-01 -2.21146286e-01 1.66578740e-01 8.65859866e-01
-3.21976356e-02 8.19856077e-02 -6.54119790e-01 4.04585823e-02
4.27788049e-01 1.11596620e+00 -8.67293119e-01 -2.43670121e-01
-6.41362369e-01 1.36644077e+00 -6.36064708e-02 7.55844653e-01
-8.93868804e-01 -1.72774747e-01 -7.67919468e-03 -3.08200091e-01
2.05241367e-01 2.00337216e-01 -3.53096992e-01 -1.39789486e+00
2.11326987e-01 -1.06059444e+00 5.60442626e-01 -6.96154654e-01
-1.73316669e+00 3.69116515e-01 -1.23045070e-03 -1.57245517e+00
-1.89637065e-01 -2.93231547e-01 -3.85057300e-01 6.82139933e-01
-1.43723428e+00 -1.62943780e+00 -2.60016501e-01 1.13758779e+00
6.81876123e-01 -2.72034943e-01 5.65476418e-01 4.71651345e-01
-6.78128362e-01 7.13520885e-01 5.94206393e-01 1.33251056e-01
9.51045632e-01 -1.12058771e+00 -4.97163534e-01 9.69366193e-01
4.85365130e-02 6.07685208e-01 4.99757886e-01 -6.71460092e-01
-1.35069263e+00 -1.23061275e+00 4.98688877e-01 -1.91067725e-01
3.64338815e-01 -4.12841439e-01 -8.65494490e-01 6.16476953e-01
4.81448889e-01 -7.21224472e-02 7.90672779e-01 -1.50263175e-01
-3.76339704e-01 -8.83573517e-02 -1.06373012e+00 4.87869829e-01
7.27002800e-01 -7.58496523e-01 -4.22903627e-01 1.89488158e-01
5.18148005e-01 -3.21347304e-02 -8.61275852e-01 3.60162288e-01
3.31151128e-01 -8.42444181e-01 8.32356691e-01 -2.67170727e-01
6.63375556e-01 -4.50316042e-01 -5.20330131e-01 -1.16172111e+00
2.58564949e-02 -2.74614960e-01 2.44398057e-01 1.48007035e+00
2.17141524e-01 -5.55588603e-01 6.75805926e-01 3.86034936e-01
-2.59740293e-01 -4.00185138e-01 -1.10424924e+00 -7.16756940e-01
3.33733588e-01 -4.37809795e-01 4.14472789e-01 1.34437430e+00
5.53419162e-03 2.30455309e-01 -6.23417258e-01 2.70834774e-01
9.02834713e-01 3.26653928e-01 5.30615449e-01 -1.02726316e+00
-1.83503792e-01 -2.58349359e-01 -3.34159642e-01 -7.86625504e-01
4.47607338e-01 -1.01763713e+00 1.48227438e-01 -1.39185143e+00
7.04763770e-01 -4.43022788e-01 -2.23533347e-01 3.36610854e-01
-2.50745445e-01 3.61612737e-01 3.45051885e-01 6.12442315e-01
-7.84553587e-01 8.25750053e-01 1.37166214e+00 -3.49464446e-01
9.48553011e-02 -1.25394672e-01 -6.87635899e-01 4.75570679e-01
7.92328894e-01 -3.44510347e-01 -7.92509437e-01 -1.44695833e-01
-7.41529325e-03 1.96369067e-01 4.95089054e-01 -1.05072987e+00
4.86384392e-01 -1.29633665e-01 4.13086414e-01 -8.63657415e-01
2.91316599e-01 -1.14608395e+00 3.49951178e-01 3.40060294e-02
-3.55453461e-01 -2.51014680e-01 -1.90589949e-01 9.87928748e-01
-6.10411048e-01 -2.12573752e-01 8.07274342e-01 -5.70697198e-03
-6.88454211e-01 2.71887183e-01 -3.03391844e-01 1.11520976e-01
1.14855111e+00 -3.58829439e-01 -2.73327529e-01 -6.13155961e-01
-6.79426968e-01 2.45726034e-01 5.98714292e-01 6.04080975e-01
7.64623523e-01 -1.59278488e+00 -6.50297821e-01 3.23851228e-01
3.18869442e-01 -1.39664769e-01 7.77316153e-01 1.03790116e+00
-1.15717910e-01 4.01612371e-01 -1.04280382e-01 -8.53683949e-01
-1.11337459e+00 7.41662145e-01 2.67513841e-01 -7.79222101e-02
-5.04669011e-01 3.09470087e-01 5.74354231e-01 -5.52140296e-01
2.04011261e-01 1.95057958e-01 -1.16291977e-01 1.58773780e-01
1.72410607e-01 -2.50378717e-02 -2.47064605e-01 -1.02499890e+00
-1.50292277e-01 6.37396514e-01 -1.27161279e-01 -2.39957824e-01
1.09094596e+00 -6.68650508e-01 -4.88912575e-02 5.24208665e-01
1.53197825e+00 -2.89318740e-01 -1.29519176e+00 -2.52624035e-01
-3.26680869e-01 -3.86717767e-01 -1.59082338e-01 -5.80249608e-01
-1.36088502e+00 8.41495454e-01 7.43631124e-01 1.87731713e-01
1.43272424e+00 3.74488011e-02 6.08459592e-01 1.83504689e-02
1.91787556e-01 -1.37896025e+00 2.51807392e-01 8.07373449e-02
6.25112951e-01 -1.47867262e+00 -1.82307407e-01 -4.44158435e-01
-1.11069000e+00 1.02257097e+00 6.16949201e-01 2.84449607e-01
6.43218458e-01 -5.37012629e-02 1.05186440e-01 -1.85907096e-01
-5.15016854e-01 -2.59831369e-01 3.05954725e-01 8.15565467e-01
1.91461086e-01 -3.59775908e-02 -1.30350173e-01 5.06118894e-01
1.75627679e-01 -3.08968663e-01 4.82091576e-01 6.66792095e-01
-7.62881264e-02 -1.05207729e+00 -6.83311820e-01 3.49151045e-01
-4.42036778e-01 1.78147271e-01 -3.53955716e-01 9.86266673e-01
3.81979793e-01 1.21599221e+00 -7.15930015e-02 -4.27222222e-01
-1.42465577e-01 7.76386037e-02 2.81346589e-01 -3.26841295e-01
-1.16322823e-01 5.15859783e-01 -3.84992599e-01 -1.88691661e-01
-9.88985956e-01 -8.92149031e-01 -1.09864366e+00 -9.22377557e-02
-4.22888815e-01 4.40652966e-02 5.61926782e-01 8.30341697e-01
2.82215565e-01 3.68777692e-01 8.55985284e-01 -6.38480186e-01
-3.92543018e-01 -7.48903811e-01 -9.94872868e-01 8.98563623e-01
8.01868662e-02 -6.39931738e-01 -6.20074749e-01 4.19384569e-01] | [8.627264976501465, 4.35468864440918] |
e682fa3c-8fc9-4e13-b321-ae2a1fe24224 | hgcn4mesh-hybrid-graph-convolution-network | null | null | https://aclanthology.org/2020.acl-srw.4 | https://aclanthology.org/2020.acl-srw.4.pdf | HGCN4MeSH: Hybrid Graph Convolution Network for MeSH Indexing | Recently deep learning has been used in Medical subject headings (MeSH) indexing to reduce the time and monetary cost by manual annotation, including DeepMeSH, TextCNN, etc. However, these models still suffer from failing to capture the complex correlations between MeSH terms. To this end, we introduce Graph Convolution Network (GCN) to learn the relationship between these terms, and present a novel Hybrid Graph Convolution Net for MeSH index (HGCN4MeSH). Basically, we utilize two BiGRUs to learn the embedding representation of the abstract and the title of the MeSH index text respectively. At the same time, we establish the adjacency matrix of MeSH terms based on the co-occurrence relationships in Corpus, which is easy to apply for GCN representation learning. On the basis of learning the mixed representation, the prediction problem of the MeSH index keywords is transformed into an extreme multi-label classification problem after the attention layer operation. Experimental results on two datasets show that HGCN4MeSH is competitive compared with the state-of-the-art methods. | ['Yujiu Yang', 'Chenhui Li', 'Miaomiao Yu'] | 2020-07-01 | null | null | null | acl-2020-6 | ['extreme-multi-label-classification'] | ['methodology'] | [ 1.60346217e-02 2.26550445e-01 -2.02634841e-01 -1.23814449e-01
-3.26643258e-01 3.44772302e-02 2.74816960e-01 4.69850659e-01
-2.13754773e-01 4.66253191e-01 3.08999807e-01 -1.45250708e-01
-4.54152972e-01 -1.06028914e+00 -5.22458494e-01 -6.61725402e-01
-1.33985028e-01 6.17859244e-01 -1.99321479e-01 -3.21168303e-02
1.48733985e-03 2.14399010e-01 -1.35304785e+00 1.77826002e-01
8.59996200e-01 1.19755805e+00 2.38602728e-01 1.42205536e-01
-6.50629818e-01 8.90953481e-01 -2.48852015e-01 -3.43135446e-01
-3.63062888e-01 -1.66757703e-01 -1.07427239e+00 -5.64649403e-02
3.08318019e-01 2.90581640e-02 -7.86133826e-01 1.42300510e+00
6.04190826e-01 9.67930183e-02 8.72355223e-01 -1.26566684e+00
-8.04501772e-01 7.08343387e-01 -7.89766669e-01 1.29663683e-02
1.99199006e-01 -4.28480655e-01 1.20676124e+00 -7.18848109e-01
5.90900123e-01 1.41944444e+00 8.56544018e-01 2.91519701e-01
-7.64596999e-01 -8.00972044e-01 1.11713193e-01 3.36001486e-01
-1.64708674e+00 1.29629657e-01 8.18601191e-01 -4.22670215e-01
7.83486903e-01 1.35654494e-01 5.17150104e-01 1.06843150e+00
4.40048575e-01 5.59495032e-01 4.67504650e-01 -5.78090735e-02
-4.25020069e-01 -1.87212065e-01 2.34882236e-01 1.32676852e+00
1.48821756e-01 -2.73001492e-01 7.20034316e-02 -3.00535142e-01
6.11824334e-01 3.78968090e-01 -4.74079907e-01 2.57567950e-02
-1.21817303e+00 7.63851821e-01 8.42707217e-01 3.99230808e-01
-4.10674900e-01 3.15963238e-01 7.05048978e-01 1.65057287e-01
7.26884604e-01 7.17109293e-02 -3.79611999e-01 3.26807678e-01
-6.04803920e-01 1.89225674e-01 7.72682488e-01 1.00140595e+00
6.19670928e-01 -2.92020619e-01 -2.94894993e-01 1.15417707e+00
6.05929196e-01 5.78611605e-02 4.99144822e-01 -3.39616090e-01
6.68563604e-01 1.02021646e+00 -6.18758917e-01 -1.62326300e+00
-7.55086482e-01 -6.67230606e-01 -1.63164759e+00 -2.76137471e-01
-1.11052886e-01 -1.81365584e-03 -8.21377158e-01 1.44392145e+00
3.65447193e-01 4.51010942e-01 -2.01544583e-01 5.33992052e-01
1.46202826e+00 7.71442890e-01 2.79003561e-01 -1.25822574e-01
1.61912525e+00 -1.06293356e+00 -1.03821182e+00 2.12356642e-01
9.31971133e-01 -7.09309936e-01 7.49637127e-01 3.85199711e-02
-7.06960738e-01 -6.05160177e-01 -8.67592156e-01 -2.86075354e-01
-7.55253851e-01 1.22302845e-01 8.71404707e-01 -1.09919995e-01
-8.86636615e-01 8.65568638e-01 -5.78665018e-01 -1.26868516e-01
5.49473345e-01 5.45004785e-01 -3.57561946e-01 -2.85590172e-01
-1.67780769e+00 6.88527882e-01 6.21842802e-01 3.75709295e-01
-3.05022061e-01 -7.37553835e-01 -1.14645767e+00 3.51892233e-01
3.86148751e-01 -7.74658203e-01 6.75638795e-01 -4.36735988e-01
-9.78299677e-01 1.02728939e+00 3.30466896e-01 1.29532889e-01
1.67363673e-01 9.26849395e-02 -6.81566596e-01 3.94308865e-02
2.36768872e-01 4.11998183e-01 3.63086730e-01 -8.62740815e-01
-4.47014689e-01 -3.42264742e-01 1.19641786e-02 1.90155178e-01
-5.66138983e-01 -1.29594654e-01 -8.84030581e-01 -8.47266734e-01
6.07031537e-03 -8.28397036e-01 -1.55410439e-01 6.06724396e-02
-7.05417871e-01 -8.88698697e-01 5.28905213e-01 -7.13586628e-01
1.48624074e+00 -2.11787271e+00 5.08754194e-01 4.13386256e-01
7.91786432e-01 7.31447712e-02 -1.81856334e-01 4.70421702e-01
-2.05330804e-01 2.31089666e-01 -1.06371641e-01 -2.84932941e-01
-1.07571915e-01 3.25871229e-01 1.44828767e-01 5.03000855e-01
9.78560150e-02 9.45985496e-01 -1.04357409e+00 -1.13412547e+00
-1.65720806e-02 6.58917785e-01 -4.19083357e-01 2.32463643e-01
-1.87045172e-01 8.30456540e-02 -6.67487979e-01 7.38396645e-01
6.15362346e-01 -8.56708825e-01 7.22814873e-02 -6.58831120e-01
2.95287877e-01 1.35080785e-01 -1.01389790e+00 1.97865927e+00
-5.25299788e-01 1.18190326e-01 3.57318260e-02 -1.21439886e+00
5.34602165e-01 4.18352991e-01 7.88647354e-01 -4.61768985e-01
4.47912127e-01 2.79864818e-01 -1.26805857e-01 -7.51961470e-01
-6.12711087e-02 5.22942729e-02 1.09577529e-01 6.97085038e-02
3.10500473e-01 3.35950732e-01 7.53444806e-02 2.07273722e-01
9.39443350e-01 -1.82758093e-01 1.45839557e-01 -4.61279988e-01
7.77830243e-01 -3.28156769e-01 4.22673881e-01 3.60179961e-01
4.50080484e-01 4.68285292e-01 6.62759423e-01 -5.33869088e-01
-5.12046695e-01 -5.12792706e-01 -2.36369997e-01 7.16038465e-01
2.39062190e-01 -5.96449673e-01 -6.69942558e-01 -8.33183229e-01
1.10256933e-01 4.77757938e-02 -9.76999164e-01 -3.42282623e-01
-5.64511657e-01 -9.42205191e-01 3.97768110e-01 3.41729045e-01
5.66674650e-01 -1.23721159e+00 3.92585546e-01 2.51515687e-01
-2.03960329e-01 -8.58679593e-01 -7.67934084e-01 4.74470966e-02
-5.67375541e-01 -1.25529265e+00 -9.00538802e-01 -1.30823541e+00
5.70372820e-01 -1.00930981e-01 1.15183115e+00 7.96595395e-01
-4.79085267e-01 4.29419130e-02 -2.01655388e-01 -1.15917578e-01
-5.34861162e-02 4.72599834e-01 -2.72606045e-01 1.13560610e-01
1.28222749e-01 -6.66998088e-01 -6.92120790e-01 -4.83556092e-02
-9.15997863e-01 2.32005715e-01 6.78876162e-01 1.07624841e+00
7.92662978e-01 3.20687681e-01 4.12715256e-01 -1.29734969e+00
7.55784869e-01 -9.03295934e-01 -3.75489503e-01 4.04789567e-01
-9.50033188e-01 1.48995612e-02 5.27941048e-01 -4.68785465e-01
-5.48488975e-01 -3.12400490e-01 -3.22666705e-01 -6.35094523e-01
2.12069213e-01 1.06391275e+00 -2.07403570e-01 -2.43518248e-01
-8.91758036e-03 1.46005660e-01 6.39345646e-02 -7.00675845e-01
2.50183433e-01 7.90044844e-01 2.19107717e-01 -3.58442038e-01
3.15859497e-01 4.34229970e-02 3.78854692e-01 -5.66994905e-01
-1.03736353e+00 -4.56745267e-01 -2.14242876e-01 9.98440236e-02
1.20199037e+00 -8.24768364e-01 -9.82185960e-01 3.19831997e-01
-1.34700775e+00 3.99366766e-02 1.98035613e-01 4.75060135e-01
-2.67798454e-01 4.57855940e-01 -8.61282110e-01 -3.92134413e-02
-6.23992801e-01 -1.30760765e+00 1.25926876e+00 8.54290724e-02
5.03220484e-02 -1.33533454e+00 5.10289147e-02 1.48125812e-01
2.46838495e-01 4.33145255e-01 1.52359068e+00 -5.83558857e-01
-1.00401171e-01 -3.87751788e-01 -7.31673956e-01 1.77384257e-01
1.75190404e-01 -2.59007722e-01 -5.65695584e-01 -1.96431056e-01
-2.29002669e-01 -8.94762948e-02 9.75845754e-01 3.42966735e-01
2.12182093e+00 -4.12397295e-01 -8.61869335e-01 8.57289374e-01
1.45593047e+00 -1.72881931e-02 5.14403045e-01 3.25674005e-02
1.51411390e+00 4.15463477e-01 2.78074265e-01 8.22664201e-02
6.41838193e-01 5.73347270e-01 5.59014142e-01 -3.94021600e-01
-3.22816253e-01 -6.27232492e-02 -4.54739064e-01 1.33677900e+00
-1.70575708e-01 -5.56044757e-01 -1.02012205e+00 3.81815255e-01
-1.94276261e+00 -5.28796434e-01 -2.61988282e-01 1.51664698e+00
8.75089288e-01 1.21721603e-01 -4.23844397e-01 -6.71087205e-02
7.42305517e-01 2.89361775e-01 -2.52311885e-01 1.00084424e-01
1.90796003e-01 2.66557544e-01 3.94775748e-01 2.79929727e-01
-1.26712728e+00 5.99591136e-01 4.95938492e+00 1.30749857e+00
-8.02240610e-01 1.21747933e-01 5.20105183e-01 3.28258276e-01
-2.51519620e-01 -3.34430844e-01 -5.78371823e-01 6.18914723e-01
6.50423884e-01 5.74879311e-02 4.14698154e-01 5.43970644e-01
-4.58494842e-01 7.03848064e-01 -9.70068634e-01 1.36930585e+00
1.22616123e-02 -1.42529726e+00 5.34043789e-01 2.92344272e-01
4.38926518e-01 -6.89566880e-02 -1.61798179e-01 6.35384500e-01
5.24188131e-02 -1.27798712e+00 -3.85945942e-03 7.46137917e-01
1.02133811e+00 -7.32484818e-01 1.20210409e+00 6.15884289e-02
-1.60443175e+00 1.03555240e-01 -4.84243304e-01 2.85663426e-01
-1.05148517e-01 5.49457252e-01 -1.22399300e-01 1.00087535e+00
5.45101881e-01 1.12472963e+00 -4.44093466e-01 8.65782022e-01
1.28152132e-01 3.07972044e-01 1.04406580e-01 -1.12950981e-01
5.08332729e-01 -2.28759155e-01 1.34985626e-01 1.21566522e+00
2.55402505e-01 2.81199187e-01 6.72834039e-01 7.79742360e-01
-7.36457586e-01 5.41628957e-01 -5.04768193e-01 -2.83011109e-01
5.99628203e-02 1.53387654e+00 -6.22299910e-01 -4.25199211e-01
-5.73791265e-01 9.89316702e-01 6.35794997e-01 2.42758691e-01
-7.50235021e-01 -8.02538991e-01 4.28798735e-01 -4.29215096e-02
-8.78003016e-02 2.99678952e-01 1.21414214e-01 -1.02121770e+00
-7.47634992e-02 -7.64852941e-01 5.87009966e-01 -4.95478630e-01
-1.58496833e+00 8.44757974e-01 -7.95854852e-02 -1.15801036e+00
3.75474751e-01 -6.92573309e-01 -4.34849322e-01 9.35678482e-01
-1.49500716e+00 -1.40045345e+00 -4.53096688e-01 5.62660754e-01
3.93089771e-01 -2.16344193e-01 1.05344558e+00 9.40010607e-01
-6.01421833e-01 6.60003006e-01 8.17919746e-02 4.29729313e-01
3.45675707e-01 -1.26031005e+00 9.95986909e-02 -1.53692707e-01
-6.13190904e-02 5.29331386e-01 9.76734608e-02 -7.76173115e-01
-1.19831061e+00 -1.32522190e+00 1.03200436e+00 7.75322020e-02
7.84183443e-01 -2.30810806e-01 -1.09924042e+00 3.42821181e-01
2.59639680e-01 3.48742157e-01 7.30734646e-01 2.31742233e-01
-2.30730727e-01 6.73328666e-03 -8.17739487e-01 5.14482498e-01
1.28560770e+00 -5.05334914e-01 -3.03832203e-01 9.21234906e-01
1.20733929e+00 -5.12931883e-01 -1.33601594e+00 6.11727178e-01
3.61194044e-01 -1.06413327e-01 1.05270839e+00 -7.83170700e-01
7.81144977e-01 -1.82112977e-02 3.65612172e-02 -1.16464007e+00
-6.86638296e-01 -6.58961153e-03 -2.46772170e-01 1.15791678e+00
2.57678479e-01 -4.62328374e-01 3.12566817e-01 2.03124266e-02
-2.70833015e-01 -1.33159184e+00 -7.96097159e-01 -2.06536248e-01
2.50666160e-02 3.38281691e-02 6.41620874e-01 1.31679153e+00
-3.96730155e-02 7.03634322e-01 -3.80684346e-01 2.35275447e-01
6.46979332e-01 2.25591227e-01 6.15681745e-02 -1.65604746e+00
-4.93977629e-02 -6.39940262e-01 -6.62259638e-01 -7.11222589e-01
5.18993914e-01 -1.43637633e+00 -2.41204470e-01 -1.93144464e+00
2.54855663e-01 -5.43348074e-01 -7.76232541e-01 5.47318697e-01
-3.98997605e-01 -2.94269361e-02 -2.96473384e-01 1.24162324e-01
-6.39229715e-01 6.56886101e-01 1.47255397e+00 -7.25083828e-01
3.02586704e-01 -2.10946083e-01 -5.80426693e-01 6.16198599e-01
3.92412454e-01 -5.51655829e-01 -3.93811613e-01 -5.06998837e-01
4.85745788e-01 2.34897897e-01 1.30955994e-01 -8.78284633e-01
2.36901268e-01 2.97820952e-04 2.64977902e-01 -6.14834368e-01
-1.49758169e-02 -9.86748457e-01 1.69456765e-01 4.31112796e-01
-4.84141409e-01 -3.82635705e-02 1.64668798e-01 8.61346424e-01
-3.20586950e-01 -2.12878585e-01 3.61303836e-01 -3.10206473e-01
-3.18518370e-01 1.05216801e+00 3.05052530e-02 1.19963787e-01
5.70570946e-01 4.42926019e-01 -5.48714399e-02 1.22116685e-01
-8.35268736e-01 6.65834308e-01 -1.40828207e-01 5.46673298e-01
6.24241769e-01 -1.67597735e+00 -5.21077454e-01 -1.58471316e-02
1.81046322e-01 3.19335490e-01 4.17899251e-01 8.59263420e-01
-6.16591811e-01 2.47842118e-01 6.88040704e-02 -2.85571009e-01
-1.19839489e+00 8.65289748e-01 4.08679545e-01 -7.46762097e-01
-7.98987091e-01 9.43804204e-01 3.58314842e-01 -5.83528340e-01
6.28865421e-01 -3.51088971e-01 -8.07473719e-01 1.88979670e-01
2.95084149e-01 1.53164029e-01 1.40431061e-01 -5.38276076e-01
-4.29781258e-01 8.68546903e-01 -3.85435760e-01 6.26125872e-01
1.35688353e+00 3.27298306e-02 -6.92388833e-01 3.53513598e-01
1.84079432e+00 -4.04605955e-01 -2.08817646e-01 -3.93944263e-01
-1.31764561e-01 9.22204256e-02 4.15637732e-01 -3.82971495e-01
-1.59827816e+00 9.20642018e-01 7.30961502e-01 2.00626031e-01
7.98060894e-01 -2.06615962e-02 1.08459997e+00 3.47185284e-01
2.57063354e-03 -7.73182213e-01 -2.33623743e-01 3.01617682e-01
9.72503245e-01 -1.15727663e+00 1.65449426e-01 -5.73288083e-01
-7.69248530e-02 1.25596786e+00 5.00180721e-01 -1.53439194e-01
1.39449477e+00 -1.36551172e-01 -2.12768808e-01 -1.08296812e+00
-3.53524506e-01 -8.82611126e-02 6.87403262e-01 4.54474986e-02
5.39231420e-01 2.80006174e-02 -5.84406435e-01 4.41681415e-01
1.04033828e-01 -1.09312467e-01 -1.85509816e-01 5.07783234e-01
1.16971955e-02 -9.18206811e-01 2.17629075e-01 9.93604600e-01
-7.48621762e-01 -4.52854037e-01 -1.22928858e-01 5.81225812e-01
3.46818238e-01 5.43631434e-01 -2.27777399e-02 -6.76128566e-01
2.76607841e-01 -4.88651507e-02 1.81002796e-01 -6.37975752e-01
-6.63704693e-01 7.98586830e-02 -1.27480835e-01 -4.67801273e-01
-3.42970967e-01 -9.48996246e-02 -1.33066630e+00 -1.49236619e-01
-4.88268942e-01 3.33418280e-01 2.77845532e-01 8.68750393e-01
3.01504701e-01 1.22048330e+00 3.10899705e-01 -6.37853324e-01
-3.37864548e-01 -1.21048498e+00 -8.27122927e-01 5.87138653e-01
2.47454762e-01 -9.83690321e-01 -2.12868020e-01 -3.67424637e-01] | [7.454965591430664, 6.32596492767334] |
f741bc00-2d55-4080-ad60-d15e7c98c9e1 | rubq-a-russian-dataset-for-question-answering | 2005.10659 | null | https://arxiv.org/abs/2005.10659v1 | https://arxiv.org/pdf/2005.10659v1.pdf | RuBQ: A Russian Dataset for Question Answering over Wikidata | The paper presents RuBQ, the first Russian knowledge base question answering (KBQA) dataset. The high-quality dataset consists of 1,500 Russian questions of varying complexity, their English machine translations, SPARQL queries to Wikidata, reference answers, as well as a Wikidata sample of triples containing entities with Russian labels. The dataset creation started with a large collection of question-answer pairs from online quizzes. The data underwent automatic filtering, crowd-assisted entity linking, automatic generation of SPARQL queries, and their subsequent in-house verification. | ['Pavel Braslavski', 'Vladislav Korablinov'] | 2020-05-21 | null | null | null | null | ['knowledge-base-question-answering'] | ['natural-language-processing'] | [-7.89701223e-01 6.79575861e-01 8.73650685e-02 -1.74057394e-01
-1.52208602e+00 -9.66129005e-01 3.72396618e-01 6.46866083e-01
-5.30210495e-01 1.45648468e+00 3.83966804e-01 -3.84525269e-01
-5.64657211e-01 -1.14925039e+00 -8.44625711e-01 5.11484087e-01
5.68156123e-01 1.43021798e+00 8.97148490e-01 -1.04767597e+00
1.06929511e-01 -1.20098423e-02 -1.25123203e+00 4.02337551e-01
1.65101659e+00 7.87900031e-01 -2.57225186e-01 8.06461632e-01
-6.24870062e-01 1.74445009e+00 -8.92751515e-01 -1.56648219e+00
-6.44339994e-03 -2.14315671e-02 -1.60687518e+00 -9.00838315e-01
9.07020509e-01 5.89494780e-02 -5.25956988e-01 9.43117917e-01
5.46140194e-01 9.14223194e-02 1.04025647e-01 -1.09057426e+00
-1.55087161e+00 7.86003709e-01 5.20888269e-01 4.26452368e-01
1.05192697e+00 -2.54337341e-01 1.44918537e+00 -9.69230592e-01
1.44732213e+00 8.13701749e-01 5.30785978e-01 4.38491344e-01
-7.98459232e-01 -3.06814760e-01 -1.05803764e+00 6.82963073e-01
-1.53694260e+00 -3.29405218e-01 7.37173632e-02 -5.06172121e-01
1.33204520e+00 4.07872945e-01 3.10058266e-01 5.48123062e-01
-5.87791800e-01 2.89733112e-01 8.06287825e-01 -5.44421732e-01
-1.42470419e-01 9.84301507e-01 7.95520246e-01 8.18072140e-01
3.80899757e-01 -4.72317845e-01 -4.12479311e-01 -3.55227530e-01
1.10585153e-01 -9.72970068e-01 -3.59771401e-01 -1.25535652e-01
-1.02799129e+00 4.84965444e-01 4.05087411e-01 2.73347311e-02
-1.72333881e-01 -3.30886632e-01 3.81763637e-01 8.51691127e-01
4.06288058e-02 1.18275404e+00 -1.19132507e+00 -3.28087844e-02
-7.85330012e-02 9.26165164e-01 1.26947188e+00 1.45427644e+00
1.16873276e+00 -6.95128977e-01 -4.33027506e-01 7.92300105e-01
1.49339080e-01 7.12288499e-01 4.00256038e-01 -8.77087414e-01
1.35628068e+00 1.31395161e+00 7.84019351e-01 -9.63205636e-01
-2.85728991e-01 1.49577767e-01 1.73337191e-01 -7.01189637e-01
8.37417841e-01 -2.31443450e-01 -3.12464416e-01 1.10320210e+00
7.59282649e-01 -4.10375983e-01 8.10396433e-01 6.42306626e-01
2.20295739e+00 3.83305430e-01 4.96896990e-02 1.55422077e-01
1.95484352e+00 -1.06251502e+00 -1.03015983e+00 4.70358104e-01
9.97429609e-01 -7.29890227e-01 1.22325885e+00 -2.56570011e-01
-1.07114756e+00 -4.76491362e-01 -5.34750283e-01 -8.11058760e-01
-1.12674880e+00 2.28271484e-01 2.12434322e-01 7.49883890e-01
-7.86563516e-01 9.05218050e-02 1.97768524e-01 -4.58570004e-01
6.13179877e-02 4.41626757e-02 -7.86088049e-01 7.66807720e-02
-1.82784879e+00 1.29067576e+00 7.72171021e-01 -2.07804903e-01
-2.32181609e-01 -1.30005431e+00 -1.00780785e+00 -1.83735728e-01
7.52769649e-01 -9.39690888e-01 1.31104994e+00 -2.23384216e-01
-1.31909347e+00 1.37020123e+00 1.03260085e-01 -7.27146447e-01
1.50826856e-01 -2.01894209e-01 -1.18206632e+00 2.98242062e-01
4.55812931e-01 1.00996599e-01 2.33939961e-01 -7.70219743e-01
-5.93089819e-01 -4.41352487e-01 4.19885069e-01 7.76271746e-02
1.82452127e-01 4.62115586e-01 -3.91996771e-01 -1.27818078e-01
-2.51947790e-01 -4.39930439e-01 2.44286925e-01 -8.48222435e-01
-4.69624281e-01 -7.06281424e-01 1.43620268e-01 -1.45335233e+00
1.46351552e+00 -1.57019401e+00 -2.32621565e-01 7.17104077e-02
3.68461698e-01 3.96594435e-01 3.36131714e-02 7.08054125e-01
-1.34887531e-01 2.12843999e-01 4.33799863e-01 6.23278260e-01
2.58416951e-01 8.24182201e-03 -6.80070817e-01 -2.48751447e-01
2.28937790e-01 1.56159544e+00 -1.14722371e+00 -9.45682824e-01
-3.15508097e-01 -4.25353289e-01 -3.97010773e-01 2.23662943e-01
-5.76847792e-01 -1.56459048e-01 -5.28616011e-01 8.20988119e-01
4.28791404e-01 -3.74268442e-01 -4.66598086e-02 -4.75947291e-01
2.72590041e-01 7.69304097e-01 -1.04169500e+00 1.29164112e+00
-2.18654275e-01 5.00832558e-01 -3.94174516e-01 -2.28898674e-01
1.08477163e+00 4.79682386e-01 8.13110843e-02 -1.23054981e+00
-2.52135724e-01 7.34960198e-01 -6.39830589e-01 -1.34682858e+00
1.31331694e+00 3.68489802e-01 -4.86069620e-01 2.41634715e-02
5.09303808e-01 -3.24162692e-01 8.41941893e-01 6.64093435e-01
1.27902031e+00 -1.90601461e-02 4.20818657e-01 -4.17316929e-02
9.47149098e-01 1.00054908e+00 2.47652292e-01 5.74067831e-01
1.28398798e-02 2.59262472e-01 3.58228832e-01 -5.63194275e-01
-1.02583706e+00 -1.19448125e+00 -2.50556916e-01 7.62462199e-01
1.07536949e-01 -7.34544694e-01 -7.35896051e-01 -9.16022897e-01
2.64469862e-01 7.61025488e-01 -6.16384208e-01 2.38124877e-01
-6.87515914e-01 3.41082923e-02 1.06411409e+00 4.50941101e-02
3.85017782e-01 -1.22785258e+00 -4.36078869e-02 2.09278807e-01
-6.83110118e-01 -1.45489788e+00 -6.90920204e-02 -3.51284653e-01
-3.87658298e-01 -1.78029525e+00 -1.94530785e-01 -8.75330627e-01
2.87114263e-01 -2.71747828e-01 2.20557237e+00 6.43114597e-02
-1.05507500e-01 5.18862307e-01 -6.18631959e-01 -2.96276569e-01
-5.97016990e-01 2.73679346e-01 -4.19764370e-01 -7.50988960e-01
8.86442184e-01 1.05803661e-01 1.33282831e-02 2.52037466e-01
-7.46441126e-01 -6.21591747e-01 -1.74355105e-01 6.62087739e-01
5.03602505e-01 -5.01446009e-01 9.68943298e-01 -1.38654816e+00
7.09466696e-01 -7.69434631e-01 -1.00468826e+00 1.22992563e+00
-2.06974000e-01 2.21810415e-02 5.09646833e-01 2.62021929e-01
-1.13450003e+00 -4.78365690e-01 -2.61300862e-01 -4.78212116e-03
1.47410780e-01 6.25018656e-01 -2.44924217e-01 -6.07237853e-02
1.45986354e+00 -2.66743928e-01 -3.92257571e-01 -5.42365193e-01
1.09673679e+00 7.56171048e-01 6.93973958e-01 -6.56023562e-01
9.79103565e-01 -2.15464205e-01 -6.90847099e-01 -5.60774744e-01
-9.81922030e-01 -6.72632873e-01 -5.00604391e-01 -1.67561412e-01
9.06247079e-01 -1.20642412e+00 -9.91946280e-01 -6.75001964e-02
-9.66294944e-01 -5.23988381e-02 -9.52845156e-01 9.34101939e-02
-5.01517594e-01 1.12271085e-02 -5.64790547e-01 -3.92646581e-01
-5.18859923e-01 -3.78192067e-01 5.18781662e-01 6.21959627e-01
-2.07829535e-01 -8.27855885e-01 7.29126513e-01 1.56734717e+00
-7.06821606e-02 -1.91142738e-01 1.09457934e+00 -1.26508701e+00
-7.18827605e-01 -2.61024803e-01 -3.92246425e-01 -5.90281785e-02
-3.77666265e-01 1.19191736e-01 -4.81640249e-01 4.05519038e-01
-8.06453943e-01 -1.09005988e+00 2.92964935e-01 -5.98206222e-01
3.38205725e-01 -3.86185884e-01 1.57733902e-01 -8.84676203e-02
1.56640232e+00 -3.26617628e-01 9.34067190e-01 7.01123655e-01
6.26462162e-01 8.92683983e-01 7.77431726e-01 -2.02002451e-01
1.25457549e+00 8.09095562e-01 1.02578245e-01 5.63405454e-01
-2.32051611e-01 -6.64677203e-01 -1.29811108e-01 1.16683066e+00
1.44461811e-01 1.93485953e-02 -1.14521587e+00 1.03226948e+00
-1.64903414e+00 -1.02796257e+00 -6.33151114e-01 1.85780108e+00
1.20283651e+00 -4.81834501e-01 1.94104359e-01 -4.17405158e-01
5.16908407e-01 -4.89197493e-01 -3.34429562e-01 -9.98515263e-02
-4.75693464e-01 6.29754245e-01 5.63510478e-01 4.57538575e-01
-7.07419872e-01 1.35267174e+00 6.73752213e+00 7.91830838e-01
1.52771235e-01 2.45541811e-01 -1.04606345e-01 4.33112025e-01
-9.71217930e-01 1.01302139e-01 -1.09764230e+00 2.54747778e-01
1.14996314e+00 -5.90118170e-01 3.30885261e-01 6.61415100e-01
-5.59268832e-01 6.44446388e-02 -3.27047139e-01 6.97994471e-01
1.28225192e-01 -1.87822080e+00 -4.55439016e-02 -5.49951196e-01
7.53518939e-01 2.21369177e-01 -6.76069438e-01 1.09577227e+00
9.85506296e-01 -8.23711336e-01 7.41718829e-01 1.00239754e+00
7.45948255e-01 -4.61600691e-01 8.77215087e-01 2.45566979e-01
-8.80288184e-01 -2.19442576e-01 -5.61985970e-01 4.59311485e-01
3.11565697e-02 4.20955151e-01 -8.41221988e-01 1.16898620e+00
9.89672899e-01 1.02625199e-01 -1.14436936e+00 7.87556231e-01
-4.66242790e-01 2.88132966e-01 -1.62522569e-01 -4.22757745e-01
-1.95071504e-01 -3.76500338e-01 3.68605167e-01 7.68990397e-01
4.41484824e-02 2.55298615e-01 -3.73015285e-01 7.24307895e-01
-7.19285846e-01 7.79854238e-01 -4.19307739e-01 -1.14183940e-01
9.07033265e-01 1.45406067e+00 9.64446273e-03 -7.31694758e-01
-3.57818216e-01 4.46079552e-01 1.16979110e+00 4.02439147e-01
-7.13626981e-01 -1.00212121e+00 4.90930825e-01 1.20422974e-01
1.83423951e-01 2.67137825e-01 3.36904749e-02 -1.50508845e+00
4.68159646e-01 -1.20591033e+00 9.85681951e-01 -1.21020317e+00
-1.49852109e+00 5.29646754e-01 -3.86382580e-01 -8.37246954e-01
-4.31291074e-01 -4.16837662e-01 1.15267849e-02 9.46820080e-01
-1.61398923e+00 -1.04280448e+00 -4.39285606e-01 9.07302797e-01
2.73951748e-03 -4.45168614e-01 8.18027794e-01 1.01771915e+00
-6.00198209e-01 6.64460182e-01 -2.16884613e-02 6.65710986e-01
9.93951261e-01 -1.46874285e+00 3.41186941e-01 3.75438422e-01
3.74196380e-01 7.00270295e-01 5.64036489e-01 -8.73117864e-01
-1.33439910e+00 -9.96379137e-01 1.78506315e+00 -1.53066909e+00
1.32305777e+00 -3.95330526e-02 -1.30396652e+00 7.72098780e-01
4.04940754e-01 7.74598941e-02 8.48869145e-01 1.20635852e-01
-7.00075686e-01 -1.40411317e-01 -1.34564602e+00 3.05683821e-01
7.45826185e-01 -1.01120150e+00 -1.22433627e+00 6.83146596e-01
1.12949169e+00 -8.78475130e-01 -1.38674784e+00 2.33624250e-01
5.57180047e-02 -4.32302564e-01 9.58014190e-01 -1.43622148e+00
2.55959958e-01 -8.94809365e-01 -1.23442180e-01 -7.87397683e-01
2.64937162e-01 -3.58723581e-01 -4.42585677e-01 1.46724021e+00
1.20795763e+00 -5.25979161e-01 7.92397380e-01 8.56221735e-01
-4.17579226e-02 -1.61640659e-01 -1.19369662e+00 -5.70256352e-01
-1.12641891e-02 -2.30904240e-02 8.31522644e-01 1.27726126e+00
3.76320958e-01 7.24330366e-01 1.75991371e-01 2.00086534e-01
1.09389335e-01 5.11796288e-02 1.15762305e+00 -9.99653280e-01
1.21288281e-02 1.11866824e-01 -4.46968615e-01 -6.43244088e-01
1.70678809e-01 -1.06993878e+00 -3.51129025e-01 -1.90412056e+00
-2.06755206e-01 -5.59909463e-01 5.90530336e-01 2.49809429e-01
-5.60283244e-01 2.30315685e-01 -1.63571700e-01 1.35505557e-01
-1.14352548e+00 4.12901312e-01 9.49592650e-01 3.67314033e-02
-3.04301437e-02 -2.59749234e-01 -6.34028554e-01 2.88245946e-01
4.82146740e-01 -5.53243995e-01 -2.44222835e-01 -3.96538168e-01
1.44752479e+00 2.40468428e-01 3.01285595e-01 -6.65442050e-01
4.94002312e-01 1.95211068e-01 -2.82700777e-01 -6.78230822e-01
5.80533929e-02 -6.32408619e-01 2.19678625e-01 -9.56174626e-04
-2.43629172e-01 4.20572698e-01 -6.82047606e-02 2.95230567e-01
-6.73642039e-01 -7.86922574e-01 2.67522424e-01 -1.59125596e-01
-7.82824457e-01 -6.95546269e-02 -3.63257667e-03 1.23135805e+00
1.01772118e+00 1.84030160e-02 -1.07979047e+00 -5.64016104e-02
-6.96238697e-01 6.74681067e-01 1.99923232e-01 4.32858557e-01
2.26203367e-01 -1.38290942e+00 -9.49380338e-01 -2.94895142e-01
8.01188648e-01 -2.57189393e-01 4.74057615e-01 4.85575736e-01
-1.09203172e+00 6.71866000e-01 -5.34254946e-02 2.06200480e-01
-1.01912761e+00 5.98085463e-01 4.45530504e-01 -6.47178471e-01
-3.81592631e-01 8.11643898e-01 -1.09819043e+00 -1.43528819e+00
-1.54578820e-01 -6.78402632e-02 -1.04085553e+00 3.71321350e-01
6.40801430e-01 6.97709084e-01 5.08523583e-01 -6.08651996e-01
-1.87508270e-01 2.60061502e-01 2.03731775e-01 -1.35916444e-02
7.84789801e-01 -1.82375506e-01 -4.42826778e-01 1.78022534e-01
8.04822981e-01 6.94375753e-01 9.74282399e-02 -6.31653249e-01
5.84640741e-01 -2.71082699e-01 -7.89285362e-01 -1.08004081e+00
-6.02443993e-01 8.65467638e-02 -1.01764398e-02 4.37052041e-01
5.74885488e-01 5.64811885e-01 1.01739609e+00 1.12914872e+00
4.42336977e-01 -1.34782648e+00 -1.60402700e-01 1.10334194e+00
7.09128082e-01 -1.27905977e+00 -2.88131595e-01 -5.76541841e-01
-5.89805424e-01 7.71711111e-01 8.99176419e-01 1.17089808e-01
5.38089633e-01 -4.33784485e-01 4.81042147e-01 -6.35185421e-01
-7.09691107e-01 -7.13141859e-01 3.75534862e-01 7.81338274e-01
1.04552448e-01 -1.58237666e-01 -3.03101212e-01 9.76000965e-01
-8.37369740e-01 1.21197581e-01 5.46545327e-01 3.41206700e-01
-2.34379396e-01 -8.34020197e-01 -1.27407864e-01 4.91763860e-01
-4.23630625e-01 -4.13472056e-01 -5.42710960e-01 9.79920864e-01
-1.35574743e-01 1.00363970e+00 -8.19443390e-02 -1.32671759e-01
1.23914444e+00 2.58943647e-01 3.54475260e-01 -6.73076987e-01
-1.28265870e+00 -1.51625526e+00 1.07627428e+00 -3.36059660e-01
-1.92218080e-01 -2.36316353e-01 -1.14678693e+00 -2.91746914e-01
-4.70734775e-01 9.88977194e-01 5.43101490e-01 9.26585734e-01
6.01557314e-01 5.51276840e-02 1.34834319e-01 8.98880661e-01
-4.09099013e-01 -9.45141137e-01 -2.96421140e-01 7.02381432e-01
-1.02782570e-01 -3.12432259e-01 -3.85556556e-02 7.67865032e-03] | [10.718945503234863, 7.940372943878174] |
e240c425-fba4-4d10-a399-d57805185500 | multi-domain-learning-for-motion | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Singh_Multi_Domain_Learning_for_Motion_Magnification_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Singh_Multi_Domain_Learning_for_Motion_Magnification_CVPR_2023_paper.pdf | Multi Domain Learning for Motion Magnification | Video motion magnification makes subtle invisible motions visible, such as small chest movements while breathing, subtle vibrations in the moving objects etc. But small motions are prone to noise, illumination changes, large motions, etc. making the task difficult. Most state-of-the-art methods use hand-crafted concepts which result in small magnification, ringing artifacts etc. The deep learning based approach has higher magnification but is prone to severe artifacts in some scenarios. We propose a new phase based deep network for video motion magnification that operates in both domains (frequency and spatial) to address this issue. It generates motion magnification from frequency domain phase fluctuations and then improves its quality in the spatial domain. The proposed models are lightweight networks with fewer parameters ( 0.11M and 0.05M). Further, the proposed networks performance is compared to the SOTA approaches and evaluated on real-world and synthetic videos. Finally, an ablation study is also conducted to show the impact of different parts of the network. | ['G. Sankara Raju Kosuru', 'Subrahmanyam Murala', 'Jasdeep Singh'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['motion-magnification'] | ['computer-vision'] | [ 2.25711420e-01 -1.08439758e-01 6.13423921e-02 3.34390849e-01
-3.01434174e-02 -4.55503374e-01 3.41037095e-01 -4.55160528e-01
-3.35241228e-01 5.92953742e-01 2.25647390e-01 -2.81732492e-02
-1.23927146e-01 -5.37880301e-01 -6.06918514e-01 -7.68953204e-01
-3.77618611e-01 -5.01264513e-01 6.52612984e-01 -1.13009445e-01
1.39158100e-01 1.28180429e-01 -1.30954766e+00 8.14546570e-02
3.00600529e-01 5.69365382e-01 2.64751613e-01 1.05362487e+00
4.89929944e-01 9.68106568e-01 -1.05245173e+00 2.12754592e-01
2.92531997e-01 -4.48190004e-01 -4.42206949e-01 -6.71442375e-02
5.59685707e-01 -6.86320066e-01 -4.03917134e-01 9.88755345e-01
8.18361998e-01 2.47224048e-01 4.03462470e-01 -1.02403879e+00
-5.20985186e-01 4.49964553e-01 -1.06829488e+00 7.41178334e-01
2.68622190e-01 4.53722864e-01 2.51191795e-01 -4.87710983e-01
4.98659939e-01 1.27463973e+00 1.11562252e+00 6.30704820e-01
-8.41854572e-01 -6.02818489e-01 -1.50172174e-01 2.72806466e-01
-9.19125259e-01 -2.24100471e-01 1.02309895e+00 -1.05349898e-01
8.65786314e-01 2.54727095e-01 5.48459113e-01 1.25576711e+00
4.62232798e-01 1.83962569e-01 7.52890468e-01 -2.47781992e-01
7.44923726e-02 -4.01766859e-02 -3.86480838e-01 4.86099660e-01
5.12549758e-01 2.00288758e-01 -4.39286590e-01 1.47071555e-01
1.15964854e+00 -4.05141823e-02 -7.47807503e-01 5.92132211e-02
-1.43388295e+00 4.00279135e-01 4.31231767e-01 5.64758778e-01
-2.50475258e-01 7.63357639e-01 3.62081856e-01 1.99145854e-01
1.52778164e-01 3.24703127e-01 -2.70558953e-01 -4.02217776e-01
-1.14578629e+00 1.56492144e-01 4.90763962e-01 5.22953987e-01
2.19528690e-01 5.19122601e-01 -1.15630709e-01 3.92344028e-01
5.19922674e-01 5.57925105e-01 7.03988194e-01 -1.10335064e+00
5.03535509e-01 2.88178362e-02 2.79189378e-01 -1.42994595e+00
-8.91262233e-01 -4.38874751e-01 -1.06729984e+00 3.23389828e-01
3.63720804e-01 -4.42230403e-01 -8.06258559e-01 1.89173090e+00
3.23388517e-01 5.67496896e-01 3.61632183e-02 1.15419805e+00
1.00186098e+00 8.54948878e-01 -1.65919691e-01 -2.76107311e-01
1.30916190e+00 -9.96721804e-01 -1.19207120e+00 -2.39453781e-02
2.91066974e-01 -9.24552977e-01 1.16939509e+00 3.70501757e-01
-1.16029525e+00 -9.33710515e-01 -1.34021115e+00 1.98294416e-01
1.11412920e-01 1.07738219e-01 2.50539452e-01 8.24730694e-01
-1.20056307e+00 1.04592991e+00 -8.48369658e-01 -3.34182978e-01
2.72239983e-01 3.92638445e-01 1.09070010e-01 4.45018172e-01
-1.31681776e+00 6.28953695e-01 1.65186942e-01 2.62092054e-01
-7.33334541e-01 -6.59444451e-01 -6.44569159e-01 -1.13127999e-01
1.89678237e-01 -8.66702974e-01 1.16737831e+00 -1.05413759e+00
-1.72125423e+00 1.90799639e-01 6.73198178e-02 -4.15058970e-01
7.65743673e-01 -4.12960738e-01 -5.61997771e-01 6.31334424e-01
-3.36215824e-01 6.88221991e-01 1.31519890e+00 -9.90519762e-01
-3.97490174e-01 2.41443083e-01 2.84357101e-01 1.10169098e-01
-3.67091447e-01 -2.42169335e-01 3.82935591e-02 -1.15400815e+00
-5.20373993e-02 -1.01087201e+00 2.87152100e-02 1.96309000e-01
-2.65268922e-01 3.11688513e-01 1.28427982e+00 -6.38916492e-01
1.48259473e+00 -2.02244210e+00 -2.46095750e-02 -1.53594449e-01
3.25704902e-01 4.86711830e-01 -8.10302719e-02 2.61809736e-01
-1.68569356e-01 2.00259537e-01 5.78296557e-02 -1.06453165e-01
-4.48373705e-01 -8.68720263e-02 1.90523431e-01 5.53496480e-01
6.90172985e-02 5.56993067e-01 -7.34145641e-01 -1.34386182e-01
3.35109383e-01 7.29115546e-01 -6.33430302e-01 1.42890038e-02
2.71066159e-01 6.20799422e-01 -2.32713297e-02 4.54596937e-01
8.11423302e-01 -2.99064964e-01 5.43900244e-02 -8.22188139e-01
-1.60437286e-01 -8.06195289e-02 -1.41357458e+00 1.44757473e+00
-4.25964683e-01 1.12364578e+00 -1.08163156e-01 -6.73782051e-01
4.16864455e-01 7.49694049e-01 3.56479764e-01 -4.63621169e-01
4.91079718e-01 8.40095878e-02 2.82112241e-01 -1.11917913e+00
4.07904595e-01 -1.68934062e-01 4.37177330e-01 3.77858639e-01
-1.86990932e-01 2.73789018e-01 4.31709960e-02 -1.28011897e-01
1.19963789e+00 1.57305986e-01 2.30720595e-01 -2.33543158e-01
4.23148304e-01 -4.14845377e-01 4.73666131e-01 5.14560878e-01
-4.71643329e-01 6.59470797e-01 2.31077522e-01 -6.50973618e-01
-1.14571297e+00 -9.43752170e-01 3.48392457e-01 4.23990309e-01
4.92699355e-01 -1.75774664e-01 -8.15394163e-01 -2.65988082e-01
-3.76898438e-01 1.21040300e-01 -4.96071875e-01 -3.95516276e-01
-9.13140476e-01 -8.92030299e-01 6.48568630e-01 6.48508787e-01
9.30116236e-01 -1.02468681e+00 -1.20824802e+00 2.20474333e-01
-4.67718899e-01 -1.29409468e+00 -5.14899611e-01 -4.13766742e-01
-1.07787192e+00 -9.10501301e-01 -9.79534924e-01 -6.84941828e-01
5.09733617e-01 4.75316375e-01 8.52651298e-01 -6.57369047e-02
-3.51485074e-01 1.66151166e-01 -3.22088152e-01 -4.16043878e-01
-3.47183675e-01 -2.35992283e-01 3.49527836e-01 -1.22427382e-01
-2.19275318e-02 -9.43248928e-01 -1.06078780e+00 2.77188301e-01
-9.85699177e-01 -1.16687082e-01 5.85990965e-01 6.04124010e-01
-4.22304459e-02 4.48088646e-01 4.55418527e-01 -4.45923150e-01
6.38228893e-01 -1.34463385e-01 -3.72885257e-01 -4.01836127e-01
-1.19005904e-01 -9.64180753e-02 7.30474055e-01 -1.16425717e+00
-8.94799590e-01 -4.19695854e-01 2.99455702e-01 -5.51609516e-01
-4.00037616e-02 6.91589564e-02 2.43923947e-01 -2.00520054e-01
8.30901802e-01 -4.23986465e-02 1.37668531e-02 -3.87648404e-01
-4.52086460e-05 3.87526631e-01 5.67525566e-01 2.32269362e-01
1.31024575e+00 8.38271797e-01 3.32318902e-01 -1.10269511e+00
-3.47704142e-01 -1.54765006e-02 -2.86559910e-01 -4.63463545e-01
1.07038498e+00 -9.97752190e-01 -7.84071147e-01 5.92490077e-01
-1.20021558e+00 -3.53744686e-01 2.05064863e-02 7.79449642e-01
-2.87337005e-01 6.80098712e-01 -8.71652007e-01 -6.30918145e-01
-5.68801701e-01 -9.47225988e-01 7.04347789e-01 6.72033727e-01
-4.86379266e-01 -1.04114223e+00 4.47469316e-02 2.04478011e-01
7.31646895e-01 5.53195000e-01 6.83444738e-01 2.61579335e-01
-7.41573513e-01 6.79894462e-02 1.43491507e-01 3.30478013e-01
3.77536356e-01 2.33860627e-01 -1.06466496e+00 -4.47281778e-01
3.24091613e-01 1.59039825e-01 5.71340740e-01 8.51807773e-01
8.11806560e-01 -5.51242590e-01 -1.49713144e-01 7.64401555e-01
1.47371745e+00 4.07315910e-01 8.54925275e-01 5.04026294e-01
7.86835015e-01 3.91852945e-01 4.50530708e-01 4.48836982e-01
-1.96579382e-01 5.10730982e-01 5.01397550e-01 -3.90161067e-01
-2.81810403e-01 3.15768182e-01 4.55708295e-01 8.41995418e-01
-4.89635140e-01 -4.63107467e-01 -4.52298939e-01 3.87984782e-01
-1.50695348e+00 -1.23675036e+00 -2.18292385e-01 1.83609462e+00
4.62514371e-01 3.17978770e-01 1.46256089e-01 5.65171659e-01
7.40573823e-01 4.37658399e-01 -3.89715821e-01 -3.10228884e-01
1.10654607e-02 1.01006322e-01 5.61487734e-01 2.96898931e-01
-1.22632813e+00 2.10562319e-01 6.12631941e+00 4.40798491e-01
-1.53118062e+00 2.01405257e-01 3.99393529e-01 -5.00833690e-01
-2.89823096e-02 -4.15247709e-01 -3.48432779e-01 6.28101587e-01
7.23047316e-01 1.95186168e-01 2.92298019e-01 4.12743151e-01
9.04022157e-01 -2.54954576e-01 -8.06698442e-01 1.24586809e+00
-7.08948672e-02 -1.22170866e+00 -2.65506282e-02 -2.76348919e-01
6.68631971e-01 -2.42147163e-01 3.66208524e-01 -1.84504628e-01
-5.84322155e-01 -9.40130830e-01 5.63978374e-01 3.02250296e-01
8.08639824e-01 -8.13114226e-01 8.53215277e-01 1.80111200e-01
-1.33425891e+00 -1.97962239e-01 -3.20488095e-01 -2.71373093e-01
2.75673747e-01 5.31503499e-01 -6.18758678e-01 2.16571733e-01
8.76117945e-01 5.66619933e-01 -3.95794809e-01 8.51945937e-01
8.24018791e-02 6.37564003e-01 -3.41334611e-01 -1.46493047e-01
3.32471371e-01 2.12360337e-01 9.18536961e-01 1.05171013e+00
5.04088104e-01 -1.09825268e-01 -3.75088125e-01 8.48249972e-01
1.25013560e-01 -4.25971389e-01 -5.52566767e-01 1.80118486e-01
3.29244643e-01 1.29011524e+00 -6.45827174e-01 -3.39282811e-01
-3.86250645e-01 8.84498596e-01 -5.69801271e-01 5.54787278e-01
-1.24286389e+00 -5.53140819e-01 5.63286424e-01 3.84663790e-01
2.35177591e-01 -2.56073415e-01 1.79341584e-01 -1.01266146e+00
2.38673180e-01 -8.66566420e-01 -1.06422037e-01 -8.29773128e-01
-7.59705961e-01 5.00633299e-01 -2.78928261e-02 -1.80212963e+00
-1.93913698e-01 -5.80633402e-01 -9.60975707e-01 5.38618505e-01
-1.31837165e+00 -5.20584762e-01 -7.24253297e-01 3.54182631e-01
6.61535382e-01 -4.88559389e-03 3.14940453e-01 7.14793146e-01
-5.35015523e-01 5.44951320e-01 -9.83572453e-02 -1.63054261e-02
8.61484349e-01 -9.54377413e-01 3.87460768e-01 1.04199994e+00
-3.25388491e-01 4.79400873e-01 1.07467711e+00 -4.22676682e-01
-1.20298254e+00 -1.12582803e+00 4.27342027e-01 -1.43763736e-01
3.13724995e-01 -1.87550306e-01 -7.66023159e-01 2.63415754e-01
3.89498711e-01 3.82715017e-02 3.20984334e-01 -7.13171303e-01
2.29830891e-01 -1.80038571e-01 -1.21390688e+00 7.96424568e-01
9.41298544e-01 -1.04609229e-01 -3.78945827e-01 6.62246495e-02
8.39831889e-01 -4.90354240e-01 -6.60446405e-01 4.43682611e-01
6.34715617e-01 -1.13777316e+00 9.36547160e-01 -6.96684942e-02
4.45821255e-01 -7.19367743e-01 2.88743436e-01 -1.19532514e+00
-4.92503315e-01 -1.18536651e+00 -4.56897438e-01 8.66476059e-01
2.63045430e-01 -4.47813779e-01 7.35285521e-01 -2.90374488e-01
2.00956985e-01 -5.76774120e-01 -7.77311563e-01 -8.67860556e-01
-4.71975118e-01 -4.03404832e-02 2.33437881e-01 1.02124310e+00
-2.05630526e-01 5.20313740e-01 -7.77314723e-01 3.68307412e-01
6.42712593e-01 -3.48150879e-01 7.18820393e-01 -8.04721117e-01
-5.44215977e-01 1.80987101e-02 -6.92753494e-01 -8.27968001e-01
-7.39319921e-01 7.62262642e-02 -2.17208907e-01 -1.21420407e+00
-3.85584533e-02 2.25858837e-01 -8.71461853e-02 -1.07120447e-01
-2.69857496e-01 5.21428585e-01 3.35956097e-01 6.56167045e-02
-2.41262600e-01 1.63168356e-01 1.47473478e+00 7.21189901e-02
-3.43535066e-01 -5.66983707e-02 -2.28057981e-01 1.00811160e+00
1.11396432e+00 -4.05034810e-01 -8.22512627e-01 -4.98506874e-01
1.91521585e-01 6.20701090e-02 4.39950168e-01 -1.59541249e+00
-2.52302689e-03 2.05572575e-01 6.67754829e-01 -6.14681900e-01
2.06014156e-01 -7.86508083e-01 4.33649123e-01 1.01146829e+00
-1.62805431e-02 5.13718247e-01 4.33040768e-01 5.60441434e-01
-5.63698374e-02 -2.03133896e-02 1.14378679e+00 -2.86958456e-01
-3.83474976e-01 -5.09811006e-02 -6.59786642e-01 -6.40736148e-02
8.61368597e-01 -5.23266077e-01 -3.17776948e-01 -7.93722272e-01
-5.98165214e-01 -3.46816301e-01 1.91134498e-01 3.92222911e-01
4.69178796e-01 -1.38293159e+00 -3.48151565e-01 -2.45261695e-02
-5.54182768e-01 5.91024272e-02 5.28774977e-01 1.09769142e+00
-1.07758760e+00 3.74516994e-01 -3.06903064e-01 -7.49781370e-01
-1.46512175e+00 5.46190560e-01 5.19521952e-01 -1.75283477e-01
-7.39320755e-01 7.59016752e-01 2.20850959e-01 2.91360885e-01
2.13389218e-01 -9.03050065e-01 -4.84122932e-01 -1.86966211e-01
7.36475110e-01 8.14689994e-01 -4.44969863e-01 -3.34800124e-01
-3.20969224e-01 1.12365711e+00 3.84661317e-01 -5.95043898e-02
9.49200690e-01 -2.94517845e-01 2.68140554e-01 3.61817837e-01
1.21378303e+00 5.17355762e-02 -1.32031822e+00 1.71915606e-01
-4.18621749e-01 -2.47457415e-01 -1.43768162e-01 -5.22214949e-01
-1.27461052e+00 9.17533040e-01 1.05267680e+00 1.55038044e-01
1.56103218e+00 -6.44018650e-01 9.65905070e-01 2.46497199e-01
-8.17134380e-02 -1.11188185e+00 6.56884372e-01 1.13736778e-01
5.21755636e-01 -9.23680544e-01 1.32239997e-01 -4.13467258e-01
-3.06030989e-01 1.36848855e+00 6.28102303e-01 -1.77442849e-01
4.26056027e-01 5.79177260e-01 3.18636835e-01 -1.19061284e-02
-7.05931544e-01 3.32387954e-01 1.83364004e-01 7.65722692e-01
5.09883642e-01 -3.58585328e-01 -3.30269396e-01 1.11335695e-01
-5.04968800e-02 2.33297706e-01 9.25811172e-01 1.06548154e+00
-4.43577975e-01 -3.79254043e-01 -6.76296651e-01 8.85486826e-02
-1.02673411e+00 1.58021122e-01 7.16328770e-02 9.91104126e-01
3.52144212e-01 9.52664375e-01 1.51805505e-01 -4.64635044e-01
3.24718744e-01 -4.52446938e-01 5.06000876e-01 -6.06335923e-02
-6.24721050e-01 5.00239551e-01 7.47791976e-02 -6.97815537e-01
-9.83023167e-01 -2.40804970e-01 -1.18895602e+00 -3.84489149e-01
-3.13305438e-01 -3.89702559e-01 3.17398727e-01 5.15391529e-01
2.22301185e-01 1.20565605e+00 5.38049817e-01 -1.03012896e+00
-4.76547390e-01 -1.24889231e+00 -4.02312815e-01 4.66195166e-01
7.43253350e-01 -5.57213485e-01 -6.81718707e-01 2.68533409e-01] | [10.876426696777344, -1.5235447883605957] |
d40d6b32-dfe6-4a5e-8ce7-01e78eda535e | synthetic-expressions-are-better-than-real | 2010.10979 | null | https://arxiv.org/abs/2010.10979v1 | https://arxiv.org/pdf/2010.10979v1.pdf | Synthetic Expressions are Better Than Real for Learning to Detect Facial Actions | Critical obstacles in training classifiers to detect facial actions are the limited sizes of annotated video databases and the relatively low frequencies of occurrence of many actions. To address these problems, we propose an approach that makes use of facial expression generation. Our approach reconstructs the 3D shape of the face from each video frame, aligns the 3D mesh to a canonical view, and then trains a GAN-based network to synthesize novel images with facial action units of interest. To evaluate this approach, a deep neural network was trained on two separate datasets: One network was trained on video of synthesized facial expressions generated from FERA17; the other network was trained on unaltered video from the same database. Both networks used the same train and validation partitions and were tested on the test partition of actual video from FERA17. The network trained on synthesized facial expressions outperformed the one trained on actual facial expressions and surpassed current state-of-the-art approaches. | ['László A Jeni', 'Jeffrey F Cohn', 'Itir Onal Ertugrul', 'Koichiro Niinuma'] | 2020-10-21 | null | null | null | null | ['facial-expression-generation'] | ['computer-vision'] | [ 4.46000487e-01 4.54523057e-01 7.39578679e-02 -4.78987902e-01
-3.77015889e-01 -3.22551459e-01 5.32018900e-01 -9.67063963e-01
-3.56453717e-01 7.80210972e-01 -7.59593546e-02 2.26934016e-01
3.59391391e-01 -5.95162094e-01 -1.02313387e+00 -8.25118482e-01
-1.30558074e-01 4.00101304e-01 2.05645729e-02 -7.02502728e-02
-1.31659999e-01 6.81584537e-01 -2.00094485e+00 7.22021699e-01
5.90790212e-02 1.48094440e+00 -1.69563562e-01 4.86645401e-01
1.30385324e-01 1.10360706e+00 -7.77824163e-01 -5.70875347e-01
5.55050790e-01 -7.62982368e-01 -6.98449671e-01 6.43766701e-01
7.40392447e-01 -5.95061898e-01 -3.77144307e-01 9.22384024e-01
2.01741487e-01 -2.48885248e-02 4.27869380e-01 -1.44372153e+00
-3.75603199e-01 1.35206878e-01 -2.41121233e-01 -4.46576886e-02
5.93986452e-01 8.78338143e-02 4.96652037e-01 -8.56831372e-01
1.13544130e+00 1.31264997e+00 6.14633322e-01 1.16184986e+00
-1.14277077e+00 -8.63744676e-01 -1.00429989e-01 1.04797721e-01
-1.27441561e+00 -8.24875295e-01 6.75496519e-01 -4.24544513e-01
8.92262101e-01 -1.08426467e-01 1.05216849e+00 1.71035612e+00
-9.10913497e-02 7.31267095e-01 1.23551285e+00 -5.01884460e-01
3.73343706e-01 1.37812356e-02 -5.84102452e-01 8.52754951e-01
-2.43237287e-01 2.87688136e-01 -5.53814948e-01 -6.19985536e-02
8.52763057e-01 -2.84752220e-01 -4.14482951e-01 -3.70029181e-01
-8.73063684e-01 7.80835986e-01 -5.28212041e-02 3.47100019e-01
-6.68471575e-01 1.52348414e-01 4.58040714e-01 5.37609577e-01
6.27763391e-01 2.67320901e-01 -3.43824059e-01 -2.93929428e-01
-1.04793763e+00 4.09086645e-01 7.91288376e-01 5.77229500e-01
6.49844348e-01 4.53034222e-01 5.86142689e-02 6.04186475e-01
1.81230605e-01 3.89441192e-01 4.54462588e-01 -1.10262394e+00
8.14384073e-02 5.96128106e-01 -1.14072993e-01 -8.32143545e-01
-1.26940534e-01 1.38835490e-01 -4.24861997e-01 6.18268847e-01
6.15392625e-01 -3.23460132e-01 -1.19038498e+00 1.84434557e+00
2.51967728e-01 3.37351769e-01 2.04861432e-01 7.51390219e-01
7.61932909e-01 6.41119719e-01 -1.08967498e-01 -2.05655187e-01
9.25904393e-01 -7.44332552e-01 -7.18628466e-01 8.94966349e-02
4.90853339e-01 -5.05107164e-01 5.62832475e-01 6.35967553e-01
-1.19895458e+00 -7.71363735e-01 -9.26473141e-01 4.36992049e-01
-1.53264940e-01 3.85237634e-01 3.04443568e-01 7.45486379e-01
-1.37475383e+00 5.62556148e-01 -5.80037534e-01 -4.39081162e-01
7.92121410e-01 5.26564240e-01 -9.92918491e-01 -6.33707792e-02
-9.69922662e-01 8.36633444e-01 2.69754559e-01 1.90466836e-01
-1.42442119e+00 -3.25028598e-01 -9.06697512e-01 -1.42161831e-01
2.09567979e-01 -3.22317898e-01 1.29497099e+00 -2.30684853e+00
-1.95603073e+00 1.32353401e+00 5.71623333e-02 -4.68923837e-01
6.99253857e-01 -1.63158864e-01 -4.73692745e-01 4.25179064e-01
-6.00763187e-02 9.38464582e-01 1.33159924e+00 -1.19283187e+00
-2.97771245e-01 -4.73741502e-01 5.59372902e-02 -2.42387012e-01
-1.25073776e-01 4.69406806e-02 -3.67688090e-01 -5.67101777e-01
-3.08870316e-01 -1.06907046e+00 1.52276844e-01 5.54162748e-02
-6.59400374e-02 -9.95908305e-02 1.06232274e+00 -6.16793871e-01
6.10865653e-01 -2.23716664e+00 3.25293362e-01 1.86920494e-01
-2.22000256e-01 5.13086677e-01 -5.12295127e-01 1.81763172e-01
-6.30173445e-01 -4.22506221e-02 2.16401834e-02 -3.18796337e-01
-4.91217464e-01 3.87787372e-01 -8.69869217e-02 5.03748119e-01
5.81766248e-01 4.86294121e-01 -7.69158721e-01 -2.31009260e-01
-6.86249556e-03 5.74303865e-01 -6.76107645e-01 5.78282833e-01
-4.06894714e-01 6.96113408e-01 -3.10080051e-01 7.95626640e-01
6.08100414e-01 2.26747468e-01 2.89814383e-01 -2.62259930e-01
2.48751178e-01 -2.50968486e-01 -9.04982567e-01 1.61708021e+00
-2.59091944e-01 7.63255537e-01 1.20513476e-01 -1.20989287e+00
1.04660785e+00 7.64340043e-01 6.78094029e-01 -6.72455013e-01
4.57302541e-01 8.34393129e-02 -1.16547249e-01 -1.03732336e+00
1.89546999e-02 -2.65724570e-01 2.67940491e-01 3.76679599e-01
6.47542000e-01 -7.97415301e-02 1.59115553e-01 -2.84290880e-01
1.20728779e+00 7.20931828e-01 1.09058805e-01 4.81738076e-02
7.16190457e-01 -3.21839154e-01 4.47432846e-01 1.87490344e-01
-6.93021491e-02 5.64823329e-01 8.49642396e-01 -1.12199855e+00
-9.99877632e-01 -7.75217116e-01 1.90146178e-01 1.01273525e+00
-4.74303961e-01 -4.31432575e-01 -1.24896395e+00 -1.17515671e+00
-2.60946929e-01 5.90570092e-01 -1.15795994e+00 -1.14809416e-01
-4.12808985e-01 -3.01610559e-01 7.33603418e-01 3.70289505e-01
5.35870194e-01 -1.42385554e+00 -7.10163414e-01 -6.53528869e-02
-1.97481327e-02 -1.39785087e+00 1.11683838e-01 -2.06259817e-01
-5.76326370e-01 -1.37187541e+00 -6.98418319e-01 -6.87002599e-01
9.26182091e-01 -2.97475636e-01 1.16127765e+00 1.71056584e-01
-1.55403703e-01 4.83366042e-01 -4.88626152e-01 -5.04159033e-01
-8.98896217e-01 -2.90235579e-01 1.01977371e-01 7.23918617e-01
3.76678944e-01 -6.62743628e-01 -1.50080979e-01 4.07854766e-01
-9.84871209e-01 -1.02807328e-01 4.85612392e-01 9.48319376e-01
3.83963436e-01 -1.58986747e-01 4.87435520e-01 -7.88002431e-01
1.70707539e-01 -3.14504623e-01 -6.43741131e-01 5.67200556e-02
4.46214378e-02 -1.09941520e-01 4.21845973e-01 -4.44299757e-01
-1.03825843e+00 7.29434013e-01 -4.29067791e-01 -1.00170553e+00
-3.66999596e-01 -1.98450405e-03 -3.42203289e-01 -1.27308458e-01
5.68370759e-01 4.17378619e-02 4.29052263e-01 -1.56027287e-01
1.29598677e-01 4.62486178e-01 4.96358037e-01 -5.52459896e-01
3.99195969e-01 4.32564110e-01 3.34203504e-02 -9.34570968e-01
-5.41679978e-01 2.02112690e-01 -8.93344343e-01 -6.63790762e-01
1.09394097e+00 -9.73357618e-01 -5.04434884e-01 8.23354721e-01
-1.25237441e+00 -4.38200414e-01 -2.14842930e-01 3.47643912e-01
-7.70151973e-01 -1.06280044e-01 -2.70981908e-01 -7.05126405e-01
-7.47589394e-02 -1.19026661e+00 1.16857278e+00 1.54302092e-02
-3.87443870e-01 -6.76702499e-01 9.15473551e-02 3.96254301e-01
1.50070459e-01 7.61826575e-01 5.85446000e-01 -5.23489177e-01
-2.77509034e-01 -3.67603570e-01 1.59730300e-01 9.10407603e-01
2.31753513e-01 4.63752508e-01 -1.26630580e+00 -5.33904061e-02
-1.96740180e-02 -9.24839973e-01 5.54354012e-01 1.31794453e-01
1.28165495e+00 -2.78103620e-01 -2.26054862e-02 5.41580617e-01
1.02820826e+00 4.67966557e-01 9.38179910e-01 1.46035284e-01
3.82874489e-01 8.41831148e-01 4.83741879e-01 3.00928563e-01
-1.40305132e-01 8.31491053e-01 7.13555634e-01 -1.08245790e-01
6.61440613e-03 -2.77505785e-01 7.50397205e-01 2.51628876e-01
-6.85729742e-01 -6.49264604e-02 -6.93945944e-01 3.30032736e-01
-1.49414372e+00 -1.22740650e+00 4.97905880e-01 1.94901609e+00
4.63410437e-01 3.99688035e-02 2.66435295e-01 5.62894382e-02
4.27204251e-01 7.45639130e-02 -3.83064032e-01 -6.01033926e-01
-4.83107232e-02 5.80853879e-01 -5.68215735e-02 1.03186756e-01
-1.18752778e+00 9.86630261e-01 6.64171124e+00 4.24369663e-01
-1.41969633e+00 8.19129124e-02 9.18829381e-01 -3.45983058e-01
2.17382506e-01 -3.97143722e-01 -4.37647700e-01 3.55834246e-01
1.18454731e+00 3.99548233e-01 3.30691457e-01 1.18195081e+00
1.10344544e-01 -6.74738139e-02 -1.29464459e+00 9.92160976e-01
5.36369145e-01 -1.32533014e+00 2.63226867e-01 2.71625668e-02
8.34065259e-01 -2.38750383e-01 -1.84159324e-01 3.24915797e-01
8.91447589e-02 -1.39824748e+00 7.78467536e-01 6.53653085e-01
1.03523695e+00 -6.40440643e-01 9.37318444e-01 3.20212618e-02
-7.75133431e-01 4.89994362e-02 3.73041518e-02 -1.56961679e-01
-1.27507731e-01 -1.06763318e-01 -8.89642060e-01 3.08902651e-01
9.02434230e-01 8.14396322e-01 -4.55288440e-01 2.11683646e-01
-2.46997043e-01 5.85102141e-01 -5.45839481e-02 2.40382954e-01
2.23438382e-01 -2.29525968e-01 2.89502323e-01 9.39000726e-01
4.44295317e-01 -2.24711336e-02 -8.03767964e-02 4.91830260e-01
-2.67897367e-01 -5.80614619e-02 -1.03186047e+00 -2.76277602e-01
-4.52741943e-02 1.28679860e+00 -5.06705046e-01 -4.43691283e-01
-6.46493912e-01 1.00674236e+00 2.43309617e-01 2.74054855e-01
-9.29113328e-01 2.40450323e-01 6.30465090e-01 3.86923879e-01
3.98699015e-01 3.14904720e-01 6.53930306e-01 -1.13856542e+00
-3.02150175e-02 -1.44357586e+00 1.14514835e-01 -1.06048131e+00
-9.43999052e-01 1.29904866e+00 9.49933976e-02 -1.16431844e+00
-8.34943950e-01 -9.80660737e-01 -5.82409084e-01 5.61302185e-01
-7.65951872e-01 -1.29803300e+00 -5.86208522e-01 7.89850712e-01
5.95793903e-01 -7.18475759e-01 1.12527907e+00 2.84769446e-01
-4.89716113e-01 4.80414331e-01 -3.95140082e-01 5.28819621e-01
5.51247716e-01 -6.64270759e-01 2.54011810e-01 5.40357292e-01
1.89468279e-01 5.04265651e-02 4.40377861e-01 -2.43132815e-01
-1.24853480e+00 -1.13147354e+00 5.79544365e-01 -3.09978664e-01
3.42284769e-01 -3.81217182e-01 -6.76542759e-01 1.00998747e+00
2.46143475e-01 3.77836645e-01 7.04386353e-01 -4.60900247e-01
-2.18711480e-01 -1.20010279e-01 -1.26679051e+00 3.67839605e-01
1.04745197e+00 -4.09996361e-01 -5.41478753e-01 1.33039042e-01
-4.00274061e-02 -4.83790696e-01 -7.89200544e-01 5.40984154e-01
8.82992327e-01 -1.26229548e+00 5.64149976e-01 -9.19127345e-01
7.13253617e-01 -1.81868989e-02 -3.56700063e-01 -1.40574396e+00
3.00976306e-01 -5.35441697e-01 1.15548722e-01 9.18886185e-01
1.31821007e-01 -4.73978430e-01 1.01948643e+00 4.33649719e-01
1.05321750e-01 -1.02176774e+00 -1.03801703e+00 -5.51369250e-01
-2.43914619e-01 -8.96987095e-02 4.88057941e-01 8.60920370e-01
-4.62739348e-01 1.77015424e-01 -4.63758379e-01 -3.51098835e-01
1.40658572e-01 -2.07773596e-01 1.17712688e+00 -1.02599716e+00
-1.85236588e-01 -1.15248039e-01 -9.54646111e-01 -2.86164939e-01
8.65340889e-01 -6.04174614e-01 -2.56495357e-01 -8.71628821e-01
-1.89206228e-01 1.57611623e-01 1.75202638e-01 8.16798985e-01
3.58240813e-01 6.46569252e-01 1.65811971e-01 -2.41821289e-01
-1.01519182e-01 6.01902008e-01 1.06146848e+00 -5.70894107e-02
1.07119776e-01 3.98815200e-02 -1.73003599e-01 9.96930778e-01
5.76018274e-01 -2.97932088e-01 -5.44181883e-01 -1.01387709e-01
-5.30381314e-02 2.21671894e-01 4.42196846e-01 -1.14960158e+00
-3.50048661e-01 1.01370037e-01 8.53055000e-01 -1.14844099e-01
6.23353362e-01 -8.98437560e-01 5.38019121e-01 3.82007569e-01
-1.72546938e-01 2.53423214e-01 3.21515739e-01 2.68953860e-01
-4.72903192e-01 -7.54522905e-02 8.36680174e-01 -3.43303710e-01
-7.83912122e-01 1.80142447e-01 -5.99350154e-01 -3.10239196e-01
1.30369639e+00 -4.17613953e-01 1.55418187e-01 -5.14753222e-01
-9.40792263e-01 -4.50786471e-01 5.76759517e-01 6.25390053e-01
6.89793050e-01 -1.39188218e+00 -7.81235397e-01 7.40793467e-01
-3.02738287e-02 -1.87878340e-01 6.50382191e-02 6.35502636e-01
-6.88204050e-01 -3.86075452e-02 -9.04443502e-01 -6.31998539e-01
-1.64769244e+00 5.06485283e-01 7.18366265e-01 4.64797253e-03
-2.63012886e-01 7.16628253e-01 2.31220484e-01 -3.21490049e-01
6.51624575e-02 -8.62942114e-02 -2.15828910e-01 1.49286851e-01
6.71079338e-01 2.08788930e-04 8.73476788e-02 -1.24955988e+00
-2.31741503e-01 5.85069358e-01 1.29287660e-01 -1.10428602e-01
1.41912997e+00 4.32646275e-01 -1.13311335e-01 2.35887721e-01
1.35346687e+00 -1.59297407e-01 -1.50907195e+00 1.14259034e-01
-2.13675722e-01 -4.49494898e-01 -2.50067174e-01 -6.19566619e-01
-1.61479008e+00 6.48142219e-01 6.46599710e-01 -7.53877014e-02
1.44661891e+00 -1.97384626e-01 3.26206416e-01 3.26240987e-01
3.97183061e-01 -9.29689884e-01 4.27736580e-01 3.25031549e-01
1.14167309e+00 -1.06355405e+00 -1.29508436e-01 -2.24248394e-01
-7.61132181e-01 1.52845371e+00 7.95925438e-01 -2.39815667e-01
5.79460561e-01 1.81721881e-01 3.44427675e-01 -3.51359338e-01
-7.35312879e-01 1.61682785e-01 1.44013569e-01 6.32701755e-01
3.45317304e-01 -2.18836442e-01 1.49234325e-01 2.91506380e-01
-1.99239552e-01 5.33344984e-01 3.47509325e-01 8.12996686e-01
1.95344552e-01 -1.00790858e+00 -3.71149451e-01 1.96625009e-01
-6.91815853e-01 3.51015389e-01 -6.86418891e-01 1.25160110e+00
4.69962686e-01 6.22844994e-01 2.97317386e-01 -6.07349694e-01
3.66111845e-01 3.43455553e-01 8.57779920e-01 -3.84630740e-01
-4.44951981e-01 -8.11312348e-02 4.07175839e-01 -9.93523955e-01
-9.40758765e-01 -8.40471864e-01 -7.51483560e-01 7.66882952e-03
2.00188503e-01 -1.95687823e-02 5.06901622e-01 9.29860234e-01
2.51694500e-01 2.74107486e-01 7.97628522e-01 -1.33675373e+00
-2.62338430e-01 -1.11215889e+00 -5.70795119e-01 7.63038218e-01
2.49278203e-01 -8.72231424e-01 -2.82288849e-01 5.92627525e-01] | [13.480672836303711, 1.5098791122436523] |
c2a81486-1e0b-4048-bdca-eaf0258ab0fd | biologically-inspired-continual-learning-of | 2211.05231 | null | https://arxiv.org/abs/2211.05231v2 | https://arxiv.org/pdf/2211.05231v2.pdf | Biologically-Inspired Continual Learning of Human Motion Sequences | This work proposes a model for continual learning on tasks involving temporal sequences, specifically, human motions. It improves on a recently proposed brain-inspired replay model (BI-R) by building a biologically-inspired conditional temporal variational autoencoder (BI-CTVAE), which instantiates a latent mixture-of-Gaussians for class representation. We investigate a novel continual-learning-to-generate (CL2Gen) scenario where the model generates motion sequences of different classes. The generative accuracy of the model is tested over a set of tasks. The final classification accuracy of BI-CTVAE on a human motion dataset after sequentially learning all action classes is 78%, which is 63% higher than using no-replay, and only 5.4% lower than a state-of-the-art offline trained GRU model. | ['Shih-Chii Liu', 'Joachim Ott'] | 2022-11-02 | null | null | null | null | ['temporal-sequences'] | ['reasoning'] | [ 1.51650667e-01 3.34032893e-01 -6.88203499e-02 2.06244081e-01
-8.50289941e-01 -3.91776472e-01 1.26298404e+00 -7.35119939e-01
-3.71731669e-01 8.21160436e-01 2.77045369e-01 -6.38406426e-02
2.83608973e-01 -6.45235777e-01 -1.18184543e+00 -1.02261710e+00
-4.12389077e-02 5.58784068e-01 4.77633625e-01 6.97630867e-02
-1.45090580e-01 1.97797075e-01 -1.69107354e+00 7.28443086e-01
2.84337133e-01 6.37356520e-01 2.98378587e-01 9.44804728e-01
2.35532507e-01 1.24807489e+00 -6.00988865e-01 -1.86349943e-01
3.30992080e-02 -1.02800477e+00 -7.17265069e-01 1.47730619e-01
7.45409727e-02 -5.68902969e-01 -4.58503723e-01 4.60858792e-01
3.12777936e-01 6.63050056e-01 1.22099888e+00 -1.24046099e+00
-8.49570572e-01 6.80593610e-01 -2.61203140e-01 2.28755116e-01
2.50195533e-01 6.91665709e-01 7.09446132e-01 -7.74121046e-01
9.78653729e-01 1.31664848e+00 6.23830080e-01 1.26240444e+00
-1.33437157e+00 -3.02202642e-01 -1.74355619e-02 3.29425186e-01
-1.12170839e+00 -2.71719038e-01 5.39922178e-01 -7.96671987e-01
1.48455811e+00 -6.90067634e-02 1.09181821e+00 2.10073066e+00
6.03405476e-01 1.28116298e+00 8.21728468e-01 -2.26663321e-01
7.64496565e-01 -4.75353450e-01 -1.25902921e-01 4.94608611e-01
-2.96012670e-01 5.08270681e-01 -6.98907495e-01 -4.66961078e-02
9.74552035e-01 -2.94927031e-01 -1.67768344e-01 -2.78668314e-01
-1.40983677e+00 8.29392791e-01 1.19658694e-01 5.68742752e-01
-7.26900339e-01 6.80364788e-01 2.71879792e-01 -6.88611716e-02
3.66678655e-01 3.79296243e-02 -2.16103598e-01 -4.67043668e-01
-1.15954816e+00 5.74907362e-01 5.09210944e-01 8.54232371e-01
4.58948076e-01 8.60620379e-01 -6.07449114e-01 6.68211520e-01
1.30520433e-01 2.71196783e-01 1.20875335e+00 -1.27095652e+00
9.08586383e-02 -1.67620227e-01 1.43012971e-01 -1.68924630e-01
-1.34352386e-01 -1.85659990e-01 -6.98813736e-01 2.20581755e-01
3.79079700e-01 -2.58455098e-01 -1.28353024e+00 1.81482732e+00
1.61256790e-01 8.62146914e-01 3.78799260e-01 8.70499790e-01
7.51690447e-01 1.14224231e+00 3.88389289e-01 -2.38911316e-01
9.69504297e-01 -1.26432610e+00 -6.58086419e-01 -4.43235971e-02
4.18850332e-01 -6.56363666e-02 8.61928046e-01 6.14970505e-01
-1.02401805e+00 -9.89915788e-01 -9.64909732e-01 1.85473844e-01
7.57823116e-04 3.52373160e-02 6.33243561e-01 6.79815710e-01
-1.19328868e+00 9.89278018e-01 -1.32372618e+00 -3.85200977e-01
3.11601579e-01 -1.05383590e-01 -2.39200726e-01 1.82589293e-01
-9.69306529e-01 7.70835400e-01 6.54601336e-01 -4.07646149e-01
-1.84746718e+00 -6.69986784e-01 -9.18625653e-01 -1.93380147e-01
-8.89336318e-03 -1.07337463e+00 1.47217000e+00 -7.89149106e-01
-2.07857275e+00 4.92889613e-01 -1.63585171e-01 -9.23806012e-01
7.80480385e-01 -3.59182358e-01 -2.33579695e-01 3.58233213e-01
-4.45398875e-02 1.17953277e+00 1.14686751e+00 -1.24823773e+00
-2.29347929e-01 1.03635758e-01 -4.02084529e-01 1.65903755e-02
7.87676573e-02 -3.83846611e-01 -2.45289803e-01 -8.88255417e-01
-3.63685966e-01 -1.23080623e+00 -1.61241323e-01 -3.50479871e-01
6.21256754e-02 -5.47212958e-01 8.38180125e-01 -8.57925355e-01
8.62898827e-01 -1.77814424e+00 5.33437848e-01 -5.61297536e-01
-2.97230780e-01 1.95747152e-01 -3.95298690e-01 4.25346941e-01
-1.48765296e-01 -9.76433530e-02 -5.08773208e-01 -8.15704525e-01
-1.42931134e-01 5.44136703e-01 -5.62561154e-01 1.30847618e-01
3.24641317e-01 1.21510839e+00 -1.02959883e+00 -2.33168587e-01
3.36158037e-01 5.83190262e-01 -5.62441230e-01 1.68752551e-01
-6.49624109e-01 7.00573027e-01 -8.78813490e-02 3.71249914e-01
2.13798285e-01 -1.58323675e-01 3.76254357e-02 3.70911390e-01
1.75666168e-01 -2.40973145e-01 -5.81982553e-01 2.03826761e+00
-2.87550896e-01 6.53552771e-01 -7.34412491e-01 -9.74064469e-01
7.17219710e-01 7.57444322e-01 6.24364257e-01 -4.19467300e-01
7.69961402e-02 4.45195846e-03 -6.04886077e-02 -6.41691506e-01
4.08166975e-01 -3.10050070e-01 7.01765195e-02 5.37056983e-01
7.87228167e-01 -1.81912541e-01 1.67643622e-01 6.84121028e-02
1.32296395e+00 1.28461587e+00 9.68863443e-02 -1.00320123e-01
3.18801761e-01 3.13668884e-02 5.04914403e-01 8.50432813e-01
-3.63405049e-01 8.71200323e-01 2.77242452e-01 -5.35999715e-01
-1.33076000e+00 -1.37459779e+00 2.64822543e-01 8.23611736e-01
-2.80544221e-01 -8.28488395e-02 -1.02097797e+00 -5.65650761e-01
-2.13203743e-01 1.50607812e+00 -9.29823577e-01 -2.48146847e-01
-6.82854831e-01 -8.42960238e-01 5.75468540e-01 7.11577058e-01
3.27849567e-01 -1.68695247e+00 -1.20211470e+00 5.33945620e-01
-5.03158152e-01 -1.24677384e+00 -2.73753610e-02 -2.18037382e-01
-1.02273846e+00 -6.87500596e-01 -1.20821869e+00 -3.91633064e-01
-1.22800516e-03 -1.74977735e-01 1.05528069e+00 -5.51779330e-01
-2.11825401e-01 8.26147735e-01 -5.53663015e-01 -2.63369203e-01
-8.98248196e-01 -2.78171897e-01 1.79219455e-01 1.53186157e-01
1.61495805e-01 -7.39364326e-01 -5.27786076e-01 7.72464201e-02
-9.97270525e-01 1.90558091e-01 4.43944842e-01 9.24117327e-01
5.72790861e-01 -2.23349601e-01 6.09512925e-01 -2.02520132e-01
2.78866708e-01 -6.82733536e-01 -2.69749761e-01 9.79805291e-02
-1.41254053e-01 1.04237042e-01 4.94116932e-01 -1.09046495e+00
-1.35190523e+00 1.57606676e-01 -1.21847540e-01 -1.07022095e+00
-3.93285662e-01 1.31822839e-01 3.50314200e-01 4.39778775e-01
8.65659177e-01 8.66249561e-01 -1.48097992e-01 -1.01512223e-01
6.11224473e-01 1.75561965e-01 7.53939331e-01 -6.03880942e-01
4.35115784e-01 5.54347157e-01 -7.57614002e-02 -1.07987201e+00
-4.39445108e-01 -1.63926542e-01 -6.93541050e-01 -7.24253058e-01
1.52194095e+00 -9.22355771e-01 -3.51185024e-01 9.54627931e-01
-1.46369779e+00 -1.23565269e+00 -4.23418820e-01 7.19013512e-01
-1.41118586e+00 4.12062883e-01 -8.16395760e-01 -1.22486687e+00
-1.44791797e-01 -9.01359260e-01 1.17074692e+00 -4.97612916e-02
-3.89084339e-01 -8.80028367e-01 4.70851392e-01 3.87360692e-01
1.07850477e-01 5.79400480e-01 6.47293448e-01 -3.07603419e-01
-5.83502471e-01 2.50540107e-01 5.48169196e-01 2.13254139e-01
-2.06932217e-01 -1.06142893e-01 -1.09133649e+00 -1.64924160e-01
6.48424104e-02 -7.07578897e-01 1.18513405e+00 7.63833284e-01
9.04265940e-01 -1.09173261e-01 -3.20888400e-01 2.84176320e-01
1.07548940e+00 5.49998641e-01 1.12629485e+00 1.67923540e-01
4.92219597e-01 4.13046241e-01 3.48224312e-01 5.34000933e-01
2.09834516e-01 5.60110390e-01 3.70183468e-01 6.31575048e-01
-3.56289536e-01 -3.94556075e-01 7.12280333e-01 5.39847255e-01
-5.78313887e-01 -5.42786539e-01 -8.66725683e-01 9.12606299e-01
-2.13105059e+00 -1.47817683e+00 1.10903174e-01 1.83154619e+00
7.01091170e-01 -2.46366430e-02 4.29087460e-01 4.41551134e-02
5.74739575e-01 1.95110723e-01 -5.25247812e-01 -1.97298810e-01
-4.00440246e-02 4.19759095e-01 -1.66624531e-01 3.58111799e-01
-1.14045680e+00 1.19192517e+00 6.49787331e+00 9.85415936e-01
-8.50129962e-01 3.46568346e-01 4.17196333e-01 5.02231857e-03
-1.62848726e-01 -2.05175668e-01 -5.37417889e-01 4.81354237e-01
1.54239261e+00 8.90322477e-02 4.90013957e-01 9.61847842e-01
2.75673062e-01 -1.62918776e-01 -1.12961614e+00 9.41278875e-01
3.43396366e-01 -1.44207811e+00 3.40865910e-01 4.82703298e-02
9.54941690e-01 3.22310217e-02 1.62500888e-01 8.55643928e-01
5.35532057e-01 -8.89548242e-01 1.15305662e+00 7.77764618e-01
4.54390109e-01 -6.09521747e-01 2.18127042e-01 7.78187156e-01
-1.00027680e+00 -2.29075179e-01 -3.77058744e-01 1.82462856e-01
5.91782033e-01 9.56930816e-02 -5.58447540e-01 5.57043672e-01
7.61657894e-01 7.08551943e-01 -2.97439367e-01 8.12995493e-01
-2.53134131e-01 9.61598635e-01 -8.19243956e-03 -7.16316514e-03
4.57851440e-01 -4.86308523e-02 7.30959475e-01 1.23335075e+00
6.60748065e-01 2.70759851e-01 -1.55503035e-01 1.05878794e+00
2.69096911e-01 -3.68083060e-01 -6.95538759e-01 -1.19556315e-01
-4.42717108e-04 8.94202232e-01 -7.44883716e-01 -7.54363060e-01
1.00568384e-02 1.52499712e+00 1.38129741e-01 4.85729545e-01
-1.19304740e+00 2.76481926e-01 2.10943624e-01 -2.53584623e-01
7.22139835e-01 -5.88004470e-01 2.68337846e-01 -1.20639813e+00
-3.13642353e-01 -6.33386314e-01 2.71271229e-01 -1.25436652e+00
-1.08343291e+00 6.76034212e-01 4.80615526e-01 -1.37296128e+00
-1.12158799e+00 -5.68570197e-01 -6.05165601e-01 4.91929591e-01
-8.59578669e-01 -1.51819837e+00 -1.26193807e-01 6.20651960e-01
9.61055696e-01 -4.40636069e-01 9.18433368e-01 -2.04101846e-01
-2.29854181e-01 2.41046175e-01 -1.41264930e-01 -9.82511863e-02
1.40102327e-01 -1.10960948e+00 7.75075853e-01 9.13315356e-01
4.80232358e-01 2.53400117e-01 8.06159854e-01 -8.08427155e-01
-1.18362379e+00 -1.35461378e+00 5.10188341e-01 -7.21158445e-01
5.40138125e-01 -2.67442614e-01 -9.17259455e-01 8.66440952e-01
5.36104500e-01 -7.38020539e-02 5.39190054e-01 -6.06841803e-01
-2.32004568e-01 6.10326171e-01 -9.52183425e-01 7.60033727e-01
1.42960668e+00 -3.88582796e-01 -8.60521376e-01 2.10821912e-01
7.85001695e-01 -2.48679012e-01 -6.82265997e-01 3.04092854e-01
6.30426228e-01 -1.15474522e+00 1.00209880e+00 -7.42470086e-01
9.31282878e-01 -1.27705783e-01 -1.80676296e-01 -1.45142162e+00
-4.57572401e-01 -6.47647858e-01 -7.76678205e-01 7.23940790e-01
6.25598207e-02 -2.86048174e-01 9.16928887e-01 1.97998837e-01
-3.56057018e-01 -3.91922534e-01 -1.17471719e+00 -1.33455646e+00
4.01156843e-01 -7.20715046e-01 2.13416606e-01 8.33179712e-01
-4.06735331e-01 2.61252284e-01 -7.02291489e-01 -2.33466700e-01
6.95465028e-01 -2.28890568e-01 8.47255111e-01 -8.27511370e-01
-9.41916645e-01 -3.74491155e-01 -4.19100225e-01 -9.72808301e-01
5.14507949e-01 -7.23631322e-01 1.45495370e-01 -1.54235756e+00
-7.01098761e-04 3.09640974e-01 -2.39351820e-02 5.10381281e-01
1.47865802e-01 2.19133645e-01 2.56339580e-01 2.83182174e-01
-3.96887809e-01 1.26556468e+00 1.27375543e+00 -1.26409009e-01
-3.38514477e-01 -6.10305965e-02 1.87919214e-01 4.77783412e-01
5.65227449e-01 -4.01252657e-01 -6.61400080e-01 -2.08664954e-01
-4.38965827e-01 5.96655428e-01 6.47664666e-01 -1.51853514e+00
-1.56271592e-01 -2.57326782e-01 5.58718264e-01 -8.71557891e-01
7.93502212e-01 -2.64036447e-01 5.62420011e-01 8.12904239e-01
-1.85830832e-01 -2.03056782e-01 3.90527815e-01 1.00150979e+00
4.58304361e-02 -8.60680416e-02 6.77141249e-01 -4.95369792e-01
-1.02250338e+00 1.91425472e-01 -1.05178833e+00 -2.46193156e-01
1.21301425e+00 -3.04634392e-01 -2.31077075e-01 -6.34887099e-01
-1.14504373e+00 -3.35338444e-01 7.63387382e-02 4.99782860e-01
7.40466714e-01 -1.46308708e+00 -8.04357111e-01 -2.35746264e-01
-8.66343547e-03 -3.08148503e-01 6.60764456e-01 4.77361500e-01
-3.20296943e-01 4.81724948e-01 -5.01947045e-01 -9.02512908e-01
-8.12082827e-01 6.91088617e-01 3.70855600e-01 -2.63844013e-01
-9.06456232e-01 8.12417626e-01 2.61502713e-01 -1.41753569e-01
-2.68790722e-01 -1.82224557e-01 -2.40817979e-01 -1.88337211e-02
4.18429524e-01 5.49992561e-01 -2.71814257e-01 -7.53147364e-01
-7.06612617e-02 3.12521815e-01 3.92436266e-01 -9.38348055e-01
1.40105951e+00 3.16840827e-01 3.75190347e-01 8.49862754e-01
8.26240838e-01 -7.02911556e-01 -1.88250542e+00 1.17332019e-01
-2.47374400e-01 -2.16280688e-02 -2.93721884e-01 -4.77413863e-01
-7.10980058e-01 1.00586116e+00 5.50134659e-01 -1.16865806e-01
6.72970891e-01 5.02961222e-03 7.43811488e-01 3.27465951e-01
7.20755875e-01 -9.45971429e-01 8.22118342e-01 6.02088988e-01
1.26373541e+00 -7.67082632e-01 -2.98399776e-01 1.78584516e-01
-1.11733985e+00 1.00495255e+00 6.21571898e-01 -3.66577774e-01
5.82542717e-01 -1.38826266e-01 -2.34619096e-01 2.27953523e-01
-1.12167430e+00 -1.70026332e-01 3.10316622e-01 9.48963821e-01
1.77969024e-01 -1.40346354e-02 -1.12644196e-01 5.38577497e-01
-2.58520454e-01 3.61732274e-01 6.28450334e-01 9.29140449e-01
-6.43177181e-02 -7.56360710e-01 -1.80758402e-01 -8.76094103e-02
-1.38595119e-01 2.23688200e-01 -2.35844534e-02 7.44361818e-01
9.90946889e-02 8.96855354e-01 2.05380499e-01 -5.01662433e-01
-1.71530709e-01 5.47085881e-01 8.86265397e-01 -4.63809073e-01
-2.71887273e-01 2.34745160e-01 -1.84889492e-02 -6.40658855e-01
-7.31669545e-01 -1.03209245e+00 -1.23105681e+00 1.38357610e-01
1.68715194e-01 -1.86767712e-01 3.86650831e-01 9.53337729e-01
1.00527987e-01 6.36997879e-01 1.02623031e-01 -1.54789746e+00
-3.96520674e-01 -1.19805455e+00 -3.35147828e-01 4.41252559e-01
1.59263968e-01 -8.62899959e-01 -3.27980071e-01 5.89071989e-01] | [7.325576305389404, -0.11761481314897537] |
aa7f8dff-dd34-4503-9256-afd94dd6e14a | isea-image-steganalysis-using-evolutionary | 1907.12914 | null | https://arxiv.org/abs/1907.12914v3 | https://arxiv.org/pdf/1907.12914v3.pdf | Evolutionary Algorithms and Efficient Data Analytics for Image Processing | Steganography algorithms facilitate communication between a source and a destination in a secret manner. This is done by embedding messages/text/data into images without impacting the appearance of the resultant images/videos. Steganalysis is the science of determining if an image has secret messages embedded/hidden in it. Because there are numerous steganography algorithms, and since each one of them requires a different type of steganalysis, the steganalysis process is extremely challenging. Thus, researchers aim to develop one universal steganalysis to detect all known and unknown steganography algorithms, ideally in real-time. Universal steganalysis extracts a large number of features to distinguish stego images from cover images. However, the increase in features leads to the problem of the curse of dimensionality (CoD), which is considered to be an NP-hard problem. This COD problem additionally makes real-time steganalysis hard. A large number of features generates large datasets for which machine learning cannot generate an optimal model. Generating a machine learning based model also takes a long time which makes real-time processing appear impossible in any optimization for time-intensive fields such as visual computing. Possible solutions for CoD are deep learning and evolutionary algorithms that overcome the machine learning limitations. In this study, we investigate previously developed evolutionary algorithms for boosting real-time image processing and argue that they provide the most promising solutions for the CoD problem. | ['M. Hadi Amini', 'Farzan Shenavarmasouleh', 'Hamid R. Arabnia', 'Farid Ghareh Mohammadi'] | 2019-07-23 | null | null | null | null | ['steganalysis'] | ['computer-vision'] | [ 9.14293051e-01 -1.77279845e-01 3.17532569e-01 2.55759358e-01
-1.85295910e-01 -4.70601350e-01 4.97854441e-01 -1.75702900e-01
-2.21271902e-01 4.43855613e-01 -6.68072462e-01 -6.12846375e-01
2.05091715e-01 -1.10327613e+00 -7.12264836e-01 -1.02062023e+00
-2.82073557e-01 2.93243736e-01 7.03813955e-02 -5.21984994e-01
6.05853677e-01 4.33706611e-01 -1.80184245e+00 2.07590684e-01
7.05786467e-01 9.36971903e-01 2.61017799e-01 9.70579326e-01
-1.73525792e-02 4.70979273e-01 -8.14353526e-01 -3.19742352e-01
5.66237926e-01 -1.00098038e+00 -4.35462862e-01 2.66810954e-01
-2.59325266e-01 3.71621437e-02 -1.58032686e-01 1.45528924e+00
2.96457082e-01 -3.41551989e-01 5.58606386e-01 -1.63040066e+00
-4.23190147e-01 3.95313382e-01 -4.49289322e-01 6.59229308e-02
1.31612673e-01 4.48233277e-01 4.39973146e-01 -3.90625745e-01
4.75518584e-01 9.68683481e-01 3.40631634e-01 4.14827377e-01
-7.31530547e-01 -9.22205329e-01 -6.53617859e-01 4.00329828e-01
-1.10947704e+00 -3.17675829e-01 9.23822761e-01 -2.44426474e-01
6.66909754e-01 7.46869445e-01 1.02838278e+00 8.10273290e-01
6.58936501e-01 4.46360320e-01 1.40661454e+00 -7.75786817e-01
9.31767896e-02 3.16666067e-01 -6.25264525e-01 8.34875524e-01
6.13720775e-01 5.23570478e-01 -9.42575112e-02 3.74014676e-02
4.11947697e-01 5.80076575e-02 -3.75318617e-01 -1.62847415e-01
-1.16307569e+00 1.18350530e+00 7.00120330e-02 5.31423986e-01
-7.46243224e-02 1.29498526e-01 2.47192979e-01 9.94272947e-01
6.18419833e-02 7.71121085e-01 -1.66781783e-01 1.03109986e-01
-8.83401334e-01 -2.46221334e-01 1.02863967e+00 5.03314793e-01
8.84764314e-01 2.06296712e-01 6.23480320e-01 1.32660240e-01
3.89984190e-01 7.54253209e-01 6.18511438e-01 -3.74582201e-01
2.07671657e-01 7.36200571e-01 -3.32618952e-01 -1.69584680e+00
-4.31810357e-02 -3.25558215e-01 -1.14514434e+00 6.46834314e-01
2.79219836e-01 -7.83509910e-02 -9.04405892e-01 1.18009102e+00
3.30350339e-01 1.31043047e-01 1.83495194e-01 6.68961465e-01
3.45372736e-01 9.70968068e-01 -4.14889842e-01 -3.83001238e-01
1.21686625e+00 -7.68435299e-01 -6.42352641e-01 -5.24238050e-01
5.37639022e-01 -9.58306134e-01 4.19188082e-01 3.55689168e-01
-6.62086189e-01 -2.39094302e-01 -1.49166024e+00 4.72410291e-01
-7.29144156e-01 -4.90607172e-01 4.74576920e-01 1.09703159e+00
-9.24255252e-01 4.63141203e-01 -3.94295365e-01 -1.19313747e-01
7.90345371e-02 7.83991456e-01 -3.45024496e-01 -4.84009013e-02
-1.39124584e+00 9.30865169e-01 7.99538910e-01 1.49925038e-01
-6.25958383e-01 5.25639057e-02 -8.21063757e-01 -7.49399215e-02
4.38704014e-01 -3.14549804e-01 3.81460637e-01 -1.62297857e+00
-1.33362496e+00 1.01667869e+00 2.41126582e-01 -5.68716168e-01
7.75208592e-01 8.13280225e-01 -7.42711127e-01 7.71066770e-02
-4.70451266e-01 2.79044449e-01 1.81860673e+00 -1.20335674e+00
-7.70366907e-01 -2.69466043e-01 -3.25612009e-01 -1.90285400e-01
-2.91221589e-01 4.51991819e-02 -1.70037150e-01 -5.57139397e-01
1.99847579e-01 -1.30739868e+00 -2.45429784e-01 -3.08368951e-01
-1.90353811e-01 2.11120427e-01 1.34707236e+00 -8.08088303e-01
1.10944283e+00 -1.93524730e+00 3.32892425e-02 6.59164786e-01
4.00627375e-01 7.78285682e-01 -8.83126408e-02 5.10133803e-01
-1.02218494e-01 4.24281806e-01 -3.21238309e-01 2.66492128e-01
-3.33558261e-01 1.51025191e-01 -2.64142267e-03 7.44971693e-01
-2.40640044e-02 9.12313461e-01 -6.70474589e-01 -5.92725396e-01
3.61100763e-01 4.13831472e-01 -2.47631028e-01 7.08415806e-02
-6.60437867e-02 3.97968978e-01 -3.43830645e-01 5.81450105e-01
8.31608534e-01 -2.68064320e-01 4.91875261e-01 1.95887566e-01
7.80836642e-02 -3.60848814e-01 -1.09342682e+00 6.98860765e-01
-3.54010582e-01 9.94369566e-01 -1.87552273e-01 -1.37102377e+00
9.99014735e-01 2.94225514e-01 4.29607987e-01 -6.42621815e-01
5.86001933e-01 5.21681488e-01 2.88694382e-01 -8.16299796e-01
3.99040282e-01 -8.88383538e-02 -1.36843875e-01 7.43561089e-01
-4.37900126e-01 -1.31912068e-01 1.58841938e-01 -2.53341049e-01
1.10409570e+00 -5.13916016e-01 2.84487426e-01 1.41302764e-01
5.84962130e-01 1.07555240e-01 3.48920226e-01 7.85245895e-01
-6.71623275e-02 2.18337953e-01 3.95826668e-01 -5.23530543e-01
-1.43979800e+00 -2.03227520e-01 5.55177689e-01 4.85929996e-01
4.52362537e-01 4.81446981e-02 -7.24886060e-01 -5.30129015e-01
-2.64638364e-01 2.41307437e-01 -5.02375126e-01 -5.10566413e-01
-6.40985727e-01 -8.63592327e-01 5.25355637e-01 -5.44509530e-01
6.47594869e-01 -1.28566551e+00 -1.06146705e+00 2.82530814e-01
-2.31956691e-01 -9.68244135e-01 -1.20742157e-01 -7.16355741e-02
-7.15212941e-01 -1.39724898e+00 -5.57161152e-01 -8.89960468e-01
9.60976183e-01 3.72760653e-01 6.73145175e-01 7.52421200e-01
-5.51587999e-01 -4.02130455e-01 -6.88617527e-01 -5.83033860e-01
-1.28439569e+00 -1.73892379e-01 -1.89223677e-01 1.89027250e-01
2.28049994e-01 -1.59836695e-01 -4.88611907e-01 2.82804608e-01
-1.13747466e+00 1.71438888e-01 8.50479901e-01 8.49306643e-01
2.16516435e-01 1.03050649e+00 1.64659828e-01 -7.57844985e-01
5.63516855e-01 -4.73993361e-01 -9.37638104e-01 4.63358998e-01
-9.03569698e-01 1.46953240e-01 5.95833480e-01 -5.96650898e-01
-3.17038566e-01 -7.92266056e-02 1.97525159e-01 -2.52371073e-01
1.72579259e-01 5.21461070e-01 -1.81317538e-01 -6.71118677e-01
2.97696471e-01 7.66618431e-01 6.12703979e-01 7.41127953e-02
-2.42966771e-01 8.21395278e-01 6.95870295e-02 4.27076429e-01
1.26308036e+00 4.26534235e-01 3.15456390e-01 -9.93320584e-01
-5.35660535e-02 -2.16307923e-01 -2.05888078e-01 -3.24152410e-01
6.89945459e-01 -2.99028456e-01 -7.49495983e-01 8.97308230e-01
-9.84061480e-01 9.41074565e-02 1.72447011e-01 2.02982232e-01
-1.60319209e-01 6.29298389e-01 2.57486757e-02 -8.00866902e-01
-3.14086258e-01 -1.44888663e+00 3.07325363e-01 5.46148717e-02
-7.87320063e-02 -9.47565138e-01 -1.84414357e-01 4.15351599e-01
4.28301185e-01 7.38872230e-01 8.08831453e-01 -5.61941504e-01
-7.52715111e-01 -6.82113409e-01 8.88665393e-03 3.61349970e-01
9.52651873e-02 1.18250117e-01 -4.18966234e-01 -6.99219704e-01
4.65370268e-01 7.91784748e-02 7.48963177e-01 1.77207381e-01
9.30642128e-01 -7.33717382e-01 -3.20206970e-01 8.34800005e-01
1.68765032e+00 7.84753203e-01 8.70684922e-01 6.03249788e-01
4.98398900e-01 7.23788798e-01 6.00468874e-01 3.39205235e-01
-8.36077929e-02 2.90565699e-01 7.53349602e-01 -2.37906232e-01
3.33490968e-01 2.92917117e-02 3.92619550e-01 7.13625133e-01
5.10992575e-03 -8.59139383e-01 -8.53368580e-01 3.68724525e-01
-1.42478454e+00 -1.15308642e+00 -2.72157520e-01 2.14912891e+00
5.10939777e-01 -3.14617151e-04 -1.85247421e-01 8.26490879e-01
1.08235765e+00 1.76845178e-01 -3.63077104e-01 -6.83808744e-01
-2.10826412e-01 1.23242386e-01 8.87016833e-01 2.26754919e-01
-1.06765950e+00 7.59618461e-01 5.49192190e+00 9.58430767e-01
-1.74062741e+00 -4.11393642e-02 4.88546371e-01 3.16049039e-01
-1.89322427e-01 2.23160550e-01 -2.50678241e-01 9.03414667e-01
1.00202894e+00 -1.59523338e-01 6.46926105e-01 4.76092696e-01
1.52057819e-02 -1.96715429e-01 -5.29060304e-01 1.23850822e+00
3.72790515e-01 -1.37696171e+00 -7.08162785e-02 6.39672875e-01
8.39143574e-01 -2.41531879e-01 2.20520288e-01 -2.44574994e-01
2.93092988e-02 -1.12332010e+00 3.95232826e-01 2.29256451e-02
9.06314909e-01 -8.09254348e-01 7.44399548e-01 4.72908199e-01
-8.01102579e-01 -2.50770658e-01 -3.77336890e-01 1.62781507e-01
4.38705795e-02 4.22182381e-01 -1.09368002e+00 2.90138096e-01
4.51155603e-01 2.27838859e-01 -4.07594502e-01 1.04032326e+00
-8.46765116e-02 4.19657469e-01 -2.35359117e-01 -3.96405131e-01
2.82193482e-01 -1.02758080e-01 8.87357056e-01 8.03494394e-01
7.65048563e-01 -1.30413305e-02 -1.77836418e-01 4.95337993e-01
4.85077053e-02 -1.14456527e-01 -8.86078179e-01 -6.64886355e-01
4.00347769e-01 9.62282836e-01 -1.11474872e+00 -1.47845775e-01
5.41933812e-02 1.21624649e+00 -5.16030550e-01 -1.00018680e-01
-7.12435424e-01 -6.15064263e-01 2.48787165e-01 1.18165083e-01
4.20723200e-01 -1.81548253e-01 -1.82874411e-01 -1.01044309e+00
-2.12077707e-01 -1.37299502e+00 1.63724512e-01 -2.26826370e-01
-6.56347632e-01 6.59107864e-01 -4.98337328e-01 -1.45016837e+00
-4.02988553e-01 -4.49715286e-01 -4.53912795e-01 3.78469437e-01
-1.56352270e+00 -1.14959979e+00 -2.76403368e-01 6.23802423e-01
4.65035200e-01 -5.21688581e-01 5.96518755e-01 4.61382605e-03
-3.47958744e-01 4.75122601e-01 3.96953136e-01 1.11996181e-01
2.39461750e-01 -6.26291275e-01 3.57942641e-01 1.10961962e+00
-2.54925136e-02 8.86018872e-02 1.18032229e+00 -7.08844125e-01
-1.83021069e+00 -8.05345714e-01 9.11154747e-01 7.19151869e-02
4.22369748e-01 -2.21547633e-01 -6.36916101e-01 2.58732706e-01
8.83602872e-02 -2.50796407e-01 5.16754150e-01 -7.86949277e-01
9.50099342e-03 1.97965100e-01 -1.37335527e+00 3.30625385e-01
5.78269422e-01 -2.24124700e-01 -1.69483021e-01 1.53002173e-01
3.14077318e-01 -2.03925580e-01 -3.53136718e-01 9.23010930e-02
6.41818941e-01 -1.03047824e+00 7.82299042e-01 -1.25755414e-01
5.07449806e-01 -2.06132486e-01 1.86041132e-01 -1.08282721e+00
1.04826771e-01 -9.13432419e-01 -5.90922162e-02 6.51712060e-01
2.95028538e-01 -9.31882381e-01 7.28478134e-01 4.35278155e-02
4.68848199e-01 -4.80145484e-01 -8.11287999e-01 -9.42403376e-01
-3.11710924e-01 -1.61918715e-01 9.46119964e-01 1.22309351e+00
-2.90389955e-01 -2.93512553e-01 -8.00013721e-01 1.40280291e-01
9.32915270e-01 2.23449901e-01 8.01117897e-01 -1.24722946e+00
-1.41612917e-01 -3.83920193e-01 -7.97183633e-01 -2.42081329e-01
-2.93510295e-02 -6.63446009e-01 -1.19430810e-01 -1.06840646e+00
-1.56304389e-01 -7.85708845e-01 5.96063845e-02 1.58035085e-01
2.23191351e-01 4.45418924e-01 1.78093314e-01 4.14621860e-01
-1.77704453e-01 1.11224227e-01 1.31755638e+00 -3.74172300e-01
-5.86839393e-02 7.02577829e-02 -4.73166645e-01 5.81712365e-01
1.02380598e+00 -9.55957234e-01 -2.57526249e-01 -1.52978405e-01
5.34214258e-01 1.84039935e-01 4.36268181e-01 -1.04760182e+00
2.60534585e-01 -3.59137595e-01 1.13467850e-01 -2.16775760e-01
2.72292554e-01 -1.08101511e+00 5.71093917e-01 1.29826725e+00
2.05220729e-01 -5.86932413e-02 -3.54931176e-01 5.60279846e-01
-2.76103944e-01 -4.94867593e-01 9.17610049e-01 -4.52973604e-01
-1.02375674e+00 1.75510049e-01 -6.62293673e-01 -2.99701601e-01
1.54511249e+00 -9.18664753e-01 -7.95866773e-02 -5.22500753e-01
-3.42561781e-01 -1.22418508e-01 6.90548420e-01 5.59763670e-01
9.93067563e-01 -7.91571498e-01 -6.34667516e-01 6.36554301e-01
-1.42759472e-01 -4.91758317e-01 2.16701299e-01 5.89448869e-01
-1.12211049e+00 2.29542091e-01 -4.51682955e-01 -3.62286389e-01
-1.84837759e+00 7.98757553e-01 2.50806689e-01 -4.79892820e-01
-3.54137689e-01 7.80721664e-01 -4.86056954e-01 1.03371359e-01
-1.86152115e-01 3.07388186e-01 -2.63792396e-01 5.45799136e-02
4.41329300e-01 4.43734020e-01 -1.48509622e-01 -8.49104941e-01
-1.01279445e-01 5.43249428e-01 1.06773883e-01 4.41572368e-02
1.29716527e+00 -2.44585082e-01 -5.40636241e-01 -2.52159834e-01
1.46980178e+00 -2.46444374e-01 -5.91154933e-01 9.83380601e-02
-5.31860217e-02 -7.15365767e-01 1.00057550e-01 -3.82667869e-01
-1.27352750e+00 8.53077590e-01 7.96589255e-01 7.09413767e-01
1.32632411e+00 -4.43152875e-01 1.23399425e+00 3.59165490e-01
4.14522439e-01 -1.20519376e+00 1.13882862e-01 2.68303603e-01
6.35937393e-01 -1.59856427e+00 -1.72442630e-01 -2.96108037e-01
-4.80566353e-01 1.18889439e+00 1.82714626e-01 -2.94584576e-02
6.90432370e-01 2.18248874e-01 -2.86499970e-02 -3.41628224e-01
-3.64099860e-01 8.38593319e-02 5.20175658e-02 5.84648669e-01
-3.79418612e-01 -3.32413465e-02 -5.06331623e-01 -3.12425166e-01
-3.95376861e-01 -3.15473378e-02 6.84030652e-01 1.17734349e+00
-6.13592029e-01 -1.42832839e+00 -7.95246243e-01 2.21508890e-01
-6.87305033e-01 1.16990209e-01 -3.92752141e-01 6.48906469e-01
4.58039343e-01 1.27089286e+00 -1.35641336e-01 -8.25363040e-01
-4.86447513e-01 -3.51885498e-01 2.90453583e-01 -1.23003334e-01
-4.53819931e-01 -2.39573851e-01 -2.75721997e-01 -1.15484565e-01
-3.42785597e-01 -2.88163573e-01 -1.02551579e+00 -8.17935467e-01
-7.07272708e-01 2.53776938e-01 1.26683617e+00 1.06760275e+00
1.72398940e-01 3.08854848e-01 1.11386192e+00 -5.56118846e-01
-4.91514623e-01 -4.25973564e-01 -4.77635980e-01 3.39823216e-01
5.17984152e-01 -2.00818494e-01 -6.53701961e-01 8.98270831e-02] | [4.301121711730957, 8.059426307678223] |
644ae718-0759-4b3d-8366-c3c6051653b6 | volumetric-medical-image-segmentation-a-3d | 2010.16074 | null | https://arxiv.org/abs/2010.16074v1 | https://arxiv.org/pdf/2010.16074v1.pdf | Volumetric Medical Image Segmentation: A 3D Deep Coarse-to-fine Framework and Its Adversarial Examples | Although deep neural networks have been a dominant method for many 2D vision tasks, it is still challenging to apply them to 3D tasks, such as medical image segmentation, due to the limited amount of annotated 3D data and limited computational resources. In this chapter, by rethinking the strategy to apply 3D Convolutional Neural Networks to segment medical images, we propose a novel 3D-based coarse-to-fine framework to efficiently tackle these challenges. The proposed 3D-based framework outperforms their 2D counterparts by a large margin since it can leverage the rich spatial information along all three axes. We further analyze the threat of adversarial attacks on the proposed framework and show how to defense against the attack. We conduct experiments on three datasets, the NIH pancreas dataset, the JHMI pancreas dataset and the JHMI pathological cyst dataset, where the first two and the last one contain healthy and pathological pancreases respectively, and achieve the current state-of-the-art in terms of Dice-Sorensen Coefficient (DSC) on all of them. Especially, on the NIH pancreas segmentation dataset, we outperform the previous best by an average of over $2\%$, and the worst case is improved by $7\%$ to reach almost $70\%$, which indicates the reliability of our framework in clinical applications. | ['Alan L. Yuille', 'Elliot K. Fishman', 'Wei Shen', 'Yingda Xia', 'Yuyin Zhou', 'Zhuotun Zhu', 'Yingwei Li'] | 2020-10-29 | null | null | null | null | ['volumetric-medical-image-segmentation', 'pancreas-segmentation'] | ['medical', 'medical'] | [-6.65471405e-02 2.77586877e-01 1.88525319e-02 -7.77837262e-02
-6.35876775e-01 -5.62932193e-01 1.67711660e-01 8.66368636e-02
-5.02848089e-01 3.54749888e-01 -1.16642630e-02 -7.34542608e-01
-1.17357805e-01 -6.09522998e-01 -7.85944700e-01 -8.76067758e-01
-3.99621636e-01 1.67869061e-01 2.59209663e-01 5.50742410e-02
6.22280538e-02 6.87676549e-01 -5.62748313e-01 -2.56679296e-01
1.00845265e+00 1.28026891e+00 -2.62844175e-01 4.46300030e-01
7.96420947e-02 4.91534919e-01 -4.20665950e-01 -5.10214865e-01
7.33407021e-01 -3.39870930e-01 -4.67746079e-01 3.15042026e-02
3.26636940e-01 -5.82985997e-01 -6.61098719e-01 1.27420104e+00
7.78378129e-01 -2.87300050e-01 5.79807997e-01 -1.01211500e+00
-5.56595325e-01 5.67977726e-01 -6.88063622e-01 2.77563244e-01
3.70011199e-03 2.26635426e-01 3.21641445e-01 -3.11587602e-01
5.92901051e-01 8.93111229e-01 8.66062343e-01 5.76617062e-01
-8.04644406e-01 -8.47067714e-01 -1.75657459e-02 -2.50507444e-01
-1.00503314e+00 1.82468891e-01 7.99370766e-01 -3.82698029e-01
4.69305754e-01 -1.01816937e-01 6.49796009e-01 1.02323902e+00
3.72334063e-01 9.35058594e-01 1.33332717e+00 -1.99849278e-01
5.26069626e-02 -3.33082467e-01 2.13672221e-01 9.03066456e-01
4.59921271e-01 4.15123969e-01 1.73050299e-01 -6.44483492e-02
1.12768328e+00 7.31666684e-02 -4.55065131e-01 -5.38075566e-01
-1.34635639e+00 9.68817115e-01 8.75945270e-01 3.47506558e-03
-3.74004841e-01 -9.55777466e-02 5.90049207e-01 2.63173431e-02
5.38302697e-02 2.19129935e-01 -2.47378066e-01 2.27566272e-01
-6.25290155e-01 5.95913082e-02 8.36229622e-01 9.18506563e-01
2.90955976e-02 -4.32237685e-02 7.03404173e-02 4.84650671e-01
3.89742076e-01 4.72211868e-01 4.07344222e-01 -8.28981280e-01
4.26092893e-01 3.78769726e-01 -1.82849139e-01 -1.01112056e+00
-7.62537777e-01 -6.13505781e-01 -1.35371971e+00 3.35110217e-01
6.17565870e-01 -2.97195762e-01 -1.32840860e+00 1.62964380e+00
4.25603569e-01 1.93775356e-01 1.28292218e-01 1.15180826e+00
1.03909433e+00 2.69012749e-01 3.26251984e-03 -1.84325695e-01
1.15548444e+00 -8.55554640e-01 -4.49655175e-01 6.91238865e-02
5.52597940e-01 -6.13058567e-01 5.44496894e-01 4.25328672e-01
-9.41451907e-01 -2.17688397e-01 -1.17979896e+00 2.03156173e-01
-2.61888430e-02 -2.79407799e-01 7.32501984e-01 7.61142612e-01
-6.99615240e-01 5.07104278e-01 -1.12391806e+00 -2.00560629e-01
8.61941636e-01 3.26523393e-01 -3.86196107e-01 -7.86388293e-02
-1.20430732e+00 8.03125441e-01 2.31293231e-01 1.88244730e-01
-1.07452619e+00 -7.00417340e-01 -8.57874870e-01 -1.73605546e-01
2.63587147e-01 -6.39035344e-01 8.76340508e-01 -3.56620461e-01
-1.45810211e+00 1.06628513e+00 6.89078450e-01 -7.43172228e-01
1.09971333e+00 -1.59594774e-01 -4.52693440e-02 5.00317872e-01
-2.14158341e-01 6.43502116e-01 4.38975483e-01 -1.25988710e+00
-3.65763575e-01 -6.16384983e-01 2.29121178e-01 5.78926988e-02
2.56454702e-02 -2.90704966e-01 -6.71917677e-01 -8.89187872e-01
4.20748681e-01 -1.25304163e+00 -5.77635586e-01 5.75915933e-01
-7.94877887e-01 2.97232777e-01 6.14409626e-01 -6.09825313e-01
6.65135503e-01 -2.31251907e+00 -2.65367329e-02 2.65919149e-01
3.75960141e-01 4.34270591e-01 6.90803379e-02 -1.70774698e-01
1.12111412e-01 1.56707823e-01 -2.97206461e-01 -5.98338321e-02
-5.55660278e-02 3.68638784e-01 -4.55396734e-02 8.87999475e-01
-2.43108124e-02 7.68979549e-01 -7.63267517e-01 -6.80770755e-01
3.49994510e-01 4.23092991e-01 -4.50492024e-01 1.60394877e-01
2.84414232e-01 5.60472727e-01 -6.72591269e-01 7.44221032e-01
1.06775272e+00 -2.44444534e-01 8.78875256e-02 -2.67796040e-01
2.30000675e-01 -2.79795766e-01 -9.10233200e-01 1.82767642e+00
-1.95159379e-03 1.53435245e-01 3.93722862e-01 -1.13543999e+00
9.66901660e-01 2.50070125e-01 7.82096565e-01 -7.38479793e-01
2.42450267e-01 3.66816670e-01 1.32330611e-01 -5.28938890e-01
-1.78844735e-01 -2.64125466e-01 -3.78374338e-01 3.08208019e-01
-1.48101762e-01 -1.65945053e-01 -6.68519288e-02 1.60585701e-01
1.03109157e+00 -4.78186980e-02 1.62312955e-01 -4.24394518e-01
4.36814100e-01 7.40303621e-02 7.33054698e-01 8.40845644e-01
-9.45469260e-01 6.94496691e-01 7.93547630e-01 -5.74262917e-01
-8.17427039e-01 -1.28574347e+00 -3.48304629e-01 2.80223131e-01
8.46619308e-01 4.30754483e-01 -8.14877987e-01 -1.25337994e+00
2.34710217e-01 3.03757340e-01 -6.57018960e-01 -2.13797584e-01
-7.92331278e-01 -1.00308120e+00 1.06162250e+00 7.47718275e-01
8.79539907e-01 -7.79093981e-01 -6.96326792e-01 3.60176936e-02
1.16987810e-01 -1.18693507e+00 -4.34858471e-01 2.54132479e-01
-1.01152003e+00 -1.34436965e+00 -1.10849118e+00 -9.89512146e-01
7.48101234e-01 1.89140886e-01 9.40030694e-01 4.61413618e-03
-3.27334434e-01 -6.29302114e-02 -3.43932837e-01 -3.91330481e-01
-5.54823995e-01 -8.77824891e-03 3.93241532e-02 -5.06705940e-01
1.15424350e-01 -4.95924145e-01 -1.02290654e+00 4.49572653e-01
-1.01589012e+00 -1.79353699e-01 8.17763984e-01 9.62005734e-01
7.27690995e-01 -1.65345058e-01 5.39387822e-01 -8.34451437e-01
3.23779702e-01 -2.93111324e-01 -5.93322098e-01 -7.62900263e-02
-4.47024822e-01 -1.55230209e-01 6.25029981e-01 -5.28443992e-01
-4.83768940e-01 2.14582920e-01 -4.24598962e-01 -8.08593690e-01
-2.78609306e-01 3.79051477e-01 -1.16362022e-02 -3.48532856e-01
4.12072152e-01 1.52508780e-01 2.94577420e-01 -4.76894349e-01
2.02023417e-01 3.63143176e-01 8.34494293e-01 -3.24251533e-01
7.87115872e-01 5.81120491e-01 2.46052667e-01 -2.63494760e-01
-6.34579003e-01 -6.65812492e-02 -4.89310473e-01 4.34208550e-02
9.76808429e-01 -8.65071177e-01 -8.21459711e-01 6.97503090e-01
-7.53593981e-01 -1.51939198e-01 -1.47323191e-01 7.77972877e-01
-5.10762393e-01 7.22041070e-01 -1.00067329e+00 -2.39881843e-01
-6.25979602e-01 -1.71893358e+00 7.87844956e-01 2.83738375e-01
2.71226853e-01 -1.00386477e+00 -1.56788349e-01 2.44972169e-01
4.78004128e-01 9.85597491e-01 1.23045707e+00 -9.72120821e-01
-5.19819021e-01 -3.50414544e-01 -4.56532598e-01 3.94390643e-01
-3.19805858e-03 -4.01155859e-01 -5.74193418e-01 -4.43384081e-01
1.49974406e-01 -2.53322184e-01 7.96425581e-01 6.76353097e-01
1.12724710e+00 -1.76611599e-02 -3.51717621e-01 1.04901278e+00
1.41721106e+00 3.91790032e-01 4.86821145e-01 4.52872694e-01
5.80631375e-01 1.22979231e-01 3.33342701e-01 2.57590830e-01
2.04186335e-01 3.33756775e-01 1.01314151e+00 -4.05915558e-01
-1.06540300e-01 -5.51436581e-02 -6.46284372e-02 8.88682246e-01
2.26046219e-01 -6.78021163e-02 -8.73087823e-01 3.81835341e-01
-1.50943124e+00 -4.66851175e-01 9.11864564e-02 1.96617341e+00
6.81813359e-01 4.40019786e-01 -1.05934655e-02 -1.05526499e-01
6.82796776e-01 1.23775057e-01 -8.01997304e-01 -2.55340636e-01
7.77312741e-02 1.82146002e-02 7.13849723e-01 1.48604751e-01
-1.57034409e+00 5.47771215e-01 6.45445442e+00 5.70966601e-01
-1.36654198e+00 -2.50794232e-01 7.42437363e-01 5.39872013e-02
1.40709862e-01 -4.00539488e-01 -3.47966582e-01 5.32122374e-01
3.10181350e-01 7.07792118e-02 1.21089481e-01 9.05107319e-01
-1.79334171e-02 7.13581890e-02 -9.54056561e-01 9.46681142e-01
9.36475098e-02 -1.25071573e+00 -1.05862021e-01 2.20899940e-01
6.74858630e-01 1.82150155e-01 1.64742172e-01 1.88966110e-01
3.43537778e-01 -1.06370127e+00 3.34204942e-01 7.57864639e-02
7.48523057e-01 -8.00451577e-01 1.04735458e+00 4.08155084e-01
-7.45305419e-01 1.16679393e-01 -1.36362091e-01 5.64797044e-01
2.25095451e-01 6.34151459e-01 -6.02839172e-01 6.27006531e-01
8.28347564e-01 4.30418015e-01 -1.18273802e-01 1.31314528e+00
-1.88750684e-01 2.85839677e-01 -6.31336570e-01 1.48843855e-01
5.41316271e-01 -9.81475934e-02 7.83431470e-01 1.11734200e+00
4.16979760e-01 3.98852713e-02 3.36215138e-01 6.95315957e-01
-3.96965265e-01 -9.17875394e-02 -5.09680748e-01 3.83073896e-01
1.42589122e-01 1.09716952e+00 -8.48291934e-01 -4.34866957e-02
-3.26059192e-01 9.15349305e-01 -2.35584289e-01 6.69334382e-02
-1.06605875e+00 -4.95518893e-01 6.28273606e-01 -4.15658116e-01
4.07582641e-01 -1.29269272e-01 -3.68634075e-01 -1.13490129e+00
-2.15949938e-02 -8.81798267e-01 4.87546802e-01 -2.83578038e-01
-1.49122334e+00 7.08421767e-01 -3.01237166e-01 -1.42076778e+00
-1.10319279e-01 -9.05136883e-01 -5.04534125e-01 6.09571576e-01
-1.63795888e+00 -1.11672413e+00 -4.37839597e-01 7.34363556e-01
1.79898679e-01 -2.43597273e-02 6.27480388e-01 4.32227671e-01
-5.13880432e-01 9.19355690e-01 3.39850843e-01 6.76785588e-01
6.37133300e-01 -1.12671173e+00 1.77066997e-01 7.52915382e-01
-3.03305358e-01 3.64752561e-01 3.35663795e-01 -5.26348472e-01
-1.64129961e+00 -1.05817175e+00 4.45415787e-02 -1.37892604e-01
4.16673362e-01 5.01751900e-03 -7.30308115e-01 5.68245053e-01
-2.11227946e-02 5.66777408e-01 5.17779589e-01 -5.26909053e-01
-3.00709367e-01 2.47808732e-02 -1.67097843e+00 5.58582127e-01
9.60494220e-01 5.91667108e-02 -6.59365952e-01 1.40866563e-01
8.09574664e-01 -1.01653898e+00 -1.31405747e+00 7.25813866e-01
5.73279500e-01 -9.89898145e-01 1.11150062e+00 -3.68191421e-01
5.03040254e-01 -2.19001293e-01 -1.76599666e-01 -1.24897957e+00
-4.67698723e-02 -5.53694069e-01 -5.42820580e-02 7.04185426e-01
1.20463789e-01 -6.72176003e-01 9.64434803e-01 3.90104294e-01
-4.48743492e-01 -1.07815397e+00 -1.15945959e+00 -6.52511597e-01
6.36728466e-01 -2.77160913e-01 4.92675513e-01 9.91631150e-01
-2.23781914e-01 -3.87101293e-01 -1.70536071e-01 4.44463819e-01
9.51921165e-01 3.06071758e-01 5.92280388e-01 -9.22644377e-01
-1.00717343e-01 -5.95449924e-01 -6.98006570e-01 -1.14405739e+00
-2.00456396e-01 -9.13411558e-01 3.09988800e-02 -1.41472530e+00
2.54925191e-01 -4.92004126e-01 -5.41025996e-01 3.43780071e-01
-7.32662529e-02 3.77315342e-01 3.99534762e-01 2.25624770e-01
-3.17626268e-01 1.63007721e-01 1.67795408e+00 -2.52696574e-01
-7.82751739e-02 4.23229262e-02 -7.79937208e-01 8.18639100e-01
5.23280621e-01 -3.91586065e-01 -3.16934511e-02 -4.88030553e-01
-4.06679720e-01 2.58503884e-01 4.41774696e-01 -9.16612089e-01
1.11862518e-01 7.41327256e-02 5.52001297e-01 -5.93752742e-01
-5.79997115e-02 -1.04057753e+00 -1.05589598e-01 8.94615233e-01
-2.87399739e-01 -3.03750932e-01 3.03054303e-01 5.32727182e-01
-1.97579116e-01 1.07948212e-02 1.21404874e+00 -2.75795996e-01
-4.16234970e-01 7.98090577e-01 2.22854726e-02 2.20137134e-01
1.28407431e+00 -1.61234975e-01 -4.16052878e-01 -1.16198212e-01
-7.68331230e-01 3.43661994e-01 4.18590218e-01 1.67708084e-01
6.03376925e-01 -1.27937877e+00 -6.55720353e-01 4.01126176e-01
-9.42373574e-02 4.83256370e-01 3.16387028e-01 1.18479228e+00
-1.15862024e+00 2.08645165e-01 -4.57087368e-01 -8.08112025e-01
-8.70272815e-01 7.59394825e-01 5.42789698e-01 -5.21731675e-01
-1.16519213e+00 6.85480773e-01 4.16226596e-01 -3.57931942e-01
6.01713240e-01 -4.96472955e-01 4.72364724e-02 -4.17047799e-01
2.70789325e-01 1.17137849e-01 -7.83058926e-02 -3.83571565e-01
-4.66959059e-01 6.58795774e-01 -2.69145697e-01 3.70359123e-01
1.37707651e+00 -2.14950088e-02 1.96678475e-01 -3.76759738e-01
1.31587362e+00 -2.84855306e-01 -1.47427881e+00 -1.35640711e-01
-2.89636344e-01 -4.27903116e-01 4.82372344e-02 -8.86715591e-01
-1.67407119e+00 1.03261459e+00 8.83205712e-01 2.91625917e-01
1.25997376e+00 -1.21253319e-01 1.37396562e+00 1.05769090e-01
3.43101144e-01 -5.19476891e-01 -1.16558447e-01 4.72527683e-01
5.11563957e-01 -1.29587388e+00 1.39739783e-02 -3.78485262e-01
-6.71192527e-01 1.06817961e+00 5.32195985e-01 -5.60966551e-01
8.04302216e-01 3.89049828e-01 5.70578694e-01 1.49104847e-02
8.70151911e-03 1.33038834e-01 3.32055502e-02 7.59212017e-01
-4.26270030e-02 -5.13675399e-02 -2.11261377e-01 6.79092884e-01
1.43886313e-01 5.97741529e-02 4.02824670e-01 8.82344902e-01
-1.22778870e-01 -7.48464882e-01 -3.82439405e-01 4.45372313e-01
-9.76400316e-01 1.63711429e-01 -3.37023400e-02 1.15571177e+00
-2.68609375e-02 4.23593909e-01 -1.50363982e-01 -2.38091052e-01
3.63156468e-01 -3.22866768e-01 3.61510038e-01 1.14759743e-01
-5.98191500e-01 3.61698866e-01 -1.96451262e-01 -5.58631837e-01
-3.03495169e-01 -3.68782014e-01 -1.39204609e+00 -2.95222014e-01
-1.41709760e-01 -2.25773543e-01 5.66017985e-01 6.84507251e-01
1.83717936e-01 4.20115262e-01 7.12963223e-01 -8.36052120e-01
-1.00735736e+00 -6.59481227e-01 -6.10373139e-01 7.09788144e-01
3.73703778e-01 -6.06560171e-01 -3.81174147e-01 -2.27357954e-01] | [14.470877647399902, -2.470282793045044] |
ba5489e1-46f9-4c23-a1f3-67e4f7fcc605 | universal-multi-source-domain-adaptation | 2011.02594 | null | https://arxiv.org/abs/2011.02594v1 | https://arxiv.org/pdf/2011.02594v1.pdf | Universal Multi-Source Domain Adaptation | Unsupervised domain adaptation enables intelligent models to transfer knowledge from a labeled source domain to a similar but unlabeled target domain. Recent study reveals that knowledge can be transferred from one source domain to another unknown target domain, called Universal Domain Adaptation (UDA). However, in the real-world application, there are often more than one source domain to be exploited for domain adaptation. In this paper, we formally propose a more general domain adaptation setting, universal multi-source domain adaptation (UMDA), where the label sets of multiple source domains can be different and the label set of target domain is completely unknown. The main challenges in UMDA are to identify the common label set between each source domain and target domain, and to keep the model scalable as the number of source domains increases. To address these challenges, we propose a universal multi-source adaptation network (UMAN) to solve the domain adaptation problem without increasing the complexity of the model in various UMDA settings. In UMAN, we estimate the reliability of each known class in the common label set via the prediction margin, which helps adversarial training to better align the distributions of multiple source domains and target domain in the common label set. Moreover, the theoretical guarantee for UMAN is also provided. Massive experimental results show that existing UDA and multi-source DA (MDA) methods cannot be directly applied to UMDA and the proposed UMAN achieves the state-of-the-art performance in various UMDA settings. | ['Xiaofu Wu', 'Haifeng Hu', 'Zhen Yang', 'Yueming Yin'] | 2020-11-05 | null | null | null | null | ['universal-domain-adaptation'] | ['computer-vision'] | [ 1.95597574e-01 -1.40976787e-01 -4.26375568e-01 -2.75451660e-01
-7.97488749e-01 -8.63889515e-01 9.74055976e-02 -3.09995323e-01
-7.09145293e-02 1.00922608e+00 -3.50576550e-01 -9.26228985e-02
3.67092267e-02 -7.22640038e-01 -6.19920194e-01 -8.55313778e-01
4.62922007e-01 6.60908997e-01 4.33784187e-01 -1.72729701e-01
-3.53844702e-01 3.14743668e-01 -8.94662619e-01 -2.03272030e-02
1.04628980e+00 1.07976687e+00 3.08671296e-01 2.77887702e-01
-1.33092239e-01 5.50245523e-01 -8.09670091e-01 -2.05773175e-01
4.59726244e-01 -6.66103542e-01 -8.29976439e-01 2.30076790e-01
1.08732857e-01 -4.39021707e-01 -2.57159680e-01 1.27469206e+00
5.65588593e-01 2.79021561e-01 8.80128384e-01 -1.69477558e+00
-1.07821429e+00 2.62821287e-01 -6.53580725e-01 1.66156515e-01
7.09094480e-02 -1.83805242e-01 5.80833852e-01 -5.92481434e-01
5.31854033e-01 1.28471911e+00 4.31318790e-01 9.09886956e-01
-9.84761417e-01 -1.15807056e+00 4.15711731e-01 1.44867972e-01
-1.33233690e+00 -1.72500923e-01 8.50375831e-01 -3.91007870e-01
1.88862637e-01 -3.68160218e-01 -2.15186715e-01 1.19749558e+00
-3.00778687e-01 8.30009282e-01 1.15537763e+00 -4.19388592e-01
4.17404532e-01 5.07719576e-01 2.15368327e-02 2.03136057e-01
8.94950628e-02 5.94626777e-02 -8.00734684e-02 -3.82669896e-01
8.09651971e-01 5.48831336e-02 -2.19539672e-01 -6.44244611e-01
-1.03854942e+00 8.69568050e-01 2.58075863e-01 4.84602898e-02
-1.17967583e-01 -5.59493840e-01 5.20370781e-01 6.75492227e-01
4.21087027e-01 -8.21034424e-03 -9.22590315e-01 4.12700355e-01
-3.63112718e-01 3.32030877e-02 7.33024180e-01 1.51645315e+00
9.21856046e-01 2.20859069e-02 2.22846061e-01 9.92205620e-01
4.26898658e-01 9.41587090e-01 7.51283467e-01 -7.34381378e-01
6.78250492e-01 6.39159679e-01 3.05812925e-01 -6.68684721e-01
-1.54712573e-01 -2.75300890e-01 -9.08668935e-01 1.98873311e-01
7.23470926e-01 -4.88901407e-01 -7.58494377e-01 2.09080648e+00
7.16897488e-01 5.87334573e-01 5.63588917e-01 8.06404054e-01
3.95969093e-01 7.50194430e-01 5.20314090e-02 -4.34967875e-01
1.03361940e+00 -9.49911594e-01 -5.00794053e-01 -5.30382752e-01
5.36298096e-01 -6.88632607e-01 9.65911865e-01 1.21769592e-01
-5.57959259e-01 -7.08024621e-01 -1.13553357e+00 2.11748108e-01
-2.91565031e-01 -1.58616323e-02 -9.92308604e-04 5.62437415e-01
-3.58401388e-01 5.01472950e-02 -4.41188306e-01 -4.12609518e-01
5.63700438e-01 3.19433510e-01 -4.21422750e-01 -4.62475032e-01
-1.46876216e+00 7.70211577e-01 8.22458744e-01 -5.74927628e-01
-1.04489195e+00 -6.34970307e-01 -7.93932974e-01 -1.38712034e-01
5.74677110e-01 -4.96296704e-01 1.44912040e+00 -1.30316699e+00
-1.49310219e+00 7.71468520e-01 -2.03166708e-01 -3.49938035e-01
3.24119508e-01 -4.53682654e-02 -8.33494961e-01 -1.59955323e-02
2.53614545e-01 2.87616968e-01 9.26111281e-01 -1.43262827e+00
-9.79313433e-01 -3.99178982e-01 5.27360514e-02 2.51608253e-01
-4.13469732e-01 -6.63582527e-04 -2.01155484e-01 -7.53742814e-01
-2.01244920e-01 -8.97674561e-01 2.17093304e-02 3.26962885e-03
-1.67171329e-01 -3.29619735e-01 1.26973820e+00 -5.62779963e-01
1.07423139e+00 -2.39477468e+00 1.59933418e-01 1.92208722e-01
-4.30513583e-02 5.65726519e-01 -2.70347774e-01 -3.10982559e-02
-8.97906572e-02 -2.66592741e-01 -4.91659373e-01 -4.55534458e-02
1.11480905e-02 6.01330400e-01 -6.63348377e-01 2.55303293e-01
1.10155635e-01 4.34067816e-01 -1.04222071e+00 -4.40259218e-01
-8.53718966e-02 1.20815277e-01 -2.33633712e-01 5.96411228e-01
-1.98473886e-01 7.35058129e-01 -7.19119430e-01 7.63166010e-01
1.08588040e+00 -4.21706229e-01 3.20584059e-01 4.09534015e-02
7.07826376e-01 -1.73774645e-01 -1.40809822e+00 1.57418144e+00
-5.11160851e-01 2.10102230e-01 -1.95308123e-02 -1.15383184e+00
1.22108924e+00 4.16374832e-01 3.82207185e-01 -4.90188181e-01
1.77362129e-01 3.70179474e-01 2.23447569e-02 -1.41356423e-01
-3.32039967e-02 -5.40368378e-01 -3.19211960e-01 5.56229651e-01
3.60777199e-01 2.07715139e-01 -1.73993856e-01 -6.84783906e-02
6.78142250e-01 9.98490397e-03 7.04824269e-01 -5.53587871e-03
7.72274315e-01 -1.32956699e-01 1.06938565e+00 5.12139618e-01
-7.38580346e-01 5.48824847e-01 1.89024746e-01 -5.67353740e-02
-1.06001472e+00 -1.29805589e+00 -1.07332617e-01 1.29391599e+00
5.74574232e-01 3.86839241e-01 -6.98685288e-01 -1.33710003e+00
6.23694099e-02 6.12089336e-01 -4.04462993e-01 -3.72497320e-01
-4.27887619e-01 -5.37612617e-01 5.99880397e-01 5.02374053e-01
7.12470055e-01 -8.95426750e-01 -6.51582703e-03 3.33314240e-01
-2.15213031e-01 -1.31677103e+00 -7.62744367e-01 1.37909964e-01
-7.19527900e-01 -1.18183219e+00 -1.01878834e+00 -1.11029458e+00
6.96101546e-01 2.46570334e-01 9.46330070e-01 -6.01887167e-01
6.06379926e-01 2.30248094e-01 -4.38453883e-01 -4.94782001e-01
-8.38347435e-01 1.75630212e-01 5.11096776e-01 2.02501118e-01
7.63534248e-01 -6.90903246e-01 -1.95784137e-01 9.49193358e-01
-1.05425763e+00 -4.23890650e-01 5.12846529e-01 9.01501536e-01
6.87835693e-01 2.33668685e-01 1.44417787e+00 -1.07320571e+00
4.91485327e-01 -1.18203056e+00 -5.08499563e-01 5.10317206e-01
-6.17837250e-01 -1.53529271e-01 1.16043627e+00 -1.03980148e+00
-1.02249455e+00 7.45254289e-03 1.84888795e-01 -7.53488600e-01
-5.51849246e-01 3.30157250e-01 -8.68156850e-01 9.38570723e-02
7.83774495e-01 3.15895915e-01 6.14749938e-02 -5.95074832e-01
3.66079688e-01 9.58789110e-01 7.43592262e-01 -6.79726183e-01
1.12917554e+00 2.72858471e-01 -2.21394971e-01 -2.65013993e-01
-1.09573174e+00 -6.02740824e-01 -8.39826226e-01 1.98662132e-01
4.53061640e-01 -1.14199889e+00 -1.57020986e-02 7.50249386e-01
-1.00518608e+00 -3.70077550e-01 -3.12148690e-01 2.97425598e-01
-5.35138726e-01 6.37453020e-01 -1.52503297e-01 -4.59891438e-01
-2.01444298e-01 -1.01001203e+00 6.33982658e-01 5.97792983e-01
2.72788107e-02 -1.30913198e+00 3.96685721e-03 1.87150627e-01
9.62075442e-02 1.67141020e-01 9.60203111e-01 -1.28499603e+00
-9.45110545e-02 -8.87744874e-02 -2.38423035e-01 8.01810205e-01
6.67469978e-01 -7.69721866e-01 -7.22597778e-01 -6.03009343e-01
1.00780368e-01 -4.33943659e-01 1.56475201e-01 1.23811454e-01
7.73008823e-01 -3.01397949e-01 -3.64659935e-01 4.29977238e-01
1.22323477e+00 5.80957174e-01 2.94975430e-01 4.90134478e-01
6.20070875e-01 2.21330389e-01 1.02656817e+00 4.42019999e-01
4.24196661e-01 6.57515168e-01 2.73911148e-01 1.25769868e-01
-7.86554143e-02 -3.38253647e-01 6.32271469e-01 6.90836251e-01
7.23727226e-01 -4.49299991e-01 -8.08323324e-01 8.40225756e-01
-1.77224934e+00 -5.91264784e-01 3.87861311e-01 2.29438448e+00
1.05539191e+00 -7.10328817e-02 3.36794317e-01 -1.31266758e-01
1.20724583e+00 -2.08588228e-01 -1.30805397e+00 -9.77796316e-02
-2.91852176e-01 -1.47512510e-01 3.78047049e-01 3.12311441e-01
-1.37173188e+00 7.75178730e-01 6.00315952e+00 1.09196782e+00
-9.90595818e-01 4.14848179e-01 3.46011400e-01 2.01987624e-01
6.56972975e-02 -1.77273184e-01 -8.84541273e-01 8.34552169e-01
8.65207791e-01 -5.72819769e-01 3.73405039e-01 1.31229103e+00
-2.67147571e-01 3.27172399e-01 -1.14371419e+00 8.06828797e-01
-6.37933463e-02 -7.01496601e-01 4.97194752e-02 -7.26767778e-02
1.02920723e+00 -4.58347835e-02 5.62752075e-02 5.99174321e-01
6.83124602e-01 -5.15656769e-01 2.10288540e-01 -1.05055071e-01
1.19117486e+00 -9.88701284e-01 7.31868625e-01 7.12723732e-01
-1.09804487e+00 -2.82641441e-01 -6.49455905e-01 2.99294651e-01
2.77187172e-02 3.92254531e-01 -8.47442389e-01 7.85063565e-01
4.89846408e-01 6.97121382e-01 -1.71428189e-01 7.85264790e-01
-1.48739100e-01 6.57706380e-01 -3.25496733e-01 5.44577658e-01
2.00979561e-02 -4.05247584e-02 5.74283183e-01 7.98132539e-01
5.03115118e-01 8.97640828e-03 5.40432155e-01 5.60938656e-01
-4.57812011e-01 4.45146374e-02 -6.45986974e-01 2.21938416e-01
1.10559285e+00 7.48827457e-01 -1.05827518e-01 -3.33674550e-01
-5.89227974e-01 1.21005499e+00 3.04483503e-01 5.24084270e-01
-9.13251400e-01 -3.91044140e-01 9.46344614e-01 -1.22033119e-01
2.95768380e-01 1.71222255e-01 -4.00673486e-02 -1.34039009e+00
-2.76623257e-02 -1.00484395e+00 8.56247902e-01 -6.35682702e-01
-1.97873950e+00 4.40884113e-01 -2.18816474e-02 -1.74618244e+00
-3.03726047e-01 -4.41260368e-01 -3.83868188e-01 1.13900185e+00
-1.77105832e+00 -1.23346972e+00 -9.74246934e-02 1.12913287e+00
4.99397367e-01 -7.01913893e-01 9.75495100e-01 4.38352674e-01
-4.56298202e-01 1.21350408e+00 8.52078080e-01 3.92892331e-01
1.24360466e+00 -1.09530008e+00 8.21580142e-02 9.72318053e-01
-3.28249127e-01 3.12610596e-01 3.00339311e-01 -5.01765609e-01
-7.86552668e-01 -1.58698153e+00 4.21382636e-01 -3.18608671e-01
6.21012926e-01 -3.86155099e-02 -1.31907225e+00 1.02326119e+00
-1.39992326e-01 2.64435440e-01 9.41917419e-01 -1.65575087e-01
-6.96616173e-01 -1.91540137e-01 -1.58673716e+00 2.22218767e-01
6.29905820e-01 -5.69306552e-01 -7.26034403e-01 2.57682025e-01
9.06057596e-01 -4.44966406e-01 -1.09323382e+00 3.13756615e-01
1.52164266e-01 -4.60599333e-01 9.88569915e-01 -7.45098591e-01
2.48505354e-01 -5.78159928e-01 -2.47814074e-01 -1.67398131e+00
-2.92427421e-01 -2.41724536e-01 -4.65090156e-01 1.59047222e+00
1.11063980e-01 -9.47636664e-01 6.43332839e-01 3.65046382e-01
1.37131587e-01 -2.62052089e-01 -1.15465939e+00 -1.32207310e+00
6.42264605e-01 -5.03692627e-02 9.50098634e-01 1.34951699e+00
-2.57991135e-01 4.22387064e-01 -5.79898715e-01 7.07431138e-01
6.98950589e-01 1.91146776e-01 7.32917905e-01 -1.35353792e+00
-3.14457625e-01 1.84427891e-02 -2.63911575e-01 -1.40190804e+00
4.51470673e-01 -9.09064412e-01 -4.55511659e-02 -1.08465910e+00
1.06175825e-01 -7.83408105e-01 -8.27089846e-01 5.29189825e-01
-4.19658005e-01 -1.95363283e-01 2.06109032e-01 4.80546892e-01
-6.27148271e-01 5.28674066e-01 1.22433233e+00 -2.05303997e-01
-1.92138866e-01 1.61836907e-01 -1.03007507e+00 7.11771488e-01
8.62153053e-01 -6.54821217e-01 -7.95375586e-01 -4.52049047e-01
-5.37850380e-01 3.24063050e-03 1.65548213e-02 -1.05718517e+00
6.83179498e-02 -4.99159157e-01 2.40436032e-01 -2.37689927e-01
-1.37875602e-01 -1.26648200e+00 -1.73770830e-01 6.67452961e-02
-1.98906526e-01 -5.20663917e-01 2.21210435e-01 8.08525681e-01
-2.72951424e-01 -2.68542409e-01 1.34866273e+00 2.05393508e-01
-1.14172280e+00 4.92637485e-01 6.83391020e-02 3.73235732e-01
1.45582020e+00 3.77510823e-02 -4.21087682e-01 -3.24987888e-01
-6.79903209e-01 5.30379295e-01 5.09877801e-01 5.29353261e-01
4.42496747e-01 -1.72264755e+00 -6.86230242e-01 1.73940420e-01
2.90955245e-01 4.84191597e-01 4.10777956e-01 2.44516224e-01
1.14158399e-01 1.47497077e-02 -3.39378417e-01 -4.42805737e-01
-1.03288043e+00 1.06376994e+00 5.15508592e-01 -4.38577712e-01
-1.97739556e-01 7.27805495e-01 6.40907824e-01 -9.71810699e-01
1.15300179e-01 1.92015961e-01 -2.22921476e-01 -1.52919441e-01
6.99090362e-01 3.57814699e-01 -3.88382822e-01 -7.84139156e-01
-3.02759737e-01 5.40111959e-01 -2.59022802e-01 4.37443048e-01
8.73175144e-01 -6.72943711e-01 2.37070546e-01 3.70640397e-01
1.24343145e+00 -2.96829700e-01 -1.52766120e+00 -1.10076916e+00
-1.92231238e-01 -3.11015368e-01 -4.74025220e-01 -1.07173419e+00
-1.02513731e+00 7.74505079e-01 8.33849013e-01 -2.03563664e-02
1.47669733e+00 2.71732938e-02 1.11969960e+00 2.89149880e-01
4.48868215e-01 -1.18945992e+00 1.10935017e-01 6.34579539e-01
5.48502266e-01 -1.49831867e+00 -3.07855099e-01 -3.64156991e-01
-1.00636292e+00 8.76851082e-01 1.04712689e+00 -4.26195040e-02
6.72813654e-01 -1.11267762e-02 3.77932847e-01 4.39065635e-01
-3.52852166e-01 -8.03553592e-03 2.98275501e-02 1.23148549e+00
-2.71244675e-01 1.23920418e-01 2.74570048e-01 1.24702990e+00
2.71912873e-01 1.58849508e-01 3.98691177e-01 7.31715560e-01
-4.49861258e-01 -1.48612082e+00 -6.63054407e-01 7.61127472e-02
-3.74142885e-01 2.54791319e-01 -1.60242781e-01 6.83677495e-01
3.90104920e-01 1.13183355e+00 -2.72855520e-01 -4.23870564e-01
4.45457548e-01 3.32503855e-01 9.11875740e-02 -6.40326917e-01
7.24115595e-02 -1.26627147e-01 -4.25510913e-01 2.70850472e-02
-3.52692187e-01 -5.14296532e-01 -1.32133770e+00 -1.74446642e-01
-2.68606186e-01 1.50454268e-01 5.20772003e-02 1.14770508e+00
4.25950289e-01 3.42722952e-01 9.96283352e-01 -2.07806513e-01
-9.10787523e-01 -9.29413676e-01 -9.19272542e-01 5.25784433e-01
4.19759512e-01 -8.80320549e-01 -1.72137976e-01 3.89162600e-01] | [10.411643028259277, 3.1505088806152344] |
58a48d15-4770-4d6a-9845-57d2fe2e7515 | semantic-segmentation-of-rgbd-images-with | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Deng_Semantic_Segmentation_of_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Deng_Semantic_Segmentation_of_ICCV_2015_paper.pdf | Semantic Segmentation of RGBD Images With Mutex Constraints | In this paper, we address the problem of semantic scene segmentation of RGB-D images of indoor scenes. We propose a novel image region labeling method which augments CRF formulation with hard mutual exclusion (mutex) constraints. This way our approach can make use of rich and accurate 3D geometric structure coming from Kinect in a principled manner. The final labeling result must satisfy all mutex constraints, which allows us to eliminate configurations that violate common sense physics laws like placing a floor above a night stand. Three classes of mutex constraints are proposed: global object co-occurrence constraint, relative height relationship constraint, and local support relationship constraint. We evaluate our approach on the NYU-Depth V2 dataset, which consists of 1449 cluttered indoor scenes, and also test generalization of our model trained on NYU-Depth V2 dataset directly on a recent SUN3D dataset without any new training. The experimental results show that we significantly outperform the state-of-the-art methods in scene labeling on both datasets. | ['Longin Jan Latecki', 'Sinisa Todorovic', 'Zhuo Deng'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['scene-labeling'] | ['computer-vision'] | [ 3.77621919e-01 1.05392687e-01 2.24820096e-02 -9.10721660e-01
-2.47497737e-01 -6.09030008e-01 4.30942714e-01 -1.70727037e-02
-6.37975335e-01 7.12055743e-01 -2.53608495e-01 -4.23563451e-01
8.21563825e-02 -8.96770120e-01 -7.58661509e-01 -4.04494166e-01
9.41469297e-02 5.66075325e-01 5.74312210e-01 -6.65164664e-02
1.30186498e-01 5.08544266e-01 -1.46775246e+00 -1.48480251e-01
8.26702833e-01 1.07317376e+00 6.13774002e-01 5.41109204e-01
-8.52501914e-02 5.13805687e-01 -3.06252986e-01 2.89696664e-01
5.12424767e-01 -2.08056450e-01 -9.25283372e-01 4.77324784e-01
5.86716652e-01 -2.96460092e-01 5.80853187e-02 1.01937330e+00
3.02988827e-01 4.27635133e-01 5.46397209e-01 -9.79711950e-01
-2.41261959e-01 -1.08653836e-01 -6.79631114e-01 9.42904651e-02
6.38106763e-01 -2.04267085e-01 6.55127287e-01 -4.09740269e-01
7.30457842e-01 1.02646124e+00 3.96333635e-01 4.19165671e-01
-8.85597765e-01 -2.63930351e-01 6.96609378e-01 -1.14615582e-01
-1.51897132e+00 1.36855701e-02 7.74945021e-01 -3.55760425e-01
1.05407107e+00 4.72612977e-01 5.70931554e-01 7.96913266e-01
-9.04238820e-02 5.82455814e-01 1.56010461e+00 -4.75587845e-01
5.61915517e-01 -6.53501377e-02 2.86761910e-01 8.27146053e-01
-4.68446780e-03 -9.13571864e-02 -1.59534588e-01 1.05539329e-01
8.87691021e-01 -5.14384406e-03 -1.75451607e-01 -6.66691005e-01
-9.68891978e-01 5.66570282e-01 6.31415784e-01 1.53228864e-01
7.06620961e-02 -9.12016705e-02 -1.70281246e-01 -3.28273177e-01
3.88931781e-01 -1.88967884e-02 -8.18082035e-01 2.71219254e-01
-5.98675787e-01 3.49298269e-01 6.16037905e-01 1.13906884e+00
9.26207185e-01 -4.91926283e-01 2.68943787e-01 5.59896886e-01
6.39571309e-01 7.26963580e-01 1.61491320e-01 -9.84166861e-01
4.51374382e-01 5.39593697e-01 1.02830715e-01 -6.51987135e-01
-7.26331830e-01 -1.30531624e-01 -5.09419680e-01 2.16657922e-01
3.58467907e-01 1.00419492e-01 -1.54034722e+00 1.36368024e+00
6.85516000e-01 8.35889652e-02 -1.19960001e-02 1.01868010e+00
1.03152442e+00 4.71753776e-01 -2.00021163e-01 -3.35276476e-03
1.18762910e+00 -8.57479453e-01 -5.14421940e-01 -4.33041811e-01
4.00576055e-01 -6.44885063e-01 9.86911058e-01 2.79293537e-01
-4.85241503e-01 -8.30453992e-01 -1.04011858e+00 -3.06152403e-01
-6.68551803e-01 1.86977126e-02 9.58030522e-01 7.70476639e-01
-7.38695502e-01 3.90247524e-01 -1.00147998e+00 -6.33598566e-01
1.74908146e-01 3.95089030e-01 -3.76586765e-01 -3.63347054e-01
-6.82025313e-01 7.37099171e-01 5.90084553e-01 1.99735820e-01
-7.35062897e-01 -1.13294713e-01 -1.25231516e+00 -4.17656600e-01
5.83833277e-01 -5.77777505e-01 1.11575174e+00 -3.70588601e-01
-1.43108010e+00 1.27910459e+00 -1.60120428e-01 -2.77192414e-01
4.66956258e-01 -4.36862141e-01 -1.42939046e-01 -1.82123836e-02
2.48912334e-01 7.77371347e-01 2.17255026e-01 -1.62550497e+00
-4.12059963e-01 -6.23589754e-01 2.78734803e-01 5.35359621e-01
3.47175807e-01 -3.99177670e-01 -6.46918476e-01 -6.10202849e-02
6.69254243e-01 -1.05971026e+00 -7.09123552e-01 -2.28346676e-01
-8.10697794e-01 -1.14983305e-01 8.72953117e-01 -4.29747701e-01
5.36432505e-01 -1.97278309e+00 -7.90129323e-03 4.52474415e-01
-2.24964380e-01 -8.00776333e-02 2.87436754e-01 -8.75397120e-03
2.24731892e-01 6.83789030e-02 -7.08313882e-01 -5.55672467e-01
-1.42870669e-03 7.85106122e-01 -9.58898589e-02 5.58831215e-01
-2.49245763e-01 6.48030281e-01 -8.15497220e-01 -6.73365474e-01
8.11477125e-01 3.29278111e-01 -3.98254693e-01 2.59344637e-01
-5.10878205e-01 7.42913783e-01 -6.17389023e-01 9.00171041e-01
1.07715130e+00 -8.62811208e-02 2.77781874e-01 9.12812725e-02
-2.79097557e-01 3.89935195e-01 -1.23308313e+00 2.47483683e+00
-2.15596631e-01 1.40561327e-01 -1.56711414e-01 -7.44752765e-01
9.53906834e-01 -2.55652487e-01 3.90828371e-01 -8.73558164e-01
2.93607056e-01 1.24318888e-02 -4.75502431e-01 -4.54642832e-01
4.80918884e-01 3.42206545e-02 -1.87627569e-01 5.39185628e-02
-1.37151200e-02 -7.47045994e-01 -2.41273618e-03 5.51737994e-02
7.97926188e-01 7.89741397e-01 3.60234141e-01 -4.23481703e-01
3.14570069e-01 1.31085301e-02 5.56416512e-01 8.69644284e-01
-2.17513308e-01 9.32758987e-01 -8.12154412e-02 -4.12510008e-01
-5.97029865e-01 -1.27081251e+00 -2.89687902e-01 4.20281827e-01
6.77871943e-01 -2.78330028e-01 -7.31749296e-01 -8.87305498e-01
-1.57142773e-01 5.78560174e-01 -6.93567157e-01 5.50881982e-01
-3.90872389e-01 -8.28362823e-01 4.54014130e-02 4.92487878e-01
9.25173163e-01 -7.05173790e-01 -1.02286065e+00 -1.33779347e-01
-2.34419063e-01 -1.68748403e+00 5.84764518e-02 5.57300627e-01
-7.73445606e-01 -1.24976897e+00 -3.72827977e-01 -6.35307431e-01
7.22433388e-01 3.46582741e-01 1.04467332e+00 -7.01646507e-02
-4.92910951e-01 4.53516185e-01 -4.48616445e-01 -4.26933706e-01
2.48091474e-01 1.91647541e-02 -2.94229463e-02 -4.18741554e-01
3.45071703e-01 -4.99079913e-01 -3.37903380e-01 4.45847809e-01
-7.09816754e-01 3.02190542e-01 2.14571357e-01 1.01865269e-01
1.16941309e+00 1.70412779e-01 -2.21571505e-01 -8.28273237e-01
-3.66996318e-01 -1.04441583e-01 -8.35527182e-01 1.95333228e-01
-2.90744722e-01 -1.14692919e-01 8.89228750e-03 9.78725478e-02
-1.19223177e+00 6.04865491e-01 -3.41016650e-01 -2.35695258e-01
-7.43542492e-01 2.17668936e-02 -5.48636436e-01 -1.46608964e-01
4.05453742e-01 1.54535128e-02 -6.80042446e-01 -6.48454726e-01
3.52489024e-01 5.12251377e-01 8.51361990e-01 -8.41958165e-01
7.87323594e-01 9.06897843e-01 1.21719673e-01 -1.07501245e+00
-1.16418135e+00 -7.35472918e-01 -1.15732539e+00 -8.43155161e-02
1.36449242e+00 -9.96157289e-01 -5.76610386e-01 3.79342943e-01
-1.05192840e+00 -6.52842700e-01 -9.35044438e-02 5.46312451e-01
-5.48820376e-01 3.78058195e-01 -3.98178309e-01 -9.10159111e-01
2.99823701e-01 -9.95452285e-01 1.39468026e+00 4.25807565e-01
9.24392715e-02 -8.14960003e-01 1.15390360e-01 5.79645634e-01
-2.73491591e-01 8.73025656e-01 5.06117761e-01 -1.63682878e-01
-7.40780354e-01 2.60216594e-01 -2.04802737e-01 3.24257314e-01
2.80004442e-01 -2.24055365e-01 -1.27738595e+00 1.24876544e-01
1.28500342e-01 -2.54379749e-01 1.00229609e+00 4.45599616e-01
1.28124380e+00 4.15606588e-01 -3.72792721e-01 7.54104376e-01
1.74005449e+00 2.43153334e-01 7.41109371e-01 5.06828129e-01
1.02035248e+00 6.22915268e-01 9.30331767e-01 2.99991995e-01
7.70707130e-01 6.48713648e-01 8.63124192e-01 -4.19347286e-01
1.33840114e-01 -2.20755383e-01 -1.98873684e-01 4.90253329e-01
-3.47311109e-01 -3.55368048e-01 -9.37272489e-01 3.88011903e-01
-1.95293868e+00 -4.28922802e-01 -4.88024592e-01 2.08378649e+00
5.50798535e-01 2.89849520e-01 -2.92590380e-01 8.06963667e-02
3.65730762e-01 1.59923375e-01 -3.08447748e-01 -1.57846749e-01
-1.83753937e-01 2.59464592e-01 8.29668283e-01 8.25463653e-01
-1.52618468e+00 9.99939322e-01 5.89092302e+00 4.62375134e-01
-6.20039880e-01 3.71551290e-02 5.34510314e-01 1.63136169e-01
-1.48242891e-01 1.20834708e-01 -9.06076014e-01 1.33041173e-01
3.86067867e-01 6.77186072e-01 1.48684487e-01 1.04637778e+00
-6.28054813e-02 -8.03489804e-01 -1.00364757e+00 1.07532060e+00
9.19293240e-02 -9.19210553e-01 -5.75715959e-01 2.41659671e-01
9.12919700e-01 3.39294642e-01 -3.52247596e-01 -8.57362058e-03
4.06490147e-01 -1.03116477e+00 8.18173170e-01 2.99902588e-01
4.91813987e-01 -5.92711091e-01 6.73257649e-01 5.70745766e-01
-1.36174679e+00 4.61705238e-01 -4.12178963e-01 -4.02505517e-01
2.81751543e-01 6.99908733e-01 -7.63705611e-01 1.02804399e+00
1.00393438e+00 5.42250454e-01 -6.59950197e-01 8.64703953e-01
-5.55896461e-01 1.30337521e-01 -7.47724652e-01 1.71974659e-01
1.69358626e-01 -3.54262084e-01 2.34855823e-02 9.85458136e-01
4.09632586e-02 4.78942364e-01 6.26245022e-01 7.79391289e-01
1.86082318e-01 -2.43834585e-01 -7.54454672e-01 5.25506973e-01
1.72499067e-03 1.12986672e+00 -1.42915869e+00 -2.27523431e-01
-2.48747364e-01 1.30531788e+00 9.70902592e-02 3.29767197e-01
-8.63582611e-01 4.14362997e-02 3.22286636e-01 1.33587457e-02
3.52818102e-01 -7.26673126e-01 -3.60942781e-01 -1.14325023e+00
1.92244381e-01 -2.18620673e-01 1.81288600e-01 -1.01614428e+00
-9.60823774e-01 5.33395290e-01 3.72662753e-01 -8.74230444e-01
3.16472016e-02 -7.43079126e-01 -2.15680152e-01 6.07131720e-01
-1.77292657e+00 -1.27513361e+00 -6.90224051e-01 7.59790659e-01
4.69813734e-01 6.07087195e-01 9.60472763e-01 2.96490222e-01
-3.18098277e-01 -2.02345148e-01 -3.39298844e-01 1.59889624e-01
2.99701720e-01 -1.39980531e+00 3.43518674e-01 9.26618040e-01
2.49698266e-01 3.69906574e-01 6.23700738e-01 -7.70528316e-01
-1.01489365e+00 -1.04980910e+00 6.75378025e-01 -6.66027367e-01
3.46323149e-03 -8.42765272e-01 -5.04151702e-01 8.51147056e-01
-1.13740340e-01 2.87956148e-01 5.94586372e-01 4.16482687e-02
-1.22940205e-01 2.30972692e-01 -1.30568624e+00 1.06808871e-01
1.41799474e+00 -3.26496512e-01 -5.46047330e-01 6.09165013e-01
9.70609009e-01 -1.05251753e+00 -5.54688871e-01 8.47791374e-01
2.25412592e-01 -1.13537753e+00 1.07261133e+00 -7.19456971e-02
2.95126308e-02 -7.69506037e-01 -8.86779785e-01 -6.54388428e-01
2.47995600e-01 -1.59923956e-01 1.69487372e-01 1.17234719e+00
3.83293033e-02 -3.66567940e-01 9.04687285e-01 7.46483386e-01
-3.59097987e-01 -5.00299633e-01 -9.40940022e-01 -9.50682104e-01
-3.55746567e-01 -7.84696281e-01 4.96359318e-01 9.79593337e-01
-5.72759092e-01 1.88702583e-01 -1.38704315e-01 4.88514453e-01
8.42731595e-01 4.01986063e-01 8.98230195e-01 -1.21449363e+00
-3.43470991e-01 1.06506310e-01 -4.68586296e-01 -1.31650114e+00
4.11193967e-02 -5.44146240e-01 2.93668658e-01 -1.94996917e+00
3.65744755e-02 -7.70751834e-01 -3.66328061e-02 5.61673164e-01
1.52498722e-01 4.14096653e-01 1.39124528e-01 -1.24826983e-01
-8.10587704e-01 6.28192186e-01 1.13624275e+00 -2.94663594e-03
-2.87236422e-01 -3.49909179e-02 -1.36051327e-01 1.12951410e+00
7.14586258e-01 -3.78009081e-01 -5.20120740e-01 -4.07501817e-01
2.70102862e-02 -2.10950181e-01 4.64044392e-01 -1.17789030e+00
-1.60876587e-01 -4.42440301e-01 6.85180664e-01 -1.08349979e+00
5.26188433e-01 -9.69604075e-01 -8.34639147e-02 3.07669610e-01
2.73578703e-01 -2.61245608e-01 2.89634615e-01 4.83290344e-01
1.09228201e-01 -4.18634824e-02 5.92418671e-01 -5.24564445e-01
-1.09780550e+00 1.93748549e-01 7.17210257e-03 -2.24855170e-01
1.17402601e+00 -3.51683825e-01 -1.39078096e-01 1.17100999e-01
-8.07012022e-01 3.34965110e-01 7.69867063e-01 4.75338697e-01
6.62292421e-01 -1.07101357e+00 -1.73291802e-01 2.99378186e-01
1.88618958e-01 9.21269059e-01 2.05202386e-01 5.17774045e-01
-6.70146823e-01 4.33694601e-01 -7.00976923e-02 -1.02255070e+00
-1.22600639e+00 4.00557071e-01 3.10751110e-01 -1.60532013e-01
-6.93351388e-01 1.00015926e+00 4.72893983e-01 -1.02579141e+00
1.54842898e-01 -8.14072311e-01 -1.79052586e-03 -5.02622843e-01
1.14692636e-01 -9.81044695e-02 1.38643205e-01 -7.01563597e-01
-8.09429824e-01 9.23671126e-01 3.63996685e-01 -2.10668445e-01
1.07463598e+00 -3.74187738e-01 -1.18712239e-01 5.61610162e-01
8.85641575e-01 1.21479044e-02 -1.30341899e+00 -9.77504477e-02
-2.23526448e-01 -6.35224402e-01 -1.49339333e-01 -8.86458635e-01
-6.44433200e-01 7.25185752e-01 8.60772908e-01 1.95075180e-02
1.10306859e+00 2.75535375e-01 4.59314585e-01 4.24836814e-01
1.00606441e+00 -1.10105872e+00 -1.33787796e-01 6.11345291e-01
3.81043732e-01 -1.48437417e+00 3.27481002e-01 -9.22657609e-01
-3.99242967e-01 8.18794489e-01 8.10219467e-01 -1.19387478e-01
6.56601191e-01 1.12153739e-01 7.84954205e-02 -3.18050474e-01
1.15682565e-01 -6.69271946e-01 2.52686888e-01 8.25154483e-01
1.60217360e-01 3.83671135e-01 -8.70861560e-02 1.46815419e-01
-4.61081564e-01 -2.63297617e-01 3.52666318e-01 1.28744078e+00
-5.61499834e-01 -1.11480534e+00 -5.42089701e-01 -3.73345353e-02
-1.30561233e-01 1.23471044e-01 -4.41866815e-01 1.11954343e+00
7.38430202e-01 1.02582109e+00 8.26838315e-02 -3.25795680e-01
2.19801784e-01 -2.04731748e-01 9.17778432e-01 -8.66899133e-01
-5.92379831e-03 2.33892173e-01 4.75977995e-02 -7.42817581e-01
-8.93256605e-01 -6.40401363e-01 -1.85265350e+00 2.07510099e-01
-3.03875297e-01 -5.61549887e-02 1.08820963e+00 1.21161389e+00
-1.81757838e-01 6.76866591e-01 3.73432487e-01 -1.06535256e+00
2.58242965e-01 -7.70966291e-01 -6.68207169e-01 1.91715181e-01
3.09282899e-01 -8.09058011e-01 -3.11251301e-02 1.55515477e-01] | [8.455305099487305, -2.6004371643066406] |
4b6fd568-a9fb-46f8-ad84-5380f37f6491 | on-the-origins-of-the-omicron-variant-of-the | 2201.07879 | null | https://arxiv.org/abs/2201.07879v1 | https://arxiv.org/pdf/2201.07879v1.pdf | On the origins of the Omicron variant of the SARS-CoV-2 virus | A possible explanation based on first principles for the appearance of the Omicron variant of the SARS-CoV-2 virus is proposed involving coinfection with HIV. The gist is that the resultant HIV-induced immunocompromise allows SARS-CoV-2 greater latitude to explore its own mutational space. This latitude is not withoutr estriction, and a specific biophysical constraint is explored. Specifically, a nearly two- to five-fold discrepancy in backbone hydrogen bonding is observed between sub-molecules in Protein Data Bank files of the spike glycoprotein yielding two conclusions: mutagenic residues in the receptor-binding subunit of the spike much more frequently do not participate in backbone hydrogen bonds; and a technique of viral escape is therefore to remove such bonds within physico-chemical and functional constraints. Earlier work, from which the previous discussion is entirely independent, explains these phenomena from general principles of free energy, namely, the metastability of the glycoprotein. The conclusions therefore likely hold more generally as principles in virology. | ['Minus van Baalen', 'Robert Penner'] | 2021-12-08 | null | null | null | null | ['virology'] | ['miscellaneous'] | [ 4.56818998e-01 1.15790367e-01 -9.26967710e-02 -3.42921168e-01
-2.04939932e-01 -4.75965470e-01 3.76924634e-01 2.73504585e-01
-5.04682183e-01 1.26199555e+00 3.50810289e-02 -8.26663435e-01
8.96780267e-02 -4.46741134e-01 -6.21331215e-01 -1.00101125e+00
-1.94223404e-01 5.97365499e-01 5.88618740e-02 -2.96370953e-01
4.69749987e-01 8.94693434e-01 -9.83795941e-01 7.34978393e-02
1.05937386e+00 5.82324713e-02 2.17453226e-01 5.05966246e-01
-1.61641628e-01 1.57620147e-01 -3.72532368e-01 -7.36745000e-02
1.67020127e-01 -7.30329156e-01 -4.54797745e-01 -1.81190416e-01
1.21178150e-01 -1.48962587e-01 3.10685813e-01 6.19102836e-01
1.51222616e-01 -2.71786869e-01 1.16003263e+00 -9.26535964e-01
-3.28457743e-01 -3.48600686e-01 -6.81082308e-01 9.46328416e-02
3.36207151e-01 1.80319637e-01 6.48773432e-01 -7.61177957e-01
1.23212743e+00 1.12354934e+00 6.32096231e-01 3.93527836e-01
-1.49528337e+00 -3.15004826e-01 -3.56405258e-01 -1.98759750e-01
-1.47548604e+00 -3.37249219e-01 3.38928610e-01 -7.08455861e-01
1.38303745e+00 5.93196809e-01 8.32811415e-01 6.15227640e-01
1.10402203e+00 1.12268245e-02 1.18624449e+00 -2.16834456e-01
2.77493566e-01 -9.05814208e-03 6.50399625e-02 6.23032391e-01
1.15916729e+00 2.12620527e-01 -8.60272571e-02 -9.66556191e-01
6.35101795e-01 2.68861592e-01 -3.90831560e-01 -6.19589210e-01
-7.70621359e-01 7.47107804e-01 -4.54918481e-02 4.46881682e-01
-4.04394686e-01 -2.87584990e-01 2.95510560e-01 1.86746463e-01
-9.02757049e-02 3.97476047e-01 -6.09303951e-01 -2.85024401e-02
-5.79280674e-01 3.95049065e-01 8.12448084e-01 3.79628658e-01
7.16885746e-01 -2.56406754e-01 8.12594533e-01 1.30187467e-01
7.05814898e-01 5.67690074e-01 -1.74794272e-01 -3.69797587e-01
7.35226944e-02 3.41778040e-01 3.92340422e-01 -6.83156848e-01
-5.60399532e-01 1.36585847e-01 -4.46070284e-01 4.47963268e-01
4.43935007e-01 -4.48876262e-01 -8.50406706e-01 1.76256454e+00
2.73296654e-01 -2.78562576e-01 2.27517083e-01 6.38649881e-01
8.57597217e-02 5.99513829e-01 3.53081882e-01 -1.00079584e+00
1.52550161e+00 -1.12084411e-01 -5.69710314e-01 4.42760214e-02
4.93522406e-01 -6.96133316e-01 2.35043958e-01 8.77973065e-02
-6.86274052e-01 -3.36213075e-02 -1.29682004e+00 5.04080296e-01
-1.37279198e-01 -8.24199736e-01 6.01017654e-01 7.44329989e-01
-9.39974010e-01 5.89600861e-01 -9.41880047e-01 -1.21201932e+00
-1.92353189e-01 4.37955856e-01 -3.70419532e-01 3.48027915e-01
-1.03911459e+00 1.12202299e+00 3.95226270e-01 -4.45180237e-02
-1.00099392e-01 -3.43088418e-01 -4.19762999e-01 -2.07151577e-01
-1.77154765e-01 -8.55186760e-01 5.23048341e-01 -7.90957034e-01
-1.01835060e+00 8.87223780e-01 -5.19395590e-01 -1.60279736e-01
1.67674229e-01 1.92340061e-01 -5.41261911e-01 2.09770933e-01
-2.58905858e-01 2.25040644e-01 6.02368236e-01 -1.53813052e+00
-1.77990749e-01 -7.76423037e-01 -6.53886378e-01 1.52293339e-01
8.67580056e-01 4.51296717e-01 2.27044761e-01 -4.75568205e-01
1.81439787e-01 -1.10921431e+00 -4.69365388e-01 -4.15016919e-01
-1.50882294e-02 -1.42396182e-01 5.09182215e-01 -2.96150744e-01
7.31754005e-01 -1.88852501e+00 -2.44669318e-01 6.42097294e-01
2.98866540e-01 2.30804861e-01 1.02497160e-01 1.08245480e+00
-4.35055673e-01 4.97720093e-01 -9.76540148e-02 7.92189777e-01
-4.71486926e-01 1.35408998e-01 -2.96867102e-01 8.19004893e-01
7.07173496e-02 5.70700288e-01 -6.64950788e-01 -2.03306705e-01
-3.05543929e-01 5.93757570e-01 -5.44154346e-01 -1.59158245e-01
-5.60214184e-02 2.80905366e-01 -8.37131858e-01 4.31557149e-01
1.07143784e+00 -1.13342978e-01 1.06969881e+00 8.03671628e-02
-4.34292912e-01 2.78005481e-01 -4.21715051e-01 1.10240233e+00
7.30709970e-01 1.63230330e-01 1.26058802e-01 -4.18098301e-01
6.88428819e-01 4.57354516e-01 4.59033668e-01 -2.74594784e-01
1.73965305e-01 5.17171383e-01 4.88990068e-01 -3.23288441e-01
1.85972810e-01 -7.06155896e-01 4.40425634e-01 1.62334740e-01
-5.52357674e-01 2.83586264e-01 5.57171069e-02 1.47899121e-01
9.07866061e-01 2.90888697e-01 4.83451396e-01 -9.73718286e-01
3.31117898e-01 2.49227181e-01 8.94047439e-01 4.79117751e-01
-3.62470448e-01 2.44506165e-01 5.79760551e-01 -4.31766212e-01
-1.26391244e+00 -8.97658169e-01 -5.82913518e-01 3.85376006e-01
1.95051000e-01 -3.44855189e-01 -8.59485984e-01 -6.46518096e-02
2.92966366e-01 7.25279868e-01 -4.90753412e-01 1.25882059e-01
-7.71116793e-01 -1.19238138e+00 3.01207155e-01 1.33257290e-05
-3.16855192e-01 -7.32748270e-01 -8.45710278e-01 4.03683782e-01
1.58379674e-01 -3.15177828e-01 -2.71315202e-02 5.34493983e-01
-1.22060359e+00 -1.35315812e+00 -4.27782327e-01 -4.51443464e-01
7.75880277e-01 2.44102255e-01 4.75874841e-01 5.87886095e-01
-3.41451019e-01 -5.15793413e-02 3.16418298e-02 -4.41936374e-01
-7.05366671e-01 -5.06319284e-01 3.77696395e-01 -6.49246454e-01
1.02291882e+00 -5.41592181e-01 -6.19167149e-01 1.55518577e-01
-6.26848578e-01 -2.91538984e-01 4.63542044e-01 4.83841836e-01
4.32198793e-01 -3.70218247e-01 6.02048099e-01 -8.64160359e-01
3.25594127e-01 -5.72985053e-01 -5.04415870e-01 2.77275980e-01
-9.06117082e-01 1.38839871e-01 3.21479827e-01 2.19882369e-01
-1.14365053e+00 -2.61099070e-01 -2.58683324e-01 4.61726159e-01
-2.68287539e-01 1.11673132e-01 -3.28396142e-01 5.58656976e-02
5.28559983e-01 2.79861033e-01 5.14151871e-01 -3.69307011e-01
-2.69347161e-01 3.49569350e-01 2.52913814e-02 -4.23110425e-01
8.00996959e-01 5.66761971e-01 6.01029508e-02 -1.23941183e+00
1.05005056e-01 -2.90697604e-01 -1.03399232e-01 2.19044521e-01
1.15889680e+00 -6.24962032e-01 -1.04448152e+00 2.23370194e-01
-1.31341684e+00 1.27376556e-01 3.12030792e-01 8.55888963e-01
-7.33553290e-01 6.84632480e-01 -6.28631592e-01 -7.20291495e-01
-1.25700668e-01 -1.06022358e+00 5.08347213e-01 2.57577803e-02
-5.55442333e-01 -7.71764219e-01 7.23975241e-01 3.60224038e-01
2.18950197e-01 3.54653448e-01 1.59325576e+00 -9.80206013e-01
-6.93659782e-01 1.82536915e-01 -8.69520381e-02 -3.66148561e-01
2.45939851e-01 4.55591530e-01 -5.00304699e-01 -5.76438665e-01
2.52906263e-01 -6.50555966e-03 6.59880757e-01 3.11670989e-01
-1.25981927e-01 3.93553711e-02 -8.57585371e-01 3.13339263e-01
1.83445311e+00 1.17955422e+00 9.15117145e-01 3.32230300e-01
1.25449777e-01 4.91168529e-01 5.70304632e-01 2.21871093e-01
-1.68834835e-01 5.16074955e-01 1.87638327e-01 -1.51331082e-01
6.92023993e-01 1.31238982e-01 1.64252162e-01 4.49753821e-01
-6.49795353e-01 -3.52021009e-01 -9.73193884e-01 5.35233170e-02
-1.44937408e+00 -1.06437945e+00 -3.88575375e-01 2.26918626e+00
8.37589800e-01 5.87962009e-02 -5.32022752e-02 -6.44793928e-01
6.60008371e-01 -1.36669781e-02 -5.78631282e-01 -7.20119417e-01
-9.71622244e-02 -1.71078369e-01 7.40008175e-01 8.95671010e-01
-3.60781997e-01 4.00143147e-01 7.84541273e+00 3.02776098e-01
-8.81720722e-01 -4.27693546e-01 3.65303040e-01 2.71862447e-01
-7.38739491e-01 4.80737358e-01 -7.10847199e-01 3.91641587e-01
9.99030113e-01 -1.47348255e-01 7.55408183e-02 1.98542431e-01
3.97637755e-01 -3.97549659e-01 -8.94721985e-01 6.13399863e-01
-4.61988226e-02 -1.31073570e+00 2.66833603e-02 6.86811388e-01
2.98860431e-01 1.59071665e-02 -3.55300486e-01 -1.76546693e-01
-1.67062715e-01 -8.27194810e-01 2.78900694e-02 5.92199862e-01
6.79479718e-01 -9.77717102e-01 6.04073822e-01 1.81729481e-01
-8.71747792e-01 5.02807140e-01 -4.28079784e-01 1.92571566e-01
3.04125965e-01 2.31572509e-01 -9.87584889e-01 2.61834055e-01
1.49686053e-01 -2.70287752e-01 -1.03164263e-01 7.06290126e-01
3.47694486e-01 1.88660979e-01 -3.11450422e-01 -9.51782428e-03
6.20360598e-02 -8.73259068e-01 7.69977629e-01 9.99946296e-01
3.34144309e-02 6.07383490e-01 -1.05426617e-01 5.75511694e-01
4.72799420e-01 2.44072393e-01 -5.75168550e-01 -2.66444772e-01
2.28064775e-01 6.47132277e-01 -8.64993870e-01 -2.94814527e-01
-5.43153167e-01 9.76970077e-01 -6.79557472e-02 5.60963809e-01
-5.67714572e-01 -3.53973150e-01 1.18128574e+00 4.38898414e-01
3.88858497e-01 -1.00965947e-02 1.38594344e-01 -9.87307668e-01
-3.25856984e-01 -9.06739354e-01 6.44420460e-02 -3.48848194e-01
-6.37477815e-01 6.26158953e-01 1.80308558e-02 -4.05692667e-01
-4.94234115e-01 -5.10945737e-01 -4.60455358e-01 1.10892510e+00
-1.09499311e+00 -6.19792700e-01 6.61540747e-01 1.42117843e-01
1.19492628e-01 -8.83718580e-02 9.13232446e-01 -2.92124331e-01
-3.03768247e-01 7.10723922e-02 7.27714360e-01 -4.03908640e-01
7.54216671e-01 -1.03134692e+00 4.06365931e-01 6.13934219e-01
-6.71267807e-01 1.54289961e+00 1.33693290e+00 -1.47756422e+00
-1.58849084e+00 -4.98842150e-01 1.37091053e+00 -5.53301811e-01
5.82060218e-01 -2.80184984e-01 -9.07618880e-01 6.18829489e-01
7.09284186e-01 -8.75416100e-01 1.10713565e+00 4.26250212e-02
-2.14785144e-01 5.02818406e-01 -1.27792060e+00 7.54296720e-01
7.96901047e-01 -3.89493108e-01 -8.94869030e-01 4.27623451e-01
3.79955471e-01 2.21877337e-01 -7.26997674e-01 2.98296958e-01
8.19866240e-01 -1.23541033e+00 6.36774421e-01 -9.73397493e-01
-4.67399359e-01 -5.90460837e-01 -4.59013972e-03 -8.09906483e-01
-3.85182172e-01 -9.78725493e-01 5.93089938e-01 5.55796802e-01
4.59848583e-01 -9.93910491e-01 6.08103156e-01 6.68518364e-01
2.29455426e-01 -8.00633132e-01 -8.44760299e-01 -8.35477769e-01
1.11054361e-01 3.59219015e-01 3.35594714e-01 9.09259021e-01
4.09192413e-01 4.35734063e-01 -1.83953568e-01 -1.70929447e-01
6.51276231e-01 1.46469682e-01 4.52038556e-01 -1.44662404e+00
-1.96400881e-01 6.93490133e-02 -2.76520789e-01 -6.15552068e-01
-2.79915243e-01 -4.73965377e-01 -2.50465393e-01 -1.22255957e+00
5.03738582e-01 -3.68121266e-01 -2.26813629e-01 -2.87443977e-02
1.57284722e-01 -3.03269774e-01 2.62895256e-01 4.82773721e-01
7.91249052e-02 2.75714286e-02 7.42155015e-01 5.38792193e-01
-3.95971924e-01 -4.05866712e-01 -7.59718180e-01 8.74119878e-01
8.43760312e-01 -6.21224463e-01 -4.66574579e-01 2.49736950e-01
5.34999073e-01 3.11641097e-01 -2.08859280e-01 -2.40735695e-01
-4.30285662e-01 -6.16273940e-01 4.15010184e-01 -7.72303939e-01
2.68805891e-01 -1.11832476e+00 1.00975907e+00 1.12161124e+00
4.63154167e-01 2.98611701e-01 1.72987431e-01 5.84370732e-01
4.79386210e-01 -1.10183209e-01 1.04242194e+00 -1.65759951e-01
-1.02482349e-01 -6.12089820e-02 -1.15785456e+00 -1.16217330e-01
1.13898695e+00 -8.66887987e-01 -4.25624222e-01 1.23642027e-01
-7.08779037e-01 -3.44441354e-01 1.27410650e+00 1.23800253e-02
3.85779440e-01 -8.49295974e-01 -6.73866928e-01 4.04738039e-01
1.39155742e-02 -9.65712011e-01 2.69969553e-01 9.92162526e-01
-1.19756722e+00 8.98510575e-01 -4.73286092e-01 -4.74352747e-01
-1.35433674e+00 1.00835693e+00 5.20519495e-01 1.36656344e-01
-5.11233807e-01 3.97935301e-01 5.38206935e-01 -3.30482684e-02
-3.37370783e-01 -1.14676720e-02 -3.24544422e-02 -2.40806043e-02
6.55665755e-01 2.26027697e-01 -1.65556222e-01 -8.16493690e-01
-9.27188635e-01 4.46030200e-01 -5.32505512e-01 3.59499991e-01
1.22511077e+00 -1.63223684e-01 -4.69762206e-01 1.75384179e-01
1.12799847e+00 4.71475422e-01 -9.54491556e-01 3.13087195e-01
6.09962903e-02 -5.32021463e-01 -6.90524995e-01 -6.41483545e-01
-2.57006913e-01 1.24759488e-01 6.92364633e-01 -6.04146831e-02
6.80906713e-01 -3.15066725e-02 5.73975682e-01 5.41582644e-01
4.52887058e-01 -8.54403019e-01 -7.27853954e-01 1.39375016e-01
8.42812836e-01 -6.50819659e-01 1.45512313e-01 -4.26504314e-01
-5.14996469e-01 9.07328427e-01 2.41432399e-01 -1.67160966e-02
3.40072632e-01 2.93526590e-01 1.51593298e-01 -3.92434955e-01
-1.04107511e+00 2.23145172e-01 -2.94252932e-01 5.43594837e-01
7.40286231e-01 -2.10549273e-02 -1.21680844e+00 -1.66703209e-01
3.44405293e-01 -1.34759963e-01 5.49995840e-01 1.20569575e+00
-8.95338178e-01 -1.46980679e+00 -6.44481897e-01 1.61949649e-01
-3.68001193e-01 -2.28338167e-01 -6.59691334e-01 1.11202025e+00
4.95362654e-02 6.63852930e-01 1.22536324e-01 2.87965119e-01
-5.49654104e-02 4.81074989e-01 5.76690197e-01 -1.10586561e-01
-3.47318083e-01 7.22299039e-01 2.06676722e-01 -4.15814787e-01
-4.01986450e-01 -8.66129279e-01 -1.77981806e+00 -7.23026276e-01
-2.81810880e-01 7.09901273e-01 5.87856472e-01 8.63481104e-01
5.26989877e-01 -1.47892416e-01 4.52623039e-01 -2.71670789e-01
-3.79742950e-01 -4.28172201e-01 -1.09088421e+00 1.22436834e-02
3.39854509e-01 -3.26730102e-01 -4.31940556e-01 -9.41154361e-02] | [4.739315986633301, 5.139683246612549] |
c39e41be-95c3-412e-89f7-c90dd9c8af83 | generating-diversified-comments-via-reader | 2102.06856 | null | https://arxiv.org/abs/2102.06856v1 | https://arxiv.org/pdf/2102.06856v1.pdf | Generating Diversified Comments via Reader-Aware Topic Modeling and Saliency Detection | Automatic comment generation is a special and challenging task to verify the model ability on news content comprehension and language generation. Comments not only convey salient and interesting information in news articles, but also imply various and different reader characteristics which we treat as the essential clues for diversity. However, most of the comment generation approaches only focus on saliency information extraction, while the reader-aware factors implied by comments are neglected. To address this issue, we propose a unified reader-aware topic modeling and saliency information detection framework to enhance the quality of generated comments. For reader-aware topic modeling, we design a variational generative clustering algorithm for latent semantic learning and topic mining from reader comments. For saliency information detection, we introduce Bernoulli distribution estimating on news content to select saliency information. The obtained topic representations as well as the selected saliency information are incorporated into the decoder to generate diversified and informative comments. Experimental results on three datasets show that our framework outperforms existing baseline methods in terms of both automatic metrics and human evaluation. The potential ethical issues are also discussed in detail. | ['Hai-Tao Zheng', 'Piji Li', 'Wei Wang'] | 2021-02-13 | null | null | null | null | ['comment-generation'] | ['natural-language-processing'] | [ 2.22492754e-01 2.72470415e-01 -3.64425182e-01 -5.27418971e-01
-1.03909218e+00 -2.57684141e-01 7.39754975e-01 3.44425231e-01
-2.12626263e-01 6.67653739e-01 1.10404456e+00 1.22114815e-01
1.56072438e-01 -5.43344676e-01 -5.53507924e-01 -6.75522923e-01
5.21894932e-01 2.69081324e-01 4.69217032e-01 -2.69009411e-01
7.71870613e-01 -6.11832440e-01 -1.69222069e+00 4.83809233e-01
1.55865359e+00 6.75854206e-01 8.30604196e-01 3.55444521e-01
-4.07875180e-01 6.24079823e-01 -7.96875894e-01 -4.73231554e-01
-3.76644552e-01 -6.88636303e-01 -5.10889530e-01 2.83017755e-01
-9.04402509e-02 -2.91580647e-01 3.06635231e-01 1.20552850e+00
6.37942731e-01 1.92040041e-01 8.73866320e-01 -1.03720045e+00
-8.17757785e-01 1.29824960e+00 -6.80348098e-01 3.76560509e-01
3.39755327e-01 -2.89035350e-01 1.20527840e+00 -1.01627970e+00
5.91658473e-01 1.35478425e+00 8.01756326e-03 4.59396005e-01
-6.10985577e-01 -5.77580750e-01 6.08573973e-01 2.29191422e-01
-1.00313795e+00 -4.61506933e-01 1.12196255e+00 -3.41815263e-01
2.62753628e-02 3.06745768e-01 4.15013433e-01 1.32213855e+00
1.85845882e-01 1.49907708e+00 6.68668151e-01 -2.17514113e-01
2.97212869e-01 7.37796962e-01 1.50392547e-01 1.72482282e-01
2.09366396e-01 -6.73472345e-01 -8.25111926e-01 -1.72706708e-01
4.17632192e-01 1.36599854e-01 -3.34707707e-01 -1.19088896e-01
-1.53149939e+00 1.32958078e+00 2.24322811e-01 -3.78831737e-02
-5.05525768e-01 -2.64390498e-01 4.15939182e-01 -1.88037559e-01
9.17720079e-01 3.52610618e-01 -9.67101306e-02 -4.20453064e-02
-1.06076980e+00 5.66238940e-01 5.72422445e-01 1.17204332e+00
6.59814417e-01 -1.60041936e-02 -7.81981826e-01 8.26123238e-01
6.97519422e-01 6.46454751e-01 7.40846872e-01 -4.45968986e-01
6.15493000e-01 6.05521560e-01 3.96279186e-01 -1.38198602e+00
-1.08887158e-01 -8.26553106e-01 -6.30826175e-01 -4.77899939e-01
-2.53535360e-01 -4.74937975e-01 -5.14382541e-01 1.42891443e+00
2.21101284e-01 -1.02298856e-02 1.12215884e-01 1.18100405e+00
1.29907227e+00 9.52712238e-01 1.98843092e-01 -3.83535355e-01
1.62932706e+00 -1.08595526e+00 -1.01524472e+00 -2.51210749e-01
2.45044217e-01 -1.02827144e+00 1.11559141e+00 9.14770067e-02
-1.18104756e+00 -3.85266602e-01 -8.14534009e-01 -1.68088563e-02
2.03470767e-01 5.38732648e-01 4.83465821e-01 3.21259797e-01
-6.54333115e-01 -1.84791058e-01 -3.24440688e-01 -1.01990201e-01
3.90619218e-01 -2.55114913e-01 4.65560466e-01 3.44654053e-01
-1.62930858e+00 3.07416826e-01 3.69234174e-01 -3.50887477e-01
-8.33959281e-01 -6.60601735e-01 -8.14834237e-01 1.34059757e-01
5.26322186e-01 -7.13280082e-01 1.49895155e+00 -1.04973376e+00
-1.30142486e+00 4.66731757e-01 -6.78544164e-01 -5.63428521e-01
4.72273171e-01 -4.44562197e-01 -2.64671028e-01 2.56771088e-01
6.80624783e-01 8.02662671e-01 1.00331581e+00 -1.35374284e+00
-9.90202785e-01 1.59526736e-01 -1.99900448e-01 8.06478322e-01
-3.22174430e-01 2.95201868e-01 -2.41137788e-01 -1.03945172e+00
1.55002058e-01 -4.57655549e-01 -2.10723251e-01 -4.63177353e-01
-7.80445755e-01 -3.87523204e-01 7.46361971e-01 -6.22656524e-01
1.30570960e+00 -2.09965301e+00 -1.71112511e-02 -3.68247516e-02
3.39687854e-01 -1.76112622e-01 3.23117405e-01 3.36451918e-01
4.43757772e-01 7.14095533e-02 8.06133170e-03 -3.31905603e-01
5.33770844e-02 -4.88372475e-01 -8.27444732e-01 -2.28510108e-02
-1.60928685e-02 9.15044844e-01 -9.91236210e-01 -7.78660655e-01
-2.61804819e-01 1.51359826e-01 -5.90326965e-01 2.23370060e-01
-6.49951279e-01 4.73007143e-01 -1.09887838e+00 2.81534106e-01
6.12310290e-01 -5.09988666e-01 -5.03272533e-01 4.15506326e-02
-1.62305191e-01 6.51182353e-01 -1.05978107e+00 1.56010234e+00
-1.46402583e-01 4.58539695e-01 -2.82459319e-01 -6.27158642e-01
1.13088071e+00 3.17166299e-01 9.17536542e-02 -4.37536985e-01
2.57073402e-01 3.12457699e-02 -2.95255333e-01 -3.84781420e-01
1.14725316e+00 -1.70867126e-02 -3.66982847e-01 9.99326587e-01
-2.62487352e-01 9.16022435e-02 2.91097406e-02 7.28861988e-01
2.01073930e-01 -2.23871142e-01 1.48026466e-01 -5.05655527e-01
3.47870231e-01 -6.48744553e-02 4.99238789e-01 7.05175459e-01
-1.04747809e-01 8.00282776e-01 5.70518374e-01 2.04243273e-01
-8.39852214e-01 -9.74391639e-01 1.53148711e-01 1.43841553e+00
7.38074780e-01 -3.95401508e-01 -8.18657279e-01 -5.64768136e-01
-4.78353649e-01 1.25988483e+00 -6.40440166e-01 -1.33468777e-01
-8.55146870e-02 -8.76198232e-01 -1.29318729e-01 2.35470474e-01
4.91504461e-01 -1.15781963e+00 -4.94463712e-01 2.25118056e-01
-8.44962656e-01 -7.05579937e-01 -7.87943542e-01 -4.31463063e-01
-5.58387876e-01 -5.94421685e-01 -1.17243576e+00 -9.25825357e-01
6.73236191e-01 8.42963815e-01 9.44286883e-01 -2.08191633e-01
2.77492017e-01 2.02574119e-01 -8.49403203e-01 -9.75777686e-01
-3.57443750e-01 2.45287120e-01 -1.99379355e-01 2.44172528e-01
4.17073011e-01 -1.72839314e-01 -8.59132051e-01 3.11605692e-01
-8.22066605e-01 5.41925848e-01 5.96550465e-01 6.64890587e-01
3.04355264e-01 -1.67087223e-02 1.14481854e+00 -9.60294127e-01
1.24473464e+00 -9.44294930e-01 -8.61288011e-02 8.90771151e-02
-4.90499765e-01 -1.17427073e-02 2.50425965e-01 -3.02691907e-01
-1.70436585e+00 -5.74935198e-01 1.44175857e-01 1.63447410e-01
-4.25485447e-02 6.77859664e-01 -3.28044385e-01 8.16428602e-01
6.05505824e-01 6.32285833e-01 -2.01820374e-01 -2.61472940e-01
3.80493164e-01 7.47837186e-01 7.56736994e-02 -2.99560517e-01
5.01034796e-01 4.02291417e-01 -7.86991835e-01 -6.32325113e-01
-1.35987759e+00 -6.13652289e-01 -1.82226568e-01 -3.60048205e-01
9.32524920e-01 -1.26334012e+00 -7.69944787e-02 3.72145474e-01
-1.34016788e+00 4.07249272e-01 -1.18603982e-01 5.11233926e-01
-3.81107599e-01 2.97970682e-01 -3.83420616e-01 -9.66881931e-01
-5.25219440e-01 -1.24450827e+00 1.34776711e+00 5.68861783e-01
-2.39289492e-01 -1.08173370e+00 -2.79364169e-01 4.45538819e-01
3.85567993e-01 -6.58995956e-02 6.64188445e-01 -9.65137124e-01
-6.64897442e-01 -2.06228648e-03 -3.77432287e-01 -9.11516324e-02
9.40955803e-02 -2.40283489e-01 -8.06816280e-01 9.99506339e-02
2.43404225e-01 -1.80284753e-01 1.02437818e+00 6.37845397e-01
8.74013484e-01 -6.76917851e-01 -4.01796013e-01 1.43751595e-02
7.76337087e-01 -4.84225452e-02 4.68728274e-01 3.02900165e-01
5.85795581e-01 7.94915497e-01 1.01614070e+00 8.77374887e-01
8.06561112e-01 2.97173351e-01 2.23578706e-01 2.52044853e-02
2.17520058e-01 -5.64175129e-01 4.66596752e-01 1.09439111e+00
3.18567902e-01 -5.80324292e-01 -5.15571833e-01 7.65528798e-01
-1.91724086e+00 -1.15791154e+00 -4.01149541e-01 1.72596991e+00
1.04007113e+00 2.83156216e-01 1.57529160e-01 -1.57595187e-01
1.25517893e+00 3.77721936e-01 -4.18257505e-01 1.93770006e-02
-2.77645528e-01 -5.66443980e-01 -1.15957260e-02 2.98364818e-01
-1.06665349e+00 1.07788849e+00 5.72779560e+00 1.08554220e+00
-7.98804045e-01 1.74696013e-01 8.94271851e-01 -3.20349559e-02
-1.14674580e+00 -8.22700262e-02 -1.13971162e+00 8.59952867e-01
4.62152064e-01 -7.95868933e-01 -4.11655128e-01 1.11550605e+00
6.61623597e-01 -2.28239402e-01 -5.15280187e-01 7.39942074e-01
6.03646994e-01 -1.39669371e+00 5.47307372e-01 -2.55539387e-01
1.06977165e+00 -5.32584071e-01 4.40310687e-01 1.06330268e-01
3.03869873e-01 -5.89105546e-01 9.74622309e-01 5.84385216e-01
1.25593796e-01 -9.17555034e-01 9.02079582e-01 6.17488086e-01
-7.44779766e-01 1.89885706e-01 -5.67180514e-01 -4.67415079e-02
5.98140061e-01 1.01343918e+00 -1.01140738e+00 3.71234298e-01
3.93460274e-01 9.75594282e-01 -5.34758687e-01 1.08706689e+00
-4.88444835e-01 9.49957609e-01 1.97012424e-01 -6.57270610e-01
2.65750170e-01 -3.38944644e-02 8.83196235e-01 1.17545164e+00
4.57634896e-01 -1.46039262e-01 1.23865046e-01 1.12562895e+00
-1.73348503e-03 6.52120650e-01 8.67232960e-03 2.12093174e-01
6.01318538e-01 1.15549147e+00 -9.27430093e-01 -4.20233279e-01
-2.25444570e-01 8.91401291e-01 -6.80580661e-02 3.38948727e-01
-9.48049903e-01 -3.59523237e-01 2.57891983e-01 1.55917004e-01
2.60709077e-01 2.43304938e-01 -5.28016329e-01 -1.35097599e+00
-3.01471800e-02 -6.59863055e-01 1.88265756e-01 -8.41675699e-01
-1.23282206e+00 4.11053002e-01 1.43642575e-01 -1.36980009e+00
-2.42360711e-01 2.10687280e-01 -1.13606286e+00 9.37942922e-01
-1.58799243e+00 -9.54012990e-01 -3.57834309e-01 2.54579574e-01
1.19783676e+00 -3.45726013e-01 2.06157550e-01 -9.75506902e-02
-3.02417308e-01 5.19196749e-01 -1.75507814e-02 -2.45183080e-01
7.77525663e-01 -1.10331881e+00 4.48527724e-01 9.83167291e-01
-1.37738213e-01 7.06051648e-01 1.01304793e+00 -9.00387824e-01
-6.82277143e-01 -1.19263172e+00 1.07165420e+00 -3.03567320e-01
4.40613121e-01 -3.63815278e-01 -8.61331642e-01 4.07508969e-01
5.23095131e-01 -1.11555684e+00 9.70641792e-01 -2.58763004e-02
1.21559680e-01 3.09184462e-01 -6.75104737e-01 9.59261715e-01
6.63155437e-01 -1.18613251e-01 -1.05282640e+00 3.42749745e-01
1.32182157e+00 -3.20023924e-01 2.25751358e-03 -1.21556837e-02
8.31384063e-02 -9.17197466e-01 6.89991534e-01 -2.08687663e-01
8.90712619e-01 -2.71554232e-01 2.34867319e-01 -1.40917432e+00
-3.67459208e-01 -7.13094056e-01 1.27713442e-01 1.63945544e+00
6.98675275e-01 -2.49539375e-01 6.63329482e-01 3.73902828e-01
-2.99133450e-01 -4.72403407e-01 -3.89487058e-01 3.11773662e-02
-2.60925680e-01 -2.19355896e-01 5.60053289e-01 6.60836637e-01
1.76028639e-01 7.27012873e-01 -6.46955132e-01 -7.34999925e-02
5.03566027e-01 4.52863038e-01 7.23340511e-01 -1.18250489e+00
1.05042517e-01 -5.59440970e-01 8.47874209e-02 -1.52555668e+00
6.92210719e-02 -6.77221298e-01 4.29736227e-01 -1.90520310e+00
7.49175251e-01 -2.24808708e-01 9.40123014e-03 -2.02439189e-01
-8.96133482e-01 -3.04394543e-01 -1.09214798e-01 5.06455481e-01
-1.16540635e+00 1.03222203e+00 1.58018935e+00 3.47926803e-02
-1.86153755e-01 3.72671455e-01 -1.54120648e+00 7.52678156e-01
7.55812407e-01 -3.70190561e-01 -8.55545819e-01 -2.94984341e-01
4.86767262e-01 -1.74182072e-01 2.72537589e-01 -7.32979357e-01
2.79578626e-01 -3.70739400e-01 1.73376098e-01 -9.36569870e-01
3.50177698e-02 -1.22953713e-01 -5.13022363e-01 3.57893333e-02
-9.44809854e-01 -2.28704959e-01 -2.68779993e-01 9.11925972e-01
-3.81129920e-01 -4.05140281e-01 3.70118856e-01 -1.07068293e-01
-6.44877076e-01 2.15377703e-01 -5.95226109e-01 4.19833213e-01
9.49983895e-01 -2.42087580e-02 -3.98278117e-01 -1.12557280e+00
-4.01420921e-01 4.89663661e-01 2.64286309e-01 7.44026959e-01
7.49954939e-01 -1.28404069e+00 -1.21709383e+00 8.71227905e-02
2.75091797e-01 1.56838194e-01 5.80061972e-01 6.08090937e-01
-7.86053762e-03 3.26614767e-01 1.63007990e-01 -5.12270451e-01
-9.88060057e-01 4.62585896e-01 -3.50489944e-01 3.09987292e-02
-4.07688886e-01 1.09768403e+00 8.16870928e-01 1.67059287e-01
1.37377307e-01 -1.97528675e-01 -8.92280996e-01 4.67293948e-01
8.63215387e-01 1.61063999e-01 -4.90135670e-01 -7.53896475e-01
5.01463450e-02 9.51188579e-02 -3.68309945e-01 -2.12702230e-01
9.53100264e-01 -8.23260427e-01 6.72908127e-02 5.61022282e-01
7.60987759e-01 3.01724792e-01 -1.07275367e+00 -4.00383681e-01
2.41209771e-02 -3.24222922e-01 -9.41418670e-03 -5.86594701e-01
-7.18079269e-01 9.63571966e-01 -7.00747371e-02 3.89582515e-01
7.79405713e-01 3.51265818e-01 1.09749198e+00 -8.00260156e-02
5.15938550e-02 -1.22830868e+00 4.02599931e-01 4.90925103e-01
1.09323323e+00 -1.41073954e+00 6.24309247e-03 -6.94184899e-01
-1.47404981e+00 7.21470296e-01 7.13049293e-01 -7.25140050e-02
5.59444129e-01 -3.11113566e-01 3.02316755e-01 -1.27846748e-01
-8.40868652e-01 -6.98639601e-02 6.62812173e-01 4.05917346e-01
5.80299556e-01 -7.44686648e-02 -6.61190271e-01 1.30984926e+00
-5.90152919e-01 -4.64802563e-01 7.32291341e-01 5.70713222e-01
-1.10798609e+00 -5.92284262e-01 -2.66336858e-01 5.96622288e-01
-5.72542846e-01 -3.31383288e-01 -2.67530292e-01 -6.32362813e-02
-2.22026154e-01 1.17585647e+00 4.16374467e-02 -1.11465089e-01
-2.36237630e-01 -1.39802799e-01 -4.55589235e-01 -9.17278647e-01
-2.76953727e-01 4.67557698e-01 2.36726291e-02 4.36781794e-02
-4.87209767e-01 -9.05883729e-01 -1.19880772e+00 4.10254076e-02
-6.60455704e-01 8.38711798e-01 5.89802146e-01 9.84035671e-01
6.07455254e-01 7.19820261e-01 7.21590102e-01 -5.94346464e-01
-4.29168254e-01 -1.20489824e+00 -5.56554377e-01 3.77525210e-01
1.71336040e-01 -5.89507043e-01 -5.07146895e-01 2.46286690e-01] | [12.398802757263184, 9.297928810119629] |
42ff0a51-9c36-4cbe-8cfc-8745e0523098 | strategic-features-for-general-games | 2101.00843 | null | https://arxiv.org/abs/2101.00843v1 | https://arxiv.org/pdf/2101.00843v1.pdf | Strategic Features for General Games | This short paper describes an ongoing research project that requires the automated self-play learning and evaluation of a large number of board games in digital form. We describe the approach we are taking to determine relevant features, for biasing MCTS playouts for arbitrary games played on arbitrary geometries. Benefits of our approach include efficient implementation, the potential to transfer learnt knowledge to new contexts, and the potential to explain strategic knowledge embedded in features in human-comprehensible terms. | ['Eric Piette', 'Dennis J. N. J. Soemers', 'Cameron Browne'] | 2021-01-04 | null | null | null | null | ['board-games'] | ['playing-games'] | [-2.72213649e-02 2.20004752e-01 5.96283861e-02 -2.06195936e-01
-7.29118526e-01 -6.65835261e-01 3.86880845e-01 -1.64070934e-01
-4.15849239e-01 1.09445536e+00 4.21032868e-02 -3.65764856e-01
-6.75352514e-01 -1.03451014e+00 -7.05371618e-01 -3.68661344e-01
-5.34711242e-01 5.87810934e-01 8.09200048e-01 -8.07859480e-01
5.09468257e-01 5.71760118e-01 -1.61807072e+00 2.84312844e-01
2.57786930e-01 8.63005757e-01 3.64801496e-01 1.13716257e+00
3.95448238e-01 1.28133655e+00 -4.87955838e-01 -6.00798249e-01
6.63218558e-01 -7.28545785e-02 -1.04185128e+00 -5.28891198e-02
4.05907445e-02 -3.73221666e-01 -5.88108301e-01 4.30731863e-01
3.62604678e-01 4.16957438e-01 6.02482438e-01 -8.63497734e-01
1.25973493e-01 4.95189548e-01 3.42989564e-02 5.61929584e-01
5.98315120e-01 4.61981475e-01 1.23227072e+00 -3.63779277e-01
6.47335649e-01 8.38351309e-01 8.78762901e-01 3.39632183e-01
-6.53030455e-01 -6.53652906e-01 -8.13190192e-02 3.80708218e-01
-1.24887800e+00 -7.59686530e-02 6.81885242e-01 -3.92321497e-01
1.35060966e+00 4.81605828e-01 1.32757735e+00 6.43941402e-01
5.27710795e-01 6.79303586e-01 1.11533391e+00 -5.05776763e-01
5.24309576e-01 -1.64777741e-01 -4.01702851e-01 7.70751536e-01
1.06322311e-01 8.30226183e-01 -7.13723063e-01 8.56112391e-02
1.32005453e+00 -6.28652632e-01 4.79785889e-01 -5.85964620e-01
-4.25952822e-01 8.17091048e-01 3.86986196e-01 -6.20586351e-02
-1.93621740e-01 5.84765196e-01 3.31206411e-01 4.35293168e-01
3.42719436e-01 1.35436547e+00 -8.79209161e-01 -8.29060197e-01
-7.42407620e-01 9.77403700e-01 9.65501368e-01 1.00630105e+00
8.29745650e-01 -1.02598844e-02 4.24399525e-01 1.35211259e-01
-5.29712392e-03 6.67518526e-02 3.08799833e-01 -1.35085344e+00
3.04811180e-01 2.27157921e-01 1.60862535e-01 -9.38854218e-01
-7.30779231e-01 -3.94690573e-01 1.03365332e-01 6.48888469e-01
1.39056206e-01 -4.88300741e-01 -4.16361451e-01 1.09349620e+00
3.20859253e-01 3.47310565e-02 6.64355680e-02 4.92335469e-01
7.80185521e-01 2.86672354e-01 -9.55816656e-02 2.91588932e-01
8.33013654e-01 -7.90968716e-01 5.92174428e-03 -2.37801045e-01
7.98646331e-01 -5.74957550e-01 7.78694212e-01 6.70613706e-01
-1.21719408e+00 -5.27283192e-01 -1.19086325e+00 1.62152365e-01
-4.56061751e-01 -4.30800736e-01 1.38474452e+00 8.91205013e-01
-1.08755934e+00 1.05228949e+00 -7.69265175e-01 -1.72287360e-01
6.08195066e-01 1.05715239e+00 -2.02167273e-01 1.97780758e-01
-1.20494735e+00 1.13687861e+00 6.52695954e-01 -3.69504720e-01
-1.02700210e+00 -7.11658239e-01 -8.54970098e-01 -2.07816184e-01
6.10703826e-01 -5.60027122e-01 1.64659441e+00 -9.64122176e-01
-1.86076164e+00 6.88975096e-01 6.16415322e-01 -3.44834358e-01
5.93452096e-01 -2.04597190e-01 -4.24027517e-02 1.10009298e-01
5.78048788e-02 5.59485316e-01 4.85725164e-01 -9.31160152e-01
-9.95365977e-01 -1.59990937e-01 8.87849987e-01 7.50073075e-01
9.88281742e-02 -1.98434934e-01 -2.36014947e-01 -5.30302405e-01
7.01382682e-02 -8.22344899e-01 -7.32816100e-01 -4.53241140e-01
1.97443701e-02 8.40472504e-02 2.44636133e-01 -4.17306811e-01
9.13804591e-01 -1.70340955e+00 1.76246792e-01 3.72051001e-01
3.75303715e-01 -6.11523278e-02 1.29264966e-01 5.12814283e-01
1.07544571e-01 -3.22918147e-02 5.58509588e-01 4.49243665e-01
-1.62621677e-01 3.01150411e-01 6.97371783e-04 2.73670815e-02
-1.33723505e-02 1.11534238e+00 -1.08906019e+00 -4.71938431e-01
6.08542740e-01 -3.36394280e-01 -9.82111156e-01 5.92408478e-02
-3.13306361e-01 3.57632101e-01 -8.31489325e-01 3.73445064e-01
1.35670587e-01 2.68130094e-01 5.36670983e-02 5.38891554e-01
-4.04802680e-01 4.94860172e-01 -1.30066264e+00 1.49787903e+00
-7.05128610e-01 6.15663052e-01 -2.44468451e-01 -8.15194309e-01
8.20741951e-01 -1.53744817e-01 5.17356157e-01 -5.95104456e-01
2.83190370e-01 1.39472827e-01 1.11955673e-01 -6.57355309e-01
9.08835053e-01 -5.51006138e-01 -4.42542911e-01 2.29264587e-01
1.78081006e-01 -8.70782912e-01 1.16451487e-01 -9.03310031e-02
1.33850873e+00 4.21361953e-01 6.52299583e-01 -3.84578854e-01
-4.57920618e-02 5.03297627e-01 -1.51674658e-01 1.06819284e+00
1.31047651e-01 5.38609982e-01 3.74101162e-01 -9.67687905e-01
-9.65246141e-01 -9.29033339e-01 2.22454742e-01 1.40943265e+00
3.81285071e-01 -7.16002703e-01 -5.01566350e-01 -3.90866071e-01
-2.04950958e-01 5.22824168e-01 -7.94872105e-01 -9.41452906e-02
-5.32637060e-01 -1.96845934e-01 3.81220520e-01 9.58025157e-01
2.55867660e-01 -1.18475747e+00 -9.47930336e-01 4.45184588e-01
4.45853710e-01 -7.82826424e-01 8.94357637e-02 5.45006752e-01
-1.00592494e+00 -1.16850400e+00 1.35744400e-02 -8.10721219e-01
1.18971184e-01 1.13124847e-01 1.16734529e+00 1.71204716e-01
-5.55745006e-01 5.54213881e-01 -4.46847320e-01 -4.52708185e-01
-1.25007570e-01 4.32242364e-01 -1.39125273e-01 -9.05022144e-01
1.70125589e-01 -7.93141723e-01 -1.99687853e-01 3.65705341e-01
-6.16793811e-01 2.88540572e-01 5.14331520e-01 6.64722621e-01
3.55511695e-01 4.06842589e-01 -1.28412798e-01 -9.59266424e-01
5.80736041e-01 -2.25307748e-01 -7.99893558e-01 -5.72147220e-02
-1.50413305e-01 9.54407454e-02 5.14120221e-01 -1.85606331e-01
-9.63347316e-01 1.51768446e-01 -3.55529547e-01 3.20660621e-01
-1.06370986e-01 4.41057414e-01 -2.67469227e-01 -6.97541118e-01
9.49075401e-01 3.96151794e-03 -2.90335238e-01 -3.76773812e-02
4.93893415e-01 2.85751939e-01 2.77804971e-01 -1.11491513e+00
7.55738139e-01 2.78077483e-01 2.88760185e-01 -9.16935027e-01
-5.85654259e-01 -3.51449430e-01 -9.03585076e-01 -3.63031298e-01
5.68375528e-01 -9.52739000e-01 -7.51277983e-01 3.99413615e-01
-6.62041605e-01 -6.66224718e-01 -7.98629463e-01 3.66031885e-01
-1.32356906e+00 -1.13981858e-01 -3.09913546e-01 -5.74767411e-01
2.33899847e-01 -8.89222324e-01 9.28966522e-01 4.32437241e-01
-5.98483324e-01 -1.27313149e+00 3.58902544e-01 3.30361873e-01
2.27957278e-01 8.04810748e-02 6.98651254e-01 -6.15251541e-01
-8.09969962e-01 -4.07611251e-01 1.38407528e-01 -6.06887937e-02
-9.32368934e-02 -8.42692629e-02 -7.61000037e-01 -3.17397080e-02
-3.47889036e-01 -7.74902523e-01 3.83572847e-01 4.82013553e-01
9.87866640e-01 1.65117159e-02 -2.02801481e-01 6.19622946e-01
1.32776117e+00 2.23095223e-01 7.21708834e-01 8.54720592e-01
5.27486503e-01 4.12875563e-01 8.06467772e-01 6.12635732e-01
4.21569735e-01 6.52770698e-01 3.20789248e-01 2.51413807e-02
9.91198942e-02 -5.16483486e-01 -5.69164520e-03 5.16480803e-01
-5.60079753e-01 1.01329470e-02 -9.51112151e-01 3.50154191e-01
-1.71036470e+00 -1.00156236e+00 1.66199878e-01 1.66454053e+00
4.19588119e-01 7.70019829e-01 3.59055549e-01 -1.31060567e-03
3.21866006e-01 -1.92654338e-02 -3.85733902e-01 -5.80888808e-01
1.80844009e-01 1.02381647e+00 7.60592103e-01 8.26596439e-01
-1.10874891e+00 1.49025631e+00 8.38830948e+00 1.05833042e+00
-6.89806044e-01 -5.93499616e-02 4.34567988e-01 -2.70002663e-01
-1.94319516e-01 3.62829119e-01 -5.38756490e-01 -2.84697682e-01
6.50766969e-01 -2.73425072e-01 5.93496025e-01 1.02230823e+00
4.29084972e-02 -4.11811620e-01 -8.99761856e-01 6.20770752e-01
-1.92717895e-01 -1.69468641e+00 -2.44461015e-01 1.18655145e-01
8.25948119e-01 -1.44333035e-01 1.68212011e-01 5.80933869e-01
8.89337480e-01 -1.35520017e+00 9.08419073e-01 3.27334106e-01
6.59494579e-01 -9.39190388e-01 5.57670653e-01 1.81625366e-01
-1.11573577e+00 -1.50522217e-01 -4.84028280e-01 -1.22493100e+00
-4.03720707e-01 -3.05064797e-01 -9.94497240e-01 4.80138421e-01
4.65205133e-01 6.82230771e-01 -7.12486565e-01 1.15066063e+00
-3.47103328e-01 6.27113879e-01 -4.89711106e-01 -7.01796830e-01
4.78031725e-01 2.64487326e-01 3.57815802e-01 7.35904098e-01
2.30030920e-02 4.41573113e-01 2.91047007e-01 2.36092746e-01
5.22133112e-01 1.63225338e-01 -7.81741619e-01 3.28192562e-01
6.84373006e-02 9.83299315e-01 -9.60575044e-01 1.45622820e-01
-1.31604359e-01 6.50818944e-01 3.35866004e-01 -6.87610880e-02
-6.26766264e-01 -3.25900227e-01 6.82776272e-01 4.84140426e-01
6.22866511e-01 -2.03466743e-01 -6.36464298e-01 -6.94590211e-01
-8.41654241e-02 -7.87057698e-01 2.56968349e-01 -9.83031571e-01
-1.00724733e+00 4.45780724e-01 2.83178449e-01 -1.15300071e+00
-4.63692069e-01 -1.02899468e+00 -6.40406489e-01 4.47528034e-01
-9.29548204e-01 -1.20251298e+00 1.38176745e-02 4.20398384e-01
6.01255000e-01 -5.78814149e-01 7.82432497e-01 -2.61318177e-01
1.23702019e-01 3.54899466e-01 1.67469040e-01 -1.30110890e-01
-9.26723257e-02 -1.33786416e+00 6.68307185e-01 1.29034147e-01
2.69227445e-01 1.78059973e-02 9.90220070e-01 -5.85604370e-01
-1.04121506e+00 -5.60469866e-01 2.39189550e-01 -9.55100179e-01
9.73653436e-01 -3.34585816e-01 -9.38414484e-02 6.28717124e-01
-1.43757388e-01 -3.85659158e-01 6.91994011e-01 6.33464754e-01
8.19597468e-02 -7.10874200e-02 -9.11347091e-01 7.44882464e-01
1.25014555e+00 -2.73431808e-01 -6.58114493e-01 1.53784722e-01
3.53141218e-01 -9.73687053e-01 -5.30435443e-01 1.47252575e-01
7.53870606e-01 -1.04193604e+00 9.56906497e-01 -1.04613590e+00
7.34664202e-01 1.04659639e-01 -1.19513549e-01 -1.25364137e+00
-5.48182130e-01 -1.02375937e+00 3.27763200e-01 4.79611099e-01
4.97388393e-01 -1.69453472e-01 1.45734394e+00 6.68958783e-01
-5.42665459e-02 -9.10441041e-01 -1.18793631e+00 -6.37248635e-01
5.27829945e-01 -9.72756565e-01 5.70158839e-01 3.76851350e-01
4.53060567e-01 1.71995491e-01 -5.01379073e-01 2.73816437e-02
1.36264533e-01 -1.14990957e-01 9.41517174e-01 -1.08482993e+00
-8.14528108e-01 -4.18256640e-01 -1.22038853e+00 -1.09635663e+00
-4.29799855e-02 -5.76278746e-01 4.64502946e-02 -1.22203040e+00
5.10738455e-02 -4.40712959e-01 -1.65658090e-02 1.41316503e-01
4.77258302e-02 1.35962293e-01 9.52238441e-02 -1.26965791e-01
-1.16800094e+00 2.52256572e-01 1.45447075e+00 1.44477129e-01
-1.59126863e-01 2.31036365e-01 -8.39693248e-01 1.03660107e+00
8.29420924e-01 -6.01299226e-01 -5.67643762e-01 -2.00104758e-01
6.91311955e-01 5.70414104e-02 3.67892414e-01 -1.50798929e+00
2.51582056e-01 -5.18835306e-01 4.92823124e-01 -1.03439756e-01
4.10900772e-01 -5.67602932e-01 8.64929780e-02 4.73310173e-01
-3.52438748e-01 9.28149149e-02 5.79147637e-01 3.15202206e-01
5.68405986e-02 -6.11960709e-01 1.91483155e-01 -7.05647230e-01
-1.15947175e+00 -3.71887870e-02 -7.48160005e-01 4.52259928e-01
1.28433216e+00 -6.48798823e-01 3.13829333e-01 -7.47397900e-01
-8.58492076e-01 4.32070382e-02 5.22316813e-01 2.50212073e-01
5.32890499e-01 -1.00089431e+00 -5.35432994e-01 1.73913151e-01
-1.13092907e-01 -2.68965393e-01 2.35500053e-01 -6.65780064e-03
-1.34841096e+00 3.21645200e-01 -5.41032434e-01 -1.48486912e-01
-1.03853023e+00 1.59732610e-01 4.93127376e-01 -6.04672134e-01
-5.24911821e-01 1.19022954e+00 -1.25251472e-01 -4.74829823e-01
-1.77332759e-01 -1.24180771e-01 -9.33038723e-03 -4.76830751e-01
3.27653080e-01 2.89395720e-01 1.77418202e-01 -4.25420880e-01
5.85835986e-03 4.92166162e-01 -2.45219544e-02 -4.55679476e-01
1.55148232e+00 2.78873742e-01 4.76742208e-01 3.73178810e-01
5.39650798e-01 -1.80205647e-02 -1.61584640e+00 2.59904623e-01
-3.48542184e-02 -5.76725721e-01 7.89548233e-02 -8.29370677e-01
-6.68050706e-01 4.62927103e-01 3.62311333e-01 1.03426181e-01
7.13079154e-01 1.74714804e-01 4.20393974e-01 8.13089550e-01
9.78982210e-01 -1.30188632e+00 2.97311991e-01 8.29246283e-01
5.92919052e-01 -8.32992077e-01 4.44008440e-01 -5.30446053e-01
-7.04976320e-01 1.31947184e+00 8.54157627e-01 -5.97259581e-01
7.84060478e-01 6.24646187e-01 1.35627136e-01 -5.48184097e-01
-8.15017521e-01 -3.96990120e-01 1.29503414e-01 1.08323264e+00
3.47051173e-02 1.21087968e-01 3.49688269e-02 7.76866198e-01
-1.09295321e+00 1.42834673e-03 7.58475304e-01 1.17413306e+00
-5.49153328e-01 -1.04679918e+00 -1.33513242e-01 4.96671110e-01
-1.18941568e-01 -1.25239134e-01 -4.77919072e-01 1.31910253e+00
4.14763480e-01 6.54446185e-01 -7.47103766e-02 -7.31913626e-01
5.06010890e-01 -3.53566885e-01 1.16918504e+00 -1.02842212e+00
-7.53496885e-01 -3.86890411e-01 2.98219383e-01 -5.21989107e-01
-1.40229911e-01 -7.74997532e-01 -1.11618459e+00 -1.90378144e-01
-2.84492791e-01 1.99359000e-01 2.81018108e-01 1.30292892e+00
-1.04295067e-01 5.67317963e-01 4.27316695e-01 -1.02432287e+00
-3.41498107e-01 -4.65555161e-01 -8.32207382e-01 1.76350996e-01
-1.39207333e-01 -9.82713282e-01 2.93826330e-02 -3.07141244e-01] | [3.5009605884552, 1.4293394088745117] |
74769767-9618-4f68-83a3-00c7a1329e32 | make-the-blind-translator-see-the-world-a | null | null | https://aclanthology.org/2021.mtsummit-research.12 | https://aclanthology.org/2021.mtsummit-research.12.pdf | Make the Blind Translator See The World: A Novel Transfer Learning Solution for Multimodal Machine Translation | Based on large-scale pretrained networks and the liability to be easily overfitting with limited labelled training data of multimodal translation (MMT) is a critical issue in MMT. To this end and we propose a transfer learning solution. Specifically and 1) A vanilla Transformer is pre-trained on massive bilingual text-only corpus to obtain prior knowledge; 2) A multimodal Transformer named VLTransformer is proposed with several components incorporated visual contexts; and 3) The parameters of VLTransformer are initialized with the pre-trained vanilla Transformer and then being fine-tuned on MMT tasks with a newly proposed method named cross-modal masking which forces the model to learn from both modalities. We evaluated on the Multi30k en-de and en-fr dataset and improving up to 8% BLEU score compared with the SOTA performance. The experimental result demonstrates that performing transfer learning with monomodal pre-trained NMT model on multimodal NMT tasks can obtain considerable boosts. | ['Hao Yang', 'Shimin Tao', 'Min Zhang', 'Chang Su', 'Yimeng Chen', 'Jiaxin Guo', 'Minghan Wang'] | null | null | null | null | mtsummit-2021-8 | ['multimodal-machine-translation'] | ['natural-language-processing'] | [ 5.22445679e-01 3.66897732e-01 -1.69635400e-01 -2.90789604e-01
-1.37695408e+00 -6.34150863e-01 7.75513291e-01 -6.33386254e-01
-6.62913740e-01 7.21835554e-01 2.61564463e-01 -6.48179233e-01
3.71233672e-01 -2.60385215e-01 -1.22858071e+00 -7.01021850e-01
3.45040023e-01 9.23271835e-01 -2.65228212e-01 -2.55175084e-01
-2.88436025e-01 -3.84701602e-02 -8.63869905e-01 8.34709942e-01
1.13882291e+00 1.02633345e+00 4.50099766e-01 4.88897890e-01
-2.20235646e-01 5.39085150e-01 -3.63811612e-01 -7.54874289e-01
2.58510292e-01 -5.16923964e-01 -8.87682915e-01 1.80555582e-01
5.67928433e-01 -2.46219724e-01 -1.43712774e-01 8.01063478e-01
6.73860133e-01 2.31004599e-03 7.54222274e-01 -9.69720960e-01
-8.24449539e-01 9.75942612e-01 -6.34819627e-01 -2.94427097e-01
-1.83803998e-02 4.14984405e-01 9.24432099e-01 -1.24085855e+00
7.82706499e-01 1.32755661e+00 5.85774660e-01 7.66912818e-01
-1.40143311e+00 -6.14143312e-01 -1.39322430e-01 1.91472337e-01
-1.12070882e+00 -5.81693411e-01 5.21947861e-01 -7.64381662e-02
1.12752175e+00 1.25521347e-01 3.48125368e-01 1.62022948e+00
1.45439088e-01 8.30872953e-01 1.37022102e+00 -6.06436431e-01
1.16961598e-02 5.03904045e-01 -6.04982674e-01 6.72524095e-01
-5.91784120e-01 -2.83570811e-02 -4.89481419e-01 1.57513469e-01
6.00780904e-01 -4.21599001e-01 -1.20165125e-01 -4.95039523e-01
-1.60408211e+00 8.11663747e-01 4.81199205e-01 2.82152414e-01
-3.07555825e-01 -9.43322629e-02 5.30266583e-01 6.50775790e-01
3.47285420e-01 2.00402617e-01 -6.98095322e-01 1.35236904e-01
-9.09805954e-01 -4.63898003e-01 3.87974679e-01 1.08033574e+00
8.43882322e-01 4.14899625e-02 -2.95074373e-01 1.22947204e+00
1.49671614e-01 9.60771382e-01 6.42319739e-01 -5.53685129e-01
1.20658982e+00 5.70056796e-01 -3.42316180e-01 -2.40706757e-01
-1.28883839e-01 -4.58284527e-01 -8.00071776e-01 -5.73966913e-02
3.64371002e-01 -5.00570297e-01 -1.41613948e+00 1.95676732e+00
2.03591987e-01 -2.03143373e-01 4.57826912e-01 7.92988539e-01
8.14713061e-01 9.69959676e-01 1.41430721e-01 -1.02121688e-01
1.28971148e+00 -1.21575224e+00 -7.61812449e-01 -4.35445368e-01
6.93365574e-01 -9.93242562e-01 1.29034722e+00 1.45709947e-01
-1.12504458e+00 -5.60008883e-01 -9.00666475e-01 -9.75406319e-02
-3.73944551e-01 7.30969608e-01 2.97035456e-01 5.12720168e-01
-1.23180759e+00 1.32659793e-01 -7.25331604e-01 -7.04885244e-01
3.12060714e-01 6.55412674e-01 -6.35442197e-01 -2.54318476e-01
-1.09160554e+00 1.28148031e+00 6.36027336e-01 4.98455375e-01
-1.33419037e+00 -4.14645553e-01 -9.15980220e-01 2.87240446e-02
3.05580646e-01 -9.45040703e-01 1.23452795e+00 -1.43806505e+00
-1.97384930e+00 9.21210885e-01 -1.99466422e-01 -2.42038101e-01
5.28911114e-01 5.14338389e-02 -4.02318507e-01 2.68713117e-01
-8.72173384e-02 1.00437832e+00 1.07709980e+00 -1.48799169e+00
-5.09085596e-01 -2.18713254e-01 -1.45639047e-01 5.45001209e-01
-5.44019163e-01 6.64420873e-02 -4.57828045e-01 -5.22958279e-01
-1.00626647e-01 -1.09991443e+00 1.45826876e-01 -5.29116035e-01
-4.78536814e-01 4.53257337e-02 8.26207876e-01 -1.11516714e+00
7.40180910e-01 -1.84666324e+00 6.86029315e-01 9.59923565e-02
-1.15573257e-01 3.88141930e-01 -7.67341852e-01 4.46606636e-01
-2.24519789e-01 -2.30104476e-01 -2.51824349e-01 -5.04558802e-01
1.55630350e-01 1.75107166e-01 -8.67353007e-02 1.64391935e-01
2.78711319e-01 1.38521218e+00 -4.60147530e-01 -4.90498304e-01
2.32992738e-01 5.44457555e-01 -2.22486883e-01 3.46070856e-01
-3.12291235e-01 7.73943245e-01 1.69020976e-04 6.31860137e-01
6.43419802e-01 -3.01876187e-01 5.75179994e-01 -5.90862751e-01
1.45682603e-01 1.62614927e-01 -5.01541138e-01 2.09636211e+00
-6.68512881e-01 6.43205166e-01 1.10605337e-01 -1.04076743e+00
5.87961614e-01 6.77121580e-01 1.02332249e-01 -1.03783417e+00
4.35520232e-01 2.88513511e-01 3.41270566e-02 -6.20647013e-01
1.95270047e-01 -4.24469978e-01 -1.37639180e-01 4.00361389e-01
7.70519912e-01 1.18176833e-01 -1.16818562e-01 1.98477611e-01
7.43366599e-01 4.14313883e-01 -2.63137102e-01 -3.89925241e-02
5.59775651e-01 4.41091880e-02 4.35765058e-01 1.91229269e-01
1.55726939e-01 4.73842949e-01 3.00860822e-01 -3.10134087e-02
-1.27815151e+00 -1.08354807e+00 5.90408891e-02 1.31768763e+00
-3.45907182e-01 -5.39797060e-02 -1.04525328e+00 -1.08540559e+00
-4.49480802e-01 7.76125848e-01 -7.26900160e-01 -8.25824514e-02
-6.88378394e-01 -8.29438269e-01 5.20725608e-01 5.65004289e-01
7.22590566e-01 -9.28977549e-01 1.17692977e-01 -1.03334643e-01
-7.35281885e-01 -1.41158950e+00 -7.85020649e-01 4.00021046e-01
-9.38285708e-01 -5.01291692e-01 -1.05655015e+00 -1.22586715e+00
8.24793398e-01 -1.57260761e-01 8.93984079e-01 -5.11518240e-01
3.46580774e-01 2.15283900e-01 -2.26283014e-01 1.17021970e-01
-6.84422255e-01 3.58989060e-01 -1.95248216e-01 2.54876673e-01
2.64548182e-01 -4.79855508e-01 -3.59653920e-01 3.91849458e-01
-6.93699121e-01 3.47023815e-01 1.30648065e+00 1.08687103e+00
3.24287534e-01 -4.61425781e-01 4.34201211e-01 -7.42828131e-01
3.13950926e-01 -1.80295303e-01 -3.90498906e-01 6.72055542e-01
-3.98817152e-01 6.14250228e-02 4.76440281e-01 -8.78661513e-01
-1.47756612e+00 1.02073230e-01 6.80890083e-02 -6.01087034e-01
6.09506015e-03 4.44857776e-01 -5.70433497e-01 -7.18101934e-02
4.81448919e-01 3.67870867e-01 -2.71383822e-02 -3.91025901e-01
6.69441164e-01 7.19492137e-01 5.54719150e-01 -6.49929464e-01
1.02661145e+00 1.34605512e-01 -2.66563028e-01 -3.84024471e-01
-5.03478765e-01 2.25511506e-01 -8.16524267e-01 -9.99041498e-02
1.05666959e+00 -1.11850476e+00 -3.94106239e-01 3.43967646e-01
-9.92162287e-01 -7.56789386e-01 1.72711432e-01 6.42826974e-01
-6.00471258e-01 2.30440602e-01 -9.94705737e-01 -5.22842765e-01
-4.63529378e-01 -1.25477087e+00 9.44896996e-01 -2.14201324e-02
2.41314128e-01 -1.13173652e+00 3.09592672e-02 9.70388830e-01
4.53964621e-01 -1.12754710e-01 1.15798295e+00 -6.61677122e-01
-6.13937855e-01 1.12986557e-01 -3.63614410e-01 5.96158028e-01
1.47431213e-02 -3.96895230e-01 -1.05376625e+00 -5.30830562e-01
-1.94922611e-01 -6.84631050e-01 7.20191538e-01 2.22300142e-01
3.45150769e-01 -1.96984351e-01 -3.86391014e-01 6.64486885e-01
1.21963322e+00 9.06237289e-02 6.99151993e-01 3.35493505e-01
7.97698796e-01 5.43531060e-01 4.09179002e-01 -3.16259623e-01
5.06099939e-01 6.12395823e-01 2.84261405e-01 -4.56891865e-01
-3.49555254e-01 -4.49964106e-01 7.49331117e-01 1.21962583e+00
-1.41420975e-01 -2.98304826e-01 -7.78796673e-01 4.86839622e-01
-1.83394539e+00 -6.51986420e-01 2.70289928e-01 2.10298371e+00
1.01296401e+00 -1.21561669e-01 7.34185576e-02 -4.72908050e-01
7.79342651e-01 -1.96774572e-01 -4.41508800e-01 -3.33306104e-01
-4.25953358e-01 1.49397820e-01 4.38834876e-01 5.04892230e-01
-9.57896948e-01 1.39620459e+00 5.69339561e+00 1.07428586e+00
-1.33043706e+00 6.99682832e-01 5.08739054e-01 7.95724615e-02
-3.09148163e-01 1.46677243e-02 -5.70887685e-01 2.90047765e-01
1.16713965e+00 3.53230119e-01 6.88602865e-01 1.77702561e-01
1.76171679e-02 1.53128102e-01 -1.07916880e+00 8.82466614e-01
2.40766689e-01 -7.83176243e-01 3.33240330e-01 2.26619348e-01
8.25335801e-01 1.90478429e-01 4.87284660e-01 7.48762429e-01
2.47951031e-01 -1.07333148e+00 6.11298203e-01 1.35527134e-01
1.11643469e+00 -7.86950648e-01 7.28359997e-01 2.80941546e-01
-9.01918948e-01 -2.13921722e-02 -3.42655897e-01 5.65427065e-01
1.09021686e-01 8.36913362e-02 -1.16225672e+00 8.27980399e-01
3.89256865e-01 3.46776396e-01 -4.63547587e-01 6.08938515e-01
-3.49711329e-01 6.68122232e-01 -2.27513745e-01 2.51788795e-01
3.63083869e-01 -2.87728697e-01 3.61830980e-01 1.23099089e+00
5.52359402e-01 -4.78865713e-01 -2.33492881e-01 6.64866805e-01
-3.92638415e-01 3.61282051e-01 -4.32277977e-01 -1.82103962e-01
2.21353024e-02 1.54289532e+00 -4.14563149e-01 -3.89456153e-01
-6.05153978e-01 1.42340934e+00 4.87027407e-01 8.03189695e-01
-1.09811020e+00 9.71129462e-02 4.34616841e-02 -1.62956849e-01
6.71834946e-01 3.44445519e-02 2.08754675e-03 -1.53423369e+00
2.18729209e-02 -1.21986008e+00 2.86361724e-01 -1.02613342e+00
-1.19079399e+00 1.04285419e+00 -1.89553574e-01 -1.25022638e+00
-3.54560137e-01 -8.56216967e-01 -2.42333889e-01 1.10765576e+00
-1.46833432e+00 -1.98125219e+00 2.36841440e-01 9.07122731e-01
4.67619002e-01 -4.32172984e-01 8.67174208e-01 5.53864539e-01
-5.51469147e-01 9.95042384e-01 1.24844961e-01 2.08239585e-01
1.13841760e+00 -1.07217073e+00 6.63492978e-02 7.32457876e-01
4.39379886e-02 5.41243374e-01 4.62967783e-01 -7.17821181e-01
-1.45793128e+00 -1.02969837e+00 9.60918009e-01 -6.48377061e-01
5.91812134e-01 -6.11301243e-01 -7.07984805e-01 1.17858052e+00
1.02931726e+00 -3.55309546e-01 5.95642388e-01 1.15670748e-01
-5.10780871e-01 -1.10231392e-01 -1.13332772e+00 5.48505783e-01
8.16102505e-01 -7.11741447e-01 -7.53141820e-01 4.35741216e-01
4.90835518e-01 -5.19317031e-01 -9.34404492e-01 5.31643689e-01
4.37593311e-01 -4.37523156e-01 7.92183518e-01 -7.22734809e-01
3.91490161e-01 -1.79904386e-01 -3.75066727e-01 -1.63039124e+00
2.50300821e-02 -7.19773710e-01 1.92046359e-01 1.42111254e+00
1.05239451e+00 -4.88021553e-01 4.78552788e-01 3.68952066e-01
-2.03590885e-01 -2.85765678e-01 -1.03983486e+00 -6.15469694e-01
3.49357188e-01 -8.02809522e-02 2.90655494e-01 1.17559195e+00
1.86722770e-01 9.77616191e-01 -7.97505856e-01 7.07990006e-02
5.64191580e-01 -1.20053560e-01 7.65828729e-01 -6.19021654e-01
-3.46804410e-01 -1.43096730e-01 2.86902517e-01 -1.06520736e+00
2.25118533e-01 -1.23079407e+00 2.30077133e-02 -1.39384413e+00
5.91662109e-01 1.41149789e-01 -1.79187015e-01 7.79851139e-01
-1.01784870e-01 3.57352078e-01 1.25982106e-01 6.03657067e-02
-5.51343977e-01 7.98237801e-01 1.58670604e+00 -3.50188285e-01
6.41959757e-02 -3.02586526e-01 -3.31014693e-01 3.04213196e-01
4.58054423e-01 -3.49493682e-01 -4.88767087e-01 -8.07626605e-01
8.80732760e-02 2.49508828e-01 1.60160914e-01 -6.97982550e-01
6.04303889e-02 1.89561307e-01 5.77926874e-01 -5.52634060e-01
5.31078219e-01 -9.61637795e-01 1.10279001e-01 1.12145238e-01
-3.13262045e-01 1.64234981e-01 4.33236510e-01 2.50922680e-01
-4.44152094e-02 4.89307269e-02 6.79082990e-01 -7.24892840e-02
-3.91710818e-01 1.04539216e-01 -2.84003645e-01 -1.91760615e-01
6.73422754e-01 8.94827545e-02 -5.09824216e-01 -4.58462983e-01
-9.45607603e-01 2.93875277e-01 3.49113882e-01 4.56360996e-01
3.37363690e-01 -1.65620899e+00 -7.17802465e-01 1.12952970e-01
9.28269401e-02 -5.54632545e-01 3.75579149e-01 1.19997847e+00
-1.19220085e-01 5.18996358e-01 -3.59934062e-01 -7.09869087e-01
-1.11969447e+00 7.91849256e-01 4.72599149e-01 -2.90635526e-01
-3.64873022e-01 6.72951639e-01 3.68623793e-01 -1.10729516e+00
1.08357608e-01 -1.07321195e-01 4.15830836e-02 -6.90361932e-02
1.49539039e-01 -1.57547474e-01 1.39911115e-01 -9.26000655e-01
-2.81111747e-01 6.28095090e-01 -1.08172879e-01 -7.35551536e-01
1.25444627e+00 -4.00626868e-01 -4.89132069e-02 2.93676227e-01
1.40348172e+00 -8.56138617e-02 -1.13859022e+00 -5.57741165e-01
-2.82151222e-01 6.18087128e-02 -1.30113894e-02 -1.36689079e+00
-1.21392334e+00 1.12047195e+00 8.28922987e-01 -4.13707763e-01
1.25659168e+00 8.60711560e-02 9.46449399e-01 4.36812818e-01
8.13901275e-02 -1.23134995e+00 7.91026652e-02 4.21209753e-01
8.04834902e-01 -1.35745561e+00 -6.50574088e-01 1.04331998e-02
-9.42723989e-01 9.87032056e-01 5.21671891e-01 3.33027780e-01
1.60640460e-02 1.53903946e-01 5.62997699e-01 2.90126279e-02
-7.15795636e-01 -1.79560989e-01 6.28258407e-01 5.23681700e-01
3.69219273e-01 -8.78295526e-02 -1.00231200e-01 3.33639950e-01
1.57030183e-03 -1.19866252e-01 -1.19246885e-01 6.52392268e-01
-8.09533224e-02 -1.27517486e+00 -3.85755509e-01 -4.09814622e-03
-3.33774924e-01 -4.22272712e-01 -3.31336737e-01 8.02950144e-01
3.21231559e-02 9.01589215e-01 -3.84300381e-01 -5.99735975e-01
3.14366043e-01 5.01756966e-01 7.64669478e-01 -3.38171929e-01
-6.70393586e-01 6.19325697e-01 2.21136689e-01 -3.35084081e-01
-4.76455510e-01 -6.63311779e-01 -8.65196824e-01 -9.88140926e-02
-2.63586730e-01 1.08016513e-01 7.95843303e-01 1.13010967e+00
2.79996932e-01 5.12172997e-01 4.99971390e-01 -8.91426325e-01
-4.43427205e-01 -1.49651098e+00 -3.16342190e-02 3.87524426e-01
2.63481438e-01 -3.73956949e-01 -3.85989219e-01 1.88056543e-01] | [11.45297622680664, 1.5367321968078613] |
06f663dc-0a4a-4873-9f9b-057c61e7ba13 | declipping-of-speech-signals-using-frequency | 2204.04068 | null | https://arxiv.org/abs/2204.04068v1 | https://arxiv.org/pdf/2204.04068v1.pdf | Declipping of Speech Signals Using Frequency Selective Extrapolation | The reconstruction of clipped speech signals is an important task in audio signal processing to achieve an enhanced audio quality for further processing. In this paper, Frequency Selective Extrapolation (FSE), which is commonly used for error concealment or the reconstruction of incomplete image data, is adapted to be able to restore audio signals which are distorted from clipping. For this, FSE generates a model of the signal as an iterative superposition of Fourier basis functions. Clipped samples can then be replaced by estimated samples from the model. The performance of the proposed algorithm is evaluated by using different speech test data sets. Compared to other state-of-the-art declipping algorithms, this leads to a maximum gain in SNR of up to 3:5 dB and an average gain of 1 dB. | ['André Kaup', 'Jürgen Seiler', 'Markus Jonscher'] | 2022-04-07 | null | null | null | null | ['audio-signal-processing'] | ['audio'] | [ 7.39244461e-01 -4.04849984e-02 3.10831577e-01 1.30265881e-03
-8.34527314e-01 -7.81669468e-02 2.72913963e-01 1.25494286e-01
-3.83638829e-01 9.83162284e-01 2.52744019e-01 -6.41388744e-02
-1.81113213e-01 -3.46817166e-01 -4.69936311e-01 -8.73693705e-01
-2.16154516e-01 -2.08439767e-01 4.02714640e-01 2.18752295e-01
2.94138640e-01 5.05349636e-01 -1.98539913e+00 4.60898876e-01
8.92927706e-01 1.16502547e+00 4.21723366e-01 6.87616765e-01
5.04503585e-02 6.24875307e-01 -8.57968986e-01 -7.19823390e-02
-5.49297743e-02 -5.41507423e-01 -2.23369047e-01 9.96548980e-02
-1.50239155e-01 -4.51882988e-01 -2.71462977e-01 1.22008371e+00
4.08087343e-01 3.00664902e-02 6.63725913e-01 -6.88736260e-01
1.59258638e-02 4.62339580e-01 -2.43178591e-01 1.96874321e-01
5.30978262e-01 -2.40715951e-01 1.82730705e-01 -1.17976403e+00
2.92413652e-01 6.87966049e-01 5.12615681e-01 2.24947706e-01
-1.29733491e+00 -6.10146284e-01 -7.40406156e-01 7.96535790e-01
-1.71864438e+00 -9.12699461e-01 1.17174542e+00 -2.44331896e-01
3.79893303e-01 5.09222031e-01 5.72456598e-01 5.35286546e-01
2.67628461e-01 5.16454399e-01 1.21517622e+00 -6.37233078e-01
2.58606136e-01 3.10312092e-01 -2.09135219e-01 -3.45126726e-02
6.61776587e-02 2.84932286e-01 -7.33236849e-01 -1.31645888e-01
5.26723146e-01 -3.73570740e-01 -1.11989796e+00 7.50855729e-02
-7.24104464e-01 4.16171968e-01 1.66159600e-01 3.61041337e-01
-7.78168380e-01 -7.33157918e-02 3.09706151e-01 3.79996240e-01
5.39696157e-01 2.70701060e-03 3.03938627e-01 -1.53244451e-01
-1.39342976e+00 3.06140602e-01 5.71870625e-01 4.62904483e-01
2.96123564e-01 5.95580935e-01 3.04403752e-02 8.49172473e-01
3.20504308e-01 4.31920409e-01 3.35609674e-01 -6.35183692e-01
3.58496994e-01 -8.89800489e-02 3.58620346e-01 -7.90329337e-01
-1.29871725e-04 -6.15678132e-01 -9.94994521e-01 4.52379644e-01
3.45843196e-01 -2.07298435e-02 -6.05562747e-01 1.29800415e+00
2.36325905e-01 5.06202579e-01 3.61116469e-01 9.04505789e-01
5.12708247e-01 1.18303108e+00 -3.60570550e-01 -7.27170646e-01
1.22634196e+00 -2.59487242e-01 -1.14563298e+00 6.00373000e-02
-2.69831538e-01 -1.31505728e+00 5.43821573e-01 1.02469158e+00
-1.08842933e+00 -8.01381767e-01 -1.21576488e+00 3.55405003e-01
3.22715431e-01 1.87491760e-01 -2.69053787e-01 6.10469103e-01
-8.80945027e-01 6.25244498e-01 -4.84608233e-01 3.15279216e-01
1.35957703e-01 1.40973151e-01 -3.74003977e-01 -1.78278685e-02
-1.15178227e+00 5.41496992e-01 2.85818875e-01 2.28160664e-01
-7.61807442e-01 -9.30429101e-01 -4.87601995e-01 2.97931999e-01
1.63536027e-01 -6.66027889e-02 9.90357220e-01 -1.00165999e+00
-1.81223524e+00 2.51671314e-01 -2.40052953e-01 -8.41307044e-01
5.33418596e-01 -1.78944424e-01 -8.51578116e-01 6.99024618e-01
-5.15944362e-01 -9.09138620e-02 1.56641185e+00 -1.14011669e+00
-3.26091290e-01 -2.68203676e-01 -6.21308565e-01 8.82318690e-02
-1.79190338e-01 5.90672605e-02 -7.29583204e-02 -9.95242059e-01
3.16949725e-01 -4.87733006e-01 1.78932771e-01 -2.59484574e-02
-1.61151543e-01 5.86870670e-01 9.20998871e-01 -1.38101971e+00
1.22207820e+00 -2.47954774e+00 1.84958145e-01 2.48045281e-01
-2.47261629e-01 5.31840801e-01 2.85635114e-01 6.41399801e-01
-1.52290419e-01 -4.79148000e-01 -6.27275109e-01 -2.54022360e-01
-5.36658406e-01 -2.01929316e-01 -4.51086104e-01 6.48981690e-01
9.68255252e-02 -1.60878673e-01 -4.00695741e-01 -1.11904919e-01
4.38063234e-01 1.02661204e+00 -2.93128133e-01 3.25508833e-01
1.43957853e-01 6.32499814e-01 7.67717212e-02 2.81626076e-01
1.09799528e+00 5.84867954e-01 6.19538464e-02 -6.77049831e-02
-3.03326756e-01 1.97466373e-01 -1.47101867e+00 1.11311960e+00
-3.73219728e-01 8.15286934e-01 4.75523651e-01 -9.13024724e-01
1.33011496e+00 9.68422055e-01 2.96870172e-01 -3.94366771e-01
6.65137619e-02 5.18315732e-01 -1.97981726e-02 -5.63561976e-01
4.05525923e-01 -1.79077566e-01 4.27042186e-01 -6.21729493e-02
-1.76716015e-01 -1.91596061e-01 -2.79952615e-01 -2.96499312e-01
6.80581808e-01 -1.38811633e-01 3.32827866e-01 -3.13736677e-01
1.10439026e+00 -5.75028419e-01 4.12979364e-01 1.02233715e-01
1.22338250e-01 7.05253065e-01 9.62095633e-02 -6.18771911e-02
-1.13423753e+00 -9.02982295e-01 -5.18716812e-01 2.03470457e-02
-1.46912217e-01 -1.69486567e-01 -9.82717991e-01 4.38246168e-02
-3.89823079e-01 8.62708867e-01 3.06453463e-02 -8.42790902e-02
-6.12267017e-01 -3.35748583e-01 4.48606431e-01 -4.38257214e-03
6.65255845e-01 -9.33979094e-01 -6.39149785e-01 6.09025300e-01
-3.79814059e-01 -1.12750983e+00 -1.84051126e-01 -2.72035785e-02
-8.69161606e-01 -9.61225688e-01 -1.14439178e+00 -5.90268314e-01
5.27217329e-01 1.50527567e-01 5.08309543e-01 6.23725206e-02
-9.61411595e-02 4.09564562e-02 -6.39821768e-01 -3.32056552e-01
-1.00206280e+00 -5.24521410e-01 4.56852801e-02 7.66398132e-01
-8.18986073e-02 -7.49876440e-01 -6.01889074e-01 2.34707728e-01
-1.15335000e+00 -1.77255020e-01 4.51508403e-01 8.15619171e-01
4.76610005e-01 4.85740572e-01 8.05850089e-01 -4.48438674e-01
6.45516276e-01 -2.06715211e-01 -9.89221871e-01 -2.14299738e-01
-3.17184508e-01 -1.39039055e-01 1.12543559e+00 -4.12048906e-01
-1.30913460e+00 1.22837126e-01 -5.99587619e-01 -4.48268563e-01
-9.39841494e-02 4.10631835e-01 -1.31134972e-01 -3.37259233e-01
5.00841618e-01 8.88493240e-01 3.46341699e-01 -8.89266372e-01
-1.86402336e-01 1.13623607e+00 8.70949090e-01 -1.31272182e-01
5.42028666e-01 2.31892765e-01 1.44375861e-01 -1.39463472e+00
2.27449641e-01 -3.67325157e-01 -9.00639072e-02 -4.97804970e-01
4.28159505e-01 -7.00797737e-01 -4.92949575e-01 9.35672224e-01
-1.19791567e+00 1.10637762e-01 -8.70469585e-02 8.83755088e-01
-4.97021347e-01 6.98530793e-01 -5.55586636e-01 -1.37862802e+00
-3.24084789e-01 -1.15489852e+00 4.95677710e-01 8.99602547e-02
1.31089941e-01 -5.00172496e-01 -2.92006463e-01 -5.90768061e-04
4.50976551e-01 1.09081730e-01 6.69696927e-01 -2.16958016e-01
-4.56259966e-01 -3.44568700e-01 3.46371859e-01 7.61348784e-01
-4.59422842e-02 -2.46992618e-01 -1.01348186e+00 -3.90887916e-01
6.29821181e-01 2.47274816e-01 8.37873042e-01 3.97794545e-01
9.36497152e-01 -3.87782782e-01 -6.04082346e-02 4.53447163e-01
1.61428297e+00 5.37593424e-01 1.15479052e+00 -1.00872897e-01
-2.83140689e-01 6.60288215e-01 6.05650604e-01 7.09039271e-01
-4.68079716e-01 8.72268200e-01 3.78862321e-01 2.90583253e-01
-4.48965251e-01 -2.57860631e-01 3.61917675e-01 8.72302115e-01
-7.18524754e-02 -1.96299687e-01 -5.40183187e-01 5.55364192e-01
-1.31569970e+00 -1.10485649e+00 -3.71097803e-01 2.78651929e+00
6.67019010e-01 1.88273966e-01 -3.99609245e-02 1.18828297e+00
9.11779165e-01 7.47142285e-02 -1.85162380e-01 -3.04216206e-01
2.08936334e-01 5.08450687e-01 3.47802281e-01 7.64171064e-01
-6.85580611e-01 2.54746795e-01 5.80430412e+00 1.11606824e+00
-1.43380022e+00 -9.71669704e-02 1.23532116e-01 2.32907116e-01
-8.28752071e-02 -6.27734289e-02 -4.38214689e-01 7.12500691e-01
1.03917491e+00 -2.01691762e-01 6.09407544e-01 3.97368520e-01
4.72667068e-01 -3.19134831e-01 -4.88533705e-01 1.01012886e+00
1.31945342e-01 -9.07053411e-01 -1.20920807e-01 3.37205902e-02
3.21843475e-01 -7.80070245e-01 7.12994784e-02 -3.63714397e-01
-6.93624437e-01 -6.37668073e-01 8.90425265e-01 6.10204697e-01
9.13251281e-01 -1.13037801e+00 8.21786404e-01 6.36532485e-01
-1.10034311e+00 -1.08098969e-01 -4.55801755e-01 -8.13396424e-02
6.05587542e-01 1.08896840e+00 -1.07908964e+00 8.34899902e-01
4.86797392e-01 1.52823105e-01 2.04896778e-01 1.60146093e+00
-3.34828466e-01 9.62991357e-01 -3.52601320e-01 2.30051264e-01
-1.39395475e-01 -1.94433033e-01 1.13045394e+00 9.13013101e-01
9.25302505e-01 1.56856149e-01 -4.96592164e-01 5.54464102e-01
1.69058457e-01 2.76355922e-01 -4.23684627e-01 2.61382014e-01
6.57641828e-01 7.68298447e-01 -3.92464429e-01 -2.44673058e-01
-1.56781897e-01 9.34591174e-01 -4.83525336e-01 3.56733829e-01
-6.91536486e-01 -8.41480613e-01 4.58961487e-01 4.58287120e-01
6.58958793e-01 -1.42885864e-01 3.00515860e-01 -7.80806720e-01
1.87353089e-01 -9.03446853e-01 -1.43978909e-01 -7.28287041e-01
-5.95921457e-01 8.49518895e-01 -5.50599508e-02 -1.80272508e+00
-4.68612909e-01 -2.15468243e-01 -3.96142304e-01 1.01386023e+00
-1.54141033e+00 -4.54446584e-01 -6.34627864e-02 6.25698149e-01
6.33051038e-01 -2.16625944e-01 9.06036675e-01 4.86905783e-01
7.77913928e-02 3.48373383e-01 4.68524009e-01 -3.28299373e-01
3.14306706e-01 -7.15814054e-01 8.14438611e-02 1.05086875e+00
1.30261667e-02 1.51854306e-01 1.34108198e+00 -5.23449063e-01
-1.07692492e+00 -7.42255092e-01 9.14872289e-01 6.68730319e-01
-5.43192448e-03 -4.75322865e-02 -1.32517624e+00 3.03974431e-02
2.40318611e-01 -8.57317150e-02 5.22109807e-01 -8.56346786e-01
-1.08541377e-01 -4.08722609e-01 -1.50091171e+00 1.02494881e-01
2.65652031e-01 -4.24107313e-01 -6.11010492e-01 -3.48763347e-01
3.51770729e-01 -3.48878592e-01 -8.56497645e-01 3.26899379e-01
4.05519158e-01 -1.33253646e+00 9.60796773e-01 2.32877851e-01
2.19893232e-01 -5.31611979e-01 -1.90197378e-01 -1.51003754e+00
8.35817680e-02 -9.20503139e-01 -2.15696041e-02 1.30612540e+00
1.37394801e-01 -6.79494202e-01 6.36477292e-01 -8.54108483e-02
-1.23108923e-01 -3.34818214e-01 -1.25975108e+00 -7.82011032e-01
-5.32103539e-01 -4.33065623e-01 5.34141898e-01 3.88422996e-01
1.99218810e-01 -1.63975507e-01 -8.09297800e-01 3.80035102e-01
1.05754435e+00 -3.26609641e-01 4.53833431e-01 -1.04117095e+00
-5.22570550e-01 3.64838019e-02 -6.89506352e-01 -1.03071809e+00
-7.57470950e-02 -4.52751338e-01 9.69531834e-02 -1.11698067e+00
-5.11449873e-01 3.18652578e-02 -1.18997648e-01 -4.09323424e-01
1.33714989e-01 4.82362002e-01 2.68280894e-01 2.29746141e-02
5.32462120e-01 6.22024655e-01 9.21523929e-01 8.32004249e-02
-3.31840575e-01 6.48742855e-01 -1.59844264e-01 7.13306785e-01
6.23448789e-01 -6.23445153e-01 -5.66963196e-01 6.85482621e-02
-1.78546995e-01 7.12825119e-01 2.97184676e-01 -1.50685775e+00
2.13012785e-01 5.00786960e-01 2.52464145e-01 -6.71796381e-01
7.83454716e-01 -1.23797917e+00 7.65468955e-01 6.08865678e-01
-2.75753528e-01 -4.28392053e-01 2.58809388e-01 4.73946780e-01
-8.63275528e-01 -7.07173526e-01 1.04840720e+00 2.80892756e-02
-3.17143559e-01 -4.07962739e-01 -7.02936888e-01 -5.25420725e-01
8.73277128e-01 -2.39936545e-01 1.15630731e-01 -6.51250064e-01
-7.85833478e-01 -5.98011613e-01 1.55550167e-01 -2.50503212e-01
1.04214251e+00 -1.16504538e+00 -9.35317516e-01 5.31736851e-01
-3.14803302e-01 -3.14551383e-01 6.84150338e-01 7.55191386e-01
-7.07058668e-01 2.72791505e-01 -1.03941530e-01 -4.30766433e-01
-1.58524084e+00 5.09678245e-01 1.62559062e-01 1.28186643e-01
-8.37883294e-01 3.10047328e-01 -4.13007796e-01 5.51270902e-01
1.70983568e-01 -1.55639634e-01 -4.42203671e-01 -1.65879026e-01
1.14108098e+00 5.96459091e-01 3.50241274e-01 -6.73808992e-01
-1.78934067e-01 6.09183192e-01 3.39836210e-01 -4.81209040e-01
1.19486010e+00 -6.82177395e-02 -7.22023919e-02 1.88848257e-01
1.23268282e+00 5.98940134e-01 -1.18223166e+00 -1.03226446e-01
-2.50202805e-01 -8.85383248e-01 4.18322891e-01 -4.73265678e-01
-7.63197660e-01 8.72401297e-01 5.92576146e-01 4.67579514e-01
1.70086789e+00 -4.41060841e-01 7.31427133e-01 -2.14435741e-01
4.54159737e-01 -6.79711342e-01 -2.45563358e-01 -2.13650525e-01
1.26260841e+00 -4.38453197e-01 9.21474323e-02 -5.12880623e-01
-2.87262291e-01 1.38470733e+00 -3.01747590e-01 -1.81347370e-01
6.12880468e-01 4.85905558e-01 -1.21052392e-01 5.46790302e-01
-4.02834445e-01 2.76716687e-02 1.66485235e-01 6.64244056e-01
2.07764462e-01 -1.41156167e-01 -7.46580660e-01 4.92881298e-01
-1.63791060e-01 4.18981045e-01 7.51291871e-01 5.72281718e-01
-6.87870145e-01 -1.05327630e+00 -1.15111244e+00 1.21399604e-01
-7.08323658e-01 9.26211253e-02 1.75924644e-01 4.68335211e-01
-2.71863520e-01 1.19656408e+00 -1.16127826e-01 -3.22408229e-01
3.74489248e-01 2.09632680e-01 4.35705125e-01 8.62664506e-02
-3.20146590e-01 5.03259182e-01 7.92425647e-02 -1.98394343e-01
-3.01110774e-01 -6.77737296e-01 -1.14286280e+00 -2.29152054e-01
-3.60967398e-01 4.22623366e-01 1.05000222e+00 5.56943595e-01
1.16649672e-01 5.95738113e-01 9.19418335e-01 -8.16317201e-01
-6.05974674e-01 -8.40114832e-01 -8.68392944e-01 1.57927766e-01
7.57443249e-01 -3.59649956e-01 -7.45517850e-01 2.04104945e-01] | [14.919153213500977, 5.6867218017578125] |
868ef3b2-059f-4d69-aede-8f72623358b4 | generative-adversarial-reward-learning-for | 2105.00822 | null | https://arxiv.org/abs/2105.00822v2 | https://arxiv.org/pdf/2105.00822v2.pdf | Generative Adversarial Reward Learning for Generalized Behavior Tendency Inference | Recent advances in reinforcement learning have inspired increasing interest in learning user modeling adaptively through dynamic interactions, e.g., in reinforcement learning based recommender systems. Reward function is crucial for most of reinforcement learning applications as it can provide the guideline about the optimization. However, current reinforcement-learning-based methods rely on manually-defined reward functions, which cannot adapt to dynamic and noisy environments. Besides, they generally use task-specific reward functions that sacrifice generalization ability. We propose a generative inverse reinforcement learning for user behavioral preference modelling, to address the above issues. Instead of using predefined reward functions, our model can automatically learn the rewards from user's actions based on discriminative actor-critic network and Wasserstein GAN. Our model provides a general way of characterizing and explaining underlying behavioral tendencies, and our experiments show our method outperforms state-of-the-art methods in a variety of scenarios, namely traffic signal control, online recommender systems, and scanpath prediction. | ['Quan Z. Sheng', 'Wenjie Zhang', 'Aixin Sun', 'Xianzhi Wang', 'Lina Yao', 'Xiaocong Chen'] | 2021-05-03 | null | null | null | null | ['scanpath-prediction'] | ['computer-vision'] | [-1.63134903e-01 -7.39969462e-02 -4.50514257e-01 -4.72043246e-01
-4.82239902e-01 -2.18772456e-01 2.21801803e-01 -1.48782104e-01
-3.59588474e-01 8.45684171e-01 5.45786917e-02 -4.12356257e-01
-4.13100332e-01 -9.55831707e-01 -4.96177882e-01 -6.46455884e-01
-6.18464276e-02 6.85549617e-01 3.60834002e-01 -6.64028525e-01
3.02013576e-01 2.53766119e-01 -1.35754335e+00 -2.27695227e-01
1.08614790e+00 9.19690251e-01 4.08273876e-01 7.53485143e-01
-4.64210622e-02 7.74583399e-01 -4.14265931e-01 -2.43112415e-01
2.65797637e-02 -6.32270813e-01 -4.20377702e-01 -1.43593280e-02
-4.41652000e-01 -4.51565742e-01 -5.43471694e-01 6.95150673e-01
4.85573828e-01 5.28863609e-01 6.77488208e-01 -1.44407725e+00
-1.04041338e+00 7.48609722e-01 -3.33400756e-01 2.02589482e-01
1.02540910e-01 3.46574336e-01 1.34884846e+00 -2.41143703e-01
7.83303678e-02 1.18371046e+00 2.91442007e-01 8.99872780e-01
-1.26344049e+00 -5.28873622e-01 6.12380147e-01 6.79407477e-01
-7.67073095e-01 1.00229539e-01 8.05622160e-01 -2.68290967e-01
3.49927604e-01 1.70142531e-01 7.83168912e-01 1.35246575e+00
2.67356657e-03 1.14386940e+00 9.91099298e-01 -1.08746789e-01
4.92719263e-01 2.43708432e-01 5.39004356e-02 4.33341622e-01
-3.71459007e-01 3.80354971e-01 -1.59227327e-01 -2.47089937e-01
1.06789660e+00 3.23997945e-01 1.55684441e-01 -6.56220138e-01
-7.27180123e-01 1.08640075e+00 1.96114779e-01 -6.39765859e-02
-6.12476230e-01 3.42904955e-01 1.48705225e-02 3.27141285e-01
2.55023867e-01 4.02541310e-01 -5.08031487e-01 -5.46289623e-01
-4.39240724e-01 3.73857409e-01 6.89160049e-01 9.38422143e-01
8.10975373e-01 3.89193565e-01 -5.59720278e-01 1.18098283e+00
5.13945282e-01 6.64465249e-01 5.02551734e-01 -9.89018977e-01
-8.88907351e-03 1.94298223e-01 4.44627106e-01 -5.21343827e-01
-3.53602976e-01 -5.03304124e-01 -3.73193920e-01 3.74720693e-02
4.10651296e-01 -4.42744315e-01 -8.16664934e-01 1.74319732e+00
1.90318778e-01 4.75024253e-01 -1.84900209e-01 1.09159505e+00
3.30524474e-01 4.32245076e-01 1.58984199e-01 -2.28539139e-01
8.37469399e-01 -9.38206971e-01 -4.79507685e-01 -2.13337541e-02
4.33684975e-01 -3.60513717e-01 1.45788133e+00 4.61092949e-01
-8.50844800e-01 -6.14353180e-01 -6.08955145e-01 7.66944349e-01
-5.42718545e-02 -5.66236451e-02 6.24572575e-01 1.00225937e+00
-7.71669984e-01 6.04366660e-01 -7.35896289e-01 -3.45309466e-01
2.99669415e-01 4.30609941e-01 6.39361203e-01 1.86755121e-01
-1.30998969e+00 6.36157334e-01 -2.34729394e-01 -2.06965491e-01
-1.24159122e+00 -4.71397728e-01 -3.80345941e-01 3.58164370e-01
8.02536666e-01 -4.47028279e-01 1.64143550e+00 -1.04038095e+00
-2.36274862e+00 -1.63489655e-01 2.43009493e-01 -6.42239690e-01
5.24478376e-01 -2.99851149e-01 -7.35946596e-01 -3.12724739e-01
-2.94050634e-01 3.27565461e-01 1.04935849e+00 -1.24087143e+00
-8.36372793e-01 -1.77949741e-02 3.83896023e-01 2.37327233e-01
-4.61480200e-01 -2.89566755e-01 -3.49474907e-01 -4.69915032e-01
-5.87459981e-01 -1.07592487e+00 -7.69313991e-01 -5.82512558e-01
-6.11926466e-02 -3.81782323e-01 8.87864053e-01 -1.03414617e-01
1.26247942e+00 -1.96423006e+00 -8.94605219e-02 5.22826791e-01
2.83015892e-02 2.10081905e-01 -3.90252501e-01 3.57763290e-01
4.85033184e-01 -7.25272531e-03 2.92296112e-01 1.55438423e-01
3.38425070e-01 4.99427348e-01 -3.18734795e-01 7.26634935e-02
-5.83531410e-02 8.08915377e-01 -1.31750941e+00 -1.74788386e-01
4.07021880e-01 1.64297342e-01 -1.04612315e+00 7.56499469e-01
-5.88668048e-01 8.65270913e-01 -9.07169819e-01 3.57067525e-01
3.00610900e-01 -1.80218548e-01 3.44393790e-01 3.86599541e-01
1.14774019e-01 9.71315578e-02 -1.18441546e+00 1.19194865e+00
-8.39125574e-01 1.35767415e-01 -2.74248093e-01 -1.24401832e+00
1.03919005e+00 4.89760228e-02 8.14464092e-01 -8.93998861e-01
1.50624290e-01 -4.24086899e-02 3.57286394e-01 -4.51913893e-01
5.64114630e-01 1.77385792e-01 8.46709299e-04 7.39873707e-01
5.39454818e-02 2.37374097e-01 8.98893550e-02 3.48576792e-02
9.32389259e-01 3.38823706e-01 5.76367043e-02 6.45556748e-02
5.18486083e-01 -5.29193640e-01 6.68088913e-01 1.14427531e+00
-2.20242798e-01 2.08434135e-01 5.05765021e-01 -2.64921069e-01
-6.86174393e-01 -9.61146355e-01 2.17885524e-01 1.67935562e+00
3.66115391e-01 -1.08800881e-01 -6.40590429e-01 -9.24452722e-01
8.50186422e-02 9.85220253e-01 -6.06968045e-01 -4.00755107e-01
-4.59900528e-01 -7.82983959e-01 1.01573631e-01 5.54972112e-01
1.23570874e-01 -1.33939946e+00 -5.02404034e-01 5.82917154e-01
8.14546496e-02 -8.25279295e-01 -5.85367858e-01 9.12870914e-02
-7.69163787e-01 -1.03574908e+00 -7.43280113e-01 -1.07513085e-01
4.23178792e-01 3.56232554e-01 1.07715678e+00 1.31039858e-01
5.23175336e-02 9.87318814e-01 -5.52519500e-01 -2.20733643e-01
-4.57024068e-01 2.00080141e-01 2.24126279e-01 4.20970768e-01
3.85357082e-01 -6.85986400e-01 -7.90728927e-01 8.19582224e-01
-6.82393074e-01 -4.57555175e-01 7.37964213e-01 1.17145169e+00
4.74527150e-01 -3.02881598e-01 1.07657754e+00 -1.33119988e+00
1.08987510e+00 -8.12149286e-01 -5.77352405e-01 3.36032063e-01
-1.15511584e+00 2.53901690e-01 1.02357936e+00 -7.92500734e-01
-1.13965571e+00 -3.64032954e-01 -2.97127247e-01 -3.35399091e-01
-4.04593825e-01 3.08854818e-01 -2.76574548e-02 1.03097372e-01
6.71679378e-01 2.31416300e-01 -1.66418120e-01 -1.97281748e-01
6.92373037e-01 6.74956977e-01 5.95225580e-03 -8.05142105e-01
7.03713357e-01 -1.25441523e-02 -2.55778313e-01 -5.74898064e-01
-7.34273314e-01 -3.61226439e-01 -2.61535257e-01 -5.28916359e-01
5.10071099e-01 -3.90342236e-01 -1.07133234e+00 -5.27965054e-02
-4.81042147e-01 -8.24417531e-01 -4.56503987e-01 5.97894847e-01
-1.12312138e+00 1.85524344e-01 -3.89067918e-01 -1.30761492e+00
1.54958040e-01 -1.22634363e+00 6.35155618e-01 5.90958655e-01
-1.67177171e-02 -1.16000259e+00 2.52095014e-01 1.54467329e-01
6.51270509e-01 -3.93172652e-01 9.13642704e-01 -7.15986907e-01
-6.17998362e-01 1.46415904e-01 1.52202442e-01 2.20110089e-01
1.30782142e-01 -1.16363592e-01 -6.48483276e-01 -1.53388351e-01
-5.49160957e-01 -4.68672007e-01 4.46879596e-01 6.34625673e-01
1.64387333e+00 -1.57692641e-01 -1.43870324e-01 2.12126613e-01
9.95567024e-01 4.82485801e-01 7.00127959e-01 1.50152504e-01
3.99360299e-01 3.82235527e-01 8.88264000e-01 8.57165217e-01
4.79363412e-01 7.93330193e-01 7.32790947e-01 -1.96788255e-02
3.88152808e-01 -5.33560336e-01 4.53092337e-01 3.02838892e-01
-3.53822947e-01 -3.68086278e-01 -4.65914547e-01 2.35924438e-01
-2.43998742e+00 -1.06458199e+00 1.29749045e-01 2.31748581e+00
4.42944318e-01 2.86974013e-01 7.92036533e-01 -1.73616558e-01
4.89160687e-01 -4.78236899e-02 -9.86255169e-01 -4.45677996e-01
4.63844150e-01 1.36885762e-01 5.50161779e-01 3.87616932e-01
-6.54151738e-01 1.05864501e+00 6.66888475e+00 6.79956377e-01
-9.64085102e-01 3.92983258e-02 5.62762141e-01 8.97820592e-02
-4.73905981e-01 -5.20977639e-02 -6.69343174e-01 4.53862518e-01
1.13090992e+00 -7.39703141e-03 9.94885921e-01 1.03878319e+00
5.52978814e-01 6.58326000e-02 -8.90931129e-01 9.51672792e-01
-1.80374503e-01 -8.74587536e-01 -1.36035189e-01 1.95356175e-01
5.78547895e-01 -1.11384466e-01 4.52617228e-01 1.00368106e+00
1.02933693e+00 -8.59248519e-01 3.98318112e-01 6.75226390e-01
2.48197719e-01 -1.06024718e+00 4.34056252e-01 5.23072124e-01
-7.05879748e-01 -5.88578463e-01 -4.43152547e-01 6.06956221e-02
2.92742074e-01 2.69630671e-01 -8.94182324e-01 2.24641949e-01
4.32427555e-01 6.86403513e-01 -3.28048706e-01 9.66710150e-01
-2.72375643e-01 1.20330846e+00 5.80063574e-02 -5.03708184e-01
3.03072333e-01 -5.14638186e-01 3.84163827e-01 9.58064914e-01
4.15372461e-01 4.80098277e-02 5.06194293e-01 7.91015089e-01
1.55866578e-01 3.23680043e-01 -2.15000212e-01 -5.39776497e-02
3.58267307e-01 1.41937888e+00 -5.63038111e-01 -1.07598521e-01
-5.42103946e-01 7.76014507e-01 2.93687582e-01 5.96121609e-01
-1.22892463e+00 -7.59033263e-02 7.97907770e-01 2.12501541e-01
4.50328022e-01 -1.90785259e-01 1.22458272e-01 -1.05682492e+00
-5.82872987e-01 -8.87323558e-01 3.03951025e-01 -3.65525961e-01
-1.28878856e+00 3.87999237e-01 -1.21210009e-01 -1.49550116e+00
-6.07094705e-01 -5.27121246e-01 -7.11941361e-01 4.50683832e-01
-1.56120813e+00 -8.30566406e-01 -1.18148111e-01 7.11028337e-01
6.88183427e-01 -5.56619883e-01 6.99379265e-01 2.74458617e-01
-5.85663497e-01 7.43401229e-01 3.10909063e-01 -1.27261609e-01
7.78767943e-01 -1.51367950e+00 1.15351841e-01 2.50359416e-01
4.37141865e-01 3.46664280e-01 9.01294112e-01 -2.21307904e-01
-1.35694683e+00 -8.72708678e-01 -1.15759559e-01 -2.91550606e-01
7.53918886e-01 -1.96424335e-01 -6.50047123e-01 4.02493656e-01
1.23911865e-01 -4.51959530e-03 1.01207042e+00 6.42829061e-01
1.06623676e-02 -1.66831344e-01 -8.51063490e-01 8.56822133e-01
9.85408962e-01 -5.07732294e-02 7.52673075e-02 1.80816218e-01
2.65773118e-01 -2.93451041e-01 -7.75657713e-01 -9.37634036e-02
6.52377605e-01 -8.54594648e-01 8.85611475e-01 -1.14565063e+00
8.97925198e-02 -1.06525794e-02 -3.10687385e-02 -1.93968141e+00
-7.28626370e-01 -8.21386337e-01 -5.23550153e-01 8.87138784e-01
4.97747421e-01 -5.65771639e-01 9.52805281e-01 6.46920800e-01
-1.01682637e-02 -8.99933517e-01 -4.12417978e-01 -7.92360902e-01
-1.53395131e-01 -5.28464675e-01 7.79465973e-01 5.79309762e-01
6.01738542e-02 4.47448134e-01 -1.04473269e+00 -4.43367139e-02
5.54812133e-01 1.36315241e-01 1.12660003e+00 -1.30193353e+00
-1.01440704e+00 -4.77952063e-01 -8.58307481e-02 -1.69953227e+00
2.22424045e-01 -4.36538696e-01 1.14158906e-01 -1.16237056e+00
-1.02697439e-01 -7.30554938e-01 -8.30156863e-01 1.78460196e-01
-2.28962243e-01 -1.91238567e-01 2.80999243e-02 -6.05241768e-03
-1.02283919e+00 9.31642294e-01 1.46295941e+00 1.21437404e-02
-5.25449395e-01 7.54298151e-01 -9.02252018e-01 4.83546287e-01
1.05852985e+00 -4.97245193e-01 -7.91222155e-01 2.22236160e-02
1.33924156e-01 3.09397280e-01 8.53744000e-02 -6.37420118e-01
-6.65173680e-02 -1.04218793e+00 7.72719532e-02 -1.38325512e-01
2.41923556e-01 -9.01375115e-01 -3.47600549e-01 3.10856074e-01
-5.74551225e-01 -1.60317436e-01 -3.43886197e-01 1.17507780e+00
1.33799672e-01 -1.50033444e-01 7.21390009e-01 -5.87430876e-03
-7.31655002e-01 6.89693570e-01 -7.46800542e-01 1.06567807e-01
8.62085819e-01 3.01261302e-02 3.62024903e-02 -1.14352846e+00
-9.60718811e-01 5.66324055e-01 -1.48616480e-02 6.31103694e-01
5.60611010e-01 -1.39444613e+00 -4.66041088e-01 2.70740576e-02
6.77296892e-02 -7.46770024e-01 2.67325759e-01 7.53759801e-01
2.77926445e-01 1.98791251e-01 -3.15431476e-01 -4.79653865e-01
-8.76392365e-01 5.11815190e-01 3.33543301e-01 -5.65958619e-01
-2.26358384e-01 2.46255159e-01 1.34425238e-02 -5.07562459e-01
2.09131032e-01 -1.50003940e-01 -6.04883134e-01 -2.99247772e-01
2.48862296e-01 5.07642806e-01 -4.01115775e-01 -2.25892454e-01
1.57457158e-01 5.53008169e-02 -1.55377835e-01 -3.04243445e-01
1.25818825e+00 -2.29399681e-01 7.87607431e-01 4.41823840e-01
4.34104562e-01 -1.58853978e-01 -1.59333861e+00 -4.20073777e-01
-2.66612954e-02 -5.39951146e-01 1.93534363e-02 -7.07325816e-01
-1.10921180e+00 8.59995246e-01 8.16233754e-01 6.19209588e-01
9.23218191e-01 -2.30207488e-01 8.51559520e-01 6.17459714e-01
5.40295541e-01 -1.62896788e+00 5.12089074e-01 5.90443611e-01
4.49565589e-01 -1.32970369e+00 -3.87820929e-01 1.83964372e-02
-1.12476218e+00 1.00198734e+00 9.55097735e-01 -2.92133003e-01
9.22597945e-01 -1.55291557e-01 1.86741471e-01 2.46896163e-01
-8.81911755e-01 -5.78273952e-01 2.66537905e-01 8.86455536e-01
4.87846583e-01 3.22861701e-01 -2.42869526e-01 8.61759424e-01
1.33970574e-01 -7.72614703e-02 5.63165009e-01 5.91705024e-01
-5.83829522e-01 -1.75200498e+00 -8.88561979e-02 7.34663904e-01
-1.56401142e-01 2.95916051e-01 -4.37937416e-02 4.69439805e-01
-2.60917991e-01 1.06122649e+00 -1.77304327e-01 -6.13696396e-01
4.95900512e-01 -1.03733912e-01 4.02691543e-01 -5.62444746e-01
-4.47029799e-01 1.97378635e-01 -1.64990723e-01 -5.67740679e-01
-3.42797190e-01 -7.74919033e-01 -1.05014169e+00 -8.15491527e-02
-3.45308065e-01 4.64601755e-01 2.38888472e-01 9.74057615e-01
3.02503765e-01 7.98588634e-01 1.15185320e+00 -5.40099144e-01
-1.04660392e+00 -9.77902591e-01 -7.95617223e-01 4.13948089e-01
7.35644251e-02 -1.05567217e+00 6.94664344e-02 -3.06914330e-01] | [4.143524646759033, 2.256166934967041] |
3ebb8649-ea32-4302-bd8f-b1600cd577d6 | joint-super-resolution-and-synthesis-of-1-mm | 2012.13340 | null | https://arxiv.org/abs/2012.13340v1 | https://arxiv.org/pdf/2012.13340v1.pdf | Joint super-resolution and synthesis of 1 mm isotropic MP-RAGE volumes from clinical MRI exams with scans of different orientation, resolution and contrast | Most existing algorithms for automatic 3D morphometry of human brain MRI scans are designed for data with near-isotropic voxels at approximately 1 mm resolution, and frequently have contrast constraints as well - typically requiring T1 scans (e.g., MP-RAGE). This limitation prevents the analysis of millions of MRI scans acquired with large inter-slice spacing ("thick slice") in clinical settings every year. The inability to quantitatively analyze these scans hinders the adoption of quantitative neuroimaging in healthcare, and precludes research studies that could attain huge sample sizes and hence greatly improve our understanding of the human brain. Recent advances in CNNs are producing outstanding results in super-resolution and contrast synthesis of MRI. However, these approaches are very sensitive to the contrast, resolution and orientation of the input images, and thus do not generalize to diverse clinical acquisition protocols - even within sites. Here we present SynthSR, a method to train a CNN that receives one or more thick-slice scans with different contrast, resolution and orientation, and produces an isotropic scan of canonical contrast (typically a 1 mm MP-RAGE). The presented method does not require any preprocessing, e.g., skull stripping or bias field correction. Crucially, SynthSR trains on synthetic input images generated from 3D segmentations, and can thus be used to train CNNs for any combination of contrasts, resolutions and orientations without high-resolution training data. We test the images generated with SynthSR in an array of common downstream analyses, and show that they can be reliably used for subcortical segmentation and volumetry, image registration (e.g., for tensor-based morphometry), and, if some image quality requirements are met, even cortical thickness morphometry. The source code is publicly available at github.com/BBillot/SynthSR. | ['Bruce Fischl', 'Brian L. Edlow', 'Polina Golland', 'Daniel C. Alexander', 'John Conklin', 'Azadeh Tabari', 'Yael Balbastre', 'Benjamin Billot', 'Juan Eugenio Iglesias'] | 2020-12-24 | null | null | null | null | ['skull-stripping'] | ['medical'] | [ 3.69531572e-01 5.08610718e-02 1.26822650e-01 -5.06667852e-01
-6.83644056e-01 -4.52025145e-01 3.24676096e-01 2.45333359e-01
-7.68546343e-01 6.37482166e-01 -8.14137980e-02 -4.10075694e-01
-1.11800410e-01 -7.92918861e-01 -6.40072346e-01 -7.22280383e-01
-3.97956848e-01 7.26848781e-01 5.09990275e-01 6.64829761e-02
8.51534083e-02 6.83027267e-01 -1.04607725e+00 3.83751355e-02
6.38193190e-01 9.02789056e-01 3.82033527e-01 4.32351142e-01
1.01948686e-01 1.03615239e-01 -1.60018817e-01 -2.89449275e-01
3.71584415e-01 -4.00851846e-01 -9.37750399e-01 -7.70998895e-02
5.73622167e-01 -5.01324058e-01 3.30831073e-02 1.07640636e+00
6.10325694e-01 -2.95548961e-02 5.23262620e-01 -6.43407404e-01
-5.19814134e-01 7.00668693e-01 -6.86753690e-01 6.71937048e-01
-1.81494564e-01 2.54671007e-01 4.21185166e-01 -7.07921743e-01
7.57053912e-01 7.54027486e-01 5.91773510e-01 5.71004570e-01
-1.57743824e+00 -6.42822981e-01 -3.16118807e-01 -1.10755421e-01
-1.17963934e+00 -2.94317931e-01 3.71034563e-01 -7.38129795e-01
7.50275671e-01 2.62930423e-01 9.54500854e-01 8.50807130e-01
5.96589684e-01 1.02217376e-01 1.60419095e+00 3.42081971e-02
2.73339987e-01 -4.66793090e-01 -1.45931140e-01 5.21991014e-01
3.08469057e-01 -1.20245859e-01 -1.90970883e-01 1.58132352e-02
1.28079486e+00 -1.51835188e-01 -3.65038872e-01 -4.19190198e-01
-1.61964297e+00 6.36715293e-01 6.00025117e-01 6.57630563e-01
-3.75894517e-01 7.41825253e-03 4.65465844e-01 1.83238044e-01
3.51048261e-01 6.19044781e-01 -1.60178766e-01 2.73358941e-01
-1.31058657e+00 2.54309058e-01 1.16188014e-02 5.39863825e-01
5.00861883e-01 1.46948263e-01 1.75356254e-01 8.10100019e-01
-8.25876966e-02 3.40614527e-01 8.33919466e-01 -1.04030514e+00
2.47577503e-01 2.27311984e-01 -3.27288359e-01 -7.54130304e-01
-9.83123720e-01 -3.08133453e-01 -1.15281999e+00 3.69312853e-01
6.45842969e-01 -4.62496653e-02 -1.14786184e+00 1.68770409e+00
3.80641282e-01 -2.14793786e-01 -5.79949141e-01 1.33560085e+00
6.24430358e-01 -1.50720775e-01 6.07277825e-02 -2.24659383e-01
1.38843513e+00 -3.48324269e-01 -3.25392604e-01 -2.54221678e-01
6.45458579e-01 -5.76251507e-01 1.18513417e+00 2.67245114e-01
-1.44042695e+00 -4.82441336e-02 -1.01153612e+00 -2.72965096e-02
-1.61842212e-01 -4.61881369e-01 5.89707732e-01 5.87071896e-01
-1.36743045e+00 7.54473269e-01 -1.20196652e+00 -1.46211579e-01
7.53607154e-01 5.73551238e-01 -7.47041941e-01 -5.80735244e-02
-9.33541238e-01 1.06160903e+00 2.68507838e-01 1.07635826e-01
-6.93302095e-01 -1.14215088e+00 -6.34239197e-01 -1.93053454e-01
2.27566180e-03 -6.62910104e-01 1.03341627e+00 -8.15936983e-01
-1.20328629e+00 1.16027093e+00 2.67316788e-01 -5.16559064e-01
6.28880262e-01 3.81510735e-01 -1.50996819e-01 4.58397359e-01
1.64528310e-01 8.11222851e-01 5.97856641e-01 -8.86102796e-01
1.41769856e-01 -9.63520348e-01 -2.51307219e-01 9.09502581e-02
1.92933112e-01 3.37356657e-01 5.99079281e-02 -4.41317677e-01
6.86692536e-01 -6.97236180e-01 -5.61155200e-01 5.54896742e-02
-2.62092859e-01 7.27576613e-01 2.93546766e-01 -8.84614706e-01
4.63215977e-01 -1.85152340e+00 3.01200524e-02 3.35215002e-01
6.05097890e-01 -1.25078812e-01 -1.00034773e-02 -2.77695119e-01
-4.47174519e-01 3.39314848e-01 -7.76487768e-01 5.00716045e-02
-4.35494900e-01 -2.51995444e-01 3.20931911e-01 8.94550264e-01
-5.01447655e-02 9.73706126e-01 -8.15151095e-01 -5.50230563e-01
1.33666620e-01 5.61379313e-01 -5.89856386e-01 -6.83084130e-02
2.26286873e-01 1.13977790e+00 -3.47912580e-01 2.55208403e-01
7.07396328e-01 -1.48427054e-01 3.42150211e-01 -3.36380571e-01
-2.90107012e-01 2.72943139e-01 -7.13844061e-01 1.76485610e+00
-4.97915536e-01 4.70603496e-01 3.99541140e-01 -1.15442979e+00
6.92010999e-01 4.17722344e-01 8.98826659e-01 -1.03835917e+00
3.75618011e-01 5.45204520e-01 6.77341461e-01 -2.27329075e-01
-9.78915915e-02 -6.99305236e-01 3.38120311e-01 6.46078944e-01
1.08958766e-01 -5.55778980e-01 1.65528312e-01 -8.33874792e-02
1.00786173e+00 -2.98783779e-01 5.67561872e-02 -5.11314869e-01
2.15339750e-01 -9.63806659e-02 4.34357792e-01 3.80049020e-01
-1.79609716e-01 1.03136837e+00 6.19816303e-01 -5.76325297e-01
-1.61109912e+00 -1.20723343e+00 -6.18518889e-01 4.87088889e-01
-2.55017877e-01 1.99908555e-01 -1.06128788e+00 -3.79817992e-01
-2.61572778e-01 1.31321251e-01 -6.34342492e-01 7.58283064e-02
-9.03044283e-01 -1.13476360e+00 6.17545724e-01 3.64573330e-01
3.45205396e-01 -1.09623766e+00 -9.24725711e-01 3.14080089e-01
-5.17674424e-02 -1.16163743e+00 -3.25715989e-01 2.12261423e-01
-1.29719377e+00 -1.12756205e+00 -1.04234838e+00 -6.57494485e-01
8.29614282e-01 -3.67215462e-02 1.04873991e+00 2.98909515e-01
-4.82499331e-01 -4.00848091e-02 -2.59769764e-02 5.24977185e-02
-2.07026944e-01 5.99471368e-02 1.71920747e-01 -2.30479658e-01
-1.95104912e-01 -1.24782670e+00 -1.01353884e+00 2.42833704e-01
-1.00381303e+00 2.09210709e-01 6.93391562e-01 7.26172030e-01
9.23575938e-01 -2.86017209e-01 6.20824575e-01 -8.85091126e-01
5.50141752e-01 -3.21111947e-01 -5.96492946e-01 -9.36091878e-03
-5.21789610e-01 -1.07194796e-01 6.09669209e-01 -3.27155292e-01
-5.43921411e-01 -2.76457131e-01 -3.72925550e-01 -1.82002738e-01
-3.31322521e-01 4.15240794e-01 6.91662729e-02 -3.58946681e-01
7.65316606e-01 1.78677857e-01 2.60824829e-01 -3.32926065e-01
5.18442802e-02 1.84675485e-01 5.77800274e-01 -5.14463067e-01
5.97000301e-01 7.02606440e-01 2.12281957e-01 -6.25460625e-01
-2.96562284e-01 8.46031010e-02 -1.08866632e+00 -1.56099424e-01
1.11927140e+00 -3.93253416e-01 -5.25165737e-01 4.58721936e-01
-7.10136414e-01 -7.48182237e-01 -6.86321557e-02 8.15093935e-01
-6.30151212e-01 2.76714921e-01 -7.16588199e-01 -1.28165960e-01
-6.69510603e-01 -1.66945481e+00 6.54041529e-01 -4.42683557e-03
-2.46802419e-01 -8.77381265e-01 -2.17500627e-01 4.18309659e-01
7.62637973e-01 5.07165670e-01 1.27803957e+00 -3.66759926e-01
-3.31674248e-01 6.92276359e-02 -2.86174864e-01 2.17977971e-01
-1.90326031e-02 -2.83631444e-01 -6.02502465e-01 -1.34934187e-01
1.35715485e-01 -3.51041317e-01 4.94245797e-01 8.67612839e-01
1.50423002e+00 -8.16135705e-02 1.21809766e-01 8.44898701e-01
1.22112393e+00 1.46142751e-01 5.86357176e-01 3.44516426e-01
7.07556605e-01 7.57091343e-01 1.88176858e-03 -2.85057933e-03
1.81440130e-01 6.86108351e-01 3.01937282e-01 -3.33106726e-01
-3.03019226e-01 3.02644759e-01 -2.46906832e-01 8.93344283e-01
-4.17750537e-01 5.73989451e-01 -1.23011529e+00 5.61857879e-01
-9.87099111e-01 -6.83909655e-01 -2.90495634e-01 2.43418384e+00
1.07112813e+00 1.26184821e-01 2.19481066e-01 -5.29995337e-02
6.44866049e-01 9.45424810e-02 -7.92349875e-01 -2.85921693e-01
1.75932408e-04 6.14583969e-01 7.49685824e-01 5.22057414e-01
-8.02579641e-01 6.67383909e-01 6.28682804e+00 2.97490895e-01
-1.58246887e+00 5.63384116e-01 8.94124448e-01 -3.16622525e-01
-3.87597322e-01 -3.21221232e-01 -4.82506081e-02 2.67963439e-01
9.75504696e-01 -7.29742497e-02 7.20206141e-01 4.28290009e-01
4.06512648e-01 -2.13927150e-01 -1.00403583e+00 8.63877892e-01
-3.21014225e-01 -1.33566332e+00 -1.86508626e-01 1.88780829e-01
4.42266166e-01 4.56384182e-01 1.51173294e-01 -2.70830452e-01
-1.41926762e-02 -1.38705540e+00 6.44995570e-01 9.73861441e-02
1.26289451e+00 -5.06186843e-01 4.84868437e-01 5.50716706e-02
-7.63534546e-01 3.25906783e-01 -4.01092649e-01 4.31752533e-01
2.61346847e-01 7.68435180e-01 -7.16072679e-01 2.59019345e-01
8.84301126e-01 7.56349266e-02 -4.37943816e-01 1.10901213e+00
2.18842372e-01 3.41233850e-01 -2.87639260e-01 5.39798021e-01
1.63148597e-01 -4.73807961e-01 3.14631075e-01 9.55608308e-01
4.17961091e-01 2.13049874e-01 -2.09945247e-01 1.10433078e+00
-9.06910524e-02 3.48780423e-01 -3.86126697e-01 1.44577116e-01
1.77673250e-01 1.40629888e+00 -1.24037468e+00 -7.64285177e-02
-3.99605036e-01 4.78260577e-01 2.85038650e-01 2.67677397e-01
-4.96045023e-01 5.07816970e-02 4.00440395e-01 8.44552457e-01
-5.71612790e-02 -4.06519175e-01 -6.54129446e-01 -1.02797818e+00
1.49287343e-01 -9.63216364e-01 7.02918842e-02 -7.06188262e-01
-1.03258789e+00 8.59233916e-01 1.99169070e-01 -6.64961219e-01
-1.14690453e-01 -5.04734039e-01 -6.25513613e-01 1.03863144e+00
-1.26226521e+00 -7.46146798e-01 -1.60553023e-01 6.16136789e-01
1.30809769e-01 2.88497597e-01 7.15477467e-01 5.18516600e-01
-3.10406864e-01 2.88563073e-01 -2.21686184e-01 2.99879432e-01
5.68372488e-01 -1.23705840e+00 4.04649466e-01 6.41423762e-01
-1.69586435e-01 7.65582860e-01 4.85517472e-01 -6.47846401e-01
-1.13480270e+00 -9.31503952e-01 4.13943231e-01 -1.11277446e-01
8.27115297e-01 -4.68677521e-01 -1.09525478e+00 7.10156202e-01
-1.65742323e-01 4.06776994e-01 5.63213408e-01 -8.45927447e-02
-2.91788042e-01 2.53948141e-02 -1.43868744e+00 5.61235130e-01
9.23343301e-01 -2.42547452e-01 -3.90529096e-01 3.47612530e-01
3.83946478e-01 -6.49138749e-01 -1.40942395e+00 3.46186280e-01
6.67826116e-01 -1.12236416e+00 1.02671206e+00 -4.93287116e-01
5.12854457e-01 5.77594489e-02 2.24735081e-01 -1.39161193e+00
-2.35187635e-01 -1.32607901e-02 4.47121650e-01 5.62569022e-01
5.91305494e-01 -7.78882265e-01 6.67998791e-01 7.92867362e-01
-3.42254341e-01 -8.81430507e-01 -1.33551526e+00 -6.68035686e-01
6.85746491e-01 -3.03517580e-01 7.44736195e-01 1.01595986e+00
-1.69756249e-01 -1.17488839e-01 1.88427716e-01 -3.66967581e-02
8.00151527e-01 2.12441292e-02 1.86885014e-01 -1.18521726e+00
-8.43951032e-02 -7.22716451e-01 -4.44066077e-01 -3.34602535e-01
5.79321757e-02 -1.14262569e+00 -1.41763926e-01 -1.43875921e+00
2.14218348e-01 -6.88322723e-01 -8.08946863e-02 4.70202446e-01
2.38456979e-01 7.19273090e-01 -1.45927444e-01 2.37099737e-01
1.58816025e-01 2.57133693e-01 1.86269021e+00 -8.84468667e-03
-6.66617230e-02 -4.22568053e-01 -4.77982879e-01 7.80429959e-01
9.84700203e-01 -4.37482327e-01 -3.89752716e-01 -6.11919999e-01
3.85474674e-02 2.73837626e-01 5.63245058e-01 -1.00566149e+00
-1.09678000e-01 -5.87530397e-02 6.79264188e-01 -2.42410172e-02
1.24984846e-01 -5.83097279e-01 2.52111793e-01 3.75602156e-01
-2.44387224e-01 3.24601948e-01 -2.89545879e-02 -3.17164987e-01
9.22334716e-02 -6.15741909e-02 1.13089156e+00 -5.43893695e-01
5.43096922e-02 7.34868765e-01 -3.49191904e-01 1.79190814e-01
5.43517113e-01 -2.50016719e-01 -1.99196950e-01 -8.33449587e-02
-9.67375040e-01 -2.44118944e-01 7.23094404e-01 1.37844896e-02
6.00637138e-01 -1.24222851e+00 -7.97706664e-01 1.65043220e-01
-3.28816295e-01 2.18152851e-01 6.47202194e-01 1.69252062e+00
-8.03414524e-01 2.33561724e-01 -8.08555365e-01 -5.66539586e-01
-8.43670666e-01 3.10759813e-01 8.08342457e-01 -3.30732018e-01
-1.10674345e+00 4.54403490e-01 3.84546310e-01 -4.98291105e-01
-4.08132285e-01 -4.63025093e-01 -4.78754900e-02 -2.30571628e-01
6.19829893e-01 -2.18541011e-01 4.95843858e-01 -8.41266572e-01
-4.50209022e-01 5.58678746e-01 -1.44354075e-01 -3.54894042e-01
1.58448792e+00 -1.75401755e-02 -3.30898374e-01 2.43137896e-01
1.09776402e+00 -2.56155044e-01 -1.11242580e+00 4.36551608e-02
-1.69797078e-01 -4.10975009e-01 4.27962393e-01 -5.76868236e-01
-1.67038238e+00 1.06449056e+00 6.86188221e-01 7.93419927e-02
9.15960312e-01 -6.95670694e-02 8.33711386e-01 -1.60212830e-01
6.20758593e-01 -6.82628512e-01 -3.53976667e-01 2.06915393e-01
1.04065847e+00 -1.01661515e+00 -3.79669443e-02 -1.52553141e-01
-3.73015940e-01 9.46333468e-01 5.27154267e-01 -1.88642159e-01
5.90194345e-01 5.86185992e-01 1.46791279e-01 -6.08709633e-01
-2.02251941e-01 8.83574337e-02 2.66565561e-01 7.64315546e-01
7.87333608e-01 1.36440039e-01 -4.97238487e-01 3.09778690e-01
-5.86632133e-01 -2.42355257e-01 6.10747576e-01 6.17117822e-01
-2.26878420e-01 -1.01865184e+00 -4.00531262e-01 8.28987360e-01
-7.62355924e-01 -2.79859871e-01 -1.07960083e-01 7.85763621e-01
4.21188399e-02 2.90745616e-01 2.52750367e-01 -4.33435738e-02
1.81099102e-01 -6.16615964e-03 9.56668973e-01 -7.22732723e-01
-6.33827627e-01 1.09716043e-01 -2.20870718e-01 -6.91684544e-01
-3.49604785e-01 -8.07595670e-01 -1.46081972e+00 -4.24544930e-01
1.23390652e-01 -1.14330992e-01 7.79500484e-01 9.17333603e-01
1.87202543e-02 4.16431069e-01 2.40196243e-01 -1.01067591e+00
-2.13851016e-02 -9.61864293e-01 -8.09307456e-01 3.21282715e-01
1.38463601e-01 -7.47577071e-01 -1.25042856e-01 -1.27986491e-01] | [14.081799507141113, -2.314619541168213] |
245461cb-d548-4b2f-a05d-868d09023311 | coupling-top-down-and-bottom-up-methods-for | 1410.0117 | null | http://arxiv.org/abs/1410.0117v2 | http://arxiv.org/pdf/1410.0117v2.pdf | Coupling Top-down and Bottom-up Methods for 3D Human Pose and Shape Estimation from Monocular Image Sequences | Until recently Intelligence, Surveillance, and Reconnaissance (ISR) focused
on acquiring behavioral information of the targets and their activities.
Continuous evolution of intelligence being gathered of the human centric
activities has put increased focus on the humans, especially inferring their
innate characteristics - size, shapes and physiology. These bio-signatures
extracted from the surveillance sensors can be used to deduce age, ethnicity,
gender and actions, and further characterize human actions in unseen scenarios.
However, recovery of pose and shape of humans in such monocular videos is
inherently an ill-posed problem, marked by frequent depth and view based
ambiguities due to self-occlusion, foreshortening and misalignment. The
likelihood function often yields a highly multimodal posterior that is
difficult to propagate even using the most advanced particle filtering(PF)
algorithms. Motivated by the recent success of the discriminative approaches to
efficiently predict 3D poses directly from the 2D images, we present several
principled approaches to integrate predictive cues using learned regression
models to sustain multimodality of the posterior during tracking. Additionally,
these learned priors can be actively adapted to the test data using a
likelihood based feedback mechanism. Estimated 3D poses are then used to fit 3D
human shape model to each frame independently for inferring anthropometric
bio-signatures. The proposed system is fully automated, robust to noisy test
data and has ability to swiftly recover from tracking failures even after
confronting with significant errors. We evaluate the system on a large number
of monocular human motion sequences. | ['Atul Kanaujia'] | 2014-10-01 | null | null | null | null | ['3d-human-pose-and-shape-estimation'] | ['computer-vision'] | [ 4.73156840e-01 -1.46265507e-01 3.97998579e-02 -2.78232276e-01
-3.36070150e-01 -6.18225992e-01 5.65327227e-01 -1.00588202e-01
-6.01961851e-01 9.86181021e-01 1.32063344e-01 4.35353905e-01
-2.45653003e-01 -2.69606471e-01 -7.27882087e-01 -7.44605660e-01
-2.66595066e-01 7.64481604e-01 2.22388133e-01 1.84983194e-01
1.91665545e-01 6.75623596e-01 -1.90119243e+00 -1.65230855e-01
4.37898189e-01 9.05104399e-01 1.14829227e-01 1.19372928e+00
5.98993838e-01 2.37100214e-01 -5.90973437e-01 -4.76628482e-01
4.00691897e-01 -1.61767766e-01 2.19937935e-02 4.07925576e-01
6.22687578e-01 -6.88501179e-01 -1.72262222e-01 9.45758045e-01
4.34443772e-01 8.06727782e-02 6.54471219e-01 -1.19170570e+00
2.62476574e-03 -2.39905953e-01 -5.09090662e-01 2.17311278e-01
9.91158426e-01 3.21813852e-01 2.51877338e-01 -5.37044704e-01
7.11162508e-01 1.32556736e+00 8.76433730e-01 5.63940465e-01
-1.08367002e+00 -5.08754909e-01 -1.22145981e-01 2.43089288e-01
-1.22673619e+00 -4.40944225e-01 7.65954196e-01 -6.57965720e-01
3.38224113e-01 4.96664137e-01 1.14048743e+00 1.56170988e+00
2.32970327e-01 4.43368316e-01 1.11078608e+00 -1.02776036e-01
3.07577383e-02 7.02881813e-02 -2.11075649e-01 8.78897727e-01
7.28346467e-01 4.61044461e-01 -6.60064340e-01 -3.12619716e-01
7.65655875e-01 2.51033723e-01 -3.41059476e-01 -4.66543078e-01
-1.30613685e+00 3.75054866e-01 -1.11304514e-01 -1.75815582e-01
-6.12511754e-01 -2.46032812e-02 9.33243951e-04 -1.67783082e-01
3.09536397e-01 1.66537404e-01 -5.19277751e-01 -2.51335293e-01
-8.85548592e-01 4.64649141e-01 6.61584854e-01 8.31638038e-01
4.60670978e-01 -3.64058826e-04 -1.94602087e-01 3.74469817e-01
5.96971810e-01 1.19008505e+00 2.64471591e-01 -8.77809882e-01
2.57778078e-01 6.43930912e-01 5.35655499e-01 -1.41702676e+00
-6.45934105e-01 -4.02078718e-01 -5.51163137e-01 1.54776886e-01
6.58163130e-01 -2.54667044e-01 -7.24368453e-01 1.62315953e+00
9.96460974e-01 1.00063197e-01 -3.41347545e-01 1.19097233e+00
5.10331929e-01 1.54917017e-01 -4.92931064e-03 -3.94630522e-01
1.43441367e+00 -1.28271937e-01 -4.01716977e-01 -4.53432322e-01
-1.39266476e-01 -7.10297525e-01 1.42932355e-01 6.16966426e-01
-6.55181289e-01 -6.91799700e-01 -9.96759415e-01 5.21161377e-01
-9.67325196e-02 3.01272303e-01 4.14561599e-01 9.80619431e-01
-6.10270560e-01 4.77782428e-01 -1.11094570e+00 -7.21379697e-01
2.08725810e-01 5.90792894e-01 -6.08937919e-01 -5.55931106e-02
-7.63168633e-01 8.78184736e-01 2.81558573e-01 3.69772285e-01
-1.07694614e+00 -4.59754646e-01 -9.03868437e-01 -7.08321691e-01
5.84846795e-01 -7.81749487e-01 5.16163945e-01 -6.85922265e-01
-1.45537949e+00 8.52269948e-01 1.34601230e-02 -5.05276799e-01
7.57042527e-01 -5.67559183e-01 -4.04491544e-01 3.40328902e-01
-1.75989985e-01 7.65770793e-01 1.23693740e+00 -1.05644739e+00
-4.70434427e-01 -9.63974297e-01 -3.06545824e-01 3.87550384e-01
8.14708546e-02 -1.23313114e-01 -2.68260688e-01 -5.22828698e-01
2.15401426e-01 -1.02343893e+00 -1.66747905e-02 3.73531938e-01
-2.70127624e-01 2.73831189e-01 6.85319722e-01 -1.06614447e+00
6.99904621e-01 -1.68167520e+00 4.37293857e-01 -1.47870639e-02
-2.87653436e-03 2.76773721e-01 3.01223606e-01 1.59721062e-01
5.11765420e-01 -6.19323790e-01 -1.46071911e-01 -4.56748568e-02
-7.97187313e-02 1.78073749e-01 5.85727580e-02 1.08003247e+00
1.64041817e-02 6.61035120e-01 -7.58583665e-01 -6.94261968e-01
4.49314892e-01 7.30414152e-01 -4.98244941e-01 5.70266902e-01
-1.69882044e-01 1.01842284e+00 -5.18492103e-01 1.02399457e+00
5.37512898e-01 3.14952761e-01 1.18417054e-01 -4.73167717e-01
-6.95946664e-02 -6.20507121e-01 -1.20120966e+00 1.57162356e+00
3.09160918e-01 3.18637997e-01 4.25839186e-01 -9.92480278e-01
9.11303997e-01 2.39759654e-01 7.41190553e-01 -1.60052314e-01
3.82892340e-01 -1.03730276e-01 -1.85984582e-01 -8.89069200e-01
2.70078480e-01 -6.06687814e-02 1.28383227e-02 -5.75161092e-02
3.67074534e-02 9.69985351e-02 -7.88593590e-02 -4.38844919e-01
8.09276104e-01 8.05505276e-01 4.06634569e-01 1.40575077e-02
6.50200963e-01 7.59665668e-02 6.95651650e-01 4.93026286e-01
-3.72492611e-01 4.86117542e-01 -1.38311923e-01 -5.78021765e-01
-9.16997910e-01 -1.17052877e+00 -1.01041816e-01 7.66323626e-01
1.60879105e-01 1.79337978e-01 -5.11697471e-01 -3.70239258e-01
1.29503384e-01 1.79846376e-01 -6.98599637e-01 -6.20100163e-02
-5.55353940e-01 -9.45061445e-01 4.48655307e-01 1.71819329e-01
3.37664425e-01 -6.36842430e-01 -1.31474268e+00 2.02711210e-01
-7.31944963e-02 -1.27327883e+00 -1.52571008e-01 -3.87989402e-01
-8.41961980e-01 -1.41370463e+00 -8.16577852e-01 -6.84540197e-02
7.14473486e-01 -1.56055734e-01 7.11482823e-01 -6.00819141e-02
-6.63018644e-01 9.57368970e-01 -2.45039821e-01 -3.54114026e-01
-8.17818567e-02 -6.24653399e-01 5.67629099e-01 5.02502918e-01
3.47712159e-01 -5.34655273e-01 -7.37395763e-01 6.27414167e-01
-4.57715958e-01 -8.70967805e-02 4.79042470e-01 5.32891154e-01
5.09041309e-01 -1.50380865e-01 1.43029198e-01 -3.88712108e-01
8.31408724e-02 -3.75387728e-01 -8.64903569e-01 1.97856367e-01
-7.61111677e-02 -2.00401962e-01 1.94564521e-01 -7.86013782e-01
-1.01285875e+00 4.80893970e-01 2.74830371e-01 -4.23415810e-01
-7.26441622e-01 -4.91220616e-02 -1.86511949e-01 -1.19970933e-01
5.24608254e-01 3.56479377e-01 2.15933323e-01 -3.73658419e-01
5.01071215e-02 5.08511901e-01 1.05727148e+00 -6.29592717e-01
9.19492006e-01 8.09140801e-01 3.66987735e-01 -1.30901134e+00
-7.49998391e-01 -3.08685571e-01 -1.01586211e+00 -9.95202243e-01
1.18397725e+00 -9.13606346e-01 -1.07400322e+00 5.06580174e-01
-9.63660479e-01 2.54078209e-01 2.56606817e-01 9.09483433e-01
-5.16888857e-01 7.30400622e-01 -1.13610558e-01 -1.48768616e+00
-1.82436019e-01 -9.47250366e-01 1.27619278e+00 3.79201561e-01
-2.69333869e-01 -9.16482925e-01 1.36345655e-01 8.35638106e-01
8.78048316e-02 8.91245306e-01 1.19882390e-01 -4.87480879e-01
-6.19550467e-01 -5.26212513e-01 2.75870740e-01 -6.77518472e-02
-1.22249462e-01 -1.50959775e-01 -8.17323267e-01 -4.15361911e-01
1.40317872e-01 -3.01124841e-01 2.94301271e-01 5.79036236e-01
3.30242425e-01 -2.12920412e-01 -4.47396487e-01 6.50636077e-01
1.01528406e+00 1.32651493e-01 1.47937611e-01 -1.04240641e-01
6.12805963e-01 8.72683585e-01 8.14085007e-01 8.66665840e-01
2.47722983e-01 8.29148233e-01 5.87802529e-01 5.42419434e-01
2.67572980e-02 -2.62959242e-01 6.44570589e-01 3.38482410e-01
-5.03701687e-01 4.79679219e-02 -6.33154333e-01 1.51638642e-01
-1.66278470e+00 -1.07267082e+00 -1.37333855e-01 2.53684592e+00
4.12777543e-01 1.24734953e-01 2.93537676e-01 3.30443718e-02
7.76011705e-01 -2.07992837e-01 -6.19533896e-01 4.41911221e-01
6.94664009e-03 -3.15823674e-01 7.11462617e-01 3.97655189e-01
-1.17990911e+00 4.65962052e-01 6.40489101e+00 2.60567993e-01
-7.75857389e-01 -1.80522665e-01 1.20923601e-01 -1.77902862e-01
2.08831862e-01 -2.62040496e-01 -1.00153911e+00 4.61486101e-01
6.30728006e-01 3.11809063e-01 3.28616112e-01 4.36536402e-01
1.53986588e-01 -6.95007205e-01 -1.02563846e+00 1.21843076e+00
4.93640393e-01 -7.01684475e-01 -3.79084527e-01 1.59735978e-01
3.76093447e-01 -2.76371628e-01 -5.93865253e-02 -5.62751070e-02
-8.71091802e-03 -7.79784203e-01 6.68924868e-01 1.17645526e+00
5.37822247e-01 -5.42091370e-01 6.13637388e-01 7.98737109e-01
-1.02844703e+00 -9.44643617e-02 -4.73591626e-01 -1.72261894e-01
3.09712738e-01 3.82063836e-01 -1.03218424e+00 4.93241876e-01
5.46027839e-01 5.29232204e-01 -5.96046269e-01 1.12729800e+00
-2.86772419e-02 3.41058373e-01 -6.43591642e-01 -9.87073183e-02
-4.13883358e-01 -3.04554462e-01 1.11620927e+00 8.14020216e-01
4.45319980e-01 2.16841564e-01 3.99002254e-01 6.52787149e-01
6.92582726e-01 -2.81982005e-01 -5.44345319e-01 7.76528195e-03
2.13015825e-01 1.27788985e+00 -6.50879204e-01 -6.24605380e-02
-1.69379890e-01 8.16367686e-01 -1.03626223e-02 1.40385970e-01
-9.28187847e-01 4.30743366e-01 5.29781401e-01 2.19413072e-01
2.69505680e-01 -4.86864001e-01 1.48150831e-01 -1.18032384e+00
-5.34030795e-02 -7.02683210e-01 5.13778508e-01 -7.18630612e-01
-1.10300589e+00 5.23386538e-01 5.79130054e-01 -1.23910999e+00
-3.82785290e-01 -6.39022648e-01 -4.06216606e-02 4.97063816e-01
-8.06057453e-01 -1.28332341e+00 -4.12345827e-01 6.07904971e-01
4.80132341e-01 -3.35024089e-01 5.84363222e-01 2.19645888e-01
-5.00293911e-01 1.46040395e-01 -4.87178952e-01 -1.07788049e-01
7.08950520e-01 -8.26088428e-01 -3.52499098e-01 8.49877536e-01
5.73827364e-02 2.59602189e-01 1.32725930e+00 -1.28933215e+00
-1.95190752e+00 -7.19309449e-01 3.88665557e-01 -9.03485298e-01
4.08530623e-01 -3.65183532e-01 -3.51879835e-01 5.23310363e-01
-3.66078764e-01 -2.36805156e-02 6.75123274e-01 -2.15959579e-01
1.99769497e-01 -1.54105291e-01 -1.26613379e+00 2.33485922e-01
1.07802343e+00 4.66599464e-02 -8.55859578e-01 2.43264288e-01
9.58494842e-02 -4.92544353e-01 -8.20583701e-01 5.72943747e-01
9.68050659e-01 -1.14568281e+00 1.18336260e+00 -5.51062465e-01
-3.76042366e-01 -4.10024464e-01 -3.81643057e-01 -7.44926572e-01
-7.49999732e-02 -6.32315040e-01 -1.63607806e-01 1.02003372e+00
-1.73128501e-01 -2.43785650e-01 1.00367761e+00 5.85327387e-01
2.59168953e-01 -3.28402668e-01 -1.15625000e+00 -5.53889930e-01
-7.93591857e-01 -4.16692406e-01 1.15788117e-01 4.41420943e-01
-3.55976552e-01 -7.42155239e-02 -1.07953680e+00 6.62137330e-01
1.32692540e+00 6.86183944e-02 1.15697718e+00 -1.61087263e+00
-5.08086145e-01 2.12913945e-01 -7.97714651e-01 -8.98594439e-01
-8.68477523e-02 -3.70019078e-01 8.71467367e-02 -9.25833464e-01
1.32562280e-01 1.42768472e-01 3.14850688e-01 -1.90818193e-03
-1.51768699e-02 3.63425612e-01 1.02800004e-01 4.68687434e-03
-5.72238684e-01 4.81376857e-01 1.06146026e+00 1.32532999e-01
7.39022568e-02 2.94105738e-01 8.49371124e-03 9.48777556e-01
3.98544669e-01 -5.11176348e-01 -2.69336283e-01 -7.10623860e-02
2.97928125e-01 6.26002371e-01 8.55318189e-01 -1.51264465e+00
3.59011203e-01 -2.79926509e-01 1.01351094e+00 -9.21678901e-01
7.59537160e-01 -1.12602663e+00 8.09998989e-01 6.23095214e-01
1.02187440e-01 2.11744681e-02 -2.97026858e-02 1.00002158e+00
2.02154621e-01 -4.84706089e-02 6.19946063e-01 -2.69213706e-01
-4.31192577e-01 4.05099332e-01 -3.32751274e-01 -7.57960007e-02
1.07326579e+00 -6.18876219e-01 1.55819342e-01 -3.27651054e-01
-8.54273081e-01 3.75230014e-02 4.62729633e-01 3.05477202e-01
5.91290712e-01 -1.18454099e+00 -7.60816574e-01 5.52937746e-01
-1.81331053e-01 -2.91205794e-01 4.85555023e-01 1.16290057e+00
-5.90611279e-01 2.99724579e-01 -6.75965130e-01 -8.73686314e-01
-1.62026024e+00 5.94383359e-01 2.10316271e-01 7.11141452e-02
-4.38148499e-01 5.80643713e-01 -1.44875467e-01 -1.40019372e-01
3.09641004e-01 5.46290400e-03 -3.30599934e-01 1.12203546e-01
5.49507320e-01 6.86981916e-01 -5.15304625e-01 -1.04829514e+00
-6.21367872e-01 9.21153128e-01 4.83817607e-01 -2.21350104e-01
1.28865910e+00 -3.18586946e-01 3.01900148e-01 2.57245183e-01
5.30131042e-01 3.99048664e-02 -1.87152302e+00 9.53436643e-02
-1.59082338e-01 -6.76007748e-01 -4.16591465e-01 -6.71060205e-01
-8.00824702e-01 6.53663814e-01 6.17821276e-01 -1.01618372e-01
8.78875732e-01 -1.26413822e-01 4.11505610e-01 1.55426323e-01
6.44136846e-01 -1.04567814e+00 -6.33684993e-02 1.78412441e-02
8.52733791e-01 -1.20867348e+00 4.80796903e-01 -2.81935185e-01
-4.85616922e-01 1.08889461e+00 5.36036193e-01 -2.34848689e-02
5.30690610e-01 1.12420246e-01 -1.03355795e-01 -1.09262772e-01
-4.95971829e-01 -2.84218848e-01 7.47225225e-01 1.15457320e+00
2.68702861e-02 1.14596039e-01 -5.39512895e-02 4.38454539e-01
-2.27859169e-01 -1.75749645e-01 1.59097195e-01 7.97999144e-01
-6.67438984e-01 -7.95313120e-01 -1.17664301e+00 1.20127566e-01
-4.10585165e-01 6.01437271e-01 -4.25066233e-01 7.34291017e-01
4.07094032e-01 8.42488110e-01 -1.38031110e-01 -3.72846276e-01
3.02346230e-01 1.14217974e-01 9.19609666e-01 -1.60820559e-01
-2.55623132e-01 3.69542897e-01 1.59206361e-01 -5.91810942e-01
-7.78791070e-01 -1.09683025e+00 -7.13530600e-01 9.91968736e-02
-7.60812014e-02 -2.35889077e-01 6.83878362e-01 1.04252660e+00
7.77426809e-02 -1.09921340e-02 1.25006452e-01 -1.27524197e+00
-5.60545921e-01 -9.08506274e-01 -4.51425791e-01 5.64188600e-01
3.16130072e-01 -1.18321526e+00 -2.40461737e-01 2.43051529e-01] | [7.051532745361328, -0.9775857925415039] |
f6c76600-b09d-42f9-aa60-5ac2f1789334 | on-the-uncertainty-analysis-of-the-data | 2303.08455 | null | https://arxiv.org/abs/2303.08455v3 | https://arxiv.org/pdf/2303.08455v3.pdf | On the uncertainty analysis of the data-enabled physics-informed neural network for solving neutron diffusion eigenvalue problem | In practical engineering experiments, the data obtained through detectors are inevitably noisy. For the already proposed data-enabled physics-informed neural network (DEPINN) \citep{DEPINN}, we investigate the performance of DEPINN in calculating the neutron diffusion eigenvalue problem from several perspectives when the prior data contain different scales of noise. Further, in order to reduce the effect of noise and improve the utilization of the noisy prior data, we propose innovative interval loss functions and give some rigorous mathematical proofs. The robustness of DEPINN is examined on two typical benchmark problems through a large number of numerical results, and the effectiveness of the proposed interval loss function is demonstrated by comparison. This paper confirms the feasibility of the improved DEPINN for practical engineering applications in nuclear reactor physics. | ['Shiquan Zhang', 'Qiaolin He', 'Yangtao Deng', 'Qihong Yang', 'Helin Gong', 'Yu Yang'] | 2023-03-15 | null | null | null | null | ['mathematical-proofs'] | ['miscellaneous'] | [ 1.26696736e-01 6.14068024e-02 6.57499731e-02 -3.36287290e-01
-5.34656286e-01 -3.83563899e-02 2.61715800e-01 9.51581597e-02
-8.37524116e-01 1.23529792e+00 -4.11596857e-02 -2.69254714e-01
-1.00845122e+00 -7.18900204e-01 -6.51363373e-01 -1.22339320e+00
-8.48682690e-03 3.33047181e-01 -5.11700884e-02 4.26855031e-03
1.29979134e-01 8.42891514e-01 -8.25459778e-01 -3.12907875e-01
6.15937352e-01 1.13991928e+00 6.70111179e-02 3.20805132e-01
1.82273641e-01 3.71242732e-01 -6.30951524e-01 9.01467949e-02
5.01866460e-01 -9.17830244e-02 -2.58768529e-01 -3.13081503e-01
-5.60508668e-01 -3.49296480e-01 -8.73805523e-01 1.62603211e+00
1.13008177e+00 6.40612364e-01 7.31973886e-01 -9.06291664e-01
-6.54145300e-01 9.22362268e-01 -5.63025653e-01 5.24354756e-01
-4.73621488e-01 3.49715322e-01 2.70201027e-01 -8.84366691e-01
3.80804032e-01 9.80857015e-01 8.36111605e-01 4.38984096e-01
-7.39168525e-01 -6.12870753e-01 -3.97488289e-02 1.30366549e-01
-1.46619701e+00 -1.74645081e-01 1.00087726e+00 -3.83460373e-01
5.48142493e-01 -4.53584231e-02 3.05316746e-01 1.07155669e+00
4.57999527e-01 3.46627980e-01 9.88216639e-01 -3.53135139e-01
5.45109689e-01 4.01149690e-03 6.85954213e-01 4.54923451e-01
7.35011101e-01 7.90805101e-01 -1.95755228e-01 -2.67523319e-01
1.00308061e+00 1.06967278e-01 -6.25758827e-01 8.38353261e-02
-1.26414323e+00 7.24558592e-01 4.25388724e-01 4.61381614e-01
-3.99292767e-01 8.83738250e-02 7.05752313e-01 6.71070069e-03
5.73229849e-01 3.49040359e-01 -2.89216220e-01 2.29204461e-01
-6.90015495e-01 1.59781262e-01 7.32778311e-01 1.13823962e+00
2.75863588e-01 4.53399599e-01 -4.30775911e-01 4.43503290e-01
3.01314235e-01 6.66528344e-01 2.39371106e-01 -7.06816375e-01
2.80914277e-01 2.87579782e-02 8.14654380e-02 -8.94256175e-01
-7.14985371e-01 -7.43227601e-01 -1.61600041e+00 2.00556144e-01
3.92848700e-01 -6.05163753e-01 -8.58189404e-01 1.43428147e+00
2.28305936e-01 5.20146266e-02 2.50454009e-01 1.04193211e+00
8.40893090e-01 9.21811700e-01 4.81796227e-02 -4.60008353e-01
8.59114587e-01 -3.52444768e-01 -8.97291481e-01 2.14503020e-01
4.44874287e-01 -1.36324301e-01 4.55299616e-01 5.59310734e-01
-9.93654430e-01 -4.98564690e-01 -1.29930365e+00 3.76531601e-01
-1.65722042e-01 1.88313261e-01 5.35931349e-01 6.14912271e-01
-4.56320733e-01 1.12991357e+00 -7.82284081e-01 1.65494040e-01
5.30745983e-01 3.73037159e-01 -1.94722712e-02 2.68509630e-02
-1.54097080e+00 5.59087455e-01 1.08349550e+00 7.73349702e-01
-8.91021907e-01 -7.72125423e-01 -4.91950780e-01 -1.30365387e-01
3.73924017e-01 -5.68050742e-01 9.90476906e-01 -1.05458103e-01
-1.51034641e+00 2.74986476e-02 6.19470119e-01 -4.04517621e-01
7.11797595e-01 4.48375791e-02 -4.04955983e-01 1.26107439e-01
-3.79008621e-01 1.03360722e-02 3.72711897e-01 -9.17189240e-01
-4.19050977e-02 -2.09632248e-01 -4.83121574e-02 -9.66802016e-02
-3.31681341e-01 -1.08110614e-01 -9.27299336e-02 -6.19670391e-01
2.64193676e-02 -5.33663154e-01 -8.38837147e-01 6.33370727e-02
-6.54299378e-01 -3.04409355e-01 6.30544662e-01 -6.08235598e-01
7.83323288e-01 -2.15026498e+00 -2.57121189e-03 5.24273038e-01
2.00230211e-01 2.58818358e-01 2.29950905e-01 2.41988882e-01
-2.32662022e-01 -7.91746601e-02 -6.84953570e-01 2.00685918e-01
1.50383562e-01 2.87546158e-01 9.90312248e-02 7.59933174e-01
-1.20427154e-01 6.52274787e-01 -6.89947367e-01 -2.61576295e-01
1.13719366e-01 3.64065021e-01 -4.15725522e-02 -1.94239452e-01
-1.24861620e-01 6.48557067e-01 -8.94282103e-01 4.61334616e-01
1.04130340e+00 -6.43562973e-02 -2.04653248e-01 -4.39065665e-01
3.96600850e-02 -6.67746663e-01 -1.25414979e+00 1.65934575e+00
-2.06742525e-01 4.26380783e-01 2.94412702e-01 -1.27040935e+00
1.04033375e+00 2.87466437e-01 5.80931842e-01 -7.31849790e-01
8.10980558e-01 1.45644292e-01 4.54100817e-02 -5.19189358e-01
2.88354754e-01 -4.09143031e-01 -6.09584153e-02 1.51287422e-01
-1.27533630e-01 1.51844293e-01 -1.54164601e-02 -1.00497780e-02
1.12237906e+00 -3.43239397e-01 1.03296220e-01 -8.09006572e-01
5.26844203e-01 -1.63624987e-01 7.90200710e-01 9.92548823e-01
-4.98351455e-01 2.29193971e-01 4.87470143e-02 -3.87422889e-01
-1.27351356e+00 -8.89800310e-01 -4.45816547e-01 2.10712552e-01
2.66121805e-01 5.15335262e-01 -6.01826012e-01 -7.02291548e-01
4.01167125e-02 8.93075764e-01 -3.30756545e-01 -3.30262989e-01
-5.04637241e-01 -1.46547580e+00 6.94088399e-01 9.66775358e-01
6.55185223e-01 -7.49150515e-01 5.98663390e-02 3.14445436e-01
1.47310376e-01 -1.10900342e+00 -9.86860171e-02 3.71415317e-01
-8.55140924e-01 -9.26161230e-01 -9.14779186e-01 -2.84287244e-01
6.58365309e-01 -2.68244207e-01 7.74694443e-01 -7.85814598e-02
-3.86883199e-01 2.75744766e-01 -3.51495855e-02 -6.72559857e-01
-5.04834652e-01 -1.70324460e-01 4.09239143e-01 -4.79505062e-01
1.75115317e-01 -6.63986325e-01 -6.63913131e-01 1.90995097e-01
-1.04769444e+00 -5.83276868e-01 6.81055427e-01 9.44675446e-01
5.14865398e-01 7.95025826e-01 8.75876009e-01 -6.19258523e-01
8.89079154e-01 -5.54615498e-01 -9.40954566e-01 2.68607447e-03
-4.51764286e-01 1.13404214e-01 6.07300699e-01 -5.02971053e-01
-1.22707450e+00 -2.89272159e-01 -3.19135457e-01 -4.30850059e-01
-6.80365711e-02 6.90837801e-01 -2.32178852e-01 -1.54692933e-01
8.17605019e-01 9.41272359e-03 -2.50690371e-01 -5.25833130e-01
1.01815231e-01 6.14483953e-01 6.61669075e-01 -9.47630525e-01
7.01956630e-01 2.39063665e-01 4.15176153e-01 -9.79533374e-01
-6.46672070e-01 -2.58526772e-01 -3.48045260e-01 -2.74299294e-01
7.26531088e-01 -7.21267700e-01 -1.09249866e+00 3.59878182e-01
-1.36574614e+00 2.33547181e-01 -6.04675412e-01 1.20658588e+00
-2.85307407e-01 6.29433274e-01 -7.29590714e-01 -1.15643108e+00
-5.27268529e-01 -1.24641871e+00 2.91621774e-01 2.56299645e-01
5.44910431e-01 -1.05730259e+00 -1.67708680e-01 -2.81218767e-01
3.02637130e-01 6.07631803e-01 1.00982141e+00 -8.90549779e-01
-4.80384171e-01 -4.12669271e-01 -4.88998562e-01 7.05573499e-01
-2.60580748e-01 -2.45526463e-01 -8.30904901e-01 -4.51871485e-01
6.49523556e-01 -3.77381146e-02 9.29257035e-01 6.38099432e-01
1.40561187e+00 5.25469817e-02 -5.00919342e-01 3.82979631e-01
1.63646245e+00 6.98068976e-01 4.41340417e-01 8.85318667e-02
4.09404248e-01 1.79642811e-01 2.54599363e-01 6.44410551e-01
-5.53074718e-01 -4.42071185e-02 5.10597885e-01 -2.99038664e-02
2.47840390e-01 4.94527280e-01 -8.33830088e-02 1.01963770e+00
-1.78426638e-01 -5.78384399e-01 -5.99565744e-01 4.89356399e-01
-1.85320151e+00 -7.21546173e-01 -2.39146858e-01 1.98698676e+00
6.71894968e-01 2.38900945e-01 -3.74648899e-01 1.90304294e-01
1.01539648e+00 -1.01914428e-01 -1.05110121e+00 -2.04119146e-01
-1.29246429e-01 6.25571236e-02 9.32053089e-01 1.55309945e-01
-9.14501309e-01 -1.96755528e-01 6.46837091e+00 1.33606088e+00
-5.17629087e-01 2.59691119e-01 7.25248337e-01 3.81942652e-02
-8.62802491e-02 -4.09431696e-01 -8.34220707e-01 5.35512686e-01
9.10763919e-01 -3.27084154e-01 1.08364828e-01 9.05254662e-01
3.27497035e-01 -1.15827784e-01 -9.71440732e-01 7.93731093e-01
-4.35079426e-01 -1.32806599e+00 -2.03058511e-01 -3.63970622e-02
7.08365917e-01 1.30016450e-02 -1.49336293e-01 1.89655498e-01
3.56390715e-01 -7.79237211e-01 3.06039900e-01 1.14897251e+00
4.74254876e-01 -1.07013452e+00 1.14715934e+00 5.45936525e-01
-5.85553944e-01 -8.53877887e-02 -8.67255211e-01 1.19649127e-01
2.31345430e-01 1.61417449e+00 -5.23647308e-01 1.15591371e+00
2.81851977e-01 2.85894603e-01 2.32502863e-01 1.20538592e+00
1.97436854e-01 5.46989918e-01 -7.19673514e-01 -2.36345828e-01
3.14969480e-01 -1.53764948e-01 9.05825138e-01 1.04786837e+00
4.58454013e-01 3.56573522e-01 1.32125556e-01 1.15836561e+00
-1.85929358e-01 -2.29018986e-01 -5.46550453e-01 -1.46190524e-01
3.88491452e-01 9.74427462e-01 -6.33314312e-01 -4.37055714e-02
8.40352289e-03 2.77932376e-01 -9.21563804e-03 5.95871270e-01
-1.01119864e+00 -6.72277510e-01 1.72666743e-01 -3.80373150e-01
-1.02494009e-01 -5.17103970e-01 -2.88280100e-01 -6.34056568e-01
3.32053495e-03 -2.45614290e-01 2.10087553e-01 -5.66869497e-01
-1.78603542e+00 6.52948260e-01 2.90905714e-01 -8.99771094e-01
2.72721291e-01 -1.02738047e+00 -5.44061780e-01 9.28724706e-01
-1.27961481e+00 -3.62987339e-01 -8.05077851e-02 2.78611481e-01
1.92260072e-01 -3.83135736e-01 3.12461376e-01 5.07138729e-01
-9.02302265e-01 3.16645354e-01 9.02341783e-01 3.09595853e-01
1.41569123e-01 -7.96080768e-01 2.35214010e-01 8.13524425e-01
-5.45382559e-01 3.72037590e-01 8.30117822e-01 -8.46980870e-01
-1.32203805e+00 -1.10725296e+00 -1.26554400e-01 2.62572467e-01
5.75721323e-01 -5.32909706e-02 -7.55277336e-01 3.42388660e-01
2.95848161e-01 1.98197380e-01 2.34182969e-01 -4.21616703e-01
3.44845921e-01 2.59559602e-02 -1.64970791e+00 1.36738837e-01
9.17135000e-01 -8.68786350e-02 -4.01647985e-01 7.23578334e-01
7.98561275e-01 -4.22393739e-01 -1.17376471e+00 8.44259620e-01
1.10168442e-01 -6.09279275e-01 1.18491375e+00 -4.53975499e-01
2.16314301e-01 -2.03198418e-01 -2.05936402e-01 -1.30860317e+00
-2.76660830e-01 -4.50607538e-01 8.20508376e-02 1.00016749e+00
1.44900680e-01 -7.55709469e-01 7.86933720e-01 7.04862416e-01
-2.17601702e-01 -8.51884902e-01 -1.25612497e+00 -1.22506154e+00
2.32705310e-01 -6.39748573e-01 5.32469094e-01 7.73293197e-01
-2.59695977e-01 -2.85881162e-01 -1.74769610e-01 4.78561550e-01
9.22553539e-01 -6.14870369e-01 -7.11205229e-02 -1.17089200e+00
-2.55026639e-01 -1.20886132e-01 -4.29329157e-01 -8.26141477e-01
-3.43030691e-02 -7.78205037e-01 3.02050442e-01 -1.35923517e+00
6.94538802e-02 -4.97389257e-01 -7.48102248e-01 -5.46292868e-03
7.78015926e-02 -5.30121066e-02 -1.55777022e-01 5.94285131e-02
-1.18025564e-01 9.23076451e-01 1.40552068e+00 -3.57421666e-01
6.99524879e-02 5.53231947e-02 -2.57114738e-01 9.06529665e-01
8.35243881e-01 -7.67396212e-01 -3.95025462e-01 -2.50760078e-01
1.02683820e-01 2.45084748e-01 5.33968627e-01 -1.38182402e+00
4.77615863e-01 -3.81477922e-02 6.55238271e-01 -9.91354764e-01
2.76994526e-01 -9.06253099e-01 3.88734937e-02 3.96832317e-01
-1.97279409e-01 -9.92154330e-02 3.76745045e-01 1.06303799e+00
1.24276094e-01 -7.93192267e-01 8.76711369e-01 -1.62115142e-01
-3.86491328e-01 4.83446509e-01 -1.75215334e-01 1.28624085e-02
1.03442228e+00 -1.23331286e-02 -2.44696975e-01 6.91049099e-02
-9.97537374e-01 6.92095011e-02 -2.84882486e-01 -4.01239544e-01
5.58062673e-01 -1.35395777e+00 -5.75277865e-01 -1.19448118e-02
-2.73888022e-01 5.37965298e-02 7.74613440e-01 8.08502913e-01
-2.32523993e-01 3.61857742e-01 1.51576817e-01 -6.07079625e-01
-4.00681108e-01 6.83594882e-01 6.26780272e-01 -3.08289558e-01
-7.98582554e-01 9.18323696e-01 -2.36149002e-02 -5.15387952e-01
3.89731616e-01 -6.49906814e-01 8.55723619e-02 -4.00831282e-01
4.15115952e-01 7.88875997e-01 2.80543447e-01 7.61604309e-02
-8.21844637e-02 1.42674297e-01 -1.80321589e-01 1.73933367e-04
1.36392128e+00 3.43985073e-02 8.05510730e-02 2.00716212e-01
9.55300927e-01 -4.20028895e-01 -1.25928831e+00 -5.40037811e-01
-1.69201076e-01 3.43469717e-02 4.92088616e-01 -8.67231667e-01
-1.24760067e+00 6.18557572e-01 7.57914126e-01 1.54338270e-01
1.07919836e+00 -5.26617825e-01 8.84254813e-01 1.03007698e+00
3.09304774e-01 -1.24744046e+00 -2.80772537e-01 5.12474239e-01
7.63070047e-01 -8.77702117e-01 4.24523890e-01 -1.37553081e-01
-2.24773973e-01 1.31091046e+00 2.34394625e-01 -1.42570913e-01
1.10696292e+00 6.21591747e-01 -3.09624821e-01 -3.38250697e-01
-5.48409224e-02 2.48105913e-01 -9.76966918e-02 5.07786632e-01
-1.23280793e-01 -3.43227327e-01 -6.48441970e-01 9.70027566e-01
4.36732084e-01 2.42144942e-01 5.41596830e-01 1.00324738e+00
-4.74088937e-01 -6.56109631e-01 -5.09180784e-01 3.61025363e-01
-3.48540604e-01 -8.28136131e-02 2.15518460e-01 9.34023619e-01
1.09965149e-02 6.73649609e-01 -3.33020717e-01 -1.46977037e-01
4.79009688e-01 -5.39451987e-02 3.11884612e-01 3.12401857e-02
-1.64107934e-01 -1.65513039e-01 -4.01709266e-02 -2.16906622e-01
-5.51383913e-01 -2.63053805e-01 -1.49987519e+00 -3.80447179e-01
-5.59361696e-01 4.16004092e-01 8.57003212e-01 1.29710960e+00
1.86693206e-01 1.04280639e+00 4.37398523e-01 -8.72109830e-01
-1.09679353e+00 -1.02629304e+00 -1.01630211e+00 -1.50908366e-01
1.30102769e-01 -9.04203117e-01 -4.85663325e-01 -6.27676845e-01] | [7.222363471984863, 3.576080560684204] |
9abdd5ba-45e1-4a56-92d8-7a0c6653b478 | complex-valued-frequency-selective | 2204.14193 | null | https://arxiv.org/abs/2204.14193v1 | https://arxiv.org/pdf/2204.14193v1.pdf | Complex-Valued Frequency Selective Extrapolation for Fast Image and Video Signal Extrapolation | Signal extrapolation tasks arise in miscellaneous manners in the field of image and video signal processing. But, due to the widespread use of low-power and mobile devices, the computational complexity of an algorithm plays a crucial role in selecting an algorithm for a given problem. Within the scope of this contribution, we introduce the complex-valued Frequency Selective Extrapolation for fast image and video signal extrapolation. This algorithm iteratively generates a generic complex-valued model of the signal to be extrapolated as weighted superposition of Fourier basis functions. We further show that this algorithm is up to 10 times faster than the existent real-valued Frequency Selective Extrapolation that takes the real-valued nature of the input signals into account during the model generation. At the same time, the quality which is achievable by the complex-valued model generation is similar to the quality of the real-valued model generation. | ['André Kaup', 'Jürgen Seiler'] | 2022-04-27 | null | null | null | null | ['miscellaneous'] | ['miscellaneous'] | [ 7.68956065e-01 -1.65045455e-01 1.04197696e-01 5.05444652e-04
-6.90488815e-01 -2.91527569e-01 4.06363487e-01 -1.50719121e-01
-4.46819186e-01 7.61642873e-01 -2.09401220e-01 -2.60793120e-01
-2.51372129e-01 -4.81826812e-01 -6.14549994e-01 -7.91035533e-01
-3.16237509e-01 -1.89812094e-01 1.81301653e-01 -2.76161402e-01
1.25827119e-01 5.91979921e-01 -1.50548291e+00 2.14900792e-01
7.30315089e-01 1.16965759e+00 1.51517689e-01 1.18876302e+00
2.81737238e-01 5.20094037e-01 -5.38733363e-01 -6.68226108e-02
3.30110461e-01 -9.05120254e-01 -4.79688853e-01 1.15309708e-01
1.17205024e-01 -1.93622082e-01 -2.86451250e-01 1.16568780e+00
4.17852432e-01 1.26530901e-01 7.56915808e-01 -1.17169714e+00
1.51560351e-01 1.58402041e-01 -5.31292617e-01 4.15940464e-01
5.72915673e-01 7.47111291e-02 4.31043625e-01 -9.38542843e-01
5.48098624e-01 6.50056005e-01 7.49531925e-01 3.49465460e-01
-1.51284575e+00 -4.23463255e-01 -4.23193753e-01 1.43746585e-01
-1.57961690e+00 -4.96157795e-01 9.96150076e-01 -2.05886841e-01
5.22572100e-01 3.96111935e-01 7.55902469e-01 7.10295796e-01
3.82393450e-01 3.98634464e-01 1.01060653e+00 -9.07505751e-01
1.36986539e-01 9.82805863e-02 2.35094018e-02 4.61268038e-01
-1.82205345e-02 2.77132779e-01 -6.56761169e-01 -4.38075244e-01
8.38084877e-01 -4.56694782e-01 -8.67554247e-01 -1.69612132e-02
-1.23846412e+00 5.00388682e-01 1.28559303e-02 4.47200894e-01
-6.40748262e-01 3.51946473e-01 3.32682461e-01 4.18173999e-01
3.68204445e-01 3.10865372e-01 4.19797637e-02 -2.71062106e-01
-1.40861475e+00 2.10141107e-01 6.67298973e-01 8.33745539e-01
4.48478311e-01 4.77837861e-01 1.93037633e-02 5.07838309e-01
1.31751433e-01 6.08193994e-01 2.64598370e-01 -9.62081850e-01
1.17363475e-01 -1.15205579e-01 3.04599792e-01 -7.40059495e-01
-3.71908993e-01 -6.67277873e-01 -1.04745150e+00 2.92756081e-01
5.92533946e-01 -1.62267163e-01 -4.48429704e-01 1.50744247e+00
3.00676763e-01 3.57740700e-01 7.27736903e-03 6.88278198e-01
3.44874620e-01 9.97312367e-01 -1.91908881e-01 -7.89838016e-01
1.13162041e+00 -2.19274864e-01 -8.16384435e-01 3.43041122e-01
7.92153254e-02 -9.19600844e-01 7.06735134e-01 8.41697812e-01
-1.22930622e+00 -6.38927162e-01 -1.31644237e+00 2.10278645e-01
4.14459497e-01 6.24821894e-02 2.60116011e-01 7.02329040e-01
-1.03718102e+00 5.72026968e-01 -4.78934318e-01 2.04016685e-01
-1.20730186e-02 3.81855160e-01 -4.20669615e-02 3.06474268e-01
-1.25102103e+00 7.39794552e-01 6.92712292e-02 7.51598999e-02
-3.02359194e-01 -9.48946357e-01 -4.26284790e-01 8.41714069e-02
1.00182056e-01 -4.61275548e-01 1.28776407e+00 -1.34473038e+00
-1.56064141e+00 4.83411640e-01 -3.15080523e-01 -6.80712879e-01
7.86016524e-01 1.91107526e-01 -7.39931881e-01 6.60886407e-01
-2.67625064e-01 3.32173318e-01 1.39894211e+00 -8.94721687e-01
-5.27786016e-01 2.34697267e-01 -1.34948432e-01 6.71794489e-02
-1.82286471e-01 -4.66200709e-02 -1.97634280e-01 -7.44838417e-01
-2.90064793e-02 -7.44400024e-01 -3.17032248e-01 1.55520782e-01
1.47060305e-01 2.95204431e-01 5.14302611e-01 -6.83884561e-01
1.31233943e+00 -2.52832794e+00 2.95512890e-03 3.35180938e-01
-1.06906751e-02 1.70305729e-01 2.26894706e-01 5.64626575e-01
-3.55387300e-01 -2.94994026e-01 -3.35064232e-01 3.27302605e-01
-5.04019022e-01 -4.29870337e-01 -4.06222701e-01 7.28191197e-01
2.24981129e-01 3.38077813e-01 -8.50494146e-01 -3.79761338e-01
2.23624274e-01 7.23965168e-01 -4.49630946e-01 -1.23810753e-01
1.41717151e-01 5.25113523e-01 -2.84263343e-01 2.24066779e-01
7.30491161e-01 6.29663914e-02 1.32051677e-01 -4.93451446e-01
-9.60282981e-02 -1.31629422e-01 -1.17951429e+00 1.18235457e+00
-6.22148812e-01 9.90561366e-01 3.16653192e-01 -9.52426493e-01
8.08491588e-01 6.86015725e-01 8.23848009e-01 -5.77185929e-01
-5.30836061e-02 4.99951512e-01 7.36841187e-02 -2.99025565e-01
4.98396814e-01 -4.92360234e-01 2.60781169e-01 3.13906252e-01
-1.63698066e-02 -3.20102572e-01 2.03296289e-01 1.27720609e-01
7.94744790e-01 9.60405171e-02 6.21310234e-01 -4.54140097e-01
7.11597919e-01 7.89373145e-02 2.63732791e-01 3.55726182e-01
9.85714942e-02 5.42282343e-01 2.99302161e-01 -3.02264065e-01
-1.07463121e+00 -9.56196666e-01 -4.53079909e-01 4.55202639e-01
1.44650191e-01 -2.42823526e-01 -8.51651788e-01 -7.83971325e-02
-3.57953042e-01 5.65176964e-01 -1.25981182e-01 -1.81343496e-01
-7.06603408e-01 -6.91498578e-01 5.42928100e-01 1.55734614e-01
4.13799167e-01 -7.91643083e-01 -1.08379233e+00 4.84785408e-01
-2.52337694e-01 -1.20705462e+00 -6.31488621e-01 1.80126712e-01
-9.87850666e-01 -9.44834709e-01 -1.01508856e+00 -4.25396651e-01
6.10678554e-01 2.05990031e-01 8.41120958e-01 -7.28988368e-03
-2.99106121e-01 5.53892434e-01 -1.88984245e-01 -2.35394031e-01
-6.36734188e-01 -5.23900330e-01 -9.18290287e-04 4.50706214e-01
-2.66361862e-01 -6.30465329e-01 -4.04351294e-01 2.98437495e-02
-1.03010654e+00 9.51077715e-02 3.48992348e-01 9.27978575e-01
4.43460047e-01 5.31809866e-01 7.08817601e-01 -5.60997427e-01
9.18892980e-01 -2.69808769e-01 -7.72185922e-01 -4.56321537e-02
-2.37155035e-01 4.69258353e-02 1.16965330e+00 -9.43007290e-01
-1.02234077e+00 3.24111104e-01 -3.01276565e-01 -2.32146367e-01
2.30735913e-01 4.55259740e-01 2.85701871e-01 -4.46816742e-01
1.03462791e+00 3.90666068e-01 -5.87712266e-02 -7.76780099e-02
1.87382936e-01 7.14860380e-01 6.95750952e-01 -2.36642286e-01
6.29055858e-01 4.17004704e-01 5.68500102e-01 -1.48914182e+00
1.96288619e-02 -3.53989065e-01 -2.53823549e-01 -6.08831644e-01
3.48478019e-01 -6.82698011e-01 -6.36970639e-01 5.81565022e-01
-1.23037004e+00 -1.83510348e-01 -4.61612761e-01 8.42446566e-01
-8.17292988e-01 5.54591238e-01 -4.08402592e-01 -1.19683349e+00
-2.91561246e-01 -1.02227867e+00 7.00657368e-01 6.71352074e-02
-2.82253295e-01 -6.72306776e-01 -3.24960411e-01 -1.83504820e-01
2.89814502e-01 3.72077644e-01 7.80550122e-01 3.00166070e-01
-4.74365056e-01 -3.36007327e-01 1.74962282e-01 3.41625452e-01
-1.95711866e-01 -5.01027144e-02 -9.41667318e-01 -8.61190930e-02
5.91066360e-01 2.94632614e-01 6.44476652e-01 6.96652949e-01
8.92670095e-01 -4.67013270e-02 -3.86859663e-02 5.43940365e-01
1.60922289e+00 4.46662337e-01 6.35391653e-01 -2.96160966e-01
-2.39320323e-02 4.06052947e-01 4.93142128e-01 4.03507829e-01
-3.95290345e-01 8.71282160e-01 -5.89705352e-03 -7.43356943e-02
-1.87307581e-01 8.43718201e-02 2.14976892e-01 5.78080952e-01
-3.49778533e-01 3.73386703e-02 -5.52158535e-01 4.61634219e-01
-1.45288813e+00 -1.21277153e+00 -5.13879240e-01 2.61849642e+00
7.08181083e-01 2.15839773e-01 3.26729089e-01 7.96552896e-01
7.63368547e-01 -2.39618048e-01 -4.56832796e-01 -4.79173899e-01
5.00440970e-02 3.36676717e-01 5.92581868e-01 5.34800529e-01
-6.59514904e-01 -4.21057455e-02 6.90225601e+00 1.05447865e+00
-1.32627177e+00 -5.62502295e-02 3.84087771e-01 2.93696318e-02
-1.90207079e-01 -2.54907876e-01 -3.63457829e-01 5.99944711e-01
1.28279030e+00 -5.03817379e-01 6.46423280e-01 4.07585710e-01
3.81650746e-01 -4.52837825e-01 -8.17562163e-01 1.13029659e+00
-2.09275290e-01 -1.13018334e+00 -2.49212116e-01 6.17249943e-02
4.68847752e-01 -5.48209846e-01 -5.93275763e-02 -2.52201587e-01
-6.43955171e-01 -8.25265706e-01 1.01889253e+00 5.10987520e-01
1.22341454e+00 -1.01734912e+00 2.18499452e-01 6.95976973e-01
-1.23477638e+00 2.50128992e-02 -2.49479562e-02 -1.18167192e-01
4.99936521e-01 9.94583189e-01 -6.05828583e-01 6.88340902e-01
1.56473458e-01 9.74945500e-02 9.90296900e-02 1.23634517e+00
1.08754836e-01 8.56279969e-01 -5.30679226e-01 -7.54736783e-03
4.66805212e-02 -3.07444662e-01 7.62610137e-01 1.12708914e+00
5.95869601e-01 3.33459616e-01 -2.28574917e-01 4.85902786e-01
2.26093292e-01 2.94712447e-02 -5.27532101e-01 7.45709687e-02
2.13204905e-01 8.45821202e-01 -8.40746343e-01 -2.28788406e-01
-4.91092086e-01 9.57136750e-01 -4.59961325e-01 5.97217143e-01
-8.58681977e-01 -7.67689884e-01 1.03199460e-01 4.38645363e-01
6.09330982e-02 -3.43816400e-01 -1.87969148e-01 -8.58938694e-01
1.58074856e-01 -8.69971156e-01 4.19137888e-02 -7.25249350e-01
-6.30701542e-01 7.83062994e-01 2.89738059e-01 -1.72089493e+00
-7.30609059e-01 -3.36242378e-01 -5.12401879e-01 1.07390332e+00
-1.33322382e+00 -4.58205879e-01 -3.01628411e-02 5.72409034e-01
4.82792199e-01 2.03034252e-01 8.02813828e-01 3.51077735e-01
2.82375634e-01 2.61095047e-01 9.42696407e-02 -2.79834628e-01
1.51496097e-01 -7.61142433e-01 1.64767161e-01 9.95014727e-01
2.93547893e-03 4.16671634e-01 1.09315586e+00 -2.34033406e-01
-1.51289940e+00 -7.06836879e-01 8.40624273e-01 4.22130287e-01
3.24060917e-01 -1.55837819e-01 -7.08739281e-01 5.34077883e-02
5.07287383e-02 1.72790527e-01 3.47204030e-01 -5.19791365e-01
1.58955812e-01 -2.05184326e-01 -1.39451611e+00 6.77204132e-01
7.00252414e-01 -5.38489103e-01 -2.03951418e-01 1.73376352e-01
3.57850820e-01 -4.77802336e-01 -7.86348522e-01 2.83650041e-01
6.22489989e-01 -1.00854087e+00 1.14859235e+00 -2.22604014e-02
1.87567532e-01 -5.14214158e-01 -2.19026785e-02 -1.13009763e+00
-2.22595081e-01 -1.12067223e+00 -2.27623940e-01 6.24791980e-01
3.07193547e-01 -6.43853903e-01 5.74263036e-01 1.44245982e-01
2.24024609e-01 -7.35614538e-01 -1.11183143e+00 -9.52487111e-01
-2.93596894e-01 -5.80205560e-01 1.54801533e-01 5.09623647e-01
2.36218527e-01 2.54495084e-01 -5.14714122e-01 6.32065460e-02
7.78455138e-01 8.81575495e-02 3.61852020e-01 -1.12797904e+00
-6.13297462e-01 -2.48950720e-01 -5.72660148e-01 -1.26252544e+00
-3.33906382e-01 -5.05574584e-01 -7.19909146e-02 -9.63069260e-01
-4.02301729e-01 -2.06886634e-01 6.38459250e-02 -3.97421688e-01
1.77657381e-01 5.10939240e-01 3.29313934e-01 -1.33231565e-01
2.25829691e-01 1.89759910e-01 1.17216969e+00 1.13633357e-01
-2.72230297e-01 6.01559997e-01 -2.52599567e-01 7.12501764e-01
6.65904999e-01 -3.41587842e-01 -7.67318368e-01 2.16667041e-01
3.75868708e-01 7.35324681e-01 4.65817690e-01 -1.22432458e+00
2.93337792e-01 4.59990054e-02 3.18705946e-01 -3.86604309e-01
6.62928581e-01 -9.94067013e-01 6.25261426e-01 6.99789107e-01
-1.42812803e-01 -1.32013500e-01 1.12026259e-01 4.83765423e-01
-3.23619217e-01 -5.49336016e-01 1.11695468e+00 -6.80185109e-02
-5.61917365e-01 -1.47373676e-01 -7.69628882e-01 -6.62822053e-02
9.28277135e-01 -5.88628829e-01 2.71584809e-01 -8.93173456e-01
-5.55903256e-01 -4.32984561e-01 1.55320570e-01 -3.05381477e-01
6.51563287e-01 -1.24282372e+00 -6.24639034e-01 4.24642414e-01
-4.97391790e-01 -3.52796823e-01 4.15609300e-01 1.13939309e+00
-5.03307045e-01 2.47002527e-01 -1.51572749e-01 -6.50002897e-01
-1.26758039e+00 3.31289560e-01 4.20414418e-01 -8.66815373e-02
-6.79094017e-01 4.48875964e-01 -1.34849563e-01 4.53753084e-01
-6.04813099e-02 -2.94026643e-01 9.37872455e-02 -1.95793152e-01
7.26325691e-01 6.04795814e-01 2.30369046e-01 -6.25843167e-01
-2.79704958e-01 7.25204408e-01 5.26162922e-01 -6.08200312e-01
1.06477857e+00 -2.62337439e-02 9.50571522e-02 3.85645896e-01
1.16348064e+00 2.62441218e-01 -1.18500757e+00 1.46945372e-01
-3.36484730e-01 -2.34976456e-01 2.34739199e-01 -4.52900082e-01
-7.08771884e-01 7.14036107e-01 3.48719597e-01 4.53482836e-01
1.86082828e+00 -6.26742363e-01 5.77457011e-01 -2.19442211e-02
6.23716235e-01 -9.11857486e-01 -3.22839737e-01 2.54898071e-02
8.41657341e-01 -5.28221488e-01 2.36126035e-01 -9.49958324e-01
-3.17019522e-01 1.29714203e+00 -2.78652847e-01 -2.83392847e-01
7.69071758e-01 5.01160085e-01 -9.29250792e-02 3.31956059e-01
-5.00982404e-01 1.52360648e-01 3.21573526e-01 7.98190415e-01
4.65669394e-01 -1.21612616e-01 -8.29332769e-01 2.00676009e-01
-3.77343111e-02 5.78676045e-01 7.86202729e-01 7.15609848e-01
-2.84264922e-01 -9.73800600e-01 -7.31186211e-01 2.99621940e-01
-6.56698525e-01 -6.85363114e-02 2.14989454e-01 5.37661433e-01
-1.41105324e-01 1.06159604e+00 -2.14734882e-01 -9.58938152e-03
1.57943740e-01 -8.67568702e-03 8.20760429e-01 1.32440431e-02
-4.94168401e-01 3.20375502e-01 1.03069350e-01 -3.37362230e-01
-4.08000588e-01 -4.52203900e-01 -1.33130825e+00 -1.38949469e-01
-3.48861247e-01 2.69815534e-01 6.99957192e-01 6.64610267e-01
8.47040117e-02 4.58230376e-01 6.30478382e-01 -1.00161862e+00
-6.70413435e-01 -5.93914449e-01 -7.43321359e-01 1.46133289e-01
6.28468454e-01 -3.45601201e-01 -3.11178327e-01 3.16614777e-01] | [6.839897155761719, 1.6017234325408936] |
2fcbab23-271f-4682-8390-87548aa76071 | labr-a-large-scale-arabic-sentiment-analysis | 1411.6718 | null | http://arxiv.org/abs/1411.6718v2 | http://arxiv.org/pdf/1411.6718v2.pdf | LABR: A Large Scale Arabic Sentiment Analysis Benchmark | We introduce LABR, the largest sentiment analysis dataset to-date for the
Arabic language. It consists of over 63,000 book reviews, each rated on a scale
of 1 to 5 stars. We investigate the properties of the dataset, and present its
statistics. We explore using the dataset for two tasks: (1) sentiment polarity
classification; and (2) ratings classification. Moreover, we provide standard
splits of the dataset into training, validation and testing, for both polarity
and ratings classification, in both balanced and unbalanced settings. We extend
our previous work by performing a comprehensive analysis on the dataset. In
particular, we perform an extended survey of the different classifiers
typically used for the sentiment polarity classification problem. We also
construct a sentiment lexicon from the dataset that contains both single and
compound sentiment words and we explore its effectiveness. We make the dataset
and experimental details publicly available. | ['Amir Atiya', 'Mohamed Aly', 'Mahmoud Nabil'] | 2014-11-25 | null | null | null | null | ['arabic-sentiment-analysis'] | ['natural-language-processing'] | [-3.85178477e-02 -1.17893614e-01 -7.44258344e-01 -9.22151983e-01
-7.16062248e-01 -1.27942705e+00 7.16696858e-01 5.70820034e-01
-1.92366660e-01 6.98985219e-01 5.22520781e-01 -3.17374974e-01
3.35715026e-01 -5.86910069e-01 -2.23527044e-01 -4.31711763e-01
1.85553938e-01 4.41691190e-01 -1.61266446e-01 -8.13194573e-01
7.09527910e-01 4.45563793e-02 -1.62966955e+00 6.65213406e-01
7.72768736e-01 1.49311030e+00 -4.98371363e-01 5.03569484e-01
1.19322561e-01 1.10503101e+00 -8.39447439e-01 -1.28759015e+00
-7.29710907e-02 -9.06982869e-02 -9.38257456e-01 5.49773090e-02
1.24964915e-01 3.36585402e-01 4.00001049e-01 8.45866680e-01
4.13136423e-01 -1.38203725e-01 9.88400698e-01 -1.33967960e+00
-7.91059673e-01 7.81795800e-01 -7.45508015e-01 7.51242926e-03
6.70565248e-01 -5.83273947e-01 1.80644393e+00 -1.20562577e+00
6.73946857e-01 7.29779005e-01 4.71755624e-01 2.95350045e-01
-5.50019085e-01 -4.13119078e-01 3.33250433e-01 2.01685742e-01
-6.35188699e-01 -2.37146392e-01 1.00498044e+00 -3.82645011e-01
8.33743811e-01 1.87157989e-01 7.03068256e-01 1.02597940e+00
1.59789901e-03 8.60175252e-01 1.74041474e+00 -5.77502787e-01
3.74081433e-01 6.61059856e-01 7.45227754e-01 6.84017614e-02
2.73661882e-01 -6.78657115e-01 -7.69174457e-01 -1.20926872e-01
-3.52648765e-01 -3.34184200e-01 -1.35240853e-01 -2.52200574e-01
-9.05349791e-01 1.04229021e+00 8.61783624e-02 -4.75021973e-02
-7.96253905e-02 -5.76503575e-01 6.56111002e-01 8.54515195e-01
8.14086735e-01 7.62522817e-01 -1.21044362e+00 -2.06774965e-01
-5.05336583e-01 2.28640765e-01 1.19941568e+00 9.37113047e-01
4.03462708e-01 -2.90943354e-01 2.70933241e-01 1.24339426e+00
1.19179592e-01 6.18179739e-01 7.08034456e-01 -4.87713546e-01
5.06565750e-01 6.74615145e-01 7.41650388e-02 -1.13214266e+00
-4.21552896e-01 -2.06762813e-02 -3.64130497e-01 -2.04285502e-01
9.08073187e-02 -3.41305703e-01 -5.83306372e-01 1.33956456e+00
1.49080500e-01 -7.71548569e-01 3.01502198e-01 5.78339279e-01
1.37279594e+00 5.31367421e-01 -2.77831018e-01 -3.43672037e-01
1.64844656e+00 -1.20813072e+00 -7.24732220e-01 -3.38474333e-01
9.84467268e-01 -9.32269573e-01 1.37203896e+00 9.72010672e-01
-8.92277062e-01 -9.71823335e-02 -1.28852665e+00 -1.70632854e-01
-9.46329176e-01 3.80352914e-01 9.29639280e-01 1.01376998e+00
-8.79901052e-01 2.07988918e-01 -2.10494086e-01 7.11115822e-02
2.67391801e-01 2.16760740e-01 -5.88510215e-01 -5.46555221e-02
-1.26028478e+00 1.08907700e+00 -3.51360291e-01 -1.85915217e-01
-2.39350080e-01 -3.76768261e-01 -1.21112692e+00 -4.78543103e-01
1.61164165e-01 9.06948224e-02 1.44191146e+00 -1.31946254e+00
-1.47139263e+00 1.31485975e+00 -3.84063423e-01 -1.22844437e-02
-1.50529563e-01 -1.20807551e-01 -6.55776680e-01 -1.18924052e-01
8.46813396e-02 1.46380946e-01 5.11733115e-01 -1.12672400e+00
-7.70463228e-01 -4.96460974e-01 4.55148488e-01 4.61402118e-01
-6.77993357e-01 4.95808631e-01 -3.74498904e-01 -8.48006785e-01
-1.53164625e-01 -1.11884415e+00 -9.46163908e-02 -9.80893910e-01
-4.10163313e-01 -2.58974969e-01 3.57680231e-01 -3.71783853e-01
1.31617546e+00 -1.79622734e+00 9.70720351e-02 2.73831725e-01
-1.02638826e-01 -3.23030561e-01 -1.79306611e-01 3.74866515e-01
-5.44825315e-01 1.37887985e-01 -7.94288293e-02 -7.10077107e-01
-4.17676792e-02 -2.48253211e-01 -6.50836349e-01 5.15037477e-01
1.16628483e-01 9.16923225e-01 -5.39039433e-01 -1.89089790e-01
-4.93126363e-01 1.19776741e-01 -4.79433477e-01 -9.76960286e-02
-7.27890851e-03 -4.56295162e-02 -3.04221213e-01 1.43337381e+00
4.76212084e-01 -1.36205897e-01 4.37080562e-01 -8.99857879e-02
1.06792590e-02 1.01804078e+00 -7.12175727e-01 9.85380769e-01
-6.35738373e-01 7.88935184e-01 -1.46526203e-01 -8.58512282e-01
1.09656465e+00 1.28469363e-01 2.24564508e-01 -7.61136293e-01
4.89964336e-01 4.28800732e-01 -1.89894035e-01 -3.35652893e-03
1.08799148e+00 -1.72040164e-01 -8.52429926e-01 6.73850536e-01
-5.27571104e-02 -5.22646785e-01 8.45457613e-01 3.55186641e-01
6.68695152e-01 -3.21865946e-01 3.85716587e-01 -4.21276301e-01
8.18479359e-01 2.62794554e-01 1.98888808e-01 1.02580547e-01
-1.30428016e-01 6.58362210e-01 1.29347920e+00 -3.29434425e-01
-7.66786516e-01 -4.81320411e-01 -4.45199847e-01 1.48703301e+00
1.15090199e-01 -8.78907323e-01 -4.17145699e-01 -1.16316271e+00
-3.87691222e-02 2.90561825e-01 -1.09463847e+00 1.88419446e-01
-1.96157515e-01 -1.42567945e+00 -5.34792542e-02 7.01103628e-01
-1.29783735e-01 -1.36537516e+00 -2.72347480e-02 -5.61021030e-01
-2.55843520e-01 -9.86519098e-01 -2.41263315e-01 6.64626479e-01
-6.73852086e-01 -1.19639468e+00 -3.21091294e-01 -1.10050106e+00
6.95422649e-01 1.21732213e-01 1.92615759e+00 -3.51180206e-03
4.47284639e-01 2.89673984e-01 -1.19799888e+00 -8.41322720e-01
-1.51080683e-01 4.39189106e-01 1.15933001e-01 -2.82730401e-01
7.35734701e-01 -1.94512606e-02 -2.10708603e-01 4.35469240e-01
-5.78244746e-01 -5.25065303e-01 2.84581810e-01 7.62293518e-01
6.41145349e-01 5.53353764e-02 1.00641859e+00 -1.61893988e+00
1.05724978e+00 -6.24900758e-01 -3.92068565e-01 2.11332683e-02
-6.90618873e-01 -4.87050116e-01 7.37691283e-01 -1.35027811e-01
-6.49618864e-01 -2.02400863e-01 -2.56923914e-01 5.52389503e-01
3.69139403e-01 1.00898302e+00 -7.04030693e-02 5.55574968e-02
4.50683385e-01 -3.48596096e-01 -3.90502334e-01 -9.35283527e-02
4.36673105e-01 1.04858303e+00 1.56871244e-01 -4.32668626e-01
4.41739112e-01 3.34375530e-01 -3.31601292e-01 -5.98655403e-01
-1.51602817e+00 -6.06854141e-01 -5.56624174e-01 -9.09102634e-02
3.29916000e-01 -1.25272727e+00 -4.78799403e-01 5.49384475e-01
-5.24271071e-01 -1.88593596e-01 -8.66978988e-02 8.64488631e-02
-3.81836504e-01 8.83609578e-02 -8.51201773e-01 -5.46962738e-01
-3.78326803e-01 -1.17037654e+00 8.84775400e-01 1.16605289e-01
-5.59820712e-01 -1.23123240e+00 3.86033505e-01 6.47762299e-01
8.86011720e-02 9.66625959e-02 6.98680043e-01 -9.10713673e-01
5.42555511e-01 -5.86955309e-01 1.57104537e-01 6.10340655e-01
6.68346658e-02 3.13878775e-01 -9.95251656e-01 -2.19183177e-01
1.55725911e-01 -1.08508420e+00 1.00142181e+00 2.44554535e-01
1.19801712e+00 2.31318474e-02 7.74417445e-02 8.37835446e-02
1.16926527e+00 3.18908431e-02 6.35057807e-01 8.76865864e-01
4.57639933e-01 1.13261735e+00 1.30267179e+00 3.29696149e-01
9.54460263e-01 1.66336581e-01 3.83927971e-01 -1.30076930e-01
5.42212546e-01 1.57716259e-01 5.76725364e-01 1.37663102e+00
-7.15023186e-03 -2.95835555e-01 -7.98301458e-01 7.64610052e-01
-1.39401639e+00 -4.97506112e-01 -1.11317769e-01 1.90830421e+00
1.22540319e+00 3.32253754e-01 4.12498504e-01 7.66746819e-01
3.63191783e-01 3.20906371e-01 -1.73402160e-01 -9.07440007e-01
-5.77064216e-01 5.03079355e-01 4.03083712e-01 2.53276795e-01
-1.50521123e+00 9.02777255e-01 7.19521904e+00 4.77681309e-01
-1.08186841e+00 -1.62117794e-01 1.15297771e+00 -1.26034573e-01
-4.81057256e-01 -1.50187328e-01 -7.68127680e-01 2.61505991e-01
8.55350077e-01 1.74595080e-02 7.12994933e-02 1.20386732e+00
-4.35314715e-01 -2.22574249e-01 -9.19891715e-01 8.04840982e-01
5.54673970e-01 -9.01895583e-01 -7.42199877e-03 -3.23201537e-01
1.17817068e+00 1.03067584e-01 4.32821661e-01 5.12111187e-01
1.94079176e-01 -1.28787315e+00 8.64907742e-01 -1.58095479e-01
1.02528751e+00 -1.05741298e+00 1.28820646e+00 -1.96751758e-01
-8.50714743e-01 -2.72672772e-01 -1.41740859e-01 -4.43905294e-01
-1.99594378e-01 8.03481281e-01 -7.83826970e-03 3.85535985e-01
1.06563699e+00 1.55867171e+00 -1.00817716e+00 2.93217719e-01
-6.37000799e-01 7.59967625e-01 1.03293873e-01 -5.14810681e-01
-1.10963203e-01 -4.98310626e-01 -1.10930353e-01 1.10379219e+00
-2.11393330e-02 -1.04757778e-01 -2.79182553e-01 -2.66689152e-01
-4.88326073e-01 5.51175296e-01 -3.82444978e-01 -2.06075609e-01
4.23812658e-01 1.90997696e+00 -8.98680270e-01 -3.89033914e-01
-6.29590511e-01 3.96039426e-01 3.96311164e-01 -6.69331998e-02
-5.57797134e-01 -9.10857975e-01 4.69198823e-01 -2.92967886e-01
2.91957438e-01 2.37642512e-01 -7.93474317e-01 -1.44100463e+00
2.96347797e-01 -1.42842221e+00 4.79868591e-01 -8.70133936e-01
-1.71734500e+00 8.66334200e-01 -3.44188273e-01 -1.22964585e+00
-1.00266449e-01 -1.33985388e+00 -4.14258599e-01 5.76952398e-01
-2.05408692e+00 -9.19417262e-01 -1.41989514e-01 3.50375116e-01
3.01359862e-01 -3.03177714e-01 7.62430251e-01 2.17013076e-01
-5.67120373e-01 6.21222973e-01 9.24373337e-04 2.03794092e-01
1.34094584e+00 -1.66314209e+00 2.47037381e-01 3.20290655e-01
-1.41663834e-01 7.24979997e-01 5.61671138e-01 -4.02348727e-01
-1.05079663e+00 -6.84912980e-01 1.24376619e+00 -1.20935500e+00
1.07416105e+00 -5.11157036e-01 -3.86164814e-01 7.64109910e-01
2.96446770e-01 -4.31405991e-01 1.24511766e+00 7.34005153e-01
-6.44839048e-01 -1.11993253e-01 -9.57560837e-01 4.73810941e-01
3.92927945e-01 -4.28292841e-01 -6.74755931e-01 1.70719206e-01
4.89085525e-01 -3.99477810e-01 -9.32461262e-01 5.48882544e-01
7.09000170e-01 -8.16346943e-01 5.43294609e-01 -5.16876996e-01
1.39259422e+00 -4.77496982e-02 -2.56502450e-01 -1.62740755e+00
1.73809528e-01 -1.17950387e-01 -4.17047106e-02 1.04530418e+00
1.23730564e+00 -6.88526452e-01 7.75963843e-01 2.29646564e-01
-1.24467775e-01 -1.44845700e+00 -2.60677397e-01 -3.61398142e-03
5.51754177e-01 -6.33799791e-01 5.74907839e-01 1.14453018e+00
7.74026752e-01 8.52636218e-01 -1.24563366e-01 -4.51831013e-01
-9.66240615e-02 6.35600924e-01 5.98916352e-01 -1.22930098e+00
4.07038219e-02 -6.03038490e-01 -2.05161027e-03 -7.75024593e-01
1.62421331e-01 -6.72000706e-01 -1.50983736e-01 -1.39463568e+00
2.93332875e-01 -4.98530358e-01 -1.52894810e-01 4.49042797e-01
-2.43265331e-01 8.36547911e-01 -9.56683755e-02 2.08766356e-01
-9.34801221e-01 2.82711476e-01 1.11198592e+00 -1.06954932e-01
-3.60657126e-02 5.71774365e-03 -1.53455830e+00 8.81297112e-01
9.69712079e-01 -2.81836033e-01 -5.28725088e-01 -6.21617888e-04
1.36968625e+00 -4.21865046e-01 -6.47530317e-01 -2.75564939e-01
-3.21084172e-01 -6.95992261e-02 4.31200564e-01 -7.59251356e-01
3.66459191e-01 -3.96377206e-01 -9.06795979e-01 -3.36767584e-01
-6.80405438e-01 5.31804442e-01 -5.87756187e-02 1.35078162e-01
-6.78548336e-01 -2.43397892e-01 6.62962914e-01 1.19160660e-01
-3.78479540e-01 1.07656881e-01 -4.27406698e-01 3.22377175e-01
9.35044289e-01 2.52316684e-01 -7.32642055e-01 -6.92880452e-01
-4.04947609e-01 5.05413949e-01 7.48960376e-01 7.13601708e-01
3.32502514e-01 -1.08168423e+00 -5.52263677e-01 -4.57323566e-02
7.05687106e-01 -3.81056964e-01 -2.05984756e-01 8.50334823e-01
-3.77025157e-01 3.21134984e-01 1.06430706e-02 4.61669713e-02
-1.39520967e+00 4.67490315e-01 3.44828330e-02 -4.76191133e-01
3.54732305e-01 6.29517257e-01 -6.74437135e-02 -8.79562736e-01
8.43779277e-03 -2.14363322e-01 -1.43284512e+00 8.76965225e-01
6.49445295e-01 1.21301465e-01 5.30099809e-01 -9.63089764e-01
-4.23658967e-01 6.38132989e-01 -2.70834833e-01 -2.00244710e-01
1.41789198e+00 -1.99421868e-01 -7.92267084e-01 8.73864055e-01
1.10646307e+00 7.19437540e-01 -2.39854500e-01 1.45941213e-01
3.62941772e-02 -8.92899781e-02 -1.13802947e-01 -1.18329847e+00
-9.98565316e-01 5.01310050e-01 -2.50943184e-01 7.82828450e-01
1.26151633e+00 7.37059042e-02 5.09326935e-01 4.73148853e-01
1.76420793e-01 -1.44828320e+00 3.20412278e-01 1.08351123e+00
7.63901651e-01 -1.50819838e+00 3.72369200e-01 -5.56117058e-01
-1.38620174e+00 8.18315148e-01 4.17222887e-01 -2.13235125e-01
9.34480667e-01 2.13681012e-01 7.61144340e-01 -4.20594215e-01
-1.13402128e+00 -4.30075601e-02 4.72597539e-01 5.38279176e-01
1.19545066e+00 6.75001070e-02 -7.23421037e-01 1.34968615e+00
-1.07197356e+00 -4.97130990e-01 1.18100584e+00 1.04914606e+00
-2.32351106e-02 -1.19589651e+00 -9.55866352e-02 6.63162351e-01
-1.00484705e+00 -2.13640064e-01 -9.61525559e-01 5.23651183e-01
-3.68876129e-01 1.43068635e+00 -6.74925745e-02 -4.21835423e-01
4.83075559e-01 -2.25161478e-01 1.71931088e-01 -6.63316727e-01
-7.04446256e-01 -3.66253138e-01 5.88903368e-01 -1.70859292e-01
-8.44205856e-01 -7.71257818e-01 -9.14797783e-01 -1.91142410e-01
-4.82341915e-01 4.63730901e-01 1.00179768e+00 6.78518653e-01
4.05681133e-03 2.62782067e-01 1.20194423e+00 -6.92958593e-01
-4.10805613e-01 -1.08444881e+00 -9.30920303e-01 4.96140182e-01
3.59943151e-01 -5.62575102e-01 -7.43220568e-01 -5.10206930e-02] | [11.236339569091797, 6.893028736114502] |
9c3e1d6a-00b0-46e9-bdc6-29de1987419c | spectrum-inspired-low-light-image-translation | 2303.10145 | null | https://arxiv.org/abs/2303.10145v1 | https://arxiv.org/pdf/2303.10145v1.pdf | Spectrum-inspired Low-light Image Translation for Saliency Detection | Saliency detection methods are central to several real-world applications such as robot navigation and satellite imagery. However, the performance of existing methods deteriorate under low-light conditions because training datasets mostly comprise of well-lit images. One possible solution is to collect a new dataset for low-light conditions. This involves pixel-level annotations, which is not only tedious and time-consuming but also infeasible if a huge training corpus is required. We propose a technique that performs classical band-pass filtering in the Fourier space to transform well-lit images to low-light images and use them as a proxy for real low-light images. Unlike popular deep learning approaches which require learning thousands of parameters and enormous amounts of training data, the proposed transformation is fast and simple and easy to extend to other tasks such as low-light depth estimation. Our experiments show that the state-of-the-art saliency detection and depth estimation networks trained on our proxy low-light images perform significantly better on real low-light images than networks trained using existing strategies. | ['Kaushik Mitra', 'Mohit Lamba', 'Sudarshan Rajagopalan', 'Kitty Varghese'] | 2023-03-17 | null | null | null | null | ['saliency-detection', 'robot-navigation'] | ['computer-vision', 'robots'] | [ 5.27642727e-01 -2.33216897e-01 3.36360708e-02 -4.21438277e-01
-5.49272358e-01 -1.95257440e-01 4.31594491e-01 -2.14046553e-01
-7.31792986e-01 9.78079379e-01 -3.97283643e-01 -1.85501575e-01
2.94331759e-01 -9.91862118e-01 -8.97497356e-01 -8.89433920e-01
3.81426156e-01 2.05356792e-01 8.52938712e-01 -2.24302724e-01
3.52927446e-01 2.77712941e-01 -2.07868147e+00 -8.35573897e-02
9.82655525e-01 8.53877246e-01 7.52148747e-01 5.86011469e-01
-6.75445721e-02 5.79114556e-01 -2.64870286e-01 -2.09482938e-01
5.32853901e-01 -3.68490756e-01 -6.29741490e-01 1.18936813e-02
8.56845796e-01 -4.82298493e-01 -8.30061510e-02 1.45738351e+00
5.37918985e-01 2.65365183e-01 2.19718710e-01 -9.94167626e-01
-4.04783636e-01 2.85386248e-03 -8.44331026e-01 3.36567611e-01
1.95827380e-01 3.08534920e-01 5.59390783e-01 -1.00497687e+00
4.29497123e-01 9.41497207e-01 7.01767087e-01 4.84338671e-01
-1.00454259e+00 -6.66611254e-01 -5.06191589e-02 4.82319236e-01
-8.45996439e-01 -5.49193203e-01 7.90985525e-01 -1.08669378e-01
7.76913941e-01 -8.39735419e-02 7.82838345e-01 5.90765655e-01
5.29795811e-02 6.46619737e-01 1.28817773e+00 -5.78369856e-01
2.62957543e-01 -3.24077271e-02 -2.07510412e-01 8.65965784e-01
6.42291844e-01 8.84569436e-02 -6.87561452e-01 3.68463874e-01
8.58200848e-01 1.54048741e-01 -4.51514989e-01 -3.77556056e-01
-1.25924158e+00 6.57815278e-01 7.25868106e-01 1.07940473e-01
-4.67870295e-01 1.84323579e-01 6.98132953e-03 -2.58539729e-02
6.85940087e-01 4.82106149e-01 -3.67611557e-01 1.06522262e-01
-1.13633025e+00 7.55822882e-02 3.43712330e-01 6.03274047e-01
1.47316372e+00 8.34974274e-02 2.54919052e-01 6.65422499e-01
6.80513754e-02 6.81586623e-01 4.46771979e-01 -1.00847661e+00
2.31098756e-01 3.42717707e-01 3.80134225e-01 -8.39806378e-01
-6.59780025e-01 -5.12743592e-01 -7.78700233e-01 5.57757497e-01
7.23818541e-01 -3.13655734e-02 -9.88938928e-01 1.47138667e+00
3.50431532e-01 3.27391773e-01 9.95786414e-02 1.32411766e+00
8.46694946e-01 6.74954236e-01 -2.11337388e-01 -2.54546463e-01
1.11391127e+00 -1.22667861e+00 -4.41065401e-01 -9.19984698e-01
3.88726801e-01 -7.47595727e-01 1.60904336e+00 3.83633763e-01
-1.28234160e+00 -4.52308476e-01 -1.13230681e+00 -5.40180385e-01
-3.99815947e-01 1.90196469e-01 9.81212795e-01 5.60168564e-01
-1.13082814e+00 2.86068559e-01 -6.04395866e-01 -3.85638386e-01
5.28789401e-01 1.97399795e-01 -1.47871077e-01 -3.33885521e-01
-1.14526176e+00 8.21940303e-01 2.10723042e-01 2.39619598e-01
-8.60520840e-01 -5.94063342e-01 -9.05791283e-01 -4.68142368e-02
5.19323826e-01 -6.38987422e-01 1.23421514e+00 -9.42128837e-01
-1.49908555e+00 1.04914618e+00 -3.49123120e-01 -4.10553992e-01
3.28387767e-01 -2.54973263e-01 1.00364380e-01 3.76188725e-01
2.67941564e-01 8.07840705e-01 1.01058972e+00 -1.30567360e+00
-9.00538445e-01 -2.46709228e-01 3.69595170e-01 3.80914748e-01
-1.80140764e-01 -2.75474072e-01 -4.20395195e-01 -2.10542399e-02
3.44748497e-01 -8.84420037e-01 -2.06551284e-01 2.84238011e-01
-6.73468709e-02 1.46466553e-01 8.01883101e-01 -2.93690592e-01
5.94792247e-01 -1.69930053e+00 -3.31518978e-01 -2.88028598e-01
9.17618498e-02 4.18615282e-01 3.82913742e-03 -1.99641854e-01
1.95886686e-01 -5.17404675e-01 -4.14816320e-01 -1.99461594e-01
-3.72794151e-01 1.10655151e-01 -2.03204766e-01 6.52919650e-01
1.01591602e-01 7.17624068e-01 -1.37184632e+00 -5.27254105e-01
6.75398707e-01 4.25886899e-01 -5.73973417e-01 -2.80703008e-02
-3.28845143e-01 5.42777240e-01 -1.08187879e-03 7.21897960e-01
8.41774940e-01 -1.94113210e-01 -3.31015527e-01 -2.39933252e-01
-3.35469604e-01 3.63704383e-01 -8.62736583e-01 1.85628259e+00
-8.62610281e-01 1.07788742e+00 1.03900358e-01 -8.84015858e-01
7.88964510e-01 -3.84474576e-01 2.70517081e-01 -9.59334731e-01
1.92556798e-01 3.92019868e-01 -2.83022195e-01 -6.05351627e-01
5.66882372e-01 -3.56729627e-01 2.45569065e-01 4.09999520e-01
-1.39580503e-01 -6.81444883e-01 3.66464585e-01 -2.23293275e-01
7.49605179e-01 3.04104209e-01 1.59148023e-01 -6.98844492e-02
5.50971568e-01 -9.24227238e-02 6.40722930e-01 6.06714070e-01
-2.52424479e-01 6.32684588e-01 1.88744739e-02 -5.37626803e-01
-9.10409749e-01 -7.90712237e-01 -1.69258267e-01 1.12625682e+00
5.38799107e-01 1.16075054e-01 -7.72171319e-01 -3.31121862e-01
-3.46781641e-01 6.52565598e-01 -3.61869305e-01 -8.67285579e-02
-3.01909655e-01 -9.39047754e-01 1.95961278e-02 2.64204502e-01
1.14237022e+00 -1.11413789e+00 -1.25200891e+00 1.74533855e-02
-4.28800851e-01 -1.27694499e+00 5.93844149e-03 4.23008561e-01
-8.18880200e-01 -1.08382380e+00 -8.12324286e-01 -9.63424563e-01
7.46268451e-01 9.59073842e-01 1.17707157e+00 1.27323344e-01
-3.47937405e-01 -5.34065850e-02 -3.89432125e-02 -6.16581678e-01
1.12747312e-01 8.18884820e-02 -1.12601526e-01 -8.80783144e-03
5.18076062e-01 -5.04728198e-01 -1.02276862e+00 2.99740702e-01
-9.25693929e-01 3.70307148e-01 8.68017137e-01 9.17383552e-01
6.76676810e-01 1.48296028e-01 3.43417436e-01 -8.08809340e-01
1.80523574e-01 1.18081629e-01 -9.34141159e-01 1.33340403e-01
-5.99741280e-01 -9.43446457e-02 5.48692346e-01 -1.83146492e-01
-1.27885127e+00 3.31144124e-01 -4.25574221e-02 -1.50882781e-01
-2.13969380e-01 1.87034056e-01 6.80661127e-02 -5.37487626e-01
9.28942263e-01 3.00211549e-01 -2.83761919e-01 -1.12472117e-01
2.68819451e-01 4.88267362e-01 8.73857796e-01 -1.74220860e-01
8.95231903e-01 8.60749602e-01 3.95123750e-01 -1.12281394e+00
-1.41640747e+00 -4.63950723e-01 -7.36606419e-01 -3.72190922e-01
7.90235102e-01 -1.02393568e+00 -5.78216672e-01 5.99373937e-01
-9.92242694e-01 -6.46468401e-01 -1.01944104e-01 5.09324729e-01
-6.04164600e-01 4.30884838e-01 -2.55541205e-01 -6.70726478e-01
-3.61595154e-01 -1.16027331e+00 1.16150093e+00 5.60902357e-01
4.06043798e-01 -8.84351909e-01 -1.10637516e-01 3.94960105e-01
3.52220088e-01 -1.35590732e-01 5.86391151e-01 4.93870646e-01
-9.95873928e-01 -1.10743038e-01 -7.62353897e-01 3.72126371e-01
9.89042744e-02 -3.18577915e-01 -1.33215129e+00 -1.50877357e-01
1.98437646e-01 -6.47174299e-01 1.09697831e+00 7.83706725e-01
1.18369579e+00 1.50332078e-01 -1.41042620e-01 9.29121733e-01
1.55276775e+00 -1.07526742e-01 6.85702741e-01 6.92751050e-01
7.65462041e-01 6.97114766e-01 8.75569105e-01 6.45063072e-02
4.70008254e-01 4.92100209e-01 8.04551780e-01 -5.90883493e-01
-2.88991928e-01 4.20054384e-02 1.05712235e-01 3.52323681e-01
-2.02511817e-01 1.10314794e-01 -8.48524332e-01 6.95459485e-01
-1.68774796e+00 -1.16753566e+00 -2.88340658e-01 2.32393909e+00
1.08755553e+00 2.39721969e-01 -6.60567433e-02 1.86110541e-01
4.33294743e-01 1.83897108e-01 -7.34639168e-01 1.23380255e-02
-2.40832925e-01 1.65490538e-01 8.14341664e-01 4.87236649e-01
-1.07552993e+00 1.12475526e+00 6.16366529e+00 4.15031403e-01
-1.48954356e+00 2.65397638e-01 6.04469895e-01 -2.13960990e-01
-7.90634677e-02 4.00801077e-02 -7.23383963e-01 4.68026012e-01
5.34905374e-01 2.10182220e-01 5.48737109e-01 8.83780003e-01
3.16451013e-01 -9.62713420e-01 -8.13706756e-01 1.37579286e+00
2.91184574e-01 -1.15334105e+00 -3.46392810e-01 -2.80028820e-01
1.01934159e+00 3.57933640e-01 1.78935334e-01 -4.63723876e-02
5.61866201e-02 -5.60902238e-01 4.67719316e-01 3.51480246e-01
7.82372594e-01 -5.54651678e-01 5.97528577e-01 4.27444309e-01
-8.57188761e-01 -3.49608473e-02 -9.54655945e-01 -2.17087984e-01
1.85105771e-01 1.02696633e+00 -7.33766973e-01 7.99141228e-02
9.83697772e-01 8.13884020e-01 -7.81694412e-01 1.35946572e+00
-4.69537675e-01 2.94756979e-01 -4.67557788e-01 -1.78741235e-02
2.55549371e-01 -3.53891939e-01 1.52792647e-01 9.04108405e-01
3.13517630e-01 -2.29594842e-01 1.23546906e-01 7.36714244e-01
-3.28613520e-02 -3.19642067e-01 -6.67020917e-01 4.48113382e-01
6.18072040e-02 1.48798859e+00 -9.68111336e-01 -3.76499712e-01
-7.19821453e-01 9.03534532e-01 6.58286288e-02 4.27034318e-01
-7.54333079e-01 -5.85815549e-01 3.32037538e-01 2.88351208e-01
5.75232543e-02 -2.24040657e-01 -5.32986760e-01 -1.32901728e+00
-5.03103342e-03 -4.60691839e-01 -8.33758563e-02 -1.46262598e+00
-8.70913684e-01 5.71254849e-01 -1.10517055e-01 -1.33648431e+00
-1.90048948e-01 -6.73279405e-01 -4.60014045e-01 7.56572485e-01
-2.51502562e+00 -1.04840267e+00 -1.07586825e+00 7.58881867e-01
7.29468048e-01 1.72063559e-01 3.84907752e-01 3.63984466e-01
-4.72300857e-01 9.61856768e-02 5.89434765e-02 -2.59542584e-01
8.67614627e-01 -1.07288241e+00 2.17620656e-01 1.32073331e+00
1.38300642e-01 2.15428159e-01 8.35212409e-01 -2.83817083e-01
-1.10574496e+00 -1.01787186e+00 8.52327347e-01 -1.27033129e-01
4.61664557e-01 -3.58955592e-01 -8.46684635e-01 4.00682151e-01
1.62631929e-01 3.09813291e-01 2.08583027e-01 -2.02192545e-01
8.36267322e-02 -4.09560889e-01 -1.01302493e+00 7.13474214e-01
1.03721118e+00 -6.57889724e-01 -5.31362414e-01 5.62360406e-01
3.56995434e-01 -4.79584455e-01 -2.12223455e-02 3.15145433e-01
3.89299244e-01 -1.41007817e+00 1.13706231e+00 2.12315574e-01
5.37779570e-01 -5.09353101e-01 1.14157125e-01 -1.22920823e+00
-7.03759342e-02 -4.33951020e-01 2.47009933e-01 6.66521728e-01
2.62142062e-01 -7.75713503e-01 1.00915313e+00 3.45822990e-01
-3.25912684e-01 -5.22563815e-01 -6.47655964e-01 -5.29270411e-01
-4.06723410e-01 -3.33561331e-01 3.00849915e-01 7.96116590e-01
-4.49862003e-01 4.10788000e-01 -2.06858829e-01 1.84389710e-01
1.06227303e+00 4.95606601e-01 8.93602490e-01 -1.49042499e+00
-2.52102632e-02 -1.94561481e-01 -1.50354519e-01 -1.05790246e+00
2.20669031e-01 -4.87618089e-01 5.97641051e-01 -1.75671446e+00
1.15294680e-01 -5.04349530e-01 -7.55859092e-02 5.95339954e-01
-1.89964950e-01 8.01176429e-01 -8.65702853e-02 2.29208812e-01
-6.45364583e-01 5.65895677e-01 1.37618661e+00 -1.52878210e-01
-2.58181572e-01 4.57088910e-02 -3.55841130e-01 1.09061015e+00
6.69033170e-01 -2.28071511e-01 -6.53354049e-01 -6.01845622e-01
5.63535631e-01 -3.57530147e-01 5.11494339e-01 -1.36691034e+00
3.36098611e-01 -3.54600608e-01 5.35416961e-01 -7.87738204e-01
5.40904403e-01 -7.46640146e-01 -4.28592741e-01 2.76823282e-01
-1.77171361e-02 -3.97586852e-01 1.25526544e-02 3.62137198e-01
-2.45390654e-01 -4.77442801e-01 1.20020294e+00 -5.29061079e-01
-1.11201000e+00 1.72889978e-01 -6.72049448e-02 -8.67719576e-02
9.92282808e-01 -4.72955912e-01 -6.58166766e-01 -4.84178454e-01
-2.27065146e-01 -4.45077233e-02 8.58482301e-01 4.45654318e-02
7.35435545e-01 -9.50913429e-01 -2.45830387e-01 3.53267521e-01
8.32512379e-02 5.69920063e-01 1.20475158e-01 1.01872504e+00
-8.87188554e-01 4.30472344e-01 -6.30611300e-01 -7.30154872e-01
-1.16155088e+00 4.91179228e-01 3.20010185e-01 1.53063968e-01
-7.05283523e-01 9.54164565e-01 4.86441463e-01 -1.21090025e-01
7.54416734e-02 -5.44429779e-01 -1.70015991e-02 -9.03161168e-02
5.49451530e-01 1.97988674e-01 2.63333231e-01 -5.30295849e-01
-2.04087481e-01 9.00653541e-01 2.82073319e-01 1.37273028e-01
1.56838918e+00 -4.50702697e-01 -1.90685838e-01 3.29680592e-01
9.66174662e-01 -4.12471801e-01 -1.62360084e+00 -3.38290006e-01
-3.95275593e-01 -8.57657969e-01 6.94525898e-01 -3.85462403e-01
-1.20051157e+00 1.27868974e+00 6.98746502e-01 8.90811533e-02
1.51285768e+00 -2.66732723e-01 9.04270887e-01 8.53293955e-01
7.00401545e-01 -1.25915718e+00 3.08786601e-01 5.01245558e-01
5.70723236e-01 -1.71373403e+00 9.98254120e-02 -4.43345010e-01
-3.46109658e-01 1.05447984e+00 7.14491010e-01 -6.46802932e-02
4.65179116e-01 3.44166532e-03 2.40664378e-01 -1.62612990e-01
-4.16643620e-01 -7.65273929e-01 1.40054166e-01 7.68081665e-01
1.91135630e-01 -4.74545419e-01 -7.63645619e-02 -8.08833539e-02
-1.55073434e-01 1.85170308e-01 8.60030711e-01 8.80643308e-01
-8.18754911e-01 -6.92411900e-01 -3.77113789e-01 4.11429822e-01
-2.82585740e-01 -1.78377897e-01 1.00269436e-03 4.48099375e-01
2.56503403e-01 8.27637076e-01 1.18993133e-01 -1.64527576e-02
6.89926222e-02 -1.01184048e-01 6.02990448e-01 -7.20356524e-01
1.30718108e-02 4.57051881e-02 -2.02300996e-01 -6.27360463e-01
-1.03717649e+00 -5.45958400e-01 -1.17004108e+00 -1.37662381e-01
-3.89623612e-01 -4.45041001e-01 8.90079916e-01 8.19524109e-01
-1.26558600e-03 2.73134112e-01 5.03624678e-01 -1.62539291e+00
1.45786889e-02 -7.53165126e-01 -4.86510068e-01 3.69037747e-01
5.71819603e-01 -9.93177295e-01 -4.55578029e-01 1.02887608e-01] | [9.837296485900879, -2.3125617504119873] |
443cd7a5-7565-4964-aa11-f081796e9659 | dr-spider-a-diagnostic-evaluation-benchmark | 2301.08881 | null | https://arxiv.org/abs/2301.08881v2 | https://arxiv.org/pdf/2301.08881v2.pdf | Dr.Spider: A Diagnostic Evaluation Benchmark towards Text-to-SQL Robustness | Neural text-to-SQL models have achieved remarkable performance in translating natural language questions into SQL queries. However, recent studies reveal that text-to-SQL models are vulnerable to task-specific perturbations. Previous curated robustness test sets usually focus on individual phenomena. In this paper, we propose a comprehensive robustness benchmark based on Spider, a cross-domain text-to-SQL benchmark, to diagnose the model robustness. We design 17 perturbations on databases, natural language questions, and SQL queries to measure the robustness from different angles. In order to collect more diversified natural question perturbations, we utilize large pretrained language models (PLMs) to simulate human behaviors in creating natural questions. We conduct a diagnostic study of the state-of-the-art models on the robustness set. Experimental results reveal that even the most robust model suffers from a 14.0% performance drop overall and a 50.7% performance drop on the most challenging perturbation. We also present a breakdown analysis regarding text-to-SQL model designs and provide insights for improving model robustness. | ['Bing Xiang', 'Patrick Ng', 'Vittorio Castelli', 'Zhiguo Wang', 'William Yang Wang', 'Steve Ash', 'Joseph Lilien', 'Jiarong Jiang', 'Sheng Zhang', 'Wuwei Lan', 'Alexander Hanbo Li', 'Henghui Zhu', 'Lin Pan', 'Mingwen Dong', 'Jun Wang', 'Shuaichen Chang'] | 2023-01-21 | null | null | null | null | ['text-to-sql', 'natural-questions'] | ['computer-code', 'miscellaneous'] | [-1.29272386e-01 -2.21315339e-01 1.21623904e-01 -4.69274610e-01
-1.31401765e+00 -9.92068410e-01 5.19232929e-01 6.17396124e-02
-1.97252259e-01 2.19306618e-01 3.83955181e-01 -5.65789104e-01
-6.57979324e-02 -6.97404742e-01 -1.11634386e+00 -3.21291275e-02
1.94427788e-01 2.82843411e-01 4.32364523e-01 -6.87907040e-01
4.60118473e-01 3.81478190e-01 -1.27884471e+00 6.75745785e-01
1.00001085e+00 7.15845048e-01 -3.88280660e-01 7.95391440e-01
-8.63017440e-02 9.89835203e-01 -1.09450579e+00 -6.42434299e-01
2.67444611e-01 1.82166412e-01 -8.07192802e-01 -4.95990455e-01
5.90668797e-01 -5.13840914e-01 -6.24994695e-01 8.36298823e-01
9.03921485e-01 -2.87062675e-03 3.94753247e-01 -1.29239130e+00
-1.06342828e+00 8.66166532e-01 -2.17333183e-01 1.82452098e-01
9.51221645e-01 8.47327292e-01 7.90493250e-01 -8.91125679e-01
3.84718567e-01 1.50914979e+00 5.74407101e-01 4.43913132e-01
-1.37298012e+00 -4.87105399e-01 1.08403526e-01 -4.90944199e-02
-1.15909815e+00 -5.81305504e-01 5.31404138e-01 -2.70571053e-01
1.26910305e+00 5.53413451e-01 -5.19357741e-01 1.61166251e+00
4.15137410e-01 5.69992006e-01 1.01160038e+00 4.86950129e-02
1.82390183e-01 1.97727308e-01 4.56116974e-01 3.72191459e-01
9.39876959e-02 1.26692101e-01 -6.34283602e-01 -3.90479088e-01
1.84986457e-01 -2.24039152e-01 -1.71732113e-01 6.18179403e-02
-1.14876616e+00 5.07299364e-01 2.08569959e-01 -1.05979241e-01
-1.32628053e-01 -4.79674749e-02 7.36213982e-01 8.43143344e-01
8.07928853e-03 1.00176775e+00 -7.66062796e-01 -4.49013472e-01
-3.31792414e-01 7.25273728e-01 1.14166164e+00 1.15854216e+00
3.49603862e-01 9.22450051e-02 -8.13524008e-01 8.12675536e-01
3.42446007e-02 6.04600549e-01 4.76705045e-01 -7.06499696e-01
9.43811238e-01 9.25388098e-01 1.61872715e-01 -1.19674087e+00
-3.50251108e-01 -1.28126785e-01 -5.72457314e-01 -2.69824237e-01
3.02932262e-01 3.66572626e-02 -8.11796010e-01 1.74982071e+00
-6.27540946e-02 -4.87362891e-01 1.63359910e-01 4.61114019e-01
1.01565194e+00 4.86615032e-01 -1.33288831e-01 2.39396974e-01
1.29098356e+00 -6.15678668e-01 -4.09779936e-01 -1.22059315e-01
4.59298134e-01 -7.17859983e-01 2.31220365e+00 4.12159801e-01
-1.13089716e+00 -6.96260095e-01 -8.10161471e-01 -4.87665506e-03
-7.03082502e-01 -3.24687183e-01 -1.33422539e-01 6.19771183e-01
-9.39232349e-01 3.13949287e-01 -6.02594495e-01 -4.87233698e-01
-1.20687388e-01 4.37014533e-04 -1.65488452e-01 -4.46204059e-02
-1.43191826e+00 1.03743458e+00 2.26221949e-01 -4.13278788e-01
-1.25167120e+00 -1.03814399e+00 -6.93341494e-01 1.20531069e-02
7.92459965e-01 -4.58702475e-01 1.53212094e+00 3.19919795e-01
-1.41531849e+00 6.53676867e-01 -4.72177751e-03 -3.86076659e-01
7.44000912e-01 -5.28124809e-01 -4.79762107e-01 -7.97852874e-02
-9.01814774e-02 1.95899010e-01 5.97721398e-01 -1.16578627e+00
9.73861292e-02 -1.58765584e-01 3.79780233e-01 -3.24275166e-01
-4.29293811e-01 3.50316733e-01 -6.46528840e-01 -8.63140702e-01
-2.25671470e-01 -7.84405828e-01 9.17606652e-02 -4.73466545e-01
-9.12545145e-01 -2.83858508e-01 6.17469311e-01 -5.88652372e-01
1.64517999e+00 -1.90716386e+00 -2.37511143e-01 1.29122669e-02
1.69594139e-01 1.42872989e-01 -5.78689694e-01 8.09533894e-01
-2.74511486e-01 6.95220172e-01 -2.32928410e-01 6.70194030e-02
4.09012496e-01 -1.33288085e-01 -9.54016864e-01 -5.90880848e-02
4.56704497e-01 1.12838662e+00 -4.28862602e-01 -2.79844344e-01
-1.74243331e-01 -7.78842568e-02 -9.10027862e-01 6.16827726e-01
-6.40878856e-01 -3.95437144e-02 -3.32124889e-01 9.27637517e-01
6.47425115e-01 -5.68395713e-03 -4.47189689e-01 -2.82193840e-01
2.94581920e-01 4.80136991e-01 -8.78464758e-01 1.29901850e+00
-2.72954434e-01 2.66987443e-01 -3.18156183e-01 -4.56021667e-01
9.93466675e-01 5.10393903e-02 2.18504239e-02 -1.03321457e+00
-1.96671844e-01 -1.05769597e-01 -5.32046705e-02 -1.09093106e+00
6.62354350e-01 3.14017504e-01 -6.34513974e-01 4.94467080e-01
-3.21701556e-01 -5.63449144e-01 4.54889655e-01 3.77531886e-01
1.66313720e+00 -2.86451668e-01 -1.63727671e-01 -9.74990800e-02
3.63573611e-01 -2.17553422e-01 1.87537223e-01 1.31029403e+00
-2.44437784e-01 8.11417460e-01 9.10810292e-01 -3.37193012e-01
-8.92848492e-01 -1.47692502e+00 4.62592170e-02 1.34478998e+00
-3.18859547e-01 -6.34707570e-01 -8.91370237e-01 -7.69706070e-01
2.52826601e-01 1.35428870e+00 -5.14517307e-01 -7.52987862e-01
-4.73372042e-01 -7.98803985e-01 1.30694270e+00 3.15369725e-01
5.78957200e-01 -1.30698967e+00 -4.42960829e-01 -8.77139345e-02
-3.60297441e-01 -1.21847689e+00 -7.04850316e-01 -6.69410229e-02
-4.93407428e-01 -1.10777915e+00 -7.52214342e-02 -3.71207267e-01
2.22243741e-01 2.86235288e-02 1.54674447e+00 -1.46212280e-01
-5.38036078e-02 5.66831827e-01 -4.25424576e-01 -3.47846031e-01
-8.20135117e-01 3.38658065e-01 2.50668526e-01 -5.09227276e-01
2.87362784e-01 -4.58133191e-01 -2.53951877e-01 9.08414721e-01
-1.54068696e+00 -6.12465739e-01 3.92664313e-01 5.32694638e-01
2.70714104e-01 6.40466064e-03 7.80280590e-01 -6.31619215e-01
1.48901856e+00 -5.78902543e-01 -4.95641947e-01 5.91326892e-01
-6.71638310e-01 3.57419372e-01 8.60688031e-01 -7.07900345e-01
-7.94854045e-01 -6.77319765e-01 -1.37168437e-01 -2.55938709e-01
-3.07676077e-01 6.72754824e-01 -5.21234512e-01 1.43898636e-01
1.49360704e+00 4.42018896e-01 -2.08078861e-01 -4.63874340e-01
5.47983348e-01 6.19352400e-01 7.37417042e-01 -1.15803921e+00
1.16279471e+00 -1.64657474e-01 -6.20249689e-01 -7.04686761e-01
-3.15541774e-01 1.27489686e-01 -1.68015920e-02 1.08232357e-01
3.07321817e-01 -5.66519499e-01 -9.97935236e-01 6.29481196e-01
-1.19157434e+00 -5.17756224e-01 -8.71901140e-02 -2.57516265e-01
-6.84734166e-01 4.95048225e-01 -7.86079109e-01 -4.70283508e-01
-4.57612842e-01 -1.30853200e+00 1.00584424e+00 5.13221771e-02
-4.82483774e-01 -3.67174506e-01 3.20657492e-01 4.54524070e-01
9.11853850e-01 2.89947361e-01 1.37702274e+00 -9.70354259e-01
-5.17853737e-01 -4.28116798e-01 -1.94950670e-01 3.30038518e-01
3.33887003e-02 3.90709847e-01 -7.15152502e-01 -2.96050340e-01
2.63915628e-01 -8.03977728e-01 5.20739675e-01 -3.69187474e-01
1.50597036e+00 -5.46801567e-01 1.97930560e-01 6.44163907e-01
1.07023072e+00 2.90073901e-02 8.99873376e-01 4.42205280e-01
3.68880540e-01 4.24971312e-01 4.29827362e-01 3.07084978e-01
6.07786298e-01 5.40941596e-01 5.18674314e-01 3.80758643e-01
3.47982317e-01 -4.21946287e-01 8.00299108e-01 5.35416305e-01
8.57151866e-01 -4.16086197e-01 -1.43713236e+00 3.93382609e-01
-1.46690333e+00 -6.32462204e-01 2.02928737e-01 2.16848826e+00
1.26708341e+00 5.25278449e-01 9.85822529e-02 3.58301327e-02
3.12798947e-01 1.50140241e-01 -9.65256691e-01 -3.95785421e-01
-1.99648097e-01 -1.10936731e-01 -4.38966751e-02 2.69445479e-01
-8.24972689e-01 1.13806295e+00 6.91455984e+00 6.90860927e-01
-1.00324929e+00 -4.25323516e-01 4.93748367e-01 -3.47629964e-01
-5.73582470e-01 -4.08493042e-01 -6.69959247e-01 5.08912206e-01
1.22755992e+00 -4.42672431e-01 6.12356722e-01 8.74679208e-01
3.46307784e-01 1.22495517e-01 -1.32965779e+00 7.82206059e-01
-4.10448648e-02 -9.57234621e-01 6.36460364e-01 -4.96732473e-01
5.21490633e-01 7.54905343e-02 3.15033853e-01 8.07336569e-01
4.49566275e-01 -1.15471685e+00 4.17701513e-01 8.53227496e-01
7.41464555e-01 -7.40435541e-01 4.21655357e-01 4.49200183e-01
-7.09863067e-01 -1.50320560e-01 -3.41045529e-01 2.61092573e-01
-1.63947955e-01 4.42771673e-01 -7.92886794e-01 3.62236708e-01
1.09137034e+00 7.07752034e-02 -1.35658109e+00 5.60189486e-01
8.24088156e-02 6.66050017e-01 -2.64875025e-01 -4.04481851e-02
-1.74994707e-01 3.13972354e-01 5.81669569e-01 1.21331704e+00
5.85584231e-02 -1.36223406e-01 -5.86839765e-02 1.29829037e+00
-3.08332920e-01 -7.60452375e-02 -8.67401719e-01 -3.36414427e-01
6.49893880e-01 8.61449361e-01 5.13501950e-02 -2.16633841e-01
-2.18351200e-01 7.76231945e-01 5.47573507e-01 6.80253565e-01
-1.17414522e+00 -7.17491329e-01 8.06027651e-01 3.07183802e-01
-2.71721631e-01 -2.04592139e-01 -4.45283979e-01 -1.23495054e+00
6.24872744e-01 -1.86806607e+00 2.30894595e-01 -8.96996856e-01
-1.63370335e+00 6.62007034e-01 2.76435435e-01 -7.68168032e-01
-2.87270129e-01 -3.78892839e-01 -7.96280146e-01 8.31382632e-01
-1.07795262e+00 -6.42747343e-01 -3.90698940e-01 6.56856537e-01
3.54373246e-01 -4.01523650e-01 7.12009609e-01 1.55841440e-01
-9.88215029e-01 1.35573959e+00 -1.29096150e-01 1.99327245e-01
1.11754584e+00 -1.12521553e+00 1.27642822e+00 9.15044844e-01
-6.00865126e-01 1.40305018e+00 8.31713200e-01 -6.15928710e-01
-1.78704834e+00 -1.29388058e+00 5.72663188e-01 -1.30830717e+00
7.65514314e-01 -4.90303099e-01 -1.39952898e+00 6.09536469e-01
1.22517332e-01 -3.63978297e-01 5.17259657e-01 -2.43840963e-01
-8.52722883e-01 -2.09590092e-01 -1.39333165e+00 1.24851429e+00
8.87445450e-01 -1.02847886e+00 -7.43705094e-01 2.23803520e-01
1.67574763e+00 -4.79857832e-01 -1.10094357e+00 7.67525971e-01
3.56117845e-01 -1.17240286e+00 1.05246460e+00 -1.18928230e+00
6.60003424e-01 -3.15953493e-01 -5.46118438e-01 -1.08756542e+00
6.32527247e-02 -9.28166926e-01 -1.60572574e-01 1.43988812e+00
6.32243693e-01 -7.88668931e-01 7.10045457e-01 1.05636263e+00
5.67776263e-02 -7.45091558e-01 -5.99145830e-01 -8.68795633e-01
3.99945170e-01 -6.09969139e-01 7.94153154e-01 7.93163180e-01
2.21008454e-02 2.42885515e-01 -1.08309984e-01 9.75119919e-02
2.77186066e-01 -3.96170378e-01 1.31806600e+00 -5.37841082e-01
-3.19364727e-01 -5.74646592e-01 2.57133722e-01 -1.04332280e+00
8.61532055e-03 -5.87906718e-01 1.32218808e-01 -9.71297443e-01
1.21677130e-01 1.73753157e-01 -5.83078787e-02 3.76751602e-01
-6.51548207e-01 -4.96308804e-01 2.46215969e-01 -1.38581291e-01
-5.33634365e-01 7.02187240e-01 7.32147992e-01 -2.85705417e-01
-2.22308293e-01 3.85062657e-02 -1.06850886e+00 2.33205736e-01
1.13633633e+00 -5.29673100e-01 -5.54126084e-01 -5.95073164e-01
5.91237962e-01 7.14193881e-02 2.08007917e-01 -1.03305674e+00
2.71776259e-01 -2.71066546e-01 -2.04997420e-01 -6.63093090e-01
-6.60980865e-02 -4.98792768e-01 -3.59600484e-01 4.50217336e-01
-7.71288097e-01 8.34995687e-01 6.30206823e-01 3.69340926e-01
-1.67018458e-01 9.82155055e-02 6.85033381e-01 2.10790373e-02
-1.63759753e-01 1.12935886e-01 -3.15781057e-01 7.28989661e-01
6.82278633e-01 2.68651962e-01 -8.02081585e-01 -6.61157906e-01
-2.39814222e-01 5.59239268e-01 5.52715600e-01 9.32704866e-01
8.29196930e-01 -1.06815696e+00 -8.68969142e-01 2.67000556e-01
5.84494472e-01 -1.19259775e-01 1.65847987e-01 3.17170560e-01
-5.12702405e-01 6.14867330e-01 7.44848102e-02 -5.01523614e-01
-8.18931341e-01 1.03826690e+00 5.32065749e-01 -4.00574833e-01
9.74670351e-02 6.55894458e-01 -4.87875678e-02 -1.31618762e+00
6.14924490e-01 -7.49170482e-01 2.16650113e-01 -3.15246522e-01
4.15834874e-01 6.48878932e-01 2.44446248e-01 5.89801818e-02
-4.34932500e-01 4.97740567e-01 -3.70582759e-01 -4.49716710e-02
9.02477145e-01 2.94733793e-01 -1.50443241e-01 2.29985684e-01
1.19969845e+00 -5.46267591e-02 -7.88283706e-01 -4.20880377e-01
3.22563499e-01 -1.44837409e-01 -7.61366069e-01 -1.13393247e+00
-6.76839173e-01 7.26328254e-01 2.59013712e-01 4.24234569e-01
1.09919834e+00 -2.82385916e-01 5.89170277e-01 1.10885608e+00
1.91544265e-01 -9.05197978e-01 6.51539266e-01 1.00034678e+00
1.65754080e+00 -1.32063985e+00 -3.64263326e-01 9.30327848e-02
-6.56484187e-01 8.53293538e-01 1.20909345e+00 1.12599336e-01
4.67269063e-01 4.20364320e-01 2.26275012e-01 -1.14828065e-01
-1.13676858e+00 3.45734358e-01 9.13467929e-02 5.26334167e-01
3.01041543e-01 -3.24547559e-01 2.89974660e-01 7.33658791e-01
-5.76880991e-01 -1.17732339e-01 6.07257843e-01 8.60346496e-01
-1.83234528e-01 -9.28477108e-01 -6.13077521e-01 4.91936505e-01
-4.13190395e-01 -3.10626477e-01 -8.74259710e-01 6.83796644e-01
-7.29316950e-01 1.48182452e+00 -4.73533571e-01 -1.02046990e+00
1.27794158e+00 2.86571294e-01 -8.73697326e-02 -5.16509354e-01
-1.01880980e+00 -7.08151996e-01 1.38627261e-01 -1.00042200e+00
7.21955657e-01 -1.79042220e-01 -9.67820406e-01 -7.52738714e-01
1.92317367e-01 -1.88368876e-02 3.42205703e-01 4.17897940e-01
7.74356604e-01 6.72710598e-01 6.12260938e-01 -2.63629675e-01
-1.49070513e+00 -1.14982712e+00 7.42938742e-02 7.12032557e-01
4.36911345e-01 -9.95072201e-02 -4.76666182e-01 -3.87721688e-01] | [6.293640613555908, 8.16315746307373] |
4e066522-be53-4d9a-9b29-b198ab5ff310 | smart-director-an-event-driven-directing | 2201.04024 | null | https://arxiv.org/abs/2201.04024v1 | https://arxiv.org/pdf/2201.04024v1.pdf | Smart Director: An Event-Driven Directing System for Live Broadcasting | Live video broadcasting normally requires a multitude of skills and expertise with domain knowledge to enable multi-camera productions. As the number of cameras keep increasing, directing a live sports broadcast has now become more complicated and challenging than ever before. The broadcast directors need to be much more concentrated, responsive, and knowledgeable, during the production. To relieve the directors from their intensive efforts, we develop an innovative automated sports broadcast directing system, called Smart Director, which aims at mimicking the typical human-in-the-loop broadcasting process to automatically create near-professional broadcasting programs in real-time by using a set of advanced multi-view video analysis algorithms. Inspired by the so-called "three-event" construction of sports broadcast, we build our system with an event-driven pipeline consisting of three consecutive novel components: 1) the Multi-view Event Localization to detect events by modeling multi-view correlations, 2) the Multi-view Highlight Detection to rank camera views by the visual importance for view selection, 3) the Auto-Broadcasting Scheduler to control the production of broadcasting videos. To our best knowledge, our system is the first end-to-end automated directing system for multi-camera sports broadcasting, completely driven by the semantic understanding of sports events. It is also the first system to solve the novel problem of multi-view joint event detection by cross-view relation modeling. We conduct both objective and subjective evaluations on a real-world multi-camera soccer dataset, which demonstrate the quality of our auto-generated videos is comparable to that of the human-directed. Thanks to its faster response, our system is able to capture more fast-passing and short-duration events which are usually missed by human directors. | ['Tao Mei', 'Jingen Liu', 'Ting Yao', 'Ning Zhang', 'Qian Bao', 'Yue Chen', 'Yingwei Pan'] | 2022-01-11 | null | null | null | null | ['highlight-detection'] | ['computer-vision'] | [-3.31443455e-03 -2.48207048e-01 1.00942282e-02 -3.89357805e-01
-9.77108300e-01 -8.25049281e-01 3.80939275e-01 -3.16560455e-02
-1.23875521e-01 2.81938072e-02 2.94580370e-01 1.49544120e-01
-2.62894388e-02 -6.89736545e-01 -9.41982388e-01 -3.66711974e-01
2.08894536e-02 7.08918452e-01 8.78358424e-01 -4.09142196e-01
2.49430701e-01 3.02161962e-01 -2.13683558e+00 8.67329955e-01
1.99083760e-01 8.43159080e-01 5.25087893e-01 1.04040146e+00
2.14026585e-01 1.20021355e+00 -4.59829330e-01 -4.10877943e-01
1.02290802e-01 -4.84187752e-01 -4.99867707e-01 5.81066787e-01
4.45412606e-01 -4.07104135e-01 -1.02306873e-01 5.97137213e-01
6.75486684e-01 2.04533085e-01 -7.71001354e-02 -1.17151046e+00
1.38223991e-01 7.03421772e-01 -5.45594513e-01 6.03679836e-01
9.96229649e-01 2.17912838e-01 9.43683863e-01 -7.82924473e-01
1.00579798e+00 1.02344978e+00 4.21558082e-01 1.31461486e-01
-9.33910728e-01 -5.26406527e-01 3.48591149e-01 6.06923342e-01
-1.16895783e+00 -5.12755930e-01 8.12728643e-01 -6.08640015e-01
6.56425059e-01 4.79950488e-01 1.09993303e+00 1.21450949e+00
-1.03088178e-01 9.13763702e-01 9.19544637e-01 -3.86471272e-01
6.85642138e-02 1.05022736e-01 -2.24009737e-01 4.90408033e-01
-3.78304452e-01 -2.08495255e-03 -1.20320082e+00 1.38830423e-01
6.47862673e-01 1.62564889e-02 -3.59844953e-01 -4.34794337e-01
-1.58559668e+00 6.52956665e-01 -3.83976042e-01 1.97835177e-01
-5.35325170e-01 9.58461408e-03 4.75611061e-01 2.00200215e-01
4.10049528e-01 2.88954407e-01 -2.12728888e-01 -6.85879827e-01
-1.14836204e+00 5.66485822e-01 9.58450556e-01 1.06195652e+00
2.55520910e-01 -2.25588143e-01 -1.45760342e-01 7.45388150e-01
-8.54259264e-03 2.38405570e-01 3.70163620e-02 -1.23637521e+00
3.94779056e-01 5.11911273e-01 4.51957956e-02 -1.03202629e+00
-4.49444294e-01 -5.49617112e-01 -1.95366830e-01 7.64236376e-02
2.98346579e-01 1.68993920e-01 -3.01220149e-01 1.29706502e+00
5.56732655e-01 2.15409428e-01 -4.26689237e-01 1.40718412e+00
8.78531575e-01 7.12599814e-01 -3.31939697e-01 -2.74890423e-01
1.93074453e+00 -9.51690614e-01 -6.40138686e-01 -2.12786105e-02
1.54384077e-01 -1.12090707e+00 1.04115248e+00 1.01781631e+00
-1.47167122e+00 -6.63736284e-01 -9.03468490e-01 1.38120353e-01
7.16863722e-02 1.32476315e-01 5.61525643e-01 2.99821347e-01
-6.46821380e-01 1.84625700e-01 -8.16111982e-01 -4.82223511e-01
2.74245255e-02 4.91826609e-02 -4.93096590e-01 -3.17688644e-01
-9.95319366e-01 5.08516014e-01 1.88580111e-01 -2.03078270e-01
-1.39805710e+00 -9.17886198e-01 -5.79863489e-01 8.35978016e-02
9.95683074e-01 -7.04318345e-01 1.50108051e+00 -1.22498286e+00
-1.48031986e+00 1.01386929e+00 1.00615904e-01 -7.14718997e-02
4.60574031e-01 -3.53132635e-01 -4.78237361e-01 6.05401635e-01
2.68874913e-01 3.69191587e-01 6.49783969e-01 -1.21927977e+00
-1.20386374e+00 -2.15634346e-01 3.63016814e-01 5.19248486e-01
-1.42610118e-01 5.60257912e-01 -1.25298548e+00 -6.96588397e-01
2.77638901e-02 -9.98457551e-01 8.01897570e-02 -3.79815131e-01
-6.55069649e-02 5.64743951e-02 6.15998387e-01 -5.32722414e-01
1.31721413e+00 -2.31599522e+00 4.11227286e-01 3.78505141e-02
1.74611852e-01 -2.20523596e-01 2.03458577e-01 6.19428694e-01
-5.98422699e-02 -5.87946177e-01 5.29738188e-01 -1.44357383e-01
-1.11350074e-01 9.59159285e-02 -1.97501302e-01 4.47877735e-01
-2.42806256e-01 1.38810277e-01 -1.11230874e+00 -7.56188035e-01
2.51559347e-01 1.83937132e-01 -7.37731755e-01 5.15487850e-01
-2.22700000e-01 4.50906396e-01 -3.19995672e-01 8.91976237e-01
3.71310145e-01 -2.51154721e-01 1.44293725e-01 -2.36775532e-01
-4.71730679e-01 -6.82567954e-02 -1.55905151e+00 2.08714867e+00
-3.20642769e-01 7.24874139e-01 3.54114115e-01 -8.05603266e-01
5.47606230e-01 4.77224708e-01 7.72796273e-01 -4.74623829e-01
-8.05478171e-02 -8.95977691e-02 -3.35536838e-01 -9.59890664e-01
8.66451740e-01 1.17694169e-01 -3.30857128e-01 3.51562768e-01
1.46223858e-01 4.23348062e-02 7.90152490e-01 6.30925477e-01
1.22081602e+00 3.38303238e-01 -3.32687311e-02 2.21557721e-01
4.35414255e-01 2.95565248e-01 6.74008965e-01 6.01309001e-01
4.80353087e-02 9.84878182e-01 7.05257714e-01 -5.19563735e-01
-8.00193012e-01 -9.77896273e-01 4.26493108e-01 1.64240837e+00
2.79821455e-01 -7.19805121e-01 -7.46779621e-01 -4.11085397e-01
-4.38346386e-01 4.43748087e-01 -1.73858956e-01 1.74345613e-01
-6.62396848e-01 -2.16204301e-01 3.34207237e-01 3.56552958e-01
1.03781894e-01 -7.91443586e-01 -8.49775016e-01 5.41544020e-01
-6.28952980e-01 -1.63132513e+00 -7.48367667e-01 -1.40277639e-01
-3.55735362e-01 -1.25899863e+00 -5.81408441e-01 -5.67870617e-01
2.31253624e-01 4.96706843e-01 1.53449929e+00 -1.27817854e-01
-3.18541944e-01 9.00599122e-01 -8.41430843e-01 -3.93823534e-01
-3.35560709e-01 -1.04807310e-01 -2.27142885e-01 2.40854621e-01
2.20805570e-01 -5.78229785e-01 -8.18664789e-01 8.55421305e-01
-9.24206674e-01 4.88919199e-01 4.86496240e-01 4.69665438e-01
8.00085187e-01 1.79982439e-01 -1.10333851e-02 -6.59326494e-01
1.87301189e-01 -4.91646111e-01 -8.02863300e-01 1.58058360e-01
-7.79494867e-02 -6.12588763e-01 3.52653950e-01 -5.77035785e-01
-1.08687627e+00 3.94409388e-01 1.62785739e-01 -6.63347661e-01
-3.61544758e-01 2.83126503e-01 -1.10588826e-01 3.51310372e-01
5.38535237e-01 2.83351004e-01 -3.33704680e-01 -3.73473138e-01
3.56947035e-01 4.48533207e-01 8.10799837e-01 -5.01307249e-01
3.58630598e-01 7.48593807e-01 -2.72384375e-01 -5.95996618e-01
-8.15669596e-01 -9.47765112e-01 -4.22338843e-01 -1.19323778e+00
1.12717235e+00 -1.44176257e+00 -1.10138273e+00 3.41704696e-01
-1.20614552e+00 -2.95578726e-02 -1.94064096e-01 7.00585246e-01
-6.90381467e-01 2.79484004e-01 -5.65120101e-01 -5.53254604e-01
1.57186821e-01 -1.29654729e+00 1.36374128e+00 1.62927642e-01
-1.19335629e-01 -4.66514885e-01 1.01465777e-01 1.00896180e+00
4.78288010e-02 1.97656319e-01 2.29926750e-01 -4.37977612e-01
-9.27642047e-01 -1.24935612e-01 1.57876030e-01 1.95865214e-01
-5.25065243e-01 1.13350071e-01 -8.22766960e-01 -4.39468063e-02
-3.64032574e-02 -5.32647781e-02 5.09952068e-01 5.12853801e-01
1.04259920e+00 3.14636886e-01 -2.08147913e-01 5.77541411e-01
1.08020949e+00 1.61376089e-01 4.32807714e-01 4.78032202e-01
5.13319314e-01 6.73212886e-01 1.14363599e+00 8.19471121e-01
7.70778477e-01 1.15251040e+00 5.38864195e-01 -1.80156991e-01
-1.21967070e-01 -2.54947543e-01 6.98113739e-01 8.83832812e-01
-5.04527688e-01 -3.74610186e-01 -8.23159814e-01 5.59206367e-01
-2.06156254e+00 -1.44489646e+00 -4.58063036e-01 1.96038890e+00
3.85185540e-01 2.51029789e-01 5.56523740e-01 5.07676788e-02
8.72794151e-01 1.37819871e-01 2.67553399e-03 -1.92291126e-01
4.98729870e-02 -1.87607631e-01 3.02665502e-01 -8.36188197e-02
-1.20771372e+00 7.85397232e-01 5.63873625e+00 9.01779175e-01
-8.86442840e-01 2.48659745e-01 2.81776994e-01 -7.74303257e-01
-7.31332786e-03 1.76627919e-01 -9.52051222e-01 4.85919327e-01
7.56584466e-01 2.99227029e-01 4.88915503e-01 9.32584822e-01
6.07657254e-01 -4.24272388e-01 -1.14129496e+00 1.15079892e+00
4.45306867e-01 -1.62206769e+00 -2.80656040e-01 -1.70292128e-02
4.37908113e-01 -9.89904329e-02 -3.77911597e-01 2.46145263e-01
1.88689038e-01 -3.61826032e-01 1.44708693e+00 6.35457933e-01
4.44914967e-01 -7.93410301e-01 4.40886140e-01 4.83988315e-01
-1.41670942e+00 -1.96772411e-01 1.21675953e-01 7.48136640e-03
7.97738135e-01 7.48567343e-01 -5.47574878e-01 6.86231434e-01
1.07034087e+00 7.92964756e-01 -3.05852264e-01 9.79615033e-01
-7.15553164e-02 5.75654209e-01 -1.62705764e-01 3.49761963e-01
-5.13133444e-02 3.06825079e-02 9.35742676e-01 1.26684999e+00
4.68596846e-01 2.55370378e-01 5.83309889e-01 4.48752552e-01
2.86961675e-01 3.17583717e-02 -4.33598280e-01 1.20939009e-01
2.15306342e-01 1.44704127e+00 -9.49959815e-01 -3.75098169e-01
-6.57752931e-01 8.25147927e-01 -1.36994556e-01 -1.20236557e-02
-1.33377385e+00 -1.32094204e-01 3.47555488e-01 5.79751253e-01
5.95087230e-01 -8.62331614e-02 3.40586841e-01 -1.21046698e+00
1.87402189e-01 -1.36109650e+00 7.46325612e-01 -1.30277431e+00
-1.03716850e+00 6.19860232e-01 2.40788415e-01 -1.61986411e+00
-2.64809787e-01 -2.73156583e-01 -3.62626284e-01 2.36542195e-01
-1.16782725e+00 -1.27309668e+00 -4.75226045e-01 9.15442586e-01
1.07776976e+00 -1.85020298e-01 4.32878703e-01 8.26214910e-01
-3.77631962e-01 3.02907582e-02 -4.20676798e-01 -1.74108654e-01
9.18335140e-01 -1.09407723e+00 -3.02731454e-01 9.21226084e-01
4.17491376e-01 5.85580058e-03 8.92388344e-01 -3.88601035e-01
-1.74118936e+00 -5.98561347e-01 6.02395415e-01 -5.35973012e-01
6.75490081e-01 -6.55698955e-01 -3.50744069e-01 6.06191814e-01
2.42809013e-01 -2.56738991e-01 7.75660336e-01 1.18124016e-01
2.12571993e-02 -4.90496248e-01 -4.41444010e-01 2.56362617e-01
1.08465648e+00 -2.80927211e-01 -5.11634290e-01 6.49326384e-01
4.73982453e-01 -9.27314878e-01 -9.83632505e-01 1.55135408e-01
6.26195550e-01 -1.43449247e+00 9.72598553e-01 -1.28796235e-01
6.45203948e-01 -4.52871919e-01 -9.74660143e-02 -9.20145810e-01
-1.77573040e-01 -7.15620637e-01 1.90018445e-01 1.43629444e+00
1.46492407e-01 1.74971178e-01 7.20884800e-01 1.14485495e-01
-4.56293017e-01 -3.78913313e-01 -7.98509717e-01 -5.20049334e-01
-1.15468621e+00 -9.29576039e-01 2.62731045e-01 7.57822573e-01
-3.52633372e-02 2.90323734e-01 -6.11986876e-01 4.52281535e-01
4.17401969e-01 4.77203995e-01 1.25311637e+00 -1.08574486e+00
-1.00419521e+00 -5.05108200e-02 -4.57977563e-01 -1.04193807e+00
-3.08875501e-01 -3.42080593e-01 1.14307754e-01 -1.29370701e+00
4.30280298e-01 1.51372313e-01 2.58621480e-02 1.95421167e-02
2.11548269e-01 9.51050147e-02 2.38923773e-01 2.74435014e-01
-1.31382394e+00 1.63677253e-03 1.21568024e+00 3.21299285e-01
-6.43282309e-02 3.08225513e-01 -4.59713370e-01 1.04433417e+00
3.74758057e-02 -6.50072753e-01 -4.18068141e-01 -4.40190047e-01
7.35449672e-01 6.78635359e-01 4.99472648e-01 -1.12150383e+00
5.74277818e-01 -1.16243139e-01 5.02391011e-02 -8.63702834e-01
4.56291676e-01 -8.31723094e-01 5.75297058e-01 3.73248085e-02
-1.45209506e-01 3.46098483e-01 -9.70159248e-02 6.31152928e-01
-6.52414799e-01 9.08856913e-02 4.21475053e-01 -4.29176033e-01
-1.02437329e+00 7.42968777e-03 -8.17522347e-01 -4.11175303e-02
1.37656510e+00 -3.39524299e-01 -1.46205306e-01 -6.79828048e-01
-1.01033998e+00 3.21439922e-01 3.29827219e-01 4.60174829e-01
4.73595858e-01 -1.01692295e+00 -7.16860712e-01 2.54762955e-02
3.07399720e-01 -1.51594177e-01 7.56401658e-01 1.08677328e+00
-7.54503727e-01 -1.73067525e-02 -5.61358929e-02 -1.14742863e+00
-1.72164214e+00 4.53768075e-01 -1.07920058e-01 -5.65612972e-01
-7.14710593e-01 7.71662533e-01 1.45469576e-01 1.96119800e-01
2.92378753e-01 -1.07735001e-01 -3.04711491e-01 4.12141114e-01
5.96260965e-01 5.05447984e-01 9.49800536e-02 -6.12065375e-01
-3.36047173e-01 4.65295911e-01 9.58664045e-02 -3.62051815e-01
1.68100488e+00 -3.52934569e-01 1.25051543e-01 4.75194752e-01
5.50457656e-01 1.49044052e-01 -1.34377027e+00 1.32736996e-01
-3.44880372e-01 -6.80007160e-01 1.95039213e-01 -6.30786359e-01
-1.25518513e+00 6.32764697e-01 5.63556492e-01 4.02777821e-01
1.35550141e+00 2.27851123e-01 8.32143784e-01 -1.35247603e-01
5.28632700e-01 -1.39841259e+00 2.73719609e-01 1.25088692e-01
8.36970627e-01 -1.09359896e+00 1.67374074e-01 -5.14785707e-01
-1.02304363e+00 1.10715675e+00 5.80009937e-01 1.42479122e-01
3.43580663e-01 4.91795272e-01 8.96949023e-02 -6.65512502e-01
-1.22979701e+00 -1.66238457e-01 1.94407314e-01 3.65459025e-01
1.92261025e-01 -2.05232516e-01 -4.13568020e-02 7.27063239e-01
-2.69297808e-02 1.87907070e-01 6.90785825e-01 9.29154217e-01
-3.65959615e-01 -9.50366974e-01 -7.12343097e-01 -2.29858160e-01
-5.91249824e-01 2.47972563e-01 -1.01869486e-01 8.35061848e-01
4.51596886e-01 1.06038833e+00 1.03836916e-01 -4.21301991e-01
1.00403261e+00 -2.75287807e-01 3.87438148e-01 -6.52368248e-01
-1.03678942e+00 5.31041861e-01 3.53478640e-01 -9.69318748e-01
-7.18274355e-01 -8.70238125e-01 -8.30983400e-01 -2.81388134e-01
-2.16259554e-01 6.24266155e-02 7.66603410e-01 6.67440236e-01
4.93167341e-01 8.12019587e-01 6.72893882e-01 -1.12435615e+00
-6.14174525e-04 -5.74439943e-01 -6.10676229e-01 6.77792847e-01
-1.87639758e-01 -6.56705499e-01 -1.28584340e-01 5.90405226e-01] | [7.94981575012207, 0.14656655490398407] |
6a7bc931-8c3c-4827-bc0c-ea5fe0b53b4c | lwposr-lightweight-efficient-fine-grained | 2202.03544 | null | https://arxiv.org/abs/2202.03544v1 | https://arxiv.org/pdf/2202.03544v1.pdf | LwPosr: Lightweight Efficient Fine-Grained Head Pose Estimation | This paper presents a lightweight network for head pose estimation (HPE) task. While previous approaches rely on convolutional neural networks, the proposed network \textit{LwPosr} uses mixture of depthwise separable convolutional (DSC) and transformer encoder layers which are structured in two streams and three stages to provide fine-grained regression for predicting head poses. The quantitative and qualitative demonstration is provided to show that the proposed network is able to learn head poses efficiently while using less parameter space. Extensive ablations are conducted using three open-source datasets namely 300W-LP, AFLW2000, and BIWI datasets. To our knowledge, (1) \textit{LwPosr} is the lightest network proposed for estimating head poses compared to both keypoints-based and keypoints-free approaches; (2) it sets a benchmark for both overperforming the previous lightweight network on mean absolute error and on reducing number of parameters; (3) it is first of its kind to use mixture of DSCs and transformer encoders for HPE. This approach is suitable for mobile devices which require lightweight networks. | ['Naina Dhingra'] | 2022-02-07 | null | null | null | null | ['head-pose-estimation'] | ['computer-vision'] | [-2.97368348e-01 4.83851105e-01 5.71189858e-02 -5.87336361e-01
-9.64538157e-01 -3.52787599e-02 5.01542747e-01 -2.17987970e-01
-7.85079062e-01 6.80747807e-01 4.59222525e-01 -1.14426069e-01
-1.10168040e-01 -5.13303995e-01 -9.50662732e-01 -4.99149472e-01
-4.95941669e-01 5.36895335e-01 2.80237377e-01 -4.64643508e-01
-8.92574936e-02 8.73412609e-01 -1.89788616e+00 -2.08969086e-01
2.86779881e-01 1.18501472e+00 6.60352930e-02 6.11884296e-01
4.81019199e-01 6.98129892e-01 -3.08750778e-01 -4.54140753e-01
2.29923561e-01 1.41121149e-01 -7.35301435e-01 -3.45518678e-01
9.10335302e-01 -7.10440934e-01 -5.94216883e-01 5.71540534e-01
1.31179690e+00 8.65708143e-02 5.53589582e-01 -1.52330518e+00
-1.82328179e-01 5.99135697e-01 -5.09166539e-01 8.43752027e-02
6.11455560e-01 7.17770755e-02 7.33366549e-01 -1.04832339e+00
3.51792872e-01 1.29420102e+00 1.13349986e+00 7.30953932e-01
-5.43943286e-01 -1.07791555e+00 2.05617458e-01 3.68335038e-01
-1.51210165e+00 -8.70483935e-01 6.32052243e-01 7.33605446e-03
1.16788554e+00 2.08188295e-01 6.16760850e-01 1.32183528e+00
-2.06568718e-01 1.24713838e+00 8.30850482e-01 -2.83057421e-01
2.02371553e-01 -3.19089621e-01 2.47861773e-01 8.50300074e-01
2.65852541e-01 1.53355941e-01 -9.92195904e-01 5.48066124e-02
3.85222793e-01 -1.87912658e-01 -4.71395671e-01 -1.41581938e-01
-8.27058911e-01 7.75664866e-01 6.38530850e-01 1.84549764e-02
-2.82044232e-01 2.38246366e-01 4.63655084e-01 2.63601076e-02
4.73052382e-01 -1.73413917e-01 -4.80078965e-01 -1.44669786e-01
-1.31161726e+00 5.67462146e-01 1.00762200e+00 1.36182988e+00
4.70268339e-01 -1.03420252e-02 8.59618261e-02 7.62530744e-01
5.52259684e-01 6.37029827e-01 7.02744424e-01 -6.16276026e-01
5.81830919e-01 2.46553108e-01 -1.50847286e-01 -6.92078054e-01
-1.15021825e+00 -1.79893672e-01 -7.72824466e-01 2.12129831e-01
2.52547652e-01 -3.90752643e-01 -1.09384620e+00 1.74477601e+00
3.77913117e-01 2.61157542e-01 -2.77818263e-01 7.51706541e-01
1.42668200e+00 2.34278709e-01 -9.72955227e-02 1.19223393e-01
1.32946289e+00 -1.01397836e+00 -6.12446666e-01 -3.62383842e-01
6.26529217e-01 -5.06825447e-01 6.68743193e-01 4.08418298e-01
-1.37395442e+00 -3.61881822e-01 -1.21796811e+00 -4.12897885e-01
-4.54499960e-01 1.14597969e-01 6.90289319e-01 8.22126865e-01
-1.65497208e+00 4.30429071e-01 -9.92487729e-01 -5.39617002e-01
4.71616179e-01 9.91627455e-01 -7.17619836e-01 1.15247093e-01
-1.10489142e+00 1.06565297e+00 1.17743857e-01 5.19991040e-01
-8.01632583e-01 -5.76991498e-01 -1.20277631e+00 -2.72789635e-02
3.45447473e-02 -5.22079527e-01 1.39377952e+00 -3.61354142e-01
-1.79276752e+00 7.25752890e-01 -1.76863059e-01 -7.62558341e-01
7.45482147e-01 -6.02969646e-01 -2.21027166e-01 -2.67123133e-02
-1.16787814e-01 1.10923374e+00 7.70734787e-01 -8.32389951e-01
-7.37237215e-01 -7.00729847e-01 -6.37402758e-02 2.38752645e-02
-4.38741952e-01 1.09697409e-01 -6.79519653e-01 -5.16360700e-01
1.98463440e-01 -1.14773440e+00 1.78250924e-01 1.10644042e-01
-6.03018343e-01 -4.22366083e-01 8.62957776e-01 -7.23859489e-01
1.02515173e+00 -1.74571836e+00 -6.01944290e-02 1.10797428e-01
3.38867128e-01 4.09295440e-01 -1.20445706e-01 2.48843879e-01
-2.74319947e-01 -4.56181794e-01 -2.66474169e-02 -1.00307202e+00
2.73752213e-01 6.63154796e-02 -1.81503128e-02 1.02153921e+00
-2.34332815e-01 9.04068470e-01 -3.76042426e-01 -4.44218129e-01
3.22946519e-01 1.01818681e+00 -7.57640123e-01 2.84231782e-01
2.25972816e-01 1.39020840e-02 -2.72827409e-02 6.23999298e-01
1.01330900e+00 3.76736075e-01 -3.17254484e-01 -5.04386067e-01
-1.50019646e-01 2.58018225e-01 -1.38171566e+00 1.64316654e+00
-5.08009195e-01 7.13791370e-01 2.09959522e-01 -6.96236551e-01
6.61782563e-01 4.51197088e-01 3.21867555e-01 -8.01698565e-01
3.95841569e-01 3.20220172e-01 -4.36520100e-01 -4.83344048e-01
4.68957782e-01 2.24688262e-01 -2.44891681e-02 2.14117751e-01
4.73230034e-01 2.63058126e-01 2.61063012e-03 -1.88213885e-01
1.07375908e+00 3.31245273e-01 1.22182585e-01 -3.65932524e-01
5.83140790e-01 -8.36646020e-01 4.31240052e-01 4.71745372e-01
-3.93247515e-01 8.03038895e-01 1.62105396e-01 -5.62522292e-01
-7.18790829e-01 -8.36564898e-01 -2.09479079e-01 1.51327658e+00
-2.59044498e-01 -4.79470074e-01 -9.65748191e-01 -4.00127500e-01
-4.25203256e-02 4.18455690e-01 -7.28201330e-01 8.42681602e-02
-9.00671959e-01 -6.31181240e-01 1.07617378e+00 1.01032054e+00
6.96058154e-01 -9.52701032e-01 -6.70965195e-01 -4.77024121e-04
-7.70665109e-02 -1.07851410e+00 -4.65048492e-01 4.48706299e-01
-4.53803033e-01 -1.17044866e+00 -1.03672063e+00 -8.21172774e-01
3.58604610e-01 -3.42927091e-02 9.00996923e-01 -1.42345130e-01
-1.42390817e-01 3.93805444e-01 -3.05191576e-01 -7.76581347e-01
3.49016488e-01 5.33851445e-01 2.58742094e-01 -3.18906277e-01
6.14390075e-01 -7.52286792e-01 -8.55305314e-01 1.86552957e-01
-4.83468324e-01 -1.78699240e-01 5.70427597e-01 5.07490993e-01
1.10849395e-01 -5.79638660e-01 3.52193832e-01 -7.42548883e-01
4.29172426e-01 -3.40119630e-01 -4.87792432e-01 -7.42059126e-02
-3.59596431e-01 1.16234995e-01 3.10126930e-01 -1.26490101e-01
-9.67651665e-01 2.79856712e-01 -8.06672871e-01 -5.73081449e-02
-4.42293674e-01 -2.26046275e-02 -1.36308119e-01 -1.92950413e-01
5.44905543e-01 1.85215145e-01 -6.59403801e-02 -6.55610263e-01
1.73865765e-01 8.56285572e-01 7.13020444e-01 -2.84319907e-01
6.91348910e-01 5.11855781e-01 6.74752966e-02 -1.08174157e+00
-6.51630640e-01 -5.41999400e-01 -9.49592233e-01 3.19610275e-02
8.71172905e-01 -1.17286694e+00 -1.19833207e+00 7.99195766e-01
-1.02527118e+00 -3.50242108e-01 1.83275893e-01 4.41822350e-01
-5.29451728e-01 1.37676358e-01 -6.34596884e-01 -9.29291129e-01
-6.93454683e-01 -1.13468039e+00 1.63372064e+00 4.16307896e-02
-2.46129051e-01 -7.48901248e-01 -5.49251102e-02 2.53498673e-01
5.69993675e-01 1.99153990e-01 2.35141218e-01 -6.10994458e-01
-1.65468425e-01 -5.21987855e-01 -1.19789995e-01 -7.19727203e-02
-2.90583372e-01 -2.35978320e-01 -1.43698239e+00 -6.94316447e-01
-3.09440076e-01 -5.18356979e-01 7.33294487e-01 5.19272387e-01
1.12191641e+00 -3.10164332e-01 -2.58181870e-01 1.06562150e+00
9.60693061e-01 -9.92586911e-02 7.58131683e-01 6.57441080e-01
7.89828241e-01 4.50327188e-01 -3.31380405e-02 5.04155874e-01
1.05072784e+00 9.50180948e-01 5.58728874e-01 -3.34625930e-01
-2.76625514e-01 -1.48332000e-01 5.01964331e-01 6.18547678e-01
-3.50508541e-01 -6.90897554e-02 -7.59138763e-01 4.59740281e-01
-1.74967360e+00 -6.71856582e-01 1.77696228e-01 2.10958076e+00
5.17626405e-01 4.82666194e-02 5.81342280e-01 3.72861207e-01
2.62662143e-01 2.48179123e-01 -4.77681965e-01 -1.94083422e-01
1.73503652e-01 6.33663237e-01 6.50957346e-01 4.03658032e-01
-1.24028361e+00 7.77599990e-01 5.78155136e+00 5.10446608e-01
-1.13769102e+00 1.98441222e-01 2.17131019e-01 -4.25226748e-01
3.46252441e-01 -6.64855838e-01 -1.40637493e+00 3.05709153e-01
1.15114808e+00 3.58232409e-01 2.17358917e-02 1.06746387e+00
-1.95744494e-03 3.38341184e-02 -1.34068429e+00 1.06647062e+00
3.71874362e-01 -7.78868735e-01 -3.40714991e-01 -4.26096171e-02
5.50586730e-02 5.50571322e-01 9.93114561e-02 4.98663604e-01
1.12090237e-01 -1.28432274e+00 9.67212856e-01 2.48652384e-01
7.82120228e-01 -1.00183034e+00 9.96643066e-01 3.41518790e-01
-1.53244507e+00 -1.32505789e-01 -2.34300792e-01 -7.50180380e-03
8.10724497e-02 1.15227409e-01 -7.04704165e-01 1.87417254e-01
1.16553450e+00 4.39437926e-01 -7.87118971e-01 1.00320041e+00
-2.20430121e-01 5.03663659e-01 -7.93347895e-01 -2.52323151e-02
1.44355997e-01 6.42676950e-01 2.43224949e-01 1.51495087e+00
2.55443275e-01 -1.43687665e-01 -8.33145157e-02 1.47831216e-01
-1.56898975e-01 -4.18004543e-02 -4.94544148e-01 8.03597093e-01
4.34169382e-01 1.19798899e+00 -5.12874067e-01 9.42349881e-02
-4.17552203e-01 8.89963031e-01 5.10905325e-01 4.97395806e-02
-8.62949312e-01 -4.61594909e-01 6.77070141e-01 3.56846958e-01
4.36652094e-01 -1.76798627e-01 2.59852596e-02 -9.16707873e-01
-4.11701761e-02 -6.50024831e-01 3.40976357e-01 -6.16598606e-01
-8.92583251e-01 8.54154587e-01 1.96207896e-01 -9.56544459e-01
-3.74884069e-01 -8.30766737e-01 -4.36847240e-01 7.17983544e-01
-1.85256886e+00 -1.68213725e+00 -6.32261276e-01 9.98148620e-01
5.56887209e-01 -1.11194506e-01 1.00566494e+00 4.93942469e-01
-7.36849904e-01 1.36060381e+00 -3.32226038e-01 2.99252778e-01
6.48397505e-01 -1.17515445e+00 5.95852196e-01 6.36787236e-01
-2.77381897e-01 7.63574302e-01 7.17331827e-01 -2.94917971e-01
-1.45819306e+00 -1.02118301e+00 9.28793073e-01 -2.87478477e-01
1.18617035e-01 -6.68088138e-01 -5.96552491e-01 1.01782715e+00
2.34055579e-01 2.67173558e-01 5.87546706e-01 2.16604918e-01
-4.08920050e-01 -2.79457599e-01 -1.25141001e+00 4.71097797e-01
1.14242125e+00 -2.84881175e-01 -5.70679188e-01 3.42614651e-01
4.56927091e-01 -8.09393525e-01 -7.08144188e-01 4.99070495e-01
7.07122505e-01 -1.24657357e+00 1.03546178e+00 -1.05267271e-01
1.95690170e-01 2.39338875e-02 -2.04822466e-01 -1.10787904e+00
-7.87074268e-02 -6.48354769e-01 -4.34852362e-01 1.05156064e+00
2.67893136e-01 -6.61898017e-01 1.12385607e+00 7.30664492e-01
-2.60549456e-01 -9.29728389e-01 -1.23134184e+00 -5.12487292e-01
1.24858469e-01 -6.37501478e-01 8.45509291e-01 3.66088182e-01
-2.56835699e-01 1.79771855e-01 -4.29649323e-01 2.62810171e-01
8.39264572e-01 -4.89661574e-01 9.12304223e-01 -1.20295465e+00
6.42968789e-02 -2.71479040e-01 -7.07878888e-01 -1.09693670e+00
2.42472127e-01 -5.94022632e-01 1.77524865e-01 -1.42095673e+00
-2.17294060e-02 -2.26395324e-01 3.43770124e-02 6.98331475e-01
1.77917331e-01 4.60616231e-01 1.84063628e-01 -1.65225640e-01
-5.24110615e-01 5.63429177e-01 6.75585985e-01 1.45714998e-01
-2.27305293e-02 2.53414810e-01 -5.03909886e-01 1.11749291e+00
3.99456412e-01 -5.44394106e-02 -3.43781769e-01 -4.01025742e-01
9.39521343e-02 -7.46964589e-02 3.78595561e-01 -1.40201998e+00
7.09127426e-01 5.93879580e-01 6.69413507e-01 -9.59927976e-01
5.86426854e-01 -7.74625123e-01 -1.52889535e-01 3.17630112e-01
-9.75921899e-02 2.71178037e-01 2.86241710e-01 1.27805144e-01
-1.44072864e-02 5.41783124e-02 8.25539351e-01 2.94658672e-02
-8.33185494e-01 5.41384518e-01 -2.29443789e-01 2.67355237e-03
8.21058452e-01 -3.65468442e-01 -1.34657919e-01 -6.27677143e-01
-5.67411423e-01 2.42014155e-01 -7.18723983e-02 4.69391555e-01
6.76380575e-01 -1.20752096e+00 -6.27817631e-01 4.89645898e-01
1.37120858e-02 4.36930776e-01 9.73182023e-02 8.46356511e-01
-6.92025900e-01 6.87364876e-01 -1.52084425e-01 -5.36544740e-01
-1.43201208e+00 9.91114322e-03 4.57852930e-01 -5.23483977e-02
-9.32548106e-01 1.38309646e+00 -9.67808291e-02 -8.51431429e-01
9.86056030e-01 -3.71919662e-01 -4.54486251e-01 3.98652524e-01
7.89287090e-01 6.69664800e-01 6.10474825e-01 -1.03845155e+00
-5.84988773e-01 5.15865922e-01 -2.49019831e-01 5.40442280e-02
1.87253463e+00 -1.38821006e-01 1.12109542e-01 -3.84416543e-02
1.57505429e+00 -2.70489067e-01 -1.32072711e+00 -1.19301632e-01
-7.80723384e-03 7.42156059e-02 2.03621596e-01 -5.29355109e-01
-1.22877240e+00 9.17549849e-01 1.08299160e+00 -4.18710232e-01
1.16505563e+00 -1.10854983e-01 1.16128993e+00 5.93138814e-01
6.29754126e-01 -1.11753714e+00 -1.05841294e-01 5.99579453e-01
9.05449331e-01 -1.21648872e+00 1.46184405e-02 -1.05200864e-01
-2.57572889e-01 1.13486469e+00 7.31341600e-01 -1.27167791e-01
9.41828370e-01 7.80916989e-01 -2.49304958e-02 -2.88422585e-01
-3.20415318e-01 -3.66913885e-01 2.44382456e-01 7.04335570e-01
7.03825831e-01 -1.39327392e-01 4.54006754e-02 5.80810487e-01
-8.57324421e-01 2.34585688e-01 7.03565925e-02 1.03694487e+00
-3.65381390e-01 -7.73201048e-01 -3.24736595e-01 3.12488854e-01
-4.75040317e-01 -5.70441224e-02 1.46737294e-02 1.09082067e+00
1.95704445e-01 6.66039944e-01 -1.01317905e-01 -3.52444887e-01
6.59659386e-01 -4.16998416e-02 6.14919662e-01 -3.29143852e-01
-6.44880772e-01 -7.81664848e-02 -3.24690975e-02 -8.50417495e-01
-3.83495271e-01 -6.23116493e-01 -1.05759990e+00 -4.71584171e-01
-2.42620766e-01 -2.79446840e-01 6.39711380e-01 1.05458641e+00
1.54677704e-01 6.01119459e-01 3.01224589e-01 -1.62397528e+00
-4.57319111e-01 -1.23945153e+00 -6.59949481e-01 1.93682000e-01
5.99833488e-01 -8.87338519e-01 -2.73459315e-01 -3.17244530e-01] | [13.658293724060059, 0.2882228493690491] |
b177d1fc-f130-4877-b6d8-09fe06080221 | beit-bert-pre-training-of-image-transformers | 2106.08254 | null | https://arxiv.org/abs/2106.08254v2 | https://arxiv.org/pdf/2106.08254v2.pdf | BEiT: BERT Pre-Training of Image Transformers | We introduce a self-supervised vision representation model BEiT, which stands for Bidirectional Encoder representation from Image Transformers. Following BERT developed in the natural language processing area, we propose a masked image modeling task to pretrain vision Transformers. Specifically, each image has two views in our pre-training, i.e, image patches (such as 16x16 pixels), and visual tokens (i.e., discrete tokens). We first "tokenize" the original image into visual tokens. Then we randomly mask some image patches and fed them into the backbone Transformer. The pre-training objective is to recover the original visual tokens based on the corrupted image patches. After pre-training BEiT, we directly fine-tune the model parameters on downstream tasks by appending task layers upon the pretrained encoder. Experimental results on image classification and semantic segmentation show that our model achieves competitive results with previous pre-training methods. For example, base-size BEiT achieves 83.2% top-1 accuracy on ImageNet-1K, significantly outperforming from-scratch DeiT training (81.8%) with the same setup. Moreover, large-size BEiT obtains 86.3% only using ImageNet-1K, even outperforming ViT-L with supervised pre-training on ImageNet-22K (85.2%). The code and pretrained models are available at https://aka.ms/beit. | ['Furu Wei', 'Songhao Piao', 'Li Dong', 'Hangbo Bao'] | 2021-06-15 | beit-bert-pre-training-of-image-transformers-1 | https://openreview.net/forum?id=p-BhZSz59o4 | https://openreview.net/pdf?id=p-BhZSz59o4 | iclr-2022-4 | ['self-supervised-image-classification', 'document-image-classification', 'document-layout-analysis'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.84448850e-01 4.23196554e-01 -1.96380928e-01 -4.13997769e-01
-8.59788060e-01 -4.97652203e-01 2.12069571e-01 -4.95762467e-01
-6.09219074e-01 2.88046002e-01 -1.96052209e-01 -4.25853342e-01
5.91813266e-01 -7.38360941e-01 -1.41117430e+00 -5.06131232e-01
3.23809206e-01 2.91813254e-01 1.75527081e-01 1.94996700e-01
-4.70644683e-02 -1.79862872e-01 -1.30832183e+00 5.71785390e-01
8.22508693e-01 1.25636446e+00 5.51560044e-01 6.74744844e-01
-1.98947445e-01 1.09177721e+00 -5.77371299e-01 -4.96280670e-01
2.84129322e-01 -1.28436282e-01 -1.13782263e+00 4.49613363e-01
6.83435917e-01 -5.58446050e-01 -5.51407218e-01 1.24465060e+00
5.85948639e-02 -2.29119584e-01 4.13544059e-01 -1.08474219e+00
-1.19944155e+00 9.64681745e-01 -7.03795850e-01 -1.62876964e-01
-1.86154231e-01 5.09703934e-01 8.55320930e-01 -1.14449275e+00
4.60543007e-01 1.17777383e+00 4.53584939e-01 7.41711915e-01
-1.23823023e+00 -8.18465889e-01 3.87544692e-01 2.07532838e-01
-1.35866511e+00 -4.10363466e-01 4.86731470e-01 -3.58904868e-01
9.71936882e-01 -3.88490334e-02 5.89289129e-01 1.03832722e+00
-2.77100593e-01 1.22637582e+00 1.04036653e+00 -2.47151807e-01
-6.73883632e-02 -1.36633990e-02 1.87678993e-01 8.61493707e-01
-2.50479728e-02 -1.26146391e-01 -2.68184870e-01 4.31148916e-01
9.33395624e-01 7.51566216e-02 -1.57736927e-01 -1.13406330e-02
-1.30258417e+00 6.04713976e-01 9.18871462e-01 8.92517939e-02
-3.22993398e-01 6.26293480e-01 3.85996044e-01 1.58331662e-01
3.73537660e-01 4.90647145e-02 -6.58877730e-01 1.42143562e-01
-8.28154385e-01 -1.77283466e-01 2.50860035e-01 1.20192552e+00
1.20083475e+00 2.32477993e-01 -3.10032934e-01 1.02735865e+00
2.50339717e-01 7.15787113e-01 5.39084911e-01 -1.05997777e+00
5.88648915e-01 5.33683956e-01 -2.02339873e-01 -3.91135305e-01
1.40019193e-01 -3.17105561e-01 -1.00764287e+00 7.20252469e-02
2.92464852e-01 -4.46345508e-02 -1.80298185e+00 1.65038729e+00
3.52296121e-02 4.42987889e-01 3.11752021e-01 9.44509506e-01
9.54445422e-01 8.41647089e-01 1.51392177e-01 2.85296112e-01
1.58386934e+00 -1.48585594e+00 -3.28510225e-01 -6.32470548e-01
2.95400828e-01 -7.38906384e-01 1.27275872e+00 3.24602515e-01
-1.15790272e+00 -8.00598860e-01 -8.06020260e-01 -3.18338126e-01
-2.87620366e-01 5.80309749e-01 4.46494728e-01 2.37409920e-01
-1.22965240e+00 1.69433191e-01 -9.28472996e-01 -6.86331689e-02
8.95388126e-01 3.18958193e-01 -1.99908033e-01 -3.52593333e-01
-8.97930205e-01 5.12952566e-01 5.09736478e-01 1.93579748e-01
-1.57961059e+00 -8.66935432e-01 -1.08944547e+00 1.81778502e-02
4.56277013e-01 -6.20326221e-01 1.57605731e+00 -1.39983857e+00
-1.46869969e+00 1.16561186e+00 -2.26393208e-01 -7.31450915e-01
5.15608132e-01 -2.05312267e-01 3.89351845e-02 2.68149525e-01
4.22593504e-01 1.29861200e+00 1.17773461e+00 -1.45195973e+00
-5.82755566e-01 4.04672371e-03 1.49854258e-01 -1.35257039e-02
-2.23112360e-01 -2.27897570e-01 -1.18936515e+00 -5.96501410e-01
-6.80045858e-02 -8.92115831e-01 -3.30484688e-01 -7.54278377e-02
-8.88496816e-01 -1.39552951e-01 7.80010998e-01 -7.16479480e-01
4.85906661e-01 -2.32205057e+00 -1.19440459e-01 8.55706912e-03
4.26509380e-01 5.77298939e-01 -4.72330481e-01 -1.52013928e-01
-2.23803550e-01 2.00677127e-01 -4.87981915e-01 -7.52907217e-01
-3.82719263e-02 5.22641659e-01 -4.83271599e-01 1.83559954e-01
4.45572257e-01 1.34197259e+00 -7.68062294e-01 -5.40464342e-01
4.99338120e-01 4.27950919e-01 -5.81068397e-01 3.33809972e-01
-4.84748244e-01 2.52606153e-01 -3.22336525e-01 6.13704562e-01
8.17729294e-01 -5.52199781e-01 6.47676131e-03 -3.78399789e-01
-8.34332332e-02 2.35031411e-01 -5.96915126e-01 1.89624441e+00
-5.48137367e-01 6.51262164e-01 6.39774799e-02 -1.24432409e+00
5.97993314e-01 1.08824857e-01 2.74957418e-01 -8.95497441e-01
2.24175870e-01 -3.95846814e-02 -3.44949961e-01 -4.57446843e-01
4.47503239e-01 2.41585657e-01 -1.14442393e-01 1.13199964e-01
4.25042480e-01 7.81093165e-02 4.60645296e-02 3.25157255e-01
9.16956604e-01 1.05811819e-01 -2.71857381e-01 1.23000786e-01
2.99916148e-01 1.34270087e-01 6.12282991e-01 7.58117855e-01
-4.85942923e-02 9.06430840e-01 4.70476270e-01 -1.83181331e-01
-9.98033404e-01 -1.09768867e+00 2.74864584e-02 9.04383540e-01
3.18637043e-01 -4.58922714e-01 -1.02387154e+00 -9.10287023e-01
4.65188101e-02 4.55058664e-01 -8.65892529e-01 -1.53677166e-01
-4.67028797e-01 -4.16472822e-01 7.13561952e-01 7.12916911e-01
1.07889116e+00 -1.27014935e+00 -3.36133629e-01 -5.06616607e-02
-3.31361890e-01 -1.39083922e+00 -6.14339948e-01 3.03301781e-01
-5.78809977e-01 -1.06286657e+00 -7.99488366e-01 -1.21577156e+00
9.40669775e-01 5.37385285e-01 1.21661210e+00 1.76888153e-01
-3.15069526e-01 3.49461764e-01 -3.57500792e-01 -1.23068415e-01
-3.19310427e-01 1.25044256e-01 -5.16447246e-01 8.61540213e-02
7.86329731e-02 -2.94274837e-01 -8.44443381e-01 1.52803645e-01
-1.00506306e+00 4.68542010e-01 8.92485023e-01 9.55350280e-01
9.77008283e-01 -5.56448400e-02 2.09314287e-01 -9.30157363e-01
-1.00390010e-01 -2.80248672e-01 -5.42238474e-01 3.04679334e-01
-4.04290885e-01 -4.95499112e-02 6.55495107e-01 -4.57016051e-01
-9.17815566e-01 2.16169164e-01 -2.08272308e-01 -1.04836607e+00
-2.97591090e-01 3.64215195e-01 -3.24083120e-01 6.51418939e-02
1.62572548e-01 4.96792555e-01 -9.02793780e-02 -4.76096213e-01
7.00837314e-01 6.37639642e-01 9.93626058e-01 -5.86701393e-01
8.88655126e-01 4.53088492e-01 -8.58493090e-01 -5.98461986e-01
-1.05315375e+00 -2.73098975e-01 -3.60491157e-01 2.94411927e-02
1.26123977e+00 -1.24253404e+00 -7.02025473e-01 8.63990486e-01
-1.07887626e+00 -9.30509686e-01 -4.16790575e-01 1.74756616e-01
-4.67397720e-01 2.33444989e-01 -8.29781711e-01 -3.14864218e-01
-4.52363819e-01 -1.40065241e+00 1.18697393e+00 2.74387747e-01
3.87965560e-01 -7.34478056e-01 -4.59866285e-01 7.97303200e-01
2.96209246e-01 -1.69851318e-01 5.88709354e-01 -2.54870504e-01
-9.45232868e-01 2.48853266e-01 -7.12563753e-01 9.81564462e-01
7.90337995e-02 -1.07722342e-01 -1.28993666e+00 -2.64975041e-01
-3.87412846e-01 -6.91671789e-01 1.43938625e+00 4.39021945e-01
1.67468548e+00 -2.97590554e-01 -2.71075130e-01 9.63087380e-01
1.53291988e+00 -2.99538765e-02 7.11913884e-01 2.98189312e-01
1.19202411e+00 2.31878400e-01 3.88153225e-01 1.96680084e-01
7.38883376e-01 3.10403645e-01 7.46672750e-01 -5.60902357e-01
-4.99309778e-01 -5.54622591e-01 4.79560673e-01 7.04861581e-01
3.32005173e-01 -1.07321106e-01 -7.92484105e-01 7.26352215e-01
-1.92740846e+00 -5.64458132e-01 1.55674919e-01 1.80167627e+00
1.06861472e+00 2.18051404e-01 -1.14429355e-01 -2.07995713e-01
7.06658423e-01 1.68473810e-01 -9.77306068e-01 -4.60207723e-02
-7.93378502e-02 3.76624852e-01 1.04963088e+00 5.85317791e-01
-1.18361843e+00 1.59517920e+00 5.41186047e+00 8.96026075e-01
-1.24953365e+00 9.34923738e-02 9.20754254e-01 1.68680102e-01
-2.65334874e-01 5.55524267e-02 -6.41691148e-01 6.31561160e-01
6.01062655e-01 7.60417134e-02 4.81836468e-01 8.60755146e-01
4.27176319e-02 5.53476885e-02 -8.52602839e-01 9.77837801e-01
-3.20173125e-03 -1.30885983e+00 3.03025454e-01 -3.68341170e-02
7.32986093e-01 4.77546245e-01 3.44729930e-01 5.20089269e-01
6.94070101e-01 -1.17839038e+00 1.04966247e+00 1.56018957e-01
1.18102896e+00 -4.16007489e-01 4.98691767e-01 1.57675579e-01
-1.20586336e+00 9.47520733e-02 -5.73250413e-01 2.02314869e-01
-1.03021741e-01 6.49873972e-01 -7.48317063e-01 4.71119106e-01
1.11727166e+00 1.18217731e+00 -5.25014997e-01 8.12861741e-01
-5.73353291e-01 1.02122819e+00 -3.08946967e-01 5.75365424e-01
4.21371520e-01 -1.22025736e-01 -5.35564013e-02 1.23555386e+00
3.69870923e-02 -1.03898905e-01 3.51196349e-01 1.05505335e+00
-5.46679556e-01 -4.26088959e-01 -2.91061342e-01 1.29492283e-01
3.98375869e-01 1.26792705e+00 -6.98790550e-01 -8.41161370e-01
-5.39247990e-01 1.41986465e+00 4.14074928e-01 6.68773353e-01
-1.09120822e+00 -2.99399048e-01 8.42814922e-01 -2.76448816e-01
8.83578241e-01 -1.01069761e-02 -1.67997345e-01 -1.42503560e+00
5.12621291e-02 -8.73503506e-01 2.01913983e-01 -9.99061704e-01
-1.15095937e+00 7.41211891e-01 -1.99183971e-01 -9.90604103e-01
2.06248835e-02 -7.90667653e-01 -5.11630177e-01 7.17993319e-01
-1.80626321e+00 -1.58598554e+00 -4.61210102e-01 7.69291759e-01
7.12973118e-01 1.08562097e-01 5.61130345e-01 2.90840864e-01
-9.01040494e-01 8.68482709e-01 -5.86495362e-02 7.11570680e-01
5.90107799e-01 -1.44172406e+00 7.83264816e-01 9.55087245e-01
1.50625914e-01 3.27260882e-01 1.04783215e-01 -4.61547852e-01
-1.25943375e+00 -1.65366483e+00 4.76542383e-01 -2.77365625e-01
5.93271017e-01 -6.02079332e-01 -9.02759433e-01 1.04317176e+00
4.62202370e-01 3.42140526e-01 2.38036796e-01 -3.78475010e-01
-6.94526494e-01 -1.64868370e-01 -9.32014287e-01 4.47637171e-01
1.04184544e+00 -6.94134712e-01 -4.59913611e-01 2.61929631e-01
1.08850157e+00 -6.65287554e-01 -6.80104077e-01 3.20045590e-01
2.46929109e-01 -6.21828973e-01 1.23004866e+00 -2.78196692e-01
5.45675397e-01 -4.17033404e-01 -1.20031431e-01 -1.21382046e+00
-1.76916122e-01 -3.30054313e-01 2.24524960e-01 1.37670279e+00
4.56798345e-01 -5.05925715e-01 9.56494331e-01 3.73781562e-01
-2.98102528e-01 -6.73717141e-01 -6.28621638e-01 -5.67446828e-01
1.99155807e-01 -6.90710008e-01 4.76352185e-01 7.98834562e-01
-6.73432410e-01 3.20991933e-01 -3.84486318e-01 3.05333138e-01
7.98598349e-01 1.64469555e-01 7.06583440e-01 -5.82481265e-01
-3.48602206e-01 -2.39410400e-01 -7.71319866e-02 -1.57533598e+00
3.70798051e-01 -1.02422512e+00 2.76404798e-01 -1.59731650e+00
3.57481569e-01 -6.54892683e-01 -4.45668668e-01 1.19603109e+00
-3.78224254e-01 6.66285276e-01 3.35014999e-01 3.25473279e-01
-7.15683401e-01 6.28643811e-01 1.44658911e+00 -7.27487624e-01
1.19148329e-01 -2.80416876e-01 -8.41149509e-01 6.16783619e-01
9.04063106e-01 -3.44480366e-01 -5.06428242e-01 -1.06434596e+00
-2.25149155e-01 -2.32732430e-01 7.52199292e-01 -7.84589052e-01
8.22657719e-02 4.10737619e-02 4.44091886e-01 -5.82903624e-01
2.62871325e-01 -6.44816458e-01 -2.34396443e-01 3.17868590e-01
-4.16214049e-01 -1.82072520e-01 2.46341050e-01 4.40192282e-01
-3.62857044e-01 -5.11218235e-02 7.47735083e-01 -2.19368786e-01
-1.02657759e+00 5.42022526e-01 -3.29794407e-01 8.47591609e-02
8.17515969e-01 -1.84986055e-01 -5.00727534e-01 -1.10391364e-01
-6.46635592e-01 4.56206799e-01 5.82990825e-01 4.33905751e-01
8.78519654e-01 -1.02685082e+00 -5.84482253e-01 3.40438217e-01
1.29149780e-01 6.56394780e-01 3.09582651e-01 5.59343219e-01
-4.35292870e-01 1.65159658e-01 -2.16181148e-02 -7.57469773e-01
-1.07909667e+00 5.60205877e-01 4.41043288e-01 -4.45974693e-02
-8.22452068e-01 1.09387088e+00 7.51285791e-01 -3.43798697e-01
3.70662272e-01 -8.33479345e-01 1.13211937e-01 -3.11635375e-01
4.62586433e-01 -2.99353093e-01 -7.64688924e-02 -4.55261081e-01
-4.41556603e-01 5.99745095e-01 -4.70716357e-01 -8.00779015e-02
1.17287290e+00 -1.09343976e-01 -1.08342476e-01 1.72297016e-01
1.48211074e+00 -4.03817385e-01 -1.70601869e+00 -5.04664779e-01
-3.31695557e-01 -2.10244805e-01 5.88160492e-02 -7.85104036e-01
-1.72557890e+00 9.44365919e-01 4.12579834e-01 -1.25648394e-01
1.23198807e+00 2.97363222e-01 1.00991702e+00 3.64818364e-01
1.53892472e-01 -7.61037767e-01 2.72192895e-01 5.91183543e-01
5.77238202e-01 -1.29013729e+00 -5.15136719e-01 -4.58031803e-01
-7.28019655e-01 6.61953986e-01 8.24521720e-01 -3.84157300e-01
5.17929554e-01 3.03579569e-01 4.76591200e-01 -9.73187312e-02
-7.89451599e-01 -4.49746490e-01 4.22702022e-02 6.06953323e-01
1.20527260e-01 1.39433265e-01 4.67133492e-01 6.27045453e-01
-2.79102862e-01 7.17965588e-02 3.61299753e-01 6.51024282e-01
-2.00539872e-01 -8.92012239e-01 -2.00263247e-01 3.68752480e-01
-3.72643530e-01 -4.24455762e-01 -2.48620629e-01 5.66413343e-01
3.61696303e-01 8.71059418e-01 2.49158695e-01 -6.29006624e-01
2.62892604e-01 -2.38614306e-01 3.47917259e-01 -6.63453937e-01
-5.46598256e-01 5.48224039e-02 -1.72144324e-01 -6.70077145e-01
-3.54014277e-01 -2.76225746e-01 -1.35450399e+00 -1.86033458e-01
-1.19394749e-01 8.09369683e-02 5.53651273e-01 8.36010575e-01
3.60220879e-01 8.34400475e-01 5.84750295e-01 -8.91166329e-01
-2.24289075e-01 -8.93645704e-01 -2.14362308e-01 3.91885787e-01
4.94867355e-01 -3.19511533e-01 -1.63688272e-01 6.20771825e-01] | [9.606143951416016, 0.8276678323745728] |
1c2e5d71-f393-404d-8189-b5270ed0e05d | show-and-write-entity-aware-news-generation | 2112.05917 | null | https://arxiv.org/abs/2112.05917v2 | https://arxiv.org/pdf/2112.05917v2.pdf | Show and Write: Entity-aware Article Generation with Image Information | Many vision-language applications contain long articles of text paired with images (e.g., news or Wikipedia articles). Prior work learning to encode and/or generate these articles has primarily focused on understanding the article itself and some related metadata like the title or date it was written. However, the images and their captions or alt-text often contain crucial information such as named entities that are difficult to be correctly recognized and predicted by language models. To address this shortcoming, this paper introduces an ENtity-aware article Generation method with Image iNformation, ENGIN, to incorporate an article's image information into language models. ENGIN represents articles that can be conditioned on metadata used by prior work and information such as captions and named entities extracted from images. Our key contribution is a novel Entity-aware mechanism to help our model better recognize and predict the entity names in articles. We perform experiments on three public datasets, GoodNews, VisualNews, and WikiText. Quantitative results show that our approach improves generated article perplexity by 4-5 points over the base models. Qualitative results demonstrate the text generated by ENGIN is more consistent with embedded article images. We also perform article quality annotation experiments on the generated articles to validate that our model produces higher-quality articles. Finally, we investigate the effect ENGIN has on methods that automatically detect machine-generated articles. | ['Bryan A. Plummer', 'Yiwen Gu', 'Zhongping Zhang'] | 2021-12-11 | null | null | null | null | ['news-generation'] | ['natural-language-processing'] | [-3.96145172e-02 2.69289047e-01 -3.07072401e-01 -2.98933715e-01
-9.50222194e-01 -7.07635045e-01 1.18990290e+00 2.09746882e-01
-6.88249946e-01 8.53001118e-01 7.06637919e-01 -8.29686690e-03
5.37470102e-01 -7.16453910e-01 -1.28165925e+00 -4.37045470e-02
4.03280616e-01 5.41839242e-01 1.71189785e-01 6.59131408e-02
4.14735317e-01 -9.54876244e-02 -1.35451257e+00 5.17498374e-01
1.06007075e+00 6.10264063e-01 5.03209889e-01 7.14441597e-01
-6.11516476e-01 1.05144715e+00 -6.51299000e-01 -9.82317090e-01
1.08389765e-01 -3.19520682e-01 -6.13197863e-01 2.29992270e-01
8.55145037e-01 -3.99134159e-01 -4.25818980e-01 1.20357049e+00
8.21833387e-02 -3.65038604e-01 1.00947154e+00 -1.29429853e+00
-1.21005905e+00 1.15890074e+00 -6.69622123e-01 5.07221483e-02
3.20135921e-01 2.14928076e-01 1.28432405e+00 -1.23922217e+00
1.31539249e+00 1.19707632e+00 4.42967057e-01 2.69268751e-01
-1.00046694e+00 -4.00717407e-01 6.87201023e-02 2.38895267e-02
-1.30433917e+00 -4.75248158e-01 4.66367543e-01 -7.92070746e-01
6.34440482e-01 7.89036155e-02 4.67295676e-01 1.39723420e+00
1.44966841e-01 1.12192702e+00 8.49381030e-01 -5.80531299e-01
-1.61852241e-01 6.16253316e-01 2.50771254e-01 8.14316750e-01
6.95055246e-01 -4.09573525e-01 -5.12072980e-01 8.50492641e-02
5.54498434e-01 -3.30367535e-01 -3.24496210e-01 -1.98083907e-01
-1.62299645e+00 9.14714634e-01 2.81404346e-01 5.10519817e-02
-5.83728135e-01 1.27877831e-01 2.61050552e-01 -1.78347453e-01
5.34677207e-01 8.39553118e-01 -8.32244605e-02 -2.54223477e-02
-1.19442832e+00 3.04957867e-01 1.02948081e+00 1.33131266e+00
7.91894376e-01 -3.02788913e-01 -4.09934402e-01 8.94621313e-01
4.53128189e-01 9.62943733e-01 4.25769001e-01 -7.63117194e-01
7.34560370e-01 5.94844103e-01 3.80748659e-01 -1.19262326e+00
4.55389125e-03 -5.07853508e-01 -3.45892847e-01 -4.16073412e-01
2.93167263e-01 -2.00541466e-01 -1.12023854e+00 1.43435800e+00
-1.85794041e-01 -1.00766189e-01 2.69944221e-01 6.95314169e-01
1.33222687e+00 9.80390668e-01 1.35079607e-01 6.45651296e-02
1.69835365e+00 -1.22198665e+00 -7.72265911e-01 -5.23137748e-01
4.79527324e-01 -1.13967705e+00 9.37102616e-01 -2.50757664e-01
-8.11317861e-01 -3.72292340e-01 -7.93366492e-01 -8.61418992e-02
-4.56314296e-01 8.22202802e-01 2.04450950e-01 3.13326977e-02
-8.13158691e-01 -9.12665799e-02 -3.99679333e-01 -6.85637414e-01
4.73454118e-01 -3.16507399e-01 -4.23897654e-01 -6.44356608e-02
-1.15661824e+00 9.32076037e-01 5.33285499e-01 -4.52913731e-01
-5.46615422e-01 -6.00637734e-01 -9.80634332e-01 -1.91148117e-01
4.85614538e-01 -9.89597023e-01 1.41097856e+00 -1.10494339e+00
-7.89039791e-01 1.17232692e+00 -2.73031235e-01 -8.39514971e-01
4.57249045e-01 -3.17447990e-01 -3.66709858e-01 3.91182840e-01
6.31158471e-01 1.11848545e+00 8.50675702e-01 -1.53049374e+00
-7.52682805e-01 -8.42900127e-02 6.92443252e-02 3.26768428e-01
-3.42420340e-01 1.05806619e-01 -1.10585010e+00 -9.63136435e-01
-4.02235240e-01 -1.13419640e+00 2.00952992e-01 5.01185134e-02
-9.55045521e-01 -1.32839218e-01 4.88830119e-01 -9.83573854e-01
1.11794353e+00 -1.90765011e+00 -2.22454607e-01 4.36524339e-02
4.33726490e-01 -5.66256680e-02 -3.25947136e-01 5.16572893e-01
5.00696421e-01 4.03773725e-01 1.32014789e-02 -1.80399448e-01
1.69178098e-01 1.59372300e-01 -5.84943891e-01 5.79470545e-02
9.26289260e-02 1.13861573e+00 -7.44022846e-01 -8.68659258e-01
-2.43660852e-01 3.88177067e-01 -2.84440488e-01 3.01637333e-02
-5.13290346e-01 -1.18473202e-01 -3.91346306e-01 5.12082100e-01
4.17489648e-01 -6.75683737e-01 -2.34269738e-01 -6.39078021e-01
-9.85828936e-02 1.76468700e-01 -8.02241147e-01 1.07805622e+00
-3.58134568e-01 1.14811027e+00 -1.57545477e-01 -3.20540398e-01
5.75390875e-01 7.71011859e-02 1.77311748e-01 -6.82353556e-01
-1.73214171e-02 1.38984904e-01 -3.89035583e-01 -6.60546243e-01
1.05479550e+00 2.70184070e-01 -1.44555822e-01 4.65776861e-01
2.40103286e-02 -1.56229036e-02 7.79939532e-01 9.33393896e-01
8.65084231e-01 -5.54560870e-02 4.48589958e-02 2.20710039e-01
2.39190713e-01 5.70076585e-01 2.75523931e-01 1.23011577e+00
1.52851671e-01 6.77476108e-01 4.33827043e-01 -6.04984276e-02
-1.46908569e+00 -8.81139815e-01 -3.38967447e-03 7.46125996e-01
2.46552750e-01 -7.05379903e-01 -7.09450603e-01 -8.90591264e-01
1.12401821e-01 1.08287644e+00 -5.25605381e-01 2.10377276e-01
-3.70931000e-01 -6.30717158e-01 6.21021152e-01 4.02408630e-01
4.02320355e-01 -9.90676820e-01 -1.90782189e-01 1.58501521e-01
-6.91780269e-01 -1.69083118e+00 -7.98943043e-01 -4.39255714e-01
-3.10304493e-01 -1.02475202e+00 -1.04949069e+00 -8.67162943e-01
1.07383072e+00 9.81194824e-02 1.51861811e+00 -2.58466274e-01
-1.44923910e-01 7.10928440e-01 -5.55254459e-01 -6.59633517e-01
-6.53834403e-01 3.41924205e-02 -2.57971853e-01 -2.62902025e-03
4.90099698e-01 3.01243424e-01 -3.48731399e-01 -5.24943173e-02
-9.45851266e-01 4.99395043e-01 1.03898048e+00 8.05246651e-01
4.92636234e-01 -2.55008131e-01 2.39686877e-01 -1.15546536e+00
8.32643390e-01 -5.08679211e-01 -5.02576828e-01 5.01590312e-01
-6.71461225e-01 4.19435024e-01 2.52281904e-01 -3.68250340e-01
-1.09145737e+00 3.15117836e-02 1.07749522e-01 -1.76886335e-01
1.60646155e-01 9.87217665e-01 2.32999563e-01 3.56784940e-01
6.06299281e-01 4.33890015e-01 9.78874639e-02 -3.00768346e-01
5.37641823e-01 7.55725920e-01 6.42602086e-01 -3.89645904e-01
9.33112741e-01 2.47931466e-01 -6.26118243e-01 -9.33291137e-01
-1.00862098e+00 -4.88924325e-01 -2.68736899e-01 -2.95412719e-01
8.03075016e-01 -1.44208455e+00 5.01953736e-02 4.46292818e-01
-1.46314681e+00 1.62173077e-01 -2.26635709e-01 6.24559760e-01
-1.40438959e-01 3.38609278e-01 -7.64724433e-01 -4.23541725e-01
-3.83465439e-01 -1.05579829e+00 1.07549810e+00 2.37255931e-01
-2.61876225e-01 -8.48688900e-01 -3.77641618e-02 6.17063105e-01
2.02733427e-01 -8.51166174e-02 7.34723806e-01 -9.07670856e-01
-7.70897031e-01 -2.38908261e-01 -6.90294087e-01 2.65668809e-01
-2.29658544e-01 4.85794246e-01 -4.83288497e-01 1.17705971e-01
-6.37380898e-01 -1.47237718e-01 1.18767500e+00 3.05330127e-01
6.43298149e-01 -9.33578789e-01 -4.60904509e-01 1.14869967e-01
1.42543244e+00 -2.05741003e-01 6.33798122e-01 6.25133872e-01
9.32651579e-01 5.87316751e-01 5.74486613e-01 3.72404069e-01
8.77534270e-01 5.44691980e-01 2.77038127e-01 -4.41203564e-02
-3.41541082e-01 -5.67426324e-01 5.05141914e-01 8.79394293e-01
2.10122332e-01 -8.02307785e-01 -1.09900367e+00 8.50713491e-01
-1.69702327e+00 -1.24500215e+00 -3.06608230e-01 1.67491794e+00
1.09911406e+00 1.79728821e-01 -7.11293295e-02 -7.20061004e-01
9.74965751e-01 8.88459384e-02 -2.68178642e-01 1.40355617e-01
-5.78718722e-01 -3.96150559e-01 9.25590754e-01 3.20241153e-01
-1.19896305e+00 1.04974699e+00 5.81205082e+00 5.56489706e-01
-1.00239098e+00 -1.53391913e-01 4.63522971e-01 1.14739142e-01
-6.07444286e-01 2.35695504e-02 -1.17977273e+00 8.29886436e-01
7.07732081e-01 -4.90463197e-01 -6.92967102e-02 9.54094052e-01
2.22433642e-01 -1.27982929e-01 -9.22876894e-01 1.10047483e+00
6.44077539e-01 -1.83432102e+00 6.01864517e-01 1.57345489e-01
8.86277616e-01 3.10979784e-01 1.18437715e-01 3.84271771e-01
5.32151401e-01 -6.79582596e-01 1.11684597e+00 8.33110929e-01
5.68882465e-01 -3.51950109e-01 9.61009800e-01 1.53143570e-01
-6.65381968e-01 2.34667644e-01 -2.99529135e-01 4.20551926e-01
1.34513170e-01 9.19314027e-01 -1.31109548e+00 1.39821410e-01
5.01323044e-01 8.61113310e-01 -1.05387771e+00 1.19443619e+00
-4.92909789e-01 6.72048867e-01 -1.57107860e-01 -4.18607593e-01
3.91153097e-01 7.28218928e-02 7.31297314e-01 1.47353780e+00
5.28345406e-01 -2.89355010e-01 1.20497502e-01 9.55132782e-01
-8.87093246e-01 3.93388808e-01 -7.38772750e-01 -7.97206342e-01
6.71163499e-01 1.31203985e+00 -7.00608671e-01 -6.95477664e-01
-7.63365448e-01 1.01618683e+00 1.55903846e-01 3.96142840e-01
-8.88013184e-01 -4.93315548e-01 2.24428639e-01 4.37555552e-01
5.15011728e-01 -7.17591196e-02 -3.56715582e-02 -1.56205213e+00
2.46370256e-01 -8.66417944e-01 3.77926826e-02 -1.37124741e+00
-1.45465469e+00 6.56870663e-01 -2.26754591e-01 -1.09399569e+00
-2.93843776e-01 -5.26751935e-01 -2.71442980e-01 5.12658954e-01
-1.46686697e+00 -1.38960564e+00 -3.35877329e-01 1.61994129e-01
6.57987416e-01 -3.65418851e-01 3.23105901e-01 2.55037546e-01
-4.70417619e-01 4.65585977e-01 3.86368722e-01 7.95088947e-01
1.19927716e+00 -1.25330114e+00 2.95783043e-01 1.12393522e+00
5.93250692e-01 6.60163879e-01 9.03568327e-01 -1.07693124e+00
-1.22617519e+00 -1.37378085e+00 1.12327194e+00 -7.16373444e-01
8.72338772e-01 -1.05401516e-01 -6.05308890e-01 8.46925616e-01
6.60051286e-01 -2.47808129e-01 5.58151066e-01 -7.95018971e-02
-5.01577556e-01 1.70670331e-01 -8.23190868e-01 7.28667796e-01
6.37958944e-01 -5.38729370e-01 -8.54815960e-01 4.29563016e-01
6.87524796e-01 -3.98694605e-01 -5.08648396e-01 1.37592077e-01
5.31658471e-01 -3.25065374e-01 7.90839791e-01 -3.30673426e-01
9.61449385e-01 -3.19771141e-01 -3.54977115e-03 -1.26248860e+00
-1.49673566e-01 6.61908314e-02 -2.77560145e-01 1.52400887e+00
8.81929398e-01 -2.13325143e-01 5.99383295e-01 6.58211768e-01
1.31345941e-02 -3.71347517e-01 -3.31820935e-01 -5.94021082e-01
-2.59824395e-01 -2.15067223e-01 3.46708328e-01 9.45658326e-01
-3.54899615e-01 7.33426511e-01 -3.59122157e-01 2.86884978e-02
6.07726038e-01 1.27066121e-01 8.85321558e-01 -1.14977980e+00
1.14982881e-01 -3.46254349e-01 -3.43813151e-01 -1.06721830e+00
2.34937921e-01 -9.40249860e-01 2.55300790e-01 -2.15323448e+00
6.90328002e-01 -2.25301102e-01 2.05442920e-01 5.52957833e-01
-3.22121263e-01 4.19061780e-01 2.65090495e-01 6.25503361e-01
-1.03709078e+00 4.18754071e-01 1.15118086e+00 -5.29166341e-01
2.35554695e-01 -4.99281555e-01 -9.51635480e-01 7.79863000e-01
4.45815086e-01 -5.36784351e-01 -1.53870314e-01 -5.58923244e-01
5.05405784e-01 -2.63030469e-01 5.27979016e-01 -8.58814836e-01
4.17593986e-01 -4.62977812e-02 4.66974348e-01 -7.68371165e-01
2.10929923e-02 -5.20620346e-01 -8.30166489e-02 6.61552474e-02
-5.79064190e-01 1.04405396e-01 -1.75658882e-01 6.76395595e-01
-4.73171204e-01 -4.12069976e-01 4.49293613e-01 -4.15553689e-01
-9.83651638e-01 1.03674486e-01 -5.42204976e-01 4.65514094e-01
9.51945186e-01 -9.02739167e-02 -7.32787013e-01 -7.28809893e-01
-5.36119521e-01 3.36643577e-01 8.29663932e-01 6.21634007e-01
6.62223637e-01 -1.17490077e+00 -1.11764121e+00 -1.94108397e-01
6.05637193e-01 -4.78139222e-01 -1.38479844e-01 7.32044101e-01
-7.02787876e-01 6.02628112e-01 8.65670890e-02 -5.08802533e-01
-1.33027363e+00 3.19236904e-01 -1.65390730e-01 -2.78363049e-01
-4.35334533e-01 7.83983707e-01 1.55633390e-01 -2.68971443e-01
1.17666580e-01 -1.91696689e-01 -4.15289402e-01 4.12543893e-01
5.51248908e-01 -1.78504929e-01 -3.48668039e-01 -1.01004326e+00
-1.22370571e-01 2.36305296e-01 -5.23329854e-01 -2.03251034e-01
1.40301025e+00 -3.46820563e-01 -1.83274418e-01 2.48832762e-01
1.07001114e+00 5.34627855e-01 -9.51445639e-01 -4.26626265e-01
1.26957938e-01 -2.69300342e-01 6.54067025e-02 -9.79682326e-01
-9.20568585e-01 3.47512156e-01 1.18736900e-01 5.54676205e-02
5.06840467e-01 4.37855572e-01 9.59100246e-01 5.65663099e-01
1.30583644e-01 -1.31774008e+00 1.50653556e-01 7.15301335e-01
9.37107563e-01 -1.58025706e+00 1.41093001e-01 -3.09318751e-01
-1.03643787e+00 1.02575815e+00 7.15117753e-01 3.33373249e-01
3.59332800e-01 5.29482774e-02 7.69878328e-02 -3.22568119e-01
-6.39666736e-01 -3.07572782e-01 6.11752093e-01 2.80241936e-01
3.56308907e-01 -9.48521271e-02 -2.68597156e-01 5.85222483e-01
-2.07556561e-01 -1.59750193e-01 9.35922801e-01 7.87039638e-01
-4.65313971e-01 -8.72016191e-01 -3.46084327e-01 9.92844760e-01
-7.30784774e-01 -4.62786078e-01 -5.45579672e-01 6.88546181e-01
-6.47141263e-02 5.83643854e-01 2.38455296e-01 -6.29829541e-02
-8.50486383e-03 2.10160203e-02 1.92778379e-01 -8.25353920e-01
-1.90022156e-01 -9.30991620e-02 5.13305426e-01 -7.69780725e-02
-5.53815365e-01 -6.29732013e-01 -1.18796146e+00 -1.39109030e-01
-2.95031548e-01 3.62356305e-01 8.91526759e-01 7.64656901e-01
7.03440309e-01 3.48152310e-01 1.13431737e-01 -3.19481939e-01
-8.74083340e-02 -8.72461200e-01 -2.51945138e-01 6.53576374e-01
1.32875830e-01 -2.97833592e-01 -4.12747979e-01 7.30630815e-01] | [10.974937438964844, 1.0738558769226074] |
d4131f11-0ebe-4ede-b58a-938bc3d96cf9 | an-attribute-enhanced-domain-adaptive-model | null | null | https://aclanthology.org/C18-1160 | https://aclanthology.org/C18-1160.pdf | An Attribute Enhanced Domain Adaptive Model for Cold-Start Spam Review Detection | Spam detection has long been a research topic in both academic and industry due to its wide applications. Previous studies are mainly focused on extracting linguistic or behavior features to distinguish the spam and legitimate reviews. Such features are either ineffective or take long time to collect and thus are hard to be applied to cold-start spam review detection tasks. Recent advance leveraged the neural network to encode the textual and behavior features for the cold-start problem. However, the abundant attribute information are largely neglected by the existing framework. In this paper, we propose a novel deep learning architecture for incorporating entities and their inherent attributes from various domains into a unified framework. Specifically, our model not only encodes the entities of reviewer, item, and review, but also their attributes such as location, date, price ranges. Furthermore, we present a domain classifier to adapt the knowledge from one domain to the other. With the abundant attributes in existing entities and knowledge in other domains, we successfully solve the problem of data scarcity in the cold-start settings. Experimental results on two Yelp datasets prove that our proposed framework significantly outperforms the state-of-the-art methods. | ['Tieyun Qian', 'Zhenni You', 'Bing Liu'] | 2018-08-01 | an-attribute-enhanced-domain-adaptive-model-1 | https://aclanthology.org/C18-1160 | https://aclanthology.org/C18-1160.pdf | coling-2018-8 | ['spam-detection'] | ['natural-language-processing'] | [-3.98035258e-01 -5.96909702e-01 -3.26343834e-01 -7.83359587e-01
-4.58535224e-01 -4.43861574e-01 6.18300974e-01 1.33200198e-01
-4.10634816e-01 8.62391293e-01 6.85554296e-02 -1.45924345e-01
3.83616090e-02 -8.02926958e-01 -3.96032572e-01 -5.69589257e-01
4.68900114e-01 3.15587550e-01 3.46684247e-01 -2.45617285e-01
1.65294617e-01 -3.39318141e-02 -1.40776157e+00 7.02058896e-02
1.44375300e+00 1.12238240e+00 -5.08232079e-02 -1.43389374e-01
-5.15120924e-01 7.02624261e-01 -7.49399006e-01 -7.42602170e-01
3.43462884e-01 -2.60703832e-01 -4.87026989e-01 -1.57309294e-01
2.05259204e-01 -6.23584092e-01 -7.12194502e-01 1.02719021e+00
4.04318482e-01 1.90692678e-01 6.33883893e-01 -1.23105633e+00
-1.37913549e+00 5.59343219e-01 -6.61007106e-01 2.56542057e-01
4.46985550e-02 5.44787124e-02 1.08963573e+00 -8.11221182e-01
5.16906381e-01 1.04949284e+00 3.47776175e-01 3.96054655e-01
-5.22270560e-01 -9.27538395e-01 8.09555769e-01 4.47111636e-01
-1.05784452e+00 -1.21069714e-01 6.99008048e-01 -1.59400210e-01
6.13940954e-01 -1.54743195e-01 3.96379352e-01 1.28238392e+00
-1.47555366e-01 1.15292275e+00 9.21636224e-01 -3.50020118e-02
2.28108257e-01 4.95885551e-01 5.09035885e-01 2.25080594e-01
6.17403686e-01 -2.09459066e-01 -4.12217319e-01 -2.50333846e-01
9.13051143e-02 5.69704950e-01 -9.21217129e-02 -3.49835694e-01
-8.36107612e-01 9.81068611e-01 5.32045305e-01 1.52544856e-01
-2.62932599e-01 -1.61200225e-01 6.87392652e-01 4.92154092e-01
8.04481745e-01 3.89933020e-01 -7.57464945e-01 1.15775377e-01
-7.22948015e-01 3.58823478e-01 1.01717043e+00 1.18132377e+00
8.21509540e-01 -1.25261351e-01 -5.50821386e-02 8.41403425e-01
3.29832792e-01 6.22067213e-01 6.55552626e-01 -2.96106964e-01
5.47555447e-01 1.04564261e+00 4.57020044e-01 -1.17852998e+00
-3.06391031e-01 -6.91610396e-01 -8.11659992e-01 -6.28714740e-01
2.16462120e-01 -2.99077213e-01 -9.24193084e-01 1.59137309e+00
3.77663702e-01 2.76272565e-01 -2.15972047e-02 1.13504183e+00
1.16273177e+00 5.44189394e-01 1.68963447e-01 -7.09924623e-02
1.21193767e+00 -1.06840336e+00 -7.56739676e-01 -5.99545538e-01
7.59321332e-01 -6.21273756e-01 9.11424994e-01 1.60562754e-01
-5.54938078e-01 -3.27882051e-01 -9.85076010e-01 -1.96732804e-01
-8.67863834e-01 5.26575208e-01 9.29961085e-01 4.77810562e-01
-3.67627323e-01 4.05208588e-01 -2.72000223e-01 -3.09506118e-01
6.71856940e-01 3.65756601e-01 -1.62713304e-01 -2.25645423e-01
-1.95361030e+00 8.76105070e-01 7.79688805e-02 2.28229195e-01
-3.67931306e-01 -3.81991625e-01 -5.28363287e-01 3.18114728e-01
4.78171170e-01 -6.63849831e-01 1.26760757e+00 -1.21307409e+00
-1.10789251e+00 7.66964555e-01 -9.95923057e-02 -4.34245467e-01
2.34239385e-01 -4.35631841e-01 -8.18520725e-01 -7.90638328e-02
3.05553138e-01 -2.69241333e-02 1.05316734e+00 -1.03177726e+00
-7.80068040e-01 -4.25331980e-01 1.56969473e-01 1.44783258e-01
-7.57954240e-01 5.18833883e-02 -4.14316148e-01 -6.05686903e-01
-6.65228963e-02 -7.85545230e-01 -1.01966769e-01 -2.30166346e-01
-1.51397735e-01 -5.07869542e-01 8.67369652e-01 -5.48984647e-01
1.56115985e+00 -1.90797710e+00 -2.33908221e-01 2.17994805e-02
4.58812654e-01 7.43292570e-01 -1.17059462e-01 3.59607160e-01
1.55624822e-01 8.38043466e-02 3.24284993e-02 -3.26460861e-02
1.87732011e-01 4.90017682e-02 -5.80474436e-01 4.70000297e-01
2.00968489e-01 9.58427906e-01 -1.04606378e+00 -3.95877391e-01
-4.07925956e-02 1.75279379e-01 -4.75240760e-02 2.72117108e-01
-5.37100583e-02 -2.38368390e-04 -9.83039618e-01 9.36226547e-01
1.07338095e+00 -4.10135329e-01 -7.56276473e-02 -6.11977577e-02
2.86270082e-01 5.10837436e-01 -9.00805891e-01 1.30965495e+00
-4.58278447e-01 3.04878414e-01 1.90136388e-01 -1.10687983e+00
1.18915713e+00 4.85755168e-02 2.73863286e-01 -6.87017918e-01
3.56099248e-01 5.78166962e-01 1.38471827e-01 -5.57492137e-01
4.21562940e-01 -1.44658992e-02 -2.73108065e-01 6.01872623e-01
-7.72472695e-02 2.18909919e-01 4.20415998e-01 4.13804203e-01
1.17848051e+00 -1.27371743e-01 1.74406692e-01 -9.03059170e-02
8.76472056e-01 -2.13875756e-01 8.34989965e-01 8.74284685e-01
-5.60139120e-01 4.01045203e-01 4.92677927e-01 -5.26115477e-01
-8.53976965e-01 -9.61838007e-01 -9.97321010e-02 1.20586836e+00
5.71471512e-01 -3.43628466e-01 -3.84400547e-01 -1.30642605e+00
4.92790699e-01 3.39448154e-01 -5.69033444e-01 -5.34337282e-01
-5.90283155e-01 -1.00915563e+00 1.49766162e-01 5.79739213e-01
5.38554430e-01 -8.92745316e-01 4.91088219e-02 2.90352345e-01
-9.61357057e-02 -1.00900126e+00 -5.07746398e-01 1.51864380e-01
-6.71331048e-01 -1.04904199e+00 -7.30022669e-01 -8.77349317e-01
6.90817058e-01 5.11340857e-01 1.31222224e+00 2.49944001e-01
-2.14616694e-02 -5.98424189e-02 -7.77084947e-01 -5.51887274e-01
4.59933579e-02 6.12458467e-01 2.66615182e-01 6.68487996e-02
1.28440392e+00 -3.96473467e-01 -8.84528995e-01 6.29590750e-01
-8.60126019e-01 -5.87571621e-01 8.04998457e-01 1.01220739e+00
-5.22843376e-02 -4.30150628e-02 1.53883827e+00 -1.28220379e+00
1.00913954e+00 -9.83013928e-01 -3.49166393e-01 3.27716500e-01
-7.14289069e-01 -1.20977446e-01 1.03267300e+00 -4.07553762e-01
-1.19592917e+00 -1.44018635e-01 1.95436075e-01 -5.38526662e-02
-2.68457569e-02 5.54451704e-01 -4.89041656e-01 2.32713476e-01
4.43270773e-01 2.31505379e-01 -2.39599735e-01 -5.49891472e-01
3.37027967e-01 1.20692241e+00 2.14481726e-01 -5.63326836e-01
8.79551113e-01 3.63852471e-01 -5.53152740e-01 -2.96559155e-01
-1.24919319e+00 -8.89509499e-01 -7.10419297e-01 2.62375444e-01
1.05503969e-01 -1.05382884e+00 -5.47996044e-01 7.23536432e-01
-1.06296861e+00 5.96959829e-01 4.65972126e-02 2.71365672e-01
2.47657448e-02 6.99661493e-01 -4.52867031e-01 -6.67348087e-01
-7.05602407e-01 -9.67926204e-01 9.21435595e-01 5.88302374e-01
1.74432352e-01 -8.53577375e-01 -2.75080293e-01 3.12913865e-01
5.93972564e-01 -4.15521979e-01 8.76981497e-01 -1.22880125e+00
-6.72166467e-01 -6.95409417e-01 -5.96595287e-01 4.28882092e-01
2.59180516e-01 -2.71756679e-01 -7.34066725e-01 -1.93097502e-01
1.35611787e-01 -3.84136260e-01 1.28757679e+00 4.40947786e-02
1.01881719e+00 -2.52078742e-01 -5.19646645e-01 4.50043738e-01
1.07149053e+00 6.76709339e-02 1.65800110e-01 4.81597036e-01
6.20506942e-01 6.84474885e-01 9.68359649e-01 4.70070750e-01
4.94016469e-01 3.06719124e-01 3.13707858e-01 1.12649716e-01
3.80550444e-01 -1.23079106e-01 1.42493635e-01 8.28729868e-01
5.08510530e-01 -3.89313549e-01 -6.62347972e-01 7.65007675e-01
-2.17633080e+00 -8.07885289e-01 -1.17402934e-01 1.96540833e+00
9.16164577e-01 1.63393259e-01 -5.11659756e-02 -3.67918581e-01
8.90261889e-01 1.91366434e-01 -1.01270163e+00 -2.72290438e-01
-2.30156645e-01 -5.37594408e-02 4.67319995e-01 -1.04135796e-01
-1.60830462e+00 1.04973042e+00 5.26876831e+00 7.40840197e-01
-8.23499322e-01 -1.33885816e-01 5.44016123e-01 -2.17270195e-01
-2.73893714e-01 8.47546477e-03 -1.07108104e+00 1.04655421e+00
7.54466891e-01 -6.04462087e-01 1.64219528e-01 1.13632751e+00
5.15455864e-02 -1.40140355e-01 -9.65392947e-01 7.45255828e-01
1.58893049e-01 -9.15875137e-01 6.98338524e-02 -2.69417077e-01
9.14900243e-01 2.56226119e-02 2.08210811e-01 7.69368291e-01
3.05193424e-01 -8.17317903e-01 1.33725628e-01 4.90546256e-01
2.43589178e-01 -7.28962839e-01 1.25692320e+00 4.18097287e-01
-1.01266372e+00 -3.88639659e-01 -7.47983336e-01 -3.17196772e-02
1.27365711e-05 9.53269482e-01 -4.69812632e-01 5.71481586e-01
8.09177041e-01 1.23271525e+00 -8.12816560e-01 1.09195685e+00
-3.34762126e-01 6.24253094e-01 -2.26849779e-01 -4.13950235e-01
4.30879235e-01 -4.45996106e-01 1.88812360e-01 9.74199653e-01
1.80094570e-01 -2.19037503e-01 3.11955571e-01 7.18243241e-01
-5.37134051e-01 3.37919533e-01 -5.04500568e-01 -1.42192960e-01
6.66783810e-01 1.37271595e+00 -1.67011857e-01 -4.25519168e-01
-8.03170979e-01 7.20574498e-01 5.01052320e-01 4.22080725e-01
-7.13428080e-01 -8.22689414e-01 8.52359831e-01 2.62654200e-02
3.85981053e-01 1.22247025e-01 -3.71038646e-01 -1.52816641e+00
3.30694288e-01 -8.58962476e-01 3.67240041e-01 -4.22200918e-01
-2.17004895e+00 4.14811015e-01 -3.93827468e-01 -1.36167824e+00
-1.04461148e-01 -6.06919825e-01 -6.41296208e-01 1.21004713e+00
-1.81262708e+00 -1.38187981e+00 -3.18682700e-01 2.07709491e-01
4.49061006e-01 -5.21143198e-01 3.88022780e-01 4.86678630e-01
-7.68157065e-01 5.77685654e-01 7.28189409e-01 3.29546064e-01
1.30401158e+00 -1.27942514e+00 6.27601743e-01 7.15755105e-01
-2.95770079e-01 1.19342625e+00 4.34428692e-01 -7.46292055e-01
-1.21567571e+00 -1.03041065e+00 9.49672878e-01 -5.96941352e-01
9.00290787e-01 -4.39276367e-01 -1.21369958e+00 3.04445028e-01
7.83984065e-02 6.00326024e-02 7.45866418e-01 3.93200576e-01
-5.00512302e-01 -4.37106550e-01 -9.47463453e-01 3.05030435e-01
7.86572754e-01 -4.20212716e-01 -7.13435709e-01 4.66005862e-01
6.08127534e-01 -1.06763773e-01 -4.19717968e-01 2.69535810e-01
5.16126454e-01 -6.65674567e-01 8.14211130e-01 -9.06540453e-01
6.04704618e-01 -2.43837848e-01 3.31371695e-01 -1.44964838e+00
-3.27007920e-01 -1.77045658e-01 -4.17302459e-01 1.40477884e+00
3.22180241e-01 -7.12542474e-01 8.80100727e-01 7.41242886e-01
9.20128897e-02 -9.83674586e-01 -6.16226971e-01 -5.86760223e-01
2.91008502e-01 1.36591762e-01 8.42858851e-01 8.43009770e-01
-2.68064048e-02 6.40920699e-01 -6.31477118e-01 -1.84835330e-01
1.43906862e-01 7.09123015e-01 7.48734117e-01 -1.65939176e+00
2.43323416e-01 -2.69461125e-01 -1.41947433e-01 -1.34459484e+00
6.16117775e-01 -8.94851804e-01 -2.63615008e-02 -1.67580485e+00
2.57962525e-01 -5.27387261e-01 -6.48411393e-01 2.35573903e-01
-7.52469122e-01 -3.59110713e-01 -1.30462691e-01 3.12690794e-01
-9.36275840e-01 8.13842893e-01 1.10783458e+00 -3.53165537e-01
-4.89939153e-02 2.76279360e-01 -1.12480783e+00 7.44030118e-01
7.95486510e-01 -4.90205407e-01 -3.62227470e-01 -5.02616227e-01
4.07393813e-01 -6.15592718e-01 5.74027523e-02 -5.66693425e-01
3.94989550e-01 -1.12249397e-01 4.36648935e-01 -8.39892745e-01
1.00958683e-01 -1.03912783e+00 -8.19356024e-01 -4.90346625e-02
-3.65080088e-01 3.99969565e-03 -1.85424641e-01 1.19718719e+00
-3.27455461e-01 -4.80438590e-01 5.24041295e-01 -2.47237295e-01
-7.44586766e-01 4.73937094e-01 -1.00193791e-01 3.79774123e-01
9.01935101e-01 2.50874430e-01 -8.38654220e-01 -4.88482982e-01
-3.19481254e-01 8.52035642e-01 4.15658236e-01 8.70625675e-01
4.58205968e-01 -1.19438207e+00 -6.80543721e-01 1.17596723e-01
4.25241411e-01 -1.71341598e-01 4.10543233e-01 6.33799493e-01
-1.85019538e-01 5.09670198e-01 9.62389112e-02 -2.05604419e-01
-9.21385407e-01 8.60821009e-01 1.68836698e-01 -3.55779767e-01
-2.10329831e-01 6.94850326e-01 3.10346842e-01 -6.60631120e-01
2.61015862e-01 5.11739627e-02 -3.17168385e-01 3.13769281e-01
3.95611852e-01 2.62347430e-01 1.37987405e-01 -4.59984124e-01
-5.26680648e-01 1.80433899e-01 -7.85675168e-01 6.58725977e-01
1.32957804e+00 -3.49073112e-01 -1.71584949e-01 6.46779910e-02
1.21425700e+00 -1.15483336e-01 -6.86518073e-01 -7.27936864e-01
2.40429655e-01 -6.24024153e-01 -9.43518579e-02 -9.57422912e-01
-1.10908353e+00 1.03301072e+00 1.83349282e-01 2.65772104e-01
9.15706158e-01 -1.58598796e-01 1.24678802e+00 7.99522638e-01
6.83434606e-02 -1.71849954e+00 -1.52783245e-01 6.82363331e-01
3.31319869e-01 -1.66555941e+00 2.36840650e-01 -3.45224828e-01
-7.34872758e-01 8.36660266e-01 1.18811274e+00 -1.67857438e-01
7.49686658e-01 -1.24395907e-01 1.78709209e-01 -1.04653977e-01
-8.70047271e-01 -3.42724353e-01 2.03531533e-01 5.60815692e-01
5.53250372e-01 -1.29281133e-01 -7.77912378e-01 1.24091291e+00
2.65080065e-01 -3.01840017e-03 5.46283662e-01 8.11082244e-01
-5.77303648e-01 -1.19498801e+00 1.09744743e-01 1.21651602e+00
-4.83913749e-01 -2.20065027e-01 -6.87250257e-01 5.02183437e-01
1.45798326e-01 1.04921138e+00 -2.45076358e-01 -2.18218520e-01
3.61715555e-01 9.00915712e-02 -2.56372064e-01 -8.12622190e-01
-8.01639616e-01 -3.74691755e-01 -1.97728574e-02 -1.76382232e-02
-3.50169748e-01 -5.25248587e-01 -8.74385893e-01 -1.50909662e-01
-6.85753465e-01 3.82914513e-01 4.02993947e-01 8.83621097e-01
6.64643943e-01 2.24006534e-01 9.24578607e-01 -1.06622294e-01
-1.05427992e+00 -9.68883038e-01 -7.88728297e-01 6.71714187e-01
2.23506734e-01 -8.69556427e-01 -6.07625067e-01 -4.50081617e-01] | [7.904204368591309, 9.938251495361328] |
9b71fdf6-c1c9-4618-bffd-c15bd48bdb0a | motion-adjustable-neural-implicit-video | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Mai_Motion-Adjustable_Neural_Implicit_Video_Representation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Mai_Motion-Adjustable_Neural_Implicit_Video_Representation_CVPR_2022_paper.pdf | Motion-Adjustable Neural Implicit Video Representation | Implicit neural representation (INR) has been successful in representing static images. Contemporary image-based INR, with the use of Fourier-based positional encoding, can be viewed as a mapping from sinusoidal patterns with different frequencies to image content. Inspired by that view, we hypothesize that it is possible to generate temporally varying content with a single image-based INR model by displacing its input sinusoidal patterns over time. By exploiting the relation between the phase information in sinusoidal functions and their displacements, we incorporate into the conventional image-based INR model a phase-varying positional encoding module, and couple it with a phase-shift generation module that determines the phase-shift values at each frame. The model is trained end-to-end on a video to jointly determine the phase-shift values at each time with the mapping from the phase-shifted sinusoidal functions to the corresponding frame, enabling an implicit video representation. Experiments on a wide range of videos suggest that such a model is capable of learning to interpret phase-varying positional embeddings into the corresponding time-varying content. More importantly, we found that the learned phase-shift vectors tend to capture meaningful temporal and motion information from the video. In particular, manipulating the phase-shift vectors induces meaningful changes in the temporal dynamics of the resulting video, enabling non-trivial temporal and motion editing effects such as temporal interpolation, motion magnification, motion smoothing, and video loop detection. | ['Feng Liu', 'Long Mai'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['motion-magnification'] | ['computer-vision'] | [ 6.11470640e-01 1.02083497e-01 -5.77216893e-02 -2.21853834e-02
-2.77197987e-01 -6.99648261e-01 8.40754151e-01 -1.71777889e-01
-1.26549050e-01 3.63425881e-01 4.65601742e-01 -3.03298645e-02
-4.82134223e-02 -7.31048226e-01 -1.09120500e+00 -8.52724910e-01
-3.55080873e-01 -2.74283469e-01 1.96921110e-01 -2.42809966e-01
3.92953038e-01 3.94702554e-01 -1.45109546e+00 2.92863160e-01
2.64751941e-01 7.92971253e-01 1.79366261e-01 8.85306239e-01
2.87244380e-01 8.87704790e-01 -5.66026986e-01 1.10267259e-01
3.58332038e-01 -6.12569511e-01 -4.65395898e-01 2.35064283e-01
5.53278446e-01 -4.20948237e-01 -8.44220281e-01 6.82638943e-01
6.74712285e-02 3.07303905e-01 5.97185493e-01 -9.75293338e-01
-1.06289613e+00 3.22984636e-01 -4.82680023e-01 1.68615639e-01
5.67937434e-01 3.50122362e-01 8.28583121e-01 -7.84418046e-01
9.16526258e-01 1.11704755e+00 7.37074554e-01 3.56851339e-01
-1.53785706e+00 -8.80829990e-02 2.69789398e-02 3.20106864e-01
-1.29040992e+00 -2.97218144e-01 9.85475361e-01 -4.74668622e-01
7.91067004e-01 3.14373672e-01 9.62337315e-01 7.75405526e-01
3.69699746e-01 3.47980559e-01 5.58922946e-01 -5.79726934e-01
6.56213164e-02 -3.62762988e-01 -4.22826141e-01 5.11125803e-01
-7.51796439e-02 4.16631520e-01 -6.40474260e-01 2.26408422e-01
1.25164032e+00 -7.60567114e-02 -8.78750443e-01 -7.29898810e-01
-1.51958835e+00 6.33516133e-01 6.44941151e-01 4.76039886e-01
-3.73858929e-01 5.22792161e-01 2.53915936e-01 1.64040938e-01
1.22496769e-01 6.97734535e-01 -2.54492939e-01 -4.91030961e-02
-8.14882636e-01 3.01245362e-01 4.54643369e-01 7.46320724e-01
9.55516756e-01 3.60680699e-01 -3.04552406e-01 1.83411136e-01
1.93430349e-01 3.07147235e-01 7.23252118e-01 -1.29326606e+00
1.71037074e-02 1.81318000e-01 3.27778369e-01 -1.42722559e+00
-9.97439027e-02 -4.86544669e-02 -6.70134425e-01 8.93466175e-02
4.08744454e-01 8.35645795e-02 -8.90286922e-01 2.08571720e+00
8.42662081e-02 6.72865629e-01 -1.90384597e-01 1.03071523e+00
2.95457065e-01 1.00358605e+00 -8.65812227e-03 -4.93444055e-01
1.18135393e+00 -5.27483821e-01 -9.03426290e-01 9.47487634e-03
4.59807694e-01 -6.07322395e-01 1.02289510e+00 -1.23071820e-02
-1.20799851e+00 -7.13403583e-01 -1.23994005e+00 -1.41618967e-01
-1.82562560e-01 -2.14896083e-01 1.81374595e-01 1.36193691e-03
-1.25009537e+00 7.51245141e-01 -6.20411336e-01 -1.26603723e-01
-1.36722401e-01 4.88653556e-02 -4.15693969e-01 2.77671903e-01
-1.29769814e+00 6.02301419e-01 1.70459434e-01 1.18382685e-01
-4.58745271e-01 -1.01551926e+00 -1.15942848e+00 5.55595057e-03
7.65423775e-02 -8.22137356e-01 1.20810938e+00 -1.39404881e+00
-1.46925139e+00 5.86392105e-01 -3.36617023e-01 -5.35229564e-01
2.79018968e-01 6.13018358e-03 -2.27807283e-01 5.99202514e-01
3.17139998e-02 8.44447255e-01 1.48665559e+00 -1.09975600e+00
-2.71165073e-01 8.56580883e-02 1.36993274e-01 3.32620084e-01
-1.76627606e-01 -4.16460305e-01 -4.25036848e-01 -1.10692191e+00
1.87462866e-01 -9.29238617e-01 -1.40887666e-02 4.24625099e-01
4.77871485e-02 2.22960129e-01 1.11444330e+00 -6.29670501e-01
1.37344027e+00 -2.45533228e+00 5.13791859e-01 -1.89123482e-01
1.17091604e-01 7.06499219e-02 -3.53975505e-01 3.37700218e-01
-4.75710660e-01 -3.41045558e-02 -3.53401333e-01 8.54420289e-02
-2.40212053e-01 9.50641260e-02 -4.19802666e-01 5.72688997e-01
6.70615792e-01 9.57413554e-01 -1.21855068e+00 5.25629930e-02
3.62352103e-01 1.02532578e+00 -8.14348340e-01 1.69154570e-01
-3.36342603e-01 5.57189286e-01 1.54557091e-03 -1.03461752e-02
3.88547599e-01 -8.31536800e-02 3.48340780e-01 -7.51998365e-01
-2.81329781e-01 2.58840889e-01 -9.36739326e-01 1.60590029e+00
-2.99132079e-01 1.33082259e+00 -1.67015865e-01 -1.10266566e+00
5.45730650e-01 4.47855353e-01 5.96979678e-01 -9.28690851e-01
5.66526428e-02 -1.30453631e-01 -4.30107266e-02 -5.44523478e-01
6.36852741e-01 -5.32356501e-02 1.13251328e-01 2.79666841e-01
-1.06189631e-01 -1.95348382e-01 1.93362162e-01 1.07561454e-01
9.68081653e-01 4.34982210e-01 8.95292088e-02 -1.93241090e-02
4.29151148e-01 -2.70405501e-01 2.71867752e-01 3.09189171e-01
-2.43193340e-02 9.82690215e-01 4.08494830e-01 -6.00106120e-01
-1.26894319e+00 -1.16111267e+00 2.10107788e-02 7.62537658e-01
1.25384450e-01 -4.40877587e-01 -6.95396900e-01 -1.03805400e-01
-1.81183353e-01 5.20150185e-01 -8.27425241e-01 -3.76036465e-01
-1.02723181e+00 -1.81667671e-01 1.37968451e-01 3.88089269e-01
2.37782270e-01 -9.58354473e-01 -9.38641846e-01 4.01942521e-01
-3.33088070e-01 -1.04961491e+00 -9.15112793e-01 1.18310556e-01
-7.77771473e-01 -9.64406073e-01 -7.83370912e-01 -9.79550600e-01
7.31374800e-01 4.57916260e-01 8.87338698e-01 -1.51944280e-01
-3.45430821e-01 4.79829729e-01 -3.56130689e-01 3.71438265e-01
-3.46792400e-01 -6.20561182e-01 -1.03938408e-01 2.61031955e-01
-2.58501232e-01 -6.23358309e-01 -8.18621755e-01 2.57955164e-01
-1.44301236e+00 3.24863255e-01 2.17189416e-01 8.26190114e-01
5.55022418e-01 -1.30941927e-01 2.17277721e-01 -4.24045444e-01
4.25355405e-01 -2.73043394e-01 -2.06416398e-01 -1.18606456e-01
-8.87220502e-02 2.30657443e-01 5.90162277e-01 -1.09451032e+00
-7.18996167e-01 -1.17150195e-01 2.38799825e-01 -7.57821739e-01
1.62708506e-01 5.08690119e-01 5.34677915e-02 6.40873238e-02
7.58463800e-01 4.60586399e-01 7.57158026e-02 -5.04937721e-03
7.86707878e-01 1.51166216e-01 9.09693420e-01 -2.36002371e-01
9.50466156e-01 4.48390424e-01 -5.80128655e-02 -9.62710977e-01
-4.70449209e-01 -1.40984980e-02 -5.43163776e-01 -2.47637659e-01
9.56525803e-01 -8.62350166e-01 -3.46810341e-01 5.67314923e-01
-1.14654601e+00 -4.89822239e-01 -5.83686590e-01 3.97004157e-01
-8.36923957e-01 4.64378119e-01 -7.19590127e-01 -1.66461661e-01
6.98016286e-02 -1.04724050e+00 1.02126181e+00 1.72414586e-01
-6.42030537e-01 -1.05493021e+00 9.22153890e-02 -2.61780322e-01
3.62008333e-01 3.98209661e-01 1.20723569e+00 3.19466203e-01
-9.31541800e-01 -3.40149812e-02 -4.35542315e-03 2.45682165e-01
4.04564053e-01 2.26006851e-01 -6.04953945e-01 -1.97615579e-01
9.51538235e-02 1.93832785e-01 4.96404558e-01 6.87502146e-01
1.01047957e+00 -7.13426888e-01 -3.69529543e-03 8.49644661e-01
1.28326523e+00 4.03032929e-01 9.38843191e-01 3.17119539e-01
7.98259974e-01 4.61474657e-01 2.71542639e-01 2.58963853e-01
9.42444801e-02 9.28426206e-01 1.18545949e-01 1.03838094e-01
-4.47626889e-01 -4.27297682e-01 5.68618715e-01 7.45358169e-01
4.48391885e-02 -7.90307373e-02 -3.97137523e-01 5.23876965e-01
-1.71470153e+00 -1.28084779e+00 2.53349572e-01 2.15129852e+00
9.60936069e-01 -3.85718830e-02 -1.46272808e-01 2.18703106e-01
6.68439209e-01 4.98805046e-01 -5.22677898e-01 -3.87711495e-01
-8.41390528e-03 1.58148721e-01 1.70026451e-01 5.28939009e-01
-9.58196998e-01 6.03785932e-01 6.50837564e+00 4.71998334e-01
-1.74581468e+00 -3.85086834e-01 3.95088941e-01 -1.04078189e-01
-6.22461617e-01 4.47763689e-02 4.16503772e-02 5.87615907e-01
9.07366514e-01 -3.73153657e-01 6.64077580e-01 3.06154639e-01
5.79140961e-01 1.12989895e-01 -1.38475335e+00 9.07499671e-01
-3.09108067e-02 -1.81198645e+00 4.86257970e-01 -1.13466017e-01
6.40207827e-01 -6.01096034e-01 2.94077694e-01 -3.70342954e-04
-5.14825821e-01 -9.62226272e-01 1.19200599e+00 6.46992922e-01
1.07968664e+00 -4.07935649e-01 1.74970776e-02 -1.16137443e-02
-1.38600123e+00 -1.61725543e-02 5.93351871e-02 -1.51207194e-01
1.53320059e-01 2.01753899e-01 -6.32571101e-01 1.40172720e-01
4.28969562e-01 1.04206896e+00 -7.21771568e-02 6.38665140e-01
-1.47937015e-01 2.84785926e-01 -1.34534845e-02 3.08954030e-01
3.46475132e-02 -4.56317842e-01 5.78067005e-01 1.03356874e+00
4.50280309e-01 1.79651812e-01 -4.42777574e-01 7.97600269e-01
-8.73951763e-02 -4.65900570e-01 -7.31592536e-01 -2.24115714e-01
4.73732144e-01 8.16119671e-01 -3.97449791e-01 -1.92769527e-01
-4.33117360e-01 9.87817049e-01 1.13229994e-02 8.06992948e-01
-9.47336197e-01 -2.42076129e-01 7.98113108e-01 5.93534470e-01
6.17976487e-01 -4.62396413e-01 1.71002477e-01 -9.39034581e-01
-2.70334594e-02 -7.54827559e-01 -2.01928809e-01 -1.33685875e+00
-7.80522108e-01 4.70068008e-01 5.48188090e-02 -1.52740920e+00
-6.25694215e-01 -4.47282195e-01 -5.64394832e-01 6.33165658e-01
-1.33877575e+00 -7.61333764e-01 -9.16913375e-02 4.89555061e-01
6.53082311e-01 3.29471767e-01 6.74723864e-01 1.50161415e-01
-6.38855249e-02 3.22100669e-01 -1.22784786e-01 2.42593080e-01
6.31431520e-01 -8.90552342e-01 2.88801134e-01 8.67333233e-01
1.82253957e-01 7.70731866e-01 1.01586330e+00 -2.08418787e-01
-1.70173645e+00 -1.08741844e+00 4.40891176e-01 -3.23824137e-01
6.97572947e-01 -1.20076410e-01 -1.05314624e+00 7.83242106e-01
3.88454705e-01 1.42992094e-01 3.63002628e-01 -7.32327044e-01
-5.29325783e-01 -9.01602656e-02 -8.54156077e-01 9.25216615e-01
9.25971389e-01 -9.38386083e-01 -6.37185335e-01 -5.29432930e-02
1.01213324e+00 -5.33875585e-01 -5.77660799e-01 3.99911791e-01
6.45039439e-01 -9.11480367e-01 1.29216337e+00 -2.44042262e-01
7.68501818e-01 -6.75669611e-01 -2.87230387e-02 -1.41605902e+00
-6.55688703e-01 -7.10394025e-01 -4.19310898e-01 7.36662030e-01
3.64649780e-02 -4.73634809e-01 3.93418878e-01 4.45917249e-01
-2.85702152e-03 -6.78926647e-01 -8.23618352e-01 -3.49649787e-01
-1.19830802e-01 -2.14068025e-01 2.89173156e-01 9.32515442e-01
9.95291513e-04 1.07333675e-01 -2.45107278e-01 1.92359522e-01
3.31909895e-01 7.50376359e-02 4.86675650e-01 -6.47507548e-01
-4.26647186e-01 -3.08980256e-01 -5.83132744e-01 -1.43288755e+00
-1.19773500e-01 -4.54244822e-01 1.64524242e-01 -1.09814239e+00
-1.68403879e-01 -8.10623989e-02 -6.39881939e-02 1.61102340e-01
-6.97377697e-02 4.19057876e-01 4.36682284e-01 2.24985987e-01
1.83650386e-02 6.17572784e-01 1.56809866e+00 -2.52136350e-01
-1.56970143e-01 -5.81265986e-01 -4.51726735e-01 6.13189101e-01
5.16654015e-01 -1.86865792e-01 -6.86093330e-01 -5.73823810e-01
2.82821685e-01 3.66943151e-01 4.40613866e-01 -9.72532034e-01
-7.29550002e-03 -1.78333238e-01 5.47095597e-01 -1.73848078e-01
4.43867862e-01 -7.28234351e-01 4.98187780e-01 4.49633986e-01
-4.90085363e-01 1.87770709e-01 1.94970205e-01 6.04990780e-01
-2.51810282e-01 -3.09046134e-02 6.18421555e-01 4.68761995e-02
-8.91942680e-01 2.10166514e-01 -7.03269958e-01 -1.02477901e-01
9.80813682e-01 -6.34328187e-01 -1.03057161e-01 -5.80700994e-01
-6.94268942e-01 -3.78609598e-01 6.24681234e-01 5.89083791e-01
6.20091558e-01 -1.41665530e+00 -3.66270155e-01 6.75834239e-01
-1.42427936e-01 1.10671381e-02 3.25670451e-01 5.35062432e-01
-6.28125191e-01 2.57045716e-01 -2.12641716e-01 -8.01446319e-01
-9.09245014e-01 4.53962654e-01 5.44918299e-01 1.08481340e-01
-6.36607945e-01 5.00098050e-01 4.93568987e-01 1.30503565e-01
-2.98695583e-02 -7.30691850e-01 -9.67719704e-02 5.25077283e-02
6.02347434e-01 -9.78863388e-02 -3.53723764e-01 -7.36027300e-01
6.57123607e-03 9.26596522e-01 4.36334044e-01 -3.40151966e-01
1.09599984e+00 -2.12847561e-01 4.00490202e-02 5.55715144e-01
1.54975080e+00 -2.69697290e-02 -1.86368895e+00 -3.52023505e-02
-4.00232494e-01 -5.33477008e-01 -5.80336004e-02 -3.82917732e-01
-9.62860405e-01 6.15638196e-01 4.26077157e-01 3.80821824e-01
1.19608438e+00 -3.69674027e-01 6.44012153e-01 -7.60828494e-04
2.80532569e-01 -7.17419147e-01 4.97407317e-01 6.04071259e-01
9.77439582e-01 -6.62685990e-01 -1.22988470e-01 -3.35611314e-01
-2.07183316e-01 1.25878358e+00 2.59848416e-01 -2.78508157e-01
5.44833601e-01 3.05900246e-01 5.24028689e-02 1.36089966e-01
-6.80137455e-01 2.26773784e-01 2.89361238e-01 7.61195362e-01
4.87970442e-01 -4.76145837e-03 -1.36700556e-01 -1.93268135e-01
-1.78755730e-01 2.16614138e-02 7.24467218e-01 8.74049962e-01
-2.26005554e-01 -9.33306515e-01 -4.05977339e-01 -7.68327713e-02
2.96282172e-02 1.78929325e-02 1.57423049e-01 7.53167450e-01
1.28867283e-01 5.69394350e-01 6.61005855e-01 -2.12912321e-01
2.74166107e-01 -7.10403994e-02 8.84333551e-01 -4.73188937e-01
-1.62227452e-01 1.38163611e-01 -2.56515026e-01 -7.22959816e-01
-6.05222702e-01 -5.40472746e-01 -1.51682007e+00 -2.75157481e-01
2.72209257e-01 -9.61186066e-02 4.04600203e-01 6.76266909e-01
5.38035095e-01 7.07375169e-01 6.42844439e-01 -1.33408594e+00
-1.78716779e-01 -6.10717118e-01 -2.94094950e-01 8.48450541e-01
8.64645958e-01 -4.37007308e-01 -4.99481976e-01 7.22256005e-01] | [10.814191818237305, -0.7737213969230652] |
9d4f67fa-f375-4298-9943-db2b31f815a8 | the-art-of-embedding-fusion-optimizing-hate | 2306.14939 | null | https://arxiv.org/abs/2306.14939v1 | https://arxiv.org/pdf/2306.14939v1.pdf | The Art of Embedding Fusion: Optimizing Hate Speech Detection | Hate speech detection is a challenging natural language processing task that requires capturing linguistic and contextual nuances. Pre-trained language models (PLMs) offer rich semantic representations of text that can improve this task. However there is still limited knowledge about ways to effectively combine representations across PLMs and leverage their complementary strengths. In this work, we shed light on various combination techniques for several PLMs and comprehensively analyze their effectiveness. Our findings show that combining embeddings leads to slight improvements but at a high computational cost and the choice of combination has marginal effect on the final outcome. We also make our codebase public at https://github.com/aflah02/The-Art-of-Embedding-Fusion-Optimizing-Hate-Speech-Detection . | ['Sanyam Goyal', 'Mohit Jain', 'Neemesh Yadav', 'Mohammad Aflah Khan'] | 2023-06-26 | null | null | null | null | ['hate-speech-detection'] | ['natural-language-processing'] | [-1.92226142e-01 -1.84215769e-01 -2.04721585e-01 -9.36813578e-02
-9.51074958e-01 -5.29391706e-01 8.06997895e-01 4.19095427e-01
-5.30293584e-01 2.40139246e-01 7.34487712e-01 -3.21322143e-01
2.89360493e-01 -4.71685410e-01 -2.64094442e-01 -4.25399095e-01
2.45287046e-01 -2.52026528e-01 9.55836996e-02 -2.03637332e-01
3.02104920e-01 3.77770811e-01 -1.23229015e+00 3.61090988e-01
8.00970018e-01 1.89496294e-01 9.74539369e-02 8.68624091e-01
-1.90285951e-01 8.72727334e-01 -5.85786283e-01 -7.43707836e-01
-1.76724121e-02 -1.67943999e-01 -7.05499470e-01 -1.59479827e-01
4.45462644e-01 -4.87174362e-01 -5.32704055e-01 1.10084736e+00
7.09804654e-01 1.23328879e-01 7.00457811e-01 -1.17151833e+00
-1.25205076e+00 4.91671801e-01 -5.37518084e-01 6.55554593e-01
2.81376451e-01 2.69691020e-01 1.30555749e+00 -1.14328039e+00
5.84906995e-01 1.44456255e+00 6.53000295e-01 6.80674851e-01
-1.23025179e+00 -7.24731863e-01 9.77433324e-02 2.30775282e-01
-1.24529111e+00 -6.41544998e-01 8.60431135e-01 -4.44774181e-01
1.30980086e+00 4.01866660e-02 7.38440380e-02 1.45608389e+00
-1.72075443e-02 9.83967960e-01 1.14750600e+00 -6.09780371e-01
-2.01251462e-01 4.63863403e-01 5.60689867e-01 7.94910669e-01
3.41762990e-01 -2.62031883e-01 -6.47600532e-01 -3.25182527e-01
1.52933747e-01 1.57137811e-01 -1.05866693e-01 1.36754692e-01
-7.60008097e-01 1.30231690e+00 3.90202790e-01 6.87075913e-01
-1.78131297e-01 1.39065444e-01 5.83226860e-01 1.09875038e-01
8.87404501e-01 7.70070374e-01 -1.39823884e-01 -2.82352269e-01
-8.75135303e-01 2.59363979e-01 5.16558826e-01 4.01547462e-01
6.86728299e-01 5.51923364e-03 -1.85191885e-01 1.20531082e+00
1.27563551e-01 5.12295365e-01 4.32349831e-01 -6.28164351e-01
4.47542131e-01 2.85502255e-01 3.19683179e-02 -1.14712954e+00
-3.67180496e-01 -8.50567669e-02 -2.30103076e-01 1.49370208e-02
3.51025879e-01 -5.07113934e-01 -9.08729792e-01 1.66490412e+00
2.84521449e-02 2.30972636e-02 -2.24525452e-01 8.25538576e-01
8.01097691e-01 8.22987258e-01 6.20752037e-01 2.95051724e-01
1.48522818e+00 -1.18730664e+00 -1.03913343e+00 -6.39271915e-01
1.00288105e+00 -1.12007499e+00 1.22311521e+00 1.26661986e-01
-8.47205281e-01 -1.62671998e-01 -1.04936826e+00 -6.25518918e-01
-8.11533630e-01 1.46182641e-01 5.00041962e-01 8.26832592e-01
-9.57896233e-01 4.05259550e-01 -7.18780518e-01 -5.60077429e-01
5.65758646e-01 -9.67889726e-02 -4.12605733e-01 -2.90931225e-01
-1.28321195e+00 1.34996152e+00 -1.58131626e-02 -3.65884244e-01
-6.78360105e-01 -7.61511981e-01 -9.43006754e-01 -1.21736839e-01
4.23201054e-01 -3.10906678e-01 1.12790692e+00 -7.17559159e-01
-1.02520168e+00 8.29050362e-01 -5.63052595e-01 -4.38286692e-01
-8.15018713e-02 -6.51636720e-01 -3.84736747e-01 2.87886500e-01
1.07180104e-01 6.70495629e-01 7.82787085e-01 -1.02458572e+00
-1.45062879e-01 -2.81256914e-01 1.27698481e-01 1.15240119e-01
-8.93464088e-01 7.24764287e-01 -1.14797212e-01 -9.34463382e-01
-6.71917439e-01 -9.00525928e-01 6.74315766e-02 -3.03167962e-02
-2.60094523e-01 -3.71489316e-01 9.29292440e-01 -1.12013805e+00
1.54396009e+00 -2.30808067e+00 -7.51607344e-02 -3.87533516e-01
2.07313895e-01 7.77549982e-01 -3.63159388e-01 9.42563295e-01
1.12033747e-01 6.26889825e-01 -2.59691715e-01 -8.25759113e-01
1.90826342e-01 -5.47652021e-02 -3.94707501e-01 4.67130631e-01
6.73410296e-01 8.88033628e-01 -8.62917840e-01 -2.67658263e-01
4.33051974e-01 1.08989429e+00 -6.47404909e-01 2.72629291e-01
-8.36842693e-03 -1.77394003e-01 -2.24993616e-01 5.26024282e-01
5.98337770e-01 -5.21489456e-02 -4.02244218e-02 9.31723788e-02
-2.64859721e-02 7.82503486e-01 -6.56579256e-01 1.36950541e+00
-5.88939905e-01 9.90613222e-01 1.31508887e-01 -6.25258446e-01
7.75091350e-01 2.74087697e-01 -7.98228532e-02 -3.92669022e-01
1.54175624e-01 2.69693751e-02 5.25242612e-02 -5.81008255e-01
5.43224692e-01 -3.74900103e-01 4.27312851e-02 4.84391034e-01
6.74616098e-02 1.82130679e-01 6.50877059e-02 2.96752244e-01
1.25696898e+00 -1.44017875e-01 3.27086657e-01 -2.08286285e-01
1.96152836e-01 7.28278756e-02 2.46089622e-01 5.07315099e-01
-6.24265611e-01 6.07287526e-01 5.89684844e-01 2.23006140e-02
-1.01503491e+00 -9.29622412e-01 -4.31455337e-02 1.48409617e+00
-3.16337854e-01 -7.23738253e-01 -7.18349755e-01 -7.87961662e-01
1.42553404e-01 1.22879219e+00 -6.28422022e-01 -2.40151539e-01
-4.04752642e-01 -7.77656615e-01 7.11882949e-01 5.17192125e-01
-2.67248396e-02 -1.07720459e+00 -4.59498107e-01 -1.66105241e-01
-1.26926266e-02 -1.17259693e+00 -5.14168382e-01 1.81005999e-01
-4.81794119e-01 -7.81586945e-01 -6.61808491e-01 -6.11807466e-01
5.48659861e-01 6.51417315e-01 7.68374681e-01 2.77398944e-01
-3.62339616e-01 2.30628744e-01 -6.63710356e-01 -5.67556620e-01
-4.79335189e-01 7.89427012e-02 -1.19264677e-01 -2.37285882e-01
9.71147180e-01 -2.47412473e-01 -3.62536579e-01 -4.19999242e-01
-9.59832847e-01 -2.16133967e-01 2.78827846e-01 6.13418758e-01
-2.37138227e-01 -1.16130777e-01 3.71641934e-01 -1.01745701e+00
1.05439544e+00 -7.27997780e-01 2.78832838e-02 -6.98210374e-02
-3.09568465e-01 -5.62077165e-02 6.29017770e-01 -4.08435404e-01
-1.01729727e+00 -3.73233438e-01 -3.10928285e-01 -4.90761638e-01
-4.42820072e-01 4.62555379e-01 2.51172602e-01 3.88943821e-01
6.39693320e-01 -1.47788852e-01 6.27327710e-02 -7.50644863e-01
6.07911110e-01 7.73739755e-01 5.88510856e-02 -4.36325550e-01
8.14884663e-01 3.18470776e-01 -6.54523790e-01 -1.24732518e+00
-8.62611234e-01 -7.70739317e-01 -4.80008543e-01 1.99175060e-01
1.04355657e+00 -1.07527328e+00 2.84204483e-02 3.95059347e-01
-1.31563902e+00 -3.12046200e-01 3.25779647e-01 2.83030331e-01
5.95681518e-02 4.76306856e-01 -9.22290802e-01 -9.75677669e-01
-3.94283891e-01 -1.13808942e+00 1.01625979e+00 -4.73805889e-02
-6.09101355e-01 -1.19899321e+00 2.07294121e-01 7.02946603e-01
5.68456292e-01 -7.29516074e-02 9.18975234e-01 -9.89904344e-01
-4.15863134e-02 -1.73121586e-01 -3.33062172e-01 6.16418660e-01
2.72902071e-01 1.74846485e-01 -1.42135549e+00 -2.20875546e-01
-2.61333704e-01 -4.45829839e-01 1.15031445e+00 2.34553486e-01
7.87925363e-01 -4.68789160e-01 -8.56332108e-02 1.12289973e-01
1.44600570e+00 -2.20577672e-01 5.93836904e-01 3.15027237e-01
7.36640811e-01 7.86355913e-01 2.76321083e-01 3.71347219e-01
2.64230639e-01 5.16505837e-01 -1.29866146e-03 6.61556721e-02
-5.38046896e-01 -3.31659853e-01 8.95102024e-01 7.63037741e-01
3.58066350e-01 -3.94246191e-01 -9.99981344e-01 8.83755803e-01
-1.56067657e+00 -1.12127697e+00 8.02530572e-02 1.71470654e+00
8.34348321e-01 1.13359131e-02 3.57921660e-01 1.14724994e-01
8.27211797e-01 7.40670979e-01 -2.81920228e-02 -9.29296434e-01
3.72237265e-02 1.63010195e-01 3.46333474e-01 9.22683060e-01
-1.23838830e+00 1.22613025e+00 5.99023390e+00 9.20602858e-01
-1.05505157e+00 5.55011213e-01 3.29595327e-01 -1.85471863e-01
-4.29783374e-01 -1.74051046e-01 -8.45900834e-01 4.42936391e-01
1.27734041e+00 -5.92625663e-02 5.86338282e-01 7.05485523e-01
2.53418446e-01 1.11856282e-01 -5.50677180e-01 7.24506855e-01
3.86496723e-01 -1.16899621e+00 -3.36614391e-03 1.90172493e-02
5.16640782e-01 3.59247178e-01 3.69893909e-01 3.33128959e-01
3.07911247e-01 -9.51989651e-01 4.87026334e-01 8.63154009e-02
3.76867324e-01 -7.56456375e-01 7.73970842e-01 1.64372101e-01
-7.03192770e-01 -1.80400267e-01 -3.05726945e-01 -1.38591707e-01
1.33524805e-01 6.02920949e-01 -9.29266155e-01 -9.76229534e-02
5.10435164e-01 6.06199384e-01 -8.40565801e-01 5.52387416e-01
-5.67649782e-01 9.19211209e-01 -4.83691543e-02 -2.88682371e-01
3.97713929e-01 1.27056494e-01 6.06213987e-01 1.81399405e+00
2.18255311e-01 -1.41086027e-01 -1.00693226e-01 9.46138799e-01
-7.31888339e-02 2.77552307e-01 -1.09961367e+00 -7.71166980e-01
7.08723962e-01 1.21402895e+00 -4.50308532e-01 -2.62520760e-01
-8.61507416e-01 8.54454279e-01 8.04557681e-01 2.77280509e-01
-7.46926367e-01 -3.73794347e-01 1.19292855e+00 2.08956242e-01
3.37216914e-01 -5.23770988e-01 -3.37902397e-01 -1.19484866e+00
-1.83501124e-01 -8.60844731e-01 3.13249737e-01 -7.91856050e-01
-1.45762539e+00 3.28928798e-01 -1.06381834e-01 -5.68038523e-01
-1.29451513e-01 -8.71422827e-01 -7.29067981e-01 7.79757977e-01
-1.54163098e+00 -1.19860196e+00 3.07631344e-01 2.49770314e-01
8.33402038e-01 1.56318285e-02 8.68274689e-01 2.74378747e-01
-8.93383026e-01 6.92001402e-01 -1.88214034e-02 5.11979759e-01
9.63040113e-01 -1.14335978e+00 3.98120016e-01 1.10178602e+00
1.95104435e-01 1.04266977e+00 8.27579737e-01 -6.38869286e-01
-1.20458663e+00 -9.85449851e-01 1.38609600e+00 -8.98376882e-01
1.07909131e+00 -5.81946611e-01 -1.16922021e+00 6.69609427e-01
7.74286032e-01 -3.00717205e-01 1.03398335e+00 5.37337065e-01
-9.49594855e-01 4.19811577e-01 -1.04199815e+00 8.55074525e-01
7.12463140e-01 -1.08310258e+00 -9.58264112e-01 2.08894908e-01
8.80826116e-01 2.05882609e-01 -6.45979762e-01 -4.07791063e-02
3.42437148e-01 -8.66605699e-01 8.74238908e-01 -5.84749579e-01
7.56708443e-01 -6.79096505e-02 -3.02108824e-01 -1.45599651e+00
-5.47086179e-01 -3.09583336e-01 -5.59813380e-02 1.42679131e+00
4.63600934e-01 -6.84951544e-01 3.24306011e-01 7.31368423e-01
-1.05868056e-02 -6.17958963e-01 -5.27726412e-01 -6.72835529e-01
4.98078793e-01 -5.45899332e-01 2.24031806e-01 1.20803177e+00
3.21643054e-01 4.89614844e-01 -4.77840066e-01 2.78361797e-01
3.19312990e-01 -3.98193032e-01 6.57332122e-01 -7.09639132e-01
-5.13059944e-02 -6.41167402e-01 -2.59736687e-01 -4.26228225e-01
5.40782392e-01 -1.02812433e+00 -3.13742399e-01 -1.64581537e+00
4.68981206e-01 1.44560307e-01 -4.38634694e-01 7.23797441e-01
-4.97885257e-01 2.55698085e-01 5.13024509e-01 -1.49640338e-02
-5.36629200e-01 4.57777768e-01 6.71979487e-01 -1.24289520e-01
4.90045175e-02 -6.75767422e-01 -9.94640648e-01 6.61217630e-01
1.14000201e+00 -6.04573071e-01 -3.49899083e-01 -7.42221296e-01
-6.44259974e-02 -6.01113021e-01 4.24880207e-01 -6.82318151e-01
-1.03237078e-01 -1.02772713e-02 1.68751866e-01 -1.58975661e-01
6.31466150e-01 -3.52947563e-01 -5.57034850e-01 3.41172963e-01
-6.03079617e-01 1.54868484e-01 4.22632694e-01 3.48394930e-01
-1.24525376e-01 -5.56363106e-01 7.59474635e-01 -6.26295134e-02
-7.92682230e-01 -2.13502031e-02 -6.36755466e-01 2.02058226e-01
6.21389866e-01 2.80695975e-01 -7.60071099e-01 -4.80937451e-01
-3.51286411e-01 -1.26537830e-01 6.00300908e-01 9.05168056e-01
5.06905138e-01 -1.25997388e+00 -7.17071891e-01 -1.90698318e-02
1.94661528e-01 -7.98561633e-01 6.38715625e-02 6.67875826e-01
-2.97072619e-01 6.41047001e-01 -5.45659056e-03 1.58211872e-01
-1.48971426e+00 8.28253269e-01 -1.14997171e-01 -1.59985214e-01
-4.97308284e-01 7.77761579e-01 1.68019637e-01 -4.25030947e-01
1.70411706e-01 1.36892304e-01 -1.93324625e-01 2.78592288e-01
7.93248236e-01 4.38729942e-01 -2.93248087e-01 -9.13206458e-01
-5.67101657e-01 2.36993358e-01 -2.33789816e-01 -1.47966281e-01
1.40163839e+00 -1.45442694e-01 -1.36141837e-01 4.78833526e-01
1.68605554e+00 3.05600613e-01 -7.86849022e-01 -1.59015015e-01
1.05780222e-01 -6.56200767e-01 3.15471590e-01 -5.98904371e-01
-8.35163653e-01 1.29668951e+00 3.13168436e-01 2.82390803e-01
6.81773305e-01 1.33764029e-01 9.28418756e-01 2.26413965e-01
-7.46222287e-02 -1.11477625e+00 2.28913233e-01 5.38181841e-01
7.79766619e-01 -1.37968373e+00 1.44826591e-01 -3.96928668e-01
-1.01089251e+00 9.20861900e-01 5.07623792e-01 -3.74464244e-01
7.76644051e-01 1.80435136e-01 2.80174017e-01 -2.40982220e-01
-8.94195080e-01 -5.59484899e-01 2.23720238e-01 7.07136214e-01
9.98292744e-01 2.10554913e-01 -4.45968926e-01 4.61078703e-01
-6.35720119e-02 -4.14258480e-01 4.79445040e-01 8.50671291e-01
-5.48493683e-01 -1.08955252e+00 -4.45845097e-01 3.71923029e-01
-9.06382263e-01 -4.71151084e-01 -7.59843469e-01 7.21307576e-01
3.01934388e-02 1.31545413e+00 -1.13572553e-01 -4.38408911e-01
5.37523488e-03 5.15117109e-01 1.92353949e-01 -8.79089177e-01
-5.31194210e-01 1.09486818e-01 5.82504928e-01 -3.16752732e-01
-2.81892568e-01 -7.51863003e-01 -1.10049927e+00 -5.19086242e-01
-1.78818837e-01 -1.19513929e-01 5.20204008e-01 7.50480592e-01
6.07064128e-01 1.77797094e-01 3.79883975e-01 -6.95993423e-01
-5.37953615e-01 -1.09821081e+00 -3.66474509e-01 5.71876407e-01
6.60452902e-01 -6.51727200e-01 -6.96328342e-01 -2.29490250e-01] | [8.819266319274902, 10.55526351928711] |
f34dd132-6600-492a-a398-4d9a2aac2546 | augmented-message-passing-stein-variational | 2305.10636 | null | https://arxiv.org/abs/2305.10636v1 | https://arxiv.org/pdf/2305.10636v1.pdf | Augmented Message Passing Stein Variational Gradient Descent | Stein Variational Gradient Descent (SVGD) is a popular particle-based method for Bayesian inference. However, its convergence suffers from the variance collapse, which reduces the accuracy and diversity of the estimation. In this paper, we study the isotropy property of finite particles during the convergence process and show that SVGD of finite particles cannot spread across the entire sample space. Instead, all particles tend to cluster around the particle center within a certain range and we provide an analytical bound for this cluster. To further improve the effectiveness of SVGD for high-dimensional problems, we propose the Augmented Message Passing SVGD (AUMP-SVGD) method, which is a two-stage optimization procedure that does not require sparsity of the target distribution, unlike the MP-SVGD method. Our algorithm achieves satisfactory accuracy and overcomes the variance collapse problem in various benchmark problems. | ['Yue Qiu', 'Jiankui Zhou'] | 2023-05-18 | null | null | null | null | ['bayesian-inference'] | ['methodology'] | [-4.43681389e-01 -4.64443505e-01 6.78014830e-02 1.41348457e-02
-7.33917415e-01 -4.71258201e-02 5.98133802e-01 -6.44130707e-02
-4.05872494e-01 1.06653368e+00 6.89276308e-02 -1.61623225e-01
-2.71567762e-01 -8.22748780e-01 -6.49692893e-01 -1.09626687e+00
1.67843238e-01 6.76613390e-01 5.08180320e-01 1.68987334e-01
2.76620835e-01 2.55289942e-01 -1.14033198e+00 -6.05393708e-01
9.65329051e-01 6.12646818e-01 4.34660196e-01 5.34116685e-01
-3.04102339e-02 2.46133819e-01 -4.68557984e-01 -6.87620342e-02
1.65016279e-02 -4.20098543e-01 -2.34361306e-01 -1.33602321e-01
1.30039409e-01 -3.69678825e-01 -1.59301594e-01 1.50255859e+00
6.07828319e-01 4.56270516e-01 9.04600203e-01 -1.18894160e+00
-3.92573535e-01 4.59160149e-01 -9.06754434e-01 3.13514680e-01
-3.97057459e-02 -2.70922989e-01 7.92089403e-01 -1.24397802e+00
3.46493989e-01 1.43320608e+00 9.14931178e-01 2.80516267e-01
-1.01837146e+00 -5.72565317e-01 3.39853317e-02 -1.24273367e-01
-1.62363589e+00 -1.44286618e-01 6.02234066e-01 -5.73210835e-01
3.42015743e-01 2.52191186e-01 7.53780663e-01 7.25730598e-01
2.27405086e-01 8.79140973e-01 9.35076237e-01 -1.61445215e-01
5.03596604e-01 -1.86799154e-01 4.39671606e-01 7.58725762e-01
8.27269793e-01 1.12407222e-01 -3.41168880e-01 -6.41683578e-01
8.23415220e-01 7.80334473e-02 -2.91028500e-01 -2.80539066e-01
-1.03054821e+00 1.19973946e+00 3.30379046e-02 5.87729961e-02
-2.13631794e-01 2.97705680e-01 7.59264231e-02 -3.38549644e-01
8.11408579e-01 -1.77900583e-01 -8.47726092e-02 -2.35450789e-01
-1.13176441e+00 6.50917888e-01 8.94686282e-01 6.63619518e-01
6.06927395e-01 3.64590362e-02 -5.63197136e-01 7.17167735e-01
1.01825106e+00 1.06124651e+00 5.60679100e-03 -9.16088760e-01
3.68915081e-01 6.35657534e-02 5.49209952e-01 -1.22161829e+00
-3.54097456e-01 -5.50242603e-01 -1.14704967e+00 5.78457825e-02
5.79949796e-01 -4.57992584e-01 -5.99843323e-01 1.62941873e+00
8.65713656e-01 6.27851009e-01 -2.19144940e-01 1.09481764e+00
5.64691603e-01 9.43087816e-01 -3.24631393e-01 -4.79527116e-01
1.08046865e+00 -8.55839312e-01 -6.05731785e-01 3.65669839e-02
1.53341308e-01 -5.92297733e-01 8.71866047e-01 1.58719525e-01
-1.10670567e+00 -2.88806140e-01 -1.10509038e+00 2.98299283e-01
3.11018497e-01 1.23488847e-02 4.93391544e-01 7.49124050e-01
-6.57022834e-01 6.00000501e-01 -1.27235246e+00 3.86334062e-02
5.23957789e-01 5.01449667e-02 3.52148861e-01 -2.41541229e-02
-9.09724236e-01 4.52114433e-01 1.14010662e-01 2.18346298e-01
-8.67264569e-01 -1.04657876e+00 -5.46136916e-01 7.39324912e-02
6.01053834e-02 -8.01495135e-01 9.21495914e-01 1.52373224e-01
-1.74375892e+00 1.23749860e-01 -6.36489570e-01 -3.29591393e-01
8.47385287e-01 -1.96383387e-01 4.41199057e-02 -1.81400210e-01
1.12417430e-01 -8.54276791e-02 1.13797438e+00 -1.13813913e+00
-4.87520248e-01 -3.88280451e-01 -4.03567284e-01 8.94618183e-02
-4.78911661e-02 -3.26846361e-01 -5.33134162e-01 -4.83505875e-01
4.17504549e-01 -1.11659765e+00 -4.01511163e-01 -2.34896496e-01
-5.42217195e-01 -2.86351293e-01 6.92799628e-01 -2.20743239e-01
1.28756368e+00 -2.25173640e+00 3.59715405e-03 4.39599842e-01
4.58999991e-01 1.25730082e-01 3.87794703e-01 2.42745906e-01
7.58618295e-01 -1.20242767e-01 -2.72252023e-01 -6.69839144e-01
2.58835524e-01 2.73716688e-01 -3.31160754e-01 1.09277833e+00
-4.92642432e-01 4.47745979e-01 -1.14467907e+00 -4.79326636e-01
2.13264868e-01 4.62425888e-01 -6.98712766e-01 -1.35338500e-01
-1.37826353e-01 5.64607680e-01 -7.76260376e-01 1.60295650e-01
1.37992775e+00 -6.06052339e-01 -7.81229362e-02 -1.47758171e-01
-2.20747143e-01 -5.59295937e-02 -1.60242510e+00 1.23659253e+00
-2.33000088e-02 3.54936093e-01 3.53496283e-01 -7.53539085e-01
6.91129327e-01 -4.26868647e-02 5.88474214e-01 8.53831470e-02
1.53290331e-01 2.45152816e-01 -2.35193491e-01 -6.59827515e-02
5.35540283e-01 -3.11210573e-01 2.21714467e-01 5.98847210e-01
-4.32117611e-01 3.01160431e-03 3.94222111e-01 2.49090508e-01
7.01341212e-01 -3.48346174e-01 -9.45729297e-03 -6.51402712e-01
5.68471313e-01 -1.88081771e-01 7.09144831e-01 1.22821164e+00
-4.65490408e-02 6.27026081e-01 2.35164493e-01 -6.64080903e-02
-9.78323698e-01 -1.41692376e+00 -6.28830791e-01 7.97880650e-01
5.39130449e-01 -4.02647793e-01 -5.87697387e-01 -2.59491861e-01
3.34267974e-01 4.76384729e-01 -4.52266723e-01 4.41331640e-02
-3.22453082e-01 -1.50300586e+00 2.77108788e-01 1.57822222e-01
6.92662477e-01 -2.92777330e-01 -2.89157271e-01 3.57638806e-01
-2.32804194e-02 -8.17303777e-01 -7.30946839e-01 -5.10327399e-01
-8.26643586e-01 -1.04977715e+00 -7.76722610e-01 -4.03386384e-01
7.57299185e-01 3.05167139e-01 8.82702529e-01 -2.60391571e-02
1.01497993e-01 6.96651340e-02 -7.17857946e-03 -3.06338847e-01
-3.52093071e-01 -2.30128035e-01 3.08495939e-01 -1.44855738e-01
2.73237228e-01 -3.43098313e-01 -6.82935357e-01 4.03845727e-01
-3.38938922e-01 -6.19193614e-02 1.09763265e-01 9.31067586e-01
6.87974215e-01 3.68188888e-01 1.50700733e-01 -6.50312185e-01
9.67598975e-01 -4.95772123e-01 -1.16824389e+00 -6.15631379e-02
-6.01206064e-01 7.83395842e-02 2.87513763e-01 -3.68170053e-01
-1.02246284e+00 -2.49629959e-01 -1.89723328e-01 -3.42753202e-01
5.82868636e-01 7.16227353e-01 2.12112740e-01 -1.68473214e-01
2.96860874e-01 3.17035884e-01 3.11220258e-01 -6.16848171e-01
1.61224186e-01 4.51209456e-01 2.69904763e-01 -7.53991961e-01
7.51885235e-01 9.11337376e-01 4.21181649e-01 -1.00700462e+00
-1.12325525e+00 -4.49054241e-01 -6.01381958e-02 -5.93448281e-02
5.82638144e-01 -8.47156227e-01 -1.11102104e+00 5.06530941e-01
-1.02858937e+00 -2.39732757e-01 -1.15184940e-01 1.05719793e+00
-2.62641728e-01 7.00012982e-01 -7.77721345e-01 -9.73045409e-01
-3.11652601e-01 -1.07147408e+00 8.10939610e-01 2.26769373e-01
1.81070641e-01 -1.20049858e+00 5.48966348e-01 -6.72877431e-02
2.16236383e-01 4.80985679e-02 4.80889827e-01 -2.75981963e-01
-4.08535242e-01 -8.04355890e-02 -3.09658915e-01 2.08645955e-01
-3.98458727e-02 1.04016013e-01 -2.23813623e-01 -7.05717564e-01
4.15240943e-01 2.28189275e-01 8.51348460e-01 1.10047877e+00
7.77483642e-01 -1.83060393e-01 -5.92605889e-01 6.98712945e-01
1.34344494e+00 -2.78594434e-01 3.68839562e-01 1.19058639e-01
6.25630915e-01 -9.82872099e-02 5.43397903e-01 8.96609426e-01
3.59465003e-01 3.32203388e-01 1.94642022e-01 1.03407137e-01
5.38175851e-02 -1.92695901e-01 3.19304377e-01 1.05639839e+00
1.27711352e-02 -3.27881426e-01 -6.83742046e-01 2.70263463e-01
-2.14900756e+00 -9.48969960e-01 -6.65076375e-01 2.27707386e+00
7.85091400e-01 -2.90138274e-02 6.06659092e-02 -1.55537561e-01
1.09465277e+00 1.97538957e-01 -6.35596156e-01 1.61935404e-01
4.09750156e-02 -1.46220595e-01 7.33352721e-01 8.85719180e-01
-9.87570524e-01 6.06961310e-01 7.38726473e+00 1.18317819e+00
-6.04061663e-01 4.94689137e-01 2.36111153e-02 1.96138285e-02
-1.09137870e-01 -1.05105251e-01 -1.49751306e+00 1.06141973e+00
4.82188642e-01 -3.76526326e-01 2.18031317e-01 7.37080038e-01
4.46183830e-01 -2.72985280e-01 -6.06349170e-01 9.93373096e-01
-1.67004168e-01 -1.51856542e+00 -2.15248808e-01 2.74685025e-01
1.05709732e+00 3.04825038e-01 3.42206173e-02 8.14451724e-02
7.87665486e-01 -7.07434952e-01 3.40140432e-01 5.06898582e-01
1.82585880e-01 -7.83489943e-01 6.06777787e-01 5.94591737e-01
-1.12085664e+00 2.52486944e-01 -1.02210355e+00 -1.60604984e-01
6.71580076e-01 1.38677657e+00 -5.29962420e-01 1.45677358e-01
5.52513182e-01 6.57629967e-01 -1.18938414e-02 1.26809323e+00
-1.25709817e-01 7.50709653e-01 -9.51419890e-01 -4.17223632e-01
3.31668615e-01 -8.06732714e-01 1.18872976e+00 9.07582521e-01
6.83049262e-01 -1.09860249e-01 3.81445736e-01 9.62886214e-01
1.31974980e-01 -8.83937776e-02 -1.66122522e-02 9.71317813e-02
8.15228522e-01 7.99121678e-01 -6.38737082e-01 -4.32414860e-01
-3.85150284e-01 4.58893210e-01 1.65007398e-01 3.51048023e-01
-8.97567749e-01 -1.94448367e-01 7.07551181e-01 -4.84589376e-02
7.03359604e-01 -4.70302969e-01 -1.79268718e-01 -1.29359186e+00
-8.12584236e-02 -3.51060182e-01 2.77150333e-01 -3.37137789e-01
-1.55117428e+00 6.00023568e-03 2.19137059e-03 -1.09426367e+00
-7.78835192e-02 -2.38234594e-01 -6.39957309e-01 7.41372705e-01
-1.10663915e+00 -6.85239434e-01 -2.72471905e-01 6.00741148e-01
1.12264052e-01 -8.52384046e-02 2.43420482e-01 3.11625481e-01
-5.72541356e-01 3.22827250e-01 1.05358446e+00 -3.55111063e-02
2.57222027e-01 -1.01744258e+00 1.45658940e-01 8.67775261e-01
-9.53631178e-02 7.21839786e-01 1.02851379e+00 -9.34411705e-01
-1.42939258e+00 -1.03095853e+00 4.05726850e-01 -1.82457790e-01
7.55506754e-01 -1.97170496e-01 -8.24759424e-01 3.91857862e-01
-3.03431451e-01 7.70088136e-02 5.48717082e-01 2.30327189e-01
1.80485010e-01 1.20515294e-01 -1.02898884e+00 6.26282096e-01
7.21057892e-01 -1.01375647e-01 -5.71305573e-01 7.03882992e-01
3.41210335e-01 -5.00650406e-01 -9.46840048e-01 2.42143020e-01
4.06021148e-01 -8.85809898e-01 1.21819699e+00 -2.37798125e-01
1.40444292e-02 -7.27106988e-01 -1.14140891e-01 -1.27244580e+00
-4.25333291e-01 -5.73185325e-01 -4.96255636e-01 1.06950665e+00
1.28091678e-01 -8.04297328e-01 1.03746784e+00 2.59088606e-01
1.44253932e-02 -5.47853827e-01 -1.04470706e+00 -1.01799405e+00
2.43035108e-01 -4.92273241e-01 5.47227621e-01 5.39189577e-01
-2.45483339e-01 2.40317568e-01 -6.07017457e-01 4.74882036e-01
1.45916736e+00 2.87712723e-01 8.58994305e-01 -1.33031595e+00
-4.62343156e-01 -3.45559984e-01 -1.29472926e-01 -1.60970449e+00
1.46851346e-01 -7.46301174e-01 2.07192972e-01 -1.55114937e+00
4.69629377e-01 -5.66830337e-01 -1.85712576e-02 -3.35151702e-01
-4.83455628e-01 2.27018923e-01 -8.31808299e-02 5.78702509e-01
-6.39004409e-01 8.67542267e-01 1.42350769e+00 1.03892013e-01
-2.35983610e-01 2.71962643e-01 -3.02980453e-01 9.30266082e-01
6.29768968e-01 -7.46954322e-01 -3.83707464e-01 -2.96634078e-01
3.88087064e-01 -5.53732961e-02 2.32318327e-01 -9.36166048e-01
5.05574286e-01 -1.40757784e-01 4.03576970e-01 -1.11855936e+00
3.95683318e-01 -3.30490470e-01 3.04026157e-01 6.97808444e-01
1.42552420e-01 -4.29061592e-01 -1.59918904e-01 9.91214573e-01
2.88889073e-02 -4.90071863e-01 1.05083084e+00 3.89619842e-02
-7.24039450e-02 6.70043170e-01 -5.58480859e-01 3.55621904e-01
7.08407223e-01 1.21628821e-01 -3.18289906e-01 -3.04661602e-01
-6.44662440e-01 4.75833088e-01 3.42956334e-01 -3.46941680e-01
5.53736329e-01 -1.50247741e+00 -8.83044541e-01 4.53413613e-02
-4.38208073e-01 1.23398222e-01 2.94276983e-01 1.20285571e+00
-5.23135126e-01 1.78915516e-01 4.82785404e-01 -9.52708602e-01
-1.01641178e+00 2.93885857e-01 1.54568166e-01 -1.91863284e-01
-8.93105805e-01 9.34610724e-01 1.06724523e-01 -3.60393822e-01
-5.01011759e-02 -5.67410104e-02 -2.96417158e-02 -2.05421463e-01
6.87231481e-01 8.87701869e-01 -4.13561881e-01 -4.09017652e-01
-2.95762211e-01 6.48631454e-01 2.66877990e-02 -2.50557244e-01
1.16083467e+00 -3.90660554e-01 -3.13395768e-01 4.22738850e-01
1.29048491e+00 2.81533182e-01 -1.53397810e+00 -1.98903009e-01
-3.50017071e-01 -7.65510798e-01 4.17875797e-01 1.28507704e-01
-9.79747415e-01 5.93003631e-01 3.48949015e-01 3.43832642e-01
2.21645862e-01 -3.48549918e-03 1.03247917e+00 3.68403465e-01
3.33563268e-01 -1.13341510e+00 -1.68796778e-01 8.10520768e-01
4.18198705e-01 -1.09887946e+00 6.28261685e-01 -4.27615225e-01
-2.38694951e-01 7.85492539e-01 1.53291211e-01 -4.32188094e-01
1.28967059e+00 2.08319932e-01 -5.34615934e-01 -2.59865880e-01
-3.73337179e-01 1.94224622e-02 1.93178236e-01 3.58695626e-01
4.51110117e-02 2.25616489e-02 -7.31111825e-01 5.31303346e-01
-2.69328028e-01 -4.61554267e-02 3.05848479e-01 8.38294744e-01
-7.68440902e-01 -7.54742324e-01 -5.22678018e-01 7.11397767e-01
-4.84578580e-01 -9.52785239e-02 4.04996306e-01 6.40185118e-01
-3.37685823e-01 9.16351140e-01 1.61128178e-01 9.96890664e-02
-1.07881933e-01 -4.90265518e-01 6.00344360e-01 -3.20414096e-01
7.00168014e-02 3.71609807e-01 -2.15094775e-01 -2.81688988e-01
-3.70116472e-01 -8.71437192e-01 -1.14125192e+00 -8.62279832e-01
-5.37791312e-01 6.89433753e-01 6.13070607e-01 1.11916506e+00
2.52305210e-01 2.20476002e-01 5.04953504e-01 -8.25149000e-01
-9.30998504e-01 -8.58107746e-01 -1.05653131e+00 6.59014881e-02
3.00556511e-01 -1.14756536e+00 -9.64247704e-01 -4.79155689e-01] | [6.871944427490234, 4.049733638763428] |
83604973-a0cc-458c-b7d7-44c74d843fd2 | ev-tta-test-time-adaptation-for-event-based | 2203.12247 | null | https://arxiv.org/abs/2203.12247v2 | https://arxiv.org/pdf/2203.12247v2.pdf | Ev-TTA: Test-Time Adaptation for Event-Based Object Recognition | We introduce Ev-TTA, a simple, effective test-time adaptation algorithm for event-based object recognition. While event cameras are proposed to provide measurements of scenes with fast motions or drastic illumination changes, many existing event-based recognition algorithms suffer from performance deterioration under extreme conditions due to significant domain shifts. Ev-TTA mitigates the severe domain gaps by fine-tuning the pre-trained classifiers during the test phase using loss functions inspired by the spatio-temporal characteristics of events. Since the event data is a temporal stream of measurements, our loss function enforces similar predictions for adjacent events to quickly adapt to the changed environment online. Also, we utilize the spatial correlations between two polarities of events to handle noise under extreme illumination, where different polarities of events exhibit distinctive noise distributions. Ev-TTA demonstrates a large amount of performance gain on a wide range of event-based object recognition tasks without extensive additional training. Our formulation can be successfully applied regardless of input representations and further extended into regression tasks. We expect Ev-TTA to provide the key technique to deploy event-based vision algorithms in challenging real-world applications where significant domain shift is inevitable. | ['Young Min Kim', 'Inwoo Hwang', 'Junho Kim'] | 2022-03-23 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Kim_Ev-TTA_Test-Time_Adaptation_for_Event-Based_Object_Recognition_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Kim_Ev-TTA_Test-Time_Adaptation_for_Event-Based_Object_Recognition_CVPR_2022_paper.pdf | cvpr-2022-1 | ['event-based-vision'] | ['computer-vision'] | [ 4.48079199e-01 -8.44595969e-01 3.54250637e-03 -6.19037390e-01
-7.08248734e-01 -5.18673301e-01 7.06869781e-01 1.58986479e-01
-4.01992351e-01 4.62273806e-01 -2.08003018e-02 8.04090966e-03
-3.07416677e-01 -6.43031836e-01 -8.64693642e-01 -8.19513202e-01
-2.59046406e-01 3.23654383e-01 5.39747417e-01 5.67392148e-02
1.16672300e-01 8.55943918e-01 -1.72382903e+00 3.47666860e-01
4.51286733e-01 1.29941428e+00 2.05181360e-01 7.31809974e-01
-3.13779153e-02 7.92875171e-01 -7.35967159e-01 6.59839958e-02
4.93234903e-01 -2.95585066e-01 -2.16501877e-01 3.78542483e-01
4.88872141e-01 -1.83983073e-01 -6.56544983e-01 6.67397201e-01
5.53712368e-01 5.03369033e-01 5.07045865e-01 -1.39949405e+00
-3.72012913e-01 -1.76624149e-01 -4.55303043e-01 7.14261889e-01
3.96572620e-01 4.04640764e-01 4.83694136e-01 -9.73021924e-01
4.98377323e-01 9.75731254e-01 6.77261889e-01 3.17246884e-01
-1.13400316e+00 -6.09077752e-01 4.94623780e-01 9.67854321e-01
-1.13751769e+00 -3.74840587e-01 9.05374467e-01 -4.50207263e-01
1.11627829e+00 9.67754349e-02 6.47711337e-01 1.57708180e+00
1.54837906e-01 6.81799293e-01 9.08166289e-01 -1.18490197e-01
5.40638089e-01 -3.20947081e-01 -5.44626452e-02 1.83933362e-01
1.50155770e-02 1.22816972e-01 -9.69974995e-01 1.56480502e-02
5.59302509e-01 4.07649755e-01 -3.30364764e-01 -5.11322975e-01
-1.43577969e+00 4.84341413e-01 2.74377704e-01 -2.38122880e-01
-6.18210733e-01 -1.67386662e-02 6.96333230e-01 5.71174502e-01
2.03319669e-01 2.43507177e-01 -5.86827815e-01 -2.22723976e-01
-7.27092028e-01 6.33417442e-02 5.96945763e-01 8.47306192e-01
5.57765305e-01 3.98777425e-01 -3.56714785e-01 6.85823798e-01
-2.46251840e-02 5.99700809e-01 5.21712780e-01 -7.29378998e-01
5.23841083e-01 3.83500457e-01 3.14947069e-01 -9.21920955e-01
-3.13718289e-01 -4.56261843e-01 -8.22640538e-01 2.08124951e-01
5.93268394e-01 2.54536811e-02 -9.64029133e-01 1.68075252e+00
5.42637944e-01 7.80802250e-01 -4.36486267e-02 9.24003303e-01
3.75425428e-01 7.40634024e-01 2.21501775e-02 -5.54015696e-01
1.02006900e+00 -5.00274599e-01 -4.71963704e-01 -4.20107305e-01
-4.76266183e-02 -6.70128047e-01 9.82699275e-01 4.39318419e-01
-7.71378577e-01 -6.96143508e-01 -1.01384604e+00 4.62061077e-01
-3.37092221e-01 -3.23449194e-01 6.03737116e-01 1.11856349e-01
-5.38450718e-01 4.10651475e-01 -1.06160915e+00 -6.40106499e-01
3.68321598e-01 3.00314993e-01 -4.23457175e-01 -2.53016055e-01
-8.30065370e-01 6.14271045e-01 1.43725082e-01 2.84693658e-01
-1.15473509e+00 -7.86859870e-01 -8.56004179e-01 -2.27054417e-01
4.02085960e-01 -3.95783037e-01 1.20189834e+00 -8.43447447e-01
-1.28084064e+00 6.94260359e-01 -3.62488925e-01 -6.30375743e-01
5.62256634e-01 -1.38616547e-01 -7.44530082e-01 1.86797678e-01
5.96760139e-02 2.73864478e-01 1.04979765e+00 -8.11287463e-01
-7.33989716e-01 -2.67875403e-01 -2.32499272e-01 1.66811213e-01
-3.26309532e-01 -9.77432937e-04 -4.11795497e-01 -7.98653066e-01
1.61993474e-01 -7.05932856e-01 -1.43339500e-01 3.01194340e-01
1.77101821e-01 -1.28785029e-01 1.19188857e+00 -1.56604558e-01
7.94166744e-01 -2.22923923e+00 -1.59130171e-01 1.66154429e-02
-3.11554164e-01 2.40787432e-01 -3.39004427e-01 5.53850532e-01
-5.94860725e-02 -5.78414202e-01 -1.32172123e-01 -1.96352322e-02
-1.42234311e-01 5.09541214e-01 -8.63645613e-01 5.12830317e-01
5.33084452e-01 4.45092410e-01 -1.02114570e+00 -2.00073108e-01
5.32273889e-01 3.76880944e-01 -2.75570422e-01 4.08538431e-01
-3.16494882e-01 7.30960011e-01 -3.76939476e-01 7.10029304e-01
5.22701085e-01 -2.90511668e-01 -1.63544506e-01 -3.03358048e-01
3.17936763e-02 2.24369958e-01 -1.31097591e+00 1.67307258e+00
-3.37683231e-01 9.77574766e-01 -2.00798914e-01 -1.38834798e+00
9.57069933e-01 1.98890120e-01 9.09906328e-01 -1.02981305e+00
-1.22499041e-01 -9.69490260e-02 -1.55669689e-01 -6.56729162e-01
2.71327049e-01 -1.00562524e-03 -9.04689580e-02 1.96962133e-02
2.49838363e-02 4.97643687e-02 1.98162958e-01 -1.60529673e-01
1.51915097e+00 1.87954456e-01 3.02704632e-01 4.13009137e-01
3.16993952e-01 7.73298070e-02 1.09667671e+00 1.04156017e+00
-5.00739276e-01 7.86460578e-01 -1.47008419e-01 -7.60146618e-01
-8.90207350e-01 -1.43356049e+00 -3.85348856e-01 1.07038784e+00
4.07016784e-01 3.56901214e-02 1.01545699e-01 -6.25823498e-01
1.08759455e-01 7.07213044e-01 -4.28493679e-01 -3.89159054e-01
-7.25435793e-01 -1.03364837e+00 1.79262266e-01 8.75885546e-01
6.35103703e-01 -9.78414714e-01 -8.83956850e-01 5.75180709e-01
-1.83700800e-01 -1.38766599e+00 -4.13871288e-01 4.94190693e-01
-6.85684025e-01 -1.09811532e+00 -3.46258670e-01 -6.40953302e-01
3.38022560e-01 5.40935576e-01 1.11534107e+00 -6.60127044e-01
-5.48166871e-01 8.59426439e-01 -5.00028789e-01 -6.11275673e-01
7.22381994e-02 -6.90350711e-01 2.34598935e-01 4.72987592e-01
5.62738061e-01 -8.93526256e-01 -8.73228431e-01 7.33363390e-01
-8.44923675e-01 -3.93359870e-01 1.96063817e-01 1.00178266e+00
9.01667714e-01 1.59354091e-01 5.67057073e-01 -2.94664174e-01
-1.16320685e-01 -6.34280801e-01 -6.49107456e-01 2.86079288e-01
-3.08526605e-01 -2.71687150e-01 7.05191314e-01 -1.12168777e+00
-1.13163233e+00 1.20820336e-01 3.27022135e-01 -7.07651019e-01
-1.53021157e-01 2.41851211e-01 -7.96208680e-02 -5.82335070e-02
7.00878620e-01 5.21857858e-01 -5.65486252e-01 -1.97819948e-01
-1.33205369e-01 3.64348203e-01 8.59418392e-01 -5.94964445e-01
8.59752834e-01 8.40198040e-01 1.07084252e-01 -6.94635928e-01
-9.19914544e-01 -8.45826089e-01 -3.61305892e-01 -4.53669906e-01
5.90281248e-01 -1.10238099e+00 -5.74030578e-01 6.50002062e-01
-9.85275686e-01 -4.39293027e-01 -6.27207875e-01 7.46544719e-01
-5.47834039e-01 1.73286602e-01 -2.90463269e-01 -7.20325351e-01
5.69125463e-04 -7.05550432e-01 1.01043546e+00 3.84489596e-01
9.26467106e-02 -8.14618886e-01 1.41300797e-01 9.00762826e-02
1.26417086e-01 4.06226426e-01 5.06866992e-01 -7.18885183e-01
-6.39506519e-01 -3.89714330e-01 -1.50475755e-01 2.95702934e-01
2.69617736e-01 -1.55434357e-02 -8.79605770e-01 -4.52084899e-01
2.75144279e-01 -3.83987695e-01 8.95453632e-01 3.77438337e-01
1.27926803e+00 7.70225748e-02 -1.34162039e-01 9.06516731e-01
1.33898377e+00 6.24625742e-01 5.19862533e-01 4.62514281e-01
5.06751001e-01 1.86198041e-01 9.16678965e-01 9.83149290e-01
3.18511128e-01 8.22072804e-01 5.49861670e-01 1.33925050e-01
-1.89972579e-01 -1.54545501e-01 6.13161445e-01 4.63286161e-01
2.56810457e-01 -3.97623688e-01 -9.84548748e-01 8.35956097e-01
-2.03174949e+00 -1.21906722e+00 6.61922023e-02 2.49692082e+00
6.32754624e-01 2.09910393e-01 5.70021830e-02 1.17285147e-01
6.33930743e-01 2.43102506e-01 -1.15967035e+00 5.82073256e-02
-3.95261407e-01 8.10330361e-02 4.16846812e-01 -2.46453375e-01
-1.23580325e+00 4.35197681e-01 6.46892595e+00 4.81716007e-01
-1.42905271e+00 1.33850705e-02 1.53323621e-01 -4.82729673e-01
2.59194732e-01 -1.17347516e-01 -6.21081412e-01 6.06601477e-01
9.48258579e-01 -1.53431222e-01 1.81874812e-01 7.09530473e-01
3.54687810e-01 -6.01433516e-02 -1.38478291e+00 1.15950727e+00
2.19925083e-02 -1.16643798e+00 -4.10374403e-02 -4.04199839e-01
6.11115277e-01 3.64345580e-01 1.84670463e-02 3.57775807e-01
3.12235862e-01 -5.94024897e-01 7.88556099e-01 5.03048241e-01
5.46113253e-01 -3.15608501e-01 3.02018583e-01 1.80202961e-01
-1.33947051e+00 -2.85987884e-01 -4.23822463e-01 -1.29774958e-01
1.39750570e-01 6.63190484e-01 -8.67133617e-01 3.66764009e-01
9.23958123e-01 1.00863385e+00 -3.74991149e-01 1.39316714e+00
1.32381856e-01 9.06197667e-01 -5.17066658e-01 3.61350417e-01
-1.83150973e-02 2.39482000e-02 9.08392072e-01 1.23016024e+00
5.15534818e-01 -1.29659122e-04 5.30945778e-01 2.93940693e-01
1.37248769e-01 -3.83215725e-01 -6.76788330e-01 2.44036287e-01
6.13110244e-01 8.80624413e-01 -5.22860646e-01 -2.43416056e-01
-6.83439553e-01 1.16130996e+00 1.92392677e-01 7.53633499e-01
-1.03914511e+00 -2.16328219e-01 8.65707338e-01 -7.71867065e-03
7.74580538e-01 -4.67794031e-01 -1.16897739e-01 -1.44853854e+00
3.19801986e-01 -7.45411217e-01 6.87171817e-01 -8.17590833e-01
-1.59416127e+00 2.39400968e-01 -9.06479603e-04 -1.78772140e+00
-2.58410871e-01 -6.65929258e-01 -8.22341502e-01 3.90359163e-01
-1.61302340e+00 -9.62061465e-01 -4.96585280e-01 9.96213019e-01
9.63961482e-01 -2.46384591e-01 5.84428132e-01 3.30806702e-01
-7.18987763e-01 4.56712335e-01 3.21870208e-01 9.77124721e-02
1.09174061e+00 -1.11152613e+00 1.69799253e-01 1.15897322e+00
3.16845387e-01 9.42282304e-02 8.88487518e-01 -5.52758634e-01
-1.74400032e+00 -1.54830468e+00 3.66810888e-01 -3.23473215e-01
7.75059938e-01 -3.45157117e-01 -1.11289012e+00 7.13516057e-01
-1.85415372e-01 6.50040507e-01 6.15233481e-01 3.18251620e-03
-8.43677104e-01 -8.02694380e-01 -9.84149456e-01 3.58146042e-01
1.16331327e+00 -6.61681116e-01 -6.29147768e-01 5.32102346e-01
3.76593769e-01 -3.86992753e-01 -8.15202177e-01 6.50713086e-01
3.83212417e-01 -8.22479248e-01 1.07578647e+00 -6.30734205e-01
-6.60006627e-02 -5.21268010e-01 -4.05124635e-01 -1.12074494e+00
-3.36021721e-01 -5.89894652e-01 -4.16479707e-01 1.08598757e+00
-5.58139458e-02 -6.61104143e-01 6.10052645e-01 3.85077357e-01
-7.03610107e-02 -2.41741881e-01 -1.17393649e+00 -1.15039968e+00
-4.94021565e-01 -8.01375866e-01 2.83535391e-01 8.92279267e-01
-3.76690865e-01 -1.27363995e-01 -3.61218870e-01 6.50935292e-01
7.62029827e-01 3.88239473e-01 6.51026666e-01 -1.08647585e+00
-4.31748629e-01 -5.38270315e-03 -6.99865222e-01 -9.54426765e-01
-1.40501767e-01 -4.77863014e-01 3.65956724e-01 -1.07375622e+00
2.37856749e-02 -2.40293428e-01 -9.19768810e-01 4.11020488e-01
-1.22751586e-01 2.80541480e-01 8.78826305e-02 3.95455748e-01
-9.31149364e-01 6.41979456e-01 8.32567036e-01 -2.47992858e-01
-1.35884538e-01 9.35209617e-02 -2.86697801e-02 7.52768636e-01
5.21944880e-01 -4.57824498e-01 -6.28520429e-01 -5.55964649e-01
7.63764083e-02 1.31479993e-01 7.27347791e-01 -1.39607847e+00
4.97132510e-01 -7.15452790e-01 6.99595869e-01 -7.02797353e-01
4.85630631e-01 -1.00770652e+00 3.10371011e-01 1.48477167e-01
-1.49554506e-01 9.68338177e-02 3.57982397e-01 1.13202512e+00
-4.71881539e-01 3.24112952e-01 8.15374196e-01 6.15030825e-02
-1.19137001e+00 5.13454437e-01 -4.42238957e-01 1.16183959e-01
1.29352593e+00 -6.07787073e-01 -1.36705935e-01 -2.85214514e-01
-6.03213787e-01 2.54950345e-01 3.23817521e-01 3.96066427e-01
6.67021215e-01 -1.32267106e+00 -5.36572635e-01 3.22631925e-01
3.45153123e-01 9.32536572e-02 4.14217561e-01 6.79837108e-01
-1.04593575e-01 -1.21517424e-02 -3.47291052e-01 -1.02084208e+00
-1.09937656e+00 5.23441195e-01 2.36071646e-01 -8.80903900e-02
-8.64199758e-01 7.71620333e-01 3.68505001e-01 4.47198823e-02
4.67875183e-01 -2.21081018e-01 1.85923830e-01 1.06102459e-01
6.33086145e-01 3.34513783e-01 1.98325694e-01 -2.59409547e-01
-5.91616511e-01 4.62787956e-01 -1.99769378e-01 1.04241230e-01
1.53199124e+00 -2.14564249e-01 5.97474158e-01 6.98856175e-01
9.67049778e-01 -4.26058680e-01 -2.01765704e+00 -5.24752557e-01
-1.02441214e-01 -5.01332700e-01 -6.90718414e-03 -8.47625852e-01
-9.02906895e-01 8.07380795e-01 8.24521184e-01 -7.53929690e-02
1.76467228e+00 -3.16447884e-01 7.39837110e-01 6.68075800e-01
4.91631210e-01 -1.12760627e+00 3.87166649e-01 6.20090127e-01
8.12867165e-01 -1.46408463e+00 -1.76443845e-01 -1.71952825e-02
-5.00230730e-01 1.02205765e+00 6.25157893e-01 -1.47623569e-01
5.21125495e-01 3.23382288e-01 -2.41035279e-02 1.34654582e-01
-1.06339622e+00 -2.43274659e-01 2.53639460e-01 9.83384132e-01
-2.65477866e-01 -2.87556678e-01 2.53656745e-01 1.74363717e-01
3.45795512e-01 -4.93043102e-02 2.72035778e-01 9.89449322e-01
-2.24015027e-01 -7.50364840e-01 -5.66490293e-01 3.51072818e-01
-6.42368272e-02 2.86074013e-01 -9.01619345e-03 6.62581623e-01
1.69219092e-01 9.66743112e-01 4.15710509e-01 -3.24335933e-01
6.07821465e-01 7.70174414e-02 4.81588691e-01 -3.64782929e-01
-3.87887061e-01 -1.60539225e-02 -1.29802495e-01 -8.67919683e-01
-5.08405924e-01 -1.09943235e+00 -1.17862785e+00 -5.80460168e-02
7.53154978e-02 -4.01564509e-01 4.89918083e-01 9.19481516e-01
5.67110121e-01 5.95597863e-01 9.37958598e-01 -8.52732837e-01
-5.88479996e-01 -6.12285256e-01 -3.46121669e-01 8.31823289e-01
6.06688142e-01 -8.35589051e-01 -3.37477297e-01 3.89372557e-01] | [8.476905822753906, -0.9137821793556213] |
fa6d08ae-4eed-4ef6-bdcd-7c8d093cb1d6 | introducing-model-inversion-attacks-on | 2301.03206 | null | https://arxiv.org/abs/2301.03206v1 | https://arxiv.org/pdf/2301.03206v1.pdf | Introducing Model Inversion Attacks on Automatic Speaker Recognition | Model inversion (MI) attacks allow to reconstruct average per-class representations of a machine learning (ML) model's training data. It has been shown that in scenarios where each class corresponds to a different individual, such as face classifiers, this represents a severe privacy risk. In this work, we explore a new application for MI: the extraction of speakers' voices from a speaker recognition system. We present an approach to (1) reconstruct audio samples from a trained ML model and (2) extract intermediate voice feature representations which provide valuable insights into the speakers' biometrics. Therefore, we propose an extension of MI attacks which we call sliding model inversion. Our sliding MI extends standard MI by iteratively inverting overlapping chunks of the audio samples and thereby leveraging the sequential properties of audio data for enhanced inversion performance. We show that one can use the inverted audio data to generate spoofed audio samples to impersonate a speaker, and execute voice-protected commands for highly secured systems on their behalf. To the best of our knowledge, our work is the first one extending MI attacks to audio data, and our results highlight the security risks resulting from the extraction of the biometric data in that setup. | ['Konstantin Böttinger', 'Ugur Sahin', 'Franziska Boenisch', 'Karla Pizzi'] | 2023-01-09 | null | null | null | null | ['speaker-recognition'] | ['speech'] | [ 1.11009920e+00 3.53711843e-01 7.20955804e-02 -2.71496415e-01
-1.01657939e+00 -9.08332288e-01 3.86074007e-01 -1.84865400e-01
-1.25739127e-01 3.96457464e-01 -7.20246062e-02 -4.39632565e-01
3.32158282e-02 -4.91077214e-01 -7.97393620e-01 -7.99559653e-01
-1.85136393e-01 1.84149653e-01 -1.34754643e-01 1.91391557e-01
2.69000858e-01 8.09519410e-01 -1.63321555e+00 6.38265073e-01
3.26650113e-01 9.74947453e-01 -5.46924770e-01 9.88222241e-01
2.25691661e-01 1.96984038e-01 -9.42433238e-01 -7.18919158e-01
3.23058903e-01 -3.75214756e-01 -7.42385268e-01 -1.93299249e-01
5.27572274e-01 -3.27372372e-01 -1.86395794e-01 1.02125573e+00
5.02158165e-01 -2.84558624e-01 5.73375165e-01 -1.66667855e+00
1.96299896e-01 7.75741518e-01 -3.22133690e-01 -2.27616221e-01
6.98575974e-01 -1.78917289e-01 7.12695360e-01 -6.56763732e-01
3.59431386e-01 1.17945981e+00 7.19593585e-01 7.72380650e-01
-1.54482746e+00 -1.11712694e+00 -3.89885157e-01 1.31600663e-01
-1.70227468e+00 -1.07941175e+00 8.77713263e-01 -1.83357522e-01
4.67222959e-01 9.17154253e-01 3.64492089e-01 1.39239860e+00
-9.90961492e-02 8.30954790e-01 1.07931638e+00 -6.21204674e-01
5.63252047e-02 6.19320333e-01 8.73447880e-02 4.04129863e-01
1.12942398e-01 2.82704204e-01 -7.89095402e-01 -7.36455977e-01
1.65050864e-01 -3.45091462e-01 -5.07785082e-01 -1.15999274e-01
-8.98428619e-01 5.18893123e-01 -2.82433510e-01 1.47318348e-01
-7.82253221e-02 1.41696958e-02 3.19648206e-01 5.46684146e-01
1.60278212e-02 1.07381605e-01 -2.49608919e-01 -1.01062931e-01
-1.01056874e+00 1.42914861e-01 1.38467884e+00 6.21577621e-01
5.15850127e-01 -7.59883821e-02 2.91563898e-01 4.90156233e-01
3.12430978e-01 7.45346129e-01 3.13999861e-01 -6.98392391e-01
4.68455195e-01 -1.76408753e-01 -6.93834499e-02 -9.81891215e-01
6.53725043e-02 -1.75791577e-01 -7.04502285e-01 6.95360005e-02
6.64340258e-01 -7.42960721e-02 -3.65494967e-01 1.88346899e+00
2.54488766e-01 5.07606208e-01 2.84624279e-01 2.53098488e-01
4.51973349e-01 3.74948204e-01 -3.29221606e-01 -2.28476882e-01
1.39242315e+00 -7.53656477e-02 -6.02788508e-01 7.44932815e-02
1.59716725e-01 -9.23241317e-01 7.58486032e-01 8.26063812e-01
-8.17870617e-01 -4.24588203e-01 -1.16973960e+00 3.36820245e-01
-3.26395810e-01 8.12668279e-02 3.44840527e-01 1.53027833e+00
-7.54072070e-01 5.41651845e-01 -7.57175505e-01 4.85414602e-02
4.70461339e-01 9.49368000e-01 -7.61563897e-01 1.54060334e-01
-1.21851671e+00 2.85975605e-01 -1.10673115e-01 1.88276887e-01
-8.72947395e-01 -6.42686784e-01 -7.64189184e-01 5.43040931e-02
2.42818803e-01 -1.15367122e-01 1.19306028e+00 -7.80302525e-01
-1.71816027e+00 7.74856389e-01 -4.39730912e-01 -5.66838861e-01
5.02098203e-01 -7.18262792e-02 -7.18525648e-01 3.01679164e-01
-3.29049319e-01 2.20343485e-01 1.58717334e+00 -1.30338192e+00
-2.98673630e-01 -5.02021253e-01 -3.22534382e-01 -6.23410404e-01
-4.83938187e-01 2.09790215e-01 9.70172808e-02 -5.16251385e-01
9.76723880e-02 -1.10615551e+00 2.72747070e-01 -2.86362439e-01
-9.60996628e-01 4.54179823e-01 9.71538126e-01 -8.47408891e-01
1.15973055e+00 -2.51364708e+00 2.49898229e-02 7.64579296e-01
1.99339781e-02 4.96140093e-01 6.11390695e-02 5.77590764e-01
-1.89087510e-01 3.25931966e-01 -4.46049988e-01 -6.54387534e-01
-6.07020333e-02 5.31090312e-02 -9.36010361e-01 7.09171593e-01
5.41818887e-02 4.59059656e-01 -3.95848125e-01 -2.22885326e-01
-1.74054697e-01 8.70345175e-01 -6.88777387e-01 7.29621500e-02
3.13891947e-01 5.75923204e-01 -1.52148232e-01 5.95356643e-01
1.09068346e+00 3.70658726e-01 3.65019262e-01 -2.80983061e-01
2.86856860e-01 5.04621387e-01 -1.39076614e+00 1.14839244e+00
-4.35209960e-01 7.81279147e-01 5.41296601e-01 -7.94850945e-01
7.95821965e-01 7.44900405e-01 1.78317055e-01 2.03441948e-01
1.77805096e-01 2.87505865e-01 1.09647764e-02 -4.08491343e-01
1.66100323e-01 -1.59628749e-01 -2.86075711e-01 7.44746447e-01
-3.28724226e-03 -7.78798088e-02 -6.08079612e-01 -1.58589616e-01
7.72074044e-01 -2.45280892e-01 2.45212346e-01 -2.15985849e-02
8.58437777e-01 -8.95898581e-01 3.27697426e-01 8.24392796e-01
-1.30540103e-01 4.90455717e-01 5.21654069e-01 4.83603179e-02
-4.65275973e-01 -1.15659904e+00 -3.06379139e-01 7.57302940e-01
-2.90428877e-01 -6.15480959e-01 -1.09062409e+00 -7.71727860e-01
1.42404705e-01 6.23700857e-01 -4.07894969e-01 -3.08612466e-01
-7.33852565e-01 -4.15994167e-01 1.50184035e+00 1.52891964e-01
2.49086246e-01 -6.79191351e-01 -4.53439206e-01 4.99579795e-02
-2.62206674e-01 -1.11804569e+00 -4.11309451e-01 5.26757259e-03
-5.88870883e-01 -1.14080381e+00 -8.46753120e-02 -4.97690409e-01
4.90110070e-01 3.98515500e-02 5.16650259e-01 -9.66641456e-02
-3.38274360e-01 5.49547434e-01 -4.41156290e-02 -6.91789031e-01
-9.99312580e-01 1.76071435e-01 4.29912776e-01 7.72891045e-01
4.25787479e-01 -8.63553107e-01 -1.01380453e-01 3.45244557e-01
-1.16194630e+00 -3.87940317e-01 1.82756975e-01 8.25225353e-01
6.46769255e-02 1.57865644e-01 4.18729007e-01 -9.13222849e-01
4.08761293e-01 -2.23522618e-01 -4.10243303e-01 3.93706173e-01
-1.58632174e-01 2.45963484e-02 5.40307403e-01 -6.85692370e-01
-8.43986034e-01 2.82157242e-01 -4.12580699e-01 -1.84560344e-01
-2.92717367e-01 1.45857930e-01 -7.25669026e-01 -4.87046480e-01
4.84609187e-01 4.50722337e-01 3.76560330e-01 -7.09362566e-01
2.49541640e-01 1.34030747e+00 7.29396343e-01 -6.46452785e-01
9.81418014e-01 6.06880248e-01 2.18223572e-01 -1.27796113e+00
-1.90941900e-01 -1.83676317e-01 -6.27784967e-01 -4.48455401e-02
2.88191915e-01 -4.69227225e-01 -1.21545386e+00 7.33793676e-01
-1.20320725e+00 1.46183103e-01 -4.04388160e-02 4.74118531e-01
-5.41193306e-01 7.72369623e-01 -4.76418585e-01 -1.17032206e+00
-3.22652996e-01 -1.22395778e+00 1.18082869e+00 -1.70613274e-01
-3.51585597e-01 -6.18567884e-01 -1.32306427e-01 3.84679884e-01
2.04014435e-01 1.71055868e-01 5.33894598e-01 -1.05057132e+00
-3.18610579e-01 -7.29313791e-01 3.85034293e-01 5.83304226e-01
2.58273900e-01 7.74816871e-02 -1.88727283e+00 -5.43226361e-01
4.11468446e-01 -7.60563090e-02 6.46800876e-01 -5.83421551e-02
1.20254302e+00 -7.60803998e-01 -3.47402573e-01 7.46880233e-01
8.61166537e-01 3.21396776e-02 7.74845123e-01 -2.27110937e-01
4.43467259e-01 8.58993709e-01 2.27616534e-01 4.88445550e-01
-1.34720892e-01 8.28954041e-01 2.95174181e-01 4.18776065e-01
3.08174282e-01 -3.54151934e-01 7.60293901e-01 6.02202713e-01
1.83234170e-01 -3.15855481e-02 -6.28173888e-01 1.57345906e-01
-1.12248647e+00 -1.08306003e+00 1.56384319e-01 2.55209303e+00
8.64523232e-01 -1.18358850e-01 1.00171916e-01 8.22679698e-01
7.12041855e-01 4.44759848e-03 -2.17875510e-01 -2.55463779e-01
1.40468583e-01 7.33001709e-01 4.16311324e-01 8.28840554e-01
-1.18979633e+00 6.65267527e-01 5.95005560e+00 7.48381913e-01
-1.31661642e+00 2.13105162e-03 2.98783153e-01 -1.30061522e-01
-1.75164878e-01 6.27582595e-02 -9.20382321e-01 3.47287387e-01
1.39455521e+00 -2.09976360e-01 6.45236611e-01 4.46146667e-01
-3.10584426e-01 2.84161210e-01 -1.45451880e+00 9.91111636e-01
3.08149427e-01 -8.40990484e-01 4.95493077e-02 6.45720124e-01
-7.22307041e-02 -5.53475916e-01 4.26922828e-01 -5.23565374e-02
-2.12137371e-01 -1.17092049e+00 6.16479754e-01 4.69224215e-01
9.83609021e-01 -8.92425299e-01 4.93362427e-01 3.97406220e-01
-1.02935326e+00 -1.50709227e-01 -1.08190164e-01 3.21054131e-01
-1.19161401e-02 3.91191989e-01 -1.28859627e+00 6.24723196e-01
3.01416665e-01 1.57693252e-01 -4.31440353e-01 6.59641087e-01
-2.04063114e-02 9.41875577e-01 -7.03531742e-01 2.60513723e-01
-4.73570794e-01 1.00714073e-01 9.18201625e-01 1.18545496e+00
5.32486379e-01 -2.83271700e-01 -3.11812669e-01 6.60838127e-01
-1.13589384e-01 -8.36439282e-02 -8.02599788e-01 -1.09717041e-01
7.02170074e-01 8.90222371e-01 -2.05360323e-01 -7.99423680e-02
5.21818288e-02 1.01737082e+00 -3.19473952e-01 1.55793250e-01
-4.73906606e-01 -6.18006289e-01 9.67914522e-01 4.56659589e-03
2.58680671e-01 -5.57761677e-02 -1.74013346e-01 -1.24211144e+00
1.79620475e-01 -1.27177811e+00 3.09231430e-01 -9.70320851e-02
-8.84499550e-01 6.07115924e-01 -9.60676372e-02 -1.32391381e+00
-7.12999284e-01 -4.36238825e-01 -4.19659883e-01 8.60519528e-01
-1.19110012e+00 -1.17224407e+00 1.82148576e-01 5.97907484e-01
-9.31272432e-02 -3.79398257e-01 1.33802271e+00 9.67614129e-02
-4.33114320e-01 1.16788745e+00 9.06032100e-02 2.96818912e-01
6.07392251e-01 -7.80813396e-01 4.77537543e-01 7.01601624e-01
6.30077064e-01 9.86752331e-01 7.66952157e-01 -2.54454404e-01
-1.61887753e+00 -6.76420093e-01 1.02225006e+00 -3.97281945e-01
3.00880611e-01 -8.05078268e-01 -1.01989686e+00 8.20687294e-01
-3.88784334e-02 -1.83422327e-01 1.50860643e+00 -2.76944757e-01
-7.28117764e-01 -3.20759535e-01 -1.57560825e+00 2.73335814e-01
5.29571176e-01 -1.13431668e+00 -4.53600675e-01 -1.77446548e-02
3.86808068e-01 -8.27827901e-02 -8.76470923e-01 1.63538620e-01
1.13938069e+00 -9.75480556e-01 1.25787711e+00 -5.95342338e-01
-1.75873280e-01 -2.42736340e-01 -4.58596230e-01 -8.52456987e-01
5.30934334e-01 -1.20476663e+00 -3.67087960e-01 1.68505502e+00
2.76625246e-01 -9.64543521e-01 7.96370387e-01 5.47316313e-01
5.84644735e-01 -1.06692031e-01 -1.42753911e+00 -7.26383567e-01
-8.15046877e-02 -8.06213260e-01 1.12527621e+00 6.77940011e-01
1.54769734e-01 -2.84588188e-01 -6.15337193e-01 7.01124787e-01
9.56050515e-01 -1.28374770e-01 1.11181402e+00 -1.32424939e+00
-5.99426150e-01 -5.19709103e-02 -5.81417859e-01 -7.86378205e-01
4.54685897e-01 -8.97285521e-01 -1.39228523e-01 -1.45193845e-01
-1.96110114e-01 -3.42709541e-01 -2.48531163e-01 2.52369255e-01
2.52725035e-01 4.90816355e-01 3.14297944e-01 1.07610948e-01
3.03095967e-01 1.13491327e-01 4.26341146e-01 -1.05316006e-01
-2.25782558e-01 7.20452964e-01 -7.70348132e-01 4.64068681e-01
9.35546875e-01 -7.24331617e-01 -3.79049242e-01 2.17633009e-01
-1.66049097e-02 1.79394513e-01 6.11736655e-01 -1.11932659e+00
1.79732203e-01 1.50515854e-01 1.30102307e-01 -2.41321042e-01
5.78360677e-01 -1.08631623e+00 4.25652981e-01 5.94475389e-01
-4.35034215e-01 -6.29846334e-01 3.05446893e-01 5.45718491e-01
-2.21375272e-01 -4.12430435e-01 6.16449535e-01 2.71567881e-01
2.18940958e-01 2.19208617e-02 -4.37438965e-01 -3.91911924e-01
7.74566054e-01 -1.66937485e-01 2.83143297e-02 -5.22149563e-01
-8.31483006e-01 -4.76129115e-01 3.32785517e-01 1.01184532e-01
6.38609946e-01 -1.02339041e+00 -6.16466939e-01 1.04309571e+00
-1.31669074e-01 -4.10982132e-01 1.17617026e-01 6.83744669e-01
-1.98624268e-01 2.60579467e-01 6.07097382e-03 -6.32752955e-01
-2.05389667e+00 4.17861760e-01 2.01091409e-01 7.58222193e-02
-4.04010981e-01 8.22204173e-01 5.13919890e-02 -3.85803640e-01
2.96961993e-01 -9.11160707e-02 9.45102647e-02 2.37219602e-01
1.00489068e+00 2.92316258e-01 2.73955613e-01 -7.72086024e-01
-5.38749337e-01 4.31981713e-01 -1.60368919e-01 -6.52930796e-01
1.08022285e+00 -3.52628641e-02 -1.96113899e-01 2.33315498e-01
1.67959857e+00 7.32518971e-01 -9.59956110e-01 -1.82377383e-01
-3.19721885e-02 -8.60698760e-01 -3.66609365e-01 -3.17883819e-01
-8.76454532e-01 1.18608248e+00 4.06666994e-01 4.18161392e-01
1.18810034e+00 -2.30830029e-01 8.06734681e-01 3.39779288e-01
5.53231120e-01 -5.17026901e-01 -4.33305502e-01 2.59992983e-02
7.81251669e-01 -5.39884567e-01 -2.18674406e-01 -6.46696091e-01
-3.81372422e-01 1.33793497e+00 -2.89981008e-01 2.43844390e-01
9.50689971e-01 5.11552155e-01 5.27952127e-02 2.59012818e-01
-4.90329117e-01 4.43826705e-01 2.48409703e-01 9.54051852e-01
5.23425173e-03 1.65235475e-01 2.53862530e-01 8.59494269e-01
-7.47978449e-01 -3.93918194e-02 6.66044772e-01 1.04834032e+00
-4.56858128e-02 -1.60633874e+00 -8.61544013e-01 2.05556750e-01
-7.75484979e-01 -8.49285070e-03 -6.61581337e-01 3.69242430e-01
-1.71039596e-01 1.15124607e+00 -1.93052515e-01 -7.22661972e-01
1.19026169e-01 6.54473901e-01 5.15224516e-01 -2.69368261e-01
-6.82277262e-01 -1.08377792e-01 1.16934955e-01 -3.62141341e-01
-2.62655675e-01 -9.90182340e-01 -8.21125567e-01 -5.82048416e-01
-2.96764910e-01 2.06153691e-01 9.92560446e-01 7.46494114e-01
4.34487760e-01 4.29147668e-02 1.10948253e+00 -9.68244612e-01
-9.24792409e-01 -7.64736176e-01 -8.17694783e-01 1.48892611e-01
8.87169421e-01 -3.53341967e-01 -7.72692144e-01 1.89767897e-01] | [13.988228797912598, 5.842278957366943] |
94f8a0dc-bb64-47b0-92cc-59b45ad671a9 | from-the-detection-of-toxic-spans-in-online-1 | null | null | https://aclanthology.org/2022.acl-long.259 | https://aclanthology.org/2022.acl-long.259.pdf | From the Detection of Toxic Spans in Online Discussions to the Analysis of Toxic-to-Civil Transfer | We study the task of toxic spans detection, which concerns the detection of the spans that make a text toxic, when detecting such spans is possible. We introduce a dataset for this task, ToxicSpans, which we release publicly. By experimenting with several methods, we show that sequence labeling models perform best, but methods that add generic rationale extraction mechanisms on top of classifiers trained to predict if a post is toxic or not are also surprisingly promising. Finally, we use ToxicSpans and systems trained on it, to provide further analysis of state-of-the-art toxic to non-toxic transfer systems, as well as of human performance on that latter task. Our work highlights challenges in finer toxicity detection and mitigation. | ['Ion Androutsopoulos', 'Jeffrey Sorensen', 'Alexandros Xenos', 'Leo Laugier', 'John Pavlopoulos'] | null | null | null | null | acl-2022-5 | ['toxic-spans-detection'] | ['natural-language-processing'] | [ 5.01375854e-01 -1.00666150e-01 -9.70639661e-02 2.61684597e-01
-1.26515377e+00 -1.09673822e+00 6.59580410e-01 6.04148984e-01
-3.40303153e-01 9.98862922e-01 6.38816178e-01 -4.33224142e-01
-1.99782886e-02 -6.55829430e-01 -8.72187376e-01 -5.10463595e-01
-4.27874029e-02 2.78171122e-01 5.01584709e-01 -1.56366169e-01
4.86415595e-01 4.75214183e-01 -1.04253161e+00 8.77331436e-01
5.64369261e-01 6.62653565e-01 -2.33839065e-01 6.75203502e-01
2.06334695e-01 9.29040551e-01 -8.69682252e-01 -5.72990060e-01
4.55776043e-02 -2.79058576e-01 -1.16915679e+00 -4.10850585e-01
6.84882641e-01 6.45510852e-02 -1.91969782e-01 7.81739771e-01
7.97386467e-01 -1.88183725e-01 1.20862532e+00 -1.22916222e+00
-7.27136016e-01 1.15623260e+00 -3.83597642e-01 5.34644783e-01
7.38257110e-01 6.32510126e-01 1.16163492e+00 -7.03570724e-01
8.53676558e-01 1.07758594e+00 1.18086958e+00 5.85641503e-01
-1.09680390e+00 -7.07206070e-01 -7.75083378e-02 2.95056969e-01
-1.06726408e+00 -3.85627925e-01 1.95693344e-01 -9.09319103e-01
1.23770940e+00 3.84313047e-01 3.08543921e-01 1.76213634e+00
4.62051183e-01 7.75668383e-01 1.24429786e+00 -1.80215493e-01
1.08446635e-01 -2.91572064e-01 3.56392175e-01 5.61748981e-01
1.46620169e-01 3.21152690e-03 -9.69270289e-01 -4.86007273e-01
-3.58574450e-01 -3.16758990e-01 -4.45635229e-01 5.60211837e-01
-7.90042818e-01 7.08155274e-01 2.61806637e-01 2.34822884e-01
-4.39293444e-01 2.85466701e-01 8.62009346e-01 5.61873370e-04
7.71041632e-01 7.12192893e-01 -6.63927197e-01 -2.04779357e-02
-9.38181221e-01 4.19194013e-01 8.77710104e-01 1.07444215e+00
1.77378446e-01 -4.69781548e-01 -8.60384405e-01 5.08881390e-01
-5.12814894e-02 5.44903427e-02 1.48287266e-01 -2.89763838e-01
7.28996098e-01 4.09809321e-01 9.49818045e-02 -1.90150574e-01
-7.07415581e-01 -1.54486150e-01 -3.86171676e-02 -1.20994434e-01
3.52626264e-01 -2.25215107e-01 -8.96759152e-01 1.52906561e+00
8.53107274e-02 6.38316497e-02 -1.90840334e-01 2.17150629e-01
7.42093205e-01 3.41568202e-01 8.33774090e-01 -1.79920256e-01
1.63494146e+00 -6.27355039e-01 -3.58921200e-01 2.70897169e-02
1.04360914e+00 -7.73094058e-01 1.15289986e+00 6.07892156e-01
-8.69942546e-01 2.26607576e-01 -1.13303554e+00 2.51740422e-02
-9.03004050e-01 -4.87106323e-01 2.90824413e-01 8.59636188e-01
-7.46558785e-01 1.22856605e+00 -1.47881284e-01 -5.89607954e-01
6.32492363e-01 3.03815631e-03 -2.43781984e-01 1.12349940e-02
-1.51347220e+00 1.34596789e+00 3.41352612e-01 -6.62237167e-01
-1.32511926e+00 -1.27393806e+00 -6.11583769e-01 -1.99412048e-01
1.63379297e-01 -5.29522181e-01 1.55719221e+00 -1.28843963e-01
-6.21583700e-01 8.59367728e-01 4.16437030e-01 -5.31624258e-01
6.01829827e-01 -5.54912806e-01 -3.09105903e-01 2.17963681e-01
3.52924138e-01 2.99414873e-01 4.65512842e-01 -7.01014876e-01
-4.89922941e-01 -1.84183300e-01 -4.47779931e-02 -1.29004836e-01
-6.92226589e-01 7.44275451e-01 1.14325903e-01 -6.29653215e-01
-8.62793148e-01 -8.72113705e-01 -4.08064686e-02 -3.47463936e-01
-1.20903432e+00 -6.73986137e-01 7.76485622e-01 -8.71329844e-01
1.38519156e+00 -1.52572942e+00 -2.74820566e-01 -1.08894922e-01
3.92252058e-02 4.10548039e-02 -3.54055941e-01 1.00815940e+00
-3.36239696e-01 7.03678727e-01 -3.71642858e-01 -1.73066184e-01
1.66037470e-01 -5.27455986e-01 -6.13939404e-01 5.88319123e-01
3.05682391e-01 8.29853952e-01 -1.01430392e+00 -4.73859489e-01
-3.89164656e-01 6.86846226e-02 -1.28615901e-01 1.27318487e-01
-7.39938438e-01 1.94864690e-01 -4.59521979e-01 8.52420866e-01
3.59490752e-01 -9.31818038e-03 -1.58111557e-01 -6.38866276e-02
-2.10067779e-01 9.31916356e-01 -3.49462658e-01 1.36173785e+00
-2.69141883e-01 3.45943779e-01 -4.53474909e-01 -3.38987201e-01
2.92675674e-01 5.69549024e-01 4.83281285e-01 -4.60370421e-01
3.96092385e-02 1.08963564e-01 -3.49661469e-01 -7.64766037e-01
4.10521775e-01 -3.15727770e-01 -3.60682964e-01 6.98711157e-01
-9.83113050e-02 1.16696082e-01 6.23228312e-01 3.64711910e-01
2.13495278e+00 3.36381793e-01 2.75472760e-01 -3.16068262e-01
2.29101524e-01 2.69753218e-01 3.06700051e-01 5.00844777e-01
-9.20384973e-02 5.92028201e-01 6.60809636e-01 -8.39653704e-03
-9.66789126e-01 -7.69443750e-01 -1.41314372e-01 1.37826884e+00
-4.55073148e-01 -8.13421547e-01 -1.08858335e+00 -1.45902967e+00
1.21717371e-01 9.58597243e-01 -9.40939069e-01 -1.64369121e-01
-3.85795057e-01 -9.75018024e-01 1.35632515e+00 7.52664030e-01
-5.33594713e-02 -1.38146448e+00 -5.85012376e-01 3.53432596e-01
-2.02106029e-01 -7.64682412e-01 -7.70072222e-01 8.26241434e-01
-3.74147743e-01 -1.55517793e+00 -4.42667544e-01 -2.07480907e-01
-9.40474719e-02 -9.39970687e-02 1.33391798e+00 2.65994608e-01
-5.56495070e-01 1.24961302e-01 -4.96995836e-01 -7.35670924e-01
-5.18573821e-01 2.31041119e-01 -1.52371764e-01 -6.67250633e-01
4.82066065e-01 -5.74050069e-01 -4.94876653e-01 1.94824189e-01
-1.08970857e+00 -4.64126378e-01 3.25237662e-01 3.35391998e-01
1.46370709e-01 -2.94381320e-01 8.59410822e-01 -1.45323241e+00
8.88899684e-01 -9.27406073e-01 4.41176438e-04 3.18845153e-01
-5.87667763e-01 -9.83112976e-02 8.73904705e-01 -3.01579922e-01
-7.53021896e-01 2.22034872e-01 -4.98760641e-01 -1.18420146e-01
-2.23214298e-01 6.02880478e-01 -3.10925633e-01 2.68710047e-01
1.38562953e+00 -9.14090499e-02 -7.29172528e-01 -6.92608476e-01
5.92190802e-01 7.77311862e-01 2.73706764e-01 -5.87634563e-01
7.85926521e-01 8.86216089e-02 1.60124470e-02 -5.77455521e-01
-1.18847609e+00 -7.12664962e-01 -6.25312507e-01 -1.89844146e-02
8.82345855e-01 -7.63424039e-01 -5.82370877e-01 2.86314815e-01
-1.48561728e+00 -5.24756372e-01 -8.23425949e-02 -2.68043518e-01
-5.93167424e-01 4.30140674e-01 -1.13245440e+00 -6.37064695e-01
-8.42649698e-01 -6.01518095e-01 9.95238125e-01 -3.46443594e-01
-9.31886256e-01 -8.90895903e-01 4.39726830e-01 2.25394368e-01
2.57038735e-02 6.43595338e-01 1.31968832e+00 -1.05713725e+00
-1.41341791e-01 -2.19667684e-02 7.62895048e-02 -1.22136779e-01
-1.33869380e-01 -1.25634726e-02 -1.22285235e+00 -4.06802678e-03
-4.12234515e-01 -7.60542691e-01 1.27036643e+00 -3.75395305e-02
1.01500046e+00 -6.22569144e-01 -6.50386214e-01 1.12099037e-01
1.26956129e+00 2.23389622e-02 8.95348549e-01 5.92240274e-01
5.90339065e-01 8.57422948e-01 6.69068038e-01 4.70230728e-01
-5.25701940e-02 4.31087434e-01 5.23547053e-01 2.33792081e-01
-4.17725146e-01 -5.08826911e-01 7.88377941e-01 1.25235707e-01
3.60667825e-01 -7.76218653e-01 -8.85414004e-01 7.64543831e-01
-1.68079507e+00 -1.10066152e+00 -5.87815404e-01 1.64686978e+00
1.33723176e+00 1.11777425e-01 5.75062573e-01 3.87700379e-01
6.43674970e-01 -1.16233779e-02 -4.95996237e-01 -5.71924925e-01
-3.70616093e-02 2.93764919e-01 6.95582807e-01 -2.77478341e-02
-1.44934678e+00 9.93370175e-01 7.92622852e+00 1.31187344e+00
-6.90304518e-01 2.92387158e-01 4.33714360e-01 -2.93405838e-02
-4.93762970e-01 -2.72249579e-01 -9.48068678e-01 6.67360306e-01
1.39673829e+00 -9.65927988e-02 9.73592401e-02 7.22308934e-01
2.29344100e-01 2.71021388e-02 -1.65239727e+00 2.11976051e-01
1.46622986e-01 -1.23796606e+00 1.91781402e-01 9.27484874e-03
4.76153761e-01 7.40371719e-02 5.46513200e-02 1.79652512e-01
8.02892208e-01 -1.26830161e+00 1.17122042e+00 2.64062792e-01
8.30972850e-01 -7.88085222e-01 4.03501391e-01 3.17533344e-01
-9.81845617e-01 -1.32925004e-01 -2.06231728e-01 7.59188905e-02
-6.93522319e-02 8.11590791e-01 -1.00243592e+00 4.23960894e-01
7.75675714e-01 1.01483881e+00 -9.49532509e-01 1.31795430e+00
-7.53622472e-01 9.28781688e-01 -7.08363159e-03 -1.64061591e-01
8.00251663e-02 3.85861546e-01 4.63282973e-01 1.86036432e+00
3.94832969e-01 9.91645679e-02 9.41705331e-02 8.41106474e-01
-4.77560967e-01 2.82395691e-01 -1.03503895e+00 -2.42446691e-01
3.93302083e-01 1.36187804e+00 -7.25285649e-01 -3.97709638e-01
-1.07224472e-01 8.63493502e-01 4.74409342e-01 -9.70231295e-02
-8.71177733e-01 -3.89344931e-01 4.95104373e-01 1.72237873e-01
9.01088342e-02 3.54931027e-01 -7.50851035e-01 -4.89995658e-01
-1.95353210e-01 -8.23777854e-01 1.01662517e+00 -8.95560145e-01
-1.83870459e+00 4.58144814e-01 -1.39041319e-01 -1.29281151e+00
1.85234815e-01 -6.31924748e-01 -9.73831594e-01 7.67400086e-01
-1.24660337e+00 -1.24920321e+00 1.63228989e-01 3.76166880e-01
5.54591060e-01 3.05196226e-01 7.79102564e-01 1.13682158e-01
-4.66739804e-01 5.63936293e-01 -2.51188278e-01 -1.99648455e-01
1.12036884e+00 -1.33177125e+00 6.01828396e-01 7.21778750e-01
-1.22075461e-01 6.18029118e-01 9.36708033e-01 -1.27902436e+00
-9.72501874e-01 -1.60707641e+00 1.05570316e+00 -1.19776285e+00
1.00513768e+00 -4.44119453e-01 -7.77145088e-01 4.72070247e-01
4.99583870e-01 -6.64807379e-01 9.59050715e-01 1.33291602e-01
-9.43085551e-01 7.97320127e-01 -1.23225582e+00 6.83053553e-01
1.52633417e+00 -5.85539222e-01 -6.93801224e-01 1.00281620e+00
8.29652190e-01 -2.80625615e-02 -7.49602854e-01 8.27086270e-02
2.75039941e-01 -6.20475352e-01 7.71500707e-01 -9.23665106e-01
1.07128215e+00 -1.24373265e-01 6.56947419e-02 -1.55145633e+00
-4.63768691e-01 -6.80971682e-01 -8.62728222e-04 1.37565720e+00
9.23178494e-01 -8.27710852e-02 7.22302139e-01 3.72042835e-01
-5.03961146e-01 -6.99901223e-01 -7.06664503e-01 -1.17961228e+00
5.91602266e-01 -4.50820982e-01 4.83078003e-01 8.98637354e-01
5.91981649e-01 6.88499689e-01 -3.05817157e-01 -1.54006124e-01
5.82167506e-01 -2.77892798e-01 1.38768286e-01 -1.10005581e+00
-1.13928244e-01 -3.43210548e-01 -1.84041113e-02 -3.03127617e-01
3.44042599e-01 -1.19521904e+00 3.96041334e-01 -1.62327361e+00
7.97739625e-01 -1.26755796e-02 -4.08981144e-01 1.13675344e+00
-4.22387332e-01 8.04294646e-01 1.02913998e-01 1.62913680e-01
-5.97362459e-01 2.63610005e-01 5.68548203e-01 -6.02926910e-01
1.11319358e-02 -3.42755109e-01 -1.06200778e+00 6.45357966e-01
8.84888947e-01 -9.73131597e-01 -5.24352863e-02 7.62073696e-03
3.26098412e-01 -4.61671531e-01 -3.34171057e-02 -8.41116071e-01
6.05375916e-02 -2.33242840e-01 4.06226993e-01 -8.26688945e-01
-9.28524658e-02 -2.84406841e-01 3.58869843e-02 5.39471984e-01
-6.42472386e-01 5.93081303e-02 2.77816623e-01 6.40119374e-01
5.70881128e-01 -5.98978758e-01 6.77005768e-01 -2.27837861e-01
-3.01608235e-01 3.58454704e-01 -8.42832565e-01 3.44319791e-01
9.71218765e-01 1.94245115e-01 -8.97702754e-01 -7.23908022e-02
-3.81437510e-01 1.14677295e-01 5.47140479e-01 2.87256330e-01
3.78425866e-01 -1.03700733e+00 -8.54941666e-01 -9.01096940e-01
5.68608701e-01 -8.52268636e-01 -1.05494365e-01 8.36165011e-01
-2.45256871e-01 3.29862386e-01 -8.64117593e-03 1.17490716e-01
-1.54417944e+00 1.01823568e+00 1.84164068e-03 -6.16815567e-01
-6.36921525e-01 7.79068351e-01 2.07298934e-01 -1.73263580e-01
1.87539563e-01 1.29764587e-01 -4.93938506e-01 4.23155785e-01
8.08124244e-01 6.54010713e-01 5.72022140e-01 -4.05274004e-01
-6.45070255e-01 1.77055076e-01 -5.56370653e-02 -3.91163304e-02
1.35646665e+00 4.34889913e-01 -1.51369601e-01 1.70785934e-01
1.09082031e+00 1.45683214e-01 -8.96784723e-01 4.45233583e-02
6.91756904e-01 -6.06655106e-02 -4.41630065e-01 -1.48789656e+00
-4.21582311e-01 9.05091166e-01 1.77138001e-01 1.57748476e-01
7.94454217e-01 7.43959041e-04 9.85087037e-01 2.31928676e-01
1.41806290e-01 -1.06031299e+00 3.88143748e-01 7.75149167e-01
1.20000994e+00 -4.34058636e-01 2.67549396e-01 -7.49277294e-01
-4.61327583e-01 9.79277074e-01 3.59004706e-01 1.56143561e-01
2.85448551e-01 4.63694155e-01 -1.49240822e-01 -4.01722252e-01
-1.08503973e+00 -3.93224329e-01 -7.55063817e-03 8.74644578e-01
6.43077195e-01 -1.56950671e-02 -6.35926187e-01 7.41127849e-01
-2.07146555e-01 -1.40547276e-01 7.09298313e-01 1.08206975e+00
-5.26950598e-01 -1.11654794e+00 -4.73889142e-01 5.72739601e-01
-1.06133568e+00 -6.35843337e-01 -1.23170030e+00 3.95490974e-01
2.60776252e-01 1.15124822e+00 -7.19273269e-01 -6.24947965e-01
6.44176006e-01 5.98226130e-01 4.87755299e-01 -1.08780754e+00
-1.39298201e+00 -5.43773808e-02 8.67927611e-01 -3.61856997e-01
1.64474785e-01 -6.26427352e-01 -1.21633101e+00 -2.11573750e-01
-2.83932924e-01 2.41369829e-01 3.81170869e-01 8.62894356e-01
2.91470364e-02 4.87017125e-01 3.24529976e-01 -5.40028632e-01
-7.36677170e-01 -1.18059015e+00 -8.07126880e-01 6.25596881e-01
-1.33068457e-01 -3.40761483e-01 -2.95491278e-01 3.95504028e-01] | [8.968360900878906, 10.620347023010254] |
f6ec61bd-c446-4607-8954-7b858f965de2 | no-regret-constrained-bayesian-optimization | 2305.03824 | null | https://arxiv.org/abs/2305.03824v1 | https://arxiv.org/pdf/2305.03824v1.pdf | No-Regret Constrained Bayesian Optimization of Noisy and Expensive Hybrid Models using Differentiable Quantile Function Approximations | This paper investigates the problem of efficient constrained global optimization of composite functions (hybrid models) whose input is an expensive black-box function with vector-valued outputs and noisy observations, which often arises in real-world science, engineering, manufacturing, and control applications. We propose a novel algorithm, Constrained Upper Quantile Bound (CUQB), to solve such problems that directly exploits the composite structure of the objective and constraint functions that we show leads substantially improved sampling efficiency. CUQB is conceptually simple and avoids the constraint approximations used by previous methods. Although the CUQB acquisition function is not available in closed form, we propose a novel differentiable stochastic approximation that enables it to be efficiently maximized. We further derive bounds on the cumulative regret and constraint violation. Since these bounds depend sublinearly on the number of iterations under some regularity assumptions, we establish explicit bounds on the convergence rate to the optimal solution of the original constrained problem. In contrast to existing methods, CUQB further incorporates a simple infeasibility detection scheme, which we prove triggers in a finite number of iterations (with high probability) when the original problem is infeasible. Numerical experiments on several test problems, including environmental model calibration and real-time reactor optimization, show that CUQB significantly outperforms traditional Bayesian optimization in both constrained and unconstrained cases. Furthermore, compared to other state-of-the-art methods that exploit composite structure, CUQB achieves competitive empirical performance while also providing substantially improved theoretical guarantees. | ['Joel A. Paulson', 'Congwen Lu'] | 2023-05-05 | null | null | null | null | ['bayesian-optimization'] | ['methodology'] | [ 2.42216155e-01 8.33609328e-02 -2.57460088e-01 -1.06134355e-01
-1.16829407e+00 -6.29310906e-01 1.31133124e-01 2.42773309e-01
-2.81321377e-01 1.19648850e+00 -3.32687467e-01 -5.18333614e-01
-8.36985230e-01 -6.75392151e-01 -9.97070014e-01 -1.02010858e+00
-9.05569941e-02 3.87022108e-01 -2.52194136e-01 3.72723630e-03
2.49152511e-01 3.31941038e-01 -8.78313720e-01 -6.00091338e-01
9.20211375e-01 1.41302145e+00 2.02481136e-01 4.98928130e-01
4.06802922e-01 2.42348522e-01 -3.31409544e-01 -6.63767830e-02
5.19474864e-01 -3.16129088e-01 -4.05735910e-01 3.11044574e-01
-1.73117727e-01 -8.12815055e-02 2.20445961e-01 1.43160605e+00
4.66888130e-01 5.20071805e-01 5.61734200e-01 -1.11338103e+00
-3.23055774e-01 4.65790212e-01 -6.69936538e-01 -4.95491549e-02
-6.80865161e-03 2.71500442e-02 9.01542246e-01 -8.43038678e-01
1.17176749e-01 1.12621474e+00 6.38904393e-01 2.29893327e-01
-1.49731791e+00 -5.43750405e-01 4.53545749e-01 -5.48432887e-01
-1.34893775e+00 -2.02600047e-01 4.46808130e-01 -4.84425306e-01
6.53487682e-01 3.28904182e-01 2.30300575e-01 6.83421671e-01
1.61806986e-01 6.62429929e-01 9.20251608e-01 -2.90147007e-01
5.32496691e-01 9.54545289e-02 1.78310648e-01 6.90324843e-01
4.77936625e-01 4.86622691e-01 -2.05600590e-01 -2.92229980e-01
7.25043476e-01 -4.55786996e-02 -6.09444618e-01 -4.86343473e-01
-1.15060115e+00 1.06700230e+00 1.75611168e-01 -1.89140603e-01
-5.02916813e-01 3.61161917e-01 1.60889626e-01 2.51674592e-01
5.49223244e-01 6.53183520e-01 -7.21563935e-01 6.93679526e-02
-6.14622295e-01 3.92721295e-01 1.05572474e+00 1.30060041e+00
4.05358493e-01 1.59265310e-01 -3.93470496e-01 6.93242133e-01
3.65023345e-01 7.73173153e-01 -6.78966716e-02 -1.00127947e+00
6.94587529e-01 9.05611813e-02 8.23424280e-01 -7.41868794e-01
-2.86599189e-01 -7.63769746e-01 -6.80232882e-01 9.42463279e-02
6.81529343e-01 -4.68058974e-01 -5.69276452e-01 1.58932471e+00
4.70796108e-01 -6.29215240e-02 6.06280416e-02 9.63592112e-01
-1.87807828e-01 8.83807302e-01 -3.99797350e-01 -7.95002103e-01
9.75811303e-01 -7.09601700e-01 -9.04793382e-01 -5.55528626e-02
2.80897468e-01 -6.24456763e-01 8.52755547e-01 6.01066530e-01
-1.08724689e+00 -9.58491955e-03 -1.13282740e+00 4.71297801e-01
-6.13269620e-02 -8.42734948e-02 5.86388886e-01 9.53930616e-01
-3.74381542e-01 6.86238170e-01 -7.34729886e-01 1.52507573e-01
2.14979202e-01 5.00863254e-01 8.85938704e-02 -3.02633010e-02
-8.21331143e-01 7.02129900e-01 4.25254375e-01 5.63819587e-01
-9.92181957e-01 -9.30394530e-01 -8.93795311e-01 2.43267342e-01
1.27252936e+00 -6.72881484e-01 1.45971572e+00 -4.75074142e-01
-1.98518598e+00 -9.23218504e-02 -1.09995209e-01 -2.72175044e-01
7.43207276e-01 -3.23233128e-01 1.14563629e-01 -1.59189090e-01
5.03951963e-03 -3.64514947e-01 6.87004030e-01 -1.07240963e+00
-5.26429057e-01 -2.74129719e-01 7.50387311e-02 -1.10427365e-01
-3.87121066e-02 -8.32175836e-02 -2.44100317e-01 -5.84807396e-01
5.10834204e-03 -6.65719509e-01 -8.81344438e-01 -1.24010947e-02
-5.64868152e-01 -1.62008256e-01 2.61939824e-01 -3.82461041e-01
1.07254827e+00 -1.95256042e+00 2.23442718e-01 6.64832830e-01
-3.47756684e-01 -1.67752117e-01 1.01367831e-01 5.06581664e-01
8.95209014e-02 1.06249139e-01 -6.16189420e-01 -3.64883780e-01
4.43196058e-01 3.14088404e-01 -3.43109280e-01 8.81547987e-01
2.81336337e-01 6.70407176e-01 -1.18518841e+00 3.63785997e-02
2.57734448e-01 -9.89653170e-02 -6.68360353e-01 3.67174186e-02
-3.20520848e-01 2.72220403e-01 -6.39871180e-01 8.19533050e-01
6.08521104e-01 -3.56548101e-01 3.44345242e-01 5.38170487e-02
-1.26642078e-01 -1.48360014e-01 -1.52849388e+00 1.35676491e+00
-7.45793521e-01 8.37973952e-02 4.91218477e-01 -1.54216254e+00
7.34627545e-01 1.93554878e-01 4.45900321e-01 -2.10689589e-01
4.14332151e-01 2.95822799e-01 -4.68382955e-01 -2.96748906e-01
3.13747555e-01 -5.33793211e-01 -3.66820335e-01 -2.08081398e-02
-2.18938872e-01 -1.50239974e-01 2.17587367e-01 -2.60199845e-01
8.14510167e-01 1.05164647e-01 5.66219926e-01 -7.28460073e-01
4.67723846e-01 -2.54416436e-01 9.36729550e-01 9.64744151e-01
-2.80853733e-02 3.09197783e-01 7.94768751e-01 1.47643685e-01
-7.48406410e-01 -9.93260980e-01 -2.93798476e-01 7.01433659e-01
3.02615106e-01 -7.31660202e-02 -2.86526442e-01 -5.26433349e-01
5.77295125e-01 7.22381473e-01 -5.44778228e-01 5.26769236e-02
-2.36758769e-01 -1.20782042e+00 -1.72141880e-01 5.82828939e-01
1.53207034e-01 -1.65184766e-01 -2.69879371e-01 6.34405315e-01
2.90810883e-01 -1.15634990e+00 -6.60174072e-01 4.76386458e-01
-7.53396034e-01 -1.09787381e+00 -7.33936906e-01 -2.16368362e-01
7.80772209e-01 -2.84333676e-02 7.99365342e-01 -5.20015836e-01
-1.15464985e-01 3.48802567e-01 -2.16972772e-02 -6.11884534e-01
-1.03595786e-01 -2.51734853e-01 1.02454454e-01 2.56257951e-01
-2.87050188e-01 -2.67961025e-01 -3.27670991e-01 4.55789924e-01
-7.19634950e-01 -5.02113163e-01 5.04375398e-01 1.23874605e+00
9.30164456e-01 2.18176916e-01 9.30118382e-01 -9.38220859e-01
5.65906584e-01 -6.81544721e-01 -1.67647612e+00 3.14920336e-01
-8.59044313e-01 2.15224192e-01 7.12887228e-01 -4.67348754e-01
-9.09472287e-01 2.97946870e-01 2.89717019e-01 -5.34832597e-01
5.16183317e-01 9.13680494e-01 -2.50230670e-01 -2.59963535e-02
2.33392954e-01 3.19036767e-02 1.84565797e-01 -5.55226684e-01
1.70881137e-01 4.61188793e-01 4.82598126e-01 -1.08785021e+00
8.23903084e-01 1.96452290e-01 4.61477757e-01 -7.25830078e-01
-1.00113261e+00 -5.61572969e-01 1.48462960e-02 -1.63720578e-01
4.18314099e-01 -7.09963560e-01 -1.35822952e+00 -9.50249135e-02
-7.82974303e-01 -3.75366330e-01 -4.45417970e-01 6.97251379e-01
-8.33488226e-01 4.15382504e-01 -1.58841804e-01 -1.69762158e+00
2.75002494e-02 -1.06939828e+00 1.01309586e+00 4.80805859e-02
1.82595491e-01 -9.69236016e-01 -2.17715442e-01 6.32201061e-02
2.54020214e-01 8.06065142e-01 6.28206909e-01 -2.35790148e-01
-7.15335727e-01 -3.55317533e-01 -2.61617973e-02 5.26981711e-01
9.99607220e-02 -1.74866572e-01 -3.22038025e-01 -5.64521968e-01
1.40925080e-01 -1.57629728e-01 7.30368078e-01 7.17110872e-01
1.15114403e+00 -4.72285300e-01 -3.55263650e-01 7.04171956e-01
1.70343304e+00 2.85858572e-01 1.14276215e-01 1.95700899e-01
1.31507546e-01 4.64403152e-01 9.10293162e-01 9.49163198e-01
-2.00053751e-01 4.95338738e-01 6.36076927e-01 1.69523388e-01
7.65171349e-01 1.17801011e-01 3.97846371e-01 5.34639835e-01
1.56637579e-02 -2.60805964e-01 -3.54070008e-01 5.79247117e-01
-2.20746493e+00 -6.83800757e-01 -5.97867705e-02 2.66452026e+00
8.22408438e-01 4.09980305e-02 3.71753797e-02 1.36616245e-01
6.75872266e-01 -2.35724658e-01 -8.30180049e-01 -4.02837515e-01
-1.95242893e-02 2.99062788e-01 1.23643291e+00 7.29835510e-01
-1.06920457e+00 2.42024943e-01 6.76235962e+00 7.96228349e-01
-5.41987002e-01 1.44215375e-01 5.95435083e-01 -3.17326665e-01
-9.55070704e-02 -9.27591622e-02 -8.57394457e-01 5.03808081e-01
8.02210331e-01 -5.32758296e-01 7.31641471e-01 1.05678260e+00
4.62119013e-01 -2.13017896e-01 -1.07013643e+00 6.97573125e-01
-4.47488844e-01 -1.10477591e+00 -6.87583327e-01 3.91023159e-01
1.02426922e+00 -3.25941443e-01 1.02694079e-01 2.62894899e-01
6.34871960e-01 -8.87037098e-01 8.61352980e-01 5.16066134e-01
6.43324852e-01 -1.11241949e+00 8.46396327e-01 4.37955439e-01
-9.97468054e-01 -6.45588040e-01 -5.30829370e-01 -9.51038823e-02
4.90844578e-01 1.04397237e+00 -4.08648193e-01 1.09600186e+00
2.74439573e-01 3.07267398e-01 3.30972731e-01 1.16504931e+00
-1.51140586e-01 4.91676956e-01 -7.71644413e-01 -3.83534938e-01
3.27372253e-01 -4.90893334e-01 7.16409028e-01 9.78368819e-01
3.98933828e-01 1.48648977e-01 4.59260315e-01 9.53659654e-01
1.52611315e-01 6.85882792e-02 -3.94985616e-01 -6.97764903e-02
5.20512760e-01 8.90181780e-01 -4.54425097e-01 -7.28457943e-02
-4.49378043e-01 5.11728525e-01 1.57239735e-02 6.51690006e-01
-9.51092303e-01 -5.45690298e-01 7.21374214e-01 -3.96468639e-01
6.74939752e-01 -3.88459653e-01 -2.91354418e-01 -8.77692163e-01
2.37825543e-01 -5.70320725e-01 4.62826073e-01 -1.58053830e-01
-1.46898246e+00 -1.76018439e-02 1.58690736e-01 -9.76158321e-01
-3.99923354e-01 -8.59722257e-01 -1.46322638e-01 1.00847018e+00
-1.65131879e+00 -5.63844502e-01 1.14927508e-01 3.35658342e-01
4.68458265e-01 -1.29440827e-02 5.14958739e-01 1.15078799e-01
-9.93490696e-01 4.68927592e-01 9.18955207e-01 -3.61785978e-01
2.37652823e-01 -1.39984071e+00 -2.84040511e-01 9.18847561e-01
-5.20657301e-01 7.08528161e-01 8.76469791e-01 -4.76165980e-01
-1.69502544e+00 -1.09518540e+00 2.64066696e-01 -1.21303186e-01
8.91539514e-01 -2.65405089e-01 -3.52644235e-01 4.83240843e-01
-2.33445868e-01 1.29855677e-01 5.17174482e-01 1.18187673e-01
1.00720428e-01 -3.25505346e-01 -1.17251003e+00 2.39877552e-01
6.31686985e-01 -1.28348067e-01 -1.57962367e-01 7.03826487e-01
7.03918278e-01 -5.44142425e-01 -1.13082850e+00 5.68216145e-01
5.14889061e-01 -3.90089035e-01 9.27741408e-01 -8.40439141e-01
1.36466175e-01 -3.93150926e-01 -5.86046576e-01 -1.23921573e+00
-1.70779854e-01 -1.19739830e+00 -4.16276991e-01 9.92065609e-01
4.04027283e-01 -1.05157232e+00 4.49069679e-01 6.42052472e-01
-2.53239095e-01 -1.12960231e+00 -1.15673912e+00 -1.62632632e+00
9.67949554e-02 -3.99441898e-01 4.83626753e-01 6.88687205e-01
-9.07102525e-02 4.63174656e-02 -5.11945128e-01 6.66700244e-01
9.40026760e-01 3.12023014e-01 5.90754628e-01 -9.33447361e-01
-7.04497039e-01 -4.80949491e-01 2.75159106e-02 -1.28269112e+00
5.09472229e-02 -6.51702583e-01 4.24294293e-01 -1.15329921e+00
-1.19597100e-01 -4.24118549e-01 -5.36139548e-01 1.10142305e-01
-2.58295983e-01 -2.51972407e-01 1.43492103e-01 -3.23647797e-01
-2.58843094e-01 8.22158992e-01 1.21345580e+00 -2.06481367e-01
-2.54946798e-01 4.74555910e-01 -6.78657532e-01 4.42172289e-01
5.60854018e-01 -4.62628841e-01 -3.44296634e-01 -3.24992239e-02
2.31894776e-01 4.14196372e-01 2.29784250e-01 -5.93892753e-01
4.19794805e-02 -6.50918365e-01 6.30494356e-02 -5.45429945e-01
2.87993044e-01 -9.99954641e-01 1.70469776e-01 5.17567039e-01
-1.53253853e-01 -2.23148763e-01 -3.46809626e-04 1.05511463e+00
4.80762944e-02 -7.13901401e-01 8.17909241e-01 1.26738012e-01
-2.65120920e-02 3.27547282e-01 -3.13605964e-01 -1.05763793e-01
1.07294381e+00 1.87508464e-01 -7.62343407e-03 -4.12051827e-01
-7.55472660e-01 8.14928412e-01 -7.65504688e-02 -1.43965572e-01
3.85511965e-01 -1.15213096e+00 -6.06920063e-01 -1.51759177e-01
-2.67430514e-01 2.87981272e-01 -5.22762164e-02 1.00446916e+00
-3.62618789e-02 4.79696751e-01 6.03805602e-01 -5.50192833e-01
-5.54892063e-01 8.27832222e-01 3.09418589e-01 -4.21560019e-01
-2.33367890e-01 6.77441239e-01 9.14578438e-02 -2.36017272e-01
2.57746220e-01 -7.21342087e-01 5.26801705e-01 -1.01870216e-01
2.67796159e-01 7.05324411e-01 -1.95668004e-02 1.21140480e-01
-1.95643172e-01 3.24602604e-01 3.26590419e-01 -1.79350421e-01
1.38211346e+00 -1.55690148e-01 1.61627337e-01 4.71839070e-01
1.14673638e+00 2.05122139e-02 -1.59148443e+00 -3.34438831e-01
2.00079843e-01 -6.22390151e-01 1.23167731e-01 -7.62257874e-01
-1.10150492e+00 4.07012403e-01 2.18797833e-01 3.36136013e-01
1.09045982e+00 -3.25404346e-01 5.93065023e-01 7.38221765e-01
4.92908478e-01 -1.16062248e+00 -1.00112632e-01 5.04064798e-01
9.10810232e-01 -1.03870094e+00 2.66613811e-01 -6.09147370e-01
-1.91127390e-01 1.00105202e+00 1.72851324e-01 -2.22401336e-01
6.47316933e-01 4.16475236e-01 -4.27960873e-01 2.08500907e-01
-5.49593151e-01 -1.86292037e-01 8.01672414e-02 2.66763091e-01
-9.80490670e-02 5.58886155e-02 -4.80831355e-01 9.26141918e-01
1.02815233e-01 -1.35094285e-01 3.94937366e-01 1.02423561e+00
-3.67650956e-01 -8.55919838e-01 -5.27173638e-01 3.57069671e-01
-6.63018525e-01 8.30257162e-02 2.29331732e-01 9.59885359e-01
-4.96147454e-01 1.08623481e+00 -2.31796265e-01 1.82936385e-01
5.09591281e-01 -1.61245674e-01 5.49440086e-01 -4.48564470e-01
-2.69049078e-01 4.03926939e-01 1.46994919e-01 -6.88537002e-01
-2.82703549e-01 -6.41058862e-01 -9.38388407e-01 -1.27853127e-02
-1.03604627e+00 2.31161833e-01 7.32629180e-01 9.62451637e-01
3.67619209e-02 6.82037890e-01 1.06389821e+00 -8.06823552e-01
-1.47826362e+00 -7.90265739e-01 -9.13127363e-01 -2.17839807e-01
5.27322948e-01 -8.84145796e-01 -4.75161940e-01 -2.08262399e-01] | [5.660253047943115, 3.585454225540161] |
64aba5b3-41de-40c8-a756-3e4e5d2578cd | sequential-skip-prediction-with-few-shot-in | 1901.08203 | null | https://arxiv.org/abs/1901.08203v2 | https://arxiv.org/pdf/1901.08203v2.pdf | Sequential Skip Prediction with Few-shot in Streamed Music Contents | This paper provides an outline of the algorithms submitted for the WSDM Cup 2019 Spotify Sequential Skip Prediction Challenge (team name: mimbres). In the challenge, complete information including acoustic features and user interaction logs for the first half of a listening session is provided. Our goal is to predict whether the individual tracks in the second half of the session will be skipped or not, only given acoustic features. We proposed two different kinds of algorithms that were based on metric learning and sequence learning. The experimental results showed that the sequence learning approach performed significantly better than the metric learning approach. Moreover, we conducted additional experiments to find that significant performance gain can be achieved using complete user log information. | ['Sungkyun Chang', 'Kyogu Lee', 'Seungjin Lee'] | 2019-01-24 | null | null | null | null | ['sequential-skip-prediction'] | ['time-series'] | [ 3.08855057e-01 -2.05248430e-01 -1.19307160e-01 -7.09639907e-01
-1.16078758e+00 -4.46568727e-01 2.68016458e-01 -1.66795492e-01
-4.90478635e-01 6.85104787e-01 3.09754610e-01 -2.47542962e-01
-2.05967933e-01 -1.03285201e-01 -4.53571677e-01 -4.79944319e-01
-4.50971484e-01 1.72019541e-01 6.64971709e-01 -1.91961288e-01
3.28450561e-01 9.27251801e-02 -1.69893348e+00 6.80236697e-01
3.08363467e-01 1.15145135e+00 5.96075058e-01 1.05810153e+00
5.39859496e-02 8.01960826e-01 -6.41079843e-01 3.35394144e-02
2.71006554e-01 -7.50459373e-01 -7.60522783e-01 -4.04747188e-01
2.39548698e-01 -1.49329573e-01 -4.67966169e-01 4.98635083e-01
7.89568424e-01 4.88065034e-01 9.86011997e-02 -1.40871179e+00
4.94149923e-01 1.10591328e+00 -1.54998854e-01 4.65621233e-01
8.82850826e-01 -1.72742665e-01 1.42443025e+00 -7.48963714e-01
2.68417031e-01 8.21641684e-01 7.93036282e-01 5.66804469e-01
-1.04283047e+00 -7.45579004e-01 1.06052145e-01 9.04677808e-01
-1.48013806e+00 -6.23457611e-01 7.43453205e-01 -2.53150791e-01
8.01447749e-01 6.60997212e-01 3.50797385e-01 9.87786233e-01
-3.43946606e-01 1.11155236e+00 7.44564533e-01 -5.47577977e-01
4.38405454e-01 2.22177040e-02 2.85617620e-01 3.10179651e-01
-4.87600118e-01 3.29910487e-01 -1.21510196e+00 -2.52280980e-01
-1.80380419e-01 -5.04229486e-01 -2.83456624e-01 -7.49930367e-02
-1.27559102e+00 5.30871868e-01 -1.63461208e-01 5.58199823e-01
-1.27226099e-01 -8.31245258e-02 5.85103273e-01 5.70609272e-01
1.06563784e-01 2.75873452e-01 -5.94683468e-01 -9.50096309e-01
-1.45997369e+00 2.35841230e-01 1.06708908e+00 8.68309379e-01
5.68662524e-01 -1.62731960e-01 -4.40611392e-01 7.53642797e-01
1.03133410e-01 2.08686635e-01 5.15622318e-01 -1.15353954e+00
5.94250917e-01 -2.19384115e-02 3.98018271e-01 -6.48448527e-01
-6.94572926e-01 2.49872729e-02 -1.77772462e-01 -4.12522674e-01
5.16028106e-01 -4.53523993e-01 -6.55300736e-01 1.71890891e+00
-8.83362591e-02 7.74113715e-01 -3.55902046e-01 8.60271871e-01
6.41609490e-01 5.88024616e-01 -1.58513874e-01 -5.27275383e-01
7.94420600e-01 -7.81998575e-01 -7.74746895e-01 6.99571846e-03
8.29998612e-01 -1.03278387e+00 1.05781019e+00 5.85306048e-01
-9.57227111e-01 -8.85439277e-01 -1.38446903e+00 6.16673172e-01
2.02432185e-01 7.59476572e-02 4.07819957e-01 9.16214526e-01
-9.52378750e-01 8.71795952e-01 -1.00304008e+00 -3.91430527e-01
-3.31851333e-01 4.18255687e-01 7.10218996e-02 -4.78855632e-02
-1.23611736e+00 5.38138628e-01 3.19704354e-01 -1.03159264e-01
-8.60173762e-01 -6.98597789e-01 -4.98331696e-01 -3.15461424e-03
3.39278162e-01 1.22620955e-01 1.86203718e+00 -4.33395326e-01
-1.52881646e+00 6.06136680e-01 -5.16247571e-01 -6.69227242e-01
4.56880569e-01 -3.23217988e-01 -6.68599248e-01 -1.83314323e-01
-2.80117035e-01 -5.44284098e-03 3.74672472e-01 -7.73650467e-01
-1.14722955e+00 -9.53455269e-02 9.32768658e-02 3.74060124e-01
-3.48002285e-01 2.63571411e-01 -5.78220129e-01 -3.74626547e-01
8.64408761e-02 -1.10650325e+00 -4.91236039e-02 -5.24435818e-01
-4.77942973e-01 -2.83986419e-01 7.17571616e-01 -6.66632712e-01
1.73851526e+00 -2.24574232e+00 3.12427767e-02 2.27152750e-01
-3.58620673e-01 2.43915662e-01 -6.51353896e-02 6.88337684e-01
1.14770412e-01 3.02316248e-02 1.12287156e-01 -3.69356632e-01
7.13073388e-02 2.46256758e-02 -3.35429400e-01 2.09183410e-01
-7.90826440e-01 8.42912123e-02 -7.99022317e-01 -2.98334628e-01
2.32439026e-01 -9.60425511e-02 -2.84787774e-01 5.95684409e-01
-3.87131944e-02 4.59051847e-01 -1.94495291e-01 4.43621486e-01
4.55131650e-01 2.48328626e-01 4.17655557e-01 -3.68590802e-01
-2.93071389e-01 6.78396344e-01 -1.44930279e+00 1.99208367e+00
-5.88297784e-01 8.14033985e-01 -7.94558153e-02 -1.06210971e+00
7.06061482e-01 5.37406325e-01 7.83792615e-01 -9.39402163e-01
-1.37506232e-01 1.94719821e-01 1.98121786e-01 -5.69401741e-01
4.33628350e-01 2.42658164e-02 -5.10186255e-02 6.20062232e-01
4.20498773e-02 1.91840127e-01 5.28952241e-01 1.39218912e-01
1.36053312e+00 1.61495432e-01 5.12867570e-02 -2.25230977e-02
6.31720185e-01 -2.12884650e-01 7.05967069e-01 1.04644692e+00
-4.95727062e-01 6.80253804e-01 9.43479985e-02 -2.05724537e-01
-6.22677684e-01 -9.87148643e-01 1.93264216e-01 1.63621545e+00
1.85936436e-01 -7.82978117e-01 -5.89382350e-01 -5.09288013e-01
-3.38162631e-01 8.79907131e-01 -4.25457776e-01 -1.60932034e-01
-6.99944377e-01 -6.24136746e-01 6.02118373e-01 5.81807137e-01
2.50723630e-01 -9.28491175e-01 -3.44162971e-01 1.53546602e-01
-7.30002105e-01 -1.37110686e+00 -8.25163543e-01 3.16288829e-01
-6.20550334e-01 -1.05547476e+00 -2.92551547e-01 -7.18821287e-01
-2.39779919e-01 3.75933737e-01 7.83954859e-01 -1.35944292e-01
-2.76205659e-01 3.72193992e-01 -7.55985737e-01 -2.34954238e-01
-2.15981558e-01 2.46934086e-01 3.81418258e-01 1.59553260e-01
4.70796078e-01 -6.22001469e-01 -6.18188441e-01 8.20265949e-01
-2.24648952e-01 -3.12926412e-01 3.06707501e-01 5.34645498e-01
3.15902114e-01 -1.07782632e-01 5.25774777e-01 -6.41684651e-01
4.26527530e-01 -4.10366476e-01 -1.32629499e-01 2.42036968e-01
-4.82671529e-01 -6.32826388e-02 3.19873750e-01 -2.60957241e-01
-9.10353303e-01 2.37423941e-01 -6.16424263e-01 9.44057778e-02
-3.89840594e-03 4.52451229e-01 -2.48327136e-01 1.49561435e-01
6.61407590e-01 4.36146826e-01 -3.36602718e-01 -7.42780924e-01
-1.14515707e-01 9.89912987e-01 3.85309279e-01 -3.25576633e-01
6.04327500e-01 1.95083156e-01 -4.33797568e-01 -1.00225496e+00
-9.59021628e-01 -1.02615464e+00 -6.19614124e-01 -4.59637254e-01
4.63800728e-01 -7.94967175e-01 -8.92995715e-01 2.61895478e-01
-7.03178287e-01 -4.30020630e-01 -1.06223598e-01 8.85998607e-01
-8.13084483e-01 4.44099545e-01 -2.36989826e-01 -1.06504488e+00
-1.02646723e-01 -6.20885670e-01 6.19993806e-01 2.10809201e-01
-5.57815075e-01 -6.44573689e-01 3.67840409e-01 3.65014583e-01
2.77765542e-01 -3.07994872e-01 4.25898135e-01 -1.11137807e+00
-1.63707539e-01 -4.53928620e-01 3.01638871e-01 3.96877944e-01
1.33541048e-01 -1.35662705e-01 -1.09814107e+00 -4.11105663e-01
-1.99448034e-01 -2.94709504e-01 6.31812572e-01 3.46194863e-01
1.28146279e+00 1.70884714e-01 -4.83015627e-01 3.36119533e-01
1.03789055e+00 4.99801844e-01 6.35248721e-01 1.27507359e-01
2.02201352e-01 3.64475578e-01 9.89943027e-01 7.21455336e-01
2.51962423e-01 1.27264595e+00 1.89568833e-01 3.73527974e-01
-2.65004218e-01 -3.77112508e-01 6.24825418e-01 8.91020358e-01
-6.92757294e-02 -3.76905501e-01 -6.06090546e-01 6.42773271e-01
-1.98446417e+00 -1.24770677e+00 -1.48195818e-01 2.57807255e+00
7.55702496e-01 3.60406071e-01 7.39248693e-01 6.94569707e-01
6.38322413e-01 1.05733618e-01 -2.33178288e-01 -3.68295610e-01
1.44994333e-01 2.13330880e-01 2.21547768e-01 5.25361180e-01
-1.01694345e+00 4.52576131e-01 7.38950300e+00 8.69209051e-01
-6.29639864e-01 4.82740216e-02 -2.03041751e-02 -5.49383938e-01
2.72551738e-02 -1.29394889e-01 -1.02774620e+00 6.76547229e-01
1.62789488e+00 1.73974037e-03 3.36469173e-01 3.69001746e-01
5.78010976e-01 -2.63484180e-01 -1.22206295e+00 8.98993492e-01
2.69876812e-02 -1.05959260e+00 -6.05484009e-01 -3.00065339e-01
4.10315990e-01 3.07915211e-01 -2.36471832e-01 5.80280721e-01
2.46007636e-01 -6.29309714e-01 5.86663365e-01 6.03894174e-01
4.43498760e-01 -6.87176883e-01 7.78820634e-01 5.75258255e-01
-1.31513965e+00 -2.53125370e-01 -2.10966855e-01 -2.20208585e-01
4.93725985e-01 5.12825847e-01 -9.68669713e-01 5.59607983e-01
8.50628674e-01 4.38337982e-01 -2.65618652e-01 1.73803735e+00
-4.49129641e-02 1.40816498e+00 -4.52533066e-01 -1.31185561e-01
-2.82814112e-02 2.90693402e-01 6.40568912e-01 1.30687022e+00
5.20927727e-01 -5.00902534e-02 4.44649130e-01 9.56159923e-03
2.13889897e-01 1.73845798e-01 -2.54821122e-01 1.55876920e-01
6.39295459e-01 9.49025929e-01 -2.89139032e-01 7.73317590e-02
-5.32577395e-01 9.96015608e-01 -1.63144290e-01 2.10947692e-01
-9.04121876e-01 -6.48477554e-01 6.48657382e-01 5.68144247e-02
5.25709569e-01 -1.84886739e-01 -1.69056639e-01 -7.86004126e-01
-6.09418303e-02 -4.84974921e-01 6.25518560e-01 -4.19460922e-01
-9.19379711e-01 3.97685260e-01 -2.10435718e-01 -1.39697289e+00
-3.83206666e-01 -1.80984542e-01 -7.39529610e-01 6.64263546e-01
-9.94568646e-01 -7.16255844e-01 -3.03142846e-01 4.14576709e-01
7.09310949e-01 -3.43757868e-01 9.91616011e-01 6.72174156e-01
-2.62227416e-01 1.12724400e+00 3.27515244e-01 -6.78019002e-02
8.66558313e-01 -1.29757285e+00 9.04594064e-02 3.81861717e-01
6.22182786e-01 1.47386178e-01 1.13480282e+00 -2.38030404e-01
-1.10243618e+00 -5.32093644e-01 1.23573792e+00 -2.73482472e-01
5.78227162e-01 -3.99881840e-01 -4.80112076e-01 5.25622308e-01
7.60903805e-02 -3.06987196e-01 1.27355671e+00 5.95857918e-01
4.98069078e-02 -3.08235019e-01 -7.69627213e-01 2.84270644e-01
1.05525005e+00 -4.47899491e-01 -4.97996837e-01 2.08363011e-01
5.54613590e-01 -4.67936844e-01 -6.15612447e-01 4.97964203e-01
8.50359321e-01 -8.80671203e-01 8.04640949e-01 -5.05900621e-01
-2.56679267e-01 -3.08615297e-01 -7.70001233e-01 -1.11995363e+00
-7.02251792e-02 -9.55992579e-01 -3.94477576e-01 1.13604927e+00
6.72858179e-01 2.60392427e-01 9.74394679e-01 3.35783929e-01
-2.62345374e-01 -3.49786133e-01 -1.08941102e+00 -1.01417744e+00
-4.40023959e-01 -8.15449536e-01 1.86927423e-01 2.85828680e-01
3.74190152e-01 3.47910404e-01 -9.07118261e-01 -2.38779164e-03
4.50452328e-01 1.42040566e-01 6.82178497e-01 -1.11511636e+00
-5.20893514e-01 -1.17574662e-01 -5.24904251e-01 -1.26123393e+00
-1.15674093e-01 -7.50659525e-01 3.84393573e-01 -1.20778751e+00
8.88849050e-02 -1.36508435e-01 -8.85864973e-01 3.36629927e-01
-8.80698562e-02 3.93134445e-01 3.13290805e-01 -1.01710930e-01
-9.31654036e-01 2.05830723e-01 5.86908162e-01 -4.24873978e-02
-4.28073943e-01 1.08749592e+00 -4.02724624e-01 3.94705057e-01
8.01344693e-01 -3.79526228e-01 -5.99495828e-01 -1.34095162e-01
7.01741967e-03 3.16265404e-01 -1.00430347e-01 -1.32273018e+00
4.26768959e-01 -1.98640808e-01 -1.92898467e-01 -9.62870717e-01
5.36732793e-01 -5.94107568e-01 -1.38561949e-01 4.14803535e-01
-6.06944442e-01 -4.11024123e-01 2.86292225e-01 7.25510120e-01
-2.40440637e-01 -3.90963465e-01 4.32727247e-01 2.52488256e-01
-1.03336525e+00 1.65495593e-02 -6.35272741e-01 -2.92596612e-02
9.83242393e-01 -1.98974401e-01 5.27418315e-01 -6.91092074e-01
-1.37038696e+00 2.88941890e-01 -3.42614681e-01 5.92662215e-01
3.43644261e-01 -1.25521028e+00 -7.57590890e-01 -9.61847752e-02
6.83018044e-02 -9.64464128e-01 4.13164675e-01 9.09493089e-01
2.73132045e-02 5.64341187e-01 3.67942043e-02 -6.43431604e-01
-1.60260153e+00 2.69538015e-02 2.99652308e-01 -3.61696988e-01
-2.82189012e-01 1.09217811e+00 -4.48294342e-01 -4.58981007e-01
8.59838247e-01 3.44416052e-02 -1.21353991e-01 1.98309094e-01
8.45770121e-01 6.64639354e-01 4.25095528e-01 -2.91190326e-01
-6.28456056e-01 1.64780766e-01 -1.62347659e-01 -4.14472044e-01
1.12832844e+00 -4.93781567e-01 5.03915370e-01 8.32757175e-01
1.25927556e+00 -1.31212980e-01 -9.93799627e-01 -5.75227439e-01
4.90117729e-01 -4.57217813e-01 3.39178331e-02 -1.07435298e+00
-6.55980647e-01 7.80276597e-01 9.54467535e-01 2.29035228e-01
1.18272185e+00 -1.47280768e-01 1.02080035e+00 4.22603160e-01
6.24820769e-01 -1.46398866e+00 -6.25158399e-02 5.41353941e-01
5.60063660e-01 -1.15050673e+00 -1.42147854e-01 -1.49628535e-01
-6.74430549e-01 8.79222035e-01 6.19774163e-01 3.81902933e-01
8.97984922e-01 3.18425924e-01 -5.28680533e-02 2.51874894e-01
-8.93681765e-01 -3.91229481e-01 4.22113448e-01 6.92714870e-01
6.75259531e-01 1.68499261e-01 -5.23428738e-01 9.58077073e-01
-3.26409727e-01 -2.37109270e-02 3.30817699e-01 1.01781750e+00
-7.24251628e-01 -1.36677957e+00 5.57616875e-02 4.66215193e-01
-3.20187449e-01 3.14054117e-02 -3.18418801e-01 3.97421926e-01
-7.58579299e-02 1.35881984e+00 -1.74611837e-01 -1.10958183e+00
6.84186578e-01 5.00362635e-01 4.61048633e-01 -5.66071451e-01
-4.12584066e-01 -4.18111905e-02 4.32899445e-01 -8.39999497e-01
-3.44869703e-01 -1.01204622e+00 -1.27275395e+00 -3.87252361e-01
-2.54714340e-01 5.32705188e-01 7.13066041e-01 1.10439515e+00
4.93695289e-02 5.20776868e-01 1.06915760e+00 -6.68199420e-01
-7.32558012e-01 -1.22430396e+00 -7.46614039e-01 6.71747476e-02
2.76474953e-01 -4.42914665e-01 -4.45563465e-01 1.31645635e-01] | [15.590384483337402, 5.135326385498047] |
8e9ead2a-56d1-4f68-b650-7907965fb725 | romnet-renovate-the-old-memories | 2202.02606 | null | https://arxiv.org/abs/2202.02606v2 | https://arxiv.org/pdf/2202.02606v2.pdf | ROMNet: Renovate the Old Memories | Renovating the memories in old photos is an intriguing research topic in computer vision fields. These legacy images often suffer from severe and commingled degradations such as cracks, noise, and color-fading, while lack of large-scale paired old photo datasets makes this restoration task very challenging. In this work, we present a novel reference-based end-to-end learning framework that can jointly repair and colorize the degraded legacy pictures. Specifically, the proposed framework consists of three modules: a restoration sub-network for degradation restoration, a similarity sub-network for color histogram matching and transfer, and a colorization subnet that learns to predict the chroma elements of the images conditioned on chromatic reference signals. The whole system takes advantage of the color histogram priors in a given reference image, which vastly reduces the dependency on large-scale training data. Apart from the proposed method, we also create, to our knowledge, the first public and real-world old photo dataset with paired ground truth for evaluating old photo restoration models, wherein each old photo is paired with a manually restored pristine image by PhotoShop experts. Our extensive experiments conducted on both synthetic and real-world datasets demonstrate that our method significantly outperforms state-of-the-arts both quantitatively and qualitatively. | ['Hongkai Yu', 'Jiaqi Ma', 'Zibo Meng', 'Jinlong Li', 'Xiaoyu Dong', 'Yuanqi Du', 'Zhengzhong Tu', 'Runsheng Xu'] | 2022-02-05 | null | null | null | null | ['colorization'] | ['computer-vision'] | [ 3.15357059e-01 -2.96447098e-01 3.32489252e-01 -1.33843899e-01
-6.71758235e-01 -3.22024524e-01 5.52599013e-01 -1.74304679e-01
-3.38649005e-01 9.00188506e-01 2.16146365e-01 7.14284703e-02
3.51979494e-01 -7.90442586e-01 -1.00285101e+00 -9.00401771e-01
2.70726949e-01 -1.21622257e-01 4.26757693e-01 -1.99583143e-01
2.28144452e-01 4.32374775e-01 -1.87194443e+00 1.01706095e-01
1.31319749e+00 1.01729310e+00 4.02442962e-01 6.02644563e-01
2.42850810e-01 7.96619713e-01 -5.08835971e-01 -5.26450872e-01
3.84151459e-01 -2.91311324e-01 -4.16422039e-01 3.89207363e-01
9.44527805e-01 -5.81536055e-01 -8.28149199e-01 1.26410675e+00
6.49284065e-01 3.02475065e-01 4.48441863e-01 -1.43894660e+00
-1.28594208e+00 2.14770377e-01 -5.11605322e-01 1.67605355e-01
2.76670933e-01 5.26089787e-01 4.95327026e-01 -9.14590776e-01
6.66339338e-01 1.29533947e+00 6.45145535e-01 3.40656400e-01
-1.14485407e+00 -8.38912904e-01 3.07495203e-02 7.72166073e-01
-1.01666713e+00 -5.14026463e-01 9.50220108e-01 -2.47981384e-01
4.59422737e-01 -8.99705477e-03 8.61398339e-01 1.16923475e+00
3.80010933e-01 5.45221150e-01 1.54499221e+00 -2.63706654e-01
3.71173978e-01 -1.38391033e-01 -1.27581373e-01 4.80658054e-01
1.00000463e-01 4.02232915e-01 -2.83389449e-01 3.23450714e-01
6.43727899e-01 2.03027889e-01 -5.87419748e-01 -4.94757593e-01
-1.07330155e+00 9.50785950e-02 8.84060740e-01 4.77367267e-02
-3.76216441e-01 4.94400747e-02 4.13617678e-02 4.12167549e-01
3.87794465e-01 -4.67159413e-02 -1.79748297e-01 3.60616148e-01
-7.44749606e-01 -1.45886317e-01 4.01095331e-01 8.11488032e-01
1.04547024e+00 2.02323914e-01 -5.18077612e-02 1.05387008e+00
1.13144860e-01 7.41115272e-01 4.98330563e-01 -1.17990613e+00
3.67554963e-01 3.61164480e-01 3.10937017e-01 -1.15107429e+00
-1.08756199e-01 -5.45444369e-01 -1.29851043e+00 5.62416494e-01
2.83111274e-01 2.58799046e-01 -1.27083004e+00 1.72561347e+00
2.87852794e-01 4.75929767e-01 2.80831784e-01 9.92747366e-01
6.23039067e-01 7.44432628e-01 7.38973096e-02 -3.95567715e-01
9.87220228e-01 -1.25918496e+00 -5.28981686e-01 -5.55674553e-01
-3.82449478e-01 -9.99492109e-01 1.23824728e+00 4.85281140e-01
-1.06860876e+00 -9.98470306e-01 -1.33622289e+00 -4.08568025e-01
-2.59163231e-01 1.07148208e-01 3.42863441e-01 2.56380141e-01
-1.27222133e+00 6.87339664e-01 -4.06437248e-01 -6.35824502e-01
5.98361850e-01 -1.10624172e-01 -5.73550940e-01 -8.35770547e-01
-9.52007532e-01 1.04788268e+00 3.40189248e-01 5.04425287e-01
-1.32826161e+00 -6.11740708e-01 -6.30271018e-01 2.15377240e-03
1.95359677e-01 -8.23704839e-01 8.15826058e-01 -1.23560309e+00
-1.33489203e+00 9.22641098e-01 2.73488671e-01 -1.34621441e-01
6.08165026e-01 -2.19210759e-01 -8.04484904e-01 8.43514036e-03
1.61531866e-01 6.37196004e-01 1.10403895e+00 -1.86065936e+00
-4.76694345e-01 -4.72323239e-01 -4.94380221e-02 1.42697528e-01
-3.87345463e-01 -3.99007440e-01 -7.71824121e-01 -7.06560612e-01
1.33106291e-01 -7.19335914e-01 -5.81629723e-02 4.73557025e-01
-2.80948102e-01 3.82358998e-01 6.78301394e-01 -1.17559528e+00
7.04136729e-01 -2.24358010e+00 1.77603960e-01 -5.25593087e-02
8.24384019e-02 2.23995641e-01 -5.42710125e-01 3.39804024e-01
-2.35985860e-01 -3.41120631e-01 -4.17979985e-01 -2.78621048e-01
-1.10116921e-01 1.54305860e-01 -3.55426937e-01 7.14296460e-01
-2.19326317e-01 7.10000634e-01 -1.07534850e+00 -4.63850915e-01
3.43144327e-01 7.74699390e-01 -9.03799385e-02 5.58536530e-01
-1.98255718e-01 4.00577486e-01 2.03558594e-01 8.09843183e-01
1.20872903e+00 1.35535851e-01 -1.29375898e-03 -6.53290927e-01
2.84932423e-02 -5.23312211e-01 -1.13870203e+00 1.99305439e+00
-4.07650709e-01 4.03340429e-01 9.20293182e-02 -6.72399402e-01
9.35795426e-01 -2.11865440e-01 7.66137242e-02 -1.06194139e+00
1.02381660e-02 2.27736682e-01 -4.39870983e-01 -6.46537006e-01
5.76949894e-01 -8.65184590e-02 1.28353477e-01 1.26268387e-01
4.63925451e-02 -1.92090571e-01 2.30755195e-01 1.80434138e-01
1.01394320e+00 3.06658477e-01 1.45433017e-03 1.06770456e-01
6.33064926e-01 -3.45346779e-01 8.09069514e-01 3.90861779e-01
-3.31182092e-01 8.03609312e-01 6.79610856e-03 -3.50681514e-01
-1.35814524e+00 -1.65633142e+00 3.48891973e-01 8.50530267e-01
6.30169570e-01 9.88600329e-02 -5.61932087e-01 -2.04480067e-01
-1.40500125e-02 5.15015423e-01 -5.96505105e-01 -3.16902280e-01
-4.00003761e-01 -6.35969698e-01 2.16056809e-01 5.24092972e-01
1.02256060e+00 -1.18022728e+00 -6.42705336e-02 6.68286383e-02
-4.08174157e-01 -1.07589829e+00 -7.05040336e-01 -1.25486612e-01
-7.99823582e-01 -1.36652637e+00 -7.77444899e-01 -1.13147473e+00
7.98296750e-01 7.15895832e-01 9.59976256e-01 1.40953630e-01
-4.38913852e-01 4.86827701e-01 -2.19540745e-01 1.39738649e-01
-5.51198661e-01 -4.60971057e-01 -1.88640773e-01 3.12986255e-01
-2.57143229e-01 -1.03789544e+00 -1.06181192e+00 2.21172348e-01
-1.24835134e+00 2.12966830e-01 1.02485013e+00 1.01587057e+00
5.90824127e-01 2.07854629e-01 4.83390778e-01 -4.37067479e-01
2.97409654e-01 -3.36295575e-01 -4.01366770e-01 6.58142328e-01
-8.56235206e-01 -1.57831192e-01 6.75834358e-01 -3.91774207e-01
-1.48264027e+00 1.99296653e-01 4.43895310e-02 -5.32671869e-01
-2.07517654e-01 1.14165703e-02 -6.38564169e-01 -2.27791458e-01
5.87626338e-01 4.24981862e-01 -1.41577376e-02 -6.14074647e-01
8.25635850e-01 5.80542207e-01 1.59637737e+00 -4.44888651e-01
1.23070228e+00 5.23215830e-01 -2.76754647e-01 -6.11135840e-01
-7.35171318e-01 -1.95891678e-01 -6.97627842e-01 -4.87981498e-01
8.41093421e-01 -1.20491004e+00 -4.26323146e-01 1.13448572e+00
-9.97089684e-01 -3.55035603e-01 -2.45850950e-01 8.78091156e-02
-5.58471560e-01 8.44247580e-01 -6.95774019e-01 -3.41991723e-01
-4.82581049e-01 -7.84857869e-01 7.58105576e-01 5.94317853e-01
6.59490764e-01 -5.36512852e-01 1.18163891e-01 5.74073136e-01
5.11172771e-01 3.35313141e-01 8.90314579e-01 4.82982188e-01
-8.74639392e-01 2.12937463e-02 -5.71079731e-01 8.89890075e-01
8.37745965e-02 -1.39028147e-01 -1.07234907e+00 -5.04815876e-01
-1.81422308e-01 -4.06595856e-01 1.25864899e+00 -4.27011475e-02
1.04182851e+00 -1.21538468e-01 -4.68082540e-02 7.72002101e-01
1.79232085e+00 4.96159233e-02 1.27228892e+00 5.25730729e-01
8.70460212e-01 3.42547238e-01 5.41547954e-01 3.17466587e-01
5.65989673e-01 4.01318073e-01 7.03709900e-01 -3.05985510e-01
-8.03628385e-01 -4.34805006e-01 5.37322044e-01 7.41328478e-01
3.63429189e-02 -2.44669616e-01 -5.49836397e-01 6.14991009e-01
-1.74922812e+00 -9.30444121e-01 8.11791420e-02 2.23634434e+00
9.41129804e-01 -1.50212497e-01 -4.67791975e-01 5.10761850e-02
1.02359152e+00 2.26366431e-01 -9.01528060e-01 2.41973132e-01
-6.82391107e-01 4.96181399e-02 3.42130572e-01 2.90007770e-01
-1.09391761e+00 9.57668006e-01 5.11498451e+00 7.26489484e-01
-9.49398458e-01 1.79576859e-01 6.84581459e-01 2.83792913e-01
-9.07282755e-02 1.66040018e-01 -4.09056172e-02 4.64091033e-01
4.30443972e-01 -3.65087315e-02 8.36832345e-01 5.94330132e-01
-1.30369803e-02 -2.38826528e-01 -8.18707526e-01 1.28273368e+00
6.27876997e-01 -1.05342329e+00 2.34038010e-02 -2.92883724e-01
1.04165304e+00 -7.54022449e-02 1.36859193e-01 1.02316976e-01
4.06069368e-01 -6.98869288e-01 9.82131302e-01 9.89455104e-01
8.78703654e-01 -5.64775527e-01 5.73035479e-01 -1.84778675e-01
-1.13635612e+00 -4.91031677e-01 -6.16968036e-01 1.15401872e-01
1.47323683e-01 6.53966606e-01 -1.53717428e-01 6.29970908e-01
1.00234067e+00 1.15643084e+00 -1.22406149e+00 1.45212996e+00
-5.77166617e-01 2.44621456e-01 2.03050226e-01 7.56642759e-01
-3.98815393e-01 -2.85427988e-01 2.25026920e-01 7.01404274e-01
4.73840415e-01 3.57093811e-02 1.59320161e-01 7.32722163e-01
-2.98174322e-01 -1.61169082e-01 -2.54776597e-01 2.67644614e-01
5.18549860e-01 1.36001408e+00 -5.77304721e-01 -3.36044103e-01
-2.86433011e-01 1.51534307e+00 1.77941799e-01 6.75308049e-01
-6.41312301e-01 -1.84828639e-01 4.59553272e-01 -1.17598310e-01
2.37073332e-01 -1.07640669e-01 -1.17675915e-01 -1.38906384e+00
1.97417438e-01 -8.61913562e-01 3.56317192e-01 -1.56561732e+00
-1.85313070e+00 7.49683321e-01 -3.48796308e-01 -1.36885846e+00
3.23898256e-01 -2.80599058e-01 -7.36874521e-01 5.94426215e-01
-1.68819273e+00 -1.54718184e+00 -1.08881629e+00 8.03654790e-01
3.00079286e-01 -1.27900511e-01 4.40157652e-01 5.71737468e-01
-6.59633398e-01 3.60367298e-01 5.52187860e-01 -6.52801692e-02
1.31517971e+00 -1.10033011e+00 3.63045722e-01 1.33373475e+00
-3.05007249e-01 8.85997638e-02 8.24069619e-01 -6.91648483e-01
-1.51761734e+00 -1.35681295e+00 2.21659437e-01 1.57849476e-01
4.64727610e-01 -1.43309653e-01 -1.03717053e+00 3.75371188e-01
4.45829302e-01 1.30488068e-01 6.39091656e-02 -3.85881782e-01
-6.41847610e-01 -8.08444619e-01 -1.14711237e+00 5.49330175e-01
1.13645625e+00 -5.31017065e-01 -5.15547931e-01 2.08619654e-01
7.16559052e-01 -2.37734184e-01 -7.75274277e-01 1.97238997e-01
6.35674357e-01 -1.20414364e+00 1.37796748e+00 1.34558082e-01
5.10844648e-01 -7.16627479e-01 -2.69969761e-01 -1.61375880e+00
-2.52906919e-01 -2.46906817e-01 1.30834416e-01 1.62596178e+00
-1.71323955e-01 -5.88702619e-01 4.20883477e-01 4.53643262e-01
-4.88467485e-01 -2.21021652e-01 -8.25700343e-01 -7.12576926e-01
-1.57413688e-02 -7.83422142e-02 6.15591168e-01 7.61590958e-01
-6.75139427e-01 3.01297337e-01 -7.09576726e-01 3.39821607e-01
1.15529692e+00 2.19088048e-01 8.60327065e-01 -1.12621379e+00
-2.56159693e-01 -1.56949028e-01 -5.12139320e-01 -6.44632339e-01
1.51612788e-01 -6.06772602e-01 4.02441829e-01 -1.78301466e+00
3.86250705e-01 -3.26905221e-01 -4.12541687e-01 4.21816170e-01
-2.41493776e-01 6.56989992e-01 2.68960029e-01 2.12590441e-01
-6.23376787e-01 1.08972216e+00 1.35996997e+00 -6.50382400e-01
-8.12541321e-02 -5.62157273e-01 -6.98014855e-01 4.89779592e-01
6.44339979e-01 -2.81208038e-01 -4.10546869e-01 -5.34327328e-01
-1.08369393e-02 -3.96451242e-02 8.02833974e-01 -1.40680194e+00
3.75944555e-01 -2.12929785e-01 8.19776833e-01 -5.23049653e-01
4.12940413e-01 -7.67514586e-01 5.09391725e-01 3.41793954e-01
6.77143112e-02 -2.27622446e-02 1.09491169e-01 8.88072073e-01
-1.21994652e-01 2.52970427e-01 1.21955264e+00 -1.04987785e-01
-1.21629560e+00 3.88616532e-01 1.74661905e-01 -6.46741688e-02
1.06068575e+00 -2.29022756e-01 -9.66622353e-01 -2.82127172e-01
-4.94941384e-01 1.24259241e-01 9.96635377e-01 5.49445391e-01
9.02002573e-01 -1.38456237e+00 -6.88714623e-01 1.43512532e-01
2.07344532e-01 -1.68038979e-01 1.03113031e+00 4.73868161e-01
-6.21067405e-01 -5.07778764e-01 -9.12529767e-01 -2.97047228e-01
-1.19491446e+00 1.06321979e+00 3.27012956e-01 1.70042664e-01
-8.93503010e-01 5.10373354e-01 1.02049902e-01 -1.95122153e-01
3.99617434e-01 -1.69762313e-01 5.85225113e-02 -6.59541711e-02
5.79674363e-01 5.83082736e-01 -6.82812631e-02 -7.95109808e-01
-9.86187980e-02 6.76364422e-01 3.92498709e-02 3.61302383e-02
1.59323585e+00 -6.47700071e-01 -4.95771587e-01 1.72692358e-01
1.08379614e+00 -4.25821602e-01 -1.67827916e+00 -4.81224149e-01
-5.21486402e-01 -8.32091153e-01 8.56413916e-02 -1.19229114e+00
-1.45754039e+00 8.04924309e-01 1.19003022e+00 -4.76365238e-01
1.79915822e+00 -2.51060694e-01 1.07499921e+00 2.18730673e-01
3.80208045e-01 -1.14801836e+00 5.92628539e-01 -4.06609923e-02
1.25805914e+00 -1.15884984e+00 1.46808103e-01 -3.26170325e-01
-5.31950951e-01 1.00684285e+00 9.14320290e-01 -2.71843642e-01
4.32278693e-01 -2.92701870e-01 2.56706923e-01 2.21868262e-01
-4.22427893e-01 -2.19329864e-01 3.60460319e-02 9.45618510e-01
-2.44295940e-01 -1.58314124e-01 -4.99049611e-02 4.47401702e-01
8.03509355e-03 1.62547633e-01 8.23147416e-01 6.65784836e-01
-4.91874546e-01 -1.04264855e+00 -6.54053330e-01 1.30751580e-01
1.10810496e-01 -9.16159526e-02 -7.59405717e-02 4.23640162e-01
4.50321943e-01 1.11749077e+00 -1.75138459e-01 -6.49404228e-01
4.06680644e-01 -2.24369153e-01 5.92755616e-01 -4.79937010e-02
-1.62609145e-01 -1.63450018e-01 -2.96190172e-01 -6.18869781e-01
-5.66153526e-01 -3.73134524e-01 -9.10377026e-01 -4.57611173e-01
4.94415127e-02 -2.40138337e-01 5.30294001e-01 6.24186814e-01
3.15440595e-01 5.42076230e-01 7.74878860e-01 -1.22970867e+00
-3.79948437e-01 -1.00259888e+00 -8.19145203e-01 7.43735373e-01
3.85897040e-01 -7.30858207e-01 -4.24779862e-01 5.16234934e-01] | [11.098012924194336, -2.1017887592315674] |
5bd0c38c-bc21-4b5f-ad0f-52ce6b5c1579 | reconstructing-undersampled-photoacoustic | 2006.00251 | null | https://arxiv.org/abs/2006.00251v1 | https://arxiv.org/pdf/2006.00251v1.pdf | Reconstructing undersampled photoacoustic microscopy images using deep learning | One primary technical challenge in photoacoustic microscopy (PAM) is the necessary compromise between spatial resolution and imaging speed. In this study, we propose a novel application of deep learning principles to reconstruct undersampled PAM images and transcend the trade-off between spatial resolution and imaging speed. We compared various convolutional neural network (CNN) architectures, and selected a fully dense U-net (FD U-net) model that produced the best results. To mimic various undersampling conditions in practice, we artificially downsampled fully-sampled PAM images of mouse brain vasculature at different ratios. This allowed us to not only definitively establish the ground truth, but also train and test our deep learning model at various imaging conditions. Our results and numerical analysis have collectively demonstrated the robust performance of our model to reconstruct PAM images with as few as 2% of the original pixels, which may effectively shorten the imaging time without substantially sacrificing the image quality. | ['Junjie Yao', 'Maomao Chen', 'Daiwei Li', 'Jianwen Luo', 'Dong Zhang', 'Anthony DiSpirito III', 'Tri Vu', 'Roarke Horstmeyer'] | 2020-05-30 | null | null | null | null | ['3d-human-action-recognition'] | ['computer-vision'] | [ 5.50579309e-01 -2.29256123e-01 4.39119041e-01 1.40163690e-01
-7.56587148e-01 -3.85892332e-01 2.98683941e-01 -3.28283668e-01
-8.25650454e-01 8.64986002e-01 -1.13274328e-01 -4.45494503e-01
7.09475996e-03 -6.49106801e-01 -7.17978716e-01 -1.08842599e+00
1.91382170e-02 -1.92259118e-01 3.93868059e-01 3.99484396e-01
2.97223747e-01 8.26388121e-01 -1.19734144e+00 1.31115735e-01
7.57105589e-01 9.49645817e-01 6.35322392e-01 8.91225696e-01
-2.54041492e-03 7.07660913e-01 -4.81412888e-01 -5.15038520e-02
3.58632386e-01 -3.28401476e-01 -6.79038584e-01 -2.40603000e-01
4.54167426e-01 -6.50717854e-01 -6.59733653e-01 1.01534259e+00
9.19151127e-01 9.73228365e-02 4.86907959e-01 -5.50652683e-01
-7.38310456e-01 5.17320037e-01 -7.66033649e-01 8.25490892e-01
-2.58290768e-01 4.89039034e-01 4.11343604e-01 -7.12777317e-01
4.00241941e-01 7.15218067e-01 6.67251110e-01 7.97489166e-01
-1.45505893e+00 -7.14087963e-01 -4.76652801e-01 1.29631028e-01
-1.09229469e+00 -6.63313508e-01 5.32587528e-01 -3.98777097e-01
4.88304108e-01 6.96668215e-03 6.72396421e-01 7.55646229e-01
6.34221792e-01 1.49656743e-01 1.37857091e+00 -3.36397916e-01
2.16680974e-01 -3.53296936e-01 -2.88769066e-01 5.90699732e-01
2.96964049e-01 2.16526538e-01 -3.00222307e-01 4.07484882e-02
1.62698233e+00 -2.75680751e-01 -7.62791395e-01 1.14252670e-02
-1.24140513e+00 4.37534869e-01 4.03737903e-01 4.34729636e-01
-1.84556350e-01 3.00898343e-01 1.03607439e-01 -1.36625022e-01
-4.60106656e-02 8.25861156e-01 -2.10697219e-01 -1.59618203e-02
-8.24994922e-01 -2.44894251e-01 3.29622298e-01 3.61415178e-01
4.91287023e-01 3.01649451e-01 -4.66382876e-02 7.81756163e-01
-1.82818845e-01 2.28832036e-01 6.92337990e-01 -1.57386291e+00
-1.23838201e-01 -1.61183793e-02 1.67667806e-01 -5.83204448e-01
-3.47135663e-01 -3.50478381e-01 -9.61297870e-01 6.00927889e-01
8.01061809e-01 -3.51709157e-01 -1.01762605e+00 1.68104720e+00
1.93788961e-01 3.75945449e-01 -3.01318653e-02 1.12389839e+00
8.66840065e-01 5.42573333e-01 2.01928586e-01 -2.84880489e-01
1.16707706e+00 -6.25915229e-01 -5.95384657e-01 -4.62784357e-02
3.19764256e-01 -6.86296999e-01 1.15898812e+00 1.16424888e-01
-1.36871195e+00 -4.39069182e-01 -1.30635893e+00 -1.30568519e-01
1.83316413e-02 2.52099991e-01 3.91496420e-01 3.67931217e-01
-1.23324013e+00 9.41655099e-01 -1.00114524e+00 3.56218917e-03
7.67651737e-01 2.54153043e-01 -4.52772349e-01 -1.83986891e-02
-8.74497950e-01 7.31391311e-01 7.43590072e-02 3.33939344e-02
-1.13303924e+00 -1.40266287e+00 -4.15863842e-01 1.12019137e-01
-2.54202187e-01 -8.43095720e-01 1.28292775e+00 -2.27992162e-01
-1.80849218e+00 7.91013837e-01 5.40107451e-02 -7.18964040e-01
5.24298549e-01 1.57887861e-01 -8.39308649e-02 6.28983200e-01
-1.12427011e-01 7.79946983e-01 5.23113251e-01 -1.26772428e+00
-4.38355953e-01 -1.42546698e-01 -1.86022088e-01 -8.94825831e-02
-9.14021581e-02 -1.86264768e-01 -8.21109302e-03 -2.71655947e-01
-1.95273291e-02 -4.50804293e-01 -2.78157413e-01 5.67168772e-01
-4.41364798e-04 5.45874119e-01 9.08988953e-01 -5.11499941e-01
6.56803727e-01 -1.95667410e+00 -1.55153170e-01 -4.05575126e-01
5.50328076e-01 5.56902111e-01 -3.77551377e-01 3.64714339e-02
-1.35005653e-01 2.29044929e-02 -3.54600608e-01 -3.98414023e-02
-7.37803936e-01 1.17122799e-01 -2.00948924e-01 8.11988652e-01
1.42781645e-01 9.88084078e-01 -8.50147009e-01 -5.65526307e-01
4.34262335e-01 8.84381831e-01 -3.19697917e-01 2.87385076e-01
-1.76253505e-02 9.68808115e-01 -1.68594718e-01 4.18119639e-01
8.72887075e-01 -4.36148703e-01 1.12825140e-01 -5.78798473e-01
-5.81818759e-01 -1.03133127e-01 -5.67933619e-01 1.45612192e+00
-6.93454802e-01 1.10621727e+00 4.36603874e-01 -8.41532171e-01
6.18680120e-01 2.25422516e-01 7.90373623e-01 -9.60257292e-01
3.60296458e-01 1.64572820e-01 1.51067242e-01 -6.47873998e-01
-2.30200309e-02 -4.79936302e-01 6.23899162e-01 4.59058642e-01
-1.28511442e-02 -2.37635538e-01 -8.39554518e-02 -3.28737825e-01
9.08236265e-01 -1.31938308e-01 1.72138944e-01 -6.16549730e-01
3.72066170e-01 -6.41074702e-02 1.60358533e-01 6.71801805e-01
-4.69545513e-01 8.75702858e-01 4.53638196e-01 -5.48215091e-01
-1.37624705e+00 -1.02936196e+00 -5.35596073e-01 5.77447295e-01
1.95398822e-01 4.57801908e-01 -7.34094560e-01 -9.65481624e-02
-4.08315182e-01 9.90109071e-02 -5.64928532e-01 8.45483541e-02
-7.46182859e-01 -9.54137921e-01 8.47335100e-01 3.22744906e-01
7.91229725e-01 -8.77332509e-01 -1.07662928e+00 2.96402574e-01
-1.50426120e-01 -1.44190073e+00 -1.50466710e-01 3.53862017e-01
-5.75150728e-01 -1.17708802e+00 -9.97605681e-01 -8.02279353e-01
7.49674141e-01 4.64626551e-01 7.05349386e-01 -1.10632449e-01
-6.83203816e-01 -5.45895332e-03 6.45171255e-02 -2.67091632e-01
-3.47903728e-01 -3.25664997e-01 1.37798404e-02 -3.81269813e-01
-4.12728302e-02 -9.45641160e-01 -1.10451913e+00 1.80935740e-01
-1.13857162e+00 -1.10641450e-01 7.72843719e-01 9.34141695e-01
7.28235304e-01 1.93903923e-01 5.79305172e-01 -4.85269994e-01
4.35859442e-01 -5.86913899e-02 -7.99876034e-01 -1.36181906e-01
-2.14263305e-01 -8.59753042e-02 8.33413780e-01 -7.37256944e-01
-1.04132760e+00 -1.51560381e-02 -3.39824259e-01 -4.80453640e-01
-2.44970426e-01 3.42077553e-01 8.24629441e-02 -7.00226128e-01
8.60666633e-01 5.21951377e-01 3.84420037e-01 -1.01620756e-01
1.49611622e-01 3.54999095e-01 9.31568205e-01 -5.70929527e-01
5.45092702e-01 1.00983441e+00 2.16357574e-01 -1.09140420e+00
-6.38014436e-01 -5.63131385e-02 -5.19111931e-01 -2.59793460e-01
9.38243270e-01 -8.78135383e-01 -8.72559071e-01 6.07241809e-01
-9.37287688e-01 -9.04781282e-01 -2.29047984e-01 5.60844243e-01
-4.27367508e-01 7.68405974e-01 -1.02658129e+00 -4.35283422e-01
-4.03085232e-01 -1.15738332e+00 6.37001634e-01 5.59339404e-01
1.20746277e-01 -9.40690577e-01 -6.25476912e-02 2.30485231e-01
7.70279109e-01 3.39827120e-01 1.15381610e+00 1.01698041e-01
-5.59089243e-01 1.15802117e-01 -7.07922399e-01 2.10852399e-01
2.62912095e-01 3.40871066e-02 -1.26610863e+00 -4.04392898e-01
1.15654528e-01 -4.09628451e-01 9.15474415e-01 1.03707135e+00
1.70215666e+00 -1.48295769e-02 -1.23250335e-01 1.02014172e+00
1.54971993e+00 2.50470430e-01 8.96360099e-01 2.86152303e-01
3.43504190e-01 3.31267029e-01 -2.55396334e-03 2.74200201e-01
-2.04435512e-01 3.84209543e-01 4.82254028e-01 -3.95708054e-01
-6.21643126e-01 -1.07927583e-01 -1.11400135e-01 4.62163955e-01
-1.53456470e-02 -7.96946213e-02 -6.42596066e-01 6.42909586e-01
-1.15741122e+00 -6.79443181e-01 1.19126454e-01 2.03619504e+00
9.96379256e-01 -2.11508408e-01 -5.45144267e-02 -3.54500264e-02
6.27192318e-01 5.73814213e-02 -6.24060392e-01 1.45307705e-01
-5.09101897e-02 5.02100408e-01 6.00650787e-01 5.12689888e-01
-9.61154878e-01 5.85347772e-01 7.67680311e+00 5.71892560e-01
-1.75826299e+00 -7.58728758e-02 8.19864631e-01 -1.88128635e-01
-1.35847539e-01 -5.17374754e-01 -5.37800133e-01 5.24077117e-01
8.25765789e-01 -1.42886445e-01 3.01599354e-01 3.85928005e-01
3.58653426e-01 -1.17899608e-02 -8.96208882e-01 8.52812886e-01
-4.38419431e-01 -1.61385286e+00 6.95081875e-02 3.13798666e-01
6.87356412e-01 1.37089849e-01 3.32541823e-01 -3.54002774e-01
2.88828462e-01 -1.31169295e+00 2.17867002e-01 2.44375274e-01
1.35337317e+00 -4.29078609e-01 6.83226049e-01 2.54487395e-01
-8.35290194e-01 -1.04484804e-01 -8.03094268e-01 -6.12378120e-03
1.14643402e-01 7.62292981e-01 -6.98095441e-01 1.30842283e-01
8.08852553e-01 3.65796447e-01 -9.29583013e-02 1.21746075e+00
2.14124337e-01 3.23776603e-01 -2.36093536e-01 1.75699607e-01
9.02148411e-02 -1.13073997e-01 3.40109110e-01 1.03030562e+00
6.63869917e-01 3.03470969e-01 -4.96074110e-01 1.20177042e+00
-1.58787280e-01 -3.69940013e-01 -4.03272033e-01 -1.42539879e-02
7.63381660e-01 1.37187231e+00 -6.57074511e-01 -1.26293115e-03
-4.01697636e-01 4.48382914e-01 3.96902502e-01 4.73156422e-01
-6.75374269e-01 -2.15050191e-01 8.25981796e-01 3.76504928e-01
3.34529668e-01 -3.53294611e-01 -3.51753801e-01 -9.00693953e-01
-3.01343322e-01 -3.40967208e-01 7.11628050e-02 -7.36527622e-01
-1.12808454e+00 4.46299165e-01 -2.60243684e-01 -1.05905068e+00
3.97570908e-01 -7.93933630e-01 -8.64413977e-01 9.82759058e-01
-2.04349542e+00 -6.88992918e-01 -5.05950093e-01 2.23830074e-01
3.22299510e-01 1.64504603e-01 6.48402452e-01 4.22196448e-01
-6.03687704e-01 2.37212345e-01 1.10184707e-01 1.21954326e-02
5.30914485e-01 -1.00281608e+00 1.54680014e-01 9.78698492e-01
-3.47482920e-01 6.13752425e-01 7.81576037e-01 -1.96184486e-01
-1.14747834e+00 -1.09292102e+00 1.19192608e-01 -9.77270827e-02
7.05373228e-01 2.59096116e-01 -9.92949307e-01 3.40609729e-01
2.55156875e-01 3.64309609e-01 7.82196045e-01 -9.39945161e-01
-1.68394491e-01 3.07166390e-03 -1.32632470e+00 7.37384200e-01
5.86895585e-01 -3.78108621e-01 -4.77590263e-02 -6.00086078e-02
6.25873744e-01 -5.52576780e-01 -1.10649967e+00 3.90712738e-01
7.67274678e-01 -9.93144810e-01 1.28542209e+00 -2.03177944e-01
1.06813896e+00 -3.48697901e-01 6.26537427e-02 -1.43455374e+00
-4.88149762e-01 -3.67026716e-01 -6.22950979e-02 6.59195840e-01
2.68823296e-01 -6.83839679e-01 1.08197141e+00 4.01082635e-01
-2.70488113e-01 -8.79668593e-01 -1.12376678e+00 -4.92901266e-01
5.67386329e-01 6.22065440e-02 5.00946045e-01 6.90423369e-01
-1.08220175e-01 -2.24868193e-01 -1.33868828e-01 3.06965172e-01
9.09820974e-01 -4.64804545e-02 2.99482942e-01 -8.70893359e-01
-9.20235142e-02 -6.25749171e-01 -3.63805234e-01 -1.23120809e+00
4.79990020e-02 -4.75784659e-01 4.83009256e-02 -1.38455260e+00
2.71363556e-01 -3.53205383e-01 -2.19987318e-01 7.68034011e-02
-6.59907609e-02 5.48649013e-01 -3.08498293e-01 3.47645700e-01
1.71883881e-01 2.09569886e-01 1.96279776e+00 -7.60689005e-02
-4.89781275e-02 -2.11874574e-01 -9.41370547e-01 3.49917799e-01
8.70445073e-01 3.74625623e-02 -4.60197121e-01 -8.65203023e-01
-4.04753327e-01 2.31754243e-01 5.52738726e-01 -1.28607905e+00
4.20139283e-01 -1.35680944e-01 8.34814668e-01 -1.47537127e-01
3.05620611e-01 -7.62003481e-01 1.02795295e-01 5.41174293e-01
-4.80194777e-01 -6.13806903e-01 3.17103744e-01 3.43161643e-01
6.58233911e-02 -1.14785455e-01 1.65495121e+00 -3.25659275e-01
-5.57675719e-01 4.23326463e-01 -4.70685869e-01 -2.70967126e-01
9.55918014e-01 -3.32965314e-01 -8.45049739e-01 -1.56646281e-01
-5.81375241e-01 -1.82342976e-01 5.72193086e-01 -3.02903771e-01
7.21444130e-01 -1.04650390e+00 -4.88162071e-01 2.74202704e-01
-2.74453759e-01 1.56267747e-01 7.37767339e-01 9.11994040e-01
-1.13420701e+00 3.87593210e-01 -6.80880129e-01 -5.57600319e-01
-9.59058404e-01 3.53041291e-01 8.97293091e-01 7.65978917e-02
-8.70553374e-01 9.44161594e-01 3.30688238e-01 -1.43729791e-01
5.09088598e-02 -2.72069395e-01 -3.33171368e-01 -6.93949580e-01
8.39502871e-01 2.67984092e-01 -2.00657576e-01 -2.08615825e-01
3.41555178e-02 7.42028475e-01 1.82095747e-02 1.18972972e-01
1.34080803e+00 -2.85116792e-01 3.70821394e-02 -4.31784317e-02
1.17995322e+00 -2.50851870e-01 -1.93219674e+00 -1.00274161e-01
-6.53219044e-01 -6.52583480e-01 4.22533900e-01 -5.56424320e-01
-1.33753967e+00 1.07727146e+00 5.27459204e-01 1.31541356e-01
9.98717964e-01 -3.68655212e-02 1.00386119e+00 -1.26505028e-02
2.31539398e-01 -7.02746451e-01 3.17840219e-01 -5.74620925e-02
4.56294030e-01 -1.02466702e+00 9.60762948e-02 -5.09515643e-01
-2.03910708e-01 1.24601364e+00 7.47416079e-01 -8.48954096e-02
5.67767859e-01 7.49267280e-01 3.64133596e-01 -2.37630323e-01
-6.27255321e-01 9.23940390e-02 -3.41923624e-01 1.02126038e+00
4.61638182e-01 -1.48496300e-01 -6.05932698e-02 3.72322232e-01
-1.03016831e-01 2.39228144e-01 8.87283027e-01 6.71117485e-01
-6.66247666e-01 -5.48468411e-01 -7.73738176e-02 4.91703451e-01
-8.41370463e-01 1.54054984e-01 9.59971398e-02 6.76738262e-01
-1.37289122e-01 5.67533374e-01 6.21228218e-02 -1.02813922e-01
1.11729436e-01 -4.56698895e-01 7.47653544e-01 -2.28193119e-01
9.03744921e-02 1.32685199e-01 -3.26466769e-01 -3.32560122e-01
-4.37090367e-01 -2.77438104e-01 -1.15940988e+00 -4.57313746e-01
-1.94990113e-01 -4.33076397e-02 5.28116524e-01 7.44818151e-01
2.19383672e-01 7.72237420e-01 5.55741966e-01 -9.88839686e-01
-4.05641526e-01 -7.89196432e-01 -7.44480550e-01 -1.26564398e-01
6.27625346e-01 -4.60567147e-01 -7.11519599e-01 3.44232112e-01] | [12.91319465637207, -2.675550699234009] |
29dece8c-3848-4af5-9888-2dc94a9d9dfa | automatic-graph-partitioning-for-very-large | 2103.16063 | null | https://arxiv.org/abs/2103.16063v1 | https://arxiv.org/pdf/2103.16063v1.pdf | Automatic Graph Partitioning for Very Large-scale Deep Learning | This work proposes RaNNC (Rapid Neural Network Connector) as middleware for automatic hybrid parallelism. In recent deep learning research, as exemplified by T5 and GPT-3, the size of neural network models continues to grow. Since such models do not fit into the memory of accelerator devices, they need to be partitioned by model parallelism techniques. Moreover, to accelerate training for huge training data, we need a combination of model and data parallelisms, i.e., hybrid parallelism. Given a model description for PyTorch without any specification for model parallelism, RaNNC automatically partitions the model into a set of subcomponents so that (1) each subcomponent fits a device memory and (2) a high training throughput for pipeline parallelism is achieved by balancing the computation times of the subcomponents. In our experiments, we compared RaNNC with two popular frameworks, Megatron-LM (hybrid parallelism) and GPipe (originally proposed for model parallelism, but a version allowing hybrid parallelism also exists), for training models with increasingly greater numbers of parameters. In the pre-training of enlarged BERT models, RaNNC successfully trained models five times larger than those Megatron-LM could, and RaNNC's training throughputs were comparable to Megatron-LM's when pre-training the same models. RaNNC also achieved better training throughputs than GPipe on both the enlarged BERT model pre-training (GPipe with hybrid parallelism) and the enlarged ResNet models (GPipe with model parallelism) in all of the settings we tried. These results are remarkable, since RaNNC automatically partitions models without any modification to their descriptions; Megatron-LM and GPipe require users to manually rewrite the models' descriptions. | ['Kentaro Torisawa', 'Toshihiro Hanawa', 'Kenjiro Taura', 'Masahiro Tanaka'] | 2021-03-30 | null | null | null | null | ['graph-partitioning'] | ['graphs'] | [-3.50261331e-01 8.12723301e-03 -1.08765624e-01 -4.01175112e-01
-9.91391912e-02 -4.20380622e-01 4.16539967e-01 -1.83795229e-01
-8.46285999e-01 1.93240792e-01 -3.77727956e-01 -7.61190832e-01
-4.24414054e-02 -1.02765119e+00 -6.85259104e-01 -4.24014300e-01
-6.44748136e-02 9.51504171e-01 8.05345714e-01 -2.57187635e-01
-1.81989282e-01 8.10184300e-01 -1.72771025e+00 5.62269270e-01
4.17309463e-01 6.67062163e-01 3.14696580e-01 9.38352346e-01
-2.43998870e-01 8.30480933e-01 -5.54961443e-01 -1.43289208e-01
5.15384078e-01 -5.70141189e-02 -8.58970881e-01 -4.81420785e-01
4.04474229e-01 -2.89788932e-01 -2.70010114e-01 6.69241607e-01
5.54059267e-01 7.00041801e-02 1.45920470e-01 -1.31947446e+00
3.54389548e-01 9.94571924e-01 -4.43587422e-01 3.04356962e-01
-3.77837867e-01 2.93535650e-01 4.96111006e-01 -5.80485642e-01
4.28461164e-01 1.19828308e+00 8.34640443e-01 8.46959472e-01
-1.22660816e+00 -8.58058035e-01 -1.13646701e-01 4.35157567e-02
-1.22872376e+00 -6.06101990e-01 5.59353717e-02 -2.91408241e-01
1.62532127e+00 4.59536046e-01 8.09618890e-01 9.07570839e-01
2.85392791e-01 6.17439687e-01 7.96152115e-01 -4.78186131e-01
3.77803832e-01 2.12064564e-01 6.99029982e-01 5.90949476e-01
-3.22331884e-03 2.39704385e-01 -2.75230020e-01 -1.17830902e-01
9.91732597e-01 -2.57313132e-01 -3.71710211e-02 8.92643183e-02
-1.00965953e+00 5.99764585e-01 5.23181379e-01 4.79292840e-01
-2.51253545e-01 3.06332856e-01 8.96803796e-01 3.31513256e-01
-1.32236943e-01 4.63809758e-01 -8.31164718e-01 -3.69203299e-01
-1.10963619e+00 2.10446343e-01 1.00263703e+00 1.21962953e+00
8.69411886e-01 2.02993333e-01 1.56265929e-01 7.23676920e-01
7.81408921e-02 2.55373064e-02 1.02715349e+00 -6.52803659e-01
4.85110790e-01 6.13366842e-01 -4.76712912e-01 -3.02346885e-01
-9.64312136e-01 -5.96871912e-01 -1.24419451e+00 1.64341286e-01
3.33344668e-01 -3.31797272e-01 -1.05527675e+00 1.48544359e+00
1.78625390e-01 2.53865361e-01 2.86026001e-01 6.86188638e-01
1.03262568e+00 8.25065970e-01 2.51589090e-01 3.52680594e-01
1.55449677e+00 -1.37352443e+00 -2.24630963e-02 -4.74587947e-01
1.31803513e+00 -6.05008841e-01 1.00805247e+00 4.40657228e-01
-1.17375576e+00 -9.63256836e-01 -1.20151389e+00 -1.45427004e-01
-3.41196030e-01 8.58747661e-02 8.65555525e-01 6.80577397e-01
-1.63946521e+00 8.35465848e-01 -1.37391388e+00 -1.54473200e-01
1.50323570e-01 9.23145473e-01 -3.38701248e-01 5.79153523e-02
-9.41369414e-01 8.36247265e-01 9.87822950e-01 -3.19099147e-03
-6.66423380e-01 -8.40434492e-01 -6.31434083e-01 3.93932790e-01
-1.24981642e-01 -7.50133753e-01 1.47181892e+00 -1.03380573e+00
-1.56764805e+00 6.76534951e-01 1.72826812e-01 -8.31148505e-01
3.76181394e-01 1.34099528e-01 -3.68904948e-01 -2.15948924e-01
-5.03440738e-01 1.22529495e+00 4.31803137e-01 -6.77536368e-01
-5.98634005e-01 -1.11135028e-01 3.03698868e-01 -2.62223426e-02
-4.22920972e-01 1.65263250e-01 -7.82964468e-01 -9.38017815e-02
-3.15268710e-02 -1.17597353e+00 -4.91008282e-01 -4.35840815e-01
-1.97641984e-01 -3.71317416e-01 8.58188093e-01 -1.99910313e-01
1.01869857e+00 -2.13003898e+00 -4.47085761e-02 1.84900299e-01
4.10836339e-01 7.49280870e-01 -2.75210351e-01 1.89201131e-01
-3.87203902e-01 2.82891039e-02 1.29263818e-01 -5.56230426e-01
-3.92696895e-02 6.73152030e-01 -1.64682075e-01 -2.74967607e-02
-9.88947079e-02 8.02629471e-01 -5.44092238e-01 -2.49060109e-01
1.27181739e-01 4.46423084e-01 -8.32677841e-01 4.53879349e-02
-2.46845111e-01 2.82549039e-02 -2.23210156e-01 7.50297457e-02
8.53505731e-01 -3.39617759e-01 1.51300758e-01 -2.09015518e-01
-3.13531965e-01 3.75223279e-01 -8.77748132e-01 1.38240635e+00
-6.79252684e-01 7.16595292e-01 -3.12372334e-02 -7.56811500e-01
7.99956560e-01 4.12207127e-01 2.35144794e-01 -6.44590914e-01
1.63030043e-01 1.98440269e-01 3.01566780e-01 -1.41312823e-01
6.96708858e-01 -5.58116240e-04 2.36378923e-01 4.92311656e-01
3.48503947e-01 1.15857236e-01 3.93795103e-01 -4.96299230e-02
1.35982502e+00 4.49710526e-02 -1.41967684e-01 -3.36497754e-01
3.15364659e-01 1.92287356e-01 5.80513895e-01 8.03037107e-01
2.08615080e-01 3.69984031e-01 5.02765417e-01 -7.52140045e-01
-1.26385128e+00 -7.08893359e-01 -3.18293452e-01 1.28939080e+00
-4.21099961e-01 -8.97953391e-01 -1.05004609e+00 -5.66960633e-01
-4.55927819e-01 7.59395063e-01 -1.83108196e-01 3.47810611e-02
-8.78650129e-01 -1.08401346e+00 1.02172768e+00 7.25692034e-01
8.37127686e-01 -1.23904467e+00 -9.67698693e-01 2.30521917e-01
4.69946951e-01 -1.07285380e+00 -9.65225697e-02 7.92646945e-01
-1.31610250e+00 -8.61914456e-01 -2.78782874e-01 -7.26512849e-01
6.38993204e-01 3.76234911e-02 1.31460834e+00 5.00617802e-01
-2.60698609e-02 -2.50548869e-01 -1.70378909e-01 -2.61433095e-01
-6.16312265e-01 5.56713343e-01 4.21698876e-02 -6.49383783e-01
3.88308436e-01 -7.07069814e-01 -2.35503495e-01 3.58844429e-01
-1.11046422e+00 9.64272261e-01 5.48625767e-01 7.65765607e-01
3.72659594e-01 1.73196509e-01 3.05233836e-01 -1.13253009e+00
2.53350049e-01 -3.53628516e-01 -9.48630393e-01 -4.06876840e-02
-5.30541658e-01 2.67606020e-01 1.10411990e+00 -6.23788953e-01
-7.10452974e-01 8.85172114e-02 -8.42472613e-01 -7.18262076e-01
-2.64057010e-01 5.97184181e-01 -1.06942942e-02 -6.36257827e-02
9.86113369e-01 -3.76660936e-02 -1.43000469e-01 -6.73057914e-01
3.69312800e-02 4.88237381e-01 2.74826884e-01 -5.73695540e-01
4.00737286e-01 -1.56104982e-01 -1.97611824e-01 -6.52279615e-01
-3.27945530e-01 -3.09585571e-01 -6.31470025e-01 1.93516254e-01
6.76304400e-01 -1.04157495e+00 -8.53852987e-01 5.97736001e-01
-1.24708712e+00 -9.83894348e-01 -3.06200236e-01 5.51478803e-01
-1.65547237e-01 -1.40940085e-01 -1.10134268e+00 -2.64036596e-01
-5.87349534e-01 -1.54279947e+00 7.62847722e-01 3.78373504e-01
-8.00954029e-02 -1.13633156e+00 -2.05150563e-02 2.26748213e-02
7.18953371e-01 -2.68323660e-01 1.05854082e+00 -8.18380892e-01
-3.34220737e-01 -1.67462617e-01 -2.93686122e-01 4.20421958e-01
-4.82363492e-01 6.38018129e-03 -1.25633359e+00 -4.21333492e-01
4.73940261e-02 -1.62060589e-01 7.17065215e-01 9.15736184e-02
1.38705814e+00 -2.38816559e-01 -3.98743540e-01 8.18343103e-01
1.35158050e+00 2.46626198e-01 8.68728638e-01 5.52507818e-01
7.60005772e-01 1.59910217e-01 -1.26566246e-01 1.19946413e-01
1.88823283e-01 8.70312870e-01 2.61459857e-01 -3.46386969e-01
-1.11577339e-01 1.03586957e-01 5.19398093e-01 1.16767740e+00
-7.44751245e-02 -1.26866847e-01 -1.25022519e+00 -4.40034047e-02
-1.77411997e+00 -6.97423458e-01 -5.30495703e-01 2.13831258e+00
8.55109155e-01 4.91539687e-01 1.27012730e-01 3.47502977e-02
3.01659137e-01 -2.01389745e-01 -5.21566093e-01 -8.69222462e-01
1.37889981e-01 6.06185436e-01 7.61010170e-01 3.04558456e-01
-7.73965836e-01 1.17622650e+00 6.34813881e+00 1.00321102e+00
-1.34389067e+00 3.27310771e-01 6.34495676e-01 -1.10454150e-01
1.02047466e-01 1.00080609e-01 -1.39542735e+00 3.01522762e-01
1.83819973e+00 1.04053043e-01 3.11705291e-01 1.10857689e+00
-8.44980106e-02 -6.56598657e-02 -1.41355693e+00 8.17720950e-01
-5.53243101e-01 -1.57329798e+00 -1.66113511e-01 1.05352826e-01
4.78726834e-01 7.34193385e-01 -4.42169636e-01 9.32303846e-01
5.65334260e-01 -9.42888319e-01 6.62931979e-01 1.80226102e-01
7.17924714e-01 -8.60629022e-01 9.26546991e-01 8.35695803e-01
-1.00619268e+00 3.36012095e-02 -7.52202451e-01 -2.04252720e-01
-1.47059783e-01 5.48046350e-01 -8.95211816e-01 4.59827513e-01
6.49219751e-01 4.36690062e-01 -8.27408552e-01 1.04420042e+00
-5.31495363e-02 8.38161290e-01 -6.03096068e-01 1.82059348e-01
3.60368371e-01 1.05965622e-02 -2.67192665e-02 1.33743203e+00
1.46637335e-01 -4.44447286e-02 6.23209076e-03 7.00358629e-01
-8.98863748e-02 -1.38209760e-01 -1.22675754e-01 2.77688056e-01
4.15849864e-01 1.49347413e+00 -8.28033626e-01 -6.80788219e-01
-3.77096772e-01 6.63430631e-01 5.43382704e-01 6.94725290e-02
-8.47788036e-01 -1.75951973e-01 4.21587348e-01 1.35838285e-01
-4.09211330e-02 -2.34223545e-01 -3.02354068e-01 -1.11496210e+00
-3.92719060e-01 -9.57334697e-01 3.24341595e-01 -7.37125576e-01
-8.56047690e-01 1.27556264e+00 2.96031445e-01 -8.85558367e-01
-4.01894748e-01 -7.88665593e-01 -7.15037167e-01 9.76750731e-01
-9.62453783e-01 -9.67378259e-01 -2.76445597e-01 3.99747491e-01
5.48559189e-01 -2.67812252e-01 1.27835906e+00 4.61197078e-01
-8.85300219e-01 7.28228390e-01 1.33166108e-02 7.95911029e-02
2.99521178e-01 -9.78987634e-01 1.07437789e+00 6.12489343e-01
1.00234240e-01 6.07291400e-01 4.43798423e-01 -2.63212174e-01
-1.40015852e+00 -1.19764745e+00 6.96138442e-01 -7.87331387e-02
5.38133085e-01 -6.14602149e-01 -1.09676075e+00 9.82686102e-01
2.09230423e-01 5.39141260e-02 6.80232048e-01 3.99810046e-01
-2.29890868e-01 -9.39974114e-02 -8.48410606e-01 5.65441668e-01
8.31215620e-01 -1.10842988e-01 -2.56574512e-01 5.74487448e-01
7.24105656e-01 -9.52673972e-01 -9.39021170e-01 3.20429057e-01
4.97963816e-01 -1.20820034e+00 7.37100244e-01 -3.45118582e-01
2.33129025e-01 5.48851974e-02 1.14013292e-01 -1.02007031e+00
-3.08108121e-01 -3.01443249e-01 -1.23954624e-01 8.91108751e-01
3.77416611e-01 -7.73369193e-01 1.07719374e+00 7.65470862e-01
-5.83747923e-01 -8.97061586e-01 -8.04464519e-01 -7.90789008e-01
1.58863589e-01 -7.55379915e-01 8.56903613e-01 8.52500975e-01
-2.63767630e-01 5.46980083e-01 -2.11236943e-02 2.12925021e-02
1.11178301e-01 -3.88991714e-01 9.62296665e-01 -1.15607035e+00
-7.78865576e-01 -5.56203187e-01 -3.83462220e-01 -1.19018614e+00
1.35539666e-01 -1.09116316e+00 -1.86232343e-01 -1.32219326e+00
-2.13274453e-02 -1.15720654e+00 1.90463811e-02 1.18552208e+00
3.38560998e-01 8.27456489e-02 4.43053722e-01 4.61678147e-01
-2.21856341e-01 1.04339890e-01 8.50950301e-01 1.63380489e-01
-4.93602365e-01 2.21003711e-01 -2.76340753e-01 8.10350955e-01
8.97345126e-01 -5.17761469e-01 -5.27798474e-01 -8.02511632e-01
2.76559085e-01 9.53435227e-02 3.35959911e-01 -1.73424327e+00
4.64947194e-01 3.17863464e-01 4.38665032e-01 -5.82432568e-01
3.14856678e-01 -7.76989341e-01 6.07122481e-01 4.78469849e-01
-1.45139620e-01 5.52819192e-01 9.16153133e-01 -2.91958690e-01
-3.80316190e-02 -6.24197900e-01 8.73874664e-01 -3.81589718e-02
-6.02476180e-01 3.21606427e-01 -5.92452109e-01 -3.18965226e-01
6.73114002e-01 -1.84065685e-01 -5.70709348e-01 3.90221030e-01
-8.25766027e-01 2.94231921e-01 3.28890115e-01 3.86657268e-01
3.84714812e-01 -1.09281027e+00 -2.86840439e-01 5.71518838e-01
-2.70671248e-01 6.21729791e-01 3.54414284e-01 8.93065035e-01
-9.81859028e-01 4.64573294e-01 -3.72261554e-01 -7.27955043e-01
-1.31562245e+00 3.53737026e-01 6.05905056e-01 -7.78107047e-01
-9.08289492e-01 8.04919839e-01 2.88067728e-01 -8.29161882e-01
1.38613045e-01 -6.22162104e-01 -1.02576740e-01 -2.86841869e-01
7.05212057e-01 1.67788222e-01 6.93701327e-01 -2.95550704e-01
-2.04984844e-01 -6.66069286e-03 -5.40808260e-01 8.42301846e-02
1.44278002e+00 6.18819654e-01 -1.60899922e-01 2.41932154e-01
1.12520456e+00 -5.68386197e-01 -1.12252307e+00 -4.99705411e-02
5.56903444e-02 1.55668959e-01 2.83329964e-01 -6.58877015e-01
-1.29765332e+00 9.42383945e-01 7.08979666e-01 1.84524674e-02
1.21878815e+00 -3.34164739e-01 7.34456837e-01 6.10088289e-01
6.27252638e-01 -7.12684870e-01 -3.84874254e-01 1.05481446e+00
4.86569375e-01 -6.97103739e-01 -1.85452461e-01 -1.77987441e-01
-2.61831015e-01 1.38215709e+00 1.04251373e+00 -1.08941413e-01
6.78169310e-01 7.44747758e-01 -1.00527890e-01 -1.94378749e-01
-1.16108978e+00 2.89019138e-01 8.18858668e-02 2.87650198e-01
4.24480885e-01 7.97408575e-05 1.44681819e-02 7.36661375e-01
-7.54407287e-01 7.49489143e-02 4.42459702e-01 8.83314312e-01
-3.86621505e-01 -1.36219192e+00 -2.85678893e-01 6.36506617e-01
-7.36009628e-02 -3.46980304e-01 2.94168741e-01 1.03226209e+00
3.89989585e-01 5.33552766e-01 4.09793884e-01 -7.08308816e-01
2.80544221e-01 1.31448777e-02 4.22312140e-01 -8.60093713e-01
-1.39104044e+00 3.51325050e-02 1.64677247e-01 -4.90017295e-01
1.73503116e-01 -1.09592766e-01 -1.64945519e+00 -7.19040513e-01
-2.63627112e-01 2.28107318e-01 9.02042747e-01 1.00048912e+00
3.80953610e-01 8.06520462e-01 1.19272597e-01 -1.30374980e+00
-4.40871418e-01 -1.04537213e+00 -4.00602609e-01 -2.98407763e-01
-2.12063536e-01 -1.82372332e-01 -2.99617779e-02 -1.68885782e-01] | [8.527721405029297, 3.1388299465179443] |
abfe1110-db8b-4a30-b795-3d7d62553ec4 | conception-multilingually-enhanced-human | null | null | https://aclanthology.org/2020.coling-main.291 | https://aclanthology.org/2020.coling-main.291.pdf | Conception: Multilingually-Enhanced, Human-Readable Concept Vector Representations | To date, the most successful word, word sense, and concept modelling techniques have used large corpora and knowledge resources to produce dense vector representations that capture semantic similarities in a relatively low-dimensional space. Most current approaches, however, suffer from a monolingual bias, with their strength depending on the amount of data available across languages. In this paper we address this issue and propose Conception, a novel technique for building language-independent vector representations of concepts which places multilinguality at its core while retaining explicit relationships between concepts. Our approach results in high-coverage representations that outperform the state of the art in multilingual and cross-lingual Semantic Word Similarity and Word Sense Disambiguation, proving particularly robust on low-resource languages. Conception {--} its software and the complete set of representations {--} is available at https://github.com/SapienzaNLP/conception. | ['Roberto Navigli', 'Simone Conia'] | 2020-12-01 | null | null | null | coling-2020-8 | ['word-similarity'] | ['natural-language-processing'] | [-2.74343520e-01 -2.94336230e-01 -5.17252028e-01 -2.32053652e-01
-8.62931192e-01 -8.57733846e-01 8.83236408e-01 7.47854054e-01
-6.27562225e-01 6.00000143e-01 6.17030084e-01 -3.29741180e-01
-1.27180800e-01 -7.03565180e-01 -1.79857612e-01 -1.94221616e-01
2.29296014e-01 6.77591205e-01 -7.65706673e-02 -7.83749461e-01
3.38555366e-01 3.12440842e-02 -1.52643514e+00 1.65811628e-01
9.37557042e-01 3.86093527e-01 5.08093894e-01 1.10189252e-01
-7.55154788e-01 2.49217525e-01 -3.48628759e-01 -3.96710724e-01
-1.06571920e-01 -8.93692523e-02 -9.72588062e-01 -5.20678997e-01
3.70102495e-01 5.49036920e-01 -2.04232544e-01 1.12550437e+00
4.94046509e-01 2.36938059e-01 6.18909359e-01 -9.68902946e-01
-1.07977223e+00 8.91194522e-01 -2.49434188e-01 4.01024312e-01
5.88699281e-01 -4.59116966e-01 1.48268294e+00 -1.15631795e+00
9.76196945e-01 1.32120478e+00 7.59441316e-01 4.11605537e-01
-1.18377769e+00 -7.21246243e-01 1.77366272e-01 3.88750494e-01
-1.66979480e+00 -4.15127128e-01 6.39237523e-01 -5.65701127e-01
1.64720368e+00 -5.46403043e-02 5.32910228e-01 1.25727630e+00
-9.19366162e-03 6.18866801e-01 8.55421126e-01 -7.53149927e-01
5.70807867e-02 1.67181879e-01 4.56174433e-01 3.91243815e-01
6.71115696e-01 -2.61604607e-01 -3.58689874e-01 -3.39225858e-01
7.18388081e-01 -5.36432490e-02 -9.68140215e-02 -5.17891824e-01
-1.34712815e+00 1.15510726e+00 3.07543963e-01 1.21367812e+00
-4.71257091e-01 -1.89573139e-01 6.72527552e-01 2.98716217e-01
6.83944702e-01 8.37515533e-01 -5.95030963e-01 -1.43469080e-01
-8.70104730e-01 3.10731083e-01 6.15500093e-01 8.57337058e-01
9.85056162e-01 1.89668238e-01 1.98207527e-01 1.43056571e+00
2.64477611e-01 4.74841237e-01 1.11236012e+00 -5.08012533e-01
2.62640625e-01 7.46586680e-01 -1.87072173e-01 -1.18587458e+00
-4.72711891e-01 -4.02087241e-01 -4.28067416e-01 -3.60677749e-01
-7.72746876e-02 6.91536590e-02 -6.86049342e-01 1.79006290e+00
1.95215315e-01 1.80614322e-01 4.67974275e-01 7.54297733e-01
9.41424489e-01 4.70846206e-01 4.97744441e-01 1.38753891e-01
1.55619013e+00 -5.45741022e-01 -7.01359630e-01 -5.91423035e-01
9.56676006e-01 -1.02659738e+00 1.02822244e+00 -2.24526674e-02
-7.10398853e-01 -3.98015738e-01 -1.07238233e+00 -2.05679104e-01
-9.89406407e-01 -1.25677064e-01 1.17717016e+00 5.37993014e-01
-1.12142301e+00 2.90893137e-01 -4.95280564e-01 -9.22019780e-01
2.13307500e-01 3.90407583e-03 -7.59126246e-01 -3.84545356e-01
-1.69147062e+00 1.32917714e+00 7.38601089e-01 -5.76476216e-01
-2.93018341e-01 -8.43034685e-01 -1.33279574e+00 -1.10272914e-01
2.63008475e-01 -5.32070100e-01 9.62900579e-01 -1.02069247e+00
-6.72617018e-01 1.10554540e+00 -1.59431070e-01 -4.20160085e-01
-2.31155396e-01 -3.11501473e-01 -8.20252776e-01 -1.63708270e-01
6.32193089e-01 4.99118090e-01 1.57754615e-01 -1.19965589e+00
-5.11422873e-01 -4.49530721e-01 5.18950373e-02 3.77561688e-01
-4.99476969e-01 2.35502288e-01 -3.34535807e-01 -9.23684657e-01
-6.71411604e-02 -6.01892114e-01 -4.17867959e-01 -4.51305419e-01
1.32764623e-01 -5.88577926e-01 1.87899962e-01 -6.41375065e-01
1.26370144e+00 -1.92290294e+00 1.45470604e-01 2.33961232e-02
4.77804504e-02 5.46956241e-01 -4.28652972e-01 9.40134227e-01
-2.60303885e-01 1.13128774e-01 -9.73811150e-02 -1.37864292e-01
6.08711541e-02 5.58014274e-01 -2.06293643e-01 4.67520684e-01
2.95407753e-02 8.38619411e-01 -1.36911237e+00 -5.02460063e-01
5.65665543e-01 8.87330353e-01 -4.77954954e-01 -2.75269628e-01
2.76045222e-02 -1.53910920e-01 -4.96401250e-01 5.86285412e-01
4.41651434e-01 6.91600190e-03 8.57840478e-01 -1.39406487e-01
-1.92812294e-01 4.48715985e-01 -1.27596879e+00 2.34924150e+00
-6.54023707e-01 6.27917051e-01 -1.41196296e-01 -1.14570069e+00
9.98812616e-01 3.89041483e-01 6.91168189e-01 -8.28628719e-01
1.67502657e-01 4.31255966e-01 -2.27658182e-01 -2.03307211e-01
9.90013957e-01 -3.02649230e-01 -3.85543913e-01 1.50247604e-01
6.16617382e-01 -5.93542233e-02 2.70221949e-01 3.27669770e-01
6.31231964e-01 1.73718527e-01 9.01614964e-01 -6.94263041e-01
4.63079542e-01 3.11040372e-01 5.86427510e-01 2.90239662e-01
2.48692445e-02 3.12685728e-01 -6.03804961e-02 -3.62157583e-01
-1.00294709e+00 -1.12820387e+00 -4.03273195e-01 1.20523858e+00
5.72985373e-02 -9.23443854e-01 -3.88007909e-01 -2.43020311e-01
2.56571144e-01 9.44233954e-01 -5.19819438e-01 7.79803842e-02
-3.30741793e-01 -5.20819068e-01 6.78452611e-01 4.26189601e-01
-2.68889010e-01 -1.03033245e+00 -1.56379640e-01 4.14176971e-01
-2.59514868e-01 -9.66403604e-01 -3.51507664e-02 -1.09301452e-02
-4.61764246e-01 -1.10077441e+00 -5.08253098e-01 -8.62528324e-01
1.49704665e-01 3.69830459e-01 1.59340334e+00 -8.65202248e-02
-5.10176599e-01 4.05273438e-01 -6.44673467e-01 -4.23883408e-01
-1.59379527e-01 2.09446818e-01 4.23610121e-01 -3.82060170e-01
1.02770448e+00 -4.42993551e-01 -2.08396047e-01 -2.27232069e-01
-1.01758289e+00 -4.68779117e-01 3.18032712e-01 6.95940256e-01
7.52782881e-01 -2.80289888e-01 6.73629701e-01 -8.68529141e-01
9.15970206e-01 -9.27434742e-01 -1.43011957e-01 2.37167105e-01
-8.76921594e-01 1.46184132e-01 2.50346571e-01 -2.21953139e-01
-7.57677674e-01 -3.47383499e-01 -2.62604207e-01 -3.52764696e-01
-3.02443862e-01 8.58684361e-01 6.41583130e-02 2.66415179e-01
8.80257130e-01 1.44383684e-01 -2.32297987e-01 -7.31999993e-01
9.47485149e-01 6.93817258e-01 2.73996651e-01 -7.70686746e-01
4.26018894e-01 2.11251259e-01 -5.65753937e-01 -1.03392625e+00
-7.38172293e-01 -1.10116529e+00 -8.75539720e-01 2.09316671e-01
7.28785336e-01 -1.27247739e+00 5.82489297e-02 -1.27108157e-01
-9.42198873e-01 2.57010043e-01 -3.13501000e-01 6.57691121e-01
-2.97104299e-01 3.50541890e-01 -2.01842800e-01 -4.31028873e-01
-5.02449870e-01 -7.40391910e-01 8.82170379e-01 -7.48551358e-03
-7.67837942e-01 -1.54305017e+00 6.26367986e-01 1.59660652e-01
6.16339684e-01 1.30498067e-01 8.88152599e-01 -1.05914104e+00
2.47298598e-01 -1.84823573e-01 -1.39005883e-02 2.01052874e-01
3.10193837e-01 -3.15385371e-01 -7.30419755e-01 -4.38220531e-01
-5.78682840e-01 -1.81108624e-01 7.74737298e-01 1.17824905e-01
4.12765324e-01 -9.70555916e-02 -5.43933213e-01 3.59071493e-01
1.89690828e+00 -1.84495449e-01 2.70564348e-01 5.90278149e-01
7.04804718e-01 4.88393784e-01 5.76321483e-01 3.43893945e-01
7.01536536e-01 7.20923245e-01 -1.90717503e-01 -1.54700642e-02
-6.76082149e-02 -2.00199872e-01 8.98493752e-02 1.27080882e+00
-3.28424983e-02 2.63861343e-02 -1.34138477e+00 1.24107587e+00
-1.78213167e+00 -1.08651686e+00 -3.15345377e-02 2.15645432e+00
8.38323832e-01 -3.58387083e-01 -1.01745605e-01 -1.91588663e-02
4.52399045e-01 3.82536650e-01 -5.93912713e-02 -5.83593249e-01
-3.50168407e-01 6.24653339e-01 4.24880415e-01 7.76504934e-01
-8.89996231e-01 1.57547116e+00 6.06789875e+00 9.41501439e-01
-9.62228537e-01 4.92163450e-01 -2.78654158e-01 -1.00931838e-01
-8.87623489e-01 -6.52712723e-03 -6.30841434e-01 1.20994836e-01
9.28731143e-01 -6.82599962e-01 1.57660902e-01 8.21784377e-01
-2.40962788e-01 3.34591627e-01 -6.50631189e-01 1.08956409e+00
4.42764789e-01 -1.43814039e+00 3.93471330e-01 -1.80829510e-01
5.63894749e-01 5.94590962e-01 -2.89842516e-01 5.99818230e-01
6.48226500e-01 -1.00733662e+00 4.73460078e-01 2.93053001e-01
8.72083664e-01 -1.00152433e+00 6.19129777e-01 -1.47024438e-01
-1.46026397e+00 2.25905418e-01 -6.69198513e-01 -4.77308296e-02
1.23887435e-01 5.62800109e-01 -5.78394949e-01 9.72055018e-01
4.85251307e-01 1.08326173e+00 -5.58776915e-01 5.36666691e-01
-1.51256412e-01 2.57678866e-01 -1.20340253e-03 4.41496596e-02
4.30120289e-01 -1.61488339e-01 4.87386674e-01 1.74826527e+00
3.47115397e-01 -2.09070221e-01 4.99100447e-01 4.53236997e-01
8.84584785e-02 8.40996146e-01 -9.12906885e-01 -3.35368484e-01
8.27325344e-01 1.16458714e+00 -4.13621128e-01 -3.75181943e-01
-6.24221742e-01 9.10013020e-01 7.10812211e-01 2.37647846e-01
-2.54090190e-01 -4.05254036e-01 1.47591484e+00 -7.15739653e-02
5.05037196e-02 -5.17871797e-01 -1.19980566e-01 -1.50191164e+00
-1.86954126e-01 -8.65601897e-01 6.83151066e-01 -4.98324722e-01
-1.52605450e+00 7.37008393e-01 8.05894961e-04 -9.69677448e-01
-5.37991583e-01 -7.85907090e-01 -1.49041161e-01 1.03690732e+00
-1.52697885e+00 -1.27340388e+00 2.69559324e-01 8.50747287e-01
6.82501435e-01 -4.95981723e-01 1.54004419e+00 3.95023763e-01
-1.24595456e-01 5.76036036e-01 2.36739904e-01 9.51752216e-02
7.64514148e-01 -1.12309003e+00 5.98139405e-01 5.95641196e-01
6.97872996e-01 1.05423141e+00 6.31470859e-01 -6.30371034e-01
-1.30140936e+00 -7.44260192e-01 1.58106458e+00 -6.43025339e-01
1.22557986e+00 -3.73825401e-01 -1.00992465e+00 8.02595496e-01
4.04969662e-01 -2.38944098e-01 1.30777729e+00 8.46728861e-01
-9.47874844e-01 3.08240980e-01 -9.08734500e-01 5.04066527e-01
1.02177036e+00 -8.68619144e-01 -1.25387061e+00 3.90672445e-01
5.82222462e-01 7.73076937e-02 -1.15692949e+00 7.19535723e-02
4.45987284e-01 -4.97263581e-01 1.36038089e+00 -8.80576909e-01
1.69421047e-01 -1.32012248e-01 -7.21120834e-01 -1.65979075e+00
-5.28698981e-01 -6.52959347e-02 3.58720213e-01 1.27162433e+00
5.44078887e-01 -7.18974710e-01 1.21305361e-01 -7.85158798e-02
-2.54657734e-02 -4.16270882e-01 -9.41044033e-01 -8.58370483e-01
6.30741477e-01 -7.65088975e-01 6.30874336e-01 1.85454297e+00
4.75357771e-01 5.97383857e-01 -1.15003012e-01 -2.76285540e-02
4.58059222e-01 1.06390722e-01 1.64502919e-01 -1.56566226e+00
1.28476262e-01 -5.72128117e-01 -8.02097797e-01 -3.19780886e-01
6.52843893e-01 -1.50332868e+00 -3.35859239e-01 -1.96101308e+00
6.58864304e-02 -5.53330064e-01 -6.89333320e-01 7.66470373e-01
-2.91337043e-01 9.72300023e-02 1.62206545e-01 2.77310580e-01
-4.62536782e-01 3.51133108e-01 5.33439696e-01 -4.90547642e-02
2.52881236e-02 -8.04642498e-01 -1.08595610e+00 5.19734800e-01
9.20694411e-01 -3.92152697e-01 -4.87744302e-01 -7.59619892e-01
3.51257414e-01 -4.65312421e-01 2.58777291e-02 -8.05252254e-01
-3.08064278e-02 -1.75527513e-01 2.82826722e-01 -3.75756994e-02
4.42710876e-01 -5.46921372e-01 6.29371479e-02 1.79308876e-01
-2.22638145e-01 3.76977295e-01 4.38094169e-01 3.75610918e-01
-4.49791878e-01 -1.33030310e-01 4.64697957e-01 -4.70018804e-01
-1.32840550e+00 2.21218333e-01 -4.63832557e-01 3.49161714e-01
7.92173326e-01 1.14492916e-01 -1.96862116e-01 4.69980277e-02
-5.01438558e-01 7.01526403e-02 6.80086911e-01 1.24878192e+00
3.93349111e-01 -1.65150023e+00 -1.02073085e+00 1.64898291e-01
7.09576726e-01 -6.07427895e-01 1.43127412e-01 1.64164394e-01
-1.70047551e-01 7.51105666e-01 -4.16894674e-01 -2.05055118e-01
-1.02567542e+00 6.87432885e-01 1.12789154e-01 -6.40281215e-02
-5.80170751e-01 7.71461904e-01 -1.31572589e-01 -8.42003047e-01
-3.06547582e-01 1.96301177e-01 -7.01270044e-01 5.38686872e-01
5.61493695e-01 8.71665999e-02 8.92364979e-02 -1.29609060e+00
-7.67830491e-01 5.39823294e-01 -3.24173197e-02 -1.58633068e-01
1.36151767e+00 -4.21520323e-02 -9.92063954e-02 6.40108466e-01
1.20880687e+00 -2.29656082e-02 -2.12332785e-01 -5.77248037e-01
3.72569948e-01 -4.52445120e-01 1.77203327e-01 -5.91771185e-01
-7.67907739e-01 6.07895076e-01 5.79936266e-01 -1.56105220e-01
6.29676819e-01 3.00880551e-01 5.62910974e-01 2.88218498e-01
5.98700702e-01 -1.17319703e+00 -5.84228337e-01 8.19268405e-01
7.56742001e-01 -1.08464813e+00 1.11473594e-02 -1.99997440e-01
-8.02270293e-01 7.52919257e-01 2.37597182e-01 -2.78778166e-01
7.31085896e-01 4.26650569e-02 4.32307988e-01 -5.02043486e-01
-6.25375748e-01 -6.23086333e-01 4.08609509e-01 9.24795389e-01
1.08393383e+00 5.23258269e-01 -7.05180883e-01 6.68510258e-01
-3.83021384e-01 -2.46968165e-01 1.39181823e-01 7.92050540e-01
-4.22542751e-01 -1.65903461e+00 -1.23978846e-01 2.03330144e-01
-5.22689641e-01 -7.11902142e-01 -4.52881366e-01 8.42847168e-01
1.59111515e-01 9.75442410e-01 1.68391526e-01 -2.62568057e-01
3.66013080e-01 3.65990669e-01 2.07339555e-01 -8.20970476e-01
-6.66439056e-01 -1.07626416e-01 4.03006166e-01 -7.82689154e-01
-5.11033773e-01 -9.01555419e-01 -1.25773406e+00 -2.61393726e-01
9.54472572e-02 3.79270941e-01 7.00548947e-01 9.12725866e-01
5.01552582e-01 3.15603226e-01 9.37145054e-02 -7.60764897e-01
-1.51210278e-01 -1.05549324e+00 -6.60927176e-01 7.13377059e-01
-1.30777985e-01 -8.80135357e-01 -2.26390436e-02 -3.50343376e-01] | [10.722278594970703, 9.761646270751953] |
ff1b5434-b2d0-429d-96a6-788aef9e7a00 | a-knowledge-graph-embeddings-based-approach | 2201.09555 | null | https://arxiv.org/abs/2201.09555v3 | https://arxiv.org/pdf/2201.09555v3.pdf | A Knowledge Graph Embeddings based Approach for Author Name Disambiguation using Literals | Scholarly data is growing continuously containing information about the articles from a plethora of venues including conferences, journals, etc. Many initiatives have been taken to make scholarly data available as Knowledge Graphs (KGs). These efforts to standardize these data and make them accessible have also led to many challenges such as exploration of scholarly articles, ambiguous authors, etc. This study more specifically targets the problem of Author Name Disambiguation (AND) on Scholarly KGs and presents a novel framework, Literally Author Name Disambiguation (LAND), which utilizes Knowledge Graph Embeddings (KGEs) using multimodal literal information generated from these KGs. This framework is based on three components: 1) Multimodal KGEs, 2) A blocking procedure, and finally, 3) Hierarchical Agglomerative Clustering. Extensive experiments have been conducted on two newly created KGs: (i) KG containing information from Scientometrics Journal from 1978 onwards (OC-782K), and (ii) a KG extracted from a well-known benchmark for AND provided by AMiner (AMiner-534K). The results show that our proposed architecture outperforms our baselines of 8-14% in terms of the F1 score and shows competitive performances on a challenging benchmark such as AMiner. The code and the datasets are publicly available through Github: https://github.com/sntcristian/and-kge and Zenodo:https://doi.org/10.5281/zenodo.6309855 respectively. | ['Mehwish Alam', 'Harald Sack', 'Aldo Gangemi', 'Silvio Peroni', 'Genet Asefa Gesese', 'Cristian Santini'] | 2022-01-24 | null | null | null | null | ['knowledge-graph-embeddings', 'knowledge-graph-embeddings'] | ['graphs', 'methodology'] | [-6.70652926e-01 -1.07267149e-01 -3.40286821e-01 5.30819148e-02
-7.32719183e-01 -6.08234286e-01 7.57381618e-01 3.48669142e-01
-4.17014837e-01 8.82823467e-01 4.82173324e-01 -1.56871945e-01
-5.25452971e-01 -9.00424421e-01 -6.46834075e-01 -5.90463042e-01
1.07244939e-01 8.09541702e-01 -1.05806246e-01 3.01490799e-02
7.95607030e-01 6.53394282e-01 -1.14056051e+00 -3.34341884e-01
1.00268316e+00 6.43710077e-01 2.51517206e-01 4.69979823e-01
-4.70656484e-01 4.44251776e-01 -4.90386665e-01 -5.42189181e-01
5.67894466e-02 -1.37429954e-02 -1.02536774e+00 -1.68588489e-01
3.24683309e-01 3.58945608e-01 -6.87537134e-01 1.13003027e+00
6.33812308e-01 3.33087623e-01 7.78983831e-01 -1.42806566e+00
-1.33383048e+00 9.36026692e-01 -7.63779342e-01 2.55576849e-01
2.54148126e-01 -2.24893749e-01 1.30395377e+00 -1.00142372e+00
1.10434484e+00 1.33178425e+00 4.91424084e-01 1.11379996e-01
-6.71889901e-01 -7.48525143e-01 -1.08213700e-01 6.98214471e-01
-1.79962921e+00 -7.48450011e-02 8.09597552e-01 -3.66261810e-01
6.41822577e-01 2.09437549e-01 4.87108618e-01 1.14235187e+00
3.92526835e-02 5.76668143e-01 1.20034051e+00 -4.01159793e-01
7.32743666e-02 1.08324371e-01 5.94753921e-01 7.46726573e-01
8.53431165e-01 -4.92654949e-01 -7.72422552e-01 -2.15474308e-01
8.01610053e-01 -1.52087256e-01 -4.93368626e-01 -6.95440769e-02
-1.31355643e+00 6.96372628e-01 4.60078984e-01 4.84855086e-01
-3.62507999e-01 4.87499610e-02 2.10633278e-01 1.85611978e-01
8.22850540e-02 8.11749160e-01 -1.85206190e-01 -1.00105166e-01
-7.43029654e-01 2.03340590e-01 1.02432036e+00 1.21191287e+00
7.32057035e-01 -2.18371078e-01 1.51063809e-02 1.05368638e+00
4.35777009e-01 4.86572564e-01 6.71114564e-01 -7.59051442e-01
3.42920750e-01 7.85513759e-01 -5.78128174e-02 -1.35584939e+00
-2.47712091e-01 -6.11891806e-01 -5.75767457e-01 -4.64352101e-01
1.41523495e-01 8.79108533e-02 -1.01511467e+00 1.24790359e+00
1.89767703e-01 2.40403920e-01 9.70078334e-02 9.62882578e-01
1.62508535e+00 7.24721730e-01 -6.91946968e-02 1.64009303e-01
1.48651624e+00 -9.73443389e-01 -9.37880039e-01 2.17065051e-01
3.88494998e-01 -9.04173732e-01 8.03930998e-01 2.79789388e-01
-6.82320237e-01 -2.23319113e-01 -9.52548623e-01 -4.03246522e-01
-1.18689132e+00 2.02994660e-01 5.57468235e-01 1.57849982e-01
-1.24070442e+00 4.69678491e-01 -4.41886216e-01 -6.50191963e-01
4.54471380e-01 3.33232045e-01 -5.31670928e-01 -2.65814185e-01
-1.39946210e+00 5.93327165e-01 6.46795869e-01 -3.06993406e-02
-4.59693491e-01 -6.48722410e-01 -5.59336603e-01 7.53448159e-02
5.71835577e-01 -7.48651087e-01 4.63934928e-01 -2.41060033e-01
-1.13795078e+00 1.06093562e+00 1.32959142e-01 7.28028342e-02
3.47217053e-01 -3.28515500e-01 -7.65855432e-01 6.45338818e-02
4.07664686e-01 4.06077772e-01 2.89035678e-01 -1.45056498e+00
-5.50270796e-01 -5.25411546e-01 -1.81739002e-01 2.82657117e-01
-6.01294100e-01 -1.59664005e-01 -1.16073072e+00 -7.99860358e-01
-4.92513962e-02 -1.06033075e+00 2.69080937e-01 -6.90620065e-01
-9.20043290e-01 -7.30882943e-01 6.34813488e-01 -8.54457378e-01
1.39372361e+00 -1.74345422e+00 3.04422170e-01 4.68469143e-01
4.28816170e-01 1.37833551e-01 -5.22470176e-02 7.69896865e-01
-3.32452804e-02 4.66210514e-01 1.46276966e-01 -2.04686105e-01
1.90352708e-01 9.46314111e-02 -1.86185896e-01 3.03687543e-01
-3.83877486e-01 1.05847526e+00 -1.02333736e+00 -7.86587417e-01
-7.11240573e-03 2.51091182e-01 5.44957742e-02 9.62675661e-02
1.05048150e-01 3.28672379e-02 -7.75437176e-01 1.20700073e+00
6.33264601e-01 -3.68173093e-01 3.37149888e-01 -2.91626662e-01
-2.44066581e-01 -7.38928914e-02 -1.33019400e+00 1.57819498e+00
6.03638180e-02 6.37466609e-01 -1.49998933e-01 -8.73874784e-01
1.01014936e+00 7.66652301e-02 3.82056147e-01 -4.98568356e-01
3.91747504e-01 4.89768922e-01 -3.23179483e-01 -4.11503881e-01
9.59683955e-01 4.14484948e-01 -2.69132555e-01 8.88674259e-02
2.28356749e-01 2.29275957e-01 5.35419166e-01 9.22372997e-01
1.28738582e+00 -6.56862333e-02 1.08862154e-01 -4.32784855e-01
4.43768471e-01 2.02679202e-01 4.73010838e-01 7.79202521e-01
-4.53051226e-03 3.48844141e-01 5.46591640e-01 -5.80432154e-02
-6.57073855e-01 -1.09184635e+00 -2.83402532e-01 6.10189021e-01
3.73503238e-01 -6.36084795e-01 -3.13288718e-01 -4.91565019e-01
6.15006149e-01 6.65534556e-01 -6.67352080e-01 4.98641208e-02
-2.49256447e-01 -7.40770936e-01 6.78181589e-01 3.08030367e-01
5.42622507e-01 -1.10459375e+00 3.22299033e-01 -1.33945808e-01
2.34399624e-02 -1.11423326e+00 -3.85403574e-01 3.37899514e-02
-5.82766652e-01 -1.36310828e+00 -7.54352331e-01 -7.71082103e-01
5.99465609e-01 1.75881803e-01 1.15901852e+00 5.24691343e-02
-3.63201529e-01 8.35230708e-01 -5.03086686e-01 -2.48603970e-01
1.33920506e-01 3.59126627e-01 1.40944913e-01 -1.60883844e-01
5.35260081e-01 -5.03638089e-01 -4.44515407e-01 -2.64346711e-02
-8.42568040e-01 -2.78901339e-01 9.60424483e-01 6.57979727e-01
7.11149454e-01 1.46470010e-01 8.89601529e-01 -1.27099729e+00
8.59241605e-01 -8.77312243e-01 -4.52612430e-01 3.54819119e-01
-1.23575497e+00 1.82489417e-02 5.99008620e-01 -6.23741038e-02
-8.73459220e-01 -6.18274689e-01 2.66737580e-01 -6.28889561e-01
6.15070760e-02 8.53759944e-01 -2.12877378e-01 -1.24127381e-01
2.41066411e-01 9.42983627e-02 -3.41326058e-01 -8.05219948e-01
5.66521168e-01 1.18253243e+00 7.20848024e-01 -9.29968655e-01
9.49758768e-01 8.17616060e-02 3.80591266e-02 -9.14742410e-01
-5.06734192e-01 -8.32256019e-01 -4.47083324e-01 -2.57225394e-01
6.02674067e-01 -9.15345848e-01 -6.89851940e-01 3.64057332e-01
-8.36601377e-01 2.56594956e-01 1.65669516e-01 4.65280592e-01
9.03569683e-02 4.56808180e-01 -6.31521344e-01 -1.74367622e-01
-4.57877874e-01 -7.72472203e-01 6.31879210e-01 7.20828533e-01
-4.89130095e-02 -1.22460413e+00 -2.34695477e-03 8.04911494e-01
1.99748740e-01 3.75353605e-01 9.49839532e-01 -1.06165612e+00
-7.45683908e-01 -1.81516930e-01 -4.96687680e-01 8.26452225e-02
1.93051949e-01 1.42023608e-01 -5.24261951e-01 -1.85546666e-01
-8.88733864e-01 -1.12557754e-01 9.89041090e-01 4.55943830e-02
1.17990613e+00 -1.00286223e-01 -6.90801799e-01 4.79691893e-01
1.70162296e+00 1.38516963e-01 5.50649941e-01 8.22811306e-01
1.39254570e+00 1.08012781e-01 3.74783009e-01 3.76891226e-01
5.91948211e-01 4.39250350e-01 2.57093877e-01 2.90059507e-01
-3.40059489e-01 -9.01454613e-02 -6.96895346e-02 1.38122725e+00
-4.37619388e-01 -5.27608514e-01 -1.25619733e+00 9.38586056e-01
-1.86451387e+00 -6.49610519e-01 -5.36792040e-01 1.91167450e+00
9.03699577e-01 -8.68944526e-02 -2.19799235e-01 -2.77971029e-01
8.08137715e-01 2.09382832e-01 -4.07177418e-01 -3.24989289e-01
-2.99670249e-01 1.81049511e-01 8.62906396e-01 3.58925909e-01
-9.16535020e-01 1.12818658e+00 4.58110380e+00 1.05099642e+00
-7.92892158e-01 -1.14563063e-01 2.55881876e-01 8.61240104e-02
-3.82865369e-01 6.40990166e-03 -9.94656563e-01 6.99601471e-01
8.48318815e-01 -6.64324641e-01 4.77545053e-01 6.99103177e-01
-1.75451070e-01 2.35450305e-02 -6.81676686e-01 1.19948757e+00
3.03790212e-01 -1.45242333e+00 3.72551739e-01 1.69170901e-01
8.55797708e-01 6.64700046e-02 -6.53243065e-02 4.31647450e-01
4.90257025e-01 -1.12103057e+00 3.09085369e-01 6.66996956e-01
5.47074974e-01 -7.83014715e-01 9.02050674e-01 -1.31291449e-01
-1.13330150e+00 1.77829504e-01 -3.30725938e-01 4.30956006e-01
-2.80038506e-01 8.36689115e-01 -7.22347260e-01 1.26192474e+00
9.58182156e-01 1.12238288e+00 -9.00050342e-01 1.01380527e+00
-3.48526806e-01 5.34170270e-01 -2.15321798e-02 -2.14461133e-01
2.05312505e-01 -4.55528885e-01 7.89260924e-01 1.32219100e+00
3.46355647e-01 1.15326181e-01 -3.37533727e-02 7.57453322e-01
-9.17682767e-01 3.51761878e-01 -4.40457135e-01 -5.81406951e-01
1.03972840e+00 1.70661843e+00 -8.70268524e-01 -3.79964709e-01
-3.61133158e-01 7.90124953e-01 5.25939167e-01 4.85349208e-01
-6.98972940e-01 -8.34785104e-01 2.71078736e-01 5.86955398e-02
1.37026042e-01 -2.56111100e-02 -9.85670984e-02 -1.15481853e+00
-1.35881320e-01 -6.65688515e-01 7.22271442e-01 -8.49083304e-01
-1.52413332e+00 3.60096216e-01 -9.99211296e-02 -8.00863981e-01
4.48550820e-01 -7.97525585e-01 -3.35112840e-01 9.13962781e-01
-1.53934121e+00 -1.00032973e+00 -5.37326336e-01 3.88897866e-01
1.51832640e-01 -3.74708205e-01 5.70266664e-01 5.30129194e-01
-9.63850856e-01 5.77779233e-01 7.58766294e-01 3.53996962e-01
1.04971433e+00 -1.53630304e+00 -2.13611096e-01 5.62128544e-01
1.76922485e-01 8.92398298e-01 4.91671145e-01 -1.02548361e+00
-1.86565697e+00 -1.21644926e+00 8.49481642e-01 -7.24683404e-01
1.05368412e+00 2.14390941e-02 -1.00216365e+00 8.64404738e-01
5.27499497e-01 -1.12585172e-01 6.76967025e-01 3.05868775e-01
-2.51961827e-01 -4.64821085e-02 -8.25992227e-01 5.98699152e-01
1.11537564e+00 -1.99953303e-01 -7.61970460e-01 4.31922197e-01
3.13990027e-01 -4.48321074e-01 -1.43316817e+00 9.42319408e-02
3.29112977e-01 -2.88307607e-01 9.47432101e-01 -4.53336030e-01
5.93659401e-01 -4.02884632e-01 1.09806154e-02 -1.29780293e+00
-5.99192679e-01 -2.85765290e-01 -2.39461839e-01 1.72968602e+00
7.81682059e-02 -7.98713326e-01 5.36692023e-01 3.75803620e-01
-3.22204202e-01 -1.04708385e+00 -8.01462829e-01 -8.22546065e-01
4.39326130e-02 1.52018979e-01 4.45441484e-01 1.58169782e+00
-1.80153802e-01 4.33648556e-01 3.93917635e-02 2.85082072e-01
7.16151714e-01 5.23526549e-01 5.08914530e-01 -1.51684284e+00
2.69583613e-01 -5.89293599e-01 -5.00133991e-01 -5.18472612e-01
3.24594915e-01 -1.34863245e+00 -5.21521509e-01 -2.13411951e+00
4.89781708e-01 -4.29947466e-01 -6.56910717e-01 6.71360433e-01
-2.65948683e-01 -1.29900072e-02 -7.16304332e-02 7.10678101e-01
-9.06916797e-01 5.13061285e-01 1.11277413e+00 -1.18025303e-01
-7.87926465e-02 -8.83629858e-01 -1.01895416e+00 4.96057481e-01
6.90423727e-01 -4.39169914e-01 -9.10622030e-02 -2.50234038e-01
1.83566839e-01 -2.77413040e-01 3.72619808e-01 -6.76312864e-01
4.53437001e-01 -1.34703696e-01 4.35388327e-01 -5.40570319e-01
3.53388265e-02 -4.97977495e-01 3.32753927e-01 -4.04573325e-03
-1.24967918e-02 1.55419335e-01 1.50628403e-01 8.19420755e-01
-2.62614042e-01 -2.66844153e-01 3.00038725e-01 -1.78275421e-01
-1.22768843e+00 2.25021705e-01 2.57631224e-02 3.20962280e-01
9.28451657e-01 9.95067228e-03 -8.79394889e-01 9.98238027e-02
-5.92325449e-01 6.28875315e-01 5.83031774e-01 6.86562777e-01
4.79250252e-01 -1.50611198e+00 -6.09900594e-01 -5.33809245e-01
2.13468209e-01 -2.02123180e-01 5.03565781e-02 7.48191297e-01
-7.08510995e-01 6.30679250e-01 -1.72818556e-01 5.96207716e-02
-1.17668104e+00 3.26732695e-01 -2.42436096e-01 -4.08370972e-01
-5.87998092e-01 7.67381132e-01 -3.85249168e-01 -3.48764539e-01
1.51116014e-01 1.87264204e-01 -6.57649338e-01 4.81421173e-01
-1.00490160e-01 6.66528344e-01 7.21641332e-02 -8.60467851e-01
-7.09120631e-01 6.08818114e-01 -2.25809440e-01 2.03995883e-01
1.53741527e+00 6.32145489e-03 -5.19046307e-01 4.71547544e-01
1.13408995e+00 3.40420455e-01 -1.02450401e-01 -3.25257301e-01
2.71013379e-01 -4.59949851e-01 2.26169303e-01 -9.24058616e-01
-1.27061188e+00 2.45253190e-01 3.87434214e-01 9.94343087e-02
7.24290371e-01 1.58145234e-01 7.93471932e-01 3.78653318e-01
2.94058323e-01 -1.30448055e+00 -3.37168872e-02 4.00858343e-01
9.01878715e-01 -1.02990496e+00 3.82189035e-01 -3.12738329e-01
-4.24068898e-01 1.14576662e+00 4.50885028e-01 1.23803532e-02
6.17900372e-01 -2.43126631e-01 3.95771861e-02 -5.70323765e-01
-3.20085824e-01 -7.97965303e-02 4.74104196e-01 1.10120736e-01
3.04424793e-01 1.67676479e-01 -6.62540138e-01 9.91907954e-01
-1.95306689e-01 -2.77893171e-02 5.96547246e-01 9.45366740e-01
-1.92200020e-01 -1.17222583e+00 -4.83854473e-01 8.38795960e-01
-5.71081758e-01 -9.08572823e-02 -6.92728460e-01 1.11552656e+00
1.27269393e-02 6.97753549e-01 -4.36743587e-01 -2.53898948e-01
2.25820139e-01 1.18145242e-01 9.75348651e-02 -4.64991063e-01
-2.68477768e-01 -2.75912136e-01 1.50073305e-01 -1.48513407e-01
-2.97458082e-01 -4.78167593e-01 -1.41568434e+00 -6.63616598e-01
-1.89076781e-01 7.47614920e-01 5.54986179e-01 5.36351323e-01
6.70915544e-01 9.40876126e-01 2.76086867e-01 -5.24617136e-01
-1.29834674e-02 -9.36817110e-01 -1.05223358e+00 6.18223369e-01
-3.63750428e-01 -9.73923564e-01 -6.16775215e-01 -8.52154847e-03] | [9.332148551940918, 8.132402420043945] |
543f3696-d9f4-481c-b2fb-c264a5e7e743 | comparing-humans-and-models-on-a-similar | 2305.15389 | null | https://arxiv.org/abs/2305.15389v1 | https://arxiv.org/pdf/2305.15389v1.pdf | Comparing Humans and Models on a Similar Scale: Towards Cognitive Gender Bias Evaluation in Coreference Resolution | Spurious correlations were found to be an important factor explaining model performance in various NLP tasks (e.g., gender or racial artifacts), often considered to be ''shortcuts'' to the actual task. However, humans tend to similarly make quick (and sometimes wrong) predictions based on societal and cognitive presuppositions. In this work we address the question: can we quantify the extent to which model biases reflect human behaviour? Answering this question will help shed light on model performance and provide meaningful comparisons against humans. We approach this question through the lens of the dual-process theory for human decision-making. This theory differentiates between an automatic unconscious (and sometimes biased) ''fast system'' and a ''slow system'', which when triggered may revisit earlier automatic reactions. We make several observations from two crowdsourcing experiments of gender bias in coreference resolution, using self-paced reading to study the ''fast'' system, and question answering to study the ''slow'' system under a constrained time setting. On real-world data humans make $\sim$3\% more gender-biased decisions compared to models, while on synthetic data models are $\sim$12\% more biased. | ['Gabriel Stanovsky', 'Gili Lior'] | 2023-05-24 | null | null | null | null | ['coreference-resolution'] | ['natural-language-processing'] | [ 3.47412407e-01 6.75283015e-01 -1.42101824e-01 -5.29088140e-01
-2.18368292e-01 -4.57147866e-01 1.04496241e+00 5.36790013e-01
-9.13496077e-01 7.02185810e-01 2.55573452e-01 -5.30435622e-01
1.60552785e-01 -6.91755831e-01 -4.88649040e-01 -4.95924354e-01
3.88132393e-01 8.52189720e-01 1.38737457e-02 -5.60045779e-01
7.39137530e-01 -9.38958153e-02 -1.56432557e+00 1.43017992e-01
8.48135412e-01 3.76076132e-01 -4.41466458e-02 6.15740120e-01
-1.34469509e-01 1.04150534e+00 -7.68737018e-01 -8.89587641e-01
2.31755804e-02 -4.11617875e-01 -1.12027991e+00 -2.30716154e-01
3.00120741e-01 2.27507919e-01 5.17289080e-02 1.31046259e+00
6.43586993e-01 1.74432993e-01 4.88672644e-01 -1.28878105e+00
-7.10807621e-01 7.37773061e-01 -6.17872894e-01 5.65695703e-01
7.18108177e-01 3.86153311e-01 8.29899788e-01 -5.21955967e-01
7.09872723e-01 1.99897552e+00 5.94189346e-01 8.63458753e-01
-1.70012510e+00 -9.23461854e-01 2.52546668e-01 2.13833060e-02
-1.40339243e+00 -9.03927803e-01 3.78360808e-01 -8.62080157e-01
5.96701562e-01 5.04652858e-01 3.95741045e-01 1.46503890e+00
1.99179232e-01 3.79245907e-01 1.81838977e+00 -3.24386597e-01
2.84051359e-01 4.31080133e-01 3.43994558e-01 4.27679747e-01
5.25949419e-01 3.15642089e-01 -1.01370132e+00 -7.18270242e-01
4.28262949e-01 -6.08657002e-01 -1.86350301e-01 1.91933557e-01
-1.21746373e+00 9.86124218e-01 9.19860899e-02 2.57154584e-01
-3.96205664e-01 -1.51848765e-02 1.37416378e-01 2.94879824e-01
5.27065158e-01 9.47185338e-01 -2.29269654e-01 -3.15235466e-01
-6.70715332e-01 8.77815664e-01 1.12344623e+00 4.84110504e-01
5.86865425e-01 -3.86727422e-01 -4.46356565e-01 7.35953569e-01
2.46362895e-01 7.08170831e-01 4.81648713e-01 -1.27088559e+00
3.54785591e-01 5.00910819e-01 9.15607512e-01 -1.35499382e+00
-4.43823159e-01 -4.34323885e-02 -5.36730766e-01 7.55185857e-02
1.01180518e+00 -4.24625427e-02 -4.07423377e-01 2.09868336e+00
2.31296107e-01 -6.78321004e-01 -2.26919115e-01 1.15378702e+00
2.06052080e-01 1.59773260e-01 8.07335556e-01 -4.17744845e-01
1.63804793e+00 -1.49854019e-01 -6.67465270e-01 -5.87839961e-01
7.01951325e-01 -8.95764172e-01 1.22998512e+00 1.18002437e-01
-1.13053167e+00 -4.32794213e-01 -5.61360300e-01 6.84838369e-02
-8.74822810e-02 -3.92541558e-01 6.72806799e-01 9.76608336e-01
-9.14021790e-01 6.60822988e-01 -3.46083432e-01 -7.48134553e-01
1.97665587e-01 2.24798352e-01 9.98739079e-02 1.59483358e-01
-1.42199373e+00 1.26561928e+00 7.29544386e-02 -8.57335404e-02
-4.30485219e-01 -3.81279856e-01 -4.58338469e-01 -4.26526785e-01
5.42482376e-01 -8.14845383e-01 1.50597656e+00 -1.51761866e+00
-1.09384906e+00 1.62794602e+00 -7.62978137e-01 -4.75459427e-01
7.85036266e-01 -4.08813264e-03 -4.09902036e-01 -2.36588180e-01
5.47483981e-01 7.23495960e-01 6.97935164e-01 -1.29251373e+00
-4.82832611e-01 -6.25185251e-01 3.02392319e-02 3.23126704e-01
5.63582063e-01 5.20800591e-01 5.34521520e-01 -5.54008365e-01
2.56630238e-02 -1.19031703e+00 -1.41154677e-01 -2.42060974e-01
-2.79366523e-01 -5.66368222e-01 4.21977900e-02 -6.37062848e-01
1.22106302e+00 -1.70576406e+00 -1.18611686e-01 1.96598038e-01
7.09549487e-01 1.26023293e-01 3.19457173e-01 3.13046575e-01
-2.13759039e-02 6.87214673e-01 -6.14075400e-02 -8.54085013e-02
2.39532590e-01 7.31613263e-02 -2.72048414e-01 5.39148331e-01
-7.21463487e-02 7.57231414e-01 -1.01054454e+00 -5.64579487e-01
-2.87499696e-01 -1.58262551e-01 -3.19865167e-01 3.26360427e-02
-1.13100454e-01 5.27976751e-01 -9.59173590e-02 3.59335870e-01
4.07980978e-01 -1.84794262e-01 3.71550024e-01 2.66434848e-01
-3.87392163e-01 3.63806397e-01 -7.31874466e-01 9.41483259e-01
6.13271864e-03 7.32882500e-01 3.19465905e-01 -6.23127759e-01
9.57979202e-01 6.35436326e-02 -3.52023184e-01 -1.05436182e+00
3.55194688e-01 1.90902844e-01 7.17742622e-01 -5.16466439e-01
8.36996019e-01 -6.15266025e-01 -2.34398752e-01 6.97382212e-01
-5.77730954e-01 -5.68981729e-02 8.66942108e-02 3.17212164e-01
8.54135811e-01 -2.34652653e-01 3.05129826e-01 -8.18693340e-01
3.68410408e-01 2.80990571e-01 9.07881558e-01 1.06649566e+00
-7.17530310e-01 1.52362898e-01 9.90080476e-01 -6.20244563e-01
-9.76163626e-01 -8.30324948e-01 4.63333465e-02 1.57254016e+00
4.42588001e-01 -9.16611180e-02 -9.83915925e-01 -1.88380688e-01
1.31434232e-01 1.32960212e+00 -9.25201118e-01 -2.41730273e-01
-3.91735643e-01 -1.04868650e+00 7.14919567e-01 3.72252643e-01
5.86475194e-01 -6.82552218e-01 -8.83970141e-01 4.29421663e-02
-5.97689271e-01 -9.51303303e-01 -2.03855082e-01 -4.86971796e-01
-4.85764146e-01 -1.21705186e+00 -2.55473524e-01 -2.14945022e-02
6.18478239e-01 1.30267888e-01 1.50740612e+00 5.54904342e-01
7.96568543e-02 2.68000036e-01 -1.30045757e-01 -9.83708024e-01
-7.33943224e-01 -1.48168162e-01 4.69202667e-01 -1.17581323e-01
1.09753931e+00 -1.47688657e-01 -6.91852152e-01 6.74975812e-01
-5.41417360e-01 2.16841429e-01 3.11858982e-01 4.71210569e-01
-1.35353148e-01 -6.02609575e-01 3.52305800e-01 -1.27892780e+00
1.27421618e+00 -5.54483414e-01 -3.36090147e-01 1.34914279e-01
-1.02004731e+00 3.51906270e-02 2.14786008e-01 -4.68095392e-01
-1.27882004e+00 -6.41298771e-01 4.25689340e-01 1.76977381e-01
-1.69800296e-01 5.23079447e-02 1.66256085e-01 2.02102333e-01
1.21385300e+00 -1.87038377e-01 1.34398609e-01 -2.69334704e-01
-9.52161774e-02 7.24581242e-01 3.12908471e-01 -1.26393735e+00
7.10606933e-01 4.92112577e-01 -3.98156643e-01 -7.13920951e-01
-8.12086284e-01 -1.41201511e-01 -3.07422221e-01 -3.02958637e-01
9.15035486e-01 -9.82479632e-01 -1.23726153e+00 2.53172576e-01
-1.10132253e+00 -5.51867723e-01 2.60114335e-02 1.37164876e-01
-4.10495222e-01 8.69336501e-02 -5.61605990e-01 -1.18648565e+00
-1.26899838e-01 -8.20899725e-01 8.89637768e-01 4.50629711e-01
-1.36830389e+00 -7.94365942e-01 7.11015016e-02 7.31826603e-01
5.12405336e-01 1.17588408e-01 1.03362751e+00 -9.91150737e-01
1.01965144e-02 3.40711363e-02 -2.70374775e-01 -2.17352450e-01
-4.41768229e-01 2.47459859e-02 -1.10265970e+00 -4.30557989e-02
2.01076135e-01 -1.54887214e-01 3.63636166e-01 -7.50540495e-02
6.61107302e-01 -4.02237356e-01 -4.82124329e-01 -2.69356221e-01
9.21335161e-01 1.73757538e-01 5.35182118e-01 2.25811645e-01
4.18506771e-01 1.30487168e+00 7.39153564e-01 2.85117686e-01
8.08206439e-01 5.70197821e-01 -2.02132985e-01 3.31260532e-01
4.43279415e-01 -5.08551538e-01 1.73437297e-01 3.55641425e-01
-4.66043800e-01 4.77178441e-03 -1.51912630e+00 3.73172581e-01
-1.92369223e+00 -9.28606510e-01 -4.85678822e-01 2.36642432e+00
7.26264477e-01 2.53550678e-01 1.95155799e-01 -1.07283242e-01
1.27849603e+00 2.72712469e-01 -2.25542858e-01 -7.62439668e-01
-4.96955439e-02 -1.75550822e-02 4.40658569e-01 6.46000266e-01
-5.02862334e-01 8.99908781e-01 6.17286777e+00 4.49581116e-01
-8.98071766e-01 2.87455589e-01 9.66445923e-01 -3.89129482e-02
-4.39378023e-01 1.85720921e-01 -4.65589911e-01 7.43807197e-01
1.26379061e+00 -6.54267073e-01 3.81818086e-01 3.34367692e-01
4.95772630e-01 -8.52981865e-01 -1.31002867e+00 9.62064743e-01
-5.32251559e-02 -8.40726495e-01 -1.67415068e-01 1.18757360e-01
4.44806844e-01 -4.90989208e-01 -1.46610066e-01 4.64047313e-01
7.35171556e-01 -1.15830255e+00 1.20848274e+00 6.21557117e-01
5.38906217e-01 -3.42624605e-01 6.67218626e-01 7.34951019e-01
-3.50852698e-01 -4.39249985e-02 -2.93194622e-01 -7.51435161e-01
-1.38626620e-01 3.42802227e-01 -6.87182486e-01 -3.03083956e-01
7.28205979e-01 -2.35747308e-01 -6.30827069e-01 2.28614047e-01
-2.31667221e-01 5.90698183e-01 5.45589672e-03 -3.16346347e-01
-2.13530689e-01 -3.98563221e-02 6.73307180e-01 9.56273437e-01
-2.34974489e-01 5.12847066e-01 -2.21150026e-01 1.14968884e+00
4.89419745e-03 9.44641829e-02 -4.28397626e-01 -1.34395257e-01
7.33751357e-01 9.09446061e-01 -8.84979546e-01 -5.74982166e-01
1.34015054e-01 6.36392653e-01 1.75948933e-01 3.26381594e-01
-6.31524622e-01 7.72889778e-02 7.03201532e-01 4.08163339e-01
-6.95532739e-01 -5.43655343e-02 -5.78474522e-01 -1.06404579e+00
-3.13284069e-01 -1.13528931e+00 3.70734036e-02 -8.19110930e-01
-1.08443820e+00 3.56535204e-02 2.55087167e-02 -3.61394614e-01
-2.22547740e-01 -4.10353839e-01 -4.13052559e-01 1.19011450e+00
-8.60406935e-01 -4.39539582e-01 -3.82447243e-01 1.74011409e-01
3.65154564e-01 1.48323208e-01 8.07948709e-01 -2.09429368e-01
-6.46421373e-01 2.47200862e-01 -4.21122104e-01 -1.20748198e-02
1.05068588e+00 -1.22013283e+00 5.99078417e-01 5.35574615e-01
-4.00806636e-01 9.45450723e-01 1.34992957e+00 -6.68495715e-01
-9.95241523e-01 -5.25083721e-01 1.44660413e+00 -1.12897336e+00
5.32118022e-01 -4.67707306e-01 -9.51533318e-01 5.65464497e-01
1.36450291e-01 -5.67711949e-01 9.40515518e-01 3.93740743e-01
-4.03979987e-01 2.54518896e-01 -1.32781851e+00 8.79423618e-01
1.33041012e+00 -6.93192661e-01 -1.06339014e+00 2.78974265e-01
4.24668878e-01 -3.04648042e-01 -5.95537663e-01 -1.93995833e-02
5.87380767e-01 -1.28167939e+00 4.40527081e-01 -6.88848853e-01
5.48799872e-01 5.00021987e-02 -5.41844331e-02 -1.20277309e+00
-3.85611504e-01 -7.98144400e-01 3.84032756e-01 1.13226151e+00
3.46591711e-01 -8.39740336e-01 3.92864645e-01 1.54149938e+00
6.19781435e-01 -4.48118955e-01 -7.18938112e-01 -2.60599673e-01
3.64283234e-01 -3.81474406e-01 6.55143380e-01 1.21169281e+00
1.23652462e-02 4.98001516e-01 1.90369394e-02 -3.62455137e-02
6.56412363e-01 -1.86860204e-01 7.91867316e-01 -1.64343393e+00
1.03046730e-01 -4.90355819e-01 -6.33946015e-03 -4.95814413e-01
2.79584765e-01 -3.96387488e-01 4.39839140e-02 -7.81882644e-01
3.10166359e-01 -4.57510412e-01 3.00990194e-01 8.47392809e-03
-5.05144536e-01 -1.52370691e-01 3.17808986e-01 6.44037247e-01
-5.57488024e-01 1.21745013e-01 7.06214190e-01 2.39485711e-01
-1.22547392e-02 -1.89953879e-01 -1.35168922e+00 1.03371644e+00
7.45660663e-01 -5.98672569e-01 -7.41576329e-02 -3.54275644e-01
9.58582163e-01 1.86891034e-01 6.35071993e-01 -5.32793581e-01
2.45345056e-01 -5.40799022e-01 2.52372235e-01 3.84856433e-01
8.57640877e-02 -2.02007279e-01 6.30409271e-02 5.68110943e-01
-6.19623363e-01 3.72272849e-01 -2.97040748e-03 6.43034816e-01
1.19252823e-01 2.21293680e-02 8.14658105e-01 -5.68964481e-01
-4.84390587e-01 -4.33127582e-01 -7.35941231e-01 4.16221946e-01
8.15883279e-01 -2.78810322e-01 -6.06678784e-01 -5.65380037e-01
-7.02581286e-01 4.57461387e-01 8.92774582e-01 5.20706892e-01
-8.45460147e-02 -8.59292805e-01 -7.14741051e-01 -2.62502462e-01
1.29873738e-01 -4.00307268e-01 1.01525456e-01 1.04083931e+00
-4.87562329e-01 3.57535690e-01 -7.82896858e-03 -4.00974840e-01
-1.07421088e+00 3.52143824e-01 4.97681022e-01 7.01672882e-02
2.17781156e-01 9.53617215e-01 2.89394259e-01 -3.75439048e-01
-2.79759735e-01 6.90899342e-02 -6.59762919e-02 4.27286893e-01
4.38926101e-01 8.14031661e-01 -4.50633734e-01 -9.50723529e-01
-3.59119892e-01 7.50736669e-02 -1.91195477e-02 -4.49059814e-01
5.35697877e-01 -5.52080750e-01 -3.49441051e-01 7.78919756e-01
3.36821586e-01 4.90759090e-02 -6.51546359e-01 -1.21843405e-01
2.77452052e-01 -7.95895100e-01 -5.62567413e-01 -9.21739399e-01
-9.96232033e-02 6.44147992e-01 4.28583533e-01 5.30698299e-01
3.32356244e-01 -9.88396257e-02 2.15416014e-01 3.59876007e-01
6.76887393e-01 -1.58570647e+00 -3.38358790e-01 2.17953816e-01
8.11597764e-01 -1.42732966e+00 -3.56771648e-02 -3.79748881e-01
-8.58293653e-01 5.29656529e-01 8.45148683e-01 2.76722964e-02
5.30474856e-02 -3.53645176e-01 2.59947449e-01 -2.89478213e-01
-1.19317031e+00 2.84687318e-02 -1.93394750e-01 4.31373388e-01
7.63369739e-01 5.18559277e-01 -8.67105365e-01 7.32051373e-01
-7.01099515e-01 3.20254639e-02 6.96506560e-01 5.69451809e-01
-3.96280020e-01 -8.39382291e-01 -9.14801359e-01 5.57269156e-01
-6.86406910e-01 1.18435182e-01 -1.04971015e+00 8.41920853e-01
5.33007562e-01 1.25652134e+00 3.70015264e-01 -3.41778755e-01
4.71355796e-01 4.18154240e-01 3.63389015e-01 -5.42455733e-01
-8.73253226e-01 -3.78199995e-01 5.76801956e-01 -7.08292007e-01
-5.43016851e-01 -1.09502721e+00 -8.65879893e-01 -9.96592581e-01
6.51760772e-02 2.53994823e-01 2.67106175e-01 1.09655476e+00
2.47176677e-01 -1.97742030e-01 1.71230271e-01 -6.11350596e-01
-6.85734034e-01 -1.12878251e+00 -4.70802575e-01 8.63957047e-01
-2.34715864e-01 -6.70689523e-01 -6.23207211e-01 -1.05592482e-01] | [9.962124824523926, 7.927724361419678] |
7405ffcf-a59d-4f3e-8a62-37f6727979a8 | moving-window-regression-a-novel-approach-to | 2203.13122 | null | https://arxiv.org/abs/2203.13122v1 | https://arxiv.org/pdf/2203.13122v1.pdf | Moving Window Regression: A Novel Approach to Ordinal Regression | A novel ordinal regression algorithm, called moving window regression (MWR), is proposed in this paper. First, we propose the notion of relative rank ($\rho$-rank), which is a new order representation scheme for input and reference instances. Second, we develop global and local relative regressors ($\rho$-regressors) to predict $\rho$-ranks within entire and specific rank ranges, respectively. Third, we refine an initial rank estimate iteratively by selecting two reference instances to form a search window and then estimating the $\rho$-rank within the window. Extensive experiments results show that the proposed algorithm achieves the state-of-the-art performances on various benchmark datasets for facial age estimation and historical color image classification. The codes are available at https://github.com/nhshin-mcl/MWR. | ['Chang-Su Kim', 'Seon-Ho Lee', 'Nyeong-Ho Shin'] | 2022-03-24 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Shin_Moving_Window_Regression_A_Novel_Approach_to_Ordinal_Regression_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Shin_Moving_Window_Regression_A_Novel_Approach_to_Ordinal_Regression_CVPR_2022_paper.pdf | cvpr-2022-1 | ['age-and-gender-classification', 'age-estimation', 'age-estimation'] | ['computer-vision', 'computer-vision', 'miscellaneous'] | [ 9.57804248e-02 -2.94274420e-01 -6.29442036e-01 -6.38795495e-01
-1.19783342e+00 9.28555205e-02 3.99251282e-01 -2.44959332e-02
-2.47702524e-01 6.33533239e-01 4.98425029e-02 -3.52273621e-02
-5.71166217e-01 -5.98633349e-01 -2.98541874e-01 -5.84897399e-01
-4.10676330e-01 1.81693628e-01 -2.15628631e-02 -3.10128722e-02
3.98112684e-01 2.83548057e-01 -1.81623888e+00 1.79593414e-01
1.00632739e+00 1.27889776e+00 -2.77524114e-01 5.16487122e-01
2.71523416e-01 7.23249972e-01 -3.96516442e-01 -6.40833080e-01
1.53826609e-01 -5.26108503e-01 -6.20506644e-01 -1.09127991e-01
4.56953853e-01 -2.99767286e-01 -2.25601539e-01 8.63542378e-01
3.10757101e-01 2.73549616e-01 7.65613317e-01 -1.32978272e+00
-6.32395387e-01 6.10205770e-01 -1.21120393e+00 5.77089610e-03
5.77538133e-01 -3.34291637e-01 1.16679430e+00 -1.22331500e+00
3.88590693e-01 1.30458450e+00 4.64222550e-01 7.00345516e-01
-1.09302974e+00 -1.27452862e+00 3.24107409e-01 5.09692430e-01
-1.69549060e+00 -4.89010543e-01 9.18095648e-01 -3.48284334e-01
2.50877380e-01 5.75119972e-01 3.85732144e-01 7.49623060e-01
-9.12353396e-02 7.62026131e-01 1.45485163e+00 -2.34080106e-01
-1.64070323e-01 -2.40786120e-01 3.62243235e-01 1.09690571e+00
-1.13927230e-01 1.59778610e-01 -8.21416080e-01 -2.84865379e-01
6.17232859e-01 -7.29565844e-02 4.12560999e-02 -9.71939508e-03
-6.88957036e-01 7.74411500e-01 2.11713538e-01 -6.05010875e-02
-6.76665381e-02 2.26569578e-01 2.96491414e-01 2.68188536e-01
7.83181012e-01 -2.05534771e-01 -3.30645859e-01 9.30019692e-02
-1.03635359e+00 -6.34442642e-02 9.32847112e-02 6.12191200e-01
6.48094535e-01 -1.90718740e-01 -1.53285101e-01 1.41626585e+00
2.61351943e-01 2.15436503e-01 3.59487653e-01 -1.09312272e+00
2.87972480e-01 6.24886990e-01 -1.68803427e-02 -9.73068416e-01
-2.51114875e-01 -1.93387926e-01 -8.63875210e-01 2.72475988e-01
5.32109261e-01 3.20791095e-01 -8.67697239e-01 1.73393571e+00
3.74315321e-01 5.16506553e-01 -3.14708054e-01 6.11866295e-01
1.00293875e+00 5.59210181e-01 1.66516960e-01 -6.63362920e-01
1.37589431e+00 -7.93471754e-01 -2.66642630e-01 6.57450780e-02
1.45799279e-01 -6.68995619e-01 9.98068988e-01 7.26112187e-01
-1.23517871e+00 -6.15968823e-01 -8.73956740e-01 4.55737822e-02
-2.39377230e-01 5.01337945e-01 6.50965631e-01 4.94040549e-01
-8.48911762e-01 7.58291006e-01 -6.48004949e-01 -1.03935480e-01
4.16765094e-01 6.53238118e-01 -3.98585796e-01 -1.30026668e-01
-1.00608206e+00 2.93864131e-01 -1.19169816e-01 2.27234408e-01
-6.60423636e-01 -7.28864849e-01 -7.70576656e-01 -4.02461708e-01
3.65515620e-01 -1.85027778e-01 8.74064267e-01 -8.21924627e-01
-1.36001301e+00 1.30278563e+00 -5.44466674e-01 7.90619776e-02
5.37939548e-01 -1.26322925e-01 -7.20080793e-01 -1.11692689e-01
-5.83226830e-02 5.03347516e-01 1.01341546e+00 -1.33138633e+00
-8.91604543e-01 -6.94900453e-01 -2.33422056e-01 3.67992744e-02
-2.16402292e-01 5.61087549e-01 -7.64622629e-01 -7.10434020e-01
3.96405041e-01 -7.30374873e-01 -1.34243220e-01 -2.10775554e-01
-2.92121381e-01 -7.13222146e-01 2.43562058e-01 -9.48746264e-01
1.91133106e+00 -2.12251782e+00 2.02417776e-01 5.78883171e-01
3.82846832e-01 -1.29302382e-01 -1.70721501e-01 -1.04602538e-01
-5.16988516e-01 1.01120658e-01 -8.31131637e-02 -2.88718998e-01
-2.84867376e-01 -2.19904378e-01 -1.41423792e-01 5.82507730e-01
-6.70845993e-03 4.48756367e-01 -7.71446645e-01 -7.11233497e-01
-6.50219321e-02 3.81170154e-01 -2.16391310e-01 2.77051717e-01
2.76130319e-01 2.51167655e-01 -4.36126024e-01 1.15688574e+00
7.33095467e-01 -7.03550279e-02 6.48786649e-02 -1.95916384e-01
-1.96775794e-01 -2.23964557e-01 -1.19776309e+00 1.16032338e+00
-1.94870174e-01 3.01514059e-01 -1.69035375e-01 -8.67905676e-01
1.34218550e+00 -1.19626664e-01 8.20283771e-01 -7.22815335e-01
1.48481578e-01 2.95174986e-01 -2.72919685e-01 -1.43313468e-01
6.04909658e-01 3.80441174e-02 -1.71403408e-01 9.48257372e-02
-2.75705069e-01 4.60237622e-01 3.76286119e-01 -8.56645107e-02
9.09980059e-01 2.50781506e-01 3.56811553e-01 -8.15384388e-02
1.06245160e+00 -4.55709308e-01 9.69216228e-01 2.86673784e-01
-2.48372838e-01 6.94012582e-01 7.68860221e-01 -4.70078558e-01
-7.38761306e-01 -1.29069161e+00 -2.70107001e-01 1.73725343e+00
1.42748833e-01 -4.71186101e-01 -4.19207722e-01 -6.94053650e-01
1.29040822e-01 2.61555523e-01 -9.90184307e-01 5.35747111e-02
-6.94067717e-01 -9.95856702e-01 1.66821033e-01 4.32925612e-01
2.92735964e-01 -8.04422259e-01 -2.37821385e-01 -1.92031085e-01
-6.17091246e-02 -5.91171980e-01 -6.20829284e-01 -2.73258090e-01
-9.20439124e-01 -1.20467877e+00 -7.22949624e-01 -7.23702252e-01
1.05626166e+00 -4.93847802e-02 1.02338946e+00 3.09034765e-01
-5.56203842e-01 2.00974584e-01 -4.91864592e-01 4.88281995e-02
4.04055789e-02 -1.53941065e-01 4.92901206e-02 1.69503957e-01
1.52358606e-01 -4.17561918e-01 -9.43079650e-01 7.68140733e-01
-3.81555736e-01 -3.89689766e-02 5.70725799e-01 8.37325513e-01
9.26358998e-01 -1.54011711e-01 6.17390633e-01 -1.13891101e+00
5.07333755e-01 -3.74100834e-01 -7.99997211e-01 4.09303635e-01
-8.95661950e-01 -5.74816465e-02 4.39917803e-01 -5.93266070e-01
-7.53160477e-01 -3.75820920e-02 1.80395823e-02 -2.59022117e-01
3.31422389e-01 3.69818866e-01 6.23883829e-02 2.35533193e-01
3.39875162e-01 -8.36153794e-03 -1.78164735e-01 -4.22236979e-01
3.12062949e-01 5.95625043e-01 6.76867902e-01 -7.73463130e-01
9.35848713e-01 3.69609356e-01 9.88874659e-02 -3.85776252e-01
-6.39863908e-01 -4.91882145e-01 -5.19021034e-01 -5.27843356e-01
5.42629957e-01 -8.59514058e-01 -8.26317430e-01 5.38747072e-01
-5.05927503e-01 -2.86106288e-01 1.70464665e-01 2.65358776e-01
-3.71982783e-01 2.44319171e-01 -4.40502286e-01 -1.08588970e+00
-4.78466332e-01 -8.83062303e-01 8.27655256e-01 4.69580144e-01
-1.81684136e-01 -4.38094616e-01 -1.55644149e-01 5.83697200e-01
-1.79101884e-01 6.79309368e-01 8.43906343e-01 -2.51446635e-01
-2.57865876e-01 -2.40493104e-01 -4.65807021e-01 1.93985850e-01
8.07873160e-02 6.05565548e-01 -6.41614079e-01 -3.67221266e-01
-7.88303852e-01 -2.49672711e-01 1.00186968e+00 4.94868517e-01
1.76064646e+00 -2.21060246e-01 -2.64681131e-01 7.57402837e-01
1.23656642e+00 3.54486763e-01 8.31046641e-01 8.72274786e-02
4.89874333e-01 7.63536453e-01 1.18090534e+00 7.64719844e-01
4.23178166e-01 6.09169006e-01 1.59550697e-01 -3.31404507e-02
1.17349084e-02 -2.22274587e-01 4.35240418e-01 9.19441760e-01
-6.18561268e-01 3.42464060e-01 -8.42027664e-01 3.43393505e-01
-1.89713943e+00 -7.63887405e-01 8.56481940e-02 2.46488523e+00
9.17890370e-01 1.24751702e-01 5.12822509e-01 2.01456517e-01
9.90355253e-01 3.70028555e-01 -6.05540156e-01 -5.26447177e-01
2.32657015e-01 4.59561855e-01 4.76544619e-01 5.62749565e-01
-9.20206666e-01 9.42950368e-01 6.02682734e+00 1.19809139e+00
-8.84418547e-01 -3.34938243e-02 1.30559015e+00 -1.74384683e-01
4.02118005e-02 -2.27781743e-01 -8.20838153e-01 5.69892704e-01
6.97393775e-01 -1.74991325e-01 5.97212911e-01 9.16342378e-01
1.60149932e-01 -2.03715235e-01 -9.46522772e-01 1.19623041e+00
1.08758688e-01 -7.08701909e-01 -2.45302483e-01 -3.19648981e-02
5.76057613e-01 -4.74426270e-01 6.16985679e-01 3.48477662e-01
4.57121849e-01 -1.26571691e+00 4.94155377e-01 7.00899780e-01
1.47714651e+00 -1.19821453e+00 3.34168673e-01 -2.12800339e-01
-1.65673685e+00 -3.55089068e-01 -3.16150695e-01 2.11156785e-01
-4.07314688e-01 5.17290115e-01 -2.48736471e-01 4.36252177e-01
1.00718725e+00 8.11011314e-01 -9.39244926e-01 7.55753577e-01
-1.80666372e-01 4.04866666e-01 9.87854972e-02 5.16046099e-02
-3.43168765e-01 -4.26142931e-01 1.28292456e-01 8.00441802e-01
4.05145407e-01 3.54960948e-01 -9.10067707e-02 2.31077716e-01
-3.06148648e-01 3.96176547e-01 -6.22675801e-03 3.55148017e-01
5.65582097e-01 1.45537949e+00 -6.45939410e-01 5.80633879e-02
-3.28422815e-01 6.21898353e-01 3.71285349e-01 1.34272099e-01
-1.07301092e+00 -4.80516165e-01 7.00917542e-01 1.93767369e-01
-1.52940124e-01 -4.66818847e-02 -1.54004917e-01 -9.69508886e-01
-5.22320904e-02 -8.69688749e-01 1.11176169e+00 -7.42042482e-01
-1.18272257e+00 4.17072952e-01 2.63572305e-01 -1.35342693e+00
-2.72076819e-02 -6.26013279e-01 -5.91888189e-01 5.00731409e-01
-1.58697355e+00 -1.22550762e+00 -5.36550522e-01 6.24195158e-01
5.61066270e-01 -2.38989413e-01 5.47118843e-01 2.74681896e-01
-9.64379370e-01 1.18280935e+00 7.46595114e-02 2.66498744e-01
8.43882859e-01 -1.06750870e+00 -1.50009736e-01 6.08727336e-01
-1.89541504e-01 6.16731465e-01 4.30240810e-01 -5.18101454e-01
-1.00397360e+00 -9.18719530e-01 5.67108870e-01 -3.79735202e-01
6.54335618e-01 -1.96585566e-01 -6.82651401e-01 3.83724213e-01
-3.18754435e-01 1.42861605e-01 8.68718684e-01 5.76481521e-01
-7.23593295e-01 -8.03912222e-01 -1.21955800e+00 6.91355526e-01
1.18803918e+00 -3.76093864e-01 -1.49334267e-01 -6.78251460e-02
5.37961185e-01 -3.05089533e-01 -1.27607071e+00 7.74669349e-01
1.26108539e+00 -9.44041073e-01 1.17695236e+00 -3.00505638e-01
5.20607769e-01 -2.18088403e-01 -1.33497879e-01 -7.01176524e-01
-3.83745760e-01 -6.30532503e-01 -2.95846522e-01 1.50751483e+00
5.61561823e-01 -4.94417459e-01 9.53986943e-01 5.97806334e-01
2.77862519e-01 -1.43005979e+00 -9.63212013e-01 -6.41141713e-01
1.71882510e-02 -3.70416611e-01 6.97739005e-01 8.84768546e-01
-7.71010444e-02 -3.45562547e-02 -5.39130628e-01 -3.43320556e-02
8.35422456e-01 5.14411815e-02 4.69621211e-01 -1.31667399e+00
-1.27690002e-01 -6.67981207e-01 -3.48796874e-01 -3.99898559e-01
2.29361787e-01 -6.36186898e-01 -1.92381918e-01 -1.25438797e+00
5.14853060e-01 -5.88752091e-01 -1.04901409e+00 4.88221258e-01
-3.76488328e-01 6.33667707e-01 -1.89615577e-01 2.68822968e-01
-7.61821449e-01 2.85925746e-01 8.12359095e-01 8.81918147e-02
-4.07590032e-01 1.83326855e-01 -8.05102944e-01 6.86152279e-01
9.48647976e-01 -3.80467057e-01 -1.26588598e-01 -3.21968086e-02
3.64270627e-01 1.79704309e-01 6.00952953e-02 -8.25175524e-01
2.63582319e-02 -4.85562265e-01 7.37429857e-01 -7.56167352e-01
2.86202520e-01 -4.05572027e-01 2.36866884e-02 4.28497821e-01
-5.69373250e-01 2.36197293e-01 -2.43986949e-01 2.98142552e-01
-3.46711762e-02 6.22067079e-02 1.15233922e+00 3.12174261e-01
-7.68153489e-01 5.98693788e-01 3.01499546e-01 -9.13655981e-02
1.05406332e+00 -3.11374664e-01 -2.49133483e-01 -1.38858736e-01
-7.63545275e-01 4.57789570e-01 3.02664042e-01 5.04744172e-01
9.68417108e-01 -1.62324357e+00 -9.60816801e-01 6.28298707e-03
4.72777843e-01 -4.11283523e-01 1.74150050e-01 7.28629112e-01
-2.76140392e-01 -1.74537644e-01 -1.41270295e-01 -2.76694864e-01
-2.01313448e+00 3.05285245e-01 -2.04568766e-02 -4.19662118e-01
-7.33801574e-02 1.15819335e+00 -5.16983196e-02 -1.91175252e-01
2.51570076e-01 -5.92108518e-02 -5.39286554e-01 3.40954334e-01
4.86590862e-01 8.48278761e-01 -4.06482816e-01 -7.43830323e-01
-5.08964956e-01 1.14575601e+00 -9.49748531e-02 -9.96567160e-02
1.38715422e+00 -1.05346955e-01 -4.11839783e-01 4.47234839e-01
1.26556385e+00 1.24397665e-01 -1.18087339e+00 -1.41350657e-01
1.30544543e-01 -8.06074142e-01 -2.91943163e-01 -6.07875764e-01
-1.15367472e+00 3.36920023e-01 8.23539317e-01 -3.26280326e-01
1.67798936e+00 2.26917818e-01 4.67492193e-01 -1.97927237e-01
2.38257125e-01 -1.36310101e+00 2.86252439e-01 1.31969512e-01
7.95535505e-01 -1.15692854e+00 5.66959918e-01 -5.25242388e-01
-6.21261477e-01 9.66349125e-01 8.48628163e-01 -2.04662070e-01
5.80154240e-01 -2.72325277e-01 -1.60619348e-01 1.55443236e-01
-8.25135052e-01 -2.04349220e-01 6.19117856e-01 2.66206712e-01
6.58860207e-01 1.77898526e-01 -7.52448261e-01 8.28794479e-01
-2.52303034e-01 -2.37174943e-01 5.90024516e-02 4.94813204e-01
-3.52028191e-01 -1.38714337e+00 -3.53257775e-01 7.79481947e-01
-5.32889068e-01 1.01594761e-01 -2.93890685e-01 7.55321026e-01
4.02773052e-01 9.55626726e-01 1.28094390e-01 -8.85766387e-01
1.93182439e-01 -1.16899526e-02 5.65354168e-01 -2.55229384e-01
-3.76506418e-01 1.89944893e-01 5.49916476e-02 -7.18856633e-01
-2.22604990e-01 -8.82410228e-01 -1.27971089e+00 -3.86162311e-01
-1.66231588e-01 1.36373729e-01 5.18028319e-01 3.06799084e-01
6.83657639e-03 2.94993341e-01 1.27583742e+00 -3.04270595e-01
-1.93471730e-01 -8.40632558e-01 -6.95644438e-01 3.35189521e-01
7.26095289e-02 -7.00646877e-01 -3.07131082e-01 1.30184349e-02] | [13.687438011169434, 0.8507369160652161] |
20423c1b-92db-4f38-8765-aaf83dbad5a7 | progressive-point-cloud-deconvolution | 2007.05361 | null | https://arxiv.org/abs/2007.05361v1 | https://arxiv.org/pdf/2007.05361v1.pdf | Progressive Point Cloud Deconvolution Generation Network | In this paper, we propose an effective point cloud generation method, which can generate multi-resolution point clouds of the same shape from a latent vector. Specifically, we develop a novel progressive deconvolution network with the learning-based bilateral interpolation. The learning-based bilateral interpolation is performed in the spatial and feature spaces of point clouds so that local geometric structure information of point clouds can be exploited. Starting from the low-resolution point clouds, with the bilateral interpolation and max-pooling operations, the deconvolution network can progressively output high-resolution local and global feature maps. By concatenating different resolutions of local and global feature maps, we employ the multi-layer perceptron as the generation network to generate multi-resolution point clouds. In order to keep the shapes of different resolutions of point clouds consistent, we propose a shape-preserving adversarial loss to train the point cloud deconvolution generation network. Experimental results demonstrate the effectiveness of our proposed method. | ['Jian Yang', 'Rui Xu', 'Le Hui', 'Jianjun Qian', 'Jin Xie'] | 2020-07-10 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2354_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123600392.pdf | eccv-2020-8 | ['point-cloud-generation'] | ['computer-vision'] | [-1.08969674e-01 -2.33530417e-01 3.14304829e-01 -1.84853643e-01
-7.95303047e-01 -4.35593456e-01 4.63355601e-01 -4.71423388e-01
-2.44943589e-01 6.25112832e-01 -4.66358736e-02 1.56137258e-01
-5.55157773e-02 -1.18465006e+00 -1.19240177e+00 -7.41833031e-01
2.00581640e-01 2.51970619e-01 1.28141701e-01 8.92759040e-02
2.60826916e-01 9.96279359e-01 -1.39265645e+00 3.99778962e-01
1.05497122e+00 9.07033503e-01 4.55097377e-01 4.47373569e-01
-1.87107071e-01 2.54321069e-01 -2.83649504e-01 -1.71209067e-01
5.01758814e-01 8.44080523e-02 -5.53416967e-01 2.40176529e-01
5.66923082e-01 -7.09845722e-01 -3.73663336e-01 1.12298071e+00
2.63421565e-01 -2.45174617e-02 5.34276783e-01 -1.05800426e+00
-1.11748040e+00 -5.04508764e-02 -6.49460971e-01 -3.21651399e-01
6.75436333e-02 1.76577479e-01 6.73081696e-01 -1.32291079e+00
4.90443677e-01 1.54728913e+00 8.27765346e-01 2.48611525e-01
-1.31039691e+00 -9.08582985e-01 1.93667598e-02 -2.20248938e-01
-1.76496112e+00 -8.57311189e-02 1.15903544e+00 -5.11989653e-01
4.07212079e-01 1.94782719e-01 3.59582096e-01 5.44231415e-01
4.80128899e-02 3.13726038e-01 8.58807623e-01 -2.08656080e-02
-1.07284710e-01 -4.15857136e-02 -6.48755670e-01 5.86229086e-01
-2.16734223e-02 2.19797954e-01 -1.57728761e-01 -4.55135822e-01
1.70473933e+00 6.34412050e-01 -3.41026157e-01 -2.10456803e-01
-1.40045154e+00 8.95113111e-01 9.83932257e-01 2.15120196e-01
-6.55454755e-01 3.55978727e-01 -5.17290458e-02 1.49180675e-02
7.87615299e-01 1.74389675e-01 -1.81343675e-01 5.77743113e-01
-1.06907797e+00 4.96737152e-01 1.93319377e-02 1.13989818e+00
1.15176046e+00 9.69970897e-02 -1.66735530e-01 6.94375575e-01
6.58404887e-01 4.74856943e-01 3.34913373e-01 -1.28818715e+00
6.39184296e-01 6.06251597e-01 4.87000585e-01 -1.14407027e+00
9.96616110e-02 -2.49528155e-01 -1.23777819e+00 7.48796582e-01
-7.08265789e-03 -2.89701931e-02 -8.96167457e-01 1.37300873e+00
2.99080431e-01 8.60774398e-01 1.22710548e-01 1.01971388e+00
6.01357341e-01 1.07943118e+00 -7.72177652e-02 3.33738774e-02
1.09895849e+00 -6.55421674e-01 -4.67076749e-01 9.60573852e-02
-2.13738889e-01 -7.42533147e-01 7.38969266e-01 -1.84179828e-01
-1.37528038e+00 -7.24851072e-01 -1.13733959e+00 -3.96299958e-01
-1.23120435e-01 -3.60495644e-03 1.42300352e-01 -2.05949187e-01
-8.76512647e-01 9.12081659e-01 -8.87783051e-01 4.34496790e-01
5.63521743e-01 1.42065793e-01 -3.85704905e-01 2.18606070e-02
-1.03907263e+00 4.20917690e-01 2.19154045e-01 3.22551996e-01
-6.42323852e-01 -1.02063215e+00 -8.43057752e-01 1.84167072e-01
-3.53061706e-01 -1.01437604e+00 8.23391020e-01 -6.80856228e-01
-1.37390387e+00 6.78404808e-01 -3.55357051e-01 -9.31954384e-02
6.61717474e-01 7.69756511e-02 -7.44821802e-02 6.40717596e-02
3.94664407e-01 7.77133405e-01 1.05676460e+00 -1.59243953e+00
-7.48799026e-01 -2.76656419e-01 -2.83992738e-01 2.54698455e-01
1.39526933e-01 -6.57804385e-02 -4.15005773e-01 -7.16352522e-01
4.99835581e-01 -5.60970366e-01 -3.75307024e-01 3.80892843e-01
-2.54528672e-01 1.72232743e-02 8.27202439e-01 -8.59536409e-01
5.27998507e-01 -2.37248087e+00 2.62033761e-01 1.36231512e-01
2.59898782e-01 -2.49237105e-01 -1.27286047e-01 -8.76159593e-02
-1.42186418e-01 3.31428528e-01 -6.02698982e-01 -5.10224342e-01
-1.03331052e-01 1.57928780e-01 -7.15734243e-01 3.59878987e-01
5.83674490e-01 9.40888584e-01 -7.73999691e-01 -2.82966048e-01
3.02927226e-01 9.81282771e-01 -5.78347623e-01 3.54532123e-01
-2.04706758e-01 9.23156679e-01 -6.36107683e-01 3.70246410e-01
1.33389544e+00 -2.38846093e-01 -8.11568141e-01 -2.43270054e-01
-5.42827785e-01 -1.35601103e-01 -1.07931149e+00 1.86302423e+00
-5.32549798e-01 2.31338382e-01 2.00944915e-01 -3.25385720e-01
1.14084470e+00 4.16805297e-01 4.02913541e-01 -3.08514386e-01
-2.01101184e-01 3.43662351e-01 -6.10578179e-01 -5.81529886e-02
4.81259018e-01 -6.02752924e-01 6.36411682e-02 -7.71936402e-02
-2.82455593e-01 -4.57473129e-01 -6.52200639e-01 -3.33043396e-01
5.22861481e-01 1.43548667e-01 -4.13377076e-01 2.95759737e-01
7.62352169e-01 -1.80351138e-01 6.04348242e-01 1.67024493e-01
3.80736440e-01 1.21006835e+00 1.08069263e-01 -3.91469628e-01
-1.63396788e+00 -1.46176589e+00 -3.79859626e-01 4.55923080e-01
2.82391191e-01 3.34219545e-01 -4.65531051e-01 -1.51414514e-01
2.68788576e-01 4.74878490e-01 -3.26650262e-01 -8.20341427e-03
-8.04445326e-01 -2.97496527e-01 2.84896433e-01 5.80152810e-01
7.53574729e-01 -1.14431345e+00 -2.13379674e-02 2.56526887e-01
-9.58962068e-02 -1.06428933e+00 -6.06692791e-01 -5.24006426e-01
-1.35501099e+00 -4.78336632e-01 -1.24683344e+00 -1.07492077e+00
9.56726670e-01 2.32429489e-01 7.22051263e-01 -4.77209128e-02
2.29672566e-01 -1.60847187e-01 1.93544269e-01 -6.67018965e-02
-3.11921865e-01 -2.55498439e-01 -8.18753168e-02 1.93179756e-01
-1.11792022e-02 -1.07780600e+00 -6.03488922e-01 2.56343544e-01
-9.91164565e-01 3.59409750e-01 7.09841490e-01 7.46802092e-01
1.04948235e+00 2.40471900e-01 4.39762145e-01 -3.62024724e-01
7.16873586e-01 -5.59011579e-01 -9.54570115e-01 -1.84442222e-01
-7.32027888e-02 7.25045130e-02 7.48749077e-01 -2.43451774e-01
-9.41631973e-01 1.89965636e-01 -3.41496527e-01 -1.25159478e+00
-1.26996726e-01 3.15613985e-01 -2.72903800e-01 -3.52858245e-01
4.28130120e-01 5.69732487e-01 -1.43501133e-01 -7.46668756e-01
6.48438871e-01 4.54540163e-01 8.43345284e-01 -5.24532735e-01
1.33029032e+00 8.21315229e-01 -6.79355338e-02 -5.66342056e-01
-3.44464958e-01 -2.44422972e-01 -8.15289497e-01 1.83959514e-01
1.17021251e+00 -1.20656681e+00 -6.36308312e-01 5.48663437e-01
-1.76240993e+00 1.07838310e-01 -1.92630649e-01 2.72107601e-01
-7.17422724e-01 3.81412089e-01 -7.48254955e-01 -5.43180346e-01
-4.28930104e-01 -1.12215257e+00 1.31548774e+00 2.57151902e-01
4.10970092e-01 -6.77498937e-01 -2.40965965e-04 -1.24665082e-01
1.89529091e-01 6.14469707e-01 6.81065857e-01 1.71621189e-01
-1.46083283e+00 -1.08730204e-01 -6.16207123e-01 5.22660673e-01
2.09464088e-01 -1.43594667e-01 -8.67398679e-01 -1.70843974e-01
1.90949425e-01 2.03346193e-01 6.01147294e-01 3.42985660e-01
1.45322406e+00 -5.10304928e-01 -2.03452364e-01 1.21190488e+00
1.65385556e+00 -4.74296212e-02 6.86199605e-01 2.72933245e-01
1.14336979e+00 3.40390831e-01 3.06132019e-01 3.76279712e-01
3.60891342e-01 4.65639979e-01 5.78478098e-01 -1.42989218e-01
1.58097312e-01 -5.66156745e-01 -1.26140369e-02 8.52868855e-01
-4.43861574e-01 5.29611826e-01 -6.57388389e-01 6.97467923e-01
-1.73663938e+00 -9.68861401e-01 -2.67268764e-03 2.24449468e+00
7.19596267e-01 -1.40945688e-01 -4.09015268e-01 -3.22710901e-01
1.01334512e+00 2.16097862e-01 -5.97099304e-01 -2.37085018e-02
5.23870960e-02 1.74245477e-01 6.67039692e-01 6.40663028e-01
-9.83011842e-01 7.64263630e-01 5.54269981e+00 7.31491983e-01
-1.21333277e+00 1.33177981e-01 1.95103735e-01 1.24268137e-01
-6.71919703e-01 -2.36518756e-01 -5.10571241e-01 8.34271431e-01
4.59169507e-01 -3.04834336e-01 5.85139215e-01 8.19438994e-01
2.26364523e-01 7.50836194e-01 -8.32522869e-01 1.16245317e+00
-3.39993834e-01 -1.76988959e+00 2.57049769e-01 2.98082791e-02
8.77013028e-01 2.15108305e-01 3.04392725e-01 -4.29551750e-02
3.11849982e-01 -1.06133378e+00 7.67938316e-01 9.14718211e-01
1.01909506e+00 -9.86809552e-01 3.65972668e-01 5.43801963e-01
-1.36201799e+00 1.71687737e-01 -8.37886035e-01 1.65683508e-01
3.49024415e-01 4.47209686e-01 -4.45562810e-01 6.77062213e-01
7.14967787e-01 7.52842367e-01 -4.97414507e-02 1.05350518e+00
-3.25885803e-01 -1.25093773e-01 -2.93212324e-01 6.44618452e-01
1.84810206e-01 -7.11380839e-01 6.48067296e-01 7.99749434e-01
7.69130766e-01 2.49746799e-01 -1.18377239e-01 1.74954140e+00
-3.08580726e-01 -2.31446236e-01 -6.74110115e-01 3.85923415e-01
6.45726383e-01 1.25976622e+00 -3.81422229e-02 -1.86772048e-01
-3.44996244e-01 1.10949087e+00 3.32118183e-01 5.60200036e-01
-7.46327758e-01 -4.69778717e-01 9.02636766e-01 2.24799767e-01
4.52877522e-01 -5.19351780e-01 -5.82025588e-01 -1.14491725e+00
2.78655887e-01 -2.66160280e-01 -2.46611848e-01 -1.30455697e+00
-1.56649876e+00 6.62490010e-01 -1.31324202e-01 -1.62799656e+00
-9.13125277e-02 -2.80158311e-01 -9.07472193e-01 1.77170062e+00
-1.81805849e+00 -1.32403564e+00 -5.01875281e-01 7.03983426e-01
1.60059109e-01 5.26563153e-02 5.57992697e-01 3.58309180e-01
-1.89411715e-02 2.13173121e-01 3.17418993e-01 2.59324878e-01
4.88293409e-01 -1.05149651e+00 7.55990684e-01 7.14085519e-01
-4.06129628e-01 6.27355993e-01 3.69798988e-01 -6.63674116e-01
-1.00300503e+00 -1.63028824e+00 5.75103343e-01 -2.48294175e-01
4.13086802e-01 -2.05837473e-01 -1.55607533e+00 8.69132936e-01
-1.03039116e-01 2.48467803e-01 6.97847903e-02 -5.67955673e-01
-2.15353370e-01 -6.88163564e-02 -1.47631741e+00 6.52473569e-01
8.06876123e-01 -8.04912686e-01 -8.04793000e-01 3.20747852e-01
1.25458741e+00 -7.57499993e-01 -1.18930089e+00 4.98739511e-01
9.44273025e-02 -5.70968568e-01 1.51646519e+00 -2.37130865e-01
8.50439906e-01 -7.08516181e-01 8.71075615e-02 -1.31041801e+00
-7.72684216e-01 -2.55349398e-01 2.25358650e-01 1.16877961e+00
2.04090834e-01 -8.50923240e-01 8.60419452e-01 5.63213348e-01
-3.44162107e-01 -5.00006914e-01 -1.09834075e+00 -5.87390363e-01
5.33636093e-01 -8.50175247e-02 1.11933208e+00 9.88211572e-01
-6.54040396e-01 -1.77977476e-02 -3.50196064e-01 9.38306689e-01
9.50483978e-01 3.71360213e-01 6.56405449e-01 -1.13462973e+00
5.52541465e-02 -3.43417048e-01 -3.82510722e-01 -1.19963527e+00
2.72295982e-01 -9.33983147e-01 1.67878307e-02 -1.58927107e+00
-1.16697423e-01 -6.20319843e-01 -1.45920560e-01 2.23344177e-01
-1.03742816e-01 2.66682804e-01 2.28332371e-01 6.54719532e-01
1.89604238e-01 8.48400891e-01 1.84937060e+00 -9.76103023e-02
-1.92724884e-01 -2.88803428e-02 -4.65690643e-01 7.43030071e-01
5.83409250e-01 -4.28277850e-01 -5.67177385e-02 -7.88909912e-01
2.22867206e-02 2.14380518e-01 8.27067912e-01 -9.62285995e-01
3.19414794e-01 -2.74502605e-01 7.57841468e-01 -9.55137372e-01
6.08621895e-01 -1.07016563e+00 5.25515139e-01 1.77477837e-01
-3.94963212e-02 -1.23284280e-01 2.62416482e-01 4.70582962e-01
-4.16950971e-01 -4.31800559e-02 8.16496670e-01 -1.71805605e-01
-3.54070336e-01 9.67817605e-01 2.77360737e-01 -4.62055624e-01
8.14476311e-01 -1.75776109e-01 -8.42365921e-02 2.02928316e-02
-7.12259054e-01 2.36879483e-01 8.46790850e-01 4.37297732e-01
1.02877724e+00 -1.96202600e+00 -9.68374789e-01 5.24480879e-01
1.61230918e-02 9.85430896e-01 4.37450379e-01 2.98549622e-01
-8.28873336e-01 9.03089792e-02 -5.22805333e-01 -7.34025896e-01
-8.23425472e-01 6.23664200e-01 4.52350706e-01 9.02564228e-02
-8.66474152e-01 6.48834050e-01 4.78100568e-01 -7.12651074e-01
-3.94782782e-01 -4.40210104e-01 -5.00816219e-02 -5.35142601e-01
4.59695309e-01 1.63032904e-01 -3.57721388e-01 -6.73402488e-01
-8.53912979e-02 1.00197387e+00 2.52599299e-01 -1.91019237e-01
1.47829330e+00 4.86705154e-02 -4.88497645e-01 3.09241265e-01
1.49475014e+00 1.95897549e-01 -1.73409033e+00 -3.19618076e-01
-4.58409607e-01 -8.55910540e-01 1.61938056e-01 -6.37139156e-02
-1.16431034e+00 9.31638360e-01 4.80147570e-01 1.51206236e-02
9.77000654e-01 -8.31976980e-02 9.23027337e-01 -1.06199011e-01
3.56920034e-01 -3.71021032e-01 -4.17881638e-01 4.75320160e-01
1.37237322e+00 -1.03178442e+00 -7.49847367e-02 -4.91026908e-01
-2.16523319e-01 9.91043985e-01 4.34680879e-01 -7.41603255e-01
5.05865753e-01 2.41295621e-02 -2.05046952e-01 -5.30023724e-02
-4.20689434e-01 2.48574570e-01 3.93067300e-01 4.26666617e-01
1.06066473e-01 5.93366995e-02 1.55723527e-01 5.56685686e-01
-1.99924693e-01 3.25408876e-01 2.39045903e-01 4.57109869e-01
-4.35345232e-01 -9.15541470e-01 -8.45781446e-01 1.13588333e-01
-4.03758064e-02 -1.50569260e-01 1.63324788e-01 3.17848355e-01
2.76102424e-01 3.05111438e-01 5.68645954e-01 -2.63773263e-01
5.06341457e-01 -1.84097126e-01 1.79472476e-01 -4.11645502e-01
-1.31708488e-01 -2.42329147e-02 -7.18472719e-01 -4.43033218e-01
-8.68418813e-02 -4.30550486e-01 -1.49246001e+00 -4.50285196e-01
4.19102237e-02 3.29757631e-02 6.81431174e-01 5.85718095e-01
5.78361988e-01 5.69706380e-01 9.41372871e-01 -1.48368931e+00
-3.37825835e-01 -7.40970731e-01 -5.79319179e-01 4.80295449e-01
6.72266245e-01 -5.20247579e-01 -5.42936981e-01 1.98033586e-01] | [8.396364212036133, -3.599618434906006] |
daa71a4e-99a7-4f97-8f63-406195214872 | perspective-taking-and-pragmatics-for | 2109.08828 | null | https://arxiv.org/abs/2109.08828v2 | https://arxiv.org/pdf/2109.08828v2.pdf | Perspective-taking and Pragmatics for Generating Empathetic Responses Focused on Emotion Causes | Empathy is a complex cognitive ability based on the reasoning of others' affective states. In order to better understand others and express stronger empathy in dialogues, we argue that two issues must be tackled at the same time: (i) identifying which word is the cause for the other's emotion from his or her utterance and (ii) reflecting those specific words in the response generation. However, previous approaches for recognizing emotion cause words in text require sub-utterance level annotations, which can be demanding. Taking inspiration from social cognition, we leverage a generative estimator to infer emotion cause words from utterances with no word-level label. Also, we introduce a novel method based on pragmatics to make dialogue models focus on targeted words in the input during generation. Our method is applicable to any dialogue models with no additional training on the fly. We show our approach improves multiple best-performing dialogue agents on generating more focused empathetic responses in terms of both automatic and human evaluation. | ['Gunhee Kim', 'Byeongchang Kim', 'Hyunwoo Kim'] | 2021-09-18 | null | https://aclanthology.org/2021.emnlp-main.170 | https://aclanthology.org/2021.emnlp-main.170.pdf | emnlp-2021-11 | ['empathetic-response-generation', 'recognizing-emotion-cause-in-conversations'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.74744442e-01 7.63551533e-01 1.28265381e-01 -6.42122507e-01
-5.89070141e-01 -4.58031088e-01 6.09902620e-01 4.54704575e-02
-3.94611567e-01 9.01900470e-01 8.61362338e-01 1.88702390e-01
1.95045233e-01 -5.63625336e-01 7.67767131e-02 -4.43393350e-01
5.85871518e-01 6.85076416e-01 -3.97719622e-01 -7.65983403e-01
2.54294366e-01 1.63181171e-01 -1.10316157e+00 6.12470567e-01
9.51409698e-01 6.02678478e-01 -1.49999475e-02 8.91736388e-01
-2.88600445e-01 1.61267579e+00 -9.44868445e-01 -8.56273651e-01
-3.18992466e-01 -1.14587593e+00 -1.35206938e+00 4.17802967e-02
-3.74487996e-01 -4.78407145e-01 3.50875586e-01 8.63749027e-01
6.80549502e-01 2.95380622e-01 7.80400336e-01 -1.23330355e+00
-4.21222627e-01 1.04365849e+00 -1.08387187e-01 -2.44548127e-01
8.63201439e-01 1.04644001e-01 9.71184075e-01 -5.24254560e-01
5.39403975e-01 1.62012780e+00 4.29374188e-01 1.29730153e+00
-7.56008685e-01 -4.31943208e-01 1.02826156e-01 1.04535565e-01
-5.85358441e-01 -5.51320255e-01 1.04847360e+00 -4.82218266e-01
9.26934004e-01 2.74867117e-01 5.94651699e-01 1.47911251e+00
-2.03057632e-01 9.54145908e-01 1.27179337e+00 -3.94669175e-01
1.77528903e-01 4.08701897e-01 2.65131295e-01 4.62069690e-01
-6.91319704e-01 -3.26472342e-01 -7.23809719e-01 -4.07843769e-01
4.30771828e-01 -5.50265133e-01 -2.31104732e-01 1.00394242e-01
-1.20719516e+00 1.38293993e+00 4.85419184e-02 3.23251277e-01
-8.29724014e-01 1.41876549e-01 5.78255355e-01 3.41425419e-01
5.51235497e-01 9.18541849e-01 -2.78352827e-01 -7.84726441e-01
-3.41626763e-01 3.17251444e-01 1.32409227e+00 5.96785545e-01
4.86787677e-01 -2.19358400e-01 -3.28519881e-01 1.17020524e+00
2.08113611e-01 3.58896703e-01 4.96573538e-01 -1.22090042e+00
1.49616420e-01 6.11285150e-01 3.86229068e-01 -8.47141206e-01
-6.48690999e-01 -4.30225581e-02 -3.48176092e-01 -1.51780099e-01
4.72043186e-01 -7.44381011e-01 -2.94428673e-02 2.11406684e+00
5.76673210e-01 -3.88594359e-01 6.20290995e-01 1.11529005e+00
1.15730703e+00 5.01559079e-01 1.87966079e-01 -3.20230067e-01
1.62097323e+00 -9.85663295e-01 -8.48529041e-01 -2.93799639e-01
9.96659040e-01 -7.14013219e-01 1.07444632e+00 3.31489444e-01
-1.15882838e+00 -1.89906314e-01 -5.84458888e-01 -2.20745102e-01
3.93308252e-02 1.17413059e-01 9.07768667e-01 4.93699133e-01
-6.76343739e-01 2.89593846e-01 -2.21679524e-01 -5.52855253e-01
-9.69819948e-02 1.79116398e-01 -1.75097615e-01 5.51034689e-01
-1.60753512e+00 1.24213648e+00 1.16580036e-02 2.05297936e-02
-4.31859046e-01 -3.90521854e-01 -7.87967682e-01 -3.48850004e-02
1.57948405e-01 -7.82593071e-01 1.67205870e+00 -1.41304195e+00
-2.34148693e+00 9.14522767e-01 -1.31924883e-01 -3.48687261e-01
5.55994451e-01 -3.25005949e-01 -5.56360744e-02 4.14165020e-01
6.65510015e-04 9.52139974e-01 5.77875674e-01 -1.01718736e+00
-2.77623922e-01 -1.03046738e-01 4.61181313e-01 5.80298364e-01
-2.26113051e-01 4.53693449e-01 2.48138696e-01 -4.32523131e-01
-3.83479506e-01 -9.42587256e-01 -8.29828680e-02 -4.05518055e-01
-4.58245605e-01 -7.28170872e-01 3.30438614e-01 -5.00983477e-01
8.01717341e-01 -1.87737727e+00 2.86484540e-01 -1.75206691e-01
3.45441759e-01 1.37032360e-01 -1.80858865e-01 7.59733856e-01
1.05341949e-01 -2.07975850e-01 4.30581979e-02 -4.72487152e-01
3.71177465e-01 4.67947647e-02 -6.49683714e-01 1.12482317e-01
2.56306857e-01 9.04396713e-01 -1.19562030e+00 -6.27520859e-01
4.97189648e-02 2.95793831e-01 -6.57310605e-01 7.72416949e-01
-2.93077171e-01 8.35498989e-01 -6.67877853e-01 -9.70441010e-03
1.55476287e-01 -8.52010101e-02 1.08427830e-01 1.22823194e-01
1.42283767e-01 6.11758947e-01 -5.53050876e-01 1.44970858e+00
-7.47166455e-01 4.77294922e-01 5.65209202e-02 -6.39803827e-01
1.02950406e+00 5.94922006e-01 1.66655734e-01 -3.59198183e-01
4.52667743e-01 9.24133658e-02 4.08279508e-01 -8.85927498e-01
5.06890774e-01 -6.63856089e-01 -4.95947748e-01 1.28422773e+00
-7.93948248e-02 -4.29900587e-01 -5.40805608e-02 4.30140853e-01
7.66857505e-01 9.22937021e-02 5.42622924e-01 -9.53248516e-02
6.31521106e-01 -5.22779711e-02 4.76298302e-01 5.79164922e-01
-3.80250037e-01 1.04196601e-01 1.03793371e+00 -2.93650717e-01
-4.90576059e-01 -4.02489960e-01 2.26371333e-01 1.54129076e+00
8.18246379e-02 -1.72413573e-01 -1.16316915e+00 -5.29798925e-01
-4.82171625e-01 1.17085135e+00 -7.72572339e-01 -9.12943110e-02
-3.53685498e-01 -4.65625197e-01 8.36657703e-01 3.59955698e-01
3.28001261e-01 -1.59673011e+00 -1.07436526e+00 2.91770816e-01
-6.59690976e-01 -1.18890619e+00 -2.31805399e-01 -1.27974838e-01
-3.04317802e-01 -8.05529654e-01 -5.28369009e-01 -3.71281654e-01
4.39878017e-01 -2.93754824e-02 9.70996380e-01 1.58099487e-01
2.06552148e-01 6.07844830e-01 -6.07188523e-01 -6.19795680e-01
-9.13529992e-01 1.49920536e-02 -1.48799911e-01 7.15077482e-03
5.09701371e-01 -3.95319790e-01 -4.04670805e-01 3.09445709e-01
-4.58651036e-01 4.97628868e-01 2.02564508e-01 8.57879639e-01
-4.75554019e-01 -6.18481219e-01 9.80130255e-01 -1.01141274e+00
1.48092449e+00 -5.97027123e-01 3.86528611e-01 2.12256014e-01
8.34121183e-03 -2.28532664e-02 6.93732023e-01 -5.32141089e-01
-1.41961014e+00 -2.37744167e-01 -2.88699210e-01 1.37553275e-01
-3.70416880e-01 1.54355302e-01 -4.36065346e-02 2.36514375e-01
6.60214603e-01 -1.47711441e-01 2.75206566e-01 -4.90182154e-02
6.74415648e-01 1.03835464e+00 2.87601501e-01 -9.34761584e-01
1.09400742e-01 1.91064537e-01 -4.46737230e-01 -7.84697831e-01
-1.13644111e+00 -3.76722515e-01 -3.60776603e-01 -5.38721085e-01
1.08775508e+00 -8.02007914e-01 -1.24314356e+00 2.59098023e-01
-1.65739429e+00 -5.53092897e-01 9.87721328e-03 5.78954697e-01
-9.49247539e-01 3.98522437e-01 -8.97222877e-01 -1.17596579e+00
-5.43593645e-01 -1.02564490e+00 1.02304101e+00 2.54208028e-01
-1.37117445e+00 -1.12992370e+00 2.40019798e-01 7.97532856e-01
3.02514315e-01 3.93865667e-02 9.39740479e-01 -1.15821147e+00
2.87001997e-01 6.56480864e-02 -7.80039583e-04 1.95284963e-01
-2.25594593e-03 -1.47892550e-01 -1.07826877e+00 3.41260612e-01
4.94617194e-01 -1.08444881e+00 2.61284024e-01 -1.58281460e-01
5.39109945e-01 -5.82517207e-01 2.02502891e-01 -1.61187239e-02
5.35166323e-01 5.59380054e-02 4.84210044e-01 -1.36016861e-01
3.45981807e-01 1.48702335e+00 8.21660340e-01 7.68158674e-01
7.45431840e-01 5.49453914e-01 2.14568958e-01 7.30294734e-03
2.51807362e-01 -1.66182771e-01 6.61240041e-01 6.89686596e-01
7.08174556e-02 -3.07259709e-01 -7.20270574e-01 4.97443557e-01
-2.07390475e+00 -1.22759843e+00 -2.40429446e-01 1.49059546e+00
1.23115802e+00 -3.22077602e-01 2.68190265e-01 -3.12754542e-01
7.44220257e-01 2.23071322e-01 -3.94212484e-01 -1.12449992e+00
-3.05297994e-03 7.56939501e-02 -4.58609432e-01 8.91393602e-01
-6.06918275e-01 1.24161601e+00 5.75411510e+00 2.95517445e-01
-9.53147471e-01 1.55397236e-01 5.73947787e-01 -1.06117621e-01
-4.54651028e-01 -3.02192736e-02 -3.48220617e-01 2.35501021e-01
7.40983725e-01 -1.59788892e-01 3.92226785e-01 7.77891755e-01
3.15515071e-01 -1.09983072e-01 -1.26603556e+00 8.53583217e-01
3.45216483e-01 -6.40676320e-01 -1.65246665e-01 -2.27517545e-01
4.44255173e-01 -5.51247835e-01 -2.78829575e-01 4.78521168e-01
5.59084773e-01 -1.05281091e+00 5.18981993e-01 6.93997264e-01
1.80681199e-01 -6.65536284e-01 9.06016588e-01 6.33831024e-01
-2.45251432e-01 1.66568071e-01 -1.57463580e-01 -4.77059633e-01
3.48068833e-01 1.27264723e-01 -1.28707743e+00 8.15856233e-02
-8.85482132e-02 3.68296236e-01 9.92415193e-03 1.15738906e-01
-8.80216777e-01 4.78530288e-01 4.03178409e-02 -7.37935960e-01
2.26502389e-01 -2.00642303e-01 6.64274156e-01 1.23420048e+00
-1.26979901e-02 5.62439501e-01 7.95695260e-02 9.97821271e-01
2.16785073e-02 4.18821484e-01 -3.45236450e-01 -2.36373663e-01
4.23504084e-01 1.55308497e+00 -3.98410261e-01 -4.16982055e-01
1.71224866e-02 1.10444999e+00 4.44648325e-01 1.29598439e-01
-8.97076547e-01 -1.35787964e-01 6.68589294e-01 -4.13128436e-01
-2.06892908e-01 2.33727828e-01 -8.31660703e-02 -1.03133619e+00
-4.12746012e-01 -1.10662901e+00 9.28796455e-02 -1.13541031e+00
-1.41004121e+00 6.88930869e-01 -2.42635876e-01 -7.62058973e-01
-9.51839805e-01 -4.62320566e-01 -8.37201715e-01 8.35705161e-01
-1.14640069e+00 -1.09678292e+00 -3.12735885e-01 3.98680806e-01
5.45794785e-01 1.14872798e-01 1.22103739e+00 -3.56062829e-01
-3.92586768e-01 4.02433336e-01 -8.42990339e-01 2.72104412e-01
1.09444857e+00 -1.21778846e+00 -2.93494351e-02 3.07860196e-01
-2.89471865e-01 7.09566772e-01 1.03899932e+00 -3.86670142e-01
-9.72282350e-01 -4.40985620e-01 1.31134570e+00 -5.77580512e-01
8.98866475e-01 -2.44008973e-01 -6.91368639e-01 4.32799041e-01
6.83004200e-01 -8.17790568e-01 1.28018594e+00 2.86616445e-01
-1.82195127e-01 6.32245064e-01 -1.20577574e+00 8.05002391e-01
9.18559074e-01 -6.73054218e-01 -1.14228809e+00 6.08790696e-01
6.80655003e-01 -4.00987923e-01 -8.15234303e-01 -4.13767397e-02
4.67636496e-01 -1.11712360e+00 3.80020618e-01 -1.01575017e+00
9.41794217e-01 8.62206519e-02 1.40602738e-01 -1.48472726e+00
1.67120397e-01 -1.10822785e+00 1.79563165e-01 1.25046909e+00
2.74474472e-01 -7.10858226e-01 4.56070453e-01 1.15608931e+00
7.36921728e-02 -7.27283657e-01 -5.45149803e-01 -1.09434336e-01
1.36193782e-01 -3.18758190e-01 6.40763640e-01 1.04761946e+00
8.42229903e-01 1.02078414e+00 -6.84221625e-01 -3.36307466e-01
1.24823719e-01 3.25182021e-01 1.17837346e+00 -1.19052541e+00
-3.20885032e-01 -4.70068127e-01 5.52397594e-02 -1.09526491e+00
7.82891512e-01 -5.65634310e-01 4.03424859e-01 -1.39648724e+00
1.78623781e-01 -1.67793334e-01 3.45626622e-01 4.41546082e-01
-4.37342495e-01 -9.65551510e-02 3.26248318e-01 8.42316821e-02
-7.94220448e-01 8.61938298e-01 1.27369106e+00 3.18618804e-01
-9.15202424e-02 -4.38955314e-02 -1.13742352e+00 1.02361357e+00
8.61290097e-01 -4.18239564e-01 -4.43003446e-01 -2.14950070e-01
4.75479782e-01 5.07869005e-01 3.75249624e-01 -3.41895700e-01
3.30237120e-01 -3.97276461e-01 -1.84393570e-01 -9.85019803e-02
6.39773846e-01 -2.73645222e-01 -2.61917025e-01 3.54848236e-01
-9.41702366e-01 -1.37041673e-01 -2.31431171e-01 1.42199829e-01
-1.15415953e-01 -6.43494725e-01 7.59385467e-01 -2.70527512e-01
-4.01822358e-01 -3.85605365e-01 -6.67323589e-01 2.14269534e-01
1.03690529e+00 -8.60870350e-03 -4.03094083e-01 -1.20412672e+00
-6.00028276e-01 3.20470273e-01 4.64236856e-01 3.58769923e-01
4.37020957e-01 -1.03957593e+00 -8.71432245e-01 -3.83000493e-01
9.08561498e-02 -4.73596573e-01 4.31667775e-01 8.81936908e-01
-2.18735725e-01 3.28597546e-01 -1.74392030e-01 -1.62705451e-01
-1.33882868e+00 9.10036117e-02 4.19618040e-01 -2.52249837e-01
-2.89088905e-01 1.08653021e+00 4.44181830e-01 -6.22082472e-01
7.42834285e-02 2.16345116e-01 -6.82226479e-01 5.31440079e-01
6.58068001e-01 1.23001076e-01 -4.26156193e-01 -8.78078580e-01
-2.02671483e-01 1.72276974e-01 -6.76144883e-02 -4.60162908e-01
1.09943962e+00 -2.20365927e-01 -4.32611465e-01 6.52445018e-01
9.22112048e-01 2.98352510e-01 -9.55645621e-01 -6.39828742e-02
-1.41064197e-01 -1.01833507e-01 -4.98039961e-01 -9.87267613e-01
-4.80649948e-01 9.12579060e-01 -2.81299412e-01 3.39976460e-01
7.53541112e-01 1.93282664e-02 9.82423604e-01 4.88002181e-01
3.11357945e-01 -1.36526155e+00 4.07343477e-01 6.39553189e-01
1.12818003e+00 -1.15125871e+00 -2.15588346e-01 -4.83840853e-01
-1.52254999e+00 1.36108732e+00 7.07754731e-01 7.06939586e-03
-4.61650155e-02 1.31748766e-02 6.47183716e-01 -3.67147744e-01
-1.22457325e+00 -1.11153357e-01 -2.75077298e-02 4.32953835e-01
9.01551247e-01 1.85471475e-01 -5.98413527e-01 8.61698329e-01
-7.08482563e-01 -4.10451800e-01 7.92401016e-01 4.52930808e-01
-3.50102365e-01 -1.10937119e+00 -1.04630947e-01 7.09450468e-02
-4.31434989e-01 -3.10581159e-02 -1.30368936e+00 4.55786288e-01
-2.46801913e-01 1.44577730e+00 -1.22616403e-01 -2.27112517e-01
1.50716990e-01 3.27894568e-01 3.49673450e-01 -8.50609422e-01
-1.09632862e+00 -3.05296212e-01 7.35307336e-01 -5.34579217e-01
-7.55336642e-01 -7.33686090e-01 -1.48994923e+00 -3.41469020e-01
-2.22212926e-01 5.49194157e-01 5.06825924e-01 1.29243505e+00
2.54065454e-01 2.58229643e-01 6.76667690e-01 -5.38408577e-01
-7.06400216e-01 -1.37312543e+00 -8.91985372e-02 6.62785530e-01
-9.35914554e-03 -4.76148158e-01 -4.11536187e-01 -2.66005099e-01] | [13.073188781738281, 7.701498031616211] |
8ca77a1c-5bef-42f2-b8ce-2b51ff635848 | dualconv-dual-mesh-convolutional-networks-for | 2103.12459 | null | https://arxiv.org/abs/2103.12459v2 | https://arxiv.org/pdf/2103.12459v2.pdf | Dual Mesh Convolutional Networks for Human Shape Correspondence | Convolutional networks have been extremely successful for regular data structures such as 2D images and 3D voxel grids. The transposition to meshes is, however, not straight-forward due to their irregular structure. We explore how the dual, face-based representation of triangular meshes can be leveraged as a data structure for graph convolutional networks. In the dual mesh, each node (face) has a fixed number of neighbors, which makes the networks less susceptible to overfitting on the mesh topology, and also al-lows the use of input features that are naturally defined over faces, such as surface normals and face areas. We evaluate the dual approach on the shape correspondence task on theFaust human shape dataset and variants of it with differ-ent mesh topologies. Our experiments show that results of graph convolutional networks improve when defined over the dual rather than primal mesh. Moreover, our models that explicitly leverage the neighborhood regularity of dual meshes allow improving results further while being more robust to changes in the mesh topology. | ['Edmond Boyer', 'Jakob Verbeek', 'Adnane Boukhayma', 'Nitika Verma'] | 2021-03-23 | null | null | null | null | ['3d-shape-representation'] | ['computer-vision'] | [-8.09323639e-02 3.64804059e-01 -7.79789016e-02 -2.79731095e-01
-1.59333125e-02 -6.92249417e-01 7.94198513e-01 3.07103932e-01
-1.47600815e-01 4.66568708e-01 2.07964882e-01 -2.19626904e-01
2.58681495e-02 -1.31915295e+00 -8.24107468e-01 -2.64887542e-01
-4.35599416e-01 6.86042666e-01 1.95904478e-01 -3.84386957e-01
-4.44396287e-02 1.09287012e+00 -1.59334922e+00 3.18622351e-01
2.37590045e-01 1.03059566e+00 -5.19048750e-01 4.23654586e-01
-2.86571592e-01 1.01352945e-01 4.77291420e-02 -3.46453190e-01
4.55395818e-01 -2.74240598e-02 -8.47323716e-01 6.45708516e-02
9.68654275e-01 3.87320966e-02 -1.82029441e-01 7.07083642e-01
1.79383114e-01 -8.76730233e-02 7.74812758e-01 -1.09269190e+00
-6.70590997e-01 3.01394969e-01 -6.68505311e-01 -1.35166589e-02
2.16468275e-01 1.23237424e-01 1.10614610e+00 -9.85420287e-01
7.62835145e-01 1.60193849e+00 1.33882558e+00 1.99988678e-01
-1.66816092e+00 -4.51303214e-01 9.66849849e-02 -4.64907318e-01
-1.29233634e+00 -5.24054587e-01 6.93090081e-01 -4.14267391e-01
9.11133349e-01 1.54740989e-01 7.90578067e-01 5.58672905e-01
2.78823167e-01 2.32116848e-01 9.94152725e-01 -1.35528684e-01
-6.09804653e-02 -3.93388212e-01 -1.28146753e-01 1.08239925e+00
2.85070240e-01 8.59528556e-02 -1.47176847e-01 -2.47913033e-01
1.34330738e+00 -7.56591558e-02 -8.33931938e-02 -8.38321984e-01
-1.06859303e+00 7.08571851e-01 7.83886373e-01 2.84150392e-01
-3.48308206e-01 5.87977290e-01 4.20547932e-01 4.09329176e-01
6.69814527e-01 3.90133619e-01 -3.16246867e-01 5.51544502e-02
-7.98826158e-01 3.88715714e-01 8.69960189e-01 7.25127041e-01
1.05742872e+00 7.20510408e-02 3.16270180e-02 6.97715342e-01
2.65630513e-01 3.14649791e-01 -3.02108020e-01 -1.09733510e+00
2.46914282e-01 8.28361571e-01 -2.76281238e-01 -1.37426496e+00
-7.29758024e-01 -2.40054503e-01 -8.45601380e-01 5.94162524e-01
5.72469175e-01 3.58468518e-02 -1.20130968e+00 1.56314647e+00
3.76569867e-01 2.69370586e-01 -3.44626248e-01 6.22775316e-01
9.92315471e-01 1.21923953e-01 -1.02159813e-01 3.67851913e-01
1.27097738e+00 -4.70738888e-01 -1.29983589e-01 1.23120509e-01
5.89916170e-01 -6.53690040e-01 9.35099363e-01 7.13496059e-02
-1.30400264e+00 -3.27322274e-01 -9.28432465e-01 -3.10484350e-01
-4.20841575e-01 -3.37735832e-01 7.54272819e-01 4.60734874e-01
-1.44134629e+00 1.04986453e+00 -9.71874952e-01 -3.85220617e-01
8.51273775e-01 6.09416485e-01 -6.13347709e-01 2.83276141e-02
-7.88818717e-01 7.19884455e-01 1.12240233e-01 -1.71068832e-01
-4.80003178e-01 -1.18327796e+00 -1.07723343e+00 1.95443943e-01
5.79126514e-02 -7.24902868e-01 8.71910512e-01 -8.61343384e-01
-1.22833860e+00 1.09324002e+00 1.17628828e-01 -2.18687907e-01
6.23094499e-01 4.12991941e-01 -1.00852132e-01 5.88482060e-02
1.97861204e-03 8.24838221e-01 8.71929288e-01 -1.24777329e+00
-1.19351119e-01 -5.03603220e-01 4.62684900e-01 3.40160877e-02
-1.45053685e-01 -5.27016461e-01 -3.25394094e-01 -6.24437273e-01
2.04689920e-01 -9.68120039e-01 -2.12753877e-01 5.77473938e-01
-2.93729126e-01 -2.21329615e-01 9.99242961e-01 -1.79396108e-01
7.69934058e-01 -2.13033533e+00 8.42756853e-02 6.45882666e-01
5.39801002e-01 -5.66182919e-02 -2.91123420e-01 3.50035191e-01
-2.21797183e-01 4.63592380e-01 -2.11055428e-01 -2.94007093e-01
-9.08963382e-02 5.07667005e-01 1.21847905e-01 5.79794645e-01
5.39679945e-01 9.71618056e-01 -8.43368471e-01 -2.93895483e-01
1.05196670e-01 6.93511486e-01 -8.19688618e-01 -1.92254648e-01
-3.19936633e-01 2.99252301e-01 -4.22458708e-01 3.76582444e-01
6.95770562e-01 -4.77512330e-01 1.64945081e-01 -3.19934905e-01
-1.18683100e-01 2.92641491e-01 -1.29240453e+00 1.61434734e+00
-7.00536489e-01 4.40725565e-01 4.97496456e-01 -7.11605370e-01
7.47592747e-01 2.90195048e-01 6.51688159e-01 -6.18260920e-01
5.48104830e-02 1.87805817e-01 -9.68368649e-02 -1.33166974e-02
2.63524711e-01 -1.51684850e-01 2.85160601e-01 5.04563391e-01
-2.13675469e-01 -2.95351207e-01 2.05382183e-02 5.04288189e-02
1.02228725e+00 1.39043093e-01 -2.14600246e-02 -7.41925597e-01
2.33925298e-01 -9.82443467e-02 3.01174670e-01 3.49827677e-01
4.89003509e-01 8.24583888e-01 7.09316552e-01 -7.58777380e-01
-1.25641775e+00 -1.12995231e+00 -5.68232656e-01 7.19625294e-01
-8.03003162e-02 -5.56189179e-01 -5.39969683e-01 -6.32654548e-01
6.00942850e-01 1.05830118e-01 -9.20743823e-01 6.73179328e-02
-8.02250803e-01 -2.48218343e-01 6.96570277e-01 7.02190399e-01
3.48267823e-01 -8.65843952e-01 -5.54668069e-01 1.14510305e-01
4.34045225e-01 -9.84852374e-01 -4.99500811e-01 1.74769342e-01
-9.98888910e-01 -1.24685013e+00 -3.84501457e-01 -8.58797610e-01
7.61270523e-01 3.76693942e-02 1.37358618e+00 7.04141974e-01
-1.76050365e-01 4.89365637e-01 -1.73592754e-02 -2.10268393e-01
-4.65479463e-01 1.80751935e-01 -7.93140978e-02 -8.86337683e-02
8.64505209e-03 -1.07998276e+00 -4.73414302e-01 1.93981186e-01
-9.21479464e-01 1.08475663e-01 7.28644356e-02 7.88438797e-01
6.54100835e-01 -2.90019244e-01 3.73407423e-01 -1.24002481e+00
4.28031832e-01 -5.15535474e-01 -5.61493993e-01 -1.28392234e-01
-5.51166475e-01 2.89492965e-01 6.44049704e-01 -1.80434719e-01
-4.29287285e-01 -1.40660107e-02 -1.48627564e-01 -6.13231301e-01
-1.09196335e-01 5.06487310e-01 2.06700534e-01 -5.24386942e-01
7.04760373e-01 -5.18597901e-01 5.36855578e-01 -5.46249330e-01
4.97201860e-01 6.20295852e-03 2.96789855e-01 -8.57050359e-01
7.93146074e-01 7.09750473e-01 6.60287738e-01 -1.01881707e+00
-2.82676250e-01 7.44742379e-02 -7.89372742e-01 1.25409722e-01
7.94433415e-01 -5.60488582e-01 -7.56879151e-01 2.47130916e-01
-1.21138012e+00 -4.60227996e-01 -4.00246650e-01 -5.67831919e-02
-6.17348135e-01 2.73216486e-01 -7.04665303e-01 -2.71268785e-01
-4.79658619e-02 -1.12492275e+00 1.14565444e+00 -1.92030758e-01
-2.35271588e-01 -1.39968526e+00 -1.16930567e-01 -3.69515210e-01
5.29430032e-01 8.56797397e-01 1.48176539e+00 -3.02701950e-01
-5.47190368e-01 1.36166560e-02 -3.59012246e-01 1.63745299e-01
3.59545767e-01 6.39476776e-02 -9.60036993e-01 -3.62581104e-01
-4.89214301e-01 -2.36038089e-01 7.78996706e-01 1.97237253e-01
1.26837814e+00 -1.44946292e-01 -3.79719913e-01 9.54172373e-01
1.37065542e+00 -5.14132440e-01 6.03210211e-01 2.99606565e-03
1.07444918e+00 5.19935131e-01 -2.84490526e-01 1.60845384e-01
4.82877940e-01 6.75867498e-01 6.26039505e-01 -6.29755855e-01
-4.84335035e-01 -1.48474038e-01 -1.66762680e-01 5.32594264e-01
-2.92551249e-01 6.15714863e-02 -1.20720518e+00 4.33088541e-01
-1.48933947e+00 -7.32099116e-01 -4.30813372e-01 2.11873317e+00
7.32839465e-01 1.07408285e-01 1.59760132e-01 -8.91996920e-02
5.10226369e-01 1.78417951e-01 -4.14775401e-01 -6.19530380e-01
-4.18086462e-02 6.37684226e-01 5.23451984e-01 5.56531072e-01
-9.94460166e-01 7.48591125e-01 6.94036293e+00 2.40832314e-01
-1.20507181e+00 -1.31884322e-01 6.86735868e-01 3.76768857e-02
-6.19616210e-01 -9.40327719e-02 -4.41534251e-01 1.57251596e-01
6.62565947e-01 1.20970234e-01 5.56022048e-01 4.77665186e-01
-1.45147726e-01 2.80841023e-01 -1.51307487e+00 8.55018437e-01
-1.71526641e-01 -1.75855196e+00 2.03906894e-01 4.51150745e-01
7.19519377e-01 2.22335547e-01 1.24819197e-01 1.69995110e-02
4.92937148e-01 -1.57958376e+00 5.94780147e-01 2.65051425e-01
1.19719696e+00 -8.48586679e-01 2.67190456e-01 1.47222308e-02
-1.49681413e+00 3.11901331e-01 -2.51292586e-01 -1.08040087e-01
-1.18639402e-01 3.47220004e-01 -8.06587696e-01 4.65722531e-01
7.43321836e-01 8.90199780e-01 -5.44359207e-01 6.83291316e-01
2.62595862e-01 2.23477334e-01 -6.04189575e-01 4.88956958e-01
3.93059939e-01 -2.83732325e-01 4.07539546e-01 1.15948141e+00
2.15499252e-02 1.50517933e-02 5.36405742e-01 1.11334574e+00
-3.39702576e-01 6.06140643e-02 -1.18270731e+00 6.29439205e-02
4.37819779e-01 1.24380541e+00 -9.98794019e-01 8.58499259e-02
-6.02458715e-01 4.15169269e-01 6.26932859e-01 3.68177980e-01
-4.43933845e-01 -1.17857672e-01 9.79370773e-01 7.95283973e-01
4.90039200e-01 -4.84903693e-01 -5.50644696e-01 -7.60872483e-01
4.34940383e-02 -7.63685107e-01 2.34242246e-01 -4.50366229e-01
-1.43031633e+00 5.63107252e-01 -5.58881741e-03 -9.23058510e-01
4.48583625e-02 -7.53219426e-01 -6.91033423e-01 9.14725065e-01
-1.49153399e+00 -1.38418400e+00 -1.75541639e-01 5.56092739e-01
-6.75669685e-02 1.96608275e-01 8.49960625e-01 2.14322150e-01
-8.86922702e-02 6.48589790e-01 -2.50724256e-01 3.65039766e-01
2.73387820e-01 -1.25593364e+00 8.07680786e-01 2.74486452e-01
1.49528563e-01 6.60537362e-01 3.49088132e-01 -6.88909411e-01
-1.62614799e+00 -1.12336528e+00 3.51276219e-01 -5.91551304e-01
5.56747615e-01 -5.12105525e-01 -1.05379593e+00 8.30743492e-01
-4.64849547e-03 5.48848331e-01 2.74624616e-01 4.31736708e-01
-6.09368384e-01 6.87615424e-02 -1.19021392e+00 6.45674825e-01
1.31880665e+00 -5.31867087e-01 -2.68124074e-01 1.67890027e-01
6.27323270e-01 -5.38600206e-01 -1.37994087e+00 5.97899795e-01
6.06300414e-01 -8.51205409e-01 1.14746845e+00 -7.81381726e-01
4.26329404e-01 -1.88131005e-01 -6.25101998e-02 -1.40369773e+00
-5.47048807e-01 -3.37462693e-01 1.88561510e-02 9.10736084e-01
3.96933407e-01 -6.11193955e-01 9.09612656e-01 3.65709215e-01
-2.77982712e-01 -1.06510890e+00 -1.07871449e+00 -6.87261045e-01
5.67589700e-01 -3.58448833e-01 8.85343611e-01 1.15818822e+00
-3.79229188e-01 1.51216432e-01 2.73875594e-01 1.11192167e-02
3.79993111e-01 5.93961403e-02 7.45445371e-01 -1.79371428e+00
-1.31500540e-02 -7.65568078e-01 -6.45757854e-01 -8.30366433e-01
2.64460862e-01 -1.48156023e+00 -2.81340390e-01 -1.41500795e+00
-4.51075852e-01 -1.08320272e+00 6.46219477e-02 8.06348503e-01
2.50574112e-01 6.41823471e-01 9.41102058e-02 -1.41817629e-01
-7.78602287e-02 3.81107450e-01 1.57037449e+00 -6.08721972e-02
-4.54249531e-02 -4.39849019e-01 -3.66598189e-01 8.65632951e-01
6.00241959e-01 -3.34242940e-01 -3.17060590e-01 -8.32611680e-01
3.67248625e-01 -3.30254696e-02 5.50660372e-01 -8.73139262e-01
-8.06244463e-02 -4.45570378e-03 6.15654349e-01 -2.53119707e-01
3.99049371e-01 -8.74469519e-01 3.50182921e-01 2.47451484e-01
-1.06040291e-01 5.62078595e-01 4.78522152e-01 4.51081455e-01
1.39461994e-01 2.05976442e-01 9.95751977e-01 -3.93482804e-01
-1.85517788e-01 6.27798557e-01 8.60928148e-02 2.22399086e-01
8.01296890e-01 -6.02006435e-01 -6.47963732e-02 -2.31840596e-01
-6.57036901e-01 2.74640918e-01 1.09448981e+00 4.68691230e-01
4.60617781e-01 -1.64257634e+00 -6.95912242e-01 5.34668326e-01
-4.47530225e-02 4.07490999e-01 -2.45560065e-01 7.94767380e-01
-6.91862643e-01 -9.28169116e-03 -3.82811755e-01 -8.63580525e-01
-1.16130078e+00 3.62031132e-01 6.20138049e-01 1.70524064e-02
-9.11646426e-01 6.25754595e-01 2.39853874e-01 -5.67769527e-01
6.59504905e-02 -4.88419414e-01 3.45132016e-02 -8.07730258e-02
1.69301584e-01 2.70964921e-01 3.43562633e-01 -8.21937144e-01
-3.65590572e-01 8.02838206e-01 5.50583452e-02 2.95946956e-01
1.58560860e+00 2.27852657e-01 -5.01280904e-01 2.96628326e-01
1.28718376e+00 6.53131083e-02 -1.19983399e+00 -1.04407266e-01
2.45165173e-03 -4.43606555e-01 1.39860064e-02 -3.71075720e-01
-1.43245566e+00 8.15849662e-01 2.92182267e-01 4.35491830e-01
4.72217500e-01 -4.91501614e-02 7.34288752e-01 1.81017235e-01
5.58290243e-01 -6.44015789e-01 1.14420727e-02 8.35297406e-01
9.31016684e-01 -8.27548206e-01 1.78653523e-02 -5.40839791e-01
7.55679384e-02 1.48752904e+00 5.70943296e-01 -4.98541981e-01
1.01291573e+00 4.99164820e-01 -1.16765812e-01 -7.83264697e-01
-5.11676908e-01 -9.55195352e-02 4.67443556e-01 5.81279933e-01
6.81215227e-01 7.31772631e-02 4.73717079e-02 -1.16179965e-01
-2.45865703e-01 -2.87349492e-01 4.01459485e-01 9.24152434e-01
-1.93297848e-01 -1.22671402e+00 -1.66220009e-01 8.25506508e-01
-3.65089267e-01 1.41529297e-03 -3.12641650e-01 1.04827166e+00
2.74829686e-01 5.11322200e-01 5.88409066e-01 -2.47524515e-01
4.56005633e-01 -1.46815358e-02 6.74303412e-01 -1.04821002e+00
-6.21794641e-01 -2.73826361e-01 1.34983525e-01 -6.73236907e-01
-2.79641956e-01 -5.50313175e-01 -1.55443728e+00 -7.17494845e-01
-1.20466528e-02 -9.77615416e-02 4.40247834e-01 6.90355062e-01
6.65299654e-01 3.76632452e-01 3.60719532e-01 -1.23149061e+00
-2.28569031e-01 -5.34755468e-01 -3.32556307e-01 8.12520027e-01
4.54263330e-01 -8.72933626e-01 -4.54524159e-03 -2.42583007e-01] | [8.409745216369629, -3.7405660152435303] |
c8cd18bb-27ed-4ef1-83ae-cb13e675cce5 | mined-semantic-analysis-a-new-concept-space | 1512.03465 | null | http://arxiv.org/abs/1512.03465v3 | http://arxiv.org/pdf/1512.03465v3.pdf | Mined Semantic Analysis: A New Concept Space Model for Semantic Representation of Textual Data | Mined Semantic Analysis (MSA) is a novel concept space model which employs
unsupervised learning to generate semantic representations of text. MSA
represents textual structures (terms, phrases, documents) as a Bag of Concepts
(BoC) where concepts are derived from concept rich encyclopedic corpora.
Traditional concept space models exploit only target corpus content to
construct the concept space. MSA, alternatively, uncovers implicit relations
between concepts by mining for their associations (e.g., mining Wikipedia's
"See also" link graph). We evaluate MSA's performance on benchmark datasets for
measuring semantic relatedness of words and sentences. Empirical results show
competitive performance of MSA compared to prior state-of-the-art methods.
Additionally, we introduce the first analytical study to examine statistical
significance of results reported by different semantic relatedness methods. Our
study shows that, the nuances of results across top performing methods could be
statistically insignificant. The study positions MSA as one of state-of-the-art
methods for measuring semantic relatedness, besides the inherent
interpretability and simplicity of the generated semantic representation. | ['Wlodek Zadrozny', 'Walid Shalaby'] | 2015-12-10 | null | null | null | null | ['implicit-relations'] | ['natural-language-processing'] | [ 0.250764 0.5082999 -0.09519564 -0.23893288 -0.35648647 -0.5564849
1.0415523 1.0312507 -0.4573531 0.56449413 0.7891666 -0.32968047
-0.65847415 -0.9479786 -0.21348159 -0.35172358 -0.06764056 0.3813442
0.14891885 -0.74797046 1.1033715 -0.19489565 -1.9598472 0.44145548
0.9111608 0.5095898 0.32281104 0.32116777 -0.9969038 0.78049856
-0.7228601 -0.46815565 -0.2655031 -0.52683556 -1.1999636 -0.25534755
0.10965797 0.60556084 0.08851232 1.1871698 -0.07469203 0.25608236
0.9899986 -1.4187697 -0.84486187 1.0375481 -0.40087584 0.2582633
0.7383544 -0.7368832 1.5638763 -1.0294269 0.88932645 1.5034022
0.5524115 0.29857033 -1.1236324 -0.59291375 -0.07376416 0.16996783
-1.340962 0.09629441 0.6067789 -0.64412016 1.2742096 0.21569696
0.19936094 0.8651229 -0.02761347 0.25691974 0.93731743 -0.9684873
0.48695838 0.16014001 0.61215603 0.38928464 0.7059123 -0.42652047
-0.7396403 -0.34225634 -0.04245241 -0.1010122 0.16598651 -0.3026143
-1.1515813 1.1158148 0.15468073 0.7552722 -0.25951388 -0.04470409
0.6528108 0.32705414 0.6661311 0.8930363 -0.2940021 -0.11302141
-0.84692985 0.534247 0.69905 1.1755983 0.7186025 -0.5133487
-0.06047357 0.94295496 0.5186237 0.07086266 1.1621585 -0.6398166
0.6001172 0.9479315 -0.28446317 -1.534291 -0.35591203 -0.29904133
-0.16316877 -0.3364536 -0.16536063 0.09121992 -0.5316404 1.5731356
0.04440182 -0.07058488 0.50721085 0.5857129 1.0831183 0.3561315
0.5053335 -0.1111488 1.5060811 -0.7692819 -0.74207103 -0.26523992
1.1715939 -0.7765052 1.1397569 0.02424394 -0.62677026 -0.3692445
-1.1295856 -0.06408236 -1.0548071 -0.45222434 0.7182818 0.6563643
-1.0250385 0.66749334 -0.26818913 -1.0036055 0.2057423 -0.16140391
-0.29747424 0.20454934 -1.4005587 1.0372365 1.1420475 -0.8802764
-0.22559346 -0.7033657 -0.9225201 0.23546678 0.48967543 -0.6517175
0.69775397 -0.8785484 -0.67590797 1.0431379 -0.08834662 -0.635461
-0.2954198 -0.2519073 -0.6452552 0.36800015 0.6962484 0.58015746
0.41305014 -1.3368186 -0.77720255 -0.31233862 0.2005036 0.32728454
-0.74529016 0.41130418 0.03606405 -0.96946794 0.48275638 -0.6274685
-0.05492098 -0.5538099 -0.37141928 -0.6091653 0.69539034 -0.39984146
1.6566854 -1.8700936 -0.02845687 0.3741021 0.5346918 -0.25752038
-0.01364543 1.07568 -0.49593166 0.61755943 -0.3358411 -0.10064579
0.17066593 0.06064798 -0.30801114 -0.01309203 -0.11167144 0.74843293
-1.2801653 -0.69546276 -0.08790744 -0.06570648 -0.4600509 -0.2537423
-0.18154483 -0.25859204 -0.5820667 0.52528685 0.08963912 -0.28041318
0.50816554 -0.0624603 0.05292159 0.7199195 -0.7525837 1.8221483
-0.27943346 0.8119914 -0.80476236 -1.3380717 1.1498392 0.3295869
0.37101585 -0.6771091 0.02459933 0.16056347 -0.05587038 -0.4494034
0.9156605 -0.35844728 -0.23406085 0.66941 0.25961596 -0.06358376
0.5161496 0.82198757 1.0541316 -0.12880443 0.642371 -1.0099263
0.534814 0.38911313 -0.03334432 0.49917027 0.21938643 0.24208933
0.5114298 0.08136763 -0.9019145 -1.1716812 0.02495849 1.2940872
0.27602574 -1.3053823 -0.8707435 -0.64224344 0.10006156 1.2574342
-0.8048672 -0.30160657 -0.0955405 -0.5596647 0.60274327 0.5571225
-0.05944958 -0.9307689 -0.5988061 0.08461671 -0.48877606 -0.9087973
0.08116243 -0.13987099 -0.9955058 -1.2530283 -0.08320646 -0.7801559
0.7132878 0.5272228 1.4800893 0.21023948 -0.20599653 0.5256101
-1.1246594 -0.72737235 -0.48330006 -0.11895773 0.13835205 -0.40983152
0.98327714 -0.57925165 -0.42160603 0.0851708 -1.1016185 -0.04890127
0.05158106 0.50896347 0.10815651 0.438081 0.86128014 -1.0167651
1.2828071 -1.0284832 0.12292074 0.25810292 -1.1128566 0.3196554
0.10324465 0.03550163 -1.0782082 -0.6235976 0.47447202 0.08613434
0.02248772 0.99684584 0.34361592 0.4455159 1.0321044 0.1117032
0.05564719 -0.4072687 0.72896063 0.8096586 0.50542104 -0.87196356
0.6681471 0.5773075 -0.12597051 -0.81911993 -0.957154 -1.0943274
-0.650917 -0.04697501 0.8794958 -0.7646877 -0.28640088 -0.4016507
-1.0405613 0.5031845 -0.1692762 0.43051884 -0.51881313 0.5365618
-0.22313826 -0.6298127 -0.4935044 -0.35894364 0.91335624 0.0376573
-1.1054313 -1.3873491 0.07537005 0.63954216 0.18182383 0.36617225
1.3845395 -1.1306348 0.3330219 0.04373468 -0.14901596 0.09650422
0.3978564 -0.12421062 -0.6582314 -0.24945286 -0.07127584 -0.15220308
0.71692586 -0.08062829 0.9785527 -0.42902088 -0.5437473 -0.34765813
1.5595896 0.20261893 0.63712287 0.8351593 0.5154043 1.333488
1.0106307 0.43646115 0.34740725 0.31611794 0.1490008 0.41526794
-0.0162293 -0.4707514 0.0901144 1.0983995 -0.07876322 -0.15401195
-1.2361726 0.7718214 -1.8686107 -1.0414642 -0.38357157 2.1301463
0.7095516 0.272404 0.01668628 0.42226014 0.8224733 -0.17103635
0.22127213 -0.43263206 -0.15727136 0.49358857 0.2775491 0.24829604
-0.6508016 1.0859543 6.2111588 0.9115367 -0.5449325 0.0453638
0.11351772 0.10882257 -0.69615096 0.36482096 -0.5125346 0.51603746
0.9706167 -0.80611306 -0.1458103 0.7794003 0.17553693 -0.14769188
-1.0262859 0.90358704 0.54683715 -1.4027345 0.44284922 -0.20345148
0.77273166 -0.31717268 -0.34154028 0.27878383 0.34547222 -1.0322926
0.76247805 0.27240193 0.40028003 -0.69931096 0.8240356 -0.0138836
-1.1069359 -0.09153198 -0.53918374 -0.24932513 -0.06991243 0.40105352
-0.8926854 0.797055 0.80021065 0.85344106 -1.0142422 0.42362884
-0.13961017 0.6715838 0.2926291 -0.39800087 0.5035463 -0.24604775
0.5858489 1.5645027 0.37065744 0.06007008 -0.18392289 0.9438484
0.14275427 0.55375755 -0.72031593 -0.5940532 0.83292055 0.9441375
-1.1684941 -0.5027052 -0.48905018 0.6598733 0.28505623 0.15274066
-0.3789925 -0.75885797 0.5127893 0.2351142 -0.06902447 -0.06691753
-0.5277416 -0.8070275 -0.04897973 -0.49688628 0.84367394 -1.0438017
-1.5016176 0.25246596 0.42038348 -1.1964695 -0.29460567 -0.5970712
-0.5336021 0.7380499 -0.96561444 -0.8009479 -0.215025 0.48080108
0.7358109 -0.53895205 0.97555244 -0.10882518 0.20099148 0.16134205
0.23036753 0.07189762 0.6771804 -1.4757198 0.49857584 0.57115024
0.21436007 1.1145624 0.9371135 -0.9997578 -0.7487442 -0.6934773
1.5410236 -0.79343253 1.1837496 -0.27416977 -0.9456835 0.3882087
0.2887514 -0.67185754 1.2933493 0.42023647 -0.8092854 0.40336898
-0.89784354 0.70871 1.2118179 -0.75381273 -1.6827673 0.20584323
1.1283442 0.37749553 -0.68118143 0.06827131 0.29633144 -0.8279506
1.1248192 -1.073085 1.0791079 -0.24240308 -0.49253997 -1.4072417
-0.36779097 -0.15840162 0.02344698 1.4686232 0.60685694 -0.43025088
0.7025276 0.2998837 -0.14735305 -0.21433093 -0.38851398 -0.96910536
0.29013312 -0.48968232 0.52818817 1.6126341 0.9216611 0.7081535
0.4524102 -0.29596218 0.6460716 -0.10236751 0.19143714 -1.5422152
-0.00958521 -0.671371 -0.8544151 -0.01592942 0.3887883 -1.3814402
-0.2371353 -1.7576385 0.4497674 -0.26949227 -0.43380854 0.2652438
-0.34729624 -0.16648297 0.2387189 0.26007694 -0.60956126 0.3354919
0.7214988 -0.09399277 -0.06076384 -0.6612771 -1.2352594 0.67539984
0.9321052 -0.71876746 -1.0782906 -0.1339869 0.6142092 -0.5254631
0.25306568 -0.74930775 0.25033963 -0.28639677 -0.2559679 -0.41120952
-0.20901111 -0.5466362 -0.06102026 0.3652663 -0.8023454 0.3278488
0.14692731 0.721768 -0.55312556 -0.7731851 0.26598674 -0.24068876
-1.0046331 -0.49871534 -0.4160432 0.4459509 0.8247503 -0.5044093
-0.33793002 -0.16419491 -0.3470762 0.02261934 0.09821993 1.0467285
0.53881645 -1.3087462 -0.7669014 -0.5729105 0.82442933 -0.32940054
-0.11137759 0.19224928 -0.59532386 0.72564816 -0.08222142 -0.18843071
-1.4283129 0.7308431 -0.34089124 0.03638921 -0.65179193 0.6030301
0.2917656 -0.42857656 -0.14838459 0.10711512 -0.7286146 0.3951202
0.52354926 0.5520611 0.03476609 -0.62186867 -0.43729088 0.4331878
-0.03382159 -0.3249956 1.1849107 -0.33954257 -0.5947219 0.5315852
1.2209847 -0.14902103 -0.03051415 -0.28128225 1.0195596 -0.35157284
-0.07485115 -0.6560042 -0.28434184 0.526711 0.26193365 0.31099135
0.7860783 0.21712041 0.19781417 0.46980447 0.27302396 -1.4978154
0.33184814 0.51792383 0.84696347 -1.0129796 0.03158016 -0.94654673
-0.5843855 1.173627 0.43698323 0.23308064 0.6696871 -0.17936307
-0.16639161 -0.765163 -0.5638509 -0.1309167 0.44651565 0.6412859
0.90057445 0.19190954 -1.153857 0.85300434 -0.74755126 -0.427534
0.54748404 1.1904516 -0.59636366 -1.0851126 -0.25321603 0.51112837
-0.39688927 -0.58994913 -0.64430076 0.83545583 -0.12161401 1.3432677
0.3290766 -0.4372322 -0.00981861 0.23737216 -0.04970904 -1.1157839
-0.41140017 -0.5109061 0.40639263 -0.34632045 -0.8277979 -0.4904952
-1.6927203 -0.16924272 -0.22376584 0.7136431 0.78744084 1.3594135
0.32621557 0.43870136 0.3313243 0.03357516 -0.12198616 -1.0113693
-0.5303748 1.0092251 -0.4210371 -0.87853295 -0.35289782 0.39240086] | [10.252057075500488, 8.732568740844727] |
08b7a618-92a4-475a-a700-def4eb6d8172 | point-level-temporal-action-localization | 2012.08236 | null | https://arxiv.org/abs/2012.08236v1 | https://arxiv.org/pdf/2012.08236v1.pdf | Point-Level Temporal Action Localization: Bridging Fully-supervised Proposals to Weakly-supervised Losses | Point-Level temporal action localization (PTAL) aims to localize actions in untrimmed videos with only one timestamp annotation for each action instance. Existing methods adopt the frame-level prediction paradigm to learn from the sparse single-frame labels. However, such a framework inevitably suffers from a large solution space. This paper attempts to explore the proposal-based prediction paradigm for point-level annotations, which has the advantage of more constrained solution space and consistent predictions among neighboring frames. The point-level annotations are first used as the keypoint supervision to train a keypoint detector. At the location prediction stage, a simple but effective mapper module, which enables back-propagation of training errors, is then introduced to bridge the fully-supervised framework with weak supervision. To our best of knowledge, this is the first work to leverage the fully-supervised paradigm for the point-level setting. Experiments on THUMOS14, BEOID, and GTEA verify the effectiveness of our proposed method both quantitatively and qualitatively, and demonstrate that our method outperforms state-of-the-art methods. | ['Qi Tian', 'Yanfeng Wang', 'Ya zhang', 'Peisen Zhao', 'Chen Ju'] | 2020-12-15 | null | null | null | null | ['weakly-supervised-action-localization'] | ['computer-vision'] | [ 2.27570921e-01 -1.44302929e-02 -7.56191552e-01 -2.48012125e-01
-9.86901104e-01 -1.82458207e-01 5.54164231e-01 -2.94800643e-02
-3.03592652e-01 4.97082680e-01 1.59234837e-01 1.21027626e-01
1.03695981e-01 -1.42745152e-01 -8.13625872e-01 -6.73293114e-01
-1.56608313e-01 4.26371917e-02 9.13469493e-01 1.97854325e-01
2.07447574e-01 2.54822165e-01 -1.30723417e+00 2.94033170e-01
5.66999793e-01 1.24380183e+00 3.79617751e-01 3.35758865e-01
3.79426718e-01 1.41205549e+00 -2.24077806e-01 -9.51813832e-02
3.67603272e-01 -2.76084781e-01 -6.34302378e-01 3.76536757e-01
5.42925596e-01 -6.21990025e-01 -6.02757573e-01 7.76389718e-01
1.69494212e-01 1.88780740e-01 1.20328695e-01 -1.49741900e+00
-9.46193486e-02 3.24815303e-01 -7.70599484e-01 1.85669512e-01
5.16499579e-01 2.15482131e-01 1.10272145e+00 -1.03117085e+00
6.50459051e-01 1.03250790e+00 6.94464445e-01 2.89235592e-01
-9.04708803e-01 -6.05445206e-01 6.06139183e-01 5.62744617e-01
-1.43097293e+00 -5.32577395e-01 9.12263691e-01 -3.70418876e-01
8.70317101e-01 -1.28901646e-01 6.79705679e-01 1.15712798e+00
5.46156392e-02 1.19751811e+00 8.29062164e-01 -1.49228156e-01
2.46593252e-01 -3.11330944e-01 -2.97942311e-01 9.48082209e-01
-3.04121375e-01 -2.10639182e-02 -9.23465490e-01 5.79631655e-03
9.65820789e-01 4.84958619e-01 -2.84790486e-01 -5.80529332e-01
-1.59485352e+00 5.03727674e-01 4.13019449e-01 1.79518774e-01
-4.52692002e-01 4.47407842e-01 4.50571626e-01 -9.78865921e-02
5.28672099e-01 4.56714556e-02 -5.93846858e-01 -4.64771450e-01
-1.08499944e+00 -9.10800789e-03 4.63466316e-01 1.12925029e+00
8.31734598e-01 -2.35138133e-01 -2.60983407e-01 4.26810384e-01
3.26390564e-01 2.56319970e-01 3.24988276e-01 -1.37897122e+00
6.68767929e-01 6.63087904e-01 2.73485541e-01 -9.97005165e-01
-2.55571306e-01 -1.60929129e-01 -4.69077498e-01 5.57846539e-02
4.43070382e-01 2.75640306e-03 -6.79806471e-01 1.65234482e+00
6.16259933e-01 1.06908429e+00 -1.53967738e-01 1.09510386e+00
3.18869919e-01 5.30473948e-01 6.02898821e-02 -2.75423795e-01
9.85408902e-01 -1.54347897e+00 -6.41322553e-01 -2.46653512e-01
9.11992788e-01 -4.86957103e-01 9.83882308e-01 1.76326677e-01
-9.13760185e-01 -6.04461670e-01 -8.45237255e-01 -1.68671668e-01
1.27089947e-01 5.24197221e-01 4.50171888e-01 -3.11189909e-02
-7.27728844e-01 5.29864132e-01 -1.41304040e+00 -3.76373112e-01
6.29204333e-01 4.06991281e-02 -5.56088388e-01 -1.05814084e-01
-9.33082104e-01 6.20958984e-01 2.58024007e-01 1.32179990e-01
-1.02847207e+00 -6.03988528e-01 -1.09136415e+00 -6.07379042e-02
1.01491320e+00 -3.84651601e-01 1.37868607e+00 -8.53163362e-01
-1.57335472e+00 4.50518072e-01 -4.73775387e-01 -6.42233074e-01
6.14105880e-01 -4.87003297e-01 -2.55229846e-02 5.56308210e-01
3.97436082e-01 7.33414531e-01 8.92702460e-01 -1.07598376e+00
-1.05863833e+00 -1.52462393e-01 3.83697271e-01 2.22169131e-01
-1.75265521e-01 -1.10009588e-01 -7.59141505e-01 -7.52283990e-01
2.64426589e-01 -1.06177831e+00 -2.65864015e-01 3.04072738e-01
-2.52788454e-01 -3.66699994e-01 1.03186679e+00 -5.76815546e-01
1.02119982e+00 -2.13472271e+00 1.86332434e-01 -2.00723276e-01
9.31635797e-02 4.24363539e-02 4.84198444e-02 4.28608745e-01
7.20030069e-02 -4.00132269e-01 9.20171663e-03 -8.99475634e-01
3.59873846e-02 3.19855094e-01 -2.79019684e-01 5.96585870e-01
2.15453282e-01 9.81438935e-01 -1.20996666e+00 -8.17363560e-01
4.77217466e-01 4.05047208e-01 -6.47892118e-01 1.73147887e-01
-3.11002165e-01 6.55089378e-01 -6.36656106e-01 9.26309168e-01
1.21990971e-01 -4.57311183e-01 1.32996112e-01 -4.86888856e-01
-2.43321627e-01 1.93329781e-01 -1.11813927e+00 2.39255095e+00
-2.06456944e-01 4.32473481e-01 -1.32002801e-01 -1.03770304e+00
5.08224487e-01 4.73395735e-01 1.02085102e+00 -4.10259277e-01
-1.65712193e-01 9.10777673e-02 -3.35482061e-01 -4.88109618e-01
3.63035232e-01 7.63608590e-02 -5.90918213e-02 1.56107858e-01
1.59374520e-01 4.65371549e-01 5.32062836e-02 2.62462765e-01
1.41545975e+00 9.06890452e-01 4.84753609e-01 2.60092765e-01
5.12498856e-01 -2.94943731e-02 1.09081566e+00 6.76968336e-01
-6.19238317e-01 5.89382350e-01 3.87672901e-01 -4.13474470e-01
-7.00124741e-01 -6.27809703e-01 1.48685768e-01 1.17359674e+00
4.85851169e-01 -8.21795464e-01 -5.36765993e-01 -1.10411537e+00
-2.19000161e-01 2.43740916e-01 -4.14765656e-01 8.52160156e-02
-7.45465457e-01 7.96224773e-02 3.86985004e-01 8.05067420e-01
6.18802071e-01 -7.78049409e-01 -8.98981988e-01 1.73228726e-01
-4.71391976e-01 -1.69126582e+00 -7.32181787e-01 5.01084365e-02
-9.18110788e-01 -1.10729825e+00 -6.26217782e-01 -5.58068871e-01
6.04486227e-01 4.98821884e-01 6.28946483e-01 1.45385668e-01
1.35168061e-01 4.90316957e-01 -7.00913191e-01 4.71097790e-02
1.77866146e-01 -8.40307474e-02 1.69249743e-01 2.65435338e-01
2.96399653e-01 -6.38395965e-01 -8.97551537e-01 5.71429491e-01
-4.94677246e-01 3.11228693e-01 7.55768955e-01 6.23763621e-01
1.04784501e+00 4.02260870e-02 3.83887678e-01 -4.47566330e-01
-3.66769761e-01 -3.27114314e-01 -3.94345254e-01 2.14710638e-01
-2.67185330e-01 -2.79370308e-01 4.85434741e-01 -3.52660596e-01
-9.98252988e-01 6.70055568e-01 2.05184907e-01 -1.02511322e+00
-3.12655181e-01 3.45771581e-01 -2.81214982e-01 -6.36536554e-02
1.59287333e-01 2.33448744e-01 -2.52631813e-01 -5.47991633e-01
3.81703854e-01 3.03934067e-01 7.27461636e-01 -5.20958781e-01
7.90247202e-01 9.26466525e-01 9.03362185e-02 -4.25299525e-01
-1.24311364e+00 -8.14790666e-01 -1.08138943e+00 -3.90727341e-01
9.56028938e-01 -1.26991189e+00 -6.05912626e-01 4.98137742e-01
-1.10431349e+00 -5.67241251e-01 -3.83069962e-01 6.33288741e-01
-8.86185646e-01 6.72938824e-01 -5.87099135e-01 -7.11989522e-01
1.09267347e-01 -1.08298993e+00 1.71937537e+00 -4.09693271e-02
-2.56469324e-02 -8.01193357e-01 -1.04211561e-01 5.01566350e-01
-3.43360156e-01 3.53111923e-01 -9.80804674e-03 -4.90212053e-01
-9.06339586e-01 -3.10092568e-01 -7.67647028e-02 2.27083921e-01
1.98727816e-01 -3.80310833e-01 -7.96475351e-01 -2.40422964e-01
8.52480456e-02 -4.34554577e-01 8.65058422e-01 1.95731401e-01
1.26452076e+00 -2.59862244e-01 -5.33421636e-01 5.89114666e-01
1.10191596e+00 -7.18517927e-03 4.44754958e-01 3.28532785e-01
1.04668498e+00 2.68430680e-01 1.35741651e+00 5.78085959e-01
6.98860109e-01 1.06035900e+00 4.38119382e-01 1.38553195e-02
-1.31311476e-01 -6.34353876e-01 5.68735719e-01 6.07261956e-01
-2.03890711e-01 -1.14435859e-01 -7.18775451e-01 4.03587162e-01
-2.56211185e+00 -1.23937011e+00 1.73139691e-01 1.89620996e+00
6.78966582e-01 1.22562498e-01 2.37969279e-01 1.27249539e-01
6.48056746e-01 5.08590579e-01 -5.62260985e-01 6.49565935e-01
1.57558441e-01 -3.38311851e-01 4.91660804e-01 3.08666676e-01
-1.54373276e+00 1.03525531e+00 4.93959475e+00 8.28919411e-01
-8.77337158e-01 4.05933142e-01 3.31590235e-01 -2.88860947e-01
5.23316205e-01 3.75291258e-01 -8.25896680e-01 7.25787878e-01
5.53468406e-01 3.06633174e-01 3.29058543e-02 1.03171682e+00
6.77792072e-01 -2.23011181e-01 -1.45044279e+00 1.06075299e+00
1.07217155e-01 -1.33837390e+00 -4.63015854e-01 5.94720580e-02
7.53875077e-01 1.17321931e-01 -3.26874435e-01 2.41736799e-01
-1.03740320e-01 -5.00946939e-01 9.30162609e-01 6.65623605e-01
5.42670548e-01 -5.31693637e-01 5.82324743e-01 5.19333661e-01
-1.63278699e+00 -2.34542459e-01 -2.00336441e-01 -3.09369564e-01
6.87881052e-01 3.38628441e-01 -3.97571594e-01 5.60822785e-01
8.28369617e-01 1.64527655e+00 -3.50262225e-01 9.83220100e-01
-5.12699604e-01 6.74603581e-01 -2.95111358e-01 4.47308064e-01
4.00889546e-01 5.61515130e-02 4.73942369e-01 9.25441265e-01
3.40908527e-01 3.70884180e-01 1.00741065e+00 3.85724485e-01
9.58350226e-02 -1.94052130e-01 -3.37872088e-01 1.88752800e-01
4.96061146e-01 1.20243621e+00 -7.25556552e-01 -3.20475966e-01
-8.03091109e-01 1.23297369e+00 3.07239205e-01 2.97097862e-01
-1.11696684e+00 7.35921711e-02 5.79278588e-01 2.51869321e-01
5.81325114e-01 -5.56385517e-01 5.68590239e-02 -1.48780823e+00
3.12682867e-01 -7.42374122e-01 4.47768092e-01 -7.61793852e-01
-1.06206977e+00 6.36939704e-02 1.15311332e-01 -1.77718163e+00
-2.14455351e-01 -2.48132482e-01 -5.66983402e-01 2.81264454e-01
-1.64971471e+00 -1.48686755e+00 -3.69359821e-01 7.81481445e-01
8.49716723e-01 -2.02393319e-04 4.25207496e-01 2.89437592e-01
-6.82236910e-01 3.74818861e-01 -2.42326334e-01 2.36494005e-01
7.36697257e-01 -1.03583145e+00 -3.48158367e-02 9.93627012e-01
3.11055243e-01 3.26838434e-01 3.60393494e-01 -7.30370045e-01
-1.34397173e+00 -1.27553976e+00 6.94673240e-01 -6.56935036e-01
9.04370904e-01 -2.52234578e-01 -7.86417007e-01 9.26234484e-01
-2.61762738e-01 5.10445893e-01 3.87547970e-01 -1.82682529e-01
-1.67225301e-01 -9.30711329e-02 -6.44685328e-01 5.67128718e-01
1.38846755e+00 -5.54604053e-01 -6.06231391e-01 5.78026175e-01
8.66556704e-01 -5.97103298e-01 -1.00294721e+00 4.10139322e-01
4.71251637e-01 -9.48368788e-01 1.06122553e+00 -1.66433737e-01
4.51826185e-01 -6.79902077e-01 -2.40919232e-01 -7.99402416e-01
-1.38868541e-01 -7.58360267e-01 -7.39807367e-01 1.23759162e+00
-7.42349327e-02 -2.56165504e-01 9.49703336e-01 4.31861252e-01
-2.39868551e-01 -9.63911831e-01 -1.24043345e+00 -8.48879695e-01
-5.45217812e-01 -5.83931088e-01 2.88095653e-01 7.70141602e-01
1.08558550e-01 1.94497198e-01 -8.91456068e-01 4.21943128e-01
5.94789565e-01 2.28062421e-01 1.02530503e+00 -8.79509389e-01
-4.66214895e-01 2.29190513e-02 -8.23347032e-01 -1.74832535e+00
4.47551072e-01 -4.10701483e-01 3.41454715e-01 -1.36278176e+00
1.30646542e-01 -3.95641953e-01 -4.31358814e-01 7.98572659e-01
-1.97018594e-01 2.97818512e-01 4.38322008e-01 5.83157301e-01
-1.34410608e+00 8.21870863e-01 1.11091268e+00 8.10807273e-02
-1.00441322e-01 -4.36233804e-02 -2.15245262e-01 1.08382189e+00
5.94602048e-01 -5.80317318e-01 -6.89207792e-01 -4.10798997e-01
-2.50642419e-01 1.47004902e-01 7.73418307e-01 -1.25341737e+00
5.02914965e-01 -4.13723171e-01 1.43765762e-01 -7.12861776e-01
6.37182534e-01 -1.09834266e+00 -1.85685173e-01 1.60504416e-01
-3.10331285e-01 -1.47354662e-01 -2.80095428e-01 1.10426497e+00
-3.28203261e-01 9.58270654e-02 5.67187786e-01 4.46160650e-03
-1.18188500e+00 6.87949240e-01 6.67168275e-02 -3.55807245e-02
1.40103734e+00 -3.32614779e-01 -1.25395566e-01 -3.60868752e-01
-5.23672879e-01 3.93570155e-01 6.14484131e-01 3.76793057e-01
5.87102234e-01 -1.46765554e+00 -3.41913074e-01 -1.61308527e-01
2.06303805e-01 2.27705777e-01 2.47388318e-01 1.53424048e+00
-6.82906285e-02 3.07273388e-01 1.45949766e-01 -9.77443755e-01
-1.12875831e+00 5.52174628e-01 2.59955585e-01 -2.78246343e-01
-1.11320066e+00 6.70803666e-01 3.35085750e-01 -7.45197944e-03
4.23275024e-01 -3.53232205e-01 4.68883850e-02 -1.48431510e-01
5.21021545e-01 3.86124820e-01 -3.48634154e-01 -1.04537213e+00
-5.38238227e-01 6.26992524e-01 5.04713766e-02 -4.46374789e-02
1.18283582e+00 -2.97686100e-01 2.10804150e-01 5.61113536e-01
1.15226412e+00 -2.92726070e-01 -2.05723715e+00 -4.90643352e-01
2.36187596e-02 -7.79842198e-01 1.12489667e-02 -4.17520881e-01
-9.85952437e-01 7.93252528e-01 2.88014710e-01 -2.80405194e-01
1.17228591e+00 9.01853070e-02 1.10701370e+00 4.35794175e-01
6.95921779e-01 -1.14568055e+00 5.53649604e-01 3.64099413e-01
6.63997471e-01 -1.41945338e+00 4.46933657e-02 -4.72863585e-01
-6.30638659e-01 7.77854383e-01 9.07239079e-01 -1.04228109e-01
5.95599473e-01 2.58822106e-02 -7.96541497e-02 -3.58600318e-02
-8.49254787e-01 -2.74011672e-01 1.72939897e-01 5.21817386e-01
7.41343051e-02 -2.89230347e-01 -1.40927464e-01 5.55791199e-01
4.60515410e-01 3.36355299e-01 1.96309671e-01 1.35168755e+00
-3.69306713e-01 -9.92005587e-01 -1.09881431e-01 1.35157347e-01
-3.32225770e-01 1.25539899e-01 -1.75917670e-01 8.07957649e-01
2.91011512e-01 9.76714790e-01 -6.07114546e-02 -4.25328314e-01
2.52699554e-01 -4.00152914e-02 2.51932591e-01 -6.34297907e-01
-1.98766485e-01 3.00000101e-01 -8.87804851e-02 -1.24468255e+00
-1.08204162e+00 -9.31770265e-01 -1.46382082e+00 9.42926705e-02
-2.90131539e-01 -3.12739387e-02 1.26526803e-01 1.17298377e+00
5.14778614e-01 3.76215667e-01 5.72027564e-01 -1.33783185e+00
-4.32746440e-01 -6.95641279e-01 -4.67242450e-01 4.35476184e-01
4.10949677e-01 -9.20892835e-01 -1.65728837e-01 3.35613310e-01] | [8.401805877685547, 0.5583764314651489] |
f88b40d9-3169-4ed9-94fa-d6fd7af5bfc8 | notes-on-the-interpretation-of-dependence | 2004.07649 | null | https://arxiv.org/abs/2004.07649v1 | https://arxiv.org/pdf/2004.07649v1.pdf | Notes on the interpretation of dependence measures | Besides the classical distinction of correlation and dependence, many dependence measures bear further pitfalls in their application and interpretation. The aim of this paper is to raise and recall awareness of some of these limitations by explicitly discussing Pearson's correlation and the multivariate dependence measures: distance correlation, distance multicorrelations and their copula versions. The discussed aspects include types of dependence, bias of empirical measures, influence of marginal distributions and dimensions. In general it is recommended to use a proper dependence measure instead of Pearson's correlation. Moreover, a measure which is distribution-free (at least in some sense) can help to avoid certain systematic errors. Nevertheless, in a truly multivariate setting only the p-values of the corresponding independence tests provide always values with indubitable interpretation. | ['Björn Böttcher'] | 2020-04-16 | null | null | null | null | ['detection-of-higher-order-dependencies', 'detection-of-dependencies'] | ['methodology', 'methodology'] | [-1.90941721e-01 -1.08992003e-01 -2.75296092e-01 -4.12822008e-01
-2.65323430e-01 -6.30608261e-01 6.46843672e-01 5.14827430e-01
-7.35664785e-01 1.22284973e+00 -6.61584884e-02 -6.63349509e-01
-9.17134941e-01 -7.64532447e-01 -1.27255693e-01 -8.30394745e-01
-4.20855224e-01 4.40464586e-01 1.00383036e-01 -2.21041769e-01
4.30052221e-01 7.08063841e-01 -1.53743744e+00 -4.25404310e-01
1.13999438e+00 6.84982419e-01 -1.40930607e-03 5.84632635e-01
2.63532661e-02 4.59715515e-01 -9.02180612e-01 -7.58335114e-01
-9.08675194e-02 -4.39083874e-01 -4.50725049e-01 -1.24613434e-01
-2.69557983e-01 -5.50852753e-02 3.12119871e-01 9.15515721e-01
1.67119026e-01 1.09477393e-01 1.32040906e+00 -1.29483092e+00
-6.68334067e-01 5.64608395e-01 -5.84488034e-01 5.25499284e-01
4.37001079e-01 -6.25679567e-02 8.34591985e-01 -5.68376064e-01
1.80915728e-01 8.99313629e-01 5.90588570e-01 -1.34423465e-01
-1.35169959e+00 -3.68951380e-01 -3.28753829e-01 5.17083751e-03
-1.51559448e+00 5.20851230e-03 3.41134787e-01 -4.30039346e-01
6.66953743e-01 3.70651096e-01 5.94160020e-01 9.21141207e-01
4.84485149e-01 2.23384470e-01 1.67607319e+00 -5.84111691e-01
1.54655769e-01 5.41824460e-01 2.92700797e-01 -5.70030697e-02
1.16566348e+00 6.38782233e-02 -3.60274464e-02 -2.72566140e-01
8.27175498e-01 -3.06274772e-01 -2.21685395e-01 -4.74852473e-01
-9.09847736e-01 8.42642128e-01 -8.85990039e-02 8.68917346e-01
-1.84450775e-01 -2.62778848e-01 2.52495021e-01 1.86346695e-01
9.10411328e-02 1.62304744e-01 -4.34060186e-01 -2.71245003e-01
-5.97902656e-01 1.43330634e-01 1.12546253e+00 7.02216268e-01
5.13980329e-01 -2.86345363e-01 1.72897339e-01 4.97043550e-01
2.08811358e-01 6.27972484e-01 1.75739288e-01 -4.29474771e-01
3.66171569e-01 2.42231607e-01 2.08055511e-01 -1.10618615e+00
-6.52524054e-01 -5.57799995e-01 -6.93899393e-01 1.78021848e-01
8.82942617e-01 -2.75715828e-01 -3.50741670e-02 1.44674194e+00
-9.98314247e-02 -3.70816231e-01 1.16253905e-01 6.18049860e-01
2.84616292e-01 1.87285990e-01 9.07541215e-02 -5.29454589e-01
1.14466023e+00 -1.94235519e-01 -9.70626712e-01 2.45624766e-01
6.68788910e-01 -9.06340659e-01 8.84291708e-01 5.25118828e-01
-9.73567665e-01 -1.66313082e-01 -9.11276162e-01 3.37044954e-01
-4.71291333e-01 -1.38148814e-01 6.28887713e-01 1.03776896e+00
-6.70896590e-01 8.01179945e-01 -6.90659463e-01 -4.52986151e-01
-2.58306921e-01 4.10779566e-01 -2.65613645e-01 2.30915383e-01
-9.41513956e-01 1.34654415e+00 2.29096696e-01 5.10542952e-02
1.51978314e-01 -2.96379983e-01 -6.28429115e-01 1.78065002e-01
1.48137122e-01 -2.18518138e-01 7.10977018e-01 -7.48864889e-01
-9.99493003e-01 7.97079802e-01 1.07902341e-01 -9.03454721e-02
5.85874081e-01 1.34713009e-01 -7.77243257e-01 1.99685302e-02
7.00942874e-02 -3.17968994e-01 2.06081867e-01 -1.19091547e+00
-4.05759394e-01 -5.02500951e-01 -9.49576944e-02 -1.36429846e-01
3.51677202e-02 1.14491813e-01 -9.71525908e-03 -2.95958191e-01
-1.57039762e-02 -5.61112642e-01 -2.14284927e-01 -5.62354028e-01
-2.40685374e-01 -1.73031449e-01 1.58090502e-01 -4.73563790e-01
1.39589131e+00 -2.02242780e+00 -1.76189750e-01 5.30538619e-01
-7.36266449e-02 -6.48666471e-02 2.68116921e-01 8.14218044e-01
-2.65558809e-01 3.58634889e-02 -2.08423913e-01 2.54475549e-02
6.44978806e-02 5.71788788e-01 1.14650235e-01 9.65924859e-01
2.40954682e-01 5.70703447e-01 -6.70635760e-01 -7.22463250e-01
3.95754695e-01 6.47518158e-01 -1.91363722e-01 -1.65002570e-01
4.50831801e-01 4.77038771e-01 -2.88263321e-01 3.65559906e-01
8.50328207e-01 -9.00603458e-02 4.58432347e-01 1.26831442e-01
-5.24322331e-01 3.00336540e-01 -1.02141404e+00 9.74938154e-01
-2.47059092e-01 5.72791338e-01 -3.52156341e-01 -1.13951886e+00
1.25240850e+00 1.39783084e-01 3.66279691e-01 -5.99728703e-01
3.29874754e-01 6.38811767e-01 5.07365286e-01 -3.75039756e-01
5.42264998e-01 -3.64837348e-01 3.33212651e-02 3.71282578e-01
-1.35056050e-02 1.35222077e-01 3.85506511e-01 -2.34677538e-01
6.95827663e-01 -1.33226112e-01 7.50460446e-01 -7.43494749e-01
6.97150230e-01 -3.92694652e-01 2.64042407e-01 3.71550351e-01
-3.71281058e-02 5.17326236e-01 1.21561313e+00 2.69477844e-01
-8.27141762e-01 -1.41761923e+00 -9.49689984e-01 2.07943037e-01
8.50880519e-04 -2.07544357e-01 -1.76532209e-01 -5.22826076e-01
2.49766484e-01 9.78165567e-01 -8.74585330e-01 8.48424211e-02
-2.07752913e-01 -1.16007173e+00 3.55235249e-01 5.08591592e-01
2.98399869e-02 -4.92828816e-01 -6.55128837e-01 -1.66748494e-01
2.25470856e-01 -6.50055408e-01 4.06927735e-01 5.83155990e-01
-1.01121211e+00 -1.26387012e+00 -8.31827879e-01 -1.42337531e-01
5.30686498e-01 1.53031603e-01 1.12455165e+00 4.65963222e-02
1.11030012e-01 4.08689916e-01 -4.90340501e-01 -1.94068447e-01
-4.29868728e-01 -2.00692430e-01 -1.76696062e-01 -3.91646832e-01
7.94828773e-01 -8.66848826e-01 -4.20976639e-01 4.55130100e-01
-9.03817832e-01 -1.21809053e+00 6.74293041e-01 5.90245605e-01
2.11242780e-01 8.24859962e-02 5.20861447e-01 -6.92917228e-01
8.33061159e-01 -5.45375109e-01 -6.39193416e-01 1.60997421e-01
-9.63279665e-01 6.78844154e-02 3.41334492e-01 -1.06977202e-01
-9.49722767e-01 -7.82995701e-01 8.20926428e-02 5.24319075e-02
-1.78935707e-01 4.76068109e-01 -4.89014424e-02 1.58430234e-01
7.20854998e-01 7.22029135e-02 5.70427999e-02 -5.24856448e-01
-8.72459039e-02 5.47456205e-01 5.28777063e-01 -6.55935526e-01
5.47820032e-01 2.66949058e-01 5.33599734e-01 -9.33364391e-01
-4.80992138e-01 -4.01326299e-01 -9.19466197e-01 -5.88508099e-02
7.05259681e-01 -4.69057292e-01 -8.63575399e-01 2.53947526e-02
-1.02418494e+00 4.65783440e-02 -3.02921653e-01 9.94663954e-01
-4.99855667e-01 6.45585656e-01 -2.52566993e-01 -1.31659591e+00
4.36129451e-01 -8.13596308e-01 3.79319519e-01 3.54007661e-01
-3.56966674e-01 -1.65313923e+00 2.99527526e-01 6.49405792e-02
2.22112343e-01 3.97665352e-01 8.35049212e-01 -9.62391794e-01
1.32239357e-01 -5.43192983e-01 -3.19159597e-01 4.32622492e-01
2.68359214e-01 5.79540610e-01 -7.15557754e-01 -5.84202558e-02
1.23947270e-01 2.03131095e-01 3.88863117e-01 5.88262498e-01
5.69503367e-01 6.26643896e-02 -2.21131459e-01 1.48557127e-01
1.69616723e+00 4.15845871e-01 7.75278986e-01 3.42920870e-01
4.81242277e-02 9.73541439e-01 7.49417722e-01 7.56036818e-01
2.59558380e-01 4.28977430e-01 3.25621068e-01 1.64011702e-01
4.31297451e-01 -9.59446188e-03 8.97875950e-02 8.26879203e-01
-5.71654737e-01 -2.25862414e-01 -7.93145597e-01 2.29455888e-01
-1.51451492e+00 -8.69203031e-01 -1.03866041e+00 2.91094494e+00
3.27735007e-01 3.26655835e-01 5.33377349e-01 3.77287209e-01
8.00726473e-01 -6.32893816e-02 1.84323236e-01 -7.51033247e-01
-5.41331410e-01 6.98414966e-02 7.27350652e-01 5.43935359e-01
-5.31455457e-01 2.39828024e-02 7.64976120e+00 5.57368815e-01
-7.25245059e-01 -4.60692076e-03 3.85185629e-01 2.64349312e-01
-4.22820061e-01 1.92598581e-01 -5.85667193e-01 5.64986408e-01
1.01468098e+00 -3.84438485e-01 -2.32553646e-01 4.59183037e-01
1.66234121e-01 -9.04436648e-01 -8.34583402e-01 6.78401351e-01
-2.38648877e-02 -4.69011754e-01 -6.32064998e-01 4.40982312e-01
5.81827998e-01 -4.00198251e-01 2.83467621e-01 1.08123505e-02
-2.67188232e-02 -1.13051271e+00 2.68942297e-01 5.23031592e-01
5.35779774e-01 -1.08987725e+00 1.23684633e+00 5.57582900e-02
-7.56596386e-01 2.11047038e-01 -5.98500431e-01 -6.18158579e-01
2.39663690e-01 1.10111034e+00 -2.80822963e-01 8.19421113e-01
3.17132413e-01 3.95161748e-01 -4.74312663e-01 1.12217093e+00
-3.65668356e-01 3.68443757e-01 -4.08060282e-01 -4.12257046e-01
2.01462165e-01 -8.32328558e-01 3.30480516e-01 1.29298139e+00
4.97945577e-01 -2.44386882e-01 -7.67007470e-01 8.80854845e-01
9.01425540e-01 4.30371523e-01 -6.94332302e-01 -1.33094564e-01
4.32438821e-01 1.18466389e+00 -1.06898654e+00 -2.36201249e-02
-1.02471805e+00 5.86085260e-01 -3.54247503e-02 1.27158821e-01
-9.15831804e-01 -3.31753731e-01 4.99344647e-01 1.53684452e-01
3.50024074e-01 -3.39655042e-01 -6.30241930e-01 -1.00945032e+00
5.77101558e-02 -1.74925208e-01 5.01190901e-01 -4.18809712e-01
-1.30448306e+00 3.05888802e-01 6.34059429e-01 -1.13583422e+00
-4.56178933e-01 -9.11498427e-01 -6.90046072e-01 1.01027989e+00
-1.06095695e+00 -4.29246426e-01 1.84886336e-01 4.80918825e-01
-5.10507822e-01 8.60209465e-02 7.95165598e-01 2.83143550e-01
-3.53078842e-01 3.92701685e-01 4.83785421e-01 -2.09500372e-01
6.79618120e-01 -1.50658190e+00 -2.44653165e-01 4.90558743e-01
-2.48961166e-01 6.90741360e-01 1.28226066e+00 -4.89958525e-01
-7.85815895e-01 -3.98302302e-02 1.28923428e+00 -4.88397717e-01
1.08091819e+00 5.95331863e-02 -7.39602029e-01 5.52367687e-01
1.39367700e-01 -4.14510518e-01 9.77811754e-01 6.38588667e-01
-1.62733614e-01 -1.07267343e-01 -1.03058493e+00 4.86702651e-01
6.45125508e-01 -1.67841166e-01 -7.84059465e-01 2.12665290e-01
-3.45804207e-02 3.56674522e-01 -1.34181738e+00 2.45254368e-01
7.81392336e-01 -2.05545664e+00 7.00902224e-01 -1.76381320e-01
3.06661338e-01 9.94961262e-02 -1.43250525e-01 -1.03649986e+00
-8.43632668e-02 -2.32251689e-01 6.96693301e-01 1.35642052e+00
5.52607298e-01 -1.24430263e+00 4.12251770e-01 6.61551118e-01
1.98622227e-01 -7.05187082e-01 -1.02008927e+00 -1.30919409e+00
4.98525620e-01 -6.49873435e-01 4.39217806e-01 1.02868652e+00
6.48375213e-01 1.83037847e-01 -7.05615655e-02 -9.40758288e-02
6.51930988e-01 3.76651883e-02 6.20821297e-01 -1.44776452e+00
-5.35290897e-01 -5.22470236e-01 -6.95099473e-01 -5.68692625e-01
-2.77142916e-02 -4.52870876e-01 -5.42250693e-01 -1.19141960e+00
1.55587897e-01 -4.36629206e-01 -2.43449032e-01 -2.71671355e-01
2.27798209e-01 -3.98838073e-02 -2.45241076e-02 8.61429498e-02
-1.05966762e-01 2.00343475e-01 1.21095932e+00 5.45016766e-01
-1.44338921e-01 5.23437500e-01 -6.14938080e-01 6.57558978e-01
8.51372421e-01 -4.77734387e-01 -3.69592339e-01 1.67560801e-01
5.09577274e-01 4.16496873e-01 2.74707437e-01 -8.19337189e-01
-5.66659607e-02 -2.20610827e-01 3.36567402e-01 -5.71805179e-01
2.11226240e-01 -8.81137192e-01 4.07062948e-01 6.14767730e-01
7.13244677e-02 4.44792420e-01 3.91096584e-02 2.58222461e-01
-2.22343788e-01 -9.12334621e-01 5.20105660e-01 2.02029757e-02
7.94149861e-02 -3.61765325e-01 -5.53616226e-01 -3.06616157e-01
9.94251370e-01 -4.60153103e-01 -3.45245928e-01 -4.74390775e-01
-6.94910765e-01 -3.33266705e-02 7.18186140e-01 -1.06136695e-01
1.08326823e-01 -1.08623528e+00 -5.42894781e-01 -1.76364973e-01
9.19132158e-02 -5.62449872e-01 1.74450964e-01 1.65582752e+00
-4.16952908e-01 8.10189009e-01 -4.00513411e-01 -3.06649297e-01
-1.13530374e+00 8.86359036e-01 5.99065833e-02 -2.76650876e-01
-2.11493313e-01 1.98929876e-01 2.44510919e-01 4.36269529e-02
-1.35565758e-01 -2.58741677e-01 -2.78119594e-01 1.92355439e-01
2.63973445e-01 9.39449072e-01 -1.54184312e-01 -7.97224641e-01
-5.88774443e-01 4.78420228e-01 3.14830154e-01 -4.04215127e-01
1.04117525e+00 -3.68841201e-01 -4.00064975e-01 1.07530975e+00
1.10138857e+00 5.51664889e-01 -7.79998004e-01 4.30451512e-01
2.56195307e-01 -5.64817250e-01 -4.81899440e-01 -7.79895604e-01
-6.01082921e-01 7.72613764e-01 7.51958564e-02 8.92652750e-01
1.10444331e+00 7.51996413e-02 -1.99342355e-01 6.56393319e-02
1.61155626e-01 -9.40112293e-01 -3.56910169e-01 1.21115863e-01
6.48226559e-01 -1.00748158e+00 3.34265828e-01 -6.10329330e-01
-5.79668224e-01 1.34470630e+00 1.77339733e-01 -3.15174967e-01
6.82592392e-01 4.79350537e-01 -8.83283913e-02 8.47325996e-02
-5.31459510e-01 -5.36457956e-01 1.90251142e-01 1.12031019e+00
9.45392430e-01 1.88040644e-01 -1.50471461e+00 4.83184278e-01
-2.95191377e-01 -3.21616977e-01 6.03514135e-01 6.53444767e-01
-2.42471904e-01 -1.40619290e+00 -6.76202655e-01 4.47958142e-01
-5.48711300e-01 2.18195140e-01 -3.79326016e-01 1.57996416e+00
-9.46369395e-02 1.19093466e+00 2.77094454e-01 -2.76721895e-01
2.19789445e-01 -1.61100850e-01 7.36927927e-01 -7.84433261e-02
-3.52801025e-01 2.91961163e-01 2.16194794e-01 -1.58605561e-01
-8.13193500e-01 -1.07001436e+00 -8.98847401e-01 -6.44082963e-01
-5.80914021e-01 4.85165000e-01 7.67283857e-01 1.01828074e+00
-4.07590717e-01 1.94195643e-01 4.44921583e-01 -2.96408802e-01
-6.50747180e-01 -1.05108345e+00 -1.40064728e+00 -2.43258458e-02
-9.10464749e-02 -7.94497669e-01 -8.77400458e-01 -6.76873386e-01] | [7.544607639312744, 4.550337314605713] |
6928bbe7-8700-40ea-b8b4-3ee083f219f2 | evolutionary-computation-in-action | 2303.00943 | null | https://arxiv.org/abs/2303.00943v2 | https://arxiv.org/pdf/2303.00943v2.pdf | Evolutionary Computation in Action: Feature Selection for Deep Embedding Spaces of Gigapixel Pathology Images | One of the main obstacles of adopting digital pathology is the challenge of efficient processing of hyperdimensional digitized biopsy samples, called whole slide images (WSIs). Exploiting deep learning and introducing compact WSI representations are urgently needed to accelerate image analysis and facilitate the visualization and interpretability of pathology results in a postpandemic world. In this paper, we introduce a new evolutionary approach for WSI representation based on large-scale multi-objective optimization (LSMOP) of deep embeddings. We start with patch-based sampling to feed KimiaNet , a histopathology-specialized deep network, and to extract a multitude of feature vectors. Coarse multi-objective feature selection uses the reduced search space strategy guided by the classification accuracy and the number of features. In the second stage, the frequent features histogram (FFH), a novel WSI representation, is constructed by multiple runs of coarse LSMOP. Fine evolutionary feature selection is then applied to find a compact (short-length) feature vector based on the FFH and contributes to a more robust deep-learning approach to digital pathology supported by the stochastic power of evolutionary algorithms. We validate the proposed schemes using The Cancer Genome Atlas (TCGA) images in terms of WSI representation, classification accuracy, and feature quality. Furthermore, a novel decision space for multicriteria decision making in the LSMOP field is introduced. Finally, a patch-level visualization approach is proposed to increase the interpretability of deep features. The proposed evolutionary algorithm finds a very compact feature vector to represent a WSI (almost 14,000 times smaller than the original feature vectors) with 8% higher accuracy compared to the codes provided by the state-of-the-art methods. | ['H. R. Tizhoosh', 'Abtin Riasatian', 'Taher Dehkharghanian', 'Shahryar Rahnamayan', 'Azam Asilian Bidgoli'] | 2023-03-02 | null | null | null | null | ['whole-slide-images'] | ['computer-vision'] | [ 3.13365430e-01 -1.07651040e-01 -4.39484380e-02 -4.91870455e-02
-9.26742673e-01 -2.58545816e-01 2.04090849e-01 6.49969876e-01
-5.75203955e-01 5.65483272e-01 -1.95106976e-02 -1.57940984e-01
-8.94374192e-01 -7.59376049e-01 -2.92490065e-01 -1.31562459e+00
-4.04844791e-01 3.96935582e-01 -1.22360438e-01 -1.24778099e-01
2.48148084e-01 8.02804470e-01 -1.71909893e+00 2.50472933e-01
8.96567047e-01 1.17856562e+00 2.36531839e-01 7.70781338e-01
-1.86334223e-01 -9.21853930e-02 -4.79191273e-01 -3.27686101e-01
1.14049315e-01 -1.88731879e-01 -5.09777606e-01 -2.23468617e-01
2.00352073e-01 1.87233165e-01 -1.39027163e-01 1.14600229e+00
8.24661136e-01 -5.85336722e-02 8.56656194e-01 -8.93883109e-01
-3.97636294e-01 4.29367900e-01 -6.15736425e-01 3.00033391e-01
-1.12385511e-01 2.29759350e-01 9.22641516e-01 -7.93515801e-01
1.07914996e+00 9.69918787e-01 7.07733512e-01 3.12856287e-01
-1.17048550e+00 -2.55380511e-01 -2.95102745e-01 3.73335779e-01
-1.68932784e+00 9.39063057e-02 8.01729739e-01 -3.13550562e-01
7.88240731e-01 8.71783018e-01 1.04423726e+00 7.07056999e-01
9.02788222e-01 5.05388677e-01 8.37848246e-01 -4.86522228e-01
4.09703970e-01 8.85492340e-02 2.92332530e-01 1.00521624e+00
4.48830456e-01 7.19860941e-02 -2.88799107e-01 -3.83920491e-01
3.66299242e-01 2.97452867e-01 -2.99302071e-01 -3.52842927e-01
-1.22771585e+00 9.26623881e-01 5.70430696e-01 6.83080375e-01
-3.05439144e-01 -2.48238705e-02 7.05355883e-01 1.52808040e-01
2.87382573e-01 5.38989484e-01 4.20769677e-02 1.13788597e-01
-9.52684462e-01 1.85389206e-01 3.70675862e-01 1.10015631e-01
5.19050360e-01 -2.52494842e-01 -2.40832463e-01 6.92940235e-01
6.93411604e-02 2.81102657e-01 1.03024399e+00 -3.65434945e-01
-7.63253644e-02 1.03920925e+00 -4.95505750e-01 -1.27496254e+00
-1.03182924e+00 -8.34295690e-01 -1.38317764e+00 3.46074760e-01
2.36492366e-01 2.22354159e-01 -8.56676519e-01 1.19953096e+00
6.34805799e-01 -1.52699590e-01 -1.75606050e-02 7.04139054e-01
5.81931412e-01 3.90404910e-01 -1.84898317e-01 -8.03382695e-02
1.74157333e+00 -4.47621316e-01 -5.31494796e-01 3.90870541e-01
1.11034703e+00 -2.01895803e-01 8.95728588e-01 3.36546957e-01
-6.51961207e-01 -2.27702573e-01 -1.14972246e+00 -2.68034786e-02
-6.61315978e-01 5.69589317e-01 6.33542836e-01 6.45046592e-01
-1.17579114e+00 5.01635015e-01 -8.61549377e-01 -4.16959286e-01
7.61483252e-01 5.60118735e-01 -5.18031895e-01 2.00891774e-03
-9.33962941e-01 6.85868859e-01 4.74792689e-01 5.89206330e-02
-5.15854597e-01 -8.54453743e-01 -9.03583705e-01 2.69350201e-01
-9.54686180e-02 -6.23641729e-01 2.59236455e-01 -8.13894033e-01
-1.36218095e+00 9.78266358e-01 3.49942595e-02 -4.85513300e-01
5.84262252e-01 4.73693192e-01 -2.79907376e-01 4.16245520e-01
-2.68641591e-01 5.21174967e-01 7.40362406e-01 -8.00825238e-01
-5.04793584e-01 -6.48605466e-01 -4.23634261e-01 1.78346425e-01
-7.50809789e-01 -3.41714293e-01 -1.90697283e-01 -6.95311546e-01
-7.60424137e-02 -1.01741934e+00 -4.45626467e-01 4.47980493e-01
-4.92710799e-01 -1.68409765e-01 6.13998413e-01 -6.60039127e-01
1.43193913e+00 -2.46281767e+00 5.11660457e-01 6.48287117e-01
5.64963520e-01 3.24356198e-01 -9.08619240e-02 6.69771433e-02
-1.87150806e-01 2.28171512e-01 -3.78636688e-01 -2.48804227e-01
-6.49730265e-02 3.14943455e-02 3.04583818e-01 7.97617078e-01
3.11712205e-01 8.73718798e-01 -7.37540245e-01 -6.97957158e-01
1.16143979e-01 4.35709745e-01 -5.54693580e-01 -1.42925441e-01
2.33010992e-01 4.98031378e-02 -3.56058031e-01 7.46421456e-01
8.71541202e-01 -3.84916931e-01 3.71920973e-01 -4.02752697e-01
-7.29775354e-02 -4.55480427e-01 -1.04902184e+00 1.76817632e+00
-2.95412958e-01 6.12424493e-01 -1.57021880e-01 -9.38152194e-01
9.68260407e-01 -8.59215632e-02 6.19301617e-01 -5.46791434e-01
3.37691456e-01 2.63203830e-01 2.54094750e-01 -4.41791236e-01
2.72082686e-01 3.44723873e-02 -6.90563931e-04 4.12965178e-01
1.65515229e-01 2.21845865e-01 4.11108375e-01 -3.81335504e-02
1.20639348e+00 -4.47982162e-01 5.21767855e-01 -4.96842623e-01
6.12697065e-01 -1.96833024e-03 4.47906673e-01 4.88417745e-01
-2.66400218e-01 4.20271039e-01 5.63934088e-01 -7.70217299e-01
-9.72486794e-01 -7.68920898e-01 -5.01380980e-01 6.34894013e-01
-7.53945783e-02 -1.75212085e-01 -5.18039227e-01 -7.94882059e-01
2.99441397e-01 3.46758336e-01 -1.12862957e+00 -3.60443205e-01
-5.46759367e-01 -1.30959880e+00 5.20655990e-01 2.89816111e-01
2.47653034e-02 -6.38557911e-01 -9.28247869e-01 9.57824737e-02
3.85994554e-01 -1.83713064e-01 -1.63040116e-01 3.55163455e-01
-7.88705766e-01 -1.02727914e+00 -9.21271861e-01 -6.22354925e-01
9.54258084e-01 -1.08471431e-01 5.33262253e-01 1.70768932e-01
-1.23067415e+00 -8.66674036e-02 -2.23374024e-01 -2.00998485e-01
-4.93525118e-01 9.81249660e-02 -8.76741186e-02 1.57575861e-01
2.26134464e-01 -4.21596646e-01 -7.55598724e-01 -9.60346386e-02
-1.14802313e+00 3.02071311e-02 9.39189255e-01 1.21542740e+00
9.80314553e-01 7.98866898e-02 4.79991227e-01 -5.04427254e-01
5.61144233e-01 -4.26169813e-01 -4.37739998e-01 4.28853154e-01
-4.25656617e-01 3.14860374e-01 7.80022085e-01 -3.78896654e-01
-6.84689879e-01 -8.87540430e-02 -9.16851163e-02 -3.55234683e-01
1.68029085e-01 6.91095471e-01 2.44447380e-01 -5.19675136e-01
8.63593221e-01 4.86640334e-01 4.61905539e-01 -2.92982221e-01
1.44603044e-01 4.64815885e-01 5.02293818e-02 -7.34744407e-03
5.05372643e-01 5.97393930e-01 3.54568362e-01 -7.32626021e-01
-7.33718798e-02 -1.84344783e-01 -5.73949218e-01 -1.46407515e-01
6.89684510e-01 -3.83452147e-01 -8.52531135e-01 3.55844289e-01
-7.11003125e-01 2.09816620e-01 -4.62197602e-01 4.84370679e-01
-4.04178500e-01 3.85302335e-01 -5.66839457e-01 -5.81620514e-01
-8.03228974e-01 -1.36802304e+00 1.04575801e+00 3.03704441e-01
-1.88982874e-01 -8.79138410e-01 2.62727499e-01 -1.43509626e-01
4.08346444e-01 5.72755992e-01 1.55188704e+00 -8.51288259e-01
-1.02263592e-01 -6.18351936e-01 -1.92117065e-01 2.05579195e-02
7.11891577e-02 3.87948692e-01 -6.03996158e-01 -5.02955079e-01
-2.20650271e-01 -8.95636231e-02 9.12987232e-01 5.47196329e-01
1.56987154e+00 -1.03093259e-01 -7.30725646e-01 9.31031942e-01
1.72577500e+00 2.59916812e-01 3.92923057e-01 4.70604569e-01
3.20606589e-01 3.43460053e-01 3.98058265e-01 6.22962713e-01
1.68150887e-01 4.91028965e-01 4.06515956e-01 -1.90187797e-01
-1.99742019e-01 2.34567299e-01 -1.00775763e-01 7.92774200e-01
4.37450521e-02 -1.25236154e-01 -1.07259750e+00 5.34472466e-01
-1.43854415e+00 -7.15789735e-01 2.86987782e-01 2.10851073e+00
5.86786985e-01 -4.18557264e-02 -2.10141256e-01 4.55710381e-01
7.63316035e-01 2.22371779e-02 -7.02672184e-01 -5.58669984e-01
-2.98014879e-01 1.95847169e-01 2.84815222e-01 1.00582421e-01
-1.02571785e+00 1.47634760e-01 5.14939213e+00 1.35798740e+00
-1.46669912e+00 -7.43352771e-02 9.58968699e-01 -4.29172426e-01
-2.77882069e-01 -7.03860819e-01 -8.09918404e-01 5.32035470e-01
8.75869513e-01 -4.11782771e-01 -2.70743817e-02 7.43499577e-01
-3.16934846e-02 -1.18684821e-01 -9.48707163e-01 1.10959351e+00
6.23517670e-02 -1.97440994e+00 2.07400411e-01 3.19370955e-01
5.13099670e-01 -2.00579271e-01 4.33547497e-01 1.81976277e-02
-2.27181375e-01 -9.69603360e-01 3.17143828e-01 6.25309229e-01
1.19840658e+00 -1.07517302e+00 1.03845930e+00 -2.62210146e-02
-1.06406045e+00 -4.83287066e-01 -4.01998639e-01 7.50253797e-01
-2.33677223e-01 8.49068165e-01 -9.76644099e-01 8.93035352e-01
6.01321995e-01 5.00562906e-01 -8.30216229e-01 1.19351888e+00
6.35306835e-01 1.76312014e-01 -3.95439088e-01 -4.70509768e-01
2.28791639e-01 9.27053913e-02 6.97722971e-01 1.44157040e+00
7.58686006e-01 -1.27229735e-01 -1.77371293e-01 5.45149744e-01
1.36723053e-02 3.97393405e-01 -4.77737635e-01 -1.50195554e-01
3.74925584e-01 1.57276869e+00 -9.54852283e-01 -1.71160087e-01
1.98898777e-01 8.05034697e-01 3.79185736e-01 -9.99586806e-02
-5.44561386e-01 -7.64333606e-01 6.29970253e-01 -8.09260234e-02
1.85590580e-01 1.04495555e-01 -3.38933557e-01 -9.61492121e-01
-1.65229574e-01 -8.41147304e-01 7.00436115e-01 -1.37890264e-01
-1.11944664e+00 8.90903234e-01 -3.62879395e-01 -1.49980617e+00
-2.77360111e-01 -7.62503088e-01 -6.38606369e-01 6.60180807e-01
-1.46217120e+00 -9.58449900e-01 -2.66853958e-01 3.82942647e-01
2.85322249e-01 -2.08691180e-01 1.16942990e+00 1.59798771e-01
-7.16021299e-01 1.01242256e+00 6.66771114e-01 -1.67224064e-01
3.13810617e-01 -1.08984613e+00 -2.16814131e-01 3.28652024e-01
-3.07098955e-01 2.82635659e-01 3.85300338e-01 -3.83729875e-01
-1.71573305e+00 -1.12474442e+00 5.83479822e-01 2.33442053e-01
5.77679515e-01 -1.73758224e-01 -7.75634587e-01 -3.39598544e-02
-2.02922761e-01 2.10843205e-01 1.28442335e+00 -3.31475019e-01
1.62678540e-01 -2.89390028e-01 -1.59833169e+00 6.49477363e-01
7.02929258e-01 -1.07538395e-01 -5.26987910e-02 2.86076397e-01
4.68149513e-01 -2.49637216e-01 -1.01859045e+00 3.33331883e-01
7.03399837e-01 -7.13720977e-01 8.43678474e-01 -4.29739118e-01
3.13468009e-01 -2.02437386e-01 -9.87931713e-02 -1.37686455e+00
-5.82267284e-01 -1.93452284e-01 -7.97774701e-04 6.34408355e-01
2.61707902e-01 -6.47629201e-01 8.37635756e-01 1.39261916e-01
-1.66414931e-01 -1.55410266e+00 -1.42188120e+00 -5.71798861e-01
-8.99404436e-02 1.11709330e-02 5.83419859e-01 7.79123724e-01
8.44729170e-02 -5.63461483e-01 2.25513160e-01 -1.57553479e-02
7.28821099e-01 2.86177695e-01 2.79206723e-01 -1.30346191e+00
-1.98844582e-01 -8.59386683e-01 -1.20282435e+00 2.81010699e-02
-1.37320697e-01 -1.35146821e+00 -3.36628139e-01 -1.14965439e+00
3.38434875e-01 -3.97453457e-01 -6.23517275e-01 2.63283908e-01
-6.63824901e-02 2.36402705e-01 1.31007940e-01 3.40320617e-02
-3.14108670e-01 4.94339943e-01 1.18718779e+00 -3.31725001e-01
-1.94430068e-01 -4.62505430e-01 -6.40336931e-01 4.60657656e-01
4.99595493e-01 -4.97222632e-01 4.85061631e-02 -1.11232132e-01
1.43075943e-01 -2.28996631e-02 2.55113959e-01 -1.11812377e+00
2.72039533e-01 8.57625827e-02 7.75713742e-01 -7.00620174e-01
2.21236780e-01 -6.20278835e-01 4.35416788e-01 1.06359160e+00
-2.96850860e-01 -2.90911011e-02 2.96977073e-01 6.03233337e-01
-2.64953643e-01 -2.84620225e-01 8.63940597e-01 1.31600976e-01
-3.79728258e-01 3.34911197e-01 -4.08676863e-01 -6.30658150e-01
1.30012679e+00 -5.40751517e-01 -1.91966444e-01 4.64015424e-01
-8.15039933e-01 4.88723628e-02 3.22686702e-01 -5.54571673e-02
9.05131161e-01 -1.38975728e+00 -8.47012997e-01 3.71197969e-01
2.82677650e-01 -1.63880125e-01 8.78238738e-01 1.03682637e+00
-7.38276780e-01 3.09811980e-01 -4.57701057e-01 -7.79797673e-01
-1.40653670e+00 3.18069905e-01 3.40419292e-01 -8.87533069e-01
-6.82514668e-01 1.17836738e+00 8.32554474e-02 -1.85877189e-01
1.51640084e-02 -3.46531153e-01 -3.89362842e-01 4.47275728e-01
7.38081455e-01 4.59736407e-01 4.74777788e-01 -3.09844702e-01
-5.41743755e-01 6.83786988e-01 -3.77966613e-01 2.79494345e-01
1.61004114e+00 1.37393281e-01 -1.87589437e-01 1.55939236e-01
1.67581964e+00 -2.47868165e-01 -8.46944213e-01 -9.76538062e-02
-1.11449160e-01 -4.48646843e-01 2.45676816e-01 -8.83066297e-01
-1.12709260e+00 8.50382507e-01 1.30693877e+00 -7.53268152e-02
1.40735245e+00 -2.75373042e-01 6.50815248e-01 2.71708429e-01
2.11071968e-01 -9.66023624e-01 -1.32825762e-01 -3.20419017e-03
8.75499725e-01 -9.48541224e-01 -1.88105684e-02 -6.82522729e-02
-4.72385824e-01 1.62122905e+00 1.15729563e-01 -1.16751753e-01
6.80265129e-01 3.32012802e-01 -1.75832495e-01 -3.90565336e-01
-7.07041740e-01 9.96764749e-02 3.24234366e-01 4.18765366e-01
9.84311476e-02 2.09076300e-01 -6.02756739e-01 8.14073563e-01
8.28241929e-03 4.65813503e-02 1.99413955e-01 8.22591484e-01
-4.08973485e-01 -7.41696596e-01 -2.09601104e-01 8.46998513e-01
-3.43800545e-01 8.67005512e-02 7.94892386e-02 5.86232126e-01
1.62123874e-01 1.77427664e-01 1.29447207e-01 -5.32678366e-01
-2.19390523e-02 -5.70008792e-02 3.31802100e-01 -1.42586544e-01
-7.25735545e-01 -1.32184520e-01 -3.49433362e-01 -4.06811148e-01
1.58110894e-02 -4.46251869e-01 -1.20551836e+00 -1.29273281e-01
-3.63275170e-01 8.23162124e-02 9.72098589e-01 5.82359195e-01
7.00166523e-01 6.66638851e-01 7.85720050e-01 -1.06249595e+00
-6.00454748e-01 -4.92844284e-01 -7.78097451e-01 2.17469409e-01
3.37511897e-01 -6.94826663e-01 -3.61570090e-01 -3.70271176e-01] | [15.123404502868652, -2.7586817741394043] |
fb0bd82d-93e4-4919-9128-d449f2923d5e | wavemix-lite-a-resource-efficient-neural | 2205.14375 | null | https://arxiv.org/abs/2205.14375v4 | https://arxiv.org/pdf/2205.14375v4.pdf | WaveMix: A Resource-efficient Neural Network for Image Analysis | We propose WaveMix -- a novel neural architecture for computer vision that is resource-efficient yet generalizable and scalable. WaveMix networks achieve comparable or better accuracy than the state-of-the-art convolutional neural networks, vision transformers, and token mixers for several tasks, establishing new benchmarks for segmentation on Cityscapes; and for classification on Places-365, five EMNIST datasets, and iNAT-mini. Remarkably, WaveMix architectures require fewer parameters to achieve these benchmarks compared to the previous state-of-the-art. Moreover, when controlled for the number of parameters, WaveMix requires lesser GPU RAM, which translates to savings in time, cost, and energy. To achieve these gains we used multi-level two-dimensional discrete wavelet transform (2D-DWT) in WaveMix blocks, which has the following advantages: (1) It reorganizes spatial information based on three strong image priors -- scale-invariance, shift-invariance, and sparseness of edges, (2) in a lossless manner without adding parameters, (3) while also reducing the spatial sizes of feature maps, which reduces the memory and time required for forward and backward passes, and (4) expanding the receptive field faster than convolutions do. The whole architecture is a stack of self-similar and resolution-preserving WaveMix blocks, which allows architectural flexibility for various tasks and levels of resource availability. Our code and trained models are publicly available. | ['Amit Sethi', 'Anandu A S', 'Kavitha Viswanathan', 'Pranav Jeevan'] | 2022-05-28 | null | null | null | null | ['spatial-token-mixer'] | ['computer-vision'] | [-5.44499494e-02 -2.55832821e-01 -1.30702332e-01 -1.78809881e-01
-4.78759348e-01 -4.21552122e-01 4.89557087e-01 -5.56450672e-02
-7.34792352e-01 2.19652116e-01 9.20371935e-02 -2.84557700e-01
-8.38244110e-02 -1.14489651e+00 -6.93673611e-01 -6.91305578e-01
-5.42819351e-02 1.42131478e-01 8.07558715e-01 1.23756481e-02
-8.66202191e-02 7.51109838e-01 -1.48931766e+00 3.12766016e-01
6.79443538e-01 1.21194363e+00 2.94199735e-01 5.39715350e-01
-2.55425781e-04 7.78360784e-01 -7.94261172e-02 -4.14919406e-01
4.93931472e-01 -1.85525820e-01 -5.94859719e-01 -3.02063599e-02
8.24151874e-01 -3.59924406e-01 -8.54335189e-01 1.08735263e+00
5.59939384e-01 1.98864773e-01 3.56935710e-01 -1.07667601e+00
-8.00407588e-01 3.50414246e-01 -8.17492127e-01 2.89754242e-01
-4.48335648e-01 3.58742505e-01 7.83073783e-01 -7.98400760e-01
4.65418816e-01 1.11664951e+00 1.13201702e+00 3.42503071e-01
-1.06216860e+00 -7.10045695e-01 1.10307343e-01 2.93796957e-01
-1.28587794e+00 -5.07339120e-01 4.72811967e-01 -1.38738081e-01
1.20563412e+00 3.04117054e-01 9.31336164e-01 9.70215678e-01
4.79557179e-02 7.36217797e-01 8.83037627e-01 -2.53548592e-01
3.74624223e-01 -3.87365013e-01 7.75793046e-02 1.01466632e+00
4.05743390e-01 -1.36947483e-01 -4.84117717e-01 2.06272349e-01
1.05985427e+00 1.62521333e-01 -1.40780330e-01 -1.60100758e-01
-1.23105514e+00 7.18211174e-01 7.31770754e-01 4.27417070e-01
-4.57935989e-01 5.58361411e-01 3.02234322e-01 -2.74662562e-02
4.24342066e-01 2.50328392e-01 -3.65209520e-01 -8.88621435e-02
-1.27429056e+00 1.09348334e-01 6.10253334e-01 7.32276976e-01
7.31835485e-01 2.61947483e-01 -1.06532507e-01 7.54584908e-01
1.78526461e-01 6.36113763e-01 5.02314389e-01 -1.21130037e+00
4.86798435e-01 6.20667040e-01 -2.73757160e-01 -9.87809300e-01
-6.72333300e-01 -6.82774663e-01 -1.30047917e+00 3.65636408e-01
4.36314702e-01 -7.39594549e-02 -1.25479174e+00 1.60210419e+00
5.23267984e-02 5.36019325e-01 -2.52495296e-02 8.74895990e-01
1.10242832e+00 8.16512525e-01 -1.43577792e-02 1.75476402e-01
1.64263654e+00 -1.19810784e+00 -3.08718026e-01 -6.37428463e-01
1.93520069e-01 -6.49170458e-01 9.43087697e-01 1.67944759e-01
-1.29254293e+00 -5.51556945e-01 -1.08828473e+00 -3.43338639e-01
-5.09025216e-01 1.66598976e-01 9.54483390e-01 6.39312565e-01
-1.23508704e+00 4.86535490e-01 -1.21668327e+00 -2.46355996e-01
7.60247767e-01 2.52515733e-01 -4.43652123e-01 -2.29890555e-01
-7.78869748e-01 6.84566915e-01 2.39403665e-01 4.02195118e-02
-6.03137136e-01 -9.62515175e-01 -9.62478161e-01 4.90986437e-01
-2.80596036e-02 -5.81901371e-01 1.00136101e+00 -7.67001987e-01
-1.45048928e+00 6.38161361e-01 -4.67355885e-02 -5.54988325e-01
4.88736123e-01 -2.00485475e-02 -3.01607996e-01 3.49909931e-01
4.47696969e-02 1.04678750e+00 8.02040279e-01 -7.71249592e-01
-7.02829480e-01 -2.38950476e-01 -1.91808015e-01 2.23970130e-01
-5.49580991e-01 -2.14738667e-01 -1.05843985e+00 -5.36642075e-01
1.67527556e-01 -7.19245970e-01 -3.94597769e-01 3.23128760e-01
-1.63946524e-01 9.51543897e-02 9.55044031e-01 -5.19870520e-01
8.46024752e-01 -2.55376267e+00 -2.36055091e-01 1.13389514e-01
4.33178842e-01 6.29458129e-01 -4.42702234e-01 1.53071001e-01
8.06803182e-02 -9.97467712e-02 -3.88169914e-01 -3.33907247e-01
-2.65686661e-01 1.92612082e-01 -6.73329458e-02 5.63210070e-01
2.03495443e-01 1.01842070e+00 -6.37543619e-01 -3.31974715e-01
5.03040433e-01 7.28428781e-01 -5.64879835e-01 -3.41415405e-01
1.90598980e-01 -2.11792849e-02 -3.72894615e-01 5.87700963e-01
8.83637011e-01 -2.57111311e-01 -1.99798092e-01 -3.78180534e-01
-3.77401441e-01 -6.08716905e-02 -1.28126597e+00 1.68071222e+00
-4.25254583e-01 9.76651907e-01 2.51350224e-01 -1.01398242e+00
8.19434702e-01 8.60684067e-02 4.54942286e-01 -1.04436696e+00
3.95928808e-02 1.59004211e-01 -2.70311534e-01 -4.29139763e-01
3.77646178e-01 3.85281920e-01 2.58132517e-01 1.44688845e-01
2.47133553e-01 -1.11584738e-01 4.30689633e-01 -1.55832723e-01
1.31671333e+00 -1.92762554e-01 -3.98673750e-02 -3.24138880e-01
8.71353671e-02 -2.81552151e-02 6.16008818e-01 7.55529463e-01
-5.17791584e-02 7.59813607e-01 2.30796784e-01 -8.00098777e-01
-9.94849920e-01 -1.15819085e+00 -2.49333054e-01 9.22027051e-01
2.78436631e-01 -8.82427171e-02 -7.57423162e-01 -2.01876476e-01
-7.01171085e-02 4.09211993e-01 -5.05099237e-01 1.87763676e-01
-6.90883517e-01 -8.89150620e-01 9.55025494e-01 8.66153061e-01
1.33104002e+00 -9.36335325e-01 -1.02323031e+00 1.54825717e-01
-1.08908348e-01 -1.18210948e+00 -4.57919657e-01 3.82578939e-01
-9.18485582e-01 -9.75172639e-01 -9.54294920e-01 -1.02833617e+00
5.98795235e-01 5.11927783e-01 1.04885089e+00 -7.86133185e-02
-5.26851177e-01 7.30584562e-02 -9.19831917e-03 -1.92384794e-01
2.56856441e-01 1.59617871e-01 -3.54749888e-01 -1.36355385e-01
1.73553720e-01 -7.27040589e-01 -8.34156156e-01 1.57750294e-01
-8.62505674e-01 1.87665135e-01 8.54482949e-01 7.48779416e-01
5.52241147e-01 1.86470523e-01 1.54702470e-01 -6.31323516e-01
3.91373694e-01 -2.05274209e-01 -6.93344057e-01 1.52478412e-01
-4.25246328e-01 -4.71982658e-02 5.48953712e-01 -3.86760771e-01
-1.12536025e+00 1.08763516e-01 -3.91086519e-01 -2.81026453e-01
-1.62108392e-01 2.56336242e-01 1.59683108e-01 -2.59780437e-01
7.54539728e-01 3.41292113e-01 -3.44237573e-02 -3.74159217e-01
5.59977472e-01 4.02228773e-01 9.62972403e-01 -3.13447446e-01
6.91720963e-01 8.21468592e-01 9.49520543e-02 -9.95506704e-01
-5.87071121e-01 -5.16206026e-01 -4.72630650e-01 7.73177519e-02
1.03404546e+00 -1.13222623e+00 -6.99584305e-01 7.76835620e-01
-1.02325916e+00 -4.87222284e-01 -4.35070008e-01 5.12013376e-01
-1.85254902e-01 1.48649812e-01 -9.68659401e-01 -3.70021731e-01
-4.81276274e-01 -1.13443816e+00 7.40707874e-01 7.22972333e-01
1.59768641e-01 -8.44281554e-01 -3.23125005e-01 9.46070328e-02
8.09994161e-01 3.25065911e-01 7.32026756e-01 -1.48909479e-01
-6.48482859e-01 -1.47979930e-01 -7.81432688e-01 2.74682611e-01
-1.75910130e-01 -2.83917896e-02 -9.85267699e-01 -1.69245243e-01
-3.55481327e-01 -2.69854993e-01 1.33641160e+00 8.37808609e-01
1.20254028e+00 -1.65236637e-01 -3.74602258e-01 1.11788630e+00
1.38249934e+00 1.47003122e-03 7.05900848e-01 4.73260939e-01
6.50840938e-01 2.93723047e-01 -1.66508868e-01 2.48627558e-01
4.98291522e-01 3.08328748e-01 5.91756523e-01 -8.84530723e-01
-4.58982378e-01 2.52409242e-02 8.36667567e-02 7.32757449e-01
-2.82581061e-01 -1.69357792e-01 -8.19064736e-01 8.75231385e-01
-2.00122023e+00 -9.89859402e-01 -2.87329942e-01 1.82081807e+00
5.84529519e-01 4.03584950e-02 -3.97773609e-02 7.51590803e-02
4.77603704e-01 3.35108280e-01 -5.59307396e-01 -2.36925170e-01
-1.79257408e-01 4.98211175e-01 9.18292820e-01 1.27457649e-01
-1.09375846e+00 8.17121506e-01 6.35125351e+00 8.85919154e-01
-1.25225985e+00 1.82921410e-01 7.89687276e-01 -1.20967284e-01
8.19606893e-03 -3.21060598e-01 -4.86251265e-01 1.30524963e-01
5.31643093e-01 2.86953717e-01 5.97346067e-01 9.22894835e-01
-6.93154112e-02 9.10002738e-03 -7.12085247e-01 1.02306926e+00
-1.32453561e-01 -1.84330320e+00 -2.45564267e-01 5.40755084e-03
7.06125557e-01 7.75529861e-01 1.78086683e-01 2.19108254e-01
4.13239449e-01 -1.00377166e+00 9.08757389e-01 7.89219737e-02
9.37303960e-01 -9.26335931e-01 8.19882035e-01 2.35475540e-01
-1.41007996e+00 -1.21224321e-01 -6.00964367e-01 2.29495000e-02
-5.31553961e-02 8.06988657e-01 -3.21781263e-03 3.59584451e-01
1.17001534e+00 5.61899662e-01 -5.20863771e-01 1.18421543e+00
-1.04027934e-01 7.01645494e-01 -5.64683020e-01 2.62036294e-01
6.95747435e-01 -2.83308290e-02 1.69652477e-01 1.61299837e+00
4.14383203e-01 -4.08856235e-02 -8.09090736e-04 7.13530481e-01
-2.04682410e-01 -3.23449641e-01 -3.16015571e-01 2.44975999e-01
5.65117002e-01 1.48772407e+00 -1.27550268e+00 -2.25299507e-01
-6.19749665e-01 8.69569540e-01 1.42250106e-01 4.63104248e-01
-7.39067554e-01 -5.67671061e-01 6.51569724e-01 6.82441378e-03
5.15870929e-01 -3.31969589e-01 -4.23231125e-01 -9.86669481e-01
-5.71865439e-02 -5.63322842e-01 2.04685301e-01 -5.67677081e-01
-1.08922648e+00 8.30540299e-01 -1.90412819e-01 -8.81731331e-01
1.43353686e-01 -7.43287683e-01 -7.54836679e-01 6.41590953e-01
-1.76100671e+00 -1.30586958e+00 -6.60618007e-01 7.29124725e-01
5.60628235e-01 -1.15126230e-01 8.41449320e-01 4.04593796e-01
-5.77738464e-01 4.82223600e-01 1.33803234e-01 5.13062656e-01
3.93151373e-01 -1.04763329e+00 9.08893228e-01 1.10791576e+00
2.92280555e-01 4.03258234e-01 2.18583137e-01 -1.37572929e-01
-1.31701028e+00 -1.32571292e+00 5.77281177e-01 1.05915762e-01
4.65714067e-01 -2.84870267e-01 -7.80320287e-01 5.42302728e-01
3.53285670e-01 5.65582573e-01 3.95753831e-01 4.60041054e-02
-5.64593971e-01 -3.34146947e-01 -1.20243680e+00 6.47052586e-01
9.76190329e-01 -2.08130375e-01 -2.33071610e-01 1.42682001e-01
5.57271540e-01 -5.88802159e-01 -5.81338286e-01 2.08298191e-01
5.65688968e-01 -1.22502589e+00 1.18741584e+00 6.62731305e-02
3.57702762e-01 -2.63883829e-01 -5.50219864e-02 -1.22810411e+00
-8.53714943e-01 -4.24664646e-01 -6.16317913e-02 1.08675575e+00
4.04080570e-01 -7.92794406e-01 8.30659270e-01 1.96863085e-01
-3.18131685e-01 -8.64618063e-01 -1.17898059e+00 -7.48201847e-01
-1.64778814e-01 -4.99224633e-01 6.26304984e-01 8.96341324e-01
-5.49605191e-01 1.90937549e-01 -2.66033299e-02 2.86245614e-01
6.63795769e-01 -3.94926667e-02 3.11016083e-01 -1.12883556e+00
-3.72408509e-01 -7.40107954e-01 -3.69892269e-01 -1.13749766e+00
-2.10767388e-01 -7.13234007e-01 -7.62238428e-02 -1.81717300e+00
3.78057882e-02 -4.86037493e-01 -1.68151051e-01 9.84007359e-01
1.42140344e-01 8.47535133e-01 2.39711672e-01 2.84037173e-01
-4.24851865e-01 3.16197842e-01 1.19725931e+00 -2.95415610e-01
-1.28414989e-01 -9.98118594e-02 -5.24122477e-01 9.12289798e-01
6.97280228e-01 -2.38291264e-01 -4.47669715e-01 -8.97651851e-01
5.78440018e-02 -2.34127954e-01 4.43106800e-01 -1.39686382e+00
3.31487298e-01 -6.45361394e-02 6.28163874e-01 -4.01364297e-01
4.34359968e-01 -7.00834513e-01 -9.77961719e-02 6.26021445e-01
7.09344149e-02 8.59995633e-02 5.88032603e-01 2.88167745e-01
-2.56053120e-01 -7.33992755e-02 1.07411933e+00 -1.68425143e-01
-9.61914241e-01 4.09952760e-01 -5.76406240e-01 5.96322864e-02
9.17509675e-01 -3.75038445e-01 -6.15565121e-01 -1.15480058e-01
-1.85144186e-01 3.05387415e-02 2.58016467e-01 3.33527237e-01
4.84238863e-01 -1.14491510e+00 -7.44510293e-01 5.10424554e-01
-8.42626765e-02 4.61610883e-01 3.25313747e-01 7.04941273e-01
-1.10101199e+00 2.03053162e-01 -4.86323208e-01 -7.27936089e-01
-9.26958978e-01 1.93033621e-01 2.64565587e-01 -3.24638277e-01
-1.10278952e+00 1.09695911e+00 1.35388881e-01 -2.61076778e-01
1.63257301e-01 -6.06078148e-01 4.55053598e-02 -7.80143365e-02
7.23284602e-01 4.45262760e-01 2.88468808e-01 -2.99337089e-01
-3.89206499e-01 6.31995440e-01 -2.57336050e-02 -7.89603312e-03
1.61814213e+00 2.06285372e-01 -7.09438547e-02 -1.32194847e-01
8.33969891e-01 -1.46863580e-01 -1.68631792e+00 -1.30690277e-01
-3.22537959e-01 -2.22589791e-01 5.61476707e-01 -6.95864916e-01
-1.55075288e+00 9.07292962e-01 7.19860971e-01 2.80699849e-01
1.37496436e+00 -2.46249706e-01 1.01698577e+00 3.18611413e-01
1.37076870e-01 -9.84107435e-01 -1.24256045e-01 5.52213430e-01
5.56080341e-01 -7.84617007e-01 -1.84399143e-01 -3.29318404e-01
-4.58819717e-01 1.13124418e+00 5.53308070e-01 -2.36639008e-01
5.89430153e-01 7.57736742e-01 6.29268512e-02 -1.88638017e-01
-4.87575740e-01 -2.86681652e-01 2.96578288e-01 8.34449470e-01
2.20085561e-01 9.40995887e-02 1.55559063e-01 3.24789524e-01
-8.73139799e-02 -6.56483471e-02 6.68329149e-02 6.70164049e-01
-4.40948427e-01 -6.14894748e-01 -2.80066490e-01 3.90226930e-01
-3.57693136e-01 -3.00117284e-01 -2.87899897e-02 8.68920505e-01
3.79634678e-01 7.72670627e-01 3.31831783e-01 -1.55906990e-01
2.48642892e-01 -3.55492443e-01 4.39637929e-01 -2.55021334e-01
-5.20921886e-01 2.98220925e-02 -6.97091743e-02 -7.17434943e-01
-2.11226940e-01 -2.63777137e-01 -1.24611521e+00 -4.86198872e-01
-1.23230159e-01 -2.70129919e-01 7.59142935e-01 8.40964675e-01
5.11151075e-01 7.16742456e-01 1.36595443e-01 -1.09657335e+00
-1.96363613e-01 -8.63944411e-01 -4.82494265e-01 1.36503085e-01
1.49413571e-01 -3.83907497e-01 -3.37635763e-02 4.91439030e-02] | [9.227431297302246, 1.0667768716812134] |
5087b6c3-f0c6-47a1-934d-2f4f2f5f7c57 | gcn-sem-at-semeval-2019-task-1-semantic | null | null | https://aclanthology.org/S19-2014 | https://aclanthology.org/S19-2014.pdf | GCN-Sem at SemEval-2019 Task 1: Semantic Parsing using Graph Convolutional and Recurrent Neural Networks | This paper describes the system submitted to the SemEval 2019 shared task 1 {`}Cross-lingual Semantic Parsing with UCCA{'}. We rely on the semantic dependency parse trees provided in the shared task which are converted from the original UCCA files and model the task as tagging. The aim is to predict the graph structure of the output along with the types of relations among the nodes. Our proposed neural architecture is composed of Graph Convolution and BiLSTM components. The layers of the system share their weights while predicting dependency links and semantic labels. The system is applied to the CONLLU format of the input data and is best suited for semantic dependency parsing. | ['Sara Mo{\\v{z}}e', 'Shiva Taslimipoor', 'Omid Rohanian'] | 2019-06-01 | null | null | null | semeval-2019-6 | ['semantic-dependency-parsing'] | ['natural-language-processing'] | [ 4.69196402e-02 7.51337767e-01 -3.33005823e-02 -7.92398036e-01
-3.72637540e-01 -6.74496293e-01 2.56977439e-01 3.72145414e-01
-6.49909556e-01 5.53871810e-01 1.87104478e-01 -5.93928039e-01
3.04218024e-01 -1.01058912e+00 -7.16721237e-01 -3.48893464e-01
-1.30979747e-01 8.10534120e-01 2.52181232e-01 -2.72610962e-01
-2.82254636e-01 1.29351020e-01 -1.05466473e+00 7.58288562e-01
5.70562243e-01 1.07693410e+00 5.38306117e-01 1.11704195e+00
-5.88714659e-01 8.70482147e-01 -6.73768222e-01 -7.89674342e-01
-2.24793687e-01 -4.97789651e-01 -1.35381758e+00 -6.64513707e-01
2.48938173e-01 2.01910362e-01 -1.12280048e-01 1.18838859e+00
1.48188263e-01 2.21469894e-01 5.29255748e-01 -9.73745883e-01
-7.33144701e-01 1.28298235e+00 2.27884784e-01 4.82253522e-01
1.20170452e-01 -5.68194807e-01 1.60469091e+00 -4.89565432e-01
7.46447682e-01 1.45457113e+00 5.51569939e-01 7.67475069e-01
-9.07350421e-01 -4.26060498e-01 2.03613222e-01 4.55397040e-01
-8.47128451e-01 -2.74005413e-01 6.55312300e-01 -2.32533112e-01
1.77310610e+00 -2.18874350e-01 2.54113376e-01 1.14835703e+00
1.75926983e-01 3.57126504e-01 9.19681847e-01 -6.87532663e-01
1.02322809e-01 -6.88285977e-02 8.59881639e-01 9.69498217e-01
7.15026930e-02 -1.59647241e-01 -3.20904762e-01 2.11775735e-01
5.44941604e-01 -7.01445997e-01 3.08717877e-01 4.26123925e-02
-6.44562960e-01 9.84286129e-01 8.89361799e-01 4.99212772e-01
-1.35717392e-01 3.52718085e-01 7.01501667e-01 2.74132371e-01
6.75274551e-01 1.57755278e-02 -9.64906454e-01 4.18174416e-01
-2.95077056e-01 -1.73455879e-01 1.00425982e+00 9.88214672e-01
5.91280997e-01 -7.04775602e-02 1.40827492e-01 9.34027374e-01
4.98758256e-01 2.18525991e-01 4.40081745e-01 -6.19757652e-01
7.64014482e-01 5.88185191e-01 -4.51693565e-01 -7.19115198e-01
-7.42708206e-01 -2.57656187e-01 -3.99626672e-01 6.48530899e-03
3.65120798e-01 -4.74358946e-01 -1.05604565e+00 1.88598549e+00
4.88578826e-02 -2.25795247e-02 6.08872116e-01 4.90358829e-01
1.33129191e+00 6.43399894e-01 9.82556224e-01 2.02067107e-01
1.53777182e+00 -9.10307050e-01 -8.22034061e-01 -5.33872664e-01
1.15386224e+00 -6.25551224e-01 5.83675385e-01 -3.43623906e-01
-9.67532396e-01 -7.07108021e-01 -1.03128183e+00 -4.05639589e-01
-1.01826572e+00 1.63085103e-01 4.95799601e-01 3.16741168e-01
-1.26893890e+00 7.76437819e-01 -8.07829797e-01 -5.43497562e-01
2.63487995e-01 4.18206036e-01 -4.73914981e-01 -1.41639793e-02
-1.45161355e+00 1.43321514e+00 1.03372073e+00 1.81346372e-01
-4.63434726e-01 -2.10814610e-01 -1.23433626e+00 1.47530511e-01
3.24625559e-02 -5.08112848e-01 1.32055330e+00 -1.11198795e+00
-1.22442508e+00 1.19598639e+00 -8.48986506e-02 -6.68858707e-01
-1.11253008e-01 -1.09598063e-01 -4.63467449e-01 -1.09051988e-01
1.97617963e-01 6.76428020e-01 3.24171960e-01 -1.04124284e+00
-8.61119807e-01 -5.14364004e-01 8.73985793e-03 1.87486202e-01
7.22857863e-02 3.24056953e-01 -2.60743648e-01 -4.74147379e-01
-6.37257770e-02 -7.95367956e-01 -2.79845208e-01 -1.02809942e+00
-5.21127462e-01 -6.54937506e-01 6.09760642e-01 -1.12029898e+00
9.21773374e-01 -1.98383617e+00 2.31709555e-01 -2.67152861e-02
-2.43458986e-01 1.31473675e-01 -2.80735940e-01 2.84473121e-01
-4.51270610e-01 3.28302756e-02 -2.32474908e-01 -3.48636061e-01
-4.33969982e-02 7.41499960e-01 1.15406081e-01 3.56955901e-02
1.98886618e-01 1.00212646e+00 -7.14967966e-01 -4.92818743e-01
1.55256763e-01 3.21175456e-01 -1.89218059e-01 7.19100296e-01
-6.23669147e-01 3.32571983e-01 -5.70257723e-01 9.01486576e-02
2.66000718e-01 4.87675369e-02 6.04595661e-01 -3.60872537e-01
1.60592288e-01 8.92555714e-01 -6.08759403e-01 1.94411612e+00
-7.43783891e-01 3.28994364e-01 3.06563452e-02 -1.20247030e+00
1.05828655e+00 4.71382797e-01 -1.45428300e-01 -6.59022868e-01
4.92998123e-01 2.59153843e-01 1.97897345e-01 -4.34921443e-01
1.35704383e-01 -1.71702921e-01 -4.67704594e-01 3.47123832e-01
8.31599832e-01 8.65762681e-02 2.01096267e-01 2.07237422e-01
1.06893432e+00 5.77482462e-01 2.93826461e-01 -4.89035606e-01
6.11238599e-01 2.47409955e-01 2.80502409e-01 1.39921755e-01
1.50453985e-01 -8.75727646e-03 6.34180129e-01 -6.68127775e-01
-9.70755279e-01 -8.87506008e-01 1.53795794e-01 1.56301558e+00
-3.30315918e-01 -3.74525398e-01 -1.03752530e+00 -1.15241122e+00
-3.70092452e-01 1.11937165e+00 -8.02760243e-01 4.12485935e-02
-8.75218451e-01 -6.28842235e-01 6.17776394e-01 7.94376731e-01
3.10858965e-01 -1.53426075e+00 -3.66367757e-01 3.82486910e-01
-2.39875123e-01 -1.69933164e+00 6.73789009e-02 9.77970123e-01
-7.60615230e-01 -1.33575916e+00 1.41276568e-01 -1.39473653e+00
5.51898360e-01 -6.56308889e-01 1.45941257e+00 2.46648267e-02
-5.91720045e-02 1.41014867e-02 -4.17856872e-01 -4.08776969e-01
-7.33054221e-01 3.12227964e-01 -5.12628973e-01 -2.58214653e-01
5.43389618e-01 -5.94559133e-01 -8.82816762e-02 -2.97367036e-01
-3.90166879e-01 1.15888648e-01 2.66513586e-01 5.89451253e-01
5.20223975e-01 -2.22989962e-01 4.41130966e-01 -1.54662633e+00
4.48301613e-01 -5.68987906e-01 -5.28995633e-01 2.70922601e-01
-3.23351622e-01 4.00599301e-01 8.72352898e-01 3.30562174e-01
-1.37332201e+00 3.76351088e-01 -5.47078013e-01 1.72431003e-02
-5.86860001e-01 4.60003883e-01 -4.96149778e-01 4.04698849e-01
3.96248728e-01 -4.63978797e-01 -5.67570269e-01 -8.35107505e-01
7.89191663e-01 4.29483950e-01 8.49749267e-01 -6.61402464e-01
6.60574734e-02 -1.80089265e-01 1.28218770e-01 -4.97505456e-01
-1.32541847e+00 -4.82182503e-02 -1.31359828e+00 1.94401264e-01
1.58240807e+00 -6.89391077e-01 -3.70771497e-01 2.14735553e-01
-1.68543577e+00 -5.23437023e-01 -1.17418692e-01 2.42422149e-01
-5.36847770e-01 -5.42478412e-02 -1.05329931e+00 -3.12970340e-01
-6.05547965e-01 -7.13074625e-01 5.65934479e-01 1.54950321e-01
-7.75562301e-02 -1.56269956e+00 3.13008785e-01 1.75165802e-01
1.20822489e-01 1.31283119e-01 1.70470655e+00 -1.40429759e+00
6.23236410e-02 -1.47833809e-01 -2.78795779e-01 6.21042430e-01
-9.48377997e-02 -2.00953558e-01 -1.16890275e+00 1.85140714e-01
-3.74824196e-01 -2.22275093e-01 7.48994589e-01 4.33032393e-01
8.69642258e-01 -1.26308337e-01 -3.44840199e-01 6.16723180e-01
1.49288142e+00 1.50193676e-01 2.10953191e-01 1.25306562e-01
9.57040370e-01 1.05758643e+00 1.20070286e-01 -2.82478690e-01
7.11915970e-01 5.14087856e-01 7.12330163e-01 9.22122505e-03
-4.75343347e-01 -3.26308995e-01 3.83459300e-01 1.05357111e+00
7.20968097e-02 -3.91289920e-01 -1.04066253e+00 5.19258618e-01
-1.84978402e+00 -2.36857787e-01 -6.79262817e-01 1.61232007e+00
5.50971568e-01 2.15513781e-01 -2.95553297e-01 -3.60317975e-01
9.26895380e-01 9.83561799e-02 4.82387980e-03 -1.34535694e+00
-3.56887989e-02 9.08388078e-01 6.13098264e-01 8.17100585e-01
-1.26264381e+00 1.70501041e+00 6.51674604e+00 3.04529607e-01
-6.07818842e-01 5.09771228e-01 4.95395750e-01 4.69731450e-01
6.18916042e-02 1.06931128e-01 -9.21702147e-01 5.73301576e-02
1.76811028e+00 1.14266723e-01 8.74893367e-02 7.07009017e-01
-3.86872113e-01 1.17807118e-02 -1.25199008e+00 3.06430042e-01
6.08745962e-03 -1.02512372e+00 1.21930979e-01 -5.27462363e-01
1.29631713e-01 4.61422443e-01 -5.23677170e-01 3.72290283e-01
1.19965732e+00 -1.22746873e+00 3.95127416e-01 1.33503571e-01
7.49388933e-01 -8.15265059e-01 1.09065104e+00 6.64574578e-02
-1.24363649e+00 2.01337710e-01 -3.52770507e-01 -1.78927124e-01
1.69252425e-01 2.60565579e-01 -8.30224276e-01 7.00304031e-01
8.16874623e-01 7.79462039e-01 -6.17698371e-01 1.85913324e-01
-9.31893349e-01 7.27423489e-01 -1.43490061e-01 2.27314122e-02
3.38258892e-01 -2.04401925e-01 1.47254810e-01 1.67917299e+00
1.87110603e-02 6.75521418e-02 1.16135225e-01 5.43848515e-01
-3.03491950e-01 3.99800986e-01 -7.24420607e-01 -1.03649259e-01
2.23311782e-01 1.23021960e+00 -7.81918585e-01 -5.62872529e-01
-5.07668078e-01 1.14641047e+00 1.01071322e+00 1.48322165e-01
-4.41291809e-01 -2.88521469e-01 4.58563358e-01 -4.76479352e-01
4.27855819e-01 -9.80890244e-02 -3.38160276e-01 -8.14096689e-01
-3.50562036e-01 -1.19921803e-01 1.08165324e+00 -8.09851110e-01
-1.33598077e+00 1.21599233e+00 -1.58650681e-01 -1.52710184e-01
-3.46650571e-01 -1.07412291e+00 -7.32081890e-01 1.34099674e+00
-1.51473475e+00 -1.57545662e+00 1.09911777e-01 6.84051931e-01
5.22458673e-01 -9.07620341e-02 1.60833859e+00 7.48150200e-02
-5.92340171e-01 2.29233310e-01 -3.03952038e-01 7.41685212e-01
3.85460258e-01 -1.49122036e+00 9.83303428e-01 8.89990807e-01
2.94369280e-01 8.31617787e-02 5.47995567e-01 -7.55923629e-01
-8.11342478e-01 -1.23752260e+00 1.61149478e+00 -6.11586273e-01
1.00779712e+00 -5.22170901e-01 -8.57348561e-01 1.24131083e+00
4.61301297e-01 2.44440645e-01 7.41523921e-01 2.33343616e-01
-5.75271249e-01 2.51171827e-01 -7.68638849e-01 2.72226930e-02
9.90821362e-01 -5.53292930e-01 -9.85838592e-01 2.89018005e-01
9.25548553e-01 -3.96715581e-01 -9.93040681e-01 1.56765938e-01
8.90920907e-02 -4.68027502e-01 6.79864705e-01 -1.18626058e+00
6.23078525e-01 2.27911979e-01 -3.69271159e-01 -1.66704106e+00
-2.82671750e-01 1.82954118e-01 3.01181048e-01 1.25171065e+00
9.80309010e-01 -3.63675237e-01 4.86409992e-01 3.07754606e-01
-3.62465143e-01 -1.14163257e-01 -1.05542386e+00 -3.60006869e-01
2.71097600e-01 -5.49064219e-01 2.26529688e-01 8.59035492e-01
-5.72303720e-02 1.07350993e+00 1.02903411e-01 3.18947941e-01
6.86696231e-01 9.40126926e-02 8.68215188e-02 -1.41539633e+00
-2.51673460e-01 -1.53836355e-01 -1.44566953e-01 -3.94561112e-01
8.80395889e-01 -1.51323581e+00 1.95785761e-01 -1.85081542e+00
-1.87616840e-01 -6.81414068e-01 -5.17417490e-01 8.87148023e-01
-1.30786002e-01 9.12770107e-02 1.36790112e-01 -1.47710547e-01
-4.12298650e-01 5.32842539e-02 7.47163594e-01 8.09853524e-02
5.01617603e-02 -1.56342924e-01 -3.02794456e-01 9.22353029e-01
9.18374658e-01 -7.00807095e-01 -1.01137972e-02 -8.69244456e-01
2.21989214e-01 1.88407257e-01 1.32771522e-01 -6.44869149e-01
-1.27379475e-02 2.45736510e-01 3.29647183e-01 -3.04289281e-01
1.08164497e-01 -9.00453866e-01 -6.33644536e-02 4.43706125e-01
-5.31005323e-01 2.76173323e-01 1.73532635e-01 2.43087664e-01
-2.37274170e-01 -3.69887352e-01 6.45323157e-01 -4.00823265e-01
-7.91772306e-01 7.60636553e-02 -1.92672148e-01 5.88226393e-02
7.10302293e-01 4.87228006e-01 -3.47494960e-01 1.73635095e-01
-1.53888321e+00 2.41796285e-01 -6.54466227e-02 6.55471921e-01
1.30320340e-01 -9.92843091e-01 -6.51447833e-01 -2.88871042e-02
-7.39475861e-02 4.77679074e-02 -5.16751632e-02 1.37181640e-01
-5.21036744e-01 5.89275360e-01 -5.53880930e-01 -2.20021023e-03
-1.31214464e+00 5.29916048e-01 4.66162860e-01 -4.11075115e-01
-7.30522215e-01 1.04411030e+00 1.73915625e-02 -6.49272561e-01
7.40407854e-02 -2.91656911e-01 -8.68396759e-01 8.65060687e-02
9.57511663e-02 -9.83957201e-02 3.64008635e-01 -9.35112417e-01
-4.44984823e-01 3.22633892e-01 1.29763559e-01 2.27473360e-02
1.61829627e+00 5.74464537e-02 -4.35277462e-01 4.78920758e-01
1.27619112e+00 -4.16249961e-01 -7.39699364e-01 -3.50452781e-01
5.83542347e-01 4.21286255e-01 -1.95063297e-02 -1.03488624e+00
-1.11951864e+00 1.12785017e+00 3.23466927e-01 1.52277276e-01
6.62950039e-01 5.04684448e-01 7.67063797e-01 2.75647789e-01
1.16098225e-01 -1.08095336e+00 -5.62066317e-01 9.73848760e-01
4.18847859e-01 -1.01778615e+00 -5.21743357e-01 -6.50220513e-01
-5.68771720e-01 1.28977644e+00 6.17824554e-01 -5.62885463e-01
7.09802330e-01 3.96123230e-01 2.78980166e-01 -4.88154739e-01
-8.79493415e-01 -4.01847750e-01 6.26614690e-02 7.94105411e-01
6.94631338e-01 5.01969814e-01 -4.99613136e-01 1.10197306e+00
-4.86574382e-01 -5.31015635e-01 2.24860549e-01 5.80479801e-01
-3.90428841e-01 -1.41707659e+00 9.86241251e-02 1.35181069e-01
-7.83881128e-01 -4.43574160e-01 -6.11144900e-01 6.14241898e-01
3.36721390e-01 8.79109085e-01 2.35163748e-01 -2.26514816e-01
5.28830886e-01 7.61615753e-01 4.45253849e-01 -1.03195906e+00
-9.16141510e-01 -2.54355818e-01 6.91884637e-01 -5.83273053e-01
-3.35302353e-01 -3.72086674e-01 -1.84623396e+00 3.15470904e-01
3.62161696e-02 3.46618146e-01 1.11395288e+00 1.22435856e+00
2.09879354e-02 9.22179699e-01 2.26752877e-01 -4.29848820e-01
7.87877217e-02 -1.18009031e+00 -3.23717594e-01 5.15299320e-01
-1.50031880e-01 -2.98585713e-01 -6.65758252e-02 3.68968338e-01] | [10.379762649536133, 9.582059860229492] |
2fbca6ce-cd39-4398-8e0d-042b17302339 | self-promoted-prototype-refinement-for-few-1 | 2107.08918 | null | https://arxiv.org/abs/2107.08918v1 | https://arxiv.org/pdf/2107.08918v1.pdf | Self-Promoted Prototype Refinement for Few-Shot Class-Incremental Learning | Few-shot class-incremental learning is to recognize the new classes given few samples and not forget the old classes. It is a challenging task since representation optimization and prototype reorganization can only be achieved under little supervision. To address this problem, we propose a novel incremental prototype learning scheme. Our scheme consists of a random episode selection strategy that adapts the feature representation to various generated incremental episodes to enhance the corresponding extensibility, and a self-promoted prototype refinement mechanism which strengthens the expression ability of the new classes by explicitly considering the dependencies among different classes. Particularly, a dynamic relation projection module is proposed to calculate the relation matrix in a shared embedding space and leverage it as the factor for bootstrapping the update of prototypes. Extensive experiments on three benchmark datasets demonstrate the above-par incremental performance, outperforming state-of-the-art methods by a margin of 13%, 17% and 11%, respectively. | ['Zheng-Jun Zha', 'Jie Cheng', 'Wei Zhai', 'Yang Cao', 'Kai Zhu'] | 2021-07-19 | self-promoted-prototype-refinement-for-few | http://openaccess.thecvf.com//content/CVPR2021/html/Zhu_Self-Promoted_Prototype_Refinement_for_Few-Shot_Class-Incremental_Learning_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Zhu_Self-Promoted_Prototype_Refinement_for_Few-Shot_Class-Incremental_Learning_CVPR_2021_paper.pdf | cvpr-2021-1 | ['few-shot-class-incremental-learning'] | ['methodology'] | [ 2.22253099e-01 1.16848700e-01 -5.12740910e-01 -4.62605685e-01
-3.16829622e-01 -2.00462878e-01 5.76764047e-01 2.68556893e-01
-3.12640220e-01 7.57377684e-01 5.71398772e-02 1.11131713e-01
-2.65370667e-01 -8.83799672e-01 -3.95782232e-01 -8.39391649e-01
-5.57998642e-02 5.43671668e-01 5.38991511e-01 -1.28700301e-01
2.47307152e-01 3.06323171e-01 -1.77136874e+00 1.39555469e-01
9.71175790e-01 9.51859117e-01 1.02671288e-01 2.03962654e-01
-2.88946867e-01 5.91201425e-01 -9.13841277e-02 -2.66459644e-01
3.97717476e-01 -2.83188194e-01 -5.10366380e-01 3.41912985e-01
-1.12394579e-01 -2.78531551e-01 -3.53918582e-01 7.95325816e-01
4.82531428e-01 5.05984366e-01 3.55888546e-01 -1.04362750e+00
-6.20221496e-01 8.19988906e-01 -4.80673879e-01 2.55912155e-01
2.71342516e-01 1.27655461e-01 1.01791883e+00 -1.25106168e+00
7.14682221e-01 7.95487881e-01 5.21169782e-01 5.33686042e-01
-1.19471133e+00 -7.94726968e-01 5.72529435e-01 6.60323262e-01
-1.66298544e+00 -5.32616496e-01 1.04023111e+00 -3.28276068e-01
8.60350072e-01 -3.19193816e-03 7.92596221e-01 9.02742624e-01
-3.89981717e-01 6.99148953e-01 7.21888006e-01 -4.43203896e-01
5.41437685e-01 5.83132267e-01 6.42489731e-01 7.26597846e-01
3.55191767e-01 -1.35113336e-02 -5.39810181e-01 -1.95021883e-01
3.17308038e-01 4.25878853e-01 -3.07109952e-01 -8.14272821e-01
-9.31717098e-01 8.16281617e-01 3.89186889e-01 2.35535368e-01
-4.79484975e-01 -4.82458770e-01 4.66705978e-01 3.72813523e-01
2.68038273e-01 4.70250696e-01 -5.20263672e-01 -7.50500560e-02
-6.88690245e-01 -3.48397233e-02 5.78108430e-01 9.36926067e-01
1.19158435e+00 -1.84797570e-02 -2.28304610e-01 1.03788412e+00
-9.07548070e-02 6.56994209e-02 8.96874905e-01 -2.69433528e-01
3.11281919e-01 1.25205088e+00 -2.18651354e-01 -7.43968964e-01
-1.88097060e-01 -7.14801013e-01 -8.78943920e-01 -1.82266772e-01
-2.67848551e-01 -1.11006953e-01 -9.83090878e-01 1.65670907e+00
8.63960743e-01 6.16279423e-01 -4.88324538e-02 3.98905873e-01
6.23986423e-01 6.33608639e-01 -7.70178288e-02 -6.08911991e-01
9.05323803e-01 -1.03208613e+00 -3.75620276e-01 -2.20472127e-01
3.67767423e-01 -2.47912496e-01 9.80490506e-01 -5.97581267e-03
-5.87449193e-01 -8.31462383e-01 -1.27864873e+00 3.82951915e-01
-2.70174056e-01 5.56018986e-02 8.72694612e-01 3.64689708e-01
-4.42151487e-01 7.22777188e-01 -8.59781861e-01 -3.65765065e-01
6.69180810e-01 4.17769343e-01 -3.19864243e-01 -4.45490032e-02
-9.69373524e-01 4.72535640e-01 7.88166523e-01 -4.49136645e-02
-6.33586466e-01 -8.71479452e-01 -6.86989844e-01 1.76662281e-01
7.15464234e-01 -5.36107421e-01 8.83036971e-01 -8.12005401e-01
-1.60459554e+00 3.51111293e-01 -6.45110011e-02 -3.80474657e-01
2.65784502e-01 -8.13076720e-02 -5.12106836e-01 -4.04743440e-02
-2.37919539e-02 3.61001015e-01 9.02220666e-01 -8.43233347e-01
-8.59327078e-01 -3.90533864e-01 -9.54416767e-02 4.83998418e-01
-8.62997472e-01 -5.13238430e-01 -5.11926591e-01 -3.91325206e-01
3.05823237e-01 -9.80159760e-01 -4.03062314e-01 -1.08709127e-01
-3.26463252e-01 -2.61195511e-01 7.30981350e-01 -8.67848322e-02
1.58712101e+00 -2.19132161e+00 2.65734106e-01 2.33493179e-01
1.33533865e-01 4.50055003e-01 -7.34999478e-02 2.25200072e-01
-3.88647839e-02 -3.16026807e-01 -2.32283980e-01 -1.36549786e-01
-1.95288941e-01 2.00869069e-01 -4.24897313e-01 1.10308409e-01
1.02706887e-01 6.82433546e-01 -9.53262568e-01 -2.82903284e-01
1.00939117e-01 2.23276839e-01 -5.93134403e-01 3.84908170e-01
1.55344382e-01 1.26553312e-01 -4.52918261e-01 6.12781584e-01
4.06493813e-01 -3.75233859e-01 3.37316543e-01 -1.21664293e-01
2.37857252e-02 1.27292171e-01 -1.56460440e+00 1.54811525e+00
-2.12856665e-01 1.28974497e-01 -6.42921448e-01 -1.12276781e+00
1.16675365e+00 4.65524085e-02 3.40000540e-01 -4.81269509e-01
-1.32040456e-01 2.53712069e-02 1.12624601e-01 -3.98181349e-01
4.48187530e-01 -9.81980637e-02 1.25253886e-01 4.78678584e-01
8.13862681e-02 4.33049500e-01 2.25995928e-01 1.07031845e-01
9.48200166e-01 -1.80055097e-01 6.14561021e-01 6.31973799e-03
6.63604140e-01 -2.71542013e-01 1.03461874e+00 7.17466533e-01
-1.25627548e-01 3.06550205e-01 2.10009381e-01 -7.61096120e-01
-8.20464075e-01 -9.85415101e-01 -6.33413717e-02 1.18467987e+00
2.73999333e-01 -5.13662696e-01 -1.38912424e-01 -9.89531457e-01
2.48335395e-02 6.76180899e-01 -9.07168984e-01 -7.19650626e-01
-4.93831635e-01 -9.52701569e-01 -2.11725488e-01 6.25226140e-01
5.07437646e-01 -9.99811769e-01 -6.67825103e-01 3.46681237e-01
2.02565283e-01 -6.96790934e-01 -3.75575989e-01 3.32259685e-01
-1.01925385e+00 -1.17950189e+00 -4.11278516e-01 -9.18979287e-01
9.24662471e-01 2.80532539e-01 4.88711655e-01 8.52501765e-02
-2.52801895e-01 5.01346998e-02 -5.21113217e-01 -7.52574056e-02
-5.88458627e-02 4.79459435e-01 1.74037784e-01 3.40790123e-01
3.39727074e-01 -8.66222560e-01 -6.49233758e-01 1.53550431e-01
-5.71497798e-01 9.32980701e-02 6.84311509e-01 1.33493626e+00
7.66683757e-01 1.65536016e-01 8.43128860e-01 -1.26051903e+00
5.14315665e-01 -5.82354009e-01 -3.18262756e-01 4.66192365e-01
-1.15802169e+00 1.91919431e-01 6.64349914e-01 -1.03245342e+00
-1.15121377e+00 2.72271067e-01 2.72293955e-01 -4.91779625e-01
2.11145356e-01 5.08037567e-01 -1.01384543e-01 9.96225849e-02
5.86863041e-01 5.87831736e-01 -2.20079511e-01 -3.65467370e-01
5.98290384e-01 5.27363420e-01 3.51159126e-01 -3.72926205e-01
1.03773201e+00 3.49634498e-01 -3.72024745e-01 -6.72032952e-01
-7.68046498e-01 -5.44520438e-01 -8.14026475e-01 1.41104922e-01
2.16256499e-01 -7.89163172e-01 -2.19200969e-01 1.73202276e-01
-5.97681046e-01 -1.44378558e-01 -8.39407563e-01 3.66779715e-01
-1.09547637e-01 2.33993784e-01 -3.85855138e-01 -5.88950574e-01
-6.45285606e-01 -9.42311883e-01 4.92627919e-01 5.24148285e-01
7.84380138e-02 -6.08055472e-01 4.23615903e-01 -1.69718251e-01
3.88454556e-01 5.02650626e-03 1.03128815e+00 -9.73847747e-01
-4.80390549e-01 -3.51642579e-01 -3.08685619e-02 1.35262311e-01
3.70167315e-01 -8.44315290e-02 -7.69829273e-01 -5.94633102e-01
-2.15457603e-01 -2.31777728e-01 9.57124770e-01 -1.59709707e-01
9.75327671e-01 -2.95136392e-01 -5.65808892e-01 6.48481309e-01
1.32887995e+00 5.35717428e-01 3.66140127e-01 2.66787648e-01
6.38826013e-01 2.52446413e-01 6.47959650e-01 7.82479227e-01
3.96659851e-01 6.31792247e-01 -1.28963152e-02 4.38342541e-01
-2.39096388e-01 -3.47503215e-01 7.17307925e-02 1.08009589e+00
1.45922944e-01 3.09261560e-01 -6.91977143e-01 5.30880570e-01
-1.99777889e+00 -9.56335902e-01 6.10062957e-01 2.39279556e+00
1.01760590e+00 4.35962021e-01 -3.46467793e-02 1.89708799e-01
7.90407419e-01 1.48640990e-01 -1.01081622e+00 -6.65992871e-03
1.75950795e-01 2.68090248e-01 -1.08012632e-01 2.07609087e-01
-1.07605672e+00 1.01583791e+00 5.57103157e+00 6.36654139e-01
-1.23009336e+00 8.51204852e-04 5.26748478e-01 1.92782488e-02
-3.01731117e-02 2.80496836e-01 -1.03527617e+00 4.90165472e-01
5.87009907e-01 -5.85568190e-01 3.97490948e-01 1.18425167e+00
-4.73435849e-01 2.11710140e-01 -1.06017351e+00 9.38029766e-01
1.73113704e-01 -1.32257342e+00 1.43952429e-01 -2.45473400e-01
9.14055943e-01 -1.72267482e-01 -8.96541309e-03 8.92644525e-01
1.22222394e-01 -2.48696536e-01 2.78209984e-01 6.14561021e-01
6.31938636e-01 -8.71280849e-01 5.49245775e-01 3.13307643e-01
-1.55249119e+00 -5.15065253e-01 -6.17683291e-01 5.61041699e-04
-9.48228389e-02 5.53569973e-01 -1.05862558e+00 4.08192217e-01
5.04006088e-01 8.02464008e-01 -9.67704177e-01 1.22693264e+00
-3.39345753e-01 4.91635948e-01 -2.14647174e-01 -9.39645544e-02
-1.18569277e-01 5.46184042e-03 5.08573174e-01 8.54315698e-01
2.12460458e-01 2.16097295e-01 4.28425819e-01 4.27909523e-01
-1.51753321e-01 2.91562706e-01 -3.74801308e-01 1.64390549e-01
1.00355351e+00 1.31851363e+00 -6.52993619e-01 -6.57166481e-01
-3.10892075e-01 1.09403765e+00 7.30847895e-01 2.47926384e-01
-6.38887823e-01 -8.03183675e-01 4.53302801e-01 -1.63658131e-02
8.03925574e-01 1.70762971e-01 5.45542762e-02 -1.38442242e+00
9.80224162e-02 -6.03835225e-01 6.62469089e-01 -2.36329302e-01
-1.09001505e+00 6.84590161e-01 -3.66547070e-02 -1.31626928e+00
-3.99340123e-01 -1.52179133e-02 -7.34182715e-01 2.49932870e-01
-1.40280497e+00 -1.03089070e+00 -4.75188553e-01 4.69823271e-01
6.36000991e-01 -4.89181131e-01 8.57591569e-01 1.79410622e-01
-1.03152537e+00 9.91025686e-01 2.28894442e-01 -1.69719368e-01
5.66252649e-01 -9.35969949e-01 1.94975585e-01 8.04247141e-01
2.18877912e-01 8.34485412e-01 4.18854833e-01 -6.03812993e-01
-1.25464320e+00 -1.29184294e+00 6.94301784e-01 7.41572678e-02
6.76826060e-01 -3.25180382e-01 -1.18084097e+00 5.95213771e-01
-2.14126632e-01 3.29257548e-01 8.87306392e-01 3.31420511e-01
-6.17293060e-01 -5.99190235e-01 -8.84740353e-01 5.68031192e-01
1.08707869e+00 -3.86176556e-01 -8.04765463e-01 1.61997616e-01
7.81226933e-01 -4.97485660e-02 -5.94290197e-01 5.22711456e-01
6.83877051e-01 -7.67105103e-01 9.46357906e-01 -7.21800625e-01
-5.76117150e-02 -3.16604614e-01 -7.85673410e-02 -1.33992219e+00
-6.56101465e-01 -5.42290211e-01 -5.76801121e-01 1.42170453e+00
3.14258873e-01 -7.62304544e-01 8.68973672e-01 5.49711764e-01
1.76909100e-02 -1.12626469e+00 -8.39248300e-01 -8.45240057e-01
-3.79406601e-01 -7.42170662e-02 8.93428564e-01 1.04286039e+00
2.22010523e-01 6.00489140e-01 -3.87321264e-01 1.97893120e-02
5.12742102e-01 6.85907722e-01 8.10248852e-01 -1.29794657e+00
-4.75995928e-01 -2.57734865e-01 -5.94124854e-01 -8.21610987e-01
-8.57880618e-03 -9.20420468e-01 -2.79912371e-02 -1.15258563e+00
6.67183578e-01 -6.38126135e-01 -8.95998597e-01 6.77758276e-01
-5.72349548e-01 -1.44083381e-01 3.24654095e-02 4.14569318e-01
-1.02256274e+00 8.65425706e-01 8.34823668e-01 -1.32820010e-01
-7.72436380e-01 1.43597499e-01 -9.45169687e-01 5.82493305e-01
7.03651607e-01 -4.78087962e-01 -8.36080372e-01 -3.03538237e-02
-1.72998577e-01 -2.85797834e-01 -2.32626066e-01 -1.16879690e+00
6.35974526e-01 -2.06856057e-01 4.03431594e-01 -4.44107711e-01
1.16536081e-01 -6.37336791e-01 2.18757376e-01 6.14140809e-01
-4.47315931e-01 -2.40091860e-01 -1.77336261e-02 9.71125901e-01
7.20044300e-02 -1.65299043e-01 7.93070972e-01 2.41956133e-02
-1.00987148e+00 7.46165991e-01 1.59067482e-01 -6.50133118e-02
1.30917108e+00 -3.62283081e-01 -1.84590012e-01 1.53796777e-01
-7.13060617e-01 3.46040159e-01 3.53745073e-01 4.40599412e-01
7.75418162e-01 -1.51371670e+00 -4.02060330e-01 4.90595698e-01
5.64414144e-01 9.57870670e-03 2.86609083e-01 6.05844498e-01
1.15033336e-01 7.93692470e-02 -1.07158549e-01 -3.48758787e-01
-1.14473200e+00 7.40111828e-01 -4.03020643e-02 -4.21029657e-01
-7.77382493e-01 9.28163469e-01 6.20021671e-02 -4.00482506e-01
3.00367475e-01 2.66777277e-02 -4.23018068e-01 3.33962381e-01
8.00203085e-01 4.37148184e-01 -6.57331049e-02 -2.85614610e-01
-3.78718495e-01 4.89416897e-01 -7.15130031e-01 3.62298429e-01
1.64290190e+00 -1.91491127e-01 1.93456635e-01 6.96779191e-01
1.14791584e+00 -2.90489912e-01 -1.35284269e+00 -7.59623170e-01
8.14499334e-02 -4.51113909e-01 -1.84350073e-01 -6.34262145e-01
-1.00240827e+00 5.05953312e-01 7.08795309e-01 -2.43215803e-02
1.20791793e+00 -1.36533305e-01 6.26576900e-01 8.01783025e-01
4.21979040e-01 -1.10037601e+00 2.65018016e-01 3.24941814e-01
5.69648743e-01 -1.15461195e+00 3.27901632e-01 -2.85075694e-01
-6.48972332e-01 1.06362283e+00 8.01772654e-01 -3.34983282e-02
8.17219198e-01 -1.99662626e-01 -4.47335005e-01 -5.58890067e-02
-9.95633245e-01 -2.17687488e-01 2.50334561e-01 3.94190401e-01
-7.68982247e-02 5.53975925e-02 -2.47706503e-01 7.05041826e-01
1.20929912e-01 -8.87573585e-02 1.81270957e-01 1.02467477e+00
-6.46833420e-01 -1.21256280e+00 3.22923571e-01 7.45616078e-01
2.18763262e-01 -1.04647819e-02 -1.25724003e-01 6.77577019e-01
2.60525942e-01 5.11117995e-01 -8.39374866e-03 -7.49902129e-01
4.42094475e-01 2.52240688e-01 2.30756760e-01 -9.70832884e-01
-2.70021230e-01 -1.37626454e-01 -2.19350055e-01 -3.76749814e-01
-9.93704572e-02 -8.67456675e-01 -1.11801445e+00 1.39548391e-01
-7.25183964e-01 3.08727086e-01 2.04512849e-01 7.56712556e-01
6.87056005e-01 4.10784185e-01 1.11235034e+00 -4.93962288e-01
-7.87610233e-01 -8.47336411e-01 -5.13010800e-01 4.80455220e-01
1.07729062e-01 -8.98760021e-01 -4.02001143e-01 -1.66230157e-01] | [9.8102388381958, 3.4112679958343506] |
f2f2c58f-a9c9-44df-9510-abe2a892c678 | 3dpct-3d-point-cloud-transformer-with-dual | 2209.11255 | null | https://arxiv.org/abs/2209.11255v2 | https://arxiv.org/pdf/2209.11255v2.pdf | 3DGTN: 3D Dual-Attention GLocal Transformer Network for Point Cloud Classification and Segmentation | Although the application of Transformers in 3D point cloud processing has achieved significant progress and success, it is still challenging for existing 3D Transformer methods to efficiently and accurately learn both valuable global features and valuable local features for improved applications. This paper presents a novel point cloud representational learning network, called 3D Dual Self-attention Global Local (GLocal) Transformer Network (3DGTN), for improved feature learning in both classification and segmentation tasks, with the following key contributions. First, a GLocal Feature Learning (GFL) block with the dual self-attention mechanism (i.e., a novel Point-Patch Self-Attention, called PPSA, and a channel-wise self-attention) is designed to efficiently learn the GLocal context information. Second, the GFL block is integrated with a multi-scale Graph Convolution-based Local Feature Aggregation (LFA) block, leading to a Global-Local (GLocal) information extraction module that can efficiently capture critical information. Third, a series of GLocal modules are used to construct a new hierarchical encoder-decoder structure to enable the learning of "GLocal" information in different scales in a hierarchical manner. The proposed framework is evaluated on both classification and segmentation datasets, demonstrating that the proposed method is capable of outperforming many state-of-the-art methods on both classification and segmentation tasks. | ['Jonathan Li', 'Linlin Xu', 'Qian Xie', 'Kyle Gao', 'Dening Lu'] | 2022-09-21 | null | null | null | null | ['point-cloud-classification'] | ['computer-vision'] | [-1.01373680e-02 -2.79449791e-01 1.04239091e-01 -4.44103956e-01
-9.81716633e-01 -1.01778872e-01 4.73955572e-01 2.58528113e-01
1.80758238e-01 1.12983912e-01 -4.08381931e-02 -1.03670105e-01
-1.86945543e-01 -1.07697988e+00 -8.58513832e-01 -8.02553594e-01
-1.87106058e-01 2.97224343e-01 3.65806192e-01 -1.37088716e-01
2.92198658e-01 1.05093002e+00 -1.83827364e+00 3.41133863e-01
1.00552034e+00 1.53730571e+00 5.91136336e-01 5.00926137e-01
-4.62085575e-01 7.08563626e-01 -3.67220610e-01 -2.56692022e-02
2.47091100e-01 -1.20281026e-01 -7.37110972e-01 1.83601469e-01
5.68360090e-01 -1.85781002e-01 -1.60401359e-01 7.06279218e-01
5.64779162e-01 1.32449806e-01 5.74488640e-01 -9.77107406e-01
-7.47042000e-01 -4.44846228e-02 -5.53826809e-01 2.47870728e-01
1.32448077e-01 9.85515490e-02 1.13370132e+00 -1.11816347e+00
3.60124260e-01 1.31540060e+00 6.70904934e-01 -1.93501171e-02
-7.15535343e-01 -6.06756926e-01 1.03117779e-01 2.71678478e-01
-1.49643683e+00 2.65980870e-01 1.13456285e+00 -4.52670276e-01
1.44738889e+00 -2.30989382e-02 1.05056071e+00 3.94683301e-01
6.80048764e-01 8.83096337e-01 9.74039912e-01 -1.40552193e-01
1.14004105e-01 -1.58236593e-01 2.37482674e-02 9.05762136e-01
-1.76785752e-01 1.02449566e-01 -6.54911399e-01 1.55381933e-01
1.14817262e+00 2.36300260e-01 1.22908413e-01 -4.80660141e-01
-7.28015959e-01 9.16990280e-01 1.09730315e+00 4.42704678e-01
-6.77320600e-01 2.52750397e-01 3.22331518e-01 2.29357988e-01
9.15643871e-01 1.36810457e-02 -4.30389524e-01 -4.03991714e-03
-9.13276732e-01 2.11136416e-01 2.86304891e-01 1.14630604e+00
1.24545395e+00 1.30527347e-01 -2.54276901e-01 8.32732320e-01
3.88369441e-01 6.47795260e-01 3.37439537e-01 -3.11853915e-01
5.30413985e-01 1.16271818e+00 -4.37770575e-01 -1.09385562e+00
-4.53315079e-01 -8.58909070e-01 -7.65370131e-01 3.88551950e-01
-3.50133538e-01 4.05447841e-01 -1.08756113e+00 1.28450918e+00
4.91800070e-01 4.22349244e-01 -2.59503275e-01 9.42219853e-01
1.09849966e+00 7.89669633e-01 5.60258143e-02 3.00253600e-01
1.34918416e+00 -8.49021792e-01 -3.76258731e-01 -1.40563756e-01
5.14762759e-01 -7.20513165e-01 9.99769449e-01 9.94698033e-02
-1.10148680e+00 -9.04223323e-01 -1.20066941e+00 -6.34684801e-01
-6.81386173e-01 -6.36603907e-02 7.33899832e-01 2.03970790e-01
-1.07438123e+00 5.16934931e-01 -9.06524420e-01 -3.70420843e-01
8.18867505e-01 6.07799530e-01 -2.64669091e-01 -2.18301028e-01
-7.78387964e-01 6.46838427e-01 1.91706270e-01 1.87222004e-01
-9.23474550e-01 -9.04088795e-01 -1.13511038e+00 2.80849248e-01
-5.55335954e-02 -7.92408764e-01 8.72150898e-01 -6.61083698e-01
-1.42945194e+00 8.30556691e-01 -1.50532901e-01 -3.35306913e-01
-1.98119178e-01 -2.75491357e-01 -3.97196552e-03 3.13255429e-01
1.91664457e-01 7.44877160e-01 1.00061429e+00 -1.16346121e+00
-8.10911000e-01 -6.98737562e-01 -1.35009125e-01 5.56868374e-01
-1.83570459e-01 -1.43751115e-01 -7.05243289e-01 -7.01856434e-01
3.08620185e-01 -3.79643947e-01 -7.52732530e-02 -1.21607862e-01
-3.68225947e-02 -4.87524390e-01 1.21552873e+00 -5.83858311e-01
1.03770018e+00 -2.07898355e+00 2.38151714e-01 3.90703738e-01
4.08863246e-01 2.62028009e-01 -1.84667498e-01 3.94992471e-01
-1.31516546e-01 -1.05195366e-01 -2.33931109e-01 -5.79407156e-01
-2.00902581e-01 3.74468327e-01 -5.65277860e-02 3.59935194e-01
6.98157489e-01 1.23880148e+00 -7.73248196e-01 -5.19403636e-01
8.76315475e-01 7.44176984e-01 -5.44464827e-01 3.20333511e-01
-4.00512889e-02 4.62744445e-01 -8.20622623e-01 1.09753525e+00
8.32171082e-01 -3.17865014e-01 -6.27863109e-01 -3.31581652e-01
-3.95897508e-01 1.75878823e-01 -8.25956225e-01 1.80658340e+00
-8.34595919e-01 2.35496908e-01 -4.77733873e-02 -9.24572587e-01
1.25710595e+00 1.28513545e-01 6.67850435e-01 -8.34827244e-01
4.18719023e-01 2.29997963e-01 -5.28889954e-01 -3.34969074e-01
3.25020790e-01 -1.60523266e-01 -3.95087488e-02 -1.06561265e-03
4.92722511e-01 -4.36122894e-01 -3.69265020e-01 -6.36755535e-03
1.12236023e+00 7.36353397e-02 1.58421785e-01 -2.86202252e-01
8.61376762e-01 -3.59431840e-02 4.50129181e-01 4.86260027e-01
-1.98646355e-02 6.52964473e-01 1.54289678e-01 -5.28683305e-01
-9.15119767e-01 -9.05895472e-01 -6.73853531e-02 7.86850572e-01
3.65459323e-01 -4.71393764e-01 -6.23133957e-01 -6.40592635e-01
2.79476881e-01 2.40901500e-01 -5.45873344e-01 -1.43614754e-01
-4.25290197e-01 -2.97065645e-01 1.44944474e-01 8.90616834e-01
8.81860495e-01 -1.10989928e+00 -4.89711165e-01 2.08488688e-01
2.55197763e-01 -1.07572126e+00 -3.46959889e-01 4.48272139e-01
-1.10829580e+00 -7.97607183e-01 -4.00144756e-01 -8.06331158e-01
4.66318160e-01 5.48888743e-01 1.00051963e+00 8.97267237e-02
-2.86285430e-01 4.19592083e-01 -5.91297448e-01 -4.14428115e-01
2.88966238e-01 3.09485346e-01 -5.13253391e-01 1.41723350e-01
4.85165477e-01 -7.74206996e-01 -4.51098561e-01 7.29022995e-02
-7.42480040e-01 7.09928423e-02 9.24528241e-01 1.00475788e+00
1.12691534e+00 -4.24438529e-02 3.43187869e-01 -5.73002458e-01
4.13149208e-01 -4.71842766e-01 -6.82620585e-01 3.26027311e-02
-4.69751775e-01 -9.14925858e-02 6.93586588e-01 2.82875806e-01
-9.18586373e-01 7.50321522e-02 -6.11682177e-01 -8.82191300e-01
-1.97868645e-01 4.18908507e-01 -3.77873838e-01 -7.07934380e-01
1.94379583e-01 3.28884512e-01 -8.68707746e-02 -7.72442043e-01
3.60671431e-01 6.48917735e-01 4.16384548e-01 -5.85534871e-01
9.35653031e-01 3.95911157e-01 1.67364284e-01 -9.71508086e-01
-7.62995958e-01 -7.81142175e-01 -9.67646837e-01 -1.84210420e-01
1.11870503e+00 -1.07474208e+00 -6.23846412e-01 7.45164990e-01
-1.05082309e+00 -1.67519093e-01 -5.50290942e-01 2.62878954e-01
-6.53915644e-01 2.36182421e-01 -5.68338811e-01 -5.99346340e-01
-6.61811948e-01 -1.31470740e+00 1.71784890e+00 3.55107784e-01
5.28274179e-01 -1.08746445e+00 -2.10203066e-01 3.44496191e-01
4.17537272e-01 4.59313869e-01 9.60967779e-01 -3.24451149e-01
-1.12087882e+00 -2.27245376e-01 -5.52965820e-01 5.50333977e-01
7.58773163e-02 -3.28517348e-01 -1.00078154e+00 -1.96198553e-01
2.06585694e-02 -2.25072235e-01 6.43507004e-01 3.73297304e-01
1.26350796e+00 1.63337022e-01 -3.39126468e-01 1.01109767e+00
1.70941603e+00 1.43110484e-01 5.65145314e-01 8.45702738e-02
1.09631348e+00 2.82981023e-02 5.86093366e-01 4.02382076e-01
6.52795732e-01 6.03192329e-01 6.68315768e-01 -4.65448618e-01
-3.48529249e-01 -3.02011400e-01 2.74865180e-02 1.13661134e+00
-1.51817426e-01 -5.91605715e-02 -9.05293107e-01 6.30676270e-01
-1.79520714e+00 -5.66683054e-01 -1.07067868e-01 1.92428327e+00
2.41749644e-01 1.04680676e-02 -2.74389088e-01 1.07177883e-01
4.41728890e-01 2.93960661e-01 -5.27304173e-01 -6.45138741e-01
-4.39055860e-02 9.63839650e-01 4.21640247e-01 4.53854740e-01
-1.27575660e+00 1.24238324e+00 5.30224466e+00 1.17318702e+00
-1.25180590e+00 1.95586890e-01 3.53402138e-01 3.36729050e-01
-2.35775158e-01 -1.71076298e-01 -7.00216234e-01 2.12770015e-01
5.88191688e-01 2.14593425e-01 1.38573170e-01 1.04198706e+00
1.19436622e-01 9.46270078e-02 -6.97043240e-01 1.19637942e+00
9.46642011e-02 -1.38186109e+00 2.93901354e-01 1.10307612e-01
5.60125709e-01 4.05285925e-01 -9.77394283e-02 2.33659327e-01
-2.93098539e-02 -9.05400157e-01 6.54575288e-01 5.63058972e-01
7.57349133e-01 -1.01058722e+00 9.62344050e-01 1.09048605e-01
-1.87071121e+00 -1.32761687e-01 -3.98338497e-01 3.08369938e-02
6.06094226e-02 7.38332570e-01 -4.84715462e-01 1.21787500e+00
9.25432682e-01 1.32948256e+00 -7.11284995e-01 1.07171166e+00
-1.92975476e-01 3.71674508e-01 -4.34165776e-01 9.83582959e-02
6.97937369e-01 -1.36372939e-01 5.24335861e-01 1.13892472e+00
3.47981393e-01 9.90275890e-02 1.59290612e-01 9.22164083e-01
3.71335745e-02 2.69343376e-01 -5.47831357e-01 2.23230600e-01
2.89529145e-01 1.38743818e+00 -5.88885307e-01 -2.02834681e-01
-7.09910691e-01 8.27428937e-01 5.40968657e-01 3.97849195e-02
-5.67946255e-01 -5.67947924e-01 8.09638560e-01 1.16474321e-02
7.25992382e-01 -4.01354700e-01 -5.34414172e-01 -9.71539915e-01
1.00394699e-03 -3.37739199e-01 3.72289360e-01 -9.15179789e-01
-1.46083689e+00 7.00258553e-01 -1.47285536e-01 -1.39644623e+00
1.51233017e-01 -6.69286430e-01 -7.87750065e-01 1.12185133e+00
-1.94686580e+00 -1.79703820e+00 -6.96479678e-01 9.55150008e-01
6.06511235e-01 -5.70494458e-02 4.73894417e-01 4.57782924e-01
-3.80073845e-01 5.05770504e-01 -1.27266482e-01 -1.29537329e-01
1.78795591e-01 -1.33895183e+00 3.37946922e-01 7.28204608e-01
3.20424251e-02 3.44066679e-01 -5.54819256e-02 -6.70533538e-01
-1.59269607e+00 -1.34692407e+00 7.23994732e-01 -9.98403952e-02
3.23458910e-01 -5.66468358e-01 -9.60833132e-01 4.87205565e-01
-1.63746774e-02 2.74103075e-01 2.86367118e-01 -9.13688987e-02
-5.08956127e-02 -3.42872620e-01 -1.07721388e+00 3.61326151e-02
1.08771288e+00 -6.90358162e-01 -5.54965258e-01 2.22510576e-01
8.68251085e-01 -6.04843020e-01 -1.14255810e+00 5.80668509e-01
1.36314601e-01 -9.00069654e-01 1.00254011e+00 -6.36317357e-02
3.18055630e-01 -4.79564339e-01 -2.85338372e-01 -1.01412582e+00
-5.26849151e-01 -2.16370761e-01 -2.06450552e-01 1.26553369e+00
-6.62215799e-02 -5.11301577e-01 7.74580359e-01 4.31199446e-02
-8.15049112e-01 -1.28410661e+00 -1.22591197e+00 -5.96730888e-01
-2.60076616e-02 -6.78837836e-01 8.96786511e-01 6.28295422e-01
-3.94171178e-01 4.28540111e-01 -8.94412473e-02 3.65605742e-01
5.67356884e-01 3.90825093e-01 6.36583865e-01 -1.21147084e+00
1.67671621e-01 -3.41220021e-01 -9.83531356e-01 -1.40164959e+00
-8.08247253e-02 -1.26294255e+00 -8.04032460e-02 -1.69501960e+00
-8.92901197e-02 -5.83851159e-01 -3.82097930e-01 4.65581089e-01
-4.98726889e-02 1.93624884e-01 1.00895420e-01 -2.75109289e-03
-4.95502740e-01 1.09035873e+00 1.46528399e+00 -2.91302241e-02
-2.37304524e-01 -6.00992851e-02 -4.57770824e-01 4.03502822e-01
3.69029224e-01 -2.08570004e-01 -3.02941203e-01 -4.84566957e-01
-2.51425713e-01 -3.23082693e-02 6.71700656e-01 -1.32269418e+00
2.74676561e-01 1.89600423e-01 4.64496285e-01 -1.29788554e+00
4.54791784e-01 -9.17336166e-01 -2.75006801e-01 9.14157107e-02
2.50711203e-01 1.32623494e-01 3.33112806e-01 5.03723502e-01
-5.80301464e-01 1.61679804e-01 6.89395547e-01 -1.02643058e-01
-9.90739703e-01 8.17498565e-01 -2.77104247e-02 -2.39208892e-01
9.06928480e-01 -3.91801983e-01 3.19947004e-02 5.76136261e-02
-4.74023581e-01 4.50491756e-01 2.50566244e-01 4.77053255e-01
1.02658319e+00 -1.54718947e+00 -5.66352665e-01 7.10586011e-01
2.59032190e-01 4.68603224e-01 5.81273377e-01 7.07338035e-01
-7.80671179e-01 4.72190291e-01 -2.94773489e-01 -1.00620198e+00
-8.67362559e-01 4.78972673e-01 3.56356561e-01 -4.38769579e-01
-9.61880803e-01 9.80204940e-01 3.38504791e-01 -4.26497102e-01
9.28640179e-03 -6.56607985e-01 -1.23843804e-01 -1.66302308e-01
1.72869146e-01 1.97399631e-01 4.75109279e-01 -8.30810308e-01
-3.74491423e-01 1.27259600e+00 -2.33521219e-02 3.95371437e-01
1.55036759e+00 -9.79548395e-02 -1.39513046e-01 1.95942476e-01
1.40366793e+00 -3.69565427e-01 -1.20482826e+00 -4.13184136e-01
-2.32249543e-01 -5.74563384e-01 4.59589183e-01 -5.37457883e-01
-1.30985880e+00 1.37354290e+00 7.84054160e-01 1.24484682e-02
1.46671712e+00 2.41906047e-01 1.09264553e+00 1.32343182e-02
4.15798783e-01 -8.53568256e-01 2.02560294e-02 7.49735653e-01
9.85934377e-01 -1.07338464e+00 -1.02284014e-01 -6.08739436e-01
-4.48173970e-01 1.07519424e+00 6.33883536e-01 -5.37694395e-01
9.48457360e-01 1.72841892e-01 -2.28348419e-01 -7.19932795e-01
-5.48369586e-01 -6.35747254e-01 6.39186919e-01 5.48951805e-01
1.78867757e-01 -1.40593410e-01 6.70002773e-02 6.77509665e-01
-8.47829208e-02 -2.88928431e-02 -1.76938429e-01 8.79216373e-01
-4.47983503e-01 -9.51305449e-01 -2.04516485e-01 5.50604463e-01
-4.57840301e-02 -1.02168888e-01 -1.43139347e-01 8.62420321e-01
5.72376311e-01 6.75829530e-01 3.12022477e-01 -7.54610121e-01
4.93221492e-01 -3.45400274e-02 3.55942219e-01 -7.45253205e-01
-1.06928062e+00 1.99008569e-01 -4.96791005e-01 -8.81424546e-01
-4.29175615e-01 -4.96602982e-01 -1.27558243e+00 -1.50582880e-01
-3.79164964e-01 4.50731814e-03 5.87050498e-01 1.03780925e+00
6.30872726e-01 9.14024413e-01 8.26549351e-01 -1.21126199e+00
8.66903365e-02 -7.93720424e-01 -6.72359526e-01 2.80431509e-01
3.37336272e-01 -9.88546371e-01 -1.25085115e-01 -3.00578386e-01] | [7.946136951446533, -3.463895082473755] |
82b2296f-49aa-433a-9c4a-e8e22ae6c7ec | simple-and-effective-few-shot-named-entity | 2010.02405 | null | https://arxiv.org/abs/2010.02405v1 | https://arxiv.org/pdf/2010.02405v1.pdf | Simple and Effective Few-Shot Named Entity Recognition with Structured Nearest Neighbor Learning | We present a simple few-shot named entity recognition (NER) system based on nearest neighbor learning and structured inference. Our system uses a supervised NER model trained on the source domain, as a feature extractor. Across several test domains, we show that a nearest neighbor classifier in this feature-space is far more effective than the standard meta-learning approaches. We further propose a cheap but effective method to capture the label dependencies between entity tags without expensive CRF training. We show that our method of combining structured decoding with nearest neighbor learning achieves state-of-the-art performance on standard few-shot NER evaluation tasks, improving F1 scores by $6\%$ to $16\%$ absolute points over prior meta-learning based systems. | ['Arzoo Katiyar', 'Yi Yang'] | 2020-10-06 | null | https://aclanthology.org/2020.emnlp-main.516 | https://aclanthology.org/2020.emnlp-main.516.pdf | emnlp-2020-11 | ['few-shot-ner'] | ['natural-language-processing'] | [ 5.21685667e-02 2.32723564e-01 -3.99962455e-01 -7.66678512e-01
-1.32244575e+00 -3.78805667e-01 6.08927429e-01 4.01389420e-01
-8.79191101e-01 7.68053174e-01 4.41637456e-01 1.18509561e-01
6.28437102e-02 -9.94967818e-01 -4.57981229e-01 -1.77146211e-01
-1.05083026e-01 5.99011481e-01 2.94143766e-01 -2.40720659e-01
3.28270197e-01 8.30453560e-02 -1.48892558e+00 3.14296842e-01
7.08111405e-01 7.81729400e-01 -2.97075324e-02 5.28938830e-01
-7.03687370e-01 9.20246661e-01 -4.81098294e-01 -8.00149500e-01
-1.24468409e-01 -3.60539556e-01 -1.02967703e+00 -5.80952823e-01
3.31541449e-01 -1.13101818e-01 -3.42932254e-01 8.67447615e-01
6.39923692e-01 5.78101993e-01 8.47500026e-01 -6.87880218e-01
-8.19976687e-01 7.76443183e-01 1.23597696e-01 1.12611920e-01
3.03028136e-01 -3.91795516e-01 1.06559074e+00 -9.52711701e-01
7.23766863e-01 5.90373993e-01 1.06963003e+00 8.18923593e-01
-1.18512976e+00 -5.52697718e-01 -3.24982107e-01 5.02953418e-02
-1.48756540e+00 -7.73827434e-01 1.77824885e-01 -1.86589703e-01
1.66356969e+00 -2.52837956e-01 7.93431625e-02 9.45133150e-01
1.41858354e-01 4.46104139e-01 8.95435750e-01 -9.07888055e-01
6.41554654e-01 -7.46839098e-04 6.73831582e-01 8.33291650e-01
9.85240331e-04 3.34479026e-02 -4.56355214e-01 -5.39148986e-01
1.81064755e-01 2.45567113e-01 2.51178652e-01 -1.23291194e-01
-8.85727525e-01 9.20978904e-01 1.75677806e-01 6.61645234e-01
-1.85675621e-01 1.06244221e-01 2.69503564e-01 6.16913997e-02
7.21224070e-01 6.52085185e-01 -8.44118416e-01 -4.23537046e-01
-1.00374997e+00 -3.10717851e-01 1.30079019e+00 1.14026105e+00
9.26570177e-01 -2.70545691e-01 -3.73711199e-01 1.08250988e+00
1.69801340e-01 3.28986049e-01 5.71808279e-01 -1.14123440e+00
2.41243020e-01 3.75045925e-01 2.52716482e-01 -2.46880636e-01
-2.31656253e-01 -6.64019436e-02 -4.39348429e-01 -9.83151719e-02
1.32007420e-01 -3.92829895e-01 -1.13225305e+00 1.43637764e+00
1.86997309e-01 4.34786856e-01 3.97713393e-01 8.07683691e-02
6.98374867e-01 7.11481392e-01 6.91374302e-01 -1.76438257e-01
1.32694447e+00 -8.62259030e-01 -5.56385279e-01 -2.52272040e-01
1.24925995e+00 -5.34970045e-01 6.61832392e-01 -1.57282114e-01
-6.34249568e-01 -3.03882509e-01 -8.34075391e-01 -2.13959649e-01
-7.12040544e-01 -1.62208810e-01 7.01539397e-01 9.06890750e-01
-7.12320805e-01 1.15584743e+00 -8.87028396e-01 -5.51038444e-01
3.34052950e-01 2.96741515e-01 -5.65466464e-01 -3.38167459e-01
-1.22986424e+00 1.06300318e+00 5.85069537e-01 -6.41858459e-01
-5.78476429e-01 -7.05427766e-01 -1.14514852e+00 2.17327908e-01
1.59384191e-01 -3.61001968e-01 1.49065137e+00 -2.39411071e-02
-1.46489251e+00 9.10742760e-01 -5.51982343e-01 -4.01114047e-01
-3.71963859e-01 -3.54300559e-01 -7.91667402e-01 -1.85443506e-01
2.75208861e-01 4.73982364e-01 2.51019746e-01 -7.43200660e-01
-7.18182981e-01 -3.49068105e-01 -3.05903137e-01 -3.44770309e-03
-4.10820812e-01 2.16427222e-01 -5.27171604e-02 -3.39503497e-01
-1.85083523e-01 -6.09661102e-01 -3.74462306e-01 -4.49441731e-01
-7.88850039e-02 -4.77534771e-01 3.10995549e-01 -4.79366481e-01
1.19056070e+00 -2.00446773e+00 -4.25375074e-01 -1.55605555e-01
7.18900487e-02 4.19380397e-01 -2.40056038e-01 4.65757549e-01
1.80222914e-01 2.47881904e-01 -2.30113044e-01 -3.80658001e-01
2.86478996e-01 2.97322065e-01 -1.61286935e-01 1.99446648e-01
-1.32767064e-02 9.64939535e-01 -1.04001689e+00 -5.75485349e-01
1.29423276e-01 6.38762474e-01 -2.85208702e-01 3.75060737e-01
1.25385933e-02 -1.83354825e-01 -2.27009252e-01 4.54965085e-01
1.78810418e-01 -1.92223907e-01 1.55271083e-01 8.18316638e-02
-1.35390863e-01 6.30532622e-01 -9.48587060e-01 2.10748363e+00
-6.25889480e-01 3.41410041e-01 -5.90540767e-01 -8.60245049e-01
1.10227275e+00 4.95488465e-01 2.03448087e-01 -7.27157652e-01
8.42095464e-02 2.23839819e-01 -6.13283217e-01 -2.39532873e-01
4.06091422e-01 -5.14689744e-01 -6.11394227e-01 3.49699378e-01
8.64248931e-01 2.62208134e-01 1.56430066e-01 -4.94878478e-02
1.62491715e+00 1.92278042e-01 7.68425465e-01 -3.16524692e-02
-1.61285102e-01 2.65049905e-01 9.09018219e-01 1.08934236e+00
-2.64950722e-01 3.66227597e-01 -7.12007508e-02 -5.09288788e-01
-1.04579127e+00 -1.07017064e+00 -3.11887741e-01 1.72900951e+00
-9.20448080e-02 -6.71188116e-01 -8.21806669e-01 -1.05713701e+00
-4.35628325e-01 1.33245325e+00 -5.49356401e-01 -6.29801750e-02
-3.63342375e-01 -5.24869442e-01 9.81559515e-01 7.60838211e-01
4.95608032e-01 -1.09563148e+00 -2.36480162e-01 4.30943459e-01
-8.77998695e-02 -1.06804872e+00 -3.38481218e-01 8.15636158e-01
-1.04377449e+00 -5.54860055e-01 -7.17485726e-01 -1.12115300e+00
3.09518576e-01 -2.90473588e-02 1.21553779e+00 -3.89487207e-01
-2.61045277e-01 2.52554804e-01 -6.65864050e-01 -1.61323264e-01
-2.83213258e-01 3.68173063e-01 -4.09625173e-02 -6.14894211e-01
1.28273022e+00 -4.93899226e-01 -1.56369239e-01 4.11820523e-02
-4.35615689e-01 -5.26081800e-01 7.18590438e-01 9.53976750e-01
6.49810493e-01 -1.56303734e-01 3.67893517e-01 -1.22895098e+00
1.56660974e-01 -6.55127823e-01 -2.57196724e-01 5.60764432e-01
-8.12132180e-01 5.33304572e-01 5.12157798e-01 -1.92492306e-01
-1.31322277e+00 3.88742298e-01 -3.44467908e-01 -4.31795567e-02
-8.60730529e-01 2.66080230e-01 -1.79946020e-01 1.82461753e-01
1.04390633e+00 1.31331190e-01 -8.12907279e-01 -7.63070643e-01
7.29656816e-01 1.11245322e+00 5.27535737e-01 -3.52428555e-01
4.69611704e-01 2.21314341e-01 -4.82647568e-01 -8.02978992e-01
-1.37233400e+00 -8.43658507e-01 -9.59021986e-01 4.40460652e-01
1.12954021e+00 -9.73769844e-01 -1.61208317e-01 9.19094309e-02
-1.24045420e+00 -3.10666531e-01 -5.77469766e-01 6.58331931e-01
-5.59507787e-01 1.00036383e-01 -9.01755452e-01 -7.10835695e-01
-5.00846267e-01 -3.17899317e-01 1.11242533e+00 3.89276832e-01
-2.66722172e-01 -1.08528531e+00 7.63677239e-01 2.30137482e-01
3.37717623e-01 -2.45667204e-01 7.87108779e-01 -1.53968966e+00
1.37245983e-01 -3.98971617e-01 -9.71785635e-02 9.50888842e-02
4.64020669e-03 -7.38609970e-01 -1.13988733e+00 6.88015819e-02
-2.07508996e-01 -2.45669484e-01 9.36418772e-01 -7.03346580e-02
3.24099809e-01 -2.43481919e-01 -7.00589240e-01 4.24685031e-01
1.65342808e+00 2.83919722e-01 7.84975529e-01 1.58624828e-01
5.96281826e-01 3.21675986e-01 4.37213808e-01 2.90979952e-01
4.94539082e-01 4.20189202e-01 -3.80119324e-01 2.69426942e-01
-9.06830952e-02 -5.68776131e-01 3.49151552e-01 7.58659482e-01
1.28061146e-01 -1.53591231e-01 -1.14103508e+00 8.84183347e-01
-1.69348490e+00 -1.25698102e+00 2.96282023e-01 2.15199089e+00
9.02635515e-01 -5.10015935e-02 -1.79096162e-01 -3.94335806e-01
1.13733077e+00 -4.70848158e-02 -2.05742061e-01 -4.06860769e-01
1.51855484e-01 7.51576483e-01 6.45367265e-01 3.10069621e-01
-1.34683287e+00 1.29176104e+00 6.99553108e+00 8.99926782e-01
-4.68462974e-01 6.41043663e-01 2.19340801e-01 2.13491321e-01
2.82670255e-03 3.71874213e-01 -1.33034790e+00 3.60594124e-01
1.78931332e+00 -2.02146489e-02 2.76531428e-01 1.23669088e+00
-5.06728590e-01 6.34725168e-02 -1.12437451e+00 8.07911277e-01
3.01844001e-01 -1.44445395e+00 -4.14544553e-01 6.10659011e-02
8.37550938e-01 6.93614841e-01 -6.40485227e-01 9.02876914e-01
1.06524324e+00 -9.18198347e-01 1.14860334e-01 6.20704353e-01
8.90193403e-01 -8.06115210e-01 8.76526892e-01 4.86774534e-01
-1.19743121e+00 -9.64946598e-02 -6.71881020e-01 -1.12542413e-01
3.19368720e-01 7.08399594e-01 -8.33000124e-01 2.68508255e-01
7.14174449e-01 5.31314969e-01 -1.94023207e-01 1.14015031e+00
-2.75145620e-01 6.79578125e-01 -2.88914651e-01 -2.45201007e-01
-8.31082687e-02 4.06906217e-01 2.12183520e-01 1.62977576e+00
4.43583369e-01 5.73083520e-01 1.51336953e-01 4.93741423e-01
-4.66111034e-01 2.47234643e-01 -9.56739187e-01 -2.71568507e-01
7.47180760e-01 1.11229193e+00 -6.78406656e-01 -5.62439084e-01
-5.45590460e-01 1.43058372e+00 7.05541134e-01 -4.30390276e-02
-4.67455775e-01 -1.36835730e+00 4.84722614e-01 -2.00776562e-01
8.98273587e-01 -2.66473055e-01 -1.46582779e-02 -1.48574591e+00
-4.20345724e-01 -1.10199302e-01 6.05226755e-01 -5.28017879e-01
-1.73930466e+00 6.55911803e-01 -2.55270004e-01 -9.70910668e-01
-5.50032556e-01 -4.54442918e-01 -4.92877543e-01 5.24766922e-01
-1.32935321e+00 -9.76345360e-01 1.32106077e-02 4.15755361e-01
2.49324873e-01 -1.17629319e-01 1.62857234e+00 2.83039868e-01
-1.93399295e-01 6.41739309e-01 5.01336634e-01 7.39692450e-01
8.19134533e-01 -1.28623748e+00 6.36838794e-01 5.65964222e-01
7.20770955e-01 6.62583649e-01 2.51119792e-01 -7.53259838e-01
-9.61993873e-01 -1.19577348e+00 1.83966434e+00 -9.11470115e-01
4.95192677e-01 -4.14045066e-01 -7.81376183e-01 8.52895081e-01
2.04227552e-01 3.69670689e-01 1.37680233e+00 5.19663274e-01
-8.61586034e-01 1.77924708e-01 -1.43401134e+00 1.49797693e-01
1.28512204e+00 -8.12718868e-01 -1.38176060e+00 1.41569167e-01
9.00476933e-01 3.39126661e-02 -1.36823809e+00 2.36493528e-01
5.16480744e-01 -6.14114940e-01 6.40559912e-01 -8.45900059e-01
1.56556517e-01 -1.86043456e-02 -6.69136643e-01 -1.16215456e+00
-5.63208401e-01 -2.51630783e-01 -1.81030825e-01 1.65738082e+00
5.58927298e-01 -2.63338000e-01 7.15091705e-01 9.33065653e-01
-2.57499656e-03 -1.88982129e-01 -1.07030857e+00 -1.04994297e+00
2.72572666e-01 -5.81466198e-01 2.99519658e-01 1.10530853e+00
5.92817485e-01 7.59848416e-01 -2.59240121e-01 1.11851506e-01
8.10543537e-01 -1.62138447e-01 1.14545643e-01 -1.51870990e+00
-2.30025947e-01 2.22671285e-01 -7.22953260e-01 -7.87035942e-01
5.73351443e-01 -9.43869174e-01 4.43882227e-01 -1.73795509e+00
4.43364382e-01 -4.31469738e-01 -5.66909909e-01 9.92748618e-01
1.12799123e-01 4.82564867e-01 1.61714908e-02 2.01171875e-01
-1.24644673e+00 3.49018067e-01 6.83746338e-02 -1.25083998e-02
-1.72425181e-01 -3.16691458e-01 -6.48211241e-01 7.09118783e-01
5.56292176e-01 -1.17497373e+00 1.11583605e-01 -5.20076573e-01
-5.15918992e-02 7.59279579e-02 -1.45218194e-01 -1.16668260e+00
5.78003645e-01 1.48164453e-02 5.00564754e-01 -1.70502827e-01
2.02619702e-01 -4.92120922e-01 -2.13314533e-01 1.33685708e-01
-5.90661049e-01 -4.46340501e-01 -1.48356512e-01 6.42332852e-01
-6.05735667e-02 -7.09363997e-01 6.48011386e-01 -3.88894022e-01
-1.32349002e+00 5.68730421e-02 -3.60813737e-01 5.65376461e-01
8.25734496e-01 8.46343115e-02 -2.14693382e-01 -6.28372878e-02
-9.22369957e-01 -2.55260050e-01 4.69825804e-01 2.32723415e-01
3.07834536e-01 -1.25951457e+00 -5.20120978e-01 -1.50472403e-01
4.43996698e-01 -6.34976089e-01 1.86972633e-01 2.18719587e-01
8.11451450e-02 4.00399566e-01 3.42255980e-02 -6.74962997e-02
-8.49948525e-01 4.27531451e-01 1.17427297e-01 -4.36345130e-01
-5.45196116e-01 7.62581289e-01 -6.32905126e-01 -9.13258910e-01
1.46793842e-01 2.36677155e-01 -1.91383481e-01 -3.82026844e-02
7.97809243e-01 6.92825437e-01 1.70279413e-01 -6.07093632e-01
-5.81560135e-01 4.32368994e-01 -4.22225520e-02 -3.49863559e-01
1.57034278e+00 -1.58794731e-01 1.89090610e-01 6.74267948e-01
1.48906851e+00 -5.63269369e-02 -7.99564004e-01 -4.68375802e-01
4.88424987e-01 -6.70382231e-02 1.80475190e-01 -9.76768017e-01
-2.35379055e-01 8.11836481e-01 7.78891385e-01 -8.37622806e-02
6.48336530e-01 3.92053992e-01 1.07254720e+00 1.03678346e+00
8.79678965e-01 -1.26489353e+00 -4.67093706e-01 9.21691716e-01
-2.88376302e-01 -1.39596987e+00 -2.55646884e-01 9.64280497e-03
-5.66833019e-01 1.01635444e+00 2.24974602e-01 -1.20192498e-01
9.04008448e-01 4.63390023e-01 1.40432239e-01 -1.56560779e-01
-6.73156261e-01 -6.28797293e-01 2.44271919e-01 8.17772686e-01
7.33352780e-01 -7.82746822e-02 -1.63928986e-01 1.06168330e+00
1.71331733e-01 3.33852410e-01 1.58962503e-01 1.16648805e+00
-9.49695528e-01 -1.39317691e+00 9.61795598e-02 4.77699846e-01
-8.38333964e-01 -5.65500140e-01 -1.69820487e-01 3.99377376e-01
3.18582594e-01 1.17822492e+00 2.03633636e-01 -4.81749773e-01
3.15610230e-01 8.50192726e-01 3.39922726e-01 -1.10112309e+00
-5.39137423e-01 -3.32023233e-01 4.04657364e-01 -5.68783581e-01
-4.64389026e-01 -4.34161603e-01 -1.59469128e+00 -1.18771508e-01
-5.59373856e-01 6.06186390e-01 7.51943231e-01 1.38865590e+00
7.53388941e-01 -1.38283640e-01 4.81666863e-01 -6.28407121e-01
-4.54139709e-01 -8.31696868e-01 -7.56027520e-01 2.70686775e-01
-1.55952349e-01 -4.25102800e-01 -2.25349098e-01 8.79169852e-02] | [9.709789276123047, 9.40416145324707] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.