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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
948227c4-7e7c-44b6-b5cd-5dcae61b7959 | bilingual-word-embeddings-from-non-parallel | null | null | https://aclanthology.org/P15-2118 | https://aclanthology.org/P15-2118.pdf | Bilingual Word Embeddings from Non-Parallel Document-Aligned Data Applied to Bilingual Lexicon Induction | null | ['Marie-Francine Moens', "Ivan Vuli{\\'c}"] | 2015-07-01 | bilingual-word-embeddings-from-non-parallel-1 | https://aclanthology.org/P15-2118 | https://aclanthology.org/P15-2118.pdf | ijcnlp-2015-7 | ['multilingual-word-embeddings'] | ['methodology'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.201005458831787, 3.8050780296325684] |
49123d51-9f8d-4149-8155-3cd665953079 | reusable-workflows-for-gender-prediction | null | null | https://aclanthology.org/L18-1082 | https://aclanthology.org/L18-1082.pdf | Reusable workflows for gender prediction | null | ['Matej Martinc', 'Senja Pollak'] | 2018-05-01 | reusable-workflows-for-gender-prediction-1 | https://aclanthology.org/L18-1082 | https://aclanthology.org/L18-1082.pdf | lrec-2018-5 | ['gender-prediction'] | ['computer-vision'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.358174800872803, 3.82206654548645] |
f2c16c8c-043e-46b9-a60a-bd12430cecf6 | a-knowledge-driven-generative-model-for-multi | null | null | https://aclanthology.org/2020.emnlp-main.116 | https://aclanthology.org/2020.emnlp-main.116.pdf | A Knowledge-driven Generative Model for Multi-implication Chinese Medical Procedure Entity Normalization | Medical entity normalization, which links medical mentions in the text to entities in knowledge bases, is an important research topic in medical natural language processing. In this paper, we focus on Chinese medical procedure entity normalization. However, nonstandard Chinese expressions and combined procedures present challenges in our problem. The existing strategies relying on the discriminative model are poorly to cope with normalizing combined procedure mentions. We propose a sequence generative framework to directly generate all the corresponding medical procedure entities. we adopt two strategies: category-based constraint decoding and category-based model refining to avoid unrealistic results. The method is capable of linking entities when a mention contains multiple procedure concepts and our comprehensive experiments demonstrate that the proposed model can achieve remarkable improvements over existing baselines, particularly significant in the case of multi-implication Chinese medical procedures. | ['Chengqing Zong', 'Yu Zhou', 'Lu Xiang', 'Yining Wang', 'Jinghui Yan'] | null | null | null | null | emnlp-2020-11 | ['medical-procedure'] | ['medical'] | [ 5.17697573e-01 1.89525738e-01 -3.12573373e-01 -3.65885317e-01
-9.53078449e-01 -6.07250571e-01 4.32852745e-01 5.63538373e-01
-7.97675729e-01 9.04150426e-01 5.08101523e-01 -4.62839991e-01
-7.57042784e-03 -6.85021222e-01 -4.11494881e-01 -5.58401525e-01
3.20683032e-01 4.92542326e-01 -1.71625540e-01 -1.62461087e-01
1.25466868e-01 4.01924878e-01 -7.68797278e-01 1.54833660e-01
1.26262712e+00 1.14033781e-01 -8.88799727e-02 5.15777290e-01
-3.06546003e-01 7.36892819e-01 -6.78022563e-01 -9.29399848e-01
-5.25730997e-02 -5.05394399e-01 -6.84601307e-01 -8.36227313e-02
-1.02720559e-02 9.79513079e-02 -1.40797853e-01 1.33979797e+00
7.53603458e-01 1.81790099e-01 9.25289094e-01 -7.67052174e-01
-8.19422781e-01 8.99140835e-01 -5.01803219e-01 2.96639413e-01
4.07310069e-01 -4.81128931e-01 6.79844499e-01 -6.08081222e-01
8.67278934e-01 1.02422512e+00 6.01266384e-01 1.04989529e+00
-7.35943079e-01 -6.09264433e-01 2.38354489e-01 -1.27336711e-01
-1.60382152e+00 -3.71546626e-01 3.33714813e-01 -4.11967039e-01
8.24276745e-01 3.67098480e-01 1.17035292e-01 9.31060255e-01
4.05381083e-01 5.42745233e-01 5.13365865e-01 -4.79930073e-01
5.05713709e-02 5.88840432e-02 1.64534494e-01 6.35711491e-01
7.80439496e-01 -3.20616156e-01 -5.83021063e-03 -2.05412164e-01
6.45825565e-01 8.20386186e-02 -2.57562131e-01 1.67830124e-01
-1.48792541e+00 7.01988220e-01 9.84442234e-02 5.65020978e-01
-3.95337939e-01 -1.80464834e-01 5.49388528e-01 -2.94287324e-01
2.90594041e-01 7.24886000e-01 -4.69883382e-01 -1.01277074e-02
-9.90354657e-01 2.00485229e-01 8.64494979e-01 1.70762348e+00
2.34742016e-01 -1.34421691e-01 -6.69524968e-01 6.89715743e-01
-5.07346801e-02 2.81794429e-01 4.92699862e-01 -4.09592003e-01
5.95267236e-01 6.43339634e-01 -1.32182032e-01 -9.83166158e-01
-6.40775919e-01 -5.10988891e-01 -1.10558069e+00 -8.76301527e-01
-8.25615376e-02 -4.90183711e-01 -1.33280718e+00 1.61625707e+00
3.43225658e-01 2.97154993e-01 2.95667768e-01 4.67406183e-01
1.20560026e+00 2.59868562e-01 8.27490151e-01 -5.74188352e-01
1.63517964e+00 -1.11888385e+00 -1.46521866e+00 -7.92862326e-02
9.50185597e-01 -9.34367895e-01 4.11129326e-01 -2.89818197e-01
-9.67531443e-01 -1.74229831e-01 -6.84331536e-01 -2.53799617e-01
-4.96493578e-01 3.23572189e-01 7.16748476e-01 7.48580873e-01
-6.54071927e-01 2.30230570e-01 -1.09056938e+00 -4.70838577e-01
3.52934629e-01 4.77383047e-01 -4.47858125e-01 -8.49975795e-02
-1.08814991e+00 1.01272106e+00 7.91038394e-01 9.55438390e-02
-2.83272654e-01 -7.44043887e-01 -1.44713306e+00 2.25636899e-01
6.32739484e-01 -1.10067141e+00 1.31025946e+00 -2.14906946e-01
-1.02324831e+00 8.89908016e-01 -4.62056816e-01 -1.37452975e-01
2.23421335e-01 -9.11180973e-02 -8.90071154e-01 -1.28168166e-01
3.40645343e-01 2.89256334e-01 1.91475555e-01 -9.64317799e-01
-7.02953279e-01 -8.47604722e-02 -2.05846369e-01 3.76929492e-01
-3.62615645e-01 3.33149940e-01 -1.09817863e+00 -1.00503826e+00
2.64102429e-01 -1.00436676e+00 -8.00673664e-01 -6.81666970e-01
-8.08190823e-01 -1.96483195e-01 -9.45536122e-02 -8.10955107e-01
1.85547173e+00 -1.94636381e+00 -5.86313270e-02 2.04561725e-01
2.03437328e-01 3.00278157e-01 1.81696430e-01 2.83051252e-01
-2.72558928e-01 5.53289413e-01 -5.78496039e-01 -4.60202008e-01
-2.84664184e-02 2.66911745e-01 1.59386545e-01 2.98101157e-01
3.89106035e-01 1.07081389e+00 -9.78420973e-01 -1.08992612e+00
-8.07990357e-02 4.29965168e-01 -7.43048966e-01 1.99397504e-01
1.36129692e-01 4.72436041e-01 -7.41470575e-01 9.63357627e-01
7.44417667e-01 -3.51336032e-01 5.13840675e-01 -2.20104411e-01
-1.92704871e-02 2.17636675e-01 -9.96945202e-01 2.00762534e+00
2.21876595e-02 -1.51199266e-01 -2.42612392e-01 -6.69981658e-01
4.88503546e-01 5.69376707e-01 5.51359832e-01 -3.15393239e-01
3.61246258e-01 1.69656098e-01 6.31618947e-02 -6.76806152e-01
8.26993227e-01 -3.94095123e-01 -5.59037626e-01 -3.27069551e-01
1.63024798e-01 7.33583793e-02 6.62788033e-01 4.26618636e-01
1.01158023e+00 -2.04604492e-03 8.93794239e-01 -3.16555440e-01
7.61749446e-01 5.04100770e-02 1.16608548e+00 6.37670875e-01
-2.95431376e-01 6.20608032e-01 4.17516083e-01 1.39053985e-01
-7.47914553e-01 -8.24095011e-01 -2.74977267e-01 6.41373694e-01
1.25868320e-01 -6.43944323e-01 -7.38551497e-01 -9.94937122e-01
-3.33554029e-01 8.52793396e-01 -6.63394690e-01 -1.98827267e-01
-9.42841351e-01 -1.29312789e+00 7.39297748e-01 8.13843548e-01
6.99161217e-02 -7.94616818e-01 2.05760747e-02 4.34838653e-01
-5.18466711e-01 -1.42979372e+00 -8.99040341e-01 2.58616097e-02
-7.21878827e-01 -1.08357954e+00 -8.02921176e-01 -1.18138480e+00
1.16738021e+00 -2.66559988e-01 1.08807981e+00 7.58552104e-02
-2.32478157e-01 2.02788621e-01 -2.71598190e-01 -6.63673341e-01
-5.24671078e-01 3.74601960e-01 -8.62198621e-02 -3.20137739e-01
6.62447870e-01 -5.30021563e-02 -4.32002872e-01 -1.11314468e-01
-1.07469976e+00 4.30654064e-02 7.40452588e-01 7.41826653e-01
6.59713447e-01 -2.32792556e-01 3.99964422e-01 -1.71700561e+00
6.13376260e-01 -6.03505313e-01 -9.47954133e-02 7.13132024e-01
-6.98737979e-01 6.79632723e-02 2.80823469e-01 -2.68193215e-01
-1.33393490e+00 1.96798399e-01 -3.89744371e-01 5.17494828e-02
-3.37303072e-01 8.25172722e-01 -3.28723013e-01 2.17411324e-01
4.12256688e-01 5.88543452e-02 -7.46496022e-01 -2.85594523e-01
2.66884804e-01 4.14191455e-01 7.84441829e-01 -6.62393808e-01
6.16863370e-01 1.78067729e-01 -1.10221863e-01 -3.53468090e-01
-7.73659825e-01 -8.83576572e-01 -7.10940957e-01 2.95410633e-01
1.25179636e+00 -1.08657002e+00 -5.50388932e-01 8.63403305e-02
-1.24927676e+00 4.56856579e-01 1.08699324e-02 8.44802678e-01
-5.04979864e-02 5.96261084e-01 -9.79044139e-01 -5.52644014e-01
-5.39312482e-01 -1.04481256e+00 9.72236931e-01 6.48967028e-01
-1.71988398e-01 -1.08417034e+00 1.93920776e-01 8.87007192e-02
2.30312586e-01 5.37965655e-01 1.05100322e+00 -1.11488736e+00
-2.50947565e-01 -4.75481421e-01 -1.62758693e-01 -8.05313289e-02
4.98464048e-01 8.54592118e-03 -4.60693836e-01 -1.35129184e-01
-8.64056274e-02 4.96191293e-01 5.88149250e-01 2.51191705e-01
1.04258645e+00 -1.36544138e-01 -8.41591895e-01 7.15907395e-01
1.39582980e+00 4.02131647e-01 7.56716669e-01 1.69159159e-01
1.02060461e+00 2.39933372e-01 6.84163868e-01 3.13704014e-01
6.15084171e-01 2.88024634e-01 -1.55999497e-01 -3.35160464e-01
7.13071004e-02 -1.85275525e-02 -2.18506426e-01 1.36163867e+00
-2.89705217e-01 -4.84232783e-01 -1.10166955e+00 7.44144142e-01
-1.78406680e+00 -7.74032414e-01 -2.52148211e-01 1.87775087e+00
1.25523579e+00 -4.03264873e-02 -5.19869745e-01 -3.58025074e-01
1.05255282e+00 -3.64310145e-01 -1.96169987e-01 -3.38115633e-01
-1.62825331e-01 4.03790176e-01 6.35626554e-01 2.24153548e-01
-1.42336977e+00 1.14546633e+00 6.15797186e+00 7.25736439e-01
-7.12349117e-01 1.72050431e-01 3.62272233e-01 2.03601032e-01
-2.55234867e-01 -1.46184385e-01 -1.17575526e+00 2.81648576e-01
8.10029745e-01 -4.93826777e-01 -3.26205224e-01 6.66543901e-01
-9.27909315e-02 2.41255373e-01 -1.08264840e+00 8.94957185e-01
3.83767158e-01 -1.25188851e+00 2.76197463e-01 -1.36718199e-01
9.74774182e-01 -5.69103241e-01 -2.86381930e-01 7.13426948e-01
3.71528298e-01 -8.68288219e-01 1.97242245e-01 5.82117081e-01
8.27550054e-01 -7.01177359e-01 1.09976423e+00 1.34793103e-01
-1.28466690e+00 2.90792078e-01 -1.89218670e-01 5.39301872e-01
4.97024924e-01 4.23606783e-01 -8.95272613e-01 1.08504999e+00
3.19704503e-01 7.01661825e-01 -5.05777121e-01 1.27799332e+00
-3.06220949e-01 5.26411474e-01 -3.26244161e-02 1.68289661e-01
1.90974995e-02 -8.75857770e-02 2.74255544e-01 1.81955147e+00
3.01562995e-01 4.98776227e-01 2.63390869e-01 5.40131986e-01
-3.91167104e-01 6.47943735e-01 -4.17130828e-01 -1.65187806e-01
4.20097917e-01 1.35996139e+00 -9.59384680e-01 -6.32716954e-01
-5.53772926e-01 1.01388776e+00 3.17021072e-01 3.42796266e-01
-1.15527475e+00 -4.48746145e-01 3.92435908e-01 -4.22788590e-01
1.57350868e-01 -8.77990052e-02 -3.79525185e-01 -1.52297091e+00
-6.61705360e-02 -8.36130798e-01 8.97116601e-01 -2.41182640e-01
-1.12902927e+00 7.12756217e-01 -5.14944568e-02 -1.28473580e+00
-1.37451783e-01 -3.62893105e-01 -3.48144263e-01 7.23288476e-01
-1.69282377e+00 -1.32243586e+00 -9.81516321e-04 7.42690980e-01
2.85585254e-01 1.62435353e-01 1.24884367e+00 1.02509689e+00
-8.68277013e-01 1.06306970e+00 -1.27060041e-01 6.11048222e-01
1.07951760e+00 -1.34072042e+00 2.47798502e-01 1.20051014e+00
-2.20165670e-01 1.33510029e+00 5.21934628e-01 -1.11073470e+00
-8.71958613e-01 -1.27116120e+00 1.55971313e+00 -4.63138908e-01
2.77827263e-01 -2.42186517e-01 -8.83621216e-01 9.35125649e-01
1.97731435e-01 -2.09124476e-01 1.27622604e+00 1.33314386e-01
-7.76019916e-02 4.53995794e-01 -1.01062799e+00 8.38759243e-01
1.00437546e+00 -2.89261252e-01 -8.52073193e-01 5.31242073e-01
8.72564793e-01 -9.80693460e-01 -1.05030107e+00 8.38509083e-01
1.59949604e-02 1.00507334e-01 8.97859931e-01 -1.03982925e+00
5.04916966e-01 -4.06027436e-01 1.87921092e-01 -8.95699203e-01
-5.13866246e-01 -5.76665699e-01 2.36933976e-02 1.44094229e+00
8.26310217e-01 -2.10375756e-01 5.62258005e-01 1.05292761e+00
-4.68333483e-01 -3.67475450e-01 -6.12316787e-01 -4.50059503e-01
1.03988931e-01 -7.81422332e-02 4.48750913e-01 1.47367060e+00
1.67478487e-01 3.23188514e-01 -3.04391652e-01 4.49579865e-01
3.37245107e-01 -3.38839367e-02 2.49291971e-01 -8.74012709e-01
-2.67867092e-03 -3.55880082e-01 -3.83023560e-01 -5.64154744e-01
1.83759362e-01 -1.00834846e+00 2.14045718e-01 -1.77463543e+00
5.70301592e-01 -2.62557358e-01 -4.58845913e-01 6.12500370e-01
-9.25830424e-01 -1.07155433e-02 4.26835939e-02 -8.46658498e-02
-7.56876349e-01 2.58753181e-01 1.16900599e+00 -6.28954321e-02
-1.70022771e-01 4.42673229e-02 -1.08327186e+00 7.31141627e-01
7.20482230e-01 -8.75581145e-01 5.49123064e-02 -4.95877832e-01
1.43252522e-01 5.75913191e-02 -2.66266257e-01 -7.76749849e-01
6.50363982e-01 -2.89399803e-01 7.84933642e-02 -4.66440976e-01
-1.77685082e-01 -7.43442178e-01 1.24125622e-01 4.14063573e-01
-3.43315631e-01 1.10864520e-01 4.32971865e-01 6.14081979e-01
-2.79263586e-01 -3.48010331e-01 4.70722944e-01 -1.93822652e-01
-7.46178865e-01 3.59864205e-01 -4.76303577e-01 2.40670666e-01
9.80682492e-01 1.95151299e-01 -2.81054378e-01 1.34832248e-01
-1.00485647e+00 3.44426692e-01 1.35484561e-01 5.32936692e-01
3.30290109e-01 -1.18325830e+00 -9.16996717e-01 -2.46407449e-01
3.32603216e-01 1.43744707e-01 4.44608957e-01 8.95232797e-01
-6.30845726e-01 5.78995764e-01 2.03811690e-01 -1.09589785e-01
-1.33318794e+00 8.91596496e-01 8.73433426e-02 -7.09840953e-01
-3.66163135e-01 9.53098416e-01 3.45480025e-01 -7.22979903e-01
2.31125250e-01 -3.90575886e-01 -6.95307314e-01 -5.65953963e-02
4.15946454e-01 -3.69347855e-02 3.65805864e-01 -6.80115223e-01
-6.79131627e-01 4.71546888e-01 -4.33343440e-01 3.58040333e-01
1.16472602e+00 -1.68186389e-02 -3.77350926e-01 -2.45958250e-02
1.02464235e+00 4.13433701e-01 -1.87699273e-01 -2.47711390e-01
3.51193190e-01 1.05553921e-02 -2.39765361e-01 -8.27908933e-01
-9.45122778e-01 3.42471719e-01 2.62754291e-01 -3.91793489e-01
1.25483012e+00 -1.06776118e-01 5.99151969e-01 3.24207664e-01
7.97329098e-02 -1.10479534e+00 -7.64231980e-01 6.25512004e-01
4.27874714e-01 -1.23548961e+00 2.68146545e-01 -1.22559595e+00
-7.61702538e-01 9.13322330e-01 6.81716442e-01 2.12815255e-01
6.89715803e-01 4.83454704e-01 1.64145589e-01 -1.68945283e-01
-3.68502885e-01 -4.59367365e-01 5.76651514e-01 2.54712671e-01
8.50403011e-01 1.98193133e-01 -1.07259703e+00 1.13060582e+00
-2.01881472e-02 1.28517691e-02 5.08637249e-01 1.36115825e+00
2.55458236e-01 -1.30578673e+00 -1.08787902e-01 5.43249011e-01
-1.16396022e+00 -7.40226924e-01 8.72733518e-02 7.02092826e-01
2.78798759e-01 8.35217595e-01 -1.63342804e-01 -5.46329245e-02
6.77497089e-01 1.33756354e-01 2.39686191e-01 -1.06695628e+00
-9.10487771e-01 3.61569524e-01 1.59292012e-01 -2.04048127e-01
-7.46718585e-01 -7.14934647e-01 -1.66574478e+00 -5.21697775e-02
-6.42962873e-01 3.06164235e-01 4.19870347e-01 9.19369936e-01
3.79299194e-01 1.05671561e+00 4.76788841e-02 1.15255706e-01
-1.02058552e-01 -7.97517776e-01 -3.88997853e-01 5.88916421e-01
1.37035437e-02 -3.99470389e-01 4.44566868e-02 5.96892774e-01] | [8.637228965759277, 8.868586540222168] |
75a15c87-45bb-488b-bfa3-12c4288b4fa9 | learning-visual-motion-segmentation-using | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Mitrokhin_Learning_Visual_Motion_Segmentation_Using_Event_Surfaces_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Mitrokhin_Learning_Visual_Motion_Segmentation_Using_Event_Surfaces_CVPR_2020_paper.pdf | Learning Visual Motion Segmentation Using Event Surfaces | Event-based cameras have been designed for scene motion perception - their high temporal resolution and spatial data sparsity converts the scene into a volume of boundary trajectories and allows to track and analyze the evolution of the scene in time. Analyzing this data is computationally expensive, and there is substantial lack of theory on dense-in-time object motion to guide the development of new algorithms; hence, many works resort to a simple solution of discretizing the event stream and converting it to classical pixel maps, which allows for application of conventional image processing methods. In this work we present a Graph Convolutional neural network for the task of scene motion segmentation by a moving camera. We convert the event stream into a 3D graph in (x,y,t) space and keep per-event temporal information. The difficulty of the task stems from the fact that unlike in metric space, the shape of an object in (x,y,t) space depends on its motion and is not the same across the dataset. We discuss properties of of the event data with respect to this 3D recognition problem, and show that our Graph Convolutional architecture is superior to PointNet++. We evaluate our method on the state of the art event-based motion segmentation dataset - EV-IMO and perform comparisons to a frame-based method proposed by its authors. Our ablation studies show that increasing the event slice width improves the accuracy, and how subsampling and edge configurations affect the network performance.
| [' Yiannis Aloimonos', ' Cornelia Fermuller', ' Zhiyuan Hua', 'Anton Mitrokhin'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['motion-segmentation'] | ['computer-vision'] | [ 2.10432664e-01 -3.69597644e-01 1.30455913e-02 -7.40996450e-02
-8.39951560e-02 -6.16312444e-01 5.01337647e-01 2.77877659e-01
-7.56258607e-01 1.78989589e-01 -1.92441549e-02 -3.12656432e-01
-1.56617150e-01 -9.95027304e-01 -8.18949759e-01 -5.67415476e-01
-2.16125011e-01 2.91674733e-01 8.45481873e-01 9.78017375e-02
3.68107371e-02 9.44828212e-01 -1.41095674e+00 1.52112618e-01
2.49286070e-01 1.05617869e+00 7.86552802e-02 9.74745572e-01
-2.39696816e-01 8.86668980e-01 -3.64790142e-01 3.23677175e-02
3.69115621e-01 -5.59281051e-01 -9.09403324e-01 2.92908818e-01
5.75398386e-01 -3.29483628e-01 -7.99831033e-01 7.79290199e-01
8.53756592e-02 5.78167498e-01 6.24948025e-01 -1.29556715e+00
-2.25498259e-01 1.09180354e-01 -3.84238303e-01 5.21567225e-01
2.24341139e-01 1.42989248e-01 6.90086067e-01 -5.08863807e-01
1.14736855e+00 9.78242636e-01 8.66743028e-01 2.94190884e-01
-1.14135182e+00 -1.74871311e-02 1.38279870e-01 4.87444311e-01
-1.24029160e+00 6.15849160e-03 9.74264920e-01 -6.32941961e-01
1.34964514e+00 4.06473503e-02 1.15564716e+00 9.06940818e-01
3.08477998e-01 5.82240164e-01 4.40943837e-01 -3.04976761e-01
5.62536120e-01 -7.23351955e-01 1.89045981e-01 5.96709073e-01
6.76172748e-02 8.19229186e-02 -4.55960631e-01 1.90205291e-01
1.04987204e+00 1.26487911e-01 -4.32281613e-01 -7.68009126e-01
-1.32153869e+00 7.06908524e-01 5.40264189e-01 3.86563718e-01
-4.12752330e-01 6.53835416e-01 3.82953137e-01 4.61428650e-02
4.65936959e-01 -1.41775021e-02 -2.97384858e-01 -2.24062666e-01
-1.21910357e+00 3.38605732e-01 7.48265743e-01 6.58338904e-01
7.24450529e-01 1.06948957e-01 3.15746218e-02 1.79646119e-01
1.09858997e-01 9.82496738e-02 3.32647592e-01 -1.37026203e+00
2.07982332e-01 7.83808589e-01 -4.17317897e-02 -1.14859080e+00
-6.74486160e-01 1.27504826e-01 -7.20717847e-01 4.65724319e-01
8.84353638e-01 -1.24358371e-01 -9.80494678e-01 1.44724154e+00
4.03437257e-01 4.08265620e-01 -2.52349913e-01 1.10066915e+00
6.94874883e-01 8.68874907e-01 -1.06939096e-02 -1.83853924e-01
1.02631927e+00 -6.20645285e-01 -5.37576020e-01 -2.10817412e-01
4.87005562e-01 -2.84339786e-01 7.79866576e-01 2.67486453e-01
-1.17698622e+00 -5.95626771e-01 -1.05693555e+00 -2.89732158e-01
-4.60826457e-01 -3.31463069e-01 6.87024355e-01 4.15277034e-01
-1.05118001e+00 8.72155607e-01 -1.23865449e+00 -6.64624631e-01
5.18203080e-01 3.87299627e-01 -4.90259975e-01 -5.04427589e-02
-7.86057234e-01 6.67574823e-01 6.13882005e-01 -5.37715107e-02
-7.47655690e-01 -6.81673050e-01 -1.07191896e+00 1.08866312e-01
2.33452842e-01 -8.06019843e-01 1.22342229e+00 -8.46264243e-01
-1.15247989e+00 7.38091052e-01 -1.37617901e-01 -6.76507115e-01
6.71172321e-01 1.29879966e-01 -2.56025326e-02 4.61906224e-01
-7.70522505e-02 8.49448621e-01 6.90565467e-01 -8.39743197e-01
-7.85272896e-01 -2.66117930e-01 2.95458764e-01 2.49665245e-01
-1.38021661e-02 -2.07125396e-01 -7.63644814e-01 -4.81446356e-01
2.73340791e-01 -8.11985910e-01 -4.12940502e-01 2.35236004e-01
-3.77937928e-02 -2.27038041e-02 1.12967062e+00 -4.03567195e-01
8.89553964e-01 -2.26625490e+00 1.21367857e-01 -1.64480414e-02
2.34085605e-01 -2.45666341e-03 1.28376454e-01 3.12047124e-01
-1.87120542e-01 -5.00166649e-03 -6.66182935e-01 -3.54501396e-01
-3.14245194e-01 3.41646880e-01 -1.78481549e-01 8.89375985e-01
1.30293369e-01 9.80587602e-01 -9.00817573e-01 -5.47109723e-01
8.05717170e-01 6.76830888e-01 -4.55432117e-01 -1.44623429e-01
-4.45829719e-01 5.14296234e-01 -2.61656314e-01 2.90216357e-01
4.89946604e-01 -3.63408506e-01 -1.94611773e-01 -2.51066834e-01
-3.56180638e-01 4.80512530e-02 -1.31251156e+00 2.08867168e+00
-4.90263896e-03 1.21476305e+00 -1.72927588e-01 -1.04513097e+00
4.39478517e-01 2.62623757e-01 1.09799802e+00 -7.50941753e-01
2.83440053e-01 -2.90622443e-01 -1.32973969e-01 -4.24800038e-01
6.21099412e-01 4.68645990e-02 8.14664140e-02 2.64532417e-01
-2.85148751e-02 -3.44538391e-01 4.63684618e-01 1.27593577e-01
1.22913516e+00 3.64461929e-01 2.28434607e-01 -8.39095339e-02
2.21709475e-01 6.88753963e-01 4.46558297e-01 6.68992221e-01
-2.55534828e-01 9.71674681e-01 4.45778221e-01 -6.71642482e-01
-1.04251623e+00 -1.01644194e+00 -6.06358945e-02 5.42104959e-01
5.60216010e-01 -2.66977459e-01 -8.86471868e-01 -4.59458858e-01
-1.60399184e-01 6.65245950e-01 -6.90676928e-01 5.30626699e-02
-8.93641829e-01 -7.63938725e-01 2.54614204e-01 7.38730788e-01
5.94824910e-01 -1.08111715e+00 -1.32423818e+00 5.92219532e-01
-8.11169744e-02 -1.43470657e+00 -5.27962863e-01 2.20032915e-01
-1.02882826e+00 -1.18169606e+00 -3.74296755e-01 -5.19728482e-01
4.50746417e-01 3.07996005e-01 1.13229835e+00 -1.73557073e-01
-4.68412012e-01 8.55416834e-01 -3.14292282e-01 -2.00140417e-01
-1.90599620e-01 -1.55377820e-01 -3.69722486e-01 1.84218824e-01
2.47224778e-01 -6.72393918e-01 -8.34755361e-01 8.67597610e-02
-1.27297699e+00 1.06927395e-01 -1.85636550e-01 2.18331501e-01
7.48564184e-01 3.15259397e-01 -2.33507544e-01 -4.19800967e-01
1.28549814e-01 -2.64903367e-01 -8.53414357e-01 -1.17918901e-01
-9.83988196e-02 -2.62727767e-01 4.58460361e-01 -4.88135546e-01
-8.29164743e-01 6.32064164e-01 1.62903175e-01 -7.35470951e-01
-2.66647339e-01 3.99520546e-01 2.07492456e-01 -1.01851620e-01
6.06035590e-01 3.26473862e-02 -2.09505975e-01 -1.16428569e-01
5.82381666e-01 -1.11078799e-01 7.79688299e-01 -2.45020360e-01
4.02133942e-01 1.14015043e+00 3.89729917e-01 -9.73257184e-01
-3.90474349e-01 -6.95844769e-01 -1.05972075e+00 -5.24632514e-01
1.51802814e+00 -6.16572917e-01 -6.95298731e-01 4.98347491e-01
-1.48953915e+00 -7.22807288e-01 -7.59006202e-01 5.93706667e-01
-8.24211299e-01 2.82853931e-01 -6.03778601e-01 -5.65934062e-01
1.64090902e-01 -1.03296006e+00 9.87146854e-01 1.52536497e-01
-1.76301241e-01 -1.25969887e+00 1.50920808e-01 -6.03091344e-03
-1.55616133e-03 6.79994166e-01 7.16982007e-01 -1.60066903e-01
-9.66481566e-01 -6.97107315e-02 -1.77133158e-01 5.11668958e-02
-1.50797904e-01 3.03657502e-01 -7.07033873e-01 8.92660320e-02
2.83132434e-01 2.31827170e-01 8.44135404e-01 1.06164110e+00
1.14230442e+00 2.29485661e-01 -3.47260892e-01 9.23372030e-01
1.60090506e+00 3.82464111e-01 7.48939395e-01 2.72882968e-01
1.07316279e+00 4.83503997e-01 2.22287223e-01 2.88860321e-01
3.31885606e-01 6.51276708e-01 6.64927304e-01 -1.83808386e-01
-4.34101433e-01 7.69329220e-02 2.65151650e-01 3.75241101e-01
-1.09668128e-01 -5.13549507e-01 -1.11622834e+00 7.76394367e-01
-2.01472044e+00 -9.77216125e-01 -5.88967443e-01 2.05168509e+00
2.39549085e-01 1.18472084e-01 1.66965872e-01 2.17957035e-01
6.04308844e-01 2.78382957e-01 -5.56196570e-01 -2.71962851e-01
-2.14962870e-01 1.76675633e-01 8.70737910e-01 5.15043914e-01
-1.12069058e+00 8.08890939e-01 6.28664970e+00 4.74060416e-01
-1.34599471e+00 1.00182937e-02 5.10184348e-01 -2.62432843e-01
2.48150840e-01 1.24641759e-02 -4.26027834e-01 2.69852370e-01
8.91791165e-01 -2.44056106e-01 3.64448875e-01 5.42891979e-01
4.25132304e-01 -3.09394300e-01 -1.45532191e+00 1.10533011e+00
-4.74389904e-04 -1.58875191e+00 -5.92906848e-02 -7.19531178e-02
5.58884203e-01 3.64826858e-01 -3.39715064e-01 -2.59883881e-01
2.14943200e-01 -9.52061772e-01 9.62298870e-01 4.26203668e-01
4.47117537e-01 -5.42825997e-01 1.52105272e-01 2.52934545e-01
-1.40899491e+00 8.35854784e-02 -3.15115079e-02 -1.69755727e-01
5.90193570e-01 4.79591936e-01 -7.04179645e-01 6.41541302e-01
8.71810853e-01 9.18906033e-01 -5.23459136e-01 1.16749442e+00
2.63237208e-01 5.45730293e-01 -5.61097026e-01 2.68435687e-01
4.23047066e-01 -3.52899879e-01 6.57283962e-01 1.21389365e+00
3.46599221e-01 2.30791062e-01 2.16223106e-01 1.00227880e+00
7.06206262e-02 -3.40692401e-01 -7.56744623e-01 1.14221275e-01
-3.36804762e-02 9.15792584e-01 -1.36349988e+00 -2.35973671e-01
-6.13755763e-01 1.08287156e+00 5.65714128e-02 6.47708058e-01
-7.93158293e-01 -6.82531148e-02 5.41423023e-01 3.41978729e-01
6.24592900e-01 -8.82685184e-01 -3.94248128e-01 -9.69794512e-01
1.05973020e-01 -3.06472450e-01 5.29296696e-01 -8.89838099e-01
-9.96307909e-01 3.07527304e-01 3.48707587e-01 -1.18353736e+00
-3.44965518e-01 -6.42696977e-01 -5.23321033e-01 4.96988803e-01
-1.16265976e+00 -8.46903086e-01 -4.74309623e-01 8.90928686e-01
6.77344322e-01 4.49362099e-01 2.18093500e-01 1.93520308e-01
-3.86544228e-01 -2.14995518e-01 -5.90148307e-02 3.86717558e-01
-1.40333092e-02 -1.11025786e+00 7.45738924e-01 1.12349868e+00
4.17012572e-01 -2.68522967e-02 3.84708554e-01 -6.44092500e-01
-1.49890387e+00 -1.34200966e+00 5.35149992e-01 -6.99569821e-01
5.98583639e-01 -3.12818825e-01 -9.41354632e-01 7.57513821e-01
-2.16957256e-02 4.57999080e-01 -1.22816730e-02 -5.57181656e-01
3.08524612e-02 4.89389338e-02 -9.66628909e-01 7.60417044e-01
1.29152918e+00 -5.00953972e-01 -3.96739691e-01 7.94635639e-02
8.08660984e-01 -6.65592968e-01 -5.98153651e-01 3.26700032e-01
2.06884503e-01 -1.13373971e+00 1.16655827e+00 -3.70739728e-01
2.53668368e-01 -5.40076971e-01 -4.80635986e-02 -8.49055231e-01
-2.77609974e-01 -4.95188683e-01 -1.98084787e-01 5.97736239e-01
-1.53907701e-01 -2.81069696e-01 1.08221388e+00 5.82996488e-01
-2.58568019e-01 -4.35589820e-01 -1.24617314e+00 -7.28402674e-01
1.86901167e-02 -9.10206437e-01 2.34725952e-01 8.52230906e-01
-5.36416829e-01 1.19510099e-01 3.20955217e-01 2.00832129e-01
5.46210587e-01 1.37370065e-01 5.04679441e-01 -1.12330759e+00
-2.73797791e-02 -5.52418828e-01 -7.85494208e-01 -1.08162868e+00
4.84751426e-02 -7.42830276e-01 -2.18176991e-02 -1.75077820e+00
-2.74445087e-01 -6.30975747e-03 1.27684727e-01 1.60563007e-01
1.92189962e-01 2.75349110e-01 1.97221979e-01 4.69877310e-02
-4.96420443e-01 2.39875481e-01 1.17268145e+00 -1.76565170e-01
-4.03373152e-01 -4.99866903e-01 2.70792931e-01 8.36534917e-01
5.14591217e-01 -4.54491079e-01 -5.94590724e-01 -6.59764946e-01
1.30210400e-01 2.71211952e-01 7.94218302e-01 -1.32357228e+00
5.84519625e-01 -1.87957212e-01 3.36288661e-01 -8.20505321e-01
5.24858117e-01 -1.08127427e+00 5.33725560e-01 2.90847212e-01
-1.72390081e-02 2.65107453e-01 3.94288480e-01 7.22855628e-01
-2.07194969e-01 -1.78316966e-01 6.87886775e-01 -3.57233644e-01
-1.12606359e+00 6.54085338e-01 -4.61635441e-01 5.34691401e-02
1.10478759e+00 -7.68018723e-01 9.38632246e-03 -2.79319018e-01
-7.43219316e-01 -1.07611142e-01 7.34936774e-01 2.46263638e-01
5.55120528e-01 -1.16276324e+00 -5.01110137e-01 2.74821110e-02
-1.31326213e-01 4.68209773e-01 3.44228983e-01 6.89184964e-01
-1.09598804e+00 3.21553051e-01 -2.60562867e-01 -1.16517127e+00
-1.03512025e+00 5.80183983e-01 6.44637883e-01 6.92657828e-02
-1.18987203e+00 5.10257304e-01 4.43541020e-01 1.11146323e-01
1.46344408e-01 -8.51934195e-01 -5.84240630e-02 2.37094015e-02
1.63546726e-01 4.93787944e-01 3.00266981e-01 -6.48730576e-01
-3.90496701e-01 7.89403021e-01 3.63713056e-01 -3.77187192e-01
1.27906454e+00 -7.93802589e-02 1.12708481e-02 6.83703244e-01
1.25654221e+00 -5.47029495e-01 -1.58935547e+00 1.40914440e-01
-5.95858470e-02 -3.75377029e-01 3.28404188e-01 -1.37167886e-01
-1.13999331e+00 8.25974882e-01 5.99415064e-01 4.77023512e-01
1.16749203e+00 -9.77870971e-02 8.68480921e-01 1.35529429e-01
2.05901101e-01 -9.00904596e-01 -1.16003104e-01 6.97587729e-01
4.98199999e-01 -1.00777221e+00 6.46031229e-03 -5.03056467e-01
-2.32472479e-01 1.21433735e+00 2.68584043e-01 -2.81198114e-01
5.97622156e-01 2.28042915e-01 -2.06109941e-01 -4.65296298e-01
-3.66273344e-01 -3.09795588e-01 3.63264173e-01 7.44803250e-01
1.42128095e-01 -2.06518814e-01 -1.02805467e-02 -8.62933099e-02
-8.02740976e-02 2.06572101e-01 6.35790586e-01 1.00687063e+00
-6.25440627e-02 -7.42172122e-01 -4.08009708e-01 1.52639404e-01
-2.88306028e-01 1.59934208e-01 -2.74639338e-01 1.07979357e+00
3.38191420e-01 8.58126581e-01 6.65197968e-01 1.25355804e-02
3.87053251e-01 -4.22013145e-05 7.83549190e-01 -4.46348101e-01
-5.04787683e-01 -1.72837794e-01 -3.45413573e-02 -9.09575999e-01
-7.10513890e-01 -9.14589643e-01 -1.61840463e+00 -2.11732283e-01
1.03257090e-01 -3.57489079e-01 7.60950387e-01 8.94141436e-01
2.92664528e-01 6.45192385e-01 1.75787732e-01 -1.17219234e+00
3.44986230e-01 -2.60302722e-01 -3.65647733e-01 7.15949774e-01
5.62153041e-01 -4.77248728e-01 -2.41650701e-01 6.36356533e-01] | [8.606322288513184, -1.3034435510635376] |
4b523ae6-6cf2-45e1-8faa-aa9ed23f8f1a | infrastructure-crack-segmentation-boundary | 2306.09196 | null | https://arxiv.org/abs/2306.09196v1 | https://arxiv.org/pdf/2306.09196v1.pdf | Infrastructure Crack Segmentation: Boundary Guidance Method and Benchmark Dataset | Cracks provide an essential indicator of infrastructure performance degradation, and achieving high-precision pixel-level crack segmentation is an issue of concern. Unlike the common research paradigms that adopt novel artificial intelligence (AI) methods directly, this paper examines the inherent characteristics of cracks so as to introduce boundary features into crack identification and then builds a boundary guidance crack segmentation model (BGCrack) with targeted structures and modules, including a high frequency module, global information modeling module, joint optimization module, etc. Extensive experimental results verify the feasibility of the proposed designs and the effectiveness of the edge information in improving segmentation results. In addition, considering that notable open-source datasets mainly consist of asphalt pavement cracks because of ease of access, there is no standard and widely recognized dataset yet for steel structures, one of the primary structural forms in civil infrastructure. This paper provides a steel crack dataset that establishes a unified and fair benchmark for the identification of steel cracks. | ['Yu-Hsing Wang', 'Jian Zhang', 'Wang Chen', 'Zhili He'] | 2023-06-15 | null | null | null | null | ['crack-segmentation'] | ['computer-vision'] | [-5.55925369e-02 -1.28326580e-01 -6.35601953e-02 -1.95039600e-01
-5.21820664e-01 -8.23670030e-02 -9.16480795e-02 2.14647949e-01
-1.80998117e-01 4.65400219e-01 -2.14678422e-01 -3.34001422e-01
-3.49122077e-01 -1.21190631e+00 -2.51982599e-01 -8.28929186e-01
2.32641175e-01 2.85420477e-01 4.61246252e-01 -2.54863858e-01
9.42887306e-01 7.29088724e-01 -1.49618721e+00 6.73046634e-02
1.19868886e+00 9.81682777e-01 -4.06869175e-03 4.47687298e-01
5.07050157e-02 3.07435423e-01 -1.48084193e-01 6.48296624e-02
-7.00865313e-02 -2.64605079e-02 -9.51480627e-01 5.93452081e-02
3.81449936e-03 -4.06757832e-01 -2.74422437e-01 8.38335872e-01
5.76453447e-01 -1.17203876e-01 8.43964458e-01 -9.18863833e-01
-4.22674477e-01 6.59808934e-01 -9.09639060e-01 1.73862502e-01
2.52294421e-01 2.25586221e-01 1.13988435e+00 -1.14116693e+00
2.90747434e-01 7.45833039e-01 1.01209092e+00 1.54263452e-01
-5.99204898e-01 -2.84344316e-01 -1.85629457e-01 4.94395226e-01
-1.19201970e+00 -1.30591951e-02 1.27188528e+00 -7.35422313e-01
5.44801772e-01 3.80699307e-01 6.41950428e-01 3.22816223e-01
1.48080617e-01 4.75389570e-01 7.29085803e-01 -4.51340735e-01
3.49165559e-01 -3.27379882e-01 7.86552131e-01 8.25402498e-01
5.83517134e-01 -2.22484525e-02 -6.75190985e-02 2.53314197e-01
7.23797083e-01 -1.56200424e-01 -3.86149138e-01 -2.85915062e-02
-7.74508238e-01 7.12580442e-01 6.47070289e-01 5.23790777e-01
-1.43171266e-01 2.06407756e-01 1.56046972e-01 9.12099704e-02
4.65678461e-02 2.16803122e-02 -7.51016885e-02 7.63337240e-02
-9.02869523e-01 9.55594033e-02 3.33321750e-01 4.89430457e-01
9.51757789e-01 3.18458676e-01 5.72721288e-02 8.22943330e-01
7.32304335e-01 4.96710747e-01 3.51821445e-02 -9.56459880e-01
2.61606991e-01 9.82574165e-01 -2.83750117e-01 -1.27198148e+00
-7.02694654e-01 -3.70189965e-01 -7.46864498e-01 3.37203979e-01
2.31501758e-01 -2.73956716e-01 -6.71107829e-01 7.28428781e-01
4.97601300e-01 4.54530157e-02 -3.15281272e-01 8.06142628e-01
8.70809913e-01 3.81931007e-01 -3.33564103e-01 3.00339442e-02
1.16417348e+00 -9.01903570e-01 -6.06522441e-01 -1.41914994e-01
5.59088707e-01 -5.75453222e-01 9.07074273e-01 3.45363617e-01
-8.16850722e-01 -3.88225526e-01 -1.23536777e+00 1.78520918e-01
-3.73849392e-01 2.58017331e-01 6.27507985e-01 9.96806204e-01
-6.76117003e-01 3.49149108e-01 -6.31416380e-01 -6.76386282e-02
4.79478091e-01 5.38124703e-03 8.34705904e-02 -1.88849881e-01
-1.09926474e+00 6.61046267e-01 1.87873557e-01 6.40845716e-01
-7.45977938e-01 -8.06712031e-01 -6.85475051e-01 -2.42884934e-01
1.13845691e-01 -3.09140801e-01 6.42539024e-01 1.02302223e-01
-1.14296460e+00 6.00543141e-01 1.99477419e-01 -6.64023533e-02
2.72634536e-01 -1.77242279e-01 -2.87987679e-01 6.28661394e-01
9.77732390e-02 9.96373594e-02 7.49630392e-01 -1.76706791e+00
-5.88118374e-01 -2.76452124e-01 8.93080682e-02 -2.73008823e-01
-3.69186312e-01 -2.39800602e-01 -1.45548388e-01 -6.14266992e-01
5.43829978e-01 -4.75829035e-01 -3.92026216e-01 -1.32894337e-01
-6.70410454e-01 -8.81320909e-02 1.13280571e+00 -8.45945537e-01
1.69868958e+00 -1.84976971e+00 -2.14729428e-01 7.69248307e-01
2.01120004e-01 -1.52115673e-01 4.24275666e-01 5.81568360e-01
-9.87429991e-02 1.63646594e-01 -1.17234588e+00 1.09884508e-01
-1.09982371e-01 1.60014763e-01 4.24716204e-01 5.13703763e-01
1.24985337e-01 5.40362060e-01 -6.55158103e-01 -6.64267182e-01
8.93019810e-02 -9.72567350e-02 -5.97708702e-01 -3.11288297e-01
2.12091729e-01 3.21908414e-01 -7.47583508e-01 1.29951286e+00
9.55107808e-01 1.70345545e-01 -5.47516286e-01 -7.28246987e-01
-5.87896943e-01 -7.42740691e-01 -1.36255741e+00 1.52442527e+00
-1.26561627e-01 3.25367928e-01 2.36196682e-01 -1.16244173e+00
1.05139732e+00 4.72149909e-01 9.07101870e-01 -3.81093860e-01
4.01268125e-01 4.09530640e-01 -3.10039312e-01 -9.64048684e-01
3.89200002e-01 1.98035911e-01 1.60056225e-03 2.94833779e-01
-5.82621574e-01 -6.54098570e-01 3.11099559e-01 9.19529125e-02
1.11563170e+00 -1.64835393e-01 -6.03268564e-01 -5.84407389e-01
5.84420025e-01 3.44948947e-01 4.49529439e-01 3.62774998e-01
-2.57398069e-01 1.03862381e+00 -4.89556640e-02 -2.78222769e-01
-8.29340577e-01 -9.79682624e-01 -5.13791740e-01 1.72229782e-01
4.81379002e-01 3.36618513e-01 -1.07565594e+00 -2.65876770e-01
1.81755409e-01 3.42279077e-01 -5.89658797e-01 -2.01584473e-02
-5.92430174e-01 -1.27413535e+00 7.12873876e-01 6.61510527e-01
9.13700521e-01 -7.47074544e-01 -7.51562178e-01 2.07676664e-01
-3.02928805e-01 -8.71157944e-01 -2.67991364e-01 -1.60304025e-01
-1.08145106e+00 -1.58233559e+00 -4.81538206e-01 -1.01749432e+00
8.77646208e-01 2.27506757e-01 7.77924716e-01 8.48256111e-01
-8.90636206e-01 4.13821131e-01 -4.36792821e-01 5.16432747e-02
-1.20785691e-01 -5.63217327e-02 -4.96064365e-01 1.68135330e-01
-2.18866408e-01 -2.46866509e-01 -9.25881386e-01 2.75201827e-01
-9.04126823e-01 -2.68334746e-01 5.28663158e-01 6.20031297e-01
2.66603887e-01 6.04136586e-01 7.86382139e-01 -8.04723859e-01
6.35047436e-01 -4.73191172e-01 -2.57342905e-01 2.64498234e-01
-1.02616620e+00 -4.98941571e-01 -2.15332553e-01 4.55151856e-01
-1.39322877e+00 -1.61643967e-01 -5.52592576e-01 1.73604816e-01
-4.10480827e-01 1.06403363e+00 -4.66447435e-02 -2.65901595e-01
4.57687140e-01 -2.16830105e-01 1.80086272e-03 -6.10296786e-01
1.24258161e-01 7.29450464e-01 8.16234350e-01 -1.09956408e+00
1.03352821e+00 7.28778541e-01 6.82725012e-02 -1.10096037e+00
-4.53333497e-01 -7.95518696e-01 -8.31135988e-01 -9.76588309e-01
1.05921721e+00 -4.69532371e-01 -6.11591995e-01 1.04317665e+00
-1.07159650e+00 -5.41681945e-02 -1.86673999e-01 3.29861253e-01
-2.15983003e-01 7.60464013e-01 -1.09196687e+00 -7.63105273e-01
-3.71256173e-01 -1.25971162e+00 6.88337445e-01 4.36472267e-01
3.41148853e-01 -9.62870359e-01 8.61266106e-02 9.44879234e-01
2.87473828e-01 6.81520820e-01 1.22476971e+00 3.59675795e-01
-5.94176352e-01 -2.69501388e-01 -3.31093103e-01 3.65807503e-01
1.04642496e-01 9.00335133e-01 -8.78359020e-01 -6.05540723e-02
-6.99743405e-02 7.51781389e-02 8.17656636e-01 5.67805469e-01
6.59428954e-01 4.42859799e-01 -4.83471572e-01 -1.30017987e-02
2.02463961e+00 3.73959631e-01 1.02449429e+00 3.99836600e-01
7.77324080e-01 6.93912804e-01 6.86128974e-01 5.08897483e-01
5.21054506e-01 1.78145289e-01 8.36144626e-01 -4.23157424e-01
-8.73415694e-02 5.51536977e-01 -2.05456674e-01 1.12376606e+00
-5.33216774e-01 1.99338853e-01 -1.39183080e+00 8.72440100e-01
-1.57027388e+00 -9.13911104e-01 -1.18164527e+00 1.69498658e+00
6.37188256e-01 2.67393202e-01 -2.33687729e-01 1.04039192e+00
8.88123870e-01 -4.16380256e-01 -3.13139051e-01 -2.35366255e-01
-1.35198325e-01 2.22005919e-01 3.84876341e-01 3.68317693e-01
-9.85103667e-01 3.08372617e-01 6.22439957e+00 7.86830485e-01
-6.53760314e-01 7.05094561e-02 5.39418399e-01 7.39398539e-01
-4.64589149e-01 6.70207739e-02 -4.32068527e-01 3.33855659e-01
2.34684616e-01 4.37025607e-01 -7.78367743e-02 4.77362692e-01
4.35694396e-01 -3.74272883e-01 -5.19175529e-01 5.57401180e-01
-3.66193026e-01 -1.40718663e+00 -4.32180762e-01 -2.14950845e-01
8.82450402e-01 -2.51752317e-01 -1.12387635e-01 -3.32096875e-01
-3.10742147e-02 -5.84363103e-01 9.01613116e-01 6.51410997e-01
4.61694926e-01 -7.76157320e-01 7.89145589e-01 -6.70503313e-03
-1.50857127e+00 -4.78261620e-01 -1.90908045e-01 -9.57685709e-03
4.91686881e-01 1.09523535e+00 -9.38769132e-02 1.17016637e+00
1.03578806e+00 7.73892343e-01 -5.23015559e-01 1.49948299e+00
-1.94145236e-02 8.26099813e-01 -4.95876312e-01 5.24537206e-01
3.37749720e-01 -4.60154444e-01 1.83375373e-01 1.03231645e+00
2.91270614e-01 1.65544674e-01 9.87531617e-02 8.10689390e-01
2.45939180e-01 8.65707174e-02 -2.22974494e-01 5.56904554e-01
6.87265456e-01 1.13771641e+00 -1.11986887e+00 2.54878461e-01
-6.51270747e-01 2.10106224e-01 -3.01885009e-01 3.32819730e-01
-1.02191710e+00 -3.82256031e-01 3.36340487e-01 1.32322237e-01
2.75489496e-04 -2.03884706e-01 -1.11063778e+00 -5.17250121e-01
-2.07849666e-01 -3.83115202e-01 3.78894389e-01 -6.31525338e-01
-1.12159741e+00 8.37883651e-02 1.20974630e-02 -1.00316095e+00
6.73499823e-01 -5.07247090e-01 -1.20280540e+00 2.60482699e-01
-1.69264996e+00 -1.35549510e+00 -6.13733649e-01 4.16576624e-01
6.48551881e-01 3.24841291e-01 1.08955048e-01 7.98605740e-01
-1.32117748e+00 3.24646711e-01 3.16532463e-01 4.49684352e-01
5.84642589e-02 -1.01481056e+00 -1.27049228e-02 1.05124402e+00
-4.90219027e-01 2.01498866e-01 6.46588445e-01 -7.30925560e-01
-1.33067811e+00 -8.00037205e-01 1.08240090e-01 -9.86551642e-02
5.35100877e-01 4.25185293e-01 -1.02191889e+00 1.67931437e-01
1.47924677e-01 -2.04453573e-01 2.13971406e-01 -3.64873290e-01
2.65806079e-01 -5.10720760e-02 -1.32725871e+00 2.71862864e-01
8.44221175e-01 -1.68251440e-01 -4.39193934e-01 -1.82426702e-02
2.99527317e-01 -2.99271613e-01 -1.36264503e+00 8.93272460e-01
4.93679017e-01 -1.18219304e+00 1.25627816e+00 2.59665817e-01
4.40646321e-01 -3.20276350e-01 1.10811464e-01 -7.08709538e-01
-2.55406916e-01 1.54209640e-02 8.99565443e-02 1.51447082e+00
3.99923205e-01 -6.62498295e-01 1.27462852e+00 6.42780602e-01
-9.89345610e-01 -9.04250026e-01 -9.57762182e-01 -3.15142483e-01
1.56451270e-01 -4.85596180e-01 7.27381587e-01 8.91107261e-01
-7.45941252e-02 -2.19431385e-01 3.29615176e-01 6.10493600e-01
9.78329778e-01 3.05625498e-01 2.39366695e-01 -1.61472690e+00
3.33635598e-01 -6.81175053e-01 -2.01014459e-01 -7.43401885e-01
-3.44676614e-01 -2.84436464e-01 3.76545161e-01 -1.88922215e+00
-1.47126555e-01 -1.12801063e+00 -1.05162680e-01 2.22264245e-01
-1.15289040e-01 2.67559856e-01 -4.76433039e-01 2.68154711e-01
2.32719049e-01 3.28540742e-01 1.42594969e+00 -4.95094270e-01
3.10574055e-01 1.37183696e-01 -2.43242607e-01 8.50125432e-01
8.47066045e-01 -4.03949589e-01 -2.99239159e-01 -5.27529955e-01
1.04294486e-01 4.95819487e-02 3.44361126e-01 -1.34965873e+00
4.02117372e-01 -2.09098935e-01 -1.65441662e-01 -1.07117581e+00
-8.84965733e-02 -1.19062483e+00 2.42435277e-01 8.81328285e-01
2.46747583e-01 -2.04275548e-02 -2.21516788e-02 5.89181185e-01
-3.97871673e-01 -8.88626993e-01 1.02198470e+00 -9.65127163e-03
-9.60311174e-01 2.01748386e-01 -5.17149925e-01 1.35266930e-02
1.24925852e+00 -9.36508358e-01 -4.54870522e-01 3.93471986e-01
-5.48226118e-01 3.13187987e-01 4.68464792e-01 -8.45696703e-02
9.11397219e-01 -9.81023610e-01 -7.25998104e-01 1.50063559e-01
-4.02891040e-02 3.93649310e-01 6.25490606e-01 8.14071178e-01
-1.43015075e+00 5.52338958e-02 -9.56854895e-02 -7.01611936e-01
-7.57805169e-01 1.22630261e-01 3.56952786e-01 -7.05754980e-02
-5.40746093e-01 8.87439728e-01 -3.70472610e-01 -5.68394400e-02
-1.97988719e-01 -6.45826995e-01 -4.51181710e-01 2.16797531e-01
-2.20011532e-01 1.07594395e+00 5.62467217e-01 -6.18261516e-01
-2.24676028e-01 1.19746363e+00 4.79267865e-01 4.26408172e-01
1.25417495e+00 -4.11825329e-01 -3.43852431e-01 1.40228316e-01
7.12535024e-01 -1.65967479e-01 -1.03140116e+00 1.51158184e-01
3.71689707e-01 -2.40431681e-01 2.98707843e-01 -3.53710145e-01
-1.78772235e+00 8.73550653e-01 6.59552813e-01 2.82398432e-01
1.15548348e+00 -2.37147480e-01 1.15314972e+00 1.43649697e-01
5.34717381e-01 -1.65296185e+00 2.51054853e-01 1.91558838e-01
5.77175558e-01 -1.09774768e+00 2.01602086e-01 -8.40934694e-01
-1.43184811e-01 1.29009366e+00 6.57717049e-01 4.93040793e-02
1.10814822e+00 5.24339378e-01 1.64024904e-01 -7.97982872e-01
1.06958682e-02 -5.52012809e-02 -1.74980447e-01 4.87142146e-01
5.90819716e-02 -3.65283817e-01 -3.82879794e-01 3.72483909e-01
4.54886734e-01 -3.76777411e-01 5.75899184e-01 1.46853495e+00
-1.04312658e+00 -8.83669853e-01 -7.87841499e-01 4.19251084e-01
-2.72363186e-01 2.03384683e-01 2.49318466e-01 7.57426322e-01
2.05313221e-01 1.24398792e+00 -3.36662143e-01 -2.39064112e-01
3.31138730e-01 -2.43799224e-01 1.20500728e-01 -4.01496738e-01
-5.92062593e-01 -4.31288421e-01 -1.17672123e-02 -3.24848630e-02
-5.60400844e-01 -5.73119819e-01 -1.79232240e+00 -1.64115220e-01
-8.81365538e-01 3.02814275e-01 6.79465294e-01 9.84751463e-01
-1.61557421e-01 5.93705297e-01 9.89519477e-01 -7.06442952e-01
-9.87325311e-02 -6.13900363e-01 -8.07398260e-01 3.61193031e-01
-1.24652922e-01 -9.86425996e-01 -2.69022226e-01 4.75519150e-01] | [7.494196891784668, 1.451162576675415] |
462d897a-c243-4a75-8f37-5dbac0feedaf | cloudattention-efficient-multi-scale | 2208.00524 | null | https://arxiv.org/abs/2208.00524v1 | https://arxiv.org/pdf/2208.00524v1.pdf | CloudAttention: Efficient Multi-Scale Attention Scheme For 3D Point Cloud Learning | Processing 3D data efficiently has always been a challenge. Spatial operations on large-scale point clouds, stored as sparse data, require extra cost. Attracted by the success of transformers, researchers are using multi-head attention for vision tasks. However, attention calculations in transformers come with quadratic complexity in the number of inputs and miss spatial intuition on sets like point clouds. We redesign set transformers in this work and incorporate them into a hierarchical framework for shape classification and part and scene segmentation. We propose our local attention unit, which captures features in a spatial neighborhood. We also compute efficient and dynamic global cross attentions by leveraging sampling and grouping at each iteration. Finally, to mitigate the non-heterogeneity of point clouds, we propose an efficient Multi-Scale Tokenization (MST), which extracts scale-invariant tokens for attention operations. The proposed hierarchical model achieves state-of-the-art shape classification in mean accuracy and yields results on par with the previous segmentation methods while requiring significantly fewer computations. Our proposed architecture predicts segmentation labels with around half the latency and parameter count of the previous most efficient method with comparable performance. The code is available at https://github.com/YigeWang-WHU/CloudAttention. | ['Federico Tombari', 'Benjamin Busam', 'Nassir Navab', 'Yige Wang', 'Mahdi Saleh'] | 2022-07-31 | null | null | null | null | ['scene-segmentation'] | ['computer-vision'] | [-2.74817813e-02 -1.14021257e-01 1.43630952e-01 -4.66299355e-01
-9.60620999e-01 -6.68278813e-01 2.67447352e-01 5.22851050e-01
-3.59877884e-01 -7.58407405e-03 -2.14090809e-01 -3.34277987e-01
3.28386156e-03 -8.95387352e-01 -1.01619244e+00 -5.35119653e-01
5.43774255e-02 8.17243278e-01 5.81605196e-01 2.15798974e-01
5.38460433e-01 1.05001307e+00 -1.53326881e+00 2.83725590e-01
1.00076592e+00 1.34262860e+00 5.18409491e-01 6.91705108e-01
-5.18001616e-01 2.00601891e-01 -2.43699849e-01 -2.50711828e-01
3.53694975e-01 3.38508606e-01 -9.54458773e-01 1.02065951e-01
8.94477427e-01 -1.55313104e-01 -1.65304571e-01 9.39690828e-01
5.33202171e-01 9.64275077e-02 4.60369498e-01 -1.28912139e+00
-8.18084478e-01 4.11225498e-01 -7.36535609e-01 2.92407513e-01
-1.78978160e-01 4.36416417e-02 1.01605761e+00 -1.32385600e+00
1.42136514e-01 1.25026596e+00 7.46544957e-01 6.13923036e-02
-9.97237861e-01 -6.42750859e-01 4.62083787e-01 2.32527226e-01
-1.65104437e+00 -2.53593773e-01 7.55450010e-01 -3.57296050e-01
1.37829125e+00 3.22622597e-01 8.45314443e-01 1.72062114e-01
-1.26628950e-01 9.70462143e-01 5.60621858e-01 -6.04148880e-02
2.61404753e-01 -2.90173769e-01 1.49304733e-01 7.34869659e-01
1.95432782e-01 -5.14881015e-01 -2.53791004e-01 -1.47933960e-02
9.42670107e-01 3.70100468e-01 2.58342177e-02 -4.27771300e-01
-1.12579560e+00 7.25325167e-01 1.02419889e+00 1.96137175e-01
-4.91014212e-01 5.37322283e-01 2.59559304e-01 -6.57546371e-02
7.02033460e-01 2.27237254e-01 -6.03557765e-01 8.34735706e-02
-1.00077426e+00 8.88072625e-02 4.08724248e-01 1.32757211e+00
8.06172669e-01 1.01510007e-02 -9.28106457e-02 7.31747568e-01
1.28280908e-01 6.79619193e-01 7.61468410e-02 -9.38445270e-01
3.49506140e-01 9.30317819e-01 -1.21564858e-01 -9.30832267e-01
-4.49314982e-01 -5.14764011e-01 -8.14711273e-01 7.01137707e-02
9.97440442e-02 3.28505188e-01 -1.25159144e+00 1.23534644e+00
6.73534036e-01 5.16593397e-01 -4.87224191e-01 9.60374713e-01
7.10764170e-01 6.88881576e-01 7.42197484e-02 2.10660666e-01
1.42519569e+00 -1.13476706e+00 -1.48934379e-01 -1.17437065e-01
4.49914813e-01 -7.83127487e-01 1.12716007e+00 1.10901363e-01
-1.30584741e+00 -5.23619533e-01 -7.23061502e-01 -7.04194844e-01
-5.19719601e-01 6.37630075e-02 8.98482382e-01 4.29785728e-01
-1.19018531e+00 5.49704492e-01 -1.14592874e+00 -3.95948857e-01
1.03303277e+00 7.21591532e-01 -4.18511778e-02 4.53316346e-02
-3.16230863e-01 4.32447404e-01 1.73014961e-02 2.12006494e-01
-5.83586037e-01 -9.19257522e-01 -7.90740252e-01 2.83714443e-01
2.41028070e-01 -6.11162722e-01 1.27471888e+00 -5.31421185e-01
-1.24135470e+00 7.67302871e-01 -4.57316399e-01 -2.99977481e-01
1.51549861e-01 -3.53945106e-01 3.65766943e-01 1.25423640e-01
1.78663939e-01 1.07731950e+00 7.59349883e-01 -1.13331127e+00
-5.79279304e-01 -6.77477121e-01 -1.03382021e-01 3.63479137e-01
-1.96843550e-01 -4.09571230e-02 -9.32952166e-01 -4.50712651e-01
5.87325037e-01 -8.68745089e-01 -4.53227550e-01 1.58170357e-01
-3.48906308e-01 -4.50582743e-01 1.03697252e+00 -3.05327564e-01
8.26443017e-01 -2.10426569e+00 2.07552817e-02 1.81636617e-01
4.21166837e-01 1.69489637e-01 -8.36973935e-02 1.38299957e-01
1.20907649e-01 2.12937608e-01 -4.92607504e-01 -8.06513309e-01
1.54453600e-02 2.18932509e-01 -3.85243446e-01 5.38962662e-01
3.42096686e-01 1.13882685e+00 -5.81776917e-01 -6.45300627e-01
3.68715703e-01 4.64735389e-01 -8.55070174e-01 2.42126663e-03
-2.23602444e-01 2.36207053e-01 -5.44854581e-01 1.08392727e+00
9.86835599e-01 -6.40318036e-01 -3.95406216e-01 -3.04328650e-01
-4.33257014e-01 1.74403518e-01 -9.97173548e-01 1.83304286e+00
-5.13497710e-01 4.15123492e-01 1.52163893e-01 -1.01075733e+00
7.79964328e-01 -2.56600999e-03 6.51181996e-01 -3.26079190e-01
2.17322990e-01 2.15753376e-01 -1.80534452e-01 -1.60438120e-01
6.61061585e-01 3.22531044e-01 -8.44800323e-02 2.94978946e-01
-1.16237719e-02 -5.81441700e-01 -1.38664007e-01 6.25430495e-02
9.57867622e-01 -2.43207384e-02 7.62461573e-02 -3.14450622e-01
2.68739372e-01 1.71643034e-01 4.96926099e-01 7.32157171e-01
-1.56689033e-01 6.97835624e-01 2.85816640e-01 -6.22905731e-01
-1.22728467e+00 -9.38390493e-01 -1.94972172e-01 1.27524495e+00
3.03421199e-01 -2.87020504e-01 -6.90136552e-01 -3.83677274e-01
2.95648336e-01 3.07980061e-01 -4.08359110e-01 4.17582422e-01
-6.92419350e-01 -4.74863499e-01 3.21062654e-01 9.30007875e-01
3.80850077e-01 -1.03700578e+00 -8.44952524e-01 1.14414498e-01
2.38588288e-01 -1.30021572e+00 -6.63998067e-01 1.41781107e-01
-1.13060653e+00 -7.51447678e-01 -6.01515591e-01 -8.97612572e-01
8.25493693e-01 5.04749417e-01 1.07968664e+00 2.01783076e-01
-3.97164345e-01 2.95916528e-01 -2.51182854e-01 -6.50284529e-01
5.52652419e-01 4.52336073e-01 -1.98961303e-01 -4.26973961e-02
4.72092152e-01 -7.05173552e-01 -7.45285332e-01 6.04689643e-02
-7.14198411e-01 1.13382101e-01 7.05443501e-01 5.54411650e-01
1.09758162e+00 -3.58347893e-01 2.12875620e-01 -7.49394715e-01
1.68806344e-01 -3.95073444e-01 -9.06463802e-01 -1.06487423e-02
-1.56182259e-01 -1.04102544e-01 5.00011444e-01 -2.06915408e-01
-6.06910110e-01 4.33995396e-01 -2.42920443e-01 -9.55885828e-01
-2.75071174e-01 1.04283869e-01 -4.00647931e-02 -5.11003852e-01
3.77383120e-02 1.96395770e-01 -3.04416150e-01 -6.14687979e-01
4.38853472e-01 5.39518952e-01 3.67367685e-01 -5.99259198e-01
5.76902986e-01 6.54375196e-01 3.54741998e-02 -1.00144815e+00
-7.26305604e-01 -6.23061419e-01 -9.38577712e-01 8.61501042e-03
9.65573728e-01 -7.94316530e-01 -1.05212927e+00 5.13479292e-01
-1.45807433e+00 -3.87596279e-01 -3.31423283e-01 1.37205333e-01
-4.48870242e-01 3.07168007e-01 -6.27095997e-01 -7.98437715e-01
-5.96316218e-01 -1.28557956e+00 1.65510225e+00 4.14392427e-02
2.27564275e-01 -5.98388493e-01 -2.64042050e-01 2.86737114e-01
4.82928932e-01 1.34907126e-01 8.32646012e-01 -5.97877562e-01
-1.27511358e+00 -7.38199726e-02 -7.19319940e-01 -1.62728325e-01
-1.72364861e-01 -7.24176541e-02 -1.01155210e+00 -2.77342379e-01
-1.35242060e-01 -1.67957470e-01 8.65193009e-01 6.40154958e-01
1.90622342e+00 -1.51351660e-01 -4.18961197e-01 1.04855239e+00
1.51614368e+00 6.94925571e-03 1.93473592e-01 -1.40196219e-01
1.20744073e+00 3.04181993e-01 3.44706923e-01 5.46439230e-01
5.79928577e-01 5.62593400e-01 6.50071859e-01 -3.57164472e-01
-3.89741920e-02 8.36023763e-02 -2.38935381e-01 1.08325112e+00
-1.00086167e-01 -1.24761149e-01 -1.29884350e+00 9.38484907e-01
-1.80836594e+00 -6.01160645e-01 -2.85830915e-01 2.02229905e+00
4.30806011e-01 6.19017072e-02 -7.30460435e-02 -7.11510256e-02
5.90479732e-01 9.73684937e-02 -7.63534606e-01 -4.55347985e-01
1.97518766e-01 5.71899772e-01 8.39330554e-01 7.37439871e-01
-1.26813281e+00 1.33162439e+00 5.63056374e+00 8.18554044e-01
-1.23184383e+00 2.68733531e-01 6.18169546e-01 -2.92463988e-01
-2.33711362e-01 -1.94599718e-01 -6.94083154e-01 2.72189349e-01
5.82252562e-01 3.34158652e-02 3.11035991e-01 1.08225060e+00
-8.14125594e-03 1.38822362e-01 -9.42056596e-01 1.10258782e+00
3.52812484e-02 -1.31246543e+00 1.24001309e-01 4.88451831e-02
5.96484184e-01 6.35769904e-01 9.99110416e-02 5.69604076e-02
1.48860797e-01 -9.06378269e-01 7.47638702e-01 2.77648062e-01
6.98224783e-01 -7.96589792e-01 6.18116379e-01 2.51934439e-01
-1.75447798e+00 -2.84708068e-02 -5.75784028e-01 6.21611765e-03
1.21194713e-01 6.53297722e-01 -7.44727015e-01 3.07849109e-01
9.54635799e-01 5.02050459e-01 -6.05438113e-01 1.28948927e+00
2.93453544e-01 3.02828878e-01 -8.70657980e-01 -1.22591645e-01
4.78851378e-01 -1.49124742e-01 4.37287867e-01 1.29237008e+00
5.78949928e-01 3.53049695e-01 3.90262544e-01 8.46801281e-01
-1.19300388e-01 1.47985086e-01 -6.15683138e-01 5.35194576e-02
7.84388423e-01 1.21963799e+00 -1.15830290e+00 -5.27020633e-01
-4.15491074e-01 9.93529081e-01 5.54022431e-01 1.29389331e-01
-8.21893454e-01 -4.19952065e-01 7.68237591e-01 1.44697860e-01
7.21313417e-01 -5.66648781e-01 -8.80212903e-01 -8.02452803e-01
6.96340948e-02 -2.45072067e-01 1.85507149e-01 -7.15333104e-01
-1.08696139e+00 7.15985775e-01 -1.02050677e-01 -1.00231671e+00
4.28019732e-01 -6.05223894e-01 -5.60069561e-01 6.95457399e-01
-1.47058427e+00 -1.45772040e+00 -6.45260632e-01 6.36168599e-01
8.13489437e-01 2.87016153e-01 5.12004197e-01 4.32307512e-01
-4.69608754e-01 5.25060296e-01 -2.67463416e-01 3.99088748e-02
2.51622736e-01 -1.42036581e+00 9.11738276e-01 7.19225407e-01
1.56988427e-01 5.17590404e-01 2.11371124e-01 -5.90291440e-01
-1.49938595e+00 -1.27708423e+00 9.15691376e-01 -3.75483096e-01
4.46215570e-01 -6.30009055e-01 -1.02912641e+00 7.14123011e-01
9.79694799e-02 3.15768242e-01 4.08118457e-01 -2.77900286e-02
-2.17730239e-01 -6.24865592e-02 -1.12439156e+00 4.36215103e-01
1.27340794e+00 -4.64257926e-01 -3.20511907e-01 4.87934142e-01
1.10211396e+00 -7.03818023e-01 -7.68133521e-01 3.77643287e-01
2.36390322e-01 -7.11596012e-01 1.05743825e+00 -2.77383149e-01
1.05791435e-01 -4.52809781e-01 -2.34266698e-01 -8.44176829e-01
-6.52019024e-01 -3.02601904e-01 -5.47689665e-03 8.99319947e-01
2.85768241e-01 -5.79426527e-01 1.08508253e+00 5.81674039e-01
-6.89188302e-01 -1.07482457e+00 -9.63421285e-01 -4.99038607e-01
2.49320101e-02 -6.94024503e-01 8.41214120e-01 8.06686580e-01
-4.84747648e-01 1.29553959e-01 2.14746460e-01 6.27498627e-01
8.49901080e-01 6.30604208e-01 6.40546918e-01 -1.27084482e+00
8.62249881e-02 -6.05663598e-01 -4.61550593e-01 -1.41820359e+00
-4.36244495e-02 -1.09868956e+00 -4.95532295e-03 -1.65516818e+00
2.91609820e-02 -7.27241457e-01 -9.45584103e-02 7.69782543e-01
5.48581742e-02 4.81670976e-01 3.98560882e-01 2.17465222e-01
-7.44378984e-01 6.32405162e-01 1.17370284e+00 -1.49064064e-01
-2.33687848e-01 -8.35520551e-02 -4.29907084e-01 8.14132869e-01
8.50314498e-01 -4.69786644e-01 -2.83663392e-01 -9.64889467e-01
-9.05558690e-02 -2.04608038e-01 3.57364357e-01 -1.16129601e+00
7.00270593e-01 -8.85055810e-02 3.52798581e-01 -1.25418925e+00
6.54981017e-01 -9.48380351e-01 -1.39309794e-01 2.05049992e-01
-2.15720367e-02 4.65715647e-01 3.86635751e-01 3.36775929e-01
-1.07208751e-01 4.17768545e-02 7.86893547e-01 -2.05704033e-01
-6.33010447e-01 9.18944001e-01 8.22203085e-02 -1.22185335e-01
1.00088727e+00 -3.36135358e-01 -6.00839555e-02 7.34629929e-02
-6.84737265e-01 4.27878052e-01 5.08076668e-01 3.10850888e-01
8.41458559e-01 -1.24851918e+00 -5.24632931e-01 3.37760389e-01
-1.33774281e-01 7.85635650e-01 5.09010077e-01 8.29171360e-01
-8.44289601e-01 5.99029958e-01 -9.67187136e-02 -1.03948283e+00
-1.13321078e+00 5.74962854e-01 8.02503973e-02 6.38038814e-02
-7.38063991e-01 1.21965408e+00 3.08652699e-01 -4.39732254e-01
1.57479852e-01 -8.13747585e-01 1.92796290e-01 -1.17958616e-02
2.24496916e-01 2.49171853e-01 3.60734612e-01 -5.25257289e-01
-5.47814608e-01 1.11141503e+00 -8.03786814e-02 2.69042104e-01
1.44682801e+00 8.56591910e-02 -2.69768834e-01 3.35241973e-01
1.16275597e+00 -1.31076828e-01 -1.25924289e+00 -1.82422593e-01
-1.04663558e-01 -6.50393546e-01 2.66019583e-01 -2.79721648e-01
-1.26418400e+00 1.05502868e+00 5.53877652e-01 3.15794855e-01
1.02795672e+00 1.84405714e-01 9.68532622e-01 3.48261505e-01
3.98515701e-01 -7.94245660e-01 -2.09901854e-01 7.17469096e-01
9.67133582e-01 -1.24381089e+00 -2.98017599e-02 -7.05033779e-01
-3.96113843e-01 7.18373358e-01 7.72399902e-01 -4.52222466e-01
8.42951894e-01 5.65427601e-01 -2.02372313e-01 -4.30740416e-01
-7.10512578e-01 -3.33464980e-01 2.31987372e-01 4.22114313e-01
2.57285029e-01 1.50133982e-01 1.07169688e-01 3.50261867e-01
-1.91968694e-01 -2.34216094e-01 1.06060699e-01 8.88768911e-01
-6.21422231e-01 -7.76580036e-01 -3.46526563e-01 5.57390332e-01
-2.44633555e-01 -2.53417403e-01 -2.61921912e-01 4.87124681e-01
1.24000371e-01 4.78322417e-01 5.96103847e-01 -1.37871787e-01
4.03530747e-01 6.93139061e-03 3.82623941e-01 -7.11801291e-01
-6.44869208e-01 1.39570147e-01 -4.37691420e-01 -8.48140180e-01
-1.71588361e-01 -6.39567614e-01 -1.45265591e+00 -4.17980671e-01
-2.04096124e-01 -1.04200326e-01 7.22692549e-01 6.64808810e-01
9.07988667e-01 5.34916103e-01 4.24539447e-01 -1.53286254e+00
-2.76924729e-01 -8.22079659e-01 -2.19244748e-01 1.55364186e-01
3.33116084e-01 -6.36624873e-01 -4.66632880e-02 -1.41813368e-01] | [7.957892417907715, -3.480340003967285] |
c58a1655-5fd4-4639-8388-cbaffa4b900c | model-based-deep-autoencoder-networks-for | 2104.08409 | null | https://arxiv.org/abs/2104.08409v1 | https://arxiv.org/pdf/2104.08409v1.pdf | Model-Based Deep Autoencoder Networks for Nonlinear Hyperspectral Unmixing | Autoencoder (AEC) networks have recently emerged as a promising approach to perform unsupervised hyperspectral unmixing (HU) by associating the latent representations with the abundances, the decoder with the mixing model and the encoder with its inverse. AECs are especially appealing for nonlinear HU since they lead to unsupervised and model-free algorithms. However, existing approaches fail to explore the fact that the encoder should invert the mixing process, which might reduce their robustness. In this paper, we propose a model-based AEC for nonlinear HU by considering the mixing model a nonlinear fluctuation over a linear mixture. Differently from previous works, we show that this restriction naturally imposes a particular structure to both the encoder and to the decoder networks. This introduces prior information in the AEC without reducing the flexibility of the mixing model. Simulations with synthetic and real data indicate that the proposed strategy improves nonlinear HU. | ['Deniz Erdoğmuş', 'José Carlos Moreira Bermudez', 'Pau Closas', 'Tales Imbiriba', 'Ricardo Augusto Borsoi', 'Haoqing Li'] | 2021-04-17 | null | null | null | null | ['hyperspectral-unmixing'] | ['computer-vision'] | [ 5.80467820e-01 7.75136277e-02 2.11475626e-01 5.26024494e-03
4.55139689e-02 -4.45896327e-01 8.82034123e-01 -2.82256961e-01
-2.50012010e-01 5.47285795e-01 3.43112081e-01 -7.41780549e-02
-3.64865184e-01 -7.63118804e-01 -6.95558429e-01 -1.35765827e+00
2.12079853e-01 2.66136646e-01 -3.39600533e-01 -1.39384672e-01
-2.19504148e-01 4.04946804e-01 -1.64822793e+00 -1.32383369e-02
1.11560214e+00 6.55227542e-01 2.48725146e-01 4.82832581e-01
1.14321038e-01 8.08403254e-01 -3.81930172e-01 -2.39174023e-01
5.29948592e-01 -8.92662585e-01 -2.12431014e-01 4.68599319e-01
2.81779245e-02 -2.05996469e-01 -5.34488201e-01 1.36782622e+00
4.03162420e-01 2.53392488e-01 9.08957601e-01 -1.10656166e+00
-4.50751036e-01 9.48154628e-01 -5.46831787e-01 -3.16113502e-01
-1.93146467e-01 1.29180118e-01 8.28582942e-01 -4.79179263e-01
4.41284269e-01 7.84394562e-01 5.46937585e-01 4.23115045e-01
-1.36978364e+00 -4.15197402e-01 -1.63769305e-01 1.58980027e-01
-1.48350334e+00 -5.03813326e-01 1.01360762e+00 -6.36276603e-01
7.07078457e-01 2.14512289e-01 8.18908095e-01 9.36062634e-01
-5.18205762e-02 4.77656126e-01 1.16145575e+00 -5.76195359e-01
3.74159873e-01 1.71749275e-02 -1.08211391e-01 4.85933423e-01
4.29604858e-01 1.69725910e-01 -3.15018088e-01 -8.95292088e-02
5.63368559e-01 -6.53334856e-02 -5.43407619e-01 -6.07236266e-01
-9.31553662e-01 9.00464773e-01 4.31184381e-01 3.54104578e-01
-7.10424185e-01 -1.18288159e-01 -9.15960073e-02 1.78791642e-01
2.74346083e-01 4.54317570e-01 -1.76010076e-02 3.69738102e-01
-1.15948379e+00 -1.90811291e-01 8.58786166e-01 7.54139543e-01
9.09065247e-01 4.98527855e-01 1.66662455e-01 7.03941286e-01
4.68942761e-01 5.47558188e-01 6.27456605e-01 -8.97673905e-01
1.82516947e-01 4.55793262e-01 -1.08682588e-01 -7.96338737e-01
-2.80997992e-01 -8.26559365e-01 -1.29658139e+00 1.94152132e-01
2.81786561e-01 -2.93049991e-01 -8.28585565e-01 1.85308671e+00
5.62016815e-02 3.09087455e-01 4.48258221e-01 7.38368571e-01
3.58321667e-01 9.05853271e-01 -1.79164395e-01 -5.44694841e-01
8.83268893e-01 -1.03945351e+00 -9.87164617e-01 -1.57336622e-01
3.64976704e-01 -6.39826715e-01 4.91000205e-01 3.91179949e-01
-9.71644938e-01 -2.71317065e-01 -1.32250381e+00 1.28809795e-01
-3.06553990e-01 2.18273014e-01 3.53527784e-01 6.76849246e-01
-9.55481827e-01 8.24926436e-01 -8.98853242e-01 -3.75843138e-01
-1.04952618e-01 3.97044808e-01 -2.68813372e-01 1.52738154e-01
-1.14211702e+00 9.35364783e-01 6.54711068e-01 4.57244366e-01
-6.83418512e-01 -1.84342235e-01 -8.96736681e-01 2.52370745e-01
2.14620203e-01 -6.93909526e-01 8.61583292e-01 -1.10295713e+00
-1.95945132e+00 3.15937281e-01 -5.07445075e-02 -6.64705515e-01
6.23084366e-01 5.72993122e-02 -4.77276534e-01 2.50529706e-01
-3.43911111e-01 6.20947838e-01 1.13199186e+00 -1.36807120e+00
-1.51179105e-01 7.67715368e-03 -1.39442831e-01 3.25153202e-01
-4.95832056e-01 -5.31437993e-01 -1.25157967e-01 -6.17138267e-01
5.13432741e-01 -9.15002286e-01 -1.44051358e-01 -2.30078459e-01
-4.17446613e-01 3.23346406e-01 7.87693977e-01 -7.50461936e-01
1.13806677e+00 -2.19150996e+00 5.31477928e-01 3.52194428e-01
9.23119672e-03 1.45929053e-01 -7.30022639e-02 4.95697230e-01
-4.85638857e-01 -2.01229110e-01 -7.40555346e-01 -4.36791897e-01
5.39041087e-02 5.11688471e-01 -1.54056966e-01 7.75908589e-01
1.48358151e-01 6.13121271e-01 -6.89692795e-01 -5.12596332e-02
4.50429112e-01 7.37342417e-01 -6.19330883e-01 2.56769478e-01
-2.14862913e-01 5.35378873e-01 1.56472519e-01 5.81358448e-02
8.21097910e-01 -2.48008087e-01 5.57300746e-01 -3.87374341e-01
-3.90978396e-01 1.20384283e-01 -1.47642922e+00 1.36321807e+00
-1.04049109e-01 5.01181006e-01 3.29779983e-01 -1.26302230e+00
7.41773129e-01 6.05929136e-01 6.07240260e-01 -1.99351206e-01
1.31717071e-01 2.45409772e-01 3.56767058e-01 -3.95262688e-01
2.84087777e-01 -6.01294398e-01 4.76132423e-01 4.27959263e-01
3.54427665e-01 -1.75347924e-01 3.20035666e-01 -5.88312894e-02
6.10282063e-01 1.29417360e-01 6.04584336e-01 -5.08811653e-01
7.17175961e-01 -2.65178710e-01 4.14069831e-01 6.24694586e-01
1.57855228e-01 6.69449627e-01 3.00608873e-01 -5.03144860e-02
-1.17879081e+00 -8.80185127e-01 -1.17145687e-01 2.87926346e-01
-1.08210333e-01 -1.55500561e-01 -9.41610992e-01 -5.82140908e-02
-4.42739993e-01 6.62566185e-01 -3.70988727e-01 -2.69062936e-01
-1.90298900e-01 -1.02112019e+00 3.79609287e-01 3.09608094e-02
6.16596162e-01 -6.69396877e-01 -2.48633996e-01 3.31996024e-01
-4.61949497e-01 -1.04590094e+00 -2.05162942e-01 7.17393160e-01
-8.53391767e-01 -8.33238065e-01 -7.48735726e-01 -4.27612901e-01
7.11088538e-01 1.30694181e-01 5.06589651e-01 -3.21586013e-01
2.21071884e-01 9.35044959e-02 -1.65053099e-01 -2.09268481e-01
-8.11191380e-01 5.60649931e-02 2.66904235e-01 4.55652028e-01
2.22827926e-01 -8.49992394e-01 -3.87722611e-01 -9.38144047e-03
-1.40059268e+00 4.15654331e-01 5.13083577e-01 7.29266763e-01
2.79614866e-01 6.15169227e-01 2.62335956e-01 -7.42487311e-01
2.03927100e-01 -5.74486375e-01 -7.38670886e-01 2.05203325e-01
-7.09818482e-01 4.43879873e-01 7.75451660e-01 -4.58744556e-01
-1.18334126e+00 4.37063485e-01 -1.60525307e-01 -5.11434019e-01
-1.64564088e-01 5.96441627e-01 -5.10340989e-01 -7.88965225e-02
6.19736016e-01 4.08120483e-01 1.97890133e-01 -5.86832583e-01
3.08935881e-01 5.80670834e-01 6.50376558e-01 -3.08459371e-01
1.20181346e+00 5.50298393e-01 1.07920051e-01 -1.13750732e+00
-5.98305643e-01 -4.13210124e-01 -7.62446105e-01 -1.89523831e-01
9.72680509e-01 -9.99962091e-01 -3.30030471e-01 5.88578641e-01
-9.97774959e-01 -1.27507448e-01 -4.49924648e-01 7.23059237e-01
-5.08388281e-01 5.71918845e-01 -6.17008328e-01 -8.61923397e-01
-1.07643150e-01 -1.21578550e+00 3.68104845e-01 4.04301792e-01
4.18309048e-02 -1.07383811e+00 -1.29864216e-02 1.79390118e-01
5.53775012e-01 2.36358792e-01 9.15556252e-01 -3.71901989e-01
-4.35201943e-01 -3.75372767e-02 9.84835923e-02 6.13559902e-01
3.00929099e-01 1.25632972e-01 -1.30236161e+00 -2.33778253e-01
5.38879812e-01 2.35590950e-01 1.07069409e+00 4.58502948e-01
8.08579326e-01 -4.71105546e-01 7.78670236e-02 8.98747504e-01
1.61921251e+00 2.64488429e-01 6.55965149e-01 8.28841180e-02
6.77153528e-01 5.82380235e-01 -5.40021598e-01 5.08768439e-01
-8.15252811e-02 3.80835205e-01 4.30100262e-01 -1.85607851e-01
3.50740179e-02 -1.85294241e-01 5.82952559e-01 1.37654865e+00
-3.73806685e-01 -5.62157929e-01 -6.50814414e-01 1.34635359e-01
-1.69552922e+00 -1.04620755e+00 -2.71907836e-01 2.21537542e+00
6.82285666e-01 -2.00237364e-01 -3.56591374e-01 4.66909617e-01
7.05922544e-01 4.19839472e-01 -3.53351057e-01 -3.79282273e-02
-6.83953106e-01 1.92236662e-01 4.90604937e-01 7.62762129e-01
-1.18786216e+00 6.17842972e-01 6.44442415e+00 4.97158736e-01
-1.31561041e+00 1.72407568e-01 6.57906309e-02 8.61525610e-02
-4.67075884e-01 1.00610085e-01 -3.31284285e-01 2.96493083e-01
8.50085020e-01 1.14438504e-01 8.23266625e-01 4.37864423e-01
1.88849002e-01 -9.68999416e-02 -7.91455746e-01 7.08989441e-01
1.52219310e-01 -8.87694240e-01 2.44823709e-01 2.52234876e-01
9.43186939e-01 -8.96027759e-02 -6.66808784e-02 -7.64773265e-02
1.30501375e-01 -6.58527136e-01 6.50362372e-01 8.04671466e-01
3.55671436e-01 -5.44974089e-01 6.84358656e-01 4.86923516e-01
-9.26649332e-01 -1.07005864e-01 -4.67661321e-01 -2.56412715e-01
1.72919959e-01 7.97793269e-01 -3.84788841e-01 8.14324439e-01
2.06002042e-01 7.23560095e-01 -4.05277759e-01 8.35972071e-01
-3.34621310e-01 6.35011971e-01 -5.55240631e-01 3.71262372e-01
2.02705771e-01 -9.48782504e-01 7.47430921e-01 9.70990658e-01
6.58694923e-01 -5.05059883e-02 -1.91464961e-01 8.95779431e-01
1.04103766e-01 -4.10747007e-02 -7.97613859e-01 -4.77135211e-01
-3.70025337e-02 1.13723207e+00 -7.52100646e-01 -3.42160344e-01
-3.72180343e-01 1.08870769e+00 -7.35811442e-02 7.26508558e-01
-6.79597080e-01 2.97355857e-02 5.16190827e-01 -4.54746783e-02
4.12402689e-01 -3.39958102e-01 -6.36985302e-02 -1.49648833e+00
-2.05856889e-01 -7.94625878e-01 2.03717425e-01 -7.28252053e-01
-1.19728220e+00 4.52052057e-01 8.44700914e-03 -1.37263167e+00
-3.04039478e-01 -6.55167758e-01 -3.59688401e-01 8.66034091e-01
-1.61228514e+00 -9.28391337e-01 -2.05005601e-01 4.69981223e-01
9.22878087e-02 -4.07595523e-02 8.35174382e-01 3.05524230e-01
-7.19683647e-01 7.46472254e-02 6.28001094e-01 -2.16658399e-01
5.08837581e-01 -1.27271068e+00 -3.43111694e-01 1.28234410e+00
2.27674142e-01 6.98636055e-01 9.44838524e-01 -5.91924906e-01
-1.20552230e+00 -9.22578871e-01 6.39838815e-01 9.14624110e-02
6.94738865e-01 -2.84416676e-01 -9.58134055e-01 5.67944169e-01
6.12900972e-01 -2.43727922e-01 8.55004549e-01 -3.49796712e-01
-1.74267784e-01 -1.56404957e-01 -7.50538111e-01 5.72701097e-01
7.72615314e-01 -6.57869995e-01 -2.87354708e-01 1.18010469e-01
4.56660897e-01 2.34844107e-02 -8.00640762e-01 4.12350714e-01
5.33659816e-01 -9.95200574e-01 7.63677835e-01 -7.92066753e-02
5.56961894e-01 -6.73731208e-01 -2.77415544e-01 -1.47563744e+00
-4.36776429e-01 -4.73092914e-01 -4.47875887e-01 1.08218992e+00
3.69994134e-01 -6.58385038e-01 3.98270667e-01 2.85065114e-01
-2.27372512e-01 -1.02470905e-01 -7.00044930e-01 -7.43975163e-01
-6.20108703e-03 -2.04618290e-01 5.37293553e-01 1.05595481e+00
-1.73230663e-01 2.36983359e-01 -7.41977751e-01 6.37112200e-01
7.32840955e-01 -2.00003125e-02 2.95474797e-01 -1.34797180e+00
-6.68443143e-01 -5.19854546e-01 -4.13127150e-03 -9.54242766e-01
1.95813686e-01 -9.70499694e-01 2.85720497e-01 -1.37320399e+00
-3.20250131e-02 1.43416017e-01 -3.23511004e-01 2.85803735e-01
3.07323448e-02 1.70093685e-01 6.10898435e-02 3.81107539e-01
1.67925656e-01 8.17108750e-01 9.62811887e-01 -2.89271861e-01
-1.60175279e-01 -5.25598638e-02 -4.27469134e-01 7.47526526e-01
9.29053724e-01 -4.31083083e-01 -5.69305420e-01 -3.86787087e-01
2.28134722e-01 -4.84144427e-02 8.23707134e-02 -1.29069114e+00
4.97054040e-01 5.04218414e-02 3.55758339e-01 -2.47074232e-01
4.55924809e-01 -1.07492542e+00 7.47063875e-01 5.88325858e-01
-9.79517326e-02 -3.39985490e-01 -5.50101064e-02 5.33953786e-01
-3.13785076e-01 -7.04263628e-01 9.11543965e-01 -9.59343091e-02
-2.96628118e-01 1.19538411e-01 -8.00209045e-01 -6.61926806e-01
5.88083208e-01 -2.70240493e-02 -8.97691548e-02 -5.92182755e-01
-8.83192956e-01 -1.16027236e-01 5.20832300e-01 -7.66075104e-02
1.07553951e-01 -1.13634193e+00 -4.85663563e-01 5.45223832e-01
-1.18978195e-01 -5.56471385e-02 3.17413598e-01 1.05270112e+00
-5.86313248e-01 2.65434444e-01 -1.66370809e-01 -4.26250279e-01
-7.83190072e-01 6.17155671e-01 6.92391038e-01 -1.99950829e-01
-4.55493718e-01 3.95114124e-01 2.08697870e-01 -5.89095712e-01
6.28832728e-02 -1.07471660e-01 -2.83892900e-01 2.07199708e-01
2.30208665e-01 2.50088662e-01 5.06539224e-03 -8.71398747e-01
1.17111601e-01 4.05798078e-01 4.88545835e-01 -2.93716103e-01
1.38089001e+00 -3.04707557e-01 -3.99331421e-01 6.18265390e-01
1.05763316e+00 -5.90194426e-02 -1.28206062e+00 -2.83902019e-01
-1.40955644e-02 1.39896274e-01 2.93687284e-01 -3.24218929e-01
-1.07075858e+00 9.95827556e-01 4.88697380e-01 4.30853605e-01
1.40432048e+00 -4.38979477e-01 2.88223863e-01 4.19187129e-01
-5.60622215e-02 -9.99274373e-01 -3.37664783e-01 4.51629370e-01
6.43788576e-01 -1.08169830e+00 1.18727819e-03 -3.71873558e-01
-3.20652455e-01 1.28793621e+00 2.48671398e-01 9.21937525e-02
7.75540352e-01 1.79048702e-01 8.49211067e-02 1.10954680e-02
-4.36119676e-01 -3.88876230e-01 2.05446288e-01 3.61999840e-01
3.29864234e-01 1.21532097e-01 -3.02769333e-01 3.00714970e-01
-2.11548701e-01 -2.24835336e-01 6.82669520e-01 6.06125891e-01
-3.64094079e-01 -1.10709798e+00 -4.14558411e-01 1.66621104e-01
-5.18490411e-02 -1.74077228e-01 -2.44132951e-01 6.08855128e-01
3.19406629e-01 9.39782441e-01 7.40465745e-02 -3.44242543e-01
-4.19876203e-02 6.58221602e-01 2.71922439e-01 -3.11621517e-01
-2.22751364e-01 6.06354654e-01 -2.39005134e-01 1.47835948e-02
-8.39922428e-01 -6.53489113e-01 -8.84886026e-01 -2.24565759e-01
-6.14889145e-01 3.86538148e-01 7.06580877e-01 1.01366234e+00
1.81368962e-01 4.57209557e-01 6.53934836e-01 -7.79775500e-01
-5.09656429e-01 -1.24517035e+00 -9.80036438e-01 2.73174673e-01
4.31848228e-01 -4.94928658e-01 -6.85269415e-01 3.40961784e-01] | [10.053913116455078, -2.093348503112793] |
de4eefdf-d31c-4899-aacf-a7b413529e30 | skinned-motion-retargeting-with-residual | 2303.08658 | null | https://arxiv.org/abs/2303.08658v1 | https://arxiv.org/pdf/2303.08658v1.pdf | Skinned Motion Retargeting with Residual Perception of Motion Semantics & Geometry | A good motion retargeting cannot be reached without reasonable consideration of source-target differences on both the skeleton and shape geometry levels. In this work, we propose a novel Residual RETargeting network (R2ET) structure, which relies on two neural modification modules, to adjust the source motions to fit the target skeletons and shapes progressively. In particular, a skeleton-aware module is introduced to preserve the source motion semantics. A shape-aware module is designed to perceive the geometries of target characters to reduce interpenetration and contact-missing. Driven by our explored distance-based losses that explicitly model the motion semantics and geometry, these two modules can learn residual motion modifications on the source motion to generate plausible retargeted motion in a single inference without post-processing. To balance these two modifications, we further present a balancing gate to conduct linear interpolation between them. Extensive experiments on the public dataset Mixamo demonstrate that our R2ET achieves the state-of-the-art performance, and provides a good balance between the preservation of motion semantics as well as the attenuation of interpenetration and contact-missing. Code is available at https://github.com/Kebii/R2ET. | ['Zhigang Tu', 'Jue Wang', 'Ying Shan', 'Linchao Bao', 'Xuefei Zhe', 'Shaoli Huang', 'Fang Zhao', 'Di Kang', 'Junwu Weng', 'Jiaxu Zhang'] | 2023-03-15 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Zhang_Skinned_Motion_Retargeting_With_Residual_Perception_of_Motion_Semantics__CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_Skinned_Motion_Retargeting_With_Residual_Perception_of_Motion_Semantics__CVPR_2023_paper.pdf | cvpr-2023-1 | ['motion-retargeting'] | ['computer-vision'] | [ 2.05473602e-01 7.35496953e-02 -2.77132601e-01 -3.50019604e-01
-5.50361335e-01 -4.75761592e-01 5.73963881e-01 -1.78438678e-01
-3.24623346e-01 3.85867834e-01 4.39944476e-01 3.41980159e-02
1.65949594e-02 -9.76389706e-01 -1.00832105e+00 -6.39912784e-01
1.97656110e-01 1.34715527e-01 7.52102673e-01 -4.79549050e-01
2.58838952e-01 4.05459553e-01 -1.40750945e+00 2.65525937e-01
1.11015272e+00 5.82547307e-01 1.25982255e-01 7.66601145e-01
1.26315849e-02 6.30226851e-01 -1.63369983e-01 -4.73536819e-01
3.42933089e-01 -3.60301852e-01 -4.27879244e-01 -2.06099361e-01
5.76949596e-01 -3.15832943e-01 -7.58595884e-01 1.07002854e+00
7.59824574e-01 3.82828385e-01 5.48381686e-01 -9.22671735e-01
-8.55119884e-01 8.92513454e-01 -8.97172451e-01 1.04001492e-01
1.77152604e-01 6.34694397e-01 8.99616838e-01 -9.02142823e-01
5.35225511e-01 1.52576184e+00 7.07924128e-01 6.47054613e-01
-1.26229739e+00 -6.53060794e-01 4.53298330e-01 1.41984954e-01
-1.56723320e+00 -3.43154907e-01 9.93450999e-01 -3.92521590e-01
5.68096101e-01 3.32292378e-01 7.59961367e-01 1.09978259e+00
3.13279241e-01 6.96718633e-01 9.69223529e-02 -5.09600192e-02
2.06046656e-01 -4.31426257e-01 -2.08721161e-01 6.58845246e-01
1.58816427e-01 2.39828482e-01 -6.07952356e-01 1.71945885e-01
1.11310446e+00 -3.96858789e-02 -2.49094874e-01 -5.33628166e-01
-1.40562236e+00 6.34467721e-01 7.32064903e-01 4.85367924e-02
-2.76112370e-02 5.98057628e-01 3.31845820e-01 -1.64335564e-01
4.05002475e-01 2.99288660e-01 -4.40973550e-01 -1.15011990e-01
-9.26474273e-01 4.97834861e-01 2.29680806e-01 1.06297600e+00
6.57792926e-01 1.00858040e-01 -5.28631926e-01 6.97804630e-01
6.14732563e-01 4.18334216e-01 4.82571304e-01 -1.20023739e+00
6.69532418e-01 6.68670833e-01 -8.84679481e-02 -1.35940039e+00
-3.99896741e-01 -4.39702004e-01 -1.01367617e+00 1.03698865e-01
3.51486415e-01 -1.61483102e-02 -9.43129122e-01 2.10171151e+00
5.46141744e-01 4.15609598e-01 -1.95380598e-01 1.10291016e+00
9.53161955e-01 6.19113088e-01 1.96331143e-01 2.59504974e-01
1.13850129e+00 -1.22956145e+00 -3.61210555e-01 -2.58944422e-01
6.48227811e-01 -6.41994238e-01 1.23373270e+00 -1.70986261e-02
-1.40926111e+00 -7.54987478e-01 -1.15362036e+00 -4.08005446e-01
6.02358878e-02 1.44897282e-01 4.22377169e-01 2.99268901e-01
-8.41615498e-01 9.96384382e-01 -1.10388911e+00 -9.78395119e-02
3.65168959e-01 2.21301153e-01 -9.69155803e-02 3.31533179e-02
-1.09894443e+00 4.03200150e-01 2.87170619e-01 3.47549051e-01
-8.50139499e-01 -9.08562541e-01 -1.08774817e+00 -9.91654000e-04
2.60850519e-01 -1.10622239e+00 1.05379462e+00 -9.17498291e-01
-1.66771102e+00 5.16478002e-01 2.02949736e-02 5.84918670e-02
1.00869107e+00 -2.05552638e-01 -1.82897568e-01 -5.69500625e-02
1.22247033e-01 1.15520585e+00 6.94692850e-01 -1.05927551e+00
-5.73756397e-01 -2.72051036e-01 -4.85579409e-02 4.05856788e-01
-1.27516329e-01 -3.90033722e-01 -1.01110780e+00 -1.35529637e+00
3.29460144e-01 -9.84229207e-01 -4.27146345e-01 4.28379714e-01
-5.13185441e-01 1.05542496e-01 5.84837615e-01 -6.67425156e-01
1.35852587e+00 -2.12403631e+00 7.03495324e-01 8.28044862e-03
1.76477224e-01 6.10409752e-02 -4.70993727e-01 1.35132611e-01
-6.78477734e-02 2.11362079e-01 -5.56690574e-01 -4.06598300e-01
-8.65862891e-02 1.86624378e-02 -2.09059104e-01 4.20540124e-01
3.39440078e-01 1.11407566e+00 -9.61913824e-01 -3.78966928e-01
2.08779827e-01 6.82176530e-01 -9.37965930e-01 7.42553249e-02
-3.52595121e-01 6.21105015e-01 -4.88809347e-01 4.34190989e-01
8.10514212e-01 -3.34474556e-02 -4.88804206e-02 -3.39704871e-01
-2.13784918e-01 2.02291921e-01 -1.26421797e+00 2.18562078e+00
3.03528234e-02 2.77270198e-01 -1.86457351e-01 -5.36310494e-01
8.42485368e-01 -6.44481629e-02 4.33614969e-01 -6.03576362e-01
-5.33774942e-02 8.87785330e-02 -3.77273113e-02 -3.35441411e-01
5.97450972e-01 1.65255576e-01 -7.02640638e-02 9.38634649e-02
-4.80497479e-01 -4.90688495e-02 -8.30456540e-02 1.45288512e-01
9.58135545e-01 6.91938043e-01 4.19878587e-02 -3.72571439e-01
4.82281983e-01 -6.35290071e-02 1.01077211e+00 3.34622622e-01
-8.87293220e-02 1.10281420e+00 2.45331064e-01 -5.20065129e-01
-1.06646502e+00 -1.24388397e+00 2.58000821e-01 1.18125522e+00
6.80583298e-01 -3.08639437e-01 -7.24448025e-01 -5.08507192e-01
-9.93280858e-02 5.90971053e-01 -7.23002076e-01 -6.10666931e-01
-9.54090834e-01 -7.74451911e-01 7.13072121e-01 7.40418911e-01
5.55049658e-01 -8.51190984e-01 -5.14784217e-01 1.47700801e-01
-2.73138911e-01 -7.09538519e-01 -1.18775451e+00 -2.96417236e-01
-9.22921956e-01 -7.17468083e-01 -1.00856006e+00 -7.00891554e-01
6.90807104e-01 3.14464688e-01 8.54469061e-01 2.65352190e-01
-1.08518884e-01 -2.65010715e-01 -4.04582441e-01 1.01425096e-01
-2.63699502e-01 2.78647959e-01 -1.75068110e-01 -1.52520612e-01
-1.96420848e-01 -7.67148614e-01 -1.00771999e+00 5.48869610e-01
-1.10381150e+00 5.92693090e-01 4.98480886e-01 4.48442191e-01
7.14511156e-01 -8.53325427e-02 2.49076977e-01 -3.93350422e-01
5.11573479e-02 -4.73727196e-01 -3.57928574e-01 7.39846975e-02
-3.53128254e-01 3.07237327e-01 4.71456587e-01 -7.15940535e-01
-1.01594865e+00 2.06644803e-01 -2.61244804e-01 -5.05965412e-01
4.41511646e-02 2.14183673e-01 -6.69172883e-01 1.28183290e-02
4.32850122e-01 1.96672574e-01 -4.90240693e-01 -5.50125599e-01
6.81885779e-01 5.37001230e-02 7.93252766e-01 -6.52023911e-01
9.52664852e-01 5.16152918e-01 -5.78657165e-02 -4.73011166e-01
-5.94710588e-01 -3.02862283e-02 -7.68749118e-01 -6.63828999e-02
1.01024532e+00 -9.71300423e-01 -4.23337400e-01 7.23190665e-01
-1.10769188e+00 -6.43448591e-01 -4.77060318e-01 3.65100294e-01
-5.55377364e-01 6.28585696e-01 -6.65923059e-01 -2.60990411e-01
-4.11466092e-01 -1.25515735e+00 1.06597388e+00 4.27473098e-01
-1.31481960e-01 -7.67243922e-01 3.34341466e-01 1.52712345e-01
2.16746792e-01 3.12722474e-01 8.20673704e-01 -9.42290425e-02
-6.60836399e-01 2.67417103e-01 -2.17065349e-01 -1.43367767e-01
1.83114335e-01 3.12794358e-01 -6.03330791e-01 -3.59558821e-01
-4.48470503e-01 6.53551444e-02 1.06203759e+00 5.55843472e-01
1.12160218e+00 -4.16405022e-01 -1.34551659e-01 1.30242169e+00
1.05696177e+00 4.58202846e-02 9.88044262e-01 3.38275343e-01
1.31360209e+00 6.49845362e-01 5.33124089e-01 6.27138972e-01
8.60811770e-01 9.10425246e-01 4.61290061e-01 -2.57208887e-02
-5.57835460e-01 -7.40607381e-01 5.14991283e-01 8.89331877e-01
-1.54612914e-01 -3.08158129e-01 -7.66226828e-01 5.49605787e-01
-2.15649104e+00 -9.41228271e-01 -2.09602207e-01 2.06863904e+00
1.01058233e+00 1.53418094e-01 1.85377985e-01 -1.84777111e-01
8.68672907e-01 2.96982974e-01 -9.20305073e-01 -7.78940171e-02
-1.78870887e-01 -1.99641407e-01 3.78782690e-01 6.79567754e-01
-9.53238308e-01 1.28452408e+00 5.52643013e+00 1.02158856e+00
-1.03459239e+00 -6.75101653e-02 7.05578208e-01 -2.19014451e-01
-8.31368029e-01 8.13262984e-02 -7.99244404e-01 6.52436018e-01
4.49233383e-01 -8.26504361e-03 4.25718397e-01 5.66302538e-01
6.03824854e-01 2.76246339e-01 -9.31473970e-01 8.72468948e-01
-2.33429968e-01 -1.40940237e+00 4.63792533e-01 -1.71284989e-01
7.63854980e-01 -2.70248681e-01 -1.85573362e-02 1.83583051e-01
3.07504863e-01 -1.02644730e+00 1.29520547e+00 8.10617149e-01
8.50122392e-01 -8.81979167e-01 2.49843881e-01 1.80106416e-01
-1.79333353e+00 1.68511480e-01 -2.81010002e-01 3.86623256e-02
2.62735933e-01 4.57892209e-01 -1.12685114e-01 5.24637282e-01
6.33744717e-01 9.67513800e-01 -5.50195277e-01 9.13043261e-01
-3.96206707e-01 2.88832903e-01 -1.28437594e-01 3.86012673e-01
3.08732484e-02 -2.57210821e-01 7.72607327e-01 1.20335817e+00
3.80091935e-01 1.08547121e-01 1.03088044e-01 1.17570627e+00
9.47848428e-03 -3.57900970e-02 -1.56172201e-01 2.80230880e-01
6.68804705e-01 1.01856089e+00 -6.66951954e-01 -1.62812978e-01
-4.15514968e-02 1.21075821e+00 3.18801820e-01 4.18126017e-01
-1.25760746e+00 -3.49199265e-01 8.88597310e-01 2.95962930e-01
3.97148728e-01 -3.06249619e-01 -5.24184704e-01 -1.31730711e+00
8.04653242e-02 -7.44737446e-01 2.63085306e-01 -5.70068479e-01
-1.07256675e+00 3.24675947e-01 4.40038256e-02 -1.26648664e+00
1.21140525e-01 -1.62365600e-01 -9.46416438e-01 6.29334748e-01
-1.20438063e+00 -1.37429011e+00 -4.85340714e-01 4.76534188e-01
6.88345253e-01 1.67179048e-01 8.74083489e-02 3.81928325e-01
-9.34989214e-01 8.76815200e-01 -3.00434470e-01 4.08408158e-02
7.33764052e-01 -9.82690275e-01 8.88310730e-01 1.12159574e+00
-2.15552002e-01 4.69163895e-01 6.12876475e-01 -9.63908792e-01
-1.27250385e+00 -1.49588060e+00 4.32074904e-01 -3.91862899e-01
3.52266878e-01 -9.07503068e-02 -9.66572404e-01 6.19884491e-01
-3.17631513e-01 -5.45669161e-03 3.66824061e-01 -5.36428273e-01
-4.39702243e-01 3.32870632e-02 -9.29119408e-01 1.12420428e+00
1.59453416e+00 -3.83150615e-02 -4.16744858e-01 -2.25318223e-01
1.23147190e+00 -5.40254056e-01 -8.28213274e-01 6.94532931e-01
5.29364645e-01 -9.00809467e-01 1.21233106e+00 -4.09273237e-01
7.01451898e-01 -6.99355185e-01 -1.79137200e-01 -1.21300185e+00
-6.70753896e-01 -7.97491431e-01 -3.30114514e-01 1.27536345e+00
2.55655527e-01 -1.64364249e-01 9.27147925e-01 5.91839373e-01
-3.89602751e-01 -8.33485484e-01 -8.92580628e-01 -5.75934231e-01
2.36726850e-01 -5.34389138e-01 7.82360494e-01 8.39019537e-01
-2.93748200e-01 2.42152542e-01 -4.34389919e-01 1.38802350e-01
5.83515882e-01 -5.98005913e-02 7.77395070e-01 -7.42154777e-01
-5.14676511e-01 -7.35787034e-01 -2.66161561e-01 -1.64705992e+00
-1.23083621e-01 -8.93762052e-01 1.70612589e-01 -1.45171022e+00
2.99675554e-01 -4.28690463e-01 4.43027802e-02 5.55801630e-01
-4.87736285e-01 3.17644477e-01 2.90071249e-01 3.76568675e-01
-3.99708718e-01 1.18765306e+00 1.63961768e+00 -1.26163572e-01
-7.04399288e-01 -1.29261851e-01 -8.98979306e-01 1.04764462e+00
6.98080719e-01 -4.12716657e-01 -5.02271175e-01 -8.89310420e-01
2.74364233e-01 -1.18707560e-01 4.46245492e-01 -1.05116987e+00
1.96313471e-01 -4.94511664e-01 3.68042976e-01 -5.71731925e-01
2.60064155e-01 -3.90338480e-01 2.86803275e-01 6.10079587e-01
-5.02211869e-01 1.74319133e-01 1.81074694e-01 7.63112605e-01
2.16484025e-01 1.48355424e-01 1.01236272e+00 9.18662027e-02
-7.70074129e-01 6.25787556e-01 -6.22350648e-02 1.21726699e-01
9.47565496e-01 -4.36105371e-01 -2.75170594e-01 -2.18819499e-01
-5.32140851e-01 3.99383903e-01 7.49797583e-01 6.33351803e-01
8.18196595e-01 -1.60931218e+00 -7.61751711e-01 5.05868010e-02
-1.27455769e-02 5.36218286e-01 6.03917480e-01 6.99081779e-01
-8.43803763e-01 -9.15987119e-02 -1.61927015e-01 -5.28781474e-01
-9.82621551e-01 3.61867964e-01 3.20960432e-01 -1.28487855e-01
-8.12449872e-01 1.01473570e+00 6.18230402e-01 -5.52330732e-01
2.10727587e-01 -5.49107254e-01 -3.10379807e-02 -3.46641898e-01
5.96052587e-01 6.24654710e-01 -4.26868498e-01 -7.41624296e-01
-4.20502990e-01 9.82240140e-01 8.40467811e-02 -1.68150410e-01
1.11923957e+00 -3.38187873e-01 1.58636987e-01 1.28793925e-01
9.23971355e-01 1.54535115e-01 -2.01658225e+00 3.82329486e-02
-2.30757236e-01 -5.69340944e-01 -1.19676977e-01 -4.17065263e-01
-1.28163314e+00 7.29457974e-01 4.09421951e-01 -5.00975132e-01
1.10840535e+00 -1.32929623e-01 1.08767903e+00 3.87904495e-02
1.01470500e-01 -8.65079284e-01 3.75059336e-01 6.40523911e-01
1.00576055e+00 -8.08355391e-01 -8.14591721e-02 -6.55001879e-01
-6.60786688e-01 9.11708355e-01 9.83621180e-01 -2.47580960e-01
4.57123011e-01 1.21558927e-01 -1.12217851e-01 7.12954178e-02
-6.07197523e-01 4.27000336e-02 6.38634562e-01 5.00214815e-01
4.96540070e-01 -7.49686658e-02 -2.61970282e-01 7.43624866e-01
-2.87712395e-01 -2.64475524e-01 3.50964040e-01 6.98022008e-01
-4.56184626e-01 -9.57064211e-01 -3.05616796e-01 -7.61143640e-02
-1.24435820e-01 -1.39882281e-01 -3.65406781e-01 5.66537678e-01
4.04252470e-01 6.82388425e-01 1.06821403e-01 -7.14744449e-01
4.92816746e-01 -5.94982207e-01 3.16379815e-01 -6.09384000e-01
-4.27274406e-01 2.15033621e-01 -2.64737517e-01 -9.09391284e-01
-1.82359695e-01 -3.53099406e-01 -1.51087749e+00 -6.08839810e-01
6.50150850e-02 -3.49811018e-01 1.61290854e-01 6.05904162e-01
6.34661615e-01 7.92489707e-01 4.79401290e-01 -1.25648844e+00
-4.33344930e-01 -7.36546576e-01 -2.60302097e-01 5.38379788e-01
1.02856293e-01 -5.37813842e-01 -1.97853550e-01 1.83095723e-01] | [7.390151500701904, -0.45035308599472046] |
cc8e8037-62f6-4913-ad2e-5009b507881d | interpreting-and-generalizing-deep-learning | 2307.04569 | null | https://arxiv.org/abs/2307.04569v1 | https://arxiv.org/pdf/2307.04569v1.pdf | Interpreting and generalizing deep learning in physics-based problems with functional linear models | Although deep learning has achieved remarkable success in various scientific machine learning applications, its black-box nature poses concerns regarding interpretability and generalization capabilities beyond the training data. Interpretability is crucial and often desired in modeling physical systems. Moreover, acquiring extensive datasets that encompass the entire range of input features is challenging in many physics-based learning tasks, leading to increased errors when encountering out-of-distribution (OOD) data. In this work, motivated by the field of functional data analysis (FDA), we propose generalized functional linear models as an interpretable surrogate for a trained deep learning model. We demonstrate that our model could be trained either based on a trained neural network (post-hoc interpretation) or directly from training data (interpretable operator learning). A library of generalized functional linear models with different kernel functions is considered and sparse regression is used to discover an interpretable surrogate model that could be analytically presented. We present test cases in solid mechanics, fluid mechanics, and transport. Our results demonstrate that our model can achieve comparable accuracy to deep learning and can improve OOD generalization while providing more transparency and interpretability. Our study underscores the significance of interpretability in scientific machine learning and showcases the potential of functional linear models as a tool for interpreting and generalizing deep learning. | ['Bei Wang', 'Pania Newell', 'Lingxiao Yuan', 'Amirhossein Arzani'] | 2023-07-10 | null | null | null | null | ['operator-learning'] | ['miscellaneous'] | [ 2.42994353e-01 2.68416405e-01 -4.68486771e-02 -4.95836586e-01
-3.78092021e-01 -4.84191239e-01 4.70535815e-01 3.69643033e-01
5.04348613e-02 9.78140295e-01 3.48553360e-02 -6.73523605e-01
-5.40590405e-01 -6.49127901e-01 -1.08119762e+00 -8.25846553e-01
-3.71472239e-01 6.36954427e-01 -3.74207884e-01 -1.11125834e-01
6.64917901e-02 1.05463862e+00 -1.37289596e+00 5.07952869e-01
1.12544954e+00 1.11235106e+00 -1.35281995e-01 5.72417557e-01
-1.88288882e-01 6.49217725e-01 -6.29818499e-01 -9.87597555e-02
2.98085719e-01 -1.63825870e-01 -8.30922544e-01 -2.50995725e-01
4.11509067e-01 -1.09700233e-01 -2.84581929e-01 6.67357564e-01
2.40238741e-01 1.11745469e-01 1.06220543e+00 -1.50348413e+00
-1.28985178e+00 3.26094359e-01 -1.24794126e-01 1.13872863e-01
7.75853870e-03 3.27431768e-01 9.16520357e-01 -8.62797320e-01
1.61966294e-01 1.39641821e+00 7.56226063e-01 5.18075466e-01
-1.64520037e+00 -4.08145428e-01 3.99242975e-02 -7.42522106e-02
-9.71782029e-01 -8.28619897e-02 5.72544634e-01 -7.49500513e-01
8.16401303e-01 5.29559851e-01 6.68727517e-01 1.17982101e+00
8.67365301e-01 6.63773179e-01 1.20293605e+00 -3.43547493e-01
4.68935609e-01 1.02220431e-01 4.32406068e-01 7.69691586e-01
5.26313663e-01 4.48057711e-01 -6.17856562e-01 -2.66109467e-01
8.96824777e-01 1.33348443e-02 -3.35909784e-01 -3.53127301e-01
-1.00062585e+00 8.84500504e-01 4.21633720e-01 -4.56559695e-02
-3.25307161e-01 6.28204226e-01 3.96531105e-01 4.33457762e-01
7.33761132e-01 1.10914528e+00 -6.55699909e-01 2.12976828e-01
-5.55476487e-01 4.15605038e-01 9.91584063e-01 5.41877925e-01
7.78212428e-01 2.93226361e-01 -7.65817612e-02 4.03457642e-01
3.72800142e-01 5.77279568e-01 1.07977048e-01 -1.04462934e+00
-1.24989999e-02 7.14620471e-01 3.14073153e-02 -8.21989894e-01
-5.97335219e-01 -5.67865312e-01 -9.69621181e-01 6.61414981e-01
4.15602416e-01 1.64287806e-01 -1.02527106e+00 1.39645290e+00
2.25362197e-01 -8.14500637e-03 1.34099275e-01 1.05623055e+00
8.83404553e-01 7.98823476e-01 2.76312023e-01 1.30685508e-01
1.02764511e+00 -5.25894642e-01 -4.98166591e-01 1.09754659e-01
7.34647572e-01 -3.45132142e-01 1.34627306e+00 5.55784345e-01
-9.05442476e-01 -4.79013771e-01 -1.11630690e+00 -4.20359761e-01
-6.10360980e-01 -1.38338000e-01 1.13334119e+00 4.11785692e-01
-9.20355380e-01 1.07754838e+00 -9.55380559e-01 -7.72052780e-02
5.46192586e-01 7.34754384e-01 -4.14266169e-01 2.65225977e-01
-1.10074162e+00 9.00306702e-01 2.11697325e-01 2.24708915e-01
-8.08187842e-01 -1.37752438e+00 -8.09401572e-01 1.32233337e-01
-4.82794754e-02 -9.55144227e-01 9.29274797e-01 -8.23463440e-01
-1.55855250e+00 5.28213441e-01 -1.52961984e-01 -5.99952340e-01
4.85909313e-01 -3.49305928e-01 -2.89856076e-01 -1.13367394e-03
-1.27948344e-01 4.78050590e-01 7.56468058e-01 -1.35724282e+00
9.96845290e-02 -1.87961027e-01 2.79001355e-01 -2.70235658e-01
-2.17309102e-01 -6.57050014e-01 3.08393002e-01 -5.68194866e-01
-2.46157683e-02 -8.71362686e-01 -1.25996366e-01 5.76024473e-01
-5.56606531e-01 -2.33575702e-01 1.04388630e+00 -5.01681089e-01
9.19317961e-01 -1.80366051e+00 3.25979382e-01 2.64827222e-01
7.96914399e-01 1.04230173e-01 -5.01421429e-02 3.24949354e-01
-3.75488818e-01 4.37762380e-01 -5.89169085e-01 -9.32700485e-02
1.33957580e-01 4.87109989e-01 -5.29925644e-01 5.88766694e-01
6.92784607e-01 1.20375860e+00 -8.44308257e-01 -3.23131792e-02
4.80031043e-01 5.35852075e-01 -5.74782729e-01 2.46194810e-01
-4.88715500e-01 1.00507522e+00 -5.48525810e-01 5.26058376e-01
6.79689407e-01 -5.22058666e-01 -1.48521870e-01 -3.45752567e-01
7.33808950e-02 3.15456353e-02 -8.63349676e-01 1.39174283e+00
-8.77912462e-01 9.17852640e-01 -1.61449552e-01 -1.56618559e+00
9.47464585e-01 4.49081436e-02 7.41333008e-01 -5.41773498e-01
-8.77679065e-02 3.52759391e-01 2.68626928e-01 -7.56351411e-01
1.47848561e-01 -4.09001470e-01 -2.21245252e-02 4.35724169e-01
-2.91816499e-02 -2.58357108e-01 -1.36778444e-01 -1.61535382e-01
9.44798172e-01 2.59336203e-01 2.04030603e-01 -1.02019536e+00
6.74986720e-01 2.71021072e-02 5.79662155e-03 7.12678075e-01
1.48740694e-01 4.32095319e-01 6.51821673e-01 -1.00619864e+00
-1.14861989e+00 -1.34589183e+00 -6.44281149e-01 8.72982860e-01
1.02990709e-01 -9.23822969e-02 -4.50400144e-01 -5.05916059e-01
6.32940292e-01 7.60674477e-01 -9.28356647e-01 -4.45825547e-01
-5.56148112e-01 -9.02160883e-01 5.89487970e-01 6.55935287e-01
1.81723341e-01 -7.92143166e-01 -3.24473113e-01 6.74891099e-02
4.06037897e-01 -7.67254591e-01 -4.81563900e-03 5.67635894e-01
-8.60191107e-01 -1.07836986e+00 -2.68865407e-01 -2.39316463e-01
7.73732662e-01 -1.38728708e-01 1.16668463e+00 1.54592037e-01
-4.35130894e-01 4.03058857e-01 1.26405939e-01 -7.03771710e-01
-9.30495381e-01 2.21741498e-02 3.54788452e-01 -1.66612387e-01
1.86952069e-01 -5.79837561e-01 -5.81459045e-01 3.36734280e-02
-1.17362666e+00 1.02326743e-01 5.76326370e-01 9.88209546e-01
4.83606517e-01 -2.31760100e-01 7.52018273e-01 -9.37234879e-01
7.45160997e-01 -6.31255090e-01 -4.21498030e-01 1.83271945e-01
-7.11201847e-01 7.70587265e-01 1.19166875e+00 -3.86736006e-01
-8.89208734e-01 -1.90714791e-01 -7.81210065e-02 -4.46426868e-01
-1.03689887e-01 5.98547518e-01 -1.23523317e-01 -3.79350305e-01
1.05285239e+00 -4.01785821e-02 3.32054824e-01 -5.08985996e-01
2.92900920e-01 4.82351691e-01 2.70758212e-01 -1.23789620e+00
6.96213722e-01 5.13940573e-01 7.17508316e-01 -9.88687694e-01
-8.41731548e-01 3.48770656e-02 -6.92267179e-01 -7.97639117e-02
7.11652100e-01 -4.94260699e-01 -9.92364287e-01 1.16874501e-01
-1.20876205e+00 -3.63251776e-01 -6.57019317e-01 4.46061611e-01
-4.30819631e-01 2.66395926e-01 -4.72483635e-01 -7.03360677e-01
-3.15495461e-01 -1.33216882e+00 1.36866248e+00 6.58064261e-02
-5.74787498e-01 -1.75560212e+00 -2.04582766e-01 6.92098141e-02
4.76212144e-01 6.60742044e-01 1.41453230e+00 -7.49366343e-01
-5.77246666e-01 -4.79101911e-02 -2.51275688e-01 3.50498646e-01
2.27509215e-01 7.73329362e-02 -1.24961483e+00 -1.63984179e-01
-4.18994240e-02 -2.92623818e-01 8.69830310e-01 6.16315544e-01
1.90979135e+00 -5.17432690e-01 -2.08298266e-01 8.03190053e-01
1.13571358e+00 -1.90413408e-02 2.68473625e-01 6.27169339e-03
8.34677398e-01 6.65442228e-01 -3.13330777e-02 1.95316523e-01
-1.46509767e-01 4.29237664e-01 3.04610223e-01 -4.29926008e-01
8.98068957e-03 -6.62729740e-02 1.71057478e-01 4.12270546e-01
-1.81342542e-01 -3.62092614e-01 -8.03057253e-01 -5.97846881e-02
-1.79615533e+00 -5.65153956e-01 -5.05594313e-01 1.98984444e+00
7.19029248e-01 2.28082463e-02 -1.86002836e-01 6.54188320e-02
3.67048562e-01 -2.17798516e-01 -1.11867476e+00 -8.40966821e-01
-9.28638652e-02 6.42793953e-01 3.74984086e-01 6.54944301e-01
-9.18867230e-01 3.98727298e-01 6.92724371e+00 6.37494922e-01
-1.57741201e+00 -1.02396645e-01 1.02103698e+00 1.35258302e-01
-6.98159218e-01 -2.89164275e-01 -4.53081101e-01 4.07447249e-01
1.08645630e+00 -2.74471968e-01 3.41644734e-01 7.60007560e-01
5.11747122e-01 2.42944866e-01 -1.83534348e+00 6.59470797e-01
-4.47394967e-01 -1.73762846e+00 4.42731440e-01 2.99786568e-01
5.86133063e-01 -3.11173141e-01 2.78263271e-01 9.69222412e-02
2.03888074e-01 -1.72498047e+00 5.81058919e-01 7.18964815e-01
8.87852609e-01 -3.14108908e-01 5.61551630e-01 1.52738154e-01
-8.74277771e-01 1.75377890e-01 -3.76844645e-01 -3.48832458e-01
-1.69289961e-01 9.09299076e-01 -7.46759713e-01 4.64998990e-01
4.25845206e-01 8.43273818e-01 -2.38256216e-01 6.33795917e-01
1.48500487e-01 6.54603958e-01 -5.15917301e-01 -1.00495473e-01
5.72809689e-02 -2.13737458e-01 5.54127097e-01 1.19599557e+00
2.13646531e-01 -1.09556757e-01 -9.62460935e-02 1.57844424e+00
-3.04762297e-03 -1.61943123e-01 -6.53300583e-01 -1.88457966e-01
-1.02239475e-02 1.09888303e+00 -5.15508890e-01 -1.76542550e-01
-1.91631734e-01 6.61118984e-01 2.63628572e-01 6.10906243e-01
-9.50815976e-01 3.96119691e-02 9.45629060e-01 3.10138285e-01
-3.28151762e-01 -3.57518822e-01 -8.50066423e-01 -1.06233573e+00
-6.37857839e-02 -4.61380810e-01 9.09139663e-02 -5.53906083e-01
-1.72369862e+00 1.85222626e-01 2.73165494e-01 -9.88338768e-01
-2.86279488e-02 -1.42748272e+00 -6.54806077e-01 1.00302494e+00
-1.25669515e+00 -1.01109779e+00 -2.73436904e-01 1.61451567e-02
2.38495484e-01 -5.54358251e-02 8.73796821e-01 6.05328791e-02
-3.71220112e-01 3.40350121e-01 2.96460748e-01 -2.62503058e-01
3.16913694e-01 -1.53505945e+00 3.74512702e-01 2.39986256e-01
-8.46979618e-02 8.71484935e-01 9.51069951e-01 -4.46713597e-01
-1.78314114e+00 -1.07284474e+00 2.18891487e-01 -5.99497020e-01
7.58707106e-01 -4.21624988e-01 -1.27245760e+00 4.16270792e-01
-2.25809097e-01 1.88375488e-01 8.92615736e-01 6.39600307e-02
-3.27709243e-02 -6.08467907e-02 -1.14443588e+00 4.27434653e-01
1.09841895e+00 -4.95586216e-01 -3.35364640e-01 5.10772765e-01
7.97693789e-01 -6.14944100e-01 -1.21735907e+00 5.49795747e-01
6.19103670e-01 -6.80678964e-01 1.16525757e+00 -1.12932277e+00
8.05896103e-01 -1.71799377e-01 1.38163343e-01 -1.38013375e+00
-2.44319797e-01 -4.64738339e-01 -6.61520422e-01 5.45104265e-01
4.46072519e-01 -8.35628092e-01 7.05715120e-01 1.11841083e+00
-3.95739555e-01 -1.28436255e+00 -9.52126086e-01 -8.55137408e-01
6.76626921e-01 -4.99447286e-01 5.67015290e-01 8.22289526e-01
-1.49141923e-01 1.78910851e-01 -1.04510702e-01 3.25938255e-01
5.42907774e-01 1.72754869e-01 5.74616790e-01 -1.52705562e+00
-3.92863184e-01 -4.82906669e-01 -4.81981426e-01 -1.07393599e+00
5.22466660e-01 -1.23645985e+00 -2.84499049e-01 -1.37985742e+00
-1.80380076e-01 -5.26564121e-01 -2.24188581e-01 4.31144774e-01
1.38547141e-02 -2.66997330e-02 -4.82687727e-02 1.97431087e-01
-6.70485152e-03 6.46644473e-01 1.59665334e+00 -3.62809092e-01
-1.35689497e-01 2.58962275e-03 -7.43497968e-01 6.21398568e-01
7.46994138e-01 -2.17955574e-01 -4.63872403e-01 -3.39353442e-01
1.45179629e-01 -1.12669252e-01 9.25191820e-01 -8.07268441e-01
-2.14807481e-01 -4.55037445e-01 7.56308436e-01 -4.28435691e-02
2.31168300e-01 -8.31411898e-01 2.21686199e-01 6.35651946e-01
-6.13220870e-01 -1.91730559e-01 6.15565717e-01 6.08487785e-01
1.33761153e-01 -1.42800018e-01 5.54629982e-01 2.12922506e-02
-4.80016112e-01 4.20200318e-01 -4.02618438e-01 1.11924924e-01
7.67988503e-01 -1.21818632e-01 -3.74201566e-01 -1.98702440e-01
-7.85895944e-01 6.13403916e-02 2.95200735e-01 2.65039206e-01
5.29310703e-01 -1.10831487e+00 -5.28543413e-01 3.09402734e-01
5.10068145e-04 2.77787507e-01 3.89308184e-02 4.72430676e-01
-9.20398891e-01 5.79404950e-01 -1.36993647e-01 -8.74979913e-01
-4.55854088e-01 2.70903707e-01 6.11270368e-01 -1.52373448e-01
-5.92044592e-01 5.50459266e-01 6.77316010e-01 -5.39250493e-01
-1.16903529e-01 -9.01405454e-01 3.20286900e-01 -6.64985299e-01
5.26014976e-02 3.90878588e-01 2.45271996e-02 -1.43625498e-01
-2.27899075e-01 4.30981517e-01 2.10840955e-01 3.33043903e-01
1.47160828e+00 3.47017556e-01 -1.12604000e-01 6.24335587e-01
1.45339453e+00 -2.48609275e-01 -1.47410738e+00 1.91836879e-01
-1.32129490e-01 -1.22038640e-01 -4.87951972e-02 -7.67686665e-01
-9.46282744e-01 1.24819243e+00 3.68738264e-01 3.94699574e-01
6.92107201e-01 5.26932217e-02 6.51410639e-01 8.62332404e-01
2.67055202e-02 -7.37839818e-01 1.57628685e-01 2.77478725e-01
1.22191489e+00 -1.39786553e+00 2.63000011e-01 -4.83676612e-01
-2.52972573e-01 1.66398406e+00 6.97960615e-01 -2.35759795e-01
8.56940866e-01 4.21498984e-01 -2.66331255e-01 -3.66997808e-01
-4.50750381e-01 4.26471025e-01 7.02128232e-01 5.77128112e-01
6.17107511e-01 -1.85688939e-02 -7.67865479e-02 6.00017250e-01
-3.05688173e-01 7.73822591e-02 4.45154369e-01 6.72130644e-01
-2.41453871e-01 -9.34317768e-01 -2.21628845e-01 7.31234908e-01
-2.15829521e-01 -5.32437451e-02 -3.02678376e-01 8.77698541e-01
7.92813823e-02 4.98629749e-01 8.40150192e-02 -9.42156240e-02
5.58170304e-02 4.01441045e-02 4.34491098e-01 -5.16152442e-01
-3.21807027e-01 -5.53249061e-01 -8.79466683e-02 -4.66816127e-01
-4.02805150e-01 -2.14963034e-01 -1.57875109e+00 -6.93018794e-01
-7.88452625e-02 1.89142287e-01 4.92119789e-01 1.32083762e+00
6.39766395e-01 7.57369399e-01 1.60026342e-01 -9.13771033e-01
-5.43970942e-01 -5.57434082e-01 -4.48056430e-01 5.03358364e-01
7.08751202e-01 -1.02379084e+00 -5.27236581e-01 2.53990889e-01] | [6.4864420890808105, 3.594923257827759] |
e2020852-2b6c-4039-8dd6-6c89ac6246c3 | eye-tracking-guided-deep-multiple-instance | 2304.12719 | null | https://arxiv.org/abs/2304.12719v1 | https://arxiv.org/pdf/2304.12719v1.pdf | Eye tracking guided deep multiple instance learning with dual cross-attention for fundus disease detection | Deep neural networks (DNNs) have promoted the development of computer aided diagnosis (CAD) systems for fundus diseases, helping ophthalmologists reduce missed diagnosis and misdiagnosis rate. However, the majority of CAD systems are data-driven but lack of medical prior knowledge which can be performance-friendly. In this regard, we innovatively proposed a human-in-the-loop (HITL) CAD system by leveraging ophthalmologists' eye-tracking information, which is more efficient and accurate. Concretely, the HITL CAD system was implemented on the multiple instance learning (MIL), where eye-tracking gaze maps were beneficial to cherry-pick diagnosis-related instances. Furthermore, the dual-cross-attention MIL (DCAMIL) network was utilized to curb the adverse effects of noisy instances. Meanwhile, both sequence augmentation module and domain adversarial module were introduced to enrich and standardize instances in the training bag, respectively, thereby enhancing the robustness of our method. We conduct comparative experiments on our newly constructed datasets (namely, AMD-Gaze and DR-Gaze), respectively for the AMD and early DR detection. Rigorous experiments demonstrate the feasibility of our HITL CAD system and the superiority of the proposed DCAMIL, fully exploring the ophthalmologists' eye-tracking information. These investigations indicate that physicians' gaze maps, as medical prior knowledge, is potential to contribute to the CAD systems of clinical diseases. | ['Jiang Liu', 'Mengdi Gao', 'Xiaoqing Zhang', 'Chen Tang', 'Jingqi Huang', 'Hongyang Jiang'] | 2023-04-25 | null | null | null | null | ['multiple-instance-learning'] | ['methodology'] | [ 2.03730330e-01 -7.79955909e-02 -3.13372947e-02 -2.63229579e-01
-4.32946503e-01 -9.16503072e-02 5.28759323e-02 -3.68140996e-01
-4.56997365e-01 7.76084304e-01 -6.93551823e-02 -4.08882231e-01
-1.56852990e-01 -4.86525685e-01 -4.54277694e-01 -9.08185363e-01
3.89179796e-01 -5.47338603e-03 -4.18081805e-02 -6.22299239e-02
1.36435479e-01 4.01199847e-01 -1.77734435e+00 2.19828993e-01
1.48303831e+00 8.78506005e-01 1.86756570e-02 5.32288730e-01
-7.41588101e-02 8.86976302e-01 -5.46083808e-01 -4.94040102e-01
3.37399989e-02 -4.97401565e-01 -4.03157383e-01 1.71745881e-01
4.96191859e-01 -4.81269062e-01 -3.84873450e-01 1.13986087e+00
9.79901493e-01 -3.23679149e-01 3.96030784e-01 -1.22723532e+00
-1.09507751e+00 -4.48369682e-02 -9.87334251e-01 3.34697247e-01
4.57476340e-02 8.35798681e-01 4.56587911e-01 -6.83354974e-01
4.17471647e-01 1.12173450e+00 4.95621264e-01 9.07346308e-01
-8.00833106e-01 -8.91181409e-01 1.66320503e-01 3.34599018e-01
-1.28053832e+00 -3.80859405e-01 5.54151058e-01 -6.52584434e-01
3.99373263e-01 1.85293809e-01 6.69930875e-01 1.11465001e+00
-1.20802801e-02 9.63149548e-01 1.15651357e+00 -3.83261919e-01
-3.70624959e-01 3.33702207e-01 2.85810709e-01 7.68654704e-01
5.37262082e-01 4.98379081e-01 -5.05022295e-02 4.43401337e-02
9.45864558e-01 2.74581134e-01 -6.07334316e-01 1.48270518e-01
-1.14550340e+00 4.77624744e-01 7.34444678e-01 2.69042607e-02
-2.78383523e-01 -3.68562490e-01 4.87854004e-01 -6.99173585e-02
4.13341254e-01 3.02369207e-01 -2.57615179e-01 2.72787839e-01
-5.68146169e-01 1.49763096e-02 2.11557120e-01 9.23813879e-01
3.00235748e-01 -9.07913316e-03 -5.78521967e-01 4.90188390e-01
3.09318423e-01 5.16544461e-01 5.79262435e-01 -6.17917299e-01
2.37667605e-01 9.86801744e-01 9.06416327e-02 -7.71844149e-01
-3.87209535e-01 -7.02392876e-01 -1.25912714e+00 1.95392251e-01
1.95852801e-01 -3.80642354e-01 -1.16976011e+00 1.50531602e+00
3.26224029e-01 6.88442469e-01 1.78785939e-02 1.18663120e+00
1.15080631e+00 2.05257773e-01 2.82802403e-01 -2.78555095e-01
1.39841151e+00 -7.76522279e-01 -9.12224174e-01 6.80534840e-02
8.00838053e-01 -6.88057899e-01 1.19788265e+00 2.65638262e-01
-8.25617313e-01 -8.34513843e-01 -7.95295954e-01 -2.08794549e-01
-1.30913500e-02 7.18314946e-01 5.97579122e-01 5.80534220e-01
-7.66803861e-01 -2.71452107e-02 -7.01431453e-01 -9.04044062e-02
1.10382390e+00 2.76843131e-01 -6.86855242e-02 -2.76361525e-01
-1.17682219e+00 7.47208655e-01 1.68018281e-01 5.68753362e-01
-6.27282262e-01 -8.67152870e-01 -6.96178854e-01 -2.18695566e-01
3.14577818e-01 -1.08781433e+00 9.65860546e-01 -9.54876304e-01
-1.21777284e+00 1.19338667e+00 -3.77325922e-01 -4.55511063e-01
5.67148387e-01 -3.57990146e-01 -7.27516055e-01 9.90429446e-02
-1.45507067e-01 7.37223983e-01 8.78324747e-01 -9.09655929e-01
-7.98237026e-01 -3.01235884e-01 -1.38727367e-01 -8.28581527e-02
-1.65585205e-01 -1.30294427e-01 -3.56438398e-01 -6.39014423e-01
-3.80448878e-01 -8.34662259e-01 -1.27746701e-01 3.35012197e-01
-6.06842935e-01 -3.10148627e-01 6.75390780e-01 -7.66577661e-01
1.41606593e+00 -2.42709398e+00 -1.58747524e-01 3.43667273e-03
6.43403351e-01 1.07778561e+00 -2.37475678e-01 -2.68393815e-01
-1.68480098e-01 3.27143908e-01 -7.37097561e-02 -1.41277686e-01
-4.77375031e-01 -9.53588355e-03 -1.34387270e-01 4.19499636e-01
8.76409948e-01 1.11169755e+00 -7.16046691e-01 -5.48742354e-01
2.54851401e-01 5.41973114e-01 -5.42020380e-01 3.22475016e-01
-1.69890001e-01 7.32326329e-01 -4.25038636e-01 7.74782062e-01
8.37502003e-01 -6.83539569e-01 -3.35201025e-01 -4.69041228e-01
7.10365921e-02 -1.85057312e-01 -6.61695123e-01 1.30521131e+00
-8.35660025e-02 5.61063945e-01 -3.77463311e-01 -4.44165617e-01
8.20564270e-01 2.54514098e-01 1.73859194e-01 -6.85356855e-01
3.55839878e-01 8.45643431e-02 3.09401989e-01 -1.18443871e+00
-1.68903336e-01 1.34558659e-02 6.61446333e-01 -2.41816062e-02
-3.75516206e-01 6.54640615e-01 -5.04173487e-02 8.88671260e-03
6.86607540e-01 1.08798146e-01 1.70409888e-01 2.48327240e-01
8.45451295e-01 1.06539938e-03 6.75140083e-01 3.46252024e-01
-5.11855185e-01 5.71085393e-01 4.14212823e-01 -3.82039696e-01
-7.89511621e-01 -7.18001723e-01 -4.27849114e-01 3.10397774e-01
4.24868792e-01 1.94274724e-01 -6.14495933e-01 -7.86614656e-01
-7.13774189e-02 3.88890713e-01 -6.17451727e-01 -3.97541404e-01
-2.09050596e-01 -8.13743651e-01 6.73619568e-01 5.04690230e-01
7.56121576e-01 -8.41983438e-01 -3.07883441e-01 -8.34018439e-02
1.67749599e-02 -8.41124654e-01 -6.26240432e-01 -7.66709030e-01
-7.56464005e-01 -1.55532789e+00 -1.10794079e+00 -8.22907209e-01
8.83703530e-01 2.18020409e-01 7.90092230e-01 2.27092490e-01
-8.34989846e-01 6.75546303e-02 -2.03212380e-01 -6.59308672e-01
-2.79152811e-01 -2.02688098e-01 -7.57517219e-02 3.83620113e-01
9.54187214e-01 -2.02297047e-01 -1.09483981e+00 1.70876503e-01
-6.98954582e-01 1.75749123e-01 1.01472831e+00 1.16820800e+00
5.95357835e-01 -3.32878679e-01 6.17271364e-01 -1.22476566e+00
5.99354804e-01 -3.21118295e-01 -6.29389048e-01 3.64848584e-01
-6.91300750e-01 -1.62530467e-01 2.48673543e-01 -6.18956029e-01
-1.17691910e+00 -1.02823883e-01 -1.54270634e-01 -9.05874312e-01
-3.33744496e-01 3.66506875e-01 -1.77260086e-01 -2.85427839e-01
7.28219032e-01 2.06607357e-01 4.29652721e-01 -4.48378623e-01
1.99847877e-01 8.93966079e-01 6.56245172e-01 -9.66857225e-02
4.50076044e-01 2.62939394e-01 -1.49953226e-02 -4.99450058e-01
-1.00847471e+00 -4.08080637e-01 -1.69797882e-01 -6.65848255e-02
8.18774462e-01 -1.24704814e+00 -1.15960741e+00 8.66811991e-01
-1.27267635e+00 3.65016162e-01 1.66543182e-02 5.95085502e-01
-7.05015287e-03 2.98363864e-01 -5.94974935e-01 -7.99846172e-01
-5.10494888e-01 -1.26069725e+00 8.66502941e-01 8.10962737e-01
1.34268045e-01 -7.22791135e-01 -1.68503270e-01 4.58797574e-01
1.53827876e-01 2.71258295e-01 9.77608442e-01 -5.47195077e-01
-8.42615068e-01 -2.05142409e-01 -7.45269239e-01 7.47642875e-01
2.08949253e-01 2.27721453e-01 -1.24405503e+00 -2.81835258e-01
-1.76522389e-01 -5.31604476e-02 6.92278504e-01 6.28287375e-01
1.32485497e+00 -2.49646336e-01 -5.77616632e-01 7.99659729e-01
1.29836142e+00 3.49154234e-01 1.03314841e+00 7.36074224e-02
9.70541954e-01 4.35540527e-01 7.56786704e-01 3.56169552e-01
4.33610469e-01 1.05946936e-01 3.63992065e-01 -7.52398193e-01
-3.55872393e-01 -1.12312689e-01 -9.15547982e-02 4.64600712e-01
-2.01946154e-01 -2.80924857e-01 -7.44974136e-01 5.80041647e-01
-1.55628228e+00 -7.07514226e-01 -4.01770055e-01 1.90885532e+00
9.71684873e-01 -8.88056383e-02 4.64226492e-02 -1.62332892e-01
1.10187840e+00 -3.54344040e-01 -1.04721868e+00 2.58484364e-01
-1.55334234e-01 2.91006029e-01 2.83693045e-01 9.90416482e-03
-1.03054392e+00 5.57655036e-01 5.12698126e+00 7.56731391e-01
-1.34815311e+00 1.25734052e-02 8.21157217e-01 -2.64847934e-01
-1.29055411e-01 -5.22229910e-01 -1.00341856e+00 8.80235016e-01
5.29237688e-01 4.16030772e-02 2.35032644e-02 7.65044928e-01
3.95833045e-01 2.44269669e-01 -8.58593047e-01 1.17963529e+00
-8.21667984e-02 -1.58764088e+00 2.41055399e-01 1.12725860e-02
5.55292428e-01 -1.88036427e-01 4.35464829e-01 1.85272589e-01
-9.65180546e-02 -1.00693834e+00 -2.17574552e-01 9.82003629e-01
1.25787556e+00 -7.21524894e-01 1.08169460e+00 1.31298974e-01
-7.24510133e-01 7.21605588e-03 -1.86222270e-01 3.35983157e-01
-3.20104100e-02 4.88802224e-01 -7.68539488e-01 5.77418268e-01
6.44826949e-01 1.06581390e+00 -6.79416776e-01 1.49788439e+00
-1.97880492e-01 5.64844668e-01 1.22867348e-02 1.19333036e-01
-2.34499350e-02 -7.06745014e-02 5.57364404e-01 8.29894185e-01
2.31416270e-01 2.91027695e-01 -6.05733395e-02 1.13310635e+00
-1.19264439e-01 -1.22296505e-01 -3.52972120e-01 -1.86261743e-01
5.45231938e-01 1.00107598e+00 1.25780746e-01 -1.37237892e-01
-5.43905377e-01 5.91950238e-01 -8.33650455e-02 6.02012873e-01
-9.07353997e-01 -7.15693951e-01 9.28584218e-01 2.29244143e-01
2.05354422e-01 5.03910482e-01 -3.91319513e-01 -1.12820721e+00
4.87016141e-02 -1.05255949e+00 3.17005724e-01 -1.03958845e+00
-1.63526380e+00 6.54772282e-01 -6.74330771e-01 -1.66684473e+00
1.75236851e-01 -7.30777085e-01 -5.02901614e-01 1.30693424e+00
-2.12859201e+00 -1.32203567e+00 -5.71059644e-01 8.02749097e-01
1.22096211e-01 -4.66986835e-01 4.22066212e-01 6.55392468e-01
-1.10180581e+00 1.02505350e+00 -1.20536527e-02 3.03353190e-01
9.91116941e-01 -7.97790647e-01 -4.33629751e-03 9.26170707e-01
-4.23148990e-01 8.99628103e-01 2.29872182e-01 -6.00938737e-01
-9.14844871e-01 -1.38513601e+00 5.48789561e-01 -3.15864056e-01
3.26054424e-01 3.66832256e-01 -1.21612966e+00 5.92759490e-01
-8.77545327e-02 1.67745516e-01 6.75717652e-01 -1.87423304e-01
-1.86722144e-01 -1.34366542e-01 -1.13162935e+00 6.28674448e-01
9.07493949e-01 -5.61707616e-01 -5.51732779e-01 2.46693224e-01
9.84492421e-01 -6.94801390e-01 -6.90915108e-01 6.50536001e-01
3.43041450e-01 -7.69701004e-01 9.65703726e-01 -8.57884347e-01
6.20773673e-01 -5.00533819e-01 6.28781319e-01 -8.73541415e-01
-1.72573492e-01 -5.60960472e-01 -1.92637399e-01 1.24793553e+00
9.10944194e-02 -9.16777730e-01 5.45108140e-01 5.37821531e-01
-1.51865602e-01 -9.16351438e-01 -5.41120350e-01 -3.04601490e-01
-2.60886580e-01 -3.21116000e-02 6.66971982e-01 8.07661831e-01
-6.31640971e-01 2.82134473e-01 -4.36739177e-01 4.95680869e-01
5.02689540e-01 -1.93981454e-01 6.69234097e-01 -1.25201190e+00
-1.28530696e-01 -3.26734722e-01 -4.78062868e-01 -9.39233243e-01
-1.77395537e-01 -7.21893489e-01 -2.91192144e-01 -1.15106249e+00
1.41444374e-02 -4.55690324e-01 -4.66766238e-01 3.38170469e-01
-8.02551985e-01 1.03893057e-01 -5.53022325e-02 3.88718665e-01
-2.35353619e-01 3.95196497e-01 1.88581061e+00 -1.41081512e-01
-2.07333252e-01 2.16907993e-01 -9.50232089e-01 5.41510284e-01
3.95029694e-01 -2.56874144e-01 -4.32412624e-01 -4.13473099e-01
5.99485934e-02 -1.73436388e-01 7.26601064e-01 -7.40133822e-01
3.96303415e-01 1.34529218e-01 3.43267947e-01 -6.56512499e-01
-7.95295537e-02 -5.73715985e-01 -1.21412985e-01 4.86386895e-01
-1.64824799e-01 -5.72183490e-01 4.37091500e-01 7.85802424e-01
-4.02237236e-01 1.24088719e-01 9.07305956e-01 5.00690229e-02
-7.26799548e-01 7.83211589e-01 1.44520432e-01 1.83357343e-01
1.16286182e+00 -3.12332630e-01 -7.16639817e-01 1.35219231e-01
-6.30434334e-01 4.97596711e-01 1.58207029e-01 2.05267966e-01
8.22003603e-01 -1.08451295e+00 -7.80135989e-01 6.63361967e-01
4.21088517e-01 4.86581028e-01 8.85458231e-01 1.24783540e+00
-5.86390316e-01 2.66801953e-01 -1.84575334e-01 -7.84031630e-01
-1.37907851e+00 8.22285295e-01 5.53706825e-01 3.28338481e-02
-5.77806950e-01 1.13371706e+00 4.71412420e-01 -3.65371928e-02
3.95141214e-01 -4.23377961e-01 -4.38351572e-01 -1.02938890e-01
8.19355905e-01 3.40370804e-01 -1.15437098e-01 -9.70860049e-02
-2.34972358e-01 6.77697599e-01 -6.37278318e-01 6.91908181e-01
1.04124928e+00 -2.49421701e-01 -8.52713808e-02 -3.81388068e-02
7.80376852e-01 -2.99916357e-01 -1.17937052e+00 -4.34380084e-01
-3.22877377e-01 -4.60235804e-01 1.53900221e-01 -1.09901547e+00
-1.17174256e+00 1.31756485e+00 1.10740709e+00 -1.91647306e-01
1.40810513e+00 -3.78256470e-01 9.82968867e-01 4.17694002e-02
-1.10131748e-01 -2.82420188e-01 -1.63262829e-01 -5.58444373e-02
5.63488066e-01 -1.51905608e+00 -1.91178918e-01 -3.95704240e-01
-9.76846337e-01 1.01399970e+00 9.34853554e-01 6.48073256e-02
4.79978472e-01 -3.16031389e-02 3.88066322e-01 -7.82942101e-02
-5.14219582e-01 -4.18639868e-01 5.53207338e-01 8.60416591e-01
2.27194980e-01 -2.33355671e-01 -1.36698604e-01 9.15166199e-01
4.34223622e-01 7.98464775e-01 3.56655508e-01 5.64395487e-01
-1.15399845e-01 -7.92270541e-01 -1.34208381e-01 6.22087657e-01
-3.54170531e-01 -5.06537378e-01 -8.28473270e-02 8.10324788e-01
5.52216947e-01 6.59977376e-01 1.59908578e-01 -4.30429786e-01
3.92750770e-01 -1.67965427e-01 3.21014166e-01 -5.56026161e-01
-3.92249912e-01 8.67232308e-02 -2.16586545e-01 -2.96487957e-01
-5.02796531e-01 -3.41861576e-01 -9.45376933e-01 -2.86903560e-01
-4.66631919e-01 -3.77413392e-01 -1.48982275e-02 9.96985495e-01
9.40177739e-01 8.71943831e-01 5.07351458e-01 -2.19113499e-01
-4.01552022e-01 -8.94947171e-01 -4.55346435e-01 2.15757266e-01
8.29705596e-01 -5.96207023e-01 -3.12605083e-01 3.17646384e-01] | [15.781620025634766, -3.9573569297790527] |
6a4162f1-0fb8-4faa-872d-e29739423226 | leveraging-artificial-intelligence-on-binary | 2210.05103 | null | https://arxiv.org/abs/2210.05103v1 | https://arxiv.org/pdf/2210.05103v1.pdf | Leveraging Artificial Intelligence on Binary Code Comprehension | Understanding binary code is an essential but complex software engineering task for reverse engineering, malware analysis, and compiler optimization. Unlike source code, binary code has limited semantic information, which makes it challenging for human comprehension. At the same time, compiling source to binary code, or transpiling among different programming languages (PLs) can provide a way to introduce external knowledge into binary comprehension. We propose to develop Artificial Intelligence (AI) models that aid human comprehension of binary code. Specifically, we propose to incorporate domain knowledge from large corpora of source code (e.g., variable names, comments) to build AI models that capture a generalizable representation of binary code. Lastly, we will investigate metrics to assess the performance of models that apply to binary code by using human studies of comprehension. | ['Yifan Zhang'] | 2022-10-11 | null | null | null | null | ['compiler-optimization'] | ['computer-code'] | [ 1.96365044e-01 2.09876865e-01 -7.91348338e-01 -4.28860098e-01
-2.35640422e-01 -1.03789032e+00 3.56469303e-01 6.84087515e-01
7.54772723e-02 7.76661411e-02 3.55070323e-01 -1.21534693e+00
3.93573552e-01 -9.78978813e-01 -7.01755047e-01 4.53422546e-01
1.65485919e-01 9.29928198e-02 4.91626188e-02 -3.62255186e-01
6.83634162e-01 6.76110312e-02 -1.09316278e+00 2.82714516e-01
1.33635700e+00 3.46276104e-01 2.63852507e-01 5.64827859e-01
-6.46398008e-01 1.04665899e+00 -7.19347000e-01 -6.48824096e-01
-1.05143853e-01 -2.56833434e-01 -9.11619186e-01 -1.79463997e-01
1.23319373e-01 -1.33397833e-01 1.25089020e-03 1.43667853e+00
-3.60197186e-01 -6.10738277e-01 6.85085654e-01 -1.36391306e+00
-1.05830121e+00 1.15677214e+00 -3.39077115e-01 4.00303528e-02
6.32906079e-01 3.76673609e-01 1.12527084e+00 -4.41069335e-01
6.28309906e-01 1.15094209e+00 5.27106285e-01 4.96148705e-01
-1.25147331e+00 -4.68764871e-01 1.12179600e-01 1.84834778e-01
-8.97047818e-01 -1.32506162e-01 7.76188791e-01 -1.03707635e+00
1.06310689e+00 1.35783240e-01 5.64742088e-01 1.06987786e+00
4.61252332e-01 6.07278526e-01 9.64115560e-01 -3.77523750e-01
1.31717443e-01 4.65986699e-01 9.98738825e-01 9.53245342e-01
5.32511175e-01 9.54526849e-03 -4.76793461e-02 -3.82559806e-01
1.57587916e-01 -6.05837665e-02 -2.30834097e-01 -2.92263269e-01
-1.00004399e+00 1.19792664e+00 5.13234198e-01 2.18142986e-01
2.98838075e-02 1.50624841e-01 6.00166440e-01 5.17537355e-01
8.41197670e-02 1.09500802e+00 -5.47129750e-01 -4.27123547e-01
-6.58884645e-01 1.02981981e-02 1.51773143e+00 9.64223862e-01
9.32434678e-01 -8.84287357e-02 3.95590961e-01 4.93495464e-01
5.95569372e-01 6.34144723e-01 6.51810706e-01 -7.44847000e-01
5.20179451e-01 9.67618227e-01 -4.12482828e-01 -1.23413110e+00
-1.77624360e-01 -5.20152509e-01 -2.04432279e-01 9.08823907e-02
9.63450223e-02 1.35000601e-01 -6.52080178e-01 1.54077578e+00
-3.51892442e-01 -3.30713332e-01 1.25836939e-01 3.80589217e-01
7.23638117e-01 4.31463987e-01 1.11878011e-02 9.00959820e-02
1.24363112e+00 -1.05763209e+00 -3.05849135e-01 -9.80559409e-01
8.78336251e-01 -5.08054852e-01 1.02067900e+00 1.37363285e-01
-6.05244398e-01 -4.21633452e-01 -1.03999352e+00 -4.00903150e-02
-5.74710429e-01 -1.87072471e-01 8.80623043e-01 1.05153191e+00
-8.59135866e-01 1.92679271e-01 -7.34538794e-01 -2.49906421e-01
2.62095451e-01 1.27374707e-02 -8.25130939e-02 -1.68375507e-01
-8.43146503e-01 9.79649246e-01 6.60767615e-01 -6.36368096e-01
-1.06791520e+00 -8.42366636e-01 -1.38037670e+00 1.70313001e-01
4.33645457e-01 -5.59712946e-01 1.31747007e+00 -1.28618884e+00
-1.10536134e+00 6.45316958e-01 -4.49446380e-01 -4.96745497e-01
-1.54908672e-01 -9.15166587e-02 -2.31068507e-01 -2.62409806e-01
1.89671218e-01 3.75986099e-01 9.18075204e-01 -1.28817463e+00
-1.93595275e-01 -1.84593171e-01 5.41486979e-01 -3.41673225e-01
-4.82856452e-01 3.40712219e-01 1.82908811e-02 -5.67077518e-01
-3.07817727e-01 -8.25695455e-01 -9.46335793e-02 -2.88389385e-01
-2.74626851e-01 -4.03741114e-02 7.96838760e-01 -1.19892907e+00
1.45133543e+00 -2.15366483e+00 5.32291293e-01 2.63596684e-01
7.42879748e-01 -5.48991077e-02 -2.95765072e-01 1.94748595e-01
-1.08674303e-01 8.12470615e-01 -3.48319441e-01 2.56547660e-01
1.88985869e-01 -5.12072295e-02 -3.60154599e-01 -6.97437152e-02
2.73921281e-01 1.24487281e+00 -9.22089219e-01 -3.43305856e-01
-1.32674724e-01 -4.36991267e-02 -1.12714016e+00 2.37832949e-01
-7.89811075e-01 1.09987818e-01 -4.17361975e-01 7.92236626e-01
2.44363368e-01 -4.96425092e-01 8.38488191e-02 -7.10730627e-03
6.67140782e-02 5.06906450e-01 -3.31124604e-01 1.36742568e+00
-8.82540166e-01 1.03414023e+00 -1.46973073e-01 -7.80391157e-01
8.70207191e-01 -1.31452441e-01 -2.16850907e-01 -4.90974277e-01
2.29231939e-01 7.48769790e-02 3.82656425e-01 -6.79070592e-01
4.25504506e-01 9.15365070e-02 -5.72928250e-01 8.57568979e-01
-2.38005355e-01 -7.36758232e-01 2.42230237e-01 4.27376896e-01
1.17493832e+00 -2.86076307e-01 6.20708108e-01 -9.72040594e-02
4.44553196e-01 5.25042355e-01 3.04179966e-01 6.14910543e-01
-2.38167606e-02 -5.20973559e-03 8.46686661e-01 -1.10026091e-01
-1.04472041e+00 -8.50057602e-01 2.22571045e-01 1.00738299e+00
6.86834827e-02 -9.33054864e-01 -8.94926846e-01 -9.81632292e-01
-8.42015743e-02 1.06192458e+00 -3.77960652e-01 -6.17120922e-01
-5.38776934e-01 -4.40684438e-01 6.70220733e-01 6.53751254e-01
3.20145071e-01 -6.57670975e-01 -6.98210597e-01 7.64174759e-02
-4.32729036e-01 -8.64049137e-01 -2.84615099e-01 1.99151427e-01
-6.63611472e-01 -1.11913037e+00 3.06791831e-02 -8.00817788e-01
5.62464058e-01 -4.43289317e-02 1.64938164e+00 7.21652210e-01
-2.08785534e-01 3.88683498e-01 -5.63635826e-01 -4.46818680e-01
-1.13640118e+00 3.92246217e-01 -4.34181213e-01 -7.60752082e-01
5.30897379e-01 -2.03266144e-01 1.00270733e-01 2.11163625e-01
-9.05835867e-01 9.41217914e-02 5.62662899e-01 6.84439957e-01
-3.05523783e-01 9.90124196e-02 9.35744122e-02 -9.71393228e-01
8.86747539e-01 -8.92967522e-01 -5.77077568e-01 3.49449486e-01
-4.59468782e-01 1.59862354e-01 9.13287222e-01 -3.97310644e-01
-8.85214806e-01 -2.63396144e-01 -1.15759149e-01 1.75838798e-01
-1.41341999e-01 1.12159884e+00 -6.98166713e-02 -2.79174656e-01
1.00988865e+00 1.74858198e-01 5.70195056e-02 -1.89897150e-01
5.98162472e-01 7.41818130e-01 4.19196606e-01 -8.06334496e-01
1.15321696e+00 -3.91375162e-02 -5.12372732e-01 -6.46387517e-01
-5.28463125e-01 -3.56748663e-02 -5.36418080e-01 1.51537627e-01
7.80401230e-01 -5.55194438e-01 -2.57331729e-01 3.11810642e-01
-1.28762031e+00 -4.60932344e-01 2.51251310e-01 -8.83725509e-02
-3.35981041e-01 5.20261049e-01 -6.49987817e-01 -1.03857532e-01
1.05230726e-01 -1.63442922e+00 7.86285281e-01 3.36747915e-01
-7.47517288e-01 -1.27908957e+00 4.99704815e-02 5.56836069e-01
7.91794360e-01 -2.70549711e-02 1.78080881e+00 -8.05922210e-01
-7.68364072e-01 5.53665049e-02 -3.88287485e-01 2.68413305e-01
2.44219437e-01 1.17859006e-01 -3.99074614e-01 -1.77839622e-01
1.29940301e-01 -3.01903665e-01 3.71047229e-01 -1.74208224e-01
1.16268563e+00 -4.03566808e-01 -5.37832618e-01 5.29352307e-01
1.32103407e+00 2.80448735e-01 4.38033491e-01 3.90657365e-01
7.39526570e-01 6.05911911e-01 4.01600748e-01 1.52898937e-01
7.36199081e-01 2.69657761e-01 4.36810791e-01 2.41820619e-01
-1.50318757e-01 -3.05492878e-01 4.90994900e-01 1.07278907e+00
5.70702553e-01 -5.86363524e-02 -1.66208625e+00 4.80746210e-01
-1.20933843e+00 -5.56528986e-01 -2.34963760e-01 1.54358006e+00
8.79510343e-01 2.78561980e-01 -2.42665067e-01 -1.01360366e-01
3.95690590e-01 -7.05545321e-02 -4.45351213e-01 -7.39411354e-01
2.72867560e-01 2.44857118e-01 5.00697970e-01 6.62193120e-01
-7.43206799e-01 1.01479518e+00 6.84547281e+00 4.72699314e-01
-1.25851047e+00 1.37569204e-01 3.19087029e-01 7.49016643e-01
-8.65979373e-01 6.30581260e-01 -5.53609431e-01 5.22404552e-01
1.11318588e+00 -7.44694889e-01 8.80742729e-01 1.23658681e+00
-2.30236322e-01 -5.77844074e-03 -1.33146954e+00 3.87660891e-01
4.67173129e-01 -1.17937374e+00 5.71833923e-02 -6.58083409e-02
6.36422992e-01 -6.09370805e-02 4.84960005e-02 5.27523160e-01
6.55884922e-01 -1.36588597e+00 6.47676826e-01 2.35401109e-01
4.18240935e-01 -3.62465978e-01 5.99154115e-01 5.08662045e-01
-1.13250828e+00 -4.70655501e-01 -2.16648489e-01 -3.29370290e-01
-1.55196026e-01 3.15265477e-01 -1.04995108e+00 3.39409709e-01
1.96636781e-01 8.87530863e-01 -1.37386739e+00 8.04113925e-01
-5.23003757e-01 6.34976089e-01 2.38464981e-01 -2.20809698e-01
-7.81983212e-02 1.94799230e-01 4.33051765e-01 1.18365610e+00
9.36210901e-02 -1.92032725e-01 4.55122352e-01 1.40999651e+00
-4.66564707e-02 -5.87998517e-02 -8.43966246e-01 -8.36136222e-01
4.92554784e-01 8.00846636e-01 -6.40733063e-01 -4.44897413e-01
-7.51852572e-01 6.67541504e-01 3.08495343e-01 2.72193998e-01
-8.28238189e-01 -6.10783041e-01 7.89625525e-01 -1.22793056e-01
-5.85336275e-02 -6.03600442e-01 -6.51247084e-01 -1.41536558e+00
-1.74964845e-01 -1.59194922e+00 -3.95130441e-02 -7.71531940e-01
-1.08972073e+00 5.16577840e-01 1.45230517e-01 -6.71580076e-01
-3.65406603e-01 -7.65404105e-01 -7.55806744e-01 6.69056654e-01
-1.42609298e+00 -8.79279196e-01 -5.31263113e-01 8.02327618e-02
6.28694057e-01 -2.79624999e-01 5.56127787e-01 9.56137553e-02
-3.99263561e-01 6.03551507e-01 -2.54979491e-01 3.52329701e-01
3.06318551e-01 -1.08986211e+00 9.35967445e-01 1.06207001e+00
1.50250895e-02 1.30158973e+00 7.42603660e-01 -1.01326811e+00
-1.70027995e+00 -1.01818514e+00 7.54327714e-01 -8.87921274e-01
1.13710713e+00 -2.66323328e-01 -1.08037448e+00 1.07992923e+00
1.24290004e-01 -6.43103421e-01 6.78637385e-01 7.09670335e-02
-9.54047382e-01 4.90885586e-01 -8.59779477e-01 6.23239338e-01
9.22604501e-01 -1.03372848e+00 -1.03997052e+00 1.50796458e-01
1.12544799e+00 -2.96132922e-01 -9.06780183e-01 1.83754712e-01
3.63301098e-01 -6.49326265e-01 7.91266203e-01 -5.98832548e-01
1.02242935e+00 -3.65228981e-01 -2.87171364e-01 -1.45668089e+00
-2.03498051e-01 -2.31583372e-01 -1.89047288e-02 1.20713830e+00
6.83280647e-01 -7.28956699e-01 5.64582169e-01 7.83209622e-01
6.24254392e-03 -1.36073261e-01 -8.57326686e-02 -6.72200739e-01
3.83222312e-01 -6.04085982e-01 7.53945708e-01 1.03877878e+00
4.36295658e-01 2.10945472e-01 2.57203430e-01 1.56724766e-01
4.18223172e-01 3.54594171e-01 8.48091424e-01 -1.28517401e+00
-7.85814822e-01 -8.07407737e-01 -5.30378222e-01 -1.09460509e+00
9.38362002e-01 -1.29896772e+00 9.21076611e-02 -1.38528287e+00
5.92922628e-01 -5.43032289e-01 3.21553737e-01 6.16615653e-01
-3.14995766e-01 -4.75999378e-02 2.95305908e-01 8.51472765e-02
-1.40931621e-01 7.30761141e-02 5.42789102e-01 -6.60555303e-01
2.62640148e-01 -3.44721556e-01 -1.00706732e+00 8.29382181e-01
8.75666857e-01 -6.63460135e-01 -3.53011072e-01 -7.13966548e-01
4.76657927e-01 1.32606581e-01 3.62116128e-01 -6.39349937e-01
1.55577958e-01 -5.11091888e-01 -1.41281515e-01 1.05004199e-01
-2.17465535e-01 -6.69238806e-01 1.00010950e-02 8.25806618e-01
-3.94702435e-01 3.55820656e-01 1.99368462e-01 3.00648987e-01
-2.86393940e-01 -8.76912177e-01 5.50204873e-01 -3.57770264e-01
-1.04202247e+00 -1.80409446e-01 -6.63973153e-01 3.20212245e-01
9.97123837e-01 -5.62306494e-02 -8.72732997e-01 -2.29384840e-01
-2.81649470e-01 1.59615383e-01 1.31316817e+00 8.80054116e-01
4.80134696e-01 -8.07679474e-01 -3.95594895e-01 3.77943784e-01
4.10837591e-01 -6.75327063e-01 -5.05260587e-01 3.84252965e-01
-7.17158616e-01 4.33406681e-01 -3.22768837e-01 -4.79360610e-01
-1.44876254e+00 7.40927517e-01 2.56030172e-01 -1.69175237e-01
-1.27460033e-01 5.67537367e-01 2.53350705e-01 -7.29656339e-01
-2.06039473e-02 -4.56397772e-01 -1.34947717e-01 -3.46262634e-01
3.75586689e-01 -8.80792364e-02 -1.08985588e-01 -3.57759506e-01
-3.96954596e-01 6.17330909e-01 -9.52613726e-02 2.21721724e-01
1.00404894e+00 6.41070232e-02 -7.66957462e-01 2.03535736e-01
1.22836196e+00 2.91599810e-01 -5.08947611e-01 -3.87580916e-02
6.59442782e-01 -5.83748221e-01 -2.83872753e-01 -9.25210297e-01
-7.16491938e-01 9.21433866e-01 9.57701504e-02 2.41715655e-01
8.48939896e-01 2.85606712e-01 5.31886935e-01 5.55929422e-01
5.18281937e-01 -3.61969084e-01 3.13193738e-01 8.73530686e-01
4.90145326e-01 -1.24833655e+00 -2.31013507e-01 -5.57221711e-01
-4.01579410e-01 1.37169480e+00 8.72114420e-01 2.30196655e-01
5.05877733e-01 7.02923477e-01 8.89048055e-02 -2.73903817e-01
-7.57212818e-01 1.64888531e-01 2.06993714e-01 7.59368956e-01
7.19976425e-01 1.79693401e-01 -3.02845627e-01 4.92658108e-01
-5.37067175e-01 -1.71819746e-01 1.08633471e+00 1.23096466e+00
-4.52704817e-01 -1.58139408e+00 -3.15933168e-01 6.96407676e-01
-3.71737868e-01 -5.56786656e-01 -8.30352426e-01 5.10528743e-01
-6.44072443e-02 1.15270233e+00 -2.57933021e-01 -5.10750949e-01
-1.06069230e-01 1.33111924e-01 3.71658355e-01 -1.20738196e+00
-7.64876306e-01 -8.62651765e-01 1.82220161e-01 -3.71372193e-01
9.55516174e-02 -2.98380703e-01 -1.11146331e+00 -3.81480366e-01
-2.53356159e-01 2.15712577e-01 8.44145954e-01 9.82419729e-01
1.78240061e-01 5.82123697e-01 3.15626740e-01 -2.27809519e-01
-4.07871693e-01 -5.76178074e-01 3.40311944e-01 3.70284349e-01
2.62000680e-01 -3.56939465e-01 -3.21633548e-01 4.31461185e-01] | [7.539938449859619, 7.847000598907471] |
2a1fc522-e719-4248-b4b5-8d3e25ab7ddf | accurate-word-segmentation-and-pos-tagging | null | null | https://aclanthology.org/D14-1011 | https://aclanthology.org/D14-1011.pdf | Accurate Word Segmentation and POS Tagging for Japanese Microblogs: Corpus Annotation and Joint Modeling with Lexical Normalization | null | ['Masaru Kitsuregawa', 'Nobuhiro Kaji'] | 2014-10-01 | null | null | null | emnlp-2014-10 | ['lexical-normalization'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.323692798614502, 3.647221565246582] |
241aace5-40af-4f53-a908-4b03b4513cda | gtnet-graph-transformer-network-for-3d-point | 2305.15213 | null | https://arxiv.org/abs/2305.15213v2 | https://arxiv.org/pdf/2305.15213v2.pdf | GTNet: Graph Transformer Network for 3D Point Cloud Classification and Semantic Segmentation | Recently, graph-based and Transformer-based deep learning networks have demonstrated excellent performances on various point cloud tasks. Most of the existing graph methods are based on static graph, which take a fixed input to establish graph relations. Moreover, many graph methods apply maximization and averaging to aggregate neighboring features, so that only a single neighboring point affects the feature of centroid or different neighboring points have the same influence on the centroid's feature, which ignoring the correlation and difference between points. Most Transformer-based methods extract point cloud features based on global attention and lack the feature learning on local neighbors. To solve the problems of these two types of models, we propose a new feature extraction block named Graph Transformer and construct a 3D point point cloud learning network called GTNet to learn features of point clouds on local and global patterns. Graph Transformer integrates the advantages of graph-based and Transformer-based methods, and consists of Local Transformer and Global Transformer modules. Local Transformer uses a dynamic graph to calculate all neighboring point weights by intra-domain cross-attention with dynamically updated graph relations, so that every neighboring point could affect the features of centroid with different weights; Global Transformer enlarges the receptive field of Local Transformer by a global self-attention. In addition, to avoid the disappearance of the gradient caused by the increasing depth of network, we conduct residual connection for centroid features in GTNet; we also adopt the features of centroid and neighbors to generate the local geometric descriptors in Local Transformer to strengthen the local information learning capability of the model. Finally, we use GTNet for shape classification, part segmentation and semantic segmentation tasks in this paper. | ['Ying He', 'Xinzhe Shi', 'Weiwei Jin', 'Qian Wang', 'Wei Zhou'] | 2023-05-24 | null | null | null | null | ['3d-point-cloud-classification', 'point-cloud-classification'] | ['computer-vision', 'computer-vision'] | [-4.27964061e-01 -1.70564115e-01 3.37857194e-02 -3.37092310e-01
3.32729854e-02 -2.39930823e-01 1.94480926e-01 2.01832235e-01
-2.53733039e-01 1.19765490e-01 -1.00888409e-01 1.10036194e-01
-2.94842929e-01 -1.42999041e+00 -7.05595493e-01 -7.15446770e-01
1.57640040e-01 4.34405565e-01 7.53907502e-01 -3.51756722e-01
2.70831317e-01 8.49465668e-01 -1.25376832e+00 1.07335009e-01
7.96668351e-01 1.12829697e+00 3.84058952e-01 -9.53521281e-02
-7.26200581e-01 3.34281832e-01 -6.46105945e-01 -2.12292727e-02
2.38588542e-01 -1.49401039e-01 -4.30941343e-01 9.07338038e-03
2.69754291e-01 -1.17116749e-01 -6.66755319e-01 1.20171154e+00
5.42966425e-01 2.26295054e-01 4.59455699e-01 -1.45090961e+00
-1.03788221e+00 4.41377342e-01 -6.22207761e-01 2.14668185e-01
3.22125815e-02 2.08719045e-01 8.39323044e-01 -6.73405230e-01
3.36669385e-01 1.33431935e+00 7.66285062e-01 9.00723711e-02
-7.82853663e-01 -9.17451620e-01 7.38516808e-01 3.44270647e-01
-1.54955626e+00 1.95104226e-01 1.28882980e+00 -1.26587838e-01
1.01055419e+00 -1.56815555e-02 1.14996171e+00 3.81792605e-01
1.96569353e-01 6.54485226e-01 4.91310358e-01 7.28444904e-02
-1.68058425e-01 -3.62544209e-01 -1.16745755e-02 9.08825934e-01
5.96310832e-02 -1.18425556e-01 -7.19061047e-02 1.47935271e-01
1.11052489e+00 5.50711155e-01 -3.23050141e-01 -5.68692505e-01
-1.16702020e+00 8.04854333e-01 1.35806191e+00 6.69749856e-01
-2.56731004e-01 2.03630805e-01 1.90593526e-01 2.88924098e-01
4.42655176e-01 1.62614018e-01 -5.50679922e-01 3.92211825e-01
-5.06779730e-01 -4.86178733e-02 2.83919007e-01 1.02788687e+00
1.39854097e+00 -1.66596174e-02 -1.54249668e-01 7.39093006e-01
4.82892931e-01 5.98480523e-01 6.85473859e-01 -3.42402101e-01
5.75160623e-01 1.40861738e+00 -5.94688892e-01 -1.74736249e+00
-5.64135253e-01 -5.14319599e-01 -8.54626119e-01 7.48077706e-02
1.62552763e-03 1.47995323e-01 -1.24375033e+00 1.57993758e+00
3.34764421e-01 4.14307475e-01 -4.12771255e-01 9.53228056e-01
1.28602171e+00 7.62594640e-01 -7.13895485e-02 -1.06486129e-02
8.89172375e-01 -7.95302689e-01 -4.25127149e-01 -1.95185263e-02
3.21901530e-01 -5.64724743e-01 1.01227450e+00 -1.43161163e-01
-7.29354739e-01 -7.75795996e-01 -9.77790058e-01 -2.55519271e-01
-8.14045668e-01 -1.19162098e-01 6.60548568e-01 -6.66171536e-02
-9.90519226e-01 7.58859634e-01 -8.15293133e-01 -1.60121918e-01
5.86539984e-01 7.87511051e-01 -2.35218182e-01 -6.85626045e-02
-1.10738885e+00 6.05989635e-01 3.40326726e-01 3.65774840e-01
-4.24252361e-01 -7.13587165e-01 -9.78260458e-01 3.31645697e-01
1.53294832e-01 -6.39276266e-01 6.16358519e-01 -9.14291859e-01
-1.23712993e+00 6.84177339e-01 3.21588069e-02 1.90838143e-01
1.48223400e-01 1.72706187e-01 -3.70668560e-01 -1.41594922e-02
2.94723332e-01 7.16573060e-01 7.09350705e-01 -1.07315087e+00
-5.39395571e-01 -6.90398753e-01 -4.00998890e-02 4.08869386e-01
-1.93724275e-01 -5.44701278e-01 -8.71173501e-01 -3.83262843e-01
7.67236710e-01 -6.69648528e-01 -2.02567860e-01 -1.82662308e-01
-3.21905315e-01 -6.38038039e-01 1.35040927e+00 -2.08072335e-01
1.01858950e+00 -2.36516809e+00 1.23966277e-01 5.16459584e-01
5.10277569e-01 1.58550203e-01 -3.68584901e-01 1.82454959e-01
-1.31055608e-01 1.50746197e-01 -1.40775889e-01 2.56265819e-01
-2.62821674e-01 3.24430496e-01 6.34764358e-02 2.43819579e-01
2.80657202e-01 1.15695906e+00 -9.15223777e-01 -5.97068369e-01
7.12943733e-01 4.73841816e-01 -4.49189365e-01 -5.69235766e-03
-1.70043126e-01 1.26172081e-01 -1.04229629e+00 3.77413660e-01
1.04264855e+00 -1.56317905e-01 -2.88510889e-01 -7.09390342e-01
-1.30298242e-01 -1.92098022e-02 -1.02188015e+00 1.78037214e+00
-4.22694355e-01 1.03608415e-01 -1.02215342e-01 -1.04548383e+00
1.30721748e+00 -1.02555372e-01 8.01124275e-01 -6.64946496e-01
2.89787292e-01 3.15076858e-02 6.68300465e-02 -3.43734831e-01
-3.21473926e-02 1.09188624e-01 2.19564751e-01 1.02260366e-01
1.71170741e-01 -3.26126724e-01 -2.59717792e-01 1.30187631e-01
1.06592846e+00 -2.83040032e-02 -2.26342469e-01 -1.00379862e-01
6.62971497e-01 -1.51243091e-01 6.22857809e-01 2.70703256e-01
3.21663842e-02 6.83107853e-01 3.71130705e-01 -6.10500634e-01
-6.58552110e-01 -1.09058309e+00 -1.86964869e-02 7.54875481e-01
8.60953271e-01 -3.64482939e-01 -5.28919935e-01 -9.71483409e-01
3.77915531e-01 3.05929720e-01 -6.89507067e-01 -5.59739769e-01
-6.28966391e-01 -6.36701524e-01 7.07474053e-02 4.80854362e-01
1.04803526e+00 -1.36627817e+00 -4.21543382e-02 2.99146652e-01
5.27540557e-02 -8.10497522e-01 -6.29877210e-01 6.43471628e-02
-9.99327600e-01 -1.26193440e+00 -5.53801000e-01 -1.07257354e+00
1.01373065e+00 4.22659039e-01 8.34428012e-01 5.62573910e-01
1.60659142e-02 2.61680692e-01 -3.76098782e-01 -2.31271207e-01
4.14979875e-01 3.58891338e-01 -3.83201510e-01 2.65694894e-02
6.82491422e-01 -9.18801010e-01 -6.17953897e-01 3.97105783e-01
-7.20180869e-01 -1.54853016e-01 5.48229635e-01 7.32398212e-01
7.76136994e-01 2.32050017e-01 3.45739245e-01 -5.09867609e-01
7.05978930e-01 -3.68389547e-01 -5.40270329e-01 6.33084178e-02
-3.45154643e-01 -9.94310156e-02 7.05617249e-01 -2.97892839e-01
-6.37959123e-01 -3.14271897e-02 -1.68600604e-01 -9.41072941e-01
1.34912372e-01 5.74736595e-01 -5.61351299e-01 -5.11377454e-01
1.07973062e-01 3.62803131e-01 -2.78521981e-03 -3.32119614e-01
1.74808979e-01 2.18011886e-01 1.78345874e-01 -3.46987039e-01
9.21495795e-01 2.63122946e-01 3.05598732e-02 -4.25520122e-01
-6.55282557e-01 -3.48382235e-01 -7.71588445e-01 -1.42198369e-01
1.08309126e+00 -6.20956123e-01 -9.05600488e-01 7.33730912e-01
-1.16641808e+00 -2.61400580e-01 -2.68177420e-01 3.56821686e-01
-1.60172492e-01 3.03826898e-01 -5.23066461e-01 -2.19838083e-01
-1.99421629e-01 -1.17197049e+00 1.32743990e+00 5.59350073e-01
4.92010176e-01 -1.05032229e+00 -2.08390996e-01 -2.54362166e-01
2.26849571e-01 2.03754231e-01 9.43161666e-01 -4.05073434e-01
-8.74322236e-01 -6.81674406e-02 -4.91531551e-01 2.27725089e-01
5.39530396e-01 7.30689093e-02 -5.88681757e-01 -2.77963132e-01
3.09857260e-02 2.06142500e-01 8.63574207e-01 5.56356668e-01
1.44927645e+00 -4.20788676e-03 -6.67217612e-01 9.50457096e-01
1.52772903e+00 3.58985007e-01 5.52492321e-01 2.67038852e-01
1.38237178e+00 1.99630648e-01 3.53727192e-01 -1.86134521e-02
3.95575821e-01 3.99599522e-01 6.16315842e-01 -2.81007707e-01
-7.15708286e-02 -6.04932487e-01 -1.86004877e-01 1.06457412e+00
-2.96608090e-01 -4.99376059e-02 -7.93099582e-01 5.16160727e-01
-1.94387329e+00 -7.08230078e-01 -8.47844109e-02 2.12184548e+00
2.10099697e-01 1.93212643e-01 -3.22335869e-01 -1.44504294e-01
1.00327325e+00 2.44187549e-01 -7.44424164e-01 5.56024201e-02
-1.29437059e-01 2.62360305e-01 4.87821937e-01 2.43034646e-01
-9.51419950e-01 1.24355483e+00 5.07006168e+00 9.82835352e-01
-1.45080471e+00 1.34356562e-02 4.93543446e-01 9.92175713e-02
-5.34317851e-01 -8.58905818e-03 -4.57544774e-01 6.06493056e-01
-1.65340513e-01 -1.35797769e-01 4.20478344e-01 9.55265582e-01
-6.52960166e-02 2.98646301e-01 -7.22415209e-01 1.23744309e+00
-6.80741370e-02 -1.16164577e+00 3.84242952e-01 1.11328647e-01
5.86678863e-01 4.82233495e-01 -1.29562289e-01 3.77353013e-01
1.65381372e-01 -8.85113358e-01 2.87167251e-01 5.35510898e-01
5.51240206e-01 -9.67643082e-01 8.24806094e-01 2.28435710e-01
-1.76359892e+00 -8.81885178e-03 -9.22179401e-01 6.68121576e-02
-3.06995418e-02 7.33504713e-01 -3.01541120e-01 7.81333029e-01
8.29961717e-01 1.08522952e+00 -7.23680854e-01 9.98585641e-01
-2.96165615e-01 8.56815726e-02 -6.98077023e-01 -3.45468633e-02
5.13561606e-01 -6.32927477e-01 4.81382489e-01 7.28187144e-01
4.04300839e-01 2.36489803e-01 5.11270761e-01 1.07104623e+00
-1.04206689e-01 3.17561209e-01 -7.55189419e-01 3.13916773e-01
5.89104056e-01 1.29125369e+00 -8.58796179e-01 -3.37675393e-01
-4.19689089e-01 7.52654850e-01 4.33945537e-01 4.86848474e-01
-6.96219862e-01 -7.35193968e-01 4.23108131e-01 1.95666224e-01
5.06789982e-01 -3.47160190e-01 -1.15509667e-01 -1.07619786e+00
8.81455019e-02 -3.10964167e-01 3.37360322e-01 -9.45165575e-01
-1.38528001e+00 6.45977855e-01 -7.65677989e-02 -1.23710227e+00
4.43560004e-01 -5.63252687e-01 -1.12479734e+00 9.43128705e-01
-1.40354979e+00 -1.28913546e+00 -7.08336473e-01 1.00593770e+00
1.19019426e-01 -4.08236198e-02 3.04219157e-01 3.27177018e-01
-3.76055300e-01 4.47688073e-01 -2.49713689e-01 3.73087943e-01
2.80695587e-01 -1.08990812e+00 3.52010459e-01 4.58423197e-01
1.26834400e-02 6.65650606e-01 -1.28205568e-01 -9.00569558e-01
-1.13910115e+00 -1.18395126e+00 5.39096296e-01 -1.49201989e-01
4.87082362e-01 -2.23542765e-01 -1.10862386e+00 7.25852430e-01
-2.24324867e-01 3.84108543e-01 2.48710308e-02 1.97051868e-01
-1.09672561e-01 -4.68200415e-01 -1.10361254e+00 4.30970788e-01
1.39138651e+00 -4.54512507e-01 -5.79477072e-01 4.48192835e-01
1.04451764e+00 -4.99364406e-01 -8.38255346e-01 7.76725650e-01
7.64512569e-02 -9.47375655e-01 8.29151273e-01 -2.07419410e-01
1.21972024e-01 -7.09528625e-01 2.73209006e-01 -1.44298518e+00
-7.94952214e-01 5.32140248e-02 4.83474076e-01 1.28930247e+00
2.25121409e-01 -1.02953815e+00 8.90174985e-01 1.38826638e-01
-5.92789829e-01 -8.23147833e-01 -9.45415378e-01 -5.55735767e-01
5.43297417e-02 -2.88536578e-01 1.29719186e+00 1.06095636e+00
-2.76230484e-01 4.79441613e-01 3.45029801e-01 2.38846764e-01
2.69923747e-01 4.15554941e-01 5.63412070e-01 -1.31009209e+00
2.52913177e-01 -6.73950970e-01 -9.22837913e-01 -1.17579496e+00
1.13864690e-01 -1.19872844e+00 -1.99686825e-01 -1.75867116e+00
1.33344978e-02 -8.01775873e-01 -6.19907439e-01 6.81231678e-01
-2.28267759e-01 8.45332071e-02 3.47469002e-01 1.93038180e-01
-4.53090906e-01 8.82246256e-01 1.91458809e+00 -4.56466734e-01
-3.63792658e-01 -4.56970446e-02 -4.37685400e-01 6.74042761e-01
6.12105787e-01 -2.97146291e-01 -5.94243288e-01 -7.10321128e-01
7.81225413e-02 -3.34808081e-01 4.02720779e-01 -1.16312563e+00
4.77276921e-01 -7.78884813e-02 7.73985147e-01 -7.37124860e-01
8.72478038e-02 -1.03299844e+00 -2.34306082e-02 2.65086353e-01
2.49569789e-01 1.85010284e-01 2.95506030e-01 4.72680509e-01
-4.39403951e-01 3.03941183e-02 5.28290451e-01 -3.95314217e-01
-8.20376337e-01 1.13293707e+00 9.46634412e-02 -1.22851670e-01
9.45719898e-01 -5.07885396e-01 -1.21856384e-01 -1.05214976e-01
-8.10465872e-01 5.01817107e-01 4.46721017e-01 6.46356821e-01
9.05889630e-01 -1.66090035e+00 -3.26608777e-01 4.72303599e-01
-4.81858514e-02 7.02722609e-01 3.76984566e-01 6.08955920e-01
-5.65467596e-01 5.18744029e-02 -3.66316140e-01 -9.01553094e-01
-9.32217062e-01 7.01972187e-01 6.91074908e-01 -8.81622136e-02
-7.48062789e-01 9.07570422e-01 6.48058116e-01 -8.01624656e-01
-2.35670626e-01 -5.01819730e-01 -4.19605970e-01 -1.86483070e-01
-8.31679031e-02 5.08109713e-03 9.82649177e-02 -6.51037693e-01
-5.06534934e-01 1.27360427e+00 1.02277519e-02 3.46717894e-01
1.21558952e+00 9.79874879e-02 -4.15322393e-01 3.53765875e-01
1.39544392e+00 2.97456756e-02 -9.46526825e-01 -1.09182112e-01
-5.58148682e-01 -4.44075584e-01 3.20821136e-01 -4.96239275e-01
-1.85176122e+00 9.35451269e-01 5.09595931e-01 2.71199346e-01
1.18371713e+00 5.47486469e-02 9.59198356e-01 2.13500038e-01
4.07035589e-01 -7.91655540e-01 -3.90459001e-02 6.87933624e-01
8.50528836e-01 -1.02065575e+00 -9.22873318e-02 -6.95888042e-01
-1.80935323e-01 1.12286437e+00 1.17619491e+00 -5.56616187e-01
1.06659746e+00 -7.73358047e-02 -4.09621261e-02 -6.80916190e-01
-2.50403285e-01 -2.96051174e-01 4.06759143e-01 6.80256903e-01
8.28650817e-02 -1.32695466e-01 -1.99927300e-01 4.62022603e-01
-3.26261401e-01 -1.39395803e-01 -1.68193519e-01 5.83064973e-01
-3.97574723e-01 -9.18668032e-01 -6.74493611e-02 6.08261883e-01
1.12864435e-01 -1.37123436e-01 -4.36231405e-01 1.05178583e+00
7.03879058e-01 5.99275589e-01 6.23155296e-01 -7.43664980e-01
5.32947838e-01 -3.74875635e-01 4.46717441e-01 -6.69117391e-01
-6.48609102e-01 1.45873457e-01 -5.41046500e-01 -6.17599010e-01
-3.27464342e-01 -4.10875052e-01 -1.69530869e+00 -3.05331379e-01
-5.18060088e-01 2.79481947e-01 3.60368013e-01 9.62786198e-01
4.53481972e-01 8.03881228e-01 7.53393173e-01 -8.41758072e-01
1.37613013e-01 -9.40742373e-01 -7.01191485e-01 5.23283780e-01
3.81360538e-02 -8.23179364e-01 -4.06686485e-01 -4.73498881e-01] | [7.951376438140869, -3.5528030395507812] |
a7ec5dcd-5812-4857-9906-c0f7ef023693 | a-big-data-intelligence-marketplace-and | 2111.09872 | null | https://arxiv.org/abs/2111.09872v1 | https://arxiv.org/pdf/2111.09872v1.pdf | A big data intelligence marketplace and secure analytics experimentation platform for the aviation industry | The unprecedented volume, diversity and richness of aviation data that can be acquired, generated, stored, and managed provides unique capabilities for the aviation-related industries and pertains value that remains to be unlocked with the adoption of the innovative Big Data Analytics technologies. Despite the large efforts and investments on research and innovation, the Big Data technologies introduce a number of challenges to its adopters. Besides the effective storage and access to the underlying big data, efficient data integration and data interoperability should be considered, while at the same time multiple data sources should be effectively combined by performing data exchange and data sharing between the different stakeholders. However, this reveals additional challenges for the crucial preservation of the information security of the collected data, the trusted and secure data exchange and data sharing, as well as the robust data access control. The current paper aims to introduce the ICARUS big data-enabled platform that aims provide a multi-sided platform that offers a novel aviation data and intelligence marketplace accompanied by a trusted and secure analytics workspace. It holistically handles the complete big data lifecycle from the data collection, data curation and data exploration to the data integration and data analysis of data originating from heterogeneous data sources with different velocity, variety and volume in a trusted and secure manner. | ['Konstantinos Perakis', 'Domenico Messina', 'Fenareti Lampathaki', 'Dimitrios Alexandrou', 'Dimitrios Spyropoulos', 'Stamatis Pitsios', 'Dimitrios Miltiadou'] | 2021-11-18 | null | null | null | null | ['data-integration'] | ['knowledge-base'] | [-6.23346686e-01 -2.88680315e-01 8.51048678e-02 -2.45192200e-01
-1.29047111e-01 -9.28547502e-01 8.65456700e-01 9.68703330e-01
-3.25677395e-01 4.57316041e-01 2.35148929e-02 -3.13519835e-01
-6.37042403e-01 -1.04486752e+00 -1.15581535e-01 -7.59744883e-01
-1.76256225e-01 7.33489394e-01 1.55095845e-01 -5.07429779e-01
-1.25899315e-01 8.02157283e-01 -1.63370299e+00 1.33854628e-01
6.28865361e-01 1.80121779e+00 -3.73314053e-01 1.30548745e-01
1.09989494e-01 6.54418409e-01 -5.99299610e-01 -2.99069434e-01
8.61967564e-01 5.43727517e-01 -6.36321187e-01 -4.38982189e-01
-4.63349462e-01 -2.70951629e-01 4.26181853e-01 8.72362554e-01
2.73800135e-01 -1.77841708e-01 -3.72812152e-01 -1.88990772e+00
-4.01344091e-01 7.61174485e-02 -8.23562518e-02 -1.25176638e-01
9.81430709e-02 4.62901533e-01 3.41013998e-01 -5.42459607e-01
8.41913223e-01 5.31024456e-01 1.49427861e-01 -2.48660535e-01
-5.52503347e-01 -6.06422067e-01 -3.14285696e-01 1.61464334e-01
-1.24172425e+00 -2.19562501e-01 1.95644274e-01 -5.90376556e-01
5.75773001e-01 7.50598311e-01 7.57647455e-01 3.96423161e-01
5.98561049e-01 -2.57343829e-01 1.17868829e+00 -8.34539682e-02
4.34310645e-01 6.04335964e-01 3.39904189e-01 -3.49380225e-02
9.33836102e-01 2.56175637e-01 -4.44573611e-01 -1.99434459e-01
-1.24666527e-01 8.07230353e-01 -1.35252744e-01 -4.76250619e-01
-1.36609864e+00 2.40223914e-01 2.53272027e-01 4.08678204e-01
-9.99534786e-01 -4.69937205e-01 7.10792482e-01 7.92787611e-01
1.69824809e-01 3.11641604e-01 -6.99324727e-01 -2.91065425e-01
-2.71467745e-01 3.00530970e-01 9.82354045e-01 1.26659667e+00
5.93043268e-01 5.11450432e-02 4.44113135e-01 -3.73983562e-01
5.02699077e-01 7.33971953e-01 6.23699948e-02 -4.36524659e-01
5.13265014e-01 1.23784709e+00 3.65919292e-01 -1.15059090e+00
-3.27615857e-01 -2.17738360e-01 -9.44863558e-01 7.62462258e-01
2.72336155e-01 -1.76728308e-01 -4.80929762e-01 9.54992056e-01
1.03517592e+00 -9.66845214e-01 4.50656682e-01 1.06705713e+00
6.75195217e-01 4.16482121e-01 -1.84724271e-01 -5.56065403e-02
1.73719501e+00 -1.05036244e-01 -1.28046393e+00 3.89688700e-01
3.04548353e-01 -8.95651281e-01 5.53246081e-01 4.85957712e-01
-9.54442382e-01 -2.42646307e-01 -1.09236670e+00 -7.96296746e-02
-1.25610566e+00 -8.55955124e-01 4.36009079e-01 6.28608346e-01
-4.45327729e-01 1.38796628e-01 -5.94518542e-01 -1.90112919e-01
2.77978659e-01 5.58898687e-01 -8.06551397e-01 4.87717874e-02
-1.30769396e+00 9.35405254e-01 6.55182600e-01 2.98403859e-01
-7.70311356e-01 -8.96783173e-01 -5.11381865e-01 -1.31830901e-01
3.83725524e-01 -5.08336782e-01 6.04791820e-01 -2.87932038e-01
-7.83873260e-01 4.89698082e-01 6.86950564e-01 -6.74539208e-01
3.91073197e-01 -2.31792957e-01 -8.59014094e-01 -7.81071410e-02
-1.88906491e-01 -4.67885613e-01 2.90761501e-01 -1.11740494e+00
-1.01940107e+00 -9.57312167e-01 -1.48917466e-01 -4.23902750e-01
-1.48009136e-01 4.56498712e-01 3.44559848e-01 -6.38453066e-02
-8.58267993e-02 -8.28642309e-01 -7.91566223e-02 -3.14282566e-01
1.47862434e-01 4.06144053e-01 1.39117122e+00 -6.90181494e-01
1.01832008e+00 -2.33134174e+00 -5.20300940e-02 7.39333630e-01
6.84607923e-01 5.66193581e-01 7.29137897e-01 1.14843225e+00
2.92180330e-02 3.52039963e-01 3.45186055e-01 2.62495577e-01
1.83202893e-01 4.58692387e-02 -6.42573714e-01 4.83401567e-01
-1.36327446e-01 6.54004216e-01 -5.31941652e-01 -7.18071014e-02
4.60351884e-01 4.85650420e-01 -9.87386424e-03 3.86685431e-01
-2.80364156e-02 6.29654348e-01 -6.33238673e-01 1.04647756e+00
9.46380436e-01 -1.54656563e-02 2.51363039e-01 -2.03858271e-01
-7.71776915e-01 -3.18326026e-01 -1.36664224e+00 1.08726549e+00
5.09795323e-02 -1.03454307e-01 9.38597560e-01 -1.65459543e-01
1.13265133e+00 7.47467399e-01 8.31698418e-01 -6.93692803e-01
5.36501646e-01 5.81478715e-01 -1.23592556e-01 -6.20829701e-01
1.04016733e+00 -1.81725845e-01 -4.18980867e-01 7.50513077e-01
-3.18507075e-01 -1.06467560e-01 -3.01197410e-01 2.15319932e-01
7.27412701e-01 -2.87340701e-01 2.03673542e-01 -6.45997077e-02
3.68074656e-01 1.29687190e-01 5.72252452e-01 -3.06036323e-01
-1.91395849e-01 -2.14512557e-01 3.77963990e-01 -8.69847655e-01
-1.08780468e+00 -5.96044898e-01 -3.28483880e-01 3.38177234e-01
1.61965460e-01 -7.54840076e-01 -3.15556973e-01 -3.15751493e-01
2.41190732e-01 3.49568814e-01 -2.63218522e-01 -5.85538223e-02
-1.02180988e-04 -1.75959542e-01 1.75578266e-01 8.42920914e-02
6.38249636e-01 -6.03743970e-01 -1.12235498e+00 -6.96966946e-02
2.81715989e-01 -9.15990829e-01 1.31484717e-01 9.54273418e-02
-4.34593022e-01 -1.25182915e+00 3.86132002e-01 -6.30507991e-02
4.33548391e-01 7.09770545e-02 6.78753018e-01 1.49909496e-01
-1.70030296e-01 -2.58460362e-03 -4.03653145e-01 -1.26305199e+00
-3.67906392e-01 -2.32511789e-01 2.87897438e-01 1.59063950e-01
4.13981766e-01 -1.38862506e-01 -5.66791773e-01 4.02014673e-01
-1.34489536e+00 -2.00051799e-01 1.06718481e-01 3.43780637e-01
6.52662754e-01 5.19555807e-01 8.63360763e-01 -6.69071615e-01
6.10546887e-01 -8.45916808e-01 -9.60382283e-01 2.41542488e-01
-1.11415172e+00 -3.96279037e-01 7.97982156e-01 5.82176983e-01
-6.90888345e-01 -5.08832037e-01 2.28515148e-01 -2.89566666e-01
-3.49884570e-01 6.46579206e-01 -5.15118003e-01 -2.56349266e-01
9.17689055e-02 -3.60242069e-01 8.21050048e-01 -5.34137189e-01
3.26209128e-01 1.14460993e+00 7.77578056e-02 -6.48961589e-02
9.90042925e-01 8.28633904e-01 7.52184764e-02 -3.12113792e-01
-6.34296387e-02 -3.12200278e-01 -6.30097508e-01 -3.61747921e-01
9.42872941e-01 -6.05458796e-01 -1.18368363e+00 4.89709079e-01
-7.45134413e-01 4.12541270e-01 -9.43390429e-01 6.17771268e-01
8.05140287e-02 4.53744717e-02 -1.55221701e-01 -6.45827055e-01
-7.93349028e-01 -1.04565334e+00 4.90395218e-01 -1.52527606e-02
-2.43518010e-01 -4.82599854e-01 -2.05169600e-02 7.17229247e-01
1.17482710e+00 1.08170187e+00 6.92774296e-01 -1.05030465e+00
-1.30323029e+00 -1.14606261e+00 -1.07668608e-01 3.65116179e-01
4.67074603e-01 3.30234677e-01 -6.73645139e-01 -6.54795170e-01
4.43458915e-01 -3.11512709e-01 -3.44470263e-01 -5.28324842e-01
6.49133623e-01 -6.04304194e-01 1.61785990e-01 4.73828554e-01
1.26700592e+00 3.25960934e-01 5.79687953e-01 4.54483062e-01
3.08446735e-01 1.01033187e+00 1.36665654e+00 9.07322466e-01
4.47613060e-01 2.59121537e-01 1.07100511e+00 -3.59270349e-02
6.67789340e-01 1.99496835e-01 -4.00784731e-01 8.71141255e-01
-3.69134188e-01 2.40529045e-01 -9.86883819e-01 1.84717178e-01
-1.92980063e+00 -1.05029655e+00 -3.75309944e-01 2.63432217e+00
3.16929460e-01 -8.37672949e-02 8.46667811e-02 2.06579238e-01
1.73660308e-01 1.77155897e-01 -2.80261129e-01 -3.86647522e-01
7.00062290e-02 -7.51778781e-02 5.62030792e-01 -7.16028064e-02
-5.66488504e-01 4.05014992e-01 5.56368828e+00 5.72463162e-02
-1.17582071e+00 2.51243114e-01 -1.45065919e-01 -2.29671076e-01
-4.93841171e-01 1.54964447e-01 -4.71443951e-01 3.45977038e-01
1.48792362e+00 -8.80609155e-01 3.45205545e-01 7.80658007e-01
2.80680537e-01 8.95730257e-02 -5.80312133e-01 4.46374506e-01
-5.14165461e-01 -1.59360576e+00 -1.88758940e-01 7.54293025e-01
4.97404039e-01 2.78269559e-01 -1.07201830e-01 -3.52730393e-01
1.75103068e-01 -7.04739690e-01 9.83752608e-01 8.40110540e-01
3.15788209e-01 -1.14436340e+00 1.20268488e+00 2.41433710e-01
-1.08361137e+00 -5.25775671e-01 -1.60771206e-01 -1.31030843e-01
4.75258887e-01 6.06054068e-01 -5.46397865e-01 1.29131627e+00
9.25489068e-01 2.17196226e-01 -3.32247347e-01 7.14328051e-01
5.12128353e-01 -7.88612515e-02 -1.44440085e-01 2.74390191e-01
-1.22928722e-02 -6.41772807e-01 3.64657909e-01 2.11097211e-01
2.10108727e-01 -9.39009637e-02 1.27850443e-01 5.64940810e-01
1.99938849e-01 2.58382857e-01 -1.07955396e+00 -6.25208378e-01
5.46111822e-01 1.52960730e+00 -1.74728826e-01 -1.15311548e-01
-3.22460622e-01 -1.95999309e-01 -2.20553622e-01 5.76023459e-02
-4.04039234e-01 -6.47079825e-01 1.07479119e+00 6.32210732e-01
-1.44459829e-01 -7.26523519e-01 -3.45856786e-01 -8.01642060e-01
2.26132572e-01 -1.22182405e+00 7.52641976e-01 -4.23786581e-01
-1.12954903e+00 7.59160280e-01 -1.30642831e-01 -1.61449599e+00
-2.78461933e-01 -3.44864786e-01 -1.96111485e-01 8.16898704e-01
-1.36148202e+00 -1.47649813e+00 -5.87520957e-01 9.07340169e-01
-5.02343059e-01 -5.27382910e-01 1.28375220e+00 4.58689988e-01
-3.55759889e-01 -2.44858071e-01 2.95828998e-01 -1.46861628e-01
5.44873357e-01 -8.80761087e-01 3.28161009e-02 7.63415515e-01
-6.21435106e-01 9.61945534e-01 3.38123858e-01 -8.16036463e-01
-2.28308940e+00 -9.63823915e-01 7.50332236e-01 -5.61306000e-01
1.02439094e+00 -4.52128589e-01 -8.98794472e-01 8.03193927e-01
5.41351855e-01 1.19463913e-02 1.25586522e+00 -5.32984398e-02
-2.10552722e-01 -9.80568171e-01 -1.62365770e+00 1.28567666e-01
5.63733131e-02 -6.31754875e-01 -5.26344776e-01 3.42593908e-01
8.55376959e-01 -1.32227018e-02 -1.72389126e+00 3.40480626e-01
4.72736061e-01 -1.12100792e+00 5.40072083e-01 -8.94464016e-01
-2.43102893e-01 -7.91156054e-01 -3.59848112e-01 -8.38805616e-01
1.05034173e-01 -9.37659979e-01 1.37128696e-01 1.47925055e+00
-5.51685207e-02 -1.53902876e+00 6.57284260e-02 1.63455451e+00
-1.25868231e-01 -3.83177698e-01 -1.02019048e+00 -6.67199671e-01
-5.21512508e-01 -3.85551393e-01 1.80348635e+00 1.07407188e+00
4.54849601e-01 -4.25755739e-01 -2.79544562e-01 4.70256060e-01
8.71586919e-01 5.24568319e-01 1.30655622e+00 -1.41234660e+00
4.74009067e-01 4.07062978e-01 -9.22591925e-01 4.41400349e-01
-7.10376918e-01 -6.96876049e-01 -6.29173636e-01 -1.46266353e+00
-6.31731391e-01 -8.86976004e-01 -2.03427657e-01 3.75751525e-01
7.73696363e-01 -2.23359764e-01 4.94960248e-01 5.73615074e-01
-3.97372007e-01 3.73488724e-01 1.06977642e+00 3.65175545e-01
9.14255232e-02 -6.51042312e-02 -7.11383700e-01 3.78637701e-01
7.17727900e-01 -3.10560167e-01 -2.75016576e-01 -2.74866641e-01
6.26966894e-01 3.45424004e-02 4.39793617e-01 -8.82653773e-01
5.78435779e-01 -5.71932077e-01 -6.16045706e-02 -8.34644139e-01
1.20126158e-01 -2.00996113e+00 1.26582193e+00 4.92845863e-01
2.95070767e-01 5.64515829e-01 -3.21214125e-02 3.55923355e-01
-5.68945050e-01 2.83352464e-01 4.60699320e-01 1.49896935e-01
-3.25704873e-01 5.17678738e-01 1.95727982e-02 -3.29701722e-01
2.06136608e+00 -9.93829444e-02 -1.00413704e+00 5.21657616e-02
-4.35853839e-01 5.96772194e-01 1.02023911e+00 6.35604620e-01
4.30811107e-01 -1.19933021e+00 -4.13998872e-01 1.11988533e+00
3.86852324e-01 3.68970722e-01 4.33926105e-01 8.80889714e-01
-5.12491167e-01 4.42868382e-01 -8.06223631e-01 -2.27232426e-01
-1.16555595e+00 9.48979259e-01 -4.18422446e-02 -5.81874661e-02
-4.89251494e-01 -1.77976891e-01 -7.22278595e-01 -3.36421907e-01
-1.00340554e-02 -2.48581976e-01 1.67339712e-01 4.79160070e-01
9.67901647e-01 6.68448091e-01 4.12046015e-01 -7.39939570e-01
-4.59670663e-01 3.98716480e-02 -3.98811549e-02 1.17031232e-01
1.54042983e+00 -4.27185953e-01 -7.34537065e-01 6.18206263e-01
6.15059137e-01 3.49201024e-01 -5.66303551e-01 2.30731398e-01
-1.13610290e-02 -1.26296616e+00 -7.97202438e-02 -9.14753675e-01
-1.34284997e+00 7.26367235e-01 3.90112728e-01 1.15480375e+00
9.16281402e-01 -1.12985432e-01 7.48906612e-01 -6.39937893e-02
8.17884088e-01 -1.23316550e+00 -5.51191330e-01 1.00204282e-01
1.07191861e+00 -9.30084467e-01 1.76002234e-01 -5.54605238e-02
-8.76966834e-01 8.08573425e-01 1.70049742e-01 4.06015933e-01
1.12684536e+00 5.87096453e-01 3.84061277e-01 -9.15463328e-01
-8.02816868e-01 1.62806615e-01 -2.08562151e-01 6.90500140e-01
-1.48357272e-01 1.78060517e-01 -1.95578843e-01 8.66395056e-01
-2.46745646e-01 4.39760029e-01 3.83921295e-01 1.47882700e+00
-2.11573854e-01 -1.43711114e+00 -5.48675776e-01 2.90478379e-01
-5.90630352e-01 3.72216344e-01 -2.84593105e-01 7.34391928e-01
2.39017904e-01 9.68535721e-01 -1.19750716e-01 -5.24779201e-01
6.70695186e-01 -3.57410870e-04 -7.83617318e-01 -1.15440294e-01
-1.33531880e+00 -2.91056633e-01 2.54379332e-01 -4.68337089e-01
-1.35115638e-01 -4.79074776e-01 -1.63372660e+00 -9.28175390e-01
-1.81222260e-01 6.03366733e-01 1.55889750e+00 4.29707855e-01
1.15056920e+00 3.91789794e-01 8.07381988e-01 -1.28686756e-01
-7.02407479e-01 -3.76548022e-01 -1.17416716e+00 5.00390708e-01
2.34054551e-01 -1.79266423e-01 -2.98746437e-01 -2.57421464e-01] | [5.955741882324219, 6.6802215576171875] |
fe890059-8ea2-4a53-8971-057c6378a84c | reinforced-clarification-question-generation | 2212.10409 | null | https://arxiv.org/abs/2212.10409v3 | https://arxiv.org/pdf/2212.10409v3.pdf | ClarifyDelphi: Reinforced Clarification Questions with Defeasibility Rewards for Social and Moral Situations | Context is everything, even in commonsense moral reasoning. Changing contexts can flip the moral judgment of an action; "Lying to a friend" is wrong in general, but may be morally acceptable if it is intended to protect their life. We present ClarifyDelphi, an interactive system that learns to ask clarification questions (e.g., why did you lie to your friend?) in order to elicit additional salient contexts of a social or moral situation. We posit that questions whose potential answers lead to diverging moral judgments are the most informative. Thus, we propose a reinforcement learning framework with a defeasibility reward that aims to maximize the divergence between moral judgments of hypothetical answers to a question. Human evaluation demonstrates that our system generates more relevant, informative and defeasible questions compared to competitive baselines. Our work is ultimately inspired by studies in cognitive science that have investigated the flexibility in moral cognition (i.e., the diverse contexts in which moral rules can be bent), and we hope that research in this direction can assist both cognitive and computational investigations of moral judgments. | ['Chandra Bhagavatula', 'Yejin Choi', 'Liwei Jiang', 'Ximing Lu', 'Vivek Srikumar', 'Jena D. Hwang', 'Valentina Pyatkin'] | 2022-12-20 | null | null | null | null | ['question-generation'] | ['natural-language-processing'] | [ 4.05848771e-01 8.48306715e-01 1.48102000e-01 -8.34137082e-01
-4.06132281e-01 -6.31486714e-01 6.16562545e-01 1.24135628e-01
-5.69142044e-01 1.05609059e+00 7.21016407e-01 -5.54410458e-01
-1.74337611e-01 -7.93841660e-01 -4.37535226e-01 -5.94001412e-01
4.96059060e-01 4.15856093e-01 -1.12079784e-01 -7.20825076e-01
7.51448750e-01 1.24686353e-01 -1.40464103e+00 4.61658150e-01
1.20362902e+00 4.47128236e-01 -2.31263608e-01 3.39909166e-01
4.00605351e-01 1.37024236e+00 -4.36067581e-01 -1.20888448e+00
-9.92363542e-02 -8.17482233e-01 -1.52568185e+00 -4.08297688e-01
3.61231267e-01 -7.86757529e-01 2.62016356e-01 1.07086575e+00
3.00659597e-01 3.29220921e-01 9.73386705e-01 -1.22382116e+00
-8.89256656e-01 8.09013426e-01 1.05583645e-01 1.48602739e-01
7.79953659e-01 5.29463410e-01 1.33132553e+00 -1.43130496e-01
6.21344984e-01 1.67007041e+00 2.47514978e-01 1.27270412e+00
-1.24216998e+00 -4.55550969e-01 2.50383224e-02 5.68189383e-01
-3.29835653e-01 -4.33042169e-01 8.78096342e-01 -5.45423925e-01
5.52488565e-01 4.15418029e-01 7.06417978e-01 1.47631860e+00
6.37298286e-01 5.62657356e-01 1.50055027e+00 -8.64217058e-02
6.20479286e-01 -1.77378319e-02 1.82167187e-01 2.16215298e-01
2.71466762e-01 4.95192915e-01 -5.29233515e-01 -5.06444871e-01
2.56536365e-01 -6.22909427e-01 2.69980635e-02 -3.21646005e-01
-1.01250279e+00 1.13480473e+00 3.65602165e-01 1.01408977e-02
-5.87910831e-01 3.33841830e-01 4.50470448e-01 4.36497808e-01
-1.84896603e-01 1.16786087e+00 -1.88251406e-01 -3.02625895e-01
-2.57634073e-01 8.40186417e-01 9.94031370e-01 1.92396998e-01
3.76568228e-01 -3.19609433e-01 -4.44719076e-01 5.93653917e-01
4.15214121e-01 2.93105453e-01 -3.11214421e-02 -1.92030466e+00
2.76873466e-02 4.05097127e-01 6.10626996e-01 -1.00575173e+00
-5.09697258e-01 1.74335957e-01 -2.50504822e-01 3.87956738e-01
5.50922155e-01 -4.66132522e-01 2.22709522e-01 2.11114192e+00
5.81488609e-01 -4.65987056e-01 3.71906936e-01 1.07436240e+00
7.28494704e-01 1.86647698e-01 5.56957006e-01 -1.14803039e-01
1.52423155e+00 -9.51619893e-02 -6.74961686e-01 -4.83809739e-01
7.32499063e-01 -4.80275154e-01 1.30002904e+00 2.15288803e-01
-1.17023253e+00 2.48648182e-01 -6.19999170e-01 -3.69990796e-01
9.74817127e-02 -4.82574195e-01 8.80352020e-01 6.57259583e-01
-7.90502429e-01 7.29547143e-01 -7.27859065e-02 -5.04633188e-01
5.74309170e-01 -2.14125782e-01 -3.40263434e-02 3.28574955e-01
-1.84261358e+00 1.45036507e+00 3.56146097e-01 2.49167625e-02
-7.20732391e-01 -3.10608476e-01 -6.66092217e-01 1.21328220e-01
6.51675999e-01 -8.46018612e-01 1.38181615e+00 -1.18595207e+00
-1.38882542e+00 1.34285498e+00 1.77505732e-01 -6.99616313e-01
6.66637957e-01 -3.53317946e-01 -1.19601980e-01 3.45289111e-01
3.96092474e-01 1.11634278e+00 5.13972104e-01 -1.06019568e+00
-4.48432267e-01 -2.63514012e-01 7.86928892e-01 5.79721332e-01
1.32832602e-01 4.11466092e-01 1.02358615e+00 -4.21075344e-01
-4.14185315e-01 -9.92041171e-01 -7.35909641e-02 -3.27857807e-02
-3.08400542e-01 -7.33172119e-01 2.17529118e-01 -3.57657731e-01
7.35537052e-01 -1.80898416e+00 -3.24551195e-01 -1.60277203e-01
1.37383426e-02 -9.09139216e-02 2.00960517e-01 4.52285826e-01
-4.62040082e-02 3.26079369e-01 -2.56768107e-01 7.29136050e-01
5.29657900e-01 2.15299278e-01 -5.60350895e-01 2.96530157e-01
1.95687890e-01 9.59218204e-01 -1.08433223e+00 -7.20253170e-01
-1.28580987e-01 5.65192588e-02 -1.08329046e+00 2.33097762e-01
-4.00294155e-01 2.94190705e-01 -3.96629721e-01 2.04505742e-01
3.15808922e-01 3.15252505e-02 3.18942547e-01 1.75961226e-01
2.95070559e-01 7.65385091e-01 -5.74132025e-01 8.14350605e-01
-5.03994413e-02 6.34768069e-01 -5.03900833e-02 -7.80405283e-01
8.01299334e-01 2.33291641e-01 -2.66590565e-01 -5.52954972e-01
3.16641271e-01 1.96517576e-02 3.54953855e-01 -7.44305909e-01
4.90088999e-01 -8.68649542e-01 -3.88075471e-01 9.43393707e-01
-5.82159519e-01 -6.89227521e-01 5.37081286e-02 5.03568828e-01
7.24275649e-01 6.38571620e-01 5.97178280e-01 -7.10797191e-01
5.53662121e-01 -7.68726179e-03 8.86946440e-01 8.58842313e-01
-8.66761863e-01 -2.07549438e-01 1.31660199e+00 -7.57515132e-01
-7.35096812e-01 -1.08043587e+00 -1.97323069e-01 1.43040490e+00
3.59992206e-01 2.64773846e-01 -1.11486268e+00 -8.42907608e-01
1.49135645e-02 1.92372239e+00 -9.27244842e-01 -5.73472202e-01
-5.37009299e-01 -4.45053786e-01 5.76598048e-01 2.46215686e-01
4.46416497e-01 -1.70776236e+00 -1.38241875e+00 -8.21256265e-02
-7.02091873e-01 -6.48725092e-01 -3.55479211e-01 8.70433673e-02
-6.36036694e-01 -1.37637877e+00 9.78635177e-02 -3.62728477e-01
4.27037477e-01 -1.83153644e-01 1.42668903e+00 2.84513265e-01
2.34850153e-01 5.95988452e-01 -3.18407297e-01 -3.87244523e-01
-8.49293411e-01 -6.97597861e-01 -1.15945376e-02 -3.61713767e-01
9.55535889e-01 -5.18907607e-01 -7.95810342e-01 4.11948621e-01
-8.23791444e-01 -7.61971846e-02 -1.22725569e-01 8.07955384e-01
-2.12939814e-01 -5.33481002e-01 1.10057843e+00 -1.07147133e+00
1.48940969e+00 -7.22920895e-01 -2.49138042e-01 3.24101985e-01
-6.15074933e-01 1.79830894e-01 6.80407882e-01 9.54143777e-02
-1.63157892e+00 -5.45533836e-01 3.14949080e-02 7.16943860e-01
-5.62559187e-01 -3.96288075e-02 -8.81152079e-02 2.92798460e-01
1.21299839e+00 -1.94032609e-01 -1.15161024e-01 2.14908093e-01
4.72753108e-01 4.11527485e-01 3.96010101e-01 -1.45690143e+00
4.33772594e-01 2.78590977e-01 -3.37098539e-02 -3.76571625e-01
-1.07155514e+00 2.06582442e-01 -6.92868978e-02 -6.24746382e-01
1.03400326e+00 -4.15473908e-01 -1.57786679e+00 2.47698482e-02
-1.15152955e+00 -3.71843249e-01 -2.18672127e-01 2.24889785e-01
-1.24769962e+00 5.51197648e-01 -4.79578316e-01 -8.66392732e-01
-2.15164587e-01 -9.98811424e-01 5.49270034e-01 3.06560993e-01
-1.01119125e+00 -8.10968518e-01 -8.71094167e-02 8.55256319e-01
2.13618472e-01 2.62541533e-01 1.20440578e+00 -7.94543564e-01
-4.67908941e-02 4.59046185e-01 3.33241746e-02 2.29025185e-01
-1.56065017e-01 -3.47557366e-02 -6.52265847e-01 2.65168875e-01
3.54620159e-01 -1.12382555e+00 5.03342450e-01 1.95155457e-01
8.26738477e-01 -8.09366047e-01 1.72230989e-01 5.35590202e-02
8.42362225e-01 3.61530244e-01 9.14779186e-01 5.07606208e-01
-1.58868909e-01 1.37417173e+00 9.91801500e-01 4.64746863e-01
8.85229230e-01 3.64190757e-01 5.10491848e-01 5.84600806e-01
5.37388146e-01 -2.52982289e-01 4.97537076e-01 -5.47944248e-01
5.02122082e-02 6.82276636e-02 -1.00700581e+00 4.69974995e-01
-1.73415494e+00 -1.62446856e+00 1.71882603e-02 1.97473538e+00
1.31345463e+00 -1.55366911e-02 3.20725851e-02 -3.10582936e-01
7.83899724e-01 2.30425939e-01 -8.14366579e-01 -1.33525264e+00
-2.78776735e-02 -3.58264595e-01 -3.16526949e-01 6.95344567e-01
-9.10251975e-01 1.15394902e+00 6.19728327e+00 2.54263997e-01
-3.19439828e-01 -3.02425236e-01 1.07528210e+00 -1.20390385e-01
-9.07334447e-01 3.19485158e-01 -2.04883575e-01 2.98388690e-01
5.69977701e-01 -4.00724590e-01 4.20780331e-01 8.37253928e-01
3.69415849e-01 -4.78077084e-01 -1.61651289e+00 3.45535845e-01
-1.37639448e-01 -9.65685129e-01 -7.45936856e-02 -2.04240635e-01
4.85866219e-01 -7.18374014e-01 -1.32786622e-02 3.45569789e-01
9.84475195e-01 -1.06784451e+00 7.61450231e-01 5.34073830e-01
1.27583236e-01 -8.46080780e-01 6.89445734e-01 6.51500821e-01
9.08430442e-02 -7.79502094e-02 -2.72812456e-01 -5.56849778e-01
1.01555921e-01 -1.04152772e-03 -9.04413223e-01 -2.81075835e-01
6.62499189e-01 2.12362140e-01 -1.67898953e-01 2.31307954e-01
-8.76079082e-01 5.37074804e-01 6.26122653e-02 -6.74238622e-01
2.92419165e-01 -3.58296305e-01 5.50087631e-01 7.67862797e-01
-3.58156741e-01 7.43830562e-01 -3.88131887e-01 1.32695270e+00
-3.32670137e-02 1.19907424e-01 -6.50624573e-01 2.73569316e-01
4.45504934e-01 9.68674481e-01 -4.19332355e-01 -1.99791059e-01
1.32425547e-01 6.59197807e-01 2.71283656e-01 2.94713199e-01
-8.53887856e-01 -5.14186136e-02 8.04174781e-01 -7.14591369e-02
-3.94728392e-01 5.54854333e-01 -4.50956821e-01 -1.05843949e+00
-1.33974493e-01 -1.04300535e+00 4.76717681e-01 -1.17677379e+00
-1.35320711e+00 3.12194712e-02 5.91626391e-02 -5.49048066e-01
-5.41225433e-01 -4.26195264e-01 -8.03390443e-01 6.35998011e-01
-1.18134344e+00 -6.72573686e-01 1.18343078e-01 1.96760312e-01
7.16317073e-02 3.99650723e-01 7.51167357e-01 -6.02597296e-01
-2.94192582e-02 3.14718246e-01 -4.51784581e-01 9.62040946e-02
9.78899658e-01 -1.30721080e+00 2.12314054e-01 5.19918859e-01
-5.46782792e-01 8.25859606e-01 1.09307086e+00 -7.21169531e-01
-7.75747299e-01 -3.29255044e-01 1.19294417e+00 -7.56444097e-01
5.21836519e-01 5.21324798e-02 -8.98132384e-01 7.06316650e-01
4.99260783e-01 -6.49051487e-01 1.11706936e+00 2.22637340e-01
-5.39723039e-01 2.49470800e-01 -1.73114848e+00 1.27543688e+00
9.81912255e-01 -5.26848614e-01 -1.56937778e+00 3.30341041e-01
6.48348033e-01 8.31571594e-02 -4.74309564e-01 -3.23869810e-02
6.29205167e-01 -1.56276000e+00 8.44137609e-01 -1.25706673e+00
1.11615312e+00 -3.05406749e-02 -1.82736948e-01 -1.28821778e+00
-4.27240103e-01 -6.38316453e-01 4.38339293e-01 9.62839484e-01
3.07644993e-01 -7.97832668e-01 6.62676930e-01 1.49678898e+00
2.25785553e-01 -7.42612004e-01 -1.09552264e+00 -1.69742957e-01
8.09713125e-01 -2.65676886e-01 7.71940410e-01 1.03031445e+00
7.19559669e-01 3.58447015e-01 -3.62408072e-01 -2.66793638e-01
7.80795872e-01 2.37292066e-01 3.72676820e-01 -1.17305267e+00
3.60300317e-02 -6.35516584e-01 1.20254047e-01 -6.17941499e-01
6.30608261e-01 -7.77792871e-01 1.81749508e-01 -1.43055105e+00
3.17684054e-01 -1.66872516e-01 8.94417614e-02 4.58880782e-01
-6.17949069e-01 -3.93416464e-01 3.55491668e-01 -1.34729877e-01
-6.74306750e-01 7.21661985e-01 1.34542584e+00 1.64482117e-01
4.83326763e-02 -3.11670065e-01 -1.55095637e+00 1.22103083e+00
1.14817297e+00 -3.52186739e-01 -3.06968153e-01 -1.30620506e-03
8.68164182e-01 3.63684237e-01 8.04877877e-01 -3.00632715e-01
-1.34667847e-02 -1.02372217e+00 1.64266437e-01 -5.67283109e-02
1.33854523e-01 -6.46607280e-01 -3.53374958e-01 6.30719244e-01
-1.23117137e+00 -2.24314809e-01 -2.00168341e-01 3.26441854e-01
2.18834952e-01 -5.50580978e-01 9.95122015e-01 -2.84477919e-01
-3.93685967e-01 -4.63681847e-01 -6.87561154e-01 7.45140612e-01
1.22990525e+00 -2.08860517e-01 -6.94446862e-01 -8.49329531e-01
-5.13312936e-01 4.45930272e-01 6.79828405e-01 3.15694660e-01
6.37614369e-01 -1.20309019e+00 -1.05424821e+00 -6.93676114e-01
1.16571926e-01 -5.95683217e-01 3.00723433e-01 4.31098521e-01
-4.46339011e-01 2.38919556e-01 -4.46373791e-01 1.50740713e-01
-9.23578143e-01 4.40331519e-01 6.37840152e-01 1.13474809e-01
-2.58948386e-01 9.27232742e-01 5.24463832e-01 -5.88904560e-01
-1.56065553e-01 5.79961874e-02 -6.32613063e-01 2.84236163e-01
4.36632484e-01 4.39086020e-01 -6.75220788e-01 -5.28400481e-01
-6.07196271e-01 7.45219067e-02 -1.20862342e-01 -1.83600411e-01
1.07224071e+00 -1.05467983e-01 -4.20541286e-01 1.56799391e-01
4.81165439e-01 -8.19963962e-02 -1.30829358e+00 1.57468051e-01
5.11950674e-03 -5.96154690e-01 -4.47542012e-01 -1.32096469e+00
-2.27637544e-01 7.74381995e-01 -1.38317764e-01 3.68583500e-01
9.09246147e-01 5.00863977e-02 3.79238486e-01 7.20533669e-01
5.35393298e-01 -1.57758403e+00 1.30335227e-01 3.73482406e-01
1.28521585e+00 -1.30042195e+00 -2.86467895e-02 1.83465257e-02
-1.37804520e+00 1.13917184e+00 9.18868721e-01 -8.97300914e-02
4.49094810e-02 -1.96800888e-01 1.86223328e-01 -2.49893025e-01
-1.22552431e+00 2.58643746e-01 -1.74556643e-01 6.77386463e-01
6.59684539e-01 5.22419691e-01 -9.81671393e-01 5.78461826e-01
-6.72929108e-01 -9.87847894e-02 1.11746085e+00 6.67974949e-01
-6.46201372e-01 -8.14026117e-01 -6.38845205e-01 5.30157268e-01
-4.88197803e-01 -2.71594048e-01 -1.09465241e+00 2.93824345e-01
-1.18445314e-01 1.35304952e+00 -3.15932482e-01 4.31846306e-02
1.12788863e-01 1.83573171e-01 2.99967974e-01 -3.04045737e-01
-9.72348452e-01 -5.87183535e-01 5.68364382e-01 -8.90103817e-01
-3.95385921e-01 -1.09410095e+00 -1.43966770e+00 -6.00681961e-01
6.99377954e-02 3.04963231e-01 -4.44292389e-02 1.03031850e+00
1.02032706e-01 5.52759506e-02 3.86948526e-01 4.06293049e-02
-1.00833058e+00 -4.70521003e-01 -2.59317547e-01 7.78383911e-01
1.63787112e-01 -5.74129105e-01 -6.61490083e-01 -1.37969866e-01] | [11.551856994628906, 7.9893412590026855] |
d09fd09e-882e-4f93-a43c-e27e4501e193 | data-aware-low-rank-compression-for-large-nlp | null | null | https://openreview.net/forum?id=_sSHg203jSu | https://openreview.net/pdf?id=_sSHg203jSu | Data-aware Low-Rank Compression for Large NLP Models | The representations learned by large-scale NLP models such as BERT have been widely used in various tasks. However, the increasing model size of the pre-trained models also brings the efficiency challenges, including the inference speed and the model size when deploying the model on devices. Specifically, most operations in BERT consist of matrix multiplications. These matrices are not low-rank and thus canonical matrix decomposition could not find an efficient approximation. In this paper, we observe that the learned representation of each layer lies in a low-dimensional space. Based on this observation, we propose DRONE (data-aware low-rank compression), a provably optimal low-rank decomposition of weight matrices, which has a simple closed form solution that can be efficiently computed. DRONE is generic, could be applied to both fully-connected and self-attention layers, and does not require any fine-tuning or distillation steps. Experimental results show that DRONE could improve both model size and inference speed with limited loss of accuracy. Specifically, DRONE alone achieves 1.92x faster on MRPC task with only 1.5%loss of accuracy, and when combined with distillation, DRONE achieves over 12.3x faster on various natural language inference tasks. | ['Cho-Jui Hsieh', 'Inderjit S Dhillon', 'Hsiang-Fu Yu', 'Patrick Chen'] | 2021-01-01 | null | null | null | null | ['low-rank-compression'] | ['computer-code'] | [-1.68627545e-01 3.22763398e-02 -1.85250312e-01 -3.56611103e-01
-6.22620821e-01 -4.63607162e-01 3.30369174e-01 1.12729799e-02
-5.46194136e-01 5.21097243e-01 1.54882789e-01 -4.92298543e-01
-3.01253319e-01 -8.59390974e-01 -1.10748172e+00 -3.72135669e-01
3.14812176e-02 6.97812974e-01 -1.79597273e-01 -1.07978567e-01
-1.26475736e-01 3.63791347e-01 -1.37269330e+00 4.17260498e-01
9.93938804e-01 1.09391892e+00 5.58046937e-01 6.41959488e-01
3.40785906e-02 8.01803052e-01 -3.70304167e-01 -6.35100842e-01
3.14072490e-01 3.54917884e-01 -6.37952030e-01 -4.36297119e-01
7.62262464e-01 -7.74785876e-01 -9.09376323e-01 8.68959308e-01
2.80944675e-01 2.43062004e-01 4.81416434e-01 -1.20103621e+00
-5.33397377e-01 8.67999613e-01 -3.84019345e-01 2.82086432e-01
-1.20205410e-01 -4.00438197e-02 1.28708768e+00 -1.08046067e+00
2.17746094e-01 1.45059896e+00 5.48461318e-01 4.18329149e-01
-1.20868421e+00 -9.64251339e-01 2.26087287e-01 2.92665929e-01
-1.55341530e+00 -5.10963321e-01 3.01387906e-01 1.79539938e-02
1.28743351e+00 2.83436388e-01 5.28306127e-01 1.04401791e+00
-8.62441137e-02 1.06023526e+00 6.33216918e-01 -5.64954430e-02
5.63240275e-02 -1.78275973e-01 2.14808270e-01 1.02146804e+00
7.71676242e-01 -3.51281077e-01 -5.84593832e-01 -1.72129557e-01
6.74674809e-01 3.65735352e-01 -1.65942326e-01 -2.15263255e-02
-1.18446279e+00 8.47117722e-01 5.92742920e-01 7.74230510e-02
-3.06786448e-01 7.48129308e-01 4.10952061e-01 8.30439702e-02
2.59094834e-01 3.65557492e-01 -7.53594160e-01 -3.50517333e-01
-8.63607466e-01 2.74268240e-01 8.19623590e-01 1.09602129e+00
7.95981467e-01 1.88501067e-02 -2.16110647e-02 8.68377030e-01
1.97569340e-01 8.10057640e-01 6.55424953e-01 -9.59555805e-01
1.08792007e+00 5.48605144e-01 -1.97534263e-01 -1.33930993e+00
-3.81736606e-01 -6.15847349e-01 -1.22515440e+00 -6.64178729e-01
2.31142625e-01 -2.38182515e-01 -5.47361076e-01 1.74048269e+00
8.54180828e-02 4.29596186e-01 -1.25887068e-02 6.77054644e-01
4.98114288e-01 9.61371660e-01 -1.98295996e-01 1.04188375e-01
1.45761192e+00 -1.21836650e+00 -7.55371451e-01 -6.74882233e-01
8.47447217e-01 -5.16318500e-01 1.27679515e+00 5.66125870e-01
-1.22928989e+00 -4.62545723e-01 -1.23662329e+00 -6.62625611e-01
-3.16865921e-01 5.29744327e-01 1.12839413e+00 3.26521426e-01
-7.42275417e-01 7.18114555e-01 -1.05683494e+00 1.55725732e-01
5.66213667e-01 4.76986885e-01 -2.29973480e-01 -5.64446449e-01
-1.14856970e+00 7.42675066e-01 5.02096832e-01 2.96081781e-01
-7.11778939e-01 -9.41508651e-01 -7.78318226e-01 5.70860267e-01
5.09537637e-01 -8.16993356e-01 1.12436724e+00 -1.48891434e-01
-1.29001963e+00 2.50359088e-01 -5.06824195e-01 -9.11711276e-01
1.92262456e-01 -8.62813056e-01 -2.30921268e-01 1.55238554e-01
-2.27855876e-01 5.61628222e-01 7.68146694e-01 -5.73873580e-01
-5.68688691e-01 -3.66969615e-01 3.47009033e-01 1.12701379e-01
-8.34865689e-01 -4.47801203e-01 -9.31928933e-01 -5.40628552e-01
6.94077760e-02 -1.06247890e+00 -2.19217494e-01 1.12345800e-01
-1.34906650e-01 -3.73696417e-01 5.95620334e-01 -6.18227839e-01
1.62285686e+00 -2.12877941e+00 -4.85924669e-02 1.85955688e-01
4.91187662e-01 5.16903639e-01 -2.29441658e-01 3.02512407e-01
3.87844741e-01 1.42839193e-01 -1.19444281e-01 -3.84994268e-01
1.79014102e-01 5.66615999e-01 -6.47546053e-01 2.82584637e-01
1.14026755e-01 1.10580695e+00 -9.39760566e-01 -2.79392689e-01
6.44208565e-02 6.39899015e-01 -1.08116007e+00 7.48707205e-02
-2.27357283e-01 -4.51280594e-01 -2.55985647e-01 2.66513407e-01
7.89103687e-01 -5.87577045e-01 3.75914991e-01 -5.17233431e-01
3.27513576e-01 7.23580420e-01 -1.18631256e+00 1.85179377e+00
-1.07905829e+00 8.18606853e-01 -4.34156470e-02 -1.00153053e+00
5.87116182e-01 9.22411028e-03 6.00143783e-02 -4.97393966e-01
7.06288293e-02 2.17443913e-01 2.12427136e-02 -2.74618745e-01
4.73164052e-01 3.19449395e-01 1.50000691e-01 5.27219951e-01
1.57373592e-01 1.61412671e-01 5.05720079e-01 4.82565016e-01
1.11345220e+00 -4.22602981e-01 1.79009885e-01 -8.93447772e-02
2.78772444e-01 -4.41115379e-01 5.05237758e-01 7.23014534e-01
4.31765586e-01 3.14364016e-01 4.17701215e-01 -3.74744982e-01
-7.68502891e-01 -9.20954108e-01 -7.90990964e-02 1.18581498e+00
-2.10224912e-01 -1.06495965e+00 -6.31306350e-01 -4.12719071e-01
7.47877061e-02 6.47146881e-01 -2.57774651e-01 -4.28340167e-01
-7.16258645e-01 -8.26847553e-01 6.63055658e-01 7.59985089e-01
7.60708332e-01 -4.85249311e-01 -3.97985190e-01 2.66448706e-01
-3.72227520e-01 -1.40325856e+00 -5.10492027e-01 6.76997602e-02
-1.28170526e+00 -1.07972336e+00 -2.27873236e-01 -5.74368596e-01
8.85552168e-01 5.93961358e-01 1.03629541e+00 2.89766371e-01
-1.64982632e-01 4.10456210e-02 -5.47149219e-02 -3.73139083e-01
1.62760183e-01 3.82421285e-01 1.55360833e-01 -6.89930320e-02
4.48228896e-01 -7.81294525e-01 -5.43958247e-01 2.18025912e-02
-8.63417327e-01 8.44981968e-02 8.61408889e-01 9.28729534e-01
5.10475695e-01 2.99253404e-01 3.03269923e-01 -1.01863337e+00
6.29266381e-01 -3.95114362e-01 -6.68209612e-01 2.76652485e-01
-5.81721187e-01 5.01669765e-01 1.01153755e+00 -4.67851251e-01
-8.31854701e-01 2.28877068e-02 -1.36536956e-01 -6.45597875e-01
4.46716905e-01 7.65784204e-01 -8.82112086e-02 3.64591897e-01
5.24352670e-01 2.19700530e-01 -9.79473218e-02 -8.19226682e-01
6.90230966e-01 6.36355102e-01 5.09230196e-01 -7.76954532e-01
9.39152956e-01 4.89896178e-01 -1.17454425e-01 -8.46517086e-01
-1.28017962e+00 -3.54580104e-01 -3.66631687e-01 4.54557240e-01
3.41894627e-01 -1.44103312e+00 -9.15903389e-01 -1.88260917e-02
-1.17727077e+00 -2.73346722e-01 -9.84571036e-03 6.89863861e-01
-2.83971161e-01 3.03298950e-01 -7.23002136e-01 -5.51741421e-01
-8.53616774e-01 -8.36433947e-01 8.61607194e-01 -1.44923359e-01
-8.87675583e-02 -7.46546745e-01 -4.33423609e-01 6.35888815e-01
4.25981581e-01 -3.88590008e-01 1.16102290e+00 -4.89178032e-01
-7.77062654e-01 -3.27859104e-01 -4.90867823e-01 5.02985954e-01
-2.62336820e-01 -2.33904451e-01 -9.58060682e-01 -5.71511447e-01
-1.09831654e-01 -5.13529718e-01 9.95033681e-01 6.84701502e-02
1.83277595e+00 -7.74089396e-01 -3.20052385e-01 8.82136226e-01
1.29669058e+00 -2.52647728e-01 4.64792639e-01 -1.72427356e-01
9.45380270e-01 5.81120811e-02 5.03636718e-01 5.00757575e-01
4.36946660e-01 5.09517074e-01 3.61743093e-01 1.60709754e-01
-1.26150262e-03 -6.87123179e-01 4.49366063e-01 1.34496951e+00
-1.00817665e-01 -1.45619303e-01 -7.93311596e-01 4.16740984e-01
-1.97154999e+00 -7.31019974e-01 1.03951871e-01 2.20127392e+00
9.18809116e-01 1.47332951e-01 -2.69613415e-01 2.60925293e-01
3.60960841e-01 1.64370477e-01 -7.10159361e-01 -3.39947075e-01
1.21202193e-01 4.48327214e-01 7.41712451e-01 4.22898769e-01
-8.53729606e-01 9.73001897e-01 5.99629021e+00 1.15032363e+00
-9.76451576e-01 1.78889185e-01 3.93052757e-01 -5.89940131e-01
-1.23731606e-01 -1.62163287e-01 -1.12481427e+00 4.47511345e-01
1.40631986e+00 -2.84597844e-01 9.17809188e-01 1.21080256e+00
-5.34592718e-02 4.77620751e-01 -1.33029997e+00 1.34301305e+00
-2.53446847e-02 -1.59088266e+00 3.37701529e-01 1.88885361e-01
6.17305577e-01 4.13357258e-01 3.30041628e-03 8.16344559e-01
2.92536110e-01 -1.02157199e+00 3.52190793e-01 1.64819077e-01
7.45494902e-01 -9.38773572e-01 5.31682491e-01 7.13103890e-01
-1.19279218e+00 -3.42259377e-01 -1.12612510e+00 -2.78361648e-01
-4.09563035e-02 8.13465893e-01 -7.31704116e-01 1.45941421e-01
6.67678416e-01 6.93726540e-01 -4.64749575e-01 6.22031271e-01
-2.59077907e-01 7.64001608e-01 -7.65762269e-01 -1.58025220e-01
2.12377265e-01 -1.57997057e-01 4.99013588e-02 1.17017913e+00
4.13559198e-01 1.65895566e-01 -3.94452037e-03 6.73089385e-01
-5.77746212e-01 2.05390397e-02 -4.82621491e-01 -2.85489053e-01
8.27840269e-01 1.17875481e+00 -5.14993183e-02 -5.58055460e-01
-5.10242999e-01 8.70630801e-01 8.92069340e-01 1.95477322e-01
-1.08531392e+00 -3.84071410e-01 9.22176480e-01 2.25361630e-01
4.84328955e-01 -4.23864394e-01 -9.19706598e-02 -1.55005884e+00
2.51089394e-01 -9.47748840e-01 2.98288703e-01 -6.56594038e-01
-1.07374454e+00 3.28811824e-01 1.78415552e-02 -8.60856354e-01
-1.53876647e-01 -7.10084796e-01 -2.52779573e-01 4.52788949e-01
-1.48095322e+00 -8.05368125e-01 -1.78776845e-01 6.23379767e-01
4.83170837e-01 -1.20389231e-01 8.94877732e-01 7.28308499e-01
-8.75209570e-01 1.05896878e+00 4.61025774e-01 2.13115707e-01
3.17752898e-01 -1.03888011e+00 5.14654458e-01 7.92450249e-01
5.01566529e-01 1.24986613e+00 2.49700665e-01 -2.62918919e-01
-2.04979563e+00 -1.37298632e+00 1.11930120e+00 -1.57005280e-01
7.53668129e-01 -6.56355381e-01 -7.86984265e-01 8.40561211e-01
-1.10737041e-01 2.22392410e-01 6.24246478e-01 4.56289440e-01
-6.04452312e-01 -4.85589474e-01 -8.04929972e-01 7.57250488e-01
1.35732722e+00 -7.08497047e-01 -4.93984401e-01 7.16320097e-01
9.64478016e-01 -5.45009017e-01 -8.67839575e-01 2.39182115e-01
4.56106752e-01 -3.98823708e-01 1.25379050e+00 -8.23997438e-01
5.23099899e-01 -2.34884601e-02 -3.49500179e-01 -9.72244442e-01
-4.79020625e-01 -6.54648125e-01 -9.88611877e-01 9.41219449e-01
5.49664795e-01 -5.98049045e-01 7.96006680e-01 8.24632287e-01
9.33123156e-02 -1.11824763e+00 -7.35057056e-01 -8.61350358e-01
-9.10881385e-02 -6.92323446e-01 7.60464430e-01 6.37809873e-01
-1.86223492e-01 7.77640522e-01 -5.72086811e-01 3.08651268e-01
4.45591211e-01 2.27431551e-01 9.64085340e-01 -1.25877881e+00
-5.58058619e-01 2.33452930e-03 -3.50588821e-02 -1.93778932e+00
1.61036655e-01 -1.23025024e+00 -2.93256938e-01 -1.53314173e+00
2.91544441e-02 -5.28936744e-01 -2.37828240e-01 6.54866159e-01
-1.36499688e-01 -9.79329795e-02 3.18880022e-01 1.83766589e-01
-5.76890767e-01 6.66368067e-01 9.86512125e-01 -2.58523643e-01
9.17018652e-02 -1.72106922e-01 -8.56595516e-01 7.85899222e-01
8.89196992e-01 -5.36596894e-01 -7.20957994e-01 -1.05315590e+00
6.08192325e-01 -2.62170792e-01 3.58222798e-02 -1.05676293e+00
3.92130226e-01 1.92611486e-01 2.46395126e-01 -6.63703799e-01
6.68204725e-01 -9.45123911e-01 -2.06593171e-01 5.83504975e-01
-3.02367628e-01 1.91559613e-01 2.16852054e-01 7.80511022e-01
-6.99558184e-02 -2.68477947e-01 5.29239595e-01 9.87712992e-04
-3.04958314e-01 5.91919720e-01 -1.41867772e-01 1.87744498e-01
4.77274299e-01 3.12144369e-01 -4.15076107e-01 -1.93811908e-01
-2.82396168e-01 2.46609911e-01 -1.57634228e-01 3.93280983e-01
6.57766402e-01 -1.23740005e+00 -5.20681620e-01 2.12948009e-01
-2.16503039e-01 5.11801124e-01 2.57738173e-01 6.25341654e-01
-6.31032169e-01 8.50087345e-01 2.29220524e-01 -2.32885227e-01
-1.02348840e+00 6.30944073e-01 -1.12562925e-01 -6.25353396e-01
-6.75173283e-01 9.00341988e-01 1.31593823e-01 -3.94984275e-01
3.98725301e-01 -7.65815914e-01 1.46931037e-01 9.28570982e-03
8.40544939e-01 4.87689674e-01 1.35783166e-01 -1.05715491e-01
-1.02545686e-01 3.36381018e-01 -4.53100592e-01 3.01895201e-01
1.49020004e+00 1.24559395e-01 -1.60767704e-01 1.53391436e-01
1.46523499e+00 -5.66965491e-02 -1.13352072e+00 -6.07673943e-01
-3.30984354e-01 -5.64123988e-01 4.10933793e-01 -4.46262389e-01
-1.43825638e+00 1.18104613e+00 1.91995353e-01 -2.62986898e-01
9.05243158e-01 -4.13224727e-01 1.31944633e+00 1.37046182e+00
3.90555143e-01 -1.11088288e+00 -3.88225950e-02 7.01138973e-01
9.41163599e-01 -9.96839345e-01 2.21132159e-01 -4.43695903e-01
-3.88695896e-01 9.01006341e-01 6.20693266e-01 -1.87959209e-01
7.77000248e-01 1.52639017e-01 -5.75814545e-01 -1.08873166e-01
-1.36837184e+00 1.97675541e-01 2.10142910e-01 2.65607595e-01
3.18722367e-01 1.01889111e-01 -2.62612760e-01 8.28700542e-01
-5.30756116e-01 -4.47164997e-02 2.25201637e-01 5.50275385e-01
-4.29980606e-01 -8.06556225e-01 3.53277922e-02 8.33838105e-01
-2.53216088e-01 -4.83341366e-01 1.12800561e-02 5.76617777e-01
-1.17891416e-01 8.83163273e-01 2.03662410e-01 -4.66994882e-01
2.24703312e-01 -2.25828309e-02 3.16892833e-01 -5.44448256e-01
-2.95169115e-01 -3.37364823e-01 8.40810165e-02 -7.08781242e-01
1.80975527e-01 -2.11445108e-01 -1.35454309e+00 -8.10930610e-01
-3.64719361e-01 2.37119496e-02 7.72539675e-01 7.29516149e-01
8.53762746e-01 3.68714750e-01 2.99846441e-01 -5.62242091e-01
-9.66474175e-01 -1.07399023e+00 -3.12660724e-01 1.80905268e-01
-2.65900120e-02 -4.66136783e-01 -3.47534001e-01 -1.90638408e-01] | [8.738411903381348, 3.649918794631958] |
5d85cc06-d78a-4e84-b1b8-dfec7831c5ec | how-to-make-them-stay-diverse-counterfactual | 2303.04579 | null | https://arxiv.org/abs/2303.04579v1 | https://arxiv.org/pdf/2303.04579v1.pdf | "How to make them stay?" -- Diverse Counterfactual Explanations of Employee Attrition | Employee attrition is an important and complex problem that can directly affect an organisation's competitiveness and performance. Explaining the reasons why employees leave an organisation is a key human resource management challenge due to the high costs and time required to attract and keep talented employees. Businesses therefore aim to increase employee retention rates to minimise their costs and maximise their performance. Machine learning (ML) has been applied in various aspects of human resource management including attrition prediction to provide businesses with insights on proactive measures on how to prevent talented employees from quitting. Among these ML methods, the best performance has been reported by ensemble or deep neural networks, which by nature constitute black box techniques and thus cannot be easily interpreted. To enable the understanding of these models' reasoning several explainability frameworks have been proposed. Counterfactual explanation methods have attracted considerable attention in recent years since they can be used to explain and recommend actions to be performed to obtain the desired outcome. However current counterfactual explanations methods focus on optimising the changes to be made on individual cases to achieve the desired outcome. In the attrition problem it is important to be able to foresee what would be the effect of an organisation's action to a group of employees where the goal is to prevent them from leaving the company. Therefore, in this paper we propose the use of counterfactual explanations focusing on multiple attrition cases from historical data, to identify the optimum interventions that an organisation needs to make to its practices/policies to prevent or minimise attrition probability for these cases. | ['Andreas Gregoriades', 'André Artelt'] | 2023-03-08 | null | null | null | null | ['counterfactual-explanation'] | ['miscellaneous'] | [ 4.78548855e-01 5.77249527e-01 -4.60447699e-01 -3.37935030e-01
-1.74263511e-02 -8.76902137e-03 6.73837185e-01 3.38123977e-01
-5.53177178e-01 1.02284992e+00 4.15749729e-01 -8.01112950e-01
-7.80336261e-01 -6.82773173e-01 -4.19739872e-01 -5.04319370e-01
4.98344392e-01 6.31347775e-01 -5.40680110e-01 -2.23494828e-01
7.32395768e-01 7.57654667e-01 -1.73751092e+00 2.81010538e-01
7.43048787e-01 2.84914017e-01 3.43404949e-01 5.53960204e-01
-3.09475988e-01 7.58346260e-01 -5.66119611e-01 -3.54714990e-01
6.55734241e-02 -6.34825766e-01 -1.06477082e+00 6.42569140e-02
-1.57030016e-01 -8.47399086e-02 2.35804394e-01 5.10535896e-01
4.61034954e-01 1.77856907e-01 7.25403309e-01 -1.01454675e+00
-2.36789912e-01 5.65611005e-01 -3.23844016e-01 2.94442534e-01
2.85773098e-01 9.06653851e-02 8.02835882e-01 -2.03616127e-01
3.12556863e-01 1.25741792e+00 3.36084634e-01 4.99561936e-01
-1.43015516e+00 -6.05658710e-01 9.90141407e-02 2.58291453e-01
-5.47661841e-01 -4.26307917e-01 7.05630183e-01 -4.61904943e-01
9.26511884e-01 4.50399965e-01 6.11486137e-01 7.65712500e-01
5.32638669e-01 2.58112550e-01 1.15004337e+00 -9.61716056e-01
6.74894154e-02 4.43310827e-01 -1.50433034e-01 2.04054639e-01
7.47627676e-01 3.75637650e-01 -2.76145488e-01 -4.11670562e-03
4.74807024e-01 2.09052607e-01 -9.03783515e-02 -1.18253440e-01
-8.52603853e-01 1.11815023e+00 4.32310194e-01 6.83021486e-01
-7.79867887e-01 2.30291992e-01 3.83226216e-01 1.37781426e-01
1.45668119e-01 8.94239366e-01 -5.33158600e-01 -1.96152329e-01
-7.30836749e-01 4.45398957e-01 5.54208517e-01 5.74399158e-02
6.45811856e-01 3.87454107e-02 -1.06120624e-01 3.84585410e-01
1.92623302e-01 1.50909215e-01 5.11898935e-01 -9.62924063e-01
3.61925483e-01 1.11657977e+00 2.90992975e-01 -8.83071959e-01
-3.60558480e-01 -6.36159658e-01 -6.05352938e-01 3.71489882e-01
3.80059838e-01 -6.22138344e-02 -6.88153982e-01 1.25182116e+00
-6.72895387e-02 -1.14969194e-01 1.34978697e-01 6.15698636e-01
1.83857337e-01 4.20716435e-01 3.05559039e-01 -4.97328311e-01
1.16411650e+00 -1.80233508e-01 -7.26547122e-01 -6.66752756e-01
1.17888010e+00 -6.32038355e-01 8.48540425e-01 2.49938309e-01
-8.99309099e-01 -5.79221010e-01 -6.90660954e-01 5.23273408e-01
-2.12300017e-01 2.37580258e-02 7.44145632e-01 7.89695859e-01
-3.54357213e-01 1.06249762e+00 -6.71905398e-01 -2.01570183e-01
4.55993593e-01 6.66904509e-01 -8.13056007e-02 -9.29118246e-02
-1.18012702e+00 1.11364448e+00 6.67410672e-01 4.69719559e-01
-2.06228092e-01 -4.68590051e-01 -4.75169986e-01 3.26117784e-01
5.27934670e-01 -8.80098939e-01 1.21847045e+00 -1.16319537e+00
-7.39440978e-01 6.69605911e-01 -2.18365893e-01 -8.39232445e-01
4.85697716e-01 -7.41171604e-03 -1.07743666e-01 -5.56447446e-01
2.91055977e-01 2.87869990e-01 3.64077628e-01 -1.10321069e+00
-9.68793988e-01 -6.70946300e-01 -1.62031487e-01 1.68206260e-01
-2.26112437e-02 1.88555494e-02 5.65262973e-01 -3.02520245e-01
-5.98045699e-02 -1.17966855e+00 -4.86003369e-01 -1.26807690e+00
-3.21528882e-01 -1.82621136e-01 7.02709317e-01 -5.58394551e-01
1.18116581e+00 -1.53818369e+00 -6.71056658e-02 2.72208806e-02
-1.05373487e-01 4.40564841e-01 5.41344285e-01 4.28660929e-01
-1.39404520e-01 3.06976765e-01 3.69908810e-02 1.05660565e-01
-1.62266850e-01 4.28536236e-01 -1.47593245e-01 9.19729397e-02
2.75673032e-01 5.99500060e-01 -5.26095033e-01 -5.86214587e-02
3.95501405e-01 1.79383323e-01 -3.85546505e-01 1.98954672e-01
1.43321648e-01 5.72711647e-01 -5.02124131e-01 2.63927907e-01
2.82134116e-01 3.10406715e-01 4.31087971e-01 4.01549250e-01
-2.96595514e-01 4.68193799e-01 -9.37929749e-01 5.76095700e-01
-7.94256806e-01 7.13644683e-01 -2.76013434e-01 -1.43567967e+00
1.19112766e+00 4.66471553e-01 1.75736874e-01 -7.20309019e-01
2.00538710e-01 5.93384743e-01 4.27429110e-01 -5.10279596e-01
4.29771900e-01 -8.13679814e-01 1.38394777e-02 4.68633622e-01
-6.05654299e-01 1.75631210e-01 3.03812083e-02 -1.89105794e-01
8.06287706e-01 -2.87894513e-02 5.73047101e-01 2.76590474e-02
8.53276432e-01 9.91177335e-02 7.66004562e-01 6.88604593e-01
-7.66641945e-02 3.72507609e-02 5.86742043e-01 -7.98770428e-01
-8.98707211e-01 -1.97680742e-01 1.28714323e-01 6.34963274e-01
-4.64320928e-01 4.84487802e-01 -4.63370800e-01 -6.17767870e-01
5.46139516e-02 1.48564684e+00 -6.36913300e-01 -5.83235919e-01
-6.46114469e-01 -7.37388492e-01 -4.78726141e-02 3.84934992e-01
4.76561189e-01 -1.53079808e+00 -1.23325062e+00 5.03170788e-01
-2.12205514e-01 -4.47591156e-01 1.22334950e-01 3.03486079e-01
-1.24852395e+00 -1.20344579e+00 -4.76601213e-01 -2.43166029e-01
6.06996655e-01 2.28774816e-01 8.97801042e-01 2.16399595e-01
-1.88663498e-01 9.42401066e-02 -1.33751497e-01 -1.13113379e+00
-7.69508779e-01 2.52306044e-01 2.65677840e-01 -1.55435920e-01
5.91932356e-01 -2.88886309e-01 -4.74028230e-01 3.02473307e-01
-7.41175234e-01 4.65870043e-03 9.53129530e-01 8.13104331e-01
3.73625547e-01 4.98930365e-01 1.01241791e+00 -1.25089061e+00
8.27636003e-01 -2.49799818e-01 -1.92427859e-01 2.59708136e-01
-1.16955745e+00 3.29993159e-01 4.00777429e-01 -3.25534046e-01
-1.41110325e+00 -2.40578026e-01 1.40990838e-01 1.61727056e-01
-4.64975089e-01 5.97677290e-01 -1.99423730e-01 5.25940776e-01
5.61570287e-01 6.05726540e-02 1.45040704e-02 -2.89336026e-01
-3.49994227e-02 7.13444889e-01 2.43135810e-01 -3.07156414e-01
2.82159746e-01 2.75254995e-01 2.94246972e-01 -5.23554325e-01
-1.05768287e+00 -4.88880366e-01 -7.12826192e-01 -3.78604949e-01
5.54457963e-01 -4.39770788e-01 -8.36154640e-01 -2.96141416e-01
-9.26453233e-01 3.92152406e-02 -3.00506353e-01 6.27037704e-01
-8.12967658e-01 -3.14683020e-01 3.38131249e-01 -1.28341973e+00
-1.24721907e-01 -8.98030698e-01 3.67179096e-01 4.81400341e-01
-7.86933362e-01 -1.07771468e+00 -1.84969187e-01 8.35676551e-01
3.79622757e-01 3.60883445e-01 9.94797230e-01 -7.95676887e-01
-2.71719873e-01 -4.03613299e-01 1.56153873e-01 2.64728487e-01
3.92783880e-01 -3.47378910e-01 -5.33501744e-01 -1.99910939e-01
2.91698009e-01 1.52713612e-01 7.35475838e-01 8.30963314e-01
7.85241961e-01 -6.75797045e-01 -5.71624339e-01 -8.74906182e-02
1.15516245e+00 6.42833889e-01 6.93077266e-01 1.00624049e+00
3.36604923e-01 1.34291720e+00 1.02046132e+00 3.74263674e-01
-3.76193114e-02 6.75981641e-01 5.57829082e-01 8.10511783e-03
4.58955765e-01 8.75148550e-03 1.99279606e-01 -5.41336648e-02
-5.94948769e-01 9.76738632e-02 -7.92573273e-01 6.20515645e-01
-1.95162153e+00 -1.39661574e+00 -6.41333044e-01 2.52344251e+00
1.76722959e-01 6.56349480e-01 2.52483845e-01 6.25906229e-01
6.05647027e-01 -2.57179350e-01 -1.73690185e-01 -1.27905595e+00
3.35910946e-01 -1.47508457e-01 6.62175298e-01 2.83095300e-01
-5.15431404e-01 4.26473796e-01 4.92157745e+00 2.20019698e-01
-7.14634657e-01 -4.18514550e-01 9.82996702e-01 -1.94357723e-01
-2.78161049e-01 5.05717695e-01 -1.07791233e+00 2.41756529e-01
1.51421702e+00 -4.13906544e-01 1.72009662e-01 5.67194760e-01
1.05567610e+00 -2.88569719e-01 -8.78613949e-01 1.87834010e-01
-4.53542173e-01 -1.29677272e+00 -1.18387245e-01 5.76954246e-01
6.72423124e-01 -7.84442067e-01 -5.55168837e-02 3.65000069e-01
9.68463346e-02 -1.26778352e+00 3.93749714e-01 9.55146611e-01
9.74596590e-02 -1.49332666e+00 1.39231443e+00 8.72272730e-01
-4.25246239e-01 -6.51861191e-01 -3.61055225e-01 -7.83499539e-01
1.97195664e-01 4.76396561e-01 -1.62161970e+00 5.05637467e-01
4.15366083e-01 -3.36682349e-02 -1.46105215e-01 6.05573893e-01
-1.87294409e-01 6.77178502e-01 3.27134699e-01 -9.18692052e-02
4.05992955e-01 -2.67473400e-01 2.89428622e-01 7.31386721e-01
5.34080088e-01 -7.21639842e-02 -4.11492646e-01 9.15459037e-01
3.71528000e-01 7.67154768e-02 -9.14285064e-01 -2.09031463e-01
3.04135501e-01 7.87929595e-01 -7.49184191e-01 2.50257067e-02
-1.24789953e-01 6.99443996e-01 4.32651117e-02 -3.26942354e-02
-3.77063274e-01 -2.97712684e-01 4.62917328e-01 8.35918486e-01
-1.61424931e-02 1.90085530e-01 -4.51947212e-01 -3.60144228e-02
-2.65296698e-01 -8.38623941e-01 2.51694381e-01 -3.12723428e-01
-5.69982111e-01 1.07924007e-01 4.74658795e-02 -6.83090985e-01
-5.71483195e-01 -3.78426969e-01 -8.05089772e-01 1.24864066e+00
-1.26363111e+00 -9.04300928e-01 1.85214281e-01 -3.28465253e-01
5.27296126e-01 -2.42425978e-01 6.53984845e-01 -1.98049217e-01
-4.64191347e-01 -3.37152705e-02 6.98327571e-02 -4.85335648e-01
3.40157986e-01 -1.23881924e+00 -1.76916383e-02 5.43484986e-01
6.91461712e-02 6.19615734e-01 1.10350490e+00 -8.89703631e-01
-7.31860399e-01 -8.91468942e-01 1.59100628e+00 -2.58259982e-01
5.94408885e-02 2.35461697e-01 -9.19125378e-01 6.80508733e-01
-1.08124584e-01 -7.15819001e-01 5.97485602e-01 4.70858037e-01
6.93006098e-01 -1.02015696e-01 -9.60293531e-01 6.00630879e-01
5.00845194e-01 3.26612182e-02 -7.46758282e-01 1.13541838e-02
3.21970284e-01 1.41794726e-01 -4.47396547e-01 2.77731299e-01
4.06926304e-01 -1.53621554e+00 8.16094577e-01 -9.29467738e-01
4.41829264e-01 1.93385705e-01 2.53523231e-01 -1.35746288e+00
-3.32532525e-01 -3.76676917e-01 3.04658026e-01 1.28480899e+00
3.26232016e-01 -7.81839192e-01 1.06700492e+00 9.35983598e-01
1.22041985e-01 -1.01566744e+00 -9.90862072e-01 -6.42254710e-01
-1.87095240e-01 -3.95574063e-01 8.26159418e-01 7.74181247e-01
-3.34822088e-01 3.56887013e-01 -4.94175375e-01 -2.44913280e-01
3.75272155e-01 5.33873476e-02 1.00763988e+00 -1.61695540e+00
-8.65015686e-02 -6.50216222e-01 -3.03801596e-01 -4.76061404e-02
1.88518330e-01 -6.24520659e-01 -3.68546098e-01 -1.72902322e+00
2.34691665e-01 -3.09606016e-01 -2.38727033e-01 2.98649162e-01
-2.70149142e-01 -4.70714301e-01 2.28788272e-01 2.47872651e-01
2.28271335e-01 3.20843339e-01 1.07342589e+00 1.20303757e-01
-7.70925224e-01 9.36192691e-01 -1.30904925e+00 8.05458665e-01
1.00497758e+00 -7.31481850e-01 -2.35250145e-01 3.07809800e-01
3.88472259e-01 4.24991310e-01 4.75424826e-01 -7.10163534e-01
-2.47429669e-01 -5.99629998e-01 3.49915534e-01 -3.60089749e-01
-1.17360845e-01 -9.26872611e-01 6.71082020e-01 1.15293515e+00
-5.18779457e-01 8.32394287e-02 2.01287016e-01 6.27094150e-01
-1.46514237e-01 -7.88717330e-01 4.91357476e-01 -2.16073662e-01
-2.42909595e-01 -2.84902960e-01 -5.28476655e-01 -5.92916787e-01
1.14730489e+00 -5.05153239e-01 5.92476875e-02 -4.81508046e-01
-9.04180646e-01 1.79226249e-01 2.08709389e-01 4.69432175e-01
3.83305758e-01 -1.20604098e+00 -7.98074126e-01 -2.94941634e-01
-1.53472647e-01 -2.21314892e-01 3.68030906e-01 8.98091018e-01
-3.00941497e-01 1.21202815e+00 -3.14753741e-01 5.79768829e-02
-1.51773965e+00 4.47677851e-01 3.07665825e-01 -9.51797128e-01
-3.00415188e-01 5.02107322e-01 -1.24773523e-02 -2.84835935e-01
-2.50035942e-01 1.47255376e-01 -7.18459487e-01 8.40901434e-02
3.14089745e-01 8.66711318e-01 1.69714317e-01 -6.07383013e-01
-2.61940241e-01 -1.62657589e-01 -3.61828685e-01 1.70231476e-01
1.62767065e+00 -1.37240231e-01 8.76001194e-02 5.52376866e-01
6.62331641e-01 -9.33757946e-02 -1.13338029e+00 2.07018390e-01
5.33344746e-01 -7.47999966e-01 2.87604004e-01 -8.42312157e-01
-5.65909386e-01 8.72091472e-01 5.40469110e-01 5.61493099e-01
1.10876155e+00 -2.76059508e-01 2.65828669e-01 1.00657791e-01
2.00537238e-02 -1.26880908e+00 -8.80469233e-02 -1.15856811e-01
9.51678693e-01 -1.33957326e+00 2.41813034e-01 -2.22164281e-02
-7.54168808e-01 1.07897699e+00 3.07314873e-01 3.06629747e-01
1.07154809e-01 -5.89652300e-01 -1.89724341e-01 -3.05444002e-01
-7.05937624e-01 -8.88512060e-02 3.12213898e-01 4.18656349e-01
8.45434487e-01 3.98216784e-01 -8.15218806e-01 4.95257080e-01
-2.18641877e-01 9.26115587e-02 6.85815990e-01 7.02189505e-01
-6.30536258e-01 -1.52480245e+00 -7.97339976e-01 1.13002121e+00
-7.18793273e-01 3.54918867e-01 -6.22288465e-01 1.09361005e+00
3.68016183e-01 9.31126058e-01 -1.62180051e-01 -5.22291735e-02
7.08413422e-01 2.82710314e-01 9.20894593e-02 -5.78913927e-01
-4.46106851e-01 -1.94263622e-01 3.39251071e-01 -1.73551980e-02
-6.13773167e-01 -1.09455252e+00 -1.31892765e+00 -3.55636358e-01
-4.82976079e-01 2.18286440e-01 7.38781989e-01 1.37290001e+00
2.90501803e-01 1.01694977e+00 6.43349409e-01 -7.34089971e-01
-7.10118532e-01 -9.58045781e-01 -5.26924372e-01 2.11842403e-01
-1.35747269e-01 -8.39908302e-01 -3.85365188e-01 -3.88305992e-01] | [8.62459945678711, 5.562601089477539] |
ce3e28f9-aef8-4abd-926a-5a773c2d0d8a | black-scholes-option-pricing-revisited | 2202.05671 | null | https://arxiv.org/abs/2202.05671v2 | https://arxiv.org/pdf/2202.05671v2.pdf | Black-Scholes-Merton Option Pricing Revisited: Did we Find a Fatal Flaw? | The option pricing formula of Black and Scholes (1973) hinges on the continuous-time self-financing condition, which is a special case of the continuous-time budget equation of Merton (1971). The self-financing condition is believed to formalize the economic concept of portfolio rebalancing without inflows or outflows of external funds, but was never formally derived in continuous time. As a much bigger issue, however, we discover a timing mistake in the model of Merton (1971) and show that his self-financing condition is misspecified both in discrete and continuous time. Our results invalidate seminal contributions to the literature, including the budget equation of Merton (1971), the option pricing formula of Black and Scholes (1973), the continuous trading model of Harrison and Pliska (1981), and the binomial option pricing model of Cox, Ross and Rubinstein (1979). We also show that Black and Scholes (1973) and Harrison and Pliska (1981) implicitly assumed their replication result. | ['Frans J. de Weert', 'Mark Mink'] | 2022-01-09 | null | null | null | null | ['mathematical-proofs'] | ['miscellaneous'] | [-5.85782230e-01 2.19170168e-01 -3.85974318e-01 -9.24223810e-02
1.71023309e-01 -9.64713931e-01 5.82684934e-01 -2.46523749e-02
-5.98141849e-01 1.25847924e+00 -5.86294718e-02 -1.20525575e+00
-4.29188788e-01 -9.63479280e-01 -1.82396114e-01 -2.45608106e-01
-1.39419269e-02 4.78004664e-01 2.17490479e-01 9.36686993e-02
4.93590474e-01 5.03848255e-01 -1.02470672e+00 -3.76705110e-01
5.87699175e-01 1.02831423e+00 -3.07596713e-01 6.54297411e-01
-4.63008821e-01 6.69131815e-01 -1.19343497e-01 -7.55872011e-01
5.29323280e-01 -5.53732336e-01 -5.58403671e-01 -2.66041487e-01
-4.64319706e-01 -6.30774617e-01 -2.30315030e-01 9.99888837e-01
-1.45825282e-01 -9.84531567e-02 8.17376733e-01 -1.44927609e+00
-7.78917253e-01 7.59408116e-01 -7.15894043e-01 4.83054191e-01
-3.60171080e-01 -1.77603941e-02 1.01518309e+00 -7.19542921e-01
2.78644741e-01 8.48305404e-01 7.36860096e-01 3.58980000e-01
-1.05779696e+00 -6.86108470e-01 -1.81208123e-02 -4.08776134e-01
-1.05923438e+00 -1.23927653e-01 2.16417894e-01 -5.66018164e-01
5.48916698e-01 4.90633786e-01 9.37986612e-01 4.27592844e-01
4.31571305e-01 6.98912382e-01 1.25834286e+00 -7.03143418e-01
3.67823064e-01 1.51992943e-02 2.67000526e-01 -1.43325729e-02
9.40908909e-01 6.04814529e-01 -6.22579306e-02 -5.78578591e-01
1.56147432e+00 3.46502602e-01 -1.46295175e-01 -2.95468271e-01
-1.07281423e+00 1.21495891e+00 -3.92464697e-01 3.21192890e-01
-5.36384225e-01 2.40199745e-01 4.17995870e-01 7.84923315e-01
3.24637622e-01 -8.55490714e-02 -4.66554284e-01 -2.88286090e-01
-1.07118273e+00 5.11254787e-01 1.15950227e+00 9.41840470e-01
-1.46929948e-02 2.47341275e-01 -1.05373241e-01 -9.17294470e-04
6.99170530e-01 5.21301329e-01 5.52272022e-01 -1.50314975e+00
1.72845483e-01 -3.40059400e-01 6.92737818e-01 -3.30289871e-01
-1.49996698e-01 -4.51808959e-01 -3.06731135e-01 3.64202231e-01
8.89756978e-01 -2.95250148e-01 -3.02138962e-02 1.81405103e+00
-9.21887755e-02 -1.70097202e-01 1.17070764e-01 4.55709398e-01
-5.19683838e-01 4.09640968e-01 -2.21478879e-01 -9.64313328e-01
8.18792343e-01 -8.10205400e-01 -6.57384872e-01 8.02726090e-01
2.24859938e-01 -6.77383780e-01 5.03419399e-01 3.64702970e-01
-1.47303283e+00 -3.98675120e-03 -5.72135389e-01 5.70195615e-01
3.55539232e-01 -8.45122278e-01 1.05126417e+00 1.01957536e+00
-1.25465751e+00 8.87735128e-01 -6.95982873e-01 -1.11094460e-01
2.79256463e-01 1.10779069e-01 3.18863511e-01 6.39721930e-01
-1.03985810e+00 6.93733990e-01 -1.44021422e-01 -1.80521607e-01
-5.61424911e-01 -5.56144416e-01 -6.77725598e-02 2.69790113e-01
6.74634695e-01 -7.56084859e-01 1.64681315e+00 -1.20459890e+00
-1.61613524e+00 4.07554924e-01 5.68208098e-01 -8.93956780e-01
9.99299765e-01 2.56163269e-01 -2.02560455e-01 1.53847530e-01
1.49921328e-01 -2.23957404e-01 3.59463334e-01 -6.35870337e-01
-6.34037912e-01 -4.53461736e-01 -2.79710088e-02 -9.40229464e-03
1.96012855e-01 3.77510041e-01 7.83359051e-01 -1.07522094e+00
7.34818578e-02 -5.53184867e-01 -6.46862686e-02 8.00304562e-02
2.03245670e-01 -1.09941036e-01 -9.37934518e-02 -3.93028110e-01
1.05295742e+00 -1.92764699e+00 -4.98207688e-01 2.44504422e-01
3.33261117e-02 -4.00699347e-01 3.64456117e-01 7.06905603e-01
-6.31491244e-01 4.84975904e-01 -3.74154747e-01 2.64315426e-01
4.70017225e-01 5.29764928e-02 -8.43084574e-01 6.16869748e-01
-3.56535077e-01 8.97812784e-01 -6.51364148e-01 -7.70812556e-02
-2.27739707e-01 -1.03586964e-01 -4.14650142e-01 -3.89784813e-01
-3.15439655e-03 -3.04764640e-02 -2.72734761e-01 5.34934044e-01
5.25750041e-01 -1.74743026e-01 2.29132265e-01 6.27277315e-01
-1.00920594e+00 4.98399913e-01 -1.28473842e+00 7.77762711e-01
1.83257889e-02 4.49304253e-01 -2.69825794e-02 -8.61647844e-01
5.16059399e-01 7.32394695e-01 4.70567465e-01 -5.67067862e-01
2.85374582e-01 6.00567698e-01 -4.99212518e-02 -4.35349680e-02
5.78152478e-01 -1.00682783e+00 3.05931363e-02 1.24786735e+00
-5.76818109e-01 5.85355647e-02 1.52613208e-01 2.20883694e-02
7.93582082e-01 1.47715390e-01 5.14856994e-01 -6.51185989e-01
-7.93260429e-03 6.30938858e-02 4.73802745e-01 7.53687978e-01
-1.87614724e-01 4.15197521e-01 9.72078621e-01 -1.66099310e-01
-1.30119550e+00 -1.12527645e+00 -5.48899174e-01 4.39899385e-01
-4.23453182e-01 3.88018876e-01 -4.78347212e-01 -5.29053360e-02
6.11480892e-01 1.04301798e+00 -7.69467413e-01 4.66106832e-01
-2.29282990e-01 -6.29077077e-01 5.79095900e-01 4.30216253e-01
2.20431536e-01 -7.90387511e-01 -1.33812892e+00 4.15706307e-01
2.92062640e-01 2.97292788e-02 -6.04791582e-01 3.79714072e-01
-1.23317230e+00 -1.05345750e+00 -1.48487115e+00 1.92491040e-01
3.27512443e-01 -1.11151993e-01 7.02008843e-01 -6.79848045e-02
1.57770291e-01 6.92962825e-01 -2.06117034e-01 -6.67545974e-01
-4.15906399e-01 -8.20164979e-01 -5.49288988e-02 -7.31181446e-03
2.95428246e-01 -5.15563428e-01 -5.01209497e-01 3.38460594e-01
-1.03533292e+00 -4.01819766e-01 9.15927738e-02 1.04303491e+00
2.74810970e-01 -4.33074161e-02 9.87094104e-01 -6.14473701e-01
5.67957163e-01 -5.57792425e-01 -1.38116705e+00 3.19326162e-01
-1.35174000e+00 -5.07501364e-02 -7.30625167e-02 -5.24288237e-01
-1.08260179e+00 -9.66520846e-01 3.15608382e-01 -1.40635207e-01
4.76127923e-01 7.04807699e-01 2.03967005e-01 -1.18033387e-01
-4.74498793e-02 1.08988240e-01 2.21088037e-01 -8.52837086e-01
6.94069266e-02 6.13318384e-01 3.44290406e-01 -6.01150990e-01
5.21399200e-01 5.93641460e-01 6.01226576e-02 -2.89505005e-01
-2.45900974e-01 -1.53248233e-03 -2.42011443e-01 1.49710238e-01
5.70799887e-01 -7.33218133e-01 -1.15272510e+00 4.59827274e-01
-9.44528937e-01 -5.00177085e-01 -1.19517994e+00 8.45533550e-01
-1.03567612e+00 3.67190301e-01 -7.75689185e-01 -1.72138965e+00
2.55264372e-01 -3.77319843e-01 -2.54083388e-02 2.41046712e-01
-2.85148829e-01 -1.23654819e+00 4.79270726e-01 -1.46945342e-01
8.30183148e-01 3.90167654e-01 8.48854125e-01 -6.40748084e-01
-5.16745150e-01 -2.34336808e-01 -2.00768277e-01 1.10843681e-01
-1.81099474e-01 2.09703237e-01 -4.12082732e-01 -1.07234009e-01
9.71712112e-01 6.44713938e-02 6.16147876e-01 5.29590666e-01
4.22943622e-01 -6.13521516e-01 2.83042431e-01 2.81800538e-01
1.56443727e+00 7.52411306e-01 6.40482783e-01 7.28017151e-01
-3.25914145e-01 8.32826793e-01 3.58014971e-01 7.62023330e-01
5.42027891e-01 2.08607674e-01 1.66696131e-01 4.32375729e-01
6.99873269e-01 -2.94874847e-01 3.55489612e-01 4.41695154e-01
-2.93006063e-01 4.65358160e-02 -5.52806675e-01 6.85414195e-01
-1.89503253e+00 -1.48676777e+00 -2.87793279e-01 2.66428638e+00
9.27785039e-01 3.25749934e-01 9.34955835e-01 1.76459923e-01
6.19926155e-01 -2.37159282e-01 -5.01519620e-01 -6.88221693e-01
-3.01250279e-01 2.02762201e-01 1.15300477e+00 5.85558474e-01
-2.42073461e-01 1.66712329e-01 7.34043980e+00 2.57975310e-01
-6.27743959e-01 2.07398146e-01 8.60763192e-01 -1.04086481e-01
-8.96976233e-01 6.09685957e-01 -6.16817772e-01 7.55384028e-01
1.09732735e+00 -1.32113194e+00 2.82676041e-01 6.11338437e-01
-2.18868535e-02 -3.69468689e-01 -9.24183130e-01 4.47584301e-01
-1.56274542e-01 -1.01404619e+00 -4.92153674e-01 7.03193367e-01
7.32978284e-01 -4.02574241e-01 2.58733034e-01 1.68050840e-01
5.69469512e-01 -1.07065201e+00 1.11063898e+00 7.57934570e-01
3.93693537e-01 -9.11129892e-01 5.00227630e-01 4.83454227e-01
-6.43079460e-01 -1.86780527e-01 -4.52164590e-01 -5.07670224e-01
5.00661552e-01 5.94021022e-01 -1.51089625e-02 4.48096514e-01
9.49497074e-02 -4.05304506e-02 1.34820983e-01 1.17338371e+00
1.24673493e-01 7.49605775e-01 -4.95032758e-01 1.73178822e-01
2.87469298e-01 -6.67982638e-01 2.62676060e-01 4.73256052e-01
4.84102577e-01 4.92978811e-01 -6.58817112e-01 1.26805794e+00
3.37336302e-01 -1.25197053e-01 -2.50918984e-01 -6.13463044e-01
1.21405400e-01 5.16978562e-01 -9.60388303e-01 -1.00812659e-01
-1.02235651e+00 4.61551160e-01 -7.31466055e-01 5.43387592e-01
-6.75025880e-01 -2.95884967e-01 1.06496572e-01 5.87920547e-01
5.72721362e-01 -2.51014292e-01 -7.20946670e-01 -1.12728417e+00
3.22006494e-02 -4.12497342e-01 4.23949778e-01 -3.13000292e-01
-1.28844583e+00 -4.05733436e-02 4.49145645e-01 -7.31346965e-01
-5.09288728e-01 -5.57173312e-01 -8.72027934e-01 1.52243745e+00
-1.61677682e+00 -5.36020875e-01 9.12474811e-01 3.03825349e-01
1.34328112e-01 -8.62470418e-02 4.55317229e-01 -7.71865994e-02
-6.82945997e-02 3.01098645e-01 8.52050185e-01 -7.02344999e-02
3.32223326e-01 -1.40174150e+00 6.16543531e-01 5.06265759e-01
-4.84480783e-02 6.81251645e-01 5.23556471e-01 -1.00885987e+00
-9.62740183e-01 -1.80093706e-01 1.20507514e+00 -2.41685286e-01
1.20045304e+00 2.91962832e-01 -7.84063995e-01 9.85557854e-01
3.93036693e-01 -4.07085091e-01 7.25477457e-01 -4.01984572e-01
1.22644696e-02 1.99421361e-01 -1.06629777e+00 3.01989585e-01
4.65742767e-01 -2.39489004e-01 -7.85621762e-01 3.18287089e-02
7.08937585e-01 8.92185867e-02 -7.37965167e-01 -9.21914130e-02
1.09689760e+00 -1.33163643e+00 6.11113131e-01 -4.59992528e-01
1.25776008e-01 2.50338525e-01 -1.93064749e-01 -5.06787658e-01
-2.52143115e-01 -1.10711515e+00 3.68481517e-01 1.10076308e+00
2.87255853e-01 -1.49481547e+00 2.98666894e-01 8.31140041e-01
2.69243062e-01 -6.17614329e-01 -1.24202561e+00 -1.39016807e+00
9.40984845e-01 -1.68256387e-02 6.70690119e-01 1.14294422e+00
1.19418107e-01 -6.04943931e-01 -1.53906837e-01 -8.15788031e-01
1.02900386e+00 2.45146200e-01 1.62638456e-01 -1.38358474e+00
-7.82247305e-01 -8.56796741e-01 2.91789472e-01 -7.84565985e-01
-1.07455425e-01 -6.71362698e-01 -5.38708568e-01 -1.05229616e+00
2.18292519e-01 -5.46820104e-01 -5.24540603e-01 1.02052696e-01
5.07780313e-01 -3.32144946e-01 5.02671897e-01 5.89609385e-01
2.26753756e-01 3.82482678e-01 8.09391081e-01 2.59761214e-01
-6.51519075e-02 3.42333972e-01 -1.03557229e+00 8.41910064e-01
5.29657066e-01 -6.77776515e-01 -4.78397280e-01 -7.19483420e-02
7.63674080e-01 9.46905136e-01 7.07281351e-01 -2.08226740e-02
3.65179300e-01 -8.52107465e-01 8.30183178e-02 -7.40184605e-01
-3.47030997e-01 -5.79784632e-01 7.95252204e-01 7.80468643e-01
-2.00582772e-01 7.03370333e-01 8.95856470e-02 5.83129585e-01
1.02269270e-01 -1.00613737e+00 4.84824836e-01 -4.46905345e-01
2.61332184e-01 6.99959397e-02 -3.55503678e-01 3.55211169e-01
1.17903686e+00 -2.36620560e-01 -2.72734016e-01 -4.94527847e-01
-7.87029088e-01 2.17989445e-01 9.55535114e-01 -3.46481323e-01
4.36771631e-01 -1.42457783e+00 -6.59686685e-01 9.41397622e-02
-7.58009672e-01 -3.22405368e-01 -1.20391920e-01 1.27648628e+00
-7.28593946e-01 5.97560167e-01 -3.39646220e-01 1.81453064e-01
-5.74068725e-01 5.06391823e-01 3.47435683e-01 -1.68616727e-01
-6.35519445e-01 6.04039252e-01 3.61637145e-01 7.23065674e-01
1.39612742e-02 -2.38958120e-01 2.18444824e-01 2.29117036e-01
6.31994128e-01 7.74442673e-01 -8.17354500e-01 -2.80670524e-01
-2.62681425e-01 1.96003869e-01 1.73353478e-01 -1.13785958e+00
1.52523077e+00 -4.57280546e-01 -3.78489435e-01 1.20556355e+00
5.76235056e-01 7.20401853e-02 -9.39813256e-01 1.72645092e-01
3.07278037e-01 -5.48541546e-01 -3.43618333e-01 -5.40884852e-01
-6.02205157e-01 7.89241195e-01 -3.50959823e-02 6.64870203e-01
6.58081174e-01 -5.32400787e-01 7.34160006e-01 -9.99830514e-02
5.52676797e-01 -1.09150720e+00 -4.23515052e-01 2.04865709e-01
9.76906657e-01 -5.28394520e-01 2.27266774e-02 2.40893021e-01
-6.34658992e-01 1.14035165e+00 -3.91796231e-01 -2.14594781e-01
1.14650559e+00 3.39835525e-01 -4.35759574e-01 1.27983406e-01
-7.69947648e-01 2.99162120e-01 -3.10310274e-01 2.07159773e-01
2.94994593e-01 2.01619908e-01 -1.10466111e+00 1.21693778e+00
-3.23386379e-02 3.78996760e-01 1.19317901e+00 1.23463321e+00
-3.90254080e-01 -1.01867402e+00 -6.31891370e-01 4.28425103e-01
-8.56301129e-01 -1.84373125e-01 -1.74960390e-01 9.19461608e-01
-3.50236714e-01 2.86599904e-01 5.12305617e-01 6.08375490e-01
1.43907756e-01 4.83369380e-01 4.36047524e-01 -4.50196028e-01
-4.86808866e-01 8.12561631e-01 -3.32136333e-01 1.09169662e-01
-3.96282881e-01 -1.09672451e+00 -1.01534414e+00 -9.39534724e-01
-5.37646592e-01 4.62088674e-01 4.35387671e-01 8.46455038e-01
-3.60312819e-01 1.34647980e-01 6.67868495e-01 -2.09496785e-02
-1.76545548e+00 -4.72515017e-01 -1.56886482e+00 -3.37458432e-01
2.57193539e-02 -3.52075696e-01 -8.87131512e-01 5.82797872e-03] | [4.8506646156311035, 3.95963191986084] |
860f722a-4b06-4620-99ee-c7c9221380d7 | bo-muse-a-human-expert-and-ai-teaming | 2303.01684 | null | https://arxiv.org/abs/2303.01684v2 | https://arxiv.org/pdf/2303.01684v2.pdf | BO-Muse: A human expert and AI teaming framework for accelerated experimental design | In this paper we introduce BO-Muse, a new approach to human-AI teaming for the optimization of expensive black-box functions. Inspired by the intrinsic difficulty of extracting expert knowledge and distilling it back into AI models and by observations of human behavior in real-world experimental design, our algorithm lets the human expert take the lead in the experimental process. The human expert can use their domain expertise to its full potential, while the AI plays the role of a muse, injecting novelty and searching for areas of weakness to break the human out of over-exploitation induced by cognitive entrenchment. With mild assumptions, we show that our algorithm converges sub-linearly, at a rate faster than the AI or human alone. We validate our algorithm using synthetic data and with human experts performing real-world experiments. | ['Svetha Venkatesh', 'Mahad Rashid', 'Julian Berk', 'Santu Rana', 'Hung Le', 'Majid Abdolshah', 'Shannon Ryan', 'Arun Kumar A V', 'Alistair Shilton', 'Sunil Gupta'] | 2023-03-03 | null | null | null | null | ['experimental-design'] | ['methodology'] | [-1.19984865e-01 5.84137380e-01 1.18049853e-01 1.19984031e-01
-2.11910993e-01 -7.57890940e-01 1.83805451e-01 9.73632112e-02
-6.62465751e-01 7.55891383e-01 -1.71311602e-01 -2.03569397e-01
-4.06791508e-01 -3.38017493e-01 -6.53602123e-01 -5.03817022e-01
-2.17103466e-01 8.75172913e-01 -1.09829403e-01 -2.87955999e-01
5.54591060e-01 1.74025059e-01 -1.26105785e+00 -2.15641186e-01
1.00020480e+00 7.36907482e-01 -2.08725035e-02 7.23053515e-01
5.04759431e-01 1.24395645e+00 -8.29818606e-01 -3.64177138e-01
8.29984546e-01 -7.47532308e-01 -7.39357710e-01 3.95758569e-01
-3.97203863e-02 -5.28276376e-02 -1.03950202e-01 1.11460197e+00
5.22485375e-01 1.46347806e-01 3.69538516e-01 -1.33952951e+00
-4.57741946e-01 7.28167951e-01 -4.50904429e-01 2.61178523e-01
4.90034193e-01 6.70863867e-01 9.75322306e-01 -6.37702525e-01
6.18953109e-01 1.14129710e+00 5.44837415e-01 5.18724501e-01
-1.37738788e+00 -4.81200665e-01 9.35733244e-02 -8.57067704e-02
-1.45504391e+00 -4.04555172e-01 6.26313448e-01 -6.88410282e-01
7.59221375e-01 7.60888830e-02 9.07198846e-01 8.08355272e-01
1.93752721e-01 8.22475374e-01 9.80704725e-01 -4.82020974e-01
6.59148872e-01 5.19247532e-01 -2.89108098e-01 7.87682474e-01
4.88936305e-01 5.39021432e-01 -7.68899798e-01 -4.34576869e-01
8.73272419e-01 -2.54361212e-01 -2.92040467e-01 -4.38567042e-01
-1.17245221e+00 6.82742655e-01 5.25233805e-01 4.82279539e-01
-9.30989563e-01 1.65772974e-01 -1.49371997e-01 9.12300408e-01
2.33738109e-01 1.41875625e+00 -3.21076423e-01 -1.51770622e-01
-7.48434842e-01 5.17567456e-01 9.56896603e-01 6.95400238e-01
4.55739051e-01 -1.12235978e-01 2.00753897e-01 1.80039719e-01
-3.35741043e-02 8.66907462e-02 6.18360758e-01 -1.47495556e+00
2.65644165e-03 9.09578621e-01 6.92323446e-01 -7.93679178e-01
-1.84831873e-01 -7.10514426e-01 -2.61064708e-01 8.01328182e-01
5.89602053e-01 -5.14446676e-01 -4.36242312e-01 1.59451437e+00
1.74436972e-01 -1.32081956e-01 1.09579775e-03 1.01757741e+00
-4.27936077e-01 3.35996181e-01 -3.68774325e-01 -4.64306444e-01
1.05712748e+00 -1.19909310e+00 -3.10948074e-01 -5.11923850e-01
4.98311400e-01 -1.97341576e-01 8.59026372e-01 9.18912232e-01
-1.52154970e+00 -3.56152475e-01 -1.05427670e+00 4.26237434e-01
-1.15615718e-01 -3.06980401e-01 4.75887865e-01 3.01359564e-01
-9.87404168e-01 9.34899926e-01 -5.49334884e-01 -3.79470587e-01
5.51353335e-01 3.90139461e-01 -1.47647589e-01 1.28545225e-01
-7.89584100e-01 9.61978972e-01 4.81903255e-01 1.75817221e-01
-1.10916245e+00 -5.79422116e-01 -3.60015690e-01 1.69207305e-01
1.23845172e+00 -1.00807369e+00 1.36765194e+00 -1.51926088e+00
-1.41903782e+00 6.79168701e-01 3.54975224e-01 -6.01904690e-01
7.88897693e-01 -3.03582758e-01 2.49585837e-01 -9.97419208e-02
8.49474818e-02 3.95240545e-01 7.94776797e-01 -1.33218420e+00
-5.03162205e-01 -5.17809689e-01 2.54552841e-01 3.07091892e-01
-2.47584864e-01 6.49815500e-02 2.48449147e-01 -5.75677097e-01
-1.49355620e-01 -1.12438536e+00 -4.96050566e-01 -1.67511534e-02
-9.98668522e-02 -2.24873498e-01 7.27250725e-02 -3.49888563e-01
1.35720718e+00 -1.95808327e+00 7.29641616e-01 4.48342800e-01
9.05161381e-01 1.21661238e-01 -1.19844424e-02 4.92449492e-01
1.62077248e-02 1.86667398e-01 -1.59856915e-01 5.42282574e-02
1.73889443e-01 -1.54111432e-02 4.02657874e-02 3.27239901e-01
-1.20829985e-01 1.05538964e+00 -1.16325593e+00 -3.76776040e-01
-4.49708104e-01 -3.20725888e-01 -6.54322386e-01 4.01325107e-01
-2.99510747e-01 4.29853857e-01 -7.01266825e-01 7.01496363e-01
-1.04243290e-02 -4.00218904e-01 2.65415639e-01 6.25701547e-01
-5.32897711e-02 -1.18586883e-01 -1.05295837e+00 1.32153225e+00
-1.43553987e-01 5.89903176e-01 4.00453269e-01 -8.86571169e-01
5.26464403e-01 3.50285292e-01 2.26220056e-01 -3.87827069e-01
3.70754540e-01 2.96114236e-01 5.40046215e-01 -4.40104663e-01
1.04645886e-01 -3.10064018e-01 -3.78381424e-02 9.97864127e-01
-8.46103355e-02 -1.10001370e-01 1.94346786e-01 9.83850732e-02
1.45777977e+00 -9.67631042e-02 4.73452866e-01 -5.11110961e-01
1.27720982e-01 4.86207634e-01 5.10825694e-01 1.07630372e+00
-1.48771361e-01 -1.38747528e-01 6.33763075e-01 -6.52812302e-01
-1.05354190e+00 -8.89394701e-01 5.22016227e-01 1.24799788e+00
-7.65517056e-02 -3.35829109e-01 -7.29987860e-01 -6.53590739e-01
1.55726001e-01 9.77860868e-01 -9.24734771e-01 -4.01390731e-01
-4.48532373e-01 -2.62063920e-01 3.53319794e-01 4.53757882e-01
4.46801007e-01 -1.24683297e+00 -1.32714880e+00 3.98346722e-01
2.12102115e-01 -4.89982009e-01 -4.76291448e-01 2.46672839e-01
-7.16254234e-01 -8.53121698e-01 -7.28853464e-01 -3.91887546e-01
9.69991207e-01 -6.66911379e-02 1.22797382e+00 5.51476181e-01
-5.01418054e-01 3.87949705e-01 -2.43224483e-02 -7.66972482e-01
-5.73825419e-01 -2.30017364e-01 2.42989808e-01 -3.33855242e-01
2.66178250e-01 -8.02161753e-01 -4.80790377e-01 4.60202903e-01
-6.16491079e-01 5.68625936e-03 6.21798277e-01 1.05144370e+00
-1.25254631e-01 6.03325129e-01 2.93154389e-01 -5.80926299e-01
1.09741986e+00 -4.86981750e-01 -7.85393357e-01 3.41604203e-01
-8.54589283e-01 9.41620395e-02 3.47734213e-01 -7.42728233e-01
-8.51280451e-01 -1.35926425e-01 1.04929507e+00 -6.11269474e-01
1.94938108e-01 6.83424234e-01 4.59092557e-02 -9.08461139e-02
1.17543972e+00 -4.78357496e-03 1.23820506e-01 -3.99636179e-01
1.28240108e-01 4.06245112e-01 4.50193822e-01 -8.38393331e-01
9.37968612e-01 -2.87142545e-02 -2.93307066e-01 -4.46721584e-01
-8.82461250e-01 -3.34947050e-04 -5.67373633e-01 -3.31746846e-01
6.51420653e-01 -6.50443554e-01 -1.11876810e+00 2.52911538e-01
-9.06403899e-01 -5.32965004e-01 -6.44629478e-01 1.38493553e-01
-5.06402373e-01 -1.68657586e-01 -1.73444510e-01 -1.05952120e+00
4.12943400e-02 -7.30925024e-01 4.06528831e-01 3.82911503e-01
-4.91750091e-01 -8.33342969e-01 3.12339216e-01 5.54594815e-01
5.29281974e-01 2.16205627e-01 7.30132520e-01 -1.03011489e+00
-6.33078933e-01 -3.94372791e-01 3.94245863e-01 2.58058041e-01
-2.55740464e-01 -3.66017550e-01 -5.34799099e-01 -1.86420515e-01
3.61853898e-01 -5.33952057e-01 3.80341440e-01 -9.94759649e-02
6.20587170e-01 -5.76880276e-01 -3.56352895e-01 1.00529157e-01
9.62776124e-01 4.15219575e-01 2.64503419e-01 6.44334733e-01
3.11967224e-01 8.16421628e-01 5.05715489e-01 7.30525792e-01
9.21534076e-02 3.80920082e-01 4.61982451e-02 1.41033545e-01
6.02982223e-01 -2.30845690e-01 3.28209698e-01 9.01815221e-02
-4.35417414e-01 -3.95659983e-01 -1.15342093e+00 4.99370724e-01
-2.32430577e+00 -9.12142098e-01 6.94675267e-01 2.30747986e+00
8.22147787e-01 4.84447539e-01 6.08005404e-01 -6.76413700e-02
5.93539417e-01 -5.14415562e-01 -8.35711181e-01 -2.36125693e-01
1.18493676e-01 -4.09815945e-02 3.38515073e-01 3.43197852e-01
-5.51155865e-01 8.65868032e-01 6.75948381e+00 3.55302513e-01
-4.96006042e-01 -6.72025755e-02 5.56992769e-01 -4.99720991e-01
-1.22906743e-02 3.24422896e-01 -2.02754676e-01 2.59603083e-01
7.92353988e-01 -8.22329819e-01 1.16059721e+00 8.01659942e-01
-1.10208705e-01 -3.77821892e-01 -1.50613582e+00 4.98692274e-01
6.21093214e-02 -8.45539629e-01 -5.58290482e-01 4.11009341e-01
6.07573330e-01 -3.81981581e-01 -1.02694243e-01 1.02971986e-01
8.91469717e-01 -1.25441742e+00 8.40179980e-01 5.52478790e-01
-1.10686861e-01 -6.76020741e-01 6.21750832e-01 1.01069617e+00
-4.35271442e-01 -7.72726953e-01 -1.74219877e-01 -7.70157337e-01
-6.48497045e-02 8.83713439e-02 -9.99068081e-01 2.06210628e-01
4.11727220e-01 1.64349273e-01 -6.12823844e-01 7.82543361e-01
-2.75966078e-01 2.56520957e-01 -4.40428704e-01 -1.53349698e-01
3.04978818e-01 -1.92691103e-01 7.75016785e-01 3.85900795e-01
3.37470621e-02 3.42188507e-01 3.56624305e-01 1.32773077e+00
1.47759065e-01 -1.35407791e-01 -6.84660077e-01 -6.73285723e-01
4.75621909e-01 9.03674066e-01 -6.77715838e-01 -4.77009535e-01
2.56970078e-01 9.57572401e-01 3.33394140e-01 3.70256901e-01
-4.76986498e-01 -4.11771834e-01 3.36573690e-01 2.55557358e-01
1.34442523e-01 -1.97431386e-01 -2.66275197e-01 -1.01041424e+00
4.36306931e-02 -1.33060265e+00 1.98133245e-01 -8.38652194e-01
-1.19818175e+00 5.75598896e-01 6.36062771e-02 -7.07556784e-01
-3.65530431e-01 -4.04808372e-01 -7.00165570e-01 6.52920604e-01
-5.44930816e-01 -5.07233977e-01 -1.37392953e-01 2.88690776e-01
4.08954144e-01 -5.10205269e-01 3.15751880e-01 -3.31747681e-01
-5.19643724e-01 3.77613127e-01 -2.74771810e-01 -2.29066998e-01
3.97766620e-01 -1.06783867e+00 3.22984070e-01 6.00896299e-01
1.04782686e-01 8.66254210e-01 1.09050322e+00 -7.11652637e-01
-1.35371506e+00 -1.29573286e-01 5.85404634e-01 -7.37849593e-01
1.09498107e+00 -4.16680247e-01 -6.43872082e-01 6.61451340e-01
2.74745226e-01 -4.77947295e-01 4.81779486e-01 -7.24515459e-03
-7.68943271e-03 2.24053904e-01 -1.19092107e+00 7.75922954e-01
1.16183579e+00 -2.58044422e-01 -8.93697858e-01 4.67476547e-01
6.64826632e-01 2.76888581e-03 -4.56187069e-01 -9.51268151e-02
5.10597587e-01 -8.90747070e-01 7.14206457e-01 -7.83259511e-01
2.99416721e-01 -2.98121423e-01 1.45445570e-01 -1.47561991e+00
-4.21163976e-01 -1.28046298e+00 6.32495880e-02 6.86391473e-01
3.31310809e-01 -6.50669932e-01 4.96865571e-01 1.01909566e+00
3.44268858e-01 -5.41017175e-01 -7.60099947e-01 -7.48988569e-01
3.21467000e-04 1.38920635e-01 2.43239015e-01 8.96234870e-01
4.96358186e-01 4.70763147e-01 -3.44083518e-01 -1.39325038e-01
8.80755365e-01 -1.00130670e-01 9.00127232e-01 -1.58656454e+00
-8.49784553e-01 -6.75889194e-01 -2.20544875e-01 -7.83374965e-01
1.20472394e-01 -5.60444832e-01 1.87573656e-01 -7.11115122e-01
2.08241880e-01 -1.27436489e-01 -2.60968685e-01 6.27309263e-01
-1.13417506e-01 -1.61892697e-01 3.45756471e-01 2.94460326e-01
-7.74717093e-01 3.25290114e-01 1.07450700e+00 4.66199890e-02
-5.51641047e-01 2.97420360e-02 -1.22535360e+00 1.16121793e+00
6.43866777e-01 -6.79710984e-01 -4.50447917e-01 -1.67822123e-01
7.81665206e-01 3.20590258e-01 4.71447110e-01 -9.10758257e-01
7.15629697e-01 -4.07139093e-01 2.26321831e-01 3.09624016e-01
3.14472951e-02 -1.04135609e+00 1.62267298e-01 8.10755968e-01
-4.98605728e-01 3.16852629e-01 3.56492065e-02 4.23654228e-01
1.62628815e-01 -3.50060195e-01 5.71795464e-01 -5.63885331e-01
-1.61669940e-01 -4.26218629e-01 -5.69124401e-01 1.14554420e-01
1.24085891e+00 -1.33339271e-01 -3.36270422e-01 -5.00771523e-01
-7.85250604e-01 7.30048239e-01 7.54924893e-01 -2.61182375e-02
3.43912840e-01 -9.36564565e-01 -7.01350570e-01 1.41996115e-01
-5.51125780e-02 -2.58693159e-01 -1.60455525e-01 8.91598403e-01
-4.19987351e-01 4.00452651e-02 -3.46901298e-01 -2.21913233e-01
-1.03703701e+00 9.90056336e-01 4.80512679e-01 -2.15875149e-01
-3.87132674e-01 9.66027856e-01 2.81902879e-01 -3.63367125e-02
3.22673023e-01 2.01961383e-01 3.40749383e-01 -1.71832502e-01
4.32265818e-01 5.72448850e-01 -4.91032869e-01 9.44498825e-05
-3.16732287e-01 2.30945796e-01 -9.28809345e-02 -5.31454861e-01
1.30966377e+00 1.69403777e-02 -9.44291875e-02 5.38364589e-01
3.72393280e-01 -8.49755183e-02 -1.11346388e+00 -3.09187561e-01
3.92295927e-01 -6.87959075e-01 7.40279853e-02 -1.20922625e+00
-4.46966678e-01 4.77196485e-01 2.04947412e-01 3.43112111e-01
9.94199157e-01 3.28514948e-02 2.85883576e-01 1.08906841e+00
6.99220359e-01 -1.31414855e+00 6.87507212e-01 5.39080240e-02
9.81281102e-01 -1.06255007e+00 1.53766498e-01 2.66918153e-01
-1.04786551e+00 6.99722886e-01 7.94671893e-01 -3.76858652e-01
3.60007882e-01 2.25118786e-01 -9.10350978e-02 -5.32170177e-01
-1.25227869e+00 6.77145720e-02 6.32312940e-03 2.34493002e-01
-3.94364446e-02 -1.73183024e-01 -1.79488081e-02 5.96623361e-01
-2.60191768e-01 2.79402107e-01 6.14048481e-01 1.37429404e+00
-8.63612950e-01 -9.21323001e-01 -4.73996162e-01 -7.02010235e-03
-2.84524113e-01 6.76298514e-02 -9.29152310e-01 8.16622674e-01
2.00100183e-01 7.89345026e-01 1.06355781e-02 -4.20622677e-01
3.48273993e-01 3.18246275e-01 5.28228879e-01 -4.73205835e-01
-9.91983294e-01 2.53861487e-01 -6.14238158e-02 -6.75462186e-01
-5.67967631e-02 -6.15726531e-01 -8.67543340e-01 -3.31686258e-01
-2.09405467e-01 4.38949257e-01 3.65530908e-01 8.49071503e-01
3.46202344e-01 3.30483466e-01 7.86018670e-01 -7.70396829e-01
-9.70988870e-01 -9.38103855e-01 -5.32927334e-01 7.70152807e-02
2.24993870e-01 -6.23925328e-01 -7.59625494e-01 -6.25703633e-02] | [4.187417984008789, 1.726456880569458] |
517ff363-6615-4f41-9c48-cb74e5760368 | iterative-in-iterative-super-resolution | 2306.14487 | null | https://arxiv.org/abs/2306.14487v1 | https://arxiv.org/pdf/2306.14487v1.pdf | Iterative-in-Iterative Super-Resolution Biomedical Imaging Using One Real Image | Deep learning-based super-resolution models have the potential to revolutionize biomedical imaging and diagnoses by effectively tackling various challenges associated with early detection, personalized medicine, and clinical automation. However, the requirement of an extensive collection of high-resolution images presents limitations for widespread adoption in clinical practice. In our experiment, we proposed an approach to effectively train the deep learning-based super-resolution models using only one real image by leveraging self-generated high-resolution images. We employed a mixed metric of image screening to automatically select images with a distribution similar to ground truth, creating an incrementally curated training data set that encourages the model to generate improved images over time. After five training iterations, the proposed deep learning-based super-resolution model experienced a 7.5\% and 5.49\% improvement in structural similarity and peak-signal-to-noise ratio, respectively. Significantly, the model consistently produces visually enhanced results for training, improving its performance while preserving the characteristics of original biomedical images. These findings indicate a potential way to train a deep neural network in a self-revolution manner independent of real-world human data. | ['Xun Guan', 'Jian Song', 'Shijie Deng', 'Ying Xiao', 'Benqi Zhao', 'Xinyue Wang', 'Yuanzheng Ma'] | 2023-06-26 | null | null | null | null | ['super-resolution'] | ['computer-vision'] | [ 6.91052496e-01 3.08855146e-01 2.31055722e-01 -3.02037716e-01
-1.16177213e+00 -1.34627381e-02 3.00341159e-01 -8.59283283e-03
-3.67061228e-01 7.49811590e-01 2.42251195e-02 1.50537446e-01
-1.74582005e-01 -7.74085879e-01 -5.06475925e-01 -9.05094683e-01
-1.03008904e-01 1.97858408e-01 1.52982816e-01 -9.75621268e-02
7.90013000e-03 6.56116724e-01 -1.35759890e+00 6.03541374e-01
1.04379737e+00 8.15109313e-01 5.27852595e-01 5.55335641e-01
2.11887255e-01 7.00364470e-01 -6.08390689e-01 -9.10472199e-02
2.08787799e-01 -5.42681813e-01 -6.50644600e-01 1.10331960e-01
5.17548978e-01 -4.99398798e-01 -3.16335797e-01 1.03852832e+00
9.84276354e-01 -1.51362553e-01 3.54433864e-01 -2.31671676e-01
-1.01916993e+00 5.79637885e-01 -8.59057426e-01 6.26237929e-01
2.01619074e-01 2.25655094e-01 3.64762276e-01 -7.59421170e-01
8.20116699e-01 8.70433509e-01 6.46252453e-01 7.91879416e-01
-1.43205380e+00 -7.10443139e-01 -3.89948159e-01 1.20521116e-03
-1.38086867e+00 -4.30885851e-01 7.03342199e-01 -3.22932065e-01
3.84470522e-01 3.77321780e-01 4.20899153e-01 9.69375968e-01
4.44805712e-01 3.70062381e-01 1.28198159e+00 -2.21880823e-01
2.04820987e-02 2.60257632e-01 -2.42504850e-01 6.11083746e-01
4.02752161e-01 1.21752203e-01 -3.86221081e-01 -4.91103418e-02
1.33695972e+00 -1.06750906e-01 -4.93193924e-01 -6.69193789e-02
-1.19434774e+00 4.60734904e-01 6.54489875e-01 7.92701781e-01
-5.39026141e-01 -3.04161191e-01 1.01884723e-01 -1.02067411e-01
4.64141965e-01 8.02338958e-01 -1.05174258e-01 3.04170698e-01
-9.03818667e-01 -4.89297733e-02 -1.06172442e-01 2.88905472e-01
2.15844527e-01 3.58192474e-01 -3.88283908e-01 9.20161128e-01
-1.90152586e-01 4.30447519e-01 8.54599953e-01 -1.05807889e+00
-1.55209228e-01 5.64736009e-01 1.90868735e-01 -1.05483043e+00
-5.78898728e-01 -1.23662794e+00 -1.18565726e+00 2.49489948e-01
1.83738470e-01 -1.04080096e-01 -1.06404400e+00 1.54443860e+00
3.86930734e-01 2.62979269e-01 9.88161638e-02 1.01763988e+00
1.04568958e+00 2.13851333e-01 8.38750750e-02 -4.60270405e-01
1.23151577e+00 -4.44146246e-01 -9.21960294e-01 3.79449688e-02
3.18612099e-01 -4.50164318e-01 1.04225004e+00 3.42877328e-01
-1.30642748e+00 -7.43677139e-01 -1.17064452e+00 2.30676159e-01
1.40021279e-01 2.23150849e-01 2.50924647e-01 3.64576757e-01
-1.21208477e+00 8.31341267e-01 -7.02056646e-01 -3.76923382e-02
8.29916060e-01 3.41705561e-01 -4.50791121e-01 -8.09367970e-02
-1.10475588e+00 7.80481339e-01 2.15165913e-01 2.77696550e-02
-8.09373200e-01 -1.12028122e+00 -5.65816402e-01 -1.51637763e-01
-2.23426551e-01 -8.06112289e-01 9.36671793e-01 -7.80207753e-01
-1.27680385e+00 1.13560247e+00 -2.60770805e-02 -4.35101658e-01
5.84349394e-01 -1.63159892e-01 -6.17955327e-01 5.68995714e-01
1.88079208e-01 5.38798869e-01 8.19605708e-01 -1.53839886e+00
-6.19683564e-01 -6.00806713e-01 -4.59464252e-01 1.17488481e-01
-3.69065255e-01 -2.60434821e-02 -1.63072392e-01 -5.98294377e-01
1.83424294e-01 -6.77513719e-01 -5.37297308e-01 -3.73845138e-02
-1.65811449e-01 4.94415760e-01 6.47016525e-01 -7.52363503e-01
1.00762415e+00 -2.01584435e+00 -1.74328163e-01 3.46434154e-02
6.71843886e-01 6.00497544e-01 -2.34535560e-01 -3.39246064e-01
-2.71852523e-01 9.83617008e-02 -1.88512847e-01 4.89203483e-02
-9.32882071e-01 -3.50938767e-01 1.91808373e-01 5.93403578e-01
1.16764203e-01 9.65124786e-01 -1.09789455e+00 -6.59964919e-01
2.42171377e-01 9.83774185e-01 -2.43490174e-01 2.04568073e-01
3.26620162e-01 1.13968873e+00 -3.46227169e-01 4.41691667e-01
7.69820511e-01 -7.86899984e-01 2.77408689e-01 -4.61815804e-01
1.15035363e-01 -3.52120101e-01 -8.74973595e-01 1.50078905e+00
-3.37350011e-01 6.59278035e-01 5.10105342e-02 -7.98276007e-01
9.99820769e-01 4.87636656e-01 1.04040253e+00 -1.24526405e+00
1.35703847e-01 2.23939847e-02 1.28569752e-02 -5.43464124e-01
6.19908124e-02 -3.60791773e-01 4.47649211e-01 2.07627162e-01
-2.33759269e-01 2.79819787e-01 -1.77566767e-01 -1.08001128e-01
1.02595150e+00 -3.09721053e-01 1.78705174e-02 -5.15438914e-02
5.00450790e-01 -1.59076199e-01 3.94495547e-01 7.49582529e-01
-1.99736938e-01 8.83555651e-01 -1.21206596e-01 -5.79520524e-01
-1.20218277e+00 -1.17911375e+00 -5.28627753e-01 6.08964264e-01
-1.79580990e-02 1.79520443e-01 -8.09897661e-01 -3.07853460e-01
-4.56105709e-01 3.34580928e-01 -7.11994588e-01 -2.63853550e-01
-6.62778854e-01 -1.03505576e+00 3.97538424e-01 3.50790590e-01
6.61958098e-01 -9.92025375e-01 -1.04312623e+00 3.43499959e-01
-3.71580303e-01 -1.40727806e+00 -1.86616853e-01 -3.34372222e-01
-1.12296033e+00 -8.55943382e-01 -1.21753728e+00 -8.82844627e-01
9.25536752e-01 1.92116410e-01 9.77307379e-01 9.56472103e-03
-9.18539941e-01 2.93527055e-03 -2.02980310e-01 -2.76952296e-01
-6.31348431e-01 -3.21162671e-01 -6.01238795e-02 -2.72042081e-02
-1.34646446e-01 -5.45248151e-01 -1.14199352e+00 1.31747708e-01
-7.14202642e-01 3.13586086e-01 9.70993161e-01 1.01408279e+00
9.82220590e-01 7.59438798e-02 7.34833539e-01 -9.63583469e-01
6.34594023e-01 -5.77116907e-02 -3.12720984e-01 1.40592977e-01
-7.35345781e-01 6.87237084e-02 5.44725299e-01 -5.72346985e-01
-1.29660451e+00 1.13181345e-01 -7.61997402e-02 -5.12761295e-01
-8.64256471e-02 3.37566674e-01 2.94256270e-01 -2.58401543e-01
1.13939416e+00 3.75892073e-01 2.27782145e-01 -2.49708593e-01
2.09187582e-01 6.16224468e-01 8.38780284e-01 -3.29368785e-02
6.55062973e-01 7.35174000e-01 5.52836806e-02 -7.53293395e-01
-9.27937269e-01 -2.48456970e-01 -5.03398180e-01 -2.35967785e-01
9.10390854e-01 -9.89222646e-01 -3.59567285e-01 3.87906998e-01
-7.61320472e-01 -1.66812763e-01 -3.51430446e-01 4.94309574e-01
-2.47348234e-01 2.95073390e-01 -7.75112629e-01 -5.54792047e-01
-9.42042232e-01 -1.01981568e+00 1.05375910e+00 4.10753608e-01
-1.72665194e-01 -7.25017130e-01 3.12976725e-02 5.41797042e-01
6.72454178e-01 8.10995221e-01 8.25370073e-01 -2.95036107e-01
-4.47742105e-01 -1.99843332e-01 -4.32014376e-01 2.67465770e-01
6.08023584e-01 -4.14804578e-01 -1.10595977e+00 -4.36119854e-01
1.71731964e-01 -2.75149196e-02 5.08162200e-01 7.19417870e-01
1.26490474e+00 -1.24449655e-01 -3.24200898e-01 7.43412256e-01
1.43615842e+00 2.97618568e-01 6.45961583e-01 2.87472814e-01
6.38349116e-01 3.74119073e-01 4.82940465e-01 4.05658633e-01
-1.40406832e-01 5.06849468e-01 2.47747406e-01 -8.39977443e-01
-3.93878311e-01 -5.43658622e-03 -2.24332914e-01 3.42592269e-01
-2.08182871e-01 3.62703204e-01 -7.81613052e-01 6.61340773e-01
-1.36560726e+00 -1.04909253e+00 -4.55772243e-02 2.23801661e+00
1.02999461e+00 2.98458003e-02 -7.01646507e-02 -3.06070931e-02
8.92246306e-01 -2.52677292e-01 -8.21229637e-01 1.15889810e-01
-8.54882449e-02 4.52292323e-01 4.16229725e-01 1.82255462e-01
-9.88086224e-01 6.04646981e-01 7.00535822e+00 6.23486876e-01
-1.58214176e+00 1.13924682e-01 1.02699435e+00 -1.38956740e-01
-3.19705993e-01 -8.42088938e-01 -4.01787281e-01 2.39851579e-01
1.01072776e+00 -2.79911101e-01 1.08358137e-01 5.77875197e-01
6.61918998e-01 1.64802119e-01 -8.61504316e-01 1.18238735e+00
2.86721233e-02 -1.67517102e+00 2.64411062e-01 9.02703032e-02
9.47446167e-01 -1.33011237e-01 4.93798822e-01 -2.67294735e-01
-1.47364300e-03 -1.36504066e+00 1.13979630e-01 5.92802763e-01
1.41126060e+00 -6.78990841e-01 7.42108643e-01 9.09506232e-02
-8.55037332e-01 -5.81971668e-02 -1.68916553e-01 5.74908912e-01
-7.46650472e-02 8.12450469e-01 -1.13542521e+00 6.14932001e-01
7.96545565e-01 3.84978771e-01 -6.40739679e-01 9.73048806e-01
2.08886266e-01 3.64923090e-01 2.66901106e-01 5.13089597e-01
-1.80685356e-01 1.10100411e-01 5.53732812e-01 1.15096903e+00
3.53756487e-01 2.52866715e-01 -1.65455624e-01 9.31858957e-01
1.22218937e-01 4.19548713e-02 -3.72685611e-01 1.28954142e-01
4.30962592e-01 1.43423140e+00 -8.20768297e-01 -4.08020914e-01
-1.45251617e-01 8.44406545e-01 -2.33767759e-02 3.36195797e-01
-8.61423075e-01 -1.88143417e-01 2.28505194e-01 6.61858857e-01
3.80699337e-02 2.14653432e-01 -5.67699969e-01 -8.12949359e-01
-1.23444133e-01 -1.06993806e+00 3.41058910e-01 -8.73066485e-01
-1.08511269e+00 1.09891772e+00 -1.85731575e-01 -1.29361236e+00
-2.57463425e-01 -2.76854038e-01 -3.81091475e-01 8.74439001e-01
-1.40202320e+00 -1.02297544e+00 -6.42750323e-01 5.32446742e-01
3.03452641e-01 -1.19209416e-01 8.01990926e-01 4.71869886e-01
-3.85019332e-01 6.86124027e-01 2.12827772e-01 1.07838057e-01
6.83687747e-01 -1.14117336e+00 5.03639057e-02 8.27271461e-01
-1.16774917e-01 5.59232414e-01 9.29457664e-01 -4.48452085e-01
-1.03237295e+00 -1.29581988e+00 2.67142326e-01 -2.44440764e-01
1.78853214e-01 2.29638323e-01 -1.12866807e+00 1.73747748e-01
-4.24619950e-02 1.61890760e-01 6.01445198e-01 -3.04208189e-01
1.48895532e-01 -3.30128282e-01 -1.64642787e+00 5.81508219e-01
8.22403431e-01 -3.51350546e-01 -3.25747669e-01 2.26192310e-01
6.53643548e-01 -6.70261025e-01 -1.15110123e+00 6.68021798e-01
6.01693451e-01 -9.31427658e-01 1.19446063e+00 -2.57555068e-01
5.63998997e-01 -1.53956965e-01 2.94873983e-01 -1.13970137e+00
-6.24221981e-01 -3.40117723e-01 -6.43502101e-02 6.89689994e-01
4.10332710e-01 -5.36014438e-01 9.44971621e-01 5.20525038e-01
-7.76233152e-02 -9.54593420e-01 -8.01673710e-01 -4.57587004e-01
8.03857818e-02 1.28507286e-01 3.71022224e-01 9.22851682e-01
-2.46024355e-01 9.41361040e-02 -2.25738585e-01 3.07391018e-01
9.60718513e-01 -8.23012646e-03 3.28922808e-01 -1.14166522e+00
-1.29834354e-01 -3.11798573e-01 -3.26666355e-01 -4.62996364e-01
-2.82931447e-01 -7.88131177e-01 -8.22453871e-02 -1.79262030e+00
5.64190269e-01 -2.36047193e-01 -6.98491454e-01 1.38910159e-01
-3.46774966e-01 6.83665514e-01 -2.83037871e-01 1.80444002e-01
-3.34266871e-01 3.13331597e-02 1.98953152e+00 -8.14194232e-02
-3.25631857e-01 -1.78700224e-01 -1.09271729e+00 4.77344751e-01
8.95199895e-01 -1.66615441e-01 -5.61697662e-01 -3.06627542e-01
-2.46975884e-01 2.47062683e-01 3.21279317e-01 -1.29360127e+00
-2.35520322e-02 1.16210110e-01 1.00564003e+00 -2.87250102e-01
2.00787857e-01 -5.18261075e-01 4.15574700e-01 6.87279344e-01
-5.09326816e-01 -2.43177027e-01 2.86161065e-01 3.14421773e-01
-3.72593999e-02 1.90282047e-01 1.30878866e+00 -4.25201535e-01
-4.48150814e-01 3.28955948e-01 -1.73880309e-01 -1.08109348e-01
1.04950559e+00 -4.18040544e-01 -1.31230071e-01 -3.57136995e-01
-8.82725596e-01 -1.49729490e-01 2.39018068e-01 1.87767729e-01
1.05611920e+00 -1.11047673e+00 -1.06379485e+00 1.86691687e-01
-6.22567125e-02 9.04430896e-02 8.10909390e-01 7.99542487e-01
-3.96979928e-01 2.66806602e-01 -5.05193889e-01 -8.51656199e-01
-1.38180757e+00 4.51661497e-01 5.96915841e-01 -5.32909274e-01
-1.08693182e+00 7.04094112e-01 3.23449254e-01 7.34120905e-02
-3.30826193e-02 -1.09691255e-01 -4.14491802e-01 -4.12000895e-01
9.48055804e-01 2.14075834e-01 8.45544562e-02 -5.65431237e-01
-2.40969375e-01 5.64832687e-01 -3.86456221e-01 5.33578545e-02
1.40352464e+00 -1.76629201e-01 2.62678206e-01 1.28750922e-02
9.18210328e-01 -1.05725080e-01 -1.33680296e+00 -3.00372392e-01
-2.88380325e-01 -4.42545027e-01 4.00949836e-01 -1.10690808e+00
-1.36985052e+00 5.75414777e-01 1.34105432e+00 -1.31276414e-01
1.36773086e+00 5.33464439e-02 8.85577142e-01 -2.36792583e-02
3.67499769e-01 -8.58310223e-01 3.82607043e-01 -1.78774133e-01
9.34165120e-01 -1.21050537e+00 1.82595879e-01 -2.37147823e-01
-7.20919132e-01 8.88636053e-01 6.89585268e-01 -2.37659905e-02
3.14009190e-01 3.58546972e-01 3.44091475e-01 -2.45803386e-01
-4.59561825e-01 9.54621285e-02 3.89472961e-01 9.15994942e-01
6.00268364e-01 -2.69232634e-02 -7.68084824e-02 4.66619551e-01
-2.45349184e-02 4.03206259e-01 4.93596971e-01 5.95236540e-01
-4.83497858e-01 -6.01798654e-01 -4.26661491e-01 6.00886524e-01
-8.08487356e-01 2.46990882e-02 1.54798716e-01 5.55999994e-01
1.07504435e-01 7.72908211e-01 2.36735959e-03 -2.63525397e-01
3.81119192e-01 -4.81175154e-01 6.22162521e-01 -5.32846391e-01
-2.01172262e-01 2.59733766e-01 -2.10602939e-01 -5.20118654e-01
-4.05619025e-01 -3.24096620e-01 -1.35191107e+00 -1.15257679e-02
-1.12294145e-01 -1.38184533e-01 4.26795304e-01 6.94749773e-01
6.36567175e-01 1.06208611e+00 6.86248541e-01 -7.40552604e-01
-3.26501667e-01 -8.31058204e-01 -3.98314774e-01 4.81997937e-01
4.67396379e-01 -4.13745373e-01 -3.39762010e-02 3.64528447e-01] | [13.660624504089355, -2.3107597827911377] |
b229d5d9-44cd-4eb6-886d-ca0654a0e336 | deep-multi-task-learning-to-recognise-subtle | null | null | http://openaccess.thecvf.com/content_ECCV_2018/html/Guosheng_Hu_Deep_Multi-Task_Learning_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Guosheng_Hu_Deep_Multi-Task_Learning_ECCV_2018_paper.pdf | Deep Multi-Task Learning to Recognise Subtle Facial Expressions of Mental States | Facial expression recognition is a topical task. However, very little research investigates subtle expression recognition, which is important for mental activity analysis, deception detection, etc. We address subtle expression recognition through convolutional neural networks (CNNs) by developing multi-task learning (MTL) to effectively leverage a side task: facial landmark detection. Existing MTL methods follow a design pattern of shared bottom CNN layers and task-specific top layers. However, the sharing architecture is usually heuristically chosen, as it is difficult to decide which layers should be shared. Our approach is composed of (1) a novel MTL framework that automatically learns which layers to share through optimisation under tensor trace norm regularisation and (2) an invariant representation learning approach that allows the CNN to leverage tasks defined on disjoint datasets without suffering from dataset distribution shift. To advance subtle expression recognition, we contribute a Large-scale Subtle Emotions and Mental States in the Wild database (LSEMSW). LSEMSW includes a variety of cognitive states as well as basic emotions. It contains 176K images, manually annotated with 13 emotions, and thus provides the first subtle expression dataset large enough for training deep CNNs. Evaluations on LSEMSW and 300-W (landmark) databases show the effectiveness of the proposed methods. In addition, we investigate transferring knowledge learned from LSEMSW database to traditional (non-subtle) expression recognition. We achieve very competitive performance on Oulu-Casia NIR&Vis and CK+ databases via transfer learning. | ['Timothy Hospedales', 'Zhihong Zhang', 'Guosheng Hu', 'Yongxin Yang', 'Yang Hua', 'Neil Robertson', 'Zehao Yu', 'Yang Yuan', 'Li Liu', 'Fumin Shen', 'Ling Shao'] | 2018-09-01 | null | null | null | eccv-2018-9 | ['deception-detection'] | ['miscellaneous'] | [ 1.51221350e-01 -3.02249551e-01 -2.25869343e-01 -7.86807001e-01
-7.32859075e-01 -2.01573208e-01 5.30840039e-01 -4.82427776e-01
-4.74350214e-01 5.51227629e-01 1.82136953e-01 1.85812369e-01
-3.84060070e-02 -2.80007571e-01 -3.34054440e-01 -9.95382547e-01
-1.64626703e-01 4.25508618e-02 -4.95669127e-01 -6.90324008e-01
1.67324056e-03 7.87483096e-01 -1.33620632e+00 5.82046926e-01
3.30973685e-01 1.79744124e+00 -6.75341666e-01 4.28858876e-01
-1.01828679e-01 1.36101925e+00 -6.76539838e-01 -6.98616207e-01
1.09120049e-01 -4.61441755e-01 -8.32698166e-01 2.77030077e-02
5.31344056e-01 -1.38970762e-01 1.74106807e-01 9.07243311e-01
7.63723135e-01 3.42306239e-03 5.42070627e-01 -1.64025033e+00
-5.38578451e-01 -3.86603619e-03 -8.83879185e-01 6.59546629e-02
2.31324703e-01 1.12268105e-01 8.25054526e-01 -9.56542790e-01
5.77901900e-01 1.08211648e+00 8.99594426e-01 7.67948747e-01
-1.11385405e+00 -1.18339550e+00 -1.35554358e-01 3.25082779e-01
-1.64587033e+00 -7.77168095e-01 1.19431269e+00 -3.58622104e-01
7.85216272e-01 2.69181699e-01 5.81948936e-01 1.73035979e+00
-1.35448337e-01 1.06022346e+00 1.62524724e+00 -9.93386656e-02
9.38832313e-02 7.47427121e-02 -3.15284878e-02 8.49550188e-01
-6.46146476e-01 -2.77564913e-01 -9.99508381e-01 -3.58423322e-01
3.15706104e-01 -1.50266528e-01 -2.29718447e-01 -1.43512547e-01
-6.54251516e-01 8.21887314e-01 2.61755824e-01 3.62139702e-01
-4.46458429e-01 2.75994748e-01 9.06019390e-01 7.36833274e-01
7.50623703e-01 2.86057323e-01 -4.58296210e-01 -4.76606965e-01
-9.98640776e-01 2.31667116e-01 7.87774026e-01 5.23629606e-01
9.31770980e-01 2.45625079e-01 -3.47095191e-01 1.27510571e+00
4.70708869e-03 4.01660681e-01 5.15701711e-01 -7.55935848e-01
1.04912870e-01 5.87664902e-01 -3.32399338e-01 -1.26646399e+00
-5.26681840e-01 -9.20588523e-02 -1.03662539e+00 3.73712629e-01
1.25266060e-01 -3.12733322e-01 -6.15405738e-01 2.09637570e+00
6.72553703e-02 2.14616582e-01 -3.89851332e-02 1.01300108e+00
7.78576791e-01 4.07013983e-01 1.82898819e-01 -2.26578876e-01
1.27081132e+00 -9.18303847e-01 -8.41592133e-01 -9.82036628e-03
9.94930625e-01 -4.53960449e-01 9.89018321e-01 7.30731249e-01
-8.55360627e-01 -3.37440789e-01 -8.73267174e-01 -1.58147782e-01
-5.05401254e-01 3.13702077e-01 1.08945239e+00 7.86568940e-01
-1.10734713e+00 3.30445439e-01 -3.55795503e-01 -1.09320708e-01
9.94804859e-01 4.78826672e-01 -7.95506060e-01 2.31258065e-01
-1.14307475e+00 9.50094044e-01 -6.97891600e-03 5.59714317e-01
-8.53236496e-01 -5.59446931e-01 -1.03081000e+00 -1.65220365e-01
3.00788969e-01 -1.91195861e-01 1.01271582e+00 -1.96103930e+00
-1.94588399e+00 1.34479201e+00 -1.40759647e-01 -1.21988893e-01
3.75069052e-01 -2.29103386e-01 -5.81926346e-01 1.44690961e-01
-1.78214446e-01 4.99567509e-01 1.12867200e+00 -9.62801874e-01
6.80762809e-03 -5.94145834e-01 -1.77422360e-01 -1.45196870e-01
-7.52077460e-01 5.64893126e-01 -2.07212865e-01 -5.66856921e-01
-4.04968023e-01 -9.06825364e-01 3.00937612e-02 3.80744755e-01
-2.90734589e-01 -4.03853863e-01 1.03602397e+00 -4.73944694e-01
9.05299485e-01 -2.38822460e+00 2.85031348e-01 3.71129304e-01
3.39979857e-01 2.79618621e-01 -4.80248064e-01 -3.91989127e-02
-3.75639617e-01 -7.38277435e-02 -3.05018928e-02 -8.09061825e-01
2.04393581e-01 2.90661246e-01 -3.34522218e-01 5.92277348e-01
5.54474413e-01 1.22499120e+00 -6.54259562e-01 -4.30582762e-01
-2.27787048e-01 4.86880273e-01 -3.20445597e-01 2.05342591e-01
8.44817236e-02 3.13568562e-01 -4.52487558e-01 1.20853245e+00
7.00650275e-01 -8.77316017e-03 -7.53840953e-02 -4.17447716e-01
1.46108255e-01 -4.03933913e-01 -7.71562457e-01 1.84465063e+00
-6.40672684e-01 6.89424038e-01 3.31308931e-01 -1.23933578e+00
1.24956763e+00 3.27842355e-01 6.83851659e-01 -8.74956548e-01
5.09766519e-01 2.28370368e-01 -2.95205086e-01 -6.86455548e-01
2.72625387e-01 -4.94179338e-01 -3.87457877e-01 4.46271271e-01
5.38297713e-01 1.21322908e-01 -1.67390004e-01 -1.07107215e-01
1.05525494e+00 1.98610529e-01 3.12828928e-01 -1.48422450e-01
5.71166277e-01 -4.47851866e-01 8.42879236e-01 2.68006176e-01
-5.55376947e-01 2.64619321e-01 8.97816420e-01 -8.24016869e-01
-4.92938727e-01 -7.84653723e-01 -8.32993761e-02 1.54792857e+00
-3.25158417e-01 -5.25479436e-01 -4.65038121e-01 -7.55031526e-01
-1.45271227e-01 1.80864498e-01 -1.18687963e+00 -4.48220551e-01
-2.67888337e-01 -8.66356075e-01 1.28653002e+00 4.81890440e-01
6.88950479e-01 -1.25899696e+00 -4.06729788e-01 -1.76314726e-01
-1.46253422e-01 -1.17654538e+00 -2.28183240e-01 2.98123986e-01
-2.72123843e-01 -8.91964734e-01 -7.05816031e-01 -4.17490572e-01
3.87979209e-01 -2.58993775e-01 1.05974019e+00 6.48753569e-02
-4.15979862e-01 3.03135484e-01 -3.06628197e-01 -4.66167837e-01
-8.19053203e-02 -1.56791762e-01 4.58732806e-02 9.02632177e-01
6.68881774e-01 -7.72540331e-01 -3.32940519e-01 2.28562161e-01
-7.58417606e-01 -1.85742781e-01 7.00975776e-01 9.26129460e-01
3.04678679e-01 -5.44089317e-01 4.94279861e-01 -5.80286384e-01
7.55313218e-01 -4.27008331e-01 -1.40331045e-01 2.67286003e-01
-8.60582367e-02 -1.36638135e-01 4.35965627e-01 -6.62162066e-01
-1.01292121e+00 7.96837360e-02 -2.36773714e-01 -9.54914510e-01
-2.48220280e-01 3.66155297e-01 -2.17962757e-01 -5.83313882e-01
6.36128306e-01 1.83099553e-01 2.75893956e-01 -1.95954293e-01
3.07080060e-01 6.31414592e-01 4.05139387e-01 -7.83015966e-01
3.72756898e-01 5.37705362e-01 1.86007917e-01 -9.16228056e-01
-1.01041806e+00 -3.36179644e-01 -7.09564149e-01 -3.43347877e-01
9.06582654e-01 -9.26813006e-01 -1.08979023e+00 7.65590191e-01
-1.02677298e+00 -5.48300028e-01 -8.16533715e-02 -1.50964558e-02
-6.41499162e-01 1.18648149e-01 -7.04150498e-01 -8.59292626e-01
-4.78246719e-01 -9.97998297e-01 1.39844310e+00 -1.70744330e-01
-6.03640854e-01 -9.40989494e-01 2.06258342e-01 4.54156041e-01
6.58434510e-01 9.12364602e-01 5.68855882e-01 -4.78674769e-01
5.25429435e-02 -2.00362012e-01 -2.63651222e-01 6.17932975e-01
-6.52038902e-02 -1.09477313e-02 -1.39790893e+00 -8.13910365e-02
6.41067848e-02 -1.38577414e+00 9.18624878e-01 -1.12129718e-01
1.52004862e+00 -3.16702694e-01 9.08136442e-02 1.13297677e+00
1.03656566e+00 -1.96695015e-01 8.99369180e-01 2.71877527e-01
7.39060581e-01 7.10634947e-01 2.56920159e-01 6.57368064e-01
2.35814661e-01 9.43050146e-01 3.50508034e-01 -4.26229358e-01
1.12465434e-01 3.58955324e-01 6.37249887e-01 5.90209246e-01
-4.58789438e-01 3.66042286e-01 -7.70875752e-01 2.96531230e-01
-1.74633694e+00 -1.15933728e+00 2.84659296e-01 1.51475835e+00
1.02581799e+00 -3.68664026e-01 1.33969277e-01 -2.49993149e-02
-1.25874877e-02 4.53574836e-01 -4.39728081e-01 -9.40269411e-01
-4.96256888e-01 6.57566130e-01 -6.19595945e-02 1.53667312e-02
-1.13596749e+00 9.87317324e-01 5.30186367e+00 1.12439775e+00
-1.69423306e+00 4.47400838e-01 8.17752004e-01 -5.40410578e-01
1.94557562e-01 -6.24211669e-01 -4.63163316e-01 2.65755087e-01
8.79230440e-01 1.23130567e-01 2.36268401e-01 9.82612252e-01
9.94873270e-02 1.98314622e-01 -9.13022697e-01 1.58595610e+00
5.95373154e-01 -1.02648723e+00 -3.83701846e-02 -5.35870753e-02
5.84367573e-01 -5.77928722e-02 2.55934924e-01 7.29211986e-01
-1.09462790e-01 -1.45380712e+00 5.70652246e-01 5.80674410e-01
9.66759264e-01 -9.93091285e-01 7.26689935e-01 7.74557004e-03
-9.03036356e-01 -2.03735247e-01 -9.96362194e-02 -1.12026542e-01
-5.94308600e-02 3.77308726e-01 -3.18959236e-01 3.41512024e-01
8.58119130e-01 9.96533453e-01 -5.81739724e-01 2.66666412e-01
-3.29504430e-01 4.65733618e-01 -1.70331582e-01 -2.43133511e-02
4.31186587e-01 -2.47206777e-01 9.89366397e-02 1.50909889e+00
1.96692944e-01 2.70407405e-02 4.22907621e-02 8.89391541e-01
-3.80972892e-01 2.56881624e-01 -6.62511408e-01 1.03241240e-03
-3.08056682e-01 1.80929995e+00 -1.78481370e-01 -1.45066440e-01
-3.82388502e-01 1.41441786e+00 6.72185421e-01 3.46914440e-01
-8.72122467e-01 -1.93428248e-01 1.24679947e+00 -3.50909948e-01
-8.42407048e-02 1.09873172e-02 -3.36153507e-02 -1.17301226e+00
2.95649022e-02 -1.05999315e+00 3.27464849e-01 -7.95779347e-01
-1.35039330e+00 7.98520505e-01 -3.26017708e-01 -9.16084766e-01
-2.57679731e-01 -9.08457458e-01 -6.43139660e-01 7.26628423e-01
-1.75628376e+00 -1.60904920e+00 -5.47928631e-01 1.11077154e+00
1.48680598e-01 -2.29676992e-01 1.26341426e+00 5.03923953e-01
-7.74974644e-01 9.95400131e-01 -4.27363932e-01 5.36130011e-01
8.64774168e-01 -1.03193021e+00 -4.54628646e-01 3.67386490e-01
1.85615838e-01 3.68267089e-01 1.52967915e-01 3.10311578e-02
-1.34692514e+00 -1.07373011e+00 5.78575194e-01 -4.86682326e-01
7.33347178e-01 -6.87282264e-01 -8.90829325e-01 6.30523264e-01
6.18637092e-02 4.68795836e-01 1.25468874e+00 3.78264725e-01
-8.41035306e-01 -2.86597729e-01 -1.05091548e+00 5.80067039e-01
1.00139487e+00 -9.96178508e-01 -1.85169369e-01 2.78905749e-01
-3.49258073e-02 -2.49881521e-01 -1.01338816e+00 5.99982679e-01
7.88154364e-01 -1.10356331e+00 6.98518157e-01 -8.03829551e-01
6.68776870e-01 8.50046799e-02 -1.59354582e-01 -1.24743688e+00
-1.04255769e-02 -8.53187203e-01 -1.97925642e-01 1.11064422e+00
1.89596146e-01 -4.79032397e-01 8.19659412e-01 6.11019611e-01
5.63940220e-02 -1.15815902e+00 -1.06838930e+00 -6.68823123e-01
-1.23105220e-01 -6.81077600e-01 4.53991413e-01 1.48974609e+00
-6.32381067e-02 4.71395016e-01 -7.85214543e-01 -2.95680404e-01
3.29972833e-01 2.38929853e-01 1.11314261e+00 -9.96346176e-01
-5.05716093e-02 -8.30519736e-01 -7.61611462e-01 -5.71485102e-01
8.25364172e-01 -9.42288101e-01 -1.95194200e-01 -5.89408994e-01
2.86795527e-01 -4.97884043e-02 -5.60513675e-01 1.19964278e+00
2.03156129e-01 7.42175817e-01 1.29233107e-01 -1.30227476e-01
-8.61983657e-01 1.03203011e+00 1.07977581e+00 -2.23468810e-01
1.40316077e-02 -3.00526947e-01 -7.30750203e-01 9.02614355e-01
6.82949603e-01 -1.62131503e-01 -2.56789595e-01 -9.36250091e-02
4.05597210e-01 -2.80529499e-01 6.29919767e-01 -6.64000869e-01
-1.20379059e-02 -6.14383146e-02 5.18220484e-01 -1.94853377e-02
7.77559876e-01 -6.26057088e-01 -1.70580193e-01 -3.02857403e-02
-2.87747055e-01 -2.39755899e-01 3.92375648e-01 4.90308739e-02
-4.64835256e-01 2.38060400e-01 8.25997174e-01 1.62045479e-01
-1.13739133e+00 6.55610263e-01 -1.74130976e-01 -3.41583043e-02
1.02378678e+00 -2.66557455e-01 -9.72392187e-02 -4.47548151e-01
-7.13808417e-01 6.05868995e-02 1.15642950e-01 5.80912888e-01
7.16131806e-01 -1.65415299e+00 -7.28154242e-01 2.57219881e-01
6.50997818e-01 -4.44976658e-01 1.37675658e-01 1.31168628e+00
-1.23421371e-01 -2.24338755e-01 -5.90088189e-01 -5.37684202e-01
-1.50078881e+00 1.62483290e-01 5.81945181e-01 -1.28574103e-01
-2.30494514e-01 1.07713020e+00 7.35364109e-02 -5.03223062e-01
-2.54445951e-02 1.33454464e-02 -1.89001143e-01 4.09543723e-01
8.37035179e-01 2.05982123e-02 1.69477202e-02 -1.02106965e+00
-3.51019531e-01 6.10150039e-01 1.55021972e-03 1.73111752e-01
1.47747254e+00 1.82697207e-01 -5.37019849e-01 4.29336697e-01
1.82153177e+00 -2.29205400e-01 -1.09337091e+00 -2.71669954e-01
1.78634912e-01 -4.19789940e-01 1.38016656e-01 -7.08437383e-01
-1.30298960e+00 1.05843902e+00 5.82965612e-01 -3.94463390e-01
1.56368852e+00 -2.75929179e-02 7.41497636e-01 5.27310073e-01
3.23196828e-01 -1.11153746e+00 6.08958960e-01 5.16455233e-01
1.19408190e+00 -1.23889649e+00 -3.37954551e-01 3.33133265e-02
-8.21578383e-01 1.19866490e+00 6.35131001e-01 4.38317060e-02
6.05529428e-01 4.45977330e-01 4.49995130e-01 -7.60602534e-01
-7.30354488e-01 -2.10632265e-01 2.07989752e-01 4.45512444e-01
2.90299416e-01 -1.21440321e-01 2.68198490e-01 7.88659930e-01
-1.96879119e-01 3.25926632e-01 2.04358213e-02 7.61592269e-01
1.78362668e-01 -8.72456312e-01 -1.14084207e-01 2.41210133e-01
-6.87391996e-01 1.47411048e-01 -7.31326640e-01 7.58950770e-01
2.98616618e-01 5.32627046e-01 -6.83739111e-02 -3.53844523e-01
3.41195554e-01 3.56134683e-01 5.31405210e-01 -2.63221949e-01
-7.08451629e-01 -2.77519792e-01 1.36765063e-01 -1.16697764e+00
-7.04214036e-01 -4.15476739e-01 -7.87831008e-01 -2.94945568e-01
7.57146776e-02 -1.40100941e-01 5.29083431e-01 1.09654093e+00
3.31861496e-01 3.44846725e-01 7.26754129e-01 -1.04286766e+00
-3.76066238e-01 -9.36598897e-01 -8.58072042e-01 8.94574583e-01
3.97139192e-01 -7.84875095e-01 -2.39483207e-01 -9.62909162e-02] | [13.585808753967285, 1.805151104927063] |
0ed8efb2-87bf-46b8-9200-70be90936cbd | multipath-cnn-with-alpha-matte-inference-for | 2109.14249 | null | https://arxiv.org/abs/2109.14249v1 | https://arxiv.org/pdf/2109.14249v1.pdf | Multipath CNN with alpha matte inference for knee tissue segmentation from MRI | Precise segmentation of knee tissues from magnetic resonance imaging (MRI) is critical in quantitative imaging and diagnosis. Convolutional neural networks (CNNs), which are state of the art, have limitations owing to the lack of image-specific adaptation, such as low tissue contrasts and structural inhomogeneities, thereby leading to incomplete segmentation results. This paper presents a deep learning based automatic segmentation framework for knee tissue segmentation. A novel multipath CNN-based method is proposed, which consists of an encoder decoder-based segmentation network in combination with a low rank tensor-reconstructed segmentation network. Low rank reconstruction in MRI tensor sub-blocks is introduced to exploit the structural and morphological variations in knee tissues. To further improve the segmentation from CNNs, trimap generation, which effectively utilizes superimposed regions, is proposed for defining high, medium and low confidence regions from the multipath CNNs. The secondary path with low rank reconstructed input mitigates the conditions in which the primary segmentation network can potentially fail and overlook the boundary regions. The outcome of the segmentation is solved as an alpha matting problem by blending the trimap with the source input. Experiments on Osteoarthritis Initiative (OAI) datasets and a self prepared scan validate the effectiveness of the proposed method. We specifically demonstrate the application of the proposed method in a cartilage segmentation based thickness map for diagnosis purposes. | ['Weitian Chen', 'Yongcheng Yao', 'Basim Azam', 'Sheheryar Khan'] | 2021-09-29 | null | null | null | null | ['image-matting'] | ['computer-vision'] | [ 4.60917413e-01 3.15206409e-01 1.43041342e-01 -1.99097365e-01
-8.39865685e-01 -5.95683195e-02 1.35942593e-01 -4.93346713e-02
-4.70171899e-01 6.31964326e-01 1.80946872e-01 -5.81341572e-02
-2.94656634e-01 -5.38369536e-01 -5.98654032e-01 -8.41247201e-01
-2.00851083e-01 5.64701140e-01 5.59228539e-01 -8.48984271e-02
2.91840076e-01 3.61394972e-01 -1.28439558e+00 6.26483321e-01
1.01165366e+00 1.10669494e+00 3.08405846e-01 4.96685237e-01
-2.36205444e-01 4.79829401e-01 -4.14135695e-01 -2.02596132e-02
3.66148502e-01 -3.06676030e-01 -8.58789563e-01 3.01312655e-01
3.24435920e-01 -2.70399451e-01 -2.14161724e-02 1.02140474e+00
4.31654900e-01 -2.97818631e-01 7.60450065e-01 -8.35379422e-01
-1.17002159e-01 5.32630622e-01 -6.83031917e-01 3.50519568e-01
-2.90733099e-01 1.88369632e-01 6.38276100e-01 -8.98087800e-01
6.32367194e-01 8.15888107e-01 8.33573401e-01 2.23351136e-01
-1.21502304e+00 -4.52785581e-01 -3.39725822e-01 -1.81920063e-02
-1.32190621e+00 -1.26772532e-02 8.47173452e-01 -7.68570721e-01
7.45176136e-01 2.14732096e-01 9.41551268e-01 5.21964073e-01
7.15948999e-01 6.65147841e-01 1.38233340e+00 -2.34379217e-01
1.57976076e-01 -2.60266364e-01 -6.03744611e-02 6.68183565e-01
1.36447877e-01 -2.15065237e-02 -2.51842111e-01 2.02708021e-01
1.20632279e+00 -3.77230078e-01 -1.55131981e-01 -4.05436903e-01
-1.43080378e+00 5.79173028e-01 7.97781527e-01 4.82466161e-01
-5.76164365e-01 6.00427240e-02 4.85327005e-01 -4.99027967e-02
4.71152872e-01 3.84108961e-01 -1.93604484e-01 2.92020649e-01
-1.49023461e+00 8.69118869e-02 4.59715754e-01 3.98480564e-01
6.18865967e-01 1.57544017e-01 -1.51009008e-01 9.63255107e-01
3.08225363e-01 6.70180023e-02 6.74938321e-01 -9.23548996e-01
2.25025088e-01 7.60908008e-01 -4.93503183e-01 -9.38458800e-01
-6.17902696e-01 -8.02810669e-01 -1.02307630e+00 3.48709911e-01
4.50390190e-01 1.57233246e-03 -1.41628706e+00 1.18994498e+00
4.81291294e-01 1.42917410e-01 -2.64995962e-01 1.34425759e+00
5.95549047e-01 2.91407079e-01 -1.57487571e-01 5.68564497e-02
1.29788232e+00 -9.44940567e-01 -5.17513394e-01 -1.43077448e-01
7.53786027e-01 -7.60208666e-01 7.59714246e-01 4.56681818e-01
-1.15424156e+00 -4.03149396e-01 -1.19295144e+00 -1.96320098e-02
-1.29183173e-01 2.78479159e-01 4.96177375e-01 4.61348712e-01
-1.12326407e+00 4.90292490e-01 -1.00725567e+00 1.06724702e-01
5.07226765e-01 7.63815880e-01 -3.58530074e-01 1.20719239e-01
-1.03353846e+00 7.23984540e-01 5.50110996e-01 6.79052353e-01
-6.96196795e-01 -8.22923720e-01 -5.48782706e-01 -3.15163106e-01
2.65255451e-01 -7.74058402e-01 8.27815294e-01 -1.25903809e+00
-1.37737000e+00 8.45479190e-01 3.52078170e-01 -5.25090158e-01
6.82044029e-01 -1.02497995e-01 -1.59324205e-03 5.94444990e-01
2.63619155e-01 8.94947588e-01 8.13876152e-01 -1.32947004e+00
-3.05453777e-01 -4.91597921e-01 -2.57068902e-01 1.83855250e-01
-9.02389083e-03 -2.99949408e-01 -4.11289632e-01 -9.33017850e-01
5.85588276e-01 -9.76081550e-01 -5.70490539e-01 5.49579151e-02
-7.04442620e-01 5.21834433e-01 5.84703922e-01 -1.15916955e+00
1.06285477e+00 -1.91263962e+00 2.47286633e-01 6.33848786e-01
3.92828643e-01 1.53908618e-02 1.08876646e-01 -1.09613806e-01
-1.96254909e-01 1.46247623e-02 -7.21811235e-01 -9.72191431e-03
-5.75295150e-01 2.25642040e-01 4.47505146e-01 5.01358032e-01
3.52760613e-01 6.66804135e-01 -6.98877454e-01 -9.22700346e-01
2.11913496e-01 4.48060483e-01 -7.08686233e-01 9.36743915e-02
-1.25419214e-01 9.38205957e-01 -3.11533540e-01 6.76727057e-01
6.56244218e-01 -4.28427868e-02 1.64091006e-01 -6.72571957e-01
-2.33499452e-01 -6.09739684e-03 -1.13591337e+00 1.82887375e+00
-4.33699548e-01 2.18322039e-01 4.79592353e-01 -1.20456409e+00
7.52272725e-01 4.17627722e-01 8.37718725e-01 -6.20464385e-01
2.38255963e-01 5.88292956e-01 4.15725827e-01 -7.31316686e-01
2.19006523e-01 -2.48605311e-01 2.60491937e-01 1.62696838e-01
1.24354526e-01 -8.89916122e-02 1.94170579e-01 1.01699159e-01
8.44564021e-01 2.77728736e-01 -2.83043414e-01 -4.26417977e-01
7.29526401e-01 2.34280840e-01 6.17115796e-01 3.02969813e-01
-1.33367032e-01 1.05788112e+00 6.01568341e-01 -3.18089008e-01
-1.28444004e+00 -1.06231439e+00 -1.53743729e-01 3.56683254e-01
-1.29433140e-01 -1.68987364e-02 -1.02571487e+00 -4.67106611e-01
-2.27201730e-01 -5.37516810e-02 -6.66441441e-01 3.35450843e-02
-9.28953946e-01 -8.85550678e-01 4.14807022e-01 5.11731207e-01
6.55513525e-01 -9.75771010e-01 -6.28937721e-01 2.88992167e-01
-3.47553253e-01 -9.91223097e-01 -3.33092123e-01 2.37739608e-01
-1.50287640e+00 -1.03943932e+00 -1.00918126e+00 -9.36194360e-01
9.74344790e-01 -2.81304549e-02 7.66533196e-01 3.18045378e-01
-6.07605219e-01 -3.32840867e-02 2.89237853e-02 2.12072566e-01
-4.61204052e-01 1.99573338e-01 -2.99252152e-01 1.29129544e-01
-3.12172771e-01 -6.15698397e-01 -9.76554573e-01 1.26037508e-01
-1.07212591e+00 6.09595478e-01 1.16926003e+00 1.15283573e+00
8.37775707e-01 4.83071767e-02 4.94979739e-01 -9.23231661e-01
3.45278054e-01 -4.41344768e-01 -2.49866337e-01 7.16893654e-03
-5.05727232e-01 9.46703926e-02 2.63597488e-01 -3.02801251e-01
-8.92939866e-01 2.50983834e-01 -3.05282325e-01 -2.98278302e-01
-8.13301802e-02 8.77502382e-01 1.47544116e-01 -2.61331260e-01
4.22719240e-01 2.13599727e-01 5.78120351e-01 -5.00276566e-01
3.92059684e-02 3.87587994e-01 6.24739587e-01 -5.62198460e-01
3.64965439e-01 5.92126727e-01 2.18257979e-01 -7.86789298e-01
-5.02924323e-01 -5.20730555e-01 -1.01228106e+00 -7.50047982e-01
1.10122859e+00 -6.22697651e-01 -1.85636923e-01 5.95781326e-01
-9.29082930e-01 -2.74223626e-01 -5.45111150e-02 6.52302206e-01
-4.87084299e-01 4.51980650e-01 -9.29350615e-01 -3.89655352e-01
-5.31386197e-01 -1.64334178e+00 1.03776634e+00 1.39219478e-01
6.86776191e-02 -9.24635231e-01 -1.42978519e-01 8.15985978e-01
3.69762838e-01 4.59808052e-01 1.15042293e+00 -4.01127517e-01
-5.70223391e-01 9.46526509e-03 -2.21261591e-01 7.47248650e-01
-2.17430070e-02 -2.86714453e-02 -6.28623307e-01 -1.06599070e-01
-3.96797881e-02 -1.04940660e-01 8.59480619e-01 7.12974787e-01
1.05215907e+00 -2.13714197e-01 -1.44474775e-01 5.64956665e-01
1.54641092e+00 -6.99008629e-02 7.12934971e-01 5.40115893e-01
9.61367846e-01 8.67610931e-01 3.96607429e-01 7.04089031e-02
1.60787597e-01 5.47746599e-01 5.89397907e-01 -6.36462986e-01
-4.78619665e-01 2.83639669e-01 3.63561809e-02 1.25091994e+00
-2.06984997e-01 5.18029094e-01 -1.09938514e+00 7.91789353e-01
-1.56347895e+00 -2.87029237e-01 -5.57205021e-01 2.14248514e+00
9.75107431e-01 3.98219109e-01 6.30462170e-02 2.96265960e-01
6.01484656e-01 -2.97240883e-01 -2.38215014e-01 -1.85091510e-01
7.08713233e-02 5.84276974e-01 6.45910263e-01 4.95797396e-01
-1.05792928e+00 6.93974316e-01 5.72614288e+00 8.99811506e-01
-1.31648254e+00 1.81145489e-01 6.34390175e-01 9.39184502e-02
-1.53186858e-01 -7.17543140e-02 -1.39168561e-01 3.93910289e-01
5.70326984e-01 5.49845278e-01 3.47560532e-02 3.68328810e-01
4.10109758e-01 -4.50801492e-01 -6.03700817e-01 5.35553634e-01
-9.95413736e-02 -1.29666984e+00 1.48180217e-01 7.68804774e-02
6.52344644e-01 -5.15987910e-03 7.80898705e-02 -1.00430794e-01
-2.25633979e-01 -9.23330665e-01 6.64457262e-01 6.76256895e-01
5.37626863e-01 -7.22271919e-01 1.04233658e+00 4.10032878e-03
-9.06945109e-01 1.03542671e-01 -2.44610962e-02 2.72090316e-01
1.45206362e-01 9.59756851e-01 -1.17904043e+00 7.85839438e-01
6.27378881e-01 3.18380773e-01 -6.69087470e-01 1.39227295e+00
8.34892839e-02 7.53245533e-01 -2.77034461e-01 4.80804950e-01
4.80843037e-01 -3.97739559e-01 6.71491086e-01 1.24801528e+00
1.58978030e-01 -1.92121089e-01 2.54242986e-01 9.07062590e-01
2.69932628e-01 2.80123860e-01 -2.70010412e-01 1.77437782e-01
-1.69127986e-01 1.49383271e+00 -1.29863775e+00 -3.15468192e-01
-2.41249979e-01 6.55743837e-01 -5.06398715e-02 3.01461548e-01
-4.66042012e-01 9.55120698e-02 1.41174018e-01 5.68246186e-01
5.66215962e-02 -3.95531952e-01 -8.38808537e-01 -8.02742839e-01
-2.24839542e-02 -8.42652380e-01 2.90879607e-01 -7.42873549e-01
-1.02941048e+00 5.46805084e-01 4.73188655e-03 -1.19654679e+00
4.70942305e-03 -4.85699087e-01 -3.38967413e-01 8.03216338e-01
-1.15362620e+00 -1.28763163e+00 -2.00632825e-01 3.48428607e-01
5.88766515e-01 1.21606901e-01 2.46793911e-01 5.80990732e-01
-4.63878602e-01 1.29938066e-01 -1.91879079e-01 2.33333409e-01
4.21665102e-01 -1.32410765e+00 -2.72287816e-01 7.39008725e-01
-3.95197928e-01 5.60966909e-01 5.32970130e-01 -1.13065910e+00
-1.02208066e+00 -1.02003074e+00 4.61561739e-01 3.48262697e-01
7.47116745e-01 -1.47593349e-01 -1.02410841e+00 2.18087122e-01
9.02181715e-02 -1.97314061e-02 5.21951437e-01 -4.47379351e-01
1.64403543e-01 -4.44805585e-02 -1.07727766e+00 6.12775445e-01
4.45813388e-01 -2.07391009e-01 -3.97903830e-01 2.66731888e-01
3.19994181e-01 -5.80867112e-01 -1.13829529e+00 6.58427417e-01
7.10502207e-01 -1.00736618e+00 1.08722758e+00 -1.99315175e-01
8.04224253e-01 -3.67204070e-01 2.33441249e-01 -9.97589707e-01
-8.62067342e-02 -7.77199417e-02 2.15378523e-01 8.30276251e-01
3.90424222e-01 -3.14236552e-01 9.62691903e-01 3.20124328e-01
-6.48882151e-01 -1.14679575e+00 -1.24016500e+00 -5.75900137e-01
1.49286151e-01 -2.54023314e-01 -6.61676005e-02 7.08089948e-01
-5.16194284e-01 -1.04952179e-01 -4.47114110e-02 8.27906355e-02
6.86175942e-01 -2.85710335e-01 2.74226576e-01 -1.00328672e+00
-1.08266339e-01 -4.67602104e-01 -4.67553914e-01 -6.75860763e-01
-2.33893648e-01 -1.19190645e+00 2.82694340e-01 -1.62068272e+00
1.02525063e-01 -6.71346068e-01 -3.58477831e-01 2.59939492e-01
6.11669347e-02 7.67248929e-01 -9.20547172e-03 4.37544167e-01
-3.06410640e-02 2.90758818e-01 1.78983986e+00 -2.92360395e-01
-1.09983675e-01 -1.40477106e-01 -1.25953302e-01 6.89626694e-01
7.37753332e-01 -4.72363055e-01 -2.91882783e-01 -3.10054004e-01
1.12032153e-01 2.23014399e-01 7.67208099e-01 -1.30141211e+00
2.32230932e-01 3.30473751e-01 5.12976408e-01 -7.42818236e-01
2.23970205e-01 -6.95147872e-01 2.33300433e-01 8.03327382e-01
-3.06215227e-01 7.40614757e-02 6.96415156e-02 2.56458670e-01
-3.78764093e-01 -2.54909664e-01 8.78849328e-01 -4.92503583e-01
-3.01724553e-01 1.12121917e-01 -5.66921771e-01 -1.79332823e-01
6.84110940e-01 -6.53969109e-01 2.40761355e-01 5.05075343e-02
-1.08484185e+00 5.70718162e-02 6.59752339e-02 6.54097646e-02
8.51993084e-01 -1.19774997e+00 -7.88123965e-01 1.92743868e-01
-3.01483780e-01 2.26278305e-01 5.44478595e-01 1.67881107e+00
-9.91349995e-01 1.44638032e-01 -5.78275084e-01 -1.01340556e+00
-9.88453686e-01 1.98798299e-01 6.49962604e-01 -6.18655682e-01
-8.96932662e-01 6.89917505e-01 1.89102232e-01 -2.91962355e-01
-4.52132933e-02 -5.37886798e-01 -2.74957478e-01 -1.16400264e-01
-7.31817260e-02 2.60629475e-01 4.13976908e-01 -8.44523847e-01
-2.31785625e-01 5.63403249e-01 -1.82250172e-01 -4.23917085e-01
1.20560396e+00 -1.52577430e-01 -4.53690141e-01 3.64095062e-01
1.14640725e+00 -3.61750841e-01 -1.15918076e+00 4.37968737e-03
1.91926509e-01 -2.46852681e-01 5.30454695e-01 -8.96191716e-01
-1.56742024e+00 9.76915836e-01 8.60344887e-01 -1.71934485e-01
1.07062054e+00 -3.07233810e-01 1.13682151e+00 -3.29743534e-01
2.28138968e-01 -1.30255175e+00 1.53388202e-01 1.63272619e-01
9.26657677e-01 -9.83194709e-01 -1.22639528e-02 -6.04784071e-01
-5.20454884e-01 1.40014422e+00 6.57190859e-01 -2.73299724e-01
5.42709112e-01 4.89182770e-01 2.41057113e-01 -5.12936234e-01
-3.49666268e-01 -1.59340262e-01 3.91334981e-01 4.73797858e-01
5.28040051e-01 -2.52139587e-02 -6.65434659e-01 4.37897414e-01
4.89594182e-03 8.03487077e-02 4.29599553e-01 9.99975085e-01
-3.11326802e-01 -9.07685935e-01 -5.83210349e-01 7.79502094e-01
-7.66846001e-01 -1.04713775e-01 -2.06903934e-01 8.11926365e-01
4.70033407e-01 4.34647650e-01 9.30385143e-02 -3.47887129e-01
-9.00141448e-02 -1.72776759e-01 4.87053096e-01 -5.87044597e-01
-1.09348822e+00 5.45535922e-01 1.64963100e-02 -4.85521823e-01
-3.99763674e-01 -5.49072683e-01 -1.51004100e+00 2.95292884e-01
-2.25798145e-01 -9.02670845e-02 8.48415136e-01 1.02089036e+00
5.09914532e-02 9.25473034e-01 4.17038947e-01 -9.08193529e-01
-1.61911622e-01 -9.44896698e-01 -4.87916291e-01 2.63904303e-01
2.41401926e-01 -8.33613455e-01 -1.62813872e-01 2.44659796e-01] | [14.383228302001953, -2.4150338172912598] |
2723689d-1670-4bee-947c-6d31f598d524 | pansharpening-by-convolutional-neural | 2111.08334 | null | https://arxiv.org/abs/2111.08334v3 | https://arxiv.org/pdf/2111.08334v3.pdf | Pansharpening by convolutional neural networks in the full resolution framework | In recent years, there has been a growing interest in deep learning-based pansharpening. Thus far, research has mainly focused on architectures. Nonetheless, model training is an equally important issue. A first problem is the absence of ground truths, unavoidable in pansharpening. This is often addressed by training networks in a reduced resolution domain and using the original data as ground truth, relying on an implicit scale invariance assumption. However, on full resolution images results are often disappointing, suggesting such invariance not to hold. A further problem is the scarcity of training data, which causes a limited generalization ability and a poor performance on off-training test images. In this paper, we propose a full-resolution training framework for deep learning-based pansharpening. The framework is fully general and can be used for any deep learning-based pansharpening model. Training takes place in the high-resolution domain, relying only on the original data, thus avoiding any loss of information. To ensure spectral and spatial fidelity, a suitable two-component loss is defined. The spectral component enforces consistency between the pansharpened output and the low-resolution multispectral input. The spatial component, computed at high-resolution, maximizes the local correlation between each pansharpened band and the panchromatic input. At testing time, the target-adaptive operating modality is adopted, achieving good generalization with a limited computational overhead. Experiments carried out on WorldView-3, WorldView-2, and GeoEye-1 images show that methods trained with the proposed framework guarantee a pretty good performance in terms of both full-resolution numerical indexes and visual quality. | ['Giuseppe Scarpa', 'Giovanni Poggi', 'Antonio Mazza', 'Sergio Vitale', 'Matteo Ciotola'] | 2021-11-16 | null | null | null | null | ['satellite-image-super-resolution', 'pansharpening'] | ['computer-vision', 'computer-vision'] | [ 4.70909238e-01 -3.47913504e-01 -2.23099977e-01 -9.71666202e-02
-7.97778130e-01 -3.11340094e-01 3.55190992e-01 -1.18511997e-01
-5.33040106e-01 8.30934703e-01 -3.70104849e-01 6.15484407e-03
-5.25256872e-01 -1.29898310e+00 -5.36901236e-01 -1.12298274e+00
1.10278971e-01 4.06778008e-02 1.23686954e-01 -4.18638617e-01
-2.83057690e-01 7.80907929e-01 -1.54063678e+00 -1.74145083e-04
1.18097568e+00 1.09356821e+00 3.93180817e-01 3.54242265e-01
2.61610121e-01 2.53956944e-01 -5.40022671e-01 -2.47029603e-01
5.73681891e-01 -3.36570799e-01 -4.01160330e-01 1.60952374e-01
5.73910534e-01 -3.17634434e-01 -2.50732332e-01 1.44919062e+00
3.54780793e-01 2.18300208e-01 3.13308775e-01 -6.78684711e-01
-6.03972495e-01 1.51331469e-01 -8.65475655e-01 1.12018168e-01
-2.16263399e-01 -6.68942630e-02 1.05296683e+00 -5.55833817e-01
2.38821253e-01 8.15782726e-01 7.13100314e-01 4.39690426e-02
-1.44897282e+00 -4.66649294e-01 -1.83589697e-01 1.92285001e-01
-1.40110171e+00 1.12010650e-01 9.53541875e-01 -2.66663253e-01
4.92063910e-01 2.82913178e-01 6.96589828e-01 7.74714530e-01
2.14532003e-01 2.00832337e-01 1.46648550e+00 -4.45996583e-01
2.55336538e-02 4.76122387e-02 -1.64892688e-01 2.84100860e-01
3.96393359e-01 5.30526698e-01 1.83183677e-03 1.29195273e-01
1.03663385e+00 -2.03209799e-02 -7.02440619e-01 -4.07635003e-01
-8.43575895e-01 8.73101056e-01 1.04990709e+00 6.82954431e-01
-5.92840791e-01 -3.29738379e-01 2.89370790e-02 3.73861909e-01
5.48292875e-01 4.94431496e-01 -2.94099778e-01 3.55275482e-01
-1.26754808e+00 1.71476781e-01 1.79505810e-01 2.52862215e-01
9.91665006e-01 2.10910246e-01 4.02428359e-01 1.18624890e+00
-6.11941591e-02 6.73568606e-01 4.17517573e-01 -6.97196066e-01
2.35404179e-01 4.37396735e-01 4.35493328e-02 -1.26350629e+00
-4.87161547e-01 -7.82533705e-01 -1.24246836e+00 7.48680830e-01
2.08613217e-01 -1.19346641e-01 -8.83058786e-01 1.79966331e+00
2.40077063e-01 -6.47472739e-02 1.47936329e-01 1.36776221e+00
4.05936033e-01 9.28419292e-01 -6.72777221e-02 -4.51284409e-01
1.09648144e+00 -7.48903692e-01 -6.17339671e-01 -2.27821440e-01
6.60579056e-02 -7.79645085e-01 1.07340944e+00 5.54795325e-01
-7.29500711e-01 -8.57521236e-01 -1.27087402e+00 1.91799060e-01
-5.71597397e-01 3.03208560e-01 5.08187830e-01 6.28900290e-01
-8.57232928e-01 7.78084338e-01 -6.01663053e-01 -2.96141654e-01
6.09771758e-02 4.68293577e-02 -5.11856914e-01 -6.64181560e-02
-1.40931654e+00 9.41798210e-01 7.63531804e-01 4.15762842e-01
-2.65202671e-01 -6.09573126e-01 -6.07949793e-01 2.26468563e-01
2.77740210e-01 -2.02962965e-01 7.20744312e-01 -1.40424216e+00
-1.35080886e+00 6.72850966e-01 5.62730014e-01 -4.70613718e-01
6.36433542e-01 -1.60748482e-01 -9.64223027e-01 2.76586473e-01
-3.81479673e-02 3.17563951e-01 1.09763789e+00 -1.43227148e+00
-6.31218374e-01 -2.71316290e-01 1.35051645e-02 2.74530768e-01
-4.21031296e-01 -3.14148009e-01 -2.09011510e-01 -8.27498794e-01
3.21783245e-01 -6.89999580e-01 -4.12126444e-02 -4.84978445e-02
-3.01851928e-02 3.44811141e-01 7.40918159e-01 -8.69819045e-01
8.57995212e-01 -2.23265743e+00 -4.70029190e-02 2.71368057e-01
-2.05780163e-01 5.40671766e-01 -3.15574050e-01 4.20930535e-02
-4.31726724e-01 -2.46777937e-01 -6.64959729e-01 2.29938358e-01
-3.60604763e-01 1.53427362e-01 -3.48214656e-01 6.29554033e-01
2.67065585e-01 4.04531598e-01 -5.94719291e-01 -3.65319908e-01
4.71388251e-01 7.97847450e-01 -2.68469125e-01 1.27256274e-01
-1.75154760e-01 3.83384854e-01 -1.96410656e-01 4.86517936e-01
1.22533131e+00 3.30030662e-03 -2.36738343e-02 -4.68164504e-01
-4.21411783e-01 -3.08512598e-01 -1.27378118e+00 1.40103638e+00
-7.13518143e-01 6.49285614e-01 3.84195596e-01 -1.04746234e+00
1.11841273e+00 2.74902403e-01 5.32465875e-01 -1.16637051e+00
-1.13696881e-01 3.56096715e-01 -3.22156176e-02 -1.94198340e-01
5.66792607e-01 -4.12258416e-01 3.05760682e-01 1.51425928e-01
-1.15913659e-01 -4.21289504e-01 5.12889102e-02 -4.64986116e-01
2.73022145e-01 1.81249544e-01 3.74692708e-01 -1.85349613e-01
7.28669763e-01 1.24817461e-01 4.63912278e-01 5.23122966e-01
1.34985000e-01 6.98104262e-01 1.54993013e-01 -3.67927670e-01
-1.14923704e+00 -9.85149026e-01 -5.59860110e-01 8.64398897e-01
1.88085467e-01 1.97843432e-01 -6.79224968e-01 -2.81147450e-01
-2.66092986e-01 6.42547071e-01 -5.64035594e-01 -2.65972875e-02
-5.50540686e-01 -1.10466874e+00 2.13099569e-01 1.91491053e-01
1.05975306e+00 -9.92443144e-01 -7.81279981e-01 1.99736521e-01
-2.30498090e-01 -1.04133987e+00 1.33336514e-01 4.12143767e-02
-1.03334713e+00 -9.32780862e-01 -9.95942175e-01 -4.51644391e-01
3.03302467e-01 4.16321695e-01 7.27597773e-01 -2.52934605e-01
-1.41480207e-01 -1.77292340e-02 -2.29928568e-01 4.92149182e-02
-4.04712558e-01 4.31969054e-02 -2.45853394e-01 2.16599643e-01
1.37934443e-02 -6.95226610e-01 -4.25743759e-01 3.69433880e-01
-1.15745366e+00 2.17192233e-01 9.73513067e-01 9.99538004e-01
7.75917530e-01 6.45180464e-01 4.57711726e-01 -5.08509636e-01
2.79722303e-01 -7.65206525e-03 -1.07025564e+00 1.76787749e-01
-7.29719520e-01 -2.99605727e-01 8.66837859e-01 -3.39056492e-01
-1.35902441e+00 -9.96067896e-02 -3.21341127e-01 -3.38468701e-01
-3.34060252e-01 7.09382296e-01 -1.72896355e-01 -3.23917657e-01
9.37001824e-01 2.93238401e-01 -1.97604284e-01 -6.04813278e-01
2.25856632e-01 5.27463317e-01 7.77714074e-01 -2.54528612e-01
1.13421452e+00 5.99410832e-01 1.19989909e-01 -1.34792972e+00
-8.16215158e-01 -3.14033359e-01 -5.42126417e-01 -2.63265073e-01
7.66628742e-01 -9.17323589e-01 -1.01456299e-01 5.78394592e-01
-7.36242056e-01 -2.93857783e-01 -3.57474506e-01 5.61922669e-01
-4.14453864e-01 5.23778498e-01 -2.73631603e-01 -6.41588211e-01
-3.10891896e-01 -9.01523292e-01 6.95279598e-01 4.46149528e-01
3.99540037e-01 -8.45520318e-01 2.08336681e-01 2.43101582e-01
5.89645505e-01 3.59784991e-01 1.00972974e+00 -1.69421956e-01
-4.40142393e-01 -3.67771864e-01 -6.49573147e-01 6.76236987e-01
2.92270929e-01 -1.52265921e-01 -1.09225261e+00 -4.71718848e-01
3.91388297e-01 -3.09643090e-01 9.81321633e-01 5.20083904e-01
1.10633838e+00 -1.57686934e-01 1.29044011e-01 9.54841912e-01
1.83001399e+00 -3.96830961e-02 6.50528610e-01 5.75666606e-01
4.81123805e-01 4.85813826e-01 7.82471597e-01 8.95700157e-02
-2.99449176e-01 7.40706563e-01 7.50024378e-01 -6.65563107e-01
-1.28944784e-01 -1.00181423e-01 1.33655639e-02 1.52482390e-01
-5.08543372e-01 -7.35380575e-02 -6.49795473e-01 4.68045264e-01
-1.53556025e+00 -1.06001949e+00 -1.95007697e-02 2.60773396e+00
7.34483302e-01 8.95640478e-02 -3.05856969e-02 4.62580591e-01
7.84597397e-01 6.68339431e-01 -3.95500928e-01 -6.50934801e-02
-5.45627356e-01 3.95949692e-01 6.41340911e-01 5.91280103e-01
-1.44651306e+00 6.57257915e-01 5.05879307e+00 8.04940164e-01
-1.73859239e+00 6.55491725e-02 4.07912940e-01 2.39936963e-01
-1.13157239e-02 -2.36826897e-01 -2.04809904e-01 2.26649418e-01
6.46653593e-01 -1.09375902e-01 4.90921974e-01 8.74250829e-01
2.42558599e-01 -1.95930138e-01 -4.60730076e-01 9.91154790e-01
-1.23156726e-01 -1.05253124e+00 -6.89356402e-02 4.98172045e-02
8.04272413e-01 1.20109826e-01 1.93027750e-01 5.63527495e-02
-1.74343824e-01 -9.44461226e-01 4.16526228e-01 3.30173194e-01
1.02060354e+00 -8.77288163e-01 7.98245132e-01 3.11372340e-01
-1.12789094e+00 -1.40518770e-01 -6.13932490e-01 2.88012177e-01
1.34066632e-02 7.82385290e-01 -2.76643336e-01 1.04592943e+00
7.06045806e-01 5.26102901e-01 -2.90891916e-01 1.09648275e+00
-3.90940785e-01 4.21025634e-01 -3.87451589e-01 5.22934437e-01
3.04945737e-01 -7.07401395e-01 5.05039930e-01 1.31349182e+00
4.59523529e-01 2.10086495e-01 2.57715881e-02 8.51041734e-01
1.43971622e-01 1.57289132e-01 -4.16353106e-01 1.56581268e-01
-1.11311786e-01 1.47369218e+00 -5.36234617e-01 -1.05545089e-01
-4.33587730e-01 8.11290622e-01 -1.22936651e-01 6.76066995e-01
-8.32988024e-01 -1.88999861e-01 3.78134489e-01 7.10288361e-02
3.97372484e-01 -9.89756435e-02 -2.62102783e-01 -1.11069381e+00
-8.82866755e-02 -1.06055439e+00 4.46595132e-01 -7.09389687e-01
-1.04628348e+00 7.98637211e-01 4.17179130e-02 -1.35924196e+00
-5.77720776e-02 -6.32598042e-01 -3.78278613e-01 1.23951566e+00
-1.89814401e+00 -1.24331796e+00 -7.64331758e-01 5.55642784e-01
2.56531656e-01 9.35620889e-02 8.12567949e-01 4.24998790e-01
-4.25356120e-01 2.65938461e-01 2.61804670e-01 -2.73841750e-02
5.04248857e-01 -1.02732396e+00 -7.10398108e-02 1.19297862e+00
1.02918833e-01 2.90053245e-02 9.56066549e-01 -3.47872227e-01
-8.13506424e-01 -1.08886755e+00 4.81243670e-01 5.27864039e-01
5.23971438e-01 1.12178952e-01 -1.36757839e+00 2.97880381e-01
1.64203383e-02 4.61983196e-02 2.26421565e-01 -1.76054969e-01
-2.95178026e-01 -5.61477482e-01 -1.16063905e+00 4.43676293e-01
4.47197437e-01 -5.78459859e-01 -5.21848857e-01 2.32978851e-01
2.08764657e-01 -3.06408435e-01 -8.01168323e-01 7.71999836e-01
4.91403490e-01 -1.35629034e+00 9.78555262e-01 -2.24338770e-02
4.24679905e-01 -3.62051666e-01 -1.94876835e-01 -1.43614638e+00
-5.65168738e-01 -8.94581825e-02 2.45252967e-01 8.51866126e-01
2.93974489e-01 -8.02972198e-01 6.40895545e-01 -1.55017413e-02
7.40215033e-02 -3.74059707e-01 -8.75305355e-01 -8.57502341e-01
1.44100323e-01 -2.58406520e-01 4.15584207e-01 1.15767252e+00
-6.08133256e-01 1.33964330e-01 -4.96410906e-01 7.33724296e-01
7.64690399e-01 6.40045464e-01 5.96045196e-01 -1.36503375e+00
-4.93441880e-01 -5.71873605e-01 -1.54907331e-01 -7.90088773e-01
-3.76077630e-02 -3.91953021e-01 -3.75067629e-02 -1.46696913e+00
-1.23089075e-01 -4.29778069e-01 -4.04245287e-01 4.05061513e-01
-1.14684552e-02 4.90388274e-01 6.81680441e-02 2.31793165e-01
3.91517937e-01 5.68513513e-01 1.28656471e+00 -1.47729829e-01
-2.42460832e-01 1.69502363e-01 -1.70866534e-01 7.48497605e-01
9.64881539e-01 -2.39590064e-01 -3.10009778e-01 -3.99379671e-01
9.10425112e-02 1.93068102e-01 5.33957124e-01 -1.31644213e+00
-1.18195079e-02 -3.01613122e-01 4.44978386e-01 -5.02571642e-01
5.96207082e-01 -1.13933873e+00 4.60811377e-01 3.74368012e-01
-7.09634274e-02 -4.11016196e-01 2.76605308e-01 2.96917886e-01
-5.23880959e-01 -1.67674467e-01 1.51908076e+00 -9.39002261e-03
-8.09694886e-01 3.45654458e-01 1.48736209e-01 -4.65083390e-01
6.92685485e-01 -2.88785368e-01 -2.05764636e-01 -2.63795137e-01
-5.72659791e-01 -3.18275899e-01 6.89972758e-01 1.61330074e-01
4.54665363e-01 -1.10109854e+00 -7.79611945e-01 4.68308568e-01
9.58278924e-02 -1.44793332e-01 5.75476289e-01 7.52184212e-01
-6.32484138e-01 1.90265402e-01 -5.32526970e-01 -5.07587254e-01
-9.82506216e-01 6.05680227e-01 8.53247166e-01 -2.36929923e-01
-7.33320951e-01 3.23979527e-01 2.87111580e-01 -1.96900636e-01
-1.80190191e-01 -1.87708110e-01 -3.41372997e-01 1.01348855e-01
5.21211445e-01 2.33157679e-01 2.86541730e-01 -8.79824102e-01
-5.58811463e-02 8.45502734e-01 2.49898583e-01 -8.06292742e-02
1.39763105e+00 3.84735800e-02 -1.20647445e-01 2.67980963e-01
9.70947266e-01 1.10222086e-01 -1.31878316e+00 -3.95117670e-01
-2.69296348e-01 -7.26789474e-01 5.36794126e-01 -8.37752879e-01
-1.28201020e+00 1.01885521e+00 1.01518357e+00 5.04361331e-01
1.63803720e+00 -4.60405827e-01 4.31711793e-01 2.47746900e-01
2.00814530e-01 -1.16286922e+00 -2.31664538e-01 2.23878399e-01
1.00180626e+00 -1.30811965e+00 1.45549193e-01 -2.69423008e-01
-2.87990570e-01 1.23537350e+00 5.11171520e-01 -1.29842818e-01
4.23237354e-01 -1.20686673e-01 1.93516329e-01 4.61780801e-02
-3.93177532e-02 -1.92967594e-01 3.00976992e-01 5.16597092e-01
3.13711435e-01 1.47178307e-01 -2.27083847e-01 1.52660891e-01
-9.61866155e-02 -1.14709951e-01 3.24402422e-01 3.09916854e-01
-6.20591879e-01 -7.75107205e-01 -8.03386509e-01 1.72027335e-01
-3.80810767e-01 4.81923260e-02 -7.09880702e-03 1.17965972e+00
2.52314150e-01 7.37124622e-01 4.85833995e-02 8.92481208e-02
4.98650759e-01 -3.21089089e-01 3.42260182e-01 -1.70120955e-01
-9.48565453e-02 2.90254354e-01 -1.49994865e-01 -4.11703736e-01
-6.42690480e-01 -2.22482368e-01 -7.79585302e-01 -3.74516279e-01
-4.85117674e-01 2.36746892e-01 5.97255826e-01 6.75711632e-01
-2.58318305e-01 3.06578249e-01 7.83324301e-01 -9.16532338e-01
-5.91634095e-01 -8.96221876e-01 -9.20345008e-01 2.29443192e-01
6.33432686e-01 -5.61655521e-01 -4.30977434e-01 -2.15600118e-01] | [10.109939575195312, -1.9317055940628052] |
1a217eaf-bb69-4295-bbc7-8242c71be4f6 | gender-and-interest-targeting-for-sponsored | 1606.07189 | null | http://arxiv.org/abs/1606.07189v1 | http://arxiv.org/pdf/1606.07189v1.pdf | Gender and Interest Targeting for Sponsored Post Advertising at Tumblr | As one of the leading platforms for creative content, Tumblr offers
advertisers a unique way of creating brand identity. Advertisers can tell their
story through images, animation, text, music, video, and more, and promote that
content by sponsoring it to appear as an advertisement in the streams of Tumblr
users. In this paper we present a framework that enabled one of the key
targeted advertising components for Tumblr, specifically gender and interest
targeting. We describe the main challenges involved in development of the
framework, which include creating the ground truth for training gender
prediction models, as well as mapping Tumblr content to an interest taxonomy.
For purposes of inferring user interests we propose a novel semi-supervised
neural language model for categorization of Tumblr content (i.e., post tags and
post keywords). The model was trained on a large-scale data set consisting of
6.8 billion user posts, with very limited amount of categorized keywords, and
was shown to have superior performance over the bag-of-words model. We
successfully deployed gender and interest targeting capability in Yahoo
production systems, delivering inference for users that cover more than 90% of
daily activities at Tumblr. Online performance results indicate advantages of
the proposed approach, where we observed 20% lift in user engagement with
sponsored posts as compared to untargeted campaigns. | ['Nemanja Djuric', 'Mihajlo Grbovic', 'Vladan Radosavljevic', 'Narayan Bhamidipati', 'Ananth Nagarajan'] | 2016-06-23 | null | null | null | null | ['gender-prediction'] | ['computer-vision'] | [-1.76840723e-02 2.63244927e-01 -1.12167990e+00 -7.28440046e-01
-1.20517266e+00 -6.57380462e-01 8.37058783e-01 1.03491418e-01
-2.44938582e-01 4.36860144e-01 4.65549767e-01 -8.53660479e-02
6.25934526e-02 -1.05546725e+00 -8.41837049e-01 -1.84637174e-01
-2.84624659e-02 1.00425422e+00 7.02849329e-02 -4.23854053e-01
2.41039351e-01 1.39246270e-01 -1.68955255e+00 5.47005177e-01
1.00052409e-01 1.59526038e+00 7.59236887e-02 1.67241216e-01
-3.95105809e-01 8.47965181e-01 -5.38883269e-01 -9.31868315e-01
1.26796648e-01 1.10046774e-01 -4.30237204e-01 8.08255374e-02
5.12718439e-01 -4.11984921e-01 -5.30507624e-01 6.53235495e-01
4.35753047e-01 2.03661732e-02 2.95686066e-01 -1.20395553e+00
-6.44035995e-01 1.42056465e+00 -7.32074320e-01 7.06342161e-02
2.95765817e-01 -6.07573271e-01 1.39495516e+00 -6.33285522e-01
8.10245931e-01 1.39041126e+00 4.33788896e-01 3.51034820e-01
-1.10664999e+00 -1.21991634e+00 2.70549417e-01 -2.32238665e-01
-1.29383218e+00 -4.48887408e-01 7.62732148e-01 -4.82018143e-01
1.76079378e-01 5.84948778e-01 6.95290387e-01 1.63410509e+00
5.58232591e-02 8.85179996e-01 1.13368094e+00 3.61881293e-02
1.98773891e-02 7.89538324e-01 1.85078487e-01 4.98851091e-01
-2.82052793e-02 -1.88210666e-01 -1.10233891e+00 -4.37405288e-01
5.85701466e-01 -1.58226892e-01 5.74115634e-01 1.01103567e-01
-8.12935472e-01 1.48601139e+00 -1.85688287e-02 -3.86603549e-02
-1.80414453e-01 3.46909076e-01 3.25878859e-01 -1.19138934e-01
1.16953623e+00 2.51177132e-01 -2.13805005e-01 -1.98788837e-01
-1.01389682e+00 5.84292531e-01 7.21467555e-01 1.01184320e+00
5.40827155e-01 -2.06824273e-01 -4.32385385e-01 1.16538370e+00
4.61238861e-01 5.80380201e-01 6.17333889e-01 -9.36653495e-01
2.82002181e-01 4.03046578e-01 1.83886111e-01 -1.42382467e+00
6.46008477e-02 -8.74645114e-01 -1.95361212e-01 -6.25783682e-01
1.87474757e-01 1.27621621e-01 -6.90332413e-01 1.63997889e+00
1.48121864e-01 -1.37193963e-01 -3.67319435e-01 7.29990780e-01
1.07767177e+00 6.47300005e-01 2.23635837e-01 -8.23894963e-02
1.73674691e+00 -6.46999121e-01 -8.36915076e-01 -4.72381473e-01
1.40282989e-01 -9.49003696e-01 1.08105469e+00 4.58042294e-01
-1.19727862e+00 -4.86464500e-01 -6.61240101e-01 -2.63167154e-02
-5.49158514e-01 3.35618347e-01 1.25472236e+00 8.58427227e-01
-5.67863703e-01 2.12499857e-01 -2.57271111e-01 -2.39774138e-01
5.22738397e-01 2.51798481e-01 1.38544500e-01 2.10378811e-01
-1.25267351e+00 6.39212728e-02 2.76201870e-02 -5.02649605e-01
-9.46401596e-01 -8.06047797e-01 -6.22228920e-01 -1.81432918e-01
4.65030640e-01 3.32326069e-02 1.54255176e+00 -7.47568846e-01
-1.04159784e+00 1.16551852e+00 -9.95771959e-02 -6.41430318e-01
3.78207207e-01 -1.22715332e-01 -7.74097264e-01 -1.22321114e-01
3.81158710e-01 7.36735642e-01 8.18437040e-01 -1.09065211e+00
-1.09690273e+00 -5.74642777e-01 1.02454014e-01 -1.68605298e-01
-7.45514274e-01 3.45663965e-01 -6.89485490e-01 -7.58951783e-01
3.25586587e-01 -8.12918007e-01 4.07061353e-02 -6.21976793e-01
-2.83396006e-01 -3.42860162e-01 7.00784743e-01 -8.84301066e-01
1.23662353e+00 -2.18877387e+00 -4.99878258e-01 3.51577342e-01
2.41073608e-01 -5.62784970e-01 3.54978472e-01 4.23639089e-01
2.27708772e-01 1.12971507e-01 7.88700700e-01 -3.36991131e-01
2.47225031e-01 -2.52882410e-02 -7.98877418e-01 1.06482074e-01
-4.30235922e-01 8.10338020e-01 -6.60079956e-01 -4.39929545e-01
-4.65024501e-01 2.54365414e-01 -5.28287530e-01 -1.11835897e-01
-6.81934953e-01 3.18173096e-02 -6.05917633e-01 1.07877886e+00
4.73068953e-01 -2.29305223e-01 2.64027834e-01 -3.54274392e-01
-1.09616101e-01 5.61277688e-01 -8.14100325e-01 1.17513609e+00
-5.87581933e-01 6.24878645e-01 1.06099732e-01 -6.98015571e-01
1.33256853e+00 -6.58724178e-03 7.66936302e-01 -9.20386195e-01
3.69897097e-01 5.61453030e-02 -5.54069519e-01 -2.44559377e-01
9.35735524e-01 8.06826632e-03 -8.16736281e-01 5.98305345e-01
-1.12176538e-01 5.54757774e-01 1.39849082e-01 4.30110484e-01
5.71390510e-01 -1.56732336e-01 -2.68076271e-01 3.00367810e-02
1.70530528e-02 2.32231244e-01 2.53726870e-01 7.47251809e-01
1.64779693e-01 1.49734303e-01 4.08527166e-01 -2.66774386e-01
-6.85969830e-01 -1.09289813e+00 -2.47751608e-01 1.91321361e+00
5.40506057e-02 -4.56195354e-01 -3.59059125e-01 -5.75179935e-01
2.15999275e-01 9.72525537e-01 -6.69321954e-01 -4.05707955e-03
-4.54718739e-01 -7.32701004e-01 4.04566586e-01 2.04923123e-01
5.85527480e-01 -5.64914405e-01 1.05840236e-01 1.80548057e-01
-5.81903636e-01 -1.32908928e+00 -6.51086092e-01 -1.70429990e-01
-5.25733531e-01 -5.62563896e-01 -4.55478042e-01 -8.10491264e-01
2.30586305e-01 1.71747878e-01 1.23916364e+00 -6.46601796e-01
-1.48292795e-01 4.73634541e-01 -2.96049267e-01 -4.21060055e-01
-4.11551803e-01 3.34440142e-01 -6.67786449e-02 4.33234423e-01
6.69867814e-01 -5.64838350e-01 -4.76510078e-01 7.53737628e-01
-4.53754812e-01 -1.28664225e-01 6.43546999e-01 3.22174668e-01
4.58920926e-01 1.75381795e-01 8.04560602e-01 -1.22538149e+00
7.11167753e-01 -7.35644639e-01 -4.17435080e-01 -1.88598350e-01
-6.79008484e-01 -5.68593740e-01 -2.41507679e-01 -8.74045670e-01
-8.71763349e-01 -4.55879606e-02 -2.65329480e-01 1.13904156e-01
3.56155723e-01 5.49165905e-01 2.64566317e-02 1.35049552e-01
5.10456681e-01 2.84127295e-01 -2.29502935e-02 -8.27708840e-01
4.30180967e-01 1.44895041e+00 3.45929533e-01 -6.76702976e-01
5.45624495e-01 5.58774590e-01 -5.96154273e-01 -6.70695662e-01
-1.26119018e+00 -8.05464745e-01 1.99246436e-01 -4.88399118e-01
5.69419384e-01 -1.29636717e+00 -1.06113815e+00 4.01866995e-02
-5.04269779e-01 2.05245793e-01 -1.21007830e-01 3.41866285e-01
-3.57062191e-01 -1.36461943e-01 -7.60096550e-01 -1.01199186e+00
-2.79526651e-01 -8.32631528e-01 1.07003176e+00 1.59928165e-02
-2.77803183e-01 -6.26796782e-01 -3.58874083e-01 1.06005967e+00
4.29822445e-01 1.66415721e-02 7.84631610e-01 -1.22518909e+00
-4.69844759e-01 -7.47898519e-01 -3.43932718e-01 1.55075835e-02
-1.20526291e-01 -3.58688742e-01 -9.23911154e-01 -8.08916688e-02
-2.12424934e-01 -4.04024988e-01 7.03135073e-01 4.22363788e-01
1.55559576e+00 -6.38034940e-01 -5.10859609e-01 6.59741983e-02
1.00520802e+00 2.67683268e-01 5.18734336e-01 3.17533612e-01
3.32827359e-01 7.27239668e-01 1.03689039e+00 7.25360990e-01
5.88209450e-01 1.14399552e+00 4.08897251e-01 -9.19305813e-03
7.87890926e-02 -8.21648240e-01 3.81314814e-01 3.96170616e-01
3.73010226e-02 3.06051914e-02 -2.78056413e-01 3.31379563e-01
-1.70079994e+00 -7.76517868e-01 2.62093581e-02 2.11640859e+00
9.86147821e-01 4.75796074e-01 7.71934807e-01 -3.63152713e-01
6.86342537e-01 1.85818717e-01 -3.15084994e-01 -2.46013179e-01
2.77146921e-02 3.27349640e-02 9.81119037e-01 1.82224199e-01
-1.13346434e+00 9.75852370e-01 6.55612040e+00 1.26351810e+00
-1.01190186e+00 5.04819214e-01 1.09196663e+00 -3.03533703e-01
-3.06614876e-01 -3.76878947e-01 -1.55961621e+00 5.98071039e-01
1.08330083e+00 -1.81403503e-01 4.95203644e-01 1.32503200e+00
3.45752448e-01 1.52385816e-01 -6.67559266e-01 9.95478988e-01
2.49335498e-01 -1.76643729e+00 4.51093493e-03 5.90724051e-01
5.06579936e-01 -3.37024808e-01 4.83314365e-01 7.31608689e-01
3.81933987e-01 -6.78287864e-01 1.11556387e+00 9.96533632e-02
8.12966168e-01 -8.27620447e-01 4.32916582e-01 1.11898303e-01
-8.05135369e-01 -3.83926481e-01 -2.08834976e-01 1.68694898e-01
2.52520263e-01 7.19953179e-01 -8.46906602e-01 3.81760160e-03
7.90957034e-01 6.32028878e-01 -5.40522337e-01 4.20795798e-01
3.90964895e-01 8.52818727e-01 -1.15916654e-01 -2.16280639e-01
8.63142088e-02 -2.45832764e-02 2.66391337e-01 1.00639629e+00
4.79886293e-01 -3.18454027e-01 2.88460314e-01 6.61739707e-01
-4.55059379e-01 4.51132625e-01 -1.95089936e-01 -7.75074601e-01
5.51147699e-01 1.60645282e+00 -7.90637970e-01 -2.93383688e-01
-1.94403514e-01 4.45275813e-01 -1.74959376e-01 8.11744034e-02
-1.20379257e+00 1.85632691e-01 5.36241889e-01 1.30454051e+00
1.91751987e-01 -1.21284693e-01 -1.20174535e-01 -6.10570252e-01
-3.89074415e-01 -9.58771527e-01 4.36910599e-01 -6.68839693e-01
-1.15173423e+00 3.09551865e-01 2.32471451e-01 -7.32254148e-01
-8.29609111e-02 -5.59833765e-01 -3.08854468e-02 5.25386095e-01
-8.72671545e-01 -1.69718242e+00 -1.30461380e-01 1.98717311e-01
7.76520073e-01 -6.94637775e-01 5.82275689e-01 7.76654422e-01
-1.79714739e-01 8.14068377e-01 -2.85164952e-01 -3.33312005e-02
7.45656550e-01 -8.21419001e-01 2.56375223e-01 -3.77332941e-02
2.43167281e-01 6.77990258e-01 9.14150715e-01 -9.80502725e-01
-1.36015248e+00 -1.09589016e+00 1.08036602e+00 -3.04684341e-01
1.02282941e+00 -1.11059630e+00 7.10374415e-02 9.38758194e-01
-1.88369200e-01 -6.32886887e-01 1.02921963e+00 4.92050439e-01
-3.61160338e-01 -3.55033457e-01 -1.16117918e+00 4.80865419e-01
9.88250971e-01 -3.34682107e-01 -3.80129516e-02 7.26060927e-01
1.01761234e+00 -3.14535409e-01 -8.61095428e-01 -6.13707751e-02
8.88509452e-01 -4.74955618e-01 1.28014255e+00 -5.37759840e-01
7.38360286e-01 3.27300072e-01 -6.99273884e-01 -5.82549930e-01
-2.02605486e-01 -4.91888881e-01 -3.28189760e-01 1.78241456e+00
5.59110820e-01 -4.27716494e-01 1.12047350e+00 4.43436682e-01
1.24562159e-01 -7.01521516e-01 -5.59706032e-01 -4.18632150e-01
-5.14877021e-01 -6.43173039e-01 5.56608081e-01 6.80061579e-01
-3.13763738e-01 4.93579328e-01 -7.39425480e-01 -1.20234318e-01
6.53653860e-01 1.79985434e-01 7.51114309e-01 -1.34314680e+00
-6.57961547e-01 -1.19448461e-01 -2.14703679e-01 -1.21475458e+00
-9.26164538e-02 -8.32000375e-01 -3.85234773e-01 -1.21558988e+00
4.79235500e-01 -8.03900421e-01 1.40248090e-01 4.08209532e-01
7.43886054e-01 8.83986354e-01 -3.31470013e-01 2.46392816e-01
-4.03692424e-01 8.10646638e-02 9.74736214e-01 -3.81505936e-01
-1.15220495e-01 7.55405486e-01 -1.44562614e+00 4.67323720e-01
6.96226060e-01 -4.60612863e-01 -3.03275257e-01 2.96269566e-01
6.70868039e-01 -1.73851162e-01 2.19648212e-01 -3.80846173e-01
-1.99203491e-01 -1.64542794e-01 4.01584864e-01 -9.36123490e-01
8.24169993e-01 -6.00140631e-01 1.97929710e-01 4.16475609e-02
-8.04810822e-01 -3.32960993e-01 -4.51781861e-02 6.46687150e-01
7.84860551e-03 -5.45285121e-02 2.78825432e-01 -1.05567880e-01
-5.34158766e-01 1.92714080e-01 -3.62885594e-01 5.92486653e-03
8.23135018e-01 -6.73123151e-02 -2.87759334e-01 -8.81079853e-01
-9.86875355e-01 -2.44874269e-01 -2.07860574e-01 8.47610176e-01
1.62388057e-01 -1.45129573e+00 -5.02605319e-01 1.30404383e-01
2.18743831e-01 -8.74559462e-01 1.15342401e-02 4.58863527e-01
1.46355867e-01 4.94870037e-01 8.55746791e-02 -3.99251252e-01
-1.35183334e+00 3.23351651e-01 -2.14959636e-01 -1.45523399e-01
-2.27473274e-01 9.01903868e-01 1.15611583e-01 -3.12531680e-01
6.10998809e-01 9.09114480e-02 -4.54634577e-01 6.97182417e-01
5.66046536e-01 2.11830124e-01 -5.56354690e-03 -8.24165940e-01
-9.89395455e-02 1.22382879e-01 -3.57246935e-01 -3.62399399e-01
1.30096197e+00 -3.70864689e-01 8.79498273e-02 4.86876547e-01
1.07061028e+00 4.16820943e-01 -7.90084422e-01 -2.21194014e-01
-2.44842499e-01 -7.02991068e-01 5.32292485e-01 -1.08045256e+00
-1.21046233e+00 2.84321338e-01 7.29875505e-01 4.83849645e-01
6.37674391e-01 6.33669734e-01 1.14120865e+00 -1.44674554e-01
5.90147257e-01 -1.30489373e+00 3.46807361e-01 -1.03676967e-01
6.28930569e-01 -1.15421414e+00 -9.53809023e-02 -7.81450391e-01
-6.11252606e-01 6.16254628e-01 2.16619030e-01 3.61124992e-01
7.59916306e-01 5.60193770e-02 -8.74944162e-05 -4.60615009e-01
-5.20924091e-01 -7.96503667e-03 3.30493599e-01 3.79972577e-01
5.87399065e-01 1.68243006e-01 -5.56109965e-01 1.34675980e+00
-5.93833208e-01 -1.34706110e-01 1.92531466e-01 5.05390465e-01
-7.05584288e-01 -1.09145105e+00 -2.73562878e-01 9.00741100e-01
-1.04537988e+00 7.06820711e-02 -2.92200595e-01 6.64453030e-01
6.96445107e-02 9.23573136e-01 2.58636504e-01 -7.75021434e-01
2.49495015e-01 -1.73585042e-01 1.19807824e-01 -5.13974547e-01
-5.70267498e-01 5.46834886e-01 7.18078554e-01 -4.08580929e-01
-3.73209119e-02 -7.16588140e-01 -6.78314686e-01 -4.11716491e-01
1.66956391e-02 3.26503038e-01 1.17284715e+00 5.16112864e-01
2.92793602e-01 4.70662832e-01 8.88139307e-01 -6.29023433e-01
-5.40867925e-01 -8.59702468e-01 -9.66774881e-01 3.39320511e-01
-5.68034708e-01 -8.44702721e-01 -2.36852720e-01 1.28274128e-01] | [9.958996772766113, 5.861457824707031] |
25594811-40e5-462e-9b9f-ab9eb3173c48 | simultaneous-identification-of-models-and | 2305.15174 | null | https://arxiv.org/abs/2305.15174v1 | https://arxiv.org/pdf/2305.15174v1.pdf | Simultaneous identification of models and parameters of scientific simulators | Many scientific models are composed of multiple discrete components, and scien tists often make heuristic decisions about which components to include. Bayesian inference provides a mathematical framework for systematically selecting model components, but defining prior distributions over model components and developing associated inference schemes has been challenging. We approach this problem in an amortized simulation-based inference framework: We define implicit model priors over a fixed set of candidate components and train neural networks to infer joint probability distributions over both, model components and associated parameters from simulations. To represent distributions over model components, we introduce a conditional mixture of multivariate binary distributions in the Grassmann formalism. Our approach can be applied to any compositional stochastic simulator without requiring access to likelihood evaluations. We first illustrate our method on a simple time series model with redundant components and show that it can retrieve joint posterior distribution over a set of symbolic expressions and their parameters while accurately capturing redundancy with strongly correlated posteriors. We then apply our approach to drift-diffusion models, a commonly used model class in cognitive neuroscience. After validating the method on synthetic data, we show that our approach explains experimental data as well as previous methods, but that our fully probabilistic approach can help to discover multiple data-consistent model configurations, as well as reveal non-identifiable model components and parameters. Our method provides a powerful tool for data-driven scientific inquiry which will allow scientists to systematically identify essential model components and make uncertainty-informed modelling decisions. | ['Jakob H. Macke', 'Cornelius Schröder'] | 2023-05-24 | null | null | null | null | ['bayesian-inference'] | ['methodology'] | [ 3.87071341e-01 -3.12537760e-01 -8.44314098e-02 -1.52038202e-01
-7.93523967e-01 -8.63647342e-01 8.43948066e-01 8.20914656e-02
-3.17118585e-01 8.33800495e-01 -2.06400886e-01 -5.88563502e-01
-4.97496903e-01 -4.30512577e-01 -7.26016819e-01 -8.13255727e-01
-1.61108181e-01 9.77876902e-01 2.38936380e-01 1.00722440e-01
3.37580442e-01 6.53780580e-01 -1.46246672e+00 -2.64701426e-01
8.11759233e-01 5.07869601e-01 2.83098072e-01 6.22164428e-01
9.89843458e-02 4.20971841e-01 -6.41667783e-01 -1.49156615e-01
-1.63526669e-01 -4.13401157e-01 -5.50634384e-01 -1.84642255e-01
-2.11095631e-01 -7.51384944e-02 1.83460951e-01 1.14857531e+00
3.44874293e-01 1.67883523e-02 1.10655165e+00 -1.27286065e+00
-2.00407192e-01 7.34502256e-01 -4.48067188e-01 1.76700920e-01
5.46494387e-02 3.06861401e-01 7.10893631e-01 -4.25089717e-01
4.93990511e-01 1.37079132e+00 6.30548835e-01 1.88387975e-01
-2.14282560e+00 -6.61759794e-01 1.04032099e-01 3.96568738e-02
-1.74013674e+00 -4.30566549e-01 6.65855110e-01 -7.84372330e-01
6.59368396e-01 8.16770047e-02 7.04873502e-01 1.40904582e+00
3.37514460e-01 5.38737118e-01 1.28402817e+00 -3.55808794e-01
7.97227919e-01 -4.80448455e-03 3.72496933e-01 4.35174286e-01
4.68555510e-01 2.87869573e-01 -3.48379344e-01 -6.94650829e-01
8.68038177e-01 -2.84655184e-01 -1.34328038e-01 -2.81052709e-01
-1.22113729e+00 6.53069496e-01 -5.30934334e-01 1.12895414e-01
-3.86734873e-01 6.75045252e-01 -1.09194823e-01 -5.15719093e-02
1.05088465e-01 4.05239373e-01 -4.53862309e-01 -2.25016043e-01
-1.06741953e+00 7.54862070e-01 8.96242678e-01 9.35638964e-01
5.90895832e-01 -7.64291957e-02 -9.10940468e-02 5.84628224e-01
5.65353513e-01 6.53707922e-01 2.41228372e-01 -1.43121958e+00
-3.40811551e-01 -4.06231396e-02 5.50965905e-01 -6.54911637e-01
-3.04408580e-01 -3.67779046e-01 -6.56258583e-01 1.80830434e-01
6.33939922e-01 -3.63240466e-02 -1.00267053e+00 2.19957638e+00
4.09529253e-04 2.16490656e-01 -3.04618090e-01 5.69985807e-01
4.50980254e-02 4.44723636e-01 1.31581917e-01 -3.65758538e-01
1.34485126e+00 8.23415965e-02 -6.60984516e-01 -5.32239601e-02
2.81392515e-01 -5.07079601e-01 8.40894759e-01 6.17955625e-01
-1.23176575e+00 -1.10367201e-01 -9.31276321e-01 3.19411099e-01
-6.96386769e-02 -3.09134983e-02 6.61568046e-01 7.47008741e-01
-1.09698749e+00 8.66265118e-01 -1.30274689e+00 -2.83176482e-01
3.64560753e-01 4.85723138e-01 1.26818508e-01 1.28933221e-01
-1.03443003e+00 1.02411962e+00 3.19659650e-01 -3.14049423e-02
-1.24969685e+00 -6.53475583e-01 -3.49729478e-01 5.23327030e-02
2.75831461e-01 -9.56744134e-01 1.40695250e+00 -5.67953646e-01
-1.51047921e+00 4.53097165e-01 -4.49292451e-01 -3.77620518e-01
2.48740405e-01 4.17106271e-01 -2.86286086e-01 -1.98487332e-03
-1.12870418e-01 4.63582218e-01 7.98340976e-01 -1.32199311e+00
1.67462170e-01 -1.61179781e-01 -1.35424599e-01 -2.60764331e-01
2.67884523e-01 2.07424372e-01 -3.38088542e-01 -5.18145382e-01
3.97600353e-01 -1.06287992e+00 -3.57162952e-01 1.28757833e-02
-4.64759558e-01 2.48884354e-02 1.00973405e-01 -3.77197921e-01
1.04544687e+00 -1.64182508e+00 3.86034578e-01 5.21489561e-01
2.41891623e-01 -2.97903150e-01 4.22951877e-02 4.87704843e-01
-8.97076130e-02 3.13581526e-01 -8.34732592e-01 -2.99824148e-01
3.71724516e-01 3.25006366e-01 -3.56370479e-01 4.47844565e-01
3.42721164e-01 5.59860468e-01 -7.22165287e-01 -3.45157295e-01
1.70474257e-02 5.84894657e-01 -4.49709713e-01 8.77689477e-03
-5.47914326e-01 3.90664101e-01 -5.17225266e-01 5.02249181e-01
6.32906556e-01 -4.22236204e-01 4.25087303e-01 -8.37953836e-02
4.21747612e-03 2.15053484e-01 -1.42137897e+00 1.23913443e+00
-1.55383185e-01 3.46065074e-01 9.38296095e-02 -1.15026236e+00
6.95081890e-01 3.91493440e-02 3.19201350e-01 -6.17206469e-02
4.05040145e-01 1.80329934e-01 3.65750283e-01 -2.78965741e-01
1.45426333e-01 -4.35920447e-01 -4.92919125e-02 8.39927018e-01
1.62642464e-01 -5.07676542e-01 1.73218206e-01 1.49820819e-01
1.05273902e+00 4.72870737e-01 2.05277532e-01 -6.48263097e-01
-1.31569237e-01 -1.13968730e-01 4.64908600e-01 1.07033181e+00
-3.40399556e-02 4.78958875e-01 8.52305532e-01 4.87567522e-02
-1.08959937e+00 -1.36008143e+00 -5.76580942e-01 4.48993415e-01
-5.13597727e-02 -3.96720886e-01 -6.79259598e-01 7.79677480e-02
-2.73400228e-02 1.22219050e+00 -6.38142109e-01 -2.17538849e-01
-8.40420499e-02 -1.32211494e+00 5.43052614e-01 3.78277510e-01
-2.17951298e-01 -5.65804958e-01 -6.73093259e-01 3.13073695e-01
-9.66124135e-05 -8.01828444e-01 -4.49826615e-03 5.85976958e-01
-8.56652141e-01 -9.49539304e-01 -4.73574638e-01 -3.85105684e-02
5.49290836e-01 -3.23207498e-01 1.06348348e+00 -2.92129278e-01
-3.37447912e-01 4.43798631e-01 3.38862896e-01 -4.71882135e-01
-7.58379996e-01 -4.65238392e-01 2.75312245e-01 -3.08015913e-01
3.06508452e-01 -9.33457077e-01 -2.26296082e-01 3.05807948e-01
-1.03459120e+00 1.36756152e-01 2.51410723e-01 7.62052536e-01
6.78712845e-01 1.15696840e-01 4.36416835e-01 -6.95845127e-01
8.51789534e-01 -7.13112295e-01 -1.19756162e+00 3.46465915e-01
-7.56138802e-01 5.68143845e-01 2.89303005e-01 -7.40330100e-01
-9.69101727e-01 -4.15025502e-02 1.78388447e-01 -4.65499133e-01
-1.97278649e-01 9.05995309e-01 -2.48221874e-01 2.69259006e-01
6.22881413e-01 2.06386164e-01 1.18108563e-01 -4.67486322e-01
2.85463363e-01 1.88221514e-01 3.74598652e-01 -1.26942313e+00
4.17603880e-01 3.63103241e-01 3.54964375e-01 -7.23665714e-01
-2.36606970e-01 1.92990914e-01 -4.36470479e-01 -7.18132481e-02
5.92369854e-01 -6.78919613e-01 -9.25991118e-01 3.82286757e-01
-1.29379857e+00 -4.43773568e-01 -5.97405769e-02 6.50288641e-01
-9.08474326e-01 1.91547483e-01 -3.95166367e-01 -1.27200508e+00
4.57368761e-01 -1.46132195e+00 1.07270062e+00 -6.43518567e-02
-5.87147593e-01 -1.07360291e+00 2.26933435e-01 -1.57161072e-01
3.98757666e-01 1.24807628e-02 1.20831907e+00 -6.86550438e-01
-7.71067321e-01 -2.32257575e-01 1.85056161e-02 -1.09235503e-01
-2.38811493e-01 4.88619745e-01 -9.82881010e-01 -3.34365405e-02
1.37036309e-01 -4.61471826e-03 8.51760566e-01 7.55740941e-01
1.35477877e+00 -8.62703845e-02 -6.40189588e-01 4.06813294e-01
1.07145739e+00 2.58390725e-01 4.35427397e-01 -1.23366103e-01
3.28141898e-01 6.86872661e-01 -5.29998392e-02 4.81900394e-01
2.01879531e-01 6.82716787e-01 1.24549247e-01 5.85073352e-01
4.20223981e-01 -1.53866217e-01 2.47533455e-01 5.84604263e-01
2.29959395e-02 -3.98437500e-01 -1.07926011e+00 4.02810842e-01
-1.79751372e+00 -1.03774726e+00 4.34733033e-02 2.45811820e+00
1.12027657e+00 3.12385231e-01 2.09751412e-01 -1.81647524e-01
8.54827344e-01 -4.52259004e-01 -9.26136315e-01 6.55977651e-02
-2.59347439e-01 4.09176916e-01 4.51911122e-01 6.25334859e-01
-6.40881896e-01 6.38865590e-01 7.95579338e+00 8.70581210e-01
-7.50473142e-01 -1.83033809e-01 5.31187236e-01 -2.36916468e-01
-7.54308164e-01 4.82215762e-01 -7.21036971e-01 4.77608860e-01
1.51267815e+00 -5.51120937e-01 6.01714432e-01 3.33404332e-01
4.60687548e-01 -4.97606188e-01 -1.30927646e+00 8.07605267e-01
-3.21270466e-01 -1.32151306e+00 -8.47224370e-02 2.61705816e-01
4.27802503e-01 -1.48484319e-01 1.64690323e-03 -1.51445523e-01
1.02274990e+00 -1.26536834e+00 9.42001104e-01 1.01664650e+00
5.03502309e-01 -5.06940186e-01 3.00796837e-01 4.31597054e-01
-4.71263200e-01 2.31890634e-01 -2.95578599e-01 -1.75129935e-01
1.40684247e-01 8.13599408e-01 -6.78815246e-01 1.13718785e-01
3.93907189e-01 4.11598623e-01 -2.86680430e-01 1.04420960e+00
5.35035990e-02 1.12407184e+00 -9.22743917e-01 -1.22271523e-01
-4.08014804e-01 -4.16665196e-01 6.06797159e-01 1.02868867e+00
5.29375196e-01 1.09218229e-02 -2.01945990e-01 1.73968947e+00
4.58926946e-01 -5.12450457e-01 -4.30947095e-01 -4.14099365e-01
9.58611608e-01 8.26884925e-01 -1.13508499e+00 -2.16792524e-01
1.46016568e-01 3.88194442e-01 1.17561825e-01 7.39575267e-01
-8.81613731e-01 -7.34200031e-02 6.44607306e-01 -1.10218704e-01
3.27136189e-01 -5.43660343e-01 -2.19028547e-01 -1.15936291e+00
-1.78528890e-01 -9.38925087e-01 -6.98302910e-02 -9.75762367e-01
-1.28499746e+00 1.80071592e-01 8.00960720e-01 -8.46395433e-01
-6.37109458e-01 -6.30691290e-01 -5.05609095e-01 1.29898620e+00
-8.82152438e-01 -6.95393205e-01 4.02662098e-01 2.44378582e-01
-1.64995939e-01 6.86866567e-02 9.27254677e-01 -2.18712330e-01
-7.56048024e-01 1.25574276e-01 3.57966781e-01 -4.88987565e-01
4.59703147e-01 -1.16103077e+00 2.40703613e-01 7.09036827e-01
-1.18077956e-02 1.19537520e+00 1.35003531e+00 -7.67114162e-01
-1.58374929e+00 -5.81094027e-01 3.70455474e-01 -6.25909746e-01
9.86487567e-01 -4.16918576e-01 -9.45558786e-01 7.21425831e-01
-1.37909099e-01 -4.09056783e-01 6.66554749e-01 1.01690255e-01
-2.97261149e-01 3.42304111e-01 -9.76848781e-01 8.86380374e-01
8.23447406e-01 -5.48835576e-01 -4.97750282e-01 2.57254899e-01
4.66158539e-01 -4.63770255e-02 -8.66663218e-01 1.28352195e-01
6.55756235e-01 -8.30893755e-01 8.72202456e-01 -6.01346731e-01
7.59667158e-02 -5.84907949e-01 -3.69612455e-01 -1.30905771e+00
-2.03812689e-01 -8.30806673e-01 -1.18782178e-01 1.11051810e+00
4.55622435e-01 -6.30564392e-01 3.19468915e-01 1.07375991e+00
1.37499109e-01 -3.82020444e-01 -8.35022509e-01 -7.97452450e-01
4.31620598e-01 -1.06394374e+00 6.94916725e-01 6.88981473e-01
-1.81186140e-01 1.77744120e-01 4.30899672e-02 2.65470535e-01
1.03453183e+00 7.07604066e-02 4.51773673e-01 -1.46505702e+00
-6.45130754e-01 -8.62874329e-01 -1.47425070e-01 -8.55717421e-01
2.24237919e-01 -8.07622433e-01 1.23949632e-01 -1.09509873e+00
3.76538217e-01 -4.96360242e-01 -6.67546466e-02 2.26457328e-01
3.04931626e-02 -2.01060146e-01 -1.60205513e-01 2.74808407e-01
-1.66646600e-01 5.74083269e-01 8.00065935e-01 -8.99126008e-03
9.62418541e-02 -7.03781992e-02 -8.13945949e-01 9.11976218e-01
5.12496948e-01 -7.02019572e-01 -6.26363397e-01 -4.85262834e-02
4.52966064e-01 3.76941144e-01 8.54810596e-01 -7.52454937e-01
2.16602921e-01 -5.38537681e-01 5.14373541e-01 -5.63311815e-01
3.07553023e-01 -4.83028352e-01 8.26400936e-01 1.82524860e-01
-5.40547252e-01 -1.24365620e-01 4.82503176e-01 6.58679783e-01
2.98262954e-01 -4.81543750e-01 6.05434239e-01 -7.41958804e-03
-7.49124587e-03 3.55222523e-02 -9.12993729e-01 -2.67371237e-02
5.77878952e-01 1.04833812e-01 -1.57045811e-01 -5.17740071e-01
-1.00681794e+00 2.19270155e-01 6.29771292e-01 -7.82101080e-02
3.28509122e-01 -9.71643806e-01 -5.74050605e-01 1.67735517e-01
-8.06796774e-02 -2.53944993e-01 1.38407126e-01 8.23432803e-01
-1.79730132e-01 1.68588713e-01 5.26159182e-02 -8.13740373e-01
-6.16436899e-01 5.08544624e-01 4.69942987e-01 1.25422180e-01
-9.49268416e-02 5.29806972e-01 1.91486865e-01 -2.03299671e-01
-6.94240108e-02 -5.95388591e-01 1.98434234e-01 -9.56218913e-02
4.04533625e-01 1.99790448e-01 -3.01212996e-01 -2.72638679e-01
-3.52104038e-01 4.04569596e-01 1.17835782e-01 -7.10122228e-01
1.16140342e+00 -2.11858138e-01 -1.47103533e-01 1.00815332e+00
7.91008115e-01 -2.19891295e-01 -1.40048778e+00 2.85056178e-02
5.12088463e-02 4.60566282e-02 9.99764726e-02 -9.40949440e-01
-3.12332958e-01 8.87474716e-01 3.90126944e-01 2.97083795e-01
8.49932790e-01 1.34239942e-01 -4.03302759e-02 3.26883763e-01
3.92726392e-01 -7.31251478e-01 -3.72619778e-01 2.37567440e-01
7.38730550e-01 -8.59267950e-01 1.43583983e-01 -5.74829765e-02
-2.85155922e-01 1.05386436e+00 1.26603916e-01 -6.40497357e-02
1.00347126e+00 6.24448061e-01 -4.20467168e-01 -2.11027041e-01
-1.10397065e+00 -2.49051824e-02 2.28810295e-01 6.75945878e-01
4.15267080e-01 8.28287303e-02 -9.25135240e-02 1.04516506e+00
-2.10778415e-01 2.45936379e-01 8.00358951e-01 8.35691988e-01
-2.82139987e-01 -1.31132567e+00 -6.17058218e-01 5.04574180e-01
-3.39761913e-01 -2.94217527e-01 -6.52548000e-02 7.50259101e-01
-2.46160313e-01 7.94560373e-01 1.16349101e-01 -1.25281826e-01
-9.48673040e-02 4.19744253e-01 8.64820778e-01 -4.22422379e-01
7.45545924e-02 5.17654538e-01 5.13706170e-02 -2.85699576e-01
-4.23850596e-01 -1.16021156e+00 -1.07372272e+00 -3.28966826e-01
-3.19683075e-01 1.00621454e-01 7.89622962e-01 1.13037944e+00
4.68998969e-01 5.24687767e-01 8.24178979e-02 -9.87622321e-01
-7.79877782e-01 -8.63182187e-01 -6.88271224e-01 -4.92269434e-02
2.28461757e-01 -1.04719543e+00 -5.17906427e-01 1.98509127e-01] | [6.831654071807861, 3.9442431926727295] |
1eee3c4a-5865-4ebd-82a2-8a5393e82a4c | a-faster-lighter-and-stronger-deep-learning | 2211.14864 | null | https://arxiv.org/abs/2211.14864v1 | https://arxiv.org/pdf/2211.14864v1.pdf | A Faster, Lighter and Stronger Deep Learning-Based Approach for Place Recognition | Visual Place Recognition is an essential component of systems for camera localization and loop closure detection, and it has attracted widespread interest in multiple domains such as computer vision, robotics and AR/VR. In this work, we propose a faster, lighter and stronger approach that can generate models with fewer parameters and can spend less time in the inference stage. We designed RepVGG-lite as the backbone network in our architecture, it is more discriminative than other general networks in the Place Recognition task. RepVGG-lite has more speed advantages while achieving higher performance. We extract only one scale patch-level descriptors from global descriptors in the feature extraction stage. Then we design a trainable feature matcher to exploit both spatial relationships of the features and their visual appearance, which is based on the attention mechanism. Comprehensive experiments on challenging benchmark datasets demonstrate the proposed method outperforming recent other state-of-the-art learned approaches, and achieving even higher inference speed. Our system has 14 times less params than Patch-NetVLAD, 6.8 times lower theoretical FLOPs, and run faster 21 and 33 times in feature extraction and feature matching. Moreover, the performance of our approach is 0.5\% better than Patch-NetVLAD in Recall@1. We used subsets of Mapillary Street Level Sequences dataset to conduct experiments for all other challenging conditions. | ['Songzhi Su', 'Ze Huang', 'Rui Huang'] | 2022-11-27 | null | null | null | null | ['camera-localization', 'loop-closure-detection', 'visual-place-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-1.58187389e-01 -4.80067372e-01 -4.83843714e-01 -3.30874085e-01
-6.55349255e-01 -5.15799344e-01 5.38725376e-01 -2.69421432e-02
-5.93492031e-01 5.38451910e-01 3.77421714e-02 -3.10548156e-01
-2.11007252e-01 -9.07373369e-01 -1.02172971e+00 -7.02772439e-01
3.39409411e-02 4.00912128e-02 6.32076383e-01 -3.92996222e-02
5.79857349e-01 7.97702551e-01 -1.57040715e+00 -1.01287186e-01
7.44487345e-01 1.29720676e+00 4.54090804e-01 3.85377467e-01
2.53415793e-01 7.17269063e-01 -1.89099759e-01 9.74496379e-02
2.54931092e-01 1.48439974e-01 -4.42338437e-01 -1.12469517e-01
9.94341254e-01 -2.67200083e-01 -7.88115025e-01 1.07729590e+00
5.76397240e-01 4.07146990e-01 4.73390788e-01 -1.32482708e+00
-7.63146102e-01 1.88307390e-02 -7.32590020e-01 3.73786122e-01
5.09387851e-01 1.73928216e-01 9.66521442e-01 -1.04129970e+00
5.34609795e-01 1.25724733e+00 7.62035549e-01 2.02763990e-01
-9.99513924e-01 -8.80461752e-01 3.28350067e-01 5.09453893e-01
-1.88462627e+00 -4.48998243e-01 6.36500478e-01 -2.20239177e-01
1.19765925e+00 5.90599552e-02 5.96916974e-01 1.00912154e+00
2.92489082e-01 9.08954084e-01 7.87341416e-01 -1.55648738e-01
9.91212279e-02 -1.49696872e-01 1.04243262e-02 1.09305859e+00
1.87098697e-01 1.13839068e-01 -6.14697099e-01 -1.47575021e-01
1.15970588e+00 4.99423057e-01 -3.21658760e-01 -4.16553289e-01
-1.51354873e+00 8.37189555e-01 1.05435300e+00 1.77148461e-01
-2.98689634e-01 5.63528061e-01 7.30290189e-02 1.63928084e-02
-5.80894947e-02 1.55701488e-01 -1.18263103e-01 9.17278305e-02
-7.35753298e-01 3.66178483e-01 3.50144088e-01 1.41511559e+00
9.93971050e-01 -1.52092531e-01 -2.35016152e-01 7.34091580e-01
5.50842524e-01 8.60840380e-01 5.62256277e-01 -6.94349945e-01
5.70514798e-01 7.12599218e-01 7.13392496e-02 -1.61294794e+00
-4.05286133e-01 -1.78969368e-01 -1.01592135e+00 -8.97379369e-02
-8.46164450e-02 8.89997557e-02 -1.03889298e+00 1.50263405e+00
1.98188186e-01 7.52799034e-01 -1.51211992e-01 8.90357077e-01
9.82168496e-01 8.07383716e-01 -5.26198335e-02 3.78475606e-01
1.31385803e+00 -1.25206316e+00 -4.18738335e-01 -5.19836426e-01
5.30655026e-01 -7.55064666e-01 6.48013175e-01 1.28142655e-01
-5.90895176e-01 -8.72578204e-01 -1.34079981e+00 -2.24800631e-01
-5.40264249e-01 4.59726304e-01 9.19470131e-01 1.30165577e-01
-1.12174213e+00 4.99570489e-01 -6.76367283e-01 -5.46172619e-01
4.43263680e-01 5.18127918e-01 -6.38436735e-01 -2.80711770e-01
-8.10271084e-01 7.27595687e-01 4.52506393e-01 3.10732782e-01
-7.52839327e-01 -2.24302307e-01 -1.33097672e+00 5.44589646e-02
1.78300336e-01 -5.67390680e-01 9.08863842e-01 -2.64175773e-01
-1.14614618e+00 6.59646869e-01 -4.89751905e-01 -4.51661170e-01
1.50913477e-01 -1.14098854e-01 -4.13952529e-01 -3.25303674e-02
3.27856719e-01 9.79135513e-01 6.01863801e-01 -7.06259608e-01
-9.25638318e-01 -4.15681779e-01 -3.91075164e-02 1.65323928e-01
9.00823995e-02 -2.88794398e-01 -6.87548220e-01 -4.33754623e-01
5.59937716e-01 -1.09351802e+00 -3.54484320e-01 3.65536541e-01
-1.90405548e-01 -4.45069641e-01 8.64217758e-01 -1.86501384e-01
1.03166389e+00 -2.41284013e+00 -3.11694264e-01 2.72915870e-01
2.41436988e-01 4.13318455e-01 -2.17528790e-01 3.56149286e-01
2.93322355e-01 -1.56498194e-01 -2.30658101e-03 -2.20684454e-01
7.03492165e-02 1.70615047e-01 -4.10859317e-01 7.77100563e-01
5.73867485e-02 9.99922752e-01 -9.12040889e-01 -5.66253901e-01
6.40745580e-01 3.66158903e-01 -5.73794961e-01 1.74377300e-02
1.49620697e-01 3.77832986e-02 -6.54960871e-01 8.41355741e-01
8.89262974e-01 -2.79755533e-01 -2.60143429e-01 -2.01873556e-01
-2.54074186e-01 7.41794631e-02 -1.41872001e+00 1.97170138e+00
-5.29764473e-01 9.35509562e-01 -3.59803259e-01 -8.69329631e-01
1.32823133e+00 -1.52187154e-01 6.49900436e-02 -9.67621267e-01
1.30320400e-01 2.23977908e-01 -1.70258790e-01 -4.31926996e-01
5.01692891e-01 6.26252353e-01 -2.09646359e-01 -1.39545694e-01
1.87670678e-01 2.23301843e-01 -3.28406021e-02 -1.09769925e-01
1.08492637e+00 5.68597466e-02 3.27703387e-01 -3.61923277e-01
7.03650653e-01 -5.49884923e-02 6.98030770e-01 9.35985208e-01
-3.25575352e-01 5.20382166e-01 -2.33827457e-02 -7.60519326e-01
-7.07381129e-01 -8.97158325e-01 -2.54576623e-01 7.86133647e-01
9.21118736e-01 -4.69319344e-01 -2.99494356e-01 -4.22587931e-01
2.46057853e-01 2.62291551e-01 -4.87466604e-01 -2.26850495e-01
-5.95035970e-01 -4.32418883e-01 7.42636740e-01 8.01787674e-01
1.18081963e+00 -1.11542010e+00 -5.29332399e-01 1.75133407e-01
-1.21424533e-01 -1.22698069e+00 -3.57448816e-01 -2.68947575e-02
-7.23736167e-01 -1.16182506e+00 -4.64608997e-01 -1.28860247e+00
7.39962637e-01 6.53566241e-01 7.89852679e-01 7.99996778e-02
-3.30719829e-01 -6.09667506e-03 -2.07345024e-01 -2.20254824e-01
6.30039692e-01 3.11933637e-01 1.35828005e-02 -1.91107169e-02
7.60225832e-01 -5.05835533e-01 -7.68698514e-01 5.13524175e-01
-5.40167093e-01 -1.58178985e-01 8.36260974e-01 9.54756081e-01
8.76383126e-01 -2.51369476e-01 8.62281322e-02 -3.98238093e-01
2.75607854e-01 -3.77255470e-01 -9.72202718e-01 8.57515186e-02
-4.62931484e-01 1.60243198e-01 6.13366306e-01 -3.30137789e-01
-5.41866660e-01 3.30762208e-01 -2.16995969e-01 -5.01547277e-01
-2.65309572e-01 2.03493372e-01 -1.25696614e-01 -6.83552563e-01
5.78399718e-01 5.68133652e-01 -3.89979243e-01 -3.99228454e-01
2.22114787e-01 6.60262108e-01 6.33866549e-01 -3.44149798e-01
9.36405003e-01 5.00122547e-01 1.34034038e-01 -9.05454338e-01
-6.34350955e-01 -8.20009530e-01 -5.45856297e-01 1.97041094e-01
7.83280432e-01 -1.14543676e+00 -1.06339574e+00 5.55813670e-01
-1.28201079e+00 6.75958097e-02 2.59516180e-01 6.89785659e-01
-4.61802602e-01 1.09242514e-01 -2.84153551e-01 -4.46295679e-01
-3.35487783e-01 -1.19907844e+00 1.28399503e+00 6.13261938e-01
3.50325525e-01 -9.09101009e-01 9.36077833e-02 -6.58724383e-02
3.67927909e-01 2.82480806e-01 3.83320034e-01 -5.35238266e-01
-1.14780653e+00 -2.85929233e-01 -5.07730186e-01 7.39624575e-02
-6.52115745e-03 -1.75002575e-01 -9.11415875e-01 -3.62656295e-01
-4.33254153e-01 -1.38354778e-01 1.05478525e+00 2.16847032e-01
1.34361887e+00 -2.19036609e-01 -7.42477834e-01 1.19197726e+00
1.76591444e+00 3.04903328e-01 7.47275531e-01 5.38877606e-01
8.28230739e-01 -2.08466537e-02 8.81202281e-01 2.80341297e-01
6.16535842e-01 7.65360355e-01 5.15236795e-01 -2.89993763e-01
5.29760309e-02 -5.95447898e-01 1.88047379e-01 6.20267749e-01
6.79705143e-02 -1.31609723e-01 -8.43396604e-01 7.27201819e-01
-2.37419605e+00 -1.01237547e+00 -3.52814496e-02 2.03478622e+00
2.96986133e-01 1.33853108e-02 -3.47090662e-01 -2.24412516e-01
7.13069081e-01 5.00119448e-01 -4.47930157e-01 -3.92794758e-02
-1.12600803e-01 1.74349889e-01 8.40511918e-01 4.49066490e-01
-1.49978256e+00 1.32581449e+00 6.03592396e+00 8.60747635e-01
-1.19964767e+00 -2.64493465e-01 3.15461874e-01 2.86943525e-01
1.45812005e-01 4.94185314e-02 -1.15359652e+00 4.66108412e-01
3.44983369e-01 8.06000158e-02 4.32066053e-01 1.23942208e+00
-1.14974625e-01 -1.08914204e-01 -1.18213534e+00 1.59900117e+00
3.44948232e-01 -1.52305162e+00 -1.33051001e-03 -1.12471161e-02
6.08811319e-01 5.10902524e-01 -8.84947926e-03 3.57333750e-01
5.02285548e-02 -1.16482115e+00 5.16489029e-01 5.24095893e-01
6.03531539e-01 -8.39069247e-01 1.03268886e+00 4.45494860e-01
-1.76522970e+00 -1.40398070e-01 -9.46762502e-01 -1.35022551e-01
-1.69213220e-01 1.50548652e-01 -7.23717451e-01 4.65230435e-01
7.84285486e-01 1.10307705e+00 -8.22430730e-01 1.53503084e+00
-1.82095945e-01 7.25594088e-02 -5.15279114e-01 -2.38830507e-01
6.65656269e-01 -7.40936324e-02 1.81861028e-01 1.19974113e+00
4.39511538e-01 -1.62569229e-02 5.04166484e-01 7.30630875e-01
2.33942084e-02 3.37439291e-02 -1.06199551e+00 4.39601779e-01
9.79524136e-01 1.35173297e+00 -5.02786934e-01 -2.99030662e-01
-3.51133168e-01 9.52101171e-01 4.56798255e-01 2.07428902e-01
-1.05956089e+00 -1.02510655e+00 7.19950318e-01 -1.66816428e-01
6.52759731e-01 -3.62953275e-01 2.51644909e-01 -1.16580832e+00
1.99016169e-01 -5.66411257e-01 1.20117441e-01 -8.58481467e-01
-9.83083367e-01 6.20162606e-01 -2.12885752e-01 -1.36828125e+00
-2.36900240e-01 -8.73974860e-01 -6.42614245e-01 8.01241875e-01
-1.81843567e+00 -1.15683818e+00 -8.94320607e-01 1.00631440e+00
4.63855267e-01 -5.39970100e-02 6.48821533e-01 5.03733158e-01
-5.59123337e-01 7.49988377e-01 8.78904313e-02 4.53088373e-01
6.15968704e-01 -8.67429435e-01 7.26304173e-01 8.99654627e-01
4.32164550e-01 8.66970897e-01 9.05906036e-02 -2.89012015e-01
-1.64447856e+00 -1.28388357e+00 9.32073653e-01 -2.68884063e-01
4.47970539e-01 -4.17062283e-01 -5.04416943e-01 7.26315200e-01
-3.94408964e-02 4.87768710e-01 2.34618694e-01 -8.58521312e-02
-3.90568107e-01 -4.30960208e-01 -1.09274411e+00 5.69814563e-01
1.39352417e+00 -6.37169778e-01 -6.52454197e-01 1.78575099e-01
6.59042239e-01 -6.26739860e-01 -6.00850880e-01 4.20658350e-01
3.97209376e-01 -9.05096352e-01 1.11697757e+00 -1.23022959e-01
-6.78393021e-02 -7.82697260e-01 -4.97437298e-01 -8.92246425e-01
-8.84196579e-01 -3.13448250e-01 5.62207587e-02 1.15939701e+00
2.58295357e-01 -8.41986775e-01 6.20263696e-01 2.80186571e-02
-1.55079380e-01 -8.41894448e-01 -9.31507707e-01 -8.24009657e-01
-5.10081887e-01 -1.88314959e-01 7.13216543e-01 8.11104178e-01
-3.69942039e-01 3.50514799e-01 -3.18967372e-01 5.14832914e-01
5.25763333e-01 2.73457378e-01 8.74954164e-01 -1.26020110e+00
5.59528098e-02 -2.31019616e-01 -1.10856378e+00 -1.76666176e+00
1.40721738e-01 -8.65155578e-01 1.94395870e-01 -1.70359957e+00
5.85616305e-02 -6.40237093e-01 -4.68929887e-01 6.92694247e-01
6.81119487e-02 3.37957084e-01 1.19282126e-01 1.21027306e-01
-9.74748313e-01 6.56330228e-01 1.00872564e+00 -3.49776119e-01
-1.62961353e-02 -2.18581975e-01 -4.02840018e-01 8.33037078e-01
6.56629384e-01 -3.99826229e-01 -4.44135010e-01 -5.44308662e-01
-6.91505298e-02 -3.18537980e-01 6.27991736e-01 -1.40586591e+00
6.57390356e-01 -1.83021992e-01 7.46987522e-01 -9.41614628e-01
4.37138170e-01 -1.01299214e+00 -1.52664501e-02 3.98228109e-01
9.43726003e-02 4.82664794e-01 2.47728348e-01 6.22313321e-01
-3.71452540e-01 -6.20446652e-02 5.55745959e-01 -1.15128495e-01
-1.43017650e+00 7.13342190e-01 -4.94627170e-02 -1.04037113e-01
1.12357593e+00 -4.25370663e-01 -4.38745528e-01 5.89816757e-02
-1.66426063e-01 3.46493155e-01 4.95818913e-01 6.71097994e-01
9.87400770e-01 -1.72790658e+00 -3.02323341e-01 4.72071528e-01
4.60673332e-01 3.46548438e-01 1.00229800e-01 6.30090415e-01
-9.02068317e-01 7.41744816e-01 -2.70776212e-01 -9.42254722e-01
-9.51081872e-01 5.89208066e-01 3.02734911e-01 1.27460584e-01
-7.66606271e-01 8.44253540e-01 2.04830155e-01 -4.30798411e-01
3.95941585e-01 -4.02711034e-01 -1.50814310e-01 -2.81129926e-01
5.72586060e-01 4.15699273e-01 -1.20385163e-01 -7.66402483e-01
-8.09935570e-01 1.08650565e+00 -6.53107166e-02 2.79660821e-01
1.19319117e+00 6.15878357e-03 -1.02328531e-01 9.21248496e-02
1.33455253e+00 -1.33767024e-01 -1.23750889e+00 -3.62546623e-01
-9.37162489e-02 -6.99239612e-01 1.26020253e-01 -3.35652590e-01
-1.00137913e+00 7.23379016e-01 8.24751079e-01 -2.34127223e-01
8.36854517e-01 -6.94102645e-02 8.22148979e-01 9.54154849e-01
8.03445935e-01 -8.87201011e-01 -1.72564864e-01 6.27383828e-01
6.90589786e-01 -1.34565389e+00 -3.23902778e-02 -3.02014381e-01
-2.37399012e-01 9.02309954e-01 8.89406145e-01 -7.87545979e-01
5.50248384e-01 -6.83160648e-02 -9.94437113e-02 -2.07793608e-01
-5.17505348e-01 -2.29895219e-01 4.08162177e-01 6.33502960e-01
7.38130361e-02 -1.37142405e-01 1.10819213e-01 2.51785785e-01
-1.28835633e-01 -1.31068677e-01 2.13739043e-03 9.02671456e-01
-6.00675881e-01 -7.38518834e-01 -1.31081209e-01 2.23751441e-01
-3.88920233e-02 -2.53720134e-01 -1.36876062e-01 8.74098420e-01
3.35945338e-01 7.68277943e-01 2.37475693e-01 -5.95645070e-01
4.16207671e-01 -3.76241297e-01 3.80523592e-01 -2.60419935e-01
-2.96817154e-01 -2.20496133e-01 -2.19114631e-01 -1.12767041e+00
-3.22176933e-01 -3.56330097e-01 -1.13418615e+00 -5.47691882e-01
-4.25104439e-01 -1.13203106e-02 5.47093689e-01 6.90267861e-01
8.33717704e-01 2.65786141e-01 6.60333931e-01 -9.59190905e-01
-2.57562220e-01 -7.23942459e-01 -3.61104906e-01 5.70382476e-02
5.53659618e-01 -9.77936804e-01 -2.43611559e-01 -2.70550638e-01] | [7.783807754516602, -1.9022700786590576] |
1447e64f-5335-4be4-90f9-a413a045d8ac | intrinsic-image-harmonization | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Guo_Intrinsic_Image_Harmonization_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Guo_Intrinsic_Image_Harmonization_CVPR_2021_paper.pdf | Intrinsic Image Harmonization | Compositing an image usually inevitably suffers from inharmony problem that is mainly caused by incompatibility of foreground and background from two different images with distinct surfaces and lights, corresponding to material-dependent and light-dependent characteristics, namely, reflectance and illumination intrinsic images, respectively. Therefore, we seek to solve image harmonization via separable harmonization of reflectance and illumination, i.e., intrinsic image harmonization. Our method is based on an autoencoder that disentangles composite image into reflectance and illumination for further separate harmonization. Specifically, we harmonize reflectance through material-consistency penalty, while harmonize illumination by learning and transferring light from background to foreground, moreover, we model patch relations between foreground and background of composite images in an inharmony-free learning way, to adaptively guide our intrinsic image harmonization. Both extensive experiments and ablation studies demonstrate the power of our method as well as the efficacy of each component. We also contribute a new challenging dataset for benchmarking illumination harmonization. Code and dataset are at https://github.com/zhenglab/IntrinsicHarmony. | ['Bing Zheng', 'Zhaorui Gu', 'Yufeng Jiang', 'Haiyong Zheng', 'Zonghui Guo'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['image-harmonization'] | ['computer-vision'] | [ 3.31916183e-01 -3.23083192e-01 3.66188765e-01 -2.35950038e-01
-5.38490832e-01 -6.46783352e-01 2.94330835e-01 -4.61066872e-01
-6.78722486e-02 5.19941330e-01 3.32811959e-02 2.28139713e-01
8.61766338e-02 -7.61902511e-01 -8.67134988e-01 -1.17034256e+00
8.38099301e-01 -6.50127307e-02 -2.59107560e-01 -1.57480195e-01
-1.03204310e-01 3.82161766e-01 -1.62728095e+00 1.07353531e-01
1.18043149e+00 7.79137790e-01 5.37761487e-02 4.88006592e-01
4.48457077e-02 5.14850020e-01 -4.42905098e-01 -3.40835333e-01
5.62573373e-01 -6.99146926e-01 -3.63725990e-01 5.12848914e-01
6.85211837e-01 -2.45796949e-01 -3.28564435e-01 1.18308508e+00
5.79237640e-01 2.11968757e-02 4.47034985e-01 -1.41250503e+00
-1.21279013e+00 1.75804049e-01 -9.15942192e-01 -3.78488958e-01
-5.22511080e-02 5.20793855e-01 8.40526402e-01 -6.36663139e-01
2.49558851e-01 9.85072792e-01 4.33490545e-01 3.29174370e-01
-1.57658470e+00 -7.31544793e-01 1.72516741e-02 -2.08363440e-02
-1.40820277e+00 -4.57670152e-01 1.23939049e+00 -3.17858785e-01
2.70467818e-01 4.10048723e-01 7.77651489e-01 9.19231117e-01
-3.39563610e-03 7.29491174e-01 1.26196611e+00 -3.84867400e-01
-1.50885075e-01 3.01626533e-01 -3.07460368e-01 5.47601283e-01
2.31404454e-01 2.18313619e-01 -3.42937529e-01 1.25838831e-01
8.47725272e-01 -4.72339317e-02 -6.93843722e-01 -3.16774786e-01
-1.12625849e+00 5.24249911e-01 5.03490031e-01 6.08689114e-02
-3.24145496e-01 -1.38057753e-01 -5.88381812e-02 7.72473365e-02
1.87791616e-01 4.45185930e-01 -2.55291790e-01 5.65207601e-01
-4.24003392e-01 -5.24869785e-02 4.43554968e-01 8.26815009e-01
1.16301930e+00 1.34907752e-01 -3.14788014e-01 1.09531844e+00
2.47364730e-01 7.56661773e-01 2.51665384e-01 -1.16634822e+00
2.78205164e-02 6.44512534e-01 2.41800770e-01 -1.20834708e+00
-2.07763806e-01 -2.76803464e-01 -1.21714783e+00 2.11123824e-01
2.25054771e-01 -1.99117288e-01 -6.63902223e-01 1.96589684e+00
5.71056247e-01 3.63764524e-01 4.95724902e-02 1.12682080e+00
9.39012110e-01 7.44499147e-01 -1.35539427e-01 -3.01686823e-01
1.32057512e+00 -1.13469136e+00 -6.77448273e-01 6.59327721e-03
-1.16538845e-01 -1.09314895e+00 1.30043221e+00 3.92677993e-01
-1.41276932e+00 -7.53518164e-01 -1.00910616e+00 -3.34544331e-01
-1.85494293e-02 3.89866769e-01 3.93120557e-01 3.95237923e-01
-9.01686966e-01 2.04243451e-01 -5.05995035e-01 3.46942879e-02
2.73515701e-01 9.77659971e-02 -7.87795112e-02 1.77095801e-01
-1.11240888e+00 5.68139195e-01 1.86475828e-01 2.23530024e-01
-5.84022641e-01 -7.67469108e-01 -7.31906474e-01 -1.12328477e-01
1.05511986e-01 -1.03134978e+00 7.45582044e-01 -1.27938271e+00
-1.83189690e+00 1.05882382e+00 2.22983174e-02 1.18674345e-01
4.16478306e-01 -1.15884967e-01 -4.92957622e-01 -2.50183772e-02
-1.72191277e-01 5.42661726e-01 1.16708100e+00 -1.88405478e+00
-4.75898474e-01 -1.66780099e-01 -1.15386611e-02 4.26356763e-01
-4.78028834e-01 -1.79535791e-01 -9.89323437e-01 -5.29910982e-01
2.42562108e-02 -7.85316348e-01 1.58292904e-01 2.99007539e-02
-6.31538570e-01 2.74880201e-01 6.79579377e-01 -6.93401456e-01
1.04894519e+00 -2.20446229e+00 3.07410866e-01 4.58435155e-02
3.92579705e-01 8.00317302e-02 -6.18559837e-01 -6.35579303e-02
-2.10411042e-01 -1.88621119e-01 -4.41909373e-01 -2.24184960e-01
-6.38502762e-02 1.29160985e-01 -2.37771243e-01 6.33303940e-01
3.16469133e-01 9.60106552e-01 -6.65807188e-01 -4.55500901e-01
3.84962231e-01 9.88561571e-01 -3.60884875e-01 5.04966378e-01
-4.16290425e-02 8.89375508e-01 -2.36961126e-01 6.12274528e-01
1.04458630e+00 -1.61364436e-01 2.10754154e-03 -1.00498080e+00
-2.06414938e-01 -3.59846741e-01 -1.09673476e+00 1.50897777e+00
-7.29251981e-01 3.92307460e-01 3.21127206e-01 -8.99782658e-01
9.74410176e-01 -1.05248392e-03 5.73531270e-01 -9.62576449e-01
3.51568282e-01 3.53706144e-02 -2.78233945e-01 -3.88291657e-01
2.27877125e-01 -8.67306720e-03 2.75802016e-01 5.73500216e-01
-2.49793127e-01 -4.15233672e-01 -1.27940461e-01 -3.32688987e-01
4.34573233e-01 2.46781468e-01 -2.33986545e-02 -7.97122270e-02
7.71168590e-01 -5.30680358e-01 6.19539320e-01 3.80503118e-01
-1.26899958e-01 8.95231783e-01 1.78923160e-01 -4.58777905e-01
-1.04565299e+00 -1.17643523e+00 -2.47440070e-01 7.65418053e-01
7.09638476e-01 1.27494201e-01 -8.62334549e-01 -1.17440864e-01
-1.69117555e-01 5.14936745e-01 -5.73403060e-01 -2.75619179e-01
-7.37559915e-01 -9.55146194e-01 3.36384237e-01 1.11553326e-01
7.85760581e-01 -8.73794198e-01 -4.26528066e-01 -1.32439777e-01
-3.98514420e-01 -9.55828011e-01 -7.32560635e-01 -1.95612520e-01
-2.65790641e-01 -1.13920283e+00 -6.70814157e-01 -7.00897336e-01
7.66846716e-01 6.46922231e-01 1.39791226e+00 2.01890677e-01
-5.92035711e-01 3.53127539e-01 2.42603514e-02 -2.88316190e-01
-2.68595546e-01 -3.20205182e-01 -1.60570621e-01 5.95272183e-01
-6.43197149e-02 -7.83105016e-01 -1.02204692e+00 6.41421854e-01
-1.10349023e+00 4.02682722e-01 5.90293109e-01 9.34801221e-01
8.83682787e-01 3.32881480e-01 -4.35223654e-02 -5.96580923e-01
5.12303770e-01 -2.06456691e-01 -8.92005920e-01 5.31877220e-01
-3.13735425e-01 -2.13601038e-01 5.88535309e-01 -4.07652050e-01
-1.37511921e+00 -2.63184635e-03 1.27014011e-01 -5.83641052e-01
-2.54081517e-01 -1.86541334e-01 -7.11506963e-01 -2.77580947e-01
5.34369528e-01 3.62262785e-01 -5.59569113e-02 -1.32936031e-01
6.52867854e-01 4.14332926e-01 7.84249187e-01 -7.55405366e-01
9.75920677e-01 6.83174253e-01 -1.26352236e-01 -6.89877093e-01
-7.47508943e-01 -5.17562851e-02 -3.69062155e-01 -2.73321360e-01
9.83724058e-01 -9.29534614e-01 -9.68750536e-01 8.60730231e-01
-1.18688202e+00 -5.92044950e-01 -2.71992683e-01 1.63208351e-01
-4.16933626e-01 2.92210728e-01 -6.00753486e-01 -5.17488062e-01
-4.13264900e-01 -1.12510455e+00 1.20071816e+00 5.74581921e-01
3.30961287e-01 -9.36587751e-01 2.27547899e-01 5.59320271e-01
5.13940990e-01 5.03532708e-01 7.17915177e-01 3.39102417e-01
-7.49831319e-01 2.10401751e-02 -6.32863939e-01 5.49123764e-01
5.62296569e-01 4.19342548e-01 -1.06658721e+00 -3.01428765e-01
1.52225599e-01 -2.39057809e-01 8.58235240e-01 4.70807076e-01
1.21410143e+00 -3.47579271e-01 1.57397345e-01 1.08069897e+00
1.60620677e+00 6.30717278e-02 7.17766345e-01 2.94335425e-01
1.05971348e+00 8.86785090e-01 2.82853216e-01 5.35736442e-01
4.17028010e-01 8.16521704e-01 3.85127693e-01 -8.39275718e-01
-4.53811795e-01 4.55573387e-02 9.62090865e-02 6.93313956e-01
4.30109241e-04 -2.69713193e-01 -6.12325311e-01 1.19115867e-01
-1.69949412e+00 -7.41953850e-01 -7.06609935e-02 2.17563939e+00
1.13080680e+00 -4.14625168e-01 -2.12849639e-02 -3.29556376e-01
1.02163649e+00 4.08578850e-02 -7.96484649e-01 2.73975879e-01
-5.93918622e-01 4.41380106e-02 2.89398491e-01 7.03523099e-01
-1.05103230e+00 8.24710548e-01 5.38461065e+00 6.65084898e-01
-1.41455352e+00 1.08206831e-01 9.27160263e-01 -1.01036653e-01
-6.60068572e-01 -1.87373951e-01 -3.61804187e-01 6.07824981e-01
1.37878627e-01 3.18544321e-02 9.59672511e-01 2.93388814e-01
1.10775098e-01 1.68332741e-01 -6.61211908e-01 1.17263901e+00
2.15383664e-01 -1.04426837e+00 6.76118284e-02 -3.02516073e-01
9.96790588e-01 -3.26177895e-01 4.75115091e-01 -1.15581118e-01
3.04121822e-01 -8.17767143e-01 6.06896400e-01 6.48890793e-01
7.15997219e-01 -5.91126978e-01 3.53809595e-01 -1.81649789e-01
-1.13317800e+00 1.36658102e-01 -3.58131617e-01 3.92577171e-01
8.18804726e-02 5.89839876e-01 1.34010583e-01 6.29184604e-01
7.26122558e-01 8.02712977e-01 -3.96062642e-01 6.26252174e-01
-4.81928110e-01 8.38273540e-02 -2.03856841e-01 6.26599848e-01
-3.03766996e-01 -8.07719767e-01 3.57098281e-01 8.04409146e-01
1.58579618e-01 2.01825231e-01 1.30980358e-01 1.49853432e+00
-5.26633263e-02 1.72184315e-02 -2.11274013e-01 2.33413294e-01
4.74005044e-01 1.61797583e+00 -3.31260651e-01 -1.16493173e-01
-4.93210584e-01 1.14059854e+00 1.56108215e-01 7.76335359e-01
-1.14916027e+00 -2.95818090e-01 7.85949707e-01 -2.04078689e-01
-5.45582809e-02 5.07825278e-02 -3.92041415e-01 -1.36692727e+00
1.51219368e-01 -9.34692681e-01 2.84831263e-02 -1.04857314e+00
-1.45522249e+00 6.13238752e-01 -2.58760840e-01 -1.31848907e+00
6.60219729e-01 -4.13203776e-01 -8.15977931e-01 1.03197324e+00
-1.73491657e+00 -1.51683509e+00 -9.58044469e-01 9.32784081e-01
6.63359091e-02 9.47233066e-02 5.67083240e-01 4.33340997e-01
-1.13670862e+00 5.73720932e-01 4.87562716e-02 -5.76131307e-02
1.09150481e+00 -1.05110347e+00 1.36708379e-01 7.96946704e-01
1.75429601e-02 5.10379136e-01 5.73743820e-01 -2.32085019e-01
-1.58138311e+00 -1.22390997e+00 1.18669905e-01 -3.18626553e-01
4.72856700e-01 -1.30261675e-01 -9.88695323e-01 3.81086200e-01
5.07679522e-01 -1.75822396e-02 6.46097600e-01 -2.78443784e-01
-5.70666909e-01 -5.45129240e-01 -8.12597096e-01 8.73375416e-01
7.38490641e-01 -5.32691419e-01 -1.03473626e-01 4.22037452e-01
7.82257080e-01 -3.75149190e-01 -8.32408786e-01 4.23891127e-01
6.04813635e-01 -1.26270318e+00 1.15466475e+00 -7.08459020e-02
5.58665872e-01 -7.06530571e-01 -1.00272074e-01 -1.23426080e+00
-4.87508297e-01 -6.53091848e-01 1.06825076e-01 1.45611441e+00
1.69837967e-01 -7.17467666e-01 4.03240949e-01 5.62942505e-01
-5.20071089e-02 -5.42874873e-01 -2.40954369e-01 -4.65521723e-01
3.10211219e-02 6.66157752e-02 8.52232516e-01 1.11903226e+00
-7.58511543e-01 3.81237656e-01 -5.37000299e-01 4.17883724e-01
8.81781638e-01 5.41770577e-01 1.07565892e+00 -7.55871236e-01
-5.49500823e-01 -7.37669110e-01 2.65452474e-01 -7.10272312e-01
3.16875041e-01 -5.38424790e-01 3.05113733e-01 -1.21647072e+00
4.41262484e-01 -3.98149937e-01 -3.85246962e-01 5.88112295e-01
-4.22135502e-01 6.69542491e-01 1.59353152e-01 3.54589611e-01
-4.55243111e-01 8.67911160e-01 1.62625134e+00 -3.32428187e-01
-4.92167383e-01 -4.35622513e-01 -9.41011012e-01 6.38871074e-01
8.68730664e-01 8.24956670e-02 -4.58190978e-01 -8.40439916e-01
1.23323329e-01 -1.25152066e-01 5.17471850e-01 -8.00646186e-01
1.32566303e-01 -4.41947728e-01 4.03514653e-01 -1.95544794e-01
2.00533897e-01 -7.43587315e-01 5.19348443e-01 1.55074820e-01
-2.63083756e-01 -3.83439213e-01 2.13075608e-01 3.79564524e-01
-2.88445950e-01 3.98041099e-01 1.14718902e+00 4.21640091e-02
-3.60258400e-01 5.70105731e-01 3.24692309e-01 -4.16597314e-02
8.70394051e-01 -8.88502747e-02 -5.29135406e-01 -3.07814270e-01
-2.58910447e-01 1.02655061e-01 8.22500706e-01 2.32513994e-01
4.67047662e-01 -1.47036779e+00 -8.77010047e-01 4.17919457e-01
8.18600506e-02 7.03187212e-02 7.24144101e-01 1.03398204e+00
-5.81816971e-01 -1.84144944e-01 -4.40266401e-01 -5.44867337e-01
-1.15110993e+00 5.64778268e-01 5.50249636e-01 1.25417769e-01
-4.37547475e-01 8.62048149e-01 8.80382180e-01 -4.04255718e-01
1.67031601e-01 1.15187332e-01 7.44653589e-05 -1.85205325e-01
4.74205703e-01 2.47631714e-01 -1.84688121e-01 -6.74041331e-01
-1.06110454e-01 1.07965863e+00 1.41237542e-01 2.16463581e-01
1.05195415e+00 -4.75588083e-01 -4.88728315e-01 1.35936335e-01
1.29403496e+00 9.56510454e-02 -1.49303949e+00 -2.52052873e-01
-8.46939206e-01 -6.19490027e-01 1.44436449e-01 -6.10823631e-01
-1.56240618e+00 7.62991428e-01 6.15232766e-01 1.77879483e-01
1.78787863e+00 -2.39780858e-01 9.23503578e-01 2.65402589e-02
-2.08467245e-01 -8.91067028e-01 3.36614609e-01 1.31852299e-01
1.01948047e+00 -1.33044183e+00 -1.92010701e-02 -4.11742210e-01
-5.71553528e-01 9.47001815e-01 7.85382688e-01 -7.16007352e-02
3.98493797e-01 2.85428613e-01 4.50779676e-01 -2.29174476e-02
-3.74005735e-01 -7.37704560e-02 5.00428438e-01 5.05537570e-01
4.55425203e-01 1.74181648e-02 5.89567935e-03 4.66185242e-01
-1.61577649e-02 -4.66441572e-01 1.07418276e-01 3.25132161e-01
-9.06398669e-02 -1.01921213e+00 -5.05791068e-01 -1.40515760e-01
-9.98713728e-03 -1.47003382e-01 -3.60816747e-01 5.48711836e-01
3.58696520e-01 8.86159420e-01 1.47442490e-01 -3.56250763e-01
2.56787062e-01 -5.18823266e-01 5.15334189e-01 -1.34290934e-01
-2.38510013e-01 5.07254541e-01 -5.78564107e-01 -6.75045013e-01
-6.39653623e-01 -3.98773611e-01 -8.60835493e-01 -3.06504071e-01
-2.61208832e-01 -1.84422269e-01 3.85824263e-01 4.93803650e-01
4.17921871e-01 5.55676341e-01 1.07590461e+00 -9.91598129e-01
-1.89305991e-01 -5.10144174e-01 -6.55587077e-01 6.73184216e-01
3.91169548e-01 -4.19772208e-01 -4.65384781e-01 3.18861693e-01] | [11.196016311645508, -1.3158165216445923] |
0cc03e56-c333-4672-b601-45cb7727493b | simple-question-answering-with-subgraph | 1904.04049 | null | http://arxiv.org/abs/1904.04049v1 | http://arxiv.org/pdf/1904.04049v1.pdf | Simple Question Answering with Subgraph Ranking and Joint-Scoring | Knowledge graph based simple question answering (KBSQA) is a major area of
research within question answering. Although only dealing with simple
questions, i.e., questions that can be answered through a single knowledge base
(KB) fact, this task is neither simple nor close to being solved. Targeting on
the two main steps, subgraph selection and fact selection, the research
community has developed sophisticated approaches. However, the importance of
subgraph ranking and leveraging the subject--relation dependency of a KB fact
have not been sufficiently explored. Motivated by this, we present a unified
framework to describe and analyze existing approaches. Using this framework as
a starting point, we focus on two aspects: improving subgraph selection through
a novel ranking method and leveraging the subject--relation dependency by
proposing a joint scoring CNN model with a novel loss function that enforces
the well-order of scores. Our methods achieve a new state of the art (85.44% in
accuracy) on the SimpleQuestions dataset. | ['Tagyoung Chung', 'Angeliki Metallinou', 'Anuj Goyal', 'Wenbo Zhao'] | 2019-04-04 | simple-question-answering-with-subgraph-1 | https://aclanthology.org/N19-1029 | https://aclanthology.org/N19-1029.pdf | naacl-2019-6 | ['fact-selection'] | ['natural-language-processing'] | [-4.10019308e-02 4.65302050e-01 -2.07394406e-01 -3.38367909e-01
-8.03021193e-01 -5.96937895e-01 5.21828175e-01 5.36644220e-01
-3.10404241e-01 7.31417954e-01 2.70831406e-01 -5.34617484e-01
-4.81057167e-01 -1.09834421e+00 -8.90126169e-01 -1.96224377e-01
2.07251370e-01 2.25034758e-01 7.53949761e-01 -4.77073550e-01
1.69143900e-01 1.55731291e-01 -1.45862103e+00 3.39005172e-01
1.12368810e+00 1.01080728e+00 -1.72242612e-01 4.16939944e-01
-3.59415799e-01 1.22534454e+00 -4.53071415e-01 -7.75983214e-01
-1.29437834e-01 -5.95393419e-01 -1.47733879e+00 -2.56069094e-01
7.61887074e-01 -1.57885343e-01 -4.58559334e-01 9.33274925e-01
3.09415936e-01 2.76123375e-01 2.94868290e-01 -1.06941938e+00
-7.78609872e-01 6.15485847e-01 -1.23073533e-01 3.88719916e-01
6.98010504e-01 -2.21629620e-01 1.75600564e+00 -4.87677664e-01
5.90924859e-01 8.87619853e-01 4.70493674e-01 3.67119282e-01
-8.28927279e-01 -1.27414748e-01 5.53219199e-01 8.32524061e-01
-1.19588888e+00 -1.57215089e-01 9.21387494e-01 -1.68774575e-01
1.05455279e+00 5.10583341e-01 5.53970158e-01 6.03941381e-01
-1.43658191e-01 7.24761069e-01 1.12149823e+00 -4.44204271e-01
2.71079421e-01 1.46473303e-01 7.22420335e-01 9.47215974e-01
3.24421972e-01 -5.25957406e-01 -6.60792053e-01 -1.68707430e-01
3.08813453e-01 -3.80940080e-01 -5.25970280e-01 -5.30095935e-01
-9.46151972e-01 8.83597314e-01 6.47795618e-01 4.91120338e-01
-2.31397867e-01 1.96198851e-01 9.87795368e-02 5.20858705e-01
1.87828839e-01 7.51857102e-01 -6.62758589e-01 9.69848037e-02
-7.58141816e-01 4.72858727e-01 1.23655987e+00 7.53575027e-01
9.43724990e-01 -5.03330767e-01 -5.09403765e-01 5.52138686e-01
2.90333152e-01 1.04119278e-01 1.72002539e-02 -7.42884517e-01
4.23989207e-01 1.24030769e+00 -5.73572442e-02 -1.19868898e+00
-5.13720930e-01 -6.88726783e-01 -3.53845358e-01 -2.93943286e-01
6.34806812e-01 1.05518900e-01 -5.76459348e-01 1.90004659e+00
7.00933516e-01 2.10436389e-01 -4.71822210e-02 9.18230176e-01
1.28777027e+00 4.66182917e-01 -6.58109710e-02 -9.91980582e-02
1.73484766e+00 -1.22845352e+00 -5.60848534e-01 2.68888585e-02
7.19691038e-01 -2.87928402e-01 1.23774529e+00 1.29440039e-01
-9.26548302e-01 -1.54119119e-01 -9.52724636e-01 -6.44091427e-01
-6.93249226e-01 -2.04148248e-01 7.53319144e-01 5.66048324e-01
-1.25357151e+00 3.49222302e-01 -3.94630253e-01 -5.13460994e-01
2.52512455e-01 1.85576409e-01 -2.88731426e-01 -3.28364104e-01
-1.56494975e+00 1.02557087e+00 2.79131889e-01 -1.10511810e-01
-5.28980911e-01 -8.14925313e-01 -5.97606897e-01 3.63695592e-01
1.11094463e+00 -1.24751830e+00 1.13927591e+00 -3.93400997e-01
-1.44277656e+00 6.80959821e-01 -1.64299771e-01 -4.35590804e-01
2.50534147e-01 -3.03587914e-01 -3.60638440e-01 4.16067183e-01
9.98509303e-03 7.76515007e-02 5.05341113e-01 -1.20131505e+00
-5.10557592e-01 -4.98599380e-01 8.15813601e-01 3.45363647e-01
-5.73719561e-01 1.83816310e-02 -6.71069980e-01 -2.61357486e-01
8.07579607e-03 -5.64339876e-01 -1.25417233e-01 -1.70076057e-01
-4.94778752e-01 -6.50294840e-01 4.94387358e-01 -8.33901465e-01
1.65775156e+00 -1.60917044e+00 1.34954065e-01 2.26333618e-01
6.49294555e-01 3.44690651e-01 -5.87721728e-02 5.42289078e-01
1.12340003e-01 1.86117068e-01 -2.07357422e-01 1.89231724e-01
2.62705266e-01 -2.70104725e-02 -3.13224673e-01 7.85061391e-04
3.31735492e-01 1.26317644e+00 -1.19688773e+00 -3.99007052e-01
-1.26093805e-01 1.85527131e-01 -6.89703405e-01 1.27566233e-01
-7.40065217e-01 5.01495302e-02 -8.02661657e-01 5.76466501e-01
4.43224162e-01 -6.97727323e-01 2.70793200e-01 -3.79383057e-01
2.20356658e-01 6.97737157e-01 -1.02237844e+00 1.60072803e+00
-2.85982609e-01 1.20904930e-01 -2.51538932e-01 -1.07285392e+00
7.29827702e-01 1.17066883e-01 2.88292736e-01 -6.36296511e-01
-1.28850281e-01 3.24500084e-01 -4.56537381e-02 -7.39491940e-01
4.53080326e-01 -4.24447171e-02 -7.79528916e-02 3.99709284e-01
1.95197985e-01 -1.47550359e-01 5.09146631e-01 5.86313188e-01
1.57410944e+00 9.09224153e-02 4.17486519e-01 -2.45067462e-01
6.63291514e-01 1.42301098e-01 1.62007615e-01 8.36561382e-01
-3.25834155e-02 4.54120338e-01 8.11394095e-01 -2.70926446e-01
-5.17782032e-01 -8.93770874e-01 2.49706551e-01 1.05096543e+00
6.62888363e-02 -6.57015085e-01 -7.39385247e-01 -1.22548532e+00
6.21649325e-02 7.10398674e-01 -7.89913833e-01 -2.74639785e-01
-5.08781791e-01 -4.52242732e-01 3.12947869e-01 2.95744359e-01
5.81775486e-01 -9.41342473e-01 -2.51021624e-01 1.48457974e-01
-4.24383670e-01 -1.33896470e+00 -2.20696330e-01 -6.76574558e-02
-7.15415895e-01 -1.45832241e+00 -4.93972689e-01 -7.26666510e-01
3.85082722e-01 3.97986948e-01 1.59264135e+00 3.72581631e-01
3.08891404e-02 7.18797386e-01 -5.92699707e-01 -1.93589002e-01
7.82624707e-02 4.51418608e-01 -4.97230411e-01 9.19322371e-02
3.90005052e-01 -7.13776112e-01 -7.12681651e-01 1.31901011e-01
-1.04353547e+00 -2.45817557e-01 5.93371451e-01 4.59995717e-01
3.42006564e-01 -3.03087831e-02 9.24008965e-01 -9.91011739e-01
8.24624062e-01 -6.41293645e-01 -4.09857303e-01 8.71284962e-01
-7.08867788e-01 1.35620460e-01 6.66746557e-01 1.07668582e-02
-9.51645553e-01 -4.08877879e-01 -1.84735924e-01 1.23121120e-01
-2.83906143e-02 9.71246064e-01 -1.76636145e-01 -2.34963551e-01
5.83281457e-01 6.50361478e-02 -3.80855799e-01 -4.15978879e-01
6.83877528e-01 2.23471835e-01 2.41673067e-01 -6.15332127e-01
8.91546905e-01 4.19155359e-01 1.06108718e-01 -6.52154326e-01
-1.42062330e+00 -7.29505718e-01 -2.56973714e-01 -1.36392862e-01
7.05990016e-01 -6.40493274e-01 -9.34036493e-01 6.59643412e-02
-9.72236574e-01 5.60722686e-02 -4.40982938e-01 2.34644160e-01
-3.07894051e-01 7.06337094e-01 -3.82678658e-01 -6.13756418e-01
-3.37601691e-01 -6.97501540e-01 7.20055103e-01 1.96971878e-01
-4.53296304e-02 -8.97631645e-01 2.29938909e-01 9.16523218e-01
5.21007419e-01 1.30429760e-01 1.30287623e+00 -9.79200959e-01
-1.07782602e+00 -2.35413164e-01 -4.65330273e-01 1.26602456e-01
-2.42554545e-02 -3.58540475e-01 -8.11184704e-01 2.15068236e-02
1.59323157e-03 -3.93605292e-01 1.17865121e+00 -1.01576142e-01
1.02666044e+00 -3.66540253e-01 -3.15911509e-02 2.22508937e-01
1.39560390e+00 -2.74814576e-01 7.06911981e-01 3.84556025e-01
7.27527380e-01 6.81815922e-01 2.25214154e-01 1.51727512e-01
1.10687852e+00 6.71835065e-01 4.97227013e-01 2.60983318e-01
-3.34740371e-01 -4.22882318e-01 1.05499653e-02 9.97867584e-01
-1.50709838e-01 -3.01833838e-01 -8.02857816e-01 6.91371322e-01
-1.96105587e+00 -7.51022816e-01 -3.95296812e-01 2.01011491e+00
9.34546769e-01 1.35718221e-02 3.43756899e-02 1.35096833e-01
3.52151453e-01 3.69155288e-01 -5.03710091e-01 -9.98040959e-02
-2.64680594e-01 3.32873076e-01 -3.58315930e-02 5.04188716e-01
-8.81149650e-01 9.86034811e-01 5.55706167e+00 9.18833494e-01
-6.65609956e-01 -3.03273611e-02 3.14812005e-01 3.46845329e-01
-8.58809352e-01 3.01534295e-01 -7.23686993e-01 1.48385838e-01
6.11482203e-01 -2.03349382e-01 4.85164613e-01 5.57612777e-01
-2.30609611e-01 -2.63329118e-01 -8.73871028e-01 5.26644766e-01
3.26853901e-01 -1.35251451e+00 3.56387645e-01 -3.39547038e-01
5.72202563e-01 -2.51334906e-01 -1.77493885e-01 5.69230974e-01
2.47368902e-01 -7.96450734e-01 4.02237475e-01 7.07804203e-01
2.27116823e-01 -3.76741350e-01 7.50328839e-01 3.11136395e-01
-1.18694270e+00 5.48775531e-02 -8.72700810e-02 -1.24666944e-01
-1.87650379e-02 9.15966690e-01 -6.66834116e-01 1.29871881e+00
7.59849191e-01 3.64902824e-01 -7.33354807e-01 1.09812391e+00
-6.68636024e-01 9.27126884e-01 -3.05780679e-01 -3.95059645e-01
1.84933260e-01 -8.97191018e-02 3.98121387e-01 8.86916101e-01
5.96810430e-02 3.56019318e-01 4.24977019e-03 6.99250162e-01
-4.56136167e-01 4.71669793e-01 -2.98133492e-01 -1.55508026e-01
6.90174252e-02 1.27575123e+00 -5.58687270e-01 -2.39375487e-01
-5.62717021e-01 6.16720378e-01 1.00457191e+00 4.50343311e-01
-6.54593110e-01 -4.83367890e-01 2.21275896e-01 1.53517351e-01
4.68645990e-01 -1.69357985e-01 -2.17489824e-01 -1.43463433e+00
5.74011803e-01 -8.84200811e-01 7.75345147e-01 -5.93999147e-01
-1.34672308e+00 4.70876336e-01 -3.48972119e-02 -5.93132913e-01
6.57014623e-02 -4.54541266e-01 -5.53286374e-01 5.45649409e-01
-2.17607236e+00 -1.26183343e+00 -4.66065198e-01 6.43458426e-01
-1.48239493e-01 2.78931946e-01 7.92458475e-01 2.48719186e-01
-4.89928454e-01 5.53052545e-01 -2.69616961e-01 -1.24189928e-01
4.60260510e-01 -1.36346197e+00 2.24207669e-01 7.39354968e-01
2.86244780e-01 8.33405435e-01 4.74574000e-01 -4.00296271e-01
-1.58743608e+00 -8.21226180e-01 1.48941779e+00 -6.28543735e-01
9.45661426e-01 -2.58939594e-01 -1.27865386e+00 4.98673022e-01
3.13236266e-01 -1.28668193e-02 6.74065650e-01 5.61695099e-01
-6.89063251e-01 -2.65536129e-01 -9.07926500e-01 6.21713877e-01
1.21106088e+00 -5.97663641e-01 -7.16686010e-01 4.21779811e-01
1.02062118e+00 -1.44046903e-01 -8.43979895e-01 6.06833935e-01
1.91374809e-01 -1.01537216e+00 8.41852188e-01 -9.97794628e-01
3.31140101e-01 -4.36551929e-01 -3.84628922e-02 -9.57887709e-01
-2.99772590e-01 -7.01610982e-01 -6.67665243e-01 1.09098291e+00
6.67986333e-01 -8.24762881e-01 9.09196854e-01 4.74958748e-01
-1.37803033e-01 -1.21265936e+00 -1.00504398e+00 -7.16679692e-01
-5.98587561e-03 -3.38786960e-01 7.16589034e-01 1.00934970e+00
2.96045899e-01 7.35653043e-01 -1.29818991e-01 2.76511878e-01
3.13366503e-01 5.77180445e-01 7.13701785e-01 -1.23459423e+00
-2.75990069e-01 -5.29494941e-01 -1.85093880e-01 -1.17288256e+00
4.82000560e-02 -1.00838602e+00 -2.82428682e-01 -2.32718992e+00
3.37827921e-01 -1.14270046e-01 -4.84608173e-01 4.90695417e-01
-7.00109243e-01 -6.99280351e-02 4.28253412e-02 -1.20322175e-01
-1.18833554e+00 5.92474461e-01 1.35223615e+00 -1.79074388e-02
2.54540920e-01 -7.44872913e-02 -1.06339514e+00 4.96896535e-01
8.04215789e-01 -2.30845526e-01 -5.02743423e-01 -3.13288331e-01
8.67601693e-01 -3.05058598e-03 4.74274844e-01 -7.92053223e-01
5.74599504e-01 -1.20382063e-01 -3.62124354e-01 -3.14768225e-01
2.52647519e-01 -6.01239800e-01 -3.38938862e-01 2.32716918e-01
-3.63232404e-01 -1.11315280e-01 -8.77024308e-02 8.77428830e-01
-4.52900589e-01 -1.48410752e-01 2.85601825e-01 -9.34666991e-02
-7.80445993e-01 3.43466699e-01 1.68805271e-01 5.94804704e-01
8.11449707e-01 4.97829951e-02 -6.04251504e-01 -5.79522908e-01
-5.82016885e-01 5.21345437e-01 3.91370729e-02 3.50817859e-01
5.75057507e-01 -1.09079194e+00 -8.17922950e-01 -3.49034935e-01
4.63209987e-01 -3.69892977e-02 3.95966679e-01 8.61018419e-01
-3.27543765e-01 6.93889081e-01 3.08281064e-01 -1.71688609e-02
-9.23889697e-01 5.87238073e-01 2.87519395e-01 -8.39187086e-01
-2.73218006e-01 1.01616478e+00 -2.73905061e-02 -6.46101296e-01
1.70316935e-01 -3.95022392e-01 -6.10980868e-01 -1.20699108e-02
2.51647562e-01 4.05110627e-01 4.12372202e-01 -1.68945670e-01
-4.45295274e-01 4.54330266e-01 -7.83315450e-02 3.01865697e-01
1.13346159e+00 -4.95164134e-02 -3.72534215e-01 2.25519627e-01
9.84340370e-01 1.10997841e-01 -6.15236700e-01 -4.96875316e-01
3.16179007e-01 -2.26892114e-01 -5.17503731e-02 -1.10960352e+00
-8.65713537e-01 6.39470041e-01 -1.29613191e-01 7.76724219e-01
1.03937447e+00 2.99812526e-01 9.77456927e-01 7.94375300e-01
3.41847777e-01 -9.34606433e-01 2.78410882e-01 7.84814239e-01
8.89235437e-01 -1.15168250e+00 -6.92356902e-04 -8.47881854e-01
-2.76976556e-01 9.06750202e-01 5.77550173e-01 -6.51711151e-02
6.74775660e-01 -3.10524315e-01 -1.31543085e-01 -6.26918137e-01
-8.16872835e-01 -6.45324886e-01 7.21748173e-01 2.39532933e-01
3.83737504e-01 -1.74916789e-01 -6.22123539e-01 7.11034596e-01
-2.20494300e-01 6.53120801e-02 2.42956132e-01 8.57845962e-01
-5.81073105e-01 -9.12316859e-01 7.36981705e-02 6.86237037e-01
-5.85474670e-01 -2.79721260e-01 -6.41964674e-01 6.36381924e-01
-2.58303016e-01 1.25922918e+00 -6.04441345e-01 -3.35663766e-01
6.74220920e-01 1.48106262e-01 5.09594858e-01 -7.14093208e-01
-7.96126008e-01 -6.94375336e-01 4.46745008e-01 -5.29562056e-01
-4.75824296e-01 -2.91830182e-01 -1.01436639e+00 -1.81663245e-01
-6.12070501e-01 4.48316902e-01 4.38001037e-01 1.18209064e+00
5.50207138e-01 5.91302991e-01 4.84065562e-01 2.43898794e-01
-7.06827402e-01 -6.73002958e-01 -5.17626464e-01 3.86374056e-01
1.25984326e-01 -4.33448225e-01 -3.47788125e-01 -3.47118080e-01] | [10.503904342651367, 7.934709072113037] |
5284888c-c977-4bce-8eda-78a3ac2ec4e8 | bmn-boundary-matching-network-for-temporal | 1907.09702 | null | https://arxiv.org/abs/1907.09702v1 | https://arxiv.org/pdf/1907.09702v1.pdf | BMN: Boundary-Matching Network for Temporal Action Proposal Generation | Temporal action proposal generation is an challenging and promising task which aims to locate temporal regions in real-world videos where action or event may occur. Current bottom-up proposal generation methods can generate proposals with precise boundary, but cannot efficiently generate adequately reliable confidence scores for retrieving proposals. To address these difficulties, we introduce the Boundary-Matching (BM) mechanism to evaluate confidence scores of densely distributed proposals, which denote a proposal as a matching pair of starting and ending boundaries and combine all densely distributed BM pairs into the BM confidence map. Based on BM mechanism, we propose an effective, efficient and end-to-end proposal generation method, named Boundary-Matching Network (BMN), which generates proposals with precise temporal boundaries as well as reliable confidence scores simultaneously. The two-branches of BMN are jointly trained in an unified framework. We conduct experiments on two challenging datasets: THUMOS-14 and ActivityNet-1.3, where BMN shows significant performance improvement with remarkable efficiency and generalizability. Further, combining with existing action classifier, BMN can achieve state-of-the-art temporal action detection performance. | ['Xiao Liu', 'Tianwei Lin', 'Errui Ding', 'Shilei Wen', 'Xin Li'] | 2019-07-23 | bmn-boundary-matching-network-for-temporal-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Lin_BMN_Boundary-Matching_Network_for_Temporal_Action_Proposal_Generation_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Lin_BMN_Boundary-Matching_Network_for_Temporal_Action_Proposal_Generation_ICCV_2019_paper.pdf | iccv-2019-10 | ['temporal-action-proposal-generation'] | ['computer-vision'] | [ 2.03141317e-01 -9.22446847e-02 -5.74662387e-01 -1.97754085e-01
-1.10920310e+00 -2.07322780e-02 6.48842454e-01 -2.11075768e-01
-4.17371511e-01 7.30149031e-01 6.57603621e-01 2.97569185e-01
3.17702085e-01 -5.91446459e-01 -4.02844191e-01 -4.87830341e-01
-1.24639817e-01 2.55214125e-01 1.11695111e+00 1.31440058e-01
1.67925030e-01 1.10500440e-01 -1.34733856e+00 5.22531688e-01
5.98273575e-01 1.23048711e+00 1.56396896e-01 4.25208718e-01
1.66469038e-01 8.80330443e-01 -5.55233061e-01 -2.30299145e-01
2.24232346e-01 -4.59413558e-01 -6.80215299e-01 5.03941402e-02
4.68380153e-01 -8.54248345e-01 -5.93496919e-01 7.32972324e-01
5.68733573e-01 3.47186834e-01 6.50689662e-01 -1.33339882e+00
-1.89910367e-01 5.58220088e-01 -7.05267668e-01 4.86723185e-01
4.67538148e-01 4.27858680e-01 9.40698445e-01 -1.10020101e+00
8.30296278e-01 1.25740957e+00 5.08178830e-01 7.46139944e-01
-8.13027978e-01 -7.15988636e-01 6.29216552e-01 5.43980718e-01
-1.35236704e+00 -3.91440868e-01 5.54443538e-01 -1.19130217e-01
9.26825166e-01 1.49169527e-02 7.83523858e-01 1.24652195e+00
6.79677352e-02 1.56043565e+00 4.67261255e-01 1.10893182e-01
2.85459191e-01 -7.11632431e-01 -4.75908518e-01 5.26027977e-01
-3.41189443e-03 -4.00416069e-02 -7.51671195e-01 -1.44385070e-01
1.07433569e+00 1.91419572e-01 -1.91563085e-01 -2.83247888e-01
-1.77974927e+00 5.67939818e-01 5.98927021e-01 2.44375780e-01
-5.97329855e-01 4.89862412e-01 5.26760519e-01 -3.32523525e-01
3.30379575e-01 1.49250766e-02 -1.90615296e-01 -3.37166250e-01
-1.14206040e+00 3.76432717e-01 4.98777442e-02 1.06327391e+00
4.13110197e-01 -1.72113776e-01 -1.18786454e+00 7.50196218e-01
3.90608817e-01 2.12079316e-01 6.81251168e-01 -1.22198427e+00
7.61185408e-01 6.28314793e-01 4.63683218e-01 -8.20553005e-01
-1.81761280e-01 -1.89075813e-01 -6.87256753e-01 -1.36151657e-01
2.59517223e-01 6.47560284e-02 -9.62695897e-01 1.84260559e+00
7.72899926e-01 7.09061086e-01 -2.57348537e-01 9.96394336e-01
8.73015821e-01 7.87813723e-01 4.58372414e-01 -3.28441054e-01
1.22856677e+00 -1.49985147e+00 -6.14608228e-01 -1.88425452e-01
5.42720497e-01 -4.47810292e-01 6.41686380e-01 2.31557265e-01
-1.35493648e+00 -7.57946849e-01 -8.10437918e-01 -5.99209312e-03
2.76145965e-01 5.37207723e-01 6.82968140e-01 -1.50557682e-01
-9.71330285e-01 6.73137903e-01 -1.08704007e+00 -2.62617260e-01
8.17372978e-01 3.28850858e-02 -3.70226711e-01 -7.30887726e-02
-1.00863755e+00 6.37880445e-01 5.86816370e-01 2.41394699e-01
-1.24993551e+00 -3.58079433e-01 -8.91804099e-01 -6.50653467e-02
5.58756471e-01 -7.57205367e-01 1.68201768e+00 -6.79902315e-01
-1.32288325e+00 4.79132801e-01 -3.28542590e-01 -5.94897985e-01
8.06029797e-01 -2.40439579e-01 -4.64653820e-01 4.60495204e-01
6.72257960e-01 1.49380767e+00 7.01197267e-01 -5.18287361e-01
-1.18399179e+00 -2.82345396e-02 -2.73126394e-01 2.77814656e-01
-1.38450161e-01 1.47176996e-01 -8.58302593e-01 -9.04469728e-01
5.51969111e-01 -5.12235880e-01 -4.92906481e-01 4.03323293e-01
-2.03011781e-01 -8.67824376e-01 5.69764137e-01 -5.08929789e-01
1.56646502e+00 -1.86474931e+00 -1.16132312e-01 -1.80265278e-01
2.57509742e-02 3.41512620e-01 -3.77015859e-01 1.60078555e-01
2.36854568e-01 -2.32956186e-01 -3.98236979e-03 -3.31130177e-01
9.99898836e-02 -3.56518999e-02 -3.90506536e-01 2.79665023e-01
5.00728250e-01 1.19648743e+00 -1.30212104e+00 -9.12475586e-01
1.77941352e-01 1.69530883e-01 -4.99161601e-01 3.12681884e-01
-3.67314816e-01 2.28259742e-01 -8.69064212e-01 9.49603736e-01
3.47750425e-01 -2.90004432e-01 4.75852676e-02 -2.22134158e-01
5.37296571e-02 4.33100611e-01 -1.18201292e+00 2.00987720e+00
2.01698005e-01 4.20721650e-01 -4.20814663e-01 -7.63067067e-01
9.26092386e-01 3.85966510e-01 7.19367087e-01 -8.83955479e-01
-1.91608772e-01 1.35565445e-01 -3.81566077e-01 -3.76073390e-01
6.16533875e-01 -3.52899432e-02 -1.78175375e-01 5.42380452e-01
9.65515226e-02 2.73161680e-01 3.73437643e-01 2.61776626e-01
1.50999057e+00 7.57080674e-01 4.45716709e-01 3.48759800e-01
5.45144379e-01 -2.60732889e-01 1.14044988e+00 7.14481592e-01
-9.18224156e-01 7.93051124e-01 4.15483952e-01 -6.69468641e-01
-8.79187584e-01 -1.14665890e+00 2.04725474e-01 1.05201077e+00
3.35307389e-01 -4.40890759e-01 -5.45102179e-01 -1.18309927e+00
-2.93547571e-01 3.58043849e-01 -5.26616633e-01 -2.88827479e-01
-9.63280439e-01 -2.96733946e-01 3.09088200e-01 1.02817869e+00
8.97170186e-01 -1.68376744e+00 -6.28816187e-01 6.18425131e-01
-7.37405777e-01 -1.04313719e+00 -9.72114205e-01 -5.46690226e-01
-1.12726796e+00 -9.92253006e-01 -1.06004357e+00 -5.70174634e-01
5.45050263e-01 4.35836375e-01 1.24041581e+00 -1.82310585e-02
-2.58585691e-01 2.23109320e-01 -5.13916552e-01 -1.42648362e-03
1.85330901e-02 -2.26221815e-01 -1.03009500e-01 -1.45027703e-02
4.13852245e-01 -4.75133568e-01 -1.18209457e+00 8.83998394e-01
-7.50384212e-01 1.57440186e-01 7.65663445e-01 5.51249743e-01
8.84741724e-01 -3.21966618e-01 7.68606782e-01 1.08466610e-01
2.48651043e-01 -4.25048292e-01 -3.22550744e-01 3.77063483e-01
-1.18168235e-01 -7.14986995e-02 -2.69235224e-02 -6.04479611e-01
-1.01726437e+00 1.43120453e-01 -1.78141505e-01 -5.34772873e-01
-1.36036515e-01 1.47108972e-01 -5.41573726e-02 4.02187437e-01
5.20210207e-01 4.44896996e-01 -3.79390001e-01 -3.09338868e-01
3.37362319e-01 2.60863125e-01 7.85871029e-01 -4.88050133e-01
3.26981753e-01 7.07935929e-01 -2.57233232e-01 -2.50680614e-02
-1.01518846e+00 -7.79336512e-01 -6.22011840e-01 -5.72582185e-01
7.88747728e-01 -1.10153067e+00 -5.44649661e-01 4.92486060e-01
-1.24299383e+00 -4.06266004e-01 -3.80254179e-01 6.97772026e-01
-8.70364368e-01 6.29801512e-01 -5.14201760e-01 -8.03801358e-01
-4.88721460e-01 -8.40216398e-01 1.44571507e+00 3.07999849e-01
-2.60527819e-01 -3.84899288e-01 2.52375364e-01 3.54632974e-01
8.83401483e-02 1.65190786e-01 4.80288379e-02 -3.46542895e-01
-8.99323583e-01 -3.56925040e-01 -3.15124243e-01 6.06629737e-02
-3.86586227e-02 -3.12228799e-02 -6.14130080e-01 -1.44597113e-01
-3.52118671e-01 -5.24200737e-01 1.28643560e+00 6.10500395e-01
1.17040741e+00 -1.62705407e-01 -6.56595409e-01 1.44923151e-01
8.21162522e-01 6.63602874e-02 9.50343609e-01 2.06958532e-01
2.22518712e-01 2.31848165e-01 1.37571514e+00 8.33053291e-01
4.01255071e-01 1.08968365e+00 4.46486056e-01 1.70954108e-01
-2.24562585e-01 -6.57229245e-01 6.43240869e-01 3.58184218e-01
-2.88989305e-01 -1.54253542e-01 -3.22548985e-01 7.49340653e-01
-2.41747618e+00 -1.47085977e+00 1.15822323e-01 2.19591689e+00
8.50193143e-01 3.93064946e-01 5.26811779e-01 -1.19881399e-01
9.36737716e-01 3.49609613e-01 -5.59459925e-01 3.83533180e-01
1.42896637e-01 -5.73542789e-02 -3.35398018e-02 -7.91502893e-02
-1.47485030e+00 1.00262117e+00 6.22748089e+00 1.29312110e+00
-7.06282079e-01 2.36491159e-01 6.72549963e-01 -2.84019530e-01
6.21586991e-03 -6.49464354e-02 -9.86938775e-01 6.43722832e-01
4.60545272e-01 1.66386724e-01 -3.26505184e-01 1.02375817e+00
5.03991544e-01 -3.31811756e-01 -1.04691672e+00 9.21329081e-01
-6.94320947e-02 -1.47993910e+00 1.01015627e-01 -2.05723315e-01
8.37217212e-01 4.47819717e-02 -2.59894490e-01 6.86491609e-01
2.25653574e-01 -6.21110141e-01 1.00786877e+00 7.54984617e-01
5.04270554e-01 -4.69168544e-01 5.58809578e-01 3.16143215e-01
-1.77863681e+00 -1.44709244e-01 -5.37310600e-01 5.88244237e-02
6.60308897e-01 4.35113162e-01 -7.21874833e-01 3.42210829e-01
5.91815889e-01 1.13144255e+00 -2.96952635e-01 1.65126669e+00
-5.37744164e-01 5.25095999e-01 -2.58322090e-01 -1.66092291e-01
5.22609532e-01 1.23933345e-01 3.49229157e-01 1.09828281e+00
6.07321024e-01 3.76346409e-01 3.49694967e-01 8.79490793e-01
-1.92352775e-02 -1.30104989e-01 -1.20766506e-01 1.06335491e-01
5.08624315e-01 1.38487148e+00 -8.95168602e-01 -7.02859879e-01
-2.11555809e-01 1.01633108e+00 4.54259783e-01 1.04814351e-01
-1.26228833e+00 7.22449571e-02 4.89324450e-01 8.01511779e-02
5.96310556e-01 -1.34299491e-02 3.03098619e-01 -1.09438038e+00
4.21976000e-01 -4.86792654e-01 7.68531621e-01 -1.07567346e+00
-1.15321183e+00 2.75580704e-01 3.32240388e-02 -1.91670644e+00
-1.20830707e-01 -1.99388325e-01 -1.08336222e+00 4.11353797e-01
-1.21506274e+00 -1.13893008e+00 -3.98831338e-01 5.25262296e-01
9.29172873e-01 4.84691262e-02 2.86705166e-01 3.46673608e-01
-6.59710407e-01 4.43091005e-01 -5.25488079e-01 1.62866771e-01
6.87182426e-01 -8.64236772e-01 6.45459354e-01 9.04447734e-01
2.26169720e-01 2.47186869e-01 1.00817822e-01 -9.45966601e-01
-6.08807206e-01 -1.43082559e+00 1.03139114e+00 -4.20411646e-01
4.89306748e-01 -5.77341346e-03 -6.74772024e-01 4.33262914e-01
-2.19994858e-01 3.90757859e-01 1.55746877e-01 -3.60325694e-01
-2.08643556e-01 -2.29316745e-02 -8.56812894e-01 8.28867197e-01
1.62355566e+00 -1.20552100e-01 -6.43463731e-01 5.60043693e-01
6.91621780e-01 -3.47011477e-01 -5.88886380e-01 6.85818434e-01
6.43520415e-01 -1.25745440e+00 1.04070318e+00 -6.09515429e-01
6.12085164e-01 -5.91732502e-01 1.40016332e-01 -6.03506625e-01
-4.59410310e-01 -7.81406760e-01 -4.92768615e-01 1.06665266e+00
3.87547314e-01 -3.24450210e-02 1.01043594e+00 4.99167860e-01
-4.26466793e-01 -1.07629204e+00 -1.22799993e+00 -7.39171445e-01
-5.36689162e-01 -6.98394895e-01 5.48580289e-01 3.61374676e-01
1.24447837e-01 -6.40435442e-02 -4.80393678e-01 -2.61989444e-01
3.06050122e-01 5.95078468e-02 8.13997805e-01 -7.24822521e-01
-2.13466108e-01 -6.11397803e-01 -6.53370380e-01 -1.70459819e+00
-1.36615232e-01 -4.55481678e-01 5.01961768e-01 -1.81932795e+00
5.18359244e-01 -3.53752553e-01 -5.10867894e-01 6.57047987e-01
-4.73464429e-01 4.41870213e-01 5.80299944e-02 4.08135533e-01
-1.54589450e+00 1.06344247e+00 1.53342724e+00 -1.14258185e-01
-8.81110877e-02 2.67590880e-01 -2.09052831e-01 7.97043204e-01
4.47058797e-01 -4.98456240e-01 -3.71931434e-01 -4.96967994e-02
-2.35802960e-02 2.72300482e-01 4.32016402e-01 -1.34128284e+00
2.47180268e-01 -4.52020288e-01 5.29634237e-01 -1.17516279e+00
4.27478284e-01 -3.27544928e-01 -4.55502868e-02 3.90285313e-01
-3.97532970e-01 -1.05727375e-01 -1.30128935e-01 8.78485441e-01
-3.46995741e-01 -6.05260879e-02 5.12969315e-01 -2.24832028e-01
-1.14804423e+00 8.57333899e-01 -2.01664180e-01 8.50275233e-02
1.30445540e+00 -4.75612849e-01 -2.27870643e-01 -2.92292565e-01
-6.88178718e-01 4.15017784e-01 5.86586185e-02 4.35939640e-01
9.23505068e-01 -2.03368330e+00 -7.63648927e-01 -1.57128185e-01
2.94535279e-01 1.08102337e-01 4.89848346e-01 1.16881728e+00
-1.77523792e-01 3.51776898e-01 -1.38536215e-01 -6.72194898e-01
-1.07052708e+00 4.66553271e-01 2.44525671e-01 -4.34224099e-01
-7.62460887e-01 1.19170856e+00 3.46014917e-01 -4.54131328e-03
4.18472230e-01 -5.54802835e-01 -2.04674527e-01 -9.95999277e-02
8.40908229e-01 4.01648015e-01 -4.48670179e-01 -5.20871401e-01
-3.86961132e-01 4.37923104e-01 3.07160392e-02 -3.10661137e-01
1.02051508e+00 1.84359789e-01 2.12609023e-01 -6.05954193e-02
6.49705946e-01 -5.90292871e-01 -1.83938801e+00 -3.74998659e-01
-3.03954426e-02 -6.39374077e-01 -3.00411820e-01 -5.12059748e-01
-1.02912188e+00 6.58956528e-01 3.51828933e-01 -2.59782910e-01
1.08535814e+00 1.77660674e-01 1.06824625e+00 2.36907691e-01
5.72949111e-01 -1.47674143e+00 8.81142557e-01 1.55645952e-01
9.85109985e-01 -1.25858378e+00 7.67652690e-02 -2.45667860e-01
-8.36183310e-01 8.65822136e-01 1.06780732e+00 2.77681034e-02
2.98366785e-01 -3.47729713e-01 -3.98956329e-01 -8.44009221e-02
-1.01153874e+00 -2.95621276e-01 5.88317037e-01 4.97255772e-01
2.12932229e-01 -1.42010422e-02 -5.99510312e-01 6.54132247e-01
5.31911671e-01 3.82325560e-01 8.40855092e-02 1.03983748e+00
-7.83905685e-01 -8.61680686e-01 -1.09974973e-01 3.92411768e-01
-1.45721912e-01 5.74273132e-02 -2.18311056e-01 4.94682640e-01
1.77025914e-01 8.00055444e-01 8.24203044e-02 -4.08097118e-01
2.20411554e-01 -1.72723383e-01 3.56089085e-01 -5.01473010e-01
-2.09548414e-01 3.08577329e-01 1.64346606e-01 -1.19595850e+00
-6.48745358e-01 -9.80025291e-01 -1.38701057e+00 2.04681978e-01
-4.26649392e-01 -7.08716642e-03 6.65185750e-02 9.86209810e-01
5.23314714e-01 3.23084414e-01 4.45642531e-01 -1.09652507e+00
-6.67111695e-01 -1.28765059e+00 -3.71083140e-01 4.44593102e-01
-1.95743829e-01 -8.48463416e-01 3.17721888e-02 2.19605956e-02] | [8.391603469848633, 0.4504220187664032] |
fef921c3-2b92-4288-9dfc-afac9a9c4084 | conversational-semantic-parsing-using-dynamic | 2305.06164 | null | https://arxiv.org/abs/2305.06164v1 | https://arxiv.org/pdf/2305.06164v1.pdf | Conversational Semantic Parsing using Dynamic Context Graphs | In this paper we consider the task of conversational semantic parsing over general purpose knowledge graphs (KGs) with millions of entities, and thousands of relation-types. We are interested in developing models capable of interactively mapping user utterances into executable logical forms (e.g., SPARQL) in the context of the conversational history. Our key idea is to represent information about an utterance and its context via a subgraph which is created dynamically, i.e., the number of nodes varies per utterance. Moreover, rather than treating the subgraph as a sequence we exploit its underlying structure, and thus encode it using a graph neural network which further allows us to represent a large number of (unseen) nodes. Experimental results show that modeling context dynamically is superior to static approaches, delivering performance improvements across the board (i.e., for simple and complex questions). Our results further confirm that modeling the structure of context is better at processing discourse information, (i.e., at handling ellipsis and resolving coreference) and longer interactions. | ['Mirella Lapata', 'Parag Jain'] | 2023-05-04 | null | null | null | null | ['semantic-parsing'] | ['natural-language-processing'] | [ 1.30878448e-01 9.93732989e-01 -3.92058305e-02 -4.03698474e-01
-4.88339067e-01 -9.69463170e-01 5.25342107e-01 6.90527856e-01
-2.29430497e-01 6.56805098e-01 4.99983281e-01 -6.26954138e-01
2.78926194e-01 -1.18993771e+00 -7.76346862e-01 5.96134923e-02
-2.56149322e-01 8.29740584e-01 5.48456252e-01 -6.95502281e-01
-9.03995708e-02 6.76115453e-02 -1.31451535e+00 6.13193989e-01
7.29233563e-01 4.91898000e-01 2.39870980e-01 8.48622918e-01
-6.85655951e-01 1.00770497e+00 -9.57382262e-01 -7.48379648e-01
-3.89231235e-01 -3.94488454e-01 -1.73308468e+00 -6.10806085e-02
3.25549603e-01 -2.12337196e-01 -9.57050920e-02 9.51800048e-01
1.38237640e-01 3.10616821e-01 1.19243205e-01 -1.13975203e+00
-3.39726478e-01 1.14991736e+00 1.65691912e-01 2.88071245e-01
9.46215987e-01 -5.03399633e-02 1.36143792e+00 -2.53705919e-01
1.04169118e+00 1.58267772e+00 4.85179067e-01 7.04701483e-01
-1.11945808e+00 -1.49123725e-02 4.11984146e-01 2.74005115e-01
-9.24166262e-01 -6.03811681e-01 4.55686539e-01 -3.07612568e-01
1.46111071e+00 5.32821000e-01 3.61708373e-01 8.58105719e-01
-4.16219905e-02 6.67661130e-01 5.23158371e-01 -6.94573700e-01
2.31555834e-01 1.50074303e-01 7.01648116e-01 9.22336638e-01
9.68056992e-02 -5.84209085e-01 -6.39659166e-01 -3.89188081e-01
3.74589026e-01 -6.93671167e-01 -3.42766643e-01 -2.87989259e-01
-8.07822645e-01 8.81088495e-01 2.42944077e-01 3.57231528e-01
-2.23896593e-01 7.80669004e-02 5.55434942e-01 5.68210363e-01
2.08628967e-01 7.01499104e-01 -6.12639308e-01 -2.16116324e-01
-4.40479070e-02 4.16599900e-01 1.69167423e+00 1.18444633e+00
8.48598421e-01 -4.26379412e-01 -5.48016168e-02 8.37512374e-01
3.20634656e-02 2.46620178e-03 3.51390153e-01 -1.03555095e+00
8.51087391e-01 8.41849446e-01 3.02539021e-01 -9.31480706e-01
-7.31769204e-01 -2.44395100e-02 -1.78068921e-01 -4.21334982e-01
6.85855508e-01 -4.68946129e-01 -4.26186860e-01 2.22053313e+00
6.15994036e-01 5.79281151e-02 6.53828740e-01 5.65098286e-01
1.19525170e+00 5.77366531e-01 4.85825598e-01 -2.76868761e-01
1.91345727e+00 -6.49834454e-01 -8.23992789e-01 -6.91685200e-01
1.07653022e+00 -2.72056818e-01 1.00854313e+00 -1.46427676e-01
-1.15720963e+00 -1.67815790e-01 -7.49455988e-01 -3.07179272e-01
-5.45013666e-01 -6.20846450e-01 8.97059858e-01 3.68384421e-01
-1.27746379e+00 4.47620362e-01 -7.95638800e-01 -6.77087784e-01
-2.83346057e-01 1.73462823e-01 -2.82184333e-01 3.78829204e-02
-1.72624218e+00 9.61346686e-01 8.12418461e-01 -2.13844329e-01
-1.71624050e-01 -3.89742494e-01 -1.35838079e+00 3.19730371e-01
8.49271417e-01 -8.85734320e-01 1.68657875e+00 -7.99310803e-01
-1.33470452e+00 8.10692012e-01 -3.59325856e-01 -5.88240623e-01
-1.94116998e-02 4.60926630e-02 -3.38379413e-01 2.29208291e-01
-6.53537735e-03 5.78641653e-01 1.54066607e-01 -1.07851326e+00
-6.18474662e-01 -5.56964278e-01 1.05629420e+00 5.15366971e-01
-1.73174977e-01 2.93544084e-01 -5.94084382e-01 -7.49270320e-02
2.63513982e-01 -9.18433249e-01 -4.85073365e-02 -6.04435861e-01
-5.02189159e-01 -6.35873973e-01 6.18625283e-01 -1.00724196e+00
1.52197778e+00 -1.88242352e+00 4.01752025e-01 6.01597130e-02
2.91426212e-01 9.14575830e-02 5.04366495e-02 7.00090289e-01
2.04372287e-01 3.65500748e-01 -1.04669027e-01 -5.17122969e-02
8.13448429e-02 5.75457573e-01 9.68868062e-02 -3.46128374e-01
1.85626030e-01 1.20447266e+00 -1.00644696e+00 -4.90386575e-01
-2.35847801e-01 6.29598200e-02 -7.71047175e-01 3.25043797e-01
-9.76421535e-01 1.91970721e-01 -8.01643789e-01 2.57842839e-01
1.75616711e-01 -5.54372847e-01 9.77488637e-01 -7.51163140e-02
2.16861963e-01 8.70316625e-01 -1.15837336e+00 1.71842909e+00
-7.19404936e-01 4.14208353e-01 3.63189638e-01 -7.41866827e-01
4.87391680e-01 4.59185541e-01 -2.52920389e-01 -3.30179632e-01
-1.06038243e-01 -1.63509816e-01 -8.23273063e-02 -7.08689928e-01
6.69944644e-01 1.95841920e-02 -4.57254648e-01 6.30644619e-01
6.68354630e-02 6.58849403e-02 3.22597474e-01 7.41491199e-01
1.33138871e+00 -2.81967998e-01 4.25341219e-01 -9.64749306e-02
4.69335467e-01 3.09590697e-01 3.53577584e-01 5.92392206e-01
1.86922997e-01 -2.46515796e-01 1.14801002e+00 -3.13024789e-01
-5.73051751e-01 -6.89786255e-01 1.71571612e-01 1.38831937e+00
2.16358751e-01 -6.51487947e-01 -1.05803144e+00 -6.85181797e-01
2.02172007e-02 9.93061602e-01 -3.07056546e-01 -3.08118071e-02
-1.01779544e+00 -4.16800708e-01 4.06475008e-01 4.76907581e-01
4.12669450e-01 -1.36766922e+00 -7.33458877e-01 5.62455535e-01
-7.93803513e-01 -1.43406308e+00 -8.32488090e-02 5.89691587e-02
-6.60560310e-01 -1.26989651e+00 2.99743623e-01 -8.36714625e-01
3.59206438e-01 -4.18386869e-02 1.60383415e+00 4.34776485e-01
4.21922319e-02 6.89076185e-01 -4.93982643e-01 -2.20013350e-01
-8.62625182e-01 3.03624332e-01 -5.34832835e-01 -3.47877711e-01
5.11487842e-01 -5.27300715e-01 -2.03657627e-01 -2.44212942e-03
-8.61962855e-01 1.35378331e-01 -1.55121740e-02 7.03710139e-01
1.15713309e-02 1.07040212e-01 6.05599701e-01 -1.56132603e+00
9.81418490e-01 -6.22594833e-01 -5.02151668e-01 7.39024997e-01
-1.95883587e-01 3.05581957e-01 4.86376673e-01 -7.09691346e-02
-1.31203496e+00 -2.97316819e-01 -1.10760838e-01 3.47024113e-01
-2.56259173e-01 8.10197413e-01 -3.35523725e-01 2.49015242e-01
6.23569608e-01 -2.41098106e-01 4.44079787e-02 -4.22812015e-01
6.46463335e-01 6.86008930e-01 5.87996483e-01 -1.00458694e+00
2.90173858e-01 -1.19245946e-01 -3.41484964e-01 -7.57767498e-01
-7.66391873e-01 -5.24283171e-01 -5.08685410e-01 -3.21790949e-02
1.02000189e+00 -7.22780108e-01 -1.06911445e+00 2.83914536e-01
-1.48060012e+00 -6.02246940e-01 -1.16259931e-02 -3.92321050e-02
-4.40622687e-01 4.49092299e-01 -1.05552793e+00 -7.05579877e-01
-2.84226120e-01 -8.28467488e-01 9.71481204e-01 1.83691666e-01
-6.44849241e-01 -1.27987373e+00 -1.85952723e-01 6.03718817e-01
2.49701217e-01 2.88083971e-01 1.48567891e+00 -9.92181122e-01
-6.79946959e-01 2.74046212e-02 -1.74543746e-02 -1.58157885e-01
3.26013193e-02 -4.03598338e-01 -8.05061579e-01 -1.35156929e-01
-1.24487579e-01 -3.48702788e-01 9.42338407e-02 -2.56312519e-01
8.70437324e-01 -8.00323188e-01 -4.28752333e-01 6.40725270e-02
1.22034812e+00 2.89676100e-01 5.36225677e-01 2.15764925e-01
4.68375444e-01 1.02623594e+00 3.62656891e-01 1.11927085e-01
1.00879872e+00 8.39115620e-01 2.38142595e-01 3.69212955e-01
4.03280556e-02 -3.74496043e-01 -2.48978455e-02 4.93852615e-01
2.95924157e-01 -5.13947487e-01 -1.12186611e+00 3.92402440e-01
-1.95008981e+00 -8.71046484e-01 -1.71790253e-02 1.87104988e+00
1.15656018e+00 1.32992610e-01 -9.83584579e-03 -3.83674413e-01
7.14604616e-01 2.26320416e-01 -5.54328263e-01 -7.02056885e-01
1.30392745e-01 2.18624592e-01 8.85888562e-02 1.13258338e+00
-7.31112242e-01 1.26934934e+00 5.91132641e+00 1.50780054e-02
-7.25605488e-01 -3.02114300e-02 3.91633451e-01 3.89837980e-01
-4.41498488e-01 2.41596717e-02 -7.67524421e-01 1.81908011e-01
1.25614440e+00 -2.89638609e-01 7.45051265e-01 6.22612834e-01
-3.06633145e-01 -1.12997048e-01 -1.27365673e+00 2.56834120e-01
-8.98320377e-02 -1.39235806e+00 6.71635121e-02 -3.83680880e-01
1.21656805e-01 -2.43861884e-01 -6.95107460e-01 6.06910586e-01
7.04540968e-01 -6.73072338e-01 5.03798306e-01 1.41389057e-01
4.65404451e-01 -3.47278804e-01 7.06127882e-01 6.38913274e-01
-1.05917728e+00 -6.31361157e-02 -1.30717680e-01 1.19009018e-02
1.44445375e-01 -1.31096631e-01 -1.05351317e+00 8.60715330e-01
5.30548871e-01 1.66921005e-01 -4.10369217e-01 2.89509028e-01
-3.34280163e-01 3.51181388e-01 -3.69533658e-01 -2.39078894e-01
6.48076832e-02 3.85729335e-02 5.31488776e-01 1.36126339e+00
-1.80115268e-01 8.49552333e-01 3.25290769e-01 5.59706032e-01
-1.82942644e-01 1.19640782e-01 -5.09629607e-01 -3.90073806e-02
8.19922805e-01 1.03129125e+00 -4.02236372e-01 -6.94639981e-01
-5.36238551e-01 7.53343463e-01 8.40717137e-01 5.44932842e-01
-1.74217075e-01 -2.27201685e-01 6.41498864e-01 1.15911365e-01
1.88272104e-01 -8.78255218e-02 2.35035002e-01 -1.13785422e+00
2.09817514e-01 -1.03615570e+00 9.66307044e-01 -9.89533246e-01
-9.30341959e-01 6.85107291e-01 3.38442296e-01 -1.87985599e-01
-7.47441113e-01 -4.84585643e-01 -7.69088447e-01 8.63360465e-01
-1.29559338e+00 -9.23135519e-01 -2.86682963e-01 4.42934453e-01
4.50392753e-01 4.20672864e-01 1.18856764e+00 -9.25443918e-02
-5.26009619e-01 4.11814511e-01 -5.35577834e-01 3.09813350e-01
1.18200541e-01 -1.43354106e+00 6.82694495e-01 5.67114890e-01
5.79142049e-02 9.34463859e-01 8.51107001e-01 -6.50190890e-01
-1.67114031e+00 -7.69176304e-01 1.35297143e+00 -6.29689276e-01
9.29604769e-01 -5.36640346e-01 -1.37838840e+00 1.17092776e+00
3.35524440e-01 -4.39329326e-01 5.59604049e-01 6.34217024e-01
-4.57221389e-01 2.38630801e-01 -1.19903266e+00 5.37109852e-01
1.30834055e+00 -6.96386874e-01 -1.00560462e+00 4.73328531e-01
1.25549161e+00 -9.88508463e-01 -9.08207059e-01 2.37691551e-01
3.11706979e-02 -6.18830264e-01 8.42029333e-01 -1.19949508e+00
1.17494345e-01 7.29560703e-02 -7.98858702e-02 -1.30495489e+00
-4.07274626e-02 -7.69782782e-01 -4.84132141e-01 1.17071128e+00
7.55445480e-01 -7.87448525e-01 8.33600819e-01 1.11584580e+00
-2.55586207e-01 -5.39159417e-01 -8.47136438e-01 -3.64984065e-01
-2.43248671e-01 -2.12869927e-01 9.05495524e-01 1.10837829e+00
7.84288168e-01 8.98125291e-01 2.21421510e-01 5.10016263e-01
1.39261529e-01 3.03933948e-01 5.76599419e-01 -1.25388575e+00
-5.23828864e-01 -9.58572701e-02 -2.06296876e-01 -1.00863802e+00
5.36371887e-01 -8.73137474e-01 2.28422019e-03 -1.81058705e+00
-1.66033700e-01 -5.11235118e-01 9.36209559e-02 6.55488074e-01
-3.95367950e-01 -6.02032602e-01 3.42324749e-02 -5.07998951e-02
-6.80598378e-01 -1.20524596e-02 9.57896411e-01 -5.44808879e-02
-2.81574488e-01 -5.37449531e-02 -8.07754815e-01 6.76255465e-01
8.59635592e-01 -1.55532867e-01 -6.46399617e-01 -5.73834360e-01
2.78776318e-01 7.58236885e-01 3.66366178e-01 -4.87091631e-01
4.53238070e-01 -1.81667060e-01 -3.74003798e-01 -4.26673591e-02
4.70632493e-01 -5.35830855e-01 1.31302178e-01 1.62665159e-01
-5.79913437e-01 2.69208848e-01 3.35274935e-01 5.56822956e-01
-3.38828295e-01 -4.18897271e-01 1.37836739e-01 -4.59183991e-01
-9.68764782e-01 -9.26577225e-02 -1.24099731e-01 5.13691127e-01
7.08314002e-01 7.38310441e-02 -6.67332351e-01 -6.37410462e-01
-1.18713737e+00 6.31095231e-01 3.30284864e-01 5.50298333e-01
1.48379624e-01 -8.08747470e-01 -3.60279530e-01 -8.55860934e-02
2.79503286e-01 2.70963311e-01 1.20692097e-01 6.87729865e-02
-4.87326920e-01 4.97359216e-01 1.16581924e-01 -3.36363465e-01
-1.61238348e+00 3.44190717e-01 4.06762570e-01 -5.20774186e-01
-5.69524765e-01 8.96900535e-01 3.48407924e-02 -6.12107813e-01
4.50123757e-01 -3.46194774e-01 -6.06135666e-01 1.42227948e-01
5.34738004e-01 -1.26960218e-01 1.77076310e-01 -3.72939110e-01
-3.04519862e-01 1.66324243e-01 -8.35121050e-02 6.23431690e-02
1.22973537e+00 -1.67362407e-01 -5.30110478e-01 2.72269279e-01
1.02796531e+00 -8.96795765e-02 -6.25146687e-01 -6.44715071e-01
5.01325846e-01 -1.08953662e-01 -5.11289239e-01 -8.40534031e-01
-5.56317210e-01 4.39120114e-01 -2.29505077e-01 8.18431437e-01
6.87385857e-01 5.87202787e-01 7.05582619e-01 9.39002097e-01
7.07424521e-01 -7.69149840e-01 -1.07566923e-01 6.71825290e-01
7.67767966e-01 -1.03721726e+00 -2.86239237e-01 -9.69546318e-01
-6.19030297e-01 1.07789040e+00 6.75966799e-01 5.94507396e-01
2.12972298e-01 2.21248031e-01 9.77025479e-02 -7.33254313e-01
-1.07399189e+00 -7.90420920e-02 -1.14839785e-01 4.82095182e-01
4.25140381e-01 2.51443267e-01 -1.94207355e-01 4.77640688e-01
-4.52134609e-01 -3.80610734e-01 5.50508320e-01 1.07136524e+00
-6.80858314e-01 -1.30777276e+00 -3.58288400e-02 3.54393154e-01
-4.16941613e-01 -2.15302810e-01 -6.30503833e-01 9.64933217e-01
-2.58759141e-01 1.24077892e+00 9.88412276e-02 -1.19492039e-01
6.44303739e-01 6.43748522e-01 3.13566387e-01 -1.19276261e+00
-8.70711923e-01 -6.05655491e-01 1.10978186e+00 -5.79634666e-01
-1.30274981e-01 -5.27919292e-01 -1.65413725e+00 -3.40609342e-01
-2.67065257e-01 6.49666369e-01 5.76786399e-01 8.09947133e-01
3.24142218e-01 4.01751012e-01 2.38276169e-01 -1.43834483e-02
-6.21568680e-01 -7.31739581e-01 -7.25479946e-02 7.29407251e-01
1.24565586e-01 -3.65905046e-01 -2.36412153e-01 -8.42942763e-03] | [12.254355430603027, 8.104039192199707] |
6263e006-7437-41a4-8bfb-ac8e2e007e3a | gtr-ctrl-instrument-and-genre-conditioning | 2302.05393 | null | https://arxiv.org/abs/2302.05393v1 | https://arxiv.org/pdf/2302.05393v1.pdf | GTR-CTRL: Instrument and Genre Conditioning for Guitar-Focused Music Generation with Transformers | Recently, symbolic music generation with deep learning techniques has witnessed steady improvements. Most works on this topic focus on MIDI representations, but less attention has been paid to symbolic music generation using guitar tablatures (tabs) which can be used to encode multiple instruments. Tabs include information on expressive techniques and fingerings for fretted string instruments in addition to rhythm and pitch. In this work, we use the DadaGP dataset for guitar tab music generation, a corpus of over 26k songs in GuitarPro and token formats. We introduce methods to condition a Transformer-XL deep learning model to generate guitar tabs (GTR-CTRL) based on desired instrumentation (inst-CTRL) and genre (genre-CTRL). Special control tokens are appended at the beginning of each song in the training corpus. We assess the performance of the model with and without conditioning. We propose instrument presence metrics to assess the inst-CTRL model's response to a given instrumentation prompt. We trained a BERT model for downstream genre classification and used it to assess the results obtained with the genre-CTRL model. Statistical analyses evidence significant differences between the conditioned and unconditioned models. Overall, results indicate that the GTR-CTRL methods provide more flexibility and control for guitar-focused symbolic music generation than an unconditioned model. | ['Mathieu Barthet', 'Zack Zukowski', 'CJ Carr', 'Yu-Hua Chen', 'Adarsh Kumar', 'Pedro Sarmento'] | 2023-02-10 | null | null | null | null | ['music-generation', 'genre-classification', 'music-generation'] | ['audio', 'computer-vision', 'music'] | [ 2.36733034e-01 -3.87976259e-01 -9.29232612e-02 -1.11889370e-01
-9.17236388e-01 -1.05894637e+00 5.60825944e-01 -1.07282609e-01
-4.01031692e-03 5.42849004e-01 1.83402985e-01 -1.73519805e-01
-4.23382491e-01 -7.80191064e-01 -7.08854377e-01 -6.30716264e-01
2.00052876e-02 5.20185709e-01 -3.99233758e-01 -4.39079314e-01
3.02738070e-01 3.45905811e-01 -1.70688093e+00 7.24133790e-01
4.04668510e-01 1.13358939e+00 5.30943125e-02 9.67291772e-01
1.53995290e-01 9.37947929e-01 -1.30560040e+00 -6.79704696e-02
2.91812330e-01 -7.71876276e-01 -6.83616579e-01 -8.22162688e-01
3.94425720e-01 -1.88546747e-01 5.25558880e-03 5.30844212e-01
9.02007043e-01 3.17294478e-01 8.34242582e-01 -1.23956740e+00
-2.92851865e-01 1.51971138e+00 -1.14700645e-01 -1.49795219e-01
6.64057493e-01 2.36173213e-01 1.24272692e+00 -3.13959748e-01
4.88223314e-01 9.51565921e-01 6.84496999e-01 4.48948503e-01
-1.57446516e+00 -1.12980819e+00 -4.24063087e-01 -4.69587520e-02
-1.22086060e+00 -2.40043193e-01 9.04657364e-01 -5.94383240e-01
8.14888477e-01 6.28702044e-01 9.18825150e-01 1.53238606e+00
1.57624528e-01 7.64029264e-01 1.02005243e+00 -5.03677368e-01
2.39870101e-01 -2.66054571e-01 -4.55169469e-01 1.03932880e-01
-4.94978905e-01 6.95517182e-01 -1.05165207e+00 5.69641367e-02
9.13107634e-01 -7.06936836e-01 -3.31353443e-03 3.85092646e-01
-1.25584185e+00 7.49673903e-01 1.59646794e-01 5.94976425e-01
-2.61452109e-01 4.52846348e-01 6.96947753e-01 5.35413742e-01
8.06871522e-03 1.41656423e+00 -5.04546225e-01 -6.12068832e-01
-1.26844096e+00 8.76920581e-01 7.64994323e-01 8.37873280e-01
2.61598855e-01 7.02805698e-01 -7.57759452e-01 9.74273682e-01
-2.55559623e-01 5.80692708e-01 7.41009831e-01 -1.08674061e+00
2.14001670e-01 1.26471341e-01 -1.50258645e-01 -7.53056407e-01
-2.75817305e-01 -5.85923851e-01 -6.09987557e-01 2.62618750e-01
4.51338887e-01 -1.47897243e-01 -6.57600105e-01 1.97616756e+00
-2.34457210e-01 1.78932875e-01 -1.61067545e-01 7.89920211e-01
8.61559868e-01 6.70974791e-01 -3.39481235e-02 -7.30203539e-02
9.22825694e-01 -4.88426864e-01 -5.93363881e-01 3.03405404e-01
5.77204227e-01 -9.78210211e-01 1.47826898e+00 9.75912809e-01
-1.03918839e+00 -9.35786843e-01 -1.00019169e+00 3.12507711e-02
-2.95031548e-01 3.34493339e-01 6.70338392e-01 6.78209603e-01
-6.40420794e-01 1.12681651e+00 -4.12644774e-01 3.12473625e-01
-1.55020252e-01 3.82444233e-01 9.52851549e-02 7.20833480e-01
-1.58288717e+00 4.43289727e-01 5.07876039e-01 -6.66489154e-02
-1.09437513e+00 -7.96000302e-01 -6.25413537e-01 1.65439427e-01
8.40895344e-03 -3.81644577e-01 1.69730914e+00 -7.98596799e-01
-2.05389619e+00 6.23028755e-01 5.90675771e-01 -4.11729276e-01
2.89322019e-01 -2.52874017e-01 -4.09975857e-01 -3.53861004e-02
1.38824433e-01 5.44265389e-01 8.83097649e-01 -9.43134904e-01
-3.24600726e-01 1.45423397e-01 -9.81224179e-02 4.06205468e-02
-6.01136163e-02 -1.17567843e-02 8.58535469e-02 -1.26474798e+00
-1.88457564e-01 -1.13199162e+00 3.80768418e-01 -6.18450642e-01
-8.84134173e-01 -2.15792477e-01 4.21636403e-01 -3.98933470e-01
1.45657647e+00 -2.09347510e+00 2.37571329e-01 2.16280431e-01
-3.73069704e-01 9.17785615e-02 -2.14561418e-01 5.76586425e-01
-4.02503461e-01 1.55229673e-01 1.85734347e-01 -1.99977428e-01
4.19087410e-01 1.06091477e-01 -8.72256935e-01 -2.03574389e-01
-2.10338850e-02 8.28906178e-01 -6.58496082e-01 -3.13021660e-01
1.50009111e-01 2.99413949e-01 -8.56964767e-01 3.32073957e-01
-6.39226377e-01 6.96499169e-01 -2.96228170e-01 6.77298844e-01
-4.48326878e-02 4.60173666e-01 -1.22597873e-01 -3.46630096e-01
-3.32499623e-01 6.83398068e-01 -9.52763438e-01 1.93920100e+00
-6.14528298e-01 4.34497982e-01 -3.78312051e-01 -4.49247211e-01
1.15212047e+00 5.45820594e-01 3.19310844e-01 -6.37942433e-01
5.70720553e-01 3.22515965e-01 3.18388164e-01 -2.24246189e-01
7.21089542e-01 -5.53267777e-01 -6.99126303e-01 4.25381809e-01
1.73784539e-01 -7.52163410e-01 3.22899729e-01 -1.58142120e-01
8.55199277e-01 7.70554364e-01 -2.38017645e-02 9.57022831e-02
1.08435780e-01 -2.61802405e-01 2.50260115e-01 9.86592829e-01
4.52244073e-01 6.58398986e-01 5.60561478e-01 6.87725395e-02
-7.68038034e-01 -1.19537044e+00 1.75096855e-01 1.56500351e+00
-4.43998516e-01 -5.94323218e-01 -6.57027543e-01 -4.13070549e-04
-6.93262741e-02 1.29720795e+00 -7.49718964e-01 -5.07499635e-01
-6.38105810e-01 -1.13572367e-01 1.41945589e+00 6.53595388e-01
3.75240184e-02 -1.73877728e+00 -6.76372588e-01 2.01999381e-01
-3.48360032e-01 -4.62511450e-01 -5.83933890e-01 5.92704058e-01
-6.87800229e-01 -6.48140907e-01 -4.83091712e-01 -5.09092331e-01
-3.55636418e-01 -7.57634580e-01 9.54739332e-01 -4.40086246e-01
-3.64943177e-01 8.72931406e-02 -5.27335048e-01 -6.18643522e-01
-6.53676152e-01 3.65547091e-01 2.12907031e-01 -1.54368654e-01
-3.03102434e-01 -8.16529512e-01 -2.53427535e-01 8.51951316e-02
-8.70393634e-01 1.86289057e-01 3.50071073e-01 7.65697658e-01
6.78547144e-01 -7.96578452e-02 6.14627659e-01 -4.90970045e-01
9.59276497e-01 -7.92977959e-02 -3.21676105e-01 -1.39318556e-01
-4.07658905e-01 1.95343211e-01 8.66813898e-01 -9.75697398e-01
-7.45079458e-01 -1.78395152e-01 -3.19736302e-01 -8.83320153e-01
-8.40136185e-02 8.20869863e-01 -1.04508109e-01 3.98556322e-01
8.97903264e-01 1.68710962e-01 -3.83337200e-01 -5.11456847e-01
4.14036751e-01 5.55993259e-01 8.99592221e-01 -1.32904518e+00
7.16760159e-01 -3.16983968e-01 -1.25494311e-02 -4.45172042e-01
-8.45838726e-01 1.97922006e-01 -3.88901442e-01 -5.47870696e-01
5.87477565e-01 -3.62924337e-01 -1.06800616e+00 5.58655798e-01
-8.16778421e-01 -7.16974676e-01 -7.54787147e-01 5.26058018e-01
-1.19101787e+00 -3.05458337e-01 -7.88058281e-01 -8.61473322e-01
-6.17406547e-01 -9.19471860e-01 1.17601335e+00 8.37461278e-02
-8.54985774e-01 -5.64047873e-01 2.65190005e-01 -2.16362365e-02
3.51529539e-01 6.97272480e-01 1.10189545e+00 -5.93648434e-01
-9.48644057e-02 -3.09955031e-01 5.19105911e-01 3.29212397e-01
3.56580503e-02 5.56569174e-02 -1.14179432e+00 1.69445872e-02
-1.40451789e-01 -8.35704863e-01 5.31057596e-01 3.17424387e-01
1.22597003e+00 -3.79064381e-01 3.04906100e-01 7.60042071e-01
9.10124123e-01 5.83896995e-01 6.20043099e-01 3.19056034e-01
4.69195038e-01 3.03318888e-01 7.77445674e-01 6.08521879e-01
-4.46411282e-01 9.45711434e-01 1.49086520e-01 2.25102350e-01
-2.16635957e-01 -8.41338217e-01 6.62621260e-01 6.90711200e-01
-2.78003395e-01 -3.89955491e-01 -5.49624145e-01 2.44071320e-01
-1.25059235e+00 -1.27596283e+00 3.99741140e-04 2.21437001e+00
1.34768474e+00 2.51681715e-01 2.38455266e-01 9.06571627e-01
2.83624500e-01 -2.54623089e-02 -3.88497561e-01 -8.40674460e-01
-2.57929593e-01 1.04027784e+00 -1.63988203e-01 1.63113385e-01
-7.96426713e-01 1.05971026e+00 6.51980543e+00 1.18615794e+00
-1.55653167e+00 -4.02303487e-01 1.63985506e-01 -4.79727864e-01
-2.14571521e-01 -1.07333116e-01 -4.82066721e-01 4.15530950e-01
9.45930362e-01 -1.01374201e-01 9.17692482e-01 7.48256385e-01
4.17838097e-01 1.27031192e-01 -1.48009968e+00 9.49955702e-01
-1.22165538e-01 -1.21405029e+00 3.14207166e-01 -1.56259790e-01
5.83137810e-01 -4.91560549e-01 5.84951222e-01 8.42872083e-01
-4.07813378e-02 -1.21222687e+00 1.38313830e+00 5.98465025e-01
1.34474885e+00 -8.40544641e-01 3.10673356e-01 5.43724298e-02
-1.18605256e+00 2.54669376e-02 2.47760609e-01 -3.08009028e-01
-4.22106609e-02 3.05237360e-02 -1.15598130e+00 5.38084209e-01
3.65447402e-01 3.87054384e-01 -3.46233368e-01 7.08803058e-01
-4.15680438e-01 1.23471737e+00 -2.64378846e-01 -1.71375591e-02
-6.62455931e-02 9.33301300e-02 6.23249531e-01 1.22073293e+00
5.60976326e-01 -2.61821002e-01 -2.13364158e-02 1.29902864e+00
1.87744573e-01 2.60351330e-01 -2.86996990e-01 -7.29843020e-01
5.21903455e-01 9.95663047e-01 -4.88553315e-01 -1.65957212e-01
5.56089997e-01 8.23758125e-01 -2.01793477e-01 1.86267912e-01
-9.51396644e-01 -5.32417297e-01 3.59536856e-01 9.89814252e-02
1.60261020e-01 -3.61431725e-02 -4.64392841e-01 -6.22456968e-01
-5.92172146e-01 -1.19826019e+00 3.26751888e-01 -1.34111011e+00
-1.06435513e+00 5.64737856e-01 1.45274445e-01 -1.28422654e+00
-8.93184483e-01 -5.52883148e-01 -6.42682195e-01 9.50980783e-01
-5.48302889e-01 -1.26564097e+00 5.32715470e-02 5.75839341e-01
3.35622936e-01 -3.11507404e-01 1.40160918e+00 4.48068865e-02
-2.44326413e-01 9.38357353e-01 -2.83001602e-01 2.52687037e-01
8.71534348e-01 -1.42847919e+00 -3.20984684e-02 2.18480974e-01
4.49843585e-01 8.34299266e-01 1.03976452e+00 -4.38899875e-01
-1.12056255e+00 -9.60516691e-01 6.45265698e-01 -3.12248200e-01
5.15901148e-01 -3.57275575e-01 -4.05595690e-01 5.37667513e-01
6.98974878e-02 -8.38032007e-01 9.02431667e-01 3.49706978e-01
-6.21199548e-01 -1.72399625e-01 -7.66433477e-01 6.93996072e-01
6.79319084e-01 -5.61962843e-01 -6.68781102e-01 -7.34540597e-02
6.52867734e-01 -6.60209239e-01 -9.74347651e-01 4.84957159e-01
9.71704662e-01 -8.85950923e-01 6.24994695e-01 -4.75427598e-01
7.74044394e-01 -3.08593869e-01 -2.45707303e-01 -1.43920946e+00
-2.47927845e-01 -7.98691809e-01 3.58598307e-02 1.31490886e+00
3.86723071e-01 3.51770110e-02 3.77658606e-01 -1.21673472e-01
-3.73044282e-01 -4.72418576e-01 -7.43486404e-01 -9.37752783e-01
2.18884677e-01 -8.65779459e-01 6.58739865e-01 9.56534028e-01
1.50704339e-01 5.97766042e-01 -5.63899398e-01 -3.86442751e-01
8.21439102e-02 4.50522661e-01 8.15632045e-01 -1.12196577e+00
-8.98678482e-01 -8.46984327e-01 -1.17081948e-01 -3.60424846e-01
3.27237323e-02 -1.36996841e+00 8.86644423e-02 -1.07379842e+00
-2.66067147e-01 -4.26675469e-01 -5.50319970e-01 9.30592835e-01
3.01299602e-01 4.25523698e-01 5.46103299e-01 -4.83440161e-02
2.52915807e-02 5.33349216e-01 1.04805255e+00 -2.40595281e-01
-6.19770169e-01 2.42421687e-01 -3.92454118e-01 3.67589712e-01
9.14937556e-01 -5.13960898e-01 -4.62838948e-01 -1.26720062e-02
4.46462452e-01 4.49621677e-01 3.66961032e-01 -1.32883048e+00
-1.56818554e-01 -6.72899783e-02 4.55060720e-01 -6.00013852e-01
6.16745234e-01 -9.99657586e-02 4.24854100e-01 3.20476651e-01
-9.40608621e-01 -2.22184598e-01 6.52191460e-01 -2.10923273e-02
-1.82675347e-01 -1.41878411e-01 5.41107833e-01 1.62192494e-01
6.16174936e-03 -1.14851080e-01 -5.22533178e-01 1.73389077e-01
3.14223468e-01 -1.05797075e-01 1.70457870e-01 -5.82728207e-01
-1.01589656e+00 -4.44327712e-01 -1.27259940e-02 7.11252213e-01
4.24562901e-01 -1.44718802e+00 -6.35433078e-01 2.71347523e-01
1.43702134e-01 -3.77345324e-01 2.34996080e-01 5.91024518e-01
-1.95228621e-01 2.86610931e-01 -3.34568232e-01 -3.88012022e-01
-1.15717149e+00 3.80597979e-01 3.08087111e-01 -2.70975083e-01
-2.49428704e-01 1.00924635e+00 -5.88852353e-02 -4.51536894e-01
3.20305467e-01 -5.39991856e-01 -1.90109342e-01 2.81169027e-01
1.83734864e-01 2.72618949e-01 -1.15567207e-01 -2.76934415e-01
2.04244349e-02 3.55167598e-01 3.00938368e-01 -7.26032794e-01
9.87344027e-01 8.11088324e-01 3.97171192e-02 1.11318386e+00
7.72567511e-01 4.39440787e-01 -5.89416921e-01 5.28094292e-01
-2.89011776e-01 -8.14419910e-02 -5.73421828e-02 -1.47325647e+00
-7.84835458e-01 7.59559393e-01 3.43402296e-01 1.11172058e-01
1.13900006e+00 -2.45567262e-01 6.17478788e-01 2.18536541e-01
3.98063511e-01 -9.92775500e-01 1.74811125e-01 6.96926475e-01
1.47947454e+00 -1.01218171e-01 -4.08869267e-01 1.63792074e-01
-6.90330863e-01 1.06454575e+00 5.85186839e-01 -5.02536930e-02
2.28484437e-01 4.93339747e-01 4.85044569e-02 1.60883501e-01
-6.93804204e-01 4.56687212e-02 3.81718069e-01 4.34551209e-01
8.70341122e-01 3.03593397e-01 -2.11964652e-01 1.20755470e+00
-1.41521931e+00 1.10343911e-01 1.87990472e-01 5.04890800e-01
-2.16845497e-02 -1.29746437e+00 -5.21550536e-01 4.73276556e-01
-2.99377918e-01 -2.92304158e-01 -6.92637086e-01 7.50184655e-01
5.91860235e-01 9.59151804e-01 -4.13529575e-02 -1.03296185e+00
4.46439564e-01 3.72459799e-01 7.95030832e-01 -7.15529919e-01
-1.20585203e+00 5.19763529e-01 1.54008001e-01 -3.35004330e-01
1.06204011e-01 -7.40591407e-01 -1.45330572e+00 2.14264542e-03
-3.03990960e-01 2.39390492e-01 4.76430476e-01 5.58280289e-01
-4.45736311e-02 1.07871473e+00 5.36687434e-01 -1.20035207e+00
-5.97940624e-01 -1.33872247e+00 -8.95763755e-01 3.66836488e-01
-1.76585376e-01 -6.21000707e-01 -4.85114723e-01 -1.86374399e-03] | [16.04645538330078, 5.486169338226318] |
98979816-cf37-42fb-97c0-baf1f692ec29 | multi-hop-selector-network-for-multi-turn | null | null | https://aclanthology.org/D19-1011 | https://aclanthology.org/D19-1011.pdf | Multi-hop Selector Network for Multi-turn Response Selection in Retrieval-based Chatbots | Multi-turn retrieval-based conversation is an important task for building intelligent dialogue systems. Existing works mainly focus on matching candidate responses with every context utterance on multiple levels of granularity, which ignore the side effect of using excessive context information. Context utterances provide abundant information for extracting more matching features, but it also brings noise signals and unnecessary information. In this paper, we will analyze the side effect of using too many context utterances and propose a multi-hop selector network (MSN) to alleviate the problem. Specifically, MSN firstly utilizes a multi-hop selector to select the relevant utterances as context. Then, the model matches the filtered context with the candidate response and obtains a matching score. Experimental results show that MSN outperforms some state-of-the-art methods on three public multi-turn dialogue datasets. | ['Songlin Hu', 'Mingming Li', 'Chunyuan Yuan', 'Wei Zhou', 'Shangwen Lv', 'Jizhong Han', 'Fuqing Zhu'] | 2019-11-01 | null | null | null | ijcnlp-2019-11 | ['conversational-response-selection'] | ['natural-language-processing'] | [ 2.07687601e-01 -7.69936517e-02 -2.65891433e-01 -6.59030318e-01
-1.08073843e+00 -3.68016481e-01 6.70922577e-01 7.77396094e-03
-5.24038494e-01 7.40731120e-01 8.17655802e-01 6.59475103e-02
-1.46703990e-02 -6.52630150e-01 1.49747685e-01 -5.90420246e-01
4.85052139e-01 3.74093711e-01 6.98584914e-01 -8.15931320e-01
5.72020650e-01 -1.55386627e-01 -1.34119713e+00 1.01539612e+00
9.68324244e-01 7.82688737e-01 5.45450747e-01 4.59249556e-01
-8.56633723e-01 7.18890071e-01 -1.00974572e+00 -2.39026859e-01
-1.74485847e-01 -8.25386345e-01 -1.26842153e+00 -6.50181845e-02
3.32691781e-02 -4.00861084e-01 -2.57961214e-01 9.07486677e-01
7.64086187e-01 6.49812818e-01 9.48964283e-02 -8.04332674e-01
-3.32279146e-01 9.43542838e-01 -1.28653437e-01 2.43539095e-01
9.65470493e-01 -5.73011190e-02 1.19441497e+00 -1.06961346e+00
5.25041401e-01 1.73141038e+00 2.84418046e-01 7.35710442e-01
-7.69261301e-01 -3.09377521e-01 5.78870773e-01 2.97971159e-01
-9.77728963e-01 -5.23625851e-01 9.18126464e-01 1.72741413e-01
9.47045028e-01 6.83797240e-01 4.40062195e-01 9.74834204e-01
-1.37245640e-01 1.45347559e+00 9.29898560e-01 -6.02421224e-01
-1.24890156e-01 1.54126003e-01 5.82002342e-01 3.65566015e-01
-7.74423003e-01 -5.45724094e-01 -6.57594502e-01 -5.65772831e-01
2.43310958e-01 6.54181540e-02 -3.76625538e-01 4.77772176e-01
-1.16707182e+00 9.67958570e-01 1.29923850e-01 5.94855547e-01
-4.04069841e-01 -5.14896452e-01 5.84349453e-01 7.64888704e-01
5.40743768e-01 4.79538888e-01 -4.51714486e-01 -3.35165471e-01
-4.03960735e-01 3.21939915e-01 9.66253996e-01 9.81876671e-01
8.62361848e-01 -5.60787261e-01 -6.35561347e-01 1.44524407e+00
2.18104631e-01 2.67039597e-01 5.62732935e-01 -9.14703071e-01
7.82324195e-01 9.10078824e-01 1.03749216e-01 -1.12263572e+00
-5.36518037e-01 1.03250228e-01 -5.26417851e-01 -8.82210314e-01
2.32485503e-01 -4.04574573e-01 -1.31212845e-01 1.42823172e+00
6.77431047e-01 -1.01288006e-01 1.89470887e-01 1.25023234e+00
1.33473349e+00 7.87343204e-01 -8.98115560e-02 -5.92893600e-01
1.40190351e+00 -1.29991436e+00 -9.35228646e-01 -1.44566417e-01
5.66629052e-01 -1.23661768e+00 1.50686562e+00 2.69738078e-01
-7.64986575e-01 -3.19168478e-01 -5.81140876e-01 -2.86346581e-02
-5.15250117e-02 8.39495882e-02 5.81868827e-01 2.25079894e-01
-7.45579243e-01 2.01028436e-01 -3.22067514e-02 -3.37043405e-01
-4.39893663e-01 2.12499455e-01 -4.36039306e-02 -1.49505302e-01
-1.79900682e+00 8.31746280e-01 2.64037922e-02 1.19199388e-01
-3.34462076e-01 -1.85200915e-01 -7.16031611e-01 -7.22032553e-03
8.18848193e-01 -2.46031821e-01 1.69372797e+00 -8.47391546e-01
-1.85060251e+00 1.42918348e-01 -6.30248725e-01 -5.91058098e-03
2.18289912e-01 -2.62061387e-01 -5.78231692e-01 2.80610025e-01
8.31284821e-02 2.91506588e-01 4.99669492e-01 -9.42941487e-01
-1.09545732e+00 -1.03276230e-01 4.45357442e-01 6.07079268e-01
-2.08540484e-01 1.60893202e-01 -5.49452186e-01 -3.00154775e-01
4.06526715e-01 -7.79119492e-01 -1.74050853e-01 -9.36941326e-01
-3.50567222e-01 -9.62496936e-01 1.00970399e+00 -3.27465475e-01
1.62189746e+00 -1.75145066e+00 -2.35437993e-02 -6.59290254e-02
1.32486895e-01 3.61635178e-01 -3.89860302e-01 9.38410461e-01
5.98112941e-01 -1.17013268e-01 8.25410187e-02 -1.12957343e-01
-6.19194992e-02 7.36230984e-02 -1.39192238e-01 -1.39579192e-01
-9.23013315e-02 6.35413289e-01 -1.24219358e+00 -5.97550988e-01
1.16132312e-01 1.83925703e-01 -3.96192580e-01 4.09989655e-01
-4.55667198e-01 4.95908380e-01 -9.24391508e-01 4.31010753e-01
3.79389346e-01 -6.84434101e-02 4.92985129e-01 -1.49270847e-01
-1.00484133e-01 9.18004572e-01 -8.79139185e-01 1.59717619e+00
-7.85140097e-01 4.29475337e-01 1.07219078e-01 -7.09118605e-01
1.01805067e+00 4.31530148e-01 3.59582424e-01 -7.11420357e-01
1.72702909e-01 9.97220948e-02 6.25481978e-02 -9.43359196e-01
8.97823095e-01 9.56945196e-02 -4.14565027e-01 7.58041680e-01
-1.91362977e-01 1.18685551e-01 3.33904028e-02 4.28475887e-01
8.81139934e-01 -2.80620158e-01 2.06295237e-01 -5.17818853e-02
1.08858240e+00 -1.64461523e-01 7.79736996e-01 7.01180875e-01
-4.16346312e-01 1.44645095e-01 4.06772375e-01 -3.93651158e-01
-3.01236302e-01 -2.90722132e-01 3.49916518e-01 1.66271019e+00
3.75629693e-01 -5.71482003e-01 -6.90782130e-01 -1.01163626e+00
-1.84672624e-01 6.03564978e-01 -3.39251131e-01 -1.73126422e-02
-7.14961171e-01 -4.69730616e-01 3.62759352e-01 8.45113676e-03
8.37937534e-01 -1.13158238e+00 -7.91562870e-02 5.89455485e-01
-1.05104232e+00 -8.96925986e-01 -8.95926237e-01 -5.05216837e-01
-3.69200706e-01 -1.04225516e+00 -3.76664847e-01 -7.19447792e-01
1.83180109e-01 9.60359156e-01 1.11256337e+00 5.47218680e-01
2.58955508e-01 2.32987180e-01 -9.32378173e-01 -1.65138245e-01
-5.02614558e-01 3.72022599e-01 3.89278657e-03 1.78198814e-01
9.46462452e-01 -2.93117285e-01 -6.69673502e-01 8.03590596e-01
-5.81398487e-01 -1.64486706e-01 3.94954503e-01 1.19092643e+00
2.50628173e-01 -1.54601038e-01 1.32017756e+00 -8.89024496e-01
1.35382271e+00 -3.88969511e-01 -1.68311045e-01 6.83801770e-01
-4.35951144e-01 -1.48818627e-01 6.38345122e-01 -3.77076983e-01
-1.46794605e+00 -9.12310258e-02 -2.32318118e-01 2.16066033e-01
-2.94020493e-02 5.72181344e-01 -3.43221217e-01 8.98229331e-02
3.46730322e-01 2.87714243e-01 -1.80028141e-01 -5.20743310e-01
2.68827200e-01 1.24828386e+00 -7.31272101e-02 -7.18765378e-01
-2.46680565e-02 -4.34135646e-02 -5.41130364e-01 -8.43461573e-01
-9.69511271e-01 -9.74712849e-01 -3.78735334e-01 -5.57382584e-01
5.32760918e-01 -5.53312659e-01 -9.38825667e-01 1.36751667e-01
-1.32242560e+00 1.20658942e-01 3.70660663e-01 5.80115676e-01
-2.26910919e-01 6.05219901e-01 -7.85111487e-01 -1.10042286e+00
-5.98430455e-01 -1.27626514e+00 8.51916552e-01 6.29875958e-01
-1.89465761e-01 -7.34302759e-01 -1.22095935e-01 5.22320390e-01
4.63793188e-01 -5.04641414e-01 7.18596041e-01 -1.01688743e+00
-3.77579153e-01 -1.25544474e-01 -4.69050817e-02 -4.67452891e-02
4.45739746e-01 -2.51085520e-01 -1.01736391e+00 1.51801659e-02
2.27899671e-01 -4.66963053e-01 8.49784374e-01 8.33791122e-03
6.96547985e-01 -5.92708707e-01 -2.67483741e-01 -2.50544965e-01
8.25553417e-01 3.68076682e-01 3.03373188e-01 1.31773919e-01
4.09598023e-01 1.01519430e+00 1.22649574e+00 3.75301540e-01
7.00598121e-01 6.19102180e-01 5.13935275e-02 2.06043527e-01
2.98612386e-01 -1.29245803e-01 1.53459579e-01 1.28734207e+00
2.75528103e-01 -2.16822803e-01 -6.27371252e-01 5.62582731e-01
-2.16949916e+00 -1.12306273e+00 -6.06025085e-02 1.94594932e+00
1.05057096e+00 8.57302994e-02 3.04116786e-01 -1.54565632e-01
9.96435702e-01 3.48920882e-01 -3.67983758e-01 -4.66502905e-01
-1.03702441e-01 -2.84624934e-01 -3.55376393e-01 8.84067237e-01
-8.75864506e-01 1.15924239e+00 5.76015568e+00 9.89473104e-01
-8.67666960e-01 6.92860857e-02 4.02582884e-01 -1.12351432e-01
-4.10837799e-01 5.69451563e-02 -9.76885319e-01 4.24067885e-01
5.19893169e-01 -3.87108892e-01 4.51148540e-01 7.74011254e-01
3.98936361e-01 -3.37781101e-01 -8.43085408e-01 6.99175715e-01
1.61322221e-01 -1.08686304e+00 1.72657833e-01 -4.48288321e-01
4.14746284e-01 -1.87216446e-01 -3.76694560e-01 6.48092866e-01
2.83771455e-01 -4.70430851e-01 -9.48837250e-02 6.37021363e-01
2.58870274e-01 -9.62759793e-01 1.02384603e+00 6.12311721e-01
-1.17041397e+00 -2.11250886e-01 -5.22534847e-01 4.03917534e-03
1.15671270e-01 5.25626898e-01 -1.10906756e+00 7.58094251e-01
2.80550867e-01 2.62769759e-01 -2.22986996e-01 6.75025165e-01
-1.03730947e-01 5.69946289e-01 -2.86162555e-01 -6.78556383e-01
5.72556019e-01 -1.78754061e-01 6.96020901e-01 1.46387744e+00
-1.05941333e-01 6.26271009e-01 7.75462329e-01 2.32469648e-01
-6.39566779e-02 6.10296071e-01 -3.45486760e-01 4.03226644e-01
1.08000517e+00 1.33315575e+00 -3.76305699e-01 -5.58312595e-01
-6.02852345e-01 7.72366881e-01 2.97708899e-01 2.17681557e-01
-2.64780164e-01 -7.27360964e-01 5.08092284e-01 -4.20338601e-01
-5.28895073e-02 1.36315584e-01 1.79683641e-01 -1.09185004e+00
1.13754548e-01 -1.12282848e+00 5.64413071e-01 -3.45050752e-01
-1.38988709e+00 7.47098744e-01 -2.32886806e-01 -1.33869219e+00
-4.39924777e-01 1.31562412e-01 -8.20222139e-01 9.62970614e-01
-1.41517460e+00 -9.01562452e-01 -1.12541959e-01 5.56199074e-01
1.30098867e+00 -2.24198669e-01 8.91517103e-01 2.41393074e-01
-5.75955689e-01 8.32276940e-01 -1.71979472e-01 1.89880997e-01
9.47542846e-01 -1.05718279e+00 7.89054334e-02 4.34702784e-01
-2.67531455e-01 9.72991407e-01 5.42861342e-01 -5.19314408e-01
-1.31040752e+00 -7.96373665e-01 1.28523231e+00 -2.86598712e-01
3.46358150e-01 -3.81242409e-02 -1.03593302e+00 2.24654898e-01
4.45415914e-01 -5.70274591e-01 8.98368537e-01 4.02402937e-01
-6.97723776e-02 -6.66113049e-02 -1.14482796e+00 7.62785733e-01
7.78593659e-01 -7.29684770e-01 -9.38674748e-01 3.89408708e-01
1.22660160e+00 -4.94905770e-01 -7.64177740e-01 1.84201717e-01
6.01972759e-01 -1.03726864e+00 6.48918450e-01 -6.04938388e-01
3.06794226e-01 -1.58529282e-01 -2.99179882e-01 -1.44819057e+00
-7.31221586e-02 -8.87944758e-01 9.01300162e-02 1.46314621e+00
5.96486866e-01 -5.46944678e-01 4.54862475e-01 7.89623559e-01
-2.48012915e-01 -9.10665393e-01 -9.11875904e-01 -4.08811927e-01
-1.83300301e-01 -8.57649222e-02 1.16198170e+00 1.18277192e+00
7.49629378e-01 1.03201878e+00 -4.54933077e-01 -1.92543447e-01
-1.61277968e-02 5.68311751e-01 8.62582386e-01 -9.27129328e-01
-1.41932061e-02 -4.03032660e-01 2.29833692e-01 -1.73784411e+00
2.46187389e-01 -4.13085669e-01 3.27417433e-01 -1.42973828e+00
4.03592288e-02 -4.18266028e-01 -2.49299973e-01 1.60161957e-01
-8.87612760e-01 -4.14985389e-01 1.37381732e-01 2.60373473e-01
-1.17400098e+00 6.95019901e-01 1.42169440e+00 -1.08174264e-01
-6.26647890e-01 4.19718325e-01 -6.24523640e-01 6.53801382e-01
8.68297458e-01 -3.07426453e-01 -5.45936108e-01 -4.32533532e-01
-3.60431671e-02 6.12223446e-01 -3.09811532e-01 -3.46229106e-01
5.35825014e-01 -4.91562992e-01 -7.67950416e-02 -9.35643375e-01
4.58297819e-01 -5.89384139e-01 -6.73204243e-01 -7.83040002e-02
-9.04970765e-01 7.18457773e-02 -2.39490092e-01 6.00280881e-01
-4.40467417e-01 -3.61694783e-01 2.24141359e-01 -4.34534878e-01
-7.16733575e-01 -3.21064815e-02 -5.65739989e-01 7.20670447e-02
3.32667023e-01 3.49450022e-01 -5.18245459e-01 -8.40338290e-01
-6.31719649e-01 7.82822192e-01 1.01658581e-02 6.88446939e-01
6.69219077e-01 -1.34336710e+00 -6.74431026e-01 -1.53645113e-01
2.79640615e-01 -1.33168146e-01 4.49248821e-01 6.26391053e-01
2.77458489e-01 5.55324078e-01 3.50909144e-01 -5.58836460e-01
-1.64646876e+00 2.57596731e-01 2.47289076e-01 -3.11575830e-01
-5.02996981e-01 9.65768218e-01 -1.03589527e-01 -8.05290043e-01
2.52122313e-01 -4.23049107e-02 -7.80398607e-01 3.38097394e-01
1.02230740e+00 1.56844631e-01 7.41512775e-02 -6.24038577e-01
-2.40788639e-01 1.88872814e-01 -4.44865227e-01 -1.65636748e-01
8.90614867e-01 -7.66552389e-01 -3.16521972e-01 5.06597221e-01
1.10429144e+00 9.07323807e-02 -7.58615613e-01 -8.78277600e-01
1.78668395e-01 -6.96780384e-01 -3.23628157e-01 -8.66892636e-01
-7.85047054e-01 6.40905440e-01 2.22395994e-02 6.87205136e-01
1.05859983e+00 -2.43818536e-01 1.24282420e+00 1.13744891e+00
4.86743093e-01 -1.53024435e+00 1.98901713e-01 1.06031597e+00
9.63489115e-01 -1.36602938e+00 -4.97269072e-02 -5.62893510e-01
-9.72900212e-01 9.88943636e-01 1.03424644e+00 2.60805219e-01
4.06868339e-01 -3.04064095e-01 5.63847244e-01 -2.28844449e-01
-1.19907367e+00 -4.04824674e-01 -7.56833553e-02 3.83122593e-01
5.96571565e-01 -2.76163391e-05 -9.05975699e-01 7.85950601e-01
-1.36359185e-01 -4.00285184e-01 6.02246940e-01 9.86810327e-01
-9.20083284e-01 -1.39405739e+00 -2.76332617e-01 4.82380331e-01
-3.97521406e-01 -2.71274298e-01 -9.53871548e-01 4.12895113e-01
-3.71442109e-01 1.76424026e+00 -4.16252315e-01 -8.05037260e-01
5.17966211e-01 4.34536994e-01 -8.58621448e-02 -6.94167376e-01
-1.12284386e+00 2.23986343e-01 6.92849874e-01 -4.49947029e-01
-6.21126294e-01 -4.82799113e-01 -1.13229036e+00 -1.71222165e-01
-9.71335649e-01 6.38828576e-01 3.20695490e-01 1.18851125e+00
4.13023889e-01 2.85784721e-01 1.38593578e+00 -4.40875828e-01
-7.25575626e-01 -1.40867674e+00 -1.54956713e-01 4.01166290e-01
3.37090582e-01 -5.60651779e-01 -2.23560289e-01 -5.41769803e-01] | [12.514565467834473, 7.794970989227295] |
27067dff-2c9f-48e4-9d6c-02c69d2b718b | road-segmentation-on-low-resolution-lidar | 2005.13102 | null | https://arxiv.org/abs/2005.13102v1 | https://arxiv.org/pdf/2005.13102v1.pdf | Road Segmentation on low resolution Lidar point clouds for autonomous vehicles | Point cloud datasets for perception tasks in the context of autonomous driving often rely on high resolution 64-layer Light Detection and Ranging (LIDAR) scanners. They are expensive to deploy on real-world autonomous driving sensor architectures which usually employ 16/32 layer LIDARs. We evaluate the effect of subsampling image based representations of dense point clouds on the accuracy of the road segmentation task. In our experiments the low resolution 16/32 layer LIDAR point clouds are simulated by subsampling the original 64 layer data, for subsequent transformation in to a feature map in the Bird-Eye-View (BEV) and SphericalView (SV) representations of the point cloud. We introduce the usage of the local normal vector with the LIDAR's spherical coordinates as an input channel to existing LoDNN architectures. We demonstrate that this local normal feature in conjunction with classical features not only improves performance for binary road segmentation on full resolution point clouds, but it also reduces the negative impact on the accuracy when subsampling dense point clouds as compared to the usage of classical features alone. We assess our method with several experiments on two datasets: KITTI Road-segmentation benchmark and the recently released Semantic KITTI dataset. | ['Santiago Velasco-Forero', 'B Ravi Kiran', 'Andres Serna', 'Nagarjuna Vemuri', 'Leonardo Gigli', 'Thomas Paul', 'Beatriz Marcotegui'] | 2020-05-27 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 2.67026871e-01 -1.83526129e-02 1.06359422e-01 -6.98408067e-01
-5.02094626e-01 -3.62619400e-01 8.73726606e-01 2.36287236e-01
-7.25308716e-01 4.54908192e-01 -4.64019477e-01 -4.57765847e-01
3.94211337e-03 -1.27352798e+00 -1.08103919e+00 -2.89695263e-01
1.10662594e-01 1.03418398e+00 7.94085681e-01 -3.33544344e-01
3.49588543e-01 1.04635561e+00 -2.44436789e+00 -1.60750464e-01
8.98695171e-01 1.26411402e+00 2.87617326e-01 5.49963772e-01
-5.93451560e-01 -1.13508657e-01 -2.92078942e-01 7.74608254e-02
7.44582772e-01 5.35089552e-01 -1.53962016e-01 -1.11252904e-01
1.08688653e+00 -2.72314817e-01 2.36627370e-01 9.87900257e-01
1.92402706e-01 1.23011395e-01 5.42656481e-01 -1.43166482e+00
5.99670485e-02 -2.11294759e-02 -6.98035657e-01 8.10134411e-02
3.04636043e-02 2.53949314e-01 5.82907498e-01 -1.06992650e+00
6.29330754e-01 1.53616226e+00 8.49661231e-01 2.04540920e-02
-1.27607167e+00 -1.03874135e+00 6.57387674e-02 6.76747710e-02
-1.70717573e+00 -3.97991717e-01 5.67339480e-01 -5.00437558e-01
1.15340507e+00 5.99160306e-02 5.14460325e-01 4.15472507e-01
2.80203819e-01 -2.12426811e-01 1.19708920e+00 6.37451038e-02
3.35735649e-01 2.01331288e-01 2.37024039e-01 5.38938344e-01
8.15524697e-01 4.95168358e-01 -4.56401914e-01 -5.13165966e-02
6.55623317e-01 2.26461552e-02 2.47991636e-01 -8.71257305e-01
-9.19251502e-01 1.03625011e+00 9.08794045e-01 -2.21362695e-01
-3.58716041e-01 3.91113937e-01 1.55721292e-01 1.66051522e-01
4.15806264e-01 4.59325761e-02 -1.94421470e-01 1.97784707e-01
-9.55111802e-01 4.47725087e-01 3.70376229e-01 1.20855856e+00
1.71649230e+00 -9.68728885e-02 3.64125580e-01 5.09945571e-01
3.46941411e-01 9.13498223e-01 -2.10061613e-02 -1.20583713e+00
5.94585538e-01 8.96904290e-01 -2.38433364e-03 -6.29635513e-01
-4.08066243e-01 -1.42055884e-01 -4.91115004e-01 8.21885943e-01
1.71604797e-01 1.75267637e-01 -1.31506217e+00 9.32660460e-01
5.09338439e-01 2.43082002e-01 1.43148169e-01 6.75444603e-01
8.53738308e-01 4.32244152e-01 -1.88916773e-02 4.52817380e-01
1.31105566e+00 -3.32333595e-01 -1.66626900e-01 -4.98029053e-01
5.14324188e-01 -6.24545991e-01 9.02308941e-01 1.56758741e-01
-7.45062411e-01 -8.42969120e-01 -1.29040885e+00 -5.87775648e-01
-9.24446046e-01 -7.87060186e-02 4.72340494e-01 6.86185300e-01
-1.21822417e+00 4.25778300e-01 -7.38662064e-01 -3.44521552e-01
5.73353231e-01 6.49726689e-01 -3.65864128e-01 -3.04732978e-01
-7.60578275e-01 1.05493534e+00 3.39469492e-01 -9.58130881e-02
-3.46092641e-01 -9.32375312e-01 -1.02528226e+00 -1.29154682e-01
3.00283581e-01 -6.79631174e-01 8.12095404e-01 -1.36244506e-01
-1.10372090e+00 8.50462735e-01 -3.33925068e-01 -8.22401822e-01
4.92696971e-01 -1.14536509e-01 -6.32394701e-02 1.16099030e-01
3.73719901e-01 1.46208584e+00 7.46058524e-01 -1.42523634e+00
-1.11559725e+00 -7.26457238e-01 -1.78041026e-01 1.80579811e-01
6.69297993e-01 -6.48688257e-01 -5.06372117e-02 3.49797934e-01
4.83831763e-01 -1.05923498e+00 -3.72141302e-01 1.82898223e-01
-1.34606853e-01 -2.59273440e-01 1.34945178e+00 1.50365606e-01
1.36096090e-01 -2.26572633e+00 -5.63382030e-01 4.15559918e-01
8.60043913e-02 7.28562772e-02 3.28145456e-04 3.98373902e-02
1.25635639e-01 4.17240001e-02 -3.03079158e-01 -3.15863699e-01
-1.03843644e-01 6.96847916e-01 -4.64621425e-01 5.57770729e-01
4.21206862e-01 7.51998067e-01 -5.21988034e-01 -4.82235104e-01
7.62186050e-01 7.72161841e-01 -3.62736046e-01 -2.15470016e-01
-2.83704966e-01 2.19598502e-01 -3.85239750e-01 4.48625743e-01
1.22399938e+00 4.26320165e-01 -5.53892136e-01 -9.64492932e-02
-7.51417220e-01 5.08912623e-01 -1.09838891e+00 1.57025516e+00
-5.44805348e-01 7.31474817e-01 9.00596455e-02 -3.46027493e-01
1.34812009e+00 -2.11130798e-01 1.56269029e-01 -7.70402372e-01
8.82796049e-02 2.97388971e-01 -7.22356373e-03 2.42560625e-01
9.34388518e-01 -1.63444445e-01 -5.16562909e-02 -6.22464195e-02
-3.61524254e-01 -9.36661601e-01 3.99416573e-02 2.73366924e-03
6.90600693e-01 1.72123373e-01 -1.53053096e-02 -3.10104489e-01
4.56644326e-01 5.24063945e-01 2.70087063e-01 5.88042080e-01
1.56715199e-01 6.08607590e-01 2.17168465e-01 -3.41973543e-01
-1.13581336e+00 -1.14269257e+00 -6.54173851e-01 5.77426553e-01
3.41231078e-01 -1.26968369e-01 -4.09374982e-01 -1.63662717e-01
6.97568059e-01 9.04086828e-01 -3.81360024e-01 3.25266309e-02
-5.26687503e-01 -2.72210687e-01 3.86579901e-01 5.67515194e-01
6.25053048e-01 -6.94274127e-01 -1.28642535e+00 -3.47223096e-02
4.88214999e-01 -1.46159935e+00 4.10278887e-01 4.54379022e-01
-1.11990726e+00 -9.17094707e-01 3.60901989e-02 -2.06223950e-01
3.43763977e-01 7.23871052e-01 9.27270234e-01 -1.08468480e-01
-2.37617463e-01 3.28474343e-01 -4.23707515e-02 -8.80964279e-01
-1.76622733e-01 1.76622182e-01 1.28461989e-02 -4.40883368e-01
8.15441668e-01 -6.35682523e-01 -3.27101380e-01 2.88422287e-01
-7.37235427e-01 -6.69101775e-02 6.10368133e-01 6.11621961e-02
9.31218028e-01 -1.21424600e-01 1.22387975e-01 -7.74048328e-01
4.24801670e-02 -3.39689165e-01 -1.16724217e+00 -6.50345147e-01
-6.44680083e-01 7.70523548e-02 1.46556541e-01 1.82435468e-01
-4.93230551e-01 4.52122182e-01 -1.28442690e-01 -7.51599908e-01
-5.13753355e-01 -1.12666786e-02 6.80510923e-02 -4.68860954e-01
6.26319945e-01 -1.09641984e-01 2.11699665e-01 -4.95590627e-01
5.37073135e-01 6.04831755e-01 6.74716711e-01 -3.97270292e-01
1.12537110e+00 9.81248677e-01 5.90544224e-01 -1.22548449e+00
-2.28359997e-01 -7.68395305e-01 -9.90148246e-01 7.76827857e-02
9.14200187e-01 -1.14828336e+00 -5.36210477e-01 -2.58420873e-02
-1.16514409e+00 -5.57036065e-02 -5.90590954e-01 3.08740139e-01
-6.26099229e-01 1.33830383e-01 2.72727162e-01 -7.14054406e-01
2.58895140e-02 -1.39043057e+00 1.51200926e+00 1.58454761e-01
2.56009370e-01 -3.26106727e-01 -5.59724122e-02 2.27101117e-01
2.92402953e-01 3.60263407e-01 8.92453492e-01 -2.41263345e-01
-1.11973631e+00 -1.99859902e-01 -6.80492699e-01 2.16743141e-01
-2.08880842e-01 9.03707072e-02 -1.25597370e+00 1.31773859e-01
-2.48147264e-01 7.40851015e-02 1.12989724e+00 2.30395272e-01
8.04643512e-01 5.78575790e-01 -4.74800766e-01 7.77925014e-01
1.61144066e+00 -1.55204460e-01 6.26929224e-01 4.06091571e-01
8.51525426e-01 7.59714663e-01 9.18293774e-01 -1.65825393e-02
6.23862624e-01 6.62172735e-01 9.70253468e-01 -7.04428926e-02
-2.09122345e-01 -1.51718572e-01 1.42610306e-02 5.47237471e-02
-4.40622158e-02 3.95663917e-01 -1.19687200e+00 4.63076890e-01
-1.36953342e+00 -4.90872324e-01 -6.61127448e-01 2.35920548e+00
2.68647611e-01 4.81951565e-01 -5.79511449e-02 1.78900555e-01
4.66256052e-01 6.52198493e-02 -5.62900484e-01 -7.28246629e-01
9.11206007e-02 5.82310677e-01 1.36933279e+00 6.51880383e-01
-9.06300128e-01 1.10085058e+00 5.10911417e+00 5.16362309e-01
-1.19056356e+00 2.74155531e-02 -3.01902872e-02 -9.65356678e-02
-1.90901905e-01 7.62860551e-02 -1.37694609e+00 1.39536411e-01
1.42874718e+00 9.63821039e-02 6.15139008e-02 1.02193439e+00
1.77865550e-01 -5.31862676e-01 -9.78719473e-01 9.08179939e-01
-3.95563126e-01 -1.22521174e+00 2.03986675e-01 6.34972453e-01
2.82783270e-01 8.67804527e-01 1.18963838e-01 2.40412489e-01
2.89262086e-01 -1.12563241e+00 8.48126233e-01 1.72345698e-01
1.23274112e+00 -8.89778614e-01 5.78781426e-01 5.44072568e-01
-1.32541668e+00 8.76122862e-02 -8.79596114e-01 -1.55301630e-01
1.18313543e-01 4.56408352e-01 -1.32350564e+00 4.66765136e-01
7.65435398e-01 5.48131526e-01 -8.26960504e-01 8.54109466e-01
8.61816332e-02 1.47067815e-01 -1.01095080e+00 3.85054767e-01
5.75977266e-01 -3.92421663e-01 6.24708593e-01 1.03316259e+00
3.16074342e-01 -1.47634253e-01 1.27034351e-01 9.51417804e-01
1.31830350e-01 -1.97052732e-01 -1.16083419e+00 4.30762202e-01
5.20245254e-01 1.16967511e+00 -7.65178978e-01 -3.94867480e-01
-2.99354255e-01 1.07684303e-02 4.46133167e-02 3.54782715e-02
-3.89875412e-01 -3.70220125e-01 1.13374734e+00 8.19636762e-01
6.97433412e-01 -6.74731672e-01 -6.53020740e-01 -4.25826192e-01
-1.31239235e-01 -2.54804224e-01 -2.47848034e-01 -9.38716173e-01
-6.78138316e-01 5.21198571e-01 3.77697885e-01 -1.20375562e+00
-4.05080974e-01 -6.36295855e-01 -3.39192271e-01 1.22388232e+00
-2.22398353e+00 -1.13613284e+00 -8.38777602e-01 5.26146352e-01
4.04111654e-01 2.20975041e-01 5.90916932e-01 1.27250105e-01
-5.10338787e-03 -2.08706871e-01 -2.46662483e-01 -3.19054037e-01
3.63687515e-01 -1.24782848e+00 9.43948209e-01 5.40692687e-01
-6.47262633e-02 3.80021691e-01 7.26281047e-01 -7.88351059e-01
-1.27141786e+00 -1.44380212e+00 3.95531774e-01 -5.73077381e-01
3.18072617e-01 -5.54654896e-01 -1.09803808e+00 6.61406517e-01
-2.81264365e-01 2.31316224e-01 9.65137258e-02 -2.50573456e-01
-3.50085050e-01 -2.89203137e-01 -1.45097983e+00 2.28799686e-01
1.03725064e+00 -3.78988564e-01 -5.12780368e-01 5.09623773e-02
7.42167592e-01 -4.42618579e-01 -7.03062832e-01 7.68426359e-01
3.99843991e-01 -1.18290544e+00 1.30084276e+00 8.92855600e-02
1.09961793e-01 -6.50833368e-01 -4.02293295e-01 -9.40176010e-01
-8.32164288e-02 4.97950278e-02 5.09903193e-01 7.49248147e-01
1.13445885e-01 -9.29820120e-01 9.94237840e-01 1.57138586e-01
-4.06287849e-01 -2.89440483e-01 -1.41964352e+00 -6.99590266e-01
5.67627884e-02 -7.28897452e-01 1.01899767e+00 3.02281499e-01
-1.04147398e+00 3.52596462e-01 7.28645444e-01 5.89257836e-01
8.52025270e-01 5.40120006e-02 1.31745672e+00 -1.98339403e+00
6.03729010e-01 -4.01327282e-01 -9.69151795e-01 -8.20667386e-01
2.31861006e-02 -8.49720418e-01 1.47256300e-01 -1.55718970e+00
-7.07012594e-01 -9.12573814e-01 2.75771618e-01 8.56686682e-02
3.69812280e-01 5.34197986e-01 1.17664434e-01 2.90533006e-01
-1.70081686e-02 4.05157298e-01 7.43546486e-01 7.28840455e-02
-1.23452775e-01 2.05456540e-02 -1.65177539e-01 7.56768703e-01
6.42457187e-01 -6.18972600e-01 -3.14351290e-01 -4.13663536e-01
1.14559129e-01 -3.16051811e-01 6.66164458e-01 -1.50000656e+00
1.91930473e-01 8.89387075e-03 3.45008880e-01 -1.44578218e+00
1.00107205e+00 -1.15041220e+00 1.17446944e-01 3.78326893e-01
3.30248535e-01 1.15495443e-01 4.82230276e-01 5.76622427e-01
6.92357793e-02 -8.42423514e-02 9.98345256e-01 -1.16647005e-01
-8.63819301e-01 1.70394927e-01 -7.11760521e-02 -4.40504998e-01
1.03142762e+00 -9.43683207e-01 -2.33691737e-01 7.62137249e-02
-1.58322841e-01 2.91822433e-01 1.03149867e+00 4.32414383e-01
6.85883105e-01 -9.39657629e-01 -5.88382900e-01 8.02590132e-01
2.52539426e-01 8.13234746e-01 -1.94540739e-01 6.44796133e-01
-8.63453567e-01 8.19239378e-01 -3.39342386e-01 -1.25902081e+00
-1.25647950e+00 3.41710150e-01 2.78733581e-01 3.94225955e-01
-9.56716776e-01 5.07189214e-01 2.20010430e-01 -4.33994889e-01
-9.66383964e-02 -9.99507368e-01 -2.65925646e-01 4.10060845e-02
6.67177737e-02 4.12875623e-01 4.14020598e-01 -1.01628339e+00
-4.31175232e-01 9.88527060e-01 5.06858975e-02 -1.78296015e-01
1.04586220e+00 -1.70554191e-01 5.45876063e-02 6.59358323e-01
8.76298547e-01 -8.45548660e-02 -1.20444119e+00 -1.95942506e-01
1.84014533e-02 -7.15246320e-01 3.63262713e-01 -2.22005814e-01
-7.59955645e-01 1.20514977e+00 8.73939991e-01 1.92894220e-01
3.70709211e-01 2.89807096e-02 6.94415450e-01 4.32777077e-01
7.36461937e-01 -7.53431976e-01 -8.86681259e-01 5.36422014e-01
5.65579951e-01 -1.17104244e+00 2.88340777e-01 -8.13678026e-01
-3.65512818e-01 9.16156292e-01 3.32453221e-01 -5.73423564e-01
5.56058884e-01 2.78957456e-01 1.11511640e-01 -5.89213610e-01
-5.65909445e-01 -6.46913052e-01 1.11847006e-01 9.03509855e-01
-1.11152656e-01 2.36499250e-01 1.93666592e-01 -1.25268951e-01
-8.27913582e-01 -7.11828321e-02 5.40485382e-01 9.02187943e-01
-1.07450485e+00 -8.64084423e-01 -6.83262289e-01 6.86842978e-01
2.52056003e-01 1.10807583e-01 -2.32481048e-01 1.07608247e+00
5.42335331e-01 7.91276157e-01 8.70115519e-01 -1.98308527e-01
6.12283051e-01 3.48331518e-02 2.59965062e-01 -9.86371219e-01
-3.34971756e-01 -3.00278544e-01 -9.89913493e-02 -6.77176833e-01
-5.90530187e-02 -7.96678782e-01 -1.49538219e+00 -3.46503794e-01
-1.38382171e-03 -5.38095385e-02 1.57616651e+00 6.48464262e-01
5.57574868e-01 2.48587564e-01 2.18850181e-01 -1.44704688e+00
-2.97135502e-01 -8.59104335e-01 -4.77012992e-01 -5.41908257e-02
4.54851806e-01 -1.10618353e+00 -3.55528116e-01 -3.52224290e-01] | [7.952734470367432, -2.7723844051361084] |
1dc7fceb-289f-49c1-97a4-58bcb12874cd | a-point-set-generation-network-for-3d-object | 1612.00603 | null | http://arxiv.org/abs/1612.00603v2 | http://arxiv.org/pdf/1612.00603v2.pdf | A Point Set Generation Network for 3D Object Reconstruction from a Single Image | Generation of 3D data by deep neural network has been attracting increasing
attention in the research community. The majority of extant works resort to
regular representations such as volumetric grids or collection of images;
however, these representations obscure the natural invariance of 3D shapes
under geometric transformations and also suffer from a number of other issues.
In this paper we address the problem of 3D reconstruction from a single image,
generating a straight-forward form of output -- point cloud coordinates. Along
with this problem arises a unique and interesting issue, that the groundtruth
shape for an input image may be ambiguous. Driven by this unorthodox output
form and the inherent ambiguity in groundtruth, we design architecture, loss
function and learning paradigm that are novel and effective. Our final solution
is a conditional shape sampler, capable of predicting multiple plausible 3D
point clouds from an input image. In experiments not only can our system
outperform state-of-the-art methods on single image based 3d reconstruction
benchmarks; but it also shows a strong performance for 3d shape completion and
promising ability in making multiple plausible predictions. | ['Leonidas Guibas', 'Hao Su', 'Haoqiang Fan'] | 2016-12-02 | a-point-set-generation-network-for-3d-object-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Fan_A_Point_Set_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Fan_A_Point_Set_CVPR_2017_paper.pdf | cvpr-2017-7 | ['3d-object-reconstruction', '3d-object-reconstruction-from-a-single-image'] | ['computer-vision', 'computer-vision'] | [ 2.65351385e-01 2.70358622e-01 2.60173082e-01 -3.69784087e-01
-8.84559751e-01 -4.86716598e-01 7.67991126e-01 -1.03486791e-01
2.01909430e-02 5.34852743e-01 1.11773880e-02 -2.68801779e-01
-1.79458018e-02 -9.71374929e-01 -1.18080115e+00 -3.57786298e-01
1.44891858e-01 8.99916410e-01 6.91179410e-02 -2.51682281e-01
2.00809434e-01 9.70022023e-01 -1.69079363e+00 -6.17047548e-02
7.63594210e-01 1.22212613e+00 1.74708486e-01 2.46782228e-01
-5.07151723e-01 -2.46785395e-02 -3.14858615e-01 -3.68489087e-01
6.13680482e-01 6.28800467e-02 -6.12854302e-01 2.81322807e-01
7.25565493e-01 -2.64266878e-01 3.12302634e-02 9.75340962e-01
4.60410595e-01 -1.89692646e-01 1.02678549e+00 -1.17972922e+00
-9.37961936e-01 3.85416113e-02 -5.98807931e-01 -4.72791493e-01
2.91408181e-01 1.96466714e-01 7.79600382e-01 -1.39015913e+00
5.88310897e-01 1.60016131e+00 9.45035696e-01 4.19156373e-01
-1.46967793e+00 -4.49145854e-01 4.44753207e-02 -4.14737821e-01
-1.44198704e+00 -1.67843610e-01 9.64001119e-01 -5.36314905e-01
7.84310997e-01 2.21795440e-01 7.22824275e-01 1.05033600e+00
1.23377264e-01 6.61577463e-01 1.00135684e+00 -3.19646716e-01
1.67817101e-01 -6.71700686e-02 -3.44269514e-01 5.81183374e-01
3.11373264e-01 1.42431453e-01 -1.75864726e-01 -1.32266477e-01
1.14587891e+00 1.58488676e-01 -1.63230330e-01 -7.83191741e-01
-1.25606465e+00 7.48441398e-01 6.48717880e-01 -1.31132603e-01
-5.78569233e-01 2.33752251e-01 -2.42445804e-03 4.60514538e-02
5.83860636e-01 3.93801928e-01 -3.72597188e-01 3.18680763e-01
-7.42881835e-01 7.04390764e-01 5.27721167e-01 1.06362677e+00
9.71879780e-01 3.47305775e-01 9.64749902e-02 4.65456516e-01
4.53337759e-01 7.11462855e-01 -2.04899963e-02 -7.59166181e-01
4.92304832e-01 8.00609589e-01 1.39593199e-01 -1.07235885e+00
-2.66990185e-01 -4.38261688e-01 -1.14516568e+00 6.53291881e-01
3.08167964e-01 1.85399324e-01 -1.10324728e+00 1.57066810e+00
5.47342718e-01 1.34653240e-01 -5.29923625e-02 9.00400817e-01
8.23311031e-01 5.15330017e-01 -1.95916981e-01 2.03128457e-01
9.33991790e-01 -3.59824330e-01 -2.51623333e-01 -1.57410532e-01
-1.84383646e-01 -8.68743062e-01 1.10111415e+00 2.79601604e-01
-1.26708555e+00 -6.89511240e-01 -9.21917558e-01 -2.84975499e-01
-2.61565745e-01 -3.98834012e-02 5.76724112e-01 2.73982704e-01
-1.03235233e+00 6.80394650e-01 -6.16554916e-01 -2.48363584e-01
5.39025366e-01 3.90353173e-01 -4.51685131e-01 -9.11205262e-02
-6.57839417e-01 8.20988655e-01 2.44902432e-01 5.09945191e-02
-7.39374101e-01 -8.62005472e-01 -8.66101146e-01 -8.13999772e-02
2.12123886e-01 -1.13881373e+00 9.83543873e-01 -6.66957259e-01
-1.11582696e+00 9.81269002e-01 2.55886838e-02 -2.97391236e-01
7.64285266e-01 -1.66981936e-01 3.08628939e-02 -2.17488661e-01
1.57353535e-01 8.61781120e-01 1.24426377e+00 -1.75008261e+00
-1.47966951e-01 -7.16739655e-01 -1.74426362e-01 2.30674818e-01
2.03116611e-01 -6.00642920e-01 -2.58138627e-01 -6.92569971e-01
6.81402028e-01 -9.27328646e-01 -3.83473843e-01 4.35820490e-01
-5.31744063e-01 -2.89480776e-01 7.10286856e-01 -2.34852612e-01
4.71521616e-01 -1.93402100e+00 3.50037180e-02 1.35434732e-01
1.44380450e-01 4.96486127e-02 -8.23142081e-02 2.99390376e-01
-8.60678181e-02 4.01501268e-01 -3.52969289e-01 -6.30393445e-01
1.24077082e-01 2.91673541e-01 -7.26778746e-01 4.42170709e-01
6.42501354e-01 9.98059034e-01 -6.62409484e-01 -3.12609613e-01
4.04027313e-01 7.65738904e-01 -5.53247035e-01 4.11049396e-01
-5.13975799e-01 6.15899622e-01 -5.56195378e-01 6.94416463e-01
1.10094857e+00 -3.76947105e-01 -3.58383924e-01 -3.19656134e-01
-1.67598113e-01 2.14437433e-02 -1.23082805e+00 2.02240062e+00
-4.27324593e-01 1.59439608e-01 -1.02710634e-01 -7.65856862e-01
1.38636220e+00 1.67887315e-01 4.28270817e-01 -2.18210146e-01
-2.54136398e-02 3.26035649e-01 -4.59138334e-01 -2.22897187e-01
7.23612368e-01 -3.58571589e-01 -2.28152163e-02 3.58382434e-01
-4.28806022e-02 -9.69536662e-01 -5.20341158e-01 -6.18006140e-02
6.19675457e-01 4.77806807e-01 2.22241580e-01 -1.51265800e-01
3.37809771e-01 -1.54006565e-02 4.30361301e-01 5.22850752e-01
2.33665600e-01 1.38788450e+00 1.97127655e-01 -8.69804144e-01
-1.44839513e+00 -1.44449615e+00 -1.88374758e-01 1.94505677e-01
2.09564388e-01 -5.04470654e-02 -3.78382206e-01 -4.62827384e-01
3.91389519e-01 5.67808151e-01 -5.99076211e-01 8.74668434e-02
-5.81981421e-01 -4.26645070e-01 3.42159599e-01 3.19069415e-01
3.75675499e-01 -1.05934942e+00 -3.88782829e-01 1.72270074e-01
-3.43861133e-02 -1.16484714e+00 -2.62556851e-01 -5.01566455e-02
-1.15635979e+00 -9.25196171e-01 -7.71498680e-01 -6.40324593e-01
9.13393736e-01 3.87392133e-01 1.37146711e+00 -4.72806208e-02
-2.46721908e-01 3.34335268e-01 -1.64762046e-02 -7.45745718e-01
-4.79521215e-01 -4.82420437e-02 9.07838717e-03 1.51790500e-01
5.43912053e-02 -9.02579308e-01 -6.23338640e-01 9.40188318e-02
-1.05562139e+00 2.66295999e-01 6.95453346e-01 6.52029932e-01
1.17821550e+00 -2.19226867e-01 5.62601507e-01 -6.35355830e-01
5.90410411e-01 -4.49784011e-01 -6.12773657e-01 8.85001719e-02
-4.95825857e-01 3.41314554e-01 6.45094216e-01 -1.75150841e-01
-9.43797410e-01 4.28886920e-01 -3.49407345e-01 -9.65206981e-01
-4.11777526e-01 1.06279902e-01 -4.29672897e-02 6.85390905e-02
8.24111819e-01 3.94911855e-01 2.54348189e-01 -7.21255422e-01
3.76870781e-01 2.46650785e-01 6.48682773e-01 -7.32690275e-01
1.15562344e+00 7.66554654e-01 4.12803262e-01 -8.68837357e-01
-7.71668255e-01 -1.09393090e-01 -7.46155620e-01 -1.16545886e-01
8.32540393e-01 -1.02499890e+00 -5.08911908e-01 3.43186349e-01
-1.56221771e+00 7.26697072e-02 -3.77454698e-01 1.33126259e-01
-7.25684881e-01 4.25992459e-01 -5.87268844e-02 -8.41025352e-01
-5.77749491e-01 -1.16867852e+00 1.53645360e+00 -6.83880895e-02
-1.47760466e-01 -6.18737638e-01 -2.81174909e-02 1.82719573e-01
3.36637676e-01 8.99786890e-01 1.02711535e+00 -7.24800229e-02
-1.11976779e+00 -2.10279301e-01 -3.04228634e-01 2.72579581e-01
1.26275152e-01 -1.26935482e-01 -1.06745517e+00 -1.91654786e-01
-3.36101912e-02 -3.77547115e-01 5.85811555e-01 4.45394188e-01
1.28579569e+00 -2.09505245e-01 -9.74164605e-02 8.53021741e-01
1.62123573e+00 -3.65674198e-01 4.78847533e-01 2.20272550e-03
7.41116464e-01 5.33515990e-01 3.95953149e-01 5.45129299e-01
4.51321572e-01 5.80825210e-01 1.05687451e+00 -7.17063472e-02
-1.06396757e-01 -6.77955866e-01 -1.82913974e-01 6.06944382e-01
-2.90802896e-01 -5.86454198e-02 -8.52607369e-01 4.49211478e-01
-1.78090394e+00 -6.57158196e-01 -2.45486349e-01 2.34874320e+00
4.42970991e-01 1.19402073e-01 -1.31350964e-01 5.67879900e-02
5.47068894e-01 1.06538683e-01 -6.81152761e-01 -1.37327716e-01
-2.35164449e-01 2.40838483e-01 3.06137949e-01 2.42376417e-01
-8.43608022e-01 7.70428121e-01 5.87041569e+00 6.84881032e-01
-1.17201924e+00 -2.72783071e-01 6.88127041e-01 2.38660753e-01
-7.36865580e-01 -1.00115880e-01 -7.02759743e-01 2.17591226e-01
1.52785778e-01 4.56302520e-03 2.78947741e-01 9.10432100e-01
2.46431101e-02 1.70732692e-01 -1.14868367e+00 1.30213249e+00
1.87812895e-02 -1.61480689e+00 6.08832300e-01 1.97062269e-01
8.62041950e-01 1.54733837e-01 1.72057524e-01 1.50958383e-02
1.62217796e-01 -1.26457143e+00 1.05538428e+00 7.32849658e-01
1.05750620e+00 -7.31663764e-01 3.52509230e-01 6.28900468e-01
-1.06909811e+00 3.30108047e-01 -5.40198684e-01 1.34137738e-02
1.83942631e-01 6.38396502e-01 -9.79001820e-01 7.22902417e-01
6.68423891e-01 5.12651026e-01 -3.72296602e-01 1.07857132e+00
-8.94491300e-02 -4.68288995e-02 -4.98460382e-01 1.32366151e-01
2.29896098e-01 -2.70893782e-01 5.88004410e-01 7.83814311e-01
6.86600447e-01 1.03594720e-01 1.66128144e-01 1.50339985e+00
-1.08607993e-01 -6.66584074e-02 -1.25764835e+00 3.53136837e-01
4.81685728e-01 1.13023722e+00 -5.32780528e-01 -3.30350697e-02
-1.83652848e-01 6.07340932e-01 4.28884894e-01 1.88687339e-01
-5.01003921e-01 1.50252432e-01 6.65503085e-01 4.18352306e-01
3.26372385e-01 -4.79067504e-01 -6.88615322e-01 -1.01374257e+00
2.55070865e-01 -5.36077023e-01 -5.26733436e-02 -9.73333895e-01
-1.47917044e+00 7.65200913e-01 -4.10655923e-02 -1.61659038e+00
-2.34994739e-01 -5.62112570e-01 -6.30963743e-01 1.03192747e+00
-1.57718313e+00 -1.41806316e+00 -4.86665547e-01 6.42624199e-01
6.00667715e-01 -8.70265216e-02 8.87291431e-01 -4.17535240e-03
2.76241452e-02 2.25359753e-01 -1.66377321e-01 -1.75171420e-01
3.74950588e-01 -1.29098916e+00 9.86271918e-01 5.90631902e-01
2.40626231e-01 4.89353299e-01 5.63909769e-01 -6.43522978e-01
-1.88576066e+00 -1.38235092e+00 5.91900170e-01 -6.74303830e-01
8.02935287e-02 -3.40854973e-01 -8.87117147e-01 5.38150609e-01
-2.37252668e-01 2.42436469e-01 9.13364366e-02 -2.83225298e-01
-2.69830495e-01 5.29947970e-03 -1.26659167e+00 5.60879111e-01
1.18439078e+00 -3.21008176e-01 -6.13994062e-01 2.21555784e-01
8.00052702e-01 -6.88348234e-01 -6.96367383e-01 5.46845496e-01
3.85555089e-01 -1.04578638e+00 1.48820901e+00 -4.20296669e-01
6.69497371e-01 -3.91918212e-01 -2.25544646e-01 -1.25653863e+00
-1.62446201e-01 -5.75738609e-01 5.43256812e-02 9.16054606e-01
2.12757856e-01 -3.21872205e-01 9.70637560e-01 6.42771542e-01
-4.06532735e-01 -9.59430516e-01 -9.73889947e-01 -7.30864704e-01
2.97929019e-01 -6.38384640e-01 1.05321217e+00 6.92876041e-01
-7.90433288e-01 3.57950240e-01 -4.03503776e-01 2.92172104e-01
9.72151697e-01 5.31606376e-01 1.20543194e+00 -1.56646168e+00
6.49216324e-02 -2.94766665e-01 -3.23079616e-01 -1.18564975e+00
3.16813886e-02 -1.03082299e+00 -1.68123525e-02 -1.63208532e+00
-3.14689636e-01 -7.77716041e-01 2.68493235e-01 2.04735845e-01
9.16795135e-02 2.87535608e-01 2.54684210e-01 2.16660723e-01
-2.08721459e-01 9.24394965e-01 1.58856177e+00 -1.05291344e-01
-9.65730622e-02 1.14129476e-01 -5.71602821e-01 8.73533785e-01
6.07751369e-01 -3.21460605e-01 -3.31135392e-01 -8.00582409e-01
2.49329090e-01 1.46473289e-01 7.87472427e-01 -8.99136305e-01
-4.21789177e-02 -1.17561400e-01 6.44549549e-01 -1.14698887e+00
7.53484368e-01 -1.05718160e+00 5.07528722e-01 1.18897371e-01
1.28801197e-01 2.80562758e-01 1.01087190e-01 7.13015437e-01
-1.27096334e-02 3.27250287e-02 6.95801914e-01 -5.31671882e-01
-4.88349050e-01 8.49766910e-01 2.08511978e-01 3.52375619e-02
6.40814304e-01 -5.58514714e-01 1.63864106e-01 -2.25961372e-01
-4.68156576e-01 -1.05671704e-01 6.78497851e-01 5.86808860e-01
1.09571373e+00 -1.62209702e+00 -8.15545261e-01 6.23049438e-01
5.96035868e-02 7.95700371e-01 6.46981820e-02 2.59270906e-01
-5.60041904e-01 1.87025741e-01 -2.71448433e-01 -9.84209836e-01
-7.81578362e-01 4.17093277e-01 2.59996086e-01 6.51820004e-02
-9.42764223e-01 5.91639459e-01 3.63244325e-01 -7.91254818e-01
1.00080401e-01 -6.20006919e-01 1.63289085e-01 -3.01645786e-01
4.18168455e-02 -9.23650414e-02 1.28804460e-01 -7.36299098e-01
-2.48946194e-02 9.28344011e-01 2.69278467e-01 1.38458863e-01
1.59256136e+00 6.73843920e-02 9.80262086e-02 3.52995038e-01
1.13060617e+00 -2.77142912e-01 -1.38576341e+00 -3.10568720e-01
-2.94210792e-01 -7.80119598e-01 -1.97590530e-01 -4.87961084e-01
-9.32885647e-01 9.72534359e-01 4.26281780e-01 3.37411076e-01
6.62939548e-01 4.85058455e-03 7.67028928e-01 3.32481414e-01
6.48832858e-01 -5.86984634e-01 5.74280433e-02 6.70688272e-01
1.47277951e+00 -1.51205683e+00 5.63700646e-02 -4.38753217e-01
-3.28122824e-01 1.21356452e+00 5.33409894e-01 -4.98249382e-01
6.62049055e-01 -5.99349961e-02 -8.91803578e-02 -4.17877197e-01
-4.21810091e-01 4.70633693e-02 5.07745326e-01 7.55049050e-01
2.15374976e-01 1.97698735e-02 1.38065562e-01 4.12106544e-01
-5.13908029e-01 -1.78264216e-01 2.83849031e-01 5.22436023e-01
-2.21512899e-01 -1.00927794e+00 -6.46116376e-01 4.25489336e-01
-2.09246621e-01 1.34185478e-01 -2.75546819e-01 9.04475570e-01
1.08559750e-01 2.83895969e-01 1.62767231e-01 -2.19761461e-01
5.21420658e-01 -9.34026465e-02 5.24131179e-01 -5.72829485e-01
-2.24373609e-01 -4.86017801e-02 -2.37923011e-01 -3.44521493e-01
-2.19864503e-01 -6.36066020e-01 -1.10070729e+00 -2.38809168e-01
-1.34310275e-01 -2.22194970e-01 7.79717088e-01 7.23129153e-01
4.84807014e-01 8.24720785e-02 6.28533423e-01 -1.45553792e+00
-8.09688568e-01 -6.74481690e-01 -5.02740204e-01 6.86950386e-01
5.12741864e-01 -7.64800549e-01 -1.82987764e-01 -1.06679253e-01] | [8.626834869384766, -3.545947313308716] |
dfc82c93-e219-4d02-a6c3-03689a4b696e | p-transformer-towards-better-document-to | 2212.05830 | null | https://arxiv.org/abs/2212.05830v1 | https://arxiv.org/pdf/2212.05830v1.pdf | P-Transformer: Towards Better Document-to-Document Neural Machine Translation | Directly training a document-to-document (Doc2Doc) neural machine translation (NMT) via Transformer from scratch, especially on small datasets usually fails to converge. Our dedicated probing tasks show that 1) both the absolute position and relative position information gets gradually weakened or even vanished once it reaches the upper encoder layers, and 2) the vanishing of absolute position information in encoder output causes the training failure of Doc2Doc NMT. To alleviate this problem, we propose a position-aware Transformer (P-Transformer) to enhance both the absolute and relative position information in both self-attention and cross-attention. Specifically, we integrate absolute positional information, i.e., position embeddings, into the query-key pairs both in self-attention and cross-attention through a simple yet effective addition operation. Moreover, we also integrate relative position encoding in self-attention. The proposed P-Transformer utilizes sinusoidal position encoding and does not require any task-specified position embedding, segment embedding, or attention mechanism. Through the above methods, we build a Doc2Doc NMT model with P-Transformer, which ingests the source document and completely generates the target document in a sequence-to-sequence (seq2seq) way. In addition, P-Transformer can be applied to seq2seq-based document-to-sentence (Doc2Sent) and sentence-to-sentence (Sent2Sent) translation. Extensive experimental results of Doc2Doc NMT show that P-Transformer significantly outperforms strong baselines on widely-used 9 document-level datasets in 7 language pairs, covering small-, middle-, and large-scales, and achieves a new state-of-the-art. Experimentation on discourse phenomena shows that our Doc2Doc NMT models improve the translation quality in both BLEU and discourse coherence. We make our code available on Github. | ['Min Zhang', 'Hao Yang', 'Shimin Tao', 'Jing Jiang', 'Junhui Li', 'Yachao Li'] | 2022-12-12 | null | null | null | null | ['nmt'] | ['computer-code'] | [ 2.65906096e-01 1.46113455e-01 -1.61557764e-01 -2.30125025e-01
-1.25109112e+00 -7.70841599e-01 9.29812193e-01 -3.03530276e-01
-2.77139187e-01 9.14373159e-01 5.44371367e-01 -6.96005285e-01
3.23728502e-01 -6.14579916e-01 -1.08034861e+00 -7.08130836e-01
3.93245369e-01 6.78447008e-01 -1.14337578e-01 -5.54300129e-01
5.82984015e-02 -6.58477768e-02 -8.57903361e-01 5.64834774e-01
1.23901451e+00 5.77205896e-01 5.52322805e-01 7.09716499e-01
-2.06089363e-01 5.36179423e-01 -7.38309205e-01 -5.69587469e-01
2.38969177e-01 -7.52114832e-01 -8.87851596e-01 -3.04985374e-01
4.38592881e-01 -3.77432138e-01 -2.82132059e-01 8.91665161e-01
9.60621536e-01 -1.74983144e-01 5.33026993e-01 -6.19876266e-01
-1.16295600e+00 8.77117038e-01 -5.34479618e-01 1.57142982e-01
6.96578100e-02 3.83282095e-01 1.03268003e+00 -1.21307981e+00
6.15650356e-01 1.35139012e+00 3.99134845e-01 5.90384066e-01
-1.18856001e+00 -4.68554348e-01 2.02539973e-02 2.90407855e-02
-9.88514662e-01 -4.90565181e-01 4.83664393e-01 -2.41518229e-01
1.44068360e+00 3.88532788e-01 3.57300520e-01 1.45868170e+00
3.63863707e-01 9.41421270e-01 1.08520973e+00 -4.38428044e-01
-1.79043174e-01 -3.64872366e-02 -1.42915308e-01 3.87657374e-01
-2.45528877e-01 1.73105627e-01 -4.33924764e-01 2.37204224e-01
6.73029006e-01 -3.08431506e-01 -4.54594076e-01 3.39364380e-01
-1.64216840e+00 6.94923282e-01 3.00352991e-01 4.92546767e-01
-3.17271799e-01 4.28220406e-02 5.72286010e-01 6.04289412e-01
5.44974387e-01 6.73443377e-01 -6.54047847e-01 -4.57121760e-01
-7.41745710e-01 1.26012951e-01 5.52826226e-01 1.16931963e+00
4.83959258e-01 1.25427544e-01 -7.26233602e-01 8.86834800e-01
-2.03332037e-01 8.44635427e-01 9.52936530e-01 -7.01813519e-01
1.13502908e+00 3.61252546e-01 9.63282306e-03 -6.59066856e-01
2.08997317e-02 -7.45606661e-01 -1.08769333e+00 -3.88213426e-01
8.67651179e-02 -3.34297508e-01 -8.13017190e-01 1.97419107e+00
5.50561175e-02 -2.39134029e-01 3.77697855e-01 1.02447379e+00
8.33868682e-01 1.13365424e+00 -4.13653344e-01 -4.95182604e-01
1.19525981e+00 -1.46592283e+00 -8.90851080e-01 -2.58472294e-01
8.72798800e-01 -9.86795545e-01 1.38946676e+00 -8.09025019e-02
-1.39126682e+00 -7.92613268e-01 -9.54598486e-01 -5.06088316e-01
-1.94982335e-01 4.99826968e-01 1.35037586e-01 9.99557152e-02
-1.05223382e+00 6.97240233e-01 -6.93367183e-01 -6.51068464e-02
-9.82624129e-04 2.21552521e-01 -3.38540763e-01 -1.48876905e-01
-1.59402871e+00 9.37675595e-01 1.49519593e-01 3.41706276e-01
-7.15044975e-01 -6.79609716e-01 -7.20797420e-01 1.08620018e-01
-3.35149504e-02 -9.34572160e-01 1.36250222e+00 -8.77690673e-01
-1.90477347e+00 5.43786943e-01 -4.21692580e-01 -3.12646687e-01
6.15689039e-01 -4.15220559e-01 -1.71764269e-01 -5.48647679e-02
2.01962963e-01 7.52262950e-01 6.52459204e-01 -7.81438351e-01
-2.06275821e-01 -2.02876747e-01 -2.13044509e-02 5.62596679e-01
-4.64546531e-01 -4.19811867e-02 -3.86422247e-01 -7.74912417e-01
-6.45157043e-03 -8.28209817e-01 1.88315272e-01 -5.64490497e-01
-6.79863513e-01 -3.77922177e-01 6.30669057e-01 -9.05816495e-01
1.43906236e+00 -1.95418429e+00 6.48161829e-01 -5.42053759e-01
-1.34808555e-01 4.51612055e-01 -5.58654368e-01 8.65656674e-01
-7.13485805e-03 2.23644793e-01 -2.98402309e-01 -4.71107960e-01
-4.67225388e-02 2.27553606e-01 -5.63761353e-01 4.26642224e-02
5.29316425e-01 1.43285847e+00 -1.06542742e+00 -3.72325093e-01
-1.36726558e-01 4.53222215e-01 -5.23964405e-01 3.34088981e-01
-2.65944570e-01 4.82560009e-01 -2.49507323e-01 2.51803547e-01
5.53262651e-01 -3.09318990e-01 2.88705468e-01 -5.32656573e-02
-2.82919675e-01 1.06943321e+00 -2.70782560e-01 1.86802101e+00
-6.77522838e-01 8.29848230e-01 -3.09430301e-01 -8.61124635e-01
8.14971745e-01 5.46177626e-01 1.74444597e-02 -9.80943680e-01
3.57932374e-02 5.43112576e-01 2.42119193e-01 -4.78092104e-01
8.05639803e-01 -1.19396791e-01 9.18948557e-03 5.57501674e-01
1.49697766e-01 4.80844080e-02 2.99299389e-01 1.89709395e-01
9.06242013e-01 2.29732811e-01 -1.72272213e-02 -2.41007417e-01
4.34042841e-01 -9.12102014e-02 3.64110172e-01 4.32429284e-01
2.55492598e-01 8.60059381e-01 6.55353069e-01 3.78362387e-02
-1.25578713e+00 -8.44476581e-01 6.15583397e-02 9.81199265e-01
-1.30753055e-01 -4.49806243e-01 -9.09977913e-01 -8.46601784e-01
-2.27087393e-01 8.50154459e-01 -5.49227297e-01 -1.67536467e-01
-1.10052288e+00 -7.11699307e-01 6.24114215e-01 5.64057052e-01
4.81930345e-01 -9.92180288e-01 -2.86906734e-02 4.37525898e-01
-6.65859878e-01 -1.03311586e+00 -1.20935464e+00 2.48508081e-01
-1.01187289e+00 -4.69001561e-01 -1.06083715e+00 -8.54565024e-01
5.85792184e-01 3.42506140e-01 1.02941060e+00 -1.49088174e-01
4.83953536e-01 -3.38961691e-01 -3.49384874e-01 1.33272912e-02
-5.63412786e-01 4.45965350e-01 -5.81927635e-02 -2.84440458e-01
1.56211779e-01 -5.53363025e-01 -5.71232557e-01 3.05821031e-01
-5.67409515e-01 4.01936084e-01 9.95391846e-01 1.33069146e+00
4.42702919e-01 -3.84817064e-01 6.75435722e-01 -6.28473341e-01
8.98988187e-01 -3.33883315e-01 -2.56844640e-01 1.91062883e-01
-5.33996642e-01 2.32886091e-01 1.02391529e+00 -5.64568102e-01
-9.02462840e-01 -4.99315262e-01 -2.65864879e-01 -3.92333537e-01
3.97353798e-01 7.01499522e-01 -3.18198502e-01 6.95424497e-01
5.90664029e-01 7.85271227e-01 6.34125620e-03 -6.43710434e-01
3.83940041e-01 8.93991053e-01 4.75511760e-01 -5.71565628e-01
7.21844435e-01 -1.23704962e-01 -5.31032681e-01 -3.85703355e-01
-7.21946895e-01 -3.95037746e-03 -6.00599527e-01 3.08426708e-01
6.90975070e-01 -9.59544003e-01 -3.53844285e-01 4.48528945e-01
-1.69919193e+00 -5.09914100e-01 -9.32287872e-02 4.62434858e-01
-3.71795744e-01 2.11635455e-01 -8.76981258e-01 -4.02816445e-01
-8.05333972e-01 -1.32128537e+00 1.33304906e+00 -9.40393582e-02
-1.13999888e-01 -8.81264746e-01 9.60652530e-02 4.00417775e-01
4.30342585e-01 -2.22188145e-01 1.06792808e+00 -5.60638964e-01
-6.30767047e-01 7.10594803e-02 -3.22212130e-01 6.75436616e-01
1.44666672e-01 -2.05152363e-01 -6.88004434e-01 -4.22963738e-01
-8.48048180e-02 -2.95601934e-01 8.60757411e-01 2.01821908e-01
8.24026704e-01 -7.39900291e-01 -4.80181724e-02 4.76085097e-01
1.08843374e+00 5.77449799e-02 7.62571990e-01 1.25260815e-01
6.45267427e-01 4.30262506e-01 5.22748113e-01 -8.22264552e-02
4.89407122e-01 9.51495886e-01 1.55163314e-02 -9.18781608e-02
-5.46404839e-01 -5.69915354e-01 9.21402276e-01 1.65620172e+00
-9.11922529e-02 -6.27246857e-01 -5.69273174e-01 6.22167587e-01
-1.76991570e+00 -8.81769776e-01 -1.82226643e-01 2.10920572e+00
1.50282300e+00 1.81297019e-01 -3.03286284e-01 -1.57429725e-01
6.60245419e-01 1.80056304e-01 -3.89470369e-01 -6.94103956e-01
-3.79192799e-01 8.50501135e-02 2.30604783e-01 7.04793096e-01
-6.31134212e-01 1.12722778e+00 5.31956911e+00 1.02204394e+00
-1.46460998e+00 4.49684352e-01 5.31044960e-01 -2.08863303e-01
-5.58014154e-01 -1.64502397e-01 -8.28199148e-01 7.92532265e-01
1.14694202e+00 -2.15650067e-01 4.73876327e-01 5.52038848e-01
3.44900012e-01 4.96936232e-01 -1.21983612e+00 6.17965698e-01
-6.60810471e-02 -1.35976481e+00 2.46302456e-01 1.47752568e-01
8.81481707e-01 2.17928827e-01 1.82664365e-01 6.77375674e-01
6.27771914e-02 -1.02778196e+00 8.43749583e-01 1.77568346e-01
1.23225439e+00 -4.90671039e-01 8.78554642e-01 4.65232998e-01
-8.21369171e-01 1.08140655e-01 -4.59196001e-01 -7.52147660e-02
3.29742879e-01 5.48380017e-01 -1.04396260e+00 8.96684885e-01
2.04916552e-01 7.53910005e-01 -1.14306547e-01 3.19751292e-01
-5.66024303e-01 7.38101840e-01 -1.15288816e-01 -2.07967356e-01
5.50565660e-01 -1.32037997e-01 6.06186032e-01 1.26656258e+00
5.95881760e-01 2.77098129e-03 -3.24873745e-01 8.50592315e-01
-3.45858455e-01 1.04889214e-01 -3.39004040e-01 -3.25506032e-01
5.43109357e-01 8.81276369e-01 7.10638985e-02 -5.00057817e-01
-1.43228173e-01 1.39145732e+00 4.73264694e-01 5.34960985e-01
-7.83751249e-01 -5.15188813e-01 7.57131696e-01 -3.73733826e-02
5.52278280e-01 -2.70929962e-01 -2.82412052e-01 -1.30035925e+00
3.40305239e-01 -1.21683264e+00 -2.12637618e-01 -7.75885642e-01
-1.04222167e+00 9.91756678e-01 -2.62807637e-01 -1.30036211e+00
-5.21627128e-01 -3.80413443e-01 -6.02401257e-01 1.23674119e+00
-1.71312487e+00 -1.21234357e+00 2.30677322e-01 1.87825665e-01
8.93090427e-01 -1.70227736e-01 8.55964065e-01 4.43448454e-01
-6.74412131e-01 9.85723257e-01 4.57222730e-01 1.65028393e-01
8.82846534e-01 -1.19316173e+00 8.59855115e-01 7.84855723e-01
5.11403307e-02 9.19953644e-01 4.84249890e-01 -5.59523582e-01
-1.48977900e+00 -1.30151975e+00 1.54287922e+00 -6.28578544e-01
4.44658041e-01 -5.82759738e-01 -8.74273241e-01 5.45661509e-01
6.29924893e-01 -3.20964009e-01 2.12285146e-01 -9.58731622e-02
-3.34107995e-01 -4.76124063e-02 -5.06191134e-01 6.58444285e-01
1.11301112e+00 -7.05577016e-01 -6.81265950e-01 5.07292926e-01
1.39478147e+00 -8.11623454e-01 -7.72091925e-01 3.38755906e-01
4.34623927e-01 -6.77143455e-01 6.56137526e-01 -6.46458030e-01
1.07789528e+00 -2.22511947e-01 -1.75387576e-01 -1.67681324e+00
-3.93467039e-01 -1.00289416e+00 -1.76925346e-01 1.12504303e+00
8.43002856e-01 -6.26619756e-01 1.77800521e-01 -2.29590908e-01
-8.50432098e-01 -1.12368739e+00 -1.12669504e+00 -9.54760075e-01
6.96454883e-01 1.41492542e-02 7.96691835e-01 8.39893758e-01
1.55184835e-01 8.77729893e-01 -5.59020936e-01 -1.03028141e-01
6.82604536e-02 2.19267279e-01 6.93393469e-01 -6.07821584e-01
-4.69784915e-01 -4.57929999e-01 1.91941604e-01 -1.88202322e+00
7.49940285e-03 -1.09826159e+00 1.00402519e-01 -1.59212399e+00
2.51412123e-01 -1.87381670e-01 -1.05015688e-01 4.69665766e-01
-4.79941964e-01 1.70216441e-01 1.26159891e-01 4.21675205e-01
-9.11210552e-02 9.74923730e-01 1.84901464e+00 -2.24641964e-01
-9.95493978e-02 -3.40369463e-01 -7.30785608e-01 -5.77090830e-02
6.24675155e-01 -3.89716238e-01 -3.43732506e-01 -1.05830646e+00
1.57557040e-01 1.99025065e-01 9.17894021e-02 -4.83722955e-01
3.16268094e-02 2.28360854e-02 1.29986390e-01 -6.64882064e-01
2.48801008e-01 -2.33238161e-01 -2.11516142e-01 3.77179503e-01
-5.26097953e-01 3.61236006e-01 1.29291937e-01 2.43403107e-01
-3.89474720e-01 -2.41114385e-02 4.02701408e-01 -2.99472897e-03
6.20005950e-02 1.06615603e-01 -5.97780235e-02 2.02712581e-01
3.66251290e-01 1.03846543e-01 -7.13412702e-01 -3.54817986e-01
-1.58625990e-01 3.03056598e-01 3.01606297e-01 6.99110448e-01
3.63163680e-01 -1.56756651e+00 -1.13688505e+00 1.77995056e-01
-5.36493547e-02 -9.90609452e-03 2.01955408e-01 1.20430124e+00
-7.95670673e-02 9.60450888e-01 9.02726129e-02 -6.87807441e-01
-1.00091588e+00 3.14994991e-01 2.70227790e-01 -5.76258183e-01
-5.26171327e-01 9.34755027e-01 3.82070899e-01 -8.12967002e-01
-9.00347158e-02 -4.08760548e-01 1.51920542e-01 -5.30651100e-02
4.88583684e-01 4.57862839e-02 2.15906888e-01 -5.10290325e-01
-1.20904393e-01 4.11260784e-01 -4.02979612e-01 -2.64713824e-01
1.15583038e+00 -2.04086870e-01 -1.95432127e-01 3.07766765e-01
1.53075278e+00 1.96762988e-03 -1.04845071e+00 -2.45188013e-01
-4.38092887e-01 -2.73593247e-01 -6.84972629e-02 -1.04086244e+00
-8.84377539e-01 1.33539963e+00 1.81982592e-01 -5.70826344e-02
9.58740771e-01 -2.12151498e-01 1.33956492e+00 3.60949785e-01
-7.54541624e-03 -9.72421288e-01 1.69982910e-01 9.09177959e-01
1.30532491e+00 -1.13903558e+00 -3.65649730e-01 -3.98896039e-02
-6.63617671e-01 1.11839640e+00 6.19934142e-01 3.48336011e-01
1.81755163e-02 1.47521153e-01 1.20392255e-01 2.55282015e-01
-1.20916533e+00 1.77856952e-01 2.77242243e-01 2.98784137e-01
8.54589999e-01 2.98701663e-04 -6.61510348e-01 6.16490483e-01
-5.96749961e-01 -1.79532215e-01 4.26775634e-01 5.34704387e-01
-1.59694254e-01 -1.33604085e+00 -1.76488012e-02 8.55219364e-02
-3.87403071e-01 -6.73460424e-01 -3.59245390e-01 5.92939615e-01
-1.16402090e-01 7.45808244e-01 6.94723353e-02 -5.61445117e-01
3.17272305e-01 9.16117057e-02 3.83456856e-01 -5.44432938e-01
-7.39696264e-01 2.80322641e-01 3.14078212e-01 -3.29373956e-01
-8.03098176e-03 -5.90535402e-01 -1.15500975e+00 -3.82560968e-01
-4.74176288e-01 2.89687067e-01 6.80424571e-01 8.36950660e-01
8.14183116e-01 8.02877367e-01 8.13651502e-01 -6.78468168e-01
-1.02654898e+00 -1.67773843e+00 -4.82028211e-03 8.14104676e-02
5.52641809e-01 -1.23503484e-01 -3.30224633e-01 -2.84562744e-02] | [11.724536895751953, 9.85677433013916] |
521e3808-6550-425a-b210-7c2357f290b2 | text-alignment-is-an-efficient-unified-model | 2307.02729 | null | https://arxiv.org/abs/2307.02729v1 | https://arxiv.org/pdf/2307.02729v1.pdf | Text Alignment Is An Efficient Unified Model for Massive NLP Tasks | Large language models (LLMs), typically designed as a function of next-word prediction, have excelled across extensive NLP tasks. Despite the generality, next-word prediction is often not an efficient formulation for many of the tasks, demanding an extreme scale of model parameters (10s or 100s of billions) and sometimes yielding suboptimal performance. In practice, it is often desirable to build more efficient models -- despite being less versatile, they still apply to a substantial subset of problems, delivering on par or even superior performance with much smaller model sizes. In this paper, we propose text alignment as an efficient unified model for a wide range of crucial tasks involving text entailment, similarity, question answering (and answerability), factual consistency, and so forth. Given a pair of texts, the model measures the degree of alignment between their information. We instantiate an alignment model (Align) through lightweight finetuning of RoBERTa (355M parameters) using 5.9M examples from 28 datasets. Despite its compact size, extensive experiments show the model's efficiency and strong performance: (1) On over 20 datasets of aforementioned diverse tasks, the model matches or surpasses FLAN-T5 models that have around 2x or 10x more parameters; the single unified model also outperforms task-specific models finetuned on individual datasets; (2) When applied to evaluate factual consistency of language generation on 23 datasets, our model improves over various baselines, including the much larger GPT-3.5 (ChatGPT) and sometimes even GPT-4; (3) The lightweight model can also serve as an add-on component for LLMs such as GPT-3.5 in question answering tasks, improving the average exact match (EM) score by 17.94 and F1 score by 15.05 through identifying unanswerable questions. | ['Zhiting Hu', 'Ruichen Li', 'Yichi Yang', 'Yuheng Zha'] | 2023-07-06 | null | null | null | null | ['text-generation', 'question-answering'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.91192076e-01 9.20546576e-02 -2.53143638e-01 -3.98392320e-01
-1.43744671e+00 -7.47836113e-01 8.08225453e-01 2.36406952e-01
-4.64419365e-01 6.88602805e-01 3.49265069e-01 -8.02812040e-01
-2.40038425e-01 -5.09897292e-01 -7.24153996e-01 -1.80278033e-01
4.08091366e-01 9.03832197e-01 1.83032617e-01 -5.59305906e-01
4.08202380e-01 -2.40301311e-01 -1.13236618e+00 6.33822739e-01
1.19487035e+00 8.31289947e-01 3.11653107e-01 7.80084133e-01
-5.12150109e-01 8.10682893e-01 -6.15927935e-01 -9.69663322e-01
2.09259037e-02 -3.39894235e-01 -1.35236466e+00 -3.80965233e-01
9.27222490e-01 -4.40758839e-02 1.20534368e-01 6.88826263e-01
5.30453146e-01 1.58010751e-01 4.73118782e-01 -1.12907457e+00
-7.98292100e-01 9.92717624e-01 -3.06602448e-01 1.86255306e-01
6.73782885e-01 8.76686797e-02 1.62717474e+00 -8.69972110e-01
4.18888390e-01 1.35467350e+00 8.56905520e-01 5.37032545e-01
-1.14347005e+00 -5.46737611e-01 1.75686687e-01 2.60184735e-01
-9.85980332e-01 -4.39062774e-01 7.14481995e-02 -2.32889056e-02
1.64188933e+00 6.35435104e-01 8.41225386e-02 1.24861240e+00
2.26127490e-01 9.53427970e-01 9.61773098e-01 -5.29153049e-01
-2.40298823e-01 7.70633817e-02 3.42853844e-01 5.19787550e-01
8.62084925e-02 -3.25548828e-01 -5.36225379e-01 -4.45159107e-01
-4.31931466e-02 -4.33793277e-01 -1.83937624e-01 5.40777266e-01
-1.38780248e+00 9.51611161e-01 8.29506963e-02 4.23272461e-01
-4.69957441e-02 4.92435805e-02 3.53265017e-01 5.88459969e-01
4.51700658e-01 8.66235375e-01 -8.12962413e-01 -2.82637239e-01
-9.77323711e-01 8.19363058e-01 1.12412751e+00 1.04464269e+00
5.28245389e-01 -2.18700573e-01 -5.35445750e-01 1.06689131e+00
1.76368449e-02 4.44666088e-01 9.27424133e-01 -9.28783059e-01
1.02099001e+00 5.60554862e-01 -1.59224235e-02 -7.73270428e-01
-4.46165234e-01 -4.04086500e-01 -7.56363809e-01 -6.75929666e-01
6.31463587e-01 -1.85487997e-02 -6.94560826e-01 1.99600399e+00
1.90019161e-01 -1.97408199e-01 -1.28178999e-01 4.03507859e-01
7.69871473e-01 8.79464507e-01 1.94772497e-01 -1.05793096e-01
1.58010840e+00 -1.15192187e+00 -6.10770702e-01 -7.72913456e-01
1.05302286e+00 -1.26381636e+00 1.53978872e+00 2.74091601e-01
-1.30166614e+00 -6.84709489e-01 -6.12565577e-01 -4.32153553e-01
-3.73040974e-01 -1.43602133e-01 5.78757703e-01 4.83451515e-01
-1.14610505e+00 4.83627468e-01 -1.61705181e-01 -5.68183303e-01
-1.29637256e-01 1.39908135e-01 -1.28511563e-01 -2.49538496e-01
-1.55440116e+00 1.31057441e+00 2.60540485e-01 -2.39135638e-01
-3.51868868e-01 -1.00356376e+00 -7.92303145e-01 1.62724759e-02
2.44356006e-01 -1.18387544e+00 1.55752933e+00 -3.86046320e-01
-1.39555633e+00 8.90187144e-01 -4.88041162e-01 -6.70221210e-01
4.58533019e-01 -4.92506474e-01 -3.52382481e-01 -3.21063131e-01
2.84362078e-01 7.26871789e-01 5.62125146e-01 -6.68121397e-01
-6.81028366e-01 -6.49585873e-02 1.20301284e-01 2.94040859e-01
-2.80206144e-01 2.47454643e-01 -1.98779732e-01 -6.81925833e-01
2.84107607e-02 -8.27948570e-01 -1.09670766e-01 -6.94128752e-01
-4.93752509e-01 -7.81643331e-01 4.66445476e-01 -7.90025413e-01
1.64252329e+00 -1.66143620e+00 -2.91426871e-02 -2.16545880e-01
-1.18838415e-01 3.88451993e-01 -6.90361202e-01 6.72360957e-01
-1.11588568e-03 3.14081252e-01 -3.46991628e-01 -5.73933065e-01
2.95491576e-01 2.82059014e-01 -6.91681564e-01 -6.28833547e-02
1.77126616e-01 1.38507426e+00 -7.81086922e-01 -5.50929785e-01
-9.36330855e-03 -4.86492515e-02 -5.86476922e-01 1.70271203e-01
-5.23972094e-01 7.45768100e-03 -2.57941216e-01 4.16811943e-01
3.29468638e-01 -5.14741480e-01 7.91416243e-02 1.98525488e-01
2.13226676e-01 1.05391657e+00 -9.48154509e-01 1.64984477e+00
-7.77132571e-01 4.53820556e-01 -2.96307355e-01 -8.52401793e-01
7.59275734e-01 4.38780993e-01 1.53481901e-01 -8.25877488e-01
-2.13384122e-01 3.20853233e-01 1.95551872e-01 -6.96139038e-01
8.11135828e-01 -1.16183512e-01 -2.69187629e-01 7.81564236e-01
5.50208054e-02 -5.14401853e-01 4.55202878e-01 3.64527553e-01
1.16600561e+00 -1.55989945e-01 3.87017459e-01 -2.89879292e-01
6.42072141e-01 2.95090843e-02 1.63299248e-01 1.03662980e+00
1.12139478e-01 4.27761793e-01 3.55589956e-01 -3.29822481e-01
-9.85613644e-01 -7.81036854e-01 -2.04681709e-01 1.56420946e+00
-3.41235191e-01 -7.51391709e-01 -7.43329227e-01 -6.85340941e-01
5.12199439e-02 1.22320998e+00 -3.12723249e-01 -4.68599200e-02
-9.92652655e-01 -9.32401717e-01 9.09906745e-01 3.84241998e-01
3.68333131e-01 -1.03118169e+00 -2.61400566e-02 2.68391967e-01
-9.73463893e-01 -1.30682075e+00 -6.82904124e-01 -3.68524306e-02
-7.82012641e-01 -8.06405067e-01 -4.28180248e-01 -7.94721007e-01
2.09029198e-01 2.48985291e-01 1.73330998e+00 1.95992604e-01
8.19613039e-02 1.76055744e-01 -3.20032448e-01 -3.56994867e-01
-7.49477804e-01 5.43204188e-01 -4.40987013e-02 -3.95193428e-01
4.66008663e-01 -5.36459446e-01 -3.42763990e-01 3.73912454e-01
-8.51579785e-01 -8.59761834e-02 5.56193590e-01 8.60902011e-01
3.32842857e-01 -4.07456726e-01 8.28530192e-01 -1.02112520e+00
1.07392704e+00 -5.45904696e-01 -2.30160311e-01 6.35261297e-01
-8.44903827e-01 1.31700933e-01 8.58700931e-01 -4.54288363e-01
-9.84345734e-01 -6.57515168e-01 -3.95799637e-01 2.57209241e-01
5.49064428e-02 4.87182766e-01 -3.01759876e-02 3.37136567e-01
8.08157623e-01 2.50340849e-01 -2.35127598e-01 -4.76842284e-01
6.10591352e-01 6.14927828e-01 4.59662408e-01 -7.96023846e-01
8.43290865e-01 -4.06336375e-02 -3.93971145e-01 -4.89944130e-01
-1.36745024e+00 -5.30549705e-01 -2.26000980e-01 4.06786382e-01
5.23849845e-01 -7.54117489e-01 -6.38503134e-01 1.78125143e-01
-1.39737213e+00 -4.00280058e-01 -5.40961232e-03 2.14948580e-01
-3.32249790e-01 6.41163290e-01 -7.69680202e-01 -5.45286536e-01
-8.26825678e-01 -8.29370916e-01 1.15951157e+00 -1.37422279e-01
-9.78514194e-01 -1.28863716e+00 1.60773590e-01 8.78111541e-01
5.71777105e-01 -2.88204998e-01 1.49340236e+00 -1.05696535e+00
-2.15869024e-01 -2.37255901e-01 -5.36520481e-02 4.27051723e-01
-6.78137615e-02 -1.97132215e-01 -7.87489533e-01 -5.35136126e-02
1.04603224e-01 -5.17333806e-01 8.16502452e-01 2.76473254e-01
1.19482434e+00 -6.16764307e-01 -1.43200457e-01 2.64846802e-01
1.07311416e+00 -2.95963168e-01 5.46664059e-01 3.45804662e-01
4.28474098e-01 7.75052369e-01 6.12947226e-01 8.13912302e-02
7.36625612e-01 6.60988808e-01 1.68934450e-01 3.55181396e-01
-2.73588538e-01 -5.40790677e-01 6.18761003e-01 1.21831095e+00
1.87369019e-01 -5.31053007e-01 -8.83336663e-01 5.69509268e-01
-1.84656525e+00 -1.02294159e+00 -6.08668268e-01 2.11744404e+00
1.25473130e+00 8.46857950e-02 -9.28040519e-02 -7.98906609e-02
4.51822072e-01 2.82064885e-01 -3.05558026e-01 -6.92928791e-01
-2.81974405e-01 5.25188267e-01 3.09390072e-02 8.08507979e-01
-7.22427011e-01 1.08888817e+00 6.89898300e+00 1.28229594e+00
-5.89346111e-01 1.96391985e-01 5.96555650e-01 -1.14934616e-01
-6.01907492e-01 8.63635167e-02 -1.21047390e+00 4.96085256e-01
1.14189434e+00 -3.10948879e-01 4.60234731e-01 7.32782364e-01
-1.58622935e-01 -1.07457660e-01 -1.32190502e+00 7.47424483e-01
1.69282407e-01 -1.15496600e+00 3.69592905e-01 -1.65243819e-01
9.27542329e-01 1.08260512e-01 9.41477790e-02 7.76406527e-01
4.78931218e-01 -1.28985679e+00 6.45064056e-01 5.89059703e-02
4.74962980e-01 -5.01585662e-01 7.91055679e-01 8.36299300e-01
-8.83131921e-01 -5.00080734e-02 -4.89514560e-01 -2.73205489e-01
2.97852308e-01 5.59880555e-01 -8.61854076e-01 5.14464617e-01
4.99624312e-01 1.94206730e-01 -6.68591917e-01 5.76090217e-01
-4.37867373e-01 7.66881108e-01 -3.42810482e-01 -2.88426846e-01
5.07989824e-01 3.93570811e-02 4.61076379e-01 1.47277105e+00
3.37461233e-01 -1.16053373e-01 -1.90385729e-01 7.14964688e-01
-2.46014491e-01 3.10588300e-01 -3.24129701e-01 1.53202787e-01
6.51536822e-01 1.41499472e+00 6.38449891e-03 -5.15592039e-01
-3.13419878e-01 9.12379503e-01 5.54575443e-01 1.35399684e-01
-1.00686479e+00 -1.89894348e-01 5.95892131e-01 -4.60379906e-02
1.21664248e-01 3.29702459e-02 -4.16808218e-01 -1.00906157e+00
3.94041002e-01 -1.41168213e+00 7.42334127e-01 -6.94896519e-01
-1.74448919e+00 6.96272552e-01 -1.81338023e-02 -7.36295462e-01
-7.05980837e-01 -5.93048573e-01 -6.98432267e-01 1.02675557e+00
-1.61158061e+00 -1.13647914e+00 -2.58656759e-02 4.88122761e-01
7.76305854e-01 -4.30253260e-02 9.44504142e-01 2.69912571e-01
-4.53463376e-01 9.27619278e-01 -3.69075313e-02 -7.39355609e-02
1.00881541e+00 -1.35441434e+00 9.02487993e-01 7.42853522e-01
5.35451055e-01 8.33528280e-01 5.61791778e-01 -4.38325942e-01
-1.16150415e+00 -1.11628604e+00 1.96901441e+00 -1.11120844e+00
9.77766633e-01 -2.49083430e-01 -1.05401528e+00 7.78469563e-01
3.90612960e-01 -5.19904494e-01 6.59735978e-01 4.22377944e-01
-5.75088799e-01 2.89500691e-02 -7.46038079e-01 7.17403591e-01
1.07070220e+00 -6.32521749e-01 -9.92504120e-01 8.27056944e-01
1.00708091e+00 -5.17188668e-01 -8.01895440e-01 3.21738631e-01
4.22479123e-01 -7.78646410e-01 8.98551822e-01 -1.05645812e+00
7.32819617e-01 1.37662113e-01 -3.01428825e-01 -1.20477259e+00
-3.78289074e-01 -1.02050018e+00 -1.20669991e-01 1.41058958e+00
8.48489285e-01 -7.46658385e-01 3.91025722e-01 7.68010437e-01
-1.81185365e-01 -9.08598661e-01 -1.03166604e+00 -8.59244525e-01
6.14690363e-01 -7.77092338e-01 6.86347544e-01 9.12320614e-01
9.48864594e-02 8.06149960e-01 -1.67036816e-01 -1.38450518e-01
2.94570416e-01 2.05741391e-01 7.44726717e-01 -7.65667975e-01
-4.77292776e-01 -7.04966843e-01 3.64877433e-01 -1.40134656e+00
4.14443463e-01 -1.21566546e+00 -9.91554335e-02 -1.53274751e+00
2.85784364e-01 -4.78813201e-01 -1.05830468e-02 4.90943968e-01
-7.40956187e-01 3.99784232e-03 2.26756558e-02 7.91366994e-02
-5.70990682e-01 2.91882634e-01 1.01833224e+00 -2.13400908e-02
1.10671386e-01 1.09529510e-01 -9.78336811e-01 6.08437777e-01
8.68907094e-01 -3.67920965e-01 -3.45338315e-01 -9.27587271e-01
6.34718359e-01 -5.31337149e-02 2.90271789e-01 -5.74605346e-01
3.19090933e-01 -3.80708203e-02 -2.63691507e-03 -4.95652795e-01
1.16013497e-01 -3.77527148e-01 -1.39828950e-01 3.22746098e-01
-6.03233397e-01 4.52910811e-01 1.88104287e-01 2.73045838e-01
-2.01320246e-01 -5.14048338e-01 3.59524429e-01 -2.42966682e-01
-4.06982660e-01 1.64398924e-01 -7.93617591e-02 7.25307882e-01
4.42031592e-01 -1.44541683e-02 -6.09589815e-01 -4.55126464e-01
-1.58931807e-01 4.00624305e-01 -6.27844259e-02 6.82363391e-01
3.34717661e-01 -1.20206678e+00 -1.11902249e+00 -1.39555201e-01
1.86698601e-01 5.49009144e-02 2.43355691e-01 9.37729716e-01
-2.84240600e-02 9.44329202e-01 4.83524323e-01 -4.79788691e-01
-1.18958664e+00 2.44365498e-01 2.31066987e-01 -1.05552459e+00
-1.72265738e-01 1.10041082e+00 1.46754339e-01 -7.89972186e-01
8.84359106e-02 -5.37465692e-01 1.02739446e-01 -1.74498148e-02
6.53821290e-01 3.82841498e-01 4.06916112e-01 -2.28101492e-01
-2.56343424e-01 4.71815825e-01 -3.41886133e-01 -2.39378363e-02
1.12506938e+00 -1.45476326e-01 -4.19963181e-01 1.70500159e-01
1.10491395e+00 1.36763686e-02 -5.82619190e-01 -4.65242058e-01
3.34444284e-01 -1.20514847e-01 -4.48824555e-01 -1.08114564e+00
-3.91290784e-01 8.63004148e-01 -2.57233024e-01 2.23142385e-01
8.17356169e-01 1.28109634e-01 1.36764228e+00 6.48689806e-01
1.81332275e-01 -9.79549885e-01 1.60427719e-01 8.89078617e-01
9.74095225e-01 -1.28466034e+00 -1.16025999e-01 -1.27652064e-01
-6.32193208e-01 9.03231859e-01 6.27139151e-01 3.33169520e-01
7.86076114e-02 1.41497254e-01 -7.52207488e-02 -3.38931754e-02
-1.31207943e+00 2.83639431e-02 5.88446975e-01 2.89033264e-01
7.44506836e-01 1.40980561e-03 -3.68364453e-01 7.01679885e-01
-8.46554756e-01 -4.17532653e-01 3.23563069e-02 3.42074752e-01
-4.71467704e-01 -1.29459596e+00 -1.98326871e-01 4.73350614e-01
-6.14294410e-01 -6.93435252e-01 -4.03122991e-01 8.92236948e-01
-3.64795625e-02 1.29131138e+00 -7.28055611e-02 -2.58638293e-01
2.82314360e-01 4.34196323e-01 4.68647659e-01 -5.45386016e-01
-1.03811061e+00 -4.49517906e-01 4.86360461e-01 -5.29149115e-01
-8.38056952e-02 -5.05257189e-01 -9.12953258e-01 -7.13022053e-01
-3.54341567e-01 1.57193109e-01 4.64903474e-01 1.17632735e+00
3.95783424e-01 7.58704916e-02 3.38595390e-01 2.45860778e-03
-1.17028224e+00 -1.41132939e+00 -1.23265974e-01 4.70955580e-01
-4.28203568e-02 -1.20131625e-02 -3.41330141e-01 -4.67457846e-02] | [11.094555854797363, 8.494844436645508] |
85a86665-88ca-4922-ba71-1cb2c306dcb5 | federated-learning-as-variational-inference-a | 2302.04228 | null | https://arxiv.org/abs/2302.04228v1 | https://arxiv.org/pdf/2302.04228v1.pdf | Federated Learning as Variational Inference: A Scalable Expectation Propagation Approach | The canonical formulation of federated learning treats it as a distributed optimization problem where the model parameters are optimized against a global loss function that decomposes across client loss functions. A recent alternative formulation instead treats federated learning as a distributed inference problem, where the goal is to infer a global posterior from partitioned client data (Al-Shedivat et al., 2021). This paper extends the inference view and describes a variational inference formulation of federated learning where the goal is to find a global variational posterior that well-approximates the true posterior. This naturally motivates an expectation propagation approach to federated learning (FedEP), where approximations to the global posterior are iteratively refined through probabilistic message-passing between the central server and the clients. We conduct an extensive empirical study across various algorithmic considerations and describe practical strategies for scaling up expectation propagation to the modern federated setting. We apply FedEP on standard federated learning benchmarks and find that it outperforms strong baselines in terms of both convergence speed and accuracy. | ['Eric P. Xing', 'Yoon Kim', 'Andrew Gelman', 'Hongyi Wang', 'Philip Greengard', 'Han Guo'] | 2023-02-08 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [-6.56808019e-01 1.39818251e-01 -7.27008209e-02 -6.81317210e-01
-1.65389538e+00 -4.71281081e-01 6.67631686e-01 -4.29300033e-02
-3.10103685e-01 7.56355464e-01 4.49457526e-01 -1.78119496e-01
-1.99546665e-01 -6.49528027e-01 -1.05614400e+00 -9.76634979e-01
-1.22076176e-01 8.41439068e-01 -2.89844751e-01 5.25770903e-01
-2.69413382e-01 1.34084195e-01 -1.31136489e+00 4.19857264e-01
2.72691518e-01 8.53414595e-01 -1.54610321e-01 8.09878767e-01
-3.47604841e-01 1.16869915e+00 -2.58535177e-01 -8.91890645e-01
1.44360691e-01 -3.37478489e-01 -1.08530831e+00 -2.25227624e-01
5.09073436e-01 -4.89202350e-01 -2.35550240e-01 1.25976777e+00
5.89427590e-01 2.08345756e-01 6.79652035e-01 -1.38786280e+00
-4.06250417e-01 9.06116068e-01 -4.19248521e-01 -1.18078075e-01
4.71260883e-02 8.13734904e-02 1.45806289e+00 -1.02041936e+00
5.68414330e-01 1.32756948e+00 1.01022804e+00 2.89146334e-01
-1.58823252e+00 -3.40617031e-01 3.95326763e-01 2.94942707e-01
-1.23429334e+00 -5.65019548e-01 4.02488738e-01 -2.61393636e-01
8.19088817e-01 2.81921089e-01 8.93123597e-02 1.06991303e+00
-1.24840811e-02 1.26604712e+00 7.10146248e-01 -3.20202827e-01
6.95875168e-01 2.26275057e-01 1.94229856e-01 6.22654855e-01
1.68260839e-02 -1.04509421e-01 -8.54477167e-01 -8.31915319e-01
-7.52740800e-02 2.71954894e-01 -2.56335586e-01 -5.46713114e-01
-5.75781286e-01 1.18533051e+00 1.13106303e-01 -3.45685691e-01
-8.15599203e-01 3.70258570e-01 6.45069957e-01 2.93188721e-01
9.47445273e-01 -3.52798492e-01 -8.45148087e-01 -2.46962398e-01
-1.27998507e+00 4.80741143e-01 1.50389779e+00 6.34552538e-01
1.07148468e+00 -3.40533108e-01 -3.52616847e-01 8.88402164e-01
8.21162760e-01 4.01441157e-01 3.49116586e-02 -1.55506027e+00
3.48948747e-01 -3.41886724e-03 2.11132824e-01 -3.24478537e-01
4.08333063e-01 -2.17955455e-01 -4.55218673e-01 3.57070059e-01
4.71320927e-01 -6.40809476e-01 -1.75722510e-01 1.96143019e+00
8.11509132e-01 1.71977043e-01 -8.05693790e-02 8.01146984e-01
1.01820000e-01 7.76732326e-01 5.41220568e-02 -2.10686326e-01
8.67493272e-01 -1.22194815e+00 -5.56404769e-01 2.61894673e-01
4.98199761e-01 -4.85909224e-01 6.80559635e-01 8.47325683e-01
-1.20335746e+00 2.94200301e-01 -5.48326433e-01 -8.44126865e-02
-1.29692256e-01 -3.88440371e-01 6.27000451e-01 6.80840731e-01
-1.31523967e+00 8.42000961e-01 -1.24651074e+00 -1.65866986e-01
7.20890820e-01 3.62754501e-02 -1.45864353e-01 -2.96816379e-01
-6.02508843e-01 5.44725478e-01 4.53105904e-02 -2.82911569e-01
-1.36938953e+00 -1.16062200e+00 -5.17212987e-01 3.10814470e-01
4.66653481e-02 -1.14622784e+00 1.97155857e+00 -6.59250140e-01
-1.51821768e+00 5.80243945e-01 -3.80024165e-01 -5.44814169e-01
8.03678632e-01 -3.84469926e-01 9.26140621e-02 -1.16072446e-01
-1.29871741e-01 -1.91947892e-01 7.62106061e-01 -1.16519928e+00
-8.04559290e-01 -6.35732412e-01 -3.48111659e-01 -9.61086713e-03
-4.32042539e-01 9.60430279e-02 -5.30715525e-01 -2.02376530e-01
-3.03404808e-01 -4.27553743e-01 -4.46473956e-02 3.65442216e-01
-2.09648907e-01 -6.58417046e-01 1.08147085e+00 -6.86565936e-01
8.09031546e-01 -1.97521579e+00 1.62614986e-01 1.48414478e-01
5.06554604e-01 -3.54524434e-01 8.01909044e-02 6.46886230e-01
3.55518699e-01 -1.49031237e-01 -1.23813450e-01 -1.29325759e+00
8.52368951e-01 2.67278135e-01 -5.89599788e-01 1.03338456e+00
-5.46926558e-01 8.35468471e-01 -7.56494880e-01 -7.98222601e-01
-4.82949317e-02 8.30102921e-01 -1.10547781e+00 5.25033712e-01
-8.66814196e-01 3.28054056e-02 -5.35852671e-01 4.14720654e-01
7.32010186e-01 -5.97658217e-01 3.96407753e-01 5.75842634e-02
-2.68423147e-02 3.52375984e-01 -1.29007256e+00 1.89220691e+00
-6.67962432e-01 3.13037068e-01 1.03518665e+00 -9.04094458e-01
4.44034100e-01 5.43010414e-01 8.51491153e-01 2.16547344e-02
-6.24858625e-02 -1.29574418e-01 -1.15160477e+00 -2.78975517e-01
1.41479298e-01 -3.20031106e-01 2.81023026e-01 1.25162685e+00
4.44101781e-01 2.61043996e-01 -3.46321583e-01 5.32499433e-01
1.12713110e+00 3.95133287e-01 -3.21488112e-01 -2.05642432e-01
1.86866954e-01 -1.23434953e-01 5.70118546e-01 1.05551541e+00
-2.59772062e-01 3.40571433e-01 3.90518010e-01 -4.03337926e-01
-9.35567975e-01 -1.41487706e+00 -7.61282668e-02 1.78304815e+00
-4.31312472e-01 -7.34850049e-01 -8.14688623e-01 -8.83669496e-01
3.72172236e-01 9.55025375e-01 -4.03236479e-01 2.47989222e-01
1.53990500e-02 -8.57016504e-01 4.47565049e-01 2.12568417e-01
2.12060750e-01 -5.41454613e-01 -2.97202647e-01 1.97679222e-01
-3.83172482e-01 -5.63578010e-01 -5.55030584e-01 2.32635841e-01
-8.88116181e-01 -8.69681656e-01 -5.05627871e-01 -2.41904274e-01
1.06099196e-01 -2.70968646e-01 1.34407747e+00 -4.44769651e-01
-2.63209403e-01 8.12886178e-01 2.34797060e-01 -2.45789096e-01
-2.89456666e-01 -2.80934632e-01 6.07986487e-02 2.78731823e-01
6.35664761e-01 -7.10421085e-01 -5.30875802e-01 -1.41136333e-01
-5.98836064e-01 -2.54754782e-01 7.77130574e-02 8.17215800e-01
7.39279449e-01 -1.41482383e-01 2.11952925e-01 -1.01244485e+00
5.80850184e-01 -1.08009183e+00 -5.40261388e-01 5.48413634e-01
-9.17150795e-01 4.11954522e-01 7.36805439e-01 -1.19673051e-01
-1.56058156e+00 -1.39763340e-01 -1.69540673e-01 -8.24101806e-01
-2.28211302e-02 3.45528752e-01 -1.81317672e-01 2.54025400e-01
6.23737633e-01 2.90928662e-01 1.42371818e-01 -1.08273256e+00
7.88095117e-01 7.71960199e-01 4.77192253e-01 -1.37092459e+00
2.88019449e-01 6.71621740e-01 -4.65048641e-01 -4.10331726e-01
-1.05262363e+00 -4.23368543e-01 -3.80899347e-02 -1.40667990e-01
4.28711385e-01 -1.06978440e+00 -1.25672519e+00 2.52189219e-01
-1.19869137e+00 -5.32849371e-01 -6.47061646e-01 4.76699501e-01
-9.76558208e-01 2.95098364e-01 -1.02635932e+00 -1.00158799e+00
-8.78072500e-01 -7.58084297e-01 1.03341424e+00 -9.52860340e-02
1.08405473e-02 -1.46913481e+00 9.43598688e-01 3.76826733e-01
7.24500537e-01 -3.76013108e-03 7.45500982e-01 -7.85613716e-01
-5.50748110e-01 -7.31410161e-02 -7.58213848e-02 3.77728254e-01
-3.43417495e-01 -1.83780845e-02 -1.17601252e+00 -4.85695660e-01
3.16582352e-01 -8.43042672e-01 7.58046865e-01 3.47545385e-01
1.36801338e+00 -7.41494477e-01 -1.91079617e-01 1.03519297e+00
1.72286761e+00 -9.31737125e-01 -2.17622332e-02 -8.25945288e-02
2.69206613e-01 2.50986516e-01 -9.77065563e-02 1.05915749e+00
5.56572556e-01 1.76327109e-01 5.31827271e-01 5.19559681e-01
1.20427102e-01 -5.70240855e-01 4.86684382e-01 6.32425845e-01
2.99380809e-01 -1.33170769e-01 -7.14446247e-01 6.28905714e-01
-2.22566557e+00 -1.14595044e+00 1.18252404e-01 1.93922794e+00
1.14027274e+00 -6.17767155e-01 2.66067356e-01 -5.48949659e-01
6.11424088e-01 9.39847380e-02 -9.12011743e-01 -3.40220511e-01
1.18112952e-01 2.98660547e-01 4.18910146e-01 6.73971236e-01
-8.18032384e-01 5.18113554e-01 7.07865047e+00 7.13772357e-01
-6.00961208e-01 8.08926761e-01 6.64112926e-01 -6.15410566e-01
-3.93013299e-01 2.49227956e-02 -8.06635499e-01 4.55007911e-01
1.24506795e+00 -3.80765140e-01 8.69200468e-01 1.25583565e+00
2.97083594e-02 1.93559915e-01 -1.38614738e+00 9.73074019e-01
-1.74883634e-01 -1.53893173e+00 -3.05232376e-01 2.87371486e-01
8.97950947e-01 1.03449523e+00 -4.25404571e-02 3.86189580e-01
1.58017302e+00 -9.74716842e-01 5.87748051e-01 9.08958852e-01
3.69875759e-01 -7.84010172e-01 2.65167147e-01 7.51689315e-01
-7.68980443e-01 -1.03878044e-02 -2.99557745e-01 9.22880694e-02
1.99933842e-01 5.98607421e-01 -4.77802813e-01 3.39866459e-01
1.02031529e+00 5.53472221e-01 1.33772463e-01 8.83974552e-01
8.48471001e-02 1.10741627e+00 -7.20082879e-01 1.94266900e-01
-3.38080004e-02 -3.34057480e-01 2.92117715e-01 1.35347199e+00
-2.07775105e-02 -3.57281715e-01 1.92045301e-01 1.35250223e+00
-4.91532087e-01 1.23406962e-01 -1.80771068e-01 2.59299785e-01
4.00659859e-01 1.23223960e+00 1.64850131e-01 -3.44209075e-01
-5.56254923e-01 9.28535640e-01 1.00284314e+00 6.47927701e-01
-7.09640265e-01 -1.50683029e-02 1.31243443e+00 -4.17201757e-01
7.19141006e-01 1.89627290e-01 -1.13923840e-01 -1.41640842e+00
-2.09876537e-01 -9.38110173e-01 9.69374776e-01 -3.36002469e-01
-2.06731558e+00 1.04524747e-01 -6.22593723e-02 -1.49496898e-01
-4.76472527e-01 -2.73657918e-01 -9.49886680e-01 1.25972068e+00
-1.30489075e+00 -1.10119963e+00 1.07803456e-01 1.14067590e+00
2.69415706e-01 1.10688061e-02 1.05311155e+00 2.72320747e-01
-6.50916994e-01 6.88713968e-01 9.11408424e-01 5.60572604e-04
5.30865550e-01 -1.25068879e+00 -4.70275208e-02 7.94197142e-01
3.18725437e-01 4.65593040e-01 9.20973539e-01 -3.19025487e-01
-1.93616033e+00 -1.31273484e+00 7.98844814e-01 -6.03059769e-01
8.56979609e-01 -2.11218998e-01 -7.85124302e-01 1.21663952e+00
3.64991069e-01 4.94773120e-01 8.34354639e-01 4.74631250e-01
-8.07210982e-01 -1.75424173e-01 -1.42211318e+00 6.92297742e-02
5.94865263e-01 -8.00156891e-01 -2.71731377e-01 7.35989332e-01
5.53277552e-01 1.51929269e-02 -1.20646548e+00 -3.13199550e-01
3.96610796e-01 -1.01165664e+00 7.72082031e-01 -1.02557206e+00
1.04450114e-01 9.30596143e-02 -6.51565611e-01 -1.18552160e+00
-1.65457398e-01 -1.16002202e+00 -1.08284044e+00 1.20995986e+00
9.36008319e-02 -7.38121450e-01 1.01031446e+00 1.02548897e+00
3.02059442e-01 -6.57768786e-01 -9.35593605e-01 -4.59538043e-01
5.22378087e-01 -7.67590880e-01 6.65925443e-01 7.59573638e-01
7.66710714e-02 -9.84798223e-02 -1.87750116e-01 1.87244907e-01
1.52869737e+00 3.01324308e-01 7.64436007e-01 -1.00498152e+00
-8.98286581e-01 -2.66678035e-01 1.85225844e-01 -1.12686861e+00
6.02728724e-01 -1.06720781e+00 1.00804910e-01 -1.32924592e+00
7.10191309e-01 -2.97103643e-01 -4.17136967e-01 6.50843859e-01
1.42136142e-01 -2.23331600e-01 1.36385068e-01 2.76653171e-01
-1.15709114e+00 8.24677467e-01 3.61177266e-01 -9.12970901e-02
2.15920836e-01 1.80242807e-01 -7.29956925e-01 4.58049595e-01
4.21902210e-01 -6.94761217e-01 -2.60945439e-01 -6.58302665e-01
2.49847934e-01 2.44896129e-01 4.34361130e-01 -3.64526957e-01
8.84175777e-01 -2.02833131e-01 1.07478924e-01 -5.92280924e-01
2.70397335e-01 -6.37822211e-01 3.79510820e-01 1.75443828e-01
-4.39482331e-01 -1.69099763e-01 -1.99614689e-01 8.75533521e-01
-8.92676562e-02 -3.30510102e-02 7.25805998e-01 -5.35598360e-02
-1.03496023e-01 6.96043372e-01 -9.77659971e-03 6.26232266e-01
8.44461322e-01 4.75585103e-01 -1.03605390e-01 -6.02757931e-01
-8.10531676e-01 5.16751051e-01 5.52224755e-01 -1.76207468e-01
3.03174734e-01 -9.40894425e-01 -1.02139866e+00 -1.03677744e-02
-3.36862803e-01 3.93185578e-02 1.00440979e-01 8.05880189e-01
-1.90718308e-01 2.62370199e-01 6.89246416e-01 -6.21121764e-01
-9.46901381e-01 4.36366856e-01 6.65871382e-01 -2.03151315e-01
-9.15685654e-01 1.08274949e+00 -2.65102178e-01 -7.41273701e-01
9.24972475e-01 3.66160542e-01 7.55199134e-01 -2.83030391e-01
7.18456864e-01 6.07179940e-01 1.35877177e-01 -5.82074709e-02
-2.88721085e-01 -2.75629312e-01 -5.98527640e-02 -3.33998501e-01
1.80090177e+00 -2.16901585e-01 -4.56191570e-01 5.07672369e-01
1.68197107e+00 -2.12512612e-01 -1.74360120e+00 -7.74917960e-01
-3.74954939e-02 -3.77094090e-01 5.23423493e-01 -8.09426129e-01
-1.21780705e+00 5.68653643e-01 3.31991047e-01 -1.43465951e-01
7.16975331e-01 3.45738769e-01 6.99175894e-01 4.88268137e-01
4.48202878e-01 -9.99912977e-01 -3.89516294e-01 4.17969942e-01
2.73693055e-01 -1.03500128e+00 -7.67959878e-02 6.10916138e-01
-3.44617695e-01 9.16745961e-01 1.77919015e-01 -5.45401908e-02
1.23697174e+00 5.93299687e-01 -5.85110560e-02 -2.97748923e-01
-1.53659976e+00 4.80691701e-01 -1.93065017e-01 3.04166347e-01
3.90058845e-01 -1.88089628e-02 1.54233083e-01 6.85968399e-01
-3.08166631e-02 1.33507788e-01 -2.64334291e-01 8.56944263e-01
-2.28790656e-01 -9.91624355e-01 -5.32618880e-01 3.90546501e-01
-7.80178785e-01 1.34627417e-01 8.69084001e-02 1.81831568e-02
-2.90834844e-01 9.01679814e-01 1.13080293e-01 1.01879023e-01
-3.29846382e-01 7.65671253e-01 3.90377522e-01 -4.27111536e-01
-6.44418955e-01 -1.07910417e-01 -2.11072505e-01 -9.10235047e-01
-1.88631520e-01 -8.94648075e-01 -9.29076612e-01 -8.01424325e-01
-1.64908051e-01 6.72835171e-01 1.16596496e+00 8.53320062e-01
5.32186031e-01 -1.03629278e-02 7.31975973e-01 -6.86499774e-01
-1.86072934e+00 -6.95273459e-01 -7.44802594e-01 3.68001729e-01
4.94639814e-01 -4.85254191e-02 -6.95110142e-01 8.54819939e-02] | [5.838528156280518, 6.302443981170654] |
05938eb2-d6f4-4999-86d2-8d0969a08ac3 | pd-quant-post-training-quantization-based-on | 2212.07048 | null | https://arxiv.org/abs/2212.07048v3 | https://arxiv.org/pdf/2212.07048v3.pdf | PD-Quant: Post-Training Quantization based on Prediction Difference Metric | Post-training quantization (PTQ) is a neural network compression technique that converts a full-precision model into a quantized model using lower-precision data types. Although it can help reduce the size and computational cost of deep neural networks, it can also introduce quantization noise and reduce prediction accuracy, especially in extremely low-bit settings. How to determine the appropriate quantization parameters (e.g., scaling factors and rounding of weights) is the main problem facing now. Existing methods attempt to determine these parameters by minimize the distance between features before and after quantization, but such an approach only considers local information and may not result in the most optimal quantization parameters. We analyze this issue and ropose PD-Quant, a method that addresses this limitation by considering global information. It determines the quantization parameters by using the information of differences between network prediction before and after quantization. In addition, PD-Quant can alleviate the overfitting problem in PTQ caused by the small number of calibration sets by adjusting the distribution of activations. Experiments show that PD-Quant leads to better quantization parameters and improves the prediction accuracy of quantized models, especially in low-bit settings. For example, PD-Quant pushes the accuracy of ResNet-18 up to 53.14% and RegNetX-600MF up to 40.67% in weight 2-bit activation 2-bit. The code is released at https://github.com/hustvl/PD-Quant. | ['Wenyu Liu', 'Xinggang Wang', 'Dawei Yang', 'Zhihang Yuan', 'Lin Niu', 'Jiawei Liu'] | 2022-12-14 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Liu_PD-Quant_Post-Training_Quantization_Based_on_Prediction_Difference_Metric_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Liu_PD-Quant_Post-Training_Quantization_Based_on_Prediction_Difference_Metric_CVPR_2023_paper.pdf | cvpr-2023-1 | ['neural-network-compression', 'neural-network-compression'] | ['methodology', 'miscellaneous'] | [-7.85321295e-02 -2.65737236e-01 -3.81265670e-01 -4.47934031e-01
-5.44950306e-01 -1.13762915e-01 4.80029061e-02 1.50391519e-01
-8.62407446e-01 6.61889553e-01 -1.36984894e-02 -4.14187402e-01
-3.56050804e-02 -1.04160190e+00 -6.70155823e-01 -6.51955843e-01
2.46693268e-01 9.30051580e-02 3.69139135e-01 -1.96174383e-01
2.36809269e-01 4.23649013e-01 -1.49639630e+00 3.31600100e-01
7.80932963e-01 1.39778578e+00 3.07399750e-01 4.12632108e-01
-1.66746408e-01 5.25800288e-01 -8.35081220e-01 -3.92865449e-01
5.00515580e-01 -2.04096764e-01 -5.81006646e-01 -7.15713918e-01
3.13741595e-01 -6.99004531e-01 -4.77387190e-01 1.43031681e+00
6.13039672e-01 -2.16335636e-02 4.48563695e-01 -1.19685459e+00
-4.38297361e-01 8.95137310e-01 -2.90861547e-01 5.64057410e-01
-3.74641091e-01 2.25914977e-02 6.73478603e-01 -6.52638435e-01
7.58039206e-02 1.29787540e+00 9.08947289e-01 5.12499750e-01
-1.21324468e+00 -1.06088114e+00 -4.63033617e-01 4.14674819e-01
-1.87934935e+00 -4.60875124e-01 4.66945916e-01 -1.44076720e-01
1.11100316e+00 -6.92015048e-03 6.31984413e-01 5.41211486e-01
4.71175015e-01 2.45812818e-01 6.52098298e-01 -4.86513495e-01
4.94554400e-01 -1.65197000e-01 2.00415954e-01 5.48696816e-01
3.41185540e-01 8.48480389e-02 -4.87158924e-01 -3.64158899e-02
9.87834573e-01 8.55320618e-02 -3.61138821e-01 2.68947959e-01
-8.08867157e-01 9.38490510e-01 7.08068490e-01 3.27914119e-01
-4.51113403e-01 5.13731599e-01 5.57077527e-01 2.09638938e-01
1.26533553e-01 3.56733471e-01 -5.17299473e-01 -5.02162457e-01
-9.89670753e-01 1.52033880e-01 4.43489879e-01 7.14869440e-01
9.05626595e-01 3.85465562e-01 -1.16738394e-01 1.04009330e+00
1.45498648e-01 3.17875475e-01 1.03941023e+00 -1.02848721e+00
7.32622445e-01 3.57899517e-01 -2.34688908e-01 -1.12198520e+00
-4.07388866e-01 -3.32811266e-01 -1.26266611e+00 1.62958294e-01
3.22583497e-01 -3.73289853e-01 -8.85009229e-01 1.76987565e+00
-5.25096469e-02 5.20548271e-03 -2.81282123e-02 9.60914493e-01
7.22874880e-01 9.50678945e-01 6.17684014e-02 8.96762311e-03
1.33822894e+00 -7.31707156e-01 -8.11309636e-01 -2.18200430e-01
7.69359052e-01 -5.48109591e-01 1.20271218e+00 3.48545253e-01
-1.04676294e+00 -7.13315606e-01 -1.44533527e+00 -9.42569003e-02
-3.96640420e-01 2.60871232e-01 1.79563433e-01 7.57914722e-01
-1.23743629e+00 1.09515488e+00 -9.33890104e-01 1.16738878e-01
4.14790869e-01 7.39679456e-01 -3.50349322e-02 9.54981819e-02
-1.68431759e+00 7.89149523e-01 9.09393191e-01 -1.32578328e-01
-1.51145369e-01 -7.38770068e-01 -6.48999095e-01 5.14728367e-01
-9.87800583e-02 -2.42084205e-01 1.35948229e+00 -9.41046178e-01
-1.54958105e+00 1.86593644e-02 -4.90799993e-02 -1.00766850e+00
6.58213273e-02 -6.86121583e-02 -3.66793245e-01 1.63606092e-01
-3.94036710e-01 1.15978146e+00 7.61811376e-01 -4.07483429e-01
-5.02697468e-01 -2.27201626e-01 -2.39693612e-01 1.02088206e-01
-8.09953332e-01 -2.90724933e-01 -5.09015143e-01 -7.87697315e-01
2.61218756e-01 -6.91608548e-01 -1.64638266e-01 4.21483293e-02
-2.53129993e-02 -1.54193342e-01 6.07034266e-01 -6.42561078e-01
1.47420180e+00 -2.31716228e+00 -5.58958769e-01 2.17413604e-01
1.26603052e-01 7.09775031e-01 -9.81838778e-02 -6.86749816e-02
-7.93516636e-02 4.69260991e-01 3.65930051e-02 -2.06901535e-01
2.36403011e-02 3.13095808e-01 -1.57851130e-01 2.48582959e-01
1.55489191e-01 7.35483468e-01 -4.81688738e-01 -4.70993459e-01
1.77281067e-01 8.93829942e-01 -7.61125863e-01 -1.49149030e-01
1.28739551e-01 -4.10388738e-01 -8.42834339e-02 2.52300411e-01
8.76612484e-01 -2.92329580e-01 6.33871630e-02 -5.48612535e-01
4.00768965e-02 6.95002615e-01 -1.24375331e+00 1.16375172e+00
-3.08498949e-01 7.18630731e-01 -2.71833837e-01 -7.66011417e-01
1.15166020e+00 3.56853247e-01 3.17552418e-01 -1.05772257e+00
4.17348415e-01 1.35013849e-01 2.32626628e-02 5.76765463e-02
7.93230951e-01 -1.65838838e-01 1.87411919e-01 1.71689332e-01
3.32566686e-02 9.52258408e-02 1.11044236e-01 -1.71652019e-01
7.57252455e-01 -4.98030275e-01 2.62564361e-01 -8.02586749e-02
1.25091448e-01 -3.53390664e-01 8.35546434e-01 5.09323716e-01
-3.28907162e-01 6.88557386e-01 4.98216838e-01 -3.62871885e-01
-1.26745486e+00 -6.22502804e-01 -4.25931841e-01 8.64707887e-01
-1.13417871e-01 -7.63980567e-01 -9.97258127e-01 -1.73329547e-01
-1.07186727e-01 6.70933962e-01 -2.77584255e-01 -4.92064148e-01
-3.97990018e-01 -8.96817923e-01 8.38454187e-01 5.28868914e-01
9.01506305e-01 -8.61629665e-01 -6.83038771e-01 2.58167237e-01
-1.69566274e-01 -1.02695966e+00 -5.34258604e-01 5.64984143e-01
-1.16592395e+00 -5.70296288e-01 -5.60452938e-01 -5.93277335e-01
5.23934484e-01 -1.01281844e-01 9.05612528e-01 2.46592671e-01
2.47044474e-01 -6.02277577e-01 -2.89353698e-01 -3.01833898e-01
-4.03587610e-01 3.33362401e-01 2.03446120e-01 -3.18770796e-01
4.66985852e-01 -5.26628017e-01 -5.75847805e-01 2.32224718e-01
-9.95391250e-01 9.42891464e-02 5.49815714e-01 7.78148770e-01
7.95143008e-01 4.09579486e-01 5.63334048e-01 -5.53932548e-01
8.73188257e-01 -1.96765020e-01 -7.97680616e-01 1.93957873e-02
-8.99152458e-01 4.31346297e-01 1.04472935e+00 -5.41821122e-01
-3.96975547e-01 -4.17014062e-02 -5.12089968e-01 -7.74273574e-01
2.53370196e-01 4.08057660e-01 4.35061455e-02 -1.94072306e-01
8.91212761e-01 5.48399724e-02 8.53370801e-02 -3.68359566e-01
2.21151784e-02 7.57722497e-01 3.93458217e-01 -1.73896298e-01
3.95349354e-01 -2.16657355e-01 -1.69718966e-01 -5.44169188e-01
-4.02236670e-01 -4.81601283e-02 -4.23718989e-01 1.94794774e-01
4.71119523e-01 -8.78821254e-01 -6.03901744e-01 2.97722965e-01
-1.03891253e+00 -5.46467066e-01 -3.30443770e-01 5.28005421e-01
-1.54710770e-01 2.88711816e-01 -8.33913028e-01 -4.51360434e-01
-7.84558713e-01 -1.37098646e+00 4.98400390e-01 2.39741266e-01
-1.98360071e-01 -6.67010009e-01 -4.52588826e-01 -3.01641434e-01
6.56271398e-01 -2.21909240e-01 9.01205063e-01 -3.38046491e-01
-2.27061316e-01 -1.06334850e-01 -3.16363961e-01 7.37883806e-01
1.47004910e-02 -4.18309905e-02 -8.04665327e-01 -2.02941850e-01
-8.92503858e-02 -2.46642560e-01 7.52752960e-01 5.17282009e-01
1.53819621e+00 -6.67964995e-01 8.82188678e-02 9.13491607e-01
1.50408137e+00 3.40759277e-01 8.88239503e-01 2.85783499e-01
4.15987730e-01 -1.42925039e-01 4.49432284e-01 6.10811472e-01
1.29506916e-01 8.16987097e-01 4.25129056e-01 2.29058266e-01
-3.52265328e-01 -3.35167617e-01 2.71093071e-01 1.01445580e+00
2.41962776e-01 -8.60698745e-02 -9.97382045e-01 2.32206449e-01
-1.39709961e+00 -7.61190653e-01 1.82117507e-01 2.35571814e+00
1.30719197e+00 4.40134853e-01 -6.66854233e-02 7.28820920e-01
7.17596650e-01 -7.90209174e-02 -5.33118546e-01 -7.09874034e-01
1.19793184e-01 3.71211499e-01 9.90391374e-01 4.98921931e-01
-8.04069102e-01 8.61191452e-01 6.32441378e+00 1.29258752e+00
-1.52197397e+00 1.82353839e-01 8.00194323e-01 -1.83732316e-01
2.04472281e-02 -3.39820206e-01 -1.25893080e+00 8.79216790e-01
1.54979289e+00 -6.70850500e-02 6.17262900e-01 9.58453119e-01
1.84787020e-01 6.10624366e-02 -8.26896906e-01 1.32279539e+00
-3.13938916e-01 -1.56097388e+00 2.37989798e-01 -3.24070267e-02
5.08456707e-01 3.24096382e-01 2.97054835e-02 3.27896386e-01
-1.11770608e-01 -1.07855558e+00 8.51041317e-01 2.51367897e-01
1.06613791e+00 -1.02226925e+00 1.00256777e+00 3.74433577e-01
-1.11601114e+00 -2.44204208e-01 -9.17166948e-01 -1.33702978e-01
-2.64523029e-01 8.02048624e-01 -1.08676791e+00 -1.71906188e-01
8.48796606e-01 3.36769700e-01 -5.47463715e-01 9.96753812e-01
-7.40366876e-02 7.47874558e-01 -6.91451967e-01 -2.82506138e-01
1.04246765e-01 -5.51193841e-02 -1.91860527e-01 1.02034974e+00
5.20301163e-01 1.84005350e-01 -3.36831093e-01 6.98041499e-01
-3.56616586e-01 2.82288939e-02 -4.74254228e-02 1.42287374e-01
1.08146417e+00 8.26400220e-01 -5.10843933e-01 -4.03462440e-01
8.75543803e-03 5.60693264e-01 4.05678064e-01 1.70848608e-01
-8.71476829e-01 -6.91922247e-01 5.51911056e-01 2.50529230e-01
3.70287180e-01 -1.88107982e-01 -5.23417056e-01 -7.34043598e-01
-1.31225824e-01 -8.42781603e-01 1.42035276e-01 -6.90475941e-01
-6.36836946e-01 6.36829138e-01 -8.64587352e-02 -1.22891331e+00
-3.56573075e-01 -4.63313043e-01 -2.25467563e-01 9.20616508e-01
-1.49427199e+00 -2.11702526e-01 -2.48641819e-01 4.56493855e-01
3.65279794e-01 -2.78102420e-02 8.08988631e-01 5.81197619e-01
-6.06748998e-01 1.22516584e+00 3.61138940e-01 1.59384519e-01
4.60516155e-01 -7.30152071e-01 4.11386460e-01 6.86581910e-01
-1.68077797e-01 5.64296007e-01 5.22515833e-01 -2.49261588e-01
-1.14970815e+00 -1.22345734e+00 8.58421862e-01 3.80392045e-01
2.25871027e-01 -9.70019624e-02 -1.13977003e+00 3.40221584e-01
-2.46490791e-01 1.93592861e-01 4.94486392e-01 -2.62182564e-01
-1.23728380e-01 -5.89071214e-01 -1.37621903e+00 6.13858640e-01
4.13215727e-01 -3.92880172e-01 -2.32713714e-01 8.20585936e-02
8.37302268e-01 -5.42610288e-01 -1.18863213e+00 3.20046216e-01
4.86582041e-01 -9.51039732e-01 8.76433492e-01 1.36739552e-01
4.41307604e-01 -1.48738071e-01 -3.09237391e-01 -1.27469730e+00
-3.83869439e-01 -3.07004631e-01 -2.21977934e-01 1.02264571e+00
4.45733786e-01 -7.79142737e-01 8.90170336e-01 6.47539318e-01
1.13748349e-02 -9.85717058e-01 -1.12227392e+00 -6.91432655e-01
1.28399402e-01 -5.19134820e-01 9.94156420e-01 6.42805696e-01
-6.59983605e-02 3.88892069e-02 -8.68563801e-02 4.64495048e-02
3.29494298e-01 -5.42103589e-01 3.55472207e-01 -9.42526579e-01
-1.37809709e-01 -6.09249532e-01 -4.92227882e-01 -1.07794273e+00
-1.06159084e-01 -7.56016374e-01 -4.43382934e-03 -1.13344944e+00
-3.64039868e-01 -7.87609041e-01 -4.38417643e-01 9.06298935e-01
1.28988981e-01 4.85997736e-01 4.57149833e-01 3.27100098e-01
-1.86218306e-01 6.42262697e-01 1.07392764e+00 -7.56774144e-03
-2.58042127e-01 -2.08828241e-01 -4.90045518e-01 6.07095182e-01
1.43438685e+00 -6.70127749e-01 -4.90540475e-01 -5.77563643e-01
1.43988058e-01 1.59027278e-02 1.85159594e-01 -1.46184337e+00
3.61154258e-01 2.31148619e-02 5.82129180e-01 -5.22430599e-01
4.45132732e-01 -7.59385109e-01 1.33598968e-01 7.20947325e-01
-3.55404943e-01 3.85908276e-01 4.03452545e-01 1.00455739e-01
-2.62014449e-01 -6.62380219e-01 1.01717508e+00 2.27309048e-01
-6.57960773e-01 2.88987666e-01 -2.41678461e-01 -1.50097653e-01
4.86240566e-01 -3.35591108e-01 -3.96835774e-01 -2.95451552e-01
-3.37534875e-01 -7.71977082e-02 2.81631112e-01 1.96287826e-01
7.55547643e-01 -1.54505515e+00 -3.57608527e-01 6.02023125e-01
-3.97548169e-01 7.43718818e-02 6.62700757e-02 5.22300720e-01
-8.34255517e-01 5.68117201e-01 -3.26923460e-01 -3.61836076e-01
-1.21196771e+00 2.55459428e-01 5.99726081e-01 -1.09648347e-01
-4.80177552e-01 7.28781939e-01 -3.84799212e-01 -4.04380746e-02
5.23149848e-01 -8.55101228e-01 -2.52177954e-01 -2.05532134e-01
7.86562443e-01 3.45499843e-01 4.37125772e-01 -4.59341645e-01
-1.98897660e-01 4.39051926e-01 -5.17487638e-02 3.49553749e-02
1.14956355e+00 1.62872389e-01 1.89569462e-02 1.49561241e-01
1.49362373e+00 -5.77029049e-01 -1.35550952e+00 -1.80852607e-01
-3.67019564e-01 -2.76103377e-01 5.72408855e-01 -4.87012714e-01
-1.49595916e+00 1.07428563e+00 1.03370786e+00 1.49388656e-01
1.34115672e+00 -6.26061976e-01 1.02521861e+00 4.55471456e-01
2.73269087e-01 -1.16496682e+00 -1.35349602e-01 7.39934802e-01
6.69844866e-01 -8.69259894e-01 1.05861127e-01 -9.71517712e-02
-2.32952639e-01 1.25264001e+00 6.17065191e-01 -7.67161995e-02
8.32726300e-01 5.22072196e-01 -6.42382726e-02 3.46743524e-01
-7.14409232e-01 4.01944339e-01 4.82042693e-02 3.03044140e-01
3.69499892e-01 8.77957270e-02 -3.68452787e-01 7.29478180e-01
-8.31479490e-01 2.89981127e-01 4.59009916e-01 7.54040956e-01
-6.09804928e-01 -1.01727462e+00 -4.98426944e-01 6.61303461e-01
-6.28489435e-01 -3.86454731e-01 2.34181821e-01 4.35392499e-01
3.13602746e-01 6.76240683e-01 4.54399347e-01 -8.41498196e-01
1.50680050e-01 4.25100848e-02 1.08975716e-01 -2.09497482e-01
-5.64263344e-01 -1.56499535e-01 -2.13094801e-01 -4.01674330e-01
2.24434696e-02 -1.42483175e-01 -1.66020370e+00 -8.59742522e-01
-4.29019302e-01 1.37883380e-01 8.43248785e-01 6.43320382e-01
6.07223570e-01 6.82292819e-01 4.08494473e-01 -8.08206141e-01
-9.76243377e-01 -1.00648558e+00 -5.46425223e-01 4.40577306e-02
3.38031381e-01 -4.33169991e-01 -3.89794052e-01 -5.18286712e-02] | [8.598514556884766, 3.0421090126037598] |
461eabdc-2347-4924-9d59-e7da01d58dc9 | emog-synthesizing-emotive-co-speech-3d | 2306.11496 | null | https://arxiv.org/abs/2306.11496v1 | https://arxiv.org/pdf/2306.11496v1.pdf | EMoG: Synthesizing Emotive Co-speech 3D Gesture with Diffusion Model | Although previous co-speech gesture generation methods are able to synthesize motions in line with speech content, it is still not enough to handle diverse and complicated motion distribution. The key challenges are: 1) the one-to-many nature between the speech content and gestures; 2) the correlation modeling between the body joints. In this paper, we present a novel framework (EMoG) to tackle the above challenges with denoising diffusion models: 1) To alleviate the one-to-many problem, we incorporate emotion clues to guide the generation process, making the generation much easier; 2) To model joint correlation, we propose to decompose the difficult gesture generation into two sub-problems: joint correlation modeling and temporal dynamics modeling. Then, the two sub-problems are explicitly tackled with our proposed Joint Correlation-aware transFormer (JCFormer). Through extensive evaluations, we demonstrate that our proposed method surpasses previous state-of-the-art approaches, offering substantial superiority in gesture synthesis. | ['Jianxin Lin', 'Xin Jin', 'Bohan Li', 'Wei Zhao', 'Jinming Liu', 'Tianyu He', 'Yijun Wang', 'Lianying Yin'] | 2023-06-20 | null | null | null | null | ['gesture-generation'] | ['robots'] | [-1.28792197e-01 -2.49970675e-01 -4.21878621e-02 1.33455917e-01
-8.34755123e-01 -3.40108812e-01 8.02285492e-01 -6.44296765e-01
-9.28400531e-02 4.57074255e-01 6.90684080e-01 3.05514168e-02
-1.21138565e-01 -3.97614300e-01 -2.89085001e-01 -8.93671334e-01
1.08428188e-01 3.31133902e-01 2.30157092e-01 -3.51391584e-01
-1.14413813e-01 2.14371786e-01 -1.32849157e+00 -2.18586363e-02
8.98802936e-01 7.78257728e-01 2.06956640e-01 8.93773973e-01
-1.54449478e-01 8.57604980e-01 -6.35288537e-01 -2.98741490e-01
-5.27454540e-02 -9.10759926e-01 -4.20790404e-01 2.16029868e-01
-1.07041650e-01 -5.06292224e-01 -4.49371845e-01 8.16685200e-01
7.63487995e-01 5.25897622e-01 9.69041884e-01 -1.12603426e+00
-6.41014993e-01 4.68243539e-01 -7.15568066e-01 -1.35253087e-01
3.16866130e-01 2.12922543e-01 8.77063990e-01 -8.55073690e-01
7.07958281e-01 1.56341493e+00 3.83199096e-01 9.40641165e-01
-8.00152183e-01 -5.23647130e-01 5.77501476e-01 1.44553050e-01
-1.23673356e+00 -4.00263608e-01 8.81740630e-01 -4.87626195e-01
5.58828533e-01 2.72911370e-01 7.74599016e-01 1.56294346e+00
-2.49467477e-01 1.06204855e+00 9.26333487e-01 -1.42800033e-01
2.11418703e-01 -6.44041896e-01 -1.11054674e-01 4.80406761e-01
-2.95756519e-01 1.53209576e-02 -8.01565588e-01 -1.37308776e-01
9.72920775e-01 -8.35680366e-02 -3.55055153e-01 8.93412381e-02
-1.32447469e+00 6.53723478e-01 -9.29351430e-03 4.93907988e-01
-5.50897360e-01 5.95614374e-01 1.53406337e-01 -5.12543209e-02
3.85949761e-01 -1.52820796e-01 -2.63019875e-02 -6.14377201e-01
-1.08295393e+00 6.80641532e-01 8.19133282e-01 9.42571104e-01
1.20037086e-01 2.31675938e-01 -3.85592908e-01 8.10885549e-01
6.51722431e-01 5.93837500e-01 5.06997526e-01 -8.77968013e-01
6.99472964e-01 9.36565474e-02 1.46386325e-01 -1.02486801e+00
-3.32254618e-01 -2.23604947e-01 -1.14589870e+00 -1.62480306e-02
6.23444378e-01 -3.59331369e-01 -7.84014046e-01 1.96503639e+00
3.13616604e-01 5.06412923e-01 9.18637663e-02 1.25648522e+00
9.03448880e-01 7.04081655e-01 1.54833451e-01 -3.54292661e-01
1.34046793e+00 -1.33834732e+00 -1.26960015e+00 -3.37646604e-02
2.61236608e-01 -1.01582301e+00 8.58153522e-01 2.49689952e-01
-1.26469409e+00 -4.62012291e-01 -7.70511925e-01 1.13492711e-02
2.58614570e-01 3.17239314e-01 6.09669209e-01 4.31681961e-01
-8.63071680e-01 5.78969538e-01 -1.17230165e+00 -2.39673302e-01
-6.66839536e-03 1.34874880e-01 -5.94063289e-02 1.78753927e-01
-1.03328300e+00 6.87118471e-01 -1.78940833e-01 2.02612683e-01
-5.58583915e-01 -3.39697987e-01 -6.31824374e-01 -1.83589179e-02
4.89752144e-01 -9.42484319e-01 1.30513442e+00 -6.31732166e-01
-2.02181435e+00 6.28451332e-02 -5.16160369e-01 1.49142027e-01
9.16569412e-01 -5.33641279e-01 -2.00401232e-01 2.58863457e-02
-2.49836326e-01 5.01580000e-01 1.01433372e+00 -1.30219293e+00
-4.00165617e-01 -2.04284921e-01 -3.13254178e-01 6.19114339e-01
-3.19805652e-01 -4.24457230e-02 -9.46123183e-01 -1.44364417e+00
2.48737589e-01 -1.13563931e+00 -2.27707803e-01 4.94204573e-02
-4.71447349e-01 -1.87746227e-01 9.54889357e-01 -9.20926213e-01
1.45609665e+00 -1.91563559e+00 7.35347450e-01 1.17523922e-02
1.76460341e-01 3.04499894e-01 -2.98175246e-01 5.40524781e-01
3.32543582e-01 -1.53559789e-01 -7.23408461e-02 -8.74636531e-01
2.47786358e-01 3.09276819e-01 -3.82739395e-01 1.50982514e-01
1.04805775e-01 9.33035672e-01 -9.44388986e-01 -3.80238950e-01
1.23380437e-01 7.19728351e-01 -5.86193621e-01 4.09928411e-01
-2.10398734e-01 8.21798027e-01 -7.55300105e-01 6.51728749e-01
5.19866467e-01 -1.17560923e-02 2.47256488e-01 -3.17711800e-01
8.85414556e-02 2.70835087e-02 -1.32093632e+00 1.98215532e+00
-2.99184412e-01 2.59806544e-01 2.42589936e-01 -5.54981649e-01
8.65839660e-01 6.19398355e-01 6.09418988e-01 -1.44969270e-01
1.75690085e-01 3.08400065e-01 6.52843341e-02 -8.40100288e-01
5.37385345e-01 -2.02946723e-01 1.20014111e-02 6.08074009e-01
2.18445566e-02 -2.10994646e-01 7.89088458e-02 7.11949766e-02
1.10341704e+00 5.05270004e-01 -6.78505301e-02 2.01270029e-01
3.83252859e-01 -3.53853464e-01 6.45999491e-01 4.28132027e-01
-2.67932117e-01 8.51016998e-01 4.25523579e-01 7.62082860e-02
-7.00861514e-01 -9.86076951e-01 7.49491632e-01 6.82991326e-01
2.32138902e-01 -4.22487110e-01 -9.32539046e-01 -5.27173162e-01
-3.24594587e-01 4.45957959e-01 -4.60205793e-01 1.76556021e-01
-7.05255866e-01 -8.04171920e-01 6.81905329e-01 7.87139416e-01
4.60562468e-01 -8.95069957e-01 -4.01995033e-01 2.95540303e-01
-6.46704972e-01 -1.26149011e+00 -9.15058792e-01 -3.42316091e-01
-7.83116460e-01 -7.47567177e-01 -1.39924479e+00 -4.95297253e-01
3.10509175e-01 3.01682353e-01 6.71066463e-01 1.44947678e-01
-1.18943639e-01 4.15728301e-01 -7.16245651e-01 -4.82808426e-03
-3.05986911e-01 -1.02909930e-01 1.70904145e-01 6.05444387e-02
8.47980678e-02 -8.13950419e-01 -7.56663978e-01 3.92620325e-01
-9.06644225e-01 1.67907372e-01 5.95251620e-01 9.44675148e-01
3.05731535e-01 -1.46213785e-01 3.53095055e-01 -1.60919473e-01
9.51263130e-01 -2.11412817e-01 -2.51916740e-02 2.06861570e-01
-1.08893938e-01 9.65334624e-02 3.54917020e-01 -9.38973308e-01
-1.26046002e+00 1.07232153e-01 -3.72918427e-01 -5.47826052e-01
2.31066495e-02 3.93485904e-01 -1.85129791e-01 3.07272345e-01
3.18906009e-02 2.82709688e-01 1.56560034e-01 -5.65872610e-01
6.55705273e-01 4.10851687e-01 6.63584471e-01 -7.47495174e-01
9.58746612e-01 4.96548176e-01 4.37498540e-02 -8.71061981e-01
-3.41605186e-01 -2.71465987e-01 -7.12169349e-01 -3.48328710e-01
1.22537482e+00 -8.64160359e-01 -7.49544203e-01 7.58925378e-01
-1.44869709e+00 -5.56665659e-01 -5.92172220e-02 7.31083691e-01
-7.00905681e-01 6.45709336e-01 -9.32435513e-01 -1.16587007e+00
-2.05853269e-01 -1.29619193e+00 1.26325548e+00 2.71720052e-01
-5.03737509e-01 -9.05552685e-01 1.43008634e-01 4.05797094e-01
4.08256024e-01 2.33811915e-01 6.39350712e-01 -1.48234680e-01
-5.66618681e-01 1.26178682e-01 2.84184292e-02 -2.76302788e-02
3.00806910e-01 1.10682480e-01 -6.35562837e-01 3.68628949e-02
-8.76184404e-02 -2.06333309e-01 6.71939015e-01 3.70808214e-01
6.41780436e-01 -1.86229780e-01 -1.15990981e-01 5.16747177e-01
6.74938500e-01 2.31218129e-01 6.52389765e-01 -9.20105055e-02
7.88260996e-01 7.94740677e-01 7.58944273e-01 6.22277677e-01
6.58500195e-01 1.08844173e+00 2.15786770e-01 -1.11889383e-02
-5.38148344e-01 -4.36385185e-01 5.73703170e-01 1.50017083e+00
-6.26087248e-01 -4.19580042e-01 -4.76909578e-01 3.87649983e-01
-2.47273946e+00 -1.05677283e+00 -2.09709212e-01 1.50211906e+00
5.58049619e-01 -3.52250129e-01 3.40216249e-01 -1.04417847e-02
4.53564823e-01 5.38056254e-01 -3.72053593e-01 2.48876885e-02
-7.42000714e-02 -2.13122051e-02 -1.75612479e-01 4.66443658e-01
-9.69362974e-01 1.09069645e+00 6.03267050e+00 1.12046289e+00
-1.02855146e+00 8.85678381e-02 3.22635829e-01 -1.58082202e-01
-3.72369587e-01 -2.50380397e-01 -6.91190839e-01 4.75657940e-01
2.77251810e-01 3.35074872e-01 5.46180427e-01 4.70703363e-01
4.78771120e-01 1.78799823e-01 -7.56193399e-01 1.07489836e+00
1.92027420e-01 -9.84117210e-01 6.46725384e-05 2.16108467e-03
6.42202735e-01 -5.63983500e-01 7.35050738e-02 1.07688829e-01
2.66098440e-01 -8.56425226e-01 1.09367859e+00 7.11272299e-01
4.68909949e-01 -4.70088542e-01 3.97189409e-01 4.92540747e-01
-1.59552920e+00 1.54274032e-01 -2.48603094e-02 8.52466747e-02
7.88160801e-01 4.51973379e-01 -1.20033167e-01 6.59914076e-01
4.94955719e-01 6.68598473e-01 1.12981036e-01 6.56889796e-01
-5.38053274e-01 4.45932597e-01 -1.63845673e-01 -2.21105799e-01
2.93156743e-01 -4.21452969e-01 7.83203840e-01 1.12993395e+00
9.06892121e-01 5.03723204e-01 1.18597023e-01 9.70797002e-01
2.81988174e-01 -7.46282041e-02 -2.14510322e-01 -2.97281593e-01
3.59262675e-01 1.06286311e+00 -8.11480343e-01 -2.91818142e-01
-3.56204957e-01 1.46442294e+00 1.36204092e-02 7.49830544e-01
-8.45250010e-01 -3.21123749e-02 9.15412962e-01 -2.84986198e-01
3.84291887e-01 -7.23391294e-01 -1.84962288e-01 -1.40265715e+00
2.16189131e-01 -9.95539427e-01 5.97654097e-02 -6.21958196e-01
-1.22328460e+00 5.41125298e-01 -1.44131184e-01 -1.18253434e+00
-6.04618192e-01 -4.87276316e-01 -6.94154203e-01 8.37933421e-01
-1.17362297e+00 -1.42039692e+00 -3.27889830e-01 6.00070357e-01
7.58181572e-01 3.87378433e-03 7.10681677e-01 5.37598372e-01
-5.48947036e-01 4.05276746e-01 -1.39792845e-01 4.84313294e-02
6.33313835e-01 -9.66054022e-01 4.82533365e-01 9.27985489e-01
1.52512267e-01 7.15463638e-01 6.00070655e-01 -7.50611782e-01
-1.63742781e+00 -7.30661392e-01 8.02685440e-01 -2.32596949e-01
7.65174448e-01 -2.90302604e-01 -6.90153718e-01 3.15451890e-01
-1.21606998e-02 -2.25984201e-01 5.10525823e-01 -1.55495003e-01
-2.30579287e-01 4.33871001e-01 -5.28397501e-01 9.80844498e-01
1.30799341e+00 -2.99066693e-01 -4.53707159e-01 -1.02771118e-01
5.00960946e-01 -6.55922890e-01 -5.84493399e-01 3.18195790e-01
8.26621652e-01 -7.71353483e-01 9.64518666e-01 -3.94871324e-01
6.35498643e-01 -4.59866017e-01 -2.79334515e-01 -1.46594715e+00
-3.47402506e-02 -1.15912354e+00 -6.76334620e-01 1.40186870e+00
7.61486068e-02 -1.67199582e-01 8.62151384e-01 5.95440626e-01
-4.42338586e-02 -7.84337103e-01 -9.61067557e-01 -6.82929039e-01
-3.00125149e-03 -5.38624525e-01 6.38336897e-01 8.66768718e-01
-1.52934967e-02 2.91487336e-01 -1.07022953e+00 -9.36087500e-03
4.87306446e-01 1.01623900e-01 9.47667778e-01 -7.61932731e-01
-8.95898998e-01 -6.50107086e-01 -5.20911179e-02 -1.71680796e+00
-3.72984931e-02 -2.18480170e-01 4.21521902e-01 -1.60918808e+00
1.78581193e-01 -3.04361284e-01 3.32880259e-01 2.33796313e-01
-4.94021207e-01 1.21296927e-01 5.61974704e-01 3.00678879e-01
-3.44969004e-01 1.03356898e+00 1.60648298e+00 -1.94708188e-03
-3.25864106e-01 8.00112039e-02 -4.04165685e-01 7.68812776e-01
3.32742125e-01 -2.44088754e-01 -5.82098186e-01 -5.16991496e-01
-1.33172378e-01 5.89980185e-01 3.69581103e-01 -8.80723596e-01
2.91128546e-01 -2.49808446e-01 5.21371467e-03 -6.82258189e-01
8.46768916e-01 -6.11617744e-01 1.86567485e-01 4.53229010e-01
-1.72482118e-01 1.23250552e-01 -2.01235205e-01 7.01060057e-01
-3.04513335e-01 4.69842181e-02 3.75675619e-01 1.68589756e-01
-3.86830956e-01 3.70492369e-01 -7.17666090e-01 -6.67086691e-02
7.63342857e-01 1.54591605e-01 -2.58086652e-01 -1.07591116e+00
-1.03485000e+00 1.92957953e-01 1.23122461e-01 6.35666668e-01
5.70959270e-01 -1.66000044e+00 -5.69462776e-01 -1.75046355e-01
-3.12758505e-01 -5.46732619e-02 4.49834853e-01 1.10725391e+00
-3.67965311e-01 1.22764751e-01 1.18219711e-01 -4.23148334e-01
-1.40434468e+00 9.03326049e-02 -3.91319096e-02 -4.51950222e-01
-7.11581767e-01 9.61098373e-01 1.94012567e-01 -2.54636973e-01
4.78457510e-01 -3.39313358e-01 -1.16452269e-01 1.11484066e-01
5.91797173e-01 5.38319588e-01 -4.27747577e-01 -8.87728035e-01
-2.36600056e-01 9.27250326e-01 3.16214532e-01 -7.03139901e-01
1.29022646e+00 -2.03656256e-01 2.23734558e-01 3.26909006e-01
7.99937963e-01 -5.70227839e-02 -1.57091463e+00 -4.45400588e-02
-2.15825796e-01 -3.15777898e-01 -2.43665740e-01 -6.52765930e-01
-1.18739069e+00 1.13167119e+00 2.73836911e-01 -1.38901383e-01
1.09328508e+00 -2.54834712e-01 1.32356060e+00 6.23995103e-02
1.64607882e-01 -1.21142316e+00 5.72536290e-01 7.23076522e-01
9.05439675e-01 -7.67258048e-01 -1.65458098e-01 -7.08491445e-01
-9.78224099e-01 1.15734267e+00 4.34752494e-01 -4.04520817e-02
6.43505752e-01 3.77708465e-01 3.73163581e-01 -3.73558365e-02
-6.93758428e-01 -3.52541834e-01 6.29454434e-01 6.16522908e-01
5.62286556e-01 -1.66892409e-02 -6.00071967e-01 9.39336240e-01
-1.36528254e-01 2.01155543e-02 8.22243020e-02 7.52393663e-01
-4.75988835e-02 -1.52282131e+00 -5.98947942e-01 -1.20920517e-01
-2.71624953e-01 9.90888029e-02 -5.34397006e-01 6.48323238e-01
-1.29891589e-01 1.20466316e+00 -2.39583254e-01 -6.74493670e-01
3.34726542e-01 7.15486007e-03 5.83244920e-01 -1.17230304e-01
-4.24978644e-01 8.39120686e-01 1.44945309e-01 -6.91230297e-01
-5.93313575e-01 -7.54202843e-01 -1.23758531e+00 -1.82352290e-01
-2.91499376e-01 -1.60491914e-01 6.22682214e-01 1.17437947e+00
1.91835776e-01 8.51259828e-01 2.52713770e-01 -1.38768613e+00
-7.11243153e-01 -1.03051150e+00 -6.56899214e-01 5.61152875e-01
2.30190024e-01 -8.48246515e-01 -2.26657242e-01 8.74963701e-02] | [5.74577522277832, -0.2007274180650711] |
ce4fba60-b2a6-4c48-9457-383f4342f3c0 | on-graph-reconstruction-via-empirical-risk | null | null | http://papers.nips.cc/paper/6588-on-graph-reconstruction-via-empirical-risk-minimization-fast-learning-rates-and-scalability | http://papers.nips.cc/paper/6588-on-graph-reconstruction-via-empirical-risk-minimization-fast-learning-rates-and-scalability.pdf | On Graph Reconstruction via Empirical Risk Minimization: Fast Learning Rates and Scalability | The problem of predicting connections between a set of data points finds many applications, in systems biology and social network analysis among others. This paper focuses on the \textit{graph reconstruction} problem, where the prediction rule is obtained by minimizing the average error over all n(n-1)/2 possible pairs of the n nodes of a training graph. Our first contribution is to derive learning rates of order O(log n / n) for this problem, significantly improving upon the slow rates of order O(1/√n) established in the seminal work of Biau & Bleakley (2006). Strikingly, these fast rates are universal, in contrast to similar results known for other statistical learning problems (e.g., classification, density level set estimation, ranking, clustering) which require strong assumptions on the distribution of the data. Motivated by applications to large graphs, our second contribution deals with the computational complexity of graph reconstruction. Specifically, we investigate to which extent the learning rates can be preserved when replacing the empirical reconstruction risk by a computationally cheaper Monte-Carlo version, obtained by sampling with replacement B << n² pairs of nodes. Finally, we illustrate our theoretical results by numerical experiments on synthetic and real graphs. | ['Stephan Clémençon', 'Aurélien Bellet', 'Guillaume Papa'] | 2016-12-01 | null | null | null | neurips-2016-12 | ['graph-reconstruction'] | ['graphs'] | [ 2.50415683e-01 7.06213892e-01 -2.35861421e-01 -1.18336074e-01
-5.89890838e-01 -5.44061720e-01 4.93036866e-01 7.59887397e-01
-4.07314301e-01 9.28667724e-01 -3.45848143e-01 -3.84478360e-01
-6.11949444e-01 -9.21505094e-01 -8.94721627e-01 -7.71768630e-01
-6.92892313e-01 7.98262894e-01 3.16138208e-01 9.63292718e-02
2.64961839e-01 4.47800130e-01 -1.21248949e+00 -5.02533376e-01
6.94781125e-01 9.08390343e-01 -1.95998311e-01 8.27021301e-01
1.66004553e-01 4.45854545e-01 -2.99170434e-01 -8.42261553e-01
2.77254224e-01 -4.49441999e-01 -1.02481282e+00 1.56284586e-01
3.33504558e-01 6.28961399e-02 -3.60878170e-01 1.25371754e+00
1.71848387e-01 7.84360319e-02 9.06206191e-01 -1.20858622e+00
-2.05210730e-01 8.52375805e-01 -7.28878200e-01 1.91059336e-01
2.97234565e-01 -3.80315989e-01 1.25750113e+00 -4.17802006e-01
4.44847047e-01 1.10346353e+00 9.05547380e-01 3.14478904e-01
-1.71604490e+00 -4.83451396e-01 8.37888918e-04 7.19864294e-02
-1.59329522e+00 -1.57500073e-01 5.87128103e-01 -5.72316885e-01
3.08205932e-01 2.19042137e-01 5.00272095e-01 7.08465338e-01
6.62128553e-02 4.51391965e-01 9.00690317e-01 -6.00608468e-01
2.89626330e-01 1.19399406e-01 8.92469212e-02 9.78877425e-01
7.38832891e-01 -1.08958341e-01 -1.81860283e-01 -4.87187207e-01
6.81725383e-01 -1.05457626e-01 -1.59078568e-01 -5.96573055e-01
-9.31655765e-01 9.08768594e-01 4.46487427e-01 1.48695529e-01
-1.63547859e-01 1.94237918e-01 3.25693190e-01 4.83688712e-01
6.53591037e-01 2.22022057e-01 -3.53335530e-01 2.26281151e-01
-7.59831905e-01 -5.73306866e-02 1.13820171e+00 9.96274054e-01
8.33888650e-01 -2.43866801e-01 3.65610003e-01 6.19788229e-01
3.11923563e-01 2.13062197e-01 -8.32716748e-02 -7.22548425e-01
3.61048251e-01 3.85579616e-01 -5.23180664e-02 -9.77800608e-01
-3.58397663e-01 -4.69326586e-01 -1.35062647e+00 -8.21071342e-02
8.04015219e-01 -2.58397877e-01 -2.85214156e-01 1.67626047e+00
3.75268072e-01 2.01910794e-01 -2.83765197e-01 3.91611099e-01
2.42847294e-01 4.16345209e-01 -1.84413388e-01 -5.64324915e-01
8.35113764e-01 -5.08968234e-01 -2.30853677e-01 1.00610994e-01
6.81113005e-01 -4.71682817e-01 6.03469968e-01 5.09420037e-01
-8.98699224e-01 -1.79612309e-01 -6.74194038e-01 3.18574041e-01
-1.69316411e-01 -1.56813443e-01 7.76516318e-01 7.13023841e-01
-1.25345445e+00 1.11447358e+00 -6.39514565e-01 -5.41199803e-01
5.87568820e-01 5.17135024e-01 -3.97986561e-01 7.09427744e-02
-7.97799885e-01 5.53588212e-01 2.77713597e-01 8.84165149e-03
-6.62141800e-01 -5.12413740e-01 -6.50512815e-01 1.57304015e-02
6.59320951e-01 -3.78437489e-01 7.94675589e-01 -8.75699103e-01
-1.21847105e+00 7.14530885e-01 1.45357428e-02 -5.91856539e-01
6.84443057e-01 2.70256817e-01 8.69655833e-02 -6.00369275e-02
-5.72106428e-02 1.99426040e-01 5.54862857e-01 -1.08904922e+00
-2.53783941e-01 -5.71535170e-01 -4.94046398e-02 -1.35721117e-01
-2.71414995e-01 -3.30478042e-01 -1.99583441e-01 -3.62505347e-01
4.87517230e-02 -9.69129324e-01 -5.34648657e-01 1.59463137e-01
-7.52724528e-01 -3.67203534e-01 1.44168302e-01 -5.39413810e-01
1.02051044e+00 -1.87138224e+00 4.37758803e-01 8.03152442e-01
5.93037248e-01 5.57173379e-02 8.32629274e-04 8.94711554e-01
-6.40758127e-02 2.08785996e-01 -4.08031583e-01 -2.71689713e-01
-1.32653654e-01 6.55622333e-02 1.01192482e-01 9.05347884e-01
-2.06514686e-01 5.74676573e-01 -8.91307116e-01 -5.00489593e-01
6.51832372e-02 2.62509286e-01 -3.61472368e-01 -5.89358360e-02
-2.37497482e-02 1.95336893e-01 -3.91361982e-01 1.34759918e-01
4.87158179e-01 -8.54796648e-01 7.13998377e-01 2.10505664e-01
3.50231856e-01 1.26456380e-01 -1.29635787e+00 1.07079613e+00
-2.96741039e-01 5.75274885e-01 1.67469665e-01 -1.42128503e+00
8.70403230e-01 9.36745331e-02 5.36442399e-01 -6.42625391e-02
1.77532777e-01 2.22664811e-02 -5.53227775e-02 -9.47282836e-02
2.05084905e-02 -2.09832653e-01 -2.31515598e-02 5.09527743e-01
-5.95694557e-02 1.69838995e-01 1.93193004e-01 4.26879734e-01
1.32878184e+00 -4.11797106e-01 5.69107711e-01 -3.71418118e-01
4.04252946e-01 -3.53874177e-01 1.71655312e-01 8.29507768e-01
1.03371479e-02 2.54654437e-01 1.16830611e+00 -1.51723742e-01
-1.07760823e+00 -1.08791006e+00 -2.29465038e-01 7.88695693e-01
1.66303769e-01 -5.87261796e-01 -8.95781338e-01 -7.70726323e-01
4.23403941e-02 2.84337074e-01 -8.33507538e-01 -4.28196117e-02
-2.62256742e-01 -7.82689154e-01 2.91519970e-01 1.88369974e-01
2.00556546e-01 -7.15988159e-01 3.11037749e-02 -5.01047308e-03
1.80907771e-01 -8.22126687e-01 -2.52758145e-01 1.08197525e-01
-1.20014822e+00 -1.42201090e+00 -3.95199984e-01 -7.53901601e-01
8.26170743e-01 2.11927947e-02 1.07902586e+00 3.08682829e-01
-1.85614467e-01 3.80104601e-01 -1.44471243e-01 -8.21559057e-02
-4.73690718e-01 4.13980514e-01 1.08581267e-01 -6.86478317e-02
7.10074678e-02 -8.01889062e-01 -4.64040339e-01 2.44195357e-01
-7.90669978e-01 -5.53173088e-02 6.39853835e-01 7.74482012e-01
7.15609014e-01 3.01278710e-01 5.06766140e-01 -1.38054752e+00
6.28179073e-01 -7.15662062e-01 -9.53291118e-01 4.29432273e-01
-8.77930880e-01 2.11675555e-01 9.00337279e-01 -3.66380513e-01
-3.45867842e-01 1.23098724e-01 1.02305949e-01 -4.61649448e-02
1.02217376e-01 4.57350492e-01 1.08240671e-01 -3.99082959e-01
4.93491679e-01 1.26259282e-01 3.27575266e-01 -3.78467560e-01
1.87286973e-01 4.13938195e-01 2.00181052e-01 -2.79478639e-01
8.63515735e-01 2.97208965e-01 7.04367459e-01 -9.53407526e-01
-8.90085161e-01 -3.51270974e-01 -7.39679217e-01 -2.40822613e-01
4.50042009e-01 -4.32255030e-01 -1.14318621e+00 1.58613950e-01
-7.89264560e-01 -3.28457117e-01 -1.62934050e-01 3.23068172e-01
-6.17266059e-01 6.41646802e-01 -5.76134622e-01 -9.97578025e-01
-7.46596381e-02 -5.82215905e-01 7.84416378e-01 -4.53923233e-02
-1.27350781e-02 -1.49716115e+00 2.41937697e-01 1.22844689e-01
7.91481882e-02 4.24249113e-01 1.00745404e+00 -6.99335396e-01
-5.30357003e-01 -3.71575862e-01 -2.07479373e-01 2.58892804e-01
5.84973283e-02 -1.36498839e-01 -3.99180621e-01 -4.87299800e-01
-3.55463892e-01 -9.81609300e-02 8.93751323e-01 4.08155262e-01
1.28585911e+00 -7.09313035e-01 -6.36018693e-01 3.11053991e-01
1.64890635e+00 -4.81017888e-01 4.03661460e-01 -3.05673659e-01
7.29906499e-01 6.72047317e-01 2.34428808e-01 4.71117526e-01
2.57210821e-01 5.08286953e-01 4.85006720e-01 1.58898264e-01
1.50514573e-01 -4.63815987e-01 -2.53922194e-02 8.44491303e-01
-1.01055451e-01 -2.96057999e-01 -7.97124743e-01 3.29027057e-01
-1.95097065e+00 -7.65381992e-01 -2.20754713e-01 2.76304030e+00
7.20880687e-01 2.11523548e-01 5.56904852e-01 2.73334146e-01
1.04789805e+00 -1.77890677e-02 -5.56750894e-01 -7.38572627e-02
1.64131060e-01 2.37207755e-01 9.08235848e-01 9.51751769e-01
-8.40783417e-01 5.51446617e-01 6.80560970e+00 9.67658520e-01
-5.61817169e-01 -1.85637757e-01 9.64182436e-01 9.40811262e-02
-2.13648677e-01 1.72537029e-01 -6.32569849e-01 5.04058361e-01
1.02669668e+00 -2.11098492e-01 7.64159501e-01 6.77335680e-01
-1.74638927e-01 -2.46209413e-01 -1.19022763e+00 8.55958879e-01
-5.39522469e-02 -1.21573162e+00 -2.08410814e-01 5.56639016e-01
7.85404801e-01 -2.02740007e-03 -1.75617233e-01 8.24697781e-03
5.51550448e-01 -1.08285773e+00 1.27659500e-01 5.39005756e-01
7.21999347e-01 -8.82147253e-01 6.59176350e-01 6.43689036e-01
-1.10108745e+00 8.73765126e-02 -5.50166726e-01 -9.40099135e-02
-1.27002031e-01 1.07517731e+00 -1.06108952e+00 5.75697482e-01
3.88328493e-01 7.17866063e-01 -5.39633572e-01 1.24219060e+00
-1.47108380e-02 7.67516375e-01 -6.42688513e-01 -5.60737610e-01
-1.67203262e-01 -5.27159810e-01 4.66259062e-01 7.78997898e-01
1.00200236e-01 -3.42586376e-02 2.21524276e-02 4.61792797e-01
-4.80034977e-01 1.49522141e-01 -7.48217762e-01 1.04890607e-01
5.44408798e-01 1.35219789e+00 -1.03598046e+00 -1.71437133e-02
-2.23431274e-01 6.88809633e-01 9.28087890e-01 1.36211187e-01
-5.27599275e-01 -4.31587577e-01 1.46885470e-01 5.90386033e-01
4.24556732e-01 -2.41529718e-01 -2.85877176e-02 -7.80371547e-01
8.20185021e-02 -3.38981599e-01 4.65360224e-01 -3.50821055e-02
-1.72031939e+00 2.70624399e-01 1.31627515e-01 -8.96764576e-01
-1.99056491e-01 -6.59615934e-01 -4.89816040e-01 5.34829021e-01
-1.14482522e+00 -5.72435915e-01 6.46010116e-02 2.82525778e-01
-4.58354801e-02 1.66549414e-01 6.98924899e-01 9.41123664e-02
-5.78136683e-01 5.83200514e-01 5.23459077e-01 1.26338944e-01
1.82174891e-01 -1.52242160e+00 3.00725013e-01 5.93870699e-01
2.84293383e-01 1.73031926e-01 7.06829786e-01 -6.05457246e-01
-1.26687455e+00 -9.10396695e-01 9.93216634e-01 -3.15613180e-01
9.00314689e-01 -5.13639867e-01 -8.14461887e-01 7.07984805e-01
-2.53295779e-01 2.82448053e-01 6.62969351e-01 3.98514777e-01
-1.83099464e-01 -2.42908329e-01 -1.24035811e+00 3.95983607e-01
1.17905843e+00 -2.76882350e-01 1.35509804e-01 6.94039464e-01
2.95968831e-01 5.37606515e-02 -1.09270298e+00 2.04163149e-01
3.72895032e-01 -9.83670056e-01 8.02007794e-01 -6.15630090e-01
2.09708035e-01 6.88968673e-02 1.93494651e-03 -1.18201828e+00
-2.64054030e-01 -8.86205018e-01 -1.43654540e-01 1.06730115e+00
4.76055145e-01 -7.31132507e-01 1.05725026e+00 1.35082752e-01
6.86926186e-01 -9.63966966e-01 -1.14960134e+00 -7.65424132e-01
2.61630923e-01 -8.96538273e-02 2.31490731e-01 8.63660276e-01
6.27855286e-02 5.54861486e-01 -3.67332608e-01 3.29938307e-02
1.07467616e+00 -9.76894423e-02 7.48379290e-01 -1.76415789e+00
-6.25900984e-01 -4.79263425e-01 -6.68070614e-01 -1.01395094e+00
2.98598826e-01 -1.07357252e+00 -8.20961148e-02 -1.48679411e+00
3.61270607e-01 -6.32846117e-01 2.76048873e-02 7.31746182e-02
-6.16861731e-02 9.25621912e-02 -2.13557750e-01 1.21382691e-01
-9.36572790e-01 4.25935686e-01 1.12186730e+00 2.34161869e-01
7.63979554e-02 6.11239731e-01 -6.40465438e-01 7.57352412e-01
6.62052631e-01 -5.36144078e-01 -3.15896869e-01 2.16020867e-02
5.80013394e-01 3.95306438e-01 3.53090852e-01 -7.91217089e-01
2.64867038e-01 5.51148281e-02 2.68267542e-01 -2.33577564e-01
1.11877739e-01 -7.08672047e-01 1.22973286e-01 5.56338131e-01
-6.04529321e-01 4.08693962e-02 -2.37575009e-01 1.08809555e+00
2.20964834e-01 -5.34170091e-01 8.31165850e-01 -5.78145012e-02
2.71519311e-02 4.90620971e-01 -2.41292998e-01 1.51602656e-01
1.13993776e+00 2.98260543e-02 -2.60654509e-01 -6.59599423e-01
-7.29452133e-01 1.87737152e-01 3.16123188e-01 -3.01398516e-01
4.68398422e-01 -1.07672656e+00 -7.38957942e-01 -1.18000023e-01
-1.41484305e-01 1.38014853e-01 -6.13802066e-03 9.93650019e-01
-4.83537018e-01 3.71816382e-02 2.56953895e-01 -5.34147441e-01
-1.23811460e+00 5.24029136e-01 3.13014358e-01 -4.84162301e-01
-4.26762670e-01 7.99553275e-01 -2.15135142e-01 -3.95046413e-01
3.01448196e-01 2.10831851e-01 2.70463191e-02 -1.37280867e-01
2.27767646e-01 6.85327828e-01 -1.41935408e-01 -3.11963290e-01
-2.90675819e-01 4.38643873e-01 -2.52337128e-01 2.03854128e-04
1.41815901e+00 -1.23657718e-01 -3.02111596e-01 6.15389407e-01
1.22173679e+00 9.18146521e-02 -1.13614583e+00 -5.02606988e-01
2.63682216e-01 -3.25334132e-01 -2.47503921e-01 -2.87171423e-01
-1.03416693e+00 5.97208560e-01 2.46135637e-01 9.68424082e-01
9.18102145e-01 6.15535736e-01 3.10975611e-01 5.38132131e-01
4.93830323e-01 -8.27201962e-01 -7.28263035e-02 2.09911928e-01
4.84370500e-01 -1.30109894e+00 3.31411451e-01 -6.48451507e-01
-2.09170431e-01 1.03268266e+00 2.22948879e-01 -5.70016325e-01
1.13732886e+00 -3.25775053e-03 -7.99082518e-01 -1.80856243e-01
-7.48175561e-01 -6.56042323e-02 2.03280658e-01 5.92773080e-01
3.20646733e-01 2.33526230e-01 -2.85939664e-01 6.75035566e-02
-3.54821414e-01 -3.41988593e-01 5.76984763e-01 4.00375068e-01
-4.85759407e-01 -1.21647811e+00 2.15395793e-01 1.03505909e+00
-3.61251295e-01 -2.62569208e-02 -6.68791115e-01 7.62050509e-01
-4.28269714e-01 7.53077626e-01 6.03007078e-02 -3.00435454e-01
-4.25428972e-02 -2.21463859e-01 7.41267145e-01 -4.96124446e-01
-2.85281509e-01 -2.73414522e-01 1.09298721e-01 -2.73992568e-01
-3.51749718e-01 -8.96476746e-01 -8.60508978e-01 -8.38515937e-01
-5.04157424e-01 3.89205009e-01 5.56137264e-01 9.01657581e-01
1.08732238e-01 1.39950499e-01 8.64228904e-01 -5.15382409e-01
-7.00741708e-01 -9.19050276e-01 -9.52627361e-01 1.14748783e-01
2.91494787e-01 -4.09737885e-01 -8.05404305e-01 -3.79200548e-01] | [6.942177772521973, 5.341022968292236] |
01738b31-9a96-4834-ae54-6b94f7ea0079 | left-corner-transitions-on-dependency-parsing | null | null | https://aclanthology.org/C14-1202 | https://aclanthology.org/C14-1202.pdf | Left-corner Transitions on Dependency Parsing | null | ['Hiroshi Noji', 'Yusuke Miyao'] | 2014-08-01 | left-corner-transitions-on-dependency-parsing-1 | https://aclanthology.org/C14-1202 | https://aclanthology.org/C14-1202.pdf | coling-2014-8 | ['transition-based-dependency-parsing'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.3298020362854, 3.778496265411377] |
fc0d2e6f-ebd0-4639-8db2-0ee8bc030287 | excess-risk-bound-for-deep-learning-under | 2302.07503 | null | https://arxiv.org/abs/2302.07503v1 | https://arxiv.org/pdf/2302.07503v1.pdf | Excess risk bound for deep learning under weak dependence | This paper considers deep neural networks for learning weakly dependent processes in a general framework that includes, for instance, regression estimation, time series prediction, time series classification. The $\psi$-weak dependence structure considered is quite large and covers other conditions such as mixing, association,$\ldots$ Firstly, the approximation of smooth functions by deep neural networks with a broad class of activation functions is considered. We derive the required depth, width and sparsity of a deep neural network to approximate any H\"{o}lder smooth function, defined on any compact set $\mx$. Secondly, we establish a bound of the excess risk for the learning of weakly dependent observations by deep neural networks. When the target function is sufficiently smooth, this bound is close to the usual $\mathcal{O}(n^{-1/2})$. | ['William Kengne'] | 2023-02-15 | null | null | null | null | ['time-series-prediction'] | ['time-series'] | [-1.05940923e-01 7.95452967e-02 -6.16393127e-02 -3.96431178e-01
-7.36003697e-01 -2.46816948e-01 2.30146617e-01 7.88183287e-02
-4.65459883e-01 1.00978029e+00 -2.03631774e-01 -3.95984113e-01
-6.61245823e-01 -9.18478072e-01 -9.98869300e-01 -1.16852915e+00
-8.29445899e-01 3.71703595e-01 -2.01820374e-01 -2.76302606e-01
-1.16890728e-01 5.35895228e-01 -1.12912774e+00 -4.74109441e-01
5.97860396e-01 1.68320251e+00 -3.66221547e-01 8.50053012e-01
1.85148731e-01 8.44564676e-01 -4.77822423e-01 -6.07955575e-01
5.43479979e-01 -2.54444927e-01 -3.80011261e-01 -3.03513825e-01
7.25580901e-02 -1.61674604e-01 -5.55281699e-01 1.22966516e+00
5.93703806e-01 5.34943342e-01 1.20260894e+00 -8.60339046e-01
-5.73687375e-01 7.87816107e-01 -6.05529189e-01 6.56399846e-01
-5.85974514e-01 1.43916681e-01 6.24522984e-01 -6.31411374e-01
-1.13322914e-01 9.80355620e-01 1.06541800e+00 3.15950036e-01
-1.10668349e+00 -1.02235162e+00 1.18169442e-01 -2.16592297e-01
-1.24983501e+00 -1.41367480e-01 6.34704947e-01 -5.32096148e-01
5.39927185e-01 1.38424546e-01 7.05809295e-02 1.20893300e+00
3.89344335e-01 2.87193269e-01 8.95278156e-01 -2.06507802e-01
4.46632564e-01 2.69190799e-02 4.61822540e-01 5.74991584e-01
3.07096332e-01 2.39404365e-01 7.08755702e-02 -1.93144917e-01
1.01551032e+00 2.27041259e-01 -1.16033308e-01 2.02314556e-01
-6.42417252e-01 9.78009999e-01 6.13083988e-02 4.95856553e-01
-5.40982068e-01 5.33063233e-01 3.55687767e-01 7.13923156e-01
8.82698894e-01 2.71507829e-01 -8.77408028e-01 1.59671441e-01
-6.05057478e-01 5.97387627e-02 1.00193620e+00 8.97367299e-01
3.31072837e-01 5.54007471e-01 -2.23141044e-01 6.39282227e-01
-2.32207309e-02 7.03014076e-01 2.31221821e-02 -9.96526659e-01
6.31637335e-01 -1.67972833e-01 2.23364666e-01 -5.68628669e-01
-4.96970385e-01 -5.87796509e-01 -1.62206757e+00 9.91103575e-02
6.66661620e-01 -9.17639077e-01 -6.37485206e-01 2.04972935e+00
-9.50561929e-03 1.35805815e-01 -2.06702556e-02 3.21952581e-01
2.01092824e-01 8.53793025e-01 1.55835837e-01 -8.12796652e-01
8.55729282e-01 -3.01090956e-01 -6.34314656e-01 1.75188348e-01
2.10287884e-01 -3.38810176e-01 7.20181048e-01 4.77802426e-01
-1.45303226e+00 -7.77577877e-01 -8.19842398e-01 2.79141665e-01
-1.98130310e-01 -3.48012269e-01 5.73995411e-01 6.04569197e-01
-1.00609004e+00 9.82826650e-01 -7.29111493e-01 1.61874667e-01
6.96625292e-01 6.96761787e-01 -7.01940805e-02 1.00372463e-01
-1.30564010e+00 5.54353058e-01 1.81133837e-01 1.62467271e-01
-1.16801655e+00 -8.82104456e-01 -5.25182962e-01 2.29583874e-01
1.70774348e-02 -4.12815154e-01 1.03634369e+00 -8.85135055e-01
-1.38726568e+00 6.18634999e-01 2.06466153e-01 -8.65460753e-01
6.79651558e-01 -2.58081675e-01 -5.75516045e-01 8.99369828e-03
-8.46829191e-02 -1.45145029e-01 1.33580709e+00 -6.93940997e-01
-5.51786244e-01 -6.19092286e-01 -4.04073950e-03 -3.23855579e-01
-2.47325569e-01 1.29308283e-01 2.40511045e-01 -7.03699887e-01
-3.45408946e-01 -7.32121587e-01 -6.64294899e-01 -2.07336515e-01
-1.14190474e-01 -4.66615051e-01 4.78924245e-01 -5.89078724e-01
1.05484056e+00 -2.29952908e+00 -1.51069298e-01 4.24969763e-01
1.88783810e-01 -8.18668082e-02 2.40990117e-01 2.78766274e-01
-4.96298015e-01 7.41130933e-02 -5.02922058e-01 -4.79958743e-01
1.02176085e-01 7.60448202e-02 -4.00464863e-01 1.00184190e+00
-1.25099979e-02 6.34478569e-01 -3.33619297e-01 4.33033444e-02
6.58664629e-02 3.08124393e-01 -1.73539326e-01 2.82390237e-01
-3.32638383e-01 5.68186700e-01 -2.65885562e-01 2.93593228e-01
8.00077856e-01 -3.99871916e-01 -2.27608129e-01 2.97688663e-01
-1.63738728e-01 -3.88883382e-01 -1.07484388e+00 1.10555780e+00
-6.60692215e-01 5.82808137e-01 3.71927500e-01 -1.91015041e+00
1.05839515e+00 6.20400786e-01 7.02298343e-01 -2.37267569e-01
6.02613807e-01 3.43235433e-01 3.71016981e-03 -3.63885045e-01
-3.29679221e-01 -9.64112520e-01 -1.83952287e-01 1.95761636e-01
1.83904037e-01 1.13300472e-01 -5.64452447e-02 -5.45509219e-01
1.11961782e+00 -3.75164062e-01 1.62928045e-01 -7.88995683e-01
5.49739659e-01 -6.09624982e-01 2.76399910e-01 8.51137996e-01
-2.90659368e-01 1.91320941e-01 8.13046932e-01 -4.69421655e-01
-1.00498092e+00 -1.21633720e+00 -5.25614083e-01 1.37379944e+00
-1.70412272e-01 5.21036029e-01 -4.37303960e-01 -4.37174141e-01
4.59238589e-02 7.40293622e-01 -9.73554730e-01 -3.78280818e-01
-7.35949218e-01 -1.24954772e+00 7.34463096e-01 8.02394867e-01
5.59090495e-01 -1.19252264e+00 -6.99429661e-02 2.84185559e-01
3.70985866e-01 -7.23765850e-01 -1.91459090e-01 1.01716983e+00
-1.02263546e+00 -6.99220836e-01 -1.09390807e+00 -9.67954278e-01
4.32672799e-01 -4.40048993e-01 1.03128147e+00 -6.68069303e-01
7.80310258e-02 2.54030615e-01 2.62610435e-01 -6.76775753e-01
-1.90966412e-01 -1.29555389e-01 3.91443729e-01 1.11237220e-01
3.43013465e-01 -9.03036177e-01 -7.75183380e-01 1.36464238e-01
-9.06321049e-01 -1.05108392e+00 3.58022034e-01 6.11003697e-01
5.84423125e-01 5.73394239e-01 8.28394830e-01 -6.15952551e-01
6.86373770e-01 -7.50159800e-01 -7.25018322e-01 -6.93165809e-02
-3.43973547e-01 1.28509805e-01 1.11953866e+00 -7.05215394e-01
-1.03239202e+00 -2.98060685e-01 -3.78869265e-01 -6.54664099e-01
-2.25009903e-01 4.12460685e-01 2.59275809e-02 1.96446925e-01
8.35715771e-01 1.22903243e-01 -1.44298658e-01 -5.26801169e-01
2.06127375e-01 3.04835856e-01 5.32333434e-01 -6.51354611e-01
6.22736156e-01 6.58620477e-01 5.19538641e-01 -9.24742103e-01
-9.89112735e-01 -1.19738996e-01 -3.71765912e-01 1.71333492e-01
8.74146104e-01 -8.11835706e-01 -1.28450525e+00 3.62665772e-01
-9.50836897e-01 -6.09949291e-01 -8.06815028e-01 9.19719219e-01
-9.68894899e-01 1.88183170e-02 -1.01817250e+00 -1.23246396e+00
-3.86649698e-01 -5.28530240e-01 3.19348991e-01 -9.14173201e-02
2.49765255e-02 -1.37773263e+00 1.52706146e-01 -1.17573224e-01
4.02111322e-01 4.96415794e-01 1.14883792e+00 -1.03553867e+00
2.04643253e-02 -6.18610799e-01 -8.85497630e-02 1.01811981e+00
-9.10173208e-02 -1.50245458e-01 -9.01998341e-01 -2.18479559e-01
9.41762984e-01 -1.49229527e-01 8.85960281e-01 1.23106349e+00
1.67264950e+00 -5.87461174e-01 -4.61948700e-02 7.76881754e-01
1.40907848e+00 6.33730292e-01 4.61759925e-01 -1.59308597e-01
4.76738632e-01 6.18375361e-01 -5.03669903e-02 5.89131653e-01
-2.68769592e-01 -3.23991865e-01 5.20348907e-01 -5.62722497e-02
6.72293544e-01 3.08385730e-01 5.34128487e-01 9.71839666e-01
-4.30166304e-01 -2.36543953e-01 -6.09007657e-01 4.46385086e-01
-1.65635967e+00 -1.04921520e+00 -2.87259847e-01 2.39388537e+00
5.93103945e-01 4.79321808e-01 1.72688410e-01 5.52572794e-02
1.02036667e+00 2.41035447e-02 -5.38417399e-01 -5.08623481e-01
-1.78866029e-01 1.08632100e+00 8.21538031e-01 4.63110328e-01
-1.31607699e+00 3.48238707e-01 6.58958578e+00 1.01112449e+00
-1.08222389e+00 3.11339587e-01 1.06743157e+00 -1.56950682e-01
4.70513590e-02 -5.93972147e-01 -6.38995171e-01 6.13284945e-01
1.38305974e+00 -9.68908295e-02 5.40773906e-02 8.90991628e-01
9.12445635e-02 4.48931724e-01 -1.06318104e+00 9.20295477e-01
-4.30470318e-01 -1.06310773e+00 -4.33498055e-01 2.59541780e-01
9.28327680e-01 -1.33563634e-02 6.21990681e-01 7.58622706e-01
7.44622886e-01 -1.38922763e+00 2.75249511e-01 5.03106236e-01
8.70313585e-01 -1.34774280e+00 8.28645170e-01 5.00370502e-01
-1.11801839e+00 -3.66039604e-01 -4.38468635e-01 -1.72167867e-01
2.92349085e-02 7.62378931e-01 -7.48722777e-02 3.19284081e-01
6.76434875e-01 4.66796696e-01 2.56023049e-01 6.52511656e-01
2.57048696e-01 7.86062121e-01 -5.38402736e-01 -2.95068771e-02
4.68096793e-01 -4.09216374e-01 4.00200874e-01 1.27638805e+00
6.76589429e-01 2.59727478e-01 9.22995526e-03 5.29616654e-01
-2.41701841e-01 3.34333256e-02 -8.58987749e-01 1.02520466e-01
-1.07601017e-01 6.93639457e-01 -5.82720160e-01 -3.07775408e-01
-2.33741626e-01 6.91216230e-01 1.29431471e-01 5.31721056e-01
-9.69044209e-01 -4.52214420e-01 6.22476280e-01 9.18261558e-02
5.28126001e-01 -2.48903781e-01 -5.51962793e-01 -9.43197072e-01
-2.51633059e-02 -3.50381136e-01 7.61903405e-01 -4.06921953e-02
-1.92850804e+00 5.33058226e-01 1.12913279e-02 -1.07081270e+00
-1.59817189e-01 -8.36713731e-01 -6.72217727e-01 1.04074645e+00
-9.89256620e-01 -4.13077325e-01 3.41267318e-01 1.03170538e+00
4.23566133e-01 -3.16209555e-01 5.84164023e-01 2.38597974e-01
-6.94317579e-01 3.37518692e-01 5.89558244e-01 8.68511349e-02
2.41230056e-01 -1.28996146e+00 9.38179940e-02 6.98092461e-01
-3.19342375e-01 4.53949481e-01 8.72186542e-01 -4.32695508e-01
-9.74416018e-01 -1.13237655e+00 3.00496429e-01 -3.92129660e-01
1.00887573e+00 -1.88314930e-01 -9.75451291e-01 7.76455641e-01
1.89796403e-01 3.48659933e-01 8.35127294e-01 7.02709779e-02
-1.10552542e-01 -4.15046036e-01 -1.15082216e+00 3.96038592e-01
8.24560106e-01 -3.43873441e-01 -3.73885423e-01 4.15755093e-01
6.52508259e-01 1.07838959e-01 -1.31977117e+00 6.63208842e-01
2.90487051e-01 -8.49299908e-01 1.01925933e+00 -1.10612369e+00
4.47464496e-01 3.57549906e-01 -4.38292950e-01 -1.07824206e+00
-4.92068708e-01 -1.01778626e+00 -5.20702481e-01 7.33115911e-01
4.79072660e-01 -7.25757003e-01 5.93879163e-01 4.73575145e-01
-1.19842969e-01 -6.63136661e-01 -1.29058444e+00 -9.20694709e-01
7.52500057e-01 -6.38816059e-01 2.07177788e-01 1.00941396e+00
-3.45017135e-01 2.35828027e-01 -4.37799484e-01 1.92194536e-01
8.63396943e-01 -2.56347179e-01 2.16300398e-01 -1.53647935e+00
-4.95677710e-01 -6.98626041e-01 3.94218564e-02 -1.03875971e+00
3.58022451e-01 -6.70091212e-01 1.19344816e-01 -1.00636184e+00
-3.95549029e-01 -6.00202084e-01 -8.24695826e-01 -3.23013395e-01
7.01002032e-03 -2.68079281e-01 -2.42515087e-01 7.43852556e-02
-2.88535565e-01 6.20913982e-01 1.18088901e+00 6.64140657e-02
-1.54458374e-01 9.44215596e-01 -5.95282853e-01 1.02719784e+00
9.61995959e-01 -3.74597192e-01 -3.67853880e-01 3.41478549e-02
1.75385907e-01 7.58826554e-01 1.72725290e-01 -9.52125847e-01
-1.03504464e-01 -1.52833179e-01 5.41350067e-01 -2.54696548e-01
4.16452348e-01 -1.01434004e+00 -1.16226889e-01 6.72877789e-01
-7.19560325e-01 -5.10791428e-02 6.49501905e-02 9.12213683e-01
3.60204955e-03 -3.63176584e-01 1.14239073e+00 -2.78826445e-01
-2.70771682e-02 9.04486358e-01 -3.34854722e-01 2.39132166e-01
9.84486282e-01 2.65460998e-01 8.18576589e-02 -7.36829162e-01
-9.95318115e-01 -9.39352810e-02 -5.18142641e-01 -1.81545705e-01
1.53645083e-01 -1.23273373e+00 -7.32483149e-01 3.76843326e-02
-2.66866386e-01 2.07459554e-01 3.10515344e-01 9.84472930e-01
-2.90479034e-01 8.46156254e-02 -6.45775720e-02 -3.32493126e-01
-3.50528210e-01 9.07025814e-01 5.17202795e-01 -3.83258015e-01
-4.70852256e-01 1.12242091e+00 4.99155939e-01 8.11484680e-02
3.88376087e-01 -5.86547911e-01 -6.94380850e-02 -2.22877786e-02
4.66604680e-01 8.03018510e-01 -8.02587420e-02 -1.08739346e-01
7.08943186e-03 2.45422065e-01 1.69166774e-01 -1.92170605e-01
1.49153316e+00 3.87382992e-02 -2.16944888e-01 9.58960950e-01
1.63573635e+00 -2.17070535e-01 -1.64166605e+00 -1.99844375e-01
-1.00990437e-01 2.90133059e-01 -3.89478922e-01 -1.82438239e-01
-1.00947654e+00 1.10065198e+00 7.04905510e-01 9.36452389e-01
9.19131875e-01 -7.77424723e-02 6.38859332e-01 4.65956360e-01
5.73096462e-02 -1.12765896e+00 6.48927838e-02 7.49221861e-01
6.87276244e-01 -1.14330816e+00 -4.74126279e-01 1.00245483e-01
-3.57741624e-01 1.17861748e+00 4.18248743e-01 -7.76647627e-01
1.45687556e+00 2.92358041e-01 -4.02849019e-01 -6.70925304e-02
-5.73446691e-01 -1.97404802e-01 1.63617283e-01 4.82225925e-01
3.21827441e-01 1.49436697e-01 -1.87465519e-01 7.35119641e-01
-1.36151016e-02 -6.35555536e-02 7.94239715e-02 4.12593812e-01
-5.13373911e-01 -3.45814377e-01 -2.74964273e-01 6.40392065e-01
-1.14701521e+00 -6.40315115e-02 3.93080711e-01 8.00452173e-01
1.16707906e-01 7.83010364e-01 3.63854349e-01 1.53788358e-01
1.53478205e-01 1.61630139e-01 3.66845608e-01 -2.43809894e-01
-5.00728190e-01 1.07171088e-01 -2.74938047e-01 7.48589216e-03
-2.00492650e-01 -3.66212308e-01 -8.87390673e-01 -4.43040580e-01
6.39024749e-02 3.52861993e-02 3.55792493e-01 1.06575179e+00
-2.03342646e-01 3.07806283e-01 9.99967217e-01 -5.47713041e-01
-9.11211491e-01 -1.25189531e+00 -1.07495821e+00 3.13099563e-01
7.22870588e-01 -4.14016306e-01 -8.25866699e-01 -6.06931262e-02] | [7.189160346984863, 3.8323276042938232] |
b1a7fb49-9cd7-40e3-b3ae-df98c30b4b46 | soft-gripping-specifying-for-trustworthiness | 2307.01159 | null | https://arxiv.org/abs/2307.01159v1 | https://arxiv.org/pdf/2307.01159v1.pdf | Soft Gripping: Specifying for Trustworthiness | Soft robotics is an emerging technology in which engineers create flexible devices for use in a variety of applications. In order to advance the wide adoption of soft robots, ensuring their trustworthiness is essential; if soft robots are not trusted, they will not be used to their full potential. In order to demonstrate trustworthiness, a specification needs to be formulated to define what is trustworthy. However, even for soft robotic grippers, which is one of the most mature areas in soft robotics, the soft robotics community has so far given very little attention to formulating specifications. In this work, we discuss the importance of developing specifications during development of soft robotic systems, and present an extensive example specification for a soft gripper for pick-and-place tasks for grocery items. The proposed specification covers both functional and non-functional requirements, such as reliability, safety, adaptability, predictability, ethics, and regulations. We also highlight the need to promote verifiability as a first-class objective in the design of a soft gripper. | ['Shane Windsor', 'Kerstin Eder', 'Jonathan Rossiter', 'Graham Deacon', 'John Downer', 'Jonathan Ives', 'Alix J. Partridge', 'Arianna Manzini', 'Peter D. Winter', 'Greg Chance', 'Nguyen Hao Le', 'Dhaminda B. Abeywickrama'] | 2023-07-03 | null | null | null | null | ['ethics'] | ['miscellaneous'] | [-4.02152129e-02 5.02145469e-01 -3.52570936e-02 -3.68442893e-01
5.92542946e-01 -8.16349566e-01 2.25574404e-01 -2.13668749e-01
-1.78539138e-02 7.57902861e-01 -3.14475119e-01 -6.39728010e-02
-5.56194007e-01 -7.33103991e-01 -4.62239087e-01 -5.98336875e-01
1.25638142e-01 2.98461884e-01 6.02200449e-01 -5.39769769e-01
8.87016878e-02 2.84988940e-01 -1.32433176e+00 -2.58064628e-01
8.13444018e-01 1.10977578e+00 5.86584210e-01 5.78213297e-02
3.71754855e-01 6.50811732e-01 -3.56038898e-01 -3.53179991e-01
2.83574998e-01 6.71633333e-02 -8.53788495e-01 -2.27345124e-01
-6.63898706e-01 -5.15930474e-01 1.30975112e-01 9.89409089e-01
-6.77940100e-02 -3.45417172e-01 4.58919227e-01 -1.49708366e+00
-6.85200334e-01 9.41807508e-01 2.31039450e-01 -8.84073257e-01
4.45911944e-01 4.00707386e-02 6.70181692e-01 -3.78469914e-01
8.13301444e-01 7.08920896e-01 2.74801731e-01 4.45641786e-01
-5.45746565e-01 -4.18043643e-01 -6.44608438e-02 -3.94669563e-01
-8.85270774e-01 -2.72505730e-01 5.31773984e-01 -3.56657952e-01
6.99337125e-01 1.46728635e-01 5.50253808e-01 9.89328027e-01
1.16703212e+00 2.62157649e-01 1.09206760e+00 -3.41143250e-01
5.85053623e-01 4.12726611e-01 1.82593450e-01 1.15107454e-01
9.88159716e-01 1.00112408e-01 9.67836678e-02 2.91765798e-02
7.59869814e-01 -4.93140966e-02 1.02616018e-02 -7.16374934e-01
-1.03260040e+00 2.57737577e-01 1.66617736e-01 7.22517669e-01
-2.51384884e-01 3.57611299e-01 3.01765025e-01 2.86039114e-01
-1.84419498e-01 7.07283914e-01 -3.94433379e-01 -5.58202207e-01
-7.12831467e-02 1.90802753e-01 1.39048052e+00 1.22997856e+00
-9.94445384e-02 6.10132031e-02 5.61682105e-01 5.19953728e-01
6.70193672e-01 5.88356316e-01 -8.78910571e-02 -8.87131155e-01
4.82377931e-02 6.78949356e-01 7.54481256e-01 -7.88257897e-01
-3.17541778e-01 -3.15066241e-02 -3.09789658e-01 7.23274350e-01
-5.99806570e-02 -1.13429710e-01 -6.22144401e-01 1.17696655e+00
8.96095335e-02 -1.09411824e+00 2.69250184e-01 1.07668805e+00
5.74860334e-01 4.60782647e-01 -1.18626885e-01 -1.59309357e-01
9.40533578e-01 -3.60975146e-01 -8.22702527e-01 -1.39858246e-01
1.18689351e-01 -8.77091765e-01 7.75305510e-01 7.25715876e-01
-9.72437024e-01 -4.17667598e-01 -1.44800806e+00 4.05746013e-01
-2.17532381e-01 -9.10572112e-02 7.42754400e-01 6.00045323e-01
-6.05410516e-01 6.97161019e-01 -9.40647125e-01 -5.79179525e-01
-3.64606321e-01 6.48253977e-01 -3.26369792e-01 3.00264321e-02
-1.22865355e+00 1.63769066e+00 5.12079477e-01 3.16363841e-01
-5.05184412e-01 -4.24598232e-02 -5.61412215e-01 -3.01706284e-01
3.54046762e-01 -1.60984844e-01 1.40060687e+00 -6.67803824e-01
-2.08366013e+00 4.87929791e-01 9.93202806e-01 -8.53113011e-02
4.85480785e-01 -4.54914391e-01 -5.70052624e-01 -1.09181050e-02
1.29681677e-02 1.42140806e-01 7.93322325e-01 -1.36604476e+00
-5.18422663e-01 3.10514390e-01 5.54804504e-01 -3.07619154e-01
-1.11490168e-01 2.78411448e-01 3.52964282e-01 -2.45517150e-01
5.80770969e-02 -1.47901833e+00 -2.35017519e-02 2.22172320e-01
-1.35556608e-01 -3.26981664e-01 1.15764070e+00 -2.46305674e-01
5.80289900e-01 -2.30197573e+00 1.32364616e-01 4.91059899e-01
3.92736122e-02 4.29803044e-01 3.78723800e-01 6.46867394e-01
4.59093243e-01 -2.48047456e-01 6.26394409e-04 3.24089378e-01
5.10565162e-01 6.98307157e-01 -3.41415942e-01 4.92959708e-01
1.35418653e-01 6.09454155e-01 -1.01098013e+00 -1.29667416e-01
4.13100600e-01 1.93987042e-01 6.61431625e-02 2.35331237e-01
-2.65518636e-01 -5.01439571e-02 -9.62371886e-01 1.03430367e+00
6.14192426e-01 1.16925091e-01 3.02208364e-01 1.05294310e-01
-4.92442280e-01 -3.54165807e-02 -1.15903127e+00 1.01022828e+00
-4.02131885e-01 -5.51276244e-02 6.71676695e-01 -4.77047950e-01
1.31362069e+00 5.90982974e-01 6.07752323e-01 -5.64382911e-01
5.51244676e-01 9.90902305e-01 1.62713155e-01 -8.35565805e-01
9.05333579e-01 -2.94582844e-01 -5.27842939e-01 4.66204345e-01
-2.66782433e-01 -8.42827260e-01 1.94169417e-01 -2.38235250e-01
8.87250066e-01 8.74289870e-01 2.83538610e-01 -6.23909414e-01
9.22533348e-02 -1.55810833e-01 5.96821547e-01 5.86139597e-02
-8.22028443e-02 1.60220519e-01 3.06988180e-01 -3.93147022e-01
-1.03880572e+00 -8.99726987e-01 -1.46426842e-01 7.46856034e-01
7.56420553e-01 -1.39511883e-01 -6.03917956e-01 -3.71046960e-01
3.57924342e-01 5.55226743e-01 -2.23118722e-01 -2.06121027e-01
-3.89123112e-01 1.41087577e-01 1.88590795e-01 5.79792380e-01
3.36529434e-01 -1.01687026e+00 -1.55094087e+00 5.33639610e-01
2.60408640e-01 -1.22929025e+00 1.09186850e-01 4.87874836e-01
-3.70627612e-01 -1.01945829e+00 -4.48458225e-01 -6.44001901e-01
8.12145114e-01 3.22939217e-01 3.86676192e-01 2.90621102e-01
7.82415196e-02 4.27650154e-01 -1.06203866e+00 -5.69282770e-01
-8.81998360e-01 4.65934277e-02 6.05218828e-01 -7.00534523e-01
-2.37433463e-01 -4.59071130e-01 -2.29569808e-01 9.07377243e-01
-9.18855011e-01 -8.59810412e-02 8.87432516e-01 3.93022746e-01
7.07592443e-02 2.59977549e-01 6.61955416e-01 -3.16655666e-01
7.02288270e-01 -3.08652878e-01 -6.17408037e-01 4.24409926e-01
-4.65154648e-01 -1.81545556e-01 6.27389908e-01 -4.31928694e-01
-9.59002197e-01 1.38753399e-01 3.73876877e-02 3.88750285e-01
-4.59533483e-02 6.88993752e-01 -1.46040410e-01 -4.12107825e-01
4.41188157e-01 -3.70085120e-01 3.07024032e-01 -7.09725395e-02
1.37630468e-02 1.11087275e+00 2.57467479e-01 -1.02751803e+00
7.39980817e-01 1.29571334e-01 4.13123630e-02 -7.51985848e-01
1.82335377e-02 1.11648723e-01 -3.44338000e-01 -5.61097443e-01
5.65001369e-01 -1.93512917e-01 -9.07384038e-01 2.89005786e-01
-7.84469068e-01 -3.89746189e-01 -2.27398232e-01 4.65665817e-01
-6.90818667e-01 3.32697928e-01 -4.16617811e-01 -9.35464144e-01
-4.57613289e-01 -1.42765832e+00 6.79615438e-01 1.82077095e-01
-9.33655679e-01 -3.92502278e-01 -2.12585896e-01 6.54029548e-02
8.46072555e-01 5.73018491e-01 4.83456969e-01 -9.78094786e-02
-5.11091292e-01 -7.04371810e-01 2.56293207e-01 5.91996789e-01
6.65005028e-01 7.33139157e-01 -4.21664715e-01 -3.83231968e-01
1.11110941e-01 -5.06393015e-01 -2.14933574e-01 1.62758052e-01
2.96768337e-01 -3.30884039e-01 -4.26830500e-01 -2.09312618e-01
1.17926228e+00 7.08901286e-01 7.36141622e-01 2.92125672e-01
7.97325522e-02 7.37536669e-01 1.27323031e+00 1.85518041e-01
-7.65742287e-02 5.81604302e-01 7.79932618e-01 3.57413471e-01
3.17630291e-01 1.89395994e-01 6.51794791e-01 7.08892226e-01
-6.49637938e-01 6.84112683e-02 -8.05359185e-01 1.18587293e-01
-1.82053161e+00 -5.58401942e-01 -4.13249880e-02 2.15000701e+00
5.07731199e-01 6.17334485e-01 2.70224847e-02 5.26607513e-01
4.25987780e-01 -5.06247282e-01 -1.08507633e-01 -6.63520932e-01
3.76567036e-01 -5.60021661e-02 4.43463117e-01 2.94583201e-01
-8.11627150e-01 7.01884389e-01 6.05836821e+00 9.62080657e-02
-1.41102779e+00 -3.71786118e-01 -2.44812861e-01 2.67646283e-01
-4.32316333e-01 3.05997849e-01 -2.91284263e-01 6.24235392e-01
2.38241017e-01 -7.49807358e-02 1.56721920e-01 1.35013556e+00
5.80426194e-02 -5.57563305e-01 -1.01162088e+00 3.15130562e-01
-3.80149096e-01 -8.85176420e-01 -7.02797949e-01 -7.08038583e-02
4.44394469e-01 -4.34909523e-01 -1.12238877e-01 -4.55312468e-02
3.38961065e-01 -8.29184949e-01 1.56945539e+00 3.61709625e-01
5.14161825e-01 -4.86242235e-01 1.10656238e+00 8.72585401e-02
-9.61313248e-01 -1.76973850e-01 -4.44718689e-01 -5.52755594e-01
5.26810706e-01 7.90784180e-01 -7.93062747e-01 5.69742322e-01
8.38858724e-01 2.15987578e-01 3.76813889e-01 6.67538345e-01
-4.56769288e-01 9.73099172e-02 -4.73953664e-01 -9.27994847e-01
-6.96923882e-02 -1.31526530e-01 3.81426543e-01 7.16312647e-01
4.08964306e-01 1.93066895e-01 1.26159310e-01 9.57230568e-01
6.69036329e-01 -2.92106330e-01 -6.00060225e-01 -3.86635989e-01
5.05093694e-01 1.35006154e+00 -9.32135582e-01 3.05015475e-01
-1.65496200e-01 3.69948179e-01 -4.20386374e-01 -1.52551189e-01
-7.79781282e-01 -7.16852188e-01 4.19469416e-01 3.91452372e-01
6.59257639e-03 -8.05780530e-01 -4.47709769e-01 -3.79127145e-01
3.78828704e-01 -5.11449993e-01 -5.01474917e-01 -8.36783171e-01
-1.08760893e+00 3.43803227e-01 1.83523387e-01 -1.66412091e+00
-3.78992289e-01 -7.96976507e-01 -2.45778590e-01 4.54259694e-01
-9.39821959e-01 -1.51096523e+00 -3.33468795e-01 -1.48080900e-01
-4.34446782e-02 1.07625432e-01 8.16194534e-01 -1.98954642e-01
6.09483905e-02 6.57101795e-02 -1.35659128e-01 -2.76269734e-01
7.91336298e-01 -6.71083331e-01 -1.40582159e-01 6.03871942e-01
-1.12954438e+00 1.02189040e+00 1.15137911e+00 -1.08487117e+00
-2.02028394e+00 -5.65508246e-01 5.39808512e-01 -1.65330231e-01
7.54137993e-01 -2.65268534e-01 -4.10443902e-01 4.53736365e-01
-1.00238509e-02 -1.64059460e-01 2.78553903e-01 -3.19222718e-01
3.01962793e-01 -3.17520201e-01 -1.56593192e+00 4.51214910e-01
6.87115014e-01 -1.50404781e-01 -8.18507314e-01 -4.40096892e-02
5.99738717e-01 -6.32362902e-01 -1.29845285e+00 5.65049529e-01
1.04076791e+00 -6.02579713e-01 3.88235569e-01 2.51094937e-01
5.28377950e-01 -6.85166717e-01 -1.77083626e-01 -9.41932082e-01
-5.46149135e-01 -8.33309770e-01 8.46993178e-02 1.23147321e+00
2.79194146e-01 -9.30352867e-01 3.35601151e-01 1.22464812e+00
-6.50354922e-01 -5.67026377e-01 -8.22836876e-01 -1.26317596e+00
-1.40228599e-01 -8.28591734e-02 7.58326948e-01 7.12934494e-01
1.06581795e+00 -2.58060843e-01 -3.14990848e-01 1.85823262e-01
2.46348768e-01 1.27037372e-02 5.37461817e-01 -1.27213848e+00
-1.43559858e-01 9.08829272e-03 -6.08554423e-01 -2.76537806e-01
-8.18835422e-02 -3.02700162e-01 6.80345953e-01 -1.92107499e+00
-1.47245780e-01 -7.93120027e-01 1.14123456e-01 7.34338045e-01
4.72603798e-01 -1.91654742e-01 3.41394752e-01 1.24839857e-01
-3.17751259e-01 1.52646765e-01 1.31036961e+00 3.68621796e-02
-1.80713028e-01 1.33592859e-01 -6.72519982e-01 5.41573584e-01
9.72747505e-01 -1.44164309e-01 -3.52772355e-01 -2.51879334e-01
7.42297828e-01 6.35799840e-02 1.72781080e-01 -1.11470401e+00
1.64884180e-01 -9.15119469e-01 -1.13056488e-01 -3.37439403e-02
1.78091764e-01 -1.65957582e+00 8.06954324e-01 8.67265701e-01
1.28394410e-01 -2.28556484e-01 -2.15398937e-01 -1.23290963e-01
8.64738524e-02 -4.68312532e-01 6.12163424e-01 5.53352339e-03
-5.49604774e-01 -1.68372259e-01 -6.26067162e-01 -8.26886117e-01
1.78388512e+00 -7.09422469e-01 -6.04363799e-01 7.09130391e-02
-3.31526756e-01 2.13213310e-01 1.06423175e+00 6.60692811e-01
5.83104074e-01 -1.02694988e+00 -6.93501979e-02 -1.48537094e-02
1.04344413e-01 -1.46079492e-02 -8.08868781e-02 5.04313529e-01
-7.85854757e-01 3.08150023e-01 -8.31351340e-01 -2.39631489e-01
-1.05268455e+00 4.70719397e-01 -1.12101808e-01 3.73108268e-01
-5.75113952e-01 2.36666352e-01 -6.23462439e-01 -1.67868003e-01
1.96359623e-02 -8.47738504e-01 2.65562654e-01 -7.51800001e-01
1.83968037e-01 4.79318798e-01 -4.29196609e-03 -3.10057938e-01
-6.80683434e-01 5.96960008e-01 2.48369008e-01 8.81664827e-03
1.51654172e+00 1.27867162e-01 -5.02397537e-01 3.06793571e-01
2.35091925e-01 1.58522367e-01 -1.20047987e+00 6.00800455e-01
-1.36313379e-01 -3.26399952e-01 -3.89199793e-01 -8.35314691e-01
-7.34607100e-01 1.90751448e-01 2.16123581e-01 6.30632162e-01
5.53874731e-01 3.80785316e-02 6.76438034e-01 5.05276144e-01
1.34529352e+00 -1.82881939e+00 1.17773049e-01 4.65723932e-01
1.29727292e+00 -6.81680560e-01 3.15574586e-01 -9.75957870e-01
-5.96902788e-01 1.45277870e+00 5.37646592e-01 -2.56496996e-01
7.30934560e-01 9.40935254e-01 -2.25920677e-01 1.09279245e-01
1.56097459e-02 3.49725068e-01 -2.05746397e-01 8.52546394e-01
3.78645480e-01 5.28192759e-01 -7.07281649e-01 8.34470332e-01
-2.13802591e-01 8.11060250e-01 7.00152397e-01 1.96052289e+00
-8.84041429e-01 -1.33870041e+00 -5.23854315e-01 2.69466341e-01
-2.34106392e-01 8.71303678e-01 -4.90139455e-01 7.01409101e-01
3.21637154e-01 1.14032757e+00 -5.28171360e-01 -8.02334845e-01
4.44698691e-01 -3.62314671e-01 5.42201102e-01 -3.93380493e-01
-3.88102740e-01 -4.40312564e-01 6.46473527e-01 -3.38066161e-01
-3.93953323e-01 -3.27404171e-01 -1.51004291e+00 -4.04350817e-01
-2.81167299e-01 1.48238987e-01 1.16480899e+00 1.02487922e+00
9.42949131e-02 3.06114495e-01 4.37090576e-01 -8.30812871e-01
-8.93837988e-01 -7.06946015e-01 -8.29171240e-01 -1.58383660e-02
-1.83875412e-01 -1.15312159e+00 -1.17472922e-02 -1.21216059e-01] | [4.8872904777526855, 1.1082592010498047] |
151890f4-e316-49c3-ae18-0f4127cae0d4 | open-set-multi-source-multi-target-domain | 2302.00995 | null | https://arxiv.org/abs/2302.00995v2 | https://arxiv.org/pdf/2302.00995v2.pdf | Open-Set Multi-Source Multi-Target Domain Adaptation | Single-Source Single-Target Domain Adaptation (1S1T) aims to bridge the gap between a labelled source domain and an unlabelled target domain. Despite 1S1T being a well-researched topic, they are typically not deployed to the real world. Methods like Multi-Source Domain Adaptation and Multi-Target Domain Adaptation have evolved to model real-world problems but still do not generalise well. The fact that most of these methods assume a common label-set between source and target is very restrictive. Recent Open-Set Domain Adaptation methods handle unknown target labels but fail to generalise in multiple domains. To overcome these difficulties, first, we propose a novel generic domain adaptation (DA) setting named Open-Set Multi-Source Multi-Target Domain Adaptation (OS-nSmT), with n and m being number of source and target domains respectively. Next, we propose a graph attention based framework named DEGAA which can capture information from multiple source and target domains without knowing the exact label-set of the target. We argue that our method, though offered for multiple sources and multiple targets, can also be agnostic to various other DA settings. To check the robustness and versatility of DEGAA, we put forward ample experiments and ablation studies. | ['Ritik Agrawal', 'Muhammed Abdullah Shaikh', 'Aadhithya Iyer', 'Arihant Gaur', 'Rohit Lal'] | 2023-02-02 | null | null | null | null | ['multi-target-domain-adaptation'] | ['computer-vision'] | [ 5.13348579e-01 -1.98293179e-02 -4.41584557e-01 -4.89755094e-01
-6.36730671e-01 -8.84925306e-01 6.16835356e-01 -8.98731593e-03
-7.54698068e-02 8.24312210e-01 -2.00908005e-01 -1.41408052e-02
-2.01953590e-01 -6.40920222e-01 -5.42854488e-01 -5.77794373e-01
2.88295180e-01 9.65884328e-01 6.34636819e-01 -5.17759800e-01
-1.37706280e-01 2.07912624e-01 -1.18238270e+00 2.18989208e-01
9.19802964e-01 7.34629929e-01 2.08904475e-01 3.23689133e-01
-2.17222527e-01 4.07472849e-01 -8.67338181e-01 -5.34501731e-01
3.43572795e-01 -6.45459890e-01 -8.98404002e-01 1.18738011e-01
3.33793163e-01 1.39575601e-01 -1.47123039e-02 1.11801624e+00
6.16067588e-01 2.29588702e-01 9.70429003e-01 -1.81857634e+00
-1.10103071e+00 4.41121429e-01 -5.91528833e-01 3.36852968e-01
2.27287322e-01 1.62773002e-02 9.10065532e-01 -5.81361055e-01
7.29095578e-01 1.42444372e+00 6.64461553e-01 8.63603711e-01
-1.08868933e+00 -8.34021866e-01 6.63355529e-01 3.01852494e-01
-1.09414017e+00 -3.46414328e-01 8.34808469e-01 -2.04726174e-01
6.50801241e-01 3.34767848e-02 1.46705583e-01 1.72763789e+00
-1.82958588e-01 6.94219470e-01 1.23519957e+00 -5.18343806e-01
2.12019727e-01 3.65176469e-01 1.27957448e-01 6.15837276e-02
1.84298918e-01 4.30271812e-02 -2.35762462e-01 -1.18828617e-01
6.83440268e-01 -1.09076686e-01 -2.28026077e-01 -7.21158743e-01
-1.36349559e+00 1.03254724e+00 3.98178160e-01 4.70431417e-01
-4.96887006e-02 -3.85948360e-01 5.86480975e-01 6.16193175e-01
5.96362412e-01 3.69692773e-01 -7.35532403e-01 1.84740916e-01
-4.50641185e-01 -4.34500240e-02 6.92259312e-01 1.31768858e+00
7.73299813e-01 -1.42022385e-03 -1.08312547e-01 1.15995622e+00
1.43176883e-01 5.22305012e-01 7.79303014e-01 -5.06981075e-01
5.06975770e-01 9.03939664e-01 8.81188512e-02 -7.69142866e-01
-4.38521057e-01 -5.26578724e-01 -7.60911107e-01 -1.73198637e-02
7.64992237e-01 -1.73261374e-01 -9.73130345e-01 2.27057362e+00
5.81061423e-01 2.98972577e-01 3.75159681e-01 9.14121389e-01
7.57159352e-01 3.37826490e-01 2.74071485e-01 -6.11311346e-02
1.16363001e+00 -1.07970524e+00 -3.51560771e-01 -7.44015336e-01
7.05543637e-01 -4.78496104e-01 1.23048985e+00 -9.41741746e-03
-4.36405808e-01 -4.73797768e-01 -8.18884969e-01 1.32708430e-01
-8.34396958e-01 -1.98775679e-01 3.76327813e-01 6.81835592e-01
-8.74971032e-01 1.08957492e-01 -1.71529084e-01 -8.73139977e-01
4.18255001e-01 3.72312039e-01 -6.04989886e-01 -4.44554955e-01
-1.53244519e+00 1.17246854e+00 5.98538578e-01 -5.06624997e-01
-1.03921402e+00 -5.74351847e-01 -7.71915495e-01 -3.30512635e-02
6.77735448e-01 -8.21783006e-01 1.27483189e+00 -1.51803493e+00
-1.42979896e+00 1.07015467e+00 -6.68500364e-02 -2.12524399e-01
1.99455023e-01 1.11899026e-01 -8.86191666e-01 -1.10671394e-01
5.29192448e-01 5.47476649e-01 9.28357124e-01 -1.31541836e+00
-7.79846609e-01 -5.40188730e-01 2.43210360e-01 5.45174658e-01
-4.24448520e-01 2.10305280e-03 -1.45592332e-01 -6.32133186e-01
-2.02857256e-01 -8.08309734e-01 -1.01825111e-01 -2.22367838e-01
-3.84516746e-01 -4.20160472e-01 1.12166107e+00 -2.12479874e-01
8.85296464e-01 -2.09857464e+00 2.30581924e-01 -7.60176554e-02
1.12539239e-01 5.47086239e-01 -6.14636540e-01 4.33185965e-01
-3.87031257e-01 -1.24147199e-01 -4.80069816e-01 -1.27363801e-01
1.30634665e-01 4.57242399e-01 -1.20891690e-01 3.04232210e-01
3.25826436e-01 9.54843819e-01 -1.10778427e+00 -3.85305375e-01
2.22801231e-03 2.78291285e-01 -2.36193135e-01 1.52948901e-01
-5.02024293e-01 7.79151618e-01 -4.62184131e-01 7.77958035e-01
6.71085000e-01 -4.98514831e-01 1.79158553e-01 3.52535099e-02
5.42276084e-01 -6.31102696e-02 -1.21477246e+00 1.66926062e+00
-3.62827629e-01 2.87696481e-01 1.43217012e-01 -1.38832521e+00
1.02493024e+00 2.75280207e-01 5.19841790e-01 -7.27624059e-01
1.37334047e-02 3.48359048e-01 1.78995028e-01 -2.34987304e-01
9.59780514e-02 -6.00785673e-01 -3.89907420e-01 3.84003997e-01
3.50099891e-01 2.13538319e-01 1.27319187e-01 1.95499152e-01
1.00356424e+00 1.19320668e-01 6.54673576e-01 -1.46492511e-01
5.72640717e-01 1.85343161e-01 6.67795479e-01 6.41385734e-01
-6.24634027e-01 5.21893382e-01 4.98376936e-01 -9.44231004e-02
-1.04778206e+00 -9.13136482e-01 1.08398840e-01 1.71388626e+00
5.30112684e-01 8.57236609e-02 -6.36627316e-01 -1.37863219e+00
7.36901164e-02 6.66051090e-01 -6.31407261e-01 -5.17929494e-01
-2.75765806e-01 -6.51079059e-01 6.15127444e-01 3.38446915e-01
4.34901297e-01 -1.04440534e+00 -2.27354124e-01 2.54403293e-01
-3.51711005e-01 -1.31969666e+00 -6.41177356e-01 3.01013142e-01
-5.73308408e-01 -1.28049326e+00 -1.09927320e+00 -9.54763412e-01
4.18347269e-01 2.42549375e-01 1.42344642e+00 -4.32222098e-01
3.99945468e-01 6.55053496e-01 -6.59102082e-01 -5.35566807e-01
-6.94694102e-01 4.21673417e-01 1.65504232e-01 3.52863401e-01
7.45357275e-01 -7.76450694e-01 -2.31090873e-01 8.73964965e-01
-9.04276311e-01 -3.42689067e-01 7.01817811e-01 9.08997059e-01
4.23627049e-01 -3.27648297e-02 1.19120014e+00 -1.29568446e+00
7.53341377e-01 -1.04797244e+00 -2.60581613e-01 4.46901351e-01
-6.51392758e-01 -1.38019994e-01 9.37138140e-01 -9.94651139e-01
-9.44041908e-01 -1.05906695e-01 1.65933877e-01 -5.01694202e-01
-6.50648355e-01 3.70645493e-01 -7.30326593e-01 -5.90001680e-02
9.53112841e-01 2.22998768e-01 -9.15926024e-02 -5.51443517e-01
4.55632955e-01 7.34973669e-01 5.15371978e-01 -5.29860318e-01
8.82820725e-01 2.39435971e-01 -3.01775783e-01 -4.48734611e-01
-1.03503394e+00 -7.75076449e-01 -8.61068070e-01 1.93441566e-02
4.72642243e-01 -8.45008910e-01 -1.28984317e-01 5.57893157e-01
-9.94779348e-01 -4.28762347e-01 -3.81210506e-01 3.57739367e-02
-5.26041806e-01 4.30906951e-01 8.29916224e-02 -4.58014160e-01
-4.18782420e-02 -8.67428660e-01 1.05776525e+00 1.56339005e-01
-8.32695067e-02 -1.51883113e+00 1.47692457e-01 1.66012838e-01
4.82294589e-01 3.02145898e-01 9.26286817e-01 -1.51494372e+00
-1.75682679e-02 -6.36241212e-02 -3.99052352e-01 3.35882604e-01
4.61081386e-01 -7.64322579e-01 -7.53803551e-01 -5.06987453e-01
-4.82300296e-02 -3.86480182e-01 4.47498083e-01 1.50084913e-01
5.00095725e-01 -3.02687317e-01 -6.20267391e-01 3.36281687e-01
1.47432005e+00 2.32550278e-01 3.88667017e-01 5.44077575e-01
8.43738317e-01 5.46124220e-01 7.69150138e-01 1.55207306e-01
5.42567134e-01 8.72298121e-01 5.54568470e-01 -1.60931632e-01
-3.74645174e-01 -1.79880127e-01 6.04161561e-01 4.74275470e-01
4.50962663e-01 -6.21192634e-01 -1.01800084e+00 1.06679916e+00
-1.72298229e+00 -7.17247069e-01 4.44833189e-02 2.12478256e+00
7.93841541e-01 1.69143394e-01 6.46179974e-01 -6.62590191e-02
1.14367616e+00 -1.52010232e-01 -1.08581161e+00 -2.88607627e-01
-3.97584468e-01 4.10904735e-03 4.22153860e-01 2.26028129e-01
-1.14639997e+00 1.11073816e+00 5.97047091e+00 1.12358344e+00
-1.01468015e+00 4.39519227e-01 2.81148553e-01 2.49903977e-01
-1.71431601e-01 -7.04307621e-03 -9.07384038e-01 7.45438755e-01
1.09123719e+00 -3.18401664e-01 2.18379319e-01 1.00181985e+00
-4.40673232e-01 2.67904222e-01 -1.28083062e+00 7.68031061e-01
9.63354483e-02 -5.97669005e-01 1.44922301e-01 -1.16268471e-01
6.05335116e-01 9.52276066e-02 4.47444171e-02 7.45187283e-01
6.91328526e-01 -8.60432565e-01 3.65251064e-01 -8.67965966e-02
1.04861093e+00 -5.63661516e-01 6.61622703e-01 6.26558006e-01
-1.10903859e+00 -3.10283720e-01 -3.82713348e-01 7.64971599e-02
1.31360684e-02 3.62107426e-01 -9.67065871e-01 8.05750072e-01
5.68018734e-01 9.71842051e-01 -6.40004158e-01 9.73029315e-01
-1.08303353e-01 4.82753128e-01 -4.43085492e-01 2.01357827e-01
2.68514723e-01 4.20170948e-02 7.62275994e-01 1.16506612e+00
3.24891388e-01 -8.74559581e-02 3.67646039e-01 6.42171323e-01
-2.11129654e-02 1.14635140e-01 -1.00922894e+00 -9.09603685e-02
5.75430214e-01 8.88477385e-01 -5.97546101e-01 -3.51930559e-01
-5.27866900e-01 1.15170372e+00 3.36632431e-01 5.90193987e-01
-9.04885292e-01 -2.81133503e-01 8.16564441e-01 6.34044111e-02
2.79061496e-01 8.60925391e-02 -1.84519917e-01 -1.30306840e+00
-6.07664101e-02 -1.12125683e+00 8.68143737e-01 -5.80918670e-01
-1.93822181e+00 5.91984034e-01 2.41253912e-01 -1.49392962e+00
-1.70433670e-01 -5.66345215e-01 -2.80457258e-01 8.35657299e-01
-1.87218690e+00 -1.45327950e+00 -1.08100183e-01 1.08793426e+00
7.17056751e-01 -4.66315389e-01 9.83962655e-01 4.42935348e-01
-4.95940775e-01 9.24196839e-01 3.61703902e-01 9.02630538e-02
1.11832893e+00 -1.22007573e+00 4.56344932e-01 7.92925477e-01
9.71559808e-02 3.69311333e-01 5.57593763e-01 -6.29706562e-01
-7.57187843e-01 -1.31641591e+00 6.24910474e-01 -8.04902077e-01
6.17219448e-01 -4.64037508e-01 -1.10266304e+00 1.02558982e+00
1.47754848e-01 1.27008364e-01 5.59247077e-01 1.12096153e-01
-6.67361856e-01 2.73562912e-02 -1.36119521e+00 3.90398413e-01
1.24025977e+00 -1.90601975e-01 -6.03543222e-01 4.79086995e-01
8.90933156e-01 -3.05704802e-01 -9.69784617e-01 5.07490098e-01
-1.12669110e-01 -9.02603149e-01 1.09473181e+00 -7.84142911e-01
1.91089511e-01 -1.32016510e-01 -1.94385827e-01 -1.74575782e+00
-4.47016567e-01 -3.88154536e-01 -1.10508412e-01 1.40987957e+00
3.92945856e-01 -1.05334520e+00 6.18170440e-01 2.25994706e-01
-2.08915979e-01 -2.34233737e-01 -1.18282890e+00 -1.33563781e+00
4.90368456e-01 -1.42887935e-01 8.23445678e-01 1.43556821e+00
-1.38678849e-01 8.66765201e-01 -4.69309777e-01 2.73104489e-01
6.27931058e-01 1.68085113e-01 7.02517569e-01 -1.67649639e+00
-2.05118075e-01 -3.91906857e-01 -3.25868815e-01 -1.05486608e+00
3.44904125e-01 -9.84354913e-01 -1.64992139e-01 -1.42861438e+00
5.22281528e-02 -7.62844980e-01 -4.79084104e-01 8.01301301e-01
-2.34629408e-01 2.67825574e-01 2.08612476e-02 4.02614266e-01
-7.73165226e-01 5.40668368e-01 1.28992343e+00 -1.94858834e-01
-9.18636471e-02 1.78744912e-01 -1.11393440e+00 4.60964710e-01
7.44122505e-01 -6.74153149e-01 -7.60713339e-01 -3.57315540e-01
-2.34784573e-01 -2.43956730e-01 2.43839562e-01 -9.01529968e-01
7.14830086e-02 -4.07625020e-01 4.41088788e-02 3.76316980e-02
2.13714167e-01 -1.08190656e+00 -1.07503580e-02 -1.38802752e-01
-2.34341592e-01 -1.67571411e-01 2.38998666e-01 9.66762662e-01
-1.92609236e-01 -2.52023906e-01 1.02017045e+00 -3.13575894e-01
-1.33069253e+00 3.96915615e-01 1.03838055e-03 4.71570373e-01
1.37319756e+00 -4.90183145e-01 -4.70250398e-01 -3.23578537e-01
-8.72673213e-01 3.69175047e-01 5.79180837e-01 7.29086578e-01
3.23090881e-01 -1.47624099e+00 -8.54355812e-01 1.47280380e-01
6.32922053e-01 7.73604438e-02 1.63920090e-01 6.57547593e-01
3.07338476e-01 3.45912725e-01 -3.77410889e-01 -6.93716943e-01
-1.02599192e+00 9.45798993e-01 3.91680092e-01 -4.71921176e-01
-5.32411039e-01 8.99237812e-01 5.25543690e-01 -1.00318217e+00
5.73711023e-02 1.73387349e-01 -4.53509897e-01 5.34040965e-02
1.15447789e-01 1.46254376e-01 -9.16713327e-02 -9.63455260e-01
-5.20160317e-01 5.83213210e-01 -9.49610025e-02 1.87111065e-01
1.05469942e+00 -4.70402151e-01 2.15603516e-01 4.57659751e-01
1.07108867e+00 -4.78175759e-01 -1.17886496e+00 -6.14091575e-01
3.12542349e-01 -2.86887437e-01 -5.23008287e-01 -1.26792550e+00
-9.38139915e-01 7.83585489e-01 6.05249107e-01 3.83360744e-01
1.34398019e+00 1.59399673e-01 7.86216080e-01 1.10668302e-01
3.88583392e-01 -9.43771303e-01 2.38482937e-01 5.89053750e-01
7.24189997e-01 -1.52235734e+00 -5.58289170e-01 -5.15887141e-01
-9.44232345e-01 7.07473576e-01 1.05707574e+00 8.63809809e-02
2.56565511e-01 -1.40232995e-01 2.42374003e-01 -1.08018711e-01
-6.12411499e-01 -3.86912614e-01 2.00585186e-01 1.43245113e+00
3.24382745e-02 -9.07644108e-02 2.01130494e-01 7.07086504e-01
2.29361773e-01 -4.62589189e-02 3.94713283e-01 5.79563677e-01
-3.03744227e-01 -1.47578394e+00 -5.18993974e-01 1.81334332e-01
-1.71041429e-01 7.81105980e-02 -6.84942544e-01 9.82349813e-01
2.58505344e-01 1.04666960e+00 -3.76656115e-01 -3.02635729e-01
6.98047042e-01 3.34814638e-01 2.82787412e-01 -9.18893039e-01
-3.94391596e-01 -1.91979498e-01 -4.25009876e-02 -2.49815300e-01
-6.26078427e-01 -3.57127309e-01 -7.96965241e-01 6.87626330e-03
-3.82364422e-01 2.01915830e-01 2.31557682e-01 9.43223417e-01
5.46141744e-01 3.60586762e-01 3.97026002e-01 -4.72086042e-01
-6.21996284e-01 -1.08234751e+00 -6.92282975e-01 7.93463528e-01
3.11414391e-01 -1.12237406e+00 -3.26277465e-01 -3.68817449e-02] | [10.337457656860352, 3.0387630462646484] |
e2ee08d5-a441-4a53-9e0b-160effee01dd | pointpwc-net-a-coarse-to-fine-network-for | 1911.12408 | null | https://arxiv.org/abs/1911.12408v2 | https://arxiv.org/pdf/1911.12408v2.pdf | PointPWC-Net: A Coarse-to-Fine Network for Supervised and Self-Supervised Scene Flow Estimation on 3D Point Clouds | We propose a novel end-to-end deep scene flow model, called PointPWC-Net, on 3D point clouds in a coarse-to-fine fashion. Flow computed at the coarse level is upsampled and warped to a finer level, enabling the algorithm to accommodate for large motion without a prohibitive search space. We introduce novel cost volume, upsampling, and warping layers to efficiently handle 3D point cloud data. Unlike traditional cost volumes that require exhaustively computing all the cost values on a high-dimensional grid, our point-based formulation discretizes the cost volume onto input 3D points, and a PointConv operation efficiently computes convolutions on the cost volume. Experiment results on FlyingThings3D outperform the state-of-the-art by a large margin. We further explore novel self-supervised losses to train our model and achieve comparable results to state-of-the-art trained with supervised loss. Without any fine-tuning, our method also shows great generalization ability on KITTI Scene Flow 2015 dataset, outperforming all previous methods. | ['Li Fuxin', 'Zhuwen Li', 'Zhiyuan Wang', 'Wei Liu', 'Wenxuan Wu'] | 2019-11-27 | null | null | null | null | ['scene-flow-estimation'] | ['computer-vision'] | [-2.22196728e-01 -3.53777111e-01 1.98596939e-02 -4.16276604e-01
-5.67003369e-01 -4.64880049e-01 4.73041922e-01 -1.45444423e-01
-4.66783255e-01 6.87707722e-01 2.00512931e-01 -2.21352682e-01
-7.29294717e-02 -1.08785272e+00 -9.99179006e-01 -2.87130654e-01
-2.94222534e-01 6.04052305e-01 4.64344412e-01 -9.92629528e-02
1.85306653e-01 7.66581833e-01 -1.48384118e+00 1.24075934e-01
1.10696769e+00 1.16327202e+00 2.65973760e-03 9.03768957e-01
-1.83027133e-01 7.73655355e-01 -3.50946844e-01 -2.17901319e-01
8.12392473e-01 -5.76938502e-02 -9.19500411e-01 2.88677253e-02
1.09396219e+00 -8.21882904e-01 -5.99197090e-01 7.16136396e-01
3.76519501e-01 4.63342696e-01 5.17831028e-01 -1.19024181e+00
-5.05235374e-01 -1.70536116e-01 -6.93813920e-01 2.17613131e-01
1.67592332e-01 5.29814482e-01 9.03946877e-01 -1.12641251e+00
7.50214815e-01 1.51173282e+00 8.89297128e-01 2.63692588e-01
-1.23501050e+00 -7.29021907e-01 2.14962095e-01 -5.81237786e-02
-1.14370263e+00 -7.13144243e-02 7.40471780e-01 -5.57708621e-01
1.29572821e+00 -1.35950312e-01 9.31718707e-01 5.60141027e-01
9.86586958e-02 6.58768535e-01 5.26427448e-01 1.04036443e-01
2.04122469e-01 -6.36603415e-01 -2.51869142e-01 1.00299060e+00
4.44256999e-02 4.50779140e-01 -5.01525939e-01 -7.34296069e-02
1.25096416e+00 2.14009523e-01 -3.33670586e-01 -6.06332123e-01
-1.39668250e+00 9.62742031e-01 1.21089363e+00 -3.20361853e-01
-3.43930095e-01 6.23323619e-01 4.98696178e-01 1.86643526e-01
7.22659707e-01 3.65168810e-01 -6.44374907e-01 -3.64020675e-01
-1.22872531e+00 6.43507659e-01 6.07888877e-01 1.09171879e+00
1.05984938e+00 9.60712880e-02 -1.14518277e-01 3.65580112e-01
1.49242133e-01 5.45931697e-01 -5.29548042e-02 -1.48426557e+00
8.13837409e-01 5.57604671e-01 1.12205662e-01 -8.26435506e-01
-4.25387204e-01 -4.62290436e-01 -9.77715731e-01 6.74997330e-01
3.91275287e-01 -1.83941096e-01 -1.01561260e+00 1.47574866e+00
6.32542074e-01 7.00901866e-01 -2.26977214e-01 1.21305609e+00
6.51961029e-01 1.06238604e+00 -1.48106605e-01 2.52301455e-01
8.59418571e-01 -1.33132088e+00 -1.79959342e-01 -1.25210851e-01
6.78536415e-01 -4.58315611e-01 1.30206466e+00 7.77046382e-02
-1.34679353e+00 -7.30124533e-01 -9.87556756e-01 -6.69998884e-01
1.70058552e-02 -1.54065490e-01 1.05145371e+00 2.63089575e-02
-1.08846295e+00 1.14570272e+00 -1.10184979e+00 -5.84927127e-02
9.68279660e-01 2.22259387e-01 -3.76315415e-01 -1.24990687e-01
-8.09364796e-01 6.77011311e-01 -8.24464336e-02 -1.85264274e-02
-9.95028794e-01 -1.69546068e+00 -1.12260020e+00 8.16340223e-02
1.28051406e-02 -1.45411205e+00 1.18118262e+00 -2.11055174e-01
-1.61823750e+00 7.32678771e-01 -3.04523498e-01 -6.19939029e-01
8.17026973e-01 -4.90262777e-01 2.34622359e-01 2.53738016e-01
3.21539730e-01 1.13834250e+00 7.67444253e-01 -9.27623987e-01
-7.49793053e-01 -1.65807128e-01 1.92689493e-01 3.08573544e-01
1.93217471e-01 -4.63058054e-01 -3.54317844e-01 -5.94276071e-01
-1.13389537e-01 -7.61903524e-01 -3.38364691e-01 8.61148477e-01
-6.37198389e-02 -1.85852796e-01 8.88570309e-01 -2.51774967e-01
7.10431099e-01 -2.01390028e+00 2.24041164e-01 -2.23858908e-01
4.54374373e-01 3.46119314e-01 -7.74203092e-02 1.17707089e-01
1.58856526e-01 -6.21090531e-02 -7.05822110e-01 -6.28035128e-01
-6.21108189e-02 2.57321864e-01 -5.03778934e-01 5.59575200e-01
6.32794261e-01 9.77044582e-01 -1.15491545e+00 -3.96197289e-01
8.03071856e-01 6.75177515e-01 -1.31012869e+00 1.57335192e-01
-8.46531838e-02 5.09629130e-01 -4.99429345e-01 5.04992723e-01
9.35606718e-01 -5.36588073e-01 -9.08905208e-01 -1.17121041e-01
-2.27238953e-01 5.35172164e-01 -8.72851253e-01 2.47884488e+00
-6.88630104e-01 7.19267309e-01 -4.17669713e-02 -6.60502374e-01
8.25729489e-01 -1.92728668e-01 6.44913733e-01 -2.64551252e-01
-1.26870796e-01 8.05570856e-02 -2.00228989e-01 -1.75118536e-01
5.53313375e-01 -1.61285192e-01 -1.87890008e-02 -6.42834008e-02
2.61542290e-01 -8.30973446e-01 9.73984152e-02 2.35929683e-01
1.16031086e+00 5.47921479e-01 -1.41958341e-01 -2.74605483e-01
5.91866016e-01 4.26642001e-01 6.45576060e-01 3.73136312e-01
-3.48209679e-01 8.35026860e-01 4.13709074e-01 -8.26882184e-01
-1.05008662e+00 -1.19528055e+00 -1.46193877e-01 5.70472598e-01
3.83817255e-01 -3.14496368e-01 -3.98665428e-01 -6.53997064e-01
5.25932848e-01 3.91771942e-01 -5.00386834e-01 6.55528381e-02
-7.65337288e-01 -3.13552707e-01 4.17753011e-01 7.43033469e-01
1.00431991e+00 -8.71953189e-01 -9.00213540e-01 3.35892439e-01
-5.10030724e-02 -1.17943490e+00 -8.32002223e-01 -5.18195257e-02
-1.19343126e+00 -9.80364442e-01 -7.50200093e-01 -6.80018604e-01
5.75981259e-01 2.75585651e-01 1.36327124e+00 1.94843356e-02
-3.56843770e-01 1.10907063e-01 -1.42659515e-01 -2.37555638e-01
1.69351488e-01 1.08818278e-01 -7.50443563e-02 -4.14418697e-01
1.32535234e-01 -7.79970884e-01 -1.07473576e+00 8.97218361e-02
-5.60587883e-01 1.68136694e-02 3.12228918e-01 7.26975381e-01
6.69043422e-01 -1.09639384e-01 1.42442524e-01 -4.67707455e-01
1.39258325e-01 -1.11924678e-01 -8.90628278e-01 -3.40663493e-01
-5.93645424e-02 1.10322468e-01 7.20644474e-01 -1.75547495e-01
-8.96874249e-01 2.12674692e-01 -3.49053174e-01 -1.17699504e+00
-1.21063748e-02 -8.27607438e-02 2.46165097e-01 -5.81957459e-01
6.01665437e-01 -1.81384265e-01 -9.41363126e-02 -2.73089051e-01
5.53863645e-01 -1.69482693e-01 8.58107924e-01 -5.72783411e-01
1.20227945e+00 9.69924033e-01 4.84378636e-01 -6.50708556e-01
-8.86740863e-01 -5.28278768e-01 -8.12130332e-01 -1.16330743e-01
9.64482367e-01 -1.17032111e+00 -9.13618803e-01 5.49619555e-01
-1.39233339e+00 -6.67945504e-01 -8.10990751e-01 6.01433992e-01
-8.57441962e-01 1.48714930e-01 -8.12500477e-01 -2.61488050e-01
-4.84160006e-01 -1.15861595e+00 1.47208023e+00 -4.48371023e-02
-1.40056789e-01 -9.31862354e-01 1.97888389e-01 8.18082467e-02
4.12472844e-01 6.34349406e-01 3.88599902e-01 2.50723988e-01
-9.58844543e-01 3.36845554e-02 -4.96507496e-01 3.35227281e-01
-1.80555463e-01 1.84899583e-01 -8.21850359e-01 -2.55889475e-01
-1.25985071e-01 -5.84300220e-01 1.18307769e+00 5.99547744e-01
1.32409096e+00 -3.83703485e-02 -1.73107803e-01 1.62884498e+00
1.44360566e+00 -2.18056306e-01 5.12542307e-01 1.59539163e-01
1.05920076e+00 3.40929598e-01 5.74358404e-01 4.60439175e-01
6.27111077e-01 3.43995094e-01 8.25536430e-01 -3.17026913e-01
-2.09475666e-01 -5.73289692e-01 -2.04416905e-02 4.67957497e-01
-2.02714354e-01 -1.34236127e-01 -7.91659892e-01 6.03063226e-01
-1.73888969e+00 -8.50494802e-01 -2.09150210e-01 1.94512975e+00
5.78672409e-01 2.45880127e-01 -4.85374518e-02 -2.45901570e-02
1.22329049e-01 5.88606417e-01 -7.86759138e-01 -1.86411470e-01
2.33777300e-01 4.92474586e-01 7.76861310e-01 9.11213934e-01
-1.19807422e+00 1.18999445e+00 6.07843065e+00 5.32272756e-01
-1.13312721e+00 -6.36966601e-02 3.82226139e-01 -5.59804916e-01
-2.00628474e-01 1.40921965e-01 -7.27637887e-01 4.00761157e-01
3.69031489e-01 -7.96705708e-02 4.17130917e-01 9.35199678e-01
2.14201331e-01 1.16317637e-01 -1.20995414e+00 1.15598440e+00
-3.07066351e-01 -1.88408267e+00 1.19740188e-01 6.00175299e-02
8.09042454e-01 5.37374914e-01 -1.64765611e-01 3.70689392e-01
4.49547857e-01 -9.00142848e-01 7.23362684e-01 2.75056541e-01
8.46823275e-01 -7.37613976e-01 3.09563786e-01 2.25409672e-01
-1.33331323e+00 2.39068553e-01 -7.19334722e-01 -5.37385285e-01
6.64152622e-01 6.50851905e-01 -3.21824223e-01 5.07161677e-01
9.03296232e-01 1.18094563e+00 -1.09028570e-01 1.16206694e+00
-7.74480104e-02 2.60122955e-01 -5.45869768e-01 2.23193362e-01
7.28940904e-01 -2.94980079e-01 7.27035105e-01 1.05928874e+00
4.22751009e-01 1.86313078e-01 2.77665883e-01 1.24312007e+00
-3.13882351e-01 -3.90470684e-01 -6.06049180e-01 5.24859548e-01
3.65904450e-01 1.04969549e+00 -4.32557195e-01 -4.47896749e-01
-4.46574718e-01 1.06181502e+00 5.27267039e-01 3.50813121e-01
-7.56374061e-01 -6.28486097e-01 1.30287409e+00 2.43437037e-01
5.33964634e-01 -4.55622017e-01 -3.66008222e-01 -1.32040572e+00
1.36878982e-01 -4.04688865e-02 1.13548070e-01 -7.43600786e-01
-1.40241086e+00 5.70584536e-01 -8.87100399e-03 -1.60903764e+00
-1.64428532e-01 -5.85012555e-01 -7.42782891e-01 1.03049934e+00
-1.97687042e+00 -8.96477222e-01 -6.55400097e-01 7.15256333e-01
5.82517326e-01 2.71000594e-01 3.23802054e-01 2.57754862e-01
-5.73845357e-02 2.44125053e-01 -3.52613956e-01 1.29802674e-01
5.75087011e-01 -1.20483148e+00 1.01653183e+00 8.82505417e-01
-1.61349103e-01 2.09450319e-01 2.89561391e-01 -5.12223065e-01
-1.23018408e+00 -1.47597647e+00 7.49214530e-01 -6.56625152e-01
5.25650203e-01 -3.69889468e-01 -9.75727320e-01 6.50386095e-01
-9.16707516e-03 8.64852846e-01 4.15245723e-03 -2.77223080e-01
-2.73373336e-01 -1.66715100e-01 -1.31549323e+00 3.05837899e-01
1.65146923e+00 -2.60737002e-01 -5.39815187e-01 3.86366993e-01
1.27716339e+00 -9.83534396e-01 -8.47487926e-01 6.52289450e-01
1.69413999e-01 -1.10370266e+00 1.31738174e+00 -5.38506567e-01
8.77265751e-01 -4.90986228e-01 8.25305507e-02 -1.43763590e+00
-6.08943522e-01 -8.74550343e-01 -2.62940526e-01 5.40538013e-01
5.72644845e-02 -5.95304608e-01 1.05007041e+00 4.72279191e-01
-6.48965240e-01 -8.40733826e-01 -1.04239357e+00 -7.62974977e-01
4.28345889e-01 -5.51656067e-01 8.04249942e-01 7.95617580e-01
-3.64316106e-01 1.38635352e-01 -2.09592566e-01 2.44230121e-01
8.68360281e-01 2.36990839e-01 1.13035607e+00 -1.01234698e+00
-9.24066082e-02 -5.42613268e-01 -7.30851471e-01 -1.78630137e+00
1.19397707e-01 -8.91688049e-01 -4.50946465e-02 -1.66690338e+00
-5.15594125e-01 -5.00485897e-01 2.72145510e-01 3.19587350e-01
-2.02534944e-01 2.17144266e-01 4.02055204e-01 1.73703849e-01
-3.49015862e-01 9.88672614e-01 1.96309686e+00 -1.05786920e-01
-4.44146276e-01 -1.92641392e-01 -2.21444651e-01 8.73040020e-01
3.13372850e-01 -2.75810033e-01 -6.09917462e-01 -9.80626166e-01
-1.42474920e-01 2.37993509e-01 7.08800435e-01 -1.31403875e+00
3.17015976e-01 -1.93402156e-01 5.13982892e-01 -1.11690652e+00
5.18544316e-01 -8.02313983e-01 -3.12302023e-01 3.57830048e-01
-1.77761152e-01 1.18694909e-01 3.80889058e-01 5.58430851e-01
-2.40220472e-01 4.02585626e-01 8.99667323e-01 -1.80508196e-01
-6.26012981e-01 1.14550376e+00 2.42706120e-01 3.97289455e-01
9.56618428e-01 -1.36376664e-01 -1.76574439e-01 -1.75085723e-01
-3.22669297e-01 5.94292581e-01 5.82862437e-01 4.84742761e-01
7.22225308e-01 -1.43468714e+00 -7.30336547e-01 4.70230043e-01
-6.41775802e-02 1.21825707e+00 3.39205772e-01 4.70017314e-01
-1.23171890e+00 2.74076432e-01 -2.71339566e-01 -9.93729532e-01
-4.33295906e-01 4.62739199e-01 4.21709687e-01 -2.51854897e-01
-1.18145740e+00 1.18226433e+00 4.68213320e-01 -6.89732730e-01
2.13835955e-01 -9.33784544e-01 4.52996641e-01 -3.90159965e-01
4.41623181e-01 5.06809235e-01 4.80374042e-03 -5.17246783e-01
-5.17471313e-01 1.08228195e+00 2.32105955e-01 9.84301418e-03
1.35372150e+00 2.23345250e-01 1.47857264e-01 1.63652465e-01
1.65475786e+00 -3.61365020e-01 -2.06973839e+00 -1.03511766e-01
-7.25290358e-01 -1.07230973e+00 2.74006724e-01 -3.38231266e-01
-1.34407866e+00 1.18138051e+00 7.86255524e-02 -3.52872461e-01
1.00598347e+00 -2.73793608e-01 1.17239714e+00 3.64256442e-01
4.39640850e-01 -5.26156127e-01 2.86974455e-03 7.94883549e-01
9.52489197e-01 -1.34229863e+00 4.23661694e-02 -4.91345078e-01
-4.87262279e-01 9.68260884e-01 7.43472874e-01 -1.01866996e+00
8.83465469e-01 2.57624745e-01 -2.15417609e-01 -2.44771555e-01
-7.57412612e-01 -1.34211227e-01 2.80823946e-01 6.56571865e-01
-3.83934528e-02 -1.45914331e-01 2.54503727e-01 -1.21613525e-01
-5.01615584e-01 4.32548553e-01 2.45830759e-01 8.62906456e-01
-2.09593147e-01 -5.56708992e-01 -1.31230056e-01 4.48907077e-01
5.94380535e-02 -2.61798911e-02 5.76542579e-02 8.74830365e-01
4.24988903e-02 5.57568789e-01 5.59544683e-01 -3.64031345e-01
7.63879597e-01 -3.86889458e-01 4.65543658e-01 -4.12772954e-01
-4.86867517e-01 -3.81925642e-01 -2.16255680e-01 -1.12736177e+00
-4.08572435e-01 -5.87920785e-01 -1.50352907e+00 -7.44372487e-01
1.32740349e-01 -2.24051848e-01 3.16846341e-01 6.59687757e-01
5.95116198e-01 5.66366136e-01 7.23229647e-01 -1.60206962e+00
-3.80373776e-01 -6.27857089e-01 -2.02498361e-01 5.06569505e-01
9.56746876e-01 -7.10066676e-01 -6.37225091e-01 -5.15520088e-02] | [8.557684898376465, -2.0580077171325684] |
300bbb1c-2f21-4865-a574-cfc5382d0e06 | small-moving-object-detection-algorithm-based | 2301.01917 | null | https://arxiv.org/abs/2301.01917v2 | https://arxiv.org/pdf/2301.01917v2.pdf | Small Moving Object Detection Algorithm in Surveillance Video Based on Motion Information | A Small Moving Object Detection algorithm Based on Motion Information (SMOD-BMI) is proposed to detect small moving objects with a low Signal-to-Noise Ratio (SNR) in surveillance video. Firstly, a ConvLSTM-PAN model structure is designed to capture suspicious small moving objects, in which the Convolutional Long and Short Time Memory (ConvLSTM) network aggregated the Spatio-temporal features of the adjacent multi-frame small moving object and the Path Aggregation Network (PAN) located the suspicious small moving objects. Then, an object tracking algorithm is used to track suspicious small objects and calculate their Motion Range (MR). At the same time, the size of the MR of the suspicious small moving object is adjusted adaptively according to its speed of movement (specifically, if the object moves slowly, its MR will be expanded according to the speed of the object to ensure the necessary environmental information of the object). Adaptive Spatio-temporal Cubes (ASt-Cubes) of the small moving objects are generated to ensure that the SNR of the moving objects is improved, and the necessary environmental information is retained adaptively. Finally, a LightWeight U-Shape Net (LW-USN) based on ASt-Cubes is designed to detect small moving objects, which rejects false detections and returns the position of small moving objects. This paper uses the bird in the surveillance video as the experimental data set to verify the algorithm's performance. The experimental results show that the proposed small moving object detection method based on motion information in surveillance video can effectively reduce the missed and false detection rate of small moving objects. | ['Haiyan Zhong', 'Hengcao Li', 'Zexi Hua', 'Ziwei Sun'] | 2023-01-05 | null | null | null | null | ['moving-object-detection', 'small-object-detection'] | ['computer-vision', 'computer-vision'] | [ 8.82706642e-02 -5.00465512e-01 7.75807425e-02 -6.66068727e-03
-1.69457961e-02 -6.47047833e-02 1.25641987e-01 -3.77891690e-01
-6.28263116e-01 2.63293162e-02 7.25797787e-02 -1.96384430e-01
-8.63355249e-02 -8.17041278e-01 -3.87334675e-01 -1.14666378e+00
-6.09389842e-01 -3.72799963e-01 1.33467805e+00 1.60296962e-01
-4.38029319e-02 3.99536520e-01 -1.28269958e+00 5.19151807e-01
4.39393252e-01 1.40563083e+00 6.34707391e-01 8.11874211e-01
2.55185068e-01 1.05159521e+00 -7.77915597e-01 4.49334115e-01
5.26909173e-01 -3.20998907e-01 -1.57826334e-01 8.08787271e-02
3.12114745e-01 -8.43331397e-01 -6.01941407e-01 1.24971569e+00
5.33196807e-01 3.90743732e-01 1.40250325e-02 -1.04062855e+00
-1.61782533e-01 4.42032635e-01 -6.35403037e-01 1.08807087e+00
-1.22786909e-01 3.62124979e-01 2.93506943e-02 -5.98492026e-01
2.58991063e-01 1.34513748e+00 5.91297507e-01 4.90020066e-01
-1.55911908e-01 -8.53823304e-01 6.36198342e-01 4.71036226e-01
-1.30036700e+00 -1.31133944e-01 2.99975574e-01 -5.99914305e-02
6.50295138e-01 3.59303981e-01 9.86923337e-01 6.91051781e-01
5.98424375e-01 9.11430240e-01 1.96450157e-03 9.28721502e-02
6.80901408e-02 -2.90259570e-01 7.42922351e-02 8.12458634e-01
3.09094518e-01 2.90912330e-01 -1.48251234e-02 7.65950456e-02
7.37687886e-01 4.87073511e-01 -4.76324588e-01 9.63561237e-02
-1.53724444e+00 5.02471745e-01 7.14718401e-01 5.06720781e-01
-3.84091854e-01 9.50466767e-02 5.96910775e-01 2.38982320e-01
2.32408375e-01 -4.51916814e-01 -5.07046640e-01 5.79409078e-02
-8.78747344e-01 -2.00359076e-01 2.95351237e-01 9.74696815e-01
1.02525890e-01 3.40179205e-01 -5.05086005e-01 2.07107425e-01
4.91962373e-01 8.36557806e-01 5.76275766e-01 -7.83106685e-01
5.03104687e-01 7.63211548e-01 1.02961123e-01 -1.46565604e+00
-4.85335827e-01 -3.39325994e-01 -1.07961631e+00 6.94807246e-02
1.40173972e-01 -3.29103589e-01 -9.93776321e-01 1.29193532e+00
6.42707407e-01 7.89911211e-01 4.61630300e-02 1.18551195e+00
9.80123162e-01 1.23742676e+00 1.72334805e-01 -8.61199319e-01
1.48357904e+00 -6.18400097e-01 -6.35364056e-01 -2.09833995e-01
4.02059913e-01 -4.24335986e-01 4.13197249e-01 1.84433639e-01
-7.85932302e-01 -9.86975491e-01 -7.54599750e-01 6.55272067e-01
-1.30855739e-01 1.87817872e-01 1.52900487e-01 4.47077721e-01
-6.53046072e-01 1.65563554e-01 -9.84756589e-01 -2.33716562e-01
4.95999694e-01 2.58201271e-01 1.24788731e-01 -1.16141625e-01
-1.29737830e+00 5.40083170e-01 7.76850462e-01 7.54081130e-01
-1.27978587e+00 -3.07466745e-01 -9.54341769e-01 8.91326666e-02
6.78929090e-01 -5.34222901e-01 9.50468719e-01 -1.32292974e+00
-8.12179267e-01 1.90015808e-01 1.08578026e-01 -4.22213823e-01
4.58300710e-01 1.26343638e-01 -8.68388414e-01 4.28247422e-01
2.36252304e-02 6.06714785e-01 9.70832705e-01 -5.07677019e-01
-1.37965333e+00 -2.39481032e-01 -2.52898633e-01 2.55861342e-01
-4.33646291e-01 5.47993422e-01 -5.90772331e-01 -6.70959115e-01
1.70504063e-01 -5.99317610e-01 -4.06910241e-01 6.94269538e-02
4.73999754e-02 -9.84734744e-02 1.51622868e+00 -5.69425404e-01
1.56971598e+00 -2.38464212e+00 -1.58856079e-01 1.37775347e-01
1.77021667e-01 8.67368281e-01 -3.36864918e-01 -5.33362746e-01
2.11508498e-01 -5.20513952e-01 -9.79292579e-03 5.80193996e-01
-7.63625443e-01 -5.34260571e-02 -3.93096060e-02 4.73258823e-01
-3.29236835e-02 6.10103488e-01 -1.10619259e+00 -6.68727577e-01
4.75952953e-01 4.30536330e-01 6.69895019e-03 7.98778012e-02
-8.51087123e-02 3.50891016e-02 -1.00946140e+00 7.17949033e-01
9.31062341e-01 1.05095752e-01 -4.11529064e-01 -3.09905499e-01
-5.75342119e-01 -3.76523644e-01 -1.49405181e+00 8.01563144e-01
2.52326041e-01 6.24782801e-01 3.63410622e-01 -4.47853416e-01
7.12674499e-01 3.01586598e-01 6.27610624e-01 -5.88846266e-01
4.25675243e-01 -1.33665875e-01 1.73315197e-01 -1.05527425e+00
3.03461164e-01 2.85746306e-01 2.94978172e-01 -9.93745774e-02
-5.92788994e-01 6.89830780e-01 1.53429970e-01 8.88426974e-02
1.31927490e+00 -1.61917433e-01 -1.72224622e-02 -2.51493547e-02
6.49254203e-01 7.26309940e-02 9.87300634e-01 6.32405937e-01
-3.77290487e-01 9.10248831e-02 -1.85200900e-01 -9.34119046e-01
-4.42657411e-01 -9.28629577e-01 3.69263068e-02 1.02121973e+00
6.20812297e-01 1.16155177e-01 -6.80453122e-01 -6.56106114e-01
-4.62536991e-01 3.51165354e-01 -4.62646306e-01 -4.84940261e-01
-9.00035560e-01 -1.02902150e+00 3.05518299e-01 5.80534160e-01
8.67920578e-01 -1.51365197e+00 -1.39393497e+00 3.31216067e-01
-3.69920544e-02 -9.22385812e-01 -8.11085999e-01 -3.84819597e-01
-7.16527224e-01 -1.33386755e+00 -6.63402379e-01 -1.18288624e+00
8.31711292e-01 8.73758972e-01 1.10993274e-01 2.30962694e-01
-1.98342010e-01 3.59708257e-02 -4.23545063e-01 -3.48617494e-01
-2.58269578e-01 -5.89612067e-01 7.55875334e-02 2.43215710e-01
4.02012527e-01 2.16931611e-01 -7.47369349e-01 6.91781402e-01
-1.24441433e+00 -4.58457917e-02 6.66513383e-01 3.98754179e-01
3.43837291e-01 5.33133447e-01 1.95818469e-01 -3.97279337e-02
2.39441827e-01 -4.25170571e-01 -9.36881125e-01 1.29114360e-01
2.09927678e-01 -7.15578854e-01 7.73582995e-01 -1.20821369e+00
-8.25292051e-01 -5.11815660e-02 2.53584802e-01 -7.60701716e-01
-7.53247226e-03 2.40954638e-01 -4.17835444e-01 -2.81149391e-02
8.11780468e-02 5.10102928e-01 -2.38921687e-01 1.16884135e-01
-1.29413605e-01 6.72488689e-01 5.80071628e-01 6.84207618e-01
4.65256512e-01 4.66952145e-01 -4.14316021e-02 -7.96386361e-01
-4.42614853e-01 -5.61273098e-01 -3.15862268e-01 -6.59387767e-01
1.07709718e+00 -8.68199766e-01 -9.44072306e-01 7.83276081e-01
-1.08737588e+00 -5.92012927e-02 -6.55666590e-02 9.90949810e-01
6.98326975e-02 4.67574894e-01 -8.42395723e-01 -8.54176581e-01
-5.88320553e-01 -1.05932677e+00 8.37774336e-01 6.79012954e-01
4.08717990e-01 -6.77493036e-01 -4.63024437e-01 -1.19217098e-01
4.95659411e-01 2.67582983e-01 2.25424230e-01 -4.20293421e-01
-9.17092204e-01 -4.62811917e-01 -2.68010139e-01 1.95163205e-01
2.14930937e-01 1.12654485e-01 -4.59178716e-01 -3.83415401e-01
4.12111193e-01 6.37648344e-01 1.06357718e+00 1.17385721e+00
8.50191057e-01 -5.91230750e-01 -8.17159951e-01 6.59415603e-01
9.02753830e-01 7.91173637e-01 5.48526049e-01 3.69517446e-01
7.57731915e-01 3.30669612e-01 1.12264156e+00 4.25837815e-01
1.41173508e-02 2.26703286e-01 7.40872145e-01 -1.09287754e-01
1.37184545e-01 2.96313912e-01 9.01061237e-01 5.53930581e-01
4.57913391e-02 -5.22184253e-01 -3.61710280e-01 3.79979223e-01
-1.83861959e+00 -1.57568872e+00 -6.12845182e-01 2.24769783e+00
1.19757168e-01 3.17304939e-01 3.79697941e-02 -6.69447631e-02
1.07817245e+00 3.17767471e-01 -5.84640443e-01 3.44192326e-01
-1.55934647e-01 -7.45592594e-01 5.15289128e-01 1.00523040e-01
-1.35195136e+00 7.48252988e-01 4.69000769e+00 8.46126318e-01
-1.22153652e+00 -5.94004430e-02 5.45816481e-01 -3.85226279e-01
4.71157283e-01 -4.36362207e-01 -9.55464900e-01 8.78780603e-01
6.11369193e-01 1.77912444e-01 2.71582250e-02 8.41854632e-01
8.42005610e-01 -2.92853504e-01 -4.04329300e-01 6.77567661e-01
1.38050601e-01 -8.72900903e-01 1.19690135e-01 -3.13297391e-01
3.77024561e-01 1.17041461e-01 1.74993739e-01 4.54502255e-02
-1.40803054e-01 -2.46498063e-01 6.42171443e-01 6.38779759e-01
3.40902984e-01 -6.52747691e-01 1.15060019e+00 6.03810549e-01
-1.71371210e+00 -6.56077862e-01 -8.09427917e-01 1.62496239e-01
3.51972908e-01 4.27926034e-01 -4.60014105e-01 2.81552494e-01
1.05810547e+00 6.45784676e-01 -4.36293989e-01 1.42395377e+00
1.65524989e-01 5.63923180e-01 -5.29013991e-01 -2.42238075e-01
7.49320626e-01 -9.20910165e-02 1.06222153e+00 1.21744895e+00
5.66952288e-01 5.08487999e-01 5.45853436e-01 4.11959916e-01
4.10546839e-01 -2.26107925e-01 -3.01138639e-01 4.95309711e-01
3.04867148e-01 1.36031067e+00 -1.02230513e+00 -8.24441850e-01
-3.70827496e-01 6.18531704e-01 -5.24435282e-01 2.14978963e-01
-1.07461381e+00 -4.21338350e-01 6.39659986e-02 6.66012913e-02
7.70049989e-01 -2.90533826e-02 7.09956110e-01 -7.52898276e-01
1.33702144e-01 -4.62913185e-01 9.72321391e-01 -8.33500922e-01
-7.51479268e-01 7.58490145e-01 -6.40065521e-02 -1.70045292e+00
1.71164319e-01 -4.75810438e-01 -1.04968488e+00 3.68592650e-01
-9.66320515e-01 -1.06895518e+00 -6.86752081e-01 7.05154121e-01
7.94056177e-01 -2.70355642e-01 5.86430356e-02 2.23826140e-01
-7.24527717e-01 8.28994811e-02 -1.49913346e-02 3.62922847e-01
2.20165133e-01 -3.92675489e-01 1.26197442e-01 1.34693038e+00
-5.55033147e-01 3.04509729e-01 3.72166723e-01 -1.18476510e+00
-1.16221249e+00 -1.66273320e+00 3.13277423e-01 -2.99413830e-01
4.26590741e-01 -3.02591231e-02 -9.04723346e-01 5.19749403e-01
-2.60300368e-01 4.40486401e-01 8.68576840e-02 -1.09025645e+00
3.81912768e-01 -2.46178493e-01 -1.15562296e+00 4.62178022e-01
8.75064135e-01 4.90196168e-01 -5.05073130e-01 2.59060651e-01
1.01996827e+00 -3.69048744e-01 -3.09996247e-01 6.63982451e-01
4.30128366e-01 -6.84929609e-01 9.05084312e-01 -5.37036180e-01
-2.67117500e-01 -1.03686929e+00 1.81108996e-01 -9.18048203e-01
-6.29045486e-01 -2.88002610e-01 -1.97780848e-01 7.30538011e-01
-2.04457104e-01 -2.22098634e-01 4.16802943e-01 2.67352641e-01
-4.38432336e-01 -5.71434915e-01 -1.12939084e+00 -6.29794836e-01
-9.23536956e-01 -3.13657075e-01 2.35810280e-01 7.11362183e-01
-3.84574741e-01 -7.08444715e-02 -4.40181136e-01 7.87026763e-01
3.12218308e-01 -5.63012064e-02 2.44096458e-01 -8.27606201e-01
2.62915015e-01 -3.15096945e-01 -6.09055102e-01 -1.14319944e+00
-3.71828318e-01 -3.28610331e-01 1.47722021e-01 -1.44466436e+00
2.35077530e-01 2.41017845e-02 -5.34524143e-01 2.31146380e-01
-5.18230617e-01 -3.01689375e-02 2.69400571e-02 3.59383002e-02
-1.18950319e+00 3.54403824e-01 1.31249726e+00 -3.12141240e-01
-5.43669164e-01 6.77162588e-01 1.49097517e-01 8.91466260e-01
4.66530263e-01 -3.67113829e-01 -1.44510493e-01 -4.12466854e-01
-4.32098389e-01 3.07784110e-01 5.87308943e-01 -1.22039807e+00
7.62160659e-01 -1.96762174e-01 1.02043200e+00 -8.54493260e-01
1.37129068e-01 -1.17668080e+00 2.60226950e-02 1.28057671e+00
-1.02477856e-01 1.58432245e-01 4.43223208e-01 6.60345793e-01
1.18174329e-01 2.01172792e-04 8.29885602e-01 -1.22212991e-01
-1.02239203e+00 6.86765254e-01 -9.72325742e-01 -3.39252681e-01
1.46458268e+00 -5.63156486e-01 -2.63106436e-01 -1.64603945e-02
-4.65348303e-01 6.72333360e-01 -2.27565795e-01 6.18365228e-01
1.20938277e+00 -1.21753478e+00 -6.71197295e-01 4.62726414e-01
-9.84205678e-02 8.74373466e-02 6.95166409e-01 1.06103718e+00
-5.71660936e-01 2.37425625e-01 -1.31868884e-01 -5.99427342e-01
-1.79525602e+00 1.05510712e+00 6.32031918e-01 3.09675366e-01
-7.78670430e-01 9.88887548e-01 2.53598809e-01 3.21011245e-01
3.64131182e-01 -8.42851579e-01 -5.29855430e-01 -6.31709471e-02
1.34049129e+00 5.46052039e-01 -4.52131271e-01 -7.88422465e-01
-5.03961027e-01 5.13821542e-01 -5.97884171e-02 4.98422354e-01
1.15634036e+00 -9.05215964e-02 -1.32568225e-01 2.30738409e-02
7.79326379e-01 -1.29716262e-01 -1.59054661e+00 -1.06433794e-01
-3.63069862e-01 -6.34490013e-01 2.11089939e-01 -4.10371125e-01
-1.39945865e+00 5.53566813e-01 9.86171007e-01 2.81185657e-01
1.50250971e+00 -1.34672627e-01 1.06808507e+00 3.04989785e-01
2.78236806e-01 -7.81368017e-01 -4.49553318e-02 4.36823010e-01
3.36347848e-01 -8.83259118e-01 -1.06073104e-01 -1.18606769e-01
-6.98780894e-01 1.22692442e+00 9.75597262e-01 1.22534581e-01
4.36003774e-01 3.67903113e-01 -3.49648707e-02 -1.80304736e-01
-6.12788260e-01 -3.61602157e-02 2.64677376e-01 6.77899241e-01
-4.43940848e-01 -2.35402688e-01 2.99920052e-01 4.61584985e-01
4.52163219e-01 -1.44555449e-01 3.49348783e-01 8.29522669e-01
-1.11953640e+00 6.98284209e-02 -7.32601225e-01 5.25617003e-01
-2.81622797e-01 -4.98246253e-02 7.83487968e-03 5.73351085e-01
6.03276849e-01 1.15228665e+00 4.82756883e-01 -5.65323412e-01
3.53698015e-01 -8.48918200e-01 -1.41001381e-02 -1.30285129e-01
-6.00550115e-01 4.37475622e-01 -3.46913904e-01 -6.18929565e-01
-4.21441317e-01 -3.75026017e-01 -1.53773177e+00 -1.51672810e-01
-6.32034481e-01 9.75497663e-02 -4.25291136e-02 7.29015350e-01
5.92960864e-02 7.40787923e-01 6.86236501e-01 -6.74396336e-01
-1.69020757e-01 -9.59329545e-01 -4.93118465e-01 -2.12699268e-02
5.90067506e-01 -3.25646549e-01 -2.43889883e-01 7.90483132e-02] | [8.753193855285645, -0.6483418941497803] |
b959879b-4f36-4cee-96c4-b320df2ab625 | segment-anything-model-sam-meets-glass-mirror | 2305.00278 | null | https://arxiv.org/abs/2305.00278v1 | https://arxiv.org/pdf/2305.00278v1.pdf | Segment Anything Model (SAM) Meets Glass: Mirror and Transparent Objects Cannot Be Easily Detected | Meta AI Research has recently released SAM (Segment Anything Model) which is trained on a large segmentation dataset of over 1 billion masks. As a foundation model in the field of computer vision, SAM (Segment Anything Model) has gained attention for its impressive performance in generic object segmentation. Despite its strong capability in a wide range of zero-shot transfer tasks, it remains unknown whether SAM can detect things in challenging setups like transparent objects. In this work, we perform an empirical evaluation of two glass-related challenging scenarios: mirror and transparent objects. We found that SAM often fails to detect the glass in both scenarios, which raises concern for deploying the SAM in safety-critical situations that have various forms of glass. | ['Choong Seon Hong', 'Sung-Ho Bae', 'Seungkyu Lee', 'Yuna Jung', 'Maryam Qamar', 'Yu Qiao', 'Chaoning Zhang', 'Dongsheng Han'] | 2023-04-29 | null | null | null | null | ['transparent-objects'] | ['computer-vision'] | [ 5.21865547e-01 1.31882191e-01 1.32130936e-01 -1.96795374e-01
-6.92758679e-01 -6.14336669e-01 2.97782689e-01 -2.37227201e-01
-2.43896753e-01 2.36351416e-01 -2.76866704e-01 -1.22511439e-01
3.16330403e-01 -5.16580760e-01 -6.95371389e-01 -6.34996712e-01
3.36554497e-01 6.61464632e-01 7.87139893e-01 -2.73513068e-02
3.71797919e-01 3.27195108e-01 -1.54704618e+00 4.39774483e-01
8.11987042e-01 9.90333974e-01 3.78868788e-01 4.95836258e-01
6.38579205e-02 6.61374211e-01 -5.54978192e-01 -6.96067274e-01
8.61330748e-01 -2.53397882e-01 -1.01438332e+00 4.71703887e-01
1.17414963e+00 -4.48559493e-01 -3.29723023e-02 1.13175154e+00
3.46053571e-01 1.74390852e-01 7.30618238e-01 -1.31001770e+00
-4.89940494e-01 6.07713759e-01 -6.11952782e-01 5.69694757e-01
1.34571092e-02 1.01430488e+00 1.02607322e+00 -5.44851005e-01
4.71991748e-01 1.35120130e+00 8.69982958e-01 8.35397005e-01
-1.04284346e+00 -3.86539042e-01 2.98763543e-01 -7.92334527e-02
-8.62161219e-01 -2.78800607e-01 3.50165159e-01 -4.28702444e-01
1.09725654e+00 3.06047887e-01 8.23262930e-01 1.08956695e+00
2.09632471e-01 1.24445975e+00 1.24340737e+00 -5.79829738e-02
2.32092798e-01 -7.81603232e-02 3.66581440e-01 7.67780006e-01
5.16993284e-01 -7.01409206e-02 -3.12330633e-01 1.17863111e-01
7.39523232e-01 -1.58242777e-01 -1.00699924e-02 6.61938339e-02
-1.19214559e+00 7.11809039e-01 8.55179489e-01 5.97978681e-02
-1.29237413e-01 3.16310287e-01 -6.53168708e-02 2.08847895e-02
3.42877030e-01 8.40338588e-01 -9.25943628e-02 1.30687386e-01
-1.09105122e+00 2.47767761e-01 8.31344724e-01 1.17258823e+00
4.87900198e-01 3.10046464e-01 -4.94620949e-01 5.21498144e-01
2.25879371e-01 3.12013239e-01 2.79481828e-01 -9.85454142e-01
5.11106789e-01 5.56368530e-01 1.25628918e-01 -3.00445020e-01
-5.20442665e-01 -2.65625149e-01 -3.06879848e-01 2.82836825e-01
6.51367426e-01 -2.08163750e-03 -1.68545103e+00 1.33076322e+00
4.06693041e-01 1.19792424e-01 -6.64651617e-02 1.18791091e+00
1.25295913e+00 4.27540421e-01 2.09167585e-01 4.77046639e-01
1.38188946e+00 -1.54246902e+00 -3.93572032e-01 -1.10926902e+00
2.51892898e-02 -8.31594825e-01 1.48052084e+00 3.07126433e-01
-1.17698181e+00 -5.21088839e-01 -9.50747073e-01 -1.02651864e-01
-2.09936008e-01 -1.97311953e-01 9.65542376e-01 7.37424314e-01
-9.44602728e-01 4.45858002e-01 -7.25578010e-01 -8.06255579e-01
8.63074183e-01 3.53561282e-01 1.23671532e-01 -2.62804717e-01
-4.68063712e-01 8.89159620e-01 8.07191655e-02 1.70567140e-01
-1.32023978e+00 -5.62744379e-01 -5.77362597e-01 -2.09043562e-01
5.52087665e-01 -1.10429311e+00 1.48589134e+00 -9.87396657e-01
-1.24402475e+00 1.36937547e+00 1.11193180e-01 -6.09469652e-01
8.95124018e-01 -5.60179174e-01 -1.56984359e-01 3.70660186e-01
1.78428248e-01 9.26945806e-01 1.42058134e+00 -1.15973413e+00
-3.07503104e-01 -3.63461465e-01 1.68843031e-01 2.65466332e-01
1.40257835e-01 1.45777300e-01 -3.41700912e-01 -5.12400150e-01
1.18184634e-01 -1.17952657e+00 -2.76342481e-01 -4.11893278e-02
-7.88530648e-01 -1.79938182e-01 8.67799461e-01 -2.52734303e-01
3.83127481e-01 -1.99158776e+00 -2.44035617e-01 -2.31581897e-01
3.17674577e-01 4.01531398e-01 2.13319529e-02 2.42858350e-01
3.92552376e-01 1.11021683e-01 -8.02440643e-01 -3.34167272e-01
-1.70229986e-01 1.65391695e-02 -3.56274009e-01 4.47090894e-01
4.49806064e-01 1.29913735e+00 -6.21337175e-01 -6.24462605e-01
2.55402088e-01 -1.13713741e-03 -3.88110846e-01 2.17597216e-01
-6.12830818e-01 5.54203749e-01 -4.70004290e-01 1.17597890e+00
5.63467324e-01 -4.31985646e-01 -5.01896620e-01 -2.72972640e-02
1.16145574e-01 -6.17428757e-02 -5.86131752e-01 1.52922547e+00
2.32241467e-01 5.85411191e-01 3.73671995e-03 -5.91375589e-01
6.25716448e-01 9.15386379e-02 3.07856500e-01 -5.31476974e-01
3.80665451e-01 -5.41879097e-04 3.44249696e-01 -8.91201854e-01
4.26985383e-01 -2.87485003e-01 4.13507149e-02 3.60229343e-01
-1.80054635e-01 -1.10219908e+00 1.48326695e-01 2.50375401e-02
1.34367275e+00 1.67288929e-01 -2.04287589e-01 -2.50084698e-01
-1.43109933e-01 3.89696449e-01 3.52596134e-01 9.56524312e-01
-8.03327799e-01 1.36553192e+00 -1.27356917e-01 -6.07072711e-01
-6.78047419e-01 -1.50570250e+00 -1.94671288e-01 9.26481664e-01
6.42405272e-01 1.08072534e-01 -1.06681287e+00 -7.74875224e-01
6.67651743e-02 6.72882974e-01 -6.59305811e-01 -2.86043763e-01
-5.31634092e-01 -1.04717433e+00 5.20630598e-01 6.31318152e-01
9.87235606e-01 -1.61881804e+00 -1.13868070e+00 2.00587913e-01
-6.09089285e-02 -1.41707695e+00 -5.09195209e-01 1.06053203e-01
-8.49455237e-01 -1.29470778e+00 -8.04910004e-01 -7.29415178e-01
5.18033803e-01 5.71564853e-01 1.46768165e+00 1.93217024e-02
-7.46643126e-01 5.30224025e-01 -1.45710334e-01 -7.72002757e-01
-3.46660793e-01 -3.19992453e-02 -3.67357790e-01 2.28477512e-02
4.15322930e-01 -1.04450531e-01 -9.00368929e-01 5.67387044e-01
-8.52980196e-01 1.52178019e-01 5.92047334e-01 1.89060662e-02
3.87573302e-01 -4.62598264e-01 3.94248903e-01 -8.63308668e-01
6.21985137e-01 -3.50605011e-01 -3.41235161e-01 2.79653341e-01
-4.16362286e-02 -3.36644799e-01 7.25063384e-02 -4.20732021e-01
-9.04353678e-01 1.10466138e-01 -3.95013541e-02 -5.40271878e-01
-4.12842929e-01 -1.25923932e-01 1.58190116e-01 -7.70189241e-02
6.47879958e-01 -6.22807592e-02 -3.98965806e-01 -3.04225266e-01
2.06598669e-01 3.90836954e-01 5.20813823e-01 -4.37644392e-01
7.58022308e-01 7.62515604e-01 -2.64281392e-01 -9.64898229e-01
-1.47424173e+00 -6.61564052e-01 -5.60030639e-01 -2.71641314e-01
1.48805785e+00 -7.27642059e-01 -4.71005112e-01 6.58893049e-01
-8.58273506e-01 -9.05831456e-01 -4.87979472e-01 1.94067746e-01
-7.05698252e-01 8.13842490e-02 -6.50864601e-01 -7.69448459e-01
-5.98685145e-01 -1.34715843e+00 1.46821165e+00 4.08373982e-01
-4.52418327e-01 -5.57061434e-01 -2.40393564e-01 9.45741236e-01
4.00973260e-01 2.68206924e-01 6.35934114e-01 -3.10781777e-01
-1.00559640e+00 -7.48522580e-02 -2.92578101e-01 2.71034598e-01
1.43189698e-01 9.01535600e-02 -1.46180749e+00 -1.99378431e-01
1.14964642e-01 -5.01374364e-01 1.39846480e+00 5.20126641e-01
1.15322196e+00 -1.79411191e-02 -2.93987900e-01 5.31695127e-01
1.07752514e+00 2.04340518e-01 6.22100353e-01 1.56829104e-01
9.39549685e-01 5.37277758e-01 4.82649595e-01 -2.29578331e-01
8.74550194e-02 4.96355206e-01 8.00248444e-01 -1.07798472e-01
-6.61681116e-01 2.66989004e-02 2.92695731e-01 1.99171349e-01
-1.14860497e-01 -4.20276046e-01 -1.02061653e+00 6.56371117e-01
-1.75205374e+00 -9.81070101e-01 -1.35807931e-01 1.94947350e+00
4.51856017e-01 4.32587266e-01 4.36446160e-01 -1.47266537e-01
7.18197465e-01 6.71977848e-02 -1.14437437e+00 -4.16833341e-01
-2.60625854e-02 7.75014237e-02 3.43740761e-01 -6.33216798e-02
-1.37789857e+00 1.38308871e+00 7.29410982e+00 3.35581452e-01
-1.14948165e+00 2.37954006e-01 6.89204693e-01 -1.33151948e-01
1.99311435e-01 -2.91758657e-01 -7.51573503e-01 5.57300806e-01
5.03906488e-01 1.23106375e-01 1.98288843e-01 8.71454895e-01
-1.64268389e-01 -2.87153125e-01 -1.15426767e+00 6.52264178e-01
1.51530430e-01 -1.07193577e+00 5.48067763e-02 -1.35344684e-01
1.02613986e+00 5.78638196e-01 4.73392695e-01 1.03144303e-01
5.87458849e-01 -1.13658857e+00 1.04446685e+00 5.36146425e-02
4.16920930e-01 -1.56475127e-01 3.19966644e-01 1.79284260e-01
-9.47214842e-01 -1.15851507e-01 -4.20864433e-01 -3.88098955e-02
2.14573443e-01 5.52196741e-01 -1.08765626e+00 5.96225671e-02
1.06743860e+00 5.71369886e-01 -6.91587448e-01 1.56657636e+00
-1.59144640e-01 8.63304675e-01 -4.78231549e-01 2.29288310e-01
6.28652573e-01 -1.36083364e-01 6.27178788e-01 1.07046521e+00
-9.85837057e-02 -5.00668809e-02 4.42722589e-01 1.17969513e+00
-2.08907023e-01 -4.55334127e-01 -7.52633393e-01 -3.69915366e-02
-7.73005635e-02 1.24638486e+00 -1.51205492e+00 -3.46659720e-01
-3.01037729e-01 1.08305466e+00 1.43449336e-01 4.78069246e-01
-8.81693065e-01 9.96233150e-02 5.35876513e-01 2.88275480e-01
4.45222706e-01 -1.88958943e-01 -5.79321086e-01 -9.68697846e-01
1.86560117e-02 -6.68176055e-01 3.48993123e-01 -8.70238125e-01
-1.52034700e+00 8.00312877e-01 5.93331866e-02 -1.11700785e+00
3.33696485e-01 -5.01798332e-01 -1.01470315e+00 3.33440185e-01
-1.19045115e+00 -1.45691323e+00 -7.54633009e-01 6.00343347e-01
9.81704056e-01 1.73539609e-01 3.50771576e-01 -1.18679486e-01
-6.64765656e-01 2.45296821e-01 -4.08602506e-01 -7.66315162e-02
4.75470752e-01 -1.24611628e+00 9.44583774e-01 8.36303532e-01
2.99126774e-01 3.41977715e-01 8.59590173e-01 -7.58096874e-01
-1.24422121e+00 -1.37866282e+00 -1.47804186e-01 -9.13860202e-01
2.99978584e-01 -4.19760853e-01 -8.21243525e-01 9.24417078e-01
2.84672379e-01 1.69191435e-01 1.73809946e-01 -3.87332559e-01
-2.24724084e-01 3.00993711e-01 -1.55229795e+00 7.60850489e-01
1.27893615e+00 -1.42500877e-01 -8.73765707e-01 7.09368467e-01
8.12242746e-01 -4.22431380e-01 -3.57216239e-01 5.27729750e-01
3.24677199e-01 -1.15228152e+00 9.60246325e-01 -4.21780825e-01
3.82476598e-01 -1.06971242e-01 7.55302235e-02 -9.64067876e-01
5.74337784e-03 -5.64809561e-01 -8.85879174e-02 9.78001893e-01
2.28098124e-01 -6.67204857e-01 9.02298510e-01 8.08029056e-01
-5.67567587e-01 -5.95389068e-01 -9.06528473e-01 -9.55872655e-01
-5.83043694e-03 -4.01916057e-01 2.54764616e-01 6.47558749e-01
-7.30936050e-01 2.86554098e-01 -1.18400417e-01 2.26812586e-02
6.15173459e-01 6.36987984e-01 7.22261608e-01 -1.21598887e+00
-2.72267878e-01 -4.51221883e-01 -4.61121351e-01 -7.71246850e-01
2.64177825e-02 -8.41777623e-01 3.86232495e-01 -1.96026993e+00
2.83637851e-01 -3.78306925e-01 -1.58449143e-01 6.22034729e-01
-3.65914136e-01 8.09931934e-01 3.65485758e-01 2.78484523e-01
-6.49804235e-01 3.56728256e-01 1.56372023e+00 -6.86712682e-01
-8.79642516e-02 4.22761261e-01 -5.93863785e-01 1.11593306e+00
1.05785894e+00 -4.52282488e-01 -2.77025491e-01 -6.44749105e-01
-1.53386772e-01 -6.39220655e-01 4.70998555e-01 -1.55146396e+00
-3.00005317e-01 -3.17225337e-01 1.55357465e-01 -4.65204358e-01
7.46174157e-01 -7.74246037e-01 -1.85330257e-01 4.98507649e-01
-2.07294505e-02 -6.90398738e-02 5.12847364e-01 5.10116339e-01
7.71382973e-02 -3.52172047e-01 1.03056550e+00 -6.56231046e-01
-1.01321757e+00 4.99518752e-01 -3.48565996e-01 4.30022597e-01
1.33435965e+00 -6.66315734e-01 -6.23283625e-01 -3.23791965e-03
-6.52474105e-01 1.05688825e-01 6.88225031e-01 6.73903227e-01
6.21941149e-01 -6.22145653e-01 -5.49503148e-01 7.13803321e-02
2.04577789e-01 3.05829078e-01 1.37636870e-01 8.93577814e-01
-6.30313635e-01 -2.21335720e-02 -1.61209479e-01 -7.76002467e-01
-1.23612785e+00 4.72528875e-01 3.66530567e-01 6.66742176e-02
-1.18961799e+00 1.19743299e+00 7.83396423e-01 5.37290089e-02
1.25236571e-01 -6.91081405e-01 3.06536078e-01 -2.63199180e-01
4.14237976e-01 2.55530268e-01 2.09565043e-01 -4.13593024e-01
-2.74615139e-01 6.95923507e-01 -1.22949205e-01 -5.71859861e-03
1.21086621e+00 7.82038346e-02 1.98054656e-01 7.02381253e-01
7.64666736e-01 -3.80622685e-01 -1.53207994e+00 1.33969665e-01
-6.34747520e-02 -3.52125973e-01 -2.51866221e-01 -8.13742101e-01
-1.27958047e+00 1.06851029e+00 5.84541380e-01 2.90863216e-01
8.16312253e-01 3.51313442e-01 1.03075457e+00 5.22747576e-01
5.00744283e-01 -8.81463885e-01 4.37282711e-01 3.37209612e-01
9.50902522e-01 -1.54399252e+00 -4.30362001e-02 -6.80278122e-01
-8.76718104e-01 6.54052019e-01 1.04179966e+00 -1.92469224e-01
1.77524984e-01 2.33470783e-01 2.85588354e-01 -7.50383675e-01
-3.07279527e-01 -7.04722226e-01 3.51798743e-01 7.40144730e-01
1.00276150e-01 2.40835059e-03 2.95686126e-01 1.10855661e-01
-4.27672535e-01 -2.18181282e-01 4.12657529e-01 1.08753145e+00
-6.68014705e-01 -4.01928008e-01 -4.06209111e-01 8.39596093e-01
-4.48734403e-01 -1.14830203e-01 -8.87556791e-01 6.94992959e-01
-3.23424727e-04 1.08575821e+00 2.54147887e-01 -3.89050618e-02
2.25918472e-01 -5.14516979e-02 8.48924994e-01 -9.71490622e-01
-1.05864549e+00 -2.65059352e-01 -5.63639551e-02 -7.48636305e-01
-5.71502805e-01 -5.94577432e-01 -1.18795562e+00 2.02435195e-01
-2.34154433e-01 -3.27729523e-01 4.08410579e-01 9.92885590e-01
-2.36663651e-02 6.29329443e-01 -6.54104576e-02 -1.20256698e+00
-3.33611637e-01 -1.05691743e+00 -5.76072812e-01 6.03528380e-01
2.32172459e-01 -7.50610769e-01 -2.06725001e-01 8.54425803e-02] | [9.598631858825684, 0.35209304094314575] |
504be558-a0a6-47f7-b1de-c63ef7492ddb | ab-initio-potential-energy-surfaces-by-1 | 2110.05064 | null | https://arxiv.org/abs/2110.05064v3 | https://arxiv.org/pdf/2110.05064v3.pdf | Ab-Initio Potential Energy Surfaces by Pairing GNNs with Neural Wave Functions | Solving the Schr\"odinger equation is key to many quantum mechanical properties. However, an analytical solution is only tractable for single-electron systems. Recently, neural networks succeeded at modeling wave functions of many-electron systems. Together with the variational Monte-Carlo (VMC) framework, this led to solutions on par with the best known classical methods. Still, these neural methods require tremendous amounts of computational resources as one has to train a separate model for each molecular geometry. In this work, we combine a Graph Neural Network (GNN) with a neural wave function to simultaneously solve the Schr\"odinger equation for multiple geometries via VMC. This enables us to model continuous subsets of the potential energy surface with a single training pass. Compared to existing state-of-the-art networks, our Potential Energy Surface Network PESNet speeds up training for multiple geometries by up to 40 times while matching or surpassing their accuracy. This may open the path to accurate and orders of magnitude cheaper quantum mechanical calculations. | ['Stephan Günnemann', 'Nicholas Gao'] | 2021-10-11 | ab-initio-potential-energy-surfaces-by | https://openreview.net/forum?id=apv504XsysP | https://openreview.net/pdf?id=apv504XsysP | iclr-2022-4 | ['variational-monte-carlo'] | ['miscellaneous'] | [ 2.36184031e-01 -1.03766821e-01 -3.91049217e-03 -1.00295760e-01
-7.82795310e-01 -2.52604455e-01 3.64761055e-01 3.37951511e-01
-6.83214784e-01 1.25352287e+00 -5.28745830e-01 -7.37813532e-01
-5.53704314e-02 -1.15042925e+00 -9.24443364e-01 -9.16753173e-01
-1.96334288e-01 6.83685303e-01 -9.07665938e-02 -3.88331771e-01
3.80812943e-01 6.33551359e-01 -1.36693501e+00 9.43118557e-02
1.03736377e+00 7.60788679e-01 2.31821612e-02 5.58761239e-01
7.17103481e-02 5.49837112e-01 -5.83553165e-02 -2.47170791e-01
-1.36295304e-01 -4.29620862e-01 -8.72698307e-01 -9.58078206e-01
4.73250777e-01 2.95669865e-02 -5.29486418e-01 1.13665903e+00
7.33697116e-01 4.54871625e-01 8.37741673e-01 -6.89479589e-01
-8.43538165e-01 6.11905456e-01 -3.83432657e-02 -1.24737784e-01
2.25089550e-01 3.57160747e-01 1.23432267e+00 -8.06513608e-01
8.50157499e-01 6.95091605e-01 9.70142543e-01 7.83648312e-01
-1.65474534e+00 -4.86338705e-01 -5.67569017e-01 6.26801252e-01
-1.77482116e+00 -3.62812221e-01 7.73420513e-01 -6.29650205e-02
1.94123340e+00 -1.74809426e-01 8.95800412e-01 8.79593194e-01
7.80498624e-01 1.52191430e-01 8.50214958e-01 -4.48786467e-01
4.38042313e-01 -3.54992390e-01 8.10222626e-02 7.43801355e-01
4.35299665e-01 4.37936753e-01 -4.18303996e-01 -2.73412824e-01
3.13005030e-01 -1.09985583e-01 -8.26449916e-02 -4.27438021e-01
-8.45797360e-01 1.17985845e+00 7.78385043e-01 1.72692746e-01
-6.29057348e-01 7.72957385e-01 2.51805216e-01 5.52876405e-02
5.23491614e-02 1.06749380e+00 -2.00398967e-01 -1.42767251e-01
-1.01957417e+00 6.22541189e-01 1.06248260e+00 4.17981207e-01
1.04174232e+00 4.11980718e-01 3.17429781e-01 1.55298263e-01
1.48364261e-01 3.71870726e-01 -1.57814130e-01 -8.74031067e-01
1.48199975e-01 3.83908600e-01 1.25012830e-01 -8.03408742e-01
-6.50143385e-01 -2.54112363e-01 -1.13554811e+00 2.46929914e-01
3.63197923e-01 -3.39131832e-01 -6.95078433e-01 1.55249500e+00
2.32675686e-01 -1.16467856e-01 5.24727069e-02 5.66949666e-01
8.39348495e-01 9.51172888e-01 5.00235520e-02 -3.04236412e-01
8.78611803e-01 -5.90658605e-01 -5.24272561e-01 1.35514379e-01
9.69345212e-01 -4.41187173e-01 5.61232805e-01 5.57807267e-01
-1.39121962e+00 -3.17787528e-01 -1.34901059e+00 -2.42437601e-01
-7.49144375e-01 -3.40585351e-01 1.08391070e+00 7.54040480e-01
-1.22373641e+00 1.62811446e+00 -1.06674445e+00 2.14522630e-01
4.17143583e-01 9.29823279e-01 -4.15538907e-01 -3.64345200e-02
-1.40509808e+00 1.26299727e+00 6.09050274e-01 1.51681855e-01
-8.41210544e-01 -6.99823022e-01 -8.64279866e-01 1.60315007e-01
3.00088853e-01 -7.87030578e-01 9.45342481e-01 -2.00245574e-01
-1.65333724e+00 2.53678262e-01 -2.88051724e-01 -5.80765545e-01
1.07108437e-01 3.69758755e-01 -2.63873041e-01 3.59784737e-02
-3.19642782e-01 7.70215452e-01 1.18041925e-01 -7.83325791e-01
4.87596542e-01 -1.82983190e-01 -8.45283568e-02 -8.79489481e-02
7.30464906e-02 -2.81668723e-01 1.39117867e-01 1.38727769e-01
-6.20425604e-02 -9.39588547e-01 -5.55214167e-01 -4.56740439e-01
-4.56659257e-01 -2.97525436e-01 1.73014402e-01 -3.17952633e-01
1.20828176e+00 -1.43161809e+00 4.02707785e-01 3.94486189e-01
4.08328593e-01 3.91737252e-01 -1.02820255e-01 9.91501570e-01
-3.17836642e-01 1.43005997e-01 -3.41978759e-01 -2.94580519e-01
4.60226387e-02 1.18762918e-01 3.60834748e-01 4.73097950e-01
3.37825596e-01 1.30897832e+00 -6.22917831e-01 -2.26954967e-01
1.82699591e-01 1.01001883e+00 -1.08896494e+00 -2.49726787e-01
-3.95463854e-01 3.64549130e-01 -1.89809263e-01 2.33759910e-01
7.42152214e-01 -6.90290034e-01 5.08969784e-01 -1.79440916e-01
-3.99675012e-01 5.84639192e-01 -1.20118093e+00 1.78189778e+00
-2.50825405e-01 4.09174502e-01 -2.01901317e-01 -1.25173926e+00
6.56305730e-01 2.76945978e-01 4.87833917e-01 -8.13294649e-01
3.37649554e-01 4.66331333e-01 5.31268299e-01 -1.50245309e-01
6.24548674e-01 -4.64016438e-01 2.18491122e-01 2.20975295e-01
2.81835198e-01 -4.34572577e-01 4.43661600e-01 3.58732790e-02
9.01212096e-01 1.59377947e-01 3.78508627e-01 -4.46459353e-01
4.52170819e-01 2.47874618e-01 1.81593031e-01 8.21743250e-01
5.39935045e-02 2.72871077e-01 4.54107046e-01 -9.09335971e-01
-1.26422322e+00 -6.69314384e-01 -3.98960888e-01 5.86172104e-01
-6.25998080e-02 -6.83284342e-01 -1.02144969e+00 1.25155210e-01
-2.70764325e-02 7.94686556e-01 -5.90177715e-01 -4.22980934e-01
-6.07525706e-01 -1.04959798e+00 4.80289340e-01 2.00818047e-01
2.17384815e-01 -1.21276593e+00 -4.10661727e-01 7.71988928e-01
3.65231335e-01 -8.12027395e-01 2.17322540e-02 5.96189201e-01
-6.73411965e-01 -9.67386186e-01 -3.92113090e-01 -3.47953290e-01
2.25400090e-01 -2.64537215e-01 1.20510745e+00 1.01198159e-01
-6.83098435e-01 -4.54424590e-01 2.27700248e-01 -1.55814573e-01
-7.37778723e-01 3.74197632e-01 2.17304811e-01 -5.44848144e-01
4.51214999e-01 -8.71126115e-01 -8.15627933e-01 -3.27782452e-01
-5.77181220e-01 5.86831681e-02 4.59083289e-01 8.72026086e-01
7.76464343e-01 9.51334611e-02 3.66692394e-01 -7.19251633e-01
6.97955608e-01 -2.87862062e-01 -8.02838027e-01 1.29176853e-02
-9.22691107e-01 5.37187219e-01 1.04231071e+00 -2.19993517e-01
-5.69376230e-01 1.44445300e-01 -6.15794003e-01 -4.43216152e-02
-2.62912642e-02 8.14972818e-01 2.90476739e-01 -7.73638070e-01
8.57958317e-01 2.21062854e-01 -1.43758263e-02 -1.84565350e-01
3.67708147e-01 2.09503174e-01 1.78067625e-01 -6.16761804e-01
3.98277491e-01 1.19742714e-01 8.88276696e-01 -7.51055658e-01
-5.29113233e-01 -1.22463899e-02 -7.45765030e-01 1.29270956e-01
1.17247045e+00 -4.96028692e-01 -1.68104529e+00 3.17946285e-01
-1.30770731e+00 -2.71807075e-01 -3.85508034e-03 5.51179886e-01
-4.62987006e-01 4.11420405e-01 -7.04744816e-01 -9.16609824e-01
-4.80975896e-01 -1.48253238e+00 6.51665032e-01 1.75260469e-01
-8.56353268e-02 -1.10764790e+00 3.16120327e-01 9.27071795e-02
6.68133438e-01 4.50537384e-01 1.00036609e+00 -2.98464566e-01
-6.17296398e-01 -5.01233876e-01 -1.99508980e-01 8.25124010e-02
-5.20917356e-01 6.55293167e-02 -7.38170505e-01 -2.66932338e-01
-4.15742606e-01 -4.56170917e-01 1.13441479e+00 6.28716171e-01
1.13943219e+00 -7.66743720e-02 -3.05184215e-01 7.41035044e-01
1.64153910e+00 3.13351274e-01 6.47526741e-01 -2.06783459e-01
9.36395764e-01 7.10143372e-02 -4.07366991e-01 2.92774379e-01
4.56314892e-01 5.88620245e-01 5.44131279e-01 8.32646489e-02
3.41860950e-01 -1.33405671e-01 8.26872960e-02 8.16027582e-01
-7.27364242e-01 -2.51499325e-01 -1.27597034e+00 1.05223330e-02
-1.49007642e+00 -1.25739419e+00 -5.03339946e-01 2.09941912e+00
9.82929289e-01 6.83243051e-02 -1.60116211e-01 7.32878968e-02
2.35654324e-01 1.70621738e-01 -8.53727937e-01 -7.44869351e-01
9.60611105e-02 9.41699207e-01 6.88989460e-01 7.38754928e-01
-9.11238134e-01 9.64885890e-01 7.06515694e+00 8.48563790e-01
-1.17133141e+00 2.93044839e-02 3.98503423e-01 -8.00332874e-02
-2.10677743e-01 1.52981773e-01 -8.63963723e-01 2.73623317e-01
1.55292225e+00 4.78776917e-02 1.10354686e+00 7.10758448e-01
7.78717399e-02 -1.24583311e-01 -1.06283581e+00 1.02105725e+00
-2.25392938e-01 -2.13206744e+00 3.45850401e-02 3.75033885e-01
8.58266890e-01 5.04699469e-01 -1.94586053e-01 3.26973170e-01
2.79919297e-01 -1.65051186e+00 2.46269852e-01 5.76058447e-01
9.62985456e-01 -9.63184416e-01 6.52003288e-01 3.38405132e-01
-1.12696218e+00 3.56260419e-01 -6.26515210e-01 -4.87298459e-01
1.82705626e-01 2.85771996e-01 -4.67016131e-01 7.67201006e-01
2.42077261e-01 5.22035003e-01 -2.50978112e-01 7.89978325e-01
1.15431495e-01 3.22878540e-01 -3.14764053e-01 -7.42510438e-01
5.28555572e-01 -6.41938984e-01 2.51000151e-02 7.63752639e-01
3.17292362e-01 1.22639149e-01 -4.08921428e-02 1.51461613e+00
-2.18370587e-01 5.67810312e-02 -6.38333201e-01 -4.80804443e-01
4.30365592e-01 1.09319782e+00 -2.94104964e-01 -2.73331940e-01
-2.44634017e-01 7.64193356e-01 6.59399092e-01 3.14612389e-01
-8.06060195e-01 -7.49488533e-01 3.88043195e-01 -1.03472531e-01
1.68125600e-01 -6.05543673e-01 -1.61600322e-01 -1.09159768e+00
-3.02559465e-01 -5.46545088e-01 -9.85238329e-02 -5.28518260e-01
-1.01395762e+00 4.71774429e-01 -1.87216997e-01 -3.52802753e-01
-3.58024567e-01 -1.22048545e+00 -7.22645581e-01 1.24311399e+00
-1.61309731e+00 -7.15381145e-01 9.58500803e-02 3.50215137e-01
-4.17326659e-01 5.68109527e-02 1.28560805e+00 3.11787516e-01
-7.60065734e-01 3.94745767e-01 4.90348756e-01 -1.38501152e-01
1.21229745e-01 -1.39782369e+00 6.66480482e-01 4.15047914e-01
-5.22269048e-02 1.02185071e+00 7.24847615e-01 -4.72098738e-01
-2.07219028e+00 -6.41042471e-01 9.67188060e-01 -2.51232117e-01
7.10151017e-01 -3.50696832e-01 -1.11454582e+00 1.75336376e-01
2.80345261e-01 9.32670459e-02 6.75146878e-01 1.33611754e-01
-1.15414798e-01 3.33151907e-01 -9.21908319e-01 3.69013518e-01
1.01691258e+00 -6.56974256e-01 -1.58175156e-01 5.29035985e-01
4.02883202e-01 -3.98782670e-01 -1.05467403e+00 2.28577986e-01
5.34157157e-01 -9.77076530e-01 1.19074774e+00 -8.09214413e-01
5.34085333e-01 -3.98220606e-02 -4.24853750e-02 -1.27589226e+00
-2.21346438e-01 -9.20668542e-01 -1.71475753e-01 2.08573222e-01
6.61102712e-01 -7.71126330e-01 8.14444542e-01 7.01856732e-01
-3.57848376e-01 -9.17752743e-01 -1.22919333e+00 -6.54694140e-01
8.48927319e-01 -5.46789885e-01 5.43989539e-01 7.59411633e-01
1.36041135e-01 4.03304040e-01 -3.70851427e-01 -1.16252579e-01
7.32023358e-01 -1.21865079e-01 2.75461555e-01 -1.41676319e+00
-4.07704175e-01 -6.66664481e-01 -2.13097885e-01 -6.87045634e-01
4.51282769e-01 -1.31852245e+00 -1.57184750e-01 -1.60230505e+00
2.97626942e-01 -9.02705044e-02 -2.39105135e-01 3.77565473e-01
2.38771029e-02 3.83677810e-01 -1.72273651e-01 -8.86099190e-02
-5.18514454e-01 6.65307820e-01 1.29514360e+00 -2.82962024e-02
-2.14624658e-01 -5.32576680e-01 -3.63254607e-01 3.63377899e-01
8.33189487e-01 -5.61398208e-01 1.75075848e-02 -1.27497703e-01
9.98552024e-01 1.87478438e-01 5.92561662e-01 -1.27417910e+00
4.20121640e-01 -7.02206492e-02 3.15144360e-01 -5.49598336e-01
5.90578318e-01 -2.55939782e-01 2.55321234e-01 7.44649887e-01
-4.52904701e-02 9.10182483e-03 2.32066810e-01 3.41075748e-01
-8.35348219e-02 -3.91786456e-01 1.05205345e+00 -4.19314235e-01
-2.43446827e-01 5.17214894e-01 -4.16165620e-01 -2.15198770e-01
6.19543731e-01 -2.03832686e-01 -1.96073636e-01 -6.15929179e-02
-8.31118584e-01 7.23289847e-02 4.32436705e-01 -4.66385037e-01
4.19597954e-01 -1.10572946e+00 -3.98433417e-01 2.65713364e-01
-1.78412840e-01 -1.92948833e-01 5.93444407e-01 7.56889999e-01
-9.95479107e-01 7.72476137e-01 -2.15903655e-01 -2.92741567e-01
-7.72388518e-01 5.46588421e-01 8.48151028e-01 -3.90160412e-01
-2.71230489e-01 6.43396914e-01 -5.64150095e-01 -5.61899602e-01
-3.19243520e-01 -1.52130932e-01 3.92685011e-02 -1.23692274e-01
2.57430762e-01 3.88507724e-01 2.24952444e-01 -6.03316486e-01
-5.06416380e-01 5.90509295e-01 7.95102585e-03 7.17399642e-02
1.54813159e+00 5.40026665e-01 -2.40637466e-01 4.53342833e-02
1.45839798e+00 -3.58005196e-01 -1.02016652e+00 9.09091607e-02
-2.54146755e-01 2.83935577e-01 3.53427917e-01 -6.46490693e-01
-5.80104947e-01 1.38140631e+00 3.45171243e-01 3.28052074e-01
4.09409612e-01 -3.21637839e-01 7.75884748e-01 1.21295226e+00
3.54446054e-01 -1.11735678e+00 -2.93482155e-01 7.58244216e-01
3.78369451e-01 -1.41398692e+00 1.00235924e-01 -5.18416166e-02
-2.32494801e-01 1.47910440e+00 1.72369748e-01 -3.68574768e-01
6.46483481e-01 1.18056156e-01 -5.89976490e-01 -6.16457760e-01
-6.72256827e-01 -2.80382168e-02 5.02870739e-01 3.44053507e-01
8.07775259e-01 2.12260231e-01 -2.32788458e-01 1.93730697e-01
-3.35541278e-01 9.06958506e-02 6.32096291e-01 6.53738737e-01
-4.47806805e-01 -1.31802881e+00 2.20529765e-01 2.89140314e-01
-4.48595941e-01 -6.74030483e-01 -2.29039788e-01 9.14476573e-01
-5.87947071e-02 8.16605151e-01 -1.60734117e-01 -2.01239079e-01
1.03324294e-01 4.55161542e-01 7.51370311e-01 -4.63538527e-01
-5.14927328e-01 -4.01323557e-01 1.16414301e-01 -7.24574566e-01
-2.86558926e-01 -4.16960627e-01 -1.51112366e+00 -9.63679492e-01
-5.40397406e-01 3.90821278e-01 7.55803227e-01 9.16991234e-01
6.26300871e-01 5.71488440e-01 1.04712777e-01 -1.24246573e+00
-7.35423267e-01 -7.41518915e-01 -4.78672355e-01 1.17741637e-01
2.44412541e-01 -6.22247279e-01 -1.87891945e-01 -5.61142802e-01] | [5.3609395027160645, 5.250645160675049] |
23fdfb5f-2b88-44c3-ab64-ef6461e5f079 | understanding-the-ability-of-deep-neural | 2101.01386 | null | https://arxiv.org/abs/2101.01386v1 | https://arxiv.org/pdf/2101.01386v1.pdf | Understanding the Ability of Deep Neural Networks to Count Connected Components in Images | Humans can count very fast by subitizing, but slow substantially as the number of objects increases. Previous studies have shown a trained deep neural network (DNN) detector can count the number of objects in an amount of time that increases slowly with the number of objects. Such a phenomenon suggests the subitizing ability of DNNs, and unlike humans, it works equally well for large numbers. Many existing studies have successfully applied DNNs to object counting, but few studies have studied the subitizing ability of DNNs and its interpretation. In this paper, we found DNNs do not have the ability to generally count connected components. We provided experiments to support our conclusions and explanations to understand the results and phenomena of these experiments. We proposed three ML-learnable characteristics to verify learnable problems for ML models, such as DNNs, and explain why DNNs work for specific counting problems but cannot generally count connected components. | ['Murray Loew', 'Shuyue Guan'] | 2021-01-05 | null | null | null | null | ['object-counting'] | ['computer-vision'] | [-1.45134866e-01 -3.06814518e-02 4.10278663e-02 -2.61999458e-01
3.49979788e-01 -5.57834089e-01 4.95522350e-01 1.44031644e-01
-7.45396197e-01 7.09475577e-01 -2.33493730e-01 -3.04163843e-01
1.23684093e-01 -1.42302537e+00 -7.44110942e-01 -3.74819845e-01
-1.62989095e-01 1.07070661e+00 3.50625932e-01 2.61582494e-01
2.56437987e-01 6.18858278e-01 -1.44490838e+00 4.41027433e-01
5.35362542e-01 5.67985713e-01 2.21528411e-01 1.05558741e+00
-3.69251430e-01 1.48522270e+00 -8.16904485e-01 -2.62962252e-01
3.94293994e-01 -5.87953985e-01 -6.89926863e-01 -1.95397362e-01
1.00385392e+00 -7.25666463e-01 -5.69697320e-01 1.20332229e+00
6.54920191e-02 1.59757677e-02 7.22543478e-01 -1.46409464e+00
-1.03145123e+00 1.06746829e+00 -5.25902331e-01 8.68338525e-01
-9.49302316e-02 3.06160897e-01 9.75520968e-01 -5.91542959e-01
3.42434168e-01 1.76858175e+00 8.04398894e-01 8.11175644e-01
-8.78741086e-01 -7.90254951e-01 -9.89805013e-02 -1.44790247e-01
-1.20928478e+00 -8.26824165e-04 3.50919999e-02 -3.07928622e-01
1.23119617e+00 2.27379370e-02 1.07188666e+00 6.58786654e-01
-2.97479481e-02 1.08015549e+00 1.01059854e+00 -6.20467186e-01
2.06894696e-01 -2.88347095e-01 5.77756464e-01 1.10089326e+00
1.26366544e+00 -6.83712363e-02 -2.44569898e-01 2.93109179e-01
1.72032690e+00 1.58434436e-01 4.31850582e-01 2.12747008e-01
-1.28381407e+00 9.05605435e-01 5.95118582e-01 6.32030666e-01
9.64973047e-02 7.53024518e-01 1.83246762e-01 7.33597279e-02
1.80820227e-01 7.74973154e-01 -4.85608995e-01 1.45248815e-01
-7.59947240e-01 2.04251826e-01 1.03932071e+00 1.13812780e+00
5.30766189e-01 1.81578875e-01 -2.21605882e-01 3.21369171e-01
-1.80238962e-01 6.36664331e-01 2.58949578e-01 -1.23422492e+00
5.55095732e-01 8.89402449e-01 -1.45392612e-01 -9.37416553e-01
-8.37636352e-01 -1.58139810e-01 -9.59204257e-01 1.62906975e-01
1.13027704e+00 -4.35330048e-02 -8.68745625e-01 1.83373272e+00
-3.73117626e-01 1.01251140e-01 -2.98678339e-01 7.01169074e-01
8.43449712e-01 4.08501297e-01 3.07770640e-01 1.15600809e-01
1.42468715e+00 -6.19983792e-01 -5.99066257e-01 -5.48584044e-01
1.04648018e+00 -1.68592602e-01 1.28546607e+00 2.68817544e-01
-1.25361693e+00 -7.22363830e-01 -1.17426813e+00 -3.67919117e-01
-4.56571192e-01 2.91010439e-01 1.38034320e+00 7.81566262e-01
-1.06607485e+00 8.23538721e-01 -1.11481464e+00 -4.70309019e-01
9.66523349e-01 6.07147813e-01 1.61128148e-01 -3.69982906e-02
-7.73511767e-01 1.25231147e+00 8.76098335e-01 -2.33400270e-01
-8.10837686e-01 -3.07637304e-01 -7.14769185e-01 4.44613993e-01
1.01954758e-01 -9.40739214e-01 1.38413990e+00 -7.89480448e-01
-7.13712454e-01 1.18627524e+00 -1.96934521e-01 -7.27385402e-01
4.14210886e-01 -1.87314644e-01 -1.58561051e-01 1.30251840e-01
3.95078920e-02 1.07762182e+00 8.57291743e-02 -9.02263284e-01
-6.89751506e-01 -4.17918742e-01 2.98735470e-01 -1.35177153e-03
-4.40448701e-01 -1.32244602e-01 1.26564756e-01 -2.43922755e-01
1.75825208e-01 -6.42669737e-01 2.86152121e-02 3.64138931e-01
-1.83345735e-01 -7.28916109e-01 5.70586920e-01 -1.47438664e-02
8.94878864e-01 -1.70535648e+00 -3.95683825e-01 -3.57931644e-01
1.00908113e+00 2.08625257e-01 -4.62649345e-01 -2.16625109e-01
1.92625135e-01 4.38029528e-01 -3.64563018e-02 -1.28818303e-01
2.92254076e-03 6.16908371e-01 1.26364855e-02 4.90448982e-01
3.90465647e-01 1.13444245e+00 -1.24748755e+00 -8.30414951e-01
2.68004030e-01 4.17619981e-02 -6.16525769e-01 -1.50288969e-01
-3.88685495e-01 -9.40943435e-02 -1.65650621e-02 5.15169621e-01
8.65073144e-01 -6.74094856e-01 1.53017774e-01 6.76379949e-02
-3.30366800e-03 2.53003865e-01 -1.10200906e+00 8.71684492e-01
-3.29817712e-01 1.26978171e+00 -5.49804568e-01 -9.28133130e-01
6.29946709e-01 -1.27403602e-01 -1.61501300e-02 -9.03814375e-01
5.57009399e-01 2.03094497e-01 6.65871441e-01 -1.88238516e-01
3.94523531e-01 -2.32893631e-01 1.86653867e-01 5.89527309e-01
2.89466560e-01 -3.75990510e-01 6.31534517e-01 3.81728888e-01
1.28336954e+00 -7.28277028e-01 5.37232816e-01 -3.45607460e-01
1.50821030e-01 3.25790700e-03 1.95724949e-01 1.33073318e+00
-3.21159929e-01 3.03670615e-01 6.42451465e-01 -8.63107443e-01
-1.35300577e+00 -1.15482390e+00 -2.60792762e-01 1.19576108e+00
1.26868770e-01 -1.13478817e-01 -7.89230406e-01 -4.01876450e-01
-3.29341069e-02 8.76496673e-01 -6.45796537e-01 -3.28836814e-02
-1.00218761e+00 -8.11242104e-01 8.92315686e-01 1.12333143e+00
1.00976884e+00 -1.41320479e+00 -7.37067699e-01 -4.19573253e-03
2.95847263e-02 -1.45664239e+00 -8.74881819e-02 2.88401842e-01
-1.35677743e+00 -1.31487203e+00 -4.58378851e-01 -1.16525137e+00
8.38379502e-01 2.88353413e-01 1.64960170e+00 7.46717155e-01
-5.37638247e-01 3.08915138e-01 2.18819782e-01 -8.01387608e-01
-5.11573255e-01 5.83217703e-02 3.00023437e-01 -1.17413318e+00
1.02572727e+00 -2.55484313e-01 -4.23783123e-01 1.89493150e-01
-8.24270964e-01 -2.20187992e-01 5.09236455e-01 4.83169138e-01
2.11533904e-01 2.56001681e-01 3.22957456e-01 -7.70895958e-01
7.38583267e-01 -1.47531912e-01 -8.64241004e-01 7.61288553e-02
-2.21274376e-01 3.38245153e-01 7.93738067e-01 -8.87639880e-01
-4.57643002e-01 -3.98605503e-02 2.77011424e-01 1.82424951e-02
-3.94775979e-02 -8.30057338e-02 1.39822990e-01 6.96123466e-02
8.01975846e-01 9.24892128e-02 -3.82763058e-01 -8.90036672e-02
3.17921102e-01 1.42277271e-01 4.12877977e-01 -5.91314197e-01
5.20176351e-01 7.42118239e-01 3.53999019e-01 -7.39006341e-01
-1.28543675e+00 -2.63568133e-01 -8.78179431e-01 -6.60696477e-02
1.01106739e+00 -7.37272441e-01 -1.41567910e+00 4.95379865e-01
-1.47595561e+00 -6.07849598e-01 -5.26548266e-01 6.07360244e-01
-4.60734397e-01 7.69333318e-02 -1.25213575e+00 -9.26249921e-01
-1.01461321e-01 -4.20414388e-01 6.13177121e-01 4.14334536e-01
-3.80864829e-01 -1.15625191e+00 -5.79849891e-02 -1.41086742e-01
1.84884325e-01 -1.14419907e-01 1.06878853e+00 -7.55760014e-01
-8.29672575e-01 -1.68775707e-01 -8.62112939e-01 4.71843809e-01
-1.10293463e-01 -2.41277926e-02 -8.07799160e-01 1.16097331e-02
5.06111234e-02 -5.01438141e-01 1.18384612e+00 7.39977419e-01
1.55892777e+00 -5.06248355e-01 -3.85653287e-01 3.85083973e-01
1.44297504e+00 -7.76940659e-02 7.57824719e-01 1.08855814e-01
7.80648947e-01 3.61554213e-02 -9.52164158e-02 3.10517341e-01
3.17524046e-01 -1.03178054e-01 3.86962980e-01 -4.80446965e-03
-3.74548078e-01 -4.32358772e-01 -1.09499522e-01 9.37814832e-01
-4.28547978e-01 -5.37239134e-01 -9.48513687e-01 4.50589418e-01
-1.55363035e+00 -1.28721333e+00 -5.55942953e-01 1.82130897e+00
7.66243815e-01 5.82978547e-01 1.96663558e-01 2.35556677e-01
7.26503015e-01 -1.01697899e-01 -5.78453600e-01 -3.63456398e-01
-3.98397356e-01 4.82396662e-01 7.75627077e-01 2.66878635e-01
-7.40431905e-01 1.15982068e+00 7.80704355e+00 6.08211517e-01
-4.44053859e-01 2.39789747e-02 6.36892855e-01 1.02317117e-01
1.20245174e-01 -1.56860456e-01 -1.02199602e+00 7.79056773e-02
7.48986542e-01 -1.31375372e-01 3.28068793e-01 1.06119347e+00
-2.92645305e-01 -3.44209313e-01 -1.77723360e+00 1.05523419e+00
-8.70346874e-02 -1.42218935e+00 3.22716653e-01 -7.39502907e-02
6.54940844e-01 -4.40463517e-03 -9.40949935e-03 7.15664864e-01
9.74562466e-01 -1.43066275e+00 6.15465164e-01 1.71344057e-01
6.14995062e-01 -4.76540625e-01 6.95396364e-01 7.44532704e-01
-1.13154149e+00 -8.92930776e-02 -1.29703057e+00 -9.55475032e-01
-3.55223298e-01 6.43397391e-01 -1.03263044e+00 -7.66263485e-01
3.93482298e-01 3.90266538e-01 -6.68811023e-01 1.15799725e+00
-4.64664400e-01 6.95589960e-01 -5.82836568e-01 -6.98276699e-01
4.45705056e-02 1.84542567e-01 -3.07952352e-02 1.40867543e+00
2.39842787e-01 5.31921208e-01 -1.41341031e-01 1.55197501e+00
-4.84509885e-01 -3.73178214e-01 -6.11047089e-01 -2.52239406e-01
6.66606605e-01 9.93683279e-01 -1.54621506e+00 -8.17841530e-01
-2.21564025e-01 6.24479651e-01 4.76634800e-01 -1.42776176e-01
-8.11698854e-01 -3.09898774e-03 1.68190762e-01 2.37953469e-01
1.30745843e-01 -6.17894113e-01 -8.10146093e-01 -1.14692748e+00
-2.20263883e-01 -4.68266845e-01 4.89082634e-01 -7.87815511e-01
-1.47432542e+00 2.02457532e-01 -1.97037980e-02 -8.12463999e-01
2.22585872e-01 -1.23729551e+00 -6.16987169e-01 3.19784224e-01
-9.54158187e-01 -6.68763041e-01 -4.25968021e-01 4.47376996e-01
4.90441382e-01 3.34654259e-03 7.62107134e-01 4.63145562e-02
-3.55255753e-01 5.41559875e-01 -3.00260037e-01 8.41200709e-01
8.33632483e-04 -1.29855263e+00 6.70055926e-01 7.40915537e-01
4.69036967e-01 7.97374547e-01 3.35265368e-01 -5.38825750e-01
-9.82378125e-01 -7.50705302e-01 8.54274511e-01 -1.02626491e+00
5.53377748e-01 -5.87150753e-01 -6.75282717e-01 9.29029167e-01
-1.78506777e-01 3.64802152e-01 2.28509441e-01 3.55957121e-01
-5.98035932e-01 2.62122124e-01 -1.30524135e+00 5.62157631e-01
1.60657680e+00 -3.27607870e-01 -8.78688395e-01 6.37601137e-01
6.33623183e-01 -3.05764377e-01 -2.32055396e-01 1.69986989e-02
4.31029797e-01 -1.07547247e+00 1.05419326e+00 -6.82942271e-01
7.39632130e-01 2.76670810e-02 6.14541098e-02 -8.15848589e-01
-6.82297170e-01 3.89916599e-01 -5.16653597e-01 6.18453681e-01
2.32994705e-01 -5.75612664e-01 9.37377512e-01 4.57575619e-01
2.61674047e-01 3.27272974e-02 -7.32800663e-01 -1.34861910e+00
5.58666110e-01 -4.26892936e-01 4.41282988e-01 1.00856447e+00
-1.66014060e-02 3.92644286e-01 2.37337023e-01 7.09334165e-02
7.76938021e-01 -2.23344103e-01 3.88345480e-01 -1.59188569e+00
-3.20060104e-02 -7.76516199e-01 -7.51923501e-01 -1.28538299e+00
7.98117518e-02 -1.05966425e+00 -2.24995106e-01 -1.71277666e+00
8.80687594e-01 -3.66569966e-01 6.12356588e-02 4.46728110e-01
-9.75172445e-02 1.00354731e-01 4.94118035e-01 1.42209709e-01
-9.17130411e-01 -2.69047171e-01 1.60782981e+00 -2.17819139e-01
2.62161475e-02 -3.26992236e-02 -6.02518678e-01 1.21408761e+00
8.97014439e-01 -8.20969045e-01 -1.66612148e-01 -8.41248512e-01
7.60483384e-01 -2.29914740e-01 5.60598969e-01 -1.50667596e+00
2.77951926e-01 -1.21365488e-01 1.13084388e+00 -8.07202041e-01
-1.98947415e-02 -6.17676377e-01 -7.79688358e-01 9.32317078e-01
-4.08231616e-01 4.19055149e-02 4.52270746e-01 1.74943611e-01
1.82946846e-01 -5.34574747e-01 1.00309885e+00 -7.98271477e-01
-6.36676669e-01 2.83698499e-01 -5.93943954e-01 5.62422633e-01
7.02301264e-01 -4.71199483e-01 -6.87539160e-01 -1.57929555e-01
-4.84466881e-01 9.22131836e-02 2.66649991e-01 -8.98799772e-05
5.73049366e-01 -1.27469587e+00 -7.26526916e-01 -9.19562429e-02
-2.73419529e-01 2.15260014e-01 -2.70027786e-01 1.82956487e-01
-1.03864467e+00 5.78326881e-01 -4.37395841e-01 -5.87109029e-01
-8.47045958e-01 8.37239623e-01 3.84789616e-01 -2.74499238e-01
-5.88994503e-01 1.25109291e+00 3.29604030e-01 -3.85258734e-01
3.40573996e-01 -1.24345553e+00 -1.77215990e-02 -6.11227714e-02
8.22002411e-01 6.96208179e-01 -4.05045480e-01 1.45066902e-01
-3.37657243e-01 4.37584728e-01 1.17956102e-01 2.58172095e-01
1.31656289e+00 3.13860774e-01 -1.76904276e-01 5.82006931e-01
1.03548348e+00 -4.35545176e-01 -9.03293610e-01 3.02525070e-02
7.20769316e-02 -2.76623964e-01 -3.93280059e-01 -5.57522058e-01
-1.10141420e+00 1.14809120e+00 4.15094703e-01 3.78555775e-01
6.54998481e-01 3.89159173e-01 6.46384656e-01 9.20628667e-01
4.34091657e-01 -1.13951433e+00 6.86715186e-01 1.06813478e+00
3.80186021e-01 -1.42201078e+00 1.58425808e-01 -1.93695799e-01
-2.50265837e-01 1.48665714e+00 1.15082288e+00 -6.12810731e-01
3.89599979e-01 7.93688118e-01 -5.47808290e-01 -4.80534196e-01
-5.57419419e-01 -3.59861493e-01 -1.24251381e-01 7.73433030e-01
5.34947693e-01 3.33543658e-01 -4.82037403e-02 3.38862956e-01
-5.16421795e-01 3.35480869e-01 7.63834476e-01 6.55103862e-01
-9.74760413e-01 -3.59742463e-01 -3.94383520e-01 8.38305652e-01
-6.92763701e-02 -4.55436021e-01 -5.50674915e-01 1.15153456e+00
4.92069393e-01 6.63334548e-01 7.40598142e-01 -1.10334896e-01
5.57667725e-02 -1.15697011e-01 1.31109858e+00 -9.38891470e-01
-2.39955977e-01 -8.27804565e-01 -2.50310749e-01 -1.52270019e-01
-3.90870690e-01 -3.51057410e-01 -1.64411414e+00 -1.03502619e+00
-6.11917555e-01 -5.96632123e-01 1.09968379e-01 9.56301987e-01
-4.72300112e-01 5.05609810e-01 -7.92407244e-02 -5.45616031e-01
-5.37434995e-01 -1.09563768e+00 -7.97430933e-01 3.31126690e-01
1.67383939e-01 -3.70579302e-01 -7.19829857e-01 -1.64607465e-01] | [8.914536476135254, 0.29966971278190613] |
b7f7dade-de09-43b4-aa31-72b47d0886a7 | learning-to-read-by-spelling-towards | 1809.08675 | null | http://arxiv.org/abs/1809.08675v2 | http://arxiv.org/pdf/1809.08675v2.pdf | Learning to Read by Spelling: Towards Unsupervised Text Recognition | This work presents a method for visual text recognition without using any
paired supervisory data. We formulate the text recognition task as one of
aligning the conditional distribution of strings predicted from given text
images, with lexically valid strings sampled from target corpora. This enables
fully automated, and unsupervised learning from just line-level text-images,
and unpaired text-string samples, obviating the need for large aligned
datasets. We present detailed analysis for various aspects of the proposed
method, namely - (1) impact of the length of training sequences on convergence,
(2) relation between character frequencies and the order in which they are
learnt, (3) generalisation ability of our recognition network to inputs of
arbitrary lengths, and (4) impact of varying the text corpus on recognition
accuracy. Finally, we demonstrate excellent text recognition accuracy on both
synthetically generated text images, and scanned images of real printed books,
using no labelled training examples. | ['Ankush Gupta', 'Andrew Zisserman', 'Andrea Vedaldi'] | 2018-09-23 | null | null | null | null | ['unsupervised-text-recognition'] | ['computer-vision'] | [ 1.08373284e+00 -1.43925980e-01 -2.06551462e-01 -5.00786006e-01
-6.35992348e-01 -7.87405729e-01 8.64790499e-01 1.82040080e-01
-5.28112411e-01 7.81039357e-01 -3.47819924e-02 -4.00543779e-01
2.03794297e-02 -5.58514953e-01 -9.70616281e-01 -6.25106990e-01
3.09927464e-01 8.02570462e-01 2.12091297e-01 2.72385478e-01
6.95849478e-01 4.92207319e-01 -1.80131149e+00 4.40055043e-01
6.79369152e-01 7.21777916e-01 1.91230446e-01 1.23592925e+00
-4.74881202e-01 6.30683303e-01 -8.12305570e-01 -4.26537573e-01
1.45599157e-01 -7.34195590e-01 -6.09919250e-01 7.14230657e-01
7.27910936e-01 -1.62369579e-01 -1.42077073e-01 1.16714311e+00
2.43182793e-01 -9.43969041e-02 1.33911788e+00 -8.90514255e-01
-7.53714323e-01 7.14007914e-01 -3.12259197e-01 -5.91035858e-02
3.97847384e-01 1.06195882e-01 7.56982803e-01 -6.91232979e-01
7.63084531e-01 9.58476543e-01 3.11284453e-01 3.14150125e-01
-1.48227167e+00 -4.12545651e-01 -2.60458022e-01 -1.10858127e-01
-1.14786816e+00 -5.46746612e-01 4.98285204e-01 -7.93615818e-01
8.10523629e-01 3.48395288e-01 3.11625779e-01 1.35091853e+00
7.80872554e-02 7.55624115e-01 1.07655728e+00 -1.31562293e+00
3.08072418e-01 5.31375289e-01 2.41984352e-01 4.18818146e-01
6.60194159e-02 6.88663730e-03 -5.49853384e-01 8.09201971e-02
7.52084911e-01 -5.91626465e-01 -1.41271353e-01 -5.57639778e-01
-1.15841937e+00 5.00389278e-01 -5.50797701e-01 4.55998659e-01
2.22357213e-01 -2.04319417e-01 5.30256808e-01 3.65715623e-01
2.13887900e-01 1.06668986e-01 -1.21500954e-01 -2.14667693e-01
-1.17005849e+00 -2.55792856e-01 9.73979175e-01 1.26482892e+00
4.73191589e-01 2.23359734e-01 3.65457721e-02 9.33362126e-01
2.86529232e-02 6.72556460e-01 9.19816494e-01 -5.46807274e-02
6.90884948e-01 4.24639821e-01 8.22623894e-02 -9.21506345e-01
6.80769533e-02 9.34589729e-02 -7.10125387e-01 2.90363848e-01
7.75404572e-01 3.71754728e-02 -1.01060402e+00 1.40020788e+00
-2.40877509e-01 -4.04571444e-01 1.56195447e-01 5.81176639e-01
2.74481267e-01 7.82450676e-01 -3.21074843e-01 -4.59582686e-01
8.78575087e-01 -6.30891562e-01 -5.29972374e-01 -2.28029892e-01
5.79267621e-01 -9.42751110e-01 1.43447280e+00 4.56764430e-01
-9.40720916e-01 -6.87554419e-01 -1.18423772e+00 9.87350345e-02
-3.64001632e-01 7.75843859e-01 -1.11263819e-01 9.61774230e-01
-7.28259861e-01 4.54923928e-01 -5.25115669e-01 -5.44235051e-01
1.30743548e-01 4.44860697e-01 -4.07320470e-01 2.97401667e-01
-6.40234113e-01 6.33377612e-01 6.93709016e-01 -9.04886276e-02
-3.64918023e-01 -5.15577421e-02 -6.00096762e-01 1.49503544e-01
1.38385817e-01 2.67313905e-02 9.92501080e-01 -1.77675700e+00
-1.60414135e+00 1.10281789e+00 -3.94537419e-01 -2.94409662e-01
9.21219826e-01 1.03614949e-01 -4.41988170e-01 1.04802147e-01
-1.43155321e-01 6.08809054e-01 1.21597612e+00 -1.17042351e+00
-4.60470825e-01 -2.39474639e-01 -9.95266736e-01 2.88600147e-01
-4.12654579e-01 -2.65363753e-01 -3.94290417e-01 -7.66148508e-01
1.60609104e-03 -6.23792946e-01 3.18773270e-01 3.89804924e-03
-5.51766217e-01 -6.83861822e-02 6.01868749e-01 -7.75879562e-01
7.84820616e-01 -2.25272942e+00 9.43480060e-02 5.75543165e-01
-5.30219018e-01 9.10470411e-02 -1.43114790e-01 5.26766896e-01
-1.77338094e-01 -7.39007667e-02 -1.60723686e-01 -1.43240139e-01
8.70960206e-02 6.61051795e-02 -6.39562726e-01 3.79316837e-01
2.65843540e-01 6.77544653e-01 -3.95484924e-01 -6.53689384e-01
2.09440470e-01 -2.08508410e-02 6.12963326e-02 2.28362054e-01
-4.50502425e-01 1.50159702e-01 -2.04762612e-02 2.22141936e-01
2.51213044e-01 -1.23026550e-01 4.73877758e-01 -3.44094932e-02
-5.57933487e-02 9.18632299e-02 -1.35351527e+00 1.10643148e+00
-1.18768148e-01 1.20189941e+00 -4.31684107e-01 -1.03921020e+00
1.36813462e+00 2.45793313e-02 -1.93216503e-01 -8.29918087e-01
2.20575601e-01 1.01095088e-01 -1.08917579e-01 -4.08004344e-01
6.42982185e-01 7.09951296e-02 4.54126894e-02 8.42620432e-01
2.37956941e-01 9.41510871e-02 5.03993988e-01 -1.77595895e-02
4.10693288e-01 2.36373618e-01 1.48961142e-01 -2.34901309e-01
4.60355401e-01 -6.07537255e-02 -1.96176797e-01 9.78402972e-01
2.18886867e-01 6.48243666e-01 8.28701258e-01 -4.06488329e-01
-1.63837719e+00 -8.00877333e-01 -3.29308242e-01 1.09118819e+00
4.88943942e-02 -4.74031270e-02 -7.87402689e-01 -4.80326384e-01
-2.10830406e-03 8.46187055e-01 -6.38386130e-01 1.81757167e-01
-4.90771085e-01 -5.61047614e-01 7.48177767e-01 5.13597846e-01
1.60208613e-01 -1.05797827e+00 -6.82327330e-01 -1.37953669e-01
7.95707926e-02 -1.12881863e+00 -4.65839803e-01 4.74937081e-01
-8.07401001e-01 -8.07259202e-01 -6.81525886e-01 -1.19214308e+00
9.90728736e-01 -4.27723974e-01 8.16224277e-01 -1.67582855e-01
-4.31500822e-01 5.14646888e-01 -1.53636038e-01 -1.10528015e-01
-9.52634931e-01 6.58171577e-03 -6.92208707e-02 2.40368024e-01
2.76110381e-01 -2.40194991e-01 7.07892701e-02 5.23798883e-01
-9.58951652e-01 2.63471872e-01 6.92379236e-01 1.18166173e+00
5.19758761e-01 7.95756560e-03 2.15838283e-01 -8.59636009e-01
6.56177998e-01 2.04681531e-01 -9.12576675e-01 6.71892107e-01
-5.47930360e-01 2.74102628e-01 8.03780437e-01 -7.90008307e-01
-9.61186230e-01 2.90836185e-01 2.02603877e-01 -5.96869141e-02
-5.77589452e-01 6.11347519e-02 -6.78993315e-02 1.37929484e-01
9.34668779e-01 9.56998527e-01 6.92024305e-02 -1.22924402e-01
3.68515611e-01 1.09636140e+00 8.11245799e-01 -6.88038468e-01
6.00339532e-01 9.44728851e-02 -2.78398693e-01 -1.18415987e+00
-1.54595509e-01 -9.75940749e-02 -1.01739061e+00 -1.26768917e-01
5.95243990e-01 -5.07310271e-01 -4.70808327e-01 6.66280687e-01
-9.41019058e-01 -5.10431051e-01 -1.44309059e-01 4.04991925e-01
-7.23630607e-01 7.41885364e-01 -4.92525846e-01 -1.01716328e+00
-1.46968603e-01 -9.66333926e-01 8.76600146e-01 4.71021459e-02
-4.20923352e-01 -9.74521160e-01 -7.11581111e-02 1.02563970e-01
-3.82161252e-02 -3.27320509e-02 1.33034229e+00 -1.03442204e+00
-3.35035324e-01 -3.92593533e-01 -2.62429625e-01 4.37956870e-01
1.86423242e-01 3.53756934e-01 -8.11527073e-01 -3.56487453e-01
-2.59704798e-01 -7.55174816e-01 4.23770726e-01 5.79162166e-02
8.40882003e-01 -3.68790716e-01 -2.92824984e-01 2.51070946e-01
1.25320601e+00 3.68804604e-01 5.98812163e-01 3.08163971e-01
4.58633333e-01 7.22831428e-01 3.68812352e-01 3.02282900e-01
-3.59278858e-01 4.02688086e-01 -7.47155249e-02 5.63810952e-02
4.47599776e-02 -4.46002483e-01 2.84510225e-01 7.93442607e-01
5.28157115e-01 -6.69862628e-01 -9.64265704e-01 4.83431876e-01
-1.55272222e+00 -9.03343737e-01 -3.23759653e-02 2.42998624e+00
8.84521961e-01 5.40297806e-01 2.43755460e-01 4.09929454e-01
1.28418159e+00 -1.69817045e-01 -6.14509583e-01 -4.99188632e-01
-3.77149403e-01 1.31725490e-01 5.06369352e-01 5.12952507e-01
-9.79977310e-01 7.36797512e-01 6.93961573e+00 9.70679224e-01
-1.29729080e+00 -5.45139551e-01 8.52430463e-01 5.39024808e-02
-8.41059834e-02 -1.09401807e-01 -8.16413999e-01 6.52719796e-01
8.13706756e-01 -1.63183004e-01 3.24379325e-01 6.18198216e-01
-3.51086199e-01 -3.50742161e-01 -1.50204861e+00 9.47396517e-01
5.28725743e-01 -1.14297199e+00 5.00815868e-01 -1.21644169e-01
7.25967884e-01 -4.59908873e-01 -1.08977978e-03 -1.03859678e-01
-8.71571228e-02 -1.20681179e+00 9.84715223e-01 4.13094163e-01
1.14232218e+00 -5.51922619e-01 4.85333800e-01 4.90801871e-01
-6.34769499e-01 1.80849433e-01 -3.90363991e-01 1.75524056e-01
-3.67235631e-01 2.16977477e-01 -1.08844483e+00 3.14752400e-01
1.76903412e-01 4.92574900e-01 -6.83261454e-01 6.93996191e-01
-1.26825562e-02 6.80181444e-01 -4.36232924e-01 -6.23938084e-01
6.00666553e-02 -3.12596321e-01 2.01954901e-01 1.63310409e+00
2.79291272e-01 -4.12929833e-01 -1.95334524e-01 8.11975718e-01
1.00976974e-01 3.33773732e-01 -5.83276629e-01 -5.07567286e-01
3.98880005e-01 8.25723171e-01 -1.29143775e+00 -3.74764949e-01
-2.60118723e-01 1.10907257e+00 1.49556726e-01 4.25581038e-01
-2.69830942e-01 -7.22613931e-01 -3.32036853e-01 -1.40251145e-01
5.42400837e-01 -9.73429382e-02 -5.34457624e-01 -9.40878868e-01
3.24112833e-01 -1.12152088e+00 1.28775999e-01 -8.52430582e-01
-1.06647813e+00 4.27967846e-01 -1.40704826e-01 -1.22839379e+00
-5.33393085e-01 -8.75279069e-01 -4.54573601e-01 1.08968031e+00
-7.25636125e-01 -7.47033834e-01 7.22698867e-02 2.33827204e-01
7.16326118e-01 -6.18467629e-01 7.96752691e-01 -2.38509849e-01
-3.67393374e-01 9.63812590e-01 6.23703301e-01 3.81993234e-01
7.16657043e-01 -1.12976682e+00 5.34302235e-01 8.40793014e-01
5.21552444e-01 2.14159250e-01 6.42099857e-01 -6.93861723e-01
-1.13673496e+00 -7.04377413e-01 7.15938509e-01 -4.23409730e-01
5.53254366e-01 -8.25931191e-01 -8.99900675e-01 6.32542014e-01
1.64681911e-01 -5.21814585e-01 5.02231359e-01 -2.22966552e-01
-2.62640208e-01 1.46734595e-01 -8.65973413e-01 7.38145888e-01
5.76727927e-01 -6.60860121e-01 -6.10379398e-01 3.22870970e-01
1.82769448e-02 -4.35167432e-01 -5.20755589e-01 -1.30996093e-01
8.32556248e-01 -9.81591523e-01 3.84235024e-01 -4.19354469e-01
6.53204262e-01 -6.74658716e-02 -1.06597558e-01 -8.03354561e-01
-4.87795100e-03 -3.80232871e-01 3.79459977e-01 1.29255009e+00
5.30653954e-01 -3.32055271e-01 1.01838541e+00 3.44198883e-01
3.95597070e-01 -2.62448698e-01 -9.02888894e-01 -6.89548254e-01
8.87264907e-02 -1.71103165e-01 4.96188179e-02 9.52165723e-01
2.52568454e-01 5.45425177e-01 -4.63223666e-01 1.25976279e-02
4.12770599e-01 1.48556769e-01 8.59917402e-01 -9.78804708e-01
-5.69157124e-01 -4.56669152e-01 -6.40194416e-01 -1.17194092e+00
1.44345194e-01 -8.16718519e-01 2.59849072e-01 -9.03581977e-01
1.88857466e-01 -6.33436218e-02 4.51279491e-01 2.22777963e-01
1.06326200e-01 1.77254174e-02 5.03293388e-02 4.52851623e-01
-2.16115355e-01 3.11899483e-01 7.82795906e-01 -1.63029611e-01
-2.55391914e-02 -1.51421204e-01 -2.59500537e-02 4.71973687e-01
5.73734462e-01 -3.53846103e-01 -5.11575341e-01 -3.76453817e-01
1.05432741e-01 2.28319868e-01 1.26244381e-01 -8.51405382e-01
4.00596410e-01 -1.20533323e-02 9.49452043e-01 -7.45288730e-01
-1.07305869e-01 -7.36784279e-01 5.58770038e-02 2.51632452e-01
-8.48178446e-01 1.38806403e-01 3.35611641e-01 5.68325043e-01
-7.90606588e-02 -9.24424410e-01 7.02256620e-01 1.45055711e-01
-4.60657865e-01 -5.90471387e-01 -5.25888026e-01 -6.98637515e-02
8.77506256e-01 -8.46005738e-01 -1.97311565e-01 -2.67339081e-01
-6.19096458e-01 -9.36599895e-02 7.44025826e-01 2.36401007e-01
3.15147936e-01 -1.02488041e+00 -5.78673780e-01 7.43606508e-01
1.39947996e-01 -3.09049904e-01 -2.60158360e-01 2.94677317e-01
-6.12475991e-01 3.92660260e-01 -3.54720592e-01 -1.02087986e+00
-1.40742350e+00 3.91268373e-01 1.55559793e-01 -2.09526736e-02
-5.04175425e-01 5.08243203e-01 3.21517214e-02 -3.53963107e-01
7.04757214e-01 -1.22165762e-01 5.95909311e-03 -1.16977289e-01
2.84957379e-01 9.67520624e-02 1.72911391e-01 -4.20206130e-01
1.05366699e-01 7.49424696e-01 -3.36308300e-01 -3.74132633e-01
6.99620426e-01 5.02854623e-02 1.90320253e-01 9.10692632e-01
9.77747619e-01 -6.75141588e-02 -1.20473325e+00 -2.91586537e-02
2.43558735e-01 -3.03294599e-01 -4.47709113e-01 -8.35657418e-01
-3.81917089e-01 8.93192112e-01 7.87413895e-01 2.28342041e-01
8.21979046e-01 -3.43747050e-01 7.08713308e-02 6.62945151e-01
1.20225307e-02 -1.22494698e+00 1.78773150e-01 4.31214184e-01
5.32902360e-01 -9.42581415e-01 -8.06509182e-02 -1.78095907e-01
-6.54079437e-01 1.70942402e+00 4.46270972e-01 6.17339881e-03
-9.86196920e-02 3.80529970e-01 1.25477880e-01 2.16371149e-01
-6.79538846e-01 1.81732506e-01 2.29665235e-01 5.15898764e-01
4.46125984e-01 -3.38187397e-01 -1.07471041e-01 5.24448119e-02
-2.79108912e-01 -8.02349374e-02 7.04167545e-01 1.05179119e+00
-3.19587737e-01 -1.05007374e+00 -5.30135930e-01 6.49863422e-01
-2.17623308e-01 -8.36092606e-02 -7.17501104e-01 7.62623131e-01
-2.35601842e-01 4.35233295e-01 5.77657938e-01 -1.60449490e-01
2.32812166e-01 4.93800044e-01 6.72404349e-01 -3.56833428e-01
-2.41940409e-01 5.79323284e-02 9.23939124e-02 3.32762450e-01
-8.90699700e-02 -7.80767560e-01 -6.93005502e-01 7.77081549e-02
-6.00665271e-01 2.55521894e-01 6.94204748e-01 9.58831728e-01
5.75586148e-02 6.90161213e-02 5.23331642e-01 -8.68897140e-01
-7.88829327e-01 -9.88366127e-01 -5.35916150e-01 4.59032416e-01
7.09624290e-02 -1.34273767e-01 -5.25111735e-01 7.84951508e-01] | [11.819461822509766, 2.35841965675354] |
e119eace-6e56-484a-b155-2e40cdee2768 | unsupervised-audio-visual-subspace-alignment | 2102.03673 | null | https://arxiv.org/abs/2102.03673v1 | https://arxiv.org/pdf/2102.03673v1.pdf | Unsupervised Audio-Visual Subspace Alignment for High-Stakes Deception Detection | Automated systems that detect deception in high-stakes situations can enhance societal well-being across medical, social work, and legal domains. Existing models for detecting high-stakes deception in videos have been supervised, but labeled datasets to train models can rarely be collected for most real-world applications. To address this problem, we propose the first multimodal unsupervised transfer learning approach that detects real-world, high-stakes deception in videos without using high-stakes labels. Our subspace-alignment (SA) approach adapts audio-visual representations of deception in lab-controlled low-stakes scenarios to detect deception in real-world, high-stakes situations. Our best unsupervised SA models outperform models without SA, outperform human ability, and perform comparably to a number of existing supervised models. Our research demonstrates the potential for introducing subspace-based transfer learning to model high-stakes deception and other social behaviors in real-world contexts with a scarcity of labeled behavioral data. | ['Maja J Matarić', 'Leena Mathur'] | 2021-02-06 | null | null | null | null | ['deception-detection'] | ['miscellaneous'] | [ 3.61680567e-01 2.11100932e-02 -2.94925183e-01 -5.84077060e-01
-1.17072809e+00 -5.58930635e-01 4.24911886e-01 -2.62276441e-01
-4.28579479e-01 5.98009109e-01 6.56864464e-01 -2.60427028e-01
6.60943761e-02 1.70767531e-02 -4.90669727e-01 -1.38765633e-01
2.18401924e-01 -1.50182202e-01 -3.96348149e-01 -1.64323047e-01
2.26446688e-01 1.46914020e-01 -1.15850949e+00 9.52584624e-01
5.44210851e-01 6.01830542e-01 -7.71958530e-01 9.07381773e-01
9.24247026e-01 1.32749236e+00 -9.37526166e-01 -9.36615229e-01
1.29001454e-01 -5.75414658e-01 -6.85894310e-01 2.72980481e-02
1.08127749e+00 -1.10760057e+00 -5.67235351e-01 8.42825294e-01
6.79464340e-01 -1.65421993e-01 7.76907623e-01 -1.74996209e+00
-9.97031391e-01 -1.75949052e-01 -4.22542661e-01 3.77725869e-01
1.11144924e+00 4.74056214e-01 6.31998777e-01 -5.97215295e-01
7.87732422e-01 1.31180787e+00 9.74226832e-01 9.11743402e-01
-1.18229342e+00 -1.06265509e+00 -6.33724630e-01 5.20393431e-01
-9.42409098e-01 -7.30960310e-01 1.02171409e+00 -7.71876454e-01
6.08662128e-01 3.22045803e-01 7.90626109e-01 2.34020519e+00
2.80205905e-01 7.33121097e-01 1.30844676e+00 1.82789396e-02
-1.89106097e-03 6.83467016e-02 -2.13995799e-01 7.82424033e-01
4.26003747e-02 1.78006500e-01 -1.13609600e+00 -7.41116524e-01
4.03888047e-01 1.33486032e-01 -3.84331308e-02 -3.52909863e-01
-1.34749234e+00 8.66532445e-01 2.44609356e-01 3.72886546e-02
-2.50531226e-01 1.64773762e-01 7.08934486e-01 4.79437321e-01
6.09867990e-01 7.20026195e-01 -4.79259677e-02 -6.99587226e-01
-1.14403164e+00 1.30087197e-01 5.24913609e-01 4.96082187e-01
4.81518395e-02 1.44396976e-01 -2.27465510e-01 6.54092848e-01
-1.13301374e-01 6.83416486e-01 4.67765838e-01 -1.40909445e+00
4.19529021e-01 3.97184700e-01 3.28436345e-01 -1.65518999e+00
-3.95942122e-01 1.02067299e-01 -4.16522056e-01 1.32567286e-01
4.40426975e-01 -3.53323579e-01 -5.13738394e-01 1.54583085e+00
2.81685859e-01 4.88985181e-01 -1.12799808e-01 9.96730685e-01
7.36698270e-01 2.93065041e-01 1.36996165e-01 -1.16088927e-01
1.05463231e+00 -4.58258450e-01 -6.73322678e-01 -2.18345404e-01
8.90826821e-01 -4.87141848e-01 1.21619165e+00 6.18498504e-01
-7.92987704e-01 -2.64965352e-02 -8.41168463e-01 -1.33008078e-01
-1.47371486e-01 -8.75833631e-03 4.68020797e-01 1.36063933e+00
-7.45851099e-01 2.90373862e-01 -4.57781255e-01 -4.86705303e-01
9.39236879e-01 -3.89406569e-02 -8.41359735e-01 -2.99047172e-01
-1.14878273e+00 1.02225137e+00 -3.46515685e-01 6.23181313e-02
-1.32447016e+00 -5.72131693e-01 -1.15245926e+00 -2.04049900e-01
2.67627388e-01 -5.59402943e-01 1.20542455e+00 -1.62846279e+00
-1.12100565e+00 1.36796260e+00 2.22863406e-01 -2.51747161e-01
5.93749702e-01 -2.57850200e-01 -7.40380943e-01 6.29942179e-01
1.24969855e-01 4.93023872e-01 1.34449661e+00 -9.87692833e-01
5.04893512e-02 -4.67381895e-01 3.29907313e-02 2.72992015e-01
-1.02957737e+00 5.44802606e-01 8.95930529e-01 -2.73158342e-01
-7.11248398e-01 -8.18403721e-01 4.40545678e-01 2.16060113e-02
-7.90158808e-02 2.51099497e-01 1.10720491e+00 -1.13410854e+00
7.99422443e-01 -2.24028230e+00 -5.12727015e-02 -2.23910689e-01
4.70314801e-01 5.82759738e-01 -1.21246964e-01 3.20774108e-01
-1.12740390e-01 2.21203074e-01 1.36792898e-01 -2.81682312e-01
-3.94462049e-02 -1.64595842e-01 9.24140885e-02 9.85977352e-01
1.46194875e-01 1.03281951e+00 -1.32140124e+00 -6.11637294e-01
1.26808763e-01 1.17245398e-01 -6.11542284e-01 6.27522022e-02
5.90336025e-01 5.06253064e-01 -6.92530796e-02 1.02812576e+00
1.88729823e-01 2.28503868e-01 4.22518067e-02 6.45817295e-02
4.90972430e-01 -1.32801607e-01 -2.84196973e-01 1.48838735e+00
6.78336918e-02 1.09600616e+00 4.04953599e-01 -9.78355050e-01
4.91046935e-01 4.80952948e-01 5.16326368e-01 -5.96247137e-01
1.61714032e-01 4.65952791e-02 -6.56516701e-02 -1.11009181e+00
3.29592258e-01 -6.61589146e-01 -6.12335622e-01 2.72743940e-01
3.08282346e-01 -2.20551252e-01 -6.98666632e-01 4.59294885e-01
1.53723192e+00 3.11304666e-02 2.83936888e-01 3.32352638e-01
1.45746037e-01 1.31946653e-01 4.77124661e-01 5.87041497e-01
-1.04898357e+00 7.65240908e-01 8.48479152e-01 -5.34845829e-01
-9.26569283e-01 -9.10344005e-01 1.90838858e-01 1.34684908e+00
-1.92069292e-01 -7.33180419e-02 -8.51085782e-01 -9.77045059e-01
1.03200674e-01 8.06759715e-01 -6.70901835e-01 -8.36058974e-01
-1.27964422e-01 -4.23063159e-01 1.17027235e+00 1.31959960e-01
2.26320684e-01 -1.08163834e+00 -8.73848379e-01 -2.82359749e-01
-6.17779493e-01 -1.37270308e+00 -5.58353722e-01 -8.18880677e-01
-3.50177377e-01 -1.45121062e+00 -7.56273389e-01 -3.02442104e-01
3.25635105e-01 3.33651543e-01 8.97876561e-01 -1.78723544e-01
-4.88032460e-01 1.26340663e+00 -3.54024053e-01 -6.39353275e-01
-4.76182163e-01 -5.16600490e-01 4.41397905e-01 3.63387048e-01
6.58156753e-01 -2.79828310e-01 -4.74266440e-01 3.46695632e-01
-7.57216275e-01 -3.88326555e-01 3.20722610e-01 9.56643224e-01
-3.93020034e-01 -7.10971534e-01 9.37010825e-01 -5.34466386e-01
1.16394925e+00 -8.19754303e-01 1.22815043e-01 9.70487110e-03
-1.32644132e-01 -1.07181239e+00 2.89525151e-01 -8.12651813e-01
-8.43218505e-01 -1.36558220e-01 4.62781310e-01 -7.57576942e-01
-5.80531061e-01 3.63479376e-01 2.15244576e-01 -2.38849849e-01
1.11513984e+00 -4.37099896e-02 1.87482432e-01 1.51554987e-01
-3.40686576e-03 1.09277821e+00 6.35647357e-01 -2.83521950e-01
6.84170663e-01 4.30550337e-01 -8.62212926e-02 -1.03211546e+00
-8.65054667e-01 -6.76885664e-01 -6.57846510e-01 -7.73007393e-01
8.11165035e-01 -1.09134984e+00 -9.83870029e-01 6.66289508e-01
-8.94563556e-01 -2.83219635e-01 -1.29448995e-01 5.76079726e-01
-7.90658593e-01 7.06344306e-01 -6.13705516e-01 -1.15253496e+00
-8.74488801e-02 -5.17909527e-01 1.29829443e+00 -4.05465662e-01
-6.57324076e-01 -6.38120890e-01 2.38062799e-01 1.47956669e+00
7.51944855e-02 8.32664073e-01 3.65092039e-01 -7.86374092e-01
2.29864880e-01 -7.96673119e-01 -1.16554163e-01 5.91738105e-01
-1.27365282e-02 -3.55931312e-01 -1.04464614e+00 -3.27100545e-01
1.35041624e-01 -1.52560580e+00 4.02804285e-01 6.13027960e-02
1.00901341e+00 -5.72043478e-01 7.83953220e-02 2.58314222e-01
3.92584592e-01 -1.89945087e-01 8.23145926e-01 2.93846847e-03
8.18401098e-01 6.73547447e-01 6.55413270e-01 3.49355906e-01
4.69739556e-01 4.96094793e-01 4.93939370e-01 -1.41196474e-01
1.50872216e-01 -4.19910759e-01 9.89691496e-01 4.21306878e-01
2.22544931e-02 -5.59638292e-02 -1.10440814e+00 6.76905572e-01
-1.68069768e+00 -1.59661138e+00 -7.01959059e-02 2.02011895e+00
6.14191651e-01 -2.64784575e-01 6.46007240e-01 6.18282855e-02
5.56248784e-01 1.97252244e-01 -5.42106271e-01 -8.33581507e-01
1.06923595e-01 -4.55899268e-01 4.03421149e-02 7.94267133e-02
-1.21692443e+00 6.57722712e-01 7.16041422e+00 5.33339977e-01
-8.97732735e-01 5.30989647e-01 8.81607890e-01 -8.88969839e-01
-6.16859347e-02 -5.73025346e-01 7.87583143e-02 4.89396155e-01
1.40744841e+00 1.68981507e-01 3.26385528e-01 6.93358898e-01
5.43649733e-01 2.57341638e-02 -1.33828330e+00 1.36134255e+00
1.08039427e+00 -7.67517984e-01 -4.20180380e-01 2.46002033e-01
6.96987212e-01 -4.17431802e-01 4.34780508e-01 4.40727711e-01
2.22363584e-02 -1.46989644e+00 6.15911901e-01 4.10141677e-01
1.17962897e+00 -7.01762557e-01 8.61192822e-01 6.28981829e-01
1.30420193e-01 -4.44828570e-01 -1.52973995e-01 -3.22131157e-01
-3.51967942e-03 1.29205227e-01 -1.05382490e+00 -4.87903543e-02
6.33122027e-01 8.11543822e-01 -4.89040643e-01 6.70258462e-01
9.06320848e-03 9.49976146e-01 8.51497352e-02 -1.20359398e-01
3.37139815e-01 3.18396062e-01 5.68297088e-01 1.37735927e+00
2.54130393e-01 9.91460681e-02 -2.15662226e-01 5.64787328e-01
-3.51855785e-01 -1.29698455e-01 -1.27514195e+00 -1.61053494e-01
-1.42622618e-02 1.16968703e+00 1.67538613e-01 -1.05548836e-01
-2.26810008e-01 1.38595629e+00 2.56474137e-01 2.24929616e-01
-9.28990841e-01 3.07617728e-02 5.14963210e-01 3.46510440e-01
-4.64191079e-01 -1.37123495e-01 -5.18879388e-03 -1.49044061e+00
5.59308641e-02 -1.46563900e+00 5.19837976e-01 -1.22113740e+00
-1.41669321e+00 -2.18989670e-01 2.08118349e-01 -1.33007979e+00
-4.57696468e-01 -5.79580843e-01 -3.08958262e-01 1.87000617e-01
-9.50938284e-01 -1.64440799e+00 -3.28232378e-01 1.02243936e+00
2.93223143e-01 -3.67716402e-01 1.02054346e+00 6.54735267e-02
-4.91476864e-01 6.23726606e-01 -1.38161287e-01 3.94111902e-01
1.35713494e+00 -9.21717703e-01 -2.73239344e-01 6.69232011e-01
5.70954196e-02 2.76422858e-01 4.36426550e-01 -7.14638829e-01
-1.27811933e+00 -6.54812396e-01 6.61910713e-01 -1.10916495e+00
8.58394623e-01 -7.79923975e-01 -5.24065435e-01 9.56346810e-01
1.10465527e-01 7.88022056e-02 1.36972451e+00 3.22445184e-01
-9.01990592e-01 -4.90854010e-02 -1.88990235e+00 5.72844267e-01
1.12650144e+00 -1.06959927e+00 -8.34426045e-01 6.49269044e-01
8.78654271e-02 -1.92071095e-01 -7.70601153e-01 -3.12570930e-02
1.10599542e+00 -1.00399363e+00 1.11584544e+00 -1.34963632e+00
1.03030384e+00 4.53091711e-01 6.81233332e-02 -1.53377426e+00
-2.26767555e-01 -7.59313285e-01 -1.47339150e-01 9.38454151e-01
-3.61182503e-02 -3.82168174e-01 8.21899056e-01 9.95656133e-01
1.20814562e-01 -4.06671986e-02 -1.40511441e+00 -6.55430436e-01
1.46066979e-01 -3.95499617e-01 2.84162629e-02 1.62786937e+00
7.96684682e-01 5.62358834e-02 -1.23964870e+00 4.66799177e-02
6.95020258e-01 -5.74085355e-01 9.06373858e-01 -1.04590750e+00
1.43537775e-01 4.25422452e-02 -1.19696093e+00 -8.66388604e-02
5.36676168e-01 -4.21330571e-01 -2.44690418e-01 -8.64450932e-01
7.29383349e-01 4.92948681e-01 -9.91540179e-02 6.02824748e-01
1.50681455e-02 6.41527534e-01 1.86443567e-01 7.39398897e-02
-8.98193657e-01 4.22078997e-01 1.08541989e+00 -3.89876485e-01
2.28707895e-01 -3.90904218e-01 -8.51060688e-01 9.39464748e-01
6.76619411e-01 -6.29264295e-01 -3.55378538e-01 -2.88075238e-01
1.15380995e-01 3.34361345e-01 1.19885242e+00 -9.10475135e-01
9.37111601e-02 -4.79011118e-01 7.37144589e-01 2.19656050e-01
8.10569227e-01 -8.23008120e-01 -3.73407274e-01 3.10030222e-01
-4.09446537e-01 -1.64818108e-01 8.37343708e-02 7.83756375e-01
-2.70038731e-02 -1.66625500e-01 5.69307208e-01 -3.26575369e-01
-2.09933430e-01 -3.61468136e-01 -9.35533524e-01 1.23121716e-01
1.36210203e+00 -3.86597663e-01 -5.95740080e-01 -1.29380524e+00
-7.15020835e-01 -6.48166463e-02 4.34079349e-01 5.14456987e-01
1.01057744e+00 -1.37019169e+00 -1.02145803e+00 -5.82200326e-02
4.28959429e-01 -8.22438955e-01 3.23616773e-01 8.61138165e-01
-2.25672305e-01 1.33547336e-01 -5.98978937e-01 -3.21163923e-01
-1.62357962e+00 6.28837109e-01 3.45789760e-01 1.31384268e-01
-1.27748191e-01 5.28304160e-01 -1.89687565e-01 -5.12568712e-01
-1.04345217e-01 3.47756207e-01 8.68272856e-02 2.26756129e-02
5.90170741e-01 7.89323390e-01 -2.79126734e-01 -1.20117211e+00
-4.12088543e-01 -1.32536903e-01 2.52336025e-01 -2.15875074e-01
1.21213579e+00 1.16401859e-01 3.50813955e-01 5.46143115e-01
1.33392787e+00 -1.67864293e-03 -8.72481465e-01 2.88576216e-01
-2.43307710e-01 -8.31734121e-01 -7.12782666e-02 -9.65338469e-01
-5.39005637e-01 9.14210856e-01 4.87658262e-01 -7.53791630e-02
9.97932196e-01 -2.76903421e-01 6.93887472e-01 4.97719675e-01
4.43311483e-01 -1.26201463e+00 8.86891007e-01 -1.68745071e-01
1.12223542e+00 -1.42641425e+00 1.22276835e-01 -1.67363659e-01
-1.18338788e+00 7.25559473e-01 6.04974926e-01 -1.56055108e-01
1.87587589e-01 -3.37537706e-01 1.58927262e-01 -4.48802739e-01
-3.83431822e-01 4.25856233e-01 3.33777130e-01 1.20058525e+00
-8.79640784e-03 2.55271316e-01 1.39015922e-02 5.31947017e-01
-1.64291263e-01 2.85639524e-01 8.99032056e-01 7.14776993e-01
1.02926731e-01 -5.39035320e-01 -6.82218254e-01 5.80807447e-01
-8.55248213e-01 1.68687910e-01 -1.34663522e+00 5.48090935e-01
1.13724411e-01 1.29115272e+00 -1.75436497e-01 -6.01191223e-01
3.54806900e-01 4.15700138e-01 4.31743383e-01 -5.65235555e-01
-8.10321689e-01 -2.65517920e-01 6.41122580e-01 -1.00541294e+00
-5.41055083e-01 -1.06836903e+00 -1.06394820e-01 -4.58042979e-01
-4.43408638e-03 -5.64082265e-01 2.92053789e-01 7.96560287e-01
5.48709989e-01 -1.30774155e-01 5.31301439e-01 -9.22841847e-01
-6.33818924e-01 -9.94158387e-01 -6.01580322e-01 1.08103919e+00
7.13600636e-01 -4.68769461e-01 -4.69364405e-01 1.14916302e-01] | [13.266210556030273, 2.035550832748413] |
33084920-73b3-44d2-9a19-7f3f8b7c3754 | choral-collecting-humor-reaction-labels-from | null | null | https://aclanthology.org/2021.emnlp-main.364 | https://aclanthology.org/2021.emnlp-main.364.pdf | CHoRaL: Collecting Humor Reaction Labels from Millions of Social Media Users | Humor detection has gained attention in recent years due to the desire to understand user-generated content with figurative language. However, substantial individual and cultural differences in humor perception make it very difficult to collect a large-scale humor dataset with reliable humor labels. We propose CHoRaL, a framework to generate perceived humor labels on Facebook posts, using the naturally available user reactions to these posts with no manual annotation needed. CHoRaL provides both binary labels and continuous scores of humor and non-humor. We present the largest dataset to date with labeled humor on 785K posts related to COVID-19. Additionally, we analyze the expression of COVID-related humor in social media by extracting lexico-semantic and affective features from the posts, and build humor detection models with performance similar to humans. CHoRaL enables the development of large-scale humor detection models on any topic and opens a new path to the study of humor on social media. | ['Julia Hirschberg', 'Shayan Hooshmand', 'Zixiaofan Yang'] | null | null | null | null | emnlp-2021-11 | ['humor-detection'] | ['natural-language-processing'] | [-5.53648472e-01 1.05164915e-01 -3.04333363e-02 -6.67595267e-02
8.33763480e-02 -5.83521426e-01 7.86603272e-01 3.54899764e-01
-4.07521240e-02 5.67207575e-01 1.21586621e+00 7.92613775e-02
4.74654973e-01 -6.11669421e-01 2.90603161e-01 -1.06311319e-02
2.86969662e-01 3.08702499e-01 -2.08094995e-02 -6.42356098e-01
6.00380898e-01 2.37599500e-02 -1.35851109e+00 7.99601316e-01
6.11374199e-01 4.00421977e-01 -4.70903851e-02 6.48121357e-01
3.50467972e-02 2.08094692e+00 -6.72792017e-01 -7.33066499e-01
-1.66695014e-01 -1.00177264e+00 -1.12167311e+00 2.05260545e-01
3.93049717e-01 -1.67541161e-01 -4.32283282e-01 8.03922892e-01
5.11853755e-01 1.63439125e-01 6.10421360e-01 -1.03898430e+00
-6.95341945e-01 9.12289977e-01 -2.63450921e-01 -1.03049286e-01
1.00272965e+00 1.93078920e-01 1.31673801e+00 -1.05530810e+00
1.01873136e+00 1.26090825e+00 7.73721755e-01 6.97558701e-01
-1.13894892e+00 -3.06327879e-01 -1.03038359e+00 6.24371886e-01
-8.63216579e-01 -1.53394058e-01 1.17121446e+00 -1.06770062e+00
7.08050430e-01 5.32058537e-01 7.60894179e-01 1.20741391e+00
-5.74422441e-02 1.04545987e+00 1.40440297e+00 -4.37955648e-01
7.23880157e-02 4.97065037e-01 3.03416073e-01 7.86900461e-01
-3.42951834e-01 -8.35612059e-01 -8.64364684e-01 -4.44584578e-01
2.16499940e-01 -3.53276916e-02 -3.45518231e-01 1.30092785e-01
-8.45273852e-01 1.14378488e+00 5.09645939e-01 4.24080044e-01
-1.84662625e-01 -3.87265950e-01 1.01707697e+00 3.92564833e-01
3.11481208e-01 1.09417856e+00 1.70513660e-01 -3.91870618e-01
-1.03989887e+00 5.66982627e-01 1.40785599e+00 8.74497354e-01
4.84836340e-01 -1.12301715e-01 -1.61471277e-01 1.19426537e+00
-1.71913520e-01 2.03637213e-01 6.88868523e-01 -1.00664616e+00
7.39220828e-02 9.92096245e-01 2.65310019e-01 -1.67936611e+00
-8.35524738e-01 -2.19020143e-01 -5.23283482e-01 6.21475168e-02
5.45982838e-01 7.66646340e-02 8.21688548e-02 1.30971372e+00
-2.10912898e-01 -7.59791613e-01 -6.16797924e-01 1.24264085e+00
1.25356507e+00 4.02990729e-01 -2.17362508e-01 -3.59065741e-01
1.46300972e+00 -9.97511923e-01 -8.53396356e-01 -1.27928019e-01
7.65402436e-01 -9.52004254e-01 1.55166113e+00 5.33091605e-01
-1.13059306e+00 -6.55556144e-03 -1.00667393e+00 -6.17787659e-01
-3.82293582e-01 -2.34634414e-01 3.23029965e-01 5.33480227e-01
-4.75897431e-01 5.65877199e-01 6.17781579e-02 -5.34316242e-01
1.88555092e-01 -3.21297735e-01 -1.58038020e-01 3.14155608e-01
-1.37444615e+00 1.30897021e+00 8.86014774e-02 -5.94194293e-01
-7.19813883e-01 -3.07477027e-01 -7.90250182e-01 -1.64178163e-01
8.93922672e-02 -4.09143358e-01 1.32335138e+00 -1.21212149e+00
-1.16907120e+00 1.26329041e+00 -9.56321359e-02 -3.89842302e-01
5.51288843e-01 -1.40659571e-01 -3.08869302e-01 -8.21615383e-02
1.27877548e-01 -2.85859611e-02 9.10829306e-01 -1.05993021e+00
7.18169808e-02 -3.39934081e-02 -1.73575699e-01 -4.96642515e-02
-1.03726447e+00 6.43760443e-01 2.48277739e-01 -4.25997794e-01
-3.72339070e-01 -7.65454292e-01 2.69609571e-01 -5.56335211e-01
-3.16786855e-01 -2.31079161e-01 7.49636948e-01 -1.14126265e+00
1.90051830e+00 -2.01129413e+00 3.07070822e-01 -1.15257271e-01
9.81883764e-01 2.68589482e-02 2.78848559e-01 1.04614079e+00
2.63894260e-01 -5.96448220e-02 6.46840921e-03 -2.80385852e-01
2.61461228e-01 -1.68926150e-01 -4.34229106e-01 4.62660372e-01
-6.24022603e-01 1.00041318e+00 -1.38231122e+00 -6.83259189e-01
1.69042230e-01 -1.11721745e-02 -6.13848150e-01 3.90125364e-01
-1.40527114e-01 -1.89616252e-02 8.03299248e-02 5.05213201e-01
2.43268266e-01 -5.43326616e-01 4.39679980e-01 2.60055631e-01
-2.05929026e-01 7.60990560e-01 -1.54153243e-01 8.43245268e-01
-4.90803748e-01 1.27641940e+00 -7.69374743e-02 1.12037398e-01
1.07962883e+00 1.95979655e-01 3.94296885e-01 -4.18926060e-01
5.11360824e-01 2.13131234e-01 -1.63458347e-01 -5.03609478e-01
1.04060566e+00 -5.85133791e-01 -5.70223987e-01 6.25814915e-01
-1.09941773e-02 -4.63817120e-01 1.68891683e-01 5.29701591e-01
1.25533414e+00 -5.05800128e-01 8.83271694e-01 -3.61289322e-01
6.31803811e-01 2.07587853e-02 1.06372587e-01 4.81213897e-01
-4.44709629e-01 5.26966870e-01 8.05586696e-01 -5.96006274e-01
-1.23469591e+00 -5.34899712e-01 -1.44290894e-01 1.41634715e+00
-3.00054938e-01 -1.17320251e+00 -6.85837865e-01 -3.00610930e-01
1.41137943e-01 9.13145006e-01 -5.25721550e-01 3.20590079e-01
-3.39024127e-01 -4.69781190e-01 6.19436860e-01 2.92360429e-02
4.03645873e-01 -1.39882565e+00 -6.91574395e-01 2.18212396e-01
-6.35178983e-01 -1.01065183e+00 -3.81095439e-01 1.85141468e-03
-3.45439494e-01 -1.09328103e+00 -2.40912452e-01 -7.36090839e-01
2.62608975e-01 3.48087043e-01 1.44901395e+00 4.40076083e-01
-2.44801342e-01 6.98705465e-02 -6.40197158e-01 -4.54584993e-02
-6.08549118e-01 -2.89584007e-02 5.92674389e-02 -4.96731281e-01
8.47265840e-01 -6.20959759e-01 -3.01488906e-01 5.16202033e-01
-5.71038067e-01 2.31898353e-01 -3.53383571e-01 9.61160719e-01
-7.42509305e-01 -1.88104689e-01 5.80857992e-01 -9.15814161e-01
1.22073531e+00 -9.06592786e-01 2.33020365e-01 -6.44194901e-01
-5.10092497e-01 -5.99896848e-01 1.17987788e+00 -1.01869546e-01
-7.45095074e-01 -3.38944465e-01 3.84521723e-01 8.81313533e-02
2.63392895e-01 5.99254847e-01 3.60457778e-01 9.28440690e-02
1.50818813e+00 -2.20210195e-01 -1.05209805e-01 -3.84190738e-01
5.84645867e-01 1.32244182e+00 6.92293227e-01 -3.61308604e-01
5.37312925e-01 3.10384572e-01 -2.19424769e-01 -1.14054441e+00
-1.25439191e+00 -1.09283793e+00 -5.87216496e-01 -7.72430599e-01
6.22101903e-01 -8.47793221e-01 -1.06759846e+00 4.96669203e-01
-1.36880767e+00 -1.70043662e-01 4.48111852e-04 -3.33999455e-01
-7.11644411e-01 8.08822334e-01 -1.31179583e+00 -1.06343544e+00
-5.38255513e-01 -1.81499571e-01 1.59030780e-01 -9.29236561e-02
-1.49403346e+00 -1.02290869e+00 2.63807356e-01 1.11101198e+00
7.29539618e-02 2.87403524e-01 8.38905811e-01 -5.54410517e-01
1.03053115e-01 -2.69364208e-01 -3.23450029e-01 2.10810333e-01
-9.10526887e-02 -3.65482569e-01 -8.60162497e-01 8.71697962e-02
2.51027763e-01 -1.11864388e+00 7.77678370e-01 -4.31154698e-01
6.56015396e-01 -9.34152484e-01 3.51503074e-01 -1.76503241e-01
1.02078211e+00 -5.25806904e-01 8.21811795e-01 6.71987414e-01
6.07245326e-01 8.00590575e-01 1.49248749e-01 1.29516184e+00
4.28243130e-01 4.54426557e-01 3.56677860e-01 3.28393847e-01
-7.33440146e-02 -7.30916083e-01 6.03487194e-01 1.18253076e+00
-9.68085453e-02 -8.27027112e-02 -1.10583138e+00 7.95142174e-01
-1.99591792e+00 -1.58430779e+00 -9.89911199e-01 1.76628232e+00
1.15625405e+00 -1.17473945e-01 6.79940045e-01 3.08530360e-01
3.78947198e-01 4.48323458e-01 1.17288105e-01 -7.27813542e-01
-4.01932508e-01 -1.67365700e-01 5.57632148e-02 8.66974354e-01
-7.47498453e-01 1.09693754e+00 6.43897915e+00 4.02758241e-01
-7.13974059e-01 3.90170574e-01 5.43024614e-02 -3.25317293e-01
-3.45799297e-01 -4.74989563e-02 -2.25976452e-01 5.88154376e-01
5.87677717e-01 -2.64645368e-01 8.35541844e-01 1.21920824e+00
5.80496132e-01 -4.79479954e-02 -9.35645223e-01 1.16924047e+00
7.50643730e-01 -9.72947001e-01 -3.45495313e-01 -1.99295193e-01
1.04883063e+00 -2.15517089e-01 -3.08260202e-01 3.53808194e-01
3.14380229e-01 -1.23714578e+00 7.44682491e-01 4.37143922e-01
2.99196035e-01 -4.50623840e-01 8.24622154e-01 5.94015956e-01
-3.93400580e-01 -2.77442038e-01 -2.97311097e-01 -1.07581079e+00
3.58483374e-01 6.52440906e-01 -1.26309896e+00 -5.06543398e-01
3.22901875e-01 9.47632909e-01 -8.85477424e-01 6.41016126e-01
-5.76050460e-01 7.61511028e-01 2.10052714e-01 -7.14425862e-01
-1.14273325e-01 -3.87879536e-02 8.39674115e-01 1.83328485e+00
-1.42648116e-01 -4.88389023e-02 1.27880394e-01 1.02507424e+00
-4.40796047e-01 5.85010767e-01 -5.13533056e-01 -3.15096945e-01
3.40203762e-01 1.65007186e+00 -2.05748305e-01 -2.15693131e-01
-2.30646372e-01 1.26582074e+00 5.61273873e-01 -2.57648647e-01
-5.61961055e-01 -1.77810043e-01 2.86144435e-01 7.12343037e-01
-4.90459681e-01 -4.52429771e-01 -9.71263766e-01 -1.21672273e+00
-4.14979190e-01 -1.16073966e+00 4.60994542e-01 -1.00823486e+00
-1.82164264e+00 2.06369728e-01 -4.30618376e-01 -8.53797734e-01
-2.29694933e-01 -7.14286983e-01 -7.03621149e-01 7.50989854e-01
-6.04200184e-01 -1.18484998e+00 -8.14854681e-01 3.53143007e-01
5.18177748e-01 -1.35931611e-01 6.18373156e-01 6.90189824e-02
-2.40654442e-02 1.50946856e-01 -2.31569692e-01 2.32708767e-01
1.06456733e+00 -1.42364919e+00 1.73588227e-02 1.85642660e-01
-3.87517154e-01 7.15721309e-01 1.49213219e+00 -7.25515366e-01
-9.03969347e-01 -4.17487085e-01 1.55844152e+00 -1.32790196e+00
1.43096912e+00 -3.55431706e-01 -8.70922863e-01 3.79764438e-01
6.68819249e-01 -9.41646993e-01 1.00604296e+00 5.61272621e-01
-8.32344294e-01 7.06113458e-01 -1.03587162e+00 7.06271946e-01
9.75761890e-01 -1.11152613e+00 -9.20094252e-01 5.50967276e-01
7.69958422e-02 -3.90893035e-02 -6.97941422e-01 -2.40457252e-01
5.69559634e-01 -1.17149663e+00 2.95893729e-01 -7.46752083e-01
1.54267132e+00 -2.84893394e-01 -9.70029831e-02 -1.16016138e+00
-8.59140158e-01 -1.01064610e+00 -2.81888902e-01 1.07793581e+00
1.35172501e-01 4.86362465e-02 7.11139798e-01 7.15565622e-01
2.24020593e-02 -2.16895536e-01 -3.83243889e-01 -4.46038038e-01
1.98440671e-01 -2.92691678e-01 -5.17936703e-03 1.43273628e+00
1.66981494e+00 8.78512859e-01 -1.24189067e+00 -9.31733966e-01
4.62890655e-01 4.34391201e-02 1.07423365e+00 -1.16722548e+00
-3.06430876e-01 -7.20129609e-01 -6.27762794e-01 -9.58940208e-01
3.88459980e-01 -1.22941422e+00 -5.75024523e-02 -1.18729532e+00
1.11914849e+00 5.69646299e-01 2.82315195e-01 5.11923552e-01
9.08842534e-02 5.92129588e-01 4.95808095e-01 5.02353549e-01
-7.60023654e-01 5.72483242e-01 1.23377264e+00 1.71707058e-03
-3.16757441e-01 -7.59321511e-01 -6.80740952e-01 9.71578598e-01
7.24034965e-01 -5.76535575e-02 7.05507845e-02 5.45698762e-01
1.00019026e+00 -4.04996760e-02 5.63846529e-01 -7.64620960e-01
2.27140058e-02 -3.72238308e-01 1.02958716e-01 -3.38964254e-01
2.38195747e-01 -3.54124099e-01 -1.26836166e-01 1.25326365e-01
-3.36605310e-01 -4.17247593e-01 -2.81966478e-01 4.04118672e-02
-6.74661323e-02 -5.69419801e-01 1.14313269e+00 -3.95092875e-01
-4.20259684e-01 -6.04209006e-01 -9.30304646e-01 1.76685289e-01
5.81669033e-01 1.50678977e-01 -7.65439451e-01 -1.13918483e+00
-5.19252300e-01 -5.59548754e-03 1.01746416e+00 5.34116089e-01
4.20330763e-01 -1.38116217e+00 -8.27414334e-01 -4.59442705e-01
3.44153613e-01 -1.04512215e+00 -4.03227210e-02 7.81495869e-01
-5.92947781e-01 1.15441203e-01 -4.35592115e-01 5.35576046e-02
-1.52241051e+00 7.50931621e-01 1.27767801e-01 -7.72730559e-02
-5.39400637e-01 2.28117391e-01 -2.41391271e-01 -4.89644706e-01
-3.30180585e-01 6.78155720e-01 -3.80592942e-01 4.91682976e-01
7.67596483e-01 9.84998167e-01 -3.34967405e-01 -1.01125717e+00
-8.60914122e-03 -4.29261923e-01 1.04496241e-01 -1.52784139e-01
1.01121652e+00 -1.86629027e-01 -9.21866059e-01 9.99602377e-01
9.91095066e-01 4.90328789e-01 -2.97080755e-01 9.78183001e-02
2.93368787e-01 -7.96294034e-01 -2.24009603e-01 -8.75716865e-01
1.39624579e-02 5.85851848e-01 -3.51855695e-01 6.95755959e-01
5.95763683e-01 6.35246933e-02 1.20001638e+00 3.67395431e-01
5.01849353e-01 -1.46220803e+00 7.28530407e-01 1.21750200e+00
1.33069372e+00 -9.97488797e-01 2.09318101e-01 -2.36747399e-01
-9.67789888e-01 1.22419178e+00 7.42277086e-01 4.32500755e-03
2.89603949e-01 -1.62288964e-01 1.04470357e-01 -5.22924244e-01
-7.62062073e-01 1.07628018e-01 4.69132781e-01 -5.15479371e-02
1.29878736e+00 4.20959890e-01 -1.13116169e+00 9.19186711e-01
-9.63960230e-01 -9.73738078e-03 1.20378828e+00 3.03882837e-01
-1.03516817e+00 -6.58932984e-01 -4.02502924e-01 4.66867745e-01
-3.25435072e-01 -1.76368028e-01 -1.60692775e+00 2.20370978e-01
-1.45009100e-01 1.15716267e+00 -6.34650290e-01 -8.99190009e-01
-1.21720679e-01 3.34868342e-01 2.22177446e-01 -6.57781422e-01
-1.03856587e+00 -5.43630779e-01 8.25220108e-01 -1.95814013e-01
-2.19913051e-02 -6.53118074e-01 -1.16886461e+00 -1.22520411e+00
-2.75543351e-02 1.24118462e-01 1.16276026e-01 8.25348675e-01
-2.27286927e-02 -2.72018820e-01 7.83717632e-01 -6.72702610e-01
-3.32724661e-01 -1.22513044e+00 -6.73788905e-01 1.28199220e+00
1.13289729e-01 -1.95621580e-01 -7.27992535e-01 1.52548477e-01] | [8.892504692077637, 11.04920482635498] |
320da317-f067-4c17-b47d-f711dc524733 | wedge-a-multi-weather-autonomous-driving | 2305.07528 | null | https://arxiv.org/abs/2305.07528v1 | https://arxiv.org/pdf/2305.07528v1.pdf | WEDGE: A multi-weather autonomous driving dataset built from generative vision-language models | The open road poses many challenges to autonomous perception, including poor visibility from extreme weather conditions. Models trained on good-weather datasets frequently fail at detection in these out-of-distribution settings. To aid adversarial robustness in perception, we introduce WEDGE (WEather images by DALL-E GEneration): a synthetic dataset generated with a vision-language generative model via prompting. WEDGE consists of 3360 images in 16 extreme weather conditions manually annotated with 16513 bounding boxes, supporting research in the tasks of weather classification and 2D object detection. We have analyzed WEDGE from research standpoints, verifying its effectiveness for extreme-weather autonomous perception. We establish baseline performance for classification and detection with 53.87% test accuracy and 45.41 mAP. Most importantly, WEDGE can be used to fine-tune state-of-the-art detectors, improving SOTA performance on real-world weather benchmarks (such as DAWN) by 4.48 AP for well-generated classes like trucks. WEDGE has been collected under OpenAI's terms of use and is released for public use under the CC BY-NC-SA 4.0 license. The repository for this work and dataset is available at https://infernolia.github.io/WEDGE. | ['Ketan Kotecha', 'Rahee Walambe', 'Deva Ramanan', 'Aboli Marathe'] | 2023-05-12 | null | null | null | null | ['2d-object-detection'] | ['computer-vision'] | [ 9.14025903e-02 2.12012902e-02 3.09447408e-01 -3.93110543e-01
-7.05261469e-01 -1.12942672e+00 8.05610299e-01 -2.06574097e-01
-4.48784739e-01 6.25398815e-01 7.96347558e-02 -5.37804246e-01
5.79288602e-01 -8.79551530e-01 -9.19914126e-01 -7.14753211e-01
-2.85629243e-01 2.92659283e-01 3.39305520e-01 -4.45973933e-01
-1.85557023e-01 4.74996924e-01 -1.53354764e+00 1.63738400e-01
8.46414387e-01 7.85326838e-01 7.03788176e-02 1.41795647e+00
5.32957315e-01 3.65066111e-01 -8.22396874e-01 -5.87440729e-01
9.27900314e-01 7.36650079e-02 -1.54360205e-01 -2.54140347e-01
1.27523291e+00 -6.01090550e-01 -6.05068386e-01 1.02178812e+00
8.27899456e-01 2.01533988e-01 8.10510576e-01 -1.48008287e+00
-8.91740441e-01 -1.43422373e-02 -3.58415186e-01 6.41983151e-01
3.00229013e-01 8.65606844e-01 8.66786718e-01 -9.01867211e-01
3.89966697e-01 1.20433760e+00 7.58878469e-01 3.59201193e-01
-1.17690277e+00 -7.98746824e-01 6.75861239e-02 4.99191172e-02
-1.25276995e+00 -5.52356660e-01 -1.25557194e-02 -6.41164303e-01
9.70684886e-01 3.65276754e-01 2.91783631e-01 1.65590489e+00
2.04359174e-01 3.90025467e-01 1.39271474e+00 1.01968355e-01
2.31213436e-01 -3.93726677e-02 -1.06178358e-01 5.50798059e-01
3.71637493e-01 7.19362497e-01 -3.52409303e-01 1.73020512e-02
2.10005105e-01 -5.93198538e-01 -3.47594082e-01 -2.79710181e-02
-1.09312797e+00 9.40615714e-01 8.32041442e-01 -5.56226730e-01
-2.31000349e-01 1.86117351e-01 3.35970521e-03 4.13500011e-01
5.71748912e-01 4.36142385e-01 -2.21498787e-01 6.42778501e-02
-4.56981510e-01 8.52487862e-01 7.80535936e-01 9.03428912e-01
5.88156343e-01 3.46797526e-01 3.84624153e-02 6.44314706e-01
6.68407604e-02 1.68163157e+00 -2.36163557e-01 -1.08962226e+00
6.34116530e-01 -1.32136092e-01 5.62546134e-01 -9.68744755e-01
-5.41339338e-01 -4.11317617e-01 -4.95324343e-01 6.58877611e-01
6.11720204e-01 -4.29921150e-01 -1.40085959e+00 1.59102583e+00
4.03468549e-01 2.60402560e-01 4.12818283e-01 1.21617222e+00
9.35145974e-01 1.10242748e+00 5.30486479e-02 4.86827672e-01
1.23544431e+00 -8.07819545e-01 -2.91828811e-01 -7.30457067e-01
2.00932413e-01 -9.70418632e-01 1.20045030e+00 2.78441489e-01
-3.55163336e-01 -5.16701996e-01 -9.45576668e-01 1.63684830e-01
-9.19062674e-01 -3.57530341e-02 4.86035764e-01 6.39166176e-01
-1.10029960e+00 -2.76150256e-01 -3.89708906e-01 -6.38814151e-01
4.84019041e-01 -3.94254684e-01 -2.89834946e-01 -2.18152955e-01
-1.48815024e+00 1.13274860e+00 2.53490001e-01 1.43047303e-01
-1.75026250e+00 -8.40315104e-01 -1.11437714e+00 -4.68611330e-01
2.77377903e-01 -6.09620929e-01 1.26159942e+00 -3.06485981e-01
-7.95037627e-01 8.10186505e-01 5.67570329e-02 -8.51300538e-01
7.79810429e-01 -6.35221899e-01 -7.20018506e-01 3.07150006e-01
4.49090928e-01 1.02982175e+00 1.07905316e+00 -1.52013624e+00
-7.19297647e-01 -1.75773472e-01 3.30490470e-01 3.10796738e-01
1.70586556e-01 -1.80063099e-01 5.10550067e-02 -5.09946227e-01
-4.17800307e-01 -1.20937169e+00 -1.78059429e-01 2.48063132e-01
-6.01984203e-01 2.80657113e-01 8.04038048e-01 -8.33475530e-01
4.73885059e-01 -1.98306465e+00 -5.35179973e-01 -6.54697716e-02
7.77901262e-02 3.38407040e-01 -1.66800976e-01 4.18135464e-01
1.83600098e-01 1.27517819e-01 -4.63413745e-01 -3.05626899e-01
2.07418874e-01 3.01374078e-01 -9.89843905e-01 7.65066087e-01
3.47208261e-01 8.39979649e-01 -7.83742666e-01 1.05857015e-01
4.64924634e-01 2.36712888e-01 -2.09339440e-01 3.73842925e-01
-3.35475177e-01 3.21469963e-01 -9.65556726e-02 8.94996524e-01
9.87777233e-01 4.91504014e-01 -6.10012889e-01 4.18055654e-02
-2.79855639e-01 -1.04361430e-01 -9.73141074e-01 1.24717486e+00
-2.84536898e-01 9.45684373e-01 1.32968858e-01 -2.33401746e-01
9.58462715e-01 -5.47518209e-02 -4.43120092e-01 -9.93033528e-01
-1.59030885e-01 4.72384356e-02 -1.55251563e-01 -6.25429273e-01
6.18064225e-01 1.19916141e-01 -2.80118704e-01 5.81171066e-02
-3.06275994e-01 -8.22997451e-01 2.16028422e-01 4.45841521e-01
9.38915014e-01 3.09142377e-02 -5.66171408e-02 -1.06348425e-01
-2.89613664e-01 4.22113061e-01 3.53902996e-01 1.31126392e+00
-5.45551836e-01 8.49862218e-01 1.11515798e-01 -5.90527058e-01
-1.45184338e+00 -1.56957698e+00 -5.26949167e-01 9.75117326e-01
1.23389028e-01 1.76659554e-01 -5.57712674e-01 -6.01254761e-01
4.47182119e-01 1.22367775e+00 -9.13262188e-01 1.82271246e-02
1.61778405e-02 -9.29073572e-01 1.29251707e+00 2.92742014e-01
4.63240474e-01 -8.63210380e-01 -8.67319703e-01 -1.79528013e-01
-2.71146119e-01 -1.44631839e+00 -6.27767667e-02 7.85253048e-02
8.31261799e-02 -9.60410833e-01 -7.44665563e-01 -1.39417022e-01
3.38063359e-01 7.13960230e-01 1.10678768e+00 -4.24196035e-01
-6.64446592e-01 5.48261702e-01 -5.29905021e-01 -9.34000731e-01
-3.81754518e-01 -2.14613467e-01 3.44070971e-01 -1.88952684e-01
1.53200373e-01 -3.76426995e-01 -5.97034812e-01 4.05599177e-01
-8.43926847e-01 -2.70911574e-01 2.79143929e-01 5.43855071e-01
2.15185791e-01 -3.49511027e-01 2.94341028e-01 -2.37425804e-01
2.76389778e-01 -8.32780778e-01 -9.31293130e-01 -1.57947510e-01
-3.83753538e-01 -4.10772949e-01 4.05496746e-01 -4.49623093e-02
-1.18392324e+00 -2.93065142e-02 2.77238479e-03 -3.89227539e-01
-8.50054383e-01 -4.35444415e-02 1.32656455e-01 -1.41501456e-01
1.39098537e+00 2.95753121e-01 -2.53869534e-01 -3.20256539e-02
7.56991506e-01 7.48071253e-01 1.04656267e+00 -3.61804783e-01
1.50567174e+00 8.73684585e-01 -3.04896653e-01 -1.03523815e+00
-1.09405398e+00 -3.51030231e-01 -2.30594985e-02 -4.34867203e-01
1.12060606e+00 -1.41018867e+00 -2.82931507e-01 7.65474856e-01
-7.20662117e-01 -8.83633316e-01 4.02077958e-02 2.72141397e-01
-3.44715416e-01 1.88689768e-01 -3.35354507e-02 -9.30343628e-01
-1.43365815e-01 -7.53028631e-01 1.13425732e+00 4.87555325e-01
-1.19699240e-01 -6.98003888e-01 4.34266031e-01 4.04349327e-01
4.80839163e-01 6.48041725e-01 1.93741560e-01 -2.35102713e-01
-4.01284784e-01 -2.60091692e-01 -3.02179635e-01 2.82953799e-01
-3.29160154e-01 2.11150557e-01 -1.54000831e+00 -3.77770692e-01
-5.30300140e-01 -8.51269484e-01 1.29615247e+00 2.48634458e-01
4.62244421e-01 -1.78779021e-01 -8.43088031e-02 8.35982442e-01
1.23808098e+00 -2.34371442e-02 6.35384798e-01 5.67321897e-01
6.37186766e-01 5.91629148e-01 8.54073763e-01 4.12924826e-01
3.87161374e-01 4.75028038e-01 1.09282291e+00 -2.79502749e-01
-1.40721887e-01 -1.94628999e-01 3.58128250e-01 -5.34665823e-01
1.58429474e-01 -7.89056480e-01 -1.23730326e+00 8.34859192e-01
-1.51250970e+00 -1.06511247e+00 -4.32297438e-01 2.06468463e+00
5.16095221e-01 4.43536341e-01 -1.10736355e-01 -3.22622508e-01
4.79334027e-01 5.95499992e-01 -5.30652225e-01 -4.78041649e-01
-5.47968745e-01 -2.76480854e-01 1.07812655e+00 7.66641498e-01
-1.46205294e+00 1.01212382e+00 5.72910786e+00 3.75802577e-01
-9.75432634e-01 6.78499416e-02 5.33097863e-01 -1.01161785e-01
-2.79207408e-01 -1.86229929e-01 -9.86776292e-01 4.33330357e-01
1.06281257e+00 1.12091094e-01 2.96992391e-01 1.01898062e+00
4.12942290e-01 -4.15500760e-01 -3.99834573e-01 5.79675317e-01
8.60155746e-02 -1.15611172e+00 -1.82354614e-01 6.33417070e-02
9.27502871e-01 6.77539825e-01 4.73699927e-01 4.31930661e-01
6.96379483e-01 -9.96552229e-01 8.10574889e-01 4.49021935e-01
9.13806200e-01 -5.91846228e-01 6.04279518e-01 2.04186827e-01
-8.61239195e-01 -1.25686880e-02 -6.39815331e-01 -1.90150887e-01
1.47954106e-01 6.90651298e-01 -1.38193941e+00 4.02888060e-01
1.18898547e+00 4.29434448e-01 -6.41892433e-01 1.10898733e+00
-5.01817763e-01 1.02798617e+00 -7.79501259e-01 2.82365859e-01
6.46272719e-01 1.87427863e-01 1.01045966e+00 1.54815793e+00
9.88522470e-02 7.47659802e-02 4.44854259e-01 7.27264702e-01
2.14951873e-01 -5.48316121e-01 -1.30253923e+00 3.25849116e-01
5.00324607e-01 1.23789060e+00 -2.86355615e-01 -2.62504339e-01
-1.53739616e-01 7.64448524e-01 -9.21063200e-02 5.48559606e-01
-1.30293083e+00 -3.76581401e-01 1.15990567e+00 -6.31102324e-02
3.82177234e-01 -4.50371832e-01 -2.91239202e-01 -8.25230956e-01
-1.39701376e-02 -6.71875358e-01 1.20699458e-01 -1.23919582e+00
-1.15075147e+00 7.98553705e-01 2.76060730e-01 -1.33831882e+00
-3.74609381e-02 -5.62427580e-01 -6.07495964e-01 1.05319715e+00
-1.73335421e+00 -1.24833083e+00 -8.89554083e-01 2.94359982e-01
4.76406246e-01 -1.66505128e-01 9.13948059e-01 -5.89222237e-02
-3.44064772e-01 4.89037395e-01 1.10240690e-01 6.68305904e-02
1.00902140e+00 -1.56780648e+00 1.19055915e+00 1.43287992e+00
1.24102399e-01 1.00774735e-01 1.19386482e+00 -6.48036599e-01
-1.07762313e+00 -1.57962990e+00 5.94596624e-01 -8.19691956e-01
8.54826450e-01 -5.55876374e-01 -7.48200774e-01 6.64387107e-01
2.01513395e-01 1.42418161e-01 2.42799804e-01 -3.63467485e-01
-9.23558414e-01 -1.70805737e-01 -1.15870583e+00 8.34680736e-01
8.41165006e-01 -4.56039637e-01 -6.49822593e-01 6.60314977e-01
8.49610448e-01 -7.13500023e-01 -5.02600670e-01 4.46444929e-01
4.15402740e-01 -1.14435339e+00 9.70250249e-01 -4.08750296e-01
3.73753190e-01 -7.33357131e-01 -5.68749189e-01 -1.70270014e+00
-3.41490567e-01 -5.31366646e-01 1.54391170e-01 8.99101496e-01
7.31358647e-01 -1.00785470e+00 3.18600953e-01 1.04277544e-01
-5.14371037e-01 -2.34576687e-01 -9.54737961e-01 -9.03970480e-01
-5.40171638e-02 -6.95422530e-01 1.74388617e-01 7.81249523e-01
-8.79121065e-01 4.27135862e-02 -4.51069146e-01 1.11322927e+00
8.76024008e-01 1.44494697e-03 1.26002514e+00 -6.26672983e-01
-1.67737886e-01 -1.85600013e-01 -3.83433282e-01 -7.06346631e-01
-1.41375497e-01 -3.14211369e-01 4.57598507e-01 -1.44080603e+00
-3.74874979e-01 -5.00314534e-01 2.05471158e-01 6.84805334e-01
-2.70531535e-01 7.67825305e-01 8.51344168e-02 1.78093284e-01
-2.55608201e-01 5.76194882e-01 9.21228290e-01 -3.52399796e-01
3.59188080e-01 -8.01759288e-02 -4.85102236e-01 4.43706512e-01
1.15823746e+00 -5.16125202e-01 -3.52610677e-01 -6.27918482e-01
2.28407737e-02 -4.91695881e-01 8.59656930e-01 -1.35973096e+00
-3.44504178e-01 -3.55780542e-01 4.94915813e-01 -4.54747736e-01
6.95602238e-01 -4.18656349e-01 -1.49722502e-01 4.30924267e-01
-1.26539275e-01 2.16703676e-02 5.34645975e-01 6.06754839e-01
1.27677053e-01 1.42627984e-01 8.45364332e-01 -1.08470097e-01
-1.25020814e+00 1.69476688e-01 -4.10175562e-01 5.51359415e-01
1.20857453e+00 2.06853766e-02 -1.24155009e+00 -6.52170539e-01
-5.92428207e-01 5.24739146e-01 5.50801694e-01 6.86343610e-01
4.75787520e-01 -7.31688023e-01 -1.26906919e+00 1.95562199e-01
6.14370108e-01 -2.67638154e-02 2.78120875e-01 4.01757985e-01
-8.55722785e-01 1.58234492e-01 -1.24544233e-01 -6.92688704e-01
-8.99879754e-01 2.63966382e-01 4.84945297e-01 4.43204790e-01
-6.61091208e-01 1.03704345e+00 5.44436634e-01 -5.92680693e-01
8.59610960e-02 -2.74872541e-01 2.35890716e-01 -2.09323652e-02
8.80699337e-01 3.15046251e-01 -8.79011229e-02 -6.43402398e-01
-3.04990619e-01 1.11408964e-01 1.68480948e-01 -2.52699345e-01
8.39788616e-01 -1.32270902e-01 5.52711964e-01 3.86605799e-01
7.94063091e-01 1.16394117e-01 -1.81406069e+00 1.57728299e-01
-7.60186791e-01 -6.16792440e-01 1.02981217e-02 -1.27455211e+00
-6.56659782e-01 8.40693831e-01 9.48053956e-01 3.97514284e-01
6.16569042e-01 -8.77531245e-02 7.62707829e-01 5.52152872e-01
1.05180852e-01 -6.78886414e-01 -1.57708198e-01 8.21141422e-01
1.15721822e+00 -1.46684217e+00 -1.81725860e-01 -1.58085078e-01
-1.01362300e+00 7.19011545e-01 7.64242530e-01 -2.68039286e-01
2.98424006e-01 4.82028306e-01 7.82624841e-01 -4.92483266e-02
-6.35944843e-01 -4.59798902e-01 3.38981897e-02 1.21656418e+00
-4.30274844e-01 5.82900643e-01 5.53139687e-01 -1.49341270e-01
-5.87703049e-01 -5.69255948e-01 8.26273978e-01 7.79408991e-01
-6.61887884e-01 -3.03230733e-02 -7.63505936e-01 1.28404289e-01
-1.84939533e-01 -4.22539949e-01 -4.95353878e-01 6.87230945e-01
1.50967062e-01 1.15590096e+00 3.99095081e-02 -4.44722295e-01
4.40257907e-01 -1.61631361e-01 3.38182822e-02 -4.32561696e-01
-2.20540330e-01 -5.07696152e-01 5.44998825e-01 -6.09226763e-01
4.75354381e-02 -6.68009162e-01 -8.08753371e-01 -1.92369103e-01
1.09862037e-01 -1.51176050e-01 8.30124855e-01 5.81171989e-01
4.43734378e-01 2.91695535e-01 5.40991545e-01 -1.12412989e+00
-7.06027210e-01 -9.55671072e-01 -2.76728094e-01 2.09023893e-01
7.38993466e-01 -7.34404504e-01 -7.46753335e-01 1.07710764e-01] | [8.182232856750488, -1.3571802377700806] |
2d72f77d-74ac-4b80-871c-dce7e720b1c1 | creating-synthetic-datasets-for-collaborative | 2303.01297 | null | https://arxiv.org/abs/2303.01297v1 | https://arxiv.org/pdf/2303.01297v1.pdf | Creating Synthetic Datasets for Collaborative Filtering Recommender Systems using Generative Adversarial Networks | Research and education in machine learning needs diverse, representative, and open datasets that contain sufficient samples to handle the necessary training, validation, and testing tasks. Currently, the Recommender Systems area includes a large number of subfields in which accuracy and beyond accuracy quality measures are continuously improved. To feed this research variety, it is necessary and convenient to reinforce the existing datasets with synthetic ones. This paper proposes a Generative Adversarial Network (GAN)-based method to generate collaborative filtering datasets in a parameterized way, by selecting their preferred number of users, items, samples, and stochastic variability. This parameterization cannot be made using regular GANs. Our GAN model is fed with dense, short, and continuous embedding representations of items and users, instead of sparse, large, and discrete vectors, to make an accurate and quick learning, compared to the traditional approach based on large and sparse input vectors. The proposed architecture includes a DeepMF model to extract the dense user and item embeddings, as well as a clustering process to convert from the dense GAN generated samples to the discrete and sparse ones, necessary to create each required synthetic dataset. The results of three different source datasets show adequate distributions and expected quality values and evolutions on the generated datasets compared to the source ones. Synthetic datasets and source codes are available to researchers. | ['Luis Martínez', 'Raciel Yera', 'Abraham Gutiérrez', 'Jesús Bobadilla'] | 2023-03-02 | null | null | null | null | ['collaborative-filtering'] | ['miscellaneous'] | [-1.09674916e-01 -3.02569687e-01 -3.65103297e-02 -5.82919180e-01
-3.37997615e-01 -4.56061125e-01 5.69702089e-01 -2.03978512e-02
-2.17043400e-01 8.86423528e-01 3.23298305e-01 1.24342412e-01
-3.05334538e-01 -1.28248894e+00 -5.77985704e-01 -7.66406298e-01
2.50504196e-01 7.40191340e-01 -2.02399120e-01 -1.51957810e-01
1.11446306e-01 1.43548146e-01 -1.81199634e+00 1.77392736e-01
1.32726264e+00 8.62814963e-01 1.01117685e-01 3.90905440e-01
-5.25414944e-01 2.52052933e-01 -7.87263274e-01 -6.22927308e-01
4.05789554e-01 -7.20534861e-01 -9.13422108e-02 -4.08242382e-02
1.10527612e-01 -4.04342294e-01 6.23898357e-02 8.69767487e-01
6.28892899e-01 3.10687661e-01 9.09977257e-01 -1.33468771e+00
-1.45998240e+00 8.34700823e-01 -2.98861951e-01 -1.86383635e-01
1.14734635e-01 1.42804116e-01 7.38604724e-01 -8.13990295e-01
3.30860257e-01 9.94148552e-01 6.13729835e-01 5.81952870e-01
-1.00770652e+00 -9.79784548e-01 1.34563819e-02 7.63958469e-02
-1.26353121e+00 -4.59703468e-02 8.00549150e-01 -4.18748409e-01
1.21991463e-01 2.69345760e-01 1.02857518e+00 1.58527923e+00
-6.76737353e-02 4.31818813e-01 8.87771666e-01 -2.50194699e-01
5.14878452e-01 8.22151780e-01 6.42258748e-02 1.12767436e-01
5.80581129e-01 1.35336399e-01 -1.88887212e-02 -1.05319895e-01
7.36037910e-01 7.22853899e-01 -1.84219748e-01 -3.40446025e-01
-9.05682504e-01 1.11108696e+00 3.42688411e-01 3.86601627e-01
-5.87286472e-01 -2.15560287e-01 3.92684825e-02 3.43479425e-01
3.89688343e-01 5.68283856e-01 -3.83314937e-01 -1.57301977e-01
-9.03025389e-01 1.74541578e-01 6.41552925e-01 1.07172668e+00
6.12689435e-01 3.34457099e-01 -3.40262473e-01 9.22940314e-01
3.31208795e-01 6.26472473e-01 1.01648164e+00 -3.99412543e-01
1.88768521e-01 8.38197351e-01 2.49232382e-01 -1.22847331e+00
3.94172147e-02 -8.60575616e-01 -1.21660411e+00 -3.69825587e-02
1.85113564e-01 -3.63115698e-01 -7.76214361e-01 1.66706824e+00
4.20973063e-01 3.59690577e-01 1.44119725e-01 8.33403170e-01
8.45587134e-01 9.85950351e-01 -1.18889198e-01 -1.04425594e-01
8.95022333e-01 -9.93301332e-01 -7.72971928e-01 1.80075392e-01
2.19895855e-01 -7.71781504e-01 1.45225811e+00 4.31764841e-01
-8.11723113e-01 -8.61107111e-01 -9.87491548e-01 1.61041886e-01
-5.90363443e-01 3.61690670e-01 7.29827404e-01 8.82672906e-01
-7.22721696e-01 5.43025136e-01 -2.93000847e-01 -8.41527060e-02
4.57579762e-01 1.02924526e-01 -1.54679954e-01 -1.72558829e-01
-1.30416965e+00 4.55374360e-01 2.55711347e-01 4.80382517e-02
-7.93880284e-01 -8.89741540e-01 -6.25525892e-01 2.07101673e-01
-2.30452463e-01 -7.03714073e-01 6.82791829e-01 -1.34088659e+00
-1.58977342e+00 1.55397013e-01 4.79077697e-01 -3.94088984e-01
3.97545636e-01 -1.66886136e-01 -5.37854373e-01 -4.07276362e-01
-3.62613559e-01 3.26496631e-01 9.43885982e-01 -1.14832652e+00
-3.42954308e-01 -1.84941635e-01 -5.74913658e-02 1.05556078e-01
-7.86129653e-01 -2.90899307e-01 -2.72082448e-01 -8.87495220e-01
-2.05289677e-01 -6.63652062e-01 -2.56862074e-01 -2.48905346e-01
-8.25668871e-02 -4.40836065e-02 5.57687938e-01 -5.51497638e-01
1.16184282e+00 -2.18805552e+00 1.24465272e-01 3.74811530e-01
-9.30472091e-02 3.28431129e-01 -1.97242096e-01 4.06249493e-01
6.93280846e-02 1.31257564e-01 3.73824127e-02 -2.22250581e-01
1.23778626e-01 1.90218464e-01 -1.73922017e-01 1.71436295e-01
-5.60681037e-05 5.75196147e-01 -9.20426905e-01 -5.52619398e-02
2.57904053e-01 6.68241143e-01 -8.21246862e-01 5.83076119e-01
-2.33457327e-01 4.58110929e-01 -4.77103740e-01 3.77811462e-01
7.04376459e-01 -1.59657925e-01 -1.88042969e-02 -2.36500099e-01
2.80753970e-01 -1.64487615e-01 -1.59398854e+00 1.42741835e+00
-7.53966689e-01 1.41494945e-01 -1.90114200e-01 -8.36305261e-01
1.40697050e+00 2.37395734e-01 3.75096470e-01 -3.19632709e-01
4.19329494e-01 1.48044631e-01 1.41015891e-02 -2.92720318e-01
4.80463177e-01 9.29346085e-02 -6.38617501e-02 6.11115217e-01
1.85796857e-01 4.27495018e-02 1.32683679e-01 1.74736694e-01
6.24602079e-01 -1.55030014e-02 -2.58215666e-01 5.06877303e-02
3.66865307e-01 -2.45665193e-01 5.64728141e-01 4.07636434e-01
4.97805536e-01 6.78250670e-01 1.23917036e-01 -4.56411064e-01
-1.19901752e+00 -9.10949588e-01 -1.71990901e-01 7.53740668e-01
4.10203002e-02 -3.68286908e-01 -6.87856376e-01 -6.19680583e-01
-2.29769666e-02 7.49069631e-01 -7.09523380e-01 -4.41806942e-01
-1.12237679e-02 -7.97675014e-01 9.83632579e-02 4.23537493e-01
3.91771197e-01 -1.13576162e+00 -3.48193012e-02 1.42802566e-01
2.14958161e-01 -4.71439809e-01 -5.15694559e-01 -1.57903895e-01
-6.66488528e-01 -1.04870641e+00 -7.68664598e-01 -7.19269395e-01
8.48935425e-01 -3.91424727e-03 9.23361421e-01 -5.72753921e-02
4.98538464e-02 -4.78313714e-02 -8.37520421e-01 -4.41849291e-01
-3.48661304e-01 1.14189602e-01 2.30396420e-01 5.02663791e-01
2.58110702e-01 -8.03516626e-01 -6.43404603e-01 2.81487405e-01
-1.06245518e+00 7.45029598e-02 7.44846225e-01 1.02516007e+00
6.78648353e-01 7.82950968e-02 8.81325841e-01 -1.19576406e+00
8.66160631e-01 -9.28121030e-01 -4.43050623e-01 2.44765148e-01
-9.81890678e-01 -1.18510202e-01 1.33931589e+00 -7.12390125e-01
-8.87714684e-01 -4.16823566e-01 -2.70252019e-01 -5.91689289e-01
-1.83049485e-01 2.83554852e-01 -4.64866996e-01 2.79998302e-01
7.29981184e-01 3.04065704e-01 -4.94950637e-02 -6.91737056e-01
7.92922020e-01 1.07533026e+00 1.22142501e-01 -4.34028894e-01
8.90807748e-01 -1.08921401e-01 -5.80746114e-01 -3.77521098e-01
-6.85355723e-01 3.10825780e-02 -3.50570530e-01 1.01016782e-01
5.15538812e-01 -8.16325605e-01 -4.35577109e-02 4.58212137e-01
-7.04801619e-01 -1.18113302e-01 -8.62725139e-01 7.37008095e-01
-1.36041254e-01 -1.69982508e-01 -5.18344462e-01 -5.73426723e-01
-5.83203256e-01 -1.06790471e+00 5.20745337e-01 6.94569647e-01
1.58625357e-02 -8.80255640e-01 6.28663227e-02 2.72386938e-01
8.78712833e-01 2.77280092e-01 7.24164963e-01 -9.73384559e-01
-4.04651731e-01 -4.31273758e-01 1.37018368e-01 7.33616948e-01
4.51708615e-01 3.60672176e-01 -4.49230969e-01 -3.94415259e-01
-5.33967279e-02 -2.93153793e-01 4.15690213e-01 1.54430106e-01
1.36502838e+00 -6.75752819e-01 9.05078370e-03 9.17388439e-01
1.34365153e+00 4.32167143e-01 5.29078782e-01 2.57884283e-02
5.83938062e-01 1.58896700e-01 5.13368666e-01 5.15906811e-01
1.85295671e-01 2.57343888e-01 2.50262558e-01 -4.69736196e-02
2.45329719e-02 -6.06480718e-01 1.30914703e-01 1.06453001e+00
7.63402507e-02 -2.98458636e-01 -3.37765634e-01 4.43892419e-01
-1.46712732e+00 -1.12180841e+00 1.62026361e-01 2.24873900e+00
9.90597665e-01 -4.30691754e-04 1.07910432e-01 2.53334463e-01
6.65926993e-01 -1.40798554e-01 -6.88033521e-01 -3.59921813e-01
-3.46497111e-02 5.03757000e-01 1.24613822e-01 9.98598710e-02
-7.41113484e-01 5.73913932e-01 5.89516830e+00 8.86208057e-01
-1.23582089e+00 1.64068401e-01 7.37226725e-01 -2.77109444e-01
-1.00154078e+00 -2.87526906e-01 -7.31241405e-01 1.06027293e+00
1.14067554e+00 -2.50233918e-01 5.23622930e-01 9.48610425e-01
2.27001429e-01 4.12759483e-01 -8.02937925e-01 1.07510126e+00
1.14338249e-01 -1.37367225e+00 4.31004077e-01 -1.93048343e-01
1.22834265e+00 -4.31065470e-01 2.10500598e-01 4.94625062e-01
5.18532634e-01 -1.14500642e+00 3.41395110e-01 8.44419718e-01
6.94556296e-01 -9.38161671e-01 1.05494404e+00 4.01753932e-01
-7.35459626e-01 -2.32971624e-01 -6.02212429e-01 2.55832504e-02
-9.66154784e-02 8.33044052e-01 -3.83682609e-01 6.43224835e-01
6.38435066e-01 7.54532695e-01 -5.32584846e-01 1.04165578e+00
-1.61096424e-01 7.86650240e-01 -1.73210442e-01 -4.23812211e-01
-1.42020300e-01 -9.08852041e-01 -3.37880701e-02 8.13258708e-01
6.77914202e-01 -1.05898425e-01 -7.11802766e-02 9.55810487e-01
-2.88949430e-01 4.93497193e-01 -5.14085114e-01 -1.17374890e-01
8.28695238e-01 1.41206777e+00 -2.81616032e-01 -3.54424715e-01
-2.31505871e-01 7.42868364e-01 1.95138142e-01 2.77517080e-01
-1.04489124e+00 -6.15550637e-01 5.90730309e-01 1.42609283e-01
2.92100579e-01 -2.68293936e-02 -1.06304206e-01 -1.18718851e+00
-7.57544786e-02 -1.14928341e+00 2.68495113e-01 -5.80278635e-01
-1.71383309e+00 7.34155238e-01 -2.34235629e-01 -1.42361605e+00
-2.41196722e-01 -3.46707329e-02 -7.44697332e-01 1.18259108e+00
-1.07764161e+00 -9.80179071e-01 -5.85347831e-01 6.34384215e-01
4.34263349e-01 -5.96924245e-01 9.76897478e-01 6.23267531e-01
-7.74571896e-01 8.93206477e-01 6.46788359e-01 9.35280472e-02
6.45029426e-01 -9.86852050e-01 8.81840363e-02 6.68057382e-01
3.32330763e-01 8.25166881e-01 5.24556875e-01 -2.84580469e-01
-1.37155449e+00 -1.38029087e+00 3.54089975e-01 -2.57651299e-01
1.67103782e-01 -3.45974803e-01 -7.81072736e-01 4.99153465e-01
2.57982731e-01 -1.28771849e-02 1.04978812e+00 1.45036861e-01
-3.00993957e-02 -5.38824320e-01 -1.52010155e+00 3.53648514e-01
7.14788616e-01 8.69699121e-02 -4.97903764e-01 2.49844864e-01
7.69944608e-01 -2.61313617e-01 -9.38625634e-01 1.58330500e-01
5.17418027e-01 -9.82055187e-01 7.22701848e-01 -6.43514931e-01
5.35329878e-01 -3.27316284e-01 -1.30295530e-01 -1.73510766e+00
-3.85993749e-01 -2.00553626e-01 -1.21758342e-01 1.47993684e+00
5.08459449e-01 -5.99961519e-01 8.35703075e-01 4.38755035e-01
-2.66464502e-02 -1.07020736e+00 -3.10056418e-01 -3.68616372e-01
-3.37098353e-02 -1.00119203e-01 1.20545614e+00 1.02695668e+00
-5.45990884e-01 3.58616084e-01 -5.79399049e-01 -7.86655471e-02
4.56986994e-01 4.67918575e-01 1.07443571e+00 -1.22722173e+00
-3.95852864e-01 -1.58563137e-01 -2.56811798e-01 -8.19669902e-01
-1.88546866e-01 -8.88797760e-01 -5.11718988e-01 -1.69080412e+00
-1.08324513e-01 -9.29208338e-01 -3.81853819e-01 1.80236608e-01
-4.54869270e-02 1.98277965e-01 3.14562917e-02 5.03063202e-02
-1.53710082e-01 1.01826692e+00 1.28821194e+00 -2.06438199e-01
-3.96216393e-01 2.90359497e-01 -1.05133474e+00 3.91996652e-01
8.50060940e-01 -2.92638391e-01 -8.85161042e-01 -3.62166643e-01
1.54491082e-01 -1.89755782e-01 -1.30420208e-01 -1.10756314e+00
-1.15859367e-01 -3.31023544e-01 8.49741399e-01 -3.59833121e-01
2.36001343e-01 -1.01223814e+00 6.76447630e-01 2.37828374e-01
-4.30583149e-01 2.60960292e-02 -1.83824301e-01 4.55815822e-01
-2.63766080e-01 -2.71609277e-01 7.04935610e-01 -4.46964987e-02
-3.21848303e-01 7.06353307e-01 1.48677394e-01 3.95151749e-02
1.31282246e+00 -1.64104208e-01 -3.64519656e-02 -4.76372570e-01
-6.79071426e-01 1.39240310e-01 4.19745654e-01 7.10881948e-01
7.48706698e-01 -1.62648821e+00 -6.94413602e-01 6.50559127e-01
-1.40663236e-01 6.76734149e-02 4.82103497e-01 2.50192821e-01
-4.16417718e-01 -1.76054269e-01 -5.45081675e-01 -1.92134380e-02
-6.71392322e-01 6.26117826e-01 1.98795453e-01 -9.82783064e-02
-3.14353496e-01 6.53924584e-01 -4.14181709e-01 -6.54884636e-01
2.80616224e-01 -3.04602176e-01 -6.04264975e-01 1.75350294e-01
5.67453384e-01 2.23391414e-01 -5.15798368e-02 -2.27452964e-01
1.82167754e-01 3.22766483e-01 1.18524581e-01 2.85237938e-01
1.45906293e+00 1.67563811e-01 1.82797045e-01 3.77050221e-01
1.03768909e+00 3.24598670e-01 -9.14804041e-01 -1.40821248e-01
-7.03041077e-01 -6.51396751e-01 -2.65095741e-01 -7.74884582e-01
-1.43936169e+00 7.48059392e-01 8.28765988e-01 3.48875016e-01
1.09447098e+00 -4.37886566e-01 9.18695211e-01 -1.11026250e-01
3.46193701e-01 -7.79218674e-01 -4.85640913e-02 1.67862341e-01
8.82800221e-01 -1.01237750e+00 -1.96761891e-01 1.02670632e-01
-5.77879071e-01 8.75837266e-01 8.97191405e-01 -3.67348492e-01
8.15436840e-01 3.02591175e-02 3.06588858e-02 2.59231627e-01
-6.59332156e-01 1.93380386e-01 2.36393392e-01 7.24183679e-01
4.27088648e-01 1.47326529e-01 -5.10431230e-01 1.38089931e+00
-5.14264584e-01 1.87050421e-02 4.57487136e-01 3.44692200e-01
-2.76588649e-01 -1.51038599e+00 -1.40147030e-01 9.93950248e-01
-2.18091041e-01 -2.60702446e-02 -2.85720602e-02 5.13943911e-01
6.32212341e-01 8.20912004e-01 1.95162147e-01 -7.75387168e-01
3.68475974e-01 -1.01864219e-01 1.38727367e-01 -6.91381693e-01
-6.34856105e-01 -5.24583042e-01 -3.47933292e-01 -1.11951731e-01
-1.81565642e-01 -3.61587495e-01 -9.95262742e-01 -4.06847715e-01
-3.99870366e-01 7.53646135e-01 6.49488330e-01 5.52940369e-01
5.38218737e-01 6.96259558e-01 1.17126477e+00 -6.64820850e-01
-8.01092803e-01 -1.19148648e+00 -7.12074280e-01 8.00821722e-01
-1.99824244e-01 -6.31167173e-01 -5.03166735e-01 -1.71806157e-01] | [10.204936981201172, 5.518398761749268] |
a5fa9ed8-7e0e-4bbc-add5-8fad1a013ff3 | deformable-neural-radiance-fields | 2011.12948 | null | https://arxiv.org/abs/2011.12948v5 | https://arxiv.org/pdf/2011.12948v5.pdf | Nerfies: Deformable Neural Radiance Fields | We present the first method capable of photorealistically reconstructing deformable scenes using photos/videos captured casually from mobile phones. Our approach augments neural radiance fields (NeRF) by optimizing an additional continuous volumetric deformation field that warps each observed point into a canonical 5D NeRF. We observe that these NeRF-like deformation fields are prone to local minima, and propose a coarse-to-fine optimization method for coordinate-based models that allows for more robust optimization. By adapting principles from geometry processing and physical simulation to NeRF-like models, we propose an elastic regularization of the deformation field that further improves robustness. We show that our method can turn casually captured selfie photos/videos into deformable NeRF models that allow for photorealistic renderings of the subject from arbitrary viewpoints, which we dub "nerfies." We evaluate our method by collecting time-synchronized data using a rig with two mobile phones, yielding train/validation images of the same pose at different viewpoints. We show that our method faithfully reconstructs non-rigidly deforming scenes and reproduces unseen views with high fidelity. | ['Ricardo Martin-Brualla', 'Steven M. Seitz', 'Dan B Goldman', 'Sofien Bouaziz', 'Jonathan T. Barron', 'Utkarsh Sinha', 'Keunhong Park'] | 2020-11-25 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Park_Nerfies_Deformable_Neural_Radiance_Fields_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Park_Nerfies_Deformable_Neural_Radiance_Fields_ICCV_2021_paper.pdf | iccv-2021-1 | ['3d-human-reconstruction'] | ['computer-vision'] | [ 3.64762127e-01 1.64120674e-01 4.98841256e-01 -2.54208237e-01
-7.85751164e-01 -9.38879490e-01 5.91851294e-01 -7.23020613e-01
-1.00783393e-01 5.65744340e-01 1.69684738e-01 4.46673892e-02
-5.87535203e-02 -7.20788300e-01 -1.24580741e+00 -5.28162718e-01
2.61225581e-01 4.46144104e-01 9.80201811e-02 -2.45899737e-01
1.46734998e-01 1.07570744e+00 -1.48685694e+00 1.48601502e-01
4.92001265e-01 5.65538466e-01 2.49552932e-02 1.04568112e+00
7.05575228e-01 4.61652786e-01 -3.10506314e-01 -6.18169904e-01
7.77310550e-01 -1.47161126e-01 -8.09780300e-01 4.54475999e-01
1.21034610e+00 -6.65592611e-01 -5.26028991e-01 6.06305480e-01
2.97490150e-01 3.96246344e-01 4.19518292e-01 -8.15227866e-01
-8.52914751e-01 -4.91455048e-01 -4.49719459e-01 -2.17950523e-01
9.07576919e-01 -5.28110936e-03 5.58064103e-01 -8.63456786e-01
1.28104162e+00 1.30640543e+00 1.01779652e+00 8.37831676e-01
-1.62686169e+00 5.75925484e-02 -1.13460131e-01 -3.88908386e-01
-1.50218356e+00 -5.04659593e-01 9.30055022e-01 -5.72526574e-01
8.04994106e-01 8.37145030e-01 8.43477666e-01 1.01944280e+00
4.80275422e-01 1.94563240e-01 1.20356917e+00 -2.92448491e-01
1.19234867e-01 3.79939340e-02 -3.75646353e-01 8.62770736e-01
-1.18374571e-01 1.91402465e-01 -3.75413448e-01 -2.62063682e-01
1.48782873e+00 2.46572360e-01 -7.13513136e-01 -7.03024685e-01
-1.28353381e+00 3.46799463e-01 5.10437191e-01 -3.74652557e-02
-2.82912493e-01 1.74807101e-01 -4.35908347e-01 7.45450482e-02
5.06133497e-01 4.19780523e-01 -2.06076443e-01 7.50384331e-02
-5.99765658e-01 2.52211303e-01 6.37005508e-01 7.40715444e-01
6.81436479e-01 4.78601903e-02 2.42885754e-01 4.80201066e-01
2.03188106e-01 7.10502505e-01 1.16853468e-01 -1.61022317e+00
1.87346458e-01 3.13780963e-01 4.45291817e-01 -1.15796030e+00
-2.27289036e-01 1.48118462e-03 -6.39106631e-01 5.83873808e-01
1.29844636e-01 5.21315113e-02 -7.72128940e-01 1.62855446e+00
5.13443768e-01 2.44139954e-01 2.42571197e-02 1.06019115e+00
6.10130906e-01 5.77304184e-01 -4.43715334e-01 -1.70159653e-01
1.04846597e+00 -4.59719151e-01 -4.39009488e-01 5.47484793e-02
3.39243524e-02 -6.21542990e-01 1.20605838e+00 3.99942189e-01
-1.50976765e+00 -3.94477814e-01 -7.21823573e-01 -4.58794504e-01
2.25741163e-01 -1.30131990e-01 1.95338279e-01 4.61670518e-01
-1.30202699e+00 9.05805230e-01 -9.63307619e-01 -2.80284584e-01
-1.81329176e-02 3.17453772e-01 -7.67441690e-01 2.48908669e-01
-6.51591837e-01 8.40370059e-01 -3.36382568e-01 1.30946830e-01
-8.54357898e-01 -7.78141558e-01 -7.74410903e-01 -2.92591959e-01
7.99429417e-03 -1.03946567e+00 8.46398592e-01 -9.22233582e-01
-1.70458174e+00 1.33791065e+00 -2.16157660e-01 -4.47981283e-02
8.20492148e-01 -2.13460907e-01 -1.84484854e-01 4.50184405e-01
-4.94078606e-01 5.73931634e-01 8.61765265e-01 -1.91417909e+00
3.20931077e-01 -5.17210066e-01 3.77837151e-01 5.59665263e-01
-1.00667559e-01 -1.71218902e-01 -3.20737839e-01 -7.15724170e-01
3.40530634e-01 -1.19047701e+00 -7.07796365e-02 4.42476213e-01
-5.11970878e-01 6.77170575e-01 8.46622586e-01 -7.48366237e-01
6.26729250e-01 -1.84710920e+00 6.36620760e-01 1.99307770e-01
3.73125076e-01 -1.30552694e-01 -1.98017620e-02 4.21012491e-01
-1.36869490e-01 1.51825428e-01 -3.07111919e-01 -7.62878776e-01
-3.00437808e-01 4.19696778e-01 -4.25119281e-01 7.46930063e-01
-2.16815755e-01 9.69747901e-01 -6.49966300e-01 -1.69752970e-01
4.30084169e-01 1.14406693e+00 -8.13080847e-01 4.04525787e-01
9.99892596e-03 1.06655097e+00 -2.82048821e-01 4.37773675e-01
8.36891890e-01 -4.43811342e-02 5.30957282e-02 -4.70277190e-01
-1.47353351e-01 -1.07341431e-01 -1.24740386e+00 1.77634692e+00
-5.74449301e-01 5.26671648e-01 3.79914343e-01 -5.35619318e-01
6.53978348e-01 2.16605008e-01 5.38447261e-01 -2.74249792e-01
-1.41307255e-02 -4.99586836e-02 -8.47588480e-01 -6.34667099e-01
6.13566697e-01 -4.90140557e-01 2.03546524e-01 1.99754864e-01
-1.52670726e-01 -7.48773754e-01 -6.68701708e-01 -5.48207499e-02
8.76408458e-01 6.24835253e-01 -1.77187160e-01 -1.94765672e-01
3.81803066e-01 -3.35932106e-01 2.74303138e-01 3.27946842e-01
2.91197419e-01 1.38771975e+00 -5.52372485e-02 -7.80585170e-01
-1.29670787e+00 -1.45255756e+00 -3.81669730e-01 3.96071225e-01
3.02123815e-01 -2.04283804e-01 -1.02933764e+00 -2.32428104e-01
-1.38755769e-01 4.36901689e-01 -7.68778980e-01 6.50162697e-02
-7.78850436e-01 -4.23886001e-01 3.81089091e-01 2.47703403e-01
3.11094880e-01 -8.14084411e-01 -6.70050442e-01 -1.75543115e-01
-3.34659666e-01 -1.09252405e+00 -5.79290271e-01 -4.06552196e-01
-1.01126337e+00 -1.05088663e+00 -9.63324428e-01 -5.46001613e-01
9.04140294e-01 3.57871562e-01 1.22216249e+00 1.12226814e-01
-2.57229686e-01 9.40298557e-01 1.91420987e-01 2.98929989e-01
-5.72431922e-01 -7.32622623e-01 2.33242348e-01 3.71266276e-01
-6.28104568e-01 -9.16620314e-01 -7.87391245e-01 6.09011829e-01
-1.09169269e+00 2.22500294e-01 -2.38763854e-01 4.44776535e-01
9.25175130e-01 -4.14162546e-01 -4.30211514e-01 -6.21242285e-01
2.69145459e-01 -1.75115361e-04 -7.15902925e-01 2.88430780e-01
-1.03727216e-02 -9.08330828e-03 6.66699529e-01 -4.65127498e-01
-1.18639934e+00 2.22796738e-01 -1.56427160e-01 -9.15790617e-01
-1.38362199e-01 -1.52288079e-01 -7.03277811e-02 -7.67838299e-01
9.27988827e-01 6.42946288e-02 -1.52855307e-01 -4.90650237e-01
3.67042631e-01 1.93097487e-01 8.22797418e-01 -6.40678048e-01
1.25063586e+00 1.02718246e+00 3.05055887e-01 -9.97434735e-01
-4.77841139e-01 1.03397951e-01 -8.73707294e-01 -3.86204094e-01
1.09534788e+00 -7.99072444e-01 -1.12036312e+00 4.55726743e-01
-1.17684031e+00 -5.25259852e-01 -6.41290247e-01 4.19377774e-01
-9.19916928e-01 4.97109860e-01 -6.34733140e-01 -6.73994243e-01
3.52558270e-02 -1.09164691e+00 1.72302651e+00 1.36529906e-02
-1.53116673e-01 -1.30545914e+00 5.78580439e-01 6.78332865e-01
1.57422557e-01 8.15457046e-01 4.17758763e-01 5.25781095e-01
-9.99224186e-01 -1.60055421e-02 3.32760692e-01 3.40992898e-01
-1.29219359e-02 2.79463202e-01 -1.24523401e+00 -3.80766034e-01
3.85775983e-01 -2.60507286e-01 3.65600020e-01 3.90116006e-01
1.14130414e+00 -5.88577032e-01 -1.53457969e-01 1.31068265e+00
1.69152665e+00 2.06446648e-02 9.76453781e-01 6.54420629e-02
1.16254461e+00 5.50118923e-01 -9.70434211e-03 1.59918666e-01
4.54900026e-01 1.08953929e+00 6.56007171e-01 -2.69150734e-01
-8.23850632e-02 -3.71838450e-01 4.62471396e-01 5.75596631e-01
-8.81805301e-01 -1.68748230e-01 -5.28507113e-01 1.91674188e-01
-1.36037612e+00 -1.00977850e+00 -1.70959204e-01 2.52956533e+00
5.35507202e-01 -4.61456656e-01 -9.55754817e-02 -1.02348574e-01
4.76911336e-01 9.40606445e-02 -4.29517955e-01 -3.62232149e-01
-9.57200751e-02 2.00121060e-01 3.07058454e-01 1.20136440e+00
-6.95151210e-01 5.69003344e-01 6.69622803e+00 6.57898560e-02
-1.10766900e+00 4.84147593e-02 3.44352663e-01 -8.50260407e-02
-9.67124164e-01 -1.09444223e-02 -3.01567018e-01 1.86904836e-02
7.34965861e-01 1.95872158e-01 7.63878644e-01 4.92661804e-01
3.16902936e-01 2.43796706e-02 -1.01427460e+00 9.98117507e-01
3.04423630e-01 -1.50442052e+00 2.04684287e-01 3.16701502e-01
9.86049592e-01 -1.13683172e-01 3.36435325e-02 -5.65869629e-01
1.61101595e-01 -1.01444554e+00 8.73417020e-01 1.10543823e+00
1.05566239e+00 -3.10906708e-01 -5.37845120e-02 3.33907962e-01
-8.56344044e-01 3.06201577e-01 -3.17705572e-01 2.39123836e-01
4.59551632e-01 2.67520159e-01 -4.03388649e-01 6.55637681e-01
6.72750354e-01 6.49703562e-01 -3.66115451e-01 5.86642325e-01
1.46933809e-01 -5.77419512e-02 -3.48634511e-01 7.00649500e-01
-3.24193686e-01 -6.64739490e-01 8.74378800e-01 6.77895904e-01
4.84444052e-01 5.05946696e-01 -9.17594954e-02 8.72297347e-01
-2.28162542e-01 -2.82510221e-01 -9.29881155e-01 6.03940785e-01
-7.49401897e-02 1.26871908e+00 -4.51169908e-01 -8.24664757e-02
-1.35800950e-02 1.42325640e+00 2.60552555e-01 5.25398076e-01
-8.12452316e-01 2.17883676e-01 7.06637919e-01 5.61561704e-01
-8.27071890e-02 -4.73405331e-01 1.16473131e-01 -1.82385480e+00
4.40117657e-01 -5.60660660e-01 -1.84680954e-01 -1.56829596e+00
-9.03248847e-01 9.51198101e-01 8.22125822e-02 -1.34420228e+00
-1.13765322e-01 -5.32072186e-01 -4.60078120e-01 8.19339335e-01
-1.10521770e+00 -1.11432433e+00 -6.88107669e-01 8.90989482e-01
1.79976240e-01 4.72832680e-01 1.00056911e+00 -2.06911080e-02
-7.93341324e-02 2.44316489e-01 1.11425057e-01 -2.94208795e-01
4.50429320e-01 -1.29714286e+00 5.56106567e-01 7.84752965e-01
2.60695070e-01 7.13239193e-01 6.34609461e-01 -5.09109795e-01
-1.66759253e+00 -8.42183053e-01 5.13713777e-01 -1.14357758e+00
1.33148327e-01 -4.11310256e-01 -9.42338169e-01 1.06310725e+00
4.56737727e-02 3.59602600e-01 2.85785794e-01 -5.13434410e-01
-2.07430795e-01 -7.37007558e-02 -1.58793759e+00 6.75765216e-01
1.38338554e+00 -8.59989762e-01 -5.64517856e-01 6.31750822e-01
4.85979795e-01 -1.02009380e+00 -1.11625886e+00 2.35588208e-01
7.59260535e-01 -1.25437391e+00 1.40926349e+00 -4.25955206e-01
2.20455438e-01 -4.32210863e-01 -2.68165767e-01 -1.35433877e+00
-6.51529506e-02 -1.13332605e+00 6.60261363e-02 7.38118947e-01
-1.04596056e-01 -6.55435145e-01 6.49897337e-01 9.53948736e-01
-1.59654573e-01 -6.12935185e-01 -8.86450171e-01 -7.35514343e-01
2.09515728e-03 -2.71183670e-01 2.50426948e-01 1.02040994e+00
-5.71372926e-01 -1.81518555e-01 -5.78031778e-01 3.94297183e-01
7.51931965e-01 -1.14832097e-03 7.87053466e-01 -1.21688378e+00
-4.17038649e-01 2.88668066e-01 -4.13124770e-01 -1.11419165e+00
9.51616317e-02 -7.22228765e-01 -9.95978639e-02 -1.31349206e+00
4.11693193e-02 -2.11151063e-01 6.58035040e-01 2.15736419e-01
3.01046431e-01 8.08746159e-01 1.21588334e-01 5.22808552e-01
2.65351683e-02 4.83574331e-01 1.70305133e+00 3.83595765e-01
-2.93095291e-01 -1.45230487e-01 -3.10344040e-01 1.07060444e+00
2.66929448e-01 -1.50960565e-01 -4.28642809e-01 -6.88821614e-01
4.69171137e-01 3.19511741e-01 1.04002917e+00 -8.61431360e-01
-1.62675038e-01 -3.00250769e-01 5.32483518e-01 -8.21749121e-02
7.98473179e-01 -1.00264287e+00 1.05464387e+00 5.11521399e-02
-2.86830276e-01 4.40093935e-01 1.25639349e-01 4.40883875e-01
1.91992193e-01 6.10055663e-02 8.02955985e-01 -2.51890272e-01
-9.11061242e-02 3.41429502e-01 9.01320856e-03 4.73887585e-02
9.25429821e-01 -4.74546134e-01 -1.25311941e-01 -5.95742643e-01
-1.05542636e+00 -4.53347892e-01 1.33474135e+00 1.58026427e-01
8.01451087e-01 -1.34095371e+00 -5.43199182e-01 5.31783044e-01
-2.88845003e-01 2.06942126e-01 4.69949812e-01 5.62927186e-01
-1.18912721e+00 -5.17587177e-02 -8.57346803e-02 -7.77593434e-01
-1.40718341e+00 3.86246771e-01 9.73712742e-01 9.10327360e-02
-9.49822903e-01 5.95259726e-01 4.73210901e-01 -7.98904538e-01
-1.19397141e-01 -4.85889375e-01 2.41004080e-01 -5.04660368e-01
2.81858683e-01 3.50902289e-01 2.15833202e-01 -9.88435864e-01
-3.48724455e-01 1.40647542e+00 5.03549516e-01 -4.21750367e-01
1.45613134e+00 -3.44818950e-01 -4.07868437e-03 1.81047812e-01
1.33160901e+00 6.93930745e-01 -1.96263146e+00 1.49252668e-01
-9.14100111e-01 -7.96562195e-01 -2.14141309e-01 -5.48130810e-01
-1.15054071e+00 6.86052144e-01 3.68955374e-01 1.55185863e-01
1.09924388e+00 -4.91584092e-02 6.15623355e-01 3.29949915e-01
5.07304966e-01 -5.47676623e-01 4.24773470e-02 1.49133101e-01
1.38092351e+00 -7.78048635e-01 -3.90531607e-02 -5.35126090e-01
-4.35419351e-01 1.11852300e+00 1.88518345e-01 -5.00060678e-01
4.53495234e-01 2.14453116e-01 9.53519270e-02 -2.51204133e-01
-2.78406352e-01 4.36657935e-01 7.10103214e-01 6.42048955e-01
4.13172543e-02 -4.25215736e-02 3.73329103e-01 -3.30269188e-01
-4.18042839e-01 -1.11077480e-01 7.41338372e-01 6.99413061e-01
1.14253193e-01 -9.36265469e-01 -7.41107404e-01 -2.61299670e-01
-1.74084172e-01 1.27627984e-01 -5.08099675e-01 8.29387605e-01
-3.72381061e-02 5.77711701e-01 2.18731463e-01 -3.74726385e-01
6.65866256e-01 -9.46120396e-02 1.14652991e+00 -3.69388789e-01
-5.03453374e-01 -1.21415895e-03 -4.25935872e-02 -1.01512015e+00
-7.19514728e-01 -7.00661421e-01 -1.01142156e+00 -4.71313179e-01
1.65299252e-01 -1.90422386e-01 5.26046813e-01 7.32196867e-01
4.43705887e-01 1.27671659e-01 9.40186858e-01 -1.51020467e+00
-2.71135092e-01 -2.93271691e-01 -4.51003700e-01 8.27185154e-01
8.31264734e-01 -5.00093400e-01 -5.86956918e-01 4.50407326e-01] | [9.18468952178955, -2.862116575241089] |
9061e9da-6b2f-4c14-b3d0-7c74bc92b902 | real-time-visual-analysis-of-microvascular | null | null | http://openaccess.thecvf.com/content_cvpr_2015/html/Liu_Real-Time_Visual_Analysis_2015_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2015/papers/Liu_Real-Time_Visual_Analysis_2015_CVPR_paper.pdf | Real-Time Visual Analysis of Microvascular Blood Flow for Critical Care | Microcirculatory monitoring plays an important role in diagnosis and treatment of critical care patients. Sidestream Dark Field (SDF) imaging devices have been used to visualize and support interpretation of the micro-vascular blood flow. However, due to subsurface scattering within the tissue that embeds the capillaries, transparency of plasma, imaging noise and lack of features, it is difficult to obtain reliable physiological data from SDF videos. Therefore, thus far microcirculatory videos have been analyzed manually with significant input from expert clinicians. In this paper, we present a framework that automates the analysis process. It includes stages of video stabilization, enhancement, and micro-vessel extraction, in order to automatically estimate statistics of the micro blood flows from SDF videos. Our method has been validated in critical care experiments conducted carefully to record the microcirculatory blood flow in test animal subjects before, during and after induced bleeding episodes, as well as to study the effect of fluid resuscitation. Our method is able to extract microcirculatory measurements that are consistent with clinical intuition and it has a potential to become a useful tool in critical care medicine. | ['Chao Liu', 'Michael R. Pinsky', 'Hernando Gomez', 'Artur Dubrawski', 'Srinivasa Narasimhan', 'Brian Zuckerbraun'] | 2015-06-01 | null | null | null | cvpr-2015-6 | ['video-stabilization'] | ['computer-vision'] | [ 3.46392319e-02 -3.82997721e-01 2.95615971e-01 9.57688019e-02
1.41774088e-01 -4.58509892e-01 1.48991570e-01 6.05831385e-01
-6.26643002e-01 8.71046782e-01 1.61615312e-01 -3.21115375e-01
1.01461127e-01 -4.47120577e-01 -2.91879863e-01 -8.57605577e-01
-3.83871645e-01 1.92388549e-01 4.71216738e-01 2.49801457e-01
3.17987502e-01 9.65742826e-01 -1.10346866e+00 3.73085082e-01
5.18632412e-01 7.83675730e-01 1.13148153e-01 1.05452800e+00
-6.36601597e-02 1.09687710e+00 -4.89927053e-01 8.96358266e-02
5.18547930e-02 -7.70800829e-01 -4.36234653e-01 -8.06691721e-02
1.56758785e-01 -8.18575919e-01 -8.13141614e-02 5.53885043e-01
9.36566651e-01 1.27240655e-03 7.10609555e-01 -6.75517917e-01
2.42982551e-01 2.45042548e-01 -4.03362632e-01 8.50420773e-01
3.28194469e-01 3.92646074e-01 6.71104640e-02 -7.39601612e-01
7.71715283e-01 9.44196999e-01 2.35095367e-01 5.59186757e-01
-1.48104072e+00 -1.89832285e-01 -3.94439518e-01 3.49649459e-01
-9.34592485e-01 -4.65656340e-01 4.36486512e-01 -9.73687768e-01
6.22684956e-01 1.99261159e-01 1.05833495e+00 6.97319508e-01
7.35556781e-01 2.60568976e-01 1.10501885e+00 -5.22695422e-01
3.09155732e-01 4.95960295e-01 6.05625054e-03 4.47879463e-01
4.12483245e-01 4.61829938e-02 -3.65166783e-01 -5.49315300e-04
9.40547228e-01 3.87675799e-02 -8.83670807e-01 -5.50898433e-01
-1.08379841e+00 4.65524226e-01 -2.27456335e-02 7.19732940e-01
-4.91522700e-01 1.00668799e-02 5.25872529e-01 9.34131593e-02
2.93281257e-01 3.10534358e-01 -6.05809465e-02 -6.15300655e-01
-6.65473223e-01 -2.79047072e-01 8.60554457e-01 2.02174440e-01
9.64983925e-02 -7.83724040e-02 -4.74173343e-03 4.76286381e-01
2.22228900e-01 5.07024109e-01 5.33118129e-01 -1.09748161e+00
1.43252537e-02 4.91781533e-01 2.91766435e-01 -1.05994725e+00
-3.90145212e-01 1.09588224e-02 -7.72639334e-01 5.84486485e-01
7.89325714e-01 -2.83615440e-01 -3.52327704e-01 9.22432423e-01
4.43762928e-01 2.06008345e-01 -5.83992936e-02 1.35098338e+00
6.16931260e-01 2.62366712e-01 2.22274885e-01 -5.68523467e-01
1.13533473e+00 -3.54811579e-01 -9.05526578e-01 3.14173758e-01
6.79171026e-01 -6.85224175e-01 9.63969648e-01 4.15217757e-01
-1.13218701e+00 -3.06299537e-01 -7.39765823e-01 2.89554626e-01
-1.52224489e-03 7.36742467e-02 3.38493705e-01 5.40293574e-01
-7.66663790e-01 8.51216614e-01 -1.24500704e+00 -4.43773925e-01
4.61208493e-01 8.84699076e-02 -6.80877209e-01 -3.12192421e-02
-6.89634323e-01 1.03938448e+00 -1.44497201e-01 3.61948520e-01
-5.40748537e-01 -9.19884562e-01 -6.51079714e-01 -5.12103774e-02
-1.07211560e-01 -7.84826517e-01 8.68368447e-01 -4.53342825e-01
-1.60269344e+00 8.08598399e-01 -3.29817832e-01 -4.38260376e-01
8.27962697e-01 -2.43116736e-01 -5.11886217e-02 1.03889740e+00
-3.23308408e-01 1.63919643e-01 3.18063825e-01 -1.12361336e+00
-2.99948066e-01 -3.76386166e-01 -2.36144066e-01 -9.09656137e-02
6.85115308e-02 1.08603172e-01 1.98354632e-01 -2.35209122e-01
-8.93995538e-02 -5.59384286e-01 -2.28994399e-01 3.78846943e-01
3.34718563e-02 7.54414856e-01 6.09819949e-01 -4.41881746e-01
1.28802860e+00 -2.01469946e+00 -1.99049473e-01 1.63619682e-01
5.47299027e-01 3.65530968e-01 2.90530682e-01 4.69945401e-01
2.03113817e-02 -8.41674395e-03 1.87886879e-02 1.79739669e-01
-5.06602705e-01 -1.21842138e-01 -3.54570486e-02 9.66604352e-01
-1.84279978e-01 6.57576084e-01 -8.86197448e-01 -8.36447060e-01
8.00854146e-01 6.49987400e-01 -2.78031677e-01 2.92425454e-01
3.61570358e-01 1.12406981e+00 -4.54761356e-01 4.39715803e-01
6.26034975e-01 -9.37514231e-02 2.66577452e-01 -3.91916126e-01
-6.22041225e-01 -2.21754909e-01 -7.97687292e-01 1.11485147e+00
-3.11863065e-01 8.04432571e-01 4.92903084e-01 -8.43103409e-01
6.40135467e-01 4.57648516e-01 1.00760126e+00 -7.78280258e-01
6.48636222e-01 1.55220345e-01 1.93383709e-01 -1.19725931e+00
-1.36483341e-01 -5.97035348e-01 9.10838425e-01 4.76865798e-01
-2.52661914e-01 -9.87709910e-02 4.77892369e-01 9.85058025e-02
8.35923493e-01 -1.45860799e-02 2.91011393e-01 -5.09301603e-01
8.83385658e-01 1.57886371e-02 1.39449105e-01 1.98619485e-01
-6.22368872e-01 6.59647644e-01 9.15636063e-01 -7.78350651e-01
-7.76169002e-01 -9.11016464e-01 -3.91531974e-01 1.46617055e-01
3.18775922e-01 -3.00651789e-02 -7.51052380e-01 -1.27934799e-01
6.27899393e-02 2.95563787e-01 -6.38798296e-01 7.36898109e-02
-6.42675340e-01 -7.43682563e-01 -1.02498792e-02 3.07015181e-01
-4.82949913e-02 -9.71950591e-01 -1.51729560e+00 7.37582266e-01
-2.11481676e-01 -1.25221097e+00 8.93603638e-02 -1.59384031e-02
-1.06343877e+00 -1.60298514e+00 -8.58782172e-01 -3.56887132e-01
6.79793954e-01 2.75633693e-01 8.33438396e-01 1.44041926e-01
-9.04076099e-01 4.08867061e-01 -4.62124616e-01 -4.16816682e-01
-5.50824106e-01 -7.10318029e-01 -1.90875009e-01 7.27955028e-02
3.63826275e-01 -6.37774825e-01 -1.14536989e+00 4.34037745e-01
-6.93874180e-01 -1.71954066e-01 3.01312864e-01 4.60851341e-01
6.20442331e-01 -4.55389798e-01 2.37205848e-01 -8.70597601e-01
4.54044521e-01 -3.06774765e-01 -6.66661024e-01 -1.05253175e-01
-3.41604650e-01 -3.97947073e-01 8.84547651e-01 -2.00698406e-01
-8.16770136e-01 -2.20776230e-01 4.83818837e-02 -3.85143220e-01
-4.12257731e-01 2.04063237e-01 5.13710320e-01 -7.40174577e-02
8.28812838e-01 -8.82622898e-02 6.88558161e-01 -2.09827214e-01
-1.83648989e-01 5.16927838e-01 3.94182295e-01 -3.15147698e-01
1.38234034e-01 9.16378498e-01 4.63274866e-01 -1.18307471e+00
-2.79561281e-01 -6.80189550e-01 -9.55999613e-01 -7.50351310e-01
8.74740660e-01 -5.09714663e-01 -1.12065315e+00 2.99634069e-01
-9.53945398e-01 -5.48012972e-01 -4.13969845e-01 1.26988208e+00
-3.98857623e-01 7.28016376e-01 -8.83856356e-01 -7.94608414e-01
-3.33072960e-01 -1.18096900e+00 5.24458706e-01 1.79492980e-01
-2.78562307e-01 -1.35166180e+00 3.03625941e-01 1.25060126e-01
7.29460955e-01 8.80455971e-01 9.35835958e-01 7.57266432e-02
-3.84266287e-01 -2.18145192e-01 6.77911490e-02 3.52127343e-01
1.92335904e-01 5.64890146e-01 -7.34297812e-01 -9.90172997e-02
1.95101649e-01 1.00968376e-01 5.26970625e-01 1.11720562e+00
8.43816280e-01 7.19914911e-03 -2.53599823e-01 4.37381327e-01
1.31569052e+00 3.34261388e-01 8.18233311e-01 1.20558158e-01
1.29119083e-01 1.06343818e+00 4.87218380e-01 7.11345255e-01
-2.63465047e-01 2.26750165e-01 3.35825592e-01 -3.43025923e-01
-6.40055090e-02 3.55982304e-01 1.77450195e-01 5.03562868e-01
-3.90612125e-01 -2.74680734e-01 -7.00547993e-01 3.56018990e-01
-1.26597261e+00 -8.77100587e-01 -7.57976532e-01 2.34364891e+00
4.42557812e-01 -3.30252759e-02 2.29185715e-01 7.22296238e-02
7.32725263e-01 -4.99487251e-01 -2.04413235e-01 -4.30305094e-01
1.57431915e-01 3.48405063e-01 3.35095495e-01 6.97826743e-01
-7.49401927e-01 2.74115294e-01 6.94120598e+00 -1.84965208e-01
-1.67745745e+00 -1.88404337e-01 6.02495849e-01 -1.37699381e-01
4.74594310e-02 -4.11935262e-02 -2.84972399e-01 5.73540747e-01
7.40415812e-01 -9.97452959e-02 1.05412602e-02 3.91901970e-01
9.39873397e-01 -9.25100863e-01 -9.71936464e-01 8.00288439e-01
-2.27042571e-01 -1.27375817e+00 -2.37409562e-01 1.34183958e-01
7.59485066e-02 -3.33887666e-01 -3.38693738e-01 -4.80617017e-01
-5.40879011e-01 -8.89465034e-01 2.09642336e-01 6.65026903e-01
1.02445781e+00 -1.26886874e-01 1.18122971e+00 2.56862342e-01
-7.70008922e-01 1.32478222e-01 -2.70278513e-01 -9.25338939e-02
4.80813026e-01 1.03982902e+00 -1.00882220e+00 3.50811370e-02
3.31825495e-01 6.37637436e-01 -2.62746155e-01 1.34630680e+00
1.68039858e-01 4.15350795e-01 -1.00387067e-01 -1.84165016e-02
-1.30649135e-01 -4.24536079e-01 5.04308581e-01 1.07659853e+00
2.50398844e-01 1.58502296e-01 -2.82754481e-01 9.46728051e-01
6.12474799e-01 5.29113829e-01 -6.08074129e-01 -1.53384591e-02
-8.38616863e-03 1.29482949e+00 -1.11061692e+00 -3.52087438e-01
-4.98177618e-01 3.87911409e-01 -3.49858046e-01 3.57010722e-01
-4.94998634e-01 -1.43034607e-01 3.56788546e-01 9.39284682e-01
-2.43771404e-01 -5.15256785e-02 -3.43874469e-02 -1.09491539e+00
-1.42316073e-01 -3.05110604e-01 2.47954220e-01 -6.42687738e-01
-9.16542828e-01 4.37356055e-01 1.87185228e-01 -1.43406165e+00
-9.20889303e-02 -6.79502308e-01 -8.29390168e-01 8.70332658e-01
-1.78432512e+00 -2.83756196e-01 -7.50288308e-01 2.77394801e-01
-3.81189063e-02 2.42017001e-01 7.44956791e-01 3.52539301e-01
-4.92417485e-01 -1.49805367e-01 2.90337261e-02 -8.37636590e-02
8.02644908e-01 -1.20713449e+00 -5.16480505e-01 7.21518457e-01
-6.58790290e-01 5.51856220e-01 8.66603136e-01 -4.96517539e-01
-1.16497886e+00 -5.95709324e-01 4.73272294e-01 -2.57913321e-01
5.56904435e-01 2.69616581e-02 -7.12771297e-01 7.49420226e-02
6.22359850e-02 5.62284887e-01 1.05931056e+00 -8.46224904e-01
4.87186134e-01 -2.34463736e-01 -1.26196826e+00 2.19847143e-01
1.49712652e-01 5.61968088e-02 -3.62953275e-01 1.72299549e-01
-2.05771163e-01 -4.96521086e-01 -1.18088973e+00 3.00727755e-01
8.74446332e-01 -1.35428119e+00 7.29257464e-01 -3.72554302e-01
4.55292374e-01 -2.73550212e-01 4.38710541e-01 -1.07717001e+00
1.93168111e-02 -4.49274540e-01 8.31973329e-02 6.77515209e-01
9.59079713e-02 -9.08013225e-01 6.53585851e-01 6.92734778e-01
-4.94307317e-02 -5.26595116e-01 -7.76417613e-01 -2.83909261e-01
-3.39672156e-02 7.07083046e-02 -1.31443769e-01 4.97555405e-01
5.96909463e-01 7.19285682e-02 7.00695515e-02 -3.52713674e-01
5.48811018e-01 1.28840193e-01 7.69043386e-01 -1.22465527e+00
-9.36205909e-02 -3.62207204e-01 -5.96128464e-01 -5.35605550e-01
-2.21600562e-01 -4.21663523e-01 -1.19276553e-01 -1.53857422e+00
9.12823807e-03 -2.04700798e-01 -7.91899934e-02 -1.87090710e-01
3.10135745e-02 1.55474126e-01 -1.05956048e-01 1.17748611e-01
-3.42541598e-02 8.19476172e-02 1.88535082e+00 4.23888803e-01
-3.73904616e-01 -2.04258814e-01 -2.20408335e-01 4.57337260e-01
7.47096360e-01 -3.68772835e-01 -3.49769562e-01 1.85520098e-01
-2.45591596e-01 4.73509789e-01 4.44308639e-01 -1.04946089e+00
3.65440398e-02 5.86800650e-03 6.81223691e-01 -3.76790792e-01
7.13631660e-02 -1.03632975e+00 2.65738517e-01 9.99996960e-01
-1.00213580e-01 -1.30882606e-01 1.28196925e-01 3.57586652e-01
-5.48187912e-01 -2.76433319e-01 1.24843740e+00 -5.26768386e-01
-2.23366573e-01 -1.43517368e-02 -1.01363814e+00 1.12443365e-01
1.35551202e+00 -6.21654510e-01 -4.84409302e-01 -2.54174620e-01
-9.53172088e-01 1.45119056e-01 5.83971560e-01 -3.62396300e-01
7.25860298e-01 -5.78702092e-01 -6.22046411e-01 3.85305703e-01
-1.18774794e-01 -1.14257365e-01 7.99703836e-01 1.72658718e+00
-1.66954875e+00 3.34537268e-01 -4.57053512e-01 -8.96371841e-01
-1.28819513e+00 4.74306375e-01 5.53847551e-01 8.91786069e-02
-1.07834923e+00 4.71131533e-01 2.82951534e-01 6.75478995e-01
-2.61800718e-02 -5.69100440e-01 -6.24141216e-01 6.79085730e-03
8.03559124e-01 6.51103079e-01 -2.94089895e-02 -6.34278119e-01
-2.90112674e-01 7.88425922e-01 2.03694582e-01 2.34520033e-01
1.07210004e+00 -4.95510936e-01 -1.62760258e-01 4.76023823e-01
1.08769858e+00 1.70206413e-01 -1.32161307e+00 4.87745583e-01
-4.84173805e-01 -8.09711456e-01 -4.47814092e-02 -6.29723072e-01
-1.09228992e+00 1.35504651e+00 5.57167530e-01 4.21392381e-01
1.03200507e+00 -3.76896471e-01 6.86103284e-01 -2.09458709e-01
2.23599523e-01 -9.84023452e-01 1.57858089e-01 -1.81322157e-01
5.80540836e-01 -1.02659929e+00 7.78545439e-02 -8.17509234e-01
-5.10070503e-01 1.65586698e+00 3.66718590e-01 -3.23062725e-02
8.53403330e-01 6.18538439e-01 5.20414889e-01 -1.69977903e-01
-7.84835279e-01 6.87657893e-02 -1.50695696e-01 5.66917956e-01
6.05104864e-01 -1.88148215e-01 -5.43669939e-01 -7.37142712e-02
3.43592405e-01 3.99787188e-01 1.10741687e+00 9.03546393e-01
-6.51343703e-01 -7.60331511e-01 -2.67501175e-01 3.17039758e-01
-1.05757701e+00 2.30238110e-01 2.31265984e-02 8.04445148e-01
-1.78598061e-01 9.60051239e-01 -2.26309657e-01 2.59187698e-01
5.06538451e-01 -6.34636134e-02 7.76939452e-01 -3.68474036e-01
-3.81045491e-01 3.79965693e-01 -1.83300689e-01 -5.89192390e-01
-7.65061736e-01 -5.43468773e-01 -1.35819411e+00 -1.23405740e-01
-1.26939312e-01 3.73264581e-01 6.93767965e-01 5.58550000e-01
2.01725155e-01 4.92647201e-01 5.72496831e-01 -6.39513850e-01
1.22754686e-01 -1.00270617e+00 -1.04089332e+00 4.92256969e-01
5.07738590e-01 -6.80678844e-01 -8.46261561e-01 5.44585466e-01] | [13.794171333312988, -2.9542157649993896] |
fa3d9ebc-f895-4424-bc0e-92ebaceec00d | adanns-a-framework-for-adaptive-semantic | 2305.19435 | null | https://arxiv.org/abs/2305.19435v1 | https://arxiv.org/pdf/2305.19435v1.pdf | AdANNS: A Framework for Adaptive Semantic Search | Web-scale search systems learn an encoder to embed a given query which is then hooked into an approximate nearest neighbor search (ANNS) pipeline to retrieve similar data points. To accurately capture tail queries and data points, learned representations typically are rigid, high-dimensional vectors that are generally used as-is in the entire ANNS pipeline and can lead to computationally expensive retrieval. In this paper, we argue that instead of rigid representations, different stages of ANNS can leverage adaptive representations of varying capacities to achieve significantly better accuracy-compute trade-offs, i.e., stages of ANNS that can get away with more approximate computation should use a lower-capacity representation of the same data point. To this end, we introduce AdANNS, a novel ANNS design framework that explicitly leverages the flexibility of Matryoshka Representations. We demonstrate state-of-the-art accuracy-compute trade-offs using novel AdANNS-based key ANNS building blocks like search data structures (AdANNS-IVF) and quantization (AdANNS-OPQ). For example on ImageNet retrieval, AdANNS-IVF is up to 1.5% more accurate than the rigid representations-based IVF at the same compute budget; and matches accuracy while being up to 90x faster in wall-clock time. For Natural Questions, 32-byte AdANNS-OPQ matches the accuracy of the 64-byte OPQ baseline constructed using rigid representations -- same accuracy at half the cost! We further show that the gains from AdANNS translate to modern-day composite ANNS indices that combine search structures and quantization. Finally, we demonstrate that AdANNS can enable inference-time adaptivity for compute-aware search on ANNS indices built non-adaptively on matryoshka representations. Code is open-sourced at https://github.com/RAIVNLab/AdANNS. | ['Ali Farhadi', 'Prateek Jain', 'Sham Kakade', 'Qingqing Cao', 'Alan Fan', 'Sharan Ranjit S', 'Aditya Kusupati', 'Aniket Rege'] | 2023-05-30 | null | null | null | null | ['natural-questions'] | ['miscellaneous'] | [-1.52580306e-01 -3.73246014e-01 -5.44246912e-01 -4.76846009e-01
-1.33294261e+00 -9.01141822e-01 4.87641007e-01 2.10037038e-01
-5.30678809e-01 1.91533174e-02 2.57219225e-01 -5.75623691e-01
-2.79090792e-01 -1.14937270e+00 -1.06736839e+00 -1.58417061e-01
1.55094057e-01 6.74209774e-01 3.14927697e-01 -4.93093342e-01
3.29065800e-01 7.52972662e-01 -1.72650719e+00 4.68160003e-01
3.05149853e-01 1.51543593e+00 -1.69056818e-01 9.98499691e-01
-4.00339365e-02 4.53159988e-01 -4.21064347e-01 -3.47613782e-01
6.68101907e-01 4.73016441e-01 -1.02317595e+00 -9.30913627e-01
9.79180872e-01 -9.65708852e-01 -8.01756680e-01 5.17573118e-01
4.83441502e-01 1.27432972e-01 6.45332694e-01 -9.42987978e-01
-9.06927705e-01 4.40002531e-01 -2.60745198e-01 2.39717305e-01
2.39409000e-01 3.46721739e-01 1.31640184e+00 -9.63647246e-01
2.68781245e-01 1.23165321e+00 8.05723310e-01 3.94867569e-01
-1.21601963e+00 -7.22557425e-01 -4.15330678e-01 1.74093038e-01
-1.80951488e+00 -6.41101718e-01 6.83108047e-02 1.46247253e-01
1.36083794e+00 5.60538650e-01 5.48390269e-01 7.61995852e-01
1.16260208e-01 7.08715141e-01 6.38422132e-01 -1.39025241e-01
2.94542581e-01 -3.63895714e-01 3.98128331e-01 9.67888415e-01
2.68341422e-01 8.39966834e-02 -7.18851864e-01 -5.60451210e-01
8.07350516e-01 4.74923104e-01 4.35360372e-02 -1.93033203e-01
-9.04224575e-01 9.78491247e-01 1.06572676e+00 1.00794882e-01
-2.33620927e-01 1.00779152e+00 5.37740529e-01 4.28352505e-01
5.43764085e-02 4.52963471e-01 -6.54630423e-01 -3.83299381e-01
-1.30244744e+00 3.88651699e-01 9.41492081e-01 1.09408045e+00
9.95795488e-01 -1.29672274e-01 -3.52740824e-01 6.19254351e-01
1.48842260e-01 8.55460286e-01 6.55971885e-01 -1.17004597e+00
2.93562353e-01 7.01773226e-01 -2.05176264e-01 -7.88078666e-01
-1.11960053e-01 -4.02539581e-01 -7.27003336e-01 6.86046183e-02
1.98949948e-02 5.01021206e-01 -1.09642243e+00 1.40731382e+00
2.07734033e-01 -1.22248910e-01 5.07559627e-02 9.51843858e-01
5.94275236e-01 9.21235204e-01 -1.85686752e-01 5.14185071e-01
1.62588227e+00 -9.64641511e-01 -1.00470539e-02 -9.41317603e-02
8.85460913e-01 -8.25407684e-01 1.44022036e+00 6.33829013e-02
-1.34719074e+00 -5.93044221e-01 -1.20467782e+00 -8.64188135e-01
-6.22087419e-01 -1.78163797e-01 7.71540403e-01 4.58407402e-01
-1.58214498e+00 5.80692589e-01 -9.54064250e-01 -8.33978951e-02
2.79462188e-01 6.12351060e-01 -1.81185633e-01 -1.23202438e-02
-1.01852429e+00 5.90296686e-01 1.85171068e-01 -1.36350691e-01
-6.72481358e-01 -9.94125128e-01 -5.68914711e-01 4.27470207e-01
2.06244841e-01 -8.48977625e-01 1.55432987e+00 -4.67871547e-01
-1.16269231e+00 6.78265631e-01 -3.52028072e-01 -9.01329279e-01
-1.12396546e-01 -5.45202911e-01 -2.14825392e-01 2.71308988e-01
-1.03754722e-01 7.61500299e-01 8.35280001e-01 -6.48267448e-01
-1.63465872e-01 -3.67725104e-01 2.11011752e-01 -4.86493558e-02
-5.46997368e-01 -2.89848536e-01 -6.32877946e-01 -7.43058503e-01
2.79762745e-01 -1.15009916e+00 3.01803723e-02 6.29656255e-01
-1.07091174e-01 -4.04897451e-01 7.57691145e-01 -2.82806844e-01
1.41699624e+00 -2.16076422e+00 -2.10577458e-01 4.74898458e-01
3.97547215e-01 6.06326878e-01 -2.50081539e-01 3.87154102e-01
3.63803029e-01 2.28980899e-01 -7.33568333e-03 -1.20733101e-02
4.02182519e-01 4.01658267e-01 -7.40967751e-01 3.09091449e-01
1.04439206e-01 1.34239340e+00 -5.93295753e-01 -4.30963337e-01
-5.19902594e-02 5.27481318e-01 -9.77520287e-01 1.20282464e-01
-4.01653558e-01 -5.60140908e-01 -5.76224804e-01 1.01794124e+00
5.25522947e-01 -7.01368153e-01 -1.79949731e-01 -5.54628491e-01
1.03325874e-01 6.80464745e-01 -7.78955221e-01 1.94572282e+00
-7.79391825e-01 6.13712370e-01 -3.51516567e-02 -5.46531856e-01
6.69652402e-01 -1.36814505e-01 2.40162477e-01 -9.28317904e-01
-3.42575759e-01 7.07731783e-01 -5.30964315e-01 1.70872733e-01
1.08727908e+00 6.81044817e-01 -4.18730617e-01 5.41462421e-01
-1.05189219e-01 -5.07522523e-01 -9.29856449e-02 3.21954310e-01
1.34659827e+00 -2.35032644e-02 -3.19971889e-02 -1.80991292e-01
3.03442240e-01 1.77375406e-01 -6.78789541e-02 1.22280514e+00
2.46173009e-01 4.12782222e-01 -8.14288333e-02 -9.84622538e-01
-1.30371952e+00 -1.23165429e+00 -3.54595989e-01 1.59947038e+00
-3.94642763e-02 -9.25223470e-01 -6.10725880e-01 -3.95071864e-01
4.39423442e-01 4.57279682e-01 -1.95001051e-01 -3.54489267e-01
-9.24650192e-01 -6.29449040e-02 1.19840181e+00 7.65127480e-01
5.04527330e-01 -4.82882291e-01 -1.08036160e+00 7.38933459e-02
2.38240197e-01 -6.34702504e-01 -8.69857013e-01 3.23594600e-01
-9.14759338e-01 -7.23834097e-01 -6.33086026e-01 -5.16822875e-01
2.49659687e-01 3.14099073e-01 1.42478299e+00 3.35435420e-01
-3.38046670e-01 5.34750521e-01 -2.58095890e-01 -8.89993310e-02
-2.49374598e-01 6.49002790e-01 3.67767550e-02 -8.26096058e-01
5.07489920e-01 -6.60328805e-01 -1.19084990e+00 2.12383017e-01
-1.29116726e+00 -3.32191169e-01 9.19964254e-01 8.73665869e-01
9.10167754e-01 -6.66546762e-01 -8.90247002e-02 -3.97573829e-01
5.61087072e-01 -5.18826127e-01 -7.83984005e-01 2.76675344e-01
-1.20240712e+00 8.31374168e-01 7.46513188e-01 -4.27245080e-01
-1.07027411e-01 -1.77007526e-01 -2.94063121e-01 -8.94683659e-01
3.52888405e-01 2.41081849e-01 6.91465914e-01 -2.76414037e-01
1.03687012e+00 4.11856800e-01 4.18884680e-02 -5.30936599e-01
7.51332998e-01 8.46472561e-01 6.11410916e-01 -9.17246044e-01
7.26769984e-01 3.10567468e-01 1.15159407e-01 -3.09901565e-01
-5.01334012e-01 -4.47457314e-01 -8.82821307e-02 5.75780690e-01
3.97247285e-01 -1.02405822e+00 -1.00463390e+00 1.31274879e-01
-1.16825926e+00 -3.89859200e-01 -5.45739114e-01 -9.56559256e-02
-5.33908129e-01 4.96378876e-02 -9.13349807e-01 -3.38818103e-01
-1.12794495e+00 -1.33366501e+00 1.65651739e+00 7.00519755e-02
-2.61904299e-01 -3.66593480e-01 -7.35068768e-02 2.52037674e-01
9.49114740e-01 -3.93840641e-01 9.82905686e-01 -7.84624577e-01
-1.05373883e+00 -1.92850560e-01 -5.46596050e-01 2.04264686e-01
-4.01388079e-01 -2.72803515e-01 -8.68555725e-01 -4.05223817e-01
-3.32162648e-01 -6.47578895e-01 8.88936222e-01 -2.29610801e-02
1.63690805e+00 -9.20197904e-01 -1.87635213e-01 1.18033564e+00
1.66150832e+00 -2.30394557e-01 5.25107741e-01 2.47425035e-01
5.02038658e-01 -2.01497778e-01 3.10562819e-01 4.84310240e-01
4.08865482e-01 9.31665301e-01 5.42309999e-01 1.70412660e-01
-3.61134261e-01 -3.98165822e-01 2.88811147e-01 9.67430353e-01
2.67059296e-01 -1.71883240e-01 -9.92744744e-01 4.52440113e-01
-1.58956575e+00 -6.08241677e-01 2.49368668e-01 2.14052582e+00
1.23976803e+00 1.94273870e-02 -1.03177801e-01 -9.72376540e-02
8.85298550e-02 3.70490909e-01 -1.01012206e+00 -9.26481605e-01
4.68691476e-02 1.01381958e+00 1.16441405e+00 3.85864019e-01
-7.41640508e-01 9.67680216e-01 5.97875834e+00 1.19090068e+00
-1.13298702e+00 1.33355767e-01 5.82408190e-01 -5.33946037e-01
-5.29305518e-01 -6.09817617e-02 -1.09778309e+00 4.17106152e-01
1.60967588e+00 7.76014430e-03 9.06658828e-01 1.22483039e+00
-4.99439418e-01 3.15371007e-01 -1.30117202e+00 1.23880374e+00
-1.79272503e-01 -1.76902878e+00 4.84341025e-01 1.99355379e-01
3.90775830e-01 4.47341681e-01 4.52687502e-01 4.13636714e-01
5.24950325e-01 -1.29024339e+00 6.80646479e-01 6.26893699e-01
1.35135412e+00 -5.73321402e-01 4.87827122e-01 -9.59311426e-02
-1.38524985e+00 -1.58608034e-01 -5.06063879e-01 3.33925426e-01
-2.81269252e-01 3.51891100e-01 -8.36006999e-01 5.40379137e-02
9.78212118e-01 5.93244843e-02 -6.15007520e-01 5.55412054e-01
3.96479368e-01 4.47882295e-01 -8.53580773e-01 -4.16141480e-01
6.08897388e-01 2.70275086e-01 1.64114088e-01 1.22087944e+00
4.68213856e-01 1.27068460e-02 -2.32112110e-01 9.19101894e-01
-4.39475328e-01 -1.39453650e-01 -5.90882301e-01 1.55974517e-03
1.04519928e+00 9.48838830e-01 -4.18377295e-02 -5.35700142e-01
-1.44261986e-01 1.24884498e+00 5.28890729e-01 5.74531741e-02
-7.53972352e-01 -5.23815095e-01 1.03792596e+00 3.87360185e-01
4.11326200e-01 -2.56842375e-01 -1.34291887e-01 -8.92947137e-01
1.25210837e-01 -1.19097054e+00 4.31271702e-01 -8.99831355e-01
-1.13613141e+00 5.63231170e-01 -8.16830620e-02 -8.93790185e-01
-6.72502756e-01 -6.35089219e-01 1.07044376e-01 1.09336531e+00
-1.41820323e+00 -1.07638013e+00 -3.03588420e-01 7.13375211e-01
6.22099698e-01 4.24648495e-03 1.19098961e+00 2.14534968e-01
1.61513537e-01 1.15362799e+00 4.45192695e-01 8.81551132e-02
5.35559416e-01 -9.37050879e-01 9.18635368e-01 2.88668305e-01
4.10442710e-01 1.30383825e+00 1.03413835e-01 -2.02378899e-01
-2.23383880e+00 -8.57785881e-01 5.46780586e-01 -4.51999277e-01
7.61278272e-01 -2.03737050e-01 -7.96871662e-01 5.67179501e-01
-1.20528713e-01 6.08533859e-01 4.70257044e-01 -5.83735779e-02
-1.07655644e+00 -5.30225575e-01 -1.07474840e+00 6.70807362e-01
1.13443291e+00 -1.17896652e+00 -4.11475837e-01 4.91801411e-01
1.16191661e+00 -7.23668277e-01 -1.16110289e+00 3.50403905e-01
1.08004701e+00 -9.01982009e-01 1.55463862e+00 -4.81360465e-01
3.25143188e-01 -8.60362127e-02 -7.68988848e-01 -6.03371918e-01
-3.00995559e-01 -5.12602448e-01 -7.57200599e-01 4.71685320e-01
1.48813888e-01 -7.33191729e-01 9.32200789e-01 6.78174436e-01
4.66815792e-02 -1.21194923e+00 -1.11628056e+00 -6.71461463e-01
1.92268685e-01 -2.85660475e-01 1.12922394e+00 3.91556621e-01
-5.35733759e-01 7.12132305e-02 2.13315085e-01 2.11347908e-01
3.13501298e-01 1.41053975e-01 7.93812096e-01 -8.75293314e-01
-3.95280778e-01 -4.83445168e-01 -5.40785193e-01 -1.78950560e+00
-6.14433289e-02 -9.44263518e-01 -2.59274185e-01 -1.14206886e+00
-1.49009079e-01 -8.82457674e-01 -2.96939075e-01 7.71249413e-01
3.44717324e-01 3.42080891e-01 1.25579253e-01 7.41673231e-01
-5.30872285e-01 3.34377378e-01 7.08092809e-01 -3.12232673e-01
9.12052672e-03 -3.67443889e-01 -6.58243299e-01 2.98441201e-01
5.88184953e-01 -3.58115971e-01 -4.77368444e-01 -1.01972342e+00
4.36923146e-01 -3.45836990e-02 6.67803049e-01 -1.21715128e+00
7.32118070e-01 2.28215054e-01 4.28715497e-01 -7.68856466e-01
7.79280245e-01 -7.88017035e-01 6.44971207e-02 6.44481838e-01
-5.74862897e-01 7.81725585e-01 2.80564249e-01 4.13526177e-01
-4.42722738e-02 -2.30601341e-01 3.40906084e-01 -2.15795636e-01
-4.88432080e-01 4.19954151e-01 9.96911004e-02 9.19960886e-02
4.31856215e-01 -1.39632314e-01 -6.72445893e-01 -3.66708457e-01
-2.64260788e-02 -1.55405015e-01 6.03066444e-01 1.27783567e-01
7.22183764e-01 -1.26710665e+00 -4.25669730e-01 3.78055394e-01
8.58459063e-03 3.10181767e-01 -1.34688854e-01 2.10027754e-01
-1.03808630e+00 7.89784133e-01 2.78756350e-01 -5.87702155e-01
-9.34657991e-01 4.61158901e-01 2.09283873e-01 -3.73851269e-01
-2.48084649e-01 9.62448299e-01 -2.42176205e-01 -4.25876528e-01
9.41622928e-02 -6.23529434e-01 7.56571472e-01 -3.68105412e-01
4.77522224e-01 4.23198432e-01 3.55549276e-01 -3.63733232e-01
-4.90725636e-01 5.85573673e-01 -3.84745240e-01 -3.87970619e-02
1.18480432e+00 2.43065476e-01 -1.73034862e-01 5.63365500e-03
1.70726264e+00 -1.34056387e-03 -9.88915741e-01 -4.19551998e-01
-2.84433872e-01 -3.46508682e-01 3.11085969e-01 -6.90129757e-01
-9.77394342e-01 7.07897007e-01 9.15650964e-01 8.83801952e-02
1.03961313e+00 8.12107325e-02 1.34299815e+00 1.02000701e+00
5.61123490e-01 -8.87249172e-01 1.65413141e-01 5.40808678e-01
8.19618583e-01 -7.57407427e-01 1.74706370e-01 1.57205924e-01
1.82253439e-02 1.29767680e+00 3.01085621e-01 -3.39331359e-01
5.94029546e-01 3.83468062e-01 -3.25188845e-01 -2.85593003e-01
-8.85311067e-01 5.34088910e-02 5.32698095e-01 2.48213604e-01
1.46097362e-01 -3.46202962e-02 1.59010410e-01 2.62922317e-01
-4.96270388e-01 4.74318815e-03 -2.10884899e-01 1.11972606e+00
-3.38807225e-01 -1.03059924e+00 -3.30668718e-01 8.09421241e-01
-3.66041392e-01 -7.65097439e-01 7.12164491e-03 5.27408302e-01
-2.89687306e-01 5.82276702e-01 4.98605132e-01 -6.16801739e-01
2.24271074e-01 8.24125186e-02 2.06270128e-01 -2.68244594e-01
-8.79711986e-01 -5.00947833e-01 -1.94040284e-01 -1.23834765e+00
1.37084812e-01 -1.40449315e-01 -1.28751302e+00 -1.04310310e+00
-1.90268740e-01 6.68413565e-02 9.20323849e-01 2.73620069e-01
1.20431983e+00 6.91854879e-02 4.16943938e-01 -6.63156688e-01
-1.30204737e+00 -5.39186418e-01 -3.69346738e-02 1.39709383e-01
4.67464983e-01 -4.28155251e-02 -3.79756898e-01 -4.33476985e-01] | [8.680315017700195, 3.4382193088531494] |
776c5002-d756-4b78-afce-85a0ad56a602 | deep-cross-modal-learning-for-caricature | 1807.11688 | null | http://arxiv.org/abs/1807.11688v1 | http://arxiv.org/pdf/1807.11688v1.pdf | Deep Cross Modal Learning for Caricature Verification and Identification(CaVINet) | Learning from different modalities is a challenging task. In this paper, we
look at the challenging problem of cross modal face verification and
recognition between caricature and visual image modalities. Caricature have
exaggerations of facial features of a person. Due to the significant variations
in the caricatures, building vision models for recognizing and verifying data
from this modality is an extremely challenging task. Visual images with
significantly lesser amount of distortions can act as a bridge for the analysis
of caricature modality. We introduce a publicly available large
Caricature-VIsual dataset [CaVI] with images from both the modalities that
captures the rich variations in the caricature of an identity. This paper
presents the first cross modal architecture that handles extreme distortions of
caricatures using a deep learning network that learns similar representations
across the modalities. We use two convolutional networks along with
transformations that are subjected to orthogonality constraints to capture the
shared and modality specific representations. In contrast to prior research,
our approach neither depends on manually extracted facial landmarks for
learning the representations, nor on the identities of the person for
performing verification. The learned shared representation achieves 91%
accuracy for verifying unseen images and 75% accuracy on unseen identities.
Further, recognizing the identity in the image by knowledge transfer using a
combination of shared and modality specific representations, resulted in an
unprecedented performance of 85% rank-1 accuracy for caricatures and 95% rank-1
accuracy for visual images. | ['Narayanan C. Krishnan', 'Jatin Garg', 'Himanshu Tolani', 'Skand Vishwanath Peri'] | 2018-07-31 | null | null | null | null | ['caricature'] | ['computer-vision'] | [ 2.56088167e-01 -6.71422854e-02 -1.59754232e-02 -5.87415457e-01
-8.82625103e-01 -8.50814164e-01 9.59058642e-01 -5.35126567e-01
-2.15017758e-02 3.91276687e-01 1.42090321e-01 3.78927216e-02
-1.40967136e-02 -4.77721483e-01 -9.26134408e-01 -6.83471441e-01
1.15859054e-01 1.98891208e-01 -3.67813379e-01 -1.39080614e-01
5.54884151e-02 9.13416922e-01 -1.89833653e+00 7.98798740e-01
1.86335132e-01 1.34301710e+00 -4.52585161e-01 7.40718842e-01
1.29200980e-01 5.69844723e-01 -4.12398279e-01 -7.99177051e-01
5.94411850e-01 -2.77754754e-01 -7.89383888e-01 1.12005770e-01
1.34613192e+00 -3.22538614e-01 -4.12248552e-01 9.58971143e-01
4.23230290e-01 -2.27859765e-01 9.11670446e-01 -1.69848967e+00
-1.34804249e+00 1.53888553e-01 -7.79837668e-01 -7.49911293e-02
6.51850402e-01 9.95844901e-02 6.86347723e-01 -1.17313027e+00
5.03932893e-01 1.52222168e+00 8.87246370e-01 8.02298844e-01
-1.08956492e+00 -1.01467335e+00 -1.45379797e-01 2.44241044e-01
-1.76602948e+00 -8.11945200e-01 7.88086236e-01 -5.19092917e-01
5.83992958e-01 1.61022574e-01 1.38656959e-01 1.23308778e+00
-1.03028804e-01 4.32277471e-01 1.16397035e+00 -3.42132896e-01
-4.35372859e-01 4.84571040e-01 -2.66820103e-01 7.02498555e-01
1.09550372e-01 4.92884576e-01 -5.78811824e-01 -2.78013460e-02
6.41816735e-01 1.64839685e-01 -3.06715041e-01 -4.01771426e-01
-1.01414204e+00 6.46439850e-01 5.00588119e-01 2.69969314e-01
-7.07544561e-04 3.97351794e-02 2.33433321e-01 3.50631595e-01
-1.41793847e-01 2.15557531e-01 -2.79092342e-01 3.84046555e-01
-8.33303750e-01 5.36871552e-02 4.60728407e-01 9.05881047e-01
7.15387166e-01 1.57613695e-01 -1.65444240e-02 7.93962657e-01
5.29746473e-01 9.51250792e-01 3.06610733e-01 -5.36370456e-01
2.53925383e-01 6.57974005e-01 -1.85463905e-01 -1.27340043e+00
-7.90842101e-02 -1.53792119e-02 -9.88187671e-01 6.37597740e-01
5.41502416e-01 1.49382398e-01 -1.10467505e+00 1.90786850e+00
2.19719633e-01 2.66366810e-01 3.35668087e-01 8.39812756e-01
1.17707157e+00 2.98281610e-01 1.62106324e-02 2.40331158e-01
1.48397517e+00 -4.10954088e-01 -5.45745313e-01 6.71901703e-02
1.73711494e-01 -1.06980336e+00 6.24555945e-01 1.45307049e-01
-1.00605190e+00 -7.49017954e-01 -1.12381947e+00 -3.58951598e-01
-7.25633621e-01 3.53973389e-01 4.03393537e-01 7.84823716e-01
-1.23926473e+00 2.53161162e-01 -1.05020970e-01 -4.40746456e-01
6.95735931e-01 5.26635647e-01 -1.25814378e+00 -2.18772247e-01
-9.90073383e-01 1.11893880e+00 -8.90941266e-03 2.45481402e-01
-1.26892662e+00 -9.03654873e-01 -1.28717959e+00 -1.42364532e-01
-1.49577752e-01 -2.13352621e-01 6.60355747e-01 -1.56473196e+00
-1.09247231e+00 1.52062130e+00 -9.63648707e-02 -1.92967191e-01
4.54271734e-01 8.86163414e-02 -9.01236296e-01 4.13348377e-02
-1.16541162e-01 7.23506570e-01 1.31112444e+00 -1.58082628e+00
-3.77550691e-01 -5.46228588e-01 -1.47635505e-01 -2.38725811e-01
-1.85833052e-01 2.00189576e-01 -2.00582817e-01 -4.65376854e-01
-5.57428412e-02 -9.80427384e-01 7.29487240e-01 5.28624296e-01
-2.13295668e-01 -2.70070303e-02 1.20989668e+00 -8.21551740e-01
4.87747610e-01 -2.31346011e+00 6.34288639e-02 2.71980852e-01
2.29362454e-02 2.47096568e-01 -5.00281990e-01 1.81491137e-01
-5.00511825e-01 1.36530071e-01 5.01357689e-02 -3.69429082e-01
-3.29873487e-02 3.88143398e-02 -3.77027422e-01 6.93093896e-01
6.07468486e-01 9.22393799e-01 -5.04496336e-01 -4.29359108e-01
9.74567160e-02 1.03121209e+00 -2.99117237e-01 2.31651247e-01
4.68249828e-01 3.88915747e-01 1.62555963e-01 1.12672162e+00
1.11375105e+00 1.05079375e-01 -2.80074645e-02 -6.80776477e-01
2.64349848e-01 -5.22118211e-01 -1.11312246e+00 1.42484462e+00
-4.00372565e-01 6.71983063e-01 2.91662902e-01 -7.93741882e-01
1.06446159e+00 5.47976434e-01 1.75068304e-01 -6.62665188e-01
3.80772680e-01 1.60944477e-01 -2.61278208e-02 -3.55865389e-01
1.82917386e-01 -3.85431528e-01 -8.90208110e-02 3.37137789e-01
5.05418122e-01 -1.95827000e-02 -3.01559687e-01 3.89106087e-02
4.38039005e-01 7.89067894e-02 1.23037085e-01 -5.70215955e-02
1.03075218e+00 -3.81782979e-01 3.80565733e-01 2.76556879e-01
-4.91669655e-01 9.25279260e-01 1.40248492e-01 -5.89565635e-01
-1.02192533e+00 -1.32328391e+00 -2.64651924e-01 9.17481720e-01
-7.96627179e-02 1.98963270e-01 -4.53893781e-01 -5.82269073e-01
5.13894141e-01 1.08617440e-01 -1.21142471e+00 -3.75850558e-01
-3.14527869e-01 2.61427779e-02 1.00309002e+00 6.41770303e-01
6.72826707e-01 -7.60158241e-01 1.05079152e-01 -7.18441606e-01
9.97221842e-02 -1.27157879e+00 -7.12503970e-01 -6.05519891e-01
-2.01647446e-01 -1.52730417e+00 -5.87776542e-01 -1.04644418e+00
8.33259225e-01 2.93018669e-01 8.22982490e-01 1.70836270e-01
-6.73435688e-01 8.12146127e-01 2.88847554e-02 -3.93356740e-01
-3.96474272e-01 -5.23945749e-01 2.74543971e-01 7.52724469e-01
4.95331854e-01 -3.52002323e-01 -4.17281091e-01 2.96829909e-01
-7.82252967e-01 -3.20165753e-01 4.78820652e-01 7.62863159e-01
2.67250389e-01 -3.51473987e-01 4.17226851e-01 -4.12882537e-01
2.11251169e-01 -5.46872616e-01 -4.70122129e-01 5.80757678e-01
-4.99780476e-02 3.40704769e-02 3.32643062e-01 -4.76163566e-01
-1.00121605e+00 3.74422908e-01 1.97762609e-01 -9.55884397e-01
-2.48886079e-01 -1.32616445e-01 -2.83636212e-01 -4.74576265e-01
5.69568694e-01 1.65762141e-01 5.58553398e-01 -2.12869793e-01
5.35238802e-01 5.86935401e-01 1.08334219e+00 -5.29795587e-01
1.19196439e+00 7.10596681e-01 1.49955615e-01 -5.18432617e-01
-4.42051232e-01 -1.76457971e-01 -8.66385698e-01 -3.00126672e-01
1.03596878e+00 -1.07834852e+00 -1.14069188e+00 5.89646459e-01
-1.06308043e+00 3.58491272e-01 1.07180975e-01 2.56094605e-01
-3.62000138e-01 3.17120373e-01 -1.71495110e-01 -6.82878077e-01
-2.41683751e-01 -1.10152161e+00 1.17821741e+00 2.20576629e-01
-2.22256426e-02 -9.39942241e-01 -9.32762772e-02 6.31727099e-01
5.35786271e-01 6.31738305e-01 6.47086859e-01 -6.15043223e-01
-5.81578612e-01 -5.52717388e-01 -4.32703048e-01 5.51468492e-01
4.39475030e-01 3.59650373e-01 -1.45632327e+00 -4.21731263e-01
-4.87547517e-01 -5.65307200e-01 6.52982533e-01 -1.17740199e-01
9.42490280e-01 -1.77572310e-01 -1.05117992e-01 8.31859350e-01
1.42121565e+00 -5.93756698e-02 6.85584605e-01 -1.08161047e-01
6.79233551e-01 7.54914463e-01 8.03369880e-02 2.34706327e-01
3.11524659e-01 5.81103802e-01 6.35343432e-01 -2.22259000e-01
-4.21331197e-01 -4.78799284e-01 3.37325633e-01 7.48035163e-02
-7.43237361e-02 2.60731399e-01 -9.51248884e-01 7.27097690e-01
-1.33175099e+00 -1.30671024e+00 2.48736262e-01 2.31455755e+00
5.89704156e-01 -6.43295705e-01 -6.24472648e-02 1.27462342e-01
9.07709658e-01 -1.93845615e-01 -4.75833654e-01 -5.19801795e-01
-4.56115246e-01 1.34484813e-01 2.79168636e-01 5.06607711e-01
-1.18632817e+00 7.33726203e-01 6.57538939e+00 3.06906581e-01
-1.24009311e+00 3.62615176e-02 4.16417688e-01 -7.47893006e-02
-3.34453322e-02 -2.62568504e-01 -6.97748661e-01 2.62310833e-01
6.20657206e-01 -2.46057194e-02 5.07061899e-01 6.98247135e-01
-2.87390441e-01 3.17267150e-01 -1.41400743e+00 1.47969031e+00
7.76259124e-01 -1.27020824e+00 3.39788973e-01 1.63394865e-02
8.75006676e-01 -3.65187377e-01 6.70386672e-01 1.65944979e-01
3.17676872e-01 -1.64738262e+00 6.03849828e-01 8.16889822e-01
1.24730229e+00 -6.98332787e-01 7.90141642e-01 -2.40248024e-01
-1.22430372e+00 -1.30603582e-01 -1.68907031e-01 1.10250160e-01
-2.05169782e-01 -2.72827715e-01 -6.14303589e-01 3.30823511e-01
9.00533974e-01 7.26370454e-01 -7.86486626e-01 4.53342319e-01
8.39497223e-02 -1.38805687e-01 -1.58599198e-01 6.23279274e-01
-1.87859580e-01 1.63922146e-01 1.03936195e-01 9.88150895e-01
2.69091338e-01 -1.20588660e-01 -1.85696483e-01 8.83349299e-01
-4.36458796e-01 -1.28682524e-01 -1.12589610e+00 2.00228710e-02
5.60620666e-01 1.33006692e+00 2.02795178e-01 -1.28904164e-01
-5.21117628e-01 9.04309928e-01 2.70606488e-01 2.83023000e-01
-8.03714812e-01 -3.17725986e-02 1.10146689e+00 -9.15770829e-02
2.45888993e-01 1.41466454e-01 -1.06379740e-01 -1.03914642e+00
-5.09174503e-02 -9.67263877e-01 5.53321362e-01 -9.09206927e-01
-1.69239211e+00 7.56235719e-01 -1.17626965e-01 -1.23082125e+00
-1.88714370e-01 -9.38521624e-01 -6.34656608e-01 1.15580511e+00
-1.61520553e+00 -2.10148668e+00 -6.56775117e-01 1.30191064e+00
-1.01212196e-01 -6.60383165e-01 9.31748986e-01 3.48655403e-01
-2.65918165e-01 1.24699664e+00 -1.41990528e-01 6.80310309e-01
1.11977470e+00 -7.95237064e-01 -1.31188855e-01 6.89772129e-01
9.43204239e-02 8.10035944e-01 3.16847801e-01 -2.49025181e-01
-1.68963492e+00 -1.06236303e+00 8.18320096e-01 -8.15554380e-01
4.65747029e-01 -2.98746407e-01 -8.56281638e-01 9.76576090e-01
2.88998306e-01 6.53020382e-01 1.00707209e+00 -6.80294409e-02
-1.23763001e+00 -3.62087041e-01 -1.54922068e+00 1.82933420e-01
9.48711097e-01 -1.12644792e+00 -4.93362904e-01 1.83732376e-01
6.16158061e-02 -2.67422229e-01 -1.07993710e+00 4.58057374e-01
8.94087493e-01 -8.07586133e-01 1.10562086e+00 -8.44492435e-01
4.23547417e-01 -4.88631666e-01 -4.94488478e-01 -8.07944775e-01
-1.74733952e-01 -1.54496655e-01 1.42121881e-01 1.59515679e+00
2.95488417e-01 -5.76802909e-01 4.64757234e-01 7.26974010e-01
3.27667505e-01 -2.28022188e-01 -1.08144033e+00 -6.53451145e-01
3.16464692e-01 -2.50544012e-01 8.78292084e-01 1.16739631e+00
-3.25078368e-01 -6.75536543e-02 -6.35371983e-01 4.95303512e-01
6.62642300e-01 7.65331537e-02 9.70550299e-01 -1.09648514e+00
1.81814477e-01 -2.48142853e-01 -1.00720286e+00 -2.19915897e-01
5.00881553e-01 -1.17172682e+00 -3.25755000e-01 -8.69690120e-01
4.09441084e-01 -1.53666660e-01 -2.85537153e-01 7.57833660e-01
2.79260635e-01 8.84837627e-01 4.65923637e-01 1.76029891e-01
-7.33778998e-02 4.00576979e-01 9.78191733e-01 -5.44608951e-01
4.23840553e-01 -3.70559156e-01 -7.98147678e-01 6.40056312e-01
5.16692758e-01 7.88319632e-02 -2.61438251e-01 -5.30647874e-01
-1.04231812e-01 -2.61455327e-01 8.96402299e-01 -8.97887766e-01
2.68110931e-01 9.98923369e-03 9.22402322e-01 -4.44584399e-01
6.30775809e-01 -1.13918638e+00 3.77971739e-01 9.75205824e-02
-3.82479399e-01 2.00914398e-01 5.26546955e-01 3.78939301e-01
-4.28246558e-01 2.34664470e-01 1.11833286e+00 2.18027213e-04
-8.10728908e-01 4.24891829e-01 1.86528549e-01 -2.38066733e-01
1.29509318e+00 -4.69205648e-01 -4.78642106e-01 -2.90432990e-01
-8.30557466e-01 -3.03613730e-02 5.63906014e-01 9.94624078e-01
9.33789492e-01 -1.89795089e+00 -8.72372210e-01 6.29365027e-01
4.49485093e-01 -7.18041122e-01 4.49825108e-01 5.31999588e-01
-3.25728744e-01 3.76096249e-01 -7.65588820e-01 -6.37255907e-01
-1.76265526e+00 8.41463208e-01 8.50079000e-01 4.64079350e-01
-1.90371666e-02 8.61787319e-01 2.92620599e-01 -6.60790443e-01
-3.31912600e-02 1.61749825e-01 -2.51305878e-01 1.40603110e-01
8.07520449e-01 1.52187115e-02 -3.83615717e-02 -1.65556347e+00
-7.08766341e-01 1.05582702e+00 -1.38238162e-01 9.22639370e-02
1.00350547e+00 1.72886178e-02 -1.82926863e-01 8.75861198e-02
1.81085503e+00 -1.41149446e-01 -1.08010817e+00 -4.63110566e-01
-4.39178586e-01 -8.82899404e-01 -1.96885198e-01 -9.51981962e-01
-1.42188549e+00 9.61808741e-01 1.07429659e+00 -4.31965262e-01
1.11588323e+00 8.10036212e-02 2.79414684e-01 -1.13062657e-01
2.53856778e-01 -7.23524451e-01 1.18210956e-01 2.89625645e-01
1.35692155e+00 -1.72865307e+00 -1.01538233e-01 -1.73560485e-01
-7.81206727e-01 1.13471043e+00 7.98362911e-01 -6.08390160e-02
7.48482883e-01 -1.26876887e-02 3.65361303e-01 -1.74927786e-01
-4.96794224e-01 -8.59624669e-02 6.98228419e-01 1.10119975e+00
3.34604353e-01 -7.68362954e-02 3.79606009e-01 3.62982154e-01
-2.02086031e-01 -2.81088799e-01 2.09053472e-01 6.00996017e-01
5.84872887e-02 -9.17588532e-01 -7.24910736e-01 -8.27806443e-02
-3.79482150e-01 1.67658210e-01 -7.57949293e-01 9.50179338e-01
5.27318120e-01 9.87740993e-01 1.88842610e-01 -4.75768924e-01
3.17490637e-01 1.96920171e-01 6.04479849e-01 -2.12458774e-01
-5.69466889e-01 -7.61911631e-01 -3.96576494e-01 -5.80569327e-01
-4.68303382e-01 -6.89850986e-01 -9.16295946e-01 -6.10390425e-01
2.22917646e-02 -3.88196915e-01 7.30892181e-01 7.93997586e-01
5.39281845e-01 4.32385392e-02 7.57041812e-01 -8.83250713e-01
-3.65490198e-01 -8.00436378e-01 -5.11306524e-01 1.03971660e+00
7.57350922e-01 -9.67694640e-01 -4.78722781e-01 2.17139006e-01] | [13.177900314331055, 0.5961452126502991] |
b762e406-c37d-4862-a588-10ed167dd31e | lys-porting-a-twitter-sentiment-analysis | null | null | https://aclanthology.org/S14-2071 | https://aclanthology.org/S14-2071.pdf | LyS: Porting a Twitter Sentiment Analysis Approach from Spanish to English | null | ["Carlos G{\\'o}mez-Rodr{\\'\\i}guez", 'David Vilares', 'Yerai Doval', 'Miguel Hermo', 'Miguel A. Alonso'] | 2014-08-01 | null | null | null | semeval-2014-8 | ['twitter-sentiment-analysis'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.200911521911621, 3.8126556873321533] |
18f99984-9c89-4b4a-8681-460483be7d10 | pinqi-an-end-to-end-physics-informed-approach | 2306.11023 | null | https://arxiv.org/abs/2306.11023v1 | https://arxiv.org/pdf/2306.11023v1.pdf | PINQI: An End-to-End Physics-Informed Approach to Learned Quantitative MRI Reconstruction | Quantitative Magnetic Resonance Imaging (qMRI) enables the reproducible measurement of biophysical parameters in tissue. The challenge lies in solving a nonlinear, ill-posed inverse problem to obtain the desired tissue parameter maps from acquired raw data. While various learned and non-learned approaches have been proposed, the existing learned methods fail to fully exploit the prior knowledge about the underlying MR physics, i.e. the signal model and the acquisition model. In this paper, we propose PINQI, a novel qMRI reconstruction method that integrates the knowledge about the signal, acquisition model, and learned regularization into a single end-to-end trainable neural network. Our approach is based on unrolled alternating optimization, utilizing differentiable optimization blocks to solve inner linear and non-linear optimization tasks, as well as convolutional layers for regularization of the intermediate qualitative images and parameter maps. This design enables PINQI to leverage the advantages of both the signal model and learned regularization. We evaluate the performance of our proposed network by comparing it with recently published approaches in the context of highly undersampled $T_1$-mapping, using both a simulated brain dataset, as well as real scanner data acquired from a physical phantom and in-vivo data from healthy volunteers. The results demonstrate the superiority of our proposed solution over existing methods and highlight the effectiveness of our method in real-world scenarios. | ['Andreas Kofler', 'Patrick Schuenke', 'Christoph Kolbitsch', 'Felix F Zimmermann'] | 2023-06-19 | null | null | null | null | ['mri-reconstruction'] | ['computer-vision'] | [ 3.56031120e-01 1.49680272e-01 2.53833562e-01 -4.80791271e-01
-1.05000877e+00 -1.93374708e-01 1.67534634e-01 -1.80903137e-01
-6.43804908e-01 8.20135117e-01 1.88469827e-01 -1.67428240e-01
-3.62472922e-01 -1.52362645e-01 -7.61723816e-01 -9.67800915e-01
-4.35883135e-01 3.24549109e-01 -1.89595036e-02 5.14988005e-02
-1.71733869e-03 5.03465056e-01 -5.72319329e-01 -1.99755561e-02
9.80535746e-01 1.14400697e+00 3.41378957e-01 3.72425109e-01
3.08484077e-01 1.08210599e+00 -3.13810110e-02 2.14546084e-01
3.13248694e-01 -2.77657747e-01 -7.75687158e-01 -9.84128937e-02
4.65167761e-01 -4.88922238e-01 -5.41076660e-01 9.71206486e-01
8.62712204e-01 2.07845747e-01 4.67825681e-01 -6.04165733e-01
-3.69773597e-01 5.79692662e-01 -6.08402848e-01 3.99792671e-01
-1.75707608e-01 4.08772498e-01 5.31468451e-01 -7.79912293e-01
5.60000479e-01 5.76780438e-01 9.13574696e-01 3.95697355e-01
-1.50251925e+00 -6.37210310e-01 -1.88860118e-01 -1.94515602e-03
-1.26779127e+00 -5.06301284e-01 8.39526534e-01 -7.30464518e-01
6.66688681e-01 -3.22799645e-02 4.57768917e-01 6.92088127e-01
6.21022105e-01 3.90736312e-01 1.52568054e+00 -9.62415487e-02
1.65977925e-01 -6.79957122e-02 1.37152165e-01 6.88500702e-01
-1.37396404e-04 3.52117062e-01 -2.38611817e-01 -2.78915673e-01
1.04254401e+00 -1.54615462e-01 -6.80874705e-01 -5.99135399e-01
-1.34987283e+00 5.43566048e-01 6.93686008e-01 4.07420576e-01
-6.34773850e-01 3.94017965e-01 3.83697957e-01 -9.13749412e-02
4.31883067e-01 4.55702722e-01 -2.47292712e-01 1.77209467e-01
-1.04614353e+00 -5.95175289e-03 6.04511797e-01 4.45207834e-01
6.01577282e-01 2.50688493e-01 -3.25900108e-01 6.65340185e-01
4.16343570e-01 5.89366496e-01 4.85535115e-01 -9.08934712e-01
2.05679461e-01 9.42116007e-02 1.38151810e-01 -9.03636992e-01
-7.98208117e-01 -8.57805789e-01 -9.76600587e-01 1.45019680e-01
3.84642720e-01 -2.47289568e-01 -1.02597237e+00 1.93878376e+00
3.85004699e-01 5.05003333e-01 -2.82989889e-01 1.39721990e+00
8.00825715e-01 2.95276254e-01 1.29944742e-01 -3.07885826e-01
9.27179754e-01 -6.94161415e-01 -6.46039546e-01 -2.15328619e-01
5.26538908e-01 -4.21478927e-01 9.05975580e-01 2.50012875e-01
-1.27662551e+00 -2.16801494e-01 -1.00136375e+00 3.45198140e-02
2.77727813e-01 2.16146454e-01 5.59491456e-01 3.13744724e-01
-9.15806115e-01 7.46670783e-01 -1.20298815e+00 3.33921701e-01
4.65000272e-01 6.58830702e-01 -3.68906975e-01 -8.31817389e-02
-1.21446276e+00 9.65223491e-01 2.25536823e-01 6.61346078e-01
-1.18924165e+00 -1.31215179e+00 -6.42526090e-01 -1.34643987e-01
3.16045374e-01 -5.70062160e-01 9.08798397e-01 -7.16091812e-01
-1.64673936e+00 7.02594042e-01 1.27176434e-01 -5.11205137e-01
6.13440394e-01 -1.53363183e-01 -7.26577342e-02 4.29196358e-01
-6.06811792e-02 3.14572245e-01 7.38462567e-01 -1.14882576e+00
3.72656912e-01 -3.90597016e-01 -1.39541700e-01 -5.74798584e-02
7.67525136e-02 -2.86173314e-01 -1.57461777e-01 -5.01921952e-01
3.56427610e-01 -9.36087370e-01 -5.71264327e-01 1.27989843e-01
-3.79250050e-01 5.56628942e-01 2.31537491e-01 -1.15835631e+00
7.74301112e-01 -2.06542444e+00 1.19291626e-01 3.57453048e-01
5.75515330e-01 8.22889730e-02 -1.41285062e-01 -2.70561818e-02
-3.67887944e-01 -3.20917517e-01 -6.07943475e-01 -9.70389321e-02
-2.70609677e-01 -5.74252009e-03 3.16631980e-02 9.80343997e-01
3.30316238e-02 1.13932240e+00 -1.03392112e+00 -3.99899125e-01
2.49476746e-01 8.46881807e-01 -4.68469799e-01 3.91225219e-01
3.18669975e-02 1.33626199e+00 -5.99252760e-01 8.90673250e-02
8.18861306e-01 -4.00123626e-01 3.69635761e-01 -7.27501214e-01
-1.18447930e-01 2.54951894e-01 -8.69364321e-01 2.04170394e+00
-5.90787947e-01 3.29773635e-01 6.27426863e-01 -1.42625380e+00
5.84430695e-01 4.61168051e-01 1.15969777e+00 -9.50285077e-01
1.95920855e-01 4.43154365e-01 2.71993279e-01 -6.86164558e-01
-1.79078549e-01 -6.33957505e-01 2.78183758e-01 5.67045569e-01
1.98185444e-01 -2.59996563e-01 -2.43797779e-01 8.20122436e-02
1.12939548e+00 2.90300012e-01 -5.94438938e-03 -6.30773425e-01
6.30235732e-01 -1.15758665e-01 4.92771208e-01 8.09594393e-01
-2.91225612e-01 5.58218062e-01 3.48241061e-01 -3.11165065e-01
-9.55830574e-01 -1.11449337e+00 -3.85139734e-01 5.49424171e-01
-8.38351622e-02 2.33227953e-01 -7.57190943e-01 -3.92986506e-01
-6.43230453e-02 2.80192822e-01 -7.49643445e-01 -1.01923078e-01
-1.02656949e+00 -9.82379973e-01 2.54847944e-01 2.80174166e-01
3.72939736e-01 -9.07944679e-01 -6.74855828e-01 5.09709179e-01
-2.95692384e-01 -1.36989510e+00 -5.14002383e-01 3.27114075e-01
-9.65338647e-01 -8.63536179e-01 -6.20363116e-01 -4.90724802e-01
7.30547607e-01 -2.14421824e-02 1.10392046e+00 -3.64004932e-02
-6.12398624e-01 4.56303209e-01 1.96049348e-01 4.37955745e-02
-3.46422911e-01 1.40485205e-02 -1.58277780e-01 9.68351364e-02
-3.63133043e-01 -8.74030769e-01 -1.10279810e+00 2.48214155e-01
-1.00629950e+00 1.69470757e-01 8.88080001e-01 9.84647512e-01
7.96435595e-01 -3.81932318e-01 6.25995934e-01 -9.35361326e-01
4.34355408e-01 -4.85379547e-01 -5.98871112e-01 2.44447768e-01
-4.28613991e-01 3.37547660e-01 5.13022602e-01 -4.46474314e-01
-9.60606873e-01 2.79815942e-01 -2.73259282e-01 -3.51531744e-01
1.57058343e-01 7.95545578e-01 4.99620698e-02 -6.66817129e-01
6.28500283e-01 4.42481846e-01 3.12029332e-01 -3.04660201e-01
2.93048352e-01 2.23878220e-01 7.59273410e-01 -7.60931849e-01
6.51906610e-01 7.54846990e-01 1.69994026e-01 -6.24330878e-01
-8.65946412e-01 -4.35220480e-01 -8.30861926e-01 -3.52314264e-01
7.99900830e-01 -7.39845276e-01 -7.89655983e-01 3.62398326e-01
-8.66498530e-01 -6.49442196e-01 -2.76753575e-01 8.20791304e-01
-8.04589510e-01 6.50753260e-01 -7.75830150e-01 -4.59692627e-01
-7.47918427e-01 -1.46303570e+00 8.20922315e-01 -2.14240462e-01
2.24457711e-01 -1.19472826e+00 1.00170650e-01 3.23494792e-01
1.00861311e+00 5.95688581e-01 8.79784346e-01 -2.49200687e-01
-7.23794043e-01 -9.27571487e-03 -2.79446959e-01 3.94003421e-01
8.41072276e-02 -7.74765193e-01 -8.04874420e-01 -6.14608765e-01
5.58077633e-01 -5.07256567e-01 7.79705644e-01 8.67120445e-01
1.24694753e+00 -7.75084719e-02 -1.12658828e-01 9.61926460e-01
1.55460751e+00 -6.29273951e-02 5.31512141e-01 -1.32200643e-01
7.01054990e-01 3.91205847e-01 7.91426972e-02 3.11866492e-01
2.80790865e-01 6.22470438e-01 4.00680631e-01 -4.18689519e-01
-1.48724213e-01 5.96646667e-02 -2.61597615e-03 1.05571854e+00
-7.83604681e-02 4.67552960e-01 -1.01420641e+00 3.13539833e-01
-1.66140175e+00 -5.27302265e-01 9.95062292e-02 2.16996241e+00
1.05682981e+00 -1.34362474e-01 -2.60396808e-01 -3.48904163e-01
3.11808109e-01 -7.23869959e-03 -9.61498320e-01 2.71508157e-01
2.13138372e-01 5.55072725e-01 5.89621663e-01 7.11503446e-01
-1.05959606e+00 5.16445458e-01 6.38426065e+00 5.37929773e-01
-1.71460938e+00 4.99272823e-01 6.75512612e-01 -1.20180503e-01
-2.87764609e-01 -8.02628398e-02 -1.64849982e-01 1.97507933e-01
9.87463236e-01 1.54731125e-01 7.25142837e-01 3.18788439e-01
5.91136694e-01 -7.06480518e-02 -1.04614162e+00 8.19138765e-01
-5.06394096e-02 -1.29307353e+00 -4.55741972e-01 -2.99348161e-02
5.06981075e-01 3.54307681e-01 5.79316579e-02 6.80498108e-02
2.80627282e-03 -1.29627824e+00 5.18669128e-01 8.49410057e-01
9.00985003e-01 -2.22464547e-01 5.19391418e-01 3.27931255e-01
-7.85960674e-01 1.81505710e-01 -1.13342710e-01 2.16121241e-01
2.77051747e-01 8.92257869e-01 -7.48921216e-01 6.00720167e-01
3.79143625e-01 7.06014693e-01 -2.07540050e-01 9.92363393e-01
-9.73783433e-02 7.55439460e-01 -2.07813337e-01 4.48752463e-01
2.07605705e-01 -3.69737297e-01 4.60517615e-01 1.16017783e+00
1.25992447e-02 4.27592307e-01 1.78139046e-01 1.32460618e+00
1.71780035e-01 9.25729051e-02 -5.44389337e-02 3.33473273e-02
-1.22230865e-01 1.50573003e+00 -5.34245372e-01 -2.66730431e-02
-2.77805507e-01 4.75049406e-01 5.11418879e-01 5.62103331e-01
-8.40896904e-01 6.00971235e-03 1.43460289e-01 2.88336039e-01
1.08888648e-01 -5.92507422e-01 -2.56128758e-01 -1.26530671e+00
1.07187346e-01 -7.48983264e-01 -3.10018921e-04 -5.82852662e-01
-1.30276978e+00 7.01191723e-01 -2.05309652e-02 -9.13598120e-01
-2.22373739e-01 -6.05538189e-01 -3.96339148e-01 1.16289413e+00
-1.87647378e+00 -9.39488769e-01 -3.90004694e-01 5.63765049e-01
-1.89932823e-01 3.48978937e-01 5.18509626e-01 5.51829755e-01
-3.55254263e-01 3.19689602e-01 1.63456693e-01 2.32953310e-01
5.07764518e-01 -1.04809749e+00 -1.70666277e-01 5.70363522e-01
-4.43934560e-01 7.68547356e-01 6.19490027e-01 -4.74574983e-01
-1.68698382e+00 -9.01017904e-01 4.00288962e-02 7.22302422e-02
8.59888077e-01 -3.84596437e-01 -1.13364804e+00 5.15819311e-01
1.48759122e-04 7.67995119e-01 5.37445843e-01 -4.53105509e-01
5.11253215e-02 -2.10561827e-01 -1.21019733e+00 8.63087103e-02
7.74476528e-01 -5.32006621e-01 -2.77106762e-01 4.54730988e-01
5.26836157e-01 -7.81760573e-01 -1.30229473e+00 4.80180115e-01
6.04624510e-01 -6.37645960e-01 1.18156230e+00 -4.89073008e-01
4.59255457e-01 -1.53629676e-01 3.20486166e-02 -1.39529657e+00
-2.85712034e-01 -5.60526252e-01 5.45472763e-02 4.16828096e-01
2.90165365e-01 -7.44591653e-01 6.89912379e-01 8.29243362e-01
-3.89752597e-01 -9.83129323e-01 -1.22704113e+00 -4.92820710e-01
4.16266859e-01 -4.18292761e-01 1.69764608e-01 8.70029032e-01
-1.49961159e-01 7.39451721e-02 -4.95990932e-01 2.82326698e-01
1.07902503e+00 1.18585028e-01 2.79114783e-01 -8.76992285e-01
-5.32631874e-01 -4.13019136e-02 -2.24299893e-01 -1.03959823e+00
3.26649398e-01 -1.13403368e+00 4.23453718e-01 -1.35073543e+00
3.60936642e-01 -8.29117239e-01 -6.24870956e-01 2.80720919e-01
-8.89179111e-02 1.15662947e-01 -9.63178873e-02 3.14180672e-01
-2.69273549e-01 5.38743913e-01 1.76635277e+00 -1.45491332e-01
-1.40697703e-01 -3.44385833e-01 -4.54654694e-01 5.11857033e-01
5.55882514e-01 -5.27021766e-01 -3.92825574e-01 -6.52389407e-01
-3.19673456e-02 6.58689380e-01 6.84179068e-01 -1.05985153e+00
2.66823202e-01 -2.95609720e-02 3.69798928e-01 -2.09002197e-01
2.45404631e-01 -8.53858352e-01 1.84884980e-01 6.20082080e-01
-5.10328472e-01 -3.92328173e-01 2.08476111e-01 3.42632294e-01
-4.99557965e-02 -1.16692893e-02 1.01173329e+00 -2.73184389e-01
-2.03736201e-01 5.78791797e-01 8.90767500e-02 3.03801954e-01
5.73967934e-01 9.33767781e-02 1.81919467e-02 -1.75891176e-01
-1.16277230e+00 1.40408650e-01 1.89008284e-03 -1.06884934e-01
7.31377244e-01 -1.01059639e+00 -8.79217863e-01 2.10947067e-01
-3.21585745e-01 -6.74198195e-02 5.89405835e-01 1.68099761e+00
-5.09191096e-01 3.96441638e-01 -2.74983525e-01 -8.83676827e-01
-4.50828284e-01 2.97780156e-01 1.08249676e+00 -6.59028947e-01
-8.62663746e-01 4.87355918e-01 3.89804006e-01 -7.68258274e-01
3.18908766e-02 -5.21016836e-01 5.21171615e-02 -6.84668064e-01
1.87257692e-01 -1.45400120e-02 2.20479459e-01 -5.93643725e-01
-3.92922252e-01 5.63132644e-01 -4.76567149e-02 -8.76152441e-02
1.65863681e+00 -8.47335756e-02 -3.23720545e-01 2.61673719e-01
1.45442271e+00 -2.26236179e-01 -1.53037059e+00 -5.33934355e-01
-1.02714650e-01 -2.03070715e-01 5.32334089e-01 -9.91524875e-01
-1.42359173e+00 9.55158532e-01 8.59930396e-01 -4.53420132e-01
9.88134265e-01 -2.30758607e-01 9.36997831e-01 1.15924738e-01
5.20649910e-01 -7.03314483e-01 9.76537764e-02 2.71027952e-01
1.01131165e+00 -1.24292672e+00 1.52257800e-01 -5.30633807e-01
-4.03093964e-01 9.83114183e-01 3.34053278e-01 -2.61206537e-01
8.63655031e-01 5.45872092e-01 2.18025297e-01 -3.41222376e-01
-3.95085216e-01 2.67785221e-01 4.26962972e-01 3.24554712e-01
6.32570803e-01 -1.94311533e-02 -2.29375526e-01 5.13619184e-01
2.08689600e-01 5.00400245e-01 2.71821916e-01 8.63103688e-01
-1.75396457e-01 -6.42695606e-01 -1.48543283e-01 4.98840898e-01
-5.33407092e-01 -1.48337409e-01 3.09110671e-01 5.53189874e-01
-1.32720387e-02 3.52410734e-01 -3.07357013e-01 1.45039216e-01
2.81445265e-01 -2.14618921e-01 7.35763371e-01 -4.03978080e-01
-6.97059274e-01 1.10679179e-01 -3.94434512e-01 -7.37740040e-01
-4.46581751e-01 -5.16227245e-01 -1.52148211e+00 1.87019154e-01
-2.59100884e-01 9.98047292e-02 7.78581023e-01 1.03535771e+00
3.57061356e-01 7.00932622e-01 5.32095432e-01 -9.78094220e-01
-8.39031696e-01 -9.44264352e-01 -6.62895977e-01 4.41780329e-01
4.84752178e-01 -7.11979687e-01 -2.73053169e-01 -1.46703973e-01] | [13.482513427734375, -2.4457414150238037] |
eed67c5e-4835-4b4f-9ad9-1d5739cf75f3 | model-free-generative-replay-for-lifelong | 2208.05056 | null | https://arxiv.org/abs/2208.05056v2 | https://arxiv.org/pdf/2208.05056v2.pdf | Model-Free Generative Replay for Lifelong Reinforcement Learning: Application to Starcraft-2 | One approach to meet the challenges of deep lifelong reinforcement learning (LRL) is careful management of the agent's learning experiences, to learn (without forgetting) and build internal meta-models (of the tasks, environments, agents, and world). Generative replay (GR) is a biologically inspired replay mechanism that augments learning experiences with self-labelled examples drawn from an internal generative model that is updated over time. We present a version of GR for LRL that satisfies two desiderata: (a) Introspective density modelling of the latent representations of policies learned using deep RL, and (b) Model-free end-to-end learning. In this paper, we study three deep learning architectures for model-free GR, starting from a na\"ive GR and adding ingredients to achieve (a) and (b). We evaluate our proposed algorithms on three different scenarios comprising tasks from the Starcraft-2 and Minigrid domains. We report several key findings showing the impact of the design choices on quantitative metrics that include transfer learning, generalization to unseen tasks, fast adaptation after task change, performance wrt task expert, and catastrophic forgetting. We observe that our GR prevents drift in the features-to-action mapping from the latent vector space of a deep RL agent. We also show improvements in established lifelong learning metrics. We find that a small random replay buffer significantly increases the stability of training. Overall, we find that "hidden replay" (a well-known architecture for class-incremental classification) is the most promising approach that pushes the state-of-the-art in GR for LRL and observe that the architecture of the sleep model might be more important for improving performance than the types of replay used. Our experiments required only 6% of training samples to achieve 80-90% of expert performance in most Starcraft-2 scenarios. | ['Ajay Divakaran', 'Michael Piacentino', 'Indranil Sur', 'Abrar Rahman', 'Jesse Hostetler', 'Aswin Raghavan', 'Zachary Daniels'] | 2022-08-09 | null | null | null | null | ['starcraft'] | ['playing-games'] | [-1.22741237e-01 2.98441797e-02 -9.62765291e-02 -3.70991393e-03
-3.41436058e-01 -5.36039948e-01 9.28703845e-01 -1.34117901e-01
-5.91604412e-01 1.00493407e+00 6.92568421e-02 -5.82543127e-02
-2.27499038e-01 -5.70889354e-01 -1.07090616e+00 -1.06472230e+00
-3.17353249e-01 5.97383857e-01 3.02990586e-01 -3.36667806e-01
9.17018950e-02 5.66877484e-01 -1.78091550e+00 1.58464178e-01
7.50701010e-01 6.79175556e-01 4.48513776e-01 8.41645479e-01
1.63291261e-01 1.02530527e+00 -7.55230665e-01 1.52674660e-01
2.98091710e-01 -6.18160427e-01 -7.01616287e-01 4.82952502e-03
6.25547767e-02 -2.10426390e-01 -3.55865717e-01 5.69749475e-01
6.59459054e-01 3.99233401e-01 6.52251184e-01 -1.25408912e+00
-5.40584922e-01 4.36339617e-01 -2.06055716e-01 3.24671149e-01
-1.25860050e-01 8.16505611e-01 4.09005105e-01 -4.86099958e-01
5.15929639e-01 1.13063765e+00 7.52763152e-01 9.38759327e-01
-1.16273749e+00 -5.69066346e-01 2.95566499e-01 -2.44163461e-02
-1.03142858e+00 -5.51415443e-01 3.65105122e-01 -3.46189171e-01
1.39014935e+00 -1.24615632e-01 7.57615268e-01 1.65383637e+00
8.77056718e-01 6.40253842e-01 1.23865509e+00 -4.49074924e-01
8.29101145e-01 2.00416520e-01 2.32682247e-02 7.56949186e-01
3.09074372e-01 5.71567535e-01 -7.83813000e-01 -2.51391202e-01
8.76206040e-01 3.92899252e-02 -1.34362847e-01 -6.89246356e-01
-9.77151394e-01 6.74492180e-01 4.11371350e-01 1.64926320e-01
-4.03133065e-01 6.80149615e-01 3.47762853e-01 5.79506755e-01
4.06076163e-01 7.59778559e-01 -5.90162873e-01 -1.86385363e-01
-6.50562465e-01 2.90236950e-01 6.93636775e-01 7.16120601e-01
8.23885322e-01 5.63566864e-01 -2.90075690e-01 6.09751165e-01
1.85058601e-02 3.64471346e-01 1.16166651e+00 -1.05087984e+00
-8.46524313e-02 2.90652692e-01 2.40221485e-01 -2.05750659e-01
-6.08148634e-01 -1.00177264e+00 -3.07408214e-01 5.47070742e-01
1.95664436e-01 -4.44907248e-01 -1.12473273e+00 2.15483332e+00
1.28539115e-01 3.65553856e-01 5.73539138e-01 5.02109945e-01
3.57907087e-01 6.62965834e-01 1.41687974e-01 -2.57608533e-01
9.40766692e-01 -1.06970572e+00 -3.11883181e-01 -5.87681711e-01
7.09393144e-01 -8.99782628e-02 1.42225480e+00 4.68201160e-01
-1.00729799e+00 -7.75179744e-01 -1.27634597e+00 3.60988200e-01
-4.12501544e-01 -7.51403347e-02 6.25482976e-01 5.08433759e-01
-1.21096444e+00 9.27667677e-01 -1.19895661e+00 -5.99775910e-01
1.73593923e-01 3.98105085e-01 -9.41382125e-02 8.64548236e-02
-1.00107384e+00 1.04127777e+00 5.09016335e-01 -4.27505016e-01
-1.77547514e+00 -6.13178790e-01 -4.03161764e-01 9.45820585e-02
3.47021490e-01 -1.00621796e+00 1.33869374e+00 -1.11053908e+00
-1.80443454e+00 5.07007957e-01 1.56113699e-01 -9.10315752e-01
3.05244565e-01 -5.39839685e-01 -1.56986251e-01 -1.63947612e-01
-2.74560183e-01 7.77894795e-01 1.09563351e+00 -1.46918547e+00
-4.26407933e-01 -3.39411288e-01 -6.28363192e-02 3.80377799e-01
-2.64880329e-01 -6.91087663e-01 2.77252719e-02 -5.34031510e-01
-3.52968872e-01 -1.18002439e+00 -8.06514174e-02 -4.30249274e-01
3.41085196e-01 -1.92380562e-01 7.71514475e-01 -2.71111310e-01
8.21282148e-01 -2.23218393e+00 2.21426129e-01 -3.79723221e-01
-2.72815861e-02 3.24611694e-01 -3.56296837e-01 5.92363894e-01
8.25225413e-02 -2.49020278e-01 -5.37312143e-02 -5.43350160e-01
-1.26089543e-01 6.79094553e-01 -5.92254817e-01 2.70783961e-01
-4.43623066e-02 1.07597411e+00 -9.59281325e-01 1.62085146e-01
6.03160029e-03 3.94261807e-01 -4.59291607e-01 3.61428112e-01
-6.84272230e-01 4.60450232e-01 -5.31631932e-02 1.24486446e-01
1.74078792e-01 -2.15102986e-01 1.15203030e-01 2.69705415e-01
-2.46752761e-02 2.30780631e-01 -7.53120184e-01 2.05441475e+00
-6.15585387e-01 3.53510469e-01 -2.48932391e-01 -6.90259576e-01
8.87135088e-01 2.50072241e-01 2.31264636e-01 -8.39715004e-01
-5.80548421e-02 6.72456473e-02 1.10665351e-01 -2.41702452e-01
3.34183395e-01 -1.47538194e-02 -3.17379795e-02 7.05153584e-01
7.54540265e-01 7.07411692e-02 3.22476961e-02 7.77853131e-02
1.35538673e+00 6.31666660e-01 2.13437304e-01 -4.99930620e-01
-6.22105077e-02 2.97562033e-03 6.18131757e-01 1.21337533e+00
-6.35890514e-02 1.98835194e-01 3.64930212e-01 -7.87892401e-01
-9.77151632e-01 -1.14433002e+00 2.98903644e-01 1.49321353e+00
-2.07253799e-01 -2.81089544e-01 -7.67058790e-01 -6.66903734e-01
-3.48021761e-02 1.18782318e+00 -9.08275723e-01 -9.33345973e-01
-5.32174528e-01 -1.03608990e+00 4.42605913e-01 4.37920511e-01
5.44130981e-01 -1.41068220e+00 -1.26637030e+00 3.06717932e-01
3.75101656e-01 -6.60076439e-01 -1.00356936e-01 9.24518883e-01
-1.12810886e+00 -7.44886220e-01 -5.65902889e-01 -4.66627091e-01
4.94916230e-01 1.43547297e-01 1.15189004e+00 -8.11644867e-02
-1.61764324e-01 5.95270753e-01 -3.17462534e-01 -3.47240835e-01
-6.26915991e-01 2.77593374e-01 4.77012128e-01 -3.57478499e-01
-4.94618118e-02 -8.72330844e-01 -6.22261047e-01 1.40587270e-01
-8.33037615e-01 3.85454856e-02 8.75795841e-01 9.15687025e-01
3.75008076e-01 1.07012458e-01 7.04613745e-01 -8.40081871e-01
6.91973090e-01 -5.05116343e-01 -4.99704689e-01 2.75353521e-01
-9.95301068e-01 6.84543490e-01 6.65243924e-01 -7.48027146e-01
-1.01391411e+00 -1.18363135e-01 8.09250772e-02 -6.14312112e-01
-1.52141377e-01 7.30461180e-02 1.48847222e-01 -1.42512051e-02
1.10887229e+00 6.25496805e-01 2.02776298e-01 -4.56687361e-01
4.61959630e-01 1.13154076e-01 4.52772349e-01 -8.62320602e-01
4.13164318e-01 2.82200068e-01 -1.87607542e-01 -5.71538746e-01
-8.38360190e-01 8.88156965e-02 -3.26590776e-01 -7.93176666e-02
4.94217038e-01 -9.23793733e-01 -6.22303426e-01 5.44846892e-01
-7.47833192e-01 -1.22240114e+00 -9.66025472e-01 3.47875953e-01
-9.77460682e-01 -2.09559679e-01 -6.20006621e-01 -8.62736344e-01
-3.86395276e-01 -1.03433478e+00 7.11882234e-01 3.67809653e-01
-2.78630164e-02 -1.03969932e+00 7.27051735e-01 -2.58735240e-01
6.55452549e-01 1.43168435e-01 1.11831260e+00 -6.10628426e-01
-2.53121436e-01 3.40560883e-01 4.40665901e-01 3.46693248e-01
1.26087636e-01 -4.71845925e-01 -1.27076304e+00 -8.28987062e-01
2.72059560e-01 -6.38198316e-01 1.14231086e+00 3.64026129e-01
7.79096603e-01 -4.45291221e-01 -2.59146035e-01 5.77615440e-01
1.31978452e+00 2.71639436e-01 5.62749028e-01 4.68516886e-01
2.57184118e-01 2.83589840e-01 3.25630069e-01 5.15486300e-01
1.54698372e-01 4.10009414e-01 5.59122264e-01 8.94817412e-02
-3.05079937e-01 -4.43896562e-01 9.51106012e-01 5.49154639e-01
9.71623287e-02 -2.51067609e-01 -7.69191921e-01 5.04507720e-01
-2.05836654e+00 -8.28957438e-01 6.58859313e-01 2.41209412e+00
8.29218030e-01 3.97618979e-01 2.95651078e-01 -2.29316771e-01
1.41368926e-01 -3.40014859e-03 -1.07093930e+00 -4.27710533e-01
6.90918341e-02 3.50866675e-01 4.59716171e-01 4.57909167e-01
-7.04076946e-01 1.14024174e+00 6.68672132e+00 6.34100795e-01
-1.23890829e+00 4.32461023e-01 5.27600348e-01 -3.79050940e-01
4.29405496e-02 -4.46093269e-02 -1.01370740e+00 4.51205462e-01
1.38150692e+00 6.26687333e-02 7.62635827e-01 9.89033282e-01
-1.06413942e-02 -1.96425125e-01 -1.17990661e+00 6.72866464e-01
1.11391447e-01 -1.24810064e+00 9.32199433e-02 2.68841330e-02
8.43555868e-01 4.80200648e-01 2.95692265e-01 9.66610789e-01
9.29823756e-01 -9.50833917e-01 6.86106741e-01 7.83426583e-01
4.90424097e-01 -7.51558900e-01 4.35453892e-01 7.22973347e-01
-5.89037299e-01 -4.03947383e-01 -4.47726786e-01 -7.14735910e-02
-3.41365695e-01 2.96493202e-01 -7.92869210e-01 2.52249360e-01
7.34771967e-01 5.08954942e-01 -8.13733816e-01 7.15360582e-01
-1.94310308e-01 7.06039727e-01 -2.06509829e-01 7.76655525e-02
3.31970870e-01 2.61785001e-01 4.70306545e-01 9.92484450e-01
3.32992166e-01 -2.88074374e-01 1.30533472e-01 8.27368200e-01
1.14955656e-01 -5.26885092e-01 -7.28019118e-01 7.05654398e-02
3.32661718e-01 1.05123401e+00 -6.14613712e-01 -1.90149069e-01
4.37765494e-02 1.00384116e+00 5.75159669e-01 5.97831845e-01
-7.02816606e-01 1.61736801e-01 6.12739563e-01 1.91416025e-01
3.71541113e-01 -5.47250807e-01 1.16220064e-01 -9.82451260e-01
-4.57360953e-01 -9.88155067e-01 3.01761240e-01 -8.65863025e-01
-1.10335922e+00 7.49148965e-01 6.36595264e-02 -7.59939015e-01
-6.88795745e-01 -4.22132850e-01 -6.08185768e-01 5.11149704e-01
-1.32155073e+00 -8.53361368e-01 -3.28828514e-01 6.23154759e-01
7.08975911e-01 -5.73738098e-01 9.68655109e-01 -3.65754277e-01
-5.34813106e-01 4.79411453e-01 4.06708151e-01 -5.17904580e-01
6.34911537e-01 -1.25858641e+00 5.90218484e-01 6.69025421e-01
3.23021889e-01 6.86248481e-01 8.02904248e-01 -6.22623205e-01
-1.43331265e+00 -1.11964464e+00 5.23188822e-02 -5.43903053e-01
2.65001893e-01 -5.71002960e-01 -9.70044971e-01 8.25833917e-01
3.85425657e-01 -8.49542022e-02 4.12152946e-01 8.26915577e-02
-1.43490702e-01 -2.80289352e-01 -9.36511040e-01 6.12555265e-01
9.88248944e-01 -1.63592041e-01 -5.52057207e-01 2.65222698e-01
8.05369377e-01 -2.41834953e-01 -4.34934616e-01 2.58212000e-01
4.62644905e-01 -1.07148790e+00 7.36022472e-01 -7.26495564e-01
-1.25080526e-01 -1.58535764e-01 5.77240549e-02 -1.74528968e+00
-5.43225825e-01 -8.78579378e-01 -5.14450729e-01 7.95032740e-01
1.97045103e-01 -7.55988061e-01 7.50707507e-01 1.81309462e-01
-3.04426849e-01 -8.38166475e-01 -8.24899912e-01 -1.17495143e+00
2.71437973e-01 -3.34262289e-02 3.92594367e-01 5.82792521e-01
-3.27311546e-01 5.33690155e-01 -4.29099768e-01 -4.05353904e-02
4.24674273e-01 -1.36154503e-01 8.68297398e-01 -1.04615057e+00
-8.21760535e-01 -1.47382185e-01 9.04613733e-02 -9.08115566e-01
1.73163682e-01 -7.56490052e-01 4.89637107e-02 -1.35421824e+00
1.44096702e-01 -5.43403089e-01 -5.89142740e-01 7.43080854e-01
1.38951644e-01 -2.74647117e-01 1.17368340e-01 4.23501611e-01
-7.02580988e-01 9.19849336e-01 9.42804933e-01 1.71835676e-01
-4.98210371e-01 -8.81782174e-02 -5.64186752e-01 6.09571278e-01
1.05583978e+00 -6.87679946e-01 -9.25640821e-01 -4.59509492e-01
2.91688979e-01 -1.78019464e-01 4.21059787e-01 -1.43708658e+00
1.35837391e-01 -2.30905220e-01 5.16541183e-01 -4.01870683e-02
4.83346403e-01 -4.67241943e-01 1.45718426e-01 9.27486598e-01
-4.23805863e-01 3.05519611e-01 5.03394842e-01 6.78339541e-01
5.13046443e-01 -3.22791636e-01 9.11902010e-01 -4.39226061e-01
-7.03837693e-01 -2.60642301e-02 -4.97425377e-01 9.48894024e-02
8.85138631e-01 -4.28465232e-02 -6.32502437e-01 -2.08678171e-01
-8.12723577e-01 8.01382959e-02 5.69337845e-01 5.14014721e-01
3.77225727e-01 -9.67138171e-01 -5.36411941e-01 3.86753559e-01
-1.04117312e-01 -2.69717574e-01 -1.69431940e-02 4.72262591e-01
-2.48827189e-01 3.12447131e-01 -6.28693819e-01 -5.82109153e-01
-6.02069557e-01 7.08874106e-01 5.91712117e-01 -5.94062984e-01
-6.46628499e-01 7.54412413e-01 2.93658048e-01 -3.25314105e-01
3.21345359e-01 -2.90626347e-01 6.82056174e-02 -3.16375911e-01
3.52560580e-01 2.60174543e-01 1.73223436e-01 6.04577176e-02
-2.18698993e-01 5.62689342e-02 -4.04957354e-01 -3.44989002e-01
1.42660463e+00 1.42712697e-01 4.11409885e-01 9.07968283e-01
5.73993385e-01 -4.74852145e-01 -1.98758972e+00 -1.55116066e-01
-2.18225762e-01 1.06244288e-01 1.48722246e-01 -1.10767508e+00
-6.92083299e-01 8.67292166e-01 9.78243649e-01 -7.08790645e-02
8.83580863e-01 -1.41312286e-01 5.58563352e-01 6.36804402e-01
6.82021797e-01 -1.15447724e+00 7.08756328e-01 6.13411188e-01
9.92132485e-01 -8.51088405e-01 2.95344144e-02 6.47513092e-01
-7.58982599e-01 8.37178707e-01 8.19022179e-01 -4.38789159e-01
3.46177936e-01 3.63646686e-01 -2.74268448e-01 -9.53513458e-02
-1.44242442e+00 -1.34659812e-01 -1.26954690e-01 7.56471038e-01
5.33083174e-03 -2.65119016e-01 2.44419083e-01 4.82155979e-01
-1.33391276e-01 1.14593133e-02 3.28583837e-01 1.13191617e+00
-8.02612364e-01 -9.63052928e-01 -3.20234373e-02 2.06264079e-01
-2.39542103e-03 3.59139517e-02 -2.39562452e-01 8.62748086e-01
3.88929471e-02 5.66026568e-01 -6.65187165e-02 -2.97228009e-01
8.48217588e-03 4.15848672e-01 8.36671054e-01 -7.63399422e-01
-7.63400614e-01 -1.48690045e-01 -2.48764575e-01 -5.05986750e-01
-4.28173020e-02 -5.74296594e-01 -1.23587203e+00 -1.86017334e-01
-5.64489998e-02 2.17495471e-01 6.05281949e-01 9.38483715e-01
7.86993444e-01 7.76550114e-01 3.42006207e-01 -8.57091606e-01
-7.50266194e-01 -1.01285541e+00 -4.99535501e-01 5.26251942e-02
3.12845320e-01 -9.99559581e-01 -3.25543523e-01 6.31780848e-02] | [4.248187065124512, 1.4791216850280762] |
bb1bf2a5-84d6-4982-8786-970534bc8286 | hdr-vdp-3-a-multi-metric-for-predicting-image | 2304.13625 | null | https://arxiv.org/abs/2304.13625v1 | https://arxiv.org/pdf/2304.13625v1.pdf | HDR-VDP-3: A multi-metric for predicting image differences, quality and contrast distortions in high dynamic range and regular content | High-Dynamic-Range Visual-Difference-Predictor version 3, or HDR-VDP-3, is a visual metric that can fulfill several tasks, such as full-reference image/video quality assessment, prediction of visual differences between a pair of images, or prediction of contrast distortions. Here we present a high-level overview of the metric, position it with respect to related work, explain the main differences compared to version 2.2, and describe how the metric was adapted for the HDR Video Quality Measurement Grand Challenge 2023. | ['Param Hanji', 'Dounia Hammou', 'Rafal K. Mantiuk'] | 2023-04-26 | null | null | null | null | ['video-quality-assessment', 'video-quality-assessment'] | ['computer-vision', 'time-series'] | [ 2.57254034e-01 -6.17006242e-01 1.47178173e-01 -4.84843016e-01
-7.35500395e-01 -3.68301809e-01 3.37877542e-01 -1.04101703e-01
-3.24334390e-02 5.12774408e-01 3.16658229e-01 -2.47386277e-01
-1.43765539e-01 -1.82272524e-01 -2.74890333e-01 -2.87091881e-01
-4.55387622e-01 -3.64772975e-01 5.63998818e-01 -3.03114116e-01
5.00643492e-01 7.61509657e-01 -1.86527193e+00 6.01425052e-01
4.58695561e-01 1.47131050e+00 3.02688807e-01 1.21992457e+00
5.30521750e-01 1.03003323e+00 -6.71510577e-01 -2.94144869e-01
4.60551053e-01 -7.18289375e-01 -7.50218689e-01 1.37270391e-01
7.54717171e-01 -8.09472740e-01 -8.31691206e-01 9.09273446e-01
6.66119039e-01 1.31714717e-01 5.69129884e-01 -1.50741851e+00
-5.48225224e-01 -3.54749292e-01 -6.14361048e-01 9.86335814e-01
1.05331254e+00 4.12070334e-01 6.81114793e-01 -8.77810776e-01
7.75151253e-01 1.28025758e+00 5.72215676e-01 4.21996087e-01
-1.23466837e+00 -1.81898817e-01 -3.49603534e-01 9.79754388e-01
-1.35061264e+00 -7.10229218e-01 2.68490672e-01 -5.86164773e-01
1.19432139e+00 6.05210960e-01 7.36985683e-01 6.07424498e-01
6.56632841e-01 2.82675982e-01 1.33879924e+00 -8.72668028e-02
-5.53241149e-02 -3.25665176e-01 -3.14443558e-01 2.43493468e-01
-1.89653084e-01 5.30017436e-01 -4.49887872e-01 8.32727849e-02
8.90680969e-01 -5.58679819e-01 -8.08830380e-01 -3.75365496e-01
-1.13108349e+00 1.02595814e-01 2.23485932e-01 5.99790700e-02
-8.56610388e-02 -4.54468764e-02 4.88334239e-01 9.43345129e-01
2.59182960e-01 1.12077706e-01 -2.18820021e-01 -4.31617200e-01
-1.39356601e+00 1.93066061e-01 2.09805816e-01 1.01757753e+00
4.70210880e-01 2.72920609e-01 -4.88410741e-01 9.49210405e-01
4.97062892e-01 5.66192091e-01 1.53408393e-01 -1.66408515e+00
2.13639855e-01 -2.41837770e-01 1.73737153e-01 -1.03716433e+00
-2.12332815e-01 3.20044845e-01 -8.67847860e-01 1.03011703e+00
4.27526943e-02 3.12685996e-01 -7.50175118e-01 1.06247997e+00
-2.07647502e-01 -1.64529398e-01 -1.68156952e-01 1.34692645e+00
1.16573942e+00 7.63379455e-01 -2.43541792e-01 -6.58244669e-01
9.74804580e-01 -7.84742892e-01 -7.95935273e-01 8.82059112e-02
-1.95774943e-01 -1.01592731e+00 9.07642066e-01 5.67424178e-01
-1.68233097e+00 -1.15312517e+00 -1.40493679e+00 -3.42346191e-01
3.41988960e-03 -4.14482027e-01 -2.43303597e-01 7.14434505e-01
-1.77773321e+00 8.26186180e-01 -1.98565170e-01 -1.73767403e-01
-5.44421263e-02 1.06307520e-02 -3.75864297e-01 -2.68621176e-01
-1.33804500e+00 8.61282885e-01 2.89814658e-02 -3.15825790e-01
-9.26756024e-01 -9.31398451e-01 -8.01090419e-01 -1.35704681e-01
-2.26656929e-01 -5.66418886e-01 1.04222310e+00 -9.45916176e-01
-1.52200162e+00 1.26190698e+00 -4.48699221e-02 -2.63503373e-01
6.84716344e-01 1.62462011e-01 -1.04966104e+00 4.80096430e-01
-3.55862826e-01 8.44925046e-01 8.25779140e-01 -1.32365596e+00
-8.93058598e-01 -3.03132795e-02 7.80536886e-03 4.22190338e-01
5.88996887e-01 4.22573477e-01 -7.49257982e-01 -6.26035750e-01
-3.86805721e-02 -4.49381262e-01 1.07408285e-01 4.53441888e-01
6.03136420e-02 3.97038758e-01 8.23476017e-01 -1.04630435e+00
1.41282284e+00 -2.32004690e+00 -7.15874359e-02 6.99653253e-02
6.23027205e-01 3.39359760e-01 -3.00637662e-01 1.06110170e-01
-4.30531472e-01 3.42003107e-02 7.96154663e-02 2.89962403e-02
-1.13680296e-01 -3.13132405e-01 -1.28397703e-01 6.30786300e-01
-1.14449732e-01 7.45654702e-01 -8.70247960e-01 -4.80340332e-01
8.36252868e-01 7.05843270e-01 -1.72428980e-01 3.90221328e-01
2.61452973e-01 4.66359317e-01 5.72517097e-01 7.15941668e-01
1.05897892e+00 -1.82328820e-02 9.43472702e-03 -6.32586122e-01
-2.45721221e-01 -4.21598330e-02 -1.33823073e+00 1.24285614e+00
-1.95515946e-01 1.57610106e+00 -1.06636032e-01 -1.52857572e-01
7.51105368e-01 3.61660957e-01 6.57940269e-01 -1.46412575e+00
-2.85217375e-01 2.64687061e-01 -4.31992970e-02 -5.45495033e-01
7.47071087e-01 3.19863468e-01 3.86092514e-01 -1.06800012e-01
-9.72850695e-02 -2.02884942e-01 4.93206859e-01 1.02073759e-01
1.17995334e+00 2.07470767e-02 6.70199931e-01 -5.57192303e-02
8.38694930e-01 -6.38043880e-01 3.63057464e-01 5.76056838e-01
-8.19984615e-01 1.34593904e+00 5.18382788e-01 -3.17502648e-01
-1.57408500e+00 -1.54128075e+00 -1.93354636e-01 6.49769425e-01
6.56216979e-01 -6.30337834e-01 -2.45997041e-01 -1.89062998e-01
-2.28163004e-01 3.58542949e-01 -4.58982855e-01 1.53854415e-01
-3.10470045e-01 -9.09843296e-02 2.78437734e-01 2.00840741e-01
6.18551314e-01 -7.98959017e-01 -8.69963586e-01 -1.30612507e-01
-3.59726131e-01 -1.31981814e+00 -6.09826028e-01 -3.32068980e-01
-7.47299194e-01 -1.19396365e+00 -1.07793939e+00 -4.61050898e-01
-8.71770084e-02 5.65523803e-01 1.80768454e+00 8.04156214e-02
-5.37957907e-01 7.05045342e-01 -4.11767364e-01 5.74557066e-01
-6.44727707e-01 -1.03527367e+00 1.23560084e-02 -4.33750153e-01
-8.94706473e-02 -4.99660730e-01 -1.15037453e+00 7.56061614e-01
-8.33994806e-01 8.99250433e-03 1.69318587e-01 3.89182895e-01
9.09199595e-01 2.04800908e-02 2.09067129e-02 -1.05272330e-01
4.51121092e-01 -9.66932252e-02 -6.16775870e-01 2.93453723e-01
-8.32330942e-01 -3.92690778e-01 3.17670316e-01 -1.79684535e-02
-8.49946499e-01 -6.49973273e-01 -3.80389065e-01 -5.86098433e-01
1.55760357e-02 -3.45678508e-01 -3.33768010e-01 -3.87414515e-01
5.91041148e-01 2.10201398e-01 -2.39987552e-01 -1.67534024e-01
2.73433238e-01 6.52064860e-01 7.89169669e-01 1.59124896e-01
5.94719708e-01 2.29091212e-01 1.91714361e-01 -6.87385917e-01
-8.80831853e-03 -5.42354524e-01 -4.31134671e-01 -7.61904061e-01
7.71759331e-01 -1.00657737e+00 -6.73739374e-01 5.64419985e-01
-9.83638048e-01 -4.41721022e-01 -3.99281055e-01 3.51833791e-01
-1.06858277e+00 6.95200741e-01 -7.44856358e-01 -3.00214887e-01
-2.88688123e-01 -1.51257348e+00 7.77270257e-01 1.66976228e-01
1.33965909e-01 -9.15263295e-01 2.77809530e-01 1.27183974e-01
7.00197458e-01 2.50017464e-01 6.35962188e-01 2.34742805e-01
-4.05835807e-01 4.65947062e-01 -7.36259520e-01 8.74544084e-01
6.13418110e-02 1.99276909e-01 -8.21254551e-01 -5.62181294e-01
-5.91838583e-02 -8.34630132e-02 3.51469606e-01 6.69587851e-01
1.09494388e+00 1.11094117e-01 1.26843512e-01 9.81271803e-01
1.74222481e+00 4.17094260e-01 1.48439586e+00 3.63700718e-01
3.39462221e-01 1.41891450e-01 8.94488096e-01 6.49986327e-01
1.05247483e-01 1.22026777e+00 4.56191391e-01 -1.92983851e-01
-8.24014246e-01 1.64036900e-01 4.47211683e-01 6.29937530e-01
-2.84313560e-01 -3.88029277e-01 -5.66362619e-01 3.65446806e-02
-1.14129674e+00 -1.07087862e+00 -4.25797462e-01 2.24814630e+00
5.22061825e-01 -8.23958516e-02 2.95960307e-01 4.48099524e-01
9.79191840e-01 4.23214912e-01 -4.70083058e-01 -7.13365734e-01
-4.62594062e-01 -1.56870559e-01 5.02352595e-01 5.09985447e-01
-6.83494210e-01 3.16665649e-01 8.45190907e+00 7.77603209e-01
-8.50874186e-01 8.76836702e-02 8.76240432e-01 -1.18655242e-01
1.12515509e-01 -1.51551560e-01 -2.72399038e-01 5.95233619e-01
1.35152197e+00 -1.58304751e-01 5.79741061e-01 5.27574420e-01
6.63262904e-01 -3.83548498e-01 -1.19580114e+00 1.72599602e+00
2.73416698e-01 -1.09826016e+00 6.09512143e-02 7.81132653e-02
5.54920375e-01 2.85906065e-02 4.78646189e-01 -8.27193931e-02
-4.40476447e-01 -9.47509766e-01 8.18413258e-01 5.77826262e-01
1.69826531e+00 -5.27338564e-01 5.02413690e-01 -6.16151392e-01
-1.37583470e+00 2.17659101e-01 -4.16376531e-01 3.50911319e-01
5.02326012e-01 4.76209909e-01 1.01408713e-01 4.68930334e-01
1.20724964e+00 8.47669780e-01 -9.23712254e-01 1.55018938e+00
1.09801792e-01 2.24053916e-02 3.66667122e-01 9.28772807e-01
-3.10758293e-01 8.48026127e-02 8.13326836e-01 1.41734791e+00
4.85485584e-01 7.82015324e-02 -3.83979946e-01 4.63556588e-01
2.36055896e-01 -5.26872985e-02 -2.54876316e-01 4.79860812e-01
2.53735512e-01 1.04793572e+00 -1.81510225e-01 -3.43027890e-01
-5.57307065e-01 1.54166853e+00 -4.25549835e-01 2.21170962e-01
-1.16269469e+00 -6.35540605e-01 9.65603948e-01 4.18303549e-01
1.73843965e-01 -6.31659329e-02 1.52616650e-01 -9.19080436e-01
-1.41731724e-01 -1.22992730e+00 1.80034116e-01 -1.68286300e+00
-9.87925470e-01 7.94095516e-01 1.56525314e-01 -1.90778112e+00
-4.04065251e-01 -6.70760036e-01 -1.25970080e-01 8.89008820e-01
-1.66268277e+00 -2.72388220e-01 -8.25561821e-01 7.31505930e-01
6.77474558e-01 -6.08664975e-02 3.15287173e-01 6.91656709e-01
-9.87244770e-03 6.50424898e-01 1.90238297e-01 -2.32014462e-01
1.04218435e+00 -1.30788839e+00 4.94918287e-01 1.01013041e+00
-3.95902216e-01 -3.24642628e-01 1.10033095e+00 -2.03260764e-01
-1.13300896e+00 -7.28006423e-01 4.75721151e-01 -1.41043991e-01
3.26413125e-01 9.38977003e-02 -8.18096995e-01 2.01713562e-01
4.48213935e-01 5.37333608e-01 4.92601693e-01 -6.53507054e-01
-4.08349335e-01 -3.32378566e-01 -1.59071803e+00 5.69055378e-01
1.20712292e+00 -4.06364471e-01 -1.59096032e-01 -1.68436989e-01
7.30061650e-01 -7.28345394e-01 -1.47674906e+00 3.49540442e-01
6.50880992e-01 -1.96571469e+00 1.23469067e+00 1.54846787e-01
4.70365644e-01 -5.73182225e-01 -5.54019630e-01 -1.12746954e+00
-6.18647456e-01 -8.14077556e-01 -3.66694778e-01 9.48852420e-01
-1.03542738e-01 -4.52931039e-02 2.06291750e-01 2.87223756e-01
-1.14091925e-01 -2.23471537e-01 -1.00865364e+00 -9.65397596e-01
-4.70990121e-01 -3.12532932e-01 2.73095250e-01 3.82611662e-01
1.03358170e-02 -5.81570864e-02 -6.41540527e-01 -6.42700419e-02
7.09227800e-01 -3.01663756e-01 4.70014066e-01 -6.79969192e-01
-4.29152280e-01 -4.66647595e-01 -1.29901326e+00 -1.34855008e+00
-5.48705816e-01 -2.64578372e-01 -2.50878762e-02 -1.59852564e+00
1.61724746e-01 1.84533093e-02 -3.61035794e-01 -3.41287702e-01
4.26694565e-02 8.10834229e-01 4.79194999e-01 3.11620086e-01
-1.07770324e+00 2.26606250e-01 1.16818035e+00 -5.44578694e-02
-5.47476038e-02 -1.74610317e-01 -2.70265490e-01 2.25869775e-01
4.35103863e-01 1.00958765e-01 -3.40584874e-01 -2.34294102e-01
1.95118427e-01 5.81241786e-01 4.13863510e-01 -1.64590561e+00
-1.79863274e-01 -1.95173576e-01 7.77010262e-01 -6.62647665e-01
3.57987404e-01 -9.69489694e-01 6.25883639e-01 4.18682814e-01
-2.44218946e-01 6.74986124e-01 4.45756651e-02 3.13392371e-01
-4.97300923e-01 1.83454808e-02 1.56649899e+00 1.66116759e-01
-1.42881417e+00 2.92558700e-01 -4.41584945e-01 3.19967240e-01
9.57878709e-01 -6.77753091e-01 -5.60197413e-01 -5.85580170e-01
-6.21524036e-01 -1.10568181e-01 9.56963301e-01 5.91862857e-01
1.30136406e+00 -1.66275239e+00 -8.96977901e-01 2.82418821e-02
3.66472423e-01 -9.44130719e-01 8.17112982e-01 9.00783598e-01
-9.54766333e-01 3.63356441e-01 -8.26422811e-01 -7.74972737e-01
-1.69156682e+00 8.56725216e-01 6.64042413e-01 -1.77398384e-01
-6.82837248e-01 2.45722786e-01 1.21810518e-01 6.97420120e-01
2.06277482e-02 4.54260223e-02 -3.44578028e-01 -4.15788680e-01
1.09339225e+00 8.74618709e-01 1.50957584e-01 -1.07277894e+00
-5.34764171e-01 7.39619374e-01 2.40095839e-01 -2.79465318e-01
9.07441795e-01 -1.15956736e+00 3.91256325e-02 5.42407930e-01
1.56374872e+00 -2.05620632e-01 -1.51572311e+00 -2.36291233e-02
-3.74226332e-01 -1.07457566e+00 1.76872969e-01 -9.44809318e-01
-1.31910455e+00 7.89323986e-01 1.74282682e+00 1.47989959e-01
1.81346941e+00 -1.65142298e-01 6.07108176e-01 -4.04833466e-01
2.13717744e-01 -9.68019485e-01 3.00914079e-01 3.51291418e-01
1.08054960e+00 -1.25267482e+00 2.16054216e-01 -2.94888824e-01
-6.44060731e-01 1.19643998e+00 5.27382731e-01 9.36614275e-02
5.32652497e-01 2.86718398e-01 1.00530624e-01 2.68268615e-01
-8.62897158e-01 -1.77615747e-01 6.89548194e-01 1.32803524e+00
6.48620605e-01 -4.08740342e-01 -1.05816670e-01 -2.88730979e-01
2.74091631e-01 5.28373979e-02 6.88083589e-01 6.37514055e-01
-4.82552946e-01 -7.26941228e-01 -3.57673407e-01 1.85714856e-01
-4.42984790e-01 -3.26231360e-01 1.61311299e-01 5.29871345e-01
9.55530703e-02 1.22946990e+00 4.13271964e-01 -9.19863462e-01
6.27480745e-01 -6.43702507e-01 9.24596727e-01 4.88797948e-02
-3.11194271e-01 -1.66584730e-01 -9.29150954e-02 -1.22250259e+00
-7.09068358e-01 -4.07072723e-01 -6.42058849e-01 -9.84446883e-01
4.01128680e-01 -3.39775443e-01 5.63092291e-01 2.95364380e-01
3.57637346e-01 5.17263234e-01 8.95348191e-01 -1.09381068e+00
8.47760364e-02 -7.00519323e-01 -8.08389664e-01 5.20010293e-01
6.42917812e-01 -4.41241324e-01 -6.55239165e-01 3.33013117e-01] | [11.662485122680664, -1.8930670022964478] |
6968001f-61f2-4ff8-8613-f3e1ae19407f | visual-semantic-contrastive-alignment-for-few | 2210.11000 | null | https://arxiv.org/abs/2210.11000v1 | https://arxiv.org/pdf/2210.11000v1.pdf | Visual-Semantic Contrastive Alignment for Few-Shot Image Classification | Few-Shot learning aims to train and optimize a model that can adapt to unseen visual classes with only a few labeled examples. The existing few-shot learning (FSL) methods, heavily rely only on visual data, thus fail to capture the semantic attributes to learn a more generalized version of the visual concept from very few examples. However, it is a known fact that human visual learning benefits immensely from inputs from multiple modalities such as vision, language, and audio. Inspired by the human learning nature of encapsulating the existing knowledge of a visual category which is in the form of language, we introduce a contrastive alignment mechanism for visual and semantic feature vectors to learn much more generalized visual concepts for few-shot learning. Our method simply adds an auxiliary contrastive learning objective which captures the contextual knowledge of a visual category from a strong textual encoder in addition to the existing training mechanism. Hence, the approach is more generalized and can be plugged into any existing FSL method. The pre-trained semantic feature extractor (learned from a large-scale text corpora) we use in our approach provides a strong contextual prior knowledge to assist FSL. The experimental results done in popular FSL datasets show that our approach is generic in nature and provides a strong boost to the existing FSL baselines. | ['Ranga Rodrigo', 'Mohamed Afham'] | 2022-10-20 | null | null | null | null | ['few-shot-image-classification'] | ['computer-vision'] | [ 2.02092320e-01 1.21241860e-01 -4.47881281e-01 -5.19355774e-01
-6.24991238e-01 -5.68648994e-01 1.01375222e+00 2.02035412e-01
-4.40041631e-01 5.21193624e-01 3.52005094e-01 7.70079195e-02
1.50810421e-01 -8.05066824e-01 -8.49347770e-01 -4.19561595e-01
2.96413004e-01 2.85461724e-01 5.10396481e-01 -3.40923637e-01
1.28177851e-01 1.53836876e-01 -1.83741045e+00 5.30681908e-01
3.81709754e-01 1.12829471e+00 3.97885054e-01 4.49018657e-01
-5.63800514e-01 1.08742440e+00 -2.74681300e-01 -2.71098405e-01
1.07352778e-01 -5.97714484e-01 -8.86930764e-01 2.76033312e-01
6.51816487e-01 -7.57239908e-02 -1.84613243e-01 8.65327060e-01
3.45001668e-01 4.80533123e-01 7.49762774e-01 -1.27678454e+00
-9.14610147e-01 3.02330673e-01 -3.65293622e-01 1.44165322e-01
2.79281944e-01 4.24711794e-01 1.24335790e+00 -1.18485665e+00
9.63000953e-01 1.22005987e+00 5.50384462e-01 8.75212312e-01
-1.10435295e+00 -2.18141735e-01 2.58187771e-01 5.47400057e-01
-1.14559913e+00 -5.10615230e-01 9.37043726e-01 -4.89481866e-01
9.44464445e-01 -3.71761695e-02 6.99494123e-01 1.30143368e+00
-4.63215739e-01 9.52455342e-01 1.08577561e+00 -7.67297924e-01
4.80206072e-01 5.26254654e-01 1.22271389e-01 6.99888170e-01
-9.92009789e-02 2.05618426e-01 -6.76706553e-01 1.83680788e-01
3.48583043e-01 3.01133364e-01 -1.85845092e-01 -9.22814190e-01
-9.09031391e-01 1.01417685e+00 6.95680737e-01 4.90315706e-01
-8.42374638e-02 2.08235011e-01 5.78716278e-01 3.38765502e-01
3.52667898e-01 3.20034117e-01 -2.97257751e-01 -1.10208588e-02
-8.98111403e-01 -9.12242606e-02 7.04110265e-01 9.84212458e-01
1.19258463e+00 1.47936955e-01 -4.82560307e-01 8.85079265e-01
2.95345634e-01 4.14267987e-01 6.05393946e-01 -8.51437747e-01
1.53040320e-01 6.93972468e-01 -2.68674493e-01 -5.53749979e-01
-6.58287331e-02 -2.15850174e-01 -3.32217962e-01 3.36829096e-01
2.43524443e-02 1.06360734e-01 -1.27198422e+00 1.73272932e+00
2.10048258e-01 4.01947051e-01 8.59724060e-02 7.00749397e-01
1.02884996e+00 5.94930232e-01 2.78240114e-01 -2.03097492e-01
1.23843324e+00 -1.10668528e+00 -5.37156343e-01 -7.76140094e-01
3.12857389e-01 -4.98960197e-01 1.47154713e+00 -5.20555750e-02
-7.86345184e-01 -6.24825954e-01 -1.17231107e+00 -2.39966825e-01
-8.25938523e-01 -4.63442564e-01 5.81899285e-01 3.95448416e-01
-1.00330889e+00 3.69976848e-01 -4.86134052e-01 -7.81768203e-01
6.86866939e-01 -5.79372495e-02 -4.29458350e-01 -4.82473344e-01
-1.14542329e+00 9.98843670e-01 7.41490602e-01 -5.14933050e-01
-1.02582574e+00 -6.21552229e-01 -1.30511093e+00 1.14219725e-01
7.47171640e-01 -6.56421125e-01 1.18503976e+00 -1.34240544e+00
-1.27713084e+00 1.06609643e+00 -1.06056288e-01 -3.11969429e-01
1.28410831e-01 3.43109854e-02 -1.57260910e-01 4.66238409e-01
1.50031775e-01 7.70631492e-01 1.13870966e+00 -1.55364192e+00
-4.70976770e-01 -2.30824783e-01 1.94739625e-01 1.17823914e-01
-5.99573076e-01 -1.24789655e-01 -5.27439475e-01 -5.96912682e-01
-6.20601416e-01 -5.57145059e-01 -1.13808125e-01 3.01924497e-01
-2.38385722e-02 -2.06114650e-01 9.39690173e-01 -2.75436580e-01
9.82999027e-01 -2.31840873e+00 2.00499132e-01 6.06544726e-02
1.39906943e-01 3.99878979e-01 -3.91418010e-01 6.75624967e-01
5.05406559e-02 -2.46608317e-01 -3.88857901e-01 -3.79768372e-01
5.25425598e-02 5.14244497e-01 -2.73773491e-01 1.29828393e-01
3.53906304e-01 1.22467363e+00 -1.26342273e+00 -6.46849632e-01
5.27401865e-01 5.50882995e-01 -5.03400087e-01 3.78202736e-01
-3.83602977e-01 6.04999214e-02 -2.04401776e-01 5.73882878e-01
1.39620766e-01 -4.48986858e-01 -8.03065300e-02 -1.66218013e-01
-2.32220367e-02 -1.35982201e-01 -9.09567356e-01 2.14081979e+00
-6.17470980e-01 5.57228506e-01 -3.84182781e-01 -1.36930633e+00
8.44880223e-01 3.94149661e-01 1.85178727e-01 -6.03384495e-01
1.96992055e-01 -1.08843058e-01 -4.25477803e-01 -6.54347956e-01
-1.08156223e-02 -5.64292789e-01 -1.12061098e-01 3.87204349e-01
9.67128754e-01 -1.56225100e-01 2.90189952e-01 4.80104268e-01
1.00274026e+00 2.71449685e-01 7.57963300e-01 1.32398054e-01
3.87520969e-01 2.22412460e-02 4.67848986e-01 9.03718174e-01
-3.35013300e-01 5.72280645e-01 1.54531479e-01 -4.10681546e-01
-1.17761600e+00 -1.07901716e+00 9.31062773e-02 1.52106261e+00
1.86851136e-02 -5.86717606e-01 -3.93244058e-01 -9.64576840e-01
-3.01660039e-03 8.97685468e-01 -8.02838147e-01 -4.54715967e-01
-1.99875876e-01 -2.36054108e-01 5.78750409e-02 8.30775619e-01
2.24246606e-01 -1.24386847e+00 -6.49618864e-01 2.26375367e-02
1.74994573e-01 -1.20797694e+00 -2.29357794e-01 3.41616422e-01
-6.04210079e-01 -1.01743698e+00 -5.28495610e-01 -9.29866910e-01
3.48048002e-01 2.89472342e-01 1.10857594e+00 2.54370254e-02
-5.82988322e-01 9.47022736e-01 -6.48395121e-01 -6.32899404e-01
-2.55619615e-01 -2.95677632e-01 -2.89008200e-01 2.28324100e-01
7.95762479e-01 -5.34792602e-01 -3.38871479e-01 -3.59173894e-01
-9.52165067e-01 -5.18249385e-02 5.93951106e-01 1.03142643e+00
3.24815363e-01 -4.04409528e-01 5.68332136e-01 -1.03584850e+00
2.03724027e-01 -5.78344047e-01 -1.48493648e-01 4.07464415e-01
-5.91813445e-01 2.70673335e-01 6.32378757e-01 -5.63670337e-01
-1.12166929e+00 2.55733699e-01 5.95796742e-02 -7.67910957e-01
-5.08491695e-01 4.30571258e-01 -1.66102827e-01 -6.80007637e-02
7.04901397e-01 3.04140747e-01 -7.67709762e-02 -4.22567874e-01
9.72538888e-01 5.40690124e-01 5.86430609e-01 -3.40287745e-01
1.00513828e+00 7.00445354e-01 -1.04300037e-01 -9.17226076e-01
-1.21273649e+00 -7.96291947e-01 -8.28246355e-01 -2.59083092e-01
8.00115764e-01 -9.27753747e-01 -2.60228634e-01 -4.91619706e-02
-9.51537967e-01 -1.62130356e-01 -7.76386738e-01 3.76461983e-01
-7.91113794e-01 3.74262422e-01 -3.80227208e-01 -7.34223485e-01
-2.27473944e-01 -6.23244226e-01 1.09294403e+00 1.44494042e-01
-1.30930111e-01 -1.17931342e+00 2.59717733e-01 1.09031357e-01
4.53217268e-01 1.24814339e-01 9.22633410e-01 -8.08782518e-01
-4.34335291e-01 -1.81873456e-01 -2.74708450e-01 3.95944297e-01
-4.84522469e-02 -2.50216335e-01 -1.34036386e+00 -3.31311196e-01
3.44899409e-02 -9.66030777e-01 1.20446849e+00 -4.75005340e-03
8.66368651e-01 -3.16198468e-01 -2.53169596e-01 5.58297753e-01
1.93431592e+00 -4.29792181e-02 4.26573515e-01 2.36558124e-01
8.68485928e-01 5.46141684e-01 4.87472713e-01 4.94983047e-01
2.72613436e-01 6.16575420e-01 4.86285716e-01 -9.09929723e-02
-4.85891432e-01 -3.92216444e-01 3.25902134e-01 7.67805994e-01
-1.29254207e-01 1.09944277e-01 -8.13238919e-01 8.12277257e-01
-1.89386737e+00 -1.32503664e+00 6.26117647e-01 2.02357578e+00
1.05944729e+00 -1.04049137e-02 7.15768039e-02 8.56366009e-02
4.51384693e-01 4.21793789e-01 -4.94051754e-01 -3.56116802e-01
1.38807679e-02 4.35128868e-01 2.39739209e-01 3.91550094e-01
-1.09097493e+00 1.09479856e+00 5.99316454e+00 8.38578939e-01
-1.00934279e+00 3.46267492e-01 -2.51761023e-02 -2.69248188e-01
-3.23648125e-01 1.88841090e-01 -7.12480724e-01 1.85220689e-01
7.57796049e-01 -1.82348505e-01 4.33094829e-01 9.53927398e-01
-2.64937490e-01 3.02256525e-01 -1.35822201e+00 1.05028188e+00
6.42293751e-01 -1.39146161e+00 3.20830315e-01 -2.19639286e-01
6.27919614e-01 9.77885351e-02 1.77478523e-03 6.83521450e-01
3.81396502e-01 -9.15541589e-01 6.16162658e-01 6.09648407e-01
8.57486069e-01 -5.87296605e-01 5.01940012e-01 3.78920555e-01
-1.21602368e+00 -3.37322980e-01 -4.82871234e-01 -1.28617570e-01
1.53772041e-01 9.19282585e-02 -7.59208500e-01 3.71504635e-01
6.43537998e-01 1.16828609e+00 -8.42574179e-01 9.61860538e-01
-4.59739923e-01 5.42790055e-01 1.26889884e-01 -5.92962056e-02
5.43835700e-01 3.41374338e-01 5.12628138e-01 1.24613082e+00
2.58453395e-02 3.76130305e-02 3.53039593e-01 8.37735474e-01
-6.62222784e-03 3.19747448e-01 -9.78935778e-01 -2.18572557e-01
3.40362698e-01 1.19184113e+00 -4.28087533e-01 -6.29789889e-01
-1.11548650e+00 9.59308326e-01 5.76947093e-01 4.17050093e-01
-4.05221224e-01 -4.36555684e-01 2.54816890e-01 -3.87103818e-02
7.04661489e-01 1.75662100e-01 5.69899082e-02 -1.33051896e+00
-1.37806013e-01 -6.04595065e-01 5.23825765e-01 -9.29208338e-01
-1.68442416e+00 4.54075038e-01 -7.01947734e-02 -1.24987137e+00
-5.34210980e-01 -6.82164848e-01 -6.92242861e-01 4.94781017e-01
-1.83290267e+00 -1.66604078e+00 -3.81481975e-01 9.73165929e-01
9.10006583e-01 -3.55964273e-01 8.85714054e-01 -6.84816251e-03
-1.43394619e-02 3.65843475e-01 -1.09896116e-01 2.18395531e-01
8.05114567e-01 -1.44015932e+00 7.42416680e-02 8.02470744e-01
5.27419031e-01 4.31737810e-01 7.47836530e-01 -4.19783086e-01
-1.19568849e+00 -1.08526909e+00 9.24112856e-01 -6.59813583e-01
8.78364801e-01 -4.74705994e-01 -1.02326858e+00 7.98639119e-01
5.00008464e-01 4.00923699e-01 9.56754327e-01 1.31371722e-01
-8.11431050e-01 3.64321982e-03 -7.92423427e-01 3.40693086e-01
1.00896585e+00 -9.07359779e-01 -1.23940134e+00 1.37019292e-01
8.62379968e-01 2.70772398e-01 -4.10465151e-01 2.87714213e-01
3.35411966e-01 -7.13207543e-01 1.18374407e+00 -9.41581368e-01
4.50856090e-01 -1.50818154e-01 -3.87228549e-01 -1.39139676e+00
-3.08894128e-01 -1.30501434e-01 -3.12781632e-01 1.30975997e+00
8.11358988e-02 -1.93831116e-01 4.86035883e-01 3.69904935e-01
-1.40047789e-01 -4.65514749e-01 -6.77110791e-01 -9.10834968e-01
-2.03360260e-01 -3.84305984e-01 1.88343763e-01 1.18345845e+00
2.63433546e-01 8.70416880e-01 -5.97425103e-01 -1.67173296e-01
8.50656152e-01 3.23364258e-01 6.77131891e-01 -1.42331910e+00
-4.75009888e-01 -1.68485224e-01 -6.90766871e-01 -5.67044497e-01
4.14765835e-01 -1.27915812e+00 1.44941851e-01 -1.64479065e+00
6.46881461e-01 7.78901875e-02 -6.72698140e-01 9.11306441e-01
-1.94312871e-01 3.40355694e-01 5.26566982e-01 1.43219074e-02
-9.05973375e-01 7.57282913e-01 9.86042798e-01 -3.65859568e-01
2.33104397e-02 -3.40001345e-01 -6.73907816e-01 8.59576464e-01
3.23692262e-01 -4.07107830e-01 -6.29590213e-01 -1.59820542e-01
1.93274289e-01 -4.08832222e-01 6.08943462e-01 -8.76762986e-01
2.50238150e-01 -2.44485483e-01 3.76133323e-01 -1.83089927e-01
4.31190431e-01 -9.39576685e-01 -5.24682283e-01 2.23622188e-01
-4.19168741e-01 -5.41648388e-01 -5.87485321e-02 7.65390873e-01
-3.85443598e-01 -4.26651865e-01 8.71460915e-01 -4.79714781e-01
-1.57578027e+00 3.35945606e-01 -5.78711592e-02 3.47481400e-01
9.99560893e-01 -3.51209521e-01 -3.59416127e-01 -3.29672545e-01
-8.80778193e-01 1.56173751e-01 6.08419657e-01 6.36205077e-01
7.58507550e-01 -1.46964955e+00 -3.79333138e-01 1.80980846e-01
8.59192491e-01 -4.26224500e-01 5.34275547e-03 4.38555539e-01
1.52162552e-01 2.59193689e-01 -4.07929033e-01 -4.92514491e-01
-1.01828599e+00 1.23840821e+00 -5.26731238e-02 2.54311472e-01
-7.91575313e-01 7.52673805e-01 3.23452801e-01 -2.28401646e-01
2.71639585e-01 1.21982835e-01 -4.50930208e-01 2.50313133e-01
7.82217026e-01 4.57712002e-02 -2.39458039e-01 -7.24466681e-01
-2.86128640e-01 6.91255748e-01 2.12848186e-01 -1.53219163e-01
1.59295297e+00 -3.33905965e-01 2.12148443e-01 1.11857104e+00
1.29196942e+00 -2.12831736e-01 -1.35112774e+00 -7.88128436e-01
1.63212642e-01 -3.81775439e-01 -3.68092135e-02 -7.76930571e-01
-8.64402950e-01 1.26264894e+00 5.00621140e-01 -1.10668853e-01
1.03904068e+00 5.53597689e-01 4.99297529e-01 4.04226571e-01
2.59077340e-01 -1.34106100e+00 5.52962601e-01 4.66634482e-01
6.97415650e-01 -1.60831928e+00 -2.19613656e-01 -1.55288607e-01
-9.05022681e-01 1.03686500e+00 6.00725532e-01 -9.58768725e-02
5.37169099e-01 1.22256085e-01 9.74670649e-02 -3.48735094e-01
-9.54787493e-01 -7.79369771e-01 5.56395769e-01 8.49564612e-01
2.36999258e-01 -2.88632572e-01 7.65553787e-02 5.21138012e-01
2.87470520e-01 1.65666446e-01 2.84733951e-01 1.12001228e+00
-9.46471810e-01 -1.03614855e+00 7.89616853e-02 3.01063329e-01
-1.24789394e-01 -2.54874289e-01 -3.85540694e-01 6.93809211e-01
3.02604854e-01 6.55320108e-01 5.36434762e-02 2.77747437e-02
2.44758233e-01 5.29984832e-01 5.12542784e-01 -1.24942982e+00
-2.30070755e-01 -1.52851552e-01 -2.72741497e-01 -7.45394409e-01
-6.63836062e-01 -3.50200176e-01 -1.05586481e+00 3.82731825e-01
-2.79175006e-02 -8.42534453e-02 4.25748140e-01 1.31246364e+00
8.27178301e-04 3.43612403e-01 4.52832371e-01 -7.78584182e-01
-3.87710959e-01 -7.79757202e-01 -6.38320982e-01 8.62430155e-01
5.40794551e-01 -8.15944850e-01 -3.88271868e-01 3.21523517e-01] | [10.102428436279297, 2.4508845806121826] |
413249bd-6270-4f21-867d-3d9401be7a6b | cross-lingual-and-cross-domain-discourse | null | null | https://aclanthology.org/P17-2037 | https://aclanthology.org/P17-2037.pdf | Cross-lingual and cross-domain discourse segmentation of entire documents | Discourse segmentation is a crucial step in building end-to-end discourse parsers. However, discourse segmenters only exist for a few languages and domains. Typically they only detect intra-sentential segment boundaries, assuming gold standard sentence and token segmentation, and relying on high-quality syntactic parses and rich heuristics that are not generally available across languages and domains. In this paper, we propose statistical discourse segmenters for five languages and three domains that do not rely on gold pre-annotations. We also consider the problem of learning discourse segmenters when no labeled data is available for a language. Our fully supervised system obtains 89.5{\%} F1 for English newswire, with slight drops in performance on other domains, and we report supervised and unsupervised (cross-lingual) results for five languages in total. | ['Anders S{\\o}gaard', "Oph{\\'e}lie Lacroix", "Chlo{\\'e} Braud"] | 2017-07-01 | null | null | null | acl-2017-7 | ['discourse-segmentation'] | ['natural-language-processing'] | [ 1.95503831e-01 7.05580175e-01 -6.27657294e-01 -4.29740459e-01
-1.35526371e+00 -1.10584867e+00 7.14256346e-01 5.16041994e-01
-6.73672318e-01 1.12658393e+00 5.00234663e-01 -5.75369060e-01
3.75290155e-01 -6.20062292e-01 -5.73700011e-01 -1.14937134e-01
3.68077531e-02 8.24713051e-01 6.87085092e-01 -3.20117682e-01
1.41802549e-01 -3.20671082e-01 -8.51185679e-01 6.86588228e-01
1.17346358e+00 3.26826006e-01 1.31276175e-01 9.86389339e-01
-4.35025245e-01 1.04230559e+00 -9.11249101e-01 -4.76653785e-01
-2.53017217e-01 -7.08096504e-01 -1.55120242e+00 2.67102361e-01
2.01365352e-01 -1.21635571e-01 3.37688327e-02 8.56171012e-01
1.59519508e-01 -1.98494136e-01 6.47394598e-01 -5.22531152e-01
-3.89252752e-01 1.23061275e+00 -4.58812237e-01 3.96512508e-01
6.63388550e-01 -4.37043160e-02 1.30845106e+00 -3.95326942e-01
1.17771614e+00 1.18678200e+00 4.36666280e-01 6.29596472e-01
-1.17647052e+00 1.01331910e-02 3.64604264e-01 -1.70510679e-01
-8.08950663e-01 -5.59432745e-01 5.21121085e-01 -7.36646891e-01
1.29871190e+00 3.12883198e-01 1.91324204e-01 1.06760466e+00
-3.50668430e-01 1.00054908e+00 1.09064889e+00 -7.35373437e-01
2.08990067e-01 1.13733403e-01 5.04938602e-01 4.19335634e-01
1.35087028e-01 -4.51261103e-01 -4.58385527e-01 -4.16000432e-04
2.63393611e-01 -9.81908441e-01 -8.29222202e-02 3.99742961e-01
-1.24010706e+00 9.86645520e-01 -5.85113503e-02 6.73240960e-01
-9.80880558e-02 -3.87926787e-01 9.49967504e-01 4.55486745e-01
7.91622937e-01 6.09059572e-01 -4.26782072e-01 -4.80758995e-01
-1.13319445e+00 4.17747229e-01 1.34976375e+00 9.64227021e-01
5.07269561e-01 -1.44297346e-01 -1.46195382e-01 8.52774084e-01
4.98270728e-02 3.50575298e-01 3.85925025e-01 -1.13273084e+00
9.67963278e-01 6.88836217e-01 7.21275657e-02 -5.14181018e-01
-4.78496701e-01 7.00936541e-02 -3.39076489e-01 -1.65528446e-01
9.86145496e-01 -7.64671922e-01 -7.11489439e-01 1.62884736e+00
3.50645006e-01 -5.38756490e-01 3.43113929e-01 6.65808201e-01
1.08623219e+00 6.12055600e-01 1.94449916e-01 -5.31593084e-01
1.58316207e+00 -9.72978592e-01 -7.11679101e-01 -6.06434464e-01
1.15550292e+00 -1.00279236e+00 9.88928497e-01 2.62660444e-01
-1.32184565e+00 -2.80402750e-01 -9.45213974e-01 -5.36156118e-01
-8.22622553e-02 5.40505089e-02 3.36238712e-01 7.13142216e-01
-8.36251199e-01 3.35886598e-01 -9.66383517e-01 -2.13114113e-01
1.96914762e-01 4.74143736e-02 -5.95308281e-02 8.86850208e-02
-1.26877820e+00 9.96043205e-01 7.11446047e-01 -3.51432949e-01
-5.21817625e-01 -2.90271193e-01 -9.49165404e-01 -4.14836228e-01
5.04192531e-01 -2.16754884e-01 1.67988193e+00 -1.29335201e+00
-1.60640109e+00 1.45572555e+00 -5.04907370e-01 -6.82722032e-01
7.58101761e-01 -4.38472688e-01 -1.47148415e-01 3.15943360e-01
4.20284599e-01 3.68813187e-01 2.32353345e-01 -9.97439086e-01
-8.74986649e-01 -9.40668583e-02 4.13943172e-01 3.45281661e-01
7.29184449e-02 4.41964298e-01 -4.31460440e-01 -3.99928123e-01
-6.92600980e-02 -7.08199799e-01 -2.03622490e-01 -6.36320531e-01
-8.34680319e-01 -6.25312328e-01 6.82415426e-01 -1.32396662e+00
1.34685087e+00 -1.82024634e+00 1.11401014e-01 -7.94889852e-02
-1.88650768e-02 3.73279393e-01 7.95851573e-02 2.48708025e-01
2.79794037e-01 3.25618148e-01 -5.89692652e-01 -3.13657433e-01
6.49976209e-02 2.89755762e-01 -8.45237300e-02 4.31777030e-01
3.95866334e-01 7.55491793e-01 -1.02459228e+00 -8.37114513e-01
-4.74596657e-02 -4.70433943e-02 -6.43884599e-01 7.09428713e-02
-8.59340072e-01 6.12443388e-01 -5.14685333e-01 4.29627597e-01
2.26949975e-01 -1.48828775e-01 7.07389951e-01 4.14504647e-01
-4.34016109e-01 1.25050581e+00 -8.90129209e-01 1.72204518e+00
-3.95180255e-01 8.89360249e-01 2.81451225e-01 -1.14586306e+00
7.03966975e-01 4.64418799e-01 1.38738349e-01 -4.94452000e-01
3.29853803e-01 5.72641134e-01 3.19818556e-01 -7.06425726e-01
8.52584839e-01 -1.11123048e-01 -6.34324789e-01 5.55793822e-01
7.98831210e-02 -1.19587243e-01 8.39797497e-01 1.24978296e-01
1.09762502e+00 3.54262106e-02 4.23407823e-01 -4.22318608e-01
4.70763713e-01 7.44342923e-01 5.87163210e-01 4.87255633e-01
-1.66653395e-01 6.74591422e-01 1.08496094e+00 9.86037478e-02
-1.24865079e+00 -6.57899618e-01 -1.95813909e-01 1.49722266e+00
-1.18666790e-01 -4.62737024e-01 -1.16372705e+00 -6.88994527e-01
-3.52690965e-01 9.13188994e-01 -2.55886078e-01 6.39873862e-01
-1.21448672e+00 -6.46137655e-01 8.73753548e-01 3.10748428e-01
4.14229244e-01 -1.06563103e+00 -3.65525037e-01 4.48844403e-01
-5.13426423e-01 -1.43549848e+00 -1.04135536e-01 1.04718052e-01
-5.96693516e-01 -1.36005044e+00 -6.65716290e-01 -1.17435837e+00
2.82214582e-01 -3.88289213e-01 1.35874510e+00 2.34133396e-02
3.58402520e-01 -1.08616561e-01 -4.74874437e-01 -3.11202168e-01
-1.07203770e+00 7.84632266e-01 -4.50366348e-01 -7.26696551e-01
5.14140248e-01 3.16995904e-02 -1.21844731e-01 -4.42039855e-02
-3.25414509e-01 7.37776607e-02 1.45569414e-01 6.80354476e-01
3.01766396e-01 -4.42037076e-01 7.21428931e-01 -1.73019338e+00
7.40769565e-01 -5.78455746e-01 -5.00217378e-01 1.21435471e-01
-3.01279221e-02 -1.18999697e-01 5.96214592e-01 -7.72057101e-02
-1.39775372e+00 -1.92510039e-01 -5.56852639e-01 5.67191243e-01
-4.19740885e-01 7.41933882e-01 -2.61033475e-01 7.59084940e-01
9.88719523e-01 -3.41357708e-01 -2.56885380e-01 -5.90053380e-01
4.50374752e-01 8.74561608e-01 5.56632698e-01 -8.15827131e-01
3.68748933e-01 7.76883438e-02 -6.36781931e-01 -1.27334762e+00
-1.04880989e+00 -5.94592154e-01 -8.79767895e-01 1.78782880e-01
1.28342307e+00 -1.05269313e+00 -4.51053143e-01 2.63264447e-01
-1.49260950e+00 -7.35588312e-01 -4.79099035e-01 3.23530376e-01
-5.01341939e-01 4.13768560e-01 -9.95142877e-01 -6.25609815e-01
-2.22378850e-01 -9.24624503e-01 7.84949362e-01 2.36544356e-01
-8.74745250e-01 -1.18647277e+00 1.93814397e-01 5.13194621e-01
-9.44257677e-02 4.66376632e-01 7.02916384e-01 -1.04658258e+00
-2.28430584e-01 1.84841663e-01 -5.49018010e-02 3.79912555e-01
-2.50289366e-02 8.17677751e-02 -9.52184796e-01 -2.97228014e-03
-2.66620934e-01 -6.29680872e-01 8.41990471e-01 4.40366924e-01
5.75972855e-01 -4.61167753e-01 -3.18542749e-01 1.11609682e-01
9.86176789e-01 3.54961725e-03 3.46477449e-01 4.70001400e-01
6.49888337e-01 9.29119885e-01 6.25664651e-01 1.50428087e-01
6.90410018e-01 2.54525334e-01 -2.53501624e-01 -3.30998898e-02
-3.76840644e-02 -1.37553513e-01 4.99611586e-01 1.13747311e+00
-7.95251206e-02 -4.44035709e-01 -1.26600933e+00 1.12319398e+00
-1.71187770e+00 -8.24858367e-01 -5.51876307e-01 1.66714144e+00
1.55635083e+00 5.84258735e-01 4.88806635e-01 2.75709182e-02
8.18478525e-01 4.77985501e-01 -3.76809478e-01 -7.06335306e-01
-3.89764309e-01 3.67795855e-01 4.92187053e-01 9.81320381e-01
-1.45139527e+00 1.25729311e+00 6.70020533e+00 5.94810188e-01
-8.91989410e-01 2.17541590e-01 6.08336926e-01 1.46958888e-01
-4.00137573e-01 2.15452552e-01 -9.54255641e-01 2.85701722e-01
1.20666170e+00 -3.40529859e-01 -1.24740988e-01 9.20774817e-01
9.95512605e-02 -3.82325143e-01 -1.11736333e+00 3.09324533e-01
-2.66575992e-01 -1.47411060e+00 -6.31699979e-01 -2.06485838e-01
9.26136315e-01 3.39787692e-01 -5.28980434e-01 4.42511380e-01
7.92850316e-01 -9.18174684e-01 7.15092540e-01 -1.11794680e-01
8.33994985e-01 -5.50158262e-01 6.34420037e-01 6.76276505e-01
-7.59060442e-01 4.40972209e-01 -1.33100629e-01 -2.68947154e-01
5.30588031e-01 3.83127183e-01 -1.16057432e+00 3.63715112e-01
2.42637321e-01 6.11975849e-01 -2.18959972e-01 6.70499504e-01
-6.64513886e-01 1.27092254e+00 -4.55958486e-01 -2.23092273e-01
4.40886438e-01 2.61191037e-02 7.09223449e-01 1.84764779e+00
-1.37914985e-01 2.24791750e-01 6.48274064e-01 4.91172105e-01
-3.32221031e-01 2.10877419e-01 -3.70915622e-01 -1.58186749e-01
5.71644962e-01 8.05642664e-01 -9.46683764e-01 -5.67220986e-01
-4.56648588e-01 7.20511734e-01 3.43167305e-01 2.37483084e-01
-4.82139349e-01 -3.68776947e-01 5.40986001e-01 1.92027867e-01
1.67591438e-01 -4.82847661e-01 -5.57463944e-01 -1.06118739e+00
1.22552030e-02 -1.08846581e+00 5.43216646e-01 8.88084844e-02
-1.25623238e+00 4.94997114e-01 7.55164549e-02 -6.42613411e-01
-8.40907931e-01 -6.17241979e-01 -6.51690960e-01 8.57246101e-01
-1.63078046e+00 -1.09294772e+00 2.24760041e-01 2.72790432e-01
8.20608556e-01 1.40802199e-02 8.21487129e-01 2.33303145e-01
-4.87524629e-01 4.48233336e-01 8.13748911e-02 7.80597508e-01
7.32914567e-01 -1.58302319e+00 6.83403075e-01 9.96647060e-01
-6.45697936e-02 4.45754677e-01 9.75114405e-01 -6.26393139e-01
-7.61759758e-01 -9.91777599e-01 1.49652207e+00 -3.63100350e-01
8.32587719e-01 -3.93286616e-01 -1.21070707e+00 9.48464215e-01
6.75074995e-01 -4.57502216e-01 7.77182400e-01 6.92540765e-01
-4.34681363e-02 7.44394243e-01 -8.57388616e-01 3.49038839e-01
8.99784863e-01 -7.35484719e-01 -1.16835618e+00 5.78122199e-01
8.50912511e-01 -9.54656243e-01 -8.56754124e-01 1.24734780e-02
2.10744031e-02 -7.98011839e-01 5.93757033e-01 -7.02631950e-01
6.77884340e-01 -1.36989936e-01 -6.01952821e-02 -1.00787401e+00
1.87427565e-01 -6.72903121e-01 2.21410245e-01 1.64923739e+00
9.58782256e-01 -4.03461605e-01 7.30373979e-01 5.87929666e-01
-5.17145157e-01 -3.01414162e-01 -9.23456728e-01 -6.08488500e-01
7.00919986e-01 -5.41993916e-01 9.50047374e-02 1.13252997e+00
4.41951185e-01 7.58811891e-01 1.11363247e-01 -1.32183924e-01
2.73126632e-01 1.69412211e-01 7.26634979e-01 -1.00024939e+00
-3.55478764e-01 -3.83892506e-01 1.62131235e-01 -1.25377572e+00
5.47175944e-01 -9.40896213e-01 3.38341057e-01 -1.69308364e+00
-1.24155752e-01 -6.24850392e-01 4.28448439e-01 4.27330762e-01
-2.00797856e-01 1.02294490e-01 -4.31988426e-02 1.16031796e-01
-8.32587302e-01 1.61265984e-01 9.83577549e-01 -8.22196007e-02
-6.10038340e-01 -2.95631327e-02 -6.81700468e-01 1.28091013e+00
9.61172581e-01 -3.58190328e-01 -1.51849434e-01 -6.80630326e-01
4.54268716e-02 6.38984963e-02 -1.18570760e-01 -5.21457672e-01
1.15757294e-01 -2.89040029e-01 -2.96833813e-02 -6.76767051e-01
-3.53035964e-02 7.41056502e-02 -5.73958993e-01 1.63140476e-01
-4.15713727e-01 -1.63482040e-01 9.33373272e-02 4.71830815e-02
-5.07781625e-01 -5.23998737e-01 7.05587447e-01 -3.41952562e-01
-7.96909094e-01 -3.52899969e-01 -5.55797815e-01 1.10341561e+00
1.02329481e+00 -4.03018966e-02 -6.92507982e-01 -1.38931759e-02
-8.17299128e-01 4.27652568e-01 5.48401177e-01 2.17018902e-01
-4.05946448e-02 -8.01049352e-01 -1.09956682e+00 -2.50698894e-01
-1.90605134e-01 5.11639416e-01 -9.76983681e-02 5.66380084e-01
-8.74065936e-01 5.48274219e-01 9.09526646e-02 -6.20464444e-01
-1.27715540e+00 5.92449680e-02 8.43510777e-03 -6.44699693e-01
-5.01939356e-01 9.06346500e-01 -1.14211328e-01 -4.82992440e-01
1.44044794e-02 -4.71209586e-01 -2.84994632e-01 5.59489787e-01
4.28688228e-01 1.80897281e-01 -6.63933754e-02 -9.61168110e-01
-3.68394464e-01 6.99064732e-02 -1.59857243e-01 -4.14608449e-01
1.18003392e+00 -2.19182938e-01 -5.93962856e-02 7.30400980e-01
9.55443442e-01 3.18187773e-01 -1.03870451e+00 -3.20759118e-01
7.25914657e-01 -1.06850881e-02 -3.35935205e-01 -5.10964513e-01
-3.95966262e-01 8.80563319e-01 -2.98433065e-01 5.88907361e-01
6.82908893e-01 3.60689670e-01 1.03064764e+00 2.57191420e-01
-2.50079837e-02 -1.61759138e+00 -2.78776109e-01 1.16923165e+00
5.79443693e-01 -1.31978238e+00 8.35697807e-04 -6.26608789e-01
-7.94528782e-01 8.00538003e-01 5.38881838e-01 -2.97484249e-01
4.25065011e-01 3.23147267e-01 1.82927623e-01 -8.01965371e-02
-5.81182897e-01 -4.53331858e-01 1.62893400e-01 6.45708382e-01
1.15350914e+00 2.58777767e-01 -7.96752214e-01 6.95784867e-01
-6.29670799e-01 -3.29148412e-01 5.58736920e-01 1.04375947e+00
-7.64760852e-01 -1.42944121e+00 -2.59585321e-01 2.78317481e-01
-9.77440357e-01 -5.31594865e-02 -5.12456298e-01 1.11097193e+00
1.01505026e-01 1.14383137e+00 2.34398663e-01 1.59461513e-01
2.77514368e-01 7.38026947e-02 3.92769367e-01 -9.81016397e-01
-7.35738456e-01 3.14674020e-01 1.11493611e+00 -6.09169789e-02
-6.62918627e-01 -1.01270235e+00 -1.61606145e+00 -3.92107427e-01
-1.43178880e-01 2.42719486e-01 1.61249951e-01 1.21142721e+00
-1.87584639e-01 5.58487415e-01 1.09933890e-01 -5.63885450e-01
-1.45365000e-01 -1.14068961e+00 -2.72107035e-01 3.94789964e-01
3.96165937e-01 -2.09732920e-01 -5.03744520e-02 4.67793912e-01] | [10.851201057434082, 9.497427940368652] |
72f6e469-1f50-4e47-aa05-667bb18e49e3 | prompting-language-informed-distribution-for | 2305.14428 | null | https://arxiv.org/abs/2305.14428v1 | https://arxiv.org/pdf/2305.14428v1.pdf | Prompting Language-Informed Distribution for Compositional Zero-Shot Learning | The compositional zero-shot learning (CZSL) task aims to recognize unseen compositional visual concepts (i.e., sliced tomatoes), where the models are learned only from the seen compositions (i.e., sliced potatoes and red tomatoes). Thanks to the prompt tuning on large pre-trained visual language models such as CLIP, recent literature shows impressively better CZSL performance than traditional vision-based methods. However, the key aspects that impact the generalization to unseen compositions, including the diversity and informativeness of class context, and the entanglement between visual primitives (i.e., states and objects), are not properly addressed in existing CLIP-based CZSL literature. In this paper, we propose a model by prompting the language-informed distribution, aka., PLID, for the CZSL task. Specifically, the PLID leverages pre-trained large language models (LLM) to 1) formulate the language-informed class distribution, and 2) enhance the compositionality of the softly prompted class embedding. Moreover, a stochastic logit mixup strategy is proposed to dynamically fuse the decisions from the predictions in the compositional and the primitive logit space. Orthogonal to the existing literature of soft, hard, or distributional prompts, our method advocates prompting the LLM-supported class distribution that leads to a better compositional zero-shot generalization. Experimental results on MIT-States, UT-Zappos, and C-GQA datasets show the superior performance of the PLID to the prior arts. The code and models will be publicly released. | ['Yu Kong', 'Heng Huang', 'Lichang Chen', 'Wentao Bao'] | 2023-05-23 | null | null | null | null | ['compositional-zero-shot-learning'] | ['computer-vision'] | [ 3.68801266e-01 -1.50937498e-01 -4.19814587e-01 -1.40272960e-01
-6.28106892e-01 -8.14252853e-01 9.65443254e-01 -9.37484577e-02
-8.05385858e-02 1.83675736e-01 4.31854337e-01 -3.52723151e-01
2.41869748e-01 -5.90272903e-01 -7.85085618e-01 -8.51770282e-01
4.12569761e-01 3.12245637e-01 2.67368942e-01 -1.01042993e-01
-1.53716262e-02 6.62581101e-02 -1.61731052e+00 7.24581301e-01
6.91837847e-01 1.14375281e+00 5.20793676e-01 6.45309150e-01
-4.69862670e-01 9.87310410e-01 -1.43228531e-01 -4.77831811e-01
2.98533916e-01 -4.45559382e-01 -2.07528129e-01 -1.80126950e-02
6.71096265e-01 -4.26260680e-01 -2.75716424e-01 1.19395065e+00
2.61230141e-01 7.15572387e-02 1.00784039e+00 -1.36270142e+00
-1.33957624e+00 8.52389276e-01 -4.71840799e-01 7.42615983e-02
1.67440057e-01 7.40730286e-01 1.41413748e+00 -1.23509228e+00
6.55972123e-01 1.55681455e+00 4.51609015e-01 6.01136744e-01
-1.46371973e+00 -7.90943682e-01 5.57185233e-01 3.13253671e-01
-1.44883573e+00 -3.54244739e-01 9.63895738e-01 -8.06316495e-01
7.36222267e-01 1.07792422e-01 6.31639063e-01 1.76287186e+00
-9.26560983e-02 1.07988358e+00 1.19983435e+00 -5.36293149e-01
5.55949569e-01 2.45536104e-01 1.08412072e-01 7.81402290e-01
6.22558296e-02 2.65990824e-01 -6.99266553e-01 -1.16868056e-01
4.93245453e-01 -2.24233456e-02 -1.89467564e-01 -6.90004885e-01
-1.09586489e+00 8.02644432e-01 2.81099856e-01 7.57854432e-02
-2.42491305e-01 1.20424986e-01 3.29443783e-01 -5.05879968e-02
3.66071224e-01 2.57064193e-01 -1.42384112e-01 2.03982651e-01
-1.01272881e+00 1.19613349e-01 6.45223439e-01 1.17594552e+00
7.83262312e-01 2.87331522e-01 -5.36325932e-01 6.47052646e-01
5.07986069e-01 6.51555359e-01 2.07963377e-01 -6.74796402e-01
3.53023976e-01 4.53786910e-01 -1.91590890e-01 -6.15105629e-01
9.74331051e-02 -1.71530336e-01 -6.08525515e-01 2.55552262e-01
2.61773348e-01 8.55079889e-02 -1.33991969e+00 2.03529215e+00
1.34525791e-01 2.96884835e-01 9.31363627e-02 1.02833629e+00
8.70809853e-01 9.19566393e-01 6.28152907e-01 2.28948846e-01
1.48785901e+00 -8.93032789e-01 -5.44256330e-01 -3.58808011e-01
1.67845204e-01 -5.70934832e-01 1.57846367e+00 1.33110315e-01
-8.21012735e-01 -6.79943919e-01 -1.03323853e+00 -3.37523341e-01
-4.65846181e-01 1.42234832e-01 4.51311648e-01 5.23960769e-01
-9.42590892e-01 1.46216482e-01 -8.01908970e-01 -4.53598857e-01
4.87558901e-01 -1.31272063e-01 1.59318447e-02 -1.90721959e-01
-1.16349626e+00 5.99578083e-01 5.89890540e-01 -2.33657703e-01
-1.27945352e+00 -1.00852668e+00 -1.01792288e+00 2.71882385e-01
6.11270785e-01 -7.03518271e-01 1.09194565e+00 -1.34397185e+00
-1.43302083e+00 7.55311966e-01 -1.42150551e-01 -4.04713959e-01
3.51079285e-01 1.03897780e-01 -1.65463760e-02 3.61225784e-01
-7.02273771e-02 9.97176170e-01 1.06723940e+00 -1.58450747e+00
-4.90764558e-01 -3.03510651e-02 1.17581636e-01 2.11560950e-01
-4.17185545e-01 -1.29204720e-01 -2.77677357e-01 -7.26535141e-01
-1.33566365e-01 -8.15235853e-01 1.55279025e-01 4.03861105e-01
-4.44943935e-01 -1.96027622e-01 7.73526669e-01 -5.82459033e-01
9.97820079e-01 -2.66746378e+00 1.52952030e-01 -7.29174539e-02
1.42345175e-01 1.86525211e-01 -3.12606573e-01 4.49303240e-01
7.38650933e-02 1.07070498e-01 -2.09600568e-01 -5.02598763e-01
5.61723232e-01 3.37118536e-01 -9.35029387e-01 1.84767991e-01
3.35423142e-01 1.18302500e+00 -9.67172503e-01 -5.09865582e-01
3.26042503e-01 4.53017801e-01 -5.96608400e-01 1.80615291e-01
-6.69834673e-01 1.57552715e-02 7.15350965e-03 8.65919888e-01
4.16273624e-01 -3.67455125e-01 1.24023341e-01 -5.15490949e-01
-9.07528400e-02 6.21601567e-02 -8.33139956e-01 1.52444983e+00
-1.60071358e-01 4.75046933e-01 -1.25087500e-01 -6.27930343e-01
6.50225580e-01 1.84740260e-01 1.57345273e-02 -6.08431756e-01
-1.24332443e-01 -3.35571282e-02 -2.46373136e-02 -3.81093174e-01
5.86753190e-01 -5.19166350e-01 -2.03057140e-01 3.61356109e-01
4.44488674e-01 -1.13772266e-01 5.21995947e-02 5.00313699e-01
6.54449344e-01 4.29752856e-01 4.38213408e-01 -3.41960818e-01
2.60786749e-02 -1.10296942e-01 5.71104705e-01 8.38478446e-01
-3.50236505e-01 7.03004420e-01 5.77359438e-01 -7.94646442e-02
-1.24337113e+00 -1.65389431e+00 2.07099468e-01 1.47105515e+00
3.92324716e-01 -3.54310006e-01 -5.04934669e-01 -5.09945035e-01
9.06398520e-02 1.15816653e+00 -6.43122554e-01 -3.69932979e-01
-2.67829716e-01 -3.30178022e-01 5.15355885e-01 6.62443638e-01
1.74254552e-01 -8.49109769e-01 -4.97449487e-01 -5.71956411e-02
5.50053492e-02 -1.16670716e+00 -7.15944052e-01 2.30463669e-01
-3.13963801e-01 -7.01701760e-01 -6.17322862e-01 -8.76616955e-01
4.33433443e-01 1.95800185e-01 9.10233319e-01 -3.70414406e-01
-2.25446820e-01 6.29547954e-01 -2.73853570e-01 -2.95897216e-01
-4.42058235e-01 -2.65234619e-01 -7.74063095e-02 2.96403915e-01
3.87904674e-01 -4.76337850e-01 -4.16795790e-01 -1.15232423e-01
-9.59125757e-01 3.75858039e-01 5.69461882e-01 8.58451307e-01
5.10539472e-01 -3.43816698e-01 3.88225317e-01 -7.03144193e-01
3.54039788e-01 -6.66122615e-01 -4.14633244e-01 5.53195715e-01
-5.20962298e-01 1.60702676e-01 6.44602835e-01 -1.07834351e+00
-1.22350097e+00 -3.37835625e-02 2.29833081e-01 -1.02067876e+00
-2.02347890e-01 3.84405881e-01 -2.69818693e-01 2.52928644e-01
4.54645604e-01 4.01287884e-01 -2.98464000e-01 -2.96865106e-01
8.67364287e-01 4.19201136e-01 6.09521270e-01 -9.14879203e-01
7.61915207e-01 6.35157764e-01 -3.69824857e-01 -8.16128671e-01
-6.66678429e-01 -3.11828047e-01 -2.96794385e-01 -1.03000335e-01
1.18546343e+00 -1.11491346e+00 -4.44507450e-01 3.57013702e-01
-1.09015810e+00 -5.09821415e-01 -5.28660417e-01 2.46413007e-01
-4.82243806e-01 5.13701200e-01 -7.83870578e-01 -1.19267118e+00
-2.98882753e-01 -1.01517284e+00 1.14182794e+00 2.56016970e-01
-3.23524177e-01 -9.96629715e-01 -1.49900839e-01 1.69291154e-01
2.88068712e-01 -1.21719409e-02 1.63813937e+00 -6.09686553e-01
-9.04238403e-01 2.44655475e-01 -3.88425618e-01 3.66544664e-01
-9.98013094e-02 3.04771364e-01 -1.14764655e+00 -9.54402089e-02
-1.89030945e-01 -5.97781599e-01 1.14111912e+00 1.77239314e-01
8.70208919e-01 -3.15308243e-01 -1.01994172e-01 5.75209200e-01
1.38358593e+00 2.56313443e-01 2.66067713e-01 -2.06024826e-01
7.53990114e-01 4.30933535e-01 2.54888266e-01 4.45350975e-01
6.52428806e-01 3.47676396e-01 4.19153005e-01 4.35824282e-02
-6.20779693e-01 -9.72871959e-01 6.51825666e-01 8.09981942e-01
3.33150923e-01 -4.67155546e-01 -8.49362612e-01 6.01859689e-01
-1.75104320e+00 -1.10804546e+00 4.72730428e-01 1.83840036e+00
9.70987558e-01 5.48049323e-02 -1.75676271e-01 -3.43156099e-01
5.86268663e-01 5.63385963e-01 -7.42972374e-01 -2.27510765e-01
-4.22047943e-01 2.27517262e-01 1.25153244e-01 5.30251205e-01
-1.04253113e+00 1.19268000e+00 5.64376545e+00 1.04284286e+00
-1.22644377e+00 2.56242543e-01 4.33845997e-01 -3.89993131e-01
-7.92663753e-01 2.91940302e-01 -8.47531080e-01 6.32474124e-01
5.31164110e-01 -1.77657872e-01 6.88522637e-01 8.42690945e-01
-1.82429329e-01 -7.00993240e-02 -1.34010220e+00 9.53539550e-01
4.26064193e-01 -1.24073493e+00 3.93307626e-01 -4.25559282e-02
8.14475715e-01 -3.32669802e-02 5.68591833e-01 8.29196811e-01
5.41321039e-01 -7.44107306e-01 1.48138404e+00 5.52962244e-01
9.29732919e-01 -2.33904287e-01 1.18781172e-01 4.61302966e-01
-1.22221410e+00 -3.64857554e-01 -2.96686888e-01 7.32170045e-02
1.07526623e-01 2.88279742e-01 -7.08635211e-01 3.52657497e-01
4.01525885e-01 5.36496878e-01 -4.85984176e-01 4.93346095e-01
-3.86547416e-01 8.86493742e-01 -8.64495635e-02 -1.33623749e-01
3.55879843e-01 -1.75547734e-01 5.73543668e-01 1.24555874e+00
2.20830351e-01 9.69440043e-02 4.85785961e-01 1.51417017e+00
2.68315002e-02 -2.04408512e-01 -5.21859705e-01 -4.80615109e-01
3.13727021e-01 1.09645164e+00 -3.97940695e-01 -5.60865104e-01
-4.96910751e-01 8.05306911e-01 2.55899817e-01 8.53432536e-01
-8.64415884e-01 6.64987369e-03 7.01774597e-01 -5.88529371e-02
4.72078681e-01 -2.89444596e-01 -3.38826269e-01 -1.29594707e+00
-2.17669055e-01 -8.74244571e-01 4.30955946e-01 -9.90737557e-01
-1.76634610e+00 2.48237118e-01 2.10534111e-01 -8.69686902e-01
-8.13533440e-02 -8.40798736e-01 -6.62620246e-01 9.89154220e-01
-1.41434717e+00 -1.61898315e+00 1.55712850e-02 3.97057980e-01
7.84197748e-01 -8.74806046e-02 6.27153158e-01 8.12214389e-02
-4.07505214e-01 5.12414396e-01 -2.52657831e-01 2.50315052e-02
6.99983001e-01 -1.18893111e+00 1.64670810e-01 8.81330311e-01
3.05012137e-01 7.30290473e-01 6.99236631e-01 -6.99806690e-01
-1.50283039e+00 -1.02531505e+00 6.90666199e-01 -4.03520525e-01
9.98045623e-01 -8.99462223e-01 -9.40930486e-01 6.02828622e-01
3.17647725e-01 4.28728871e-02 7.57095098e-01 -1.07391842e-01
-1.05990899e+00 7.82536867e-04 -7.24016666e-01 9.53758776e-01
1.03353667e+00 -1.24626613e+00 -7.59320498e-01 7.17226118e-02
9.80362296e-01 -1.16889365e-01 -2.35756695e-01 3.62894893e-01
6.89449787e-01 -6.67484224e-01 9.93374228e-01 -6.96717739e-01
5.78264892e-01 -2.75183380e-01 -7.08315492e-01 -1.10476303e+00
-5.54017782e-01 -2.76737958e-01 -4.12870109e-01 1.19827628e+00
2.53141463e-01 -3.66861552e-01 3.98624182e-01 6.44736707e-01
-1.71797663e-01 -6.70286655e-01 -8.47344279e-01 -8.63335490e-01
1.34873271e-01 -4.67457891e-01 3.85419697e-01 8.62582922e-01
-1.59616828e-01 3.60947251e-01 -3.27923059e-01 3.14970523e-01
6.73921347e-01 4.41366911e-01 4.65457588e-01 -8.67641687e-01
-6.53029621e-01 -5.79969645e-01 -2.22672820e-01 -1.03041661e+00
4.18543965e-01 -1.22931457e+00 1.54636756e-01 -1.35207117e+00
5.57845652e-01 -2.42771387e-01 -4.70683247e-01 5.76631725e-01
-3.66877228e-01 -3.67224291e-02 6.71829939e-01 8.33126605e-02
-6.33994162e-01 7.60699213e-01 1.00554383e+00 -5.26043296e-01
7.02995136e-02 -5.19915879e-01 -7.43860722e-01 6.53133035e-01
4.37981397e-01 -2.47038171e-01 -6.59856796e-01 -4.24050331e-01
1.44900501e-01 -3.09134722e-01 8.35117459e-01 -6.07279301e-01
2.24023417e-01 -4.27149117e-01 2.83478230e-01 -5.32799780e-01
6.41971350e-01 -5.45489371e-01 -3.41366976e-02 1.99007392e-01
-6.16107881e-01 -3.75784874e-01 7.88565576e-02 8.61048937e-01
1.20471604e-01 -1.87352225e-01 7.48813748e-01 -2.18348801e-01
-9.01227772e-01 3.51227611e-01 -1.95180252e-01 2.01267928e-01
9.54435289e-01 -1.06073543e-01 -5.82466841e-01 -3.46982747e-01
-5.08369982e-01 1.15094833e-01 6.54260337e-01 5.79069138e-01
5.58740139e-01 -1.36870015e+00 -4.17501748e-01 3.26504767e-01
4.02703166e-01 -2.29357421e-01 4.18282866e-01 6.36934519e-01
4.03003804e-02 1.75691724e-01 -2.11145207e-02 -6.47275269e-01
-9.62641954e-01 9.57469583e-01 1.72781351e-04 5.16068675e-02
-5.26908278e-01 8.50663722e-01 9.39710855e-01 -2.36848623e-01
3.10758233e-01 -5.55397570e-01 2.53526926e-01 2.12040037e-01
4.51608270e-01 3.96806337e-02 -5.48361242e-01 -4.72729951e-01
-2.30320871e-01 3.30526948e-01 5.79777472e-02 -3.80903602e-01
9.32987571e-01 -9.21283942e-03 8.14565495e-02 1.03164601e+00
9.39756393e-01 -2.74485908e-03 -1.83165145e+00 -5.28065622e-01
-1.67246219e-02 -2.43890345e-01 -9.55381617e-02 -9.62237120e-01
-6.43570781e-01 1.32985258e+00 4.92356330e-01 -9.37939733e-02
8.33959818e-01 2.81105548e-01 5.79620719e-01 -8.65479279e-03
2.91944444e-01 -9.93889689e-01 3.27121109e-01 4.82272059e-01
7.87621558e-01 -1.08611548e+00 -3.24810565e-01 -1.58304393e-01
-1.01645327e+00 9.41385806e-01 6.69517994e-01 1.10203922e-01
3.98685873e-01 3.10054511e-01 -1.53369671e-02 7.66416565e-02
-1.05268013e+00 -4.19313967e-01 4.36368972e-01 6.14313304e-01
1.19890142e-02 2.91600555e-01 3.64042491e-01 9.64600086e-01
1.48069635e-01 -2.39260167e-01 1.41368270e-01 7.09525108e-01
-4.96039510e-01 -5.99180520e-01 -2.44494781e-01 1.95694745e-01
7.84566700e-02 -5.97692728e-01 -4.74349588e-01 4.19143617e-01
4.14470911e-01 7.18343377e-01 3.01661491e-02 -2.90067971e-01
4.72658724e-02 5.81011713e-01 4.98418838e-01 -8.07647705e-01
-3.37193608e-01 2.88353730e-02 -1.94304988e-01 -3.13818693e-01
5.39183132e-02 -5.13871491e-01 -1.17050064e+00 4.44338061e-02
4.19331379e-02 -4.72832948e-01 4.88267630e-01 8.07276785e-01
2.52867222e-01 3.42907608e-01 2.40301415e-01 -7.90976524e-01
-8.61068547e-01 -6.56607330e-01 -4.51272935e-01 5.59497297e-01
4.47107047e-01 -7.04436660e-01 -5.15695572e-01 9.94830579e-02] | [10.24728775024414, 2.133354663848877] |
ac7707f3-4089-4d55-9222-86590339b934 | dsnet-automatic-dermoscopic-skin-lesion | 1907.04305 | null | https://arxiv.org/abs/1907.04305v2 | https://arxiv.org/pdf/1907.04305v2.pdf | DSNet: Automatic Dermoscopic Skin Lesion Segmentation | Automatic segmentation of skin lesion is considered a crucial step in Computer Aided Diagnosis (CAD) for melanoma diagnosis. Despite its significance, skin lesion segmentation remains a challenging task due to their diverse color, texture, and indistinguishable boundaries and forms an open problem. Through this study, we present a new and automatic semantic segmentation network for robust skin lesion segmentation named Dermoscopic Skin Network (DSNet). In order to reduce the number of parameters to make the network lightweight, we used depth-wise separable convolution in lieu of standard convolution to project the learned discriminating features onto the pixel space at different stages of the encoder. Additionally, we implemented U-Net and Fully Convolutional Network (FCN8s) to compare against the proposed DSNet. We evaluate our proposed model on two publicly available datasets, namely ISIC-2017 and PH2. The obtained mean Intersection over Union (mIoU) is 77.5 % and 87.0 % respectively for ISIC-2017 and PH2 datasets which outperformed the ISIC-2017 challenge winner by 1.0 % with respect to mIoU. Our proposed network also outperformed U-Net and FCN8s respectively by 3.6 % and 6.8 % with respect to mIoU on the ISIC-2017 dataset. Our network for skin lesion segmentation outperforms other methods and can provide better segmented masks on two different test datasets which can lead to better performance in melanoma detection. Our trained model along with the source code and predicted masks are made publicly available. | ['Md. Kamrul Hasan', 'Prasad N. Samarakoon', 'Fakrul Islam Tushar', 'Robert Marti Marly', 'Lavsen Dahal'] | 2019-07-09 | null | null | null | null | ['melanoma-diagnosis', 'skin-lesion-segmentation'] | ['computer-vision', 'medical'] | [ 8.01394403e-01 2.35563308e-01 -8.91990885e-02 -8.68381560e-02
-8.13249588e-01 -4.65467960e-01 3.84299606e-01 4.30011190e-02
-7.39632547e-01 7.29231179e-01 -1.72980815e-01 -3.93763095e-01
-2.76793204e-02 -6.93189919e-01 -4.15790707e-01 -7.32602835e-01
2.33472824e-01 -6.00887462e-02 4.98356611e-01 -1.67214740e-02
1.54971018e-01 4.51222986e-01 -1.12586808e+00 4.55788761e-01
1.23799396e+00 1.16560078e+00 2.00203538e-01 9.27082837e-01
-5.36002815e-02 1.89564735e-01 -2.62894392e-01 -4.50989217e-01
3.59950960e-01 -4.65567738e-01 -9.66587663e-01 2.69741789e-02
5.19416213e-01 -7.15764537e-02 2.24208366e-02 1.28472400e+00
6.02494061e-01 -3.05057466e-01 8.08582902e-01 -8.05360734e-01
-2.01973304e-01 1.47982523e-01 -8.56865346e-01 -7.09877610e-02
-1.22149494e-02 6.64115250e-02 4.70491022e-01 -3.14734161e-01
8.93510401e-01 6.85778975e-01 9.22944546e-01 9.85323370e-01
-7.36384094e-01 -4.51419324e-01 -2.51388639e-01 1.08364359e-01
-1.42989230e+00 -1.68586865e-01 3.27055573e-01 -3.63578588e-01
4.97740448e-01 6.65418863e-01 4.79857445e-01 9.79032636e-01
-5.37761580e-03 7.88959682e-01 1.45809591e+00 -4.84300673e-01
1.59690797e-01 2.31552795e-01 6.24739118e-02 9.92245913e-01
3.18565726e-01 -1.32859781e-01 4.30004112e-02 2.12882236e-01
9.06448066e-01 -1.22611962e-01 -3.85462582e-01 3.22601259e-01
-7.48800397e-01 4.92883474e-01 6.82353258e-01 2.25892112e-01
-2.86244273e-01 5.46495095e-02 3.99756879e-01 -1.47720695e-01
3.85153145e-01 4.30945046e-02 -2.28989825e-01 -1.71318930e-02
-9.05799150e-01 -1.53181478e-01 6.15533829e-01 3.64977717e-01
2.54785240e-01 -2.88459629e-01 -2.84689516e-01 9.44735885e-01
6.60130614e-03 3.84235978e-01 4.49673563e-01 -5.16901433e-01
1.94503814e-01 8.23977292e-01 -1.72145918e-01 -5.93296945e-01
-5.61406374e-01 -5.34499228e-01 -1.13804340e+00 3.68402183e-01
6.88329518e-01 -3.56580168e-01 -1.62305534e+00 1.23675478e+00
2.48178259e-01 3.47986549e-01 1.29120201e-01 9.72229183e-01
9.66346323e-01 2.29903415e-01 2.80631989e-01 2.16845617e-01
1.43731821e+00 -9.18949127e-01 -4.25076127e-01 3.81005667e-02
7.22353756e-01 -8.68365049e-01 6.29664481e-01 2.95219094e-01
-8.44435751e-01 -3.00386816e-01 -9.77566659e-01 3.71833406e-02
-2.88390249e-01 5.43532729e-01 5.30952334e-01 8.80640447e-01
-1.09918189e+00 4.39065933e-01 -8.86217833e-01 -8.86682510e-01
9.44955409e-01 4.30774301e-01 -4.61533993e-01 -1.80108428e-01
-8.36597621e-01 6.66769385e-01 3.77758056e-01 9.63264555e-02
-7.19367266e-01 -7.00526416e-01 -7.10992217e-01 -4.01850313e-01
3.42808276e-01 -3.91824514e-01 8.56397629e-01 -1.17784441e+00
-1.43958402e+00 1.10054684e+00 -4.66293097e-02 -7.47845411e-01
7.94464111e-01 1.27127975e-01 -4.24197406e-01 5.74702382e-01
-4.58356366e-02 1.02385962e+00 4.20169443e-01 -1.04077685e+00
-8.98649096e-01 -2.01584712e-01 1.69028889e-03 1.92656830e-01
-1.71927780e-01 -2.48349234e-01 -7.16018200e-01 -4.31114912e-01
-1.15180440e-01 -9.08887446e-01 -2.77217925e-01 3.43019366e-01
-8.23950827e-01 9.56901610e-02 6.68970823e-01 -1.08730209e+00
1.10079896e+00 -1.95961785e+00 -1.25865772e-01 4.19650465e-01
2.05029314e-03 7.77846515e-01 -3.47958207e-01 3.89541499e-02
-5.13362177e-02 3.66548240e-01 -6.67426288e-01 -2.72681952e-01
-4.41650301e-01 -1.12647496e-01 5.03241360e-01 5.74231207e-01
3.48396510e-01 8.83080244e-01 -4.46949571e-01 -7.17521489e-01
3.62168431e-01 6.04657114e-01 -2.71904439e-01 -1.40326351e-01
-1.92497402e-01 2.33234704e-01 -1.71656087e-01 1.10627282e+00
9.95811880e-01 -4.43342924e-02 6.42619357e-02 -3.05821180e-01
6.04856275e-02 -5.09861887e-01 -1.01033664e+00 1.64168143e+00
-3.61679435e-01 5.82064211e-01 2.79636472e-01 -8.33285332e-01
7.04842865e-01 4.22395796e-01 4.01458383e-01 -7.94685483e-01
3.71942759e-01 3.21130991e-01 1.48065537e-01 -6.57828391e-01
1.13167159e-01 -6.30370975e-02 2.77222246e-01 -1.18730195e-01
-5.70658147e-02 2.00278401e-01 3.28498393e-01 6.67994842e-02
1.00618136e+00 -3.90152596e-02 2.92701513e-01 -3.60213727e-01
9.33143616e-01 1.35010689e-01 5.81486523e-01 2.97926337e-01
-4.57692087e-01 7.62060642e-01 7.08494782e-01 -1.85466215e-01
-6.37982428e-01 -9.58158016e-01 -4.67242330e-01 2.37942502e-01
2.40949452e-01 4.32624556e-02 -1.18406582e+00 -1.02767575e+00
-1.37488797e-01 2.63888448e-01 -9.28655505e-01 3.44512075e-01
-2.57928818e-01 -9.02362525e-01 7.95216918e-01 4.87319738e-01
9.56729293e-01 -8.25301945e-01 -3.67140651e-01 -1.76366851e-01
1.47698358e-01 -1.06613278e+00 -3.74557763e-01 -8.39930922e-02
-5.35251021e-01 -1.78020120e+00 -1.16295934e+00 -9.26127672e-01
1.07622814e+00 -8.32994208e-02 5.87207794e-01 1.76599935e-01
-1.11642194e+00 1.26362741e-02 -2.56495267e-01 -4.08046961e-01
-3.33851814e-01 1.88147306e-01 -6.79025829e-01 2.09516868e-01
2.85441399e-01 -3.44548523e-02 -9.10475194e-01 1.83351904e-01
-1.08896101e+00 3.43665451e-01 9.00400877e-01 9.22918200e-01
7.11610317e-01 3.60328853e-02 5.32146275e-01 -1.35075557e+00
4.25349087e-01 -3.48984092e-01 -6.08638942e-01 4.26113486e-01
-2.64060378e-01 -3.45005304e-01 4.53997225e-01 2.72287671e-02
-1.16972136e+00 4.63793606e-01 -3.99918854e-01 8.64949599e-02
-3.73204112e-01 2.65810311e-01 8.55062753e-02 -4.46129143e-01
6.07718229e-01 3.17364298e-02 2.96565205e-01 -3.56334865e-01
-2.61468329e-02 8.72474194e-01 4.75196779e-01 -6.64881617e-02
5.64620018e-01 6.21987224e-01 2.96711028e-01 -7.82887340e-01
-7.26007283e-01 -5.28490245e-01 -5.65945446e-01 -1.07656531e-01
1.29219103e+00 -7.92515516e-01 -5.80551028e-01 9.64268148e-01
-8.63220632e-01 -4.37060714e-01 6.90470040e-02 1.97097715e-02
2.36318763e-02 4.53729630e-01 -6.29903913e-01 -7.70933986e-01
-6.52324617e-01 -1.21047091e+00 1.04977548e+00 7.93271780e-01
-6.88559115e-02 -1.33616221e+00 -2.72587568e-01 5.57102799e-01
4.89577353e-01 8.71563971e-01 6.29680753e-01 -4.75379944e-01
-2.03685343e-01 -4.03901339e-01 -7.26503730e-01 6.55316591e-01
2.69902319e-01 2.87103266e-01 -1.02411747e+00 -2.04154000e-01
-6.65332794e-01 -1.95199743e-01 1.30421960e+00 5.41092217e-01
1.40048563e+00 1.00979008e-01 -5.28363645e-01 8.98146093e-01
1.98540092e+00 2.55972922e-01 6.99393570e-01 1.81835026e-01
5.51916063e-01 6.25035763e-01 4.22758698e-01 1.06661230e-01
9.29438844e-02 1.58463761e-01 6.51472867e-01 -7.22208023e-01
-6.62251174e-01 -1.15280477e-02 -1.70429572e-01 2.42380902e-01
-2.01746583e-01 -1.74102813e-01 -9.29952323e-01 7.87638605e-01
-1.40662539e+00 -5.33910930e-01 -3.44380021e-01 2.07672358e+00
8.67891192e-01 9.08718854e-02 4.89461571e-02 1.71361730e-01
8.25571239e-01 -1.80461049e-01 -4.89804566e-01 -4.93734211e-01
-1.61409885e-01 8.97046924e-01 9.11070347e-01 6.35615587e-01
-1.38535571e+00 8.46632123e-01 5.01979160e+00 1.24329984e+00
-1.34541833e+00 7.27753900e-03 1.10023439e+00 9.92871448e-02
6.46163151e-02 -3.94144922e-01 -7.19723761e-01 4.79003310e-01
6.02699995e-01 3.52938741e-01 -8.29514861e-03 3.47862750e-01
-1.33203045e-01 -5.89340687e-01 -6.11629307e-01 8.53441060e-01
2.65947431e-02 -1.47616315e+00 -2.23544151e-01 2.65801270e-02
1.06690884e+00 -2.22688556e-01 2.40973279e-01 -1.15631014e-01
-2.50071362e-02 -1.51648748e+00 -2.92054061e-02 5.61353981e-01
1.41581631e+00 -7.62865841e-01 1.23668647e+00 3.47818062e-02
-9.56880689e-01 2.12909296e-01 -2.19106868e-01 3.28150988e-01
-8.42415839e-02 5.34000814e-01 -1.04039156e+00 5.69959342e-01
3.57054770e-01 6.22188449e-01 -8.47628355e-01 1.34547126e+00
-5.15430272e-02 8.20733368e-01 -3.18448871e-01 -1.06213897e-01
4.29602474e-01 -8.16246402e-03 1.96299359e-01 1.25638044e+00
1.97732285e-01 -2.29960471e-01 -1.88794553e-01 6.06932402e-01
-1.25261337e-01 1.32228523e-01 2.98595373e-02 3.42179015e-02
2.01137409e-01 1.58751583e+00 -8.17411602e-01 -1.45504698e-01
-1.66041121e-01 1.20329094e+00 -1.32772997e-01 3.57606560e-01
-7.98497200e-01 -7.11547852e-01 5.25460958e-01 8.39815885e-02
4.36070412e-02 4.15001065e-01 -3.32956403e-01 -7.23316073e-01
-1.06551319e-01 -6.69930220e-01 4.13820833e-01 -1.14338309e-01
-1.18004763e+00 7.64322758e-01 -4.30039555e-01 -1.07455623e+00
3.06971043e-01 -1.00515985e+00 -9.36929107e-01 1.17110705e+00
-1.87922895e+00 -1.36025798e+00 -8.39917719e-01 4.85217333e-01
4.18388307e-01 -1.41027927e-01 9.02470529e-01 1.81888103e-01
-9.17983830e-01 9.03168321e-01 1.05561165e-03 3.35483581e-01
7.01391459e-01 -1.40526617e+00 2.46662647e-01 7.64983594e-01
-4.11983669e-01 6.70123994e-02 4.24822085e-02 -5.70259213e-01
-9.28700686e-01 -1.40780663e+00 3.50454301e-01 3.43087651e-02
4.20746297e-01 -1.14288270e-01 -5.73706448e-01 4.01507802e-02
3.84969771e-01 2.21418664e-01 8.08275998e-01 -3.76778543e-01
2.80316640e-03 -2.18391016e-01 -1.68397677e+00 5.72320163e-01
6.86136961e-01 -2.89281875e-01 2.03682229e-01 3.98426354e-01
4.00952280e-01 -5.99597514e-01 -8.57639432e-01 5.75503469e-01
4.69669938e-01 -9.78592098e-01 6.47062600e-01 -1.72718287e-01
5.20494759e-01 -1.29798129e-01 1.05854208e-02 -1.00998390e+00
6.15422837e-02 -2.53632903e-01 4.53718096e-01 1.09179616e+00
5.38860142e-01 -6.99103296e-01 1.29435420e+00 2.62934774e-01
-1.10970661e-01 -1.40752733e+00 -8.51867497e-01 -3.42753798e-01
2.12343514e-01 -1.95495754e-01 1.74270988e-01 4.63292837e-01
-5.64502716e-01 -3.24600995e-01 9.53298714e-03 9.25150067e-02
8.52081776e-01 -3.25683296e-01 3.10123712e-01 -9.30797279e-01
-9.85509455e-02 -5.03130555e-01 -5.54250062e-01 -5.48431456e-01
-1.91451013e-01 -1.08446515e+00 -2.25592718e-01 -1.70414495e+00
2.25471869e-01 -5.41963696e-01 -4.25379246e-01 5.67949712e-01
-2.89820462e-01 8.32962453e-01 -5.12886187e-03 -2.34659821e-01
-2.88908392e-01 -1.55939907e-01 1.49458230e+00 -2.61413664e-01
-4.73123081e-02 -2.02979194e-03 -7.01042533e-01 7.24238455e-01
1.06415236e+00 1.18289068e-01 -2.06397861e-01 -2.45645955e-01
-4.22016948e-01 -5.58173954e-02 5.46328843e-01 -1.22988653e+00
3.29127401e-01 -3.41221839e-02 7.14457273e-01 -4.92589146e-01
1.72112018e-01 -6.73917413e-01 -1.08908027e-01 8.35549057e-01
-1.93882987e-01 -7.45526195e-01 4.21242654e-01 3.71582180e-01
-2.44051084e-01 -2.72894561e-01 1.12656546e+00 -6.90431595e-02
-9.31538105e-01 3.31253767e-01 -2.77708024e-02 -1.67659059e-01
1.59687543e+00 -5.35346985e-01 -5.01374304e-01 1.86407074e-01
-7.31228769e-01 2.94752687e-01 4.36793327e-01 6.55929819e-02
6.21038914e-01 -9.15467203e-01 -8.16892803e-01 1.92684844e-01
7.75376931e-02 2.27046713e-01 5.93644619e-01 1.08373201e+00
-1.17587972e+00 4.74048704e-01 -3.62720191e-01 -5.16581595e-01
-1.61707711e+00 -1.35202259e-01 6.18416250e-01 -2.16569468e-01
-2.96899289e-01 1.31714785e+00 4.78075929e-02 -3.54409456e-01
3.39938700e-01 -2.92046934e-01 -1.80420697e-01 -2.57831156e-01
3.44801009e-01 4.55371052e-01 9.17612750e-04 -3.96907449e-01
-3.92386466e-01 7.25946128e-01 -2.51248121e-01 2.73546129e-01
1.04687083e+00 2.76523441e-01 -2.21663669e-01 -3.55351210e-01
1.27674270e+00 -2.19813049e-01 -1.30113637e+00 8.48261192e-02
-8.81729368e-03 -2.68138945e-01 8.96800011e-02 -1.55366373e+00
-1.37396061e+00 8.36914778e-01 1.16445637e+00 -2.59048223e-01
1.44634724e+00 -3.63843322e-01 1.03141510e+00 -1.86122745e-01
-6.13941886e-02 -1.13640416e+00 -2.71042436e-01 -1.09809507e-02
5.46233237e-01 -1.37771833e+00 -1.34273335e-01 -9.17753398e-01
-7.05854833e-01 1.10329139e+00 8.34339857e-01 -2.75584966e-01
5.38821399e-01 3.35621536e-01 3.36808741e-01 6.42858148e-02
-3.57781500e-01 -4.52575445e-01 4.92146283e-01 7.64017284e-01
5.47048688e-01 2.91201949e-01 -5.94195187e-01 4.94246125e-01
1.65773019e-01 8.37286785e-02 4.25764382e-01 6.18967175e-01
-2.87091106e-01 -1.24934196e+00 -1.62492320e-01 7.52323925e-01
-7.76176333e-01 9.79564805e-03 -4.85268682e-01 1.01058543e+00
6.81237757e-01 7.35318959e-01 -4.01782244e-02 -2.00940475e-01
-1.59493107e-02 -2.14745075e-01 4.45635647e-01 -5.03223300e-01
-6.41893625e-01 7.29319379e-02 2.20227435e-01 -3.98907900e-01
-3.62631708e-01 -3.11900795e-01 -1.34193909e+00 6.67928681e-02
-1.11255735e-01 -1.18247598e-01 9.29420888e-01 5.80601454e-01
9.67261568e-02 7.09509313e-01 3.40599477e-01 -1.40988559e-01
-3.65644455e-01 -9.22072589e-01 -6.66800737e-01 4.13144290e-01
1.12458028e-01 -2.73944795e-01 -3.50468419e-02 8.44868198e-02] | [15.646245002746582, -2.9501357078552246] |
98359d97-66f9-45c8-a83c-5ab7bffecef5 | socialmapf-optimal-and-efficient-multi-agent | 2210.08390 | null | https://arxiv.org/abs/2210.08390v1 | https://arxiv.org/pdf/2210.08390v1.pdf | SOCIALMAPF: Optimal and Efficient Multi-Agent Path Finding with Strategic Agents for Social Navigation | We propose an extension to the MAPF formulation, called SocialMAPF, to account for private incentives of agents in constrained environments such as doorways, narrow hallways, and corridor intersections. SocialMAPF is able to, for instance, accurately reason about the urgent incentive of an agent rushing to the hospital over another agent's less urgent incentive of going to a grocery store; MAPF ignores such agent-specific incentives. Our proposed formulation addresses the open problem of optimal and efficient path planning for agents with private incentives. To solve SocialMAPF, we propose a new class of algorithms that use mechanism design during conflict resolution to simultaneously optimize agents' private local utilities and the global system objective. We perform an extensive array of experiments that show that optimal search-based MAPF techniques lead to collisions and increased time-to-goal in SocialMAPF compared to our proposed method using mechanism design. Furthermore, we empirically demonstrate that mechanism design results in models that maximizes agent utility and minimizes the overall time-to-goal of the entire system. We further showcase the capabilities of mechanism design-based planning by successfully deploying it in environments with static obstacles. To conclude, we briefly list several research directions using the SocialMAPF formulation, such as exploring motion planning in the continuous domain for agents with private incentives. | ['Joydeep Biswas', 'Arya Anantula', 'Rahul Maligi', 'Rohan Chandra'] | 2022-10-15 | null | null | null | null | ['multi-agent-path-finding', 'social-navigation'] | ['playing-games', 'robots'] | [-1.80600762e-01 6.61436260e-01 -6.01507068e-01 -2.02242255e-01
-4.88907784e-01 -5.28841317e-01 2.81074703e-01 3.88096154e-01
-8.66345167e-01 1.35978794e+00 1.09891288e-01 -5.28668046e-01
-1.01051891e+00 -1.18300235e+00 -4.09517288e-01 -4.51751053e-01
-7.00769126e-01 1.01034284e+00 3.24696988e-01 -5.40553093e-01
3.27121913e-01 5.11924505e-01 -9.55435336e-01 -3.47167522e-01
6.47967041e-01 4.42005515e-01 4.94448870e-01 5.99836171e-01
4.25160639e-02 5.97024798e-01 -5.25896430e-01 1.27826437e-01
5.11903942e-01 1.76226825e-01 -1.00612569e+00 8.30206275e-02
-5.12023032e-01 -6.21667802e-01 -3.96005720e-01 5.44132769e-01
2.96179354e-01 3.73792052e-01 4.38274503e-01 -2.23101568e+00
2.26828307e-01 8.14186454e-01 -4.84188467e-01 3.13195437e-02
6.22629225e-01 3.04175168e-01 7.43603408e-01 -1.18365683e-01
9.27857220e-01 1.28229523e+00 2.53164619e-01 5.06097376e-01
-1.16360533e+00 -2.37808689e-01 5.12891948e-01 2.21382320e-01
-1.15319097e+00 -1.28640756e-01 2.69877940e-01 3.48623097e-02
1.43770337e+00 4.28096771e-01 5.19720852e-01 4.07806605e-01
3.89693886e-01 6.47212446e-01 8.41902316e-01 -1.54709026e-01
4.56789076e-01 -3.28490973e-01 5.25198542e-02 4.16068733e-01
4.75831091e-01 4.04609740e-01 -1.97481349e-01 -5.04562378e-01
6.97832942e-01 -1.56150818e-01 -7.12333471e-02 -7.64055490e-01
-1.34640002e+00 9.91187572e-01 3.92238319e-01 -2.45173872e-01
-7.86903977e-01 6.41425431e-01 1.14295080e-01 3.89505714e-01
-2.94352353e-01 5.85744917e-01 -2.52319843e-01 -2.55415857e-01
-3.68276328e-01 1.09127283e+00 8.26350093e-01 1.10673881e+00
6.72087073e-01 -4.46211576e-01 -5.73873036e-02 3.28018308e-01
4.34341729e-01 4.35587764e-01 -4.79549021e-01 -1.70521164e+00
9.89784718e-01 4.18055803e-01 9.63989258e-01 -9.73109484e-01
-1.10131967e+00 -3.56429927e-02 -2.25236639e-01 6.01942778e-01
5.14393568e-01 -3.62992942e-01 -4.02597129e-01 1.99572885e+00
4.49493498e-01 -2.97125965e-01 3.51295978e-01 8.40539098e-01
1.89645573e-01 7.16391563e-01 -6.48699924e-02 -4.80534554e-01
9.84048426e-01 -1.18204892e+00 -6.52111351e-01 -1.65700451e-01
8.50588918e-01 -4.74465549e-01 5.07272720e-01 -1.43665564e-03
-1.44152272e+00 3.68197590e-01 -6.89546585e-01 4.17142749e-01
-3.11625719e-01 -7.07113266e-01 7.11935103e-01 5.65246820e-01
-1.14111125e+00 3.08942080e-01 -7.28520274e-01 -5.72089493e-01
1.83170795e-01 7.30320513e-01 -2.43893579e-01 -1.43492982e-01
-8.45844328e-01 9.78989005e-01 3.99659157e-01 -3.40271056e-01
-9.22971427e-01 -1.56081632e-01 -7.41480708e-01 3.39631766e-01
1.17418504e+00 -9.12359476e-01 1.36702287e+00 1.44854143e-01
-1.18303382e+00 7.10817650e-02 7.27621987e-02 -5.29996216e-01
6.41577542e-01 3.23571295e-01 1.25931263e-01 -1.00311518e-01
7.51659095e-01 8.27317417e-01 -1.62776440e-01 -1.24400425e+00
-1.07100499e+00 -7.67091811e-02 7.84224451e-01 6.82880700e-01
1.60513058e-01 -2.17563465e-01 6.31434796e-03 -7.42382333e-02
-5.90483658e-02 -1.12080443e+00 -1.08515871e+00 -2.41972193e-01
-5.36454141e-01 -3.17484319e-01 5.14768362e-01 2.41329908e-01
1.09639359e+00 -1.66454160e+00 2.04828847e-02 6.64546013e-01
8.12158883e-02 -3.69380176e-01 -3.83465320e-01 1.06870329e+00
5.49948633e-01 1.61113843e-01 -5.91080151e-02 -1.82999566e-01
3.40258181e-01 7.39302695e-01 2.33432762e-02 3.02760750e-01
-5.57599485e-01 6.78675234e-01 -1.10643280e+00 -3.55233997e-01
2.15997100e-01 -3.74375284e-01 -7.88743973e-01 -2.09950611e-01
1.29149752e-02 3.94026041e-02 -7.83401251e-01 3.18434298e-01
7.59651542e-01 1.20805614e-01 2.65452415e-01 8.03975999e-01
-6.69007659e-01 2.64533699e-01 -1.49406850e+00 1.42157888e+00
-8.34352896e-02 1.92002118e-01 6.64699197e-01 -5.83595514e-01
4.39508259e-01 -5.04262857e-02 8.57074320e-01 -6.04853153e-01
6.13994598e-02 2.77249187e-01 -7.36151710e-02 -1.89397514e-01
9.42169964e-01 1.60307586e-01 -6.00817442e-01 9.48803782e-01
-7.42627084e-01 9.20228660e-02 4.26251411e-01 6.20865881e-01
1.41249990e+00 -4.15026635e-01 3.82875949e-01 -4.52139437e-01
2.82788396e-01 5.80931127e-01 4.83385354e-01 1.28971636e+00
-5.79380095e-01 -3.30178827e-01 4.13452506e-01 -4.76623207e-01
-7.75224626e-01 -1.04611492e+00 3.70763719e-01 8.57270837e-01
9.85975921e-01 -3.04427713e-01 -5.23282766e-01 -5.25306046e-01
3.84852350e-01 8.37162256e-01 -4.59422201e-01 3.07629883e-01
-6.88778460e-01 -5.97964704e-01 9.03983191e-02 3.49150687e-01
5.54389417e-01 -7.67019570e-01 -1.24585414e+00 6.77367806e-01
-4.38520014e-01 -9.45727468e-01 -5.90820968e-01 -5.98771386e-02
-3.65054339e-01 -1.29139829e+00 -2.96697587e-01 -3.57425630e-01
7.80818760e-01 9.75248635e-01 4.34042692e-01 1.67597979e-01
1.09415464e-01 7.32323706e-01 -1.51769668e-01 -2.72411853e-01
-8.50551203e-03 1.26313090e-01 1.76193312e-01 -5.75212777e-01
-3.92043330e-02 -1.02143802e-01 -6.77734673e-01 9.06801581e-01
-3.76385391e-01 1.40088007e-01 1.20864041e-01 5.65543056e-01
3.44967067e-01 5.83194554e-01 7.64064431e-01 -3.32306772e-01
1.03349173e+00 -8.43580008e-01 -6.63615286e-01 6.61180988e-02
-5.57848155e-01 -6.15352802e-02 1.39855653e-01 2.98200492e-02
-8.59322071e-01 -1.53371617e-01 3.87868553e-01 4.34456229e-01
1.02322653e-01 4.69750881e-01 -1.43026471e-01 -1.24322094e-01
2.17392817e-01 -3.34401548e-01 6.99965358e-02 1.07034184e-01
1.71617046e-01 2.76400208e-01 3.26881737e-01 -7.88748562e-01
5.61262310e-01 6.84855223e-01 3.66296053e-01 -3.38679612e-01
2.25099057e-01 -4.73918587e-01 2.01801121e-01 -4.06969726e-01
5.57808936e-01 -5.03527224e-01 -1.61343002e+00 1.76350623e-01
-1.11309886e+00 -6.89781308e-01 -3.35721552e-01 4.34361935e-01
-1.27236211e+00 2.14429617e-01 -3.79770368e-01 -1.23681843e+00
4.12409037e-01 -1.24143481e+00 5.05164742e-01 2.36005679e-01
-2.02926815e-01 -7.93528795e-01 1.04468688e-01 3.03080529e-01
5.13764679e-01 4.40376222e-01 5.95782161e-01 -2.01845959e-01
-1.13400912e+00 2.89734274e-01 -1.74764886e-01 -1.00933290e+00
-3.85233164e-02 -4.76908922e-01 -4.77837324e-02 -3.95037532e-01
-4.59739298e-01 1.58160418e-01 3.08749825e-01 8.46731961e-01
4.04122621e-01 -6.80370390e-01 -1.08032286e+00 -4.91501689e-02
1.37319219e+00 7.27626920e-01 6.39931858e-01 1.20069110e+00
-3.38132642e-02 1.07072592e+00 1.33610082e+00 7.70334184e-01
1.14613914e+00 1.14404476e+00 9.85939682e-01 3.76625992e-02
4.99132514e-01 -2.33438268e-01 3.02245528e-01 -5.22461951e-01
-3.75651360e-01 -6.48512423e-01 -7.65821338e-01 7.91023433e-01
-2.57789087e+00 -1.13317239e+00 -3.15285414e-01 2.19444823e+00
-3.33925411e-02 2.28253543e-01 5.86808562e-01 1.94955632e-01
7.69754052e-01 -2.51585543e-01 -5.01102269e-01 -7.79490113e-01
6.64908113e-03 -7.06067502e-01 1.31332672e+00 1.08834124e+00
-7.57711470e-01 6.86993301e-01 6.62781096e+00 4.93183255e-01
-1.73143953e-01 2.35217333e-01 4.22507375e-01 -3.45049620e-01
-3.88712347e-01 2.81945944e-01 -5.06744206e-01 2.44531780e-01
7.38841832e-01 -6.42833889e-01 8.12637568e-01 8.37943256e-01
9.00712967e-01 -6.22109950e-01 -8.66420567e-01 5.46563685e-01
-7.11335957e-01 -1.48061240e+00 -5.30467033e-01 7.60885477e-01
6.61639094e-01 -8.45436677e-02 -2.65994161e-01 6.26391023e-02
8.46651316e-01 -8.73649538e-01 8.72282147e-01 3.52881625e-02
8.38698894e-02 -1.22139132e+00 7.50432789e-01 6.29977882e-01
-1.34819460e+00 -5.73278129e-01 -3.29123549e-02 -6.22225165e-01
1.05852425e+00 9.60700773e-03 -1.17613244e+00 5.23240030e-01
5.31770349e-01 -1.19911961e-01 4.56130534e-01 1.24111021e+00
4.19940352e-02 -4.82920855e-01 -6.84729040e-01 -1.00987859e-01
7.65243173e-01 -7.16266707e-02 9.37483609e-01 6.19540513e-01
2.91753590e-01 4.47227776e-01 7.63546348e-01 5.45893192e-01
5.42344749e-01 -1.30549267e-01 -8.23247194e-01 3.53817225e-01
7.18070507e-01 7.59338260e-01 -1.00905097e+00 -1.10770211e-01
-6.41304925e-02 3.18083346e-01 -7.32542425e-02 4.71232414e-01
-1.00949895e+00 -1.97537899e-01 9.47214365e-01 3.49336356e-01
-1.18183173e-01 -5.44864714e-01 -3.73266935e-01 -4.10926908e-01
-1.94464505e-01 -2.26275429e-01 5.54388940e-01 -5.89651287e-01
-5.00238419e-01 2.22440556e-01 5.63386381e-01 -8.07528198e-01
-4.62722749e-01 -1.80347875e-01 -7.01100707e-01 4.66862023e-01
-1.74296665e+00 -7.22780764e-01 -2.51047499e-02 6.71035886e-01
5.38941324e-01 -1.46537498e-01 4.64385092e-01 1.32407919e-01
-2.72944897e-01 9.09709781e-02 1.08042704e-02 -5.19280970e-01
4.47233133e-02 -9.63334262e-01 3.10624540e-01 5.96513450e-01
-9.13996994e-01 4.51233655e-01 8.32029223e-01 -7.77395070e-01
-1.43155158e+00 -7.21882582e-01 9.00841713e-01 -1.17075928e-01
4.26880956e-01 1.33804664e-01 -1.13654219e-01 7.01289833e-01
8.24682862e-02 -5.52479506e-01 3.63794357e-01 -6.64954856e-02
5.49589157e-01 2.11751059e-01 -1.69528830e+00 8.57766449e-01
1.49596012e+00 4.00840819e-01 -1.97565630e-01 3.41339171e-01
5.12371778e-01 -3.21042538e-01 -3.66768390e-01 1.30390763e-01
4.58982438e-01 -8.84607792e-01 8.92593324e-01 -4.48487699e-01
-2.63526231e-01 -5.94008088e-01 -2.98157305e-01 -1.46051097e+00
-4.48908806e-01 -1.04744661e+00 4.58161265e-01 5.54240286e-01
4.37318653e-01 -1.11690402e+00 1.03963900e+00 1.20376956e+00
-3.01091313e-01 -5.42754948e-01 -1.43180811e+00 -1.10434318e+00
-1.89758986e-01 -3.42260212e-01 1.01470494e+00 5.97873390e-01
5.19654334e-01 -3.12521815e-01 -3.37109059e-01 4.04659897e-01
8.37438643e-01 -4.79315631e-02 1.03341877e+00 -7.05804408e-01
-1.77194759e-01 -4.85017747e-01 -8.36559087e-02 -9.32558179e-01
-1.20277824e-02 -4.81638998e-01 2.07155813e-02 -2.15038514e+00
1.13719329e-01 -1.19374132e+00 6.94555715e-02 6.15299761e-01
5.80505371e-01 -3.39894533e-01 7.22067893e-01 1.62119702e-01
-8.33082676e-01 2.84217209e-01 1.41982305e+00 -2.25370690e-01
-5.94194412e-01 9.78986546e-02 -7.13230550e-01 4.65420425e-01
9.40805137e-01 -5.48680663e-01 -6.55122757e-01 -3.57337773e-01
3.10371608e-01 8.53889644e-01 1.54583469e-01 -5.59435546e-01
5.32459080e-01 -1.27987289e+00 -6.06534004e-01 -6.69974029e-01
4.87524450e-01 -1.09284091e+00 5.57476163e-01 9.98570025e-01
-1.84811890e-01 3.89382929e-01 6.96512014e-02 6.76556706e-01
2.23958433e-01 -2.33182877e-01 1.75205261e-01 -1.47228301e-01
-7.87899554e-01 7.99473226e-02 -1.00627387e+00 -3.82135332e-01
1.80622292e+00 -5.33717930e-01 -8.14093471e-01 -8.77988756e-01
-7.02389061e-01 1.32004786e+00 5.64512432e-01 1.16704628e-01
6.94805920e-01 -1.25685251e+00 -5.11927128e-01 -2.32947856e-01
-1.77277178e-01 -1.76225513e-01 3.88249695e-01 8.62946153e-01
-4.51809436e-01 6.70758188e-01 -5.08748055e-01 -2.42942065e-01
-1.13065672e+00 6.50073946e-01 1.62519813e-01 -5.82094014e-01
-4.22243983e-01 2.68980026e-01 3.07560384e-01 -4.03157026e-01
3.57206389e-02 -3.15408379e-01 2.23710220e-02 -3.01359653e-01
4.59815592e-01 1.01472759e+00 -5.34675241e-01 -5.22667885e-01
-7.05826402e-01 3.10221970e-01 1.59527496e-01 -6.24782920e-01
1.19747269e+00 -7.08343744e-01 2.24539101e-01 -5.32678187e-01
3.09315115e-01 -6.67096898e-02 -1.30510676e+00 2.98522830e-01
-8.33493248e-02 -8.64744961e-01 -8.88629630e-02 -7.91123331e-01
-8.18099499e-01 -2.36168392e-02 1.55757219e-01 4.68473405e-01
8.81823897e-01 -7.98459575e-02 6.15713358e-01 5.35461664e-01
1.35039008e+00 -1.25445879e+00 -1.71606600e-01 5.16217113e-01
6.51993394e-01 -8.15274715e-01 -7.72187859e-02 -7.86428094e-01
-5.76945186e-01 8.41964185e-01 5.44993401e-01 6.73780441e-02
2.39849493e-01 3.40872586e-01 -3.56311172e-01 -2.64766425e-01
-8.51144075e-01 -3.59117568e-01 -9.15727317e-01 8.82764578e-01
-7.11979747e-01 6.39410317e-01 -7.13658929e-01 2.56877393e-01
-1.38221055e-01 6.76444024e-02 1.33381355e+00 1.43789446e+00
-8.99167776e-01 -1.43241060e+00 -7.82583773e-01 1.92974180e-01
2.58268923e-01 6.09188199e-01 -1.39644518e-01 1.04544044e+00
-1.27185890e-02 1.44215381e+00 1.67786822e-01 5.12076132e-02
4.37051356e-01 -5.77533782e-01 2.99164206e-01 -3.96624565e-01
-2.81519920e-01 5.21412753e-02 7.52706587e-01 -7.46500015e-01
-6.74920380e-01 -7.72769094e-01 -1.75894392e+00 -7.95620620e-01
-7.09450543e-02 4.44398344e-01 5.51437259e-01 7.04465866e-01
2.31637746e-01 3.59771341e-01 6.41519070e-01 -8.72784734e-01
-1.27883315e-01 -1.62056819e-01 -3.87766927e-01 -1.44746289e-01
2.60527968e-01 -1.19090366e+00 -4.18794043e-02 -9.48031306e-01] | [4.976403713226318, 1.7523614168167114] |
d7374ff7-336d-4259-bea3-83b667ebac42 | unsupervised-object-representation-learning | 2210.12918 | null | https://arxiv.org/abs/2210.12918v2 | https://arxiv.org/pdf/2210.12918v2.pdf | Unsupervised Object Representation Learning using Translation and Rotation Group Equivariant VAE | In many imaging modalities, objects of interest can occur in a variety of locations and poses (i.e. are subject to translations and rotations in 2d or 3d), but the location and pose of an object does not change its semantics (i.e. the object's essence). That is, the specific location and rotation of an airplane in satellite imagery, or the 3d rotation of a chair in a natural image, or the rotation of a particle in a cryo-electron micrograph, do not change the intrinsic nature of those objects. Here, we consider the problem of learning semantic representations of objects that are invariant to pose and location in a fully unsupervised manner. We address shortcomings in previous approaches to this problem by introducing TARGET-VAE, a translation and rotation group-equivariant variational autoencoder framework. TARGET-VAE combines three core innovations: 1) a rotation and translation group-equivariant encoder architecture, 2) a structurally disentangled distribution over latent rotation, translation, and a rotation-translation-invariant semantic object representation, which are jointly inferred by the approximate inference network, and 3) a spatially equivariant generator network. In comprehensive experiments, we show that TARGET-VAE learns disentangled representations without supervision that significantly improve upon, and avoid the pathologies of, previous methods. When trained on images highly corrupted by rotation and translation, the semantic representations learned by TARGET-VAE are similar to those learned on consistently posed objects, dramatically improving clustering in the semantic latent space. Furthermore, TARGET-VAE is able to perform remarkably accurate unsupervised pose and location inference. We expect methods like TARGET-VAE will underpin future approaches for unsupervised object generation, pose prediction, and object detection. | ['Tristan Bepler', 'Alireza Nasiri'] | 2022-10-24 | null | null | null | null | ['learning-semantic-representations'] | ['methodology'] | [ 3.61156374e-01 9.51948464e-02 2.57702749e-02 -2.98280627e-01
-5.65039992e-01 -9.06114042e-01 9.54701900e-01 -5.89450300e-01
-4.31270972e-02 5.36779881e-01 1.83423162e-01 4.49164361e-02
-2.62787312e-01 -7.24940658e-01 -1.05182028e+00 -9.99339461e-01
1.15366779e-01 1.13142622e+00 -6.14817254e-02 5.63062914e-02
4.01571244e-02 7.42405832e-01 -1.49843538e+00 1.23540111e-01
2.08876908e-01 7.00239420e-01 3.73993844e-01 6.71143532e-01
4.41475540e-01 5.43936789e-01 -4.36777413e-01 5.47462925e-02
2.41864562e-01 -5.51244736e-01 -1.02641618e+00 6.06706142e-01
5.65752268e-01 -3.40355188e-01 -5.01770735e-01 9.54895973e-01
6.66039959e-02 2.99785048e-01 1.14856517e+00 -1.08637965e+00
-1.01243114e+00 2.90649891e-01 -4.79294986e-01 -1.57795414e-01
1.00469604e-01 -4.96067293e-02 1.18902695e+00 -9.27907765e-01
1.00878191e+00 1.27355802e+00 4.38725978e-01 4.85253870e-01
-1.83392024e+00 -9.57926214e-02 -9.87129379e-03 8.41113329e-02
-1.38316548e+00 -3.85652989e-01 9.56050336e-01 -6.48222506e-01
7.64174581e-01 2.73869693e-01 4.78770405e-01 1.35588992e+00
1.60016298e-01 5.75705528e-01 8.82730424e-01 -3.76820654e-01
4.35783982e-01 -6.76989704e-02 -3.75638872e-01 7.37053573e-01
1.96352139e-01 -4.39037569e-02 -3.76920074e-01 -1.75445691e-01
1.00588536e+00 2.28493929e-01 -2.42134556e-01 -1.00814342e+00
-1.70944166e+00 9.30827200e-01 6.07033134e-01 6.80546537e-02
-1.78864345e-01 2.71699160e-01 1.04983434e-01 -1.59487844e-01
3.79490227e-01 8.07260811e-01 -4.02984262e-01 4.26993966e-01
-7.98726916e-01 3.82155985e-01 6.50832713e-01 1.01703525e+00
8.70504618e-01 2.63053000e-01 1.53719038e-01 3.60126972e-01
3.85596395e-01 7.31941998e-01 5.56444228e-01 -1.29723001e+00
3.31066176e-02 2.90792167e-01 1.01989031e-01 -9.33248401e-01
-3.63318115e-01 -3.27450931e-01 -6.28332853e-01 2.66598076e-01
2.12281376e-01 1.78233892e-01 -1.16002798e+00 2.02577019e+00
2.72054404e-01 2.42983505e-01 7.17018023e-02 1.05416238e+00
6.02168679e-01 6.18375361e-01 -3.24136943e-01 -8.61829612e-03
1.55692685e+00 -7.11258888e-01 -4.81350988e-01 -3.49588305e-01
2.35789433e-01 -5.62330306e-01 8.22432220e-01 1.84352741e-01
-9.81674492e-01 -3.05068612e-01 -1.11774492e+00 -2.60830641e-01
-4.58939940e-01 9.55512747e-03 6.28657758e-01 1.94633752e-01
-8.32854331e-01 5.17321348e-01 -1.12771749e+00 -3.46886396e-01
3.02708328e-01 1.75634548e-01 -6.75479829e-01 -8.02527890e-02
-7.39273369e-01 9.32170093e-01 3.31511497e-01 -2.85953302e-02
-1.10266101e+00 -6.14578903e-01 -1.21244562e+00 -2.08516829e-02
2.33992711e-01 -1.13223493e+00 1.13029945e+00 -8.15697730e-01
-1.20829487e+00 1.08709574e+00 -3.43742609e-01 -1.62684813e-01
3.84969383e-01 9.66754705e-02 -1.25184998e-01 4.58608329e-01
4.11081940e-01 9.05151427e-01 1.28247869e+00 -1.66834128e+00
4.38956767e-02 -7.52744675e-01 -6.82916567e-02 5.26821196e-01
8.40206742e-02 -3.20212722e-01 -1.95673347e-01 -6.23596132e-01
7.13879585e-01 -1.21577847e+00 -1.32454470e-01 1.27345651e-01
-4.97045875e-01 3.46889831e-02 1.25074339e+00 -5.82264304e-01
2.12254345e-01 -2.10410690e+00 8.36789310e-01 -3.71459909e-02
2.84855455e-01 -3.32625806e-01 -4.01809029e-02 1.50159642e-01
-3.88843596e-01 9.01304185e-02 -4.35269952e-01 -4.67420429e-01
1.12240933e-01 7.43605554e-01 -4.72665250e-01 8.51382375e-01
3.49757493e-01 1.04386652e+00 -9.05561507e-01 -3.26929778e-01
3.03985208e-01 5.34112215e-01 -7.14279592e-01 1.62210673e-01
-3.70681733e-01 7.47380435e-01 -3.90360653e-01 6.27482295e-01
5.28683305e-01 -5.45152068e-01 1.30497947e-01 -5.29728115e-01
9.25755724e-02 3.17617655e-01 -1.00622249e+00 1.90625298e+00
-3.15747082e-01 7.24446237e-01 1.02040894e-01 -1.21071494e+00
5.61203837e-01 3.63765240e-01 6.26986504e-01 -2.74048686e-01
-8.04797634e-02 -3.70996147e-02 -1.83283702e-01 -4.54935312e-01
6.08776629e-01 -4.37849373e-01 -9.82496440e-02 6.64088070e-01
4.42771494e-01 -5.64588904e-01 -2.25858212e-01 4.06679779e-01
8.69757414e-01 5.08052230e-01 1.21369898e-01 -1.79947004e-01
-1.35710910e-01 -8.55436251e-02 4.28696930e-01 6.73169434e-01
1.15815543e-01 1.16474247e+00 2.93883532e-01 -4.87484068e-01
-1.47048748e+00 -1.60316801e+00 -5.43915451e-01 7.63831675e-01
1.81963623e-01 -1.67241320e-02 -5.05711496e-01 -3.86840940e-01
4.01973501e-02 7.07984626e-01 -8.37449968e-01 -3.77219915e-01
-4.25637662e-01 -8.01146269e-01 3.95408899e-01 5.64268708e-01
1.34017766e-01 -1.01803124e+00 -5.32879174e-01 -1.24860136e-02
-3.17715645e-01 -1.24527955e+00 -2.72145927e-01 2.86147565e-01
-8.92059267e-01 -9.12517130e-01 -5.83841562e-01 -6.14093065e-01
1.06311953e+00 2.65734434e-01 1.08596075e+00 -3.26926976e-01
-4.20654118e-01 5.37310541e-01 -2.67257571e-01 -1.54920161e-01
-1.90362856e-01 -2.44769022e-01 4.21316683e-01 1.54796347e-01
1.61256827e-02 -8.29129457e-01 -4.93221313e-01 3.81524891e-01
-1.10456288e+00 3.57934721e-02 4.21511233e-01 1.04402900e+00
7.14476645e-01 5.75608462e-02 6.29706457e-02 -7.25868821e-01
-4.17377204e-02 -5.75033009e-01 -3.08872074e-01 3.26431282e-02
-2.95396984e-01 3.67802113e-01 3.40204775e-01 -4.40357298e-01
-9.23360705e-01 3.75537157e-01 1.26158774e-01 -8.16443980e-01
-3.03125143e-01 2.37216160e-01 -2.38680825e-01 5.28105162e-02
7.48337507e-01 4.77606952e-01 1.14793479e-01 -4.61353272e-01
6.53049409e-01 1.82308853e-01 7.55698979e-01 -7.23818064e-01
1.17957902e+00 9.83601987e-01 2.15469748e-01 -8.61754239e-01
-8.16498220e-01 -1.73459560e-01 -1.04132736e+00 1.29051194e-01
1.17511499e+00 -1.00732851e+00 -5.31719744e-01 2.65760601e-01
-1.21687281e+00 -7.44642317e-02 -4.52676743e-01 5.37646830e-01
-8.75113547e-01 3.67530763e-01 -3.65891665e-01 -3.65735918e-01
2.57578731e-01 -1.35494304e+00 1.54510093e+00 -7.43864402e-02
-2.77756751e-01 -1.03905046e+00 -1.12376899e-01 4.29411113e-01
7.98829049e-02 4.76160675e-01 9.86975312e-01 -4.38331485e-01
-8.74054551e-01 -3.68186906e-02 -3.76854911e-02 2.18533069e-01
2.26458684e-01 -6.66018501e-02 -9.35510278e-01 -3.24101508e-01
1.92604288e-01 -4.40800816e-01 8.82297277e-01 4.50117230e-01
1.12818849e+00 -5.46157718e-01 -2.91870892e-01 9.40211475e-01
1.20938814e+00 -1.62983850e-01 3.29979092e-01 8.49777311e-02
1.01718771e+00 6.62860751e-01 -6.93420041e-03 2.19886929e-01
2.43105382e-01 8.65308702e-01 7.14143038e-01 1.05278425e-01
-5.73142879e-02 -4.08116609e-01 3.51795554e-01 5.61922073e-01
-2.98109472e-01 2.22051889e-02 -6.59684896e-01 4.87535208e-01
-1.70376861e+00 -1.13895607e+00 1.44009843e-01 1.96878159e+00
5.75673640e-01 -3.36147547e-01 -2.27349088e-01 -6.50780722e-02
5.13512313e-01 4.48799849e-01 -8.61949801e-01 -1.19058217e-03
-3.20640236e-01 1.35899007e-01 4.06985521e-01 4.37827945e-01
-9.86248910e-01 7.35350847e-01 6.64204884e+00 3.34202349e-01
-1.07394433e+00 1.77502990e-01 3.05532336e-01 -1.14998817e-01
-6.54381216e-01 1.28267080e-01 -3.54571909e-01 1.73928156e-01
5.06073177e-01 2.30306946e-02 6.36982560e-01 1.04691064e+00
-1.35003045e-01 3.42161566e-01 -1.46926022e+00 8.36639166e-01
4.78082478e-01 -1.43793416e+00 3.04585040e-01 2.38479048e-01
7.47077942e-01 2.97115874e-02 3.65069956e-01 1.18355108e-02
2.97342837e-01 -1.11087620e+00 9.86975968e-01 3.63653630e-01
8.87005150e-01 -4.26986724e-01 3.46543282e-01 2.88085252e-01
-8.21711063e-01 2.22518310e-01 -4.66173172e-01 1.27710015e-01
1.63414046e-01 3.36690545e-01 -8.25989306e-01 5.71767628e-01
6.27141237e-01 7.96788573e-01 -2.76453167e-01 2.33727679e-01
-4.20921117e-01 4.35350716e-01 -3.15841496e-01 3.24313372e-01
1.54597029e-01 -3.23345691e-01 9.57369745e-01 6.75999761e-01
3.16839844e-01 2.17252180e-01 2.90159844e-02 1.54368174e+00
-1.05694361e-01 -6.47898734e-01 -8.52467656e-01 5.39954379e-03
3.10114503e-01 1.25808620e+00 -7.32439637e-01 -1.66124418e-01
-1.33955285e-01 1.20449936e+00 1.20392606e-01 7.28069603e-01
-6.72453046e-01 -8.57947022e-03 7.78173447e-01 8.92495587e-02
6.12660348e-01 -4.55908895e-01 -2.05194071e-01 -1.53176439e+00
3.85214500e-02 -6.13064349e-01 7.27934763e-02 -1.31310117e+00
-1.27561057e+00 4.65321302e-01 2.17477739e-01 -1.16726375e+00
-5.09899497e-01 -9.80821908e-01 -5.24499536e-01 6.94244444e-01
-8.64263296e-01 -1.44187188e+00 -9.02417749e-02 5.66502154e-01
4.00414288e-01 -1.13151550e-01 1.02868509e+00 -1.06600963e-01
-2.42770419e-01 1.28143549e-01 2.49342352e-01 1.88802838e-01
4.56210762e-01 -1.47566140e+00 1.47033349e-01 7.43538558e-01
7.68035650e-01 7.64688075e-01 9.21372652e-01 -3.11042190e-01
-1.62407577e+00 -1.12673318e+00 4.33966309e-01 -1.02741349e+00
5.66117704e-01 -7.70338118e-01 -7.41531134e-01 1.33547151e+00
-7.32021928e-02 3.22674513e-01 4.47232932e-01 1.24870867e-01
-6.68647766e-01 2.16664225e-01 -1.07995462e+00 5.95257640e-01
1.00456977e+00 -9.84506130e-01 -8.70626450e-01 7.77142227e-01
8.85815918e-01 -5.83738983e-01 -7.32470393e-01 2.21932411e-01
5.15550911e-01 -8.79316807e-01 1.35694754e+00 -9.10317242e-01
5.67500353e-01 -3.88219684e-01 -4.40912843e-01 -1.36681223e+00
-6.55069411e-01 -1.28801987e-01 -2.51343101e-01 6.69941902e-01
2.47471482e-01 -5.54069221e-01 7.77969778e-01 5.37727594e-01
-2.22453967e-01 -5.46670318e-01 -1.04371798e+00 -6.99516535e-01
4.52100895e-02 -2.58702129e-01 4.89243686e-01 1.10851932e+00
-4.30800140e-01 3.86956573e-01 -3.20194125e-01 5.87498605e-01
8.58458281e-01 4.22506273e-01 5.08461237e-01 -1.04952598e+00
-5.72849751e-01 -1.16517201e-01 -7.21703351e-01 -1.12201226e+00
5.12094676e-01 -1.02176571e+00 3.18355076e-02 -1.27679324e+00
3.42746556e-01 -6.78309053e-02 4.58912365e-02 5.05761147e-01
2.03232154e-01 3.96912664e-01 4.18102788e-03 5.32316804e-01
-4.25823122e-01 8.24012160e-01 1.23039782e+00 -3.19643408e-01
1.76554278e-01 -2.17347890e-01 -5.14465392e-01 7.90216088e-01
3.23433280e-01 -6.22772634e-01 -3.87973636e-01 -6.13944232e-01
2.55973071e-01 3.21902074e-02 8.85796249e-01 -6.14270747e-01
3.61385308e-02 -2.08956361e-01 7.72026122e-01 -4.44579214e-01
7.34716713e-01 -7.77939498e-01 2.72015572e-01 1.97605163e-01
-1.14540853e-01 -5.05069718e-02 -2.20709443e-01 8.03475857e-01
-1.48392230e-01 -1.69082016e-01 6.92097306e-01 -3.75126332e-01
-5.08322537e-01 3.66234750e-01 -3.54515851e-01 -1.51652381e-01
8.89774501e-01 -3.33016366e-01 -2.66553968e-01 -3.50676358e-01
-9.74403858e-01 -1.98929101e-01 6.76769555e-01 4.41722155e-01
6.42013788e-01 -1.54674459e+00 -5.89798391e-01 3.66754293e-01
2.59429812e-01 6.25558197e-01 2.00904384e-01 5.89952946e-01
-4.30811644e-01 2.13751853e-01 -3.17424148e-01 -9.57296848e-01
-9.37781334e-01 4.12361771e-01 4.93801147e-01 1.20431893e-01
-7.15355098e-01 7.10122287e-01 5.88734925e-01 -9.27060902e-01
-3.44477177e-01 -1.86840549e-01 2.00604558e-01 -1.41469076e-01
2.80006617e-01 -3.95091176e-02 -1.03427149e-01 -1.04677153e+00
-3.02518189e-01 5.30299842e-01 -5.26065845e-03 -3.76791179e-01
1.52280164e+00 2.03852523e-02 -2.72593379e-01 6.15158975e-01
1.31569195e+00 -1.01162732e-01 -1.43561733e+00 -3.57693613e-01
-5.58833480e-01 -3.47366303e-01 2.64108386e-02 -3.98135453e-01
-1.01086462e+00 7.56137609e-01 1.88360289e-01 -2.17616968e-02
7.96894968e-01 5.79749823e-01 3.10592115e-01 5.15206695e-01
3.23267996e-01 -6.87056303e-01 4.07609642e-01 4.17770028e-01
1.14459956e+00 -1.12040150e+00 1.52546197e-01 -9.31107178e-02
-5.15732229e-01 1.00707448e+00 4.89024192e-01 -1.91923186e-01
5.88582814e-01 5.17128967e-03 -1.76029831e-01 -4.93971378e-01
-5.59605241e-01 1.01080112e-01 4.94282246e-01 7.63669431e-01
8.82344767e-02 2.26050481e-01 5.34366190e-01 3.07598770e-01
-3.34845811e-01 -7.39158452e-01 4.43481892e-01 7.92597055e-01
-1.95441738e-01 -7.16120005e-01 -5.96400797e-01 1.96906328e-01
-6.07802495e-02 1.84171036e-01 -3.79079759e-01 6.53435767e-01
2.09400102e-01 4.26976919e-01 4.13170934e-01 -2.62902856e-01
5.80062978e-02 1.83208689e-01 8.27362061e-01 -8.28011334e-01
3.28678846e-01 2.71787364e-02 -2.87231088e-01 -4.70974475e-01
-5.84922493e-01 -8.70227337e-01 -1.34320307e+00 -3.01225968e-02
-1.64803490e-01 5.23402262e-03 7.37946868e-01 1.05294394e+00
3.24569732e-01 4.22993720e-01 5.96676052e-01 -1.40380442e+00
-7.39845634e-01 -9.09782231e-01 -8.40448081e-01 8.65572393e-01
5.27476966e-01 -9.08982873e-01 -6.68166220e-01 4.52905327e-01] | [8.836466789245605, -3.070441484451294] |
b24e13c3-39f1-4188-8c00-5abe1f138a95 | an-empirical-study-on-challenging-math | 2306.01337 | null | https://arxiv.org/abs/2306.01337v2 | https://arxiv.org/pdf/2306.01337v2.pdf | An Empirical Study on Challenging Math Problem Solving with GPT-4 | Employing Large Language Models (LLMs) to address mathematical problems is an intriguing research endeavor, considering the abundance of math problems expressed in natural language across numerous science and engineering fields. While several prior works have investigated solving elementary mathematics using LLMs, this work explores the frontier of using GPT-4 for solving more complex and challenging math problems. We evaluate various ways of using GPT-4. Some of them are adapted from existing work, and one is MathChat, a conversational problem-solving framework newly proposed in this work. We perform the evaluation on difficult high school competition problems from the MATH dataset, which shows the advantage of the proposed conversational approach. | ['Qingyun Wu', 'Chi Wang', 'Richard Peng', 'Yin Tat Lee', 'Yue Wang', 'Erkang Zhu', 'Hangyu Li', 'Shaokun Zhang', 'Feiran Jia', 'Yiran Wu'] | 2023-06-02 | null | null | null | null | ['elementary-mathematics'] | ['reasoning'] | [-7.76764005e-03 5.79750016e-02 -7.84022268e-03 -2.35604495e-01
-5.03947020e-01 -5.66987634e-01 6.49373770e-01 5.16530037e-01
-1.92022443e-01 7.26874471e-01 1.78546399e-01 -5.48135996e-01
-6.15626574e-01 -1.06662965e+00 -6.22832716e-01 -2.69865960e-01
-1.00968093e-01 6.66409552e-01 2.36432403e-01 -5.14878571e-01
7.68096626e-01 1.45102277e-01 -1.28954470e+00 3.62931013e-01
1.26257586e+00 2.93116093e-01 2.56415516e-01 8.30553472e-01
-6.43119633e-01 1.33679760e+00 -6.58457637e-01 -7.24622309e-01
1.28294170e-01 -3.37263405e-01 -1.31121683e+00 -2.88836747e-01
7.51095414e-01 3.06569636e-01 -1.91833928e-01 8.85751128e-01
3.44518900e-01 7.27018714e-01 2.53315985e-01 -1.34529960e+00
-8.12523961e-01 1.04036379e+00 -3.57001960e-01 4.32181746e-01
7.57629871e-01 4.39182147e-02 1.00083113e+00 -4.18825150e-01
4.10490394e-01 1.67800367e+00 4.12035853e-01 3.67006004e-01
-9.25356925e-01 -6.46538734e-01 1.82225242e-01 6.35297358e-01
-1.32773077e+00 7.60523826e-02 7.06877112e-01 -4.01696771e-01
9.60168660e-01 4.30805296e-01 8.36446822e-01 8.54301929e-01
4.34682071e-01 8.26082647e-01 1.50934160e+00 -4.55855161e-01
4.12024595e-02 1.77183107e-01 9.30463850e-01 6.43543422e-01
1.55939406e-03 -5.25139332e-01 -6.10177815e-01 -1.26593381e-01
5.12050688e-01 -3.06612998e-01 -3.33100930e-02 5.06637059e-02
-1.28635883e+00 7.29338944e-01 -8.40193480e-02 4.98792380e-01
-2.80912425e-02 -4.63574268e-02 1.58928722e-01 7.81234562e-01
5.96019566e-01 7.75295019e-01 -2.15540618e-01 -7.10057735e-01
-7.07870066e-01 7.80311048e-01 1.36327958e+00 1.00102997e+00
4.40241009e-01 -3.89916301e-01 -3.25267464e-01 9.22242463e-01
2.73614526e-01 9.26366542e-03 3.15325618e-01 -9.80115116e-01
9.03551877e-01 9.83025670e-01 -3.53525430e-01 -1.22702408e+00
-1.80090666e-01 -3.20765287e-01 -5.09588242e-01 -3.62195551e-01
4.80033606e-01 -9.43265706e-02 -2.49679923e-01 1.54800498e+00
3.37131709e-01 7.44701922e-01 2.62308776e-01 5.83362699e-01
1.44560456e+00 9.82830644e-01 4.21378523e-01 5.49735501e-02
1.30263209e+00 -1.48360467e+00 -7.20624268e-01 -1.70169279e-01
7.27949381e-01 -1.10632145e+00 9.12577391e-01 5.03992379e-01
-1.36889219e+00 -4.09764707e-01 -6.30864322e-01 -3.88874859e-01
-5.09205520e-01 -2.54393876e-01 1.06393433e+00 9.33756828e-01
-1.12459099e+00 5.18383861e-01 -4.17122126e-01 -5.75898230e-01
2.07018465e-01 2.15234324e-01 -1.18488118e-01 -2.21462935e-01
-1.10032463e+00 8.96557391e-01 3.25943053e-01 -1.36918709e-01
-4.77995723e-01 -1.09530365e+00 -7.38780975e-01 3.79892349e-01
6.04947031e-01 -5.36334038e-01 1.37166381e+00 -1.76627368e-01
-1.90916479e+00 9.05666053e-01 -9.08878371e-02 -2.97635257e-01
3.97137851e-01 -2.17503965e-01 3.51694338e-02 -1.62574023e-01
-7.64925852e-02 4.39294457e-01 -1.18837738e-03 -7.27961719e-01
-2.66350240e-01 1.73264870e-03 6.24261320e-01 3.37892652e-01
-5.41569173e-01 1.01632878e-01 -2.64306843e-01 -6.37944818e-01
5.28022796e-02 -7.91399777e-01 -4.33486760e-01 -6.70813441e-01
-2.85049617e-01 -9.61061597e-01 4.39511687e-01 -3.60652596e-01
1.38224494e+00 -1.41325569e+00 5.02046883e-01 1.04234643e-01
2.95014381e-01 3.60962242e-01 -3.02465230e-01 8.78204882e-01
-4.42592576e-02 3.55860054e-01 1.10391296e-01 -4.16032821e-01
3.34007591e-01 3.15507129e-02 -3.03745955e-01 -4.31797430e-02
-2.01135039e-01 1.12974739e+00 -8.39412689e-01 -6.95967317e-01
4.15149450e-01 2.31243204e-02 -6.83623374e-01 2.95898050e-01
-5.57817996e-01 3.45786184e-01 -7.41960108e-01 3.20581466e-01
7.76054859e-01 -3.57125461e-01 7.05729052e-02 5.59166968e-01
-1.60206839e-01 6.29059732e-01 -1.35648608e+00 1.88716781e+00
-3.92805666e-01 4.83879596e-01 -1.55056387e-01 -1.22828782e+00
7.84924448e-01 3.44788224e-01 4.32383299e-01 -4.10452902e-01
2.46227488e-01 -2.66573071e-01 3.21846634e-01 -8.50903511e-01
5.13149917e-01 1.41572326e-01 2.09004313e-01 7.79130697e-01
2.22106457e-01 -7.36013472e-01 5.88728726e-01 4.71927553e-01
1.18301988e+00 5.70633151e-02 1.13201976e-01 -5.74268997e-01
9.38266933e-01 -3.90631258e-02 1.85763970e-01 1.02340949e+00
-2.16017500e-01 -1.01081077e-02 5.00283778e-01 -3.78966212e-01
-5.15879512e-01 -6.74844205e-01 3.80783565e-02 1.21160996e+00
-1.00133955e-01 -1.14892042e+00 -6.84179306e-01 -2.72378325e-01
-2.11759940e-01 6.16188467e-01 5.98957529e-03 2.32361004e-01
-7.63952851e-01 -6.16630852e-01 3.49972576e-01 -2.46213868e-01
8.57521653e-01 -1.12491024e+00 -2.47967113e-02 2.93116599e-01
-4.10248846e-01 -1.51479089e+00 2.08278839e-02 -4.91917849e-01
-9.03375268e-01 -8.79592121e-01 -4.80902910e-01 -1.08207047e+00
2.18058065e-01 4.72537518e-01 1.44551694e+00 3.48465145e-01
-2.40354359e-01 8.79266977e-01 -4.32584316e-01 -6.19699121e-01
-4.63849872e-01 3.42131168e-01 -2.52583474e-01 -4.15729880e-01
6.97032452e-01 -9.83419955e-01 -9.70085561e-02 -1.47351176e-01
-3.75086635e-01 2.77257621e-01 4.10626203e-01 6.69207722e-02
2.42435075e-02 3.51740271e-01 5.06953537e-01 -1.05279720e+00
1.26884604e+00 -6.16935849e-01 -5.00757098e-01 3.60903740e-01
-1.84243917e-01 -2.77069032e-01 5.03754795e-01 -5.13065755e-01
-9.62560415e-01 -5.43247938e-01 -4.15305555e-01 -3.35193276e-02
-3.35585088e-01 7.98925042e-01 3.09154451e-01 -6.18605196e-01
2.50731200e-01 2.18121827e-01 -5.29307485e-01 -4.29031879e-01
1.58513069e-01 3.66371274e-01 4.32331748e-02 -1.63415670e+00
7.76547015e-01 -3.13627541e-01 1.44389600e-01 -1.15636849e+00
-1.01022089e+00 -4.41488683e-01 -2.96709061e-01 -3.70993555e-01
6.46897435e-01 -7.22472370e-01 -1.69335032e+00 3.75496507e-01
-1.37099802e+00 -2.79849857e-01 1.92343831e-01 6.03259206e-01
-2.29416579e-01 6.08284712e-01 -1.01894212e+00 -9.46022213e-01
-2.99001664e-01 -1.39012873e+00 6.99532211e-01 5.15933394e-01
-3.91191721e-01 -1.27502882e+00 3.05717528e-01 1.27165771e+00
4.93858784e-01 -5.35783693e-02 1.37164152e+00 -7.48591006e-01
-1.05129337e+00 3.35184373e-02 -3.21129039e-02 -3.15577239e-02
-3.04202348e-01 -1.32846028e-01 -8.54367077e-01 4.19073813e-02
2.31951326e-01 -6.77287579e-01 4.12204385e-01 4.59740832e-02
1.38867474e+00 -3.21435601e-01 -7.80185089e-02 -2.13459749e-02
1.11257136e+00 -5.20402659e-03 4.97989297e-01 3.85052353e-01
5.90024650e-01 6.68315232e-01 2.51240820e-01 7.23948628e-02
1.04474652e+00 3.85202050e-01 -1.36535004e-01 4.70932573e-01
2.20983610e-01 -2.39066184e-01 3.03191513e-01 1.51867199e+00
-3.34467083e-01 -1.83723167e-01 -1.36409092e+00 3.78653705e-01
-1.90481997e+00 -9.92166996e-01 -5.35163403e-01 1.51478875e+00
1.10273254e+00 1.55087665e-01 -1.84038818e-01 -1.23976700e-01
2.75172621e-01 1.67227611e-01 1.52871087e-01 -6.80173874e-01
1.49851277e-01 9.99234378e-01 -2.40280315e-01 8.42139900e-01
-8.02805603e-01 1.28262758e+00 6.47851992e+00 1.30840051e+00
-7.08967686e-01 -1.05042523e-03 5.05752325e-01 3.28602999e-01
-9.26752202e-03 9.09924507e-02 -8.00987065e-01 1.23548716e-01
1.13457608e+00 -5.79464734e-01 4.81571883e-01 6.83013499e-01
2.69634843e-01 -2.68079102e-01 -1.28823161e+00 8.92119229e-01
2.72789765e-02 -1.38989937e+00 3.74150962e-01 -1.23876475e-01
7.93628752e-01 -3.89702290e-01 -5.66874333e-02 8.92319441e-01
5.37826955e-01 -1.20088255e+00 2.47011498e-01 4.32595700e-01
-4.05260146e-01 -6.53568566e-01 5.48962712e-01 9.22745943e-01
-1.25656366e+00 9.11760181e-02 -2.42272064e-01 -9.27013874e-01
-1.67370066e-01 3.45640808e-01 -7.29482412e-01 6.64111018e-01
6.43216848e-01 7.57032394e-01 -8.26999068e-01 1.19545913e+00
-2.56609827e-01 9.59461510e-01 -1.69629812e-01 -5.56845903e-01
3.04927796e-01 -7.72813439e-01 6.24979854e-01 1.20441782e+00
6.18012287e-02 7.53351569e-01 6.74674332e-01 1.31166887e+00
-8.73282254e-02 4.80727136e-01 -5.65493524e-01 -5.02102494e-01
4.39497679e-01 1.51495016e+00 -6.64980710e-01 -2.99590498e-01
-4.71150488e-01 3.84819984e-01 5.86965442e-01 1.40915200e-01
-6.75454497e-01 -1.55433178e-01 3.80833924e-01 -1.71842545e-01
-4.57721293e-01 -7.50755131e-01 -4.16531861e-01 -1.31145597e+00
2.31310655e-03 -1.28049707e+00 3.27333152e-01 -7.04076946e-01
-1.52903390e+00 1.36501640e-01 3.41459274e-01 -6.33760273e-01
2.49516174e-01 -4.71728802e-01 -1.16525126e+00 8.73843014e-01
-1.40231597e+00 -1.06358612e+00 -2.99643427e-01 4.13772643e-01
1.04583919e+00 -1.95427552e-01 8.79204094e-01 3.70277137e-01
-6.94999874e-01 3.69479179e-01 -4.64417785e-01 -1.68131396e-01
5.39426863e-01 -1.22684658e+00 3.77439111e-01 5.69045842e-01
1.33654580e-01 9.60007489e-01 8.25507104e-01 -5.24631202e-01
-1.81179178e+00 -6.97228730e-01 1.32747495e+00 -7.98039317e-01
9.29728031e-01 -6.94192111e-01 -1.05839288e+00 8.21564257e-01
7.58789301e-01 -6.68839037e-01 8.11316729e-01 3.61473531e-01
1.11848779e-01 3.78805012e-01 -9.47168827e-01 8.08850467e-01
9.37523365e-01 -1.89934373e-01 -1.08099151e+00 8.04245234e-01
9.39143717e-01 -7.81412363e-01 -1.22128570e+00 1.10384770e-01
6.25592694e-02 -6.52982295e-01 1.22635353e+00 -7.93397129e-01
9.08274174e-01 2.32602298e-01 1.84182059e-02 -1.28032172e+00
-1.68750674e-01 -8.34773719e-01 -2.74695493e-02 1.54092097e+00
1.17182903e-01 -4.79541034e-01 1.09092844e+00 8.33443642e-01
-5.42651676e-02 -7.61921823e-01 -7.00950861e-01 -5.21002173e-01
7.53199697e-01 -6.61740661e-01 4.37527895e-01 1.47721100e+00
4.71355408e-01 6.99966013e-01 -1.24672689e-01 8.46104845e-02
5.26178479e-01 3.10102493e-01 1.08007169e+00 -1.28966057e+00
-3.45553219e-01 -5.87128997e-01 -2.87612051e-01 -1.17020285e+00
5.84299922e-01 -1.26332641e+00 -4.92693156e-01 -1.59995127e+00
2.60932595e-01 -3.87024134e-01 2.32500285e-01 3.22536111e-01
-2.45245054e-01 -1.59236580e-01 4.78007227e-01 -3.46146524e-01
-9.96176064e-01 5.77954888e-01 1.58033431e+00 -2.53522456e-01
-2.00566128e-01 6.94774557e-03 -5.82538366e-01 7.72398233e-01
8.18861663e-01 -2.33391210e-01 -4.84785318e-01 -4.34984595e-01
4.51188207e-01 2.10491538e-01 1.69640556e-01 -1.19992435e+00
6.28550231e-01 -6.65376067e-01 -3.30429487e-02 -3.71939003e-01
4.46036071e-01 -6.56624436e-01 -3.14540654e-01 1.28776371e-01
-4.77688432e-01 1.89278811e-01 5.66966951e-01 2.90494442e-01
-2.64702797e-01 -3.81031394e-01 4.48965132e-01 -5.11400282e-01
-6.37234628e-01 1.89744979e-01 -5.87071240e-01 1.96536958e-01
1.12912130e+00 1.63155019e-01 -5.96937537e-01 -3.17467809e-01
-6.72076702e-01 7.45407343e-01 -5.18635273e-01 7.13414252e-01
5.00847757e-01 -1.25635898e+00 -8.56697917e-01 -6.40354827e-02
-4.85452823e-02 -3.72296311e-02 3.50781232e-01 8.09789717e-01
-5.85404575e-01 9.48985994e-01 -1.53951392e-01 -5.30172467e-01
-1.57420003e+00 3.10433149e-01 1.42509669e-01 -6.82738781e-01
-5.50230026e-01 1.01564324e+00 -6.62614405e-02 -1.02701294e+00
4.28399712e-01 -5.31263649e-01 -5.48828483e-01 -3.75204742e-01
3.75554562e-01 5.98859251e-01 -2.06170813e-03 -8.30838904e-02
7.31823221e-02 6.30377471e-01 -7.20406622e-02 2.41740867e-01
1.72212827e+00 4.42912150e-03 -7.01725841e-01 6.17418468e-01
6.51952565e-01 1.32003287e-02 2.82556955e-02 -4.67498332e-01
3.27259809e-01 -3.60953510e-01 -1.95129156e-01 -7.75564611e-01
-5.61862230e-01 9.89002287e-01 1.39257610e-01 3.28322411e-01
5.15426219e-01 -2.29938194e-01 6.06638134e-01 9.92548585e-01
5.40815949e-01 -7.25106001e-01 2.05828026e-01 1.17639410e+00
7.66769171e-01 -1.28314412e+00 3.76969622e-03 -9.10142660e-01
-2.14136913e-01 1.25895047e+00 1.05221868e+00 -2.30762675e-01
7.33662128e-01 1.29797578e-01 -3.96604151e-01 -5.72912753e-01
-1.09492791e+00 1.92444518e-01 2.40869284e-01 -2.50106324e-02
9.78480637e-01 -5.53706288e-02 -6.74722910e-01 8.31179202e-01
-7.48524487e-01 2.28947431e-01 6.46559894e-01 1.22293961e+00
-4.77606237e-01 -1.39151430e+00 -5.15408874e-01 1.91045597e-01
-4.43794042e-01 -2.68112153e-01 -8.43743324e-01 6.43935323e-01
1.40898705e-01 1.24477506e+00 -3.25946838e-01 -2.50816673e-01
2.93739825e-01 3.07208121e-01 7.69803226e-01 -9.82354045e-01
-1.10513389e+00 -6.32534742e-01 9.31560844e-02 -3.77164006e-01
-4.96823996e-01 -2.81899512e-01 -9.63192582e-01 -1.06339943e+00
-2.99518760e-02 5.30996382e-01 2.95905679e-01 1.23817086e+00
5.64566143e-02 6.01502717e-01 4.08571325e-02 -6.43634975e-01
-3.25363308e-01 -1.06669772e+00 -3.17246318e-01 1.61595255e-01
7.81539921e-03 -3.93910587e-01 -8.92071873e-02 -1.56318188e-01] | [9.741873741149902, 7.381670951843262] |
5e4209e9-1383-4b9a-bc40-336cd2a57e72 | domainstudio-fine-tuning-diffusion-models-for | 2306.14153 | null | https://arxiv.org/abs/2306.14153v1 | https://arxiv.org/pdf/2306.14153v1.pdf | DomainStudio: Fine-Tuning Diffusion Models for Domain-Driven Image Generation using Limited Data | Denoising diffusion probabilistic models (DDPMs) have been proven capable of synthesizing high-quality images with remarkable diversity when trained on large amounts of data. Typical diffusion models and modern large-scale conditional generative models like text-to-image generative models are vulnerable to overfitting when fine-tuned on extremely limited data. Existing works have explored subject-driven generation using a reference set containing a few images. However, few prior works explore DDPM-based domain-driven generation, which aims to learn the common features of target domains while maintaining diversity. This paper proposes a novel DomainStudio approach to adapt DDPMs pre-trained on large-scale source datasets to target domains using limited data. It is designed to keep the diversity of subjects provided by source domains and get high-quality and diverse adapted samples in target domains. We propose to keep the relative distances between adapted samples to achieve considerable generation diversity. In addition, we further enhance the learning of high-frequency details for better generation quality. Our approach is compatible with both unconditional and conditional diffusion models. This work makes the first attempt to realize unconditional few-shot image generation with diffusion models, achieving better quality and greater diversity than current state-of-the-art GAN-based approaches. Moreover, this work also significantly relieves overfitting for conditional generation and realizes high-quality domain-driven generation, further expanding the applicable scenarios of modern large-scale text-to-image models. | ['Jian Yuan', 'Jiansheng Chen', 'Huimin Ma', 'Jingyuan Zhu'] | 2023-06-25 | null | null | null | null | ['image-generation'] | ['computer-vision'] | [ 2.19837442e-01 9.86654758e-02 -3.31031792e-02 -2.40180343e-01
-1.08515775e+00 -2.00046510e-01 9.66217160e-01 -6.62297964e-01
1.28483642e-02 1.01841617e+00 3.57765257e-01 3.78485292e-01
-4.58528101e-02 -1.28833401e+00 -6.45792961e-01 -1.01171422e+00
4.26424563e-01 6.35011733e-01 3.98465902e-01 -2.55224437e-01
-8.70775431e-03 2.04294324e-01 -1.55387616e+00 2.68601716e-01
1.21354055e+00 7.58807898e-01 6.14667118e-01 7.80375123e-01
-1.51985973e-01 6.34543180e-01 -9.48171616e-01 -5.43666184e-01
2.60233313e-01 -1.13725567e+00 -2.51831561e-01 4.19807613e-01
3.37988555e-01 -5.11772752e-01 -3.21663797e-01 8.27914059e-01
1.01384652e+00 4.07803021e-02 9.87998903e-01 -1.08663464e+00
-1.44999015e+00 4.28868115e-01 -5.56196868e-01 -9.05587077e-02
1.77037239e-01 5.26786149e-01 4.70074773e-01 -7.50282109e-01
1.03642976e+00 1.16289890e+00 4.57084805e-01 1.06515992e+00
-1.27087522e+00 -6.82410717e-01 -1.32381186e-01 7.29783028e-02
-1.22279799e+00 -3.96620572e-01 1.08137751e+00 -3.86366755e-01
6.55256391e-01 -6.39442876e-02 7.30952501e-01 1.82395613e+00
1.68621868e-01 6.89114630e-01 1.21649945e+00 -4.86367464e-01
5.23840904e-01 2.88979888e-01 -7.20455468e-01 3.31585467e-01
5.39469570e-02 2.52627015e-01 -8.09269607e-01 -1.60379522e-02
1.18152654e+00 -1.83569595e-01 -2.14650601e-01 -2.52329260e-01
-1.21324170e+00 1.08942270e+00 2.54563451e-01 4.99044925e-01
-6.32427633e-01 4.27373052e-02 -4.06660028e-02 1.57302096e-01
8.06140840e-01 2.40419656e-01 -8.70239139e-02 -1.55889288e-01
-1.48645914e+00 4.10288870e-01 6.72889769e-01 1.28592277e+00
7.84857571e-01 6.45385742e-01 -6.35564566e-01 1.13515389e+00
1.39567971e-01 9.02955949e-01 6.86180949e-01 -9.92093265e-01
2.11447656e-01 1.81619108e-01 -4.07011844e-02 -6.96268201e-01
3.26532751e-01 -5.74541450e-01 -1.09963989e+00 2.81856000e-01
1.27909988e-01 -3.04612547e-01 -1.39364004e+00 1.78381848e+00
3.29627365e-01 -2.50065308e-02 1.25906736e-01 6.75661147e-01
7.79635370e-01 9.33751523e-01 6.58211485e-02 -9.25829411e-02
9.42605555e-01 -9.41932857e-01 -7.40060389e-01 -1.78850189e-01
1.32104978e-01 -7.95383990e-01 1.17460096e+00 4.02428567e-01
-1.16385329e+00 -7.68557489e-01 -8.70475411e-01 1.20220549e-01
-1.73260778e-01 -2.87574483e-03 4.03177917e-01 9.03528631e-01
-1.26868236e+00 4.87819344e-01 -4.33156937e-01 -3.68232042e-01
8.01798820e-01 -9.01967511e-02 -1.29976854e-01 -5.53672135e-01
-1.23023522e+00 7.54978418e-01 2.33985946e-01 -4.00119126e-01
-1.46224689e+00 -8.32259834e-01 -7.76149929e-01 -2.88003415e-01
1.83601435e-02 -1.16445649e+00 9.25278306e-01 -9.46120381e-01
-1.86783504e+00 6.04312420e-01 -9.10887793e-02 -3.51254225e-01
7.00815499e-01 7.39089027e-02 -5.10970354e-01 2.41346240e-01
2.85316288e-01 1.18605888e+00 1.37206101e+00 -1.49653006e+00
-2.84748405e-01 -8.97269472e-02 -3.35568249e-01 9.43478197e-02
-7.44705975e-01 -3.55909020e-01 -4.70360488e-01 -1.11164606e+00
-3.32171500e-01 -6.43371463e-01 -3.78319591e-01 5.45659885e-02
-2.40773022e-01 3.93767580e-02 8.47097337e-01 -6.83233440e-01
1.03575230e+00 -1.94056940e+00 2.65121788e-01 -1.66001752e-01
1.42237633e-01 2.61752784e-01 -4.16289926e-01 5.04909515e-01
3.76282245e-01 -1.50257517e-02 -5.89670181e-01 -5.39542198e-01
4.69537862e-02 3.60009640e-01 -2.23454326e-01 -2.43210830e-02
3.41637224e-01 1.08378983e+00 -7.40563214e-01 -6.48773015e-01
2.45527416e-01 7.56177187e-01 -6.66026831e-01 3.19583565e-01
-3.89406800e-01 6.79260433e-01 -3.80149901e-01 5.37259281e-01
8.22981358e-01 -3.98263149e-02 -2.57697135e-01 -3.56617346e-02
3.52914423e-01 -3.70010227e-01 -9.78818715e-01 1.95895493e+00
-5.15859127e-01 4.87931967e-01 -1.77483365e-01 -6.90468609e-01
1.27294981e+00 2.61866212e-01 4.05914932e-01 -8.80384088e-01
-1.89578041e-01 3.12994033e-01 -3.93463194e-01 -2.62071371e-01
6.05839133e-01 -6.87750280e-01 -7.49224350e-02 2.49853879e-01
4.88693714e-01 -7.88193583e-01 4.09247905e-01 4.41331983e-01
7.79979169e-01 4.42931443e-01 -1.01321675e-01 -5.71244806e-02
5.91809265e-02 -8.23074356e-02 5.79989791e-01 7.73199201e-01
1.23991236e-01 1.29171395e+00 1.44685373e-01 4.38476261e-03
-1.50435877e+00 -1.32964742e+00 3.89267541e-02 5.26073813e-01
-6.23768312e-04 -2.13753700e-01 -9.28996801e-01 -7.09974766e-01
-4.31849331e-01 9.80039656e-01 -6.82124257e-01 -2.76727468e-01
-2.54803151e-01 -9.38346088e-01 5.22096038e-01 4.54896986e-01
8.66306961e-01 -1.09009492e+00 -1.16547853e-01 4.11169797e-01
-2.84526229e-01 -9.20688570e-01 -4.82678264e-01 -1.70980960e-01
-9.47763860e-01 -6.06967509e-01 -1.81148958e+00 -7.95354128e-01
7.13371813e-01 -3.47703621e-02 1.30987155e+00 -5.45254767e-01
-2.03827322e-01 3.31876516e-01 -4.96224344e-01 -3.92032593e-01
-7.70475209e-01 -6.32978529e-02 -2.26929769e-01 2.07390729e-02
1.27286747e-01 -8.84311616e-01 -8.01665425e-01 3.16306800e-01
-1.26789045e+00 5.45293391e-02 8.34628761e-01 9.01235998e-01
7.82047033e-01 2.70349592e-01 8.56720269e-01 -7.82960176e-01
8.13224018e-01 -5.67350030e-01 -1.79247931e-01 2.11518049e-01
-6.43634319e-01 -1.96367763e-02 6.66459918e-01 -8.47883880e-01
-1.68334818e+00 -2.18909413e-01 -3.79116237e-01 -5.36821306e-01
-2.85482287e-01 1.14090741e-01 -2.95131981e-01 2.05300488e-02
1.14084315e+00 7.72673070e-01 -1.09419815e-01 -3.35643947e-01
6.90993249e-01 5.13075113e-01 2.63719946e-01 -7.61036575e-01
8.85045052e-01 5.56876242e-01 -2.08490640e-01 -8.50662231e-01
-4.19683993e-01 1.67642701e-02 -4.43974823e-01 -2.97980011e-01
8.98151755e-01 -1.03755343e+00 3.61856848e-01 9.00005102e-01
-9.19903815e-01 -7.04020858e-01 -9.21336889e-01 2.85291970e-01
-7.80255795e-01 3.83891344e-01 -6.12806618e-01 -6.32184744e-01
-3.58494371e-01 -9.75575447e-01 1.23008013e+00 2.52784461e-01
-8.31622407e-02 -1.08621061e+00 3.18296671e-01 1.86611041e-01
9.16931868e-01 3.34271401e-01 5.39052963e-01 -8.87829512e-02
-5.83501399e-01 2.03669388e-02 4.47931550e-02 6.63509846e-01
1.63024902e-01 -1.48876160e-01 -7.34559238e-01 -1.83825642e-01
2.16206372e-01 -3.86400342e-01 9.91365373e-01 6.78462446e-01
7.74362803e-01 -2.31133237e-01 -1.91839471e-01 5.38876474e-01
1.52537751e+00 1.31289110e-01 1.05493414e+00 1.37693569e-01
5.81415117e-01 5.42106330e-01 4.03535664e-01 5.14424145e-01
2.28758216e-01 5.48481762e-01 2.38667708e-02 -3.29575449e-01
-9.24514592e-01 -6.09610081e-01 4.05100048e-01 7.49024212e-01
-7.39140110e-03 -5.53478777e-01 -4.57736164e-01 9.43992794e-01
-1.49870300e+00 -1.04517865e+00 7.72187635e-02 1.89529359e+00
1.09759068e+00 -8.26146174e-03 1.21858455e-01 -1.90779656e-01
7.00233936e-01 2.02238798e-01 -5.99530280e-01 -7.67922997e-02
-5.33654332e-01 3.47206712e-01 1.53908595e-01 1.80391029e-01
-6.59344912e-01 9.19172108e-01 6.46474218e+00 1.60114491e+00
-8.70204508e-01 5.24493992e-01 7.74719954e-01 -2.89183855e-01
-7.80570090e-01 -1.34965569e-01 -9.10292029e-01 6.89341545e-01
8.82205188e-01 -1.64624095e-01 1.81136727e-01 9.90095377e-01
-1.78850105e-03 -1.13658801e-01 -6.00278676e-01 8.37076724e-01
3.25075746e-01 -1.45686734e+00 4.55915183e-01 2.29693606e-01
1.55921948e+00 -2.46977463e-01 3.81643295e-01 2.36760929e-01
5.96903980e-01 -8.10354650e-01 7.87125349e-01 7.46408761e-01
1.09943366e+00 -6.32056594e-01 4.07651752e-01 4.89811063e-01
-7.89670885e-01 1.98161498e-01 -5.96855164e-01 4.18751478e-01
6.28802240e-01 1.18436110e+00 -4.85303968e-01 6.67054594e-01
6.29735887e-01 6.70012176e-01 -4.08445984e-01 8.10687125e-01
-4.10494477e-01 6.66509271e-01 -2.31814995e-01 2.06411093e-01
8.79639536e-02 -2.50497431e-01 5.60976565e-01 1.09634542e+00
1.00687885e+00 -1.61038056e-01 -1.78425685e-01 1.27378893e+00
4.24675010e-02 -1.33619934e-01 -8.44290912e-01 4.39865962e-02
2.35748529e-01 9.43689227e-01 -6.10758483e-01 -5.70904315e-01
-1.46449342e-01 1.37297893e+00 -2.72594243e-02 4.80483919e-01
-8.47544968e-01 -2.24417627e-01 2.85504133e-01 3.40314478e-01
6.32496417e-01 -2.49598369e-01 -2.30224371e-01 -1.17938840e+00
-1.07974581e-01 -9.97363865e-01 1.59568548e-01 -9.53167439e-01
-1.73019373e+00 9.20145869e-01 1.22250795e-01 -1.29465640e+00
-4.89341021e-01 -4.22972664e-02 -6.36331737e-01 1.05705392e+00
-1.45825469e+00 -1.74357319e+00 -3.25066715e-01 8.69634390e-01
9.28734601e-01 -4.31043655e-01 7.69980669e-01 2.17254713e-01
-9.76781547e-02 4.97928113e-01 3.53833914e-01 -3.50674778e-01
8.87868762e-01 -1.05239284e+00 3.95236850e-01 1.07056105e+00
1.59417629e-01 3.84208649e-01 6.42941833e-01 -1.01350451e+00
-9.04979467e-01 -1.13120615e+00 5.52025378e-01 -4.76799518e-01
6.26232848e-02 -3.37127984e-01 -6.93331122e-01 2.74343312e-01
5.26410699e-01 -2.52945662e-01 5.49263358e-01 -3.10933501e-01
-1.05355009e-01 -1.78532064e-01 -1.39495063e+00 6.45687163e-01
1.15763283e+00 -3.48582149e-01 -3.08531404e-01 2.92509437e-01
6.41437531e-01 -1.88201249e-01 -9.90791798e-01 2.50907660e-01
1.67718291e-01 -1.17805612e+00 1.03581238e+00 1.02823585e-01
8.03718626e-01 -1.14665657e-01 -8.40132609e-02 -1.68570399e+00
-4.57336634e-01 -7.12014675e-01 -3.01901639e-01 1.64962816e+00
2.84447253e-01 -4.16653126e-01 8.92253220e-01 4.51192319e-01
-1.49359507e-02 -5.48489034e-01 -7.33446002e-01 -9.71708238e-01
3.88927341e-01 -2.77582347e-01 6.89516425e-01 6.29872203e-01
-5.77332497e-01 1.76775530e-01 -8.36094141e-01 -3.19635719e-01
9.66167867e-01 9.00254957e-03 8.95161510e-01 -7.89798260e-01
-4.30958867e-01 -3.93805742e-01 -1.40487894e-01 -1.15611231e+00
-2.25511640e-01 -5.20077646e-01 9.21220705e-02 -1.90902233e+00
2.54853815e-01 -4.84295487e-01 1.71086088e-01 1.01441126e-02
-1.60997376e-01 5.89751303e-01 7.74799660e-02 2.87285537e-01
-3.00401419e-01 1.10093677e+00 1.79391873e+00 -1.88728377e-01
-1.38427496e-01 -1.45430759e-01 -8.17890167e-01 2.64463931e-01
7.97771454e-01 -6.09179556e-01 -8.17866445e-01 -3.94783497e-01
-1.14123523e-01 -1.96657225e-01 2.76315451e-01 -1.28703487e+00
1.55076683e-01 -2.00973287e-01 6.94926143e-01 -3.59861284e-01
4.73661095e-01 -4.37168032e-01 5.97374141e-01 2.06721187e-01
-1.19459257e-01 -5.54978848e-01 9.65621602e-03 8.21725905e-01
-3.45864832e-01 -1.67328700e-01 9.69884157e-01 -4.50918227e-01
-8.41496229e-01 4.90495235e-01 -2.42887124e-01 9.44417790e-02
1.25318336e+00 -5.49952984e-01 -1.50849611e-01 -7.59897351e-01
-5.76916218e-01 -2.86476403e-01 7.36233473e-01 4.24658120e-01
8.48711133e-01 -1.48312366e+00 -1.03945482e+00 3.28002781e-01
-5.58679923e-02 2.30698790e-02 7.36200333e-01 3.93738270e-01
-2.76153266e-01 5.13896830e-02 -4.39266920e-01 -4.86238390e-01
-7.70393789e-01 5.58915973e-01 2.80609559e-02 -4.44018453e-01
-7.07055211e-01 1.02186489e+00 4.08575267e-01 -3.42957646e-01
-3.14886332e-01 2.83587486e-01 1.22623019e-01 -5.00708260e-02
4.36123282e-01 1.98900372e-01 -2.16937914e-01 -4.16065305e-01
1.34650722e-01 7.02520072e-01 6.98495656e-02 -5.13021231e-01
1.49336529e+00 -1.22314170e-01 2.66703665e-01 -7.87438452e-02
8.94766986e-01 4.25811149e-02 -1.83561409e+00 -1.86909273e-01
-6.66883826e-01 -6.50995672e-01 2.02819603e-04 -8.79651725e-01
-1.33112502e+00 9.81188595e-01 5.37720680e-01 -3.35942097e-02
1.46006513e+00 5.97451143e-02 1.07134259e+00 -3.43477368e-01
5.97682595e-01 -1.20216799e+00 6.88275456e-01 1.63229518e-02
1.02320731e+00 -1.07765210e+00 -1.51622966e-01 -2.31804073e-01
-1.05081058e+00 8.13715577e-01 6.63562059e-01 -6.55532107e-02
4.11315858e-01 3.17432702e-01 -7.30600134e-02 -4.05533910e-02
-5.20710409e-01 -2.29982048e-01 2.50774652e-01 1.47913277e+00
5.61535954e-02 -1.81669787e-01 -2.36685097e-01 4.37449455e-01
-1.00452229e-01 3.45641255e-01 3.60832483e-01 7.69754708e-01
-2.89497584e-01 -1.43475485e+00 -3.97506297e-01 3.47253710e-01
-2.50856996e-01 -2.36061513e-01 -1.37260363e-01 5.24382412e-01
3.21033806e-01 9.26614344e-01 -2.28273720e-01 -1.91643104e-01
1.81962952e-01 -7.70363957e-02 5.93640149e-01 -4.88953054e-01
-2.30476096e-01 2.11791471e-01 -8.35163370e-02 -2.65981466e-01
-5.82652152e-01 -3.82057339e-01 -5.79088390e-01 -3.47955912e-01
-2.08502665e-01 1.48330517e-02 5.09175599e-01 5.67956448e-01
7.20601618e-01 6.72536910e-01 4.64077383e-01 -1.06768036e+00
-3.89822870e-01 -1.08841038e+00 -8.17754805e-01 5.23389935e-01
-1.22333243e-01 -5.29107511e-01 -1.67833924e-01 4.01923835e-01] | [11.491389274597168, -0.4149489998817444] |
8419c2e9-9392-455d-94b7-e984f9203764 | mean-oriented-riesz-features-for-micro | 2005.06198 | null | https://arxiv.org/abs/2005.06198v1 | https://arxiv.org/pdf/2005.06198v1.pdf | Mean Oriented Riesz Features for Micro Expression Classification | Micro-expressions are brief and subtle facial expressions that go on and off the face in a fraction of a second. This kind of facial expressions usually occurs in high stake situations and is considered to reflect a human's real intent. There has been some interest in micro-expression analysis, however, a great majority of the methods are based on classically established computer vision methods such as local binary patterns, histogram of gradients and optical flow. A novel methodology for micro-expression recognition using the Riesz pyramid, a multi-scale steerable Hilbert transform is presented. In fact, an image sequence is transformed with this tool, then the image phase variations are extracted and filtered as proxies for motion. Furthermore, the dominant orientation constancy from the Riesz transform is exploited to average the micro-expression sequence into an image pair. Based on that, the Mean Oriented Riesz Feature description is introduced. Finally the performance of our methods are tested in two spontaneous micro-expressions databases and compared to state-of-the-art methods. | ['Anne-Claire Legrand', 'Rémi Emonet', 'Olivier Alata', 'Hubert Konik', 'Carlos Arango Duque'] | 2020-05-13 | null | null | null | null | ['micro-expression-recognition'] | ['computer-vision'] | [ 1.64248720e-01 -1.83817253e-01 -1.57496959e-01 -4.76594985e-01
-3.43824565e-01 -2.26414829e-01 7.22578347e-01 -3.25405717e-01
-3.58374953e-01 6.97418630e-01 1.54835150e-01 5.54978609e-01
-7.22901197e-03 -5.28188944e-01 -1.96533948e-01 -1.17577541e+00
-1.18681893e-01 -2.43103355e-01 -1.06442250e-01 -5.00151515e-01
2.35578150e-01 8.73667955e-01 -1.74838841e+00 1.29419237e-01
3.40440303e-01 1.16052127e+00 -2.78308123e-01 3.95061225e-01
-1.13086745e-01 1.23537242e+00 -5.16540587e-01 -5.85926950e-01
2.57165208e-02 -6.36252761e-01 -6.35128736e-01 2.63395518e-01
2.95295537e-01 -1.07434511e-01 -9.58722010e-02 1.44555306e+00
4.27876800e-01 1.65592250e-03 6.42233789e-01 -9.93903577e-01
-3.25627267e-01 -2.63083309e-01 -8.64459276e-01 2.43844971e-01
8.43937814e-01 -8.55718777e-02 7.32847452e-01 -9.93815541e-01
1.07308698e+00 1.30152571e+00 5.98393500e-01 2.03757286e-01
-1.13107610e+00 -5.96722305e-01 -3.54292572e-01 5.23485243e-01
-1.45356417e+00 -5.47052920e-01 1.36285198e+00 -6.18252397e-01
5.13850808e-01 3.00839245e-01 1.00556397e+00 8.84319007e-01
3.69623184e-01 5.67074776e-01 1.51309288e+00 -5.61310172e-01
4.02271487e-02 2.31273562e-01 -3.36151980e-02 9.25258517e-01
-3.85068297e-01 -1.13019556e-01 -6.68522477e-01 -1.60572499e-01
6.56289756e-01 -1.07299678e-01 -4.73607361e-01 -3.42005789e-01
-8.68207991e-01 6.59671247e-01 2.08248615e-01 8.51830423e-01
-8.30036461e-01 -1.62380964e-01 3.86821598e-01 2.44179010e-01
7.02672243e-01 -9.69837829e-02 1.26623854e-01 -5.90040326e-01
-9.59950984e-01 2.45265558e-01 7.50671744e-01 1.97708935e-01
1.26083016e+00 1.06097206e-01 -3.00578885e-02 6.83268189e-01
1.47047460e-01 5.98155141e-01 6.05962217e-01 -9.70812738e-01
-1.68464109e-01 5.69049776e-01 5.12787793e-03 -1.76509356e+00
-3.22243363e-01 1.98395681e-02 -8.54053617e-01 4.48409677e-01
3.97785485e-01 1.02120981e-01 -2.36842871e-01 1.77347696e+00
3.92939419e-01 3.01051766e-01 -2.76892819e-02 1.15023291e+00
4.87808555e-01 6.28904641e-01 -1.70448288e-01 -7.12069809e-01
1.34281516e+00 -3.79587203e-01 -1.25049591e+00 4.40922856e-01
3.13649058e-01 -9.51027453e-01 7.65236020e-01 3.43260914e-01
-8.97930801e-01 -4.20868814e-01 -8.96079898e-01 1.51385844e-01
-2.19487116e-01 1.14631824e-01 5.81211805e-01 8.13926756e-01
-8.85647416e-01 6.73590004e-01 -5.64847648e-01 -4.60570276e-01
6.47670701e-02 3.97534780e-02 -7.51578867e-01 2.57058442e-01
-1.04645479e+00 9.46853042e-01 -2.77584285e-01 3.82002622e-01
-3.15935582e-01 -3.76485795e-01 -9.16969597e-01 -2.68539488e-01
-2.01801747e-01 -6.88642785e-02 9.60895658e-01 -1.64590394e+00
-1.94253397e+00 1.44987988e+00 -6.33625090e-01 -7.37972111e-02
4.81062293e-01 1.11328274e-01 -5.04183412e-01 5.60604095e-01
5.38996095e-03 1.16056830e-01 1.25126207e+00 -7.99324036e-01
-1.98247302e-02 -7.41147578e-01 -1.74059495e-01 -9.68321785e-03
-3.13448235e-02 3.64346564e-01 2.79293545e-02 -3.92657429e-01
2.21675605e-01 -8.09842169e-01 1.24762550e-01 -6.20441362e-02
-7.69532844e-02 -1.51640251e-01 1.24141324e+00 -7.78522193e-01
1.01327717e+00 -2.35877395e+00 1.82999715e-01 2.50442773e-01
3.24578546e-02 -1.89391465e-03 1.61318287e-01 2.55208373e-01
-2.76484877e-01 -2.33632863e-01 -1.49315208e-01 -6.67540580e-02
-1.87262133e-01 1.19882673e-01 -1.06143907e-01 1.14122319e+00
-1.33070732e-02 6.31150484e-01 -8.76872420e-01 -4.97512788e-01
2.50297904e-01 6.06482148e-01 -1.96807235e-01 2.28392228e-01
4.60350841e-01 7.71710932e-01 -3.25192809e-01 7.07641184e-01
8.70546937e-01 1.02464549e-01 2.37535127e-02 -4.77718651e-01
-2.56515145e-01 -2.22875744e-01 -7.64236450e-01 1.55847549e+00
-3.29054087e-01 9.78640497e-01 3.67450505e-01 -1.35538244e+00
1.19084215e+00 4.80409175e-01 8.40315402e-01 -7.88769782e-01
3.17528427e-01 3.26024592e-01 -3.44266742e-01 -8.55226159e-01
1.53357893e-01 -4.84348118e-01 2.15698779e-01 1.05042018e-01
9.22492612e-03 -1.46219745e-01 2.78552413e-01 -3.17920566e-01
7.55918920e-01 2.55284935e-01 7.07819879e-01 -3.25790226e-01
1.02743030e+00 -4.93741781e-01 5.16001701e-01 -3.05586420e-02
-4.19147044e-01 3.94558549e-01 7.36404717e-01 -6.79385185e-01
-7.44704247e-01 -8.09262633e-01 -9.24369022e-02 7.02677667e-01
4.26577963e-03 -2.20825896e-01 -1.04489505e+00 -4.23415065e-01
-2.35984549e-01 2.07016364e-01 -7.40738511e-01 7.90264979e-02
-4.10821587e-01 -5.91825902e-01 5.32142758e-01 -1.65538654e-01
9.41925764e-01 -1.19256234e+00 -1.05692542e+00 9.94990394e-02
-3.18670481e-01 -1.20970964e+00 -1.92879751e-01 -4.55585897e-01
-5.46678483e-01 -1.05693960e+00 -1.05753386e+00 -4.80414033e-01
4.38048959e-01 7.03819916e-02 8.59386444e-01 -3.66289496e-01
-6.66218817e-01 4.09156263e-01 -3.42628270e-01 -1.20649286e-01
-2.53770679e-01 -6.91202462e-01 1.07909389e-01 9.96061921e-01
6.90373778e-01 -9.88836288e-01 -3.21344972e-01 1.39115602e-01
-6.06089175e-01 -2.33200490e-01 4.52406406e-01 6.99854791e-01
4.19192672e-01 -2.19141990e-01 8.25121813e-03 -5.28665125e-01
5.66623390e-01 -6.43038079e-02 -4.18268681e-01 -5.51370755e-02
-1.97242111e-01 -7.44587258e-02 2.74182111e-01 -3.66632670e-01
-1.28991342e+00 1.23242373e-02 -1.59557641e-01 -6.01417542e-01
-1.30967811e-01 3.92580390e-01 -4.18832898e-02 -4.17363316e-01
7.25394189e-01 4.88477260e-01 3.93850386e-01 -1.23474672e-01
1.87474802e-01 6.44208610e-01 4.41056877e-01 -3.14040691e-01
5.36568105e-01 9.84840453e-01 3.04419070e-01 -1.69451082e+00
-5.98760545e-01 -6.35585248e-01 -6.44015372e-01 -5.41179001e-01
1.08881319e+00 -5.89599311e-01 -9.48297620e-01 7.88404524e-01
-1.27998340e+00 8.64348263e-02 -1.83723688e-01 4.66253579e-01
-7.82809913e-01 6.41953230e-01 -6.07129633e-01 -1.23132575e+00
-1.19057849e-01 -1.18573248e+00 1.13125908e+00 2.63690978e-01
-3.69232178e-01 -9.28269327e-01 2.65097648e-01 1.27884537e-01
4.43311721e-01 6.42458916e-01 4.43312138e-01 1.52394414e-01
-1.51560217e-01 -2.18511730e-01 -7.97711909e-02 4.61348653e-01
2.16038629e-01 5.61840124e-02 -1.13195086e+00 1.41561761e-01
4.53897983e-01 -2.98315585e-01 2.90856361e-01 4.23364669e-01
8.58302653e-01 -4.17562485e-01 4.30701897e-02 7.14921832e-01
1.17935538e+00 3.74913275e-01 8.75941336e-01 1.29557893e-01
3.14442515e-01 1.00483429e+00 6.23313844e-01 7.49016345e-01
-1.17381901e-01 1.05140162e+00 1.58787519e-01 -4.35658693e-02
2.63972104e-01 -3.63092707e-03 5.82902253e-01 6.66910470e-01
-7.31648684e-01 6.92228556e-01 -3.94574910e-01 2.00302303e-01
-1.70900667e+00 -1.43488777e+00 -1.81133932e-04 1.99622369e+00
5.06739438e-01 -4.04650807e-01 -3.77085507e-02 3.56115401e-01
6.13588750e-01 5.83558857e-01 -6.14820458e-02 -5.04052520e-01
-5.54662764e-01 3.01384866e-01 2.07584709e-01 3.84973586e-01
-9.87702906e-01 7.29458153e-01 6.07211304e+00 8.78024936e-01
-1.89991570e+00 2.78810799e-01 5.68148911e-01 4.42510605e-01
5.24850637e-02 -1.24539897e-01 -2.78133303e-01 3.32090378e-01
6.21755242e-01 -2.50252664e-01 2.01404884e-01 7.50726700e-01
4.52835083e-01 -2.95633227e-01 -5.52501023e-01 1.75364923e+00
2.73996115e-01 -8.87376487e-01 -2.26959839e-01 1.86655104e-01
4.19044286e-01 -5.68977714e-01 8.86444673e-02 -7.98441470e-02
-4.77931172e-01 -9.81098294e-01 5.49153745e-01 6.80409372e-01
7.17292190e-01 -7.74813592e-01 7.36042559e-01 1.11010872e-01
-1.13329577e+00 1.75522566e-01 -2.21649319e-01 -4.24322486e-01
2.30335250e-01 7.70212293e-01 -4.83912587e-01 4.22136426e-01
6.11049891e-01 1.00729537e+00 -2.16345072e-01 3.96242946e-01
-2.46457025e-01 5.19946814e-01 -2.84649193e-01 -3.12443376e-01
2.64956295e-01 -9.68631148e-01 6.99223459e-01 1.37981021e+00
4.78277802e-01 1.22590952e-01 -3.44020784e-01 8.02863061e-01
3.81498128e-01 5.01249492e-01 -9.17707562e-01 -1.45398751e-01
-2.53930748e-01 1.55004501e+00 -3.10017854e-01 -2.59990692e-01
-3.80389959e-01 1.34676123e+00 -1.80624008e-01 5.21608949e-01
-6.21210992e-01 -3.20456147e-01 7.93182492e-01 2.16691241e-01
-1.99688405e-01 -1.76788848e-02 1.87250286e-01 -1.34551489e+00
1.43098027e-01 -7.53859878e-01 -2.27874108e-02 -9.88283455e-01
-8.51265848e-01 5.18107712e-01 -1.12173885e-01 -1.20254338e+00
-6.72046900e-01 -6.56298399e-01 -5.11837423e-01 8.47453475e-01
-1.27454376e+00 -9.57485139e-01 -7.27380335e-01 8.77374113e-01
1.98436707e-01 -1.80466399e-01 1.00728846e+00 3.10455292e-01
-1.63992628e-01 2.82503277e-01 -1.69247478e-01 2.16833323e-01
4.61846352e-01 -8.71100962e-01 -4.53386515e-01 5.16027749e-01
4.26052511e-03 4.80157018e-01 9.20969546e-01 -5.87138496e-02
-1.35119987e+00 -2.99331129e-01 9.05992568e-01 1.66202515e-01
7.15002954e-01 -8.81098211e-02 -6.75567508e-01 4.63827342e-01
2.23253593e-01 4.40699369e-01 5.70283651e-01 -2.84064233e-01
-3.11989576e-01 -5.01362145e-01 -1.32046676e+00 3.68002892e-01
7.17799306e-01 -7.41920710e-01 -3.86140972e-01 2.70578302e-02
-2.48778671e-01 -1.59611374e-01 -7.64846563e-01 3.37816596e-01
8.88819039e-01 -1.72223866e+00 7.30884552e-01 -1.75187498e-01
2.45997936e-01 -7.06246048e-02 -2.91912854e-01 -1.19842732e+00
-1.26078218e-01 -1.02401137e+00 2.58699000e-01 8.37840021e-01
-2.82943547e-01 -7.01825738e-01 7.70894706e-01 1.10375080e-02
4.78487313e-01 -5.66556454e-01 -1.27655911e+00 -5.04337549e-01
-5.48365474e-01 -1.44815668e-01 1.98235258e-01 1.16258359e+00
3.40922982e-01 3.24285209e-01 -4.71507818e-01 -3.34893018e-01
8.23744714e-01 7.83963203e-02 9.48201418e-01 -1.05566359e+00
-6.20960668e-02 -5.96081614e-01 -1.26151597e+00 -7.35707641e-01
4.71480548e-01 -5.77851832e-01 -2.78490186e-01 -7.49597967e-01
4.32305336e-02 1.80095181e-01 -5.85833404e-05 2.47600414e-02
2.77373195e-01 3.85789335e-01 9.13402662e-02 1.79508060e-01
-9.14630517e-02 6.15900159e-01 1.28402352e+00 -1.53729364e-01
-7.60920197e-02 -9.40466449e-02 -1.40044838e-01 1.03140616e+00
4.13188279e-01 6.22979272e-03 -1.64215192e-01 5.18143654e-01
5.65992631e-02 2.72233039e-01 3.83425355e-01 -9.65860903e-01
-6.08646497e-02 -1.17068037e-01 1.62626281e-01 -2.24884033e-01
8.35788012e-01 -7.08797455e-01 2.91591108e-01 2.88136035e-01
4.54734974e-02 -1.19085358e-02 -3.32135677e-01 2.20498413e-01
-9.26738083e-01 -1.55094698e-01 1.19321191e+00 -1.73573643e-01
-8.86762738e-01 1.78420916e-01 -5.05314767e-01 -3.27605605e-01
1.08229816e+00 -3.78041625e-01 2.93586552e-01 -7.60195136e-01
-6.65287077e-01 -7.78483629e-01 2.98970282e-01 -1.75098795e-02
6.37559175e-01 -1.33632421e+00 -6.03621721e-01 2.80226648e-01
5.47074080e-02 -6.06345236e-01 3.70190322e-01 1.45749092e+00
-7.87675321e-01 1.77744463e-01 -5.78982711e-01 -7.50408173e-01
-1.68941510e+00 2.35380620e-01 6.03465796e-01 -1.67459756e-01
-5.88554680e-01 4.98975903e-01 3.44026417e-01 -4.93504480e-02
-2.80073047e-01 7.90685043e-02 -5.33154488e-01 4.43033367e-01
8.19908202e-01 4.15517539e-01 -1.55247614e-01 -1.49085677e+00
-4.58133489e-01 1.19956732e+00 4.68940705e-01 -3.51596713e-01
1.14464140e+00 -1.34145886e-01 -7.57148325e-01 6.91597223e-01
1.85163677e+00 3.70352596e-01 -9.39586043e-01 -8.17045476e-03
-1.38351619e-02 -7.35868037e-01 -1.97022259e-01 -1.06272502e-02
-9.89518583e-01 1.11463106e+00 8.21790993e-01 3.43857974e-01
1.46846628e+00 -2.75357395e-01 3.20020080e-01 3.08412611e-01
4.91327703e-01 -9.07596648e-01 2.47456074e-01 2.56420732e-01
1.09507954e+00 -9.26493883e-01 -1.63548470e-01 -5.39343059e-01
-5.98748088e-01 1.36949074e+00 -7.17661679e-02 -3.46829414e-01
7.47774720e-01 2.67181039e-01 4.16292399e-01 -2.04823732e-01
-1.20914698e-01 -1.71799824e-01 -4.50284965e-03 5.39627790e-01
6.76984787e-01 4.22643237e-02 -5.65117419e-01 1.21512875e-01
-2.51142651e-01 4.44595277e-01 4.53570873e-01 8.59233260e-01
-3.68353575e-01 -6.16680145e-01 -5.96842527e-01 -1.51768178e-01
-8.80928993e-01 4.73024309e-01 -1.42473087e-01 5.82232535e-01
5.65455034e-02 6.83746696e-01 -7.89770558e-02 -1.62714034e-01
4.27734375e-01 2.74035752e-01 8.23021650e-01 8.27029645e-02
-1.41886324e-01 1.26947150e-01 -1.09392568e-01 -9.26684737e-01
-1.11893106e+00 -6.17066741e-01 -9.75765824e-01 -2.10018665e-01
3.12464982e-01 2.70813078e-01 8.19454134e-01 7.21488893e-01
3.41872610e-02 -2.18263902e-02 7.37794280e-01 -1.19832742e+00
-3.60966891e-01 -1.02726424e+00 -1.15533209e+00 9.62937295e-01
5.16549230e-01 -8.80011797e-01 -6.59853399e-01 2.84994006e-01] | [13.627718925476074, 1.7915830612182617] |
5b068d61-828b-4817-807c-6278f939f766 | full-angle-quaternions-for-robustly-matching | null | null | http://openaccess.thecvf.com/content_cvpr_2014/html/Liwicki_Full-Angle_Quaternions_for_2014_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2014/papers/Liwicki_Full-Angle_Quaternions_for_2014_CVPR_paper.pdf | Full-Angle Quaternions for Robustly Matching Vectors of 3D Rotations | In this paper we introduce a new distance for robustly matching vectors of 3D rotations. A special representation of 3D rotations, which we coin full-angle quaternion (FAQ), allows us to express this distance as Euclidean. We apply the distance to the problems of 3D shape recognition from point clouds and 2D object tracking in color video. For the former, we introduce a hashing scheme for scale and translation which outperforms the previous state-of-the-art approach on a public dataset. For the latter, we incorporate online subspace learning with the proposed FAQ representation to highlight the benefits of the new representation. | ['Minh-Tri Pham', 'Maja Pantic', 'Stefanos Zafeiriou', 'Bjorn Stenger', 'Stephan Liwicki'] | 2014-06-01 | null | null | null | cvpr-2014-6 | ['3d-shape-recognition'] | ['computer-vision'] | [-3.72957230e-01 -5.23589551e-01 -3.26685667e-01 -5.83750159e-02
-6.93083763e-01 -9.13329005e-01 6.74684405e-01 -6.64143190e-02
-4.31439281e-01 7.95211196e-02 6.19929172e-02 -1.96597695e-01
6.49031810e-03 -3.74832332e-01 -7.37273455e-01 -6.75428808e-01
-2.44791508e-01 2.88542181e-01 3.92184734e-01 -2.01859623e-01
4.72591847e-01 1.19981551e+00 -1.17648149e+00 -3.58405352e-01
1.33055717e-01 9.71044302e-01 -4.17452574e-01 4.04306293e-01
1.03329107e-01 -1.53222308e-01 -2.39244521e-01 -4.90996420e-01
8.44192863e-01 1.43681720e-01 -4.96876240e-01 1.12686709e-01
6.96856737e-01 -4.47910964e-01 -6.74310684e-01 8.97907972e-01
6.10786378e-01 -1.16688102e-01 7.35480547e-01 -1.59000909e+00
-7.68546760e-01 -1.60853013e-01 -7.08608508e-01 -1.90608606e-01
7.81611264e-01 -2.39438385e-01 1.03763628e+00 -1.29870820e+00
9.12550569e-01 1.37836969e+00 7.30641961e-01 2.35351726e-01
-1.12012827e+00 -4.82781708e-01 -2.05537051e-01 2.10506260e-01
-1.85361803e+00 -1.39004067e-01 8.08610082e-01 -5.00990391e-01
8.84160757e-01 1.76434487e-01 9.01509523e-01 6.86134934e-01
1.64706886e-01 7.64153540e-01 8.58263969e-01 -3.92241836e-01
2.25724488e-01 -2.51292765e-01 -3.75065133e-02 6.61939085e-01
5.50267220e-01 2.59607822e-01 -4.36065733e-01 -4.25067067e-01
9.46973085e-01 3.47526670e-01 -1.39862388e-01 -1.36581254e+00
-1.51734650e+00 7.93673396e-01 5.23615062e-01 -1.52690962e-01
-8.90229046e-02 1.04198635e-01 1.59856796e-01 2.55838841e-01
2.13470772e-01 2.37123638e-01 -1.46372646e-01 -1.31875664e-01
-6.61414981e-01 3.38633835e-01 8.48075151e-01 1.21245694e+00
6.74941361e-01 1.79658718e-02 2.75088638e-01 4.91334736e-01
5.58225989e-01 9.89552736e-01 1.24971345e-01 -9.18067694e-01
1.95484430e-01 4.44514781e-01 2.29383767e-01 -1.01913714e+00
-2.96593666e-01 1.82186216e-02 -5.51339388e-01 1.21334724e-01
2.57462800e-01 3.49732786e-01 -5.08526325e-01 1.44865060e+00
7.97314167e-01 3.77879411e-01 9.91159976e-02 1.01251578e+00
5.37012219e-01 4.68613446e-01 -6.91739440e-01 3.22415903e-02
1.19780087e+00 -5.33327639e-01 -3.63271028e-01 3.75138462e-01
2.32728213e-01 -8.36785614e-01 3.62274468e-01 2.65219331e-01
-8.17582607e-01 -2.89824247e-01 -1.34468210e+00 -4.17937450e-02
-4.07768637e-01 -1.10793397e-01 6.40863776e-01 7.23430395e-01
-1.09045875e+00 6.08848512e-01 -9.29546475e-01 -7.04435170e-01
-6.66057467e-02 3.82411480e-01 -6.85443163e-01 -8.65727663e-04
-9.02568460e-01 9.13829505e-01 3.78575921e-02 -1.48965254e-01
-4.40947086e-01 -4.40188259e-01 -9.42338943e-01 -3.79710078e-01
2.46285554e-03 -5.70639074e-01 1.00505209e+00 9.43753496e-02
-1.57012916e+00 1.18562317e+00 -1.79909259e-01 -3.03999156e-01
3.31366956e-01 -3.37399364e-01 -1.41225994e-01 3.62143725e-01
-1.79454774e-01 5.94951928e-01 1.19754577e+00 -1.02869642e+00
-2.12632567e-01 -6.86241388e-01 -1.27835050e-02 1.77767724e-01
-1.58907250e-01 1.55094400e-01 -6.60827518e-01 -6.60980344e-01
7.81183124e-01 -1.49627900e+00 -5.52186258e-02 4.45357025e-01
-2.60645479e-01 -3.38021278e-01 9.36343789e-01 -2.77374983e-01
8.07663798e-01 -2.36582375e+00 4.17980731e-01 3.20296139e-01
7.79777616e-02 1.44511744e-01 -1.21336743e-01 6.70071363e-01
-2.06452325e-01 -1.91310719e-01 1.26164407e-04 -3.66519272e-01
3.12492251e-01 3.61406893e-01 -6.01492643e-01 1.27004659e+00
2.62575150e-01 6.72902107e-01 -8.13261330e-01 -1.86350852e-01
1.12497494e-01 6.94185376e-01 -4.57062721e-01 1.51671246e-01
5.45636714e-01 1.11461364e-01 -3.01801562e-01 9.15047765e-01
1.16211915e+00 4.74091209e-02 -1.32720664e-01 -5.01028895e-01
-2.54292011e-01 1.30118743e-01 -1.81976461e+00 1.87859809e+00
2.22001165e-01 1.96717903e-01 -1.42389879e-01 -6.85515046e-01
1.17993283e+00 1.86568081e-01 8.48874569e-01 -2.13779807e-01
-1.60277456e-01 3.38951796e-01 -5.31168461e-01 8.90606865e-02
6.63275421e-01 1.21618479e-01 -1.55744731e-01 5.05146503e-01
3.47686857e-02 -5.56006670e-01 -1.59336418e-01 1.97912067e-01
9.00589108e-01 3.83270085e-01 5.69075048e-01 -4.69802767e-02
5.47142029e-01 -3.58531922e-01 4.74754035e-01 3.49089712e-01
-5.50227284e-01 9.68091130e-01 2.00133115e-01 -4.07154858e-01
-1.23617244e+00 -1.41601729e+00 -2.46018350e-01 5.67073345e-01
3.09877396e-01 -5.27793765e-01 -1.81060791e-01 -6.02187634e-01
7.80038834e-01 -8.77129957e-02 -2.06281424e-01 -4.18990664e-02
-6.23218060e-01 -1.98737264e-01 4.26803410e-01 6.02770984e-01
1.71687156e-01 -1.96192384e-01 -5.57662487e-01 -3.80339622e-02
2.22982913e-01 -1.32761633e+00 -7.47457504e-01 3.33861522e-02
-1.08133817e+00 -1.11397684e+00 -9.25473452e-01 -6.63775802e-01
4.45210934e-01 8.72126222e-01 7.39679039e-01 -2.27380171e-01
-2.27205098e-01 1.01367331e+00 -5.54935098e-01 -2.48166502e-01
-2.06090994e-02 -2.71403641e-01 7.05975771e-01 -4.02171835e-02
4.38212335e-01 -5.14748096e-01 -6.55030429e-01 5.22664964e-01
-8.13892245e-01 -4.44414139e-01 4.49737370e-01 5.33353686e-01
7.37808585e-01 -5.75213194e-01 7.27495030e-02 -1.71621203e-01
1.86619848e-01 -1.16482124e-01 -8.99935603e-01 1.13426067e-01
-5.62183022e-01 3.16223562e-01 2.02765033e-01 -4.47632730e-01
-4.30047140e-02 4.96167243e-01 -3.34675349e-02 -9.02083993e-01
2.02117801e-01 -2.12443173e-02 6.00169823e-02 -7.30128169e-01
3.06096166e-01 3.11843693e-01 9.04137790e-02 -5.62771082e-01
7.81714797e-01 6.44311309e-01 4.88603532e-01 -5.26376963e-01
1.34811711e+00 8.33240271e-01 5.85943222e-01 -7.49645174e-01
-3.28025579e-01 -9.02375579e-01 -1.11278844e+00 1.27549097e-01
6.22785926e-01 -1.11929917e+00 -8.36820364e-01 4.93170500e-01
-1.35384774e+00 5.14241993e-01 -3.27051997e-01 7.34929681e-01
-8.99746537e-01 8.92879367e-01 -4.04196441e-01 -8.12231779e-01
-2.43558779e-01 -1.09039247e+00 1.45931435e+00 -1.60144135e-01
1.08846620e-01 -5.71678996e-01 5.58325529e-01 -3.06367695e-01
9.32994187e-02 3.00102770e-01 4.86149639e-01 -5.69349766e-01
-6.24730170e-01 -3.22955251e-01 -2.39090025e-01 3.04596424e-01
4.41607982e-02 1.34512946e-01 -6.25332236e-01 -7.05008388e-01
-3.79370525e-02 -1.36815608e-01 6.26435995e-01 -9.80819911e-02
5.36930084e-01 -1.15730576e-01 -1.49964735e-01 1.07202697e+00
1.45450604e+00 2.92008743e-04 4.58483338e-01 5.25753856e-01
7.13588953e-01 1.78526878e-01 7.13432848e-01 7.86115050e-01
5.79746723e-01 9.93520617e-01 3.98892701e-01 6.43464625e-02
1.29955381e-01 -2.14685917e-01 5.36327958e-01 1.23030269e+00
-1.50311559e-01 5.91165423e-01 -7.05216408e-01 4.18504238e-01
-1.63877654e+00 -7.06061184e-01 2.78687060e-01 2.40117025e+00
5.46546698e-01 -2.21082404e-01 3.12684089e-01 2.49324322e-01
6.57846153e-01 3.37951005e-01 -5.32083750e-01 -3.65034819e-01
-3.16762954e-01 2.56697744e-01 6.81287408e-01 2.34205171e-01
-1.34558034e+00 7.22127676e-01 7.00206900e+00 3.46741974e-01
-1.26214528e+00 -1.52197808e-01 -4.43145007e-01 3.97089899e-01
-1.97089091e-01 1.88960314e-01 -1.04944062e+00 3.04362439e-02
4.52261686e-01 -1.01976916e-01 4.23616797e-01 9.74749207e-01
-2.64659852e-01 3.39246064e-01 -1.35352087e+00 1.42990899e+00
4.48106289e-01 -9.29413319e-01 2.46018857e-01 6.84648454e-02
5.80882251e-01 3.39902230e-02 2.10095912e-01 9.89441667e-03
-1.72550365e-01 -5.29719532e-01 7.98966110e-01 2.68905103e-01
9.35378134e-01 -6.35705650e-01 2.45045960e-01 -8.62428620e-02
-1.49203253e+00 1.64474458e-01 -7.99188137e-01 3.38041902e-01
-6.16657436e-02 2.44307339e-01 -9.02112305e-01 8.28525007e-01
4.58234668e-01 1.04147506e+00 -6.13039553e-01 1.49323666e+00
-1.49427548e-01 1.03450753e-02 -6.02186084e-01 1.29440039e-01
5.82523048e-02 -4.38759416e-01 9.24462557e-01 1.17512476e+00
8.07270885e-01 9.38888192e-02 7.64794648e-02 4.64669645e-01
2.15538181e-02 9.06716138e-02 -9.94672954e-01 1.08550325e-01
6.71374261e-01 1.06655061e+00 -4.58859891e-01 -2.51050871e-02
-6.12973630e-01 1.06500947e+00 -4.98404652e-02 2.76750177e-01
-6.47729158e-01 -4.22368467e-01 9.73554850e-01 -1.35169506e-01
8.30891013e-01 -9.64879274e-01 2.49854773e-01 -1.51703250e+00
-2.09779665e-02 -7.65696049e-01 3.77403080e-01 -7.17189670e-01
-1.33197904e+00 1.20765671e-01 1.81875885e-01 -1.89359307e+00
-1.87391266e-01 -9.82189894e-01 -2.68944055e-01 5.02590179e-01
-1.69470429e+00 -1.07512093e+00 -5.56083210e-02 8.61644626e-01
-8.74747932e-02 -8.95181149e-02 9.29730594e-01 1.45074412e-01
-2.93324832e-02 5.63340485e-01 4.51876879e-01 1.44563079e-01
1.16912150e+00 -1.39204752e+00 7.62171149e-01 7.21675694e-01
5.64446568e-01 8.85182738e-01 5.37131429e-01 -3.99992764e-01
-2.57706213e+00 -7.85426617e-01 6.47949219e-01 -5.49437404e-01
7.51847863e-01 -3.30244392e-01 -7.13039756e-01 7.66290605e-01
-2.17370883e-01 3.76187503e-01 7.21937954e-01 -2.92634636e-01
-1.10170066e+00 -3.52176368e-01 -1.15086126e+00 4.15286869e-01
9.21870053e-01 -6.59376562e-01 -1.00244427e+00 2.36911848e-01
7.30147541e-01 -4.92393464e-01 -1.15425014e+00 4.06152159e-01
8.26828480e-01 -6.50469959e-01 1.53953326e+00 -4.64778334e-01
-5.05666733e-01 -8.06101501e-01 -7.05482244e-01 -1.09838855e+00
-5.10512769e-01 -7.96291232e-01 -4.19475377e-01 7.88264751e-01
-2.38039002e-01 -6.79373503e-01 6.06137633e-01 1.94579527e-01
1.37105629e-01 -4.75723684e-01 -1.38533354e+00 -1.15712726e+00
1.54313922e-01 -3.04108411e-01 8.00007939e-01 7.90806532e-01
-2.89795399e-02 -6.69356063e-02 -3.71551484e-01 3.05735469e-01
1.02086306e+00 4.81098235e-01 1.00125241e+00 -1.20101881e+00
-1.02107823e-01 -1.77310184e-01 -1.22897196e+00 -1.39494312e+00
9.54328552e-02 -9.85022902e-01 -4.24213320e-01 -9.63293970e-01
2.72773892e-01 -4.96331789e-03 -2.70532280e-01 8.56796503e-02
1.59164399e-01 4.23813283e-01 6.06311321e-01 5.52483559e-01
-7.62149453e-01 7.79692411e-01 1.12007999e+00 -3.30812261e-02
2.21894793e-02 -1.27475783e-01 -2.75336027e-01 4.77697015e-01
1.45497829e-01 -3.82400125e-01 1.52834356e-01 -3.24310035e-01
1.36087775e-01 -1.50737852e-01 3.93479943e-01 -8.56574833e-01
3.97925615e-01 -1.73096918e-02 3.46526504e-01 -1.10206866e+00
5.73357940e-01 -1.03235698e+00 2.63318419e-02 6.03582919e-01
2.60944724e-01 4.70240414e-01 -2.27771653e-03 5.88186979e-01
-2.27910772e-01 -8.46265536e-03 6.83099926e-01 2.21181512e-01
-7.31024623e-01 4.84590173e-01 1.38190389e-01 -1.91263109e-01
1.07522893e+00 -3.23143840e-01 -2.02821076e-01 -3.68827254e-01
-4.48987782e-01 -1.23436943e-01 8.99472713e-01 4.22607303e-01
1.02924991e+00 -1.91099048e+00 -6.84852123e-01 4.04274344e-01
3.53704393e-01 -3.63087386e-01 -2.31707349e-01 8.94183517e-01
-6.54644668e-01 5.85473776e-01 -2.93858230e-01 -1.05672193e+00
-1.20603538e+00 8.18311870e-01 2.37393565e-02 1.09236650e-01
-4.72110480e-01 4.18680102e-01 -2.07813680e-01 -7.24488318e-01
3.01987618e-01 -3.43838990e-01 7.73502439e-02 4.18495387e-02
3.58374536e-01 4.84002054e-01 1.23042166e-01 -1.00496173e+00
-8.16822529e-01 1.29245257e+00 3.08754861e-01 -2.08273828e-01
1.40406311e+00 -1.13903314e-01 -1.15384027e-01 5.46624303e-01
1.49051452e+00 2.23319992e-01 -1.02225351e+00 -3.94301772e-01
1.65894963e-02 -7.63842940e-01 -4.16145802e-01 -2.93604955e-02
-6.89859748e-01 8.17603528e-01 7.78028488e-01 -6.73285350e-02
8.40546966e-01 -8.10936615e-02 7.82203674e-01 7.95442462e-01
6.70579314e-01 -7.09610105e-01 7.17617348e-02 8.97808254e-01
9.24097538e-01 -9.74293232e-01 4.72254694e-01 -2.82810867e-01
-4.84655350e-01 1.48969865e+00 1.04503393e-01 -5.91316462e-01
8.60387444e-01 -2.92771775e-03 -9.56920460e-02 1.76408023e-01
-3.50719273e-01 -3.19060981e-01 6.01213276e-01 7.09139824e-01
1.10209689e-01 -6.80755675e-02 -2.34059826e-01 4.12515029e-02
-1.72498226e-02 -2.99686700e-01 4.37192738e-01 1.10971069e+00
-3.07323903e-01 -1.31708980e+00 -8.03958058e-01 -3.94763947e-01
-1.80206954e-01 2.97746599e-01 -4.61497426e-01 8.93195987e-01
-3.13441992e-01 3.87696803e-01 3.40584964e-02 -5.18010616e-01
6.57133460e-01 1.47812515e-01 9.14040923e-01 -3.75461608e-01
-2.02969134e-01 2.65569121e-01 -4.63385165e-01 -8.46878171e-01
-6.62263989e-01 -8.47046018e-01 -1.03294826e+00 -1.46588594e-01
-2.21527457e-01 -9.38688815e-02 1.13565230e+00 3.61902118e-01
4.41340566e-01 -5.00116587e-01 1.15215719e+00 -1.03559816e+00
-1.37085450e+00 -3.95467937e-01 -9.68793273e-01 6.01978779e-01
5.17274678e-01 -9.13049936e-01 -4.92851555e-01 -2.33297795e-01] | [7.803235054016113, -2.2867112159729004] |
d7aa82c3-9f41-45fa-8551-0de0ec6301b5 | self-supervised-image-clustering-from | 2209.11927 | null | https://arxiv.org/abs/2209.11927v1 | https://arxiv.org/pdf/2209.11927v1.pdf | Self-supervised Image Clustering from Multiple Incomplete Views via Constrastive Complementary Generation | Incomplete Multi-View Clustering aims to enhance clustering performance by using data from multiple modalities. Despite the fact that several approaches for studying this issue have been proposed, the following drawbacks still persist: 1) It's difficult to learn latent representations that account for complementarity yet consistency without using label information; 2) and thus fails to take full advantage of the hidden information in incomplete data results in suboptimal clustering performance when complete data is scarce. In this paper, we propose Contrastive Incomplete Multi-View Image Clustering with Generative Adversarial Networks (CIMIC-GAN), which uses GAN to fill in incomplete data and uses double contrastive learning to learn consistency on complete and incomplete data. More specifically, considering diversity and complementary information among multiple modalities, we incorporate autoencoding representation of complete and incomplete data into double contrastive learning to achieve learning consistency. Integrating GANs into the autoencoding process can not only take full advantage of new features of incomplete data, but also better generalize the model in the presence of high data missing rates. Experiments conducted on \textcolor{black}{four} extensively-used datasets show that CIMIC-GAN outperforms state-of-the-art incomplete multi-View clustering methods. | ['Limin Liu', 'Dongjin Guo', 'Xuewen Yang', 'Zhiwei Xu', 'Jiatai Wang'] | 2022-09-24 | null | null | null | null | ['incomplete-multi-view-clustering', 'image-clustering'] | ['computer-vision', 'computer-vision'] | [ 7.13727698e-02 -9.19869989e-02 -1.74678341e-01 -2.67435670e-01
-1.02026916e+00 -5.89899242e-01 4.46601301e-01 -5.01890182e-01
-1.11684268e-02 8.55991423e-01 3.07755977e-01 3.78273934e-01
-5.54092675e-02 -7.00899124e-01 -6.82000399e-01 -1.07463408e+00
4.79706705e-01 6.51325643e-01 -3.31261992e-01 2.07443178e-01
-2.86028713e-01 1.10917307e-01 -1.54848552e+00 6.01154387e-01
1.16829455e+00 5.35692334e-01 2.68895239e-01 3.07489038e-01
-1.03185654e-01 9.97861385e-01 -5.40104032e-01 -3.17069679e-01
2.69523472e-01 -9.42363918e-01 -6.57434642e-01 4.94594246e-01
2.40125418e-01 -2.76638567e-01 -2.87205964e-01 9.60657656e-01
5.17709911e-01 -6.72096014e-02 6.56187177e-01 -1.66422606e+00
-9.70320761e-01 5.89359641e-01 -8.69836152e-01 -3.56343299e-01
1.55347869e-01 1.53982453e-02 7.56428719e-01 -6.80095732e-01
8.56470406e-01 1.08209300e+00 6.49773836e-01 8.64017606e-01
-1.22330868e+00 -6.78727150e-01 9.20384377e-02 1.74904495e-01
-1.42924190e+00 -3.23151350e-01 1.07350028e+00 -3.26530755e-01
2.59337217e-01 2.62898177e-01 5.05064666e-01 1.33840930e+00
-1.63751006e-01 9.14569139e-01 1.51903152e+00 -3.27302486e-01
3.15628290e-01 1.25987098e-01 -3.59084576e-01 5.27100742e-01
3.78629528e-02 1.35498773e-02 -4.81751204e-01 -1.18689626e-01
5.67644060e-01 4.53720331e-01 -2.81886876e-01 -4.32964236e-01
-1.17262435e+00 8.62103224e-01 4.54440296e-01 5.00225484e-01
-4.30581123e-01 2.17524794e-05 2.21105486e-01 1.24652132e-01
4.61478859e-01 1.23871796e-01 -1.34790733e-01 2.22660918e-02
-1.08838904e+00 -2.42813617e-01 3.85831594e-01 1.03897750e+00
9.20848131e-01 3.81108552e-01 1.34874031e-01 9.54555213e-01
4.99386452e-02 6.57799721e-01 4.95215476e-01 -1.41955209e+00
5.94624639e-01 8.78242195e-01 -3.10218662e-01 -6.07865989e-01
-1.85478523e-01 -4.37527090e-01 -1.54564321e+00 2.38350004e-01
1.92279041e-01 -1.96331128e-01 -1.04311419e+00 1.99308455e+00
1.36690810e-01 1.14344694e-01 2.82809079e-01 7.85764277e-01
8.87035251e-01 5.47868252e-01 -2.07689375e-01 -2.35067889e-01
9.72447157e-01 -9.47750390e-01 -8.78429532e-01 -6.10303581e-02
3.38538587e-01 -6.25629127e-01 7.89383650e-01 5.06325364e-01
-1.14173234e+00 -6.24052882e-01 -1.09779179e+00 -1.40897073e-02
-4.17308867e-01 2.17624202e-01 4.54759061e-01 7.12314010e-01
-1.13681865e+00 2.10084066e-01 -1.00034726e+00 -1.29054189e-01
5.46324968e-01 3.00352663e-01 -6.17572963e-01 -7.00711071e-01
-8.55215907e-01 4.82965112e-01 3.61574233e-01 -4.38923500e-02
-9.62358892e-01 -7.15291440e-01 -8.26120973e-01 -8.34939852e-02
4.75887418e-01 -9.24114943e-01 3.67524773e-01 -1.20561683e+00
-1.13235736e+00 6.53135538e-01 -2.02958763e-01 2.74328589e-02
4.47063416e-01 3.28108110e-02 -4.09658194e-01 3.02202672e-01
2.00219274e-01 9.28270042e-01 8.34507287e-01 -2.09220648e+00
-2.86590457e-01 -6.99122787e-01 -1.51065990e-01 3.49584609e-01
-2.22573534e-01 -5.92722297e-01 -7.10188925e-01 -7.74004102e-01
3.05675715e-01 -1.14571798e+00 -4.17163186e-02 -3.42691183e-01
-5.29494524e-01 2.54528642e-01 1.22809160e+00 -9.61592674e-01
9.15702999e-01 -2.10429001e+00 6.58873796e-01 5.81504479e-02
2.79069662e-01 -7.15589337e-03 -2.02659130e-01 5.80718756e-01
2.52334494e-02 1.60228431e-01 -5.90034783e-01 -6.34335935e-01
-1.37505561e-01 6.33309364e-01 4.50818613e-02 3.96278411e-01
-4.82002832e-02 9.22158241e-01 -6.16011620e-01 -5.02907872e-01
4.75136369e-01 8.07524920e-01 -5.50221324e-01 2.18324900e-01
-1.15349986e-01 1.05907476e+00 -2.14031905e-01 8.00692797e-01
9.23419476e-01 -4.69834566e-01 3.38716298e-01 -1.72892556e-01
3.13555211e-01 -4.97370183e-01 -1.27425218e+00 2.17712903e+00
-4.54187095e-01 2.97686815e-01 2.78180450e-01 -9.94452119e-01
5.38378537e-01 4.75755364e-01 7.80732095e-01 -6.94243193e-01
-2.17037514e-01 3.33127566e-02 -2.77939588e-01 -2.68140554e-01
3.62188220e-01 -3.15854788e-01 -1.53352946e-01 5.63239515e-01
2.07526654e-01 -8.72235596e-02 -1.68841034e-02 3.75920981e-01
8.62361431e-01 2.64819771e-01 -1.04679763e-01 2.25944072e-01
5.00419855e-01 -2.05270067e-01 7.75433183e-01 5.23567319e-01
1.67231679e-01 1.22438800e+00 3.08769673e-01 -6.93529546e-02
-1.20171559e+00 -9.78347123e-01 2.74995357e-01 7.05692291e-01
6.18890300e-02 -2.70273983e-01 -8.17315936e-01 -8.13993633e-01
-2.78513640e-01 5.71530163e-01 -8.81109476e-01 -6.65686354e-02
-2.78851777e-01 -1.00419915e+00 5.26709139e-01 7.14435637e-01
7.93360174e-01 -8.06753635e-01 -1.35792941e-01 -1.06324047e-01
-6.92317724e-01 -1.16521800e+00 -2.76089519e-01 9.52755883e-02
-8.33843172e-01 -1.11804748e+00 -8.14128220e-01 -4.82578337e-01
8.84958208e-01 3.45478892e-01 1.08015561e+00 8.13991502e-02
-2.00704619e-01 7.69266784e-01 -5.73181272e-01 1.22983702e-01
-7.04308867e-01 -1.18184146e-02 -2.43015200e-01 1.67420983e-01
7.20335916e-02 -5.79258561e-01 -4.58953112e-01 3.79207879e-01
-1.36947441e+00 2.88598418e-01 5.83715916e-01 1.24558473e+00
7.66819239e-01 3.28637779e-01 6.16222262e-01 -1.23416102e+00
2.12285891e-02 -4.41143930e-01 -1.64987609e-01 6.58068776e-01
-5.92870951e-01 -1.59288775e-02 7.03076124e-01 1.68434121e-02
-1.32226205e+00 1.16881989e-01 -1.86471298e-01 -8.63057733e-01
-3.06598872e-01 3.13704789e-01 -6.21435463e-01 2.18439013e-01
1.73947752e-01 3.83122116e-01 2.54575938e-01 -3.97821397e-01
6.84190929e-01 5.00180006e-01 4.80124474e-01 -4.97831047e-01
5.74520528e-01 9.59744930e-01 -4.54459116e-02 -5.58069587e-01
-6.27363861e-01 -3.44888985e-01 -8.80012751e-01 -5.16057238e-02
1.16391075e+00 -1.36399329e+00 -4.10828471e-01 6.87923372e-01
-6.97901428e-01 -1.27074972e-01 -4.65839803e-01 3.79930735e-01
-6.54258251e-01 5.56297362e-01 -5.95334589e-01 -4.73430842e-01
-1.47610262e-01 -1.31882644e+00 1.02998030e+00 5.77930100e-02
2.71101177e-01 -1.20245397e+00 -1.18347377e-01 8.99316192e-01
2.64886409e-01 5.93796015e-01 7.98445582e-01 -3.49014789e-01
-5.82216442e-01 -9.08888280e-02 5.66657633e-02 7.20498860e-01
4.13098544e-01 -1.23635322e-01 -1.13273656e+00 -6.43671036e-01
1.47525683e-01 -5.00674129e-01 8.80509853e-01 4.60870773e-01
1.25747514e+00 -2.98245788e-01 -2.47740999e-01 7.00025141e-01
1.72652423e+00 1.06952302e-01 7.66127229e-01 -1.88829657e-02
1.10302830e+00 5.99711478e-01 3.26863319e-01 5.30929685e-01
6.39023125e-01 4.84513879e-01 6.73607111e-01 -2.72852361e-01
-4.58993107e-01 -3.81559402e-01 1.82550102e-01 1.22905970e+00
-1.28712535e-01 -3.51292640e-01 -6.60450459e-01 5.23850143e-01
-1.75281644e+00 -1.25541401e+00 -4.47238758e-02 2.05372238e+00
5.58401823e-01 -4.30918276e-01 -2.01352164e-02 2.10782439e-01
9.26200330e-01 1.31412178e-01 -6.44175768e-01 2.87357897e-01
-5.16113698e-01 -1.37405157e-01 2.86878735e-01 2.68078178e-01
-9.06692803e-01 5.11444926e-01 5.54723358e+00 8.04288149e-01
-8.62192035e-01 5.61541080e-01 8.38898838e-01 -1.01651177e-01
-6.73085570e-01 1.13147058e-01 -2.77934074e-01 6.48891985e-01
4.80387598e-01 5.02826869e-01 8.13827813e-01 4.82568473e-01
-1.96354017e-01 -8.48235339e-02 -1.00834560e+00 1.11158800e+00
3.88823122e-01 -1.12159026e+00 1.76567957e-01 2.09110186e-01
1.46244419e+00 -1.33004650e-01 2.99519926e-01 8.69035348e-02
5.69688320e-01 -9.92959499e-01 4.94635850e-01 4.99858290e-01
8.86618733e-01 -9.30026591e-01 7.53025055e-01 5.11224031e-01
-9.88956690e-01 -1.46573618e-01 -2.32441351e-01 4.18396562e-01
5.28532229e-02 6.19564950e-01 -4.58488882e-01 1.23300695e+00
6.95349216e-01 8.51383448e-01 -7.38919079e-01 4.63354528e-01
-1.74381569e-01 4.62955594e-01 -9.76742133e-02 8.02453637e-01
7.83069581e-02 -4.44881320e-01 1.74255371e-01 6.26464903e-01
3.80347610e-01 4.08263616e-02 5.97488992e-02 9.96551871e-01
-1.32865146e-01 -2.86346614e-01 -7.43295133e-01 1.13289513e-01
4.71409678e-01 1.04888034e+00 -6.27717972e-01 -3.37519169e-01
-6.65383875e-01 1.29333091e+00 1.29338697e-01 6.24664783e-01
-7.20456958e-01 4.40578550e-01 3.74573916e-02 -1.95001021e-01
3.75307322e-01 -8.42251852e-02 -4.38031673e-01 -1.40364981e+00
-4.16385047e-02 -1.11837113e+00 7.41423190e-01 -1.01206660e+00
-1.48430610e+00 5.48817217e-01 -8.37793127e-02 -1.34613252e+00
-4.03505355e-01 3.49589176e-02 -2.63892502e-01 7.36295164e-01
-1.23123407e+00 -1.93217862e+00 -5.28020084e-01 9.54563618e-01
6.29129231e-01 -3.53965670e-01 8.28716278e-01 3.30102354e-01
-5.65154791e-01 4.38913435e-01 5.57256341e-01 6.02461286e-02
7.82671213e-01 -1.35966027e+00 -1.67576909e-01 8.23784471e-01
2.91512430e-01 3.67655814e-01 3.08433175e-01 -6.38986468e-01
-1.43060172e+00 -1.04855978e+00 1.45229101e-01 -3.53508919e-01
-6.16679341e-02 -1.77401274e-01 -9.18456912e-01 7.35979617e-01
6.72891974e-01 1.59541279e-01 1.12650764e+00 -1.78102143e-02
-4.72223490e-01 -1.91421151e-01 -1.24975741e+00 1.98686585e-01
6.88913703e-01 -6.46125913e-01 -2.72216499e-01 -5.62350862e-02
4.50438976e-01 -1.92025766e-01 -1.10691726e+00 5.47665417e-01
2.02363700e-01 -1.22267544e+00 9.15087581e-01 -2.32126758e-01
5.90942740e-01 -3.75938714e-01 -4.36551601e-01 -1.52904058e+00
-1.66404575e-01 -2.16350332e-01 2.10005697e-02 1.61473787e+00
5.73586561e-02 -3.91998440e-01 8.08252752e-01 5.70434690e-01
-2.21813500e-01 -4.72745150e-01 -1.08472681e+00 -5.17050922e-01
3.34803224e-01 -5.95577210e-02 7.02350199e-01 1.39505970e+00
-3.21047962e-01 9.30427462e-02 -7.65461266e-01 1.92688152e-01
8.21446121e-01 2.78799981e-01 7.51162529e-01 -9.78330612e-01
-3.98852646e-01 6.61527878e-03 -5.47016710e-02 -5.45090318e-01
2.23892659e-01 -8.84383857e-01 -3.75821322e-01 -1.77098560e+00
7.36131966e-01 -6.17719531e-01 -1.95816159e-01 6.23384893e-01
-2.29199067e-01 5.17575681e-01 4.08252001e-01 4.98264641e-01
-6.66080117e-01 8.53319705e-01 1.42051685e+00 -3.73883396e-01
7.82066211e-02 -2.86222607e-01 -7.13496566e-01 5.01852751e-01
5.58348835e-01 -4.20610845e-01 -6.35931313e-01 -5.25833547e-01
7.23393187e-02 4.77428705e-01 5.10229468e-01 -1.15271401e+00
2.51360536e-01 3.11573595e-02 7.25311756e-01 -8.14492881e-01
6.16157949e-01 -1.13490009e+00 7.49600112e-01 2.98435062e-01
-1.24756746e-01 2.17931315e-01 1.31444298e-02 8.22921157e-01
-5.40353417e-01 -1.15718856e-01 7.23145664e-01 -5.29897153e-01
-5.90878665e-01 1.03350826e-01 -1.92575268e-02 1.06419452e-01
1.07373691e+00 -1.13106743e-01 -4.15706724e-01 -5.52629411e-01
-1.00622404e+00 2.23441407e-01 1.05714083e+00 4.46074516e-01
6.00072563e-01 -1.55269802e+00 -7.44523525e-01 2.82975346e-01
-2.77287345e-02 1.57301277e-01 9.42047119e-01 6.94075227e-01
-3.02414775e-01 8.96737054e-02 -4.02625442e-01 -8.50751340e-01
-1.11324728e+00 8.38854253e-01 1.32295057e-01 -3.75708044e-01
-4.46686119e-01 3.65679026e-01 2.57069945e-01 -6.62962735e-01
-1.75020490e-02 3.41252029e-01 5.94293363e-02 1.54076740e-01
7.63621461e-03 5.47261477e-01 2.74653789e-02 -7.45046139e-01
-1.82475358e-01 5.76553702e-01 9.54635218e-02 -2.55306333e-01
1.44470775e+00 -4.90097970e-01 -2.53555954e-01 4.53145832e-01
1.24394608e+00 4.11352888e-02 -1.42972457e+00 -1.41032666e-01
-5.63184202e-01 -4.76327688e-01 -1.69662893e-01 -9.73339438e-01
-1.69964468e+00 9.90285337e-01 6.59004033e-01 2.39419788e-02
1.47656810e+00 -1.13530997e-02 6.26208305e-01 -1.59182116e-01
2.76085705e-01 -1.17282856e+00 3.25352281e-01 -1.23333715e-01
6.88974142e-01 -1.41900802e+00 1.78375125e-01 -3.79832357e-01
-1.08534181e+00 8.27719212e-01 7.09848821e-01 1.98342726e-01
4.54880953e-01 1.43705040e-01 1.07537739e-01 -2.39118025e-01
-5.87867081e-01 -2.05608711e-01 6.75159469e-02 7.25013196e-01
2.09130883e-01 8.33015367e-02 5.39147928e-02 4.09615755e-01
2.95987040e-01 -2.28625804e-01 5.59890687e-01 8.62993598e-01
2.02299505e-01 -1.35515618e+00 -6.58276558e-01 3.50275517e-01
-5.01136839e-01 1.92772791e-01 -3.35500002e-01 7.84545183e-01
4.58502501e-01 1.04240322e+00 -1.15143470e-01 -4.41402584e-01
-1.68541804e-01 1.15650125e-01 4.74657685e-01 -3.53721708e-01
-3.89871180e-01 2.96494871e-01 -4.52636600e-01 -2.85047770e-01
-1.04175103e+00 -8.40610504e-01 -1.04727924e+00 -2.61939615e-01
-3.19178462e-01 1.07490174e-01 6.91482365e-01 8.22794795e-01
5.23362517e-01 7.53065288e-01 8.21069956e-01 -6.22926056e-01
-3.50564420e-01 -8.23624730e-01 -7.38414347e-01 8.51097882e-01
1.60818592e-01 -5.54056883e-01 -3.75534743e-01 3.10683131e-01] | [8.46117877960205, 4.520175457000732] |
01f265b0-cbda-4a61-a8c8-56454f81ea12 | visual-relationship-detection-using-scene | 2005.08045 | null | https://arxiv.org/abs/2005.08045v1 | https://arxiv.org/pdf/2005.08045v1.pdf | Visual Relationship Detection using Scene Graphs: A Survey | Understanding a scene by decoding the visual relationships depicted in an image has been a long studied problem. While the recent advances in deep learning and the usage of deep neural networks have achieved near human accuracy on many tasks, there still exists a pretty big gap between human and machine level performance when it comes to various visual relationship detection tasks. Developing on earlier tasks like object recognition, segmentation and captioning which focused on a relatively coarser image understanding, newer tasks have been introduced recently to deal with a finer level of image understanding. A Scene Graph is one such technique to better represent a scene and the various relationships present in it. With its wide number of applications in various tasks like Visual Question Answering, Semantic Image Retrieval, Image Generation, among many others, it has proved to be a useful tool for deeper and better visual relationship understanding. In this paper, we present a detailed survey on the various techniques for scene graph generation, their efficacy to represent visual relationships and how it has been used to solve various downstream tasks. We also attempt to analyze the various future directions in which the field might advance in the future. Being one of the first papers to give a detailed survey on this topic, we also hope to give a succinct introduction to scene graphs, and guide practitioners while developing approaches for their applications. | ['Ayush Mangal', 'Vipul', 'Aniket Agarwal'] | 2020-05-16 | null | null | null | null | ['visual-relationship-detection'] | ['computer-vision'] | [ 5.50423801e-01 1.46229401e-01 -1.48895621e-01 -5.46600223e-01
-1.56296670e-01 -4.66381341e-01 7.32959628e-01 5.06238580e-01
2.15784013e-02 4.81543958e-01 7.61422664e-02 -4.68766421e-01
-1.82994246e-01 -8.47727478e-01 -4.98304933e-01 -4.25683558e-01
-1.07192323e-01 4.31426078e-01 3.31394732e-01 -2.77076155e-01
4.39095318e-01 6.69320524e-01 -1.84118116e+00 5.43677449e-01
3.63851786e-01 8.13630164e-01 3.71605575e-01 7.15940952e-01
-4.55340743e-01 1.18467510e+00 -6.29560828e-01 -5.65789104e-01
-1.29296541e-01 -4.32641685e-01 -1.11537981e+00 4.26378965e-01
5.77456534e-01 3.94001380e-02 -2.94705808e-01 1.05531657e+00
1.11147240e-01 5.93995675e-02 5.47484159e-01 -1.49814558e+00
-6.45747364e-01 2.81844556e-01 -7.16083944e-01 3.81234258e-01
5.70633948e-01 -2.86650598e-01 1.04262173e+00 -5.58122456e-01
6.38295591e-01 1.34671676e+00 2.90034920e-01 2.93733150e-01
-9.68416750e-01 -3.51049244e-01 1.67793021e-01 7.21686125e-01
-1.13489687e+00 -1.78689472e-02 9.50595796e-01 -4.29506570e-01
9.99083638e-01 3.99847060e-01 7.00674713e-01 5.27983844e-01
7.98649639e-02 1.00055230e+00 1.07964826e+00 -6.92947090e-01
-7.28036314e-02 3.03035557e-01 2.97919095e-01 7.73683846e-01
5.68712167e-02 -2.30827674e-01 -3.08502793e-01 2.85321534e-01
8.54204893e-01 -1.08257808e-01 -1.72722086e-01 -6.39196515e-01
-1.13249707e+00 9.05643642e-01 9.40611422e-01 6.45872951e-01
-1.14187628e-01 1.48149967e-01 3.69365305e-01 1.67452380e-01
2.15059027e-01 5.64502478e-01 -7.39095360e-02 2.34362483e-01
-6.99765265e-01 2.10081130e-01 7.38787889e-01 8.70113850e-01
8.20717335e-01 1.65721420e-02 4.57501858e-02 8.53227973e-01
7.56818950e-02 -6.79202750e-03 1.26554906e-01 -8.59687984e-01
3.33674580e-01 8.54976535e-01 -2.01022267e-01 -1.58047140e+00
-4.28412795e-01 -3.26819003e-01 -8.97237480e-01 2.84005761e-01
4.93156582e-01 3.60348403e-01 -1.13078094e+00 1.43866289e+00
1.22101903e-01 -1.23381868e-01 -1.82364598e-01 8.72535110e-01
1.20225251e+00 8.91773999e-01 1.35618746e-01 9.11446512e-02
1.64325750e+00 -9.53292131e-01 -6.38137579e-01 -5.48712015e-01
3.91612560e-01 -9.39218700e-01 8.41521144e-01 3.70082766e-01
-8.99862051e-01 -6.87878251e-01 -1.01446509e+00 -3.99418384e-01
-8.46878469e-01 -1.75098941e-01 1.03667355e+00 4.93480355e-01
-1.24060118e+00 3.46387953e-01 -3.94105405e-01 -8.55531216e-01
6.53994322e-01 3.38474959e-01 -4.57170695e-01 -2.61072636e-01
-1.00964046e+00 9.98194873e-01 6.03198051e-01 1.47990525e-01
-5.64827442e-01 -2.42984056e-01 -9.80909467e-01 4.66801636e-02
4.41147566e-01 -7.57581353e-01 1.07825661e+00 -1.08848536e+00
-8.04510832e-01 1.45687246e+00 -1.96796730e-01 -4.45871085e-01
1.39791459e-01 1.99563038e-02 -3.42535853e-01 2.49482363e-01
-1.96766648e-02 9.90574181e-01 5.50390780e-01 -1.53289235e+00
-5.99229157e-01 -3.61157656e-01 3.98170918e-01 4.51622963e-01
-2.17166189e-02 3.53638500e-01 -6.46518528e-01 -4.49661851e-01
2.53603160e-01 -5.79500735e-01 -3.17201078e-01 1.26591012e-01
-4.83714432e-01 -3.65041852e-01 1.12744021e+00 -4.19528812e-01
8.52950156e-01 -1.80243778e+00 2.24081025e-01 2.61751730e-02
4.37621295e-01 5.63952565e-01 -8.48821849e-02 7.45580733e-01
-4.89243150e-01 8.75449553e-02 -3.83729696e-01 -1.33176461e-01
-2.95425147e-01 4.66670364e-01 -2.73958087e-01 2.25671008e-01
1.74524710e-01 1.15964949e+00 -9.25183415e-01 -5.98044395e-01
7.03224123e-01 5.79591215e-01 -8.75234157e-02 1.27371863e-01
-3.17299902e-01 4.75996792e-01 -3.09935212e-01 4.63664949e-01
4.66766685e-01 -5.12441695e-01 1.80206001e-02 -1.61106765e-01
-3.35721765e-03 -1.00367377e-02 -9.69963908e-01 1.43057823e+00
-2.49568954e-01 1.36652482e+00 -1.48189619e-01 -1.57280564e+00
1.05506372e+00 1.27606407e-01 3.09875250e-01 -9.33820009e-01
2.17228085e-01 -1.52448744e-01 5.24272360e-02 -7.49483645e-01
5.83890617e-01 -2.79245615e-01 2.50649434e-02 2.27572188e-01
-2.73875922e-01 -4.43962485e-01 3.35894972e-01 3.33578765e-01
6.74333692e-01 4.66212258e-02 7.14379489e-01 3.86879295e-02
7.76860714e-01 3.64594281e-01 -1.48597002e-01 6.10108078e-01
-2.13453293e-01 5.70451200e-01 6.62728548e-01 -7.79826105e-01
-1.04245770e+00 -9.04555023e-01 -1.78381912e-02 9.26667809e-01
5.13741314e-01 -3.47371876e-01 -5.94746053e-01 -3.39554012e-01
-3.16652834e-01 6.18096054e-01 -6.22101426e-01 1.94959879e-01
-7.26748466e-01 -6.97278202e-01 1.92591384e-01 6.08287334e-01
7.57982671e-01 -1.50318539e+00 -6.53150439e-01 -7.55603472e-03
-7.50264451e-02 -1.40068638e+00 2.51834720e-01 9.02190879e-02
-9.01053846e-01 -1.17455351e+00 -7.70412087e-01 -1.32061887e+00
7.85078943e-01 6.56397223e-01 1.36151636e+00 3.66200835e-01
-6.03289723e-01 4.50339705e-01 -5.07413745e-01 -5.02567887e-01
-2.76805520e-01 -2.33389661e-01 -6.71658993e-01 -1.89228281e-01
4.14677113e-01 -3.45003515e-01 -4.59170163e-01 8.48173127e-02
-1.26170957e+00 4.25659180e-01 4.00929600e-01 5.08696258e-01
5.29303014e-01 2.60422111e-01 8.00278410e-02 -1.05100632e+00
6.91871285e-01 -3.59703004e-01 -2.61462033e-01 3.87936294e-01
-8.95669609e-02 1.76904202e-02 5.02374768e-01 1.15940437e-01
-9.01589572e-01 6.97752982e-02 -1.97981998e-01 -1.07661247e-01
-5.76607049e-01 4.72096622e-01 -5.63223474e-02 -6.98043853e-02
5.11424303e-01 2.02484950e-01 -1.63803726e-01 -2.90011495e-01
7.69691944e-01 4.54739481e-01 6.55576706e-01 -1.46281824e-01
6.83649182e-01 5.62400937e-01 2.01121449e-01 -1.08244634e+00
-1.12866747e+00 -8.01365256e-01 -9.07366991e-01 -4.23606485e-01
1.20955515e+00 -3.22217852e-01 -5.59638262e-01 3.95936668e-01
-1.34100795e+00 -2.42077351e-01 1.33401630e-02 -1.54689923e-02
-5.82758188e-01 6.73020065e-01 -2.36199826e-01 -5.74432492e-01
-5.66137694e-02 -1.06524456e+00 1.04785669e+00 5.28298318e-01
-1.20151863e-01 -1.39523840e+00 -2.27225631e-01 6.04809761e-01
2.05238938e-01 5.06724000e-01 1.19401419e+00 -3.35484803e-01
-7.26947665e-01 -1.90866947e-01 -8.20607841e-01 1.92118242e-01
1.16466075e-01 -1.29141450e-01 -9.77960408e-01 -8.26021936e-03
-1.07440338e-01 -1.81790173e-01 8.58921826e-01 3.21654230e-01
1.43297732e+00 -1.71585549e-02 -5.30432820e-01 4.10926819e-01
1.65047753e+00 5.46667635e-01 9.02150214e-01 4.98415887e-01
9.84205127e-01 1.08095753e+00 5.74953377e-01 -1.13128766e-01
3.71764034e-01 7.54345894e-01 8.08162749e-01 -6.12790883e-01
-5.24343431e-01 -5.86577021e-02 -1.68567628e-01 5.33433616e-01
-1.48724049e-01 -5.23239732e-01 -1.14782298e+00 4.95693386e-01
-1.89839864e+00 -1.07436538e+00 -5.87502301e-01 1.96491551e+00
2.40830898e-01 -5.95794898e-03 1.03357442e-01 3.03845227e-01
9.49817300e-01 3.38180512e-01 -3.52598965e-01 -7.18126416e-01
-1.18623190e-01 2.28374258e-01 1.91402480e-01 4.04492289e-01
-1.19976163e+00 1.25588906e+00 6.56221771e+00 5.06239176e-01
-1.09039259e+00 -4.06787574e-01 8.61164093e-01 6.83445096e-01
-1.06289826e-01 2.53068022e-02 -4.57821429e-01 -4.58913781e-02
3.36205035e-01 -1.08380187e-02 2.01912686e-01 7.25596786e-01
-5.35416305e-02 -4.87768084e-01 -1.04486620e+00 1.32631683e+00
3.65583122e-01 -1.42012060e+00 3.24040085e-01 -1.27885789e-01
5.64790010e-01 -3.68188411e-01 5.87923042e-02 -3.73218238e-04
-6.53591603e-02 -1.42443466e+00 4.67698157e-01 3.33679736e-01
5.59773207e-01 -7.03725278e-01 7.34984338e-01 2.56362110e-01
-1.42718685e+00 8.83630812e-02 -4.23984677e-01 -5.09789586e-01
2.30325401e-01 3.60046357e-01 -1.02472901e+00 6.10210180e-01
6.93538189e-01 7.66967773e-01 -7.94070125e-01 1.42732477e+00
-3.92418414e-01 1.06486030e-01 8.00011829e-02 -3.21469873e-01
4.90517050e-01 -1.54419810e-01 3.94716471e-01 1.26434112e+00
-1.71828285e-01 8.37869495e-02 -3.66544835e-02 7.59741187e-01
1.60478726e-01 1.85714766e-01 -8.86604965e-01 -7.48448074e-02
-3.25531363e-02 1.30139887e+00 -1.30814505e+00 -4.75643128e-01
-5.46007276e-01 9.77855563e-01 2.70849884e-01 3.19260925e-01
-7.99371004e-01 -5.72808325e-01 3.67346644e-01 1.63686842e-01
5.57398871e-02 -5.31700611e-01 -4.22980756e-01 -8.32843900e-01
-2.50782102e-01 -7.92301536e-01 6.29524469e-01 -1.32305324e+00
-1.01846957e+00 7.43515074e-01 3.67594153e-01 -1.02554262e+00
-1.72941461e-01 -8.99028659e-01 -6.25356138e-01 6.94350183e-01
-1.53826952e+00 -1.01154447e+00 -4.99616593e-01 5.23297727e-01
7.97630250e-01 1.27157927e-01 6.82844818e-01 4.34846096e-02
-3.36347431e-01 -4.70048897e-02 -2.89732248e-01 3.46571058e-01
2.54808426e-01 -1.22556007e+00 4.43553656e-01 9.01441753e-01
7.47057796e-01 4.85065371e-01 9.37992930e-01 -3.58053863e-01
-9.31010187e-01 -7.34000623e-01 9.55235302e-01 -3.61514390e-01
5.61547339e-01 -4.14363354e-01 -9.12873268e-01 6.22151017e-01
6.18255317e-01 -3.03708255e-01 3.98345798e-01 -1.36523545e-01
-2.48453155e-01 -2.60321703e-02 -7.97515988e-01 7.99862862e-01
9.36691523e-01 -5.31527638e-01 -7.72447109e-01 5.45171916e-01
4.53195840e-01 -2.96613574e-01 -4.14522827e-01 3.48262459e-01
3.18816930e-01 -1.36683536e+00 1.13453686e+00 -5.69476426e-01
4.41228986e-01 -4.31015819e-01 2.67762523e-02 -9.44893181e-01
-2.54193157e-01 -3.06173533e-01 3.71673405e-01 9.97582197e-01
7.67356530e-02 -3.07769924e-01 9.64648426e-01 3.74214947e-01
-2.08451319e-02 -7.70827353e-01 -4.70114976e-01 -4.35570359e-01
-2.30833739e-01 -5.17849982e-01 1.84612721e-01 8.84261966e-01
-1.88067243e-01 7.02203691e-01 -3.30258250e-01 -7.52579719e-02
5.85503817e-01 4.00593787e-01 7.87792325e-01 -1.27479005e+00
1.83801837e-02 -6.39961123e-01 -8.40912759e-01 -1.15009272e+00
-1.69765186e-02 -9.20835555e-01 -8.75623003e-02 -2.30645323e+00
3.54689956e-01 -1.07447639e-01 3.91039327e-02 3.94311100e-01
-1.20634735e-01 6.69219375e-01 3.42687517e-01 5.08786216e-02
-7.20072925e-01 9.53233242e-02 1.64358854e+00 -2.03896195e-01
2.25509182e-02 1.77263655e-02 -7.56461859e-01 8.66267741e-01
1.00415516e+00 -1.98530614e-01 -5.69621146e-01 -5.12174606e-01
2.80559361e-01 -1.06281927e-02 5.46654761e-01 -9.87917364e-01
2.82189906e-01 3.51013057e-02 4.11494821e-01 -8.07819188e-01
4.60432589e-01 -7.39981413e-01 7.62781277e-02 4.41876411e-01
-3.13409328e-01 1.66437149e-01 3.35161239e-01 4.64153469e-01
-5.39777875e-01 -3.88527870e-01 9.38775182e-01 -5.08958876e-01
-1.52456307e+00 5.66871502e-02 -4.29478616e-01 -2.12103605e-01
1.29101741e+00 -6.81620896e-01 -2.93939590e-01 -7.82366633e-01
-8.25357854e-01 1.65934548e-01 3.45623136e-01 6.52383924e-01
7.57444382e-01 -8.89264405e-01 -3.84419769e-01 -1.27484784e-01
2.28822529e-01 -5.11715971e-02 2.41351724e-01 5.15197933e-01
-9.80059206e-01 6.75237656e-01 -4.11333501e-01 -7.91100502e-01
-1.66689634e+00 7.88504839e-01 2.12809831e-01 -2.57229984e-01
-7.01522350e-01 1.00044203e+00 6.35313511e-01 1.07438289e-01
3.53266776e-01 -2.10648313e-01 -8.21072400e-01 -3.31052728e-02
4.58095014e-01 2.53650188e-01 -7.29910880e-02 -9.05765891e-01
-4.33888942e-01 9.96040225e-01 5.14646880e-02 2.23424107e-01
1.27781844e+00 -2.48448715e-01 -4.53984648e-01 3.95444036e-01
1.17091739e+00 -2.89008915e-01 -8.29463601e-01 -9.29081719e-03
2.24291608e-01 -3.72719973e-01 -1.11848660e-01 -6.94311261e-01
-1.04818654e+00 1.34338140e+00 3.91017139e-01 9.56173360e-01
1.37192559e+00 4.85086024e-01 5.55999339e-01 2.60446370e-01
2.79539555e-01 -6.57957256e-01 3.93416196e-01 4.84810650e-01
1.02434051e+00 -1.36799514e+00 2.34818786e-01 -8.43229532e-01
-7.53987134e-01 1.42392194e+00 4.51400399e-01 -1.19493626e-01
3.23699772e-01 3.96645218e-02 1.84901446e-01 -6.03451133e-01
-2.65478253e-01 -7.19435990e-01 5.58309972e-01 9.29087102e-01
6.36155903e-01 -6.51606321e-02 -7.16369748e-02 -3.98480892e-01
-2.20260233e-01 -3.32884520e-01 4.19400126e-01 5.85179210e-01
-7.30802000e-01 -1.36743319e+00 -4.35107172e-01 3.47340554e-01
-3.44518721e-01 -1.38491793e-02 -8.32798064e-01 1.15285516e+00
1.57900840e-01 1.05331397e+00 2.31059358e-01 -8.68043303e-02
1.97978631e-01 -2.59185582e-01 7.53327429e-01 -7.95247138e-01
-3.51693332e-01 -2.69481778e-01 1.37416020e-01 -5.48501074e-01
-8.88626933e-01 -3.28907609e-01 -1.21249425e+00 -9.68545824e-02
1.41355479e-02 -7.22449422e-02 5.80174923e-01 9.55240786e-01
-1.72448426e-01 5.91751575e-01 1.69411346e-01 -8.65494668e-01
4.81973231e-01 -5.47791362e-01 -4.76042777e-01 5.30830264e-01
9.20313001e-02 -5.40393949e-01 3.16840112e-02 1.78186953e-01] | [10.287991523742676, 1.56173574924469] |
5113706a-2b61-4563-8888-7300d706dba1 | raw-nav-merge-seismic-data-to-subsurface | null | null | http://proceedings.neurips.cc/paper/2021/hash/498f2c21688f6451d9f5fd09d53edda7-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/498f2c21688f6451d9f5fd09d53edda7-Paper.pdf | Raw Nav-merge Seismic Data to Subsurface Properties with MLP based Multi-Modal Information Unscrambler | Traditional seismic inversion (SI) maps the hundreds of terabytes of raw-field data to subsurface properties in gigabytes. This inversion process is expensive, requiring over a year of human and computational effort. Recently, data-driven approaches equipped with Deep learning (DL) are envisioned to improve SI efficiency. However, these improvements are restricted to data with highly reduced scale and complexity. To extend these approaches to real-scale seismic data, researchers need to process raw nav-merge seismic data into an image and perform convolution. We argue that this convolution-based way of SI is not only computationally expensive but also conceptually problematic. Seismic data is not naturally an image and need not be processed as images. In this work, we go beyond convolution and propose a novel SI method. We solve the scalability of SI by proposing a new auxiliary learning paradigm for SI (Aux-SI). This paradigm breaks the SI into local inversion tasks, which predicts each small chunk of subsurface properties using surrounding seismic data. Aux-SI combines these local predictions to obtain the entire subsurface model. However, even this local inversion is still challenging due to: (1) high-dimensional, spatially irregular multi-modal seismic data, (2) there is no concrete spatial mapping (or alignment) between subsurface properties and raw data. To handle these challenges, we propose an all-MLP architecture, Multi-Modal Information Unscrambler (MMI-Unscrambler), that unscrambles seismic information by ingesting all available multi-modal data. The experiment shows that MMI-Unscrambler outperforms both SOTA U-Net and Transformer models on simulation data. We also scale MMI-Unscrambler to raw-field nav-merge data on Gulf-of-Mexico to obtain a geologically sound velocity model with an SSIM score of 0.8. To the best of our knowledge, this is the first successful demonstration of the DL approach on SI for real, large-scale, and complicated raw field data. | ['Anshumali Shrivastava', 'Antoine Vial-Aussavy', 'Anu Chandran', 'Menal Gupta', 'Zhaozhuo Xu', 'Aditya Desai'] | 2021-12-01 | null | https://openreview.net/forum?id=HLalhDvDwrQ | https://openreview.net/pdf?id=HLalhDvDwrQ | neurips-2021-12 | ['auxiliary-learning', 'seismic-inversion'] | ['methodology', 'miscellaneous'] | [ 2.34330148e-01 -1.27987072e-01 3.52062017e-01 -2.30442628e-01
-1.19026387e+00 -1.43264920e-01 5.10076821e-01 -9.31442901e-02
-6.58375263e-01 7.65523195e-01 2.15563238e-01 -6.95329487e-01
-1.99606240e-01 -1.20880115e+00 -1.09714973e+00 -9.28580463e-01
-4.05301362e-01 8.06216419e-01 6.13454878e-01 -6.40575528e-01
3.85010242e-01 6.22936547e-01 -1.50070035e+00 5.08940697e-01
9.22008514e-01 1.18676710e+00 4.77984190e-01 6.65360153e-01
-3.86117771e-02 7.13880181e-01 -3.66205662e-01 1.92777380e-01
1.60928980e-01 -8.65813792e-02 -9.58901048e-01 -4.93674219e-01
3.00786257e-01 -5.70646942e-01 -9.56762433e-02 7.40110695e-01
5.19777596e-01 1.19601905e-01 5.52535355e-01 -1.19003773e+00
-1.02477126e-01 6.53916717e-01 -6.81565344e-01 1.54088706e-01
7.50374198e-02 -1.55934244e-01 6.56542242e-01 -1.15234447e+00
2.71430254e-01 1.07094991e+00 1.17627180e+00 -1.11372568e-01
-1.16370618e+00 -6.73907697e-01 -1.32488906e-01 1.34237289e-01
-1.40893376e+00 -2.84066737e-01 7.29169607e-01 -4.94013906e-01
1.27563357e+00 2.68035382e-01 6.32510364e-01 7.57762074e-01
2.31307089e-01 6.22398496e-01 9.77756798e-01 -3.89023691e-01
2.23581165e-01 -4.32324886e-01 -8.47416818e-02 3.72480214e-01
5.52219674e-02 1.43216342e-01 -7.72393525e-01 4.97333845e-03
9.16839719e-01 -4.38701791e-05 -1.81318536e-01 1.29443735e-01
-1.38699317e+00 7.61803746e-01 4.80329007e-01 2.58792877e-01
-4.74766225e-01 3.21981758e-01 3.98985773e-01 3.78625244e-01
5.97805500e-01 3.49495202e-01 -5.58767855e-01 -2.51463324e-01
-1.30307591e+00 3.35350454e-01 7.51184464e-01 5.19820571e-01
1.21667123e+00 4.72570270e-01 5.25505424e-01 6.16692066e-01
2.05388382e-01 1.09297860e+00 3.85658145e-01 -9.02406573e-01
7.59374559e-01 1.00133136e-01 8.63285214e-02 -1.19411612e+00
-6.74114347e-01 -1.64480090e-01 -1.27036715e+00 3.90429050e-01
3.00558567e-01 -2.34924495e-01 -8.25620651e-01 1.40112543e+00
-3.79382968e-02 4.81261462e-01 1.57570809e-01 8.18534255e-01
6.66864932e-01 1.00305486e+00 -1.66163206e-01 4.40513432e-01
1.07556105e+00 -7.12297857e-01 -4.09303546e-01 -5.70943594e-01
7.36917555e-01 -4.87669051e-01 1.03516519e+00 3.26661468e-01
-8.87826204e-01 -5.60101449e-01 -1.07838786e+00 1.64699435e-01
-4.91224140e-01 -1.11186422e-01 7.45977223e-01 1.62235320e-01
-1.27312636e+00 6.48678005e-01 -1.30823040e+00 1.03611037e-01
1.07320309e-01 3.83657277e-01 -5.76533318e-01 1.93557993e-01
-1.55133295e+00 7.46744514e-01 3.68781090e-01 5.36171079e-01
-8.72034132e-01 -8.95517707e-01 -1.11556470e+00 2.69156873e-01
1.75126940e-01 -3.73310477e-01 9.82408345e-01 -6.75275385e-01
-1.30899858e+00 2.14552522e-01 -2.10129842e-01 -3.86430472e-01
3.15730453e-01 -1.79950401e-01 -2.79440969e-01 1.59864891e-02
3.20763350e-01 3.77217859e-01 5.76675236e-01 -1.40965331e+00
-5.62709510e-01 -1.94402784e-01 -1.54765353e-01 -1.55598283e-01
-2.67135769e-01 -2.62996525e-01 -2.21906692e-01 -6.52225912e-01
5.86797655e-01 -6.85166538e-01 -3.66244465e-01 -3.97007585e-01
-2.45633423e-01 3.23892325e-01 7.90808737e-01 -9.18966353e-01
1.04223168e+00 -2.05491209e+00 -8.75445157e-02 4.36243594e-01
3.28068808e-02 2.24224970e-01 -1.78644463e-01 5.42553306e-01
-9.65782776e-02 8.90620425e-02 -7.72004426e-01 -4.33580577e-01
-2.00440675e-01 3.51726979e-01 -5.05731225e-01 2.96974748e-01
1.21209219e-01 7.72817373e-01 -6.73998475e-01 -4.52458948e-01
1.69474825e-01 5.24914980e-01 -7.96373606e-01 2.24419624e-01
1.38780445e-01 5.96820891e-01 -4.03098375e-01 7.41844475e-01
1.23661840e+00 -1.17791012e-01 -1.41964510e-01 -3.97856146e-01
-4.70270097e-01 1.48733675e-01 -1.33972049e+00 1.54002035e+00
-7.84229577e-01 4.50706393e-01 2.58030862e-01 -1.55058467e+00
1.00057113e+00 2.36046001e-01 5.66518962e-01 -1.05465543e+00
-3.33652914e-01 7.17039466e-01 -1.66158870e-01 -6.03846490e-01
6.41496360e-01 -3.95312876e-01 -2.44488567e-01 5.35926223e-01
-4.48165871e-02 -4.54700559e-01 -2.77768821e-02 7.87798315e-02
9.24724817e-01 2.94072181e-01 -2.74836570e-01 -3.89756560e-01
4.63796496e-01 1.01736754e-01 5.55208206e-01 9.49039280e-01
3.89328569e-01 1.00198710e+00 3.23528737e-01 -5.27223825e-01
-1.31456184e+00 -1.03549659e+00 -1.40947014e-01 9.79508817e-01
2.81110145e-02 8.28676894e-02 -5.00009000e-01 -7.72974640e-02
-4.66071256e-02 2.41375074e-01 -5.19821286e-01 1.77209139e-01
-9.19175804e-01 -1.05356491e+00 8.92773211e-01 8.56377900e-01
7.38910317e-01 -1.13483894e+00 -5.52075624e-01 6.10466659e-01
-4.94264632e-01 -9.90598202e-01 8.14467967e-02 4.02635336e-01
-9.20316756e-01 -4.18305218e-01 -5.58728874e-01 -6.21094882e-01
3.58321577e-01 -1.91607457e-02 1.12288964e+00 4.93217334e-02
1.54212475e-01 -2.62136549e-01 -2.94098377e-01 -2.33983204e-01
-2.67857075e-01 1.34960994e-01 -8.62830654e-02 1.89768313e-03
1.91045795e-02 -9.68860328e-01 -4.29156750e-01 2.52000660e-01
-1.14558864e+00 2.67703176e-01 6.56808674e-01 1.16118753e+00
5.03362596e-01 -8.31620768e-02 6.64133430e-01 -5.16594946e-01
2.11943075e-01 -6.46588624e-01 -6.81335390e-01 8.10676441e-02
-1.53867975e-01 1.61261380e-01 5.75630486e-01 -2.11289972e-01
-1.29657853e+00 -1.03694886e-01 -9.14196014e-01 -8.37856845e-04
-7.21616298e-02 1.23620772e+00 1.14733674e-01 -2.07525179e-01
5.74419439e-01 4.17953670e-01 6.96216598e-02 -6.23527110e-01
-2.07855552e-01 9.17459309e-01 8.47207785e-01 -7.48627007e-01
6.28365636e-01 8.56486917e-01 1.44049913e-01 -1.02974987e+00
-6.95465028e-01 -3.09055239e-01 -7.91438103e-01 -1.35934949e-01
6.67268634e-01 -9.45101798e-01 -6.48626387e-01 9.20281827e-01
-1.08848679e+00 -7.99327374e-01 5.12304939e-02 6.56208694e-01
-5.23658633e-01 5.12816846e-01 -7.89956629e-01 -5.89459062e-01
-3.52830023e-01 -1.29734635e+00 1.32596982e+00 -1.82183161e-01
1.43398836e-01 -1.00008047e+00 1.08612634e-01 1.87361181e-01
8.51688087e-01 1.98482648e-01 4.99466151e-01 -2.78683305e-01
-5.80158830e-01 -2.81789631e-01 -5.00412524e-01 2.37232044e-01
-3.94199416e-02 -2.43138477e-01 -1.04134786e+00 -1.76879048e-01
7.27128536e-02 -4.20446426e-01 1.02072453e+00 5.26654899e-01
1.19374979e+00 7.78506044e-03 -6.64126202e-02 1.06607628e+00
1.47968256e+00 1.58746049e-01 9.66315448e-01 9.04451787e-01
7.95587242e-01 3.65282714e-01 7.01505125e-01 6.36947870e-01
5.37947893e-01 3.88020784e-01 6.06073976e-01 -6.51323617e-01
1.18363224e-01 -7.88129047e-02 1.29560798e-01 1.09005499e+00
-4.59540069e-01 -1.59756109e-01 -1.45005929e+00 7.28885472e-01
-1.84154081e+00 -9.55050528e-01 -2.21990049e-01 1.95284081e+00
5.73642015e-01 1.06890120e-01 -4.76792991e-01 3.10954124e-01
1.84092909e-01 2.07060799e-01 -2.26088136e-01 -2.04871923e-01
-3.36638808e-01 4.19404387e-01 8.22297096e-01 7.00851500e-01
-1.21108425e+00 8.70578170e-01 5.69915104e+00 7.10937798e-01
-1.52989626e+00 1.79894239e-01 5.03222525e-01 6.14095509e-01
-4.45005774e-01 -7.72901252e-02 -7.28287876e-01 4.68478978e-01
1.17688942e+00 4.75214869e-01 2.28657946e-01 4.85268474e-01
3.26282293e-01 -3.97913605e-01 -7.43545055e-01 6.73924744e-01
-9.99305621e-02 -1.81565547e+00 1.99557897e-02 -1.07945591e-01
5.87007046e-01 4.93813992e-01 -1.52832285e-01 2.25883350e-01
2.37037644e-01 -9.93044972e-01 8.54303896e-01 7.96177924e-01
8.22149515e-01 -6.27410948e-01 1.22046006e+00 3.92914176e-01
-1.40728474e+00 -3.69168594e-02 -1.61208287e-01 -2.95540124e-01
5.74376404e-01 6.52036965e-01 -4.98580217e-01 7.48881817e-01
1.10423946e+00 7.26024032e-01 -2.16515169e-01 8.35788131e-01
1.60323277e-01 7.20097840e-01 -7.52368212e-01 4.50505614e-01
6.61495149e-01 -1.74305707e-01 1.04955971e-01 1.26757038e+00
8.94839227e-01 7.56403133e-02 1.94892675e-01 6.07112527e-01
4.46327418e-01 -2.79656619e-01 -5.77168643e-01 4.01286691e-01
4.05619383e-01 8.03766310e-01 -5.22896051e-01 -4.80025977e-01
-5.75291872e-01 5.64688742e-01 1.80926993e-02 5.18951356e-01
-7.10199416e-01 -4.50353682e-01 4.51417267e-01 2.69417286e-01
3.07646006e-01 -5.51405549e-01 -5.15608668e-01 -9.91071045e-01
-1.46951869e-01 -7.50646114e-01 1.76440090e-01 -8.84894729e-01
-1.05007887e+00 7.18244076e-01 2.01854393e-01 -1.49042773e+00
-3.92705381e-01 -6.14189506e-01 -4.98677522e-01 1.03696465e+00
-1.88085830e+00 -1.34709644e+00 -5.82264245e-01 5.85988462e-01
2.55898654e-01 -1.03163905e-01 6.92445338e-01 6.96153641e-01
-1.61353812e-01 2.69007236e-01 3.79301637e-01 3.12458456e-01
5.53009629e-01 -1.00002789e+00 7.28769541e-01 7.64616966e-01
-3.95438462e-01 2.33587861e-01 6.68303490e-01 -6.39797390e-01
-1.37192535e+00 -8.92149448e-01 8.78087342e-01 6.26293570e-02
1.04937744e+00 -1.44296333e-01 -1.46385205e+00 8.20948839e-01
-2.21426353e-01 2.21417621e-01 2.80927628e-01 -1.83493182e-01
-7.48366788e-02 -3.74863416e-01 -9.41142380e-01 3.84169787e-01
5.99965990e-01 -5.93649745e-01 -5.18969893e-01 1.14833318e-01
5.48502028e-01 -4.07774329e-01 -1.09801364e+00 7.74197102e-01
5.15189826e-01 -1.07714224e+00 1.23291743e+00 4.46109697e-02
6.17989004e-01 -4.03553873e-01 -4.42683399e-01 -1.26796770e+00
4.50688601e-03 -1.22101054e-01 2.83807546e-01 6.63920701e-01
5.29529870e-01 -1.01175916e+00 8.15432966e-01 2.26597652e-01
-6.88589871e-01 -4.43383217e-01 -1.15596902e+00 -7.17167556e-01
3.66010249e-01 -8.23708057e-01 8.87933671e-01 8.60940218e-01
-1.19008556e-01 -7.15524629e-02 -4.60853249e-01 6.02646947e-01
7.51950741e-01 1.62937805e-01 6.52000904e-01 -1.16262126e+00
-2.37526774e-01 -1.71727642e-01 -2.08499774e-01 -1.14419591e+00
-3.64473425e-02 -4.98729497e-01 4.03548181e-01 -1.48737180e+00
-1.37871489e-01 -1.03837812e+00 -1.91834688e-01 7.14217126e-01
1.55619055e-01 5.10296464e-01 -1.87332854e-01 2.01820314e-01
1.02090620e-01 5.23028612e-01 1.14688814e+00 -2.69834250e-01
2.92894198e-03 -2.15696141e-01 1.85975776e-04 7.32697487e-01
8.07701230e-01 -3.78140986e-01 -8.25033039e-02 -9.32538569e-01
4.18374211e-01 5.99160731e-01 4.42079812e-01 -1.31692421e+00
4.23568696e-01 -2.12319687e-01 1.53402926e-03 -1.00438774e+00
4.99799401e-01 -6.99920297e-01 3.65121007e-01 4.00066257e-01
1.85516119e-01 1.48335040e-01 2.64412701e-01 2.04394571e-02
-8.45403314e-01 -1.96478367e-01 6.53472364e-01 -1.52779847e-01
-9.36146259e-01 1.93054870e-01 -4.21111912e-01 -2.64829755e-01
4.39802080e-01 -2.94914424e-01 -8.08668509e-02 -3.52317482e-01
-6.77445769e-01 8.75785723e-02 3.01191881e-02 -2.47902602e-01
6.95726335e-01 -9.95697320e-01 -9.57999349e-01 7.26094842e-01
-1.42129034e-01 5.41917562e-01 6.36252344e-01 1.02983057e+00
-1.13746285e+00 2.14322791e-01 -2.49912456e-01 -9.35748458e-01
-7.04579890e-01 9.93454456e-02 3.71525228e-01 -3.87289077e-01
-8.29263210e-01 9.40636992e-01 1.45653740e-01 -8.75926018e-01
-2.38891214e-01 -5.66106260e-01 -9.12209600e-02 1.45817012e-01
5.34931362e-01 2.36637443e-01 5.23049414e-01 -7.27174044e-01
-1.18780598e-01 7.36351013e-01 4.25328910e-01 -3.00345838e-01
1.83400548e+00 1.33785317e-02 -2.98895955e-01 3.20556849e-01
1.27447498e+00 -2.45604843e-01 -1.30639827e+00 -1.83401257e-01
-4.55657244e-02 -3.85411054e-01 2.51190037e-01 -3.82477105e-01
-1.29122901e+00 1.21670008e+00 3.64232332e-01 2.48793773e-02
1.14844799e+00 -1.39589235e-01 9.97521043e-01 6.56763315e-01
5.02924562e-01 -9.37197208e-01 7.96281472e-02 1.05824244e+00
9.83200490e-01 -1.28195727e+00 -1.18030496e-01 -1.16577193e-01
-4.59155530e-01 1.05270302e+00 4.43218082e-01 -1.89506821e-02
1.12573886e+00 1.01213205e+00 1.29292473e-01 -1.90898865e-01
-3.04070741e-01 5.42064756e-02 -2.54846662e-01 4.62895364e-01
2.01202303e-01 -2.20389031e-02 1.81963071e-01 4.76848185e-01
-3.47033978e-01 2.04465330e-01 4.69815671e-01 1.23592424e+00
-5.21785796e-01 -8.17614377e-01 -8.10032189e-01 4.12197351e-01
-1.27068028e-01 -4.02902871e-01 7.32962549e-01 6.38228416e-01
-3.12480796e-02 5.61653018e-01 5.04875720e-01 -3.91138226e-01
2.58743793e-01 -3.21515948e-01 -1.72597885e-01 -2.24800780e-01
-2.85427183e-01 2.13911176e-01 5.79941384e-02 -4.15556163e-01
-4.41717118e-01 -5.52463651e-01 -1.50729060e+00 -5.80917299e-01
-6.39260858e-02 2.83302248e-01 7.56592512e-01 1.07449841e+00
6.26718253e-02 6.15514994e-01 4.96205062e-01 -1.79361534e+00
-2.58746743e-01 -1.18341827e+00 -7.15943992e-01 1.05254717e-01
6.75061464e-01 -7.86323071e-01 -5.72297156e-01 4.78385314e-02] | [6.857578754425049, 2.5348434448242188] |
1261a96a-5ced-46da-b9f3-2a4372ba0c47 | a-stronger-baseline-for-multilingual-word | 1811.00586 | null | https://arxiv.org/abs/1811.00586v2 | https://arxiv.org/pdf/1811.00586v2.pdf | Multilingual Embeddings Jointly Induced from Contexts and Concepts: Simple, Strong and Scalable | Word embeddings induced from local context are prevalent in NLP. A simple and effective context-based multilingual embedding learner is Levy et al. (2017)'s S-ID (sentence ID) method. Another line of work induces high-performing multilingual embeddings from concepts (Dufter et al., 2018). In this paper, we propose Co+Co, a simple and scalable method that combines context-based and concept-based learning. From a sentence aligned corpus, concepts are extracted via sampling; words are then associated with their concept ID and sentence ID in embedding learning. This is the first work that successfully combines context-based and concept-based embedding learning. We show that Co+Co performs well for two different application scenarios: the Parallel Bible Corpus (1000+ languages, low-resource) and EuroParl (12 languages, high-resource). Among methods applicable to both corpora, Co+Co performs best in our evaluation setup of six tasks. | ['Hinrich Schütze', 'Philipp Dufter', 'Mengjie Zhao'] | 2018-11-01 | null | null | null | null | ['multilingual-word-embeddings'] | ['methodology'] | [-2.94389844e-01 -1.94199905e-01 -4.16980058e-01 -1.94447517e-01
-9.88806486e-01 -7.57998943e-01 1.08269536e+00 8.63822579e-01
-1.13616478e+00 8.75063300e-01 5.77459157e-01 -3.61246198e-01
2.24768505e-01 -6.25940800e-01 -4.74275678e-01 -2.50820369e-01
-2.63837159e-01 6.99569166e-01 4.29046601e-02 -4.11479652e-01
2.05958977e-01 2.14942649e-01 -1.21176863e+00 1.74512237e-01
9.98920679e-01 1.56522945e-01 5.28777599e-01 7.43357837e-01
-5.36358237e-01 3.67761075e-01 -4.26606596e-01 -3.99622500e-01
4.52733338e-02 -1.80515483e-01 -8.26513827e-01 -5.86438179e-01
4.96088415e-01 1.65229440e-01 7.61191025e-02 7.71744549e-01
7.90257931e-01 -1.47140650e-02 8.05429399e-01 -8.34844828e-01
-9.57099319e-01 8.45080853e-01 -3.63918215e-01 5.08493900e-01
5.12994528e-01 -2.18918756e-01 1.61471486e+00 -1.52054048e+00
9.18926597e-01 1.25658751e+00 6.85815990e-01 4.16683108e-01
-1.40811443e+00 -4.71317142e-01 1.83509097e-01 3.94612819e-01
-1.26282835e+00 -1.03633530e-01 6.72380090e-01 -4.40726250e-01
1.34267354e+00 2.68975878e-03 6.90419137e-01 1.35678470e+00
2.11726725e-01 8.75411987e-01 1.34573925e+00 -1.12955904e+00
2.02983052e-01 4.65597868e-01 3.70285332e-01 2.78866947e-01
2.92651534e-01 -9.27540287e-03 -6.89585507e-01 -2.77375996e-01
9.76888090e-02 -9.89158228e-02 -6.67403862e-02 -2.28666872e-01
-1.53899693e+00 1.13957965e+00 8.72757360e-02 9.41833556e-01
-2.31685683e-01 -3.27898897e-02 6.68465972e-01 4.65129316e-01
9.49172497e-01 7.43426263e-01 -7.81777620e-01 -3.72196347e-01
-9.49715853e-01 4.34883893e-01 8.93856287e-01 9.90150154e-01
8.16931903e-01 -1.74452990e-01 -2.55225211e-01 9.65918958e-01
2.80298740e-01 5.29729545e-01 8.78213942e-01 -2.14727134e-01
4.35940862e-01 1.45981938e-01 -1.22906007e-01 -7.26329207e-01
-3.03950965e-01 -2.90656805e-01 -1.77498251e-01 -3.68057013e-01
2.13157997e-01 -2.14163393e-01 -3.91214460e-01 1.70482302e+00
2.94050127e-01 2.41753384e-01 5.39179206e-01 5.96639395e-01
7.86895394e-01 8.30592930e-01 2.10038871e-01 -2.04552412e-02
1.75723743e+00 -1.06784463e+00 -7.04919279e-01 -3.45345676e-01
9.53833222e-01 -9.75781262e-01 1.27010024e+00 1.14585310e-01
-6.00037634e-01 -4.68726277e-01 -1.18769121e+00 -3.15474927e-01
-8.34083557e-01 1.77206218e-01 7.03461289e-01 4.19651628e-01
-1.06917334e+00 2.48507038e-01 -5.35431802e-01 -9.05571163e-01
-1.06774054e-01 -1.42666593e-01 -5.25164068e-01 -4.90613073e-01
-1.65113342e+00 1.27369356e+00 6.04703665e-01 -5.62428951e-01
-6.03218138e-01 -8.42003644e-01 -1.11046302e+00 -2.11785361e-01
-5.22087999e-02 -3.76617432e-01 8.03303063e-01 -7.28565812e-01
-1.12687826e+00 1.03399253e+00 -1.66529581e-01 -6.86648309e-01
-5.01971133e-02 -5.10513663e-01 -5.26705801e-01 8.73814970e-02
3.75643551e-01 7.64964759e-01 5.52623928e-01 -1.11535788e+00
-5.14085233e-01 -2.37315729e-01 1.15076408e-01 3.04577678e-01
-9.49960411e-01 3.34689230e-01 -6.16155639e-02 -6.62213385e-01
-3.00562471e-01 -6.90355062e-01 -2.80268900e-02 -2.70336956e-01
2.13283733e-01 -7.53701627e-01 5.04043400e-01 -7.91502059e-01
1.31203616e+00 -1.96119761e+00 2.25621581e-01 -3.96479517e-01
-1.19213864e-01 3.73279959e-01 -5.24211824e-01 1.19897151e+00
8.78130049e-02 2.07073629e-01 -3.28002840e-01 -5.20859897e-01
2.45717168e-01 5.47559738e-01 -2.28424415e-01 3.02707314e-01
5.23965478e-01 9.25815582e-01 -1.48183334e+00 -7.54938841e-01
1.20495498e-01 6.42929614e-01 -5.95552504e-01 9.53541473e-02
-1.49791330e-01 2.15680927e-01 2.79409643e-02 4.46151316e-01
3.24016809e-01 4.47561085e-01 5.89521170e-01 7.72984885e-03
-4.13181394e-01 4.92656857e-01 -7.75873184e-01 2.17736506e+00
-1.17687571e+00 9.51215208e-01 -2.89011985e-01 -9.17736530e-01
9.70906496e-01 6.13324165e-01 2.70178556e-01 -4.46133971e-01
-2.18995512e-01 5.05996883e-01 5.28585212e-03 -6.44252002e-01
8.75100195e-01 -2.68700093e-01 -2.95769006e-01 5.74978769e-01
7.10025191e-01 -1.54108733e-01 5.41978776e-01 3.13400686e-01
9.11052227e-01 2.73353338e-01 7.80536711e-01 -8.41323853e-01
6.17302001e-01 -1.30980194e-01 3.77931058e-01 4.41504121e-01
-3.45289826e-01 4.21510398e-01 3.50519210e-01 -3.63240063e-01
-1.21081877e+00 -1.09712100e+00 -7.12159574e-01 1.30187666e+00
-4.04086620e-01 -8.65353405e-01 -3.60677958e-01 -8.18749368e-01
1.25893310e-01 7.93457091e-01 -6.76655114e-01 3.00164342e-01
-6.04110062e-01 -6.25604093e-01 4.31654751e-01 4.90664929e-01
-2.13680685e-01 -1.13961852e+00 -9.86312032e-02 5.61853170e-01
-2.56977081e-01 -1.05982554e+00 -4.90254313e-01 2.30881304e-01
-7.16123223e-01 -6.86666846e-01 -7.21928239e-01 -1.06635404e+00
4.15994078e-01 -1.67229950e-01 1.46303892e+00 -4.03642803e-01
-2.41339013e-01 4.52613533e-01 -7.58986115e-01 -5.96945047e-01
-2.18474776e-01 2.75202453e-01 3.16606283e-01 -2.30661958e-01
9.53902364e-01 -4.91470098e-01 -1.83600038e-01 -4.49015260e-01
-8.45727861e-01 -5.52202940e-01 5.36221325e-01 1.13679075e+00
4.43643719e-01 -6.99827254e-01 6.86269403e-01 -7.89075613e-01
8.73413086e-01 -7.33051836e-01 -2.24320337e-01 1.78613096e-01
-6.46383643e-01 1.96012408e-01 6.22396946e-01 -4.87379909e-01
-4.81081009e-01 -4.34004277e-01 -3.78090173e-01 8.05537626e-02
-1.86394811e-01 7.39654779e-01 1.40571594e-01 3.81569415e-01
7.12281644e-01 3.04182857e-01 -3.34325910e-01 -6.61343694e-01
7.09299922e-01 9.77712691e-01 9.38089341e-02 -9.42680597e-01
6.08580709e-01 2.56695360e-01 -4.80177701e-01 -1.18041527e+00
-7.91938484e-01 -7.79382169e-01 -9.31114018e-01 1.06472701e-01
9.98246014e-01 -1.14162290e+00 2.66669571e-01 -1.54838473e-01
-1.43100214e+00 3.15007195e-02 -4.29625392e-01 9.86165285e-01
-2.14187160e-01 3.56394589e-01 -5.94992697e-01 -6.52774036e-01
-4.25000608e-01 -7.76975155e-01 1.11301959e+00 -1.40707970e-01
-4.03909534e-01 -1.60455751e+00 9.04179633e-01 -1.74566552e-01
3.53689313e-01 2.24074975e-01 8.34892333e-01 -1.04934263e+00
1.16073243e-01 -1.45989969e-01 5.90881780e-02 5.38924694e-01
1.42473221e-01 -1.29916102e-01 -9.29430723e-01 -6.70155644e-01
-3.53195488e-01 -5.21969676e-01 7.85280049e-01 -2.47233436e-02
4.55832839e-01 -2.20039174e-01 -1.62727281e-01 2.53942639e-01
1.81365025e+00 -2.41747394e-01 3.35421383e-01 6.26077414e-01
5.18960059e-01 5.75468123e-01 6.95093095e-01 9.87466276e-02
6.27048433e-01 6.59669816e-01 -1.64403692e-01 1.77615121e-01
-1.09245837e-01 -3.02828938e-01 8.63432765e-01 1.90347219e+00
1.88402727e-01 3.02267913e-02 -1.14605832e+00 1.17493200e+00
-1.61431181e+00 -8.67614150e-01 -8.31960738e-02 1.97740602e+00
1.31320095e+00 -2.89285723e-02 2.86323503e-02 -1.55501068e-01
4.68486965e-01 2.30991304e-01 2.80202478e-01 -9.33698893e-01
-1.34900391e-01 7.26194322e-01 1.22529566e-01 7.07594931e-01
-1.14065647e+00 1.13329983e+00 5.66181374e+00 8.20372701e-01
-8.81039798e-01 7.87053823e-01 -9.12406743e-02 5.39342314e-02
-6.59633815e-01 3.05789322e-01 -9.77749169e-01 5.91380239e-01
1.32463002e+00 -3.56001824e-01 -8.09934214e-02 7.54736006e-01
-3.55233431e-01 1.64047942e-01 -1.19581866e+00 7.04984307e-01
6.08243763e-01 -1.08304322e+00 -1.17657088e-01 -1.28664225e-01
7.68768549e-01 4.80803877e-01 -1.55432478e-01 7.26777196e-01
2.79245347e-01 -8.19320202e-01 5.12019157e-01 1.61383063e-01
1.01859140e+00 -9.00406182e-01 1.00739110e+00 1.10055454e-01
-1.25158930e+00 1.57986149e-01 -6.21268690e-01 -4.23184149e-02
2.22177893e-01 7.49728084e-01 -6.87395334e-01 9.03914034e-01
5.20969152e-01 1.21246541e+00 -6.32053077e-01 4.77140129e-01
-6.32281363e-01 9.12383854e-01 -7.32445121e-02 -3.95554572e-01
5.51481307e-01 -3.03197265e-01 7.20371962e-01 1.92370975e+00
1.31721303e-01 -5.22500515e-01 3.76344770e-01 5.90854049e-01
-2.27667093e-02 6.86577857e-01 -8.77622962e-01 -2.77924180e-01
6.35953844e-01 1.45185030e+00 -3.05958778e-01 -3.27730566e-01
-7.34543443e-01 8.05972040e-01 8.06091249e-01 9.32135582e-02
-4.19163048e-01 -6.72809660e-01 8.00940275e-01 -2.71708757e-01
3.82348835e-01 -5.37265062e-01 5.09797223e-02 -1.40715909e+00
4.07061800e-02 -7.51013517e-01 2.86864758e-01 -2.43122175e-01
-1.65817225e+00 7.85269499e-01 5.37152812e-02 -1.24898219e+00
-2.40227163e-01 -8.57113659e-01 -4.67201322e-01 9.72270429e-01
-2.02314091e+00 -1.40145934e+00 3.32189053e-01 1.54890269e-01
6.92038596e-01 -3.73500228e-01 1.34755528e+00 4.90341187e-01
-1.62648201e-01 5.58972538e-01 3.89137268e-01 3.66235614e-01
1.14509845e+00 -1.78551269e+00 4.13024485e-01 7.11756110e-01
6.95891321e-01 8.22397768e-01 5.58384180e-01 -3.37909788e-01
-1.30096936e+00 -1.00403094e+00 1.89466429e+00 -7.68786609e-01
1.19073999e+00 -9.19739664e-01 -7.58708060e-01 6.45628691e-01
8.72926176e-01 1.68068632e-02 1.11148489e+00 5.43196321e-01
-5.97620070e-01 -1.81336313e-01 -8.17760468e-01 6.37766361e-01
6.03331625e-01 -8.95370305e-01 -1.17799950e+00 5.98764181e-01
9.00579989e-01 1.03112467e-01 -1.21168983e+00 7.02811182e-02
3.73347878e-01 -5.10220289e-01 8.39450002e-01 -6.83918297e-01
5.94566822e-01 -8.44201595e-02 -4.29290920e-01 -1.73662400e+00
-1.01096183e-01 -2.43614480e-01 -9.78666842e-02 1.46438050e+00
5.83973229e-01 -8.56946647e-01 -1.49409622e-01 -4.80167538e-01
-1.90864101e-01 -8.88449728e-01 -1.07960284e+00 -1.26704288e+00
1.00334024e+00 -3.13768893e-01 3.96731466e-01 1.36485660e+00
3.96105856e-01 5.80942869e-01 -4.31572162e-02 -1.69750199e-01
4.57152009e-01 -7.18791857e-02 6.03323579e-01 -1.05833280e+00
-3.34266663e-01 -1.64815381e-01 -5.76786101e-01 -6.48333311e-01
5.27828515e-01 -1.49189758e+00 -1.48762735e-02 -1.57525849e+00
2.17691556e-01 -4.62621838e-01 -5.12003899e-01 2.31289789e-01
-5.07732630e-01 5.48780784e-02 2.41495013e-01 -5.34344204e-02
-4.44464952e-01 6.04497552e-01 6.60597444e-01 -2.41643697e-01
2.60675132e-01 -8.57218266e-01 -3.09624285e-01 2.06052393e-01
8.17174554e-01 -6.84424639e-01 -1.01314031e-01 -5.83405316e-01
2.59617895e-01 -6.39924586e-01 -1.60623174e-02 -8.52946758e-01
-7.62501284e-02 1.48215026e-01 -6.39886037e-02 -4.58459854e-01
1.44915938e-01 -6.10156655e-01 -6.39986396e-01 4.45592701e-01
-3.66215587e-01 5.23678720e-01 2.21779019e-01 4.41020727e-01
-5.39792538e-01 -5.82249522e-01 5.27697325e-01 -1.04755886e-01
-7.97362089e-01 2.35890180e-01 -3.76821220e-01 6.55897856e-01
8.92827392e-01 2.53845572e-01 -2.52813958e-02 6.49308190e-02
-3.99284393e-01 2.40757063e-01 2.26534426e-01 7.43161917e-01
5.11069357e-01 -1.68547297e+00 -1.26860595e+00 9.17913467e-02
6.85140252e-01 -5.07056832e-01 -3.05090517e-01 8.26438904e-01
-4.73169237e-01 6.45615518e-01 1.49762277e-02 -5.69841921e-01
-9.46188271e-01 6.17722392e-01 -2.23005176e-01 -5.94359815e-01
-5.11244476e-01 8.23896170e-01 -3.02278399e-01 -1.06407964e+00
-2.13456750e-01 -1.37606397e-01 -3.93842846e-01 4.43631053e-01
6.22103572e-01 5.69339544e-02 -2.33701952e-02 -6.79337621e-01
-5.30981839e-01 4.68743026e-01 -3.06243628e-01 -6.57963932e-01
1.40778351e+00 9.95472521e-02 -3.74030799e-01 1.06494713e+00
1.54247594e+00 2.69226462e-01 -7.38602996e-01 -4.59015876e-01
5.49778461e-01 -3.01875889e-01 -1.79254144e-01 -4.76268530e-01
-3.37244034e-01 1.18907428e+00 4.47797358e-01 -9.70180109e-02
6.36361122e-01 3.74263078e-02 7.80863941e-01 3.02628815e-01
5.21449685e-01 -1.26822865e+00 4.89480980e-02 9.25056696e-01
8.84813845e-01 -1.24347103e+00 -5.59077226e-02 1.29839718e-01
-5.58079541e-01 1.12070370e+00 4.41307217e-01 -2.54644215e-01
7.54464388e-01 1.96612477e-01 2.02036858e-01 3.28592248e-02
-8.53661895e-01 -4.83538985e-01 3.28897595e-01 7.59404361e-01
9.94018853e-01 1.60937265e-01 -1.12746739e+00 4.94116187e-01
-2.72853613e-01 -4.86944109e-01 3.75620157e-01 1.13895869e+00
-3.07017595e-01 -1.69297671e+00 -1.13896579e-01 1.89925656e-01
-3.87064099e-01 -7.72006094e-01 -2.83775300e-01 8.75396371e-01
2.66778260e-01 8.04140270e-01 1.91921085e-01 -3.05043943e-02
-1.17319347e-02 5.65835476e-01 4.89929378e-01 -1.16694105e+00
-6.68714166e-01 -3.23874444e-01 2.78113812e-01 -2.02554509e-01
-5.89730322e-01 -9.40204442e-01 -9.86749053e-01 7.86991194e-02
-2.79744148e-01 3.87938231e-01 8.86130571e-01 9.58599389e-01
2.03095436e-01 3.07927996e-01 5.00440836e-01 -6.03582740e-01
-5.24387896e-01 -1.47372103e+00 -5.16466856e-01 5.14183521e-01
1.93367839e-01 -5.13965368e-01 -4.18102801e-01 -6.59355447e-02] | [10.880169868469238, 9.70950984954834] |
bd4336bf-25f9-4517-a7f7-79cee9680da6 | weakly-supervised-real-time-image-cropping | null | null | https://dl.acm.org/doi/abs/10.1145/3394171.3413824?casa_token=k8h5Xio4LfQAAAAA:pYe2xwRtzJp1gymh-FHvg8oxiYBsHaPwW4dyz98tlJXg0grhocgODGRAZ7o0HCSZ6awgOHVYFo3o | https://dl.acm.org/doi/pdf/10.1145/3394171.3413824 | Weakly Supervised Real-time Image Cropping based on Aesthetic Distributions | Image cropping is an effective tool to edit and manipulate images to achieve better aesthetic quality. Most existing cropping approaches rely on the two-step paradigm where multiple candidate cropping areas are proposed initially and the optimal cropping window is determined based on some quality criteria for these candidates af- terwards. The obvious disadvantage of this mechanism is its low efficiency due to the huge searching space of candidate crops. In order to tackle this problem, a weakly supervised cropping frame- work is proposed, where the distribution dissimilarity between high quality images and cropped images is used to guide the coordinate predictor’s training and the ground truths of cropping windows are not required by the proposed method. Meanwhile, to improve the cropping performance, a saliency loss is also designed in the proposed framework to force the neural network to focus more on the interested objects in the image. Under this framework, the im- ages can be cropped effectively by the trained coordinate predictor in a one-pass favor without multiple candidates proposals, which ensures the high efficiency of the proposed system . Also, based on the proposed framework, many existing distribution dissimi- larity measurements can be applied to train the image cropping system with high flexibility, such as likelihood based and diver- gence based distribution dissimilarity measure proposed in this work. The experiments on the public databases show that the pro- posed cropping method achieves the state-of-the-art accuracy, and the high computation efficiency as fast as 285 FPS is also obtained. | ['Xiaojie Wang', 'Xujun Peng', 'Jiahui Liu', 'Peng Lu'] | 2020-10-15 | null | null | null | null | ['image-cropping'] | ['computer-vision'] | [ 3.69446099e-01 -2.67300218e-01 -1.40416458e-01 -1.93382636e-01
-4.06827539e-01 -2.44388953e-01 1.83941782e-01 1.00518661e-02
-2.89435714e-01 5.37554085e-01 -3.72812897e-01 3.17995846e-02
-1.10311493e-01 -1.12751079e+00 -8.12570512e-01 -9.74689722e-01
3.68586928e-01 -1.21252581e-01 5.87706149e-01 -1.82668120e-01
5.37233591e-01 2.98027724e-01 -1.87922394e+00 -5.07714935e-02
1.45392311e+00 1.11894071e+00 1.01199293e+00 2.88768679e-01
-1.22789547e-01 2.29393914e-01 -4.37384844e-01 -2.55356908e-01
3.76300961e-01 -4.90840167e-01 -3.43994834e-02 1.37391835e-01
1.71623766e-01 -3.13830495e-01 8.57305750e-02 1.32088745e+00
4.76397187e-01 -5.03274202e-02 6.37186825e-01 -1.27096903e+00
-8.46652091e-01 4.11710232e-01 -1.11887956e+00 -1.54265821e-01
-4.74980250e-02 -6.82565849e-03 7.09638476e-01 -1.20063829e+00
3.76629502e-01 1.16008997e+00 3.04379702e-01 6.83814511e-02
-8.38359118e-01 -9.67270374e-01 2.29790673e-01 2.44972110e-01
-1.37110937e+00 -3.74250710e-02 9.10354137e-01 -1.21913128e-01
1.65993124e-01 1.66611195e-01 7.69118965e-01 2.67594129e-01
3.20992619e-01 8.54827285e-01 1.02196467e+00 -6.63072884e-01
9.49432477e-02 3.98452789e-01 -1.08611733e-01 7.40306497e-01
3.87930423e-01 9.93285999e-02 -3.99521798e-01 2.40862608e-01
8.47032130e-01 3.52517813e-02 -2.69616365e-01 -6.80024326e-01
-1.31209397e+00 5.08789480e-01 8.29033852e-01 2.60153472e-01
-4.32165176e-01 -1.78842828e-01 1.50348455e-01 -9.77342874e-02
3.42060089e-01 2.49337509e-01 -2.42039412e-01 3.03082913e-01
-1.14495933e+00 3.17341775e-01 1.74969479e-01 9.95821714e-01
8.70689213e-01 -7.16300234e-02 -1.41158760e-01 6.69211268e-01
3.52566868e-01 8.77494335e-01 3.05107623e-01 -6.13233328e-01
6.04226232e-01 6.24271452e-01 2.00253397e-01 -1.56818616e+00
-1.17209911e-01 -3.21946919e-01 -1.01880431e+00 5.19121468e-01
2.39196226e-01 -4.95704114e-02 -9.42419887e-01 1.44188058e+00
4.70146686e-01 7.12810904e-02 -6.36148527e-02 9.67780232e-01
5.30705035e-01 8.69426787e-01 2.43804678e-02 -4.91672605e-01
1.27582014e+00 -8.50944877e-01 -8.07171047e-01 1.58164445e-02
2.04659998e-01 -1.05226040e+00 1.10680878e+00 4.21293914e-01
-1.03405011e+00 -7.81047523e-01 -1.28764355e+00 2.44951457e-01
-3.75680417e-01 8.30264449e-01 4.04392004e-01 5.69814920e-01
-7.70522237e-01 3.35754573e-01 -5.08949637e-01 -2.99920797e-01
5.40744305e-01 1.13048010e-01 -1.47411053e-03 -2.12875441e-01
-1.09183478e+00 8.58000278e-01 7.43553221e-01 3.41848314e-01
-6.81340039e-01 -6.17969751e-01 -5.70584655e-01 1.90061957e-01
2.58363575e-01 -1.91385001e-01 5.97973526e-01 -1.25219834e+00
-1.21632814e+00 5.25251448e-01 -1.59392223e-01 -1.98546916e-01
4.69694078e-01 -5.36899008e-02 -2.11919829e-01 2.20263898e-01
1.87057287e-01 1.04781651e+00 9.79606867e-01 -1.29635191e+00
-1.11759317e+00 -2.00225502e-01 -2.04936750e-02 7.31973588e-01
-5.39040565e-01 -1.26298815e-01 -5.39757311e-01 -8.00461709e-01
2.79615134e-01 -8.85965586e-01 2.09612679e-02 3.36942106e-01
-2.43560836e-01 3.31449974e-03 1.13786304e+00 -7.48617709e-01
1.27174091e+00 -2.08822227e+00 9.35966820e-02 3.05329919e-01
-1.77440152e-01 3.93177450e-01 -7.48104677e-02 -5.48346452e-02
3.17348279e-02 -4.18268293e-02 -5.51299572e-01 2.80155718e-01
-3.70858014e-01 -1.64951429e-01 -8.01816434e-02 2.05705985e-01
5.04846752e-01 5.52943110e-01 -8.30285966e-01 -7.55752504e-01
6.32272601e-01 2.90055484e-01 -2.79126614e-01 2.48284072e-01
-1.83228359e-01 1.86830074e-01 -3.59333873e-01 8.02408040e-01
1.32233942e+00 8.30008015e-02 -1.51748374e-01 -4.97241288e-01
-2.81865448e-01 -5.32312572e-01 -1.28774619e+00 1.57263398e+00
-3.93256217e-01 2.90688932e-01 -1.02048486e-01 -1.07718110e+00
1.33454418e+00 -1.14154674e-01 3.54965121e-01 -6.81174278e-01
2.11723953e-01 2.69713551e-01 -1.49633601e-01 -5.13764501e-01
6.29841685e-01 2.86468923e-01 1.34437025e-01 1.97179168e-01
-2.92156309e-01 -2.74506360e-01 2.92761505e-01 -1.94828939e-02
2.74934024e-01 6.86219931e-01 1.75901160e-01 -3.22028905e-01
7.27889001e-01 2.29019284e-01 6.81448042e-01 3.26047540e-01
-1.39541999e-01 7.11780488e-01 1.61653981e-01 -1.71042293e-01
-1.15857792e+00 -8.58649552e-01 -2.00558841e-01 7.23631680e-01
9.12504911e-01 1.22010253e-01 -1.01972044e+00 -7.08545029e-01
-3.44626188e-01 7.38255382e-01 -4.99034762e-01 -1.45929053e-01
-4.25267190e-01 -9.60120618e-01 6.83758557e-02 2.58303851e-01
1.12955582e+00 -1.31762111e+00 -9.09901977e-01 9.23226699e-02
-4.33400691e-01 -6.18955135e-01 -4.28266257e-01 -2.89607421e-02
-6.69990778e-01 -1.03966820e+00 -1.28923428e+00 -1.09973407e+00
1.03968823e+00 7.92759597e-01 6.44219398e-01 2.40941912e-01
-2.27453947e-01 -2.91998863e-01 -6.04616225e-01 -5.59178531e-01
-2.07336783e-01 8.56886357e-02 -1.18845463e-01 5.70976036e-03
4.17219669e-01 -2.22663298e-01 -7.61288524e-01 4.84311551e-01
-1.03911793e+00 4.04571563e-01 1.08871973e+00 9.94737625e-01
7.77579427e-01 5.43593943e-01 7.33356178e-01 -3.28903764e-01
3.64281923e-01 -3.57457906e-01 -8.71035755e-01 5.95325530e-01
-7.80366719e-01 -6.29360154e-02 6.54277682e-01 -5.92745423e-01
-1.34340656e+00 2.81974226e-01 2.68325895e-01 -3.67366523e-01
1.92364678e-01 3.53386998e-01 -4.54063654e-01 -2.26224840e-01
3.81231368e-01 3.93508673e-01 1.22641899e-01 -4.20596153e-02
4.03261393e-01 6.00758135e-01 3.82961839e-01 -2.37634748e-01
1.03569996e+00 1.71161845e-01 2.51190234e-02 -4.69285578e-01
-4.10681397e-01 -2.26290580e-02 -6.66819990e-01 -4.69240189e-01
9.18703973e-01 -8.23026180e-01 -5.72263479e-01 6.72149062e-01
-1.02846801e+00 2.84359723e-01 1.60923660e-01 4.81914997e-01
-3.28894615e-01 4.33572322e-01 -6.30800650e-02 -1.01116836e+00
-3.26308578e-01 -1.29288816e+00 9.98088837e-01 8.69932473e-01
4.76969749e-01 -4.65844929e-01 -6.15446210e-01 -7.75603652e-02
4.38603193e-01 2.39502132e-01 9.49356377e-01 1.10851616e-01
-8.76253068e-01 -6.93107098e-02 -7.28266716e-01 3.51010323e-01
2.78205127e-01 2.16109321e-01 -7.03387558e-01 -1.13929316e-01
-2.74981320e-01 -1.86822653e-01 7.01506078e-01 6.39963329e-01
1.06109023e+00 8.97810459e-02 -4.52276677e-01 3.80402416e-01
1.65249538e+00 4.65767562e-01 7.54185259e-01 3.91650230e-01
5.28742731e-01 6.61559463e-01 1.70364296e+00 3.07829946e-01
1.62318692e-01 4.22677457e-01 6.56236470e-01 -4.44432944e-01
7.18783662e-02 -4.95462388e-01 3.77084941e-01 5.73408544e-01
1.92787901e-01 -2.18167946e-01 -4.36582237e-01 6.19647324e-01
-1.88779128e+00 -8.27707946e-01 5.41851856e-02 2.50389981e+00
7.18322575e-01 1.26635715e-01 -1.55379906e-01 2.31844410e-01
1.16443717e+00 1.56199813e-01 -7.71327674e-01 -5.08214012e-02
-2.78331190e-01 -1.99214175e-01 6.34936392e-01 1.42353684e-01
-1.10207713e+00 8.77438962e-01 5.43251419e+00 1.22358882e+00
-1.26566851e+00 -2.97375470e-01 8.97075951e-01 2.39516005e-01
-1.58163577e-01 2.24898886e-02 -8.10302138e-01 6.77162349e-01
1.60112277e-01 -2.68381685e-01 4.53247786e-01 8.97782505e-01
3.97088826e-01 -7.40179837e-01 -3.06140900e-01 1.00483799e+00
2.78566331e-01 -1.06731808e+00 -3.82560939e-02 -1.67157084e-01
6.39007151e-01 -5.46372056e-01 2.36317620e-01 -4.91617732e-02
-7.57145556e-03 -7.14500487e-01 6.56035602e-01 6.25451088e-01
8.95093083e-01 -1.03988707e+00 7.28366494e-01 4.35805917e-01
-1.44962454e+00 -1.77954867e-01 -8.06031704e-01 2.54963398e-01
1.08203962e-01 8.49517643e-01 -6.55566394e-01 8.23487103e-01
8.59207332e-01 6.43783271e-01 -5.89408100e-01 1.12800753e+00
-1.13347374e-01 3.53485525e-01 -3.78143877e-01 -2.65680730e-01
3.53961438e-02 -5.75633645e-01 3.85449111e-01 7.88396895e-01
8.72390449e-01 2.86113378e-02 1.30967125e-01 9.56542969e-01
2.12678418e-01 3.74201477e-01 -4.45720792e-01 2.41992727e-01
7.45823324e-01 1.51646090e+00 -1.02694380e+00 -3.03835571e-01
-1.24372467e-01 8.62764657e-01 -2.01333873e-02 8.08435902e-02
-1.10476828e+00 -8.38614523e-01 -7.59494901e-02 -1.67731624e-02
4.67160672e-01 7.31695741e-02 -3.02223146e-01 -9.48972464e-01
2.90859282e-01 -7.35655844e-01 -6.56309479e-04 -1.19943237e+00
-9.70953941e-01 4.80707884e-01 1.92020368e-02 -1.70697105e+00
1.52730346e-01 -3.16265225e-01 -5.63809872e-01 1.08179092e+00
-1.68543613e+00 -1.32703078e+00 -7.08339870e-01 2.78558552e-01
6.03487253e-01 -1.08167812e-01 5.16474366e-01 2.34528095e-01
-4.92785633e-01 4.89751846e-01 1.46868795e-01 -1.44630358e-01
7.76835680e-01 -9.88675535e-01 -3.75824496e-02 1.21739173e+00
-1.33014917e-01 2.03515425e-01 7.02283859e-01 -7.54791737e-01
-1.08991361e+00 -1.13423371e+00 5.88967264e-01 8.55941996e-02
6.15034029e-02 -9.81782302e-02 -8.87591541e-01 -1.09155014e-01
1.66053489e-01 -9.34069082e-02 -4.53459192e-03 -4.71557915e-01
2.94074893e-01 -4.10359770e-01 -1.32540667e+00 5.49987972e-01
7.08767831e-01 1.79234490e-01 -4.05968070e-01 1.70325771e-01
8.43546450e-01 -3.71029824e-01 -4.96214062e-01 4.94955599e-01
6.77921891e-01 -9.76453364e-01 7.76004791e-01 2.74966300e-01
5.90728521e-01 -7.59373546e-01 5.14647625e-02 -1.31165040e+00
-3.50165308e-01 -1.53782219e-01 3.50698978e-01 1.57344735e+00
2.95414895e-01 -2.91017801e-01 7.36918032e-01 3.11462700e-01
2.79704958e-01 -8.67567241e-01 -5.14867365e-01 -4.19682622e-01
-2.15727553e-01 3.41447629e-02 8.08287859e-01 7.40889013e-01
-3.08767766e-01 -8.62555280e-02 -4.22913104e-01 3.51589799e-01
6.68510020e-01 1.98501065e-01 6.89641714e-01 -1.00655591e+00
1.00788817e-01 -9.43306610e-02 -2.11004838e-01 -1.08886874e+00
-2.00877666e-01 -5.49141824e-01 3.74439508e-01 -1.45854020e+00
9.63711292e-02 -8.50533307e-01 -4.39937353e-01 2.97986329e-01
-6.06690943e-01 1.38039440e-01 3.19924474e-01 3.20967466e-01
-2.80506581e-01 7.05356717e-01 1.43806362e+00 -1.56979024e-01
-2.83433557e-01 2.77359951e-02 -6.30704820e-01 4.87245113e-01
7.69296944e-01 -3.20934922e-01 -6.36831641e-01 -2.94572383e-01
-8.77071023e-02 7.11362483e-03 1.14981793e-01 -1.19408965e+00
1.12021312e-01 -3.58852118e-01 7.13600755e-01 -1.07594311e+00
1.38303384e-01 -1.03040040e+00 -1.68315262e-01 4.73473161e-01
-1.41156763e-01 9.44396108e-02 1.25809051e-02 5.51118493e-01
-2.08311617e-01 -3.06686699e-01 9.49337482e-01 1.19806252e-01
-8.17409515e-01 2.95337200e-01 1.05565302e-01 -3.84387732e-01
1.50929093e+00 -3.71644378e-01 -2.22053751e-02 -1.83039427e-01
-5.75558245e-02 3.16207081e-01 5.66341996e-01 5.65895140e-01
6.45095170e-01 -1.45778668e+00 -6.33308053e-01 2.58388460e-01
1.89893663e-01 6.26157746e-02 3.46905380e-01 7.67569423e-01
-6.64971650e-01 1.45381987e-01 -5.80975890e-01 -7.47937858e-01
-1.32606888e+00 7.27951884e-01 9.56525356e-02 -6.33617714e-02
-3.21546495e-01 5.25783181e-01 3.85455310e-01 -2.48471066e-01
-4.01232466e-02 -2.60643423e-01 -5.23974597e-01 7.24018067e-02
5.44344187e-01 1.37051344e-01 -1.60220921e-01 -6.33902848e-01
-1.78229734e-01 8.63993287e-01 2.22292487e-02 -7.93312714e-02
1.05865061e+00 -4.57262039e-01 -2.55393311e-02 2.17961237e-01
7.75382578e-01 -9.69371870e-02 -1.38696063e+00 -1.66991323e-01
-3.81519854e-01 -9.36132789e-01 1.30205363e-01 -8.53937089e-01
-1.24959958e+00 9.34223294e-01 1.04833782e+00 3.37793678e-02
1.64772332e+00 -6.41840041e-01 6.76315963e-01 -3.21342386e-02
4.92537737e-01 -1.29988265e+00 -1.07546411e-01 -4.37052026e-02
8.27179670e-01 -1.30937719e+00 1.13512710e-01 -5.39405048e-01
-8.56222808e-01 1.12574112e+00 9.49165404e-01 -3.54102015e-01
4.64631885e-01 9.71923843e-02 3.68627161e-02 2.56566554e-01
-1.63910165e-01 -1.09129898e-01 4.83404025e-02 7.47224092e-01
1.77164987e-01 -7.80008733e-02 -6.85718358e-01 3.47152799e-01
6.68255389e-02 2.45360970e-01 5.02749741e-01 6.64304733e-01
-7.31279910e-01 -9.09089983e-01 -6.22220099e-01 4.98776048e-01
-6.51925802e-02 -6.33524433e-02 1.29768699e-01 6.24500692e-01
4.64253724e-01 8.71481180e-01 -3.27252820e-02 -3.49025756e-01
1.70960784e-01 -4.20082510e-01 2.94329852e-01 -1.66920200e-01
-2.57074356e-01 1.61727756e-01 -4.48399514e-01 -3.06181848e-01
-7.16185927e-01 -4.76240039e-01 -1.16329110e+00 -6.73955167e-03
-1.00558507e+00 5.91719300e-02 7.55620778e-01 6.78624511e-01
4.30698514e-01 4.57504123e-01 1.03653181e+00 -9.78879213e-01
-2.98957855e-01 -7.97115088e-01 -6.34571016e-01 1.27158448e-01
-1.30729273e-01 -7.30803311e-01 -1.93521723e-01 1.93704352e-01] | [11.243192672729492, -1.0992883443832397] |
37d01ab9-4ae8-498a-a06d-6e53905034a3 | learning-spectro-temporal-representations-of | 2103.07125 | null | https://arxiv.org/abs/2103.07125v1 | https://arxiv.org/pdf/2103.07125v1.pdf | Learning spectro-temporal representations of complex sounds with parameterized neural networks | Deep Learning models have become potential candidates for auditory neuroscience research, thanks to their recent successes on a variety of auditory tasks. Yet, these models often lack interpretability to fully understand the exact computations that have been performed. Here, we proposed a parametrized neural network layer, that computes specific spectro-temporal modulations based on Gabor kernels (Learnable STRFs) and that is fully interpretable. We evaluated predictive capabilities of this layer on Speech Activity Detection, Speaker Verification, Urban Sound Classification and Zebra Finch Call Type Classification. We found out that models based on Learnable STRFs are on par for all tasks with different toplines, and obtain the best performance for Speech Activity Detection. As this layer is fully interpretable, we used quantitative measures to describe the distribution of the learned spectro-temporal modulations. The filters adapted to each task and focused mostly on low temporal and spectral modulations. The analyses show that the filters learned on human speech have similar spectro-temporal parameters as the ones measured directly in the human auditory cortex. Finally, we observed that the tasks organized in a meaningful way: the human vocalizations tasks closer to each other and bird vocalizations far away from human vocalizations and urban sounds tasks. | ['Emmanuel Dupoux', 'Anne-Catherine Bachoud-Lévi', 'Julien Karadayi', 'Rachid Riad'] | 2021-03-12 | null | null | null | null | ['sound-classification'] | ['audio'] | [-1.89287543e-01 -4.55028377e-02 4.37834591e-01 -3.95699292e-01
-4.88097131e-01 -6.15448892e-01 6.17785394e-01 1.78288043e-01
-7.13867426e-01 3.65013063e-01 3.19007725e-01 -1.84383907e-03
-5.27213037e-01 -4.35987890e-01 -3.99164677e-01 -9.32671726e-01
-6.67204797e-01 5.38004451e-02 3.87539625e-01 -1.14664339e-01
1.87801138e-01 5.86759269e-01 -1.96770024e+00 6.33381367e-01
3.05975914e-01 1.20721197e+00 2.80653685e-01 8.97410512e-01
2.10920006e-01 4.10206825e-01 -8.10487211e-01 -2.37547746e-03
-9.39950868e-02 -2.78733194e-01 -6.35839641e-01 -3.49661559e-01
3.16555232e-01 2.55395234e-01 -1.27267212e-01 8.67454946e-01
6.32354140e-01 2.74588615e-01 9.31417644e-01 -7.85810947e-01
-6.02255046e-01 9.32088792e-01 2.96670258e-01 5.88532865e-01
2.60714144e-01 3.03594232e-01 1.06221282e+00 -9.22217846e-01
1.23876981e-01 1.34514546e+00 8.14067006e-01 5.21306455e-01
-1.39318895e+00 -5.00958204e-01 -5.18407039e-02 3.92659605e-01
-1.19482613e+00 -6.71032250e-01 3.21761549e-01 -7.93607891e-01
1.05849659e+00 5.08879125e-01 8.63050997e-01 1.06532538e+00
7.15449303e-02 3.01351428e-01 1.27005398e+00 -3.67795765e-01
3.94313455e-01 2.02755570e-01 3.44686896e-01 5.08279085e-01
-2.59485841e-01 4.65285808e-01 -7.98396528e-01 -4.17902619e-02
5.50271511e-01 -5.72977662e-01 -5.56985736e-01 2.35731587e-01
-1.17640865e+00 7.44736314e-01 4.39064294e-01 8.04516196e-01
-2.25641027e-01 2.86073565e-01 5.07975221e-01 2.69694477e-01
5.20829678e-01 5.88899791e-01 -5.90752125e-01 -9.19342041e-02
-8.85742068e-01 -4.23796140e-02 8.38820338e-01 1.61410689e-01
6.09916031e-01 4.58101928e-01 -3.08960021e-01 1.15958691e+00
2.69084990e-01 4.46220309e-01 7.29691625e-01 -7.89691925e-01
-9.53929946e-02 9.77483764e-03 -2.40759984e-01 -6.94767058e-01
-9.04461920e-01 -7.83358693e-01 -6.61540210e-01 3.08023870e-01
5.38766682e-01 1.44543350e-01 -7.42934525e-01 1.73205483e+00
-1.29942104e-01 1.98791817e-01 -1.27450779e-01 1.00516295e+00
7.58660674e-01 6.31026566e-01 2.29351118e-01 -2.40271017e-02
1.74396563e+00 -5.26255369e-01 -6.35091722e-01 -1.21292062e-01
1.58808038e-01 -6.85150862e-01 1.30435932e+00 5.42009115e-01
-8.68091702e-01 -8.39787364e-01 -1.05455065e+00 2.53677279e-01
-5.79196513e-01 3.40965360e-01 5.49079061e-01 9.98118639e-01
-1.25612473e+00 8.39417994e-01 -7.29894698e-01 -2.14298338e-01
9.46050808e-02 1.75217107e-01 -3.33757132e-01 9.01999831e-01
-1.11124837e+00 8.96947563e-01 4.38537091e-01 2.19962254e-01
-1.23297000e+00 -6.24343216e-01 -5.10719717e-01 5.16504288e-01
-1.31411269e-01 -4.10446972e-01 1.43777227e+00 -8.17859530e-01
-1.69985664e+00 8.97129118e-01 1.56935468e-01 -8.11051965e-01
-4.41949926e-02 1.11325383e-01 -5.56010485e-01 1.25148863e-01
-3.88769686e-01 4.50423539e-01 1.01913691e+00 -9.67297375e-01
-2.94249684e-01 -1.76982388e-01 -2.25045264e-01 -2.85105228e-01
-3.94337833e-01 1.52040616e-01 4.07029986e-01 -9.12710726e-01
8.05412456e-02 -7.81640947e-01 2.48721242e-01 -5.92375249e-02
-1.22244179e-01 -3.18231642e-01 2.79828340e-01 -5.56475580e-01
1.14758110e+00 -2.39909410e+00 7.19062686e-02 1.75059825e-01
1.40418828e-01 3.52977306e-01 -2.40220711e-01 3.22744697e-01
-2.77654290e-01 1.45755097e-01 -2.79066086e-01 -2.10735515e-01
3.45704257e-01 4.49538603e-02 -4.96889502e-01 3.69168788e-01
1.13865159e-01 4.35007662e-01 -6.24221027e-01 1.26046672e-01
5.00040166e-02 6.36936069e-01 -6.22488558e-01 1.26998276e-01
-1.50070461e-02 2.65683144e-01 1.31962467e-02 9.84168649e-02
4.56518143e-01 3.45161110e-01 -1.58033639e-01 -3.73390079e-01
-3.73269439e-01 6.24401987e-01 -9.58260119e-01 1.29859746e+00
-7.54932046e-01 1.12232363e+00 3.91009748e-01 -1.09931600e+00
8.08463871e-01 5.77690482e-01 1.78856626e-01 -5.64705193e-01
1.21332757e-01 3.44803393e-01 5.98909974e-01 -5.25115311e-01
-7.97382295e-02 -1.81843743e-01 3.09659153e-01 4.43072408e-01
7.49362707e-01 -2.91258335e-01 -6.99584857e-02 -3.43534142e-01
8.07405889e-01 -3.74696553e-01 2.19120994e-01 -6.43442869e-01
6.56664670e-01 -7.49027252e-01 3.83075923e-02 7.12019861e-01
-1.82918727e-01 5.00577152e-01 3.23816925e-01 -4.06718314e-01
-5.93646884e-01 -1.34635746e+00 -5.37450314e-01 1.60764384e+00
-4.95459974e-01 -4.68718022e-01 -1.19794202e+00 1.19310796e-01
-3.08977902e-01 6.73823178e-01 -6.54162467e-01 -3.58466566e-01
-3.95869702e-01 -7.25266933e-01 1.07427907e+00 3.11539054e-01
1.10618703e-01 -1.51188242e+00 -7.95968235e-01 2.83663988e-01
-7.86873549e-02 -1.10202730e+00 -2.86998630e-01 7.25844979e-01
-6.86693966e-01 -8.78508270e-01 -3.45887423e-01 -6.39266908e-01
-6.46093860e-02 -1.93995491e-01 8.22638273e-01 -2.51384407e-01
-6.81229055e-01 5.56658983e-01 -1.84592679e-01 -7.83280969e-01
-5.37657440e-01 -8.83789659e-02 3.82686645e-01 2.61867076e-01
-1.20750330e-02 -8.38440597e-01 -5.20752370e-01 3.94278169e-01
-9.03121889e-01 -4.79775310e-01 3.68270934e-01 7.95098186e-01
3.02251488e-01 -1.19358189e-01 6.39111698e-01 -2.56999999e-01
8.89669001e-01 -1.83840483e-01 -2.99287409e-01 -5.50678931e-03
-1.87195748e-01 3.03857297e-01 5.58826625e-01 -6.28840387e-01
-8.03458571e-01 -1.29205510e-01 -6.05769873e-01 7.32572377e-03
-5.77024996e-01 3.15892518e-01 2.49256253e-01 -1.53381675e-01
1.16739082e+00 2.53923506e-01 -2.94556707e-01 -6.80621684e-01
9.42257568e-02 6.19115651e-01 4.31009948e-01 -4.79523271e-01
5.69888830e-01 3.75552416e-01 -3.47186714e-01 -1.43056023e+00
-6.18065655e-01 -3.31084907e-01 -3.84699166e-01 -3.35208178e-01
1.15093625e+00 -6.06702030e-01 -1.08477759e+00 6.38152838e-01
-1.26408899e+00 -3.74219298e-01 -4.31209028e-01 9.99777973e-01
-7.07781196e-01 6.46838695e-02 -7.33178794e-01 -1.27799118e+00
-3.60335857e-01 -8.75189543e-01 9.96317983e-01 4.61087078e-02
-1.67965069e-01 -9.38307941e-01 2.76910454e-01 -6.02931976e-02
6.91841900e-01 -2.72000372e-01 1.15679383e+00 -7.89509654e-01
-2.66688734e-01 2.10330278e-01 4.22479995e-02 6.56979978e-01
-8.51146355e-02 1.47025719e-01 -1.95205331e+00 -4.13965061e-02
4.10225391e-01 -1.21687144e-01 1.19943380e+00 7.80214369e-01
1.38721120e+00 -3.76631916e-01 1.58249363e-01 5.41307688e-01
7.59404957e-01 2.50832349e-01 4.44980979e-01 -1.81748644e-01
-6.66985884e-02 8.41610730e-01 -1.10349685e-01 2.99793988e-01
-2.58981496e-01 8.97335052e-01 4.21965241e-01 9.50289741e-02
-3.52534682e-01 -1.25790775e-01 7.00461447e-01 1.12040222e+00
-4.23160136e-01 -4.30464968e-02 -6.08510137e-01 3.74932200e-01
-1.50147748e+00 -1.11532140e+00 6.08639121e-02 2.31339312e+00
6.54731393e-01 2.24461347e-01 2.75853634e-01 5.55831134e-01
4.49985862e-01 -1.08948611e-02 -1.39057234e-01 -6.77876115e-01
-3.09106767e-01 7.91394293e-01 -5.87960295e-02 5.54381430e-01
-9.65462744e-01 5.55761158e-01 7.38791466e+00 9.59190845e-01
-1.28463936e+00 3.41347456e-01 1.96094856e-01 -1.69756927e-03
-1.29979163e-01 -4.44258481e-01 -5.09229124e-01 2.48366803e-01
1.36176634e+00 1.11037448e-01 8.36686730e-01 6.88536406e-01
4.21322972e-01 4.07462716e-02 -1.16573882e+00 9.99619722e-01
-2.81641722e-01 -1.19334531e+00 -1.40730694e-01 -1.00811496e-01
1.14206132e-02 1.14497408e-01 2.07907543e-01 2.82552004e-01
-1.86956704e-01 -1.32790983e+00 1.07812893e+00 6.87561989e-01
4.64621276e-01 -3.42903078e-01 4.10472333e-01 4.13546264e-01
-1.33082223e+00 -3.19531858e-01 -5.30449033e-01 -1.40346482e-01
-1.73325762e-01 5.66088319e-01 -1.04863358e+00 -3.01165022e-02
8.52512002e-01 2.55385429e-01 -6.10215902e-01 1.28653216e+00
-1.60516977e-01 9.92958128e-01 -3.77016991e-01 -3.83669823e-01
8.40994567e-02 9.40251276e-02 7.37188697e-01 1.66775119e+00
4.24053997e-01 -1.96928293e-01 -2.86390781e-01 1.02127147e+00
2.25749195e-01 2.06297636e-01 -3.97859633e-01 -1.79402649e-01
2.35023171e-01 1.30788779e+00 -7.35621870e-01 9.91025120e-02
1.99326612e-02 4.08260018e-01 1.71544209e-01 4.87565756e-01
-7.85024583e-01 -2.80429423e-01 9.39904094e-01 1.97884783e-01
1.56247213e-01 -3.10541958e-01 -1.57497767e-02 -8.94759953e-01
-1.32570669e-01 -5.32031655e-01 1.42960519e-01 -7.95102894e-01
-1.31141841e+00 9.71508622e-01 6.46724775e-02 -9.30035293e-01
-2.77959675e-01 -1.20634627e+00 -5.94784498e-01 9.42277074e-01
-1.12718320e+00 -6.54312849e-01 2.39999481e-02 6.72294974e-01
4.63850260e-01 -2.51400858e-01 1.26818526e+00 3.75866652e-01
-2.32103154e-01 5.13947010e-01 -2.00593516e-01 8.64057466e-02
3.38003337e-01 -1.14189947e+00 2.80789763e-01 4.59535360e-01
8.68243396e-01 6.28895760e-01 6.38560951e-01 1.96091965e-01
-6.85780942e-01 -5.93053401e-01 6.48999810e-01 -3.73040289e-02
7.24924326e-01 -7.36517489e-01 -9.88930047e-01 1.76217854e-01
2.91643173e-01 5.16043007e-02 7.45058239e-01 2.04802558e-01
-5.48722565e-01 -2.94058353e-01 -8.36004317e-01 2.66882837e-01
9.95834291e-01 -1.13658488e+00 -6.61765456e-01 4.28709894e-01
5.50646424e-01 1.26733288e-01 -5.21788180e-01 1.61118269e-01
7.26264477e-01 -1.37890828e+00 1.04373157e+00 -7.79545009e-01
-3.11551690e-01 -3.37840825e-01 -1.34243682e-01 -1.69595873e+00
-4.74429667e-01 -3.96260560e-01 1.76123321e-01 8.51892114e-01
4.38914955e-01 -8.81773233e-01 5.15586212e-02 -2.16706902e-01
-4.31323469e-01 -2.99415827e-01 -1.37053406e+00 -8.98156524e-01
-1.44671962e-01 -8.13436866e-01 1.03563033e-01 4.95454371e-01
4.00663503e-02 4.63685632e-01 -6.85217604e-03 1.63532212e-01
2.78887838e-01 -1.95338070e-01 2.47099120e-02 -1.42019737e+00
-6.26450062e-01 -8.52255344e-01 -7.06131041e-01 -6.85112238e-01
2.41621181e-01 -9.56105828e-01 2.61284977e-01 -9.39411759e-01
-3.20653677e-01 -1.68433562e-01 -5.31409979e-01 4.37106967e-01
2.08863348e-01 1.82323158e-01 9.93459523e-02 -2.45006010e-01
2.40400687e-01 3.59344810e-01 8.92826557e-01 -1.99262097e-01
-1.18896097e-01 4.49979037e-01 -1.31421849e-01 1.03835094e+00
7.43640244e-01 -4.64645386e-01 -3.00208092e-01 -2.42292762e-01
1.26997977e-01 -1.50958538e-01 7.34842420e-01 -1.53375983e+00
1.44001573e-01 2.99255639e-01 1.76877692e-01 -3.48732136e-02
6.83690488e-01 -5.88983059e-01 8.98412839e-02 5.58570385e-01
-6.09382391e-01 -4.65425283e-01 4.14791077e-01 4.19713438e-01
-4.27700460e-01 -4.24042732e-01 1.08480334e+00 4.88560796e-02
-5.03229320e-01 -9.17336624e-03 -1.01063824e+00 -6.70870719e-03
4.42242593e-01 9.77896503e-04 -1.42659530e-01 -4.42623436e-01
-1.45387554e+00 -6.99737728e-01 -5.02100229e-01 3.17322135e-01
4.77945656e-01 -9.59985018e-01 -7.58229911e-01 4.02969122e-01
-4.93145781e-03 -7.83605158e-01 3.64005774e-01 8.06666970e-01
-3.30624253e-01 5.55143774e-01 -4.45047617e-01 -7.44224727e-01
-1.14252925e+00 3.22006434e-01 9.34123278e-01 2.08225355e-01
-1.65624112e-01 1.06906581e+00 6.30723476e-01 -4.76150572e-01
4.85327750e-01 -1.22442579e+00 -4.65469003e-01 3.75739336e-01
6.33858383e-01 3.24208111e-01 4.95764673e-01 -4.99519080e-01
-3.63082916e-01 5.84361196e-01 6.71749115e-01 -2.98609436e-01
1.35471141e+00 2.30381802e-01 -6.99862605e-03 8.79510164e-01
9.79553580e-01 1.39995962e-01 -9.29315627e-01 4.31762915e-03
7.31298625e-02 2.37174928e-02 1.34530701e-02 -9.33564663e-01
-8.82650018e-01 1.50293338e+00 1.02790809e+00 9.26221550e-01
1.17225158e+00 -5.28645143e-03 1.91931561e-01 4.29682314e-01
2.83546180e-01 -1.07057917e+00 2.02170819e-01 6.80257320e-01
1.20888674e+00 -5.86057067e-01 -4.62541550e-01 -4.12273854e-01
-3.01006407e-01 1.55761015e+00 1.69690341e-01 9.79429344e-04
9.76339757e-01 3.69362950e-01 -3.86088304e-02 1.28109893e-02
-7.99175084e-01 -5.01631618e-01 6.26406908e-01 8.48200679e-01
5.50545990e-01 3.20565403e-01 -8.56174342e-03 9.05915678e-01
-5.09192169e-01 -5.90707779e-01 2.36475915e-01 8.08840469e-02
-8.01473439e-01 -6.27183914e-01 -4.43741977e-01 2.36284301e-01
-3.28483820e-01 -2.91543961e-01 -3.74974698e-01 5.20372391e-01
2.55910158e-01 8.85461986e-01 9.02725011e-02 -5.86264491e-01
5.33222139e-01 5.17517328e-01 6.18533850e-01 -7.76323199e-01
-1.00210547e+00 2.13230580e-01 1.36614725e-01 -3.92476887e-01
-3.60377461e-01 -3.91219884e-01 -1.21112442e+00 2.18730882e-01
-2.24694461e-01 3.11042756e-01 8.38931203e-01 8.07317674e-01
5.17503172e-02 7.36296833e-01 4.53752130e-01 -1.12292278e+00
-6.39463127e-01 -1.30936134e+00 -8.59995008e-01 1.09499902e-01
5.08379817e-01 -6.88484967e-01 -7.59976387e-01 2.12194830e-01] | [15.230658531188965, 5.5165815353393555] |
f3d7a60f-2e85-4f05-8b0f-29575d9f4a03 | semantic-segmentation-with-active-semi-1 | 2210.08403 | null | https://arxiv.org/abs/2210.08403v1 | https://arxiv.org/pdf/2210.08403v1.pdf | Semantic Segmentation with Active Semi-Supervised Representation Learning | Obtaining human per-pixel labels for semantic segmentation is incredibly laborious, often making labeled dataset construction prohibitively expensive. Here, we endeavor to overcome this problem with a novel algorithm that combines semi-supervised and active learning, resulting in the ability to train an effective semantic segmentation algorithm with significantly lesser labeled data. To do this, we extend the prior state-of-the-art S4AL algorithm by replacing its mean teacher approach for semi-supervised learning with a self-training approach that improves learning with noisy labels. We further boost the neural network's ability to query useful data by adding a contrastive learning head, which leads to better understanding of the objects in the scene, and hence, better queries for active learning. We evaluate our method on CamVid and CityScapes datasets, the de-facto standards for active learning for semantic segmentation. We achieve more than 95% of the network's performance on CamVid and CityScapes datasets, utilizing only 12.1% and 15.1% of the labeled data, respectively. We also benchmark our method across existing stand-alone semi-supervised learning methods on the CityScapes dataset and achieve superior performance without any bells or whistles. | ['Matthew Hoffman', 'Christopher Kanan', 'Aneesh Rangnekar'] | 2022-10-16 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [ 2.85352498e-01 5.89772105e-01 -4.03150946e-01 -6.75869703e-01
-1.30884123e+00 -8.12779844e-01 3.06247413e-01 2.06853077e-01
-8.10759068e-01 6.25124812e-01 -9.61255208e-02 -1.94609061e-01
1.87588677e-01 -8.87152433e-01 -6.36884272e-01 -7.46582150e-01
2.75537550e-01 7.79826462e-01 8.01644206e-01 1.06687158e-01
-1.44344449e-01 2.74681807e-01 -1.59505451e+00 1.11302927e-01
1.32385516e+00 1.05973625e+00 1.86890647e-01 3.86948526e-01
-3.70526254e-01 9.42807615e-01 -6.71207666e-01 -3.32940817e-01
1.96717903e-01 -1.40514970e-01 -1.30766487e+00 4.96814489e-01
6.53165698e-01 -1.67467237e-01 2.05995843e-01 8.44520509e-01
3.85601103e-01 2.23545507e-01 6.14024818e-01 -1.13090396e+00
-1.62829220e-01 5.99342585e-01 -4.99859095e-01 -1.01690985e-01
-1.31876511e-03 1.07753128e-01 1.15612555e+00 -6.28376722e-01
5.14343143e-01 1.06107020e+00 5.76185822e-01 5.50138414e-01
-1.16792762e+00 -6.63940668e-01 3.17322701e-01 7.72597734e-03
-1.37756538e+00 -4.97699976e-01 6.78948343e-01 -2.98409194e-01
7.21480131e-01 1.77750751e-01 6.50898635e-01 8.29787195e-01
-9.78074253e-01 1.59580576e+00 1.02275324e+00 -5.24742246e-01
4.63731319e-01 2.69274563e-01 3.02432030e-01 7.62196600e-01
7.92033151e-02 -8.13466758e-02 -3.98869246e-01 5.19461371e-02
5.58460295e-01 -9.31545943e-02 -2.52365340e-02 -5.55274725e-01
-8.12618196e-01 9.46440756e-01 4.93378609e-01 3.46910767e-02
-1.33854210e-01 -9.06747505e-02 1.38555616e-01 -8.78543705e-02
9.08872068e-01 4.29426312e-01 -8.24997067e-01 3.26217748e-02
-1.31158543e+00 -1.53248254e-02 7.82031000e-01 9.40473020e-01
1.04911709e+00 -2.73671355e-02 1.30284131e-01 1.03578353e+00
5.25140047e-01 4.68328923e-01 1.15379570e-02 -1.22693694e+00
3.81764859e-01 9.67296958e-01 8.69404301e-02 -4.36984926e-01
-2.10375220e-01 -4.88176674e-01 -3.08083564e-01 3.20644557e-01
4.70403582e-01 -2.19650179e-01 -1.32732511e+00 1.40671575e+00
5.57861209e-01 2.53213108e-01 4.08846997e-02 6.13412619e-01
1.09519351e+00 5.46064496e-01 3.34584445e-01 -6.72367439e-02
8.92372429e-01 -1.50967216e+00 -3.94783229e-01 -4.09764051e-01
8.76280606e-01 -5.34213543e-01 1.20571232e+00 3.74525100e-01
-8.48623872e-01 -6.39047384e-01 -7.96470106e-01 -2.02137195e-02
-5.83901286e-01 -5.12789972e-02 7.83522725e-01 6.36309266e-01
-9.58766401e-01 3.54160249e-01 -1.00277495e+00 -2.42891029e-01
1.06112647e+00 3.20416987e-01 -6.36665523e-02 6.00797050e-02
-8.35752547e-01 4.66430277e-01 8.09782386e-01 -2.67469078e-01
-8.69094193e-01 -7.56836057e-01 -8.85543108e-01 -1.69216037e-01
9.08553958e-01 -1.08331092e-01 1.39370441e+00 -1.20961440e+00
-1.36098087e+00 1.18594360e+00 2.40272582e-02 -3.80632490e-01
5.67284822e-01 -1.70667350e-01 -2.77019739e-01 3.47163469e-01
3.52231860e-01 1.30514371e+00 5.46403408e-01 -1.58438778e+00
-9.09428716e-01 -3.02842230e-01 2.02773929e-01 3.43775064e-01
-2.81657934e-01 -3.14733535e-01 -8.25410306e-01 -4.15590793e-01
3.23859066e-01 -9.60966587e-01 -3.42312783e-01 1.05407700e-01
-4.52522457e-01 -5.19066572e-01 1.15984607e+00 -2.60351509e-01
7.94553578e-01 -2.22049570e+00 -2.85199553e-01 1.61648244e-01
3.44400942e-01 5.50912619e-01 -9.12161171e-02 -7.56580010e-02
1.93115532e-01 1.28850326e-01 -6.60831869e-01 -7.08124876e-01
-1.73230588e-01 5.52561581e-01 -1.94302917e-01 1.26332924e-01
3.32898766e-01 1.07569325e+00 -1.17949677e+00 -9.65112329e-01
2.50024676e-01 2.42316902e-01 -4.27564740e-01 3.20228815e-01
-4.74685133e-01 5.57417631e-01 -4.01994854e-01 7.90395558e-01
4.77544397e-01 -4.92108732e-01 -1.37089133e-01 2.31871262e-01
-9.09821168e-02 2.73871899e-01 -1.22160876e+00 1.85938108e+00
-3.13154340e-01 4.09353644e-01 4.63147126e-02 -1.17481005e+00
9.35027242e-01 1.94793582e-01 5.94580531e-01 -9.70903456e-01
-1.51963860e-01 3.44723642e-01 -4.96780097e-01 -4.90294963e-01
2.72461772e-01 3.42111975e-01 4.01289128e-02 4.81982529e-01
2.51436949e-01 -4.00627196e-01 3.78000051e-01 2.90237218e-01
7.83040583e-01 3.81260812e-01 -1.63108855e-01 -2.52451718e-01
2.48563483e-01 3.79460543e-01 5.24842262e-01 8.66759956e-01
-4.42199349e-01 7.02949822e-01 2.10090116e-01 -2.40132689e-01
-5.86678624e-01 -1.17309451e+00 -1.35650188e-01 1.30861712e+00
1.57407761e-01 -1.28066033e-01 -9.87160623e-01 -1.35393405e+00
-3.31481159e-01 7.16051280e-01 -4.50970083e-01 2.17279345e-01
-2.99548388e-01 -8.11792254e-01 5.61748207e-01 6.37708783e-01
1.13539338e+00 -1.02358568e+00 -5.75106084e-01 -1.98796596e-02
-2.86360115e-01 -1.17636526e+00 -5.91256879e-02 3.66158575e-01
-1.00510895e+00 -1.13853776e+00 -8.17004979e-01 -9.52909291e-01
7.06510246e-01 2.35142589e-01 1.47040641e+00 1.01828530e-01
-5.22485003e-02 2.33532742e-01 -5.91043830e-01 -5.72412789e-01
-2.65511841e-01 4.55300838e-01 -5.88331938e-01 -1.65759593e-01
4.21899885e-01 -2.42921308e-01 -5.78553855e-01 4.43701565e-01
-9.00824070e-01 1.93326354e-01 3.59893858e-01 4.57702070e-01
8.75478327e-01 1.37467787e-01 4.52605963e-01 -1.53093886e+00
-7.71868080e-02 -2.78078139e-01 -7.52440810e-01 1.72078669e-01
-7.33659804e-01 -1.68885335e-01 2.35228881e-01 -3.00976336e-01
-1.31981623e+00 6.44352376e-01 -4.29471970e-01 1.03412922e-02
-4.92234439e-01 2.07269743e-01 -2.08944499e-01 -7.21792430e-02
8.42524290e-01 -9.06697586e-02 -4.13194299e-01 -4.87069875e-01
4.84951735e-01 9.54767644e-01 4.82066214e-01 -4.25618619e-01
6.62199855e-01 6.77012503e-01 -5.13842106e-01 -8.60448778e-01
-1.61031866e+00 -9.13906336e-01 -9.49632943e-01 -1.17063344e-01
8.78168643e-01 -1.05195415e+00 -2.41226971e-01 7.27579772e-01
-7.65736461e-01 -8.65187883e-01 -6.06346548e-01 2.41047665e-01
-4.20784324e-01 2.91646928e-01 -3.79238546e-01 -8.22134137e-01
-2.66346067e-01 -1.18410945e+00 1.35310340e+00 3.88889164e-01
-1.29166752e-01 -1.23230517e+00 3.22230756e-02 9.05699790e-01
2.18261063e-01 2.17184678e-01 5.00330150e-01 -9.69699085e-01
-6.35946929e-01 -2.57148352e-02 -3.90312076e-01 5.79185486e-01
1.58517454e-02 -1.03071094e-01 -1.36125481e+00 -9.97447222e-02
-3.03376228e-01 -7.93768883e-01 1.17744708e+00 1.88298330e-01
1.18003404e+00 -5.40565420e-03 -2.45880917e-01 5.17407060e-01
1.25403941e+00 1.78750604e-01 5.64985096e-01 3.18797082e-01
8.03869069e-01 7.43998349e-01 8.15557420e-01 6.08810708e-02
4.44997877e-01 4.75170493e-01 4.31286633e-01 -8.01979244e-01
-2.74505973e-01 -2.29842454e-01 -1.69200659e-01 6.01696074e-01
1.01096034e-01 -1.95081532e-01 -1.24010432e+00 6.85740411e-01
-1.86749732e+00 -4.75399554e-01 -3.10481995e-01 1.97755182e+00
1.06228089e+00 4.25618112e-01 1.64964497e-01 4.13134962e-01
3.71830463e-01 2.77326047e-01 -6.73182845e-01 2.65166938e-01
-1.02102861e-01 4.36130822e-01 6.34409845e-01 4.87627596e-01
-1.52429974e+00 1.44552302e+00 6.34275579e+00 1.10070109e+00
-9.33935404e-01 2.55997181e-01 8.04580450e-01 2.26315245e-01
2.82109268e-02 -7.93930609e-03 -8.53505671e-01 2.14784861e-01
5.59640169e-01 5.19893527e-01 2.46982779e-02 1.03025532e+00
1.36129633e-01 -5.29071927e-01 -8.36223066e-01 9.82125640e-01
-2.60011498e-02 -1.16108966e+00 -2.25988656e-01 -9.35789198e-02
1.16305661e+00 2.24327877e-01 -1.84338748e-01 3.06667268e-01
6.96416438e-01 -8.38572860e-01 6.76326156e-01 1.10717110e-01
5.92551887e-01 -5.68703353e-01 7.50816941e-01 5.76257050e-01
-1.04277575e+00 6.40551224e-02 -1.07539989e-01 7.81110153e-02
7.71678537e-02 6.39491081e-01 -9.13567007e-01 2.26394504e-01
8.04015100e-01 7.33274877e-01 -8.83325040e-01 1.17568755e+00
-5.99961162e-01 1.06746829e+00 -5.25144815e-01 1.40991151e-01
4.65362132e-01 -9.76272523e-02 8.06436390e-02 1.00087106e+00
-3.03581178e-01 3.40442024e-02 5.80044270e-01 8.09965611e-01
-2.24951923e-01 1.23538196e-01 -2.04078257e-01 -7.41012841e-02
5.04246294e-01 1.18756676e+00 -1.32412934e+00 -6.63006067e-01
-2.93732166e-01 1.04043913e+00 3.71419311e-01 6.07080877e-01
-7.17869580e-01 -2.78516829e-01 1.05746716e-01 -4.14183959e-02
1.56174794e-01 -1.85927257e-01 -5.21850705e-01 -9.42954838e-01
-4.00578320e-01 -5.80563843e-01 5.63338816e-01 -7.47890294e-01
-1.00117671e+00 3.34419727e-01 7.92527348e-02 -8.29400361e-01
-1.89059600e-01 -3.10198575e-01 -4.12678212e-01 4.96568739e-01
-1.65967155e+00 -1.41597533e+00 -5.09336770e-01 5.95096171e-01
7.63922453e-01 -1.55374212e-02 7.47069716e-01 3.87764484e-01
-5.52004635e-01 4.41471785e-01 -5.93628883e-02 6.25232518e-01
4.57463354e-01 -1.50939775e+00 5.41797400e-01 7.77738154e-01
5.20351529e-01 1.50541114e-02 2.67262071e-01 -4.90477920e-01
-7.08371878e-01 -1.31012177e+00 7.45528519e-01 -5.41233242e-01
4.88868684e-01 -4.69362348e-01 -8.62571836e-01 7.70059526e-01
-1.37486830e-01 5.49276881e-02 7.06109166e-01 1.89023644e-01
-2.83948511e-01 -6.72067031e-02 -1.15735340e+00 3.44041258e-01
1.19490123e+00 -4.72814947e-01 -5.87861896e-01 4.90519404e-01
9.90484655e-01 -3.69907618e-01 -4.39001411e-01 5.22007585e-01
6.16848506e-02 -8.07738423e-01 9.52361941e-01 -3.48738194e-01
1.88022166e-01 -1.23239957e-01 3.85695957e-02 -1.06616807e+00
4.44216505e-02 -2.31684804e-01 9.42574292e-02 1.48685157e+00
7.76947916e-01 -5.47513545e-01 1.35360420e+00 5.25327265e-01
-3.69861811e-01 -6.61481678e-01 -6.47986233e-01 -6.71913087e-01
-4.91353460e-02 -6.60601974e-01 2.76103973e-01 1.11816275e+00
-5.72055280e-01 3.40286225e-01 2.37358436e-01 4.73957807e-02
8.31776500e-01 5.08482242e-03 7.91993499e-01 -1.39654839e+00
-1.66795596e-01 -1.81447208e-01 -7.66194016e-02 -1.26204669e+00
2.63539761e-01 -7.65625954e-01 2.23334402e-01 -1.73126662e+00
9.41343084e-02 -9.07148659e-01 -2.65782531e-02 8.94863725e-01
-1.22580595e-01 7.29857445e-01 -6.34399354e-02 3.20899963e-01
-1.15318227e+00 3.45299363e-01 1.27678061e+00 -3.13286006e-01
-5.31374156e-01 1.71781287e-01 -5.11574268e-01 9.15855646e-01
7.45493531e-01 -5.80859184e-01 -7.70681798e-01 -5.29215574e-01
1.68243535e-02 -4.30049896e-01 2.87917256e-01 -9.82499003e-01
1.98315024e-01 -1.02608904e-01 1.64697826e-01 -7.12900639e-01
2.58681357e-01 -7.87835598e-01 -2.48187020e-01 2.01814458e-01
-4.95810568e-01 -7.68736660e-01 6.06152453e-02 3.16617578e-01
-3.33168089e-01 -3.75079215e-01 9.75280106e-01 -3.27279210e-01
-1.08702314e+00 4.05841023e-01 -1.02523349e-01 4.95693564e-01
9.85254169e-01 -3.40802938e-01 -2.58295625e-01 -1.37597591e-01
-8.07139635e-01 5.19908845e-01 4.73802000e-01 1.77027002e-01
3.60052377e-01 -8.13464820e-01 -4.58457708e-01 1.38247147e-01
2.53246278e-01 8.60965312e-01 2.94373930e-02 6.16088986e-01
-6.75622165e-01 1.61558598e-01 2.97307819e-01 -9.71585810e-01
-1.21494114e+00 2.10881338e-01 3.29242378e-01 -1.79571286e-01
-3.26318592e-01 1.07067370e+00 1.21821770e-02 -8.66199434e-01
5.54558873e-01 -4.70869355e-02 -3.94578665e-01 3.94234419e-01
2.08703697e-01 3.68296921e-01 2.28821173e-01 -4.97490823e-01
-9.68729481e-02 4.65926349e-01 9.98606160e-03 -1.37845904e-01
1.25701547e+00 -1.98046818e-01 1.81765720e-01 5.18145382e-01
1.13681734e+00 -1.46694362e-01 -1.47510028e+00 -4.57661420e-01
2.23930091e-01 -3.28946471e-01 3.96276802e-01 -1.10370123e+00
-1.29066253e+00 8.86142850e-01 7.93708682e-01 4.10428077e-01
1.03181148e+00 4.89250898e-01 8.62483203e-01 5.27602077e-01
4.41403776e-01 -1.42527783e+00 2.56329983e-01 4.82203275e-01
1.19336531e-01 -1.74396968e+00 -2.68544778e-02 -8.43469858e-01
-6.05980098e-01 7.24872053e-01 5.39031506e-01 1.33202180e-01
4.98042554e-01 1.64110541e-01 6.16946816e-01 -2.74931788e-01
-2.32389748e-01 -8.04194391e-01 2.30483517e-01 7.04326212e-01
2.84018934e-01 7.53341094e-02 7.41691962e-02 2.60236710e-01
-6.22860566e-02 1.77726382e-03 2.71229874e-02 1.10990644e+00
-7.66905189e-01 -9.36690807e-01 -1.50542811e-01 3.33304226e-01
-2.05002129e-01 -8.11834484e-02 -5.47274888e-01 8.54424417e-01
3.33192766e-01 1.19482636e+00 2.31656119e-01 2.34581647e-03
2.83953458e-01 1.75729990e-01 9.15809348e-02 -8.82537186e-01
-3.43350619e-01 2.50468612e-01 1.79059938e-01 -5.83000124e-01
-9.11330998e-01 -4.34162676e-01 -1.53515196e+00 6.29517734e-02
-4.71969366e-01 1.22056596e-01 4.97930735e-01 1.11917603e+00
1.05968066e-01 2.22245932e-01 5.61609685e-01 -5.97527027e-01
-1.96918771e-01 -7.47408926e-01 -4.51678514e-01 6.57156706e-01
-8.67296290e-03 -5.26960611e-01 -3.91702920e-01 2.66907871e-01] | [9.45173168182373, 0.7139582633972168] |
b144279f-f0bb-4ec5-aa13-954cbd200817 | learning-to-diversify-for-product-question | 2207.02534 | null | https://arxiv.org/abs/2207.02534v1 | https://arxiv.org/pdf/2207.02534v1.pdf | Learning to Diversify for Product Question Generation | We address the product question generation task. For a given product description, our goal is to generate questions that reflect potential user information needs that are either missing or not well covered in the description. Moreover, we wish to cover diverse user information needs that may span a multitude of product types. To this end, we first show how the T5 pre-trained Transformer encoder-decoder model can be fine-tuned for the task. Yet, while the T5 generated questions have a reasonable quality compared to the state-of-the-art method for the task (KPCNet), many of such questions are still too general, resulting in a sub-optimal global question diversity. As an alternative, we propose a novel learning-to-diversify (LTD) fine-tuning approach that allows to enrich the language learned by the underlying Transformer model. Our empirical evaluation shows that, using our approach significantly improves the global diversity of the underlying Transformer model, while preserves, as much as possible, its generation relevance. | ['Eliyahu Kiperwasser', 'Alexander Nus', 'Yotam Eshel', 'Uriel Singer', 'Haggai Roitman'] | 2022-07-06 | null | null | null | null | ['question-generation'] | ['natural-language-processing'] | [ 2.64818043e-01 7.03307331e-01 5.50155081e-02 -3.81698519e-01
-1.20723569e+00 -7.86938310e-01 6.00579381e-01 1.52443424e-01
8.27522278e-02 6.86012745e-01 5.94607115e-01 -1.27084360e-01
-1.18911564e-01 -9.09101665e-01 -8.60933483e-01 -1.12586789e-01
5.83375514e-01 8.26017559e-01 1.24886893e-01 -7.22586870e-01
5.06886542e-02 -3.99605781e-02 -1.63890791e+00 6.05442464e-01
1.22013152e+00 1.00214672e+00 4.67445940e-01 4.33107227e-01
-2.85053790e-01 7.63507128e-01 -6.33958936e-01 -9.04165626e-01
2.80954707e-02 -6.99079990e-01 -1.07517076e+00 3.30137283e-01
5.53851902e-01 -2.57483244e-01 -5.39390855e-02 8.51757228e-01
2.74992198e-01 3.98480557e-02 6.68543160e-01 -9.30599868e-01
-1.21043944e+00 9.99017000e-01 4.13993299e-02 -2.84694731e-01
5.26979208e-01 1.11198798e-01 1.61762583e+00 -8.30396354e-01
7.02697158e-01 1.12657559e+00 4.31937903e-01 6.66671515e-01
-1.31884527e+00 -3.29869390e-01 2.67990857e-01 1.37360059e-02
-1.15149200e+00 -2.62325585e-01 6.57372475e-01 -2.73924470e-01
8.56282830e-01 1.64887190e-01 3.92315716e-01 1.12018752e+00
4.09985669e-02 9.25141215e-01 6.39048517e-01 -1.75291717e-01
2.80956060e-01 6.16518617e-01 -2.43760291e-02 4.07901943e-01
2.62424946e-01 -2.54351676e-01 -3.52051824e-01 -6.13962449e-02
4.64940935e-01 -2.92537451e-01 -2.92932689e-01 -3.95589232e-01
-9.30090249e-01 1.08352935e+00 2.42066354e-01 4.59458888e-01
-5.11407018e-01 -7.88111761e-02 1.88406214e-01 5.74468732e-01
3.91712159e-01 1.22862661e+00 -9.92841005e-01 1.09345643e-02
-8.38571191e-01 8.07304621e-01 1.06441176e+00 1.21631253e+00
9.10467982e-01 -2.77732730e-01 -7.44215965e-01 7.02450454e-01
1.30917653e-01 1.56862587e-01 4.86711711e-01 -7.90280282e-01
5.12838364e-01 7.73134589e-01 3.16088825e-01 -6.25633001e-01
-6.66643828e-02 -8.88301432e-01 -5.63306868e-01 -2.36099094e-01
4.66821373e-01 -2.75278360e-01 -6.80092037e-01 1.95327258e+00
6.70308694e-02 -3.15751672e-01 1.75822735e-01 5.23616612e-01
7.53358066e-01 8.07870209e-01 -1.23806819e-01 1.84619918e-01
1.48871183e+00 -1.07316256e+00 -7.10274637e-01 -4.73677605e-01
6.33484066e-01 -7.44566500e-01 1.46623850e+00 3.79383624e-01
-1.14760888e+00 -6.66915596e-01 -8.11012805e-01 -2.65606403e-01
-3.22857857e-01 2.77799547e-01 4.41600263e-01 7.12092221e-01
-9.70405042e-01 4.33502644e-01 -1.84553206e-01 -4.03658271e-01
1.39076859e-01 1.58137321e-01 2.65694279e-02 -3.53946328e-01
-1.42913067e+00 1.04319966e+00 4.54728812e-01 -2.81011939e-01
-8.47924471e-01 -1.08026707e+00 -9.70476151e-01 4.36267257e-01
5.29951334e-01 -1.14355755e+00 1.57695162e+00 -8.83743227e-01
-1.37869918e+00 3.39616627e-01 -1.42809600e-01 -2.16685295e-01
1.50524482e-01 7.47964755e-02 -2.33217269e-01 -1.18052982e-01
1.95879743e-01 9.04156983e-01 6.95170283e-01 -1.38833606e+00
-6.19116187e-01 -1.59905791e-01 6.38067901e-01 3.69293749e-01
-4.48576689e-01 -2.32138932e-01 -4.74872917e-01 -5.82875013e-01
-4.49151486e-01 -7.49432862e-01 -3.97866249e-01 -3.66388738e-01
-4.39272046e-01 -3.27933967e-01 3.87310922e-01 -7.92752266e-01
1.36047900e+00 -1.81348825e+00 1.68377995e-01 -1.05342187e-01
-2.00331882e-02 1.06399402e-01 -5.26440084e-01 7.03203201e-01
1.39573112e-01 2.98509598e-01 -2.91644037e-01 -5.37633300e-01
4.17768508e-01 1.09022252e-01 -3.46270204e-01 -2.28329092e-01
4.57860440e-01 1.23199534e+00 -8.69615555e-01 -1.05731100e-01
-1.50038362e-01 4.11214143e-01 -9.74621415e-01 3.25969219e-01
-9.91789281e-01 3.87507021e-01 -5.06857097e-01 3.54039013e-01
4.62671220e-01 -4.53400433e-01 -7.30722770e-02 -2.51159549e-01
2.99559951e-01 7.81950474e-01 -8.77309084e-01 1.90182936e+00
-9.35414255e-01 3.30023229e-01 -2.28574887e-01 -4.91129011e-01
9.70710278e-01 2.74104208e-01 3.58713746e-01 -7.09958255e-01
-8.88018385e-02 2.95143574e-01 -1.34885550e-01 -5.54916263e-01
9.56757665e-01 -3.96632880e-01 -2.62537599e-01 4.30252969e-01
3.12700808e-01 -2.97392249e-01 3.98569018e-01 1.09716609e-01
1.04725921e+00 2.07321793e-02 1.56505227e-01 -1.43356785e-01
4.90621626e-01 1.79782454e-02 1.12211086e-01 6.78635776e-01
3.93292308e-01 7.06601322e-01 6.43532693e-01 1.23086207e-01
-1.08115816e+00 -7.55917192e-01 5.78610823e-02 9.83697653e-01
-9.50525627e-02 -5.74324965e-01 -8.65614712e-01 -8.88574064e-01
-5.50698303e-02 1.25698471e+00 -6.41851246e-01 -3.03502262e-01
-3.98178697e-01 -3.57316971e-01 3.88032705e-01 2.64352351e-01
1.83411241e-01 -1.10801101e+00 -3.32709223e-01 4.55237895e-01
-3.91414732e-01 -1.10967517e+00 -7.66486883e-01 -7.32141957e-02
-5.27176321e-01 -6.96981072e-01 -8.96004677e-01 -8.61933529e-01
4.93783653e-01 5.00644669e-02 1.66175425e+00 -3.27358782e-01
1.69793680e-01 3.94762993e-01 -5.95023990e-01 -1.76240221e-01
-7.02992558e-01 6.06674135e-01 -6.75909340e-01 8.56024548e-02
4.99419719e-01 -3.72072577e-01 -4.51296091e-01 1.40059009e-01
-1.18808722e+00 -1.30251840e-01 7.00875878e-01 7.76586592e-01
3.26508164e-01 3.11414748e-01 9.90338266e-01 -1.06167161e+00
1.16334701e+00 -6.87695146e-01 -4.57881600e-01 5.14751256e-01
-7.03380108e-01 4.76073474e-01 8.38836372e-01 -5.27120233e-01
-1.15658951e+00 -1.16408952e-01 -6.12836123e-01 7.18265846e-02
-8.76491591e-02 7.83163309e-01 -4.88316715e-01 2.23091438e-01
7.28557169e-01 3.28797013e-01 -2.15503991e-01 -6.23198390e-01
7.50881732e-01 5.11251509e-01 1.45682856e-01 -4.76529509e-01
6.69259310e-01 -1.44345671e-01 -4.79565352e-01 -4.81846243e-01
-1.08577788e+00 -3.73626560e-01 -1.98050424e-01 2.16551274e-01
7.39170015e-01 -8.07159424e-01 -3.77041906e-01 3.86100933e-02
-1.41138327e+00 -1.85714513e-01 -7.71038175e-01 -1.12848161e-02
-6.30576611e-01 1.22276954e-01 -4.21745121e-01 -6.06558561e-01
-2.66563833e-01 -1.28492379e+00 1.27182746e+00 -2.03259904e-02
-4.09045905e-01 -1.08314550e+00 9.80703011e-02 4.37696189e-01
6.47895634e-01 -3.87746468e-02 1.35316741e+00 -9.37914908e-01
-6.57127738e-01 -4.69062030e-02 -5.01917042e-02 5.36000967e-01
3.93380165e-01 -5.22721827e-01 -8.06935430e-01 -1.02334090e-01
2.38938004e-01 -3.09258342e-01 8.22175860e-01 7.78319612e-02
1.04845321e+00 -7.62468755e-01 8.40481594e-02 1.52689770e-01
1.39388907e+00 -3.33457161e-03 6.44457042e-01 2.93977968e-02
5.90446413e-01 1.11714160e+00 4.90075588e-01 4.64609981e-01
9.21376884e-01 8.61314595e-01 4.54572856e-01 -8.70438442e-02
-2.28394553e-01 -6.83387339e-01 4.79536623e-01 5.09536982e-01
7.37549305e-01 -4.83277202e-01 -4.71677095e-01 9.20024395e-01
-1.70139778e+00 -7.67420411e-01 3.86689007e-02 2.14839411e+00
1.10776758e+00 -7.66216442e-02 1.93720847e-01 -2.04813972e-01
4.02857602e-01 3.62145230e-02 -6.73695385e-01 -3.91117126e-01
1.57909058e-02 2.51210570e-01 3.18596840e-01 4.85009074e-01
-5.00343978e-01 8.39702904e-01 6.43677473e+00 8.51274133e-01
-6.79605126e-01 1.84732482e-01 4.89902675e-01 1.25372067e-01
-1.30468988e+00 1.65543631e-02 -9.98776972e-01 4.30084080e-01
9.87022758e-01 -4.54511076e-01 4.17525113e-01 7.19949186e-01
-3.08289425e-03 3.93798441e-01 -1.39498854e+00 7.97876418e-01
3.10372561e-01 -1.35581124e+00 4.84292328e-01 1.16418414e-01
7.75566995e-01 -4.57056284e-01 2.89857984e-01 7.13265836e-01
3.67384374e-01 -9.93358195e-01 7.94138968e-01 1.91289589e-01
6.37382567e-01 -8.62971723e-01 5.94776094e-01 5.79954684e-01
-9.55721080e-01 -2.70186782e-01 -4.71142471e-01 3.51669848e-01
4.17842835e-01 7.96357930e-01 -8.80566478e-01 5.15380085e-01
2.47549951e-01 3.65466565e-01 -7.63821900e-01 7.71298170e-01
-2.03981489e-01 3.38233680e-01 -2.10824031e-02 -2.23744780e-01
4.52948809e-01 3.55289169e-02 2.89499044e-01 1.04608870e+00
6.44109428e-01 -1.14660338e-01 7.12080598e-02 1.30342472e+00
-3.88117105e-01 2.28105098e-01 -6.82852685e-01 -2.59749621e-01
1.95659727e-01 1.08574128e+00 5.01491204e-02 -1.67186454e-01
-5.40677428e-01 1.08955014e+00 1.52064621e-01 3.18620622e-01
-6.40241742e-01 -3.57418984e-01 7.25118756e-01 4.33318794e-01
5.00486076e-01 1.07771814e-01 -1.83224112e-01 -1.27831995e+00
1.49498329e-01 -1.34549880e+00 2.72110879e-01 -7.36485243e-01
-1.43388903e+00 8.48479271e-01 -1.84799489e-02 -1.00596070e+00
-7.80453980e-01 -4.08708394e-01 -1.62193730e-01 9.65939701e-01
-1.80997705e+00 -1.40572047e+00 -6.77065970e-03 4.45841223e-01
8.77746284e-01 -1.37936950e-01 6.29759908e-01 3.23460609e-01
-5.37916943e-02 7.27623105e-01 -1.68591201e-01 -4.30244178e-01
5.85338652e-01 -1.40220559e+00 7.59404778e-01 6.38537169e-01
-3.34714092e-02 6.23184502e-01 8.26210439e-01 -6.55225754e-01
-1.28368390e+00 -1.40612137e+00 1.49637127e+00 -6.11846387e-01
5.93638361e-01 -5.51297843e-01 -9.05103087e-01 6.70960844e-01
5.02491355e-01 -6.65946305e-01 7.09864795e-01 2.62912840e-01
-4.42046732e-01 5.88438183e-04 -1.20659947e+00 6.22113287e-01
7.80240834e-01 -5.62722564e-01 -6.38544619e-01 4.18799788e-01
1.16179252e+00 -1.92083135e-01 -9.39345181e-01 1.75172433e-01
2.15650499e-01 -8.72807741e-01 7.24838614e-01 -5.44550419e-01
6.24860168e-01 -1.99529946e-01 -2.45643616e-01 -1.57468510e+00
-4.83661979e-01 -7.84037948e-01 -1.71369672e-01 1.44823968e+00
9.64291573e-01 -4.64362681e-01 8.07813048e-01 7.03843594e-01
-2.41800800e-01 -8.13859582e-01 -6.08618319e-01 -7.86589086e-01
3.69611382e-01 -2.32118115e-01 1.08611619e+00 5.78958809e-01
3.10271699e-02 9.66927946e-01 -4.67691630e-01 -8.39711800e-02
3.10011923e-01 1.03577822e-01 5.37767053e-01 -1.11557651e+00
-7.11856842e-01 -4.33078706e-01 2.75783151e-01 -1.64840424e+00
1.11940600e-01 -9.24811423e-01 1.50758788e-01 -1.84343171e+00
9.15740430e-02 -4.50514197e-01 -5.79491025e-03 3.18632215e-01
-2.96974391e-01 -6.64869919e-02 1.87411755e-01 -2.23983422e-01
-3.47812563e-01 7.38925576e-01 1.65772808e+00 -1.74166486e-01
-2.25488976e-01 1.58100471e-01 -1.67192626e+00 5.23989089e-02
6.90672934e-01 -3.79677057e-01 -8.90817881e-01 -5.92230499e-01
8.33735704e-01 1.01512045e-01 9.21791047e-02 -5.47579348e-01
4.12654988e-02 -2.19292175e-02 -1.60598487e-01 -4.70517039e-01
3.29158098e-01 -7.63578534e-01 -1.11321323e-02 1.56376913e-01
-8.38134944e-01 3.31139332e-03 2.43826807e-02 2.63569266e-01
-4.00822610e-01 -5.39465666e-01 5.90638220e-01 -3.65006775e-01
-2.81491429e-01 4.27501738e-01 -3.19910258e-01 1.23912208e-01
5.22737563e-01 -5.04519753e-02 -2.85426617e-01 -8.72780621e-01
-3.15533787e-01 3.71182710e-01 4.44587678e-01 7.43143559e-01
4.25794244e-01 -1.39240849e+00 -1.02563560e+00 3.31749655e-02
5.93890727e-01 -3.28316279e-02 1.92352414e-01 2.23368973e-01
1.24150209e-01 7.63277411e-01 8.39445964e-02 -8.17855671e-02
-6.39901280e-01 7.32997179e-01 3.07972789e-01 -8.34523261e-01
-3.14059675e-01 7.53647804e-01 4.99594390e-01 -7.35526264e-01
3.90452594e-02 -6.27528012e-01 -4.74760413e-01 2.47558534e-01
4.96073991e-01 4.71851379e-02 2.75976986e-01 -3.57665241e-01
7.22089782e-02 3.68804187e-01 -2.45439991e-01 -2.64456302e-01
1.15559566e+00 -2.92111069e-01 1.76673517e-01 4.37571332e-02
1.26092029e+00 5.28323576e-02 -1.06516957e+00 -2.56386429e-01
5.64105883e-02 -1.67097688e-01 -1.65536717e-01 -1.09118748e+00
-9.26066756e-01 7.67649293e-01 1.46424174e-01 5.95561564e-01
1.21141422e+00 2.52047867e-01 1.21271479e+00 4.54755574e-01
2.34394014e-01 -7.84474313e-01 2.73179501e-01 7.06574261e-01
1.13563180e+00 -9.47408319e-01 -5.92739522e-01 -5.05252779e-01
-9.47794497e-01 8.32559049e-01 5.51703513e-01 1.79508716e-01
2.68199265e-01 4.59641442e-02 -1.53608292e-01 -1.28945917e-01
-1.05931401e+00 -5.56017935e-01 4.25131977e-01 6.39383376e-01
4.68810886e-01 -3.01765084e-01 -1.32011607e-01 7.94354379e-01
-5.37631035e-01 3.61205190e-02 6.51835859e-01 5.66144049e-01
-4.12021011e-01 -1.43989944e+00 7.03959465e-02 5.35464704e-01
-4.39579934e-01 -3.75871301e-01 -4.40250993e-01 3.95700097e-01
1.50199503e-01 1.18529642e+00 -3.18480998e-01 -3.16525906e-01
7.18061388e-01 -8.07433948e-02 5.08899450e-01 -1.10194778e+00
-9.61430013e-01 -5.56207374e-02 4.12775010e-01 -3.02394688e-01
8.77470300e-02 -5.78166783e-01 -6.88265264e-01 1.01961643e-01
-3.15075189e-01 2.54387885e-01 6.19650126e-01 9.62488055e-01
5.31067133e-01 5.24008393e-01 6.51026547e-01 -2.31689885e-01
-9.25134540e-01 -1.04876900e+00 -5.45918226e-01 4.13147926e-01
4.55449671e-01 -3.09995204e-01 -1.85299188e-01 7.20563382e-02] | [11.624471664428711, 8.542997360229492] |
6a4b39d4-9993-4981-ae45-8021878110d9 | pushing-the-limits-of-unconstrained-face | 1804.10275 | null | http://arxiv.org/abs/1804.10275v3 | http://arxiv.org/pdf/1804.10275v3.pdf | Pushing the Limits of Unconstrained Face Detection: a Challenge Dataset and Baseline Results | Face detection has witnessed immense progress in the last few years, with new
milestones being surpassed every year. While many challenges such as large
variations in scale, pose, appearance are successfully addressed, there still
exist several issues which are not specifically captured by existing methods or
datasets. In this work, we identify the next set of challenges that requires
attention from the research community and collect a new dataset of face images
that involve these issues such as weather-based degradations, motion blur,
focus blur and several others. We demonstrate that there is a considerable gap
in the performance of state-of-the-art detectors and real-world requirements.
Hence, in an attempt to fuel further research in unconstrained face detection,
we present a new annotated Unconstrained Face Detection Dataset (UFDD) with
several challenges and benchmark recent methods. Additionally, we provide an
in-depth analysis of the results and failure cases of these methods. The
dataset as well as baseline results will be made publicly available in due
time. The UFDD dataset as well as baseline results are available at:
www.ufdd.info/ | ['He Zhang', 'Hajime Nada', 'Vishal M. Patel', 'Vishwanath A. Sindagi'] | 2018-04-26 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [ 2.07388639e-01 -5.08605480e-01 2.54090220e-01 -5.60161054e-01
-4.73696381e-01 -6.92293704e-01 6.55107439e-01 -6.58172309e-01
-1.95256904e-01 6.76622212e-01 7.44887218e-02 2.61941940e-01
1.56440318e-01 -2.42364146e-02 -2.43005037e-01 -5.74846983e-01
-2.88825303e-01 6.06494769e-02 2.56919086e-01 -8.98902267e-02
1.97547004e-01 9.62154090e-01 -1.90899539e+00 2.51223803e-01
3.34945351e-01 8.71700883e-01 4.54933010e-03 5.93184114e-01
2.09208980e-01 4.68534380e-01 -8.03976059e-01 -5.91991723e-01
4.12460804e-01 -3.25805336e-01 -5.94887078e-01 2.38243550e-01
1.19748211e+00 -7.08091080e-01 -3.29717934e-01 1.21775782e+00
1.07595217e+00 -2.40633637e-01 3.65816265e-01 -1.44865584e+00
-5.01242816e-01 -1.42185211e-01 -7.35513926e-01 5.12354612e-01
5.83031774e-01 3.47580433e-01 3.60630542e-01 -1.21687531e+00
6.99242175e-01 1.70447028e+00 7.93132305e-01 8.81285191e-01
-9.05082524e-01 -9.15073335e-01 -2.80033099e-03 1.29594356e-01
-1.38494980e+00 -1.27004349e+00 6.10139251e-01 -4.32188183e-01
9.76569951e-01 3.10665399e-01 3.27481747e-01 1.40023780e+00
-1.79550275e-01 6.57373667e-01 1.10519719e+00 -5.15378118e-01
-3.85010131e-02 1.89591452e-01 -1.24672070e-01 8.26155365e-01
4.05593455e-01 2.56217062e-01 -4.55653846e-01 -2.04673961e-01
7.27970839e-01 -2.09661812e-01 -3.15956831e-01 -3.65180522e-01
-6.95225477e-01 5.09723365e-01 8.08072537e-02 1.21778317e-01
2.11564302e-01 -1.40375271e-01 3.07842851e-01 1.66512832e-01
5.46490431e-01 1.67543665e-01 -3.75537783e-01 -8.58644396e-02
-1.08971620e+00 3.17548811e-01 7.57315457e-01 9.91001010e-01
5.75627029e-01 1.02231435e-01 -1.57342032e-01 9.26872611e-01
3.98518950e-01 6.50264680e-01 1.09626956e-01 -9.23329055e-01
1.75931856e-01 2.01713637e-01 2.54067928e-01 -1.09490049e+00
-3.79285961e-01 -8.10571536e-02 -3.65799963e-01 3.93838137e-01
5.58599174e-01 -3.01154435e-01 -8.55184257e-01 1.61422801e+00
5.70665002e-01 3.03336233e-01 -3.50863755e-01 9.04371440e-01
9.92092550e-01 2.33914435e-01 -1.69852123e-01 -5.33918858e-01
1.46424353e+00 -8.82179976e-01 -9.85881865e-01 -4.61916775e-01
5.70286177e-02 -1.29862940e+00 5.16592383e-01 3.22826803e-01
-1.00380301e+00 -4.91139829e-01 -9.30568516e-01 1.14468478e-01
-4.32814658e-01 3.15809041e-01 6.63801908e-01 1.25105000e+00
-1.38649869e+00 3.43285501e-01 -6.60732865e-01 -1.14695251e+00
5.90483725e-01 4.32115763e-01 -4.69166368e-01 -2.62199581e-01
-8.30065668e-01 1.02122426e+00 -7.23391026e-02 2.80673534e-01
-1.00587642e+00 -5.66075325e-01 -8.23836386e-01 -5.41608512e-01
5.02311051e-01 -2.37676784e-01 1.33330750e+00 -8.16838861e-01
-1.16082871e+00 1.18471289e+00 -2.19401166e-01 1.49708703e-01
6.82913542e-01 -2.91133612e-01 -8.71061206e-01 1.00531042e-01
-7.73947015e-02 6.69158578e-01 1.13425171e+00 -1.13306844e+00
-5.99133909e-01 -5.90706170e-01 -6.86965957e-02 -3.06464713e-02
-5.72091639e-01 1.11913562e+00 -9.17665541e-01 -6.10852182e-01
-4.13138390e-01 -8.41681242e-01 3.15854818e-01 4.39656794e-01
-2.36921206e-01 -1.73847556e-01 1.41248107e+00 -5.63403428e-01
1.15298450e+00 -2.18609238e+00 -1.90859944e-01 -5.12171566e-01
-1.15499482e-01 6.52453363e-01 -2.81490058e-01 3.55699480e-01
-7.78134093e-02 1.24921734e-02 -1.67732373e-01 -7.11144269e-01
-4.53225933e-02 -5.11803254e-02 -2.00789928e-01 8.84306610e-01
5.18323123e-01 4.75193083e-01 -7.84065843e-01 -3.88943136e-01
1.38956621e-01 7.00907886e-01 -2.56784230e-01 2.42539763e-01
7.90013373e-02 2.62500346e-01 -7.08348677e-02 1.34833372e+00
1.28983867e+00 3.23883384e-01 -7.01240301e-02 -3.95208210e-01
-3.24316978e-01 -1.96128368e-01 -1.43716502e+00 1.33882344e+00
1.38260797e-01 8.20361495e-01 6.75872386e-01 -5.13934016e-01
6.32409275e-01 3.21531177e-01 3.50688875e-01 -3.41624439e-01
-1.72736719e-02 1.50307313e-01 -6.89558238e-02 -7.03068197e-01
4.34398055e-01 2.14084536e-01 4.31635976e-01 2.10612759e-01
8.81772563e-02 3.82071063e-02 2.94277638e-01 -1.82986651e-02
1.03869319e+00 2.84415096e-01 2.12898135e-01 -3.02037269e-01
5.56741357e-01 -1.75217465e-01 6.06760800e-01 5.10368645e-01
-9.43762302e-01 8.20962667e-01 3.10653538e-01 -4.14094806e-01
-6.83048308e-01 -9.20260608e-01 -5.19012809e-01 8.91208827e-01
-6.81381226e-02 -3.15898687e-01 -9.64671016e-01 -6.66841328e-01
1.27456352e-01 -5.04923165e-02 -7.01206625e-01 3.25290769e-01
-5.41188717e-01 -1.02372158e+00 7.58513331e-01 4.68268633e-01
6.16169751e-01 -9.87388968e-01 -5.36911130e-01 -2.13040471e-01
6.77914098e-02 -1.42049778e+00 -5.27648985e-01 -3.30315083e-01
-5.71903288e-01 -1.26277006e+00 -8.26231837e-01 -8.29068303e-01
6.97112679e-01 6.43429577e-01 1.11652946e+00 2.81107605e-01
-1.04018867e+00 5.52716613e-01 -1.50372043e-01 -6.26443863e-01
-2.65955720e-02 -4.54652756e-01 4.29881215e-01 -1.59445032e-02
6.13093257e-01 2.36682557e-02 -8.06613266e-01 5.59014499e-01
-6.71933711e-01 -5.89035928e-01 3.19641501e-01 5.64920962e-01
-1.64336711e-02 -2.96168309e-02 5.14950454e-01 -6.45333350e-01
3.48281890e-01 -2.93117017e-01 -9.13310707e-01 7.16238692e-02
-2.75205612e-01 -5.65782309e-01 1.23379715e-02 -2.81806469e-01
-1.34337938e+00 3.31754029e-01 -1.44657835e-01 -3.03623497e-01
-5.52595735e-01 -1.24136761e-01 -3.27231854e-01 -4.50770766e-01
6.63098872e-01 -1.63452774e-01 1.49623305e-01 -6.70626938e-01
3.00175119e-02 9.27269578e-01 5.03263533e-01 -2.37880483e-01
9.32698369e-01 7.57067919e-01 -3.09030443e-01 -1.20137930e+00
-6.89665079e-01 -6.10334814e-01 -6.04438424e-01 -4.86925155e-01
5.37736773e-01 -1.01061070e+00 -4.05366153e-01 8.89858782e-01
-1.21500301e+00 5.09392954e-02 3.59902978e-01 9.32567269e-02
1.81036424e-02 5.70503891e-01 -5.88698983e-01 -1.08614445e+00
-2.04899505e-01 -1.35765135e+00 1.24098921e+00 4.17011291e-01
-3.64632793e-02 -7.91055560e-01 -1.41012594e-01 3.89003545e-01
6.40666306e-01 3.16669494e-01 1.05999298e-01 -1.79909229e-01
-4.13942814e-01 -1.85568839e-01 -3.58515322e-01 2.43080512e-01
4.11310226e-01 4.51401770e-01 -1.46681798e+00 -9.18871820e-01
2.16128267e-02 -3.41046721e-01 7.75571704e-01 3.58744919e-01
7.96617925e-01 7.70539939e-02 -5.43297648e-01 5.45114517e-01
1.27824903e+00 1.52686909e-01 6.60424769e-01 2.11172849e-01
4.01283503e-01 8.02693307e-01 6.68141305e-01 4.35528785e-01
-4.20477130e-02 1.00454116e+00 5.60448527e-01 -9.16865915e-02
-5.73694110e-01 1.84236899e-01 6.14781857e-01 1.29715681e-01
-1.72128484e-01 -9.85959843e-02 -7.13628173e-01 4.44001675e-01
-1.38848138e+00 -1.10462034e+00 -1.65531188e-01 2.27569318e+00
5.57289422e-01 -2.66400516e-01 4.39009458e-01 -8.73517692e-02
9.89620447e-01 1.03460357e-01 -4.57883269e-01 -1.05812877e-01
-2.17915922e-01 9.54072699e-02 1.76591426e-01 3.88409674e-01
-1.50695169e+00 1.05455065e+00 7.27656984e+00 5.80246031e-01
-1.15630257e+00 7.15311393e-02 5.27938485e-01 -1.40094995e-01
5.14516473e-01 -2.17913643e-01 -1.32383609e+00 4.58412617e-01
5.40917575e-01 2.63381690e-01 4.67236072e-01 7.89882302e-01
2.06392705e-01 -2.43925646e-01 -1.08734143e+00 1.24056637e+00
6.29091561e-01 -7.77913868e-01 -5.12255073e-01 1.36227220e-01
7.04162717e-01 -2.77557294e-03 2.89266795e-01 1.53188765e-01
3.41607793e-03 -1.00065124e+00 4.59285557e-01 1.34075537e-01
9.86944616e-01 -4.67123538e-01 5.82208991e-01 -1.11452065e-01
-1.16610289e+00 -5.95012978e-02 -6.06504917e-01 -1.61928564e-01
-2.01609448e-01 4.43631083e-01 -6.11396611e-01 3.65771383e-01
9.98682022e-01 6.64288461e-01 -9.46238279e-01 1.33625174e+00
-7.00664669e-02 3.89768243e-01 -2.56786913e-01 8.11656415e-02
-2.06615195e-01 8.83570462e-02 4.57904518e-01 1.51181841e+00
2.93080360e-01 -1.63389981e-01 5.82200550e-02 5.08544326e-01
-1.74193069e-01 -1.08062573e-01 -6.78316593e-01 4.29924168e-02
5.01273572e-01 1.70432913e+00 -6.45475030e-01 -4.90625612e-02
-7.55648255e-01 1.17748749e+00 1.35385126e-01 4.07964587e-01
-8.56153846e-01 -1.37549415e-01 1.24498641e+00 2.53174566e-02
1.95657045e-01 -4.30455916e-02 2.78773487e-01 -1.09927630e+00
1.08404718e-01 -1.04955852e+00 4.31138992e-01 -4.90591615e-01
-1.39518631e+00 5.85988820e-01 2.79579908e-02 -1.06475127e+00
-4.86229025e-02 -9.63188052e-01 -5.42214513e-01 7.68449426e-01
-1.65592480e+00 -1.08713686e+00 -5.10172188e-01 5.12757778e-01
8.34117234e-01 -3.64739925e-01 8.32523108e-01 6.72646344e-01
-1.01431310e+00 7.21507967e-01 3.06617618e-02 2.79426903e-01
1.35788703e+00 -8.70782912e-01 4.41031188e-01 1.09145021e+00
-3.74405012e-02 6.89989686e-01 8.86943102e-01 -4.89400774e-01
-1.70964372e+00 -1.03372228e+00 7.09965289e-01 -7.27803707e-01
5.63377202e-01 -6.38600588e-01 -5.69969535e-01 6.35998249e-01
4.20964092e-01 4.25123274e-01 3.88999552e-01 -8.31390694e-02
-2.81400919e-01 -2.87303150e-01 -1.47626030e+00 5.06990492e-01
1.21002042e+00 -3.93738687e-01 -3.51642281e-01 5.63750863e-01
2.00241841e-02 -5.54688931e-01 -3.73146981e-01 4.91361916e-01
6.82156503e-01 -1.20955718e+00 1.06610334e+00 -2.81201839e-01
-1.09055452e-01 -4.57312673e-01 -5.02053238e-02 -9.92325842e-01
-3.01737010e-01 -8.37165594e-01 -1.99311540e-01 1.68007112e+00
-1.71242300e-02 -6.15681469e-01 7.64571965e-01 6.68550074e-01
1.31482482e-01 -3.91951621e-01 -9.42870021e-01 -9.47569549e-01
-1.99161619e-01 -9.18091387e-02 3.07165116e-01 9.07296300e-01
-3.11129779e-01 6.28673434e-02 -5.85135818e-01 3.84046882e-01
9.04913127e-01 -5.01935296e-02 7.86598742e-01 -1.09406543e+00
1.44403428e-01 -2.61828363e-01 -6.45135939e-01 -7.21353173e-01
-4.28962633e-02 -3.51905435e-01 -6.38154671e-02 -1.19829917e+00
5.10477662e-01 4.88142185e-02 -3.60104702e-02 4.22593951e-01
-2.13898763e-01 7.47637689e-01 1.87101871e-01 7.68819228e-02
-5.43635190e-01 1.33684009e-01 8.63849163e-01 -9.82948318e-02
2.18205795e-01 -2.50354022e-01 -5.76427519e-01 7.45627105e-01
6.90977514e-01 -2.04456016e-01 -1.29937306e-01 -5.28877079e-01
-4.56148297e-01 -3.31667364e-01 3.53188962e-01 -1.25403678e+00
2.75308102e-01 9.55296867e-03 9.44590092e-01 -4.93244708e-01
6.88029408e-01 -7.42570341e-01 3.50890420e-02 4.54243481e-01
2.69251466e-01 1.63059935e-01 5.81495047e-01 3.50632876e-01
-1.15060270e-01 -1.75652787e-01 1.10075450e+00 -1.24952123e-01
-1.02364576e+00 3.24825704e-01 -1.86741576e-01 -9.28129703e-02
1.27063751e+00 -3.46937448e-01 -6.22045815e-01 -1.57803014e-01
-4.34653312e-01 1.19810760e-01 6.56007648e-01 7.78170884e-01
5.58550715e-01 -1.01247096e+00 -8.66761923e-01 3.35736334e-01
1.38969636e-02 -6.01352274e-01 2.35916987e-01 7.29472935e-01
-4.33357209e-01 3.52888852e-01 -3.88603449e-01 -5.72632551e-01
-1.87877882e+00 4.89110231e-01 3.28113228e-01 3.43279660e-01
-2.24270359e-01 1.08176184e+00 6.24289550e-02 -2.44771093e-01
4.76128697e-01 2.77280450e-01 -1.70135602e-01 2.09445566e-01
1.03525746e+00 3.87634516e-01 1.91835612e-01 -9.05721903e-01
-7.89740980e-01 5.36080539e-01 -1.82313502e-01 2.63265371e-01
1.16229284e+00 -1.18079677e-01 -2.63324499e-01 -1.27591223e-01
1.03652418e+00 1.83262452e-01 -1.41086161e+00 1.68827400e-02
6.93242550e-02 -1.00209081e+00 -1.25819966e-01 -1.01290905e+00
-1.26222241e+00 6.58217132e-01 1.22488439e+00 -2.52538044e-02
1.25615621e+00 3.93458232e-02 3.07930380e-01 4.81138900e-02
3.51296365e-01 -1.06489897e+00 2.01977119e-01 4.29625750e-01
1.15017200e+00 -1.51674306e+00 2.82126039e-01 -7.60690451e-01
-1.89630643e-01 1.10622227e+00 8.59542727e-01 3.44138622e-01
4.84983504e-01 6.51689827e-01 2.62327343e-01 -5.74449413e-02
-4.57700759e-01 -4.11946118e-01 7.87682384e-02 9.47598398e-01
7.43654966e-01 -3.38515908e-01 -8.08603782e-03 1.27764540e-02
1.70549661e-01 3.21538094e-03 2.82185912e-01 1.05941236e+00
-3.82091910e-01 -1.18938816e+00 -9.18255627e-01 2.27938667e-01
-6.71138704e-01 3.63217667e-02 -7.20635712e-01 7.62551904e-01
1.98874265e-01 1.31670737e+00 -7.03169405e-02 -1.25220895e-01
3.91051710e-01 -4.26358022e-02 6.88885391e-01 -6.27492309e-01
-3.25006157e-01 1.71554819e-01 1.92738175e-01 -6.40066385e-01
-4.75452691e-01 -9.70026851e-01 -6.27059996e-01 -4.39303696e-01
-4.17149246e-01 -3.48809689e-01 6.41168237e-01 6.14243031e-01
3.22997034e-01 2.11891219e-01 3.79408836e-01 -1.22615576e+00
-5.17713785e-01 -1.17993414e+00 -4.48875248e-01 2.91144639e-01
3.90372068e-01 -9.34129775e-01 -5.25152385e-01 1.32251874e-01] | [13.404772758483887, 0.7397366166114807] |
07e5302a-bfe0-4f70-8546-b0df427a9da6 | approximately-optimal-binning-for-the | 2007.12463 | null | https://arxiv.org/abs/2007.12463v1 | https://arxiv.org/pdf/2007.12463v1.pdf | Approximately Optimal Binning for the Piecewise Constant Approximation of the Normalized Unexplained Variance (nUV) Dissimilarity Measure | The recently introduced Matching by Tone Mapping (MTM) dissimilarity measure enables template matching under smooth non-linear distortions and also has a well-established mathematical background. MTM operates by binning the template, but the ideal binning for a particular problem is an open question. By pointing out an important analogy between the well known mutual information (MI) and MTM, we introduce the term "normalized unexplained variance" (nUV) for MTM to emphasize its relevance and applicability beyond image processing. Then, we provide theoretical results on the optimal binning technique for the nUV measure and propose algorithms to find approximate solutions. The theoretical findings are supported by numerical experiments. Using the proposed techniques for binning shows 4-13% increase in terms of AUC scores with statistical significance, enabling us to conclude that the proposed binning techniques have the potential to improve the performance of the nUV measure in real applications. | ['György Kovács', 'Attila Fazekas'] | 2020-07-24 | null | null | null | null | ['template-matching', 'tone-mapping'] | ['computer-vision', 'computer-vision'] | [ 6.58142447e-01 -1.43125817e-01 -3.01796436e-01 -3.55630964e-01
-7.37999380e-01 -3.16049248e-01 7.14474380e-01 7.92829618e-02
-3.91247630e-01 7.29062974e-01 1.97009757e-01 -2.88802713e-01
-7.52757370e-01 -6.76816106e-01 -3.39949697e-01 -7.11816847e-01
-3.33775520e-01 2.42942557e-01 3.03382844e-01 -1.33732513e-01
7.71848261e-01 5.60903907e-01 -1.81085300e+00 -3.49470489e-02
1.02709150e+00 1.25443494e+00 1.72582895e-01 5.87444782e-01
-8.38294253e-02 2.46035039e-01 -8.33758175e-01 -7.61549592e-01
7.20012665e-01 -6.44047439e-01 -5.70224941e-01 6.79527298e-02
4.11772221e-01 3.03764582e-01 -6.49703071e-02 1.32378411e+00
5.09201944e-01 5.33427835e-01 9.84550893e-01 -1.25397110e+00
-6.20569170e-01 4.14992869e-01 -6.36823535e-01 4.86569196e-01
4.96592283e-01 -2.37057045e-01 9.51798499e-01 -7.25515306e-01
1.72084853e-01 1.08592999e+00 7.50867367e-01 1.65549129e-01
-1.20432150e+00 -5.79342723e-01 -6.41708016e-01 4.99307334e-01
-1.66143870e+00 -2.39044726e-01 5.81299961e-01 -4.15078908e-01
8.53323460e-01 9.93536055e-01 2.11047128e-01 3.71326625e-01
7.29585290e-01 1.31085649e-01 1.61041903e+00 -8.71427357e-01
-2.50810962e-02 2.58556962e-01 4.04147655e-02 4.27851111e-01
3.91654849e-01 4.88461077e-01 -5.23453116e-01 -8.41303468e-02
7.01259613e-01 -3.57035309e-01 -2.84530878e-01 -3.23596448e-01
-1.18297946e+00 7.97097087e-01 2.96272457e-01 5.23815930e-01
-1.58921614e-01 -7.04600587e-02 2.68565804e-01 5.54530442e-01
3.34633917e-01 3.04581821e-01 1.90551266e-01 -7.93960169e-02
-1.24850619e+00 -3.89956236e-02 5.06282449e-01 9.57045197e-01
5.99729598e-01 -8.57872367e-02 -2.35447198e-01 9.67573106e-01
1.64070457e-01 4.32722628e-01 6.82281137e-01 -8.01416636e-01
2.36324728e-01 2.72008926e-01 -3.45893539e-02 -1.33669782e+00
-7.27240816e-02 -2.59286225e-01 -9.51047421e-01 2.95255035e-01
3.26109380e-01 4.96820956e-01 -8.44945312e-01 1.57672513e+00
1.17444232e-01 1.34042427e-01 -2.88391829e-01 7.77895749e-01
4.94834691e-01 3.92396182e-01 -1.28943741e-01 -5.09943068e-01
1.16174960e+00 -4.54954892e-01 -9.09009993e-01 1.87167063e-01
-1.51424527e-01 -1.27946627e+00 7.82349288e-01 2.77403325e-01
-1.02595425e+00 -7.21492290e-01 -1.30147588e+00 2.99861819e-01
-4.13275480e-01 -4.81088132e-01 3.74589175e-01 1.12987983e+00
-1.18057597e+00 6.84184372e-01 -3.52661312e-01 -6.17752910e-01
-2.07398720e-02 6.07146144e-01 -4.07472193e-01 1.20927647e-01
-1.05340278e+00 1.19423676e+00 2.79443711e-01 2.94025149e-02
-1.80284917e-01 -2.70140320e-01 -6.34757996e-01 -1.32316858e-01
-1.90060422e-01 -4.01953727e-01 6.65109098e-01 -7.61007667e-01
-1.29456580e+00 1.04440379e+00 -3.46613348e-01 -5.16271412e-01
4.78202164e-01 2.25490481e-01 -4.90143150e-01 -3.99235487e-02
-1.76175907e-02 5.65623641e-01 8.04105103e-01 -8.64250779e-01
-4.55843210e-01 -7.45531768e-02 -2.89240748e-01 8.13277215e-02
-2.66377836e-01 1.30208105e-01 -7.64128864e-02 -1.03322518e+00
6.61378205e-01 -6.01165593e-01 -1.29645020e-01 -1.60544306e-01
-1.04821093e-01 -2.29866840e-02 3.20625931e-01 -6.41094089e-01
1.53253889e+00 -1.99110889e+00 -1.70431495e-01 6.97097838e-01
-9.60443616e-02 -2.81409901e-02 -3.85872684e-02 5.05574524e-01
-3.49195898e-01 -7.12089539e-02 -4.27150965e-01 2.29550272e-01
1.15262985e-01 1.97754711e-01 -5.35486266e-02 7.72316039e-01
-1.57907054e-01 7.09147751e-01 -4.71854925e-01 -8.32531631e-01
5.00089407e-01 2.05578312e-01 -3.59049588e-01 1.03421606e-01
5.27697682e-01 1.22979484e-01 3.55999619e-01 4.71203268e-01
1.14962757e+00 1.57317474e-01 2.20107054e-03 -4.52829868e-01
-1.99436456e-01 -1.05815455e-01 -1.40906048e+00 1.06161940e+00
-1.57637805e-01 1.09058511e+00 -4.25240219e-01 -1.20904040e+00
1.29491353e+00 1.75138146e-01 5.04769742e-01 -1.00427091e+00
2.82290697e-01 2.99217314e-01 1.42855629e-01 -3.19732487e-01
6.42420232e-01 -5.23190618e-01 6.54768348e-02 1.01571023e-01
1.06685288e-01 2.24038482e-01 9.75986794e-02 -2.76211083e-01
8.17909718e-01 -4.09264863e-01 8.89095902e-01 -7.73201823e-01
8.37727606e-01 -2.41407186e-01 3.85754555e-01 9.34906185e-01
-5.91983557e-01 5.94953418e-01 -2.35597435e-02 7.31512234e-02
-1.00506198e+00 -1.26137793e+00 -7.06100941e-01 6.50649548e-01
5.04021764e-01 2.85055488e-01 -7.14963675e-01 -1.49098262e-02
4.08696197e-02 4.93073761e-01 -7.24727273e-01 5.82251772e-02
-2.26521164e-01 -8.50517869e-01 4.35854256e-01 1.88886970e-01
5.39632976e-01 -9.01170671e-01 -5.30922830e-01 1.70255125e-01
-3.21755230e-01 -8.27536523e-01 -6.35794103e-01 2.18865231e-01
-9.86480236e-01 -8.44057024e-01 -8.99253845e-01 -6.91365421e-01
3.85409266e-01 5.04831970e-01 1.11839974e+00 -7.60115758e-02
-4.74051446e-01 2.31352523e-01 -4.49628502e-01 -2.54850835e-01
-4.02402014e-01 -4.13226575e-01 3.35565507e-02 7.60145411e-02
7.43517518e-01 -7.07442880e-01 -7.01653183e-01 9.06531453e-01
-1.07121265e+00 -1.92561343e-01 5.51793277e-01 6.75753355e-01
5.19542634e-01 1.11335069e-01 5.25015533e-01 -3.80153030e-01
6.91877186e-01 -2.46330887e-01 -6.51653647e-01 2.07712129e-01
-1.01653671e+00 1.54549748e-01 -1.36333242e-01 -4.62529808e-01
-7.72003531e-01 -4.56843168e-01 1.11362092e-01 -5.45762293e-02
7.29577839e-02 1.72093153e-01 -1.67503640e-01 -5.64325154e-01
6.80749834e-01 2.46704221e-01 7.41376132e-02 -3.32059950e-01
2.33390018e-01 8.92459691e-01 9.25961137e-01 -3.55833620e-01
8.15398991e-01 4.56219375e-01 3.65185291e-01 -7.65486896e-01
-3.75546008e-01 -8.97753775e-01 -4.64155763e-01 -3.38260710e-01
8.55458379e-01 -3.07623744e-01 -8.94030213e-01 -1.00444455e-03
-8.32689941e-01 1.74275488e-01 -8.77532959e-02 6.64730906e-01
-7.99799621e-01 7.22109973e-01 -2.47888893e-01 -1.05550313e+00
-4.30122942e-01 -9.59487438e-01 6.58439040e-01 2.97649622e-01
-3.00981343e-01 -1.06497943e+00 2.20895931e-01 2.63362259e-01
5.61063170e-01 2.70490974e-01 8.55749369e-01 -5.19096434e-01
-3.81236464e-01 -3.64555389e-01 -4.91735965e-01 3.09672058e-01
2.35525414e-01 -3.38720620e-01 -8.70504320e-01 -1.15402840e-01
2.64998347e-01 4.89383489e-01 5.93945205e-01 7.41872847e-01
7.32108295e-01 -1.41385883e-01 -1.24363050e-01 5.37106991e-01
1.81857097e+00 5.20284474e-01 1.10918736e+00 5.82071960e-01
-7.92703107e-02 5.27815700e-01 9.38964009e-01 4.85850722e-01
-2.81653106e-01 8.73122752e-01 3.88188064e-01 4.16420028e-02
-2.00326979e-01 1.54656082e-01 4.80308942e-02 9.37531650e-01
-1.40916195e-03 -3.06250781e-01 -4.57687408e-01 5.32476366e-01
-1.59358346e+00 -1.35843158e+00 -2.14224353e-01 2.62890911e+00
6.39108479e-01 2.10485458e-01 4.66420762e-02 5.36572814e-01
1.14078617e+00 9.72739533e-02 -4.26313169e-02 -8.77949774e-01
-3.77109528e-01 5.29513657e-01 7.92766333e-01 7.65597224e-01
-1.05189943e+00 4.97913748e-01 7.17781210e+00 1.22777402e+00
-9.85437632e-01 1.05882592e-01 5.43959379e-01 3.86157751e-01
-2.17183501e-01 1.13364588e-02 -3.87368113e-01 5.73961437e-01
9.67797101e-01 -4.80854690e-01 4.67146665e-01 3.18611443e-01
2.31358156e-01 -4.96286422e-01 -7.74315059e-01 1.21391070e+00
1.48962796e-01 -9.67700481e-01 1.18611520e-02 3.17542791e-01
8.54263067e-01 -3.16977978e-01 1.89353526e-01 -1.99895859e-01
-3.54497254e-01 -1.12234199e+00 4.20363814e-01 7.58277833e-01
7.24579692e-01 -8.06506753e-01 9.51536119e-01 7.05070123e-02
-1.18003201e+00 2.43239969e-01 -6.70733511e-01 6.46181926e-02
-7.38422722e-02 6.63545430e-01 -7.79492199e-01 5.37463009e-01
3.98277611e-01 -6.07365333e-02 -5.45736909e-01 1.74692667e+00
5.02712727e-01 2.92132258e-01 -2.66346931e-01 -1.88427180e-01
1.35696396e-01 -5.42976439e-01 6.42640591e-01 1.46452367e+00
6.37555063e-01 -8.88549536e-02 -4.02070791e-01 6.34912014e-01
3.53792310e-01 4.77393657e-01 -5.07721424e-01 3.00669342e-01
4.19962227e-01 1.06060743e+00 -9.72762167e-01 -2.62383640e-01
-3.56845856e-01 1.05702591e+00 -3.67442578e-01 1.97011326e-02
-9.26690280e-01 -6.15113199e-01 6.41263068e-01 1.60526738e-01
2.63235152e-01 7.19123855e-02 -7.04941452e-01 -6.61759853e-01
6.00302070e-02 -9.21032131e-01 5.73983133e-01 -6.54406130e-01
-1.25379193e+00 6.26747489e-01 2.41550028e-01 -1.84993160e+00
-2.03846335e-01 -6.90937221e-01 -3.77644539e-01 1.12843812e+00
-1.24257708e+00 -4.92928207e-01 -3.49216491e-01 3.86453450e-01
4.85391200e-01 -1.32116541e-01 6.99146926e-01 4.86014664e-01
-1.20624855e-01 8.74886692e-01 3.29399228e-01 -2.04844221e-01
8.35987926e-01 -1.28031147e+00 2.50435531e-01 9.85041916e-01
3.62891734e-01 6.55643940e-01 1.23557866e+00 -3.98663819e-01
-9.39863145e-01 -5.20700455e-01 1.11778069e+00 -1.96536064e-01
5.59469283e-01 7.73804933e-02 -7.82321513e-01 5.75202703e-03
3.20420235e-01 -2.69256175e-01 9.41258788e-01 -3.15986246e-01
-2.70620733e-01 -2.45356396e-01 -1.53431606e+00 5.11340857e-01
8.80142570e-01 -3.60892326e-01 -7.13550210e-01 1.87197939e-01
2.47125715e-01 -2.20034927e-01 -1.07664657e+00 5.75421631e-01
8.09404135e-01 -1.48975384e+00 9.65005457e-01 -5.19949347e-02
-3.30466665e-02 -3.43697786e-01 -7.83796668e-01 -8.81443620e-01
-4.63658571e-01 -7.31108963e-01 4.37241852e-01 1.22158408e+00
3.77180517e-01 -6.90215051e-01 6.58650815e-01 4.51539427e-01
8.84018689e-02 -6.64470911e-01 -1.34355640e+00 -1.17959487e+00
-2.69622982e-01 -6.84744656e-01 5.83430111e-01 7.61302531e-01
2.15102077e-01 -2.58342713e-01 -4.43418592e-01 4.66248728e-02
1.00218046e+00 1.13050275e-01 4.52562094e-01 -1.16741550e+00
-3.27254891e-01 -7.64107645e-01 -1.21371853e+00 -8.49338710e-01
-2.95652330e-01 -7.10187733e-01 7.07517639e-02 -1.21168077e+00
4.66700763e-01 -2.56784469e-01 -7.54151404e-01 -3.65466535e-01
-1.34984240e-01 7.28865683e-01 8.42212439e-02 1.54115528e-01
-1.46988422e-01 2.32419632e-02 1.03246903e+00 -1.30690306e-01
8.58792104e-03 1.17896982e-01 -5.84132731e-01 3.44892591e-01
5.42442262e-01 -4.49661255e-01 -2.60900915e-01 9.58257020e-02
-1.96224123e-01 8.54979157e-02 1.89029589e-01 -1.06832087e+00
9.13629308e-02 -3.48053724e-01 3.36149931e-01 -6.41782045e-01
2.14556694e-01 -8.61826777e-01 4.66631323e-01 4.26569194e-01
-2.79391289e-01 3.77899408e-01 1.12413652e-01 4.60557729e-01
-4.15874898e-01 -5.90323627e-01 1.10348082e+00 1.91121101e-01
-7.14572132e-01 -8.41514468e-02 -1.48114175e-01 -2.51052707e-01
9.21771646e-01 -9.68110442e-01 -8.59079286e-02 -6.84614539e-01
-5.64150274e-01 -4.58831489e-01 3.15218538e-01 2.12349892e-01
6.08124852e-01 -1.60981619e+00 -4.94453132e-01 3.27035725e-01
1.70675129e-01 -1.13199699e+00 1.00342982e-01 1.01701093e+00
-6.00393116e-01 5.42982876e-01 -4.45779294e-01 -7.47997105e-01
-1.50809860e+00 6.93172812e-01 2.21869111e-01 -5.43984435e-02
-1.65316164e-01 6.97161138e-01 1.69651195e-01 2.03734443e-01
2.21562803e-01 -1.61474705e-01 -2.79521067e-02 -1.00269176e-01
4.92543668e-01 6.78546727e-01 3.22589815e-01 -8.85620713e-01
-4.37336385e-01 7.15072453e-01 3.78792316e-01 -5.42277396e-01
8.84080648e-01 -4.15142298e-01 -3.60299647e-01 3.44424129e-01
1.43327463e+00 -6.26323968e-02 -5.38835168e-01 -1.72731251e-01
4.47335899e-01 -8.74968410e-01 -1.80529222e-01 -5.24646580e-01
-6.79475069e-01 7.34062254e-01 1.28228092e+00 4.05372918e-01
1.54536641e+00 -2.46828750e-01 5.55170834e-01 1.78268608e-02
2.85820872e-01 -9.93591487e-01 -7.32928589e-02 1.74652681e-01
9.08753872e-01 -1.05063641e+00 9.99494344e-02 -5.36099851e-01
-3.07895690e-01 9.86065567e-01 1.55841261e-01 -2.01726854e-01
6.33672416e-01 2.90026218e-01 2.13241085e-01 2.05737397e-01
-2.43625596e-01 -4.56386894e-01 8.50577652e-01 8.31887424e-01
6.78665042e-01 2.55939871e-01 -1.06563032e+00 5.62436022e-02
-3.89111489e-01 -4.52149510e-01 1.87790826e-01 5.02308547e-01
-8.32113206e-01 -1.25016999e+00 -9.40103173e-01 6.00112438e-01
-5.71743548e-01 -1.89464897e-01 -1.94013968e-01 8.32855046e-01
1.74172089e-01 1.07809019e+00 1.41857877e-01 -6.58895016e-01
2.37776592e-01 -5.13401590e-02 6.79360688e-01 3.83727588e-02
-3.07179272e-01 3.37411851e-01 -2.67024785e-01 -4.40143794e-01
-6.98102534e-01 -5.27930558e-01 -6.11094654e-01 -4.48371470e-01
-6.31266952e-01 3.02825361e-01 6.17390990e-01 8.97043824e-01
-5.72254024e-02 8.81667584e-02 7.46332407e-01 -5.27598977e-01
-4.06527609e-01 -1.14078021e+00 -6.02258921e-01 5.24923682e-01
1.01274751e-01 -9.00842905e-01 -4.05175984e-01 1.09137245e-01] | [11.746587753295898, -1.9286547899246216] |
2e9174f6-5402-4fd2-9da6-0fccac184f2c | stacking-of-hyperparameter-tuned-models-for | 2306.10077 | null | https://arxiv.org/abs/2306.10077v2 | https://arxiv.org/pdf/2306.10077v2.pdf | Stacking of Hyperparameter Tuned Models for Tagging Coding Problems | Coding problems are problems that require a solution in the form of a computer program. Coding problems are popular among students and professionals as it enhances their skills and career opportunities. An AI system that would help those who practice coding problems would be highly useful and there is a huge potential for such a system. In this work, we propose a model which uses stacking of hyperparameter tuned boosting models to achieve impressive metric scores of 77.8% accuracy and 0.815 PR-AUC on the dataset that was scraped from Codeforces and Leetcode. We open source the dataset and the models developed for this work. | ['P. Shunmugapriya', 'S. Lakshmana Pandian', 'Sathya Krishnan TS'] | 2023-06-16 | null | null | null | null | ['coding-problem-tagging'] | ['natural-language-processing'] | [-2.75208503e-01 1.52209714e-01 -2.69591391e-01 -4.90385711e-01
-5.16166389e-01 -4.00717288e-01 3.16681355e-01 8.74299258e-02
-6.91450313e-02 5.13802409e-01 1.09476246e-01 -7.15749979e-01
-1.32257670e-01 -5.25305033e-01 -3.64390224e-01 -1.97071090e-01
1.13571748e-01 2.85688072e-01 8.43867734e-02 -4.88151491e-01
1.00420022e+00 5.56699820e-02 -1.52423596e+00 4.88899589e-01
1.48763192e+00 7.07884908e-01 2.95503944e-01 6.76527083e-01
-2.02700078e-01 1.30561340e+00 -5.35941601e-01 -6.74856424e-01
3.81285965e-01 -1.40998051e-01 -9.21049714e-01 -4.73172277e-01
2.16014341e-01 2.36843064e-01 1.73658341e-01 1.00353158e+00
1.58368140e-01 -1.03744335e-01 4.13524747e-01 -1.43778729e+00
-8.34873557e-01 5.48692226e-01 -8.26742768e-01 1.46437839e-01
6.62917852e-01 1.98645905e-01 9.62841570e-01 -5.94806790e-01
3.92094374e-01 8.90779018e-01 8.95404220e-01 4.67099458e-01
-1.22587931e+00 -9.89168942e-01 -4.16293859e-01 4.36609030e-01
-1.13161385e+00 -9.56249386e-02 7.96521842e-01 -1.05665004e+00
1.14689958e+00 1.68008402e-01 9.75596249e-01 6.24103367e-01
5.37986219e-01 7.18821943e-01 1.22702491e+00 -8.01078558e-01
4.74301964e-01 4.63106513e-01 4.93495911e-01 7.31392682e-01
-1.42674064e-02 -7.48551860e-02 -2.16036335e-01 -3.29549722e-02
5.62628925e-01 1.49561837e-01 4.14590091e-02 -3.54885638e-01
-8.72200370e-01 1.18950391e+00 6.15365267e-01 5.31293333e-01
-2.07519621e-01 1.80297017e-01 2.64692158e-01 4.99991924e-01
2.26648003e-01 1.08660412e+00 -2.98501104e-01 -8.41566265e-01
-9.64475453e-01 3.79771769e-01 8.48141968e-01 9.86110926e-01
4.02995974e-01 -1.20478980e-01 7.70556629e-02 1.04801667e+00
2.80062705e-01 -1.07665760e-02 7.70247340e-01 -9.89386261e-01
3.93967092e-01 1.22571266e+00 -1.97409883e-01 -9.94088054e-01
-3.02104324e-01 -3.30102354e-01 -4.04540181e-01 2.69830674e-01
1.91004500e-01 -7.17127174e-02 -9.09968019e-01 1.38205028e+00
-2.39857119e-02 7.89313391e-03 -2.07595527e-01 7.68479168e-01
7.78890014e-01 6.50782287e-01 1.88377902e-01 2.37554058e-01
1.21391714e+00 -1.21192396e+00 -4.77065027e-01 -3.88866007e-01
7.85308599e-01 -8.10028434e-01 1.13442397e+00 8.78152251e-01
-1.16232073e+00 -5.69459200e-01 -9.04493451e-01 -1.04319751e-01
-2.19113857e-01 9.70816389e-02 1.11514211e+00 9.29012060e-01
-1.08749378e+00 2.62839615e-01 -4.58788604e-01 -3.32134426e-01
5.93185484e-01 3.09862643e-01 -1.36940420e-01 -1.84354812e-01
-7.94954836e-01 1.17297697e+00 1.69419736e-01 -3.77325863e-01
-5.89587748e-01 -8.10930133e-01 -4.25485641e-01 2.27293342e-01
-6.16186764e-03 -2.74284244e-01 1.38766050e+00 -9.17762220e-01
-1.05678225e+00 9.66257870e-01 2.13763222e-01 -4.55940485e-01
4.92953986e-01 7.93104023e-02 -3.75668183e-02 -3.29392701e-01
1.11955069e-01 5.58460534e-01 5.68565369e-01 -7.08827972e-01
-7.05703616e-01 -2.53224134e-01 2.28824824e-01 -1.44151896e-01
-3.78425986e-01 2.34956890e-01 -2.72734910e-01 -6.00859642e-01
-1.58252679e-02 -1.14424527e+00 -2.52071142e-01 -2.85705507e-01
3.24617028e-02 -3.62638235e-01 3.62485915e-01 -9.12650049e-01
1.59882033e+00 -2.06007075e+00 2.60395110e-01 1.69240624e-01
1.88749284e-01 4.76287037e-01 2.64722437e-01 4.74226326e-01
-3.29504699e-01 3.30543011e-01 -7.20394254e-02 3.13638359e-01
-5.37064709e-02 -1.72558725e-01 -9.36744139e-02 3.26434150e-02
6.16770014e-02 6.36523426e-01 -7.35903203e-01 -3.27597111e-01
4.94451672e-02 2.14593366e-01 -9.93327320e-01 4.11768109e-01
-8.28413013e-03 -8.62941295e-02 -3.74570757e-01 6.57660007e-01
5.22215962e-01 -3.60499978e-01 3.55477072e-02 6.94295824e-01
6.13861308e-02 1.26133442e-01 -8.39716196e-01 1.70312786e+00
-7.66524434e-01 7.73090661e-01 -1.97148487e-01 -1.35358393e+00
1.37468398e+00 1.92288905e-01 1.77177221e-01 -6.78006411e-01
1.84095781e-02 8.02670419e-02 4.63975877e-01 -9.19594884e-01
4.35918659e-01 -9.88410786e-02 -5.37730828e-02 4.64092016e-01
-1.10092118e-01 -6.18868530e-01 4.03404206e-01 2.60608494e-01
1.54877138e+00 -4.63056304e-02 5.21165490e-01 -4.44977909e-01
4.89759952e-01 1.88748568e-01 2.44277790e-01 5.01544118e-01
-2.69678146e-01 3.87999147e-01 7.26191819e-01 -7.41225183e-01
-1.15669489e+00 -4.24442172e-01 -3.02322894e-01 1.18581951e+00
-4.27150339e-01 -6.24255776e-01 -5.97567260e-01 -4.72042918e-01
1.51339009e-01 6.45331204e-01 -7.19372213e-01 -1.63740829e-01
-1.82534352e-01 -2.24560887e-01 1.25519574e-01 5.07864594e-01
4.26446676e-01 -1.20168030e+00 -7.27534354e-01 2.05776677e-01
-5.61350510e-02 -5.41957438e-01 3.95367108e-02 4.09656048e-01
-8.06313932e-01 -9.71780837e-01 -5.28532386e-01 -1.05970681e+00
6.38583064e-01 3.06015730e-01 1.14140177e+00 6.33131802e-01
-4.69436467e-01 -1.05249956e-01 -6.41838729e-01 -7.71882653e-01
-3.06693792e-01 3.55914176e-01 -4.07218397e-01 -7.14001834e-01
7.37027347e-01 -5.45380116e-01 -3.08069974e-01 1.99400238e-04
-3.81688982e-01 4.21855092e-01 5.68273365e-01 9.47425246e-01
-2.47850880e-01 -1.19682021e-01 5.53419828e-01 -1.07186091e+00
9.62805390e-01 -9.09150839e-01 -7.59433806e-01 3.58010471e-01
-7.66856611e-01 -1.48831969e-02 5.68562269e-01 -2.85891265e-01
-8.33984315e-01 1.52034804e-01 -2.25657225e-01 -7.31569249e-03
6.98287189e-02 7.42413461e-01 5.61397731e-01 -2.37368479e-01
1.16896200e+00 -2.51624376e-01 -2.58367598e-01 -2.28533477e-01
-2.11341530e-01 1.08816290e+00 1.05702884e-01 -8.19257677e-01
5.49377441e-01 -3.23820561e-01 -3.40145051e-01 -5.22485137e-01
-6.07333481e-01 -5.57892263e-01 -7.37974122e-02 -3.84018838e-01
5.51573038e-01 -8.16062689e-01 -6.78179204e-01 2.31908392e-02
-1.05458307e+00 -1.67421043e-01 2.10796416e-01 2.40582749e-01
-4.13901299e-01 -1.78665623e-01 -1.59885257e-01 -8.79802048e-01
-2.17606410e-01 -1.31752574e+00 4.41961557e-01 5.73662400e-01
-6.59784198e-01 -7.94771135e-01 1.33936167e-01 1.08416557e+00
8.06462467e-01 2.20369026e-01 8.41274440e-01 -5.25714040e-01
-4.95148599e-01 -2.40772501e-01 -2.78954148e-01 2.34977409e-01
-3.75293732e-01 3.20456803e-01 -9.56207216e-01 -1.22461095e-01
-5.57269575e-03 -6.88940585e-01 5.31940103e-01 4.67273891e-01
1.48657084e+00 -1.27986401e-01 -1.51051134e-01 5.33154249e-01
1.54761696e+00 6.77128375e-01 5.79948187e-01 5.99869370e-01
4.52874005e-01 6.34880960e-01 4.04976159e-01 4.53948855e-01
3.36626917e-01 6.74163699e-01 1.99563429e-01 1.76782072e-01
2.69967109e-01 -1.17648229e-01 1.11258313e-01 8.57591569e-01
-1.37599841e-01 3.38928849e-01 -1.82415378e+00 7.53085673e-01
-1.91598535e+00 -8.25299799e-01 -4.61566031e-01 1.64226091e+00
8.40792954e-01 2.41595253e-01 1.48656487e-01 3.44352484e-01
4.60432589e-01 -3.84541124e-01 -2.78269023e-01 -1.27289629e+00
6.44200027e-01 4.28822339e-01 1.72130112e-02 2.07088798e-01
-7.07215250e-01 7.28014886e-01 6.86506701e+00 6.73223138e-01
-8.54603827e-01 3.22650932e-02 7.27838695e-01 1.76208466e-01
-1.15488730e-01 1.08822241e-01 -4.17097837e-01 7.73203373e-01
1.12274027e+00 -5.64110160e-01 6.95985556e-01 1.51211822e+00
-1.28888696e-01 -5.16982198e-01 -9.07153904e-01 1.00122309e+00
-4.18718494e-02 -1.33260107e+00 -6.43454790e-01 9.24326107e-02
9.72491384e-01 -1.19004041e-01 3.71055938e-02 7.98737884e-01
9.47414577e-01 -1.30165291e+00 4.46379215e-01 2.85822630e-01
4.45930183e-01 -7.31490493e-01 8.81527722e-01 5.68841219e-01
-6.17443919e-01 -5.81667840e-01 -3.62241268e-01 -8.38561177e-01
-5.00842869e-01 5.73097110e-01 -1.13383126e+00 -1.25366524e-01
7.77095854e-01 6.79430246e-01 -8.92081857e-01 1.54116619e+00
-3.51307958e-01 8.43452334e-01 -4.14606631e-02 -4.95684445e-01
-6.53548315e-02 -8.87170956e-02 -2.43117556e-01 1.14991641e+00
4.57873970e-01 2.07716614e-01 6.27703890e-02 8.30654562e-01
1.35709450e-01 2.36374423e-01 -7.55553067e-01 -5.88129163e-02
6.48189723e-01 1.25377822e+00 -6.34610951e-01 -1.43935770e-01
-5.48600852e-01 2.30434373e-01 6.50568485e-01 -2.05145463e-01
-8.33492517e-01 -5.71319699e-01 4.65028763e-01 1.00198500e-01
1.69268921e-01 1.55372947e-01 -8.83178234e-01 -1.03083920e+00
4.83574942e-02 -1.19053292e+00 1.12807624e-01 -9.44136381e-01
-1.21956050e+00 3.61375928e-01 -2.88687825e-01 -1.02406454e+00
-1.44277453e-01 -5.22888303e-01 -6.88209534e-01 7.66154706e-01
-1.01204789e+00 -1.07210827e+00 -6.37118399e-01 2.07194835e-01
4.28032100e-01 -5.16880691e-01 6.27443075e-01 1.25405923e-01
-3.72805208e-01 4.99386698e-01 6.75650267e-03 1.59183115e-01
5.21419704e-01 -1.43770123e+00 2.47197822e-01 6.91425145e-01
6.49311766e-02 9.15878534e-01 6.79170012e-01 -3.85848910e-01
-1.26891088e+00 -5.67003787e-01 9.91410673e-01 -6.85454071e-01
5.76167822e-01 -2.14171976e-01 -8.72732520e-01 5.01157820e-01
2.32142702e-01 -3.57322395e-01 8.04923058e-01 5.51736474e-01
-2.52379060e-01 -9.36329924e-03 -1.28382540e+00 1.41081393e-01
5.95884979e-01 -3.46309423e-01 -9.41235900e-01 4.24336016e-01
2.40981877e-01 -5.31376779e-01 -8.45430791e-01 -1.65770039e-01
7.51579285e-01 -1.21692312e+00 6.67387605e-01 -4.02242810e-01
1.20953429e+00 3.12282983e-03 1.18149638e-01 -1.38799977e+00
-4.84248340e-01 -4.57545877e-01 2.97236830e-01 1.08180046e+00
5.75769484e-01 -3.38238865e-01 9.08803582e-01 1.19970727e+00
-1.23673178e-01 -7.16768444e-01 -6.82325959e-01 -5.53742409e-01
3.34834158e-01 -5.81744850e-01 6.10140800e-01 1.27823699e+00
8.82263422e-01 2.72769779e-01 -1.45974085e-01 -6.49778724e-01
1.32718340e-01 -1.32573619e-02 8.84679675e-01 -1.43977547e+00
-2.61848330e-01 -4.59889054e-01 -5.18481076e-01 -4.25135344e-01
1.18467242e-01 -1.17460513e+00 -2.00710446e-01 -1.54623318e+00
5.90754926e-01 -7.71454275e-01 -4.11337540e-02 8.12697530e-01
-2.26943359e-01 -6.21323532e-04 4.55880821e-01 1.35479733e-01
-5.17711103e-01 -6.11992367e-02 9.67398584e-01 -3.01125236e-02
-9.02573839e-02 -1.14777774e-01 -1.23242998e+00 4.75805670e-01
9.89665210e-01 -6.12180829e-01 -4.12991107e-01 -4.03377175e-01
6.39801681e-01 3.33615005e-01 -1.64166227e-01 -1.29836285e+00
3.23990792e-01 -3.56408447e-01 2.85522282e-01 7.07359314e-02
-1.97010282e-02 -9.70921516e-01 3.31337214e-01 6.81048453e-01
-6.26017034e-01 4.10226434e-01 1.41897142e-01 7.51525387e-02
-1.70559421e-01 -6.73522651e-01 8.12937677e-01 -2.76452810e-01
-5.19078314e-01 -3.92309099e-01 -3.60522121e-01 8.43032748e-02
1.58782911e+00 -1.39155567e-01 -4.86165553e-01 -3.83785307e-01
-2.44334906e-01 3.91777635e-01 6.33220196e-01 6.36615813e-01
4.24370855e-01 -1.20798695e+00 -4.88085717e-01 4.46359783e-01
2.29357421e-01 -3.11929286e-01 -9.66530144e-02 4.96602684e-01
-9.08726990e-01 5.71130395e-01 -8.89679670e-01 -3.93276662e-01
-1.38595867e+00 5.20299017e-01 -1.44099191e-01 -3.41538876e-01
-2.36836731e-01 1.15982342e+00 -4.80055243e-01 -5.96765399e-01
9.94146839e-02 -4.08046901e-01 -3.20041895e-01 -1.52198419e-01
6.21612906e-01 3.44917893e-01 9.96209085e-02 -1.68368772e-01
-2.56888628e-01 2.88665444e-01 -2.66646117e-01 1.33354083e-01
1.84517324e+00 6.16873503e-01 -1.11824654e-01 1.40874445e-01
8.49981546e-01 -2.58930087e-01 -7.16232657e-01 1.43203154e-01
1.36895403e-01 -9.39499199e-01 2.98383027e-01 -1.24052942e+00
-1.04588342e+00 1.15580022e+00 5.81314087e-01 6.05418384e-01
9.33348715e-01 -2.84243017e-01 2.20383331e-01 1.87172025e-01
5.78399122e-01 -1.11627340e+00 1.36374786e-01 4.84153152e-01
8.75244677e-01 -1.57057357e+00 2.81483084e-01 -1.01564221e-01
-9.07708108e-01 1.16714144e+00 8.52352202e-01 -3.45857322e-01
5.89820683e-01 4.05694723e-01 -2.05358509e-02 -1.89314842e-01
-1.11597240e+00 2.48049006e-01 1.14559524e-01 7.25752234e-01
1.05550909e+00 2.27765858e-01 -7.50559270e-01 7.29409218e-01
-6.03958488e-01 5.66404700e-01 6.63419187e-01 1.07952595e+00
-4.45125639e-01 -1.16516936e+00 -5.40325820e-01 8.17032099e-01
-4.49093401e-01 -5.88668734e-02 -5.07297933e-01 4.96883899e-01
1.53916374e-01 9.25434768e-01 -2.45463267e-01 -7.95224845e-01
-4.35426421e-02 2.02576816e-01 2.36049190e-01 -7.45292068e-01
-1.13524401e+00 -5.44871986e-01 -4.09155488e-02 -3.58294338e-01
-7.82063752e-02 -7.08793521e-01 -9.49317038e-01 -7.52493083e-01
-4.57473278e-01 2.28150144e-01 9.98951137e-01 5.14875770e-01
2.39202857e-01 5.03552973e-01 4.97460067e-01 -4.33628708e-01
-2.63133019e-01 -1.02435255e+00 -2.48620778e-01 3.24695528e-01
-2.38548834e-02 -7.65211046e-01 -1.85098797e-01 2.83153467e-02] | [7.819547653198242, 7.703769207000732] |
a821cc37-2f18-40e7-a2b6-245f210ef14f | neural-modal-odes-integrating-physics-based | 2207.07883 | null | https://arxiv.org/abs/2207.07883v2 | https://arxiv.org/pdf/2207.07883v2.pdf | Neural modal ordinary differential equations: Integrating physics-based modeling with neural ordinary differential equations for modeling high-dimensional monitored structures | The order/dimension of models derived on the basis of data is commonly restricted by the number of observations, or in the context of monitored systems, sensing nodes. This is particularly true for structural systems (e.g., civil or mechanical structures), which are typically high-dimensional in nature. In the scope of physics-informed machine learning, this paper proposes a framework -- termed Neural Modal ODEs -- to integrate physics-based modeling with deep learning for modeling the dynamics of monitored and high-dimensional engineered systems. Neural Ordinary Differential Equations -- Neural ODEs are exploited as the deep learning operator. In this initiating exploration, we restrict ourselves to linear or mildly nonlinear systems. We propose an architecture that couples a dynamic version of variational autoencoders with physics-informed Neural ODEs (Pi-Neural ODEs). An encoder, as a part of the autoencoder, learns the abstract mappings from the first few items of observational data to the initial values of the latent variables, which drive the learning of embedded dynamics via physics-informed Neural ODEs, imposing a modal model structure on that latent space. The decoder of the proposed model adopts the eigenmodes derived from an eigen-analysis applied to the linearized portion of a physics-based model: a process implicitly carrying the spatial relationship between degrees-of-freedom (DOFs). The framework is validated on a numerical example, and an experimental dataset of a scaled cable-stayed bridge, where the learned hybrid model is shown to outperform a purely physics-based approach to modeling. We further show the functionality of the proposed scheme within the context of virtual sensing, i.e., the recovery of generalized response quantities in unmeasured DOFs from spatially sparse data. | ['Eleni Chatzi', 'Limin Sun', 'Kiran Bacsa', 'Xudong Jian', 'Wei Liu', 'Zhilu Lai'] | 2022-07-16 | null | null | null | null | ['physics-informed-machine-learning'] | ['graphs'] | [ 1.01812832e-01 5.41693211e-01 2.62823373e-01 1.22089334e-01
-2.81691104e-01 -4.15404826e-01 2.85484761e-01 -3.71726096e-01
1.17346108e-01 6.18238986e-01 1.97130844e-01 -1.67378321e-01
-8.30957651e-01 -7.78715014e-01 -1.11739075e+00 -1.14232481e+00
-2.70086914e-01 4.36749846e-01 -4.79053885e-01 -4.60574389e-01
-2.84242481e-01 7.76663840e-01 -1.35048735e+00 -2.50194907e-01
3.87528211e-01 1.02104461e+00 1.94955632e-01 6.17387652e-01
4.48928140e-02 7.97145545e-01 -1.58678859e-01 2.74370164e-01
2.08869740e-01 -2.05252543e-01 -4.69534516e-01 1.55415803e-01
-1.08193206e-02 -5.12386739e-01 -8.25123847e-01 7.71051049e-01
2.72137403e-01 3.61672312e-01 9.58513975e-01 -9.03904259e-01
-9.32658136e-01 4.67611194e-01 1.26118347e-01 -2.02582374e-01
2.59143803e-02 7.40217119e-02 8.80343080e-01 -8.79067421e-01
5.42854011e-01 1.02901220e+00 8.74895811e-01 5.29379368e-01
-1.67718399e+00 -3.50650325e-02 -1.87517434e-01 -1.18406467e-01
-1.41728640e+00 -3.20011884e-01 1.43837583e+00 -1.18229079e+00
7.10514545e-01 -5.74152358e-02 7.03762412e-01 1.40101683e+00
5.81348896e-01 1.81524843e-01 7.01464772e-01 -3.04363161e-01
6.45383298e-01 1.71063706e-01 7.29697496e-02 5.73874116e-01
-1.00721091e-01 7.75928378e-01 -5.43575644e-01 -4.45338160e-01
1.02982712e+00 1.04140058e-01 -4.63901013e-01 -9.04095531e-01
-9.51959491e-01 1.04237163e+00 5.54606974e-01 2.85775006e-01
-8.87426138e-01 2.87966847e-01 -5.22604398e-02 4.15876806e-01
3.24151516e-01 4.09986347e-01 -4.80618984e-01 1.90446407e-01
-8.38168323e-01 2.48452082e-01 1.06393576e+00 5.39446294e-01
9.47783828e-01 6.32053673e-01 4.05078411e-01 4.26279753e-01
6.27364695e-01 7.39710331e-01 2.23697364e-01 -1.16644585e+00
-1.83054999e-01 3.22283685e-01 2.19996393e-01 -7.93533027e-01
-3.81792217e-01 -6.34810567e-01 -1.20616889e+00 4.45610553e-01
-1.12904631e-01 -5.06759882e-01 -6.64944232e-01 2.11907387e+00
4.78880048e-01 2.71339834e-01 2.36309648e-01 1.01570427e+00
4.75409955e-01 8.97594512e-01 -3.50736499e-01 -3.27645272e-01
8.44143093e-01 -1.90320089e-01 -6.65672004e-01 2.94757038e-01
3.54249388e-01 2.37015262e-02 9.29213107e-01 9.95817035e-02
-1.04608595e+00 -7.04543054e-01 -1.00739348e+00 2.41455942e-01
-5.48172653e-01 4.02285764e-03 3.24333198e-02 1.55726761e-01
-1.19051301e+00 8.68271291e-01 -1.10836172e+00 -2.23961964e-01
-2.66487569e-01 2.66369671e-01 -3.12513471e-01 6.67099118e-01
-1.31901395e+00 6.88520312e-01 2.11454734e-01 4.86472964e-01
-1.49980009e+00 -1.08877110e+00 -7.18025208e-01 3.06792945e-01
1.93651006e-01 -1.06105173e+00 7.27564037e-01 -4.80498224e-01
-1.79272151e+00 2.60049880e-01 1.64582327e-01 -2.73305953e-01
2.30403587e-01 -1.22638322e-01 -2.85441637e-01 3.54366936e-02
-1.75819293e-01 1.92557827e-01 1.15323532e+00 -1.46722531e+00
3.50904405e-01 -1.07403725e-01 3.22323710e-01 -2.00155854e-01
-6.13108814e-01 -8.71496737e-01 3.66308659e-01 -5.92430174e-01
2.11047858e-01 -1.15646803e+00 -3.56218427e-01 2.00558737e-01
-4.92173553e-01 2.58566529e-01 9.33810949e-01 -7.51349449e-01
7.77950943e-01 -2.12632060e+00 1.22403288e+00 3.11607510e-01
3.85159403e-01 -3.72147471e-01 4.35153395e-02 7.83890843e-01
-4.05988991e-01 -2.63218582e-01 -6.81352139e-01 -2.55287915e-01
1.89439520e-01 5.12248218e-01 -7.03060985e-01 3.96694273e-01
3.02399933e-01 6.71007574e-01 -4.44974273e-01 5.79117350e-02
3.70041668e-01 8.10237706e-01 -5.15700996e-01 4.44687963e-01
-3.09859544e-01 1.02093065e+00 -6.23394489e-01 1.44011602e-01
3.73354673e-01 -3.56992751e-01 -9.60520580e-02 -6.00657225e-01
-1.34837732e-01 -3.23366612e-01 -1.24756026e+00 2.05583858e+00
-7.90503025e-01 3.23548764e-01 6.37796104e-01 -1.46631396e+00
7.12254167e-01 4.51300263e-01 1.11670995e+00 -2.19874695e-01
2.13674083e-01 9.77126956e-02 -1.23184614e-01 -7.74117231e-01
1.89455628e-01 -2.74559528e-01 -1.99934348e-01 2.01654196e-01
3.58230382e-01 -2.59259820e-01 -3.68368596e-01 3.92456204e-02
1.13547444e+00 4.79101270e-01 -2.83479970e-03 -4.62165058e-01
7.51232445e-01 1.00683317e-01 2.39158154e-01 5.26966512e-01
4.17960674e-01 1.90886319e-01 1.78294584e-01 -3.74095380e-01
-1.25678897e+00 -1.24753308e+00 -3.28065783e-01 5.20846725e-01
-2.08432540e-01 1.01764418e-01 -7.53547609e-01 1.94936022e-01
3.26712757e-01 5.49973607e-01 -9.04557288e-01 -5.33251703e-01
-4.98179823e-01 -5.03048241e-01 3.08973491e-01 2.08150968e-01
-1.00183457e-01 -8.51164818e-01 -7.35774100e-01 5.76716959e-01
2.52747744e-01 -1.01280212e+00 4.26020205e-01 3.89723718e-01
-8.97618294e-01 -7.96955228e-01 -7.07146168e-01 -3.34028661e-01
3.17033023e-01 -6.12457275e-01 7.14413524e-01 -4.97117162e-01
-2.72016466e-01 9.42071974e-01 8.14700779e-03 -2.15846226e-01
-7.61407495e-01 -4.76198345e-02 5.24975896e-01 4.62161630e-01
-4.74418849e-01 -1.36971271e+00 -3.03257644e-01 -1.61586940e-01
-1.10319638e+00 -1.55552730e-01 6.36505187e-01 1.00401723e+00
6.15175307e-01 7.36184269e-02 1.91384211e-01 -2.96909988e-01
4.31135029e-01 -7.46271968e-01 -7.15995669e-01 8.25432874e-03
-2.59713441e-01 5.11699438e-01 7.83221900e-01 -7.21701026e-01
-9.32316005e-01 2.01307550e-01 -1.12302490e-01 -9.55531120e-01
-2.92063296e-01 9.48471546e-01 -1.32750511e-01 -3.19727272e-01
5.25318921e-01 3.71321142e-01 2.28940830e-01 -7.44993567e-01
3.02453727e-01 4.16227341e-01 6.44862533e-01 -9.49269712e-01
1.12372124e+00 5.39085865e-01 6.08221531e-01 -1.17585754e+00
-4.70186442e-01 3.40780579e-02 -5.25486708e-01 -3.82580936e-01
7.42353499e-01 -9.20116484e-01 -8.36224675e-01 4.62566674e-01
-9.99555528e-01 -5.75664818e-01 -8.96324635e-01 6.72139049e-01
-8.55901062e-01 1.22071542e-01 -7.63031423e-01 -1.07879972e+00
-2.66314223e-02 -1.06247783e+00 1.25253987e+00 -1.95768312e-01
-1.70177140e-03 -1.33803380e+00 5.00285089e-01 -1.81193888e-01
5.25360703e-01 6.51654482e-01 1.11884844e+00 -1.08781666e-01
-4.76297975e-01 -1.21753126e-01 5.62079251e-01 4.08547282e-01
-1.43813863e-01 1.36773065e-02 -1.09944904e+00 -2.92046726e-01
6.13885701e-01 -1.37831360e-01 5.00227094e-01 7.91317344e-01
9.23407376e-01 -1.67670503e-01 -2.34979391e-01 6.48768246e-01
1.68507326e+00 -1.04551539e-01 3.12977843e-03 -2.57141382e-01
9.10593390e-01 7.47470438e-01 -1.38235942e-01 6.76281273e-01
1.61904857e-01 5.97398639e-01 8.72951508e-01 7.49683678e-02
1.73535988e-01 -4.37965617e-02 4.72901404e-01 1.20834398e+00
-1.26385972e-01 -2.50347666e-02 -7.79512167e-01 4.22693551e-01
-1.59337926e+00 -7.31379747e-01 -3.65136960e-03 1.82074475e+00
3.00774306e-01 -2.20098034e-01 -2.80242383e-01 1.09639265e-01
4.76860166e-01 1.24039650e-01 -1.03376126e+00 -2.34454840e-01
-1.43370897e-01 1.71762913e-01 1.82248935e-01 7.74368703e-01
-7.38941312e-01 1.76077813e-01 5.35017729e+00 2.12158263e-01
-1.34837306e+00 3.44674230e-01 -8.43468383e-02 5.15228286e-02
-3.20742786e-01 6.15401082e-02 -3.02864373e-01 3.47479165e-01
1.35598481e+00 1.13623433e-01 7.98522234e-01 7.57153571e-01
5.09473920e-01 5.06379485e-01 -1.21833229e+00 6.34792507e-01
-4.47348922e-01 -1.47849619e+00 -2.05466896e-02 6.13918841e-01
6.11951232e-01 9.29202884e-02 1.82021111e-01 1.71235964e-01
1.52151838e-01 -6.23474360e-01 1.18054473e+00 1.21281481e+00
6.36942208e-01 1.11482190e-02 4.03004259e-01 7.53874242e-01
-9.15671825e-01 -4.08600241e-01 -1.48980945e-01 -1.66185498e-01
5.59032321e-01 8.06094289e-01 4.70254160e-02 6.36851370e-01
7.07202911e-01 9.49155867e-01 2.50713408e-01 2.54689932e-01
3.83796096e-01 6.17976427e-01 -5.57291865e-01 1.58858821e-01
1.26581460e-01 -4.31082964e-01 1.04323971e+00 5.03319740e-01
6.28619373e-01 2.00197399e-01 6.03917539e-02 1.62718260e+00
2.45326027e-01 -4.35277730e-01 -7.87340879e-01 -1.65431634e-01
9.68848467e-02 1.16916263e+00 3.10738944e-02 -9.99529511e-02
-1.84358388e-01 4.55595970e-01 8.36821422e-02 8.60768139e-01
-8.89922142e-01 2.05447376e-01 7.15350032e-01 1.24221258e-01
4.24457788e-01 -3.96671921e-01 1.29919767e-01 -1.16604555e+00
7.33006671e-02 -4.07508165e-01 -8.99076983e-02 -1.03009772e+00
-1.45523584e+00 4.55062598e-01 3.32814097e-01 -1.29955494e+00
-5.65834343e-01 -1.03252184e+00 -4.66981888e-01 9.51215982e-01
-1.12577760e+00 -1.04753554e+00 -3.16669196e-01 6.41855717e-01
5.42335659e-02 -1.01639643e-01 9.94304001e-01 1.36468723e-01
-5.83120704e-01 -7.12059960e-02 5.67348123e-01 -1.88474730e-02
-7.43033886e-02 -1.01205647e+00 -8.87279585e-02 6.08589113e-01
-1.05192430e-01 6.52111232e-01 1.10911596e+00 -4.70048249e-01
-1.89413476e+00 -8.18885565e-01 -2.60431343e-03 -1.60129860e-01
1.05896008e+00 -3.62050951e-01 -8.92737389e-01 5.85814536e-01
-9.70113501e-02 1.54980853e-01 5.27014971e-01 -2.25736171e-01
9.57465097e-02 -9.91335735e-02 -9.57565963e-01 2.36573070e-01
8.73978138e-01 -1.03178823e+00 -5.95374823e-01 3.50641340e-01
6.99618876e-01 -4.18266982e-01 -1.53746200e+00 4.59364444e-01
7.07052350e-01 -5.91718078e-01 1.21807146e+00 -6.53577924e-01
4.67925489e-01 -3.28536987e-01 -5.27178705e-01 -1.32488585e+00
-5.06889939e-01 -6.27171934e-01 -9.36300337e-01 8.70748162e-01
1.69095665e-01 -6.40545428e-01 6.55656695e-01 4.67372894e-01
-4.59458917e-01 -7.36381531e-01 -1.04569376e+00 -7.32069075e-01
3.49354118e-01 -3.75923604e-01 6.17563307e-01 9.42415178e-01
-4.72663999e-01 3.87940295e-02 -2.30487287e-01 9.28849280e-01
8.06809306e-01 1.84639581e-02 4.60392386e-01 -1.47824335e+00
-8.78950238e-01 -1.75955534e-01 -4.49146360e-01 -7.49089956e-01
4.89744693e-01 -4.14487839e-01 3.35977897e-02 -1.02591467e+00
-3.58974934e-01 -1.27397701e-01 -3.96172285e-01 4.47462313e-02
5.54744005e-01 -4.12112415e-01 -9.53413919e-02 3.90400320e-01
4.48071957e-01 1.08599150e+00 8.82909358e-01 -2.94885039e-01
-2.22462490e-01 -2.30836436e-01 5.39921504e-03 5.04864573e-01
2.83752233e-01 -3.27222675e-01 -5.32190681e-01 -4.25477803e-01
2.87107944e-01 6.83880806e-01 1.12973750e+00 -1.43630791e+00
2.92614698e-01 -5.36504686e-02 1.38100358e-02 -3.52165818e-01
7.97557235e-01 -1.28380704e+00 9.29480731e-01 5.43528438e-01
-4.56706613e-01 8.47692695e-03 1.18934207e-01 8.65300357e-01
-2.64811367e-01 9.99145061e-02 2.99524516e-01 -1.73402391e-02
-5.33931196e-01 3.05155009e-01 -3.88513029e-01 -3.47601861e-01
5.33305824e-01 -2.08324224e-01 2.13370219e-01 -3.72012943e-01
-1.39607108e+00 -2.23914638e-01 3.49774063e-01 3.89615586e-03
2.64859200e-01 -1.22200704e+00 -4.92506206e-01 5.02867758e-01
2.16480158e-02 9.72027034e-02 7.21077442e-01 8.88795316e-01
-2.55583107e-01 1.77823737e-01 -2.59753883e-01 -9.39437151e-01
-2.46256247e-01 5.41353643e-01 8.00553024e-01 -3.64933349e-02
-6.46010280e-01 4.93454993e-01 3.08302432e-01 -7.51929641e-01
-1.39118701e-01 -6.75368547e-01 1.56286970e-01 -2.00738043e-01
-2.46398345e-01 2.67884046e-01 -1.30633280e-01 -1.00947583e+00
-7.04014450e-02 8.59059811e-01 8.20120096e-01 -3.24456006e-01
1.50943387e+00 -1.26301125e-01 -1.70610741e-01 9.75514114e-01
1.31490576e+00 -2.79777735e-01 -1.65943360e+00 -2.59393811e-01
-4.24186110e-01 5.75768948e-01 4.63160276e-01 -3.66213024e-01
-1.14018130e+00 1.05495381e+00 7.19363987e-01 6.68757617e-01
1.00835335e+00 1.19866863e-01 5.09021282e-01 6.19226933e-01
2.94211268e-01 -8.23785067e-01 -2.23467961e-01 3.86480123e-01
1.02149606e+00 -6.93151593e-01 -3.17142755e-01 1.84901822e-02
1.94132969e-01 1.23448503e+00 5.75035661e-02 -4.35587227e-01
1.39265537e+00 4.37162042e-01 -1.74325734e-01 -4.93805587e-01
-5.11006653e-01 2.13903010e-01 3.19734305e-01 3.86618257e-01
-1.18101940e-01 2.84293946e-02 4.75701004e-01 5.15060902e-01
-9.41921100e-02 -2.30661258e-02 3.72109860e-01 9.24305797e-01
-5.58935851e-02 -6.27238631e-01 -4.89249676e-01 9.92126912e-02
9.28126946e-02 2.82730490e-01 -5.51290810e-02 8.50196242e-01
2.21070245e-01 5.80199838e-01 -1.01623155e-01 -4.67855960e-01
4.39284682e-01 1.47228301e-01 1.38289690e-01 -2.17132896e-01
-1.15597546e-01 -2.94587546e-04 -4.09893930e-01 -5.95739663e-01
-6.73268259e-01 -7.32996643e-01 -1.02289832e+00 -2.36037821e-01
-1.95025757e-01 1.98204279e-01 8.83142531e-01 9.87189412e-01
5.87020576e-01 9.40996051e-01 5.05557954e-01 -1.35652268e+00
-9.33597326e-01 -8.75253677e-01 -9.78443801e-01 1.91058174e-01
9.59870160e-01 -1.11048830e+00 -6.08631551e-01 2.81248242e-01] | [6.531846046447754, 3.506882429122925] |
7e6a982d-ab3c-48c2-967e-c34ef0d265d8 | on-duality-of-multiple-target-tracking-and | 1610.04542 | null | http://arxiv.org/abs/1610.04542v1 | http://arxiv.org/pdf/1610.04542v1.pdf | On Duality Of Multiple Target Tracking and Segmentation | Traditionally, object tracking and segmentation are treated as two separate
problems and solved independently. However, in this paper, we argue that
tracking and segmentation are actually closely related and solving one should
help the other. On one hand, the object track, which is a set of bounding boxes
with one bounding box in every frame, would provide strong high-level guidance
for the target/background segmentation task. On the other hand, the object
segmentation would separate object from other objects and background, which
will be useful for determining track locations in every frame. We propose a
novel framework which combines online multiple target tracking and segmentation
in a video. In our approach, the tracking and segmentation problems are coupled
by Lagrange dual decomposition, which leads to more accurate segmentation
results and also \emph{helps resolve typical difficulties in multiple target
tracking, such as occlusion handling, ID-switch and track drifting}. To track
targets, an individual appearance model is learned for each target via
structured learning and network flow is employed to generate tracks from
densely sampled candidates. For segmentation, multi-label Conditional Random
Field (CRF) is applied to a superpixel based spatio-temporal graph in a segment
of video to assign background or target labels to every superpixel. The
experiments on diverse sequences show that our method outperforms
state-of-the-art approaches for multiple target tracking as well as
segmentation. | ['Mubarak Shah', 'Yicong Tian'] | 2016-10-14 | null | null | null | null | ['occlusion-handling'] | ['computer-vision'] | [ 0.13550648 -0.3242838 -0.48017615 -0.16084827 -0.70013624 -0.7055364
0.10148569 0.02421554 -0.29295355 0.69014186 -0.40473014 0.09388029
0.00939117 -0.53469694 -0.73263896 -0.9406489 0.21553609 0.6236501
1.0034462 0.2983701 0.04887026 0.5127754 -1.1872845 -0.07474953
0.82458866 1.035568 0.4808031 0.39072928 -0.32561314 0.7729848
-0.39842135 -0.1964145 0.3479986 -0.34683606 -0.66031253 0.7348401
0.61645937 -0.2803214 -0.04859476 1.4463474 0.14095278 0.24998413
0.45597917 -1.2916459 -0.17915463 0.34584656 -1.21567 0.387629
-0.08595196 0.22348058 0.64733225 -0.6517404 0.5452614 1.2563705
0.5764726 0.6488702 -1.3206028 -0.76531243 0.7149225 -0.17819452
-1.3477509 -0.30619657 0.652814 -0.68726134 -0.03610136 0.2050611
0.81719404 0.50256974 0.06165331 1.1604894 0.6986014 -0.17037518
-0.06878481 0.1549022 0.32451594 0.8189541 0.35232857 0.11661765
-0.05590824 0.03872391 1.0204567 0.23336735 -0.3445008 -0.55051255
-1.2676991 0.735068 0.2825931 0.3239777 -0.3193929 0.24989347
0.22233494 -0.4264313 0.36241457 -0.3504107 -0.34438428 0.44922334
-1.1968933 0.26329955 0.48131123 1.1476815 0.8038679 0.20606269
-0.6081229 0.53135633 0.7793823 0.67807317 0.02183462 -1.0149889
0.33605295 0.5492966 0.4357214 -1.0184377 -0.20379286 -0.5326417
-0.7609954 0.3109319 0.8211857 -0.25120607 -1.1108483 1.6395706
0.9159168 0.71113735 -0.31620845 1.0542053 0.6289798 0.61710155
0.30694643 -0.670573 1.4164757 -1.2221082 -0.92202514 -0.3761005
0.31657228 -0.9410806 0.25975773 0.13044478 -1.13113 -0.85176826
-0.60622036 0.40202472 0.00895085 0.24800816 0.36359507 0.48367906
-0.6538043 0.37021697 -0.8973501 -0.09280056 0.58597654 0.37891012
0.17118566 -0.13222261 -0.7738517 0.7545116 0.68958414 0.34052017
-1.0896921 -0.4511484 -0.73697704 -0.24422516 0.7681394 -0.60794413
1.067441 -1.0988559 -1.1600225 0.6918377 -0.3585099 -0.2627517
0.57942337 -0.18192895 -0.35476872 0.01057374 0.36166826 0.7223212
1.0930793 -1.5491266 -1.1636561 -0.2556581 -0.20337652 0.22877055
0.07402506 0.2701094 -1.1217697 -0.55063313 0.18279281 -0.8706271
-0.5791654 0.15783381 -0.50015754 -0.21009026 1.3353643 -0.75765073
1.3881627 -2.0171745 0.3434132 0.0708316 0.2593647 0.32124516
0.05883523 -0.40535438 0.21025865 -0.13691227 -0.02672338 -0.43419176
-0.2685129 0.25812367 -0.04077382 0.7651072 -0.02262884 0.83559227
-0.9847281 -1.1465766 0.494074 0.35694796 -0.14392486 0.09556884
-0.66173506 0.99706334 -0.80650014 0.8544584 0.8905921 -0.44310996
-0.1238851 -0.22885841 -0.3990382 -0.5706421 -1.7447095 1.4771922
0.15543595 0.20716877 0.45101643 -0.8361831 0.6336182 0.23377211
0.83580476 -0.24436519 0.16664422 -0.02943524 -0.09720549 -0.36501843
0.28751835 0.1563251 0.14092274 0.23549879 -0.33206317 0.24001576
0.46346632 0.05609117 0.54251176 0.5036748 0.04272294 -0.19999237
0.73051286 0.4060192 1.2755374 0.7444928 -0.47866148 0.4703912
-0.13983539 -0.45104718 -0.6362778 -0.9907635 -0.13671802 0.9254408
0.80215585 0.04118977 -0.7434415 -0.7362884 -0.10204221 0.27920747
-0.26308236 0.40361288 -0.9510616 -0.7611164 0.09878637 0.39248186
0.3972914 -0.9143251 -0.71891767 0.5678878 -0.36581814 -1.215347
-0.93057203 0.03683664 -1.0077273 -1.3229041 -0.8826553 -0.8906217
0.7811223 0.5420527 0.9174712 0.20157917 -0.19456886 0.15754184
-0.14505506 -0.25038344 -0.24452347 -0.36194617 -0.03501927 0.38337475
0.02463502 -0.13096848 -0.52224046 0.6365697 -0.8057359 0.17913567
0.5039968 0.48283634 1.0668973 0.12049742 -0.03231518 -0.66516256
-0.10975026 -0.32102916 -1.1981374 0.6390294 -0.07504088 -0.35878444
0.22085044 -0.74558586 -1.0766754 0.6733041 0.27724877 -0.9730479
-0.29157597 0.00728566 -0.22617322 -0.29670447 0.06150123 0.33108732
-0.06342781 -0.39767134 0.40528914 0.1276208 0.44600424 -0.5154573
1.0121875 0.6470975 0.05588332 -0.47307283 -1.2389804 -0.7457014
-0.95648104 -0.7601663 1.3132277 -0.9235091 -0.7482442 0.45110068
-1.4628154 -0.24145973 -0.25063738 0.5727807 -0.32479218 0.44504339
-0.498898 -0.9797235 -0.09560537 -1.4043221 1.2387627 0.71021485
0.18133882 -1.2277049 -0.1256809 0.31732565 0.08579356 0.49231702
0.36053857 -0.42856485 -1.2579603 -0.06292556 -0.28654805 0.10130256
0.16403438 0.1172213 -0.61213815 -0.42054233 0.04875234 0.05072041
0.74402183 0.9814379 0.83985955 0.02985873 -1.0062467 0.4000644
1.5782424 0.46803185 0.24317999 0.11310922 1.1249685 0.4708551
1.0123873 0.34449056 0.29182485 0.8850972 0.39329025 -0.31415445
-0.18956609 0.18027121 0.37956265 0.41138527 0.02501492 -0.2204864
-0.75093997 0.4375712 -2.1771698 -1.078576 -0.63148993 2.068594
0.5936689 0.06252285 0.42346472 -0.34810856 1.5118817 0.04646717
-0.745315 0.6403269 -0.03234062 -0.34744713 0.6762795 0.54564416
-1.4565725 1.140156 5.766024 0.97446316 -0.875376 0.2611038
0.6564035 0.03408551 0.25382745 0.14889565 -1.4754705 0.58266675
0.21446294 0.0548024 0.18411243 0.7306584 0.30566102 -0.32789153
-0.93306667 0.9007317 -0.10859285 -1.2318267 -0.15254393 0.07668474
0.8443048 -0.13809168 -0.16895822 0.03713942 0.65233564 -0.45002222
0.98686016 0.59392524 0.24375668 -0.44356298 0.2913878 0.57021105
-1.8261572 0.1148856 -0.24547367 0.39890352 0.69159836 0.6593687
-0.18966049 0.69507504 0.6352152 0.78203976 -0.36019954 1.4893789
-0.02109139 0.23261471 -0.37308145 0.31534246 0.23313558 -0.619177
0.706259 0.9908094 0.05039754 0.11952869 1.0885942 0.93485975
0.38856673 -0.16987526 -0.2166122 0.3146488 0.42171454 1.5811512
-1.2571667 -0.54439634 -0.5409875 0.7441718 0.1749735 0.4149232
-1.3176726 0.16578947 0.3175686 -0.03055259 0.6637847 -0.19390526
-0.02797364 -1.2384958 -0.04535244 -0.38865227 0.577661 -0.4353416
-1.1466151 0.3379965 -0.04355191 -1.492725 0.17929026 -0.24148054
-0.6153693 0.8497099 -1.6642232 -1.2439852 -0.3178611 0.7754144
0.6821813 0.26642412 0.16465168 0.7052714 -0.89835244 0.06354451
0.06469142 0.44499946 0.40121573 -0.9072236 -0.09885192 1.262898
0.12715496 0.24184768 0.49755454 -1.0571007 -1.0948882 -1.4995824
0.35203716 -0.41805872 0.6068367 -0.05177254 -0.82593143 0.82905436
0.0357324 0.38277373 0.31468213 -0.38977593 0.3596716 0.05727646
-0.9244947 0.35134956 0.9583031 0.23876575 -0.29478848 0.6079917
0.6988658 -0.81424946 -0.66241276 0.40761277 0.14403582 -0.7288493
1.0201429 -0.286239 -0.27073455 -0.96179837 0.06762214 -0.87509304
-0.41018608 -0.45512584 -0.35087815 1.5441576 0.05729266 -0.07128763
0.9246876 0.5752324 -0.20367049 -0.45484224 -0.82964385 -0.88922507
-0.24492387 -0.25937545 0.13273688 0.84408283 -0.8287522 0.22728738
-0.5115485 0.5839481 1.2847295 0.50531435 0.6741442 -1.456587
-0.19242845 -0.38291752 -0.0234858 -1.4658175 0.24302676 -0.6729823
0.3444231 -1.6163359 0.34104553 -0.78259915 -0.14907663 0.34882343
-0.32963428 0.09337153 0.35464883 0.43627623 -1.14363 0.32462573
1.6398817 -0.3482523 -0.29194278 0.43564504 -0.34246305 0.9332653
0.46946558 -0.8361538 0.01114735 -0.52571195 -0.62037325 0.4686471
0.4337991 -1.0280068 0.60774714 -0.48801064 0.52919924 -0.9619291
0.3775521 -1.0869904 0.28965637 0.58423334 0.17637765 -0.19282438
0.25090057 0.8513715 -0.24690707 -0.344259 1.062614 -0.3202643
-0.975664 0.8296623 -0.15979593 0.06756159 1.5902116 -0.47518736
-0.00792558 0.18567389 -1.0544378 0.8262221 0.29249343 0.34478894
0.39678687 -1.354141 -0.548003 -0.14271511 -0.24100609 0.45800522
0.3675707 1.143411 -0.12880033 0.152542 0.0320172 -1.0160091
-1.4795289 0.7366357 0.49287072 -0.3651688 -0.50571936 0.9223552
0.4837265 0.06089673 0.4732656 -0.16327928 -0.30797604 0.14522356
0.5038597 0.3960867 -0.64681 -1.0312523 -0.40106347 1.0466822
-0.12162094 0.15239449 0.911978 -0.4402637 -0.15887602 0.4252266
0.64468825 -0.29557034 -1.6712099 -0.3911698 0.18253864 -0.5829282
0.06747671 -0.44128498 -1.4687796 0.5268196 0.66667974 0.20396435
0.97198015 0.05273626 0.847413 -0.15551831 0.39011785 -1.0036933
0.1018511 0.4153931 0.2831715 -1.3480731 0.01408709 -0.8163346
-0.6265612 1.0771917 0.98934937 0.05030546 0.50604063 0.430736
0.03215501 -0.15992725 -0.15870512 -0.5017871 0.38299558 0.502373
0.07219979 -0.2281254 -0.01461287 0.18435806 0.7968004 -0.01667325
0.07652946 0.9998591 -0.72396135 -1.1443051 -0.8831931 0.2896682
-0.46214244 0.18968958 0.09942622 0.78177667 0.4869231 0.9549328
0.09797531 0.04796343 0.17735812 -0.22656381 0.5311301 -0.6193972
-0.5058314 0.73522353 -0.27131024 -0.42359835 -0.9211225 -1.0161822
-1.398665 0.02556826 -0.8685179 0.17537035 0.27709407 1.0250437
-0.00629802 0.7608294 0.32309353 -1.0788039 -0.11443931 -0.5331071
-0.6713155 0.29660124 0.27449703 -0.9188286 -0.05200639 0.47265688] | [6.442507266998291, -1.9932774305343628] |
bfbe3c1a-fe36-48e0-b7a4-08e189fc641a | more-than-you-ve-asked-for-a-comprehensive | 2302.12173 | null | https://arxiv.org/abs/2302.12173v2 | https://arxiv.org/pdf/2302.12173v2.pdf | Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection | Large Language Models (LLMs) are increasingly being integrated into various applications. The functionalities of recent LLMs can be flexibly modulated via natural language prompts. This renders them susceptible to targeted adversarial prompting, e.g., Prompt Injection (PI) attacks enable attackers to override original instructions and employed controls. So far, it was assumed that the user is directly prompting the LLM. But, what if it is not the user prompting? We argue that LLM-Integrated Applications blur the line between data and instructions. We reveal new attack vectors, using Indirect Prompt Injection, that enable adversaries to remotely (without a direct interface) exploit LLM-integrated applications by strategically injecting prompts into data likely to be retrieved. We derive a comprehensive taxonomy from a computer security perspective to systematically investigate impacts and vulnerabilities, including data theft, worming, information ecosystem contamination, and other novel security risks. We demonstrate our attacks' practical viability against both real-world systems, such as Bing's GPT-4 powered Chat and code-completion engines, and synthetic applications built on GPT-4. We show how processing retrieved prompts can act as arbitrary code execution, manipulate the application's functionality, and control how and if other APIs are called. Despite the increasing integration and reliance on LLMs, effective mitigations of these emerging threats are currently lacking. By raising awareness of these vulnerabilities and providing key insights into their implications, we aim to promote the safe and responsible deployment of these powerful models and the development of robust defenses that protect users and systems from potential attacks. | ['Mario Fritz', 'Thorsten Holz', 'Christoph Endres', 'Shailesh Mishra', 'Sahar Abdelnabi', 'Kai Greshake'] | 2023-02-23 | null | null | null | null | ['computer-security'] | ['miscellaneous'] | [ 2.84684241e-01 -8.92792344e-02 -3.43679994e-01 1.38159350e-01
-6.57128990e-01 -1.52698588e+00 7.93174982e-01 -8.81785378e-02
-2.75633633e-01 -5.41324802e-02 2.15421602e-01 -1.37178934e+00
2.64526635e-01 -6.53349876e-01 -7.04315722e-01 -2.06561700e-01
-3.36311191e-01 -1.81072980e-01 5.31857193e-01 -2.48331606e-01
5.75973392e-01 6.45995736e-01 -1.07305646e+00 3.91041428e-01
5.65835655e-01 4.26591039e-01 -1.35810718e-01 9.07707870e-01
-9.64015871e-02 9.16603327e-01 -1.18369675e+00 -5.46072006e-01
2.71332532e-01 8.39300826e-02 -7.83228040e-01 -5.19344509e-01
3.88296098e-02 -8.05716872e-01 -2.99526155e-01 9.91961300e-01
1.86540499e-01 -5.15131414e-01 3.84240560e-02 -1.84513903e+00
-4.36664581e-01 6.89136863e-01 -5.79608858e-01 5.04037105e-02
4.42672670e-01 1.00701332e+00 5.07382572e-01 -2.98682153e-01
3.85323912e-01 1.07786691e+00 4.96904939e-01 8.45404863e-01
-1.42872453e+00 -8.97003055e-01 -5.94962873e-02 -6.22642279e-01
-9.26752388e-01 -5.61957359e-01 2.87713319e-01 -4.32734370e-01
1.07563305e+00 7.48194635e-01 -1.67418987e-01 1.63659561e+00
7.57266343e-01 2.93829769e-01 9.71931279e-01 -3.71719390e-01
2.33200446e-01 3.30773681e-01 4.95095491e-01 3.79285395e-01
5.83563328e-01 3.17215621e-01 -5.22223413e-01 -1.24407160e+00
3.73073786e-01 -1.03715427e-01 -4.21049714e-01 1.01070605e-01
-9.83161747e-01 7.19150066e-01 -5.43074198e-02 1.30633458e-01
-3.38398129e-01 3.89726222e-01 6.57579422e-01 4.02765036e-01
-1.31988779e-01 8.24332237e-01 -7.71996200e-01 -5.98069429e-01
-4.21216220e-01 -2.54665837e-02 1.24566591e+00 8.13753486e-01
4.08197761e-01 3.79910581e-02 5.40571287e-02 -2.16595978e-02
4.25767630e-01 6.24013245e-01 2.74284899e-01 -8.54945540e-01
4.17226821e-01 5.14785528e-01 1.41891897e-01 -1.04834354e+00
-2.15276051e-02 8.04925263e-02 -4.85377163e-02 4.38080221e-01
3.97693962e-01 -3.68151516e-01 -2.69734621e-01 1.68571603e+00
2.82537192e-02 8.64835456e-02 -9.10283104e-02 4.83089566e-01
-8.56393352e-02 4.77340966e-01 2.95230925e-01 4.77661602e-02
1.47355270e+00 -1.94085404e-01 -2.88465559e-01 -1.93916440e-01
7.84355879e-01 -7.85262465e-01 1.45181322e+00 3.16359341e-01
-5.75518191e-01 8.24013352e-02 -1.00987911e+00 5.60314000e-01
-5.84216714e-01 -6.38628185e-01 5.18651068e-01 1.14878297e+00
-9.04766500e-01 1.92191154e-01 -1.06221759e+00 -1.54691264e-01
1.63735554e-01 4.50142592e-01 -2.04241276e-01 2.72358745e-01
-1.18769372e+00 7.59333432e-01 -1.51806891e-01 -2.80822814e-01
-1.10090077e+00 -9.05856371e-01 -4.85331774e-01 1.73761025e-01
4.98018503e-01 -3.58756751e-01 1.42809832e+00 -4.56113189e-01
-1.30255580e+00 4.71043289e-01 4.10121620e-01 -4.72093016e-01
9.28264931e-02 -2.90676236e-01 -4.58126336e-01 5.60115324e-03
-2.88790204e-02 6.18178919e-02 1.08972538e+00 -1.20083177e+00
-1.39096886e-01 -7.41911009e-02 4.86716181e-01 -4.70580131e-01
-7.64252007e-01 8.47693384e-01 8.00443739e-02 -3.38933498e-01
-8.35042477e-01 -1.03502238e+00 -3.38346921e-02 -3.06545764e-01
-8.06029141e-01 4.68505949e-01 1.23535848e+00 -5.71328044e-01
1.35788763e+00 -2.24027681e+00 -3.20793837e-01 3.51236612e-01
3.87580276e-01 8.93196106e-01 -3.44213158e-01 9.06055808e-01
1.73231840e-01 9.45205569e-01 -2.69386312e-03 3.79329510e-02
2.74777412e-01 3.45525183e-02 -1.14944351e+00 3.43524843e-01
6.51859194e-02 1.00088251e+00 -6.25068903e-01 1.05061240e-01
-7.94729032e-03 3.68961126e-01 -7.90410995e-01 5.19813001e-01
-4.38617051e-01 2.69792974e-01 -5.23224831e-01 6.59474075e-01
4.37912971e-01 -3.03414669e-02 2.94734418e-01 1.79018855e-01
-3.76301259e-01 7.18438864e-01 -4.30469662e-01 8.44090343e-01
-7.83065259e-01 6.27707660e-01 6.85602009e-01 -8.21676776e-02
3.77693176e-01 3.12384844e-01 -2.63466597e-01 -3.12703162e-01
2.32040882e-02 9.63104144e-02 2.08799094e-01 -5.84612429e-01
2.84450293e-01 4.43546712e-01 -6.61909103e-01 1.12124646e+00
-4.38864708e-01 -5.68158105e-02 -4.54330027e-01 6.85492814e-01
1.88230371e+00 -2.34988600e-01 1.73037320e-01 -9.19096842e-02
3.15846980e-01 4.79114726e-02 5.52511141e-02 1.24241924e+00
-2.00722605e-01 -2.96514720e-01 1.05433226e+00 -1.21550545e-01
-5.35876572e-01 -1.03120542e+00 2.07300082e-01 1.45191419e+00
-1.81138024e-01 -9.41251040e-01 -9.15213764e-01 -1.09012473e+00
1.67514712e-01 1.20492554e+00 -1.71451464e-01 -6.92926228e-01
-6.85217440e-01 -4.67085809e-01 1.41980004e+00 2.12344989e-01
2.89182812e-01 -9.57069337e-01 -9.83452201e-01 4.11722362e-02
-6.73980592e-03 -9.50497866e-01 -8.79203141e-01 5.28678410e-02
-3.85485083e-01 -1.26988852e+00 2.92875707e-01 6.08161390e-02
4.50899273e-01 4.95816469e-01 1.01220369e+00 5.51652312e-01
-4.22810882e-01 8.19063962e-01 -1.90778956e-01 -2.59452134e-01
-1.22403717e+00 4.33253385e-02 4.05810885e-02 -1.71916485e-01
2.93752164e-01 -7.20322609e-01 -2.80557483e-01 4.81184691e-01
-1.27366626e+00 -5.21702409e-01 3.99077535e-01 4.54877555e-01
-4.64634150e-01 -1.82025358e-01 5.98258138e-01 -1.15184784e+00
1.12031245e+00 -8.37298870e-01 -9.72530246e-01 9.44714993e-03
-3.42892855e-01 -2.61739064e-02 8.62798572e-01 -8.58744144e-01
-8.49116027e-01 -4.90907073e-01 -2.44602874e-01 -1.98217601e-01
-3.82050365e-01 2.87423193e-01 -4.24075067e-01 -4.80356574e-01
8.41161788e-01 2.84282207e-01 8.05392414e-02 -3.53117585e-01
4.09562796e-01 8.88904512e-01 4.03286934e-01 -1.04601085e+00
1.22540009e+00 2.29823127e-01 -4.69071239e-01 -8.13054919e-01
-1.60607360e-02 -6.72369301e-02 2.28571430e-01 -1.99288595e-02
5.58158278e-01 -2.65735000e-01 -1.30060410e+00 7.02654302e-01
-1.43769479e+00 -5.68080723e-01 5.21040499e-01 -2.47101903e-01
-1.06113680e-01 5.88222980e-01 -9.44874644e-01 -7.64176965e-01
-5.17112792e-01 -1.50174785e+00 8.15215945e-01 -9.20088664e-02
-8.13772619e-01 -6.65906131e-01 7.97805116e-02 4.16855097e-01
1.26326883e+00 3.46129835e-02 1.36026251e+00 -1.02427864e+00
-8.60792160e-01 -4.80708927e-01 -5.14522940e-02 1.74802452e-01
1.90299913e-01 4.18247700e-01 -9.07385767e-01 -3.74815494e-01
3.00369740e-01 -3.69131267e-02 -1.73686996e-01 -6.32081807e-01
8.98226023e-01 -1.08591771e+00 -3.30413491e-01 5.77681780e-01
1.22690952e+00 2.98758119e-01 4.92307693e-01 1.77527323e-01
5.81936479e-01 5.13513505e-01 1.25013202e-01 1.75618172e-01
-1.30224705e-01 5.73247015e-01 7.10724533e-01 4.30096447e-01
3.28271300e-01 -3.65374744e-01 9.72430110e-01 2.23442376e-01
5.96666157e-01 -1.44956857e-01 -1.14200759e+00 -1.38846403e-02
-1.21584463e+00 -7.76672065e-01 -2.30641007e-01 2.44204235e+00
9.26661789e-01 5.97286105e-01 -4.26279157e-02 -3.43885273e-01
5.49231350e-01 7.87664056e-02 -4.86402959e-01 -7.25097179e-01
5.65610290e-01 3.00081223e-01 7.99275339e-01 5.96960008e-01
-7.45464981e-01 8.60067487e-01 6.24477911e+00 4.55472469e-01
-1.40507376e+00 2.88023591e-01 1.54521808e-01 -3.16332243e-02
-4.08320308e-01 4.37629640e-01 -7.39890099e-01 8.09942305e-01
1.58873916e+00 -2.88872987e-01 8.60386252e-01 7.79422283e-01
3.31009805e-01 3.57266478e-02 -9.32929635e-01 1.70222074e-01
-2.58965135e-01 -1.17230809e+00 -1.12851657e-01 4.36838418e-01
-1.39912337e-01 1.42226517e-01 1.47082120e-01 4.22652215e-01
7.49756634e-01 -8.69176626e-01 4.85183865e-01 2.16137487e-02
6.52603030e-01 -6.07528090e-01 2.93062359e-01 7.99701214e-01
-6.99671447e-01 -2.32932836e-01 3.76519769e-01 -2.64376909e-01
1.32813528e-01 1.57932132e-01 -7.92707145e-01 -9.18285102e-02
3.66733849e-01 -4.10208195e-01 -6.56549335e-01 1.99557871e-01
-3.57010216e-01 1.05998492e+00 -2.54399717e-01 5.98331653e-02
8.00955445e-02 3.12859267e-01 7.30707407e-01 1.14422536e+00
-1.38439789e-01 -1.30733877e-01 2.50094868e-02 1.15566278e+00
8.14418048e-02 -5.90705812e-01 -1.08826196e+00 -4.66753811e-01
8.76416147e-01 1.35345078e+00 -4.20088470e-01 1.09260090e-01
-6.80534124e-01 8.03102434e-01 6.56014914e-03 4.39528078e-01
-9.83556271e-01 -5.70646346e-01 1.31858206e+00 2.32123509e-01
-3.49860996e-01 -4.98540103e-01 -1.82854980e-01 -1.13820696e+00
8.48732248e-04 -1.54807329e+00 2.63083745e-02 -4.76252377e-01
-1.43121445e+00 5.79578936e-01 1.28515974e-01 -5.50659359e-01
-3.18477571e-01 -3.59349489e-01 -1.06916368e+00 9.43006217e-01
-8.32368553e-01 -9.47707534e-01 2.84855425e-01 4.55579102e-01
-5.28040854e-03 -2.18337923e-02 8.39348733e-01 1.54637381e-01
-5.77389061e-01 7.56752133e-01 -3.09581101e-01 3.23938787e-01
7.12405324e-01 -7.79662132e-01 1.07159328e+00 1.20910573e+00
-1.47465467e-01 1.63077462e+00 6.58500016e-01 -8.82563233e-01
-2.04293776e+00 -9.59536493e-01 4.09787148e-01 -1.11339450e+00
1.22067451e+00 -9.24886048e-01 -1.16751182e+00 9.64412212e-01
2.83925086e-01 -2.83137411e-01 7.71605313e-01 -3.35708618e-01
-8.89817357e-01 2.05704018e-01 -1.15591192e+00 1.08424354e+00
5.88409603e-01 -1.10591435e+00 -2.71209478e-01 2.83426374e-01
1.13028407e+00 7.56365955e-02 -5.51441133e-01 -1.21084020e-01
5.19081235e-01 -1.05089450e+00 8.96784306e-01 -9.24500465e-01
2.71867737e-02 -1.71552822e-01 -2.26847515e-01 -9.19473171e-01
3.19565654e-01 -1.41974676e+00 -2.76468456e-01 1.62498808e+00
1.69117153e-01 -1.32390702e+00 2.66962320e-01 1.10791707e+00
1.48263499e-01 -3.21118683e-01 -5.90514958e-01 -6.34923935e-01
1.25147223e-01 -5.86690843e-01 7.20840335e-01 9.96675193e-01
5.45487523e-01 1.12565279e-01 -2.74715126e-01 6.48034692e-01
5.79001069e-01 -3.81721586e-01 1.03779173e+00 -4.92550820e-01
-7.06084728e-01 -5.52595079e-01 2.94132289e-02 -7.39759505e-01
1.76495001e-01 -7.42124081e-01 -6.55493140e-02 -4.04268980e-01
-6.53033778e-02 -2.37748712e-01 2.01557666e-01 7.94462621e-01
3.21757533e-02 -1.50226772e-01 5.43010771e-01 2.80850321e-01
-1.49772838e-01 -7.79056624e-02 3.47781360e-01 7.22601861e-02
-1.64703622e-01 1.03548266e-01 -9.39392567e-01 7.45490968e-01
8.67787838e-01 -7.39824295e-01 -3.85306776e-01 -2.01948419e-01
2.94778764e-01 3.76963913e-01 6.32794678e-01 -7.10380554e-01
3.02407652e-01 -4.03827906e-01 -5.11692226e-01 2.27502614e-01
-5.39335534e-02 -8.45820069e-01 4.34909090e-02 6.86769724e-01
-4.01912302e-01 4.09827799e-01 4.21431512e-01 4.08274233e-01
3.57606024e-01 -1.38215393e-01 5.04845083e-01 -1.02262467e-01
-2.56970048e-01 1.85263865e-02 -1.24034095e+00 1.08488254e-01
1.10061550e+00 4.41466957e-01 -1.12252939e+00 -5.03645539e-01
-1.12536043e-01 5.20062223e-02 9.81910288e-01 6.32073522e-01
3.27098757e-01 -6.68726623e-01 -1.14651926e-01 4.44581211e-01
-1.23395538e-02 -8.86105120e-01 -1.17120326e-01 6.35629237e-01
-4.38903123e-01 2.44432241e-01 1.84378810e-02 -2.20827624e-01
-1.34059739e+00 9.88698781e-01 3.75691265e-01 -1.50735423e-01
-1.93773612e-01 2.78410494e-01 2.77936816e-01 -5.99947512e-01
2.20013216e-01 -1.25447333e-01 5.49699068e-01 -6.10210717e-01
8.18087041e-01 1.77599937e-01 -1.53292075e-01 -1.56961929e-03
-6.11915469e-01 -2.19467133e-01 -1.83090851e-01 -1.36056796e-01
7.12525308e-01 2.01352701e-01 -6.90308809e-01 3.88221443e-02
1.04996800e+00 7.85823047e-01 -8.48380089e-01 9.47218761e-02
1.62239239e-01 -6.74210310e-01 -4.94240016e-01 -1.07409191e+00
-6.84003830e-01 8.75122547e-01 -1.50914133e-01 8.91936600e-01
9.06940341e-01 -1.00638524e-01 1.05242622e+00 2.76967585e-01
8.03556204e-01 -8.14062506e-02 1.49052829e-01 3.80467862e-01
5.50240040e-01 -7.22644448e-01 -5.56217492e-01 -4.44147140e-01
-2.25813240e-01 1.00687099e+00 8.04600060e-01 1.55764297e-02
4.77187544e-01 1.15647960e+00 2.61889160e-01 -1.05874576e-01
-1.03111625e+00 7.06222236e-01 -4.84810799e-01 8.14664125e-01
2.01248780e-01 9.06287208e-02 1.05570452e-02 7.34133959e-01
3.84928375e-01 -3.40098053e-01 1.06058216e+00 1.23550463e+00
-1.84407011e-01 -1.55074143e+00 -7.83836782e-01 4.09889758e-01
-8.28344762e-01 -3.57363373e-01 -6.88071728e-01 9.37175393e-01
-3.15942705e-01 1.23534405e+00 -4.37948257e-01 -7.28610456e-01
2.84239799e-02 3.14071894e-01 -2.87664533e-01 -7.84343183e-01
-1.18144441e+00 -5.06053984e-01 2.49881282e-01 -8.67549062e-01
5.84917009e-01 -3.62430573e-01 -9.04084265e-01 -8.03956866e-01
-1.46410272e-01 9.60626155e-02 5.93906224e-01 6.99592292e-01
9.16529298e-01 2.52584487e-01 6.23328328e-01 -6.56299174e-01
-1.30490994e+00 -5.40534317e-01 -6.07660338e-02 1.23751111e-01
4.14311856e-01 -2.65908450e-01 -8.12858939e-01 -1.94630563e-01] | [6.131111145019531, 7.824937343597412] |
6969c115-59f4-4ee7-9148-0c0e8eec4ae5 | online-detection-of-failures-generated-by | 2101.07100 | null | https://arxiv.org/abs/2101.07100v1 | https://arxiv.org/pdf/2101.07100v1.pdf | Online detection of failures generated by storage simulator | Modern large-scale data-farms consist of hundreds of thousands of storage devices that span distributed infrastructure. Devices used in modern data centers (such as controllers, links, SSD- and HDD-disks) can fail due to hardware as well as software problems. Such failures or anomalies can be detected by monitoring the activity of components using machine learning techniques. In order to use these techniques, researchers need plenty of historical data of devices in normal and failure mode for training algorithms. In this work, we challenge two problems: 1) lack of storage data in the methods above by creating a simulator and 2) applying existing online algorithms that can faster detect a failure occurred in one of the components. We created a Go-based (golang) package for simulating the behavior of modern storage infrastructure. The software is based on the discrete-event modeling paradigm and captures the structure and dynamics of high-level storage system building blocks. The package's flexible structure allows us to create a model of a real-world storage system with a configurable number of components. The primary area of interest is exploring the storage machine's behavior under stress testing or exploitation in the medium- or long-term for observing failures of its components. To discover failures in the time series distribution generated by the simulator, we modified a change point detection algorithm that works in online mode. The goal of the change-point detection is to discover differences in time series distribution. This work describes an approach for failure detection in time series data based on direct density ratio estimation via binary classifiers. | ['Andrey Ustyuzhanin', 'Maksim Karpov', 'Leonid Gremyachikh', 'Vladislav Belavin', 'Andrey Sapronov', 'Mikhail Hushchyn', 'Kenenbek Arzymatov'] | 2021-01-18 | null | null | null | null | ['density-ratio-estimation'] | ['methodology'] | [-5.22125661e-01 -5.80478251e-01 -1.08664244e-01 3.87361832e-02
-1.08863413e-01 -4.30733860e-01 3.98767382e-01 4.12990957e-01
2.14234158e-01 6.63204074e-01 -7.52528965e-01 -7.85726666e-01
-2.69257396e-01 -7.60471046e-01 -7.71715403e-01 -7.18801498e-01
-7.93663681e-01 9.61144269e-01 8.97602201e-01 1.22422175e-02
2.31141627e-01 1.08349550e+00 -2.19734430e+00 -1.98700577e-02
3.86928260e-01 1.25481939e+00 4.36235577e-01 7.79977441e-01
-3.05147231e-01 4.14394081e-01 -1.32681608e+00 9.01594400e-01
3.15706804e-02 -1.78483073e-02 -3.58002275e-01 -1.67168342e-02
-3.08369815e-01 -3.99521351e-01 -3.31274480e-01 6.00674093e-01
1.67823955e-01 -1.69819415e-01 5.62833250e-01 -1.92754996e+00
2.50525534e-01 2.48940051e-01 -3.36875856e-01 8.61076951e-01
1.42845690e-01 1.11167505e-01 -7.05235824e-03 -3.25160772e-01
2.77283549e-01 8.55699956e-01 1.73158616e-01 -1.08399719e-01
-1.31321168e+00 -3.23147058e-01 -3.87456298e-01 3.87894154e-01
-1.40867400e+00 -7.44378120e-02 5.69244981e-01 -4.88928974e-01
1.55679321e+00 3.81627142e-01 2.34960169e-01 7.42681921e-01
7.89624989e-01 1.96954966e-01 1.13914371e+00 -7.90738404e-01
6.55632555e-01 3.16835821e-01 2.96217412e-01 3.06957245e-01
1.24637797e-01 3.85119975e-01 -2.27355808e-01 -3.90744627e-01
6.29921854e-01 2.48728856e-01 6.97310418e-02 -1.71472311e-01
-8.39322209e-01 3.10595185e-01 -2.03666046e-01 7.03368127e-01
-3.05462211e-01 -2.11932063e-01 6.57001555e-01 7.54031539e-01
2.53474265e-01 1.92504302e-01 -7.27444649e-01 -3.82956952e-01
-1.09152436e+00 -2.55350173e-01 1.12342477e+00 8.37653697e-01
5.58108807e-01 1.87684536e-01 -4.19211946e-02 2.76080966e-01
1.16763383e-01 5.67310989e-01 6.44029975e-01 -5.23600936e-01
-1.20452933e-01 4.55963433e-01 3.62862766e-01 -4.82936621e-01
-5.11257470e-01 -3.34509984e-02 -6.93829596e-01 5.27152121e-01
1.59170702e-01 3.12347740e-01 -8.15217376e-01 1.21845627e+00
4.26183403e-01 5.11903107e-01 -5.23850560e-01 3.76653165e-01
-4.05399799e-01 9.09660459e-01 -3.81889284e-01 -8.13279808e-01
8.98000479e-01 -2.51619548e-01 -8.74947250e-01 5.80955267e-01
5.31401038e-01 -6.48466945e-01 8.40145528e-01 6.37872577e-01
-9.58580196e-01 -6.99009299e-01 -1.19453883e+00 9.10453379e-01
-7.00840950e-01 -1.43130377e-01 1.83294073e-01 6.04768813e-01
-1.17039096e+00 8.75589609e-01 -1.43754411e+00 -4.31185395e-01
-1.86593100e-01 4.64467168e-01 8.57453272e-02 2.98970252e-01
-7.04079390e-01 9.66421068e-01 3.34520102e-01 -5.05217731e-01
-1.11781025e+00 -6.71507716e-01 -2.89753169e-01 2.42678195e-01
4.49944317e-01 -4.75704931e-02 1.17818820e+00 -1.84511229e-01
-1.35284078e+00 5.08867264e-01 2.41930448e-02 -2.03072399e-01
1.75467908e-01 2.65978456e-01 -1.08618367e+00 -5.01459651e-03
-3.15881401e-01 -6.35894299e-01 8.87395144e-01 -9.02302742e-01
-4.09510195e-01 -5.15882552e-01 -7.26967514e-01 -7.77921557e-01
-1.80919275e-01 2.36803174e-01 3.59448325e-03 -1.83217332e-01
-2.78319389e-01 -6.78666949e-01 5.08281410e-01 -6.01169348e-01
-4.09209728e-01 -4.99606490e-01 1.43075585e+00 -4.22433794e-01
1.50672638e+00 -2.12797022e+00 -4.31914002e-01 4.05652672e-01
-3.58389974e-01 2.24998698e-01 3.56431067e-01 9.84919608e-01
-3.24484855e-01 1.25529066e-01 2.97998011e-01 -2.84839302e-01
-1.79701168e-02 3.52613926e-01 -5.34078956e-01 4.08519268e-01
3.66614126e-02 7.13166147e-02 -3.28143984e-01 -1.39507890e-01
6.64431930e-01 -1.25291303e-01 3.85097414e-01 6.85102522e-01
-5.60866117e-01 1.54710934e-01 -1.77239627e-01 8.07538629e-01
6.27420545e-01 -5.29749334e-01 1.08300969e-01 1.34711161e-01
-4.51941848e-01 3.52495432e-01 -1.27025032e+00 9.98485088e-01
-4.95003730e-01 3.69325548e-01 -9.96854454e-02 -1.23703063e+00
9.60199237e-01 2.61433482e-01 5.81042588e-01 -7.10995913e-01
1.28283843e-01 3.59562457e-01 -1.28975645e-01 -5.48694074e-01
2.47747228e-01 9.79486331e-02 1.67506024e-01 8.54554117e-01
-1.64574236e-02 1.24876484e-01 1.34566411e-01 3.08653116e-02
1.98858833e+00 -3.73210609e-01 -6.37332126e-02 -2.70844281e-01
1.66115776e-01 1.83687173e-02 1.53693736e-01 6.14530444e-01
6.60949722e-02 -3.44711132e-02 5.93280256e-01 -2.16709778e-01
-1.25574362e+00 -1.45402992e+00 -4.25122589e-01 6.68381393e-01
-1.79243699e-01 -1.95640296e-01 -1.58000708e-01 -1.74721435e-01
3.29912573e-01 8.58814120e-01 -2.25516543e-01 -2.39975274e-01
-3.62453759e-01 -7.20978320e-01 -1.23116132e-02 4.10092235e-01
-2.20528573e-01 -1.04565430e+00 -6.65845752e-01 3.31850082e-01
7.43817329e-01 -8.91103864e-01 2.36823082e-01 7.59001732e-01
-1.08463013e+00 -1.22756672e+00 2.33240202e-01 -3.56707960e-01
4.67497617e-01 4.96968254e-02 1.23708880e+00 3.00831586e-01
-9.36344802e-01 5.21346092e-01 -2.05680847e-01 -3.37377518e-01
-8.91556978e-01 -2.60434777e-01 6.32332504e-01 -4.22874391e-01
3.47492963e-01 -7.18447924e-01 -2.33666733e-01 7.00757802e-01
-1.01066160e+00 -7.82722354e-01 2.36414373e-01 5.64355969e-01
7.06422567e-01 1.12939274e+00 6.73269391e-01 -3.36522490e-01
4.67160106e-01 -1.11873448e+00 -1.28829157e+00 5.25929391e-01
-1.28215051e+00 -4.39674333e-02 9.64494765e-01 -6.22878909e-01
-4.52547699e-01 -3.65072638e-01 5.07638514e-01 -9.49130058e-01
-4.48161602e-01 3.11617792e-01 -1.25675321e-01 5.10573030e-01
3.08873743e-01 2.67970204e-01 3.71045321e-01 -6.81769133e-01
-4.26513880e-01 9.77595687e-01 5.36324739e-01 -7.54924834e-01
5.44901848e-01 1.88794896e-01 1.25443771e-01 -8.23659360e-01
1.91696882e-01 -4.52300936e-01 -3.59178692e-01 -3.31778795e-01
6.90919831e-02 -5.68107903e-01 -1.10839546e+00 7.68924713e-01
-6.24537647e-01 -5.59721947e-01 -2.98534274e-01 2.70944059e-01
-4.75546122e-01 4.44229431e-02 -7.18732297e-01 -1.10971701e+00
1.17652990e-01 -8.57067227e-01 9.52550530e-01 3.86784762e-01
3.31728309e-01 -8.59257638e-01 3.65288526e-01 -4.28583682e-01
6.82852507e-01 -3.59254554e-02 1.19894564e+00 -9.20920730e-01
-8.11342955e-01 -5.83183110e-01 2.78167009e-01 4.67157423e-01
1.89229816e-01 5.93818784e-01 -6.72705114e-01 -8.93025219e-01
3.38314682e-01 -1.31556513e-02 -1.82503432e-01 3.00803423e-01
1.67833221e+00 9.21998769e-02 -7.09967136e-01 1.77502304e-01
1.70497286e+00 5.00740528e-01 7.65541017e-01 2.27062002e-01
-1.44278288e-01 3.65181297e-01 6.84175134e-01 8.29313993e-01
-3.02613169e-01 9.08832431e-01 4.98474181e-01 4.34979260e-01
4.69658047e-01 -8.91316086e-02 6.03763163e-01 7.88000822e-01
6.65991664e-01 -5.30135989e-01 -1.06247771e+00 4.99840379e-01
-1.48070550e+00 -8.65588188e-01 -2.14568302e-01 2.79161811e+00
5.09832799e-01 5.80259502e-01 4.80700463e-01 4.60947365e-01
8.27681303e-01 -6.21659219e-01 -7.35828996e-01 -4.10610348e-01
1.79013565e-01 3.63816708e-01 3.67161512e-01 -1.20428905e-01
-4.86316085e-01 1.60750806e-01 6.44122458e+00 7.62142539e-01
-1.29834998e+00 1.44874960e-01 4.59417194e-01 -2.52049357e-01
3.85003388e-01 4.80089895e-02 -9.97974098e-01 1.30588067e+00
2.11485958e+00 -2.47097060e-01 3.94947112e-01 8.09524417e-01
5.51487267e-01 -8.05693686e-01 -1.19363391e+00 7.78833807e-01
-2.08562434e-01 -1.26650226e+00 -7.29720592e-01 2.56090373e-01
2.47599721e-01 3.81961465e-01 -2.44579345e-01 4.42839891e-01
7.09780902e-02 -7.64025569e-01 3.42081726e-01 8.75431180e-01
7.70500064e-01 -5.05586684e-01 7.27520645e-01 6.66904569e-01
-8.11379969e-01 -1.35230601e-01 -1.23462237e-01 -7.97873661e-02
4.56268400e-01 1.07778049e+00 -9.02023911e-01 3.65295708e-01
9.39343572e-01 8.00170656e-03 -4.90100265e-01 1.21196508e+00
4.91888523e-01 9.49202895e-01 -1.05008721e+00 1.80997238e-01
-6.56522810e-01 1.84008274e-02 3.45822930e-01 5.25721550e-01
5.80083191e-01 -3.60645771e-01 3.03568959e-01 8.97455573e-01
5.09351254e-01 -4.79647011e-01 -5.13310671e-01 -1.06180511e-01
1.03737807e+00 1.02175879e+00 -8.89163673e-01 -1.58877209e-01
-4.29332435e-01 5.63446045e-01 -1.25726953e-01 1.31710887e-01
-7.29284525e-01 -1.67431071e-01 4.12752599e-01 7.48519480e-01
3.03325862e-01 -5.92002392e-01 8.07903707e-02 -6.64149821e-01
2.22452119e-01 -6.29253864e-01 1.88011929e-01 -9.21276391e-01
-1.59643102e+00 4.39290226e-01 3.29934806e-01 -9.95508730e-01
-4.95194644e-01 -7.06625998e-01 -9.92954433e-01 1.04458010e+00
-8.13839078e-01 -6.58067048e-01 -1.56630218e-01 7.18901992e-01
2.95597285e-01 -5.35690784e-01 6.26870930e-01 1.96274400e-01
-8.23062301e-01 2.01286003e-01 6.84926987e-01 -7.02341914e-01
6.03563964e-01 -1.33800185e+00 9.57816318e-02 6.83302581e-01
-2.90135652e-01 3.09685588e-01 1.02718782e+00 -9.46749985e-01
-1.56361198e+00 -5.87196529e-01 2.54576206e-01 -3.96971822e-01
1.10330021e+00 -3.81299287e-01 -1.42965007e+00 4.10971433e-01
-5.35373949e-02 3.02824289e-01 4.41448897e-01 8.91968682e-02
7.29727224e-02 -2.72113204e-01 -1.11944664e+00 -3.71459186e-01
2.21627116e-01 -6.30676568e-01 -3.24083269e-01 5.89057148e-01
2.15803981e-01 -8.40739384e-02 -1.06229222e+00 3.81628662e-01
-1.15313955e-01 -9.78691757e-01 5.38794398e-01 -5.48699796e-01
-1.50691748e-01 -4.17185962e-01 -2.19066054e-01 -1.06451261e+00
2.82244124e-02 -6.67712331e-01 -9.72342670e-01 1.33737242e+00
-1.97484810e-02 -6.41837955e-01 2.77188718e-01 4.15246040e-01
-1.80423379e-01 -6.09561801e-01 -1.08682299e+00 -1.50447714e+00
-2.62202293e-01 -1.75816700e-01 6.44401312e-01 5.59456944e-01
1.97261974e-01 -1.07217029e-01 -3.47899389e-03 4.76181924e-01
3.67282778e-01 1.59585461e-01 6.67998374e-01 -1.21281123e+00
-6.75481260e-01 -1.23543449e-01 -5.39904118e-01 -4.05892640e-01
1.85694531e-01 -2.84273386e-01 -9.35439840e-02 -7.48210013e-01
-9.11013875e-03 -8.89281571e-01 -5.46914995e-01 1.77688494e-01
3.54226381e-01 -6.98590100e-01 -3.29113603e-01 6.30173028e-01
-4.32746828e-01 3.18487108e-01 1.61294322e-02 2.57454544e-01
-1.50148615e-01 3.38524610e-01 6.29461765e-01 2.13936698e-02
8.56531382e-01 -4.73521650e-01 -3.15184951e-01 1.66729003e-01
2.17296600e-01 5.15167415e-01 5.22762120e-01 -1.33251441e+00
2.94531941e-01 -7.22732395e-02 1.25077277e-01 -1.11201882e+00
-7.07274899e-02 -1.02175272e+00 7.64638364e-01 3.92512679e-01
3.41705203e-01 8.11317623e-01 4.88079131e-01 6.76194131e-01
-2.52665311e-01 -2.05166712e-01 6.74628615e-01 3.21350157e-01
-5.04225552e-01 4.28438783e-02 -5.42485535e-01 -4.73192364e-01
1.55212593e+00 9.97062698e-02 -8.37925613e-01 2.09616199e-02
-7.12777913e-01 2.17782363e-01 8.89165819e-01 4.36236560e-01
1.05590858e-01 -8.78341675e-01 3.94342020e-02 5.17082632e-01
-4.36387509e-02 -4.71780002e-01 2.81845897e-01 9.19904947e-01
-4.45451975e-01 4.12399769e-01 -2.52060890e-01 -8.75195205e-01
-1.06668115e+00 1.21413815e+00 3.91283423e-01 -2.91927934e-01
-4.55792010e-01 7.26804361e-02 -7.27738798e-01 2.91035414e-01
1.77975714e-01 -2.15519682e-01 5.09741306e-01 -2.22274050e-01
6.41931951e-01 5.71692944e-01 6.04460716e-01 2.21861109e-01
-3.06829542e-01 -2.09710583e-01 -1.87265635e-01 2.13450924e-01
1.32660985e+00 -1.56945497e-01 -4.34952468e-01 1.24931633e+00
9.24567997e-01 -4.01035398e-01 -8.56365681e-01 6.92656487e-02
2.49477148e-01 -3.98632586e-01 1.16487406e-01 -9.06759977e-01
-6.65120363e-01 6.94696069e-01 1.23428893e+00 1.12760472e+00
1.26949668e+00 2.30955958e-01 4.29664910e-01 -6.34330884e-02
1.11047137e+00 -1.26908982e+00 1.38727250e-03 1.61235809e-01
3.95871818e-01 -1.05180061e+00 -2.32068568e-01 9.27486047e-02
3.05043340e-01 1.11799991e+00 6.77066803e-01 8.50892663e-02
1.24761641e+00 1.03515995e+00 -5.68701565e-01 -2.92441905e-01
-1.31238830e+00 1.87359244e-01 -5.47710061e-01 5.62555075e-01
-1.54337123e-01 3.48457664e-01 -7.78325051e-02 2.47507975e-01
3.23190868e-01 2.79328495e-01 6.22752845e-01 1.34658933e+00
-5.76674223e-01 -1.37743545e+00 -7.29061782e-01 8.56452525e-01
-1.97298855e-01 4.47911978e-01 2.45409921e-01 6.56721175e-01
-1.15839534e-01 9.89109993e-01 6.93888426e-01 -5.19012630e-01
1.51086152e-01 3.55213583e-01 3.62586737e-01 -5.29309809e-01
-2.13466257e-01 -2.03121573e-01 -3.93289804e-01 -7.08177507e-01
2.27795914e-01 -8.03225100e-01 -1.19203484e+00 -6.60904348e-01
-4.63805079e-01 6.30228370e-02 1.16644418e+00 8.59968841e-01
6.63058043e-01 7.41689444e-01 1.17881238e+00 -5.52999973e-01
-1.02235389e+00 -1.08639157e+00 -1.34684789e+00 1.93809509e-01
8.32777992e-02 -1.11019707e+00 -8.46811056e-01 -1.21570468e-01] | [6.749003887176514, 2.7101778984069824] |
42feebbb-b4bd-422c-bf61-7b776f9b6926 | self-vs-self-supervised-encoding-learning-for | 2303.15993 | null | https://arxiv.org/abs/2303.15993v1 | https://arxiv.org/pdf/2303.15993v1.pdf | SELF-VS: Self-supervised Encoding Learning For Video Summarization | Despite its wide range of applications, video summarization is still held back by the scarcity of extensive datasets, largely due to the labor-intensive and costly nature of frame-level annotations. As a result, existing video summarization methods are prone to overfitting. To mitigate this challenge, we propose a novel self-supervised video representation learning method using knowledge distillation to pre-train a transformer encoder. Our method matches its semantic video representation, which is constructed with respect to frame importance scores, to a representation derived from a CNN trained on video classification. Empirical evaluations on correlation-based metrics, such as Kendall's $\tau$ and Spearman's $\rho$ demonstrate the superiority of our approach compared to existing state-of-the-art methods in assigning relative scores to the input frames. | ['Mahdi Eftekhari', 'Mehrdad Hosseinzadeh', 'Kave Bahraman', 'Hojjat Mokhtarabadi'] | 2023-03-28 | null | null | null | null | ['video-classification'] | ['computer-vision'] | [ 5.02734542e-01 -1.80254206e-02 -4.46000427e-01 -5.64164579e-01
-9.75666285e-01 -3.60331327e-01 4.34306949e-01 1.29921600e-01
-3.24485838e-01 8.91570091e-01 7.16306925e-01 4.23027314e-02
9.04831886e-02 -6.38959408e-01 -8.94143581e-01 -3.04630667e-01
-5.62418513e-02 5.90978637e-02 2.88565189e-01 -1.29758520e-02
3.60785842e-01 -9.84783545e-02 -1.85390699e+00 4.46517587e-01
8.24681401e-01 1.33189929e+00 3.67447436e-02 5.28070450e-01
2.29974277e-02 1.40076780e+00 -5.80899298e-01 -5.73802650e-01
6.11243732e-02 -1.00092995e+00 -7.57906199e-01 5.39961383e-02
4.70069379e-01 -5.45071423e-01 -6.45765662e-01 1.13070285e+00
1.84064895e-01 4.15770829e-01 5.28839529e-01 -1.16132998e+00
-8.57853830e-01 3.65086883e-01 -3.73595923e-01 4.30187434e-01
6.64668620e-01 -8.33008960e-02 1.36705291e+00 -8.14612031e-01
7.75245905e-01 8.46904159e-01 7.69599378e-01 4.50552642e-01
-1.19228625e+00 -5.71284592e-01 5.65237105e-02 5.73039591e-01
-1.26336730e+00 -5.30239046e-01 7.09064007e-01 -4.98503238e-01
1.12946200e+00 1.79901809e-01 6.64762378e-01 9.45526302e-01
-1.54925242e-01 7.92183042e-01 7.05411851e-01 -2.90856093e-01
2.28551134e-01 -1.21282518e-01 -2.35706449e-01 9.75898147e-01
2.19822988e-01 -2.20552206e-01 -8.31530929e-01 1.73013553e-01
7.76012182e-01 2.88564824e-02 -2.60739326e-01 -4.52230364e-01
-1.12574923e+00 7.91059673e-01 4.71569926e-01 1.20033585e-01
-3.25511038e-01 3.72262299e-01 7.30264723e-01 2.55458355e-01
7.06414163e-01 4.31544691e-01 -2.80306965e-01 -5.75051248e-01
-1.40130591e+00 2.07430124e-01 5.96728802e-01 8.21168542e-01
7.67655730e-01 1.77130684e-01 -2.55981088e-01 7.84906209e-01
1.68939903e-01 -3.05241756e-02 7.01898754e-01 -1.24545431e+00
4.80563641e-01 7.34027445e-01 6.02547750e-02 -1.00033081e+00
-4.54480015e-02 -2.56897241e-01 -7.76857316e-01 -3.58354263e-02
-2.68668141e-02 1.50568828e-01 -5.79998791e-01 1.84952950e+00
-1.62877887e-01 4.75929677e-01 1.13670565e-01 1.00419486e+00
9.13381338e-01 6.34603262e-01 4.37724888e-02 -4.19135898e-01
1.05541646e+00 -1.23033702e+00 -4.96818781e-01 9.21927951e-03
4.54721898e-01 -4.28406864e-01 9.75225449e-01 1.27770722e-01
-1.19440377e+00 -5.81068635e-01 -1.12141967e+00 -9.20094997e-02
6.14020377e-02 6.45977110e-02 6.10076189e-01 4.11612868e-01
-1.15008366e+00 8.40447783e-01 -8.21660101e-01 -4.92495626e-01
8.53986502e-01 2.03439057e-01 -5.80344856e-01 -7.20810145e-02
-9.74802196e-01 8.23891401e-01 3.69795471e-01 -5.12918867e-02
-7.98791587e-01 -7.86183298e-01 -1.03385091e+00 3.23827744e-01
5.15676081e-01 -8.75873566e-01 1.39925110e+00 -1.32376206e+00
-1.58299935e+00 7.25979447e-01 -2.24333346e-01 -7.37125933e-01
4.14574027e-01 -4.24121648e-01 -1.36564285e-01 5.31795979e-01
3.41586322e-01 7.17744827e-01 7.16462553e-01 -9.73948002e-01
-8.19148839e-01 -6.23384193e-02 2.46141464e-01 3.33926111e-01
-4.52501237e-01 1.09470457e-01 -5.89785397e-01 -8.91764700e-01
-1.34911388e-01 -5.29106557e-01 -2.30592713e-01 -1.43013403e-01
-1.05899140e-01 -1.33971617e-01 6.37177169e-01 -8.55460048e-01
1.56965411e+00 -1.95711541e+00 4.09080982e-01 -7.94288740e-02
2.02432528e-01 3.23721975e-01 -3.60981561e-02 2.78343290e-01
-4.32344265e-02 -5.27966470e-02 -2.90701151e-01 -3.93415242e-01
-9.83514711e-02 6.75955564e-02 -1.51508212e-01 2.49045074e-01
3.92307937e-01 8.16884339e-01 -1.16773534e+00 -6.32361829e-01
2.51870602e-01 4.53537077e-01 -6.82011366e-01 3.96036506e-01
-3.00629526e-01 3.32861096e-01 -2.16596469e-01 6.43445730e-01
1.41524047e-01 -4.27261680e-01 2.56601393e-01 -3.77686411e-01
1.82195138e-02 3.86897683e-01 -7.29925156e-01 2.10131955e+00
-1.06098115e-01 8.80327880e-01 -5.70109248e-01 -1.57354593e+00
7.84688473e-01 3.64809483e-01 8.12109768e-01 -7.26322055e-01
2.30856508e-01 2.85476968e-02 -4.24341321e-01 -5.99835873e-01
7.11955309e-01 -1.67748198e-01 -6.11885898e-02 3.23952287e-01
2.64492899e-01 1.11124121e-01 4.81365830e-01 3.44117731e-01
1.45390677e+00 5.65299869e-01 5.60628474e-01 1.77306402e-03
5.66552997e-01 5.75263286e-03 7.88505018e-01 3.71512175e-01
-2.05880597e-01 8.99746120e-01 6.93741560e-01 -5.78219056e-01
-1.17435884e+00 -9.39016640e-01 3.25146675e-01 1.09478283e+00
2.00861339e-02 -8.37910771e-01 -1.02248704e+00 -7.57321537e-01
-1.44939184e-01 3.58064651e-01 -6.39374912e-01 -3.56525481e-01
-5.32021165e-01 -3.10097992e-01 5.10997772e-01 8.74544442e-01
4.91329312e-01 -1.07209468e+00 -8.90491068e-01 2.39127606e-01
-3.46656203e-01 -1.32401121e+00 -2.98995823e-01 -1.91139296e-01
-9.57417607e-01 -1.25057042e+00 -6.53345346e-01 -6.66475475e-01
5.84181011e-01 3.33786070e-01 1.34530938e+00 1.40885487e-01
-3.35634872e-02 3.02253246e-01 -6.66586399e-01 -6.20503910e-02
-3.21722239e-01 5.48150018e-02 -5.39855622e-02 2.73935287e-03
4.25229579e-01 -4.73392189e-01 -9.03949142e-01 7.10319206e-02
-9.92564082e-01 2.23354593e-01 5.13788819e-01 7.66506910e-01
6.89543784e-01 4.60151769e-02 7.49717712e-01 -9.19364512e-01
5.03531098e-01 -4.77992028e-01 -3.95126253e-01 1.66800275e-01
-3.69718403e-01 -3.70827429e-02 5.38634896e-01 -1.05663650e-01
-8.02432358e-01 -7.14618042e-02 1.91580787e-01 -6.66310012e-01
1.08926892e-01 7.99555361e-01 1.27132237e-01 2.02551246e-01
4.78004932e-01 1.91970542e-01 5.82791083e-02 -1.32771805e-01
2.14722484e-01 3.38446319e-01 8.25184941e-01 -2.99334705e-01
6.33421898e-01 2.90416002e-01 -8.06339756e-02 -7.53628492e-01
-1.06524098e+00 -3.15741450e-01 -4.22592342e-01 -4.11487132e-01
8.91533852e-01 -1.15393996e+00 -4.13944095e-01 1.33105099e-01
-9.94380891e-01 -1.51547641e-01 -4.02202338e-01 5.24888039e-01
-8.97869885e-01 5.02055228e-01 -5.03727674e-01 -5.06358266e-01
-4.18489903e-01 -1.06961179e+00 8.96058857e-01 1.91506043e-01
-4.40130144e-01 -8.65265608e-01 1.05250522e-01 7.69538939e-01
4.73443568e-01 4.08734143e-01 7.24368691e-01 -3.60327512e-01
-6.11041427e-01 -2.98087806e-01 -4.92965966e-01 7.62412906e-01
9.11747813e-02 -9.10541117e-02 -7.17679381e-01 -2.17660040e-01
-3.13896030e-01 -4.31787580e-01 1.08786893e+00 5.01875103e-01
1.30560100e+00 -4.61910874e-01 1.19183790e-02 4.16341692e-01
1.42315006e+00 1.85514279e-02 8.45122755e-01 3.98117989e-01
6.29669070e-01 2.54292190e-01 5.94965637e-01 6.67230070e-01
5.99720716e-01 5.82324207e-01 4.93088424e-01 1.14450581e-01
-2.64083445e-01 -4.28537905e-01 4.23068792e-01 8.97765160e-01
-5.19539237e-01 -2.09857762e-01 -5.95114827e-01 3.89835805e-01
-2.20153856e+00 -1.24369359e+00 3.78458828e-01 2.24724627e+00
7.68832743e-01 7.10512474e-02 3.09628904e-01 2.27811828e-01
6.75731301e-01 3.38595957e-01 -4.06138748e-01 -1.66213930e-01
-3.94759588e-02 2.09688172e-01 2.19749957e-01 3.59770469e-02
-1.15259016e+00 9.56421256e-01 6.03161240e+00 5.43245554e-01
-9.31000113e-01 6.11709021e-02 6.02752030e-01 -1.97417974e-01
-6.07447177e-02 -3.28131318e-02 -1.39529541e-01 6.92106605e-01
9.86836374e-01 -3.22229147e-01 3.59345466e-01 7.02819884e-01
6.02698922e-02 -2.34132707e-01 -1.33189785e+00 1.14528155e+00
6.39558434e-01 -1.83683348e+00 1.96795464e-01 -2.64079422e-01
9.34599876e-01 -6.32517487e-02 -8.74033347e-02 2.96943128e-01
9.82978493e-02 -1.01128876e+00 8.33367467e-01 5.67735434e-01
9.81016219e-01 -6.13620937e-01 7.96031535e-01 -9.38528925e-02
-1.40123999e+00 -8.53752568e-02 -3.08344066e-01 -2.69122720e-01
1.91655084e-01 3.78136545e-01 -6.38097048e-01 5.90395093e-01
7.29751110e-01 1.10858238e+00 -4.21737015e-01 9.45955753e-01
-1.70458525e-01 6.66575670e-01 1.73551336e-01 1.89316228e-01
1.35463834e-01 -7.56768063e-02 2.07450688e-01 1.28303850e+00
5.49852014e-01 1.51332214e-01 1.79010794e-01 5.08684397e-01
-4.90035743e-01 9.82729346e-02 -7.31646895e-01 -2.38702118e-01
3.45767766e-01 9.81266081e-01 -7.20417380e-01 -6.37730002e-01
-7.32205451e-01 1.00920963e+00 4.32222039e-01 2.67393380e-01
-1.02137125e+00 -2.96609104e-01 5.62555850e-01 1.07572071e-01
4.11656618e-01 -1.47500038e-01 -1.58714712e-01 -1.41031313e+00
1.99704632e-01 -7.34084368e-01 5.23756206e-01 -7.99002171e-01
-9.95349109e-01 6.25278473e-01 1.77635644e-02 -1.52168298e+00
-4.56525177e-01 -2.01379329e-01 -4.74338323e-01 2.80978888e-01
-1.60114491e+00 -9.89421308e-01 -4.90971982e-01 3.96046966e-01
8.36093426e-01 -5.55453360e-01 7.95568943e-01 3.27928036e-01
-5.96073389e-01 6.29336715e-01 -1.12240389e-01 9.76533368e-02
6.07894659e-01 -1.00832570e+00 8.93320143e-02 8.76008749e-01
1.70978785e-01 2.32728645e-01 7.61032581e-01 -3.28808457e-01
-1.31777179e+00 -1.11263359e+00 8.53939712e-01 -3.34085077e-01
6.94055974e-01 2.06926800e-02 -8.82964909e-01 6.90444410e-01
1.82117626e-01 1.44873098e-01 1.04414594e+00 -6.88269287e-02
-5.99523723e-01 -2.87713647e-01 -9.36009347e-01 4.09588814e-01
1.18899035e+00 -6.28673315e-01 -7.69211769e-01 1.41201958e-01
7.21187651e-01 -1.91293657e-01 -9.48809266e-01 5.05347610e-01
6.71458662e-01 -1.26314187e+00 9.70644653e-01 -7.44061172e-01
1.13333309e+00 -1.97528124e-01 -3.98306280e-01 -1.18839765e+00
-3.55586946e-01 -4.41062659e-01 -3.80693257e-01 1.23662436e+00
1.41260177e-01 -1.18577555e-01 1.02935362e+00 4.47822332e-01
-2.92001396e-01 -7.97213852e-01 -1.06288767e+00 -5.80965936e-01
-4.12525326e-01 -4.12473649e-01 4.26066816e-01 8.71764839e-01
1.99702546e-01 4.86228794e-01 -5.98689079e-01 -1.81596830e-01
5.71069419e-01 5.91314249e-02 7.46118069e-01 -1.20206571e+00
-2.03909993e-01 -5.92471480e-01 -9.31283295e-01 -6.58816040e-01
4.54969287e-01 -7.50751495e-01 -1.85568109e-02 -1.77131641e+00
5.78652203e-01 -9.30072442e-02 -6.65921092e-01 3.83685797e-01
-1.33882836e-01 7.44477272e-01 2.35802799e-01 2.87517279e-01
-1.31875420e+00 6.39518797e-01 7.98587322e-01 -1.32133603e-01
-2.50797737e-02 -3.89026433e-01 -7.96287239e-01 7.43922949e-01
8.20109606e-01 -2.13940427e-01 -5.51060319e-01 -6.74708128e-01
5.39823949e-01 2.66581595e-01 2.85113186e-01 -1.38758457e+00
1.69368505e-01 -3.34723406e-02 1.58257157e-01 -3.67531300e-01
2.47399122e-01 -5.31252980e-01 5.29341251e-02 2.85941213e-01
-4.39726382e-01 2.11545870e-01 -1.35887578e-01 6.17982268e-01
-6.38777316e-01 -7.14436099e-02 5.96445918e-01 -1.15536623e-01
-8.48563433e-01 2.62606949e-01 -2.13072419e-01 1.36500776e-01
8.85288239e-01 -5.43036520e-01 -3.20120156e-01 -5.43729544e-01
-2.08565265e-01 4.92497208e-03 5.79665661e-01 4.44616139e-01
8.29270899e-01 -1.33147228e+00 -5.58473647e-01 6.20437749e-02
2.14404672e-01 -1.42186768e-02 3.17298412e-01 8.70528042e-01
-6.28087938e-01 4.15805757e-01 -3.69740695e-01 -3.49831492e-01
-1.16965997e+00 4.17986333e-01 -6.05053417e-02 -2.37017289e-01
-7.22061813e-01 6.41937137e-01 7.49372244e-02 2.95592964e-01
2.91472226e-01 -2.67881095e-01 -2.94974804e-01 1.16567656e-01
5.07775068e-01 6.04222775e-01 -3.56333740e-02 -7.65227556e-01
-2.87500888e-01 4.42750037e-01 7.40361661e-02 6.83963671e-02
1.47689259e+00 -7.58567452e-03 -3.45453210e-02 2.57787973e-01
1.33143997e+00 -4.01490748e-01 -1.44653106e+00 -2.70302683e-01
1.10594302e-01 -6.26809776e-01 -1.30939096e-01 -4.78698164e-01
-1.08970857e+00 5.50471067e-01 3.15853208e-01 6.79246783e-02
1.20800877e+00 1.02865092e-01 9.50840235e-01 4.82482493e-01
2.33907759e-01 -1.27392578e+00 4.85269815e-01 4.26771790e-01
6.02685988e-01 -1.46880579e+00 1.99110389e-01 -2.13953882e-01
-7.50787675e-01 9.89308953e-01 5.84575534e-01 -3.40093940e-01
2.70930171e-01 -9.95039865e-02 -6.51308149e-02 -7.45826066e-02
-8.90689850e-01 9.33939684e-03 1.93452537e-01 4.76646364e-01
7.28031576e-01 -1.19266964e-01 -2.34335691e-01 5.76987326e-01
-2.78347164e-01 5.32551408e-01 4.64790940e-01 9.90816236e-01
-4.65376318e-01 -7.78233290e-01 7.31069073e-02 6.86095238e-01
-5.78635216e-01 1.13739409e-01 3.95391174e-02 4.23207462e-01
-3.35768214e-03 7.60328710e-01 2.16381401e-01 -4.97384548e-01
2.63218403e-01 -7.32680708e-02 4.91288990e-01 -6.85779393e-01
-3.67440134e-01 -1.91797137e-01 7.45500773e-02 -7.66297817e-01
-9.85279977e-01 -6.33227170e-01 -1.26397693e+00 -1.20906368e-01
1.28995880e-01 1.89746544e-01 4.00611311e-01 9.42661524e-01
3.74951988e-01 6.31278694e-01 4.72151697e-01 -1.12209010e+00
-4.07229602e-01 -7.85472512e-01 -2.64435291e-01 8.04634035e-01
1.49087340e-01 -8.63311768e-01 1.80579014e-02 3.74222070e-01] | [10.245601654052734, 0.4766032099723816] |
94f9c6ce-9ca7-4d1f-b411-4b934dfda801 | joining-the-conversation-towards-language | 2305.12235 | null | https://arxiv.org/abs/2305.12235v1 | https://arxiv.org/pdf/2305.12235v1.pdf | Joining the Conversation: Towards Language Acquisition for Ad Hoc Team Play | In this paper, we propose and consider the problem of cooperative language acquisition as a particular form of the ad hoc team play problem. We then present a probabilistic model for inferring a speaker's intentions and a listener's semantics from observing communications between a team of language-users. This model builds on the assumptions that speakers are engaged in positive signalling and listeners are exhibiting positive listening, which is to say the messages convey hidden information from the listener, that then causes them to change their behaviour. Further, it accounts for potential sub-optimality in the speaker's ability to convey the right information (according to the given task). Finally, we discuss further work for testing and developing this framework. | ['Peter McBurney', 'Dylan Cope'] | 2023-05-20 | null | null | null | null | ['language-acquisition'] | ['natural-language-processing'] | [ 3.87030959e-01 8.66821945e-01 1.16763085e-01 -4.82420474e-01
-7.10872948e-01 -6.08647585e-01 7.43551910e-01 1.64472520e-01
-4.42417055e-01 4.82644260e-01 4.13266957e-01 -3.31967115e-01
-3.33413184e-01 -4.45163965e-01 -5.18832207e-01 -7.29677260e-01
-3.72779965e-01 6.09146953e-01 2.51319349e-01 -5.23052096e-01
1.07580453e-01 -2.04510093e-02 -1.56856358e+00 8.20392519e-02
1.86132342e-01 3.99311483e-01 4.39000696e-01 1.03534901e+00
-7.02081015e-04 1.38824058e+00 -5.38068950e-01 -2.56074876e-01
-1.05038665e-01 -3.80808651e-01 -1.11327505e+00 2.81151712e-01
-3.72049183e-01 -1.67242780e-01 2.68898636e-01 8.78825307e-01
4.33540106e-01 3.23719867e-02 5.58850706e-01 -1.35424268e+00
-7.31539726e-02 1.42948592e+00 5.13179936e-02 3.83950144e-01
9.72503662e-01 -1.45415692e-02 1.37437129e+00 -3.13407272e-01
3.65376145e-01 1.46500933e+00 3.63313526e-01 7.93375075e-01
-1.09259093e+00 -5.65395117e-01 4.71851796e-01 -1.05778001e-01
-1.33663154e+00 -8.70456040e-01 8.23275149e-01 -6.39089346e-01
5.82879543e-01 4.62086290e-01 7.78140485e-01 1.07983100e+00
1.45950824e-01 1.01088834e+00 1.13669860e+00 -6.92946613e-01
4.34762686e-01 4.83418763e-01 9.30878669e-02 2.16477513e-01
-2.66593695e-01 2.68372118e-01 -1.33259535e+00 -4.47989613e-01
2.61245668e-01 -8.23009491e-01 -1.08892016e-01 -1.63877264e-01
-1.05132401e+00 8.00859869e-01 -3.19389135e-01 7.11557567e-01
-6.43658161e-01 1.63018957e-01 -9.47312415e-02 9.94460940e-01
3.45426768e-01 4.82639670e-01 -3.37938458e-01 -4.03954864e-01
-3.39049608e-01 4.77992356e-01 1.28406465e+00 6.99872017e-01
4.19028968e-01 -2.79976368e-01 4.69164588e-02 4.67350394e-01
1.12436998e+00 1.60021916e-01 2.44238630e-01 -1.17962408e+00
1.25063062e-01 9.28565934e-02 2.76291013e-01 -7.94099808e-01
-4.48514163e-01 -2.63038963e-01 3.51864286e-02 -7.85396695e-02
2.20383272e-01 -3.24650437e-01 -1.35027125e-01 2.05206251e+00
4.38428521e-01 3.01344153e-02 6.17453694e-01 6.22127712e-01
5.60641229e-01 5.68524480e-01 -7.41181970e-02 -7.56788909e-01
1.19486248e+00 -3.18028480e-01 -7.89712787e-01 -4.57839012e-01
6.56283081e-01 -7.22800493e-01 5.65731883e-01 6.22560501e-01
-1.44328761e+00 -1.10852972e-01 -8.26483965e-01 4.32801157e-01
4.28116828e-01 -8.00135314e-01 7.47223616e-01 7.65472174e-01
-1.38612092e+00 6.59640059e-02 -5.70298076e-01 -3.51608694e-01
-1.62389740e-01 5.97081959e-01 -2.65224148e-02 2.15563431e-01
-1.30233037e+00 4.83043969e-01 -6.81243017e-02 1.50489867e-01
-1.40993834e+00 -5.37855215e-02 -6.44532442e-01 -6.11877106e-02
4.64921445e-01 -4.18345779e-01 2.10547614e+00 -1.23986769e+00
-2.00763631e+00 1.04123402e+00 -2.28646979e-01 -2.38836750e-01
3.89500648e-01 1.34760857e-01 -1.03876337e-01 -1.49154216e-01
4.09590602e-01 4.10658121e-01 6.90776467e-01 -1.36287141e+00
-9.61770535e-01 -4.61789012e-01 2.20163599e-01 6.21642232e-01
7.54709169e-02 5.64379930e-01 -4.88119982e-02 -3.56226027e-01
3.22824210e-01 -9.46658790e-01 -2.61679679e-01 -6.90468431e-01
-4.44380224e-01 -8.12490106e-01 1.88555151e-01 -1.96014866e-02
8.19396973e-01 -2.44322968e+00 4.95508105e-01 5.54278314e-01
4.81370449e-01 -4.79977638e-01 2.55871955e-02 6.85475588e-01
1.40064240e-01 1.69224665e-01 2.25979984e-02 -7.09859908e-01
7.20076635e-02 3.91708702e-01 -2.27426976e-01 5.23576379e-01
-2.61817247e-01 3.22569698e-01 -6.21037543e-01 -4.40061063e-01
-4.63140696e-01 3.80766392e-02 -6.13890052e-01 7.35924065e-01
-3.69275361e-01 8.13444614e-01 -9.16080296e-01 9.44486111e-02
1.10178009e-01 1.38651714e-01 5.41961372e-01 1.25436723e+00
-4.33571815e-01 8.52328658e-01 -1.25079167e+00 1.14836597e+00
-1.96379811e-01 4.97276157e-01 9.62598264e-01 -8.60888422e-01
6.01534247e-01 8.87862682e-01 7.38447234e-02 -1.48013234e-01
3.50292712e-01 1.77538678e-01 6.69677079e-01 -5.34401894e-01
7.83127174e-02 -5.70180535e-01 -4.08876956e-01 1.13661230e+00
-2.06904396e-01 -2.89564669e-01 -3.12844813e-01 4.54886168e-01
8.88739109e-01 -5.13269365e-01 3.33728224e-01 -2.82751828e-01
4.19202238e-01 -2.96884865e-01 3.90809715e-01 1.32855988e+00
-2.70901114e-01 -3.28689367e-01 8.15515459e-01 8.10706690e-02
-2.90343642e-01 -7.22503662e-01 2.83905625e-01 1.95138061e+00
3.38587798e-02 -1.31643653e-01 -5.76852322e-01 -1.26542896e-01
-4.71343130e-01 9.51475561e-01 -5.80960810e-01 -7.83889219e-02
-2.23458990e-01 1.27771813e-02 4.22475070e-01 -1.42444472e-03
6.09510876e-02 -1.19715595e+00 -6.42990291e-01 2.18162060e-01
-2.78190494e-01 -7.95629382e-01 -3.69265676e-01 3.16731691e-01
-3.06992590e-01 -5.00853717e-01 -1.35594189e-01 -8.96412909e-01
3.44498575e-01 3.18585843e-01 7.36016631e-01 3.51506740e-01
5.17809033e-01 9.28416729e-01 -4.31023538e-01 -8.74904811e-01
-9.75020587e-01 -2.35156208e-01 3.92675966e-01 2.13640258e-01
7.09785447e-02 -5.73799908e-01 1.13432385e-01 1.61435723e-01
-6.61424220e-01 8.59356821e-02 2.08705187e-01 3.68730038e-01
-1.93119764e-01 3.77857387e-01 3.26615453e-01 -7.69157887e-01
1.11884642e+00 -5.88757455e-01 -3.77721608e-01 7.18444064e-02
-1.44975111e-01 -7.23111629e-02 -4.29845378e-02 -4.43797737e-01
-1.18668544e+00 7.18619823e-02 5.08538298e-02 4.88160789e-01
-3.06348592e-01 5.26970685e-01 -1.90918669e-01 9.07304212e-02
4.35190529e-01 3.39326859e-01 1.70210943e-01 -3.46584141e-01
9.78322625e-02 1.02325046e+00 5.79607636e-02 -7.92816818e-01
4.85920072e-01 1.38745323e-01 -4.74475294e-01 -1.01895809e+00
-9.01832819e-01 -5.81579328e-01 -1.96570963e-01 -5.20247996e-01
8.62914145e-01 -1.12825346e+00 -1.26137435e+00 5.79370379e-01
-1.32650268e+00 -6.25203192e-01 4.68027070e-02 6.67993844e-01
-7.52227783e-01 2.04814151e-02 -4.15395945e-01 -1.79800475e+00
4.45549935e-01 -1.15342557e+00 8.37682843e-01 6.41456321e-02
-6.00950003e-01 -9.88562107e-01 9.53210443e-02 5.95979154e-01
2.07681715e-01 -7.46235430e-01 5.70534647e-01 -1.25634491e+00
-5.99926293e-01 3.87003049e-02 7.76141822e-01 4.89313202e-03
-1.92214310e-01 -1.55018300e-01 -9.61726248e-01 -8.30167159e-02
5.93575001e-01 -3.73050123e-01 1.00444317e-01 3.57082486e-01
1.94644958e-01 -6.61192358e-01 -2.82612473e-01 -3.01811546e-01
3.91575992e-01 5.50268769e-01 6.53654933e-02 2.64980718e-02
6.90503567e-02 1.40284741e+00 3.40418577e-01 4.40286815e-01
9.35068846e-01 7.62271881e-01 1.81043476e-01 2.96363890e-01
6.64590538e-01 -4.18354481e-01 8.07782412e-01 7.40318775e-01
8.38177055e-02 -3.46279681e-01 -8.72930348e-01 5.50480247e-01
-1.90566778e+00 -9.73685622e-01 -4.39784974e-02 1.93904746e+00
1.16368866e+00 3.84769917e-01 4.20893490e-01 2.86168069e-01
6.76156938e-01 -2.79751495e-02 -8.18502903e-02 -6.21662498e-01
1.06825784e-01 -2.89460998e-02 1.31321907e-01 1.30490160e+00
-4.65047598e-01 1.14825618e+00 7.11441422e+00 2.75734365e-01
-6.91100538e-01 4.21160877e-01 4.50580627e-01 1.32741734e-01
-7.27365851e-01 2.50919133e-01 -7.71485329e-01 6.52974297e-04
9.13028717e-01 -3.94031227e-01 6.71112776e-01 3.66052032e-01
3.30858320e-01 -4.23076093e-01 -1.39431596e+00 3.12366873e-01
4.85055819e-02 -7.70041823e-01 -4.01596874e-01 1.73412859e-01
1.58549502e-01 -1.84204549e-01 -7.09155723e-02 2.45202094e-01
1.06079376e+00 -9.07765329e-01 1.34595287e+00 3.09758604e-01
-3.04931793e-02 -4.27884340e-01 5.26217341e-01 1.24696827e+00
-7.41289318e-01 -3.32814008e-01 3.68837774e-01 -9.51685309e-01
1.18908703e-01 4.80027832e-02 -1.20383763e+00 -9.53034461e-02
5.74541211e-01 2.43511900e-01 -5.55765182e-02 4.20511842e-01
-6.66405678e-01 1.04652762e+00 -5.06031573e-01 -5.33269525e-01
2.47097045e-01 1.40575618e-01 1.01006401e+00 7.36218035e-01
-1.06394842e-01 4.53116477e-01 4.24488097e-01 5.63338578e-01
2.69750684e-01 4.74698246e-02 -7.26358175e-01 8.11811313e-02
6.23490810e-01 6.34024858e-01 -5.17168939e-01 -2.53460348e-01
-2.70862579e-02 3.78248632e-01 1.14680380e-01 3.94884676e-01
-2.71989498e-02 3.63680273e-01 5.80928445e-01 2.93427780e-02
-1.36028742e-02 -1.00935087e-01 9.12747998e-03 -6.71552539e-01
-1.40745938e-01 -1.04297447e+00 3.27935398e-01 -6.45569503e-01
-8.87989163e-01 2.12049603e-01 -4.08670641e-02 -5.35768867e-01
-4.95726854e-01 -1.47094011e-01 -8.04805756e-01 6.36394382e-01
-9.50267673e-01 -6.62347019e-01 5.58488786e-01 4.62002039e-01
5.67268610e-01 -1.08033746e-01 7.02592552e-01 -5.17414749e-01
-3.97384316e-01 1.78107128e-01 -9.38583493e-01 -1.96921706e-01
1.24679662e-01 -1.02935195e+00 -1.50115713e-01 5.94134569e-01
1.95721760e-01 6.78828657e-01 1.22331607e+00 -4.44010973e-01
-1.37576640e+00 -2.94214904e-01 1.59031177e+00 -6.03802145e-01
8.90760839e-01 -5.93375921e-01 -6.01910889e-01 1.15676451e+00
3.71391177e-01 -8.81443143e-01 1.12964141e+00 3.82896423e-01
3.42188887e-02 2.45479852e-01 -9.30389404e-01 6.21352255e-01
1.03318202e+00 -5.22574127e-01 -9.33348954e-01 4.76825148e-01
8.59032452e-01 -2.52654236e-02 -3.95043522e-01 -1.18157364e-01
4.57746059e-01 -1.12891567e+00 3.98305386e-01 -6.39081538e-01
-1.03814512e-01 -2.24223822e-01 -4.48921531e-01 -1.13138866e+00
-2.51199722e-01 -1.34830427e+00 5.36021590e-01 1.10363829e+00
6.97304547e-01 -7.66765475e-01 4.42518473e-01 6.65675342e-01
1.46437138e-01 -9.80294198e-02 -1.49688184e+00 -9.74819735e-02
-1.10264625e-02 -1.13729179e+00 3.31977785e-01 8.78972828e-01
6.93883419e-01 7.94316411e-01 -4.08841759e-01 5.58909178e-01
4.27879333e-01 -4.06794369e-01 8.51754904e-01 -1.39167011e+00
-6.76991940e-01 -3.50591660e-01 -1.57665282e-01 -1.23306715e+00
4.43426281e-01 -7.11828887e-01 7.96125829e-01 -1.12981665e+00
9.72190425e-02 -4.90138233e-01 1.07028946e-01 3.72007966e-01
1.84374988e-01 -3.98249686e-01 1.88709930e-01 3.24827552e-01
-8.84300113e-01 2.67342001e-01 8.82725894e-01 5.10010608e-02
-5.69877803e-01 7.43820429e-01 -1.26944399e+00 9.22192752e-01
6.56965137e-01 -7.05164909e-01 -4.01366770e-01 -1.70358509e-01
8.35110664e-01 7.36715913e-01 1.88046306e-01 -2.88541943e-01
6.23928964e-01 -4.56411183e-01 -5.79709172e-01 -3.51673588e-02
3.19857419e-01 -7.50388980e-01 4.46273945e-02 5.77492595e-01
-1.12267065e+00 -1.62834376e-01 -2.12120369e-01 5.90705156e-01
-7.14247599e-02 -2.85199106e-01 3.16713065e-01 -5.25800623e-02
-2.68269598e-01 -3.54899913e-01 -1.57551897e+00 -1.91153735e-01
1.19738710e+00 -1.34910718e-02 4.22165960e-01 -1.17980576e+00
-1.10180235e+00 6.09441698e-01 3.79361957e-02 3.14774483e-01
4.69020456e-01 -7.70230889e-01 -8.07873189e-01 2.29520008e-01
3.74359190e-02 -8.24035704e-02 1.52774472e-02 9.21771944e-01
2.57491648e-01 2.88493127e-01 4.16149944e-01 -6.89009607e-01
-1.60779488e+00 1.72896143e-02 4.17751670e-01 1.30084679e-01
-6.97764307e-02 1.37689281e+00 4.39355969e-01 -3.25206667e-01
5.71514964e-01 7.32587418e-03 -5.46343386e-01 -2.99731828e-02
5.30820787e-01 -5.77282347e-02 -4.56059873e-01 -9.16440547e-01
-5.12734354e-01 1.19395837e-01 2.67058890e-02 -9.69217479e-01
1.09763575e+00 -5.33830166e-01 -3.79196852e-01 1.19832349e+00
4.90224242e-01 4.96911317e-01 -7.25129783e-01 -5.83957136e-01
1.96671650e-01 -2.27734014e-01 9.45450291e-02 -5.22152781e-01
-3.48913491e-01 3.55922997e-01 1.65589392e-01 8.75766933e-01
6.06541753e-01 8.78585577e-01 -1.34620801e-01 4.19163048e-01
5.48343837e-01 -1.08949804e+00 1.09979063e-01 4.18420732e-01
8.17104459e-01 -1.04215312e+00 -4.21479374e-01 -6.32758200e-01
-8.65521431e-01 6.24637187e-01 2.27299318e-01 5.57931423e-01
7.51322925e-01 3.17267984e-01 4.99259055e-01 -5.00594676e-01
-1.60272539e+00 -2.86159694e-01 -1.64508939e-01 7.19989777e-01
5.49228907e-01 5.66746593e-01 -7.96964169e-02 6.50432706e-01
-6.61958516e-01 -1.99057281e-01 8.63415360e-01 1.00620353e+00
-8.70016634e-01 -1.06409335e+00 -4.19126928e-01 -5.05279116e-02
-5.20267963e-01 4.86638322e-02 -1.05292642e+00 3.07376951e-01
1.18821628e-01 1.80646777e+00 1.31544009e-01 -3.70919257e-01
3.32194835e-01 7.05546662e-02 1.26258984e-01 -1.11412585e+00
-5.88860333e-01 3.06458324e-01 3.56444895e-01 -2.33867571e-01
-8.17429841e-01 -1.13487554e+00 -1.05157626e+00 -7.48436153e-02
-2.28828803e-01 7.28664160e-01 4.35254782e-01 1.36211610e+00
-1.93049446e-01 1.28622115e-01 1.16143835e+00 -4.29186255e-01
-6.43054664e-01 -1.10130692e+00 -8.03372383e-01 -1.68166608e-01
5.34018815e-01 -3.64046872e-01 -8.54934692e-01 -1.23496622e-01] | [12.577692031860352, 8.00041675567627] |
b87e37f2-bb8f-4508-9e86-130cf15b322b | topoal-an-adversarial-learning-approach-for | 2007.09084 | null | https://arxiv.org/abs/2007.09084v1 | https://arxiv.org/pdf/2007.09084v1.pdf | TopoAL: An Adversarial Learning Approach for Topology-Aware Road Segmentation | Most state-of-the-art approaches to road extraction from aerial images rely on a CNN trained to label road pixels as foreground and remainder of the image as background. The CNN is usually trained by minimizing pixel-wise losses, which is less than ideal to produce binary masks that preserve the road network's global connectivity. To address this issue, we introduce an Adversarial Learning (AL) strategy tailored for our purposes. A naive one would treat the segmentation network as a generator and would feed its output along with ground-truth segmentations to a discriminator. It would then train the generator and discriminator jointly. We will show that this is not enough because it does not capture the fact that most errors are local and need to be treated as such. Instead, we use a more sophisticated discriminator that returns a label pyramid describing what portions of the road network are correct at several different scales. This discriminator and the structured labels it returns are what gives our approach its edge and we will show that it outperforms state-of-the-art ones on the challenging RoadTracer dataset. | ['Pascal Fua', 'Mateusz Kozinski', 'Leonardo Citraro', 'Subeesh Vasu'] | 2020-07-17 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/5687_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123720222.pdf | eccv-2020-8 | ['road-segementation'] | ['computer-vision'] | [ 6.04171693e-01 5.82734883e-01 2.01816529e-01 -1.93707824e-01
-6.03926003e-01 -8.89621794e-01 5.64691901e-01 -2.74178535e-01
-2.94295430e-01 6.92911983e-01 -2.78921485e-01 -4.38677967e-01
2.86538988e-01 -1.25381804e+00 -9.97235060e-01 -7.91600108e-01
-6.93273991e-02 5.59601426e-01 9.11690831e-01 -2.52201229e-01
-2.63070837e-02 6.38299167e-01 -1.30349302e+00 2.11999401e-01
7.89049089e-01 9.13749456e-01 -2.29155436e-01 9.75906014e-01
5.01950830e-02 8.25807571e-01 -6.55017972e-01 -5.63962400e-01
7.54352748e-01 -6.55145645e-01 -1.15562367e+00 1.59794390e-01
7.31585741e-01 -3.36482108e-01 -4.46538925e-01 1.21604764e+00
8.77538249e-02 -1.15764275e-01 8.67698789e-01 -1.15781283e+00
-2.44574174e-01 4.19377804e-01 -6.29213691e-01 -2.88228355e-02
-1.54402345e-01 3.23518932e-01 9.25542295e-01 -5.37482023e-01
7.95843482e-01 9.65169370e-01 9.14502442e-01 4.10237432e-01
-1.54765642e+00 -3.32273245e-01 3.25681716e-01 -3.28341156e-01
-1.32136035e+00 -1.48732021e-01 6.93369687e-01 -6.02421165e-01
3.31326365e-01 3.37475061e-01 5.78906953e-01 7.34625101e-01
-2.05609854e-02 6.56889737e-01 1.18289447e+00 -3.04689884e-01
2.71443516e-01 6.64045662e-02 -1.29070744e-01 8.73564005e-01
1.25924930e-01 1.51349321e-01 9.68559310e-02 1.22939311e-01
9.54133928e-01 -1.67415500e-01 -4.43536520e-01 -6.39871061e-01
-1.03356898e+00 7.75183856e-01 9.17528510e-01 7.60344565e-02
-3.22217256e-01 3.78151327e-01 1.02454595e-01 2.64094621e-01
4.96525854e-01 2.61494428e-01 -2.74591476e-01 5.99239588e-01
-1.33695781e+00 3.96395713e-01 9.67421055e-01 6.24288499e-01
1.15824664e+00 -1.36557007e-02 -8.24118555e-02 4.29798663e-01
-3.30398530e-02 2.86314398e-01 -8.75173360e-02 -1.06379247e+00
2.08928093e-01 5.44031739e-01 9.20644403e-02 -9.90751386e-01
-1.70788646e-01 -4.08496082e-01 -6.28055632e-01 9.50893223e-01
8.47787142e-01 -4.21260089e-01 -1.41365421e+00 1.63947916e+00
1.99432001e-01 2.03290299e-01 -3.62938158e-02 7.89388537e-01
3.90445381e-01 6.15937948e-01 -7.36477673e-02 5.09957075e-01
9.27906275e-01 -9.60970581e-01 -2.27926373e-01 -6.21451557e-01
5.88795364e-01 -6.64912105e-01 9.29691494e-01 2.75666952e-01
-1.14072192e+00 -4.69517320e-01 -1.17849731e+00 2.79694468e-01
-6.85538888e-01 3.15460235e-01 2.95442879e-01 6.95914567e-01
-1.19227362e+00 9.24252629e-01 -6.75689101e-01 -4.08961236e-01
6.27786756e-01 3.01804006e-01 -2.65833497e-01 5.95685169e-02
-1.03176904e+00 1.13247967e+00 3.61860603e-01 8.11524019e-02
-1.20153224e+00 -3.22851479e-01 -8.20500135e-01 -1.38089091e-01
5.48400104e-01 -7.28081286e-01 1.03755879e+00 -1.35979962e+00
-1.20745599e+00 1.22884727e+00 6.01392686e-02 -6.21303618e-01
8.74200165e-01 7.98728913e-02 3.61777395e-02 3.18712443e-01
1.07301489e-01 8.73418450e-01 9.69042361e-01 -1.67185342e+00
-7.42088675e-01 -1.98831290e-01 2.71771312e-01 2.07327306e-03
3.44531596e-01 -2.83318460e-01 -2.68616408e-01 -6.76695824e-01
1.81016028e-01 -1.00793982e+00 -3.73218238e-01 3.33575577e-01
-7.21477687e-01 3.36615682e-01 1.01561284e+00 -6.94837391e-01
9.05802965e-01 -2.01249599e+00 7.54441172e-02 4.71271813e-01
2.02465236e-01 3.57399255e-01 -5.38965128e-02 1.37307003e-01
-1.45567685e-01 2.70297080e-01 -9.77658927e-01 -2.72591859e-01
-1.40400946e-01 3.59369069e-01 -4.87435430e-01 7.10706890e-01
6.46527171e-01 9.43830490e-01 -9.03657258e-01 -3.72761607e-01
3.30022603e-01 4.50928897e-01 -3.18524092e-01 1.61789790e-01
-3.67660105e-01 3.28167379e-01 -2.97247797e-01 4.63746727e-01
8.37043703e-01 4.84360382e-02 -1.28082886e-01 -6.15047105e-02
-9.16630849e-02 5.88828586e-02 -1.18544066e+00 1.25123537e+00
-1.41468972e-01 7.52689302e-01 3.70345324e-01 -1.05291879e+00
8.32300544e-01 -4.00674045e-02 1.57254547e-01 -1.48832604e-01
1.00369744e-01 3.07573110e-01 -3.79468352e-02 -2.79474318e-01
2.93891996e-01 -2.19019517e-01 5.41823953e-02 2.47586176e-01
-1.42000183e-01 -4.99859989e-01 -2.92755719e-02 2.20918611e-01
1.32065332e+00 4.60517973e-01 -1.34648487e-01 -3.08283865e-01
4.82545257e-01 3.97145748e-01 3.35825861e-01 6.66417718e-01
-3.67067605e-02 9.97892976e-01 8.53468239e-01 -4.16872889e-01
-1.08522189e+00 -1.00378656e+00 4.00447138e-02 6.84613585e-01
1.12335205e-01 7.43868053e-02 -1.25099754e+00 -1.07574928e+00
-1.41122460e-01 4.60426986e-01 -9.75091994e-01 -2.37683039e-02
-5.99374473e-01 -5.74699104e-01 7.65983641e-01 4.77458477e-01
6.43681109e-01 -9.83200848e-01 -7.57472277e-01 7.54494295e-02
5.67180701e-02 -1.03742063e+00 -2.11684138e-01 5.57925701e-01
-6.50484860e-01 -1.34397030e+00 -9.91001070e-01 -6.68747962e-01
1.06297016e+00 5.99061772e-02 1.28718531e+00 6.10669777e-02
-2.69461095e-01 -1.25652164e-01 -1.41821489e-01 -2.00622037e-01
-5.37689030e-01 2.38700077e-01 -7.37230718e-01 1.70522496e-01
-1.19485678e-02 -5.60543478e-01 -4.55957860e-01 2.40454316e-01
-9.35730875e-01 -1.04045972e-01 6.51100934e-01 7.22274899e-01
6.66271389e-01 2.50909030e-01 7.76579082e-02 -1.43530858e+00
1.53525546e-01 -2.39298791e-01 -7.71614730e-01 2.59388924e-01
-2.74926901e-01 -2.78049484e-02 6.55799687e-01 -1.92135856e-01
-6.24932528e-01 5.63331962e-01 -2.16721028e-01 -2.66479671e-01
-3.89267713e-01 5.84669225e-02 -2.98641741e-01 -4.59521621e-01
7.35913694e-01 9.12390128e-02 -2.07213566e-01 -1.77994251e-01
4.84026521e-01 3.39904994e-01 9.16997015e-01 -2.47753277e-01
1.19752467e+00 6.61920190e-01 3.28810543e-01 -6.51348174e-01
-9.21487510e-01 -1.99453548e-01 -8.17900658e-01 -3.14315140e-01
1.13182354e+00 -6.50041759e-01 -1.42364472e-01 4.79700089e-01
-1.17277169e+00 -8.96316528e-01 -7.16622114e-01 -1.87597096e-01
-6.72672689e-01 3.34888399e-02 -3.85437667e-01 -7.35375404e-01
8.21519196e-02 -1.16096187e+00 1.06151295e+00 2.18677640e-01
8.63323882e-02 -1.01807928e+00 -1.10780923e-02 -1.01401694e-01
4.92465049e-01 8.41133714e-01 7.60514200e-01 -4.39744771e-01
-7.72894561e-01 -3.31631213e-01 -4.11744714e-01 7.09065318e-01
-1.14909269e-01 2.46751055e-01 -1.19384313e+00 8.61605406e-02
-2.14881450e-01 -1.65873215e-01 1.23310375e+00 3.24477553e-01
1.17188251e+00 -2.70297468e-01 -4.13623720e-01 7.69507110e-01
1.62355554e+00 -1.27293214e-01 1.08465409e+00 3.09176475e-01
8.97826791e-01 1.00155342e+00 3.29464674e-01 -2.55547643e-01
2.57306427e-01 4.70411658e-01 9.94719505e-01 -7.55672276e-01
-3.31467092e-01 -3.27591896e-01 3.34603667e-01 -1.60526246e-01
1.37653193e-02 -3.77491474e-01 -9.86262560e-01 6.84657872e-01
-1.81050384e+00 -9.25180554e-01 -4.37090158e-01 2.21252227e+00
6.56114876e-01 4.09972787e-01 3.16623628e-01 3.94366056e-01
6.20672882e-01 2.51690865e-01 -3.53289515e-01 -3.29471260e-01
-2.14522839e-01 3.18365782e-01 1.08687985e+00 7.16495991e-01
-1.47870159e+00 1.19667995e+00 6.71965075e+00 5.68627238e-01
-1.08953810e+00 -3.24762911e-02 8.20302844e-01 5.97226679e-01
-2.57634342e-01 2.49833807e-01 -5.89471459e-01 3.10024351e-01
7.37572312e-01 4.17029619e-01 3.17018986e-01 7.47282922e-01
-1.28058102e-02 -3.56145114e-01 -8.67739379e-01 3.79191965e-01
-5.18439226e-02 -1.09445179e+00 -3.80224250e-02 1.06846392e-01
9.42731917e-01 8.58171359e-02 -1.34697825e-01 -6.14968054e-02
7.43247628e-01 -1.23790646e+00 8.43727708e-01 6.30628765e-01
6.59592211e-01 -6.94918692e-01 8.34245384e-01 4.56617445e-01
-1.06945360e+00 1.02152891e-01 -3.47758979e-01 -1.88411307e-02
1.43557400e-01 5.59381723e-01 -5.92225969e-01 4.69088942e-01
3.63383234e-01 4.81446356e-01 -7.36486793e-01 1.00806522e+00
-4.62723404e-01 6.79637074e-01 -3.87990415e-01 6.10763431e-01
5.77228487e-01 -3.31095964e-01 4.27981824e-01 1.31783080e+00
-1.71255264e-02 -1.44830957e-01 3.94478232e-01 1.01192319e+00
-1.33136466e-01 -2.14307919e-01 -1.03882325e+00 1.80669487e-01
6.23514950e-02 1.17456961e+00 -1.26291597e+00 -4.25021768e-01
-3.32978874e-01 1.19328451e+00 3.06025669e-02 3.73917878e-01
-7.60339260e-01 -5.87698936e-01 4.81130272e-01 5.93472421e-01
4.69917268e-01 -4.71537411e-02 -4.18322593e-01 -8.35261166e-01
-1.00923069e-01 -5.96234918e-01 4.36654575e-02 -8.71752083e-01
-1.11585045e+00 7.20853150e-01 -1.13818705e-01 -1.12618697e+00
-2.91761309e-02 -7.23955989e-01 -8.57199967e-01 1.04861617e+00
-1.76302433e+00 -1.23705542e+00 -5.38002133e-01 4.88504320e-01
2.92837024e-01 2.24910304e-01 6.89105272e-01 1.25859857e-01
-3.54291618e-01 1.70153484e-01 -2.12196559e-01 4.71557766e-01
4.69744265e-01 -1.48370004e+00 5.73527396e-01 1.36818182e+00
7.00752065e-02 1.13999322e-01 7.27397263e-01 -4.94100392e-01
-5.13310313e-01 -1.50301921e+00 8.62922549e-01 -6.13986075e-01
6.55242741e-01 -1.64723709e-01 -9.33821023e-01 7.56303608e-01
7.02446476e-02 3.77710849e-01 -1.23064391e-01 -6.69154108e-01
-7.68085495e-02 -7.17635173e-03 -1.20648992e+00 4.28518236e-01
9.32038605e-01 -4.15371269e-01 -4.16396290e-01 2.57784247e-01
4.41050589e-01 -4.12077755e-01 -2.74331599e-01 4.71208036e-01
2.13638440e-01 -1.20204258e+00 8.79381061e-01 -2.84647793e-01
5.03818750e-01 -7.68049300e-01 1.98935624e-02 -1.27765799e+00
3.10687888e-02 -4.12066430e-01 3.02923769e-01 1.13901222e+00
4.37046140e-01 -5.98006070e-01 9.00939465e-01 2.10195184e-01
-2.11665303e-01 -7.31910765e-01 -6.11967504e-01 -7.35386372e-01
2.61350989e-01 -8.89109597e-02 4.35586751e-01 7.34571218e-01
-8.30204964e-01 2.09058523e-01 -1.90453753e-01 1.75953642e-01
5.77038407e-01 -6.79941848e-02 8.41795444e-01 -1.31751812e+00
-6.23198822e-02 -5.04757941e-01 -3.98375511e-01 -1.00451720e+00
1.17034197e-01 -8.19320738e-01 3.07104737e-01 -1.51932883e+00
-3.60220641e-01 -4.92715478e-01 1.65808186e-01 6.77330256e-01
-3.62244993e-02 6.97207630e-01 -5.71834892e-02 8.70310664e-02
-1.45358682e-01 8.37767646e-02 1.21625221e+00 -3.28669876e-01
-2.50369180e-02 1.99254125e-01 -5.32489061e-01 9.41276491e-01
8.61890018e-01 -6.80830479e-01 -3.50226015e-01 -3.68572265e-01
9.88778248e-02 -3.73882890e-01 9.37026083e-01 -1.01734960e+00
7.71471635e-02 -2.89943591e-02 4.14540082e-01 -3.15569848e-01
1.03916802e-01 -8.97119522e-01 2.59577841e-01 5.23578346e-01
-3.25678647e-01 -9.02434289e-02 -1.18642123e-02 4.02829051e-01
-1.90931633e-01 -3.59874099e-01 1.13827360e+00 -3.75790358e-01
-5.04036009e-01 1.54472560e-01 -3.57725382e-01 6.69884756e-02
1.12113988e+00 -4.13328737e-01 -2.19672948e-01 -3.56460541e-01
-8.24832976e-01 1.34686351e-01 1.00413620e+00 -1.98340297e-01
2.99029231e-01 -8.25452566e-01 -6.55655384e-01 2.62647510e-01
-1.62666693e-01 4.60944206e-01 -2.29157537e-01 6.78713739e-01
-9.16513681e-01 -2.25949716e-02 -1.46301389e-01 -4.76192355e-01
-9.47296500e-01 3.27068478e-01 8.65074337e-01 -3.47018510e-01
-7.76132107e-01 9.14760113e-01 3.40852499e-01 -2.77198315e-01
1.98575377e-01 -2.90172338e-01 -1.11379540e-02 6.04809448e-03
1.86333239e-01 1.42808780e-01 7.04229996e-02 -7.29476869e-01
-3.14364642e-01 7.01185763e-01 4.27405059e-01 -2.30871543e-01
1.19050670e+00 5.42494729e-02 -9.97678339e-02 2.33831003e-01
8.68872404e-01 6.40536919e-02 -1.65619254e+00 -1.80558022e-02
-1.66379839e-01 -4.69717920e-01 1.87676206e-01 -8.42345893e-01
-1.31889880e+00 1.07498002e+00 5.18871009e-01 4.84276116e-01
1.13066220e+00 -1.35836378e-01 6.40986085e-01 1.75260946e-01
2.42293000e-01 -8.71993184e-01 -2.61272579e-01 4.28541362e-01
5.92147768e-01 -1.04832470e+00 -1.53609216e-01 -5.43044865e-01
-5.16500711e-01 1.09936774e+00 2.46248439e-01 -6.72980130e-01
5.67544401e-01 4.60532248e-01 2.25389168e-01 -2.31548458e-01
-1.82180971e-01 -7.83426642e-01 1.54801622e-01 8.12853098e-01
3.53122428e-02 4.62108478e-03 1.36290267e-01 8.68417397e-02
-1.81956083e-01 -7.71661922e-02 5.38636923e-01 8.20593536e-01
-5.18350959e-01 -1.08265555e+00 -2.82607198e-01 3.57838303e-01
-4.31472093e-01 -3.27019542e-02 -8.07981193e-01 9.19901133e-01
4.81771380e-01 6.76832676e-01 -5.71206175e-02 -4.11579818e-01
5.17198563e-01 7.21860006e-02 3.54094863e-01 -6.83058262e-01
-7.34558523e-01 -7.41902143e-02 -9.57281049e-03 -6.05453491e-01
-4.62928176e-01 -3.93030286e-01 -1.10617518e+00 -3.48833591e-01
-3.06382507e-01 -9.12068188e-02 5.49571455e-01 7.93464363e-01
-1.85368672e-01 6.72924042e-01 5.77714920e-01 -1.06686914e+00
-3.05348486e-01 -6.20136082e-01 -4.64398772e-01 2.40886599e-01
5.69783151e-01 -4.26291913e-01 -6.24514699e-01 4.13461119e-01] | [9.943495750427246, 0.18685507774353027] |
961f1663-7159-4cd7-a557-72f745a108ed | improving-low-resource-rrg-parsing-with-cross | null | null | https://aclanthology.org/2022.coling-1.384 | https://aclanthology.org/2022.coling-1.384.pdf | Improving Low-resource RRG Parsing with Cross-lingual Self-training | This paper considers the task of parsing low-resource languages in a scenario where parallel English data and also a limited seed of annotated sentences in the target language are available, as for example in bootstrapping parallel treebanks. We focus on constituency parsing using Role and Reference Grammar (RRG), a theory that has so far been understudied in computational linguistics but that is widely used in typological research, i.e., in particular in the context of low-resource languages. Starting from an existing RRG parser, we propose two strategies for low-resource parsing: first, we extend the parsing model into a cross-lingual parser, exploiting the parallel data in the high-resource language and unsupervised word alignments by providing internal states of the source-language parser to the target-language parser. Second, we adopt self-training, thereby iteratively expanding the training data, starting from the seed, by including the most confident new parses in each round. Both in simulated scenarios and with a real low-resource language (Daakaka), we find substantial and complementary improvements from both self-training and cross-lingual parsing. Moreover, we also experimented with using gloss embeddings in addition to token embeddings in the target language, and this also improves results. Finally, starting from what we have for Daakaka, we also consider parsing a related language (Dalkalaen) where glosses and English translations are available but no annotated trees at all, i.e., a no-resource scenario wrt. syntactic annotations. We start with cross-lingual parser trained on Daakaka with glosses and use self-training to adapt it to Dalkalaen. The results are surprisingly good. | ['Simon Petitjean', 'Tatiana Bladier', 'Kilu von Prince', 'Jakub Waszczuk', 'Laura Kallmeyer', 'Kilian Evang'] | null | null | null | null | coling-2022-10 | ['constituency-parsing'] | ['natural-language-processing'] | [ 2.46526599e-02 4.12955880e-01 6.20646775e-02 -3.38777691e-01
-9.07036006e-01 -8.56312990e-01 2.87008613e-01 2.45922640e-01
-9.64474261e-01 9.14860666e-01 4.70085084e-01 -6.64104998e-01
1.21861257e-01 -8.81936729e-01 -5.72954297e-01 -5.03774881e-01
3.70291173e-02 7.19291031e-01 3.11691105e-01 -4.80781674e-01
-9.24058259e-02 2.25154087e-01 -1.18434668e+00 6.23333007e-02
7.91863322e-01 1.05635099e-01 6.95696533e-01 5.97024798e-01
-4.85700637e-01 2.34151989e-01 -3.83922189e-01 -8.06299686e-01
1.87221661e-01 -2.24642113e-01 -1.17472339e+00 -2.12190196e-01
1.05551086e-01 1.04985379e-01 2.46923417e-01 9.62765276e-01
2.89659649e-01 -1.54738069e-01 2.43433798e-03 -4.67472613e-01
-3.75084430e-01 1.29043007e+00 -1.86733603e-01 1.69263408e-01
2.27847710e-01 -9.99040306e-02 1.44273508e+00 -7.17647195e-01
9.80178535e-01 1.41544390e+00 5.70429206e-01 6.88780069e-01
-1.17493260e+00 -4.62613195e-01 2.07814842e-01 -1.81334391e-01
-9.13682818e-01 -3.59280676e-01 6.34846151e-01 -6.32274151e-02
1.18612874e+00 -9.37848836e-02 3.01649213e-01 9.95994389e-01
-3.93918455e-02 4.51268196e-01 1.20613778e+00 -1.01638722e+00
3.56364064e-02 1.38911843e-01 2.71077245e-01 5.46118081e-01
2.71137506e-01 3.67674828e-02 -3.01775634e-01 -1.18192211e-02
4.85263318e-01 -6.66100025e-01 5.36776148e-02 7.62247592e-02
-1.16800427e+00 9.83675897e-01 -6.15770780e-02 9.13911819e-01
-2.91974932e-01 -1.92004517e-01 7.53758192e-01 3.28492612e-01
5.76345682e-01 3.54534864e-01 -9.66828704e-01 -1.16715454e-01
-7.57875562e-01 -5.78228645e-02 9.98436213e-01 7.92773128e-01
8.52273345e-01 -7.63163567e-02 3.81206810e-01 1.10905004e+00
1.29251063e-01 2.47872591e-01 7.27544785e-01 -6.97200179e-01
9.16354358e-01 4.14830774e-01 -8.29129368e-02 -2.26736471e-01
-4.55445975e-01 1.39059015e-02 -3.53616059e-01 -1.09111145e-01
8.02025616e-01 -4.78025615e-01 -7.41527081e-01 2.18708086e+00
4.56188858e-01 -2.84961641e-01 7.40401983e-01 4.98049140e-01
5.46000242e-01 7.32540131e-01 4.11884785e-01 -3.20399046e-01
1.79816842e+00 -7.88859129e-01 -5.23726881e-01 -6.03582323e-01
1.11872578e+00 -8.92899811e-01 1.37777078e+00 1.33277476e-01
-1.07662594e+00 -5.39669216e-01 -7.80404210e-01 -2.58898973e-01
-6.62469745e-01 2.21383840e-01 5.63733339e-01 7.82646596e-01
-1.10883307e+00 6.48399234e-01 -8.49260747e-01 -6.94375992e-01
-2.18198895e-01 1.74374044e-01 -7.78174698e-01 -3.01273465e-01
-1.54637766e+00 1.05108678e+00 9.59605515e-01 2.78622117e-02
-1.87758833e-01 -3.30122262e-01 -1.02665102e+00 -1.54188082e-01
5.79368353e-01 -1.56461820e-01 1.12896967e+00 -9.19783831e-01
-1.53141630e+00 1.14887452e+00 -1.98783249e-01 -3.07879716e-01
2.24466175e-01 -2.28780717e-01 -2.85208881e-01 -2.74884462e-01
3.75296146e-01 5.46143174e-01 1.95137605e-01 -1.03372502e+00
-7.52969682e-01 -5.08828163e-01 2.57678330e-01 1.75088674e-01
-1.53212801e-01 5.36208987e-01 -3.43231529e-01 -4.08492565e-01
1.26795650e-01 -1.03869963e+00 -3.34738821e-01 -9.39859092e-01
-2.46025890e-01 -5.41363835e-01 3.41478825e-01 -1.11085880e+00
1.21866596e+00 -2.01002622e+00 2.20091552e-01 4.51230258e-02
-3.73247236e-01 3.26266497e-01 -2.46363431e-01 6.50304258e-01
-9.06004533e-02 4.96827126e-01 -4.45207268e-01 -5.55577576e-01
-8.83040056e-02 1.05223143e+00 4.24517989e-02 -2.27475651e-02
4.92318094e-01 8.08740139e-01 -9.39710319e-01 -4.84647214e-01
1.20738754e-02 1.07244544e-01 -4.61881816e-01 1.43575475e-01
-1.03707433e-01 4.67411876e-01 -3.03350836e-01 3.79376590e-01
4.91561979e-01 4.77841705e-01 8.15825641e-01 5.72694615e-02
-6.24631941e-01 7.39393532e-01 -1.19703317e+00 1.72976065e+00
-1.22230780e+00 1.50812700e-01 1.68489054e-01 -1.01853728e+00
8.19137335e-01 3.42192739e-01 4.81452830e-02 -4.51112658e-01
7.79549358e-03 6.93770766e-01 5.77449858e-01 -2.81497061e-01
6.43141150e-01 -4.00912911e-01 -5.30121207e-01 4.46956754e-01
3.81658971e-01 -1.92208558e-01 5.69659531e-01 -1.16711088e-01
1.09738433e+00 6.52790189e-01 6.89543784e-01 -5.45141518e-01
6.94114149e-01 9.98487100e-02 7.16885328e-01 3.18671316e-01
1.49232179e-01 3.02583605e-01 5.83909869e-01 -3.58241856e-01
-1.32838356e+00 -8.62478912e-01 -3.59963387e-01 1.46559107e+00
-3.07078779e-01 -6.91258490e-01 -9.30252790e-01 -8.19325924e-01
-5.10590613e-01 8.58398616e-01 -4.92515147e-01 5.36823928e-01
-1.38014925e+00 -8.25184703e-01 5.77075362e-01 4.52585161e-01
2.40504771e-01 -1.52095056e+00 -6.09457970e-01 7.39800513e-01
-3.62523869e-02 -1.47359788e+00 -4.73980978e-02 6.45615518e-01
-7.20523477e-01 -9.71505463e-01 -3.07760298e-01 -9.69283581e-01
2.09626392e-01 -4.63039130e-01 1.37328804e+00 4.43910286e-02
5.26943468e-02 1.15761630e-01 -6.37079895e-01 -3.14861953e-01
-9.94487345e-01 3.96599323e-01 -3.29837129e-02 -4.26777065e-01
3.65067095e-01 -4.16429043e-01 2.70526737e-01 4.03255224e-02
-9.63855863e-01 -7.45249987e-02 5.92192173e-01 9.04904723e-01
6.02819979e-01 -2.49768034e-01 7.03523099e-01 -1.42079997e+00
4.47084099e-01 -4.45919544e-01 -7.21072316e-01 3.30622107e-01
-1.58939570e-01 3.13628793e-01 1.12524867e+00 -2.45873258e-01
-1.20619655e+00 6.07415214e-02 -9.08077598e-01 3.03983569e-01
-3.98872226e-01 5.34793854e-01 -6.17052734e-01 4.73691016e-01
5.05332291e-01 -1.70622274e-01 -3.75783294e-01 -6.97600007e-01
6.13505840e-01 5.63659847e-01 4.16254640e-01 -9.97667730e-01
6.93351269e-01 -1.33929521e-01 -1.36430308e-01 -9.37808394e-01
-7.38799095e-01 -3.24894607e-01 -1.08097970e+00 3.84546578e-01
1.01461267e+00 -6.38677597e-01 -2.95735121e-01 1.19462818e-01
-1.45653999e+00 -4.99205023e-01 -4.25508469e-01 5.03008306e-01
-4.14313555e-01 5.16340971e-01 -7.10522950e-01 -5.49546301e-01
-2.98170775e-01 -1.11460102e+00 9.02676761e-01 -1.54657960e-02
-1.58149257e-01 -1.40404141e+00 2.59621203e-01 -7.40491599e-02
1.23329923e-01 2.13652980e-02 1.33824801e+00 -1.12727904e+00
-1.10514984e-01 1.04660168e-01 2.55211946e-02 6.01738751e-01
2.47911334e-01 -2.24269599e-01 -8.87816846e-01 -1.08125098e-01
-1.16497844e-01 -3.07603389e-01 3.95972669e-01 6.10086434e-02
3.59761387e-01 -1.39209643e-01 -1.11234048e-02 2.44379327e-01
1.56474781e+00 -2.35341769e-02 4.48205680e-01 4.96714592e-01
6.04493022e-01 1.10136414e+00 6.60692155e-01 -7.37855211e-02
6.73242211e-01 5.14600992e-01 -2.63594389e-02 5.63977882e-02
-6.11989386e-02 -3.14193755e-01 6.03341043e-01 1.25139701e+00
-8.64679366e-02 -1.30889669e-01 -1.03529346e+00 7.88067877e-01
-1.69428551e+00 -4.18310672e-01 -1.86486393e-01 2.15848207e+00
1.06584442e+00 1.95746660e-01 2.45911926e-02 2.43477337e-02
6.25182211e-01 1.58312261e-01 -4.12526913e-02 -1.00569272e+00
-1.40879720e-01 9.15913463e-01 5.82094550e-01 7.20284283e-01
-1.05154979e+00 1.70771968e+00 5.09602880e+00 6.77525163e-01
-8.25847030e-01 4.71460938e-01 2.76730835e-01 5.45351088e-01
-2.92467743e-01 3.92290264e-01 -1.19797289e+00 3.46678197e-01
1.50456965e+00 -5.97895235e-02 3.95526081e-01 6.87998474e-01
2.24077143e-02 -2.17693388e-01 -1.12031651e+00 3.67717832e-01
-2.21113935e-01 -9.49585140e-01 -3.52139503e-01 -4.95892279e-02
2.47574508e-01 3.24007988e-01 -7.29160428e-01 5.86544394e-01
7.92730451e-01 -6.15680993e-01 6.00039721e-01 -2.77223021e-01
7.32284307e-01 -6.65884316e-01 1.17743254e+00 5.95769286e-01
-1.34219027e+00 2.12310210e-01 -5.96883833e-01 8.50399584e-02
6.12349927e-01 3.68148237e-01 -6.60812914e-01 9.66251969e-01
5.76474667e-01 3.14348042e-01 -4.56467032e-01 3.55687141e-01
-7.60821998e-01 1.01709890e+00 -5.69870234e-01 4.00342755e-02
4.29039836e-01 -4.13316160e-01 5.13648510e-01 1.56034410e+00
3.38983446e-01 1.34435471e-03 4.65176433e-01 4.41521376e-01
-9.12741944e-02 7.64777303e-01 -5.84141493e-01 1.98934272e-01
3.57292444e-01 1.46976233e+00 -8.79480720e-01 -3.22148681e-01
-5.96925378e-01 8.39397788e-01 6.50498807e-01 -1.38384536e-01
-3.64857376e-01 -1.72181919e-01 4.23123598e-01 5.12960330e-02
2.91416138e-01 -4.65897679e-01 2.23129280e-02 -1.06303453e+00
1.28333390e-01 -6.21555686e-01 4.78576750e-01 -2.73043156e-01
-1.39642131e+00 1.03492641e+00 2.89664418e-01 -5.91484308e-01
-6.96999967e-01 -9.13835227e-01 -5.58550298e-01 1.37127316e+00
-1.88169193e+00 -1.39918160e+00 3.78403485e-01 2.61721164e-01
7.08538890e-01 8.42805281e-02 1.22500241e+00 1.60256237e-01
-5.08528590e-01 5.09873867e-01 -9.78993997e-02 4.29245263e-01
5.96867561e-01 -1.40311551e+00 7.63477027e-01 1.00967896e+00
4.13298696e-01 6.26883090e-01 4.21795040e-01 -5.95596373e-01
-1.02832758e+00 -9.41925764e-01 1.41903579e+00 -2.34924182e-01
1.14230216e+00 -7.35517621e-01 -1.22418749e+00 8.10843349e-01
3.42727691e-01 4.59798193e-03 6.95907295e-01 4.72345769e-01
-1.15320034e-01 2.64862031e-01 -1.04906404e+00 5.74685574e-01
1.04724026e+00 -3.48914891e-01 -9.29113328e-01 2.23139048e-01
9.10232425e-01 -3.56857389e-01 -9.75351989e-01 4.99303229e-02
3.83692741e-01 -6.04180992e-01 4.83869642e-01 -6.89738393e-01
2.85236984e-01 -3.90199684e-02 -3.95544142e-01 -1.55718315e+00
-9.39488560e-02 -5.15473187e-01 7.83369899e-01 1.67992926e+00
6.57841325e-01 -8.46578062e-01 5.36662221e-01 3.49229485e-01
-4.23023939e-01 -4.95159864e-01 -1.26961303e+00 -8.48291636e-01
7.56150663e-01 -6.12861097e-01 4.56855178e-01 7.25414991e-01
-1.43414199e-01 5.98626971e-01 -4.09291983e-02 1.81799710e-01
3.10186088e-01 -1.04776330e-01 5.71827531e-01 -1.32926345e+00
-4.56272185e-01 1.98992323e-02 -3.23125944e-02 -5.49612045e-01
8.32626641e-01 -1.15061367e+00 3.89966428e-01 -1.29628730e+00
-3.04948032e-01 -8.99408162e-01 7.87240453e-03 7.02489018e-01
-2.36880571e-01 8.27103108e-02 3.90140980e-01 -6.25234023e-02
9.13831145e-02 3.57268006e-02 7.81184733e-01 4.84568179e-01
-3.71726483e-01 -2.17470020e-01 -5.65132141e-01 9.26670194e-01
9.22062516e-01 -7.70253778e-01 6.38809204e-02 -4.79684085e-01
1.38700321e-01 7.50553161e-02 -9.40389633e-02 -5.88423550e-01
-2.11388007e-01 -3.16354707e-02 -1.86928406e-01 -7.54431486e-02
-2.55328100e-02 -6.73257411e-01 -1.77682355e-01 1.77943677e-01
6.83766231e-02 4.92611080e-01 4.14114237e-01 1.51795521e-02
-1.91804782e-01 -1.04062462e+00 7.21272111e-01 -5.55374086e-01
-9.03101206e-01 -6.75434843e-02 -3.90874833e-01 4.05919164e-01
6.00927174e-01 -1.74421981e-01 -7.46930316e-02 2.34907582e-01
-9.39647377e-01 -1.52715482e-02 4.41787362e-01 3.16760361e-01
3.79062295e-02 -9.07960057e-01 -8.57193649e-01 2.87749946e-01
9.42324921e-02 4.06074598e-02 -2.21460592e-02 4.44954365e-01
-3.92288655e-01 2.28192151e-01 -1.20193072e-01 -1.85852319e-01
-1.11132181e+00 4.37663972e-01 -1.37964338e-01 -8.56788933e-01
-5.69138944e-01 5.08958817e-01 -1.13324681e-02 -9.68993187e-01
-3.39739621e-01 -5.07212043e-01 -4.18871701e-01 1.36723876e-01
2.05974337e-02 -1.44984677e-01 3.10394645e-01 -9.93079007e-01
-4.79071081e-01 5.66336334e-01 -5.79184741e-02 -3.27725410e-01
1.49548972e+00 -1.03473917e-01 -2.94222504e-01 5.44708312e-01
9.84274566e-01 8.06565046e-01 -5.93064427e-01 -2.57875502e-01
6.67303801e-01 3.96995991e-02 -2.94162095e-01 -6.19692743e-01
-6.09367013e-01 8.62318277e-01 1.79620817e-01 1.10835396e-01
8.36722016e-01 2.84493685e-01 7.23989785e-01 3.82813096e-01
5.75844765e-01 -1.17673588e+00 -4.26053286e-01 1.07116747e+00
5.15406966e-01 -1.16032875e+00 -3.07328135e-01 -4.49602604e-01
-6.91462636e-01 1.22731519e+00 2.88253725e-01 -2.35534146e-01
5.46620786e-01 4.25991565e-01 2.55932242e-01 8.90432000e-02
-6.10285223e-01 -5.30489087e-01 -2.54943013e-01 5.57494581e-01
8.01529646e-01 2.62906075e-01 -8.84149432e-01 8.58724475e-01
-6.72705770e-01 -3.18435729e-01 6.56967282e-01 8.36544871e-01
-1.50730267e-01 -2.02212882e+00 -2.13438705e-01 4.33312654e-02
-8.28151107e-01 -6.43153846e-01 -7.81063959e-02 1.38442862e+00
3.57729077e-01 8.52985263e-01 8.55588168e-02 -2.47364584e-02
3.70788753e-01 5.08595407e-01 3.99323463e-01 -1.16226780e+00
-8.43690693e-01 1.62670359e-01 6.36265159e-01 -2.50299752e-01
-4.73407805e-01 -7.48898864e-01 -1.38384378e+00 2.41308492e-02
-3.41630042e-01 4.48223829e-01 8.04940164e-01 1.09268045e+00
-1.79139271e-01 3.45755458e-01 3.88102025e-01 -9.24555838e-01
-3.66819561e-01 -1.08865833e+00 -5.18125713e-01 2.66952872e-01
-2.93371469e-01 -3.79023463e-01 -2.57423759e-01 -7.99811184e-02] | [10.447131156921387, 9.96452808380127] |
136d230b-4c0e-4b0e-820e-6c1f68f557d8 | risk-stratification-of-lung-nodules-using-3d | 1704.08797 | null | http://arxiv.org/abs/1704.08797v1 | http://arxiv.org/pdf/1704.08797v1.pdf | Risk Stratification of Lung Nodules Using 3D CNN-Based Multi-task Learning | Risk stratification of lung nodules is a task of primary importance in lung
cancer diagnosis. Any improvement in robust and accurate nodule
characterization can assist in identifying cancer stage, prognosis, and
improving treatment planning. In this study, we propose a 3D Convolutional
Neural Network (CNN) based nodule characterization strategy. With a completely
3D approach, we utilize the volumetric information from a CT scan which would
be otherwise lost in the conventional 2D CNN based approaches. In order to
address the need for a large amount for training data for CNN, we resort to
transfer learning to obtain highly discriminative features. Moreover, we also
acquire the task dependent feature representation for six high-level nodule
attributes and fuse this complementary information via a Multi-task learning
(MTL) framework. Finally, we propose to incorporate potential disagreement
among radiologists while scoring different nodule attributes in a graph
regularized sparse multi-task learning. We evaluated our proposed approach on
one of the largest publicly available lung nodule datasets comprising 1018
scans and obtained state-of-the-art results in regressing the malignancy
scores. | ['Qi Song', 'Sarfaraz Hussein', 'Ulas Bagci', 'Kunlin Cao'] | 2017-04-28 | null | null | null | null | ['lung-cancer-diagnosis'] | ['medical'] | [ 3.16178471e-01 2.55651355e-01 -3.89903367e-01 -3.78354579e-01
-1.08106947e+00 -2.27506250e-01 3.70104581e-01 3.12215894e-01
-3.97721976e-01 5.46521723e-01 3.30876410e-01 -4.47687298e-01
-3.84375364e-01 -8.04083824e-01 -4.71324861e-01 -7.01698244e-01
8.71689022e-02 8.38636100e-01 4.34442550e-01 1.29230559e-01
-2.05414042e-01 6.97800457e-01 -1.00567424e+00 3.98336023e-01
6.78778172e-01 1.12420201e+00 3.42797041e-01 6.15548849e-01
-3.26900072e-02 8.85874391e-01 1.30494654e-01 -6.66539446e-02
2.83901244e-01 -3.17942590e-01 -9.21036065e-01 2.31846482e-01
4.96009231e-01 -2.44998947e-01 -3.30934137e-01 9.01616156e-01
4.32256490e-01 -3.53830636e-01 9.75493968e-01 -6.63699567e-01
-1.96912006e-01 5.23671210e-01 -6.25827014e-01 2.16664776e-01
-3.35668474e-01 6.60448074e-02 1.14680815e+00 -8.27788472e-01
4.40803856e-01 5.18861115e-01 7.04848945e-01 4.27878737e-01
-9.53640521e-01 -2.33501628e-01 -2.90043145e-01 -3.64771783e-02
-1.14019084e+00 -2.41471440e-04 7.21512854e-01 -5.74785173e-01
3.84211540e-01 3.38572025e-01 8.12779546e-01 8.68471920e-01
3.21352512e-01 4.97088552e-01 1.06532860e+00 -9.79154110e-02
-2.96795100e-01 1.20225504e-01 -1.48343980e-01 1.34627998e+00
5.22412062e-01 -2.12691665e-01 1.52799875e-01 -1.90733865e-01
1.02468789e+00 3.73219043e-01 -2.13262904e-02 -6.92257464e-01
-1.55710065e+00 9.44747210e-01 1.06635535e+00 4.90281194e-01
-5.31938195e-01 2.07838669e-01 3.15808475e-01 -1.10710010e-01
4.42593068e-01 2.30038077e-01 -2.29513243e-01 5.20627022e-01
-8.83738399e-01 -1.91126972e-01 8.31355155e-01 3.24369103e-01
7.11104274e-01 -6.45491034e-02 -8.13351810e-01 7.58699834e-01
3.85545164e-01 2.00741768e-01 7.41999984e-01 -5.07555008e-01
2.28979275e-01 8.08344722e-01 -4.38902825e-01 -6.05401158e-01
-8.29976201e-01 -8.93639147e-01 -1.38228762e+00 -8.21307302e-02
2.96546847e-01 3.94538604e-02 -1.21730053e+00 1.30918324e+00
2.36536354e-01 1.31515846e-01 -2.10328192e-01 7.92607248e-01
1.16272330e+00 -1.57990530e-02 1.39729278e-02 1.04856782e-01
1.61715102e+00 -1.07918108e+00 -1.69151619e-01 -1.09901205e-01
1.00798047e+00 -6.63430095e-01 7.97526777e-01 -1.71933085e-01
-7.67017424e-01 -3.57955933e-01 -7.63433695e-01 -2.30108202e-02
6.41246587e-02 6.96465015e-01 7.92272091e-01 4.73188668e-01
-9.87500906e-01 3.98884267e-01 -1.12479663e+00 -4.25812006e-01
8.00037920e-01 6.78620100e-01 -4.21449631e-01 -2.73287386e-01
-7.87769139e-01 1.01450610e+00 3.71553510e-01 -1.35386614e-02
-1.16889298e+00 -7.89878070e-01 -6.52247250e-01 9.58205238e-02
5.98737776e-01 -1.01573873e+00 1.20728588e+00 -3.89310896e-01
-1.14053118e+00 8.87707591e-01 2.85173785e-02 -3.48644853e-01
5.50746500e-01 3.37321192e-01 -3.16419709e-03 1.71396509e-01
2.29395062e-01 4.68506634e-01 8.19100976e-01 -1.00549591e+00
-3.69928479e-01 -3.62138391e-01 -3.18690360e-01 2.00095654e-01
-5.76095641e-01 -4.84743267e-01 -4.51298654e-01 -4.39923286e-01
1.59904703e-01 -1.23263717e+00 -7.47298777e-01 3.02409708e-01
-5.27400553e-01 -9.60908905e-02 8.57497334e-01 -5.85898340e-01
8.04476023e-01 -1.80231547e+00 2.89000154e-01 5.12366652e-01
8.24434519e-01 1.47784367e-01 -5.57224378e-02 -1.67079940e-01
1.05988607e-01 1.14010729e-01 -2.53525078e-01 -1.98768139e-01
-4.10274833e-01 2.11996108e-01 8.16927910e-01 4.81389999e-01
6.01338744e-01 1.30658197e+00 -6.22508466e-01 -1.13651919e+00
3.37148786e-01 4.38661307e-01 -6.92173302e-01 2.91132569e-01
-1.64422970e-02 7.14225352e-01 -9.91811156e-01 7.43957818e-01
2.61940062e-01 -9.89093721e-01 7.20522553e-02 -6.04323268e-01
2.09427834e-01 -4.24856134e-02 -5.65329850e-01 1.71829271e+00
-6.98046505e-01 2.29631081e-01 -3.36713344e-02 -1.05518329e+00
7.16719866e-01 5.34441411e-01 1.09692562e+00 -2.55629122e-01
2.25669891e-01 3.64495397e-01 4.30515081e-01 -4.64871138e-01
-4.54208553e-02 -1.10687442e-01 9.65348184e-02 4.88471270e-01
1.84040800e-01 -4.59193230e-01 -1.08623207e-01 1.07796334e-01
1.46215475e+00 -4.38421339e-01 6.86658084e-01 -3.55655551e-01
6.61305010e-01 1.07847758e-01 2.26311356e-01 5.63565612e-01
-2.71689087e-01 7.67539382e-01 5.72122693e-01 -3.84135962e-01
-1.01263082e+00 -8.16421926e-01 -1.76299974e-01 6.99106812e-01
-4.26867664e-01 1.23237528e-01 -1.61498815e-01 -1.01127660e+00
1.35918975e-01 8.83238092e-02 -1.02962756e+00 -8.10310915e-02
-7.84207880e-01 -1.03371620e+00 4.14807826e-01 5.69029331e-01
3.23523015e-01 -6.92580402e-01 -3.53153914e-01 1.82092354e-01
-1.83896691e-01 -1.09136307e+00 -4.13504601e-01 7.26794422e-01
-1.18144643e+00 -1.22231877e+00 -1.07828891e+00 -8.46645772e-01
8.75796556e-01 3.60582978e-01 1.16786313e+00 1.85119808e-01
-7.36800253e-01 2.13300958e-01 -6.76464364e-02 -2.98611909e-01
-5.14524460e-01 5.84812760e-01 -5.34637809e-01 5.47390655e-02
-2.91018151e-02 -3.12996864e-01 -7.28273213e-01 1.69984296e-01
-7.47668028e-01 1.08039454e-01 1.59863651e+00 9.40121472e-01
8.55396092e-01 -2.39887640e-01 4.67385590e-01 -1.21863377e+00
3.69391650e-01 -6.46828711e-01 -1.46531254e-01 2.42458239e-01
-4.56482738e-01 2.79598296e-01 4.19607460e-01 -2.38217056e-01
-8.56632650e-01 4.92261440e-01 -4.05779444e-02 -3.88509780e-01
2.43520923e-02 7.88632691e-01 3.29121292e-01 -5.52896976e-01
5.60831428e-01 4.28489335e-02 2.73017287e-01 -1.94840208e-01
8.41503739e-02 3.26315522e-01 1.43809587e-01 -1.98894516e-01
1.00919080e+00 5.16695619e-01 7.11971283e-01 -5.27417541e-01
-1.37767541e+00 -8.66758168e-01 -9.07981575e-01 -1.72034115e-01
1.05475724e+00 -9.77223694e-01 -3.92828763e-01 4.98358756e-02
-7.33379304e-01 -1.18838213e-01 -3.15505773e-01 6.91533029e-01
-4.02252764e-01 2.54669398e-01 -6.32605076e-01 -1.29808411e-01
-3.75973284e-01 -1.28330433e+00 1.19816780e+00 -4.22508866e-02
8.55991244e-02 -1.06187475e+00 1.58219621e-01 5.10425329e-01
8.04515779e-01 2.46233970e-01 1.05633068e+00 -1.03751707e+00
-7.16968000e-01 -4.09947395e-01 -7.24072218e-01 8.12337846e-02
5.72860718e-01 -2.82937825e-01 -6.70123279e-01 -3.73010755e-01
3.08234412e-02 -6.93376541e-01 1.25787878e+00 6.09996378e-01
1.43400741e+00 -8.20085872e-03 -5.61109483e-01 7.23962128e-01
1.51821506e+00 -4.75271106e-01 -9.93885007e-03 3.77634503e-02
1.24141264e+00 3.36492419e-01 2.29954422e-01 4.23110336e-01
3.81056100e-01 3.14499617e-01 8.10850263e-01 -4.93231416e-01
-4.94285047e-01 9.08761844e-02 -4.43948716e-01 9.86531198e-01
-4.57670450e-01 -1.19368531e-01 -1.18950367e+00 5.28128922e-01
-1.59045804e+00 -4.81521249e-01 -2.06568658e-01 1.75939000e+00
6.31833673e-01 8.06362182e-02 -2.14204937e-01 -1.60415277e-01
6.08280122e-01 1.88974544e-01 -4.68541414e-01 3.29827607e-01
1.68897584e-01 2.33647257e-01 8.56962681e-01 1.31788269e-01
-1.47677684e+00 5.32875597e-01 5.65193605e+00 6.06031179e-01
-1.21867704e+00 2.26312324e-01 7.40611851e-01 3.48053388e-02
-3.01303923e-01 -3.10727775e-01 -4.35337216e-01 -3.25539201e-01
5.47147095e-01 -9.71056670e-02 -1.92084089e-01 7.17753589e-01
8.26805755e-02 -1.63982308e-03 -1.18963504e+00 6.78595245e-01
1.51836380e-01 -1.29648423e+00 1.98374316e-01 3.68067533e-01
9.19759512e-01 4.58658487e-01 1.42203480e-01 5.80272004e-02
1.42236382e-01 -1.16542196e+00 -1.71156600e-01 5.68849623e-01
8.13096583e-01 -3.02632421e-01 1.08425653e+00 2.14814171e-01
-1.32451200e+00 7.60103241e-02 -2.44195879e-01 5.84107518e-01
-1.52195901e-01 7.41613030e-01 -1.65031481e+00 8.84161234e-01
3.68443221e-01 8.38592231e-01 -1.03074336e+00 1.18776774e+00
3.00830901e-01 6.04338765e-01 -2.22061887e-01 -1.65804610e-01
4.12340313e-01 1.54899076e-01 2.89956123e-01 9.81894970e-01
5.93778551e-01 9.93284956e-03 4.27525520e-01 5.90143442e-01
-3.52910817e-01 2.24389210e-01 -5.02809227e-01 -1.82892278e-01
-1.91033259e-01 1.69272768e+00 -9.84322786e-01 -2.13846192e-01
-7.35840142e-01 7.43668973e-01 3.52627695e-01 -1.51839092e-01
-6.75676763e-01 2.46317759e-01 -2.22489778e-02 2.63749421e-01
3.35753798e-01 -1.40196010e-01 -1.33619636e-01 -1.23874080e+00
-2.36421019e-01 -5.22853553e-01 5.34833491e-01 -2.81065822e-01
-1.45384490e+00 7.49923348e-01 -2.29760870e-01 -1.46612096e+00
-2.58247644e-01 -5.06241620e-01 -7.46270597e-01 7.07973480e-01
-1.97062218e+00 -1.59789801e+00 -8.49703908e-01 7.04590142e-01
4.37734663e-01 -2.54346609e-01 7.61010826e-01 3.08434814e-01
-2.65141696e-01 2.92975813e-01 -1.44281119e-01 3.38600397e-01
8.60724330e-01 -1.33851862e+00 -1.73652202e-01 2.55376697e-01
-5.57772256e-02 -2.19277114e-01 3.62000341e-04 -6.41098320e-01
-1.39543200e+00 -1.64521825e+00 5.78020632e-01 -3.02033335e-01
7.81521261e-01 1.26340643e-01 -9.38235164e-01 6.64758801e-01
-2.89473146e-01 6.66003823e-01 7.73605227e-01 -1.43665001e-01
-5.55445533e-03 -5.03064878e-02 -9.18512225e-01 2.14235604e-01
7.34582543e-01 -4.10996050e-01 -2.70763099e-01 5.49796999e-01
5.25014222e-01 -2.99637556e-01 -1.11019766e+00 8.56503725e-01
4.35011208e-01 -7.13159561e-01 1.05307400e+00 -4.62975621e-01
4.92056549e-01 -2.82263551e-02 -2.04320550e-02 -1.12796950e+00
-7.26032615e-01 3.03975582e-01 1.83946922e-01 3.86941254e-01
7.06628680e-01 -3.49813491e-01 1.36101329e+00 1.50266111e-01
-4.32640612e-01 -1.02101409e+00 -8.09901059e-01 -4.32031214e-01
1.53286815e-01 -2.94728745e-02 1.40172139e-01 6.65993273e-01
-6.42994642e-01 1.51377484e-01 -2.09886670e-01 1.32607654e-01
8.09353590e-01 1.91030174e-01 3.11851174e-01 -1.59856331e+00
-4.14264470e-01 -4.00769830e-01 -4.13839728e-01 -5.58285117e-01
5.81335723e-02 -1.44914949e+00 -6.69451803e-02 -1.91039526e+00
9.17809606e-01 -4.43884313e-01 -7.29617357e-01 5.21469116e-01
-2.15265319e-01 4.74544823e-01 -9.19737741e-02 5.36706746e-01
-6.07788324e-01 6.16446853e-01 1.87681890e+00 -4.19098943e-01
1.83424845e-01 3.86537760e-01 -6.00254059e-01 6.63265288e-01
6.64179206e-01 -7.19912410e-01 -2.13295162e-01 -4.30580735e-01
-1.35968834e-01 4.21529949e-01 7.02360928e-01 -1.16653872e+00
3.42577130e-01 -1.48385182e-01 6.42295778e-01 -7.66548038e-01
3.88872236e-01 -1.04998028e+00 -1.66429415e-01 9.49241042e-01
-2.97119170e-01 -3.38168651e-01 -1.25046283e-01 6.88066661e-01
-4.60836202e-01 -1.32524714e-01 7.45564818e-01 -4.84580189e-01
-2.37793043e-01 1.00517058e+00 -1.77804187e-01 -2.41588369e-01
1.19220555e+00 -9.49896351e-02 2.11775135e-02 -6.88140765e-02
-8.35844755e-01 2.16075093e-01 1.11260049e-01 2.27323081e-02
5.96187651e-01 -1.38481128e+00 -1.16746283e+00 9.73430090e-03
3.46786290e-01 2.41894066e-01 2.23575681e-01 1.19109738e+00
-4.18621749e-01 6.78840101e-01 -3.55492145e-01 -8.72732043e-01
-1.17821538e+00 2.05328763e-01 4.99965340e-01 -9.57142055e-01
-3.15021664e-01 7.57084429e-01 2.45327145e-01 -4.03123438e-01
-3.70422378e-02 -4.84530598e-01 -4.15640295e-01 -5.84046915e-03
-3.18983942e-01 -1.73898898e-02 4.29690331e-01 -3.17072779e-01
-3.94648969e-01 6.14568830e-01 -2.01928496e-01 2.26663604e-01
1.44203901e+00 2.60767728e-01 -1.85612381e-01 2.92732358e-01
1.39810860e+00 -1.67315200e-01 -7.90710151e-01 -4.37111586e-01
1.60258234e-01 -7.60820881e-02 3.66762549e-01 -5.70488274e-01
-1.29246604e+00 8.60530972e-01 6.47978187e-01 4.16579358e-02
9.70873296e-01 4.84169126e-01 6.74800694e-01 6.69366300e-01
-3.42584662e-02 -2.52722770e-01 4.90804911e-01 3.01970571e-01
7.20179379e-01 -1.96895528e+00 2.81837791e-01 -6.81620240e-01
-5.90720773e-01 1.24066412e+00 7.03385770e-01 -2.00816914e-01
9.95314002e-01 5.98840415e-02 2.49941424e-02 -4.79303271e-01
-7.50341177e-01 -5.66487789e-01 8.16653967e-01 3.16480458e-01
7.15126097e-01 2.17417702e-01 -7.26579269e-03 3.62755895e-01
1.93048909e-01 -2.50616781e-02 3.82767290e-01 8.58019054e-01
-5.99836171e-01 -1.04273641e+00 -1.10657224e-02 1.25023139e+00
-5.37849307e-01 -9.80950817e-02 -3.56070668e-01 8.68983507e-01
-2.15020865e-01 2.68671125e-01 -1.42467543e-01 -1.98147029e-01
1.01808175e-01 -1.27791226e-01 4.35346723e-01 -1.02012837e+00
-6.91389799e-01 1.13076381e-01 2.36374829e-02 -3.11313510e-01
-5.34740627e-01 -5.85166097e-01 -8.59909356e-01 3.99789184e-01
-4.41657245e-01 -1.97837614e-02 5.74606776e-01 8.12382340e-01
-2.09156126e-02 1.05807078e+00 7.72499859e-01 -6.97048426e-01
-7.90249109e-01 -9.82953012e-01 -2.92437732e-01 9.26499665e-02
4.15693998e-01 -4.66556996e-01 -2.81526268e-01 -2.92814493e-01] | [15.387168884277344, -2.1347172260284424] |
aa92c38c-91ad-4097-9598-c6b22474d69b | neural-network-compression-by-joint-sparsity | 2210.07451 | null | https://arxiv.org/abs/2210.07451v1 | https://arxiv.org/pdf/2210.07451v1.pdf | Neural Network Compression by Joint Sparsity Promotion and Redundancy Reduction | Compression of convolutional neural network models has recently been dominated by pruning approaches. A class of previous works focuses solely on pruning the unimportant filters to achieve network compression. Another important direction is the design of sparsity-inducing constraints which has also been explored in isolation. This paper presents a novel training scheme based on composite constraints that prune redundant filters and minimize their effect on overall network learning via sparsity promotion. Also, as opposed to prior works that employ pseudo-norm-based sparsity-inducing constraints, we propose a sparse scheme based on gradient counting in our framework. Our tests on several pixel-wise segmentation benchmarks show that the number of neurons and the memory footprint of networks in the test phase are significantly reduced without affecting performance. MobileNetV3 and UNet, two well-known architectures, are used to test the proposed scheme. Our network compression method not only results in reduced parameters but also achieves improved performance compared to MobileNetv3, which is an already optimized architecture. | ['Erik Meijering', 'Antonio Robles-Kelly', 'Syed S. Naqvi', 'Tariq M. Khan'] | 2022-10-14 | null | null | null | null | ['neural-network-compression', 'neural-network-compression'] | ['methodology', 'miscellaneous'] | [ 5.03090084e-01 1.87335253e-01 -5.52691817e-02 -5.67332983e-01
2.09882751e-01 -2.21831813e-01 1.98973611e-01 1.01489089e-01
-9.08563197e-01 6.16308689e-01 -6.16867580e-02 -3.91679645e-01
-2.22950056e-01 -8.44999015e-01 -6.95046902e-01 -3.78372192e-01
-1.26664117e-01 6.96629956e-02 5.88629246e-01 1.30806863e-01
3.31993699e-01 4.51632529e-01 -1.66269028e+00 3.81474823e-01
6.52103603e-01 1.12215352e+00 2.76717693e-01 5.16018748e-01
-1.31238416e-01 7.73130298e-01 -5.35231709e-01 -3.43952686e-01
6.41446114e-01 -4.08312887e-01 -8.53734553e-01 1.03713915e-01
7.46261895e-01 -1.75182536e-01 -4.27515507e-01 1.19442141e+00
3.94696891e-01 -4.48132828e-02 2.28027537e-01 -7.97832012e-01
9.97028947e-02 9.61341143e-01 -4.87496614e-01 3.40869665e-01
-4.13597226e-01 -1.04341932e-01 7.66625702e-01 -4.31162804e-01
6.46174073e-01 1.12815642e+00 7.68884599e-01 3.76096100e-01
-1.14850652e+00 -7.45215833e-01 2.10705236e-01 2.09966421e-01
-1.34897256e+00 -5.09564638e-01 6.09978795e-01 1.97148249e-02
1.13661695e+00 3.92103940e-01 8.38907957e-01 4.62349176e-01
-6.00078888e-02 5.10999978e-01 9.17346537e-01 -5.52139878e-01
3.73906046e-01 -3.40552628e-02 2.81164765e-01 9.32042003e-01
7.31075525e-01 -9.40486267e-02 -4.35023248e-01 2.98318505e-01
7.97916174e-01 -1.06301114e-01 -3.34307760e-01 -2.86761701e-01
-8.16901207e-01 6.81533396e-01 4.51036483e-01 2.43005604e-01
-2.27929577e-01 4.34962302e-01 4.73831952e-01 3.46841604e-01
1.91679001e-01 4.37544912e-01 -5.28901219e-01 -1.60322502e-01
-1.44538176e+00 1.78851500e-01 1.04193771e+00 1.05408871e+00
7.15141475e-01 2.77091384e-01 1.74636617e-02 8.71016860e-01
3.06510270e-01 -1.42106876e-01 4.17140931e-01 -1.16814721e+00
5.79289436e-01 8.97357702e-01 -5.48496425e-01 -1.09502113e+00
-3.49828750e-01 -8.97746980e-01 -7.76977539e-01 2.64665246e-01
2.46175304e-01 -3.05013359e-01 -1.14103055e+00 1.56374729e+00
-2.03912947e-02 3.76541853e-01 8.97117984e-03 6.17921114e-01
7.67549217e-01 2.49419838e-01 -1.04168326e-01 -1.57482386e-01
8.96138072e-01 -1.06823814e+00 -4.99258310e-01 -1.58762023e-01
7.07234263e-01 -8.31071496e-01 7.24840283e-01 5.81511855e-01
-1.28321528e+00 -3.86030197e-01 -1.38062334e+00 -2.41995719e-03
-2.56913185e-01 6.92633539e-02 7.50409424e-01 1.04316676e+00
-1.20876062e+00 1.00396919e+00 -9.37590122e-01 -3.17399144e-01
8.89610112e-01 7.35207558e-01 -1.78708792e-01 -4.16436382e-02
-6.36385083e-01 6.65062547e-01 7.67679155e-01 -9.48221236e-02
-7.29935169e-01 -6.99071050e-01 -6.29307091e-01 4.58433360e-01
3.62506837e-01 -2.71284223e-01 9.99451697e-01 -9.23978508e-01
-1.28913260e+00 4.46693301e-01 -6.62584603e-02 -1.00134361e+00
3.66833538e-01 -1.66328251e-01 -4.43058647e-02 3.31570774e-01
-4.57002610e-01 1.15979266e+00 5.53571284e-01 -9.41319525e-01
-7.44200885e-01 -2.92513873e-02 2.31001899e-01 8.42032656e-02
-9.06291723e-01 -1.61102504e-01 -7.50523925e-01 -7.50262022e-01
3.76021147e-01 -9.56272602e-01 -5.26556849e-01 6.77353889e-03
-3.76723319e-01 8.73199403e-02 1.09460151e+00 -2.45492622e-01
1.37045491e+00 -1.90143728e+00 -1.05884083e-01 4.30182606e-01
2.85732985e-01 6.78266108e-01 -9.96438861e-02 -2.15841401e-02
-9.05341581e-02 4.38799649e-01 -4.88354743e-01 -4.36124444e-01
-3.59743983e-01 4.46749896e-01 8.11266229e-02 2.47836679e-01
2.32719243e-01 5.08474886e-01 -3.81722420e-01 -6.46523416e-01
8.30838829e-02 5.86122692e-01 -1.14668012e+00 -4.83011276e-01
-1.19728968e-01 -7.80056566e-02 -1.10359669e-01 6.30504668e-01
7.70607173e-01 7.36235594e-03 2.83837527e-01 -3.49821478e-01
-1.38372019e-01 2.93078512e-01 -1.15104508e+00 1.70882773e+00
-1.48407534e-01 6.36656463e-01 8.81118253e-02 -1.23876691e+00
8.56065929e-01 1.60944507e-01 4.33700353e-01 -6.35559916e-01
4.57266271e-01 2.46893898e-01 3.62951875e-01 -1.49681687e-01
5.94850361e-01 4.24146622e-01 3.89489383e-01 1.09385103e-01
1.96436241e-01 2.59074748e-01 7.31032252e-01 1.53662398e-01
1.42630422e+00 2.81408895e-02 -8.24645087e-02 -5.49354970e-01
4.05625194e-01 8.29153135e-02 8.59591842e-01 6.99734747e-01
-1.20076694e-01 5.46318531e-01 6.67647362e-01 -2.47810796e-01
-9.37092543e-01 -4.71337557e-01 -2.63178080e-01 5.45141160e-01
8.57877955e-02 -7.27965057e-01 -9.64811206e-01 -5.79203367e-01
-2.66972065e-01 3.92099917e-01 -3.12547386e-01 4.67043519e-02
-7.34415174e-01 -8.50835443e-01 8.30983460e-01 4.55754489e-01
9.33435321e-01 -9.21116292e-01 -1.07021773e+00 2.71715045e-01
2.49481395e-01 -1.26467967e+00 -7.33194426e-02 5.50627351e-01
-1.54805517e+00 -1.08480597e+00 -4.89932477e-01 -8.80531013e-01
9.17921424e-01 2.01123804e-01 9.55026269e-01 4.37691510e-01
-2.81958073e-01 -1.24089569e-01 -3.98010492e-01 -4.02756602e-01
-7.53467754e-02 3.53713602e-01 -3.18779022e-01 -2.16722295e-01
3.47590655e-01 -8.61842930e-01 -5.92191815e-01 1.82808936e-01
-9.94988441e-01 1.32313251e-01 8.61682355e-01 6.20075941e-01
8.07964742e-01 3.79758954e-01 3.80019277e-01 -1.29884708e+00
4.08037305e-01 -1.84111223e-01 -6.48102224e-01 -1.04384664e-02
-8.27951491e-01 1.85419157e-01 5.14255524e-01 -4.01582986e-01
-8.00045431e-01 3.65980417e-01 -1.13856763e-01 -5.45224309e-01
-8.91529918e-02 5.95397174e-01 -9.41997617e-02 -5.06588995e-01
4.94453818e-01 4.76454720e-02 -8.05915669e-02 -5.65388143e-01
-8.63539055e-02 2.40030214e-01 3.16121101e-01 -2.75092095e-01
4.79273766e-01 4.33556229e-01 3.22884113e-01 -9.83621180e-01
-3.50505561e-01 -3.11518908e-01 -4.83173490e-01 2.61907279e-02
4.80331957e-01 -7.89482832e-01 -4.83795702e-01 2.33660698e-01
-1.05517066e+00 -2.29516864e-01 -2.34091625e-01 5.84001064e-01
-1.63882747e-01 2.59769201e-01 -5.31875610e-01 -4.71725255e-01
-3.40273172e-01 -1.23494804e+00 3.37281913e-01 3.20132673e-01
-1.91743493e-01 -6.49029851e-01 -4.40701008e-01 4.29885834e-02
7.28170991e-01 2.50147969e-01 8.96529675e-01 -5.32775939e-01
-7.50425994e-01 -1.04230784e-01 -4.28716451e-01 6.04279041e-01
-2.84538895e-01 -1.07733980e-01 -6.99803352e-01 -2.57829964e-01
4.83274572e-02 -9.49170813e-02 1.35313249e+00 4.82302636e-01
1.43010008e+00 -2.86055088e-01 -2.64038473e-01 1.05702651e+00
1.82390249e+00 3.48984301e-01 8.53519976e-01 3.32669169e-01
6.92584276e-01 4.14649487e-01 2.17659101e-01 3.79635602e-01
-9.43134427e-02 3.45081627e-01 8.15970898e-01 -5.84047027e-02
-3.91202301e-01 7.60254711e-02 -3.55742453e-03 8.91021311e-01
-1.65416166e-01 -3.07787120e-01 -6.71181440e-01 5.56819975e-01
-1.67929232e+00 -8.02515149e-01 -1.57968163e-01 1.90609121e+00
7.42934406e-01 5.01433611e-01 -2.72619337e-01 6.20572269e-01
4.36333627e-01 1.73292398e-01 -4.13489759e-01 -5.96528769e-01
-1.87857926e-01 8.17439735e-01 9.91471529e-01 3.52596194e-01
-1.09048545e+00 9.39450324e-01 6.37865543e+00 7.72680342e-01
-1.18553257e+00 4.36016023e-02 6.21113062e-01 -7.58132160e-01
-3.03376205e-02 1.45738706e-01 -1.12596476e+00 2.57918775e-01
8.00240517e-01 1.91349030e-01 1.24832064e-01 9.36993182e-01
1.95781946e-01 -2.39421412e-01 -9.10497069e-01 8.35789919e-01
-1.04619354e-01 -1.61131477e+00 1.09650798e-01 1.33274093e-01
8.01037550e-01 1.86558604e-01 -8.30047205e-02 2.86862534e-02
1.63825601e-02 -1.04694200e+00 8.20232391e-01 2.26148248e-01
5.75197518e-01 -1.23648131e+00 8.30681920e-01 1.62836850e-01
-1.08273280e+00 -9.61455554e-02 -7.34920204e-01 -1.59375787e-01
-5.06844968e-02 8.61431956e-01 -6.66442394e-01 1.15561642e-01
9.24292743e-01 7.39921868e-01 -7.44103253e-01 1.46321678e+00
-6.92607239e-02 1.11722410e+00 -5.56079984e-01 1.11998888e-02
3.04235131e-01 -4.60186973e-02 5.04678547e-01 1.49709678e+00
2.39536732e-01 -1.55726463e-01 -5.67161776e-02 7.00111091e-01
-3.88963401e-01 2.05532964e-02 -4.10120994e-01 -5.67094311e-02
4.80609030e-01 1.28719330e+00 -1.19026744e+00 -1.44003928e-01
-3.88551086e-01 6.21666431e-01 2.53030628e-01 1.71302482e-01
-7.18384564e-01 -5.53240895e-01 5.37167192e-01 3.42875928e-01
6.95598185e-01 -2.69770473e-01 -6.37845755e-01 -6.89784765e-01
9.05073509e-02 -8.17196727e-01 1.55026512e-02 1.88316274e-02
-3.98487985e-01 7.90728509e-01 -4.08849381e-02 -9.88191545e-01
2.41415069e-01 -4.15562451e-01 -5.29118061e-01 5.99296987e-01
-1.61509919e+00 -6.70586586e-01 -3.88255924e-01 3.83331150e-01
6.30169392e-01 -3.89861554e-01 5.93972445e-01 7.21334815e-01
-8.02190602e-01 8.34262311e-01 -1.91693097e-01 -9.75159407e-02
3.03600788e-01 -7.96602368e-01 1.09766364e-01 1.20859933e+00
1.82234704e-01 7.49015510e-01 6.23162746e-01 -6.01501465e-01
-1.04257917e+00 -1.38113868e+00 7.45782137e-01 3.48121345e-01
7.02461824e-02 -1.40283838e-01 -8.59356165e-01 4.17262733e-01
3.40915412e-01 2.36996096e-02 4.66783106e-01 -1.51697814e-01
-9.74087343e-02 -2.76402384e-01 -1.19987917e+00 5.65897882e-01
1.28588080e+00 1.67394340e-01 -9.86648649e-02 1.21342838e-01
6.38971388e-01 -4.14620757e-01 -5.33524334e-01 6.03776336e-01
4.55916256e-01 -1.19440198e+00 7.88694978e-01 -2.89216727e-01
6.37093961e-01 -3.16284031e-01 -1.79518461e-01 -8.08307409e-01
-8.62892792e-02 -5.18166959e-01 -1.47490501e-01 1.06649637e+00
4.63461280e-01 -4.13408697e-01 1.50178528e+00 2.91653723e-01
-3.70394260e-01 -1.06857717e+00 -7.42309093e-01 -8.25198650e-01
-3.31247091e-01 -4.85823542e-01 4.56221908e-01 6.59654915e-01
-4.64071244e-01 1.96105689e-02 -2.55052269e-01 -6.99219704e-02
5.81327021e-01 -3.09761941e-01 4.43170130e-01 -1.22559214e+00
-4.34263587e-01 -6.49923742e-01 -5.78076124e-01 -1.07559645e+00
-1.59130618e-01 -9.24941838e-01 -6.61452934e-02 -1.32587850e+00
-5.78678399e-02 -6.14393532e-01 -2.09795818e-01 6.68926656e-01
3.37319136e-01 5.63901186e-01 1.89440474e-01 -1.07095949e-02
-5.00146151e-01 9.33608040e-02 1.05998349e+00 1.17071578e-03
-2.67227024e-01 -1.87833473e-01 -7.03157365e-01 9.53507781e-01
1.07645094e+00 -6.89270973e-01 -8.14930916e-01 -7.55071998e-01
1.41165048e-01 -3.74392599e-01 1.76396325e-01 -1.67108834e+00
5.02795100e-01 1.47500321e-01 2.90543556e-01 -8.49468231e-01
2.61111021e-01 -9.79000032e-01 7.55833387e-02 7.06614435e-01
-3.32824856e-01 2.24653557e-01 3.64645600e-01 4.74300802e-01
-4.68940347e-01 -6.45322204e-01 1.02055442e+00 -2.01310143e-01
-7.73403108e-01 3.09312046e-01 -2.33854920e-01 -1.54568285e-01
9.50824201e-01 -5.73406518e-01 -2.27783948e-01 4.05019894e-02
-6.02707207e-01 1.10812120e-01 2.57287562e-01 1.14181846e-01
7.08710015e-01 -8.91945124e-01 -5.16297460e-01 2.37352520e-01
-4.79478240e-01 2.50351995e-01 -1.25442356e-01 7.85370052e-01
-1.12514842e+00 5.81886649e-01 -3.53602648e-01 -5.22105634e-01
-1.70120382e+00 7.91955963e-02 1.98757425e-01 -3.71512443e-01
-6.00503981e-01 1.17179871e+00 -1.85595751e-01 6.46669865e-02
6.84873641e-01 -4.49413300e-01 -3.02527577e-01 -8.20510909e-02
3.31836551e-01 4.58530664e-01 2.99590945e-01 -3.94137472e-01
-2.65775472e-01 3.32044870e-01 -3.48719805e-01 6.23715669e-02
1.59581161e+00 2.43816495e-01 -1.97261259e-01 -2.44980693e-01
1.24485290e+00 -3.38569164e-01 -1.18711030e+00 -1.97259754e-01
2.71921515e-01 -2.63607651e-01 3.87187332e-01 -5.46767056e-01
-1.81765783e+00 8.07435274e-01 7.77086556e-01 -7.55605847e-02
1.35171270e+00 -5.40072680e-01 8.27684343e-01 5.40687859e-01
2.59270698e-01 -1.17940938e+00 -7.98722878e-02 7.63945997e-01
5.22915363e-01 -8.49889338e-01 2.78774500e-01 -7.19524443e-01
-7.00909942e-02 1.06030607e+00 8.05607855e-01 -3.71969640e-01
7.43111253e-01 6.26817048e-01 -3.10610563e-01 -9.32806581e-02
-6.73847497e-01 -2.30573773e-01 -4.07462567e-02 4.18315172e-01
5.13349235e-01 -3.14412028e-01 -7.46794641e-01 3.17678779e-01
-4.10382599e-01 1.65197521e-01 4.54771698e-01 1.31859696e+00
-6.87515736e-01 -1.38375008e+00 4.23274823e-02 9.27749634e-01
-7.60110021e-01 -2.75850534e-01 -2.60356933e-01 7.95542598e-01
5.08238137e-01 8.11572492e-01 1.21039465e-01 -5.26036024e-01
2.89528012e-01 -2.49741450e-01 5.16499996e-01 -7.04847276e-01
-8.33289504e-01 8.37162063e-02 1.31429553e-01 -6.53482318e-01
-6.36733651e-01 -3.22120428e-01 -1.26467395e+00 -3.80794883e-01
-3.90414983e-01 -2.96187103e-02 8.10972631e-01 7.33914375e-01
3.98263425e-01 9.54434276e-01 -1.17616868e-03 -6.60655737e-01
-2.46270508e-01 -8.57938826e-01 -4.22531158e-01 3.22794281e-02
-9.18591470e-02 -4.37523514e-01 -1.30679339e-01 9.58582014e-03] | [8.599573135375977, 3.1038525104522705] |
76844ac4-df6f-4b3e-9d8b-279dc29ebfbf | modeling-dynamic-environments-with-scene | 2305.17537 | null | https://arxiv.org/abs/2305.17537v4 | https://arxiv.org/pdf/2305.17537v4.pdf | Modeling Dynamic Environments with Scene Graph Memory | Embodied AI agents that search for objects in large environments such as households often need to make efficient decisions by predicting object locations based on partial information. We pose this as a new type of link prediction problem: link prediction on partially observable dynamic graphs. Our graph is a representation of a scene in which rooms and objects are nodes, and their relationships are encoded in the edges; only parts of the changing graph are known to the agent at each timestep. This partial observability poses a challenge to existing link prediction approaches, which we address. We propose a novel state representation -- Scene Graph Memory (SGM) -- with captures the agent's accumulated set of observations, as well as a neural net architecture called a Node Edge Predictor (NEP) that extracts information from the SGM to search efficiently. We evaluate our method in the Dynamic House Simulator, a new benchmark that creates diverse dynamic graphs following the semantic patterns typically seen at homes, and show that NEP can be trained to predict the locations of objects in a variety of environments with diverse object movement dynamics, outperforming baselines both in terms of new scene adaptability and overall accuracy. The codebase and more can be found at https://www.scenegraphmemory.com. | ['Li Fei-Fei', 'Roberto Martín-Martín', 'Silvio Savarese', 'Jiajun Wu', 'Ruohan Zhang', 'Emily Jin', 'Chengshu Li', 'Tanmay Agarwal', 'Michael Lingelbach', 'Andrey Kurenkov'] | 2023-05-27 | null | null | null | null | ['link-prediction'] | ['graphs'] | [ 4.57137916e-03 2.51102328e-01 -1.04179382e-01 -2.74542183e-01
7.87028596e-02 -5.45819581e-01 5.58384478e-01 4.63357687e-01
-2.67360151e-01 6.25552297e-01 2.95970887e-01 7.56533816e-02
-2.47823179e-01 -1.21321070e+00 -1.06355727e+00 -3.21354449e-01
-8.60190153e-01 1.17826724e+00 6.41308308e-01 -2.73106903e-01
-1.94047421e-01 3.72922421e-01 -1.54730678e+00 5.51583013e-03
5.59122860e-01 7.12700486e-01 6.96141243e-01 1.00447428e+00
3.59261602e-01 1.25243950e+00 -2.14749441e-01 -6.61858767e-02
3.00206810e-01 -8.78222063e-02 -9.23355520e-01 -7.57647008e-02
1.32599592e-01 -6.74140692e-01 -9.90298927e-01 7.54460335e-01
3.23126376e-01 6.53379440e-01 4.20892835e-01 -1.68076861e+00
-8.14875364e-01 8.59836102e-01 -5.82720935e-02 4.30150211e-01
7.84173250e-01 4.87979650e-01 1.10576272e+00 -2.44841710e-01
9.68187809e-01 1.33762980e+00 5.48409522e-01 2.46709511e-01
-1.33183241e+00 -3.16517621e-01 7.19818294e-01 7.98251808e-01
-1.26232874e+00 -3.85804743e-01 7.23376513e-01 -3.69041950e-01
1.76791012e+00 1.62691459e-01 1.07034659e+00 1.20042729e+00
2.62482256e-01 8.32747102e-01 4.36810404e-01 1.67443052e-01
4.75075096e-01 -2.10186899e-01 1.10525951e-01 1.17798138e+00
2.39793360e-01 2.29407042e-01 -6.71108544e-01 -2.16098785e-01
5.80823541e-01 2.43712083e-01 -3.18948179e-01 -9.26013768e-01
-1.23768771e+00 5.25721550e-01 9.85292196e-01 -1.95215955e-01
-6.66036129e-01 6.66487813e-01 3.78532633e-02 2.38530546e-01
1.33295611e-01 3.22255850e-01 -5.52978873e-01 -1.64201781e-01
-1.03217833e-01 4.36087996e-01 1.18875825e+00 1.20348036e+00
6.73917830e-01 -1.90286383e-01 7.02577084e-02 3.21275622e-01
4.62036997e-01 6.53683126e-01 2.96647161e-01 -8.62796307e-01
5.79972982e-01 9.99603748e-01 2.11516514e-01 -1.33156180e+00
-8.90469491e-01 -2.38305762e-01 -4.35477704e-01 -6.33021221e-02
-1.71338152e-02 1.05623323e-02 -9.71619129e-01 2.04962683e+00
5.92341423e-01 6.73573792e-01 9.80049651e-03 8.83437455e-01
9.20532346e-01 7.55606472e-01 1.46738857e-01 2.01133057e-01
9.29832160e-01 -1.35985374e+00 -4.57981080e-01 -7.80430853e-01
6.31907284e-01 1.64722517e-01 6.48745537e-01 -2.84125388e-01
-1.19019508e+00 -3.06049049e-01 -9.00805354e-01 6.13185531e-03
-7.92069852e-01 -5.58110714e-01 8.08978796e-01 -1.32457852e-01
-1.54228401e+00 7.93203175e-01 -1.47097063e+00 -9.77840066e-01
3.60854298e-01 6.72675371e-01 -4.04052168e-01 -2.05644630e-02
-8.89006436e-01 1.02376890e+00 5.53367198e-01 1.15596682e-01
-1.33543491e+00 -4.99238253e-01 -1.25768971e+00 2.79284179e-01
6.06193900e-01 -1.09233797e+00 1.29297447e+00 -4.09932643e-01
-1.13567424e+00 5.22532046e-01 -5.35022514e-03 -6.70751691e-01
2.92892754e-01 -1.06849745e-01 -3.79530698e-01 6.65252209e-02
2.68508464e-01 8.30897987e-01 4.14883554e-01 -1.19007111e+00
-7.85133004e-01 -4.02716637e-01 1.87346786e-01 4.77662772e-01
6.81160986e-02 -5.36786497e-01 -5.47306001e-01 -3.30897897e-01
-1.87602695e-02 -1.20442486e+00 -4.35112149e-01 -2.07189564e-02
-5.37009597e-01 -1.91127360e-01 7.52346516e-01 -5.95854282e-01
9.80704129e-01 -1.69571066e+00 4.86798316e-01 2.95602500e-01
3.76935184e-01 -2.49341711e-01 -5.76213539e-01 6.01839900e-01
1.90460756e-01 -3.25923443e-01 -1.80645697e-02 -4.30572093e-01
2.64764309e-01 3.99219275e-01 -1.55138085e-02 3.89507949e-01
-1.87521726e-01 1.52396822e+00 -1.38716924e+00 -1.03718385e-01
1.85187429e-01 3.32790166e-01 -5.89793503e-01 1.77637056e-01
-7.93119013e-01 2.86149919e-01 -7.17712164e-01 5.26681066e-01
2.12026283e-01 -7.83075631e-01 3.55728567e-01 1.46353915e-01
3.09229881e-01 3.83713752e-01 -1.21410537e+00 1.85843933e+00
-5.00719957e-02 5.65763652e-01 -3.33499387e-02 -7.41400123e-01
4.98541474e-01 -3.77155766e-02 5.41065812e-01 -7.09811270e-01
-5.79275899e-02 -2.31998384e-01 -1.37848809e-01 -5.23445904e-01
4.91319656e-01 7.74174690e-01 -1.05445124e-01 4.32636589e-01
-5.40935993e-02 1.06686903e-02 3.32375467e-01 4.55939889e-01
1.91947556e+00 6.88982308e-02 1.27224550e-01 -1.06141143e-01
-1.03726611e-01 4.34956342e-01 3.26111823e-01 9.41260695e-01
-2.24832207e-01 -7.68874139e-02 1.59193128e-01 -7.75360823e-01
-7.52388120e-01 -1.19323146e+00 6.49120450e-01 1.17556810e+00
6.60552740e-01 -4.97795761e-01 -4.63048339e-01 -5.31460106e-01
3.10538739e-01 7.47441530e-01 -5.93425393e-01 -5.00750721e-01
-6.05112135e-01 -4.08840060e-01 -1.43577889e-01 7.76124895e-01
4.83874619e-01 -1.48539209e+00 -1.22129643e+00 3.76087457e-01
-8.12710747e-02 -1.24791443e+00 -4.54125345e-01 4.20771152e-01
-6.09645307e-01 -1.32093489e+00 -6.42399210e-03 -1.02024591e+00
7.52600491e-01 1.80163473e-01 1.47051048e+00 2.66316086e-01
-4.30990070e-01 1.07420075e+00 -2.21481815e-01 -2.31078386e-01
-2.12449074e-01 3.00878793e-01 1.76125363e-01 -4.48217630e-01
1.65500745e-01 -8.34457397e-01 -6.52141690e-01 1.83974564e-01
-3.55321825e-01 4.73023474e-01 3.23473275e-01 4.10066247e-01
5.25391698e-01 4.35314536e-01 7.20067248e-02 -6.13143384e-01
3.12423080e-01 -8.55012298e-01 -9.10771489e-01 3.95510525e-01
-3.79090399e-01 2.10760623e-01 4.56708074e-01 -4.31433260e-01
-8.00868571e-01 3.36995095e-01 5.42021215e-01 -2.98582524e-01
-1.66914642e-01 4.70019996e-01 -9.99554396e-02 4.34655435e-02
3.73186618e-01 1.58615053e-01 -3.47984403e-01 -2.12579057e-01
4.67625141e-01 -1.96242869e-01 7.42117167e-01 -4.61520731e-01
9.44486856e-01 4.16631639e-01 1.63102254e-01 -3.25789273e-01
-4.50554729e-01 -3.93617719e-01 -6.47964537e-01 -2.71676779e-01
7.12155521e-01 -9.60740924e-01 -1.20682609e+00 4.22004014e-01
-1.02010167e+00 -1.56187725e+00 -4.08803493e-01 4.00029212e-01
-6.85445249e-01 -2.42053241e-01 -6.89746141e-01 -4.73354161e-01
2.22025793e-02 -9.64911640e-01 1.08095860e+00 2.65967667e-01
-3.39399368e-01 -1.22170258e+00 2.43084788e-01 -2.49454491e-02
3.14539582e-01 4.34290677e-01 9.48964655e-01 -6.92857802e-01
-1.28119540e+00 6.84875995e-02 1.01158768e-01 -8.11566353e-01
1.27147377e-01 -4.85531747e-01 -5.23311496e-01 -5.51285982e-01
-5.98157108e-01 -1.67702809e-01 7.32120216e-01 4.05464441e-01
8.32407475e-01 -6.63130462e-01 -1.14303958e+00 5.53576589e-01
1.28398454e+00 2.81164348e-01 3.70441452e-02 4.50962186e-01
7.19553649e-01 3.30299705e-01 2.80374110e-01 5.96472383e-01
1.23033381e+00 8.16797256e-01 1.05595124e+00 1.84549272e-01
-3.27288270e-01 -6.82643712e-01 2.65395939e-01 4.08143371e-01
7.50382692e-02 -7.48799741e-01 -1.05249727e+00 8.03252518e-01
-2.43867302e+00 -1.00394237e+00 1.82771027e-01 1.68094218e+00
2.43167117e-01 -5.89072285e-03 1.46175340e-01 -4.11808938e-01
5.40203452e-01 3.27512324e-01 -1.11993515e+00 2.20149949e-01
7.38908127e-02 -3.11062157e-01 6.06296480e-01 7.82771587e-01
-1.16852486e+00 1.15951622e+00 6.09670544e+00 -7.39766583e-02
-4.89694864e-01 2.52545066e-02 4.23286051e-01 -3.91171634e-01
-1.53568268e-01 -1.88287079e-01 -6.87942028e-01 3.55895907e-01
8.44676614e-01 -1.08216792e-01 1.15371084e+00 1.09773409e+00
-2.28928532e-02 -2.16570765e-01 -1.60424125e+00 8.76729727e-01
5.74850589e-02 -1.34870017e+00 -3.00544333e-02 9.43828747e-02
7.55559087e-01 5.63915074e-01 7.89980218e-02 4.07223463e-01
1.35864830e+00 -8.85116279e-01 8.08888197e-01 6.20411634e-01
-5.93255498e-02 -3.03148270e-01 2.28523657e-01 4.42689031e-01
-1.76612592e+00 -3.90039951e-01 -1.85699522e-01 -3.50815684e-01
3.86083096e-01 -1.70140699e-01 -1.21615124e+00 2.25741744e-01
9.53922093e-01 9.95215476e-01 -7.42247462e-01 1.06647575e+00
-3.37172180e-01 1.62549272e-01 -7.16618836e-01 -2.56331116e-01
1.60288930e-01 7.27028698e-02 8.04246008e-01 6.16812110e-01
3.24457765e-01 4.62184370e-01 6.90007448e-01 9.41521704e-01
-1.58302367e-01 -4.74541247e-01 -9.41450119e-01 7.48150274e-02
8.02845597e-01 1.05038786e+00 -1.06601536e+00 -3.29198211e-01
-2.14507133e-01 1.22764754e+00 8.45592678e-01 6.13970280e-01
-8.93170893e-01 3.05908829e-01 7.34253228e-01 1.26863688e-01
3.96480113e-01 -3.00510854e-01 3.05800587e-01 -9.04947460e-01
-3.65907103e-02 -5.29594123e-01 7.00390220e-01 -1.05349708e+00
-8.61375809e-01 5.03956199e-01 1.70963943e-01 -4.79816824e-01
-3.35657954e-01 -3.51589769e-01 -5.86611629e-01 1.90428808e-01
-1.37332606e+00 -1.39148796e+00 -7.20814586e-01 7.29146957e-01
5.72695673e-01 -6.02877364e-02 8.12972069e-01 -7.50909373e-02
-6.14650905e-01 1.50362179e-01 -1.78076208e-01 3.89418006e-02
-2.09491365e-02 -1.19755185e+00 9.55502808e-01 7.73792803e-01
4.85566944e-01 1.27944946e-01 6.87425554e-01 -1.13756514e+00
-1.79649496e+00 -1.45728803e+00 6.52077615e-01 -8.49322915e-01
7.96113431e-01 -4.68552202e-01 -6.50905609e-01 1.54652214e+00
1.17153265e-02 2.71930188e-01 1.56909570e-01 -3.24786268e-02
-2.00266074e-02 1.79720342e-01 -8.80519688e-01 8.11470628e-01
1.98168170e+00 -3.03989857e-01 -2.90096521e-01 5.86581588e-01
9.82233167e-01 -6.43834293e-01 -7.19729722e-01 3.11202884e-01
2.33498544e-01 -6.03916585e-01 1.21185565e+00 -7.94383764e-01
-1.00346273e-02 -2.48776242e-01 -3.23296547e-01 -1.66313136e+00
-8.80407691e-01 -5.23579538e-01 -7.80604899e-01 7.93489993e-01
3.88406456e-01 -7.84528196e-01 1.02400339e+00 1.06433523e+00
-1.66065507e-02 -5.48391759e-01 -7.14654446e-01 -7.03360081e-01
-8.75993788e-01 -2.07850188e-01 1.11868083e+00 6.45204067e-01
6.31228648e-03 4.36119825e-01 -7.33762011e-02 8.59230459e-01
6.50215089e-01 2.79160112e-01 7.80925632e-01 -1.24993360e+00
-3.71266365e-01 -3.50656033e-01 -6.48050427e-01 -1.12531781e+00
5.59832811e-01 -1.00224984e+00 1.29139319e-01 -2.14253330e+00
3.48609954e-01 -5.29184163e-01 -1.49042577e-01 8.13939571e-01
-6.28016144e-02 -3.15051228e-01 2.56624281e-01 1.12476349e-02
-1.29855096e+00 6.75791442e-01 9.59488571e-01 -6.50410414e-01
-5.46293795e-01 -2.35258013e-01 -1.04141064e-01 8.36697876e-01
6.56263888e-01 -3.13616216e-01 -7.19064891e-01 -8.80537033e-01
3.75201166e-01 1.37989238e-01 7.91423321e-01 -1.26100862e+00
7.74347425e-01 -2.54064322e-01 4.81185794e-01 -6.91262066e-01
7.79393554e-01 -1.22853625e+00 7.38071382e-01 7.51888216e-01
-4.32062089e-01 5.52951396e-01 2.04024971e-01 1.04248226e+00
4.57403839e-01 3.21377546e-01 -1.10282712e-02 -3.57465833e-01
-1.49648261e+00 7.42037237e-01 -7.16470480e-02 -1.97302520e-01
1.32857764e+00 -8.22689608e-02 -6.23691738e-01 -6.50086522e-01
-8.81492317e-01 8.83985579e-01 6.84200406e-01 6.17666125e-01
5.84339082e-01 -1.32452679e+00 -3.31868231e-01 1.54716298e-01
3.81868593e-02 2.99719572e-01 1.16730042e-01 4.92790610e-01
-4.12075311e-01 3.38728547e-01 -2.61684265e-02 -4.39323366e-01
-1.04898036e+00 9.80994701e-01 3.98647606e-01 -4.00571167e-01
-9.24130321e-01 9.92809176e-01 3.53140563e-01 -5.84905982e-01
4.26131278e-01 -5.35224319e-01 -2.70796388e-01 -3.35415125e-01
1.84311762e-01 5.29502213e-01 -3.01641554e-01 -6.77739024e-01
-4.59362745e-01 2.89835334e-01 2.16369674e-01 2.19524965e-01
1.77376556e+00 -2.73140639e-01 -3.42890085e-03 4.02165622e-01
9.55112636e-01 -5.83401620e-01 -1.47809601e+00 -3.52302909e-01
1.03174776e-01 -3.93564373e-01 -2.06862450e-01 -7.88910151e-01
-1.07536447e+00 8.81635174e-02 5.45426190e-01 3.00483435e-01
7.53997266e-01 4.95418817e-01 9.03582454e-01 9.25224006e-01
8.75728488e-01 -9.11659896e-01 4.18422103e-01 3.48945677e-01
8.15349221e-01 -1.23049867e+00 -1.35171652e-01 -2.82110572e-01
-4.72839117e-01 5.26065886e-01 9.24185455e-01 -1.56667307e-01
7.14333534e-01 2.88580626e-01 -6.05949879e-01 -5.72521329e-01
-1.17903554e+00 -3.11836034e-01 -2.03075483e-02 9.44491148e-01
-5.93095899e-01 2.67834127e-01 9.81384933e-01 2.91744024e-01
-3.98966402e-01 -2.23341390e-01 1.08781442e-01 1.02973366e+00
-4.46664393e-01 -5.85997939e-01 -7.88898021e-02 5.45171261e-01
3.95053506e-01 1.11158915e-01 -3.45916092e-01 5.63475728e-01
-1.60518304e-01 1.00886488e+00 4.05860364e-01 -4.76172775e-01
4.14617538e-01 -2.13620409e-01 4.16103631e-01 -6.82499588e-01
-2.19379693e-01 -6.34584963e-01 2.43861869e-01 -1.17944217e+00
-2.35086739e-01 -5.54805338e-01 -1.46894097e+00 -4.39878762e-01
-8.67538303e-02 -8.42301399e-02 1.94110125e-01 5.74863374e-01
6.86409295e-01 8.40924442e-01 1.35867193e-01 -1.08550000e+00
-5.91376871e-02 -5.59083879e-01 -4.24741536e-01 6.33725345e-01
5.22396743e-01 -8.97460103e-01 5.01666293e-02 5.99144995e-02] | [4.841824531555176, 0.5528905987739563] |
4c163752-36a7-4c89-8809-f56656d90cf3 | dcase-2021-task-3-spectrotemporally-aligned | 2106.15190 | null | https://arxiv.org/abs/2106.15190v1 | https://arxiv.org/pdf/2106.15190v1.pdf | DCASE 2021 Task 3: Spectrotemporally-aligned Features for Polyphonic Sound Event Localization and Detection | Sound event localization and detection consists of two subtasks which are sound event detection and direction-of-arrival estimation. While sound event detection mainly relies on time-frequency patterns to distinguish different sound classes, direction-of-arrival estimation uses magnitude or phase differences between microphones to estimate source directions. Therefore, it is often difficult to jointly train these two subtasks simultaneously. We propose a novel feature called spatial cue-augmented log-spectrogram (SALSA) with exact time-frequency mapping between the signal power and the source direction-of-arrival. The feature includes multichannel log-spectrograms stacked along with the estimated direct-to-reverberant ratio and a normalized version of the principal eigenvector of the spatial covariance matrix at each time-frequency bin on the spectrograms. Experimental results on the DCASE 2021 dataset for sound event localization and detection with directional interference showed that the deep learning-based models trained on this new feature outperformed the DCASE challenge baseline by a large margin. We combined several models with slightly different architectures that were trained on the new feature to further improve the system performances for the DCASE sound event localization and detection challenge. | ['Woon Seng Gan', 'Douglas L. Jones', 'Ngoc Khanh Nguyen', 'Karn Watcharasupat', 'Thi Ngoc Tho Nguyen'] | 2021-06-29 | null | null | null | null | ['direction-of-arrival-estimation', 'sound-event-localization-and-detection'] | ['audio', 'audio'] | [ 1.19503532e-02 -8.30161095e-01 6.24643028e-01 -2.31730133e-01
-1.45311666e+00 -7.34983146e-01 2.56504953e-01 3.57197016e-01
-4.19741362e-01 2.65680432e-01 5.30884385e-01 -2.66119093e-01
-2.59024620e-01 -4.43810940e-01 -5.47849655e-01 -7.20629930e-01
-5.05986989e-01 -4.90411907e-01 4.46719944e-01 3.34985495e-01
1.65408477e-01 2.98310310e-01 -1.67005110e+00 4.40853298e-01
3.50204051e-01 1.44090772e+00 3.23707134e-01 1.21177745e+00
3.09086293e-01 5.87962449e-01 -8.98342311e-01 3.33887428e-01
1.35928124e-01 -7.19101429e-01 -2.37262264e-01 -5.93276799e-01
5.25968492e-01 -3.39603186e-01 -3.48319024e-01 7.76244581e-01
1.05550802e+00 4.37738419e-01 4.91796881e-01 -1.14327598e+00
-1.86032932e-02 5.80838323e-01 -1.63846344e-01 9.38850820e-01
7.13860393e-01 -2.48468563e-01 9.66754138e-01 -1.25787377e+00
-1.58234134e-01 7.84199536e-01 8.95769417e-01 -1.67077735e-01
-8.78631532e-01 -9.70367312e-01 -6.24009073e-02 5.81328332e-01
-1.42891228e+00 -6.14986718e-01 7.55368769e-01 -4.08984959e-01
1.39061153e+00 4.30375189e-01 4.85218346e-01 8.26022804e-01
1.86483730e-02 2.33263507e-01 8.10201824e-01 -5.69437385e-01
2.30486438e-01 -3.46619576e-01 8.09432119e-02 3.56961936e-01
-2.41073087e-01 4.72664475e-01 -1.01023054e+00 -4.36041355e-01
4.49152976e-01 -2.67113775e-01 -7.75251031e-01 2.53947616e-01
-1.36201155e+00 4.46860790e-01 3.59512091e-01 4.97931957e-01
-2.99071550e-01 2.03496695e-01 1.91544876e-01 2.64455557e-01
5.19398689e-01 4.08085734e-01 -6.20261312e-01 -4.47222531e-01
-7.85435855e-01 1.89084426e-01 8.22661817e-01 2.58819073e-01
4.00337219e-01 2.28150085e-01 -3.43205899e-01 1.09541297e+00
2.83169359e-01 8.42288673e-01 4.32310641e-01 -7.44331717e-01
4.88608181e-01 -2.68890500e-01 2.01426998e-01 -1.24604785e+00
-7.92566717e-01 -8.27455580e-01 -4.40753907e-01 -1.39972910e-01
6.72095776e-01 -5.71954846e-01 -6.31199896e-01 1.99022937e+00
4.45971996e-01 9.62107062e-01 -3.76092702e-01 1.12353826e+00
5.45600414e-01 7.92895555e-01 -1.71984151e-01 -3.36355805e-01
1.32078815e+00 -6.82748437e-01 -8.28122675e-01 -2.15668857e-01
2.73904055e-01 -9.61933851e-01 6.82459474e-01 2.97967404e-01
-9.00653958e-01 -6.27108693e-01 -1.02139509e+00 3.16562623e-01
-3.14467281e-01 1.64863333e-01 1.89684629e-01 7.17895627e-01
-6.35801256e-01 2.66452432e-01 -8.40320349e-01 2.63816804e-01
-2.53557533e-01 -7.25461468e-02 -2.34001756e-01 4.39520657e-01
-1.40894306e+00 2.94923246e-01 -6.59728348e-02 1.92178324e-01
-8.77913058e-01 -1.03614676e+00 -6.94242716e-01 4.39083666e-01
-2.10842341e-02 -2.60204941e-01 1.59852040e+00 -3.38092893e-01
-1.30808258e+00 2.39062592e-01 -4.12670106e-01 -5.25958300e-01
-9.34420973e-02 -3.09429437e-01 -1.09486318e+00 2.06932440e-01
3.36735845e-01 -2.19279513e-01 1.07855403e+00 -7.06148088e-01
-1.20284998e+00 -3.59499827e-02 -3.16269010e-01 3.97758484e-02
-3.26863527e-01 2.18283340e-01 -7.01725036e-02 -9.47350800e-01
4.08721775e-01 -5.82707703e-01 2.11133555e-01 -3.37265998e-01
-3.37008536e-01 -9.28784609e-02 3.37014496e-01 -7.43306398e-01
1.62043548e+00 -2.58183384e+00 -5.59018850e-01 8.87589157e-03
-7.46364892e-03 -9.53707006e-03 -2.62972981e-01 4.40320939e-01
-4.72564340e-01 -4.00613219e-01 5.80819808e-02 -1.07369170e-01
1.06846295e-01 -7.78383315e-01 -7.08996952e-01 4.82441723e-01
5.00096940e-02 9.69464779e-02 -1.21046174e+00 1.29262740e-02
1.11068599e-01 6.53716922e-01 -5.35877585e-01 4.57248896e-01
4.16641861e-01 2.96816111e-01 -2.34307274e-02 2.58914322e-01
6.31620049e-01 1.81990966e-01 -3.52744102e-01 -3.83431435e-01
-2.51571685e-01 9.02971864e-01 -1.46863186e+00 1.55514133e+00
-7.86401987e-01 9.34701979e-01 1.99302971e-01 -5.43953300e-01
7.12055445e-01 7.97706366e-01 4.66135174e-01 -7.55639672e-01
-1.65438741e-01 5.24789214e-01 1.99251443e-01 -5.03754795e-01
2.09304482e-01 3.20607498e-02 5.09303100e-02 6.23611629e-01
4.40416373e-02 6.31087124e-02 1.44269094e-01 -9.31713730e-02
1.38140428e+00 -4.09276634e-01 1.84263185e-01 7.46801496e-02
4.97709781e-01 -7.55377591e-01 5.80062568e-01 9.21824753e-01
-1.69088781e-01 9.50646281e-01 1.41094983e-01 -1.01469323e-01
-1.35543615e-01 -1.53922713e+00 -2.28529721e-01 1.60572636e+00
-1.71651214e-01 -6.94969296e-01 -6.53445125e-01 -5.93056321e-01
-1.39912799e-01 9.12848949e-01 -2.97648519e-01 -2.27981344e-01
-6.68507099e-01 -6.74684644e-01 6.74743176e-01 6.34419918e-01
2.45585993e-01 -7.81021357e-01 -5.02910137e-01 5.76215565e-01
-5.22181392e-01 -1.09730220e+00 -9.39111769e-01 5.85460246e-01
-9.53747854e-02 -1.02414274e+00 -5.14396548e-01 -7.71821082e-01
9.42718163e-02 4.54166681e-01 8.43844354e-01 -5.12412429e-01
-2.47220829e-01 6.77509606e-01 -4.87516165e-01 -7.42720604e-01
6.54820055e-02 -4.27547246e-01 1.09332353e-01 4.30358946e-01
1.27496570e-01 -1.04005837e+00 -9.65840340e-01 3.49366933e-01
-6.04923368e-01 -3.99060935e-01 2.91854203e-01 6.10200524e-01
3.60671014e-01 2.64058352e-01 9.18531358e-01 3.11531983e-02
7.62053788e-01 -4.80334014e-01 -2.94761300e-01 -2.03968123e-01
7.42148329e-03 -5.51060200e-01 6.77008033e-01 -5.64552367e-01
-1.00580323e+00 -7.31699839e-02 -3.84187460e-01 -1.59959584e-01
-1.61769718e-01 4.03412074e-01 3.88614158e-03 3.70261580e-01
7.41746545e-01 3.18116724e-01 -9.09053802e-01 -7.72078276e-01
5.08105233e-02 9.66857135e-01 8.45867872e-01 -1.09139480e-01
5.04071772e-01 2.64205277e-01 -1.58892661e-01 -7.21506715e-01
-9.15527046e-01 -8.12066376e-01 -1.10590197e-01 1.32798944e-02
7.21071422e-01 -1.12861896e+00 -3.89975458e-01 7.44360745e-01
-1.34000421e+00 -9.49325040e-03 -4.03578579e-02 1.15903187e+00
-2.52407193e-01 -1.96633320e-02 -5.29107273e-01 -9.39446747e-01
-3.43919814e-01 -7.30157614e-01 1.17172396e+00 -3.72119434e-02
-2.15235367e-01 -5.26600659e-01 4.65918839e-01 -2.07962558e-01
5.72852612e-01 1.17254937e-02 7.78063655e-01 -7.15985417e-01
-1.19841322e-01 -5.75691164e-01 1.73999257e-02 4.07717735e-01
4.86630648e-01 -3.84722114e-01 -1.37257910e+00 -9.76165608e-02
3.38759392e-01 7.20684230e-02 7.29674697e-01 6.64177716e-01
1.16697967e+00 -1.19218901e-01 -2.46721104e-01 6.31388366e-01
8.52344811e-01 4.23383713e-01 9.09357592e-02 -7.21616000e-02
4.79820192e-01 1.39403805e-01 5.12036681e-01 6.43481910e-01
2.44715974e-01 9.17579174e-01 1.56386971e-01 8.90106633e-02
-4.33315068e-01 -3.08349878e-01 4.65678513e-01 1.15912580e+00
2.75858343e-01 -4.04332340e-01 -8.49476933e-01 7.26562381e-01
-1.38620710e+00 -9.69455421e-01 -2.97247857e-01 2.44832253e+00
7.41722584e-01 -5.07240221e-02 -3.24024111e-02 6.38637424e-01
7.10205615e-01 3.21261168e-01 -1.54117033e-01 -1.21758312e-01
6.54051900e-02 4.25578266e-01 8.47108364e-02 6.09343588e-01
-1.23809004e+00 1.43704176e-01 6.04211664e+00 9.46286201e-01
-1.42154801e+00 2.93187767e-01 1.10851400e-01 -3.52428049e-01
8.00753459e-02 -3.70531976e-01 -6.02308512e-01 6.00760996e-01
1.21603060e+00 2.25262251e-02 5.34419239e-01 4.21969473e-01
4.28552181e-01 -2.63258785e-01 -1.27541947e+00 1.31024873e+00
-1.15708813e-01 -8.64523113e-01 -5.81007779e-01 -3.16913009e-01
4.31930810e-01 2.30165944e-01 1.02165349e-01 2.81655014e-01
-3.90126407e-01 -7.30132818e-01 1.10564625e+00 4.03499544e-01
7.96608508e-01 -6.03796005e-01 3.01490039e-01 2.31447428e-01
-1.70091915e+00 -2.91038990e-01 -3.69012021e-02 -1.84311241e-01
2.29373202e-01 1.06176960e+00 -1.03476143e+00 1.99980602e-01
1.01945102e+00 4.39786971e-01 -3.20095032e-01 1.44834828e+00
-1.90446466e-01 1.17700684e+00 -6.19233727e-01 1.83392882e-01
-2.19328463e-01 4.09730017e-01 9.73855555e-01 1.56346154e+00
7.97157407e-01 -3.13028581e-02 -6.65965453e-02 5.30455232e-01
7.80025050e-02 -1.18352152e-01 -1.38309792e-01 1.34100452e-01
9.67635274e-01 1.19944239e+00 -5.64287364e-01 -3.63512849e-03
-5.05485237e-01 5.87989926e-01 -1.55023828e-01 7.09014654e-01
-9.79883850e-01 -1.02517128e+00 8.43837261e-01 -1.47500979e-02
5.70212483e-01 -1.57980114e-01 9.06310901e-02 -6.76501751e-01
1.36581093e-01 -6.65952861e-01 3.45626712e-01 -8.49252403e-01
-1.27856243e+00 7.43582129e-01 -2.32655242e-01 -1.44593275e+00
-5.60458422e-01 -4.83230859e-01 -9.46677327e-01 1.12049830e+00
-1.34066439e+00 -5.40405393e-01 -1.57455225e-02 7.16876626e-01
3.24504584e-01 7.04856440e-02 1.05038929e+00 6.07534170e-01
-4.31189805e-01 6.56979978e-01 2.00226113e-01 2.31042430e-01
8.75193179e-01 -1.35078204e+00 4.56970334e-01 8.94501746e-01
5.32025397e-01 3.89872670e-01 5.99981189e-01 -2.93048203e-01
-1.05946505e+00 -1.07281113e+00 1.12812579e+00 -3.01613599e-01
7.39269972e-01 -7.54169226e-01 -7.38832176e-01 2.87566215e-01
-7.50183463e-02 4.57241356e-01 7.63636708e-01 1.41274231e-02
-3.08583915e-01 -4.60759729e-01 -5.90782225e-01 1.47475421e-01
8.78233850e-01 -1.02451801e+00 -5.15423059e-01 3.74597847e-01
6.93535864e-01 -4.27985221e-01 -4.85514343e-01 2.78175801e-01
4.56651658e-01 -1.10877454e+00 1.12174487e+00 -1.13391094e-01
-1.26217067e-01 -5.45692146e-01 -4.44695145e-01 -1.49562001e+00
-5.40974736e-01 -8.00691903e-01 -2.68875539e-01 1.42561841e+00
4.18034852e-01 -6.96944535e-01 -6.13075979e-02 -1.47680476e-01
-3.11680168e-01 -4.77141500e-01 -1.14090252e+00 -7.67692864e-01
-4.06680971e-01 -1.19766605e+00 6.90159440e-01 5.27339160e-01
-5.81680015e-02 5.32698810e-01 -9.70964879e-02 7.76380777e-01
1.88757882e-01 2.62869984e-01 3.12707126e-01 -8.64626765e-01
-5.47102153e-01 -3.06281179e-01 -1.16215304e-01 -1.19099092e+00
-3.11269760e-01 -7.25792706e-01 6.74066722e-01 -1.27977467e+00
-4.48823512e-01 -8.74059871e-02 -9.44442987e-01 2.81650394e-01
-4.82409477e-01 1.69410646e-01 -8.21369216e-02 -3.35675299e-01
-4.52115595e-01 4.14478242e-01 6.47773802e-01 1.32433340e-01
-3.21935982e-01 4.07688618e-01 -3.25199276e-01 9.27386165e-01
4.04484421e-01 -8.56436610e-01 -3.58105928e-01 -5.34157813e-01
2.41934985e-01 3.62214357e-01 5.62236905e-01 -1.39004242e+00
4.03859943e-01 1.94797009e-01 4.17622119e-01 -6.22548699e-01
4.59236234e-01 -7.04884708e-01 -3.81439961e-02 6.31401390e-02
-3.22887391e-01 -1.21000580e-01 3.03585500e-01 8.40391457e-01
-4.28420246e-01 1.15494929e-01 4.50064301e-01 3.81986052e-01
-3.94515961e-01 -1.02438793e-01 -7.73693442e-01 2.13093728e-01
5.38753390e-01 2.40336850e-01 -1.57273278e-01 -5.61761022e-01
-4.90363270e-01 -3.76196474e-01 -5.56647778e-01 5.02714097e-01
5.40089905e-01 -1.48914647e+00 -8.76821756e-01 4.26279664e-01
-1.57031775e-01 -3.71105015e-01 4.64125395e-01 9.64447439e-01
-5.47768362e-02 5.22297740e-01 2.02657923e-01 -6.29948497e-01
-1.07800794e+00 1.16859779e-01 7.46620953e-01 5.25181973e-03
-2.46100143e-01 1.47278166e+00 5.10514021e-01 -1.91178992e-01
5.63517988e-01 -7.25015342e-01 -1.44411579e-01 8.91796425e-02
9.95391965e-01 7.69206345e-01 4.94530320e-01 -5.90201378e-01
-7.37802327e-01 5.52784920e-01 4.70617861e-01 -5.72918057e-01
1.10298407e+00 -4.85507697e-01 4.10358049e-02 6.80172026e-01
1.62088442e+00 6.24059081e-01 -1.11436737e+00 -2.64952183e-01
-1.55584782e-01 -6.05964363e-01 5.30683458e-01 -1.21872818e+00
-1.03239357e+00 1.07888305e+00 1.08894348e+00 6.13025188e-01
1.64310372e+00 -2.04586536e-01 9.34463322e-01 2.45334864e-01
2.10503116e-01 -8.68281543e-01 1.66790158e-01 6.86301172e-01
9.70519125e-01 -7.83676565e-01 -4.40657020e-01 -2.09856749e-01
-2.04756781e-02 1.06599474e+00 1.98766842e-01 4.86637838e-02
1.15579724e+00 4.63531584e-01 4.08639342e-01 3.48922312e-02
-5.58439553e-01 -2.64000148e-01 6.25827610e-01 6.19802833e-01
6.10238433e-01 3.88461794e-03 1.04750596e-01 1.04865277e+00
-5.44219613e-01 -6.17547750e-01 -3.64451148e-02 7.26021647e-01
-4.65946138e-01 -7.55033553e-01 -8.61144602e-01 3.96441787e-01
-7.85560369e-01 -3.33985537e-01 -1.07995115e-01 3.76713760e-02
3.22879016e-01 1.66983128e+00 3.76055628e-01 -5.17547727e-01
7.23216414e-01 1.77809089e-01 2.62059838e-01 -4.85026091e-01
-6.62025750e-01 5.87936342e-01 1.08642563e-01 -6.76756561e-01
1.03161834e-01 -6.75396204e-01 -1.28590715e+00 3.17941368e-01
-5.26165664e-01 2.55071580e-01 7.29057729e-01 7.50521481e-01
3.46403480e-01 9.40814972e-01 9.86233473e-01 -9.56192553e-01
-3.21402401e-01 -1.10382915e+00 -8.94191921e-01 2.67725796e-01
9.64679420e-01 -3.28153312e-01 -7.52202094e-01 8.88680071e-02] | [15.156414985656738, 5.27189826965332] |
7d5b1d9e-1ca2-42f0-84ce-54fb703dafda | resolving-language-and-vision-ambiguities | 1604.02125 | null | http://arxiv.org/abs/1604.02125v4 | http://arxiv.org/pdf/1604.02125v4.pdf | Resolving Language and Vision Ambiguities Together: Joint Segmentation & Prepositional Attachment Resolution in Captioned Scenes | We present an approach to simultaneously perform semantic segmentation and
prepositional phrase attachment resolution for captioned images. Some
ambiguities in language cannot be resolved without simultaneously reasoning
about an associated image. If we consider the sentence "I shot an elephant in
my pajamas", looking at language alone (and not using common sense), it is
unclear if it is the person or the elephant wearing the pajamas or both. Our
approach produces a diverse set of plausible hypotheses for both semantic
segmentation and prepositional phrase attachment resolution that are then
jointly reranked to select the most consistent pair. We show that our semantic
segmentation and prepositional phrase attachment resolution modules have
complementary strengths, and that joint reasoning produces more accurate
results than any module operating in isolation. Multiple hypotheses are also
shown to be crucial to improved multiple-module reasoning. Our vision and
language approach significantly outperforms the Stanford Parser (De Marneffe et
al., 2006) by 17.91% (28.69% relative) and 12.83% (25.28% relative) in two
different experiments. We also make small improvements over DeepLab-CRF (Chen
et al., 2015). | ['Kevin Kochersberger', 'Aishwarya Agrawal', 'Yash Goyal', 'Stanislaw Antol', 'Dhruv Batra', 'Ankit Laddha', 'Gordon Christie'] | 2016-04-07 | null | null | null | emnlp-2016-11 | ['prepositional-phrase-attachment'] | ['natural-language-processing'] | [ 3.45667809e-01 6.11220419e-01 -5.64938448e-02 -5.10778248e-01
-1.22617054e+00 -7.99292028e-01 6.08595967e-01 -3.39352409e-04
-4.95020896e-01 5.62604487e-01 2.87132025e-01 -2.68526882e-01
1.69709086e-01 -5.30302465e-01 -8.83426368e-01 -2.90904999e-01
4.39422727e-01 9.25611198e-01 4.61390853e-01 -2.20971286e-01
2.32002586e-01 2.17442825e-01 -1.34782100e+00 4.98543978e-01
7.13098705e-01 4.41725135e-01 4.77499455e-01 6.74634635e-01
-2.86197245e-01 5.79322994e-01 -5.86188674e-01 -9.50182676e-01
1.68084115e-01 -1.01396434e-01 -1.22409844e+00 2.39250973e-01
1.05912280e+00 -1.03771210e-01 3.60662848e-01 1.27212584e+00
4.34959214e-03 -1.30317628e-01 5.12190580e-01 -9.43362653e-01
-4.16023701e-01 8.85449350e-01 -1.07508111e+00 1.97571799e-01
3.49908173e-01 -4.67802882e-02 1.24981892e+00 -9.00301516e-01
8.90902281e-01 1.65526342e+00 6.09603882e-01 5.89206219e-01
-1.26552820e+00 -7.46964991e-01 3.47718388e-01 -4.45037708e-02
-1.22902882e+00 -4.20564145e-01 3.11086267e-01 -4.33097452e-01
1.02223408e+00 1.10555820e-01 3.38303268e-01 1.05817890e+00
1.43678904e-01 8.68943155e-01 1.26728046e+00 -4.67953026e-01
-1.54505625e-01 1.13865035e-02 3.57759178e-01 8.10605228e-01
3.79200876e-01 -4.53829587e-01 -4.12339717e-01 -4.74995449e-02
7.74660707e-01 -5.87910533e-01 2.92551011e-01 1.60153136e-01
-1.28730404e+00 8.16229045e-01 3.55491817e-01 1.14287160e-01
-2.69890308e-01 2.78982222e-01 3.48580480e-02 -3.00994486e-01
7.48298094e-02 4.32213396e-01 -4.80302423e-01 1.88862488e-01
-1.14318645e+00 4.66347039e-01 8.25829744e-01 9.26498592e-01
6.80406809e-01 -3.97509307e-01 1.00928351e-01 7.63643086e-01
5.97053349e-01 5.57746589e-01 -9.11476165e-02 -1.52977991e+00
5.93643963e-01 1.85057402e-01 1.87780991e-01 -9.89641964e-01
-4.10327822e-01 -2.49841332e-01 -3.05593461e-01 8.44180807e-02
7.06777215e-01 -2.40446866e-01 -1.16473055e+00 1.84169018e+00
1.54140159e-01 1.68857686e-02 2.40859225e-01 9.48390901e-01
7.81446457e-01 6.28323734e-01 7.59340823e-01 -4.56312001e-02
2.09688425e+00 -1.08987522e+00 -5.46432972e-01 -1.01168656e+00
-1.46267852e-02 -1.24134207e+00 8.92374694e-01 3.26015264e-01
-1.29734886e+00 -4.34661746e-01 -8.47071767e-01 -4.66729671e-01
6.51907995e-02 1.30744427e-01 6.73919320e-01 2.46412024e-01
-1.08700502e+00 1.06467128e-01 -8.54384363e-01 -6.56754911e-01
3.75131905e-01 2.84334332e-01 -2.28977591e-01 -8.56399629e-03
-9.98414993e-01 1.00654411e+00 3.41960192e-01 -1.52814627e-01
-3.64191532e-01 -2.61379451e-01 -8.52766275e-01 -2.42995936e-02
6.31637096e-01 -9.04159427e-01 1.53583980e+00 -1.11199570e+00
-9.22698140e-01 1.36145067e+00 -4.84628230e-01 -7.36928284e-01
3.73630881e-01 -4.35308218e-01 -8.22503567e-02 5.29849470e-01
7.89418399e-01 1.54805958e+00 7.46473491e-01 -1.42575979e+00
-8.86407971e-01 -4.30046767e-01 2.28286430e-01 4.49719042e-01
4.58879948e-01 5.72912633e-01 -7.10455894e-01 -5.53758442e-01
4.55193698e-01 -9.41340148e-01 -2.51013219e-01 -1.09739728e-01
-4.29938138e-01 -1.85045943e-01 4.07409012e-01 -1.09427011e+00
5.55956542e-01 -1.81363976e+00 4.18675542e-02 -2.27886721e-01
1.33773774e-01 1.67418495e-01 -1.55651465e-01 -6.54740110e-02
8.27321261e-02 4.50516105e-01 -4.83507276e-01 -4.80322063e-01
-1.29827499e-01 4.68016177e-01 -5.26553512e-01 -8.63486086e-04
5.30627251e-01 9.48313415e-01 -8.86145175e-01 -9.81491387e-01
-1.32153466e-01 3.96753699e-01 -3.24855328e-01 -3.50569487e-01
-5.76641321e-01 3.34876269e-01 -3.30841362e-01 7.87781000e-01
6.91393971e-01 -2.75415748e-01 4.43797350e-01 -2.81124204e-01
-8.14081654e-02 3.96787226e-01 -1.07335019e+00 1.55683911e+00
-2.05549479e-01 5.87073624e-01 3.97586405e-01 -4.97252882e-01
6.99483752e-01 1.84812874e-01 8.84816349e-02 -5.96529305e-01
-7.05952570e-02 1.30059104e-02 -1.26201153e-01 -4.84671980e-01
6.19885445e-01 -3.67560148e-01 -3.89229774e-01 1.87706113e-01
-4.30496894e-02 -2.61261165e-01 3.03741872e-01 5.03521442e-01
8.10956657e-01 4.88851160e-01 3.40718836e-01 -3.21366012e-01
1.99523300e-01 5.35885513e-01 9.47569132e-01 7.65111685e-01
-3.16231221e-01 7.85034478e-01 6.43830776e-01 -1.95793346e-01
-9.68347371e-01 -1.49991727e+00 -5.32719158e-02 1.07193732e+00
3.96268874e-01 -3.67553800e-01 -9.96387243e-01 -7.14404047e-01
-3.93566102e-01 9.19028282e-01 -2.04509482e-01 6.12125814e-01
-7.53842950e-01 -7.00280786e-01 6.42165482e-01 6.04910553e-01
6.22590184e-01 -1.09081173e+00 -6.89711392e-01 5.55128418e-02
-7.00624347e-01 -1.58070457e+00 -2.25357622e-01 -6.33786023e-02
-7.09504604e-01 -8.94722760e-01 -2.66285658e-01 -9.98124361e-01
5.80517530e-01 1.61138847e-01 1.33628464e+00 3.89894284e-02
-2.04448383e-02 4.42104816e-01 -1.44793168e-01 -2.82076746e-01
-4.82638657e-01 -1.41724765e-01 -2.25517973e-01 -4.01921540e-01
4.70209956e-01 -2.43877307e-01 -2.27402642e-01 1.07945107e-01
-5.77574134e-01 4.47273433e-01 7.07064390e-01 5.99773824e-01
8.60235274e-01 -1.36768430e-01 8.98019001e-02 -1.10538554e+00
2.20910132e-01 -3.35741013e-01 -4.98283565e-01 3.89864028e-01
-2.45303363e-01 8.14580843e-02 2.60202270e-02 -1.18197370e-02
-1.40256035e+00 3.28224063e-01 -2.59628564e-01 -6.46824837e-02
-6.58487618e-01 1.78272277e-01 -1.82922736e-01 3.98733437e-01
3.89464915e-01 -3.17277074e-01 -2.66064584e-01 -4.89533037e-01
4.64828044e-01 2.56169975e-01 9.58877563e-01 -8.11579168e-01
6.61166012e-01 5.42176187e-01 -1.91319391e-01 -7.87204564e-01
-1.36908710e+00 -4.36082572e-01 -8.06704879e-01 1.47948325e-01
1.51943266e+00 -1.19058490e+00 -3.67915332e-01 1.60818368e-01
-1.72988331e+00 -5.59039563e-02 8.26259926e-02 2.08443210e-01
-4.90382820e-01 4.46480244e-01 -7.45191336e-01 -6.24592364e-01
-1.87499538e-01 -1.07890630e+00 1.24828804e+00 5.54981649e-01
-5.97903907e-01 -6.01273417e-01 -3.14216971e-01 1.15392184e+00
-1.91502944e-01 1.58370152e-01 7.81206608e-01 -5.49492836e-01
-6.63275599e-01 2.66765296e-01 -5.93614638e-01 -7.29106069e-02
-1.93353951e-01 1.48497343e-01 -8.96796346e-01 1.67319268e-01
5.56911016e-03 -1.41249910e-01 1.04172313e+00 4.11592096e-01
4.90833431e-01 -3.34049433e-01 -4.02871817e-01 1.91731513e-01
1.34865093e+00 -5.97659424e-02 7.35006273e-01 3.81966621e-01
6.75894499e-01 8.65652919e-01 9.24816966e-01 -2.87305653e-01
7.78213263e-01 5.42067826e-01 2.44984806e-01 1.76699106e-02
-3.00438493e-01 -3.16729009e-01 4.29992616e-01 2.68732458e-01
9.36827138e-02 -2.81795025e-01 -1.22333622e+00 7.26169407e-01
-1.99415052e+00 -1.03721678e+00 -4.50191587e-01 1.61457098e+00
8.07075679e-01 3.70971352e-01 4.01368178e-02 -5.09230793e-01
1.09623408e+00 -5.07662781e-02 -2.25018099e-01 -6.08783841e-01
-2.83460885e-01 3.78853410e-01 5.01026094e-01 9.07999337e-01
-1.36263311e+00 1.59841740e+00 5.98106813e+00 6.36624336e-01
-6.15646601e-01 2.29611456e-01 6.67089045e-01 2.48066097e-01
-2.30530486e-01 4.32904899e-01 -1.12761998e+00 9.68162715e-02
7.18543589e-01 3.25328171e-01 1.23844936e-01 5.75889170e-01
4.54899482e-02 -6.20488167e-01 -7.31862366e-01 7.64609694e-01
2.58347422e-01 -1.23354483e+00 7.54482150e-02 -1.92107499e-01
5.38779795e-01 1.33813247e-01 -9.66174081e-02 -1.71764828e-02
6.60044968e-01 -1.04511225e+00 1.09343231e+00 4.85357881e-01
1.71674639e-01 -5.34605801e-01 5.42754710e-01 3.97549152e-01
-9.56918895e-01 2.39754960e-01 -1.95355251e-01 -1.63645074e-01
3.71134937e-01 2.93468535e-01 -9.36883748e-01 2.74650007e-01
9.01326060e-01 3.43691856e-01 -7.07115769e-01 5.60949266e-01
-9.41440046e-01 6.11512601e-01 -4.90933776e-01 3.22242320e-01
3.01456422e-01 -9.60266143e-02 8.39482725e-01 1.28356421e+00
-1.26461953e-01 3.47391456e-01 3.18153799e-01 1.07470036e+00
1.57091856e-01 -1.88519552e-01 -8.65279883e-02 -6.85446709e-02
4.13965821e-01 1.42603648e+00 -1.33665061e+00 -3.75668317e-01
-4.14272249e-01 1.02652776e+00 2.41151020e-01 2.75643706e-01
-8.21533442e-01 1.42101169e-01 5.78110754e-01 1.57239616e-01
5.87110996e-01 -4.34568495e-01 -4.68062580e-01 -1.05192983e+00
-2.93100812e-02 -6.60869658e-01 6.38701022e-01 -1.13567841e+00
-1.26330137e+00 4.88906682e-01 1.21647701e-01 -6.08820260e-01
-3.17651302e-01 -5.53063631e-01 -4.40010518e-01 8.06958139e-01
-1.37854517e+00 -1.63331020e+00 8.02759528e-02 5.52815758e-03
8.26492250e-01 4.22400892e-01 7.93071151e-01 -1.19646169e-01
-4.53961998e-01 2.35597372e-01 -6.67511344e-01 3.44287515e-01
6.92942441e-01 -1.24675405e+00 7.10233450e-01 1.26255560e+00
4.63480651e-01 7.79500902e-01 9.87158418e-01 -8.97026420e-01
-9.09391224e-01 -9.76691246e-01 1.15273106e+00 -8.31258595e-01
6.65860474e-01 -2.70818472e-01 -8.24595392e-01 9.59933937e-01
5.06628454e-01 -4.86908972e-01 4.31849718e-01 5.71895763e-02
-5.15110552e-01 3.76693189e-01 -1.28253806e+00 8.50215733e-01
8.70456636e-01 -2.44543597e-01 -1.17344308e+00 2.54670262e-01
8.54008973e-01 -4.64880019e-01 -5.24124622e-01 3.69755924e-01
3.60215664e-01 -8.87024999e-01 1.01479959e+00 -4.14577067e-01
6.32481337e-01 -4.98945147e-01 -2.91782290e-01 -5.75834394e-01
-1.68913886e-01 -3.28866452e-01 4.25340712e-01 1.46953464e+00
7.39157319e-01 -2.44698778e-01 6.50461733e-01 7.96377182e-01
-3.11121549e-02 -3.37318540e-01 -7.54345357e-01 -3.98094684e-01
2.13505626e-01 -5.35834551e-01 9.89538208e-02 8.06909859e-01
-2.97868460e-01 7.27931738e-01 -1.46239493e-02 6.49811208e-01
8.15908372e-01 3.41758907e-01 4.86868024e-01 -1.10127676e+00
-3.68323267e-01 -2.42077157e-01 -1.35332301e-01 -8.66170108e-01
3.71362418e-01 -8.54519546e-01 3.94157648e-01 -1.95381832e+00
3.52275252e-01 -2.14021444e-01 7.81749487e-02 7.71519840e-01
-1.52686596e-01 4.98179466e-01 4.65384930e-01 2.32726827e-01
-7.87472427e-01 -2.27119491e-01 1.02951801e+00 -1.32036835e-01
2.32654642e-02 -2.12236464e-01 -1.14801943e+00 1.34286928e+00
8.48823130e-01 -5.12238801e-01 -1.07324749e-01 -7.55499482e-01
3.39067936e-01 6.34604841e-02 7.80434847e-01 -7.56155372e-01
3.08341593e-01 -3.61311585e-01 4.02242839e-01 -5.37148237e-01
4.09286588e-01 -3.91395807e-01 -5.81694059e-02 2.93219000e-01
-1.89129174e-01 4.41597924e-02 5.00557423e-01 4.74713862e-01
-1.58286933e-03 -3.15718859e-01 7.61739433e-01 -4.65445161e-01
-1.17157471e+00 -2.63792008e-01 -4.20329511e-01 2.16264993e-01
7.71296680e-01 -1.59147590e-01 -3.86834443e-01 -9.20980722e-02
-1.03902948e+00 3.95926684e-01 5.73691726e-01 4.65327919e-01
4.94933635e-01 -7.91518927e-01 -7.29430020e-01 -2.68273652e-01
-2.50701845e-01 2.04949304e-01 3.82693339e-04 6.98626459e-01
-5.38255692e-01 1.75050244e-01 -1.87291011e-01 -5.93988538e-01
-1.41241288e+00 1.47900388e-01 9.75140780e-02 -7.90096223e-02
-5.34335911e-01 9.96268928e-01 1.10819861e-01 -2.19928652e-01
-1.07115664e-01 -2.51412302e-01 -2.12698892e-01 2.27488130e-01
2.68236130e-01 -6.12859018e-02 -2.96313018e-01 -1.00936127e+00
-6.85065627e-01 8.28320146e-01 -2.93451041e-01 -5.35893977e-01
1.18073010e+00 -4.71548408e-01 -4.05978113e-01 2.36374125e-01
5.13830066e-01 3.24034482e-01 -1.16427636e+00 -1.71834994e-02
1.52247116e-01 -1.97238982e-01 -2.17462644e-01 -1.17608082e+00
-6.93111479e-01 6.76365614e-01 1.16052203e-01 -4.75236028e-02
8.88706326e-01 6.35482371e-01 9.23431098e-01 3.24738503e-01
3.51063371e-01 -1.05135727e+00 -2.22248614e-01 5.47283471e-01
7.51167953e-01 -1.19970512e+00 6.92371232e-03 -9.28119242e-01
-9.64324474e-01 9.34580982e-01 7.90075660e-01 -1.71649769e-01
1.24707073e-01 5.50779343e-01 1.59483388e-01 -2.94688255e-01
-5.88587105e-01 -3.78223509e-01 2.73402452e-01 4.94552076e-01
4.06343758e-01 3.22699904e-01 -3.72458935e-01 8.34858358e-01
-4.29639697e-01 -4.59074080e-01 5.95079780e-01 7.77766526e-01
-7.38235831e-01 -1.04678488e+00 -6.98842466e-01 1.43255308e-01
-5.74078143e-01 -2.42977992e-01 -6.77539468e-01 8.75918806e-01
4.61134225e-01 1.12034976e+00 3.95153612e-01 2.40861569e-02
-2.50727311e-02 2.36760885e-01 6.36228561e-01 -9.13415551e-01
-3.29014391e-01 4.34359729e-01 3.58550131e-01 -3.74867976e-01
-7.90201664e-01 -1.01242042e+00 -1.90169489e+00 1.13721408e-01
-3.90868038e-02 -9.08454880e-02 5.61668038e-01 1.30063617e+00
1.92555577e-01 1.64815828e-01 -3.23809445e-01 -4.99723405e-01
-1.12925604e-01 -4.91126895e-01 2.45228633e-02 1.90389082e-01
-6.94937259e-03 -4.67856109e-01 -1.61780879e-01 5.94605982e-01] | [10.63011360168457, 1.518006682395935] |
fb60389d-faa8-46c6-a191-298ada5e1f54 | lightweight-uncertainty-aware-conformalized | 2303.02207 | null | https://arxiv.org/abs/2303.02207v1 | https://arxiv.org/pdf/2303.02207v1.pdf | Lightweight, Uncertainty-Aware Conformalized Visual Odometry | Data-driven visual odometry (VO) is a critical subroutine for autonomous edge robotics, and recent progress in the field has produced highly accurate point predictions in complex environments. However, emerging autonomous edge robotics devices like insect-scale drones and surgical robots lack a computationally efficient framework to estimate VO's predictive uncertainties. Meanwhile, as edge robotics continue to proliferate into mission-critical application spaces, awareness of model's the predictive uncertainties has become crucial for risk-aware decision-making. This paper addresses this challenge by presenting a novel, lightweight, and statistically robust framework that leverages conformal inference (CI) to extract VO's uncertainty bands. Our approach represents the uncertainties using flexible, adaptable, and adjustable prediction intervals that, on average, guarantee the inclusion of the ground truth across all degrees of freedom (DOF) of pose estimation. We discuss the architectures of generative deep neural networks for estimating multivariate uncertainty bands along with point (mean) prediction. We also present techniques to improve the uncertainty estimation accuracy, such as leveraging Monte Carlo dropout (MC-dropout) for data augmentation. Finally, we propose a novel training loss function that combines interval scoring and calibration loss with traditional training metrics--mean-squared error and KL-divergence--to improve uncertainty-aware learning. Our simulation results demonstrate that the presented framework consistently captures true uncertainty in pose estimations across different datasets, estimation models, and applied noise types, indicating its wide applicability. | ['Amit Ranjan Trivedi', 'Theja Tulabandhula', 'Danilo Erricolo', 'Alex C. Stutts'] | 2023-03-03 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [-1.35662019e-01 4.08080548e-01 -3.06749314e-01 -3.52041185e-01
-1.14347196e+00 -3.16362321e-01 3.50450575e-01 7.66492262e-02
-5.62768698e-01 1.09739661e+00 6.59498498e-02 -9.57732722e-02
-5.09350419e-01 -5.60927153e-01 -1.10846090e+00 -7.09860146e-01
-2.68130660e-01 7.17879653e-01 3.94201986e-02 1.68774307e-01
1.66491747e-01 4.95274633e-01 -1.34419298e+00 -4.80981916e-01
1.09813046e+00 1.43938613e+00 1.75346300e-01 4.78115261e-01
4.46174711e-01 3.22255045e-01 -3.60492080e-01 -3.65014464e-01
3.39980781e-01 1.48267637e-03 -5.94467595e-02 -3.52959365e-01
2.43558556e-01 -5.31424642e-01 -3.27353925e-01 9.02044594e-01
6.22921705e-01 -2.12758742e-02 9.30189550e-01 -1.41596377e+00
-3.96123379e-01 5.03134847e-01 -2.09464669e-01 -1.48359597e-01
-1.85825810e-01 2.71447748e-01 5.14197111e-01 -7.88063288e-01
7.66790807e-01 7.84205079e-01 1.09119380e+00 4.32910323e-01
-1.00975144e+00 -5.39049387e-01 6.48725173e-03 1.09265670e-02
-1.44901514e+00 -3.88495624e-02 6.37036264e-01 -5.48976779e-01
6.23569727e-01 -1.62959725e-01 8.38187337e-01 1.34090436e+00
1.18339789e+00 6.18321419e-01 6.83727145e-01 1.72563806e-01
6.62750423e-01 -6.74528182e-02 -4.76101249e-01 7.89483130e-01
6.83345020e-01 5.29786527e-01 -5.52651763e-01 -1.21340506e-01
1.01649117e+00 1.20723687e-01 -2.17293575e-01 -1.05930865e+00
-1.08737195e+00 7.48125076e-01 7.54704297e-01 -3.61604154e-01
-2.62265354e-01 7.61711299e-01 2.74787456e-01 -2.88037598e-01
3.69686186e-01 2.61858851e-01 -3.91486257e-01 -2.86067665e-01
-5.11980712e-01 2.23929346e-01 6.57092929e-01 1.39214563e+00
4.61481303e-01 1.49663821e-01 -2.48897612e-01 3.54765505e-01
5.59370816e-01 5.09955108e-01 1.45219713e-01 -9.78181064e-01
2.85792381e-01 4.11004066e-01 2.47594550e-01 -7.66980112e-01
-7.96417773e-01 -9.47183967e-01 -9.04351056e-01 6.56536758e-01
1.71180412e-01 -2.11560816e-01 -1.31024110e+00 1.88809216e+00
3.91165972e-01 1.38874009e-01 -1.18866146e-01 1.06172884e+00
4.87428635e-01 2.32134666e-02 7.42759407e-02 2.79873013e-02
1.11789060e+00 -3.84502590e-01 -5.44541478e-01 -3.80983084e-01
3.98014933e-01 -1.64906427e-01 6.76162064e-01 3.36147100e-01
-8.60971212e-01 -9.81748551e-02 -1.37861335e+00 -1.30864143e-01
-4.43344004e-02 9.69695896e-02 8.07685375e-01 5.28649807e-01
-7.19958305e-01 8.00008178e-01 -1.31119037e+00 1.05952486e-01
8.30720425e-01 5.51821530e-01 -3.22813481e-01 -3.30426581e-02
-1.03299522e+00 1.05200374e+00 3.60542417e-01 1.28655270e-01
-1.45016515e+00 -1.02609479e+00 -1.06911528e+00 -7.78945386e-02
3.69639099e-01 -1.03606975e+00 9.79498208e-01 -1.66874126e-01
-1.54830647e+00 2.68456221e-01 3.98496717e-01 -7.88728654e-01
8.62188995e-01 -4.99269336e-01 1.60986245e-01 -1.26879349e-01
-1.31770313e-01 1.00930774e+00 9.17805970e-01 -1.16387796e+00
-2.84011304e-01 -6.28812671e-01 -1.85059249e-01 3.42743635e-01
2.17934269e-02 -9.07509863e-01 -2.72946715e-01 -4.37879235e-01
5.18763363e-01 -1.11379123e+00 -5.30804813e-01 6.78472400e-01
-2.92178512e-01 2.55528927e-01 2.93198943e-01 -4.99767095e-01
7.80580163e-01 -2.01276088e+00 2.73366272e-01 1.72573656e-01
3.44251007e-01 -3.87573004e-01 4.68134552e-01 1.18034624e-01
5.48115313e-01 -2.16944396e-01 -6.41394913e-01 -6.43850565e-01
-5.00415526e-02 4.86835152e-01 -1.63439617e-01 7.44539738e-01
1.80077419e-01 8.73152971e-01 -9.89784718e-01 -4.84676749e-01
3.76862764e-01 7.24053562e-01 -7.13559926e-01 -1.03330657e-01
-3.70817363e-01 7.28015184e-01 -3.56743306e-01 7.08662927e-01
5.21552622e-01 -7.95443729e-02 -2.36864552e-01 -1.37739226e-01
1.38303339e-01 -4.46740575e-02 -1.12729585e+00 2.17346716e+00
-5.78764915e-01 5.23563802e-01 -1.78642854e-01 -3.53858143e-01
9.27852988e-01 1.30437668e-02 4.37918484e-01 -4.04091142e-02
4.30700004e-01 4.35744911e-01 -1.71475649e-01 -3.78449857e-02
4.62935001e-01 -1.16589956e-01 -3.01264495e-01 -2.84552544e-01
3.43540758e-01 -5.65482914e-01 -4.31170404e-01 -1.91464871e-02
9.16321933e-01 6.06191993e-01 2.75209874e-01 -1.52026936e-01
-8.84279460e-02 3.44603732e-02 9.03071165e-01 5.58499932e-01
-3.71263534e-01 6.76950693e-01 4.12320614e-01 -3.55240941e-01
-1.01633179e+00 -1.47673523e+00 -5.66252708e-01 2.07993209e-01
4.51464564e-01 5.89727722e-02 -5.10984600e-01 -5.88355184e-01
5.05887508e-01 7.74850070e-01 -6.66650951e-01 -6.48516893e-01
-1.18265472e-01 -7.29702175e-01 3.82468849e-01 9.26216424e-01
1.98001578e-01 -3.98532271e-01 -7.07114041e-01 2.51159430e-01
1.58778355e-01 -9.28508162e-01 -2.54363596e-01 5.81360936e-01
-1.05936205e+00 -8.71762931e-01 -5.32550275e-01 -3.15056860e-01
7.62707889e-01 -3.71957839e-01 7.57352650e-01 -6.42875910e-01
-5.10065198e-01 2.96494454e-01 -2.39396226e-02 -8.25274348e-01
-8.38189796e-02 -1.41044989e-01 3.99812669e-01 -4.18871731e-01
-1.22963786e-01 -7.15926528e-01 -8.72002304e-01 3.06678742e-01
-6.72308266e-01 -1.51027769e-01 9.44665492e-01 8.90905619e-01
1.04210639e+00 -4.59842622e-01 6.25882387e-01 -4.26247299e-01
3.75707060e-01 -6.09889030e-01 -9.13271546e-01 -4.14802581e-02
-8.10184121e-01 4.69930708e-01 3.58912498e-01 -3.97504985e-01
-8.13435733e-01 2.19925448e-01 1.84767377e-02 -8.50043833e-01
1.89863622e-01 4.53916788e-01 -4.67041768e-02 -4.40843016e-01
9.41457987e-01 -1.31442413e-01 1.78971529e-01 -4.27572764e-02
3.45590800e-01 4.03670251e-01 7.65948892e-01 -3.95183325e-01
5.44220924e-01 7.63722777e-01 5.03794611e-01 -4.03368592e-01
-6.65092051e-01 -1.64173275e-01 -1.74117029e-01 -3.50519955e-01
6.94315493e-01 -1.07302821e+00 -8.87427032e-01 2.31259063e-01
-8.94983709e-01 -1.22495331e-01 -3.68513465e-01 1.02362120e+00
-1.04287362e+00 1.45352855e-01 -3.18323284e-01 -9.68979776e-01
-3.95937622e-01 -1.41042459e+00 1.34989285e+00 2.74924904e-01
-3.87733094e-02 -7.76838779e-01 4.30111103e-02 -7.07101682e-03
3.51270080e-01 6.37651563e-01 5.21020889e-01 -3.60175759e-01
-9.65412438e-01 -4.47178125e-01 2.59301923e-02 1.66814759e-01
-2.13291183e-01 -2.98682451e-01 -8.53593826e-01 -2.23873392e-01
3.06642298e-02 -2.30911002e-01 7.72062242e-01 1.00571382e+00
1.12122166e+00 -1.01026008e-03 -6.46558225e-01 9.78941739e-01
1.39616668e+00 1.75061859e-02 5.65677881e-01 6.65956289e-02
4.53390241e-01 1.61733299e-01 7.05608010e-01 7.19786227e-01
5.12822568e-01 4.62783962e-01 1.13831294e+00 5.37320733e-01
1.06178060e-01 -4.69819248e-01 8.33339095e-02 3.07583123e-01
2.42946856e-02 -1.15919262e-01 -7.44503498e-01 4.91202503e-01
-1.86907709e+00 -4.40594196e-01 3.35444987e-01 2.49673295e+00
5.44744313e-01 4.60918814e-01 -3.29565346e-01 -2.83583552e-01
4.98505354e-01 -2.62240350e-01 -1.13751447e+00 -1.84971951e-02
3.00329626e-01 -3.22264761e-01 1.03817236e+00 4.14164931e-01
-1.06589627e+00 5.93403876e-01 5.48440599e+00 6.86818898e-01
-9.03697014e-01 -9.43984389e-02 6.47520959e-01 -1.47610396e-01
-3.96320164e-01 -9.75219831e-02 -1.02051461e+00 4.59935874e-01
7.78101325e-01 -4.20310088e-02 2.10024059e-01 1.38088560e+00
-1.38160914e-01 -3.62258196e-01 -1.21666729e+00 1.00052249e+00
6.58225566e-02 -1.40805423e+00 -3.51943314e-01 1.45825267e-01
7.62901306e-01 3.06236714e-01 3.68240088e-01 2.05209315e-01
4.17758077e-01 -8.69286537e-01 9.54719543e-01 8.64304781e-01
9.18019474e-01 -8.37777078e-01 8.99154842e-01 4.40413386e-01
-8.44001710e-01 -2.20060959e-01 -5.55769563e-01 3.07203054e-01
3.51712853e-01 8.78430665e-01 -1.15682757e+00 5.86425245e-01
6.89106345e-01 4.60532337e-01 -2.37212493e-03 1.41937947e+00
-4.08844501e-01 -4.78326716e-02 -7.56246865e-01 -1.84068963e-01
-7.44555146e-02 -8.09516013e-02 8.87390733e-01 4.05750871e-01
5.57618260e-01 -2.68059134e-01 -4.59136181e-02 1.08341527e+00
6.67625666e-03 -5.55613756e-01 -6.35590315e-01 1.71249673e-01
6.23616278e-01 1.05535686e+00 -5.62246561e-01 3.48725408e-01
-4.64441180e-02 7.55913258e-01 4.13021207e-01 2.57266662e-03
-9.86680806e-01 -3.35023314e-01 9.60074484e-01 -1.02207303e-01
1.12812079e-01 -5.81129491e-01 -7.00725853e-01 -8.00055385e-01
1.80100575e-01 -8.50084499e-02 2.84603953e-01 -7.55459130e-01
-1.13314903e+00 5.07553756e-01 2.20600531e-01 -1.55999720e+00
-5.40435493e-01 -8.36018264e-01 -2.09265471e-01 5.83234251e-01
-1.36237097e+00 -1.14051402e+00 -3.87989074e-01 -6.40366301e-02
3.44632298e-01 1.09166294e-01 5.37082613e-01 -7.74420798e-02
-1.47449642e-01 5.62760293e-01 2.08549708e-01 -2.36998156e-01
6.53678894e-01 -1.08290100e+00 1.63569301e-01 5.68900645e-01
-1.31283626e-01 4.41630661e-01 8.75600219e-01 -9.82447684e-01
-1.51318514e+00 -1.09292293e+00 8.81747380e-02 -6.63408875e-01
4.98096704e-01 -3.74325395e-01 -4.51524913e-01 7.72900760e-01
-6.38292670e-01 3.96704704e-01 3.43609452e-01 -4.38005812e-02
1.35494396e-01 -1.53477296e-01 -1.41919065e+00 6.74330115e-01
1.21179736e+00 -2.25163192e-01 -3.65360379e-01 1.34876624e-01
8.69277418e-01 -9.94930983e-01 -9.64223683e-01 1.03312564e+00
7.78902233e-01 -7.76733398e-01 9.30346787e-01 -3.18345636e-01
2.13510543e-01 -2.66675681e-01 -1.36203080e-01 -1.36838531e+00
1.77984815e-02 -5.59598982e-01 -4.45828736e-01 5.81588924e-01
3.00090909e-01 -6.03678942e-01 1.08513367e+00 7.92841256e-01
-8.26926947e-01 -1.17739010e+00 -1.49552524e+00 -9.87605274e-01
1.58073567e-02 -7.20430911e-01 2.81715482e-01 2.18065977e-01
-2.96506453e-02 -3.04899037e-01 -2.63220280e-01 5.60912788e-01
9.95680451e-01 -2.74956048e-01 4.95723516e-01 -1.31892562e+00
-2.04390556e-01 -2.19425917e-01 -1.06971323e+00 -7.76525378e-01
-6.77097067e-02 -6.98327363e-01 4.76725489e-01 -1.63579333e+00
-2.11649776e-01 -5.18430829e-01 -8.81991237e-02 1.18976973e-01
-5.96784316e-02 -4.94128168e-02 -2.53010005e-01 8.62317160e-03
-3.71345997e-01 9.64601576e-01 1.12649333e+00 7.58577958e-02
-2.17019185e-01 1.65030494e-01 -3.21124464e-01 7.76613295e-01
5.31322539e-01 -5.94734371e-01 -6.24974966e-01 -3.42630237e-01
4.41543579e-01 2.09545717e-01 5.62002540e-01 -1.47835994e+00
4.71402019e-01 6.19861297e-02 6.43383563e-01 -5.62866509e-01
5.23846328e-01 -1.02212477e+00 3.51202697e-01 6.32885933e-01
-8.84650573e-02 -3.88762206e-01 4.59546238e-01 1.17402816e+00
2.32703507e-01 -2.54961979e-02 7.44917274e-01 2.72265762e-01
-5.60201883e-01 6.95390522e-01 -7.18465596e-02 2.97744479e-02
1.25735176e+00 -1.78857833e-01 -2.25502148e-01 -3.55164289e-01
-7.16841638e-01 3.28667790e-01 5.86066127e-01 3.11278164e-01
8.76261532e-01 -1.23993361e+00 -5.72371900e-01 1.79530457e-01
3.66924912e-01 5.62406898e-01 4.41264689e-01 9.32420313e-01
-6.33303046e-01 2.14615583e-01 -1.70999855e-01 -9.73220170e-01
-3.80503088e-01 3.86530936e-01 4.22029972e-01 -3.93456407e-02
-5.69060028e-01 1.10356271e+00 1.68999031e-01 -5.11938751e-01
4.74641442e-01 -6.98333859e-01 2.85173059e-01 -4.19979215e-01
1.00803755e-01 3.83445412e-01 9.18453485e-02 -1.60305761e-02
-3.71675432e-01 3.32918048e-01 2.13032335e-01 -1.43513009e-01
1.26857781e+00 -9.57289264e-02 5.15746474e-01 5.58629870e-01
7.62286901e-01 -3.95182103e-01 -2.06397128e+00 1.63283020e-01
-2.11550072e-02 -2.27432922e-01 2.59786934e-01 -8.02279115e-01
-6.26869678e-01 7.11846530e-01 8.12134922e-01 -6.19336367e-01
6.06945753e-01 7.34185353e-02 3.92346382e-01 2.98428327e-01
1.01723289e+00 -1.02672553e+00 -2.09392771e-01 2.62968153e-01
9.68040705e-01 -1.39146924e+00 2.38952875e-01 -3.54451686e-01
-6.75757945e-01 7.94759095e-01 6.19062662e-01 -2.22712830e-01
7.36796021e-01 4.12833989e-01 -2.25055367e-01 6.80088252e-02
-5.87481856e-01 4.42405902e-02 4.60271776e-01 7.71453142e-01
-1.49412200e-01 -3.83171141e-02 -1.80478200e-01 8.07010055e-01
-6.46181405e-02 1.69022232e-01 2.93559581e-01 9.02786076e-01
-5.20181656e-01 -5.20715892e-01 -1.81027874e-01 6.48660421e-01
-1.12296484e-01 5.01958188e-03 2.15103522e-01 7.96634912e-01
-1.69442132e-01 2.59120166e-01 9.63690802e-02 -3.38767707e-01
9.53358486e-02 -4.45274264e-03 5.89713216e-01 -3.29388559e-01
3.97286080e-02 -2.41703913e-01 -1.28795728e-01 -7.65538573e-01
2.34304205e-01 -4.43743140e-01 -1.48014987e+00 8.87740925e-02
-4.58988011e-01 -1.08516343e-01 1.29512358e+00 8.33902180e-01
6.50840461e-01 6.08461142e-01 2.15529636e-01 -1.13029099e+00
-9.66998398e-01 -7.50556707e-01 -5.50117552e-01 -2.14678407e-01
3.06117415e-01 -1.32104850e+00 -5.15660465e-01 -4.93487269e-01] | [7.535497665405273, -1.4412871599197388] |
8a01a481-0472-45e1-8ee3-c361e8a6292d | imagined-speech-classification-using-eeg | null | null | https://www.researchgate.net/publication/309967859_Imagined_Speech_Classification_using_EEG | https://www.researchgate.net/publication/309967859_Imagined_Speech_Classification_using_EEG | Imagined speech classification using EEG | The objective of this work is to assess the possibility of using (Electroencephalogram) EEG for communication between different subjects. Here EEG signals are recorded from 13 subjects by inducing the subjects to imagine the English vowels ‘a’, ‘e’, ‘i’, ‘o’ and ‘u’ through visual stimulus. These recorded signals are then processed to remove artifacts and noise. Common features: Average power, Mean, Variance and Standard deviation are computed and classified using bipolar neural network. This method yields maximum classification accuracy of 44%. The result shows that EEG has some distinctive information for across subject classification. | ['Shenbaga Devi. S.', 'Madan Raj. M.', 'Rajkumar R.', 'Kamalakkannan Ravi'] | 2014-12-01 | null | null | null | advances-in-biomedical-science-and | ['eeg', 'classification-1', 'eeg'] | ['methodology', 'methodology', 'time-series'] | [ 3.99705589e-01 -2.70612448e-01 5.74952960e-01 -4.12674725e-01
-2.90047657e-02 -6.44356370e-01 3.97169113e-01 6.40848577e-02
-4.51525360e-01 1.43385684e+00 -3.13889161e-02 -7.36224204e-02
-3.35480869e-01 -1.78593531e-01 -2.32211724e-01 -5.99088550e-01
-4.68393713e-01 -1.89502105e-01 -2.77314931e-01 5.59528396e-02
6.58610106e-01 4.73858178e-01 -1.37446809e+00 2.55893260e-01
7.19703257e-01 1.07903969e+00 2.14009166e-01 6.24731183e-01
4.28943515e-01 9.69338790e-02 -1.30694056e+00 1.90125987e-01
7.77401850e-02 -6.70953572e-01 -7.70928562e-01 -1.82990953e-02
-1.63912088e-01 -9.37616639e-03 5.86144030e-02 1.27511251e+00
6.83171213e-01 8.61721635e-02 1.01495266e+00 -1.33786929e+00
-8.18394065e-01 4.37150747e-01 -6.31302059e-01 7.50275195e-01
9.21966791e-01 -1.04743667e-01 2.71166623e-01 -6.27256155e-01
1.58505678e-01 5.92426002e-01 5.30439138e-01 2.34707892e-01
-1.32420802e+00 -1.07756913e+00 -5.52018940e-01 4.01981860e-01
-1.71323311e+00 -3.08372885e-01 7.63153672e-01 -4.65157419e-01
1.33914554e+00 6.66757286e-01 9.12392914e-01 1.12838447e+00
9.30046141e-01 -1.56671684e-02 1.85042715e+00 -4.16098982e-01
1.72990173e-01 6.22061193e-01 6.32505119e-01 -7.82257318e-02
1.99790925e-01 -6.36388659e-02 -4.87191439e-01 -1.92723006e-01
2.70238608e-01 -3.61486912e-01 -8.02883208e-01 6.99813664e-01
-1.17880428e+00 2.88492680e-01 -4.59673675e-03 9.69940066e-01
-8.27448010e-01 -2.86799341e-01 3.63963693e-01 5.57014227e-01
-1.97397858e-01 7.41678476e-01 -3.52146268e-01 -4.03416604e-01
-7.86868751e-01 -1.23853378e-01 7.13978171e-01 8.34407568e-01
1.60601661e-01 9.18696374e-02 -5.09950519e-02 6.35711074e-01
1.16590902e-01 4.62940156e-01 9.25974369e-01 -4.61773574e-01
1.21353582e-01 2.13667795e-01 2.14661777e-01 -1.15014994e+00
-8.43066335e-01 -2.15041608e-01 -7.58927763e-01 2.70602614e-01
-1.12674966e-01 -5.30121565e-01 -7.47390151e-01 1.21947312e+00
-4.73538309e-01 1.08472236e-01 1.84348106e-01 8.28275740e-01
7.57835388e-01 7.28650808e-01 1.48342118e-01 -7.07657993e-01
1.62162995e+00 -2.69234955e-01 -1.13302922e+00 -1.13718235e-03
-2.55755484e-01 -9.22929525e-01 7.93872535e-01 7.97389209e-01
-1.01268089e+00 -6.29695058e-01 -1.28132463e+00 5.81735432e-01
-5.38483679e-01 -3.28330323e-03 5.29592812e-01 9.33357716e-01
-1.04606938e+00 6.30451739e-01 -4.31542754e-01 -3.42733890e-01
1.77764520e-01 7.93238163e-01 -6.93696499e-01 8.89894187e-01
-1.27142680e+00 1.07067847e+00 5.37653387e-01 2.21663833e-01
-3.64988595e-01 -3.42410028e-01 -3.59675050e-01 -9.08574928e-03
-6.68969572e-01 -2.42527604e-01 5.90713501e-01 -1.17234075e+00
-1.57895517e+00 6.83033764e-01 -1.17117651e-01 -1.98330700e-01
-1.14468083e-01 1.43127739e-01 -8.65517259e-01 3.07079226e-01
-2.27413550e-01 3.60128552e-01 7.30017245e-01 -9.33513999e-01
-3.56415808e-01 -7.83982515e-01 -5.60069323e-01 1.61252599e-02
1.96008608e-01 4.40407783e-01 6.96641624e-01 -5.08051336e-01
2.76378274e-01 -7.06110656e-01 5.66079557e-01 -9.50121045e-01
-5.06173909e-01 -3.05279672e-01 4.86275643e-01 -8.87030065e-01
1.12080669e+00 -2.32080054e+00 -1.30451545e-01 7.84747660e-01
-1.41286440e-02 -4.32423539e-02 4.09524351e-01 4.90948975e-01
-5.44195056e-01 -6.02322891e-02 -3.33374947e-01 4.54300582e-01
-1.52429178e-01 -2.19648048e-01 9.46847498e-02 5.95636249e-01
-6.28680214e-02 4.37086195e-01 -3.90157163e-01 -1.05942443e-01
2.52079308e-01 7.25941777e-01 1.25135973e-01 1.88733205e-01
9.62588668e-01 7.70177364e-01 -1.61336679e-02 4.32028651e-01
6.61838949e-01 3.77462596e-01 -1.33613899e-01 -4.07926381e-01
-3.18496644e-01 3.86942327e-01 -1.15128076e+00 1.13287330e+00
-5.08812629e-02 1.23410213e+00 -9.19878334e-02 -1.13811851e+00
1.09335566e+00 9.00569499e-01 1.10876471e-01 -8.08888853e-01
7.43291140e-01 1.10882580e-01 6.92693591e-01 -8.91265512e-01
6.88516200e-02 -2.02326283e-01 1.84620336e-01 4.85388726e-01
9.64017585e-02 -1.30377665e-01 -5.86768985e-02 -2.14580283e-01
8.20178449e-01 -4.63957429e-01 5.75574756e-01 -6.36199057e-01
6.92493021e-01 -6.67917371e-01 -2.32698880e-02 4.11034822e-01
-5.74753642e-01 1.21326104e-01 3.74595135e-01 2.66467966e-02
-5.85439861e-01 -1.02523673e+00 -7.54770100e-01 4.35475588e-01
-1.57252662e-02 1.31507218e-01 -1.06814384e+00 8.26951861e-02
-2.58032888e-01 1.00620735e+00 -6.84590936e-01 -2.31496483e-01
-1.83950290e-02 -9.06660318e-01 4.63097930e-01 3.90432566e-01
5.81230700e-01 -1.35544074e+00 -1.21793783e+00 3.29974115e-01
-8.54173210e-03 -7.56464362e-01 -6.56624138e-02 6.22340083e-01
-7.07676470e-01 -6.79270029e-01 -5.34352064e-01 -7.23451316e-01
6.01145327e-01 -4.38851655e-01 6.85411274e-01 -1.85886443e-01
-5.28659105e-01 3.30248296e-01 -2.43184716e-01 -7.28550076e-01
9.07333121e-02 -5.23503959e-01 3.91684413e-01 -1.05787499e-03
8.56094837e-01 -1.24666715e+00 -6.48879051e-01 1.46629699e-02
-4.58785892e-01 -5.16970575e-01 5.42116106e-01 2.50028521e-01
6.43791109e-02 1.94860220e-01 7.59217441e-01 -2.92301387e-01
1.40490377e+00 -2.31285527e-01 -4.97137755e-02 -8.33330005e-02
-2.64809728e-01 -3.93809468e-01 6.23053432e-01 -4.04092699e-01
-6.97166383e-01 -5.57386041e-01 5.52721024e-02 3.47908914e-01
-7.83796906e-01 1.40244663e-01 -8.97851959e-02 -2.20964521e-01
7.13028729e-01 5.17702878e-01 -6.06895089e-01 -9.96306017e-02
-5.06510317e-01 1.32885742e+00 7.93764591e-01 5.53446673e-02
2.07402468e-01 -2.08851434e-02 -1.83763161e-01 -1.03364253e+00
8.28473121e-02 -9.42015797e-02 -7.50598729e-01 -3.19466025e-01
8.82873774e-01 -5.60197711e-01 -9.47475493e-01 5.57904720e-01
-1.25404453e+00 2.41638124e-01 4.88137096e-01 1.22361326e+00
-1.92940101e-01 -3.27680856e-02 -5.01443624e-01 -9.74113703e-01
-8.38869214e-01 -1.10750246e+00 3.18431914e-01 3.92848760e-01
-9.15040910e-01 -6.01405561e-01 -3.44149441e-01 -1.28733382e-01
3.66697669e-01 9.10445973e-02 7.89096653e-01 -7.48237491e-01
3.78686368e-01 -3.43345851e-01 -1.05046414e-01 4.25924659e-01
4.98642206e-01 -1.07520089e-01 -1.15883827e+00 -4.21229899e-02
6.26548350e-01 -2.15607043e-02 8.85338895e-03 5.19359827e-01
1.08931124e+00 -1.80526823e-01 -2.68265635e-01 4.51935172e-01
1.43559003e+00 1.36014664e+00 9.16292667e-01 1.49580454e-02
-5.88360690e-02 3.24445188e-01 -3.23947608e-01 2.07937226e-01
-1.89244032e-01 1.60306647e-01 -2.10740611e-01 5.75399756e-01
4.26850230e-01 5.20870149e-01 2.28072144e-02 1.09618795e+00
-4.24697846e-01 -3.95054258e-02 -7.63127387e-01 2.87024468e-01
-7.86419630e-01 -8.66505623e-01 -2.22102046e-01 2.02650166e+00
6.77427709e-01 3.22192132e-01 -1.53341135e-02 9.00456786e-01
7.17581511e-01 -5.07550001e-01 -6.74909577e-02 -9.37962174e-01
1.05692230e-01 7.95764387e-01 4.06470180e-01 3.12778890e-01
-5.75272858e-01 6.80300891e-02 6.88664865e+00 2.61028975e-01
-1.44084787e+00 8.27663485e-03 2.96554983e-01 1.36410221e-01
3.52746665e-01 -4.95239556e-01 -2.94435561e-01 1.10052848e+00
1.32371473e+00 -3.90389174e-01 6.44251883e-01 2.41251186e-01
3.15152138e-01 -8.11654747e-01 -1.16646135e+00 1.48955929e+00
3.28330934e-01 -5.52004814e-01 -6.06782697e-02 -1.53412610e-01
3.92393023e-01 -3.76858383e-01 -3.44076380e-02 2.42469143e-02
-5.21006703e-01 -1.30867183e+00 4.24290091e-01 6.97490394e-01
6.95414186e-01 -8.84532928e-01 6.59587502e-01 1.31556407e-01
-8.98718774e-01 -1.46130538e-02 -4.29477513e-01 -5.53416193e-01
-6.36545718e-02 1.96225703e-01 -5.01743138e-01 3.49966198e-01
6.49907112e-01 2.58954823e-01 -4.03109878e-01 1.12301981e+00
-1.09008521e-01 6.98527396e-01 -3.11955541e-01 -4.29913938e-01
2.37593696e-01 -4.99489456e-01 5.35508573e-01 1.22588968e+00
4.56627637e-01 6.91126943e-01 -5.60475707e-01 7.43104279e-01
1.91203088e-01 1.73180476e-01 -8.34317207e-01 7.43948966e-02
7.27578461e-01 9.45715487e-01 -1.16690016e+00 -1.80719912e-01
-1.04467601e-01 1.19497454e+00 -2.45820686e-01 4.56355035e-01
-6.55874133e-01 -1.24460864e+00 9.35615152e-02 -3.27694386e-01
-5.02848744e-01 4.48320732e-02 -6.31201684e-01 -4.54614341e-01
5.61932027e-02 -7.06917107e-01 -3.21227163e-02 -1.20928884e+00
-8.82090449e-01 9.60231841e-01 2.16057509e-01 -7.86685646e-01
1.77849591e-01 -6.02086782e-01 -7.76562989e-01 1.27834821e+00
-4.72607106e-01 -6.26042858e-02 -1.90990135e-01 6.35837853e-01
2.37190932e-01 -3.92507464e-01 9.22465146e-01 2.83523828e-01
-3.41864347e-01 3.58324140e-01 -1.39018491e-01 1.27462208e-01
3.84678036e-01 -1.30160296e+00 -2.34786212e-01 3.73692989e-01
-1.39140068e-02 9.30651188e-01 1.03405178e+00 -2.85020411e-01
-1.03682172e+00 -4.39866364e-01 1.01061642e+00 -9.88595039e-02
3.45684916e-01 -2.13683203e-01 -7.93553352e-01 5.78710735e-01
9.94811654e-01 -5.43681562e-01 1.14569116e+00 -3.45516324e-01
3.24377447e-01 -1.03955097e-01 -1.32163584e+00 4.97411519e-01
4.16876733e-01 -5.55220604e-01 -1.30519676e+00 1.32149950e-01
-1.59272432e-01 -4.03576382e-02 -9.64617968e-01 3.18916701e-03
8.80777657e-01 -1.05392051e+00 4.93800461e-01 -3.07270139e-01
-2.31008045e-02 2.72473078e-02 -9.50457994e-03 -1.86150730e+00
-2.99709886e-01 -7.17607439e-01 7.60207653e-01 9.51503813e-01
5.48552573e-01 -1.19308043e+00 -9.81110185e-02 5.56375861e-01
-1.06566940e-02 -2.97569543e-01 -9.01554644e-01 -5.68378747e-01
-1.62838042e-01 -4.74779665e-01 4.24234807e-01 1.01050854e+00
9.00422633e-01 7.19802260e-01 -1.79994568e-01 2.59635337e-02
4.67798322e-01 -3.75345588e-01 -6.17999658e-02 -1.33269131e+00
2.04464495e-01 -3.77639711e-01 -1.01624703e+00 -1.88433871e-01
1.01083882e-01 -7.82012165e-01 -1.74291819e-01 -1.47940469e+00
1.29162699e-01 2.22816616e-01 -5.65196455e-01 1.26325220e-01
-3.54619026e-02 6.00768328e-01 -1.26614749e-01 -2.71428853e-01
2.92674452e-01 -3.49595211e-02 8.34059894e-01 1.62294835e-01
-3.42681289e-01 -1.87447354e-01 -3.83429050e-01 6.72545373e-01
1.07572627e+00 -4.37240422e-01 -5.23524225e-01 -1.23660542e-01
1.24147208e-02 1.49493203e-01 2.31466785e-01 -1.45727003e+00
1.98685035e-01 3.34531933e-01 1.04218602e+00 -3.74744058e-01
2.83462554e-01 -1.07547069e+00 6.11028194e-01 4.75739300e-01
-2.45760351e-01 6.35014176e-01 3.50566357e-01 1.34990290e-01
-1.96013719e-01 -5.21392226e-01 7.90197790e-01 -2.04228401e-01
-3.35633188e-01 -3.84519190e-01 -8.48978281e-01 -3.43347639e-01
1.38255298e+00 -6.17663383e-01 -6.13859221e-02 -2.77509898e-01
-1.02899826e+00 -2.14314878e-01 -1.04505651e-01 -1.38269231e-01
6.39186859e-01 -1.19580448e+00 -5.19116461e-01 6.91079080e-01
-1.27565637e-01 -9.28272069e-01 9.68000591e-02 1.32530725e+00
-4.92484868e-01 6.62120223e-01 -1.12303460e+00 -3.71841818e-01
-1.47096717e+00 1.92358911e-01 5.22427917e-01 6.24316752e-01
-7.18163490e-01 9.81260061e-01 -3.21821868e-01 5.32053709e-01
2.39310771e-01 -2.54626304e-01 -7.25257993e-01 2.92252153e-01
7.74662077e-01 4.26569879e-01 2.42573813e-01 -7.37626016e-01
-6.05937183e-01 4.68658626e-01 3.14865381e-01 -4.36228722e-01
1.08287716e+00 -1.10117771e-01 -2.58155584e-01 1.01620853e+00
1.29835403e+00 -2.12997962e-02 -2.62209207e-01 7.27811694e-01
-1.09918406e-02 -5.43907285e-01 -1.09525301e-01 -1.25420141e+00
-6.71504736e-01 9.00366008e-01 9.58396375e-01 5.69698632e-01
1.25115192e+00 -4.59564626e-01 3.36682558e-01 3.35867703e-01
5.64708173e-01 -1.15863812e+00 -5.61270416e-01 5.68046607e-02
1.05614328e+00 -6.55269921e-01 -7.20371827e-02 8.52229819e-02
-6.82323635e-01 1.14835763e+00 1.61760271e-01 -4.16680008e-01
8.05748880e-01 5.51831782e-01 6.19703904e-02 -3.14549834e-01
-5.97974300e-01 2.32332647e-01 3.68126482e-01 9.49171364e-01
8.77135396e-01 3.28632653e-01 -1.01348507e+00 5.96880317e-01
-7.90615201e-01 1.70023963e-01 7.92726099e-01 7.57132590e-01
-4.39374417e-01 -3.35010111e-01 -6.95177615e-01 9.49540079e-01
-9.70241189e-01 -5.51753417e-02 -5.11665881e-01 9.21539783e-01
9.36734006e-02 1.34799206e+00 1.99579149e-01 -4.13173884e-01
5.57898700e-01 5.38367689e-01 8.07653189e-01 -1.99221820e-01
-8.34857047e-01 7.64750550e-03 -1.93684235e-01 -1.21198587e-01
-5.30307531e-01 -4.11827594e-01 -1.21426594e+00 1.06973916e-01
-1.50978282e-01 5.60986459e-01 9.10191894e-01 1.01564443e+00
6.63042860e-03 7.86208391e-01 4.69870299e-01 -7.64380097e-01
-1.89665809e-01 -1.73402381e+00 -1.16746843e+00 5.27564108e-01
1.91544339e-01 -7.69200921e-01 -6.94172144e-01 2.17160136e-01] | [13.27430248260498, 3.3155486583709717] |
b33186ab-25c1-4349-9fbb-589460eb81df | samplernn-an-unconditional-end-to-end-neural | 1612.07837 | null | http://arxiv.org/abs/1612.07837v2 | http://arxiv.org/pdf/1612.07837v2.pdf | SampleRNN: An Unconditional End-to-End Neural Audio Generation Model | In this paper we propose a novel model for unconditional audio generation
based on generating one audio sample at a time. We show that our model, which
profits from combining memory-less modules, namely autoregressive multilayer
perceptrons, and stateful recurrent neural networks in a hierarchical structure
is able to capture underlying sources of variations in the temporal sequences
over very long time spans, on three datasets of different nature. Human
evaluation on the generated samples indicate that our model is preferred over
competing models. We also show how each component of the model contributes to
the exhibited performance. | ['Aaron Courville', 'Rithesh Kumar', 'Jose Sotelo', 'Ishaan Gulrajani', 'Soroush Mehri', 'Shubham Jain', 'Yoshua Bengio', 'Kundan Kumar'] | 2016-12-22 | null | null | null | null | ['audio-generation'] | ['audio'] | [ 2.55818784e-01 1.87561139e-01 1.78711027e-01 -3.59546125e-01
-8.27646375e-01 -2.59084225e-01 7.01324403e-01 -2.88867146e-01
4.43206728e-02 1.04853725e+00 7.44499147e-01 -7.37694651e-02
-2.32389793e-01 -7.10101902e-01 -7.47523844e-01 -5.33743024e-01
-4.07110006e-01 1.56445783e-02 3.36149424e-01 -3.63737315e-01
6.45003701e-03 4.24592555e-01 -1.87914109e+00 5.69573224e-01
3.40974748e-01 9.97853279e-01 9.29079428e-02 1.07270122e+00
1.71433687e-01 1.27744746e+00 -7.85045087e-01 -1.39004201e-01
-4.81130742e-02 -5.72136819e-01 -7.58509159e-01 -5.09171933e-03
-3.23512629e-02 -1.88127786e-01 -3.45478475e-01 3.76498252e-01
5.66297591e-01 1.20769799e-01 7.17239380e-01 -9.89476681e-01
-3.96028370e-01 1.39001739e+00 1.41890392e-01 8.85166973e-02
4.14316624e-01 6.97475001e-02 1.03312898e+00 -8.19625199e-01
4.26574022e-01 1.18539071e+00 9.50714946e-01 6.40734553e-01
-1.55090117e+00 -4.91082728e-01 5.56364171e-02 7.60407969e-02
-1.28759134e+00 -7.63300359e-01 1.15318406e+00 -5.02029836e-01
1.15508687e+00 2.72987783e-01 5.14619529e-01 1.63931954e+00
3.98396820e-01 6.93195879e-01 7.68757761e-01 -4.96227324e-01
2.47951433e-01 -1.29110888e-02 1.46079868e-01 1.04218259e-01
-2.75884569e-01 3.56795043e-01 -9.07198012e-01 -3.66832316e-01
7.20843732e-01 -2.48431280e-01 -1.38431415e-01 2.95006018e-02
-9.49021637e-01 6.15422845e-01 1.30061239e-01 5.34016013e-01
-8.61764848e-01 4.65346992e-01 3.37410033e-01 3.71314019e-01
4.43383992e-01 3.35533470e-01 -4.49750125e-01 -3.23257625e-01
-1.08538890e+00 2.79391855e-01 7.56907165e-01 7.26971745e-01
3.47931653e-01 9.22237694e-01 -3.39263380e-01 8.44536483e-01
1.36906698e-01 2.52127737e-01 9.56062615e-01 -8.79886508e-01
1.86182976e-01 1.43774375e-01 1.67923525e-01 -6.90711677e-01
-4.05782133e-01 -6.64004147e-01 -1.09567630e+00 -2.24908944e-02
1.03098797e-02 -5.09568155e-01 -8.53026211e-01 2.00065923e+00
-2.83413231e-01 4.03716832e-01 2.94393599e-01 4.28566247e-01
5.65281928e-01 8.85108411e-01 7.28568584e-02 -3.65924835e-01
8.02732587e-01 -6.80668771e-01 -6.93879008e-01 7.41125196e-02
-1.56963449e-02 -6.00897908e-01 5.00777304e-01 6.32126153e-01
-1.55176139e+00 -9.14888084e-01 -1.06542265e+00 3.99954379e-01
-2.65346408e-01 1.30500391e-01 4.04469311e-01 6.33305252e-01
-1.46373236e+00 1.00474441e+00 -5.68189383e-01 7.33629316e-02
-3.73642236e-01 9.94680524e-02 -1.92899238e-02 9.08119142e-01
-1.74333513e+00 4.74437892e-01 5.76897323e-01 2.93243408e-01
-1.13684535e+00 -5.44217229e-01 -5.64448893e-01 1.63541347e-01
-1.79716483e-01 -6.80486202e-01 1.63507402e+00 -8.09189677e-01
-1.91807008e+00 1.61861077e-01 -3.33293349e-01 -1.22378242e+00
3.28217834e-01 -3.43960792e-01 -6.14077866e-01 1.20234832e-01
-3.55154723e-01 5.40338635e-01 1.27758241e+00 -1.13317430e+00
-5.54219604e-01 1.30984068e-01 -3.27710062e-01 -2.78376192e-01
-2.70756215e-01 4.41053929e-03 2.32878298e-01 -8.26201618e-01
-7.29062930e-02 -9.64301169e-01 -5.43650746e-01 -9.43328738e-01
-4.73317206e-01 -2.61411369e-01 4.76802021e-01 -7.64224410e-01
1.52980614e+00 -2.12004638e+00 2.58678973e-01 2.05582812e-01
-2.38398209e-01 -1.10078469e-01 -4.84495163e-02 7.95270979e-01
-3.51314455e-01 1.23069532e-01 -5.38601466e-02 -4.99442339e-01
4.89486530e-02 1.14857011e-01 -1.15485501e+00 -1.23896018e-01
3.33104938e-01 7.10410535e-01 -6.47096634e-01 -1.87852398e-01
7.78613985e-03 7.09977925e-01 -2.93449849e-01 3.29085618e-01
-3.11034411e-01 3.19332808e-01 -5.95831238e-02 2.63811022e-01
1.20543219e-01 2.64502037e-02 6.03972375e-02 1.91173583e-01
-1.64242297e-01 5.25241494e-01 -1.10873294e+00 1.46945083e+00
-4.15645927e-01 6.13367856e-01 -2.52002954e-01 -6.99093461e-01
1.15853250e+00 1.11941326e+00 2.46308059e-01 -3.82028371e-01
-8.09892789e-02 1.29398584e-01 -1.35560438e-01 -1.38901964e-01
7.47073770e-01 -3.73664796e-01 -1.32807672e-01 4.15391982e-01
4.14559305e-01 -3.27951129e-04 2.17702076e-01 8.34279656e-02
1.06539285e+00 6.81871325e-02 1.07451908e-01 8.85776058e-02
2.96079367e-01 -6.32531047e-01 5.34996688e-01 9.93115425e-01
1.66864753e-01 7.42484391e-01 4.04840618e-01 -3.21683228e-01
-1.19701362e+00 -1.31607068e+00 -3.82012837e-02 1.09721339e+00
-7.19351649e-01 -3.96960199e-01 -3.90056670e-01 9.18323174e-02
-4.60096687e-01 9.07061458e-01 -5.17149210e-01 -2.54447222e-01
-5.59063971e-01 -5.00585854e-01 9.41285491e-01 7.07575500e-01
-5.60904928e-02 -1.49632239e+00 -8.39556336e-01 7.41810143e-01
-2.90218145e-01 -7.92068779e-01 8.84942636e-02 4.79677349e-01
-1.07156301e+00 -2.01359138e-01 -9.05198693e-01 -3.65725905e-01
-1.84608370e-01 -1.25374705e-01 1.35921836e+00 -5.03785014e-01
4.49986644e-02 3.38196337e-01 -4.81074482e-01 -6.14884317e-01
-7.61673629e-01 1.90020412e-01 2.92972356e-01 2.41299734e-01
-2.24624023e-01 -1.31753707e+00 -3.95233363e-01 -1.79668292e-01
-1.04909348e+00 -1.11476565e-02 5.06298244e-01 6.94743097e-01
1.46307603e-01 1.69510439e-01 1.04503071e+00 -5.63381016e-01
1.07425416e+00 -6.82118297e-01 -1.68899849e-01 -1.80267811e-01
-4.49973762e-01 2.57378250e-01 6.03316665e-01 -5.86616516e-01
-1.13162637e+00 1.14961587e-01 -3.30619752e-01 -4.35870230e-01
-3.46855640e-01 5.30526638e-01 1.24038339e-01 5.89059055e-01
7.72085309e-01 6.20700538e-01 -3.00666869e-01 -6.15657032e-01
3.75466347e-01 4.21655804e-01 6.09320223e-01 -4.08902973e-01
5.13255477e-01 1.76756755e-01 -1.03594862e-01 -9.04487967e-01
-4.33167070e-01 -1.03106752e-01 -5.71807027e-01 -3.53043139e-01
5.92331886e-01 -1.01028025e+00 -4.51096296e-01 6.43090725e-01
-1.47599089e+00 -9.17444602e-02 -5.65271795e-01 4.87837136e-01
-9.54273224e-01 -4.10531685e-02 -1.10260904e+00 -1.57046127e+00
-4.14639980e-01 -5.33293664e-01 8.01279426e-01 3.43555436e-02
-5.76938927e-01 -8.93730104e-01 5.98887980e-01 -3.64720345e-01
6.40675426e-01 3.29164416e-01 8.36507261e-01 -6.72438622e-01
-3.38653594e-01 -2.39643231e-01 4.52410430e-01 5.76258242e-01
5.99082280e-03 3.11291695e-01 -1.30085909e+00 3.75862457e-02
1.53336540e-01 -1.66462094e-01 8.86426806e-01 5.14879107e-01
8.22259009e-01 -5.14770448e-01 1.52796134e-01 7.00570792e-02
1.10015845e+00 4.81898248e-01 7.81335711e-01 -1.87212437e-01
2.17745334e-01 6.26632273e-01 1.07863313e-02 7.18162239e-01
1.12437442e-01 5.99551618e-01 3.41100901e-01 1.17394768e-01
1.28637567e-01 -4.95060712e-01 8.18947852e-01 1.21929681e+00
-3.23848635e-01 -1.38547555e-01 -6.67922556e-01 9.18513238e-01
-1.91331959e+00 -1.40456522e+00 -1.67550698e-01 1.93903816e+00
1.01683533e+00 4.70916122e-01 4.88113314e-01 6.02455795e-01
5.83154798e-01 2.62222558e-01 -1.83477297e-01 -6.50564253e-01
-3.65881979e-01 4.30816829e-01 -2.79120021e-02 2.67033845e-01
-8.31240356e-01 4.91931438e-01 8.11062050e+00 7.23302782e-01
-1.01466298e+00 -2.09419221e-01 4.09759730e-01 -2.30984002e-01
-2.69740641e-01 -1.80765226e-01 -6.15456283e-01 4.37760025e-01
1.93298244e+00 -3.36714953e-01 8.89788046e-02 7.09230423e-01
4.68277514e-01 4.85207558e-01 -1.12230659e+00 6.00765526e-01
-2.02101678e-01 -1.35316575e+00 4.72146302e-01 -1.35882363e-01
6.46710932e-01 -2.25538447e-01 3.20417196e-01 3.62578422e-01
5.81630707e-01 -1.09087551e+00 1.09000564e+00 1.07953191e+00
2.37083197e-01 -7.94645965e-01 5.87047756e-01 4.28071558e-01
-1.24944520e+00 -2.02727482e-01 -1.54206052e-01 -3.94751877e-01
3.84683996e-01 7.62495339e-01 -1.01419520e+00 7.04897404e-01
5.33908486e-01 5.12746990e-01 -4.69629765e-01 9.90601659e-01
-1.24075226e-01 1.13308859e+00 -4.79821824e-02 2.55356096e-02
2.94355065e-01 2.24327222e-01 8.26883137e-01 1.51257992e+00
6.49811625e-01 -3.66716385e-01 -1.60291910e-01 8.98987234e-01
3.01108807e-01 -1.35544986e-01 -8.26035142e-01 8.95764381e-02
4.47442025e-01 9.30068016e-01 -2.66976088e-01 -4.20664847e-01
1.66186839e-01 6.12666667e-01 1.17994174e-01 5.30102551e-01
-6.83388233e-01 -2.80541688e-01 2.75894612e-01 1.11474119e-01
4.84861046e-01 -2.07022205e-01 -4.50953189e-03 -8.78107369e-01
1.41158164e-03 -7.30046749e-01 3.13072234e-01 -1.14551282e+00
-1.27816796e+00 1.02161479e+00 -3.99619155e-02 -1.35679841e+00
-1.43619740e+00 -7.44399652e-02 -7.29798257e-01 9.83215690e-01
-1.04057181e+00 -9.87235367e-01 3.61571729e-01 6.58106744e-01
6.44373357e-01 -1.40583470e-01 1.09629047e+00 7.59335011e-02
-3.53575915e-01 1.91408843e-01 -8.42743441e-02 -8.58322755e-02
2.69581586e-01 -1.31407642e+00 4.05400097e-01 8.75989020e-01
5.39119720e-01 7.49917328e-01 1.07814300e+00 -3.83481264e-01
-8.31985056e-01 -1.01370120e+00 1.13547385e+00 -3.07615429e-01
8.81675720e-01 -1.76673785e-01 -1.08079088e+00 6.97235286e-01
5.54465950e-01 -7.32819021e-01 8.14132154e-01 1.54970482e-01
-3.45883101e-01 -3.78797911e-02 -6.57284439e-01 4.93531913e-01
7.18199909e-01 -7.40286291e-01 -9.14439380e-01 -2.61657268e-01
1.14936411e+00 -5.68969846e-02 -8.33259225e-01 4.72813606e-01
6.77675664e-01 -1.30485702e+00 9.38338935e-01 -8.70097041e-01
6.52030647e-01 -6.98821321e-02 -2.12109521e-01 -1.32605755e+00
-4.73566860e-01 -1.08868599e+00 -5.74739635e-01 1.54001534e+00
7.18257308e-01 -4.69896078e-01 5.39044976e-01 4.36928898e-01
-9.76403579e-02 -3.55879337e-01 -1.18264604e+00 -8.12461555e-01
1.08136721e-01 -8.42998922e-01 6.78084373e-01 2.87702292e-01
-7.25702569e-02 5.69743097e-01 -1.05433595e+00 7.48759359e-02
3.27148497e-01 -2.19803192e-02 5.54245591e-01 -1.30438125e+00
-5.85229874e-01 -3.94319981e-01 -3.91827255e-01 -8.05062592e-01
1.91235408e-01 -3.67701054e-01 2.45782003e-01 -1.04885125e+00
-1.37975603e-01 5.97089017e-03 -8.20173144e-01 1.23810716e-01
1.92779452e-01 1.25484273e-01 2.25972340e-01 8.74085799e-02
-3.10999185e-01 6.99431419e-01 3.73572320e-01 5.55066392e-02
-4.10259783e-01 4.53729689e-01 -6.41028702e-01 8.79529536e-01
1.00307095e+00 -3.62427771e-01 -4.52142864e-01 -5.42658642e-02
3.48588675e-01 6.00098073e-01 5.54201007e-01 -1.46097958e+00
2.11867064e-01 8.92030075e-02 3.99714410e-01 -7.32345998e-01
7.26230323e-01 -6.83944702e-01 7.03779280e-01 5.13265550e-01
-7.46573985e-01 1.06709160e-01 1.78952321e-01 5.49202800e-01
-5.02062798e-01 -2.87271231e-01 3.59659880e-01 -1.29897356e-01
-3.45610291e-01 -1.27569467e-01 -7.98555315e-01 -3.49291176e-01
4.25366342e-01 7.72482306e-02 9.64541435e-02 -8.80156398e-01
-1.11487710e+00 -3.55563730e-01 -2.99076229e-01 5.85251272e-01
7.49622226e-01 -1.49824250e+00 -9.72266734e-01 1.21781692e-01
-2.39712477e-01 -6.51624739e-01 4.43533719e-01 3.67114812e-01
2.39803299e-01 7.53593862e-01 -2.17997059e-01 -4.95795310e-01
-1.07867980e+00 4.25731450e-01 2.17319801e-01 -6.60760760e-01
-4.54803795e-01 6.76618695e-01 -2.70945847e-01 -1.26942454e-04
3.54326397e-01 -3.73608172e-01 -3.96473438e-01 2.19621077e-01
5.56483448e-01 3.91654819e-01 -1.19378403e-01 -5.62063336e-01
2.82408390e-02 1.26317754e-01 7.95701891e-02 -7.02384949e-01
1.37228751e+00 -1.70857340e-01 3.27330939e-02 1.49916530e+00
7.47516632e-01 4.76022884e-02 -1.07816947e+00 -2.16923445e-01
2.09026843e-01 7.70599693e-02 -3.45990181e-01 -5.97970307e-01
-5.93762040e-01 8.02178919e-01 3.88498813e-01 8.90947700e-01
1.23928940e+00 -1.70825824e-01 7.26373196e-01 2.83290565e-01
5.17947674e-01 -9.21643913e-01 8.66968408e-02 6.14546895e-01
1.13217437e+00 -4.53076541e-01 -4.74789470e-01 2.01409712e-01
-4.79278952e-01 1.04730022e+00 6.33654743e-02 -3.73086840e-01
5.99873245e-01 3.42503011e-01 -7.97337964e-02 3.42024237e-01
-1.66539359e+00 -2.82994211e-02 2.85988539e-01 6.41340077e-01
6.65314674e-01 1.65029630e-01 -2.91235726e-02 8.85600388e-01
-7.65637279e-01 1.66819215e-01 7.25284576e-01 7.74298489e-01
-4.64004517e-01 -1.07439375e+00 -4.95336622e-01 2.38415852e-01
-6.10568225e-01 -1.06824480e-01 -4.90508616e-01 4.26114619e-01
-1.26211390e-01 1.23875403e+00 -1.10422850e-01 -5.53898275e-01
3.23994875e-01 5.95167279e-01 8.40095133e-02 -1.78695604e-01
-8.58679891e-01 3.33730787e-01 1.98862553e-01 -3.57966483e-01
-4.28445071e-01 -4.88004208e-01 -8.09385717e-01 -6.00272417e-03
4.62752022e-02 3.73381048e-01 3.81824344e-01 4.90710646e-01
4.54507202e-01 9.71406639e-01 9.73720372e-01 -1.09976196e+00
-7.30648994e-01 -1.31767023e+00 -8.29061329e-01 2.00972080e-01
6.43752158e-01 -2.23647743e-01 -5.20260870e-01 5.09990335e-01] | [15.618600845336914, 5.748224258422852] |
8f3c40d0-29ae-4c89-a9e0-8c669c62fd75 | algorithmic-optimisations-for-iterative | 1304.7211 | null | http://arxiv.org/abs/1304.7211v1 | http://arxiv.org/pdf/1304.7211v1.pdf | Algorithmic Optimisations for Iterative Deconvolution Methods | We investigate possibilities to speed up iterative algorithms for non-blind
image deconvolution. We focus on algorithms in which convolution with the
point-spread function to be deconvolved is used in each iteration, and aim at
accelerating these convolution operations as they are typically the most
expensive part of the computation. We follow two approaches: First, for some
practically important specific point-spread functions, algorithmically
efficient sliding window or list processing techniques can be used. In some
constellations this allows faster computation than via the Fourier domain.
Second, as iterations progress, computation of convolutions can be restricted
to subsets of pixels. For moderate thinning rates this can be done with almost
no impact on the reconstruction quality. Both approaches are demonstrated in
the context of Richardson-Lucy deconvolution but are not restricted to this
method. | ['Martin Erler', 'Martin Welk'] | 2013-04-26 | null | null | null | null | ['image-deconvolution'] | ['computer-vision'] | [ 2.19637722e-01 -1.88564450e-01 5.91594875e-01 -1.10108331e-02
-2.36520439e-01 -5.87210357e-01 6.12489581e-01 -1.01856947e-01
-9.78965402e-01 9.53559399e-01 2.06936151e-03 -3.78078938e-01
-3.54320824e-01 -5.60829520e-01 -3.55567694e-01 -8.50662231e-01
-3.52433980e-01 3.65626872e-01 4.92214024e-01 -1.17941007e-01
5.20094991e-01 7.34073281e-01 -1.39297342e+00 9.41430554e-02
8.61500263e-01 6.07924938e-01 5.03780723e-01 7.89297938e-01
-2.43026465e-01 3.94348353e-01 -3.09425116e-01 -2.35762715e-01
3.61217886e-01 -5.82364857e-01 -8.13427746e-01 2.12844580e-01
-2.50452101e-01 -3.61636758e-01 1.31531358e-01 1.11771369e+00
4.51849639e-01 2.08041266e-01 7.59209752e-01 -3.28796089e-01
8.01521689e-02 4.49411392e-01 -5.82344294e-01 2.22462982e-01
2.79409111e-01 -5.24251051e-02 3.11537087e-01 -1.02359891e+00
4.38639879e-01 8.82786453e-01 1.13052440e+00 1.35954946e-01
-1.50429285e+00 -2.06224203e-01 -5.20142078e-01 3.60118113e-02
-1.35565627e+00 -6.25740826e-01 4.37898576e-01 -2.47892335e-01
7.79024303e-01 4.32902485e-01 5.53977847e-01 1.16261780e-01
-1.95961315e-02 2.45632291e-01 1.36963391e+00 -7.31293797e-01
4.08556551e-01 1.03950262e-01 -6.44769296e-02 3.79014552e-01
3.12508851e-01 1.67615399e-01 -3.80315259e-02 -1.46439478e-01
1.10490513e+00 -2.68208444e-01 -7.22114742e-01 -1.02723047e-01
-1.26357055e+00 6.83376193e-01 2.64023721e-01 5.48715234e-01
-6.66710734e-01 1.10609792e-01 3.63333762e-01 5.12589872e-01
7.29450405e-01 4.25043821e-01 -2.27761477e-01 -1.79331154e-02
-1.44991374e+00 2.84275293e-01 8.00531745e-01 3.88016433e-01
8.47374141e-01 -2.46275380e-01 -3.82855944e-02 7.74912655e-01
1.38657719e-01 1.31607741e-01 2.77315527e-01 -1.10058427e+00
4.87364158e-02 7.42296800e-02 5.52369177e-01 -2.25706056e-01
-4.94181305e-01 -3.30750585e-01 -7.42732108e-01 9.14780617e-01
7.52440274e-01 -3.66594881e-01 -9.21946883e-01 8.99333596e-01
2.88177967e-01 3.72144133e-01 1.18247822e-01 9.33436751e-01
3.89916509e-01 4.57229704e-01 -3.04493129e-01 -5.23645282e-01
1.43685389e+00 -4.63784307e-01 -6.02524102e-01 2.40068913e-01
5.28260887e-01 -1.33380342e+00 4.58321244e-01 4.41420317e-01
-1.33882594e+00 -2.17717484e-01 -8.73354137e-01 1.34562209e-01
-8.57292488e-02 2.61591852e-01 3.95221800e-01 8.82063091e-01
-1.25201988e+00 1.21405816e+00 -7.00294137e-01 -2.43145421e-01
2.51116931e-01 3.92275006e-01 -6.67027086e-02 -5.12484908e-02
-7.32495248e-01 9.87476349e-01 3.01293135e-01 1.99989766e-01
-2.13669345e-01 -7.16794193e-01 -3.15884888e-01 2.20924467e-01
-1.73150986e-01 -6.42498195e-01 1.04822564e+00 -1.10613048e+00
-1.57926011e+00 7.58388162e-01 -2.42430881e-01 -6.95263863e-01
8.62501740e-01 -7.01167732e-02 -1.71488784e-02 4.06401783e-01
-2.43834138e-01 2.55821258e-01 1.18286300e+00 -1.25880265e+00
-1.91536903e-01 4.18729931e-02 -1.71097606e-01 3.17869842e-01
8.32921453e-03 3.97245228e-01 -1.24837518e-01 -6.15226448e-01
3.68255764e-01 -7.27094829e-01 -3.82460982e-01 1.53001904e-01
-1.86326019e-02 1.43893793e-01 3.57907861e-01 -8.75670552e-01
7.71642208e-01 -2.22314024e+00 -7.33643025e-02 2.49526307e-01
1.41327080e-04 3.94441038e-01 2.15062290e-01 4.27252918e-01
-2.52088606e-01 -3.56034756e-01 -6.05688870e-01 -5.32691658e-01
-3.65099251e-01 6.05435371e-02 -3.03359091e-01 9.16762948e-01
4.08795616e-03 4.04942572e-01 -5.00330567e-01 -3.06202829e-01
2.61372417e-01 4.56651300e-01 -4.68124062e-01 -1.20499171e-01
1.54566079e-01 5.52140117e-01 -1.06261052e-01 7.36310259e-02
1.14863527e+00 -1.32721961e-02 -1.26296535e-01 -6.26212806e-02
-7.31741309e-01 1.25902340e-01 -1.34421539e+00 1.34172702e+00
-4.81412828e-01 7.05156505e-01 5.85655034e-01 -1.11567605e+00
6.58097863e-01 6.28669143e-01 2.66669989e-01 -2.73802817e-01
-1.02511272e-02 4.88503128e-01 1.17739411e-02 -3.48608106e-01
6.20603085e-01 -7.26550281e-01 8.10620487e-01 5.03106475e-01
-1.23699091e-01 -8.84144455e-02 2.54794419e-01 -2.63285488e-01
9.45270240e-01 1.26599789e-01 3.48754585e-01 -6.59257591e-01
6.04235113e-01 1.40465006e-01 -1.56778772e-03 5.87902725e-01
2.36645415e-01 9.31756139e-01 3.58114421e-01 -1.36566564e-01
-1.20519483e+00 -7.36788154e-01 -5.65817237e-01 5.85589647e-01
1.09896220e-01 -5.46807423e-02 -9.62376595e-01 -1.60436898e-01
-4.97110665e-01 5.12861431e-01 -3.07506174e-01 4.58463758e-01
-5.86408734e-01 -1.05829549e+00 2.61173278e-01 1.50236905e-01
5.17967641e-01 -9.44715500e-01 -7.79752254e-01 2.36619517e-01
2.46290773e-01 -8.16413283e-01 -1.13430724e-01 2.81401694e-01
-1.44238472e+00 -1.02056181e+00 -1.32286954e+00 -5.59324324e-01
8.55012774e-01 4.35396105e-01 7.34568119e-01 3.04310471e-01
2.78087165e-02 3.40157181e-01 -3.72797161e-01 -1.42174795e-01
-4.29403007e-01 -5.62534630e-01 -1.57950431e-01 7.48765245e-02
1.50001403e-02 -8.65712643e-01 -6.93324685e-01 3.12990487e-01
-8.53893459e-01 -8.93366486e-02 5.23412526e-01 8.01305473e-01
3.09782535e-01 3.89008224e-01 8.16765577e-02 -7.69646049e-01
6.73723400e-01 -1.35343701e-01 -8.63941193e-01 -1.20455310e-01
-3.71297836e-01 8.76498595e-02 7.49635160e-01 -3.23493958e-01
-1.06825089e+00 2.04528734e-01 -2.13051513e-01 -4.28994656e-01
-1.86688140e-01 2.98985541e-01 5.26222765e-01 -7.26916790e-01
8.64052951e-01 3.31972986e-01 3.09331715e-01 -8.45372319e-01
1.10721946e-01 5.08167148e-01 4.24540848e-01 -2.52129316e-01
8.63888741e-01 7.42232203e-01 3.32395971e-01 -1.21519756e+00
1.95686758e-01 -6.68378234e-01 -4.79137123e-01 -1.39653951e-01
6.73391402e-01 -7.20870614e-01 -5.48106074e-01 4.69769895e-01
-1.45115256e+00 -3.30449879e-01 -4.45638746e-01 8.47340107e-01
-6.83044612e-01 6.60487652e-01 -5.34421444e-01 -1.08220029e+00
-4.34951261e-02 -9.84309494e-01 5.51324069e-01 2.16454893e-01
-1.42176583e-01 -1.01040840e+00 -2.11781897e-02 -7.63881654e-02
5.89911163e-01 -3.93824905e-01 6.10493481e-01 -3.94308686e-01
-4.81003106e-01 -1.17294811e-01 -4.30630505e-01 4.25153553e-01
-5.77548333e-02 -2.91722596e-01 -1.04112625e+00 -3.05441260e-01
5.52618444e-01 3.07360619e-01 1.04331398e+00 6.60829365e-01
8.70973766e-01 -1.21205889e-01 -2.97171056e-01 7.38835990e-01
1.44038355e+00 -1.00009993e-01 1.10419595e+00 2.51015455e-01
1.72120720e-01 7.11210310e-01 2.94093043e-01 3.86184067e-01
-3.19491297e-01 6.01580501e-01 1.86351687e-01 -2.37022832e-01
-2.49832243e-01 5.09486556e-01 1.41173765e-01 4.34646606e-01
-8.51635158e-01 1.10467814e-01 -5.85337162e-01 6.02761269e-01
-1.48122001e+00 -1.12739182e+00 -8.03944290e-01 2.54911947e+00
4.99555290e-01 -7.95794874e-02 7.84190297e-02 4.07364935e-01
8.36616814e-01 -4.32873741e-02 1.04974434e-01 -3.10633957e-01
1.46825984e-01 5.51963687e-01 9.91477013e-01 7.41524398e-01
-9.34303463e-01 3.37214649e-01 6.81851959e+00 7.53561974e-01
-9.85509276e-01 4.00241375e-01 2.32856244e-01 -8.20220932e-02
-1.90540627e-01 2.62480736e-01 -2.70437926e-01 6.02662921e-01
7.44522274e-01 2.51935661e-01 7.01866031e-01 3.21252674e-01
3.80644470e-01 -6.69321299e-01 -5.19774139e-01 9.78303254e-01
-3.88183475e-01 -1.28969693e+00 -5.79899669e-01 2.22458452e-01
5.03954053e-01 5.18351095e-03 -3.06436121e-01 -3.78010690e-01
-1.07875608e-01 -1.02089727e+00 4.58619297e-01 6.23479724e-01
8.02229881e-01 -6.93850875e-01 6.79113507e-01 3.22046846e-01
-8.32750678e-01 1.02157637e-01 -5.28677583e-01 -1.11557983e-01
4.59014982e-01 9.46783364e-01 -4.80589211e-01 5.45653999e-01
5.71616054e-01 9.54914913e-02 2.84118839e-02 1.69451833e+00
1.08028743e-02 6.77119792e-01 -6.50007188e-01 1.98117226e-01
1.95526570e-01 -6.73835158e-01 9.44643319e-01 1.29951978e+00
7.94513643e-01 4.15680170e-01 -5.74557602e-01 9.06469345e-01
4.31136519e-01 6.63799644e-02 -3.45609039e-01 4.15760815e-01
1.51892798e-02 1.14232850e+00 -1.08517480e+00 -2.13382289e-01
-5.60722530e-01 1.35292172e+00 -2.13511512e-01 5.80070138e-01
-4.14747894e-01 -5.10453105e-01 5.63669503e-01 5.22142053e-01
8.04435134e-01 -3.28826129e-01 -5.31024873e-01 -9.50023770e-01
-1.64283589e-01 -3.60494107e-01 5.28016686e-02 -7.73506761e-01
-7.79860020e-01 5.36333263e-01 -6.65673837e-02 -1.24282372e+00
-9.61662605e-02 -6.91487730e-01 -1.00808215e+00 1.46680152e+00
-1.60899961e+00 -6.03507340e-01 1.00566052e-01 7.42554009e-01
2.58311749e-01 3.58783826e-02 8.13934028e-01 2.17294410e-01
8.72998461e-02 -3.61048356e-02 6.31641865e-01 -3.41248095e-01
4.91819918e-01 -1.14875352e+00 1.22687913e-01 1.00646126e+00
-3.53594543e-03 7.17620850e-01 1.15432525e+00 -4.82309163e-01
-9.02524769e-01 -5.64056993e-01 9.01728034e-01 1.55052587e-01
4.42021459e-01 -3.67512256e-02 -1.18580270e+00 2.56756395e-01
3.61635387e-01 9.14120227e-02 2.62299716e-01 -2.74733789e-02
2.91887432e-01 1.71103448e-01 -1.22782481e+00 3.41028780e-01
5.74293315e-01 -3.95625979e-01 -4.47079182e-01 3.72747391e-01
-1.61385611e-02 -3.10433388e-01 -5.35721123e-01 5.87788585e-04
1.85112923e-01 -1.38064361e+00 1.08826745e+00 7.31517896e-02
1.93969846e-01 -6.17073178e-01 2.52184778e-01 -1.31970191e+00
-3.08866233e-01 -9.68799889e-01 2.36923456e-01 9.24980462e-01
1.81597993e-01 -9.12701964e-01 5.34120381e-01 1.55205756e-01
-3.50432768e-02 -2.06265494e-01 -1.17832887e+00 -7.89952099e-01
-1.11894645e-01 -2.75679141e-01 2.39925787e-01 7.06122220e-01
-1.08210348e-01 -1.77652746e-01 -2.42588133e-01 2.72250086e-01
7.97444761e-01 1.19555593e-01 3.13658893e-01 -1.25296140e+00
-5.30567050e-01 -5.05983233e-01 -3.61747324e-01 -9.90180314e-01
-3.31024498e-01 -6.73102975e-01 5.80507219e-02 -1.31707156e+00
-1.49404809e-01 -5.24621189e-01 9.69067812e-02 -2.13079154e-03
-1.93598159e-02 6.39471948e-01 -5.74881919e-02 4.12049383e-01
1.59658462e-01 7.55867586e-02 1.16278279e+00 5.19373775e-01
-2.41592795e-01 3.30413550e-01 -1.91025540e-01 7.81295180e-01
6.29104495e-01 -3.45609903e-01 -3.31456095e-01 -2.27387741e-01
1.69451311e-01 1.62065789e-01 6.67759776e-01 -1.08025336e+00
4.41094577e-01 1.21916793e-01 3.41736406e-01 -3.78862590e-01
4.42119360e-01 -8.18852901e-01 4.60278332e-01 4.00514364e-01
2.17499733e-01 -3.51993889e-01 2.30851248e-01 3.28128964e-01
-1.47638589e-01 -1.01949620e+00 1.18755221e+00 -3.57863247e-01
-3.78780454e-01 -2.54136235e-01 -5.86718559e-01 -6.37764454e-01
7.68685520e-01 -4.55208689e-01 7.16286749e-02 -4.21691120e-01
-8.00050914e-01 -4.02714342e-01 6.67653382e-01 -6.22859836e-01
4.17229652e-01 -9.87469852e-01 -7.91987777e-01 1.44027367e-01
-4.87905741e-01 -1.07001923e-01 2.48051807e-01 1.44287181e+00
-1.03320026e+00 2.62755603e-01 -3.74330766e-02 -2.75825232e-01
-1.42238688e+00 5.78880131e-01 4.97777611e-01 -8.94735754e-03
-8.58949661e-01 8.48859191e-01 2.18890995e-01 1.78785965e-01
-2.35450819e-01 -1.78925827e-01 -2.63943046e-01 1.18037581e-01
8.02780151e-01 6.07148290e-01 1.44313216e-01 -5.20640552e-01
-1.84146076e-01 6.93550229e-01 1.65908828e-01 -4.42688853e-01
1.45957029e+00 -2.47605890e-01 -5.43934703e-01 -2.90850699e-02
9.02047336e-01 2.84850359e-01 -1.45464242e+00 1.97251402e-02
-1.99141055e-01 -4.58370775e-01 3.97710025e-01 -4.84968573e-01
-8.03111076e-01 8.32478166e-01 5.96035361e-01 5.74622750e-01
1.44508350e+00 -1.96695164e-01 4.21337217e-01 -5.44621050e-03
2.22333983e-01 -8.44014525e-01 -7.13809431e-01 4.11117136e-01
7.89019525e-01 -7.25630581e-01 3.86972606e-01 -5.44241369e-01
-8.64546672e-02 1.52352452e+00 -3.72203112e-01 -4.04620945e-01
4.93749440e-01 3.26268762e-01 -1.73713088e-01 -1.40766233e-01
-8.52588117e-02 -4.85539377e-01 1.67778015e-01 5.34766197e-01
4.64722067e-01 -1.27998665e-01 -9.47407067e-01 7.10658878e-02
1.27803877e-01 1.51914880e-01 6.90331578e-01 6.72035396e-01
-7.07825422e-01 -1.17623878e+00 -1.05061257e+00 4.35626894e-01
-5.32032788e-01 -5.22851765e-01 1.69053599e-01 5.14975190e-01
3.93882319e-02 7.08768666e-01 9.19405464e-03 2.66988218e-01
3.74836475e-03 1.99575499e-01 8.43519390e-01 -2.32447237e-01
-6.62173092e-01 3.92895877e-01 2.36082554e-01 -3.68050575e-01
-5.94305754e-01 -8.53835881e-01 -1.27184606e+00 -4.66856003e-01
-3.82087678e-01 2.02240020e-01 8.81715775e-01 1.03250670e+00
-1.20033309e-01 2.43002847e-01 3.87536824e-01 -1.29317605e+00
-3.55921447e-01 -9.01389122e-01 -1.03812075e+00 -4.31553610e-02
4.76505578e-01 -4.18061435e-01 -6.30779266e-01 9.40043777e-02] | [11.64954662322998, -2.601712703704834] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.