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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
015113fa-1b9e-4d7b-a9d0-e9543f067cbd | neuspell-a-neural-spelling-correction-toolkit | 2010.11085 | null | https://arxiv.org/abs/2010.11085v1 | https://arxiv.org/pdf/2010.11085v1.pdf | NeuSpell: A Neural Spelling Correction Toolkit | We introduce NeuSpell, an open-source toolkit for spelling correction in English. Our toolkit comprises ten different models, and benchmarks them on naturally occurring misspellings from multiple sources. We find that many systems do not adequately leverage the context around the misspelt token. To remedy this, (i) we train neural models using spelling errors in context, synthetically constructed by reverse engineering isolated misspellings; and (ii) use contextual representations. By training on our synthetic examples, correction rates improve by 9% (absolute) compared to the case when models are trained on randomly sampled character perturbations. Using richer contextual representations boosts the correction rate by another 3%. Our toolkit enables practitioners to use our proposed and existing spelling correction systems, both via a unified command line, as well as a web interface. Among many potential applications, we demonstrate the utility of our spell-checkers in combating adversarial misspellings. The toolkit can be accessed at neuspell.github.io. Code and pretrained models are available at http://github.com/neuspell/neuspell. | ['Graham Neubig', 'Danish Pruthi', 'Sai Muralidhar Jayanthi'] | 2020-10-21 | null | https://aclanthology.org/2020.emnlp-demos.21 | https://aclanthology.org/2020.emnlp-demos.21.pdf | emnlp-2020-11 | ['spelling-correction'] | ['natural-language-processing'] | [ 4.60583299e-01 -2.18868732e-01 2.45136060e-02 -5.54823913e-02
-1.37722087e+00 -9.52695906e-01 5.70063889e-01 2.05792516e-01
-5.06661952e-01 7.76647091e-01 5.07286608e-01 -6.46516681e-01
4.34203923e-01 -5.04873991e-01 -9.67409670e-01 -3.14291567e-01
3.90364826e-01 2.21148565e-01 1.41753659e-01 -6.03071988e-01
5.15074670e-01 2.17764959e-01 -1.09144402e+00 5.35120904e-01
1.11353612e+00 -1.66543126e-02 1.41638234e-01 1.05078769e+00
5.90133816e-02 7.24511385e-01 -9.86018062e-01 -7.18504250e-01
2.99286842e-01 -2.95298636e-01 -8.05858850e-01 -5.77367187e-01
6.25389457e-01 -2.04531759e-01 -3.29304397e-01 1.25394058e+00
6.49982095e-01 3.46438661e-02 3.67435545e-01 -8.82694781e-01
-1.09988785e+00 8.87878060e-01 -2.77771592e-01 2.94647962e-01
6.18804097e-01 3.50115567e-01 9.22572196e-01 -1.05136216e+00
3.13137114e-01 1.03907382e+00 1.04146421e+00 1.02158260e+00
-1.08386815e+00 -7.72430539e-01 -1.63536537e-02 2.64579263e-02
-1.24814963e+00 -6.01814985e-01 3.88755620e-01 -1.59950227e-01
1.12471843e+00 5.49946606e-01 1.46136895e-01 1.45801914e+00
1.85527176e-01 9.65254068e-01 8.46648395e-01 -7.96385825e-01
-1.98932230e-01 -4.06297408e-02 1.43391237e-01 5.26660800e-01
2.30702221e-01 3.03872496e-01 -4.62606668e-01 -2.49622360e-01
3.27419996e-01 -1.99395344e-01 -3.63963127e-01 2.71815270e-01
-1.12648869e+00 5.68453133e-01 1.79536015e-01 2.44550079e-01
9.73882005e-02 3.79362434e-01 4.97523278e-01 3.51546258e-01
4.35426593e-01 8.67155373e-01 -6.30933464e-01 -3.76706302e-01
-1.08525634e+00 5.91055214e-01 6.64527774e-01 1.05283320e+00
4.93166149e-01 2.20010608e-01 -2.20960736e-01 9.05433178e-01
-1.01954982e-01 6.43614948e-01 7.67109513e-01 -6.02647543e-01
6.86840773e-01 7.20153525e-02 3.96193266e-01 -7.26313353e-01
-4.53448370e-02 -3.69542807e-01 -3.53741586e-01 2.88713239e-02
5.78398347e-01 -3.26404601e-01 -1.01572967e+00 1.60328209e+00
-1.49849281e-01 1.04260407e-01 -5.31702936e-02 6.27911866e-01
6.55294061e-01 5.96061051e-01 -1.90214452e-03 2.78181136e-01
8.72174501e-01 -1.23774374e+00 -5.13196945e-01 -5.57605326e-01
1.00512218e+00 -1.17112637e+00 1.50315177e+00 4.25802410e-01
-1.10096419e+00 -2.13635772e-01 -8.87572110e-01 4.90849353e-02
-5.04499078e-01 9.64030102e-02 1.44845337e-01 8.08464587e-01
-1.08170807e+00 1.02322376e+00 -8.34994674e-01 -2.93847680e-01
5.02488464e-02 6.92230240e-02 -1.36314422e-01 -3.85623910e-02
-1.46383572e+00 1.05518782e+00 2.24574566e-01 -2.01206684e-01
-8.09594572e-01 -9.72073019e-01 -9.31759834e-01 -1.78360149e-01
1.85926735e-01 -3.50907266e-01 1.81350219e+00 -1.09941435e+00
-1.23794913e+00 7.16714084e-01 -2.83952445e-01 -4.42229748e-01
6.39023006e-01 -8.37259412e-01 -6.49286926e-01 -2.14678720e-01
1.45366848e-01 1.28242195e-01 6.94254100e-01 -1.10394394e+00
-4.18554544e-01 2.44468719e-01 -1.63862482e-01 1.98316082e-01
-1.94977865e-01 5.07155895e-01 -3.46995562e-01 -1.23496425e+00
-5.39524198e-01 -9.50677335e-01 -2.56083995e-01 -5.56470692e-01
-6.76793575e-01 2.76287943e-01 5.76453209e-01 -1.14629257e+00
1.48702872e+00 -1.98576045e+00 -1.16235457e-01 9.39196125e-02
-2.76782691e-01 1.01974440e+00 -3.82052243e-01 7.44292200e-01
-3.90342385e-01 5.73453307e-01 -5.49353421e-01 -3.58133405e-01
-7.94384703e-02 -1.37778908e-01 -6.53952837e-01 1.72596306e-01
2.20095143e-01 9.11594570e-01 -1.24842417e+00 1.28585875e-01
1.51685372e-01 2.30936870e-01 -8.49092364e-01 1.37048751e-01
-1.42716870e-01 2.63936877e-01 1.24276660e-01 5.95826805e-01
5.37757814e-01 1.39235735e-01 5.40783023e-03 3.11530650e-01
2.93574389e-02 8.58727098e-01 -1.03556073e+00 1.69447851e+00
-6.08591378e-01 6.78420484e-01 -1.55788139e-01 -3.43174577e-01
7.67011881e-01 2.07594350e-01 -2.31144980e-01 -2.09306091e-01
-1.78243071e-01 4.42219973e-01 -2.17326298e-01 -2.38998428e-01
9.51951623e-01 1.17483020e-01 -3.94893408e-01 5.95686674e-01
-6.10924996e-02 -3.58597994e-01 2.80993998e-01 5.19557595e-01
1.29873765e+00 2.05758125e-01 3.37361306e-01 9.16188490e-03
3.90995622e-01 2.59252964e-03 6.10925615e-01 1.13061166e+00
-1.60851747e-01 7.49789596e-01 2.02508092e-01 -3.85884017e-01
-1.18035579e+00 -1.18760669e+00 1.39387503e-01 1.28765130e+00
-2.73718119e-01 -9.12770092e-01 -9.89359856e-01 -6.93854213e-01
8.28108117e-02 1.26840889e+00 -5.76654494e-01 -3.53199750e-01
-7.10713208e-01 -6.64607048e-01 1.19444311e+00 6.12782478e-01
1.80277169e-01 -1.24519360e+00 -5.50378859e-02 1.19251855e-01
-2.92381376e-01 -6.58608675e-01 -9.93859708e-01 1.21820785e-01
-4.01301354e-01 -8.76474082e-01 -5.85349262e-01 -6.60093904e-01
5.49577296e-01 1.24201544e-01 1.21031451e+00 5.98735034e-01
-1.60247073e-01 1.69161439e-01 -6.78432643e-01 -5.23806334e-01
-9.80025530e-01 2.10258141e-01 3.15332890e-01 -5.88913143e-01
4.29834634e-01 -3.83713186e-01 -2.13420734e-01 8.04941580e-02
-1.01804447e+00 1.75195988e-02 2.86260337e-01 9.54703927e-01
1.92253590e-01 -5.33629775e-01 5.18963933e-01 -1.22193635e+00
8.99864137e-01 -3.87170047e-01 -4.95552301e-01 2.39213794e-01
-5.10517478e-01 6.26976117e-02 8.79134417e-01 -2.52849013e-01
-8.38684022e-01 -1.74164295e-01 -6.34639204e-01 -2.52607912e-01
-4.12271053e-01 3.51524234e-01 1.23196011e-02 -5.72678298e-02
1.08508992e+00 3.15618753e-01 -3.83907825e-01 -6.68225944e-01
4.54108864e-01 7.93523133e-01 8.14390063e-01 -7.84574866e-01
1.06600046e+00 -2.44198933e-01 -9.00121629e-01 -5.29948115e-01
-8.16444457e-01 -2.13952973e-01 -4.55104470e-01 1.42250657e-01
3.35848600e-01 -7.05514848e-01 -1.34688079e-01 7.02598035e-01
-1.17110944e+00 -7.44727254e-01 -1.37023494e-01 6.25475794e-02
-3.66266221e-01 5.77588141e-01 -8.68379891e-01 -5.28071761e-01
-5.16157269e-01 -1.10682917e+00 8.37175369e-01 1.60762593e-01
-6.58582926e-01 -8.51525843e-01 2.38293990e-01 2.45223969e-01
4.53540206e-01 4.20568213e-02 7.68030167e-01 -1.00091934e+00
-1.60194948e-01 -3.58155012e-01 1.66199341e-01 4.22382325e-01
3.03748995e-01 3.50309968e-01 -9.60330009e-01 -3.05584371e-01
-6.03153646e-01 -3.79806250e-01 9.00746346e-01 -4.34724614e-02
1.29111540e+00 -6.43153787e-01 -1.31788969e-01 9.13573444e-01
1.08583093e+00 -3.23358513e-02 6.59253597e-01 5.74786544e-01
7.78750420e-01 1.31022066e-01 5.37928760e-01 3.84013444e-01
1.28695980e-01 4.45229888e-01 2.57670373e-01 2.36118361e-01
-5.86365946e-02 -7.93691278e-01 6.56815886e-01 9.33849633e-01
1.25455959e-02 -3.47839892e-01 -1.27474713e+00 6.69440031e-01
-1.59934819e+00 -1.24115288e+00 -1.46441702e-02 2.27537322e+00
1.25469196e+00 2.21524626e-01 2.00507976e-02 1.99837178e-01
8.45789969e-01 2.01269612e-01 -3.83499652e-01 -9.64316845e-01
1.55476583e-02 4.12112147e-01 7.76930451e-01 9.88569379e-01
-1.17240024e+00 1.56133640e+00 6.39318609e+00 8.88802707e-01
-9.47278023e-01 -7.31115639e-02 3.36716443e-01 -3.86786699e-01
-5.86064100e-01 1.25068594e-02 -8.60019445e-01 6.33522093e-01
1.12157714e+00 -3.85224283e-01 7.77874410e-01 7.15653300e-01
2.08490625e-01 3.05152357e-01 -8.49669278e-01 6.10701382e-01
2.28274882e-01 -1.37361324e+00 9.07017663e-02 -4.00344431e-01
8.02895069e-01 1.62137613e-01 1.11463010e-01 5.16141236e-01
1.07758307e+00 -1.18438911e+00 9.33061004e-01 2.79182911e-01
1.01015794e+00 -9.37394738e-01 4.85100955e-01 4.33831155e-01
-7.27869332e-01 1.38990223e-01 -2.18044698e-01 -1.46103770e-01
8.20869654e-02 4.10618752e-01 -1.04586673e+00 4.48451549e-01
6.61846876e-01 7.68905520e-01 -8.71280551e-01 1.05517817e+00
-7.90169835e-01 1.01519036e+00 -4.62625958e-02 -7.87047520e-02
-1.67180840e-02 3.02921981e-01 7.82234788e-01 1.80965161e+00
2.57754624e-01 -1.74483165e-01 -9.34976041e-02 5.97901523e-01
-2.01131061e-01 5.66154607e-02 -6.61181152e-01 -2.37261146e-01
8.29279542e-01 9.59783018e-01 -1.26118571e-01 -2.51809478e-01
-1.77537978e-01 1.37605190e+00 4.66703713e-01 4.47666854e-01
-9.93990600e-01 -8.94953251e-01 9.54596281e-01 -4.01372761e-02
1.07828997e-01 -2.36477047e-01 -3.98749739e-01 -1.46087503e+00
-1.85126230e-01 -1.49909520e+00 1.62479654e-01 -8.63687336e-01
-1.09190917e+00 5.82299113e-01 -2.16084197e-01 -1.18388855e+00
-3.67839128e-01 -5.91626108e-01 -8.96168828e-01 1.06138349e+00
-1.24246883e+00 -9.11270142e-01 1.75853223e-02 3.80799890e-01
7.70754039e-01 -1.60279408e-01 9.82465744e-01 1.79949090e-01
-6.90338790e-01 1.21049976e+00 2.16848224e-01 4.65732843e-01
1.30220759e+00 -1.44088650e+00 1.36907101e+00 1.43422413e+00
3.01687598e-01 1.03106546e+00 9.50415432e-01 -9.29854333e-01
-7.63377130e-01 -1.34404910e+00 1.19656348e+00 -9.44849908e-01
7.92600453e-01 -3.55933040e-01 -1.00211632e+00 9.79918540e-01
3.51147085e-01 -2.52648354e-01 7.90054917e-01 3.85385230e-02
-6.32190764e-01 3.02181989e-01 -1.04214299e+00 1.05233288e+00
1.02652764e+00 -6.27697349e-01 -8.78127933e-01 3.26031029e-01
9.27624524e-01 -8.29544187e-01 -3.17628354e-01 1.42895103e-01
3.88048947e-01 -8.04572999e-01 8.47416520e-01 -8.95483136e-01
7.12570190e-01 -3.05747688e-01 -1.46628842e-01 -1.94657195e+00
-4.00912315e-01 -9.85182583e-01 4.64401320e-02 1.33969450e+00
8.02619874e-01 -5.51531374e-01 4.31197613e-01 5.51238358e-01
-3.47508073e-01 -4.46700662e-01 -3.98734719e-01 -7.59878337e-01
5.33191741e-01 -6.64285302e-01 8.30450356e-01 9.74246085e-01
1.79608196e-01 -1.85350507e-01 -6.70081019e-01 3.67352843e-01
2.50381023e-01 -4.34791654e-01 8.33003223e-01 -4.53677088e-01
-5.30871332e-01 -3.21237892e-01 1.32543206e-01 -8.81727993e-01
1.56466797e-01 -1.18364418e+00 2.71521866e-01 -1.22210908e+00
-2.23530233e-01 -4.17085558e-01 -2.66523272e-01 8.26631069e-01
-6.55041099e-01 3.05960000e-01 5.02424240e-01 1.55137405e-01
-1.96952343e-01 2.38057435e-01 7.43382931e-01 -7.29526654e-02
-4.11301740e-02 8.49331170e-03 -1.08607602e+00 9.74429846e-01
1.32903111e+00 -7.42474556e-01 1.68311313e-01 -8.38767409e-01
2.97201216e-01 -3.58087271e-01 3.65438640e-01 -1.05482197e+00
3.25760208e-02 -2.47668132e-01 3.15916210e-01 -2.04188719e-01
8.76053050e-02 -2.35567659e-01 -2.83339113e-01 4.59121138e-01
-7.20560431e-01 4.84314144e-01 5.56941688e-01 3.76641929e-01
1.23103850e-01 -6.17831290e-01 7.31870234e-01 -2.66535699e-01
-6.35846674e-01 1.08804982e-02 -6.51445925e-01 3.40648741e-01
6.08407497e-01 7.65818954e-02 -5.33797681e-01 -4.47538674e-01
-4.57511663e-01 -3.18852775e-02 8.66169751e-01 5.63719273e-01
4.88977462e-01 -1.23087442e+00 -7.23467410e-01 4.04108644e-01
5.88013381e-02 -4.08505827e-01 -9.80016217e-02 4.39680666e-01
-9.07246053e-01 1.64543003e-01 1.28740938e-02 6.59884065e-02
-1.29972339e+00 4.34983909e-01 4.11819994e-01 -2.46697694e-01
-3.17283809e-01 1.16832876e+00 -2.07130581e-01 -8.93299460e-01
1.04817025e-01 -3.10809642e-01 1.64270967e-01 -5.19633889e-01
9.85822380e-01 2.32834294e-01 2.74807572e-01 -4.18745935e-01
-5.60915411e-01 1.53355509e-01 -4.15546805e-01 -2.29402617e-01
1.14648831e+00 8.75147283e-02 1.99519306e-01 2.02977881e-01
8.16294670e-01 9.09302950e-01 -1.06948304e+00 -1.26882359e-01
4.51290831e-02 -6.77437305e-01 -4.32412475e-01 -1.17893648e+00
-5.03480315e-01 7.44932950e-01 3.20906416e-02 -1.14005640e-01
9.63320434e-01 -4.52508211e-01 9.14448202e-01 4.90140617e-01
1.95339620e-01 -9.80519116e-01 -9.60047916e-02 1.05368149e+00
1.01363575e+00 -9.61345434e-01 -2.73622006e-01 -5.39767705e-02
-8.81163538e-01 1.02534044e+00 5.55064499e-01 -5.27628720e-01
1.99458271e-01 5.59490085e-01 3.70538205e-01 4.85788733e-01
-7.32207358e-01 1.67939574e-01 1.05879135e-01 7.61418760e-01
8.01991582e-01 1.39362082e-01 -1.17663316e-01 7.31043875e-01
-6.96159720e-01 -1.66231140e-01 9.40371573e-01 9.28997159e-01
-4.16358083e-01 -1.56271899e+00 -6.45701826e-01 4.13528830e-01
-7.96519816e-01 -8.79095614e-01 -6.55966938e-01 3.85227889e-01
-1.35329783e-01 1.00241697e+00 -3.24396104e-01 -6.29080176e-01
2.56840736e-01 1.82888031e-01 2.48640582e-01 -9.60694373e-01
-1.04296398e+00 -3.61600161e-01 3.94356936e-01 -5.63822210e-01
3.14169854e-01 -7.41123796e-01 -9.35295105e-01 -6.31734133e-01
-5.69707081e-02 2.05308154e-01 3.46648008e-01 7.61170268e-01
3.28572184e-01 4.00241762e-01 3.25509787e-01 -9.36969399e-01
-7.58676469e-01 -1.01626015e+00 -1.51499361e-01 4.53539819e-01
5.23446441e-01 -1.00007780e-01 -5.24134040e-01 1.41098440e-01] | [11.030938148498535, 10.367570877075195] |
a5d40e63-f1a7-4ddd-803b-77c326a11da7 | complex-reading-comprehension-through | 2211.03277 | null | https://arxiv.org/abs/2211.03277v1 | https://arxiv.org/pdf/2211.03277v1.pdf | Complex Reading Comprehension Through Question Decomposition | Multi-hop reading comprehension requires not only the ability to reason over raw text but also the ability to combine multiple evidence. We propose a novel learning approach that helps language models better understand difficult multi-hop questions and perform "complex, compositional" reasoning. Our model first learns to decompose each multi-hop question into several sub-questions by a trainable question decomposer. Instead of answering these sub-questions, we directly concatenate them with the original question and context, and leverage a reading comprehension model to predict the answer in a sequence-to-sequence manner. By using the same language model for these two components, our best seperate/unified t5-base variants outperform the baseline by 7.2/6.1 absolute F1 points on a hard subset of DROP dataset. | ['Gholamreza Haffari', 'Yuan-Fang Li', 'Xiao-Yu Guo'] | 2022-11-07 | null | null | null | null | ['multi-hop-reading-comprehension'] | ['natural-language-processing'] | [ 4.96408552e-01 4.95212555e-01 1.52302459e-01 -5.01154244e-01
-1.71356642e+00 -1.15248275e+00 2.92281568e-01 4.56970394e-01
-4.02649015e-01 6.69349849e-01 4.28141057e-01 -1.00397873e+00
5.92344180e-02 -8.75910401e-01 -1.16703546e+00 1.55414253e-01
4.54846859e-01 5.92687011e-01 5.02390742e-01 -5.26625335e-01
2.07265422e-01 -4.22024816e-01 -1.21951079e+00 8.42914701e-01
1.38366222e+00 6.74788356e-01 2.65415728e-01 1.28699362e+00
-5.39073348e-01 1.39191687e+00 -4.27340329e-01 -7.30133116e-01
-1.66297421e-01 -7.36931205e-01 -1.42496192e+00 -2.18411833e-01
8.83707821e-01 -7.14683831e-01 -1.85578182e-01 7.04886854e-01
9.70971286e-02 1.97831783e-02 5.02653241e-01 -5.19182861e-01
-7.70332932e-01 9.45363045e-01 -3.07948440e-01 2.46064618e-01
1.06505239e+00 2.18803525e-01 1.60606956e+00 -6.81145966e-01
2.88544208e-01 1.23773360e+00 5.12921333e-01 6.01781547e-01
-1.33446920e+00 -1.70335248e-01 5.59970021e-01 4.19212520e-01
-6.24596775e-01 -2.94228703e-01 2.50091702e-01 -1.69044808e-01
1.34962201e+00 4.26003784e-01 2.54331768e-01 1.05577719e+00
2.78264731e-01 9.55367565e-01 1.07895684e+00 -5.29878438e-01
4.36525792e-02 -3.73605788e-01 8.73118043e-01 1.08113742e+00
9.68086720e-03 -6.92687154e-01 -4.36985493e-01 4.21940908e-02
-6.35529384e-02 -2.56320298e-01 -2.88028300e-01 3.87515396e-01
-8.96693349e-01 7.94645131e-01 1.26266569e-01 3.79997480e-04
-2.19457462e-01 2.17996687e-01 1.57500088e-01 6.78036749e-01
1.20584741e-01 7.62623131e-01 -1.06708884e+00 -2.97172695e-01
-8.05493474e-01 4.86724555e-01 1.37052596e+00 6.09899521e-01
4.99139398e-01 -5.61866105e-01 -3.90216082e-01 6.63111389e-01
2.78323263e-01 5.78290820e-01 2.86974669e-01 -1.23019731e+00
1.01350808e+00 6.20766938e-01 -2.70125866e-02 -5.86712062e-01
-3.06677073e-01 -4.40214723e-01 -3.41133773e-01 -3.43798488e-01
8.60342503e-01 -2.76025623e-01 -8.60224545e-01 2.00596547e+00
1.45479843e-01 -2.42175221e-01 1.43896788e-01 4.62076098e-01
9.46784973e-01 9.73874092e-01 1.16708301e-01 1.13374688e-01
1.80964494e+00 -1.38775325e+00 -4.34568971e-01 -6.21649623e-01
6.69841945e-01 -6.69110417e-01 1.52641761e+00 5.02758920e-01
-1.65479851e+00 -4.73608315e-01 -1.01261270e+00 -9.46162581e-01
-4.95777540e-02 1.16257751e-02 -2.76061986e-02 7.96941519e-02
-1.17072797e+00 1.03355087e-01 -4.94398922e-01 -6.46754727e-02
1.27066344e-01 -1.28509803e-02 -5.28144836e-02 -6.17181182e-01
-1.25401878e+00 9.68822479e-01 7.61757046e-02 -3.15717220e-01
-1.00140822e+00 -9.84079361e-01 -7.98630416e-01 4.51798052e-01
6.66791439e-01 -1.16908050e+00 1.93562317e+00 -5.11941671e-01
-1.48322546e+00 7.23143518e-01 -7.94618309e-01 -4.50115263e-01
4.09460962e-01 -8.11498046e-01 6.05638511e-02 4.23994690e-01
2.64377773e-01 5.50072253e-01 6.48773551e-01 -9.56875622e-01
-5.17153144e-01 -1.87337443e-01 7.53859401e-01 1.23845361e-01
2.13525325e-01 -1.87730759e-01 -3.88932347e-01 -4.23483372e-01
2.65556782e-01 -6.91768527e-01 9.37810838e-02 -5.73351681e-02
-4.78982270e-01 -4.21847105e-01 3.07581902e-01 -1.37519813e+00
1.23717380e+00 -1.60986364e+00 4.39181030e-01 -2.29377434e-01
6.20308280e-01 7.70699084e-02 -7.07732022e-01 5.50137758e-01
1.45019859e-01 1.09552965e-01 -3.75316322e-01 -4.32368457e-01
2.83773124e-01 1.91117883e-01 -6.25146508e-01 -3.76622111e-01
4.66216385e-01 1.38293505e+00 -9.65253651e-01 -3.08937401e-01
-2.71567643e-01 -1.17363051e-01 -9.71231341e-01 6.08471274e-01
-1.25689209e+00 1.97426498e-01 -3.43682677e-01 3.55118006e-01
5.53845406e-01 -6.03548825e-01 1.89839140e-01 2.72137940e-01
3.45420897e-01 1.20662344e+00 -5.99393487e-01 1.87177777e+00
-6.78503692e-01 5.46288967e-01 1.29558474e-01 -7.77001619e-01
5.18726587e-01 1.52610764e-01 -2.71001995e-01 -8.65171790e-01
-2.25884393e-01 2.36732751e-01 1.97869632e-02 -9.55778003e-01
2.28693843e-01 -2.16981366e-01 -2.78691947e-01 8.68388176e-01
1.72821194e-01 -2.40742296e-01 3.29997122e-01 6.49309874e-01
1.52590537e+00 1.66510493e-01 1.96805485e-02 1.11945651e-01
8.52153003e-01 -8.52564350e-02 -3.36653963e-02 1.01059413e+00
2.76531398e-01 5.53125381e-01 8.83168101e-01 -1.74697489e-01
-7.52852917e-01 -1.39667630e+00 5.64689755e-01 1.63464391e+00
-2.12466389e-01 -6.17004752e-01 -1.01093841e+00 -8.29918683e-01
4.10632323e-03 1.24605060e+00 -4.27599251e-01 -1.54760987e-01
-8.58647168e-01 -1.13186739e-01 8.75213742e-01 6.31194293e-01
5.01794815e-01 -6.55913174e-01 -4.19530779e-01 4.40742403e-01
-8.35328221e-01 -1.19313157e+00 -4.78653103e-01 2.62994826e-01
-6.95631146e-01 -1.16988802e+00 -3.47723007e-01 -8.77032101e-01
4.03417587e-01 2.15330407e-01 1.69156098e+00 6.15896821e-01
1.68797582e-01 3.96627873e-01 -5.19622326e-01 -2.71293223e-01
-5.61558962e-01 5.73531508e-01 -6.69385135e-01 -3.88015658e-01
3.60558093e-01 -4.69627947e-01 -3.42974514e-01 -1.90937370e-01
-8.24179471e-01 2.83064067e-01 6.13907516e-01 6.74216747e-01
3.28967661e-01 -5.28271377e-01 6.20526195e-01 -8.89587760e-01
1.06180823e+00 -6.63531899e-01 -4.49599981e-01 8.52217853e-01
-2.14351296e-01 6.56185269e-01 9.15395379e-01 -2.40575045e-01
-1.06709218e+00 -3.80635560e-01 -5.72159350e-01 3.81549925e-01
-1.67923003e-01 6.88973784e-01 -2.00523049e-01 4.68166441e-01
6.04147315e-01 5.04299641e-01 -8.06258097e-02 -5.68179846e-01
8.38352799e-01 3.11097801e-01 7.75972128e-01 -1.08660913e+00
9.68437850e-01 -2.06088737e-01 -3.29324722e-01 -3.97510469e-01
-1.63501954e+00 -1.55342788e-01 -3.38792235e-01 3.38261336e-01
1.19580078e+00 -8.89900684e-01 -8.90085220e-01 2.90818781e-01
-1.64205623e+00 -6.81504965e-01 -2.36319862e-02 6.13387069e-03
-4.18612957e-01 4.07251656e-01 -9.85133290e-01 -4.65486050e-01
-5.24851680e-01 -9.20730233e-01 9.41016555e-01 1.16467431e-01
-5.77491224e-01 -9.26869214e-01 9.47876945e-02 1.24238503e+00
4.13017184e-01 -3.28538477e-01 1.75756764e+00 -9.09143269e-01
-9.23836529e-01 1.91194251e-01 -3.31380188e-01 6.09538436e-01
-1.55700684e-01 -4.35718387e-01 -8.02726448e-01 6.94748759e-02
3.35440457e-01 -9.67446327e-01 1.03324115e+00 -1.20323017e-01
1.21720278e+00 -6.00582302e-01 2.76770353e-01 3.53114188e-01
1.12355328e+00 -3.47031742e-01 4.76189047e-01 6.84892088e-02
5.52112937e-01 7.24034429e-01 3.31635587e-02 -1.47611082e-01
1.27974796e+00 -3.38496342e-02 2.64379442e-01 3.69518310e-01
-2.30931565e-01 -7.41294682e-01 4.83482659e-01 1.22086740e+00
4.92550462e-01 -3.10092300e-01 -1.02867925e+00 4.14567530e-01
-1.54987669e+00 -8.33287358e-01 -2.84595549e-01 1.55294907e+00
1.38237309e+00 3.03812176e-01 -1.76023349e-01 2.90578301e-03
-2.54092924e-02 1.29743055e-01 -7.33331144e-01 -6.19003117e-01
-8.68888348e-02 6.99678183e-01 -1.58747911e-01 1.28382742e+00
-5.78257859e-01 1.00357568e+00 6.45874262e+00 5.67557514e-01
-6.96842492e-01 -8.87654442e-03 5.40018082e-01 2.10838336e-02
-9.22955811e-01 2.25744680e-01 -6.60105109e-01 1.86578795e-01
1.16124845e+00 5.69436923e-02 7.35349357e-01 1.33223340e-01
-1.15953505e-01 -3.84849072e-01 -1.33202100e+00 3.89385968e-01
2.40202621e-01 -1.18225813e+00 4.03827012e-01 -5.55891991e-01
4.69961733e-01 -7.92472586e-02 -6.33961484e-02 6.29325271e-01
5.06003320e-01 -1.46962881e+00 7.34889865e-01 6.92667365e-01
4.74960238e-01 -1.76616102e-01 4.96463150e-01 8.89530778e-01
-9.99285996e-01 -2.53739029e-01 -1.70507319e-02 -6.08574986e-01
2.83498764e-01 2.63272792e-01 -6.19370103e-01 5.03694057e-01
2.64069229e-01 2.80185789e-01 -9.21339631e-01 5.26565850e-01
-9.74193573e-01 1.08819139e+00 -2.51345634e-01 -3.21800768e-01
1.96246460e-01 -1.03221595e-01 2.01581076e-01 1.05448365e+00
7.76810423e-02 6.93402588e-01 -2.34804507e-02 8.96750271e-01
-4.74503994e-01 -1.30138755e-01 1.65180117e-01 -1.08887114e-01
3.42943162e-01 8.40835929e-01 -1.89628301e-03 -6.17492378e-01
-8.31201553e-01 1.06379914e+00 8.32349122e-01 4.30343449e-01
-6.91144764e-01 -4.67727572e-01 4.52380270e-01 1.13560222e-01
3.46360862e-01 -4.00419325e-01 -4.70788389e-01 -1.55892634e+00
4.44216162e-01 -1.63569450e+00 5.91372371e-01 -1.16942430e+00
-1.34994817e+00 4.10704523e-01 -1.45395249e-01 -3.12260807e-01
-3.99551392e-01 -7.25692868e-01 -6.58710301e-01 1.09921718e+00
-1.73264349e+00 -1.13424468e+00 -2.61817992e-01 4.69040483e-01
8.66516888e-01 3.21818650e-01 8.03951383e-01 -9.25222337e-02
-2.54107833e-01 6.80288017e-01 -2.44183779e-01 1.52622730e-01
6.03970289e-01 -1.60208356e+00 7.05034614e-01 7.81710565e-01
1.08177997e-01 8.52706730e-01 6.26402438e-01 -3.55628520e-01
-1.62293077e+00 -7.66289294e-01 1.45603490e+00 -1.03449488e+00
8.59995663e-01 -4.86644655e-01 -1.35369718e+00 8.88617337e-01
6.75467968e-01 -6.97904110e-01 7.69147336e-01 2.18080133e-01
-1.05973458e+00 -5.82200103e-02 -8.04793477e-01 6.99509740e-01
9.00415778e-01 -1.07485676e+00 -1.66128349e+00 2.03208372e-01
1.41416240e+00 -4.48148161e-01 -6.62481487e-01 1.06802717e-01
4.92480278e-01 -7.66315579e-01 8.22302043e-01 -1.00726533e+00
9.90050316e-01 -2.11783811e-01 -3.50629717e-01 -1.01822495e+00
-8.02564397e-02 -5.77799439e-01 -3.73558193e-01 8.02030563e-01
9.22525108e-01 -6.79833531e-01 5.11216640e-01 5.72145283e-01
-8.76508206e-02 -1.05110204e+00 -6.61841571e-01 -2.27404520e-01
7.23732173e-01 -5.91319382e-01 6.60145581e-01 3.70429397e-01
1.10785224e-01 9.81753945e-01 1.09697483e-01 1.54148981e-01
4.26816076e-01 2.68759251e-01 6.56536937e-01 -9.67090249e-01
-8.31567764e-01 -3.15890878e-01 5.62023222e-01 -1.95060575e+00
3.14510465e-01 -1.01684642e+00 1.76594943e-01 -1.98517156e+00
2.88417757e-01 1.19695164e-01 -6.05335552e-03 3.35219085e-01
-7.14296460e-01 -3.31003726e-01 2.36350507e-01 -2.47364670e-01
-9.89758790e-01 3.10673267e-01 1.23217142e+00 -2.38811225e-01
2.60722399e-01 -1.51891395e-01 -1.17752814e+00 5.79471052e-01
7.03872800e-01 -2.39450768e-01 -6.74729705e-01 -1.26619363e+00
7.44017482e-01 4.98319864e-01 3.53615224e-01 -7.81455398e-01
4.37921226e-01 -9.12370384e-02 9.88919362e-02 -4.97159094e-01
1.46239519e-01 -2.21170753e-01 -6.62737250e-01 5.50671041e-01
-1.05314291e+00 4.28951442e-01 2.51740128e-01 4.19785976e-01
-1.84925660e-01 -3.59433889e-01 3.20537388e-01 -3.59365284e-01
-2.29445741e-01 -9.44700390e-02 -4.25669253e-01 8.15365016e-01
5.40550113e-01 3.48294318e-01 -7.22137630e-01 -8.10388565e-01
-6.42546058e-01 7.70013511e-01 1.96874529e-01 3.69389355e-01
5.29117405e-01 -5.70930123e-01 -1.04416883e+00 -4.07003947e-02
-2.84417626e-02 2.49366239e-01 1.69191167e-01 5.79066157e-01
-6.76967978e-01 6.36252284e-01 2.51296967e-01 -1.79622263e-01
-1.00521457e+00 4.17672157e-01 3.85907352e-01 -6.92139030e-01
-2.27344036e-01 1.15673232e+00 1.04235813e-01 -8.01540375e-01
2.50724196e-01 -9.10664201e-01 5.77015467e-02 -3.37210437e-03
7.01621413e-01 1.52432397e-01 6.00391626e-03 4.98002544e-02
-2.18172953e-01 7.24300027e-01 -2.71476924e-01 -3.98676932e-01
1.09270477e+00 -3.14095557e-01 -4.29083735e-01 2.92402685e-01
1.23912299e+00 1.75173566e-01 -1.03928339e+00 -3.76304001e-01
3.01312119e-01 1.35115415e-01 -5.91751575e-01 -1.44082272e+00
-2.09214807e-01 1.20387411e+00 -3.87011200e-01 1.61428928e-01
1.09338605e+00 4.07162577e-01 1.32148910e+00 9.22155499e-01
-1.99283995e-02 -5.64766288e-01 6.06754959e-01 1.13600612e+00
1.01766419e+00 -1.08036590e+00 -3.48596394e-01 -3.43721569e-01
-3.30140889e-01 1.11739051e+00 7.69944727e-01 -2.24963576e-03
1.06242582e-01 2.10312575e-01 2.24815369e-01 -4.28971909e-02
-1.51791525e+00 -5.94298914e-02 4.99077499e-01 7.97212571e-02
6.33419335e-01 -1.28480688e-01 -1.65236592e-01 1.01743054e+00
-5.77054918e-01 -7.42631704e-02 4.10218179e-01 9.08440769e-01
-8.08890939e-01 -1.23164248e+00 -2.15352982e-01 5.61151683e-01
-4.77715492e-01 -5.48186481e-01 -4.51620728e-01 3.05558145e-01
-1.92822948e-01 1.36153436e+00 -1.93129465e-01 -3.01900476e-01
4.64229822e-01 6.12959504e-01 5.92019379e-01 -7.13291228e-01
-6.94588780e-01 -5.40949225e-01 3.90526831e-01 -5.72720647e-01
3.14847648e-01 -3.65845025e-01 -1.37934899e+00 -4.04168904e-01
1.20505668e-01 1.46492124e-01 2.38323495e-01 1.49860168e+00
3.89553696e-01 5.34206986e-01 1.11738920e-01 1.33023337e-01
-1.22910619e+00 -9.45722222e-01 2.54509032e-01 3.01340103e-01
7.98026621e-01 1.86073333e-01 -3.92488778e-01 9.83364806e-02] | [11.140464782714844, 7.9442925453186035] |
5d35a58c-d18d-4cde-8077-7e8b99131454 | boundary-aware-instance-segmentation | 1612.03129 | null | http://arxiv.org/abs/1612.03129v2 | http://arxiv.org/pdf/1612.03129v2.pdf | Boundary-aware Instance Segmentation | We address the problem of instance-level semantic segmentation, which aims at
jointly detecting, segmenting and classifying every individual object in an
image. In this context, existing methods typically propose candidate objects,
usually as bounding boxes, and directly predict a binary mask within each such
proposal. As a consequence, they cannot recover from errors in the object
candidate generation process, such as too small or shifted boxes.
In this paper, we introduce a novel object segment representation based on
the distance transform of the object masks. We then design an object mask
network (OMN) with a new residual-deconvolution architecture that infers such a
representation and decodes it into the final binary object mask. This allows us
to predict masks that go beyond the scope of the bounding boxes and are thus
robust to inaccurate object candidates. We integrate our OMN into a Multitask
Network Cascade framework, and learn the resulting boundary-aware instance
segmentation (BAIS) network in an end-to-end manner. Our experiments on the
PASCAL VOC 2012 and the Cityscapes datasets demonstrate the benefits of our
approach, which outperforms the state-of-the-art in both object proposal
generation and instance segmentation. | ['Zeeshan Hayder', 'Mathieu Salzmann', 'Xuming He'] | 2016-12-09 | boundary-aware-instance-segmentation-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Hayder_Boundary-Aware_Instance_Segmentation_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Hayder_Boundary-Aware_Instance_Segmentation_CVPR_2017_paper.pdf | cvpr-2017-7 | ['object-proposal-generation'] | ['computer-vision'] | [ 5.11478305e-01 4.25421298e-01 -8.54879618e-02 -7.35847354e-01
-9.45599258e-01 -4.99414384e-01 4.46029156e-01 -1.83290225e-02
-5.57181001e-01 3.68547350e-01 -2.28973970e-01 -1.80889536e-02
1.09996669e-01 -6.50292516e-01 -1.23033106e+00 -5.04448354e-01
3.54137450e-01 7.42623448e-01 7.32718349e-01 2.34648362e-01
1.85647577e-01 2.07756042e-01 -1.44116127e+00 6.01583779e-01
1.12079477e+00 1.32104385e+00 5.03409147e-01 3.79917800e-01
-1.07176811e-01 3.63874197e-01 -6.24618053e-01 -4.08474535e-01
3.92755687e-01 -1.07934199e-01 -8.95755589e-01 5.52594125e-01
7.87439585e-01 -3.56110781e-01 -2.42623431e-03 1.04036224e+00
7.47235715e-02 4.36992086e-02 6.24851882e-01 -9.09682631e-01
-6.36393428e-01 7.41013885e-01 -6.61141992e-01 -6.48248047e-02
-4.35773313e-01 1.90499827e-01 1.04942727e+00 -1.13453686e+00
3.95546556e-01 1.30183530e+00 5.27654052e-01 5.44204712e-01
-1.49502075e+00 -5.22555590e-01 7.57847905e-01 -5.10219410e-02
-1.55570149e+00 -4.62761045e-01 6.37485325e-01 -4.63756740e-01
5.75964510e-01 9.44966599e-02 4.79909748e-01 8.47513914e-01
-2.67706037e-01 1.23627543e+00 7.84970164e-01 -1.16805069e-01
2.13517502e-01 1.38824314e-01 5.09386510e-02 6.99299872e-01
2.82854974e-01 -1.78448394e-01 -1.74652800e-01 2.81578779e-01
8.67170811e-01 2.26342216e-01 -2.58113235e-01 -3.86844128e-01
-1.28241050e+00 6.84853375e-01 9.06936884e-01 1.13834739e-01
-3.58768672e-01 3.45790833e-01 -1.76305193e-02 -3.78227681e-01
7.48699069e-01 1.90067887e-01 -6.60490930e-01 7.15196431e-01
-1.27982378e+00 3.77537519e-01 3.47266078e-01 8.95489395e-01
9.53779399e-01 -2.52729326e-01 -5.53002775e-01 8.54623675e-01
5.16044497e-01 1.03101924e-01 2.72152752e-01 -8.76661003e-01
5.82176149e-01 7.47901022e-01 2.88933784e-01 -4.97484773e-01
-3.45025063e-01 -7.24381745e-01 -4.52910215e-01 1.68128803e-01
4.60616767e-01 4.19039503e-02 -1.44808197e+00 1.56533897e+00
7.13393688e-01 5.69168150e-01 -1.38348877e-01 1.21165717e+00
8.78601193e-01 6.13372684e-01 1.52968019e-01 2.63293147e-01
1.49994743e+00 -1.36687505e+00 -3.27968776e-01 -7.79367805e-01
3.29174191e-01 -5.33059359e-01 8.72840524e-01 2.05828890e-01
-1.11938632e+00 -7.51897275e-01 -7.99719214e-01 -1.60472766e-01
-1.66350231e-01 5.12092829e-01 4.37192887e-01 2.46141404e-01
-9.33210969e-01 6.23841643e-01 -1.03103173e+00 2.01485425e-01
9.89717841e-01 3.38963479e-01 5.77987544e-02 6.62238151e-02
-6.56880200e-01 5.85972965e-01 7.43779600e-01 4.60801095e-01
-9.10584152e-01 -8.31717193e-01 -9.29908454e-01 1.97599649e-01
8.21221411e-01 -6.46457970e-01 1.33924484e+00 -1.29713809e+00
-1.06138694e+00 9.61322427e-01 -2.73998439e-01 -7.02479661e-01
5.86890757e-01 -1.56503528e-01 1.37592390e-01 8.08912218e-02
4.07291114e-01 1.30194759e+00 1.10174036e+00 -1.42656398e+00
-9.83071685e-01 -3.66943955e-01 -9.17749330e-02 1.94734745e-02
1.50584370e-01 -8.53325576e-02 -7.17547417e-01 -7.12921083e-01
3.92997682e-01 -7.47483790e-01 -5.22891402e-01 1.91148698e-01
-7.86735892e-01 -4.30071235e-01 6.82551682e-01 -6.56729877e-01
8.17861497e-01 -2.05610418e+00 9.08961669e-02 1.22595169e-01
2.40809321e-01 3.01964998e-01 -1.48397952e-01 -4.23716933e-01
1.68105245e-01 2.54068404e-01 -8.33951533e-01 -7.74411321e-01
-3.66812944e-02 6.66263029e-02 -3.43730181e-01 3.46708864e-01
6.95612490e-01 1.20035362e+00 -6.14469767e-01 -4.96320426e-01
1.36786580e-01 4.89542782e-01 -6.39319241e-01 2.39812240e-01
-7.34196723e-01 5.93744516e-01 -5.86716831e-01 6.74361706e-01
7.80882418e-01 -4.17005241e-01 -1.17565192e-01 -2.34452426e-01
-1.21110126e-01 3.14653397e-01 -1.21778643e+00 1.65079522e+00
-2.81652898e-01 3.86557281e-01 1.91592112e-01 -1.15720069e+00
8.44003260e-01 1.90200713e-02 3.06258380e-01 -4.81233180e-01
7.11171282e-03 4.63395983e-01 2.30296887e-02 -3.56087029e-01
3.57528895e-01 -1.65766925e-02 -1.61564853e-02 8.93402770e-02
1.28584355e-01 -1.30401328e-01 1.72190964e-01 2.69177854e-02
6.24806881e-01 4.59846377e-01 -6.76353946e-02 -1.61749944e-01
5.13971269e-01 -1.14250630e-01 6.92546844e-01 7.13151157e-01
-7.90467113e-02 1.07324803e+00 4.23330545e-01 -3.75246584e-01
-8.43748569e-01 -9.98794138e-01 -3.87378782e-01 9.58124816e-01
3.52495223e-01 5.64729795e-02 -1.01911926e+00 -1.03499615e+00
1.39198408e-01 6.11313522e-01 -5.99219799e-01 2.28086971e-02
-7.87815809e-01 -6.84226155e-01 2.40383893e-01 6.35500491e-01
5.74532866e-01 -1.19638598e+00 -6.26307607e-01 4.23856676e-01
-2.83120155e-01 -1.39683342e+00 -7.15691090e-01 1.27223596e-01
-8.49932492e-01 -1.05136156e+00 -7.07169712e-01 -9.43197966e-01
8.15288424e-01 1.65882647e-01 9.36562896e-01 1.75738916e-01
-2.79970855e-01 -1.11235954e-01 -1.30470589e-01 -3.77641648e-01
-2.42891178e-01 1.03105068e-01 -4.65656549e-01 5.90491772e-01
4.42428179e-02 -1.00306608e-01 -8.99626970e-01 3.66058022e-01
-1.04096949e+00 2.40113020e-01 7.99087763e-01 7.31947482e-01
1.07942843e+00 -1.75227523e-01 7.43653119e-01 -1.01962876e+00
-8.50599557e-02 -4.07484412e-01 -8.95219624e-01 2.80974627e-01
-3.56083959e-01 3.80854756e-02 3.78621072e-01 -3.29354912e-01
-1.12253237e+00 4.03501481e-01 -1.69770360e-01 -4.92806107e-01
-3.81785601e-01 3.07586581e-01 -2.97324568e-01 7.99553990e-02
3.67510229e-01 1.11660235e-01 -2.60265082e-01 -7.21609533e-01
4.89456952e-01 5.32533467e-01 7.06509411e-01 -3.81908804e-01
6.01905286e-01 7.18818367e-01 -3.49759936e-01 -4.18821543e-01
-1.40709782e+00 -5.91269791e-01 -7.51760662e-01 -5.59374206e-02
1.06794024e+00 -1.15114284e+00 -4.23843890e-01 3.64683896e-01
-1.41788340e+00 -5.91934919e-01 -3.64526808e-01 1.08716793e-01
-5.11102378e-01 8.08438845e-03 -3.95646214e-01 -6.91179693e-01
-2.79922515e-01 -1.33047259e+00 1.68344784e+00 3.16232502e-01
1.52121887e-01 -5.76161742e-01 -4.74404186e-01 6.32507026e-01
1.66311920e-01 2.30433896e-01 6.03840411e-01 -6.40361249e-01
-1.07927275e+00 -1.20558180e-01 -5.79275846e-01 3.74593258e-01
-2.20279306e-01 -1.75494432e-01 -1.12987745e+00 -7.39316866e-02
-1.37040779e-01 -1.73906624e-01 1.62622857e+00 6.54010296e-01
1.67922199e+00 -4.15396512e-01 -5.54317892e-01 8.02219331e-01
1.38403344e+00 -7.09688738e-02 3.69358510e-01 7.94384331e-02
9.56173122e-01 7.51235306e-01 5.71658671e-01 1.55441612e-01
5.01455545e-01 7.98445046e-01 7.57782400e-01 -2.92525858e-01
-3.29610169e-01 -1.58325389e-01 1.19689040e-01 -8.09293892e-03
2.35966414e-01 -3.41314524e-01 -7.30310678e-01 8.14025223e-01
-1.96600378e+00 -4.82037276e-01 -3.30230474e-01 2.01360559e+00
7.84938276e-01 2.77770549e-01 1.17765419e-01 -2.00512186e-01
9.73349869e-01 1.81815937e-01 -9.06510234e-01 1.41498908e-01
-3.99157591e-02 3.56535137e-01 5.52178442e-01 4.62447196e-01
-1.44191110e+00 1.39898109e+00 4.74064064e+00 6.77974224e-01
-1.02126849e+00 3.11336398e-01 1.08008552e+00 -5.66996913e-03
-5.81066124e-02 -4.15351652e-02 -1.05529845e+00 5.37557125e-01
5.46113431e-01 4.19086665e-01 3.02344441e-01 8.65715921e-01
1.59663260e-01 -2.89094504e-02 -1.12656021e+00 5.77768147e-01
-4.45430391e-02 -1.36371052e+00 -7.90224131e-03 -8.18743557e-02
9.82802808e-01 3.02119792e-01 1.42382354e-01 1.94334894e-01
3.02501265e-02 -1.00500131e+00 1.40782869e+00 3.75655830e-01
6.00065827e-01 -4.10481423e-01 5.70356369e-01 4.45772916e-01
-1.07731533e+00 -2.52195150e-01 -5.02068162e-01 3.35896224e-01
2.64713705e-01 7.43910253e-01 -8.31301451e-01 1.95012733e-01
6.28330588e-01 5.75825393e-01 -5.78399122e-01 1.39160776e+00
-3.90121311e-01 6.96991026e-01 -3.32789928e-01 3.78571510e-01
3.91975164e-01 -1.84321463e-01 3.27015668e-01 1.03005755e+00
1.42037958e-01 5.97997345e-02 4.61986780e-01 1.73713160e+00
-3.72705519e-01 -2.32107773e-01 6.12126850e-02 3.08019698e-01
3.32168162e-01 1.39628482e+00 -1.19221544e+00 -5.27745962e-01
-2.88713992e-01 1.01748884e+00 4.78899986e-01 5.00262141e-01
-9.46759403e-01 -1.28284097e-01 6.07926548e-01 1.27093047e-01
1.02091098e+00 1.44072324e-01 -6.32672131e-01 -1.12249994e+00
3.84601295e-01 -4.71727878e-01 1.10646605e-01 -4.93415445e-01
-1.12968695e+00 4.09880668e-01 -1.17473297e-01 -8.08559835e-01
2.24568605e-01 -6.22665882e-01 -6.95922136e-01 7.60020912e-01
-1.78569305e+00 -1.30808938e+00 -2.53900647e-01 1.66388139e-01
9.23374534e-01 4.86659199e-01 1.60162032e-01 3.10533643e-01
-7.48041749e-01 3.55623156e-01 -2.33790845e-01 2.75256991e-01
3.77484977e-01 -1.22953272e+00 7.73354948e-01 9.56019819e-01
3.72751981e-01 1.93310097e-01 2.70146459e-01 -6.40612006e-01
-7.25272477e-01 -1.73472798e+00 6.62934542e-01 -5.68251431e-01
2.59304643e-01 -6.35309100e-01 -1.09946060e+00 6.48635685e-01
-3.61272901e-01 5.96171260e-01 -5.90019859e-02 -2.96449423e-01
-2.20646858e-01 -1.17723979e-01 -1.06320512e+00 3.18085164e-01
1.16359270e+00 -1.26150429e-01 -3.65268260e-01 4.59576666e-01
1.04889452e+00 -4.81003523e-01 -4.30553406e-01 5.60181797e-01
2.45488524e-01 -9.08757925e-01 1.16338074e+00 -5.41884720e-01
4.36881095e-01 -5.49643397e-01 1.36763483e-01 -1.07121289e+00
-1.92122191e-01 -2.07915455e-01 -1.25798211e-01 1.13326097e+00
6.80081189e-01 -3.72987717e-01 9.45879817e-01 5.16234934e-01
-6.01836205e-01 -9.81640577e-01 -1.06790388e+00 -5.79796433e-01
-3.53600308e-02 -5.98299801e-01 7.75984466e-01 4.73576128e-01
-7.91215301e-01 2.23133624e-01 -4.95713241e-02 4.55305785e-01
6.70175612e-01 3.63621354e-01 4.56959784e-01 -1.24258256e+00
-3.50417972e-01 -6.84718490e-01 -1.54635146e-01 -1.39338982e+00
3.74741018e-01 -1.04053831e+00 6.69961810e-01 -1.75757575e+00
1.11708909e-01 -7.13433325e-01 -2.57357419e-01 5.73394299e-01
-6.52886033e-01 5.71759403e-01 1.82290256e-01 2.69854814e-01
-8.92970085e-01 5.05111158e-01 1.29947531e+00 -3.05927664e-01
-2.48737186e-01 3.31156611e-01 -7.57427156e-01 9.55083847e-01
6.21809483e-01 -6.74736142e-01 -2.11620647e-02 -6.43796504e-01
-1.95361182e-01 -1.93259403e-01 8.90050888e-01 -7.81188846e-01
-1.95323862e-02 5.61905243e-02 5.35878956e-01 -6.85487807e-01
3.16094130e-01 -7.66664803e-01 -3.90372992e-01 3.14062685e-01
-3.70647162e-01 -6.37969196e-01 -8.35415423e-02 7.38987267e-01
-7.31798355e-03 -2.13242009e-01 8.56663764e-01 -2.19722062e-01
-6.58805311e-01 5.07677794e-01 8.70275795e-02 1.53719515e-01
9.71823812e-01 -2.51956433e-01 -1.24417588e-01 3.07327539e-01
-7.33678341e-01 4.73673850e-01 2.83676296e-01 4.07100350e-01
6.41292870e-01 -1.04118240e+00 -8.69867325e-01 3.13627005e-01
9.23053548e-02 9.10079718e-01 1.71245977e-01 8.65964353e-01
-2.29668826e-01 2.71892637e-01 3.64259928e-01 -9.43757057e-01
-8.68826330e-01 4.64874357e-01 5.67073345e-01 -6.57662451e-02
-7.07568169e-01 1.28273857e+00 8.34933043e-01 -5.04865170e-01
2.87002981e-01 -7.39623606e-01 -7.52313510e-02 -1.01304479e-01
3.12151670e-01 -6.92594647e-02 7.58911297e-02 -7.00666964e-01
-2.76941597e-01 5.81761777e-01 -5.63852601e-02 -1.87170226e-02
1.23682642e+00 -4.25628833e-02 -2.93050986e-02 2.28414953e-01
9.98157382e-01 -4.17159826e-01 -1.90774190e+00 -3.46471012e-01
2.92670816e-01 -4.74677891e-01 1.00426838e-01 -7.70898581e-01
-1.28707623e+00 7.83138692e-01 4.33068901e-01 9.48061328e-03
8.51913691e-01 4.81000215e-01 9.75015223e-01 1.37424275e-01
1.05343938e-01 -9.90184665e-01 7.67980963e-02 2.65205055e-01
7.93658316e-01 -1.52089381e+00 -3.29646587e-01 -6.53669953e-01
-5.51129162e-01 8.59528005e-01 7.40981042e-01 -3.13128978e-01
4.10555393e-01 -1.11755222e-01 -1.39543444e-01 -2.21289471e-01
-5.64651191e-01 -5.33549428e-01 7.28892088e-01 3.42048943e-01
2.46464923e-01 2.25920945e-01 -2.80958898e-02 8.35363626e-01
1.00171529e-01 -2.50398070e-01 3.25139284e-01 5.04722893e-01
-7.15498269e-01 -7.92468905e-01 -5.10198772e-01 5.84136963e-01
-5.41195095e-01 -1.57222763e-01 -3.15473616e-01 4.58146721e-01
6.96850538e-01 7.78203964e-01 4.54397589e-01 8.33198354e-02
3.49983305e-01 4.89472337e-02 2.52493441e-01 -9.21173096e-01
-5.82631230e-01 3.26105386e-01 -2.24529281e-01 -6.25029564e-01
-4.29000378e-01 -6.72983110e-01 -1.53460073e+00 5.22171259e-01
-6.60248220e-01 -3.79500017e-02 7.65139222e-01 1.13216555e+00
3.23033959e-01 7.32131660e-01 3.82685930e-01 -1.26276481e+00
-4.69457150e-01 -9.63793814e-01 -2.78443992e-01 3.29048187e-01
5.34429729e-01 -5.00594914e-01 -4.00248654e-02 2.09733129e-01] | [9.484508514404297, 0.46579593420028687] |
34fa43ef-b593-4bca-9e0d-5a768244bded | scale-recurrent-network-for-deep-image | 1802.01770 | null | http://arxiv.org/abs/1802.01770v1 | http://arxiv.org/pdf/1802.01770v1.pdf | Scale-recurrent Network for Deep Image Deblurring | In single image deblurring, the "coarse-to-fine" scheme, i.e. gradually
restoring the sharp image on different resolutions in a pyramid, is very
successful in both traditional optimization-based methods and recent
neural-network-based approaches. In this paper, we investigate this strategy
and propose a Scale-recurrent Network (SRN-DeblurNet) for this deblurring task.
Compared with the many recent learning-based approaches in [25], it has a
simpler network structure, a smaller number of parameters and is easier to
train. We evaluate our method on large-scale deblurring datasets with complex
motion. Results show that our method can produce better quality results than
state-of-the-arts, both quantitatively and qualitatively. | ['Xin Tao', 'Xiaoyong Shen', 'Jiaya Jia', 'Jue Wang', 'Yi Wang', 'Hongyun Gao'] | 2018-02-06 | scale-recurrent-network-for-deep-image-1 | http://openaccess.thecvf.com/content_cvpr_2018/html/Tao_Scale-Recurrent_Network_for_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Tao_Scale-Recurrent_Network_for_CVPR_2018_paper.pdf | cvpr-2018-6 | ['image-relighting'] | ['computer-vision'] | [ 1.61152050e-01 -6.43882394e-01 -1.37867659e-01 1.52226880e-01
-8.64169896e-01 -1.72065362e-01 3.92911851e-01 -5.21578550e-01
-2.28233427e-01 8.44277382e-01 7.34637082e-01 1.75073985e-02
-4.43179421e-02 -3.96734595e-01 -6.61445618e-01 -7.27158666e-01
1.38939187e-01 -7.49103650e-02 4.46092606e-01 -1.53368592e-01
3.95095915e-01 3.86324584e-01 -1.05847812e+00 1.30325392e-01
8.70187163e-01 7.12999701e-01 3.12750518e-01 7.35102057e-01
6.34833694e-01 1.09213901e+00 -3.41471970e-01 -1.02376819e-01
8.40123296e-02 -6.02158904e-01 -8.37571502e-01 7.68822432e-02
8.77854526e-01 -7.01581776e-01 -9.05674756e-01 1.29612136e+00
7.50713646e-01 1.60039827e-01 5.39799273e-01 -3.62778127e-01
-1.30066729e+00 3.59633088e-01 -9.43550110e-01 6.20263577e-01
1.49088308e-01 -1.85335487e-01 6.42848015e-01 -9.01248932e-01
5.52848518e-01 1.24848461e+00 9.61141407e-01 5.59354603e-01
-1.32853734e+00 -5.74187934e-01 -1.83770731e-01 5.67549825e-01
-1.33576322e+00 -5.93053341e-01 7.42537558e-01 -2.21236780e-01
7.86575556e-01 7.61010032e-03 3.57269228e-01 1.04788685e+00
5.62933564e-01 7.61305690e-01 1.31224775e+00 -4.20579731e-01
9.16498527e-02 -6.10195935e-01 7.20661804e-02 4.07102495e-01
2.33470276e-01 4.58553016e-01 -6.46229148e-01 -1.57036092e-02
1.36910415e+00 1.18449412e-01 -7.74885297e-01 -1.76327631e-01
-1.31643987e+00 5.73222339e-01 5.46393692e-01 4.58715141e-01
-5.95720589e-01 4.66314852e-01 2.56603479e-01 4.23231035e-01
1.04970610e+00 3.93148065e-01 -3.21015209e-01 -6.25430653e-03
-1.58499241e+00 1.95405066e-01 3.07099283e-01 3.12776953e-01
5.36642373e-01 3.84310246e-01 -1.95012972e-01 1.04341030e+00
5.24118170e-02 3.59281629e-01 7.91574478e-01 -1.28148818e+00
2.30301619e-01 -1.53422505e-01 3.78730178e-01 -1.15757620e+00
-2.51889169e-01 -4.44963753e-01 -1.65123713e+00 3.29554230e-01
1.03157520e-01 -1.78374082e-01 -1.06482267e+00 1.59575951e+00
-9.21050757e-02 7.27236390e-01 3.38164046e-02 1.25518465e+00
5.56384265e-01 8.92103314e-01 -3.39805365e-01 -4.00480270e-01
1.04707932e+00 -1.28195119e+00 -1.08042741e+00 -2.62391180e-01
5.37176915e-02 -8.46384048e-01 7.74733961e-01 5.82821369e-01
-1.25891519e+00 -6.34728670e-01 -1.00234389e+00 -1.06379434e-01
8.79389793e-03 2.99728960e-01 3.59446466e-01 4.10041422e-01
-1.71439707e+00 1.19095635e+00 -8.25472355e-01 -2.39995599e-01
3.40447366e-01 5.15473858e-02 -2.18617395e-01 -4.03167874e-01
-1.26797318e+00 1.08571744e+00 1.68253347e-01 1.20283678e-01
-1.03681946e+00 -5.83924174e-01 -7.54383206e-01 6.85838088e-02
2.17877463e-01 -8.21584821e-01 1.14921987e+00 -8.67637455e-01
-1.76938510e+00 5.55217803e-01 -3.10113788e-01 -5.39768875e-01
4.49790776e-01 -7.56126523e-01 -3.45022529e-01 1.37161434e-01
-1.85253546e-01 5.86842716e-01 1.45808363e+00 -1.18932879e+00
-3.86830717e-01 -2.97705829e-02 -3.26871991e-01 1.03507355e-01
-2.53463566e-01 3.09094816e-01 -4.13960397e-01 -1.33508801e+00
2.28838131e-01 -8.19380581e-01 -3.23264778e-01 -1.75877139e-01
-2.49359772e-01 2.61447370e-01 8.51882100e-01 -1.20547140e+00
1.45504892e+00 -1.91578901e+00 6.13107085e-01 -3.62166315e-01
4.78550613e-01 4.95440960e-01 -2.43408993e-01 2.84758896e-01
-3.78787994e-01 -4.21930552e-02 -2.33424947e-01 -5.19668102e-01
-3.82766157e-01 -1.40532807e-01 -3.42820495e-01 8.04739952e-01
-1.15855634e-01 1.00022602e+00 -7.26177394e-01 -1.05667986e-01
1.96127281e-01 6.84531212e-01 -5.01267254e-01 2.05990881e-01
1.46090746e-01 3.97310495e-01 -1.10432923e-01 6.11405849e-01
7.96069503e-01 -5.61457038e-01 -2.14908749e-01 -5.75329483e-01
-2.37070277e-01 -2.68290311e-01 -9.64961648e-01 1.75627744e+00
-4.97560591e-01 1.15108728e+00 1.77915335e-01 -8.14217210e-01
5.67296267e-01 5.10061383e-01 2.06412718e-01 -4.11578745e-01
6.72590733e-02 2.51655072e-01 -4.07129467e-01 -4.95078981e-01
9.54734683e-01 -4.54006270e-02 3.44404310e-01 4.52660382e-01
-2.47761771e-01 -1.06254995e-01 -6.20866530e-02 5.90718258e-03
1.12168562e+00 3.64354216e-02 2.05435038e-01 -4.41881567e-01
5.37257373e-01 -3.25452864e-01 4.06832784e-01 8.63019109e-01
-2.90075004e-01 1.21513999e+00 -7.41698965e-02 -5.19251645e-01
-1.33677673e+00 -7.83672750e-01 2.01827556e-01 6.76775575e-01
4.48727220e-01 -2.03670934e-01 -9.18863475e-01 -1.76141217e-01
-4.85220134e-01 3.99171710e-01 -5.94374835e-01 -7.89365545e-02
-8.83698821e-01 -8.48877728e-01 3.51899594e-01 4.30365890e-01
8.94687295e-01 -8.37646008e-01 -1.99460223e-01 2.18966395e-01
-6.20838940e-01 -1.02380335e+00 -9.98329937e-01 -3.20708036e-01
-1.23361659e+00 -6.39612734e-01 -1.44613612e+00 -9.34837461e-01
4.41227913e-01 6.67162836e-01 1.19546342e+00 -6.90150727e-03
3.87540311e-02 -4.90485132e-03 -3.60299349e-01 4.83961523e-01
-4.62337017e-01 -1.83139369e-03 6.57413974e-02 -2.31448971e-02
-2.50716150e-01 -9.66487706e-01 -9.10183072e-01 4.14270312e-01
-1.17919660e+00 8.64862576e-02 7.21860588e-01 1.13158679e+00
4.74171638e-01 2.16529727e-01 5.74221790e-01 -3.23906362e-01
9.96803761e-01 -1.56970873e-01 -3.85369956e-01 1.12293936e-01
-7.90129662e-01 -8.51447508e-03 6.11656368e-01 -8.18886757e-01
-1.01712859e+00 -3.64928365e-01 4.64361422e-02 -8.05561244e-01
-1.04166999e-01 4.94334757e-01 3.26450735e-01 -4.58544433e-01
7.40426421e-01 5.62321186e-01 -1.27889410e-01 -9.42144036e-01
3.86634886e-01 4.88908917e-01 9.00179863e-01 -2.41803005e-01
9.20496762e-01 4.91304487e-01 -2.30582431e-01 -7.11489856e-01
-6.85349584e-01 -2.64122903e-01 -5.06482780e-01 -2.04137281e-01
9.22065079e-01 -1.17091775e+00 -2.71263450e-01 1.14889503e+00
-1.40048194e+00 -4.90143955e-01 -1.15495618e-03 4.83567476e-01
-5.25888562e-01 8.12450886e-01 -1.33166575e+00 -2.94323236e-01
-4.99958515e-01 -1.16106737e+00 9.29457068e-01 5.12073517e-01
2.96886824e-02 -1.00363386e+00 3.42922926e-01 1.90616608e-01
1.17867243e+00 -1.65346302e-02 4.63322341e-01 7.22126290e-02
-5.82926154e-01 -2.38094404e-02 -7.08312154e-01 5.97975373e-01
4.27001566e-01 -4.22543913e-01 -5.52074432e-01 -7.18558192e-01
2.17454001e-01 -3.44096243e-01 1.20032096e+00 9.75815594e-01
1.05673969e+00 -4.37834829e-01 -1.35019377e-01 7.02921987e-01
1.52506900e+00 -2.69625396e-01 1.07403564e+00 5.27086258e-01
7.94476807e-01 1.17098473e-01 1.78524345e-01 1.49982542e-01
3.08468282e-01 8.40864539e-01 2.71865278e-01 -4.49079692e-01
-6.94704175e-01 1.36848181e-01 6.29784942e-01 8.91802013e-01
-2.92318761e-01 -3.10474545e-01 -5.63444495e-01 7.55832136e-01
-2.01257205e+00 -9.30609584e-01 8.24150369e-02 1.89013803e+00
1.16354465e+00 -2.48686403e-01 -2.32140332e-01 -2.35245705e-01
1.09876645e+00 7.00124562e-01 -4.41281468e-01 3.28283124e-02
-4.37709510e-01 6.42039925e-02 5.93050420e-01 8.05681050e-01
-1.18827975e+00 1.19597781e+00 7.18611288e+00 1.22461402e+00
-1.16716039e+00 3.63247275e-01 8.35062385e-01 6.75876718e-03
1.08968571e-01 -1.96221262e-01 -4.01946127e-01 3.57609004e-01
7.32163966e-01 7.20802024e-02 8.31928432e-01 3.58282983e-01
6.55756354e-01 -9.04425904e-02 -7.73582339e-01 1.06654859e+00
3.59363377e-01 -1.56771779e+00 -1.10127404e-01 -2.67556012e-01
1.17621958e+00 3.22704375e-01 2.09726840e-01 -1.54201657e-01
3.61038506e-01 -1.10624254e+00 7.02828944e-01 6.92493677e-01
9.19624209e-01 -4.91634279e-01 8.29385161e-01 1.55116767e-01
-9.84766603e-01 1.33939549e-01 -3.43278438e-01 4.45038565e-02
5.92483938e-01 8.71904373e-01 3.63110304e-01 6.48162007e-01
9.89000499e-01 1.12957168e+00 -5.13829827e-01 1.22685850e+00
-2.46096894e-01 8.20223927e-01 4.96251062e-02 5.19502521e-01
3.66435498e-02 -2.67912656e-01 7.98622012e-01 1.26410067e+00
5.92606723e-01 8.67875367e-02 -1.50040507e-01 8.23849082e-01
-1.67823061e-01 -3.97534907e-01 -1.48442954e-01 3.47916186e-01
1.87521726e-01 9.73729312e-01 -4.67162758e-01 -5.98573506e-01
-1.78104997e-01 1.55815339e+00 1.01324655e-01 6.75945222e-01
-6.75955415e-01 -2.54649729e-01 5.55144429e-01 -2.53819767e-02
7.18114734e-01 -4.52158362e-01 4.46976833e-02 -1.62891805e+00
-1.46359771e-01 -1.09325826e+00 -2.86691580e-02 -1.17103124e+00
-1.56088376e+00 8.29435945e-01 -7.89453611e-02 -1.38284016e+00
-1.23486593e-01 -3.40525538e-01 -4.89295781e-01 8.97603691e-01
-1.90413415e+00 -1.02491450e+00 -3.50377768e-01 4.81705606e-01
8.18276227e-01 3.01933400e-02 3.64409357e-01 4.80825245e-01
-6.59361005e-01 2.99662352e-01 5.76522171e-01 3.83540727e-02
1.05858302e+00 -8.53145003e-01 7.19649494e-01 1.32477593e+00
-2.04457372e-01 4.43314075e-01 9.67552006e-01 -8.24193358e-01
-1.18688130e+00 -1.07696128e+00 7.27357268e-01 -6.77695870e-02
7.52072871e-01 3.28318119e-01 -1.18366885e+00 4.36626434e-01
7.75919378e-01 1.75502032e-01 3.11502516e-02 -4.43936646e-01
-3.44973177e-01 3.90751027e-02 -9.31434929e-01 6.20920539e-01
7.99409151e-01 -5.13257980e-01 -6.24272525e-01 1.02129713e-01
7.25865722e-01 -6.22023761e-01 -8.86915028e-01 4.58128303e-01
4.17991877e-01 -1.25332010e+00 1.22800994e+00 -3.01284879e-01
9.05961692e-01 -3.59125704e-01 -1.75282452e-02 -1.74287915e+00
-9.16662991e-01 -8.84799659e-01 -4.50166613e-01 8.94411385e-01
7.55715817e-02 -7.37509847e-01 6.36940897e-01 -3.69390920e-02
-5.03762253e-02 -5.24666667e-01 -1.02325153e+00 -8.88766646e-01
2.27674723e-01 1.69538409e-01 2.45474488e-01 9.80689347e-01
-3.67210329e-01 1.66746318e-01 -1.17938888e+00 4.14934248e-01
8.62310767e-01 8.07939172e-02 3.02985579e-01 -7.68378854e-01
-4.20195997e-01 -5.62841773e-01 -4.92512397e-02 -1.58166885e+00
4.16931957e-02 -3.11688870e-01 1.76348850e-01 -1.50706542e+00
4.21152890e-01 -4.48076203e-02 -2.41553068e-01 2.55688250e-01
-4.27249908e-01 6.12844348e-01 4.06157151e-02 7.92358518e-01
-5.21469295e-01 7.38796413e-01 1.46640623e+00 -2.07395151e-01
-1.74994245e-01 -3.56429875e-01 -4.87141371e-01 6.62898421e-01
6.90117836e-01 -5.59835076e-01 1.30811661e-01 -7.26825774e-01
-7.13915378e-02 1.94873050e-01 4.35867131e-01 -1.02618539e+00
3.76077771e-01 4.33628336e-02 3.54082882e-01 -6.21108890e-01
2.74205267e-01 -5.09426713e-01 1.97725624e-01 4.57265943e-01
-2.88144767e-01 2.92356580e-01 1.86877653e-01 5.85529923e-01
-4.86999601e-01 -1.39275640e-02 1.07170916e+00 -3.22614536e-02
-5.91567039e-01 2.46289924e-01 -5.64958572e-01 -1.16234921e-01
2.72718728e-01 -1.53951079e-01 -5.63358605e-01 -7.97796845e-01
-4.99810129e-01 -2.80441284e-01 7.97636032e-01 2.87167132e-01
6.82193458e-01 -1.48275697e+00 -1.13550794e+00 -1.66335568e-01
-6.06459975e-01 -2.42270902e-01 4.28462178e-01 1.07993257e+00
-6.25922799e-01 1.52904317e-01 -1.59919590e-01 -4.56978321e-01
-1.19534898e+00 5.01542389e-01 5.03696561e-01 -6.17798388e-01
-7.03485489e-01 6.67289376e-01 2.18573928e-01 1.11040846e-01
-1.77095309e-02 -1.95258453e-01 -2.37414762e-01 -2.75721222e-01
8.10643792e-01 6.40833020e-01 -1.62717730e-01 -7.45954514e-01
1.36920940e-02 1.04451632e+00 -2.55053133e-01 -1.28869846e-01
1.51609969e+00 -5.48996270e-01 -4.58399922e-01 -5.25986552e-02
1.06608319e+00 4.85564396e-02 -1.55106330e+00 -5.61295390e-01
-2.49944597e-01 -6.51919603e-01 7.09059536e-01 -6.45194709e-01
-1.45718157e+00 5.97615838e-01 8.99097800e-01 9.90045890e-02
1.44543958e+00 -1.08336471e-01 1.05704141e+00 9.73805338e-02
8.86374712e-02 -8.28992069e-01 4.45410639e-01 6.25632167e-01
1.18053091e+00 -1.11361241e+00 4.20162499e-01 -1.95410639e-01
-2.67365903e-01 1.09960997e+00 2.99066126e-01 -4.23950821e-01
5.61849713e-01 1.85198024e-01 1.30264848e-01 -1.54235093e-02
-5.21014035e-01 1.27626255e-01 4.31972265e-01 3.73588830e-01
3.51583630e-01 -3.48770767e-01 -1.11048087e-01 2.31207177e-01
2.37899989e-01 4.23180282e-01 8.29057038e-01 6.19784951e-01
-5.91774285e-01 -9.25316989e-01 -7.34249115e-01 1.26586705e-01
-6.17734790e-01 -5.47360718e-01 1.31035019e-02 3.56952190e-01
-2.46861219e-01 9.23213959e-01 -2.52354771e-01 -3.63340199e-01
8.01757425e-02 -5.86900353e-01 5.62625825e-01 -7.19071254e-02
-3.30985457e-01 3.11987460e-01 -8.30516070e-02 -7.77426362e-01
-7.21474051e-01 -5.02776265e-01 -4.64681208e-01 -5.27227759e-01
-6.42872155e-01 -2.38725096e-01 1.33545086e-01 8.29265237e-01
5.99697411e-01 5.32062411e-01 6.21001363e-01 -1.70057142e+00
-5.72448432e-01 -1.17417383e+00 -5.52798092e-01 9.61833596e-02
9.10370052e-01 -3.04391116e-01 -5.77684522e-01 3.50230366e-01] | [11.510047912597656, -2.5393240451812744] |
776e4b9b-b438-43b8-b1a4-faee3366d96d | an-investigation-of-end-to-end-models-for | 2102.06237 | null | https://arxiv.org/abs/2102.06237v1 | https://arxiv.org/pdf/2102.06237v1.pdf | An Investigation of End-to-End Models for Robust Speech Recognition | End-to-end models for robust automatic speech recognition (ASR) have not been sufficiently well-explored in prior work. With end-to-end models, one could choose to preprocess the input speech using speech enhancement techniques and train the model using enhanced speech. Another alternative is to pass the noisy speech as input and modify the model architecture to adapt to noisy speech. A systematic comparison of these two approaches for end-to-end robust ASR has not been attempted before. We address this gap and present a detailed comparison of speech enhancement-based techniques and three different model-based adaptation techniques covering data augmentation, multi-task learning, and adversarial learning for robust ASR. While adversarial learning is the best-performing technique on certain noise types, it comes at the cost of degrading clean speech WER. On other relatively stationary noise types, a new speech enhancement technique outperformed all the model-based adaptation techniques. This suggests that knowledge of the underlying noise type can meaningfully inform the choice of adaptation technique. | ['Rajbabu Velmurugan', 'Preethi Jyothi', 'Archiki Prasad'] | 2021-02-11 | null | null | null | null | ['robust-speech-recognition'] | ['speech'] | [ 5.79912245e-01 2.73043979e-02 3.52295548e-01 -4.67200249e-01
-1.36012435e+00 -4.89001364e-01 6.29452944e-01 -4.00468297e-02
-7.03749359e-01 3.35866421e-01 5.37878573e-01 -7.26952970e-01
3.10718175e-02 -9.28777456e-02 -4.04484659e-01 -7.28862464e-01
1.82698727e-01 1.90934658e-01 9.51993465e-02 -5.06748557e-01
1.36757791e-01 4.35633838e-01 -1.38524270e+00 3.25343668e-01
5.34912825e-01 6.81810856e-01 2.75924176e-01 1.40537131e+00
-1.30459219e-01 5.00603437e-01 -1.09301651e+00 -2.67341256e-01
2.99830556e-01 -4.94806856e-01 -6.24034464e-01 1.87741682e-01
2.47915536e-01 -6.44500181e-03 -5.41054785e-01 9.92305219e-01
1.21252990e+00 6.51691437e-01 3.51792157e-01 -4.81960475e-01
-5.79384387e-01 6.37314081e-01 -1.63216397e-01 5.34432471e-01
4.33768451e-01 1.31621331e-01 4.60969269e-01 -9.28276539e-01
2.06183478e-01 1.38602471e+00 5.85022390e-01 9.02284384e-01
-1.12880647e+00 -5.11761010e-01 1.38406858e-01 -2.56790612e-02
-1.04213405e+00 -1.35152853e+00 7.81588316e-01 5.11672087e-02
1.24381721e+00 5.25337040e-01 7.93930143e-02 1.34386384e+00
-1.77274197e-01 7.20262289e-01 1.27455902e+00 -8.04251075e-01
2.15534672e-01 -2.48682071e-02 -6.19507618e-02 2.19117254e-02
-2.55272746e-01 5.29186845e-01 -5.63312888e-01 -4.97363247e-02
1.94314808e-01 -6.56903327e-01 -3.30456227e-01 2.92041332e-01
-1.01919520e+00 4.56873953e-01 5.48219606e-02 4.40859854e-01
-4.03522074e-01 -4.63935966e-03 7.60430515e-01 7.66973019e-01
6.87719405e-01 5.46820641e-01 -7.02881873e-01 -4.13783014e-01
-1.34716880e+00 1.11884266e-01 5.59511244e-01 6.94697559e-01
3.16610187e-01 8.79167318e-01 -2.09587350e-01 1.37272763e+00
4.36428517e-01 4.44670022e-01 6.13818765e-01 -7.42827356e-01
6.15810633e-01 -2.58910537e-01 -1.50229186e-01 -2.14288414e-01
-1.42250225e-01 -4.62769479e-01 -6.57909870e-01 6.16879940e-01
4.13501441e-01 -3.33630860e-01 -1.52191854e+00 1.47805333e+00
3.47402915e-02 1.19105965e-01 5.70657969e-01 6.94840252e-01
6.74612522e-01 7.45421708e-01 1.80365607e-01 -2.97521770e-01
1.01642799e+00 -9.48511302e-01 -1.06933999e+00 -6.11997783e-01
3.46222997e-01 -1.34524488e+00 9.21773314e-01 4.75045681e-01
-1.28883076e+00 -6.67556167e-01 -1.14383149e+00 1.17148831e-01
-4.82696891e-01 -8.25286284e-02 -1.92681272e-02 1.14730299e+00
-1.16653550e+00 3.89212787e-01 -8.49392891e-01 -1.87526837e-01
9.77835283e-02 4.23590690e-01 -4.07138109e-01 -1.00701913e-01
-1.23189020e+00 1.16723001e+00 3.43163937e-01 5.86087704e-02
-7.45957196e-01 -4.82774884e-01 -1.02354586e+00 -2.10780784e-01
3.55797738e-01 -4.00581360e-01 1.70580196e+00 -1.13955486e+00
-2.12046909e+00 4.86946732e-01 -3.75547558e-01 -5.24000168e-01
3.10786963e-01 -2.34316468e-01 -9.35524821e-01 -2.48311564e-01
-5.68042517e-01 2.38194719e-01 1.37630856e+00 -1.32656872e+00
-3.45893085e-01 -2.49015480e-01 -3.00778925e-01 4.46742743e-01
-6.69905618e-02 6.93174958e-01 -2.23068640e-01 -1.14558744e+00
1.55165698e-02 -7.66444087e-01 -4.45194751e-01 -4.50628042e-01
-1.15317762e-01 1.03131801e-01 1.02814615e+00 -1.11802447e+00
1.18610787e+00 -2.29664826e+00 -1.42760694e-01 1.48592800e-01
-5.23876011e-01 9.74669874e-01 -4.38794941e-01 3.22738767e-01
-3.79910618e-01 2.68194914e-01 -3.36519301e-01 -6.88723385e-01
-1.21105693e-01 2.35672966e-01 -2.87123710e-01 2.38207638e-01
2.59880543e-01 5.24110138e-01 -8.52100253e-01 4.99584572e-03
5.41046262e-01 9.55375612e-01 -3.03597987e-01 3.80810231e-01
9.03602317e-02 3.52862448e-01 3.66031639e-02 3.65681648e-01
5.48566341e-01 7.64300764e-01 -2.24303573e-01 7.97164589e-02
7.55948434e-03 7.96466172e-01 -1.34769475e+00 1.50307322e+00
-7.32776046e-01 7.60345459e-01 5.27321935e-01 -8.10414970e-01
1.11428225e+00 8.90616119e-01 -1.64576992e-02 -6.36628270e-01
-1.62962377e-01 4.65540886e-01 3.27080369e-01 -4.56212103e-01
5.43366015e-01 -3.98144037e-01 1.72647282e-01 1.26763314e-01
8.01620334e-02 -3.61561447e-01 -2.41030738e-01 -1.60135373e-01
1.18541062e+00 -7.67821679e-03 3.64305705e-01 1.28018618e-01
6.49122179e-01 -4.11969125e-01 2.71238387e-01 8.38859081e-01
-3.70139271e-01 9.63032305e-01 -3.62565696e-01 7.56118745e-02
-1.12565815e+00 -9.65976834e-01 1.16669379e-01 1.27148032e+00
-5.40929139e-01 -1.49672642e-01 -8.68189037e-01 -5.48800468e-01
-5.19658506e-01 1.06155229e+00 -1.84197918e-01 -1.91064283e-01
-6.78885996e-01 -6.11589670e-01 8.71342719e-01 4.45876092e-01
2.90430516e-01 -1.07712758e+00 -3.33740725e-03 3.76379848e-01
-1.05147831e-01 -9.93734419e-01 -7.88105845e-01 6.83898866e-01
-7.18090653e-01 -2.76196897e-01 -9.59704518e-01 -7.59980977e-01
4.04407740e-01 2.96263874e-01 8.16641629e-01 -4.62862253e-02
2.68049210e-01 4.73601490e-01 -6.44677997e-01 -5.51163256e-01
-1.33875561e+00 -8.09422508e-02 3.92565906e-01 -3.33754420e-02
3.14029604e-01 -4.02281702e-01 -2.17684075e-01 2.46479347e-01
-9.10844743e-01 -5.71526289e-01 6.81231797e-01 8.55317056e-01
4.23988372e-01 2.24478245e-01 9.06463027e-01 -6.35838270e-01
9.39861774e-01 -2.00300425e-01 -2.10410759e-01 1.88341051e-01
-5.49809396e-01 1.61560789e-01 6.67984068e-01 -7.33980834e-01
-1.40081227e+00 1.38323367e-01 -9.63987052e-01 -4.05999213e-01
-6.69843614e-01 5.30252874e-01 -4.23202068e-01 -3.61376405e-02
8.19391727e-01 2.43421972e-01 -8.83460511e-03 -6.82314038e-01
3.78667593e-01 1.25951767e+00 6.16683424e-01 -2.70051807e-01
9.32180882e-01 -2.04898044e-01 -5.70649981e-01 -1.39327109e+00
-4.39863503e-01 -7.49335706e-01 -4.70065325e-01 8.95979907e-03
5.92504323e-01 -8.59026074e-01 1.28160760e-01 5.88452637e-01
-1.12295437e+00 -6.14831448e-01 -3.19275260e-01 5.60315847e-01
-6.06817782e-01 4.96219754e-01 -3.41031849e-01 -1.18492723e+00
-4.93126899e-01 -1.35135686e+00 8.68705392e-01 -7.01717064e-02
-2.84723282e-01 -9.38811362e-01 6.60505220e-02 5.50596416e-01
9.86159503e-01 -5.16750336e-01 5.28233528e-01 -1.09413362e+00
2.64568757e-02 -4.61603254e-01 3.79667133e-01 8.78579021e-01
3.30869406e-01 -4.85990271e-02 -1.41566503e+00 -4.36994135e-01
2.56953448e-01 -1.25711709e-01 7.97128201e-01 4.24214274e-01
9.40904677e-01 -2.81752676e-01 1.90213904e-01 3.56497467e-01
9.50328410e-01 5.73943436e-01 9.64026213e-01 4.20614660e-01
3.98763716e-01 4.08357531e-01 4.59697485e-01 -1.04403190e-01
-2.13531554e-01 7.80195236e-01 -5.70253879e-02 -1.77240357e-01
-6.17321253e-01 5.26534989e-02 8.65080893e-01 9.35419083e-01
4.88934256e-02 -5.12051642e-01 -8.35307896e-01 5.57077408e-01
-1.23042178e+00 -1.18549418e+00 2.51328886e-01 2.26541042e+00
8.67451131e-01 3.38900000e-01 1.25786781e-01 6.03598833e-01
7.93665111e-01 3.86702329e-01 -2.66097814e-01 -8.54867160e-01
-1.88488647e-01 4.33987767e-01 4.95865822e-01 8.78840566e-01
-1.00109863e+00 8.75815272e-01 7.00077200e+00 8.89804721e-01
-1.08320630e+00 2.06898436e-01 4.72537071e-01 -1.07344933e-01
-1.67061582e-01 -1.12419017e-01 -5.86143553e-01 1.63259685e-01
1.44834089e+00 1.40885204e-01 7.98645735e-01 5.74023306e-01
7.10218191e-01 2.26153687e-01 -7.45780408e-01 8.29403818e-01
1.30729929e-01 -7.70065248e-01 -1.23857595e-01 -2.28792295e-01
5.33898175e-01 2.10517406e-01 2.32668102e-01 4.91768718e-01
1.13648027e-01 -9.40410495e-01 6.40940845e-01 1.77422658e-01
6.19593859e-01 -7.57870734e-01 6.79793119e-01 1.47999287e-01
-9.66265678e-01 -2.53549572e-02 -5.92660494e-02 2.65609443e-01
3.29644173e-01 2.51233160e-01 -8.37143302e-01 4.40773934e-01
4.91279215e-01 -2.15845648e-03 -4.54561800e-01 1.11730218e+00
-1.63596123e-01 1.22701287e+00 -4.09834802e-01 3.02429378e-01
1.55726492e-01 2.73303062e-01 1.03450263e+00 1.77802956e+00
3.16043913e-01 2.31540147e-02 -2.28201896e-02 1.57269567e-01
7.98763856e-02 7.77763203e-02 -5.47775865e-01 -1.35397777e-01
4.90532756e-01 9.11436319e-01 -3.30390602e-01 -1.87047318e-01
-5.27039886e-01 1.12997520e+00 -8.89009163e-02 7.44011521e-01
-3.11695814e-01 -5.24453461e-01 7.27874160e-01 1.21193700e-01
3.17306459e-01 -4.07430917e-01 -4.03513044e-01 -7.53759563e-01
-6.05182210e-03 -1.55718470e+00 1.61293939e-01 -6.32527769e-01
-1.13301086e+00 9.48308706e-01 -1.58441961e-01 -9.11954105e-01
-4.82757866e-01 -5.72897732e-01 -7.57967412e-01 1.16679466e+00
-1.48280501e+00 -1.05197883e+00 2.60457873e-01 4.49032217e-01
1.22472286e+00 -5.18389225e-01 9.18628395e-01 3.28487486e-01
-4.35183555e-01 9.22612667e-01 1.88017562e-01 1.12250658e-04
8.41027677e-01 -1.38270068e+00 9.52571452e-01 1.41106522e+00
2.32938811e-01 6.14836395e-01 1.04697478e+00 -5.14461815e-01
-1.13451099e+00 -1.10755575e+00 7.39898145e-01 -3.39248717e-01
4.52919543e-01 -2.55385339e-01 -1.36553097e+00 5.00377417e-01
4.75336969e-01 -1.84593699e-03 6.53907895e-01 1.18710019e-01
-2.94054776e-01 -8.89838189e-02 -9.77458298e-01 7.20235229e-01
7.33594596e-01 -7.58871853e-01 -7.80809164e-01 -6.55487180e-02
8.26349497e-01 -5.21582425e-01 -7.94374287e-01 3.99377763e-01
1.05785653e-01 -5.10919750e-01 1.03230357e+00 -4.73820299e-01
-2.38112524e-01 -3.73638660e-01 -4.35207903e-01 -1.83481050e+00
-1.66336685e-01 -1.16330945e+00 -1.31246045e-01 1.54283261e+00
8.39967012e-01 -4.93079960e-01 3.98175925e-01 5.90153337e-01
-7.74087250e-01 -3.74346197e-01 -1.04772556e+00 -9.93123829e-01
3.70222479e-02 -8.96433771e-01 3.65499914e-01 5.20904779e-01
-2.17466652e-01 2.31484622e-01 -5.21702886e-01 4.30168837e-01
3.16719592e-01 -7.95566618e-01 6.86093152e-01 -4.94524151e-01
-4.16481495e-01 -4.14012045e-01 -2.77662903e-01 -8.43575954e-01
1.57918721e-01 -7.24383056e-01 5.80921531e-01 -1.50011837e+00
-5.65402865e-01 -2.59550750e-01 -3.63833547e-01 4.60446656e-01
-5.54414213e-01 -3.91772725e-02 1.19856894e-01 -2.75640309e-01
-1.45464420e-01 4.99147117e-01 7.88049459e-01 -1.24037951e-01
-5.57955027e-01 5.20992815e-01 -6.65772736e-01 4.98584121e-01
1.09013891e+00 -4.57237393e-01 -4.97064948e-01 -4.94740486e-01
-3.89774263e-01 -8.44726488e-02 8.13036412e-02 -1.03478467e+00
2.74264038e-01 2.37199068e-02 1.66335836e-01 -2.82267213e-01
5.38779616e-01 -7.64674306e-01 -3.35871875e-02 6.72039688e-02
-6.48255348e-01 2.10196432e-02 6.53281152e-01 6.10416651e-01
-3.05554271e-01 -4.68067139e-01 1.16452682e+00 6.01093434e-02
-7.21373439e-01 -1.10009186e-01 -9.84179556e-01 1.60280559e-02
4.67050374e-01 -3.81827384e-01 5.43961488e-02 -6.78452134e-01
-9.24705088e-01 -3.47994119e-01 1.02973536e-01 7.79811859e-01
7.72632360e-01 -9.11637723e-01 -9.63841140e-01 4.02701616e-01
-7.34953359e-02 -2.80971646e-01 2.22000226e-01 3.34250242e-01
4.82159480e-03 1.56279474e-01 3.00392807e-01 -2.02538356e-01
-1.66313446e+00 6.70473218e-01 6.50353849e-01 -3.54114845e-02
-3.45148921e-01 7.38052130e-01 -4.57011670e-01 -5.04909456e-01
6.04186773e-01 -4.85333838e-02 -4.06180024e-02 -3.72176915e-01
7.78184950e-01 3.29016864e-01 6.13475442e-01 -8.84510994e-01
-1.64094657e-01 1.95449531e-01 -3.26577812e-01 -6.75591767e-01
1.19831979e+00 -3.52032602e-01 4.99680758e-01 4.28184688e-01
9.77449834e-01 1.84889659e-01 -9.76945043e-01 -2.31862918e-01
1.22999325e-01 -4.33372885e-01 5.28557718e-01 -1.08292091e+00
-7.59433746e-01 8.76803160e-01 8.61972630e-01 1.61424652e-01
1.41485679e+00 -3.77434611e-01 6.70723736e-01 4.21300143e-01
-8.41043666e-02 -1.29544640e+00 1.10543808e-02 7.78230071e-01
1.04600298e+00 -1.33055532e+00 -2.92996734e-01 -1.01204798e-01
-7.00569630e-01 9.68627512e-01 4.00083989e-01 2.61166543e-01
6.24522746e-01 5.97757995e-01 6.08954370e-01 2.65652090e-01
-5.25703430e-01 -5.12680411e-01 4.30609822e-01 1.03726363e+00
5.74450612e-01 -2.19995245e-01 1.56977519e-01 1.74020097e-01
-2.49839872e-01 -3.93850625e-01 3.04092318e-01 8.39991450e-01
-4.70319748e-01 -1.46803069e+00 -7.74745762e-01 2.30862811e-01
-8.98521245e-01 -2.92724609e-01 -3.76578957e-01 3.58141631e-01
-2.41678417e-01 1.53314388e+00 -3.24632198e-01 -4.05908614e-01
6.71616673e-01 4.65122789e-01 2.40561068e-01 -7.18368292e-01
-8.99468958e-01 3.75018358e-01 4.63063717e-01 -1.43370321e-02
-3.57654393e-01 -6.58468902e-01 -9.28046346e-01 -1.68235824e-02
-6.13381207e-01 -5.33451959e-02 1.10286999e+00 1.03907764e+00
9.96135920e-02 7.51848698e-01 6.06606364e-01 -9.70643997e-01
-8.37438583e-01 -1.20272553e+00 -1.36246189e-01 3.98347884e-01
7.84569979e-01 -1.24151602e-01 -5.28758228e-01 1.09200992e-01] | [14.725558280944824, 6.339413166046143] |
f178940a-2e4c-4ea5-80c2-43b3db5860d4 | multi-view-temporal-alignment-for-non | 2012.15184 | null | https://arxiv.org/abs/2012.15184v1 | https://arxiv.org/pdf/2012.15184v1.pdf | Multi-view Temporal Alignment for Non-parallel Articulatory-to-Acoustic Speech Synthesis | Articulatory-to-acoustic (A2A) synthesis refers to the generation of audible speech from captured movement of the speech articulators. This technique has numerous applications, such as restoring oral communication to people who cannot longer speak due to illness or injury. Most successful techniques so far adopt a supervised learning framework, in which time-synchronous articulatory-and-speech recordings are used to train a supervised machine learning algorithm that can be used later to map articulator movements to speech. This, however, prevents the application of A2A techniques in cases where parallel data is unavailable, e.g., a person has already lost her/his voice and only articulatory data can be captured. In this work, we propose a solution to this problem based on the theory of multi-view learning. The proposed algorithm attempts to find an optimal temporal alignment between pairs of non-aligned articulatory-and-acoustic sequences with the same phonetic content by projecting them into a common latent space where both views are maximally correlated and then applying dynamic time warping. Several variants of this idea are discussed and explored. We show that the quality of speech generated in the non-aligned scenario is comparable to that obtained in the parallel scenario. | ['Phil D. Green', 'Jose L. Perez-Cordoba', 'Alejandro Gomez-Alanis', 'Miriam Gonzalez-Atienza', 'Jose A. Gonzalez-Lopez'] | 2020-12-30 | null | null | null | null | ['multi-view-learning'] | ['computer-vision'] | [ 4.42304164e-01 2.90779471e-02 -1.14498168e-01 1.19301446e-01
-8.39160264e-01 -5.98113418e-01 7.10319579e-01 -2.75921464e-01
-1.68543011e-01 4.50552970e-01 5.36495686e-01 -6.09529763e-02
-2.26991221e-01 -3.63694757e-01 -3.56887251e-01 -1.04782963e+00
1.84324473e-01 3.17174554e-01 4.34687734e-02 -3.83539274e-02
9.13254097e-02 4.61518735e-01 -1.81850421e+00 2.27501526e-01
3.02128851e-01 1.72592565e-01 5.60397267e-01 9.03727293e-01
-2.46906787e-01 1.42987862e-01 -5.24857223e-01 -5.03425971e-02
2.23479614e-01 -7.09476829e-01 -5.47529817e-01 2.86535531e-01
5.46489730e-02 -7.78725967e-02 6.41489327e-02 7.33886957e-01
6.55125558e-01 1.31081030e-01 6.56153023e-01 -8.75752449e-01
2.76504736e-02 1.62894219e-01 -4.08602238e-01 -5.81631623e-02
5.97776055e-01 -2.92666823e-01 6.52406156e-01 -7.89758742e-01
7.44438648e-01 1.29539609e+00 2.91337669e-01 9.00450408e-01
-1.22480547e+00 -1.32214844e-01 -8.65550265e-02 3.28065008e-01
-1.01641321e+00 -8.77996624e-01 1.10662401e+00 -5.37322104e-01
7.30971277e-01 3.28720152e-01 5.82006276e-01 1.12167168e+00
2.08820626e-01 5.39951503e-01 1.10578752e+00 -9.24079418e-01
1.24676630e-01 3.07053514e-02 -2.93007225e-01 2.38617167e-01
-3.76209915e-01 3.79659176e-01 -7.15118825e-01 -1.14269875e-01
3.05074364e-01 -2.21963391e-01 -2.36509979e-01 -4.15048093e-01
-1.19089890e+00 6.27653480e-01 -3.88396204e-01 6.71226621e-01
-5.23239076e-01 -2.99457014e-01 4.96986121e-01 3.92845482e-01
4.26729023e-01 1.48515776e-01 -1.08509839e-01 -3.69339168e-01
-9.14922178e-01 -4.97733019e-02 6.66671574e-01 2.56530493e-01
2.07179502e-01 2.18309999e-01 5.05093217e-01 1.07563710e+00
3.61268520e-01 5.82892060e-01 8.40438068e-01 -8.93847644e-01
3.19025129e-01 2.35835128e-02 -6.27740100e-02 -6.19311213e-01
-3.00659865e-01 1.51105538e-01 -4.57208604e-01 5.41271627e-01
4.44148481e-01 -2.33764470e-01 -6.78970873e-01 1.73565245e+00
3.81306142e-01 2.91251361e-01 5.64519584e-01 6.55551016e-01
2.23321572e-01 8.15471590e-01 -3.81101996e-01 -9.55215931e-01
1.08192694e+00 -9.00787234e-01 -1.09669042e+00 -1.70984343e-01
3.01030219e-01 -1.28560376e+00 7.29645431e-01 4.75239307e-01
-1.22991419e+00 -5.88386953e-01 -9.97069001e-01 3.27133536e-01
-8.51073638e-02 -1.04334677e-05 -2.11051241e-01 6.58718050e-01
-9.31508780e-01 5.28428972e-01 -1.03190815e+00 -3.64068478e-01
-4.73983824e-01 2.82070667e-01 -5.43606281e-01 1.82497159e-01
-9.09126163e-01 9.23803151e-01 1.80310220e-01 4.04568277e-02
-5.53476274e-01 -2.23182678e-01 -8.19102287e-01 -2.26234123e-01
1.82636708e-01 -4.91779029e-01 1.15871561e+00 -8.20290029e-01
-1.99523342e+00 6.94238365e-01 -6.56476021e-01 -8.45122561e-02
5.28818965e-01 -2.73995310e-01 -6.32509410e-01 2.31351346e-01
-5.39046787e-02 1.08707868e-01 1.34570706e+00 -1.29965448e+00
-3.94956589e-01 -5.74090123e-01 -4.41180080e-01 5.06611347e-01
-2.33329862e-01 3.17497216e-02 -4.07082707e-01 -6.02245450e-01
3.18054438e-01 -1.18786633e+00 1.74205974e-01 -3.15551758e-01
-2.02916846e-01 -2.18239859e-01 1.04875302e+00 -8.97336006e-01
1.04350901e+00 -2.20106745e+00 6.61762953e-01 8.12653154e-02
-3.77135277e-01 4.23427641e-01 -1.36282012e-01 8.79514515e-01
-4.66906548e-01 -2.73362309e-01 -1.20616354e-01 -6.58038914e-01
-5.25571525e-01 2.82082170e-01 -4.17521775e-01 5.81225276e-01
-3.73696715e-01 2.40452722e-01 -7.99821317e-01 -4.00679916e-01
5.90097666e-01 5.44962823e-01 -2.34112263e-01 3.43753874e-01
1.44411027e-01 6.87690198e-01 -1.62776321e-01 1.42526790e-01
1.90478727e-01 5.14805496e-01 3.97714227e-01 -2.90922560e-02
-3.94895434e-01 3.01620960e-01 -1.10539317e+00 1.69651473e+00
-8.89902472e-01 6.72851086e-01 2.86373228e-01 -1.14505732e+00
9.91386592e-01 1.06413758e+00 8.15461159e-01 -4.56730753e-01
-2.61786878e-01 4.41121429e-01 1.21848971e-01 -9.03476954e-01
7.37476274e-02 -6.31259203e-01 2.63762444e-01 5.03028989e-01
-1.38899416e-01 -2.74648905e-01 6.59169331e-02 -3.84563953e-01
6.33864999e-01 2.83497870e-01 5.16518712e-01 7.83372074e-02
8.92157257e-01 -4.03953165e-01 2.36140922e-01 1.90739185e-01
-4.31919023e-02 6.32358611e-01 8.08996707e-02 -1.53058320e-01
-1.41566896e+00 -1.20035458e+00 -8.77854042e-03 6.75387979e-01
-2.46541142e-01 -2.67925739e-01 -8.00061822e-01 -1.50213331e-01
-4.60928857e-01 6.06780887e-01 -1.91309318e-01 -8.69450718e-02
-9.94900942e-01 -1.93637088e-01 2.17296019e-01 2.61160314e-01
-1.84845671e-01 -1.24490571e+00 -5.59317529e-01 4.84301746e-01
-3.46358269e-01 -1.00745642e+00 -3.93202603e-01 -1.36138096e-01
-9.69767809e-01 -7.75295675e-01 -9.02387142e-01 -7.88136005e-01
3.82865578e-01 3.48454148e-01 2.48851016e-01 -3.94124001e-01
-9.36650559e-02 5.30731201e-01 -2.83543020e-01 -3.81976694e-01
-1.02300656e+00 -1.90636277e-01 5.27583480e-01 1.38102338e-01
1.28400475e-01 -8.28670502e-01 -2.85377741e-01 3.70552927e-01
-7.06291437e-01 7.11510554e-02 2.16277465e-01 9.24261570e-01
4.57686692e-01 -7.30500445e-02 9.34419632e-01 -2.58763254e-01
6.60445869e-01 -1.79803103e-01 -4.30929840e-01 2.28289589e-01
-3.73679757e-01 2.47333758e-02 7.56961048e-01 -5.25901735e-01
-1.07951093e+00 3.09817225e-01 -4.91124958e-01 -6.01920187e-01
-2.91978866e-01 3.03769320e-01 -9.67134759e-02 4.99966949e-01
4.34035897e-01 4.07811165e-01 4.57202315e-01 -5.46000838e-01
4.16640878e-01 9.87381637e-01 8.12811613e-01 -2.07063347e-01
7.25995839e-01 6.05951548e-01 -6.42558783e-02 -1.32071710e+00
-2.48632520e-01 -8.57007861e-01 -8.83525074e-01 -5.01880884e-01
9.60418344e-01 -5.63846231e-01 -5.15163422e-01 6.28153861e-01
-1.19627106e+00 -2.22627353e-02 -2.44204164e-01 9.41106141e-01
-1.13355970e+00 6.70068920e-01 -2.00322136e-01 -1.09778380e+00
-1.57788724e-01 -1.19649982e+00 9.89063501e-01 9.12981704e-02
-4.42988604e-01 -1.13240004e+00 5.34581125e-01 5.54151237e-01
3.84836420e-02 3.91009599e-02 9.19273198e-01 -5.86818695e-01
1.46682814e-01 -4.55111191e-02 7.92032123e-01 4.64395076e-01
7.96597600e-01 -1.37881726e-01 -1.16756713e+00 -4.76606250e-01
4.43032801e-01 8.05656388e-02 4.31163162e-01 6.31608009e-01
5.18122137e-01 -2.24990755e-01 -3.58224094e-01 1.28891662e-01
9.57963049e-01 8.52261543e-01 4.07803595e-01 4.64051915e-03
4.66783881e-01 1.21509039e+00 4.59452152e-01 9.97827947e-02
-1.08256772e-01 1.20099664e+00 1.37789220e-01 2.33006269e-01
-3.36269170e-01 -1.67169645e-01 6.98661506e-01 1.49597764e+00
-1.54858485e-01 -2.15371877e-01 -7.49973416e-01 9.29569721e-01
-1.56952894e+00 -1.24334788e+00 1.81424692e-01 2.41690588e+00
5.86916268e-01 -8.20286199e-03 1.29699811e-01 5.22358775e-01
9.93492484e-01 3.05799603e-01 -3.77158344e-01 -7.33747602e-01
1.66803524e-01 1.40649304e-01 -3.35389897e-02 9.22017157e-01
-7.95234561e-01 5.79588354e-01 5.94740486e+00 6.10712707e-01
-1.39744246e+00 1.25296101e-01 -5.50919510e-02 -2.00377017e-01
-2.86395848e-01 -2.27559641e-01 -4.22006339e-01 2.93549031e-01
1.22973108e+00 -2.72435516e-01 5.23585677e-01 5.61424136e-01
8.05610061e-01 1.06762707e-01 -1.06293404e+00 1.03935325e+00
3.37264955e-01 -1.00290334e+00 -2.59254426e-01 1.62590012e-01
4.89949793e-01 -2.51528561e-01 2.01369636e-02 -2.67271429e-01
-4.90727693e-01 -7.04625189e-01 4.94652003e-01 4.62273777e-01
7.53887236e-01 -6.43566251e-01 3.09484184e-01 6.76135719e-01
-1.22411811e+00 -3.04916278e-02 1.28451377e-01 1.19887337e-01
7.97653615e-01 1.09963141e-01 -1.12118614e+00 6.36735082e-01
1.81879327e-01 4.28581804e-01 2.32593104e-01 7.91434288e-01
-1.11642443e-01 4.93780226e-01 -2.91062623e-01 2.87018359e-01
2.63475776e-02 -3.42902362e-01 1.20118701e+00 7.36097217e-01
8.08400869e-01 -8.24767202e-02 -2.65404619e-02 2.07932919e-01
4.57891881e-01 3.92166495e-01 -1.18794191e+00 4.60068733e-02
3.71080756e-01 8.23676586e-01 -5.75308681e-01 -1.66121453e-01
-4.97031718e-01 1.05892861e+00 -2.31418744e-01 1.81545496e-01
-2.11687565e-01 -3.94187942e-02 5.11809886e-01 1.06619246e-01
1.37529418e-01 -3.58361602e-01 1.40250877e-01 -8.67040694e-01
6.57849237e-02 -8.09547484e-01 1.19017102e-01 -7.40731001e-01
-8.91716778e-01 6.13125384e-01 2.55874544e-01 -1.40313435e+00
-1.25262010e+00 -3.93478394e-01 -6.92686439e-01 9.37790453e-01
-1.09134221e+00 -9.60093975e-01 2.72212833e-01 6.16861224e-01
1.20776439e+00 -4.73444253e-01 1.13186920e+00 2.11311325e-01
-9.43515375e-02 1.71937957e-01 3.58797312e-01 -3.22400600e-01
8.22373867e-01 -1.16106176e+00 2.13930458e-01 8.27687263e-01
4.28972751e-01 4.47878629e-01 8.54586720e-01 -5.72890222e-01
-1.11864901e+00 -4.74430412e-01 1.21014869e+00 -1.00255080e-01
4.16228205e-01 -1.40989736e-01 -9.09329891e-01 3.45020115e-01
2.64658540e-01 -2.85727441e-01 7.71521091e-01 -1.46255165e-01
2.28893876e-01 -5.40278703e-02 -7.60420203e-01 6.49713695e-01
6.56448185e-01 -8.77596855e-01 -9.68960166e-01 2.19837949e-01
3.22856158e-01 -2.15478435e-01 -7.95658648e-01 1.11969665e-01
8.46487701e-01 -9.16252911e-01 9.54223752e-01 -3.51166636e-01
6.76094070e-02 -2.59859413e-01 -6.54369146e-02 -1.63273263e+00
2.28372872e-01 -9.42526996e-01 4.68897074e-02 1.12340033e+00
8.36872309e-02 -4.77037549e-01 5.37876368e-01 -1.51680663e-01
-3.20060819e-01 -3.82557660e-01 -1.41259015e+00 -8.39584410e-01
-9.26073757e-04 -3.84683102e-01 2.33724788e-01 7.34843016e-01
1.85828671e-01 2.59573698e-01 -7.57364511e-01 1.79904059e-01
4.02339578e-01 1.84246823e-01 5.73877513e-01 -1.09065819e+00
-3.14315796e-01 -1.92451045e-01 -1.80605695e-01 -6.84506834e-01
4.04902160e-01 -6.87693655e-01 1.91115543e-01 -1.37134922e+00
-2.39486575e-01 -9.75448638e-02 1.19835787e-01 7.54539371e-02
1.80828094e-01 -1.47801518e-01 4.01301384e-01 4.07102585e-01
5.70918024e-01 3.95034313e-01 1.24276733e+00 1.64286971e-01
-6.40741885e-01 5.62608123e-01 1.00975581e-01 9.89691496e-01
7.06357241e-01 -3.46129477e-01 -6.94252551e-01 -1.33528471e-01
-2.70437717e-01 6.97996318e-01 4.43085656e-02 -8.60166430e-01
2.46010602e-01 -3.53913233e-02 -2.26062357e-01 -7.32936323e-01
7.46032357e-01 -8.88303041e-01 3.74898732e-01 6.25068486e-01
-2.10073978e-01 2.90114373e-01 -4.37549129e-03 5.09161353e-01
-4.22837108e-01 -6.57098174e-01 6.89086437e-01 -6.55329302e-02
-5.27634323e-01 -2.54599661e-01 -6.92982018e-01 -5.40323853e-01
1.05373275e+00 -2.84720361e-01 -8.61916840e-02 -6.24937415e-01
-1.19406116e+00 -3.70199621e-01 1.78021550e-01 7.28970170e-01
5.98999083e-01 -1.30119050e+00 -6.35554612e-01 3.18203688e-01
-2.19369642e-02 -4.09478247e-01 3.95242304e-01 7.74711847e-01
-2.14865714e-01 6.14332974e-01 -2.04626739e-01 -7.03956366e-01
-1.75070596e+00 6.60563409e-01 2.41435274e-01 -6.75736368e-02
-8.96721959e-01 2.18350485e-01 4.55481634e-02 -2.38206267e-01
8.59498698e-03 1.57126695e-01 -4.98941898e-01 3.70475829e-01
4.00249541e-01 4.51075941e-01 1.13587119e-01 -1.09193778e+00
-2.02914283e-01 8.90859604e-01 2.00275347e-01 -8.19484353e-01
1.30409300e+00 -4.59340990e-01 1.31611973e-01 8.76785219e-01
1.29644918e+00 4.99500692e-01 -1.03298569e+00 -8.01491961e-02
-1.18699558e-01 -3.02222610e-01 -1.54445946e-01 -3.53855640e-01
-6.47889316e-01 1.09533513e+00 8.68401766e-01 3.26692522e-01
1.12900090e+00 7.00586736e-02 7.45318055e-01 7.89663419e-02
9.47100669e-02 -1.09622324e+00 2.81459719e-01 1.49047732e-01
1.02895665e+00 -7.43908584e-01 -3.40755552e-01 -2.87086844e-01
-4.50363964e-01 1.43238044e+00 6.60404488e-02 2.45433941e-01
5.58042049e-01 2.25310087e-01 3.87742579e-01 1.54748932e-01
-6.72390103e-01 -1.40517533e-01 2.83639878e-01 8.51205826e-01
3.41329277e-01 3.78338955e-02 -4.97481436e-01 -2.33366564e-01
-4.47036117e-01 -3.11734408e-01 6.38266027e-01 7.31063426e-01
-1.79352641e-01 -1.61647499e+00 -7.41519630e-01 -6.34896606e-02
-3.96985680e-01 2.57887334e-01 -3.64420228e-02 5.67936718e-01
4.64915521e-02 1.04623377e+00 7.84197450e-02 -1.51044503e-01
4.33047950e-01 6.84711516e-01 5.21926701e-01 -5.95875502e-01
-6.09382689e-02 9.57809210e-01 1.53308719e-01 -2.74114847e-01
-8.72157395e-01 -1.26380289e+00 -1.09341991e+00 2.24522978e-01
-2.42270321e-01 1.08510271e-01 1.07280505e+00 1.20077085e+00
-9.55268294e-02 5.51241398e-01 8.66879702e-01 -8.79249454e-01
-4.81625885e-01 -7.47492850e-01 -4.31865066e-01 4.22464430e-01
7.35878646e-01 -5.33996344e-01 -3.98154318e-01 3.94012690e-01] | [14.574737548828125, 5.47404146194458] |
34a35dbb-96c6-4aad-b18f-4955c63b707e | computable-phenotypes-to-characterize | 2303.05504 | null | https://arxiv.org/abs/2303.05504v1 | https://arxiv.org/pdf/2303.05504v1.pdf | Computable Phenotypes to Characterize Changing Patient Brain Dysfunction in the Intensive Care Unit | In the United States, more than 5 million patients are admitted annually to ICUs, with ICU mortality of 10%-29% and costs over $82 billion. Acute brain dysfunction status, delirium, is often underdiagnosed or undervalued. This study's objective was to develop automated computable phenotypes for acute brain dysfunction states and describe transitions among brain dysfunction states to illustrate the clinical trajectories of ICU patients. We created two single-center, longitudinal EHR datasets for 48,817 adult patients admitted to an ICU at UFH Gainesville (GNV) and Jacksonville (JAX). We developed algorithms to quantify acute brain dysfunction status including coma, delirium, normal, or death at 12-hour intervals of each ICU admission and to identify acute brain dysfunction phenotypes using continuous acute brain dysfunction status and k-means clustering approach. There were 49,770 admissions for 37,835 patients in UFH GNV dataset and 18,472 admissions for 10,982 patients in UFH JAX dataset. In total, 18% of patients had coma as the worst brain dysfunction status; every 12 hours, around 4%-7% would transit to delirium, 22%-25% would recover, 3%-4% would expire, and 67%-68% would remain in a coma in the ICU. Additionally, 7% of patients had delirium as the worst brain dysfunction status; around 6%-7% would transit to coma, 40%-42% would be no delirium, 1% would expire, and 51%-52% would remain delirium in the ICU. There were three phenotypes: persistent coma/delirium, persistently normal, and transition from coma/delirium to normal almost exclusively in first 48 hours after ICU admission. We developed phenotyping scoring algorithms that determined acute brain dysfunction status every 12 hours while admitted to the ICU. This approach may be useful in developing prognostic and decision-support tools to aid patients and clinicians in decision-making on resource use and escalation of care. | ['Tezcan Ozrazgat-Baslanti', 'Azra Bihorac', 'Parisa Rashidi', 'Katharina Busl', 'Carolina B. Maciel', 'Benjamin Shickel', 'Rayon Uddin', 'Ziyuan Guan', 'Tyler J. Loftus', 'Yuanfang Ren'] | 2023-03-09 | null | null | null | null | ['icu-mortality'] | ['medical'] | [-2.35523611e-01 -2.49045789e-01 1.59241065e-01 -4.10268679e-02
-3.51328969e-01 -6.49742246e-01 -1.73214421e-01 7.13225424e-01
-8.34062397e-01 1.13747478e+00 5.03259480e-01 -7.66597509e-01
-4.08649921e-01 -6.56624615e-01 -1.61881279e-02 -3.98178130e-01
-3.01164895e-01 9.36045766e-01 -4.05874014e-01 4.30698305e-01
-8.48187879e-02 5.88838398e-01 -7.61208296e-01 7.47865736e-02
1.08924866e+00 6.68110490e-01 1.37693822e-01 7.09362388e-01
7.24250749e-02 9.77845490e-01 -6.25472188e-01 2.65882641e-01
1.32883444e-01 -6.63904309e-01 -5.98331094e-01 -1.61775768e-01
-4.50539261e-01 -7.85432756e-01 -3.85065377e-01 4.60929066e-01
6.75201654e-01 -3.49492542e-02 1.06170440e+00 -1.19755876e+00
-1.35536343e-01 3.35776269e-01 8.56466889e-02 4.14327979e-01
3.65656704e-01 5.69505632e-01 3.11539441e-01 -6.58692777e-01
1.01887703e-01 6.12416267e-01 4.79885340e-01 7.34449744e-01
-9.82336342e-01 -6.61934853e-01 -2.13522434e-01 5.41614071e-02
-1.26978648e+00 -1.85294241e-01 -2.35383466e-01 -9.88423586e-01
9.57349539e-01 -4.42580506e-02 1.15927255e+00 5.60925901e-01
5.95684230e-01 -1.33764818e-01 1.23632503e+00 3.70687783e-01
4.23555255e-01 -4.75157171e-01 7.15357482e-01 5.47073007e-01
7.37509966e-01 -1.16748169e-01 -2.77392894e-01 -7.63697267e-01
6.60377800e-01 8.98924351e-01 -8.06046903e-01 8.64371434e-02
-1.65660250e+00 5.20061672e-01 -3.08563951e-02 -3.77705693e-01
-7.64121652e-01 -2.48449981e-01 4.06988829e-01 2.89725244e-01
-2.62953043e-01 3.42733741e-01 -6.21925175e-01 -5.99819958e-01
-6.07571721e-01 -1.95189118e-01 8.01378012e-01 9.37418342e-01
4.45623338e-01 -1.53424039e-01 -2.03754559e-01 7.33185649e-01
-1.63888186e-02 6.71285868e-01 7.59022474e-01 -1.06191492e+00
2.88402349e-01 8.37752819e-01 5.10107458e-01 3.25890183e-02
-1.06217563e+00 -2.17882842e-01 -9.55577731e-01 -9.64320004e-02
3.18734825e-01 -7.76537240e-01 -7.85834372e-01 1.69970453e+00
-3.37929308e-01 -2.16981068e-01 3.59829187e-01 5.77183664e-01
2.96227932e-01 1.72078699e-01 1.74869940e-01 -6.08715236e-01
1.42869306e+00 -2.80481905e-01 -2.74103105e-01 1.96431011e-01
7.71502674e-01 -3.17714781e-01 1.09297442e+00 3.47474605e-01
-1.20817983e+00 1.72026549e-02 -5.96654713e-01 7.33710051e-01
1.21796623e-01 -2.69093007e-01 -6.59176633e-02 4.67181742e-01
-1.23951006e+00 4.58332092e-01 -9.57214653e-01 -6.84155405e-01
6.02058530e-01 4.96485174e-01 -3.22669894e-01 -4.20663893e-01
-8.29137623e-01 1.09927619e+00 3.77909929e-01 -5.32137930e-01
-1.03254580e+00 -9.71986890e-01 -5.47492802e-01 2.27055147e-01
-3.03698033e-01 -1.44476628e+00 8.97933304e-01 -8.84214975e-03
-7.36437500e-01 4.17369962e-01 -3.09655547e-01 -3.61974210e-01
4.64444757e-01 -1.34871632e-01 -3.02237511e-01 3.98184747e-01
1.65357143e-01 1.15798779e-01 4.58107851e-02 -7.14046836e-01
-5.59238195e-01 -6.86023533e-01 -3.18879068e-01 6.35276496e-01
-8.55431110e-02 8.32384676e-02 5.08439958e-01 -3.02790195e-01
5.31759746e-02 -9.19700623e-01 -1.15470812e-01 -3.83533865e-01
-1.64996870e-02 1.64220989e-01 7.96105206e-01 -5.22968590e-01
1.10630655e+00 -2.00423527e+00 -2.25258902e-01 -1.33508652e-01
7.68707216e-01 -1.75850734e-01 3.68714571e-01 2.27108866e-01
-1.70332074e-01 3.38713005e-02 -6.91021681e-01 -1.70293108e-01
-2.16765091e-01 2.78816998e-01 -1.21301070e-01 5.47622561e-01
-3.22121717e-02 5.78443468e-01 -1.07783973e+00 -4.70571443e-02
4.65120792e-01 1.92143366e-01 -3.65063518e-01 5.62769830e-01
7.23575950e-01 8.17770302e-01 8.11978206e-02 8.37665141e-01
3.96661729e-01 -3.50285530e-01 -6.29021274e-03 7.62600839e-01
-1.06496416e-01 1.19863011e-01 -2.40978405e-01 8.89360130e-01
3.24496813e-02 3.79670203e-01 4.04478833e-02 -6.79295123e-01
7.72871017e-01 5.72351336e-01 8.07845950e-01 -2.49068916e-01
4.20101106e-01 2.73254097e-01 4.65502203e-01 -3.20647955e-01
-1.49615183e-01 -7.57853270e-01 3.09773814e-02 8.41269732e-01
-4.97087449e-01 -5.33594824e-02 -3.98361012e-02 1.29239902e-01
1.80162323e+00 -6.52928829e-01 4.46092665e-01 -5.94593048e-01
1.08042084e-01 1.29137307e-01 7.79227316e-01 6.43415928e-01
-8.19677949e-01 7.91058779e-01 6.73950493e-01 -2.92053938e-01
-7.52702236e-01 -1.56638741e+00 -5.32771170e-01 4.70992416e-01
-1.64871678e-01 1.44180760e-01 -6.77157283e-01 -6.92455694e-02
-6.35916926e-03 9.92276669e-01 -3.09976399e-01 -5.60987294e-01
-3.32563400e-01 -1.15517879e+00 5.67007184e-01 6.19653761e-01
7.08885610e-01 -1.00962532e+00 -1.13761663e+00 5.09095669e-01
-4.13344622e-01 -4.29957241e-01 -3.98256034e-01 5.66490352e-01
-1.20681190e+00 -1.42916322e+00 -1.38195229e+00 -4.42216009e-01
7.50630498e-01 2.01579496e-01 8.46753955e-01 -4.51455750e-02
-2.38877386e-01 5.77455759e-01 -1.66988477e-01 -3.11505854e-01
-2.90436924e-01 -5.16036272e-01 6.99461818e-01 -4.44970310e-01
5.03978610e-01 -5.29433370e-01 -1.19035363e+00 2.92015821e-01
-5.37249744e-01 -3.08938384e-01 4.61034864e-01 6.92557395e-01
3.19477707e-01 -5.28022289e-01 9.18560147e-01 -4.09649700e-01
8.50308895e-01 -1.01342666e+00 -1.68770254e-02 -4.79637971e-03
-1.15978789e+00 -4.87293422e-01 7.61180341e-01 -3.35788429e-01
-3.93361896e-01 -2.65396804e-01 3.71554643e-01 -4.97675925e-01
-4.03515428e-01 5.30148089e-01 1.22613192e-01 8.98371696e-01
4.30306494e-01 3.22031468e-01 1.34282693e-01 -1.68614790e-01
-5.73491454e-01 9.60196495e-01 5.99631548e-01 -4.89471287e-01
3.43240909e-02 2.56184071e-01 -9.14847106e-02 -6.58477843e-01
-1.71194583e-01 -6.92164540e-01 -5.45382798e-01 -2.95981653e-02
1.21327353e+00 -1.14727211e+00 -8.58439505e-01 5.68902671e-01
-5.72554469e-01 -9.12466049e-01 -2.52653807e-01 9.53980982e-01
-4.97301877e-01 4.91347641e-01 -4.90762949e-01 -6.83106363e-01
-6.16590202e-01 -1.33834219e+00 2.01624334e-01 1.01052657e-01
-7.85924017e-01 -8.16074193e-01 1.60535291e-01 2.44694531e-01
4.58829612e-01 3.69367152e-01 1.74152696e+00 -6.04060531e-01
-1.25662517e-02 2.27077920e-02 -3.56334507e-01 3.50497693e-01
7.22599745e-01 -3.65140349e-01 -1.39967710e-01 -5.81725895e-01
4.85238731e-02 1.17096871e-01 2.83130974e-01 7.14930296e-01
8.42921197e-01 1.01622939e-01 -1.54497907e-01 6.32952571e-01
1.13483310e+00 1.16652226e+00 5.60286701e-01 1.36408970e-01
3.25169861e-01 2.95284212e-01 7.64644444e-02 8.78311098e-01
7.33188331e-01 -2.63247997e-01 1.96515933e-01 2.20173538e-01
2.05981627e-01 3.19913983e-01 5.34318030e-01 8.17774534e-01
-2.10214436e-01 -1.90569103e-01 -1.49006009e+00 5.64212263e-01
-1.47587967e+00 -6.94914639e-01 -5.08810639e-01 2.65176415e+00
5.36586761e-01 -1.20106637e-01 3.00352991e-01 -2.61285871e-01
7.50253141e-01 -8.37777495e-01 -1.03829634e+00 -3.36799234e-01
-7.62049928e-02 8.99638087e-02 4.89721239e-01 1.10772803e-01
-3.60125482e-01 2.94164330e-01 6.55357361e+00 -7.54238605e-01
-9.97314692e-01 -2.60019917e-02 9.32481527e-01 -5.00185788e-01
3.26707393e-01 1.11743003e-01 -2.30325773e-01 7.26542056e-01
1.27511072e+00 -4.51336443e-01 4.85984981e-01 6.66684747e-01
6.97798073e-01 -3.66455168e-01 -1.17682171e+00 1.26185453e+00
-1.28530994e-01 -8.87051105e-01 -1.38118342e-02 2.10406855e-01
6.17218077e-01 3.32800001e-01 -2.10450560e-01 2.44066849e-01
5.18634737e-01 -1.15795684e+00 4.76035446e-01 9.33898687e-01
1.64538062e+00 -8.45542729e-01 9.70244110e-01 2.53431350e-01
-8.78168821e-01 -3.50256115e-01 -2.64428109e-01 -4.78930712e-01
2.70812154e-01 4.24736410e-01 -9.82734680e-01 -1.04883820e-01
8.81762803e-01 2.02997357e-01 -1.03798881e-01 1.01435435e+00
2.90880259e-02 6.75695240e-01 -2.87334800e-01 8.83843750e-02
-4.67179380e-02 -4.57569778e-01 4.50280458e-01 7.93208420e-01
6.03336215e-01 7.66858399e-01 -1.08288631e-01 6.58520103e-01
-1.77842453e-01 -1.15352862e-01 -6.55824542e-01 -2.81349141e-02
4.75322992e-01 6.03232026e-01 -7.11384296e-01 -7.89253414e-01
-4.49810117e-01 9.64010656e-01 -1.23664243e-02 6.75084472e-01
-7.11928844e-01 -5.56876242e-01 1.01142883e+00 2.39239395e-01
-6.71665251e-01 -1.35845900e-01 -7.73082197e-01 -1.32675421e+00
-4.30716157e-01 -6.29474819e-01 5.40706813e-01 -9.36443329e-01
-1.18719816e+00 5.33969641e-01 -1.18403211e-01 -9.73932743e-01
-2.80793279e-01 -4.95966405e-01 -7.83041656e-01 1.38413346e+00
-9.90887582e-01 2.06307635e-01 -7.70895720e-01 9.12758946e-01
1.90530509e-01 -2.08727777e-01 1.14838099e+00 2.73020472e-02
-7.66623378e-01 7.51655102e-02 2.97619671e-01 1.51563689e-01
7.73488164e-01 -1.13965321e+00 -4.27056879e-01 4.16051745e-01
-1.36593497e+00 1.03959990e+00 8.61469377e-03 -9.29990411e-01
-1.01701391e+00 -1.36283958e+00 7.03417420e-01 -5.58930039e-01
4.73132968e-01 3.06239635e-01 -9.66532588e-01 6.01265609e-01
5.34554757e-02 -4.44167048e-01 1.17895377e+00 -3.37339699e-01
2.88001209e-01 4.88028556e-01 -1.34187675e+00 7.37145483e-01
8.28518987e-01 -4.43977982e-01 -1.08051109e+00 9.10983458e-02
1.06136717e-01 -2.56523006e-02 -1.27412212e+00 4.33873057e-01
4.32403862e-01 -7.84398615e-01 5.12898624e-01 -4.88638222e-01
2.96816349e-01 -2.28195846e-01 -1.20994009e-01 -1.40871215e+00
-5.43311000e-01 -6.04401290e-01 4.19863872e-02 3.82016689e-01
2.02114224e-01 -1.40115941e+00 3.78360927e-01 1.37570941e+00
-7.16995001e-01 -9.13103700e-01 -9.20074999e-01 -5.48232496e-01
6.83935642e-01 2.48329509e-02 7.61138380e-01 8.12124908e-01
4.47031230e-01 2.59887993e-01 3.36851001e-01 -5.45260757e-02
4.12426502e-01 -1.35721102e-01 2.56539106e-01 -1.27388525e+00
1.46476999e-01 -7.37888873e-01 -1.31897733e-01 -3.93083185e-01
-2.19167769e-01 -9.10195291e-01 -3.99571717e-01 -2.29252338e+00
9.12009716e-01 -4.24486607e-01 -6.53017223e-01 7.05210507e-01
-3.04087728e-01 -4.18545604e-01 -1.70467839e-01 5.44729948e-01
1.43026993e-01 3.96045774e-01 8.75445306e-01 2.06588864e-01
-4.68473583e-01 -2.31175810e-01 -8.79300237e-01 5.78640759e-01
1.03264809e+00 -3.41341168e-01 -4.80058879e-01 -1.93087846e-01
-3.15492779e-01 6.56966925e-01 1.88126802e-01 -1.00329816e+00
8.81048366e-02 -4.36804712e-01 7.20242381e-01 -7.24847555e-01
2.69226055e-03 -6.04809701e-01 4.14332241e-01 1.11158168e+00
3.00398797e-01 6.44648850e-01 2.50066489e-01 1.20362952e-01
1.30516827e-01 2.88218800e-02 1.03151238e+00 -1.20223098e-01
-1.33980855e-01 5.72182417e-01 -1.35950100e+00 4.50508237e-01
1.23861146e+00 -4.21399474e-01 -5.21368444e-01 -3.31500739e-01
-9.97761488e-01 4.90246475e-01 9.53877270e-01 -1.91776931e-01
9.26245332e-01 -8.42111588e-01 -7.87381053e-01 5.16947433e-02
1.07849345e-01 1.13677509e-01 3.91949803e-01 1.35183704e+00
-9.45629418e-01 3.48696262e-01 -2.73550689e-01 -3.08254600e-01
-7.82022536e-01 3.78415793e-01 4.89877105e-01 1.92489251e-01
-9.48998511e-01 4.15513098e-01 4.75811154e-01 1.08951502e-01
1.91542014e-01 -6.66045904e-01 3.44400376e-01 -8.72623641e-04
5.80271721e-01 7.25082874e-01 -1.87608793e-01 -4.78210300e-01
-5.52883923e-01 1.21011712e-01 2.36434475e-01 -4.65004817e-02
9.77937818e-01 -2.39006951e-01 -3.54594588e-01 5.43388367e-01
1.08916545e+00 -5.36997736e-01 -1.12433732e+00 4.71502900e-01
-4.51536447e-01 1.37098640e-01 -2.16768727e-01 -9.81149912e-01
-4.96374458e-01 8.90471160e-01 6.42456114e-01 -2.31290564e-01
1.18968594e+00 -3.40705588e-02 9.08303380e-01 4.67609942e-01
6.10618293e-01 -6.95435464e-01 -2.44847879e-01 8.21768701e-01
5.17853379e-01 -7.36001015e-01 -4.07946289e-01 6.09184980e-01
-7.21592069e-01 1.14431214e+00 2.38971993e-01 -3.94042321e-02
6.23030782e-01 1.05267286e-01 2.15059310e-01 -3.21239173e-01
-7.95935869e-01 2.01712355e-01 -5.29319406e-01 5.11857569e-01
3.08223695e-01 5.96020401e-01 -2.77553260e-01 7.66222060e-01
-2.92924732e-01 -1.23808347e-02 9.80075836e-01 1.07034206e+00
-5.32279551e-01 -6.19300842e-01 -4.49473560e-01 1.42758548e+00
-2.97196448e-01 -4.26287919e-01 -3.30754220e-01 4.31681097e-01
-2.15644184e-02 1.00274408e+00 2.63939410e-01 -1.99496716e-01
4.55524623e-01 9.78664756e-01 1.11373022e-01 -7.16266930e-01
-3.79209220e-01 5.63634075e-02 -2.11642429e-01 8.46880972e-02
1.91777200e-01 -1.06180120e+00 -2.08395624e+00 -4.15934294e-01
2.83585429e-01 1.51287898e-01 1.77719533e-01 8.04717362e-01
6.36837900e-01 7.04216480e-01 3.38905752e-01 -1.09366566e-01
-4.33696449e-01 -1.08822310e+00 -8.70691597e-01 -1.44551611e-02
5.78463137e-01 -7.68574834e-01 -8.61195028e-01 9.34244245e-02] | [8.009265899658203, 6.136665344238281] |
f9baf5b6-b74e-4fad-9d4a-511bed4e4dc3 | zbs-zero-shot-background-subtraction-via | 2303.14679 | null | https://arxiv.org/abs/2303.14679v1 | https://arxiv.org/pdf/2303.14679v1.pdf | ZBS: Zero-shot Background Subtraction via Instance-level Background Modeling and Foreground Selection | Background subtraction (BGS) aims to extract all moving objects in the video frames to obtain binary foreground segmentation masks. Deep learning has been widely used in this field. Compared with supervised-based BGS methods, unsupervised methods have better generalization. However, previous unsupervised deep learning BGS algorithms perform poorly in sophisticated scenarios such as shadows or night lights, and they cannot detect objects outside the pre-defined categories. In this work, we propose an unsupervised BGS algorithm based on zero-shot object detection called Zero-shot Background Subtraction (ZBS). The proposed method fully utilizes the advantages of zero-shot object detection to build the open-vocabulary instance-level background model. Based on it, the foreground can be effectively extracted by comparing the detection results of new frames with the background model. ZBS performs well for sophisticated scenarios, and it has rich and extensible categories. Furthermore, our method can easily generalize to other tasks, such as abandoned object detection in unseen environments. We experimentally show that ZBS surpasses state-of-the-art unsupervised BGS methods by 4.70% F-Measure on the CDnet 2014 dataset. The code is released at https://github.com/CASIA-IVA-Lab/ZBS. | ['Jinqiao Wang', 'Ming Tang', 'Chaoyang Zhao', 'Haiyun Guo', 'Tao Yu', 'Xu Zhao', 'Yongqi An'] | 2023-03-26 | null | http://openaccess.thecvf.com//content/CVPR2023/html/An_ZBS_Zero-Shot_Background_Subtraction_via_Instance-Level_Background_Modeling_and_Foreground_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/An_ZBS_Zero-Shot_Background_Subtraction_via_Instance-Level_Background_Modeling_and_Foreground_CVPR_2023_paper.pdf | cvpr-2023-1 | ['zero-shot-object-detection', 'foreground-segmentation'] | ['computer-vision', 'computer-vision'] | [ 2.37974495e-01 -4.71735865e-01 4.88981903e-02 -3.29532146e-01
-4.31988984e-01 -1.60409153e-01 4.95730251e-01 -2.17784151e-01
-4.20451343e-01 6.14287615e-01 -3.09358746e-01 -3.18418175e-01
4.64810401e-01 -8.33706617e-01 -6.77326441e-01 -1.17196655e+00
2.01219767e-01 5.60982227e-02 1.11997283e+00 7.77529366e-03
-7.24117011e-02 4.59911168e-01 -1.52635622e+00 6.26473203e-02
6.99353635e-01 9.38019931e-01 3.78457427e-01 6.84134066e-01
-3.74201566e-01 9.41195428e-01 -8.44918430e-01 -2.05563962e-01
3.68874490e-01 -3.94760281e-01 -2.45215982e-01 2.10985348e-01
5.74472427e-01 -7.87043214e-01 -5.36727965e-01 1.28787887e+00
4.94167924e-01 2.47818813e-01 3.36735874e-01 -1.45774579e+00
-1.71683684e-01 1.90385029e-01 -7.20055103e-01 8.38616490e-01
-2.22916380e-01 3.05989981e-01 4.45728660e-01 -8.25302303e-01
4.61308032e-01 1.23442543e+00 4.75908607e-01 6.49745405e-01
-1.04281282e+00 -9.24078047e-01 4.08780873e-01 3.97172540e-01
-1.34969318e+00 -4.74908203e-01 4.91460234e-01 -5.35390079e-01
5.01785934e-01 3.16856503e-01 6.53834760e-01 1.07839775e+00
-7.73611069e-02 1.27299845e+00 1.00067413e+00 -3.00614059e-01
3.61164182e-01 -1.08997948e-01 4.93558437e-01 4.43709075e-01
6.95692003e-01 2.06021946e-02 -3.11201781e-01 2.10614681e-01
7.15407789e-01 5.10094523e-01 -3.80907565e-01 -3.68977040e-01
-1.16045630e+00 6.36634707e-01 2.67487526e-01 2.44451955e-01
-1.20533325e-01 5.75015461e-03 2.48594731e-01 -1.29017904e-01
5.58511496e-01 -2.87234217e-01 -4.55466241e-01 1.16915688e-01
-1.14840961e+00 7.36766607e-02 6.86355114e-01 1.09399307e+00
7.75715351e-01 4.02611792e-01 -3.03036898e-01 6.07031345e-01
2.33586788e-01 6.90635383e-01 3.11033398e-01 -8.00809264e-01
7.24076778e-02 3.81911129e-01 2.01168120e-01 -9.63477790e-01
-2.54163116e-01 -4.15884495e-01 -9.55943048e-01 3.20315272e-01
5.49955726e-01 -3.08823764e-01 -1.26126873e+00 1.38204527e+00
6.63820386e-01 7.80066192e-01 1.08687086e-02 9.96127188e-01
1.26466751e+00 8.06753278e-01 5.46118803e-02 -3.57269585e-01
1.07457232e+00 -1.01021826e+00 -8.46686602e-01 -3.90141815e-01
2.05395192e-01 -5.26636183e-01 7.99621224e-01 4.32591796e-01
-5.72414935e-01 -6.55594349e-01 -1.11608422e+00 2.23378137e-01
-3.99430543e-01 2.12326813e-02 5.11333764e-01 9.26376820e-01
-8.65571082e-01 2.98683465e-01 -1.09042454e+00 -4.40980524e-01
7.03672349e-01 2.22511515e-01 9.35848802e-02 -1.03702255e-01
-1.08238304e+00 5.07241011e-01 5.94551086e-01 1.07607082e-01
-1.31631589e+00 -3.61116499e-01 -7.74848700e-01 -6.29435033e-02
1.01624191e+00 -3.90495956e-01 1.33346903e+00 -1.22371399e+00
-1.26778936e+00 8.24547052e-01 -3.47178787e-01 -5.85317612e-01
6.55655503e-01 -3.40219796e-01 -5.30577302e-01 1.45106018e-01
1.55908585e-01 4.54812497e-01 9.78232503e-01 -1.20493448e+00
-9.29132521e-01 -1.99343756e-01 8.10360238e-02 -1.26646370e-01
-1.81183591e-01 3.28531563e-01 -8.80325675e-01 -6.94435716e-01
5.75758852e-02 -5.70899963e-01 -2.40446627e-01 4.83274683e-02
-4.48496759e-01 6.22622408e-02 1.23026729e+00 -6.92109466e-01
9.45186675e-01 -2.30965137e+00 -3.86879265e-01 -2.15402186e-01
4.65906173e-01 7.88783729e-01 1.24177255e-01 -1.77049786e-01
2.30245426e-01 -1.13415487e-01 -3.80724698e-01 -1.02352552e-01
-1.92668915e-01 1.80651411e-01 -1.85269713e-01 5.17035961e-01
1.21296898e-01 6.06192589e-01 -1.07818723e+00 -6.95868909e-01
5.42043388e-01 2.04552591e-01 -1.35278031e-01 1.03864476e-01
-1.29794985e-01 5.67112267e-01 -1.97952211e-01 8.51075053e-01
1.05306208e+00 -1.23995222e-01 -2.34159771e-02 1.12962775e-01
-1.11448631e-01 -1.09148890e-01 -1.41563702e+00 1.12718344e+00
3.05817008e-01 9.99271810e-01 1.30565524e-01 -1.09057009e+00
7.03628898e-01 -3.15525271e-02 3.11304748e-01 -6.23006165e-01
2.51007169e-01 -2.07872819e-02 1.69883147e-02 -3.45526367e-01
1.76286057e-01 -4.64865044e-02 2.97733486e-01 -1.98111713e-01
2.51299858e-01 3.18555146e-01 5.47268689e-01 3.20261866e-01
9.46586549e-01 1.42801613e-01 2.89027572e-01 -2.07689047e-01
5.56528032e-01 -1.84443489e-01 1.09309983e+00 1.14721608e+00
-6.69781029e-01 5.24360657e-01 2.33263850e-01 -3.23368013e-01
-3.86196315e-01 -1.26948321e+00 -1.93519853e-02 1.16579521e+00
7.19191968e-01 -1.81529641e-01 -1.03877831e+00 -6.04820728e-01
-1.34826094e-01 7.83773541e-01 -3.57629269e-01 -5.96358068e-02
-4.28699970e-01 -1.08267736e+00 3.94629270e-01 5.91586292e-01
9.29654896e-01 -1.06037593e+00 -5.70768595e-01 2.32332677e-01
-2.81746387e-01 -1.38724995e+00 -9.12774727e-02 3.81962508e-02
-7.06838727e-01 -1.20512402e+00 -8.51348519e-01 -6.71498358e-01
4.53331918e-01 8.86795163e-01 9.83915448e-01 5.22796847e-02
-5.25540233e-01 3.59483272e-01 -3.62548053e-01 -9.23592329e-01
-2.89775968e-01 -3.61924171e-01 7.46023357e-02 4.54809368e-01
7.49291182e-01 -1.77891105e-01 -6.48661077e-01 5.70065677e-01
-8.94960761e-01 2.08201513e-01 3.98986667e-01 4.32228088e-01
6.91078663e-01 3.86836767e-01 3.08436692e-01 -7.68966675e-01
-1.07978411e-01 -3.44448388e-01 -9.76043105e-01 1.90245613e-01
-1.06486879e-01 -7.02476323e-01 2.49596879e-01 -5.01689076e-01
-1.31411672e+00 -8.67393389e-02 1.07854225e-01 -4.36605185e-01
-5.55727184e-01 -2.09776655e-01 -6.77107751e-01 -7.43461773e-02
3.07610005e-01 3.24404657e-01 -4.41722929e-01 -4.78860587e-01
2.51391530e-01 6.33079052e-01 6.55508518e-01 -1.93278998e-01
9.08510864e-01 8.13246071e-01 -3.15496355e-01 -1.23723650e+00
-7.95359552e-01 -9.34057176e-01 -6.68197751e-01 -4.81317908e-01
1.05834043e+00 -1.09389246e+00 -3.12043637e-01 1.10889053e+00
-9.49473262e-01 -6.38666332e-01 -1.26204088e-01 4.10695225e-01
-2.47493774e-01 5.60662866e-01 -6.25922322e-01 -1.21297967e+00
-9.09722373e-02 -9.75232661e-01 1.05194199e+00 6.64415002e-01
2.78445065e-01 -6.45437896e-01 -4.53691989e-01 2.90783733e-01
1.63303375e-01 3.56869459e-01 1.73662663e-01 -7.74210155e-01
-1.00207365e+00 -5.57209887e-02 -2.93122709e-01 6.24256194e-01
1.95561185e-01 2.99271971e-01 -1.08474350e+00 -1.79485992e-01
1.91558227e-01 1.94630146e-01 1.30077088e+00 6.26410246e-01
1.12624884e+00 -1.85875241e-02 -3.99849772e-01 7.51260936e-01
1.29641342e+00 6.19861543e-01 8.58802259e-01 5.39996445e-01
8.90303671e-01 2.18090996e-01 8.30427289e-01 4.16473866e-01
-2.62875892e-02 4.06695276e-01 4.69469577e-01 -4.76265639e-01
-1.44113660e-01 4.01768714e-01 5.19541800e-01 4.61516678e-01
-1.29259611e-02 -3.32480729e-01 -9.96080577e-01 5.04004657e-01
-1.89464247e+00 -1.11390436e+00 -4.33857620e-01 2.11860061e+00
6.69538617e-01 5.51257133e-01 1.53654188e-01 1.66188050e-02
1.24447846e+00 2.70430505e-01 -6.52965605e-01 2.50753492e-01
-3.29273522e-01 1.42162085e-01 6.90344393e-01 1.05351515e-01
-1.68442976e+00 1.16352189e+00 5.26218653e+00 9.43736374e-01
-1.01392424e+00 3.30475211e-01 6.76553249e-01 -1.73324198e-01
5.60407758e-01 -1.83278188e-01 -1.14861298e+00 7.57301331e-01
6.55306518e-01 -7.51727149e-02 9.57563519e-02 9.25196290e-01
2.77469814e-01 -3.27351779e-01 -7.09490716e-01 1.04704142e+00
6.61698356e-02 -1.21150351e+00 -1.22473374e-01 -2.52113342e-01
8.03895235e-01 1.89224243e-01 -3.16210538e-01 6.08277380e-01
2.21656114e-01 -4.37638372e-01 7.91414499e-01 3.92348260e-01
5.07941902e-01 -4.17458445e-01 8.56357992e-01 5.00699103e-01
-1.17164505e+00 -6.28552511e-02 -5.26901364e-01 -8.98533687e-02
2.04001442e-01 6.57159328e-01 -5.20224154e-01 4.54680175e-01
1.02417111e+00 6.51166677e-01 -6.78128541e-01 1.51409888e+00
-3.09397727e-01 1.01271248e+00 -3.32833797e-01 1.07760087e-01
2.57616848e-01 -2.88919598e-01 6.80276752e-01 1.59072292e+00
5.55295534e-02 1.83924124e-01 4.35900122e-01 7.20823228e-01
1.06838502e-01 -1.54644707e-02 -3.37549865e-01 3.09755653e-02
1.10823266e-01 1.24772978e+00 -1.36805344e+00 -9.60118532e-01
-6.99450016e-01 9.43151653e-01 -3.53104681e-01 6.19832516e-01
-1.19316745e+00 -4.72595155e-01 7.24152148e-01 2.08458737e-01
6.99646831e-01 -1.71393901e-01 2.92735621e-02 -1.40029132e+00
-1.58486739e-01 -8.01658869e-01 3.49401146e-01 -8.69321465e-01
-1.12483478e+00 1.67816639e-01 1.40151426e-01 -1.16156542e+00
4.49990720e-01 -8.19243908e-01 -8.68761301e-01 4.21293765e-01
-1.50063646e+00 -7.70116627e-01 -8.82307112e-01 6.77858710e-01
8.38500857e-01 -2.01926365e-01 3.18778396e-01 5.40808558e-01
-1.00981164e+00 1.48851573e-01 4.30194855e-01 7.08670497e-01
7.24544942e-01 -1.20537138e+00 6.46730185e-01 1.57680380e+00
1.57438874e-01 4.33360189e-01 6.07501090e-01 -8.94530892e-01
-9.81944144e-01 -1.42486954e+00 1.15241282e-01 -9.29792672e-02
6.40862525e-01 -4.86281514e-01 -1.02123642e+00 5.94239235e-01
-5.10313548e-02 3.96667063e-01 5.44584632e-01 -3.90776783e-01
-1.17199808e-01 -2.90858179e-01 -9.32193935e-01 7.01829195e-01
1.08849835e+00 -2.97678132e-02 -5.41551650e-01 3.24317694e-01
7.33692050e-01 -4.30713415e-01 -6.64755479e-02 6.14617229e-01
1.28257856e-01 -1.09048319e+00 9.96212780e-01 -4.76610750e-01
-1.03460550e-01 -5.92096865e-01 -2.72201210e-01 -6.61628246e-01
-2.85097271e-01 -6.47481263e-01 -3.26238185e-01 1.34199333e+00
-2.51037121e-01 -6.75280035e-01 7.02851057e-01 2.84166068e-01
-7.48355612e-02 -1.73651323e-01 -7.95120955e-01 -1.12300086e+00
-3.16058457e-01 -4.84003335e-01 4.27856445e-01 8.01258802e-01
-8.96252275e-01 -3.80466208e-02 -3.05883318e-01 4.71728474e-01
9.62581396e-01 1.07317530e-01 1.16673911e+00 -1.32011485e+00
-1.90647408e-01 -4.64776784e-01 -7.14127839e-01 -1.05607212e+00
-1.28272608e-01 -4.56283599e-01 3.25347900e-01 -1.66059303e+00
3.63633633e-01 3.77343240e-04 -6.60840988e-01 3.59223247e-01
-5.29552698e-01 3.98866057e-01 3.22489530e-01 5.82931526e-02
-1.19980633e+00 3.49937171e-01 9.17231560e-01 -3.50560397e-01
-1.23398066e-01 1.77022666e-01 -3.94554079e-01 1.19566882e+00
1.16693342e+00 -4.45650965e-01 -1.86942413e-01 -1.05577789e-01
-7.66941607e-01 -3.72849107e-01 6.35886431e-01 -1.17983329e+00
6.96351007e-02 -4.28347588e-01 6.32206500e-01 -8.30784976e-01
1.29714057e-01 -6.19409800e-01 -6.30014315e-02 4.67451841e-01
3.62108558e-01 -7.15843678e-01 4.54918146e-01 8.84393632e-01
-2.35937703e-02 -1.02349751e-01 8.80325317e-01 -2.35103860e-01
-1.49970579e+00 4.05649364e-01 -7.91288972e-01 2.21990004e-01
1.16666210e+00 -5.57585716e-01 -3.58017683e-01 -1.61918342e-01
-7.14055955e-01 1.11191079e-01 3.46460730e-01 4.44596767e-01
5.35338283e-01 -1.02447891e+00 -6.08324945e-01 2.75444686e-01
6.63853213e-02 1.99323535e-01 2.44333088e-01 1.00364017e+00
-7.28206217e-01 2.22004894e-02 -1.41615734e-01 -8.66201818e-01
-1.55613267e+00 6.36089504e-01 3.95987511e-01 2.09044918e-01
-9.99117792e-01 8.53628218e-01 9.96009231e-01 1.29177600e-01
4.92754072e-01 -2.53033340e-01 8.93828347e-02 -3.44459526e-02
9.71674144e-01 5.29011428e-01 -4.89210971e-02 -6.97890282e-01
-4.98012722e-01 1.39350891e-01 6.95779547e-02 1.95206538e-01
1.11011291e+00 -9.89752412e-02 -3.01618665e-03 5.84725261e-01
8.50632131e-01 -9.47180092e-02 -1.54324830e+00 -3.71057361e-01
1.00475132e-01 -6.98885083e-01 1.21116184e-01 -5.15143454e-01
-1.37855577e+00 9.04228687e-01 7.60767162e-01 7.14644119e-02
1.26314223e+00 -1.96492642e-01 8.10810924e-01 4.21579868e-01
4.03185546e-01 -1.02673435e+00 -1.02912076e-02 3.84943813e-01
2.52883434e-01 -1.47632706e+00 -8.58678818e-02 -5.53792357e-01
-5.08655369e-01 9.54189718e-01 8.00289929e-01 3.29514220e-02
5.81691861e-01 4.36205685e-01 1.63083732e-01 6.22004718e-02
-1.75252855e-01 -7.79279292e-01 7.75393546e-02 7.43448913e-01
-1.38753980e-01 9.89905298e-02 6.80682287e-02 7.47146606e-01
3.97307724e-01 1.68624651e-02 4.95521426e-01 1.03088605e+00
-8.30160499e-01 -5.47938764e-01 -6.01306379e-01 3.90515327e-01
-5.99836946e-01 -1.18277103e-01 -4.03935015e-01 8.27982008e-01
4.72729236e-01 1.10172474e+00 4.33411188e-02 -9.18821692e-02
7.80218840e-02 1.10864654e-01 3.48463804e-01 -5.60018361e-01
-3.23583931e-02 4.04735655e-01 -1.91448227e-01 -6.50570452e-01
-6.38495862e-01 -6.87431037e-01 -1.24714446e+00 -2.46422261e-01
-6.40680254e-01 -2.05286115e-01 4.31840092e-01 8.57297480e-01
-3.61558758e-02 7.10921407e-01 1.57795191e-01 -1.00444102e+00
4.58652563e-02 -7.12235451e-01 -7.22945154e-01 4.92728591e-01
3.35453480e-01 -9.42373574e-01 -4.66009110e-01 3.52216810e-01] | [9.020655632019043, -0.5850842595100403] |
2f8b3a8d-3246-48df-ab3f-226084a5d950 | hprn-holistic-prior-embedded-relation-network | 2112.14608 | null | https://arxiv.org/abs/2112.14608v2 | https://arxiv.org/pdf/2112.14608v2.pdf | HPRN: Holistic Prior-embedded Relation Network for Spectral Super-Resolution | Spectral super-resolution (SSR) refers to the hyperspectral image (HSI) recovery from an RGB counterpart. Due to the one-to-many nature of the SSR problem, a single RGB image can be reprojected to many HSIs. The key to tackle this ill-posed problem is to plug into multi-source prior information such as the natural spatial context-prior of RGB images, deep feature-prior or inherent statistical-prior of HSIs, etc., so as to effectively alleviate the degree of ill-posedness. However, most current approaches only consider the general and limited priors in their customized convolutional neural networks (CNNs), which leads to the inability to guarantee the confidence and fidelity of reconstructed spectra. In this paper, we propose a novel holistic prior-embedded relation network (HPRN) to integrate comprehensive priors to regularize and optimize the solution space of SSR. Basically, the core framework is delicately assembled by several multi-residual relation blocks (MRBs) that fully facilitate the transmission and utilization of the low-frequency content prior of RGBs. Innovatively, the semantic prior of RGB inputs is introduced to mark category attributes, and a semantic-driven spatial relation module (SSRM) is invented to perform the feature aggregation of clustered similar range for refining recovered characteristics. Additionally, we develop a transformer-based channel relation module (TCRM), which breaks the habit of employing scalars as the descriptors of channel-wise relations in the previous deep feature-prior, and replaces them with certain vectors to make the mapping function more robust and smoother. In order to maintain the mathematical correlation and spectral consistency between hyperspectral bands, the second-order prior constraints (SOPC) are incorporated into the loss function to guide the HSI reconstruction. | ['Qian Du', 'Yunsong Li', 'Rui Song', 'Jiaojiao Li', 'Chaoxiong Wu'] | 2021-12-29 | null | null | null | null | ['spectral-super-resolution'] | ['computer-vision'] | [ 5.97181261e-01 -4.16268110e-01 1.80527329e-01 -3.53603035e-01
-6.21316671e-01 -2.96959937e-01 2.67946482e-01 -4.02700514e-01
-1.57074258e-01 6.25450373e-01 8.78957435e-02 8.83494597e-03
-6.29877627e-01 -1.14335215e+00 -5.06822169e-01 -1.22682178e+00
3.42073888e-01 -4.89919305e-01 -9.25743505e-02 -4.32721734e-01
-2.83211842e-03 5.56827724e-01 -1.56072056e+00 1.17708087e-01
1.45861101e+00 1.36951315e+00 6.37457728e-01 7.23072421e-03
-2.53850549e-01 4.04671818e-01 -2.08028734e-01 2.63413112e-03
4.45014060e-01 -4.94343758e-01 -4.46392328e-01 3.41842055e-01
-1.37747824e-01 -2.57687718e-01 -3.68546158e-01 1.60016787e+00
3.90679181e-01 2.96473235e-01 3.91414970e-01 -1.09667242e+00
-7.90231824e-01 2.53135502e-01 -7.38339067e-01 -2.62399644e-01
1.88005388e-01 3.40024203e-01 7.72897065e-01 -8.09707046e-01
1.26177803e-01 8.70773673e-01 6.96173549e-01 -1.52428776e-01
-1.11780405e+00 -6.84663534e-01 8.61355215e-02 2.76234984e-01
-1.74264872e+00 -2.47444496e-01 1.11088550e+00 -1.03539929e-01
3.21766913e-01 4.19072926e-01 8.76914918e-01 7.21394777e-01
-3.31538022e-01 3.08583021e-01 1.20756674e+00 -3.12669069e-01
-1.35345668e-01 1.40471514e-02 -1.23294108e-01 3.71847153e-01
2.01568857e-01 1.10110424e-01 -3.21638674e-01 3.08077067e-01
9.02324319e-01 2.46968016e-01 -8.40276182e-01 -1.06302783e-01
-1.01727057e+00 6.11721635e-01 9.32674646e-01 2.76578367e-01
-4.39133376e-01 -4.08814341e-01 -7.89323077e-02 -7.76992217e-02
2.90647656e-01 2.41356999e-01 -2.73297101e-01 4.67240989e-01
-8.70331824e-01 -2.33199000e-01 1.28190413e-01 1.02610016e+00
1.40955889e+00 1.20493829e-01 -1.74455702e-01 8.78702700e-01
5.19028664e-01 4.82578933e-01 3.03112596e-01 -8.93379509e-01
4.08400297e-01 8.91480088e-01 8.91821682e-02 -1.21204770e+00
-3.82333815e-01 -8.49728048e-01 -1.34468806e+00 -6.68638125e-02
-2.06700072e-01 2.54157931e-01 -8.85009527e-01 1.65542555e+00
3.76842648e-01 2.25554511e-01 8.44130963e-02 1.24120903e+00
8.10120165e-01 8.05428982e-01 -1.24590121e-01 -4.61885810e-01
1.25192630e+00 -5.21719158e-01 -6.77784324e-01 -8.04974213e-02
5.51942103e-02 -6.55995369e-01 1.31186306e+00 2.30224907e-01
-7.30472922e-01 -7.41000354e-01 -1.15455258e+00 -1.42231688e-01
-5.35263598e-01 3.42512488e-01 7.01075912e-01 4.28720564e-01
-7.97796488e-01 5.30068517e-01 -4.60973382e-01 -6.19336180e-02
3.71529430e-01 1.57709479e-01 -2.85837680e-01 -2.39572391e-01
-1.45097804e+00 7.12931395e-01 7.05054760e-01 9.09748077e-01
-4.54804838e-01 -7.60529935e-01 -7.14179754e-01 8.52063000e-02
4.27325249e-01 -5.19759655e-01 2.34483138e-01 -1.01728535e+00
-1.57360971e+00 5.39110601e-01 2.09665541e-02 3.02875787e-01
1.76868141e-01 1.57692030e-01 -7.49490917e-01 3.21185231e-01
3.53429243e-02 2.19875023e-01 7.76451409e-01 -1.47988462e+00
-4.74233240e-01 -1.57928914e-01 1.22708455e-01 4.08491850e-01
-5.94720304e-01 -2.38937765e-01 -6.23181820e-01 -6.10547602e-01
6.71022356e-01 -5.13052166e-01 -1.72416449e-01 -2.40299720e-02
-5.32205164e-01 3.60400230e-01 5.47614574e-01 -9.13899660e-01
1.08323908e+00 -2.51776385e+00 2.93446749e-01 4.42156523e-01
-8.35293159e-02 1.31543115e-01 -1.67649209e-01 1.49274841e-01
-3.83582473e-01 8.09567496e-02 -8.70587409e-01 1.43582318e-02
-6.87014014e-02 3.37725729e-01 -2.24069268e-01 5.88018894e-01
3.70892733e-01 6.00884795e-01 -7.76905775e-01 -3.92495275e-01
4.24489796e-01 6.77308261e-01 -3.04198474e-01 3.90719622e-01
-8.08554590e-02 8.06784332e-01 -6.25991583e-01 6.63101017e-01
1.40634596e+00 -2.21830338e-01 -1.92823321e-01 -1.06787586e+00
-5.17272413e-01 -3.56458910e-02 -1.53577054e+00 2.03740621e+00
-4.21198666e-01 -2.18271047e-01 5.72029471e-01 -9.20103967e-01
1.00464928e+00 6.93724528e-02 6.56085134e-01 -6.63160503e-01
-4.35819626e-02 2.63561606e-01 -3.93664241e-01 -5.85330784e-01
3.69839191e-01 -2.06289351e-01 3.76357108e-01 -1.59990937e-01
-2.13609233e-01 -2.87297130e-01 -2.73596406e-01 -1.05127066e-01
4.06083018e-01 5.10032892e-01 5.93649969e-02 -2.63154715e-01
1.00298488e+00 4.16599289e-02 8.47932339e-01 2.99212247e-01
1.96162477e-01 8.02694798e-01 1.55917317e-01 -8.43669176e-02
-8.36013734e-01 -9.53415155e-01 -4.97900754e-01 6.88806176e-01
5.56596994e-01 7.47235343e-02 -4.56050217e-01 -1.55298308e-01
-1.46629840e-01 5.72882652e-01 -4.07917947e-01 -3.27788740e-01
-1.62711933e-01 -1.21643436e+00 2.13698253e-01 1.93115473e-01
1.17595017e+00 -8.26128483e-01 -1.64549604e-01 1.54365703e-01
-3.52981776e-01 -9.63586032e-01 -7.41937608e-02 3.02498668e-01
-5.85694134e-01 -1.05401623e+00 -4.89383370e-01 -3.41814518e-01
6.00058973e-01 4.96098489e-01 4.80739146e-01 5.86657971e-02
-2.40309447e-01 5.35295065e-03 -6.76807106e-01 1.50056586e-01
1.73794046e-01 -1.19659774e-01 -3.23991925e-01 4.50336218e-01
-1.59259737e-02 -8.73279572e-01 -7.12137043e-01 2.44594023e-01
-1.38948488e+00 3.01145673e-01 7.55361736e-01 9.91827250e-01
7.23584294e-01 5.67652225e-01 2.12139860e-01 -3.86744827e-01
1.60874575e-01 -5.48966825e-01 -7.02112615e-01 4.86707687e-01
-5.79941034e-01 -2.88366258e-01 7.96558380e-01 -2.16260031e-01
-1.42901695e+00 1.85747713e-01 -1.88209429e-01 -5.02938151e-01
-1.88908800e-01 7.64269590e-01 -8.90844703e-01 -4.38919485e-01
2.84286976e-01 7.62361825e-01 -6.00459203e-02 -5.31947017e-01
4.89152610e-01 6.01638377e-01 7.70710707e-01 -5.30203938e-01
1.13444221e+00 5.39756596e-01 2.18023136e-01 -6.56391680e-01
-9.48946297e-01 -4.53257889e-01 -6.56961679e-01 -1.41096354e-01
9.16969001e-01 -1.05498672e+00 -7.04370737e-01 5.86189747e-01
-9.55177605e-01 -1.70111954e-01 -1.04803152e-01 4.47081208e-01
-2.29839563e-01 5.81779182e-01 -4.30113673e-01 -7.08295584e-01
-5.70989810e-02 -1.22902572e+00 9.78056014e-01 5.81446886e-01
7.92098582e-01 -6.69936776e-01 -4.60989922e-01 2.01326504e-01
6.63109481e-01 3.15157771e-01 7.95702398e-01 2.11741909e-01
-8.09306562e-01 3.02102357e-01 -8.73394251e-01 6.52774036e-01
4.14373398e-01 -2.51053646e-02 -1.16698742e+00 -1.79315850e-01
3.30455124e-01 -8.95926729e-02 9.67445016e-01 3.51724654e-01
1.48159242e+00 -1.92639887e-01 -2.04829480e-02 1.46518886e+00
2.01451850e+00 1.26263633e-01 9.99488533e-01 5.29523015e-01
9.82541203e-01 7.92657793e-01 6.42389596e-01 5.84318221e-01
2.97340244e-01 3.05413276e-01 9.10486162e-01 -4.90064830e-01
3.10258325e-02 -3.38925794e-02 3.52672003e-02 6.10425472e-01
-3.53578061e-01 -1.09262474e-01 -3.87608707e-01 1.39173359e-01
-1.59793913e+00 -8.11363339e-01 -3.45753491e-01 2.04057717e+00
8.85770500e-01 -3.12683135e-01 -4.98117447e-01 1.93903103e-01
9.84234691e-01 3.29692930e-01 -6.51713192e-01 6.01593852e-01
-5.99379361e-01 1.71707213e-01 5.78264952e-01 3.27270687e-01
-1.03713274e+00 7.46277690e-01 4.82637453e+00 1.14468098e+00
-1.25427330e+00 9.97144207e-02 4.17482913e-01 4.68134612e-01
-7.06723750e-01 2.26275012e-01 -3.86516035e-01 4.32232827e-01
2.04817444e-01 2.58295953e-01 9.72291052e-01 4.20181274e-01
3.55257213e-01 -2.13946342e-01 -3.34973395e-01 1.12968576e+00
-2.23418504e-01 -9.65111911e-01 6.01091841e-03 8.95065740e-02
6.08195186e-01 -2.38322854e-01 1.89275056e-01 -6.54028654e-02
-7.76936486e-02 -8.78029108e-01 7.51918674e-01 9.79213595e-01
1.21965599e+00 -7.39300370e-01 6.83721721e-01 1.62306264e-01
-1.61179924e+00 -1.71253145e-01 -7.25863039e-01 2.85337001e-01
7.44885802e-02 9.22148705e-01 -6.18711486e-02 1.50645542e+00
7.73949206e-01 9.35499668e-01 -4.14726555e-01 9.11051333e-01
-3.44153136e-01 1.85006738e-01 -3.28656167e-01 6.91873074e-01
2.61863798e-01 -9.68650460e-01 3.55650157e-01 9.40198660e-01
6.31195724e-01 5.35499811e-01 1.74339756e-01 1.33607507e+00
2.38467917e-01 -9.30875726e-03 -1.48805857e-01 7.88976029e-02
4.74197388e-01 1.69304585e+00 -5.55883586e-01 2.01323777e-02
-5.12532949e-01 9.87138212e-01 -1.12296812e-01 8.15075099e-01
-8.63594890e-01 -4.43385541e-01 4.87568915e-01 -1.10233776e-01
2.77878910e-01 -1.35950699e-01 -2.88845718e-01 -1.13324285e+00
1.30674273e-01 -7.77090907e-01 2.74779946e-01 -1.25135469e+00
-1.48658013e+00 5.81873655e-01 -1.21229649e-01 -1.55489385e+00
5.78061938e-01 -6.18195534e-01 -5.55008888e-01 1.54158485e+00
-2.24066472e+00 -1.59229457e+00 -1.05961263e+00 1.04445195e+00
-1.00362204e-01 1.63056165e-01 6.86661363e-01 4.47874010e-01
-9.90936935e-01 6.78229108e-02 1.21048845e-01 -9.24009383e-02
7.21219540e-01 -8.52946758e-01 -5.90394020e-01 1.10539091e+00
-6.14855230e-01 7.82812893e-01 4.20510322e-01 -4.89131778e-01
-1.51834261e+00 -1.32777929e+00 -5.84349222e-02 3.23003113e-01
6.56056106e-01 1.35577217e-01 -1.20408106e+00 3.93549711e-01
-8.22005346e-02 2.83813745e-01 4.79652494e-01 -2.91352034e-01
-3.33786786e-01 -7.00874865e-01 -1.11092162e+00 2.41569355e-01
1.02155018e+00 -7.66454339e-01 -2.25183412e-01 3.43672663e-01
9.33649600e-01 -2.91927814e-01 -1.09946465e+00 7.74276495e-01
2.25746647e-01 -1.22301626e+00 1.22625673e+00 -5.29422984e-02
4.25836027e-01 -9.23253238e-01 -4.42183942e-01 -1.14295709e+00
-5.96797347e-01 -3.17631185e-01 3.57289314e-01 1.61594427e+00
5.97852245e-02 -8.59887302e-01 3.83266032e-01 3.55825096e-01
-6.73871100e-01 -4.47922170e-01 -7.89761364e-01 -5.90498447e-01
-3.88043374e-01 -3.58027488e-01 1.12336290e+00 1.21838641e+00
-5.05575657e-01 -2.52498180e-01 -1.78851292e-01 8.52811754e-01
6.48983955e-01 3.12329024e-01 3.13645780e-01 -1.16823626e+00
-8.87187123e-02 -6.29126668e-01 7.10102245e-02 -8.45067203e-01
6.27454044e-03 -8.00792873e-01 1.70285612e-01 -1.65897918e+00
1.54201791e-01 -7.96581686e-01 -6.14142716e-01 5.03579021e-01
-3.32619518e-01 1.50244772e-01 -4.77711931e-02 4.35331672e-01
-1.98097333e-01 1.01012361e+00 1.52813482e+00 -1.53057098e-01
-2.09359542e-01 -1.45916000e-01 -7.93401361e-01 5.32962382e-01
6.03628993e-01 -1.36733398e-01 -4.11383003e-01 -4.57751662e-01
3.67289037e-01 7.05054253e-02 7.10787356e-01 -9.33084130e-01
4.05278206e-01 -5.35280049e-01 5.17744541e-01 -6.88205779e-01
4.11318988e-01 -1.06226468e+00 4.67665821e-01 1.50837228e-02
1.61789373e-01 -5.60911834e-01 1.23451538e-02 4.20672596e-01
-4.03816789e-01 -1.69891149e-01 8.50154400e-01 -2.76014209e-01
-8.26115012e-01 5.98054230e-01 2.49732465e-01 -4.12559897e-01
5.98825812e-01 -3.78581822e-01 -2.16914982e-01 -1.10206120e-02
-3.61282349e-01 2.31232092e-01 6.39666140e-01 -2.40114778e-01
6.24829531e-01 -1.36453116e+00 -4.96137381e-01 4.35368299e-01
2.22805575e-01 5.16811192e-01 7.85713673e-01 9.82698262e-01
-5.21806777e-01 -4.63623703e-02 -2.91490167e-01 -5.67250013e-01
-4.13112015e-01 5.64967930e-01 5.14908433e-01 -5.56362607e-02
-6.52901649e-01 7.65918434e-01 1.65429384e-01 -4.55077738e-01
-6.04617707e-02 -1.80015862e-01 -4.64299798e-01 1.56757638e-01
3.63836020e-01 2.43155196e-01 -1.68135334e-02 -7.68880010e-01
-2.30080053e-01 7.49863923e-01 6.62646472e-01 5.67670688e-02
1.50062573e+00 -4.56869364e-01 -7.80202031e-01 1.08011700e-01
1.14478660e+00 7.44798258e-02 -1.46234477e+00 -3.25742543e-01
-4.84537721e-01 -5.27968705e-01 4.02721137e-01 -6.76305652e-01
-1.25685334e+00 8.29017341e-01 4.41036761e-01 2.39186630e-01
1.85399079e+00 -4.05942947e-01 5.61561763e-01 2.00196244e-02
2.78493851e-01 -9.89597261e-01 -2.36614019e-01 4.03125644e-01
8.75042617e-01 -1.03586698e+00 1.56204581e-01 -7.71810889e-01
-3.57006282e-01 1.30034566e+00 5.21460354e-01 2.08963618e-01
5.15511632e-01 -4.76364940e-02 -2.50678927e-01 -2.07779240e-02
3.22910547e-01 -5.35567701e-01 3.64713877e-01 6.56728148e-01
1.03821345e-01 -1.48440138e-01 3.15293856e-02 9.06394064e-01
5.03948592e-02 -1.63922340e-01 1.81833550e-01 4.37415272e-01
-4.17923748e-01 -8.32016587e-01 -7.21213996e-01 1.79570049e-01
1.96149096e-01 -4.29834455e-01 2.51426786e-01 5.28793156e-01
7.07849741e-01 1.09472382e+00 -2.05662504e-01 -5.18638313e-01
2.33235151e-01 -2.15055883e-01 1.05204672e-01 -4.42783475e-01
-1.82988793e-01 5.11529326e-01 -3.71233761e-01 -5.74039638e-01
-9.35738504e-01 -3.49581361e-01 -1.16830969e+00 -1.42097831e-01
-5.53369641e-01 7.47988820e-02 5.58509707e-01 9.36394393e-01
5.65961152e-02 7.52421260e-01 1.03366435e+00 -9.88288581e-01
-3.20387840e-01 -9.74514902e-01 -1.15358829e+00 2.15001702e-01
3.48857611e-01 -7.67047405e-01 -5.59538007e-01 -5.37353493e-02] | [10.262887001037598, -1.9799680709838867] |
43aee0e9-b2ac-452f-b092-9573d390bc60 | reinforcement-learning-for-survival-a | 2207.08040 | null | https://arxiv.org/abs/2207.08040v2 | https://arxiv.org/pdf/2207.08040v2.pdf | Reinforcement Learning For Survival, A Clinically Motivated Method For Critically Ill Patients | There has been considerable interest in leveraging RL and stochastic control methods to learn optimal treatment strategies for critically ill patients, directly from observational data. However, there is significant ambiguity on the control objective and on the best reward choice for the standard RL objective. In this work, we propose a clinically motivated control objective for critically ill patients, for which the value functions have a simple medical interpretation. Further, we present theoretical results and adapt our method to a practical Deep RL algorithm, which can be used alongside any value based Deep RL method. We experiment on a large sepsis cohort and show that our method produces results consistent with clinical knowledge. | ['Thesath Nanayakkara'] | 2022-07-17 | null | null | null | null | ['clinical-knowledge'] | ['miscellaneous'] | [ 1.82172313e-01 1.63920373e-01 -4.42023277e-01 -9.07633528e-02
-6.59644604e-01 -2.96672225e-01 -1.57074139e-01 3.18319231e-01
-5.42577922e-01 1.27635646e+00 3.57021213e-01 -5.37288904e-01
-7.78316617e-01 -3.10659707e-01 -3.69385779e-01 -8.79177153e-01
-4.04673725e-01 6.86333120e-01 -3.64689201e-01 -7.00919032e-02
1.37415871e-01 5.33143997e-01 -5.85358381e-01 9.47443470e-02
6.56554997e-01 9.41330850e-01 -1.14947520e-01 7.88548589e-01
4.13242370e-01 1.18290126e+00 -2.61521399e-01 1.35390818e-01
3.25538844e-01 -7.50742853e-01 -7.62379289e-01 -1.55636549e-01
-3.42963099e-01 -5.77290893e-01 -9.04016420e-02 4.17172462e-01
1.25759673e+00 4.18519557e-01 7.46197939e-01 -9.62106645e-01
-1.01841576e-01 5.49274802e-01 -2.39764139e-01 4.11466956e-01
-4.41263728e-02 6.63576007e-01 9.20697808e-01 -1.50125697e-01
3.07030708e-01 1.21760237e+00 6.95044041e-01 8.68104219e-01
-1.25726295e+00 -2.92724758e-01 1.88201845e-01 -2.86341995e-01
-8.02745461e-01 -3.72889012e-01 3.83712560e-01 -5.32341719e-01
8.90472293e-01 -1.26415238e-01 8.47374022e-01 1.14513922e+00
6.36900961e-01 6.89976513e-01 1.14163613e+00 -3.75302792e-01
5.40317416e-01 -3.79029155e-01 -1.29334062e-01 5.81795156e-01
3.11421484e-01 6.98889434e-01 -1.34979039e-01 -3.08779418e-01
9.62382138e-01 2.46460065e-01 -5.76510131e-01 -7.03394651e-01
-1.19995987e+00 1.37735534e+00 1.15201607e-01 -2.20028967e-01
-9.33139324e-01 7.71275103e-01 6.04138851e-01 1.16572283e-01
3.41228873e-01 7.82459915e-01 -8.93245101e-01 -3.45471233e-01
-4.99533713e-01 4.41545993e-01 8.96592617e-01 3.85325044e-01
-2.36361980e-01 1.49809957e-01 -4.98964667e-01 3.83117706e-01
2.35221505e-01 6.18734062e-01 3.05299461e-01 -1.79180753e+00
2.02770948e-01 -1.32304236e-01 5.42357981e-01 -5.28443694e-01
-9.25181448e-01 -4.95154828e-01 -7.68139958e-01 2.45390639e-01
3.44599873e-01 -9.21368241e-01 -2.62308210e-01 1.80671775e+00
3.56889404e-02 1.49610296e-01 3.36904645e-01 9.15565133e-01
5.84363863e-02 1.07748844e-01 2.16362894e-01 -6.47137284e-01
8.49518657e-01 -7.07244635e-01 -7.63946414e-01 -7.71440640e-02
7.81843424e-01 -1.86970621e-01 9.56734836e-01 6.62164271e-01
-1.34711528e+00 2.57793307e-01 -4.93570566e-01 5.68797171e-01
4.08615828e-01 -3.09294581e-01 7.03492403e-01 3.83287281e-01
-1.03682053e+00 9.08359647e-01 -9.33882475e-01 -4.71874982e-01
5.75632393e-01 4.85352248e-01 2.55466551e-01 5.95172793e-02
-1.28066635e+00 1.07857478e+00 3.30328107e-01 1.23767659e-01
-1.02434874e+00 -9.03856039e-01 -5.50584197e-01 1.61837682e-01
6.03727937e-01 -1.29664767e+00 1.68534124e+00 -5.53246498e-01
-1.83840752e+00 3.73473942e-01 2.91924417e-01 -8.93851101e-01
7.64308453e-01 -1.74673185e-01 2.57085174e-01 4.38419372e-01
-2.67811745e-01 3.88206571e-01 5.32297671e-01 -1.16852069e+00
-4.79276210e-01 6.33714721e-02 1.03324816e-01 2.01199725e-01
1.70598432e-01 8.52492973e-02 6.29149914e-01 -6.63476825e-01
-4.98647451e-01 -9.42592144e-01 -8.48848104e-01 -1.62154380e-02
-1.14354966e-02 2.68498249e-02 9.01722759e-02 -3.36803317e-01
1.21332228e+00 -1.63747680e+00 1.76922783e-01 6.10820502e-02
4.89735514e-01 -7.31053576e-02 3.69044282e-02 5.23147106e-01
-3.55126522e-02 2.13234648e-01 -4.36844200e-01 5.82915545e-02
-2.47240476e-02 5.02347291e-01 -4.06556964e-01 6.51220143e-01
2.79141337e-01 1.21638691e+00 -1.28862214e+00 -5.10247648e-01
2.17833966e-01 8.92601609e-02 -1.01118505e+00 5.56078434e-01
-2.57185876e-01 7.71092415e-01 -9.30813313e-01 1.33532435e-01
1.06746346e-01 -4.11970615e-01 4.12330031e-01 4.26201105e-01
3.58101763e-02 -1.31680802e-01 -5.88347554e-01 1.30727482e+00
-5.86329877e-01 2.61474490e-01 7.68868029e-02 -1.31570411e+00
4.44993168e-01 4.62725967e-01 1.12036419e+00 -1.14743330e-01
4.77548540e-01 -3.56782712e-02 1.99381366e-01 -8.44291985e-01
-2.96033055e-01 -1.14253247e+00 2.30026878e-02 6.61872447e-01
-3.33609164e-01 -3.68526071e-01 -2.52082378e-01 -1.61470771e-01
1.39560783e+00 9.20291990e-02 6.67367101e-01 -6.73316240e-01
5.60018644e-02 1.55962296e-02 4.46483582e-01 1.03835046e+00
-5.70708215e-01 5.57587981e-01 8.96537364e-01 -4.27845240e-01
-6.60734892e-01 -1.10068023e+00 -1.43997103e-01 7.49035895e-01
-3.45261306e-01 2.93261796e-01 -3.68180931e-01 -6.91552997e-01
4.16942745e-01 8.50775540e-01 -7.52357066e-01 -4.85553175e-01
-8.84166062e-01 -1.03629911e+00 4.04181123e-01 5.96246421e-01
-1.92267448e-01 -1.18631208e+00 -1.25673723e+00 6.34522915e-01
-1.32661089e-01 -7.21099615e-01 -5.70370078e-01 4.59058940e-01
-1.18456411e+00 -1.15777540e+00 -8.80844235e-01 -2.54428655e-01
3.97846431e-01 -2.39939809e-01 1.24366665e+00 -3.16909179e-02
-3.03533882e-01 8.59739125e-01 -1.04343176e-01 -7.96296120e-01
-3.93233865e-01 -1.33214757e-01 3.50454658e-01 -2.68333405e-01
-1.46235302e-01 -2.27344245e-01 -1.22513223e+00 -1.75279621e-02
-7.07874060e-01 -3.24410349e-01 3.91008258e-01 9.74810362e-01
4.45495367e-01 -2.25036412e-01 1.15407360e+00 -8.20135295e-01
1.13446450e+00 -7.32311845e-01 -5.17637193e-01 -8.17763507e-02
-1.08482575e+00 3.91935527e-01 7.64593482e-01 -4.92662102e-01
-7.26823211e-01 -1.50480554e-01 1.07316785e-01 -6.84773266e-01
2.34734327e-01 6.52556837e-01 4.83868450e-01 3.13752770e-01
4.74278748e-01 -2.21366078e-01 3.83058488e-01 -2.93702513e-01
3.35516155e-01 2.28450477e-01 8.55367780e-02 -8.51071775e-01
-3.85774970e-02 2.18814880e-01 4.38620567e-01 -2.64578134e-01
-9.70997870e-01 -1.63487196e-01 -3.01264673e-01 -1.10420950e-01
9.52944219e-01 -6.39612257e-01 -1.42891490e+00 -6.58359230e-02
-1.01565623e+00 -1.02899218e+00 -1.01053524e+00 8.70147943e-01
-1.29446554e+00 1.23401254e-01 -5.83039701e-01 -1.10023606e+00
-3.44028026e-01 -9.86631036e-01 8.38495255e-01 -2.06279978e-01
-2.83755511e-01 -1.51852405e+00 4.91419047e-01 -1.66304067e-01
5.75272322e-01 6.33366644e-01 9.91776228e-01 -5.07291377e-01
-2.34492406e-01 1.41428143e-01 -1.78776216e-02 4.27993804e-01
7.63398111e-02 -2.43131027e-01 -4.81515795e-01 -5.08807778e-01
4.19926256e-01 -7.04551101e-01 7.63195634e-01 1.15994573e+00
1.21636486e+00 -4.73708510e-01 -2.05649421e-01 5.34690797e-01
1.46788931e+00 4.93385553e-01 1.88707754e-01 1.42602772e-01
3.13569248e-01 5.93755782e-01 3.92540753e-01 1.04635692e+00
2.53645778e-01 1.99332446e-01 3.91693234e-01 -7.82416686e-02
4.39112216e-01 -1.32899471e-02 1.34313568e-01 6.57549620e-01
-2.24005386e-01 -6.19821548e-01 -1.03059578e+00 2.21647426e-01
-2.12486839e+00 -8.56406629e-01 2.62981772e-01 2.21022105e+00
9.22070146e-01 -4.26291302e-02 1.65831074e-01 -2.20446572e-01
2.71882474e-01 -1.09142825e-01 -7.62722969e-01 -7.51378775e-01
7.65852705e-02 4.97976929e-01 7.73734748e-01 4.59938914e-01
-8.36484492e-01 4.75324005e-01 8.29604435e+00 3.98657084e-01
-9.76750791e-01 5.14695160e-02 8.77115667e-01 -3.67334992e-01
1.14494279e-01 -3.20768714e-01 -2.34897435e-01 1.18778639e-01
1.31442690e+00 -4.64828700e-01 6.63615406e-01 3.72107089e-01
1.08192241e+00 -6.25100508e-02 -1.40839338e+00 7.83493161e-01
-4.49531555e-01 -1.17480254e+00 -3.48447621e-01 5.82034653e-03
7.56949246e-01 -6.13660105e-02 -7.84120616e-03 3.25543672e-01
9.22248721e-01 -1.32413554e+00 2.57642806e-01 9.05458927e-01
6.97682321e-01 -7.30496049e-01 9.59501803e-01 2.76166588e-01
-4.56572235e-01 -3.75361830e-01 -8.57022181e-02 -1.79678485e-01
1.93049476e-01 3.63952279e-01 -8.22430849e-01 6.55055344e-02
1.63249597e-01 6.70906544e-01 3.21163267e-01 8.92614663e-01
-1.42073169e-01 7.60985255e-01 -7.98132867e-02 -1.44209772e-01
3.74948263e-01 -1.09819863e-02 3.52800876e-01 1.12043655e+00
2.14422539e-01 6.19636059e-01 3.68513763e-01 7.93778896e-01
4.70604189e-02 1.72804490e-01 -7.29802728e-01 -1.22710548e-01
-1.97345793e-01 7.17721283e-01 -4.82989699e-01 -1.28303006e-01
-1.02886595e-01 4.74628180e-01 1.29470170e-01 4.95371431e-01
-8.03040028e-01 -3.22836312e-03 6.82652056e-01 1.74084708e-01
6.60660788e-02 -8.18332434e-02 -3.19883466e-01 -1.04481184e+00
-5.39625883e-01 -8.51712048e-01 6.66180432e-01 -7.01854050e-01
-1.44182253e+00 1.05768628e-01 2.11461172e-01 -1.04130805e+00
-5.91097176e-01 -6.80653036e-01 -4.67766762e-01 9.31368709e-01
-1.54360986e+00 -2.15947852e-01 2.23394796e-01 3.29496533e-01
4.39561278e-01 1.90463930e-01 6.70817912e-01 -1.38147920e-01
-5.95237315e-01 1.18342794e-01 4.34224337e-01 -1.70029014e-01
3.95327628e-01 -1.30454135e+00 -3.56804013e-01 2.11011887e-01
-8.78074586e-01 3.07366282e-01 6.64312422e-01 -6.80424154e-01
-1.48159170e+00 -1.09381139e+00 2.90845484e-01 -5.86120605e-01
9.10553277e-01 2.05060765e-01 -5.08240402e-01 5.52549839e-01
2.66300440e-01 8.48859921e-03 5.00642300e-01 -3.27943444e-01
6.41127765e-01 1.51363850e-01 -1.32502568e+00 4.81562346e-01
9.85381544e-01 7.39489421e-02 -5.29462576e-01 5.50160229e-01
8.81171048e-01 -3.01858127e-01 -1.04418373e+00 6.26989007e-01
5.62141776e-01 -4.74360675e-01 9.36958611e-01 -1.31068170e+00
5.15984237e-01 3.81509483e-01 -3.92087828e-03 -1.77576506e+00
-2.65038103e-01 -9.25471604e-01 -2.55770832e-01 3.78426075e-01
8.70879889e-02 -1.02071321e+00 4.37718838e-01 8.29311311e-01
8.25391859e-02 -1.37350821e+00 -9.48802173e-01 -7.26099670e-01
8.00649941e-01 -2.12338060e-01 1.74164921e-01 7.20862091e-01
1.23442762e-01 2.07636923e-01 -3.80199820e-01 -1.37100697e-01
7.21447587e-01 9.97761488e-02 9.96559039e-02 -1.02370203e+00
-5.14059603e-01 -7.24794269e-01 2.60946810e-01 -7.22277284e-01
2.46375680e-01 -8.26497555e-01 2.19880834e-01 -1.57362568e+00
2.83117086e-01 -5.71469784e-01 -7.52128005e-01 3.94817322e-01
-1.90983027e-01 -4.96185005e-01 1.93964705e-01 -2.52008419e-02
-4.87624019e-01 7.60555983e-01 1.44910538e+00 1.65386498e-01
-4.97265816e-01 8.25581104e-02 -7.21195400e-01 5.80837786e-01
1.13656104e+00 -7.34896123e-01 -6.45330489e-01 -1.32829949e-01
2.25418925e-01 8.32174838e-01 2.99614251e-01 -4.52853531e-01
-2.74963051e-01 -9.35718894e-01 1.36214823e-01 -1.53435534e-02
-1.93175465e-01 -7.43484735e-01 -1.10403560e-01 1.08988273e+00
-8.75701010e-01 2.61842668e-01 2.86379814e-01 6.16958082e-01
3.45430404e-01 -2.60473549e-01 1.05592716e+00 -3.80289882e-01
1.47121459e-01 5.43470263e-01 -5.67989290e-01 9.86813784e-01
9.51369882e-01 4.35748369e-01 -1.41983241e-01 -6.35417879e-01
-1.06154370e+00 6.11821115e-01 1.02247775e-01 -8.17288831e-02
8.73225510e-01 -1.05642378e+00 -1.00674319e+00 -1.83440670e-01
-2.97041982e-01 -2.33182058e-01 -6.95154741e-02 1.13892102e+00
-6.91019475e-01 6.01975203e-01 1.83238655e-01 -2.80004025e-01
-3.53474766e-01 9.70910430e-01 8.95969450e-01 -5.37329495e-01
-7.55431890e-01 3.10128570e-01 3.82972598e-01 -5.02531156e-02
1.62283465e-01 -7.85390675e-01 4.01321016e-02 -7.92983621e-02
2.06348404e-01 6.03193045e-01 -4.31301832e-01 2.24324930e-02
-3.84306371e-01 4.13540572e-01 5.25338292e-01 -1.78863525e-01
1.46021008e+00 -2.54884183e-01 7.72442073e-02 5.63185334e-01
1.04636753e+00 -3.95252436e-01 -1.43350267e+00 1.04838036e-01
2.20156498e-02 -1.74182773e-01 2.85050988e-01 -9.27131355e-01
-1.28644371e+00 9.01879966e-01 7.20551074e-01 8.44607502e-02
1.14444935e+00 -2.13424951e-01 7.28614986e-01 5.18670857e-01
3.89546663e-01 -7.87350118e-01 8.30793977e-02 2.54713655e-01
1.03473222e+00 -1.23595095e+00 6.26428192e-03 3.21028531e-01
-9.02758718e-01 1.05710828e+00 1.66313156e-01 -3.39357138e-01
8.31398249e-01 3.78947467e-01 1.01843797e-01 -1.29131734e-01
-1.23448050e+00 -1.18508182e-01 -3.01087856e-01 5.08027077e-01
3.13268870e-01 2.73382872e-01 -5.49342334e-01 6.26743913e-01
2.85113275e-01 7.10712373e-01 7.45993137e-01 1.11647522e+00
-3.32208037e-01 -8.82199407e-01 -1.19990967e-01 7.62964368e-01
-8.63738418e-01 -3.30129594e-01 9.32727456e-02 6.46899223e-01
-2.67491251e-01 9.99857962e-01 -2.34416157e-01 1.45815462e-01
4.96827185e-01 1.43519610e-01 5.14721513e-01 -5.29351771e-01
-5.87115765e-01 4.40619849e-02 3.13920975e-02 -5.99959373e-01
-3.77201587e-01 -8.28655660e-01 -1.57539070e+00 -1.26973987e-01
9.39146504e-02 1.02930255e-01 1.73338413e-01 7.21766770e-01
3.29176694e-01 6.90952480e-01 9.56742644e-01 -5.84256887e-01
-1.20508552e+00 -5.39339066e-01 -6.24047875e-01 7.68279433e-02
8.19769204e-01 -7.10978448e-01 -5.91489911e-01 -2.14886039e-01] | [4.007367134094238, 2.7264153957366943] |
8765cac3-18ee-47b2-84da-248900a33504 | miipher-a-robust-speech-restoration-model | 2303.01664 | null | https://arxiv.org/abs/2303.01664v1 | https://arxiv.org/pdf/2303.01664v1.pdf | Miipher: A Robust Speech Restoration Model Integrating Self-Supervised Speech and Text Representations | Speech restoration (SR) is a task of converting degraded speech signals into high-quality ones. In this study, we propose a robust SR model called Miipher, and apply Miipher to a new SR application: increasing the amount of high-quality training data for speech generation by converting speech samples collected from the Web to studio-quality. To make our SR model robust against various degradation, we use (i) a speech representation extracted from w2v-BERT for the input feature, and (ii) a text representation extracted from transcripts via PnG-BERT as a linguistic conditioning feature. Experiments show that Miipher (i) is robust against various audio degradation and (ii) enable us to train a high-quality text-to-speech (TTS) model from restored speech samples collected from the Web. Audio samples are available at our demo page: google.github.io/df-conformer/miipher/ | ['Michiel Bacchiani', 'Ankur Bapna', 'Wei Han', 'Yu Zhang', 'Nobuyuki Morioka', 'Kohei Yatabe', 'Yifan Ding', 'Shigeki Karita', 'Heiga Zen', 'Yuma Koizumi'] | 2023-03-03 | null | null | null | null | ['speech-enhancement', 'speech-denoising'] | ['speech', 'speech'] | [ 4.28842217e-01 2.29411870e-02 2.06000641e-01 -3.99050087e-01
-1.59914148e+00 -3.84496927e-01 3.42069030e-01 -5.12056053e-01
1.25331417e-01 5.93960643e-01 8.81381035e-01 -5.18437147e-01
1.66727945e-01 -5.30264795e-01 -8.11753988e-01 -5.33526361e-01
2.14736894e-01 -8.95529389e-02 1.00274086e-01 -3.60906273e-01
-9.42642912e-02 3.20957899e-01 -1.82612705e+00 6.78142846e-01
1.19981277e+00 9.28547204e-01 6.97145402e-01 1.01880848e+00
-4.65570437e-03 5.47665417e-01 -9.58250046e-01 4.76425178e-02
2.40705743e-01 -7.33380318e-01 -6.67587578e-01 1.38899788e-01
2.41382763e-01 -2.44972959e-01 -5.36560535e-01 1.03475678e+00
9.25673962e-01 2.97600925e-01 2.69611180e-01 -9.36908722e-01
-8.13701451e-01 7.40537107e-01 9.54248309e-02 3.63319904e-01
8.75890493e-01 1.53552398e-01 8.15996230e-01 -1.16957760e+00
5.97531497e-01 1.49260771e+00 2.85045087e-01 7.24286556e-01
-1.08631229e+00 -6.02612019e-01 -1.05811931e-01 2.85801083e-01
-1.21145403e+00 -1.32480049e+00 8.49142730e-01 -1.14718303e-02
9.24789250e-01 6.42039061e-01 2.86702037e-01 1.45259750e+00
-1.23145478e-02 9.82819557e-01 1.13001335e+00 -6.62659347e-01
2.56397635e-01 -1.46559060e-01 -1.08738035e-01 1.14914238e-01
-5.25141537e-01 3.21202725e-01 -9.93291557e-01 9.75655578e-03
4.38063711e-01 -5.63718081e-01 -8.97433281e-01 5.77267826e-01
-9.87818539e-01 3.85868877e-01 1.40385211e-01 3.78220797e-01
-3.26684356e-01 -2.51641303e-01 1.11209087e-01 8.75106037e-01
6.01492345e-01 -2.54421588e-02 -4.40554529e-01 -4.53793973e-01
-9.80054080e-01 -9.50742662e-02 5.21578133e-01 1.08478773e+00
3.75466496e-01 6.40518129e-01 -1.94188833e-01 1.52267587e+00
3.65605801e-01 9.76263821e-01 8.85672748e-01 -8.52634847e-01
8.44883144e-01 -2.29325473e-01 -6.84664771e-02 -4.46991533e-01
3.51858176e-02 -3.66338342e-01 -6.76317394e-01 2.37861145e-02
-1.14869334e-01 -2.94874161e-02 -1.03737414e+00 1.57447588e+00
8.45583081e-02 1.31259501e-01 3.22059661e-01 8.97007525e-01
1.00387108e+00 1.17802644e+00 -4.70823884e-01 -5.10533750e-01
1.00129366e+00 -8.88868511e-01 -9.60500300e-01 -3.91271085e-01
1.00560315e-01 -1.00142634e+00 1.59630001e+00 7.11716473e-01
-1.17887056e+00 -9.39661205e-01 -9.93414581e-01 -1.34371236e-01
-2.52878666e-02 3.32169145e-01 -2.09965095e-01 7.34032393e-01
-1.46138537e+00 6.76704824e-01 -6.16087377e-01 -2.25268885e-01
1.09254159e-01 -1.97100546e-02 -5.37842095e-01 -1.93916515e-01
-1.30607033e+00 5.47380090e-01 -1.01178586e-01 1.37454107e-01
-1.19000196e+00 -3.37829977e-01 -9.94454145e-01 2.02905238e-02
1.32916182e-01 -3.53949577e-01 1.35626709e+00 -9.09825623e-01
-2.08781433e+00 2.53233045e-01 -4.30733413e-01 -1.19733818e-01
1.76073745e-01 -3.42266440e-01 -1.14356613e+00 3.60767633e-01
1.34991072e-02 1.62996843e-01 1.37830627e+00 -1.45057726e+00
-5.68006516e-01 -3.08867633e-01 -4.26780492e-01 8.22258219e-02
-3.28935832e-01 3.16366285e-01 -3.17451864e-01 -1.05131280e+00
3.26662183e-01 -5.56309223e-01 2.90045023e-01 -3.65462929e-01
-5.59431195e-01 4.67824824e-02 8.20823371e-01 -1.46908379e+00
1.21219099e+00 -2.45257664e+00 5.62110431e-02 4.51904768e-03
-4.32715267e-01 3.96407276e-01 -6.23432577e-01 5.80230534e-01
-2.55791277e-01 1.64461628e-01 -2.77425677e-01 -6.96443379e-01
2.12908890e-02 1.77156925e-01 -6.70395553e-01 2.18682066e-01
2.65528888e-01 4.25704330e-01 -8.63937199e-01 -2.69416481e-01
1.08243607e-01 7.00573385e-01 -3.07189137e-01 6.01955473e-01
1.31939813e-01 5.70660412e-01 -1.29743740e-01 5.58160305e-01
6.43600762e-01 4.73997444e-01 -3.01984027e-02 -1.40064523e-01
1.13823814e-02 1.02666032e+00 -1.14140368e+00 1.62237084e+00
-7.77049065e-01 6.51800931e-01 3.00616235e-01 -4.86320227e-01
1.20864856e+00 6.79201424e-01 -6.90794662e-02 -9.61293638e-01
-1.40191913e-01 3.65613699e-01 -3.07957858e-01 -5.30237019e-01
4.60466832e-01 -1.05404794e-01 2.75631875e-01 2.92598307e-01
2.11991310e-01 -5.70399940e-01 2.06151411e-01 1.31764978e-01
1.19763291e+00 4.47995178e-02 -1.82562277e-01 7.83557538e-03
6.55588865e-01 -7.32328176e-01 6.47399664e-01 3.87921482e-01
1.50544895e-02 1.00378859e+00 -5.27571887e-04 3.61586958e-01
-9.54552650e-01 -1.49533784e+00 -3.06733400e-02 8.34657967e-01
-4.11919266e-01 -4.05452728e-01 -8.85903299e-01 -2.23213255e-01
-3.53878468e-01 9.82279301e-01 4.37609069e-02 -2.94142395e-01
-5.11068523e-01 -1.73430040e-01 6.23270154e-01 2.82668859e-01
2.33432934e-01 -1.22418630e+00 3.37802380e-01 2.84914851e-01
-8.36049855e-01 -1.06966007e+00 -9.58970606e-01 1.90904036e-01
-8.11861753e-01 -5.10187507e-01 -6.11649811e-01 -7.44881630e-01
3.95646244e-01 4.35338020e-01 8.62102270e-01 -6.70802891e-02
1.73904732e-01 2.70659566e-01 -8.81978214e-01 -5.97870238e-02
-1.12037051e+00 -2.29057506e-01 5.64060986e-01 2.55637407e-01
-2.23945871e-01 -8.85073364e-01 -2.89316744e-01 2.97434121e-01
-1.04332256e+00 -1.10246062e-01 3.71493816e-01 6.73340619e-01
5.26689053e-01 1.97479293e-01 1.09206963e+00 -2.31127664e-01
9.22656357e-01 -1.83286190e-01 -3.43612224e-01 -1.00098647e-01
-3.93029422e-01 -1.93632081e-01 9.35004890e-01 -4.66195732e-01
-1.29252064e+00 -2.92964190e-01 -8.55467021e-01 -4.34618950e-01
-1.87129423e-01 5.75479209e-01 -7.44621933e-01 4.37304348e-01
8.11111152e-01 6.05605781e-01 -5.33258133e-02 -1.02056348e+00
4.15757984e-01 1.67090476e+00 9.05562341e-01 -5.74769199e-01
9.37128067e-01 1.20769143e-01 -7.68269956e-01 -1.25759518e+00
-4.73866493e-01 -3.11582893e-01 -3.64737958e-01 -1.85071692e-01
3.06903154e-01 -1.07113540e+00 -1.04504861e-01 5.70306778e-01
-9.70733643e-01 -6.46550834e-01 -4.56669092e-01 5.52136183e-01
-7.10601628e-01 5.98195553e-01 -1.01112604e+00 -1.00428307e+00
-5.25963366e-01 -1.06537008e+00 1.17119277e+00 3.03520020e-02
5.63878529e-02 -3.94654244e-01 -2.46458147e-02 6.88975036e-01
3.35400105e-01 -4.35694307e-01 7.22955763e-01 -3.51332277e-01
-2.04541624e-01 -5.19468486e-02 1.82402790e-01 1.09686732e+00
5.88053226e-01 5.26110940e-02 -1.33809137e+00 -5.26825368e-01
2.61997432e-01 -1.44213185e-01 5.03323376e-01 2.74678022e-01
9.11749423e-01 -4.85251784e-01 2.39904806e-01 5.52306533e-01
9.20357168e-01 3.27156156e-01 1.04460537e+00 9.25368890e-02
2.88983852e-01 4.23225313e-01 6.83170617e-01 3.20081711e-01
2.60706127e-01 8.84218514e-01 8.39248970e-02 5.63946255e-02
-8.96137238e-01 -5.51798999e-01 1.23477542e+00 1.87595737e+00
2.24880531e-01 -4.54025120e-01 -6.11833811e-01 6.58046365e-01
-1.26547062e+00 -9.25512016e-01 -2.04167172e-01 2.23597074e+00
1.29025757e+00 1.15474500e-01 3.04266904e-03 5.93291461e-01
7.24348128e-01 1.85205579e-01 -2.49053538e-01 -2.93435842e-01
-3.46363902e-01 2.54463911e-01 -2.23334670e-01 7.81577110e-01
-5.28360665e-01 8.33176315e-01 6.11268377e+00 1.12437057e+00
-1.13833904e+00 2.54519433e-01 3.72867316e-01 -1.19413517e-01
-5.12626827e-01 -2.44897604e-01 -5.50733864e-01 7.35346377e-01
1.52396619e+00 -2.92481512e-01 8.04980516e-01 5.27313948e-01
9.30871964e-01 1.88763022e-01 -8.30195010e-01 9.42552507e-01
1.78735137e-01 -8.34438026e-01 1.78445742e-01 -9.32828784e-02
4.89956945e-01 1.02330774e-01 7.21697211e-02 2.10613072e-01
1.91971943e-01 -8.30944717e-01 1.09521902e+00 4.41527754e-01
1.27137947e+00 -5.16066790e-01 4.59650427e-01 3.45040321e-01
-1.21906233e+00 -1.94616243e-01 -2.69695014e-01 3.15448999e-01
3.75331163e-01 7.52566040e-01 -7.59001553e-01 9.02487099e-01
8.20566416e-01 6.46351755e-01 -3.82552713e-01 8.17007601e-01
-5.68802893e-01 1.04822719e+00 -1.79018572e-01 6.95778370e-01
-4.13358420e-01 -9.94233787e-02 8.18803728e-01 1.14014149e+00
7.51658618e-01 -4.78692017e-02 -1.54170230e-01 5.94132602e-01
-1.05181485e-01 1.47817163e-02 -4.29510742e-01 -2.28940099e-02
6.76999629e-01 9.06563103e-01 -1.05981223e-01 -1.06396757e-01
-1.66029319e-01 1.08703816e+00 -1.61384389e-01 6.50460780e-01
-3.93018156e-01 -4.59751874e-01 8.10913026e-01 3.77322920e-02
2.70272344e-01 -2.26574972e-01 3.62068005e-02 -1.20669281e+00
3.56085986e-01 -1.41263831e+00 -8.28900039e-02 -1.45378983e+00
-1.24520576e+00 1.13222694e+00 -4.49278563e-01 -1.56061423e+00
-4.66029167e-01 -3.27909708e-01 -4.67642784e-01 1.09974968e+00
-1.58972085e+00 -9.87986147e-01 -7.97447339e-02 8.05519402e-01
1.00603330e+00 -2.70561308e-01 7.41132021e-01 5.21721184e-01
-4.59704638e-01 5.41097641e-01 3.07418734e-01 -4.56085503e-02
7.74250686e-01 -1.06602633e+00 8.34613144e-01 1.31575584e+00
3.52846533e-01 3.77224803e-01 8.04668128e-01 -6.29558861e-01
-1.42201936e+00 -1.10635269e+00 9.86178219e-01 -3.05899054e-01
5.41823983e-01 -4.34781581e-01 -1.04614866e+00 5.39792299e-01
2.16569994e-02 -1.82900861e-01 4.99140888e-01 -2.95272082e-01
-3.04420650e-01 -4.15992767e-01 -1.00834846e+00 5.69317400e-01
1.14274299e+00 -1.20039570e+00 -7.67122447e-01 1.09252870e-01
1.21641541e+00 -3.54734182e-01 -1.01565015e+00 1.38510659e-01
1.28128722e-01 -6.47960782e-01 8.70645702e-01 -9.20835435e-02
2.54696399e-01 -5.63672006e-01 -5.74153841e-01 -1.82813358e+00
3.54417786e-02 -1.05200517e+00 -9.19567570e-02 1.59522402e+00
5.04516125e-01 -6.55553997e-01 9.51153412e-02 -6.20172136e-02
-7.71810770e-01 -1.73902690e-01 -1.26386988e+00 -1.26986992e+00
-1.49310023e-01 -8.07106495e-01 7.43605673e-01 5.53822398e-01
1.75616026e-01 3.55204076e-01 -4.52744871e-01 5.30123174e-01
3.33285809e-01 -2.07669616e-01 7.19421864e-01 -9.09263909e-01
-3.16023499e-01 3.81065793e-02 9.30208564e-02 -1.32186067e+00
5.49430884e-02 -8.85488808e-01 5.40859401e-01 -1.77410495e+00
-4.32165146e-01 -2.11505070e-01 -4.32644784e-03 4.13363069e-01
-1.13704704e-01 2.38084537e-03 4.46442932e-01 2.04807237e-01
1.02954410e-01 9.60993350e-01 1.17146575e+00 -7.86495283e-02
-3.10236216e-01 2.27324978e-01 -6.58212662e-01 2.51910061e-01
6.63706601e-01 -4.91408557e-01 -4.92521644e-01 -4.33405638e-01
-7.54704252e-02 5.79932868e-01 6.97952807e-02 -1.03809500e+00
3.20211053e-04 4.92824726e-02 -5.88070489e-02 -3.99787188e-01
7.00505078e-01 -6.14206135e-01 2.18457565e-01 8.24366137e-02
-1.74341977e-01 -2.65054017e-01 2.00792730e-01 3.60916525e-01
-5.16203046e-01 -1.63881436e-01 6.87014759e-01 3.24758083e-01
-2.27361038e-01 8.65767226e-02 -5.39444447e-01 -1.08302258e-01
5.32094128e-02 -1.04697026e-01 -7.15951145e-01 -8.38204384e-01
-9.07995701e-01 -1.66804060e-01 3.73188585e-01 7.83664525e-01
9.24709976e-01 -1.45928347e+00 -1.01343691e+00 4.78666067e-01
-1.64568853e-02 -2.83417463e-01 3.12661082e-01 4.45474863e-01
-2.40226239e-02 1.68751016e-01 1.36706129e-01 -3.56509686e-01
-1.31462145e+00 3.74607325e-01 2.07039118e-01 2.86827296e-01
-8.12927961e-01 6.13543212e-01 -2.47177660e-01 -5.58117807e-01
1.30126223e-01 -4.61718202e-01 -8.11198279e-02 -1.98509052e-01
8.09024155e-01 3.29225928e-01 4.47629690e-01 -8.15127194e-01
-2.53207475e-01 1.82130262e-01 3.24503332e-01 -6.40195608e-01
1.56166267e+00 -6.17969692e-01 4.54091430e-02 5.02093792e-01
1.14264810e+00 3.76703829e-01 -1.26945496e+00 -2.92999327e-01
-8.23893324e-02 -6.96102083e-01 2.42605761e-01 -8.62439930e-01
-1.01999044e+00 7.54051030e-01 5.71041405e-01 3.55981439e-01
1.50309789e+00 -1.28917411e-01 1.04605114e+00 4.73853946e-02
5.05094945e-01 -1.25260997e+00 1.17339671e-01 4.35196310e-01
1.61502647e+00 -7.54002035e-01 -4.36477035e-01 -4.84054416e-01
-5.83902419e-01 1.04720676e+00 1.21948421e-01 3.22742641e-01
4.92571920e-01 4.72731590e-01 4.89511728e-01 3.81016880e-01
-8.97878826e-01 -3.25369120e-01 1.65540636e-01 1.06836939e+00
3.29786032e-01 -3.11948750e-02 -5.56313479e-03 5.47074020e-01
-6.44048750e-01 -3.08378879e-02 7.30592906e-01 5.93202293e-01
-5.10323048e-01 -1.35839808e+00 -6.93229258e-01 2.25900337e-01
-2.41728857e-01 -3.70089173e-01 -1.60519302e-01 -6.33641481e-02
-3.74484897e-01 1.83881950e+00 -2.95765549e-01 -5.96000671e-01
7.35059559e-01 2.06389114e-01 4.25362438e-02 -6.42444134e-01
-4.12176877e-01 3.97263855e-01 4.74497825e-01 -4.75079924e-01
2.26460863e-02 -6.47434175e-01 -1.07311785e+00 -1.42774120e-01
-2.43396163e-01 2.18414947e-01 8.61314714e-01 7.77853787e-01
4.46091622e-01 8.09907615e-01 1.11762977e+00 -9.43073273e-01
-6.88040078e-01 -1.32612979e+00 -7.59630263e-01 2.90433407e-01
7.26611555e-01 -1.23580039e-01 -8.19326401e-01 5.10485411e-01] | [14.882107734680176, 6.118956565856934] |
ab49cc9a-a752-414a-bec6-4c295db564d7 | network-intrusion-detection-system-in-a-light | 2210.03254 | null | https://arxiv.org/abs/2210.03254v1 | https://arxiv.org/pdf/2210.03254v1.pdf | Network Intrusion Detection System in a Light Bulb | Internet of Things (IoT) devices are progressively being utilised in a variety of edge applications to monitor and control home and industry infrastructure. Due to the limited compute and energy resources, active security protections are usually minimal in many IoT devices. This has created a critical security challenge that has attracted researchers' attention in the field of network security. Despite a large number of proposed Network Intrusion Detection Systems (NIDSs), there is limited research into practical IoT implementations, and to the best of our knowledge, no edge-based NIDS has been demonstrated to operate on common low-power chipsets found in the majority of IoT devices, such as the ESP8266. This research aims to address this gap by pushing the boundaries on low-power Machine Learning (ML) based NIDSs. We propose and develop an efficient and low-power ML-based NIDS, and demonstrate its applicability for IoT edge applications by running it on a typical smart light bulb. We also evaluate our system against other proposed edge-based NIDSs and show that our model has a higher detection performance, and is significantly faster and smaller, and therefore more applicable to a wider range of IoT edge devices. | ['Marius Portmann', 'Siamak Layeghy', 'Liam Daly Manocchio'] | 2022-10-06 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [ 1.82832390e-01 -2.30885878e-01 -4.93536472e-01 6.08801506e-02
5.34759760e-01 -2.85123736e-01 3.28900427e-01 -2.05359221e-01
-1.65283948e-01 5.02338350e-01 -3.86998743e-01 -8.88807952e-01
-2.10584998e-01 -1.00020778e+00 1.67354479e-01 -5.43385029e-01
-2.76196413e-02 1.11709125e-01 6.84746325e-01 -6.97192922e-03
3.71688008e-02 5.66507220e-01 -1.44721901e+00 -2.83665299e-01
3.17137390e-01 1.55441713e+00 -3.19684297e-01 3.90895307e-01
2.14691296e-01 3.84451210e-01 -4.63930070e-01 -2.07737848e-01
4.48529005e-01 -1.98907122e-01 -3.14273924e-01 -4.02226001e-01
-1.89661130e-01 -2.29605168e-01 -4.28420931e-01 1.00906563e+00
8.25101197e-01 -1.82593226e-01 2.96155065e-01 -1.92344272e+00
-5.59081852e-01 5.45245886e-01 -6.83836341e-01 3.46520841e-01
2.71055251e-01 5.08256853e-02 5.70375204e-01 2.25381851e-02
1.62181631e-01 7.54052401e-01 6.26853108e-01 7.64213681e-01
-5.68063021e-01 -1.12111986e+00 -1.23138078e-01 4.91100311e-01
-1.11687803e+00 -4.14813340e-01 1.01878798e+00 3.08399230e-01
1.27605140e+00 2.19562203e-01 6.50463581e-01 1.06064498e+00
5.61907768e-01 4.71059471e-01 1.24242747e+00 -2.93510795e-01
7.06195891e-01 1.11384861e-01 1.19344018e-01 4.72763896e-01
8.11182857e-01 1.40096143e-01 -1.46763191e-01 -1.26509935e-01
3.84614557e-01 4.82269764e-01 8.41298178e-02 -2.93095205e-02
-6.53509796e-01 5.42560577e-01 4.70845342e-01 5.77436209e-01
-2.06592306e-01 1.70762882e-01 5.80236375e-01 1.75033867e-01
1.56935513e-01 -1.58099934e-01 -7.84185946e-01 -3.21474999e-01
-3.85597199e-01 -6.95985317e-01 1.14513540e+00 9.33346987e-01
2.58542478e-01 6.58438951e-02 4.25286114e-01 1.84135303e-01
5.20285189e-01 6.08464122e-01 1.74585983e-01 -4.40166324e-01
8.61475170e-02 8.68200243e-01 -2.12134108e-01 -7.94515729e-01
-9.09632981e-01 -2.59883255e-02 -1.19857764e+00 4.57406133e-01
-1.57180041e-01 -2.54994333e-01 -6.73178196e-01 1.34843326e+00
4.90493000e-01 7.42398143e-01 1.03630558e-01 3.76954019e-01
7.53287017e-01 3.69365752e-01 1.87471613e-01 -3.53377074e-01
1.64794767e+00 -5.55960596e-01 -6.05145097e-01 -2.26512700e-01
4.21156079e-01 -6.34420156e-01 6.29651427e-01 3.69060725e-01
-5.87545991e-01 -4.35784519e-01 -1.39212751e+00 3.45606387e-01
-7.87837446e-01 -4.45225775e-01 1.03426981e+00 1.65248263e+00
-7.94839680e-01 1.55417234e-01 -8.75202417e-01 -8.94098103e-01
7.57919073e-01 8.13199937e-01 1.37501538e-01 1.55904606e-01
-8.92344713e-01 5.17502785e-01 2.55376011e-01 -2.60372877e-01
-2.94230700e-01 -1.70493111e-01 -3.84006232e-01 -4.20350395e-03
3.16675216e-01 -9.04872298e-01 8.86520803e-01 -2.23408908e-01
-1.78580904e+00 4.35537428e-01 2.76858836e-01 -4.18067545e-01
-3.61815020e-02 2.10049063e-01 -1.20004630e+00 -6.73093647e-02
-2.56198466e-01 4.23767477e-01 8.16064894e-01 -8.42760563e-01
-7.78039753e-01 -5.58161974e-01 1.80452600e-01 -4.69718993e-01
-8.73980641e-01 2.54135281e-01 8.13841298e-02 -1.58252284e-01
-1.18695855e-01 -1.20552921e+00 -1.40154466e-01 -7.97619075e-02
-4.27926004e-01 -5.11463344e-01 1.73195791e+00 1.38298601e-01
1.31363308e+00 -2.01929426e+00 -8.65847945e-01 1.84543282e-01
1.57211050e-01 7.45136321e-01 1.91800654e-01 9.31975469e-02
2.35803545e-01 2.75203705e-01 1.95941478e-01 -7.10303336e-02
9.96179059e-02 3.01847219e-01 -7.12464228e-02 4.16458219e-01
-4.64974552e-01 8.30966651e-01 -7.50683665e-01 -2.98533112e-01
7.75093853e-01 7.68501878e-01 -2.65306860e-01 -1.86977610e-01
6.84118271e-02 4.29825932e-01 -7.88518608e-01 1.02749050e+00
6.61325872e-01 -2.43565097e-01 1.58909485e-01 -4.18425739e-01
-1.01395383e-01 3.01869869e-01 -1.22150302e+00 1.14226496e+00
-5.26719570e-01 2.75242776e-01 -3.81038003e-02 -9.78664696e-01
8.50031972e-01 4.84920144e-01 7.66021490e-01 -9.21529353e-01
7.65224636e-01 1.07285291e-01 6.38864562e-02 -3.41368586e-01
-1.70004815e-01 2.14579068e-02 -1.92680210e-01 7.57782996e-01
-4.32552934e-01 1.78064078e-01 5.31236408e-03 -1.00875176e-01
1.75853789e+00 -3.07369977e-01 4.67857748e-01 -1.07752703e-01
5.47103107e-01 -4.83336806e-01 4.53261346e-01 7.78384209e-01
-6.86000407e-01 -3.08584720e-01 -2.09456638e-01 -7.49694049e-01
-5.84899187e-01 -9.50145245e-01 -3.24970990e-01 7.45939493e-01
5.78664541e-01 -4.25355852e-01 -4.86423165e-01 -8.99195194e-01
-1.90926656e-01 5.21245122e-01 9.61276069e-02 -3.17535311e-01
-2.22123936e-01 -1.21842515e+00 6.73327804e-01 5.30447900e-01
1.17124760e+00 -1.13314927e+00 -1.15494823e+00 2.13607579e-01
4.87417459e-01 -1.52763820e+00 1.08373156e-02 2.25530297e-01
-9.67532814e-01 -1.13662100e+00 5.10155439e-01 -9.01304066e-01
4.43659693e-01 4.77528065e-01 9.31156635e-01 2.25329652e-01
-5.51626801e-01 5.36714494e-01 -4.65238988e-01 -8.55327308e-01
-1.10499106e-01 3.85815352e-02 6.78346753e-01 -1.59271508e-01
1.09028733e+00 -1.03696465e+00 -9.49250042e-01 5.49507678e-01
-7.30362415e-01 -4.42044139e-01 7.16725767e-01 2.62004942e-01
1.72243387e-01 1.00463974e+00 7.32593775e-01 -5.40350735e-01
4.97997403e-01 -5.20554483e-01 -7.47902870e-01 -3.04010883e-02
-1.14571548e+00 -3.59160453e-01 8.56929958e-01 -5.83097816e-01
-6.80298805e-01 -1.17781475e-01 -2.58388311e-01 1.81726292e-01
-4.64226872e-01 -3.38327438e-01 -5.27342021e-01 -5.44995844e-01
2.48478845e-01 -3.69277447e-01 -6.33712530e-01 -2.52382606e-01
7.68184811e-02 1.29741085e+00 2.00603813e-01 -1.40567571e-01
8.85581434e-01 8.69197667e-01 4.94792998e-01 -9.43910897e-01
-3.67908478e-01 -4.81813908e-01 -5.85362390e-02 7.02140182e-02
8.07704449e-01 -5.49687505e-01 -1.49962521e+00 3.05149823e-01
-7.23467171e-01 -1.03508838e-01 -1.72797412e-01 4.30285275e-01
-9.25088972e-02 3.14789265e-01 -5.79340637e-01 -9.38942492e-01
-9.91629243e-01 -1.09204435e+00 8.68466794e-01 6.42465651e-01
-2.81319201e-01 -9.65359747e-01 -8.13511312e-02 2.28982791e-01
6.68530226e-01 7.01738638e-04 7.60622919e-01 -6.42936409e-01
-7.28744388e-01 -4.65560138e-01 -3.52541715e-01 1.17050529e-01
5.99696279e-01 -2.92147219e-01 -9.48426962e-01 -4.69234765e-01
5.04521668e-01 1.80415168e-01 3.30444992e-01 1.60976395e-01
9.60219800e-01 -1.31692141e-01 -9.14229810e-01 6.35979712e-01
1.51419914e+00 8.26040745e-01 7.58227050e-01 4.83129144e-01
6.10737026e-01 -3.19835722e-01 6.08506016e-02 4.24644470e-01
2.54719444e-02 4.22810495e-01 7.77698398e-01 2.80974079e-02
6.31962940e-02 1.00161538e-01 3.41890633e-01 7.47409105e-01
1.75326765e-02 -6.99568033e-01 -5.65271437e-01 1.80838123e-01
-1.49043667e+00 -9.00383770e-01 1.67556237e-02 1.86716580e+00
1.44421179e-02 6.71940148e-01 8.04373994e-02 6.02842629e-01
7.05030739e-01 7.89070651e-02 -9.89864528e-01 -5.13457179e-01
3.32791477e-01 6.45830393e-01 6.83023751e-01 -1.49721310e-01
-1.03972185e+00 5.39539337e-01 6.41762543e+00 5.84585547e-01
-1.13815343e+00 3.97973716e-01 3.46563667e-01 2.02388406e-01
3.29518586e-01 4.19764258e-02 -8.60911727e-01 7.12828815e-01
1.01486433e+00 1.73315480e-01 4.88390893e-01 1.26194310e+00
1.48584113e-01 9.36981738e-02 -9.25101817e-01 1.30328941e+00
-1.24439128e-01 -8.46575856e-01 -3.49061579e-01 3.05289775e-01
6.49236023e-01 2.09314674e-01 -1.15233749e-01 2.05645800e-01
3.14754605e-01 -5.89977205e-01 -9.24605057e-02 -3.38673353e-01
6.61426544e-01 -7.78493643e-01 8.55536520e-01 1.69358298e-01
-1.52060032e+00 -4.64412481e-01 -3.80677938e-01 -5.12765586e-01
3.04074973e-01 8.41207385e-01 -6.31186426e-01 2.30285808e-01
9.94373202e-01 4.36947584e-01 -2.22101405e-01 8.02628279e-01
-2.83732921e-01 5.59312880e-01 -7.22262263e-01 -5.10439277e-01
-2.67288804e-01 -6.09067902e-02 3.77306879e-01 7.80840755e-01
4.24162298e-01 1.55067697e-01 3.85289729e-01 3.99179071e-01
-3.18271369e-01 -3.27111065e-01 -7.38310933e-01 9.62993652e-02
7.47329295e-01 1.67560661e+00 -1.35256910e+00 -1.98546976e-01
-9.42954779e-01 9.29065406e-01 -4.70673352e-01 -1.34341300e-01
-8.67185414e-01 -4.33377534e-01 9.10005391e-01 9.41725746e-02
1.21735133e-01 -1.76082551e-01 -4.43878144e-01 -9.28317428e-01
-2.92515397e-01 -6.36980712e-01 4.76804972e-01 -5.07242322e-01
-1.53457725e+00 6.56597793e-01 -2.49590024e-01 -1.17219901e+00
2.26981267e-01 -9.34060335e-01 -9.76993322e-01 1.04722463e-01
-1.31115174e+00 -1.10600543e+00 -4.55028772e-01 7.22246766e-01
2.74164855e-01 -1.41889423e-01 8.98913503e-01 5.76434076e-01
-6.88115656e-01 5.40724337e-01 -9.43258256e-02 -4.16967794e-02
2.95348346e-01 -7.07345545e-01 7.42402673e-01 8.35242748e-01
-3.41922678e-02 6.13876820e-01 4.33994085e-01 -7.62355626e-01
-1.88865948e+00 -8.70918751e-01 2.57086396e-01 -2.90715247e-01
7.29208708e-01 -3.82913798e-01 -8.80671218e-02 7.26223946e-01
2.15562060e-01 3.45740139e-01 7.68660069e-01 -2.14274257e-01
-1.17558300e-01 -4.86978680e-01 -1.67564523e+00 6.93887055e-01
1.54735136e+00 -4.37107861e-01 -1.35055155e-01 1.93066761e-01
4.56633806e-01 1.99930534e-01 -7.54629970e-01 7.64059186e-01
5.73569953e-01 -1.00677574e+00 9.77630794e-01 -9.39089879e-02
-8.19551587e-01 -5.67989767e-01 -5.77235892e-02 -7.04359710e-01
-3.03597867e-01 -8.73706818e-01 -5.25183201e-01 1.44515979e+00
-1.70085877e-01 -1.19145405e+00 1.07012534e+00 5.06172955e-01
1.35386661e-01 -5.39281785e-01 -1.23032188e+00 -1.01445103e+00
-7.95207858e-01 -7.21796453e-01 8.17964613e-01 6.98867142e-01
3.13053548e-01 6.01487219e-01 -1.14882894e-01 4.53032643e-01
7.10423112e-01 -5.93378432e-02 6.11084759e-01 -1.56548572e+00
-6.74635544e-02 -2.88838983e-01 -1.02881193e+00 -9.28534687e-01
-2.14053884e-01 -5.96976459e-01 -4.62178886e-01 -1.39788353e+00
-1.77067183e-02 -8.17419291e-01 -6.18164957e-01 6.86936617e-01
4.01666135e-01 7.57557094e-01 -2.16053382e-01 -6.18925095e-02
-9.03287768e-01 3.91426608e-02 4.73625690e-01 -2.58882523e-01
-3.64614069e-01 1.22426599e-01 -7.41603315e-01 1.08944309e+00
1.34508610e+00 -6.54197454e-01 -6.60291910e-01 7.94756636e-02
2.56950766e-01 -6.22347057e-01 1.49728522e-01 -1.47159684e+00
4.64276463e-01 1.36733158e-02 3.96649122e-01 -2.61962086e-01
1.12354599e-01 -1.60485578e+00 3.46952349e-01 9.04770494e-01
9.54735398e-01 1.43770292e-01 1.10439450e-01 4.00088131e-01
5.77744067e-01 1.05698377e-01 6.57168627e-01 9.86627340e-02
-7.89265215e-01 5.64667523e-01 -2.75376111e-01 -3.26209873e-01
1.45052361e+00 -5.28726637e-01 -6.54219031e-01 2.97909435e-02
-1.62533537e-01 -3.13593447e-01 6.48870051e-01 5.46566188e-01
3.71936917e-01 -1.11686945e+00 2.25728258e-01 5.92072308e-01
-7.54459053e-02 -4.06184644e-01 -1.60832554e-01 5.47780454e-01
-2.51499414e-01 4.47814226e-01 -2.32837528e-01 -4.82733160e-01
-1.20713806e+00 1.10718298e+00 5.72777819e-03 -2.25096941e-01
-7.83114433e-01 3.27479959e-01 -3.15326750e-01 5.82850352e-02
2.69849747e-01 -2.61815608e-01 -1.25017324e-02 -5.20027041e-01
7.41999328e-01 9.08292711e-01 1.70960367e-01 -4.03994054e-01
-7.35921741e-01 8.85066509e-01 2.77732730e-01 4.71485645e-01
1.13806784e+00 -2.74763525e-01 -1.68924257e-01 -8.69622305e-02
1.01361704e+00 8.94470699e-03 -6.57638609e-01 3.18533421e-01
-3.91358044e-04 -1.14064865e-01 2.76813716e-01 -4.80378419e-01
-1.32379115e+00 3.11824262e-01 1.31883597e+00 7.91026890e-01
1.58926034e+00 -6.42377809e-02 1.52135658e+00 5.19231737e-01
1.07470214e+00 -9.88979161e-01 1.76094666e-01 1.40864104e-01
-4.48783100e-01 -1.12099504e+00 1.22278199e-01 -6.84616327e-01
3.63238275e-01 7.62789488e-01 5.73809206e-01 4.28106412e-02
1.17718041e+00 8.95738125e-01 -1.66322768e-01 -3.38838160e-01
-2.29407132e-01 -2.77447134e-01 -5.15725911e-01 9.09138501e-01
-2.53443748e-01 9.98891219e-02 -2.29779109e-01 3.67374539e-01
2.31664911e-01 4.28061128e-01 2.20234558e-01 1.35573578e+00
-4.84529048e-01 -1.48688114e+00 -3.01984489e-01 5.32104075e-01
-7.82474041e-01 -3.08856405e-02 2.72913948e-02 4.77786690e-01
5.76899648e-01 1.61207223e+00 1.72551051e-02 -8.17289829e-01
1.46679759e-01 -1.99210830e-02 2.43663862e-01 -2.23661840e-01
-2.01861084e-01 -5.65449357e-01 -2.15088576e-01 -4.81776804e-01
-5.75751722e-01 -5.68976738e-02 -1.22751880e+00 -7.36299276e-01
-5.45491993e-01 -1.76712945e-01 1.06620371e+00 8.63471985e-01
6.45324230e-01 6.24156356e-01 7.32345343e-01 -6.92761421e-01
8.83069783e-02 -5.97351074e-01 -5.09587646e-01 2.38659099e-01
-1.18134856e-01 -7.46362746e-01 -7.18802154e-01 -3.90621990e-01] | [5.178332805633545, 7.154537677764893] |
685ea557-9dba-4789-9ca4-127a9287feca | dds3d-dense-pseudo-labels-with-dynamic | 2303.05079 | null | https://arxiv.org/abs/2303.05079v2 | https://arxiv.org/pdf/2303.05079v2.pdf | DDS3D: Dense Pseudo-Labels with Dynamic Threshold for Semi-Supervised 3D Object Detection | In this paper, we present a simple yet effective semi-supervised 3D object detector named DDS3D. Our main contributions have two-fold. On the one hand, different from previous works using Non-Maximal Suppression (NMS) or its variants for obtaining the sparse pseudo labels, we propose a dense pseudo-label generation strategy to get dense pseudo-labels, which can retain more potential supervision information for the student network. On the other hand, instead of traditional fixed thresholds, we propose a dynamic threshold manner to generate pseudo-labels, which can guarantee the quality and quantity of pseudo-labels during the whole training process. Benefiting from these two components, our DDS3D outperforms the state-of-the-art semi-supervised 3d object detection with mAP of 3.1% on the pedestrian and 2.1% on the cyclist under the same configuration of 1% samples. Extensive ablation studies on the KITTI dataset demonstrate the effectiveness of our DDS3D. The code and models will be made publicly available at https://github.com/hust-jy/DDS3D | ['Jinghua Hou', 'Zhe Liu', 'Jingyu Li', 'Dingkang Liang'] | 2023-03-09 | null | null | null | null | ['pseudo-label'] | ['miscellaneous'] | [ 2.04939824e-02 1.71769366e-01 -2.12522373e-01 -4.45546210e-01
-6.59624696e-01 -3.87545884e-01 4.52539057e-01 -9.32026953e-02
-3.60037774e-01 4.45085198e-01 -8.76460671e-02 -1.66652381e-01
3.63204211e-01 -5.07486343e-01 -6.02102339e-01 -8.21933031e-01
1.06673345e-01 3.64120871e-01 7.37089157e-01 9.52826217e-02
-5.27146682e-02 4.60345447e-01 -1.78456295e+00 2.00930294e-02
7.12087572e-01 1.00642812e+00 2.65595645e-01 3.28445941e-01
4.69125947e-03 2.96549886e-01 -4.21416312e-01 -1.49150997e-01
5.77374339e-01 -2.41225049e-01 -4.58688140e-01 3.42252016e-01
5.26522875e-01 -3.86370897e-01 -3.76914710e-01 1.17002690e+00
8.57979298e-01 -1.42146125e-01 8.31696868e-01 -1.22099054e+00
-3.72205019e-01 4.68010724e-01 -8.94579530e-01 -1.41785888e-03
1.14082679e-01 1.40330568e-01 6.86376393e-01 -1.08737028e+00
5.86905122e-01 1.35306883e+00 6.09383821e-01 6.77241206e-01
-1.01861465e+00 -9.86713409e-01 2.68762857e-01 1.35097921e-01
-1.64813077e+00 -3.73572052e-01 7.86620378e-01 -4.07712489e-01
4.28414941e-01 7.10706264e-02 6.12352610e-01 1.12436140e+00
-3.42963308e-01 1.14953697e+00 1.41554844e+00 -6.93023741e-01
1.57221466e-01 4.72319365e-01 4.86856818e-01 9.00688589e-01
4.40472096e-01 3.38538736e-01 -4.81697530e-01 1.85110793e-02
5.76805115e-01 -1.71237215e-01 1.07633367e-01 -6.25124812e-01
-1.04230785e+00 6.73250616e-01 3.48578632e-01 1.46089531e-02
-3.48761491e-02 -3.41980755e-01 3.20011258e-01 -8.61147195e-02
7.72699356e-01 -1.84994236e-01 -2.57874578e-01 1.61263019e-01
-8.74582052e-01 1.27765745e-01 5.83060801e-01 1.18138790e+00
6.83541238e-01 -2.34594688e-01 -3.24578047e-01 7.48726428e-01
4.60499048e-01 8.05613339e-01 5.95631078e-02 -7.20670640e-01
2.61665374e-01 6.72683060e-01 2.94381917e-01 -7.18761563e-01
-4.37801749e-01 -6.99327469e-01 -7.47878134e-01 5.10115147e-01
4.59311843e-01 -1.42500535e-01 -1.35650408e+00 1.50404429e+00
8.07338715e-01 3.02523106e-01 -2.86545455e-01 1.25154030e+00
1.16892493e+00 2.78800547e-01 7.10523427e-02 -1.01819653e-02
1.37987113e+00 -1.05526888e+00 -5.01673818e-01 -2.67498672e-01
6.38769269e-01 -7.78787076e-01 8.88516247e-01 1.49657845e-01
-7.29426444e-01 -8.48018646e-01 -1.14343679e+00 7.00584799e-02
-3.02058220e-01 4.98592883e-01 5.69576800e-01 8.92876744e-01
-8.39593172e-01 1.26361489e-01 -8.29347730e-01 -4.71240938e-01
6.58308923e-01 1.50126651e-01 -2.06173986e-01 -1.27706394e-01
-9.83715415e-01 8.86917531e-01 4.05564308e-01 -8.39095414e-02
-1.01511347e+00 -5.46578944e-01 -7.04595387e-01 -3.19138497e-01
5.47628880e-01 -4.06959713e-01 1.20110774e+00 -5.77963710e-01
-1.40704393e+00 1.35907102e+00 -1.98870286e-01 -3.13955843e-01
6.96200907e-01 -2.69065201e-01 -1.03490844e-01 8.27795938e-02
2.78331161e-01 9.98592496e-01 7.59513199e-01 -1.36949456e+00
-7.63076603e-01 -3.26962173e-01 -2.57863700e-01 3.15578669e-01
-1.92444071e-01 4.24948297e-02 -8.37765098e-01 -4.84068751e-01
3.09683383e-01 -1.20888698e+00 -2.48894989e-01 2.38115907e-01
-6.99740112e-01 -4.32434738e-01 8.32390726e-01 -3.38089883e-01
8.22386801e-01 -2.34366798e+00 -3.04457732e-02 1.37848705e-01
3.09956580e-01 6.78420246e-01 -1.43996514e-02 -1.83629125e-01
8.00710693e-02 -1.59242019e-01 -2.78839231e-01 -6.95758224e-01
4.25274856e-02 -9.47661176e-02 -6.30725175e-02 6.01850152e-01
2.44212762e-01 7.59749770e-01 -9.18320596e-01 -7.97673047e-01
4.64175314e-01 5.70033133e-01 -7.60546401e-02 1.05681077e-01
-6.64300174e-02 3.30862701e-01 -6.65865421e-01 9.11253512e-01
1.02063084e+00 -2.03794643e-01 -1.34012654e-01 -2.15389714e-01
-1.42889023e-01 1.99758634e-01 -1.55933726e+00 1.58575046e+00
1.17487453e-01 1.73247010e-01 1.01994589e-01 -8.72039795e-01
1.10129201e+00 -4.34399918e-02 3.34454089e-01 -6.03252411e-01
3.18088979e-01 2.06164628e-01 -5.04615247e-01 -1.09175548e-01
3.07179183e-01 -1.18785519e-02 -4.67217043e-02 4.69950259e-01
1.32754773e-01 1.87899962e-01 1.63809642e-01 3.08610320e-01
7.95131862e-01 4.47304964e-01 1.92183003e-01 -2.83492386e-01
4.11733270e-01 9.46297646e-02 6.82724655e-01 7.87512839e-01
-6.52583539e-01 6.25518262e-01 2.33837858e-01 -2.98432529e-01
-8.17291379e-01 -1.17308795e+00 -3.95733148e-01 7.48751700e-01
5.31606197e-01 -2.36008421e-01 -7.09109843e-01 -9.56777632e-01
1.64562404e-01 5.70970893e-01 -4.43234414e-01 -3.45575921e-02
-3.17041129e-01 -9.31958675e-01 4.71197486e-01 3.35574985e-01
7.42858529e-01 -8.27775776e-01 -4.81875241e-01 -9.61796418e-02
-1.70916259e-01 -1.29977596e+00 -3.76046568e-01 3.33637685e-01
-6.62125826e-01 -1.00754905e+00 -8.25624228e-01 -8.45754385e-01
8.11647296e-01 7.09674299e-01 9.22102988e-01 -1.45483956e-01
-3.84228110e-01 1.14280753e-01 -5.69962382e-01 -6.21399164e-01
-4.12234604e-01 2.11595342e-01 1.89726040e-01 -3.61612290e-02
6.75537825e-01 -4.63059396e-01 -5.94514012e-01 4.87619847e-01
-3.52874011e-01 2.58470595e-01 8.33795190e-01 4.65034693e-01
9.11997437e-01 -6.80414587e-02 4.71014291e-01 -8.84714186e-01
1.59026626e-02 -2.80227631e-01 -7.51964569e-01 -2.70039905e-02
-6.63658679e-01 -9.97520834e-02 9.62212086e-02 -5.43998301e-01
-1.05767334e+00 6.06941164e-01 -1.13582417e-01 -5.20239234e-01
-6.41381204e-01 -3.41772467e-01 -2.14877784e-01 -1.57052010e-01
7.15606153e-01 1.27355054e-01 5.10446262e-03 -8.36072564e-01
4.58217174e-01 9.97620821e-01 3.15772921e-01 -3.11611444e-01
1.09882748e+00 7.62888730e-01 7.31406137e-02 -6.95074737e-01
-1.34079564e+00 -7.59054780e-01 -8.56242359e-01 -3.09831858e-01
8.97420287e-01 -1.18554986e+00 -3.81712556e-01 7.77402043e-01
-8.79916608e-01 -2.60929227e-01 -2.32793808e-01 5.30363142e-01
-1.90617457e-01 5.39757490e-01 -5.92067838e-01 -9.91380274e-01
-3.48476857e-01 -9.34586704e-01 1.32821703e+00 2.68392444e-01
1.25467345e-01 -4.14522886e-01 -1.91891283e-01 4.69740570e-01
8.36434364e-02 2.94990510e-01 3.01214278e-01 -5.53034246e-01
-5.80935776e-01 -2.45535746e-01 -6.31644130e-01 6.04212642e-01
1.23588085e-01 -4.21054125e-01 -1.14139903e+00 -2.96652764e-01
-1.99164525e-01 -5.31618357e-01 1.02266645e+00 4.26601052e-01
6.66526377e-01 2.54729867e-01 -5.51601350e-01 3.73839766e-01
1.06089723e+00 -2.06181198e-01 2.69097298e-01 3.08287382e-01
7.61598468e-01 5.02631128e-01 9.94013309e-01 4.98543233e-01
4.00296330e-01 7.20316470e-01 3.86829376e-01 -3.05820584e-01
-6.19660735e-01 -3.20180804e-01 3.51142675e-01 5.59839725e-01
1.33976758e-01 8.98235142e-02 -7.63928592e-01 4.82132554e-01
-1.77190602e+00 -7.63258636e-01 -2.51232803e-01 2.23344612e+00
8.63767445e-01 5.29346943e-01 4.66727018e-01 1.93963841e-01
1.02217972e+00 5.61118647e-02 -5.50731659e-01 3.22680384e-01
-2.56361425e-01 -2.13346761e-02 6.73605084e-01 4.65138108e-01
-1.49198198e+00 1.10546565e+00 5.44493580e+00 1.00699949e+00
-8.83071065e-01 2.72078663e-01 4.94154483e-01 -2.26621822e-01
3.46510202e-01 -9.77920964e-02 -1.58331144e+00 4.93943900e-01
4.68324274e-01 3.39324564e-01 -1.11982457e-01 1.00066352e+00
2.20928758e-01 -2.02472419e-01 -8.35563004e-01 9.43954051e-01
1.37369365e-01 -8.24342847e-01 -8.53313431e-02 -1.26332166e-02
6.39264703e-01 1.90948024e-01 -1.02121204e-01 2.04684511e-01
3.23277891e-01 -5.51674902e-01 1.06700623e+00 1.73357770e-01
8.03305209e-01 -5.97991943e-01 6.71189785e-01 5.59541881e-01
-1.18325281e+00 2.14301467e-01 -3.91416162e-01 5.90496138e-02
2.46315062e-01 1.12120152e+00 -7.14888334e-01 4.98915583e-01
8.43905926e-01 7.59967804e-01 -7.85817504e-01 1.10716939e+00
-5.56389093e-01 8.18439603e-01 -6.60076261e-01 -3.89708988e-02
7.21903369e-02 -9.12265107e-03 7.01900482e-01 1.39320529e+00
7.00170994e-02 1.08629681e-01 4.71721977e-01 7.05521941e-01
1.50751635e-01 2.62296684e-02 -3.33117127e-01 4.53079879e-01
6.44845366e-01 1.29495239e+00 -8.32996786e-01 -3.35332096e-01
-3.27690363e-01 8.91287565e-01 1.71549782e-01 8.57195556e-02
-7.81122029e-01 -3.40995550e-01 3.38749826e-01 3.21420342e-01
5.33162057e-01 -2.48745397e-01 -2.33613506e-01 -1.03506637e+00
2.73898423e-01 -5.39398849e-01 3.39969873e-01 -4.64901716e-01
-1.26644659e+00 4.87300575e-01 2.51241177e-01 -1.35820222e+00
2.52293468e-01 -6.40657365e-01 -2.20313132e-01 7.41037428e-01
-1.76973581e+00 -1.47925544e+00 -5.16822040e-01 4.02184129e-01
3.26909721e-01 -3.33122648e-02 5.95820248e-01 5.22275031e-01
-5.73542953e-01 5.34175277e-01 -2.32343495e-01 2.26800859e-01
8.51460218e-01 -1.02605534e+00 5.29079974e-01 9.79037344e-01
9.68116075e-02 1.19409971e-01 6.19561970e-01 -6.73810899e-01
-9.49310541e-01 -1.22354996e+00 7.76876628e-01 -4.00601476e-01
2.55491495e-01 -7.24710107e-01 -6.75895274e-01 4.26756769e-01
-1.74358994e-01 1.88379675e-01 4.23263311e-01 1.35376845e-02
-4.02425021e-01 -5.71782850e-02 -1.15857792e+00 5.00868559e-01
1.52954257e+00 -1.28970638e-01 -4.80615109e-01 5.30169487e-01
8.08645964e-01 -5.21661043e-01 -5.03005207e-01 5.88016152e-01
2.74676383e-01 -1.00128508e+00 1.10292315e+00 -2.06072791e-03
-1.22656755e-01 -6.78049028e-01 -9.93429571e-02 -8.46101940e-01
-3.31482679e-01 -4.08202797e-01 -1.59788743e-01 1.34366822e+00
2.05189735e-01 -4.26402390e-01 1.04334652e+00 1.73320487e-01
-1.59670189e-01 -5.72080970e-01 -1.08813345e+00 -9.78417993e-01
-2.41780728e-01 -4.93026614e-01 2.92855889e-01 6.36114478e-01
-4.60691988e-01 5.03155291e-01 -5.49790084e-01 4.04336065e-01
1.16127980e+00 2.25376159e-01 7.74072647e-01 -1.28193462e+00
-2.10633948e-01 -1.98607013e-01 -3.87991220e-01 -1.40338433e+00
1.25159889e-01 -9.84067559e-01 1.44102186e-01 -1.36187565e+00
3.96860242e-01 -8.33556354e-01 -2.91176617e-01 7.00320065e-01
-1.73593253e-01 5.20008326e-01 1.36676669e-01 2.55985975e-01
-9.29558992e-01 6.63498998e-01 1.24492371e+00 -1.53412363e-02
-3.14200431e-01 1.94636703e-01 -7.12773442e-01 7.25721300e-01
1.00055599e+00 -7.63477862e-01 -2.34235957e-01 -1.40269116e-01
-4.06333596e-01 -5.75653911e-01 5.61046481e-01 -1.03656280e+00
1.18826605e-01 8.09942409e-02 2.71075487e-01 -9.99503970e-01
3.42031479e-01 -5.24984896e-01 -2.57368326e-01 4.99896884e-01
1.67137217e-02 -5.48809946e-01 1.92588896e-01 5.79623103e-01
2.52832957e-02 -1.99910149e-01 1.05732465e+00 -7.47815520e-02
-9.32911813e-01 2.29524270e-01 -1.23185933e-01 6.73317164e-02
1.28381741e+00 -1.98590800e-01 -1.83330432e-01 2.88243163e-02
-5.33265352e-01 3.71660292e-01 4.30217713e-01 4.77095902e-01
4.77602214e-01 -1.32723117e+00 -8.77569735e-01 2.50215262e-01
1.95567623e-01 2.43210167e-01 1.51614517e-01 6.20172918e-01
-1.45158738e-01 4.06771600e-01 -1.26845136e-01 -8.49717617e-01
-1.38186884e+00 4.48444426e-01 1.35002106e-01 -3.10575217e-02
-8.01154196e-01 8.01785171e-01 9.83497277e-02 -5.09651780e-01
5.02762198e-01 8.27105902e-03 -2.47441661e-02 3.68344365e-04
4.47879672e-01 3.07146668e-01 -4.85371314e-02 -6.17424667e-01
-5.79925299e-01 6.31795824e-01 -1.67622998e-01 2.73200404e-02
1.19870138e+00 -1.60251483e-01 3.32205653e-01 3.00900102e-01
8.83999109e-01 -9.97268260e-02 -1.46321797e+00 -4.25754309e-01
-6.08189590e-02 -3.92496854e-01 2.09436715e-01 -7.49718785e-01
-9.32248235e-01 6.26441121e-01 1.02020216e+00 -1.57342792e-01
9.97224748e-01 3.81898493e-01 7.13596463e-01 2.50928342e-01
4.26533937e-01 -9.52215552e-01 7.26004466e-02 2.45675057e-01
4.57475603e-01 -1.39933991e+00 1.80536225e-01 -8.27462137e-01
-6.09824717e-01 4.89129752e-01 7.16309369e-01 -4.98245507e-02
7.53582478e-01 3.37466598e-01 2.91520953e-01 -8.21531117e-02
-3.07509243e-01 -5.92939854e-01 2.27338269e-01 6.34711444e-01
5.60714379e-02 9.24470276e-02 -2.29632080e-01 4.36898500e-01
1.13714024e-01 -2.67349137e-03 2.51795828e-01 9.14896905e-01
-6.54688478e-01 -1.18278027e+00 -4.59358543e-01 3.20917547e-01
-9.18561518e-02 1.29020542e-01 -4.09730464e-01 7.88019717e-01
4.14670378e-01 9.21661139e-01 -3.60177606e-01 -4.65529710e-01
5.34830272e-01 1.72712684e-01 3.85087579e-01 -7.63419688e-01
-1.64830178e-01 1.08579621e-01 1.07035980e-01 -5.12257755e-01
-7.53409028e-01 -9.24860716e-01 -1.18589866e+00 -1.03445120e-01
-6.67848945e-01 -6.57149777e-02 5.38744569e-01 6.84665143e-01
5.52374065e-01 1.17227554e-01 4.59689468e-01 -1.03497779e+00
-6.04938626e-01 -1.02557230e+00 -6.66136622e-01 4.11844254e-01
1.35504559e-01 -1.08708370e+00 -4.70439374e-01 -5.35230804e-03] | [9.14746379852295, 1.2379863262176514] |
6345f278-a782-441c-8375-10eefdcdc0bf | frequency-enhanced-hybrid-attention-network | 2304.09184 | null | https://arxiv.org/abs/2304.09184v3 | https://arxiv.org/pdf/2304.09184v3.pdf | Frequency Enhanced Hybrid Attention Network for Sequential Recommendation | The self-attention mechanism, which equips with a strong capability of modeling long-range dependencies, is one of the extensively used techniques in the sequential recommendation field. However, many recent studies represent that current self-attention based models are low-pass filters and are inadequate to capture high-frequency information. Furthermore, since the items in the user behaviors are intertwined with each other, these models are incomplete to distinguish the inherent periodicity obscured in the time domain. In this work, we shift the perspective to the frequency domain, and propose a novel Frequency Enhanced Hybrid Attention Network for Sequential Recommendation, namely FEARec. In this model, we firstly improve the original time domain self-attention in the frequency domain with a ramp structure to make both low-frequency and high-frequency information could be explicitly learned in our approach. Moreover, we additionally design a similar attention mechanism via auto-correlation in the frequency domain to capture the periodic characteristics and fuse the time and frequency level attention in a union model. Finally, both contrastive learning and frequency regularization are utilized to ensure that multiple views are aligned in both the time domain and frequency domain. Extensive experiments conducted on four widely used benchmark datasets demonstrate that the proposed model performs significantly better than the state-of-the-art approaches. | ['Victor S. Sheng', 'Fuzhen Zhuang', 'Jianfeng Qu', 'Guanfeng Liu', 'Pengpeng Zhao', 'Huanhuan Yuan', 'Xinyu Du'] | 2023-04-18 | null | null | null | null | ['sequential-recommendation'] | ['miscellaneous'] | [-1.72236994e-01 -6.53810501e-01 -3.97406787e-01 -2.51394242e-01
-1.41562864e-01 -2.44931877e-01 4.27862763e-01 -9.45776850e-02
2.01480854e-02 2.47701243e-01 6.08272433e-01 1.38801515e-01
-4.65057909e-01 -6.97363555e-01 -7.17067540e-01 -6.84373260e-01
2.44208537e-02 -2.57139474e-01 4.11039650e-01 -3.44869763e-01
2.26636693e-01 -9.03377011e-02 -1.57667553e+00 2.98942745e-01
1.27699578e+00 9.33776975e-01 2.18661025e-01 2.61807680e-01
-1.07427187e-01 5.58467388e-01 -3.03913414e-01 4.69975173e-02
4.31534275e-02 -5.03925800e-01 -4.10326600e-01 9.52209458e-02
2.54436284e-01 -1.98780373e-01 -5.93193531e-01 1.05592012e+00
4.04549062e-01 3.27305526e-01 2.05587283e-01 -7.52911270e-01
-1.17111194e+00 5.78988254e-01 -8.37082207e-01 6.83883727e-01
5.13359666e-01 -2.89583392e-02 1.03514528e+00 -8.85812879e-01
1.02517962e-01 1.18955016e+00 7.14138627e-01 2.14963295e-02
-1.03843009e+00 -6.41558409e-01 8.75196517e-01 6.07915699e-01
-1.05655694e+00 1.08331241e-01 1.26425993e+00 -3.51382226e-01
7.38560915e-01 1.06215924e-01 8.07700753e-01 1.25073397e+00
2.76872039e-01 6.77429259e-01 9.52470958e-01 -2.61607617e-01
-3.13203156e-01 6.87102750e-02 5.33667207e-01 3.68823022e-01
1.35157602e-02 2.35906333e-01 -5.35703182e-01 9.48905200e-03
9.59074020e-01 6.76438034e-01 -6.06452823e-01 -9.50971320e-02
-1.25618565e+00 7.54900873e-01 5.09767234e-01 8.36797118e-01
-4.94176239e-01 -8.59453678e-02 4.06697005e-01 3.45661789e-01
6.54768050e-01 1.20140694e-01 -3.23132187e-01 2.46906698e-01
-5.65809309e-01 1.12208426e-01 4.19879705e-01 7.58658648e-01
4.84244108e-01 2.07679957e-01 -4.86092895e-01 9.01287258e-01
4.25614387e-01 2.81894386e-01 9.87672091e-01 -4.09339845e-01
2.38910511e-01 4.82424557e-01 1.24873474e-01 -1.38462627e+00
-4.35009688e-01 -1.00353229e+00 -1.02864778e+00 -4.46934044e-01
2.35812157e-01 -4.53126766e-02 -5.65810263e-01 1.81901884e+00
4.26576376e-01 8.18736613e-01 -2.68180281e-01 1.25230694e+00
8.56449425e-01 8.06513250e-01 -2.38847658e-02 -6.68407023e-01
1.57889736e+00 -1.25073433e+00 -1.22513998e+00 2.53793206e-02
-8.16137567e-02 -7.34307230e-01 1.28051329e+00 3.39541316e-01
-9.85892594e-01 -1.25654006e+00 -1.24232197e+00 1.99765954e-02
-1.37353852e-01 8.11739042e-02 6.87673032e-01 3.19944739e-01
-5.38157165e-01 7.56209433e-01 -7.46629119e-01 -2.71678984e-01
-5.69333462e-03 1.83247834e-01 1.37819335e-01 1.70660928e-01
-1.66607904e+00 3.91195625e-01 1.85996786e-01 6.79602399e-02
-2.92791784e-01 -7.57864416e-01 -3.75430822e-01 3.20784211e-01
4.62718666e-01 -4.93830115e-01 9.45926845e-01 -1.17276394e+00
-1.41284657e+00 8.35983530e-02 -5.58090806e-02 -1.61607772e-01
1.18423820e-01 -4.96579230e-01 -1.06155789e+00 7.67382160e-02
-1.48676291e-01 -1.90630659e-01 1.02020955e+00 -9.94928360e-01
-7.26842523e-01 -3.38452935e-01 2.56828576e-01 3.18100989e-01
-8.69508684e-01 -2.43196130e-01 -5.88807106e-01 -1.12339318e+00
1.55499533e-01 -6.89178526e-01 8.61311331e-02 -4.44856763e-01
-4.14718166e-02 -2.28185713e-01 9.36980009e-01 -6.35610640e-01
1.90411234e+00 -2.33060670e+00 1.96551070e-01 8.07538554e-02
1.15557969e-01 2.64032155e-01 -1.47696421e-01 6.00636542e-01
-2.76561648e-01 -1.91655427e-01 1.97664306e-01 4.27215779e-03
-1.87774122e-01 7.86819458e-02 -3.95066738e-01 5.17209053e-01
-1.39833940e-02 5.82081258e-01 -9.60928977e-01 -1.53142542e-01
1.36874303e-01 7.76979089e-01 -7.81093776e-01 2.78345197e-01
-3.81559804e-02 7.61080086e-01 -6.35900080e-01 1.77715242e-01
7.28061438e-01 -5.65109313e-01 1.37229323e-01 -6.62380874e-01
-3.48755181e-01 2.79038012e-01 -1.22997761e+00 1.63815367e+00
-3.60027790e-01 -2.90824082e-02 -1.35058299e-01 -1.08762968e+00
7.35282898e-01 4.92017984e-01 8.66772950e-01 -8.60647261e-01
-5.66525087e-02 -4.39029373e-03 8.99023041e-02 -7.88250506e-01
3.39344114e-01 -8.69922563e-02 3.56108725e-01 3.97381663e-01
2.64405996e-01 6.07050419e-01 1.68074682e-01 -1.85366929e-01
5.97106636e-01 2.64677107e-01 1.23158999e-01 -3.57625246e-01
8.85053933e-01 -5.26519835e-01 7.92713583e-01 4.75216150e-01
3.66149694e-02 5.68924606e-01 8.12515467e-02 -5.01692235e-01
-8.09170485e-01 -5.38287461e-01 -2.26203203e-01 1.31747794e+00
6.30244017e-01 -4.74326700e-01 -4.29161578e-01 -7.28575230e-01
1.04312161e-02 3.91977221e-01 -7.40232348e-01 -4.81473893e-01
-6.05511606e-01 -9.57973480e-01 -2.69200891e-01 6.43765867e-01
5.87264240e-01 -8.98761928e-01 -2.51865476e-01 5.80752492e-01
-3.90522510e-01 -8.22530150e-01 -1.19369113e+00 -5.92750721e-02
-7.67354548e-01 -9.97235477e-01 -8.94600868e-01 -7.37482667e-01
3.19982439e-01 7.49445081e-01 9.08465266e-01 2.76426613e-01
6.36713803e-02 3.57641429e-01 -8.19964886e-01 -1.98710188e-01
3.60411882e-01 4.07225117e-02 -1.89356669e-03 6.28542840e-01
5.18637121e-01 -9.64907050e-01 -9.24089253e-01 3.94211113e-01
-8.60229373e-01 -1.12177536e-01 4.80408967e-01 1.10250580e+00
5.32102704e-01 1.69193283e-01 9.04598475e-01 -8.70963216e-01
6.80191696e-01 -8.49263966e-01 -2.70629764e-01 8.83525461e-02
-7.50940800e-01 -2.26676345e-01 1.08290362e+00 -9.77257788e-01
-1.18299818e+00 -4.90985841e-01 -9.36446413e-02 -5.98094404e-01
5.65646682e-04 7.46988952e-01 6.69417605e-02 -1.21024027e-02
3.69898140e-01 4.87723440e-01 -1.62232384e-01 -7.28926897e-01
6.86611608e-03 3.73273700e-01 2.84194589e-01 -4.43850875e-01
6.66094244e-01 4.72021520e-01 -4.32053328e-01 -5.09203970e-01
-1.23082852e+00 -6.01944387e-01 -1.86766788e-01 -1.33998111e-01
6.75390601e-01 -8.36021543e-01 -7.40056813e-01 3.15906852e-01
-8.93669069e-01 2.07151473e-01 -2.34976113e-01 9.18618619e-01
-3.36986810e-01 6.12435043e-01 -6.52875721e-01 -7.71624863e-01
-4.22383755e-01 -9.41217244e-01 7.05476999e-01 6.14046037e-01
1.51399553e-01 -1.00182760e+00 1.56307861e-01 1.61083043e-02
5.95932424e-01 -1.49214625e-01 8.47840428e-01 -6.60117269e-01
-3.53566498e-01 2.08035320e-01 -1.93825543e-01 -8.30813032e-03
2.31559068e-01 -1.79221958e-01 -7.05112457e-01 -3.24276149e-01
5.21549165e-01 9.49114338e-02 8.10604453e-01 4.58387136e-01
1.39018071e+00 -3.70712101e-01 -1.53430521e-01 6.86711311e-01
1.47051978e+00 4.07950580e-01 5.26972473e-01 2.55166292e-01
8.00970495e-01 5.46111286e-01 4.90372360e-01 6.62717342e-01
4.20432329e-01 8.31573129e-01 2.21778482e-01 -1.14050008e-01
-1.75118092e-02 -2.61908680e-01 1.77006975e-01 1.27012002e+00
-4.62662160e-01 -2.12733090e-01 -2.93140233e-01 5.04233956e-01
-2.29610085e+00 -1.29083347e+00 -2.17125997e-01 2.09380221e+00
5.92784047e-01 -7.76892155e-02 2.48382375e-01 1.07528538e-01
7.88501322e-01 2.54856259e-01 -5.84038496e-01 4.93505970e-02
-6.04551770e-02 -1.11531422e-01 1.76131874e-01 9.17135030e-02
-1.25654984e+00 2.88954407e-01 5.86269522e+00 8.64257753e-01
-1.23577297e+00 5.18034026e-02 4.51284081e-01 -5.31407520e-02
-4.79883730e-01 -2.69643158e-01 -5.11318088e-01 9.54402387e-01
7.51894832e-01 -2.18259886e-01 5.74240506e-01 4.61396784e-01
3.17544550e-01 4.12259549e-01 -7.52302051e-01 8.38437796e-01
1.05606869e-01 -8.95118713e-01 -2.74513811e-01 -6.60378411e-02
7.50821352e-01 -3.37362558e-01 1.99587688e-01 5.00059724e-01
-4.85423096e-02 -7.71164775e-01 5.84053516e-01 9.10029173e-01
4.64579850e-01 -8.12630951e-01 7.56393552e-01 3.90238971e-01
-1.70986784e+00 -3.20622653e-01 -2.25566730e-01 1.14158764e-02
1.28457397e-01 6.15039647e-01 2.36154556e-01 1.23148632e+00
1.00019920e+00 1.30356443e+00 -1.09225564e-01 9.90545154e-01
1.43308878e-01 7.55045950e-01 -1.28812373e-01 1.68761939e-01
1.47040993e-01 -5.61097622e-01 5.78458905e-01 9.87539053e-01
3.69791031e-01 3.27160120e-01 5.09973645e-01 5.62629342e-01
3.33336681e-01 1.89300865e-01 -8.65721256e-02 6.12177246e-04
2.13134617e-01 1.00283456e+00 -3.77682954e-01 -2.17386812e-01
-1.01011658e+00 7.29148686e-01 1.82455987e-01 5.13821959e-01
-1.20487249e+00 -3.19518179e-01 3.40773165e-01 1.01133026e-01
6.90280676e-01 -7.86029324e-02 6.59178048e-02 -1.58440828e+00
1.24360405e-01 -9.90073979e-01 4.65155691e-01 -3.41412693e-01
-1.75960410e+00 5.28445005e-01 -1.82904631e-01 -1.70273006e+00
1.12760223e-01 -7.24223703e-02 -7.25130141e-01 9.24221396e-01
-1.77941751e+00 -1.21286595e+00 -4.22132999e-01 8.42704892e-01
7.34835982e-01 2.53237151e-02 6.60256326e-01 7.65253365e-01
-5.35404325e-01 5.94822943e-01 1.89900085e-01 -3.22155178e-01
7.70615160e-01 -1.04996932e+00 -1.35278016e-01 6.60502017e-01
7.82526210e-02 9.76058424e-01 5.93058884e-01 -6.11900330e-01
-1.56671834e+00 -9.76034641e-01 4.56922859e-01 1.46435305e-01
6.68412864e-01 -3.92028727e-02 -1.38704455e+00 5.03089905e-01
3.29089642e-01 2.79741168e-01 6.93675339e-01 3.17605108e-01
-2.85573453e-01 -3.07000607e-01 -7.43423879e-01 3.31346214e-01
1.06776679e+00 -4.83892620e-01 -6.98644042e-01 2.76063085e-01
8.29925537e-01 -2.93848425e-01 -1.06338942e+00 5.13434350e-01
7.29850590e-01 -1.15404654e+00 1.09528387e+00 -4.12261307e-01
3.54151398e-01 -4.57612753e-01 6.88092113e-02 -1.27422845e+00
-1.11303020e+00 -6.10550404e-01 -5.76298773e-01 1.37824714e+00
-1.61430046e-01 -6.05268836e-01 3.70480895e-01 -2.49514833e-01
-2.55529135e-01 -9.44523573e-01 -5.23473799e-01 -6.71331882e-01
-3.61180991e-01 6.78496435e-02 6.37169600e-01 1.06899762e+00
1.75027236e-01 4.88418132e-01 -8.22691679e-01 1.99632898e-01
3.77149522e-01 6.09865546e-01 2.09412590e-01 -1.31032932e+00
-6.62659764e-01 -4.17993814e-01 5.25505617e-02 -1.41090441e+00
-1.60701156e-01 -4.32436198e-01 -3.06660116e-01 -1.23961473e+00
3.98711383e-01 -2.11340025e-01 -8.66997719e-01 5.62935472e-02
-5.90704560e-01 3.01769748e-02 -5.19165471e-02 4.96108472e-01
-4.62682873e-01 8.98789346e-01 1.46597040e+00 4.19579959e-03
-2.42848217e-01 -2.21611839e-03 -8.93875957e-01 7.60130644e-01
3.71519089e-01 -1.85524851e-01 -6.63724363e-01 -4.12854016e-01
-1.75501689e-01 1.48264050e-01 3.59863862e-02 -9.49245393e-01
2.26334497e-01 -2.77475059e-01 3.74981284e-01 -7.06269562e-01
1.04572497e-01 -1.14215565e+00 1.84430137e-01 2.70567060e-01
-2.74945915e-01 1.23928953e-02 9.28046554e-02 1.05364943e+00
-4.34949875e-01 1.79598182e-01 7.45384812e-01 -9.69420448e-02
-5.64633489e-01 6.66086614e-01 -1.90225691e-02 -1.78718686e-01
8.52961600e-01 -5.20235598e-02 -2.06514582e-01 -3.14161181e-01
-6.29703403e-01 3.13857645e-01 1.02713056e-01 6.59919143e-01
3.09808195e-01 -1.66674256e+00 -5.91062903e-01 2.53415704e-01
4.51824926e-02 -4.73585784e-01 1.10142970e+00 1.16229153e+00
2.31806517e-01 4.65138584e-01 -1.42737508e-01 -4.41258937e-01
-8.20915461e-01 1.02550483e+00 1.48795739e-01 -5.87943375e-01
-7.96526372e-01 4.71807808e-01 5.16556025e-01 -3.10458932e-02
1.82852149e-01 -1.78229004e-01 -9.69980776e-01 1.59629866e-01
7.95363605e-01 1.79801553e-01 -1.50739387e-01 -6.02799952e-01
-3.04299146e-01 9.82660294e-01 -2.14869663e-01 2.88286239e-01
1.45555162e+00 -4.69236076e-01 1.30543470e-01 6.58720732e-01
1.04810119e+00 2.71492124e-01 -1.24203002e+00 -4.51668203e-01
-3.31653297e-01 -3.95748496e-01 2.58220673e-01 -5.06167769e-01
-1.20488942e+00 6.39143407e-01 6.06394112e-01 8.04682255e-01
1.41710985e+00 -4.39761013e-01 9.72594023e-01 -2.31043607e-01
8.79357979e-02 -1.06160545e+00 2.85416782e-01 5.60435832e-01
8.44500124e-01 -1.14435112e+00 4.19912487e-02 -4.69739050e-01
-3.08420211e-01 1.16164231e+00 8.29426706e-01 -5.54981768e-01
9.49247360e-01 1.55066941e-02 -2.24198759e-01 -6.33616671e-02
-7.53268719e-01 -1.19649969e-01 7.85582781e-01 2.84662068e-01
7.63690472e-01 -3.64828944e-01 -8.18412185e-01 1.20939374e+00
2.85683006e-01 1.21811353e-01 2.10167080e-01 7.78681934e-01
-2.79177487e-01 -9.60678041e-01 -3.53169292e-01 3.74598831e-01
-7.65876472e-01 1.86063461e-02 2.84436464e-01 4.44225311e-01
3.18411797e-01 1.04537892e+00 1.05567463e-01 -6.35971129e-01
5.15631199e-01 -1.62650138e-01 2.18405530e-01 -4.62639898e-01
-8.29128444e-01 7.46425390e-01 -2.47279599e-01 -5.39222956e-01
-6.49639428e-01 -2.97818661e-01 -9.27706122e-01 -7.13456571e-02
-6.24451876e-01 2.67293602e-01 2.11996377e-01 8.74646485e-01
6.05130315e-01 1.15739238e+00 9.97684240e-01 -8.84183824e-01
-6.16434991e-01 -1.12396407e+00 -6.94723964e-01 6.78526103e-01
4.28918660e-01 -9.31576371e-01 -3.71625513e-01 1.92031916e-02] | [10.137913703918457, 5.552881240844727] |
ee902da4-f3ec-4c52-ad8f-96e979794a1c | s2d2net-an-improved-approach-for-robust-steel | null | null | https://ieeexplore.ieee.org/document/9506405 | https://ieeexplore.ieee.org/document/9506405 | S2D2Net: An Improved Approach For Robust Steel Surface Defects Diagnosis With Small Sample Learning | Surface defect recognition of products is a necessary process to guarantee the quality of industrial production. This paper proposes a hybrid model, S2D2Net (Steel Surface Defect Diagnosis Network), for an efficient and robust inspection of the steel surface during the manufacturing process. The S2D2Net uses a pretrained ImageNet model as a feature extractor and learns a Capsule Network over the extracted features. The experimental results on a publicly available steel surface defect dataset (NEU) show that S2D2Net achieved 99.17% accuracy with minimal training data and improved by 9.59% over its closest competitor based on GAN. S2D2Net proved its robustness by achieving 94.7% accuracy on a diversity enhanced dataset, ENEU, and improved by 3.6% over its closest competitor. It has better, robust recognition performance compared to other state-of-the-art DNN-based detectors. | ['Chiranjoy Chattopadhyay', 'Vikanksh Nath'] | 2021-08-23 | null | null | null | ieee-international-conference-on-image-3 | ['small-data'] | ['computer-vision'] | [ 3.02442044e-01 4.72871840e-01 3.62253606e-01 -2.08059132e-01
-6.15338743e-01 5.12964129e-02 2.48747185e-01 -2.30431810e-01
-3.23896185e-02 8.77509937e-02 -5.86168528e-01 2.26447508e-01
-5.97940125e-02 -8.90968680e-01 -4.81380105e-01 -8.24734509e-01
8.54260176e-02 5.47486722e-01 4.72263753e-01 -5.15250087e-01
2.44421899e-01 7.62454093e-01 -1.53725982e+00 5.54193914e-01
5.34079254e-01 1.53125811e+00 4.93243784e-01 3.60801578e-01
2.70298511e-01 7.32073545e-01 -9.67340946e-01 -1.24873780e-01
5.03974378e-01 -1.70231670e-01 -4.67885137e-01 2.33495776e-02
2.15471923e-01 -1.14641771e-01 -4.65070099e-01 1.38103485e+00
7.41630495e-01 -1.54474840e-01 7.03682005e-01 -9.22137082e-01
-8.48813415e-01 4.74411756e-01 -4.56632853e-01 -3.39207575e-02
-1.73854139e-02 1.11997440e-01 6.12938106e-01 -1.13817871e+00
7.71356285e-01 1.12356079e+00 7.69632280e-01 7.03508854e-01
-5.54264545e-01 -5.02521455e-01 -3.13050896e-01 7.87615627e-02
-1.18409097e+00 1.04592532e-01 9.49258089e-01 -3.17959338e-01
1.12699020e+00 4.70201932e-02 5.29542148e-01 9.79833007e-01
5.04544437e-01 1.08171928e+00 6.50782526e-01 -5.23873985e-01
3.13992351e-01 -3.78683299e-01 4.35703434e-02 1.02730536e+00
4.03561503e-01 3.60470504e-01 -3.73707235e-01 4.13991928e-01
1.32365334e+00 -2.89026964e-02 -4.66571338e-02 -1.88485861e-01
-8.44113767e-01 6.57110751e-01 5.74841917e-01 6.87503994e-01
-4.84967023e-01 -4.89147156e-02 5.65518379e-01 7.70802557e-01
4.81214434e-01 7.35239625e-01 -6.62817955e-01 5.54172732e-02
-4.75856185e-01 -6.14432469e-02 5.55490673e-01 8.91011775e-01
2.64020145e-01 7.39412844e-01 -5.50039858e-02 1.17858493e+00
2.62183696e-01 6.67161345e-01 3.99412662e-01 -8.01301837e-01
1.59646422e-01 9.46340740e-01 -1.84066787e-01 -1.13068151e+00
-4.48184818e-01 -6.73782229e-01 -1.17059386e+00 7.76113510e-01
-3.43698710e-02 5.98832667e-02 -1.33024979e+00 1.10101259e+00
-6.68392144e-03 -4.30598497e-01 3.99021395e-02 7.72824347e-01
1.10429573e+00 4.62121308e-01 -5.79558730e-01 4.37577993e-01
9.62133050e-01 -1.06064117e+00 -5.17599821e-01 -1.54912025e-01
6.30932093e-01 -5.98455727e-01 4.22326207e-01 9.91270721e-01
-8.63087237e-01 -9.70968962e-01 -1.54109657e+00 5.73733151e-01
-1.50753096e-01 7.79927611e-01 4.85430062e-01 5.12200773e-01
-9.18991506e-01 8.34030449e-01 -8.84261012e-01 -3.46261591e-01
4.91549134e-01 5.26642084e-01 -6.58318520e-01 -6.11939132e-01
-8.92022908e-01 9.86607909e-01 2.85567760e-01 3.76662880e-01
-1.26566923e+00 -3.96272629e-01 -8.89717162e-01 -2.20390692e-01
2.25448254e-02 -4.98496383e-01 1.12545121e+00 -6.89041853e-01
-1.63234377e+00 1.07726419e+00 6.38844550e-01 -2.89401501e-01
3.66463006e-01 -3.69113892e-01 -8.62283528e-01 3.82982910e-01
3.94360051e-02 3.75116438e-01 8.57369959e-01 -1.25369477e+00
-6.87126935e-01 -4.83546406e-01 -3.36322457e-01 -2.77665943e-01
4.61381041e-02 1.31513728e-02 -4.25445974e-01 -7.46812940e-01
6.17262423e-01 -7.96572268e-01 -1.89030152e-02 -4.08116244e-02
-5.29373765e-01 -6.61437809e-01 1.00815344e+00 -9.63929296e-01
4.39227432e-01 -2.36796832e+00 -2.37091139e-01 3.10729802e-01
2.95618057e-01 7.74228036e-01 -5.41242242e-01 3.73907983e-01
-2.37763569e-01 -4.38277394e-01 -1.89908743e-01 -1.09752536e-01
2.07761154e-02 -1.36292234e-01 5.39750755e-01 5.29965520e-01
4.64921176e-01 7.44680464e-01 -7.16677725e-01 1.55734792e-01
4.45436597e-01 2.48178676e-01 -2.63956487e-01 2.36843735e-01
2.32736349e-01 1.22956894e-01 -4.30937320e-01 1.22806680e+00
8.07842016e-01 1.88362718e-01 -8.30999315e-02 -4.84933525e-01
3.05354178e-01 -3.70058715e-01 -9.27992821e-01 1.98620701e+00
-2.74228066e-01 6.09336078e-01 2.63284951e-01 -1.50431454e+00
2.05455017e+00 2.61401415e-01 6.33480132e-01 -8.27540994e-01
5.81652164e-01 4.55611765e-01 5.97652495e-02 -6.37017369e-01
5.06880619e-02 5.82949184e-02 -1.03554845e-01 -3.88560712e-01
3.99763435e-01 -4.38260496e-01 -2.73513943e-02 -2.34779820e-01
1.68006122e+00 -1.07933797e-01 -2.83214211e-01 -3.71631712e-01
4.84187245e-01 -2.98607409e-01 6.84001803e-01 7.13215232e-01
-1.95326418e-01 9.97033119e-01 1.37412190e-01 -6.43839777e-01
-1.07962906e+00 -1.05564368e+00 1.09059535e-01 1.80946793e-02
9.81122106e-02 1.73743278e-01 -6.58262849e-01 -9.90031660e-01
3.92143935e-01 5.74006438e-01 -6.24868810e-01 -7.49996901e-01
-3.37906241e-01 -4.55886543e-01 5.67627788e-01 7.04947472e-01
9.52631474e-01 -1.19183576e+00 -2.75223464e-01 2.02403352e-01
4.54974711e-01 -8.35526168e-01 1.90392762e-01 3.33883584e-01
-7.35885024e-01 -1.30222881e+00 -6.94051266e-01 -1.34419036e+00
7.50835896e-01 -2.00867757e-01 9.79355574e-01 1.48481160e-01
-6.91758096e-01 1.36237428e-01 -4.49816197e-01 -5.55436432e-01
-6.47471845e-01 -7.92682543e-02 2.47519277e-02 -1.16455197e-01
5.16543448e-01 -1.82242408e-01 -3.02009672e-01 3.03588331e-01
-6.44591570e-01 -4.51308429e-01 9.21698570e-01 1.27759135e+00
3.65644693e-01 6.60222888e-01 9.66886401e-01 -6.82812214e-01
3.22879881e-01 -1.87977299e-01 -4.93282259e-01 1.17481723e-01
-7.73420393e-01 -2.73131698e-01 3.81281555e-01 2.96588182e-01
-1.20764601e+00 5.78640886e-02 -6.81748927e-01 -6.40870452e-01
-3.66829783e-01 3.23824376e-01 -2.88588405e-01 -2.55309701e-01
5.12348950e-01 -1.10500064e-02 3.72247636e-01 -8.32228243e-01
-2.04613581e-01 9.13091183e-01 8.99780869e-01 -2.39104569e-01
5.56240559e-01 2.40255862e-01 -6.94914237e-02 -9.75954950e-01
-7.68849909e-01 -5.59867918e-01 -5.68128467e-01 -4.31517243e-01
7.65610993e-01 -9.05512691e-01 -5.78580737e-01 1.35476565e+00
-1.13443351e+00 -5.89091443e-02 -5.99819720e-01 5.46144009e-01
-4.41152900e-01 3.49683136e-01 -9.65182841e-01 -6.33816183e-01
-6.18233085e-01 -8.13852489e-01 1.06736708e+00 -1.71439692e-01
1.84649795e-01 -6.89594507e-01 -2.08997637e-01 4.60303634e-01
3.48643124e-01 5.97217560e-01 8.45255136e-01 -8.19272220e-01
-1.76124617e-01 -7.95730829e-01 -9.23562497e-02 1.31516838e+00
3.99709105e-01 -3.26399691e-02 -8.02592576e-01 -4.27703291e-01
3.77537727e-01 -2.65393347e-01 1.22265458e+00 3.60323578e-01
9.19503629e-01 3.46196264e-01 -2.03908384e-01 3.08289111e-01
1.51115525e+00 7.47144401e-01 9.38265085e-01 4.93188590e-01
6.43492341e-01 3.72910023e-01 8.30637395e-01 3.12607735e-01
-3.02934319e-01 3.77895981e-01 9.51719403e-01 -4.31574374e-01
-5.32128572e-01 -3.16121988e-02 3.90865386e-01 1.33216202e+00
1.13546349e-01 -2.04349011e-01 -8.86192441e-01 7.34068036e-01
-1.55259991e+00 -5.77853739e-01 -2.08710417e-01 1.50701642e+00
2.27066860e-01 4.91074592e-01 -4.18577075e-01 6.38282835e-01
1.03784299e+00 -3.33492815e-01 -6.95119858e-01 -3.53211701e-01
-4.74626184e-01 5.13758898e-01 4.87701535e-01 -9.38493833e-02
-1.06571186e+00 6.90131187e-01 6.32325077e+00 1.12908423e+00
-9.33717370e-01 3.42725404e-02 2.36899778e-01 2.94807553e-01
1.21868692e-01 -7.51919985e-01 -5.15177131e-01 2.75812656e-01
3.70070547e-01 6.39702916e-01 -1.52692467e-01 1.15977645e+00
-4.61179674e-01 1.45539805e-01 -1.00706804e+00 1.12320256e+00
3.14275354e-01 -1.28552675e+00 -2.24306345e-01 -3.18894893e-01
9.72095847e-01 3.24081063e-01 -3.73229384e-02 2.42016599e-01
6.43183768e-01 -8.26158464e-01 6.47481799e-01 4.20981795e-01
9.78700519e-01 -9.15380776e-01 1.54911041e+00 1.29191801e-02
-8.67918432e-01 -2.52641350e-01 -6.05564535e-01 1.52617350e-01
4.72371839e-03 1.21360266e+00 -7.89568007e-01 1.07266641e+00
9.52202320e-01 1.07360506e+00 -3.07809770e-01 8.89624596e-01
-3.83692384e-01 6.28545761e-01 -3.15224439e-01 1.92013234e-01
8.51784647e-02 5.90489954e-02 3.10951114e-01 8.55768442e-01
6.09836161e-01 -3.24064702e-01 5.15441336e-02 6.68954968e-01
4.32923771e-02 -4.03817654e-01 -8.27441692e-01 1.50430724e-01
4.65473861e-01 1.05420363e+00 -5.14845431e-01 -3.77244093e-02
-2.09267035e-01 9.77054477e-01 -7.23671392e-02 -9.45757180e-02
-4.71244454e-01 -8.32707942e-01 5.25852680e-01 -3.99690717e-01
7.45806217e-01 3.08144074e-02 -2.13233575e-01 -5.96849382e-01
2.77654678e-02 -9.13786352e-01 1.81227207e-01 -8.47572684e-01
-1.64726412e+00 6.84525430e-01 -6.29566491e-01 -1.32081783e+00
3.17780942e-01 -1.20751965e+00 -3.83097202e-01 4.80482340e-01
-1.14710784e+00 -1.35350454e+00 -3.56221884e-01 2.39784524e-01
6.32921994e-01 -9.90709782e-01 7.14139462e-01 2.65340537e-01
-7.04621673e-01 5.47875226e-01 3.69586408e-01 3.48099619e-01
1.85721964e-01 -1.06609976e+00 7.62950897e-01 6.58300936e-01
-2.50805259e-01 -2.98947468e-02 2.65757680e-01 -7.75770068e-01
-1.46116447e+00 -1.24453986e+00 5.58759212e-01 -1.01564609e-01
1.25141859e-01 -1.14307692e-02 -7.66467512e-01 2.75651813e-01
7.28237405e-02 8.00210759e-02 9.85980853e-02 -3.56345683e-01
-1.04966074e-01 -3.82099122e-01 -1.70673275e+00 -1.66925192e-01
1.20146227e+00 -4.70105588e-01 -7.42950439e-01 2.66998976e-01
5.38226843e-01 -3.68443787e-01 -1.35384560e+00 9.10793722e-01
2.35260606e-01 -8.14162910e-01 7.98636794e-01 -4.87831868e-02
4.14695323e-01 -4.30335075e-01 -2.89241821e-01 -1.39258993e+00
-6.66591644e-01 -3.20186913e-02 4.05875593e-02 1.23102736e+00
1.17340311e-01 -7.53887177e-01 1.00972652e+00 -1.28583729e-01
-9.07165229e-01 -9.94173408e-01 -1.01001179e+00 -1.31692374e+00
-2.15712607e-01 -4.91272956e-01 5.85849464e-01 8.51275384e-01
-4.17461783e-01 1.11255310e-02 -4.04149964e-02 2.55920708e-01
4.20273244e-01 -2.82763988e-01 2.88303792e-01 -1.45367241e+00
9.65891853e-02 -2.12789074e-01 -1.16972065e+00 -9.60580289e-01
3.26453596e-02 -9.79833245e-01 3.05737048e-01 -1.80112171e+00
-3.94960940e-01 -4.67663109e-01 -8.08765352e-01 4.96634036e-01
4.57868129e-01 3.72320116e-01 -9.18535218e-02 -6.57670945e-02
-2.39135668e-01 7.41104722e-01 1.46953690e+00 -7.48312056e-01
3.76729459e-01 -5.02681173e-02 -3.52404296e-01 5.63177109e-01
1.00014913e+00 -4.50645655e-01 -5.12518361e-02 -4.37693954e-01
-1.04750209e-01 -1.91039965e-01 2.52198130e-01 -1.45900393e+00
1.14977077e-01 4.86883938e-01 5.48965394e-01 -9.89610136e-01
1.06605552e-01 -8.50658536e-01 1.33053154e-01 1.06414545e+00
1.42721478e-02 -1.05813757e-01 3.66726443e-02 3.31232160e-01
-6.53872967e-01 -5.10380983e-01 8.08363318e-01 -2.55950153e-01
-1.10317993e+00 9.29016769e-02 -3.20595145e-01 -4.54652280e-01
1.13258553e+00 -4.35706556e-01 -1.61582068e-01 1.97560072e-01
-6.62675023e-01 -9.48601589e-02 3.49548757e-01 7.81103671e-01
1.11365449e+00 -1.52519858e+00 -7.61284888e-01 7.58540034e-01
2.08493590e-01 1.91726059e-01 3.61729026e-01 2.64929146e-01
-7.61142850e-01 3.30446303e-01 -4.00070071e-01 -8.37918878e-01
-8.26644838e-01 3.31697673e-01 5.63255787e-01 -2.60599613e-01
-6.94558859e-01 1.48701155e+00 -4.51523364e-02 -7.19171703e-01
-4.24818322e-03 -2.90573597e-01 -2.45497018e-01 -3.83804768e-01
-7.00703189e-02 6.76911056e-01 8.25637579e-01 -1.74979001e-01
-4.57756519e-01 6.90931499e-01 -7.17955753e-02 6.15090013e-01
1.56242979e+00 4.73355651e-01 -7.85946324e-02 1.44543648e-01
1.14660609e+00 -5.96905053e-01 -1.14916468e+00 -8.39904621e-02
3.98893729e-02 -2.92225689e-01 2.96720713e-01 -1.16650879e+00
-1.96088493e+00 5.93092918e-01 9.69864368e-01 1.71692520e-01
1.23806155e+00 -1.00058272e-01 1.00220311e+00 5.14211416e-01
6.63383245e-01 -1.49293375e+00 3.50667357e-01 5.79043329e-01
1.16292429e+00 -1.21018040e+00 -3.81729424e-01 -5.11940837e-01
-3.91756833e-01 1.29506671e+00 9.19435441e-01 -6.77074432e-01
8.45034540e-01 4.09771204e-01 2.35022560e-01 -9.91066992e-01
-3.54488015e-01 2.40122586e-01 1.06430143e-01 9.47797239e-01
-3.16918339e-03 -4.72184978e-02 8.72680545e-02 5.53362846e-01
4.38056327e-02 6.40624911e-02 -1.79307219e-02 1.23035347e+00
-4.96799499e-01 -8.16389918e-01 -5.17901443e-02 6.27697945e-01
-3.87140036e-01 3.61467808e-01 -2.39516214e-01 8.08994412e-01
4.21220005e-01 1.07653153e+00 4.75030057e-02 -9.24684405e-01
8.86705697e-01 -3.98544550e-01 6.47868276e-01 -6.72123790e-01
-7.63111770e-01 -4.02732968e-01 2.17205927e-01 -6.99062109e-01
-2.77138501e-01 -6.06714189e-01 -1.24329364e+00 3.16120349e-02
-6.82093024e-01 -9.26740319e-02 9.43847001e-01 7.93890655e-01
3.83442670e-01 1.06802273e+00 1.00046778e+00 -6.89027667e-01
-8.15299690e-01 -1.14983857e+00 -1.28744447e+00 4.05856788e-01
1.92293704e-01 -7.97258079e-01 -1.90661818e-01 -1.59994319e-01] | [7.382396221160889, 1.8610985279083252] |
1b5c1e33-7bd9-4019-bdd1-9e771081878f | cbeaf-adapting-enhanced-continual-pretraining | 2211.11363 | null | https://arxiv.org/abs/2211.11363v1 | https://arxiv.org/pdf/2211.11363v1.pdf | CBEAF-Adapting: Enhanced Continual Pretraining for Building Chinese Biomedical Language Model | Continual pretraining is a standard way of building a domain-specific pretrained language model from a general-domain language model. However, sequential task training may cause catastrophic forgetting, which affects the model performance in downstream tasks. In this paper, we propose a continual pretraining method for the BERT-based model, named CBEAF-Adapting (Chinese Biomedical Enhanced Attention-FFN Adapting). Its main idea is to introduce a small number of attention heads and hidden units inside each self-attention layer and feed-forward network. Using the Chinese biomedical domain as a running example, we trained a domain-specific language model named CBEAF-RoBERTa. We conduct experiments by applying models to downstream tasks. The results demonstrate that with only about 3% of model parameters trained, our method could achieve about 0.5%, 2% average performance gain compared to the best performing model in baseline and the domain-specific model, PCL-MedBERT, respectively. We also examine the forgetting problem of different pretraining methods. Our method alleviates the problem by about 13% compared to fine-tuning. | ['Tong Ruan', 'Qi Ye', 'Kui Xue', 'Yongyu Yan'] | 2022-11-21 | null | null | null | null | ['continual-pretraining'] | ['methodology'] | [ 7.41995946e-02 3.15479130e-01 5.58626577e-02 -4.24290985e-01
-6.98905349e-01 5.15854619e-02 3.11057687e-01 4.09887219e-03
-9.55102205e-01 1.03813601e+00 3.13461035e-01 -4.44843948e-01
3.82180601e-01 -4.12806004e-01 -7.61134505e-01 -4.34274495e-01
1.99109346e-01 5.05204082e-01 3.24424595e-01 -3.95497382e-01
-1.27893358e-01 3.28746855e-01 -6.85653210e-01 3.74845147e-01
1.17673683e+00 4.82890993e-01 7.49788880e-01 4.72495854e-01
-2.14543447e-01 8.31426740e-01 -8.65713358e-01 -3.04995149e-01
-2.37050071e-01 -5.23172319e-01 -9.82922077e-01 -8.73794630e-02
-1.55075744e-01 -3.97777468e-01 -3.63665879e-01 8.00256133e-01
6.56656802e-01 2.24453062e-01 4.43486184e-01 -7.05781639e-01
-1.14456189e+00 9.27792072e-01 -3.47320020e-01 5.23443162e-01
-3.20032775e-01 1.14038669e-01 3.80990744e-01 -1.16896558e+00
4.61525500e-01 1.25821137e+00 8.38189125e-01 1.06211293e+00
-1.08467937e+00 -7.62409270e-01 5.40046275e-01 9.09824297e-02
-1.43166053e+00 -5.14452398e-01 5.22278965e-01 -1.59683928e-01
1.30606771e+00 -3.15690368e-01 2.62635022e-01 1.24083459e+00
7.09174752e-01 9.11654413e-01 8.65168929e-01 -4.49492544e-01
1.31005943e-01 2.78343678e-01 4.28272814e-01 6.20961070e-01
2.49362841e-01 -9.05019939e-02 -3.94195467e-01 -1.08374797e-01
6.36498809e-01 2.34655052e-01 -3.00465107e-01 3.16705227e-01
-1.07514942e+00 6.41917884e-01 5.69075406e-01 6.78582191e-01
-4.36364740e-01 4.70792968e-03 3.00194293e-01 3.83283108e-01
8.29729199e-01 5.70231378e-01 -1.02784705e+00 1.70666635e-01
-8.70955288e-01 -2.70899504e-01 3.81147951e-01 1.14574063e+00
5.10327935e-01 2.91813999e-01 -5.35763562e-01 9.71867323e-01
-1.61625072e-02 2.53077060e-01 1.29736972e+00 -4.04049724e-01
2.27845743e-01 3.76968294e-01 2.58605834e-02 -2.64552116e-01
-6.06775165e-01 -7.95483291e-01 -1.09795809e+00 -2.88185596e-01
8.90565664e-02 -4.94788498e-01 -1.50923598e+00 2.11770082e+00
-1.51976600e-01 2.28169635e-01 2.13081151e-01 5.50945282e-01
7.03537464e-01 5.92745602e-01 6.58975303e-01 -3.45222026e-01
1.22256374e+00 -1.50424075e+00 -9.70699072e-01 -6.59041464e-01
9.08529580e-01 -6.07488334e-01 1.56207335e+00 1.63581774e-01
-1.14772522e+00 -8.12202394e-01 -9.47209299e-01 -3.79712224e-01
-3.29912901e-01 2.39341781e-01 3.20238084e-01 1.36502907e-01
-1.32957888e+00 7.26161063e-01 -1.02182066e+00 -5.12056112e-01
2.42619336e-01 3.07246029e-01 -2.68012315e-01 -2.47097671e-01
-1.61074328e+00 1.15979421e+00 6.35670185e-01 -1.42896369e-01
-1.23029125e+00 -9.78370786e-01 -5.74948728e-01 4.05973524e-01
5.56675047e-02 -9.15025711e-01 1.50237083e+00 -8.98212969e-01
-1.55585778e+00 5.80624104e-01 -3.99547905e-01 -6.06670558e-01
3.58912140e-01 -6.29060507e-01 -5.87493896e-01 -3.15955758e-01
9.19478014e-02 7.01164067e-01 7.48534679e-01 -8.82582068e-01
-4.73095655e-01 -8.97924528e-02 -1.26879409e-01 1.29956916e-01
-5.21401346e-01 -1.59346998e-01 -6.35673225e-01 -9.55370307e-01
-1.92388743e-01 -8.04624438e-01 -5.26379287e-01 -3.06059152e-01
-2.26801157e-01 -2.81764507e-01 5.44057548e-01 -9.47100401e-01
1.57425964e+00 -2.32026839e+00 -9.78832990e-02 -4.54207927e-01
1.53115630e-01 4.69944298e-01 -4.62381124e-01 1.05801009e-01
-1.98849544e-01 1.62863731e-01 -4.54587549e-01 -6.27465308e-01
-5.78043282e-01 3.18244696e-01 -7.28949681e-02 1.02507748e-01
4.12024051e-01 1.07802665e+00 -1.11710644e+00 -2.42274925e-01
-3.06031287e-01 5.77009380e-01 -7.22097993e-01 3.17558289e-01
-1.77892089e-01 4.66902435e-01 -2.71079183e-01 3.70552659e-01
6.18900537e-01 -4.85981822e-01 1.43373072e-01 8.81379768e-02
1.51700661e-01 3.86504382e-01 -3.28209758e-01 2.02137709e+00
-6.13213778e-01 3.29274058e-01 -1.92620516e-01 -7.84365356e-01
8.56753051e-01 3.86896104e-01 -7.50790536e-03 -8.18533897e-01
3.83810624e-02 8.38371143e-02 2.51311362e-01 -2.48976693e-01
4.10532981e-01 -5.93215764e-01 -1.82546631e-01 3.35985750e-01
4.56162393e-01 4.22013402e-01 -6.39081299e-02 2.94354081e-01
1.35598457e+00 -7.83934593e-02 3.91425252e-01 -5.65174282e-01
5.94006658e-01 -6.47643954e-02 8.36326838e-01 7.81388223e-01
-4.56090569e-01 5.37129760e-01 1.01441167e-01 -3.90253723e-01
-8.96204412e-01 -8.08772683e-01 9.26290229e-02 1.43699610e+00
-2.64974684e-01 -4.04303372e-01 -7.22002268e-01 -8.81310821e-01
-1.45650446e-01 1.16767943e+00 -5.96311986e-01 -8.27403188e-01
-7.03770995e-01 -9.95145798e-01 4.70394582e-01 7.14538038e-01
7.24504173e-01 -1.18783748e+00 -2.22782373e-01 5.16099453e-01
-2.21758336e-01 -9.30948794e-01 -8.71991575e-01 3.99753392e-01
-1.15551436e+00 -4.90922779e-01 -1.30467629e+00 -1.01204872e+00
8.54556024e-01 2.20752835e-01 1.27325737e+00 1.19077906e-01
1.92800999e-01 2.16538720e-02 -2.28660151e-01 -5.58093071e-01
-4.41170067e-01 6.58151627e-01 1.27853692e-01 -2.45990902e-01
4.77165788e-01 -4.10646647e-01 -5.47307372e-01 -3.39430682e-02
-8.09255838e-01 1.71643168e-01 1.04186153e+00 1.26584959e+00
5.08679926e-01 -2.00312644e-01 9.44361329e-01 -1.34297776e+00
9.95394111e-01 -5.55044949e-01 -5.32854721e-02 3.97772014e-01
-9.12580073e-01 4.08716440e-01 9.41422939e-01 -7.24021316e-01
-1.43331897e+00 -6.19142689e-02 -3.27041656e-01 -5.02678514e-01
-5.56515902e-03 7.17665493e-01 -1.20365195e-01 4.16542321e-01
7.08932281e-01 3.77794832e-01 -2.13560507e-01 -9.81773496e-01
2.72758484e-01 4.89751041e-01 5.12027383e-01 -2.72343546e-01
3.38127583e-01 6.14019819e-02 -7.06019521e-01 -5.89463413e-01
-1.14418113e+00 -2.32591197e-01 -6.44060969e-01 3.40547532e-01
6.22130394e-01 -1.21559525e+00 -3.86881307e-02 5.00492811e-01
-1.30722249e+00 -7.55650997e-01 -2.09274396e-01 5.21641612e-01
-2.32031301e-01 6.73336387e-02 -1.02072108e+00 -4.92625833e-01
-6.71705425e-01 -8.47440958e-01 7.04950273e-01 -8.28033499e-03
-2.18243361e-01 -1.10486388e+00 -1.44362580e-02 -9.67445597e-02
7.65150011e-01 -2.25876927e-01 1.21270585e+00 -7.50907719e-01
8.27755108e-02 1.41901374e-01 -5.70792221e-02 4.75440532e-01
2.94370621e-01 -7.83887088e-01 -9.50118303e-01 -6.33800447e-01
2.64153332e-01 -1.56389654e-01 1.11405659e+00 3.51277888e-01
1.24397337e+00 -3.67308766e-01 -5.91263652e-01 4.90630955e-01
1.29006958e+00 4.18058097e-01 5.12587845e-01 2.06434742e-01
3.99519920e-01 3.90728153e-02 5.06168306e-01 1.86549634e-01
2.29797572e-01 1.83586195e-01 -1.64319590e-01 -3.62693340e-01
-4.51061755e-01 -3.47007334e-01 4.80881631e-01 1.28889966e+00
1.70553952e-01 -1.71270728e-01 -1.11523449e+00 8.25477779e-01
-1.71402955e+00 -5.14392853e-01 2.81325459e-01 1.91190100e+00
1.31642520e+00 4.17011142e-01 -3.11363935e-01 -2.54078865e-01
7.27356374e-01 -3.33820373e-01 -8.03532124e-01 -4.58425105e-01
1.39514022e-02 2.99400300e-01 3.51249725e-01 6.46251798e-01
-8.82974088e-01 1.37619531e+00 6.36420059e+00 6.88874900e-01
-1.25701046e+00 6.45525098e-01 8.90180826e-01 -1.71267912e-01
-1.17960304e-01 -2.34000266e-01 -1.02581859e+00 5.59327662e-01
1.55830228e+00 -5.55224121e-01 1.94811314e-01 9.74945009e-01
2.76475787e-01 1.99621975e-01 -1.04664135e+00 5.78571320e-01
1.23978019e-01 -9.16344047e-01 2.59138703e-01 -3.26293051e-01
7.48979449e-01 2.99037695e-01 -9.85971391e-02 1.09076595e+00
4.33065057e-01 -8.38898778e-01 4.22095358e-01 4.71047044e-01
9.27476227e-01 -7.13122845e-01 8.22406232e-01 6.82821870e-01
-7.06600130e-01 -1.99936070e-02 -7.19058394e-01 2.91523971e-02
2.22068891e-01 7.09562302e-01 -9.80868518e-01 3.86827230e-01
7.88652539e-01 5.83232880e-01 -6.57014549e-01 9.01552200e-01
-2.54712850e-01 8.84808719e-01 -3.17032598e-02 1.87861204e-01
1.98326692e-01 5.55857658e-01 1.38202175e-01 1.46748543e+00
5.31810403e-01 2.54995078e-01 -2.08203774e-02 4.98463899e-01
-3.99355054e-01 1.39869656e-03 -3.17603350e-01 -4.00490947e-02
3.46152723e-01 7.79397726e-01 -2.95425355e-01 -7.16465890e-01
-3.82625341e-01 1.39162397e+00 8.04002762e-01 4.73413885e-01
-9.50716972e-01 -5.68980575e-01 4.86093342e-01 1.86295390e-01
1.95321634e-01 -1.64258465e-01 -2.00980872e-01 -1.33326519e+00
-3.91754240e-01 -7.94625163e-01 4.41514969e-01 -8.15533876e-01
-1.36208367e+00 1.13762200e+00 -2.67425448e-01 -7.83718944e-01
-1.46720903e-02 -4.69033271e-01 -5.26613653e-01 1.19692898e+00
-1.83882737e+00 -9.31887746e-01 -1.72162756e-01 7.07654059e-01
1.00643885e+00 -1.66162118e-01 1.02269256e+00 5.28621554e-01
-6.43897772e-01 8.64655614e-01 2.53019035e-01 7.35316649e-02
1.10969472e+00 -9.26343918e-01 5.72629571e-01 8.46493602e-01
-4.19814587e-01 9.82006371e-01 4.88721192e-01 -8.28799188e-01
-8.01111460e-01 -1.60482883e+00 1.59675395e+00 -3.01746756e-01
3.53102237e-01 -3.58210683e-01 -1.35871494e+00 1.03060424e+00
4.55645174e-01 -1.22291766e-01 6.65199935e-01 2.56868750e-01
-2.32132357e-02 -3.42202932e-02 -1.05582094e+00 6.36157393e-01
1.23400557e+00 -2.43497208e-01 -9.75969732e-01 3.49621505e-01
1.31461644e+00 -2.41252884e-01 -6.99871719e-01 4.31259304e-01
1.58108640e-02 -3.66941810e-01 6.62415385e-01 -1.05388117e+00
2.71873891e-01 -5.13139293e-02 3.46832335e-01 -1.64302552e+00
-1.08551371e+00 -5.39034486e-01 -1.78536206e-01 1.09246600e+00
7.52607107e-01 -5.51730454e-01 5.01524329e-01 5.54877579e-01
-4.87102419e-01 -6.35205507e-01 -6.75237298e-01 -9.51987803e-01
5.74644864e-01 -1.61542341e-01 4.40349728e-01 1.05751872e+00
-5.13702631e-02 9.70885158e-01 -5.11817217e-01 -4.32354510e-02
1.30368516e-01 -3.77241492e-01 2.67447591e-01 -9.23469365e-01
-3.36193264e-01 -1.40557170e-01 2.96784788e-01 -1.19997478e+00
1.45380363e-01 -1.00871825e+00 1.71967715e-01 -1.48893619e+00
2.48132989e-01 -3.01088244e-01 -1.02278864e+00 9.38555002e-01
-6.19460464e-01 -2.17647269e-01 1.53672054e-01 1.59230649e-01
-5.53444564e-01 7.40998149e-01 1.28276134e+00 -3.17037851e-01
-3.71431977e-01 -6.16929829e-02 -1.08942974e+00 5.75863481e-01
1.02531135e+00 -6.87127888e-01 -5.17272949e-01 -9.67328787e-01
-2.64534950e-01 -1.81513339e-01 -2.37164035e-01 -9.95663702e-01
3.09372276e-01 -2.64993738e-02 5.98786294e-01 -4.23105866e-01
6.79763407e-02 -5.55511713e-01 -1.26907259e-01 9.32329476e-01
-4.96833354e-01 3.31266254e-01 6.26554191e-01 5.42339385e-01
-1.49751440e-01 -1.47989199e-01 1.03294885e+00 -4.35811102e-01
-8.71352136e-01 3.32923591e-01 -5.65142393e-01 8.40103403e-02
6.52717352e-01 4.41708624e-01 -2.84911066e-01 -2.54071414e-01
-1.21186531e+00 2.58218259e-01 2.09724754e-02 5.41047156e-01
7.10986912e-01 -1.36783087e+00 -8.23591650e-01 3.59276623e-01
3.19430009e-02 -1.31188586e-01 4.42335755e-01 6.62150919e-01
-2.11188480e-01 7.47008801e-01 -9.89132002e-02 -2.46901959e-01
-8.99335802e-01 1.00498569e+00 3.15089703e-01 -7.20415533e-01
-5.32894611e-01 1.11088264e+00 6.24320567e-01 -3.51782441e-01
1.99044883e-01 -6.04385853e-01 -9.57279280e-02 -4.00452524e-01
6.71104312e-01 3.40715349e-02 1.69712648e-01 -1.02350578e-01
-5.27251124e-01 -4.73643988e-02 -6.64580107e-01 1.07691139e-01
1.34944081e+00 -1.84121579e-01 1.91366926e-01 4.08078611e-01
1.03564024e+00 -2.87463337e-01 -1.11247301e+00 -5.40210426e-01
1.95612460e-01 1.39067411e-01 1.80039719e-01 -1.19793606e+00
-8.91024292e-01 1.03691828e+00 5.80823839e-01 -3.71897817e-01
1.35174704e+00 -1.06312506e-01 9.13198948e-01 5.83221495e-01
4.46917683e-01 -1.03021479e+00 6.15160353e-02 9.71658945e-01
1.10449028e+00 -1.12642586e+00 -2.83215076e-01 1.17508389e-01
-7.50366807e-01 6.85381532e-01 1.08318698e+00 -7.82120153e-02
9.38121259e-01 2.65977621e-01 1.05320692e-01 2.22970366e-01
-1.21431470e+00 -7.03116432e-02 1.36646777e-01 3.74348819e-01
7.94645190e-01 -1.88809335e-01 -6.06235445e-01 1.12896168e+00
5.17854616e-02 3.98009807e-01 2.40776166e-01 1.03278291e+00
-4.57754463e-01 -1.09946716e+00 1.22980468e-01 3.49584639e-01
-6.45008922e-01 -5.82750082e-01 -1.67034104e-01 8.14006031e-01
7.31091201e-02 6.39425993e-01 -2.81678569e-02 -3.00099373e-01
4.89635408e-01 5.48619568e-01 9.26004797e-02 -1.01523399e+00
-6.88227713e-01 2.35075310e-01 -1.23409010e-01 -2.32432976e-01
-3.99279520e-02 -1.65559262e-01 -1.33626199e+00 -1.68407306e-01
-2.29924515e-01 2.67649949e-01 1.12594292e-01 7.76281774e-01
8.39108229e-01 1.09447551e+00 1.35733724e-01 -3.91648382e-01
-5.86542904e-01 -1.46187377e+00 -3.59803855e-01 3.23600978e-01
3.79437983e-01 -4.60639715e-01 -1.49232537e-01 2.17982486e-01] | [10.412894248962402, 8.104899406433105] |
70315324-b75a-43bd-a0b5-514d4347d27f | prediction-guided-distillation-for-dense | 2203.05469 | null | https://arxiv.org/abs/2203.05469v2 | https://arxiv.org/pdf/2203.05469v2.pdf | Prediction-Guided Distillation for Dense Object Detection | Real-world object detection models should be cheap and accurate. Knowledge distillation (KD) can boost the accuracy of a small, cheap detection model by leveraging useful information from a larger teacher model. However, a key challenge is identifying the most informative features produced by the teacher for distillation. In this work, we show that only a very small fraction of features within a ground-truth bounding box are responsible for a teacher's high detection performance. Based on this, we propose Prediction-Guided Distillation (PGD), which focuses distillation on these key predictive regions of the teacher and yields considerable gains in performance over many existing KD baselines. In addition, we propose an adaptive weighting scheme over the key regions to smooth out their influence and achieve even better performance. Our proposed approach outperforms current state-of-the-art KD baselines on a variety of advanced one-stage detection architectures. Specifically, on the COCO dataset, our method achieves between +3.1% and +4.6% AP improvement using ResNet-101 and ResNet-50 as the teacher and student backbones, respectively. On the CrowdHuman dataset, we achieve +3.2% and +2.0% improvements in MR and AP, also using these backbones. Our code is available at https://github.com/ChenhongyiYang/PGD. | ['Elliot J. Crowley', 'Amos Storkey', 'Mateusz Ochal', 'Chenhongyi Yang'] | 2022-03-10 | null | null | null | null | ['dense-object-detection'] | ['computer-vision'] | [-3.53412956e-01 2.21942008e-01 -3.75592560e-01 -1.33354723e-01
-9.74296987e-01 -3.94798696e-01 7.55116761e-01 2.05370978e-01
-6.74923480e-01 4.97388661e-01 9.72292498e-02 -4.49738652e-02
2.95177281e-01 -6.37409687e-01 -8.86618614e-01 -5.66727936e-01
9.87702981e-02 6.58091009e-01 1.00089633e+00 -1.16019301e-01
-8.65543932e-02 3.88724715e-01 -1.46907473e+00 1.11061320e-01
9.96648073e-01 8.40740025e-01 3.64807397e-01 8.06957841e-01
1.39788866e-01 8.94212067e-01 -7.47783363e-01 -5.05034029e-01
3.31751287e-01 1.90978125e-02 -6.96084082e-01 -4.18116659e-01
1.06327367e+00 -8.29833627e-01 -7.12986410e-01 1.12777722e+00
8.67757857e-01 1.81272268e-01 6.75098181e-01 -1.07210934e+00
-7.35042989e-01 8.57752144e-01 -9.77701724e-01 5.92616379e-01
4.75018807e-02 3.43730152e-01 1.10155523e+00 -1.28791797e+00
4.43367541e-01 1.32869816e+00 5.39249003e-01 5.82689464e-01
-1.19971526e+00 -1.09019458e+00 3.06770980e-01 3.48736227e-01
-1.56801701e+00 -4.03487533e-01 4.16655421e-01 -4.74263370e-01
8.68577778e-01 -8.33644718e-02 5.19136906e-01 8.18955600e-01
-3.93636733e-01 1.49357533e+00 8.93991828e-01 -3.48055065e-01
-1.07378691e-01 3.32945198e-01 3.19633663e-01 9.60316837e-01
4.26062614e-01 8.86791795e-02 -4.86483902e-01 5.91939725e-02
5.79948902e-01 -1.69119641e-01 -2.58272856e-01 -2.70346403e-01
-8.51342082e-01 9.81775701e-01 8.36551070e-01 1.24750403e-03
-2.12636307e-01 2.03857958e-01 8.14847872e-02 -9.86213163e-02
6.65624022e-01 5.69861412e-01 -4.11616057e-01 -1.57547846e-01
-9.92415071e-01 3.56394917e-01 6.19908452e-01 9.60830092e-01
8.02258968e-01 -6.80180937e-02 -6.28158331e-01 8.22148740e-01
2.11709321e-01 6.85520291e-01 2.48891219e-01 -8.41833115e-01
4.70829010e-01 6.58429563e-01 2.38770336e-01 -7.31084645e-01
-1.92730322e-01 -7.00048387e-01 -2.52495646e-01 8.69061518e-03
5.20521641e-01 -1.88736945e-01 -1.23568475e+00 1.75317240e+00
8.95804882e-01 7.06433296e-01 -1.48405418e-01 7.84748375e-01
1.20844185e+00 5.48196733e-01 2.07081884e-01 2.49897018e-01
1.44058335e+00 -1.46902537e+00 -1.24824896e-01 -2.85048753e-01
6.66336477e-01 -6.98176146e-01 1.01716530e+00 2.81401843e-01
-1.02936208e+00 -5.71213782e-01 -8.98684204e-01 -1.88319117e-01
-1.14492342e-01 3.96876901e-01 3.75098199e-01 2.74513364e-01
-1.02415764e+00 2.89288610e-01 -9.32149947e-01 -2.80378491e-01
8.39038968e-01 1.92946702e-01 -5.95049448e-02 -2.57508457e-01
-9.02689338e-01 9.88531113e-01 2.60691375e-01 -3.94569010e-01
-1.42093194e+00 -1.18436301e+00 -4.96628821e-01 7.48857185e-02
5.42207658e-01 -4.13763911e-01 1.79690242e+00 -2.25791901e-01
-1.23913717e+00 7.87408173e-01 9.95329246e-02 -7.04870284e-01
7.79077828e-01 -7.72585571e-01 1.98645040e-01 2.77613252e-01
3.36363465e-01 1.19204783e+00 7.14540064e-01 -9.41915572e-01
-1.11157036e+00 -1.73303083e-01 4.93333228e-02 3.14961761e-01
-4.48302060e-01 9.94204357e-03 -8.92011821e-01 -5.91995895e-01
-1.68201134e-01 -8.81961703e-01 -3.04255486e-01 4.41188514e-02
-4.91325498e-01 -6.77861094e-01 9.10126925e-01 -4.99682069e-01
1.07374918e+00 -2.05018187e+00 -1.84041321e-01 -2.26441547e-01
7.62727559e-01 8.09718847e-01 -3.18395346e-01 -1.46106230e-02
3.97463799e-01 -1.36233360e-01 2.72950113e-01 -4.81919438e-01
-4.50711809e-02 5.56278825e-02 -3.18370014e-01 4.54811662e-01
3.82012993e-01 9.81761098e-01 -1.06239963e+00 -4.85071748e-01
8.63661915e-02 5.15406132e-01 -5.90370417e-01 4.20233369e-01
-2.84449071e-01 2.27328524e-01 -5.30931771e-01 5.83464861e-01
6.20192230e-01 -3.71502876e-01 -3.23193520e-01 9.26396698e-02
-8.68622139e-02 5.19290447e-01 -1.15941811e+00 1.45716560e+00
-1.01082571e-01 7.57892251e-01 1.03309877e-01 -8.76408994e-01
8.35715055e-01 -5.09285629e-02 1.70659587e-01 -5.15297949e-01
1.52879074e-01 5.99112548e-02 6.56811446e-02 -1.08945385e-01
7.05072463e-01 4.27777320e-01 2.56359518e-01 4.13219482e-01
2.36121580e-01 2.18630042e-02 3.27866763e-01 7.31646895e-01
1.22805560e+00 -2.25349963e-01 2.35296935e-01 -2.65338391e-01
1.04761451e-01 5.97111918e-02 6.47940934e-01 1.14871752e+00
-5.14232457e-01 5.11955082e-01 3.49874765e-01 -3.40629935e-01
-8.21691811e-01 -9.01830137e-01 -1.57957286e-01 1.56248534e+00
2.08202466e-01 -3.43408287e-01 -7.25648999e-01 -9.13497210e-01
3.09895128e-01 5.32359600e-01 -6.30708814e-01 -1.22201398e-01
-5.78796744e-01 -6.70247376e-01 9.19476748e-01 8.46870244e-01
6.44942522e-01 -6.93165362e-01 -4.76474673e-01 1.14368983e-01
-4.12140228e-03 -1.31780756e+00 -4.13847268e-01 -5.77770965e-03
-6.12386346e-01 -1.15455568e+00 -7.83638000e-01 -6.01825595e-01
6.52976930e-01 9.59215105e-01 1.21741939e+00 2.47638270e-01
-4.04390097e-01 2.23209947e-01 -3.37191254e-01 -7.72821486e-01
-2.82133192e-01 4.17949945e-01 2.14158565e-01 -4.67777491e-01
6.15575314e-01 -2.43798599e-01 -7.01844156e-01 3.02908242e-01
-3.95900965e-01 1.74323320e-01 7.04187274e-01 6.69349194e-01
4.92917210e-01 -2.83000350e-01 3.96981984e-01 -8.74132931e-01
3.11541021e-01 -4.57705230e-01 -7.52307117e-01 1.54910073e-01
-7.47853637e-01 1.00104541e-01 2.98557729e-01 -7.43164599e-01
-1.02602708e+00 1.13148652e-01 -9.18493513e-03 -7.49839962e-01
-7.40609914e-02 -6.40515387e-02 2.58608580e-01 -2.25710556e-01
1.10240567e+00 2.10560501e-01 -3.63344938e-01 -7.32074380e-01
5.32783031e-01 4.73242879e-01 6.15114987e-01 -7.27878690e-01
1.05100071e+00 2.97710747e-01 -4.99158680e-01 -5.86799681e-01
-1.25282550e+00 -7.58987486e-01 -4.60377753e-01 5.66244684e-02
4.60381687e-01 -1.57625186e+00 -4.50894982e-01 3.62674147e-01
-9.97191310e-01 -4.50038910e-01 -3.49643767e-01 4.37884718e-01
2.24800836e-02 8.97205770e-02 -7.46024251e-01 -6.65225804e-01
-5.60365558e-01 -1.19860029e+00 9.51349437e-01 5.77509403e-01
1.29994214e-01 -5.03763616e-01 2.36317858e-01 4.82825100e-01
3.73087704e-01 -2.00320154e-01 1.76658973e-01 -1.06208706e+00
-7.17865467e-01 -1.67818442e-01 -6.17585480e-01 2.54356563e-01
-4.07302886e-01 1.61510989e-01 -1.17293084e+00 -4.71637219e-01
-6.17645025e-01 -6.16811156e-01 1.39388406e+00 2.38414541e-01
9.34319019e-01 -3.58627945e-01 -6.38856769e-01 5.30682802e-01
9.78158653e-01 -2.36117452e-01 9.51379910e-02 1.30784675e-01
8.50718439e-01 2.32384130e-01 8.15161824e-01 3.43432963e-01
7.40785420e-01 7.04717040e-01 4.78470534e-01 -7.12727308e-02
-6.27951562e-01 -7.23241746e-01 3.97831202e-01 4.50318545e-01
-5.88358892e-03 8.98588970e-02 -1.12785912e+00 9.14030254e-01
-1.79957414e+00 -7.49799132e-01 -1.04519226e-01 1.98656523e+00
1.17357183e+00 2.55949825e-01 4.52696204e-01 -2.85438418e-01
7.70416796e-01 9.48268771e-02 -7.07501948e-01 1.98883370e-01
1.17285937e-01 1.70064896e-01 5.47578156e-01 5.77336550e-01
-1.32822883e+00 1.44854295e+00 5.63471794e+00 1.18931246e+00
-9.72060382e-01 3.67841393e-01 5.05729795e-01 -3.86796236e-01
1.87236607e-01 -3.21284264e-01 -1.68295228e+00 2.93581933e-01
6.75368667e-01 -9.05547738e-02 2.06817929e-02 1.39214194e+00
-2.00259518e-02 -2.29992881e-01 -1.04388106e+00 7.05542684e-01
-3.52311023e-02 -1.34725416e+00 4.00241092e-02 -6.66937754e-02
1.08875406e+00 6.02341652e-01 2.54681319e-01 8.26565504e-01
1.04422069e+00 -8.25951040e-01 6.97049916e-01 -1.13044262e-01
4.97014582e-01 -5.92070401e-01 5.09349465e-01 5.46033978e-01
-1.21242642e+00 2.74015479e-02 -6.06968582e-01 -4.10050042e-02
-4.35650289e-01 6.89789891e-01 -1.49192488e+00 -8.25346559e-02
9.07360137e-01 5.49730599e-01 -8.22820485e-01 1.33016014e+00
-8.94729018e-01 1.06440413e+00 -6.30187452e-01 -1.09270304e-01
3.83126765e-01 4.38945919e-01 5.56917548e-01 1.53507352e+00
5.12244180e-02 2.20853105e-01 3.53975743e-01 8.04917097e-01
-6.03506982e-01 -7.76535347e-02 -3.26730728e-01 3.38742197e-01
1.06792808e+00 1.46980226e+00 -4.58209097e-01 -6.35402799e-01
-4.08004552e-01 5.45017302e-01 8.10007453e-01 2.29540095e-01
-1.03224480e+00 -2.38403663e-01 8.60972524e-01 5.75355515e-02
5.87993741e-01 5.66236712e-02 5.41868359e-02 -1.11227453e+00
-1.74610123e-01 -7.16787398e-01 4.96432513e-01 -3.42001230e-01
-1.07550478e+00 2.58483946e-01 -2.86181848e-02 -7.52084255e-01
1.47218903e-04 -3.43264163e-01 -7.31288135e-01 7.56248474e-01
-1.71321917e+00 -1.22418463e+00 -4.49019164e-01 4.29510981e-01
6.16412163e-01 1.35920597e-02 3.71587068e-01 2.70847946e-01
-8.96890163e-01 9.51102018e-01 8.27410966e-02 5.20631075e-01
8.57441306e-01 -1.50984836e+00 6.63965702e-01 1.06506574e+00
5.08610845e-01 3.87960881e-01 6.26765668e-01 -5.12283564e-01
-1.07680607e+00 -1.44434536e+00 5.98279655e-01 -8.37443352e-01
6.45290971e-01 -2.25571632e-01 -1.00554204e+00 7.40639627e-01
-6.33883625e-02 3.47136408e-01 2.93112874e-01 2.79667139e-01
-7.73007572e-01 -6.71568676e-04 -1.09959304e+00 4.44199264e-01
9.75324810e-01 -2.29958773e-01 -6.51933610e-01 4.34523463e-01
9.28433836e-01 -8.12846303e-01 -6.86689138e-01 3.45057875e-01
3.85161489e-01 -7.20147908e-01 1.11840606e+00 -5.81903040e-01
3.72239202e-01 -2.89436847e-01 1.15582511e-01 -1.27881527e+00
-5.19703507e-01 -4.66514975e-01 -6.19727373e-01 1.11321771e+00
3.80936205e-01 -3.24720442e-01 1.06288803e+00 4.93627995e-01
-9.73172337e-02 -9.04756010e-01 -7.41052210e-01 -8.16316426e-01
1.33450896e-01 -3.61140728e-01 5.26783407e-01 8.74978840e-01
-3.03628385e-01 5.42710066e-01 -2.40438268e-01 4.50478256e-01
7.40495265e-01 4.13299836e-02 1.25466394e+00 -1.16670144e+00
-3.11417222e-01 -5.13900638e-01 -2.34841436e-01 -1.68544161e+00
-1.00912601e-01 -7.53126383e-01 1.37539908e-01 -1.35228205e+00
5.65778315e-01 -6.89363837e-01 -3.58919799e-01 7.38134503e-01
-6.67949617e-01 4.15586948e-01 4.88275558e-01 3.44608843e-01
-9.77360010e-01 5.43669939e-01 1.28827739e+00 -1.50962830e-01
-3.16079795e-01 6.91125840e-02 -7.85779595e-01 8.49513829e-01
7.46540427e-01 -7.51514792e-01 -1.90195531e-01 -6.44076884e-01
-2.48799533e-01 -4.46612537e-01 2.67007798e-01 -9.85996604e-01
5.45195282e-01 -9.77967530e-02 3.75822902e-01 -6.41670585e-01
3.61211121e-01 -2.74243981e-01 -6.00286245e-01 6.37757957e-01
-3.20644915e-01 -3.46816033e-01 2.48211369e-01 6.11103773e-01
7.80598149e-02 -4.76176701e-02 1.02868247e+00 2.93004178e-02
-9.18455422e-01 5.17295122e-01 1.42443582e-01 4.73486722e-01
1.14603877e+00 1.41169757e-01 -8.39069724e-01 -1.61544845e-01
-3.21554005e-01 6.43124163e-01 1.85142174e-01 4.20965463e-01
4.78136241e-01 -1.07797337e+00 -9.79596794e-01 -2.43319925e-02
1.23166874e-01 5.02842903e-01 -3.92813720e-02 8.44846070e-01
-3.73384565e-01 4.00910079e-01 1.85116619e-01 -7.08489060e-01
-1.42931986e+00 2.66696155e-01 2.50928938e-01 -3.41743916e-01
-5.99934340e-01 1.61026144e+00 4.06575114e-01 -2.93294460e-01
6.24366641e-01 -3.75735670e-01 -1.38905004e-01 -5.86385606e-03
9.52357888e-01 6.86057687e-01 -2.13030368e-01 -5.87545156e-01
-3.50748479e-01 2.14748770e-01 -7.05749273e-01 2.52836913e-01
1.28950822e+00 2.73665428e-01 4.25900817e-01 -2.36276984e-01
8.09183180e-01 7.75657818e-02 -1.74720597e+00 -8.46382618e-01
-1.26405954e-01 -5.74002981e-01 2.28210986e-01 -8.19382489e-01
-1.14627516e+00 9.79707062e-01 4.10116643e-01 -7.16136992e-02
7.85109878e-01 4.48001653e-01 6.80382073e-01 4.77301925e-01
1.96746141e-01 -8.51099789e-01 3.10418040e-01 5.30093312e-01
5.37411511e-01 -1.50887012e+00 1.53232798e-01 -3.65846932e-01
-4.44425285e-01 6.16883576e-01 1.30517948e+00 -1.89466953e-01
3.18231761e-01 2.56527573e-01 -1.43534958e-01 -1.67850450e-01
-9.71167028e-01 -5.29076636e-01 4.17923957e-01 5.06295145e-01
2.52736509e-01 2.79121697e-01 2.03181468e-02 4.94348943e-01
-2.05437452e-01 -2.50509918e-01 3.63294214e-01 6.97609186e-01
-1.14383185e+00 -6.83405221e-01 -4.44234401e-01 4.62571532e-01
-5.11458218e-01 -1.32794932e-01 -4.89849776e-01 8.61815035e-01
1.01229541e-01 8.28137457e-01 -1.86102256e-01 -4.43686515e-01
4.05509412e-01 -2.00396255e-01 2.97299117e-01 -9.53758299e-01
-4.98301446e-01 -2.08807185e-01 -9.09067094e-02 -6.68470860e-01
-8.88590366e-02 -4.42789435e-01 -1.26840019e+00 -5.46474695e-01
-6.57514572e-01 1.33547708e-01 3.32411796e-01 6.17484510e-01
4.62523311e-01 2.82638669e-01 2.83061713e-01 -6.76578760e-01
-9.90035653e-01 -1.22678638e+00 -2.49549791e-01 7.45941326e-02
3.06541234e-01 -9.07853603e-01 -1.91696063e-01 -3.67756546e-01] | [9.2761869430542, 1.3079094886779785] |
4822f522-2ca5-406e-b19c-b81c8e8942c4 | a-novel-deep-reinforcement-learning-based | 2107.00931 | null | https://arxiv.org/abs/2107.00931v1 | https://arxiv.org/pdf/2107.00931v1.pdf | A Novel Deep Reinforcement Learning Based Stock Direction Prediction using Knowledge Graph and Community Aware Sentiments | Stock market prediction has been an important topic for investors, researchers, and analysts. Because it is affected by too many factors, stock market prediction is a difficult task to handle. In this study, we propose a novel method that is based on deep reinforcement learning methodologies for the direction prediction of stocks using sentiments of community and knowledge graph. For this purpose, we firstly construct a social knowledge graph of users by analyzing relations between connections. After that, time series analysis of related stock and sentiment analysis is blended with deep reinforcement methodology. Turkish version of Bidirectional Encoder Representations from Transformers (BerTurk) is employed to analyze the sentiments of the users while deep Q-learning methodology is used for the deep reinforcement learning side of the proposed model to construct the deep Q network. In order to demonstrate the effectiveness of the proposed model, Garanti Bank (GARAN), Akbank (AKBNK), T\"urkiye \.I\c{s} Bankas{\i} (ISCTR) stocks in Istanbul Stock Exchange are used as a case study. Experiment results show that the proposed novel model achieves remarkable results for stock market prediction task. | ['Zeynep Hilal Kilimci', 'Anil Berk Altuner'] | 2021-07-02 | null | null | null | null | ['stock-market-prediction'] | ['time-series'] | [-8.90677750e-01 -1.56748921e-01 2.14262772e-02 -1.04370125e-01
6.76636351e-03 -5.97709835e-01 4.20615554e-01 -5.07534668e-02
-2.19704390e-01 8.51347327e-01 3.51260722e-01 -4.98343319e-01
-1.83741793e-01 -1.25662720e+00 -6.60911500e-01 -4.75754440e-01
-2.65855044e-01 1.73936486e-01 3.20109516e-01 -8.85694444e-01
6.61893427e-01 2.31278747e-01 -8.61831665e-01 1.64641812e-01
4.52565372e-01 1.12924325e+00 8.80286545e-02 1.57263860e-01
-2.84861356e-01 1.63834524e+00 -5.56068420e-01 -8.44875157e-01
4.98493254e-01 -4.91151363e-01 -6.72035575e-01 -3.98889408e-02
-3.72595310e-01 -5.66256464e-01 -3.25934380e-01 1.11308289e+00
3.92837405e-01 1.27084896e-01 6.46340609e-01 -1.27232456e+00
-7.81321883e-01 9.73986328e-01 -7.50251949e-01 6.12776637e-01
-3.72076511e-01 -7.23164752e-02 1.55011630e+00 -7.67179430e-01
4.48220462e-01 8.55154097e-01 5.04273355e-01 1.52785733e-01
-7.31806159e-01 -1.24523473e+00 1.37432799e-01 5.15069544e-01
-7.72466481e-01 1.16919123e-01 1.07335961e+00 -3.97354186e-01
9.17590797e-01 -4.32957619e-01 1.18050337e+00 5.02165020e-01
3.36337298e-01 9.99323547e-01 1.04230177e+00 -6.50917068e-02
2.81921834e-01 4.80302066e-01 2.92060733e-01 3.27668190e-01
2.10072875e-01 1.15405314e-01 -4.75623995e-01 1.36270091e-01
6.91352904e-01 9.26858261e-02 2.31643632e-01 -1.89661812e-02
-8.38279128e-01 1.41413128e+00 7.87459254e-01 5.16288340e-01
-6.90375090e-01 2.00568244e-01 4.58286673e-01 6.87572598e-01
6.97856486e-01 2.11792275e-01 -5.54633617e-01 -5.44802658e-02
-8.28058302e-01 4.09337848e-01 8.95273983e-01 5.00054359e-01
5.37985861e-01 7.59509921e-01 3.58676553e-01 4.79439259e-01
4.60083574e-01 2.38233536e-01 8.68922055e-01 -3.00223798e-01
3.93754512e-01 8.06551874e-01 9.49290544e-02 -1.27315295e+00
-3.52472395e-01 -7.05342889e-01 -7.08345413e-01 4.47321534e-01
2.14755889e-02 -6.82723463e-01 -3.04963380e-01 1.06023979e+00
-1.00457366e-03 4.54107910e-01 5.09306371e-01 8.12511563e-01
5.33458292e-01 9.49255049e-01 -1.20049588e-01 -2.77124941e-01
9.92427945e-01 -9.76055861e-01 -6.86680138e-01 1.62570357e-01
4.56560254e-01 -4.99951780e-01 2.89697349e-01 3.77525270e-01
-1.07861650e+00 -4.23671901e-01 -1.08660555e+00 5.64813614e-01
-4.71499532e-01 -1.66489240e-02 4.85928148e-01 4.07750934e-01
-1.18247294e+00 6.45978928e-01 -4.67221886e-01 1.37791574e-01
3.37683052e-01 5.46201169e-01 8.92649032e-03 6.18485928e-01
-1.73814678e+00 9.62332308e-01 6.61500514e-01 9.02634114e-02
-5.47616541e-01 -3.85947406e-01 -4.81121361e-01 1.91885769e-01
1.84190348e-02 1.04110688e-02 1.06254017e+00 -1.52726007e+00
-1.48986554e+00 1.88082263e-01 6.74742520e-01 -9.12495553e-01
4.95031208e-01 -5.47730811e-02 -4.67986882e-01 1.82421371e-01
-2.93951365e-03 2.94449627e-01 7.00924158e-01 -6.44183636e-01
-9.10815954e-01 -2.81171143e-01 1.95378810e-01 3.18703592e-01
-6.14162564e-01 1.04469575e-01 4.21216011e-01 -1.05445516e+00
-2.50359029e-01 -6.91566169e-01 2.72110128e-03 -8.45171452e-01
-5.43067753e-02 -3.85760069e-01 8.16971481e-01 -9.59236324e-01
1.17696428e+00 -1.92786980e+00 -1.79763988e-01 5.97721815e-01
2.55719088e-02 2.36470655e-01 3.16682942e-02 6.87584162e-01
-3.23376894e-01 -8.98902714e-02 2.06131533e-01 2.25053489e-01
-1.00743949e-01 -1.98336005e-01 -5.25125980e-01 1.86913103e-01
2.42386892e-01 9.62320745e-01 -6.99302197e-01 -2.13282704e-01
-7.43077472e-02 9.60592926e-02 -3.56528074e-01 8.96220878e-02
-3.24355543e-01 8.73000026e-02 -5.90520322e-01 3.38546574e-01
4.98257607e-01 -3.71252090e-01 9.14185047e-02 -1.31043181e-01
-1.83528930e-01 3.98120806e-02 -1.31570208e+00 7.62996256e-01
-2.72097178e-02 4.95687217e-01 -5.61868370e-01 -1.28856456e+00
1.43930185e+00 5.56370020e-01 5.21777987e-01 -8.70852888e-01
3.62147480e-01 4.38101828e-01 3.00925821e-01 -2.75949180e-01
4.35845643e-01 -4.51788068e-01 3.48121256e-01 7.34390914e-01
-6.37203828e-02 1.62544474e-02 3.35393459e-01 1.15532972e-01
6.29664958e-01 1.55161032e-02 8.86832252e-02 -2.21388429e-01
6.56037569e-01 2.18615364e-02 6.19544148e-01 -7.16893524e-02
-1.85767233e-01 -6.35346621e-02 7.87405550e-01 -5.61461270e-01
-1.06952012e+00 -5.33581853e-01 3.86384696e-01 7.24008262e-01
-1.27936810e-01 6.51656538e-02 -4.28254068e-01 -5.86566806e-01
1.99201182e-01 7.92661369e-01 -6.22576773e-01 -6.20372631e-02
-2.41766632e-01 -7.75692642e-01 1.89860523e-01 5.97360432e-01
9.24476087e-01 -1.48038757e+00 -4.41827804e-01 4.20102715e-01
3.53289515e-01 -6.98883593e-01 -1.54512137e-01 4.37091291e-02
-8.37045670e-01 -1.05552948e+00 -1.02062106e+00 -7.37197101e-01
7.59978592e-02 -2.12354455e-02 8.07761192e-01 -3.17827053e-02
2.52464056e-01 -1.16505533e-01 -4.90894705e-01 -7.31669188e-01
-2.35773385e-01 7.01291189e-02 -8.00306275e-02 4.18825924e-01
4.56461430e-01 -3.62321496e-01 -8.67316365e-01 1.31286785e-01
-7.70505250e-01 -1.01224698e-01 6.04322910e-01 8.99264157e-01
-3.07762567e-02 5.80576539e-01 1.57212055e+00 -9.46547687e-01
8.84592474e-01 -8.15012753e-01 -1.09721565e+00 5.58290295e-02
-8.81356895e-01 -4.03037714e-03 5.18217564e-01 -2.37083342e-03
-1.26555550e+00 -4.06435937e-01 1.55973941e-01 -3.81537259e-01
6.67106330e-01 1.01646590e+00 4.17499602e-01 2.32962176e-01
2.13195220e-01 4.47205573e-01 2.92214096e-01 -1.60598502e-01
-1.74581504e-03 6.15824282e-01 -2.96244442e-01 1.84525654e-01
7.86045432e-01 7.97496587e-02 -6.69560358e-02 -4.06134337e-01
-6.39488280e-01 -2.17387185e-01 -2.98256040e-01 -4.59360301e-01
8.29979420e-01 -9.80601311e-01 -9.70762849e-01 7.31887698e-01
-6.41151190e-01 -1.16926469e-01 1.96419373e-01 6.86736822e-01
-3.57986122e-01 -9.03044492e-02 -9.55013216e-01 -1.09860897e+00
-6.15214348e-01 -9.92686510e-01 1.05264857e-01 3.15366983e-01
1.08444080e-01 -1.29761434e+00 1.75791442e-01 4.18312430e-01
4.29207385e-01 7.78715834e-02 8.12746048e-01 -1.16584253e+00
-6.62915289e-01 -2.39346832e-01 -4.07287151e-01 8.91661823e-01
2.03383788e-01 -1.73502237e-01 -5.46515703e-01 -2.36516088e-01
-4.90481406e-02 -4.13457245e-01 5.50879955e-01 2.83899248e-01
4.48201388e-01 -2.26574644e-01 3.61698240e-01 -1.37675762e-01
1.68464160e+00 6.12705886e-01 5.12806952e-01 7.69963264e-01
4.09166187e-01 5.25178552e-01 6.39960885e-01 6.58408523e-01
4.68390465e-01 3.55630517e-02 3.71191561e-01 2.60189652e-01
4.70999449e-01 -2.33154252e-01 8.30044329e-01 1.14113402e+00
-1.13286830e-01 2.46391356e-01 -6.41312480e-01 4.29157495e-01
-1.93356657e+00 -1.17684495e+00 1.03181772e-01 1.59269834e+00
6.95074320e-01 3.84856761e-01 3.84866267e-01 3.71213377e-01
5.46676397e-01 4.17423025e-02 -4.96105492e-01 -2.05692261e-01
1.31780757e-02 2.58027732e-01 5.89201331e-01 1.25939175e-01
-7.98380792e-01 9.64884281e-01 4.40332651e+00 7.29564905e-01
-1.24298704e+00 -2.86850840e-01 9.22775388e-01 2.86423080e-02
-3.11321855e-01 -1.26721099e-01 -7.30263889e-01 7.56803572e-01
1.04837942e+00 -2.47750595e-01 2.31997177e-01 6.31608963e-01
4.96355623e-01 7.89307356e-02 -5.95053911e-01 6.16733730e-01
-1.95974916e-01 -1.60203004e+00 8.38817134e-02 3.09560560e-02
8.20038736e-01 2.58081555e-02 2.01927662e-01 4.92051452e-01
4.62986797e-01 -7.04227507e-01 8.38194788e-01 6.61213994e-01
-1.94850281e-01 -1.29741383e+00 1.21382558e+00 4.33682472e-01
-1.01481056e+00 -4.91398126e-01 -3.48839760e-01 -1.56778350e-01
-1.71938643e-01 2.62247473e-01 -8.42642546e-01 6.45534754e-01
6.42048061e-01 1.07377183e+00 -2.22748190e-01 6.51813209e-01
-4.94132936e-02 8.69419754e-01 1.73822686e-01 -7.30553150e-01
6.31628752e-01 -7.51610637e-01 7.71662965e-02 6.24154747e-01
3.35325450e-01 2.82549977e-01 -1.98672906e-01 6.07619286e-01
-3.63078088e-01 5.27397215e-01 -6.45872593e-01 -2.54611135e-01
8.24799389e-02 1.12109673e+00 -8.63499403e-01 -2.93833315e-01
-7.18202710e-01 4.26428080e-01 3.26919377e-01 2.30662182e-01
-7.13203311e-01 -3.81075948e-01 2.36027449e-01 1.15634874e-01
4.93584514e-01 5.27174771e-02 -4.73900437e-02 -1.03711760e+00
-2.81095713e-01 -8.32153678e-01 4.10852492e-01 -8.15771937e-01
-1.47154427e+00 5.14641941e-01 -1.72170445e-01 -1.31253362e+00
-3.20309371e-01 -7.48111784e-01 -7.67712653e-01 1.03304720e+00
-1.94723392e+00 -8.95374477e-01 3.77501220e-01 7.86846757e-01
4.95308667e-01 -1.13125348e+00 2.65188098e-01 1.52090535e-01
-6.01344287e-01 5.42125292e-02 3.32449108e-01 7.89674878e-01
1.37552559e-01 -1.38579667e+00 1.59010857e-01 6.36723638e-01
1.39515594e-01 5.54704368e-02 3.45113218e-01 -8.78404081e-01
-9.46045578e-01 -1.02969754e+00 7.67187238e-01 2.13799104e-01
1.35309637e+00 2.86002189e-01 -6.46874666e-01 8.54035079e-01
7.82815099e-01 -2.83614993e-01 6.30647659e-01 -3.96579504e-01
-1.74038354e-02 -4.63629574e-01 -8.51230979e-01 5.51586449e-01
-1.01879135e-01 -1.36926949e-01 -7.31992185e-01 2.25006714e-01
4.57843482e-01 2.13305429e-01 -8.96763623e-01 -1.28410965e-01
2.67862827e-01 -1.15738964e+00 5.70874512e-01 -5.30291975e-01
5.14736533e-01 -1.45716429e-01 5.27402014e-02 -1.53691518e+00
-8.81070867e-02 -3.56746197e-01 1.36810482e-01 9.46686447e-01
4.83078659e-01 -8.41471255e-01 1.09602487e+00 3.38089406e-01
3.26709896e-01 -5.72288573e-01 -6.37495399e-01 -5.18374622e-01
2.59242892e-01 2.45766784e-03 6.77454591e-01 1.10836208e+00
3.09637517e-01 5.79097688e-01 -2.92479306e-01 -1.30927190e-01
4.71372664e-01 1.71593010e-01 3.43545169e-01 -1.01198137e+00
-3.23853880e-01 -5.57020307e-01 -3.06389362e-01 -3.77247602e-01
2.46882334e-01 -7.77973831e-01 -7.22692192e-01 -1.35785758e+00
-1.01768412e-01 -2.69905150e-01 -7.72597909e-01 2.11808756e-01
6.70586079e-02 4.69529033e-02 4.32053596e-01 1.77462220e-01
-8.63678902e-02 7.21746325e-01 1.39775312e+00 -2.32483670e-01
-1.14417709e-01 4.44820553e-01 -4.68067586e-01 3.53850931e-01
1.12905836e+00 7.37139806e-02 -6.63551867e-01 -3.02424133e-02
9.00816500e-01 4.98194396e-01 6.76637366e-02 -6.29033923e-01
2.10610017e-01 6.39874563e-02 4.66880888e-01 -7.72979975e-01
2.20522195e-01 -8.98088872e-01 2.98447944e-02 8.62267196e-01
-3.97716582e-01 8.39734375e-01 9.78839677e-03 4.16982025e-01
-8.29615653e-01 -4.64475155e-01 5.74133515e-01 -4.41488385e-01
-8.10748756e-01 2.68643528e-01 -3.18835676e-01 -1.89042360e-01
1.34550154e+00 -9.89951640e-02 -4.28752378e-02 -8.19254279e-01
-6.29066825e-01 5.58515549e-01 -4.88500148e-01 3.03743571e-01
9.03970182e-01 -1.29083216e+00 -9.71617460e-01 -3.73176672e-02
-2.51018137e-01 -4.56960917e-01 9.55638438e-02 6.15103066e-01
-9.40565944e-01 1.95988774e-01 -5.22102714e-01 2.02656463e-01
-7.10554123e-01 4.00044233e-01 4.49575484e-01 -4.35303956e-01
-3.56530458e-01 6.49477541e-01 -8.67912844e-02 -2.84974761e-02
-8.97951890e-03 -1.41404897e-01 -1.07733083e+00 8.07173133e-01
1.14196241e-01 2.51268864e-01 1.14295557e-01 -4.70254838e-01
1.35629445e-01 2.22896665e-01 -3.00760537e-01 -2.33297095e-01
1.73366308e+00 1.19982183e-01 -2.61603087e-01 4.12907958e-01
1.03060579e+00 -2.62818038e-01 -9.57720101e-01 -3.64297181e-02
3.87880027e-01 1.40615599e-02 2.42122099e-01 -5.85500121e-01
-1.69838500e+00 8.28516722e-01 4.51664060e-01 4.63605165e-01
9.79194164e-01 -5.95753849e-01 9.41968501e-01 4.29484963e-01
1.56498119e-01 -1.44780958e+00 3.36271852e-01 5.30006170e-01
7.29757607e-01 -1.18699229e+00 -2.81857580e-01 2.78625757e-01
-1.02611887e+00 1.46140826e+00 3.95402223e-01 -7.33046472e-01
1.38272226e+00 -1.10840313e-01 3.04709971e-01 -1.76732257e-01
-8.15524936e-01 -2.11735010e-01 -1.48525909e-01 1.52053740e-02
3.47306311e-01 -1.25170112e-01 -2.03524709e-01 6.01898849e-01
-3.62180918e-01 1.34951293e-01 9.90131259e-01 8.16523552e-01
-4.17634845e-01 -8.74723017e-01 -1.43368125e-01 5.64993620e-01
-7.29443192e-01 -2.94861704e-01 3.34835649e-02 9.02967155e-01
-2.90205896e-01 6.52694941e-01 4.03325170e-01 -4.27605212e-01
2.38606364e-01 -2.96819434e-02 -2.10100740e-01 -3.09084028e-01
-1.05031765e+00 2.90952921e-01 -3.50919403e-02 1.73590362e-01
-6.51445091e-01 -6.87062800e-01 -1.29731023e+00 -3.63127500e-01
-3.12154561e-01 5.94445765e-01 5.10524273e-01 8.05378377e-01
3.66921388e-02 5.36529660e-01 1.14793372e+00 -1.68862328e-01
-9.09942925e-01 -1.10597372e+00 -1.18999684e+00 3.22262406e-01
1.68890432e-01 -6.30232990e-01 -8.37969929e-02 2.55753770e-02] | [4.373959541320801, 4.253649711608887] |
4e9c645e-89dd-4b0b-aa9f-e72754d2facb | exploring-global-and-local-information-for | 2306.02025 | null | https://arxiv.org/abs/2306.02025v1 | https://arxiv.org/pdf/2306.02025v1.pdf | Exploring Global and Local Information for Anomaly Detection with Normal Samples | Anomaly detection aims to detect data that do not conform to regular patterns, and such data is also called outliers. The anomalies to be detected are often tiny in proportion, containing crucial information, and are suitable for application scenes like intrusion detection, fraud detection, fault diagnosis, e-commerce platforms, et al. However, in many realistic scenarios, only the samples following normal behavior are observed, while we can hardly obtain any anomaly information. To address such problem, we propose an anomaly detection method GALDetector which is combined of global and local information based on observed normal samples. The proposed method can be divided into a three-stage method. Firstly, the global similar normal scores and the local sparsity scores of unlabeled samples are computed separately. Secondly, potential anomaly samples are separated from the unlabeled samples corresponding to these two scores and corresponding weights are assigned to the selected samples. Finally, a weighted anomaly detector is trained by loads of samples, then the detector is utilized to identify else anomalies. To evaluate the effectiveness of the proposed method, we conducted experiments on three categories of real-world datasets from diverse domains, and experimental results show that our method achieves better performance when compared with other state-of-the-art methods. | ['Xibin Zhao', 'Nan Wang', 'Fan Xu'] | 2023-06-03 | null | null | null | null | ['fraud-detection'] | ['miscellaneous'] | [ 1.22565567e-01 -4.23793554e-01 -2.14871820e-02 -3.02909851e-01
-3.05470884e-01 8.90398324e-02 2.78438956e-01 4.62393463e-01
-1.50470495e-01 3.67178857e-01 -1.46087427e-02 8.67287293e-02
-4.28607035e-03 -7.62478411e-01 -2.18056604e-01 -8.54524314e-01
-1.22247949e-01 1.59339324e-01 4.74133402e-01 1.00329049e-01
4.50290054e-01 3.76507044e-01 -1.43263960e+00 1.27890661e-01
1.09137750e+00 1.39206707e+00 -4.00519878e-01 -8.21749195e-02
-3.65170091e-01 6.83176160e-01 -5.51388860e-01 1.66018978e-01
2.26295322e-01 -5.58813274e-01 -6.07497916e-02 5.79884410e-01
3.48158590e-02 -4.51054245e-01 -1.00975223e-01 1.46594191e+00
2.91264087e-01 3.43173236e-01 4.90941942e-01 -1.20821738e+00
-3.03678304e-01 1.35348529e-01 -1.07996178e+00 6.57141387e-01
2.66817808e-01 1.72400966e-01 8.26121092e-01 -1.17331958e+00
1.45871773e-01 1.28002334e+00 3.73330176e-01 2.55634397e-01
-8.95435333e-01 -7.93411732e-01 6.70164824e-01 3.47328812e-01
-1.28877497e+00 -4.10563231e-01 9.84769523e-01 -2.93276012e-01
3.36153924e-01 1.52130663e-01 5.27201116e-01 8.73772919e-01
1.33719742e-01 9.71068501e-01 6.70254171e-01 -3.33840102e-02
4.20897037e-01 -1.45806625e-01 3.29747617e-01 5.20423353e-01
4.69402015e-01 -2.10600466e-01 -2.56930381e-01 -6.21608436e-01
4.09258664e-01 8.88101518e-01 -3.95128787e-01 -2.93656021e-01
-1.08520794e+00 8.53460193e-01 2.55119741e-01 3.02149296e-01
-6.68805480e-01 -6.48105741e-01 7.32151985e-01 5.22395015e-01
6.27645373e-01 -1.40329346e-01 -3.33288610e-01 9.32339206e-02
-6.12776577e-01 2.50269175e-01 4.48853642e-01 5.81078053e-01
7.15101898e-01 2.77117044e-01 -4.07788642e-02 1.01392090e+00
5.83184063e-01 3.17029536e-01 8.70287776e-01 -2.92617172e-01
6.52028203e-01 1.18423533e+00 -2.08146106e-02 -1.55288744e+00
-2.29287133e-01 -3.98190856e-01 -1.16178012e+00 -2.33722329e-01
2.77770430e-01 -2.25431666e-01 -8.27373683e-01 1.25569701e+00
5.57376802e-01 6.78817451e-01 3.99298184e-02 9.57027793e-01
5.18483162e-01 7.81954885e-01 -1.56406909e-01 -5.79989076e-01
1.04007733e+00 -6.50684237e-01 -7.77845144e-01 -3.31890792e-01
5.30761302e-01 -6.64206624e-01 8.45570028e-01 5.41512430e-01
-6.01227820e-01 -2.56071508e-01 -9.47624624e-01 7.32991159e-01
-3.40097654e-03 -1.71443015e-01 2.45622858e-01 1.43625543e-01
-1.41591087e-01 4.82422441e-01 -1.00522959e+00 -4.11950856e-01
4.98776078e-01 -1.27735436e-01 -1.80767193e-01 -3.50917757e-01
-9.47578073e-01 2.41421580e-01 6.47473454e-01 2.73213416e-01
-7.25472867e-01 -2.24078447e-01 -1.01195645e+00 3.18445936e-02
5.98521352e-01 2.45839935e-02 1.00139928e+00 -9.75432098e-01
-8.73643816e-01 4.07311887e-01 -3.96064073e-01 -2.83312410e-01
4.04013902e-01 -1.90606788e-01 -1.11792600e+00 -7.33668655e-02
2.22394094e-01 -4.24020320e-01 9.27779377e-01 -1.03697824e+00
-9.93405461e-01 -5.85733235e-01 -6.00726008e-01 1.02699585e-01
-4.59934920e-01 1.59488901e-01 -3.44014287e-01 -6.60030246e-01
8.17997992e-01 -4.24702913e-01 -4.61843461e-01 -1.65587366e-01
-5.89559853e-01 -1.36637300e-01 1.52410269e+00 -5.71998239e-01
1.35552609e+00 -2.58402586e+00 -3.89550000e-01 8.57518375e-01
2.74812490e-01 2.69114673e-01 3.35569829e-02 2.25494951e-01
-1.50770634e-01 -3.18855315e-01 -6.63419008e-01 9.16749462e-02
-2.69755661e-01 1.64433420e-01 -4.02966797e-01 6.73346579e-01
2.62346417e-01 1.82544485e-01 -9.58154500e-01 -3.29187870e-01
1.96164072e-01 -2.11386159e-01 -4.03877199e-01 4.64860439e-01
-6.18039630e-02 6.32752955e-01 -9.64251518e-01 1.06679726e+00
8.36400032e-01 -4.14879508e-02 -1.70371085e-01 2.15485334e-01
2.57443815e-01 -1.36660680e-01 -1.67076194e+00 1.06514537e+00
1.68938309e-01 2.27308515e-02 1.52632818e-02 -1.55480886e+00
1.25011837e+00 2.05984488e-01 6.19385779e-01 -7.77098835e-01
9.47503746e-02 5.00035167e-01 1.90579519e-01 -6.54996932e-01
6.61177859e-02 1.62202884e-02 6.34268066e-03 3.99935305e-01
-3.16918224e-01 6.73450351e-01 2.97291785e-01 -7.90874287e-03
1.21007645e+00 -4.66769516e-01 4.21872079e-01 -7.48061389e-02
9.81559753e-01 -3.52427304e-01 1.27438676e+00 4.75275099e-01
-4.48530614e-01 4.74936515e-01 6.99987054e-01 -6.55799866e-01
-5.95267236e-01 -9.75609779e-01 -5.93935363e-02 6.25281513e-01
3.64955783e-01 -2.38905892e-01 -3.74957234e-01 -1.03969419e+00
7.99315572e-02 5.12012362e-01 -2.75682688e-01 -5.41656971e-01
-4.74161059e-01 -1.09464180e+00 1.63556799e-01 2.89871573e-01
7.16549933e-01 -1.42542779e+00 -3.35663766e-01 3.34904850e-01
2.54227873e-02 -9.71750855e-01 -4.44124818e-01 -2.15603873e-01
-1.14737713e+00 -1.33591318e+00 -5.08410275e-01 -7.44244158e-01
1.00134957e+00 3.89395416e-01 6.93797171e-01 3.19133461e-01
-1.35009542e-01 -1.10506475e-01 -5.18310130e-01 -4.20979410e-01
1.28087569e-02 -4.69774842e-01 3.43230039e-01 8.09079170e-01
8.34848344e-01 -5.94126701e-01 -6.33384943e-01 4.33748603e-01
-1.10961246e+00 -7.71927714e-01 7.17879295e-01 1.00758588e+00
8.57845902e-01 3.45314264e-01 8.34358990e-01 -1.11986232e+00
7.85301328e-01 -1.00429094e+00 -4.34162199e-01 -1.20149791e-01
-6.80101633e-01 -2.87045181e-01 9.16854382e-01 -4.20931906e-01
-9.63287532e-01 -2.31843829e-01 -3.23275924e-02 -5.11803627e-01
-3.48852932e-01 8.01123202e-01 -4.92878139e-01 2.55440801e-01
4.22579557e-01 4.57657635e-01 6.47976249e-02 -6.14214361e-01
-2.61388063e-01 7.42397904e-01 4.35367584e-01 -4.67505217e-01
7.57940710e-01 3.01235318e-01 -1.60854191e-01 -8.68652284e-01
-7.75731444e-01 -7.98954010e-01 -1.31294161e-01 5.19890115e-02
3.46340120e-01 -7.22231507e-01 -8.10673982e-02 7.77250469e-01
-7.06833005e-01 4.30883050e-01 -3.62412572e-01 5.53242922e-01
4.92029302e-02 8.16395223e-01 -4.69673932e-01 -1.04941130e+00
-2.89917022e-01 -9.12407935e-01 7.40283728e-01 5.31645775e-01
-2.15373784e-02 -7.89871454e-01 -5.52683957e-02 -1.10427797e-01
2.95009553e-01 5.63149750e-01 7.98358440e-01 -1.39995587e+00
-2.80824363e-01 -7.44094133e-01 -1.67832077e-01 5.40599763e-01
4.84799057e-01 2.03149989e-02 -5.54839253e-01 -4.52589750e-01
3.44088823e-01 -7.85051286e-02 7.40674734e-01 1.54067390e-02
1.59548020e+00 -4.92123991e-01 -3.54378164e-01 4.39229250e-01
1.16441834e+00 4.94838744e-01 4.59752947e-01 3.66455466e-01
7.77330995e-01 3.11956584e-01 9.72424626e-01 8.94542754e-01
2.45172624e-02 2.10880503e-01 5.80375552e-01 1.59069896e-02
8.41049433e-01 -1.45761058e-01 4.30806518e-01 1.10172760e+00
1.73247948e-01 8.87268186e-02 -8.04001868e-01 6.69101536e-01
-2.17030764e+00 -1.04933918e+00 -2.62919009e-01 2.36055160e+00
3.60201657e-01 4.79783028e-01 1.15674697e-01 4.72432792e-01
9.84590709e-01 3.21640193e-01 -9.61073816e-01 -2.82874573e-02
-8.04948807e-03 -3.12238671e-02 -2.96919253e-02 -2.22561896e-01
-1.23171139e+00 4.38628137e-01 5.28463745e+00 8.20989549e-01
-1.01464844e+00 -1.30789459e-01 6.01764441e-01 -1.48716187e-02
3.26529285e-03 7.29667619e-02 -3.06260347e-01 9.02392089e-01
6.47846580e-01 -1.76571742e-01 8.06827843e-02 1.11628211e+00
4.26657796e-01 1.16922207e-01 -8.18509340e-01 1.05335760e+00
8.61211568e-02 -5.30370891e-01 2.94138137e-02 -8.82126316e-02
7.27652013e-01 -6.24257475e-02 -1.50900871e-01 3.98952544e-01
-4.61288281e-02 -5.94288409e-01 1.69947237e-01 5.79459250e-01
1.67420715e-01 -1.06719100e+00 1.13378346e+00 6.29241168e-01
-1.24391377e+00 -2.81313121e-01 -5.58070660e-01 -9.60903242e-02
-6.85085133e-02 1.29230714e+00 -3.02866131e-01 6.61766231e-01
7.50831246e-01 1.14248860e+00 -4.52273637e-01 1.32679653e+00
5.72942244e-03 7.36327529e-01 -3.72113407e-01 1.61795542e-01
2.20681921e-01 -5.84916890e-01 6.80307925e-01 7.82649934e-01
6.24681830e-01 9.30101052e-02 6.08100474e-01 5.49582541e-01
-2.13548094e-02 6.26486957e-01 -6.12835407e-01 1.29354283e-01
4.32382733e-01 1.15228748e+00 -6.46362662e-01 -4.13655639e-01
-7.75133610e-01 8.24926674e-01 2.97737755e-02 3.49447697e-01
-5.09276092e-01 -6.82852447e-01 5.44843256e-01 8.60482920e-04
1.21097893e-01 2.46190816e-01 -4.75827828e-02 -1.65051973e+00
5.90728939e-01 -1.17625928e+00 8.50741029e-01 -1.60297692e-01
-1.79902101e+00 2.61014253e-01 -3.80188793e-01 -1.83316386e+00
-5.64552136e-02 -3.24900955e-01 -1.40466392e+00 6.55369759e-01
-1.24486458e+00 -4.37277526e-01 -6.38832450e-01 8.65374684e-01
4.67286497e-01 -4.95616525e-01 6.43865645e-01 4.65802789e-01
-1.18040073e+00 5.47054589e-01 4.64537084e-01 3.63536656e-01
6.37861907e-01 -9.87703145e-01 3.21149915e-01 1.36969233e+00
-1.34123564e-01 6.09183252e-01 3.82769167e-01 -7.45366752e-01
-8.92452300e-01 -1.38185418e+00 3.62596750e-01 2.02595502e-01
5.98871708e-01 1.45540193e-01 -1.48899090e+00 6.44965231e-01
-3.79653186e-01 7.42955148e-01 6.60272837e-01 -1.17853791e-01
-1.09063946e-01 -2.92686880e-01 -1.34856129e+00 4.95634079e-01
6.83124363e-01 -1.16524175e-02 -6.92126036e-01 2.61079520e-01
2.54753143e-01 -4.13082927e-01 -5.97775578e-01 5.89635611e-01
1.38054460e-01 -1.05973649e+00 5.51768959e-01 -6.04609549e-01
2.72521734e-01 -6.08004570e-01 -9.83109847e-02 -1.36913383e+00
-1.55301958e-01 -1.08741634e-01 -5.16373813e-01 1.40392590e+00
6.09609112e-02 -1.10035193e+00 7.63542533e-01 2.26580665e-01
-2.08902672e-01 -8.88364971e-01 -8.32452059e-01 -6.55510426e-01
-5.56392372e-01 -3.08250993e-01 7.95782566e-01 1.20882130e+00
-1.91735979e-02 1.21467464e-01 -3.25570911e-01 2.69487888e-01
7.64912188e-01 2.05809727e-01 6.54955268e-01 -1.54069722e+00
-6.16081245e-02 -3.61705273e-01 -8.43569756e-01 -8.52043688e-01
-8.60052109e-02 -6.13332033e-01 4.08285111e-02 -1.17537260e+00
1.02562092e-01 -3.21052670e-01 -7.98646033e-01 2.83575654e-01
-6.56463802e-01 -5.13957217e-02 -4.54017013e-01 5.39533257e-01
-7.60383964e-01 8.13346982e-01 7.74478555e-01 -1.55449793e-01
-3.03182751e-01 3.23102504e-01 -6.93012178e-01 1.24792504e+00
7.47658789e-01 -4.54226762e-01 -3.85732621e-01 -5.90532320e-03
-3.99060935e-01 -9.92639214e-02 -6.10808237e-03 -1.24728942e+00
8.32009241e-02 -4.22172308e-01 5.92808843e-01 -5.55866778e-01
-2.71488339e-01 -1.02113044e+00 -4.14315879e-01 6.01143420e-01
-2.49754381e-03 1.37227118e-01 -2.35213861e-01 8.93773913e-01
-6.87853158e-01 -1.62896886e-01 8.08931410e-01 -9.89107601e-03
-8.84030938e-01 7.86282122e-01 -2.07106143e-01 3.36179078e-01
1.22821128e+00 -4.55529056e-02 -8.21681246e-02 -1.89483911e-01
-5.35610139e-01 5.83995521e-01 2.99858660e-01 3.78979474e-01
9.79087830e-01 -1.76921070e+00 -7.15817034e-01 6.24087214e-01
7.24595129e-01 4.69511509e-01 2.68976003e-01 9.09171283e-01
-4.87706929e-01 -1.52877495e-01 -5.02635688e-02 -7.17804074e-01
-9.44174886e-01 6.53556585e-01 1.53099969e-01 -2.29955524e-01
-9.16241705e-01 3.41147542e-01 7.90689141e-02 -3.72907430e-01
1.85923725e-01 -1.87103987e-01 -4.29180384e-01 -3.90999913e-02
8.00573766e-01 3.37395102e-01 1.52110517e-01 -6.46077871e-01
-4.99833047e-01 4.28172439e-01 -4.03844029e-01 3.61372143e-01
1.13933671e+00 6.23149574e-02 -3.05514991e-01 6.00227475e-01
9.97593343e-01 4.48875986e-02 -9.95255649e-01 -7.30149925e-01
2.20681325e-01 -7.28399158e-01 -3.40571761e-01 -9.82499719e-02
-1.48253071e+00 8.14300060e-01 6.87368929e-01 5.03644824e-01
1.40591252e+00 -4.48452234e-01 1.08457267e+00 5.07853270e-01
1.58875048e-01 -1.08297706e+00 1.58675209e-01 4.76480901e-01
3.85273039e-01 -1.34229171e+00 -1.17948778e-01 -1.79551870e-01
-7.12946594e-01 9.79809165e-01 9.82499063e-01 -5.93327105e-01
6.93738639e-01 -1.74389258e-01 5.53586408e-02 -1.60503507e-01
-4.59192395e-01 3.32515463e-02 1.66536689e-01 3.58246297e-01
3.72566469e-02 -1.63266271e-01 -4.02350187e-01 8.55029523e-01
3.54434967e-01 -3.29205275e-01 4.11085516e-01 1.11377275e+00
-7.26905286e-01 -7.43997753e-01 -6.45184398e-01 1.08923340e+00
-7.82468081e-01 2.38476336e-01 -1.28297329e-01 3.55191976e-01
1.61707744e-01 1.02040803e+00 1.53112158e-01 -4.42800432e-01
6.60692751e-01 1.97204784e-01 -4.63586658e-01 -6.26985788e-01
-1.81058783e-03 8.85837898e-02 -3.16896528e-01 -6.79771900e-01
-6.20727763e-02 -7.87173510e-01 -1.37552881e+00 -1.49811447e-01
-3.28923464e-01 3.96375060e-01 -1.61467609e-03 1.15774322e+00
1.44838661e-01 6.17850721e-01 1.10789657e+00 -5.46171963e-01
-6.31251574e-01 -9.74423707e-01 -8.95485461e-01 8.88280749e-01
4.27461594e-01 -6.14069879e-01 -7.26215422e-01 -2.16249153e-01] | [7.55881404876709, 2.3992648124694824] |
a4c3cd18-71f8-4acb-926e-88b57f021eb2 | adaptive-wavelet-transformer-network-for-3d | null | null | https://openreview.net/forum?id=5MLb3cLCJY | https://openreview.net/pdf?id=5MLb3cLCJY | Adaptive Wavelet Transformer Network for 3D Shape Representation Learning | We present a novel method for 3D shape representation learning using multi-scale wavelet decomposition. Distinct from previous works that either decompose 3D shapes into complimentary components at a single scale, or naively adopt up-/down-sampling to build hierarchies and treat all points or local regions equally, we decompose 3D shapes into sub-bands components at multiple scales and all scales form a decomposition tree in a principled manner rooted in multi-resolution wavelet analysis. Specifically, we propose Adaptive Wavelet Transformer Network (AWT-Net) that firstly generates approximation or detail wavelet coefficients per point, classifying each point into high or low sub-bands components, using lifting scheme at multiple scales recursively and hierarchically. Then, AWT-Net exploits Transformers that regard the features from different but complementary components as two holistic representations, and fuse them with the original shape features with different attentions. The wavelet coefficients can be learned without direct supervision on coefficients, and AWT-Net is fully differentiable and can be learned in an end-to-end fashion. Extensive experiments demonstrate that AWT-Net achieves state-of-the-arts or competitive performance on 3D shape classification and segmentation benchmarks. | ['Yi Fang', 'Hao Huang'] | 2021-09-29 | null | null | null | iclr-2022-4 | ['3d-shape-retrieval', '3d-shape-representation'] | ['computer-vision', 'computer-vision'] | [ 8.32062773e-03 -2.93196682e-02 -1.67784676e-01 -2.53131449e-01
-9.88172412e-01 -5.38452327e-01 1.26969799e-01 5.05874604e-02
1.81473881e-01 1.60129532e-01 3.10569882e-01 -1.31382033e-01
-7.53085539e-02 -1.19489133e+00 -5.61700881e-01 -7.44453728e-01
-2.98475593e-01 6.03269696e-01 4.48817968e-01 -1.73592091e-01
-1.12383783e-01 9.25422072e-01 -1.12472522e+00 6.36607230e-01
4.55528170e-01 1.52440751e+00 -1.29858077e-01 6.19921327e-01
-3.84503514e-01 2.54458517e-01 -2.15488225e-01 -2.24597424e-01
6.07899010e-01 1.53125480e-01 -7.79396117e-01 4.87807781e-01
5.49128830e-01 -2.88108945e-01 -2.63104856e-01 9.06901240e-01
6.05910480e-01 -1.96445182e-01 5.88711262e-01 -1.04909575e+00
-1.23290372e+00 1.56093091e-01 -1.11170149e+00 1.72392726e-01
1.84785202e-01 2.56632362e-03 1.12391531e+00 -1.14936709e+00
3.37974966e-01 1.78651559e+00 1.07683432e+00 2.44035810e-01
-1.84237266e+00 -4.89171267e-01 4.38898176e-01 -2.97331423e-01
-1.17883670e+00 -1.01417832e-01 1.31206489e+00 -2.76303291e-01
9.81191158e-01 9.61152429e-04 1.01338887e+00 7.78204560e-01
4.73710150e-02 9.08220589e-01 1.02437615e+00 -7.06383511e-02
8.40557143e-02 -6.58391476e-01 1.06230907e-01 1.07346106e+00
1.01830818e-01 -2.14389578e-01 -3.31336737e-01 -5.21472454e-01
1.25717139e+00 3.21412802e-01 -8.67561772e-02 -7.42948651e-01
-1.08610833e+00 7.56960452e-01 7.27397263e-01 4.14087057e-01
-6.38432920e-01 3.29192840e-02 4.75483179e-01 6.11954510e-01
7.89510906e-01 -3.50358158e-01 -8.33562791e-01 3.84361207e-01
-6.77013576e-01 2.05030426e-01 4.26364362e-01 9.31416333e-01
9.11612153e-01 2.80119181e-01 -1.01480164e-01 1.04778254e+00
1.65203348e-01 4.46240813e-01 4.70482171e-01 -7.42315352e-01
3.57469529e-01 8.70217025e-01 -3.08340132e-01 -1.14708686e+00
-3.15306664e-01 -4.21468198e-01 -1.24826181e+00 2.86270648e-01
-3.77922393e-02 2.50211447e-01 -1.20069647e+00 1.55262959e+00
5.65176308e-01 5.09800315e-01 -4.59547415e-02 7.78051436e-01
8.89374316e-01 5.86890399e-01 -2.00218201e-01 -9.06401649e-02
1.56268573e+00 -7.64253676e-01 -1.47336945e-01 2.72421122e-01
8.21373053e-03 -6.36924863e-01 1.03220880e+00 2.02426970e-01
-1.44576299e+00 -9.74675834e-01 -8.11913729e-01 -3.91115963e-01
-3.09821635e-01 9.02633071e-02 5.32771587e-01 3.06159198e-01
-1.03070784e+00 5.75656295e-01 -1.13087118e+00 7.02904612e-02
1.00196183e+00 2.36667961e-01 -2.80554891e-01 -2.03354862e-02
-6.21283889e-01 4.30725276e-01 2.56360210e-02 -2.82514207e-02
-8.67447793e-01 -1.10849321e+00 -9.54809904e-01 1.21962823e-01
-1.08220503e-01 -9.34040427e-01 9.37321484e-01 -5.32874525e-01
-1.37140942e+00 8.54229987e-01 -3.14798594e-01 -2.56921798e-01
1.25274748e-01 1.30163759e-01 -3.66987586e-02 3.85589778e-01
3.04597229e-01 6.40169561e-01 1.28866434e+00 -1.50314307e+00
-7.75181055e-01 -7.60002971e-01 -1.14414869e-02 7.05176033e-04
-3.46538901e-01 -3.48146647e-01 -3.11070293e-01 -1.02070284e+00
8.56583476e-01 -4.47488815e-01 -2.80276775e-01 3.45663279e-01
-9.70704388e-03 -5.88604987e-01 1.27181709e+00 -5.37809670e-01
7.29083896e-01 -2.23258495e+00 4.43565518e-01 3.10438216e-01
5.86933494e-01 -7.59809688e-02 -4.42115247e-01 1.39259174e-01
3.75095457e-02 9.90742594e-02 -5.64370215e-01 -4.49704617e-01
-1.97290611e-02 5.17729342e-01 -6.39008999e-01 2.72906929e-01
4.12195891e-01 9.25647259e-01 -8.28784645e-01 -3.50459188e-01
1.61479250e-01 8.79433632e-01 -6.24332666e-01 -1.54796783e-02
1.37581304e-01 3.08249652e-01 -6.73938513e-01 1.15992033e+00
1.06880689e+00 -1.46083102e-01 -2.72638023e-01 -6.82148516e-01
1.02848493e-01 1.83590241e-02 -1.09854066e+00 1.85551953e+00
-6.26845777e-01 2.16380805e-02 3.59432638e-01 -1.48798013e+00
1.26345086e+00 4.27748740e-01 7.67099082e-01 -2.90170819e-01
-2.05288097e-01 3.98607403e-01 -4.10568357e-01 -1.75970629e-01
-2.07394093e-01 -4.47281003e-01 5.28578982e-02 1.13054305e-01
3.87273461e-01 -5.14755130e-01 -1.50344923e-01 -4.05163050e-01
9.20270622e-01 2.22429767e-01 2.01925933e-01 -2.28416488e-01
6.96169674e-01 -2.13811398e-01 7.90239990e-01 3.29701990e-01
-8.14704597e-02 6.47147119e-01 3.41356218e-01 -1.11739004e+00
-9.93138552e-01 -1.32770979e+00 -1.84772313e-01 1.18771911e+00
-6.47760108e-02 -1.61825031e-01 -2.52669871e-01 -9.15972888e-01
6.64843619e-01 1.29466712e-01 -8.06317270e-01 1.08039796e-01
-7.71211684e-01 -4.72314447e-01 4.81422722e-01 8.08894753e-01
6.58202291e-01 -8.19139123e-01 -7.93290854e-01 4.46284890e-01
2.54395902e-01 -1.05542362e+00 -6.56326950e-01 3.65550488e-01
-1.45281208e+00 -9.36827481e-01 -9.82080281e-01 -1.28370118e+00
8.38884771e-01 7.08107293e-01 1.03813231e+00 1.31048337e-01
-3.96777272e-01 5.67495227e-01 -2.90316403e-01 -2.08362997e-01
9.67237055e-02 -3.69362421e-02 -1.32249087e-01 3.00119340e-01
1.15831502e-01 -1.43975937e+00 -5.47428310e-01 8.11049491e-02
-6.40398443e-01 -1.75337538e-01 6.77967370e-01 9.78146195e-01
1.08956718e+00 3.13496172e-01 5.32106340e-01 -4.94450837e-01
4.97452170e-01 -8.51920620e-02 -3.84204388e-01 7.06082806e-02
-9.48300883e-02 -1.04217835e-01 8.15712214e-01 -4.87141639e-01
-8.18953574e-01 5.52886911e-02 -3.70395958e-01 -1.15163445e+00
-3.86619240e-01 1.83083802e-01 8.37562233e-02 -2.25347087e-01
3.22942853e-01 6.70768857e-01 -8.15540478e-02 -8.78270805e-01
4.49050128e-01 2.38242537e-01 4.45499957e-01 -1.05924284e+00
1.16949141e+00 8.72429788e-01 1.92013517e-01 -9.37614560e-01
-1.01640773e+00 -2.87879825e-01 -1.08136427e+00 1.94184586e-01
7.12934971e-01 -9.14482951e-01 -6.51893795e-01 2.69294351e-01
-1.08453786e+00 7.29681402e-02 -6.60354495e-01 -1.06759734e-01
-6.37980700e-01 3.61533284e-01 -8.47387254e-01 -5.91975510e-01
-6.29243731e-01 -1.00444639e+00 1.68999863e+00 -4.79941778e-02
2.10277498e-01 -9.03953016e-01 -1.62545636e-01 1.52451426e-01
3.56232345e-01 6.75580382e-01 1.49702430e+00 -3.50455254e-01
-2.19819784e-01 -2.03492627e-01 -2.81616241e-01 3.09583306e-01
5.41556120e-01 -3.61647725e-01 -7.71132529e-01 -7.04900801e-01
1.66205928e-01 -5.15870631e-01 9.51449871e-01 5.08189797e-01
1.48544669e+00 -4.11497384e-01 -1.62215829e-01 7.75915205e-01
1.39509118e+00 -1.32564247e-01 -5.13172001e-02 -2.52895892e-01
7.35153973e-01 3.60880852e-01 -1.72574386e-01 3.45061213e-01
4.87732232e-01 3.62954915e-01 3.31723392e-01 -3.39388549e-01
-3.69518816e-01 -3.83340329e-01 2.02305522e-02 9.98250246e-01
-4.57498074e-01 8.14733744e-01 -6.24774218e-01 6.26389980e-01
-1.66263771e+00 -7.74424136e-01 2.95360804e-01 1.80657279e+00
7.63051391e-01 2.71817565e-01 5.00283301e-01 2.06014901e-01
4.99963522e-01 4.33530331e-01 -6.43536687e-01 -1.15705170e-01
-1.29860371e-01 5.46227813e-01 1.94035128e-01 4.21572208e-01
-1.01800275e+00 8.41781497e-01 5.93627453e+00 9.60460842e-01
-1.20318198e+00 1.45785198e-01 6.94933891e-01 1.30342215e-01
-4.27102417e-01 -7.62503296e-02 -3.14850837e-01 4.41357307e-02
-3.07722539e-02 5.53526450e-03 3.92475903e-01 9.90227044e-01
-1.27095655e-01 8.22184026e-01 -9.66687739e-01 8.63223314e-01
-4.26328897e-01 -1.35685873e+00 4.84083265e-01 2.38060467e-02
5.77401698e-01 -1.03982143e-01 6.72080442e-02 4.71154690e-01
2.63765395e-01 -7.65831649e-01 8.06126416e-01 1.60570771e-01
6.81896448e-01 -8.72722447e-01 4.21749413e-01 4.20664817e-01
-2.25332046e+00 -2.21409932e-01 -8.67328823e-01 5.59822172e-02
-9.56275389e-02 6.46052003e-01 -2.33363718e-01 8.01452756e-01
8.87413323e-01 8.96657348e-01 -1.82450190e-01 6.22327924e-01
5.12687936e-02 5.19325554e-01 -3.99164915e-01 4.06812847e-01
3.62882674e-01 -2.30401203e-01 6.30867004e-01 1.24027431e+00
4.59487915e-01 6.18694901e-01 8.08925331e-01 8.20762098e-01
-1.35398090e-01 -1.24622278e-01 -5.86232722e-01 2.72912055e-01
3.86921793e-01 1.41218245e+00 -8.93828809e-01 -5.28083324e-01
-5.54595053e-01 8.34550858e-01 4.01687741e-01 4.15675402e-01
-4.51193780e-01 -2.44387224e-01 7.14536071e-01 1.00000635e-01
9.91374910e-01 -3.16637844e-01 -4.47597653e-01 -1.12326717e+00
3.16629224e-02 -6.55730247e-01 6.30724728e-01 -4.50977027e-01
-1.85928404e+00 5.96837938e-01 -9.81591865e-02 -1.29001188e+00
5.42919397e-01 -7.57242143e-01 -7.10224628e-01 7.28263378e-01
-1.93246305e+00 -1.80446982e+00 -1.13213815e-01 7.92349875e-01
9.35980201e-01 -2.19850481e-01 9.59324658e-01 -2.19238400e-02
-1.93851724e-01 3.88508439e-01 -3.78137529e-01 3.03336322e-01
-1.21802753e-02 -1.47979391e+00 5.84835827e-01 2.77345747e-01
2.51608461e-01 4.21148092e-01 5.84043656e-03 -4.73479956e-01
-1.57601440e+00 -1.10723925e+00 5.05673051e-01 9.89817781e-05
4.60766286e-01 -2.86388546e-01 -1.08932233e+00 5.14232337e-01
1.03779763e-01 6.78004026e-01 5.95191836e-01 5.22558093e-02
-7.69327879e-01 -3.37949693e-01 -1.46858048e+00 2.44510144e-01
1.35183668e+00 -4.19078797e-01 -1.06154311e+00 1.09630860e-01
9.22340453e-01 -4.69031006e-01 -1.23192322e+00 6.60286903e-01
4.93610680e-01 -8.95739496e-01 1.53802693e+00 -8.03644001e-01
3.67567122e-01 -3.16138387e-01 -4.73664165e-01 -1.25879467e+00
-9.37092781e-01 -3.95894408e-01 -3.51394594e-01 8.14195514e-01
4.29675132e-02 -6.28018022e-01 8.35028231e-01 -6.97252899e-02
-2.80959547e-01 -1.36840487e+00 -1.41097713e+00 -7.15201676e-01
4.25866187e-01 -3.48409325e-01 1.00555861e+00 8.62436771e-01
-4.50150251e-01 3.70599180e-01 4.07150164e-02 4.31495845e-01
1.06170857e+00 9.39022779e-01 5.33170819e-01 -1.55463386e+00
-1.59459889e-01 -9.00884748e-01 -4.36368346e-01 -1.43225932e+00
1.44464925e-01 -1.22035408e+00 -3.43120158e-01 -1.55542207e+00
2.51465216e-02 -2.50969023e-01 -4.19591248e-01 9.17369604e-01
9.36950818e-02 4.98519123e-01 2.30352283e-01 3.07718366e-01
-5.19314520e-02 8.76032710e-01 1.62879729e+00 -4.11982328e-01
-1.56154126e-01 2.35732980e-02 -6.14176810e-01 1.03386116e+00
3.11345994e-01 -2.92974830e-01 -3.65423501e-01 -5.31442881e-01
-3.33522975e-01 1.77981690e-01 4.38168943e-01 -7.50643432e-01
7.09132478e-02 -2.55251415e-02 9.12072659e-01 -9.08114791e-01
2.77145416e-01 -1.20420516e+00 -2.98024863e-01 4.99090105e-01
-1.58078566e-01 6.95780888e-02 3.36423606e-01 5.87163329e-01
-2.90740460e-01 1.50201172e-01 9.46457088e-01 -3.06366712e-01
-2.23627016e-01 7.27038205e-01 9.03204679e-02 -9.20902789e-02
7.90222049e-01 -5.37489295e-01 2.94100910e-01 4.90677096e-02
-1.01331353e+00 2.02805907e-01 6.31749257e-02 3.51726502e-01
1.00766945e+00 -1.90494394e+00 -8.47607851e-01 6.11283243e-01
-1.97228879e-01 5.58186889e-01 2.57019699e-01 5.84956169e-01
-1.63520396e-01 -1.06119306e-03 -2.86809236e-01 -1.01594388e+00
-9.97403085e-01 4.92278725e-01 2.71159321e-01 -4.07856584e-01
-1.15170276e+00 8.33171725e-01 2.44397685e-01 -6.68683529e-01
2.46677965e-01 -7.11173296e-01 3.95574095e-03 2.17234239e-01
2.40712091e-01 2.36971885e-01 -1.01023793e-01 -6.14234984e-01
-3.19685966e-01 1.48398459e+00 1.44837230e-01 2.04868659e-01
1.97271597e+00 2.50362128e-01 -3.13714892e-01 3.56406897e-01
1.49465573e+00 2.89875604e-02 -1.38040113e+00 -5.49099207e-01
-2.87945718e-01 -5.81651449e-01 5.78771867e-02 -3.49001616e-01
-1.51252508e+00 8.73606324e-01 3.49396646e-01 6.75624907e-01
1.52478468e+00 1.23028651e-01 1.21382260e+00 3.49234790e-01
4.14513737e-01 -6.51300192e-01 3.62811595e-01 5.04403889e-01
1.30370200e+00 -8.39960933e-01 -1.03986137e-01 -4.76361483e-01
-3.20883155e-01 1.53876364e+00 3.40339333e-01 -7.53923416e-01
1.07096374e+00 3.32828552e-01 -1.87170908e-01 -2.53587782e-01
-6.78372920e-01 -1.89265624e-01 5.48566818e-01 6.90319359e-01
1.86696097e-01 1.77666899e-02 2.70947218e-01 7.67165363e-01
-1.01366736e-01 -1.10194467e-01 -3.14884037e-01 8.32751751e-01
-6.11548603e-01 -8.97215068e-01 -4.49197114e-01 3.76999378e-01
-5.31248972e-02 2.79959023e-01 -1.49894550e-01 5.60135126e-01
4.44070905e-01 3.51859659e-01 3.14841382e-02 -2.90153384e-01
7.87950337e-01 2.96437949e-01 6.21825159e-01 -4.78647619e-01
-4.37694222e-01 5.69908142e-01 -4.77504164e-01 -4.01672661e-01
-4.55420405e-01 -6.65927529e-01 -1.34912550e+00 -8.70968029e-02
1.91607043e-01 -5.76089174e-02 -2.38296725e-02 6.52481437e-01
4.89616483e-01 6.19182169e-01 9.92776692e-01 -1.42188585e+00
-8.52204800e-01 -8.32038045e-01 -4.03279305e-01 4.39629138e-01
7.85973907e-01 -6.77464247e-01 -2.92297035e-01 1.51869118e-01] | [8.048065185546875, -3.627959728240967] |
14ca2ea8-20d4-4c9a-8756-e01e609b64f0 | visual-localization-using-imperfect-3d-models | 2304.05947 | null | https://arxiv.org/abs/2304.05947v1 | https://arxiv.org/pdf/2304.05947v1.pdf | Visual Localization using Imperfect 3D Models from the Internet | Visual localization is a core component in many applications, including augmented reality (AR). Localization algorithms compute the camera pose of a query image w.r.t. a scene representation, which is typically built from images. This often requires capturing and storing large amounts of data, followed by running Structure-from-Motion (SfM) algorithms. An interesting, and underexplored, source of data for building scene representations are 3D models that are readily available on the Internet, e.g., hand-drawn CAD models, 3D models generated from building footprints, or from aerial images. These models allow to perform visual localization right away without the time-consuming scene capturing and model building steps. Yet, it also comes with challenges as the available 3D models are often imperfect reflections of reality. E.g., the models might only have generic or no textures at all, might only provide a simple approximation of the scene geometry, or might be stretched. This paper studies how the imperfections of these models affect localization accuracy. We create a new benchmark for this task and provide a detailed experimental evaluation based on multiple 3D models per scene. We show that 3D models from the Internet show promise as an easy-to-obtain scene representation. At the same time, there is significant room for improvement for visual localization pipelines. To foster research on this interesting and challenging task, we release our benchmark at v-pnk.github.io/cadloc. | ['Torsten Sattler', 'Zuzana Kukelova', 'Vojtech Panek'] | 2023-04-12 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Panek_Visual_Localization_Using_Imperfect_3D_Models_From_the_Internet_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Panek_Visual_Localization_Using_Imperfect_3D_Models_From_the_Internet_CVPR_2023_paper.pdf | cvpr-2023-1 | ['visual-localization'] | ['computer-vision'] | [ 5.10381535e-02 -4.69418287e-01 2.09403206e-02 -3.22556645e-01
-1.04914308e+00 -1.03433144e+00 4.21870291e-01 6.52138563e-03
-1.35616541e-01 4.02549297e-01 8.98105726e-02 -3.39207202e-01
3.06630373e-01 -6.45441294e-01 -1.04364860e+00 -2.32175305e-01
8.89010578e-02 5.32998919e-01 2.58524746e-01 -4.54229154e-02
1.98687926e-01 9.35855806e-01 -1.50500607e+00 2.01144479e-02
3.53203744e-01 1.04040992e+00 6.45345926e-01 8.28170061e-01
-1.12763107e-01 4.35072392e-01 -3.86838734e-01 -8.95608887e-02
3.97262633e-01 -1.76330265e-02 -4.89623249e-01 3.76572341e-01
6.76462054e-01 -4.87193972e-01 -2.81125546e-01 1.06512976e+00
3.54303360e-01 -1.55323213e-02 1.77623361e-01 -1.13104522e+00
-4.43483531e-01 -1.51469797e-01 -5.99154353e-01 -1.23030022e-01
9.46060121e-01 6.63491040e-02 6.61807299e-01 -1.14094794e+00
7.30626464e-01 1.00406420e+00 8.41274202e-01 4.54625897e-02
-1.16429281e+00 -3.89624864e-01 1.03700526e-01 -2.00877905e-01
-1.94642699e+00 -7.49148011e-01 7.86139190e-01 -3.80211800e-01
5.89676738e-01 6.05787098e-01 6.32749140e-01 1.01792264e+00
-1.00776084e-01 4.46031839e-01 9.49114919e-01 -5.56638598e-01
3.44495624e-01 2.48380750e-01 -9.32909027e-02 8.36332798e-01
3.53898704e-01 -4.54727262e-02 -4.71065998e-01 -3.75644594e-01
1.08131766e+00 2.90221363e-01 -6.39018953e-01 -9.07961607e-01
-1.37930155e+00 4.30975109e-01 7.75403678e-01 -2.93287672e-02
-2.97537416e-01 2.67358154e-01 -1.65918753e-01 5.63717708e-02
3.34949315e-01 2.50639290e-01 -3.46940368e-01 -1.10787556e-01
-1.02171493e+00 9.00456086e-02 6.26020968e-01 1.46150827e+00
1.13856566e+00 -2.40156084e-01 4.29124445e-01 5.48654377e-01
4.11394149e-01 7.95741856e-01 1.22850396e-01 -1.08552361e+00
4.53892916e-01 4.92435277e-01 4.25210357e-01 -1.32825446e+00
-1.68751404e-01 -9.56664756e-02 -6.49265528e-01 1.13995053e-01
4.62426841e-01 2.62802750e-01 -9.25901711e-01 1.33084583e+00
5.04072011e-01 3.85304242e-01 -2.30935290e-01 1.12093484e+00
7.00357497e-01 6.33290291e-01 -3.66305172e-01 2.28359848e-01
1.32916009e+00 -8.67908776e-01 -3.33348930e-01 -6.00653648e-01
7.65977204e-01 -1.06206155e+00 1.06091130e+00 2.24439934e-01
-9.11239088e-01 -5.00800848e-01 -9.24452662e-01 -2.10164040e-01
-4.06666696e-01 3.60246181e-01 5.99034607e-01 5.72818875e-01
-1.26411271e+00 2.98162222e-01 -8.28049183e-01 -7.43229926e-01
1.67166606e-01 1.31318912e-01 -7.94038296e-01 -7.42874503e-01
-5.17118752e-01 7.75915444e-01 3.50869708e-02 2.63365030e-01
-7.96079695e-01 -4.59157050e-01 -1.05585194e+00 -2.47105211e-01
4.35975730e-01 -7.41818368e-01 1.19220412e+00 -7.17313588e-01
-1.04713035e+00 8.71208310e-01 -4.07492220e-01 -1.24672130e-01
5.74782073e-01 -1.61821201e-01 -1.77598462e-01 1.21989839e-01
2.32522577e-01 4.65226084e-01 6.62255406e-01 -1.57623243e+00
-2.54023731e-01 -4.99248713e-01 3.01687360e-01 2.19617471e-01
1.85873523e-01 -8.91626030e-02 -9.32239234e-01 -3.04435104e-01
6.95391655e-01 -1.18290353e+00 -4.60998237e-01 4.88118827e-01
-4.54212189e-01 6.91795349e-01 7.69541800e-01 -8.06096792e-01
8.07752252e-01 -2.30803370e+00 -4.13524449e-01 3.52580309e-01
-4.78089461e-03 -1.06429636e-01 -1.98400170e-01 4.47317183e-01
1.49679473e-02 2.76716918e-01 5.78563996e-02 -5.60467720e-01
-1.54326782e-01 5.91527671e-02 -4.60776299e-01 6.48735583e-01
-1.86881468e-01 7.70198703e-01 -7.60541975e-01 -4.63109940e-01
5.97708225e-01 7.12827325e-01 -4.51052636e-01 9.86240804e-02
-1.93727404e-01 5.63857138e-01 -4.19190526e-01 9.46132123e-01
8.51388097e-01 -5.11372030e-01 2.43655145e-02 -3.17767888e-01
-1.39061615e-01 3.10738891e-01 -1.57057190e+00 2.11286187e+00
-5.74010074e-01 6.40872061e-01 3.63818966e-02 -5.07637501e-01
8.26516569e-01 7.12450892e-02 2.95524061e-01 -4.72584575e-01
-6.87713251e-02 2.40932882e-01 -5.60790598e-01 6.12751022e-02
8.64063978e-01 3.02443534e-01 -6.55451566e-02 2.56482184e-01
-3.55832279e-01 -5.47777534e-01 -3.07896703e-01 1.73197672e-01
1.14748430e+00 2.39700034e-01 4.12163049e-01 1.02111511e-01
1.27128720e-01 3.12618226e-01 2.99364716e-01 8.05815458e-01
1.24093905e-01 1.19153404e+00 -7.68949389e-02 -4.98444021e-01
-1.11903274e+00 -9.62965965e-01 -4.38446589e-02 4.30092901e-01
4.35564727e-01 -6.64780259e-01 -4.44381118e-01 -3.47242475e-01
-1.57208562e-01 4.15677994e-01 -2.58233249e-01 9.73824039e-02
-2.52584934e-01 -1.96442991e-01 2.10839480e-01 5.00987649e-01
3.83543432e-01 -5.10693133e-01 -8.46257210e-01 2.67578196e-02
-2.67447799e-01 -1.37508464e+00 -5.72541833e-01 -5.78868389e-02
-8.45746100e-01 -9.52838957e-01 -6.35945261e-01 -5.09223461e-01
9.21339869e-01 1.05449033e+00 1.29048276e+00 3.59428823e-01
-2.33972192e-01 7.85141647e-01 -3.17519188e-01 -2.89118707e-01
2.65456103e-02 -2.05912054e-01 3.73964272e-02 -1.23123154e-02
1.05843343e-01 -4.26678032e-01 -5.35471082e-01 5.24486244e-01
-6.93284035e-01 2.77761430e-01 4.03923929e-01 3.62473130e-01
1.07657313e+00 -1.05879240e-01 -3.45667094e-01 -5.03130615e-01
2.31925994e-02 -3.82852435e-01 -9.92957056e-01 3.17247450e-01
1.02787493e-02 -2.36118987e-01 3.49230379e-01 -3.93716007e-01
-6.68655217e-01 5.57403028e-01 1.00049838e-01 -9.25740302e-01
-4.48330283e-01 5.16105294e-01 -3.25256318e-01 -3.50577414e-01
6.09540105e-01 1.04696937e-01 -2.54011452e-01 -6.36400163e-01
4.05597717e-01 6.18873000e-01 4.97976273e-01 -5.97879410e-01
9.54389095e-01 7.36980796e-01 -8.43939036e-02 -1.10638154e+00
-8.49022269e-01 -7.41150618e-01 -7.47007549e-01 -3.58787812e-02
4.57328916e-01 -1.23773110e+00 -2.05352843e-01 1.41644001e-01
-1.40527523e+00 -3.81547421e-01 -2.21366957e-01 5.65488398e-01
-4.80419487e-01 3.62688184e-01 -1.70965791e-01 -8.46853733e-01
3.30339596e-02 -1.25670600e+00 1.54821539e+00 6.70515522e-02
-1.36514097e-01 -6.58819079e-01 -1.38590168e-02 3.47153991e-01
2.54985124e-01 3.34442854e-01 2.98280239e-01 -1.36311889e-01
-1.32684302e+00 -5.91715574e-01 -1.98376596e-01 -3.31053361e-02
1.13665074e-01 -1.02856560e-02 -1.10348785e+00 -2.38029659e-01
-2.43616000e-01 -7.94912949e-02 4.34623957e-01 2.49379769e-01
9.38040614e-01 -7.36712813e-02 -5.36865771e-01 8.53447318e-01
1.57179558e+00 1.32419104e-02 6.12985313e-01 3.23203653e-01
8.17051351e-01 2.80940056e-01 7.72852063e-01 4.21948463e-01
5.91964722e-01 9.93929148e-01 5.23957133e-01 -2.50045568e-01
-1.20878197e-01 -5.87800801e-01 7.99200982e-02 7.84854472e-01
1.04694188e-01 -8.54060426e-02 -1.23490429e+00 3.99761707e-01
-1.63374162e+00 -7.30479360e-01 -3.19039017e-01 2.61931849e+00
2.89145648e-01 -1.84833840e-01 -3.12905967e-01 -1.57609284e-01
5.08522034e-01 8.76240134e-02 -4.01150435e-01 1.62647545e-01
-1.09348387e-01 1.18729901e-02 7.06279516e-01 5.08985579e-01
-9.76518929e-01 9.95282233e-01 5.75677252e+00 4.29644614e-01
-1.16050422e+00 7.95497298e-02 3.10830116e-01 4.18706276e-02
-1.50791675e-01 4.74971592e-01 -7.54801512e-01 3.99890482e-01
8.16984236e-01 7.88055062e-02 4.70620304e-01 1.21453691e+00
2.12623775e-01 -4.62095857e-01 -1.08415937e+00 1.50346947e+00
1.08049639e-01 -1.34642231e+00 -1.70758039e-01 4.13093984e-01
5.83288074e-01 3.41832399e-01 -1.95274889e-01 -1.16212890e-01
2.36655399e-01 -8.22964430e-01 1.01502788e+00 4.97537404e-01
1.01898742e+00 -4.68795925e-01 6.29950285e-01 5.09476900e-01
-1.47831690e+00 2.95150906e-01 -6.34873927e-01 1.11096120e-02
2.97924191e-01 6.43649518e-01 -8.11962366e-01 5.09348750e-01
8.53114128e-01 5.52452445e-01 -8.71464849e-01 1.43875873e+00
-1.68452486e-02 2.22048283e-01 -7.64850199e-01 2.02062979e-01
2.04255302e-02 -2.37420931e-01 2.75328189e-01 7.88651526e-01
8.00590634e-01 1.04025595e-01 5.11679888e-01 6.87038183e-01
-6.28798380e-02 -7.13289774e-04 -1.22745788e+00 1.53118253e-01
6.30298734e-01 1.20868111e+00 -8.64396930e-01 -1.28350705e-01
-4.36457843e-01 1.11153460e+00 2.37937286e-01 3.67920786e-01
-8.09695899e-01 7.05443993e-02 5.90771854e-01 4.13262099e-01
1.95357632e-02 -7.68097699e-01 -3.71076018e-02 -1.40574467e+00
1.64124265e-01 -7.66352654e-01 -5.94976321e-02 -1.43365359e+00
-9.10132408e-01 5.78593433e-01 -3.87736298e-02 -1.50780463e+00
-9.15090591e-02 -4.15101618e-01 -2.00392231e-01 8.70196879e-01
-1.33658922e+00 -1.28377628e+00 -8.05231512e-01 6.51138127e-01
4.33911949e-01 3.35339725e-01 9.11715090e-01 3.08464199e-01
-1.83440909e-01 1.38653427e-01 1.11121520e-01 5.29688634e-02
6.36096120e-01 -9.28009570e-01 7.68260539e-01 8.88229191e-01
9.07364726e-01 6.77675664e-01 5.08553803e-01 -4.98381138e-01
-1.89882159e+00 -1.11206520e+00 6.42147303e-01 -1.04507256e+00
2.93496162e-01 -6.48038387e-01 -8.17623675e-01 9.98748600e-01
-3.10802042e-01 3.64290148e-01 4.51719403e-01 -3.28012630e-02
-2.94503242e-01 4.78253402e-02 -1.02665472e+00 7.48274803e-01
1.23109734e+00 -6.80608273e-01 -1.65099308e-01 5.08028448e-01
5.87066233e-01 -8.36172521e-01 -5.32557011e-01 3.97426076e-02
4.75468934e-01 -9.97403681e-01 1.30960417e+00 -4.61628698e-02
-7.37051889e-02 -7.93907523e-01 -7.58679390e-01 -1.22326422e+00
-2.07401719e-02 -3.83907527e-01 -3.30622867e-02 1.16382647e+00
1.37029231e-01 -4.53334868e-01 6.96938038e-01 9.52788293e-01
8.45177658e-03 -2.51326323e-01 -7.71033406e-01 -8.35389435e-01
-5.94064593e-01 -8.19862902e-01 7.59995282e-01 1.00905573e+00
-6.08489096e-01 2.27252543e-01 -3.47574800e-01 7.45153606e-01
5.44744551e-01 2.37454593e-01 1.36449277e+00 -1.16906559e+00
-3.31406109e-02 2.00089291e-01 -6.43368602e-01 -1.34900677e+00
-3.26909684e-02 -6.18891299e-01 4.42921221e-02 -1.69809008e+00
-1.91519096e-01 -8.03541481e-01 2.28675663e-01 3.87718529e-01
3.57972652e-01 4.24349695e-01 2.86439598e-01 3.68166119e-01
-6.41879797e-01 2.59636104e-01 8.09753954e-01 6.84570801e-03
-1.79125085e-01 6.57717639e-04 -4.36257541e-01 9.30699229e-01
7.88492322e-01 -3.65662545e-01 -3.88686627e-01 -7.42542028e-01
5.23068383e-02 1.65819556e-01 8.28639328e-01 -1.15708339e+00
3.90448719e-01 -1.14960767e-01 5.97868383e-01 -9.07600284e-01
9.30047512e-01 -1.30063748e+00 7.53583074e-01 -6.06642440e-02
3.17657411e-01 4.20043141e-01 1.93622023e-01 5.99601328e-01
-6.12340383e-02 -1.27249107e-01 4.56310600e-01 -5.49236298e-01
-8.49635065e-01 3.07855815e-01 1.80014759e-01 -2.25015581e-01
8.42732310e-01 -6.18334770e-01 -1.66340560e-01 -5.25430858e-01
-4.39071089e-01 -1.06923886e-01 1.23985636e+00 4.31338787e-01
7.48789847e-01 -1.38284004e+00 -1.96815357e-01 3.73544037e-01
3.21452796e-01 3.72067124e-01 1.32346615e-01 6.82103932e-01
-9.33187366e-01 4.30924565e-01 2.14861795e-01 -8.24812233e-01
-1.29695451e+00 7.02371240e-01 2.34313160e-01 1.07395619e-01
-7.70938456e-01 4.92440552e-01 2.94305593e-01 -5.86650372e-01
1.04807004e-01 -4.88647878e-01 5.46973169e-01 -3.36568683e-01
5.37725985e-01 1.12830900e-01 1.93799585e-01 -9.78184700e-01
-5.82044423e-01 9.37545419e-01 3.97707194e-01 -2.09917679e-01
1.09660578e+00 -3.07351559e-01 5.98125644e-02 4.69615430e-01
1.08887446e+00 3.37233394e-01 -1.12618506e+00 -3.35606337e-01
4.05903310e-02 -1.03896523e+00 1.54603541e-01 -5.22456884e-01
-9.18502450e-01 9.34828162e-01 6.19656861e-01 -2.29632016e-03
8.67273569e-01 8.72662887e-02 4.99636769e-01 4.81079608e-01
1.25094879e+00 -6.85089707e-01 -1.32517189e-01 2.45859370e-01
9.92553592e-01 -1.24825287e+00 1.02936208e-01 -4.35955375e-01
-3.34004641e-01 8.96309555e-01 3.63627076e-01 1.67312756e-01
6.52312398e-01 2.96775967e-01 2.03656167e-01 -1.60333231e-01
-2.96038777e-01 -2.32621804e-01 1.43909663e-01 7.18312562e-01
1.71487406e-01 3.26786339e-02 5.91753125e-01 1.63888872e-01
-2.46020094e-01 -1.59567371e-01 5.17310560e-01 1.00134170e+00
-2.48287439e-01 -8.24340999e-01 -8.47053349e-01 1.90720677e-01
-9.68763139e-03 -1.47595093e-01 -4.04992938e-01 7.69689798e-01
-1.90388188e-01 9.10765588e-01 -1.38182104e-01 -1.99405149e-01
5.84106565e-01 -1.71800733e-01 3.90626371e-01 -7.46047795e-01
-2.42586732e-02 8.78341869e-02 -1.70761660e-01 -8.84287477e-01
-1.77534208e-01 -6.03595734e-01 -8.90111387e-01 -4.25862610e-01
-5.42441964e-01 -7.18927085e-02 1.10446846e+00 3.95444304e-01
6.93840265e-01 -5.50385844e-03 4.29020673e-01 -1.40749335e+00
-1.82390794e-01 -5.77636540e-01 -5.03934264e-01 1.31552100e-01
5.29422760e-01 -6.32502437e-01 -2.56897509e-01 1.21528476e-01] | [7.6259446144104, -2.265016555786133] |
daea76ad-099f-4aec-991a-345971ffc290 | modeling-semantic-composition-with-syntactic | 2205.06530 | null | https://arxiv.org/abs/2205.06530v1 | https://arxiv.org/pdf/2205.06530v1.pdf | Modeling Semantic Composition with Syntactic Hypergraph for Video Question Answering | A key challenge in video question answering is how to realize the cross-modal semantic alignment between textual concepts and corresponding visual objects. Existing methods mostly seek to align the word representations with the video regions. However, word representations are often not able to convey a complete description of textual concepts, which are in general described by the compositions of certain words. To address this issue, we propose to first build a syntactic dependency tree for each question with an off-the-shelf tool and use it to guide the extraction of meaningful word compositions. Based on the extracted compositions, a hypergraph is further built by viewing the words as nodes and the compositions as hyperedges. Hypergraph convolutional networks (HCN) are then employed to learn the initial representations of word compositions. Afterwards, an optimal transport based method is proposed to perform cross-modal semantic alignment for the textual and visual semantic space. To reflect the cross-modal influences, the cross-modal information is incorporated into the initial representations, leading to a model named cross-modality-aware syntactic HCN. Experimental results on three benchmarks show that our method outperforms all strong baselines. Further analyses demonstrate the effectiveness of each component, and show that our model is good at modeling different levels of semantic compositions and filtering out irrelevant information. | ['Fuwei Zhang', 'Zijing Ou', 'Qinliang Su', 'Wanjun Zhong', 'Zenan Xu'] | 2022-05-13 | null | null | null | null | ['video-question-answering'] | ['computer-vision'] | [ 1.64325088e-01 -6.83122650e-02 -2.63978839e-01 -4.12624300e-01
-7.45904446e-01 -5.69131374e-01 6.36022329e-01 7.25119635e-02
-1.46379754e-01 1.82117522e-01 6.64228618e-01 -3.45369466e-02
3.67866084e-02 -6.26733661e-01 -8.09404552e-01 -5.21645963e-01
3.71543080e-01 3.09713602e-01 3.71143878e-01 -2.84716040e-01
2.18235344e-01 1.44356648e-02 -1.49086452e+00 7.19112396e-01
7.59163797e-01 1.00319958e+00 4.12863642e-01 2.89213747e-01
-8.70839357e-01 7.53371298e-01 -3.80151480e-01 -4.75404918e-01
1.52258752e-02 -7.15283334e-01 -9.70583916e-01 5.21899641e-01
5.11066616e-01 -1.54382199e-01 -4.31852877e-01 1.14085901e+00
1.21143326e-01 2.85688937e-01 5.91736436e-01 -1.25872099e+00
-6.53263211e-01 7.81094372e-01 -6.14388227e-01 -8.80567506e-02
4.96445477e-01 6.36672042e-03 1.41264248e+00 -9.72041726e-01
7.47366726e-01 1.39961076e+00 2.48320252e-01 4.39892173e-01
-9.77331281e-01 -5.03944874e-01 7.45662212e-01 6.25264227e-01
-1.40571165e+00 -2.99941361e-01 1.08495736e+00 -3.69809151e-01
6.99880123e-01 2.51210541e-01 7.70291328e-01 1.21407771e+00
-1.88098341e-01 8.88120294e-01 7.46493340e-01 -2.79654115e-01
-4.24031988e-02 1.68863550e-01 2.85992116e-01 8.26780379e-01
-1.25452533e-01 -5.03953815e-01 -6.19277537e-01 1.11581273e-02
4.14041966e-01 -6.27868176e-02 -4.78733242e-01 -5.83727062e-01
-1.30207229e+00 7.52634108e-01 5.98092318e-01 5.45192599e-01
-3.73468250e-01 9.80579853e-02 5.07845759e-01 2.57440135e-02
3.88087630e-01 3.04224283e-01 -1.89158574e-01 3.11849385e-01
-7.43530691e-01 2.27132544e-01 5.87577939e-01 1.19052672e+00
8.58601987e-01 -3.22558224e-01 -4.70885873e-01 9.12413836e-01
4.41636533e-01 1.81261525e-01 3.22820425e-01 -8.49815071e-01
7.44466126e-01 9.51267123e-01 -1.64717808e-01 -1.29846001e+00
-2.44704515e-01 -2.61482000e-01 -5.69997489e-01 -3.72743309e-01
3.16516906e-01 3.55814785e-01 -1.05437446e+00 1.70942307e+00
3.61411959e-01 2.40398660e-01 8.29886459e-03 1.10944557e+00
1.06158733e+00 8.94595802e-01 4.52197790e-01 4.38619219e-02
1.81849945e+00 -1.17696846e+00 -8.73965025e-01 -4.73441243e-01
5.68337262e-01 -6.30463600e-01 1.04453170e+00 -2.55383015e-01
-9.19473112e-01 -5.20899594e-01 -9.25205112e-01 -3.75464737e-01
-4.67327237e-01 -8.74697641e-02 1.65617779e-01 7.09327832e-02
-8.13823640e-01 1.71778828e-01 -5.35308361e-01 -4.46826339e-01
5.09824395e-01 -1.19977407e-01 -2.46605664e-01 -3.78902346e-01
-1.41572964e+00 7.65003026e-01 6.83854878e-01 1.57846272e-01
-8.48659635e-01 -7.63890386e-01 -1.12060928e+00 1.34371728e-01
7.64143407e-01 -9.49469447e-01 7.87387550e-01 -1.37789857e+00
-9.54432487e-01 8.21923435e-01 -4.25116718e-01 -1.09886400e-01
1.91033110e-01 -3.54258306e-02 -4.46755618e-01 6.03045046e-01
2.08108187e-01 8.09721231e-01 8.61726880e-01 -1.67239892e+00
-7.52435744e-01 -2.81483084e-01 4.21583682e-01 5.77124476e-01
-4.72407550e-01 1.03379205e-01 -1.25283611e+00 -6.30722702e-01
2.69012570e-01 -7.91276455e-01 3.38381785e-03 -4.82910313e-02
-5.82477629e-01 -2.30848461e-01 8.18644524e-01 -9.92578089e-01
1.29681575e+00 -2.08169222e+00 5.54466128e-01 3.22166562e-01
3.43909830e-01 -8.32068995e-02 -3.72718185e-01 3.74215931e-01
-1.27880991e-01 -4.68097180e-02 -3.01746249e-01 -3.62065136e-01
-4.68653813e-03 2.82795757e-01 -2.91198939e-01 2.57475078e-01
2.22031236e-01 1.15777433e+00 -9.98833597e-01 -7.83080339e-01
1.87979296e-01 5.05092263e-01 -3.94633710e-01 2.93370813e-01
-6.29758596e-01 3.31925243e-01 -7.32861698e-01 5.53857446e-01
4.10247147e-01 -5.49955487e-01 4.34440404e-01 -7.59666324e-01
2.81816870e-01 2.57675767e-01 -8.91725779e-01 1.96117580e+00
-4.02639300e-01 5.53656816e-01 -2.27867201e-01 -1.22488487e+00
7.74010658e-01 2.27866888e-01 5.92364371e-01 -7.65366614e-01
1.67819008e-01 -8.13284367e-02 -3.42477590e-01 -9.53148305e-01
5.03156841e-01 1.84502136e-02 -5.65329976e-02 2.42007822e-01
1.08956836e-01 1.50000468e-01 2.44513348e-01 5.24892688e-01
5.11908710e-01 4.21393812e-01 6.53883964e-02 -1.22743361e-01
7.78737843e-01 1.07776910e-01 3.11596215e-01 2.77112842e-01
1.06698558e-01 7.60106683e-01 6.66189909e-01 -3.29897821e-01
-8.92818630e-01 -8.41999888e-01 3.89382571e-01 1.13475335e+00
6.60757124e-01 -8.10011744e-01 -9.04574037e-01 -9.19812858e-01
-2.95183361e-01 7.69980252e-01 -7.71972060e-01 -2.29963928e-01
-6.21098578e-01 -4.10777748e-01 1.37809753e-01 5.50876141e-01
4.82240915e-01 -9.96283412e-01 -3.34175080e-01 4.79145423e-02
-9.80823934e-01 -1.56758070e+00 -7.71029353e-01 -3.11426967e-01
-5.49042761e-01 -1.15328300e+00 -6.64236724e-01 -1.00029254e+00
9.04920399e-01 5.65862536e-01 1.18827546e+00 3.98939550e-01
2.01329719e-02 5.31978846e-01 -5.75773358e-01 1.30344599e-01
-2.00144514e-01 3.84505875e-02 -5.40810764e-01 5.43381453e-01
4.38605398e-01 -2.31367096e-01 -7.25079477e-01 2.63014078e-01
-1.08724308e+00 3.30664665e-01 3.73574257e-01 5.76201081e-01
7.04322755e-01 -1.51662141e-01 2.13465199e-01 -8.04632306e-01
4.00566608e-01 -5.78240931e-01 -3.10973018e-01 6.46211505e-01
-1.69755399e-01 2.57534981e-01 6.00219667e-01 -5.21127760e-01
-1.15496814e+00 1.90026741e-02 2.72637885e-02 -6.62680209e-01
-6.11742437e-02 7.05746233e-01 -5.16875267e-01 4.54483271e-01
2.35029817e-01 2.58554548e-01 -2.13827536e-01 -3.64031911e-01
7.38338232e-01 3.60633820e-01 4.38901007e-01 -6.34190857e-01
7.95098126e-01 5.49913883e-01 -1.86700583e-01 -8.18820059e-01
-1.24426520e+00 -6.56402230e-01 -6.03142142e-01 -3.82436246e-01
1.39852893e+00 -8.03390324e-01 -1.93942115e-01 1.93965346e-01
-1.23843861e+00 5.40176108e-02 -1.29994288e-01 1.90682843e-01
-4.76904601e-01 5.53594947e-01 -2.91099131e-01 -4.65933055e-01
-7.43066072e-02 -1.27433896e+00 1.20634747e+00 1.44367024e-01
-1.52467832e-01 -1.11823177e+00 -2.75136381e-01 7.56181419e-01
1.45021966e-02 5.50664738e-02 1.30034566e+00 -6.99457288e-01
-7.31320024e-01 -2.59151347e-02 -4.66332227e-01 7.92851597e-02
9.57824960e-02 -8.47828165e-02 -8.50573003e-01 -5.34814261e-02
-1.57382652e-01 -1.52418792e-01 9.44034159e-01 2.94788152e-01
1.41480398e+00 -2.44196594e-01 -5.13337016e-01 5.91149867e-01
1.39544046e+00 -1.04269341e-01 7.22728908e-01 3.42670321e-01
1.18389118e+00 1.08519566e+00 5.43739617e-01 4.35233414e-02
8.16288233e-01 7.94790387e-01 6.18622780e-01 -6.05900139e-02
-4.01242316e-01 -5.63236356e-01 2.83992827e-01 9.71617997e-01
1.38412803e-01 -3.33848894e-01 -7.96676457e-01 7.20034480e-01
-1.92913842e+00 -9.55599666e-01 -2.83529729e-01 1.78942990e+00
6.69057906e-01 -1.18635066e-01 -3.16473804e-02 -2.74375021e-01
7.98990548e-01 4.27678406e-01 -3.03705662e-01 1.55972540e-01
-1.46490052e-01 -2.72512615e-01 1.34334043e-01 4.33731914e-01
-9.84608710e-01 1.11917377e+00 5.20875216e+00 8.38422179e-01
-8.02320898e-01 6.74955249e-02 4.93295014e-01 1.22161329e-01
-7.21050680e-01 2.47770920e-01 -6.04439020e-01 3.96476001e-01
5.71464241e-01 7.55820656e-03 2.99628943e-01 3.69391650e-01
2.32511889e-02 7.27151185e-02 -1.03463519e+00 9.81548131e-01
5.47193110e-01 -1.31949461e+00 5.84668875e-01 -2.80416518e-01
6.51394963e-01 -3.72428924e-01 -2.34084398e-01 2.13989601e-01
2.42309272e-02 -8.40361118e-01 8.47207010e-01 6.77271843e-01
5.16034901e-01 -6.80513322e-01 5.92522264e-01 2.97210086e-02
-1.60641444e+00 4.69642356e-02 -2.92852670e-01 5.49608290e-01
3.73565674e-01 1.61660284e-01 -3.07737768e-01 9.67478335e-01
7.74500728e-01 8.37966979e-01 -6.69910491e-01 7.64817417e-01
-3.58208984e-01 2.65914321e-01 5.27937748e-02 4.32557054e-02
4.46168512e-01 -3.50389123e-01 5.05255699e-01 1.23859954e+00
2.43163317e-01 9.13533121e-02 1.85069442e-01 1.02394354e+00
-2.45215237e-01 3.57312977e-01 -3.73769522e-01 -5.08487318e-03
1.65904894e-01 1.34174013e+00 -8.38309526e-01 -5.68560660e-01
-7.52836227e-01 1.04712749e+00 4.31210667e-01 7.21672416e-01
-1.00737083e+00 -3.65525261e-02 6.88131392e-01 6.71393648e-02
3.84074599e-01 -1.42707512e-01 -4.53789495e-02 -1.34913659e+00
1.83134526e-01 -9.60833609e-01 6.53968394e-01 -1.05785847e+00
-1.30859733e+00 7.01921403e-01 2.59416044e-01 -1.30296445e+00
4.69986275e-02 -4.64430004e-01 -3.92663330e-01 7.43305683e-01
-1.62464213e+00 -1.36445594e+00 -5.64968109e-01 7.78904200e-01
8.56895924e-01 3.71531770e-02 4.63758647e-01 2.60584384e-01
-4.72920388e-01 3.71352851e-01 -4.32558894e-01 1.89548373e-01
4.90943730e-01 -9.79857028e-01 1.74509987e-01 9.30517077e-01
4.12071735e-01 5.35376906e-01 6.61509335e-01 -6.71254754e-01
-1.37072909e+00 -1.21476626e+00 9.02845085e-01 -3.81668568e-01
8.52452993e-01 -4.05800909e-01 -1.32981229e+00 6.44837797e-01
4.22790498e-01 -5.00404239e-02 5.13875902e-01 -1.84136301e-01
-7.55415738e-01 -4.65628989e-02 -4.30416077e-01 9.53058481e-01
1.17922091e+00 -7.71258414e-01 -8.26084137e-01 3.83573353e-01
1.02985954e+00 -2.98536271e-01 -6.15920961e-01 2.43099198e-01
3.97172987e-01 -6.05670333e-01 1.09488344e+00 -8.51782680e-01
6.91706419e-01 -4.00245041e-01 -2.61134148e-01 -1.20571220e+00
-1.82624683e-01 -2.54194677e-01 -3.78871821e-02 1.46591651e+00
4.94774014e-01 -1.13553949e-01 4.88765568e-01 4.55857575e-01
-1.36470541e-01 -5.56974351e-01 -6.00766301e-01 -4.21301395e-01
-2.20045865e-01 -6.00181937e-01 5.83465099e-01 1.10594046e+00
4.02371353e-03 6.45969033e-01 -3.82703841e-01 2.65560597e-01
5.15520751e-01 3.99279684e-01 4.70283419e-01 -8.16271365e-01
4.21676859e-02 -5.64895928e-01 -2.47602031e-01 -1.26144087e+00
6.12999558e-01 -1.09365821e+00 1.14726000e-01 -1.92384005e+00
4.64016289e-01 -6.38736412e-02 -4.03777122e-01 3.33798140e-01
-5.40478289e-01 1.84464127e-01 2.92417556e-01 1.10440098e-01
-8.96375358e-01 6.58831298e-01 1.44429004e+00 -4.14163709e-01
4.23881300e-02 -4.25708622e-01 -7.27604926e-01 7.10932910e-01
6.68860316e-01 -3.15045565e-01 -6.02466762e-01 -7.40532994e-01
2.90490746e-01 -8.64175856e-02 4.35072005e-01 -5.82941651e-01
2.66624510e-01 -3.04098129e-01 1.39629975e-01 -6.12215459e-01
4.09364104e-01 -1.23601866e+00 5.45688495e-02 1.33065442e-02
-5.74516416e-01 -7.47044608e-02 -3.40968408e-02 5.98502874e-01
-4.08936381e-01 -2.34453797e-01 5.15568912e-01 -8.41921493e-02
-9.64466035e-01 4.65274215e-01 -4.64998521e-02 2.94845968e-01
1.08850884e+00 -2.84623921e-01 -3.81795377e-01 -4.88336891e-01
-6.51095033e-01 5.95936835e-01 3.60287488e-01 8.82878244e-01
8.74998331e-01 -1.54931748e+00 -6.32691145e-01 -6.13405183e-03
5.71536064e-01 -1.36633635e-01 5.19993722e-01 7.51543105e-01
-2.38947228e-01 3.32905412e-01 -6.36963248e-02 -6.19765103e-01
-1.32021368e+00 8.92685413e-01 4.74418789e-01 -9.09455493e-02
-8.21622610e-01 6.76317632e-01 7.57861197e-01 -8.65292251e-02
3.09094429e-01 -2.33459443e-01 -6.46268308e-01 4.25356090e-01
4.67594117e-01 -3.73685397e-02 -1.60246998e-01 -1.10081422e+00
-3.32249552e-01 9.43098009e-01 8.59892070e-02 -6.40731230e-02
1.16626990e+00 -5.22136033e-01 -2.53035009e-01 4.40916747e-01
1.42396641e+00 -2.99812168e-01 -1.24343276e+00 -5.22628307e-01
1.47769317e-01 -3.78539681e-01 -1.42788455e-01 -3.16558033e-01
-1.54644990e+00 8.49925578e-01 1.36700824e-01 1.26902774e-01
1.25115490e+00 4.86542463e-01 8.05924714e-01 -6.06588759e-02
-1.27534628e-01 -9.45202768e-01 2.91148841e-01 3.56411219e-01
1.01041150e+00 -9.29237247e-01 -1.27168402e-01 -8.09944332e-01
-9.24955249e-01 1.16795683e+00 7.47893095e-01 2.06542313e-01
3.75688046e-01 -2.20280856e-01 8.69106501e-02 -5.15752316e-01
-6.67777121e-01 -4.67316449e-01 8.73329341e-01 4.02572811e-01
1.80017516e-01 -1.97109535e-01 -1.97511777e-01 4.49629426e-01
9.74240005e-02 -4.69473988e-01 1.27801642e-01 5.96821189e-01
-2.98757762e-01 -9.86260414e-01 -2.45262712e-01 2.26803273e-01
-3.41358274e-01 -1.24318801e-01 -5.22509992e-01 6.36066973e-01
5.71127944e-02 9.06902850e-01 2.57998168e-01 -3.59933019e-01
4.81432319e-01 3.03701013e-01 3.04392815e-01 -3.99636865e-01
-4.74504888e-01 1.83740288e-01 1.69717818e-01 -8.16256523e-01
-7.42088735e-01 -4.30305272e-01 -1.35129583e+00 8.37341696e-02
-1.64349601e-01 3.27970684e-02 5.86662769e-01 1.24034274e+00
1.46361798e-01 9.20413077e-01 4.07675594e-01 -4.49297071e-01
-2.69004479e-02 -6.32104635e-01 -1.98456600e-01 9.21910167e-01
1.29518613e-01 -5.75512469e-01 -2.99177498e-01 3.45971256e-01] | [10.463918685913086, 1.2403932809829712] |
a28a10b4-192f-4513-8870-6968c24782b1 | coordinated-frequency-constrained-stochastic | 2305.11440 | null | https://arxiv.org/abs/2305.11440v1 | https://arxiv.org/pdf/2305.11440v1.pdf | Coordinated Frequency-Constrained Stochastic Economic Dispatch for Integrated Transmission and Distribution System via Distributed Optimization | When large-scale uncertain centralized and distributed renewable energy sources are connected to a power system, separate dispatching of the transmission power system (TPS) and the active distribution network (ADN) will lower the network security and frequency security of the system. To address these problems, this paper proposes a coordinated frequency-constrained stochastic economic dispatch (CFC-SED) model for an integrated transmission and distribution (ITD) system. In this model, the dynamic frequency security constraints and network security constraints of the ITD system are constructed, and the joint chance constraints are adopted to handle the uncertainty. Then, the control parameters of inverter-based resources, the base point power, and the regulation reserve of all dispatchable resources in the ITD system are jointly optimized for the minimum operating cost. TPS and ADNs can deliver base point power bidirectionally and provide frequency regulation support bidirectionally, which extend the existing reserve assumption in ITD dispatch and enhance the operational security of the ITD system. Moreover, based on the alternating direction of multiplier algorithm, a two-layer distributed optimization framework is proposed to solve the CFC-SED model. Case studies show that the CFC-SED model can fully utilize the potential of multiple regulation resources to improve the security performance of the ITD system, and TPS and ADNs can be coordinated efficiently through the proposed distributed optimization framework. | ['Zhengshuo Li', 'Ye Tian'] | 2023-05-19 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [-5.20802498e-01 -6.87446892e-02 -5.73841870e-01 3.29218119e-01
4.95323539e-02 -1.07207918e+00 8.25830549e-02 -1.46440461e-01
2.76504755e-01 1.24171770e+00 -1.06484490e-02 -4.55556810e-01
-6.80983424e-01 -1.19875288e+00 1.20847858e-01 -1.10800529e+00
-1.72834992e-01 7.30504766e-02 -3.10748279e-01 -5.19945145e-01
9.74405743e-03 6.32813156e-01 -9.36081946e-01 -4.36054647e-01
1.07482862e+00 1.38055778e+00 3.16407323e-01 -2.28770897e-01
2.33051389e-01 -2.96506640e-02 -1.12139499e+00 2.79130429e-01
3.33109736e-01 -1.65968344e-01 -1.70956533e-02 -2.32207343e-01
-1.21563506e+00 -7.15116799e-01 -2.81553179e-01 1.20872593e+00
9.31402266e-01 4.37603235e-01 5.54943025e-01 -2.12025738e+00
-4.86050844e-01 6.62758946e-01 -1.05245173e+00 3.65498841e-01
-1.07289953e-02 -1.75473928e-01 9.11379397e-01 -6.31803036e-01
1.97926939e-01 9.89906728e-01 -8.89335796e-02 2.20313281e-01
-8.73560548e-01 -6.91846788e-01 5.62276781e-01 -4.50453088e-02
-1.35390007e+00 1.72562569e-01 7.80243635e-01 -3.09743360e-02
1.32701349e+00 2.33743936e-01 9.96726692e-01 -5.33537008e-02
3.81105840e-01 5.28960109e-01 5.76441646e-01 -3.63082290e-01
2.55355418e-01 -2.00448930e-01 -3.28337699e-01 3.44427340e-02
8.16839397e-01 -3.88724804e-02 -1.80908903e-01 -8.63515213e-02
5.39244890e-01 -1.80571437e-01 -5.26778817e-01 -6.52807504e-02
-7.02352285e-01 6.81952238e-01 1.18572570e-01 3.99997145e-01
-2.86660701e-01 -2.85785705e-01 3.26969415e-01 1.66929424e-01
7.53139317e-01 -4.11277898e-02 -4.99793291e-01 2.92796075e-01
-6.93526447e-01 -1.41362384e-01 7.25974858e-01 1.20135617e+00
-3.74433808e-02 7.61398852e-01 -5.42169549e-02 3.70147318e-01
4.15381610e-01 9.65363443e-01 1.10959738e-01 -6.14611745e-01
6.66150093e-01 3.38651240e-01 3.75717789e-01 -8.54177237e-01
-4.94913965e-01 -8.06897700e-01 -9.90007401e-01 3.38602901e-01
-7.77733028e-02 -8.20193172e-01 2.83939741e-03 1.59800863e+00
4.46852893e-01 -1.80532008e-01 -7.43434131e-02 7.29247391e-01
4.33000214e-02 1.20505857e+00 -5.83746374e-01 -1.13019514e+00
1.10065246e+00 -4.29635912e-01 -1.24597371e+00 2.86203861e-01
2.72178769e-01 -7.39458799e-01 -3.68612865e-03 9.88209769e-02
-1.63445520e+00 2.87276190e-02 -1.36045408e+00 3.89146686e-01
-3.81895244e-01 3.88882071e-01 3.14279720e-02 5.87185144e-01
-7.52450287e-01 -5.66821843e-02 -4.36722189e-01 3.91013980e-01
2.28261158e-01 2.43287995e-01 2.09800377e-01 1.50074527e-01
-1.61694944e+00 1.24020433e+00 6.88039541e-01 8.61476481e-01
-4.06041175e-01 -8.30519557e-01 -7.02625930e-01 4.64634359e-01
8.27973306e-01 -6.21581495e-01 9.79524732e-01 -4.86486815e-02
-1.64919901e+00 -1.71481192e-01 2.58806378e-01 -9.37695205e-02
2.69808859e-01 4.84061807e-01 -5.39023101e-01 2.59559602e-01
9.00312960e-02 -4.15355444e-01 3.32417399e-01 -7.65471339e-01
-8.98644388e-01 -1.36521250e-01 5.48802428e-02 5.86056769e-01
-5.36547542e-01 -1.84804037e-01 4.00011361e-01 -8.81745458e-01
-3.62249166e-01 -4.71211970e-01 -3.27295512e-01 -4.10460263e-01
-3.94554257e-01 -5.75766504e-01 1.01750469e+00 -7.37314403e-01
1.41850591e+00 -2.06591105e+00 5.63458622e-01 6.66981518e-01
-3.32912803e-01 7.85570294e-02 2.36654431e-01 7.87141502e-01
-2.75642931e-01 1.68410093e-01 -6.29267544e-02 5.90728149e-02
5.56893408e-01 6.18260205e-01 -1.93222612e-01 4.08915848e-01
-7.13487044e-02 3.76125604e-01 -7.72324502e-01 9.24838930e-02
3.43504369e-01 -6.85704350e-02 -1.69781581e-01 -2.21532077e-01
-1.63179353e-01 -8.33116770e-02 -7.93849885e-01 5.95219791e-01
1.08427572e+00 2.67646253e-01 1.88175395e-01 -2.84116149e-01
-4.24946070e-01 -2.11722121e-01 -1.65206242e+00 1.32112265e+00
-8.16477895e-01 1.81460381e-01 8.08429122e-01 -1.41024840e+00
5.42817175e-01 2.78711259e-01 9.36168373e-01 -7.91462064e-01
-1.31735727e-01 3.59578997e-01 -1.95217803e-02 -1.05181850e-01
2.35592052e-01 -3.92419510e-02 -4.67526354e-02 8.00749421e-01
1.19516738e-01 -3.96724463e-01 4.39678162e-01 2.28833556e-01
1.99045390e-01 -2.31273249e-01 3.24789852e-01 -9.20866191e-01
8.80259037e-01 -4.29067194e-01 1.17209077e+00 -2.02806771e-01
1.53116688e-01 -4.84173179e-01 8.36699784e-01 3.15819234e-01
-6.90053344e-01 -8.96660686e-01 -1.65487304e-01 3.79937589e-01
5.58417261e-01 7.29884282e-02 -1.10825367e-01 -6.15233362e-01
3.12176108e-01 1.27974212e+00 1.95123500e-03 -2.70522237e-01
-2.42019102e-01 -1.05987120e+00 -2.20927998e-01 1.36620685e-01
4.92145449e-01 5.70600592e-02 -1.22721262e-01 2.46011212e-01
-2.15414073e-02 -7.36694276e-01 -7.81774282e-01 2.72155136e-01
-1.69103220e-01 -8.68477106e-01 -8.17061722e-01 -6.38169467e-01
9.87077236e-01 1.89559758e-01 6.85310960e-01 3.03907469e-02
-5.23969978e-02 2.42090598e-01 1.18281119e-01 -5.64634740e-01
2.09703341e-01 -5.21105938e-02 4.57023442e-01 -3.95596653e-01
-7.57269502e-01 -5.42592824e-01 -5.08387685e-01 5.91491878e-01
-8.45008254e-01 -1.59995809e-01 -1.69389322e-01 8.04629207e-01
5.48037946e-01 1.43349528e+00 1.42571068e+00 2.27921560e-01
9.59296703e-01 -8.68120432e-01 -1.31904542e+00 6.14084303e-01
-7.91014135e-01 -3.38157922e-01 8.49459112e-01 -1.34972662e-01
-1.32135427e+00 -4.50607508e-01 3.59457403e-01 -2.83108383e-01
9.21698928e-01 7.00384557e-01 -8.32222879e-01 -1.66316181e-01
-6.80585861e-01 -2.11253036e-02 -8.03771913e-02 -1.81919917e-01
1.83117956e-01 3.84042054e-01 3.38869914e-02 -7.22010612e-01
1.40629661e+00 5.87060340e-02 8.23951364e-01 -2.34885052e-01
-3.73704672e-01 1.86707333e-01 1.23340925e-02 -3.88849258e-01
2.38182902e-01 -1.09556639e+00 -1.33134675e+00 5.95020771e-01
-9.24681604e-01 4.09460962e-02 -4.95229453e-01 8.86833012e-01
-4.25131284e-02 3.52877975e-01 -1.29425526e-01 -1.19040334e+00
-2.75388181e-01 -1.05819392e+00 4.19776738e-01 6.28013670e-01
7.66310215e-01 -1.09163332e+00 -3.58872026e-01 -2.13189393e-01
5.20874560e-01 5.25811434e-01 9.84231532e-01 -9.90504324e-02
-5.48270345e-01 2.25063786e-01 1.30634815e-01 5.67254007e-01
4.40969408e-01 2.87941724e-01 4.00582142e-02 -8.46362054e-01
3.52688193e-01 2.70249695e-01 -3.82486470e-02 5.06009102e-01
9.88416016e-01 -5.21248400e-01 -3.92972946e-01 3.33887368e-01
1.96637523e+00 5.94489217e-01 1.67498484e-01 1.12283736e-01
-1.77030176e-01 5.68683922e-01 8.80425632e-01 9.54034984e-01
5.21988034e-01 4.68326867e-01 7.24915683e-01 -1.29526649e-02
6.96569204e-01 7.80296773e-02 3.35473567e-01 6.97853327e-01
-1.85438246e-01 -8.71705651e-01 -1.97891846e-01 3.21103275e-01
-1.68450844e+00 -9.37188685e-01 1.07046455e-01 2.06104016e+00
3.49235028e-01 5.82841747e-02 -8.42540264e-02 3.54191184e-01
1.01601923e+00 -5.73981218e-02 -7.44259536e-01 -6.06589794e-01
-6.11277521e-01 -1.68939948e-01 7.76013196e-01 3.20524305e-01
-2.33020425e-01 -2.84790546e-01 4.86251879e+00 1.40013885e+00
-8.70614469e-01 -7.93001056e-02 9.02481318e-01 -5.15067101e-01
-5.54275811e-01 -1.70534790e-01 -7.02119410e-01 1.10790658e+00
7.26793408e-01 -1.45558286e+00 7.37423897e-01 4.81614828e-01
8.17433238e-01 -5.34279287e-01 -4.66054380e-01 4.41797674e-01
-4.34848785e-01 -1.06308317e+00 -3.51746500e-01 4.50589716e-01
1.10642278e+00 -3.39114219e-01 -1.04982927e-01 2.57096812e-02
9.61520430e-03 -4.25914347e-01 7.16705739e-01 4.17241842e-01
6.12739861e-01 -1.76417780e+00 8.52899790e-01 5.11408985e-01
-1.64328551e+00 -6.91855967e-01 -1.38562217e-01 7.61648500e-03
1.03436470e+00 1.12860775e+00 1.61367267e-01 1.50336003e+00
8.44791412e-01 4.37499255e-01 4.38270390e-01 6.47423148e-01
-4.84879762e-01 5.27682016e-03 -8.74704838e-01 -1.40606118e-02
7.22375214e-02 -8.24289680e-01 6.34981811e-01 5.20647764e-02
6.27311230e-01 6.37136221e-01 4.01218206e-01 5.23838222e-01
-7.91560113e-02 -9.06230286e-02 -3.15878153e-01 5.55573143e-02
1.12358308e+00 1.55520940e+00 -6.87354565e-01 -1.16857514e-01
-3.46801609e-01 1.08342692e-01 -5.46807468e-01 5.42681932e-01
-1.01504779e+00 -7.47131109e-01 6.19765043e-01 -5.77031851e-01
6.19566552e-02 -4.07315791e-01 -4.62475151e-01 -1.18301797e+00
3.66644889e-01 -2.03208506e-01 5.61660349e-01 -7.09139884e-01
-1.52761400e+00 -3.13011631e-02 2.50297159e-01 -1.39094865e+00
-3.79301280e-01 -1.67163208e-01 -1.10262990e+00 1.48860657e+00
-2.08637285e+00 -6.38879657e-01 4.43309456e-01 8.04943919e-01
3.32145065e-01 -3.93105507e-01 3.90262961e-01 6.74185097e-01
-1.38019228e+00 1.89979821e-01 8.83121669e-01 -2.53764898e-01
-1.90419883e-01 -1.01644206e+00 -6.37172699e-01 1.27830803e+00
-7.87502468e-01 2.38401264e-01 2.71897823e-01 -3.58140439e-01
-1.62157309e+00 -6.22911811e-01 2.75938451e-01 7.67271161e-01
8.65160644e-01 -3.46778691e-01 -2.63777196e-01 1.20731033e-01
9.10620332e-01 -1.76136181e-01 6.34656608e-01 -6.88183844e-01
6.63381398e-01 -2.85985380e-01 -1.53201783e+00 5.18226683e-01
6.36073649e-01 -2.68145353e-01 -1.70583189e-01 5.88736475e-01
6.22335255e-01 -4.18033987e-01 -1.20871866e+00 3.62643480e-01
1.12965867e-01 1.93123624e-01 1.08025706e+00 -1.31859556e-01
-2.55961329e-01 -6.06980145e-01 -1.62984148e-01 -2.12665868e+00
-1.08149871e-01 -1.07655656e+00 -4.20517296e-01 1.65932274e+00
3.50999326e-01 -1.37354565e+00 3.17615680e-02 4.40836281e-01
-2.33938932e-01 -7.34645128e-01 -1.83750749e+00 -1.05333090e+00
2.21942678e-01 2.71128178e-01 1.16371429e+00 9.16313112e-01
5.33890665e-01 1.43458173e-01 -8.28974007e-04 8.63556266e-01
7.79593587e-01 2.49624550e-01 -3.18643153e-01 -6.82665586e-01
1.54693559e-01 -5.53690612e-01 4.64249790e-01 -3.97057652e-01
3.06508183e-01 -7.49130607e-01 -6.94789052e-01 -1.90349901e+00
-6.66211665e-01 -3.54208857e-01 -3.55297446e-01 4.49643552e-01
1.90408707e-01 -4.09705848e-01 3.46225917e-01 5.00145555e-02
2.93071061e-01 1.09453464e+00 1.42859328e+00 -2.65936792e-01
2.74262950e-02 3.56463641e-01 -3.38678926e-01 3.66561502e-01
8.44810605e-01 -5.22503862e-04 -1.09760725e+00 -4.41160560e-01
3.39556336e-01 8.88344288e-01 -3.96846145e-01 -3.84185821e-01
3.95432740e-01 -6.20187700e-01 2.51179159e-01 -1.08097792e+00
2.56907362e-02 -1.60261106e+00 3.18558931e-01 9.06306803e-01
3.33299845e-01 2.05533415e-01 1.19906753e-01 5.42468846e-01
-6.12216592e-02 -2.65115380e-01 3.71201783e-01 3.50334853e-01
-1.43146962e-01 3.43831956e-01 -7.84962475e-01 -2.89449930e-01
1.66321492e+00 1.85481489e-01 -7.92233169e-01 -2.99661279e-01
-6.04747772e-01 1.59516251e+00 -1.44441172e-01 4.17538643e-01
3.10801417e-01 -1.25353897e+00 -5.08918941e-01 -8.22008215e-03
-9.89168048e-01 1.42498270e-01 4.36398923e-01 6.11769617e-01
-7.85211027e-02 5.40239096e-01 -2.51819640e-01 -2.26135343e-01
-4.72342044e-01 4.19219077e-01 6.59605682e-01 -2.25188330e-01
1.37069598e-01 2.53251940e-01 -4.05463040e-01 -3.98655012e-02
-2.45514467e-01 -2.32441157e-01 -2.07866207e-01 9.05559659e-01
1.28331989e-01 1.06938338e+00 4.71326187e-02 -5.63056841e-02
-4.81191158e-01 3.06954443e-01 5.36601543e-01 -9.04248562e-03
1.60803020e+00 -8.20356131e-01 -4.74476844e-01 -2.19691128e-01
9.32718992e-01 1.57869399e-01 -8.05462182e-01 1.66043922e-01
-7.31474519e-01 -2.73058742e-01 4.64206159e-01 -9.08250451e-01
-1.83648682e+00 2.32934654e-01 4.32856753e-02 8.05560470e-01
1.69216120e+00 -8.25165331e-01 5.10245085e-01 9.66643468e-02
6.97608769e-01 -1.56781983e+00 -4.69857156e-01 2.62944013e-01
7.28776038e-01 -2.99137652e-01 3.77241582e-01 -4.88086164e-01
-2.12425232e-01 1.33533931e+00 4.42817688e-01 -8.63142759e-02
1.02001548e+00 6.14317417e-01 -9.63034332e-01 5.22819281e-01
-7.44411886e-01 2.38573954e-01 -1.90639734e-01 2.91425407e-01
-3.41755271e-01 2.41276771e-02 -9.82038617e-01 1.00385308e+00
3.17855537e-01 -1.60038114e-01 9.92089689e-01 1.07277274e+00
1.52360694e-02 -9.69980478e-01 -3.80376101e-01 3.76582921e-01
-4.13166583e-01 3.08210760e-01 5.25449395e-01 6.49440348e-01
5.07931411e-01 1.17494881e+00 1.35129526e-01 3.85316223e-01
6.02630973e-01 -4.53289062e-01 8.08233172e-02 -3.09835285e-01
-1.89305857e-01 4.12869871e-01 1.03620306e-01 2.78577004e-02
-1.98107213e-01 -6.26208484e-01 -1.54433250e+00 -4.04276252e-01
-1.17520201e+00 9.47935224e-01 9.35470343e-01 9.41571116e-01
1.21795945e-01 7.26689696e-01 1.78628111e+00 -7.20663607e-01
-7.46688068e-01 -3.27775389e-01 -1.28079689e+00 -1.04394853e+00
-7.78772533e-02 -7.25875437e-01 -7.83964396e-01 -7.41732836e-01] | [5.658140182495117, 2.5312445163726807] |
d68eddb0-f471-4eba-8afb-81ea39b61e24 | matchformer-interleaving-attention-in | 2203.09645 | null | https://arxiv.org/abs/2203.09645v3 | https://arxiv.org/pdf/2203.09645v3.pdf | MatchFormer: Interleaving Attention in Transformers for Feature Matching | Local feature matching is a computationally intensive task at the subpixel level. While detector-based methods coupled with feature descriptors struggle in low-texture scenes, CNN-based methods with a sequential extract-to-match pipeline, fail to make use of the matching capacity of the encoder and tend to overburden the decoder for matching. In contrast, we propose a novel hierarchical extract-and-match transformer, termed as MatchFormer. Inside each stage of the hierarchical encoder, we interleave self-attention for feature extraction and cross-attention for feature matching, yielding a human-intuitive extract-and-match scheme. Such a match-aware encoder releases the overloaded decoder and makes the model highly efficient. Further, combining self- and cross-attention on multi-scale features in a hierarchical architecture improves matching robustness, particularly in low-texture indoor scenes or with less outdoor training data. Thanks to such a strategy, MatchFormer is a multi-win solution in efficiency, robustness, and precision. Compared to the previous best method in indoor pose estimation, our lite MatchFormer has only 45% GFLOPs, yet achieves a +1.3% precision gain and a 41% running speed boost. The large MatchFormer reaches state-of-the-art on four different benchmarks, including indoor pose estimation (ScanNet), outdoor pose estimation (MegaDepth), homography estimation and image matching (HPatch), and visual localization (InLoc). | ['Rainer Stiefelhagen', 'Kunyu Peng', 'Kailun Yang', 'Jiaming Zhang', 'Qing Wang'] | 2022-03-17 | null | null | null | null | ['homography-estimation'] | ['computer-vision'] | [ 9.48884413e-02 -2.07641706e-01 -7.87223130e-02 -2.47957587e-01
-1.12769747e+00 -5.77942014e-01 4.95544642e-01 2.34541714e-01
-6.84598804e-01 2.97875404e-01 1.02538370e-01 -1.75273493e-02
1.58303812e-01 -9.59184110e-01 -1.11704755e+00 -3.59069347e-01
1.82735845e-01 2.52339333e-01 5.40426254e-01 -8.76856595e-02
2.21810803e-01 4.87709939e-01 -1.65991509e+00 8.49175602e-02
6.48330152e-01 1.42443001e+00 4.20676023e-01 8.29937756e-01
2.46036112e-01 4.08830464e-01 -2.03880876e-01 -5.13359189e-01
5.40579736e-01 1.00435048e-01 -5.85344017e-01 -6.84958175e-02
1.33464074e+00 -6.73352480e-01 -6.84983134e-01 7.27278709e-01
8.39616358e-01 -6.26903996e-02 8.64362940e-02 -9.47267592e-01
-1.24468461e-01 -3.48930410e-03 -7.25638866e-01 -8.70856419e-02
6.00841880e-01 3.37045342e-01 9.39038932e-01 -1.13788092e+00
8.04245651e-01 1.07905817e+00 1.12308919e+00 -9.80103612e-02
-1.17795873e+00 -6.30537570e-01 6.71070591e-02 2.05713883e-02
-1.69633675e+00 -6.56593263e-01 3.02062213e-01 -2.56095380e-01
1.28614235e+00 2.99075395e-01 7.90000200e-01 8.33201528e-01
4.53009516e-01 7.03893900e-01 8.64582121e-01 -7.14321584e-02
8.97183567e-02 -3.77971560e-01 -4.61449832e-01 8.90742123e-01
2.14473784e-01 3.86451274e-01 -8.39806676e-01 2.13932320e-01
1.04395711e+00 1.95688292e-01 -1.18403092e-01 -6.56459808e-01
-1.40107954e+00 5.29165328e-01 1.10730767e+00 -6.04961216e-02
-2.43388161e-01 6.22238517e-01 2.73917794e-01 2.79445082e-01
3.07252973e-01 5.11755824e-01 -3.79091918e-01 -4.35538553e-02
-1.41228843e+00 3.47691178e-01 5.42319536e-01 1.27191138e+00
1.21216369e+00 -2.86114722e-01 -1.20276302e-01 4.55325544e-01
2.02342436e-01 8.95029783e-01 1.27243668e-01 -9.10295188e-01
6.43991768e-01 5.03522098e-01 -1.63192488e-02 -1.24461055e+00
-5.38106084e-01 -8.46013129e-01 -7.24472880e-01 6.49135560e-02
3.86776090e-01 1.57222465e-01 -8.96507502e-01 1.34480393e+00
3.99948210e-01 3.43680941e-03 -3.48651886e-01 1.01381791e+00
7.76484191e-01 4.53838617e-01 -1.89026386e-01 3.48463863e-01
1.49073422e+00 -1.24845099e+00 -2.24408492e-01 -7.06882834e-01
4.87986445e-01 -1.11381757e+00 7.01142251e-01 1.49236843e-01
-1.04477513e+00 -6.91425443e-01 -1.25906706e+00 -7.10590899e-01
-3.01193506e-01 1.43481836e-01 7.57171929e-01 4.44311589e-01
-9.76559639e-01 5.66345513e-01 -8.94327819e-01 -4.90547568e-01
4.31057870e-01 7.18445599e-01 -7.40526378e-01 -3.82974476e-01
-6.72189534e-01 6.82787657e-01 1.49542555e-01 -2.75304783e-02
-4.82687891e-01 -8.24246883e-01 -1.13648689e+00 9.56443548e-02
3.99253935e-01 -1.14716399e+00 1.05787528e+00 -6.29941165e-01
-1.26162887e+00 6.89402997e-01 -2.98495501e-01 -3.88999671e-01
6.19021118e-01 -5.04607558e-01 1.44398168e-01 4.42628562e-02
2.91642994e-01 1.07297790e+00 7.10436761e-01 -7.89007425e-01
-7.79573798e-01 -3.76145393e-01 -3.35030220e-02 2.42486104e-01
-5.93827367e-02 -2.57096887e-01 -1.07785273e+00 -4.25753146e-01
5.17869532e-01 -8.66511703e-01 -4.06459451e-01 5.12707472e-01
-2.49075532e-01 3.77858073e-01 6.87801957e-01 -5.25700152e-01
1.09847689e+00 -2.26264167e+00 -1.05836138e-01 1.55920014e-01
2.45597348e-01 4.96188141e-02 -4.35948372e-02 4.94250357e-01
3.14636648e-01 -3.04286510e-01 3.60774398e-02 -7.03381002e-01
6.70728460e-02 5.00516780e-02 2.85970327e-02 7.41128504e-01
2.28685215e-01 1.01676047e+00 -9.68899190e-01 -6.17741346e-01
5.85596561e-01 5.58927774e-01 -1.03922164e+00 1.11307599e-01
1.10872574e-01 9.22370851e-02 -2.33600751e-01 9.28687096e-01
9.86728668e-01 -3.02037776e-01 -4.92957644e-02 -6.24857187e-01
-4.81385827e-01 2.97668308e-01 -1.48142743e+00 2.34032536e+00
-8.65927458e-01 8.32752824e-01 1.43247873e-01 -2.97146440e-01
7.81835198e-01 -1.75832883e-01 3.69781494e-01 -1.05385149e+00
1.34588867e-01 6.34367049e-01 -3.67927790e-01 -3.10096424e-02
8.93109679e-01 3.92926753e-01 -2.13733733e-01 -7.46739134e-02
2.42491856e-01 -3.20741147e-01 1.63210377e-01 7.40787759e-02
1.34649539e+00 4.75141108e-01 4.61715758e-01 -1.58551529e-01
2.66136825e-01 -1.91227645e-01 4.61829096e-01 7.61786222e-01
2.47353036e-02 9.13382292e-01 8.80468115e-02 -6.12349927e-01
-1.19235134e+00 -9.76480424e-01 -1.61289915e-01 9.63945568e-01
3.83827657e-01 -8.23686719e-01 -4.96856332e-01 -3.87547493e-01
2.18557358e-01 -1.07924260e-01 -3.47372949e-01 7.35062584e-02
-6.72458351e-01 -4.09088969e-01 3.69030774e-01 8.06796372e-01
9.59820092e-01 -4.43663001e-01 -1.06292582e+00 4.09477711e-01
-5.30697852e-02 -1.41311014e+00 -6.07391536e-01 5.67562163e-01
-5.62016368e-01 -8.90550792e-01 -5.33626795e-01 -6.42981946e-01
5.60522020e-01 2.14003846e-01 1.20682776e+00 -9.51549485e-02
-5.13718188e-01 2.25724131e-02 -1.09693408e-01 5.45695797e-02
3.77912253e-01 4.34549719e-01 -2.63313174e-01 -2.31992856e-01
-1.31873384e-01 -5.40020645e-01 -1.07989216e+00 4.87040818e-01
-5.18818974e-01 2.02410743e-01 9.52107310e-01 1.03141487e+00
8.92370582e-01 -6.08555436e-01 -2.79177874e-01 -3.48631591e-01
-2.84889787e-01 -8.69026557e-02 -9.05492306e-01 -8.70649703e-03
-3.92795682e-01 2.21862465e-01 5.39938033e-01 -2.11715832e-01
-5.95844090e-01 7.53085077e-01 -4.26723003e-01 -3.58039051e-01
1.41281262e-01 3.31913941e-02 -2.30042383e-01 -6.87027574e-01
6.09596312e-01 1.36653677e-01 -2.67253935e-01 -3.93736869e-01
3.86031955e-01 2.88581252e-01 8.34464252e-01 -3.84074062e-01
1.01189971e+00 5.17258644e-01 2.53207684e-01 -5.76276243e-01
-7.18758881e-01 -6.46375477e-01 -8.06136549e-01 -9.14871972e-03
7.77753115e-01 -1.37317252e+00 -9.45673585e-01 5.21119118e-01
-1.11262834e+00 -2.32313111e-01 -2.54967272e-01 3.85184795e-01
-6.31659329e-01 3.80659215e-02 -4.29779738e-01 -3.25877815e-01
-3.84582520e-01 -1.42710948e+00 1.75806570e+00 2.91179448e-01
-7.67552033e-02 -5.31047285e-01 -1.51301369e-01 2.71004051e-01
7.29747415e-01 3.77370417e-01 1.07517123e-01 -8.92484933e-02
-1.27449787e+00 -4.31690663e-01 -6.62699997e-01 -1.67041063e-01
-3.03724229e-01 -2.98842192e-01 -1.18669820e+00 -3.75140488e-01
-5.39538443e-01 -3.83078575e-01 9.59123015e-01 8.76609236e-02
1.04362702e+00 -9.75749344e-02 -4.14182514e-01 1.34178185e+00
1.70624769e+00 -3.78966302e-01 6.60923839e-01 5.19051731e-01
9.01615977e-01 3.90879437e-02 7.04164386e-01 4.42309558e-01
6.82846606e-01 1.16583097e+00 7.16879070e-01 -4.97024387e-01
-5.00120819e-01 -4.12947118e-01 2.09330320e-01 5.14289320e-01
6.44365028e-02 -1.38014495e-01 -9.08192873e-01 3.25799614e-01
-1.87446856e+00 -6.74915254e-01 -1.80256099e-01 2.35921454e+00
6.37758315e-01 5.69142550e-02 -2.37714052e-01 -2.74342746e-02
1.89507842e-01 3.23060572e-01 -3.80953610e-01 -1.76585078e-01
-1.83338150e-01 4.28596914e-01 1.21696222e+00 5.93932271e-01
-1.47258520e+00 1.09724128e+00 5.82479811e+00 1.07840538e+00
-1.18592250e+00 2.01656129e-02 5.19945741e-01 -1.15862064e-01
-1.90107524e-02 1.11101635e-01 -9.46699262e-01 2.67534792e-01
7.34690249e-01 4.61792082e-01 2.44306907e-01 1.04726326e+00
-2.95776486e-01 -5.36424756e-01 -1.25833547e+00 1.34068894e+00
1.97851937e-03 -1.48298681e+00 -3.45795661e-01 2.04496935e-01
6.37746871e-01 4.15489525e-01 -3.99051160e-02 8.02941993e-02
-5.48265129e-02 -9.13052738e-01 9.72331345e-01 3.86108756e-01
1.00901508e+00 -8.76629174e-01 8.35544586e-01 5.71037009e-02
-1.77754581e+00 -5.02742864e-02 -4.23345804e-01 -1.04448937e-01
4.36039083e-02 6.79591417e-01 -5.41020572e-01 7.97186136e-01
7.11922646e-01 7.24134445e-01 -8.62896442e-01 1.34720850e+00
-5.23331016e-02 -9.48744118e-02 -6.31115258e-01 1.25185698e-01
2.98199117e-01 3.63922119e-01 1.17641501e-01 1.38516819e+00
3.08753937e-01 -2.17151985e-01 4.49308515e-01 5.99399686e-01
-1.18577361e-01 -3.43925916e-02 -3.36600423e-01 3.79746556e-01
5.69204867e-01 1.32119834e+00 -7.72453427e-01 -2.29645193e-01
-3.16483974e-01 1.36508083e+00 4.31106418e-01 -1.07838899e-01
-8.46475363e-01 -5.37531614e-01 7.34356761e-01 4.62002724e-01
5.86059272e-01 -5.26773930e-01 -3.86195779e-01 -1.01404917e+00
2.43900180e-01 -7.87275553e-01 -1.02861188e-01 -5.68803251e-01
-6.88153803e-01 5.47118247e-01 -3.39002252e-01 -1.25430334e+00
-2.05460295e-01 -5.52181721e-01 -6.87700436e-02 7.00530291e-01
-1.63113129e+00 -1.51676333e+00 -7.86263227e-01 5.41880846e-01
4.58562583e-01 3.96593362e-01 7.50832200e-01 7.71002173e-01
-4.12942350e-01 8.62701118e-01 -1.08684614e-01 1.58049256e-01
8.33580256e-01 -1.08494854e+00 8.86485517e-01 9.07593489e-01
1.38476342e-01 5.39095879e-01 3.67912561e-01 -4.88618016e-01
-1.93799675e+00 -1.00585330e+00 1.12523389e+00 -4.08549249e-01
4.11032945e-01 -7.30018914e-01 -1.87809855e-01 2.55945683e-01
1.46166021e-02 5.75945795e-01 1.83610380e-01 6.96195066e-02
-6.14692211e-01 -4.20180976e-01 -9.72945213e-01 4.78821367e-01
1.34203613e+00 -6.34619534e-01 6.61981553e-02 3.72290671e-01
5.35330772e-01 -1.00633860e+00 -8.44471633e-01 3.73668224e-01
8.58963490e-01 -1.29280782e+00 1.38643384e+00 2.91680396e-01
4.80590522e-01 -5.38039923e-01 -5.28503597e-01 -6.54532850e-01
-4.72424448e-01 -6.64137542e-01 1.31128684e-01 9.38801467e-01
1.24828219e-01 -4.09028113e-01 8.13812077e-01 2.08810031e-01
-3.03955197e-01 -9.89864647e-01 -1.03803790e+00 -8.38477969e-01
-5.23846030e-01 -3.92211437e-01 6.22900724e-01 4.51611876e-01
-4.41256016e-01 1.79286718e-01 -3.77342522e-01 1.86326966e-01
6.10984504e-01 2.47572720e-01 1.16826761e+00 -8.21108222e-01
-4.37458187e-01 -3.46737266e-01 -8.16822171e-01 -1.67508006e+00
-2.42336884e-01 -6.60978854e-01 2.51504958e-01 -1.56671345e+00
9.07813683e-02 -5.45387089e-01 2.86473185e-01 6.52842402e-01
-8.06121081e-02 8.01723480e-01 3.16608518e-01 4.42621065e-03
-8.14406335e-01 3.61085266e-01 9.84216154e-01 8.14395770e-03
6.44862726e-02 -3.87655944e-01 -2.38992408e-01 5.35973370e-01
2.81451136e-01 -3.81852329e-01 3.73609848e-02 -6.54512405e-01
3.56661826e-01 -3.57211903e-02 6.28106296e-01 -1.59698153e+00
6.29230559e-01 1.75458789e-01 8.13514054e-01 -7.27030158e-01
7.47900963e-01 -8.45469177e-01 2.05453068e-01 6.70155644e-01
2.57353127e-01 2.72591382e-01 1.96272269e-01 1.84694588e-01
-2.97927737e-01 3.85661125e-01 6.31446302e-01 -3.35346758e-02
-8.81817639e-01 4.96400267e-01 9.76624712e-02 -5.70589192e-02
7.58494258e-01 -6.66765749e-01 -3.35950375e-01 -1.17699109e-01
-2.89052129e-01 1.55453056e-01 8.33739877e-01 2.93440193e-01
5.60647130e-01 -1.26850057e+00 -5.40441394e-01 5.08010745e-01
1.82276517e-01 2.42177993e-01 3.10443521e-01 1.00032842e+00
-9.33205843e-01 3.32290173e-01 -1.98842302e-01 -8.59477520e-01
-1.06923389e+00 2.27872133e-01 3.49984527e-01 -4.04994547e-01
-6.02505028e-01 1.11540961e+00 1.93127319e-01 -2.32775956e-01
2.20315516e-01 -4.39083099e-01 5.83977878e-01 1.06281072e-01
2.97997206e-01 3.11559796e-01 4.20844346e-01 -6.51520729e-01
-7.12871909e-01 9.59536552e-01 7.17895031e-02 1.19941786e-01
1.13594031e+00 -7.61485472e-02 1.29462734e-01 -1.20879963e-01
1.36832356e+00 3.06446582e-01 -1.64418519e+00 -1.95123658e-01
-1.98911399e-01 -8.56418550e-01 3.10335934e-01 -6.21833563e-01
-1.08076179e+00 7.59727299e-01 7.37680733e-01 -4.66339618e-01
1.12796974e+00 -1.81092411e-01 9.48740840e-01 3.95839959e-01
7.17085123e-01 -1.04289901e+00 1.21499993e-01 6.52481973e-01
7.87924230e-01 -1.40605092e+00 4.04869378e-01 -7.11839318e-01
-8.01025778e-02 1.08667445e+00 6.15441561e-01 -3.04174393e-01
4.23409432e-01 7.54595816e-01 -1.99618578e-01 3.11990222e-03
-4.55575913e-01 -3.71032149e-01 4.68128383e-01 3.84810179e-01
4.25324827e-01 -8.81800130e-02 1.40260488e-01 1.23884827e-01
-4.34332430e-01 -3.29230160e-01 -1.87639818e-02 1.02518821e+00
-4.63044912e-01 -1.00824225e+00 -3.71946037e-01 1.19113192e-01
-4.09297571e-02 -3.89808834e-01 -1.25990242e-01 9.02358830e-01
4.34766620e-01 7.70078540e-01 2.37081304e-01 -6.12624824e-01
4.62809920e-01 -4.51673210e-01 6.52519584e-01 -4.45468545e-01
-1.06390917e+00 2.44952530e-01 9.85082909e-02 -1.51724482e+00
-6.68758079e-02 -4.29334700e-01 -8.54292274e-01 -4.59225833e-01
-4.33137625e-01 -4.88868594e-01 8.36599112e-01 7.41428792e-01
6.76344931e-01 5.05396247e-01 4.49373484e-01 -1.24506736e+00
-1.52229071e-01 -5.72122991e-01 -1.22522376e-01 -1.91092715e-02
4.62238759e-01 -4.98837143e-01 1.31059885e-02 -3.02757740e-01] | [7.850198745727539, -2.101478338241577] |
739a9991-c5e8-477b-8cf1-2a1a0cf13156 | little-motion-big-results-using-motion | 2008.04946 | null | https://arxiv.org/abs/2008.04946v1 | https://arxiv.org/pdf/2008.04946v1.pdf | Little Motion, Big Results: Using Motion Magnification to Reveal Subtle Tremors in Infants | Detecting tremors is challenging for both humans and machines. Infants exposed to opioids during pregnancy often show signs and symptoms of withdrawal after birth, which are easy to miss with the human eye. The constellation of clinical features, termed as Neonatal Abstinence Syndrome (NAS), include tremors, seizures, irritability, etc. The current standard of care uses Finnegan Neonatal Abstinence Syndrome Scoring System (FNASS), based on subjective evaluations. Monitoring with FNASS requires highly skilled nursing staff, making continuous monitoring difficult. In this paper we propose an automated tremor detection system using amplified motion signals. We demonstrate its applicability on bedside video of infant exhibiting signs of NAS. Further, we test different modes of deep convolutional network based motion magnification, and identify that dynamic mode works best in the clinical setting, being invariant to common orientational changes. We propose a strategy for discharge and follow up for NAS patients, using motion magnification to supplement the existing protocols. Overall our study suggests methods for bridging the gap in current practices, training and resource utilization. | ['Ish K. Gulati', 'Girik Malik'] | 2020-08-01 | null | null | null | null | ['motion-magnification'] | ['computer-vision'] | [ 2.60681123e-01 -1.14462674e-01 -2.68472165e-01 -1.73000559e-01
-4.14995819e-01 -8.26502502e-01 -1.46071106e-01 6.48586228e-02
-5.27224541e-01 5.13250530e-01 2.82919168e-01 -1.74119473e-01
-1.88980937e-01 -3.11826348e-01 -5.58678925e-01 -7.05423832e-01
-1.50262892e-01 2.85843074e-01 7.76665956e-02 -1.09295622e-01
-3.31488177e-02 5.50093412e-01 -8.64697039e-01 5.45320690e-01
4.35014397e-01 6.33246541e-01 2.38004744e-01 7.58097291e-01
2.24403188e-01 7.97717035e-01 -1.39180347e-01 -1.64112076e-01
1.43432051e-01 -5.91202378e-01 -6.68627739e-01 -3.23266357e-01
2.15452790e-01 -1.01052475e+00 -3.65497857e-01 8.65486205e-01
1.05663908e+00 -4.12980080e-01 5.89622080e-01 -9.05239224e-01
-3.85230064e-01 5.04102945e-01 -5.64586520e-01 7.29923427e-01
4.44580138e-01 2.35275045e-01 2.27355808e-01 -5.36072791e-01
4.13315535e-01 6.95997000e-01 1.18822742e+00 1.21370792e+00
-1.00398278e+00 -7.30959356e-01 -6.86022758e-01 3.75097334e-01
-8.38562787e-01 -5.10292411e-01 4.98679936e-01 -6.03539050e-01
9.37558770e-01 2.54887313e-01 9.20499384e-01 1.46582437e+00
3.81118566e-01 6.66163385e-01 4.88439173e-01 -1.26707807e-01
1.05007611e-01 -3.88212591e-01 -2.07280427e-01 7.55104899e-01
5.73102891e-01 -2.25875542e-01 -6.58008397e-01 -2.26138651e-01
7.79228687e-01 3.38062406e-01 -5.27913392e-01 -2.48164594e-01
-9.62021410e-01 7.49540806e-01 -3.61269206e-01 4.96730059e-01
-5.51311612e-01 4.86971855e-01 4.75228161e-01 4.67581570e-01
1.32688731e-01 1.02979086e-01 -4.02461648e-01 -9.00471687e-01
-1.06207919e+00 -2.49044538e-01 4.17741925e-01 9.76918638e-01
1.22157484e-01 -2.47173324e-01 -4.00610864e-01 6.14759505e-01
4.31750059e-01 4.59556252e-01 9.82439041e-01 -1.11585259e+00
5.63683689e-01 5.17274618e-01 -1.46809733e-02 -4.40770417e-01
-1.10279632e+00 3.70569266e-02 -6.91805005e-01 2.32270271e-01
1.43365562e-01 -5.78728259e-01 -5.06748319e-01 1.80118632e+00
2.48884171e-01 1.63259000e-01 7.62966573e-02 4.94814426e-01
1.07875085e+00 -2.80268282e-01 -1.32213503e-01 -5.57148516e-01
1.13365316e+00 -5.54071069e-01 -1.08805954e+00 -9.07399282e-02
1.01721847e+00 -6.49418712e-01 7.15154767e-01 6.71950936e-01
-1.25995862e+00 3.09442550e-01 -1.03200459e+00 2.31422722e-01
1.73296899e-01 3.36016268e-01 5.05980253e-01 9.58364069e-01
-1.27202058e+00 1.24511075e+00 -1.73344278e+00 -7.11786032e-01
7.45817900e-01 8.62258196e-01 -5.45101643e-01 3.03999156e-01
-7.90194988e-01 1.24434531e+00 5.10784686e-02 3.17528874e-01
-6.52384758e-01 -3.19820851e-01 -7.75410354e-01 -1.52123421e-01
-1.68133646e-01 -3.08224916e-01 1.50782478e+00 -3.27276349e-01
-1.46189713e+00 6.26469970e-01 -3.88322845e-02 -3.33541006e-01
9.71458554e-01 -5.59755147e-01 -2.14494690e-01 5.23649395e-01
-1.60002142e-01 1.77007049e-01 7.07895458e-01 -1.83470875e-01
-3.08450639e-01 -2.90700406e-01 -1.39913738e-01 2.25079097e-02
-4.94358689e-01 5.52678883e-01 -7.38800913e-02 -3.61971766e-01
2.19569921e-01 -9.64611530e-01 1.41695350e-01 3.41650844e-01
1.44522682e-01 1.02223389e-01 8.47567201e-01 -4.29087877e-01
1.25047719e+00 -1.96996486e+00 -2.39707187e-01 -2.25020856e-01
3.92132580e-01 4.24308449e-01 -3.39297764e-02 4.60706383e-01
-1.93386804e-02 -3.22181523e-01 -1.65005386e-01 -4.23068963e-02
-1.65520906e-01 1.74105600e-01 4.85157430e-01 9.84565139e-01
1.70271993e-01 1.00405586e+00 -9.89058793e-01 -2.81505376e-01
1.50538430e-01 4.67943758e-01 -5.62472999e-01 1.52000502e-01
3.94079655e-01 5.06715536e-01 -2.41508707e-01 7.16692448e-01
5.05156577e-01 -2.56433070e-01 4.01814580e-02 -4.29968685e-02
-1.59860864e-01 1.38140649e-01 -7.70255029e-01 1.81623220e+00
5.09045199e-02 8.89270246e-01 9.43771452e-02 -6.60296738e-01
3.21388900e-01 9.96666551e-01 6.14922762e-01 -1.46640211e-01
5.09538293e-01 4.17549253e-01 2.53930390e-01 -1.52855337e+00
-4.92796749e-01 -2.96783239e-01 7.21685946e-01 6.32148325e-01
2.96079870e-02 3.44966024e-01 2.45798007e-01 -3.42394143e-01
1.70365441e+00 3.30773056e-01 3.90967518e-01 -1.72745004e-01
-9.34854522e-02 -2.73835063e-01 9.84795094e-02 6.14861131e-01
-4.77843881e-01 1.13944316e+00 5.67243934e-01 -4.00972277e-01
-5.47214389e-01 -6.54237509e-01 4.26715147e-03 7.37922311e-01
-2.74706244e-01 5.22876419e-02 -8.65775645e-01 -5.03178000e-01
-9.79794338e-02 1.48945048e-01 -9.51633334e-01 -3.81727457e-01
-8.06112945e-01 -9.15435731e-01 8.80116343e-01 9.41068053e-01
-3.23581360e-02 -1.20180082e+00 -1.65584314e+00 6.50618255e-01
-1.76402122e-01 -7.60385036e-01 -6.07857466e-01 3.07779133e-01
-8.90736997e-01 -1.21243536e+00 -1.14186275e+00 -6.92134440e-01
2.56139696e-01 9.44680646e-02 2.67198324e-01 1.46538690e-01
-4.87334251e-01 4.61900622e-01 -4.13807362e-01 -4.63677913e-01
-3.58373493e-01 -4.08775359e-01 5.39935589e-01 -1.94217905e-01
3.86054158e-01 -9.40565586e-01 -1.12734783e+00 -3.48605774e-02
-8.87069941e-01 -1.13354065e-01 8.00444245e-01 4.56486791e-01
-7.81886056e-02 -7.53589809e-01 4.30720925e-01 -4.75959778e-01
7.67331600e-01 -5.65959871e-01 -1.56928018e-01 1.62228361e-01
-5.10611296e-01 -3.13963629e-02 1.96000606e-01 -9.55090404e-01
-4.57541078e-01 -4.82711866e-02 -1.46701008e-01 -6.28580272e-01
-1.66627407e-01 2.82568723e-01 4.82963175e-01 -2.39439234e-01
7.34646857e-01 1.57981396e-01 1.18987866e-01 -5.39060593e-01
-4.17564422e-01 8.53051007e-01 6.58870161e-01 -2.74953083e-03
2.26366460e-01 7.00314164e-01 5.14919199e-02 -7.81788826e-01
-4.81905073e-01 -5.81736386e-01 -6.64487481e-01 -1.23097450e-01
1.09080613e+00 -5.31729937e-01 -8.40417027e-01 3.97228837e-01
-1.43930757e+00 -3.89245421e-01 1.94482327e-01 1.13340473e+00
-5.65514863e-01 6.43843353e-01 -6.90874577e-01 -5.80109417e-01
-8.79650772e-01 -1.18668616e+00 8.78888309e-01 -8.01500026e-03
-7.18732476e-01 -7.34148741e-01 5.31967044e-01 8.37631896e-02
6.08132422e-01 5.71443975e-01 7.67496884e-01 -7.29134023e-01
6.80498555e-02 -2.96731859e-01 2.04717278e-01 2.76640564e-01
6.06748939e-01 -2.78101414e-02 -8.63159299e-01 -4.49176759e-01
4.45559680e-01 -2.92725801e-01 1.32978588e-01 5.23859322e-01
6.35159671e-01 -1.99858978e-01 -1.86699405e-01 6.62235379e-01
1.15425003e+00 7.12463498e-01 2.55386293e-01 3.89333099e-01
4.28181410e-01 1.25103548e-01 2.07441553e-01 5.66313148e-01
2.84027830e-02 4.07393247e-01 5.04734814e-01 6.83798790e-02
-9.81596410e-02 4.85386103e-01 3.27846646e-01 6.56129658e-01
-6.43253624e-01 1.43790608e-02 -8.95377994e-01 4.89897668e-01
-1.79477942e+00 -9.34146702e-01 -2.07450241e-01 1.90499377e+00
8.55921865e-01 -2.66981363e-01 1.73222452e-01 8.82932469e-02
5.30466616e-01 -5.84136307e-01 -2.98202455e-01 -3.91449630e-01
3.71097505e-01 2.97344834e-01 3.83544534e-01 -4.07916456e-02
-9.31527019e-01 2.88833797e-01 6.83037949e+00 1.01848185e-01
-1.50735617e+00 4.52636063e-01 -1.89148024e-01 -5.73495805e-01
2.41046265e-01 -5.48675597e-01 -2.37456396e-01 6.27154708e-01
7.42606759e-01 2.05430850e-01 3.15587014e-01 4.81511116e-01
5.53429425e-01 1.43855423e-01 -1.60815609e+00 1.03816116e+00
-4.05492820e-02 -1.17768621e+00 -5.44341624e-01 -3.51323396e-01
3.21943253e-01 5.59323967e-01 -1.32664874e-01 -2.08103448e-01
-4.51099426e-01 -1.14623785e+00 6.91870868e-01 3.50180566e-01
1.16794920e+00 -4.02710021e-01 8.13274920e-01 2.82070637e-01
-8.94880950e-01 5.42682745e-02 -3.23363990e-02 2.07216308e-01
-2.35025138e-01 -3.97856057e-01 -1.02815628e+00 -1.16276287e-01
8.76199067e-01 3.20660740e-01 -7.73052052e-02 1.22321296e+00
-5.69160879e-02 6.01567090e-01 -6.32880032e-01 -3.02753240e-01
4.05140132e-01 2.85952818e-02 4.10424381e-01 1.50320566e+00
7.55948544e-01 3.24548811e-01 -8.06401730e-01 5.72461605e-01
2.24982724e-01 -1.34529117e-02 -7.41363406e-01 -4.70172949e-02
4.98106703e-02 1.08305097e+00 -7.50200510e-01 7.88119659e-02
-8.27627957e-01 1.00167334e+00 4.61477563e-02 -5.60309365e-02
-7.97375679e-01 -2.76455998e-01 5.52239865e-02 -9.61560756e-02
2.92266130e-01 2.92894781e-01 -2.74933487e-01 -9.69410598e-01
2.57320493e-01 -8.02078605e-01 2.22841889e-01 -6.88212395e-01
-2.83525914e-01 3.08960408e-01 -1.32346386e-03 -1.64344871e+00
-2.85571367e-01 -5.35275519e-01 -1.13556969e+00 3.37031454e-01
-1.04395950e+00 -7.56430805e-01 -2.84411222e-01 5.60669541e-01
5.55666864e-01 8.01594928e-02 9.64820683e-01 4.56802577e-01
-4.11821216e-01 7.22334802e-01 1.02831863e-01 -1.92737170e-02
4.83277529e-01 -9.08245742e-01 -3.89643051e-02 5.27100205e-01
-3.82697731e-01 7.05172598e-01 9.06023741e-01 -7.73673117e-01
-1.57924819e+00 -1.00888669e+00 6.96472168e-01 -4.35681909e-01
6.20066643e-01 1.21897841e-02 -7.21872151e-01 7.84043968e-01
2.53952086e-01 -1.62442580e-01 7.24906325e-01 -6.60234094e-01
2.23575786e-01 3.54609728e-01 -1.42101276e+00 6.11009538e-01
1.02210188e+00 -1.05105881e-02 -7.19157994e-01 5.60212016e-01
2.21284270e-01 -7.00906575e-01 -6.61935031e-01 7.86529839e-01
1.01557219e+00 -7.97895968e-01 3.06444585e-01 -6.14784837e-01
4.59963351e-01 1.33612350e-01 2.37255663e-01 -9.50627863e-01
-3.75772178e-01 -1.22440279e+00 -4.20702457e-01 5.99701583e-01
4.91714627e-01 -4.33090568e-01 8.49472165e-01 5.41491807e-01
-5.08521795e-02 -9.35274899e-01 -1.01534307e+00 -6.05217993e-01
-2.31057405e-01 -5.05164027e-01 3.74691635e-01 8.42320800e-01
7.02670038e-01 6.66433126e-02 -3.35748911e-01 1.69776618e-01
1.48055598e-01 -1.17484465e-01 -6.18015192e-02 -8.81149411e-01
-2.59870410e-01 -3.79741639e-01 -6.94460988e-01 -1.84785411e-01
-7.81568050e-01 -7.01492667e-01 -1.16398007e-01 -1.74361491e+00
4.50114280e-01 5.03319919e-01 -1.53425828e-01 9.21581924e-01
2.14014336e-01 4.16549683e-01 -3.39376897e-01 -1.77835569e-01
-4.92138833e-01 2.37455685e-02 8.59367251e-01 1.13900378e-01
-2.93926358e-01 2.32949182e-01 -5.67070127e-01 1.08892310e+00
9.97311592e-01 -9.15530145e-01 -3.28795403e-01 -7.12386727e-01
3.28255177e-01 2.92738587e-01 -1.36173189e-01 -8.87850106e-01
2.22315684e-01 -1.92896634e-01 3.65973413e-01 -6.68243766e-01
2.12338537e-01 -8.93660247e-01 2.27019608e-01 1.07000101e+00
-5.46519160e-02 3.17015946e-01 4.70329672e-01 5.40903956e-03
3.27583849e-01 -5.51524699e-01 1.02332926e+00 -7.16414899e-02
-1.03200448e-03 3.25384319e-01 -9.34623659e-01 -1.04525961e-01
8.51400435e-01 -2.68771201e-01 -1.79430097e-01 -3.92138541e-01
-1.02623260e+00 1.52502924e-01 3.03654581e-01 2.03937352e-01
8.10337603e-01 -1.28396130e+00 -6.76393211e-01 1.97642744e-01
-7.33113438e-02 -1.56084999e-01 1.11131571e-01 1.77671540e+00
-1.19336987e+00 2.21411332e-01 -1.37246072e-01 -6.13423645e-01
-1.71627676e+00 9.42095965e-02 2.31592566e-01 3.17525536e-01
-9.95275021e-01 8.58302712e-01 -2.17766717e-01 6.07610717e-02
5.95559895e-01 -7.99081624e-01 -4.84450519e-01 -3.54654565e-02
6.56229913e-01 7.22583950e-01 5.59530914e-01 -2.52865463e-01
-5.37620485e-01 6.04046524e-01 9.04577523e-02 -3.03140115e-02
1.66471457e+00 -8.56262352e-03 -1.74262851e-01 2.09425330e-01
1.20462704e+00 -1.48021400e-01 -9.42289531e-01 3.16171229e-01
1.25507750e-02 4.17986363e-02 -4.37121511e-01 -5.04068792e-01
-1.01943088e+00 8.47585201e-01 1.34639525e+00 1.88853234e-01
9.61570024e-01 5.93034215e-02 6.36211216e-01 8.17656457e-01
1.60934910e-01 -9.49221075e-01 -6.89553767e-02 2.98355281e-01
1.03155899e+00 -9.67504203e-01 -2.05704242e-01 1.71107084e-01
-4.67855066e-01 1.46310687e+00 2.32692555e-01 -4.34824079e-02
3.73050153e-01 5.91976941e-01 4.79005188e-01 -4.41535294e-01
-6.60140872e-01 3.29456091e-01 1.59425363e-01 4.81095880e-01
6.50406957e-01 -1.54087707e-01 -6.83721781e-01 7.90528059e-01
-1.28639027e-01 3.72514635e-01 7.82433689e-01 1.48875844e+00
-5.29057324e-01 -6.76195383e-01 -2.06287056e-01 7.01091290e-01
-1.22387969e+00 -1.25634283e-01 -2.26965576e-01 7.46286809e-01
2.52833009e-01 8.39134037e-01 -4.25712466e-01 -1.00464463e-01
3.66438657e-01 2.62710094e-01 1.04201996e+00 -7.51526773e-01
-5.02964199e-01 3.76684487e-01 -1.35234166e-02 -7.28642583e-01
-4.21490163e-01 -9.41742241e-01 -1.52104628e+00 3.18290323e-01
-3.08031172e-01 -1.89951360e-01 6.05100811e-01 1.10440755e+00
-9.92013514e-02 3.29918206e-01 2.70562083e-01 -6.36278629e-01
-5.41284740e-01 -1.34060884e+00 -6.57777786e-01 7.69498572e-02
1.03709769e+00 -5.54188311e-01 -3.81901562e-01 -9.64572188e-03] | [14.029666900634766, -2.2706637382507324] |
e66120c6-0be5-4297-99e0-93704c7c5019 | performer-a-novel-ppg-to-ecg-reconstruction | 2204.11795 | null | https://arxiv.org/abs/2204.11795v3 | https://arxiv.org/pdf/2204.11795v3.pdf | Performer: A Novel PPG-to-ECG Reconstruction Transformer for a Digital Biomarker of Cardiovascular Disease Detection | Electrocardiography (ECG), an electrical measurement which captures cardiac activities, is the gold standard for diagnosing cardiovascular disease (CVD). However, ECG is infeasible for continuous cardiac monitoring due to its requirement for user participation. By contrast, photoplethysmography (PPG) provides easy-to-collect data, but its limited accuracy constrains its clinical usage. To combine the advantages of both signals, recent studies incorporate various deep learning techniques for the reconstruction of PPG signals to ECG; however, the lack of contextual information as well as the limited abilities to denoise biomedical signals ultimately constrain model performance. In this research, we propose Performer, a novel Transformer-based architecture that reconstructs ECG from PPG and combines the PPG and reconstructed ECG as multiple modalities for CVD detection. This method is the first time that Transformer sequence-to-sequence translation has been performed on biomedical waveform reconstruction, combining the advantages of both PPG and ECG. We also create Shifted Patch-based Attention (SPA), an effective method to encode/decode the biomedical waveforms. Through fetching the various sequence lengths and capturing cross-patch connections, SPA maximizes the signal processing for both local features and global contextual representations. The proposed architecture generates a state-of-the-art performance of 0.29 RMSE for the reconstruction of PPG to ECG on the BIDMC database, surpassing prior studies. We also evaluated this model on the MIMIC-III dataset, achieving a 95.9% accuracy in CVD detection, and on the PPG-BP dataset, achieving 75.9% accuracy in related CVD diabetes detection, indicating its generalizability. As a proof of concept, an earring wearable named PEARL (prototype), was designed to scale up the point-of-care (POC) healthcare system. | ['Ella Lan'] | 2022-04-25 | null | null | null | null | ['photoplethysmography-ppg', 'electrocardiography-ecg'] | ['medical', 'methodology'] | [ 4.65043694e-01 -2.71832049e-01 1.16421878e-01 -2.70039141e-01
-9.16303098e-01 -3.34575087e-01 -1.11271136e-01 -1.06903493e-01
1.13554284e-01 6.72966838e-01 1.57240167e-01 -3.74652892e-01
-9.09579843e-02 -5.26283145e-01 -5.36607325e-01 -6.54311895e-01
-3.44352335e-01 -1.37098599e-02 -6.22381032e-01 2.28271961e-01
-1.45232186e-01 3.69900465e-01 -1.01536429e+00 4.79771972e-01
9.44646478e-01 1.36333346e+00 -3.97577696e-02 8.04386199e-01
4.25327271e-01 2.12468550e-01 -7.42936194e-01 -3.36321294e-01
1.84732094e-01 -9.39555228e-01 -1.66875631e-01 -3.01968217e-01
3.44178736e-01 -5.17198980e-01 -3.02874088e-01 6.14743471e-01
1.37340784e+00 -5.35061121e-01 4.16373461e-01 -8.40541601e-01
-9.13574278e-01 5.48724353e-01 -3.63164037e-01 3.99702549e-01
4.06063259e-01 3.64471674e-01 5.87697864e-01 -7.60692954e-01
2.62880921e-01 5.95772564e-01 1.27149594e+00 5.35181046e-01
-1.19744790e+00 -6.15013123e-01 -5.11851907e-01 1.93821788e-01
-1.40594172e+00 -3.84605706e-01 1.03836262e+00 -2.72427410e-01
9.80482817e-01 5.41947424e-01 1.14607096e+00 1.54760826e+00
7.29514897e-01 2.76256949e-01 1.17095947e+00 -2.08711848e-01
1.61944762e-01 -1.85534775e-01 -1.52360573e-01 1.71920300e-01
2.48804033e-01 4.17083621e-01 -5.26403129e-01 -2.69953310e-01
9.46014881e-01 1.89310968e-01 -5.72828054e-01 2.55612612e-01
-1.34779000e+00 2.72535980e-01 2.30699375e-01 2.62747258e-01
-8.03694308e-01 2.40513191e-01 5.36695182e-01 3.32183093e-01
1.87118113e-01 5.17969489e-01 -3.75842154e-01 -3.35690469e-01
-9.77742970e-01 -4.61041331e-02 7.37257659e-01 5.64919472e-01
-2.62874633e-01 4.52987492e-01 -5.66506803e-01 6.86823606e-01
2.40958750e-01 7.46786416e-01 6.28493369e-01 -9.45773840e-01
3.05817544e-01 3.72485906e-01 -2.22005900e-02 -1.08173120e+00
-6.23813510e-01 -9.09055352e-01 -1.48631823e+00 -3.61454964e-01
2.19186097e-01 -3.75285804e-01 -7.38316953e-01 1.68972981e+00
8.19717422e-02 6.37151599e-01 -6.57272637e-02 1.14256728e+00
9.93834615e-01 3.69683892e-01 1.63674682e-01 -5.43897092e-01
1.63116074e+00 -1.97445124e-01 -8.49309564e-01 -5.23574948e-02
1.89036310e-01 -5.40019095e-01 8.42740417e-01 5.37150204e-01
-1.04937577e+00 -8.22259068e-01 -1.19442761e+00 1.29797921e-01
2.63593853e-01 2.79438317e-01 3.57016265e-01 1.00110698e+00
-1.00277925e+00 7.18289733e-01 -8.69743943e-01 -3.17080706e-01
4.44083840e-01 1.18838310e-01 -1.32634327e-01 8.10840502e-02
-1.54574227e+00 7.38297164e-01 -1.71978816e-01 4.42410886e-01
-7.34605134e-01 -1.11392176e+00 -6.10204339e-01 1.33815527e-01
-3.12053710e-01 -1.05927181e+00 6.38642788e-01 -5.15687108e-01
-1.64142585e+00 7.09398568e-01 -1.11179613e-02 -5.78748345e-01
4.67050493e-01 -3.15285951e-01 -8.66149545e-01 3.59583765e-01
-1.51302233e-01 2.21041843e-01 8.88926566e-01 -5.81337631e-01
1.01305008e-01 -4.47750330e-01 -4.66985971e-01 -1.46783248e-01
-1.24787740e-01 -1.82961687e-01 -5.18717356e-02 -8.67259860e-01
1.90186724e-01 -7.60764658e-01 -1.29725397e-01 -1.71574757e-01
-2.33569905e-01 3.73214722e-01 1.10072650e-01 -1.21361196e+00
1.46525395e+00 -2.15972424e+00 -9.85227525e-02 2.54464984e-01
4.14333045e-01 4.19411033e-01 -1.13177091e-01 3.79704744e-01
-3.53891402e-01 1.39450192e-01 -4.17266041e-01 -1.32416943e-02
-1.81993008e-01 -5.91153726e-02 -1.44139171e-01 5.78435302e-01
3.73756826e-01 1.34756792e+00 -6.42421782e-01 -1.61233112e-01
2.86147445e-01 7.74637401e-01 -3.77343565e-01 1.74802303e-01
5.12082517e-01 7.79937565e-01 -2.01935321e-01 5.94459891e-01
7.13668227e-01 -4.08403516e-01 5.00492930e-01 -5.85290372e-01
2.53998101e-01 2.17630729e-01 -8.65705371e-01 1.90079343e+00
-2.76031733e-01 3.04283023e-01 -1.34591520e-01 -1.17637813e+00
1.25845230e+00 9.46371555e-01 9.05990243e-01 -9.89821553e-01
1.71981603e-01 2.55494267e-01 2.94880509e-01 -8.61707807e-01
-4.63174641e-01 -2.93999463e-01 2.04046872e-02 3.37478101e-01
-1.52148902e-01 2.04129398e-01 -1.87425673e-01 -1.72375903e-01
1.25294328e+00 2.09021345e-01 4.47176009e-01 -1.92071915e-01
3.56792629e-01 -5.40797949e-01 7.33833551e-01 8.06742549e-01
-3.54672372e-01 1.10983634e+00 3.33193064e-01 -6.06095135e-01
-8.21107388e-01 -1.07175136e+00 -4.20783401e-01 1.03547893e-01
-1.96951464e-01 -4.26258385e-01 -6.69716954e-01 -3.56558084e-01
6.83581904e-02 4.12355244e-01 -4.63413328e-01 -3.67486507e-01
-7.03052402e-01 -1.07610846e+00 1.16741741e+00 7.41655111e-01
4.46555406e-01 -9.52108681e-01 -9.65432703e-01 7.33532846e-01
-6.20457888e-01 -9.88829136e-01 -4.62398648e-01 -1.27582848e-01
-1.08727562e+00 -1.01754487e+00 -1.01984847e+00 -3.75149071e-01
9.49429199e-02 -3.18898141e-01 1.17614603e+00 -1.46769971e-01
-6.83789372e-01 2.78254896e-01 -2.38949493e-01 -3.33844543e-01
-3.63223791e-01 -1.62609920e-01 7.96952546e-02 2.26628810e-01
3.11842829e-01 -9.55040038e-01 -1.15170205e+00 8.78967270e-02
-4.97700840e-01 -8.09750110e-02 8.13110411e-01 9.13159549e-01
6.37392282e-01 -5.44128835e-01 1.21190429e+00 -5.63337505e-01
8.38500500e-01 -4.35032010e-01 5.91061730e-03 1.07277483e-01
-7.50927448e-01 -4.82246488e-01 6.37266457e-01 -4.59060699e-01
-6.48741007e-01 -1.39277056e-01 -5.49323201e-01 -4.95156676e-01
-1.52887389e-01 6.95446968e-01 -9.62276459e-02 2.55038381e-01
7.73659229e-01 4.40537006e-01 2.31701449e-01 -4.80341733e-01
-4.70431410e-02 7.21126318e-01 7.38239765e-01 -5.81836462e-01
1.80490151e-01 3.97502892e-02 1.72272414e-01 -7.29164183e-01
-3.06515545e-01 -1.27051800e-01 -3.50346535e-01 -9.11236554e-02
7.90184081e-01 -9.99638319e-01 -8.30531538e-01 4.66606826e-01
-1.09900200e+00 6.73895180e-02 -2.15171859e-01 5.82078278e-01
-3.64364922e-01 5.99014401e-01 -8.12842548e-01 -8.55365038e-01
-1.05470490e+00 -7.66614139e-01 8.55854928e-01 -1.81789175e-01
-5.80741525e-01 -5.60706019e-01 -1.69890337e-02 1.75879046e-01
7.06097960e-01 8.44667196e-01 7.74612844e-01 -2.55803138e-01
-1.37784958e-01 -3.70002300e-01 -1.50162682e-01 6.63008451e-01
2.58542329e-01 -4.90738750e-01 -1.01739335e+00 -3.77504379e-01
4.15996671e-01 7.62688741e-02 3.95217061e-01 6.99559808e-01
1.09159529e+00 -2.23166168e-01 -1.13128155e-01 8.59321654e-01
1.16269898e+00 4.52502072e-01 1.08147669e+00 -2.02737302e-01
5.67332745e-01 2.81556603e-02 -1.01082576e-02 5.34742594e-01
2.47420818e-01 5.15192032e-01 -2.72552706e-02 -4.11482304e-01
-4.18449223e-01 -3.00587583e-02 2.94583887e-01 1.05441332e+00
-3.45501900e-01 1.51231438e-01 -8.88187766e-01 3.14478606e-01
-1.46936071e+00 -9.17795777e-01 -2.72138357e-01 2.18701363e+00
1.00009775e+00 -2.36681551e-01 -4.04226705e-02 3.37789178e-01
5.34027576e-01 -2.77965933e-01 -7.25127339e-01 -3.40788275e-01
-1.19188979e-01 6.39238775e-01 8.63438398e-02 -5.37863187e-02
-8.24086845e-01 -1.19575858e-01 6.15949821e+00 2.59535193e-01
-1.55842710e+00 2.28606910e-01 7.83760965e-01 -1.87883735e-01
5.89949265e-02 -5.21386802e-01 -1.57603487e-01 8.29689384e-01
1.39940560e+00 6.57181889e-02 3.46808940e-01 3.32438797e-01
4.85577643e-01 4.54196632e-01 -1.29412174e+00 1.47145402e+00
6.45693168e-02 -1.12325442e+00 -1.62431061e-01 -1.47284478e-01
2.91933417e-01 -2.07499385e-01 -6.51152804e-02 1.81565464e-01
-8.01982164e-01 -1.13679373e+00 5.10796845e-01 8.71198475e-01
1.44790983e+00 -4.22380686e-01 9.03170407e-01 9.39393491e-02
-1.03439999e+00 -1.80287864e-02 -4.05700356e-02 -1.07062228e-01
1.70692563e-01 9.20556724e-01 -6.89428449e-01 8.40851307e-01
6.74682140e-01 7.50663102e-01 -2.86145896e-01 1.08195448e+00
1.31501788e-02 1.01615202e+00 -3.09317201e-01 3.60515386e-01
-4.74163473e-01 -1.15538962e-01 6.33350015e-01 1.26377881e+00
8.23720217e-01 2.93802708e-01 5.10299131e-02 1.15632868e+00
8.37472379e-02 -9.02264267e-02 -4.87879902e-01 -2.35708039e-02
4.73764449e-01 1.00598919e+00 -3.32792550e-01 -3.53721380e-01
-2.83278674e-01 8.06359589e-01 -2.49524847e-01 5.29928505e-01
-1.11969936e+00 -6.50138497e-01 4.17277902e-01 3.06678057e-01
-8.59459490e-02 1.20463014e-01 -8.79881740e-01 -1.06025255e+00
3.46532017e-01 -1.24154735e+00 3.23911995e-01 -5.95243335e-01
-1.39883339e+00 6.75822675e-01 -3.52741957e-01 -1.43557775e+00
-3.05037141e-01 -1.96589291e-01 -4.52710986e-01 1.29123020e+00
-1.23189712e+00 -9.97866213e-01 -5.08654177e-01 3.89748931e-01
6.13223994e-03 1.34597406e-01 1.23286808e+00 8.51744592e-01
-5.21426797e-01 7.48269558e-01 -3.27700049e-01 1.15763150e-01
7.39621818e-01 -9.85553682e-01 4.39223230e-01 7.99164832e-01
-2.10899204e-01 9.84762013e-01 3.93457383e-01 -5.88966727e-01
-1.54063964e+00 -1.09615660e+00 8.53426576e-01 -3.82135063e-01
-1.88569911e-02 -1.06095806e-01 -8.94176602e-01 2.43888244e-01
4.02093753e-02 3.35328653e-02 9.04825389e-01 -2.53350109e-01
-1.19250767e-01 -5.51444530e-01 -1.18101823e+00 3.31179202e-01
9.85806882e-01 -4.62974727e-01 -6.79694712e-01 -2.46220648e-01
4.12037522e-01 -5.17126560e-01 -1.41355073e+00 6.89612508e-01
1.12077200e+00 -7.93888211e-01 1.25038922e+00 -2.82916784e-01
5.50015569e-01 -2.71749645e-01 -3.16044949e-02 -1.19501972e+00
-5.52404940e-01 -8.19611490e-01 -3.76022935e-01 1.22179425e+00
2.15160668e-01 -9.59897578e-01 4.14371610e-01 6.15883231e-01
-3.24619979e-01 -8.99930120e-01 -9.18807745e-01 -4.67551589e-01
-2.13332161e-01 -5.36584616e-01 7.42598832e-01 9.31072891e-01
9.13930386e-02 2.88661778e-01 -7.58511901e-01 6.81907237e-02
6.47657216e-01 9.25893188e-02 1.95682213e-01 -1.20158815e+00
-5.40760815e-01 -2.31804565e-01 -3.14307868e-01 -6.26530647e-01
-5.43174088e-01 -1.07746267e+00 -2.70079762e-01 -1.47591925e+00
1.99881429e-03 -3.97389770e-01 -8.29308569e-01 3.34243178e-01
-2.82867312e-01 6.56809568e-01 1.11717947e-01 6.73060045e-02
7.42307007e-02 2.89242178e-01 1.32748294e+00 -1.42383680e-01
-4.17227805e-01 -2.02137291e-01 -8.94879162e-01 2.54927635e-01
8.25338244e-01 -3.25784862e-01 -3.43865067e-01 -1.23281203e-01
4.86236475e-02 6.44102097e-01 5.75371623e-01 -1.20232081e+00
-9.93388668e-02 4.11188394e-01 1.02209938e+00 -3.18937421e-01
2.98950702e-01 -6.23142719e-01 8.18511307e-01 7.61431515e-01
-1.83191285e-01 8.22341293e-02 2.44156897e-01 3.40867341e-01
-8.95056576e-02 4.55125034e-01 6.43882632e-01 -9.06193927e-02
-3.83453742e-02 2.92464733e-01 -3.76733392e-01 1.84271634e-01
5.89194894e-01 -2.66748130e-01 -8.19222778e-02 -1.60667986e-01
-7.55050063e-01 -1.81706443e-01 -2.00320646e-01 2.51560926e-01
1.02380061e+00 -1.45595169e+00 -1.10734737e+00 6.29627883e-01
-9.54497233e-02 -3.93737614e-01 6.24599993e-01 1.51668680e+00
-5.34077644e-01 3.32899958e-01 -4.87295181e-01 -8.59775901e-01
-9.43913579e-01 3.89643043e-01 4.82913524e-01 -1.38266057e-01
-1.18970871e+00 4.10664350e-01 -4.37447652e-02 -3.91368121e-02
3.28439921e-02 -5.77946842e-01 -8.36312845e-02 -5.56487814e-02
6.90836489e-01 4.77589250e-01 4.44276065e-01 -3.63804698e-02
-5.62725425e-01 5.95241904e-01 4.59065497e-01 1.75019607e-01
1.14878130e+00 -1.72419563e-01 -1.21371016e-01 2.53979832e-01
1.07376766e+00 -3.69985133e-01 -6.91099226e-01 1.45405844e-01
-3.62618864e-01 -2.59016544e-01 5.94572909e-02 -1.21494305e+00
-1.20808899e+00 1.17415750e+00 1.09407413e+00 5.57811856e-02
1.45513225e+00 -7.05259860e-01 1.10395682e+00 -8.80568475e-02
4.27535802e-01 -5.75257123e-01 -1.31671309e-01 -1.12490080e-01
1.06450605e+00 -6.13173187e-01 -1.86146662e-01 -1.45470604e-01
-5.63880980e-01 1.11054277e+00 9.64851826e-02 -5.36339842e-02
7.14064181e-01 3.41964036e-01 3.64043027e-01 -6.99980184e-02
-5.12601316e-01 3.98709655e-01 3.71530473e-01 8.35116565e-01
7.00029552e-01 2.49965817e-01 -5.80156505e-01 1.17894077e+00
-7.43095875e-02 6.57702744e-01 3.35967273e-01 5.90370297e-01
2.88913429e-01 -9.21624660e-01 -3.32102627e-01 7.52049208e-01
-9.06692028e-01 -3.61205459e-01 -4.57176799e-03 3.16314638e-01
1.85854405e-01 1.07287633e+00 -2.96858221e-01 -2.73747444e-01
4.15274054e-01 4.11528856e-01 5.62974215e-01 -2.41133720e-01
-7.62108445e-01 3.29724073e-01 8.64360258e-02 -6.16545975e-01
-1.60938442e-01 -5.10890067e-01 -9.00941968e-01 -4.65778112e-02
1.49668023e-01 -2.02583939e-01 4.49990481e-01 5.79120278e-01
9.84923303e-01 1.12958765e+00 4.19232309e-01 -5.94533563e-01
-7.54753590e-01 -1.12757182e+00 -7.45298028e-01 5.63884199e-01
3.83150697e-01 -2.68105358e-01 -8.25362504e-02 2.33618855e-01] | [14.239614486694336, 3.210813283920288] |
a0063075-0def-4483-bd47-965a77ff2cee | self-supervised-learning-in-remote-sensing-a | 2206.13188 | null | https://arxiv.org/abs/2206.13188v2 | https://arxiv.org/pdf/2206.13188v2.pdf | Self-supervised Learning in Remote Sensing: A Review | In deep learning research, self-supervised learning (SSL) has received great attention triggering interest within both the computer vision and remote sensing communities. While there has been a big success in computer vision, most of the potential of SSL in the domain of earth observation remains locked. In this paper, we provide an introduction to, and a review of the concepts and latest developments in SSL for computer vision in the context of remote sensing. Further, we provide a preliminary benchmark of modern SSL algorithms on popular remote sensing datasets, verifying the potential of SSL in remote sensing and providing an extended study on data augmentations. Finally, we identify a list of promising directions of future research in SSL for earth observation (SSL4EO) to pave the way for fruitful interaction of both domains. | ['Xiao Xiang Zhu', 'Lichao Mou', 'Nassim Ait Ali Braham', 'Conrad M Albrecht', 'Yi Wang'] | 2022-06-27 | null | null | null | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 3.77594978e-01 -5.03872894e-02 -9.69129279e-02 -6.12954617e-01
-3.47283423e-01 -4.49769497e-01 7.67319202e-01 1.28030330e-01
-5.54750979e-01 5.30335128e-01 -1.04868328e-02 -6.62273109e-01
-3.57427120e-01 -1.03359401e+00 -5.01298964e-01 -9.41384494e-01
-5.56042671e-01 -5.84362708e-02 -1.78378865e-01 -4.12043631e-01
-2.46083662e-01 9.14611518e-01 -1.47306705e+00 -1.79327801e-01
7.96604633e-01 7.35492289e-01 2.80938536e-01 4.95172262e-01
4.90763374e-02 5.53157628e-01 -8.18435848e-02 1.68015346e-01
5.22533000e-01 -2.91920930e-01 -9.14917529e-01 2.60547906e-01
7.99543440e-01 -7.34281912e-02 -4.92597222e-01 1.13667548e+00
6.21095598e-01 3.29988122e-01 4.00333971e-01 -1.10030961e+00
-5.40168941e-01 5.31174421e-01 -5.08035302e-01 5.31464994e-01
-4.99586523e-01 1.40155718e-01 1.10713494e+00 -1.00263762e+00
2.89375007e-01 1.01129818e+00 9.03157115e-01 1.32144764e-01
-1.05333543e+00 -3.13255161e-01 2.53702611e-01 1.67391241e-01
-1.08333480e+00 -4.37055349e-01 6.45207942e-01 -6.37958229e-01
8.99118006e-01 2.06309944e-01 8.47707808e-01 5.73745549e-01
-1.31048247e-01 9.00816917e-01 1.39448774e+00 -6.98719740e-01
2.51544952e-01 2.66489573e-02 3.87605488e-01 5.09249926e-01
1.12136014e-01 4.13764477e-01 -1.16544075e-01 1.05897598e-01
1.01280832e+00 4.94963303e-02 -1.51556805e-01 -1.90251350e-01
-1.25276959e+00 1.13670027e+00 8.13452184e-01 3.61975819e-01
-4.01643932e-01 3.18678729e-02 1.81798637e-01 3.67878437e-01
8.06270599e-01 6.20569646e-01 -5.66055775e-01 4.43451434e-01
-1.11311209e+00 9.20778066e-02 4.11332518e-01 2.65197545e-01
9.19921219e-01 6.01161182e-01 3.56931865e-01 1.03051639e+00
3.45193267e-01 8.53912473e-01 1.58809796e-01 -7.06650496e-01
-1.63142271e-02 3.90558183e-01 -9.38242599e-02 -1.02429438e+00
-6.91619396e-01 -8.58742058e-01 -1.29828584e+00 5.11098266e-01
-7.36866891e-02 -2.41234705e-01 -7.07329512e-01 1.42407465e+00
6.06054738e-02 3.56228769e-01 2.55988210e-01 7.40600348e-01
9.97177303e-01 8.68821383e-01 -1.61690474e-01 1.69283867e-01
8.14779580e-01 -9.94770885e-01 -4.06318814e-01 -7.46380866e-01
4.60128784e-01 -2.32452229e-01 8.84897351e-01 4.57057543e-02
-2.42734879e-01 -7.16991067e-01 -8.79285276e-01 1.82790250e-01
-4.85132903e-01 4.02072430e-01 1.14266801e+00 3.94115597e-01
-1.14329123e+00 5.05886078e-01 -9.88421261e-01 -7.57821739e-01
4.92458910e-01 4.03513536e-02 -2.62334675e-01 1.06697783e-01
-1.38964224e+00 1.01132667e+00 5.00952005e-01 5.11078417e-01
-8.88069510e-01 -6.01756513e-01 -9.51523542e-01 -1.08795002e-01
1.03820033e-01 -3.65453869e-01 9.75530386e-01 -8.76712680e-01
-8.70034933e-01 1.21781397e+00 2.10397691e-01 -7.98940539e-01
3.24356049e-01 -3.65680784e-01 -5.23435712e-01 -1.21491760e-01
1.48560390e-01 8.45099926e-01 5.22789478e-01 -1.06336391e+00
-7.15048611e-01 -4.58881944e-01 3.20542194e-02 1.76190048e-01
-1.74356669e-01 -1.58844963e-01 7.57218897e-02 -6.91001117e-01
3.69505942e-01 -9.52274978e-01 -6.84490979e-01 2.62629658e-01
-1.22642536e-02 -2.13709801e-01 7.61996508e-01 -4.27690268e-01
8.60246599e-01 -2.21979642e+00 -8.31575766e-02 7.36203343e-02
1.23745456e-01 5.96198678e-01 -3.76029283e-01 5.84109902e-01
-4.97477472e-01 6.39764369e-02 -6.38495266e-01 3.17888446e-02
-2.54943997e-01 3.52258623e-01 -7.00957775e-01 8.60122561e-01
2.80751467e-01 1.02206278e+00 -1.10948336e+00 4.87387320e-03
6.13380253e-01 3.74363422e-01 -6.61562830e-02 -4.69842330e-02
-1.50018141e-01 6.15642428e-01 -3.66132408e-01 7.00227439e-01
7.27967024e-01 -2.57745564e-01 -1.73891857e-01 8.37463140e-02
-5.03749788e-01 9.81016830e-02 -1.03763330e+00 1.27788389e+00
-4.50376034e-01 1.27168250e+00 1.22162730e-01 -1.28348875e+00
1.05421531e+00 9.25437063e-02 4.48378682e-01 -4.22256619e-01
-2.42992371e-01 1.54854074e-01 -7.82755092e-02 -6.13303781e-01
3.22529703e-01 -2.86729455e-01 2.91631103e-01 4.05863494e-01
-1.63340777e-01 -2.00067014e-01 -9.36145782e-02 -4.35782075e-02
5.29124558e-01 8.37730914e-02 6.25966012e-01 -3.79831016e-01
4.69272345e-01 2.59152353e-01 8.46292228e-02 1.13615549e+00
-2.24795878e-01 1.46394432e-01 -2.76690781e-01 -9.51017916e-01
-9.60397780e-01 -6.99354708e-01 -6.41680777e-01 1.25060332e+00
-3.03357750e-01 2.07765400e-01 -2.44043395e-01 -4.48205054e-01
6.42200187e-02 2.95961440e-01 -6.40844047e-01 1.83666706e-01
-2.22132280e-01 -1.50233221e+00 7.87075341e-01 5.32935083e-01
1.07501662e+00 -1.21228242e+00 -5.44250488e-01 1.71861351e-01
-4.96425293e-02 -1.07898772e+00 5.15604556e-01 2.34832600e-01
-1.36329043e+00 -7.98931420e-01 -5.68462372e-01 -7.66411901e-01
2.69466937e-01 7.85082579e-01 1.20831501e+00 1.16670914e-01
-2.33213827e-01 3.63174736e-01 -4.46895748e-01 -6.77087486e-01
-3.13291013e-01 2.44893566e-01 8.22359771e-02 -6.08996898e-02
4.02340919e-01 -6.70580387e-01 -3.84770095e-01 4.59387107e-03
-9.56974745e-01 -1.42671347e-01 6.05554879e-01 7.94096887e-01
1.65941060e-01 1.92967892e-01 4.87122595e-01 -7.12057650e-01
1.03220686e-01 -5.91338813e-01 -7.22016752e-01 1.48560870e-02
-6.32968128e-01 -2.97906816e-01 1.40551269e-01 1.94754422e-01
-8.45592976e-01 1.25273883e-01 -4.31479573e-01 1.15268506e-01
-5.18968523e-01 1.19306946e+00 1.58646554e-01 -3.26488107e-01
1.08414984e+00 4.70788628e-01 -1.53825385e-02 -6.04356229e-01
4.02164936e-01 7.88231850e-01 3.95140201e-01 -1.47170946e-01
8.48438382e-01 8.86458218e-01 1.11959562e-01 -1.58277392e+00
-1.46659863e+00 -7.02909410e-01 -7.39304960e-01 -8.66368935e-02
6.55455530e-01 -1.17630279e+00 -8.57710913e-02 8.95191908e-01
-6.39551699e-01 -4.86900330e-01 -3.88058782e-01 6.73383117e-01
-1.49171531e-01 3.52716297e-01 -2.40817472e-01 -9.30075884e-01
-2.32005090e-01 -7.14750528e-01 9.17096853e-01 2.08742633e-01
2.88395584e-01 -1.44652748e+00 4.17640209e-01 2.42316693e-01
5.43734193e-01 2.12973475e-01 5.43573618e-01 -4.76629108e-01
-4.38852727e-01 -1.25547722e-01 -3.81776690e-01 5.99746406e-01
2.41135970e-01 1.48261031e-02 -1.33784175e+00 -3.98003727e-01
-1.12986073e-01 -4.62545663e-01 1.56620562e+00 6.68284416e-01
7.75658190e-01 3.01691443e-02 -1.90884918e-01 9.37920809e-01
1.72674632e+00 -7.85336271e-02 5.23603261e-01 8.63058329e-01
7.09441423e-01 6.62003517e-01 4.35425162e-01 3.78641278e-01
2.46668696e-01 2.03138486e-01 7.10490644e-01 -8.96961331e-01
3.03275287e-02 1.09637775e-01 7.52969692e-03 4.88276780e-01
-3.69888246e-01 1.12768963e-01 -1.29812992e+00 8.41981947e-01
-1.78876686e+00 -1.17933357e+00 -4.79588866e-01 1.85326219e+00
4.44408417e-01 -4.45025146e-01 -1.93841085e-01 2.85956323e-01
5.60171664e-01 8.31791043e-01 -5.26341200e-01 2.15887442e-01
-6.36322141e-01 1.64479658e-01 6.41986907e-01 6.00265622e-01
-1.82434416e+00 1.28853011e+00 6.94097519e+00 4.57508445e-01
-1.50510454e+00 -6.90724477e-02 4.95776981e-01 4.62113112e-01
-8.61452520e-03 1.06738739e-01 -8.64226341e-01 -1.08619891e-01
4.61087286e-01 2.04742715e-01 1.96155176e-01 1.00534463e+00
6.04636252e-01 -2.83918768e-01 -7.09599733e-01 7.13698626e-01
-1.23454072e-01 -1.60843694e+00 -3.63343209e-02 3.08016445e-02
1.04750383e+00 1.07670522e+00 4.37274612e-02 2.42981330e-01
3.55560690e-01 -1.04356909e+00 3.23172092e-01 1.77077413e-01
5.65082848e-01 -3.90256196e-01 8.30418289e-01 4.76248026e-01
-1.05423450e+00 -1.65973660e-02 -6.60867929e-01 -4.88582253e-01
-2.67231792e-01 8.55017900e-01 -6.40743256e-01 4.84157085e-01
8.90964627e-01 1.27740455e+00 -7.02084064e-01 1.11999106e+00
-4.45990354e-01 1.03068590e+00 -3.95516276e-01 3.02756637e-01
7.72438705e-01 -4.51700449e-01 5.73832810e-01 1.42956984e+00
-5.63400798e-02 1.58022761e-01 3.93012851e-01 5.79416573e-01
2.30909973e-01 -5.90517819e-02 -8.17916572e-01 -2.41032526e-01
3.48314822e-01 1.20369148e+00 -6.18091464e-01 -2.24828511e-01
-4.48449582e-01 4.63717669e-01 8.39404613e-02 4.76214558e-01
-4.91229773e-01 -1.72096983e-01 6.69313312e-01 -6.92211613e-02
8.10459927e-02 -5.92662632e-01 -4.77433175e-01 -1.27284932e+00
-2.37047747e-01 -7.11302221e-01 2.94142783e-01 -8.39634299e-01
-1.26396120e+00 5.70228279e-01 -1.01655789e-01 -1.24385428e+00
1.94832399e-01 -5.81654549e-01 -5.94699383e-01 7.71832705e-01
-2.16836286e+00 -1.30672133e+00 -6.29369617e-01 1.62890241e-01
6.45814180e-01 -3.53396237e-01 8.73524725e-01 2.42613032e-02
-3.13220322e-01 -1.66771755e-01 5.89338601e-01 5.29452980e-01
3.89367998e-01 -1.04533029e+00 1.00186133e+00 9.47267652e-01
5.36420703e-01 2.69176394e-01 5.73986351e-01 -3.70247602e-01
-1.14077079e+00 -1.44368279e+00 8.42391968e-01 -3.47567424e-02
9.22627151e-01 1.57776594e-01 -1.05301225e+00 1.01819551e+00
5.11590578e-02 1.69930711e-01 6.66068375e-01 1.59695923e-01
-1.89261168e-01 -1.59805477e-01 -7.91012824e-01 3.65777999e-01
7.03545570e-01 -5.86719453e-01 -5.72580636e-01 5.69848716e-01
1.57628894e-01 -4.68686456e-03 -6.51403666e-01 4.85964239e-01
2.74577081e-01 -1.04204369e+00 1.04686284e+00 -6.37673616e-01
3.49483997e-01 -4.14967805e-01 -3.02253634e-01 -1.50380647e+00
-6.56744957e-01 -1.52171955e-01 4.39466327e-01 7.60793149e-01
2.48315603e-01 -6.26162529e-01 7.08653867e-01 -1.02486864e-01
-2.90826350e-01 -5.24438739e-01 -6.88722670e-01 -8.63275111e-01
3.23184282e-01 -7.12288976e-01 1.79563031e-01 1.48258340e+00
-4.31562334e-01 4.70641017e-01 -5.00724137e-01 5.55156589e-01
8.45428646e-01 3.19549471e-01 6.50521159e-01 -1.78991807e+00
-1.23246983e-01 -3.79496962e-01 -3.57781351e-01 -1.16892385e+00
2.07017943e-01 -1.15539360e+00 -2.30961498e-02 -1.71818173e+00
1.68140337e-01 -5.21767318e-01 -2.25944683e-01 8.19991887e-01
-8.50068852e-02 6.12162173e-01 4.77668121e-02 3.63176525e-01
-1.57981217e-01 5.79800904e-01 9.16216195e-01 -3.26710582e-01
-2.38196999e-01 1.81247920e-01 -3.87023330e-01 9.47820902e-01
1.00318003e+00 -3.04683417e-01 -1.78800449e-01 -7.70364404e-01
5.10970354e-01 -3.05985212e-01 7.02866554e-01 -9.64008808e-01
-1.59019046e-02 -3.83589476e-01 3.23821485e-01 -7.73898542e-01
-2.16026139e-02 -7.44550824e-01 -2.32658520e-01 3.42805296e-01
-1.83902085e-01 -3.37381929e-01 2.95429975e-01 4.34291482e-01
-4.13777888e-01 -1.75356820e-01 1.16986561e+00 -4.24189419e-01
-1.28147984e+00 4.34652835e-01 -6.53740466e-01 -1.78469509e-01
7.84287035e-01 -2.75738358e-01 -1.07509986e-01 -2.46879026e-01
-1.00348866e+00 3.56198668e-01 1.32135808e-01 2.61903286e-01
3.88031334e-01 -9.25781965e-01 -1.20829177e+00 4.58884031e-01
2.65281826e-01 -3.30230519e-02 2.56446987e-01 7.33142316e-01
-5.53662479e-01 5.49312174e-01 -1.99172601e-01 -7.29265332e-01
-9.68996167e-01 7.49932230e-02 6.95603907e-01 4.13380265e-02
-7.72674263e-01 9.69564855e-01 -1.05171390e-02 -6.99021220e-01
1.61619961e-01 -1.11998282e-01 -4.15356338e-01 1.20936602e-01
4.31413710e-01 4.55691278e-01 1.74175724e-01 -5.67931294e-01
-3.87218893e-01 5.30557096e-01 3.35856646e-01 -1.57190025e-01
1.94068944e+00 -2.04161391e-01 -3.26976031e-01 6.87908053e-01
7.81037569e-01 -3.14013124e-01 -1.20077753e+00 -6.33795857e-01
1.29114702e-01 -1.72143400e-01 6.35252833e-01 -6.57221258e-01
-1.13705480e+00 1.21629000e+00 6.67403996e-01 2.21915781e-01
1.04800463e+00 1.15213186e-01 3.40085924e-01 9.40151095e-01
2.15427399e-01 -7.71577895e-01 -2.21988350e-01 1.05458450e+00
8.88200700e-01 -1.79232025e+00 2.77797848e-01 1.33648226e-02
-3.94991845e-01 1.05120063e+00 1.78492606e-01 -2.65429258e-01
1.06304312e+00 1.08260855e-01 4.09040302e-01 -3.95985395e-01
-3.70298028e-01 -5.98320484e-01 1.74655586e-01 7.10599184e-01
6.34033561e-01 3.95645164e-02 5.79617545e-02 -1.51249424e-01
-4.15531434e-02 1.05294324e-02 4.35853124e-01 8.34984183e-01
-9.49979722e-01 -7.40463972e-01 -4.30801392e-01 3.64357650e-01
-2.31061757e-01 -5.47097743e-01 -3.35424244e-01 7.29556441e-01
1.30225360e-01 7.87004173e-01 2.97920942e-01 2.07195319e-02
3.99870984e-03 -5.58374636e-02 1.67086929e-01 -6.69631422e-01
-4.58344340e-01 4.68763858e-02 -2.96633281e-02 -8.64933878e-02
-9.70547736e-01 -6.93450809e-01 -8.37674975e-01 -1.87527239e-02
-2.64656335e-01 1.77293152e-01 8.09467852e-01 1.12059987e+00
4.83412072e-02 1.93064183e-01 8.03060412e-01 -9.83364403e-01
-4.67546463e-01 -1.14966452e+00 -9.53144908e-01 -2.75541395e-01
7.25785673e-01 -3.68188351e-01 -5.01542747e-01 1.54043540e-01] | [9.568696022033691, -1.4469492435455322] |
298c2963-5180-41a4-a1e4-15304f99d2e1 | crime-prediction-using-machine-learning-with | 2211.01551 | null | https://arxiv.org/abs/2211.01551v1 | https://arxiv.org/pdf/2211.01551v1.pdf | Crime Prediction using Machine Learning with a Novel Crime Dataset | Crime is an unlawful act that carries legal repercussions. Bangladesh has a high crime rate due to poverty, population growth, and many other socio-economic issues. For law enforcement agencies, understanding crime patterns is essential for preventing future criminal activity. For this purpose, these agencies need structured crime database. This paper introduces a novel crime dataset that contains temporal, geographic, weather, and demographic data about 6574 crime incidents of Bangladesh. We manually gather crime news articles of a seven year time span from a daily newspaper archive. We extract basic features from these raw text. Using these basic features, we then consult standard service-providers of geo-location and weather data in order to garner these information related to the collected crime incidents. Furthermore, we collect demographic information from Bangladesh National Census data. All these information are combined that results in a standard machine learning dataset. Together, 36 features are engineered for the crime prediction task. Five supervised machine learning classification algorithms are then evaluated on this newly built dataset and satisfactory results are achieved. We also conduct exploratory analysis on various aspects the dataset. This dataset is expected to serve as the foundation for crime incidence prediction systems for Bangladesh and other countries. The findings of this study will help law enforcement agencies to forecast and contain crime as well as to ensure optimal resource allocation for crime patrol and prevention. | ['Mohammad Shafiul Alam', 'Muhammad Ibrahim', 'Abu Ubaida Akash', 'Faisal Tareque Shohan'] | 2022-11-03 | null | null | null | null | ['crime-prediction'] | ['miscellaneous'] | [-7.91089684e-02 -5.68203866e-01 -1.82164565e-01 -5.82331419e-01
-6.68352187e-01 -3.81560922e-01 4.46374774e-01 9.07588601e-01
-7.26383150e-01 9.34478283e-01 9.16622341e-01 -4.62240934e-01
-1.93977222e-01 -1.42192411e+00 -1.11495636e-01 -6.79722309e-01
-4.06082310e-02 1.61510974e-01 -1.66037127e-01 -2.79534131e-01
8.47840965e-01 6.32820547e-01 -1.08349144e+00 2.59381562e-01
9.28649902e-01 3.39809150e-01 2.49448732e-01 4.45550591e-01
-3.50688659e-02 8.13150227e-01 -5.10328829e-01 -4.21478391e-01
6.64163977e-02 -6.07348047e-02 -8.18916917e-01 -2.87464172e-01
-3.54573786e-01 -5.64886689e-01 -2.16566756e-01 8.24779034e-01
5.24764776e-01 3.41628343e-01 7.43500531e-01 -7.50123084e-01
-7.51127481e-01 4.84710246e-01 -8.56773913e-01 7.86609352e-01
6.91574335e-01 -2.17910513e-01 5.30973971e-01 -7.35651314e-01
3.80974442e-01 9.25484300e-01 5.40933669e-01 1.38227150e-01
-6.42128587e-01 -7.78343797e-01 -4.68670070e-01 2.83386052e-01
-1.16677880e+00 -5.73951423e-01 7.84372389e-01 -7.98260391e-01
1.33371353e+00 1.05560474e-01 4.80279535e-01 7.09400833e-01
2.28209496e-01 2.35255539e-01 1.09866953e+00 -5.70236087e-01
4.47201617e-02 8.64063650e-02 5.02757847e-01 4.05409575e-01
5.41080058e-01 -7.74163082e-02 -3.49425465e-01 -1.79761335e-01
2.99921453e-01 5.34048975e-01 1.57917812e-01 7.78371036e-01
-5.15050232e-01 1.04569066e+00 1.67658225e-01 4.36288834e-01
-6.59928560e-01 -5.12043476e-01 3.77739519e-01 3.83498147e-02
7.97618866e-01 -4.69092466e-02 -3.67526472e-01 -5.91344655e-01
-7.68403709e-01 4.29876775e-01 4.19465184e-01 2.88331479e-01
7.73597121e-01 -6.48556277e-02 3.20284754e-01 1.11114204e+00
1.48771077e-01 8.33435893e-01 -3.07842251e-02 -4.31250453e-01
1.18587446e+00 7.59165823e-01 -7.13116676e-02 -1.74839270e+00
-5.47366679e-01 2.95796812e-01 -9.21826184e-01 -4.33101058e-01
3.80377829e-01 -3.82017702e-01 -7.15157390e-01 1.08069074e+00
3.38024974e-01 -6.15058467e-03 -2.03284360e-02 5.43095708e-01
7.21088529e-01 9.00927782e-01 3.73587012e-01 -3.81980419e-01
1.19561768e+00 5.00799939e-02 -7.79218674e-01 9.67275426e-02
7.77505696e-01 -7.43353188e-01 5.73882222e-01 2.26048887e-01
-7.30462313e-01 -1.50174797e-02 -4.19702381e-01 1.86895996e-01
-6.94325268e-01 -1.72164664e-02 8.30410540e-01 7.68834412e-01
-5.49166262e-01 2.90423125e-01 -8.53636622e-01 -6.09933257e-01
8.15837026e-01 1.02669246e-01 -5.50464869e-01 -3.22457612e-01
-1.04889679e+00 1.00485098e+00 4.58409548e-01 -9.45527479e-03
1.42553169e-02 -5.34089744e-01 -1.12855744e+00 -2.45991230e-01
-1.23456106e-01 2.45081782e-01 4.00064349e-01 8.01384225e-02
-4.28500026e-01 7.00156629e-01 -3.78243923e-01 -1.22261234e-01
-7.81449452e-02 6.27395511e-02 -7.98918784e-01 6.92612538e-03
9.17152047e-01 -2.91482985e-01 -3.74816544e-02 -6.85324967e-01
-1.21027899e+00 -9.49012816e-01 -1.83097735e-01 -2.56450078e-03
-6.60212100e-01 9.32052314e-01 -9.65185687e-02 -7.94461250e-01
1.01049073e-01 -5.03604352e-01 -1.58676714e-01 -1.13804936e+00
-2.19076201e-01 -2.47639045e-01 9.03134704e-01 -1.28574991e+00
1.84055436e+00 -1.92120230e+00 -5.26485384e-01 3.66559297e-01
-2.75379926e-01 2.99969196e-01 1.35333002e-01 7.72344947e-01
-1.47457421e-02 2.97315598e-01 -1.97753802e-01 9.65216290e-03
-5.45096934e-01 1.74246132e-01 -3.12993586e-01 5.97211599e-01
3.91913563e-01 2.74645418e-01 -9.11703348e-01 -5.24380445e-01
5.11909664e-01 2.99337357e-01 -4.74340022e-01 -1.25471801e-01
7.69215822e-01 4.55252200e-01 -8.03185463e-01 8.66772354e-01
7.57706821e-01 5.77970028e-01 -3.56675655e-01 4.54730392e-01
-4.55213666e-01 4.17253792e-01 -9.19919252e-01 8.32511783e-01
-2.16158316e-01 6.84297085e-01 -1.98823854e-01 -1.29546213e+00
1.11305058e+00 3.20102662e-01 7.66807139e-01 -8.67019594e-01
6.73988760e-02 6.36290386e-02 -3.46229225e-01 -9.97395635e-01
6.74212098e-01 1.37448087e-01 -2.89292395e-01 8.42769802e-01
-3.74047846e-01 1.25827447e-01 6.68003976e-01 1.70902878e-01
1.09260702e+00 -4.67146426e-01 3.56562912e-01 -1.73498482e-01
4.19232637e-01 3.03046405e-01 9.30966914e-01 4.57415938e-01
-1.44220084e-01 3.23563010e-01 4.49221075e-01 -1.09207439e+00
-1.12067795e+00 -5.38548827e-01 -4.61289406e-01 7.69200742e-01
-5.60865104e-01 -2.57172942e-01 -5.69050312e-01 -3.10229987e-01
-2.59950936e-01 6.72552586e-01 -6.12057030e-01 2.15536445e-01
-6.98846936e-01 -1.43420994e+00 4.28475380e-01 3.26218724e-01
8.51987362e-01 -1.15625632e+00 -6.74268007e-01 4.19363797e-01
-3.40151876e-01 -8.48058999e-01 3.77663150e-02 -1.86885595e-01
-6.27130330e-01 -1.26877189e+00 -2.35330790e-01 -5.51613092e-01
6.87858522e-01 4.54079449e-01 5.09453356e-01 2.88056642e-01
-4.53815967e-01 -1.53264642e-01 -7.85608590e-01 -9.25268352e-01
-1.22037478e-01 -3.75482365e-02 2.46091485e-01 -1.23721875e-01
1.14009440e+00 -6.87819064e-01 -2.53813446e-01 -3.19563806e-01
-8.79723728e-01 -1.11388125e-01 -1.30428717e-01 4.09884363e-01
-4.88639772e-02 8.16291690e-01 7.19259381e-01 -7.75962293e-01
8.90492320e-01 -1.06972742e+00 -4.47674811e-01 7.79184550e-02
-2.14314789e-01 -6.50150776e-01 3.51939112e-01 2.25220233e-01
-1.28892505e+00 -1.18600890e-01 -1.04623340e-01 6.61740124e-01
-5.82184970e-01 1.04212236e+00 1.00035191e-01 5.18567741e-01
5.79214692e-01 1.31682888e-01 -5.89728713e-01 -5.41802049e-01
-6.19116008e-01 1.21622133e+00 3.85858148e-01 -7.53579140e-01
6.74755633e-01 3.41107011e-01 -1.27703294e-01 -1.17652202e+00
-5.86131632e-01 -8.74614775e-01 -8.95913363e-01 -2.80133337e-01
9.76015925e-01 -6.19068265e-01 -4.34619486e-01 7.47602344e-01
-1.43567348e+00 2.87679672e-01 5.64402580e-01 7.48954594e-01
1.41369596e-01 5.85450716e-02 -6.37357831e-01 -1.38109589e+00
-2.42736682e-01 -6.73792899e-01 4.82561648e-01 3.55708569e-01
-1.56921104e-01 -1.20142841e+00 4.61887777e-01 6.65831149e-01
1.93787158e-01 7.17420995e-01 8.61345768e-01 -5.37043452e-01
2.24053308e-01 -3.34586710e-01 -3.55249405e-01 4.15408798e-02
6.22808039e-01 3.09444278e-01 -7.34434903e-01 -1.30258314e-03
-1.87198415e-01 -1.08364500e-01 9.50642586e-01 4.06748652e-01
8.09477031e-01 -6.25014365e-01 -1.63489163e-01 2.63903081e-01
1.59518659e+00 6.80702448e-01 5.96039057e-01 6.06835663e-01
6.09724224e-01 9.83403862e-01 5.72231412e-01 9.52531278e-01
6.93825543e-01 2.06210583e-01 -4.62398306e-02 -1.12729274e-01
7.04254031e-01 -9.76156965e-02 1.12858027e-01 6.74107671e-01
-8.66704166e-01 1.58111572e-01 -1.87407815e+00 1.01793754e+00
-1.64772427e+00 -1.64556777e+00 -4.34158415e-01 1.76627481e+00
7.23856151e-01 -2.14293808e-01 1.94585696e-01 7.50053227e-01
6.79235458e-01 -9.09928977e-03 1.60738260e-01 -7.33810127e-01
-2.16306243e-02 5.85970283e-01 5.00272751e-01 3.40820700e-01
-1.27288032e+00 9.91536021e-01 5.41853666e+00 6.17043853e-01
-8.99733782e-01 4.39833552e-02 6.98686361e-01 -6.03336208e-02
-1.51055986e-02 -7.66598284e-02 -8.12664330e-01 6.10202491e-01
8.29832077e-01 -3.40164363e-01 3.87861192e-01 5.37174165e-01
1.07138383e+00 -6.83438003e-01 -3.00707221e-01 8.57299268e-01
-1.19150475e-01 -1.47115111e+00 -1.56163380e-01 4.12958443e-01
7.91473746e-01 -1.39976636e-01 -1.36720866e-01 1.80800170e-01
5.29609025e-01 -1.03288043e+00 1.51818633e-01 4.10141259e-01
5.99807084e-01 -1.29486668e+00 9.32897210e-01 7.12198555e-01
-1.05410743e+00 -3.82392198e-01 -4.63401526e-01 -8.33538949e-01
3.25388163e-01 8.87892067e-01 -7.56993294e-01 3.76544923e-01
9.82269406e-01 9.54774737e-01 -3.75246048e-01 7.41485178e-01
-9.70443413e-02 7.59284735e-01 -1.55967534e-01 1.12346128e-01
4.05272543e-01 -4.23823267e-01 5.02869822e-02 1.32925189e+00
5.05620003e-01 1.05056059e+00 -1.33407712e-01 3.36598396e-01
2.96064377e-01 2.66342133e-01 -1.14695811e+00 6.49494678e-02
7.67056584e-01 8.49860609e-01 -7.23102808e-01 3.39433877e-03
-6.01235271e-01 2.48608574e-01 4.89853919e-01 2.58064717e-01
-5.04693985e-01 -6.49417698e-01 7.05417633e-01 2.31198847e-01
-3.60593289e-01 -2.80563533e-01 -5.10234475e-01 -1.17073071e+00
-7.83176571e-02 -4.33333218e-01 4.29316968e-01 -2.10136980e-01
-1.06947064e+00 1.91800237e-01 3.59059602e-01 -5.66159606e-01
-2.25877076e-01 -1.91938296e-01 -1.04553831e+00 1.00231314e+00
-1.32643652e+00 -1.05908191e+00 2.42920294e-01 8.73287618e-01
5.67101061e-01 -4.90784138e-01 9.72671628e-01 5.00462949e-01
-9.44666386e-01 5.03957197e-02 1.15519343e-02 1.01251900e+00
2.47262463e-01 -6.80720806e-01 2.51878023e-01 1.03862870e+00
-1.15759119e-01 8.42697263e-01 3.13100010e-01 -1.16762662e+00
-8.63760412e-01 -1.33786666e+00 1.75178266e+00 -4.97379214e-01
6.35122001e-01 -1.64793104e-01 -7.42230713e-01 5.17604530e-01
-1.67284504e-01 -3.81928712e-01 1.11076248e+00 3.51462066e-01
1.47316441e-01 -1.09649234e-01 -1.44485688e+00 2.34644130e-01
5.15803576e-01 -4.46636647e-01 -8.43042552e-01 3.18054557e-01
5.35719469e-02 6.57671019e-02 -7.27206767e-01 1.88967362e-02
3.68169993e-01 -9.20600295e-01 5.71274638e-01 -8.13260913e-01
8.11234355e-01 3.65832448e-01 -1.89250201e-01 -1.22592247e+00
-4.02350843e-01 4.11055945e-02 6.53199613e-01 1.66755390e+00
4.62336540e-01 -5.19648552e-01 6.73668444e-01 1.26381862e+00
2.16205254e-01 -5.68580329e-01 -9.38939631e-01 -2.19216049e-01
1.99019402e-01 -1.00304198e+00 9.27476764e-01 1.39862478e+00
4.25224990e-01 -8.44991487e-03 -3.79180461e-01 1.75817281e-01
6.23897552e-01 -3.86089474e-01 6.22916698e-01 -1.03373611e+00
5.96464932e-01 -6.71795383e-02 -3.66507471e-01 1.67955801e-01
1.73204783e-02 -3.58548075e-01 -7.27884114e-01 -1.53916359e+00
6.32657290e-01 -8.53592217e-01 8.14567730e-02 8.19088638e-01
-1.29706994e-01 2.41226479e-01 -3.28370705e-02 1.46558778e-02
7.08494186e-02 2.35390128e-03 7.09551871e-01 -6.07182980e-02
-6.10755026e-01 4.37386632e-02 -7.06578553e-01 8.63652647e-01
1.45206761e+00 -6.85279787e-01 -2.24893466e-01 -8.50421190e-01
3.12855333e-01 2.50960559e-01 7.66612291e-02 -8.15646470e-01
2.81375378e-01 -9.73793209e-01 4.92836446e-01 -9.19069886e-01
-4.58200201e-02 -8.75556886e-01 6.78370520e-02 1.78134471e-01
9.33917835e-02 1.80833668e-01 1.25425965e-01 7.34336749e-02
-3.37573081e-01 -8.91913697e-02 4.77262497e-01 -9.72407907e-02
-4.30443019e-01 4.51315373e-01 -6.38366222e-01 -2.21949905e-01
1.21582663e+00 -4.96137619e-01 -3.73518348e-01 -2.19000787e-01
-4.57794815e-01 2.34353334e-01 1.19303919e-01 2.91388363e-01
7.35457838e-01 -1.32336569e+00 -1.20976961e+00 4.27949131e-01
-1.62426308e-01 -8.58995914e-02 3.31995815e-01 5.68662167e-01
-7.83008158e-01 7.60613441e-01 -5.48275232e-01 4.77094874e-02
-1.23476410e+00 2.01660708e-01 -4.82781529e-01 -2.95454532e-01
-2.62039751e-01 3.69362473e-01 -5.59432209e-01 -3.41224939e-01
-2.35752687e-01 7.89629072e-02 -9.60147858e-01 2.69774020e-01
9.96672332e-01 7.10973740e-01 5.69106787e-02 -1.17673469e+00
-4.12068903e-01 4.32980448e-01 -1.38876087e-03 -2.14298427e-01
2.24721098e+00 -7.91115910e-02 -3.99684638e-01 8.79778191e-02
1.13670647e+00 1.59392074e-01 -4.96747494e-01 -5.59398346e-02
2.29618952e-01 -1.01873684e+00 4.92230654e-02 -3.31930846e-01
-1.10787511e+00 8.49508703e-01 -8.88246894e-02 2.58291185e-01
1.32823503e+00 -2.35135719e-01 6.47030115e-01 2.43972555e-01
5.94344437e-01 -1.52674353e+00 -5.57533324e-01 6.96482658e-01
6.57920003e-01 -1.47586846e+00 1.69913843e-01 -1.05704665e-01
-4.53733385e-01 1.09332716e+00 3.17514420e-01 -2.41857450e-02
9.55192268e-01 3.62863004e-01 -1.70043811e-01 -1.41971588e-01
-3.95187855e-01 -3.35442647e-02 -2.99025953e-01 1.01796532e+00
4.79169846e-01 1.65619522e-01 -7.29449987e-01 8.41426849e-01
-3.60507697e-01 -6.04843199e-02 6.63728416e-01 1.00258899e+00
-7.07121193e-01 -1.05973244e+00 -8.39972615e-01 8.40394080e-01
-1.03134179e+00 -2.85445929e-01 -1.71518296e-01 6.63041294e-01
2.98063517e-01 1.53946090e+00 1.07257307e-01 -3.05923611e-01
1.16074853e-01 -4.29466486e-01 -1.07753433e-01 -7.39388585e-01
-4.90357339e-01 -2.72717535e-01 2.69325078e-01 -9.40055251e-02
-5.79512656e-01 -9.20060396e-01 -1.12568653e+00 -1.18805254e+00
-1.07725605e-01 3.31278563e-01 8.64379823e-01 9.71676528e-01
-1.69655293e-01 -1.94909662e-01 9.72527981e-01 -6.28440022e-01
2.65104473e-01 -1.09169388e+00 -7.75037766e-01 3.69226068e-01
1.73636734e-01 -4.56408590e-01 4.10557305e-03 1.41722947e-01] | [6.763400554656982, 1.941611886024475] |
ec7ffb65-0d93-4a55-b8e0-ae588eb76dfb | dfa-nerf-personalized-talking-head-generation | 2201.00791 | null | https://arxiv.org/abs/2201.00791v1 | https://arxiv.org/pdf/2201.00791v1.pdf | DFA-NeRF: Personalized Talking Head Generation via Disentangled Face Attributes Neural Rendering | While recent advances in deep neural networks have made it possible to render high-quality images, generating photo-realistic and personalized talking head remains challenging. With given audio, the key to tackling this task is synchronizing lip movement and simultaneously generating personalized attributes like head movement and eye blink. In this work, we observe that the input audio is highly correlated to lip motion while less correlated to other personalized attributes (e.g., head movements). Inspired by this, we propose a novel framework based on neural radiance field to pursue high-fidelity and personalized talking head generation. Specifically, neural radiance field takes lip movements features and personalized attributes as two disentangled conditions, where lip movements are directly predicted from the audio inputs to achieve lip-synchronized generation. In the meanwhile, personalized attributes are sampled from a probabilistic model, where we design a Transformer-based variational autoencoder sampled from Gaussian Process to learn plausible and natural-looking head pose and eye blink. Experiments on several benchmarks demonstrate that our method achieves significantly better results than state-of-the-art methods. | ['Xiaokang Yang', 'Guangtao Zhai', 'Yichao Yan', 'RuiZhe Zhong', 'Shunyu Yao'] | 2022-01-03 | null | null | null | null | ['talking-head-generation'] | ['computer-vision'] | [ 7.06389695e-02 1.17461927e-01 -7.87122548e-02 -3.62571508e-01
-1.15894246e+00 -1.96247354e-01 6.22667611e-01 -5.59906781e-01
3.32026243e-01 7.10916042e-01 8.22460592e-01 4.37028617e-01
3.84226590e-02 -5.36411524e-01 -9.29048717e-01 -1.14514172e+00
5.37214339e-01 2.15931460e-02 -2.58325666e-01 5.43993153e-02
2.55268849e-02 1.60444126e-01 -2.02064157e+00 1.61844239e-01
8.22821736e-01 1.18417251e+00 3.89987797e-01 7.98511684e-01
1.64946601e-01 7.11165130e-01 -4.51630622e-01 -3.23586911e-01
-1.23047099e-01 -5.12727380e-01 -3.37420553e-01 2.37740368e-01
4.18192595e-01 -2.95931578e-01 -4.30657119e-01 8.16561043e-01
1.12684810e+00 3.43574613e-01 8.93817842e-01 -1.53687048e+00
-1.00192916e+00 4.86971766e-01 -4.85596776e-01 -4.36893404e-01
6.58832490e-01 6.64761126e-01 9.62697923e-01 -7.79333293e-01
3.05925101e-01 1.50676525e+00 4.46385622e-01 6.87133908e-01
-1.39112496e+00 -8.09884310e-01 5.59581891e-02 4.23014700e-01
-1.42209947e+00 -9.44418311e-01 1.07299018e+00 -3.35883588e-01
9.09310058e-02 2.75800675e-01 6.57793760e-01 1.65863574e+00
5.45353964e-02 1.06961644e+00 8.61366689e-01 -1.32955194e-01
-4.27682400e-02 4.47218232e-02 -7.46058822e-01 3.49865109e-01
-4.05457228e-01 -2.21805423e-02 -8.70982587e-01 -3.58156189e-02
5.31239688e-01 -1.50485963e-01 -7.36674726e-01 -1.76807418e-01
-1.47797620e+00 6.18701220e-01 2.64360517e-01 -7.90271610e-02
-6.72114491e-01 3.09178621e-01 -2.57651173e-02 -5.99058688e-01
3.73198569e-01 1.03807278e-01 -1.61260426e-01 -1.28901824e-01
-8.74764681e-01 4.60649937e-01 3.91274691e-01 9.25620914e-01
6.93007171e-01 1.98529437e-01 -8.37528348e-01 9.23644066e-01
5.45200944e-01 9.20129180e-01 4.31778103e-01 -1.20417178e+00
3.04625839e-01 -3.16899121e-02 1.64463043e-01 -8.69163573e-01
-1.83605507e-01 -2.07578659e-01 -1.02534676e+00 -6.36280701e-02
2.82041371e-01 -3.78784388e-01 -6.86804593e-01 2.21551466e+00
5.34757137e-01 6.05097473e-01 8.48642923e-03 1.01583934e+00
8.08638990e-01 1.05954719e+00 1.44919038e-01 -4.34782118e-01
1.37499595e+00 -7.90669620e-01 -1.18778586e+00 1.55831695e-01
-1.46248907e-01 -9.58045006e-01 1.14642525e+00 4.28131491e-01
-1.36411989e+00 -7.48779118e-01 -5.53307772e-01 -2.87551016e-01
3.97102594e-01 1.10790737e-01 3.01610351e-01 4.06518072e-01
-1.06068814e+00 2.92443573e-01 -4.00268376e-01 2.97640376e-02
2.18751341e-01 6.59442618e-02 -1.29355133e-01 5.72137535e-02
-1.20719254e+00 3.77105772e-01 -1.73507556e-01 1.30026340e-02
-1.00658751e+00 -9.31265533e-01 -9.26651299e-01 1.12178242e-02
1.98000267e-01 -1.29447436e+00 1.50252533e+00 -6.55061245e-01
-2.33230829e+00 4.80622381e-01 -7.33502626e-01 -7.80102834e-02
5.16564548e-01 -1.06045686e-01 -3.47544640e-01 1.16661347e-01
8.59113485e-02 1.02589774e+00 1.25462615e+00 -1.52431881e+00
-6.70993388e-01 -3.08709294e-01 -3.59716922e-01 4.54795480e-01
-3.78030419e-01 -2.37505242e-01 -5.50583482e-01 -5.82706869e-01
-1.10182643e-01 -9.38975394e-01 2.47696564e-01 1.14812039e-01
-8.26316714e-01 -4.71804798e-01 7.83049345e-01 -7.84613848e-01
9.25776839e-01 -1.99332464e+00 3.34337950e-01 -4.05517757e-01
3.30435783e-01 -2.24706754e-01 4.04669903e-02 1.19270496e-01
5.83926141e-02 1.70648247e-02 2.88390853e-02 -8.89945447e-01
4.23223346e-01 -4.93332855e-02 -5.79979479e-01 5.86648583e-01
3.15833203e-02 9.30371106e-01 -8.08704555e-01 -7.20634043e-01
2.87607640e-01 1.19809401e+00 -6.96545005e-01 5.29104769e-01
-2.72716641e-01 7.41276920e-01 -4.40077007e-01 4.83685523e-01
7.35634446e-01 5.84525093e-02 -3.72747511e-01 -5.54279745e-01
-8.34620837e-03 1.82751358e-01 -1.03579438e+00 1.80120313e+00
-8.46626759e-01 8.81136835e-01 1.25714138e-01 -4.12046880e-01
9.37296391e-01 7.35770345e-01 4.47792619e-01 -4.96594340e-01
1.30963534e-01 -2.50112593e-01 -5.24405658e-01 -8.99273217e-01
5.00583351e-01 -2.06080139e-01 6.27568737e-02 2.80017406e-01
-4.67818677e-02 -4.03678000e-01 -5.29955804e-01 -1.23750046e-01
3.65639091e-01 2.44686410e-01 -1.59606159e-01 1.53465629e-01
5.10507107e-01 -8.55496883e-01 5.88730216e-01 2.11650431e-01
-1.09602511e-01 1.00417387e+00 2.95010239e-01 1.47407636e-01
-9.99998868e-01 -1.41138422e+00 -1.01514114e-02 1.06783259e+00
2.81363040e-01 -1.88368291e-01 -1.22570682e+00 1.65507004e-01
-3.15093309e-01 9.24142301e-01 -6.06175363e-01 -2.26559043e-01
-2.44996592e-01 -4.14258033e-01 5.67475736e-01 3.44364524e-01
3.53197694e-01 -1.08132553e+00 -2.87031174e-01 1.47666484e-01
-9.11888659e-01 -1.13801003e+00 -9.29845929e-01 -6.51454389e-01
-1.15284599e-01 -4.47381735e-01 -1.25589669e+00 -5.60845792e-01
1.69254541e-01 8.47045705e-02 8.31720948e-01 -6.38350964e-01
-2.72201478e-01 2.72096723e-01 1.21334150e-01 -5.03970802e-01
-2.76876688e-01 -5.35676964e-02 2.69481987e-01 7.15881467e-01
-9.76213887e-02 -8.44720244e-01 -1.11324215e+00 1.49664015e-01
-6.90743685e-01 3.29236627e-01 5.43436706e-01 7.22112298e-01
7.57365882e-01 -1.32659256e-01 5.49161136e-01 -2.68023927e-02
7.76625454e-01 -6.36076212e-01 -2.35468507e-01 9.70640257e-02
-1.98694587e-01 8.41668099e-02 6.28120303e-01 -9.44437504e-01
-1.51406717e+00 -4.92987921e-03 -2.74248302e-01 -7.03160644e-01
-3.67375880e-01 -2.06927314e-01 -8.67160320e-01 5.29063880e-01
4.60500509e-01 5.34251213e-01 -6.28578812e-02 -4.06940281e-01
6.23354018e-01 9.51496184e-01 9.39835787e-01 -5.80923796e-01
5.46136737e-01 6.20483220e-01 5.37216365e-02 -1.02016675e+00
-6.62789762e-01 -1.62560657e-01 2.25668643e-02 -4.55917954e-01
1.19776249e+00 -9.45586443e-01 -1.40181530e+00 8.78794372e-01
-1.36165380e+00 -3.16047311e-01 -1.16299585e-01 3.91956270e-01
-1.03931022e+00 2.01673642e-01 -3.15452129e-01 -1.01303053e+00
-3.23908478e-01 -1.55790746e+00 1.67682385e+00 5.18952429e-01
-7.69096985e-02 -6.73523784e-01 7.87447318e-02 5.70905983e-01
2.66974688e-01 3.09141994e-01 5.93613982e-01 1.57860264e-01
-8.02756727e-01 -2.92833112e-02 5.10996543e-02 1.03711270e-01
3.07861388e-01 1.57031983e-01 -1.64489722e+00 1.77618876e-01
-5.05214445e-02 -2.11057559e-01 4.20336455e-01 8.98479044e-01
1.44423664e+00 -6.88039064e-01 -2.74681747e-01 9.16866601e-01
8.14847887e-01 -1.53953418e-01 6.76822960e-01 -3.63859743e-01
7.79372215e-01 1.01309848e+00 2.99583703e-01 7.94962525e-01
7.47132182e-01 1.04883325e+00 4.88593787e-01 1.07632525e-01
-3.72025400e-01 -8.58801842e-01 3.93146992e-01 6.00298643e-01
9.30106416e-02 -6.06399119e-01 -5.30612350e-01 5.21543920e-01
-1.64132750e+00 -1.16495919e+00 -1.97231770e-02 2.01511717e+00
9.69460130e-01 -4.03345287e-01 1.10579841e-01 1.42149702e-01
1.02939796e+00 1.60718560e-01 -7.36213684e-01 7.04933330e-02
-2.40041390e-01 -4.65437360e-02 -6.15398511e-02 6.09296501e-01
-6.76515758e-01 8.51146519e-01 5.02906704e+00 1.01948452e+00
-1.43298924e+00 9.17255953e-02 7.51184642e-01 -2.96436340e-01
-7.85797238e-01 -4.35582042e-01 -8.10327590e-01 6.77240431e-01
7.24008203e-01 -3.31391841e-01 5.06211579e-01 6.43265188e-01
7.87031949e-01 2.45962799e-01 -9.76502359e-01 1.29129720e+00
2.44333088e-01 -1.25741422e+00 7.60643557e-02 1.07515156e-01
7.13730633e-01 -4.50987905e-01 6.68257535e-01 -3.38739716e-02
5.19720167e-02 -1.18169737e+00 1.07238960e+00 1.03589272e+00
1.05863929e+00 -8.51714790e-01 -5.15598878e-02 3.74979526e-01
-1.21507597e+00 1.25825956e-01 -9.79960263e-02 4.59443033e-01
4.56042230e-01 3.48893672e-01 -8.45898271e-01 2.20768288e-01
7.49135256e-01 5.02219558e-01 1.40372708e-01 9.58447814e-01
-4.15561110e-01 4.04115915e-01 -1.35650963e-01 -2.71197818e-02
-1.59779593e-01 6.96000606e-02 8.31476927e-01 7.61990726e-01
5.75318098e-01 3.16248275e-02 -2.71285385e-01 1.26205218e+00
-2.12957904e-01 6.81338534e-02 -4.92494345e-01 3.20448071e-01
6.78252220e-01 1.28090811e+00 9.92975757e-02 5.82711175e-02
1.74078986e-01 9.89097238e-01 -1.83861449e-01 6.58725917e-01
-1.06683445e+00 -2.75515527e-01 1.16873860e+00 2.31801048e-01
7.94389695e-02 1.53594002e-01 2.90073864e-02 -1.01827633e+00
1.89653821e-02 -7.45809495e-01 -2.85451263e-01 -1.46202171e+00
-1.01129806e+00 6.41316831e-01 -2.08101615e-01 -1.08688271e+00
-6.01933897e-01 -1.77101627e-01 -7.60335147e-01 1.28487742e+00
-1.44679093e+00 -1.43870807e+00 -6.16782010e-01 1.04344296e+00
7.06238747e-01 2.55761743e-01 6.13767803e-01 9.39671695e-02
-5.54875076e-01 9.23841417e-01 -6.75543994e-02 -2.07465917e-01
1.02033305e+00 -9.99246716e-01 2.76372403e-01 4.60119545e-01
1.01633640e-02 3.09174776e-01 1.05297434e+00 -1.65509686e-01
-1.41736841e+00 -1.07868695e+00 6.62042856e-01 -3.19005758e-01
3.06420803e-01 -4.54090804e-01 -6.88420117e-01 2.96260625e-01
3.84136468e-01 -5.06166331e-02 5.76783061e-01 -3.38129461e-01
-1.73278660e-01 -2.74984956e-01 -9.00946975e-01 8.66581261e-01
9.32942331e-01 -6.94409013e-01 -2.91526437e-01 2.50235289e-01
1.03109944e+00 -3.61022890e-01 -7.18015611e-01 2.36443609e-01
6.42942488e-01 -9.85143363e-01 1.02237821e+00 -1.31351352e-01
5.03737271e-01 -2.31718719e-01 -1.80891693e-01 -1.54801631e+00
2.66593117e-02 -1.40447998e+00 -2.15453237e-01 1.56099021e+00
2.70719796e-01 -3.67281318e-01 6.65788710e-01 4.85965312e-01
-2.82161981e-01 -7.42365122e-01 -7.58529723e-01 -3.30534637e-01
-4.48821336e-02 -5.63464165e-01 1.04802668e+00 5.58168352e-01
-2.41450384e-01 5.29964924e-01 -8.34713399e-01 2.91649431e-01
9.05211985e-01 5.53423800e-02 1.13996971e+00 -1.02523458e+00
-2.53763199e-01 -4.26120937e-01 -7.59212822e-02 -1.45364523e+00
4.93168026e-01 -3.61790597e-01 4.18291599e-01 -1.50390136e+00
8.02101567e-02 -2.77846515e-01 2.37006098e-01 -6.08203933e-02
-3.82955849e-01 9.48938802e-02 1.74561828e-01 1.97351240e-02
-1.88591182e-01 1.20496082e+00 1.68617105e+00 -8.92579108e-02
-2.20204100e-01 1.62963167e-01 -7.61112273e-01 6.78712070e-01
4.51238543e-01 -1.18697256e-01 -6.65930748e-01 -4.59442139e-01
8.17792714e-02 6.45039082e-01 4.93928283e-01 -9.30936635e-01
4.56335336e-01 -3.30250025e-01 2.74828732e-01 -6.09805048e-01
1.00726974e+00 -4.34372932e-01 9.55326259e-02 -1.96499676e-01
-4.78370726e-01 -3.24850559e-01 -1.59156770e-02 6.64409220e-01
-1.65107295e-01 4.00974333e-01 7.93668687e-01 3.54140759e-01
-2.24521220e-01 8.44037354e-01 -1.66808769e-01 2.57454842e-01
8.28983366e-01 -1.93102330e-01 -3.63419652e-02 -1.13636708e+00
-5.98443747e-01 1.11210067e-02 3.01084280e-01 7.04119742e-01
5.72792232e-01 -1.52155232e+00 -8.31753790e-01 6.19341880e-02
5.34467436e-02 2.55356848e-01 8.02863777e-01 7.01208472e-01
3.54157649e-02 3.03419292e-01 1.34434691e-02 -9.99073505e-01
-1.04571199e+00 4.73926544e-01 3.58242750e-01 4.28410798e-01
-4.40529943e-01 9.65954304e-01 7.17553973e-01 -2.06728280e-02
4.25931901e-01 -1.14841007e-01 -1.71329081e-01 7.89937451e-02
5.72121084e-01 3.48995209e-01 -4.38991398e-01 -9.23126101e-01
8.83236825e-02 7.39034712e-01 5.28007209e-01 -3.01207364e-01
1.14262581e+00 -5.83554089e-01 3.02586854e-01 5.46402454e-01
1.42473280e+00 4.32050198e-01 -1.75584936e+00 -1.56131878e-01
-7.75626838e-01 -2.89861202e-01 2.58517712e-01 -4.13173705e-01
-1.07823920e+00 1.24351943e+00 4.39583898e-01 -5.55210076e-02
1.29489994e+00 9.89333913e-02 1.12193274e+00 -1.38332278e-01
-3.46720554e-02 -8.70750487e-01 4.01113003e-01 -3.09535097e-02
9.06008244e-01 -1.05070889e+00 -6.03657246e-01 -3.23398948e-01
-8.66926074e-01 7.60742009e-01 4.58646119e-01 3.97348881e-01
4.18190569e-01 2.11494472e-02 1.17400520e-01 3.09260637e-01
-8.11687291e-01 -5.05554443e-03 5.10181487e-01 8.64809215e-01
2.67817557e-01 2.53459543e-01 5.29707134e-01 8.21160078e-01
-6.37465835e-01 -8.57048407e-02 2.25127861e-01 2.14607064e-02
-1.29214376e-01 -6.66377962e-01 -7.07783103e-01 -1.30554631e-01
-4.19441164e-01 -1.15367286e-01 -2.25836542e-02 3.68479848e-01
2.25838646e-01 1.03432477e+00 1.41382247e-01 -4.10685986e-01
1.39369681e-01 1.12243138e-01 3.89451325e-01 -1.76727995e-01
6.98319301e-02 3.50593507e-01 -3.14638048e-01 -5.45181572e-01
-1.42327815e-01 -7.26918459e-01 -8.84558916e-01 -5.22579849e-01
-2.05445409e-01 -5.78480214e-02 8.38183582e-01 7.27096379e-01
4.89607751e-01 5.90769768e-01 8.49055767e-01 -1.34621668e+00
-5.33699393e-01 -1.09561229e+00 -5.16670406e-01 5.20550251e-01
7.07511485e-01 -6.98230147e-01 -5.53052723e-01 3.69420737e-01] | [13.240913391113281, -0.41317978501319885] |
f5ac9bfc-e2b1-4283-853c-79e80c50d08b | towards-understanding-the-generalization-of | 2303.12898 | null | https://arxiv.org/abs/2303.12898v1 | https://arxiv.org/pdf/2303.12898v1.pdf | Towards Understanding the Generalization of Medical Text-to-SQL Models and Datasets | Electronic medical records (EMRs) are stored in relational databases. It can be challenging to access the required information if the user is unfamiliar with the database schema or general database fundamentals. Hence, researchers have explored text-to-SQL generation methods that provide healthcare professionals direct access to EMR data without needing a database expert. However, currently available datasets have been essentially "solved" with state-of-the-art models achieving accuracy greater than or near 90%. In this paper, we show that there is still a long way to go before solving text-to-SQL generation in the medical domain. To show this, we create new splits of the existing medical text-to-SQL dataset MIMICSQL that better measure the generalizability of the resulting models. We evaluate state-of-the-art language models on our new split showing substantial drops in performance with accuracy dropping from up to 92% to 28%, thus showing substantial room for improvement. Moreover, we introduce a novel data augmentation approach to improve the generalizability of the language models. Overall, this paper is the first step towards developing more robust text-to-SQL models in the medical domain.\footnote{The dataset and code will be released upon acceptance. | ['Anthony Rios', 'Glenn Dietrich', 'Kim-Kwang Raymond Choo', 'Richard Tarbell'] | 2023-03-22 | null | null | null | null | ['text-to-sql'] | ['computer-code'] | [ 2.39054963e-01 4.69310552e-01 -1.80181354e-01 -7.39724219e-01
-1.41574597e+00 -3.81061733e-01 2.90434867e-01 7.72684872e-01
-3.77111256e-01 5.90212762e-01 3.10309172e-01 -7.53758192e-01
-1.11660525e-01 -8.67711961e-01 -5.96132457e-01 -5.90219796e-02
8.18648860e-02 8.88719201e-01 -6.78832531e-02 -4.70164359e-01
-9.15186778e-02 4.23602611e-01 -1.15070593e+00 9.63110447e-01
7.03024685e-01 8.26380372e-01 -2.11639896e-01 6.83858395e-01
-3.87350172e-01 1.08528149e+00 -5.90243578e-01 -5.27474344e-01
2.24803463e-01 -1.80132970e-01 -1.07698119e+00 -1.72653809e-01
4.21634942e-01 -1.89970911e-01 -2.73558885e-01 6.93387449e-01
6.19047582e-01 -2.91557789e-01 2.26023778e-01 -8.87231290e-01
-6.35039270e-01 7.48990715e-01 -1.37739211e-01 -1.22117274e-01
8.17903340e-01 2.88480669e-02 8.22072983e-01 -6.85627222e-01
8.97943735e-01 9.68026042e-01 8.38308752e-01 6.10446751e-01
-1.43515360e+00 -4.45143074e-01 -1.57980308e-01 -2.07457572e-01
-1.55303049e+00 -5.38157463e-01 2.19247088e-01 -4.07525331e-01
1.28982222e+00 6.12811148e-01 3.01097780e-01 7.12118924e-01
7.85013437e-02 6.82265639e-01 1.01659679e+00 -4.13486719e-01
5.81494235e-02 5.61067879e-01 2.50228584e-01 5.28895080e-01
3.94458711e-01 -1.71049982e-01 -4.49351281e-01 -5.19835770e-01
4.76390809e-01 2.47147470e-03 -6.11980781e-02 -2.52386987e-01
-1.26621091e+00 6.99104726e-01 1.24590516e-01 2.81975538e-01
-3.77946287e-01 -3.74079108e-01 5.07326841e-01 3.74375135e-01
2.51355022e-01 7.02559888e-01 -6.53661251e-01 -2.56444216e-01
-1.08631396e+00 5.21992564e-01 1.06486523e+00 1.35235322e+00
3.51279795e-01 -3.72651815e-01 -1.63862482e-01 9.10390019e-01
7.59147108e-02 2.29321703e-01 4.14272755e-01 -6.72059953e-01
6.58461452e-01 8.86876762e-01 4.10533771e-02 -8.31694961e-01
-5.85059047e-01 -1.07615441e-01 -8.81727397e-01 -2.29522482e-01
2.54305542e-01 -3.16893943e-02 -1.08005154e+00 1.52223468e+00
9.30138975e-02 -3.45113426e-01 3.64060670e-01 3.35302919e-01
1.09668362e+00 5.36341548e-01 2.61363298e-01 -2.37717614e-01
1.53779960e+00 -3.46785277e-01 -7.89699972e-01 -2.37122878e-01
1.11096048e+00 -8.49756539e-01 1.05745482e+00 4.11038607e-01
-1.21708298e+00 -4.18513566e-01 -8.46576750e-01 -3.21289301e-01
-4.65468317e-01 -1.07463732e-01 5.76593041e-01 8.45674694e-01
-1.03180373e+00 3.59061927e-01 -1.07512200e+00 -5.55996418e-01
3.57798666e-01 4.51649964e-01 -6.53393567e-01 -3.60732108e-01
-1.02462375e+00 1.01748002e+00 5.26421010e-01 -2.59793222e-01
-2.69932449e-01 -1.09394908e+00 -8.78171682e-01 -2.98117697e-01
5.16056657e-01 -7.68749535e-01 1.23543966e+00 -5.22127235e-03
-8.09964359e-01 1.09634638e+00 -2.45056435e-01 -4.23527092e-01
6.27430320e-01 -1.44479766e-01 -5.97661555e-01 -1.73907936e-01
1.08218841e-01 4.25476491e-01 -1.81751236e-01 -1.11740601e+00
-4.74395692e-01 -5.23046017e-01 -3.07682399e-02 -1.60928577e-01
-1.87426776e-01 1.56870946e-01 -6.84511364e-01 -7.73317635e-01
9.37855169e-02 -9.43571627e-01 -5.09963632e-01 -2.09942579e-01
-5.70416152e-01 -3.45267765e-02 9.10879299e-02 -7.91330218e-01
1.67912292e+00 -2.10466623e+00 -3.67629141e-01 2.51177192e-01
2.60990143e-01 2.66974866e-01 -2.90726312e-02 7.10752785e-01
-4.09733176e-01 4.22519147e-01 -2.92154282e-01 -1.68942079e-01
-1.64216772e-01 2.85229802e-01 -3.03860158e-01 -3.19544040e-02
3.27820241e-01 9.33368266e-01 -5.63054323e-01 -6.31878078e-01
-3.37529443e-02 3.53291243e-01 -7.82992899e-01 3.69852126e-01
-1.93450078e-01 7.89319724e-02 -4.27151859e-01 6.85679376e-01
5.29661417e-01 -3.97825629e-01 4.20345545e-01 -2.48578578e-01
1.99333131e-01 5.48211813e-01 -1.32183492e+00 1.63552630e+00
-3.53901505e-01 8.78892243e-02 -1.70553178e-01 -8.85662735e-01
8.52428257e-01 4.24416184e-01 8.17742050e-01 -7.00641751e-01
-2.74459153e-01 2.89251357e-01 -6.06513061e-02 -8.54156673e-01
6.16812050e-01 -4.95294333e-01 -1.56674623e-01 2.99475610e-01
-3.43569338e-01 -1.01351656e-01 2.81236053e-01 2.89444834e-01
1.20228934e+00 -2.40249932e-01 5.11330843e-01 -1.02154464e-01
2.41804659e-01 4.30987865e-01 2.89399534e-01 8.26350212e-01
3.79952967e-01 7.51866460e-01 3.63047421e-01 -4.67629224e-01
-9.48275387e-01 -1.02801299e+00 -5.31974196e-01 7.55177379e-01
-6.80671275e-01 -8.85156631e-01 -5.25006115e-01 -5.85211813e-01
2.13172540e-01 1.02219665e+00 -5.04905820e-01 -1.76127523e-01
-6.95880353e-01 -1.00902438e+00 8.56436610e-01 8.10896754e-01
-1.22516634e-04 -9.06664789e-01 -4.91003782e-01 4.23707187e-01
-4.34294850e-01 -1.31378663e+00 -2.19430849e-01 1.32115304e-01
-1.11969721e+00 -9.41163123e-01 -4.28126901e-01 -5.05221367e-01
6.41846776e-01 -3.61266673e-01 1.46661043e+00 3.34720761e-02
-6.78871036e-01 2.61696160e-01 -2.53911406e-01 -6.29039407e-01
-1.00951576e+00 2.99328506e-01 -3.47118527e-01 -5.95928729e-01
7.28153944e-01 -7.86017533e-03 -2.37309292e-01 1.00826491e-02
-1.42189109e+00 4.55533750e-02 6.48684800e-01 6.55772209e-01
6.75295293e-01 -4.42252383e-02 5.98062932e-01 -1.46938598e+00
7.19172776e-01 -4.46967840e-01 -2.13954747e-01 4.30367559e-01
-9.16726053e-01 2.99190849e-01 5.05569994e-01 -2.30893508e-01
-6.27655864e-01 1.76599085e-01 -5.28427601e-01 3.06044798e-02
-2.84993738e-01 9.84521568e-01 -1.66522369e-01 4.92253125e-01
7.75306523e-01 6.55373260e-02 1.29986331e-01 -6.60693645e-01
4.47078973e-01 7.86908686e-01 5.57036638e-01 -5.67367196e-01
4.59475487e-01 2.54489511e-01 -1.65115952e-01 -5.73831558e-01
-6.26866937e-01 -4.38669086e-01 -4.48024273e-01 5.05467236e-01
7.49981761e-01 -9.01511729e-01 -6.13658965e-01 1.34980619e-01
-7.73986697e-01 -1.94159791e-01 -5.17368555e-01 2.47415558e-01
-4.33402956e-01 3.57032835e-01 -7.02824354e-01 -6.65097475e-01
-3.71503532e-01 -9.79302585e-01 1.12640667e+00 -4.34994012e-01
-6.49819195e-01 -9.44803357e-01 -7.76430517e-02 6.34402454e-01
4.54305410e-01 2.95360506e-01 1.36828792e+00 -1.11804605e+00
-3.99501234e-01 -6.18286729e-01 -1.75413668e-01 1.08068846e-01
4.40848500e-01 -2.02666536e-01 -7.20056176e-01 -3.17065477e-01
3.67102437e-02 -2.38261551e-01 5.26628554e-01 8.28787684e-02
1.30209005e+00 -4.53550458e-01 -5.28748751e-01 4.33751315e-01
1.37199414e+00 2.52762169e-01 8.26001704e-01 2.06674159e-01
4.88865316e-01 5.55218279e-01 6.03044152e-01 5.65593004e-01
7.25567043e-01 7.29481995e-01 -1.44715607e-01 -3.75194818e-01
7.64450133e-02 -2.42233559e-01 -7.99068585e-02 6.55549407e-01
2.60813117e-01 -7.63100684e-02 -1.42975593e+00 6.17531180e-01
-1.57775998e+00 -6.07779562e-01 -1.93072110e-01 2.46678519e+00
1.53033257e+00 1.76257938e-01 7.16998726e-02 3.96570079e-02
1.45397589e-01 -3.15255344e-01 -3.68443102e-01 -2.93376356e-01
9.46044177e-02 4.65834200e-01 3.69874418e-01 5.11285841e-01
-8.62317979e-01 5.82020283e-01 7.02144766e+00 4.41025555e-01
-1.05674314e+00 -2.20123693e-01 7.07655966e-01 -2.15945914e-01
-3.00101995e-01 -2.00778440e-01 -9.51263011e-01 1.98564529e-01
1.28352284e+00 -2.73104876e-01 1.88274413e-01 9.08612072e-01
3.49923186e-02 9.05927718e-02 -1.65829098e+00 1.05926704e+00
8.98640305e-02 -1.35620487e+00 3.03190410e-01 1.68438733e-01
2.99194455e-01 -8.22488889e-02 -9.73561034e-02 3.81618589e-01
3.17783803e-01 -1.50307751e+00 2.23152846e-01 7.03283668e-01
9.65442955e-01 -5.73824942e-01 9.30778682e-01 3.13834727e-01
-7.94195950e-01 1.04217060e-01 -2.54728824e-01 3.61466497e-01
1.70726746e-01 5.50394058e-01 -1.36143923e+00 9.09060717e-01
6.28626585e-01 3.29548210e-01 -1.02497315e+00 6.35774970e-01
6.42697990e-01 1.96029902e-01 -3.77949893e-01 3.10940146e-01
-2.09607095e-01 2.61352241e-01 -1.02800922e-02 1.44130445e+00
4.98676866e-01 2.47338101e-01 2.88448244e-01 7.87528336e-01
-1.28848672e-01 3.86875182e-01 -8.46914768e-01 -3.46167713e-01
1.83606833e-01 9.00233984e-01 -2.83691317e-01 -5.61457813e-01
-5.44302523e-01 5.35898566e-01 1.59043178e-01 9.86183807e-02
-5.92829168e-01 -4.11501765e-01 4.15689319e-01 6.26538575e-01
2.32235342e-03 1.15974374e-01 -5.63330829e-01 -1.08248448e+00
3.91276240e-01 -1.59097719e+00 7.47187793e-01 -6.08797252e-01
-1.46081066e+00 7.63316751e-01 1.81735471e-01 -9.17107642e-01
-5.66029608e-01 -5.32916844e-01 2.83526689e-01 9.51766968e-01
-1.05318820e+00 -9.92586136e-01 -2.20934048e-01 5.26262641e-01
1.53916329e-01 -2.19889045e-01 1.36591041e+00 6.93161488e-01
-2.70415395e-01 8.89395356e-01 -6.86275810e-02 5.01155198e-01
8.99727046e-01 -1.20723307e+00 5.23666680e-01 5.18122554e-01
-7.37331156e-03 1.29592705e+00 5.35602093e-01 -7.87405849e-01
-1.63273847e+00 -1.13814449e+00 1.24035561e+00 -1.07293773e+00
3.62543464e-01 -3.74218911e-01 -1.16323304e+00 9.73196089e-01
-3.06525268e-02 -1.67129442e-01 1.10305142e+00 2.18913987e-01
-3.03188175e-01 -2.58863032e-01 -1.19740736e+00 4.57550377e-01
6.82555616e-01 -6.60064697e-01 -7.34934151e-01 4.28806216e-01
6.52897596e-01 -5.41394830e-01 -1.41454887e+00 7.45483994e-01
4.24258024e-01 -6.12712383e-01 1.10284257e+00 -1.15596497e+00
5.26735961e-01 -1.76565826e-01 -3.91634494e-01 -7.88675904e-01
4.40673605e-02 -5.19786000e-01 7.92763680e-02 1.10305107e+00
7.71536589e-01 -5.40769517e-01 9.17896509e-01 1.30588698e+00
-4.95387539e-02 -8.86516213e-01 -6.96234286e-01 -4.98139769e-01
2.97068805e-01 -7.95796037e-01 6.80332124e-01 1.13345814e+00
3.75933409e-01 2.56754756e-01 -4.98432852e-02 5.98073453e-02
2.02429235e-01 -2.79924553e-02 8.31761360e-01 -1.01950240e+00
-3.58156979e-01 -1.80486739e-01 -2.61822701e-01 -9.33794439e-01
-4.87403661e-01 -1.15195084e+00 -1.54493183e-01 -1.91916084e+00
4.79690790e-01 -6.82245135e-01 -1.97606385e-01 8.03632021e-01
-3.17706645e-01 4.95012403e-02 1.34609982e-01 -1.30909845e-01
-2.90607661e-01 -1.36675104e-01 8.29155982e-01 -1.05897874e-01
-2.93081313e-01 4.56742290e-03 -1.15568089e+00 3.29051465e-01
5.68513393e-01 -6.11991704e-01 -3.91352832e-01 -4.23624903e-01
3.62072796e-01 5.02094626e-01 1.47228912e-01 -7.89368868e-01
2.63452083e-01 -9.23140720e-02 2.50986516e-01 -6.84039235e-01
1.83467746e-01 -7.67651975e-01 4.58820492e-01 4.61354971e-01
-6.81666970e-01 4.86284435e-01 4.79582042e-01 1.22179352e-01
-2.23141208e-01 -4.87605669e-03 6.50841236e-01 -2.64283627e-01
-9.15372074e-02 1.04016878e-01 -2.61222273e-01 3.96605670e-01
7.13985682e-01 4.90652919e-02 -2.88510799e-01 -3.37060839e-01
-1.00828874e+00 1.97605565e-01 3.86915267e-01 5.46609998e-01
6.46087766e-01 -1.08419752e+00 -9.48759258e-01 4.11298186e-01
4.84721303e-01 1.91139013e-01 1.05647653e-01 7.01152205e-01
-6.85326397e-01 8.49444449e-01 1.22128434e-01 -5.01784325e-01
-1.47938728e+00 8.82762671e-01 2.65614063e-01 -5.34321249e-01
-5.79362094e-01 4.41473216e-01 -8.98485258e-02 -8.05103779e-01
2.16306552e-01 -8.05021584e-01 1.22640640e-01 1.22958636e-02
5.82574308e-01 1.40642598e-01 5.20581186e-01 -1.54981807e-01
-5.33091307e-01 2.61292636e-01 -4.42384064e-01 -7.07452279e-03
1.44565523e+00 1.76463842e-01 -2.52876878e-01 4.14753616e-01
1.13950884e+00 1.12260804e-01 -2.17343539e-01 -2.49179140e-01
4.19284672e-01 -2.98528790e-01 -3.20201367e-01 -1.17064869e+00
-8.56497765e-01 7.58059680e-01 6.21072650e-01 2.36436620e-01
1.18512106e+00 1.35897249e-01 7.99663365e-01 6.23727322e-01
2.44686082e-01 -8.16221416e-01 -3.22628766e-01 2.98624247e-01
7.95932651e-01 -1.33909559e+00 2.42683142e-01 -3.57201606e-01
-7.22431421e-01 9.90906358e-01 4.18771923e-01 5.06472945e-01
7.95254469e-01 7.21544325e-01 4.21320379e-01 -3.48111331e-01
-8.74967098e-01 1.39034882e-01 2.73599863e-01 4.90169913e-01
1.02607107e+00 1.24764172e-02 -8.23725462e-02 7.24955082e-01
-5.70028067e-01 4.60152388e-01 4.84320432e-01 1.12034130e+00
2.17827767e-01 -1.56834722e+00 -2.42410168e-01 7.82535017e-01
-9.93197381e-01 -4.56784964e-01 -2.52277672e-01 9.68000114e-01
-1.88148260e-01 9.87110198e-01 -2.17840195e-01 -1.75808504e-01
8.63505840e-01 4.32561189e-01 2.06418723e-01 -9.39147413e-01
-7.17023611e-01 7.01674223e-02 4.15179759e-01 -5.30409694e-01
-6.53965771e-02 -5.95150292e-01 -1.39479280e+00 -3.51692379e-01
5.65890037e-02 5.40563650e-02 4.22779173e-01 6.35138452e-01
5.66413224e-01 6.09587133e-01 -8.00635666e-02 2.64568508e-01
-7.07284212e-01 -8.44965816e-01 -1.91853806e-01 7.43671238e-01
2.54261374e-01 1.04039334e-01 3.79315257e-01 2.77939886e-01] | [8.533320426940918, 8.571063041687012] |
f08c0087-d888-4a99-8983-68f1094f161e | good-for-misconceived-reasons-revisiting | null | null | https://openreview.net/forum?id=Q9U_H8lQ4yV | https://openreview.net/pdf?id=Q9U_H8lQ4yV | Good for Misconceived Reasons: Revisiting Neural Multimodal Machine Translation | A neural multimodal machine translation (MMT) system is one that aims to perform better translation by extending conventional text-only translation models with multimodal information. Many recent studies report improvements when equipping their models with the multimodal module, despite the controversy whether such improvements indeed come from the multimodal part. We revisit the recent development of neural multimodal machine translation by proposing two \textit{interpretable} MMT models that achieve new state-of-the-art results on the standard \dataset\ dataset. To our surprise, however, while we observe similar gains as in the recent developed multimodal-integrated models, our models learn to \textit{ignore} the multimodal information. Upon further investigation, we discover that the improvements bought about by the multimodal models over text-only counterpart are in fact results of the regularization effect. We report our empirical findings which express the importance of MMT models' interpretability and set new paradigms for future MMT research. | ['Ben Kao', 'Lingpeng Kong', 'Zhiyong Wu'] | 2021-01-01 | null | null | null | null | ['multimodal-machine-translation'] | ['natural-language-processing'] | [ 5.40621758e-01 4.61040348e-01 -5.25497854e-01 -3.94923419e-01
-1.26379347e+00 -6.34897053e-01 1.00082135e+00 -1.10833198e-01
-3.32622975e-01 8.55141222e-01 4.51160640e-01 -6.46056294e-01
2.06513796e-02 -2.67415553e-01 -9.42496717e-01 -6.17667139e-01
4.68664616e-01 1.09673381e+00 -6.55421674e-01 -6.78887308e-01
-7.09148943e-02 -1.64063990e-01 -9.30746198e-01 7.95067906e-01
9.92290795e-01 6.83817863e-01 -1.27805322e-01 5.48588097e-01
-3.22558790e-01 6.57703519e-01 -2.02141404e-01 -1.25483584e+00
1.45480841e-01 -6.40347838e-01 -7.48233378e-01 -9.80670936e-03
5.22125483e-01 -2.05753490e-01 -1.88543946e-01 9.52356875e-01
5.34170628e-01 -2.81996787e-01 7.10428119e-01 -1.28212130e+00
-1.41940522e+00 9.59230900e-01 -4.34424877e-01 -3.84933203e-01
4.44051951e-01 1.23098567e-01 1.36022151e+00 -1.19422257e+00
8.55535746e-01 1.47349465e+00 5.79140782e-01 7.58850873e-01
-1.50121188e+00 -1.55442610e-01 2.83865146e-02 -2.16983026e-04
-9.95628297e-01 -6.37432337e-01 5.48845828e-01 -1.64220765e-01
9.35492277e-01 6.57437742e-01 7.58577585e-02 1.65340614e+00
3.11359644e-01 1.08687806e+00 1.29614353e+00 -7.44435966e-01
-3.65788639e-01 4.10040081e-01 -3.21461260e-02 6.52372956e-01
9.02775452e-02 -8.82903934e-02 -7.98680246e-01 2.52408702e-02
3.75012100e-01 -3.52359653e-01 -2.10412443e-01 -2.32461914e-01
-1.70486665e+00 8.69787872e-01 2.94390209e-02 3.49208713e-01
-1.18558392e-01 2.61840194e-01 4.10995036e-01 8.30419958e-01
7.14413345e-01 4.24977928e-01 -6.14008188e-01 -2.99954921e-01
-7.79912949e-01 -7.21755996e-02 7.42188752e-01 1.08400428e+00
8.91697466e-01 -8.92428756e-02 -1.96413919e-01 9.62952197e-01
3.36600602e-01 9.14330840e-01 2.40376577e-01 -9.21079874e-01
1.16784322e+00 7.23121405e-01 1.83714386e-02 -5.99875391e-01
-2.95780331e-01 -3.18083137e-01 -8.06982398e-01 -2.62496680e-01
4.75511968e-01 -4.65312213e-01 -5.86371541e-01 1.88010573e+00
-3.42202544e-01 -7.31441081e-01 3.38491023e-01 9.53244328e-01
7.77181089e-01 4.66461450e-01 1.44230574e-01 -2.03411460e-01
1.21316171e+00 -1.02817261e+00 -1.05353951e+00 -2.96026021e-01
1.02565455e+00 -1.18677878e+00 1.16141653e+00 1.53702036e-01
-1.16239536e+00 -2.36311644e-01 -7.29441047e-01 -3.51588935e-01
-6.52715802e-01 5.02311051e-01 8.67778838e-01 7.81389356e-01
-1.28733122e+00 3.78604621e-01 -4.89868879e-01 -6.78084850e-01
9.71665606e-02 6.70463622e-01 -5.29788673e-01 3.78231928e-02
-1.16574347e+00 1.40226018e+00 -7.54475817e-02 2.69474566e-01
-5.46242833e-01 -1.63353130e-01 -8.31866086e-01 -3.34533036e-01
2.56233066e-01 -1.07678378e+00 1.34219849e+00 -1.60135865e+00
-1.62259161e+00 1.03574824e+00 -6.05840862e-01 -2.01919794e-01
5.11819899e-01 -1.17179222e-01 -2.14424878e-01 -9.22687724e-03
-6.73314556e-02 9.32103097e-01 6.77481294e-01 -1.58738112e+00
-1.30245805e-01 -3.81974757e-01 -1.10617116e-01 2.91862220e-01
-5.42256773e-01 1.73907638e-01 -3.10251206e-01 -5.45481861e-01
-1.91418186e-01 -1.19484544e+00 5.07703237e-02 -2.95705646e-01
-5.93285024e-01 -1.42041937e-01 5.18305182e-01 -6.93842173e-01
1.06574106e+00 -1.80726755e+00 1.00607455e+00 -2.74748921e-01
1.88025400e-01 2.06264965e-02 -7.44740725e-01 9.72760022e-01
-1.63049430e-01 3.11592072e-01 -3.71448159e-01 -1.11045337e+00
7.66605020e-01 4.96226043e-01 -3.71639132e-01 2.32129201e-01
4.28565115e-01 1.59949553e+00 -2.48025522e-01 -4.60955679e-01
1.26057602e-02 4.30463731e-01 -1.79678872e-01 -9.27263424e-02
-4.72509712e-01 6.18503153e-01 -2.78220683e-01 8.94010067e-01
6.25813127e-01 -1.90743983e-01 2.56163925e-01 -1.16268583e-01
-8.56592208e-02 1.43399341e-02 -1.97439522e-01 2.00206828e+00
-2.20963582e-01 8.86228263e-01 1.53855085e-01 -9.72313344e-01
5.20350516e-01 5.56317151e-01 3.52082491e-01 -9.70766068e-01
4.27066654e-01 4.86868560e-01 2.14635860e-02 -7.86302388e-01
8.27103138e-01 -4.07853931e-01 -9.57164764e-02 6.54360414e-01
2.09763229e-01 1.03556765e-02 -2.13942211e-02 1.98184460e-01
5.11651754e-01 6.01926088e-01 -3.81715238e-01 7.38198236e-02
2.65094489e-01 3.07307929e-01 -7.25322515e-02 7.78600812e-01
-5.39911762e-02 5.84018409e-01 5.82637429e-01 -2.60718048e-01
-1.22649467e+00 -7.72841215e-01 -8.64577293e-03 1.41814709e+00
-1.57051787e-01 -3.69105458e-01 -7.36509502e-01 -8.21724951e-01
-2.96869487e-01 6.81213856e-01 -7.31972158e-01 -1.74513478e-02
-5.95703363e-01 -1.26695049e+00 9.32065427e-01 3.63356441e-01
1.56456992e-01 -4.58280355e-01 -2.37131063e-02 -1.06658898e-01
-7.79192746e-01 -1.25733650e+00 -4.09952611e-01 1.92752123e-01
-1.10805035e+00 -3.73483390e-01 -9.39780474e-01 -6.42228246e-01
4.84340519e-01 -3.22554223e-02 1.24818289e+00 -1.52016863e-01
4.91926223e-01 6.54236197e-01 -5.32946229e-01 -3.77691865e-01
-6.27905488e-01 4.40303057e-01 -1.42251968e-01 2.57509518e-02
4.85864699e-01 -2.87261128e-01 -1.36970371e-01 2.55443156e-01
-1.04808617e+00 3.71623099e-01 1.03706801e+00 8.27676654e-01
3.69243021e-03 -7.90889323e-01 5.82153499e-01 -6.24584734e-01
8.45327258e-01 -3.92818093e-01 -9.66380164e-03 5.55630624e-01
-6.57193899e-01 2.73355752e-01 3.14068347e-01 -5.16078413e-01
-1.23635471e+00 -2.21361473e-01 5.35474867e-02 -2.11380556e-01
-7.15238005e-02 7.98375547e-01 -2.03864239e-02 -2.17778072e-01
3.14450562e-01 2.90454745e-01 2.10894063e-01 -5.06373167e-01
6.67089343e-01 5.58682144e-01 1.41121089e-01 -8.98150742e-01
7.91386425e-01 2.59115994e-01 1.41506553e-01 -4.55171108e-01
-4.37789917e-01 1.01227753e-01 -6.88894868e-01 -1.70623526e-01
9.26142097e-01 -8.32644105e-01 -6.57909691e-01 -1.55259399e-02
-1.38730204e+00 -1.53692693e-01 1.45088002e-01 5.23698092e-01
-7.07092583e-01 6.30078793e-01 -7.83115685e-01 -9.20750558e-01
-3.13499242e-01 -1.40008605e+00 1.55341840e+00 -1.82245940e-01
-3.42359155e-01 -1.34201539e+00 2.24783912e-01 9.24118042e-01
6.20208263e-01 -1.29723847e-01 1.42981851e+00 -6.36487842e-01
-6.81988299e-01 -6.84040189e-02 -4.06814903e-01 1.12932518e-01
-1.95782304e-01 -2.15953201e-01 -1.03486001e+00 -4.64022830e-02
-2.50495762e-01 -3.70002419e-01 8.63683939e-01 4.27670144e-02
1.68998793e-01 -3.18645716e-01 -2.45502219e-02 3.82436842e-01
1.32162380e+00 -1.66123629e-01 5.18552601e-01 2.18581274e-01
6.91908836e-01 6.96295679e-01 3.09176296e-01 3.27913114e-03
7.83767879e-01 8.60023260e-01 5.75706542e-01 -4.61249530e-01
-1.83542505e-01 -1.28470019e-01 8.25409234e-01 1.21596360e+00
-4.83418524e-01 -4.52675730e-01 -8.54327619e-01 3.78555566e-01
-2.29222965e+00 -6.38033509e-01 -4.01114911e-01 1.94065082e+00
8.59091401e-01 -3.92632723e-01 -2.08702404e-03 -4.75969702e-01
2.44994998e-01 -1.40231729e-01 -5.99771040e-03 -8.89864564e-01
-7.37302780e-01 1.19001642e-01 2.84772485e-01 5.69103181e-01
-8.74222457e-01 1.03567839e+00 6.85756779e+00 6.89147472e-01
-1.04111540e+00 2.67669946e-01 5.86054087e-01 -7.12068975e-02
-7.19379306e-01 2.76175011e-02 -3.88223797e-01 1.20501202e-02
1.17412627e+00 2.73257077e-01 6.90136850e-01 1.17474891e-01
2.75479674e-01 -9.03077871e-02 -1.51214731e+00 7.43836403e-01
3.65740627e-01 -1.07173455e+00 5.46923518e-01 4.74824846e-01
7.47068584e-01 5.36744706e-02 6.09166265e-01 4.72793043e-01
-2.94768140e-02 -1.20707667e+00 7.96363056e-01 6.83419466e-01
6.59139514e-01 -4.77659404e-01 9.30622041e-01 3.00329149e-01
-5.98516345e-01 8.20120871e-02 -1.19561002e-01 -1.24466889e-01
9.18869749e-02 9.17715505e-02 -5.96656442e-01 1.10723448e+00
1.29290164e-01 6.13667309e-01 -8.93898547e-01 3.58904302e-01
1.26990542e-01 4.25533950e-01 5.18343821e-02 -5.35019767e-03
4.89376336e-01 -5.80386937e-01 7.08453298e-01 1.29372096e+00
3.08210671e-01 -4.74123836e-01 -2.39999279e-01 9.59624767e-01
-2.80803472e-01 3.67078453e-01 -8.71408343e-01 -5.44622540e-01
-3.68703157e-01 1.07454467e+00 -1.61645964e-01 -2.58469880e-01
-7.14684844e-01 1.31784701e+00 2.78986573e-01 7.21743107e-01
-9.62111533e-01 1.30357549e-01 3.17290515e-01 -4.52636391e-01
8.10461771e-03 -5.57156205e-01 -8.15528691e-01 -1.73392665e+00
1.71554402e-01 -1.20273805e+00 5.82750961e-02 -1.02923083e+00
-1.35590231e+00 6.28816783e-01 -2.32378915e-02 -1.16792786e+00
-2.98609823e-01 -8.57238591e-01 -5.62043861e-02 8.89891624e-01
-1.45663702e+00 -1.85649025e+00 4.71891195e-01 4.60178733e-01
5.17780125e-01 -1.20598987e-01 1.14160168e+00 3.79378647e-01
-6.36144280e-01 8.78103971e-01 2.76545495e-01 1.56485084e-02
1.13736272e+00 -1.12459588e+00 -6.59927949e-02 5.54680645e-01
3.11887681e-01 9.34922159e-01 8.28976512e-01 -4.48625982e-01
-2.24265194e+00 -5.39308131e-01 1.51330781e+00 -1.27927411e+00
7.43356764e-01 -4.16764736e-01 -3.40935200e-01 1.15512633e+00
1.32960021e+00 -8.60264122e-01 8.25074017e-01 1.69561759e-01
-4.88260835e-01 3.29326510e-01 -7.68147826e-01 9.30970490e-01
8.49117637e-01 -6.77152157e-01 -7.19351828e-01 2.92369604e-01
8.17724168e-01 -5.11531942e-02 -1.05638707e+00 4.85881925e-01
6.91079497e-01 -7.37779260e-01 7.55148947e-01 -9.26215112e-01
8.88062060e-01 -2.70873401e-02 -6.05968654e-01 -1.25407934e+00
2.25865349e-01 -8.03052127e-01 1.36205480e-01 1.06200242e+00
1.14954567e+00 -6.45211160e-01 4.12600845e-01 7.19703615e-01
-2.07928568e-01 -4.81063992e-01 -1.32112515e+00 -3.96682352e-01
6.90824270e-01 -5.57478547e-01 2.87154138e-01 1.16172683e+00
4.54419553e-01 7.11315513e-01 -7.71778286e-01 -1.76702976e-01
5.76432765e-01 8.22822824e-02 7.87723362e-01 -8.52999508e-01
-2.96103805e-01 -6.42289937e-01 2.17822045e-01 -1.03481185e+00
3.17515671e-01 -1.08223093e+00 -2.55622417e-01 -1.44370854e+00
5.63978434e-01 2.42703304e-01 1.57879144e-01 6.83103442e-01
3.32793407e-02 6.96264446e-01 4.75631565e-01 2.91758090e-01
-8.36352468e-01 7.26417780e-01 1.63282514e+00 -3.94647062e-01
1.64760321e-01 -2.64205456e-01 -8.08662474e-01 2.86312580e-01
5.23150682e-01 -1.95515111e-01 -9.35800672e-02 -1.25190961e+00
7.54240870e-01 2.63048112e-01 3.74052435e-01 -2.19769210e-01
8.71350169e-02 -1.26584068e-01 2.13157520e-01 -3.77812088e-01
6.78487897e-01 -9.96535599e-01 4.25722711e-02 -3.05550154e-02
-5.82808912e-01 4.66329187e-01 4.08818662e-01 3.03329945e-01
-2.18477517e-01 7.81176612e-02 1.52280107e-01 -4.47859615e-02
-9.28946957e-03 -1.79544315e-01 -8.42335403e-01 -4.48157191e-01
2.61992097e-01 -1.92711521e-02 -7.37741053e-01 -8.68392885e-01
-9.05698180e-01 4.62278873e-02 4.33887929e-01 6.61903322e-01
3.71058911e-01 -1.42342889e+00 -8.55713129e-01 -1.20286070e-01
3.43434215e-01 -7.45576203e-01 -1.32077470e-01 1.70755064e+00
2.13466175e-02 8.74030292e-01 1.85888875e-02 -5.91325521e-01
-1.10016799e+00 3.55034977e-01 3.51967305e-01 -1.70131490e-01
1.30608128e-02 2.78608769e-01 4.51674918e-03 -8.99738610e-01
2.65577901e-02 -2.51779765e-01 2.63268948e-01 6.43027276e-02
-6.35428950e-02 2.38612503e-01 1.91670097e-03 -8.90019298e-01
-1.51037589e-01 7.40651071e-01 5.35089932e-02 -7.45720148e-01
1.14492118e+00 -5.29590726e-01 -6.18394673e-01 5.62717974e-01
1.02004969e+00 3.15678567e-02 -3.70806903e-01 -2.81810760e-01
6.60039559e-02 -3.94922160e-02 -4.91451979e-01 -1.45199776e+00
-7.96468258e-01 1.07902133e+00 4.76898432e-01 9.29552242e-02
1.06944466e+00 6.48298562e-02 7.04468548e-01 6.47328675e-01
2.02264607e-01 -9.08438325e-01 -1.43388733e-01 7.01142311e-01
9.18412030e-01 -1.54348886e+00 -2.99849212e-01 -1.10324308e-01
-9.56912637e-01 1.32486439e+00 2.89473385e-01 5.05939066e-01
-7.00614378e-02 2.68486068e-02 3.69765759e-01 -1.04050450e-01
-1.00531697e+00 -3.00414264e-01 4.99374926e-01 5.03296375e-01
7.97145188e-01 1.26761526e-01 -6.00239694e-01 6.06500506e-01
-1.62192583e-02 -2.27215603e-01 2.07562715e-01 6.77346706e-01
-7.26617873e-02 -1.61870456e+00 -5.45782506e-01 -1.69896170e-01
-3.15854490e-01 -4.48867023e-01 -1.28561878e+00 1.00722384e+00
3.94072160e-02 1.29661250e+00 -4.97202218e-01 -5.94306409e-01
1.61973625e-01 6.60434604e-01 7.22641289e-01 -9.80873331e-02
-8.71698201e-01 3.46097142e-01 5.27541459e-01 -3.19044977e-01
-6.28866613e-01 -6.17791533e-01 -7.80385613e-01 -3.60539705e-01
-1.70090154e-01 3.04549327e-03 9.33492303e-01 1.19417298e+00
4.70250189e-01 1.98400751e-01 2.02676296e-01 -8.86496007e-01
-4.65977371e-01 -1.25579715e+00 1.51159823e-01 3.65357161e-01
4.72971022e-01 -1.48269966e-01 -2.54719257e-01 2.00671241e-01] | [11.466684341430664, 1.5101838111877441] |
77140edc-e5a3-4c27-8064-3b3b01a83253 | cells-a-parallel-corpus-for-biomedical-lay | 2211.03818 | null | https://arxiv.org/abs/2211.03818v1 | https://arxiv.org/pdf/2211.03818v1.pdf | CELLS: A Parallel Corpus for Biomedical Lay Language Generation | Recent lay language generation systems have used Transformer models trained on a parallel corpus to increase health information accessibility. However, the applicability of these models is constrained by the limited size and topical breadth of available corpora. We introduce CELLS, the largest (63k pairs) and broadest-ranging (12 journals) parallel corpus for lay language generation. The abstract and the corresponding lay language summary are written by domain experts, assuring the quality of our dataset. Furthermore, qualitative evaluation of expert-authored plain language summaries has revealed background explanation as a key strategy to increase accessibility. Such explanation is challenging for neural models to generate because it goes beyond simplification by adding content absent from the source. We derive two specialized paired corpora from CELLS to address key challenges in lay language generation: generating background explanations and simplifying the original abstract. We adopt retrieval-augmented models as an intuitive fit for the task of background explanation generation, and show improvements in summary quality and simplicity while maintaining factual correctness. Taken together, this work presents the first comprehensive study of background explanation for lay language generation, paving the path for disseminating scientific knowledge to a broader audience. CELLS is publicly available at: https://github.com/LinguisticAnomalies/pls_retrieval. | ['Trevor Cohen', 'Sheng Wang', 'Gondy Leroy', 'Wei Qiu', 'Yue Guo'] | 2022-11-07 | null | null | null | null | ['explanation-generation'] | ['natural-language-processing'] | [ 2.99987406e-01 9.22692418e-01 -7.22509086e-01 1.81854963e-02
-1.39136565e+00 -5.52887678e-01 6.85909808e-01 5.99886656e-01
-1.95643172e-01 1.21227229e+00 1.03600013e+00 -5.67612886e-01
-2.08107874e-01 -6.08652771e-01 -8.30737472e-01 -1.02455042e-01
5.27001381e-01 5.57968080e-01 -3.23919803e-01 -3.62826139e-01
5.28436422e-01 1.17982859e-02 -1.04900515e+00 7.30425656e-01
1.50538802e+00 1.32501215e-01 2.95556962e-01 6.57356441e-01
-2.38149047e-01 5.49873590e-01 -7.90316164e-01 -7.23988235e-01
-3.16719472e-01 -7.92772889e-01 -9.84877765e-01 -1.36008456e-01
5.76934993e-01 -2.98433959e-01 -2.10932598e-01 5.14630854e-01
7.64022946e-01 -1.18531674e-01 9.09955382e-01 -9.37654972e-01
-1.58879542e+00 1.16809821e+00 -9.84146073e-02 2.12285206e-01
6.50675654e-01 -3.72317843e-02 1.23578000e+00 -8.91762495e-01
8.32573771e-01 9.88851726e-01 4.14555818e-01 1.15247643e+00
-1.04344356e+00 -6.19366348e-01 2.53298134e-01 1.06836334e-02
-9.84689176e-01 -6.19118035e-01 4.82565165e-01 -3.50966990e-01
1.07508266e+00 3.66236806e-01 8.40297759e-01 1.68377662e+00
2.63938755e-01 8.44198048e-01 7.69366801e-01 -7.13538170e-01
5.13728745e-02 2.04147592e-01 2.30107814e-01 5.23654580e-01
8.02864194e-01 -2.41438076e-01 -8.70476961e-01 -2.03552827e-01
5.59836745e-01 -3.43120694e-01 -6.84626877e-01 3.81011993e-01
-1.51532125e+00 9.98198867e-01 3.33309352e-01 7.45521411e-02
-4.60183233e-01 -5.25301788e-03 2.43235812e-01 -5.34389652e-02
5.04013240e-01 9.66885746e-01 -3.78044158e-01 6.01848550e-02
-1.07489061e+00 5.97255051e-01 8.66357088e-01 1.09757710e+00
-4.74075861e-02 -8.90797228e-02 -5.58531880e-01 7.62499690e-01
2.07712978e-01 6.09710097e-01 4.04583395e-01 -9.04611230e-01
5.36304653e-01 5.42080343e-01 -6.80581033e-02 -1.03666270e+00
-1.78180993e-01 -6.71165407e-01 -8.12714159e-01 -4.63750869e-01
4.33249176e-01 -5.30128814e-02 -4.96064931e-01 1.83797562e+00
-7.19301403e-02 -4.29448009e-01 2.35986590e-01 6.24155104e-01
1.54702747e+00 6.16455197e-01 3.26538801e-01 -1.09038182e-01
1.60154486e+00 -1.00667858e+00 -9.59342301e-01 -3.54396284e-01
5.49145043e-01 -8.58031034e-01 1.11145616e+00 3.21238190e-01
-1.39962912e+00 -3.39959770e-01 -9.42665815e-01 -4.66269702e-01
-4.26919371e-01 4.93795097e-01 7.31078982e-01 2.80160040e-01
-1.15607047e+00 3.20077240e-01 -4.46728587e-01 -5.29715180e-01
5.51985502e-01 9.43861902e-02 -3.26579928e-01 -5.08484803e-02
-1.51273763e+00 8.78751814e-01 1.97980270e-01 -9.44675282e-02
-3.77655834e-01 -1.02818179e+00 -9.77026641e-01 -2.91733928e-02
3.02769542e-01 -1.42492211e+00 1.30562747e+00 -4.45856899e-01
-1.12422025e+00 8.48039031e-01 -3.04472655e-01 -3.78781796e-01
3.57692003e-01 -3.14974040e-01 -2.30344683e-01 2.52043486e-01
5.66078544e-01 9.78908360e-01 2.88584858e-01 -1.02839017e+00
-2.49123469e-01 -3.94697394e-03 -3.87867726e-02 2.89825082e-01
-2.00690970e-01 -1.63307711e-01 -3.92626166e-01 -1.00305200e+00
-2.76370227e-01 -5.48821449e-01 -2.31207475e-01 -1.53988034e-01
-7.55402923e-01 -2.49288321e-01 -2.29376078e-01 -1.10778904e+00
1.51353347e+00 -1.59580946e+00 6.77377582e-02 -2.21512854e-01
4.88387495e-01 -9.03699175e-02 -1.74139142e-01 8.25269163e-01
-9.23566967e-02 7.58299112e-01 -1.08392924e-01 -1.49446294e-01
-6.76985597e-03 -1.69170946e-01 -7.02796757e-01 -2.11695004e-02
4.46315110e-01 1.28088284e+00 -1.10590231e+00 -6.06636286e-01
-3.35122019e-01 4.80656266e-01 -8.35738659e-01 -1.43657662e-02
-3.20326298e-01 3.56207252e-01 -4.96868789e-01 5.28377712e-01
1.32205322e-01 -5.94763219e-01 -1.99065626e-01 -8.74580815e-02
-7.15848580e-02 8.13701808e-01 -4.54896182e-01 1.80690646e+00
-4.07837182e-01 6.18449450e-01 -4.57947403e-01 -4.61283177e-01
7.24460483e-01 5.33160627e-01 8.48840773e-02 -5.18957436e-01
-1.55271962e-01 4.79365647e-01 4.28051166e-02 -2.88523793e-01
7.97346890e-01 -1.68526575e-01 -1.19011998e-01 5.83622873e-01
-1.12194091e-01 -3.98808926e-01 5.29067397e-01 7.40301847e-01
8.09379220e-01 4.25090604e-02 5.74091256e-01 -3.49404126e-01
1.24948308e-01 3.52244347e-01 1.99507371e-01 1.07726824e+00
4.18343633e-01 8.93773019e-01 4.87432390e-01 -6.51584566e-02
-9.94113386e-01 -9.99007821e-01 -7.45530650e-02 4.87388402e-01
-2.55031049e-01 -7.24318147e-01 -8.09243083e-01 -5.25186300e-01
-1.69147581e-01 1.36278975e+00 -6.77258670e-01 -2.18990520e-01
-4.85430568e-01 -6.64558232e-01 7.62165904e-01 4.64477211e-01
1.10471226e-01 -1.07354259e+00 -4.35795814e-01 2.77341813e-01
-7.76027143e-01 -8.48935664e-01 -6.93016768e-01 -2.61627465e-01
-7.56254673e-01 -9.30595517e-01 -1.17200971e+00 -5.46234906e-01
8.22935581e-01 -1.40189203e-02 1.34939086e+00 4.49744105e-01
-1.32342383e-01 3.19090962e-01 -2.16313347e-01 -8.16654027e-01
-6.78406417e-01 3.32238227e-01 -1.86004147e-01 -8.74183953e-01
9.06161889e-02 -1.79290958e-02 -6.53930783e-01 -3.07061940e-01
-9.09045160e-01 6.49742544e-01 7.63103306e-01 1.09596336e+00
6.50695443e-01 -5.44980824e-01 9.47072446e-01 -9.88831401e-01
1.24911106e+00 -7.34155416e-01 3.31465192e-02 3.30707610e-01
-7.57758975e-01 1.10956982e-01 4.63135362e-01 -3.22447151e-01
-1.11590385e+00 -7.17758596e-01 -1.94555223e-01 4.30701435e-01
5.00205718e-02 7.70405293e-01 9.45685282e-02 6.42822862e-01
9.46489394e-01 1.88610002e-01 1.44310579e-01 -2.55707681e-01
5.14242411e-01 4.60474819e-01 5.46915233e-01 -7.75029480e-01
4.47751492e-01 8.17662179e-02 -3.06824297e-01 -7.15905666e-01
-1.06000125e+00 -1.03447124e-01 -2.70069122e-01 -3.42300572e-02
9.16355431e-01 -8.92780840e-01 -3.33661735e-01 -8.39083269e-02
-1.51589751e+00 -3.68823230e-01 -3.05903494e-01 4.25338805e-01
-4.03761059e-01 3.36776108e-01 -6.73191607e-01 -2.99265921e-01
-7.46790588e-01 -9.62492466e-01 1.20293677e+00 2.09715128e-01
-1.01216805e+00 -1.11744571e+00 3.25033208e-03 6.79545999e-01
2.58372456e-01 1.90871581e-01 1.29321301e+00 -9.30955291e-01
-4.72668082e-01 -1.70522988e-01 -1.19640402e-01 -1.32789269e-01
1.55419722e-01 1.00625664e-01 -5.33344269e-01 3.04581225e-01
-3.13733160e-01 -3.17388803e-01 8.09737146e-01 5.81884682e-01
1.14218378e+00 -6.61206305e-01 -4.37165409e-01 1.55571625e-01
7.99812317e-01 -4.45063189e-02 4.97168839e-01 2.85047948e-01
5.37550509e-01 7.39104211e-01 3.23762923e-01 2.47403219e-01
6.85631931e-01 5.46678424e-01 -2.90955573e-01 -3.70917112e-01
-5.51162660e-01 -6.15889549e-01 1.29307568e-01 1.23655570e+00
3.51705439e-02 -7.16315806e-01 -9.77189064e-01 7.01294243e-01
-1.69498467e+00 -1.08657587e+00 -2.44804710e-01 1.88816011e+00
1.12690151e+00 5.17974868e-02 -2.00502947e-01 -2.27420077e-01
2.78111845e-01 -1.78808242e-01 -2.88485587e-01 -4.44798410e-01
-3.95449907e-01 1.48977429e-01 1.76119998e-01 6.81613266e-01
-4.57976758e-01 7.54575014e-01 6.35866690e+00 8.49439740e-01
-6.67917669e-01 -8.43491256e-02 8.07966411e-01 -4.73915786e-02
-1.16591609e+00 -1.07419103e-01 -1.01435840e+00 3.48932236e-01
9.23300028e-01 -6.41223729e-01 7.09692463e-02 4.20261800e-01
4.59951252e-01 1.41903818e-01 -1.23615491e+00 6.26695931e-01
3.30509692e-01 -1.88263226e+00 8.01290631e-01 7.76450559e-02
6.87668324e-01 -5.40386498e-01 1.81907788e-01 3.03175360e-01
1.00741021e-01 -1.34363544e+00 8.79088879e-01 7.71984935e-01
8.67376745e-01 -3.54357362e-01 7.93200135e-01 1.92554891e-01
-7.13603377e-01 3.37244838e-01 -1.65311933e-01 4.39220928e-02
4.20613110e-01 5.58330119e-01 -9.70898151e-01 5.80501437e-01
2.44177043e-01 6.44006908e-01 -6.69079244e-01 7.49324799e-01
-6.03050888e-01 6.09321356e-01 8.68685097e-02 -2.72302270e-01
1.37759522e-01 2.75121123e-01 4.47204947e-01 1.32097936e+00
5.13363540e-01 3.80722880e-01 -1.70510873e-01 1.33351815e+00
-3.25534403e-01 4.27607089e-01 -7.38885641e-01 -5.23806512e-01
5.88047743e-01 9.18452799e-01 -6.13588452e-01 -4.48333889e-01
-3.01778495e-01 8.37137282e-01 3.30825388e-01 3.01193595e-01
-6.87168777e-01 -2.87263662e-01 2.93335430e-02 2.07970217e-01
-3.71044159e-01 2.26549312e-01 -8.18081379e-01 -1.21330225e+00
-1.25065252e-01 -1.29202354e+00 5.29540181e-01 -1.00538421e+00
-1.22161674e+00 5.53786635e-01 3.13611180e-01 -8.49923909e-01
-4.29101080e-01 -5.59582174e-01 -4.44372177e-01 1.07773530e+00
-1.26221073e+00 -1.24629450e+00 -6.17376901e-02 -3.75958644e-02
1.10483706e+00 -2.41890222e-01 8.37223530e-01 5.87327182e-02
-4.99779522e-01 6.64130926e-01 -3.77680689e-01 -1.65156677e-01
8.46302688e-01 -1.26212025e+00 4.90698487e-01 7.08936632e-01
-8.83021578e-03 1.32812238e+00 7.44803429e-01 -1.11190343e+00
-9.70590472e-01 -7.47225285e-01 1.57648790e+00 -8.83001864e-01
5.37576556e-01 -1.49418727e-01 -8.86575162e-01 5.35337389e-01
5.68613768e-01 -1.01212215e+00 1.02397156e+00 9.16872546e-02
5.70161976e-02 4.12684292e-01 -7.08103836e-01 1.35274959e+00
9.52653468e-01 -3.10196847e-01 -9.10649598e-01 5.69921613e-01
1.03378201e+00 -3.95749986e-01 -8.16148520e-01 2.03826457e-01
5.15354216e-01 -4.46143746e-01 1.00333929e+00 -6.38026178e-01
1.34370685e+00 -6.06477745e-02 3.31228703e-01 -1.36863661e+00
-4.18997794e-01 -5.63366771e-01 8.06204081e-02 1.17949080e+00
1.02503407e+00 -4.13746923e-01 4.07786340e-01 8.45883131e-01
-5.28047204e-01 -1.08033609e+00 -3.67611289e-01 -2.02826425e-01
3.95793438e-01 -2.27601081e-01 6.76319063e-01 7.65855551e-01
3.64634931e-01 5.98440886e-01 -1.84144527e-01 -2.70940244e-01
5.14688611e-01 -1.13702744e-01 5.74905634e-01 -8.32998097e-01
-3.79245192e-01 -7.71718383e-01 5.36227167e-01 -1.02499902e+00
2.40370736e-01 -1.27499247e+00 -1.44348085e-01 -2.26766467e+00
5.97817004e-01 -3.07226032e-02 2.49907464e-01 5.04465520e-01
-6.55515909e-01 4.69537154e-02 -1.47527400e-02 1.62457407e-01
-1.73164740e-01 4.94686961e-01 1.49913454e+00 -9.82746109e-02
-1.15975648e-01 -2.01785862e-01 -1.64358044e+00 4.27092463e-01
9.57016349e-01 -3.18473876e-01 -6.17076099e-01 -6.38377428e-01
4.82838660e-01 8.23348388e-02 3.65868092e-01 -3.66994113e-01
1.28114372e-01 -2.68644184e-01 4.66254324e-01 -4.99886423e-01
5.72334491e-02 -8.07851106e-02 1.88899919e-01 5.60992002e-01
-1.06830800e+00 3.98625195e-01 4.67632711e-01 3.61817002e-01
9.12798103e-03 -2.17267737e-01 1.46380916e-01 -4.07779396e-01
-1.86319098e-01 8.50508586e-02 -4.40762311e-01 3.35053265e-01
3.51130128e-01 -1.25286967e-01 -8.36170733e-01 -6.80285931e-01
-4.35925066e-01 1.20947354e-01 3.00279915e-01 3.95115614e-01
8.48766387e-01 -1.17163229e+00 -1.19213307e+00 -2.28830934e-01
2.38393426e-01 -9.24510509e-02 2.18939885e-01 8.92249882e-01
-4.87165272e-01 7.40720868e-01 -1.04766436e-01 4.13366780e-02
-9.01898086e-01 3.75531018e-01 -1.28985807e-01 -3.25737864e-01
-6.72299206e-01 6.91069484e-01 4.24423754e-01 -4.18159097e-01
2.19838731e-02 -3.99442405e-01 -3.56901199e-01 2.04670653e-02
7.47160792e-01 4.78894860e-01 -1.37395769e-01 -2.17208251e-01
-9.84644443e-02 4.59131487e-02 -2.26613387e-01 -3.77994180e-01
1.12737501e+00 -1.16438121e-01 -1.45153016e-01 3.68700713e-01
5.75588226e-01 5.05898833e-01 -5.00082791e-01 1.89228699e-01
-7.61507079e-02 5.07245015e-04 -2.17859149e-01 -1.17783999e+00
-4.87745911e-01 7.74944723e-01 -3.43074769e-01 -1.76378906e-01
8.13643932e-01 1.59925520e-01 8.18688214e-01 1.93615079e-01
-1.51231706e-01 -8.74464512e-01 2.31346890e-01 3.06595325e-01
1.46367943e+00 -1.03502595e+00 2.21831143e-01 -4.49776858e-01
-8.91875565e-01 1.06970692e+00 6.10777438e-01 4.56167251e-01
5.79060540e-02 -2.12105494e-02 3.35674137e-02 -4.77167398e-01
-9.00366187e-01 2.25210801e-01 8.50700736e-01 5.28777003e-01
1.00058043e+00 -3.24728452e-02 -9.07937407e-01 1.20846951e+00
-6.64081156e-01 -1.23896442e-01 7.04200923e-01 3.53464842e-01
-3.99824828e-02 -1.09736001e+00 -2.64003932e-01 6.81136131e-01
-6.89335406e-01 -6.63425863e-01 -7.12242961e-01 8.72095883e-01
-1.30762830e-01 1.08108342e+00 -2.78526485e-01 2.36066207e-01
2.08909377e-01 5.69883792e-04 3.91136706e-01 -8.67346048e-01
-6.54498518e-01 5.83709367e-02 4.17086899e-01 -3.26839536e-02
-1.72704548e-01 -5.38029432e-01 -1.21254587e+00 -3.30553859e-01
-3.94940078e-02 3.71621817e-01 3.26956660e-01 7.87679732e-01
4.77534145e-01 7.63534546e-01 -1.77776918e-01 -4.00461614e-01
-3.76162589e-01 -8.50370109e-01 -1.69369932e-02 3.66583735e-01
1.39344513e-01 -2.57355809e-01 -1.92616120e-01 3.14855725e-01] | [11.986271858215332, 9.209197044372559] |
5644f6df-4c99-4f8b-a798-6ace57d6f289 | vilas-integrating-vision-and-language-into | 2305.19972 | null | https://arxiv.org/abs/2305.19972v1 | https://arxiv.org/pdf/2305.19972v1.pdf | ViLaS: Integrating Vision and Language into Automatic Speech Recognition | Employing additional multimodal information to improve automatic speech recognition (ASR) performance has been proven effective in previous works. However, many of these works focus only on the utilization of visual cues from human lip motion. In fact, context-dependent visual and linguistic cues can also be used to improve ASR performance in many scenarios. In this paper, we first propose a multimodal ASR model (ViLaS) that can simultaneously or separately integrate visual and linguistic cues to help recognize the input speech, and introduce a training strategy that can improve performance in modal-incomplete test scenarios. Then, we create a multimodal ASR dataset (VSDial) with visual and linguistic cues to explore the effects of integrating vision and language. Finally, we report empirical results on the public Flickr8K and self-constructed VSDial datasets, investigate cross-modal fusion schemes, and analyze fine-grained cross-modal alignment on VSDial. | ['Bo Xu', 'Shuang Xu', 'Jing Shi', 'Linghui Meng', 'Ziyi Ni', 'Feilong Chen', 'Minglun Han'] | 2023-05-31 | null | null | null | null | ['automatic-speech-recognition'] | ['speech'] | [ 1.91377461e-01 -3.28035742e-01 -2.64706820e-01 -3.82938117e-01
-1.37079775e+00 -7.46631205e-01 9.03276920e-01 -1.82604060e-01
-4.09669012e-01 2.17161864e-01 5.69270015e-01 -3.36301863e-01
1.39345080e-01 -9.24663320e-02 -3.14185649e-01 -6.78343832e-01
4.99803126e-01 8.50193575e-02 1.53974518e-01 -3.48502815e-01
1.21860795e-01 4.51344430e-01 -1.71405876e+00 6.73457384e-01
7.31398463e-01 8.35117638e-01 2.91426718e-01 9.29606616e-01
-8.33820999e-02 6.02005422e-01 -6.21365905e-01 -3.05987209e-01
-3.78131233e-02 -3.05525571e-01 -6.32002532e-01 3.43703508e-01
5.72688639e-01 5.05405106e-02 -5.09901702e-01 7.49788344e-01
1.00482857e+00 3.13873053e-01 6.44123435e-01 -1.16695404e+00
-6.24492526e-01 5.94200850e-01 -6.08334243e-01 4.11450192e-02
7.40639269e-01 2.75110155e-01 9.21401739e-01 -1.05606031e+00
4.49548602e-01 1.41475332e+00 2.48423815e-01 7.95751393e-01
-1.15912950e+00 -4.20415491e-01 1.16028860e-01 4.44122881e-01
-1.47543120e+00 -1.35531628e+00 9.08089280e-01 -1.55171067e-01
9.32987988e-01 4.31617022e-01 1.32726252e-01 1.26221681e+00
-6.43101931e-01 1.36739230e+00 1.14430857e+00 -6.13393307e-01
-8.06929171e-02 1.42475560e-01 9.45338085e-02 4.13561732e-01
-4.48184460e-01 1.47207484e-01 -8.80029976e-01 6.02663904e-02
1.66175455e-01 -4.65140730e-01 -5.65105796e-01 -9.10898969e-02
-1.42307603e+00 7.11615980e-01 1.36075392e-01 4.37100261e-01
-2.25914255e-01 -1.45282358e-01 4.63616848e-01 1.77847192e-01
2.85294503e-01 1.27826944e-01 -2.20672339e-01 -2.00865552e-01
-1.03339612e+00 -2.21113294e-01 2.24495918e-01 5.81312835e-01
5.93217790e-01 1.99572191e-01 -5.64867854e-01 1.55031872e+00
6.08831406e-01 9.31136250e-01 5.58439016e-01 -9.91492033e-01
6.73963547e-01 3.47795099e-01 -1.76092654e-01 -6.09974444e-01
-2.75700182e-01 1.08521312e-01 -5.58054507e-01 -2.21193224e-01
2.06474319e-01 1.06776901e-01 -1.18328846e+00 1.61545086e+00
6.17194623e-02 2.02791039e-02 4.06569123e-01 1.02325857e+00
1.32763577e+00 5.09515285e-01 2.78275404e-02 -2.23207116e-01
1.30926478e+00 -9.70787406e-01 -9.46315944e-01 -1.31584957e-01
6.69633210e-01 -9.46920514e-01 1.23275077e+00 7.75847733e-02
-1.00803936e+00 -5.25318801e-01 -6.70019150e-01 3.07621830e-03
-4.00987893e-01 6.57518208e-01 2.20020115e-01 7.52999604e-01
-1.35201275e+00 -4.17676657e-01 -6.48466170e-01 -4.45716649e-01
1.49295747e-01 1.01616032e-01 -5.08964300e-01 -4.26199466e-01
-1.20157683e+00 7.76183486e-01 2.34219447e-01 7.25904945e-03
-8.17252934e-01 -3.07864308e-01 -1.20102835e+00 -2.31730938e-01
4.43626851e-01 -4.77573484e-01 1.19663262e+00 -8.14967811e-01
-1.69066882e+00 9.90841925e-01 -7.37930477e-01 -3.14809054e-01
3.20010215e-01 1.51944488e-01 -5.34885406e-01 3.70435029e-01
-2.99605489e-01 1.07080007e+00 7.79926360e-01 -1.57548642e+00
-4.80829626e-01 -4.46587145e-01 9.73224565e-02 4.01670873e-01
-1.87799409e-01 1.75158814e-01 -9.05175924e-01 -7.61755943e-01
1.05582573e-01 -1.07524431e+00 1.28717631e-01 -5.12511969e-01
-5.38327575e-01 -4.40960288e-01 8.16755116e-01 -9.30948615e-01
1.14922941e+00 -2.43930435e+00 3.39478284e-01 1.51562179e-03
-2.31886461e-01 5.74534774e-01 -5.30385554e-01 3.66726398e-01
2.04514101e-01 6.90707415e-02 -2.30448827e-01 -7.29041040e-01
1.35829508e-01 1.96969852e-01 -2.14352086e-01 2.69142061e-01
1.86285853e-01 1.09668171e+00 -4.83659804e-01 -6.07508898e-01
5.75339258e-01 8.59425008e-01 -3.56146991e-01 8.95786360e-02
-1.15457781e-01 2.79398084e-01 -4.78451252e-02 9.50861931e-01
5.17391980e-01 -2.05975808e-02 -4.19078767e-03 -4.36959833e-01
5.38236052e-02 1.97799295e-01 -8.43915284e-01 1.70021117e+00
-6.66662097e-01 9.53709841e-01 2.48868540e-01 -8.15326095e-01
7.57623613e-01 4.67814296e-01 2.86078721e-01 -9.57584679e-01
9.95613188e-02 4.17000148e-03 -9.67992842e-02 -5.78052402e-01
6.05258644e-01 1.71922475e-01 5.84101602e-02 1.29574701e-01
-3.70631590e-02 1.14942808e-02 1.34746343e-01 1.86631039e-01
9.10133839e-01 -2.18260273e-01 6.83845282e-02 5.23896337e-01
1.03239083e+00 -1.78666487e-01 5.17851189e-02 6.08767688e-01
-4.67650950e-01 8.25564802e-01 -5.26507497e-02 4.65085238e-01
-6.40452027e-01 -1.01320338e+00 -5.93206547e-02 1.33126843e+00
1.42336830e-01 -3.65386218e-01 -5.08480489e-01 -7.28221834e-01
-1.66992947e-01 8.60462248e-01 -2.37263858e-01 -6.21614195e-02
-3.26164871e-01 -5.27543366e-01 9.83822703e-01 5.08597553e-01
4.85738009e-01 -9.95394647e-01 -7.37110898e-02 -2.69590914e-01
-7.13309765e-01 -1.51981664e+00 -4.49069530e-01 -1.81086317e-01
-1.72567278e-01 -8.01872313e-01 -1.02634096e+00 -5.98397255e-01
3.33502412e-01 7.45945096e-01 8.08930576e-01 -1.57830477e-01
-3.74840461e-02 9.56828654e-01 -7.30485082e-01 5.52101023e-02
-6.36256039e-01 2.21088082e-02 1.42769236e-02 2.55024433e-01
2.52308697e-01 1.37745112e-03 -2.76115805e-01 5.82912385e-01
-8.40844870e-01 4.41642441e-02 5.54582179e-01 8.43909681e-01
4.09064800e-01 -2.79123247e-01 6.87838554e-01 -8.62482563e-02
6.28819287e-01 -1.55250907e-01 -7.76411518e-02 5.93281031e-01
-6.66355118e-02 2.04482488e-02 2.04127461e-01 -6.66888952e-01
-1.10510898e+00 1.33315116e-01 -4.38569099e-01 -7.21125603e-01
-4.87824529e-01 5.63530147e-01 -5.14000654e-01 -8.64232779e-02
2.37924770e-01 5.56606174e-01 5.18907532e-02 -4.71296191e-01
7.01254427e-01 1.27039051e+00 4.99982595e-01 -4.74821955e-01
3.81804347e-01 4.08585459e-01 -3.40940267e-01 -1.32794702e+00
-4.01113778e-01 -7.34952033e-01 -3.89144897e-01 -2.86336869e-01
8.05676579e-01 -1.09549236e+00 -6.88902557e-01 4.92952228e-01
-9.82783198e-01 -2.98509300e-01 1.86820537e-01 5.58816791e-01
-5.71127594e-01 5.28636336e-01 -4.35760736e-01 -1.05604160e+00
-1.73745696e-02 -1.49948263e+00 1.50599003e+00 1.24390446e-01
-4.70795557e-02 -8.35443616e-01 3.70035134e-02 9.68049169e-01
3.13310355e-01 -4.83348906e-01 4.60390329e-01 -7.89103925e-01
-4.13284659e-01 -6.33962899e-02 -4.50005859e-01 4.07330811e-01
1.65873736e-01 7.62280449e-02 -1.26166642e+00 -3.01433474e-01
-6.54223323e-01 -3.91524374e-01 1.16725600e+00 3.65868330e-01
6.79514170e-01 -9.61080939e-02 -3.50625396e-01 3.26143771e-01
8.28068733e-01 1.59909576e-01 6.41152799e-01 4.21392657e-02
7.74840474e-01 7.54904449e-01 7.85770476e-01 1.02222823e-01
7.66469419e-01 1.13377833e+00 1.75170138e-01 -1.12904809e-01
-5.76873183e-01 -7.07065612e-02 8.28691542e-01 9.35407996e-01
9.80345532e-02 -5.34711123e-01 -9.56559956e-01 6.77269995e-01
-1.73202097e+00 -9.98045087e-01 8.72781053e-02 1.90220666e+00
6.80121422e-01 -4.29360956e-01 3.12054366e-01 2.90347636e-01
8.34855199e-01 4.01955843e-01 -2.04466119e-01 -1.07019849e-01
-6.98710263e-01 -2.56159395e-01 2.73974001e-01 7.22081184e-01
-1.04002440e+00 1.04595017e+00 6.32874537e+00 1.14933932e+00
-1.34172201e+00 5.81704229e-02 3.93024176e-01 -3.97102609e-02
-5.10812283e-01 -3.17963004e-01 -7.36427426e-01 2.31757194e-01
8.18948030e-01 2.13172749e-01 4.88575280e-01 3.75146359e-01
3.07080477e-01 -8.18404257e-02 -9.18010831e-01 1.40206146e+00
5.50042450e-01 -9.78829563e-01 1.13852918e-01 4.56668846e-02
4.00495529e-01 2.66992182e-01 3.50345343e-01 2.63586849e-01
1.86598554e-01 -1.09737635e+00 4.07881707e-01 4.11284328e-01
8.63221228e-01 -6.40738189e-01 6.54687524e-01 1.85370654e-01
-1.29538310e+00 -3.96090187e-02 1.86324492e-01 6.51714981e-01
2.05397770e-01 -7.35256895e-02 -1.02849042e+00 7.89317191e-01
5.59171438e-01 6.10592842e-01 -9.37737644e-01 8.96981657e-01
-1.84907153e-01 6.65185034e-01 -2.92634159e-01 1.55151755e-01
7.20390230e-02 1.90719545e-01 8.38249266e-01 1.43714368e+00
1.78008914e-01 -1.37322932e-01 1.13907970e-01 3.98096114e-01
1.88652888e-01 3.28776628e-01 -6.42107725e-01 -2.74591327e-01
6.20166540e-01 1.06813037e+00 -3.35786939e-01 -1.80311501e-01
-5.72171688e-01 8.63563418e-01 6.80125877e-02 6.53005302e-01
-7.61758506e-01 2.49030087e-02 7.87481844e-01 -1.55077592e-01
2.66362488e-01 -4.41724092e-01 1.37881050e-02 -1.31104434e+00
-7.41518363e-02 -1.21583593e+00 4.48119611e-01 -1.04443479e+00
-1.03120649e+00 6.15326941e-01 -1.46045551e-01 -1.00174260e+00
-3.23131859e-01 -5.35632789e-01 -2.34541699e-01 6.51332140e-01
-1.70261133e+00 -1.49661803e+00 -1.50806278e-01 7.58886158e-01
8.35807741e-01 -4.36468214e-01 5.60046971e-01 4.83781964e-01
-5.50616980e-01 1.02254808e+00 -1.03993215e-01 2.22403735e-01
9.67622995e-01 -7.93114066e-01 -9.54406187e-02 8.43246400e-01
4.49614227e-01 4.58723992e-01 4.97385144e-01 -4.34682965e-01
-1.48801303e+00 -9.38884676e-01 6.95189118e-01 -4.60209697e-01
4.42120433e-01 -1.26782104e-01 -8.68520617e-01 4.98911977e-01
3.91634315e-01 -8.49935561e-02 6.15235507e-01 6.38575777e-02
-5.10299861e-01 2.93967742e-02 -9.54645276e-01 7.74338365e-01
8.98565650e-01 -1.00857258e+00 -4.56688672e-01 -1.87226623e-01
8.39255273e-01 -1.36770010e-01 -6.21402442e-01 8.92354667e-01
4.22525227e-01 -7.42252767e-01 1.26062512e+00 -4.28589284e-01
-1.25750648e-02 -6.06309712e-01 -8.27485442e-01 -1.35685825e+00
2.24847108e-01 -4.61372346e-01 -1.68292411e-02 1.77115607e+00
3.92686099e-01 -4.78629500e-01 3.49763691e-01 2.15640098e-01
-2.04797894e-01 -3.16701829e-01 -1.05617642e+00 -6.09912753e-01
-3.31586868e-01 -7.56032765e-01 3.45414251e-01 9.43507493e-01
-9.97213647e-02 3.26075196e-01 -3.86573821e-01 2.80840099e-01
3.04315567e-01 -9.46764424e-02 8.42787564e-01 -6.57283306e-01
7.48532405e-03 -5.39585114e-01 -4.52396214e-01 -1.18151736e+00
5.46206236e-01 -9.90246356e-01 2.49154314e-01 -1.50028110e+00
1.07108451e-01 -2.92935520e-01 -2.83317804e-01 5.96446037e-01
-2.42593050e-01 3.71549070e-01 4.38878804e-01 7.23934174e-02
-8.70494306e-01 8.24823797e-01 9.93918002e-01 -4.00152594e-01
-2.11772934e-01 -1.26631558e-01 -4.51133221e-01 4.61430967e-01
5.62972128e-01 1.24796793e-01 -2.94618428e-01 -3.75659794e-01
-1.76291198e-01 3.72927606e-01 5.06229758e-01 -6.15463972e-01
2.82887012e-01 -2.20232811e-02 5.11072315e-02 -1.02813923e+00
7.79457033e-01 -6.41017318e-01 -2.91296810e-01 -1.84211046e-01
-5.96014202e-01 -1.00689799e-01 3.08992833e-01 4.41906154e-01
-5.56630075e-01 1.72083199e-01 7.36904740e-01 3.02972853e-01
-8.11082244e-01 -6.26725405e-02 -4.97355044e-01 4.52080145e-02
8.11858952e-01 -3.15045528e-02 -4.58226889e-01 -6.72782004e-01
-7.31106400e-01 3.42708290e-01 5.20106673e-01 8.02021563e-01
8.57323587e-01 -1.53038442e+00 -7.85503149e-01 8.27824548e-02
5.19546151e-01 -5.12690127e-01 4.98922080e-01 9.94962931e-01
1.21618845e-01 6.34668350e-01 1.89664975e-01 -9.71514702e-01
-1.83104479e+00 5.83208382e-01 3.33748579e-01 1.61636338e-01
-1.44517004e-01 7.95037508e-01 1.39581904e-01 -5.17295539e-01
4.51243579e-01 7.62382820e-02 -5.28666377e-01 3.53620052e-01
4.96766955e-01 3.08806807e-01 -3.18184718e-02 -1.44508600e+00
-6.26081109e-01 6.71855628e-01 2.13921145e-01 -6.29325807e-01
7.95899987e-01 -8.03015113e-01 2.20188454e-01 6.48427248e-01
1.17719531e+00 3.04982901e-01 -6.75161481e-01 -6.03249788e-01
-1.20204516e-01 -3.82051796e-01 3.47942375e-02 -9.56856072e-01
-1.04803884e+00 1.10800970e+00 7.48136342e-01 1.33905157e-01
1.20608997e+00 4.24836099e-01 5.99534750e-01 2.80589044e-01
-1.77622028e-02 -9.91722465e-01 2.74084300e-01 5.13990402e-01
8.97733927e-01 -1.59195030e+00 -5.48179150e-01 -3.29072833e-01
-1.12700760e+00 8.41220200e-01 3.95007730e-01 6.48547053e-01
3.14577729e-01 1.20276682e-01 4.64150518e-01 1.94783822e-01
-7.79694259e-01 -9.91238296e-01 7.23758698e-01 7.37874806e-01
4.23866391e-01 7.10912198e-02 1.15799211e-01 4.02356297e-01
-6.10596985e-02 -2.90362120e-01 6.16026632e-02 6.01882041e-01
-2.95605093e-01 -1.00083470e+00 -5.60052991e-01 -1.46234065e-01
-3.76309276e-01 -1.22951128e-01 -6.72821939e-01 5.62251568e-01
-2.59313583e-01 1.52883661e+00 -9.41939577e-02 -5.50703168e-01
2.91268587e-01 3.00393730e-01 4.77090448e-01 -2.92358935e-01
-1.79556206e-01 3.75694394e-01 3.97680044e-01 -3.89349520e-01
-6.75287247e-01 -7.18524814e-01 -1.02333939e+00 -9.48622003e-02
-4.13651228e-01 -1.69422165e-01 7.92791903e-01 1.07001626e+00
4.78072017e-01 5.27860403e-01 5.56445718e-01 -8.48579526e-01
-1.19660415e-01 -9.78473306e-01 -7.22407624e-02 4.66776580e-01
6.54876709e-01 -6.64710939e-01 -5.83808720e-01 3.73455808e-02] | [14.2049560546875, 5.090799331665039] |
a7c01e81-021c-4555-bafb-419cbe2e2c27 | metaphorical-polysemy-detection-conventional-1 | 2212.08395 | null | https://arxiv.org/abs/2212.08395v1 | https://arxiv.org/pdf/2212.08395v1.pdf | Metaphorical Polysemy Detection: Conventional Metaphor meets Word Sense Disambiguation | Linguists distinguish between novel and conventional metaphor, a distinction which the metaphor detection task in NLP does not take into account. Instead, metaphoricity is formulated as a property of a token in a sentence, regardless of metaphor type. In this paper, we investigate the limitations of treating conventional metaphors in this way, and advocate for an alternative which we name 'metaphorical polysemy detection' (MPD). In MPD, only conventional metaphoricity is treated, and it is formulated as a property of word senses in a lexicon. We develop the first MPD model, which learns to identify conventional metaphors in the English WordNet. To train it, we present a novel training procedure that combines metaphor detection with word sense disambiguation (WSD). For evaluation, we manually annotate metaphor in two subsets of WordNet. Our model significantly outperforms a strong baseline based on a state-of-the-art metaphor detection model, attaining an ROC-AUC score of .78 (compared to .65) on one of the sets. Additionally, when paired with a WSD model, our approach outperforms a state-of-the-art metaphor detection model at identifying conventional metaphors in text (.659 F1 compared to .626). | ['Simone Teufel', 'Rowan Hall Maudslay'] | 2022-12-16 | metaphorical-polysemy-detection-conventional | https://aclanthology.org/2022.coling-1.7 | https://aclanthology.org/2022.coling-1.7.pdf | coling-2022-10 | ['word-sense-disambiguation'] | ['natural-language-processing'] | [ 2.79178340e-02 -4.47262302e-02 -1.97208747e-01 -6.80096401e-03
-3.81104708e-01 -1.02993441e+00 1.10799015e+00 5.35551310e-01
-8.33986521e-01 3.88497859e-01 5.04413009e-01 -7.48474598e-01
-9.74046141e-02 -7.34834433e-01 -1.10041887e-01 -4.01054800e-01
3.21255207e-01 4.64719713e-01 -4.03343476e-02 -7.00091124e-01
4.92444605e-01 2.45455563e-01 -1.09369397e+00 2.41715968e-01
8.40199709e-01 6.88017249e-01 4.49047059e-01 1.97553597e-02
-5.70415676e-01 2.11718127e-01 -8.76565933e-01 -1.98467225e-01
-5.24293892e-02 -2.46402502e-01 -9.35974538e-01 -5.53056240e-01
1.19851179e-01 2.66496778e-01 1.69358194e-01 8.87228072e-01
3.18217009e-01 1.36509672e-01 8.41974258e-01 -7.27488816e-01
-9.95481730e-01 6.21153116e-01 -4.44539994e-01 3.34785163e-01
7.14257658e-01 -1.69084162e-01 1.36482394e+00 -1.06860828e+00
7.06074476e-01 1.50094366e+00 5.18858850e-01 4.90472794e-01
-1.63065708e+00 -4.55094099e-01 1.99204758e-01 -6.65083379e-02
-1.35736537e+00 -1.00946754e-01 8.87794435e-01 -4.89589065e-01
1.61333632e+00 3.22853893e-01 6.04751945e-01 1.46884382e+00
1.85885698e-01 3.62856358e-01 1.46444917e+00 -9.93582547e-01
3.59086394e-01 1.96342587e-01 2.91434377e-01 3.45397800e-01
1.44009024e-01 8.75801370e-02 -5.46446323e-01 -4.14076686e-01
5.35133898e-01 -3.29940528e-01 -1.11388288e-01 1.53203920e-01
-1.24023628e+00 1.13634562e+00 4.09359068e-01 9.26148951e-01
-4.06575292e-01 -1.83835551e-01 6.41560495e-01 3.96143198e-01
5.90948403e-01 1.13508403e+00 -1.99301660e-01 -1.90197587e-01
-8.80585909e-01 4.19442981e-01 9.25692320e-01 4.84315157e-01
3.41510266e-01 -2.63141453e-01 5.65446541e-02 1.20668864e+00
7.93447569e-02 3.58881861e-01 1.12958729e+00 -4.49280083e-01
-5.61842918e-02 8.14876556e-01 1.00852162e-01 -1.17068064e+00
-5.16226888e-01 -1.49825931e-01 -4.79226887e-01 -2.59478956e-01
-6.41057938e-02 1.76931053e-01 -6.75373852e-01 2.00945663e+00
4.32159752e-02 -7.36133084e-02 1.77782420e-02 8.60715330e-01
9.06368375e-01 3.89584690e-01 3.57347071e-01 -1.13285653e-01
2.01356435e+00 -5.78987539e-01 -4.87458378e-01 -6.68803871e-01
5.52398741e-01 -9.11316156e-01 1.87118459e+00 2.90641636e-01
-5.73130727e-01 -3.63320857e-01 -1.18369293e+00 -2.12329328e-01
-1.04766583e+00 -1.54316664e-01 6.39256835e-01 6.40565753e-01
-9.47968006e-01 1.06268577e-01 -4.79450971e-01 -8.36424112e-01
-1.77549139e-01 -1.66935593e-01 -2.45722428e-01 2.61777163e-01
-1.56458116e+00 1.58884966e+00 6.02473557e-01 -5.27240813e-01
-4.20910388e-01 -6.33367360e-01 -8.75861406e-01 8.01854581e-02
3.07883620e-01 -9.03508663e-01 1.02667952e+00 -7.81962514e-01
-1.01279616e+00 1.37566686e+00 -1.85495466e-01 -5.70525289e-01
6.03680313e-02 -3.47308785e-01 -7.64464259e-01 5.61493225e-02
6.27275944e-01 7.42451668e-01 6.13431096e-01 -1.26411664e+00
-3.06258351e-01 9.18776393e-02 6.00110114e-01 6.50621206e-03
-6.22377098e-01 1.53549194e-01 1.13070287e-01 -1.00016034e+00
-8.09578896e-02 -8.83502722e-01 9.55491588e-02 -3.73920709e-01
-3.12662244e-01 -6.72558546e-01 5.56796432e-01 -3.33545357e-01
1.58325863e+00 -2.25780630e+00 -4.28073332e-02 9.96794999e-02
1.87111303e-01 1.08032048e-01 -1.95845529e-01 8.09517026e-01
-1.97652027e-01 2.83288419e-01 -2.42621467e-01 -3.54547858e-01
5.03832579e-01 5.71658790e-01 -8.84266973e-01 -3.29011120e-02
2.43523970e-01 8.25729072e-01 -1.27216613e+00 -4.41818357e-01
3.26507181e-01 3.63774449e-01 -4.46750969e-01 -1.59590408e-01
-2.61314094e-01 1.75522882e-02 -2.76348263e-01 5.81697226e-01
4.25283909e-01 -1.06246904e-01 5.65865219e-01 2.33902819e-02
-2.08955675e-01 7.88748741e-01 -5.53762674e-01 1.96740615e+00
-1.04405797e+00 5.70682943e-01 -4.78097618e-01 -1.06527495e+00
1.00302458e+00 1.82203844e-01 1.54264838e-01 -8.10426414e-01
-6.81814477e-02 5.89545906e-01 5.99794388e-02 -2.74609953e-01
7.17856944e-01 -6.58666193e-01 -5.01750290e-01 3.76348853e-01
3.35058831e-02 -1.06942952e-01 1.95253760e-01 1.46079406e-01
1.10160375e+00 -1.42202020e-01 6.84385419e-01 -9.07621741e-01
5.36876142e-01 1.27807617e-01 1.10369965e-01 8.22109222e-01
1.68759227e-01 1.54190183e-01 4.25016165e-01 -3.61483812e-01
-4.60324734e-01 -1.39768147e+00 -3.71406227e-01 1.11909008e+00
1.70029923e-01 -9.29190516e-01 -4.70499843e-01 -5.25419533e-01
1.09082103e-01 1.31072760e+00 -7.64046133e-01 -1.83810279e-01
-4.85176831e-01 -5.62414944e-01 6.94678247e-01 3.86954665e-01
2.26437584e-01 -1.11429513e+00 -1.19375587e+00 1.67145520e-01
-3.86350602e-01 -1.18980002e+00 -4.09125015e-02 2.42079377e-01
-2.03262389e-01 -8.55294168e-01 -3.29482734e-01 -7.60069609e-01
5.13967201e-02 4.26608175e-01 1.67918885e+00 1.88793808e-01
-2.24976853e-01 2.78501511e-01 -6.18560255e-01 -5.58581412e-01
-3.72549653e-01 8.87379497e-02 2.35972211e-01 -7.24340796e-01
1.06479549e+00 -7.52039373e-01 -5.12979805e-01 2.44626794e-02
-1.09572589e+00 -1.90475464e-01 3.79968643e-01 1.03784323e+00
2.49310017e-01 -1.88077003e-01 4.74902213e-01 -5.48942029e-01
1.29925346e+00 -5.15460253e-01 -2.30014920e-01 2.10309997e-01
-8.15511286e-01 1.15589788e-02 4.58528578e-01 -6.21127009e-01
-6.91545427e-01 -5.49116671e-01 -1.43159002e-01 -3.09089243e-01
-1.21053442e-01 9.69326019e-01 1.95416346e-01 2.17626482e-01
7.69169509e-01 4.00954932e-02 -2.40951657e-01 -6.71352386e-01
6.52780712e-01 6.17647946e-01 4.70811784e-01 -9.34875846e-01
5.26108384e-01 3.19681853e-01 -4.51383218e-02 -9.91313934e-01
-8.87332678e-01 -6.79926872e-01 -4.09857213e-01 3.17093283e-01
8.08734775e-01 -6.94756687e-01 -4.59193915e-01 -3.02371830e-01
-1.43489540e+00 1.61693230e-01 -2.50099182e-01 3.18593979e-01
-3.57947350e-01 3.85159343e-01 -1.75717354e-01 -7.55671859e-01
-4.00463790e-01 -7.15724766e-01 1.18640590e+00 -2.31310353e-01
-1.11304867e+00 -1.42790031e+00 3.41456950e-01 -1.24896675e-01
4.64629889e-01 4.91955966e-01 1.45028615e+00 -1.01743889e+00
5.02984703e-01 6.45134449e-02 -2.80841142e-01 1.19216688e-01
4.87990558e-01 -5.18896163e-01 -9.70560849e-01 -4.56331149e-02
2.67422348e-01 -2.57591039e-01 1.01151347e+00 5.51185720e-02
6.12732589e-01 -3.40500981e-01 -4.67645228e-01 5.27481325e-02
1.63239682e+00 4.71840501e-02 1.76744655e-01 8.10944796e-01
2.21367761e-01 6.67247772e-01 5.66649199e-01 3.07924300e-01
3.67802560e-01 8.67988169e-01 1.65508032e-01 -3.39800380e-02
1.44260406e-01 -3.09991956e-01 3.04772168e-01 5.96763790e-01
2.80784130e-01 -1.83784068e-01 -1.34417367e+00 7.52932549e-01
-1.66702306e+00 -6.60699606e-01 1.51359245e-01 1.90634680e+00
9.10511196e-01 4.43929732e-01 2.28850514e-01 -1.14882611e-01
2.47864053e-01 3.45061302e-01 -9.47332978e-02 -9.98852611e-01
-2.86061734e-01 7.05397546e-01 -4.68162224e-02 6.34657741e-01
-9.13559079e-01 1.35398233e+00 6.21263742e+00 9.82478917e-01
-1.14262772e+00 2.67843157e-01 -1.20571051e-02 8.98122042e-02
-5.26515067e-01 2.02037185e-01 -1.77961692e-01 5.21314144e-01
7.79657185e-01 -2.09189370e-01 5.26870072e-01 7.45855272e-01
5.38366176e-02 -1.98825806e-01 -1.14546120e+00 9.87396359e-01
9.80351195e-02 -1.08646131e+00 3.69475722e-01 -9.77597549e-04
2.29746431e-01 -1.19540721e-01 -3.03904545e-02 3.48436475e-01
5.31315543e-02 -1.18969727e+00 5.90991974e-01 9.61420387e-02
6.40910506e-01 -5.76750398e-01 5.80336213e-01 3.20464551e-01
-1.16134596e+00 -5.60246035e-02 -5.32245159e-01 -3.76596034e-01
1.37116298e-01 6.07508421e-01 -8.34261060e-01 6.91054583e-01
5.27650237e-01 5.00285208e-01 -5.61108351e-01 2.33825296e-01
-4.36535627e-01 5.28585494e-01 -3.16125631e-01 -5.66329136e-02
4.63738203e-01 -8.72105658e-02 7.90026307e-01 1.77218866e+00
2.35698491e-01 -4.59509194e-02 2.62539625e-01 1.29350567e+00
2.53229022e-01 2.57715225e-01 -6.62390113e-01 -4.25922930e-01
8.88609171e-01 1.24796081e+00 -8.04782391e-01 -4.27041948e-01
-2.74153411e-01 9.30355191e-01 1.98053226e-01 9.86834615e-03
-7.34827220e-01 -2.74426728e-01 8.03898990e-01 -1.16423279e-01
-1.38458967e-01 -2.27940679e-01 -3.73972476e-01 -1.08570051e+00
7.07447007e-02 -6.30623043e-01 5.43124676e-01 -7.43240714e-01
-1.86953354e+00 6.29669189e-01 4.20871586e-01 -1.17645919e+00
-4.98883158e-01 -1.09753001e+00 -7.54430890e-01 1.07437372e+00
-1.58698690e+00 -1.28056657e+00 -2.10860968e-02 3.26451361e-01
4.89696443e-01 -1.99055672e-02 1.52690578e+00 -6.10030666e-02
4.93295640e-02 4.49784935e-01 -3.34348351e-01 2.61219796e-02
8.40854943e-01 -1.49789739e+00 6.79162443e-01 5.16532838e-01
3.33304465e-01 1.37337899e+00 1.07497180e+00 -4.49516773e-01
-9.91575241e-01 -4.83662486e-01 1.46947229e+00 -5.04494369e-01
1.32243192e+00 -4.85747218e-01 -9.22508299e-01 3.37549746e-01
5.60475171e-01 -6.11344934e-01 1.00384283e+00 5.49682200e-01
-8.73821735e-01 5.51561117e-01 -9.36372638e-01 7.13612020e-01
1.18010342e+00 -9.14141834e-01 -1.58600354e+00 4.01908159e-01
1.08799267e+00 -1.75901830e-01 -6.57010913e-01 1.70017555e-01
6.96799695e-01 -7.52122104e-01 1.27665973e+00 -7.18955755e-01
4.99310881e-01 3.02049201e-02 -3.73137116e-01 -1.50950968e+00
-2.10149571e-01 -4.34954196e-01 7.59676844e-02 1.21087837e+00
2.35865280e-01 -9.81300890e-01 -8.58397633e-02 1.82566226e-01
-4.77373153e-02 -7.03819215e-01 -1.24419498e+00 -1.20580184e+00
5.99876225e-01 -2.94089973e-01 6.81281090e-01 1.44740367e+00
6.88580751e-01 8.03295135e-01 1.93581134e-01 -4.21389073e-01
-7.07097724e-02 4.29200351e-01 1.88307628e-01 -1.29627049e+00
-2.84417033e-01 -9.37903583e-01 -2.84403443e-01 -8.96116078e-01
7.53496468e-01 -9.43790138e-01 -3.10375780e-01 -1.52670562e+00
3.43097816e-03 -2.30229110e-01 -6.44252539e-01 6.61905229e-01
-8.11001807e-02 1.19496837e-01 2.79892743e-01 1.96776524e-01
-5.69102578e-02 3.72778147e-01 6.63127482e-01 -1.86092198e-01
-1.71181232e-01 -8.01219523e-01 -9.77009177e-01 7.38362074e-01
8.53072643e-01 -4.73188311e-01 -4.82491136e-01 -3.91351163e-01
4.32855844e-01 -3.67572784e-01 7.66291916e-01 -5.00536084e-01
-3.38531166e-01 -2.64721960e-01 -4.65988144e-02 -6.31244630e-02
4.80585247e-01 -6.55572474e-01 -2.74218470e-01 5.62316835e-01
-2.24372208e-01 6.33984447e-01 4.18980777e-01 2.12682888e-01
-9.99382660e-02 -2.56624967e-01 3.87480229e-01 -3.18483651e-01
-7.66170979e-01 -8.04703832e-01 -3.81984800e-01 2.85327107e-01
4.62230414e-01 -2.71146446e-01 -5.35097539e-01 1.42937481e-01
-3.69173497e-01 -8.30846801e-02 6.13435864e-01 9.64359820e-01
5.69848299e-01 -1.28046453e+00 -4.70116794e-01 -1.13813475e-01
5.09924829e-01 -5.05269110e-01 -2.84356207e-01 7.99955130e-01
-1.30936503e-01 6.08529389e-01 -1.20862946e-01 -3.82553548e-01
-9.46132123e-01 8.19793224e-01 5.42957075e-02 -3.16616923e-01
-6.38947010e-01 6.58575892e-01 3.52456540e-01 -3.98646325e-01
-3.23436320e-01 -4.59123135e-01 -3.24903011e-01 3.26791853e-01
2.72711009e-01 3.10103539e-02 -4.72852215e-02 -7.58931816e-01
-7.78730989e-01 4.25352216e-01 1.81493744e-01 -4.95898873e-01
9.70015883e-01 -3.30892988e-02 -4.89166021e-01 9.08602893e-01
1.00981057e+00 1.70151934e-01 -2.10078489e-02 -3.30747105e-02
5.22541404e-01 -5.31558514e-01 -1.72396347e-01 -1.31312990e+00
-6.07437864e-02 6.86679304e-01 4.71364558e-01 5.60204685e-01
1.06971180e+00 1.02001056e-01 8.06357443e-01 3.27009857e-01
3.88106227e-01 -1.04380035e+00 2.53198855e-02 8.27124774e-01
1.02465439e+00 -9.77199793e-01 -1.88052595e-01 -2.97120243e-01
-5.83869278e-01 1.08992243e+00 3.82764310e-01 -3.31793427e-01
5.51952899e-01 2.92721651e-02 2.59361625e-01 -5.73989868e-01
-6.37721241e-01 -1.82189494e-01 5.55728436e-01 3.42208207e-01
7.12184012e-01 3.06530684e-01 -1.11424387e+00 8.22398782e-01
-7.94168651e-01 -3.63402367e-01 4.45187539e-02 9.10795093e-01
-3.47046405e-01 -1.18339765e+00 -2.05214068e-01 1.92251101e-01
-2.76017427e-01 -6.70741737e-01 -9.84689772e-01 9.07628357e-01
2.00628474e-01 1.20566010e+00 1.81879982e-01 -4.40663099e-01
2.38479346e-01 3.21732014e-01 2.90797263e-01 -7.29566813e-01
-7.51159787e-01 2.75270659e-02 2.87934214e-01 -3.49324018e-01
-6.30889058e-01 -2.89523035e-01 -1.18970132e+00 -1.32265776e-01
-2.02595085e-01 4.35192019e-01 3.95065576e-01 1.33901572e+00
1.85030341e-01 2.89268315e-01 1.87560126e-01 -4.99359846e-01
-6.49382949e-01 -1.05873787e+00 -4.51896191e-01 4.69345212e-01
2.70587236e-01 -1.13188982e+00 -3.59533995e-01 -3.42930019e-01] | [10.575827598571777, 9.133906364440918] |
e82ffbac-918e-434a-b0f8-22ece13dfb9b | cabinet-scaling-neural-collision-detection | 2304.09302 | null | https://arxiv.org/abs/2304.09302v1 | https://arxiv.org/pdf/2304.09302v1.pdf | CabiNet: Scaling Neural Collision Detection for Object Rearrangement with Procedural Scene Generation | We address the important problem of generalizing robotic rearrangement to clutter without any explicit object models. We first generate over 650K cluttered scenes - orders of magnitude more than prior work - in diverse everyday environments, such as cabinets and shelves. We render synthetic partial point clouds from this data and use it to train our CabiNet model architecture. CabiNet is a collision model that accepts object and scene point clouds, captured from a single-view depth observation, and predicts collisions for SE(3) object poses in the scene. Our representation has a fast inference speed of 7 microseconds per query with nearly 20% higher performance than baseline approaches in challenging environments. We use this collision model in conjunction with a Model Predictive Path Integral (MPPI) planner to generate collision-free trajectories for picking and placing in clutter. CabiNet also predicts waypoints, computed from the scene's signed distance field (SDF), that allows the robot to navigate tight spaces during rearrangement. This improves rearrangement performance by nearly 35% compared to baselines. We systematically evaluate our approach, procedurally generate simulated experiments, and demonstrate that our approach directly transfers to the real world, despite training exclusively in simulation. Robot experiment demos in completely unknown scenes and objects can be found at this http https://cabinet-object-rearrangement.github.io | ['Dieter Fox', 'Adam Fishman', 'Clemens Eppner', 'Arsalan Mousavian', 'Adithyavairavan Murali'] | 2023-04-18 | null | null | null | null | ['scene-generation'] | ['computer-vision'] | [-6.1702300e-03 1.8090616e-01 4.7355068e-01 -2.2081095e-01
-7.3198307e-01 -9.3174911e-01 5.4156905e-01 2.1452406e-01
-5.1400518e-01 6.0149139e-01 -1.5025745e-02 -3.9016291e-01
-2.6929036e-02 -9.6234196e-01 -1.3191881e+00 -2.3075841e-01
-2.6701003e-01 1.3610562e+00 5.2744251e-01 -5.2619416e-01
3.8044375e-01 1.0033894e+00 -1.6444906e+00 7.2721273e-02
6.8413597e-01 4.4983146e-01 8.0332893e-01 9.7982407e-01
2.2801638e-01 4.1497591e-01 -4.4419098e-01 1.4000313e-01
5.4095381e-01 5.1266146e-01 -8.6404610e-01 9.4652781e-03
2.3126853e-01 -5.8352447e-01 -4.6944183e-01 5.7080787e-01
2.6614067e-01 3.2719758e-01 6.4045841e-01 -1.2369024e+00
1.4809671e-01 1.1038626e-01 -3.6047292e-01 -1.6320528e-01
8.4491211e-01 6.8369484e-01 4.7933832e-01 -9.4109201e-01
1.0660595e+00 1.5800779e+00 6.9061583e-01 1.2134619e-01
-1.3364568e+00 -5.6706953e-01 1.5031663e-01 -9.5471963e-02
-1.4425651e+00 -2.7730587e-01 1.8956080e-01 -4.7601783e-01
1.3659236e+00 2.5602156e-01 9.5246339e-01 1.0433780e+00
6.0634804e-01 5.7615399e-01 3.6060944e-01 -2.6638845e-02
3.0060741e-01 -1.9810477e-01 -1.7253411e-01 7.0899606e-01
3.5595259e-01 3.1257826e-01 -1.9716121e-01 -2.6829630e-01
1.2115461e+00 -3.6292981e-02 -1.3460037e-01 -1.0833685e+00
-1.7407023e+00 5.2406847e-01 6.8832392e-01 -5.8231622e-01
-3.1029999e-01 6.4753968e-01 1.9743349e-01 -9.2978299e-02
-8.8479698e-02 5.1444662e-01 -4.2649412e-01 -1.0446248e-01
-2.6410371e-01 1.1401129e+00 8.8405663e-01 1.8519785e+00
8.3328015e-01 -2.8237334e-01 4.2280412e-01 4.0991300e-01
2.7227774e-01 7.8580695e-01 -3.7842345e-01 -1.6140395e+00
6.4099503e-01 3.4811264e-01 6.7706758e-01 -9.9777186e-01
-7.5392634e-01 -2.0804696e-02 -4.6905157e-01 6.3603848e-01
2.0602840e-01 -1.2963264e-01 -9.8668265e-01 1.1049519e+00
7.6855975e-01 -1.3647807e-01 1.1689367e-01 1.0125494e+00
3.8831189e-01 7.1989679e-01 -2.3405676e-01 4.1758808e-01
1.1566669e+00 -9.1400975e-01 -1.8102145e-01 -4.2257819e-01
7.3163795e-01 -7.5924176e-01 9.5392519e-01 5.6136256e-01
-1.2049639e+00 -3.4855974e-01 -8.4541082e-01 -3.1074262e-01
-1.7973238e-01 -2.6416039e-01 7.4512875e-01 -3.2796353e-01
-1.0893043e+00 5.3164738e-01 -1.2065364e+00 -6.1313623e-01
2.2107024e-01 4.6036294e-01 -4.0658081e-01 -3.6065084e-01
-3.6694095e-01 1.0708739e+00 2.7884737e-01 1.0988522e-01
-1.3708692e+00 -7.7775860e-01 -1.0588815e+00 -2.0948711e-01
4.8910922e-01 -1.2721187e+00 1.7960728e+00 2.5356883e-02
-1.2821095e+00 5.0026113e-01 -2.5485948e-01 -3.4041506e-01
6.7811728e-01 -4.8005843e-01 3.4976512e-01 4.8267037e-02
3.3994725e-01 1.0276866e+00 1.1275398e-01 -1.9084783e+00
-6.7736018e-01 -2.3616819e-01 2.5354272e-01 5.6532580e-01
9.6437430e-01 -5.2610159e-01 -6.5337396e-01 -1.2460225e-01
4.3772569e-01 -1.2886335e+00 -9.4704616e-01 3.7403741e-01
-5.3100115e-01 2.1136016e-01 6.7553413e-01 -2.8849894e-01
2.9156756e-01 -1.9304529e+00 2.1715389e-01 4.3385732e-01
-1.2565410e-01 -4.7693017e-01 -1.5020671e-01 8.4951389e-01
3.8087553e-01 -2.9333833e-01 -3.0344301e-01 -4.6934617e-01
1.1784703e-01 5.1470083e-01 -5.3472495e-01 5.2828509e-01
-1.6991496e-01 7.7056205e-01 -1.0852842e+00 -3.2941556e-01
5.7978594e-01 4.0696502e-01 -1.0949711e+00 3.5907984e-02
-6.4195681e-01 6.4053816e-01 -4.6574810e-01 6.1803168e-01
8.7816459e-01 7.9930879e-02 1.6844022e-01 -7.1543276e-02
-3.9475533e-01 2.7343965e-01 -1.3057277e+00 2.2795908e+00
-4.5462638e-01 4.1096795e-01 3.8308221e-01 -3.1317553e-01
7.0020378e-01 -3.2084328e-01 3.5676274e-01 -3.6663952e-01
2.3207147e-02 1.5945019e-02 -3.4383163e-01 -3.3462599e-01
8.5895294e-01 3.2689255e-01 -3.7272638e-01 5.3136242e-03
-4.9315867e-01 -1.1526920e+00 -1.9514419e-02 3.6303866e-01
1.3784437e+00 5.9847546e-01 7.6410294e-02 -2.7608854e-01
-2.0146552e-01 9.6287817e-01 2.0803927e-01 6.9021910e-01
2.3396960e-01 6.4904785e-01 5.2553695e-02 -5.4775310e-01
-1.2738484e+00 -1.5481486e+00 2.5333671e-02 6.4943779e-01
9.6940947e-01 -3.6613077e-01 -5.0338107e-01 2.2626845e-02
5.0449258e-01 1.0065284e+00 -1.8577363e-01 1.2206378e-01
-9.1734964e-01 -1.6695440e-01 1.8963102e-01 6.0871935e-01
1.2979038e-01 -9.9161798e-01 -1.2343222e+00 3.8762659e-01
-2.2854982e-01 -1.0726074e+00 -1.4816603e-01 3.1419945e-01
-9.3157905e-01 -1.2313277e+00 -6.2382307e-02 -7.6352870e-01
8.6986393e-01 5.7542521e-01 1.2383986e+00 -5.2694820e-02
-7.4930698e-01 7.2706491e-01 -1.6893911e-01 -4.9959621e-01
-3.9489001e-01 -2.1476606e-01 6.7407437e-02 -1.1493032e+00
-1.0832686e-01 -6.4013451e-01 -6.5952921e-01 3.9763373e-01
-5.6431854e-01 5.0775796e-01 5.0594389e-01 2.9723123e-01
1.0175960e+00 -1.8304101e-01 -2.8218299e-01 -3.5221857e-01
2.1770674e-01 -4.9502686e-01 -1.0694600e+00 -2.2780620e-01
4.0440772e-02 -6.7566112e-02 2.4356607e-01 -1.8578997e-01
-9.6099842e-01 5.5745202e-01 6.8265349e-02 -5.5109501e-01
-3.6190763e-01 2.5383294e-01 -2.2978317e-02 4.8660867e-02
6.6989541e-01 -1.3641672e-01 -1.7693713e-01 -3.5596392e-01
5.6847626e-01 1.7860658e-02 9.1162658e-01 -1.0069468e+00
7.9193836e-01 8.3717972e-01 6.7040138e-02 -6.4343423e-01
-9.7007550e-02 -4.4526744e-01 -7.3675776e-01 -1.2486965e-01
6.7767560e-01 -9.2526805e-01 -1.3753927e+00 7.3877111e-02
-1.5193413e+00 -8.7119365e-01 -3.4451619e-01 5.0990397e-01
-1.1353462e+00 1.8648172e-02 -4.7505721e-01 -8.5546941e-01
1.5837878e-01 -1.3376018e+00 1.5345541e+00 -2.5635529e-01
-3.1988740e-01 -3.9007777e-01 1.5817341e-01 -5.8383357e-02
-8.0768518e-02 6.0440218e-01 6.0223883e-01 1.9412052e-02
-1.3543656e+00 8.1452169e-02 -9.5635206e-02 -6.1416662e-01
-2.4053037e-01 2.0523220e-02 -4.8230615e-01 -3.1143731e-01
-4.8926711e-01 -7.9195306e-02 4.0084195e-01 3.9159566e-01
1.1032023e+00 -8.5486397e-02 -1.1113119e+00 6.9713241e-01
1.5250598e+00 2.8654116e-01 6.5039498e-01 7.0090699e-01
4.8244554e-01 6.0187864e-01 1.1606746e+00 6.9350266e-01
8.6971688e-01 7.8342319e-01 1.0985372e+00 8.3562516e-02
1.8000643e-01 -4.0766525e-01 1.5000902e-01 3.5801095e-01
1.4541580e-02 -2.9242823e-01 -1.4394737e+00 6.4302611e-01
-1.9317484e+00 -7.5573629e-01 -5.6913334e-01 2.1084914e+00
3.4166893e-01 2.3213652e-01 -2.0738731e-01 -4.0358764e-01
2.6588854e-01 -3.8369390e-01 -7.0476049e-01 -2.8983149e-01
4.1331029e-01 -1.3000304e-02 9.6171999e-01 1.0158632e+00
-8.0290556e-01 1.2427419e+00 6.2972312e+00 1.7119673e-01
-5.5878043e-01 -2.1948493e-01 -1.0409758e-02 -4.0219676e-01
-2.9543090e-01 2.4779527e-01 -8.0332428e-01 -1.9237708e-02
7.0352900e-01 -1.9567775e-02 7.2329611e-01 1.1714094e+00
4.0799224e-01 -6.4505428e-01 -1.3775858e+00 9.0879613e-01
-2.3857254e-01 -1.3520660e+00 -1.0914157e-01 2.6583683e-01
3.5616630e-01 5.2640069e-01 -1.4066564e-01 3.2847780e-01
1.1556600e+00 -9.6657991e-01 1.2278856e+00 5.2971905e-01
5.5494738e-01 -8.0281210e-01 3.0993146e-01 7.5058556e-01
-1.2913141e+00 -6.6551338e-03 -4.7951570e-01 -1.3137253e-01
6.3007307e-01 2.0550935e-01 -1.4443657e+00 5.4589564e-01
9.6505940e-01 3.6054900e-01 -9.2896424e-02 1.2400461e+00
1.9646515e-01 -2.2308011e-01 -8.2951909e-01 1.7690590e-01
3.3245301e-01 -3.1717464e-01 8.2582593e-01 1.1416149e+00
3.6053684e-01 2.6071784e-01 4.5489153e-01 8.9453840e-01
4.3928993e-01 -5.2169526e-01 -9.5568287e-01 7.0354271e-01
7.8304458e-01 9.5551682e-01 -8.1792474e-01 -2.5759253e-01
1.8595123e-01 1.1105832e+00 3.3427253e-01 4.3937555e-01
-1.0810950e+00 -3.0761001e-01 1.0401045e+00 3.2323182e-01
2.0676371e-01 -8.5510802e-01 -2.5674173e-01 -7.5471735e-01
-4.0808551e-02 -4.9570021e-01 -2.0771568e-01 -1.3120875e+00
-6.8036330e-01 4.5376012e-01 4.8416558e-01 -1.3207750e+00
-3.2711843e-01 -5.7848918e-01 -1.3548672e-01 6.8671906e-01
-1.2809581e+00 -1.0712240e+00 -7.8019446e-01 5.2450472e-01
8.5114455e-01 5.0902289e-01 7.6183599e-01 -8.0639780e-02
7.7692807e-02 -1.5900822e-01 7.5890526e-02 -3.6640984e-01
5.0943345e-01 -1.2186270e+00 1.0292829e+00 3.2340756e-01
-3.3656335e-01 6.9080269e-01 1.0179045e+00 -1.0403929e+00
-1.8121568e+00 -1.1602247e+00 4.9070373e-02 -1.0100718e+00
3.0458713e-01 -6.5231586e-01 -5.5597168e-01 1.2537906e+00
-1.9982076e-01 -6.8811022e-02 9.9200225e-03 -2.6099408e-01
-1.7685877e-03 3.1431764e-01 -1.2112772e+00 1.0169821e+00
1.6503772e+00 1.3231939e-01 -4.7804597e-01 5.6624818e-01
1.0658671e+00 -1.2832011e+00 -6.0264564e-01 4.5989576e-01
6.7797428e-01 -9.0720898e-01 1.2749968e+00 -2.0479234e-01
1.2844153e-01 -5.4813993e-01 -4.8204023e-01 -1.3640974e+00
-2.2864869e-01 -5.5683076e-01 1.3821702e-01 4.3601066e-01
1.0808342e-01 -5.0850219e-01 8.7700945e-01 5.8209777e-01
-7.4934614e-01 -6.7740107e-01 -7.0800948e-01 -7.1893322e-01
-1.8538642e-01 -6.5094626e-01 6.8017352e-01 4.3647921e-01
-1.6572839e-01 7.8733698e-02 3.0803838e-01 8.2575750e-01
6.9970679e-01 1.4371222e-02 1.4909676e+00 -1.0448893e+00
1.8686131e-02 -9.4343387e-02 -1.9020349e-01 -1.5538126e+00
6.8621442e-02 -7.7506852e-01 5.9731275e-01 -2.1307166e+00
-1.2532903e-01 -1.0417911e+00 8.1095809e-01 2.9534444e-01
5.4159063e-01 -2.5507379e-01 2.5219360e-01 2.3674914e-01
-5.3583372e-01 5.9668112e-01 1.3698589e+00 6.4536557e-02
-3.3275405e-01 -2.9189548e-01 -2.0439123e-01 9.9933648e-01
6.6684902e-01 -2.3176113e-01 -4.4031036e-01 -1.0275241e+00
8.1903026e-02 4.0527141e-01 7.7748388e-01 -1.3877451e+00
3.6088938e-01 -4.9993986e-01 4.6338457e-01 -1.2316022e+00
9.9567920e-01 -1.2142920e+00 6.3756621e-01 6.0408461e-01
7.5423196e-02 5.6198221e-01 6.8018061e-01 7.4299943e-01
4.8803583e-01 8.0971718e-02 2.7518108e-01 -5.4694796e-01
-7.1491700e-01 2.1971859e-01 -6.2031668e-01 -3.8043422e-01
1.2635726e+00 -5.0251901e-01 -3.3725154e-01 -2.8363883e-01
-6.9578844e-01 5.6498796e-01 1.1323184e+00 3.3455241e-01
8.9719146e-01 -9.5592523e-01 -4.0049839e-01 1.1585876e-01
-7.9095233e-03 1.2065710e+00 1.8494184e-01 4.5413825e-01
-1.3520633e+00 4.1219947e-01 2.5049157e-02 -1.0436803e+00
-8.8932914e-01 6.2743437e-01 1.9897029e-01 7.1362451e-02
-1.0724543e+00 6.8586743e-01 5.6631398e-01 -1.0886166e+00
1.5866745e-01 -8.7511444e-01 4.3896034e-01 -8.2668406e-01
1.4040695e-01 4.9133649e-01 -1.2888649e-01 -5.3095651e-01
-4.5029145e-01 6.1163735e-01 2.0840041e-01 -4.1822839e-01
1.2796280e+00 -2.3344003e-01 6.7643508e-02 2.5392050e-01
7.7244204e-01 3.0053718e-02 -1.5801526e+00 3.7168616e-01
-1.7080984e-01 -5.7346946e-01 -5.1653302e-01 -7.2498715e-01
-2.3479274e-01 4.3082717e-01 3.0259672e-01 -3.9661482e-01
3.4765401e-01 1.6652697e-01 6.9840413e-01 9.5815432e-01
1.3010097e+00 -7.0405346e-01 3.7363339e-02 8.0776674e-01
1.3456739e+00 -8.4487683e-01 1.1175814e-02 -7.9426974e-01
-4.5367190e-01 9.4568723e-01 9.2372483e-01 -5.5223823e-01
2.4182715e-01 9.1526806e-01 -1.7938837e-01 -4.0529454e-01
-6.9055873e-01 2.4387331e-01 -4.6479455e-01 9.2995661e-01
-3.2617009e-01 3.5417192e-02 4.6122321e-01 7.7065162e-02
-7.8986621e-01 -1.4476144e-01 6.8656802e-01 1.4021512e+00
-7.0748281e-01 -6.3437831e-01 -7.5152928e-01 2.6335156e-01
4.5846584e-01 1.6646135e-01 -4.6016008e-02 1.1058044e+00
-9.6521966e-02 7.7061892e-01 6.3054895e-01 -2.9734248e-01
7.8244436e-01 -5.3470230e-01 6.4140916e-01 -8.4944183e-01
-2.9918283e-01 -6.1363243e-02 2.3141272e-01 -1.2909503e+00
1.1606343e-01 -6.8556368e-01 -1.9421051e+00 -4.3893260e-01
-3.6536533e-02 -1.6258601e-02 8.9518303e-01 4.3455496e-01
6.6186750e-01 5.1580536e-01 1.2830486e-02 -1.9679749e+00
-2.6523525e-01 -5.4853684e-01 -9.2551909e-02 3.8971141e-02
3.6386177e-01 -9.2870790e-01 -2.3723288e-02 3.7404276e-02] | [4.77144193649292, 0.6511719822883606] |
306aa651-aab0-48d9-a1e3-458c16794b14 | rethinking-semi-supervised-medical-image | 2302.01735 | null | https://arxiv.org/abs/2302.01735v4 | https://arxiv.org/pdf/2302.01735v4.pdf | Rethinking Semi-Supervised Medical Image Segmentation: A Variance-Reduction Perspective | For medical image segmentation, contrastive learning is the dominant practice to improve the quality of visual representations by contrasting semantically similar and dissimilar pairs of samples. This is enabled by the observation that without accessing ground truth labels, negative examples with truly dissimilar anatomical features, if sampled, can significantly improve the performance. In reality, however, these samples may come from similar anatomical regions and the models may struggle to distinguish the minority tail-class samples, making the tail classes more prone to misclassification, both of which typically lead to model collapse. In this paper, we propose ARCO, a semi-supervised contrastive learning (CL) framework with stratified group theory for medical image segmentation. In particular, we first propose building ARCO through the concept of variance-reduced estimation and show that certain variance-reduction techniques are particularly beneficial in pixel/voxel-level segmentation tasks with extremely limited labels. Furthermore, we theoretically prove these sampling techniques are universal in variance reduction. Finally, we experimentally validate our approaches on eight benchmarks, i.e., five 2D/3D medical and three semantic segmentation datasets, with different label settings, and our methods consistently outperform state-of-the-art semi-supervised methods. Additionally, we augment the CL frameworks with these sampling techniques and demonstrate significant gains over previous methods. We believe our work is an important step towards semi-supervised medical image segmentation by quantifying the limitation of current self-supervision objectives for accomplishing such challenging safety-critical tasks. | ['James S Duncan', 'Lawrence Hamilton Staib', 'S Kevin Zhou', 'David A. Clifton', 'Fenglin Liu', 'Yifei Min', 'Weicheng Dai', 'Chenyu You'] | 2023-02-03 | null | null | null | null | ['semi-supervised-medical-image-segmentation'] | ['computer-vision'] | [ 7.12290823e-01 4.33108151e-01 -5.92059076e-01 -4.83143479e-01
-1.11162543e+00 -4.07205462e-01 4.18249846e-01 4.64277267e-01
-3.61690074e-01 6.78682983e-01 -6.63519949e-02 -2.10193306e-01
-2.03812987e-01 -4.26942736e-01 -6.44828081e-01 -9.31368768e-01
-1.38592049e-01 6.74611390e-01 2.40144342e-01 5.51170520e-02
8.53964686e-02 3.86107415e-01 -1.27222788e+00 3.14436466e-01
1.06994998e+00 1.03288531e+00 8.60785171e-02 1.78801835e-01
-5.72286695e-02 7.77655959e-01 -2.71350950e-01 -2.37942874e-01
2.46107876e-01 -7.18250155e-01 -1.11460280e+00 4.09809023e-01
5.98729134e-01 -1.32000640e-01 1.66471079e-01 1.38017607e+00
4.28861856e-01 -2.26630121e-02 9.62986529e-01 -1.24596643e+00
-3.02215993e-01 6.27535641e-01 -7.91321218e-01 1.40235916e-01
-6.64508566e-02 1.66034192e-01 9.59748328e-01 -5.27055323e-01
6.58639848e-01 1.13020992e+00 7.82726884e-01 7.66837537e-01
-1.64135063e+00 -4.83214706e-01 1.80577487e-01 -1.33110583e-01
-1.15329230e+00 -2.82427996e-01 8.52667332e-01 -6.35169864e-01
2.61462092e-01 5.35444081e-01 4.91466880e-01 9.82746422e-01
4.81472798e-02 1.22406197e+00 1.64852595e+00 -3.29032034e-01
3.85319352e-01 1.48659974e-01 3.44834298e-01 8.25525641e-01
1.91797510e-01 4.09650430e-02 -9.28906426e-02 -1.77909568e-01
6.46575332e-01 5.95140904e-02 -4.02953893e-01 -8.70496154e-01
-1.15029967e+00 7.40999877e-01 8.28545868e-01 2.56556600e-01
-2.24850416e-01 7.80687481e-03 5.63256562e-01 2.05009780e-03
6.97425663e-01 4.82491583e-01 -1.89543992e-01 4.56155986e-01
-1.26670802e+00 -1.57884564e-02 4.56621677e-01 7.77221382e-01
5.99221230e-01 -3.00354987e-01 -3.59683901e-01 8.82562876e-01
2.35647768e-01 2.26805821e-01 4.76062298e-01 -9.50480461e-01
1.49684042e-01 5.78854859e-01 -2.82758057e-01 -7.33972430e-01
-4.91915613e-01 -6.81392372e-01 -1.04234231e+00 3.43733221e-01
7.06661344e-01 1.99848875e-01 -1.27636921e+00 1.78175974e+00
3.64281476e-01 -5.09070493e-02 -1.77163452e-01 8.89624715e-01
7.42594719e-01 -4.19644788e-02 2.73309141e-01 -5.45662522e-01
1.29260743e+00 -1.08732307e+00 -6.84778869e-01 -2.67515838e-01
8.68743896e-01 -3.72681320e-01 1.25945961e+00 3.59791517e-01
-1.14068413e+00 -4.77351218e-01 -9.84076858e-01 2.36224353e-01
-9.84901749e-03 -1.32946238e-01 7.52625823e-01 7.35537767e-01
-7.67405033e-01 8.90560031e-01 -1.11889327e+00 -2.27210224e-02
1.17280960e+00 2.04469055e-01 -2.18209863e-01 -5.17625324e-02
-8.50852370e-01 5.53946614e-01 2.49972627e-01 5.64429946e-02
-1.00415015e+00 -9.64322150e-01 -7.84564614e-01 -2.66749471e-01
5.57602286e-01 -4.99173701e-01 9.29598510e-01 -1.24140501e+00
-9.03239489e-01 1.45408535e+00 -7.91056454e-02 -5.64289391e-01
1.06219280e+00 1.52577087e-01 3.59139731e-03 4.07548368e-01
4.02587265e-01 8.55055153e-01 8.48779678e-01 -1.71238935e+00
-3.63005787e-01 -6.92695379e-01 -1.31495342e-01 1.64218470e-01
-1.13906875e-01 -3.83220583e-01 -2.64196336e-01 -6.78845227e-01
3.51289809e-01 -9.47649896e-01 -7.32231736e-01 3.88679415e-01
-6.86501145e-01 9.24391076e-02 2.69341320e-01 -4.48087186e-01
8.98367584e-01 -2.32591367e+00 4.07339353e-03 3.52671325e-01
6.04690015e-01 1.63735747e-02 2.53021330e-01 -3.31067741e-01
-1.85004368e-01 3.08274090e-01 -8.22234511e-01 -4.58355665e-01
-1.33568138e-01 1.93073899e-01 -2.08545569e-02 7.04971671e-01
1.95535228e-01 8.61933708e-01 -1.18333709e+00 -1.04835093e+00
2.39041358e-01 1.55024290e-01 -4.68191624e-01 2.68325098e-02
-5.12389317e-02 8.93835723e-01 -4.39851820e-01 6.18758976e-01
6.15831733e-01 -5.17030239e-01 1.63044289e-01 -3.36000383e-01
4.39814657e-01 6.57557091e-03 -9.11500335e-01 1.64612746e+00
-3.51644665e-01 2.71153241e-01 7.67669976e-02 -1.40280998e+00
5.57636440e-01 9.42459852e-02 6.84282720e-01 -7.38155067e-01
1.91371530e-01 3.76966774e-01 -7.68002719e-02 -3.61093670e-01
2.33522765e-02 -5.09261906e-01 1.16678111e-01 1.99859470e-01
-1.15587689e-01 -4.02084768e-01 1.28932834e-01 1.72994852e-01
7.66731143e-01 -1.00983284e-01 4.16959226e-01 -5.07198095e-01
4.23212230e-01 5.59088215e-02 5.99656403e-01 8.91657472e-01
-7.23970056e-01 9.11185086e-01 8.31860125e-01 -7.29316473e-02
-7.21678734e-01 -1.39486909e+00 -6.34481549e-01 6.44549549e-01
2.90364891e-01 -8.42790864e-03 -1.01102638e+00 -1.22902989e+00
-1.43438846e-01 5.63925862e-01 -9.51491714e-01 -1.79446250e-01
-4.01920676e-01 -9.43823934e-01 4.68810111e-01 6.68753982e-01
1.97973683e-01 -9.11280751e-01 -6.15291834e-01 -8.08663741e-02
-1.53214604e-01 -1.02819371e+00 -4.03466195e-01 5.80866337e-01
-1.23958087e+00 -1.20041263e+00 -8.91624033e-01 -9.05463159e-01
1.14394605e+00 1.25436261e-01 1.27462828e+00 2.33444884e-01
-4.37864006e-01 1.72743380e-01 -2.45940566e-01 -2.07049191e-01
-6.19568467e-01 -3.54385898e-02 -3.86631727e-01 -1.30960748e-01
-1.13494202e-01 -3.55978310e-01 -7.30183184e-01 4.08746928e-01
-8.92115414e-01 2.30908632e-01 5.44629335e-01 1.15777934e+00
1.17275202e+00 -1.82391137e-01 4.45665061e-01 -1.44013715e+00
1.94412023e-01 -3.84531409e-01 -2.06403583e-01 2.94511348e-01
-7.19769537e-01 3.19186486e-02 5.73961437e-01 -4.20073062e-01
-7.51990676e-01 1.37543574e-01 -4.29566838e-02 -3.87156397e-01
-1.82996556e-01 3.64479989e-01 7.97440410e-02 -4.29679975e-02
8.56680095e-01 1.03721336e-01 4.37314123e-01 -1.88119322e-01
2.14696318e-01 4.74895686e-01 4.94737357e-01 -5.38501143e-01
4.46948498e-01 1.11122358e+00 1.95726812e-01 -5.18323183e-01
-1.18556964e+00 -5.36305010e-01 -6.92818880e-01 -2.87556916e-01
9.01096761e-01 -6.57757163e-01 -3.22507441e-01 2.90370226e-01
-4.73049074e-01 -6.12622619e-01 -6.41497016e-01 2.83155233e-01
-7.99306214e-01 6.97168827e-01 -5.41711748e-01 -6.39517307e-01
-2.24750951e-01 -1.73742759e+00 1.24240458e+00 1.24291241e-01
-2.44131789e-01 -1.24703443e+00 -1.85961410e-01 5.33943772e-01
1.15725398e-01 4.75202292e-01 1.02940249e+00 -8.58131766e-01
-1.76349625e-01 1.26094639e-01 -1.70986176e-01 6.75876975e-01
3.63418162e-01 -3.03422868e-01 -9.71855342e-01 -4.81046200e-01
7.99125731e-02 -6.04257345e-01 1.04907346e+00 7.74034381e-01
1.66673899e+00 1.25752836e-01 -5.22891700e-01 7.06705451e-01
1.36357152e+00 -1.47209987e-01 4.12225842e-01 -2.86531840e-02
8.05845976e-01 8.09843302e-01 7.89881945e-01 4.07129601e-02
1.41444936e-01 5.22007525e-01 5.39264143e-01 -6.33014500e-01
-4.00914371e-01 -9.68373790e-02 -1.53969526e-01 5.60903490e-01
1.64366975e-01 1.14337035e-01 -9.45902050e-01 7.11238444e-01
-1.64071178e+00 -5.34082770e-01 -3.19504499e-01 2.28788710e+00
1.22418606e+00 3.17785561e-01 1.78702593e-01 2.82280326e-01
7.16373146e-01 1.67911232e-01 -7.22839475e-01 1.13440298e-01
7.86511078e-02 3.67878377e-01 6.48504317e-01 4.06467527e-01
-1.40753925e+00 7.41730571e-01 6.00827265e+00 1.18134797e+00
-1.06930244e+00 2.35600442e-01 1.27958775e+00 2.53172149e-03
-3.29473644e-01 -9.88882184e-02 -3.80885631e-01 5.19270003e-01
4.73063231e-01 2.30192333e-01 8.23433474e-02 9.24839020e-01
-2.65000220e-02 -1.84408501e-01 -1.36541367e+00 9.58588064e-01
8.08494911e-02 -1.18744946e+00 -1.32737711e-01 7.98536930e-03
8.43642056e-01 -1.80794328e-01 1.57840282e-01 6.05721846e-02
1.56311706e-01 -1.21923757e+00 6.66510582e-01 2.31957540e-01
7.81411707e-01 -5.48272014e-01 6.93492293e-01 1.69136867e-01
-8.01282227e-01 2.33149245e-01 -2.26654470e-01 4.70788389e-01
1.44081831e-01 8.08084249e-01 -6.69771552e-01 6.14017069e-01
7.01565564e-01 7.58964241e-01 -6.82325006e-01 9.22970772e-01
-7.69542754e-02 7.16599047e-01 -2.25318782e-02 2.93818146e-01
2.14669153e-01 -8.66664201e-02 3.76027763e-01 1.01804888e+00
-3.91633898e-01 -2.78456718e-01 4.12668675e-01 9.40187812e-01
4.60319687e-03 2.14387327e-01 -2.26728335e-01 2.62927353e-01
1.00507878e-01 1.29093432e+00 -1.34606874e+00 -5.48849285e-01
-1.41319424e-01 8.38189125e-01 2.43224472e-01 1.23880386e-01
-8.67894769e-01 2.15237923e-02 2.74669111e-01 1.40779287e-01
1.96022242e-02 2.88282484e-01 -6.62100792e-01 -8.98398936e-01
-2.46441197e-02 -9.73075032e-01 5.21268666e-01 -3.77179623e-01
-1.67441416e+00 4.22648549e-01 1.13562606e-01 -1.25312006e+00
2.58452110e-02 -5.29063821e-01 -1.58972263e-01 5.06431460e-01
-1.50885046e+00 -1.05085111e+00 -3.67967516e-01 4.34804916e-01
4.27057058e-01 3.63267392e-01 5.62514007e-01 3.16630721e-01
-3.48686367e-01 7.56888926e-01 -2.96032168e-02 2.25231737e-01
7.85802722e-01 -1.65721238e+00 7.79337361e-02 5.24887741e-01
1.46492645e-01 4.42827374e-01 6.08152568e-01 -6.17265403e-01
-7.51373172e-01 -1.06439853e+00 2.15978056e-01 -3.05469871e-01
5.90770543e-01 -3.29782218e-01 -1.08451259e+00 5.44470668e-01
-2.75182724e-01 5.14090180e-01 7.33388722e-01 -1.02757543e-01
-9.45749432e-02 6.82684183e-02 -1.43195593e+00 6.56598747e-01
1.28407979e+00 -4.46076781e-01 -5.58442831e-01 6.47127986e-01
4.59407300e-01 -5.10207117e-01 -1.00128460e+00 6.52345002e-01
2.76708037e-01 -1.08326042e+00 1.07937849e+00 -6.65395558e-01
5.28099597e-01 -1.30188867e-01 -2.04595421e-02 -1.21582425e+00
3.88885215e-02 -1.84552059e-01 2.20474556e-01 1.05522418e+00
4.37427104e-01 -5.43096304e-01 9.47076142e-01 3.85205209e-01
-3.52217942e-01 -9.82596517e-01 -9.02733445e-01 -8.75128686e-01
4.64312851e-01 -5.44537425e-01 1.05098344e-01 1.15122569e+00
-1.81729004e-01 -1.40917394e-02 -1.01324886e-01 -1.07096508e-01
9.60900605e-01 8.23471099e-02 3.96887064e-01 -1.15999079e+00
-3.74612898e-01 -6.19126797e-01 -5.32433987e-01 -8.51709723e-01
2.20274538e-01 -1.23375583e+00 2.97731727e-01 -1.38501310e+00
5.91157079e-01 -7.36333251e-01 -3.39128554e-01 4.61251348e-01
-4.73606616e-01 5.24731994e-01 -8.84145349e-02 3.20364624e-01
-5.81949413e-01 3.67472112e-01 1.70244682e+00 -3.81339043e-01
2.44586412e-02 1.56684667e-01 -6.58255994e-01 8.66412580e-01
5.61442196e-01 -6.06782973e-01 -6.09453321e-01 -9.52657836e-04
-1.42333865e-01 -1.21295914e-01 7.06080556e-01 -7.86984026e-01
-9.29264501e-02 8.24564397e-02 3.35801035e-01 -3.75696391e-01
-1.58506259e-01 -7.45144367e-01 -2.75387228e-01 8.18484664e-01
-7.38011599e-01 -5.08491635e-01 -4.95749339e-02 4.95641768e-01
-2.08158240e-01 -3.31794798e-01 1.15073133e+00 -2.62229919e-01
-4.03146893e-01 3.33070397e-01 1.33047746e-02 6.53754890e-01
1.03635204e+00 -3.07776690e-01 -4.93080430e-02 -3.31904069e-02
-1.02867007e+00 3.52163106e-01 5.09291053e-01 -5.20983078e-02
3.91543299e-01 -1.12609780e+00 -5.52133203e-01 6.35140166e-02
2.31065720e-01 3.61353934e-01 4.26207781e-01 1.44322109e+00
-4.69324172e-01 -4.04945798e-02 -1.03753388e-01 -1.18464518e+00
-1.20300353e+00 6.08948350e-01 4.98360723e-01 -5.15606284e-01
-6.30718410e-01 8.06269050e-01 6.17879152e-01 -4.52551603e-01
3.55134666e-01 -5.40433824e-01 -1.33827910e-01 1.05087936e-01
2.96062261e-01 1.19615763e-01 1.50043443e-01 -4.47295427e-01
-5.61488271e-01 4.59282845e-01 -2.71878392e-01 9.62305740e-02
1.12311924e+00 4.13112082e-02 4.43459563e-02 5.50508022e-01
1.28522682e+00 -1.89143389e-01 -1.39882231e+00 -2.16081545e-01
5.57381772e-02 -3.13794047e-01 1.61433652e-01 -7.46673048e-01
-1.28750789e+00 9.44825232e-01 9.06043291e-01 2.53081322e-01
1.11029005e+00 3.92991185e-01 5.29881835e-01 -1.08093381e-01
3.66684735e-01 -1.11207044e+00 7.93988258e-02 -2.86725521e-01
4.79976952e-01 -1.65848601e+00 1.94018036e-01 -7.64106154e-01
-9.02464151e-01 6.83303475e-01 4.02654141e-01 -1.31223679e-01
5.35964668e-01 2.22338527e-01 9.94978175e-02 -3.72328728e-01
-2.65088558e-01 -3.73210251e-01 4.64227259e-01 6.60390675e-01
4.61201221e-01 2.45092645e-01 -4.14407700e-01 4.04364347e-01
-9.47576948e-04 -1.94561288e-01 2.48618379e-01 9.01598155e-01
-6.38375133e-02 -8.36083174e-01 -1.56075925e-01 7.08460152e-01
-5.99034309e-01 -1.41973779e-01 -2.74299204e-01 1.00236881e+00
-6.32621273e-02 5.55633724e-01 1.24017842e-01 -9.73300356e-03
1.00018606e-01 -1.62427336e-01 6.84419692e-01 -6.37289822e-01
-6.29886270e-01 1.81377739e-01 -2.03070104e-01 -6.54353142e-01
-5.60249567e-01 -7.94753313e-01 -1.47237372e+00 2.84488410e-01
-3.31703573e-01 -1.09832533e-01 3.40394437e-01 1.01656425e+00
-1.26049578e-01 7.77361393e-01 6.68989241e-01 -6.36144459e-01
-7.79791355e-01 -6.43475175e-01 -5.58427334e-01 9.08859849e-01
3.60337347e-01 -8.38706791e-01 -6.97587192e-01 -2.29894314e-02] | [14.690028190612793, -2.152475118637085] |
ebcbd0d1-3f08-401a-a1a8-fe561c0fe5c7 | spatio-temporal-transformer-for-dynamic | 2205.04749 | null | https://arxiv.org/abs/2205.04749v1 | https://arxiv.org/pdf/2205.04749v1.pdf | Spatio-Temporal Transformer for Dynamic Facial Expression Recognition in the Wild | Previous methods for dynamic facial expression in the wild are mainly based on Convolutional Neural Networks (CNNs), whose local operations ignore the long-range dependencies in videos. To solve this problem, we propose the spatio-temporal Transformer (STT) to capture discriminative features within each frame and model contextual relationships among frames. Spatio-temporal dependencies are captured and integrated by our unified Transformer. Specifically, given an image sequence consisting of multiple frames as input, we utilize the CNN backbone to translate each frame into a visual feature sequence. Subsequently, the spatial attention and the temporal attention within each block are jointly applied for learning spatio-temporal representations at the sequence level. In addition, we propose the compact softmax cross entropy loss to further encourage the learned features have the minimum intra-class distance and the maximum inter-class distance. Experiments on two in-the-wild dynamic facial expression datasets (i.e., DFEW and AFEW) indicate that our method provides an effective way to make use of the spatial and temporal dependencies for dynamic facial expression recognition. The source code and the training logs will be made publicly available. | ['Shutao Li', 'Bin Sun', 'Fuyan Ma'] | 2022-05-10 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [ 1.11219451e-01 -4.23577428e-01 -2.93052107e-01 -8.98382366e-01
-3.13573241e-01 -2.78275013e-01 3.76079917e-01 -4.21878219e-01
-4.86244231e-01 4.54877853e-01 2.29719922e-01 1.02636479e-01
7.60799600e-03 -4.54628319e-01 -7.03436911e-01 -9.31986749e-01
-2.43822187e-01 -4.25308168e-01 -1.16361361e-02 -9.47391763e-02
-8.42942297e-02 6.54075921e-01 -1.37759686e+00 5.29514968e-01
4.60666448e-01 1.51354706e+00 2.39167169e-01 3.83911133e-01
-1.46618411e-01 1.28582156e+00 -3.46992552e-01 -3.55485767e-01
-7.83160031e-02 -5.42312324e-01 -5.94130456e-01 3.45834345e-01
2.79189497e-01 -5.82900107e-01 -6.23947442e-01 8.67650688e-01
4.68629688e-01 2.34708473e-01 2.93880165e-01 -1.48566544e+00
-5.71688294e-01 -1.59492251e-02 -8.27811539e-01 3.40832323e-01
2.68633038e-01 2.64948159e-01 9.14930642e-01 -1.09363210e+00
5.76635420e-01 1.20816433e+00 4.37218547e-01 4.85987008e-01
-9.44289565e-01 -9.78493452e-01 5.46313047e-01 5.59234977e-01
-1.39258230e+00 -7.39509523e-01 9.50128734e-01 -3.26983303e-01
9.92384195e-01 -4.83668670e-02 8.53019059e-01 1.17374480e+00
1.77741125e-01 1.04380870e+00 6.71416163e-01 -1.35693803e-01
-4.45137694e-02 -4.13037390e-01 -2.33028635e-01 8.17732215e-01
-8.02375436e-01 1.22599497e-01 -8.45870912e-01 3.38096730e-02
7.72834718e-01 2.50498384e-01 -2.18752742e-01 -2.13410854e-01
-7.68953323e-01 8.70986283e-01 5.29856205e-01 3.36047083e-01
-5.28632343e-01 3.12088192e-01 7.22907841e-01 2.07873881e-01
6.52271569e-01 -2.94423491e-01 -4.70119804e-01 -3.45004261e-01
-8.29734862e-01 7.44667202e-02 2.68365532e-01 7.57724583e-01
8.28352988e-01 -2.12643184e-02 -3.19631904e-01 1.03051841e+00
2.82873988e-01 3.02259773e-01 3.79664510e-01 -1.05876112e+00
3.16188633e-01 4.24519390e-01 -2.06547379e-01 -1.28091657e+00
-1.47519797e-01 -6.37282953e-02 -7.75222361e-01 3.23918685e-02
1.08211145e-01 -1.21893048e-01 -7.95163751e-01 2.11312556e+00
2.92037040e-01 6.76929116e-01 -2.11568609e-01 9.36678350e-01
5.54468930e-01 7.78385401e-01 3.78226191e-01 -4.47490990e-01
1.22454917e+00 -1.14194596e+00 -9.69962955e-01 -1.02167629e-01
5.53540170e-01 -5.68237484e-01 9.55769122e-01 4.28902172e-02
-1.06818986e+00 -6.50018036e-01 -7.22341180e-01 -2.48719633e-01
-9.52240899e-02 2.37842679e-01 4.84077156e-01 1.03185587e-02
-9.02455688e-01 3.18111509e-01 -1.14621794e+00 -1.06889717e-01
7.63381302e-01 3.26189101e-01 -6.29557431e-01 -8.19430351e-02
-1.12571132e+00 5.07975876e-01 1.59615483e-02 5.35364509e-01
-8.76702905e-01 -4.30739760e-01 -1.16662788e+00 2.62625925e-02
2.12843865e-01 -2.69540697e-01 1.15613937e+00 -1.70732617e+00
-1.71389639e+00 7.88929701e-01 -6.17610931e-01 -1.56975016e-01
2.42672235e-01 -3.62584978e-01 -3.95386815e-01 4.13573086e-01
2.10911431e-03 6.99515581e-01 9.13656294e-01 -8.16806614e-01
-5.56239367e-01 -4.35773373e-01 1.05134495e-01 3.29556055e-02
-6.70261025e-01 4.20779020e-01 -8.18252504e-01 -8.15819621e-01
-1.54915631e-01 -8.59303474e-01 2.24644225e-02 3.89586657e-01
-5.55440187e-02 -2.55644083e-01 1.17445982e+00 -7.47029126e-01
1.47999406e+00 -2.49469638e+00 3.36584479e-01 1.07364364e-01
-4.23757099e-02 2.67879456e-01 -3.94349784e-01 3.78301442e-02
-2.28746757e-01 -1.65183827e-01 -7.18242452e-02 -5.10445535e-01
-2.31990829e-01 5.13926029e-01 -2.29462013e-01 5.10893643e-01
4.80210900e-01 9.11312282e-01 -8.95188808e-01 -5.81038058e-01
1.64710224e-01 7.45373189e-01 -6.02623761e-01 4.24027979e-01
-1.37286350e-01 5.37775397e-01 -5.56024313e-01 5.34438252e-01
6.75001562e-01 -2.13041827e-01 9.29603651e-02 -3.17132711e-01
6.39762450e-03 7.28554130e-02 -6.07784808e-01 1.91212010e+00
-5.53494334e-01 8.30096304e-01 1.33123562e-01 -1.09712481e+00
7.76497424e-01 3.13091904e-01 7.38385022e-01 -1.06468928e+00
1.50435135e-01 6.24096356e-02 -2.69657612e-01 -9.24163520e-01
2.43838400e-01 -2.18845561e-01 1.62131548e-01 2.61597961e-01
2.39254937e-01 6.44205213e-01 8.37918967e-02 -3.61289643e-02
9.41634059e-01 4.46664780e-01 5.95268048e-02 1.60848260e-01
6.22194946e-01 -6.43576682e-01 9.48791623e-01 9.81063172e-02
-3.56429309e-01 4.56900775e-01 7.27105439e-01 -6.75596833e-01
-8.38750184e-01 -7.84702957e-01 1.16105296e-01 1.49918175e+00
-1.72837228e-02 -5.43327093e-01 -5.98700643e-01 -7.66391516e-01
-1.10315301e-01 2.22214654e-01 -9.87909198e-01 -2.50851363e-01
-7.47919858e-01 -3.18391860e-01 3.52577597e-01 7.86723971e-01
6.03460848e-01 -1.31383491e+00 -6.46600485e-01 2.36890972e-01
-3.47006291e-01 -1.28745353e+00 -9.03613865e-01 -5.01138903e-02
-5.04285991e-01 -8.69712710e-01 -7.09855020e-01 -7.24498689e-01
5.09226322e-01 1.08330637e-01 8.14588130e-01 1.55513465e-01
-1.89310551e-01 1.08761549e-01 -5.40500939e-01 1.16768494e-01
1.77662507e-01 -1.96957245e-01 -2.72555262e-01 5.28324008e-01
5.50863862e-01 -7.35982597e-01 -7.05699027e-01 3.53105932e-01
-8.71820390e-01 1.61040321e-01 2.27787748e-01 1.07261169e+00
5.93837917e-01 -3.46978813e-01 2.72836298e-01 -4.57709491e-01
8.97542015e-02 -5.47090590e-01 -3.33491743e-01 2.53635734e-01
1.15625218e-01 -1.85891539e-01 4.31787491e-01 -5.95591366e-01
-1.00467551e+00 2.16694936e-01 -2.79079825e-01 -1.08252847e+00
6.10591173e-02 5.68996370e-01 -4.42826688e-01 9.57244858e-02
6.81736246e-02 2.99473405e-01 3.75014059e-02 -2.98697680e-01
5.91587760e-02 4.66643095e-01 4.21614170e-01 -4.55067277e-01
2.35488936e-01 6.02670610e-01 -1.44363701e-01 -7.38151014e-01
-1.00015032e+00 -2.77490139e-01 -6.65909767e-01 -3.16083044e-01
1.22371030e+00 -1.03803647e+00 -7.51862645e-01 7.31220543e-01
-1.21255779e+00 -5.42021573e-01 5.58883650e-03 3.42538834e-01
-6.27469599e-01 2.44699642e-01 -7.83452570e-01 -7.25983143e-01
-9.74869356e-02 -1.26417947e+00 1.31436145e+00 2.94509917e-01
-9.40815508e-02 -1.00932789e+00 -1.14478536e-01 6.96032271e-02
2.99202204e-01 4.53550339e-01 6.91336215e-01 -5.58834933e-02
-3.26317549e-01 2.55697221e-02 -3.40443462e-01 5.52716851e-01
2.90874422e-01 3.91987234e-01 -9.19708550e-01 -2.80835748e-01
-2.15216935e-01 -5.63396096e-01 9.59945858e-01 4.23761606e-01
1.69670689e+00 -4.13018286e-01 -2.30989113e-01 9.91340220e-01
1.04281652e+00 2.94327706e-01 7.81452775e-01 1.98812693e-01
7.55876720e-01 6.57831907e-01 6.76610708e-01 7.49158502e-01
2.98005402e-01 1.05308723e+00 3.77131224e-01 -2.18556643e-01
1.64970204e-01 -3.09471846e-01 6.06065989e-01 5.59553862e-01
-1.28555939e-01 -1.32997051e-01 -5.20071268e-01 3.83956313e-01
-1.95625663e+00 -1.20897639e+00 5.06406903e-01 1.82841873e+00
7.68649876e-01 -2.60214180e-01 4.82715778e-02 -2.27361545e-01
5.58980048e-01 5.42029560e-01 -5.29761195e-01 -3.43466580e-01
-2.01782838e-01 2.25074932e-01 -1.08077386e-02 4.53308135e-01
-1.18394721e+00 1.05845416e+00 5.46336317e+00 8.84160280e-01
-1.55445290e+00 1.18185535e-01 9.29653406e-01 -3.97432685e-01
-6.94304556e-02 -2.04783753e-01 -5.10346949e-01 7.43933916e-01
8.37565243e-01 1.22166835e-01 1.94006637e-01 7.35491514e-01
5.07449389e-01 1.20904393e-01 -8.56630206e-01 1.05982077e+00
7.14946166e-02 -1.17941999e+00 3.19211595e-02 8.60259682e-02
5.70025623e-01 -1.60297468e-01 2.08127901e-01 2.49238446e-01
-7.54011562e-03 -1.12603378e+00 8.19047451e-01 6.73201561e-01
9.46750224e-01 -9.24663723e-01 5.00235438e-01 2.98157055e-02
-1.43960106e+00 -2.05826789e-01 -2.43978947e-01 1.07494473e-01
3.01925272e-01 3.29861343e-01 -5.98832071e-02 2.77322888e-01
8.99816871e-01 1.32162952e+00 -2.58800417e-01 5.51255763e-01
-2.46070012e-01 4.28303659e-01 -1.50006682e-01 2.90625960e-01
4.17396814e-01 -2.90107429e-01 1.55781537e-01 1.33374536e+00
4.09141630e-01 2.55115300e-01 1.02257378e-01 4.99647886e-01
-9.65907052e-02 -4.15052213e-02 -4.06893373e-01 -1.14272036e-01
9.46901292e-02 1.20786595e+00 -2.57169873e-01 -2.05369025e-01
-6.68024361e-01 1.37830758e+00 5.82206607e-01 6.04112804e-01
-1.11356997e+00 -2.04488352e-01 1.20987654e+00 -9.09727961e-02
6.20356441e-01 -1.87600911e-01 2.34804928e-01 -1.09214675e+00
2.23745108e-01 -8.04565310e-01 4.75409031e-01 -8.71902645e-01
-1.20214415e+00 6.47509992e-01 -1.82291195e-01 -1.13704336e+00
-3.59832942e-01 -6.21215165e-01 -6.22626007e-01 7.23972440e-01
-1.58432889e+00 -1.25028646e+00 -4.37310040e-01 9.56072807e-01
5.11430085e-01 -2.30091438e-02 7.23447859e-01 5.03410101e-01
-6.84435248e-01 7.20618367e-01 -1.00513563e-01 4.62182552e-01
6.84763968e-01 -6.91233456e-01 -9.87731740e-02 6.06217861e-01
1.30993649e-01 4.57108587e-01 3.44636649e-01 -1.94823921e-01
-1.05011463e+00 -1.08270967e+00 9.05204117e-01 -8.93857405e-02
7.11783767e-01 -5.25458694e-01 -1.10561776e+00 7.95348942e-01
8.94307047e-02 7.39233136e-01 7.66006529e-01 -5.84621057e-02
-6.49480820e-01 -4.00317132e-01 -7.45423794e-01 4.01059240e-01
1.22130346e+00 -8.85524750e-01 -2.98718438e-02 1.30173504e-01
5.60739815e-01 -3.76935959e-01 -6.10381007e-01 5.75714767e-01
8.09859455e-01 -1.10548365e+00 6.56151414e-01 -7.72406876e-01
7.78816640e-01 -1.79319561e-01 -1.93267703e-01 -9.97400641e-01
-1.82953328e-01 -5.81618309e-01 -9.37443897e-02 1.20002997e+00
1.12427123e-01 -2.10738614e-01 6.69420898e-01 4.24761415e-01
-2.38982290e-02 -1.05445671e+00 -1.11552143e+00 -4.99573976e-01
-2.32072175e-01 -4.86910254e-01 3.79516691e-01 9.76409733e-01
-8.13559294e-02 1.62646890e-01 -7.05556929e-01 -6.81719482e-02
9.81730446e-02 2.15844274e-01 6.74515367e-01 -5.77320814e-01
-2.60209739e-01 -2.82693326e-01 -6.30568147e-01 -1.25739193e+00
7.71363735e-01 -5.52502096e-01 1.31551027e-01 -8.86594415e-01
4.18269098e-01 -1.46677792e-01 -5.35175383e-01 7.83119082e-01
-2.63314873e-01 2.94584364e-01 1.05752617e-01 9.41487476e-02
-7.85092354e-01 1.07449377e+00 1.28831220e+00 -6.64710477e-02
-3.00035775e-02 -2.79289663e-01 -2.17417479e-01 6.92833185e-01
4.50106502e-01 -2.84326643e-01 -4.38440442e-01 -5.67408025e-01
-1.12386763e-01 1.70838416e-01 4.66651648e-01 -6.79093778e-01
2.14095742e-01 -4.37543720e-01 5.33798695e-01 -2.88846970e-01
6.50136054e-01 -8.68836939e-01 5.54120280e-02 1.76296756e-01
-4.87954944e-01 1.99183181e-01 2.84201473e-01 5.02803385e-01
-7.62846708e-01 2.95719564e-01 9.13240135e-01 2.21883953e-01
-8.08154881e-01 9.19396818e-01 -1.65763617e-01 -2.24318609e-01
1.15438426e+00 -1.23619467e-01 -1.81670003e-02 -5.94212472e-01
-9.07576859e-01 2.32044175e-01 3.07962894e-01 5.18274903e-01
7.35622823e-01 -1.54989994e+00 -5.60952544e-01 3.56446862e-01
2.67089397e-01 -2.27818057e-01 5.96186697e-01 9.60588157e-01
-2.73017198e-01 1.38028905e-01 -5.37372351e-01 -6.17543697e-01
-1.57582247e+00 4.41547364e-01 6.33696318e-01 -1.34444848e-01
-5.00460744e-01 1.11732948e+00 6.61202550e-01 3.78497243e-02
2.61812776e-01 -7.53312185e-02 -1.19470775e-01 1.09523520e-01
7.72018135e-01 -2.10244834e-01 -3.22212428e-01 -1.12515163e+00
-3.89198750e-01 5.81196010e-01 -1.17872559e-01 2.02932619e-02
1.48417091e+00 -2.30378628e-01 -3.59296203e-01 3.62054437e-01
1.89006865e+00 -2.62486696e-01 -1.91723812e+00 -4.29456055e-01
-1.16952814e-01 -7.85751939e-01 -5.29087037e-02 -2.91268438e-01
-1.60271192e+00 1.08213639e+00 5.94638824e-01 -2.77982116e-01
1.61963809e+00 3.63578461e-02 8.25982928e-01 -2.57011391e-02
-5.76695353e-02 -9.47963297e-01 5.84927559e-01 6.51899934e-01
9.63327289e-01 -1.00702357e+00 -3.08275491e-01 -2.90686011e-01
-6.75100982e-01 1.18915486e+00 7.54598022e-01 1.24812650e-04
7.88914621e-01 4.37090397e-01 1.30695906e-02 -1.42775908e-01
-9.01122093e-01 -8.49956796e-02 1.95961185e-02 3.72203231e-01
6.37052834e-01 -3.38024944e-01 6.37622848e-02 4.34661299e-01
2.17536420e-01 3.22263926e-01 -1.95492700e-01 8.83693278e-01
-8.51891935e-02 -9.89439845e-01 2.02490896e-01 1.75581612e-02
-5.41668832e-01 9.87391472e-02 -1.42072991e-01 6.00977242e-01
3.04685265e-01 6.83012962e-01 3.48943174e-01 -4.83447820e-01
2.68690348e-01 1.14095090e-02 4.77169603e-01 -2.92985022e-01
-3.37470025e-01 2.48833805e-01 -1.01119526e-01 -9.99311328e-01
-7.53738463e-01 -6.36434853e-01 -1.22905946e+00 -2.81107038e-01
4.67904322e-02 -1.57102793e-01 3.29958409e-01 9.33285475e-01
4.52138603e-01 4.60331440e-01 8.57011974e-01 -7.83755422e-01
-8.34958404e-02 -9.02599812e-01 -5.94164133e-01 6.75396204e-01
6.65415704e-01 -7.44034827e-01 -2.70264208e-01 4.18305337e-01] | [13.627209663391113, 1.7204114198684692] |
b72a652e-6a25-4df4-be41-1a040e7d2bcd | structure-aware-and-class-balanced-3d-object | 2205.12519 | null | https://arxiv.org/abs/2205.12519v2 | https://arxiv.org/pdf/2205.12519v2.pdf | Structure Aware and Class Balanced 3D Object Detection on nuScenes Dataset | 3-D object detection is pivotal for autonomous driving. Point cloud based methods have become increasingly popular for 3-D object detection, owing to their accurate depth information. NuTonomy's nuScenes dataset greatly extends commonly used datasets such as KITTI in size, sensor modalities, categories, and annotation numbers. However, it suffers from severe class imbalance. The Class-balanced Grouping and Sampling paper addresses this issue and suggests augmentation and sampling strategy. However, the localization precision of this model is affected by the loss of spatial information in the downscaled feature maps. We propose to enhance the performance of the CBGS model by designing an auxiliary network, that makes full use of the structure information of the 3D point cloud, in order to improve the localization accuracy. The detachable auxiliary network is jointly optimized by two point-level supervisions, namely foreground segmentation and center estimation. The auxiliary network does not introduce any extra computation during inference, since it can be detached at test time. | ['Mohan Trivedi', 'Akshay Rangesh', 'Savitha Srinivasan', 'Asfiya Baig', 'Sushruth Nagesh'] | 2022-05-25 | null | null | null | null | ['foreground-segmentation'] | ['computer-vision'] | [ 5.53238913e-02 7.79772028e-02 -3.21423709e-01 -4.76679116e-01
-5.76153040e-01 -5.67844987e-01 5.76586902e-01 1.32791117e-01
-5.07460237e-01 4.35833663e-01 -4.47011858e-01 -2.60752380e-01
-9.59165103e-04 -9.42353487e-01 -8.36349428e-01 -8.21276307e-01
1.31112486e-01 5.08609474e-01 8.10681224e-01 5.19186594e-02
2.27500692e-01 9.12351668e-01 -1.95386934e+00 -2.63176739e-01
1.08642292e+00 1.42190433e+00 3.99912238e-01 1.79327235e-01
-2.69617587e-01 3.99043471e-01 -4.41754043e-01 -3.38439345e-02
4.70075369e-01 1.05510704e-01 -9.59921777e-02 1.42035812e-01
5.43537259e-01 -5.57847857e-01 -2.01135591e-01 1.11725891e+00
4.53721911e-01 1.33867085e-01 5.15069067e-01 -1.64730394e+00
5.51596172e-02 1.04747809e-01 -8.62328470e-01 1.26090705e-01
-3.69314790e-01 1.87988400e-01 6.77610874e-01 -8.54760110e-01
4.46765989e-01 1.18451536e+00 7.15088904e-01 2.98428535e-01
-7.32487202e-01 -9.63244617e-01 4.22642857e-01 2.74876505e-01
-1.59998167e+00 -2.51413107e-01 1.01254070e+00 -4.37510401e-01
6.30705595e-01 1.86898425e-01 5.74150324e-01 8.04740012e-01
-3.10542643e-01 8.02071214e-01 8.42284560e-01 -9.03420895e-02
3.45453650e-01 2.29532123e-01 9.55150425e-02 7.27593899e-01
6.36182904e-01 7.20862076e-02 -4.95675594e-01 1.06458463e-01
7.64584661e-01 2.28294224e-01 8.75941217e-02 -8.54291856e-01
-1.07913554e+00 5.87262034e-01 6.43169224e-01 -5.41764349e-02
-9.17804465e-02 6.39760345e-02 1.30375504e-01 -3.09217572e-01
7.83181250e-01 1.58134177e-01 -4.24591601e-01 -1.00172520e-01
-7.81014681e-01 1.35107458e-01 2.01531336e-01 1.22465765e+00
1.02429330e+00 -1.74128875e-01 -1.71552654e-02 5.44860482e-01
3.40026259e-01 6.62424147e-01 -1.62885077e-02 -9.86614227e-01
7.42329836e-01 1.24696505e+00 1.18034646e-01 -1.09327960e+00
-4.98341292e-01 -6.05980635e-01 -7.91053593e-01 3.04576576e-01
3.56788635e-01 -4.75292057e-02 -1.13954043e+00 1.30214918e+00
9.05813515e-01 3.05870414e-01 -2.68464625e-01 1.19862032e+00
7.73251295e-01 3.86594951e-01 -1.91818535e-01 3.69452059e-01
1.06845808e+00 -6.94198012e-01 -3.69148850e-01 -4.93150324e-01
6.71859145e-01 -2.08858296e-01 8.23529601e-01 1.38147563e-01
-5.97484529e-01 -5.70337296e-01 -1.20989120e+00 -3.56465369e-01
-4.61439312e-01 5.18149436e-01 7.55544722e-01 6.26909733e-01
-4.98687059e-01 1.05978079e-01 -1.10351932e+00 -1.05060719e-01
8.63415360e-01 3.00616175e-01 -2.46279240e-01 -3.94691937e-02
-7.97232509e-01 8.22051108e-01 4.70505416e-01 3.20568085e-01
-6.31692410e-01 -6.64961219e-01 -9.00567174e-01 3.74893425e-03
6.23634040e-01 -3.23544741e-01 8.02299500e-01 -3.20024341e-01
-1.19267559e+00 8.33280087e-01 -1.19643383e-01 -3.22189689e-01
6.92114890e-01 -2.17608213e-02 -1.57680213e-01 1.26710236e-01
4.12275612e-01 8.65285516e-01 7.44365156e-01 -1.27747536e+00
-1.14766800e+00 -8.53406012e-01 1.08075693e-01 1.33784831e-01
-1.34243831e-01 -5.02558589e-01 -8.31297278e-01 -3.09443474e-01
6.48687720e-01 -9.19678509e-01 -3.25530171e-01 3.92107725e-01
-4.76191550e-01 -2.87162483e-01 9.59622979e-01 -3.08261931e-01
6.97547615e-01 -2.44781351e+00 -1.87931359e-01 2.75236547e-01
3.54399830e-01 4.77622785e-02 1.78865209e-01 -1.93645179e-01
4.23475444e-01 3.32468525e-02 -2.85309970e-01 -5.41551411e-01
1.00720175e-01 3.25191110e-01 -1.23999424e-01 7.00244009e-01
3.84583175e-01 5.63226104e-01 -7.00418830e-01 -6.23998284e-01
4.14818853e-01 3.97305369e-01 -5.45127869e-01 2.70482954e-02
-3.58580023e-01 4.36465830e-01 -9.02317882e-01 9.04973447e-01
1.14001238e+00 -1.10959053e-01 -3.02365482e-01 -2.75550842e-01
-4.36152220e-01 3.15802336e-01 -1.29835176e+00 1.64850163e+00
-7.83958733e-02 5.20871639e-01 1.36088282e-01 -8.95951331e-01
1.00597370e+00 -1.60958901e-01 3.97298127e-01 -7.48837948e-01
2.68452466e-01 2.44595140e-01 -1.16422817e-01 -3.79922330e-01
4.88806158e-01 3.11920971e-01 9.42170694e-02 -1.94980741e-01
-4.39660251e-01 -3.13381612e-01 8.03797916e-02 6.92151263e-02
8.33449543e-01 2.57202119e-01 -1.49758637e-01 -6.85304999e-02
3.35526437e-01 3.84501040e-01 8.59884441e-01 6.31864548e-01
-1.76510274e-01 4.18879688e-01 4.28983659e-01 -1.27389312e-01
-8.99186730e-01 -7.12913752e-01 -4.63406563e-01 5.90043247e-01
8.43904793e-01 1.03083095e-02 -4.48044926e-01 -6.80559099e-01
4.65537518e-01 6.25568271e-01 -5.69259882e-01 -2.04989657e-01
-4.30486381e-01 -6.63482428e-01 6.09576583e-01 7.22995162e-01
6.80523813e-01 -4.57284778e-01 -7.95834363e-01 -9.59735140e-02
-7.45221302e-02 -1.35416996e+00 1.70165319e-02 4.34338987e-01
-1.02768612e+00 -1.11938357e+00 -2.73853302e-01 -3.07327062e-01
8.89536083e-01 5.93855679e-01 6.14415228e-01 7.35278428e-02
-5.11916131e-02 -1.58942401e-01 -3.11787397e-01 -7.55087137e-01
9.78375822e-02 3.65983188e-01 5.59986793e-02 -5.03730476e-02
5.75291097e-01 -2.74207294e-01 -5.14285684e-01 5.27728021e-01
-5.86020231e-01 1.95902705e-01 6.82293832e-01 4.65405077e-01
9.01892722e-01 2.97969095e-02 3.14916760e-01 -6.34817004e-01
-3.64218533e-01 -3.90915453e-01 -1.05955601e+00 -2.27361187e-01
-4.50285345e-01 -1.63709283e-01 3.27359326e-02 -2.31328696e-01
-8.62273395e-01 1.93083972e-01 -2.55922619e-02 -6.71274424e-01
-2.55957514e-01 1.39148235e-01 -5.37972927e-01 -2.60312498e-01
4.00735259e-01 -3.33288759e-02 1.63159892e-02 -7.14834988e-01
1.30741879e-01 7.67743528e-01 4.74135309e-01 -3.00284207e-01
9.71187353e-01 7.45901346e-01 2.75802881e-01 -8.01208794e-01
-9.21239495e-01 -7.08147526e-01 -7.30188906e-01 -1.66269362e-01
6.55695975e-01 -1.22569180e+00 -6.59122944e-01 5.05894780e-01
-1.02498233e+00 -1.43972754e-01 -2.28597492e-01 5.98246753e-01
-1.21113837e-01 1.57233745e-01 -1.84178695e-01 -8.60623002e-01
1.95394382e-01 -1.08123791e+00 1.39323056e+00 1.37253448e-01
5.10978520e-01 -3.18999648e-01 -4.18939441e-01 4.63046789e-01
-4.74611633e-02 4.21113968e-01 6.78664625e-01 -2.94897288e-01
-1.12153518e+00 -4.50622261e-01 -6.12309813e-01 2.52353281e-01
9.99283791e-02 -7.25091249e-02 -1.14481807e+00 2.42850799e-02
-7.96784982e-02 4.17336524e-02 9.15587306e-01 2.72183925e-01
1.48572087e+00 1.32078588e-01 -7.08473206e-01 7.96563745e-01
1.08363497e+00 1.24305665e-01 1.48818508e-01 4.53899503e-01
1.08848977e+00 5.17708480e-01 1.07832181e+00 2.30318129e-01
6.18763268e-01 5.97161055e-01 1.10882497e+00 -1.95642471e-01
-1.81368977e-01 -1.59618989e-01 -2.08404642e-02 4.04166937e-01
-2.16926914e-02 -1.08840153e-01 -9.40087557e-01 5.80182552e-01
-1.83957791e+00 -5.71709692e-01 -4.13540184e-01 2.07302022e+00
5.93727589e-01 4.41900849e-01 -6.66344389e-02 2.04203263e-01
7.21565425e-01 4.77952212e-02 -9.98284936e-01 4.73961323e-01
-1.57365516e-01 -2.21616283e-01 1.10364377e+00 2.46492580e-01
-1.02695966e+00 8.46538067e-01 4.74394274e+00 8.55060756e-01
-9.91739750e-01 9.84432995e-02 4.04109955e-01 -1.74586847e-01
-1.05205774e-01 -1.21213645e-01 -1.34195924e+00 5.31540573e-01
3.88992995e-01 4.05370206e-01 -4.67447601e-02 1.19048381e+00
1.77997455e-01 -3.27714622e-01 -9.23305869e-01 9.92283165e-01
5.38403429e-02 -1.35094607e+00 -1.78502038e-01 1.44931093e-01
4.94422466e-01 4.25723106e-01 -2.19999522e-01 2.38048613e-01
-2.06466764e-01 -4.66931760e-01 9.87688541e-01 1.71713173e-01
7.92630136e-01 -8.93248796e-01 8.17165196e-01 7.99523532e-01
-1.11165857e+00 -8.42101648e-02 -5.16905189e-01 -1.33500293e-01
5.90374433e-02 7.96135664e-01 -6.42249167e-01 4.32011336e-01
8.38596523e-01 7.29187608e-01 -8.45645130e-01 1.17788279e+00
-2.07617983e-01 4.11831439e-01 -7.88449705e-01 3.96974944e-02
2.17096806e-01 -3.33736658e-01 5.96456826e-01 6.16995037e-01
2.85111248e-01 -3.43642710e-03 3.82324815e-01 9.23385799e-01
-3.22294161e-02 -3.32575232e-01 -4.94193643e-01 2.52046853e-01
8.28712940e-01 1.27528012e+00 -8.70126665e-01 -3.64649653e-01
-3.17451030e-01 4.44292516e-01 2.29315296e-01 1.42082408e-01
-8.95120561e-01 -4.92702723e-01 8.40437949e-01 4.27389532e-01
5.06303191e-01 -4.78555232e-01 -4.36090589e-01 -1.02042890e+00
2.34128296e-01 -3.23378474e-01 -9.92313176e-02 -5.91109455e-01
-9.11642790e-01 2.48131335e-01 6.03774935e-02 -1.39148128e+00
1.66812599e-01 -6.97989941e-01 -1.67569816e-01 6.76519990e-01
-1.85409808e+00 -1.21004820e+00 -8.31069708e-01 4.09755170e-01
2.84294933e-01 2.07177073e-01 4.37769406e-02 7.50360012e-01
-9.70521986e-01 3.63642037e-01 -6.74174950e-02 1.44246593e-01
5.74002922e-01 -1.08199441e+00 2.95516253e-01 9.03477550e-01
-3.27626616e-01 1.06769964e-01 3.59939188e-01 -7.98902452e-01
-1.39893460e+00 -1.48148119e+00 4.68150944e-01 -5.78675449e-01
3.85572821e-01 -6.43928051e-01 -8.20768952e-01 5.41019261e-01
-7.80911386e-01 2.95942128e-01 5.98912910e-02 -1.44188598e-01
1.14535745e-02 -4.02865767e-01 -1.15482378e+00 3.46328050e-01
1.25294340e+00 -3.23805690e-01 -1.61589950e-01 2.69704163e-01
8.94991279e-01 -9.01308775e-01 -6.36716247e-01 8.32432985e-01
2.50070095e-01 -8.34809661e-01 1.01533008e+00 -1.94823686e-02
1.27970278e-01 -7.53455937e-01 -1.37941584e-01 -7.34323442e-01
-3.47343609e-02 5.56081757e-02 -2.98011571e-01 1.35603881e+00
2.39954591e-01 -7.21484363e-01 1.20143032e+00 4.78894204e-01
-5.32179773e-01 -5.63389599e-01 -1.13070774e+00 -9.01106894e-01
-4.06672299e-01 -5.91722727e-01 1.07966757e+00 8.41711402e-01
-6.39673650e-01 2.59068221e-01 5.76016456e-02 7.04972625e-01
7.58884072e-01 2.36011803e-01 1.23534322e+00 -1.58621204e+00
2.59609222e-01 -2.50362694e-01 -5.35018206e-01 -1.29233205e+00
-1.79069370e-01 -6.89695001e-01 2.91572809e-01 -1.31801152e+00
-2.19732538e-01 -1.11357343e+00 -2.61045154e-02 4.71119732e-01
-1.46252200e-01 3.76344591e-01 2.93108802e-02 3.69573683e-01
-5.14129460e-01 8.13781559e-01 1.19470274e+00 -8.69967639e-02
-2.17432097e-01 1.91803321e-01 -2.53242195e-01 9.43065345e-01
7.63078928e-01 -6.54861212e-01 -3.87514353e-01 -6.33838117e-01
1.93123013e-01 -2.02975601e-01 7.22691119e-01 -1.06858134e+00
2.60316938e-01 -2.11128309e-01 5.58439612e-01 -1.36563051e+00
5.94421983e-01 -1.03876495e+00 -1.54666036e-01 2.09203377e-01
1.89559370e-01 -2.27870986e-01 2.46173680e-01 6.26646698e-01
-1.21079445e-01 -1.50842786e-01 6.90629125e-01 3.21151242e-02
-8.41922164e-01 6.79943085e-01 2.13692663e-03 -1.63841441e-01
1.04452789e+00 -7.14235842e-01 -3.16159517e-01 3.25120538e-01
-2.27176398e-01 4.91267383e-01 7.60012448e-01 4.37204182e-01
2.46314481e-01 -1.17603028e+00 -2.71515161e-01 3.56581807e-01
4.24595267e-01 1.07404661e+00 1.28645808e-01 9.43878472e-01
-5.04282355e-01 2.00574264e-01 6.84393644e-02 -1.12576234e+00
-8.37036908e-01 3.87575805e-01 2.01549813e-01 3.02625090e-01
-8.27513278e-01 7.00887203e-01 8.20785686e-02 -5.47496557e-01
3.30548942e-01 -7.67457426e-01 -2.25702688e-01 1.33127585e-01
2.06026986e-01 5.93388915e-01 2.45410532e-01 -5.27769685e-01
-5.79033792e-01 7.04016447e-01 1.34365037e-01 2.85118937e-01
1.14015019e+00 -3.00970644e-01 -3.69634256e-02 4.33015019e-01
7.90614545e-01 -6.00728355e-02 -1.64970553e+00 -1.64122313e-01
-4.55521494e-02 -6.40950084e-01 4.38347936e-01 -4.62569118e-01
-1.17099237e+00 9.30821657e-01 7.31150687e-01 4.34345454e-02
7.33481586e-01 4.03182860e-03 6.96923077e-01 4.09987718e-01
5.17758191e-01 -1.06599975e+00 -3.50895733e-01 3.92639548e-01
4.11303163e-01 -1.54412639e+00 1.50573686e-01 -7.46542692e-01
-1.58750668e-01 6.31403446e-01 9.86513853e-01 -6.91884803e-03
4.78678524e-01 1.68945223e-01 -4.26145047e-02 -3.34765553e-01
-2.04745144e-01 -3.22732091e-01 2.80486703e-01 5.97881913e-01
-4.13703114e-01 -3.81561965e-02 1.39957264e-01 5.47765076e-01
-1.94145590e-01 -1.95144698e-01 1.25703350e-01 8.65605235e-01
-5.79845190e-01 -6.57911122e-01 -4.27935034e-01 5.07618129e-01
3.76746394e-02 1.79740146e-01 -1.52320728e-01 9.93123174e-01
6.19630694e-01 7.93053687e-01 4.00237441e-01 -1.88714236e-01
3.98082763e-01 -3.77619147e-01 2.67968774e-01 -6.48233771e-01
-6.20839819e-02 -1.55512825e-01 -3.82112488e-02 -5.80192804e-01
-4.78099138e-01 -6.22978389e-01 -1.34835386e+00 -1.20452233e-01
-8.18043053e-01 3.24532352e-02 1.05396664e+00 9.46321845e-01
6.53819323e-01 5.81311464e-01 6.16578221e-01 -1.30868816e+00
-3.69170070e-01 -8.44965458e-01 -5.26559532e-01 -4.61739935e-02
5.08217037e-01 -1.07824373e+00 -4.33881849e-01 -2.87062407e-01] | [7.920213222503662, -2.5704421997070312] |
1ef152cc-b3b7-4327-add6-62e163c70bec | reflection-and-rotation-symmetry-detection | 2203.16787 | null | https://arxiv.org/abs/2203.16787v1 | https://arxiv.org/pdf/2203.16787v1.pdf | Reflection and Rotation Symmetry Detection via Equivariant Learning | The inherent challenge of detecting symmetries stems from arbitrary orientations of symmetry patterns; a reflection symmetry mirrors itself against an axis with a specific orientation while a rotation symmetry matches its rotated copy with a specific orientation. Discovering such symmetry patterns from an image thus benefits from an equivariant feature representation, which varies consistently with reflection and rotation of the image. In this work, we introduce a group-equivariant convolutional network for symmetry detection, dubbed EquiSym, which leverages equivariant feature maps with respect to a dihedral group of reflection and rotation. The proposed network is built end-to-end with dihedrally-equivariant layers and trained to output a spatial map for reflection axes or rotation centers. We also present a new dataset, DENse and DIverse symmetry (DENDI), which mitigates limitations of existing benchmarks for reflection and rotation symmetry detection. Experiments show that our method achieves the state of the arts in symmetry detection on LDRS and DENDI datasets. | ['Minsu Cho', 'Suha Kwak', 'Byungjin Kim', 'Ahyun Seo'] | 2022-03-31 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Seo_Reflection_and_Rotation_Symmetry_Detection_via_Equivariant_Learning_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Seo_Reflection_and_Rotation_Symmetry_Detection_via_Equivariant_Learning_CVPR_2022_paper.pdf | cvpr-2022-1 | ['symmetry-detection'] | ['computer-vision'] | [ 3.35775167e-01 3.37771922e-02 -1.69904768e-01 -5.85281312e-01
-3.22741479e-01 -8.60198975e-01 9.82645273e-01 -6.64716244e-01
2.46009752e-01 -2.08773669e-02 7.34589875e-01 -5.74934259e-02
-2.72415668e-01 -6.34495318e-01 -7.76528895e-01 -4.32525307e-01
-1.01459481e-01 5.69024444e-01 -1.40589103e-01 -4.14073318e-01
5.65981269e-01 8.31751585e-01 -1.43762362e+00 6.56534910e-01
9.89529118e-02 6.94522917e-01 -4.13890630e-01 6.23831153e-01
5.08924484e-01 4.37978446e-01 -1.51322156e-01 -8.24473351e-02
7.38635957e-01 -4.68623102e-01 -1.07885218e+00 2.71718115e-01
9.59685802e-01 -2.60923028e-01 -4.67269421e-01 5.44016302e-01
3.14130843e-01 -2.12032259e-01 8.86936784e-01 -1.10716438e+00
-5.50447762e-01 3.09587359e-01 -7.55069911e-01 -4.37620841e-02
5.84480226e-01 -1.51750550e-01 1.47803378e+00 -9.29794490e-01
1.26885986e+00 1.11125326e+00 7.27802634e-01 2.74854720e-01
-9.42253828e-01 -6.98001683e-01 -1.18303351e-01 -2.95982510e-02
-1.37120450e+00 -6.21643484e-01 8.33405614e-01 -3.89884740e-01
1.12280285e+00 2.63847470e-01 5.88696480e-01 1.00066853e+00
1.90579265e-01 4.64714676e-01 8.79146636e-01 -1.33527622e-01
-1.18622750e-01 -8.45390379e-01 -3.82791311e-01 7.73993433e-01
1.62719429e-01 -2.34785736e-01 -7.04865754e-01 2.30003953e-01
1.09204757e+00 1.78844988e-01 -1.25347495e-01 -1.10887885e+00
-1.56909049e+00 5.57934642e-01 5.38449883e-01 -1.20073996e-01
1.91671867e-02 -2.25954488e-01 3.13907236e-01 5.95000625e-01
3.09934672e-02 8.38781357e-01 -3.90391558e-01 -1.07042827e-01
-8.34849417e-01 4.42117125e-01 5.43697774e-01 8.60229254e-01
8.07713926e-01 -1.64103165e-01 -3.28562036e-02 7.00707436e-01
1.22346416e-01 3.75230581e-01 2.95975178e-01 -8.02614808e-01
5.02896011e-01 9.72071290e-01 -3.38938594e-01 -7.96301305e-01
-9.20343161e-01 -5.57512820e-01 -1.03326249e+00 2.99828589e-01
2.12783545e-01 4.77898657e-01 -8.28094006e-01 1.62932599e+00
1.40991688e-01 -1.65970102e-01 -1.06935427e-02 8.84948790e-01
7.56505191e-01 2.70285517e-01 -1.17754793e+00 4.91154552e-01
1.37006307e+00 -9.40367460e-01 -4.02991194e-03 8.91141519e-02
6.00067317e-01 -9.94367480e-01 8.96189570e-01 3.38119358e-01
-9.42374289e-01 -9.79875848e-02 -1.32644868e+00 -3.35157961e-01
-8.60522166e-02 1.88352227e-01 5.29044926e-01 3.48196834e-01
-1.15190041e+00 4.81051445e-01 -8.65159631e-01 -4.99115735e-01
3.24633718e-01 4.39334899e-01 -7.84136057e-01 1.80528909e-01
-5.25504351e-01 3.50500256e-01 -1.16772935e-01 -2.44312614e-01
-4.91581500e-01 -9.79601443e-01 -9.66966271e-01 6.81750104e-02
-8.64025801e-02 -7.63841748e-01 1.10411286e+00 -7.36035049e-01
-1.50203371e+00 1.38493145e+00 -1.84489086e-01 -1.40881777e-01
7.77181268e-01 5.95894642e-02 -3.54710996e-01 4.27091978e-02
3.63691598e-01 5.66867173e-01 1.06475699e+00 -8.01142752e-01
-3.31290007e-01 -5.99037468e-01 4.29124944e-02 1.93949714e-01
1.10727459e-01 3.05011600e-01 -3.59703213e-01 -4.58233774e-01
1.06358719e+00 -1.07469666e+00 3.38414133e-01 1.41730532e-01
-8.37536931e-01 5.08734211e-03 1.31376266e+00 -2.23609060e-01
7.08011925e-01 -1.89276528e+00 1.20433934e-01 5.71050167e-01
5.26140928e-01 -2.02018663e-01 -1.69276193e-01 6.41052425e-01
-8.77103209e-01 2.80656684e-02 4.32496220e-02 -9.24829245e-02
3.85372378e-02 -1.46850288e-01 -5.01063406e-01 9.68267858e-01
4.44082916e-01 7.68283486e-01 -5.18136561e-01 1.81639254e-01
1.31051391e-01 5.59794068e-01 -1.20369184e+00 -1.24701306e-01
2.90793151e-01 5.77504098e-01 -1.06479190e-01 6.67381525e-01
9.83038545e-01 -5.42228520e-01 3.25494617e-01 -4.81349915e-01
-4.59166646e-01 9.03648794e-01 -1.28350925e+00 1.64102030e+00
-3.35387856e-01 7.07487524e-01 -3.30743134e-01 -9.80837345e-01
1.14364135e+00 5.22027612e-02 7.12730348e-01 -9.23455000e-01
-1.71418041e-01 2.58018732e-01 2.75080353e-01 1.42398655e-01
4.44998115e-01 4.55250263e-01 -9.83296633e-02 1.00718212e+00
9.75535214e-02 -2.75812089e-01 1.40600666e-01 -2.24504564e-02
1.20379090e+00 2.34694138e-01 4.26078558e-01 -3.83180827e-01
5.02697229e-01 -7.27950275e-01 5.99086225e-01 5.01561582e-01
2.54291326e-01 1.20020115e+00 8.90880167e-01 -1.18390870e+00
-1.72659266e+00 -1.38748670e+00 -3.73394668e-01 7.23852754e-01
-4.30780388e-02 -8.31530690e-01 -4.98012900e-01 -4.75270212e-01
-9.47667286e-02 -3.86087239e-01 -6.17793024e-01 -6.99698627e-02
-9.31340039e-01 -4.18936163e-01 4.78661776e-01 2.70791471e-01
8.47694814e-01 -8.09558809e-01 -4.60945159e-01 -4.30418342e-01
-1.36183202e-02 -1.03388476e+00 -5.86083710e-01 -2.98369788e-02
-6.79251373e-01 -1.37272513e+00 -5.54116368e-01 -8.01764965e-01
9.72105503e-01 5.78776777e-01 1.05347252e+00 -1.67708099e-01
-5.15102983e-01 9.65180770e-02 -4.68934104e-02 6.92448020e-02
-1.74870789e-01 2.77138948e-01 1.76388845e-01 -3.23542603e-03
3.72280627e-02 -1.00792432e+00 -8.38358521e-01 7.66797066e-01
-7.39920557e-01 3.65621120e-01 3.82423550e-01 6.81422293e-01
6.08526886e-01 -1.75229147e-01 -1.39283419e-01 -5.10789633e-01
1.58958554e-01 2.89252754e-02 -5.98195076e-01 -2.18998827e-02
-2.19933957e-01 4.75980788e-01 6.14103198e-01 3.27187538e-01
-6.05924249e-01 1.97548732e-01 2.38629416e-01 -3.09933066e-01
1.05161116e-01 -9.43319201e-02 -1.75683528e-01 -2.40794420e-01
6.31836832e-01 2.48837948e-01 5.85664771e-02 -4.27109718e-01
1.95357963e-01 3.90563190e-01 5.83183765e-01 -5.69483817e-01
1.18194938e+00 1.04391289e+00 4.30892855e-01 -9.04620945e-01
-7.41248071e-01 -4.77549195e-01 -8.98844540e-01 2.00921565e-01
4.65414941e-01 -9.84525383e-01 -7.84589887e-01 8.72776449e-01
-9.74625885e-01 8.24264959e-02 -1.75161287e-01 3.16167891e-01
-1.02499807e+00 4.61721003e-01 -3.06782126e-01 -1.67005267e-02
-3.87154996e-01 -1.18776727e+00 1.53374398e+00 -3.37299742e-02
-7.15411365e-01 -4.93093759e-01 3.29527766e-01 3.23947370e-01
3.32431078e-01 9.24102962e-02 9.68425155e-01 -5.13312101e-01
-9.31495726e-01 -5.10868467e-02 -4.64460641e-01 1.04162190e-03
2.55553305e-01 3.47420931e-01 -7.68246531e-01 -4.29259866e-01
-2.91095734e-01 -5.12068152e-01 9.31553960e-01 1.64449006e-01
1.28342366e+00 -2.65812188e-01 8.78506973e-02 1.54286587e+00
9.63458240e-01 -3.34756434e-01 7.37215519e-01 7.89490044e-01
7.47397661e-01 2.99565136e-01 -4.84321825e-02 6.51453078e-01
3.31055909e-01 9.52569127e-01 4.84284222e-01 -3.98759544e-01
-2.27044702e-01 -4.08527613e-01 1.93383843e-01 4.59527493e-01
-5.21052897e-01 4.40977663e-02 -7.82062948e-01 1.74338356e-01
-1.52836597e+00 -1.13509560e+00 1.85510457e-01 2.28123236e+00
3.75782788e-01 -6.71736477e-03 1.48359925e-01 2.81843573e-01
5.98271966e-01 6.34732425e-01 -5.39064765e-01 -5.73450804e-01
-4.94520128e-01 3.63397926e-01 5.02715349e-01 1.52686328e-01
-1.31832922e+00 7.52043128e-01 6.59505892e+00 2.81282753e-01
-1.51880467e+00 -6.21704638e-01 7.64781311e-02 5.62155917e-02
-5.27517080e-01 1.42488807e-01 -7.07538247e-01 -4.49697748e-02
9.56545547e-02 1.68134123e-01 4.69767541e-01 7.47555554e-01
1.77646935e-01 6.68614328e-01 -1.34542692e+00 1.04972053e+00
1.96160033e-01 -1.63181782e+00 4.10872877e-01 3.26324433e-01
1.10876083e+00 3.52992624e-01 4.66591507e-01 -2.10747704e-01
1.77475840e-01 -1.04694891e+00 5.37883937e-01 1.31661683e-01
1.02076495e+00 -6.89434707e-01 1.18272848e-01 -3.58242631e-01
-1.33418536e+00 1.68531671e-01 -3.27441990e-01 -1.58745542e-01
-2.33292341e-01 1.14206858e-01 -1.02632046e+00 5.15643120e-01
7.49446273e-01 1.17247629e+00 -4.03010011e-01 7.09034383e-01
-3.61613244e-01 9.70666409e-02 -4.69861805e-01 6.48121655e-01
2.67865568e-01 -5.04852772e-01 6.39041007e-01 1.03303444e+00
4.79706436e-01 -5.00069559e-01 -2.94268966e-01 9.32709038e-01
-3.73249561e-01 -4.73590530e-02 -9.26699817e-01 1.13628320e-01
1.16986185e-01 1.19642854e+00 -8.30202639e-01 1.90380543e-01
-4.98184532e-01 1.15603185e+00 3.10760289e-01 2.51473069e-01
-5.58093131e-01 -3.89930576e-01 1.29703903e+00 2.75427014e-01
6.31480932e-01 -2.90759325e-01 -1.58563182e-01 -1.56772101e+00
4.29706305e-01 -9.52141345e-01 3.02570254e-01 -8.75213444e-01
-8.00105929e-01 2.34076023e-01 -3.45488310e-01 -1.54269707e+00
-3.88653755e-01 -8.65659177e-01 -1.06918156e+00 6.47665858e-01
-1.31525981e+00 -1.41219962e+00 -3.54914099e-01 5.88698983e-01
2.45115668e-01 -5.02893448e-01 7.88064063e-01 1.85223874e-02
-2.71631688e-01 7.77471721e-01 2.84290701e-01 5.11222541e-01
8.87368679e-01 -1.14425576e+00 1.05709755e+00 7.45162487e-01
3.66727740e-01 8.55682313e-01 3.88093382e-01 -1.22638591e-01
-1.57112706e+00 -1.03206348e+00 1.16311061e+00 -4.45012271e-01
6.12050235e-01 -7.27940679e-01 -3.34961563e-01 1.12775135e+00
-1.69731885e-01 1.04097888e-01 4.89839107e-01 2.41178736e-01
-1.27431643e+00 -1.56462833e-01 -9.39014256e-01 1.02316296e+00
1.83430147e+00 -8.09145033e-01 -3.88127685e-01 5.73073685e-01
2.77650058e-01 -5.98431051e-01 -6.18109941e-01 7.12640166e-01
9.56860721e-01 -1.41614652e+00 1.23118103e+00 -8.56501460e-01
8.09654474e-01 -4.39258426e-01 9.09190997e-02 -1.09348285e+00
-5.21036148e-01 -9.03907835e-01 4.62978870e-01 4.84803826e-01
3.65100861e-01 -6.63196981e-01 1.13561714e+00 -8.05519819e-02
-3.42404723e-01 -5.82312346e-01 -9.84600902e-01 -8.05605888e-01
-7.89170787e-02 -7.61741474e-02 9.54972744e-01 1.17304182e+00
-4.11072597e-02 7.44656846e-02 -2.25759178e-01 1.81398690e-01
3.79663020e-01 7.44609714e-01 1.28152752e+00 -1.09915423e+00
-7.78375119e-02 -6.91779792e-01 -1.24093044e+00 -1.22131264e+00
1.65087938e-01 -1.26949692e+00 -6.58360779e-01 -1.07443810e+00
3.40002596e-01 5.70663735e-02 -1.16529185e-02 4.69786733e-01
6.77098215e-01 6.31788373e-01 -1.50057197e-01 2.82640070e-01
-7.82821834e-01 5.26221573e-01 1.46365786e+00 -3.35587934e-02
-6.42204061e-02 2.07693856e-02 -6.49142385e-01 8.99893522e-01
7.62902081e-01 -1.54813007e-01 -3.81939828e-01 -2.83559233e-01
8.34084451e-01 -4.28242952e-01 2.90795207e-01 -9.10815358e-01
9.00601447e-02 2.98543274e-01 5.89275181e-01 -7.75961876e-01
2.39989623e-01 -3.48987579e-01 6.04613349e-02 1.97502673e-01
-2.03043744e-01 4.75215077e-01 -2.89757907e-01 7.39446506e-02
-2.25299358e-01 4.64962900e-01 7.58546710e-01 9.91131514e-02
-3.96856874e-01 5.30558109e-01 -1.42770395e-01 2.97168076e-01
4.81837392e-01 -6.26587451e-01 -5.68173230e-01 -5.55207253e-01
-4.44699228e-01 3.18686441e-02 1.13354087e+00 7.96297073e-01
6.66595697e-01 -1.46623921e+00 -7.66411364e-01 9.42571580e-01
5.29587984e-01 1.69895306e-01 1.05725378e-01 7.72223413e-01
-1.01884329e+00 5.76181889e-01 -7.28386343e-01 -9.47238564e-01
-1.33097410e+00 -1.96633134e-02 6.77302539e-01 -2.38381773e-01
-9.05442595e-01 5.70306420e-01 5.91521680e-01 -1.28894687e+00
-1.39280021e-01 -5.93100846e-01 -3.43669951e-02 -4.16354001e-01
3.44560444e-01 3.14746588e-01 4.14776176e-01 -9.59339201e-01
-5.12242794e-01 1.25227702e+00 -1.88778713e-01 1.37746138e-02
1.39600050e+00 2.04905838e-01 -2.74893671e-01 -1.71824872e-01
1.56514382e+00 4.14913684e-01 -1.12532699e+00 -3.01571965e-01
-2.96839088e-01 -3.93107533e-01 -5.86416960e-01 -3.98838282e-01
-1.00880623e+00 4.93859529e-01 1.15377553e-01 -3.96640271e-01
7.06615627e-01 4.48554829e-02 3.77489150e-01 9.63982522e-01
4.79023270e-02 -8.57168317e-01 1.72747165e-01 9.10338283e-01
1.28383875e+00 -1.10599017e+00 1.60730675e-01 -3.91071528e-01
-3.19397420e-01 1.59795213e+00 4.18401688e-01 -5.68571270e-01
6.58325195e-01 9.81444716e-02 2.30943978e-01 -5.51970065e-01
-3.92801940e-01 2.18061507e-01 5.93639255e-01 4.54998374e-01
6.43589377e-01 5.41912764e-02 -4.72396752e-03 -8.12433288e-02
-9.54727173e-01 -5.06921291e-01 4.47925836e-01 6.07459366e-01
-1.54130384e-01 -1.26136601e+00 -3.12586039e-01 2.03872189e-01
-2.63068110e-01 1.12747028e-01 -9.46270347e-01 8.71138096e-01
-1.62175775e-01 2.27831021e-01 7.32674658e-01 -3.92321885e-01
3.00984561e-01 -1.35358512e-01 6.24575257e-01 -5.37123919e-01
-1.45959899e-01 -2.77527690e-01 -1.95354242e-02 -1.08226967e+00
-3.13803017e-01 -8.07929933e-01 -1.25499737e+00 -2.84568548e-01
4.93037194e-01 -2.42264360e-01 7.75508821e-01 7.14556217e-01
6.76096976e-01 1.37345940e-01 1.06395364e+00 -8.17178726e-01
-4.98775214e-01 -5.45245290e-01 -5.90424001e-01 7.68466055e-01
4.12461221e-01 -4.66009408e-01 -5.26451111e-01 -2.72375882e-01] | [8.557003021240234, -2.1471669673919678] |
c359295d-8490-42eb-9bcf-1a903ed63779 | domain-adaptive-person-search | 2207.11898 | null | https://arxiv.org/abs/2207.11898v1 | https://arxiv.org/pdf/2207.11898v1.pdf | Domain Adaptive Person Search | Person search is a challenging task which aims to achieve joint pedestrian detection and person re-identification (ReID). Previous works have made significant advances under fully and weakly supervised settings. However, existing methods ignore the generalization ability of the person search models. In this paper, we take a further step and present Domain Adaptive Person Search (DAPS), which aims to generalize the model from a labeled source domain to the unlabeled target domain. Two major challenges arises under this new setting: one is how to simultaneously solve the domain misalignment issue for both detection and Re-ID tasks, and the other is how to train the ReID subtask without reliable detection results on the target domain. To address these challenges, we propose a strong baseline framework with two dedicated designs. 1) We design a domain alignment module including image-level and task-sensitive instance-level alignments, to minimize the domain discrepancy. 2) We take full advantage of the unlabeled data with a dynamic clustering strategy, and employ pseudo bounding boxes to support ReID and detection training on the target domain. With the above designs, our framework achieves 34.7% in mAP and 80.6% in top-1 on PRW dataset, surpassing the direct transferring baseline by a large margin. Surprisingly, the performance of our unsupervised DAPS model even surpasses some of the fully and weakly supervised methods. The code is available at https://github.com/caposerenity/DAPS. | ['Shouhong Ding', 'Qiong Jia', 'Fufu Yu', 'Guanshuo Wang', 'Yichao Yan', 'Junjie Li'] | 2022-07-25 | null | null | null | null | ['person-search'] | ['computer-vision'] | [-1.22253811e-02 -2.20054090e-01 -2.32892796e-01 -4.98918325e-01
-9.24724221e-01 -6.54691041e-01 6.99866772e-01 -3.15755785e-01
-7.48227119e-01 7.75213242e-01 2.67099470e-01 5.63220270e-02
2.18655378e-01 -4.66772974e-01 -5.71522832e-01 -6.73092961e-01
4.05558497e-01 7.67252803e-01 5.17956853e-01 -4.92239464e-03
-5.58246262e-02 2.05960363e-01 -1.52778101e+00 9.16243251e-03
1.04821384e+00 6.33011937e-01 2.44305223e-01 5.08015931e-01
1.54836774e-01 3.82903427e-01 -4.04986888e-01 -6.52299941e-01
4.78418529e-01 -1.08774684e-01 -7.64018953e-01 1.41332105e-01
7.01725900e-01 -5.95614493e-01 -5.10546565e-01 1.16854775e+00
9.04516637e-01 9.78243649e-02 6.07980788e-01 -1.41312015e+00
-7.57809401e-01 2.36428738e-01 -8.70253265e-01 2.13793114e-01
3.56153458e-01 3.85411501e-01 5.55254340e-01 -9.92962241e-01
3.88439804e-01 1.21609044e+00 5.97544789e-01 9.70897079e-01
-1.16191137e+00 -1.02383232e+00 4.06230450e-01 3.38661253e-01
-1.64503884e+00 -6.61429167e-01 4.25241858e-01 -4.69188452e-01
5.60501277e-01 2.85032354e-02 2.12051794e-01 1.41921842e+00
-6.79070055e-01 9.96107578e-01 1.08006537e+00 -3.55585456e-01
3.17111537e-02 3.15477729e-01 5.10780156e-01 3.19333315e-01
4.33025986e-01 2.08813816e-01 -4.53607082e-01 -6.28633201e-02
5.89064121e-01 -8.01836178e-02 -1.35672510e-01 -3.96116048e-01
-1.12271082e+00 4.79690760e-01 3.41163516e-01 -6.09381385e-02
-1.40362307e-01 -3.28007072e-01 3.75254184e-01 -1.58787481e-02
4.00823116e-01 5.00679202e-02 -3.28200310e-01 1.46384239e-01
-8.46238792e-01 3.70585799e-01 5.49063742e-01 1.42033613e+00
5.82500696e-01 -3.47915828e-01 -4.62614119e-01 1.04277289e+00
1.18946180e-01 8.12796295e-01 3.14792901e-01 -6.86187327e-01
6.78587854e-01 5.53026736e-01 4.52330619e-01 -5.37095606e-01
-2.52377152e-01 -5.46212554e-01 -7.34956861e-01 -5.72341047e-02
7.72998750e-01 -2.24316388e-01 -1.09015942e+00 1.94359529e+00
6.02329433e-01 3.22652012e-01 -6.51069358e-02 1.16426706e+00
9.55905795e-01 2.08821595e-01 5.24394929e-01 1.75798312e-01
1.66012371e+00 -1.25215304e+00 -4.34834212e-01 -6.90781236e-01
4.10486877e-01 -5.56442440e-01 9.58528876e-01 -6.65009171e-02
-9.74637747e-01 -8.08793366e-01 -9.82170045e-01 -9.38750654e-02
-3.39704156e-01 5.75561643e-01 1.05470650e-01 7.72411346e-01
-1.06482601e+00 -1.10440657e-01 -5.99851072e-01 -8.30394626e-01
4.77880478e-01 4.10153866e-01 -5.13121367e-01 -3.92949253e-01
-1.10540664e+00 8.20503473e-01 3.19664001e-01 -1.15866244e-01
-7.36162603e-01 -6.15682125e-01 -7.77055800e-01 -1.16612390e-01
4.52796876e-01 -9.87351775e-01 1.27764726e+00 -6.60292685e-01
-1.07733202e+00 1.35268867e+00 -5.21394849e-01 -4.67704475e-01
9.22796667e-01 -3.96188110e-01 -4.29115117e-01 -1.29391924e-02
6.51643693e-01 8.45773458e-01 5.60670972e-01 -1.28500140e+00
-1.07012689e+00 -6.22565866e-01 -1.46708056e-01 3.62611324e-01
-3.78165275e-01 2.83854216e-01 -1.11209357e+00 -5.61342478e-01
-1.69715315e-01 -9.55821395e-01 -1.53414130e-01 -1.31603694e-02
-4.89258319e-01 -4.29393888e-01 5.79947889e-01 -7.82784760e-01
8.33741784e-01 -2.04088998e+00 -6.22261092e-02 -1.17645442e-01
1.73891976e-01 5.58804512e-01 -2.01699838e-01 7.90976062e-02
7.33628683e-03 -2.61928529e-01 -1.50165945e-01 -7.52335727e-01
8.45809504e-02 -2.68225539e-02 -1.16157800e-01 5.31372666e-01
1.12022348e-01 9.61107910e-01 -8.61016989e-01 -6.16234899e-01
9.18222293e-02 1.66395068e-01 -3.38375717e-01 2.73794740e-01
2.14339614e-01 6.69359744e-01 -3.94197911e-01 7.03711927e-01
1.10549223e+00 -3.09907436e-01 2.70688329e-02 -4.77660336e-02
-1.45502821e-01 7.87945017e-02 -1.32869995e+00 1.63567233e+00
1.78061888e-01 2.59704024e-01 2.03209534e-01 -1.02804828e+00
7.49425948e-01 2.75192130e-02 2.46180534e-01 -9.00496900e-01
-1.12031564e-01 1.28699288e-01 -3.51401508e-01 -3.11021686e-01
4.09355134e-01 2.12204427e-01 -1.30710542e-01 2.19690457e-01
-9.15449783e-02 7.16260552e-01 5.64436167e-02 2.63223320e-01
8.55885863e-01 2.15851441e-01 1.75866351e-01 -3.07799876e-01
7.95321703e-01 2.34165147e-01 8.96050036e-01 9.62773442e-01
-7.71859467e-01 8.36899340e-01 6.93346933e-02 -2.86991417e-01
-1.06530738e+00 -1.26022887e+00 -1.32397592e-01 1.26856840e+00
6.47290647e-01 -2.36673281e-02 -9.36593175e-01 -8.61571014e-01
1.73321277e-01 4.00470674e-01 -4.46696639e-01 -5.74679971e-02
-6.92505181e-01 -8.69085729e-01 7.53502369e-01 7.28124440e-01
1.02502275e+00 -5.07849395e-01 3.76258865e-02 -1.16009220e-01
-5.88095844e-01 -1.60624993e+00 -8.45727086e-01 -3.41627300e-01
-2.85674602e-01 -1.10663009e+00 -1.29150498e+00 -1.08963001e+00
7.52975583e-01 6.95108414e-01 8.95314336e-01 -6.75368831e-02
-2.00365260e-01 4.84235734e-01 -2.95472920e-01 -3.30278367e-01
-8.58425722e-02 1.90974966e-01 4.96228009e-01 -5.59252910e-02
9.62430000e-01 -2.45495990e-01 -9.89843786e-01 9.35490370e-01
-2.09702492e-01 6.94249645e-02 5.73337734e-01 7.99793243e-01
5.52266896e-01 -1.65093422e-01 6.94384933e-01 -4.60081100e-01
2.55464911e-01 -3.87696415e-01 -5.11092901e-01 4.65002388e-01
-5.49683154e-01 -2.67022431e-01 2.07493633e-01 -5.55068254e-01
-1.20364344e+00 2.83902884e-01 -7.66025856e-02 -2.32537448e-01
-5.15692353e-01 -2.96780020e-01 -6.20215297e-01 -3.27424961e-03
7.36452043e-01 4.30735230e-01 -1.10973381e-01 -6.90934658e-01
1.70802668e-01 9.63212192e-01 1.09369695e+00 -8.32487524e-01
1.01986730e+00 6.69818878e-01 -5.32312095e-01 -4.95373517e-01
-7.94701695e-01 -1.03845978e+00 -8.80570352e-01 -8.92826915e-02
8.77793074e-01 -1.51805580e+00 -5.98033786e-01 6.71400249e-01
-1.02294433e+00 -3.12728792e-01 -8.43357015e-03 3.12846094e-01
-2.65754372e-01 8.02253187e-01 -3.95504832e-01 -8.27900112e-01
-4.47995543e-01 -9.67924774e-01 1.17457068e+00 4.56310898e-01
-4.29514647e-02 -3.90410960e-01 -1.78782523e-01 7.51737952e-01
1.87854216e-01 -4.34750654e-02 2.02075988e-01 -7.65763700e-01
-4.37451929e-01 -3.45108718e-01 -8.21312189e-01 1.90810129e-01
-2.07352545e-02 -6.82108402e-01 -1.07683086e+00 -4.91566539e-01
-3.29938889e-01 -2.16080815e-01 9.67537463e-01 2.98757970e-01
8.03855717e-01 4.73538078e-02 -8.06549191e-01 6.08665705e-01
9.79921579e-01 -1.08338758e-01 4.45009887e-01 6.29165471e-01
6.54494524e-01 7.86914945e-01 7.11430192e-01 2.34346226e-01
1.00637949e+00 1.05662811e+00 -1.90261938e-02 -1.87514946e-01
-5.73045671e-01 -5.89656055e-01 2.51819283e-01 1.57105047e-02
-1.12009525e-01 -8.61354023e-02 -1.09094656e+00 7.95510411e-01
-2.18275881e+00 -9.65206087e-01 -1.78352967e-01 2.33759737e+00
7.38110960e-01 3.51546481e-02 7.45566666e-01 -2.59390801e-01
1.28788948e+00 -2.71241248e-01 -7.55133510e-01 4.57066387e-01
-1.67203501e-01 -3.29803556e-01 7.21292198e-01 3.46593380e-01
-1.48472714e+00 1.13486969e+00 5.71628380e+00 8.27873051e-01
-5.02743185e-01 2.78724223e-01 4.37806159e-01 -8.17211494e-02
3.96331340e-01 -2.49084681e-01 -1.41943920e+00 7.71141410e-01
5.49890816e-01 -1.30478233e-01 1.02534495e-01 1.10999084e+00
8.82165208e-02 -4.92217243e-02 -1.17925918e+00 1.21002936e+00
6.33714125e-02 -7.63129234e-01 -1.14296973e-01 9.35635045e-02
5.85055947e-01 -2.12395247e-02 9.43922102e-02 5.44813097e-01
4.82516617e-01 -7.84177125e-01 6.41722023e-01 1.56403095e-01
9.46435094e-01 -5.80489516e-01 6.90785050e-01 6.40415370e-01
-1.40796304e+00 -1.13361232e-01 -4.09503490e-01 6.88306838e-02
3.25500190e-01 2.13696450e-01 -6.14668369e-01 5.27780831e-01
1.00030839e+00 5.70087254e-01 -6.78309619e-01 1.32765377e+00
-2.99172252e-01 3.60224485e-01 -4.14583027e-01 3.65475714e-01
-4.25668173e-02 6.67714030e-02 5.76490462e-01 1.49529946e+00
8.68611708e-02 1.81838181e-02 4.35099959e-01 8.54194701e-01
-9.01270658e-03 -2.10589215e-01 -2.42024079e-01 5.50619125e-01
7.44848132e-01 1.06886160e+00 -3.96361887e-01 -4.91829991e-01
-4.93867189e-01 1.33842719e+00 4.16249573e-01 5.75122833e-01
-1.06173110e+00 -9.64638963e-02 9.79869843e-01 2.59033203e-01
1.86174288e-01 -1.30759120e-01 -4.05488133e-01 -1.36927068e+00
3.32338125e-01 -7.19663858e-01 7.56981134e-01 -4.04963285e-01
-1.62858975e+00 2.85110831e-01 1.16058119e-01 -1.25140536e+00
-6.89101666e-02 -4.38740790e-01 -4.81120884e-01 1.17418456e+00
-1.88518882e+00 -1.56403959e+00 -6.67527139e-01 8.54265571e-01
6.19965136e-01 -2.23358870e-01 4.92300153e-01 7.83181012e-01
-8.79319847e-01 1.24853408e+00 -4.78437021e-02 5.78847468e-01
1.30280018e+00 -1.10007119e+00 7.73770213e-01 1.24015856e+00
-4.11521822e-01 7.25069404e-01 5.46162426e-01 -7.03482389e-01
-9.75442767e-01 -1.25873613e+00 9.41797435e-01 -8.95285785e-01
3.11050206e-01 -6.20168984e-01 -7.61361241e-01 7.41267562e-01
-1.17173433e-01 3.14161889e-02 5.69460452e-01 1.54186681e-01
-5.52963257e-01 -2.37352643e-02 -1.28870082e+00 6.03450298e-01
1.60579908e+00 -3.39369863e-01 -5.90848148e-01 3.55836481e-01
5.39419234e-01 -4.66985703e-01 -3.38903606e-01 4.53515798e-01
5.32574952e-01 -7.89918721e-01 1.51963758e+00 -4.94986415e-01
-5.89763597e-02 -5.73924899e-01 -8.47921818e-02 -9.22194421e-01
-5.70587695e-01 -3.13588858e-01 3.40450183e-02 1.57584453e+00
1.39986262e-01 -8.03797901e-01 9.70585585e-01 1.04654610e+00
7.73065165e-02 -6.70508370e-02 -1.02358687e+00 -1.09833050e+00
1.80037946e-01 -1.31654322e-01 6.57962084e-01 7.73169756e-01
-2.43607849e-01 2.77882993e-01 -5.74288905e-01 7.27644324e-01
1.09094238e+00 -9.12017655e-03 1.21357703e+00 -1.11166406e+00
-1.90586001e-01 -3.25425804e-01 -2.84757048e-01 -1.60188138e+00
1.00324020e-01 -6.57660544e-01 8.21607038e-02 -1.47558045e+00
7.10057020e-01 -5.12750626e-01 -1.87592030e-01 4.56984818e-01
-5.39490044e-01 2.29623839e-01 2.34171182e-01 6.57018006e-01
-8.76275003e-01 3.47657502e-01 9.01831388e-01 -2.46096060e-01
-5.96597642e-02 1.51479319e-01 -1.03299713e+00 5.69546759e-01
7.78013229e-01 -2.48741329e-01 -1.62297383e-01 -6.36038959e-01
-4.84927595e-01 -4.52150524e-01 7.94934511e-01 -1.05038643e+00
7.56130576e-01 -1.38119673e-02 6.19050384e-01 -5.95825493e-01
3.03276926e-01 -5.25634825e-01 -1.67371765e-01 2.66325265e-01
-4.17349897e-02 -2.87715197e-01 1.17648512e-01 7.46680319e-01
-2.59278547e-02 1.15191668e-01 1.01682580e+00 -1.51605923e-02
-1.20463192e+00 5.37884831e-01 9.28732455e-02 6.92118332e-02
1.17354405e+00 -3.82106602e-01 -6.66239917e-01 -9.42194089e-02
-5.49081624e-01 8.49749207e-01 6.84951901e-01 5.92651904e-01
2.97785103e-01 -1.29452622e+00 -1.01047373e+00 1.32477164e-01
4.77221996e-01 1.02140233e-01 4.65833217e-01 7.69948661e-01
6.18947763e-03 3.97257179e-01 -3.47634852e-02 -6.70785487e-01
-1.40146303e+00 6.22702181e-01 4.06600714e-01 -7.97483474e-02
-6.44363225e-01 9.19210732e-01 5.97913682e-01 -6.20257974e-01
4.76663977e-01 4.73587960e-01 -2.10331708e-01 -1.77418157e-01
9.06184673e-01 5.04553854e-01 -2.76625693e-01 -9.21181262e-01
-7.11763918e-01 7.14196861e-01 -4.03964192e-01 1.61221370e-01
9.40611362e-01 -4.87356186e-01 3.97413582e-01 -4.21899736e-01
8.51211786e-01 -2.75078446e-01 -1.51304948e+00 -5.74835300e-01
9.11362618e-02 -4.91689593e-01 -3.74387652e-01 -9.69978869e-01
-6.36434138e-01 6.59960330e-01 7.73878634e-01 -2.90128112e-01
9.56071794e-01 3.18044931e-01 9.54257607e-01 9.89230052e-02
4.75572169e-01 -1.31700146e+00 -1.29106879e-01 3.78101587e-01
5.78517914e-01 -1.69546103e+00 -1.72464460e-01 -6.35998189e-01
-7.58417428e-01 5.14356971e-01 9.96592999e-01 1.14141591e-01
1.59579873e-01 1.22933075e-01 7.21760392e-02 1.35325104e-01
-1.39305711e-01 -7.38918424e-01 3.26023310e-01 1.07463884e+00
-5.02468348e-02 9.05551687e-02 -1.56328186e-01 9.98747766e-01
2.42646877e-02 8.06591660e-02 -7.55112544e-02 6.57207668e-01
-4.52087492e-01 -1.20049846e+00 -6.53344154e-01 3.37741785e-02
-2.15595320e-01 1.09809659e-01 -4.92871642e-01 7.11962581e-01
2.62448400e-01 1.18089986e+00 -1.83150738e-01 -3.91708076e-01
6.76275134e-01 8.33661556e-02 1.97431430e-01 -5.36126971e-01
-2.10708678e-01 -1.71616092e-01 1.25325397e-01 -4.06103849e-01
-3.02114725e-01 -8.10321391e-01 -9.27996337e-01 -4.51097012e-01
-2.02901259e-01 -9.45004523e-02 2.36267239e-01 7.67837882e-01
5.87633789e-01 -2.76469775e-02 3.26190323e-01 -7.69202769e-01
-6.98809326e-01 -8.59062672e-01 -3.79986972e-01 6.10128343e-01
2.88973302e-01 -9.13943112e-01 -1.96507007e-01 -2.36369626e-04] | [14.83962631225586, 0.8527619242668152] |
37e25c15-edb2-471c-a5cf-36d2ed6beb81 | audio-super-resolution-using-neural-networks | 1708.00853 | null | http://arxiv.org/abs/1708.00853v1 | http://arxiv.org/pdf/1708.00853v1.pdf | Audio Super Resolution using Neural Networks | We introduce a new audio processing technique that increases the sampling
rate of signals such as speech or music using deep convolutional neural
networks. Our model is trained on pairs of low and high-quality audio examples;
at test-time, it predicts missing samples within a low-resolution signal in an
interpolation process similar to image super-resolution. Our method is simple
and does not involve specialized audio processing techniques; in our
experiments, it outperforms baselines on standard speech and music benchmarks
at upscaling ratios of 2x, 4x, and 6x. The method has practical applications in
telephony, compression, and text-to-speech generation; it demonstrates the
effectiveness of feed-forward convolutional architectures on an audio
generation task. | ['Volodymyr Kuleshov', 'S. Zayd Enam', 'Stefano Ermon'] | 2017-08-02 | null | null | null | null | ['audio-generation', 'audio-super-resolution', 'audio-super-resolution'] | ['audio', 'audio', 'music'] | [ 6.01294398e-01 -1.64501145e-02 -1.52722552e-01 -2.62571305e-01
-1.55457878e+00 -2.47926727e-01 5.81496596e-01 -3.24392110e-01
-4.18466888e-02 6.55527294e-01 7.29536474e-01 -2.38161415e-01
1.00558758e-01 -7.86005735e-01 -9.79173005e-01 -2.09436297e-01
-2.92059988e-01 3.27475935e-01 -5.16364239e-02 -2.21536860e-01
5.99248447e-02 1.56255066e-01 -1.54319763e+00 1.12368047e+00
2.34511837e-01 1.21123743e+00 -4.00875695e-02 1.24011433e+00
1.36601835e-01 7.96340108e-01 -8.46483409e-01 -2.34290525e-01
2.61147112e-01 -4.39275742e-01 -8.67282510e-01 1.10645711e-01
6.03990495e-01 -7.88631856e-01 -5.12311578e-01 7.52108395e-01
9.28212881e-01 1.09927401e-01 2.53144622e-01 -9.79750752e-01
-7.47915924e-01 1.23443508e+00 -4.24691230e-01 4.36579645e-01
6.61200583e-01 1.62899062e-01 1.13453186e+00 -1.11135340e+00
2.82317817e-01 1.56965601e+00 9.66441572e-01 4.79649305e-01
-1.33474851e+00 -1.01290619e+00 -3.15259248e-01 1.80629760e-01
-1.26693392e+00 -1.01141572e+00 5.09058297e-01 -1.17216982e-01
9.63262737e-01 1.97895512e-01 5.31689703e-01 1.17776930e+00
-1.51088953e-01 8.24052274e-01 6.43717349e-01 -2.68306494e-01
1.07463405e-01 -5.01459777e-01 -6.69939160e-01 1.08340964e-01
-4.69599187e-01 2.55099922e-01 -1.05253720e+00 -2.40378484e-01
1.29469597e+00 -4.78785306e-01 -4.16365355e-01 5.91154873e-01
-1.46864665e+00 6.10850215e-01 3.00562799e-01 1.94325522e-01
-3.33266139e-01 7.72059441e-01 6.23436868e-01 6.93215609e-01
5.88773191e-01 6.47244990e-01 -3.59070837e-01 -4.89032239e-01
-1.44043577e+00 4.81782466e-01 3.84335220e-01 1.00875032e+00
9.48284566e-02 6.33908093e-01 -1.70478895e-01 9.59133923e-01
-1.93277687e-01 2.43077517e-01 6.69311345e-01 -1.48526525e+00
5.82457840e-01 -5.54812551e-01 1.19771868e-01 -4.24400538e-01
-1.49026617e-01 -5.86749315e-01 -8.72995615e-01 1.26452416e-01
1.17591545e-01 -1.50789484e-01 -8.48738074e-01 1.49205923e+00
-1.13499686e-01 9.14373994e-01 -2.85223871e-02 1.00182498e+00
9.12364721e-01 1.03165340e+00 -4.17738557e-01 -3.74086164e-02
1.18304789e+00 -1.03163779e+00 -8.95751774e-01 2.86940616e-02
-1.27641723e-01 -1.20193648e+00 1.30122662e+00 7.24450469e-01
-1.88894427e+00 -1.00279760e+00 -1.08015811e+00 -2.85262972e-01
2.69672006e-01 -8.29215124e-02 6.46480560e-01 4.47854221e-01
-1.38078928e+00 1.10359073e+00 -4.99026746e-01 1.88564166e-01
5.70948541e-01 2.19543010e-01 -7.52098933e-02 1.53006151e-01
-1.23946404e+00 4.02927026e-02 4.59695533e-02 -3.43479395e-01
-1.30991125e+00 -1.19798100e+00 -6.95928693e-01 4.30623621e-01
-1.33858368e-01 -7.09619939e-01 1.94870222e+00 -1.01275301e+00
-1.79198265e+00 3.01565528e-01 -1.58714466e-02 -1.04896915e+00
4.30854112e-01 -5.31915843e-01 -8.60771656e-01 4.78912860e-01
7.31460303e-02 9.83344853e-01 1.45644248e+00 -8.52984548e-01
-6.40805662e-01 3.13647300e-01 -5.19815683e-02 1.34124324e-01
-7.59575218e-02 7.41056204e-02 -4.06808496e-01 -1.30587435e+00
1.03218429e-01 -5.63023150e-01 2.34885588e-02 6.48707449e-02
-1.80820152e-01 2.58425564e-01 9.08066332e-01 -6.81056201e-01
8.49425137e-01 -2.54552436e+00 -2.63474077e-01 -1.63594797e-01
1.90766864e-02 -4.84008528e-02 -5.78108490e-01 2.63378471e-01
-3.54252547e-01 2.54276991e-01 -1.30084353e-02 -3.41583341e-01
-1.07016012e-01 -2.64802992e-01 -9.37393904e-01 -1.31870043e-02
4.57817972e-01 7.40723193e-01 -9.24161136e-01 -3.47314030e-01
8.47114548e-02 8.07110310e-01 -9.85535383e-01 8.87988880e-02
-2.86896050e-01 4.42504793e-01 2.56968945e-01 5.95229387e-01
4.46644902e-01 -2.55835801e-01 -1.19620472e-01 -2.61740118e-01
2.32166573e-01 1.09305441e+00 -1.06832206e+00 2.02068067e+00
-8.58052492e-01 1.10173726e+00 1.96258381e-01 -5.15975296e-01
6.85495675e-01 8.79965663e-01 5.10408163e-01 -6.39745057e-01
-3.07365894e-01 2.76407927e-01 -7.29921684e-02 -1.31255999e-01
6.97890043e-01 -2.27765143e-01 2.03172773e-01 5.27476192e-01
2.54414320e-01 -6.08779013e-01 -6.01100735e-02 1.35990933e-01
1.16317713e+00 -1.02928177e-01 -1.93943843e-01 3.21569800e-01
5.09779751e-02 -6.84374154e-01 1.46065250e-01 7.83355176e-01
2.12199613e-01 1.02829027e+00 2.21685842e-01 -2.72535682e-01
-1.54324389e+00 -1.43195748e+00 -2.64702350e-01 1.38278532e+00
-3.86571348e-01 -6.75836980e-01 -6.58607662e-01 2.33002622e-02
-2.95885652e-01 4.64605212e-01 -1.30473748e-01 3.05903610e-03
-5.78814864e-01 -2.41035953e-01 1.13756013e+00 7.97276199e-01
5.48209965e-01 -1.14319801e+00 -5.71831279e-02 5.49208581e-01
-5.47733009e-01 -1.24304903e+00 -7.80501366e-01 3.45599093e-03
-1.02234304e+00 -6.02646053e-01 -9.16328490e-01 -8.57186437e-01
-1.30027592e-01 2.48937339e-01 1.43048906e+00 -8.58222917e-02
-8.36295858e-02 3.01816668e-02 -2.75341421e-01 -2.68851668e-01
-6.11714184e-01 -4.33664881e-02 7.80378357e-02 -9.36033055e-02
-1.03635892e-01 -1.13326311e+00 -6.70571029e-01 7.99687654e-02
-8.24708819e-01 1.40331620e-02 5.79985857e-01 1.09380853e+00
6.46607995e-01 2.22085759e-01 1.01040065e+00 -3.63775551e-01
8.91566396e-01 -1.22679561e-01 -3.22726130e-01 -5.21334469e-01
-1.55440494e-01 -2.50216812e-01 7.27090359e-01 -6.78404331e-01
-8.50035369e-01 -5.19635975e-02 -6.01351619e-01 -6.68714762e-01
-1.26683563e-01 2.40894869e-01 2.48553529e-01 2.08988816e-01
8.19563150e-01 1.66279465e-01 -2.18971923e-01 -6.74374521e-01
5.34765244e-01 8.47123325e-01 1.08634377e+00 -3.71849775e-01
5.50131500e-01 4.66273308e-01 -2.86883593e-01 -7.84384847e-01
-5.13770998e-01 -7.30183050e-02 -1.23661816e-01 9.92907360e-02
4.00898576e-01 -1.46088779e+00 -4.48758394e-01 2.17432886e-01
-1.17534792e+00 -5.47057986e-01 -7.18712568e-01 6.45373166e-01
-9.56560910e-01 1.85163140e-01 -1.30795145e+00 -6.07263505e-01
-5.87119043e-01 -9.85378981e-01 1.57570732e+00 -3.76434177e-01
-2.92328686e-01 -3.67854744e-01 -2.71804512e-01 1.35711044e-01
5.17702818e-01 -2.09062830e-01 7.88347423e-01 -2.37104326e-01
-7.17063487e-01 -2.94768121e-02 -1.77367806e-01 4.30543840e-01
-2.59529706e-02 -3.89016680e-02 -1.37794614e+00 -3.25577676e-01
-2.63099492e-01 -5.62316716e-01 1.10668218e+00 6.10401392e-01
1.80630910e+00 -5.93901515e-01 2.27835879e-01 7.31499195e-01
8.22232127e-01 4.90178727e-02 8.16342831e-01 -7.14532137e-02
1.40221924e-01 1.38655722e-01 3.29921335e-01 6.42998338e-01
-8.35839733e-02 8.37294161e-01 3.68504047e-01 -2.35522777e-01
-6.92047417e-01 -4.36166108e-01 4.99526054e-01 9.49450672e-01
1.10715516e-02 1.03170918e-02 -3.28183293e-01 7.24070132e-01
-1.37224722e+00 -1.29528749e+00 2.26511061e-01 2.12179828e+00
1.24800730e+00 3.31207544e-01 3.17385674e-01 7.49587834e-01
7.94263721e-01 2.14153975e-01 -3.44503999e-01 -3.92471015e-01
-1.08547546e-01 9.49002206e-01 7.07975999e-02 5.23978174e-01
-1.07486498e+00 7.74166822e-01 7.98145771e+00 9.38127816e-01
-1.18861866e+00 2.12020367e-01 6.60215735e-01 -7.11500287e-01
-2.57677078e-01 -5.38426399e-01 -2.99233586e-01 3.53204250e-01
1.38539433e+00 -1.78430617e-01 9.33001876e-01 7.63688684e-01
3.99643511e-01 5.89305758e-01 -1.43276525e+00 1.33862543e+00
-1.82558179e-01 -1.81869924e+00 2.93973953e-01 -1.05754912e-01
7.58790076e-01 1.92988932e-01 5.75407505e-01 3.37421179e-01
2.51883745e-01 -1.32706809e+00 8.83501768e-01 1.55228171e-02
1.39559996e+00 -8.93675327e-01 4.89282459e-01 -1.66152149e-01
-1.29028594e+00 -1.26040027e-01 -2.93746561e-01 -1.07138664e-01
4.33818430e-01 5.66731751e-01 -1.15782988e+00 -2.09720768e-02
9.50331688e-01 4.82179046e-01 -1.15109064e-01 9.66095567e-01
1.89574808e-02 9.05290246e-01 -2.90127635e-01 5.69669604e-01
1.00856155e-01 5.82317472e-01 4.82407391e-01 1.38717067e+00
7.51370192e-01 -1.69137940e-01 -1.82664707e-01 7.53455520e-01
-6.07154787e-01 -1.66861549e-01 -3.85790259e-01 -7.42619857e-02
8.65273952e-01 8.77585173e-01 -1.58365652e-01 -5.62531352e-01
-1.86268136e-01 8.73950899e-01 -3.16173136e-01 2.48689398e-01
-9.02523220e-01 -5.00581086e-01 6.93440735e-01 3.57935429e-01
4.19641376e-01 -2.44148374e-02 -3.89032602e-01 -8.52051735e-01
-4.88656610e-02 -1.23274267e+00 1.26690760e-01 -1.24987352e+00
-1.03459036e+00 7.27475524e-01 -3.85534018e-01 -1.44953227e+00
-8.58571470e-01 -1.96478054e-01 -5.25631309e-01 8.15626502e-01
-1.39944363e+00 -7.26638258e-01 -1.80026993e-01 7.88665593e-01
1.04330945e+00 -3.26313734e-01 9.96836126e-01 6.84486866e-01
2.68424302e-01 6.67114377e-01 -1.51944486e-03 1.23574130e-01
7.22227573e-01 -1.05475318e+00 1.23795724e+00 5.66634178e-01
6.12879038e-01 2.08999380e-01 7.35638261e-01 -1.89319387e-01
-9.77185309e-01 -1.34556365e+00 5.71770728e-01 4.31326926e-02
6.18348122e-01 -3.09038043e-01 -8.68117690e-01 5.99099100e-01
3.80134612e-01 -1.60738472e-02 7.03242183e-01 9.59863961e-02
-6.20600700e-01 -2.26586640e-01 -1.00158525e+00 4.85423326e-01
9.24346924e-01 -9.67414260e-01 -2.94630378e-01 4.60083276e-01
1.37537766e+00 -6.85354531e-01 -1.00818181e+00 2.76446402e-01
6.10183120e-01 -8.35052729e-01 1.27871203e+00 -5.18263042e-01
9.12512243e-01 -1.59251183e-01 -1.35925248e-01 -1.31748474e+00
-3.94028842e-01 -1.11324465e+00 -3.38141203e-01 8.77419174e-01
4.73413944e-01 9.38806459e-02 7.30501890e-01 -3.26137185e-01
-3.02892804e-01 -4.46751177e-01 -1.11703122e+00 -7.16329455e-01
-1.56539738e-01 -9.24429417e-01 1.09729397e+00 7.84126759e-01
-3.73507142e-02 3.69219661e-01 -6.62643254e-01 8.28151926e-02
3.83320749e-01 -2.17381399e-02 7.97805607e-01 -9.12641644e-01
-8.25172782e-01 -4.33656305e-01 -2.46429294e-01 -1.38167965e+00
-1.28177568e-01 -6.28066599e-01 -6.24995073e-03 -1.26388299e+00
-4.54998195e-01 -2.16924086e-01 -2.41558909e-01 2.62698472e-01
4.17086899e-01 8.14395189e-01 2.04960108e-01 9.42096338e-02
-3.37043852e-01 4.41220194e-01 1.10002613e+00 -4.56764162e-01
-3.06856900e-01 8.17794353e-02 -5.92479169e-01 6.32815540e-01
7.71000326e-01 -4.00178544e-02 -4.07474428e-01 -6.58426702e-01
2.02428952e-01 5.03878057e-01 4.70506281e-01 -1.39310110e+00
-1.27624184e-01 3.24873686e-01 5.63974261e-01 -7.38539279e-01
8.25504839e-01 -5.11409700e-01 1.85063571e-01 9.72135663e-02
-9.04410005e-01 -5.87481856e-02 3.20375293e-01 4.65221643e-01
-6.26805842e-01 9.32195485e-02 9.32665765e-01 -1.64570570e-01
-2.64393508e-01 2.34439701e-01 -3.58706087e-01 -7.92607293e-03
1.13738403e-01 8.14471766e-02 -2.04774931e-01 -1.23328614e+00
-9.02019978e-01 -3.41450810e-01 -3.75559703e-02 5.26830673e-01
9.21767175e-01 -1.81760824e+00 -9.56794381e-01 2.72593349e-01
-2.39487365e-01 9.91725698e-02 8.24606046e-02 4.41452235e-01
-4.89707589e-01 2.64882356e-01 -1.34955257e-01 -7.14101076e-01
-1.16281378e+00 4.82657909e-01 2.62513399e-01 3.80622707e-02
-7.56580651e-01 9.52219129e-01 9.67362449e-02 1.51261225e-01
6.23882055e-01 -6.19055808e-01 6.69385791e-02 -3.90377492e-01
1.16274917e+00 3.03853601e-01 2.33407542e-01 -2.86883444e-01
1.84413597e-01 1.05541078e-02 6.01428300e-02 -6.18719757e-01
1.35779142e+00 -4.10667732e-02 3.08957011e-01 4.50011939e-01
1.08041096e+00 1.64078772e-01 -1.29332256e+00 -2.44151235e-01
-3.69680703e-01 -6.04712188e-01 3.27893943e-01 -7.06920624e-01
-1.12910497e+00 1.02504575e+00 4.73937362e-01 3.69757116e-01
1.44117618e+00 -5.86684421e-02 1.19768918e+00 4.21722323e-01
1.31566331e-01 -1.07057798e+00 5.77454627e-01 4.00052398e-01
1.24115610e+00 -8.59229326e-01 -1.74446687e-01 -3.02839011e-01
-3.02000284e-01 1.19598866e+00 2.73684561e-01 -1.71258241e-01
3.99783492e-01 9.38790023e-01 -8.52002650e-02 2.67567784e-01
-1.15368605e+00 7.91745484e-02 1.57869399e-01 7.68178225e-01
7.73163438e-01 -4.25685793e-02 2.87781447e-01 5.52756786e-01
-7.71196425e-01 3.23963553e-01 6.16738498e-01 4.42093670e-01
-4.24720198e-01 -8.38613510e-01 -5.70955575e-01 3.71130764e-01
-8.19097221e-01 -6.83652997e-01 -1.97774768e-01 3.83842945e-01
-5.95215522e-02 1.08492124e+00 4.97285634e-01 -5.46737611e-01
5.60477786e-02 2.26346999e-02 4.53797370e-01 -3.86412889e-01
-6.36531711e-01 8.37544501e-01 2.42164746e-01 -6.37944102e-01
-4.19106960e-01 -3.45437676e-01 -1.08175457e+00 -4.99716163e-01
-2.83465385e-01 2.11265013e-02 5.41944325e-01 4.69446808e-01
5.52779198e-01 9.44349229e-01 7.38733172e-01 -1.23005641e+00
-5.07807910e-01 -1.23166144e+00 -5.48504472e-01 2.80161887e-01
7.92639375e-01 -9.36105661e-03 -1.17978960e-01 4.34009373e-01] | [15.550663948059082, 5.8164520263671875] |
bcce9bd4-6eb7-44da-b89d-0efb6b91a51c | inductive-detection-of-influence-operations | 2305.16544 | null | https://arxiv.org/abs/2305.16544v1 | https://arxiv.org/pdf/2305.16544v1.pdf | Inductive detection of Influence Operations via Graph Learning | Influence operations are large-scale efforts to manipulate public opinion. The rapid detection and disruption of these operations is critical for healthy public discourse. Emergent AI technologies may enable novel operations which evade current detection methods and influence public discourse on social media with greater scale, reach, and specificity. New methods with inductive learning capacity will be needed to identify these novel operations before they indelibly alter public opinion and events. We develop an inductive learning framework which: 1) determines content- and graph-based indicators that are not specific to any operation; 2) uses graph learning to encode abstract signatures of coordinated manipulation; and 3) evaluates generalization capacity by training and testing models across operations originating from Russia, China, and Iran. We find that this framework enables strong cross-operation generalization while also revealing salient indicators$\unicode{x2013}$illustrating a generic approach which directly complements transductive methodologies, thereby enhancing detection coverage. | ['Neil F. Johnson', 'David A. Broniatowski', 'Nicholas A. Gabriel'] | 2023-05-26 | null | null | null | null | ['specificity'] | ['natural-language-processing'] | [ 7.39979267e-01 4.73730683e-01 -6.38696909e-01 4.14022863e-01
-5.15949130e-01 -8.06402743e-01 1.10193801e+00 7.25758731e-01
4.30216119e-02 5.03372788e-01 9.88799334e-01 -7.37095833e-01
-5.09027958e-01 -1.08187664e+00 -6.10635221e-01 -3.74482155e-01
-5.58949232e-01 2.47894123e-01 -1.34311497e-01 -6.97761297e-01
5.63116074e-01 3.94383460e-01 -1.38197422e+00 4.85560745e-01
9.15744185e-01 5.74941635e-01 -2.91117460e-01 5.97074866e-01
8.74282867e-02 1.18929064e+00 -9.41969812e-01 -2.38580436e-01
3.04858629e-02 -5.05266964e-01 -5.73700070e-01 3.41603421e-02
-4.34361435e-02 8.13076645e-02 -2.40064889e-01 9.52429056e-01
4.49892104e-01 -2.88620204e-01 7.44448125e-01 -8.93716216e-01
-1.22546232e+00 1.06670988e+00 -5.15106916e-01 6.08445704e-01
8.55354309e-01 2.91441798e-01 1.38586628e+00 -1.05227441e-01
1.09178174e+00 1.42180574e+00 6.07090712e-01 3.21691409e-02
-1.24063158e+00 -7.72373974e-01 4.21291709e-01 1.61538981e-02
-7.46244311e-01 -2.01225698e-01 1.06450486e+00 -7.25166142e-01
7.34130204e-01 6.22999609e-01 1.26496363e+00 1.32581997e+00
2.41308376e-01 7.07334280e-01 1.17997777e+00 -3.93772244e-01
4.66320179e-02 -9.33408812e-02 -2.46150032e-01 8.16743731e-01
6.64798260e-01 1.61179945e-01 -5.84717691e-01 -3.69671881e-01
4.22912613e-02 -6.32874593e-02 -1.99484140e-01 4.37354028e-01
-1.52905631e+00 9.62708354e-01 5.99406838e-01 6.39027417e-01
-5.40058136e-01 2.01324746e-01 4.64501828e-01 4.61178154e-01
8.34811032e-01 1.23222971e+00 -2.36117408e-01 -3.91088188e-01
-3.20797086e-01 -4.34531532e-02 9.50525582e-01 7.37875402e-01
8.33591759e-01 1.47220954e-01 -2.36109763e-01 2.03470275e-01
-1.54452235e-01 9.01261747e-01 -5.42140044e-02 -5.74621558e-01
2.71160871e-01 1.40494466e+00 -8.95647258e-02 -1.82652211e+00
-6.19148612e-01 -4.91068840e-01 -4.06287909e-01 -4.74362701e-01
-4.79521602e-02 -4.72561330e-01 -5.72596490e-01 1.39836001e+00
3.35974753e-01 -3.84169668e-02 -1.99625224e-01 2.59021044e-01
7.28229344e-01 6.50149286e-01 2.89922282e-02 -3.37205052e-01
1.13518929e+00 -1.63338974e-01 -8.96476805e-01 -2.20975459e-01
1.22239697e+00 -3.68730903e-01 1.09902084e+00 1.93653226e-01
-4.76107657e-01 1.60393700e-01 -1.01635611e+00 5.75668812e-01
-8.27878237e-01 -3.62838238e-01 1.07148004e+00 6.62619889e-01
-7.78028488e-01 -5.50374836e-02 -2.17227831e-01 -2.67420739e-01
6.01419866e-01 2.04431117e-01 -2.89489259e-03 3.35050672e-01
-1.46868718e+00 9.08400238e-01 1.55177653e-01 -2.22071990e-01
-8.39387655e-01 -6.25711560e-01 -6.22869253e-01 -2.91516870e-01
7.96869338e-01 -2.14226633e-01 4.75239635e-01 -1.11041224e+00
-1.05973709e+00 9.05917406e-01 2.00389698e-01 -1.23384006e-01
3.78380343e-02 3.61843221e-02 -8.28803301e-01 1.82352990e-01
3.44247550e-01 4.61991727e-02 7.65506685e-01 -1.44449675e+00
-4.69847143e-01 -3.22859317e-01 4.08063740e-01 -1.75262839e-02
-7.96173573e-01 2.96552569e-01 5.63668668e-01 -6.30410612e-01
-1.27750784e-01 -9.36561346e-01 -1.82668108e-03 -8.29548478e-01
-7.40133524e-01 -2.72901773e-01 7.84175277e-01 -6.16199672e-01
1.59299719e+00 -1.80698812e+00 2.58487523e-01 4.83302951e-01
6.61351740e-01 1.52816072e-01 -5.60456961e-02 9.45918560e-01
2.65833259e-01 8.54871988e-01 2.18747541e-01 5.66395342e-01
4.70111072e-02 1.39878104e-02 -3.05776417e-01 7.00971067e-01
2.87238955e-01 1.18994093e+00 -1.22634244e+00 -1.14795037e-01
1.03447838e-02 -4.79027070e-02 -7.23394990e-01 -3.84580106e-01
-4.00080442e-01 3.46137166e-01 -6.34953380e-01 1.24495208e+00
-2.60912091e-01 -4.64499980e-01 4.91699547e-01 2.68945545e-02
-3.00158560e-01 3.13378453e-01 -5.94927013e-01 7.83159733e-01
-1.50648281e-01 9.32852447e-01 -9.58826542e-02 -1.13509846e+00
1.09430766e+00 -5.91442883e-02 5.82457125e-01 -8.16684842e-01
3.75940263e-01 -8.74985941e-03 2.13522926e-01 -8.51198077e-01
3.75131667e-01 3.23427916e-02 -7.19842076e-01 7.16309428e-01
-2.90312231e-01 -2.87855744e-01 1.34460062e-01 1.96701467e-01
1.43107116e+00 -4.86933440e-01 4.45967406e-01 -3.39845717e-01
2.44080737e-01 3.62079263e-01 4.60494876e-01 9.13832366e-01
-1.14427239e-01 -3.59994262e-01 9.93533373e-01 -3.04167330e-01
-6.50068879e-01 -7.30845273e-01 1.21469282e-01 1.55138671e+00
9.59741622e-02 -5.32889903e-01 -2.56596565e-01 -6.53435409e-01
3.59587967e-01 1.00651324e+00 -1.02326512e+00 -4.69138771e-01
-3.66417170e-01 -8.29513907e-01 6.71479642e-01 1.14703715e-01
2.54839242e-01 -1.07232046e+00 -1.35692716e-01 2.82429233e-02
-2.86113888e-01 -7.55927145e-01 -1.36544388e-02 -7.82288983e-02
-3.68320435e-01 -1.60194540e+00 2.87275463e-02 -4.92089957e-01
8.89819384e-01 1.20331571e-01 8.00705194e-01 1.77936226e-01
-1.36646941e-01 6.59951091e-01 -5.62163234e-01 -9.67365324e-01
-9.38773513e-01 2.12233692e-01 2.25617006e-01 2.88266242e-02
6.51341259e-01 -4.59829837e-01 -2.61743456e-01 1.27164930e-01
-8.37296546e-01 -3.42143714e-01 5.03246307e-01 4.99023646e-01
1.92064419e-02 -7.17797354e-02 1.05575001e+00 -1.04604602e+00
1.26595271e+00 -8.01928639e-01 -4.06423002e-01 2.92616069e-01
-8.28074276e-01 -4.10369337e-01 1.53594419e-01 -5.78380466e-01
-7.53251672e-01 -6.11470759e-01 4.88766164e-01 3.16378653e-01
2.62245923e-01 1.08680201e+00 3.70781481e-01 4.35169525e-02
1.19185078e+00 4.78711631e-03 1.14003018e-01 8.59056562e-02
5.01666546e-01 7.93174982e-01 8.59688148e-02 -4.74010468e-01
1.00538480e+00 6.74462080e-01 -1.82828233e-01 -9.68217909e-01
-7.80429780e-01 -4.21900183e-01 -4.70284402e-01 -9.34701562e-01
7.38642395e-01 -8.85881066e-01 -9.53390658e-01 7.37445354e-02
-7.96880186e-01 -2.29058713e-01 -1.99972317e-01 3.54709715e-01
-7.64202848e-02 -1.06585041e-01 -3.19677591e-01 -7.73526728e-01
-1.61567509e-01 -7.64826119e-01 8.70528817e-01 -8.58511999e-02
-5.16391098e-01 -1.24034071e+00 1.43618718e-01 5.21170557e-01
3.43431383e-01 1.00205410e+00 9.15486336e-01 -8.93244565e-01
-4.80715275e-01 -6.12657607e-01 -8.10441971e-02 6.66856691e-02
5.81407189e-01 2.27018937e-01 -4.88540411e-01 -2.02769652e-01
-3.15414250e-01 -2.02146873e-01 6.09864235e-01 4.02469598e-02
5.69041491e-01 -9.03817117e-01 -6.69603109e-01 2.68447921e-02
9.24566448e-01 2.71599233e-01 5.24382174e-01 6.27666295e-01
6.56991124e-01 7.78620481e-01 1.99546978e-01 5.11400521e-01
4.88709390e-01 6.06842749e-02 3.76541138e-01 -1.38468772e-01
6.06959574e-02 -4.76497978e-01 7.58687854e-01 1.09069157e+00
-3.82300168e-01 -3.38480473e-01 -1.27570748e+00 6.40338957e-01
-1.57025874e+00 -1.37662470e+00 -4.34812680e-02 1.51186883e+00
7.32844055e-01 3.44108254e-01 1.49472356e-01 6.36922643e-02
7.41606295e-01 5.81875801e-01 -3.41290355e-01 -5.26486933e-01
-4.20223445e-01 -6.10782728e-02 4.71828252e-01 3.15420181e-01
-8.56557190e-01 8.85322511e-01 7.04677916e+00 3.26023728e-01
-1.10437763e+00 1.49250388e-01 5.03307164e-01 1.11920223e-01
-9.50087011e-01 2.45091040e-03 -4.79596496e-01 1.60273433e-01
7.85127282e-01 -4.80608225e-01 4.75129187e-01 5.73466122e-01
3.90208334e-01 1.76844522e-01 -5.75942099e-01 2.79538482e-01
3.94418567e-01 -1.81819022e+00 1.43375412e-01 3.03766578e-01
1.14561498e+00 7.40719289e-02 1.47764888e-02 2.88001627e-01
5.83335698e-01 -8.25097263e-01 5.21262050e-01 4.59247828e-01
8.30329180e-01 -4.88904357e-01 3.46976936e-01 1.17688581e-01
-8.00886393e-01 -7.25902259e-01 1.23935834e-01 -5.63583612e-01
-6.86545968e-02 6.98632121e-01 -1.07288146e+00 2.48100549e-01
3.68840426e-01 1.00008678e+00 -7.21960843e-01 3.60050231e-01
-7.93750823e-01 1.01534700e+00 -6.20138571e-02 -8.06374490e-01
1.95849165e-01 -4.94103283e-02 1.15329909e+00 1.24623728e+00
4.68163900e-02 -1.82559311e-01 2.32971683e-01 7.66439319e-01
-2.23271608e-01 1.15646362e-01 -1.27320874e+00 -8.42381358e-01
6.21971965e-01 9.33159769e-01 -7.87709713e-01 -2.19352111e-01
-2.05628693e-01 2.50558585e-01 7.35821277e-02 5.35680830e-01
-6.69458270e-01 -2.57533580e-01 4.65383887e-01 4.18736368e-01
1.56415462e-01 -6.69034421e-02 -2.32360303e-01 -9.62174296e-01
-4.23656762e-01 -1.19260204e+00 3.19312334e-01 -2.38088235e-01
-1.06861544e+00 7.35621303e-02 1.23238556e-01 -9.17701662e-01
-9.71332639e-02 -4.72178727e-01 -6.05407059e-01 8.78289044e-02
-1.02264309e+00 -1.36165714e+00 -7.09710792e-02 1.66291282e-01
-1.28145382e-01 -3.27607363e-01 6.41042471e-01 -1.87248457e-02
-6.20909214e-01 2.90576160e-01 -1.09070636e-01 1.12615012e-01
2.00505406e-01 -9.97100234e-01 1.18213877e-01 6.32876694e-01
-1.87090635e-01 4.65525270e-01 5.44664264e-01 -1.10211742e+00
-1.75850129e+00 -9.68265653e-01 9.72205698e-01 -7.75709748e-01
1.35060740e+00 -5.32195032e-01 -2.99931020e-01 7.83239543e-01
1.43908486e-01 -6.01474941e-01 9.16489303e-01 7.26832330e-01
-4.41816092e-01 -1.23376496e-01 -8.79684269e-01 8.50735545e-01
1.42822599e+00 -9.12853777e-01 -8.55643511e-01 7.85906255e-01
8.86305273e-01 1.52524084e-01 -1.05414200e+00 4.22763139e-01
2.29421064e-01 -4.66774106e-01 7.48005867e-01 -8.90477657e-01
3.88348401e-01 -2.22686350e-01 -3.32806259e-02 -1.27625227e+00
-5.79993784e-01 -1.25992596e+00 -1.38384715e-01 1.09054899e+00
7.99121678e-01 -1.08160317e+00 1.28634542e-01 7.57315159e-02
-1.76025555e-01 -6.64432406e-01 -6.27800167e-01 -5.61924815e-01
-1.26845405e-01 -3.21633548e-01 6.73086286e-01 1.54310334e+00
6.32623017e-01 3.37949932e-01 -1.43541589e-01 4.44171466e-02
2.34498203e-01 4.84024175e-02 8.16610694e-01 -1.27697492e+00
8.48168805e-02 -8.39114547e-01 -6.80060923e-01 -3.31986725e-01
6.92495927e-02 -9.78826940e-01 -3.13678622e-01 -1.53623915e+00
1.39205918e-01 -3.30530733e-01 -1.84759319e-01 5.43474317e-01
-1.53351322e-01 4.03371044e-02 -1.48241252e-01 1.58715427e-01
-6.11208200e-01 3.81300986e-01 1.45866084e+00 -5.57866037e-01
-5.22176147e-01 -4.06700999e-01 -1.57315528e+00 6.42234385e-01
8.64534378e-01 -3.62093121e-01 -2.28458554e-01 3.45411003e-02
1.06480157e+00 -4.55653042e-01 3.44914585e-01 -6.08544409e-01
5.49343154e-02 -5.18974662e-01 -4.27657692e-03 -1.26784980e-01
-3.45797151e-01 -1.57686904e-01 1.48806989e-01 8.64739716e-01
-5.60487330e-01 1.99159458e-01 1.29184335e-01 7.90526688e-01
1.05237201e-01 3.89891416e-01 3.85487154e-02 -1.33787975e-01
-5.52529991e-01 -1.95096821e-01 -7.58419514e-01 2.73834050e-01
1.08957267e+00 -8.25492665e-02 -9.20219660e-01 -5.88545740e-01
-4.19012487e-01 1.17506264e-02 3.16400826e-01 7.47060299e-01
3.13919514e-01 -1.13200581e+00 -1.04577839e+00 -2.50196159e-01
3.29844326e-01 -5.95302463e-01 -2.85258204e-01 1.05729711e+00
-3.49918485e-01 1.55725539e-01 2.23519295e-01 -3.82305562e-01
-7.82622635e-01 5.55919349e-01 6.33740202e-02 -5.02568893e-02
-3.88073057e-01 6.74943447e-01 -1.88752785e-01 -3.41206878e-01
-4.75608669e-02 -9.29498971e-02 -4.49803025e-01 8.66301656e-01
4.46109891e-01 5.03269851e-01 -3.88764918e-01 -7.02261746e-01
-3.98232788e-01 4.95991707e-02 2.08169833e-01 3.30662072e-01
1.46086109e+00 4.71212380e-02 -4.84475076e-01 4.85009342e-01
9.40829456e-01 5.07737994e-01 -5.75228155e-01 -1.41372800e-01
2.77871609e-01 -3.32803696e-01 7.73572028e-02 -8.91338229e-01
-6.87607169e-01 8.15986693e-02 1.60351247e-01 8.25303495e-01
9.23653245e-01 3.70428801e-01 2.68983155e-01 4.72287714e-01
7.75090531e-02 -1.36927772e+00 5.26129782e-01 3.05783123e-01
1.28922045e+00 -9.14907813e-01 3.60335946e-01 -2.53766656e-01
-3.61445189e-01 8.03311408e-01 2.44636253e-01 4.18383703e-02
6.26527011e-01 7.54128247e-02 -2.39345670e-01 -9.30109978e-01
-6.30997300e-01 -4.28482473e-01 2.34802589e-01 6.65563345e-01
1.93970010e-01 5.30787885e-01 -5.62409699e-01 1.80066630e-01
-8.77332911e-02 -3.66898805e-01 5.57197928e-01 1.02602887e+00
-5.47986567e-01 -5.98060131e-01 -4.67680901e-01 9.19807434e-01
-2.77633637e-01 -2.45199099e-01 -1.13526165e+00 9.45504248e-01
2.77641684e-01 1.19629371e+00 -1.82964467e-02 -1.03009391e+00
3.95846367e-01 -2.70154059e-01 2.72895731e-02 -4.82776463e-01
-8.08434904e-01 -2.35989273e-01 8.63574803e-01 -2.78741866e-01
-6.32972300e-01 -9.26444769e-01 -1.00735712e+00 -6.50483847e-01
-4.93508816e-01 -3.01706195e-02 5.09404600e-01 9.16261971e-01
7.45680869e-01 7.22134590e-01 8.53683949e-01 -4.75791305e-01
-2.02214941e-01 -1.08747971e+00 -1.84681475e-01 5.08201063e-01
1.60991848e-01 -8.53794873e-01 -7.58268595e-01 -8.53463262e-02] | [8.509660720825195, 10.054767608642578] |
b67ac0fc-61ec-452b-83f1-0c1a8685ad0f | unifacegan-a-unified-framework-for-temporally | 2108.05650 | null | https://arxiv.org/abs/2108.05650v1 | https://arxiv.org/pdf/2108.05650v1.pdf | UniFaceGAN: A Unified Framework for Temporally Consistent Facial Video Editing | Recent research has witnessed advances in facial image editing tasks including face swapping and face reenactment. However, these methods are confined to dealing with one specific task at a time. In addition, for video facial editing, previous methods either simply apply transformations frame by frame or utilize multiple frames in a concatenated or iterative fashion, which leads to noticeable visual flickers. In this paper, we propose a unified temporally consistent facial video editing framework termed UniFaceGAN. Based on a 3D reconstruction model and a simple yet efficient dynamic training sample selection mechanism, our framework is designed to handle face swapping and face reenactment simultaneously. To enforce the temporal consistency, a novel 3D temporal loss constraint is introduced based on the barycentric coordinate interpolation. Besides, we propose a region-aware conditional normalization layer to replace the traditional AdaIN or SPADE to synthesize more context-harmonious results. Compared with the state-of-the-art facial image editing methods, our framework generates video portraits that are more photo-realistic and temporally smooth. | ['Jiebo Luo', 'Zhifeng Li', 'Linchao Bao', 'Sheng Wang', 'Li Shen', 'Xuan Wang', 'Hao Wang', 'HaoZhi Huang', 'Meng Cao'] | 2021-08-12 | null | null | null | null | ['face-reenactment', 'facial-editing'] | ['computer-vision', 'computer-vision'] | [ 4.67312306e-01 -2.27166042e-01 -9.58976597e-02 -5.56996524e-01
-3.48557234e-01 -3.36803317e-01 6.72768176e-01 -6.67493522e-01
-1.98509961e-01 6.87051058e-01 6.37091249e-02 2.11505234e-01
2.14543924e-01 -6.57476962e-01 -6.98485494e-01 -7.14380085e-01
5.44303417e-01 -1.30200580e-01 9.31997672e-02 -2.20423177e-01
-5.56429336e-03 6.67510688e-01 -1.58019722e+00 1.20578460e-01
9.22526598e-01 8.03598881e-01 7.24555328e-02 2.48956367e-01
-2.19728440e-01 6.30375683e-01 -1.49324924e-01 -6.53633237e-01
2.95396984e-01 -7.32439756e-01 -4.95035052e-01 5.61187983e-01
8.35789442e-01 -5.21756530e-01 -4.62039202e-01 1.10242116e+00
4.80374962e-01 2.99459875e-01 3.26303542e-01 -1.34030938e+00
-6.94096684e-01 2.38649384e-03 -1.08547187e+00 -1.70650989e-01
4.10592496e-01 -4.88067605e-02 3.54647577e-01 -1.11554313e+00
8.37633371e-01 1.43700528e+00 5.50696850e-01 8.56569529e-01
-1.34058702e+00 -1.01610816e+00 5.91870427e-01 1.91705108e-01
-1.49935961e+00 -9.16436136e-01 1.17442727e+00 -1.69002190e-01
3.31078172e-01 4.07094955e-01 8.68056595e-01 9.90444720e-01
-8.37783068e-02 5.73206782e-01 8.75181675e-01 -3.39520365e-01
-8.82608816e-02 -2.40154862e-01 -7.14813113e-01 8.30914557e-01
-1.53952762e-01 2.75247265e-02 -4.94573444e-01 -3.25640216e-02
1.25879109e+00 3.60578001e-01 -4.44350511e-01 -4.45910066e-01
-1.20458710e+00 4.67559755e-01 1.68301463e-01 1.72584385e-01
-3.39980692e-01 1.60134494e-01 4.18897957e-01 2.63597220e-01
7.45531440e-01 -2.06259280e-01 -6.97532371e-02 1.54984310e-01
-1.31033766e+00 3.99039328e-01 2.92483866e-01 9.75333571e-01
8.01770508e-01 2.74707675e-01 -3.62546384e-01 1.02850688e+00
2.47169495e-01 3.41235667e-01 2.26277471e-01 -1.20278406e+00
2.41750792e-01 3.25391442e-01 6.10688478e-02 -1.21471727e+00
5.29977977e-02 -1.02329865e-01 -8.45422387e-01 2.55928248e-01
-4.62600216e-03 1.15986899e-01 -9.12255824e-01 1.93629980e+00
7.85875618e-01 6.19189024e-01 -2.89833754e-01 8.71032834e-01
5.89154899e-01 6.15731895e-01 -4.35407739e-03 -5.69455266e-01
1.20852065e+00 -1.11864996e+00 -1.24786496e+00 2.36647993e-01
7.97405988e-02 -1.01724017e+00 9.84500170e-01 3.02649766e-01
-1.28743947e+00 -4.94584709e-01 -8.74767125e-01 -3.79790127e-01
1.58702388e-01 9.11760703e-02 5.12302697e-01 5.44330359e-01
-1.01237106e+00 3.92215759e-01 -8.12164664e-01 -2.05226168e-01
5.05410254e-01 1.25257522e-01 -7.11138010e-01 -1.12133779e-01
-8.94530475e-01 6.60866976e-01 8.41492191e-02 3.31848115e-01
-7.82444894e-01 -6.33127987e-01 -9.54621434e-01 -1.82555109e-01
5.06275177e-01 -7.52823234e-01 1.15797865e+00 -1.60303724e+00
-2.07312250e+00 9.52968419e-01 -5.86411595e-01 9.62787718e-02
9.00873303e-01 -1.33614227e-01 -5.48336864e-01 1.22928582e-01
-5.58091812e-02 7.53888965e-01 1.17695451e+00 -1.25828171e+00
-4.50488389e-01 -2.15092227e-01 2.36206520e-02 3.62063557e-01
-4.83576447e-01 2.75177181e-01 -1.11500442e+00 -1.26075125e+00
1.26635641e-01 -7.79133320e-01 -8.42184853e-03 6.57865226e-01
-8.12279731e-02 1.64363563e-01 1.20661521e+00 -7.62291551e-01
1.28863466e+00 -2.11939907e+00 4.45370734e-01 3.29960883e-02
1.88453346e-01 3.54720205e-01 -3.17174286e-01 1.77695110e-01
-2.41819099e-01 -7.00061619e-02 -4.35179025e-01 -8.52584362e-01
-2.48247176e-01 1.62496328e-01 -2.23701730e-01 5.63094437e-01
2.30060145e-01 6.43662512e-01 -9.25604165e-01 -7.32456565e-01
3.48919034e-01 9.46721375e-01 -7.54996002e-01 1.15947261e-01
-1.36316717e-01 8.02015007e-01 -2.93868929e-01 8.11652482e-01
1.13881230e+00 2.05826774e-01 2.03174502e-01 -3.65270853e-01
-2.37946853e-01 -3.38114053e-01 -9.60544467e-01 2.06871986e+00
-5.45471787e-01 4.41873282e-01 3.71591330e-01 -6.85323715e-01
7.91958094e-01 4.15976226e-01 5.55946290e-01 -5.98305285e-01
1.41505390e-01 1.02033943e-01 -4.95670944e-01 -4.09638166e-01
5.02380252e-01 -1.94082558e-01 5.44757485e-01 2.53609091e-01
-6.80450574e-02 -1.16790958e-01 7.28812516e-02 -1.02867715e-01
5.26033878e-01 7.76712477e-01 7.15777427e-02 -4.51256149e-02
8.61530006e-01 -5.05874872e-01 1.12952793e+00 -5.73885217e-02
-1.84276387e-01 1.00118876e+00 3.10568541e-01 -4.69549298e-01
-1.01441145e+00 -8.69960725e-01 -9.91454720e-02 8.38261843e-01
2.86439687e-01 -4.79792506e-01 -1.01830351e+00 -6.16857350e-01
-1.79865405e-01 3.46916020e-01 -5.96444666e-01 -1.68194827e-02
-7.67917454e-01 -3.50929499e-01 3.70067209e-01 2.02867299e-01
8.28722298e-01 -8.38106513e-01 -1.84315845e-01 2.32406601e-01
-3.07080507e-01 -1.06462717e+00 -1.21541870e+00 -7.82249987e-01
-8.29431653e-01 -9.33457077e-01 -1.06097138e+00 -9.19843495e-01
1.01359975e+00 5.50866544e-01 6.77853703e-01 1.51503697e-01
-2.28005856e-01 2.05742568e-01 -3.46644431e-01 -8.21974874e-02
-1.89728171e-01 -2.83502668e-01 2.00584963e-01 6.17694914e-01
-6.33160621e-02 -8.50934744e-01 -8.44987333e-01 5.24467885e-01
-1.18654275e+00 4.87863392e-01 2.09834099e-01 8.05870235e-01
7.55635738e-01 -1.76998228e-01 3.78212452e-01 -7.97512591e-01
2.38957241e-01 -1.43690005e-01 -5.64238787e-01 4.72407311e-01
-3.47884476e-01 -1.92593396e-01 7.37793982e-01 -6.67699754e-01
-1.54174042e+00 2.13072196e-01 -1.26550883e-01 -1.05291152e+00
1.75545335e-01 1.02897733e-02 -4.38794553e-01 -4.97402221e-01
1.96409121e-01 3.82630229e-01 3.03954512e-01 -3.97303075e-01
4.20413524e-01 2.52302438e-01 6.02129459e-01 -5.31093955e-01
1.02110362e+00 7.99338043e-01 -1.22209556e-01 -7.39416838e-01
-3.84172827e-01 1.64315328e-01 -7.58287311e-01 -5.17448246e-01
7.89528012e-01 -8.84246409e-01 -5.86240828e-01 8.02366018e-01
-1.21032655e+00 -1.40896052e-01 -4.05454710e-02 3.26877981e-01
-5.62673986e-01 5.98593652e-01 -4.21427399e-01 -6.18448019e-01
-2.20595613e-01 -1.18841946e+00 1.15742588e+00 2.67260760e-01
8.50690380e-02 -7.00811267e-01 -6.02196269e-02 1.55950442e-01
5.61995924e-01 5.77911258e-01 5.46610653e-01 3.30664486e-01
-5.40290594e-01 2.54389849e-02 -1.74974829e-01 3.24944586e-01
5.31323612e-01 3.91940922e-01 -8.06874156e-01 -2.91775912e-01
1.08035365e-02 -9.66728013e-03 7.68396795e-01 2.06461474e-01
1.36863244e+00 -4.36334699e-01 -2.58486211e-01 1.03464842e+00
1.02485502e+00 3.63166869e-01 7.98407853e-01 4.13192287e-02
9.18014228e-01 6.08941734e-01 7.23151028e-01 4.63342488e-01
2.51723230e-01 9.83618021e-01 2.60200292e-01 -2.41666377e-01
-1.68970406e-01 -4.94559705e-01 3.73725653e-01 6.94791198e-01
-3.10687810e-01 1.09291970e-04 -2.64339149e-01 3.37784201e-01
-1.85968018e+00 -1.09703219e+00 3.84517252e-01 2.20708418e+00
1.00938380e+00 -4.31529760e-01 -9.15402174e-02 -5.37949689e-02
1.01682591e+00 4.51579839e-01 -5.99000037e-01 -8.24098885e-02
-5.69098108e-02 8.65234435e-02 1.58042550e-01 5.06621897e-01
-9.28944767e-01 1.09310412e+00 5.67835140e+00 1.06308949e+00
-1.40666258e+00 2.41177946e-01 6.89494073e-01 -3.80576551e-01
-6.10701740e-01 4.18165140e-02 -4.30766076e-01 5.22280276e-01
7.57218823e-02 -1.92159280e-01 6.10926092e-01 5.61427593e-01
4.68461990e-01 1.48713231e-01 -8.14191937e-01 1.41493475e+00
3.95963132e-01 -1.38140523e+00 3.13717544e-01 -2.40491051e-02
8.38040650e-01 -7.63645947e-01 1.30774289e-01 -8.91415328e-02
-2.41200209e-01 -8.77390802e-01 9.91351545e-01 8.77920687e-01
1.12900269e+00 -9.97454703e-01 1.06336728e-01 -1.21947318e-01
-1.40243542e+00 1.98443592e-01 -8.95918757e-02 2.50920266e-01
4.43894207e-01 4.29570377e-01 -2.01589689e-02 7.08959281e-01
5.77880621e-01 9.01532769e-01 -2.31986463e-01 8.16354334e-01
-1.54855534e-01 -5.74042555e-04 -1.83098525e-01 6.06250763e-01
-9.53992754e-02 -6.18854642e-01 5.52804351e-01 7.65661538e-01
6.22201681e-01 3.59516621e-01 8.55544731e-02 8.35736156e-01
-3.11378002e-01 2.28854969e-01 -4.99627829e-01 1.56141847e-01
5.97877383e-01 1.34806144e+00 -5.08153260e-01 -1.85064465e-01
-5.07960498e-01 1.46947181e+00 1.41945034e-01 3.70506763e-01
-1.05028403e+00 -2.87990957e-01 8.64219010e-01 2.47094497e-01
1.93361625e-01 -2.21409872e-01 1.86253846e-01 -1.45355773e+00
2.25756869e-01 -8.09470713e-01 -7.59959668e-02 -6.38530374e-01
-1.10495126e+00 6.29567087e-01 -2.86826454e-02 -1.40915120e+00
-3.70174833e-02 -1.43912584e-01 -8.77293825e-01 6.90415084e-01
-1.69183397e+00 -1.43531466e+00 -4.56627846e-01 9.06273484e-01
6.09994352e-01 -2.12786142e-02 5.48603475e-01 6.02729917e-01
-7.56845891e-01 8.67188811e-01 -1.11964718e-01 -6.06114119e-02
1.08798099e+00 -5.02386987e-01 3.46338272e-01 8.72806907e-01
-2.25500792e-01 6.81311965e-01 4.70217258e-01 -5.80412328e-01
-1.35117543e+00 -1.31724870e+00 7.30355561e-01 -7.32009858e-02
1.69161946e-01 -3.47978771e-01 -1.00188518e+00 6.50789917e-01
2.67237991e-01 1.99200824e-01 4.11589533e-01 -4.21929002e-01
-3.00730467e-01 -4.10571903e-01 -1.36496139e+00 9.37067866e-01
1.49940610e+00 -4.87231106e-01 -1.27229273e-01 1.53167799e-01
6.38971150e-01 -5.72180688e-01 -7.10027158e-01 5.81457019e-01
8.11837792e-01 -9.32470739e-01 7.90801525e-01 -2.32071444e-01
2.51812935e-01 -7.12822437e-01 8.03614929e-02 -9.34705615e-01
-1.53960809e-01 -1.12847817e+00 1.41018927e-02 1.54521585e+00
-1.80817723e-01 -5.87401748e-01 6.71940684e-01 5.25584280e-01
-1.02706015e-01 -6.16587162e-01 -9.49835122e-01 -5.74384451e-01
-3.64457637e-01 -2.91542225e-02 8.33177924e-01 1.21565425e+00
-4.55902666e-01 -2.04757050e-01 -8.48128319e-01 -1.47672042e-01
6.64092004e-01 3.78407389e-02 9.00993943e-01 -9.22686934e-01
7.52282515e-02 -4.04751837e-01 -2.30507150e-01 -8.79030108e-01
3.32768500e-01 -6.60455108e-01 -8.03287774e-02 -1.06753349e+00
1.12068936e-01 -3.29574436e-01 -1.75328124e-02 5.39136887e-01
-1.08562864e-01 4.82856750e-01 2.81463206e-01 1.85830027e-01
-2.34841511e-01 9.75882828e-01 1.62421370e+00 7.64503330e-02
-2.94883341e-01 -3.01333994e-01 -5.74731648e-01 7.22769558e-01
5.12391686e-01 -8.42279941e-02 -6.29786611e-01 -6.38234019e-01
-1.10726319e-01 1.58397794e-01 3.47115934e-01 -7.68071651e-01
1.26322538e-01 -4.89748567e-01 4.36021835e-01 -4.16667074e-01
5.41828215e-01 -8.33268344e-01 5.96131742e-01 6.51059747e-02
-1.16821766e-01 2.29342014e-01 1.29982203e-01 5.82333088e-01
-4.19742346e-01 1.32323831e-01 1.10137832e+00 -5.34800859e-03
-6.05177760e-01 9.57610846e-01 -7.56809711e-02 -2.91404158e-01
1.31234682e+00 -3.47199261e-01 9.75623876e-02 -4.53934968e-01
-6.40575528e-01 1.83286563e-01 9.01632845e-01 6.62200093e-01
7.58390784e-01 -1.78150845e+00 -7.76848137e-01 5.24748266e-01
-4.65608798e-02 -2.43798993e-03 7.86322355e-01 8.71105909e-01
-6.64837420e-01 -8.34814981e-02 -4.55026269e-01 -4.06591147e-01
-1.49493349e+00 3.67489725e-01 3.93310189e-01 2.31460363e-01
-5.94485581e-01 7.21514225e-01 4.26655054e-01 -2.66905874e-01
1.74067616e-01 4.56116609e-02 -5.99744022e-02 2.87319589e-02
7.05859900e-01 1.86885387e-01 -1.07880928e-01 -8.90249252e-01
-2.34659120e-01 8.98424447e-01 -2.07819149e-01 -2.17115015e-01
1.17605925e+00 -3.04405183e-01 -3.06493044e-01 1.00369111e-01
1.09970951e+00 1.49558872e-01 -1.69762468e+00 -2.70063698e-01
-5.79559445e-01 -1.07672811e+00 -1.92744341e-02 -2.96662360e-01
-1.49165034e+00 6.15371108e-01 4.69486356e-01 -4.63474184e-01
1.52478111e+00 -5.26123106e-01 8.99061739e-01 -1.35480762e-01
2.89054036e-01 -1.08346510e+00 -9.09896567e-02 2.98189074e-01
1.18730664e+00 -9.29267228e-01 3.59081253e-02 -6.62954330e-01
-5.96116602e-01 1.11236906e+00 7.34236002e-01 -1.09039471e-02
5.37717640e-01 -1.28237018e-02 -1.34087309e-01 2.53459334e-01
-5.41115820e-01 2.10278735e-01 1.87867105e-01 4.82252061e-01
4.82880890e-01 -2.92947471e-01 -4.83854771e-01 1.36703998e-01
1.16977297e-01 3.12336415e-01 2.15029761e-01 8.34190845e-01
-1.99280456e-02 -1.40115345e+00 -3.74543965e-01 -2.35266984e-02
-3.18877161e-01 -2.76589906e-03 -8.94644260e-02 7.64799237e-01
2.14566514e-01 6.46463394e-01 1.13899410e-01 -2.66530424e-01
3.10428739e-01 -1.19451331e-02 8.46342981e-01 -4.02837038e-01
-1.89006194e-01 3.91170740e-01 -2.32340977e-01 -6.49307489e-01
-8.57022345e-01 -6.37979269e-01 -1.10581589e+00 -6.33744001e-01
-2.41921902e-01 -1.93418920e-01 4.03079599e-01 7.18365312e-01
5.01834810e-01 4.53617364e-01 8.53563547e-01 -1.32340312e+00
1.71156675e-02 -6.44154847e-01 -6.12094045e-01 3.40993136e-01
3.18123907e-01 -8.19499850e-01 -8.95327143e-03 4.00347501e-01] | [12.709600448608398, -0.23991349339485168] |
dcbf13ad-85de-4268-8ffb-d1c2eed03cec | learning-joint-intensity-in-a-multivariate | null | null | https://openreview.net/forum?id=VvpRCm5aWYr | https://openreview.net/pdf?id=VvpRCm5aWYr | Learning Joint Intensity in a Multivariate Poisson Process on Statistical Manifolds | We show that generalized additive models (GAMs) can be treated via the log-linear model on a structured sample space, which has a well established information geometric background. Connecting GAMs with multivariate stochastic processes, we present the additive Poisson process (APP), a novel framework that can model the higher-order interaction effects of the intensity functions in stochastic processes using lower dimensional projections. Learning of the model is achieved via convex optimization, thanks to the dually flat statistical manifold generated by the log-linear model. | ['Mahito Sugiyama', 'Lamiae Azizi', 'Feng Zhou', 'Simon Luo'] | 2020-10-19 | null | null | null | neurips-workshop-dl-ig-2020-12 | ['additive-models'] | ['methodology'] | [ 1.80857882e-01 1.65260136e-01 2.26939201e-01 -3.33871633e-01
-5.42382598e-01 -3.91019762e-01 1.03866434e+00 -1.97989911e-01
-2.07380995e-01 4.91813451e-01 2.69444376e-01 -2.50653573e-03
-5.36890626e-01 -9.30173755e-01 -9.39156592e-01 -1.07793880e+00
-3.37783784e-01 8.34150255e-01 -2.17582777e-01 3.69026124e-01
1.51244611e-01 5.76998619e-03 -1.02291203e+00 -1.99156508e-01
8.67014825e-01 5.64947367e-01 4.95374650e-02 7.97233284e-01
-1.49281591e-01 5.58252394e-01 3.03550679e-02 -7.44706213e-01
2.67081678e-01 -1.98231921e-01 -1.57138526e-01 7.94315189e-02
2.40233392e-01 -7.69597143e-02 -2.58870721e-01 1.12967479e+00
4.25203107e-02 1.35168418e-01 1.50438130e+00 -1.30501032e+00
-1.12748277e+00 5.60371637e-01 -9.15715933e-01 -6.87222257e-02
2.53854811e-01 -3.56383361e-02 9.02163744e-01 -1.28852379e+00
1.89127743e-01 1.94544220e+00 7.94036746e-01 1.06338277e-01
-1.81231809e+00 -6.33454993e-02 -4.14001867e-02 -2.97513694e-01
-1.38531756e+00 1.86004072e-01 5.54226160e-01 -7.98933148e-01
4.54448834e-02 2.55429924e-01 4.76752222e-01 1.19472587e+00
7.07768977e-01 1.00843680e+00 1.41102958e+00 -4.11547393e-01
5.56578994e-01 2.59123910e-02 5.46689034e-01 5.70926428e-01
4.22873378e-01 3.22283596e-01 -4.08397943e-01 -5.09964883e-01
1.20939302e+00 2.67567605e-01 1.06181785e-01 -7.47761846e-01
-7.94737875e-01 1.06981933e+00 -9.24608558e-02 -1.74249098e-01
-6.85060620e-01 5.01049459e-01 -2.89897501e-01 -1.15398318e-01
7.62560010e-01 -2.15360537e-01 -8.23935345e-02 -1.49636790e-01
-5.31392932e-01 3.91070217e-01 1.10610712e+00 1.19420540e+00
6.63196862e-01 2.36379076e-03 -3.67007077e-01 6.47025347e-01
1.03163993e+00 1.11366940e+00 -5.22376299e-01 -9.02355909e-01
4.76551712e-01 -4.43472061e-03 3.33341897e-01 -7.22829759e-01
-3.21695745e-01 -2.71023154e-01 -1.18500030e+00 2.33486742e-01
9.25893366e-01 -3.15802097e-01 -4.68252331e-01 1.77129602e+00
2.11907953e-01 4.73505914e-01 -1.80754021e-01 2.59705365e-01
2.33196020e-02 7.53514946e-01 3.67204070e-01 -2.37715736e-01
1.16659749e+00 -4.21811223e-01 -7.94959426e-01 1.33378699e-01
2.20652655e-01 -3.98805559e-01 1.04828453e+00 2.49820441e-01
-1.49678993e+00 -4.27367873e-02 -4.35004979e-01 1.10009141e-01
-1.29985347e-01 -1.66682377e-01 3.77829492e-01 1.02261209e+00
-1.32102263e+00 7.43704438e-01 -1.12533593e+00 -1.47131011e-01
6.29197359e-01 7.97291249e-02 1.35784492e-01 -2.58104742e-01
-6.77822173e-01 6.34380758e-01 -2.52479494e-01 1.73304547e-02
-1.03826153e+00 -1.37909830e+00 -7.99944937e-01 1.09231934e-01
2.44009271e-01 -1.10859847e+00 9.40703154e-01 -1.65194884e-01
-1.65838361e+00 6.65197670e-01 -3.21442634e-01 -4.36329007e-01
9.05669510e-01 -5.09483635e-01 3.16991746e-01 -8.30570906e-02
-1.64983109e-01 -4.69931997e-02 1.17963719e+00 -1.30894136e+00
-6.50516599e-02 -8.13311338e-01 -2.46931791e-01 2.65729636e-01
2.10991055e-01 2.08105356e-01 4.86521535e-02 -6.61922872e-01
8.78838636e-03 -9.07442033e-01 -6.51030481e-01 -9.42338780e-02
-5.82052886e-01 1.21816364e-03 5.27497053e-01 -6.63750291e-01
8.26462626e-01 -1.85508430e+00 2.78504819e-01 2.29362592e-01
2.09923804e-01 -7.01290131e-01 -3.46826613e-02 4.78569210e-01
2.92734690e-02 3.55730265e-01 -6.46353006e-01 -9.64481473e-01
4.66343045e-01 1.22428257e-02 -3.25557500e-01 6.71104491e-01
1.81516305e-01 9.93743837e-01 -9.95718420e-01 -1.44672140e-01
1.89815432e-01 5.69749892e-01 -4.69214380e-01 1.47885352e-01
-3.94370928e-02 5.84881544e-01 -2.97122806e-01 2.90203214e-01
1.18632114e+00 -5.08750752e-02 -3.79441768e-01 3.76339406e-01
7.47085735e-02 -3.14711064e-01 -1.50206399e+00 1.12091172e+00
-6.11058116e-01 1.53996516e-02 2.40967125e-01 -6.52242124e-01
6.63743377e-01 8.03546607e-02 4.86739218e-01 1.13279492e-01
-2.11461872e-01 -1.01517394e-01 -4.72678632e-01 4.37639318e-02
2.19029889e-01 -5.16439676e-01 -1.55958042e-01 5.79024196e-01
1.73376188e-01 -2.96287090e-01 -1.63808744e-02 2.31023818e-01
9.33138311e-01 1.91498175e-01 1.46779895e-01 -8.49629760e-01
1.60515010e-01 -7.25198328e-01 1.53911814e-01 1.18206131e+00
2.90097177e-01 6.42559171e-01 6.64791822e-01 -6.76443502e-02
-1.11072326e+00 -2.15552735e+00 -4.77071106e-01 9.49447513e-01
-1.25600591e-01 -6.67830482e-02 -7.07028627e-01 -1.87948570e-01
9.13721398e-02 1.01985466e+00 -8.55837882e-01 -5.99718792e-03
-2.20039070e-01 -1.41028011e+00 2.92991757e-01 4.68974262e-01
2.65606880e-01 -4.21075881e-01 1.72504023e-01 1.52955115e-01
4.11919564e-01 -7.12971330e-01 -7.21068442e-01 7.84797315e-03
-9.45451438e-01 -8.34544122e-01 -1.07516325e+00 -1.64433882e-01
5.28600574e-01 2.45876089e-02 1.22223520e+00 -9.28587556e-01
-1.36631846e-01 1.15640724e+00 3.21414769e-01 -8.36707473e-01
-3.62230361e-01 -7.30740428e-01 1.77305803e-01 5.50166786e-01
3.77486676e-01 -7.69410372e-01 -3.63423169e-01 3.19719374e-01
-8.24253500e-01 -1.58249781e-01 2.83613622e-01 5.20992875e-01
5.95417678e-01 -2.22863052e-02 1.75911874e-01 -7.56641388e-01
7.69161940e-01 -7.36240745e-01 -8.90316665e-01 6.11536242e-02
-4.30091172e-02 2.38841131e-01 9.99362990e-02 -2.86749989e-01
-1.59962535e+00 -2.24023193e-01 4.87949848e-01 -2.53063709e-01
-9.06396732e-02 4.13514286e-01 -5.10438204e-01 7.71945417e-02
3.82620811e-01 -8.59330781e-03 -1.03777163e-01 -4.80456769e-01
8.81682158e-01 1.34648323e-01 2.27938756e-01 -8.80857527e-01
7.42536902e-01 9.51695085e-01 9.12214935e-01 -1.08209372e+00
-7.64107943e-01 -3.16026390e-01 -9.37378764e-01 3.29774059e-02
1.19710708e+00 -8.25998247e-01 -6.32845044e-01 7.28366017e-01
-1.31615186e+00 -4.95935321e-01 -5.54722071e-01 7.35280037e-01
-1.33390248e+00 1.31266505e-01 -8.60088110e-01 -1.65987384e+00
1.96342066e-01 -6.95084989e-01 1.26057506e+00 -1.29555851e-01
3.19729783e-02 -1.95572424e+00 4.23217326e-01 -1.83243006e-01
2.88554788e-01 1.01398066e-01 8.88643146e-01 -3.54552865e-01
-8.95425797e-01 -1.01477832e-01 -7.44596198e-02 3.25206399e-01
-1.66604593e-01 8.31282437e-02 -6.92949295e-01 2.68866420e-01
7.06743300e-01 5.72974205e-01 7.39073336e-01 1.27813196e+00
1.07301724e+00 -4.04325902e-01 -3.59389901e-01 5.90800822e-01
1.51558864e+00 -1.73960045e-01 5.63960314e-01 -3.38258326e-01
1.01773584e+00 7.57949889e-01 5.06166220e-02 5.74977577e-01
3.87120098e-01 4.41940814e-01 2.72516906e-01 2.51155883e-01
4.37241912e-01 -6.16146922e-01 5.73911011e-01 5.38533926e-01
-3.45891505e-01 2.54254453e-02 -8.09416354e-01 1.32547408e-01
-1.95662129e+00 -1.30605769e+00 -8.50538850e-01 2.71499491e+00
4.24685627e-01 -2.93559253e-01 3.47458631e-01 -6.22120798e-01
8.89460504e-01 -1.52770996e-01 -2.54365325e-01 -1.99497923e-01
-4.70562249e-01 9.50369313e-02 8.51498544e-01 7.75920928e-01
-1.27971816e+00 3.87584627e-01 8.28274918e+00 8.38402510e-01
-1.64549321e-01 3.06129664e-01 8.29028070e-01 -1.20239165e-02
-4.14972603e-01 -2.17440099e-01 -1.07393730e+00 6.28167987e-01
1.15704155e+00 -6.61830902e-01 3.88458669e-01 6.71463311e-01
7.13835180e-01 -6.05838336e-02 -1.48637843e+00 1.04698515e+00
-1.83850855e-01 -1.04309678e+00 -1.38639912e-01 7.95032501e-01
1.16116095e+00 -1.78078800e-01 7.66669512e-01 1.80848926e-01
1.23087907e+00 -1.02903688e+00 5.41292608e-01 1.18086576e+00
2.29224145e-01 -6.72307014e-01 5.64466119e-01 4.71076608e-01
-8.22330534e-01 1.18394017e-01 -7.63355970e-01 5.41080460e-02
6.88413858e-01 1.10168099e+00 -4.18919772e-01 2.79230118e-01
3.53404939e-01 7.98348963e-01 -5.27721718e-02 1.32605231e+00
-8.75907689e-02 8.46998751e-01 -7.22759128e-01 2.70246029e-01
-5.19571677e-02 -1.32153249e+00 1.11318851e+00 1.16422009e+00
7.55185962e-01 -1.60369426e-01 -4.20439504e-02 1.52802396e+00
1.21596500e-01 2.45799944e-01 -9.97895539e-01 1.46934956e-01
-1.56276152e-01 1.02631056e+00 -4.65779006e-01 -3.49386446e-02
-4.54981834e-01 6.75427079e-01 -1.35013506e-01 6.84963346e-01
-6.46278679e-01 5.68163216e-01 7.93334842e-01 3.29058588e-01
1.26792222e-01 -3.37673336e-01 -9.02994096e-01 -1.14528000e+00
-5.75354546e-02 6.72891885e-02 2.51768172e-01 -7.22376168e-01
-2.21329141e+00 -1.86582118e-01 4.34093624e-01 -8.91163766e-01
-1.60226718e-01 -7.82178164e-01 -1.00047767e+00 1.20146012e+00
-6.70155406e-01 -1.17670798e+00 2.32375205e-01 4.82487619e-01
1.32506147e-01 1.28028616e-01 6.28650010e-01 -2.93074816e-01
-1.94453076e-01 6.51877820e-02 7.49188304e-01 -5.87389588e-01
1.89317808e-01 -2.09842825e+00 3.88931453e-01 3.95657063e-01
1.71730489e-01 6.56720519e-01 8.55134070e-01 -7.39407778e-01
-1.45058465e+00 -1.01783192e+00 1.88747227e-01 -1.17319536e+00
1.27286506e+00 -6.97340608e-01 -9.36758041e-01 9.01658118e-01
8.62687379e-02 -3.57278407e-01 8.88231516e-01 3.38084668e-01
-1.07471541e-01 3.62413526e-01 -1.30627382e+00 8.31897676e-01
1.09146571e+00 -2.47048005e-01 -3.92574787e-01 6.14421010e-01
5.70092797e-01 3.28586698e-02 -7.29079127e-01 -7.86437690e-02
1.76882043e-01 -6.30848289e-01 1.28636789e+00 -6.54631913e-01
5.25967240e-01 8.69590193e-02 -3.01944047e-01 -1.67643440e+00
-2.39731684e-01 -9.42300379e-01 -2.54267842e-01 1.11927068e+00
2.67264068e-01 -6.55965805e-01 6.33966148e-01 9.18556631e-01
2.94158489e-01 -5.00721037e-01 -9.53831792e-01 -1.01025379e+00
6.36421859e-01 -8.20665121e-01 4.46921349e-01 2.70790756e-01
-5.08539416e-02 2.52210289e-01 -5.44522285e-01 2.27010816e-01
1.56714106e+00 -6.33205414e-01 5.82410693e-01 -1.61498618e+00
-5.12098372e-01 -5.03771126e-01 -4.77702737e-01 -1.30347335e+00
1.14103228e-01 -8.05389524e-01 1.56878680e-01 -1.22393441e+00
5.99940717e-01 -4.04612213e-01 9.86039490e-02 -6.87793314e-01
-4.30862635e-01 -1.45616144e-01 2.91908145e-01 1.94304794e-01
-5.64882278e-01 9.06519532e-01 1.08136857e+00 1.40851110e-01
-2.74972469e-01 7.41354108e-01 -6.07713282e-01 1.21712244e+00
4.12744313e-01 -3.27614486e-01 -2.87686318e-01 -1.34912804e-01
4.10652369e-01 1.49747998e-01 5.50814331e-01 -5.72512209e-01
1.22689813e-01 -2.35131174e-01 1.57352388e-01 -6.36444747e-01
5.30467629e-01 -5.68287134e-01 2.22228616e-01 -1.42861024e-01
-5.03011763e-01 -5.32222614e-02 1.11924019e-02 1.16556883e+00
1.49494439e-01 -4.52949733e-01 7.38236845e-01 -1.04264647e-01
2.52487242e-01 6.25078619e-01 -5.31946123e-01 6.47593290e-02
1.03669059e+00 1.08923711e-01 -2.11183056e-02 -9.25798953e-01
-1.25169253e+00 4.07138616e-02 3.65686297e-01 -1.82060689e-01
3.59600544e-01 -1.38252449e+00 -8.51127207e-01 -1.85838833e-01
-3.45406383e-01 -7.67125096e-03 5.89085460e-01 1.07820761e+00
-1.09951884e-01 2.16847524e-01 2.36768216e-01 -6.92891657e-01
-8.04076135e-01 9.55046237e-01 1.61451921e-01 -5.80289066e-01
-3.44552457e-01 7.52053797e-01 1.14176190e+00 -5.19295931e-01
-2.00586319e-01 -3.95870674e-03 8.45438242e-02 -1.41793400e-01
9.32595611e-01 9.88734484e-01 -5.20762742e-01 -4.91614789e-01
1.72327384e-01 4.04562414e-01 2.43114874e-01 -6.53666437e-01
1.40947783e+00 -6.13354862e-01 -4.36867505e-01 1.22215497e+00
1.05134535e+00 6.82337582e-02 -1.76675141e+00 -3.31679076e-01
2.60680884e-01 -5.51231146e-01 -2.47329682e-01 -3.31913769e-01
-4.42862719e-01 1.17336142e+00 2.76452899e-01 4.83385414e-01
4.06429261e-01 2.41579309e-01 -2.17725128e-01 -1.49703696e-01
4.90901113e-01 -7.89466798e-01 1.18567318e-01 5.88593662e-01
1.17972016e+00 -7.89889872e-01 -4.47610110e-01 -1.07880676e+00
-5.57779074e-01 8.65801096e-01 -1.61178142e-01 -5.95889390e-01
1.35582209e+00 4.77052182e-01 -4.69731331e-01 -1.18119709e-01
-5.14882803e-01 -3.35578748e-04 3.67868364e-01 1.14226949e+00
1.25167340e-01 5.79138994e-01 -1.47401273e-01 7.84417570e-01
-2.29897752e-01 -1.75807446e-01 1.00748897e+00 4.44243938e-01
-1.44482315e-01 -9.53546882e-01 -8.55452955e-01 6.25810027e-01
-5.86818039e-01 -3.78920406e-01 4.08545732e-02 2.62183130e-01
-5.55940270e-01 8.40110779e-01 5.07481992e-01 4.91941810e-01
1.06321327e-01 2.80463398e-01 5.65169871e-01 -8.08067024e-01
2.21719235e-01 2.95689851e-01 -4.77019966e-01 -3.33594680e-01
-1.66197374e-01 -1.42556930e+00 -2.40431011e-01 -3.82703573e-01
1.83902085e-01 -2.70354688e-01 7.27901340e-01 9.10075128e-01
6.26947954e-02 2.48219520e-01 4.67876166e-01 -8.40210676e-01
-1.10862947e+00 -1.08231735e+00 -1.32208085e+00 1.13012604e-01
-1.90653175e-01 -8.09228182e-01 -6.10287428e-01 3.07265401e-01] | [6.941216945648193, 3.8454360961914062] |
8beb8432-0453-44ef-b01b-f2816c6fe1bb | end-to-end-modeling-via-information-tree-for | 2203.08013 | null | https://arxiv.org/abs/2203.08013v2 | https://arxiv.org/pdf/2203.08013v2.pdf | End-to-End Modeling via Information Tree for One-Shot Natural Language Spatial Video Grounding | Natural language spatial video grounding aims to detect the relevant objects in video frames with descriptive sentences as the query. In spite of the great advances, most existing methods rely on dense video frame annotations, which require a tremendous amount of human effort. To achieve effective grounding under a limited annotation budget, we investigate one-shot video grounding, and learn to ground natural language in all video frames with solely one frame labeled, in an end-to-end manner. One major challenge of end-to-end one-shot video grounding is the existence of videos frames that are either irrelevant to the language query or the labeled frames. Another challenge relates to the limited supervision, which might result in ineffective representation learning. To address these challenges, we designed an end-to-end model via Information Tree for One-Shot video grounding (IT-OS). Its key module, the information tree, can eliminate the interference of irrelevant frames based on branch search and branch cropping techniques. In addition, several self-supervised tasks are proposed based on the information tree to improve the representation learning under insufficient labeling. Experiments on the benchmark dataset demonstrate the effectiveness of our model. | ['Fei Wu', 'ShiLiang Pu', 'Peng Wang', 'Jin Wang', 'Wenming Tan', 'Wenqiao Zhang', 'Jiaxu Miao', 'Zhou Zhao', 'Shengyu Zhang', 'Haoyu Zhang', 'Tianbao Wang', 'Mengze Li'] | 2022-03-15 | null | https://aclanthology.org/2022.acl-long.596 | https://aclanthology.org/2022.acl-long.596.pdf | acl-2022-5 | ['video-grounding'] | ['computer-vision'] | [ 2.69329369e-01 1.71361580e-01 -4.63779658e-01 -2.80032665e-01
-8.27553213e-01 -1.41480714e-01 7.16756657e-02 7.97439516e-02
-3.11637908e-01 5.89965165e-01 2.40977228e-01 1.03925932e-02
1.74849451e-01 -6.31317437e-01 -8.46110821e-01 -4.50216055e-01
5.00001311e-02 -1.06555112e-01 8.69985819e-01 -9.36077982e-02
1.51335508e-01 -1.10499993e-01 -1.57572508e+00 6.47857428e-01
6.84195101e-01 1.01595569e+00 6.00911379e-01 3.57346565e-01
-2.93631881e-01 1.49091256e+00 -5.38887620e-01 -1.82464376e-01
2.71810055e-01 -6.44269407e-01 -8.08755219e-01 4.37070698e-01
6.38831139e-01 -7.25620389e-01 -6.05477810e-01 1.19338441e+00
3.22077543e-01 3.53269666e-01 -9.28260162e-02 -1.48339975e+00
-3.35843056e-01 6.21578395e-01 -5.88485897e-01 5.28609395e-01
6.40620410e-01 6.12387396e-02 1.04556334e+00 -1.12549019e+00
8.46103966e-01 1.19247103e+00 4.14268106e-01 5.22101283e-01
-7.05655217e-01 -6.01559877e-01 6.30227864e-01 5.55261672e-01
-1.64194500e+00 -5.96281230e-01 8.76906931e-01 -4.43623364e-01
6.95501268e-01 -1.76143609e-02 6.08505487e-01 7.26810992e-01
-3.35336179e-02 9.95277226e-01 6.24458075e-01 -4.17762309e-01
2.07434952e-01 -3.02958786e-01 2.18391996e-02 1.03898358e+00
4.95648719e-02 -1.78894728e-01 -9.93412077e-01 2.36982256e-02
7.78262973e-01 4.02693927e-01 -3.87459993e-01 -3.87278855e-01
-1.37772024e+00 6.11066103e-01 4.73698407e-01 2.13052034e-01
-3.42758119e-01 1.28947079e-01 4.81581539e-01 7.94037879e-02
4.50620443e-01 8.68733749e-02 -3.63211602e-01 7.86212161e-02
-1.08153260e+00 1.21958956e-01 4.08335418e-01 1.28477895e+00
1.12466252e+00 -1.67570878e-02 -5.18318415e-01 4.17388588e-01
1.52779564e-01 2.90748566e-01 3.18319798e-01 -8.86738122e-01
8.46469402e-01 7.29381382e-01 1.33402258e-01 -1.28987157e+00
-1.32811919e-01 -2.63599724e-01 -6.39312208e-01 -2.31886044e-01
2.09929347e-01 -7.22146705e-02 -8.59421074e-01 1.65824628e+00
4.58390296e-01 7.33505905e-01 -3.61596942e-02 1.21462750e+00
9.48408246e-01 7.15740144e-01 2.47031644e-01 -5.67908347e-01
1.34119475e+00 -1.34939611e+00 -8.93938541e-01 -3.64359140e-01
8.37948501e-01 -7.15485811e-01 1.20343518e+00 9.35838670e-02
-8.49282563e-01 -6.73273921e-01 -9.70775008e-01 -2.32870162e-01
-2.81311758e-02 5.74227348e-02 3.58580858e-01 1.81860521e-01
-7.45039344e-01 2.57558972e-01 -8.64169240e-01 -3.34772050e-01
4.79664177e-01 6.33992851e-02 -4.05151069e-01 -4.23356861e-01
-1.14526451e+00 2.99149275e-01 6.13817573e-01 8.80981088e-02
-9.62940633e-01 -4.47252929e-01 -9.69373941e-01 7.68647417e-02
1.16620207e+00 -5.68177164e-01 1.10903573e+00 -1.16133273e+00
-1.02402830e+00 6.01365030e-01 -3.99129331e-01 -5.66441894e-01
4.22092885e-01 -4.80703294e-01 -2.45485738e-01 6.37383401e-01
5.33610880e-01 6.96846247e-01 8.18677485e-01 -1.02628160e+00
-9.52701628e-01 -3.01299572e-01 3.72676760e-01 3.31876010e-01
-3.55768442e-01 1.56099975e-01 -9.26136851e-01 -8.62012029e-01
2.43852869e-01 -8.31720114e-01 -2.36697629e-01 -1.09956563e-01
-2.91040272e-01 -1.97308302e-01 1.03911376e+00 -6.59056187e-01
1.49925125e+00 -2.16189408e+00 9.51471254e-02 -2.03117341e-01
2.03642517e-01 5.18605560e-02 -1.58815518e-01 1.51926205e-01
9.41685811e-02 -1.56449437e-01 -4.41089496e-02 -1.22049496e-01
-4.31175679e-01 2.67916799e-01 -4.57528323e-01 3.39512318e-01
2.31658041e-01 6.33372724e-01 -1.27297735e+00 -9.67001677e-01
2.48757869e-01 9.27846581e-02 -7.33859301e-01 5.48744500e-01
-5.09901106e-01 4.71437603e-01 -6.05517924e-01 7.52481759e-01
3.84645045e-01 -3.35882336e-01 -8.01239349e-03 -4.02431607e-01
-3.22742574e-02 2.78811045e-02 -1.30285692e+00 2.24555588e+00
-6.96275458e-02 4.55725640e-01 -5.41789345e-02 -1.04048324e+00
6.39232337e-01 3.01997751e-01 7.82439411e-01 -6.81771994e-01
-8.51982981e-02 4.43465672e-02 -3.29708338e-01 -7.91414440e-01
4.29400414e-01 1.45528987e-01 7.73289474e-04 2.43728116e-01
1.05124570e-01 4.15877372e-01 4.12032098e-01 4.88426626e-01
1.11899316e+00 4.74241525e-01 3.58476996e-01 -2.37341430e-02
6.13012016e-01 1.72547489e-01 9.76804018e-01 5.52322149e-01
-3.40968490e-01 7.40939677e-01 3.70752066e-01 -6.53085291e-01
-6.49078488e-01 -4.82312322e-01 5.23907483e-01 1.58939993e+00
7.27966309e-01 -9.20578003e-01 -9.27193642e-01 -7.37654746e-01
-6.33535087e-01 3.93037170e-01 -2.76115924e-01 -2.12063327e-01
-7.78030276e-01 -3.15514922e-01 2.47074902e-01 4.45967257e-01
6.66511416e-01 -9.73862350e-01 -9.82432246e-01 2.48665586e-01
-6.75771594e-01 -1.55791628e+00 -7.00837612e-01 -1.11419529e-01
-6.89997554e-01 -1.29577327e+00 -6.81251645e-01 -8.02364945e-01
7.46599674e-01 9.09862697e-01 1.06173325e+00 4.84278351e-01
-2.87233926e-02 2.33460441e-01 -6.89262986e-01 -7.22754225e-02
-5.73308915e-02 6.85556382e-02 -1.03954330e-01 1.89060658e-01
4.47942585e-01 -2.48028561e-02 -7.22933769e-01 6.29031658e-01
-9.76853311e-01 3.97967637e-01 3.63987029e-01 6.28509104e-01
9.29272413e-01 2.11100847e-01 3.47443074e-01 -6.82060540e-01
-3.86759304e-02 -4.20906842e-01 -4.57990676e-01 3.83126020e-01
-1.67162776e-01 -1.35849401e-01 5.48314214e-01 -2.00461596e-01
-8.10060799e-01 3.27524036e-01 2.52211928e-01 -9.67697799e-01
3.47174481e-02 4.95064527e-01 -3.20247918e-01 1.65944606e-01
4.94247079e-01 1.70521155e-01 -2.88824618e-01 -2.73634851e-01
2.62061149e-01 3.82331729e-01 6.50609791e-01 -5.63397348e-01
5.48717380e-01 4.72815812e-01 -1.28147289e-01 -8.37816000e-01
-1.36070418e+00 -7.48107135e-01 -6.07890010e-01 -4.24048483e-01
1.04995263e+00 -1.39914596e+00 -1.48902342e-01 3.58321704e-02
-1.21168578e+00 -1.71921372e-01 -1.54762074e-01 4.58677769e-01
-5.52631676e-01 6.09834552e-01 -2.55929530e-01 -6.74327135e-01
-2.39218608e-01 -1.14560306e+00 1.25982940e+00 2.19028637e-01
-7.18559474e-02 -3.80696177e-01 -3.99399459e-01 4.24404860e-01
-9.22833309e-02 1.46306738e-01 5.47655225e-01 -4.08688426e-01
-1.06584716e+00 -2.39298716e-02 -2.08671898e-01 1.13073818e-01
1.03157751e-01 -2.01040685e-01 -6.38737619e-01 -4.33465362e-01
3.87157779e-03 -2.95150101e-01 8.51019859e-01 2.60460556e-01
1.17781436e+00 -2.84889102e-01 -2.49234408e-01 5.82193017e-01
1.29720604e+00 2.37016737e-01 4.70593721e-01 2.51613826e-01
1.06728935e+00 5.44783056e-01 1.26246440e+00 4.38406408e-01
3.29162389e-01 7.46308386e-01 5.23766041e-01 -1.55712306e-01
-2.46958137e-01 -5.87315559e-01 3.96525532e-01 6.21040344e-01
8.16657394e-03 -2.60689497e-01 -7.78956413e-01 4.72412616e-01
-2.37767267e+00 -1.21872759e+00 1.22545451e-01 2.14033461e+00
4.35922116e-01 1.04442820e-01 2.64734268e-01 1.76028118e-01
9.87950861e-01 3.15164596e-01 -5.52590072e-01 4.99815136e-01
-7.91255534e-02 -4.24496114e-01 3.33060771e-01 2.51040518e-01
-1.24210393e+00 1.25179338e+00 5.52726030e+00 7.19051838e-01
-9.87076342e-01 2.49333143e-01 6.90084517e-01 -1.96315050e-01
2.80490126e-02 2.27794707e-01 -7.96208799e-01 5.89308798e-01
4.50740695e-01 -3.52867022e-02 2.50938267e-01 1.06119180e+00
4.06352699e-01 -1.72710225e-01 -1.07368279e+00 1.25450110e+00
2.72656232e-01 -1.41581547e+00 1.52202502e-01 -3.86121958e-01
7.49030054e-01 -1.18788898e-01 -4.67529684e-01 2.80782342e-01
-1.86406538e-01 -5.14348447e-01 1.04748607e+00 2.41827011e-01
7.40052044e-01 -6.07550144e-01 7.25490093e-01 3.77981544e-01
-1.52339256e+00 -1.33056223e-01 -5.43577909e-01 -1.89027816e-01
2.59879649e-01 3.65970105e-01 -5.57518244e-01 6.11753941e-01
7.18080461e-01 9.25758541e-01 -5.14530659e-01 9.19315159e-01
-2.72701025e-01 5.27264655e-01 -2.94704974e-01 2.75259018e-01
3.66879344e-01 -2.13030875e-02 4.94130582e-01 1.11838412e+00
4.15860534e-01 2.96302915e-01 9.14984226e-01 4.50535297e-01
-2.81213336e-02 1.90708414e-01 -5.55213749e-01 7.50553459e-02
5.66942394e-01 9.29301918e-01 -1.09023249e+00 -5.28467715e-01
-8.17407131e-01 9.40599978e-01 1.39836624e-01 4.73182350e-01
-9.49800551e-01 -2.39007309e-01 2.74764597e-01 4.89315718e-01
2.81333208e-01 -1.72607467e-01 1.31525591e-01 -1.42358255e+00
2.94919610e-01 -8.85108411e-01 7.79453874e-01 -8.16564381e-01
-8.54816139e-01 5.05836070e-01 1.83482885e-01 -1.51388502e+00
-1.92648157e-01 -1.71599314e-01 -3.02423507e-01 2.10215971e-01
-1.39974940e+00 -1.06168544e+00 -7.21556187e-01 9.28889275e-01
1.02118564e+00 -2.54427284e-01 3.97689611e-01 5.28843462e-01
-7.18147397e-01 3.18136126e-01 -5.21506429e-01 2.69254595e-01
6.47749484e-01 -8.09975863e-01 4.85887863e-02 1.33233142e+00
4.28141892e-01 5.13870537e-01 5.88426292e-01 -7.07063735e-01
-1.44824553e+00 -1.37010300e+00 6.85773671e-01 -9.59174186e-02
3.96342486e-01 -1.81550637e-01 -9.01245356e-01 7.01326132e-01
6.20490126e-03 4.87917185e-01 5.75771570e-01 -2.43833050e-01
-1.72029629e-01 -8.85593444e-02 -6.64968252e-01 6.11671388e-01
1.42653549e+00 -5.34581602e-01 -6.65019035e-01 5.33566058e-01
1.06657398e+00 -5.28830409e-01 -3.38340968e-01 3.04805100e-01
2.05355674e-01 -9.29873407e-01 9.24378753e-01 -6.38642669e-01
5.27390778e-01 -8.48839045e-01 -4.04155940e-01 -7.42742121e-01
-2.23704398e-01 -9.02959645e-01 -3.03702205e-01 1.27245033e+00
-1.59122825e-01 1.76985279e-01 9.12420094e-01 5.41455626e-01
-1.35574788e-01 -6.67377532e-01 -9.99054611e-01 -7.43663490e-01
-6.75392449e-01 -3.20679545e-01 5.42504728e-01 7.62536108e-01
-6.41098320e-02 6.21955931e-01 -7.14918256e-01 1.98699564e-01
4.26279694e-01 3.19177091e-01 1.01215708e+00 -9.57122684e-01
-1.05329834e-01 9.99192074e-02 -5.59188068e-01 -1.30771959e+00
2.10455716e-01 -5.61829925e-01 3.30363125e-01 -1.62981820e+00
3.26448053e-01 -8.28628466e-02 -4.88396943e-01 5.67880154e-01
-4.73632872e-01 1.72653183e-01 3.29552442e-01 4.74822104e-01
-1.33076119e+00 4.96168822e-01 1.10529125e+00 -1.37415037e-01
-2.39466995e-01 -3.08917701e-01 -5.53775012e-01 9.81160939e-01
3.99080336e-01 -6.22422457e-01 -7.55171180e-01 -7.60210574e-01
1.06709100e-01 2.88536727e-01 3.29501867e-01 -1.08534718e+00
4.31987375e-01 -3.67106676e-01 1.24546558e-01 -8.87299538e-01
2.01588064e-01 -9.46461320e-01 -4.25838679e-02 2.41056770e-01
-2.74874717e-01 1.54154658e-01 -1.21520817e-01 8.46468031e-01
-5.23919046e-01 -1.65709928e-01 6.66320384e-01 -2.45793313e-01
-1.34426260e+00 5.61472178e-01 -7.56245181e-02 2.19247565e-01
1.33417487e+00 -2.35445157e-01 -1.58123329e-01 -2.35255584e-01
-4.72412139e-01 3.16607386e-01 5.23832381e-01 4.36380178e-01
7.14203596e-01 -1.34890044e+00 -5.48620582e-01 1.59522649e-02
3.12621385e-01 4.16603953e-01 3.66521925e-01 5.55169165e-01
-6.03678584e-01 2.68196821e-01 -1.20767750e-01 -7.20513821e-01
-1.28399396e+00 8.16880167e-01 6.49373978e-02 -1.11702852e-01
-8.41636062e-01 7.25368023e-01 4.23120439e-01 4.76931036e-01
4.19403791e-01 -3.63247693e-01 -1.72877073e-01 1.29286692e-01
8.73012245e-01 2.46055990e-01 -7.04009160e-02 -9.02461946e-01
-4.05370712e-01 5.68237185e-01 -1.07899785e-01 1.18225992e-01
1.06929314e+00 -3.86865675e-01 2.02393569e-02 3.22752535e-01
9.77110922e-01 -2.76974261e-01 -1.55761981e+00 -4.95120555e-01
7.00253844e-02 -8.66316259e-01 1.79300636e-01 -1.02057666e-01
-1.18110502e+00 7.01746166e-01 3.68015796e-01 -1.94228441e-02
1.20438445e+00 -1.80828542e-01 1.10628974e+00 6.89994156e-01
6.73673332e-01 -1.05581236e+00 4.26128328e-01 2.75190204e-01
6.25560701e-01 -1.30572665e+00 1.54893681e-01 -6.65312052e-01
-6.66213930e-01 9.69808400e-01 8.84610295e-01 7.17211515e-02
3.18925142e-01 8.05816427e-02 -5.74374460e-02 -9.76028591e-02
-6.06807649e-01 -3.96005094e-01 9.97342914e-02 4.27857995e-01
2.08500355e-01 -3.74796003e-01 -1.79964453e-01 6.39130175e-01
2.89717168e-01 3.78022045e-01 3.74131322e-01 1.12158906e+00
-6.13514543e-01 -6.71084881e-01 -3.37354571e-01 1.19024463e-01
-5.65521061e-01 8.77500772e-02 -1.64611101e-01 5.32657325e-01
3.04454029e-01 1.17340171e+00 1.33131929e-02 -2.71337360e-01
3.47470045e-01 -6.20322153e-02 2.19156101e-01 -8.66824925e-01
-1.79934487e-01 3.22251886e-01 -1.11587703e-01 -8.02824914e-01
-8.21498275e-01 -3.82290393e-01 -1.39031720e+00 1.23552874e-01
-3.77568960e-01 1.69549733e-01 -9.29044187e-03 9.53136742e-01
3.94615382e-01 4.76481348e-01 4.80163574e-01 -8.56529117e-01
-2.45449111e-01 -7.03895509e-01 -3.13072205e-01 7.30946422e-01
2.91484058e-01 -7.73304224e-01 2.77740695e-02 5.93212605e-01] | [9.71955394744873, 0.6336004734039307] |
06e864b2-3524-4ae8-bb5a-f772b0c549b5 | joint-entity-and-relation-extraction-with-set | 2011.01675 | null | https://arxiv.org/abs/2011.01675v2 | https://arxiv.org/pdf/2011.01675v2.pdf | Joint Entity and Relation Extraction with Set Prediction Networks | The joint entity and relation extraction task aims to extract all relational triples from a sentence. In essence, the relational triples contained in a sentence are unordered. However, previous seq2seq based models require to convert the set of triples into a sequence in the training phase. To break this bottleneck, we treat joint entity and relation extraction as a direct set prediction problem, so that the extraction model can get rid of the burden of predicting the order of multiple triples. To solve this set prediction problem, we propose networks featured by transformers with non-autoregressive parallel decoding. Unlike autoregressive approaches that generate triples one by one in a certain order, the proposed networks directly output the final set of triples in one shot. Furthermore, we also design a set-based loss that forces unique predictions via bipartite matching. Compared with cross-entropy loss that highly penalizes small shifts in triple order, the proposed bipartite matching loss is invariant to any permutation of predictions; thus, it can provide the proposed networks with a more accurate training signal by ignoring triple order and focusing on relation types and entities. Experiments on two benchmark datasets show that our proposed model significantly outperforms current state-of-the-art methods. Training code and trained models will be available at http://github.com/DianboWork/SPN4RE. | ['Shengping Liu', 'Xiangrong Zeng', 'Jun Zhao', 'Kang Liu', 'Yubo Chen', 'Dianbo Sui'] | 2020-11-03 | null | null | null | null | ['joint-entity-and-relation-extraction'] | ['natural-language-processing'] | [ 2.74656981e-01 5.05603611e-01 -2.78808147e-01 -5.94310880e-01
-8.50141466e-01 -4.86180753e-01 3.58796597e-01 2.53455132e-01
-3.19340616e-01 9.44818556e-01 3.28206390e-01 -2.48641342e-01
1.40565291e-01 -1.17927182e+00 -1.24474585e+00 -4.18318808e-01
9.31455866e-02 5.88987231e-01 3.95305678e-02 -3.84221673e-01
-1.23399757e-01 3.59987691e-02 -1.24559772e+00 5.74447215e-01
1.13109481e+00 9.63028133e-01 7.54706636e-02 3.07020843e-01
-2.00352177e-01 8.69424820e-01 -2.73305953e-01 -1.00233412e+00
3.11563551e-01 -5.49073160e-01 -7.64751136e-01 -3.31779599e-01
3.95476639e-01 -2.23217711e-01 -5.89876652e-01 1.04925752e+00
4.24649715e-01 -2.15209909e-02 5.07805288e-01 -1.02989137e+00
-6.70897961e-01 1.28741205e+00 -4.05208945e-01 2.93256715e-02
4.00784910e-01 -1.56296313e-01 1.64357281e+00 -1.23909605e+00
6.02327228e-01 1.04178047e+00 6.14896655e-01 4.31155205e-01
-1.15092885e+00 -8.53051543e-01 2.29535140e-02 3.79304647e-01
-1.44754243e+00 -6.14696503e-01 7.17122257e-01 -2.16116697e-01
1.15434265e+00 4.65980738e-01 5.78454792e-01 8.79944861e-01
7.73081779e-02 8.90473664e-01 4.53055799e-01 -1.93085268e-01
-6.16720617e-02 -7.20603094e-02 4.04030740e-01 7.35413253e-01
3.85001749e-01 -1.12048961e-01 -7.78623581e-01 8.66646916e-02
2.24654958e-01 -7.68367499e-02 -3.53915185e-01 -3.83308709e-01
-1.05868077e+00 4.95412380e-01 5.76608539e-01 1.02854259e-01
-4.01363641e-01 5.71309626e-02 5.18875778e-01 1.68069273e-01
3.64028394e-01 1.37835041e-01 -6.60934687e-01 -3.77485454e-02
-6.67254329e-01 1.74971119e-01 8.61379445e-01 1.18576145e+00
9.75144744e-01 -2.27357939e-01 -5.38333237e-01 6.35162055e-01
1.84324130e-01 4.40043926e-01 1.74393058e-01 -4.92931724e-01
1.05873919e+00 8.06376874e-01 -1.30784526e-01 -1.06679463e+00
-3.06615680e-01 -3.87066364e-01 -1.18217409e+00 -6.51157260e-01
1.95854306e-01 -3.27501535e-01 -8.11765254e-01 1.73982394e+00
4.31049824e-01 2.15658560e-01 2.22650245e-01 8.23695123e-01
1.10351861e+00 7.79186249e-01 -1.32268250e-01 -3.33203107e-01
1.45669770e+00 -8.67978871e-01 -8.04527581e-01 -2.79472232e-01
8.51942182e-01 -3.89332831e-01 6.01907074e-01 7.15718046e-02
-1.14290965e+00 -3.13938320e-01 -9.43966627e-01 -4.39183086e-01
-1.56148717e-01 2.41453916e-01 6.62030637e-01 4.34897453e-01
-7.11328030e-01 7.26865768e-01 -7.68634379e-01 1.30167287e-02
2.93251604e-01 5.30896664e-01 -4.47209567e-01 1.01811744e-01
-1.60563409e+00 7.89335668e-01 7.03361094e-01 7.86089838e-01
-1.65369764e-01 -9.11003828e-01 -1.16584754e+00 5.21627188e-01
6.49895608e-01 -8.19589615e-01 1.07321405e+00 -5.05091608e-01
-1.41022646e+00 4.63366151e-01 -6.35639906e-01 -6.68217599e-01
3.01011056e-01 -4.02443022e-01 -2.72711366e-01 -9.45928246e-02
1.01303644e-02 4.45041716e-01 2.82988816e-01 -9.15982187e-01
-4.98216599e-01 -1.49930388e-01 8.01674724e-02 2.97112852e-01
-3.38857681e-01 -1.02414668e-01 -5.10922670e-01 -2.69325376e-01
1.41714826e-01 -8.39570582e-01 -9.96013507e-02 -6.19795799e-01
-9.60500360e-01 -3.31863075e-01 2.73147494e-01 -6.78959250e-01
1.35326755e+00 -1.87212169e+00 2.93875545e-01 1.96474507e-01
3.16427648e-01 3.54085535e-01 -1.65128559e-02 5.68754733e-01
-2.49463752e-01 3.79881859e-01 -2.48717070e-01 -3.93452346e-01
3.87329385e-02 7.49699445e-03 -2.53926724e-01 1.38522506e-01
5.08456528e-01 1.12418473e+00 -9.55451846e-01 -5.81566334e-01
-1.83736503e-01 4.31767255e-01 -8.20176125e-01 2.35609002e-02
-2.32697636e-01 2.09497422e-01 -4.09287751e-01 2.41012827e-01
7.78310359e-01 -4.43135530e-01 7.59493828e-01 -6.80114686e-01
1.51691794e-01 8.32491875e-01 -1.03398073e+00 1.40838099e+00
-3.46316755e-01 2.24766225e-01 -4.82919455e-01 -1.06882775e+00
9.79722261e-01 3.32925171e-01 3.95568520e-01 -7.08445251e-01
-1.47577217e-02 1.64639547e-01 7.50067458e-02 -4.01405364e-01
7.47041583e-01 -3.24406959e-02 -2.69255519e-01 -3.36155519e-02
2.68489361e-01 2.94808954e-01 5.69583476e-01 3.77887517e-01
1.11830795e+00 2.07770377e-01 3.33316296e-01 2.40321100e-01
4.61792529e-01 -2.87700653e-01 1.27203119e+00 5.43963134e-01
3.52678478e-01 5.05293965e-01 8.87142956e-01 -2.73654014e-01
-1.03536701e+00 -9.61117208e-01 1.57631278e-01 5.77964783e-01
5.66331148e-02 -8.20576966e-01 -3.53218377e-01 -9.12145197e-01
-1.36824310e-01 7.13389218e-01 -4.14960325e-01 -2.95428842e-01
-8.11952412e-01 -7.16634154e-01 4.45595950e-01 4.46341187e-01
4.11257327e-01 -9.29840684e-01 -1.63261953e-03 1.68470263e-01
-6.71864688e-01 -1.25760496e+00 -7.58962631e-01 2.30160996e-01
-5.88575184e-01 -9.97160137e-01 -1.98971793e-01 -8.41015220e-01
6.48982704e-01 -1.04421504e-01 1.29671431e+00 -1.44686112e-02
2.33610317e-01 -3.43472779e-01 -4.46017832e-01 -4.46340702e-02
-1.86952934e-01 4.05165315e-01 -6.02414310e-02 1.24929659e-01
5.66914618e-01 -6.22405946e-01 -4.19028640e-01 -1.80805817e-01
-6.18169606e-01 3.58818680e-01 6.89516068e-01 1.03900504e+00
6.36658490e-01 -4.54979353e-02 4.81267869e-01 -1.35563684e+00
2.97119081e-01 -5.64988852e-01 -5.67363441e-01 4.35442716e-01
-3.66829336e-01 3.31880987e-01 8.70139658e-01 6.95526898e-02
-1.17803049e+00 2.94199824e-01 2.65972205e-02 -2.89027989e-01
2.98837453e-01 8.80676627e-01 -4.13281143e-01 5.42501330e-01
2.81097710e-01 2.97814965e-01 -3.22333604e-01 -3.27997297e-01
3.39295566e-01 4.61894602e-01 2.56429106e-01 -2.97001153e-01
8.68539870e-01 -4.33326699e-02 1.53949499e-01 -5.18858612e-01
-1.10449100e+00 -3.57536882e-01 -7.09555447e-01 8.91189799e-02
6.91743255e-01 -1.01204872e+00 -1.10127270e+00 2.49907196e-01
-1.49370754e+00 -6.15786808e-03 -2.48090774e-01 3.87088656e-01
-2.60504782e-01 4.18846279e-01 -6.92571521e-01 -7.93056965e-01
-5.92298865e-01 -8.55108917e-01 8.64056230e-01 1.65293336e-01
-1.61834240e-01 -6.89775646e-01 -1.88816879e-02 3.74369442e-01
-1.93054795e-01 -9.28433612e-02 9.09559488e-01 -1.01775289e+00
-9.71608341e-01 -8.95103142e-02 -1.59648404e-01 3.54689598e-01
1.59446653e-02 -7.19763711e-02 -5.46413839e-01 1.30595133e-01
-3.12794179e-01 -2.88516998e-01 1.08943200e+00 -1.18694464e-02
8.81474733e-01 -7.00892687e-01 -4.26984251e-01 4.98960406e-01
1.31568718e+00 -1.98814962e-02 8.06164443e-01 -7.02100173e-02
1.06271851e+00 5.23297250e-01 7.03563154e-01 4.32746261e-01
8.06817532e-01 6.44639730e-01 3.45419198e-01 1.53868586e-01
1.16331771e-01 -7.05339313e-01 4.50047553e-01 1.12482905e+00
1.41246647e-01 -5.44102669e-01 -9.04165924e-01 5.00829518e-01
-2.00608754e+00 -1.13409734e+00 -4.37830985e-01 2.20325971e+00
1.19141912e+00 2.08337218e-01 -1.04968049e-01 2.66777296e-02
7.58209765e-01 9.49152410e-02 -2.57265568e-01 -2.46922031e-01
-9.20646563e-02 2.20954359e-01 6.18493736e-01 5.70066094e-01
-9.82507288e-01 1.03641987e+00 5.07778645e+00 8.06838393e-01
-8.62129748e-01 -1.46497697e-01 5.93625009e-01 -1.48008719e-01
-6.82308495e-01 2.57701725e-01 -1.24695981e+00 5.83758414e-01
9.65787768e-01 -2.41672844e-01 2.53580987e-01 3.79766136e-01
2.20756307e-01 1.23333491e-01 -1.33064723e+00 6.01243913e-01
-8.30342919e-02 -1.43346727e+00 1.13391943e-01 -1.65231496e-01
4.25199986e-01 -7.89479464e-02 -1.65632620e-01 5.90479076e-01
4.21282589e-01 -9.61204767e-01 5.57539165e-01 7.56535411e-01
5.99367619e-01 -7.84355819e-01 9.12084281e-01 2.78617322e-01
-1.28959107e+00 -2.00929325e-02 -3.35136920e-01 -9.95090976e-02
3.51697773e-01 1.13238358e+00 -1.01288617e+00 1.04325926e+00
5.48635900e-01 8.73926580e-01 -2.59729415e-01 8.62125099e-01
-3.78533453e-01 5.92539489e-01 -4.40155625e-01 -2.66805410e-01
-1.30194589e-01 -4.33334529e-01 5.03803432e-01 1.13176298e+00
3.54019195e-01 1.60731062e-01 8.06578547e-02 7.62029052e-01
-6.10312462e-01 1.37043446e-01 -6.27058864e-01 2.48201587e-03
8.25637519e-01 1.21733403e+00 -2.58343488e-01 -2.70265073e-01
-4.68845248e-01 8.36912990e-01 8.67574692e-01 2.55154729e-01
-8.94779325e-01 -5.84294796e-01 3.83512437e-01 -1.27726004e-01
7.40677476e-01 1.04203168e-02 -3.38943213e-01 -1.47343314e+00
4.72751528e-01 -7.95525789e-01 2.40503639e-01 -4.82017785e-01
-1.12893796e+00 5.70015371e-01 -2.42890537e-01 -1.14992630e+00
-2.61509150e-01 -2.19814062e-01 -3.38883251e-01 7.42775619e-01
-1.46133280e+00 -1.08798075e+00 8.51357505e-02 5.66591024e-02
1.46780610e-01 1.94029972e-01 5.78279138e-01 5.22126138e-01
-7.96424568e-01 9.17174578e-01 -9.94040370e-02 6.38152301e-01
6.05144560e-01 -1.29357755e+00 5.27443349e-01 9.22089458e-01
1.72801882e-01 7.77060449e-01 6.61946356e-01 -8.07314396e-01
-1.45619583e+00 -1.18951142e+00 1.64497161e+00 -2.51291096e-01
6.34509683e-01 -7.23414063e-01 -9.22814667e-01 1.03699827e+00
2.00681120e-01 5.52904829e-02 5.79564452e-01 4.87792104e-01
-3.55389118e-01 -4.55276489e-01 -7.74282575e-01 6.25461161e-01
1.30325723e+00 -4.10697222e-01 -3.87143731e-01 4.27559912e-01
1.02107370e+00 -6.28472745e-01 -1.05759060e+00 5.94667912e-01
4.11399156e-01 -8.46402526e-01 7.37921476e-01 -6.25451326e-01
8.64041030e-01 -1.50541857e-01 8.54473785e-02 -1.30881155e+00
-3.06524605e-01 -6.85082138e-01 -4.74552274e-01 1.58756816e+00
1.02752304e+00 -5.80768704e-01 9.39200580e-01 5.68411350e-01
-2.62730092e-01 -1.00287020e+00 -8.21620464e-01 -6.12374663e-01
2.43780613e-02 -6.88018128e-02 7.27012992e-01 6.63022101e-01
3.42804581e-01 9.34777260e-01 -6.05064511e-01 2.61244774e-01
4.92236555e-01 2.59151459e-01 6.12341642e-01 -9.43735242e-01
-3.45552474e-01 -3.99768678e-03 -2.22189173e-01 -1.25409925e+00
3.93272549e-01 -1.23849928e+00 2.41095901e-01 -1.49467039e+00
4.08733159e-01 -5.62210858e-01 -1.87241256e-01 7.01997936e-01
-4.95116502e-01 -4.30882350e-02 2.97033489e-01 -7.46673867e-02
-7.09805131e-01 9.70965981e-01 1.04413199e+00 -1.14422515e-01
-1.05674498e-01 -1.10508904e-01 -7.35526502e-01 3.89646351e-01
7.62170851e-01 -6.24488652e-01 -3.51775676e-01 -4.78604227e-01
8.23176026e-01 3.29510361e-01 1.91890508e-01 -6.84973419e-01
4.85973775e-01 4.50654365e-02 1.32434458e-01 -7.84156978e-01
4.79838431e-01 -7.85590172e-01 1.75739303e-01 1.93433315e-01
-3.78375381e-01 -1.61289722e-01 -7.07161948e-02 5.62879622e-01
-2.94760108e-01 -1.82559535e-01 2.54844218e-01 8.57529491e-02
-3.98888409e-01 4.75527853e-01 1.42880544e-01 2.13156343e-01
8.86569381e-01 1.09942950e-01 -3.88044327e-01 -2.62480557e-01
-6.88549757e-01 5.57088912e-01 6.35133386e-02 3.39212865e-01
5.88119686e-01 -1.31296146e+00 -8.45447838e-01 -7.33858615e-04
-1.13710716e-01 3.88501436e-01 3.44136298e-01 9.22137320e-01
-1.62959605e-01 5.63613474e-01 2.72325963e-01 -3.49814892e-01
-1.25157726e+00 5.17680228e-01 2.78102249e-01 -8.06741118e-01
-3.50089729e-01 8.52606773e-01 3.98468450e-02 -7.96455622e-01
-9.22391191e-02 -2.34333843e-01 -2.96885371e-01 1.23432681e-01
2.02181533e-01 4.58910130e-02 2.95056641e-01 -5.36885560e-01
-4.24319804e-01 1.14227273e-01 -4.07684386e-01 2.24579602e-01
1.40738142e+00 4.30126972e-02 -3.18069458e-01 4.89260256e-01
1.27811360e+00 1.48635715e-01 -8.57719362e-01 -3.88460189e-01
7.54827186e-02 -1.43078595e-01 -3.05588335e-01 -5.01353264e-01
-1.12654722e+00 5.79075694e-01 -1.15376651e-01 9.19958204e-02
1.00786424e+00 -4.72420268e-02 9.87186372e-01 5.36167741e-01
1.87285900e-01 -7.20089495e-01 -2.81190664e-01 9.46172535e-01
6.04930103e-01 -1.12505305e+00 -2.01919656e-02 -9.72111046e-01
-7.57686138e-01 8.01841378e-01 7.03169167e-01 -1.99598148e-02
6.37731373e-01 2.96582818e-01 -4.42784488e-01 2.40281560e-02
-1.08495295e+00 -1.48515999e-01 3.17397743e-01 2.34497949e-01
7.36729622e-01 6.68115690e-02 -5.19238770e-01 8.40082586e-01
-4.53027189e-01 1.50701240e-01 5.44061840e-01 5.34761429e-01
-1.36328399e-01 -1.30406833e+00 1.36106521e-01 8.26720953e-01
-5.34172356e-01 -6.77286148e-01 -1.24993332e-01 4.08977509e-01
-7.35058784e-02 7.91308582e-01 1.54488519e-01 -6.50213063e-01
4.88783777e-01 1.44478843e-01 3.06507260e-01 -7.63411105e-01
-6.61478043e-01 -4.10463572e-01 6.13244355e-01 -4.18327689e-01
-2.16729239e-01 -7.14910388e-01 -1.41404474e+00 -4.44571555e-01
-5.49481273e-01 3.54781479e-01 1.33819684e-01 9.34972584e-01
7.62055099e-01 6.02225482e-01 6.22340858e-01 -3.99073392e-01
-6.27818406e-01 -1.03422093e+00 -3.88174415e-01 3.32522839e-01
2.86119223e-01 -4.41781014e-01 -8.27582553e-02 -7.92212635e-02] | [9.42734146118164, 8.667145729064941] |
7e7df375-1fee-4e55-8a2e-6947725a7d0a | vifs-an-end-to-end-variational-inference-for | 2306.05004 | null | https://arxiv.org/abs/2306.05004v1 | https://arxiv.org/pdf/2306.05004v1.pdf | VIFS: An End-to-End Variational Inference for Foley Sound Synthesis | The goal of DCASE 2023 Challenge Task 7 is to generate various sound clips for Foley sound synthesis (FSS) by "category-to-sound" approach. "Category" is expressed by a single index while corresponding "sound" covers diverse and different sound examples. To generate diverse sounds for a given category, we adopt VITS, a text-to-speech (TTS) model with variational inference. In addition, we apply various techniques from speech synthesis including PhaseAug and Avocodo. Different from TTS models which generate short pronunciation from phonemes and speaker identity, the category-to-sound problem requires generating diverse sounds just from a category index. To compensate for the difference while maintaining consistency within each audio clip, we heavily modified the prior encoder to enhance consistency with posterior latent variables. This introduced additional Gaussian on the prior encoder which promotes variance within the category. With these modifications, we propose VIFS, variational inference for end-to-end Foley sound synthesis, which generates diverse high-quality sounds. | ['Yong-Hwa Park', 'Hyeonuk Nam', 'Junhyeok Lee'] | 2023-06-08 | null | null | null | null | ['speech-synthesis'] | ['speech'] | [ 9.26905647e-02 4.06896211e-02 2.15128869e-01 -2.83577681e-01
-1.33763874e+00 -7.36400545e-01 7.19555974e-01 -6.10442102e-01
2.00056791e-01 6.86359823e-01 7.76233435e-01 -1.31783551e-02
1.93352640e-01 -4.59584564e-01 -7.99224436e-01 -6.69409871e-01
5.87453544e-01 1.52802661e-01 1.22096278e-01 -8.25011432e-02
5.22003882e-02 -7.44645372e-02 -1.69736505e+00 5.40540397e-01
6.02784455e-01 9.08398986e-01 5.26671529e-01 1.18070960e+00
-3.58726442e-01 9.33742464e-01 -1.02451754e+00 -3.06026518e-01
1.34534121e-01 -9.94179606e-01 -3.44038248e-01 -1.40303552e-01
5.33425629e-01 -1.90081134e-01 -1.45970613e-01 1.00028014e+00
7.79288590e-01 4.58858818e-01 1.07084715e+00 -1.34732521e+00
-6.83394670e-01 1.04671097e+00 -1.12765513e-01 -1.64717615e-01
4.30153549e-01 2.32058078e-01 1.27912927e+00 -1.14126968e+00
4.83534038e-01 1.45857322e+00 6.16577744e-01 8.99683475e-01
-1.28662944e+00 -8.20767462e-01 -1.26570448e-01 3.60546261e-02
-1.37286592e+00 -1.01780891e+00 8.65069032e-01 -5.56012630e-01
6.93964660e-01 6.13328338e-01 4.68200892e-01 1.54318810e+00
9.57124680e-03 7.60631800e-01 7.68394470e-01 -5.02794504e-01
3.99703473e-01 3.83119844e-03 -3.41634184e-01 7.61486515e-02
-6.01598501e-01 2.00806782e-01 -8.54274809e-01 6.30949959e-02
5.53169250e-01 -5.62771976e-01 -1.59480035e-01 1.68308064e-01
-1.26230145e+00 6.38465405e-01 -4.77806367e-02 5.01904488e-02
-4.00428981e-01 6.36723638e-01 4.89215225e-01 1.59961551e-01
2.91746110e-01 3.07490319e-01 -2.50217795e-01 -4.35753882e-01
-1.43883502e+00 5.89397907e-01 7.57279038e-01 1.11600196e+00
3.56041282e-01 8.14681590e-01 -9.83280241e-01 1.22032678e+00
6.19189441e-01 8.73248398e-01 3.40518266e-01 -1.40140748e+00
4.04551387e-01 -5.42793155e-01 2.30508983e-01 -5.59466720e-01
1.57005161e-01 -5.12913287e-01 -8.33184958e-01 1.55806229e-01
1.77541703e-01 -2.50927716e-01 -1.06403363e+00 1.98164773e+00
7.92567208e-02 5.50191402e-01 8.78730118e-02 7.68370569e-01
1.01917875e+00 1.23530078e+00 6.95729554e-02 -2.19604418e-01
1.20152247e+00 -1.11403263e+00 -1.07226121e+00 8.30470398e-02
-1.52852044e-01 -1.09248233e+00 1.43351448e+00 4.91192937e-01
-1.37798262e+00 -9.99981999e-01 -9.22054648e-01 -1.50583029e-01
2.74040308e-02 1.99841820e-02 -4.02332023e-02 6.28514826e-01
-9.43027616e-01 4.69281763e-01 -4.55389261e-01 3.06786865e-01
-2.75359228e-02 -2.99015015e-01 1.80609375e-01 5.13933182e-01
-1.37002301e+00 4.55810517e-01 1.90082550e-01 -2.04132587e-01
-1.33839679e+00 -1.16216552e+00 -7.41708219e-01 2.26216465e-01
1.60701767e-01 -8.25079858e-01 1.73366821e+00 -5.63085377e-01
-2.19863510e+00 2.26096094e-01 -3.77459764e-01 -4.60184574e-01
3.62887681e-01 -1.01613604e-01 -6.66287243e-01 1.20296262e-01
1.23383842e-01 8.64260316e-01 1.40425193e+00 -1.24667227e+00
-5.76465905e-01 4.47236210e-01 -4.38217849e-01 2.03768671e-01
2.02463686e-01 2.41106719e-01 -2.96347827e-01 -1.09981573e+00
-1.30813951e-02 -8.12454104e-01 1.36945352e-01 -2.22080484e-01
-7.17765868e-01 -2.88188964e-01 6.23254120e-01 -8.31395864e-01
1.33252978e+00 -2.51007771e+00 2.30587408e-01 -1.05374694e-01
-6.96433410e-02 -8.01382810e-02 -2.17305928e-01 4.31213886e-01
9.83687341e-02 6.42251223e-02 -2.04125643e-01 -8.89726758e-01
6.94504440e-01 7.05001876e-02 -7.56116033e-01 -1.28143832e-01
1.42072350e-01 6.19701326e-01 -8.64567935e-01 -6.22058213e-01
3.17782462e-01 6.88626587e-01 -1.01147521e+00 3.93271476e-01
-6.45026207e-01 5.55154324e-01 2.44735330e-02 3.29302013e-01
4.81713682e-01 2.98992187e-01 -4.75433290e-01 -1.12304397e-01
-2.79616416e-01 6.15922153e-01 -1.41267347e+00 1.78968608e+00
-6.18293166e-01 6.05556011e-01 1.54326677e-01 -4.90598679e-01
1.03240979e+00 8.46473277e-01 5.24871759e-02 -2.38497585e-01
-3.89566347e-02 3.13073009e-01 6.54477030e-02 -3.93324912e-01
7.66973197e-01 -5.10222733e-01 -2.01647133e-01 6.72724321e-02
2.89998144e-01 -9.31735516e-01 2.43444979e-01 1.55988663e-01
6.50427759e-01 3.45532954e-01 -1.01328954e-01 -4.81550656e-02
2.68036544e-01 -4.21043456e-01 5.49831152e-01 9.51906383e-01
-7.30508044e-02 1.27008605e+00 3.24029773e-01 4.15849507e-01
-1.13014472e+00 -1.70821297e+00 -9.10848528e-02 1.16077900e+00
-3.51729184e-01 -5.22418201e-01 -8.81381512e-01 -1.26454651e-01
-2.08953753e-01 1.35258198e+00 -2.16890827e-01 1.42397210e-02
-2.68620789e-01 -1.56031489e-01 9.19538081e-01 2.13725835e-01
1.56501874e-01 -1.45878100e+00 -1.33793488e-01 5.92271507e-01
-5.46203494e-01 -8.36837828e-01 -9.57857668e-01 1.04386203e-01
-9.12255868e-02 -3.92997384e-01 -1.12193167e+00 -6.85408831e-01
-2.42646202e-01 -2.19038650e-01 1.17932892e+00 -9.12579954e-01
2.38820463e-02 1.65993780e-01 -5.65699041e-01 -6.12917006e-01
-1.02081549e+00 -2.50040680e-01 3.50048363e-01 9.41172689e-02
-1.98344558e-01 -6.89298570e-01 -3.04535627e-01 -1.35961130e-01
-7.65283346e-01 2.59978682e-01 1.83820665e-01 7.48435974e-01
6.84322119e-01 -6.00503609e-02 9.65156794e-01 -4.57669735e-01
7.26210237e-01 -5.43437898e-01 -2.52610773e-01 -3.40605080e-02
-1.03647307e-01 -7.25959837e-02 1.02744305e+00 -6.97572827e-01
-1.10465825e+00 -3.58447015e-01 -7.17079520e-01 -7.65871465e-01
-2.29989320e-01 4.10800368e-01 -3.31031442e-01 7.45697320e-01
7.72530854e-01 4.35825855e-01 -1.04147591e-01 -6.27196252e-01
6.16125882e-01 1.08082879e+00 9.64864850e-01 -6.65484428e-01
6.40774488e-01 -1.70586690e-01 -2.49140933e-01 -1.02528870e+00
-7.48374820e-01 -5.25056683e-02 -5.11275604e-02 -1.89655989e-01
8.82454276e-01 -1.11624992e+00 -4.73684162e-01 5.74916363e-01
-1.37751913e+00 -3.77664089e-01 -5.87325573e-01 5.88596106e-01
-8.46637845e-01 1.20770529e-01 -7.07254648e-01 -9.49690640e-01
-4.32055086e-01 -1.11244142e+00 1.11700463e+00 1.50693625e-01
-5.17311335e-01 -5.02214551e-01 2.24053904e-01 3.12777348e-02
6.33148491e-01 1.11918114e-02 6.17980659e-01 -2.29558945e-01
-3.18242699e-01 1.93144828e-01 2.54941016e-01 8.55907559e-01
1.47702113e-01 4.01686400e-01 -1.15848863e+00 -2.26436183e-02
-1.05752110e-01 -1.34256393e-01 5.71740210e-01 6.71988308e-01
1.10309005e+00 -4.22062606e-01 3.00715774e-01 6.54856980e-01
9.63157713e-01 4.52989578e-01 7.17177749e-01 -2.91370511e-01
7.35426784e-01 3.24398041e-01 3.65879595e-01 6.74360335e-01
2.53651321e-01 7.25293279e-01 -4.53709960e-02 1.87536791e-01
-8.46770406e-01 -6.34530544e-01 6.83552027e-01 1.38923514e+00
3.79697323e-01 -5.17198443e-01 -3.69134873e-01 5.65928340e-01
-1.26170754e+00 -1.39742410e+00 -1.52854994e-01 1.89991891e+00
1.28164232e+00 2.01343447e-02 1.42662168e-01 3.79688412e-01
7.33557820e-01 3.40226799e-01 -3.61123830e-01 -5.45883775e-01
-5.83930835e-02 4.50197846e-01 -1.18115067e-01 7.38388121e-01
-5.86660206e-01 9.89483535e-01 6.45814133e+00 1.43461728e+00
-1.02641857e+00 1.90011725e-01 2.02331573e-01 -3.29719037e-01
-7.37882793e-01 -3.09549030e-02 -8.85575116e-01 9.19963717e-01
1.16999924e+00 -1.41101077e-01 7.90223658e-01 5.68545461e-01
6.07878566e-01 2.92063087e-01 -1.01597512e+00 9.37123477e-01
-1.18392974e-01 -1.28682959e+00 1.42183661e-01 -4.48270112e-01
8.40611815e-01 -1.98418528e-01 3.92048091e-01 6.59889340e-01
2.84829527e-01 -8.44548523e-01 1.59498370e+00 6.93112493e-01
1.12504089e+00 -7.29196131e-01 -6.09343909e-02 2.57756948e-01
-1.30861044e+00 2.14453682e-01 -2.76757538e-01 1.77606151e-01
6.29133642e-01 5.60083687e-01 -7.64796853e-01 3.48289907e-01
4.43052322e-01 3.13006282e-01 1.27321137e-02 8.64308774e-01
-2.58325398e-01 1.11166787e+00 -2.89200246e-01 -4.12189066e-02
1.27014697e-01 -8.71526599e-02 9.28291440e-01 1.40700686e+00
8.36059690e-01 -3.34200799e-01 -7.19808191e-02 1.22081089e+00
6.27756640e-02 -4.20469344e-02 -2.98879921e-01 -1.81053966e-01
1.04893899e+00 7.10811853e-01 -1.39431521e-01 -5.11078954e-01
-4.08879556e-02 1.02391446e+00 -4.78632182e-01 6.16752326e-01
-1.19198656e+00 -7.15962589e-01 5.82705975e-01 -4.44549322e-03
4.08731341e-01 -2.75954939e-02 -1.78729594e-01 -8.34605694e-01
-2.55360395e-01 -1.03170633e+00 -1.60818219e-01 -1.16566312e+00
-1.13810527e+00 7.08216310e-01 1.58969373e-01 -1.45353329e+00
-7.47444093e-01 -6.51819855e-02 -7.14132249e-01 1.21467590e+00
-1.19918251e+00 -1.06148052e+00 8.27255622e-02 4.70502168e-01
9.06609297e-01 -3.69118035e-01 7.26941884e-01 2.84604222e-01
-3.06499898e-01 7.29872942e-01 2.32097596e-01 -2.92567313e-01
8.67090166e-01 -1.38920999e+00 6.78470850e-01 8.37647736e-01
3.27175528e-01 4.69653010e-01 1.18753374e+00 -4.29452866e-01
-8.72469544e-01 -1.25453806e+00 1.07151592e+00 -3.50726098e-01
4.56212521e-01 -4.87296402e-01 -6.85696125e-01 3.42064381e-01
4.55464125e-01 -3.15428674e-01 5.93018651e-01 -8.92196000e-02
-2.96904713e-01 -1.35844229e-02 -8.28312635e-01 8.76074612e-01
7.94613838e-01 -8.72043252e-01 -6.43763840e-01 -6.87228069e-02
1.31325483e+00 -3.87601525e-01 -6.36592627e-01 1.99632034e-01
4.67376053e-01 -9.27520871e-01 1.05238736e+00 -1.61747560e-01
6.80939019e-01 -7.72231996e-01 -6.37889504e-01 -1.75434041e+00
-4.33099180e-01 -1.08606040e+00 -2.88787961e-01 1.83290613e+00
4.06519502e-01 -4.96469885e-02 3.80490780e-01 3.46673280e-02
-7.81856179e-01 -9.93102640e-02 -8.74020576e-01 -8.95785451e-01
2.62602270e-01 -8.75236094e-01 9.48222160e-01 5.25822401e-01
-2.69503981e-01 4.64772314e-01 -8.36227715e-01 9.85249057e-02
6.45294070e-01 -2.07144380e-01 8.44405532e-01 -8.23934436e-01
-8.63505542e-01 -4.62453604e-01 2.72259772e-01 -1.12278664e+00
4.80048172e-02 -8.83506894e-01 7.20052779e-01 -1.39912212e+00
-1.65766478e-01 -3.14782947e-01 -2.10420355e-01 7.02557191e-02
-1.89774916e-01 3.90331388e-01 6.19421422e-01 -7.34630376e-02
-2.45130032e-01 9.18687999e-01 1.35159421e+00 -6.99310824e-02
-2.60999203e-01 2.05122173e-01 -6.23844445e-01 4.57173049e-01
6.16569579e-01 -6.01165056e-01 -5.74894786e-01 -3.07546675e-01
7.45344162e-02 5.26914775e-01 3.29972535e-01 -1.19661415e+00
9.30318981e-02 -1.74425989e-01 -7.14381933e-02 -7.76608229e-01
7.25423753e-01 -3.05851698e-01 6.73741043e-01 5.44363074e-02
-5.30314863e-01 -5.83054960e-01 2.68241465e-02 3.08631957e-01
-4.10420746e-01 -3.71261448e-01 1.00756717e+00 -1.54792340e-02
-3.76743019e-01 8.65747556e-02 -6.84037864e-01 3.18069249e-01
4.49348152e-01 9.62790567e-03 -3.33300866e-02 -6.79427803e-01
-7.69094169e-01 -2.33569503e-01 3.62068713e-02 4.75422204e-01
5.47861934e-01 -1.73029721e+00 -1.18566215e+00 2.88703859e-01
-5.75022697e-02 4.40295134e-03 5.80525577e-01 3.22509944e-01
-2.29496792e-01 2.00729266e-01 1.35480776e-01 -5.07700443e-01
-8.39343250e-01 2.65188843e-01 1.73394397e-01 3.02231193e-01
-4.71634239e-01 1.17532575e+00 2.89787859e-01 -3.71397763e-01
5.40139079e-01 -3.86129856e-01 -2.91174185e-02 4.39131912e-03
6.29220545e-01 3.49248856e-01 -3.33745927e-01 -6.14649951e-01
-1.75486878e-01 2.63009697e-01 3.37548107e-01 -8.96512449e-01
7.87545562e-01 -2.87786067e-01 2.86554545e-01 9.54994202e-01
1.16706896e+00 5.25755763e-01 -1.53110695e+00 9.23206508e-02
-4.42299217e-01 -2.70618200e-01 7.07358122e-02 -8.98202837e-01
-6.27348721e-01 9.62263584e-01 2.28743404e-01 2.32404962e-01
8.98832619e-01 -2.49057710e-01 9.77572322e-01 -2.72923838e-02
1.08543411e-01 -1.15752912e+00 9.14394185e-02 8.18567514e-01
1.20714355e+00 -7.36412168e-01 -6.10569715e-01 -1.02478929e-01
-9.55129564e-01 6.58126593e-01 1.90181121e-01 -1.18578516e-01
6.27964079e-01 4.81344521e-01 2.33278379e-01 4.13394034e-01
-9.17247713e-01 -5.17833903e-02 4.23751086e-01 6.59135282e-01
4.60373312e-01 2.32804030e-01 5.56323007e-02 9.68100905e-01
-8.67591083e-01 -2.14478727e-02 4.86797720e-01 2.04857424e-01
-4.07314301e-01 -1.03213346e+00 -5.80351114e-01 2.06638306e-01
-6.14893913e-01 -5.68547130e-01 1.26556918e-01 1.79285839e-01
1.54817775e-01 1.39638305e+00 1.29591152e-01 -3.95539850e-01
2.45315477e-01 3.81938219e-01 3.10919315e-01 -6.80768013e-01
-5.20472050e-01 6.48313761e-01 2.03209281e-01 -5.06882034e-02
3.98366749e-02 -7.04003751e-01 -1.18598437e+00 -3.19503933e-01
-1.74888521e-01 3.08112442e-01 6.77406609e-01 7.02345490e-01
2.02136099e-01 1.20616996e+00 9.35406387e-01 -1.01015747e+00
-7.50245452e-01 -1.35899127e+00 -9.35789704e-01 3.43449712e-01
4.81748074e-01 -3.55528563e-01 -6.80601597e-01 4.27163422e-01] | [15.321371078491211, 6.321898937225342] |
33b52548-6070-4bf8-80f5-583d3689418d | improving-multiple-documents-grounded-goal | null | null | https://aclanthology.org/2022.dialdoc-1.15 | https://aclanthology.org/2022.dialdoc-1.15.pdf | Improving Multiple Documents Grounded Goal-Oriented Dialog Systems via Diverse Knowledge Enhanced Pretrained Language Model | In this paper, we mainly discuss about our submission to MultiDoc2Dial task, which aims to model the goal-oriented dialogues grounded in multiple documents. The proposed task is split into grounding span prediction and agent response generation. The baseline for the task is the retrieval augmented generation model, which consists of a dense passage retrieval model for the retrieval part and the BART model for the generation part. The main challenge of this task is that the system requires a great amount of pre-trained knowledge to generate answers grounded in multiple documents. To overcome this challenge, we adopt model pretraining, fine-tuning, and multi-task learning to enhance our model’s coverage of pretrained knowledge. We experimented with various settings of our method to show the effectiveness of our approaches. | ['Kyomin Jung', 'Hyunkyung Bae', 'Hwanhee Lee', 'Taegwan Kang', 'Hyung Joo Park', 'Dongryeol Lee', 'Yunah Jang'] | null | null | null | null | dialdoc-acl-2022-5 | ['goal-oriented-dialog', 'passage-retrieval'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.28486276e-02 4.95641500e-01 1.64901078e-01 -2.36003682e-01
-1.41636479e+00 -4.96420175e-01 1.06737137e+00 7.97635410e-04
-3.60051066e-01 1.12686038e+00 7.50720263e-01 -1.15003902e-02
2.40714014e-01 -7.89648712e-01 -6.37834549e-01 -2.97606885e-01
2.25840822e-01 1.00177705e+00 3.42477649e-01 -8.90006661e-01
2.82621741e-01 -2.42758155e-01 -1.05994654e+00 8.68280351e-01
1.08544481e+00 6.32817864e-01 3.45886528e-01 9.86400962e-01
-2.22876683e-01 1.20796835e+00 -1.02953291e+00 -4.65179682e-01
-1.15143426e-01 -7.62455225e-01 -1.46279824e+00 -1.66167721e-01
-1.50653981e-02 -5.88226199e-01 -2.48220280e-01 6.20884240e-01
7.60797679e-01 5.40983379e-01 7.67982006e-01 -9.15047765e-01
-8.03886771e-01 9.98967886e-01 -7.10291639e-02 -2.04916038e-02
6.27972901e-01 1.45351946e-01 1.07820809e+00 -7.62936831e-01
7.40098596e-01 1.47522283e+00 1.97575510e-01 9.11909103e-01
-9.54056561e-01 -9.55608711e-02 1.42263532e-01 1.68967456e-01
-9.62464631e-01 -4.39731538e-01 5.58056176e-01 -3.02358836e-01
1.19122744e+00 -1.76748578e-02 2.59707928e-01 1.30677199e+00
1.50202379e-01 1.09135664e+00 9.22883749e-01 -6.53209031e-01
3.13875228e-02 3.62945944e-01 3.30352247e-01 5.19233227e-01
-2.77918339e-01 3.02940644e-02 -4.49533880e-01 -4.09771204e-01
4.11154449e-01 -3.30395967e-01 -3.98982167e-01 2.00147167e-01
-1.18667424e+00 1.13337409e+00 2.42384568e-01 8.23643953e-02
-3.34452331e-01 -9.06640955e-04 2.49838457e-01 4.01804775e-01
5.61139822e-01 8.82344902e-01 -4.03076738e-01 -1.37691135e-02
-6.20777249e-01 9.54896688e-01 1.11116064e+00 1.15828884e+00
5.16326547e-01 -3.09879601e-01 -9.12339211e-01 1.08382463e+00
4.28201705e-01 2.94696242e-01 7.55768418e-01 -8.06360960e-01
8.64675760e-01 5.64261496e-01 5.82277238e-01 -4.48801726e-01
-4.34274614e-01 -2.46504113e-01 -5.61821342e-01 -4.43479210e-01
4.10661042e-01 -7.28329599e-01 -7.83860505e-01 1.80768001e+00
2.77429640e-01 -2.14678925e-02 6.24768198e-01 8.00514996e-01
1.24645495e+00 1.10808730e+00 1.96801499e-01 -1.58385366e-01
1.28787708e+00 -1.70997405e+00 -7.90279329e-01 -2.14979634e-01
9.81927812e-01 -8.01328003e-01 1.06931674e+00 -1.30572498e-01
-1.38494992e+00 -6.05909348e-01 -8.55625391e-01 -3.56326818e-01
-2.68637240e-01 7.28394911e-02 4.22142088e-01 -7.47437403e-02
-9.26530480e-01 2.95473158e-01 -3.45142752e-01 -3.69527817e-01
-4.04395878e-01 -1.40424091e-02 1.39330953e-01 -1.58366293e-01
-1.81364465e+00 1.10221076e+00 6.02224350e-01 4.86254171e-02
-1.14914727e+00 -3.21032256e-01 -7.88329244e-01 1.50149792e-01
2.80981332e-01 -1.05161655e+00 1.78947628e+00 -3.80021930e-01
-1.80833185e+00 5.49796462e-01 6.33936897e-02 -5.28947830e-01
5.63234150e-01 -2.16790736e-01 -1.30102038e-01 9.59306583e-02
-4.62624580e-02 8.12121212e-01 4.79332060e-01 -1.21737671e+00
-7.47882307e-01 -1.90984637e-01 5.67646921e-01 7.03647256e-01
-9.94389039e-03 6.70913681e-02 -3.59182447e-01 -4.74334896e-01
-4.31709468e-01 -9.66453433e-01 -4.05695647e-01 -1.21791911e+00
-3.99647206e-01 -7.62823462e-01 1.68947503e-01 -6.59424603e-01
1.11435592e+00 -1.51701021e+00 4.16474074e-01 -2.69630909e-01
-1.00899875e-01 1.25558704e-01 -6.97302401e-01 9.70297515e-01
5.14972568e-01 -2.93081645e-02 2.74071306e-01 -4.98276442e-01
1.24323122e-01 -1.45120144e-01 -7.42238283e-01 -4.30209488e-01
2.61872858e-01 9.50030386e-01 -1.05554354e+00 -4.16107744e-01
-4.49887723e-01 1.86715737e-01 -5.80068231e-01 8.50768149e-01
-1.04931986e+00 5.17737448e-01 -7.15964079e-01 1.98550120e-01
1.85961381e-01 -3.32390726e-01 2.07994431e-01 1.61590293e-01
1.37733743e-01 8.53148282e-01 -7.79962838e-01 1.99443400e+00
-4.18812752e-01 1.68510359e-02 -2.05682397e-01 -5.03349423e-01
8.30076218e-01 6.38773978e-01 -4.74049412e-02 -7.62903631e-01
9.07847099e-03 7.84440860e-02 2.79410519e-02 -6.04955494e-01
1.21175671e+00 -3.95025983e-02 -3.61094415e-01 9.43766356e-01
2.28008553e-01 -3.13846052e-01 4.50161695e-01 5.73854446e-01
8.67390335e-01 2.94425935e-01 -1.23786673e-01 1.18321339e-02
4.68650997e-01 2.86536545e-01 1.77941471e-01 9.59119678e-01
2.59699315e-01 4.22981083e-01 3.85858506e-01 -1.61243662e-01
-6.67915821e-01 -4.87110138e-01 4.65359271e-01 1.51534653e+00
8.77220333e-02 -5.00286162e-01 -9.31927919e-01 -8.55717540e-01
-3.57458472e-01 9.90875244e-01 -5.23391247e-01 -2.55262047e-01
-7.85488546e-01 -8.94830227e-01 4.15558010e-01 4.61952597e-01
5.09468079e-01 -1.43965709e+00 -2.30054542e-01 3.87768835e-01
-8.30749154e-01 -1.02176332e+00 -5.25636196e-01 -2.63241619e-01
-5.88204980e-01 -8.45267653e-01 -8.47925425e-01 -8.27132523e-01
3.64541262e-01 2.78366953e-01 1.39448071e+00 2.75426418e-01
3.29727888e-01 3.58076513e-01 -6.57177508e-01 -4.87374038e-01
-7.30462015e-01 6.07180595e-01 -2.55243480e-01 -3.17169309e-01
1.65674299e-01 -6.10409006e-02 -4.37853515e-01 2.53003180e-01
-8.43154907e-01 3.54377568e-01 4.37978059e-01 1.08879042e+00
2.00344473e-01 -3.86753917e-01 1.02416742e+00 -1.01160705e+00
1.45736814e+00 -6.15613103e-01 -4.25045401e-01 6.86754704e-01
-2.84345955e-01 1.99735343e-01 4.78307128e-01 -2.89111227e-01
-1.30232179e+00 -3.82383019e-01 -1.54165030e-01 2.79404372e-01
6.93922788e-02 8.71016562e-01 1.35003943e-02 4.73052025e-01
7.83925235e-01 2.28413552e-01 -1.21328279e-01 -5.20773113e-01
5.97042978e-01 8.50272179e-01 2.11607829e-01 -8.79094779e-01
3.02069038e-01 -3.27054918e-01 -5.51858008e-01 -5.38982928e-01
-1.18815005e+00 -3.18191200e-01 -2.39623979e-01 -4.65616137e-02
7.39210367e-01 -1.12964213e+00 -3.87632132e-01 2.49003351e-01
-1.61378968e+00 -9.01273727e-01 -9.86903608e-02 3.74028862e-01
-7.71449387e-01 1.73035786e-01 -9.97174203e-01 -6.90997362e-01
-7.60911167e-01 -1.01081133e+00 8.84245753e-01 2.80410200e-01
-1.05564356e-01 -1.12338984e+00 7.07014859e-01 7.72267938e-01
5.09471297e-01 -2.23000631e-01 1.01167929e+00 -1.15360248e+00
-6.26048684e-01 -1.65463552e-01 -4.56965677e-02 4.07594480e-02
-1.54889449e-01 -4.03184623e-01 -8.02220166e-01 -2.69602031e-01
-6.22688718e-02 -1.19347262e+00 9.62880552e-01 -1.78383514e-02
6.49095476e-01 -5.38204372e-01 -2.01057270e-01 -3.54086906e-02
1.01082218e+00 1.46061063e-01 6.77656412e-01 3.28070402e-01
4.62674648e-01 8.98417771e-01 9.04543459e-01 2.60239601e-01
1.02056396e+00 9.22180951e-01 9.02066305e-02 2.25277152e-02
-5.71304522e-02 -4.29983258e-01 4.02543068e-01 8.43057930e-01
-9.16333571e-02 -7.79909670e-01 -7.22976506e-01 5.86838901e-01
-2.23984861e+00 -1.02832210e+00 2.08617002e-01 1.88909256e+00
1.26313615e+00 -1.49142161e-01 3.12338412e-01 -5.88738084e-01
4.74280775e-01 9.41989496e-02 -1.44682318e-01 -4.15564716e-01
2.09513694e-01 -1.19680516e-01 -3.70951056e-01 9.92569327e-01
-7.99879134e-01 1.28630650e+00 6.81941843e+00 5.52885056e-01
-6.19807780e-01 7.53376931e-02 3.73433799e-01 -7.30562806e-02
-3.53441566e-01 -1.87291410e-02 -1.29360092e+00 3.17572773e-01
1.18546021e+00 -1.97238997e-01 3.13536495e-01 6.15057230e-01
1.13064684e-01 5.94522879e-02 -1.14817989e+00 1.97646752e-01
3.07077914e-01 -1.19370317e+00 3.98382872e-01 -1.66370198e-01
8.10124576e-01 -1.01868063e-02 -3.73022497e-01 1.06428730e+00
9.43556190e-01 -8.89571965e-01 4.22299385e-01 5.30350745e-01
1.63842887e-01 -4.78557825e-01 8.62706840e-01 7.54284501e-01
-4.76692587e-01 8.88583511e-02 -5.33703566e-01 -1.17105812e-01
3.72995287e-01 2.50180438e-02 -1.19416249e+00 6.25507772e-01
9.97304618e-02 1.58874422e-01 -2.82914996e-01 8.33308339e-01
-5.04472971e-01 3.12736690e-01 -6.90578064e-03 -2.86372960e-01
4.74847734e-01 -2.96778772e-02 2.46771336e-01 1.15857315e+00
2.35541180e-01 3.12370539e-01 6.24707997e-01 6.08899593e-01
-3.07533413e-01 1.81292340e-01 -2.91073680e-01 -8.55519623e-02
4.64603364e-01 1.30405974e+00 -2.14453433e-02 -6.16899729e-01
-2.61523187e-01 9.27106857e-01 6.60792947e-01 4.73065823e-01
-6.80333674e-01 -4.26782638e-01 3.79272886e-02 -9.13980231e-02
6.46350980e-02 -1.69455744e-02 3.04204285e-01 -1.28772378e+00
-3.74183476e-01 -1.12798035e+00 5.54232895e-01 -7.93010473e-01
-1.11551595e+00 9.70059752e-01 1.54747665e-01 -7.47694850e-01
-1.07168114e+00 -2.23275438e-01 -6.67061567e-01 1.06780350e+00
-1.84575105e+00 -1.45604050e+00 -2.52462536e-01 5.65329134e-01
9.01907980e-01 -2.73132175e-01 1.15990806e+00 1.77495007e-03
-5.41392624e-01 5.45448601e-01 4.39433865e-02 2.51238234e-02
1.01175356e+00 -1.32177508e+00 2.98065633e-01 3.83183777e-01
-8.12959820e-02 6.16156816e-01 7.82260060e-01 -6.32746279e-01
-1.08920443e+00 -1.12904966e+00 1.33936584e+00 -6.80696070e-01
5.40988326e-01 -2.57929802e-01 -7.59237468e-01 8.89966846e-01
7.65881300e-01 -6.50718033e-01 8.53978515e-01 2.63880193e-01
-6.47561997e-02 3.33540171e-01 -7.82002449e-01 6.84256196e-01
4.88435954e-01 -1.60476714e-01 -1.03540659e+00 8.11110675e-01
9.71592963e-01 -6.68736875e-01 -8.58027041e-01 1.86925098e-01
2.51229227e-01 -4.96885508e-01 7.32160330e-01 -1.03015637e+00
8.26658487e-01 -9.90302041e-02 -3.23233232e-02 -1.64643645e+00
-3.27630043e-01 -6.79187536e-01 -3.32029104e-01 1.33906257e+00
7.20444500e-01 -2.33545229e-01 4.06182885e-01 6.52254641e-01
-4.46563154e-01 -6.20535553e-01 -3.55042815e-01 -4.66641426e-01
3.12896848e-01 3.24570656e-01 7.24835694e-01 7.08362877e-01
3.86805028e-01 1.33265638e+00 -5.37222266e-01 -8.79718065e-02
2.58522570e-01 3.50937665e-01 1.08277655e+00 -9.72906888e-01
-5.84614456e-01 -4.92093600e-02 6.08844399e-01 -1.36873674e+00
1.79906726e-01 -7.80116677e-01 3.97452384e-01 -1.94884479e+00
3.54176641e-01 -3.27247411e-01 -9.50280651e-02 3.68320793e-01
-6.15533829e-01 -1.55709684e-02 2.44822979e-01 3.15175682e-01
-1.10783315e+00 8.96602213e-01 1.41135156e+00 -2.15742722e-01
-6.30583882e-01 1.65999427e-01 -9.63135898e-01 4.30962086e-01
7.82415867e-01 -3.31180602e-01 -8.09848785e-01 -7.82972813e-01
2.89103836e-01 6.03446841e-01 5.84995337e-02 -5.40261388e-01
2.78211594e-01 -2.25081429e-01 -1.18987774e-02 -6.95985794e-01
5.68924963e-01 -1.19604789e-01 -3.61570686e-01 1.91221029e-01
-9.94380116e-01 7.91018754e-02 1.06960185e-01 4.01917100e-01
-2.78465122e-01 -5.33045352e-01 3.23555619e-01 -5.38105905e-01
-3.34904939e-01 1.39893904e-01 -4.29056287e-01 2.83218592e-01
6.39395893e-01 5.49854755e-01 -7.49898672e-01 -6.78347170e-01
-5.86193144e-01 8.92934263e-01 1.12497501e-01 4.85736907e-01
5.41092575e-01 -1.26266909e+00 -1.18035126e+00 -3.63199651e-01
1.99876681e-01 6.70320317e-02 3.75406951e-01 4.35077757e-01
-2.31777981e-01 7.95318544e-01 2.35400554e-02 -1.06699184e-01
-1.01754320e+00 4.47333127e-01 2.32032433e-01 -1.14052880e+00
-3.44535768e-01 7.30965257e-01 1.15773216e-01 -6.66142642e-01
3.21450114e-01 1.39069706e-01 -8.34106803e-01 1.12288967e-01
9.58339334e-01 2.15608910e-01 -1.05146252e-01 -2.03833550e-01
2.35787734e-01 6.85660839e-02 -8.60553622e-01 -6.06305718e-01
1.08248436e+00 -2.21517593e-01 -1.68878645e-01 2.93482661e-01
6.41129434e-01 -3.46210480e-01 -8.59321237e-01 -4.71955806e-01
-6.56588795e-03 -3.46069783e-03 -1.48910806e-01 -1.19133615e+00
-2.73296505e-01 8.60788465e-01 2.01538801e-02 3.20387542e-01
6.52926862e-01 -6.17239140e-02 1.00502288e+00 8.23163211e-01
4.21018243e-01 -1.18515909e+00 4.56466377e-01 1.12326026e+00
1.29770386e+00 -1.17856979e+00 -1.99616939e-01 9.67071299e-03
-9.18007195e-01 7.84016728e-01 1.06091189e+00 2.97216438e-02
1.37208983e-01 -2.85397977e-01 2.03187436e-01 -1.27308130e-01
-1.41598356e+00 -1.49641648e-01 1.98188201e-01 4.03673291e-01
6.45927548e-01 -1.93179205e-01 -5.48223197e-01 7.58683920e-01
-3.33539605e-01 -3.31823863e-02 6.09638155e-01 7.91081071e-01
-5.69184422e-01 -1.27967608e+00 1.87923536e-02 2.20599473e-02
-4.18411970e-01 -2.62856722e-01 -7.10116446e-01 4.47115362e-01
-4.80062753e-01 1.34075403e+00 -1.92325428e-01 -3.68158311e-01
4.23911095e-01 3.55103880e-01 5.89350641e-01 -9.90039110e-01
-9.44685340e-01 1.00774199e-01 5.95751047e-01 -2.20147148e-01
-2.87453502e-01 -1.97816595e-01 -1.11033869e+00 -1.67275310e-01
-4.94039953e-01 7.44376481e-01 4.14984196e-01 1.00494623e+00
4.15352434e-01 6.03830278e-01 6.10944748e-01 -5.51506162e-01
-1.09047127e+00 -1.68893051e+00 -2.66819477e-01 3.90814334e-01
6.54094443e-02 -3.26341689e-01 -1.37503054e-02 -1.02302626e-01] | [12.392857551574707, 8.207060813903809] |
1bc3f79e-d166-4cb8-b24b-e1368b79324e | query-efficient-physical-hard-label-attacks | 2002.07088 | null | https://arxiv.org/abs/2002.07088v6 | https://arxiv.org/pdf/2002.07088v6.pdf | GRAPHITE: Generating Automatic Physical Examples for Machine-Learning Attacks on Computer Vision Systems | This paper investigates an adversary's ease of attack in generating adversarial examples for real-world scenarios. We address three key requirements for practical attacks for the real-world: 1) automatically constraining the size and shape of the attack so it can be applied with stickers, 2) transform-robustness, i.e., robustness of a attack to environmental physical variations such as viewpoint and lighting changes, and 3) supporting attacks in not only white-box, but also black-box hard-label scenarios, so that the adversary can attack proprietary models. In this work, we propose GRAPHITE, an efficient and general framework for generating attacks that satisfy the above three key requirements. GRAPHITE takes advantage of transform-robustness, a metric based on expectation over transforms (EoT), to automatically generate small masks and optimize with gradient-free optimization. GRAPHITE is also flexible as it can easily trade-off transform-robustness, perturbation size, and query count in black-box settings. On a GTSRB model in a hard-label black-box setting, we are able to find attacks on all possible 1,806 victim-target class pairs with averages of 77.8% transform-robustness, perturbation size of 16.63% of the victim images, and 126K queries per pair. For digital-only attacks where achieving transform-robustness is not a requirement, GRAPHITE is able to find successful small-patch attacks with an average of only 566 queries for 92.2% of victim-target pairs. GRAPHITE is also able to find successful attacks using perturbations that modify small areas of the input image against PatchGuard, a recently proposed defense against patch-based attacks. | ['Atul Prakash', 'Somesh Jha', 'Earlence Fernandes', 'Jiefeng Chen', 'Neal Mangaokar', 'Ryan Feng'] | 2020-02-17 | null | null | null | null | ['license-plate-recognition'] | ['computer-vision'] | [ 5.03907084e-01 -1.27527580e-01 2.29038402e-01 -6.12334646e-02
-1.33964455e+00 -1.39912319e+00 3.12594026e-01 -1.60115704e-01
-4.71882969e-01 4.40535694e-01 -4.29895729e-01 -9.44380939e-01
-2.47418769e-02 -9.73548830e-01 -8.88865948e-01 -6.79308116e-01
-2.72923440e-01 1.86301604e-01 3.94515306e-01 -3.79652619e-01
2.25712471e-02 9.05081868e-01 -1.08443308e+00 2.16192111e-01
4.25990015e-01 9.15598810e-01 -4.75939810e-01 1.22781432e+00
5.49470067e-01 2.46737048e-01 -1.23185325e+00 -7.54587770e-01
1.02109325e+00 -2.98209250e-01 -5.46244383e-01 -1.18492164e-01
6.87696159e-01 -4.32938933e-01 -4.91803110e-01 1.12818956e+00
9.28144872e-01 -1.30832091e-01 2.95144767e-01 -1.46616912e+00
-2.68549770e-01 4.46471095e-01 -5.29863894e-01 6.25824034e-02
4.77828652e-01 7.91965365e-01 8.22156310e-01 -2.47256845e-01
4.28354949e-01 1.25608861e+00 4.02361304e-01 6.42224669e-01
-1.35841513e+00 -1.21297765e+00 -2.63449937e-01 -9.26941913e-03
-1.48280609e+00 -4.12803978e-01 1.70999467e-01 -9.66674685e-02
7.35114694e-01 1.02384496e+00 5.79672493e-02 1.12652397e+00
1.70912936e-01 6.81619532e-03 1.34415829e+00 -3.80002826e-01
1.63974687e-01 2.79846162e-01 -4.95613784e-01 3.86878729e-01
3.21422130e-01 6.61547661e-01 -2.23013327e-01 -7.14598954e-01
4.39730406e-01 -3.48557055e-01 -4.51345384e-01 -4.99611907e-02
-9.50042725e-01 8.01604867e-01 3.34939212e-01 -6.68863654e-02
1.89335808e-01 3.55870485e-01 1.49112344e-01 6.88157260e-01
3.40116732e-02 7.56804943e-01 -2.63440102e-01 2.89270371e-01
-7.05038846e-01 3.50561827e-01 8.21217597e-01 1.00054455e+00
7.10957706e-01 1.97491422e-01 -2.51716286e-01 1.28862426e-01
-2.17032782e-03 1.17516112e+00 -7.25566000e-02 -7.93907046e-01
8.32333386e-01 -1.00860819e-02 5.66012114e-02 -1.04747415e+00
-8.20142552e-02 -2.79130191e-01 -4.21117812e-01 7.59864092e-01
5.82429767e-01 -5.12425959e-01 -8.89581442e-01 2.02347112e+00
5.22789538e-01 -4.17808034e-02 1.25717238e-01 4.87803280e-01
3.53415400e-01 6.88425362e-01 -3.16133440e-01 -6.58161268e-02
1.41526699e+00 -3.08127314e-01 -2.41273552e-01 -1.68541700e-01
6.96458995e-01 -1.04321492e+00 1.13677776e+00 4.11470175e-01
-9.58239734e-01 -2.89859235e-01 -1.31581450e+00 4.61619586e-01
-5.56353569e-01 -2.96752989e-01 4.47706357e-02 1.60699105e+00
-9.71464336e-01 2.26708248e-01 -3.76843184e-01 -3.96278240e-02
2.59036928e-01 7.24088669e-01 -4.95708913e-01 -2.68473700e-02
-1.31409681e+00 7.38901436e-01 7.39854127e-02 -2.71171719e-01
-1.14233303e+00 -6.61977112e-01 -6.17706597e-01 6.82770908e-02
6.95993245e-01 -4.21665430e-01 9.36342418e-01 -4.28020865e-01
-1.24137914e+00 7.12734818e-01 2.80814141e-01 -5.95314026e-01
8.20181489e-01 9.38074142e-02 -6.59463882e-01 3.13339651e-01
-2.65047610e-01 4.95118350e-01 1.14342773e+00 -1.19516540e+00
-4.54880983e-01 -2.49500960e-01 3.24996412e-01 -1.80352468e-03
-5.35934210e-01 2.93706477e-01 -2.31877223e-01 -8.27957571e-01
-3.22976738e-01 -1.36974931e+00 -2.99473912e-01 3.01577523e-02
-7.98827291e-01 6.83775961e-01 1.30200911e+00 -5.58153749e-01
1.02711880e+00 -2.40508389e+00 -5.32948554e-01 8.49677980e-01
9.89869516e-03 5.45599639e-01 -4.01535988e-01 5.79178691e-01
-1.90879330e-01 6.06330335e-01 -2.32727349e-01 -4.60843034e-02
5.86498007e-02 2.53947191e-02 -1.02223861e+00 6.24118924e-01
2.17111055e-02 6.89400613e-01 -3.84051025e-01 -7.41474032e-02
-6.40219031e-03 3.01291376e-01 -6.33678019e-01 2.48006001e-01
-1.90697774e-01 4.26033854e-01 -2.36818209e-01 5.34407020e-01
8.88137162e-01 2.80546695e-01 -2.97070928e-02 2.69757435e-02
3.12599331e-01 -3.88743822e-03 -1.49601054e+00 8.94157648e-01
-4.98989910e-01 5.85770667e-01 3.37822735e-01 -1.62674338e-01
8.42349052e-01 2.90718734e-01 -5.11385836e-02 -5.24982393e-01
1.01924017e-01 3.47069383e-01 1.50534406e-01 -2.60407035e-03
1.53186426e-01 7.92712346e-02 -5.24565995e-01 9.33400393e-01
-5.08820236e-01 -6.58293486e-01 -2.22319532e-02 2.36658648e-01
1.51397705e+00 -4.55742478e-01 -1.08381219e-01 3.04581132e-02
3.53389382e-01 -3.67611825e-01 2.91752696e-01 1.14174879e+00
1.36961579e-01 6.36861265e-01 4.66769189e-01 -3.18555593e-01
-9.67222333e-01 -1.17751479e+00 2.20737740e-01 8.56300890e-01
2.94357538e-01 -4.40684944e-01 -8.41310561e-01 -8.35723221e-01
1.30765796e-01 7.74164498e-01 -3.53271008e-01 -3.18913817e-01
-5.55182874e-01 -5.06053627e-01 1.30809891e+00 -2.51149647e-02
6.28154576e-01 -7.34241605e-01 -5.29007614e-01 -2.00871959e-01
4.48837578e-02 -1.35663962e+00 -8.14342022e-01 4.10820171e-02
-2.90880293e-01 -1.18414354e+00 -1.96453512e-01 -3.41301024e-01
8.05341959e-01 3.27092677e-01 6.52330637e-01 2.96468347e-01
-5.52644730e-01 3.94723952e-01 -2.71954387e-01 -3.88594240e-01
-8.83894384e-01 -2.70234406e-01 -2.36846711e-02 1.18931256e-01
-2.87285030e-01 -4.70070899e-01 -3.80555958e-01 9.31235909e-01
-1.31126130e+00 -5.55749118e-01 4.93543863e-01 6.65537894e-01
4.08186555e-01 5.24154007e-01 1.50865521e-02 -8.68163347e-01
6.21825159e-01 -1.94415171e-02 -9.68720853e-01 4.96072769e-01
-4.44085777e-01 -1.26530409e-01 7.70128429e-01 -8.37172806e-01
-5.50179005e-01 -1.04075700e-01 -1.32907838e-01 -3.37295294e-01
-2.50786040e-02 -2.92996734e-01 -5.20370007e-01 -8.59125614e-01
1.18826735e+00 2.12075636e-01 -1.79650888e-01 6.30593672e-02
4.32526231e-01 6.40013099e-01 5.24377525e-01 -8.71963561e-01
1.80431342e+00 5.01409948e-01 2.34653905e-01 -6.30837977e-01
-2.46107429e-01 -2.60266550e-02 1.07382014e-01 1.14831626e-02
4.75343853e-01 -5.41584790e-01 -1.03693199e+00 6.80641294e-01
-8.57406378e-01 -5.03696501e-01 -2.39882201e-01 6.37035146e-02
-4.73864406e-01 4.63282883e-01 -3.97462994e-01 -7.14248002e-01
-3.74971181e-01 -1.51057327e+00 9.15263772e-01 1.16023943e-02
-1.05274856e-01 -5.15653312e-01 -4.29814160e-01 4.90624040e-01
6.34645224e-01 6.48533285e-01 6.91161275e-01 -9.58033383e-01
-7.96196938e-01 -6.61351383e-01 -9.00588483e-02 4.73255903e-01
7.08828047e-02 -1.01137280e-01 -1.04068673e+00 -9.24282134e-01
1.89216822e-01 -3.52962554e-01 3.12794834e-01 -1.06316999e-01
1.26468623e+00 -8.19770217e-01 -1.48654267e-01 9.90743160e-01
1.24789619e+00 3.76788259e-01 8.70930552e-01 3.36718947e-01
4.98835474e-01 4.34697032e-01 7.04351485e-01 3.49562258e-01
-2.58878112e-01 9.02279735e-01 5.88169038e-01 -3.09675574e-01
1.73217803e-01 -1.87451661e-01 5.24892688e-01 -3.18642527e-01
5.14797986e-01 -5.11500001e-01 -7.39696205e-01 4.40044589e-02
-1.09568679e+00 -1.10638952e+00 7.96334371e-02 2.54781127e+00
9.34609473e-01 6.09017789e-01 1.42890260e-01 1.81331813e-01
8.63624156e-01 3.03813517e-01 -4.60001528e-01 -6.57762527e-01
-2.23848641e-01 5.78353763e-01 1.26969397e+00 6.09091163e-01
-1.13839567e+00 8.70070219e-01 6.23639345e+00 1.19558549e+00
-1.20214343e+00 8.15170258e-03 7.60836601e-01 -3.20274204e-01
-3.72957617e-01 2.66898006e-01 -8.72803986e-01 5.13173044e-01
9.33373213e-01 -2.36469194e-01 7.19144106e-01 5.49036860e-01
-8.16541612e-02 1.29686654e-01 -9.25612211e-01 6.55380726e-01
-1.15745291e-02 -1.06406808e+00 -1.31941870e-01 5.56934476e-01
6.26552463e-01 -5.78269839e-01 6.54845178e-01 4.79127988e-02
5.20899475e-01 -1.26932144e+00 7.24574327e-01 -1.64700061e-01
1.37329507e+00 -8.73341084e-01 4.98490244e-01 3.59597653e-01
-9.11138952e-01 -2.90656626e-01 -8.04175436e-02 4.24017608e-01
-2.29215203e-03 3.26846451e-01 -1.06057131e+00 3.84064794e-01
6.65131986e-01 -3.97892952e-01 -4.88751262e-01 8.21803987e-01
-5.17228901e-01 7.34133005e-01 -7.50457525e-01 1.95117012e-01
2.28801757e-01 2.16466248e-01 8.09655488e-01 1.14669454e+00
5.17251372e-01 4.76640910e-02 2.43253678e-01 6.45453990e-01
1.19804010e-01 -6.88941330e-02 -7.09576070e-01 1.68623239e-01
9.10066426e-01 1.05243349e+00 -5.19023418e-01 9.58846137e-02
2.40731552e-01 8.60357881e-01 -4.45947260e-01 4.05667096e-01
-1.22032869e+00 -8.48544598e-01 7.99754024e-01 1.51679292e-01
4.65908974e-01 -2.48208120e-01 -2.19376504e-01 -5.53211868e-01
1.18660435e-01 -1.74665368e+00 5.97471237e-01 -4.69825506e-01
-1.00081933e+00 6.39045477e-01 5.31659536e-02 -1.33352363e+00
-1.15363151e-01 -5.66402555e-01 -7.87196338e-01 1.15360105e+00
-1.03942084e+00 -1.31984353e+00 -2.52986085e-02 8.98905158e-01
-5.36268502e-02 -3.18504006e-01 8.23375404e-01 1.73335820e-01
-3.82114321e-01 1.32628572e+00 -1.23155735e-01 1.04335062e-01
8.78249884e-01 -1.02221358e+00 9.75907743e-01 1.29243326e+00
4.34619784e-01 6.87999964e-01 8.10882449e-01 -5.54757535e-01
-1.41619015e+00 -9.50043082e-01 3.02513003e-01 -4.84483510e-01
5.45041800e-01 -8.02369535e-01 -4.28685337e-01 3.95063639e-01
-2.29933426e-01 6.93894476e-02 7.21012890e-01 -4.86659229e-01
-1.01085186e+00 -3.70916575e-01 -1.68168652e+00 9.02399182e-01
7.36536741e-01 -7.54686534e-01 1.75282374e-01 5.99893332e-01
8.99177492e-01 -8.16077292e-01 -5.84902406e-01 3.67539257e-01
2.81846106e-01 -9.51690018e-01 1.20692527e+00 -2.76790351e-01
-4.20130372e-01 -5.34539580e-01 -2.58768409e-01 -9.81520414e-01
1.01107784e-01 -1.52539456e+00 1.32745355e-01 1.11517799e+00
5.72159648e-01 -1.05966759e+00 6.26255512e-01 5.80917597e-01
2.86639959e-01 -4.78935301e-01 -1.11129057e+00 -1.02953351e+00
-1.08486794e-01 -5.00356078e-01 1.04971683e+00 6.13414049e-01
-5.61690688e-01 -4.14375782e-01 -5.20509362e-01 7.55203724e-01
5.99988461e-01 -3.33322793e-01 1.28862441e+00 -4.62470084e-01
-8.15279961e-01 -2.20215812e-01 -6.08521700e-01 -6.63224101e-01
-8.62702802e-02 -4.34338778e-01 -2.00400516e-01 -6.89892530e-01
-2.53926367e-01 -5.59001207e-01 9.73754153e-02 7.41253674e-01
6.35688333e-03 5.16127586e-01 4.23754573e-01 4.55699675e-02
1.44116864e-01 -1.56748518e-01 9.01213944e-01 -2.74354637e-01
-1.22216485e-01 4.32823062e-01 -7.51432598e-01 2.74463385e-01
8.82644713e-01 -7.38575518e-01 -5.89897513e-01 -1.43804476e-01
2.04311252e-01 -4.10958938e-03 5.85166097e-01 -1.24158061e+00
8.26450959e-02 -3.79007608e-01 6.53736293e-02 -2.19760180e-01
3.91209245e-01 -1.08799005e+00 5.13754487e-01 7.07579017e-01
-2.78388441e-01 2.02588156e-01 4.75732237e-01 5.08583248e-01
2.12022319e-01 -1.69137537e-01 9.11127567e-01 2.51648992e-01
-1.01141542e-01 3.44240189e-01 -5.87898418e-02 1.48883626e-01
1.48632014e+00 -4.42697436e-01 -9.29755926e-01 -7.40360260e-01
-3.51386875e-01 4.11414653e-02 7.42435277e-01 1.73659518e-01
4.71416771e-01 -1.02956915e+00 -7.88255095e-01 4.17915285e-01
-3.12016234e-02 -4.16209251e-01 2.39945471e-01 4.00977224e-01
-7.54529655e-01 -9.09221172e-02 5.99957928e-02 -2.90015608e-01
-1.82085586e+00 6.86851025e-01 4.33773875e-01 -3.36732656e-01
-2.88090974e-01 1.03920126e+00 1.25057235e-01 -2.80934423e-01
3.82946938e-01 2.90147275e-01 5.49143493e-01 -4.96834010e-01
6.78573370e-01 1.53282434e-01 9.53357071e-02 -5.53235412e-01
-2.73818791e-01 6.30993843e-01 -4.89597842e-02 -6.21087432e-01
8.10067236e-01 2.53411025e-01 6.97020665e-02 -5.68021715e-01
1.14489281e+00 6.43706024e-01 -1.04584646e+00 -2.27760095e-02
-3.51731062e-01 -9.98619080e-01 -2.69498855e-01 -8.31530213e-01
-1.10031986e+00 7.63540268e-01 6.82334244e-01 4.84272450e-01
1.30539155e+00 -4.59975481e-01 1.05898213e+00 4.17746335e-01
5.00892758e-01 -7.60718942e-01 1.36849523e-01 1.12966396e-01
8.53190660e-01 -5.74388921e-01 9.34729353e-02 -5.99588156e-01
-4.95120525e-01 8.94393861e-01 4.99536127e-01 1.79971620e-01
4.31656122e-01 6.53935373e-01 1.72508091e-01 1.63006067e-01
-6.05323553e-01 2.74507314e-01 1.97258890e-01 8.84447098e-01
-4.05974388e-01 9.91624370e-02 2.76868761e-01 -2.16345228e-02
-6.25021160e-01 -7.84981072e-01 5.74577153e-01 9.98848975e-01
-2.78845191e-01 -1.44274044e+00 -1.14847267e+00 1.02796979e-01
-5.85297644e-01 -1.28859803e-01 -5.96042216e-01 7.03334510e-01
8.34129006e-02 1.33579850e+00 -3.81655544e-01 -7.74774134e-01
3.93771380e-01 -3.20580602e-01 1.76484376e-01 -5.37445784e-01
-1.00711119e+00 -9.31648761e-02 1.89041719e-01 -6.35223627e-01
3.31169069e-01 -3.50187510e-01 -8.53512943e-01 -6.77055359e-01
-5.23081541e-01 1.10295881e-03 6.32322550e-01 5.78915179e-01
4.73108917e-01 1.45576159e-02 1.26432025e+00 -4.46485311e-01
-1.27764869e+00 -3.47108275e-01 -4.40602899e-01 3.39592814e-01
1.81450129e-01 -1.74359739e-01 -8.06467056e-01 -2.16856331e-01] | [5.550463676452637, 7.907053470611572] |
a773869b-2b42-4e7c-a59b-238b7a54afaf | density-ratio-based-personalised-ranking-from | 2101.07481 | null | https://arxiv.org/abs/2101.07481v1 | https://arxiv.org/pdf/2101.07481v1.pdf | Density-Ratio Based Personalised Ranking from Implicit Feedback | Learning from implicit user feedback is challenging as we can only observe positive samples but never access negative ones. Most conventional methods cope with this issue by adopting a pairwise ranking approach with negative sampling. However, the pairwise ranking approach has a severe disadvantage in the convergence time owing to the quadratically increasing computational cost with respect to the sample size; it is problematic, particularly for large-scale datasets and complex models such as neural networks. By contrast, a pointwise approach does not directly solve a ranking problem, and is therefore inferior to a pairwise counterpart in top-K ranking tasks; however, it is generally advantageous in regards to the convergence time. This study aims to establish an approach to learn personalised ranking from implicit feedback, which reconciles the training efficiency of the pointwise approach and ranking effectiveness of the pairwise counterpart. The key idea is to estimate the ranking of items in a pointwise manner; we first reformulate the conventional pointwise approach based on density ratio estimation and then incorporate the essence of ranking-oriented approaches (e.g. the pairwise approach) into our formulation. Through experiments on three real-world datasets, we demonstrate that our approach not only dramatically reduces the convergence time (one to two orders of magnitude faster) but also significantly improving the ranking performance. | ["Shin'ichi Satoh", 'Mayu Otani', 'Masahiro Kato', 'Riku Togashi'] | 2021-01-19 | null | null | null | null | ['density-ratio-estimation'] | ['methodology'] | [ 1.39668211e-01 -2.08579659e-01 -3.23865652e-01 -2.97496259e-01
-1.10425842e+00 -5.22642374e-01 3.87290776e-01 3.22677195e-01
-6.42729163e-01 1.04524016e+00 -8.52062628e-02 -2.12394118e-01
-5.92277765e-01 -8.03704202e-01 -5.40666759e-01 -8.04296315e-01
-9.11202207e-02 5.69553494e-01 -1.73788741e-02 -2.21148893e-01
5.11112571e-01 1.75823346e-02 -1.75212908e+00 7.92781860e-02
1.21928978e+00 1.02597606e+00 7.55002201e-02 2.80293286e-01
-5.76951727e-02 6.71489358e-01 -5.19491732e-01 -5.29608190e-01
2.88464844e-01 -2.29095370e-01 -6.25416458e-01 -2.02466473e-01
4.04460698e-01 -4.60264742e-01 3.04463506e-01 1.17480123e+00
6.03178561e-01 3.59621882e-01 6.25344396e-01 -1.15727222e+00
-7.00131476e-01 2.22156599e-01 -7.65888453e-01 7.75924139e-03
2.32738197e-01 -4.94875163e-01 1.42787755e+00 -1.04950893e+00
8.60774219e-02 9.50104177e-01 6.07083440e-01 4.82295901e-01
-1.28783131e+00 -5.87055624e-01 4.89969641e-01 1.05976433e-01
-1.22294354e+00 -1.80960387e-01 7.49234021e-01 -3.26848239e-01
3.26769412e-01 4.10155803e-01 4.85976845e-01 7.98213899e-01
-4.32733327e-01 1.03530991e+00 1.35298836e+00 -3.75277162e-01
2.18828455e-01 5.30681551e-01 3.73710245e-01 3.71953785e-01
2.99326390e-01 1.01416796e-01 -5.37549794e-01 -4.50624317e-01
7.32799053e-01 1.32191375e-01 -2.84941494e-01 -8.57587993e-01
-9.58559275e-01 8.45294893e-01 4.40649539e-01 7.19002038e-02
-2.61913866e-01 -1.93481088e-01 4.60547835e-01 5.58738470e-01
7.86663890e-01 4.53742892e-01 -5.79689503e-01 -1.58444032e-01
-8.46812785e-01 3.53528947e-01 7.75273144e-01 4.77135390e-01
7.62705147e-01 -3.69244576e-01 -1.22525290e-01 1.28322160e+00
3.09610814e-01 3.01689804e-01 5.28175056e-01 -6.46898448e-01
5.91150224e-01 5.89802682e-01 5.86121202e-01 -1.08550739e+00
-6.45655990e-02 -5.72713673e-01 -8.50930691e-01 1.37549207e-01
6.31302655e-01 -1.21364519e-02 -4.86249983e-01 1.94120300e+00
2.57019281e-01 -9.51742455e-02 -1.93441257e-01 9.43287671e-01
4.72131431e-01 4.28113133e-01 -5.85797355e-02 -4.98505145e-01
1.03162527e+00 -7.72598147e-01 -5.72647870e-01 4.72895661e-03
5.53225398e-01 -4.66119111e-01 1.32316959e+00 7.80842543e-01
-9.53810871e-01 -2.62815356e-01 -9.10380185e-01 1.65240213e-01
-4.01135027e-01 2.80443013e-01 6.11605883e-01 7.54114270e-01
-1.21466184e+00 5.44122994e-01 -1.73287392e-01 1.71952620e-02
1.40928239e-01 7.23753333e-01 -2.19265953e-01 1.69445679e-01
-1.44161057e+00 7.34250426e-01 3.26418251e-01 3.01294059e-01
-2.78028846e-01 -7.85807431e-01 -6.86446726e-01 2.16206506e-01
5.30315459e-01 -5.40344656e-01 1.41577387e+00 -1.20537376e+00
-1.64837372e+00 4.91223425e-01 -3.02724689e-01 -2.23084837e-01
7.93631792e-01 -6.19620919e-01 6.73219264e-02 -3.04190964e-01
-5.79992719e-02 2.93528438e-01 6.40596330e-01 -1.33721960e+00
-6.35224104e-01 -4.60747391e-01 2.53296733e-01 5.41756392e-01
-8.44386220e-01 -2.35849634e-01 -1.17022835e-01 -4.13504332e-01
-9.64383781e-02 -7.19486296e-01 -3.53532255e-01 -2.14693591e-01
-4.64345179e-02 -4.55440909e-01 5.43019533e-01 -2.87114352e-01
1.30941141e+00 -2.02882338e+00 -3.05737685e-02 4.39255238e-01
3.15103382e-01 4.68656301e-01 -6.99095502e-02 4.74611640e-01
-1.37123674e-01 -3.71007994e-02 -9.91343036e-02 -2.95778811e-01
-2.05230974e-02 -1.40047804e-01 -3.07956845e-01 2.48237893e-01
1.36208773e-01 7.56422460e-01 -1.46779478e+00 -2.87996113e-01
1.61424518e-01 3.86087209e-01 -6.89629674e-01 2.46891096e-01
9.09375101e-02 5.40070981e-02 -2.88061082e-01 3.64214748e-01
6.21177316e-01 -4.74308521e-01 5.98747730e-01 6.20525666e-02
-5.56078590e-02 4.49066401e-01 -1.42795897e+00 9.97133374e-01
-5.76304972e-01 2.27076456e-01 -8.14153105e-02 -1.29347861e+00
1.03493607e+00 3.76248807e-01 5.93941033e-01 -7.46363699e-01
-2.14775160e-01 5.81018329e-01 -2.46086732e-01 -9.32365581e-02
7.79625297e-01 -3.29319149e-01 3.23227048e-02 4.91177022e-01
-2.97185630e-01 5.01076877e-01 2.40292862e-01 2.00191408e-01
6.83107555e-01 -5.93667589e-02 5.49285054e-01 -3.50659788e-01
6.85147345e-01 -4.15327311e-01 3.68867695e-01 7.18425453e-01
-6.65729418e-02 4.75672603e-01 6.70358956e-01 -2.77242899e-01
-6.52432680e-01 -1.15089178e+00 -3.77422124e-02 1.24214709e+00
2.13976756e-01 -4.43606734e-01 -4.10207182e-01 -1.00105941e+00
9.18762013e-02 2.47667506e-01 -7.22765565e-01 -2.71949545e-02
-3.54081571e-01 -1.01907408e+00 -3.14756073e-02 5.36593437e-01
3.77990723e-01 -8.73281002e-01 -1.59617797e-01 -1.42889470e-02
-2.95941830e-01 -4.21427697e-01 -3.06577981e-01 2.71547765e-01
-1.19846225e+00 -9.89222169e-01 -9.71375585e-01 -6.02075279e-01
9.26788867e-01 5.40195048e-01 1.25580943e+00 1.40562236e-01
2.84359246e-01 2.76682585e-01 -3.25723737e-01 -2.81342208e-01
1.34039357e-01 1.78865656e-01 1.63639799e-01 1.82561681e-01
5.71182072e-01 -4.24869150e-01 -8.13136101e-01 3.55807960e-01
-8.87587786e-01 -2.16465294e-01 8.84058535e-01 1.27887678e+00
3.99530560e-01 7.59022916e-03 1.24296975e+00 -1.41091692e+00
9.79550183e-01 -5.75080752e-01 -6.07210994e-01 3.04274142e-01
-1.01406384e+00 6.70337975e-02 6.88220203e-01 -4.44097608e-01
-1.00121689e+00 -4.59413566e-02 2.32544579e-02 -1.22119799e-01
3.28150481e-01 7.59904802e-01 1.55491587e-02 1.86215401e-01
4.70503151e-01 2.39550740e-01 6.61901310e-02 -4.24270719e-01
3.03437293e-01 6.88984036e-01 1.84339240e-01 -7.04047918e-01
8.38711619e-01 2.75055528e-01 -2.88226634e-01 -5.93259692e-01
-9.15762126e-01 -7.59022951e-01 -5.84012449e-01 -2.90952504e-01
1.54685855e-01 -7.20891416e-01 -8.78267169e-01 2.87150860e-01
-7.41425633e-01 -5.95089570e-02 -1.78850397e-01 4.85196143e-01
-4.68365371e-01 5.58468163e-01 -4.73625898e-01 -1.30591786e+00
-3.45668286e-01 -7.51535475e-01 7.96632349e-01 1.24680484e-02
-2.49385700e-01 -1.12857759e+00 2.81798005e-01 2.79639393e-01
4.37090516e-01 -1.58404335e-01 1.11530519e+00 -6.12002850e-01
-4.31502014e-01 -2.46628433e-01 -5.06575525e-01 5.67655802e-01
1.00910649e-01 -3.50904524e-01 -9.38197851e-01 -4.34168071e-01
-6.27088547e-02 -4.97212201e-01 7.45797157e-01 2.89283752e-01
1.09325123e+00 -2.97063828e-01 1.44247741e-01 -7.68003613e-02
1.50067461e+00 -9.96370167e-02 6.10739052e-01 4.28463846e-01
4.65580463e-01 9.19611812e-01 1.13484824e+00 4.85242039e-01
2.38139972e-01 7.09237218e-01 3.80963027e-01 -1.52502164e-01
4.78724509e-01 -3.86554003e-01 5.37122488e-01 9.81889904e-01
-1.96426660e-01 -2.70883515e-02 -3.22404921e-01 5.18985391e-01
-2.17394686e+00 -9.15555358e-01 -5.82530387e-02 2.97568464e+00
9.70112145e-01 -1.57802086e-02 3.79905939e-01 4.51373518e-01
7.22898722e-01 -6.14327006e-02 -3.23328406e-01 -3.18220258e-01
3.88817281e-01 8.39041471e-02 2.56173104e-01 4.40842867e-01
-9.77236032e-01 3.63851368e-01 6.34502888e+00 9.24076617e-01
-1.05572915e+00 -1.67943612e-02 7.62274384e-01 -3.44094247e-01
-3.60893697e-01 -2.72062957e-01 -6.18361413e-01 4.72040027e-01
7.07890570e-01 -1.48302510e-01 3.22368234e-01 9.91908669e-01
2.05564305e-01 -8.91196057e-02 -1.25662792e+00 1.08010995e+00
-1.05025820e-01 -7.12173522e-01 2.72477809e-02 3.08852971e-01
7.88661420e-01 -2.82862157e-01 2.71565855e-01 6.20591223e-01
5.72540984e-03 -9.04990315e-01 6.14517987e-01 2.42413014e-01
8.45750391e-01 -1.00272024e+00 9.74772036e-01 5.91026604e-01
-9.98036325e-01 -6.63624927e-02 -5.02854228e-01 -4.50458735e-01
-1.29322305e-01 1.01726258e+00 -6.17582738e-01 5.55201828e-01
7.08427370e-01 7.47469485e-01 -3.61904114e-01 1.15369439e+00
-1.29079238e-01 4.16563749e-01 -1.62749022e-01 -2.68305391e-01
7.09631890e-02 -6.07899249e-01 2.41861254e-01 8.97517860e-01
3.11076552e-01 -2.48069808e-01 -8.89121555e-03 4.82474983e-01
-2.04599932e-01 4.61625159e-01 -7.07328558e-01 1.84421614e-01
3.99164110e-01 1.28041780e+00 -5.04392445e-01 -3.73343229e-01
-4.05625165e-01 6.77341700e-01 8.43386650e-01 4.11147892e-01
-5.57117343e-01 -2.49941483e-01 2.95098841e-01 1.98368743e-01
2.20841929e-01 1.11972671e-02 -2.28248149e-01 -1.05603850e+00
4.17279184e-01 -8.61738026e-01 4.06444728e-01 -1.95175767e-01
-1.75632203e+00 1.98959172e-01 3.56189534e-02 -1.51091743e+00
-3.96956533e-01 -6.31922960e-01 -3.05389583e-01 9.24056590e-01
-1.69276607e+00 -4.74408686e-01 -1.14750236e-01 3.85022819e-01
3.29928219e-01 3.55814487e-01 7.79445052e-01 7.19650328e-01
-2.89149404e-01 8.39951932e-01 5.35452783e-01 -9.26577151e-02
8.64774466e-01 -1.68631339e+00 -2.73312271e-01 3.38393688e-01
4.84545678e-02 1.21798146e+00 3.88031244e-01 -3.34967881e-01
-1.08700168e+00 -8.25356185e-01 1.08048379e+00 -5.39789319e-01
5.66752851e-01 -3.88370842e-01 -9.41681087e-01 1.74600929e-01
-1.43634200e-01 -1.38727471e-01 9.33647275e-01 8.49631548e-01
-3.77082080e-01 -3.11886400e-01 -9.46167707e-01 6.00377679e-01
8.08063030e-01 -5.54194093e-01 -4.43510503e-01 1.31931946e-01
1.62512273e-01 -1.20134978e-02 -7.93543100e-01 6.12142801e-01
1.02028108e+00 -1.18898880e+00 9.45789039e-01 -3.48297715e-01
4.27171618e-01 -3.41638237e-01 4.32741866e-02 -1.31207275e+00
-2.88831651e-01 -3.56153011e-01 -3.08493048e-01 1.31852937e+00
4.98102099e-01 -1.04825294e+00 9.52618897e-01 4.35775727e-01
4.42036361e-01 -1.28853011e+00 -6.65933549e-01 -8.62394452e-01
3.78062297e-03 -3.31422649e-02 2.74996191e-01 9.40295398e-01
-8.86146724e-03 5.43566406e-01 -6.59993768e-01 -1.87831059e-01
4.50317264e-01 2.38615781e-01 7.20935941e-01 -1.54283535e+00
-5.41348577e-01 -4.00676697e-01 -8.08932781e-02 -1.47880530e+00
-2.38302469e-01 -6.17659867e-01 2.87957042e-01 -1.26576209e+00
6.10822380e-01 -8.60696673e-01 -9.53405976e-01 9.68705565e-02
-5.61017454e-01 4.05365944e-01 1.69696081e-02 3.29938680e-01
-8.04069281e-01 6.18170261e-01 9.63549495e-01 1.98781565e-01
-3.64146024e-01 4.00331378e-01 -1.11871552e+00 5.41344643e-01
6.84187829e-01 -4.87841308e-01 -8.16700220e-01 -1.71986088e-01
7.54991233e-01 -9.98604819e-02 3.42156708e-01 -5.89220941e-01
2.64179688e-02 -6.40375866e-03 3.85412574e-01 -7.09451020e-01
2.33448625e-01 -6.98330522e-01 -2.98739493e-01 2.07612500e-01
-5.97394228e-01 5.59646077e-02 -2.58222640e-01 9.09893274e-01
-4.60817099e-01 -4.12988394e-01 4.54154223e-01 -7.45727420e-02
-4.19932157e-01 9.75167602e-02 -5.33061139e-02 -1.45526856e-01
7.34909892e-01 -1.86676368e-01 -2.48715267e-01 -4.58795846e-01
-3.89401734e-01 2.01874763e-01 3.95308375e-01 2.75859475e-01
4.33808923e-01 -1.38289607e+00 -4.86103594e-01 3.18444334e-02
1.63475990e-01 -2.83488743e-02 1.00380434e-02 1.17747951e+00
5.58943115e-02 4.86044794e-01 2.30233550e-01 -4.94156331e-01
-1.30328047e+00 3.58598411e-01 1.35917172e-01 -9.50852394e-01
-1.48423597e-01 6.78215742e-01 5.59797764e-01 -7.59585619e-01
3.77322853e-01 4.59703729e-02 -4.76581365e-01 3.00454229e-01
5.26970029e-01 4.34690595e-01 2.69095153e-01 -1.91904709e-01
-1.74426034e-01 3.85697186e-01 -3.84961456e-01 -1.62502334e-01
1.18789518e+00 -2.74059363e-02 -1.66216865e-01 7.30508626e-01
1.26440179e+00 1.75820664e-01 -1.02136612e+00 -3.76808196e-01
2.51451910e-01 -7.82493114e-01 -3.99529375e-02 -8.13217700e-01
-8.28079402e-01 7.85296738e-01 4.37074006e-01 5.10402918e-01
1.08287621e+00 -3.62959981e-01 5.96957147e-01 5.69432914e-01
5.06316662e-01 -1.10268497e+00 1.68322727e-01 4.24116611e-01
6.02147996e-01 -1.39207125e+00 1.02027819e-01 -4.10217881e-01
-4.50734586e-01 8.96730721e-01 6.03773415e-01 -2.38824069e-01
4.24985945e-01 -3.44861984e-01 8.51180851e-02 -1.39905766e-01
-8.44018936e-01 -3.98327447e-02 4.20881569e-01 3.46389741e-01
8.42292607e-01 1.15028240e-01 -7.29330122e-01 3.96927416e-01
-2.71926308e-03 3.73254940e-02 7.40308315e-02 7.53194094e-01
-3.28659058e-01 -1.53379571e+00 -1.54071063e-01 8.24111581e-01
-4.67716396e-01 -2.06989750e-01 -4.19765115e-01 6.66468620e-01
-1.80479199e-01 9.65814471e-01 -1.57923684e-01 -3.60396862e-01
2.92770058e-01 3.71064581e-02 4.48413551e-01 -7.43305981e-01
-5.20833910e-01 -1.32798627e-01 -1.69513524e-02 -3.07091385e-01
-5.66407323e-01 -7.01743066e-01 -7.71632791e-01 -1.88929796e-01
-6.80097044e-01 6.11263335e-01 5.23437262e-01 7.98698664e-01
1.99467197e-01 1.40187129e-01 9.65185225e-01 -7.36320615e-01
-1.17798066e+00 -8.45181465e-01 -6.50150537e-01 4.70586091e-01
3.00274193e-01 -8.62583578e-01 -4.48561549e-01 -5.33622980e-01] | [9.95772933959961, 5.521844863891602] |
84271eff-c988-49fa-b285-75c000aaabcb | when-cnn-meet-with-vit-towards-semi | 2208.06449 | null | https://arxiv.org/abs/2208.06449v1 | https://arxiv.org/pdf/2208.06449v1.pdf | When CNN Meet with ViT: Towards Semi-Supervised Learning for Multi-Class Medical Image Semantic Segmentation | Due to the lack of quality annotation in medical imaging community, semi-supervised learning methods are highly valued in image semantic segmentation tasks. In this paper, an advanced consistency-aware pseudo-label-based self-ensembling approach is presented to fully utilize the power of Vision Transformer(ViT) and Convolutional Neural Network(CNN) in semi-supervised learning. Our proposed framework consists of a feature-learning module which is enhanced by ViT and CNN mutually, and a guidance module which is robust for consistency-aware purposes. The pseudo labels are inferred and utilized recurrently and separately by views of CNN and ViT in the feature-learning module to expand the data set and are beneficial to each other. Meanwhile, a perturbation scheme is designed for the feature-learning module, and averaging network weight is utilized to develop the guidance module. By doing so, the framework combines the feature-learning strength of CNN and ViT, strengthens the performance via dual-view co-training, and enables consistency-aware supervision in a semi-supervised manner. A topological exploration of all alternative supervision modes with CNN and ViT are detailed validated, demonstrating the most promising performance and specific setting of our method on semi-supervised medical image segmentation tasks. Experimental results show that the proposed method achieves state-of-the-art performance on a public benchmark data set with a variety of metrics. The code is publicly available. | ['Baoru Huang', 'Jian-Qing Zheng', 'Tianze Li', 'Ziyang Wang'] | 2022-08-12 | null | null | null | null | ['semi-supervised-medical-image-segmentation'] | ['computer-vision'] | [ 3.19699228e-01 4.02934104e-01 -4.49525297e-01 -5.25987267e-01
-5.58649361e-01 -2.49255344e-01 3.33205193e-01 -2.24613637e-01
-1.60954103e-01 5.56764543e-01 2.23868892e-01 -4.45745178e-02
-2.48404875e-01 -5.20251691e-01 -6.63325787e-01 -9.79383051e-01
7.65451938e-02 2.87536800e-01 6.13707066e-01 -7.58374482e-02
-1.97530016e-01 2.38879308e-01 -1.20662344e+00 2.62653530e-01
1.07321048e+00 1.28890622e+00 4.06003445e-01 7.39897089e-03
-9.68991816e-02 9.13732290e-01 1.54029593e-01 -2.33831361e-01
1.05019212e-01 -2.71321714e-01 -1.05688763e+00 5.00912547e-01
2.69709080e-01 -2.28400052e-01 -2.84767896e-01 1.14293969e+00
4.50971752e-01 -2.02692211e-01 2.88689315e-01 -9.84146118e-01
-5.58371961e-01 6.34635568e-01 -5.46730518e-01 1.82352424e-01
-1.99037790e-01 1.75455377e-01 9.39990163e-01 -8.64772081e-01
7.67008305e-01 7.35285759e-01 7.35522509e-01 4.64769095e-01
-1.10108221e+00 -5.98571658e-01 9.79310572e-02 1.40706450e-01
-1.15133655e+00 -3.02539021e-01 1.34384906e+00 -2.85977960e-01
5.28352618e-01 -6.43053055e-02 8.38494837e-01 9.05627549e-01
1.13631874e-01 9.82041180e-01 1.22677946e+00 -1.45032018e-01
6.10096343e-02 2.84589916e-01 1.59558430e-01 1.23579252e+00
-6.57200906e-03 4.81488071e-02 -5.45613945e-01 2.88404487e-02
9.27966893e-01 2.67102242e-01 -2.44498000e-01 -9.06968236e-01
-1.38136911e+00 5.74797928e-01 9.24390256e-01 3.22203040e-01
-3.74675721e-01 -6.34633601e-02 6.27713799e-01 -2.09187027e-02
5.78884125e-01 1.62927657e-01 -4.25111085e-01 2.08156675e-01
-1.07991076e+00 -3.48707527e-01 4.31487113e-01 1.03671372e+00
7.20810950e-01 -6.66941628e-02 -2.68066913e-01 9.41418529e-01
4.07575876e-01 1.39974564e-01 6.28634453e-01 -6.95789278e-01
2.28341162e-01 1.09065330e+00 -4.83884692e-01 -8.83223951e-01
-6.45902991e-01 -9.68899727e-01 -1.27683926e+00 -3.80627625e-02
2.13742599e-01 2.52298981e-01 -9.73166287e-01 1.73331678e+00
6.84032500e-01 4.06217545e-01 -1.69559404e-01 8.00950646e-01
1.02371025e+00 -5.02542369e-02 -1.20960087e-01 -4.72138613e-01
1.50176346e+00 -1.46440125e+00 -7.91613162e-01 7.94959366e-02
6.71955884e-01 -3.65705818e-01 9.83799577e-01 7.53276497e-02
-1.01185155e+00 -7.15314448e-01 -1.12632096e+00 -5.27083175e-03
-3.84324975e-02 2.65902191e-01 6.00449204e-01 2.76329577e-01
-9.30799067e-01 6.22200608e-01 -1.12007129e+00 -1.27068594e-01
1.02160251e+00 2.32738957e-01 -3.97353470e-01 -1.72568321e-01
-1.00485075e+00 6.36967719e-01 3.52728933e-01 3.93925995e-01
-8.89274716e-01 -6.90387189e-01 -8.68912280e-01 -2.81379759e-01
4.01470214e-01 -8.24540675e-01 7.76973009e-01 -9.90841866e-01
-1.24568522e+00 9.77989376e-01 -5.54630645e-02 -3.94898444e-01
7.75534630e-01 1.81397393e-01 -3.40534300e-01 6.17025018e-01
3.86712611e-01 7.25903928e-01 9.57101047e-01 -1.25860727e+00
-5.36417067e-01 -5.71111441e-01 -1.62612259e-01 3.41167092e-01
-5.62986791e-01 -4.21970129e-01 -7.59434462e-01 -6.84866011e-01
5.30004263e-01 -9.12234604e-01 -3.21992517e-01 3.12095314e-01
-5.72635889e-01 -1.24427140e-01 9.11631644e-01 -5.86433589e-01
1.17510438e+00 -2.12481451e+00 2.03833759e-01 3.26240003e-01
4.80054885e-01 2.06286043e-01 2.12522686e-01 -1.65591866e-01
-1.30059198e-01 -1.67107806e-01 -5.96386552e-01 -5.02959549e-01
-5.36352158e-01 4.79253083e-01 1.33772761e-01 6.60529792e-01
2.95752168e-01 1.28557885e+00 -1.15492213e+00 -1.11435020e+00
3.76399845e-01 5.08930683e-01 -3.36704373e-01 1.94650128e-01
9.28480700e-02 9.96012628e-01 -6.69155538e-01 9.21529949e-01
6.26662672e-01 -8.72024596e-01 1.82450414e-01 -6.59631848e-01
1.79758728e-01 1.17730699e-01 -1.09396851e+00 2.22343445e+00
-3.49250466e-01 -7.77630229e-03 -1.12324599e-02 -1.26421356e+00
7.48226404e-01 2.70354539e-01 7.52145350e-01 -6.47480369e-01
-4.62652668e-02 3.33612055e-01 -3.50749820e-01 -6.71718359e-01
-4.01741527e-02 1.69875231e-02 3.79254490e-01 3.22782546e-01
4.45920944e-01 9.13789496e-02 1.75926566e-01 3.84789377e-01
7.93276131e-01 5.29012620e-01 1.57635957e-01 -4.86369848e-01
7.49188185e-01 2.33106744e-02 7.49013066e-01 3.62016201e-01
-5.02846122e-01 6.12512648e-01 2.96695411e-01 -3.94731730e-01
-9.24420059e-01 -8.64274502e-01 -6.59932733e-01 7.93786108e-01
5.01000464e-01 -1.78865775e-01 -5.87981403e-01 -1.22645843e+00
-1.94847316e-01 1.27494961e-01 -1.02344370e+00 -3.22388560e-01
-4.67204034e-01 -6.59316003e-01 3.84839326e-01 7.53361404e-01
9.38413322e-01 -9.65399086e-01 -4.76844132e-01 1.99210763e-01
-3.97458315e-01 -1.10535610e+00 -5.04143596e-01 3.82423431e-01
-1.27044857e+00 -1.19819915e+00 -8.29580486e-01 -1.00981021e+00
1.09389460e+00 3.82721841e-01 9.16106105e-01 2.24704936e-01
-1.52534485e-01 2.26431847e-01 -4.33379471e-01 1.68875549e-02
-3.03612262e-01 7.98922032e-02 -2.01595306e-01 3.13604325e-01
-2.56851256e-01 -8.03584814e-01 -8.35928261e-01 5.52160263e-01
-8.99367571e-01 4.66419518e-01 7.28109539e-01 1.38547409e+00
8.61131310e-01 -2.10797340e-01 6.26995623e-01 -1.04360759e+00
-1.71097934e-01 -4.51191545e-01 -3.00418288e-01 4.40258265e-01
-1.03409052e+00 4.74910326e-02 5.43916523e-01 -2.26821214e-01
-1.17953777e+00 4.03842419e-01 -2.77579110e-02 -4.65253532e-01
1.44212488e-02 4.12237406e-01 -1.09577045e-01 -3.90686065e-01
4.06729043e-01 5.10958731e-01 4.27231699e-01 -3.66142213e-01
4.34174448e-01 4.22484308e-01 6.56768739e-01 -2.96165407e-01
6.82876468e-01 9.22412217e-01 4.39838767e-02 -3.36968929e-01
-1.25433898e+00 -6.42610610e-01 -8.75047803e-01 -3.85649234e-01
8.07074010e-01 -9.96787965e-01 -3.11673433e-01 6.25555277e-01
-6.22484505e-01 -1.47889256e-02 -4.80235368e-01 3.50774765e-01
-5.53948402e-01 5.63893378e-01 -6.59354925e-01 -2.28640229e-01
-5.47314107e-01 -1.42093575e+00 1.13422072e+00 4.19514120e-01
3.98129344e-01 -1.20852220e+00 -1.81992024e-01 6.66128755e-01
3.85177433e-01 2.11066380e-01 5.64959466e-01 -6.99538767e-01
-4.80096072e-01 3.07384301e-02 -5.18165290e-01 6.34437919e-01
3.16033512e-01 -4.11731482e-01 -1.00084949e+00 -3.02099168e-01
-1.15862750e-02 -5.93865752e-01 1.07962525e+00 5.66234827e-01
1.37529993e+00 -1.08677097e-01 -5.61884880e-01 9.14842784e-01
1.33510947e+00 -3.29896927e-01 3.65586102e-01 2.86343426e-01
1.15134120e+00 5.30248821e-01 4.75810319e-01 2.27677837e-01
7.33473420e-01 4.70389605e-01 6.47238791e-01 -5.57239771e-01
-4.18218374e-01 -2.25974724e-01 -1.04485497e-01 1.03485572e+00
-1.50077185e-02 5.53094566e-01 -7.37431288e-01 6.09107494e-01
-2.03276324e+00 -6.79513752e-01 -2.48113219e-02 1.74632287e+00
1.19726741e+00 1.31444618e-01 -5.69019876e-02 1.10525429e-01
5.94101906e-01 1.63242489e-01 -7.89106011e-01 4.07663167e-01
-1.74223989e-01 3.03975325e-02 4.31025505e-01 -1.41840689e-02
-1.37606502e+00 7.17825592e-01 5.30262709e+00 9.69999492e-01
-1.12083888e+00 3.75518173e-01 9.12460327e-01 3.26038182e-01
-2.86615431e-01 -6.04054220e-02 -4.57044125e-01 3.67920309e-01
2.30596691e-01 2.59796649e-01 5.94646260e-02 8.78408968e-01
2.05083340e-02 -6.63970013e-06 -9.75758076e-01 8.02106023e-01
4.92820367e-02 -1.72216082e+00 5.33555001e-02 -5.20533361e-02
8.34562182e-01 1.44924074e-01 -1.99666359e-02 -5.15920706e-02
2.28993362e-03 -6.98815525e-01 5.42856038e-01 4.93388385e-01
9.52112198e-01 -6.01901293e-01 8.76665652e-01 2.61028349e-01
-1.33938634e+00 -6.76026046e-02 -1.25712633e-01 4.92129415e-01
2.17052788e-01 8.96439672e-01 -4.99072522e-01 1.16135180e+00
9.09510016e-01 1.42553115e+00 -7.48378158e-01 9.00431693e-01
-2.33584285e-01 7.83614337e-01 -1.21276386e-01 4.07133192e-01
3.67380261e-01 -7.20638558e-02 4.95452285e-01 1.08603120e+00
-2.92547613e-01 -9.26155448e-02 4.06469107e-01 8.42037082e-01
2.78442539e-02 8.37810710e-02 -3.39620769e-01 3.58911186e-01
2.36911505e-01 1.79086721e+00 -1.02302814e+00 -4.62639362e-01
-3.26923698e-01 8.68963361e-01 4.37252551e-01 2.15570375e-01
-8.49134564e-01 -8.61654207e-02 -5.37062809e-02 -9.04529765e-02
3.38329345e-01 1.29177004e-01 -5.95058978e-01 -1.23314393e+00
2.84985006e-01 -5.22886157e-01 5.01320958e-01 -6.40706062e-01
-1.33931506e+00 7.11860299e-01 -1.01823851e-01 -1.59015882e+00
1.16891712e-01 -2.63241172e-01 -4.76433128e-01 5.56973815e-01
-1.97153020e+00 -1.68234706e+00 -4.83323634e-01 7.75791347e-01
4.01867092e-01 -1.66812003e-01 6.38009906e-01 3.48745614e-01
-6.31506085e-01 5.57490587e-01 -1.68371219e-02 2.06939727e-01
6.02476537e-01 -1.22119212e+00 -2.28878409e-01 7.38355458e-01
2.08804421e-02 5.93820572e-01 5.61209470e-02 -6.03777111e-01
-1.18673790e+00 -1.38574481e+00 4.60642368e-01 -1.56709760e-01
5.54920912e-01 -1.07425705e-01 -9.14535701e-01 4.11734253e-01
-2.90159974e-02 7.31988370e-01 5.94960630e-01 -1.57649666e-01
-3.74798506e-01 -3.03104311e-01 -1.23188114e+00 2.74205059e-01
1.19095993e+00 -5.26429057e-01 -5.28572679e-01 5.14677107e-01
1.00842214e+00 -5.35561204e-01 -1.06556547e+00 6.96130097e-01
4.43135500e-01 -9.60930705e-01 1.00307262e+00 -3.78382564e-01
5.19598961e-01 -5.35713553e-01 1.45126924e-01 -1.02105010e+00
-4.19900298e-01 -3.13506693e-01 -1.62213445e-01 1.23965406e+00
3.39411229e-01 -5.99745810e-01 6.49762690e-01 1.00983106e-01
-5.75891316e-01 -1.37151647e+00 -1.14634836e+00 -4.93013740e-01
-2.26728499e-01 -2.23832026e-01 1.97775483e-01 1.24916053e+00
-4.89582047e-02 2.25517511e-01 -2.82264352e-01 2.85116918e-02
9.08697128e-01 1.68494940e-01 1.64875120e-01 -1.07920349e+00
-2.63766557e-01 -2.58183122e-01 -2.87215739e-01 -9.57297444e-01
-3.74624915e-02 -1.36272931e+00 -1.35946006e-01 -1.53833365e+00
6.33141696e-01 -6.81629837e-01 -6.08623326e-01 7.92166233e-01
-2.63513356e-01 4.63768512e-01 -1.42693520e-01 5.40418804e-01
-9.11462486e-01 7.28701949e-01 1.76236928e+00 -2.10161284e-01
-3.49163301e-02 -1.06381424e-01 -7.11024523e-01 7.16092110e-01
4.86035019e-01 -3.44120592e-01 -5.28744817e-01 -2.17782781e-01
-4.79115918e-02 -9.18539762e-02 6.29867375e-01 -8.53325784e-01
3.71966243e-01 2.49085814e-01 4.79402333e-01 -5.85813344e-01
-6.84523806e-02 -8.69697511e-01 -2.17516214e-01 5.73941469e-01
-4.05300170e-01 -3.35362613e-01 -1.65764213e-01 7.86052346e-01
-3.39747548e-01 1.61573887e-01 9.18745279e-01 -1.58693224e-01
-6.70995355e-01 7.02697873e-01 4.85223442e-01 2.51829568e-02
9.16734874e-01 -3.54816407e-01 -4.81044129e-02 1.04084656e-01
-8.38640511e-01 6.98739290e-01 1.84359074e-01 2.98462451e-01
7.09100723e-01 -1.47780192e+00 -5.11012137e-01 2.96658695e-01
4.58173722e-01 5.64072967e-01 5.67768574e-01 1.59371340e+00
-1.59779951e-01 6.90376945e-03 -2.76495367e-01 -1.20510483e+00
-8.78744125e-01 4.08512443e-01 4.05709058e-01 -6.38906360e-01
-8.86632025e-01 9.14151371e-01 2.39029020e-01 -5.92794776e-01
3.10414046e-01 -2.65012383e-01 -3.79160762e-01 -1.40347019e-01
3.99606019e-01 8.89586136e-02 9.19097885e-02 -6.87895596e-01
-6.04163945e-01 5.83284318e-01 -1.37184098e-01 1.81848481e-01
1.35626662e+00 -3.71732354e-01 -2.38241404e-01 4.24438268e-01
1.31129241e+00 -4.82074559e-01 -1.49119389e+00 -7.86279798e-01
-2.46552452e-01 -2.62156557e-02 2.67813057e-01 -8.21210384e-01
-1.50841188e+00 7.17076898e-01 7.72244036e-01 -2.41006970e-01
1.29336321e+00 2.25722462e-01 8.51814866e-01 5.63653093e-03
3.17627400e-01 -1.16900170e+00 3.28822851e-01 1.84404686e-01
6.10477149e-01 -1.49927866e+00 -3.64208631e-02 -4.95950818e-01
-8.69241536e-01 1.00293005e+00 6.02834642e-01 -1.07168950e-01
9.62025464e-01 2.14500040e-01 -5.46941124e-02 -4.17548329e-01
-4.60285068e-01 -1.93265676e-01 5.35548270e-01 5.51695645e-01
2.55954713e-01 8.04357044e-03 -1.31286398e-01 6.19800270e-01
2.45444641e-01 1.83991000e-01 4.32008617e-02 9.50219452e-01
-1.87505603e-01 -7.54844129e-01 1.06242821e-01 4.90565777e-01
-2.35219523e-01 -1.17223248e-01 1.47505030e-01 4.88444209e-01
3.15118223e-01 6.54477537e-01 -1.29337341e-01 -4.56295699e-01
4.54718396e-02 -1.19246885e-01 3.17524254e-01 -4.36162025e-01
-6.61307931e-01 3.31677824e-01 -9.74017084e-02 -8.88835788e-01
-8.81422520e-01 -4.87901747e-01 -1.49002981e+00 3.53055984e-01
-6.69294357e-01 -8.30362737e-02 5.21620393e-01 1.13874066e+00
3.70992869e-01 8.06067288e-01 9.61836040e-01 -5.83016932e-01
-6.26600504e-01 -8.35964441e-01 -5.25975883e-01 5.51255465e-01
3.01899135e-01 -8.79503191e-01 -4.85533997e-02 2.42555469e-01] | [14.733248710632324, -2.1076555252075195] |
032708b0-4d9a-4c97-87da-d351951c3d09 | multilinear-class-specific-discriminant | 1710.10695 | null | http://arxiv.org/abs/1710.10695v1 | http://arxiv.org/pdf/1710.10695v1.pdf | Multilinear Class-Specific Discriminant Analysis | There has been a great effort to transfer linear discriminant techniques that
operate on vector data to high-order data, generally referred to as Multilinear
Discriminant Analysis (MDA) techniques. Many existing works focus on maximizing
the inter-class variances to intra-class variances defined on tensor data
representations. However, there has not been any attempt to employ
class-specific discrimination criteria for the tensor data. In this paper, we
propose a multilinear subspace learning technique suitable for applications
requiring class-specific tensor models. The method maximizes the discrimination
of each individual class in the feature space while retains the spatial
structure of the input. We evaluate the efficiency of the proposed method on
two problems, i.e. facial image analysis and stock price prediction based on
limit order book data. | ['Moncef Gabbouj', 'Alexandros Iosifidis', 'Dat Thanh Tran'] | 2017-10-29 | null | null | null | null | ['stock-price-prediction'] | ['time-series'] | [-2.89954394e-01 -6.81711257e-01 -1.85119182e-01 -4.76546347e-01
-4.46153224e-01 -6.33043706e-01 4.13743794e-01 -4.03008938e-01
-1.31645858e-01 2.57008344e-01 -1.33357763e-01 -2.11222708e-01
-5.20272374e-01 -3.47976178e-01 2.58679036e-02 -8.16426516e-01
-1.03134409e-01 3.79109114e-01 -1.36655927e-01 -8.90026614e-02
6.18835747e-01 8.63808036e-01 -1.29883230e+00 3.00988764e-01
6.03217006e-01 1.24333596e+00 -2.45469630e-01 3.00462097e-01
-6.53911307e-02 5.36620915e-01 -1.73209950e-01 -3.99142593e-01
5.92726171e-01 -2.43818566e-01 -2.79063255e-01 4.15143996e-01
5.33438206e-01 -1.16152197e-01 -2.17285454e-01 1.02320111e+00
1.42550394e-01 4.26074974e-02 1.06577873e+00 -1.47519398e+00
-8.97802413e-01 1.37070552e-01 -7.37621725e-01 3.23477387e-01
-8.28160122e-02 -4.39307928e-01 1.22605014e+00 -1.16959035e+00
2.95585841e-01 1.26984489e+00 2.08062053e-01 -3.96791361e-02
-1.49056947e+00 -5.88943958e-01 -8.94655064e-02 4.23658878e-01
-1.47078168e+00 -1.15056753e-01 1.47258925e+00 -9.40149248e-01
4.98099148e-01 4.57432061e-01 5.00921488e-01 5.09758055e-01
4.94750232e-01 6.91193640e-01 1.58001459e+00 -3.52003157e-01
4.61642891e-02 3.01103115e-01 5.18559694e-01 4.73270148e-01
2.10578144e-01 1.20217763e-01 -4.82598245e-01 -2.31162205e-01
7.68823385e-01 -4.20263372e-02 1.58026546e-01 -1.01457250e+00
-1.10149240e+00 1.02768362e+00 6.72754869e-02 5.92538178e-01
-3.93743306e-01 -4.24321383e-01 1.95240900e-01 2.96637863e-01
6.13584518e-01 2.60577917e-01 -3.88964087e-01 1.45370826e-01
-1.05955887e+00 2.24700361e-01 5.89208543e-01 6.62785470e-01
7.21788466e-01 4.52834100e-01 3.83443177e-01 8.54901195e-01
4.02834028e-01 5.64157605e-01 6.19670272e-01 -6.26733661e-01
3.47567648e-01 6.84105396e-01 -1.88758671e-01 -1.59387493e+00
-4.60165083e-01 -3.33178788e-01 -5.98799050e-01 2.85378039e-01
3.54260623e-01 3.74866575e-02 -4.41370785e-01 1.25642645e+00
2.27513656e-01 -2.00950056e-01 1.01504855e-01 1.07007360e+00
2.58778900e-01 4.78731453e-01 -3.32141697e-01 -3.49169254e-01
1.07461965e+00 -4.94972140e-01 -6.42780542e-01 4.74126726e-01
5.01968682e-01 -1.07702816e+00 6.24457777e-01 5.29175520e-01
-7.19974458e-01 -2.94269145e-01 -9.51205194e-01 4.48130220e-02
-2.76931077e-01 4.86333102e-01 7.13693857e-01 9.14890587e-01
-5.45202315e-01 3.04000556e-01 -8.23374450e-01 8.79200399e-02
-1.23994194e-01 4.57443237e-01 -4.45159167e-01 2.67802984e-01
-7.31176972e-01 7.05356002e-01 1.00917548e-01 1.91804081e-01
-5.19959450e-01 -4.07882214e-01 -4.93336648e-01 -2.43247330e-01
4.65027355e-02 1.35993743e-02 6.92764163e-01 -9.13000762e-01
-1.31032801e+00 5.81540287e-01 -9.65893194e-02 9.51261222e-02
2.75451779e-01 1.86466143e-01 -4.65744227e-01 -4.08579484e-02
1.69790927e-02 2.88573682e-01 1.15032601e+00 -9.58112538e-01
-2.94918597e-01 -1.00553203e+00 -5.92934489e-02 2.14100286e-01
-6.77909255e-01 2.80564338e-01 1.11809652e-02 -8.95234942e-01
8.68644357e-01 -1.26253331e+00 1.37100101e-01 -2.40458593e-01
-2.47497141e-01 -3.32333922e-01 1.05106938e+00 -8.77019167e-01
9.86104548e-01 -2.35305858e+00 5.74407339e-01 3.84153336e-01
2.13703769e-03 -5.64992577e-02 -5.22324145e-02 2.90469110e-01
-3.67572278e-01 -1.35241017e-01 4.72557638e-03 3.02878092e-03
2.87566707e-02 1.62173122e-01 -3.84557277e-01 6.23801947e-01
2.01138884e-01 3.50858420e-01 -2.70609200e-01 -5.52159607e-01
4.72209826e-02 4.35283989e-01 -4.70921189e-01 -6.94428012e-02
3.57633889e-01 2.06056759e-01 -6.39930665e-01 9.38316107e-01
9.62155402e-01 1.51280999e-01 1.10582791e-01 -6.94882870e-01
-2.60022551e-01 -2.02270642e-01 -1.43324816e+00 1.10365403e+00
-7.42298290e-02 4.94882911e-01 -3.26919854e-02 -1.28025138e+00
1.03244579e+00 1.96216986e-01 9.84336317e-01 -5.08228958e-01
7.33144209e-02 4.61745620e-01 2.08236203e-01 -3.76644403e-01
4.30407017e-01 -2.86722034e-01 9.55804810e-02 1.54016078e-01
-7.02117383e-02 3.00494909e-01 1.41075835e-01 -1.60581008e-01
2.22710654e-01 1.50733918e-01 3.64316371e-03 -5.56070685e-01
7.70401120e-01 -1.07691646e-01 5.12715697e-01 -2.85287857e-01
-1.97646227e-02 3.18105161e-01 6.24131441e-01 -5.32086968e-01
-1.04542994e+00 -9.27618861e-01 -7.12435663e-01 8.12980175e-01
-7.80382976e-02 -4.81364354e-02 -5.09008169e-01 -4.75752473e-01
3.64840209e-01 3.56421232e-01 -4.17690814e-01 2.05065131e-01
-2.55685598e-01 -7.37602174e-01 2.12528199e-01 3.70889664e-01
3.68155032e-01 -3.93484920e-01 -3.20855945e-01 -1.03990085e-01
2.17890650e-01 -9.83703315e-01 -5.00563264e-01 3.83720323e-02
-1.16977429e+00 -9.28177893e-01 -7.63946354e-01 -6.60765529e-01
7.34304011e-01 3.20360780e-01 3.53673786e-01 -5.12110591e-01
-1.99619517e-01 6.11956358e-01 -2.04765320e-01 -2.01917123e-02
2.65735805e-01 4.62812111e-02 4.56825703e-01 7.52038360e-01
5.54536462e-01 -3.66215438e-01 -1.40434787e-01 8.15599561e-01
-9.05510008e-01 -2.89731383e-01 6.24208033e-01 8.24097216e-01
6.06809914e-01 5.75464405e-02 2.89944410e-01 -3.01842779e-01
8.56418610e-01 -3.92466038e-01 -9.14306462e-01 3.47853273e-01
-8.03757846e-01 2.63155282e-01 3.14543724e-01 -6.13158107e-01
-7.07909048e-01 1.60635382e-01 4.29765791e-01 -7.19398379e-01
2.39598587e-01 8.48579347e-01 -1.53745010e-01 -5.52050114e-01
3.85373831e-01 3.34987879e-01 6.25037998e-02 -7.07027435e-01
-3.05234687e-03 7.99255311e-01 -1.78359542e-03 -7.39908338e-01
7.23257244e-01 2.40567520e-01 6.32491410e-01 -1.16973352e+00
-3.95676166e-01 -5.20263135e-01 -1.17950046e+00 -1.38228804e-01
8.09226573e-01 -5.61765969e-01 -5.34096599e-01 4.56691861e-01
-7.26028144e-01 4.94141757e-01 1.38543814e-01 9.29256439e-01
-5.07843912e-01 6.23027503e-01 -3.29769641e-01 -1.01496160e+00
3.14395428e-02 -1.28362644e+00 8.44183743e-01 -2.79695958e-01
2.33808398e-01 -9.37252045e-01 7.82399103e-02 4.66480672e-01
2.83997059e-01 9.84582976e-02 9.13787842e-01 -6.88482463e-01
-6.35673344e-01 -5.10699034e-01 -1.67319477e-01 6.02719426e-01
2.01071709e-01 7.26203341e-03 -6.48103416e-01 -2.96130717e-01
4.58293408e-02 -2.61374593e-01 5.72301328e-01 1.87357545e-01
8.80773008e-01 -1.52049497e-01 8.18024725e-02 5.59995353e-01
1.07251370e+00 4.22702521e-01 2.99860597e-01 2.47933701e-01
8.80023360e-01 8.01927149e-01 6.57942593e-01 3.16313088e-01
2.54885495e-01 1.06256068e+00 5.40107712e-02 2.54819304e-01
4.43003744e-01 1.12373628e-01 3.16487283e-01 1.03019321e+00
-2.98607171e-01 4.57866699e-01 -8.47086787e-01 2.28580326e-01
-1.70705283e+00 -7.68553376e-01 -1.76607311e-01 2.21868014e+00
3.46279025e-01 -3.36151183e-01 2.92088479e-01 2.48362646e-01
5.32247841e-01 7.09518641e-02 -4.97611105e-01 -4.82055962e-01
-3.18539917e-01 -3.68600488e-01 3.49917978e-01 4.17466313e-01
-1.23515534e+00 6.05951667e-01 6.28747797e+00 6.44955575e-01
-1.48437655e+00 -2.26608455e-01 4.23623085e-01 1.54422343e-01
-3.33021641e-01 7.28870258e-02 -7.55259573e-01 2.85566181e-01
8.27240407e-01 -2.51791507e-01 5.83412528e-01 9.04891491e-01
8.59951079e-02 2.28016078e-01 -9.12131906e-01 1.21096551e+00
2.87340015e-01 -7.84495652e-01 2.15398952e-01 5.61593831e-01
4.58056986e-01 -4.74732429e-01 6.11118317e-01 8.46537948e-02
-5.80026448e-01 -6.44082785e-01 5.44465244e-01 6.72718108e-01
4.36919212e-01 -7.57259786e-01 5.05540133e-01 2.94222832e-01
-1.00969434e+00 -2.48585325e-02 -5.64219415e-01 5.93982264e-02
-1.97720915e-01 4.56501484e-01 -5.63054740e-01 4.80834663e-01
4.39380527e-01 7.63326645e-01 -5.66276550e-01 8.42794359e-01
3.88753891e-01 3.23548406e-01 -3.18981439e-01 -2.83661205e-02
1.88674435e-01 -1.00017655e+00 5.67964017e-01 7.40720272e-01
5.02771080e-01 1.00235522e-01 3.01667303e-01 6.53882444e-01
4.86559242e-01 7.21964538e-01 -6.06796205e-01 -4.80102479e-01
-2.28004500e-01 1.27798724e+00 -5.98268688e-01 2.43650023e-02
-6.43957675e-01 7.14064598e-01 -1.06942616e-01 3.41077924e-01
-4.87895101e-01 -9.16789174e-02 6.27790391e-01 6.30650669e-02
3.12996149e-01 -7.65224338e-01 -4.66670513e-01 -1.56257546e+00
2.71211505e-01 -8.51401210e-01 4.08815563e-01 -6.00606799e-01
-1.38032472e+00 5.34799993e-01 3.09567124e-01 -1.52709043e+00
-3.85383844e-01 -1.14635611e+00 -1.13030329e-01 1.00977755e+00
-1.18681240e+00 -1.35628283e+00 2.64865130e-01 7.04996765e-01
2.23497748e-01 -8.28783870e-01 6.65707350e-01 4.54261243e-01
-7.12015688e-01 5.55462301e-01 3.42444509e-01 1.89276129e-01
4.80675697e-01 -1.07268393e+00 -5.28450489e-01 7.08418012e-01
2.41120487e-01 8.71545315e-01 4.90036219e-01 -5.10707736e-01
-1.60698760e+00 -5.40720582e-01 8.73273849e-01 -2.85333902e-01
9.99054134e-01 -1.38229638e-01 -7.44228184e-01 6.55782044e-01
-4.56412649e-03 6.33510426e-02 9.90104198e-01 2.42145985e-01
-5.90067685e-01 -4.43200767e-01 -1.13437545e+00 2.70213813e-01
3.92674118e-01 -7.19933450e-01 -3.81235421e-01 2.68428862e-01
-1.53240353e-01 4.53837849e-02 -1.23822439e+00 3.17845643e-01
8.42369914e-01 -6.84374213e-01 9.18997169e-01 -9.27774549e-01
3.78429443e-02 -5.38168073e-01 -7.36593485e-01 -1.14473307e+00
-5.38801789e-01 -1.86834231e-01 2.04438969e-01 1.00601149e+00
3.37499052e-01 -7.49951243e-01 8.10775936e-01 1.00783408e+00
8.49244967e-02 -6.57575309e-01 -1.10475004e+00 -9.33184445e-01
2.63788700e-01 -2.64542371e-01 3.35500687e-01 1.18533695e+00
1.14041269e-01 3.45504254e-01 -6.13184035e-01 3.37444544e-01
8.25755119e-01 5.41411281e-01 4.73792881e-01 -1.36750591e+00
-2.16470152e-01 -3.48104388e-01 -9.29356217e-01 -4.83088732e-01
3.41655821e-01 -9.76051509e-01 -6.34455860e-01 -7.75128603e-01
2.30868518e-01 -3.47185850e-01 -4.27862465e-01 3.24686497e-01
3.57758433e-01 1.31011695e-01 3.87770504e-01 6.50530875e-01
-1.30981371e-01 5.32012761e-01 1.17797184e+00 -3.40587914e-01
-1.05068818e-01 7.01282248e-02 -3.85492176e-01 4.32192534e-01
7.18232930e-01 -1.76105112e-01 -4.83100682e-01 -1.81687251e-01
-4.83700656e-04 2.21495911e-01 1.61389366e-01 -8.57258916e-01
-8.23938102e-02 -5.07489800e-01 4.48060036e-01 -6.31670058e-01
4.38175023e-01 -1.08432662e+00 4.19473976e-01 1.38475627e-01
-2.67124087e-01 4.21149164e-01 -5.32217212e-02 3.13986033e-01
-5.39211571e-01 -3.18971336e-01 6.93442106e-01 2.07877040e-01
-6.99620128e-01 1.75167888e-01 -1.93960562e-01 -5.20714760e-01
1.20278096e+00 -2.60784388e-01 5.31970449e-02 -1.53392762e-01
-7.90766835e-01 -1.59940809e-01 2.47693360e-01 4.77591485e-01
7.19985425e-01 -1.63505721e+00 -6.39713645e-01 4.91148770e-01
1.28342167e-01 -9.53436673e-01 8.80359784e-02 9.86287534e-01
-3.69775116e-01 9.66451824e-01 -6.15109622e-01 -7.05363154e-01
-1.37773097e+00 8.57994139e-01 4.36864972e-01 4.80410382e-02
-1.23744339e-01 3.29075933e-01 2.58756250e-01 -3.04530352e-01
-2.09224775e-01 -2.79016197e-01 -5.21071672e-01 5.33234537e-01
1.89327195e-01 5.89698017e-01 1.54400542e-01 -1.35228693e+00
-4.03144687e-01 1.10546756e+00 3.02709918e-02 -2.48383537e-01
1.22895420e+00 1.16767511e-01 -3.98012251e-01 6.93686962e-01
1.58091855e+00 -3.37809287e-02 -9.05709386e-01 2.30799560e-02
3.42450202e-01 -8.97314310e-01 4.98903573e-01 -3.50075722e-01
-1.17689145e+00 9.73465681e-01 9.49643791e-01 3.38905334e-01
9.97890413e-01 -4.69083995e-01 2.56101459e-01 3.87472451e-01
5.05635381e-01 -1.05175674e+00 8.58706832e-02 3.52116972e-01
1.14086699e+00 -1.06513572e+00 4.08741087e-02 -4.98218507e-01
-8.86966169e-01 1.35376370e+00 2.56059647e-01 -3.04137200e-01
1.03271651e+00 -1.03855141e-01 1.05887607e-01 -1.52706191e-01
-3.82282585e-01 1.27920419e-01 8.12746286e-01 3.92480522e-01
6.56733274e-01 3.15422088e-01 -6.77046537e-01 2.62045473e-01
-1.67157091e-02 -2.09020779e-01 3.23145688e-01 7.70617723e-01
-1.33243680e-01 -1.26835728e+00 -7.65057147e-01 4.40399677e-01
-4.23610449e-01 2.50324368e-01 -4.68381792e-01 6.56539798e-01
-1.66232243e-01 6.83034062e-01 -3.52239124e-02 -5.30900002e-01
3.88970256e-01 3.38267565e-01 4.22780752e-01 -1.90843225e-01
1.87924914e-02 5.46382844e-01 -1.53625309e-01 -3.17292392e-01
-4.95549768e-01 -1.26090944e+00 -8.32663655e-01 -1.67171121e-01
-2.00809449e-01 2.34037191e-01 9.60327148e-01 6.08916879e-01
1.52591497e-01 3.24831456e-02 9.89818394e-01 -8.07282150e-01
-8.49406302e-01 -7.80917764e-01 -1.23976898e+00 4.00483429e-01
1.99797764e-01 -1.04051065e+00 -4.45314199e-01 -6.56924024e-02] | [7.844665050506592, 4.219435214996338] |
d4283a9f-1edd-4e5a-b762-bcb1aa08fd99 | discriminative-local-sparse-representations | 1111.01947 | null | http://arxiv.org/abs/1111.1947v1 | http://arxiv.org/pdf/1111.1947v1.pdf | Discriminative Local Sparse Representations for Robust Face Recognition | A key recent advance in face recognition models a test face image as a sparse
linear combination of a set of training face images. The resulting sparse
representations have been shown to possess robustness against a variety of
distortions like random pixel corruption, occlusion and disguise. This approach
however makes the restrictive (in many scenarios) assumption that test faces
must be perfectly aligned (or registered) to the training data prior to
classification. In this paper, we propose a simple yet robust local block-based
sparsity model, using adaptively-constructed dictionaries from local features
in the training data, to overcome this misalignment problem. Our approach is
inspired by human perception: we analyze a series of local discriminative
features and combine them to arrive at the final classification decision. We
propose a probabilistic graphical model framework to explicitly mine the
conditional dependencies between these distinct sparse local features. In
particular, we learn discriminative graphs on sparse representations obtained
from distinct local slices of a face. Conditional correlations between these
sparse features are first discovered (in the training phase), and subsequently
exploited to bring about significant improvements in recognition rates.
Experimental results obtained on benchmark face databases demonstrate the
effectiveness of the proposed algorithms in the presence of multiple
registration errors (such as translation, rotation, and scaling) as well as
under variations of pose and illumination. | ['Trac. D. Tran', 'Vishal Monga', 'Umamahesh Srinivas', 'Yi Chen', 'Thong T. Do'] | 2011-11-08 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [ 6.15834773e-01 -2.84246147e-01 -2.93917984e-01 -7.50098884e-01
-4.91765827e-01 -3.83625180e-01 8.00697744e-01 -1.29904166e-01
8.05653557e-02 5.07919133e-01 2.55769700e-01 2.42718965e-01
-4.01805937e-01 -6.15460396e-01 -7.72140384e-01 -9.62931037e-01
-2.02385381e-01 2.16996491e-01 -1.93839461e-01 2.43079096e-01
3.34774286e-01 8.44577789e-01 -1.70943069e+00 2.12938607e-01
5.10457158e-01 9.06385422e-01 1.40437772e-02 2.65676320e-01
2.81153470e-01 5.79333425e-01 -3.97177905e-01 -1.12943538e-01
3.58440638e-01 -4.50364649e-01 -5.12713730e-01 7.30129242e-01
7.20712304e-01 -2.03889217e-02 -3.47303808e-01 1.27937508e+00
3.11216831e-01 -7.65366945e-03 6.08008981e-01 -9.05319214e-01
-4.63154763e-01 1.71225190e-01 -8.14172924e-01 1.57697782e-01
4.94089603e-01 -2.31847689e-01 8.27728987e-01 -1.25251412e+00
5.53705990e-01 1.17352211e+00 4.38559771e-01 2.89469570e-01
-1.48784542e+00 -4.81915325e-01 9.80564356e-02 2.58924246e-01
-1.69185388e+00 -1.09181440e+00 9.93728280e-01 -3.73593003e-01
6.35383725e-01 2.47430608e-01 2.05435351e-01 9.04936314e-01
1.58915952e-01 2.76316971e-01 1.28909552e+00 -6.20945334e-01
2.09930986e-01 1.33357123e-01 6.06854213e-03 9.57318366e-01
3.97877783e-01 7.26458803e-02 -8.46987724e-01 -4.23493177e-01
7.84860611e-01 1.19881302e-01 -5.41899681e-01 -7.64535964e-01
-9.77332950e-01 6.62075222e-01 3.23010921e-01 6.00398839e-01
-4.78846461e-01 -3.62758100e-01 -7.19244555e-02 2.24148795e-01
2.65261889e-01 -1.72914058e-01 -1.69363126e-01 4.77377385e-01
-1.02765119e+00 -2.35220969e-01 8.13223779e-01 5.37980378e-01
1.08962572e+00 2.46026397e-01 1.97011128e-01 9.20606196e-01
7.19819367e-01 4.70618248e-01 5.31754792e-01 -5.20586908e-01
3.76253903e-01 2.73154765e-01 -2.56365150e-01 -1.63678348e+00
-7.93883875e-02 -5.94142973e-01 -1.04104841e+00 8.47018957e-02
2.91311860e-01 3.09252024e-01 -8.51989686e-01 1.69959569e+00
4.26184446e-01 6.49433315e-01 -4.81159836e-02 6.81198835e-01
5.59842467e-01 3.95644546e-01 -2.13415623e-01 -5.18730640e-01
1.07010734e+00 -3.27319741e-01 -5.81846297e-01 -1.53022110e-01
2.55219996e-01 -8.22460234e-01 2.50379562e-01 4.62934196e-01
-6.83228135e-01 -6.18321419e-01 -1.06206453e+00 5.18013000e-01
7.87894875e-02 6.21120296e-02 3.09431911e-01 7.61394382e-01
-1.02524436e+00 4.71669614e-01 -7.26048112e-01 -2.66533226e-01
5.17642736e-01 6.55375659e-01 -9.41891193e-01 -7.22406864e-01
-5.37560046e-01 6.71601057e-01 -1.12139113e-01 5.41466951e-01
-9.05051649e-01 -2.25552246e-01 -1.10613966e+00 -8.85622054e-02
2.95625389e-01 -2.45753631e-01 5.16587675e-01 -1.12713349e+00
-1.26222789e+00 8.00053000e-01 -6.68819666e-01 -1.38436228e-01
-7.16209635e-02 -1.03073027e-02 -4.33147609e-01 1.88559175e-01
-3.11654676e-02 1.04740644e-02 1.39292705e+00 -1.42667079e+00
-2.70427763e-02 -8.63664925e-01 -2.37456828e-01 4.52889986e-02
-1.67047232e-01 8.66050124e-02 -3.04671228e-01 -5.91816604e-01
5.12737930e-01 -8.20794404e-01 -1.27929047e-01 -2.52195567e-01
-1.76225409e-01 8.61483887e-02 9.93514240e-01 -6.22360647e-01
6.81922853e-01 -2.14293385e+00 3.29812557e-01 8.15132797e-01
-4.12550475e-03 1.56376526e-01 -2.98420459e-01 2.26433635e-01
-4.08806443e-01 -2.14572325e-01 -3.98009539e-01 -4.10599798e-01
-4.07580882e-01 4.47671324e-01 -2.40401030e-01 1.12320006e+00
4.16555196e-01 5.05957723e-01 -6.50641918e-01 -3.68709564e-01
2.86864698e-01 8.36444080e-01 -5.17177343e-01 2.24614233e-01
2.96302944e-01 7.13953733e-01 -4.52511132e-01 7.43094683e-01
7.81556845e-01 -3.88142355e-02 5.21866977e-01 -3.03579271e-01
2.26017475e-01 2.54607141e-01 -1.49465334e+00 1.56576836e+00
-3.75393540e-01 3.07600766e-01 1.85492948e-01 -1.57347620e+00
1.04967678e+00 4.93482828e-01 5.81962943e-01 -4.74304378e-01
5.31797819e-02 -1.24359047e-02 -1.14101283e-02 -2.78935492e-01
-1.91348299e-01 -1.18668467e-01 3.30052584e-01 3.50257963e-01
5.40512621e-01 2.06147611e-01 -4.12768722e-02 -6.01760261e-02
9.47307348e-01 -2.35216282e-02 6.27058327e-01 -2.77244151e-01
9.30848956e-01 -7.76325285e-01 7.20530331e-01 3.07092756e-01
6.86748251e-02 6.53609157e-01 3.66712213e-01 -4.24089640e-01
-5.88411391e-01 -7.35213578e-01 -3.79168093e-01 6.88074827e-01
-8.84460136e-02 -4.66634244e-01 -6.24935627e-01 -8.15556765e-01
-7.01674074e-02 1.23330794e-01 -5.62843919e-01 -9.85454768e-02
-6.48225725e-01 -8.68874669e-01 4.21775803e-02 1.90535039e-01
3.98945928e-01 -7.81751513e-01 -2.37007737e-01 -3.42919901e-02
2.11424381e-01 -1.21263599e+00 -2.32750669e-01 4.61973436e-02
-8.52122664e-01 -1.12570941e+00 -3.90481353e-01 -8.67382646e-01
1.23502839e+00 5.50102174e-01 7.44631350e-01 2.22963378e-01
-4.22499359e-01 7.78002441e-01 -1.86006233e-01 2.74001420e-01
-8.95259157e-02 -6.47515416e-01 3.49633157e-01 8.90792549e-01
1.92689151e-01 -7.67679513e-01 -2.00930074e-01 4.21089113e-01
-9.76319313e-01 -4.39521879e-01 6.61329687e-01 1.10216403e+00
7.08196163e-01 2.70664215e-01 3.16895038e-01 -8.48250687e-01
6.08983971e-02 -5.92408419e-01 -5.35249054e-01 2.78324127e-01
-2.11855650e-01 9.43376720e-02 4.37987268e-01 -2.30125308e-01
-9.00157094e-01 3.81793082e-01 1.79738075e-01 -4.86878395e-01
-3.82902503e-01 7.12132215e-01 -6.02334082e-01 -6.67320848e-01
4.07267004e-01 5.29692173e-01 2.02103704e-02 -4.47976559e-01
1.24185205e-01 3.66006523e-01 4.73118037e-01 -6.55493915e-01
1.22552550e+00 5.45097888e-01 2.91545182e-01 -1.25240505e+00
-4.93895948e-01 -4.39783335e-01 -1.02624035e+00 -1.99631646e-01
3.00088286e-01 -9.09569681e-01 -4.49896663e-01 3.98386270e-01
-8.20652902e-01 2.55659789e-01 -1.44512542e-02 4.06047225e-01
-2.56387979e-01 7.07248211e-01 -1.22168906e-01 -6.81306720e-01
1.80949137e-01 -1.06293142e+00 8.49534690e-01 2.64364660e-01
6.83441162e-02 -9.66028810e-01 -7.61024356e-02 2.72875845e-01
2.32351035e-01 2.70167857e-01 7.33458221e-01 -6.03516340e-01
-6.59603477e-01 -4.60558832e-01 -7.59104325e-04 6.28797174e-01
5.96166849e-01 -1.50687704e-02 -9.59905148e-01 -5.85614264e-01
4.51340854e-01 -2.15364352e-01 7.04599440e-01 2.17677146e-01
1.05472302e+00 -5.49198508e-01 -2.21677929e-01 5.68361878e-01
1.40161026e+00 -1.09969884e-01 5.44022858e-01 -3.34383577e-01
7.93932855e-01 7.35278606e-01 3.12001646e-01 5.43044150e-01
5.74288480e-02 7.69129872e-01 3.34943235e-01 2.36654148e-01
-6.03838898e-02 -8.48987699e-02 5.56649387e-01 8.61242414e-01
-2.37409830e-01 1.46695152e-01 -6.41893625e-01 4.51971084e-01
-1.41145182e+00 -1.03633833e+00 1.78459764e-01 2.56393147e+00
5.04452288e-01 -1.74372450e-01 -1.69310942e-01 4.35370296e-01
6.92374170e-01 4.01195198e-01 -1.71162188e-01 -6.07687011e-02
-2.20483974e-01 6.15837812e-01 1.92385301e-01 5.31654894e-01
-1.01212525e+00 6.35280311e-01 5.94166946e+00 3.68286431e-01
-1.23005724e+00 -1.00271873e-01 6.46185160e-01 2.34905005e-01
-1.84921712e-01 1.08304486e-01 -6.95073962e-01 1.72628030e-01
7.49523103e-01 2.02919226e-02 4.37162846e-01 6.20232821e-01
-1.17663138e-01 -1.20642453e-01 -1.14096177e+00 1.01539373e+00
7.16403127e-01 -9.99336779e-01 1.92926839e-01 2.16683224e-01
8.62547994e-01 -2.48388216e-01 3.72899652e-01 -2.57825106e-01
1.20056514e-02 -1.26392305e+00 4.11197215e-01 4.78288531e-01
6.31212056e-01 -6.60267711e-01 5.26897728e-01 1.80424869e-01
-1.26820111e+00 5.72452620e-02 -4.00497973e-01 1.44559398e-01
-3.51894438e-01 7.58162737e-01 -9.14959371e-01 6.77135408e-01
3.77682567e-01 1.05523074e+00 -5.07870257e-01 8.63611817e-01
-2.84204990e-01 5.92665493e-01 -3.38281751e-01 5.42469680e-01
-1.34484634e-01 -2.48480991e-01 5.46838105e-01 9.60187614e-01
3.30864638e-01 2.12531418e-01 1.77040905e-01 5.72379470e-01
-1.36959925e-01 1.58038586e-01 -8.73138070e-01 1.00551307e-01
1.71150252e-01 1.34359312e+00 -5.84900737e-01 -4.71961014e-02
-7.63283014e-01 9.21796739e-01 3.06361228e-01 5.47104597e-01
-3.44973803e-01 1.24319099e-01 6.68022215e-01 -1.19931726e-02
6.83958173e-01 -5.49957871e-01 -3.07572633e-02 -1.38400841e+00
2.66602635e-01 -1.09947681e+00 2.62114555e-01 -1.13758817e-01
-1.06839025e+00 5.74235857e-01 -8.22892785e-02 -1.07232320e+00
-2.94185489e-01 -5.39018512e-01 -5.80446899e-01 9.26295877e-01
-1.66253233e+00 -1.10764778e+00 -2.27702811e-01 1.01134467e+00
4.10377026e-01 -3.84061366e-01 9.93545234e-01 2.00038090e-01
-6.16111517e-01 5.22997797e-01 7.17396960e-02 2.31108695e-01
6.36789322e-01 -7.85424829e-01 -2.09212214e-01 9.39069867e-01
9.16437507e-01 8.57733905e-01 5.53774834e-01 -5.19924045e-01
-1.73935926e+00 -9.64526474e-01 8.53840411e-01 -1.21119335e-01
3.49634260e-01 -2.75062829e-01 -1.03828871e+00 6.34854913e-01
-9.79320481e-02 7.20497906e-01 9.74517345e-01 1.24650978e-01
-8.03637266e-01 -4.01980370e-01 -1.12947321e+00 2.03709435e-02
8.63519728e-01 -7.79208541e-01 -5.85782647e-01 2.85818368e-01
-2.71228880e-01 -6.83740303e-02 -8.03346753e-01 4.69094723e-01
5.52210212e-01 -1.03367400e+00 8.99653435e-01 -4.42872733e-01
-2.60711294e-02 -4.41691130e-01 -4.86931056e-01 -1.06401277e+00
-3.32726687e-01 -6.03750587e-01 -4.18585613e-02 1.18350720e+00
-1.97398029e-02 -6.30174160e-01 6.94046259e-01 2.62897491e-01
2.97685593e-01 -5.47150970e-01 -1.24300611e+00 -6.12958193e-01
-5.31815827e-01 -8.38848948e-02 3.05178493e-01 1.02864587e+00
-5.20841554e-02 1.14700586e-01 -5.29498398e-01 6.48390889e-01
9.60485995e-01 2.81518489e-01 7.38672256e-01 -1.27306676e+00
-2.48956412e-01 -3.93073494e-03 -1.02070808e+00 -8.75521243e-01
4.38668579e-01 -8.60867441e-01 1.48715982e-02 -8.87448251e-01
2.45623574e-01 -4.15238589e-01 -4.99638945e-01 4.94257092e-01
5.38833216e-02 5.09500027e-01 -2.81715579e-02 3.45977873e-01
-3.28362167e-01 5.86993456e-01 7.98905730e-01 -8.76767933e-02
8.09021965e-02 5.02077676e-02 -6.47366583e-01 7.38213539e-01
3.85993451e-01 -5.32971740e-01 -1.46017358e-01 -3.11692268e-01
-2.86681712e-01 7.93482289e-02 4.51583028e-01 -1.11666751e+00
2.68018365e-01 -1.20196134e-01 7.23466516e-01 -6.48190379e-02
4.12174791e-01 -9.95136142e-01 1.69339254e-01 2.87894279e-01
-1.13260653e-02 -2.78110236e-01 -8.12546758e-04 7.99318254e-01
-5.10445654e-01 -1.62296250e-01 9.85277832e-01 1.06979012e-01
-4.71464992e-01 4.04889345e-01 7.22259581e-02 -5.55266380e-01
9.70561862e-01 -2.41336852e-01 2.20180348e-01 -3.45619708e-01
-8.03603232e-01 -3.45214009e-01 2.80936033e-01 2.49419034e-01
8.37416470e-01 -1.39402401e+00 -8.01666141e-01 9.52438176e-01
-1.02091692e-01 -2.07232401e-01 2.02265009e-01 8.88357222e-01
-9.61339101e-03 5.65152884e-01 -3.42566401e-01 -7.66892970e-01
-1.60383880e+00 3.97797167e-01 1.57213435e-01 -6.42461032e-02
-2.61478961e-01 8.17027569e-01 2.15543330e-01 -2.52676886e-02
8.27630162e-02 -1.05248146e-01 -2.35739022e-01 -6.13152646e-02
5.68355501e-01 -7.65370801e-02 2.75163352e-01 -1.38912642e+00
-5.56879282e-01 9.86369669e-01 -1.25555456e-01 1.30890101e-01
1.53695083e+00 -7.36124068e-02 -3.74570578e-01 1.54323667e-01
1.40556550e+00 3.64238620e-01 -9.78414714e-01 -8.52274716e-01
8.34439769e-02 -1.03412151e+00 5.31823188e-02 -2.62317896e-01
-1.37210000e+00 6.77259624e-01 3.78541231e-01 -7.04284385e-02
1.21427286e+00 1.53238373e-02 1.61123335e-01 2.62570977e-01
5.70281267e-01 -4.27768290e-01 6.54128268e-02 2.48283163e-01
1.03892231e+00 -1.22320998e+00 2.91340083e-01 -6.29005611e-01
-2.19068751e-01 1.23923075e+00 1.70976162e-01 -3.89823645e-01
9.62967992e-01 -4.16799635e-02 -2.85729676e-01 -1.76372752e-01
-5.60509801e-01 -8.99582729e-02 6.50983453e-01 6.27024591e-01
3.57410580e-01 -1.04433872e-01 4.87679094e-02 1.75453216e-01
2.46305719e-01 -2.68857449e-01 1.29234195e-01 9.94423628e-01
-3.43292952e-01 -1.17615378e+00 -6.74609959e-01 2.36715764e-01
-4.55268174e-01 1.99839458e-01 -2.85137713e-01 3.69875133e-01
9.74733084e-02 9.18862998e-01 -7.98052475e-02 -3.30034375e-01
9.22590196e-02 5.47608249e-02 8.31206679e-01 -8.54552269e-01
-4.60447930e-02 2.54484802e-01 -2.87641436e-01 -6.86296284e-01
-8.30641031e-01 -9.84900951e-01 -7.10338116e-01 5.80563881e-02
-3.76177222e-01 1.20887637e-01 7.04396963e-01 1.11459243e+00
2.83787608e-01 1.48885384e-01 1.18825650e+00 -1.10616219e+00
-4.98434395e-01 -7.91660905e-01 -7.54592657e-01 4.79802310e-01
4.28487450e-01 -8.43536973e-01 -5.75731218e-01 3.55557323e-01] | [12.728736877441406, 0.38442298769950867] |
4b475e70-91ee-4860-b0ec-6bb6ec3eeb99 | lap-net-adaptive-features-sampling-via | 2011.07915 | null | https://arxiv.org/abs/2011.07915v1 | https://arxiv.org/pdf/2011.07915v1.pdf | LAP-Net: Adaptive Features Sampling via Learning Action Progression for Online Action Detection | Online action detection is a task with the aim of identifying ongoing actions from streaming videos without any side information or access to future frames. Recent methods proposed to aggregate fixed temporal ranges of invisible but anticipated future frames representations as supplementary features and achieved promising performance. They are based on the observation that human beings often detect ongoing actions by contemplating the future vision simultaneously. However, we observed that at different action progressions, the optimal supplementary features should be obtained from distinct temporal ranges instead of simply fixed future temporal ranges. To this end, we introduce an adaptive features sampling strategy to overcome the mentioned variable-ranges of optimal supplementary features. Specifically, in this paper, we propose a novel Learning Action Progression Network termed LAP-Net, which integrates an adaptive features sampling strategy. At each time step, this sampling strategy first estimates current action progression and then decide what temporal ranges should be used to aggregate the optimal supplementary features. We evaluated our LAP-Net on three benchmark datasets, TVSeries, THUMOS-14 and HDD. The extensive experiments demonstrate that with our adaptive feature sampling strategy, the proposed LAP-Net can significantly outperform current state-of-the-art methods with a large margin. | ['Alois Knoll', 'Fan Lu', 'Jinhu Dong', 'Dan Xu', 'Guang Chen', 'Sanqing Qu'] | 2020-11-16 | null | null | null | null | ['online-action-detection'] | ['computer-vision'] | [ 4.87648994e-01 -3.59616637e-01 -4.30183709e-01 -3.17459196e-01
-4.52947229e-01 -2.49423310e-01 7.31462061e-01 -9.07188728e-02
-6.56816185e-01 5.67272604e-01 3.94675583e-01 1.05734520e-01
-3.03776175e-01 -5.40659010e-01 -3.38427752e-01 -6.88588202e-01
-4.13940817e-01 -8.81987289e-02 8.18878055e-01 1.40882041e-02
5.01663923e-01 3.90661955e-01 -1.77570808e+00 3.46382529e-01
5.39613426e-01 1.20056677e+00 2.39983737e-01 5.66151559e-01
8.60968903e-02 1.21037805e+00 -5.16471446e-01 1.16064891e-01
3.98831964e-01 -6.10098183e-01 -5.58731437e-01 5.45045018e-01
1.31257191e-01 -8.12217474e-01 -5.48562765e-01 8.42698097e-01
3.10803086e-01 5.06303787e-01 2.51179218e-01 -1.40621483e+00
-2.81948805e-01 4.58721370e-01 -5.07850230e-01 7.39189625e-01
7.83277929e-01 5.99412501e-01 9.68391597e-01 -6.08345866e-01
6.16504371e-01 1.14747953e+00 4.29736435e-01 4.35487598e-01
-5.96636653e-01 -5.87510765e-01 7.45699883e-01 7.33254135e-01
-1.22356713e+00 -4.22425985e-01 9.70361412e-01 -2.75756568e-01
7.88409472e-01 2.05089599e-01 1.00642562e+00 1.03254473e+00
2.12022439e-01 1.20562768e+00 8.57146561e-01 -1.18907392e-01
2.64601409e-01 -3.02378774e-01 -6.81537017e-02 5.37141144e-01
-2.30514884e-01 -2.69765891e-02 -6.90260649e-01 -1.09575316e-01
7.84610689e-01 3.69080782e-01 -5.23691237e-01 -1.31888673e-01
-1.60016549e+00 4.46868271e-01 1.05636135e-01 3.31253320e-01
-6.40345752e-01 1.02692924e-01 5.18844485e-01 2.00901464e-01
1.61642641e-01 -1.29546300e-02 -3.87399763e-01 -5.49669743e-01
-6.04368865e-01 1.34512067e-01 3.19610506e-01 7.28960752e-01
7.38266051e-01 -1.61005288e-01 -6.25572860e-01 4.17629153e-01
1.96221843e-02 2.03252748e-01 6.10118866e-01 -1.11853004e+00
5.79416275e-01 7.49027669e-01 4.42816705e-01 -1.26230514e+00
-4.51308817e-01 -1.61969781e-01 -5.84156811e-01 -3.22984867e-02
4.28841233e-01 -1.03351116e-01 -5.90138555e-01 1.57360768e+00
5.69093466e-01 6.23137891e-01 -4.40553464e-02 9.28367496e-01
3.23020011e-01 6.31738305e-01 3.31502669e-02 -6.48998439e-01
1.15081036e+00 -9.73560274e-01 -8.56784821e-01 -1.03545360e-01
5.99815011e-01 -5.21644473e-01 1.01487982e+00 5.74800313e-01
-8.48164439e-01 -7.62874186e-01 -9.83917832e-01 3.67961109e-01
1.15760211e-02 3.97676200e-01 6.36701405e-01 2.53594160e-01
-7.68237293e-01 9.15750444e-01 -9.84414697e-01 -2.65757591e-01
4.24733043e-01 1.48654163e-01 -2.70624131e-01 1.40816286e-01
-1.15042698e+00 3.19772959e-01 6.70041859e-01 2.23674953e-01
-8.66991758e-01 -3.19849551e-01 -6.48018301e-01 -9.24860069e-04
9.50929761e-01 -2.64965355e-01 1.32016993e+00 -1.11223090e+00
-1.60439658e+00 3.33144248e-01 -1.27431378e-01 -7.56598890e-01
7.67523348e-01 -3.26271921e-01 -5.94528735e-01 6.47214174e-01
-8.18958059e-02 4.49487239e-01 9.98634756e-01 -5.48627615e-01
-1.20142448e+00 -2.07784146e-01 4.28948522e-01 4.29583102e-01
-6.95164979e-01 -8.80711004e-02 -4.56834733e-01 -5.31786263e-01
2.83703268e-01 -8.81099403e-01 -1.85442999e-01 3.26434046e-01
-2.50201207e-02 -5.83707392e-01 9.26483750e-01 -3.39514971e-01
1.50808954e+00 -2.14773178e+00 -2.92605534e-02 -1.98707521e-01
1.20850325e-01 3.33632231e-01 2.12324690e-02 2.80303448e-01
1.80483356e-01 -2.02872857e-01 5.75117171e-02 8.99806470e-02
-1.94377184e-01 -1.34066995e-02 -1.42595887e-01 5.15255868e-01
5.87515719e-02 4.90025163e-01 -1.09330249e+00 -6.85577273e-01
6.13386869e-01 2.06678569e-01 -2.85571247e-01 1.97342455e-01
-2.89465129e-01 5.14611840e-01 -8.49370062e-01 5.71570337e-01
5.74218392e-01 -2.62866169e-01 1.35865167e-01 -2.01197356e-01
-2.86714077e-01 7.48275816e-02 -1.31849754e+00 1.62611914e+00
-1.01386644e-01 5.25181293e-01 -4.81867015e-01 -9.76298749e-01
8.80228102e-01 3.22906613e-01 9.25124764e-01 -8.38204682e-01
4.27237786e-02 7.06797913e-02 -9.50338766e-02 -7.42216587e-01
2.90756643e-01 2.07021430e-01 9.87305567e-02 3.65712762e-01
-2.37208039e-01 5.54490983e-01 3.73657227e-01 1.91018563e-02
1.36566639e+00 3.42755795e-01 6.62103772e-01 2.97491938e-01
1.06285822e+00 -1.85161769e-01 9.85473633e-01 8.64112377e-01
-7.07664788e-01 3.22215706e-01 6.17458224e-01 -8.53189230e-01
-4.49765623e-01 -7.46102810e-01 3.41307729e-01 1.03199959e+00
4.06352073e-01 -5.12637019e-01 -3.51280242e-01 -1.10070729e+00
-4.15785223e-01 4.47752208e-01 -5.78341246e-01 -7.08605647e-02
-7.86182880e-01 -3.08502018e-01 2.27006122e-01 6.23629034e-01
1.02661288e+00 -1.37950337e+00 -1.30147183e+00 3.57022792e-01
-3.27150524e-01 -1.22753203e+00 -6.63283050e-01 -2.61273205e-01
-8.39485884e-01 -1.20137227e+00 -7.33443916e-01 -3.69862258e-01
3.89371783e-01 5.65663338e-01 7.33963370e-01 5.46979085e-02
-5.95137142e-02 3.75022084e-01 -6.69475615e-01 -3.28095892e-04
-6.13788180e-02 -1.74372032e-01 -8.04924313e-03 4.48212504e-01
3.63620460e-01 -7.04173982e-01 -1.05188441e+00 5.20918250e-01
-8.96895528e-01 2.14324266e-01 6.79912746e-01 5.74094892e-01
6.55957043e-01 2.11972773e-01 4.56730604e-01 -4.28936869e-01
1.98839709e-01 -3.82606983e-01 -5.16901135e-01 4.24260855e-01
-2.48857230e-01 -7.60107562e-02 8.01358402e-01 -7.40276158e-01
-1.16903901e+00 2.33876988e-01 -3.60695198e-02 -6.17149472e-01
-1.82403430e-01 1.91165984e-01 -2.01847345e-01 2.22134575e-01
3.26643944e-01 6.54711366e-01 -2.42237613e-01 -2.69278169e-01
6.44642720e-03 5.09458780e-01 5.29716671e-01 -2.71804482e-01
4.02542979e-01 7.15181053e-01 -2.72606798e-02 -5.00189900e-01
-8.85132790e-01 -5.64221144e-01 -6.30851567e-01 -7.58906364e-01
7.91349292e-01 -7.70095050e-01 -9.30129945e-01 7.22286463e-01
-8.99844944e-01 -1.19658403e-01 -2.06934795e-01 6.66439652e-01
-7.85540998e-01 6.84546292e-01 -3.50081563e-01 -9.83943522e-01
-1.16818324e-01 -9.40652013e-01 9.23866391e-01 4.38235104e-01
6.12104172e-03 -7.20936775e-01 -1.74106106e-01 1.40581444e-01
1.69808283e-01 5.07290423e-01 2.58862704e-01 -5.49704790e-01
-7.20685482e-01 -7.84292594e-02 -3.49265374e-02 1.70056611e-01
5.00752985e-01 -1.02169059e-01 -5.92028856e-01 -1.99777767e-01
1.49121746e-01 -1.31846353e-01 8.48596156e-01 2.38046363e-01
1.36380553e+00 -4.03601646e-01 -2.33375967e-01 4.95098740e-01
1.18548810e+00 6.83268249e-01 6.98263943e-01 5.62224805e-01
3.00799102e-01 3.82725030e-01 1.24216628e+00 1.03116584e+00
1.97766304e-01 7.00747728e-01 6.20829284e-01 4.50206608e-01
9.70661640e-02 -2.17511103e-01 6.42526686e-01 3.47193718e-01
-2.51664758e-01 -3.48784119e-01 -4.93430436e-01 5.62714517e-01
-2.14084387e+00 -1.30706739e+00 2.06509948e-01 2.23159409e+00
4.95348752e-01 5.89886189e-01 2.94657677e-01 2.90100694e-01
8.59648705e-01 6.45476162e-01 -9.00885522e-01 2.14241341e-01
9.45372805e-02 -3.23144048e-01 2.23750547e-01 -6.64423406e-02
-1.41409159e+00 6.81343615e-01 5.40382290e+00 9.32198822e-01
-9.61919725e-01 3.41797364e-03 4.57043678e-01 -2.52559036e-01
1.68203473e-01 9.90026295e-02 -7.36176550e-01 6.68964207e-01
5.69709301e-01 -2.85235643e-01 1.27029166e-01 8.10991406e-01
4.32279587e-01 -3.13507706e-01 -9.51331019e-01 9.35187161e-01
-1.06765991e-02 -1.15166593e+00 4.31456678e-02 -1.88408181e-01
4.33136612e-01 -3.61682773e-01 -9.95343849e-02 3.38475496e-01
-1.08434081e-01 -3.78186077e-01 6.77069962e-01 7.98823655e-01
2.52241790e-01 -7.89914072e-01 4.70701724e-01 5.35618007e-01
-1.56536067e+00 -6.40471280e-01 -1.55842528e-01 -2.72575974e-01
3.71452540e-01 3.67574781e-01 -4.88138109e-01 5.65150440e-01
6.28354609e-01 1.30644798e+00 -5.46118140e-01 1.04969299e+00
-2.52084374e-01 4.34081942e-01 -1.66309878e-01 -1.31895602e-01
4.45713401e-01 1.18411426e-02 6.20412290e-01 7.59375870e-01
5.35216570e-01 3.50876272e-01 4.37065721e-01 3.37965667e-01
4.26078528e-01 -1.94333121e-02 -3.79663438e-01 -5.10886945e-02
2.84665406e-01 1.05576587e+00 -8.47035885e-01 -4.27760363e-01
-5.97212195e-01 1.01974332e+00 9.52579975e-02 1.21523693e-01
-1.13384521e+00 -1.65541261e-01 4.44452792e-01 1.52750000e-01
5.84755599e-01 -3.25515717e-01 4.74218875e-01 -1.21426713e+00
1.98590368e-01 -6.69225693e-01 8.11929464e-01 -7.26200879e-01
-8.79931331e-01 4.69058365e-01 1.77962616e-01 -1.97375202e+00
-2.06587568e-01 -2.49106720e-01 -6.69851124e-01 4.61358018e-02
-1.39774871e+00 -8.45024467e-01 -4.85841602e-01 8.28938127e-01
1.15677059e+00 -3.96988429e-02 2.84853369e-01 1.45671302e-02
-6.18843377e-01 2.50326186e-01 -9.67125818e-02 6.66220114e-02
4.59135681e-01 -8.64794374e-01 1.45181090e-01 8.20545673e-01
1.10481821e-01 1.94736719e-01 5.97480834e-01 -6.71406150e-01
-1.23145533e+00 -9.56391156e-01 4.62805152e-01 7.87750557e-02
6.75623953e-01 1.13177516e-01 -8.16258907e-01 6.16336644e-01
-7.84227904e-03 3.00638050e-01 3.38158846e-01 -4.20005083e-01
1.87034056e-01 -3.05645555e-01 -9.84159946e-01 5.77802420e-01
1.43933141e+00 -1.11212179e-01 -5.73468983e-01 4.19838607e-01
7.38732040e-01 -2.51314968e-01 -6.85176790e-01 5.53827286e-01
7.92972982e-01 -1.36969113e+00 8.12990129e-01 -3.07401329e-01
2.68173277e-01 -4.27883625e-01 -7.58481910e-03 -8.28865707e-01
-3.80152196e-01 -8.30666125e-01 -5.94693542e-01 1.05196071e+00
-2.14772820e-01 -5.69522142e-01 7.42288351e-01 3.59770715e-01
5.62064350e-02 -8.90713632e-01 -1.02006197e+00 -9.05065536e-01
-8.05221260e-01 -4.00741905e-01 5.71729958e-01 4.40779626e-01
-1.40613303e-01 -2.35336080e-01 -6.66670740e-01 4.56479266e-02
3.88087422e-01 1.62403435e-01 7.25787759e-01 -9.87785578e-01
-4.17743057e-01 -2.50376940e-01 -7.05139816e-01 -1.47597957e+00
-2.79698744e-02 -1.49818063e-01 7.07662944e-03 -1.25239289e+00
2.33967215e-01 -5.96855506e-02 -7.82760203e-01 4.51835990e-01
-2.80061781e-01 7.13100797e-03 2.91188389e-01 3.04587394e-01
-1.15968359e+00 7.54823804e-01 1.44154727e+00 -1.43351089e-02
-2.99767196e-01 2.80432671e-01 -2.85183489e-01 9.93929327e-01
7.35360920e-01 -3.30024749e-01 -8.05737376e-01 -1.20130956e-01
2.29791664e-02 4.31429803e-01 4.94224906e-01 -1.53491855e+00
3.58150274e-01 -5.97533941e-01 3.42826366e-01 -9.92480934e-01
3.40068340e-01 -7.89651036e-01 7.43081495e-02 5.12991369e-01
-4.58969027e-01 -2.26331446e-02 -5.04473932e-02 1.07013679e+00
-1.80746943e-01 -1.37692928e-01 4.48530823e-01 -3.52741390e-01
-1.43244529e+00 3.49332184e-01 -4.74619240e-01 -1.55251861e-01
1.50775015e+00 -3.65352571e-01 -1.14678636e-01 -4.67201710e-01
-8.69819164e-01 4.16692048e-01 1.99576050e-01 5.05400896e-01
8.42421412e-01 -1.44110954e+00 -4.57320660e-01 1.62063465e-02
1.12249382e-01 -3.59675586e-01 6.84990287e-01 1.12122047e+00
-1.94158033e-01 1.91983759e-01 -2.99976170e-01 -7.22521722e-01
-1.42228889e+00 7.26961613e-01 1.90288246e-01 -4.11676198e-01
-8.98699343e-01 4.98994440e-01 7.93269649e-02 2.45236665e-01
2.82215446e-01 -3.34269494e-01 -6.62291169e-01 1.94492564e-01
8.60208571e-01 5.57019770e-01 -3.06531638e-01 -5.63240170e-01
-4.47868526e-01 5.74204922e-01 -6.37498870e-02 4.46928330e-02
1.23478413e+00 -3.91143352e-01 3.38849068e-01 5.39450288e-01
1.05089235e+00 -4.13263857e-01 -1.93768203e+00 -4.21825796e-01
-2.65020202e-03 -9.86767292e-01 -2.09919959e-01 -3.32297474e-01
-1.26647747e+00 5.40700376e-01 7.29327619e-01 2.68946469e-01
1.72841251e+00 -2.59815484e-01 9.74168718e-01 4.28298056e-01
6.26571059e-01 -1.21161628e+00 5.34775734e-01 2.21629918e-01
6.76050544e-01 -1.14161026e+00 2.55808830e-01 -2.83712804e-01
-8.01644325e-01 1.30393994e+00 8.87831926e-01 -7.28845373e-02
5.01011252e-01 -1.44145116e-01 -3.29341322e-01 -3.67419831e-02
-9.33274746e-01 -3.61910075e-01 1.81004792e-01 3.50249887e-01
-5.95160313e-02 -3.03609490e-01 -5.64848304e-01 3.27427626e-01
4.10640955e-01 3.22977901e-01 4.51976269e-01 1.11671817e+00
-6.28213525e-01 -8.03146243e-01 -2.61769354e-01 3.39634955e-01
-3.54619682e-01 4.13274825e-01 -8.51870626e-02 7.50881553e-01
3.27100694e-01 9.65020478e-01 8.52453858e-02 -5.10434389e-01
2.02477589e-01 -8.05818141e-02 3.73343080e-01 -1.84809774e-01
-3.29025239e-01 5.17959855e-02 4.46579279e-03 -9.41472769e-01
-9.32447255e-01 -1.02309394e+00 -1.25977099e+00 1.22510552e-01
-2.69896597e-01 -1.96402639e-01 -1.77640561e-02 1.03850555e+00
3.71180832e-01 5.45708179e-01 9.66845393e-01 -8.61961424e-01
-6.08148932e-01 -8.76758218e-01 -4.03782159e-01 3.99110675e-01
3.53535175e-01 -8.59302461e-01 -4.95001107e-01 7.37664551e-02] | [8.424476623535156, 0.5343735218048096] |
15aaaa0f-98c7-4e01-bfea-7be6b52c09e4 | a-food-photography-app-with-image-recognition | null | null | https://ieeexplore.ieee.org/document/8523925 | https://www.researchgate.net/publication/328834305_A_Food_Photography_App_with_Image_Recognition_for_Thai_Food | A Food Photography App with Image Recognition for Thai Food | In this paper, we present a food photography application for smart phones, which can recognise 13 types of Thai food from photos. With this feature, the application can easily help users calculate their calories and make some suggestion, just by keep taking a photo of food they are eating. Our application uses React Native for the front-end, and Python-Flask for the back-end. For image recognition, we design a deep convolutional neural network to learn from our dataset. Moreover, we compare the result of our model with another model adapted from the famous one of Karen Simonyan andAndrew Zisserman called VGG19. We use transfer learning from the pre-trained VGG19, implementing with Keras and Tensorflow. Our result shows that the transfer learning model is better. It give us approximately 82% test accuracy or 18% top-1 error rate. Using top-3 and top-5 scores, The model reports 2.6% top-3 error rate and 1.3% top-5 error rate, which works well in our application. | ['Vacharapat Mettanant', 'Peerapon Chunpongthong', 'Ukrit Tiankaew'] | 2018-07-01 | null | null | null | ict-ispc-2018-7 | ['image-recognition'] | ['computer-vision'] | [-2.73729622e-01 7.77539611e-02 -4.60936397e-01 -5.63121080e-01
-3.95612180e-01 -3.81913006e-01 1.21425264e-01 1.64240763e-01
-5.41907728e-01 2.16168970e-01 2.29690418e-01 -2.26004586e-01
5.92988133e-01 -1.22678173e+00 -1.14123201e+00 -4.35710549e-01
1.54899769e-02 -8.21967125e-02 5.83043380e-04 -1.38138805e-03
-3.60382497e-02 -2.29194611e-01 -1.37823224e+00 8.05166543e-01
7.97445476e-01 1.42120004e+00 2.04638928e-01 1.05515456e+00
-1.83327988e-01 9.52993989e-01 -2.45431677e-01 -4.41825867e-01
4.13109362e-01 -1.51266158e-01 -5.42464554e-01 -1.16893597e-01
7.61108398e-01 -8.09474468e-01 -1.49704590e-01 1.24551034e+00
4.84323502e-01 -4.29514259e-01 4.74905074e-01 -1.12676954e+00
-1.21698809e+00 7.08550632e-01 -2.82233655e-01 -1.93628669e-01
6.08470261e-01 4.66278613e-01 3.94521683e-01 -6.19403183e-01
1.18191473e-01 1.01795697e+00 1.13012516e+00 4.77041245e-01
-7.00864971e-01 -8.60450923e-01 -7.71994144e-02 5.02152443e-01
-1.12769759e+00 -1.32894188e-01 3.66274774e-01 -4.87801462e-01
1.24289262e+00 2.20658854e-01 8.83532703e-01 1.05725121e+00
2.33741313e-01 1.05234396e+00 1.28969908e+00 -4.02124792e-01
1.67602450e-01 3.04960430e-01 2.43140548e-01 8.96786332e-01
1.95668921e-01 -3.59889045e-02 -1.36248274e-02 3.40532064e-01
7.49753356e-01 2.57846504e-01 1.12540171e-01 2.84427148e-03
-1.12472343e+00 8.44485462e-01 1.14375365e+00 8.23930800e-02
-5.14812052e-01 3.68534513e-02 3.50808412e-01 2.38957256e-01
3.99839640e-01 -1.45317182e-01 -6.80498779e-01 2.97846198e-01
-8.24640989e-01 7.43660494e-04 1.15270686e+00 1.00298858e+00
5.15604556e-01 -7.78340399e-02 5.98382093e-02 5.93852580e-01
5.21411777e-01 9.24246848e-01 6.31247878e-01 -6.09821916e-01
3.40213001e-01 8.81399214e-01 1.00987218e-01 -9.22461689e-01
-6.97238445e-01 3.05396747e-02 -9.40495729e-01 4.33600694e-01
6.80703938e-01 -1.35610864e-01 -1.20058656e+00 1.22892475e+00
4.12656039e-01 -1.19343564e-01 4.40327935e-02 1.21525085e+00
1.39861262e+00 7.04558194e-01 6.44498229e-01 3.65310431e-01
1.74395323e+00 -1.33842707e+00 -4.35779452e-01 -1.03745379e-01
7.04810441e-01 -8.38840425e-01 1.29903293e+00 6.88166857e-01
-7.88260281e-01 -8.06209922e-01 -1.24005926e+00 -4.25514251e-01
-8.64399850e-01 5.64795792e-01 6.36927783e-01 8.24807405e-01
-9.78901803e-01 7.58861661e-01 -7.77802885e-01 -7.93175519e-01
2.36342892e-01 2.25710928e-01 -2.65100569e-01 -2.21133813e-01
-1.02767467e+00 7.38888800e-01 4.44645226e-01 -1.32990941e-01
-5.81766546e-01 -6.95257187e-01 -7.21200347e-01 8.54265690e-02
9.23727974e-02 -7.79185653e-01 1.49927008e+00 -1.29787278e+00
-1.74306679e+00 9.81046498e-01 4.85063732e-01 -5.50797701e-01
5.96708536e-01 -4.28275108e-01 -6.29291713e-01 1.41696900e-01
-9.91024747e-02 8.33843470e-01 4.04429674e-01 -5.32965243e-01
-4.76940274e-01 -4.41502124e-01 2.70407230e-01 4.94220592e-02
-1.66110590e-01 -5.89621365e-02 -1.27701968e-01 -3.82845014e-01
-2.85616040e-01 -8.19197297e-01 4.48557921e-02 3.15185189e-01
-3.41074437e-01 -1.90852508e-01 3.24210703e-01 -1.25111175e+00
1.00193644e+00 -2.24503493e+00 -8.41939986e-01 -5.98820187e-02
-1.59123182e-01 6.57851636e-01 -7.90591314e-02 -1.96139794e-02
-2.75176447e-02 1.33534178e-01 1.84801176e-01 4.56602484e-01
1.77858502e-01 -6.21312819e-02 5.53245246e-01 2.75963843e-01
7.15009123e-02 9.47029591e-01 -1.02304149e+00 -2.39292935e-01
5.67484558e-01 6.27914190e-01 -5.74392915e-01 1.65491670e-01
-3.08050007e-01 -1.75682709e-01 -3.14373970e-01 6.93516970e-01
9.45849717e-01 -5.00289977e-01 2.80852973e-01 -6.47037864e-01
-1.60173327e-01 1.68419585e-01 -1.00403178e+00 1.65186870e+00
-3.69834512e-01 3.56814750e-02 3.02473512e-02 -7.70239890e-01
7.72374630e-01 1.17803529e-01 2.15179816e-01 -9.84261990e-01
3.38784218e-01 5.77323250e-02 -2.57238358e-01 -1.03180909e+00
1.02265328e-01 4.46261615e-01 1.93893448e-01 3.29300135e-01
8.73946995e-02 4.88926679e-01 1.21026278e-01 -1.93064824e-01
8.63073289e-01 5.70683122e-01 7.12954998e-01 -3.28431368e-01
3.22901905e-01 2.45267272e-01 2.19445020e-01 4.83888537e-01
-3.45115423e-01 5.28215468e-01 -3.57891768e-02 -9.62967157e-01
-1.08638799e+00 -9.17944074e-01 3.33959430e-01 1.18588984e+00
-3.17667991e-01 -2.96577871e-01 -1.09522903e+00 -8.06590319e-01
6.75910860e-02 5.60054541e-01 -4.94258374e-01 -3.76003981e-02
-3.65915805e-01 -7.03494012e-01 1.71303138e-01 7.77707994e-01
1.36706901e+00 -1.09117818e+00 -7.28980064e-01 1.43646553e-01
-2.74828017e-01 -8.25549245e-01 -5.78269422e-01 -1.76999465e-01
-6.99365854e-01 -1.45203567e+00 -8.96691024e-01 -1.00898933e+00
2.51313567e-01 3.65546852e-01 1.28087628e+00 2.96732429e-02
-2.06352472e-01 6.49224818e-02 -5.77627659e-01 -6.32228613e-01
-3.92967552e-01 1.23773247e-01 -2.25419492e-01 -3.29926521e-01
1.24736238e+00 -7.44938925e-02 -1.30793452e+00 1.89907923e-01
-6.57825470e-01 3.57539207e-01 4.70038921e-01 2.01709270e-01
4.22731161e-01 -4.14010376e-01 2.14138851e-01 -6.48005545e-01
2.29644123e-02 -5.25627375e-01 -6.12765789e-01 3.77555519e-01
-8.58841598e-01 -1.58800989e-01 6.91135406e-01 -5.40897429e-01
-6.10205948e-01 6.49066448e-01 -4.25342262e-01 -8.31033662e-02
-4.42416459e-01 2.97178477e-01 -1.37409493e-01 7.34908208e-02
8.70356023e-01 -2.45055974e-01 -2.66277678e-02 -7.75504887e-01
5.99527836e-01 9.69793499e-01 4.89612311e-01 7.88951069e-02
-4.94310893e-02 8.33688006e-02 -5.69920361e-01 -6.19680583e-01
-7.53727436e-01 -4.84021813e-01 -3.53079170e-01 -2.64133781e-01
1.27509296e+00 -1.27284968e+00 -1.09598386e+00 8.19090962e-01
-8.17666411e-01 -7.88854241e-01 1.27960205e-01 6.94766223e-01
-3.18033010e-01 8.51648226e-02 -8.13912749e-01 -4.00878727e-01
-9.65270877e-01 -8.20350528e-01 8.07800591e-01 4.16996032e-01
-1.27286047e-01 -7.69729495e-01 -1.78491339e-01 3.29466611e-01
5.77836335e-01 4.00523812e-01 5.12505949e-01 -4.33844447e-01
-2.13225469e-01 -1.28181547e-01 -6.30590141e-01 5.93032360e-01
4.60323803e-02 -6.14889823e-02 -1.13278651e+00 -1.85660664e-02
-1.49818704e-01 -2.44618163e-01 9.23972428e-01 6.65148437e-01
1.15529132e+00 -6.67719245e-01 -1.01583943e-01 7.11041808e-01
1.78088284e+00 1.43306464e-01 9.06020164e-01 5.26530385e-01
7.93099344e-01 1.07252948e-01 2.96602696e-01 1.45251155e-01
9.96291339e-01 5.54142058e-01 3.92815232e-01 -5.61948478e-01
-4.53007907e-01 -4.70382541e-01 7.30169415e-01 7.83939421e-01
-2.55879220e-02 1.42183840e-01 -6.23705745e-01 1.94286957e-01
-1.78438580e+00 -7.43518412e-01 -4.26877737e-01 2.12038255e+00
7.86281943e-01 -3.04202437e-01 7.20095217e-01 5.60651980e-02
6.49561763e-01 -4.43189919e-01 -6.70029998e-01 -6.56600893e-01
2.87263721e-01 2.64260560e-01 9.60313618e-01 3.70165199e-01
-1.47362638e+00 7.71881580e-01 6.57575846e+00 2.50322253e-01
-1.40672159e+00 9.30811539e-02 7.80491292e-01 -3.27077992e-02
4.41623360e-01 -8.12747955e-01 -5.62414467e-01 6.66370153e-01
1.31452537e+00 2.01857671e-01 6.51194394e-01 1.16715527e+00
1.51990920e-01 -2.35837758e-01 -1.06344819e+00 9.31167722e-01
2.97027171e-01 -7.75480211e-01 -2.70752370e-01 -2.58189350e-01
5.03331840e-01 3.36230546e-01 -9.92358401e-02 5.35989046e-01
6.38304412e-01 -8.40496898e-01 6.97003424e-01 5.32975912e-01
7.71883190e-01 -4.28740352e-01 7.15213597e-01 1.34603024e-01
-1.21396279e+00 4.04884592e-02 -5.38033724e-01 -2.91124880e-01
-5.04370391e-01 7.65361965e-01 -1.02587390e+00 3.17137182e-01
1.09974194e+00 5.38603604e-01 -4.96643066e-01 1.24296474e+00
-4.46852237e-01 6.09777749e-01 -5.29979646e-01 -4.15415585e-01
1.30680472e-01 -1.70060858e-01 -4.83099192e-01 1.54357517e+00
6.06620014e-01 1.77362069e-01 3.44685882e-01 5.25822103e-01
-1.19296812e-01 4.74962682e-01 -5.10845363e-01 1.34204134e-01
-1.82277381e-01 1.47916126e+00 -6.08717501e-01 -7.28226662e-01
-6.51350737e-01 1.04234242e+00 4.38619517e-02 1.08565353e-01
-9.16933775e-01 -3.33615273e-01 3.22582215e-01 1.59663651e-02
3.98083329e-01 1.10869683e-01 -1.41846210e-01 -1.18140459e+00
-1.21372536e-01 -9.12977397e-01 4.49218631e-01 -1.12118256e+00
-1.34198689e+00 8.59702975e-02 -3.24640483e-01 -7.84682989e-01
-8.22600871e-02 -1.11254835e+00 -4.15929854e-01 5.76044858e-01
-1.59681487e+00 -1.23046935e+00 -5.82794845e-01 5.41165709e-01
3.85773063e-01 2.40553394e-01 1.06083858e+00 6.90555036e-01
-3.96616101e-01 5.84205329e-01 -2.94991489e-02 6.23304904e-01
6.75278246e-01 -1.41524005e+00 7.35741794e-01 4.24960107e-01
-1.48492247e-01 2.19871238e-01 3.72505575e-01 -5.93714833e-01
-1.42465794e+00 -1.45318913e+00 1.23975551e+00 -1.05753109e-01
3.20218295e-01 -2.93785542e-01 -7.10276723e-01 5.89130759e-01
2.94944406e-01 -8.88307765e-02 7.42671549e-01 -1.54495621e-02
-6.98938370e-01 -3.55594784e-01 -1.75772595e+00 2.57469505e-01
8.66754770e-01 -2.35368535e-01 -3.50073904e-01 6.88256264e-01
6.29651368e-01 -4.40296739e-01 -9.39470947e-01 7.70946145e-02
1.04055452e+00 -8.99769187e-01 1.00890231e+00 -4.33335036e-01
6.62903368e-01 -1.63139865e-01 -2.05395922e-01 -1.39460409e+00
-6.61290586e-01 6.45836815e-02 -9.94714722e-03 8.82153392e-01
5.19699097e-01 -4.82542157e-01 8.69265318e-01 5.50099671e-01
2.96545159e-02 -2.35643014e-01 -2.42380470e-01 -7.06004381e-01
1.74942017e-01 -1.62147492e-01 9.05068159e-01 1.10276246e+00
1.43720597e-01 2.11274609e-01 -2.48766959e-01 1.24655269e-01
3.38893384e-01 -8.76755193e-02 7.39450932e-01 -7.51078963e-01
-2.27721170e-01 -1.28256440e-01 -2.01715410e-01 -7.56618977e-01
-5.41062713e-01 -1.00305831e+00 3.60583588e-02 -1.79569459e+00
2.81858981e-01 -1.63371824e-02 -5.42169631e-01 1.03179049e+00
-1.00736916e-02 5.58481336e-01 3.95085156e-01 -4.15926188e-01
-4.64273602e-01 -2.25911006e-01 1.10354948e+00 -4.42493051e-01
-9.89918187e-02 -2.78506339e-01 -7.87492573e-01 9.11660612e-01
1.20013320e+00 -2.20545098e-01 -2.94888876e-02 -7.25560069e-01
3.49399030e-01 -5.40076137e-01 5.32837570e-01 -1.34203017e+00
1.07592447e-02 -6.42949268e-02 8.74604046e-01 -3.76740038e-01
-9.34912786e-02 -9.72387850e-01 2.84031868e-01 9.11146045e-01
-1.44051880e-01 1.60072488e-03 5.51645905e-02 3.65486443e-02
3.86427313e-01 -4.57293540e-02 5.68786085e-01 -6.79181576e-01
-8.33145082e-01 1.44689694e-01 -1.08038090e-01 -4.92574811e-01
7.29514778e-01 8.74581113e-02 -6.41088963e-01 -5.97453862e-02
-8.71025681e-01 1.81981400e-01 2.54131436e-01 3.36220115e-01
3.43039602e-01 -1.66649055e+00 -6.51938677e-01 3.60697299e-01
1.36826679e-01 -6.23496950e-01 1.63330883e-01 6.24732196e-01
-9.59006190e-01 2.91864842e-01 -4.96509790e-01 -5.84003925e-01
-1.22141659e+00 8.53196681e-01 2.36856639e-01 -7.27229342e-02
-8.00444901e-01 5.17127573e-01 -3.32961567e-02 -5.00631928e-01
2.38269374e-01 -1.00890398e+00 -3.18530470e-01 -2.19264954e-01
1.12801707e+00 4.88376707e-01 2.87699699e-01 -1.97586671e-01
-3.94376963e-01 4.60791588e-01 4.01717462e-02 5.21355331e-01
1.33020258e+00 1.03143036e-01 4.22081113e-01 1.76451132e-01
1.22506452e+00 -5.31928658e-01 -1.18779624e+00 1.33099526e-01
-3.12618107e-01 -2.95659781e-01 1.54345125e-01 -1.46707010e+00
-1.33020186e+00 8.87754142e-01 1.38067961e+00 4.11568373e-01
1.18419445e+00 -5.00701904e-01 9.94236231e-01 2.22258225e-01
2.85009235e-01 -1.22609305e+00 -3.22827399e-01 3.65871191e-01
6.73268735e-01 -1.45263994e+00 -8.29438195e-02 -4.06022608e-01
-4.64089215e-01 1.18575120e+00 4.66166109e-01 -5.25249243e-01
6.93095088e-01 1.78984687e-01 4.76136148e-01 -1.04525103e-03
-3.03134948e-01 -2.32517272e-01 1.66565135e-01 9.57954168e-01
5.97359180e-01 7.61626005e-01 -3.73320758e-01 6.38614893e-01
-4.30202603e-01 9.12052453e-01 1.61429062e-01 5.79996347e-01
-5.82599103e-01 -6.69478059e-01 -3.52204770e-01 3.82244051e-01
-4.91593391e-01 -1.99140266e-01 -2.64342248e-01 8.33293617e-01
3.82265627e-01 1.15154099e+00 6.13554530e-02 -7.24967778e-01
5.31877756e-01 2.25290462e-01 6.39115274e-01 -2.12975234e-01
-1.06301987e+00 3.97122279e-02 3.21153879e-01 -9.05158877e-01
-6.06818080e-01 -3.65142882e-01 -1.02336824e+00 -7.67114222e-01
2.08729617e-02 -3.83525550e-01 1.15111482e+00 6.00077093e-01
2.73763418e-01 2.30362847e-01 4.22042847e-01 -6.08525574e-01
-4.37462479e-01 -1.14758217e+00 -3.92906994e-01 4.64609355e-01
7.45528415e-02 -1.77936628e-01 4.61804122e-02 3.78320843e-01] | [11.573399543762207, 4.404551982879639] |
a4895dda-614d-4191-9a85-2f17ffdd8271 | heterogeneous-graph-neural-network-for-1 | 2205.11343 | null | https://arxiv.org/abs/2205.11343v3 | https://arxiv.org/pdf/2205.11343v3.pdf | Heterogeneous Graph Neural Network for Personalized Session-Based Recommendation with User-Session Constraints | The recommendation system provides users with an appropriate limit of recent online large amounts of information. Session-based recommendation, a sub-area of recommender systems, attempts to recommend items by interpreting sessions that consist of sequences of items. Recently, research to include user information in these sessions is progress. However, it is difficult to generate high-quality user representation that includes session representations generated by user. In this paper, we consider various relationships in graph created by sessions through Heterogeneous attention network. Constraints also force user representations to consider the user's preferences presented in the session. It seeks to increase performance through additional optimization in the training process. The proposed model outperformed other methods on various real-world datasets. | ['Minjae Park'] | 2022-05-23 | null | null | null | null | ['session-based-recommendations'] | ['miscellaneous'] | [ 1.83847040e-01 1.68555573e-01 -5.59179604e-01 -7.37717569e-01
7.27200881e-02 -3.61970603e-01 4.69436914e-01 9.54830274e-02
-1.26446818e-03 6.30009830e-01 8.09701920e-01 -2.30849266e-01
-6.33678377e-01 -8.56852770e-01 -3.79350871e-01 -2.49988317e-01
-2.83943117e-01 4.81026232e-01 9.01423395e-02 -6.37763083e-01
6.52278185e-01 1.89600214e-01 -1.63560951e+00 6.01674855e-01
1.12274349e+00 6.52511477e-01 4.29847002e-01 5.86615503e-01
-4.05573517e-01 4.02861536e-01 -4.85866696e-01 -4.12938178e-01
1.17091782e-01 -8.45277190e-01 -6.54145181e-01 2.64796764e-01
3.26472014e-01 -3.18079382e-01 -4.63560998e-01 7.00029910e-01
2.67281234e-01 1.09532011e+00 7.71707714e-01 -1.25459254e+00
-1.58918750e+00 1.05683935e+00 -1.35985166e-01 5.33872068e-01
7.11813986e-01 -5.54430783e-01 1.35863304e+00 -7.80947864e-01
4.72183079e-01 1.07194412e+00 2.01260075e-01 8.16508412e-01
-7.92755544e-01 -3.27861279e-01 7.44043112e-01 4.50140506e-01
-8.94340992e-01 1.49441898e-01 8.61271083e-01 -1.72836855e-01
1.26980376e+00 7.13623106e-01 9.31537807e-01 1.27434886e+00
1.03629515e-01 6.80095613e-01 3.41446251e-01 -2.77236432e-01
5.83008975e-02 5.36325037e-01 9.09964502e-01 1.15808390e-01
3.67473155e-01 4.30204384e-02 -4.67403233e-01 -2.09720343e-01
7.43312240e-01 5.34144163e-01 -4.43656802e-01 -2.92203605e-01
-5.85227072e-01 9.76178646e-01 6.61321878e-01 4.32913810e-01
-5.16219079e-01 -3.78237605e-01 1.07780413e-03 4.94951278e-01
5.59809446e-01 8.11023116e-01 -1.96650371e-01 2.03383088e-01
-3.81904721e-01 3.28609720e-02 1.11177051e+00 1.08694780e+00
2.65232563e-01 1.00519396e-01 -3.74224186e-01 9.95672107e-01
6.37163758e-01 -1.70103297e-01 8.18862021e-01 -3.29228908e-01
2.12491781e-01 6.81128442e-01 1.26328230e-01 -1.12700641e+00
-2.70910800e-01 -9.43083644e-01 -6.69711947e-01 -4.95544881e-01
6.76300493e-04 -2.02546895e-01 -8.85909796e-01 1.43191135e+00
8.49003643e-02 4.25959796e-01 -1.49989538e-02 1.19338453e+00
1.17943442e+00 9.67729270e-01 1.75194785e-01 -4.90341783e-01
9.24034178e-01 -1.40880108e+00 -9.62000608e-01 -1.94312513e-01
3.27663392e-01 -4.97598320e-01 9.79967892e-01 3.10332716e-01
-9.32598591e-01 -7.16252685e-01 -9.67583835e-01 2.09340304e-01
-4.47818607e-01 -1.67819276e-01 1.02444971e+00 7.55788028e-01
-1.11465752e+00 7.32250333e-01 -1.08821899e-01 -9.77906764e-01
-1.31762803e-01 6.78736091e-01 5.84377609e-02 1.03413969e-01
-1.37335896e+00 7.85268724e-01 2.05764756e-01 1.82789445e-01
-4.79012996e-01 -5.92269599e-01 -6.11072719e-01 4.04871076e-01
1.75158352e-01 -7.69564986e-01 9.87173855e-01 -1.36167288e+00
-1.51511002e+00 8.09393898e-02 -3.46234143e-02 -2.53454298e-01
-1.79149359e-01 -2.68338293e-01 -9.71031666e-01 -4.94952410e-01
-5.74698150e-01 1.24335252e-01 4.52841938e-01 -1.19198346e+00
-7.18841016e-01 -3.20088476e-01 4.51782286e-01 8.13628197e-01
-7.31249213e-01 1.08954422e-02 -5.49609601e-01 -4.21201557e-01
5.30072190e-02 -1.02086735e+00 -4.68243271e-01 -6.41785800e-01
-6.33829832e-02 -6.53145373e-01 6.99015379e-01 -4.25520241e-01
1.65248513e+00 -1.81723237e+00 2.83717692e-01 6.18864059e-01
1.22581813e-02 2.16231123e-01 -2.71175832e-01 7.89262235e-01
5.99737354e-02 3.07835609e-01 5.25748193e-01 -1.48016617e-01
1.27995992e-02 2.83419579e-01 -1.53494030e-01 8.68580565e-02
-6.16357028e-01 6.67767882e-01 -8.31984401e-01 -1.21077579e-02
7.85481632e-02 4.38413560e-01 -9.74713504e-01 6.60307050e-01
-7.31206313e-02 2.67568022e-01 -7.62739122e-01 1.68481484e-01
3.00804168e-01 -4.95590538e-01 4.34747875e-01 -1.20574407e-01
2.58596361e-01 4.91570979e-01 -1.06727195e+00 1.65523303e+00
-3.09563488e-01 3.16084921e-01 -4.01153177e-01 -9.37603474e-01
1.01965332e+00 3.50536793e-01 5.15771687e-01 -5.61368227e-01
2.97331512e-02 -3.20278317e-01 4.42185104e-01 -5.86863399e-01
9.99515712e-01 3.48912120e-01 1.81097105e-01 7.43958592e-01
1.29351556e-01 5.45194685e-01 2.55669951e-01 5.65646350e-01
6.54074788e-01 5.96332289e-02 3.04426193e-01 -1.87180415e-01
5.05636632e-01 -3.88390034e-01 2.23434374e-01 1.07180381e+00
1.49440169e-01 3.71016443e-01 -1.92093283e-01 -2.51752257e-01
-7.58385301e-01 -7.08046019e-01 1.67195395e-01 1.68961179e+00
2.98173308e-01 -6.29144549e-01 -2.53648371e-01 -7.82796204e-01
-4.93564792e-02 9.49147761e-01 -8.96140575e-01 -3.83894742e-01
-2.42016971e-01 -6.65148377e-01 -3.35543215e-01 6.61413133e-01
-2.69018319e-02 -1.27617633e+00 1.41762078e-01 4.27668303e-01
3.01090702e-02 -4.60629195e-01 -9.50638056e-01 -1.97437748e-01
-9.15138841e-01 -8.05005252e-01 -6.07203066e-01 -9.56526637e-01
8.88649821e-01 7.40387857e-01 1.12513340e+00 5.80136299e-01
1.90596163e-01 6.53149366e-01 -9.50657964e-01 -1.93155333e-01
1.65354311e-01 2.09862709e-01 1.48358926e-01 1.65200636e-01
5.60379684e-01 -7.12743998e-01 -8.02044928e-01 4.68157500e-01
-7.10339606e-01 -1.06907219e-01 2.90209740e-01 7.28831291e-01
1.83567554e-01 -1.49387300e-01 9.37070131e-01 -1.50171852e+00
1.30044055e+00 -1.11425257e+00 1.04710124e-01 4.32845801e-01
-1.07607400e+00 -2.83022881e-01 6.41128957e-01 -5.90757191e-01
-1.42482138e+00 -5.56734800e-01 -1.48443848e-01 5.72759211e-02
-1.22755535e-01 9.40001786e-01 -5.54269552e-02 5.55308238e-02
8.05416763e-01 4.56095226e-02 -3.27295005e-01 -5.76425314e-01
4.10723448e-01 9.14635837e-01 -5.98167256e-02 -3.26127350e-01
3.50911945e-01 -2.23371148e-01 -6.03250265e-01 -6.02790654e-01
-1.14005399e+00 -8.71538758e-01 -2.37830251e-01 -4.22611594e-01
4.21668738e-01 -4.45767343e-01 -4.45625663e-01 -3.22405100e-01
-6.46487594e-01 7.92716444e-02 -8.31646994e-02 8.32094193e-01
-1.26054287e-01 2.67707020e-01 -6.67121291e-01 -9.11668181e-01
-4.77126420e-01 -5.90020776e-01 1.40673444e-01 6.75471425e-01
-3.10438126e-01 -1.33103001e+00 1.02889650e-01 4.38958496e-01
7.81897128e-01 -4.44300622e-01 6.27439022e-01 -1.29581428e+00
-4.76498276e-01 -2.69967139e-01 3.01135145e-02 5.83154000e-02
4.27518874e-01 -7.05202892e-02 -5.56565106e-01 -4.11034077e-01
-3.34218234e-01 2.04639286e-01 5.73415220e-01 3.98100317e-01
1.39802027e+00 -4.75089103e-01 -3.89560848e-01 3.39582682e-01
1.27820194e+00 7.52949357e-01 7.45141327e-01 8.68577510e-02
6.78954899e-01 6.03306770e-01 5.64266860e-01 4.54721332e-01
3.03546727e-01 6.06503725e-01 2.42828369e-01 1.79914296e-01
2.15570945e-02 -4.44578826e-01 2.09155470e-01 1.07023168e+00
-4.51380938e-01 -9.50705469e-01 -1.99262172e-01 4.22878474e-01
-2.19632578e+00 -1.20142937e+00 -1.97913244e-01 2.17353511e+00
1.86467052e-01 -3.92317921e-02 2.56290615e-01 -3.07280898e-01
7.24330246e-01 -9.00188833e-02 -6.35393023e-01 -6.68940842e-01
1.74489170e-01 -9.95295420e-02 1.44945845e-01 5.29100657e-01
-7.05066741e-01 7.13557005e-01 6.86233044e+00 1.39088348e-01
-7.62548268e-01 -2.30194256e-01 3.06034535e-01 -2.32382804e-01
-8.11286211e-01 -1.94513515e-01 -9.14554656e-01 5.20321012e-01
1.26713061e+00 -6.81593657e-01 4.90958363e-01 7.64056623e-01
2.91214198e-01 3.45529288e-01 -1.17095852e+00 5.63648582e-01
6.64787769e-01 -1.22350550e+00 4.28525835e-01 1.83248997e-01
9.32382166e-01 -3.62755418e-01 1.85203388e-01 6.78642571e-01
4.64752257e-01 -8.75683010e-01 -1.71818659e-01 6.46847427e-01
-9.33594927e-02 -6.84756339e-01 7.91375101e-01 4.07283425e-01
-1.07432461e+00 -3.60309392e-01 -6.61114216e-01 -2.31922448e-01
2.55818963e-01 -6.47498202e-03 -9.42676067e-01 6.47740781e-01
5.94126821e-01 1.16394067e+00 -5.42127013e-01 1.56290317e+00
4.49173637e-02 9.03070569e-01 5.56119829e-02 -5.31760454e-01
1.96401209e-01 -7.31571615e-01 4.16776776e-01 1.10503435e+00
7.21877635e-01 5.92617333e-01 3.90666217e-01 2.79478669e-01
-2.18033884e-02 7.31711566e-01 -7.38220334e-01 -7.49066696e-02
3.16166997e-01 1.30618787e+00 -3.72164994e-01 -2.73380816e-01
-5.16079605e-01 8.69522512e-01 4.40496802e-01 5.66229284e-01
-6.27680898e-01 -1.08277500e-02 5.24278462e-01 2.60933608e-01
8.80513340e-02 7.02121407e-02 -1.67660713e-02 -1.01937258e+00
-5.91110647e-01 -6.20636940e-01 8.11813712e-01 -9.04927194e-01
-1.65078580e+00 7.13101864e-01 -6.99630845e-03 -1.22932613e+00
-3.56739134e-01 -2.11665094e-01 -9.80304360e-01 8.65961075e-01
-1.01104438e+00 -9.84334290e-01 -3.18625420e-01 7.21015751e-01
9.58153069e-01 -3.90653670e-01 8.47093821e-01 5.49510539e-01
-6.05538368e-01 7.65856266e-01 1.59114674e-01 -4.87226486e-01
6.05459452e-01 -1.21366358e+00 7.86417276e-02 5.69546998e-01
3.45159739e-01 1.17895222e+00 1.04343736e+00 -7.74206161e-01
-1.25401366e+00 -8.80489945e-01 8.46446991e-01 -2.46980518e-01
3.76852959e-01 1.22286901e-01 -1.16310358e+00 1.06278372e+00
6.16693079e-01 -5.57730615e-01 1.40523994e+00 8.59748125e-01
6.66837171e-02 4.49838154e-02 -9.97353852e-01 5.27042389e-01
1.43962908e+00 -1.42494708e-01 -8.65202546e-01 4.71763134e-01
5.60121298e-01 -1.62667915e-01 -8.58050823e-01 -3.90009545e-02
4.48687047e-01 -5.38421094e-01 8.98059428e-01 -1.28866816e+00
9.02974159e-02 -1.18873455e-02 3.06319073e-02 -1.87418103e+00
-1.08000231e+00 -5.38846374e-01 -5.35862386e-01 1.13629293e+00
7.52083302e-01 -5.12610197e-01 1.05564606e+00 9.79585767e-01
-5.74230969e-01 -5.14477372e-01 6.45482028e-03 -3.55601519e-01
-4.76554811e-01 1.33912593e-01 7.58278191e-01 1.00082803e+00
5.75626671e-01 6.59302175e-01 -8.57746243e-01 1.35633543e-01
5.22242785e-01 2.36862287e-01 4.93957072e-01 -1.63340354e+00
-4.62448984e-01 -2.44940519e-01 7.94252828e-02 -1.25728571e+00
7.46210441e-02 -1.04736590e+00 -2.71459460e-01 -2.14066100e+00
1.61830008e-01 -3.48825276e-01 -1.03361607e+00 1.10128507e-01
-1.99670643e-01 9.42361448e-03 7.35029671e-03 9.93558317e-02
-7.74681091e-01 4.23018277e-01 1.53396344e+00 9.06271413e-02
-5.93716443e-01 4.48668689e-01 -1.31724143e+00 4.14844543e-01
1.00947297e+00 -1.98503047e-01 -1.13895178e+00 -2.23783642e-01
2.81735271e-01 1.63814783e-01 -5.07381976e-01 -6.24111772e-01
5.27332902e-01 -4.91870910e-01 4.56814826e-01 -6.25662923e-01
4.91077602e-01 -8.87236118e-01 4.64665741e-01 -1.23896196e-01
-9.28894699e-01 4.55610678e-02 -2.65640378e-01 9.02810276e-01
7.25875273e-02 -4.18997228e-01 1.48291945e-01 -1.13344520e-01
-9.97367382e-01 5.61946690e-01 -4.20081079e-01 -6.80899322e-01
9.58822608e-01 -4.61147130e-01 -1.99757218e-01 -8.60725522e-01
-1.13182783e+00 3.05346280e-01 -1.26596540e-01 1.10756350e+00
7.69903064e-01 -1.42194664e+00 -4.65690136e-01 1.47134691e-01
1.94693565e-01 -8.66249144e-01 5.77772200e-01 2.94775903e-01
5.58025017e-02 5.34580231e-01 -4.44146186e-01 5.92628382e-02
-1.44558251e+00 6.35679543e-01 1.49295166e-01 -1.75789997e-01
-4.82639492e-01 8.92923474e-01 2.23592911e-02 -2.91840851e-01
4.27327424e-01 1.79102957e-01 -1.36950445e+00 1.94720298e-01
7.42981851e-01 2.28754118e-01 -1.24708079e-01 -5.75521588e-01
7.35146850e-02 1.21550644e-02 -5.80384493e-01 2.26726264e-01
1.35637069e+00 -4.30689514e-01 2.21770212e-01 3.51545841e-01
8.77181351e-01 -2.31163532e-01 -7.13959277e-01 -2.86267757e-01
-6.97189048e-02 -8.17936242e-01 1.46069959e-01 -9.14279819e-01
-1.10991943e+00 2.98518926e-01 6.15305305e-01 7.39082634e-01
8.49019766e-01 -2.92790085e-01 6.26129746e-01 4.91867810e-01
3.65574926e-01 -1.21820903e+00 1.16514154e-01 3.65401953e-01
9.90156889e-01 -1.14767313e+00 5.14206290e-02 -5.17101526e-01
-8.75570536e-01 8.29714954e-01 1.30033696e+00 -2.44310260e-01
1.13965452e+00 -6.18230164e-01 -1.90464884e-01 -2.99128741e-02
-1.05761719e+00 -1.26796231e-01 1.05554128e+00 6.31410122e-01
1.06077969e+00 5.98332435e-02 -9.64902401e-01 1.04042935e+00
-5.86697049e-02 -2.99298298e-02 7.99940825e-01 6.91832840e-01
-7.88913488e-01 -1.31501329e+00 1.46008268e-01 1.07360387e+00
-3.66946608e-01 -8.27275813e-02 -3.20163041e-01 2.75759250e-01
-3.29768926e-01 1.34586370e+00 4.69361879e-02 -5.03281832e-01
4.24106985e-01 -8.59064460e-02 2.50189155e-01 -1.05727112e+00
-9.80523288e-01 2.66288798e-02 4.23908532e-01 -3.82285178e-01
-3.20775896e-01 -3.67580563e-01 -8.98930848e-01 -2.50555277e-01
-5.30688345e-01 9.13343728e-01 5.74954510e-01 6.84697747e-01
4.49438214e-01 7.83855796e-01 7.31478453e-01 -8.13381255e-01
-4.48430538e-01 -1.20020831e+00 -9.83433187e-01 7.42008865e-01
-2.06667751e-01 -6.44518793e-01 -3.86873335e-01 -2.52920002e-01] | [10.17075252532959, 5.617922782897949] |
819fbdef-aa77-42be-b4d7-51ec64afea6a | effective-sequential-classifier-training-for | 1706.04719 | null | http://arxiv.org/abs/1706.04719v2 | http://arxiv.org/pdf/1706.04719v2.pdf | Effective Sequential Classifier Training for SVM-based Multitemporal Remote Sensing Image Classification | The explosive availability of remote sensing images has challenged supervised
classification algorithms such as Support Vector Machines (SVM), as training
samples tend to be highly limited due to the expensive and laborious task of
ground truthing. The temporal correlation and spectral similarity between
multitemporal images have opened up an opportunity to alleviate this problem.
In this study, a SVM-based Sequential Classifier Training (SCT-SVM) approach is
proposed for multitemporal remote sensing image classification. The approach
leverages the classifiers of previous images to reduce the required number of
training samples for the classifier training of an incoming image. For each
incoming image, a rough classifier is firstly predicted based on the temporal
trend of a set of previous classifiers. The predicted classifier is then
fine-tuned into a more accurate position with current training samples. This
approach can be applied progressively to sequential image data, with only a
small number of training samples being required from each image. Experiments
were conducted with Sentinel-2A multitemporal data over an agricultural area in
Australia. Results showed that the proposed SCT-SVM achieved better
classification accuracies compared with two state-of-the-art model transfer
algorithms. When training data are insufficient, the overall classification
accuracy of the incoming image was improved from 76.18% to 94.02% with the
proposed SCT-SVM, compared with those obtained without the assistance from
previous images. These results demonstrate that the leverage of a priori
information from previous images can provide advantageous assistance for later
images in multitemporal image classification. | ['David Paull', 'Yiqing Guo', 'Xiuping Jia'] | 2017-06-15 | null | null | null | null | ['remote-sensing-image-classification'] | ['miscellaneous'] | [ 8.90579581e-01 -3.19807947e-01 -1.51826069e-01 -5.01404643e-01
-5.74493527e-01 -4.81155604e-01 4.80332494e-01 2.52199560e-01
-4.80907708e-01 9.52730000e-01 -5.21526337e-01 -5.24412215e-01
-3.96863043e-01 -8.52714241e-01 -2.68454850e-01 -9.91516709e-01
-2.20795006e-01 1.65560290e-01 2.84211874e-01 -4.02840346e-01
3.23900998e-01 8.57581735e-01 -1.90055120e+00 6.84462845e-01
1.01549447e+00 1.20930827e+00 6.81740463e-01 5.51662922e-01
1.30593911e-01 5.52040339e-01 -3.45247149e-01 2.78929055e-01
4.76025641e-01 -3.39316338e-01 -6.24092698e-01 3.44417185e-01
3.34040314e-01 -2.21902490e-01 1.51290461e-01 7.34300554e-01
1.29711911e-01 3.00014526e-01 5.78514457e-01 -1.07184672e+00
-3.69682945e-02 1.16023801e-01 -6.16919100e-01 4.33393925e-01
-9.66130346e-02 -1.27080465e-02 6.89938068e-01 -6.97926641e-01
3.28902751e-01 5.45567811e-01 1.03712392e+00 4.05382887e-02
-1.31554317e+00 -5.68597257e-01 6.11060150e-02 2.31570572e-01
-1.33241570e+00 -2.09361732e-01 5.34671128e-01 -6.43861175e-01
1.08248329e+00 4.50095654e-01 9.69077468e-01 1.35223523e-01
2.28122577e-01 3.31399709e-01 1.61749470e+00 -5.92964828e-01
6.66685551e-02 5.38987398e-01 -3.75101045e-02 4.60800439e-01
-2.03009024e-02 4.23862398e-01 -3.25318098e-01 -1.35527208e-01
6.46994889e-01 2.25060046e-01 -2.89078262e-02 -5.10973297e-02
-9.04790819e-01 1.07525742e+00 8.79014134e-01 4.03792977e-01
-9.00615931e-01 -6.60443842e-01 2.87045717e-01 4.57018733e-01
7.96741128e-01 4.38452780e-01 -7.61606514e-01 3.64506960e-01
-1.37444901e+00 -2.09415048e-01 3.25426310e-01 1.81929126e-01
1.10054207e+00 2.28043646e-01 4.55951184e-01 9.60794926e-01
7.58173019e-02 6.64642632e-01 3.69168907e-01 -3.99692923e-01
3.76669407e-01 7.91458964e-01 8.83231312e-02 -1.21756065e+00
-3.63192737e-01 -3.49976063e-01 -7.35378504e-01 2.88219661e-01
3.39126170e-01 -6.58162162e-02 -1.06436670e+00 9.31391418e-01
3.67772818e-01 5.00213839e-02 5.19216061e-01 6.15151644e-01
2.10262567e-01 1.05650604e+00 2.57413425e-02 -3.71899307e-01
6.92091703e-01 -7.39768028e-01 -2.04473451e-01 -4.80065376e-01
7.10362911e-01 -5.98330736e-01 7.20533848e-01 2.98239052e-01
-1.22833490e-01 -6.96815848e-01 -9.94584084e-01 9.08408761e-01
-6.33052051e-01 5.42942166e-01 8.27822387e-01 4.46188092e-01
-6.86789393e-01 7.17274725e-01 -7.64522374e-01 -6.88451529e-01
6.26697123e-01 3.79122853e-01 -4.61431056e-01 -8.54437128e-02
-9.63876128e-01 1.12629259e+00 7.46797442e-01 5.80659032e-01
-2.75087237e-01 -6.70565307e-01 -7.86682665e-01 -2.60722816e-01
-4.59726006e-02 3.00370365e-01 8.20030570e-01 -1.51657450e+00
-1.10308146e+00 6.80507720e-01 -5.18090911e-02 -4.99870479e-01
3.00790101e-01 4.40725274e-02 -5.75107515e-01 3.80391508e-01
4.31561232e-01 5.42445958e-01 9.07808125e-01 -1.36040366e+00
-1.18130398e+00 -3.13857168e-01 -3.54239404e-01 2.07205504e-01
-1.88527003e-01 -1.09493367e-01 6.51366889e-01 -3.36157084e-01
5.60181081e-01 -1.05696344e+00 -2.59540230e-01 -6.91731200e-02
3.94251764e-01 2.99041957e-01 1.27774751e+00 -8.04407597e-01
8.54665875e-01 -1.99985886e+00 -4.02876079e-01 3.86613101e-01
-5.12085497e-01 6.77147210e-01 -2.45373715e-02 4.55394566e-01
-1.37772262e-01 -7.59566724e-02 -5.57258666e-01 5.46216011e-01
-8.95294487e-01 5.34373343e-01 -4.75967854e-01 4.35503989e-01
4.39677149e-01 3.71058255e-01 -8.42046738e-01 -3.26131761e-01
6.04287863e-01 2.74972886e-01 5.29988334e-02 1.22253798e-01
5.98039739e-02 5.55256069e-01 -2.99952775e-01 7.49406755e-01
7.67335296e-01 -1.92456320e-01 2.96784163e-01 -1.17446922e-01
-3.51580739e-01 -2.04774335e-01 -1.03190815e+00 9.17238891e-01
-6.01786137e-01 7.82709479e-01 -2.40564495e-01 -1.54328740e+00
1.28980803e+00 3.17920715e-01 4.52589661e-01 -4.05558884e-01
-3.04080993e-01 4.58940268e-01 1.19173743e-01 -7.05895126e-01
4.90235478e-01 -4.54329580e-01 1.64056525e-01 3.64110321e-02
-3.39878082e-01 -3.00028652e-01 -2.07500070e-01 -3.89979750e-01
1.89123258e-01 3.43746483e-01 5.59580624e-01 -2.48075157e-01
6.08920455e-01 8.76820922e-01 3.49996299e-01 7.20969737e-01
-2.49725744e-01 1.76090837e-01 -3.47960353e-01 -9.33605492e-01
-1.00033355e+00 -4.31832284e-01 -6.60396814e-01 9.32981551e-01
-2.91735977e-02 4.27732766e-01 -1.91252515e-01 -6.93588734e-01
2.17714071e-01 6.51846826e-01 -7.16189861e-01 1.59510076e-01
-3.07092667e-01 -1.13498163e+00 2.76899666e-01 4.56627280e-01
8.46680522e-01 -9.97976542e-01 -1.02934635e+00 4.41306174e-01
-5.33760041e-02 -9.27558899e-01 3.67254049e-01 4.13697183e-01
-1.46751475e+00 -7.88919926e-01 -3.73133242e-01 -4.97536898e-01
8.46622467e-01 6.57626092e-01 2.67155081e-01 2.17728391e-01
-2.11121678e-01 -1.07991315e-01 -6.40628338e-01 -4.10331130e-01
-5.57905614e-01 4.62849811e-02 -1.21925153e-01 2.97513276e-01
2.82795280e-01 -3.81460607e-01 -1.39074564e-01 4.98003572e-01
-7.30113328e-01 2.35363454e-01 7.78005719e-01 1.36522508e+00
5.16855597e-01 3.75286102e-01 7.93190777e-01 -7.08097816e-01
-1.87911659e-01 -6.09976530e-01 -6.78102314e-01 3.65723670e-01
-7.03682423e-01 -3.14170510e-01 6.23677611e-01 -5.30002534e-01
-1.20901287e+00 4.73537713e-01 2.38873154e-01 -1.22643478e-01
-4.01260734e-01 1.26404929e+00 5.58415830e-01 -4.93617177e-01
9.05279398e-01 6.67841077e-01 3.87099296e-01 -2.05891788e-01
-3.13788056e-01 1.16947138e+00 2.45493785e-01 -8.68412033e-02
7.15023100e-01 5.00677586e-01 1.86802313e-01 -1.35542107e+00
-1.02588928e+00 -7.40928233e-01 -1.11325037e+00 -5.74283242e-01
4.33334500e-01 -8.84956837e-01 4.21135873e-02 6.94555342e-01
-6.02034330e-01 -4.01844025e-01 4.21491303e-02 7.31975138e-01
-1.21259317e-01 9.99300927e-02 5.32497391e-02 -1.12446594e+00
-1.12224162e-01 -8.45046163e-01 8.03676844e-01 3.39837819e-01
-1.03267334e-01 -1.14014626e+00 -3.95548791e-01 5.47783792e-01
6.20558143e-01 4.40891653e-01 5.08057654e-01 -5.80546498e-01
-2.56941974e-01 -5.88380754e-01 -2.65963316e-01 4.78708655e-01
7.00600564e-01 4.92387675e-02 -1.04858208e+00 -3.73066396e-01
-4.78013046e-02 -2.87003994e-01 7.94516504e-01 2.77533710e-01
6.34259403e-01 -2.27537572e-01 -4.43891048e-01 5.02819002e-01
1.78830659e+00 4.67861593e-01 3.55002165e-01 8.29574406e-01
3.90564889e-01 7.62899637e-01 1.37141979e+00 3.45234573e-01
9.06626508e-02 4.65855300e-01 4.06430423e-01 -2.97442704e-01
4.55442220e-01 5.75345904e-02 1.52688138e-02 2.27696165e-01
-5.43731987e-01 3.31036359e-01 -1.18406665e+00 5.94370186e-01
-1.76381195e+00 -1.23964572e+00 -3.29925448e-01 2.17878795e+00
5.38675666e-01 -2.49325365e-01 -6.13222457e-02 6.41960800e-01
7.07668245e-01 1.03370138e-01 -4.45264786e-01 -3.29021066e-01
-1.33727819e-01 2.25890651e-01 8.30918968e-01 4.78253990e-01
-1.39838648e+00 1.06638539e+00 5.68160915e+00 6.12245321e-01
-1.77996647e+00 -1.20417394e-01 4.33171272e-01 2.98053890e-01
3.59340429e-01 1.47386312e-01 -6.80166364e-01 1.50531188e-01
9.80683386e-01 4.37383279e-02 2.56377071e-01 7.07008541e-01
4.13974494e-01 -9.66726184e-01 -4.61864442e-01 5.34976482e-01
-9.02464762e-02 -1.38745487e+00 9.41291004e-02 -5.26880063e-02
9.62280810e-01 1.06381048e-02 -2.70013750e-01 -2.43214332e-02
-1.87228426e-01 -8.21652353e-01 3.51902008e-01 4.86001760e-01
6.02706969e-01 -5.61410487e-01 1.05359840e+00 6.28418565e-01
-1.30700850e+00 -4.30316001e-01 -3.25717837e-01 -5.63432395e-01
-2.24287197e-01 3.96940649e-01 -1.17973244e+00 9.53828752e-01
8.13459456e-01 8.88274252e-01 -4.93256420e-01 8.09275508e-01
-7.59149119e-02 7.98669636e-01 -4.23917383e-01 -6.78663403e-02
5.67666173e-01 -2.63010472e-01 3.00821602e-01 9.59290147e-01
5.27672291e-01 4.60187852e-01 3.36463749e-01 1.51044160e-01
8.45427632e-01 1.19893417e-01 -7.71291554e-01 6.27041459e-02
5.33413231e-01 1.17325389e+00 -9.33705270e-01 -4.22730774e-01
-3.84974688e-01 9.10164356e-01 -1.16554402e-01 2.12752566e-01
-4.80947196e-01 -2.17278466e-01 2.55551517e-01 7.31973201e-02
5.47816634e-01 -1.82565942e-01 -5.08324683e-01 -7.87606299e-01
-4.87984493e-02 -7.09296644e-01 5.15978098e-01 -8.95966291e-01
-9.84061301e-01 7.72395253e-01 8.05840120e-02 -1.60841668e+00
-5.68292812e-02 -5.13273299e-01 -3.68553817e-01 1.27530324e+00
-1.73815167e+00 -1.48240089e+00 -5.77573478e-01 3.37557048e-01
5.13618588e-01 -3.59645009e-01 1.04356682e+00 -3.11480433e-01
-2.14976922e-01 4.91541661e-02 4.11430299e-01 -1.09300137e-01
2.46891871e-01 -8.25552046e-01 -3.56855303e-01 9.89926755e-01
-1.62277594e-01 1.29314870e-01 4.80274737e-01 -8.44991744e-01
-9.33654964e-01 -1.40479875e+00 1.03738439e+00 9.00412574e-02
6.91240072e-01 2.75194019e-01 -1.11109245e+00 5.62579989e-01
-2.81053662e-01 2.10978508e-01 8.54754865e-01 -3.17062557e-01
3.94545384e-02 -6.23785973e-01 -1.43291068e+00 8.16812087e-03
2.16895446e-01 -5.53108275e-01 -6.12043083e-01 2.65919358e-01
-9.62858647e-02 -1.85989276e-01 -1.02007270e+00 5.30644715e-01
8.49874556e-01 -7.54391968e-01 7.42842436e-01 -3.03768337e-01
4.48939443e-01 -4.20914829e-01 -5.09230375e-01 -1.41710865e+00
-9.96039659e-02 6.33935109e-02 7.03239620e-01 5.60043514e-01
6.71177387e-01 -9.97574806e-01 5.84933579e-01 3.94801855e-01
-4.33416758e-03 -5.16485035e-01 -8.13400209e-01 -1.06411529e+00
-1.98811203e-01 -1.92016348e-01 3.18077832e-01 1.30161893e+00
-1.86569005e-01 -1.11641742e-01 -2.99798667e-01 8.18376839e-01
6.12358630e-01 5.08532941e-01 6.12901270e-01 -1.53745353e+00
-2.47534662e-02 2.59853583e-02 -5.60110807e-01 -3.66768509e-01
-6.47797361e-02 -9.65786576e-01 -6.78309202e-02 -1.39497304e+00
9.22186114e-03 -8.39385092e-01 -8.19749981e-02 1.06753838e+00
-2.75711149e-01 4.57374245e-01 7.90931508e-02 6.02152407e-01
5.24239600e-01 2.15790600e-01 1.08039820e+00 -1.62736624e-02
-5.24938881e-01 3.53770196e-01 -2.33737737e-01 4.26834047e-01
1.01742601e+00 -5.47696650e-01 -3.33523482e-01 -9.99960601e-02
-2.39400864e-01 3.50299001e-01 5.54021001e-01 -1.12422049e+00
6.80754930e-02 -6.62326455e-01 6.48409307e-01 -7.13875294e-01
2.23313183e-01 -1.04628789e+00 5.42411804e-01 7.99205601e-01
3.75575246e-03 -2.96918154e-01 7.35008299e-01 2.94379205e-01
-4.48777199e-01 -2.28951648e-01 1.06267500e+00 -6.02317564e-02
-1.12161863e+00 4.51843366e-02 -5.70120096e-01 -8.32009196e-01
1.38282323e+00 -9.32161152e-01 4.48338315e-02 5.71554378e-02
-9.05136168e-01 -4.62494902e-02 1.49760559e-01 6.89648837e-02
5.34508228e-01 -9.91176188e-01 -8.91451180e-01 4.74171788e-01
2.53326416e-01 -2.82274097e-01 2.97426909e-01 9.93179500e-01
-5.84978044e-01 3.37391496e-01 -6.00638747e-01 -1.02550495e+00
-1.59076738e+00 1.68652683e-01 4.45551991e-01 -3.47289331e-02
-5.85585892e-01 5.52514613e-01 -4.80512559e-01 -4.49491590e-01
-5.88987291e-01 -1.45239159e-01 -3.28630269e-01 5.39833307e-01
2.78837711e-01 1.69075340e-01 2.94724792e-01 -9.33909595e-01
-4.40677732e-01 5.96768618e-01 -5.84794357e-02 -1.72428682e-01
1.88367593e+00 1.66634321e-02 -4.01316732e-02 6.12204134e-01
9.51129615e-01 -6.05042160e-01 -1.29426610e+00 -3.63335431e-01
-1.15203727e-02 -9.53438699e-01 4.70230758e-01 -9.73586559e-01
-9.26171064e-01 7.67138720e-01 7.85703361e-01 3.32535297e-01
1.46883130e+00 -3.66085261e-01 1.61895216e-01 6.00798965e-01
5.19613445e-01 -1.04087639e+00 -3.62994313e-01 3.83717448e-01
8.66248369e-01 -1.55758703e+00 2.19114035e-01 -3.70098263e-01
-6.81502581e-01 1.46890032e+00 2.75956064e-01 1.44133985e-01
6.90921724e-01 -1.70247659e-01 3.68870825e-01 -4.09039930e-02
-3.03792566e-01 -2.55677372e-01 6.38553351e-02 8.03563952e-01
1.44511580e-01 2.67145038e-01 -4.62053753e-02 -5.95669523e-02
3.71460766e-02 2.23937482e-01 4.99181390e-01 1.24472702e+00
-7.48726547e-01 -7.52393782e-01 -6.71133041e-01 8.48323345e-01
-6.94401637e-02 -2.11152941e-01 1.94145180e-02 8.57410133e-01
5.46725653e-02 1.04366350e+00 1.72190666e-01 -4.45486844e-01
4.63754982e-02 1.17293149e-01 2.75088578e-01 -5.88215470e-01
-6.17669582e-01 -7.58189857e-02 1.80159733e-01 -9.24532413e-02
-1.07307911e+00 -1.24775493e+00 -8.08529794e-01 -7.23911598e-02
-7.17545807e-01 2.66425312e-01 8.54264498e-01 1.09641349e+00
2.35336889e-02 1.37980491e-01 1.17297328e+00 -1.25059545e+00
-6.81938410e-01 -1.14505816e+00 -7.14604080e-01 -8.16930234e-02
4.02114600e-01 -6.18775070e-01 -5.41137159e-01 2.92381853e-01] | [9.628179550170898, -1.5228917598724365] |
e7446314-e882-48d1-9f55-72f86bbd83a8 | a-dynamic-spatial-temporal-attention-network | 2106.10197 | null | https://arxiv.org/abs/2106.10197v2 | https://arxiv.org/pdf/2106.10197v2.pdf | A Dynamic Spatial-temporal Attention Network for Early Anticipation of Traffic Accidents | The rapid advancement of sensor technologies and artificial intelligence are creating new opportunities for traffic safety enhancement. Dashboard cameras (dashcams) have been widely deployed on both human driving vehicles and automated driving vehicles. A computational intelligence model that can accurately and promptly predict accidents from the dashcam video will enhance the preparedness for accident prevention. The spatial-temporal interaction of traffic agents is complex. Visual cues for predicting a future accident are embedded deeply in dashcam video data. Therefore, the early anticipation of traffic accidents remains a challenge. Inspired by the attention behavior of humans in visually perceiving accident risks, this paper proposes a Dynamic Spatial-Temporal Attention (DSTA) network for the early accident anticipation from dashcam videos. The DSTA-network learns to select discriminative temporal segments of a video sequence with a Dynamic Temporal Attention (DTA) module. It also learns to focus on the informative spatial regions of frames with a Dynamic Spatial Attention (DSA) module. A Gated Recurrent Unit (GRU) is trained jointly with the attention modules to predict the probability of a future accident. The evaluation of the DSTA-network on two benchmark datasets confirms that it has exceeded the state-of-the-art performance. A thorough ablation study that assesses the DSTA-network at the component level reveals how the network achieves such performance. Furthermore, this paper proposes a method to fuse the prediction scores from two complementary models and verifies its effectiveness in further boosting the performance of early accident anticipation. | ['Zhaozheng Yin', 'Ruwen Qin', 'Yu Li', 'Muhammad Monjurul Karim'] | 2021-06-18 | null | null | null | null | ['action-anticipation', 'accident-anticipation'] | ['computer-vision', 'computer-vision'] | [ 2.16864109e-01 -1.20131835e-01 -2.02073395e-01 -2.45002389e-01
-5.72636008e-01 1.70402214e-01 6.58139706e-01 -3.90537441e-01
-5.15480936e-01 2.36999914e-01 3.76639247e-01 -4.03240740e-01
-8.23618472e-02 -5.12436628e-01 -5.12885809e-01 -7.64045179e-01
-2.22916584e-02 -2.55775452e-01 6.74079239e-01 -3.16378385e-01
1.00764677e-01 6.01705909e-01 -2.02669048e+00 6.01805329e-01
5.02603590e-01 1.15698826e+00 5.77311456e-01 8.64355803e-01
4.11180884e-01 1.42058659e+00 -2.76070476e-01 -2.00911358e-01
2.10409418e-01 -2.84011662e-01 -2.33868644e-01 -1.42311811e-01
3.37018967e-01 -5.18281758e-01 -1.28998923e+00 6.54942155e-01
4.53782678e-01 4.14617568e-01 4.65926319e-01 -1.44895387e+00
-4.96124566e-01 -1.25816688e-02 -4.89388943e-01 8.40578735e-01
2.11181238e-01 5.36151648e-01 5.83516836e-01 -7.69581914e-01
4.12317365e-01 1.13641465e+00 4.81925547e-01 6.13000214e-01
-3.69205207e-01 -6.46176815e-01 3.55742335e-01 1.21212816e+00
-1.24559546e+00 -2.38025099e-01 8.70625615e-01 -6.08286381e-01
1.33406663e+00 3.41939539e-01 6.55718565e-01 1.30123222e+00
6.16779685e-01 1.02224243e+00 5.74254274e-01 2.15069111e-02
-3.87391895e-02 -2.43817508e-01 2.49096736e-01 5.77398837e-01
-1.90106392e-01 9.02312398e-01 -6.01859987e-01 5.40552378e-01
4.66467679e-01 6.22793972e-01 1.31124899e-01 2.33577654e-01
-1.02432597e+00 5.29823661e-01 6.18726611e-01 1.59824908e-01
-8.07369292e-01 2.15002581e-01 4.90245789e-01 6.83808699e-02
2.85909832e-01 -6.91317841e-02 1.51641127e-02 -3.10912073e-01
-6.06463253e-01 2.07140129e-02 5.65107539e-02 7.53923476e-01
4.96587306e-01 1.44690782e-01 -7.14056849e-01 5.25432110e-01
1.29615173e-01 6.22506261e-01 2.24160314e-01 -9.67900217e-01
4.93156284e-01 5.84271491e-01 -1.74725547e-01 -1.25672615e+00
-4.64263707e-01 -2.20219269e-01 -7.03172624e-01 4.79970694e-01
-2.99582141e-03 -2.29070798e-01 -8.77064764e-01 1.44766581e+00
2.08150312e-01 8.04508328e-01 -3.26905139e-02 1.16777658e+00
6.10890508e-01 9.66832280e-01 7.32946932e-01 -1.56083509e-01
1.28250837e+00 -1.04489589e+00 -8.80243063e-01 -4.84744757e-01
5.53435743e-01 -2.98993260e-01 5.52096069e-01 3.08448374e-01
-9.24509883e-01 -1.14265883e+00 -9.05612409e-01 -1.56174272e-01
-4.71427560e-01 6.99516982e-02 3.65255773e-01 1.99053615e-01
-7.73476839e-01 1.51326805e-01 -7.69113839e-01 -2.21249625e-01
5.38614869e-01 3.00856352e-01 -3.52023154e-01 -2.94198155e-01
-1.33068895e+00 1.17830217e+00 2.08038151e-01 2.71131068e-01
-1.10344672e+00 -5.76724172e-01 -8.67235422e-01 5.49447313e-02
3.21103036e-01 -5.90066373e-01 1.06372201e+00 -1.03232968e+00
-1.21308720e+00 4.65169519e-01 -3.12216461e-01 -8.77803326e-01
2.83755243e-01 -4.39052820e-01 -8.31716835e-01 3.85560811e-01
-1.10545911e-01 8.67420673e-01 9.92654085e-01 -8.02912235e-01
-1.19560397e+00 -1.60973355e-01 -7.17986599e-02 2.19789952e-01
-3.39100122e-01 3.07272077e-01 -5.11165679e-01 -6.44246221e-01
-5.79822242e-01 -8.52296591e-01 -2.56543994e-01 -2.37124652e-01
8.76026042e-03 -5.72373271e-01 1.55207980e+00 -8.40304255e-01
1.54474556e+00 -2.35668254e+00 -1.58482473e-02 -1.83780849e-01
2.74709404e-01 7.64024377e-01 -2.70188838e-01 2.08311453e-01
-2.31784135e-01 -4.52823102e-01 1.20468855e-01 -2.86948889e-01
-1.14476591e-01 2.78546005e-01 -5.19994140e-01 2.62724280e-01
5.28006196e-01 9.59100008e-01 -9.19734895e-01 -4.06405598e-01
9.41994071e-01 8.24287951e-01 -4.72218007e-01 5.80243111e-01
8.38082582e-02 4.31633919e-01 -5.30312240e-01 3.12501460e-01
3.26569259e-01 8.88251886e-02 -3.62639070e-01 -1.17365830e-01
-4.29872096e-01 -3.39521728e-02 -5.11726141e-01 1.15984201e+00
-2.49859646e-01 1.17801940e+00 -5.80221832e-01 -1.01959252e+00
7.29650736e-01 4.69758570e-01 4.45538819e-01 -1.47617459e+00
1.92022234e-01 -2.25193664e-01 -2.25488804e-02 -1.29372096e+00
3.28808874e-01 1.38392448e-01 1.91079937e-02 -1.08191893e-01
-1.67712316e-01 6.47356272e-01 1.86238930e-01 1.43447518e-01
1.39366913e+00 5.53053105e-03 -6.40036017e-02 3.63679200e-01
7.18702376e-01 -6.95691481e-02 4.97325510e-01 6.67270005e-01
-6.73585534e-01 6.28610313e-01 2.24773228e-01 -9.56397533e-01
-8.76103878e-01 -9.31865811e-01 3.36629629e-01 1.33044720e+00
3.37303609e-01 -1.62963063e-01 -7.45778561e-01 -7.17214465e-01
-2.69160390e-01 1.03567910e+00 -8.15527916e-01 -7.18298256e-01
-7.27435946e-01 -3.80238205e-01 1.79412827e-01 9.70903456e-01
6.18053257e-01 -1.28058338e+00 -1.03180671e+00 6.99030310e-02
-1.61427185e-01 -1.48132551e+00 -4.33716387e-01 -3.03044945e-01
-2.49962419e-01 -1.12062144e+00 -4.96759921e-01 -7.11881518e-01
4.23522353e-01 7.57770121e-01 6.90527618e-01 1.39495254e-01
-2.94355303e-01 5.38999856e-01 -3.37177902e-01 -6.51652396e-01
-1.58913702e-01 -3.62806529e-01 1.94506943e-02 4.51142132e-01
8.18617761e-01 -1.79090172e-01 -8.17046106e-01 4.01172459e-01
-6.55039489e-01 5.22414386e-01 7.07197070e-01 4.51583058e-01
4.16582853e-01 4.77508567e-02 6.46580219e-01 -2.06583291e-01
1.75474465e-01 -9.22619522e-01 -4.44377095e-01 -2.05611419e-02
-5.98944724e-02 -4.53668982e-01 4.64265406e-01 -2.49418557e-01
-1.29489517e+00 1.65815696e-01 -3.54814261e-01 -8.50870848e-01
-5.29871225e-01 2.61842608e-01 1.36319399e-01 2.70427167e-01
2.14742899e-01 4.19652648e-02 -4.44683917e-02 -6.97293654e-02
2.30307635e-02 6.67006850e-01 6.96203351e-01 2.99836546e-01
4.18109089e-01 4.59006190e-01 -1.00396834e-01 -9.68198657e-01
-9.62729514e-01 -6.98922813e-01 -3.90690297e-01 -1.02443182e+00
1.43411148e+00 -9.91974473e-01 -9.60102856e-01 4.84376490e-01
-1.22236001e+00 -3.09342772e-01 -2.23583579e-01 7.74485648e-01
-5.37056029e-01 1.28096685e-01 -4.03896958e-01 -9.44810808e-01
5.86326793e-02 -1.04879224e+00 1.14493430e+00 4.21327949e-01
-1.61746040e-03 -7.72349119e-01 2.11242102e-02 6.00327432e-01
2.32109889e-01 1.10197999e-01 4.90542740e-01 -3.76433879e-01
-8.29778254e-01 -5.58555424e-01 -3.78010243e-01 3.08156312e-01
-1.97799131e-01 1.50668826e-02 -1.14528990e+00 1.73632115e-01
1.59153864e-02 2.12221429e-01 1.11187851e+00 7.54540563e-01
1.45293128e+00 -7.58805349e-02 -4.68610317e-01 4.35645491e-01
7.94345379e-01 7.19447613e-01 1.13956141e+00 2.25654855e-01
7.90219724e-01 8.12066734e-01 8.46714795e-01 2.81724066e-01
5.91997862e-01 7.95399249e-01 7.46480048e-01 -2.58253157e-01
-4.20574099e-01 -2.70965010e-01 7.62709200e-01 6.42757475e-01
-3.45883816e-01 -5.10809660e-01 -9.28473771e-01 5.56799889e-01
-2.20646715e+00 -1.52505207e+00 -1.57784432e-01 1.94605184e+00
-2.83587724e-02 1.86977029e-01 2.58414865e-01 2.71920979e-01
8.25662374e-01 1.89266399e-01 -5.57453215e-01 -5.22557437e-01
-7.90455267e-02 -2.63279229e-01 3.39674026e-01 3.01232100e-01
-1.22786057e+00 8.61775637e-01 5.88088989e+00 8.44762802e-01
-1.12024128e+00 2.86436290e-01 8.15976620e-01 -1.73562497e-01
2.75218815e-01 -4.79083270e-01 -7.48083472e-01 6.87013805e-01
1.39265549e+00 7.04491735e-02 1.26818687e-01 6.80391431e-01
7.68272758e-01 -1.11929134e-01 -8.01311433e-01 9.75763738e-01
2.33673409e-01 -1.49804950e+00 1.15199707e-01 -1.73638053e-02
5.67741573e-01 3.17422122e-01 3.55846643e-01 5.81210315e-01
-3.41245458e-02 -1.09948707e+00 5.93482256e-01 1.06248868e+00
5.62686563e-01 -8.61331284e-01 8.39799345e-01 3.78251672e-01
-1.29710579e+00 -7.34641850e-01 -2.10587338e-01 -2.89770484e-01
5.02537787e-01 1.57099396e-01 -5.49965560e-01 2.72549838e-01
8.58845472e-01 1.17695379e+00 -6.89863682e-01 1.02578187e+00
-2.18034685e-01 6.88810289e-01 9.73524600e-02 1.13945611e-01
5.51769495e-01 6.48658201e-02 6.91849887e-01 1.30945182e+00
3.89826268e-01 4.32949036e-01 2.33269371e-02 4.18070793e-01
3.85904253e-01 -4.98441607e-01 -9.44197178e-01 9.40101445e-02
-4.23687659e-02 1.08303368e+00 -1.87664628e-01 -4.42642003e-01
-7.62355506e-01 6.48240387e-01 1.26771346e-01 5.50940931e-01
-1.24233770e+00 -1.70291945e-01 9.07493412e-01 2.41460785e-01
4.82534021e-01 -2.80472249e-01 -1.76709145e-01 -4.83746171e-01
-2.16039076e-01 -3.51372570e-01 6.98631346e-01 -1.24271750e+00
-1.04253578e+00 8.05605888e-01 -2.72610527e-03 -1.55890691e+00
-2.21558571e-01 -7.44634151e-01 -8.79660606e-01 5.47211826e-01
-1.74945831e+00 -1.02836573e+00 -5.11647701e-01 7.58228660e-01
1.02267957e+00 -3.76329511e-01 2.15207860e-01 5.15359044e-01
-1.05718911e+00 3.15543264e-01 -5.34961820e-01 1.10241778e-01
2.89953351e-01 -6.24479473e-01 2.78046727e-01 1.12107873e+00
-2.85633266e-01 -4.16120775e-02 5.67875624e-01 -6.55258179e-01
-1.33641350e+00 -1.49796510e+00 1.02418995e+00 -5.19909978e-01
4.68315780e-01 8.80756378e-02 -7.76739419e-01 7.13532925e-01
4.90382463e-01 1.77055195e-01 4.18203205e-01 -5.33929706e-01
-8.60514268e-02 -3.06351483e-01 -6.86306834e-01 7.73702443e-01
1.03128564e+00 -6.12288177e-01 -6.75594747e-01 2.30967864e-01
7.84785688e-01 -1.62123308e-01 -2.57277846e-01 4.02443767e-01
4.19210255e-01 -1.13312054e+00 1.10128701e+00 -7.71443188e-01
5.23169339e-01 -4.01554137e-01 3.27662602e-02 -8.38802695e-01
-6.05884552e-01 -6.06328845e-01 -5.30940890e-01 6.27537787e-01
-6.83567747e-02 -4.78271484e-01 3.49735945e-01 7.61281967e-01
-7.58458674e-01 -6.47796512e-01 -1.22084367e+00 -5.00521839e-01
-4.73395586e-01 -9.34848547e-01 2.90383160e-01 3.33294779e-01
1.60313933e-03 3.19949627e-01 -6.56737089e-01 4.11184132e-01
3.52733165e-01 -4.69917357e-01 4.24104720e-01 -9.52937186e-01
2.96532601e-01 -4.77883518e-01 -8.88143837e-01 -1.04047740e+00
1.66623846e-01 -5.37828505e-01 1.03094734e-01 -1.39867234e+00
1.40567809e-01 1.14499532e-01 -7.75976300e-01 3.18089366e-01
-3.48336846e-01 2.09821090e-01 2.72167653e-01 -1.31546617e-01
-1.03130877e+00 6.33845389e-01 1.09896815e+00 -1.10034198e-01
6.73969090e-02 -5.75403571e-02 -3.56313676e-01 7.64163554e-01
7.07659125e-01 -2.74417430e-01 -4.73243445e-01 -4.97635514e-01
-3.65435556e-02 3.18656355e-01 8.26965749e-01 -1.53968358e+00
7.17699170e-01 -2.58122534e-01 2.51067936e-01 -9.11235213e-01
5.30588627e-01 -1.05317652e+00 1.89193953e-02 5.22595823e-01
-3.76906991e-01 2.57135212e-01 4.37563747e-01 8.78634036e-01
-2.96585530e-01 3.36167157e-01 6.84648037e-01 3.85862350e-01
-1.53618479e+00 5.90370715e-01 -1.05333662e+00 -2.57411808e-01
1.59162462e+00 -4.83640075e-01 -4.21919882e-01 -3.39933753e-01
-5.28035939e-01 4.23436671e-01 -2.70035356e-01 1.08876085e+00
1.08860540e+00 -1.44405043e+00 -6.16446912e-01 5.85195601e-01
7.22302049e-02 -4.34867561e-01 9.89745796e-01 1.07391334e+00
-1.83524370e-01 6.46366298e-01 -2.93501735e-01 -7.12255955e-01
-1.12374258e+00 9.27088678e-01 3.24017018e-01 1.03909923e-02
-8.31925869e-01 7.47560441e-01 6.80680990e-01 4.05945837e-01
3.54478210e-01 -8.83598551e-02 -7.64167249e-01 -1.29361525e-01
1.05915368e+00 6.53102994e-01 -1.80013254e-01 -1.10068035e+00
-2.80616492e-01 4.24803019e-01 6.67747632e-02 3.16233337e-01
1.35452712e+00 -3.62535983e-01 4.96695459e-01 1.70660987e-01
1.09424555e+00 -5.77799380e-01 -1.93005455e+00 -6.08523712e-02
-1.12008065e-01 -4.62450176e-01 4.05368268e-01 -7.30466902e-01
-1.40076160e+00 1.25930440e+00 8.87109339e-01 8.24447572e-02
1.30878377e+00 -1.29444152e-01 1.27263772e+00 1.52975067e-01
-5.04486784e-02 -1.18031728e+00 2.43765533e-01 5.72700024e-01
1.03520107e+00 -1.35562873e+00 -5.60425818e-01 -1.57441854e-01
-1.13826346e+00 1.00699627e+00 7.72915840e-01 -1.18472286e-01
7.14884222e-01 1.95008427e-01 -5.48246875e-02 -2.28565961e-01
-1.14075339e+00 -5.12854755e-01 6.61731482e-01 7.13719666e-01
-9.91123095e-02 -4.98890541e-02 2.67023087e-01 6.17953420e-01
3.76037925e-01 1.69769675e-01 6.67852461e-02 5.88281095e-01
-4.54362810e-01 -3.73470575e-01 -1.53883651e-01 2.51958966e-01
-8.62321109e-02 1.11602515e-01 -2.95398384e-02 4.64971513e-01
4.36904103e-01 1.26481783e+00 5.29644549e-01 -8.72861385e-01
4.54956710e-01 -1.85917392e-01 -9.73612070e-02 -1.69952556e-01
-5.09167850e-01 -2.96327800e-01 -3.35088335e-02 -8.54372978e-01
-4.62213665e-01 -7.87584126e-01 -1.18319070e+00 -2.93993831e-01
2.17615113e-01 -1.84559673e-01 3.40539336e-01 1.00626230e+00
7.36618042e-01 1.03933752e+00 7.42288470e-01 -1.15903878e+00
2.17243448e-01 -5.29495358e-01 2.33155768e-02 3.29627335e-01
7.01525390e-01 -8.61326694e-01 -1.77689806e-01 1.46167323e-01] | [7.524539470672607, 0.1062919944524765] |
4ab912f3-b638-4877-be39-3e527652ed3c | the-first-cross-lingual-challenge-on | null | null | https://aclanthology.org/W17-1412 | https://aclanthology.org/W17-1412.pdf | The First Cross-Lingual Challenge on Recognition, Normalization, and Matching of Named Entities in Slavic Languages | This paper describes the outcomes of the first challenge on multilingual named entity recognition that aimed at recognizing mentions of named entities in web documents in Slavic languages, their normalization/lemmatization, and cross-language matching. It was organised in the context of the 6th Balto-Slavic Natural Language Processing Workshop, co-located with the EACL 2017 conference. Although eleven teams signed up for the evaluation, due to the complexity of the task(s) and short time available for elaborating a solution, only two teams submitted results on time. The reported evaluation figures reflect the relatively higher level of complexity of named entity-related tasks in the context of processing texts in Slavic languages. Since the duration of the challenge goes beyond the date of the publication of this paper and updated picture of the participating systems and their corresponding performance can be found on the web page of the challenge. | ['Jan {\\v{S}}najder', 'Lidia Pivovarova', 'Jakub Piskorski', 'Roman Yangarber', 'Josef Steinberger'] | 2017-04-01 | null | null | null | ws-2017-4 | ['multilingual-named-entity-recognition'] | ['natural-language-processing'] | [-9.26475972e-02 8.35381746e-02 4.99909483e-02 -5.12552977e-01
-8.92362297e-01 -1.05585742e+00 9.62132454e-01 5.73238850e-01
-9.68913257e-01 7.57156193e-01 5.19740462e-01 -4.90827322e-01
6.20508045e-02 -3.47785383e-01 -2.66864419e-01 -2.42427979e-02
5.45201860e-02 6.53933048e-01 1.03111438e-01 -3.22757244e-01
3.69971335e-01 7.10825562e-01 -1.00265419e+00 5.67622244e-01
5.81765115e-01 4.29205686e-01 -8.91182646e-02 4.83028054e-01
-6.94626093e-01 4.57597196e-01 -6.53101504e-01 -7.96127319e-01
2.58416891e-01 -3.18907410e-01 -1.39176166e+00 -1.83273077e-01
4.08701897e-01 5.67794025e-01 2.03834064e-02 9.72138286e-01
5.65558255e-01 2.32000440e-01 5.41882157e-01 -7.95543849e-01
-8.24529901e-02 9.46436644e-01 -4.78959903e-02 2.57436395e-01
5.76030970e-01 -2.86971986e-01 1.18397009e+00 -1.14876032e+00
1.27557516e+00 8.13874304e-01 5.70891142e-01 3.11570525e-01
-6.61063612e-01 -5.31123817e-01 2.07109183e-01 1.48477837e-01
-1.43116951e+00 -9.13807988e-01 9.30801705e-02 -5.63202679e-01
1.22043931e+00 -1.04765585e-02 1.08443066e-01 5.07717788e-01
-3.10437471e-01 8.87228847e-01 1.01747715e+00 -7.95435190e-01
3.12137544e-01 4.50735420e-01 2.52144724e-01 2.08945543e-01
5.25330722e-01 -2.86928236e-01 -4.75818813e-01 -2.36743823e-01
2.12287664e-01 -8.75011086e-01 1.23750955e-01 -3.14160198e-01
-1.24939203e+00 5.74101925e-01 -9.80691314e-02 1.10163486e+00
-4.52923059e-01 -3.97992730e-01 1.00700510e+00 3.64247322e-01
3.34216297e-01 5.74471176e-01 -7.48739719e-01 -2.62717336e-01
-9.97983634e-01 8.93167332e-02 1.04736888e+00 1.04424918e+00
4.59232181e-01 -1.20428994e-01 -6.05945475e-03 7.03353226e-01
-1.20162880e-02 3.04931134e-01 4.55024838e-01 -3.73149335e-01
6.94605410e-01 7.48277485e-01 1.95287347e-01 -5.22809207e-01
-4.43915755e-01 -2.08728507e-01 -2.49827653e-01 -1.18561551e-01
6.94746077e-01 -5.64367354e-01 -6.10670269e-01 1.35320222e+00
3.92683566e-01 -3.18595320e-01 4.06460822e-01 7.31729686e-01
1.03140962e+00 5.85077107e-01 3.23169529e-01 -3.94355208e-01
1.63297689e+00 -5.16913712e-01 -7.92900324e-01 -7.56043121e-02
8.16995978e-01 -1.22423136e+00 3.01278681e-01 -2.34732181e-02
-1.00836432e+00 -1.85146630e-01 -8.75193477e-01 -1.04458131e-01
-9.66998458e-01 1.93685651e-01 5.74746430e-01 7.45544434e-01
-7.49160588e-01 1.54323563e-01 -6.42268538e-01 -7.63384759e-01
-2.67137825e-01 1.38186209e-04 -6.82737291e-01 4.79070023e-02
-1.44204640e+00 1.18565309e+00 5.76706350e-01 8.81705657e-02
-1.80631075e-02 -6.08063877e-01 -9.62165952e-01 -2.71275967e-01
4.15948451e-01 4.46674526e-02 1.08151770e+00 -8.07679236e-01
-1.11221182e+00 1.47839701e+00 -1.14809409e-01 -4.21832472e-01
7.44345963e-01 -2.32356817e-01 -8.28838944e-01 -1.54118493e-01
3.77806246e-01 1.57840490e-01 1.86611831e-01 -7.00064480e-01
-1.00412869e+00 -4.50029254e-01 -4.41582687e-02 2.18804196e-01
2.52745718e-01 1.01419258e+00 -4.47103828e-01 -4.94999856e-01
1.89383999e-02 -8.49852562e-01 -1.25650913e-01 -1.19807506e+00
7.34077320e-02 -4.10135567e-01 3.20657492e-01 -9.22390044e-01
1.22493172e+00 -1.95686412e+00 -1.36089385e-01 1.26805454e-01
-3.38998109e-01 5.95239997e-01 -1.75407017e-03 1.11922944e+00
-4.02796477e-01 3.34608346e-01 -1.53233275e-01 -1.26553848e-01
1.68336838e-01 -2.70246267e-01 -3.52308363e-01 6.15509689e-01
1.87793195e-01 8.50986063e-01 -8.50361645e-01 -4.20425236e-01
2.73452103e-02 3.66865098e-01 3.10614407e-01 -3.14054102e-01
7.18740821e-02 1.05912723e-01 -2.11789906e-01 2.08979324e-01
5.81008375e-01 5.23391664e-01 5.54470658e-01 -4.62760217e-02
-7.79420495e-01 1.03103197e+00 -1.63251042e+00 1.49534237e+00
-5.68349302e-01 5.69905639e-01 3.50427568e-01 -6.76558733e-01
7.02371180e-01 8.64742279e-01 3.23789299e-01 -5.18649101e-01
6.69197515e-02 6.98644638e-01 -1.52846724e-01 -1.03835963e-01
8.75956953e-01 -2.66992822e-02 -4.43605989e-01 4.51515287e-01
3.13224912e-01 -5.28749004e-02 8.52901161e-01 1.77267075e-01
9.12083745e-01 1.90845922e-01 8.45407307e-01 -4.27848518e-01
1.05552566e+00 3.35643291e-01 6.70584679e-01 3.78092349e-01
-4.91834253e-01 2.91606545e-01 3.21118444e-01 -4.08208281e-01
-1.04029751e+00 -5.70523322e-01 -2.86129743e-01 1.16564965e+00
-4.43133950e-01 -6.40335321e-01 -4.31373894e-01 -8.19492817e-01
-3.69236320e-01 7.22743988e-01 -3.58873039e-01 6.56585157e-01
-9.86958146e-01 -1.99172989e-01 1.06625307e+00 9.49207172e-02
1.09809302e-01 -1.48930705e+00 -3.52766722e-01 3.29742998e-01
-1.97567627e-01 -1.49149680e+00 -3.62072915e-01 1.52885780e-01
-5.20047963e-01 -1.16525531e+00 -7.49916792e-01 -1.06886828e+00
3.11108172e-01 -3.74842763e-01 9.43892002e-01 -2.03865856e-01
-1.57442600e-01 3.05005699e-01 -3.73122066e-01 -6.42962873e-01
-4.99707192e-01 7.07928061e-01 -9.33995396e-02 -2.74185121e-01
5.36638677e-01 5.49352318e-02 5.03354967e-02 -5.36591560e-02
-8.70547473e-01 -2.82438934e-01 5.03709912e-01 4.83637124e-01
1.10317670e-01 -3.64714980e-01 5.75627565e-01 -1.24830914e+00
5.56697369e-01 -1.70902848e-01 -7.86095142e-01 5.41882753e-01
-4.34113026e-01 -1.13895889e-02 3.36701155e-01 1.75423071e-01
-1.26573193e+00 2.91126341e-01 -2.41826013e-01 4.23026741e-01
-7.26793826e-01 6.71021581e-01 -1.60647556e-01 1.79804280e-01
3.53817731e-01 1.06982440e-02 -6.82636559e-01 -4.70703453e-01
5.11337638e-01 7.98176885e-01 5.86076379e-01 -4.63518023e-01
8.39104235e-01 2.43927240e-02 -2.80192316e-01 -9.15236413e-01
-5.88556647e-01 -1.23400593e+00 -1.08875704e+00 2.47475818e-05
7.44841158e-01 -1.04166710e+00 -3.05768132e-01 4.70174521e-01
-1.34803128e+00 1.21313959e-01 -1.66868672e-01 5.34428000e-01
-1.24088034e-01 4.16893303e-01 -7.35371232e-01 -9.07351851e-01
-5.13102293e-01 -4.98985678e-01 6.67868137e-01 -1.31301978e-03
-4.13842767e-01 -1.22279072e+00 3.87298375e-01 5.27659595e-01
2.74487853e-01 9.66516975e-03 7.83180356e-01 -1.23548901e+00
8.51915330e-02 -5.49136817e-01 -1.63455084e-01 -3.86829115e-02
-1.21040441e-01 -8.05843715e-03 -7.79751837e-01 -1.06955610e-01
-3.29637110e-01 -3.58196013e-02 4.45981294e-01 -2.20041797e-01
-3.26468050e-01 -1.19483523e-01 -4.10360135e-02 -1.01190172e-01
1.39956832e+00 1.37212694e-01 4.59615856e-01 6.79531097e-01
2.41741583e-01 1.08551753e+00 6.21222019e-01 5.97138591e-02
4.77290541e-01 6.71972990e-01 -3.08724135e-01 1.00742348e-01
-1.65564805e-01 3.90407071e-02 3.88176769e-01 1.03733337e+00
5.10375164e-02 1.89325601e-01 -1.39189565e+00 9.50047016e-01
-1.66464937e+00 -8.03622425e-01 -5.53610325e-01 2.17429209e+00
6.57485127e-01 -5.44289651e-04 2.32231736e-01 -2.56335288e-01
9.81024981e-01 2.37611756e-01 1.88462242e-01 -7.70130455e-01
-3.96099240e-01 3.69619519e-01 6.94633424e-01 7.09246695e-01
-1.19034171e+00 1.37364769e+00 5.96536493e+00 6.66642189e-01
-1.01487470e+00 1.29666120e-01 8.26386586e-02 2.88824052e-01
6.01005927e-02 2.89153129e-01 -1.07659066e+00 -8.68203938e-02
1.26311195e+00 -5.30369699e-01 6.23230338e-02 4.99918431e-01
1.69286042e-01 -1.23588927e-01 -8.00265193e-01 6.99762881e-01
1.01142555e-01 -1.11226785e+00 -2.37566471e-01 -1.36338845e-01
7.88149595e-01 5.67656696e-01 -5.33118010e-01 4.52345222e-01
2.82916546e-01 -7.57610798e-01 8.60465944e-01 9.71477851e-02
5.87393761e-01 -7.14003384e-01 1.20300138e+00 2.15609029e-01
-1.24518728e+00 3.59219074e-01 -1.47812098e-01 -5.37143722e-02
4.49547231e-01 4.10229355e-01 -8.08301866e-01 1.06666636e+00
2.54979789e-01 2.02305332e-01 -6.75701022e-01 1.24568164e+00
-3.17729384e-01 7.14445233e-01 -2.67696798e-01 -1.21005073e-01
2.61801779e-01 -1.42953709e-01 6.88958526e-01 1.92548907e+00
-1.27303183e-01 -1.40924454e-02 8.06706920e-02 1.49820268e-01
-1.85898513e-01 1.05372369e+00 -2.81230927e-01 -2.66782522e-01
3.86854261e-01 1.34845769e+00 -9.26844835e-01 -4.99517530e-01
-2.42298707e-01 7.40216553e-01 4.66712564e-01 1.80578262e-01
-1.92647249e-01 -6.91721201e-01 2.88718611e-01 3.58755477e-02
4.16063398e-01 -4.49757546e-01 -6.01404488e-01 -1.08121967e+00
9.98440012e-02 -1.03031933e+00 6.96871758e-01 -4.16564673e-01
-9.92085218e-01 8.86998713e-01 1.46909785e-02 -6.50530279e-01
-2.96332598e-01 -4.83283222e-01 -3.07750076e-01 9.02444482e-01
-1.02512455e+00 -1.04818428e+00 4.51062083e-01 1.68996736e-01
3.11983645e-01 -2.60009915e-01 1.12827563e+00 3.86804163e-01
-3.36722940e-01 2.51872689e-01 -1.95728932e-02 6.65737748e-01
1.20938635e+00 -1.02319920e+00 6.94426894e-01 1.04135811e+00
6.05308473e-01 7.06226468e-01 6.70354068e-01 -7.17293084e-01
-8.75783741e-01 -6.51603043e-01 2.36958933e+00 -5.18028438e-01
1.13693476e+00 -3.90668273e-01 -7.04346657e-01 5.85817039e-01
6.88369572e-01 -4.23404843e-01 7.51057863e-01 3.40310305e-01
-3.43221277e-01 2.12532684e-01 -7.95989335e-01 2.87518561e-01
7.67035782e-01 -6.67077661e-01 -9.95879769e-01 2.60230154e-01
2.64708161e-01 -4.35321927e-01 -7.79126465e-01 2.66938061e-01
2.11012989e-01 -3.54390562e-01 3.26871186e-01 -9.19159174e-01
-1.28309846e-01 -3.68876100e-01 -7.26311747e-03 -8.59122694e-01
-4.94737029e-02 -5.46618044e-01 6.95463717e-01 1.84963322e+00
7.94463098e-01 -4.87256140e-01 5.64862549e-01 4.73719120e-01
9.12154764e-02 -4.68054041e-02 -1.15075684e+00 -5.77866554e-01
2.41225109e-01 -7.10254908e-01 2.94972539e-01 1.19160450e+00
2.69427210e-01 5.52676082e-01 9.00957733e-02 7.44924396e-02
3.81148964e-01 -6.55682310e-02 5.96533179e-01 -1.09384382e+00
2.47527421e-01 -2.78152674e-01 -5.96293628e-01 -1.79247767e-01
3.48767996e-01 -1.14574146e+00 -5.16812094e-02 -1.59510672e+00
-2.24156156e-01 -2.59223938e-01 -1.66177407e-01 6.68068409e-01
-9.23546031e-03 2.82650106e-02 5.78058243e-01 3.40887457e-01
-8.20527732e-01 -1.16144471e-01 2.07912847e-01 9.86318514e-02
-2.75510252e-01 7.71444663e-02 -3.94231111e-01 4.54640478e-01
5.93605220e-01 -4.94434386e-01 3.25278729e-01 -2.36779749e-01
4.35446382e-01 -1.27748579e-01 -2.91859776e-01 -7.84334064e-01
3.82559210e-01 -1.75915509e-01 6.96886778e-02 -6.01316035e-01
-7.34494403e-02 -5.16707420e-01 2.77676821e-01 3.03667068e-01
-2.85208523e-01 2.97590703e-01 3.02020431e-01 -1.77677244e-01
-5.69831610e-01 -4.61890370e-01 6.48677647e-01 -1.73928946e-01
-9.99375403e-01 -1.10626966e-01 -8.56104314e-01 4.65997279e-01
1.00144303e+00 1.40514180e-01 -9.37950313e-02 -2.28301361e-01
-7.78107882e-01 8.54485184e-02 4.00602281e-01 5.65966666e-01
-1.12803690e-01 -9.20452118e-01 -8.54582965e-01 -2.37801790e-01
2.54591912e-01 -6.04570091e-01 -6.84770104e-03 7.57847965e-01
-7.57167876e-01 9.99711096e-01 -2.30970502e-01 9.60836709e-02
-1.40050435e+00 2.88688093e-01 1.71386018e-01 -9.03517723e-01
-3.50227356e-01 3.16511869e-01 -2.54927784e-01 -9.29857373e-01
2.72337254e-02 3.33923161e-01 -5.04921079e-01 5.03745794e-01
3.98822039e-01 4.72124666e-01 6.53094649e-01 -1.19772577e+00
-8.69883537e-01 2.58329839e-01 -8.57143179e-02 -5.06516337e-01
1.29100525e+00 -2.16045514e-01 -4.19061959e-01 6.39353216e-01
9.04938400e-01 6.46294892e-01 -2.90653408e-01 -3.33553046e-01
9.40007567e-01 1.43276006e-01 -3.15146387e-01 -8.73848021e-01
-6.09603345e-01 5.10578871e-01 1.96902722e-01 -1.13492638e-01
4.77577180e-01 -5.63770756e-02 5.50447285e-01 4.49751645e-01
3.98487508e-01 -1.53716516e+00 -1.01154208e+00 1.37643063e+00
6.95568979e-01 -9.37527359e-01 8.98466036e-02 -6.26762033e-01
-6.28264666e-01 1.09636426e+00 2.01209933e-01 9.87709612e-02
5.45851409e-01 3.04216087e-01 6.19310737e-01 -7.82084838e-02
-4.97741967e-01 -4.24664378e-01 3.91337574e-01 5.02470970e-01
8.47640395e-01 8.19991529e-02 -1.28803980e+00 4.86768544e-01
-3.60332638e-01 -2.38725588e-01 4.57835793e-01 9.04103637e-01
-1.28916293e-01 -1.45685101e+00 -3.88507098e-01 -2.24801987e-01
-1.08966279e+00 -3.17156166e-01 -9.37731326e-01 8.62711489e-01
4.20486741e-02 9.36692059e-01 -2.75955081e-01 2.61660963e-01
6.89754307e-01 5.67679107e-01 1.41718134e-01 -7.30294347e-01
-1.15491760e+00 -1.85688823e-01 6.72864079e-01 -2.40755349e-01
-4.11333233e-01 -8.93802047e-01 -1.46374273e+00 -2.94668674e-01
-4.50544246e-02 7.40160406e-01 1.13097644e+00 1.05569255e+00
1.87997580e-01 -4.66747545e-02 6.88028038e-02 -3.88975829e-01
-3.18469524e-01 -9.71356273e-01 -5.66599011e-01 5.87415755e-01
-3.10578287e-01 -8.38108733e-02 -1.32161468e-01 1.40921354e-01] | [9.792044639587402, 9.708505630493164] |
77cf9595-b3e1-4b02-9d2d-bee4a6cf57ea | interpretable-automatic-fine-grained | 2305.14548 | null | https://arxiv.org/abs/2305.14548v1 | https://arxiv.org/pdf/2305.14548v1.pdf | Interpretable Automatic Fine-grained Inconsistency Detection in Text Summarization | Existing factual consistency evaluation approaches for text summarization provide binary predictions and limited insights into the weakness of summarization systems. Therefore, we propose the task of fine-grained inconsistency detection, the goal of which is to predict the fine-grained types of factual errors in a summary. Motivated by how humans inspect factual inconsistency in summaries, we propose an interpretable fine-grained inconsistency detection model, FineGrainFact, which explicitly represents the facts in the documents and summaries with semantic frames extracted by semantic role labeling, and highlights the related semantic frames to predict inconsistency. The highlighted semantic frames help verify predicted error types and correct inconsistent summaries. Experiment results demonstrate that our model outperforms strong baselines and provides evidence to support or refute the summary. | ['Heng Ji', 'Qi Zeng', 'Hou Pong Chan'] | 2023-05-23 | null | null | null | null | ['semantic-role-labeling', 'text-summarization'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.03850478e-01 7.46948898e-01 -8.05500925e-01 -5.69274187e-01
-1.10188162e+00 -5.51771581e-01 8.70084107e-01 9.09748316e-01
1.70095697e-01 1.15657544e+00 1.37078834e+00 -2.06243340e-02
-1.32101804e-01 -4.82077867e-01 -7.45205104e-01 7.15274662e-02
2.83525199e-01 4.84154880e-01 2.25988925e-01 -1.74393594e-01
1.10218251e+00 -2.80060589e-01 -1.26524770e+00 1.01988757e+00
1.35224926e+00 7.14374363e-01 -4.01812568e-02 6.79257035e-01
-2.47235477e-01 1.44410670e+00 -1.31655490e+00 -5.58018744e-01
-6.35848224e-01 -7.53972113e-01 -1.34654343e+00 3.21567267e-01
9.29371774e-01 -5.20041764e-01 1.24272376e-01 1.23709643e+00
1.13190658e-01 1.23240709e-01 6.88217640e-01 -1.41792786e+00
-6.44324243e-01 1.16455042e+00 -1.42080054e-01 6.85384095e-01
9.44671333e-01 -2.13155150e-01 1.46903825e+00 -6.81498766e-01
8.69358957e-01 1.60416067e+00 6.57805920e-01 3.90315294e-01
-6.03264511e-01 -1.41854212e-01 6.11978829e-01 6.25318289e-01
-7.36151934e-01 -9.75861669e-01 6.39895558e-01 -1.08259432e-01
1.43181038e+00 5.22954464e-01 4.42964822e-01 1.16082180e+00
2.93609351e-01 8.89638186e-01 4.62803215e-01 -2.35703945e-01
3.81731153e-01 -4.58255589e-01 6.41140819e-01 6.87847018e-01
1.05456853e+00 -8.36786091e-01 -9.41969752e-01 -4.14148480e-01
1.88474402e-01 -5.11895180e-01 -4.64860201e-01 3.74437332e-01
-1.24266887e+00 5.68245649e-01 1.71335056e-01 2.93532848e-01
-5.59670746e-01 2.96765625e-01 8.36069226e-01 -1.70598403e-02
6.29518688e-01 8.08676720e-01 -3.86541605e-01 -3.91855747e-01
-1.16056681e+00 7.66707480e-01 9.40195560e-01 1.16989315e+00
2.73628175e-01 -7.31144473e-02 -7.32197523e-01 7.19112873e-01
2.32016191e-01 2.35540971e-01 5.57200491e-01 -1.57252097e+00
1.01985633e+00 8.65130007e-01 4.17105645e-01 -1.34934735e+00
-3.84041131e-01 -1.33425087e-01 -6.46135986e-01 -6.45827174e-01
-1.02272443e-01 4.29449767e-01 -4.45100754e-01 1.33925462e+00
6.11085333e-02 2.01031044e-01 3.05291474e-01 7.38515079e-01
1.14477587e+00 4.95575041e-01 2.03431264e-01 -5.53359687e-01
1.47292101e+00 -1.28300476e+00 -1.31905711e+00 -3.58419478e-01
7.36254394e-01 -6.04257405e-01 8.95668864e-01 -1.72368169e-03
-1.53916264e+00 -3.00168186e-01 -1.28766739e+00 -4.37308371e-01
4.94930930e-02 4.16427180e-02 1.87038869e-01 -5.22672907e-02
-9.64887142e-01 7.10349441e-01 -7.26673365e-01 -2.99080640e-01
5.82504570e-01 -3.34861547e-01 -7.20272586e-02 -4.63345908e-02
-1.13160539e+00 9.88051951e-01 7.30199695e-01 -1.94422677e-01
-4.13127631e-01 -7.09475517e-01 -1.30777228e+00 4.65842068e-01
6.84253037e-01 -1.36722052e+00 1.76365709e+00 -6.21085465e-01
-8.80611300e-01 6.51347578e-01 -8.85182619e-01 -7.28595912e-01
3.85323316e-01 -4.06533659e-01 -3.72471720e-01 5.27286351e-01
8.09792519e-01 3.96548152e-01 7.38938630e-01 -1.24381673e+00
-8.89854550e-01 -2.64668465e-01 7.97016919e-02 4.07638490e-01
2.36902833e-01 8.43642354e-02 -2.41542220e-01 -1.04367125e+00
2.82694429e-01 -1.74876913e-01 1.68238997e-01 -5.25399506e-01
-1.13915169e+00 -6.80665433e-01 8.38246584e-01 -9.58576202e-01
1.67633295e+00 -1.60693610e+00 -7.98480734e-02 -3.41976970e-01
3.70556444e-01 -8.61080065e-02 -9.11054574e-03 3.44810218e-01
2.54159063e-01 6.52269721e-01 -2.81259805e-01 -3.41615647e-01
3.09013963e-01 2.57146060e-01 -8.63629758e-01 -8.94099772e-02
3.30074281e-01 1.02667940e+00 -1.36934364e+00 -9.23667252e-01
-1.10809493e-03 -3.41326475e-01 -4.46986049e-01 2.70839781e-01
-8.59727204e-01 5.11757806e-02 -7.35516429e-01 6.51223838e-01
4.46255744e-01 -5.63663006e-01 1.67357996e-01 -4.57286626e-01
1.62641555e-01 1.32103503e+00 -6.97513521e-01 1.67926717e+00
9.18424204e-02 5.18096447e-01 -3.85690659e-01 -9.59074736e-01
5.50621629e-01 3.48761231e-01 -9.74561274e-03 -6.96140230e-01
-3.67326736e-01 1.49863079e-01 -5.80403328e-01 -7.31690109e-01
1.15496278e+00 -1.49982087e-02 -5.65134704e-01 6.27629161e-01
-4.86653447e-02 -5.20887434e-01 3.68895084e-01 9.16114867e-01
1.11283767e+00 -2.93030031e-02 8.83964956e-01 -2.62065917e-01
5.40884078e-01 5.37157655e-01 7.24688351e-01 1.18865335e+00
-2.49373645e-01 5.08358300e-01 9.06417131e-01 -3.32946986e-01
-9.54574585e-01 -8.14885616e-01 3.22149128e-01 6.85530901e-01
5.71074188e-01 -1.09707332e+00 -6.47822022e-01 -1.07238019e+00
-1.49620548e-01 1.43788958e+00 -4.83121961e-01 -2.35157430e-01
-8.23974729e-01 -2.60527760e-01 5.88705599e-01 6.05652750e-01
8.32780600e-01 -9.85652387e-01 -6.14637911e-01 2.03636453e-01
-1.24496639e+00 -1.34270215e+00 -4.34425801e-01 -4.82625365e-01
-9.36994791e-01 -1.54316366e+00 2.43457049e-01 -4.06288743e-01
4.59437013e-01 4.34051216e-01 1.65251267e+00 5.35939097e-01
3.65292937e-01 4.13481086e-01 -3.74841005e-01 -3.85849655e-01
-7.67766178e-01 -2.98225611e-01 -1.22441337e-01 -7.66313136e-01
1.15513340e-01 -2.89186724e-02 -5.48760772e-01 7.01154172e-02
-7.98708081e-01 2.09925547e-01 -1.10443093e-01 8.03687990e-01
4.36773330e-01 1.44335330e-01 9.52376127e-01 -1.04820657e+00
1.04337311e+00 -4.82755303e-01 1.04678839e-01 7.19386041e-01
-6.03337884e-01 1.73066556e-01 6.58010364e-01 4.72579569e-01
-1.60030043e+00 -1.04552400e+00 -1.00462601e-01 1.82461247e-01
-3.88147496e-02 5.99171877e-01 -2.13764384e-01 9.92703855e-01
5.40943980e-01 6.70132414e-02 -6.48132324e-01 -1.58543512e-02
4.26161945e-01 4.45543647e-01 1.26665723e+00 -7.66605318e-01
1.24453560e-01 5.00421524e-01 -3.71473014e-01 -4.43112284e-01
-1.76398456e+00 -4.35062051e-01 -3.15215141e-01 -5.62020764e-02
7.10594058e-01 -1.02894485e+00 -2.33045906e-01 1.38153628e-01
-1.71663249e+00 7.69873895e-03 -4.50177342e-01 -1.41841412e-01
-8.58795285e-01 8.20448935e-01 -6.61777258e-01 -5.78982949e-01
-5.89157820e-01 -5.84604800e-01 1.53643465e+00 1.16577014e-01
-1.07648790e+00 -1.11102593e+00 -1.56072721e-01 7.42097914e-01
5.01059666e-02 2.47812018e-01 1.25381076e+00 -8.60780060e-01
-3.85267228e-01 6.56346306e-02 -1.75780535e-01 -7.33723268e-02
2.64896065e-01 2.28430122e-01 -5.96692443e-01 1.74336299e-01
1.95075929e-01 -4.43860650e-01 9.31614637e-01 6.15606666e-01
1.41771591e+00 -1.29604566e+00 -4.28306967e-01 -1.73151754e-02
7.85686076e-01 -2.70778239e-01 5.97879529e-01 4.53119189e-01
5.08957326e-01 4.88856256e-01 1.16535032e+00 7.81289399e-01
8.50951731e-01 5.52637875e-01 1.71807319e-01 4.94152933e-01
-3.33467364e-01 -7.90393114e-01 5.16060628e-02 6.36914372e-01
3.09968859e-01 -8.34097505e-01 -4.73508716e-01 6.76378608e-01
-2.41097164e+00 -1.55476093e+00 -1.84227705e-01 1.64989328e+00
1.02709234e+00 3.31502169e-01 -1.67803213e-01 1.49842411e-01
9.90613282e-01 6.25748992e-01 -3.81069154e-01 -5.26015759e-01
-4.21548456e-01 -3.97296309e-01 -3.10189366e-01 7.26653039e-01
-8.51441681e-01 1.03802550e+00 6.65297604e+00 5.49832523e-01
-4.53441203e-01 2.45941832e-04 6.55136824e-01 4.37978543e-02
-8.70374918e-01 1.34995028e-01 -7.31606543e-01 5.81951618e-01
7.76070714e-01 -5.68559289e-01 -3.21240783e-01 5.22912145e-01
7.54687488e-01 -4.90161330e-01 -1.26329184e+00 5.09198129e-01
4.31659281e-01 -2.03801584e+00 6.66603148e-01 -7.52745509e-01
8.44248533e-01 -5.11407197e-01 -4.63778496e-01 7.11175501e-02
5.13418376e-01 -7.79395223e-01 1.06265724e+00 4.54799771e-01
3.74295354e-01 -3.36667955e-01 8.09806824e-01 3.45448047e-01
-6.59057081e-01 -4.66528768e-03 -4.16629136e-01 -1.19674891e-01
5.54206431e-01 7.51830459e-01 -1.04147851e+00 6.88380480e-01
4.81062651e-01 1.32851911e+00 -5.96459448e-01 8.04234743e-01
-7.64701605e-01 6.81940258e-01 1.91609070e-01 1.00532718e-01
4.68203574e-02 3.06495488e-01 8.41361940e-01 1.44285464e+00
3.95003915e-01 3.24661434e-01 -2.09728740e-02 9.64420140e-01
-4.69496310e-01 -4.76206422e-01 -2.99042940e-01 -9.67909303e-03
1.10663319e+00 6.88134074e-01 -4.12945896e-01 -1.13687551e+00
-1.83758382e-02 1.16105628e+00 2.95607626e-01 2.22072914e-01
-7.60711372e-01 -3.27078998e-02 5.14402211e-01 -2.25388035e-01
1.17343388e-01 2.88580567e-01 -8.68075490e-01 -1.54659212e+00
2.94919372e-01 -8.48697484e-01 1.02253783e+00 -1.30389667e+00
-1.10702038e+00 1.61085948e-01 4.10392918e-02 -8.49649131e-01
-4.50297654e-01 1.50444835e-01 -1.10440767e+00 3.44093561e-01
-1.62066162e+00 -8.97989392e-01 -5.31627595e-01 1.98032796e-01
1.30253935e+00 3.49469721e-01 6.10520482e-01 -5.33142865e-01
-5.91019809e-01 3.14892799e-01 -4.59068507e-01 -2.32061774e-01
7.82536268e-01 -1.38669586e+00 6.54631436e-01 1.18441868e+00
-8.60531628e-02 6.87909663e-01 1.26792681e+00 -1.32592881e+00
-8.23621988e-01 -1.24170196e+00 1.75347924e+00 -6.21868312e-01
6.47414565e-01 3.82336140e-01 -1.14979923e+00 7.55343854e-01
3.74068588e-01 -3.60294759e-01 4.23855096e-01 -7.75205251e-03
-6.47589982e-01 3.96841228e-01 -1.21600556e+00 5.31476140e-01
1.33586442e+00 -5.17968476e-01 -1.68504739e+00 5.82350016e-01
9.81450737e-01 -5.16794086e-01 -4.07831907e-01 3.54516894e-01
7.97450021e-02 -9.67018485e-01 8.94553304e-01 -1.00018394e+00
1.22365713e+00 -2.98837930e-01 -7.07175434e-02 -1.55536139e+00
-2.83420533e-01 -2.30321258e-01 -6.72286868e-01 1.23577189e+00
2.74071813e-01 -1.44498944e-01 3.38863164e-01 7.96044886e-01
-7.55408525e-01 -1.18975237e-01 -8.92584443e-01 -5.56772649e-01
-3.05630147e-01 -2.06158474e-01 6.00368798e-01 1.10237968e+00
8.70692194e-01 4.94234234e-01 5.79480361e-03 1.96806088e-01
4.29216206e-01 4.42093343e-01 4.37764466e-01 -1.04023826e+00
3.59081142e-02 -5.69580376e-01 -6.42032251e-02 -9.93647814e-01
6.67035341e-01 -5.57039738e-01 2.57648349e-01 -2.19418859e+00
5.42074800e-01 1.44308349e-02 8.60174671e-02 4.95633721e-01
-8.16958070e-01 -2.67608881e-01 -1.02593891e-01 4.01556194e-01
-1.55085385e+00 6.14174545e-01 7.83424616e-01 -4.36036497e-01
3.68797481e-02 -2.68526196e-01 -1.28774738e+00 8.58758867e-01
5.90060532e-01 -2.99525797e-01 -4.06095833e-01 -5.39364457e-01
2.80966043e-01 3.74590963e-01 5.10081828e-01 -6.60154641e-01
3.69314462e-01 -2.97827899e-01 2.31357500e-01 -8.25077653e-01
-8.64266008e-02 -2.04582050e-01 -2.65605360e-01 3.22447419e-01
-9.15456474e-01 3.19326818e-01 3.67395692e-02 7.71070480e-01
-5.82186818e-01 -3.34522575e-01 2.86823332e-01 -2.65574485e-01
-7.79722333e-01 -3.11486572e-01 -3.54589343e-01 6.32975221e-01
5.36967635e-01 -1.32030651e-01 -1.52078390e+00 -5.69931567e-01
-3.75384122e-01 4.32960302e-01 5.09114504e-01 5.23938894e-01
8.28402519e-01 -1.09665549e+00 -9.15656626e-01 -4.84697998e-01
4.16420221e-01 2.11726874e-01 3.06270540e-01 5.02866864e-01
-1.77823618e-01 7.37827361e-01 -1.08850114e-02 -2.92929322e-01
-1.25310457e+00 3.52240890e-01 4.08404879e-02 -4.28281128e-01
-7.53948689e-01 6.01780534e-01 -7.68307447e-02 1.66608140e-01
6.85109496e-02 -6.78744316e-01 -3.23091537e-01 -7.48863444e-02
9.02157903e-01 6.21000051e-01 2.92081267e-01 -3.99661183e-01
-4.66137797e-01 -1.41873639e-02 -2.82671958e-01 7.03402385e-02
1.15203941e+00 -6.90483570e-01 -5.23364246e-01 2.17745870e-01
5.87470889e-01 1.33690625e-01 -1.10210252e+00 -5.29871173e-02
6.45765722e-01 -5.18645287e-01 -1.68002918e-01 -1.19067729e+00
-1.84850872e-01 3.05240512e-01 -7.48149753e-01 2.41880894e-01
8.76447320e-01 4.27700251e-01 7.56847262e-01 5.72076976e-01
9.88078117e-02 -1.24220514e+00 3.96957636e-01 6.14838183e-01
1.41652322e+00 -1.28879631e+00 3.50450575e-01 -8.23721409e-01
-7.94242203e-01 1.05550051e+00 9.28562164e-01 1.18109159e-01
-3.49196464e-01 8.29077139e-02 -2.24106401e-01 -5.33252358e-01
-1.22797525e+00 3.64050716e-01 3.18031758e-01 3.16343099e-01
4.47141200e-01 -2.16698330e-02 -6.05160832e-01 9.44851279e-01
-2.60066509e-01 -3.07424087e-02 1.04929829e+00 8.29445899e-01
-8.49968076e-01 -4.67194378e-01 -3.43109339e-01 6.59808338e-01
-3.58722627e-01 -1.28407195e-01 -7.98858047e-01 3.10790241e-01
-3.76237094e-01 1.39840102e+00 2.70966560e-01 1.41939506e-01
3.87355477e-01 9.69594494e-02 3.98509622e-01 -6.46778703e-01
-5.82621217e-01 -3.51206809e-01 8.76630425e-01 -7.55031049e-01
-5.24354041e-01 -6.67384565e-01 -1.68333244e+00 -3.12357545e-01
1.15205292e-02 4.37636644e-01 1.97676361e-01 1.33627629e+00
6.58374906e-01 6.17419422e-01 1.59692749e-01 -3.15406561e-01
-6.31399512e-01 -9.67837334e-01 3.34497914e-02 8.92566144e-01
5.97040534e-01 -3.76816154e-01 -4.19817269e-01 1.95528060e-01] | [12.21902084350586, 9.357685089111328] |
cb1cc23b-3474-4c1c-9ff5-81a05e4505ac | multi-hypothesis-pose-networks-rethinking-top | 2101.11223 | null | https://arxiv.org/abs/2101.11223v3 | https://arxiv.org/pdf/2101.11223v3.pdf | Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation | A key assumption of top-down human pose estimation approaches is their expectation of having a single person/instance present in the input bounding box. This often leads to failures in crowded scenes with occlusions. We propose a novel solution to overcome the limitations of this fundamental assumption. Our Multi-Instance Pose Network (MIPNet) allows for predicting multiple 2D pose instances within a given bounding box. We introduce a Multi-Instance Modulation Block (MIMB) that can adaptively modulate channel-wise feature responses for each instance and is parameter efficient. We demonstrate the efficacy of our approach by evaluating on COCO, CrowdPose, and OCHuman datasets. Specifically, we achieve 70.0 AP on CrowdPose and 42.5 AP on OCHuman test sets, a significant improvement of 2.4 AP and 6.5 AP over the prior art, respectively. When using ground truth bounding boxes for inference, MIPNet achieves an improvement of 0.7 AP on COCO, 0.9 AP on CrowdPose, and 9.1 AP on OCHuman validation sets compared to HRNet. Interestingly, when fewer, high confidence bounding boxes are used, HRNet's performance degrades (by 5 AP) on OCHuman, whereas MIPNet maintains a relatively stable performance (drop of 1 AP) for the same inputs. | ['Ambrish Tyagi', 'Amit Agrawal', 'Visesh Chari', 'Rawal Khirodkar'] | 2021-01-27 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Khirodkar_Multi-Instance_Pose_Networks_Rethinking_Top-Down_Pose_Estimation_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Khirodkar_Multi-Instance_Pose_Networks_Rethinking_Top-Down_Pose_Estimation_ICCV_2021_paper.pdf | iccv-2021-1 | ['2d-human-pose-estimation'] | ['computer-vision'] | [-1.05360352e-01 2.24793509e-01 2.07598671e-01 -2.82908976e-01
-9.59002078e-01 -3.82532269e-01 3.69955271e-01 -9.50407311e-02
-6.22101188e-01 1.03695822e+00 1.50719713e-02 2.59095103e-01
8.60167146e-02 -6.79810703e-01 -1.10436702e+00 -3.73133421e-01
-1.96900874e-01 7.83650339e-01 5.61637819e-01 -2.31324494e-01
-2.44704992e-01 4.19794172e-01 -1.66180420e+00 3.32638711e-01
5.28070152e-01 1.14510012e+00 -2.62900740e-01 8.43362629e-01
2.94883937e-01 4.16562915e-01 -1.08372784e+00 -6.93392873e-01
4.62442249e-01 1.54155448e-01 -5.11364400e-01 -1.50600538e-01
1.23722839e+00 -5.13787627e-01 -3.73452991e-01 5.91623187e-01
8.32373202e-01 2.62712508e-01 5.66254795e-01 -1.32929945e+00
4.79212627e-02 3.38143170e-01 -6.91429734e-01 1.77887112e-01
5.93487680e-01 3.35066348e-01 1.01778030e+00 -1.02964628e+00
5.36579072e-01 1.45384610e+00 9.66882348e-01 5.25370836e-01
-1.23692882e+00 -6.89873993e-01 3.38266343e-01 -1.38069183e-01
-1.66778898e+00 -4.72013086e-01 1.57794595e-01 -3.90581101e-01
1.06156492e+00 2.66792387e-01 7.88589418e-01 1.34957576e+00
1.24548383e-01 9.47866321e-01 8.03252161e-01 -2.53826063e-02
1.93224356e-01 -6.58354312e-02 6.14038855e-02 5.53307235e-01
5.49116254e-01 -8.66098776e-02 -8.88568819e-01 -1.44476891e-01
6.70614362e-01 -3.54181647e-01 -1.82061136e-01 -1.43200517e-01
-8.80976737e-01 5.43810427e-01 6.83809221e-01 -2.30205327e-01
-2.07492873e-01 4.10795450e-01 3.25526237e-01 -1.38865173e-01
3.78561318e-01 6.12319708e-01 -5.29696286e-01 -1.99885413e-01
-9.58388031e-01 9.33876455e-01 8.46120954e-01 1.27054727e+00
3.54812205e-01 -2.91417360e-01 -5.06076872e-01 6.82385147e-01
2.06865910e-02 9.81303573e-01 -1.19069211e-01 -9.63181913e-01
7.86577106e-01 3.36307108e-01 5.14349043e-01 -1.17296576e+00
-7.16898739e-01 -5.65638781e-01 -4.65522945e-01 -7.32189417e-02
8.70101988e-01 -2.32099533e-01 -9.93027568e-01 1.82495761e+00
3.92455071e-01 3.22477962e-03 -3.80166143e-01 9.90204692e-01
9.08392489e-01 4.76850063e-01 2.74296880e-01 2.12101251e-01
1.36795294e+00 -8.86061072e-01 -4.44943100e-01 -5.74636579e-01
2.16755107e-01 -3.66210580e-01 1.06977856e+00 5.66164434e-01
-1.13002014e+00 -7.36147404e-01 -1.00462043e+00 2.00092755e-02
-2.08206788e-01 9.02199522e-02 5.48765898e-01 6.56690061e-01
-7.79242158e-01 4.69303519e-01 -7.07170904e-01 -2.95378149e-01
5.28303921e-01 5.63869119e-01 -3.73412460e-01 -1.29414663e-01
-1.13626218e+00 8.64335418e-01 1.83652118e-01 5.99767789e-02
-8.47956359e-01 -7.29591668e-01 -7.73503840e-01 -8.22635740e-02
5.76110661e-01 -7.04087734e-01 1.06745744e+00 -2.75453031e-01
-1.20446754e+00 7.31047392e-01 -1.30247131e-01 -7.59143472e-01
1.21287084e+00 -1.05130041e+00 -2.44094938e-01 2.06294984e-01
3.11348200e-01 1.15113270e+00 5.93629420e-01 -1.15087748e+00
-7.07574606e-01 -3.99758518e-01 1.88991129e-01 2.37935424e-01
1.82142518e-02 -2.01014429e-01 -9.50731516e-01 -4.49805796e-01
4.76660319e-02 -1.07859063e+00 -1.92657441e-01 1.89475566e-01
-6.29541457e-01 -6.14787117e-02 3.77555430e-01 -6.24021053e-01
1.11032808e+00 -1.92409587e+00 4.80247550e-02 3.15800786e-01
3.40031922e-01 4.40946855e-02 -1.64852869e-02 1.75993051e-02
5.04041076e-01 -2.27879044e-02 -5.43590486e-02 -6.52708709e-01
5.01098335e-02 2.08305508e-01 -3.57934311e-02 5.36594629e-01
2.33987704e-01 8.70996892e-01 -6.15298390e-01 -5.93438506e-01
1.63607270e-01 5.61474741e-01 -9.95288551e-01 1.21434309e-01
-1.58635989e-01 4.61467981e-01 -1.33115485e-01 8.49042296e-01
6.58740520e-01 -3.47704262e-01 5.32911951e-03 -1.91531926e-01
1.30525529e-01 -3.38125369e-03 -1.39814794e+00 1.39558983e+00
-1.56184301e-01 6.66490316e-01 -2.36995339e-01 -2.53593832e-01
8.71344209e-01 1.12312742e-01 3.77284169e-01 -4.89468277e-01
1.55866802e-01 1.47408575e-01 -5.13520800e-02 -1.15406401e-01
6.20452881e-01 2.20171586e-01 -3.87434989e-01 -1.88016579e-01
-4.78207543e-02 -2.12495904e-02 2.90551186e-01 7.98489675e-02
1.14650118e+00 1.63314700e-01 1.95657521e-01 -1.95447490e-01
3.76407266e-01 -2.38857746e-01 4.94113564e-01 1.15632725e+00
-3.90618593e-01 9.05706406e-01 5.42239547e-01 -7.30751991e-01
-1.06184518e+00 -1.27171421e+00 -3.33633780e-01 1.19400179e+00
2.59800762e-01 -4.85848129e-01 -9.09267366e-01 -5.55279016e-01
3.52707148e-01 3.62039834e-01 -8.51381063e-01 2.12491363e-01
-8.70886207e-01 -6.51899874e-01 8.57201695e-01 8.55722129e-01
7.36979783e-01 -7.10184693e-01 -8.07652175e-01 1.71796884e-02
-3.72640431e-01 -1.45192015e+00 -2.44651482e-01 -1.99937984e-01
-3.94306004e-01 -9.98193383e-01 -7.81922758e-01 -9.45794804e-04
4.95240003e-01 -1.69246808e-01 1.20360637e+00 3.73801473e-03
-2.19016090e-01 2.50611722e-01 -1.49476036e-01 -4.94305372e-01
3.20737571e-01 2.78608859e-01 3.73158693e-01 -2.24182308e-01
2.61532009e-01 -2.78251857e-01 -5.59447646e-01 5.86805582e-01
-2.00804621e-01 -2.20180556e-01 4.23113674e-01 7.75468588e-01
6.46133006e-01 -4.79770333e-01 3.53393257e-01 -7.94347703e-01
1.56439379e-01 -2.71866560e-01 -5.94482899e-01 -2.77716294e-02
-2.72694945e-01 -1.57467306e-01 3.08311552e-01 -4.68390465e-01
-6.77387595e-01 1.75876096e-01 4.33633476e-02 -5.50527096e-01
-1.58010647e-01 -1.39815822e-01 -2.90814072e-01 -9.76823494e-02
9.96490359e-01 -2.66412139e-01 -2.64195830e-01 -3.73436183e-01
1.49582192e-01 3.25141311e-01 8.00545514e-01 -9.72258270e-01
7.40337431e-01 5.44573963e-01 -5.14032319e-03 -7.29719996e-01
-1.16823161e+00 -3.76673251e-01 -6.92625523e-01 -4.65829015e-01
1.03611243e+00 -1.11448145e+00 -1.10027456e+00 1.88599616e-01
-1.15239096e+00 -3.47801507e-01 -1.52385682e-01 3.21034551e-01
-4.53783721e-01 -2.98927929e-02 -4.50331837e-01 -9.00405526e-01
-2.12246865e-01 -1.08471966e+00 1.49549258e+00 4.24567536e-02
-6.66539371e-01 -4.93270367e-01 -1.90185949e-01 5.80839396e-01
1.71794623e-01 5.89423776e-01 1.99261606e-02 -6.17884815e-01
-5.18786252e-01 -3.97244602e-01 -2.59681851e-01 1.61240064e-02
-3.58810872e-01 -2.50386953e-01 -1.29019427e+00 -4.61070001e-01
-7.12824047e-01 -4.62116718e-01 6.48693502e-01 4.48716521e-01
1.09084427e+00 -1.81654140e-01 -5.33190906e-01 5.79036295e-01
9.32377398e-01 -1.45625278e-01 6.36119366e-01 3.61355245e-01
7.05755711e-01 5.67592621e-01 7.60769844e-01 6.50657535e-01
3.46231669e-01 9.88301277e-01 3.17355365e-01 3.16747576e-02
8.25886056e-03 -4.34711158e-01 2.38636896e-01 -3.68194282e-02
-4.83061999e-01 -3.93545866e-01 -1.12289691e+00 3.76529336e-01
-1.73506117e+00 -8.02347839e-01 -8.68409351e-02 2.22778487e+00
5.60456157e-01 6.26241446e-01 5.01132667e-01 1.64942145e-02
6.59812152e-01 -1.35696791e-02 -6.29839718e-01 8.45049471e-02
-9.20682773e-02 1.88322172e-01 6.59638047e-01 4.53005880e-01
-1.44393313e+00 1.05259883e+00 6.67763567e+00 4.78638142e-01
-7.48018920e-01 -1.06290914e-01 6.06528759e-01 -5.32919765e-01
4.27087396e-01 -5.60306013e-01 -1.30899215e+00 5.90159595e-01
7.64779150e-01 2.83606261e-01 1.52063310e-01 1.06440604e+00
-4.05846275e-02 -2.65509069e-01 -1.29540229e+00 1.22016764e+00
8.16915855e-02 -9.40487266e-01 -5.53807802e-02 2.35485345e-01
6.32133186e-01 -2.12081268e-01 8.59717429e-02 5.79418838e-01
1.29437864e-01 -1.23953235e+00 9.76566851e-01 4.05663729e-01
8.00581932e-01 -9.38760638e-01 9.88357604e-01 2.70331115e-01
-1.22316992e+00 1.23133563e-01 -3.88384491e-01 -1.57582462e-01
2.87891746e-01 3.55844200e-01 -8.67843330e-01 3.42370361e-01
1.05936182e+00 2.30390012e-01 -6.49106979e-01 9.87306297e-01
-3.85814548e-01 3.59633595e-01 -7.16549993e-01 -1.22085877e-01
7.47487471e-02 5.91441274e-01 5.88576317e-01 1.29407167e+00
-9.73941982e-02 1.79651737e-01 5.06602049e-01 6.17505848e-01
-1.65969536e-01 -9.19826552e-02 -2.83228874e-01 6.28506064e-01
7.71161616e-01 8.08027685e-01 -4.25430804e-01 -4.74508047e-01
-3.27390954e-02 9.39161062e-01 4.63127941e-01 1.80515617e-01
-1.34695327e+00 -2.21307054e-01 7.77180672e-01 3.81908506e-01
3.88125509e-01 -1.36502340e-01 -3.17820311e-01 -8.27734232e-01
3.47595334e-01 -8.63702536e-01 3.96322906e-01 -5.47077537e-01
-1.20121002e+00 5.93208075e-01 3.74855012e-01 -1.17507970e+00
-1.36791795e-01 -8.03653538e-01 -1.40975684e-01 6.43911123e-01
-7.51405001e-01 -1.15315378e+00 -6.40146255e-01 3.96971494e-01
3.26708525e-01 1.39411151e-01 5.54037750e-01 4.76850688e-01
-5.19475400e-01 1.15709138e+00 -5.96007824e-01 3.31724972e-01
7.15846181e-01 -1.20933390e+00 6.79177999e-01 6.74622178e-01
-9.39513892e-02 7.15607047e-01 9.36687171e-01 -7.47930408e-01
-1.09599745e+00 -1.07954383e+00 7.08394051e-01 -1.09341216e+00
1.96043834e-01 -5.68155050e-01 -7.13668883e-01 7.32231021e-01
-4.31494832e-01 2.80049622e-01 4.69712198e-01 4.65744525e-01
-4.54211950e-01 3.45961079e-02 -1.38015294e+00 7.15181172e-01
1.43255806e+00 -1.56131729e-01 -5.56723654e-01 3.43331784e-01
7.33903170e-01 -1.00108945e+00 -1.04165614e+00 5.96185923e-01
9.06589627e-01 -9.39590514e-01 1.15354121e+00 -6.05903208e-01
3.77426267e-01 -3.82046103e-01 -4.88378972e-01 -8.31046939e-01
-2.08869606e-01 -4.14383560e-01 -2.93713957e-01 7.29323745e-01
4.78705108e-01 -4.37753707e-01 9.90029871e-01 7.90144384e-01
6.56213611e-02 -9.01809275e-01 -1.06019866e+00 -9.49383676e-01
-5.05866073e-02 -6.22750103e-01 5.05030811e-01 4.51354504e-01
-3.87194067e-01 1.95161432e-01 -5.80437720e-01 4.11474496e-01
6.61534607e-01 -4.80709463e-01 1.42453241e+00 -1.11064756e+00
-5.19738972e-01 -1.97149798e-01 -6.44151449e-01 -1.28079939e+00
-1.64196432e-01 -2.96451271e-01 2.58016676e-01 -1.28764665e+00
1.60015672e-01 -4.99678135e-01 -3.61260362e-02 4.40123886e-01
-2.51563251e-01 5.08215308e-01 4.48034048e-01 1.13512143e-01
-8.42292726e-01 3.09879035e-01 1.13687658e+00 -1.10832062e-02
-2.16231421e-01 -3.59507129e-02 -3.46742988e-01 8.16418469e-01
6.74941480e-01 -3.11884940e-01 -2.83745565e-02 -4.20477360e-01
2.66392883e-02 -1.96243927e-01 5.93768060e-01 -1.61660898e+00
1.98310360e-01 1.78316161e-01 9.63142395e-01 -7.13689148e-01
8.55418563e-01 -5.00016689e-01 1.23358183e-01 4.99817461e-01
-1.30860701e-01 -4.83138077e-02 6.00292921e-01 5.65849125e-01
1.59758478e-01 3.84083301e-01 7.24400938e-01 -1.07340150e-01
-5.31084657e-01 2.84510165e-01 1.40691936e-01 4.11590159e-01
9.10364807e-01 -4.52781439e-01 -2.77901113e-01 -2.31238171e-01
-8.78835082e-01 4.34428662e-01 2.44695589e-01 3.91173393e-01
5.44859946e-01 -1.13717806e+00 -7.35691547e-01 8.92106965e-02
1.15738191e-01 2.78185725e-01 1.82869524e-01 7.20815063e-01
-5.54060936e-01 3.84238124e-01 -1.34417430e-01 -9.19407904e-01
-1.21376705e+00 -1.52666774e-02 5.19596994e-01 -2.34463662e-01
-4.68697071e-01 1.22308993e+00 2.01135546e-01 -3.89066666e-01
4.82427478e-01 -2.60024846e-01 5.58942221e-02 -3.90936732e-02
5.89785874e-01 7.32202291e-01 -9.52883735e-02 -6.47774279e-01
-6.48501515e-01 6.10199451e-01 -1.10958152e-01 -2.94495583e-01
1.12615979e+00 3.75631481e-01 3.90369326e-01 2.52929181e-01
8.08178306e-01 2.39950746e-01 -1.57889259e+00 1.17048681e-01
-1.94913596e-01 -5.98501384e-01 -4.96577770e-01 -8.03101003e-01
-6.81084991e-01 6.17942572e-01 6.17216408e-01 -2.90462911e-01
5.73282659e-01 1.63482904e-01 6.63346767e-01 6.54249847e-01
7.93491781e-01 -1.09237599e+00 1.74725324e-01 5.50823629e-01
1.06256425e+00 -1.14214051e+00 2.71933258e-01 -5.39118469e-01
-5.44102132e-01 6.94205761e-01 1.24960804e+00 -1.54837266e-01
2.40309775e-01 3.17511320e-01 -2.89515197e-01 -3.82437825e-01
-5.23115873e-01 -1.26644313e-01 4.93446678e-01 4.11021233e-01
2.93369353e-01 2.45647267e-01 4.89599071e-02 6.18919373e-01
-6.08508646e-01 -3.01689446e-01 4.47122529e-02 8.45387816e-01
-6.52051032e-01 -4.68881518e-01 -6.88569367e-01 4.16416764e-01
-4.35088426e-01 5.40039949e-02 -3.49126518e-01 1.26141095e+00
3.08296651e-01 7.63651669e-01 2.77067304e-01 -5.33049226e-01
8.34956825e-01 -1.08158797e-01 5.46199918e-01 -5.79184294e-01
-5.52459836e-01 4.47058603e-02 3.67355138e-01 -8.75669122e-01
-1.27348378e-01 -6.85987473e-01 -1.22386134e+00 -4.47205514e-01
-1.88425049e-01 -1.33403674e-01 2.25622416e-01 7.23559558e-01
2.37858430e-01 4.91234213e-01 1.60749361e-03 -1.02007163e+00
-4.08233166e-01 -1.05402470e+00 -1.90547839e-01 4.58121628e-01
1.21714786e-01 -1.17190683e+00 -2.97673076e-01 -1.90705478e-01] | [7.1827497482299805, -0.7394098043441772] |
1b07cd74-2944-4989-934a-ffa2d83f2cd7 | aet-efn-a-versatile-design-for-static-and | 2103.11645 | null | https://arxiv.org/abs/2103.11645v1 | https://arxiv.org/pdf/2103.11645v1.pdf | AET-EFN: A Versatile Design for Static and Dynamic Event-Based Vision | The neuromorphic event cameras, which capture the optical changes of a scene, have drawn increasing attention due to their high speed and low power consumption. However, the event data are noisy, sparse, and nonuniform in the spatial-temporal domain with an extremely high temporal resolution, making it challenging to design backend algorithms for event-based vision. Existing methods encode events into point-cloud-based or voxel-based representations, but suffer from noise and/or information loss. Additionally, there is little research that systematically studies how to handle static and dynamic scenes with one universal design for event-based vision. This work proposes the Aligned Event Tensor (AET) as a novel event data representation, and a neat framework called Event Frame Net (EFN), which enables our model for event-based vision under static and dynamic scenes. The proposed AET and EFN are evaluated on various datasets, and proved to surpass existing state-of-the-art methods by large margins. Our method is also efficient and achieves the fastest inference speed among others. | ['Ngai Wong', 'Edmund Lam', 'Xiaojuan Qi', 'Chang Liu'] | 2021-03-22 | null | null | null | null | ['event-based-vision'] | ['computer-vision'] | [ 3.91910821e-01 -9.47756767e-01 2.35784039e-01 -3.21223736e-01
-7.83477202e-02 -2.66341418e-01 6.75489247e-01 9.68023017e-02
-5.95631659e-01 7.54692852e-01 -5.23308069e-02 3.33929509e-01
-3.30393195e-01 -8.49870920e-01 -8.41077209e-01 -8.30809414e-01
7.67496824e-02 4.29684594e-02 8.86735797e-01 3.49688649e-01
3.97175461e-01 4.67464179e-01 -1.97022784e+00 2.11566582e-01
6.72288895e-01 1.31075406e+00 5.08151710e-01 5.10798752e-01
-1.75586790e-01 9.96974230e-01 -3.69194418e-01 -3.90997082e-01
-1.17845729e-01 -4.52802420e-01 -3.00158173e-01 -2.67190039e-01
3.70504379e-01 -3.32834572e-01 -1.02037859e+00 1.17261577e+00
5.19452870e-01 9.90212411e-02 2.30359420e-01 -1.43904102e+00
-8.80927265e-01 2.20165908e-01 -4.77279782e-01 6.28843069e-01
1.91481218e-01 1.80052981e-01 5.56758940e-01 -7.58276761e-01
7.93750525e-01 1.09377038e+00 5.61116159e-01 4.63648409e-01
-1.08399332e+00 -6.74228728e-01 3.08518112e-01 8.82142186e-01
-1.12285662e+00 -2.68101990e-01 7.67837763e-01 -1.62407994e-01
1.16640234e+00 9.24346820e-02 1.25776339e+00 1.36956239e+00
7.20025599e-01 7.93122411e-01 1.26128948e+00 2.38743961e-01
6.20718956e-01 -7.02277660e-01 1.06372923e-01 4.63680774e-01
2.16650769e-01 1.66367576e-01 -1.07298613e+00 -9.17403325e-02
1.11788821e+00 6.95780993e-01 -4.04363781e-01 3.31087746e-02
-1.56920159e+00 3.14375609e-01 5.34990847e-01 5.62913828e-02
-6.77214563e-01 5.30381620e-01 3.87653261e-01 -1.52046839e-02
-3.53832729e-02 -1.48546457e-01 -3.85682248e-02 -4.10263628e-01
-8.28409255e-01 3.40113550e-01 5.40810287e-01 7.44755566e-01
4.76716340e-01 5.78270480e-02 -3.39195937e-01 5.65099061e-01
2.91779280e-01 5.12996137e-01 6.32537544e-01 -8.19321632e-01
-5.28846830e-02 6.30808353e-01 -1.12974875e-01 -1.12101769e+00
-2.95317233e-01 -6.90503567e-02 -1.01328206e+00 5.55256829e-02
1.37554020e-01 2.64928311e-01 -9.98959780e-01 1.79556751e+00
2.99144030e-01 9.81743336e-01 -1.61135197e-01 1.03527665e+00
9.11085963e-01 9.51558709e-01 1.38764963e-01 -6.62128568e-01
1.28388286e+00 -5.03202617e-01 -9.09255862e-01 -1.83916077e-01
-5.63345730e-01 -6.66797936e-01 7.39222705e-01 5.94977021e-01
-1.13526189e+00 -4.74877447e-01 -1.05633152e+00 -2.92713761e-01
-1.80005968e-01 -1.77946955e-01 1.14154470e+00 1.22854203e-01
-8.82097185e-01 3.93990874e-01 -1.25700963e+00 -4.39929754e-01
5.83851635e-01 2.75891572e-01 -2.28939220e-01 -1.77539408e-01
-7.97341645e-01 5.12431324e-01 2.63075918e-01 1.97201654e-01
-1.20782840e+00 -5.20673633e-01 -5.39507806e-01 -1.24397278e-01
5.03687337e-02 -6.50495350e-01 1.18224967e+00 -6.24280751e-01
-1.39384472e+00 4.93710607e-01 -3.58258516e-01 -5.21899343e-01
1.32694542e-01 1.87411785e-01 -5.10416746e-01 3.79853427e-01
-2.01840088e-01 6.05158269e-01 7.53700316e-01 -6.18195295e-01
-5.64481795e-01 -6.75113857e-01 9.84068662e-02 6.16469532e-02
-4.14664268e-01 9.75416303e-02 -5.50820887e-01 -6.39953315e-01
3.96371126e-01 -6.13801181e-01 -5.54268882e-02 5.99692822e-01
-2.80927029e-02 -2.97063738e-01 1.01184821e+00 -2.12763116e-01
9.57399249e-01 -2.12051630e+00 1.10623822e-01 -5.16877532e-01
3.07826221e-01 4.91985939e-02 8.25237334e-02 1.96236178e-01
1.78457022e-01 -3.80959243e-01 -2.41372660e-01 -1.29044890e-01
-2.70254463e-01 5.54623544e-01 -5.79301596e-01 4.11484480e-01
2.99302936e-01 7.54501641e-01 -1.05751586e+00 -7.10015535e-01
4.77321565e-01 6.69332385e-01 -3.10966522e-01 4.13158238e-02
-3.18036586e-01 5.21648109e-01 -6.58314586e-01 7.43866324e-01
6.65079534e-01 -3.28119606e-01 -1.50842190e-01 -5.53906858e-01
-5.39264679e-01 2.32719421e-01 -1.23843002e+00 2.01948333e+00
6.95846451e-04 7.53030002e-01 -3.90750229e-01 -8.17073524e-01
8.32912326e-01 1.91217482e-01 6.61625028e-01 -1.10385156e+00
2.69018173e-01 9.28793252e-02 -3.30364518e-02 -5.10994673e-01
4.47789013e-01 7.42962882e-02 1.99813917e-01 6.54443279e-02
3.41888428e-01 2.78425485e-01 1.97920755e-01 3.36494371e-02
1.46494651e+00 1.82984471e-01 7.12136850e-02 1.32826403e-01
1.11152373e-01 -2.40834877e-01 1.07525551e+00 5.91345668e-01
-4.44081813e-01 7.29742050e-01 1.31158724e-01 -5.43905437e-01
-7.11406827e-01 -1.44686186e+00 -4.63129461e-01 4.96223986e-01
7.10262120e-01 -2.55657792e-01 -4.35012132e-01 -3.00695971e-02
-1.77955598e-01 2.37593651e-01 -2.28436679e-01 -2.03015715e-01
-3.39352816e-01 -1.15895200e+00 4.70929891e-01 5.88814735e-01
9.91041005e-01 -1.12109077e+00 -9.56226230e-01 5.97040832e-01
-1.91274762e-01 -1.33800292e+00 -2.11494833e-01 1.11159869e-01
-9.93598759e-01 -9.32857752e-01 -4.22761083e-01 -7.99049079e-01
4.45031255e-01 1.95357278e-01 7.38869548e-01 -5.72302580e-01
-6.25467598e-01 3.44683051e-01 -1.84187621e-01 -5.76489747e-01
5.76566756e-01 -7.27428138e-01 1.03523076e-01 3.92300487e-01
4.19149071e-01 -1.14381075e+00 -1.07568598e+00 2.79981136e-01
-1.13378596e+00 1.61894456e-01 4.50954914e-01 6.55378938e-01
1.35947526e+00 8.49145502e-02 4.51079279e-01 -3.53368163e-01
2.58662790e-01 -4.54398721e-01 -6.20604455e-01 1.26972631e-01
-4.15577769e-01 -1.69971526e-01 7.88798034e-01 -7.75024831e-01
-1.13209522e+00 9.12438184e-02 9.75402351e-03 -6.59966171e-01
7.51086771e-02 3.88564557e-01 -7.75887817e-02 -3.76903772e-01
3.94808799e-01 5.68985224e-01 -4.94523466e-01 -2.37579197e-01
-3.42231616e-02 2.93323725e-01 9.62888539e-01 -5.39350927e-01
1.97711080e-01 9.68613148e-01 1.20770290e-01 -6.43013418e-01
-4.13580894e-01 -1.92226961e-01 -2.26791844e-01 -4.15985614e-01
8.21743608e-01 -8.87312651e-01 -9.46006298e-01 1.13189650e+00
-1.46067762e+00 6.13904931e-03 -3.07923257e-01 7.63218224e-01
-5.34923613e-01 1.91140294e-01 -7.94311106e-01 -7.89777398e-01
-3.19035500e-01 -1.07225811e+00 9.30667520e-01 8.10216129e-01
3.91203970e-01 -5.02627969e-01 1.95553362e-01 -1.63664356e-01
2.70611972e-01 4.42320406e-01 5.99903345e-01 -2.73217522e-02
-1.14854169e+00 -3.27777565e-02 -4.85317975e-01 -7.80713046e-03
-2.28951931e-01 2.24803120e-01 -9.84075010e-01 2.49052271e-02
4.24761862e-01 -2.14258656e-01 8.39338005e-01 4.90624607e-01
1.66770387e+00 -4.25118096e-02 -3.41749370e-01 9.51172113e-01
1.70789242e+00 4.78955418e-01 1.00222242e+00 -8.58462676e-02
5.70479214e-01 2.95821484e-02 2.44275406e-01 7.52332926e-01
6.07446074e-01 4.26846355e-01 7.78196573e-01 5.02829552e-01
-2.40540370e-01 -3.02328825e-01 3.56924206e-01 1.13487339e+00
-2.49933481e-01 -3.47994000e-01 -5.83494782e-01 8.37458432e-01
-2.24251914e+00 -1.26006567e+00 -3.48585576e-01 2.05184984e+00
7.85740077e-01 1.43371016e-01 -2.90075153e-01 1.04136974e-01
7.78616548e-01 3.12220693e-01 -1.15646958e+00 3.09669133e-03
-3.38751793e-01 4.31222558e-01 3.35787207e-01 -4.84726369e-01
-7.58850098e-01 4.58910793e-01 6.19719553e+00 5.59276581e-01
-1.33141947e+00 2.57698774e-01 1.32500783e-01 -4.25613374e-01
-1.29796043e-01 1.48687392e-01 -7.37593949e-01 9.45142627e-01
1.01598835e+00 -4.14632380e-01 5.33022523e-01 5.91645002e-01
1.09335639e-01 -1.48930639e-01 -1.07779408e+00 1.69889843e+00
3.53711993e-02 -1.45569646e+00 1.72017321e-01 -2.79651672e-01
6.42887533e-01 2.38480359e-01 -8.40830952e-02 -1.84434634e-02
6.77677384e-03 -6.55731201e-01 8.27631116e-01 9.95438218e-01
6.25080228e-01 -4.57047284e-01 3.62410367e-01 1.36328280e-01
-1.38617718e+00 1.37232970e-02 -6.86645269e-01 -3.54050249e-01
2.08796367e-01 9.02596295e-01 5.27898930e-02 2.60317415e-01
1.13714588e+00 1.19007909e+00 -2.57968038e-01 1.56169164e+00
6.63759261e-02 5.57197630e-01 -5.63152194e-01 -2.85467058e-01
-1.22605804e-02 -3.56984660e-02 6.23496354e-01 8.13767850e-01
5.53210855e-01 2.05886811e-01 1.85965315e-01 1.07431591e+00
-2.67894894e-01 -4.90515471e-01 -3.86996657e-01 -1.58561423e-01
7.31163800e-01 1.30486166e+00 -7.87607312e-01 -4.74781170e-02
-5.25550663e-01 1.19856048e+00 2.09121332e-01 2.06774339e-01
-1.02935529e+00 -4.75377768e-01 7.13263214e-01 -2.24781230e-01
2.64682442e-01 -3.37294638e-01 -1.53170988e-01 -1.34053862e+00
3.40330780e-01 -3.76066983e-01 2.82500476e-01 -1.14789855e+00
-1.47360206e+00 3.98974091e-01 -2.32992828e-01 -1.20554483e+00
2.84798920e-01 -7.43199646e-01 -7.33489990e-01 4.18968320e-01
-1.54234040e+00 -9.00296211e-01 -6.82142258e-01 1.10616052e+00
3.78420323e-01 1.56413943e-01 6.15459740e-01 6.45204186e-01
-7.08466411e-01 1.89340040e-01 1.36513814e-01 -8.65579173e-02
5.15858710e-01 -9.84946549e-01 1.51522666e-01 1.04830277e+00
2.26092488e-01 5.80786288e-01 3.97235721e-01 -6.47117078e-01
-2.25886559e+00 -1.25492179e+00 4.37529027e-01 -1.62851080e-01
6.28428936e-01 -2.58139014e-01 -8.94928455e-01 5.49180806e-01
-1.14195861e-01 6.22856796e-01 3.31697315e-01 -3.19015205e-01
-3.92824709e-01 -5.29036105e-01 -1.21163273e+00 6.10623717e-01
1.53465676e+00 -7.52488673e-01 -6.29275680e-01 2.58168340e-01
6.73262775e-01 -4.39803511e-01 -8.12244534e-01 2.33715683e-01
6.16348565e-01 -9.89059150e-01 9.24599946e-01 -8.45928118e-02
4.29654747e-01 -6.95217133e-01 -2.40339234e-01 -7.60787249e-01
-4.88251299e-01 -4.36707616e-01 -5.17143905e-01 1.29319966e+00
-3.39266926e-01 -6.32923543e-01 3.74008298e-01 1.65875703e-01
-2.88896561e-01 -6.90940559e-01 -1.30830836e+00 -7.19426811e-01
-6.59193754e-01 -6.67095721e-01 5.03648281e-01 5.57221234e-01
-3.54294479e-01 1.92916587e-01 -1.26524001e-01 2.74454057e-01
8.96019995e-01 3.25148940e-01 5.43373451e-02 -1.35534728e+00
-1.36842832e-01 -2.31529295e-01 -1.10031819e+00 -9.27894533e-01
-2.80371457e-01 -8.30491304e-01 1.98200420e-01 -1.69897294e+00
5.23330808e-01 -1.06361263e-01 -7.52430677e-01 3.63950968e-01
-7.35326996e-03 3.54103684e-01 -7.03895316e-02 4.03817624e-01
-8.93219411e-01 8.51711929e-01 1.17233300e+00 -8.31170157e-02
1.60906777e-01 -6.46217465e-01 -2.84793377e-01 8.34686160e-01
3.01614583e-01 -3.66342217e-01 -5.61289549e-01 -8.13215077e-01
3.53527367e-01 1.14011727e-01 1.10811758e+00 -1.52068269e+00
9.82232094e-01 -2.73782343e-01 6.46335483e-01 -7.76093543e-01
4.58212256e-01 -7.69588828e-01 5.31184137e-01 2.11116910e-01
5.45957945e-02 2.67975926e-01 1.78059623e-01 1.04029894e+00
-3.45191002e-01 6.37047291e-02 6.86300516e-01 -3.06599885e-02
-1.10707498e+00 1.00820756e+00 -1.92235306e-01 5.50643206e-02
9.96463239e-01 -3.39572757e-01 -6.47616982e-01 1.79107815e-01
-2.05725119e-01 -4.65196967e-02 3.00328404e-01 3.46055359e-01
1.03129303e+00 -1.54807019e+00 -2.20063373e-01 2.02190027e-01
1.44315153e-01 1.00995958e-01 6.15202665e-01 7.40320742e-01
-3.68139684e-01 3.35401893e-02 -7.15713739e-01 -8.65921855e-01
-8.49769294e-01 4.84152555e-01 1.51973888e-01 1.95336014e-01
-8.88274193e-01 8.01671505e-01 1.66212723e-01 1.73515230e-01
2.59938627e-01 -4.46276337e-01 7.62258917e-02 -1.24522373e-01
7.08045781e-01 3.36556256e-01 -8.37610289e-02 -2.31237352e-01
-4.38002527e-01 6.89027905e-01 -3.56104076e-02 -7.88606927e-02
1.31471050e+00 3.25051434e-02 -3.13820302e-01 7.96629488e-01
9.38087642e-01 -6.60737216e-01 -1.52090311e+00 -3.55017662e-01
-3.89786124e-01 -4.50176537e-01 3.55466336e-01 -5.58390021e-01
-1.05451369e+00 9.98394132e-01 8.98907006e-01 1.14113078e-01
1.47758949e+00 -3.90367180e-01 1.35316348e+00 3.31008941e-01
8.20885897e-01 -1.00526357e+00 1.29210636e-01 4.65783894e-01
7.37897158e-01 -8.57802749e-01 -1.23288482e-01 -1.90159887e-01
-1.21004023e-01 1.05154252e+00 6.73365653e-01 -4.56771344e-01
6.57302499e-01 3.54765624e-01 -7.36246288e-01 -2.17059627e-01
-9.75415289e-01 -1.33278921e-01 -7.96243101e-02 6.26719534e-01
3.91538553e-02 -1.03223048e-01 -4.18237746e-01 7.40012586e-01
4.49406169e-02 6.26907110e-01 1.97068244e-01 9.92669761e-01
-2.98282474e-01 -7.46164382e-01 -1.57459676e-01 7.79717326e-01
-4.24132288e-01 7.37090707e-02 -6.12310655e-02 3.18548322e-01
4.99492586e-01 7.50685453e-01 4.56339747e-01 -4.14565980e-01
2.97055036e-01 -1.53050512e-01 8.69694054e-01 -2.05263421e-01
-2.99270421e-01 -1.84342265e-01 -5.03410518e-01 -8.32105458e-01
-6.85498536e-01 -7.24395871e-01 -1.55965400e+00 -3.17525446e-01
-2.99259908e-02 -4.99818563e-01 6.43470883e-01 7.75534987e-01
6.42259836e-01 8.27747643e-01 3.96740377e-01 -5.66550970e-01
-5.89187164e-03 -5.44502378e-01 -5.04608095e-01 3.55375707e-01
1.78267196e-01 -8.00814211e-01 -1.44810706e-01 2.69072056e-01] | [8.679582595825195, -1.1901817321777344] |
1f852c4c-0302-4a34-b3b4-a7f03da58e51 | randomized-conditional-flow-matching-for | 2211.14575 | null | https://arxiv.org/abs/2211.14575v1 | https://arxiv.org/pdf/2211.14575v1.pdf | Randomized Conditional Flow Matching for Video Prediction | We introduce a novel generative model for video prediction based on latent flow matching, an efficient alternative to diffusion-based models. In contrast to prior work that either incurs a high training cost by modeling the past through a memory state, as in recurrent neural networks, or limits the computational load by conditioning only on a predefined window of past frames, we efficiently and effectively take the past into account by conditioning at inference time only on a small random set of past frames at each integration step of the learned flow. Moreover, to enable the generation of high-resolution videos and speed up the training, we work in the latent space of a pretrained VQGAN. Furthermore, we propose to approximate the initial condition of the flow ODE with the previous noisy frame. This allows to reduce the number of integration steps and hence, speed up the sampling at inference time. We call our model Random frame conditional flow Integration for VidEo pRediction, or, in short, RIVER. We show that RIVER achieves superior or on par performance compared to prior work on common video prediction benchmarks. | ['Paolo Favaro', 'Sepehr Sameni', 'Aram Davtyan'] | 2022-11-26 | null | null | null | null | ['video-prediction'] | ['computer-vision'] | [ 2.84915298e-01 8.59105811e-02 -2.18623579e-01 3.96690052e-03
-5.21301985e-01 -3.51495564e-01 7.16151237e-01 -1.26884148e-01
-4.20846105e-01 6.25038087e-01 4.48107660e-01 -2.70291001e-01
2.77898163e-01 -1.06797743e+00 -9.38131928e-01 -5.94704092e-01
6.01442484e-03 4.01365846e-01 3.13153118e-01 3.90433013e-01
9.86248776e-02 3.72594357e-01 -1.34769011e+00 1.60961658e-01
5.49119115e-01 9.16280091e-01 3.07365894e-01 1.04090500e+00
8.11367407e-02 1.46527636e+00 -3.21596980e-01 -4.98712033e-01
2.82350868e-01 -9.19602036e-01 -8.05776536e-01 3.27626169e-01
3.47722769e-01 -8.86638701e-01 -7.82460272e-01 8.14526260e-01
1.46535590e-01 4.15779769e-01 4.69868779e-01 -1.01968277e+00
-2.02242956e-01 6.19355381e-01 -1.24371305e-01 2.18143806e-01
1.93360239e-01 1.91533878e-01 9.40099120e-01 -5.81394076e-01
1.09287000e+00 9.53202724e-01 5.62309742e-01 6.22680783e-01
-1.30010760e+00 -4.57380474e-01 5.23252249e-01 3.26468021e-01
-1.12988257e+00 -6.98019803e-01 7.48701572e-01 -4.37063366e-01
9.19050395e-01 -2.01503746e-03 1.06516051e+00 1.11501372e+00
9.59122702e-02 8.58841360e-01 4.48889256e-01 -1.67446747e-01
4.80998069e-01 -3.05434018e-01 -3.00958484e-01 9.32759166e-01
-4.04471070e-01 5.32054380e-02 -6.52347684e-01 -1.96469814e-01
1.06003249e+00 2.26797968e-01 -5.25520563e-01 -2.31988505e-01
-1.06711364e+00 9.08220589e-01 1.36106923e-01 -3.71705219e-02
-5.35753191e-01 6.96938574e-01 3.17253113e-01 2.12183416e-01
5.74226379e-01 1.46854641e-02 -8.99677724e-02 -5.66949487e-01
-1.51557755e+00 1.75071031e-01 8.73244941e-01 6.52324319e-01
9.23939824e-01 2.77477860e-01 -1.51448235e-01 2.86771029e-01
2.56378651e-01 3.06781113e-01 2.93020189e-01 -1.69375396e+00
4.39525098e-01 1.48819201e-02 3.48497868e-01 -7.72341311e-01
1.80681482e-01 -4.35367882e-01 -7.60842383e-01 8.18492621e-02
5.01100659e-01 -2.84749329e-01 -8.69077861e-01 1.82035005e+00
1.25893950e-01 1.01152527e+00 -1.33397833e-01 6.78430438e-01
-8.08143616e-02 1.17733002e+00 -1.05782367e-01 -6.21317029e-01
6.76227808e-01 -1.22344828e+00 -4.47652340e-01 -1.21506959e-01
6.34688318e-01 -5.59251010e-01 5.66611886e-01 4.62427527e-01
-1.23069477e+00 -5.28739989e-01 -7.85357177e-01 3.66783217e-02
3.46926481e-01 -3.27513963e-02 6.73306286e-01 4.66546327e-01
-1.14431679e+00 1.08388162e+00 -1.53357291e+00 7.64795318e-02
4.45279390e-01 1.19952120e-01 -1.21288076e-01 -2.08676934e-01
-8.58120978e-01 4.22372431e-01 6.37394041e-02 7.09148264e-03
-1.23097658e+00 -7.87326336e-01 -8.70966494e-01 2.10965931e-01
3.50401610e-01 -7.85787046e-01 1.20627820e+00 -1.16508830e+00
-1.90113425e+00 9.29452255e-02 -6.14338934e-01 -1.02104437e+00
8.16130817e-01 -4.31571841e-01 2.14548647e-01 4.36286002e-01
-1.96784288e-01 8.46841753e-01 1.07837808e+00 -6.83703125e-01
-3.51110637e-01 2.81388283e-01 2.22867355e-01 1.69626385e-01
-6.13699220e-02 -2.65038043e-01 -7.35111535e-01 -5.79037547e-01
-1.70610800e-01 -1.15037775e+00 -4.92681563e-01 1.99540973e-01
1.24318875e-01 1.02815919e-01 7.31290936e-01 -7.25097775e-01
1.16542220e+00 -2.13427520e+00 4.43008989e-01 9.66557413e-02
2.19775349e-01 2.18859941e-01 -5.51750278e-03 2.56979913e-01
1.74729258e-01 3.02375145e-02 -1.51639268e-01 -7.85002053e-01
-2.01463595e-01 5.55675685e-01 -7.94032812e-01 4.91523713e-01
-6.64502382e-04 8.08840096e-01 -1.12477541e+00 -5.05256653e-01
4.69047159e-01 7.89281785e-01 -1.12519062e+00 2.52264351e-01
-4.91617024e-01 6.41853750e-01 -3.24030101e-01 -1.68998301e-01
2.54313499e-01 -5.17322183e-01 3.10022831e-01 -6.05595335e-02
-5.62439859e-02 6.98618591e-01 -1.19946265e+00 1.80945444e+00
-7.36328542e-01 6.84126496e-01 -3.46590012e-01 -1.00169384e+00
5.08013248e-01 5.97128987e-01 6.79068089e-01 -3.65658373e-01
-1.16303042e-02 -1.18582472e-01 -3.15186828e-01 -8.50159302e-02
4.38963771e-01 8.12810287e-03 4.90361512e-01 6.34362578e-01
2.29501143e-01 2.30484545e-01 3.58441889e-01 3.68528903e-01
1.08991611e+00 7.29012847e-01 -9.98972654e-02 2.99659878e-01
3.35378706e-01 -3.35285187e-01 8.39069605e-01 8.06355834e-01
1.84925273e-02 6.91207290e-01 7.54609585e-01 -3.49646509e-01
-1.23364949e+00 -8.47304702e-01 4.19559479e-01 6.77747786e-01
-1.62331983e-01 -8.65665913e-01 -7.14378655e-01 -6.08823955e-01
-5.32262444e-01 5.66205680e-01 -5.48748195e-01 7.31814140e-03
-1.08376575e+00 -4.33345437e-01 1.83180884e-01 6.69465125e-01
4.40321147e-01 -9.63877678e-01 -8.90064299e-01 5.91388404e-01
-2.47075424e-01 -1.15972888e+00 -6.46941423e-01 -7.68703371e-02
-1.09585547e+00 -6.57875776e-01 -8.27842772e-01 -2.82085449e-01
4.06629980e-01 -1.65467024e-01 1.22904456e+00 -7.90056214e-03
8.73585269e-02 3.11687976e-01 -2.59703696e-01 4.37011540e-01
-5.25748909e-01 2.79933903e-02 -3.60644430e-01 1.73841804e-01
-3.22218120e-01 -6.71379566e-01 -7.64057577e-01 -2.28572339e-02
-8.23444009e-01 4.32975113e-01 1.95202634e-01 9.84274089e-01
6.32675290e-01 -9.31994542e-02 2.15812889e-03 -8.67573559e-01
-1.04805633e-01 -4.31419909e-01 -8.46054912e-01 4.27230522e-02
-3.43986541e-01 3.54462594e-01 7.04926133e-01 -5.90931952e-01
-1.05900240e+00 1.93788067e-01 -2.75927216e-01 -9.64809656e-01
2.18397409e-01 4.57720906e-01 3.12404424e-01 2.05456465e-01
1.66951492e-01 3.92887145e-01 -7.98472166e-02 -3.23419422e-01
4.35488850e-01 -1.79474913e-02 4.33460534e-01 -4.10478562e-01
5.08038938e-01 6.10716343e-01 1.71362251e-01 -6.33794367e-01
-6.59291685e-01 -1.37853861e-01 -3.91434520e-01 -2.91880518e-01
8.72959256e-01 -1.04209590e+00 -7.41323233e-01 3.79640996e-01
-1.17746198e+00 -7.01468468e-01 -5.65139532e-01 6.60520256e-01
-7.26077378e-01 4.81708109e-01 -1.13127041e+00 -8.69237721e-01
-1.77189305e-01 -1.18262899e+00 8.25824082e-01 1.03749372e-02
-3.40166152e-01 -1.17263937e+00 1.21282592e-01 -4.33291718e-02
3.29216927e-01 1.57818854e-01 5.65249622e-01 -4.72917110e-02
-1.17859101e+00 -9.33157429e-02 9.26114470e-02 3.16631824e-01
-1.77767336e-01 2.96036094e-01 -6.49642885e-01 -1.91326827e-01
2.09367946e-02 -3.27866077e-02 1.16235745e+00 4.06821996e-01
7.90579617e-01 -5.17974019e-01 -2.95802921e-01 7.11821377e-01
1.31361735e+00 1.14294343e-01 6.35800064e-01 -9.74515006e-02
6.69245362e-01 1.91619113e-01 3.55361432e-01 7.22163677e-01
2.57472128e-01 6.84420168e-01 2.43441001e-01 2.61867642e-01
-3.46741587e-01 -7.14249492e-01 5.86680472e-01 8.22018743e-01
-2.70509332e-01 -3.73481244e-01 -5.95670402e-01 6.66136384e-01
-2.02488041e+00 -1.23380017e+00 3.09613347e-01 2.19615602e+00
6.22143090e-01 2.37754136e-01 1.07079275e-01 1.45289108e-01
3.83278072e-01 4.54903483e-01 -4.27682996e-01 -6.64257184e-02
1.57807469e-01 5.02232015e-01 6.03711426e-01 1.04033899e+00
-9.12597060e-01 9.30200994e-01 6.22534275e+00 5.99087358e-01
-1.38059628e+00 2.49385431e-01 6.36334896e-01 -3.51881146e-01
-4.23067927e-01 4.10853982e-01 -8.32468569e-01 4.39844877e-01
1.23364949e+00 1.23927027e-01 6.69670701e-01 6.38518155e-01
1.62916780e-01 -9.22892690e-02 -1.05399358e+00 8.13414931e-01
-1.46539971e-01 -1.68513787e+00 1.36556849e-01 1.99089810e-01
7.32850730e-01 2.71640241e-01 -8.77195597e-02 9.63723101e-03
3.39533538e-01 -7.54659712e-01 8.81851912e-01 7.43528664e-01
4.70161557e-01 -6.38505638e-01 3.97303462e-01 3.48592281e-01
-1.23398578e+00 -5.06665297e-02 -3.12529087e-01 -2.02264935e-01
5.26924253e-01 5.84027171e-01 -5.56369185e-01 2.68186033e-01
3.70098263e-01 9.99161541e-01 -1.04962349e-01 8.70085359e-01
-2.41097167e-01 1.16264856e+00 -5.88149905e-01 3.43640953e-01
3.06888938e-01 -3.59617412e-01 4.75115448e-01 9.55994070e-01
4.75470155e-01 -1.84870735e-01 2.39465684e-01 8.10470641e-01
-1.04816724e-02 -2.40868106e-01 -3.91819566e-01 -6.53101057e-02
3.01373512e-01 7.84727275e-01 -6.48323357e-01 -5.85090220e-01
-4.67031598e-01 1.27226174e+00 2.38822356e-01 5.55102885e-01
-9.32237566e-01 8.18563551e-02 4.36872005e-01 1.44500032e-01
8.43183994e-01 -5.57361364e-01 1.66396841e-01 -1.52777743e+00
-6.85794838e-03 -4.17494506e-01 1.64100885e-01 -6.24607921e-01
-5.32821536e-01 5.27660966e-01 -2.11978272e-01 -1.11516726e+00
-1.00384748e+00 -1.02892444e-01 -5.67904890e-01 8.40170979e-01
-1.53228188e+00 -8.59355569e-01 7.09706992e-02 5.37689269e-01
5.73780298e-01 2.68933922e-01 7.59807944e-01 3.18555474e-01
-3.75529587e-01 1.35418773e-01 -1.01324273e-02 1.49828658e-01
4.51566756e-01 -9.85481143e-01 6.77313089e-01 1.11551630e+00
4.53369886e-01 3.55660915e-01 7.38231599e-01 -5.89516938e-01
-1.14516544e+00 -1.00659609e+00 9.62718725e-01 -7.45068565e-02
5.64013660e-01 -3.65938902e-01 -9.03032959e-01 9.39238369e-01
1.40004709e-01 2.66195267e-01 3.54409188e-01 -2.52258539e-01
-2.10062712e-01 -8.17263797e-02 -5.54976046e-01 5.30640900e-01
9.78072822e-01 -6.10338330e-01 -3.75853889e-02 1.17557622e-01
5.65879762e-01 -4.65746820e-01 -8.17004204e-01 1.01889737e-01
5.22448003e-01 -1.09276521e+00 8.13594222e-01 -1.94359660e-01
5.62474787e-01 -3.58943641e-01 1.25585543e-02 -9.09168899e-01
-2.10789353e-01 -1.03717482e+00 -6.98615313e-01 1.05980837e+00
2.23543599e-01 -3.08317900e-01 1.27401197e+00 6.48300469e-01
1.30966082e-01 -6.92118347e-01 -8.43549848e-01 -4.03706700e-01
-1.55352652e-01 -6.22527838e-01 3.59185457e-01 5.64607859e-01
-3.27907771e-01 1.18755944e-01 -8.37476313e-01 1.05303191e-01
5.68314075e-01 2.65094787e-01 6.45181894e-01 -7.14017332e-01
-1.05447721e+00 -1.96334988e-01 -3.18596393e-01 -1.73425806e+00
2.68859893e-01 -5.53110301e-01 -1.26999542e-01 -1.20310259e+00
-3.81907411e-02 -2.57440627e-01 -8.58018994e-02 1.81573123e-01
-5.70325814e-02 3.50909442e-01 4.46966976e-01 3.62439573e-01
-5.98705173e-01 6.13661230e-01 1.15007544e+00 1.19543560e-01
-1.53502405e-01 2.00740788e-02 1.04509622e-01 7.32264400e-01
4.89108831e-01 -4.43667173e-01 -6.44027829e-01 -4.29774374e-01
3.84595305e-01 8.05003166e-01 4.83055472e-01 -1.13066328e+00
3.36929888e-01 -9.58152339e-02 3.14611971e-01 -4.02286887e-01
6.50425255e-01 -5.20406365e-01 5.08473217e-01 6.31342769e-01
-4.53365147e-01 -4.61532176e-02 -2.56326526e-01 6.84011579e-01
-3.28467220e-01 -3.15570682e-01 6.79159939e-01 -2.19888568e-01
-4.47532952e-01 5.60479999e-01 -7.07033753e-01 -1.92630440e-01
6.20851338e-01 -1.14802998e-02 5.69220707e-02 -8.79525661e-01
-8.93805623e-01 -6.76482096e-02 5.34174263e-01 1.24732018e-01
5.10439157e-01 -1.11351824e+00 -2.84295678e-01 1.78837612e-01
-4.85304028e-01 -5.41041307e-02 3.33680481e-01 7.18252957e-01
-6.54065847e-01 2.16320679e-01 6.38929605e-02 -5.14635623e-01
-9.61754560e-01 4.64479834e-01 4.46528137e-01 -5.32641709e-01
-8.67599845e-01 7.99151301e-01 1.37612134e-01 2.75936663e-01
1.33375704e-01 -1.85895473e-01 -1.42806610e-02 -1.87814891e-01
5.14749944e-01 3.71955216e-01 -3.90425354e-01 -6.05957031e-01
3.10239214e-02 4.89664197e-01 6.75949678e-02 -4.49180067e-01
1.24536633e+00 -1.97498441e-01 3.50357443e-02 5.38929224e-01
1.22179949e+00 -3.22311558e-03 -2.09019136e+00 -2.87421376e-01
-3.16629022e-01 -4.33943093e-01 2.67744839e-01 -2.02259555e-01
-1.27114749e+00 9.36334133e-01 3.28425795e-01 -7.48139992e-02
9.18401182e-01 -3.08704019e-01 1.06628883e+00 3.11015427e-01
2.10952684e-01 -7.14923501e-01 -9.85428542e-02 5.05798995e-01
2.40584359e-01 -7.29133189e-01 -6.51098862e-02 -3.53181809e-01
-3.75202745e-01 1.21758044e+00 1.69309214e-01 -5.72353959e-01
7.69235849e-01 4.15093541e-01 -1.79463044e-01 3.62641245e-01
-1.25665164e+00 1.85838878e-01 5.04113957e-02 3.35673273e-01
4.45615202e-01 -2.92519033e-01 1.39572382e-01 -1.45901233e-01
-9.18432474e-02 4.76782531e-01 5.65383971e-01 9.10748899e-01
-3.41048330e-01 -1.23904133e+00 1.49398863e-01 2.91583121e-01
-6.62605584e-01 -2.95596331e-01 3.66676390e-01 3.37219954e-01
-2.27142069e-02 6.30844831e-01 4.87726033e-01 -1.98159851e-02
-3.06184083e-01 1.42217472e-01 6.42330825e-01 -2.99758703e-01
-2.21069753e-01 3.99405420e-01 -2.66616307e-02 -8.61677647e-01
-6.17881298e-01 -7.76777744e-01 -9.47324634e-01 -5.17403305e-01
-1.80090636e-01 2.80658662e-01 2.61153340e-01 1.12373245e+00
3.67606252e-01 3.99182022e-01 4.82476622e-01 -1.01513898e+00
-1.49376333e-01 -7.41100013e-01 -1.89954638e-01 2.74867147e-01
4.57920551e-01 -2.72073269e-01 -2.24521041e-01 5.11573017e-01] | [10.682145118713379, -0.7093892097473145] |
64d2ce42-3b1c-4d23-8df4-557487d61d06 | interactive-ontology-debugging-two-query | 1107.4303 | null | http://arxiv.org/abs/1107.4303v2 | http://arxiv.org/pdf/1107.4303v2.pdf | Interactive ontology debugging: two query strategies for efficient fault localization | Effective debugging of ontologies is an important prerequisite for their
broad application, especially in areas that rely on everyday users to create
and maintain knowledge bases, such as the Semantic Web. In such systems
ontologies capture formalized vocabularies of terms shared by its users.
However in many cases users have different local views of the domain, i.e. of
the context in which a given term is used. Inappropriate usage of terms
together with natural complications when formulating and understanding logical
descriptions may result in faulty ontologies. Recent ontology debugging
approaches use diagnosis methods to identify causes of the faults. In most
debugging scenarios these methods return many alternative diagnoses, thus
placing the burden of fault localization on the user. This paper demonstrates
how the target diagnosis can be identified by performing a sequence of
observations, that is, by querying an oracle about entailments of the target
ontology. To identify the best query we propose two query selection strategies:
a simple "split-in-half" strategy and an entropy-based strategy. The latter
allows knowledge about typical user errors to be exploited to minimize the
number of queries. Our evaluation showed that the entropy-based method
significantly reduces the number of required queries compared to the
"split-in-half" approach. We experimented with different probability
distributions of user errors and different qualities of the a-priori
probabilities. Our measurements demonstrated the superiority of entropy-based
query selection even in cases where all fault probabilities are equal, i.e.
where no information about typical user errors is available. | ['Kostyantyn Shchekotykhin', 'Gerhard Friedrich', 'Philipp Fleiss', 'Patrick Rodler'] | 2011-07-20 | null | null | null | null | ['fault-localization'] | ['computer-code'] | [-4.62718569e-02 3.16705912e-01 1.30557254e-01 -3.03289443e-01
-5.53520620e-01 -5.31206608e-01 1.42067358e-01 6.44178391e-01
-2.64959574e-01 7.20006406e-01 -3.65971744e-01 -3.16688687e-01
-4.37035918e-01 -9.36129034e-01 -5.03394008e-01 -3.33507150e-01
1.18814548e-02 7.55725384e-01 6.61524415e-01 -1.72493890e-01
3.22780490e-01 2.70444393e-01 -1.94941676e+00 1.16670541e-01
1.18942559e+00 7.99448252e-01 3.25072259e-01 5.29437542e-01
-3.45048517e-01 9.32139814e-01 -1.00377131e+00 -3.25875223e-01
1.95904169e-03 -3.94585252e-01 -1.11817932e+00 1.84706643e-01
-1.24043390e-01 -9.64443833e-02 6.01745099e-02 1.61379635e+00
1.12349071e-01 -8.69227573e-02 1.45293787e-01 -1.52969205e+00
-7.48788193e-02 4.89457250e-01 1.71491459e-01 2.04300940e-01
9.91732717e-01 -2.63885800e-02 9.27365303e-01 -4.34093654e-01
7.62669861e-01 9.88263488e-01 5.15086889e-01 2.20132887e-01
-1.14545071e+00 -2.39849105e-01 -5.76351620e-02 5.12699187e-01
-1.65604448e+00 -3.61480623e-01 3.62585813e-01 -4.81557548e-01
1.05676031e+00 6.35237634e-01 2.60188580e-01 5.20100951e-01
1.94126904e-01 6.29443824e-02 9.17425156e-01 -7.50998437e-01
5.46898186e-01 7.64333248e-01 1.20757200e-01 8.86369288e-01
7.30279624e-01 -2.76648879e-01 -5.11283815e-01 -7.22945571e-01
3.57456833e-01 -1.54461727e-01 -4.74982381e-01 -4.37090158e-01
-8.63245726e-01 5.23780525e-01 -1.80167928e-01 5.27609229e-01
-3.33478630e-01 -6.47568628e-02 2.84773409e-01 7.08850384e-01
1.25239417e-01 7.59452462e-01 -6.45168424e-01 -2.33343422e-01
-6.98544383e-01 3.97019535e-01 1.32921660e+00 1.10283685e+00
8.60133469e-01 -3.70993257e-01 2.38935545e-01 6.46288812e-01
3.64623249e-01 2.61895567e-01 5.17764688e-01 -8.68343294e-01
3.27710986e-01 9.92618442e-01 5.10867059e-01 -7.32056797e-01
-5.81943803e-02 -1.39246643e-01 -2.05211677e-02 4.31555599e-01
6.73565328e-01 1.05950207e-01 -5.69165289e-01 1.58534181e+00
3.44786912e-01 -1.15304396e-01 3.50838788e-02 5.57935297e-01
1.77516844e-02 1.19026847e-01 -3.41121346e-01 -2.79161900e-01
1.61166251e+00 -3.13921094e-01 -7.38518178e-01 -1.21573687e-01
8.01155746e-01 -5.97281575e-01 8.76700282e-01 6.43203437e-01
-9.25063908e-01 4.03405829e-05 -1.10297811e+00 3.62935871e-01
-4.64679629e-01 -3.15969706e-01 1.57364771e-01 7.83248544e-01
-9.70278740e-01 7.91259944e-01 -8.26827705e-01 -4.89118338e-01
-5.39152101e-02 3.36031973e-01 -3.34180772e-01 -1.91723540e-01
-1.20912993e+00 1.05155349e+00 6.49955332e-01 -3.00544232e-01
-7.62852490e-01 -2.91280180e-01 -7.34388113e-01 4.05968338e-01
8.72151971e-01 -4.59618956e-01 1.50128925e+00 -7.84864008e-01
-9.28597748e-01 5.51037669e-01 -3.84180039e-01 -3.97077024e-01
5.62022805e-01 -1.03422347e-02 -4.88263130e-01 1.16990887e-01
2.84718990e-01 -2.42242724e-01 5.49161375e-01 -1.14014959e+00
-7.60391057e-01 -2.93643206e-01 4.74919081e-01 -9.78633910e-02
-4.19180952e-02 8.41378942e-02 -3.41289401e-01 -1.85527522e-02
3.93441767e-01 -7.67060816e-01 -2.67207175e-02 8.50943178e-02
-4.04736817e-01 -1.29047096e-01 6.13985598e-01 -6.02587163e-01
1.32945776e+00 -1.86342657e+00 -5.68327568e-02 6.14468992e-01
2.46057332e-01 -9.17561129e-02 4.90052193e-01 8.16378713e-01
-1.80784747e-01 3.81206512e-01 -1.65252522e-01 -1.85699463e-02
5.64648546e-02 4.32358056e-01 -1.93566337e-01 3.06865633e-01
1.24494098e-01 -4.89305984e-03 -9.75116134e-01 -5.02188981e-01
3.98044772e-02 -5.01338057e-02 -6.71493173e-01 4.41323161e-01
-6.92961156e-01 2.07337618e-01 -5.15857816e-01 6.14207983e-01
2.70397544e-01 -4.45681572e-01 5.54226756e-01 1.11009873e-01
2.84549557e-02 5.56355774e-01 -1.54500151e+00 1.26972628e+00
-5.05906165e-01 2.74587810e-01 -1.87824130e-01 -8.47039580e-01
7.20438004e-01 8.14330876e-01 4.06707853e-01 -4.86649334e-01
-8.43774620e-03 6.63609147e-01 -8.29930082e-02 -8.80786419e-01
2.09807515e-01 1.04004052e-02 3.32723074e-02 3.04178864e-01
-2.73071881e-02 6.03363216e-02 4.23885375e-01 5.52342571e-02
1.53832006e+00 -1.06870525e-01 8.03204596e-01 -3.13327909e-01
6.39888704e-01 2.01843232e-01 7.24965632e-01 7.19800532e-01
5.12583032e-02 2.46718600e-01 1.15197814e+00 -2.23820463e-01
-8.63232374e-01 -6.44588828e-01 -1.07519694e-01 4.53163177e-01
2.82107502e-01 -7.58167207e-01 -7.24851370e-01 -7.81967938e-01
-6.30617142e-02 1.10835409e+00 -3.09413463e-01 -2.46010825e-01
-1.68647692e-01 -2.08377630e-01 4.59236264e-01 1.36717726e-02
1.88926771e-01 -8.12797904e-01 -1.15742564e+00 5.63441515e-02
-2.78893650e-01 -9.36115384e-01 1.29309610e-01 2.25951687e-01
-7.96269119e-01 -1.60390103e+00 7.02477023e-02 -1.98842198e-01
8.41971099e-01 -1.68473050e-01 1.16680586e+00 6.40668154e-01
-8.76647234e-02 3.54156554e-01 -5.68282843e-01 -2.64566481e-01
-7.52266288e-01 -2.63427883e-01 -5.52577451e-02 -2.78812110e-01
3.74611050e-01 -4.96990651e-01 1.88665576e-02 3.82422984e-01
-1.20438600e+00 -4.15799588e-01 3.32320511e-01 7.39384413e-01
2.48976514e-01 6.96244538e-01 2.66108006e-01 -1.21170866e+00
6.26454830e-01 -7.24705577e-01 -9.65791762e-01 6.12867773e-01
-1.03101397e+00 5.49259245e-01 5.61312139e-01 -2.41541684e-01
-6.88874304e-01 -3.10679287e-01 -5.75819127e-02 -3.02305222e-01
-1.88137695e-01 7.58113623e-01 -4.21048939e-01 2.50948686e-02
6.50546372e-01 -1.08046465e-01 -2.12064371e-01 -5.11917233e-01
-3.27411979e-01 6.03563428e-01 2.18115509e-01 -6.80097520e-01
5.29449821e-01 6.10228442e-02 -2.92957336e-01 -6.83953881e-01
-3.12802106e-01 -4.17908072e-01 -6.93594813e-02 -3.48187536e-02
5.89090288e-01 -4.92648929e-01 -7.70342588e-01 2.14775413e-01
-1.19086087e+00 4.83503155e-02 -2.54840106e-01 1.71869814e-01
-4.46718872e-01 3.87182236e-01 -1.83797240e-01 -1.04242849e+00
2.54721284e-01 -1.55021703e+00 8.38482082e-01 3.22422944e-02
-5.59292793e-01 -8.21471989e-01 -6.46244884e-02 8.79300088e-02
3.31538707e-01 2.31563859e-02 1.22076643e+00 -1.28631794e+00
-7.03466415e-01 -5.37591577e-01 1.85671911e-01 2.45218381e-01
1.55015126e-01 -1.88283548e-01 -9.58049655e-01 -1.74713597e-01
2.26376891e-01 7.54898489e-02 -5.92072643e-02 -3.51660490e-01
1.00022602e+00 -5.25525570e-01 -3.32375258e-01 -9.10202190e-02
1.72478652e+00 4.11398590e-01 7.59367883e-01 4.27237809e-01
2.98531741e-01 6.28199458e-01 6.34351671e-01 6.72834694e-01
2.16313899e-01 8.86718869e-01 5.30425668e-01 4.91489470e-01
5.41933537e-01 3.01899575e-02 1.82282045e-01 2.45465174e-01
2.32705623e-01 -4.37277555e-01 -1.17668688e+00 5.82253873e-01
-1.64313984e+00 -7.73464143e-01 1.93090588e-01 2.78812575e+00
8.97421002e-01 4.21669185e-01 -1.73708633e-01 5.14665186e-01
7.09416509e-01 -3.23910296e-01 -3.22085172e-01 -1.35546878e-01
4.29300785e-01 4.13428396e-02 4.53338325e-01 8.00470173e-01
-5.12127221e-01 2.71875799e-01 5.46706724e+00 3.52408975e-01
-8.57867301e-01 1.62596732e-01 -2.07756534e-01 1.92510530e-01
-5.14032900e-01 5.07906854e-01 -5.70699692e-01 7.06594229e-01
1.11334598e+00 -4.33176965e-01 4.64281380e-01 9.03669298e-01
-2.71987915e-02 -5.71423054e-01 -1.29008734e+00 6.00668490e-01
-8.31286609e-02 -9.77548838e-01 -3.67155552e-01 1.29538193e-01
2.12642401e-01 -1.97717831e-01 -7.30975091e-01 -9.21549834e-03
1.86804622e-01 -6.37885690e-01 9.57128227e-01 7.58461833e-01
5.02009690e-01 -5.69704294e-01 1.16031957e+00 5.27301729e-01
-8.61622572e-01 -2.06029356e-01 -2.40319613e-02 4.35525272e-03
-1.42156351e-02 8.25391769e-01 -1.06610966e+00 7.69828260e-01
7.30637372e-01 -8.44813511e-02 -4.32743043e-01 1.25391567e+00
-1.75867096e-01 3.00031871e-01 -4.83800262e-01 1.30888879e-01
-4.17380750e-01 -2.52088010e-02 7.78478980e-01 8.33925307e-01
5.74818254e-01 -2.10892826e-01 2.92338282e-01 1.02291286e+00
1.18308865e-01 -1.73249915e-01 -6.52749240e-01 -1.81801349e-01
9.46421742e-01 6.72656715e-01 -6.47993565e-01 -5.32488465e-01
-4.40582275e-01 8.01477492e-01 6.35296777e-02 2.80530512e-01
-7.12757230e-01 -6.59640491e-01 6.33980632e-01 3.88199568e-01
2.02062339e-01 8.09078012e-03 4.81546931e-02 -1.12071979e+00
4.37749594e-01 -1.17960238e+00 3.85129064e-01 -8.29483747e-01
-8.92404795e-01 5.59937239e-01 4.38313216e-01 -1.09238350e+00
-5.90978444e-01 -3.82074654e-01 -4.20860767e-01 1.03539145e+00
-1.12887132e+00 -3.22734296e-01 -3.31266284e-01 3.65017474e-01
2.42126957e-01 2.60854542e-01 1.07912886e+00 4.59705889e-01
-3.89338553e-01 3.11464667e-01 -2.98590302e-01 -2.40857810e-01
6.39443457e-01 -1.43348742e+00 -1.25325531e-01 1.00592506e+00
4.05604802e-02 8.16661000e-01 1.26924670e+00 -8.58291149e-01
-1.22012866e+00 -6.47298574e-01 1.40332866e+00 -1.79472908e-01
6.51229322e-01 -6.51508868e-02 -1.29668903e+00 7.41880417e-01
1.04076518e-02 -1.40058309e-01 3.86713237e-01 -7.53574073e-02
-2.31989726e-01 -1.43663466e-01 -1.54962420e+00 4.30373132e-01
5.83515942e-01 -7.20702231e-01 -6.87721014e-01 5.43780684e-01
6.73345983e-01 -3.74438822e-01 -1.06709516e+00 2.04358011e-01
2.52306819e-01 -1.33523309e+00 4.39780354e-01 -3.87672991e-01
2.91624330e-02 -6.75938368e-01 -2.37481162e-01 -1.01476371e+00
1.70353979e-01 -4.72859859e-01 -7.94200301e-02 1.13475251e+00
4.50256556e-01 -1.10622919e+00 3.17302734e-01 9.22118545e-01
-7.92617798e-02 -6.24165595e-01 -9.68845248e-01 -7.43431687e-01
-7.23853588e-01 -4.79444891e-01 8.15016508e-01 9.71586823e-01
3.27886969e-01 -1.72564656e-01 8.59523490e-02 8.14744830e-01
2.55727857e-01 -1.90446407e-01 3.88682604e-01 -1.52628875e+00
-5.94262421e-01 -2.11385012e-01 -7.02278316e-01 -3.70852202e-01
-1.20024323e-01 -5.56968153e-01 1.51001453e-01 -1.24068487e+00
-2.51770020e-01 -6.06739521e-01 -5.82359284e-02 5.75617254e-01
-4.31052148e-02 -4.89942312e-01 -1.75728768e-01 2.45256156e-01
-5.12799799e-01 -7.58443475e-02 3.56371582e-01 1.72176123e-01
-1.33223264e-02 5.03984243e-02 -4.55461174e-01 7.18370795e-01
6.65025473e-01 -8.98058116e-01 -6.00827038e-01 -1.42399311e-01
7.51319885e-01 4.57452893e-01 4.56826776e-01 -1.19898868e+00
4.44755524e-01 -2.56088018e-01 -3.34843606e-01 -7.13065341e-02
1.12914450e-01 -1.25535774e+00 6.09243870e-01 5.37839949e-01
-2.15925798e-01 3.72575939e-01 -5.19474931e-02 5.36220372e-01
-3.84922862e-01 -9.43026006e-01 5.15366018e-01 -3.19027305e-01
-5.74663877e-01 -3.24733227e-01 -4.34537053e-01 -1.86883509e-02
1.03616321e+00 -1.03114799e-01 -1.79192632e-01 -2.90124416e-01
-7.81268299e-01 1.51795402e-01 9.32694435e-01 4.56582196e-02
3.44104767e-01 -8.12172771e-01 -1.86236829e-01 3.08352441e-01
4.52480137e-01 6.93363184e-03 -2.37442225e-01 9.64355886e-01
-7.09765255e-01 2.37338528e-01 9.42336544e-02 -4.20942873e-01
-1.30323935e+00 5.62373161e-01 6.11606836e-01 -4.05022860e-01
-4.06295270e-01 5.63987732e-01 -3.71859759e-01 -2.07810566e-01
2.82194316e-01 -5.97851276e-01 -4.64854240e-02 -2.02904761e-01
4.77259189e-01 4.19503301e-01 4.99697685e-01 -1.09569676e-01
-4.91797239e-01 5.71993366e-02 7.05497786e-02 -1.79544970e-01
9.09391344e-01 -1.34034902e-01 -4.43493962e-01 7.97767043e-01
6.15948081e-01 3.35545182e-01 -4.92972583e-01 -1.73875362e-01
6.85496747e-01 -6.89621866e-01 -2.37511903e-01 -8.31745565e-01
-7.49932587e-01 5.17190576e-01 3.48675251e-01 9.28381085e-01
1.15679598e+00 3.54334190e-02 3.15858245e-01 5.95473111e-01
8.72449160e-01 -8.19654286e-01 -3.99428159e-01 2.21405491e-01
6.88588023e-01 -9.60728467e-01 -1.13047855e-02 -6.29781544e-01
-4.99657273e-01 1.14433169e+00 4.85535175e-01 1.13636747e-01
3.58792275e-01 2.37711772e-01 -1.02962643e-01 -5.20950019e-01
-9.20130908e-01 -1.05302013e-01 -7.29352608e-02 2.71190912e-01
2.96827495e-01 -1.23303011e-01 -3.28622043e-01 3.47558916e-01
-7.61675909e-02 1.73756078e-01 7.28162706e-01 1.41501355e+00
-4.91403699e-01 -1.43382096e+00 -6.66507304e-01 4.93593693e-01
-4.68879730e-01 9.55819190e-02 -1.95425659e-01 7.73621321e-01
3.68423462e-01 1.11277020e+00 -1.63161844e-01 -3.48902196e-01
5.77827096e-01 5.95288515e-01 3.96393269e-01 -8.85470092e-01
-4.19249654e-01 -2.39955485e-01 5.63957930e-01 -6.89724326e-01
-3.85365523e-02 -5.87056160e-01 -1.31704104e+00 -1.84786811e-01
-7.19856203e-01 5.14240980e-01 5.61146796e-01 1.30770469e+00
3.10009450e-01 5.53283095e-01 2.03847110e-01 -1.78627729e-01
-9.73531783e-01 -8.29287648e-01 -7.09546387e-01 4.53727096e-01
3.29955041e-01 -9.29037571e-01 -3.94912660e-01 9.52773541e-02] | [5.473856449127197, 2.8131134510040283] |
a61089d8-7715-4a2c-aeb6-2fd84ca3d7d8 | synthetic-medical-images-from-dual-generative | 1709.01872 | null | http://arxiv.org/abs/1709.01872v3 | http://arxiv.org/pdf/1709.01872v3.pdf | Synthetic Medical Images from Dual Generative Adversarial Networks | Currently there is strong interest in data-driven approaches to medical image
classification. However, medical imaging data is scarce, expensive, and fraught
with legal concerns regarding patient privacy. Typical consent forms only allow
for patient data to be used in medical journals or education, meaning the
majority of medical data is inaccessible for general public research. We
propose a novel, two-stage pipeline for generating synthetic medical images
from a pair of generative adversarial networks, tested in practice on retinal
fundi images. We develop a hierarchical generation process to divide the
complex image generation task into two parts: geometry and photorealism. We
hope researchers will use our pipeline to bring private medical data into the
public domain, sparking growth in imaging tasks that have previously relied on
the hand-tuning of models. We have begun this initiative through the
development of SynthMed, an online repository for synthetic medical images. | ['Tejpal S. Virdi', 'Peter S. Li', 'John T. Guibas'] | 2017-09-06 | null | null | null | null | ['medical-image-generation'] | ['medical'] | [ 4.89488125e-01 6.03057683e-01 2.08398774e-01 -6.61356807e-01
-1.08083975e+00 -7.00860441e-01 4.00638759e-01 1.11084342e-01
-3.22871029e-01 7.76137888e-01 2.65328437e-01 -7.34765053e-01
3.33275259e-01 -7.18691230e-01 -6.35985255e-01 -4.43219841e-01
2.25370899e-01 4.14139986e-01 -1.15843542e-01 -4.55213450e-02
1.81498863e-02 5.03812969e-01 -1.08501101e+00 4.49438572e-01
8.02134514e-01 7.07043231e-01 -2.64750272e-01 7.77299821e-01
4.13299829e-01 8.38947535e-01 -6.17334306e-01 -6.46151066e-01
6.12824559e-01 -7.51896679e-01 -7.98085272e-01 2.08324015e-01
4.57865149e-01 -8.23241174e-01 -2.83878535e-01 1.02882957e+00
7.55887032e-01 -3.09087068e-01 3.79155815e-01 -1.21347535e+00
-7.68763721e-01 3.76935422e-01 -4.28174198e-01 -8.68280903e-02
1.10210061e-01 5.88218510e-01 6.16531432e-01 -4.31562901e-01
9.74433482e-01 7.95999646e-01 4.98622477e-01 6.96831167e-01
-1.24834108e+00 -6.11368656e-01 -3.89588028e-01 -4.83682603e-01
-1.02525091e+00 -7.09716320e-01 3.89476866e-01 -8.19327652e-01
3.37496310e-01 3.81518543e-01 7.66920805e-01 1.18057692e+00
1.26744285e-01 3.97634000e-01 1.37229061e+00 -1.63503304e-01
8.85502174e-02 2.70848006e-01 -4.16887999e-01 8.37434828e-01
4.96882915e-01 2.07439080e-01 -1.90986469e-01 -4.01740491e-01
1.06962919e+00 -2.09509254e-01 -3.96093547e-01 -3.73337269e-01
-1.19916749e+00 9.15546596e-01 3.47410053e-01 -5.73409833e-02
-2.13412702e-01 1.33326687e-02 2.99794469e-02 1.57404795e-01
4.86157894e-01 6.31935596e-01 5.04530892e-02 6.52544051e-02
-1.02075839e+00 5.76133311e-01 5.88924587e-01 7.81040430e-01
1.90995604e-01 -2.39300668e-01 -9.53431204e-02 5.07046700e-01
3.07379305e-01 2.07288057e-01 4.87484604e-01 -1.09491539e+00
1.44644544e-01 4.68206197e-01 2.75977761e-01 -9.12036538e-01
-2.54397422e-01 -2.62094498e-01 -6.78685427e-01 4.96014714e-01
5.80293655e-01 -3.94766480e-01 -1.09612846e+00 1.49488962e+00
3.67063046e-01 -1.26730815e-01 1.19051106e-01 1.00213802e+00
1.01467168e+00 1.63009703e-01 2.04089671e-01 2.00827911e-01
1.34146309e+00 -5.11320651e-01 -3.13960671e-01 -6.60455674e-02
6.04864180e-01 -8.43225121e-01 7.43919730e-01 4.01827157e-01
-1.52779877e+00 -1.39768288e-01 -9.71911371e-01 -3.52814347e-01
-7.97757655e-02 -4.15570736e-02 7.47830689e-01 8.19326162e-01
-1.12382352e+00 5.60492516e-01 -9.92066026e-01 -3.06298465e-01
1.05070698e+00 2.84652859e-01 -5.12629688e-01 -2.16034085e-01
-8.27980697e-01 7.05634058e-01 4.85549569e-02 1.03103094e-01
-8.28410089e-01 -8.16099524e-01 -8.88581693e-01 -3.07581216e-01
-2.16932539e-02 -1.17793286e+00 1.23501277e+00 -9.31301653e-01
-9.62554097e-01 1.37816048e+00 1.77166492e-01 -4.70351994e-01
1.05937839e+00 2.87412941e-01 -2.94656277e-01 1.73982888e-01
8.22844207e-02 1.10904527e+00 7.29021788e-01 -1.04976594e+00
-4.18102592e-01 -3.58205378e-01 -6.22869767e-02 -4.25787922e-03
1.35752484e-01 5.46186231e-02 -2.47873858e-01 -7.74835229e-01
-8.88804197e-02 -1.06517220e+00 -6.15670383e-01 3.55101854e-01
-6.23659074e-01 4.41892952e-01 5.49908102e-01 -9.00749028e-01
6.01031423e-01 -2.17146516e+00 -3.38542908e-01 1.47721007e-01
3.38986218e-01 3.70929450e-01 1.72118153e-02 6.06972389e-02
-1.08559504e-01 4.93884951e-01 -3.54849100e-01 -2.00672701e-01
-1.99948370e-01 -2.13983685e-01 -3.53799760e-01 4.54001665e-01
3.52697045e-01 1.15383828e+00 -8.00908923e-01 -5.69609523e-01
3.99452355e-03 4.14053589e-01 -8.67117584e-01 2.71253437e-01
-1.24500223e-01 1.03114033e+00 -4.45848137e-01 6.53808653e-01
5.27377844e-01 -5.24409294e-01 5.48877604e-02 -2.61175800e-02
7.46221468e-02 2.11356401e-01 -7.28020549e-01 1.73332047e+00
1.90495793e-02 5.48997283e-01 1.09599814e-01 -4.03732806e-01
4.55520421e-01 4.75883663e-01 5.85024595e-01 -1.44111112e-01
3.75857711e-01 2.32280686e-01 3.39288175e-01 -5.95214367e-01
3.35875452e-01 -3.73935431e-01 3.66342925e-02 5.63918710e-01
-2.15611443e-01 -5.99958718e-01 6.93593696e-02 2.38171309e-01
1.18260229e+00 3.71851087e-01 -7.43074641e-02 7.59106651e-02
-1.29225686e-01 4.59964484e-01 4.18331236e-01 5.21853387e-01
-3.04804981e-01 1.25255835e+00 6.13201499e-01 -3.91252190e-01
-1.33797729e+00 -1.05388880e+00 -2.59223610e-01 4.31566119e-01
-3.93442124e-01 -9.58615839e-02 -9.66997445e-01 -6.18561387e-01
-1.59628659e-01 5.05513310e-01 -5.97618341e-01 -2.20864832e-01
-2.58756399e-01 -8.42541695e-01 6.25456274e-01 3.12528461e-01
3.04220647e-01 -1.02856255e+00 -8.29750478e-01 1.06470965e-01
-8.20780620e-02 -1.01405823e+00 -5.98467112e-01 -3.80811393e-01
-6.83514178e-01 -1.14339721e+00 -1.03972065e+00 -6.32930934e-01
9.51570988e-01 -1.63267270e-01 9.99635637e-01 1.25725731e-01
-9.48001206e-01 2.46636853e-01 -1.92477524e-01 -8.74314904e-01
-8.41121316e-01 -4.17359285e-02 -4.31946218e-01 -4.28396165e-02
2.66519119e-03 -4.90844131e-01 -1.14542103e+00 4.24852595e-02
-1.11439025e+00 3.23316574e-01 5.79126596e-01 6.64859116e-01
5.71171820e-01 -4.24955398e-01 4.26651299e-01 -1.38298225e+00
5.79253078e-01 -3.85153204e-01 -6.97478175e-01 -2.39192583e-02
-5.36417127e-01 -9.77119133e-02 1.47872627e-01 -1.69594795e-01
-9.58448112e-01 2.32416362e-01 -2.54101813e-01 -3.69787604e-01
-2.40122810e-01 3.81852031e-01 -2.10515987e-02 -3.19511384e-01
7.73826957e-01 -2.44209796e-01 3.27493817e-01 -1.23081952e-01
2.11467251e-01 7.07500517e-01 6.36528313e-01 -1.91361919e-01
5.94785094e-01 4.94124144e-01 -4.70054429e-03 -5.13280749e-01
-7.76905119e-01 1.84948057e-01 -1.52351469e-01 9.17379558e-02
1.01306093e+00 -9.60589945e-01 -5.47301412e-01 4.22147036e-01
-8.12310457e-01 -2.11865187e-01 -4.98912781e-01 3.64766687e-01
-5.36456466e-01 2.64781803e-01 -5.96200466e-01 -3.36242676e-01
-2.75300473e-01 -1.47786224e+00 1.00872195e+00 1.75329447e-01
-3.74250531e-01 -7.05396473e-01 3.93830873e-02 9.00920093e-01
4.44690406e-01 7.30746269e-01 9.10151184e-01 -5.54531872e-01
-1.03171062e+00 -4.81232584e-01 -1.76759928e-01 3.86143267e-01
4.30596024e-02 1.33378550e-01 -1.03085303e+00 -2.42349207e-01
-8.51934552e-02 -7.11510062e-01 4.32200462e-01 4.91574615e-01
1.36971247e+00 -3.34486008e-01 -2.46369347e-01 8.74429822e-01
1.03110099e+00 1.44501582e-01 8.38413775e-01 -3.08019444e-02
6.40793025e-01 8.09961796e-01 3.36379439e-01 3.28878433e-01
5.26002526e-01 2.53450871e-01 2.94611424e-01 -5.47806621e-01
-1.15158506e-01 -3.69542241e-01 -2.05964088e-01 2.79587477e-01
-1.75544005e-02 -2.23578915e-01 -1.02728069e+00 6.52072132e-01
-1.51216924e+00 -7.00744569e-01 -1.67588353e-01 2.17657328e+00
1.04723203e+00 7.56691094e-04 2.97846682e-02 -4.72470164e-01
5.46991587e-01 -3.23247075e-01 -6.32886350e-01 -1.53195620e-01
7.44966641e-02 3.62726480e-01 6.65552139e-01 2.18916491e-01
-1.07854414e+00 7.78362870e-01 6.86514759e+00 1.90114513e-01
-1.23866725e+00 -4.15623710e-02 1.34862435e+00 -1.02672733e-01
-5.31259477e-01 6.64971471e-02 -2.81409681e-01 3.86677474e-01
9.39753711e-01 -2.46935979e-01 4.64936569e-02 6.45332992e-01
3.24952245e-01 -1.52628809e-01 -1.04476619e+00 7.72036910e-01
-2.06951007e-01 -1.60340166e+00 1.03778243e-01 4.36981291e-01
6.20178163e-01 8.73791948e-02 3.47973943e-01 -3.11309010e-01
7.15336084e-01 -1.53871703e+00 4.42159027e-01 6.32204533e-01
1.07693911e+00 -5.36750674e-01 5.24046481e-01 1.08011439e-01
-2.54344881e-01 3.40924829e-01 3.06464452e-02 3.32992077e-01
3.08818698e-01 3.86869907e-01 -1.19787383e+00 2.16501743e-01
5.17633319e-01 3.98336709e-01 -5.65980971e-01 1.01960278e+00
-3.23296562e-02 4.41452324e-01 5.46011701e-02 3.99619550e-01
4.67812531e-02 -2.17517555e-01 3.45836937e-01 6.91662610e-01
3.41964662e-01 3.00692528e-01 -1.18138798e-01 1.07821381e+00
-2.74629563e-01 7.41472691e-02 -6.78246975e-01 -4.60919499e-01
4.36202101e-02 1.40908241e+00 -7.86609888e-01 -9.83786508e-02
-3.71333152e-01 8.46556365e-01 -7.31598400e-03 3.80639248e-02
-6.61782742e-01 -1.37357473e-01 4.87589151e-01 5.68550050e-01
3.50137288e-03 9.65538844e-02 -4.16473120e-01 -1.16721678e+00
-1.33092389e-01 -1.27829850e+00 2.48342767e-01 -1.10540521e+00
-1.21796691e+00 6.79761171e-01 -2.04020694e-01 -1.12965405e+00
-5.33209085e-01 -4.93845493e-01 -2.86270678e-01 1.15638041e+00
-1.07487774e+00 -1.35583901e+00 -4.03684795e-01 5.06089330e-01
-1.47422165e-01 -7.97948390e-02 8.95992339e-01 2.66058683e-01
-2.80688077e-01 5.83617151e-01 -3.53616238e-01 3.34071189e-01
9.55749869e-01 -1.02902973e+00 6.87130928e-01 7.26366699e-01
8.11416283e-02 7.55913794e-01 6.25524998e-01 -6.73614502e-01
-9.82441127e-01 -1.21096647e+00 7.51750410e-01 -7.06019163e-01
3.80607605e-01 -2.21811548e-01 -6.78260148e-01 9.61303115e-01
2.05968156e-01 2.14319795e-01 8.95213842e-01 -5.22377074e-01
-1.14995427e-01 1.08210668e-01 -1.62785864e+00 7.69938409e-01
7.45057166e-01 -3.58593196e-01 -2.49219194e-01 5.81204057e-01
5.64235449e-01 -6.62185609e-01 -1.06038356e+00 4.07347411e-01
4.52062637e-01 -9.51251745e-01 7.06426561e-01 -6.71332777e-01
8.15821350e-01 -2.63727576e-01 2.52753139e-01 -1.23197448e+00
1.46999270e-01 -9.41121936e-01 6.15768254e-01 1.04999125e+00
7.54131734e-01 -6.80036843e-01 1.00349545e+00 1.19374955e+00
5.16465046e-02 -5.70629776e-01 -5.55844426e-01 -1.01391219e-01
3.05185199e-01 -2.48317078e-01 5.60659945e-01 1.07073522e+00
-2.79137373e-01 9.72017795e-02 -1.97481558e-01 -4.47782092e-02
4.67918217e-01 8.97919536e-02 8.65873694e-01 -8.35989058e-01
-4.45601434e-01 -2.70024329e-01 -5.21861315e-01 -3.83496433e-01
-3.35246503e-01 -1.03404045e+00 -6.93341494e-02 -1.48236692e+00
3.12488765e-01 -6.67938232e-01 1.49839684e-01 5.95185578e-01
-9.71581787e-02 7.10873902e-01 -3.83997820e-02 2.82407463e-01
7.05897883e-02 9.99829397e-02 1.57798362e+00 -6.79913387e-02
5.90639971e-02 -2.53598560e-02 -1.25145769e+00 6.08160734e-01
7.00097799e-01 -4.28350776e-01 -5.92271090e-01 -3.38490307e-01
1.44696236e-01 1.67486742e-01 7.54507303e-01 -1.01329279e+00
-2.36525238e-02 -3.34279910e-02 5.45049787e-01 -2.09921345e-01
1.72411367e-01 -6.15975976e-01 5.98409712e-01 3.18644851e-01
-4.47931498e-01 -1.22547597e-01 9.19243246e-02 7.85340220e-02
-1.75830632e-01 -1.18946750e-02 9.37780082e-01 -4.77148354e-01
5.69175184e-02 5.33097923e-01 -2.25613127e-03 3.28330308e-01
1.10963047e+00 -3.56615707e-02 -3.02105814e-01 -6.43673360e-01
-9.53178823e-01 -3.89595027e-03 9.32915568e-01 1.49960265e-01
4.78347272e-01 -9.33464706e-01 -9.30534959e-01 2.75606275e-01
1.02984771e-01 3.35112184e-01 2.99880952e-01 7.97309518e-01
-9.09529984e-01 5.41054867e-02 -2.07861915e-01 -5.11155427e-01
-1.06195354e+00 5.54614425e-01 4.82007951e-01 -1.00776166e-01
-6.52611315e-01 8.02083015e-01 3.17879051e-01 -3.55942041e-01
-1.83134139e-01 -2.61587560e-01 3.69797587e-01 -2.61251360e-01
4.24377263e-01 -6.81804717e-02 6.95517361e-02 -6.05038643e-01
5.89918196e-02 -3.61415558e-02 -1.39409021e-01 -4.64966357e-01
1.35419619e+00 2.21467078e-01 -3.99758182e-02 -6.47413719e-04
1.00571358e+00 3.53443734e-02 -1.46224904e+00 3.63718808e-01
-4.14402544e-01 -4.74779129e-01 -3.85624543e-02 -1.01536846e+00
-1.18514717e+00 7.08201587e-01 5.45500755e-01 7.57846311e-02
1.03222930e+00 -6.80810064e-02 8.22715998e-01 -1.72234833e-01
3.88258517e-01 -6.66189790e-01 -2.74773180e-01 -1.73814476e-01
9.05282617e-01 -1.19190204e+00 1.54018864e-01 -4.62302685e-01
-7.42856681e-01 6.37104392e-01 3.99739444e-01 -1.17747970e-02
5.96358716e-01 5.32740712e-01 5.51384211e-01 -2.13173419e-01
-5.56886196e-01 2.14192152e-01 3.33351016e-01 6.89619422e-01
6.95887983e-01 -7.17110839e-03 -2.96228439e-01 4.61109847e-01
-5.97464561e-01 4.24001187e-01 7.82487154e-01 1.09673762e+00
8.06732550e-02 -1.41431379e+00 -1.65557846e-01 6.47301078e-01
-9.42565560e-01 -2.93556213e-01 -5.48316777e-01 7.12356567e-01
2.00873271e-01 7.40773797e-01 9.25152972e-02 7.64074400e-02
1.88459992e-01 -1.12715624e-01 5.46123564e-01 -7.61367559e-01
-5.49618959e-01 2.07313560e-02 2.01964781e-01 -5.03453672e-01
-2.19516203e-01 -8.41314375e-01 -9.51239944e-01 -1.52221069e-01
1.96722537e-01 -2.46657178e-01 7.75375545e-01 6.36784554e-01
7.20474243e-01 3.07927281e-01 3.53916436e-01 -4.80824709e-01
-2.50525445e-01 -6.66114867e-01 -2.88359821e-01 5.38992047e-01
3.36206466e-01 -5.28607704e-02 -4.83789714e-04 4.19516534e-01] | [14.302605628967285, -1.862199306488037] |
29a083d4-97fb-4f76-a07e-134b63739520 | deep-learning-approaches-in-food-recognition | 2004.03357 | null | https://arxiv.org/abs/2004.03357v2 | https://arxiv.org/pdf/2004.03357v2.pdf | Deep learning approaches in food recognition | Automatic image-based food recognition is a particularly challenging task. Traditional image analysis approaches have achieved low classification accuracy in the past, whereas deep learning approaches enabled the identification of food types and their ingredients. The contents of food dishes are typically deformable objects, usually including complex semantics, which makes the task of defining their structure very difficult. Deep learning methods have already shown very promising results in such challenges, so this chapter focuses on the presentation of some popular approaches and techniques applied in image-based food recognition. The three main lines of solutions, namely the design from scratch, the transfer learning and the platform-based approaches, are outlined, particularly for the task at hand, and are tested and compared to reveal the inherent strengths and weaknesses. The chapter is complemented with basic background material, a section devoted to the relevant datasets that are crucial in light of the empirical approaches adopted, and some concluding remarks that underline the future directions. | ['Stella Markantonatou', 'George Pavlidis', 'Chairi Kiourt'] | 2020-04-04 | null | null | null | null | ['food-recognition'] | ['computer-vision'] | [ 2.08076596e-01 -2.25179732e-01 -5.10256886e-01 -4.17942405e-01
-2.37023145e-01 -7.88396537e-01 4.46838677e-01 7.22085297e-01
-3.59578133e-01 2.25286305e-01 8.40442404e-02 1.22510396e-01
-2.38133878e-01 -8.46367121e-01 -7.18491852e-01 -1.05743659e+00
-3.61492068e-01 2.50986844e-01 -2.37173632e-01 -2.35740870e-01
1.60541743e-01 1.37790114e-01 -1.92679667e+00 6.56131208e-01
4.81612086e-01 1.29719400e+00 4.36764270e-01 6.05061531e-01
-2.76856005e-01 8.01276147e-01 -4.12985563e-01 -3.06879073e-01
2.81563967e-01 -3.32933426e-01 -4.73725885e-01 3.56002897e-01
4.56094712e-01 -2.30105802e-01 1.71975270e-02 1.11011612e+00
2.49307409e-01 -1.17925340e-02 8.44807804e-01 -8.15463006e-01
-9.14772511e-01 7.82029569e-01 -6.38812035e-02 1.88025191e-01
5.22465229e-01 7.99925029e-02 6.67390704e-01 -7.04637825e-01
3.11371237e-01 9.72831368e-01 9.31277454e-01 4.62145001e-01
-1.06722927e+00 -2.17252135e-01 3.27330977e-01 3.39781225e-01
-1.09805691e+00 -2.20269442e-01 8.17377746e-01 -7.15204120e-01
1.13622642e+00 2.52964526e-01 1.03835440e+00 1.00907815e+00
4.83238623e-02 1.11558664e+00 1.09414327e+00 -6.54945076e-01
2.43045270e-01 2.95176804e-01 4.63239431e-01 4.57920104e-01
6.14962578e-01 1.31415084e-01 -1.74677186e-02 2.27291718e-01
5.56831300e-01 -5.02065942e-02 1.32795155e-01 -8.51478219e-01
-1.07028627e+00 1.27254581e+00 6.17034674e-01 9.75847781e-01
-4.83889014e-01 -4.29361314e-01 6.42419159e-01 2.35982835e-01
5.21984160e-01 1.94799900e-01 -7.48590112e-01 4.47952837e-01
-7.44794071e-01 2.72871047e-01 8.23881388e-01 6.80249691e-01
3.07470828e-01 1.75870463e-01 1.62266642e-01 7.47217715e-01
4.74643171e-01 6.04683101e-01 4.44967210e-01 -3.40937346e-01
5.27086258e-02 5.92617571e-01 1.74333528e-01 -1.33418632e+00
-7.09142208e-01 -8.24957937e-02 -7.45271802e-01 2.14186430e-01
5.90182900e-01 7.90297538e-02 -8.49340439e-01 1.20130253e+00
4.64221656e-01 -3.94549161e-01 2.38984421e-01 9.10808742e-01
1.52074587e+00 5.04883647e-01 6.41405344e-01 3.04706069e-03
1.64213288e+00 -1.06532955e+00 -7.86263168e-01 -1.22532114e-01
6.08033776e-01 -9.41439390e-01 8.31326723e-01 4.47980851e-01
-1.03681540e+00 -8.16694260e-01 -1.10398078e+00 -3.67734507e-02
-9.37748432e-01 2.69424230e-01 8.77857804e-01 9.15767729e-01
-6.42431140e-01 6.66789651e-01 -8.99156868e-01 -7.00908840e-01
2.28495687e-01 5.42921841e-01 -3.03273410e-01 -6.60255477e-02
-7.85365343e-01 8.27388704e-01 8.78169715e-01 2.68057466e-01
-6.03460729e-01 -6.51764929e-01 -1.22975254e+00 -3.37830275e-01
1.98028818e-01 -5.08412361e-01 1.16831279e+00 -1.46104896e+00
-1.59463632e+00 1.29869032e+00 5.49587965e-01 -4.38018918e-01
3.67652982e-01 -1.67760521e-01 -4.34425145e-01 1.50123700e-01
-1.44422352e-01 7.02988029e-01 6.77841127e-01 -1.15640140e+00
-6.52647197e-01 -3.49666893e-01 3.02683890e-01 7.90178105e-02
-1.17090739e-01 2.97670305e-01 2.11020648e-01 -7.22567499e-01
-2.29525492e-01 -7.01717913e-01 -4.39089276e-02 -4.43097064e-03
-1.75837770e-01 -4.84810531e-01 3.55984271e-01 -5.93864143e-01
9.89755392e-01 -2.52855253e+00 -3.65252756e-02 -1.26487523e-01
-2.65846610e-01 6.05606675e-01 -1.67130783e-01 4.08131003e-01
-2.22198904e-01 -2.25087732e-01 1.88795775e-02 1.64863437e-01
1.98423862e-01 2.85515904e-01 2.43874416e-01 5.87260902e-01
3.74830887e-02 1.03943479e+00 -1.07985425e+00 -4.01842922e-01
9.69035506e-01 6.44879878e-01 -7.32849315e-02 -4.70546409e-02
-4.61077005e-01 3.05159867e-01 -4.87443864e-01 8.19289684e-01
7.10225284e-01 -1.85858086e-01 2.40035623e-01 -7.11972952e-01
-4.28068519e-01 2.46325508e-02 -1.25632453e+00 1.56044352e+00
-4.36953269e-02 3.59086283e-02 2.67511129e-01 -1.38938177e+00
8.24047148e-01 2.93748587e-01 7.13892400e-01 -7.61577904e-01
4.17167544e-01 3.96303803e-01 -9.88476165e-03 -8.63814175e-01
2.83315897e-01 -5.37228361e-02 -8.89742002e-03 1.96280833e-02
1.50108755e-01 1.40476927e-01 5.82583010e-01 -6.27457917e-01
3.15342993e-01 5.02388179e-01 7.96748340e-01 -7.16512859e-01
7.40898550e-01 6.02419913e-01 8.51653591e-02 2.93461025e-01
-3.24072391e-01 2.02693328e-01 -2.09650859e-01 -1.19302762e+00
-1.00614130e+00 -8.39253306e-01 -2.62449086e-01 1.07919407e+00
1.17716879e-01 1.02445101e-02 -9.88948226e-01 -7.24832952e-01
2.13872045e-01 2.86566943e-01 -1.02894723e+00 4.44045328e-02
-6.77508712e-01 -1.01355159e+00 -9.88348648e-02 5.95564127e-01
5.80513179e-01 -1.34672844e+00 -8.09734404e-01 3.59100044e-01
-2.00555131e-01 -7.43752897e-01 -5.40368259e-02 2.89164722e-01
-9.78206158e-01 -1.41893625e+00 -8.86996686e-01 -1.23374438e+00
5.45233548e-01 5.83618164e-01 1.40980172e+00 1.81878567e-01
-4.80597138e-01 4.84520912e-01 -6.86443210e-01 -6.63642228e-01
-6.03984118e-01 1.87338274e-02 -1.83947369e-01 -2.36567184e-01
1.03937781e+00 3.31321061e-02 -9.84221578e-01 1.98926434e-01
-1.04656065e+00 -2.42336020e-01 4.59628642e-01 7.03427076e-01
8.52167189e-01 -8.39481130e-02 3.99618566e-01 -5.47787249e-01
5.01313731e-02 -3.33408773e-01 -3.33234042e-01 3.40936154e-01
-2.26499438e-01 -2.74812728e-01 5.09496570e-01 -6.07589364e-01
-7.39937723e-01 4.14808631e-01 -3.53211403e-01 9.82421860e-02
-7.49974549e-01 4.01356906e-01 -2.15126693e-01 -1.46377608e-01
8.38630080e-01 -4.42624241e-02 1.79481551e-01 -7.44507194e-01
2.61416554e-01 4.43492591e-01 2.52168953e-01 -3.81846756e-01
1.89338088e-01 4.19272929e-01 -3.00586343e-01 -1.01944172e+00
-1.07103848e+00 -5.98570108e-01 -8.79772544e-01 -2.00737774e-01
1.17246294e+00 -8.18002284e-01 -4.80269760e-01 6.33341253e-01
-5.08701742e-01 -5.89700162e-01 -5.98117471e-01 5.68123341e-01
-7.17584908e-01 2.97177315e-01 -8.15331578e-01 -4.26795214e-01
-5.98804176e-01 -1.06008732e+00 8.45908761e-01 2.59605050e-01
-2.30642378e-01 -1.46922839e+00 1.51019469e-01 4.00038272e-01
3.32856327e-01 7.14836955e-01 9.69806314e-01 -6.83338642e-01
4.81599420e-02 -1.32310122e-01 1.15589276e-02 2.53274560e-01
5.77471972e-01 2.57598842e-03 -8.79520535e-01 -3.90787035e-01
5.52256033e-02 -2.83730447e-01 6.82425618e-01 8.37137938e-01
6.45912230e-01 2.09831148e-02 -3.12042594e-01 2.04971224e-01
1.66524994e+00 3.87802154e-01 4.41877753e-01 6.94613874e-01
3.84629607e-01 7.83917248e-01 7.98446417e-01 1.33466065e-01
1.83211148e-01 8.08656454e-01 5.74489176e-01 -3.73324156e-01
-5.61474025e-01 1.77943751e-01 3.74987930e-01 7.54394174e-01
-1.15466148e-01 -1.37932897e-01 -2.17454419e-01 4.04411972e-01
-1.64749217e+00 -8.55782926e-01 -4.56423372e-01 1.95066822e+00
4.15266633e-01 -3.43653113e-01 6.83672547e-01 5.37688613e-01
5.88537216e-01 -2.33898699e-01 -4.94232148e-01 -5.45859158e-01
-2.77905613e-01 2.78849930e-01 2.66344637e-01 1.44678459e-01
-1.83286047e+00 6.10594869e-01 7.38494825e+00 3.94533336e-01
-1.09009302e+00 -3.77580430e-03 5.22720814e-01 3.00950646e-01
2.76974231e-01 -9.70304251e-01 -6.25031173e-01 4.65927571e-01
8.80183041e-01 3.94872248e-01 3.15250427e-01 8.16544056e-01
-2.29591201e-03 -1.69285238e-01 -1.10472155e+00 7.78606713e-01
3.01677614e-01 -9.80268478e-01 -7.82177821e-02 -2.01631755e-01
5.64066887e-01 1.76487714e-02 2.11427435e-01 1.54991835e-01
1.78505823e-01 -6.02731109e-01 9.92345333e-01 2.00886935e-01
3.62957180e-01 -5.82261622e-01 7.35352159e-01 4.79766773e-03
-1.45504856e+00 -3.34051251e-01 -6.89030945e-01 -1.18839860e-01
-1.70262784e-01 6.52332485e-01 -6.99041113e-02 7.97861218e-01
1.09377253e+00 8.74466300e-01 -2.89571196e-01 1.10368478e+00
-9.55764726e-02 3.14376652e-01 -5.01378536e-01 -2.41843343e-01
4.01025802e-01 -3.51877272e-01 5.35969995e-02 1.48257792e+00
1.66579753e-01 -8.61148536e-02 5.01887679e-01 5.01656830e-01
5.92940688e-01 7.11647630e-01 -6.01821840e-01 -1.70601457e-01
-3.81939143e-01 1.47545433e+00 -1.25232983e+00 -3.35211933e-01
-7.45755255e-01 6.42233253e-01 -9.85994637e-02 -1.89766973e-01
-5.58929622e-01 3.77772637e-02 6.39636040e-01 -4.42725569e-02
5.91015458e-01 -1.31276429e-01 -1.72379743e-02 -9.83175933e-01
-1.66773587e-01 -9.91404772e-01 7.60427177e-01 -2.88662493e-01
-1.52623844e+00 3.95139873e-01 2.62806803e-01 -1.03639364e+00
-1.14833422e-01 -1.18040800e+00 -3.91646512e-02 2.77285874e-01
-1.49278140e+00 -1.42125249e+00 -3.01196426e-01 4.81551915e-01
6.75934911e-01 1.08091623e-01 1.30146742e+00 5.78186214e-01
-4.84778911e-01 3.37220550e-01 4.32438582e-01 1.70335978e-01
2.98570395e-01 -1.12101221e+00 1.58926956e-02 1.68296456e-01
3.75931978e-01 1.47488475e-01 8.53530288e-01 -4.80473191e-01
-1.46488607e+00 -8.78232896e-01 7.19242692e-01 -2.50855118e-01
5.68602860e-01 -3.01011711e-01 -7.40584850e-01 3.75634402e-01
3.39382678e-01 -7.65671907e-03 1.07803380e+00 -1.15067653e-01
-1.55065984e-01 -5.90526275e-02 -1.47158957e+00 1.57366022e-01
7.35771120e-01 -2.96080881e-03 -5.24275601e-01 6.75455987e-01
5.71546517e-02 -3.08858603e-01 -1.04771245e+00 4.12167460e-01
7.26372004e-01 -9.09988761e-01 1.28423381e+00 -5.21027863e-01
1.99804366e-01 -9.50744599e-02 -1.51822388e-01 -1.13876736e+00
-8.11542451e-01 6.11254834e-02 -1.08608902e-01 9.71512675e-01
1.50883064e-01 -3.51609677e-01 9.43992019e-01 1.69268191e-01
-2.24435210e-01 -3.79188031e-01 -4.80368227e-01 -7.87327290e-01
2.53669322e-01 7.88014233e-02 5.44709265e-01 9.42032039e-01
-7.07763731e-02 9.23585445e-02 7.75384083e-02 -9.80404168e-02
4.21056688e-01 3.69141251e-01 4.17828709e-01 -1.18827832e+00
-4.15521190e-02 -5.91152430e-01 -5.97762883e-01 -6.03345394e-01
-1.04504131e-01 -8.01789165e-01 4.81487811e-02 -1.65281820e+00
2.92300940e-01 -2.86118388e-01 -5.15623271e-01 4.10802454e-01
-9.39152483e-03 4.98261690e-01 2.03263476e-01 2.05739345e-02
-3.19502473e-01 -1.63492471e-01 1.28277624e+00 -6.46449089e-01
-1.67606413e-01 -2.51379833e-02 -5.88981450e-01 8.07216525e-01
1.35795975e+00 -2.16263592e-01 -3.58038217e-01 -4.98231858e-01
1.79161608e-01 -7.48919606e-01 2.17307791e-01 -1.03882170e+00
-3.99233907e-01 -4.00826298e-02 5.47398388e-01 -4.85627383e-01
1.33909211e-01 -1.23971677e+00 5.00749886e-01 8.87014925e-01
-1.69977665e-01 1.22760544e-02 3.58864039e-01 2.83673912e-01
-1.32654130e-01 -5.46319067e-01 8.57584536e-01 -5.44870853e-01
-9.80333388e-01 6.84459880e-02 -4.42931563e-01 -5.66639900e-01
1.37776434e+00 -4.64318991e-01 1.69249550e-01 3.01182657e-01
-1.15775430e+00 -2.38639191e-01 2.95735747e-01 4.30028439e-01
1.07949145e-01 -1.33228731e+00 -6.43152714e-01 2.78766811e-01
2.68294424e-01 -3.23338151e-01 4.28874105e-01 7.23338783e-01
-7.28587508e-01 6.51289225e-01 -5.09155691e-01 -7.22661316e-01
-1.46383750e+00 1.38663065e+00 3.74073237e-01 -2.68688887e-01
-9.01414275e-01 4.41392064e-01 4.55933183e-01 -1.51400164e-01
1.77219763e-01 -8.03440273e-01 -8.98249626e-01 1.19462967e-01
8.48262012e-01 2.03881234e-01 4.90372211e-01 -8.16156387e-01
-1.49142057e-01 8.27577174e-01 3.54077704e-02 8.36868048e-01
1.49080491e+00 -1.60463646e-01 2.10472003e-01 3.70129943e-01
8.41477752e-01 -7.00920045e-01 -1.16642535e+00 -5.49538024e-02
2.72150300e-02 -1.07980184e-01 -2.74306778e-02 -9.92105782e-01
-1.27812791e+00 9.63978529e-01 1.15138996e+00 7.00076044e-01
1.17106616e+00 2.88279373e-02 6.91081643e-01 1.11643538e-01
5.09483099e-01 -9.59596217e-01 -1.98466629e-02 2.09754616e-01
7.01492786e-01 -1.37870479e+00 -4.84030060e-02 -8.06055188e-01
-7.83880204e-02 1.39711380e+00 5.91948889e-02 -3.96776140e-01
8.07539582e-01 4.33164835e-01 3.44258875e-01 -3.20933163e-01
6.26309812e-02 -3.18039298e-01 2.94328809e-01 1.03433359e+00
6.66847229e-01 3.58826429e-01 -3.29451948e-01 4.41712588e-01
-4.92995903e-02 1.60062537e-01 -2.36673340e-01 1.11247671e+00
-5.08103967e-01 -1.30868042e+00 -5.62471628e-01 1.43348187e-01
-7.18751967e-01 2.39098698e-01 -2.47172371e-01 1.03746021e+00
5.27212799e-01 1.09392321e+00 1.47286102e-01 -8.35480466e-02
5.38798690e-01 -3.67239676e-02 1.06915653e+00 -3.69114906e-01
-1.10498714e+00 1.40997514e-01 7.26834387e-02 -2.29148194e-01
-1.25577986e+00 -8.72841895e-01 -9.07782197e-01 -3.15725952e-01
-3.35950464e-01 4.05579470e-02 8.18140984e-01 8.12763274e-01
-1.19030468e-01 3.02313000e-01 3.31213564e-01 -1.21851635e+00
-3.56680423e-01 -9.06409144e-01 -6.54121399e-01 6.96651518e-01
1.34017363e-01 -8.15626919e-01 -3.86899002e-02 5.30870259e-01] | [11.562455177307129, 4.3987603187561035] |
b1a6cce8-ad00-4df4-a2bd-46d8f9fe927b | aspect-sentiment-classification-with-aspect | 2010.02696 | null | https://arxiv.org/abs/2010.02696v2 | https://arxiv.org/pdf/2010.02696v2.pdf | Aspect Based Sentiment Analysis with Aspect-Specific Opinion Spans | Aspect based sentiment analysis, predicting sentiment polarity of given aspects, has drawn extensive attention. Previous attention-based models emphasize using aspect semantics to help extract opinion features for classification. However, these works are either not able to capture opinion spans as a whole, or not able to capture variable-length opinion spans. In this paper, we present a neat and effective structured attention model by aggregating multiple linear-chain CRFs. Such a design allows the model to extract aspect-specific opinion spans and then evaluate sentiment polarity by exploiting the extracted opinion features. The experimental results on four datasets demonstrate the effectiveness of the proposed model, and our analysis demonstrates that our model can capture aspect-specific opinion spans. | ['Fei Huang', 'Wei Lu', 'Lidong Bing', 'Lu Xu'] | 2020-10-06 | aspect-sentiment-classification-with-aspect-1 | https://aclanthology.org/2020.emnlp-main.288 | https://aclanthology.org/2020.emnlp-main.288.pdf | emnlp-2020-11 | ['extract-aspect'] | ['natural-language-processing'] | [-2.75894776e-02 2.12065428e-01 -7.14036644e-01 -7.09250987e-01
-6.26276612e-01 -6.93408549e-01 4.84042019e-01 2.49248102e-01
-9.38952863e-02 6.65914059e-01 6.14218771e-01 -3.04134637e-01
2.59706020e-01 -9.27628636e-01 -2.27280498e-01 -3.66039395e-01
1.34077922e-01 3.65811914e-01 -8.24404508e-02 -2.82357872e-01
5.03223717e-01 -2.91129146e-02 -1.34749293e+00 2.49281421e-01
1.03952980e+00 1.22788668e+00 -3.71967822e-01 4.55208540e-01
-5.86265624e-01 7.03466773e-01 -7.64310002e-01 -7.20542967e-01
-1.27089813e-01 -2.60068089e-01 -6.71692014e-01 1.78227559e-01
1.10743321e-01 6.12521432e-02 4.09111232e-01 7.95813978e-01
1.78554609e-01 -1.60570860e-01 8.05623293e-01 -1.13448775e+00
-1.04322660e+00 7.52567291e-01 -6.89358175e-01 -3.92268561e-02
4.13261205e-01 -5.96024506e-02 1.75054789e+00 -7.75586069e-01
2.96990544e-01 9.58167374e-01 6.95674598e-01 2.02505350e-01
-7.92781055e-01 -4.62983370e-01 7.77789354e-01 9.57643017e-02
-5.42412102e-01 7.59176612e-02 1.29136896e+00 -3.16631705e-01
1.24487174e+00 1.05141416e-01 1.38056147e+00 9.07855511e-01
5.29545248e-01 1.14656210e+00 1.48343873e+00 -2.18634680e-01
3.03046137e-01 2.77376890e-01 9.75315869e-01 4.22762066e-01
3.45228702e-01 -3.32830966e-01 -6.03047609e-01 -1.47119030e-01
1.53253721e-02 8.49145651e-02 -2.31066663e-02 -6.97616935e-02
-6.96857989e-01 1.10759389e+00 3.27847183e-01 2.27414861e-01
-4.69821155e-01 -2.41761938e-01 3.67064327e-01 2.52063483e-01
1.19375682e+00 6.49112940e-01 -9.70945954e-01 -2.11689875e-01
-7.54256785e-01 4.62735891e-02 1.08252990e+00 1.03363693e+00
9.64184344e-01 4.48087119e-02 -3.16571742e-01 6.42682195e-01
3.71452808e-01 6.85864806e-01 4.82984483e-01 -3.75297755e-01
1.46989435e-01 1.22051454e+00 -9.85198170e-02 -8.79841626e-01
-6.28646731e-01 -4.44449633e-01 -6.13328218e-01 -7.51159936e-02
-8.42501819e-02 -5.12920201e-01 -1.12067699e+00 1.53971922e+00
1.35837451e-01 -2.01626450e-01 1.10587798e-01 5.86384535e-01
8.21738303e-01 3.82850915e-01 2.94998288e-01 -3.40669483e-01
1.80761540e+00 -1.17906725e+00 -8.27946424e-01 -5.36161721e-01
3.08912277e-01 -7.02517629e-01 1.23027098e+00 2.32614413e-01
-8.54385376e-01 -3.59153092e-01 -1.03953755e+00 4.78043109e-02
-6.49635017e-01 -1.25581101e-01 1.28869057e+00 7.15129614e-01
-8.04905713e-01 -2.67597549e-02 -6.03852928e-01 -2.11142357e-02
4.57288593e-01 4.38128024e-01 -9.00800154e-02 2.27548316e-01
-1.34937990e+00 6.50885344e-01 1.11572659e-02 -3.06275743e-03
-8.72840881e-02 -4.75284457e-01 -1.14517653e+00 2.18053818e-01
2.43975714e-01 -1.06659698e+00 1.04963577e+00 -1.34005153e+00
-1.36910570e+00 5.02698600e-01 -7.36257493e-01 -3.36863369e-01
-2.96965629e-01 -4.40474927e-01 -5.17148137e-01 2.25685760e-02
1.56936884e-01 4.52573597e-01 6.85742795e-01 -1.11739409e+00
-7.03821898e-01 -5.88674366e-01 6.54323936e-01 4.17616844e-01
-7.10726023e-01 -5.65739237e-02 -5.20573556e-01 -6.29392982e-01
-2.35968977e-01 -8.39185119e-01 -5.91288269e-01 -7.87530720e-01
-4.25541073e-01 -4.40640897e-01 7.33386040e-01 -1.37119055e-01
1.35311186e+00 -1.60903215e+00 -2.62109220e-01 -2.17914232e-03
1.47807837e-01 1.78336993e-01 -1.85451359e-01 3.72297585e-01
2.44299650e-01 3.55599821e-01 -3.91701549e-01 -4.39106256e-01
1.11543186e-01 7.87908509e-02 -5.22706270e-01 -9.54698324e-02
5.49008965e-01 1.29624093e+00 -6.39882743e-01 -4.75722611e-01
-1.45542890e-01 5.60105205e-01 -7.09251821e-01 2.63608724e-01
-5.42113662e-01 1.01070188e-01 -8.30833375e-01 9.39730525e-01
6.83497071e-01 -3.53682458e-01 1.22523502e-01 -1.84732959e-01
6.38389289e-02 4.73323673e-01 -4.76427883e-01 1.18799734e+00
-7.79351890e-01 4.54844981e-01 -2.64284164e-01 -7.18141496e-01
9.45329726e-01 1.75338805e-01 4.89634424e-01 -5.70524156e-01
2.20911011e-01 -6.38052225e-02 -2.75564998e-01 -3.16027552e-01
9.50716197e-01 -5.00860453e-01 -5.27086616e-01 7.26154566e-01
3.14882919e-02 -3.77962232e-01 5.49722612e-01 4.60590050e-02
6.33235276e-01 9.20518488e-02 6.35405183e-01 -1.67198420e-01
7.28551745e-01 7.15244487e-02 6.98184550e-01 3.77227634e-01
-5.51902540e-02 6.17449880e-01 8.86002481e-01 -4.68004674e-01
-5.74196577e-01 -5.80160439e-01 -1.27457127e-01 1.03958941e+00
2.31392130e-01 -5.81158519e-01 -5.57682931e-01 -1.38137341e+00
-2.06524134e-01 6.20885611e-01 -6.81708872e-01 6.81116953e-02
-4.12111074e-01 -1.02266741e+00 -4.63628434e-02 7.29971647e-01
3.42041016e-01 -1.54153359e+00 -3.36279064e-01 7.07264766e-02
-2.12697104e-01 -9.38711643e-01 -4.36248511e-01 3.24274778e-01
-7.86846578e-01 -1.00873780e+00 -4.58941311e-01 -8.26219320e-01
6.98991477e-01 1.54567987e-01 1.50979340e+00 -5.77383861e-02
3.92496973e-01 4.11353737e-01 -5.62516689e-01 -7.57326424e-01
3.68513376e-01 5.45198798e-01 -2.35108674e-01 1.27517968e-01
1.07025480e+00 -6.91209674e-01 -7.05345392e-01 -1.11429598e-02
-7.95171022e-01 -4.12216455e-01 8.36102247e-01 8.06267083e-01
6.83848381e-01 -3.68739635e-01 9.78046119e-01 -1.52222657e+00
1.03935301e+00 -6.29428566e-01 -1.94900095e-01 1.25070676e-01
-1.21190691e+00 4.66253944e-02 7.96307683e-01 -2.75233835e-01
-1.01629221e+00 -3.85945976e-01 -5.15638888e-01 2.12692320e-01
-1.91184148e-01 1.02279723e+00 -3.91563773e-01 4.51409429e-01
-1.11352131e-01 2.26892442e-01 -3.91734213e-01 -1.88572437e-01
4.98409331e-01 7.69193351e-01 -2.13383466e-01 -3.65130335e-01
4.15861905e-01 5.06758928e-01 -2.91835010e-01 -5.74104190e-01
-1.52331531e+00 -5.95603287e-01 -4.33870465e-01 1.68251339e-02
8.00476968e-01 -9.79041815e-01 -4.49777871e-01 4.33187366e-01
-1.15964377e+00 2.55012363e-01 -3.87526810e-01 4.05258894e-01
-4.71142858e-01 2.60993004e-01 -5.81805527e-01 -1.03208065e+00
-8.90155137e-01 -1.15904665e+00 1.33150923e+00 4.91570622e-01
-5.04155755e-01 -1.37891757e+00 3.83134723e-01 3.82094115e-01
5.29107094e-01 3.52988057e-02 8.44875455e-01 -7.81470954e-01
-4.04699206e-01 -3.89102072e-01 -1.07142024e-01 2.32167974e-01
2.30187654e-01 -1.52262613e-01 -1.05654204e+00 1.77484274e-01
1.98071718e-01 -3.42845112e-01 1.24296391e+00 5.03289580e-01
8.26517224e-01 -4.43719715e-01 -1.64444923e-01 5.63455582e-01
1.36490452e+00 -2.59041134e-02 5.81403136e-01 5.16296089e-01
6.58934534e-01 5.80955207e-01 8.57888222e-01 2.07908645e-01
7.62539685e-01 3.23204249e-01 3.59105319e-01 -1.75239950e-01
2.17376515e-01 -4.79108952e-02 4.35099959e-01 1.23227084e+00
4.60661277e-02 -4.04285163e-01 -3.94316524e-01 1.00359893e+00
-1.64495647e+00 -8.70743394e-01 -2.00290442e-01 1.36479926e+00
7.49256372e-01 3.82520735e-01 -5.65547943e-02 -5.73123395e-02
2.58139133e-01 8.10971737e-01 -5.62562108e-01 -9.34625030e-01
-2.84217983e-01 3.55750799e-01 2.35744659e-02 5.63509583e-01
-1.20918190e+00 1.05852830e+00 6.63848782e+00 5.67636967e-01
-1.07795823e+00 -5.80181926e-02 6.07183158e-01 4.92804162e-02
-1.04509604e+00 3.48097324e-01 -8.38991940e-01 3.51562649e-01
6.54374599e-01 -3.22771281e-01 -2.72067636e-01 1.13606071e+00
-3.16930637e-02 -4.74312119e-02 -5.27792990e-01 2.81123638e-01
4.28659648e-01 -8.63075614e-01 3.14525872e-01 3.00959628e-02
9.78161395e-01 -1.81754038e-01 2.86606818e-01 5.16554713e-01
3.56285751e-01 -8.87337148e-01 4.79213923e-01 6.63487792e-01
3.80114347e-01 -9.72058117e-01 1.22284436e+00 -6.35239556e-02
-1.48020804e+00 3.71885486e-02 -3.16904247e-01 -4.39044684e-01
6.74809635e-01 1.12098169e+00 -5.05441904e-01 5.78188956e-01
4.26282108e-01 1.12322736e+00 -6.06350064e-01 6.71678483e-01
-7.79620886e-01 9.50747013e-01 4.11412530e-02 -5.54891706e-01
2.82191277e-01 -3.98403227e-01 4.49690968e-01 1.28658116e+00
1.94663316e-01 -1.64063945e-01 1.66735083e-01 3.93310368e-01
-6.23736121e-02 4.28481817e-01 -5.89980662e-01 -2.19391763e-01
-1.05966859e-01 1.62704575e+00 -7.67003179e-01 -4.35273468e-01
-7.91333437e-01 6.33712530e-01 5.15762746e-01 4.00420129e-01
-6.52880430e-01 -4.15919453e-01 9.81552124e-01 -1.72722265e-01
7.58968651e-01 -1.50008917e-01 -9.22039568e-01 -1.37787569e+00
2.22894087e-01 -9.10372853e-01 3.63690168e-01 -5.58490992e-01
-1.49976599e+00 1.05504429e+00 -4.28948313e-01 -1.28092015e+00
-2.73171604e-01 -5.67748785e-01 -9.40463066e-01 7.43086636e-01
-1.85511935e+00 -1.65999794e+00 4.66215350e-02 2.66062498e-01
6.44835472e-01 -6.20647967e-02 8.72479916e-01 -2.12675408e-01
-2.99317211e-01 4.94134188e-01 -5.26581764e-01 5.68015054e-02
5.05861461e-01 -1.62103319e+00 3.52590710e-01 6.46260917e-01
1.71579883e-01 1.10425389e+00 6.00662470e-01 -6.32266939e-01
-1.19628966e+00 -9.28697944e-01 1.33138561e+00 -8.63443196e-01
7.51033247e-01 -1.06797136e-01 -6.64734542e-01 7.23237038e-01
7.22288907e-01 -3.37398678e-01 1.30694008e+00 8.83387446e-01
-5.39936543e-01 -1.50368169e-01 -8.13282251e-01 5.63068867e-01
7.77016938e-01 -4.84140873e-01 -9.26045001e-01 -2.28261724e-01
1.07267737e+00 2.00470403e-01 -8.27364802e-01 5.87738454e-01
9.00509834e-01 -1.04218292e+00 5.36133349e-01 -6.04325950e-01
8.39322865e-01 -2.23957047e-01 1.09506041e-01 -1.49344242e+00
-6.73204362e-02 -4.15867530e-02 -3.66873682e-01 1.52925479e+00
9.39878643e-01 -8.44800234e-01 7.62088478e-01 4.40450877e-01
-5.00350371e-02 -1.25852799e+00 -3.63714010e-01 -1.50587112e-01
1.36022627e-01 -7.60265410e-01 1.06003821e+00 6.57105446e-01
4.28035587e-01 1.04519379e+00 -4.03132439e-01 5.16181327e-02
1.64765373e-01 1.02677095e+00 5.22451580e-01 -1.10624194e+00
-2.62852043e-01 -7.07697690e-01 -1.96087167e-01 -1.23702252e+00
4.39426154e-01 -6.00117922e-01 -1.61474302e-01 -1.92417562e+00
4.08635706e-01 -2.75014371e-01 -5.03121972e-01 4.27645981e-01
-8.58213127e-01 4.93625641e-01 5.05211540e-02 -9.23343375e-02
-9.93305743e-01 7.65658379e-01 1.31559098e+00 -4.95939374e-01
-1.57945573e-01 3.74532759e-01 -1.55670905e+00 9.08310711e-01
9.47280467e-01 -3.82612079e-01 -4.87208277e-01 -2.45084867e-01
8.36469650e-01 -2.84081280e-01 -5.01285851e-01 -5.48525572e-01
2.82223988e-03 -1.95951581e-01 2.58895129e-01 -9.52963173e-01
1.86017528e-01 -7.18710124e-01 -5.63859463e-01 -2.01178109e-03
-1.78950742e-01 1.80902898e-01 -7.12755919e-02 7.06258833e-01
-7.57558763e-01 -1.05141774e-01 7.95152262e-02 -5.13829999e-02
-6.19699597e-01 5.43292999e-01 -4.00844246e-01 1.69492811e-01
6.88227415e-01 -1.82031486e-02 -2.38912314e-01 -4.99169111e-01
-4.12259966e-01 3.54175508e-01 5.22020161e-01 4.84977245e-01
5.09262383e-01 -1.20715463e+00 -4.63523656e-01 2.87365943e-01
4.00478512e-01 -8.48996863e-02 9.18763429e-02 7.03505516e-01
9.87865478e-02 5.32781005e-01 1.81526080e-01 -3.65180194e-01
-1.09206712e+00 7.13883400e-01 -9.78336930e-02 -1.03893793e+00
-1.91067711e-01 5.92597425e-01 2.32432827e-01 -6.54031098e-01
-2.45289415e-01 -3.75326663e-01 -9.91858363e-01 5.60004294e-01
4.41418618e-01 -2.90783048e-01 -1.06645308e-01 -5.07325411e-01
-2.91874677e-01 1.13012135e+00 -1.32268727e-01 -1.60013050e-01
1.35437751e+00 -3.09437335e-01 -2.53045321e-01 5.11868596e-01
9.95274246e-01 6.23891115e-01 -9.95191395e-01 3.03911753e-02
5.47357537e-02 -9.08284262e-02 -2.24885136e-01 -7.62218714e-01
-1.22707689e+00 8.22667301e-01 -1.26171336e-01 6.57375157e-01
1.39430177e+00 1.04309343e-01 1.02769816e+00 9.51699466e-02
3.40292305e-01 -8.35217595e-01 1.35981012e-02 1.01545465e+00
5.13520062e-01 -1.25092053e+00 4.11611497e-02 -2.65953273e-01
-1.07197917e+00 8.86487603e-01 7.74944127e-01 -1.64252952e-01
9.41113591e-01 4.80678767e-01 4.37341124e-01 -4.35685575e-01
-1.05603111e+00 -7.52501190e-01 5.15012145e-01 6.23488605e-01
7.47992516e-01 1.38418466e-01 -9.44999278e-01 1.22575974e+00
-5.62883496e-01 -2.04824999e-01 4.29109722e-01 8.00743401e-01
-4.85283107e-01 -1.24090946e+00 4.36028764e-02 7.48641491e-01
-8.79755974e-01 -5.18882990e-01 -6.19543254e-01 3.72563392e-01
4.83413413e-02 1.21232319e+00 -4.07734746e-03 -3.36592525e-01
3.88213336e-01 2.04790562e-01 -1.11378670e-01 -7.26616979e-01
-8.27711821e-01 9.59919184e-04 2.44712502e-01 -2.56647825e-01
-8.25823545e-01 -3.27233553e-01 -8.92924428e-01 1.03498287e-01
-5.12655497e-01 5.12324750e-01 5.28867900e-01 1.09794080e+00
4.50463265e-01 6.02603912e-01 9.24897969e-01 -3.82820576e-01
-3.30784172e-02 -1.20805943e+00 -5.28895915e-01 2.80328095e-01
3.83811474e-01 -4.02284026e-01 -4.60193485e-01 -1.61289796e-01] | [11.431374549865723, 6.683012008666992] |
3b73e4db-11d4-4475-a7ec-c5b295736bd9 | tecno-surgical-phase-recognition-with-multi | 2003.10751 | null | https://arxiv.org/abs/2003.10751v1 | https://arxiv.org/pdf/2003.10751v1.pdf | TeCNO: Surgical Phase Recognition with Multi-Stage Temporal Convolutional Networks | Automatic surgical phase recognition is a challenging and crucial task with the potential to improve patient safety and become an integral part of intra-operative decision-support systems. In this paper, we propose, for the first time in workflow analysis, a Multi-Stage Temporal Convolutional Network (MS-TCN) that performs hierarchical prediction refinement for surgical phase recognition. Causal, dilated convolutions allow for a large receptive field and online inference with smooth predictions even during ambiguous transitions. Our method is thoroughly evaluated on two datasets of laparoscopic cholecystectomy videos with and without the use of additional surgical tool information. Outperforming various state-of-the-art LSTM approaches, we verify the suitability of the proposed causal MS-TCN for surgical phase recognition. | ['Nassir Navab', 'Matthias Keicher', 'Walter Simson', 'Magdalini Paschali', 'Seong Tae Kim', 'Tobias Czempiel', 'Hubertus Feussner'] | 2020-03-24 | null | null | null | null | ['surgical-phase-recognition'] | ['computer-vision'] | [ 5.84700167e-01 4.85858619e-01 -6.04425132e-01 -3.61672521e-01
-3.38291943e-01 -4.79923874e-01 7.09758043e-01 4.41858947e-01
-6.42250597e-01 3.23515952e-01 4.55795497e-01 -7.08382726e-01
-6.92834198e-01 -2.64780581e-01 -5.56857467e-01 -5.81135511e-01
-5.18912554e-01 3.95689636e-01 1.02802873e-01 1.27059266e-01
2.45192781e-01 7.15045094e-01 -1.14920843e+00 9.60670471e-01
4.44580704e-01 1.32857680e+00 1.47038043e-01 7.49263048e-01
1.09646633e-01 1.09466898e+00 -8.02322179e-02 6.87296689e-02
3.41419101e-01 2.88395882e-02 -9.13638234e-01 -3.30959439e-01
2.37293065e-01 -8.93250033e-02 -4.81281847e-01 6.07411027e-01
3.38142097e-01 5.85714690e-02 1.79310754e-01 -5.17856896e-01
2.74408638e-01 9.05756235e-01 8.90314355e-02 3.37472081e-01
4.75458391e-02 2.77763993e-01 3.55473697e-01 -4.15327907e-01
9.16500628e-01 6.15559459e-01 8.43045175e-01 5.89292347e-01
-1.04664290e+00 -6.30807221e-01 8.62178802e-02 1.73132718e-01
-8.30711067e-01 -3.71631920e-01 3.86185527e-01 -7.58997142e-01
1.09504330e+00 2.48792648e-01 1.15230596e+00 1.31556511e+00
1.06306291e+00 8.77212107e-01 9.49336708e-01 -1.31796032e-01
2.61301547e-02 -2.95289993e-01 -1.18644275e-01 9.92800951e-01
-2.80531123e-02 6.01947188e-01 -6.65400386e-01 8.76483992e-02
8.50874543e-01 4.21293944e-01 -3.95577908e-01 -5.70495427e-01
-1.89402568e+00 4.12569046e-01 6.23241842e-01 6.14493489e-01
-6.54659271e-01 4.34704661e-01 6.38622940e-01 -4.29940969e-02
6.11649714e-02 6.84324205e-01 -6.45237923e-01 -4.61261541e-01
-1.06596255e+00 -2.44052678e-01 9.09386516e-01 8.09174359e-01
-5.40921241e-02 -2.12662652e-01 -6.97258949e-01 6.38544708e-02
-1.12469830e-01 -3.36226940e-01 7.91041613e-01 -1.01630235e+00
6.01861849e-02 7.05835938e-01 -1.57058612e-02 -3.78425509e-01
-1.17398477e+00 -6.80782974e-01 -1.11727154e+00 4.08806168e-02
1.81937873e-01 9.53594819e-02 -1.23390448e+00 1.10338092e+00
-1.50162831e-01 3.78905177e-01 -3.49487662e-02 5.14201641e-01
8.89846623e-01 -4.60627005e-02 4.33633238e-01 -4.83906835e-01
1.29407668e+00 -9.81183648e-01 -8.02480519e-01 -2.29669645e-01
8.47519934e-01 -5.84459126e-01 3.02804291e-01 8.12900364e-01
-8.89126956e-01 -4.56143975e-01 -7.91806519e-01 1.17580906e-01
-1.27118677e-01 4.59164172e-01 1.06786585e+00 1.08672939e-01
-9.43841577e-01 1.08705986e+00 -1.50990105e+00 -3.04651886e-01
4.09515202e-01 8.56434822e-01 -5.97670376e-01 1.24691382e-01
-8.69322300e-01 1.10765040e+00 7.68391550e-01 7.49492526e-01
-1.21486235e+00 -8.09578001e-01 -9.15013671e-01 9.76060424e-03
3.74292493e-01 -9.09076393e-01 1.62022471e+00 -7.96052039e-01
-1.66280246e+00 8.12838554e-01 -1.41369417e-01 -9.54528809e-01
5.68461835e-01 -4.36178982e-01 -2.27119863e-01 1.30317703e-01
-4.81642127e-01 7.06651330e-01 7.21144855e-01 -8.17583382e-01
-6.03046536e-01 -1.88217565e-01 -7.80684948e-02 1.41959399e-01
-5.29085957e-02 -3.55868936e-01 -4.81288612e-01 -5.26469886e-01
1.67022273e-02 -1.25510645e+00 -8.22980583e-01 1.14707008e-01
-1.81048185e-01 6.88616335e-02 3.61719161e-01 -7.96199977e-01
1.48304868e+00 -1.95998526e+00 3.35376948e-01 7.78501779e-02
2.87544161e-01 4.82779115e-01 2.77098119e-01 7.83280432e-02
-2.59116709e-01 -2.66293436e-01 -4.07407284e-02 -3.08516264e-01
-4.46835488e-01 5.09966791e-01 3.96318398e-02 5.61975062e-01
5.09068277e-03 1.09136760e+00 -1.04371071e+00 -5.51004291e-01
7.23963976e-01 2.93656915e-01 -4.64566827e-01 2.07749754e-01
-9.67787206e-02 9.09064233e-01 -2.87656844e-01 5.50860047e-01
-5.15412204e-02 -2.58579135e-01 4.38550860e-01 -4.04114902e-01
-5.31445384e-01 2.31573254e-01 -5.80194950e-01 2.55342102e+00
-7.74620712e-01 6.59427524e-01 -9.14442316e-02 -8.15019667e-01
4.52273667e-01 6.51514053e-01 8.68252814e-01 -7.95852840e-01
2.60244995e-01 1.94981322e-01 3.86390597e-01 -6.47185922e-01
5.32915890e-01 -1.56801745e-01 -1.31694138e-01 -2.80459106e-01
5.01828909e-01 2.95654446e-01 5.58512583e-02 -1.37895346e-01
1.26480091e+00 4.59766090e-01 6.33531451e-01 -3.49749446e-01
3.55842888e-01 1.89677194e-01 6.67644680e-01 8.82227719e-01
-3.76083940e-01 2.33828306e-01 4.71945465e-01 -9.94677067e-01
-5.51708758e-01 -8.60938847e-01 -1.68043211e-01 5.89930356e-01
-4.26463457e-03 -2.66615242e-01 -1.92707881e-01 -8.44281256e-01
-1.43192992e-01 6.19315982e-01 -1.15482974e+00 -2.65153140e-01
-9.01326180e-01 -3.10462326e-01 5.39484978e-01 6.64885342e-01
-9.01345536e-02 -1.22470164e+00 -1.06878626e+00 6.01958275e-01
-4.40348573e-02 -1.23503363e+00 -2.49976646e-02 7.53485501e-01
-1.35257721e+00 -1.13842225e+00 -3.94805759e-01 -8.40855837e-01
5.28482676e-01 -2.38704219e-01 1.10850334e+00 -2.77907342e-01
-7.02610970e-01 1.56587452e-01 -4.31244671e-02 -3.55947584e-01
-5.41552365e-01 2.09339961e-01 -2.10871547e-01 -3.35835844e-01
-3.08682732e-02 -2.12079301e-01 -1.06445467e+00 7.10626170e-02
-5.26675344e-01 6.40592635e-01 1.08711135e+00 9.83893335e-01
6.74022257e-01 -4.47738230e-01 -1.91851079e-01 -1.11683714e+00
2.00394377e-01 -3.14264387e-01 -6.62792504e-01 2.36718595e-01
-8.00229490e-01 3.41485143e-01 4.71337289e-01 -3.52617562e-01
-1.06220472e+00 4.15145218e-01 7.42976144e-02 -7.93204665e-01
-2.60612130e-01 7.46896565e-01 7.43347049e-01 -1.85959175e-01
4.77885127e-01 1.75041646e-01 1.94343626e-02 -2.77133644e-01
9.76415500e-02 9.07403082e-02 7.67003953e-01 -1.24584414e-01
8.79862905e-02 5.57582676e-01 4.86148328e-01 -4.14132208e-01
-9.61005330e-01 -8.48501921e-01 -7.33297408e-01 -3.07085663e-01
1.01838279e+00 -8.16369116e-01 -1.20636976e+00 1.36610880e-01
-9.44602013e-01 -7.01948225e-01 -2.01341003e-01 6.95114613e-01
-7.09239006e-01 5.75806014e-02 -9.33077335e-01 -4.38331693e-01
-5.86418211e-01 -1.24824870e+00 9.37507749e-01 1.18992306e-01
-2.70304859e-01 -1.12836480e+00 -1.10286148e-02 8.43954682e-02
4.44282532e-01 5.26941478e-01 4.96158898e-01 -6.76829576e-01
-7.58240581e-01 -4.00560021e-01 1.49422184e-01 -5.56309223e-02
4.99420539e-02 -5.04072290e-03 -7.07699418e-01 -1.77239239e-01
-1.31767973e-01 1.91550508e-01 9.19478118e-01 7.15085208e-01
1.57223618e+00 -3.11896294e-01 -6.42750800e-01 1.04388845e+00
1.24174011e+00 2.35232785e-01 4.68720824e-01 5.27919173e-01
5.28104603e-01 3.78162116e-01 8.73078525e-01 4.76729840e-01
1.16058469e-01 3.04949760e-01 6.48987830e-01 -1.90350786e-01
-1.83786061e-02 -3.64319468e-03 1.65344067e-02 7.71109760e-01
-2.89960802e-01 8.67255852e-02 -1.24864542e+00 4.94730830e-01
-2.01950479e+00 -8.96176815e-01 -2.12258417e-02 2.13165736e+00
6.76535308e-01 2.72437990e-01 -6.19243801e-01 -1.95597976e-01
1.00414775e-01 -1.11298915e-02 -1.41134545e-01 -4.52356607e-01
5.71322441e-01 5.71559608e-01 9.90699053e-01 3.30322415e-01
-1.57915890e+00 7.05193758e-01 5.65415239e+00 5.18256187e-01
-1.42791235e+00 8.94682575e-03 3.58731925e-01 -2.57961005e-01
3.59907329e-01 2.75219642e-02 -4.29386795e-01 1.22607075e-01
9.93738949e-01 2.25180343e-01 2.22890541e-01 7.69765377e-01
3.89267385e-01 -2.94692427e-01 -1.51228476e+00 6.59661531e-01
-3.34746599e-01 -1.89903057e+00 -2.31586993e-01 -1.39007717e-01
5.37002981e-01 3.30675691e-01 -1.21575072e-01 3.71001899e-01
1.09354369e-01 -1.30403256e+00 4.22563970e-01 8.68485510e-01
7.40236521e-01 -3.13165665e-01 1.02853394e+00 3.83199334e-01
-1.15415895e+00 -4.11529392e-01 4.39150065e-01 1.15628473e-01
7.11589679e-02 6.73636422e-02 -1.28504491e+00 7.75671959e-01
7.05569267e-01 9.58499491e-01 -2.07724974e-01 1.33952332e+00
-2.12812707e-01 3.86951536e-01 -2.22012758e-01 2.94697583e-01
5.01376688e-01 3.98645818e-01 4.40562129e-01 1.52827680e+00
3.24175030e-01 -4.54613157e-02 1.97151288e-01 2.92391598e-01
8.02107230e-02 -2.89779902e-01 -3.86887044e-01 -1.77576497e-01
-1.54619843e-01 1.36995327e+00 -7.62150407e-01 -2.54626870e-01
-8.91194195e-02 8.54774177e-01 2.13144973e-01 1.15059875e-01
-5.61550975e-01 1.18371785e-01 3.34856510e-01 -4.88245673e-02
1.84470952e-01 -3.13072205e-01 -4.57150996e-01 -9.50343549e-01
-2.20832467e-01 -5.49641252e-01 6.36918664e-01 -2.81417519e-01
-4.21417087e-01 6.18256450e-01 -1.78558215e-01 -1.71490586e+00
-5.15547991e-01 -8.21385741e-01 -3.83665055e-01 5.19951701e-01
-1.69325829e+00 -1.33905911e+00 -5.72970390e-01 5.52040637e-01
5.59942842e-01 1.69349179e-01 1.21554995e+00 1.56856373e-01
-2.89317250e-01 1.37553930e-01 -2.52418578e-01 1.07160985e-01
7.05367625e-01 -1.07506442e+00 -3.72712575e-02 6.92352653e-01
-3.17676395e-01 8.85823607e-01 6.67549431e-01 -5.50343633e-01
-1.40251303e+00 -1.32250047e+00 7.90346265e-01 -3.13305080e-01
6.40498161e-01 3.35868895e-02 -4.80356514e-01 9.69127357e-01
2.81219274e-01 1.90307289e-01 8.43069851e-01 1.16560869e-01
2.45040298e-01 -1.27763912e-01 -7.30637908e-01 4.06238735e-01
9.10580814e-01 -8.97342935e-02 -6.74003541e-01 2.64956534e-01
5.11289120e-01 -1.16060984e+00 -1.47031653e+00 1.11002183e+00
8.36220086e-01 -9.57216620e-01 8.81313205e-01 -4.41193670e-01
5.60663462e-01 -1.11620845e-02 6.21040761e-01 -1.01744616e+00
-2.64771491e-01 -8.12842429e-01 -4.53819633e-01 1.38377488e-01
2.69633174e-01 -1.67248487e-01 8.97594988e-01 4.43433106e-01
-8.70735884e-01 -1.03638148e+00 -1.01411045e+00 -2.56814659e-01
-3.98163348e-01 -4.70709264e-01 -2.94520140e-01 7.61566758e-01
1.12760611e-01 -2.77070850e-01 -2.74617195e-01 1.63741693e-01
3.45458925e-01 2.38556579e-01 3.20581377e-01 -1.16358137e+00
-3.57531607e-01 -7.23710120e-01 -3.39218915e-01 -6.01156950e-01
-1.06700085e-01 -8.80659759e-01 2.68902957e-01 -1.86913502e+00
6.55434951e-02 -3.98056507e-01 -7.59399414e-01 8.03387582e-01
1.27184555e-01 5.45083135e-02 1.14383483e-02 1.85811684e-01
-6.37930810e-01 2.19062388e-01 1.37749207e+00 -5.89156747e-02
-3.68585080e-01 3.08130056e-01 7.61625096e-02 8.76337230e-01
5.68957269e-01 -4.80735064e-01 -9.54582170e-02 -2.66412884e-01
2.88689464e-01 6.01045430e-01 2.75407225e-01 -1.07678545e+00
6.19364798e-01 -3.49428765e-02 5.04406035e-01 -5.49045146e-01
2.39340812e-01 -8.42659950e-01 3.56458157e-01 1.27691495e+00
-4.63195711e-01 -1.63536459e-01 6.13772035e-01 5.25399566e-01
-4.38170135e-01 2.45777130e-01 5.27411163e-01 -4.11759555e-01
-1.02538836e+00 4.93228137e-01 -6.68823481e-01 -4.97477770e-01
1.08970761e+00 -1.44067436e-01 -1.28112748e-01 3.31755161e-01
-1.16318429e+00 3.14674705e-01 4.55442891e-02 5.81041276e-01
6.53469265e-01 -7.52772152e-01 -2.58699894e-01 1.27924532e-01
2.95922160e-01 3.94766718e-01 6.10649765e-01 1.59738314e+00
-8.32211852e-01 8.91640127e-01 -3.07610750e-01 -8.70889604e-01
-1.17786551e+00 6.08757734e-01 5.42667687e-01 -8.94113481e-01
-8.65231514e-01 7.73885369e-01 2.87116975e-01 -3.38865727e-01
5.84246278e-01 -9.41913426e-01 -4.45821285e-01 -6.21523038e-02
6.42239302e-02 -1.76587000e-01 3.66366416e-01 2.99122557e-02
-3.30241531e-01 6.78680167e-02 -1.08057708e-01 3.27286690e-01
1.33892775e+00 3.01359385e-01 -9.46093351e-02 4.96528417e-01
6.81483686e-01 -5.72853804e-01 -1.18810999e+00 -1.96608026e-02
3.15623522e-01 -6.08819835e-02 2.88851857e-01 -1.08381104e+00
-8.20588589e-01 8.07424605e-01 8.16186726e-01 -4.46864247e-01
1.17297947e+00 -2.60908961e-01 5.74887335e-01 4.19605553e-01
4.67973083e-01 -7.79614091e-01 -1.66590407e-01 5.19938648e-01
7.41628945e-01 -1.25566673e+00 -8.52908194e-02 -2.17063650e-01
-6.22390866e-01 1.46752203e+00 5.07356465e-01 4.58722562e-03
5.71664989e-01 4.08723921e-01 8.72240812e-02 -1.76088929e-01
-9.75862324e-01 6.19141422e-02 6.55955136e-01 -1.77447591e-02
6.48071587e-01 2.32215345e-01 -3.43485028e-01 4.77791041e-01
2.58240879e-01 6.80920959e-01 2.67101943e-01 1.25832343e+00
1.64733641e-02 -8.45242918e-01 3.84841293e-01 7.21509635e-01
-7.55765438e-01 -2.91043937e-01 3.53209734e-01 8.52804303e-01
1.42460659e-01 2.95126081e-01 -6.09329976e-02 -3.21289927e-01
2.83095360e-01 -2.93324701e-02 7.10547566e-01 -5.64628065e-01
-1.43012500e+00 1.57605246e-01 2.24691153e-01 -1.32470965e+00
-4.92568612e-01 -5.34997165e-01 -1.30891299e+00 3.29246819e-01
1.14870049e-01 -2.03888193e-01 8.41777980e-01 9.05222118e-01
3.97424102e-01 1.26545858e+00 -7.48617016e-03 -1.21602666e+00
-3.46054554e-01 -9.08389330e-01 7.29147792e-02 1.29753381e-01
6.32177591e-01 -5.49046993e-01 1.38243467e-01 1.02425247e-01] | [14.072769165039062, -3.364346504211426] |
1e45ee16-362e-4221-8b23-7b7fc1f2f8cc | nefii-inverse-rendering-for-reflectance | 2303.16617 | null | https://arxiv.org/abs/2303.16617v1 | https://arxiv.org/pdf/2303.16617v1.pdf | NeFII: Inverse Rendering for Reflectance Decomposition with Near-Field Indirect Illumination | Inverse rendering methods aim to estimate geometry, materials and illumination from multi-view RGB images. In order to achieve better decomposition, recent approaches attempt to model indirect illuminations reflected from different materials via Spherical Gaussians (SG), which, however, tends to blur the high-frequency reflection details. In this paper, we propose an end-to-end inverse rendering pipeline that decomposes materials and illumination from multi-view images, while considering near-field indirect illumination. In a nutshell, we introduce the Monte Carlo sampling based path tracing and cache the indirect illumination as neural radiance, enabling a physics-faithful and easy-to-optimize inverse rendering method. To enhance efficiency and practicality, we leverage SG to represent the smooth environment illuminations and apply importance sampling techniques. To supervise indirect illuminations from unobserved directions, we develop a novel radiance consistency constraint between implicit neural radiance and path tracing results of unobserved rays along with the joint optimization of materials and illuminations, thus significantly improving the decomposition performance. Extensive experiments demonstrate that our method outperforms the state-of-the-art on multiple synthetic and real datasets, especially in terms of inter-reflection decomposition. | ['Xin Yu', 'Changjie Fan', 'Yongqiang Zhang', 'Lincheng Li', 'Zhipeng Hu', 'Haoqian Wu'] | 2023-03-29 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wu_NeFII_Inverse_Rendering_for_Reflectance_Decomposition_With_Near-Field_Indirect_Illumination_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wu_NeFII_Inverse_Rendering_for_Reflectance_Decomposition_With_Near-Field_Indirect_Illumination_CVPR_2023_paper.pdf | cvpr-2023-1 | ['inverse-rendering'] | ['computer-vision'] | [ 3.38043630e-01 -4.91812974e-01 7.58607864e-01 -3.81558657e-01
-6.25810206e-01 -4.32512581e-01 6.22320890e-01 -4.13117886e-01
-2.17928812e-02 5.97594202e-01 3.09678733e-01 -9.45081338e-02
6.37435541e-02 -1.08849585e+00 -7.66193330e-01 -9.70910847e-01
5.08689940e-01 2.80086100e-01 -8.40576179e-03 8.41040537e-02
1.26495108e-01 6.04411304e-01 -1.69353235e+00 3.36504519e-01
1.07434499e+00 8.39266419e-01 2.09838256e-01 6.99963570e-01
-1.51863679e-01 7.91259944e-01 -1.86228231e-01 -2.18832061e-01
3.42434615e-01 -3.54830116e-01 -5.10341287e-01 7.34248683e-02
6.53807819e-01 -7.41470635e-01 -1.72994778e-01 7.83573210e-01
4.28881168e-01 3.93322974e-01 5.78649700e-01 -7.45603561e-01
-5.67988217e-01 -4.06632245e-01 -7.90590584e-01 -3.83615315e-01
4.33522612e-01 3.40335339e-01 7.74554431e-01 -9.47665095e-01
3.76382768e-01 1.22439754e+00 7.59341896e-01 3.01428527e-01
-1.41575944e+00 -6.26234412e-01 3.91309261e-01 -1.96351428e-02
-1.30913281e+00 -3.94088596e-01 1.14607430e+00 -7.73074999e-02
7.80804873e-01 5.37883699e-01 8.36964846e-01 9.42319334e-01
8.59892070e-02 5.53689897e-01 1.40084815e+00 -3.48717690e-01
2.44945362e-01 2.30023358e-02 -1.73993170e-01 7.76622117e-01
-1.43271521e-01 3.55935544e-01 -6.37043238e-01 -4.04925168e-01
8.79645109e-01 8.97644609e-02 -7.92407393e-01 -3.18358451e-01
-1.02212954e+00 4.10784751e-01 6.02843225e-01 -4.53145832e-01
-6.32387042e-01 2.38231733e-01 -2.28780195e-01 -3.35097104e-01
9.31915343e-01 3.24621913e-03 -4.82353598e-01 7.71154016e-02
-5.84016204e-01 2.02323273e-01 7.30400264e-01 6.39904797e-01
1.13644516e+00 -1.05814539e-01 -1.97869428e-02 8.63115966e-01
7.44508266e-01 7.66416669e-01 -3.63292843e-01 -1.35278010e+00
2.84464478e-01 3.74617666e-01 2.44796947e-01 -7.69022703e-01
-1.92230687e-01 -5.63229859e-01 -8.02317381e-01 4.54678386e-01
3.26190859e-01 2.20841721e-01 -1.08585513e+00 1.48340559e+00
7.52521694e-01 4.66606855e-01 -2.00822651e-01 1.11988711e+00
6.57271206e-01 8.83063078e-01 -2.17254132e-01 -7.20570460e-02
1.29767907e+00 -1.17390203e+00 -5.19616663e-01 -9.19685215e-02
1.13471776e-01 -9.43064928e-01 1.13388169e+00 7.77749717e-01
-1.22057521e+00 -4.09306496e-01 -6.21588528e-01 -3.60347807e-01
1.34379447e-01 1.32677019e-01 8.36004794e-01 5.46008468e-01
-8.13219488e-01 6.49830163e-01 -1.18206453e+00 1.77736506e-01
4.20840412e-01 -7.74130300e-02 1.84716716e-01 -4.93579775e-01
-6.57109201e-01 3.97175074e-01 -4.41189289e-01 2.23134875e-01
-9.41025913e-01 -1.31498063e+00 -6.25507116e-01 -1.21683255e-01
2.35370964e-01 -1.23664582e+00 7.18500793e-01 -7.33566284e-01
-1.86046779e+00 4.06871766e-01 -3.92649502e-01 4.48015422e-01
4.82295752e-01 -3.02516162e-01 -1.12593964e-01 2.78881669e-01
-7.38044381e-02 2.62802809e-01 7.95617700e-01 -1.91632974e+00
-2.83088714e-01 -4.22850996e-01 2.35680714e-01 5.58080792e-01
-5.47039136e-02 -3.19549829e-01 -6.88005209e-01 -3.71968001e-01
4.45167720e-01 -8.16353500e-01 -2.48925179e-01 4.93608147e-01
-4.39355731e-01 4.81410414e-01 4.20787454e-01 -7.18403935e-01
6.20047331e-01 -1.93550801e+00 9.13628861e-02 1.57345146e-01
3.22843790e-01 -2.92827457e-01 -2.03606393e-02 1.04018316e-01
9.53783691e-02 -3.81791830e-01 -4.06213641e-01 -9.20820236e-01
-2.07770821e-02 3.49567354e-01 -4.17754382e-01 6.26740456e-01
-1.98506653e-01 5.99617362e-01 -9.23233211e-01 -1.53912932e-01
5.29234767e-01 1.44185603e+00 -9.27004755e-01 4.06073868e-01
-2.73944765e-01 1.06870234e+00 -4.30728376e-01 5.46190023e-01
1.27892017e+00 -3.18699569e-01 -2.67467171e-01 -7.12464392e-01
-2.81129718e-01 4.13529277e-01 -1.10804355e+00 2.04200840e+00
-1.08464265e+00 2.72746772e-01 4.34737653e-01 -2.54869491e-01
5.11773944e-01 4.66940030e-02 4.88923848e-01 -8.04262996e-01
-8.65000710e-02 -1.06369227e-01 -6.56978786e-01 -4.59697880e-02
4.94554400e-01 -1.65348694e-01 7.25579381e-01 5.70293725e-01
-5.50333202e-01 -5.29003382e-01 -4.47835207e-01 -1.34625202e-02
7.51460373e-01 1.01984441e+00 -1.80162311e-01 1.17681161e-01
5.09195387e-01 -2.84928739e-01 3.51505697e-01 5.24822533e-01
5.01860917e-01 1.03582120e+00 -1.91855654e-01 -4.16045874e-01
-6.96700275e-01 -1.49240315e+00 -8.40456113e-02 9.83724415e-01
2.05059260e-01 -1.74161538e-01 -8.91961455e-01 -2.08532035e-01
-2.83941507e-01 1.01297104e+00 -3.91029775e-01 1.76049143e-01
-5.58018804e-01 -1.04996669e+00 -1.50653154e-01 1.08106449e-01
5.57127357e-01 -5.90890467e-01 -7.67560244e-01 7.04777315e-02
-4.62251335e-01 -1.09981227e+00 -3.03170979e-01 -6.05840571e-02
-8.34394157e-01 -9.46392834e-01 -7.29365170e-01 7.15417638e-02
9.30239916e-01 7.24949419e-01 1.46754980e+00 7.14772940e-02
-5.52210033e-01 6.73405468e-01 -3.83062521e-03 -7.18283132e-02
-1.29357606e-01 -4.93917972e-01 -4.65536267e-01 2.92313486e-01
-1.79312348e-01 -9.01263237e-01 -1.13911128e+00 4.61640090e-01
-9.44739640e-01 7.17268109e-01 2.91446269e-01 5.11571109e-01
9.53577101e-01 -5.09811044e-02 -5.72265804e-01 -6.88971400e-01
1.30593196e-01 -2.48822823e-01 -8.19001257e-01 2.27283075e-01
-4.58544493e-01 -4.12930734e-02 7.65007675e-01 -2.32175395e-01
-1.73967755e+00 -8.78481716e-02 -2.50996560e-01 -5.08142829e-01
-3.16565722e-01 1.48522551e-03 -2.74656832e-01 -4.27880228e-01
4.05199647e-01 4.42149997e-01 -4.44129109e-01 -7.07121432e-01
5.16855717e-01 1.54495969e-01 3.27458769e-01 -9.31227505e-01
7.42865443e-01 1.26025903e+00 3.31839681e-01 -9.40092146e-01
-1.15950954e+00 -2.83146203e-01 -2.72847086e-01 -3.32063138e-01
7.16248930e-01 -1.08154988e+00 -9.38458204e-01 5.29320300e-01
-1.29745471e+00 -5.30808270e-01 -1.72726408e-01 3.91397148e-01
-5.08459866e-01 4.69871879e-01 -5.34745097e-01 -9.14205730e-01
-2.73138106e-01 -1.34325039e+00 1.48252213e+00 2.18849108e-01
2.21249893e-01 -1.00869203e+00 1.41228274e-01 4.63896155e-01
4.45462137e-01 2.96791255e-01 6.37151539e-01 8.08316410e-01
-1.21252286e+00 3.21016878e-01 -5.70960045e-01 4.11393434e-01
1.44522056e-01 -9.05483514e-02 -1.42666745e+00 -1.09828256e-01
3.73912960e-01 -5.86699471e-02 7.39939690e-01 5.16581893e-01
1.37171209e+00 -1.21364266e-01 -9.78654027e-02 1.52765667e+00
1.84321916e+00 -3.06849480e-01 7.90841281e-01 2.05286518e-01
1.17909575e+00 7.43411064e-01 3.76640499e-01 7.67551720e-01
7.56983161e-01 7.03878939e-01 8.37316632e-01 -3.96839917e-01
-4.82038230e-01 -9.85041782e-02 7.67639205e-02 7.90985823e-01
-4.33731973e-01 -2.62862295e-01 -4.99040663e-01 1.63781017e-01
-1.38527644e+00 -6.47320092e-01 -6.05334878e-01 2.30866408e+00
7.43021369e-01 -4.16172713e-01 -4.65170979e-01 -2.67271310e-01
7.54055008e-02 2.50404686e-01 -5.74544191e-01 -4.86854929e-03
9.98707023e-03 3.41102093e-01 4.56037313e-01 8.99557531e-01
-4.15414959e-01 6.19474888e-01 5.36960030e+00 5.90193689e-01
-1.02044380e+00 3.48458230e-01 7.31307447e-01 -3.15483928e-01
-1.08427978e+00 2.05603302e-01 -6.56899512e-01 2.73837864e-01
5.32044172e-01 6.22149467e-01 1.14715385e+00 3.24773014e-01
3.91011328e-01 -3.35360229e-01 -7.60141134e-01 9.39611316e-01
-1.05800088e-02 -1.02903771e+00 4.18908894e-02 1.50610626e-01
9.84469593e-01 2.56584346e-01 6.88934401e-02 -2.40301281e-01
3.76297742e-01 -6.10029101e-01 6.80920243e-01 8.90989780e-01
5.90240717e-01 -7.58571446e-01 -1.06753029e-01 2.45976433e-01
-1.10505342e+00 3.71229798e-01 -3.63219023e-01 1.12722494e-01
6.22921586e-01 9.94392991e-01 -1.36410892e-01 8.88944745e-01
7.07228005e-01 5.13652563e-01 -1.41250446e-01 6.63763106e-01
-5.32686412e-01 4.28172201e-01 -5.28736770e-01 4.24734175e-01
8.04978237e-03 -9.25219119e-01 4.98795211e-01 8.93153250e-01
3.11088234e-01 2.38622501e-01 8.71692523e-02 1.47209334e+00
2.05611423e-01 -2.30713621e-01 -8.88398290e-02 5.53714037e-01
-6.32279590e-02 1.48022723e+00 -7.36574829e-01 -6.49064183e-02
-5.31647801e-01 1.39980400e+00 1.52949587e-01 9.72010911e-01
-1.03873158e+00 2.38669552e-02 9.36461508e-01 3.00444663e-01
9.05632004e-02 -2.94463634e-01 -3.93146545e-01 -1.22178948e+00
1.00369751e-01 -5.08120120e-01 -2.76327878e-01 -1.24637353e+00
-1.16611874e+00 4.54396307e-01 -8.42770785e-02 -1.02959728e+00
2.30622873e-01 -6.03538036e-01 -4.66663808e-01 1.26798892e+00
-2.20306730e+00 -1.46155453e+00 -7.41078615e-01 7.18678594e-01
2.55078077e-01 6.51164949e-01 7.79800117e-01 2.25582063e-01
-2.17732444e-01 -5.14276773e-02 2.44051903e-01 -4.83553082e-01
6.32084370e-01 -1.07297862e+00 4.08096492e-01 7.07487822e-01
-2.76675820e-02 7.80236304e-01 6.41927123e-01 -4.47292894e-01
-1.81225586e+00 -1.05062103e+00 3.55806500e-02 -3.56643289e-01
2.47706190e-01 -2.98906654e-01 -8.25906813e-01 3.48196924e-01
8.02060366e-02 5.13517559e-01 6.60216987e-01 -9.47172567e-02
-4.38672274e-01 -1.84600830e-01 -1.18522859e+00 7.68482387e-01
1.19096899e+00 -6.15980029e-01 8.14406425e-02 4.51764017e-01
5.35730958e-01 -6.46075964e-01 -7.51895964e-01 2.32006878e-01
7.51572430e-01 -1.46798730e+00 1.61686873e+00 1.34658575e-01
6.48671269e-01 -6.27056718e-01 -3.14295202e-01 -1.40978897e+00
-2.04245493e-01 -6.94719315e-01 -3.70353192e-01 1.12398553e+00
-1.61852136e-01 -8.23599398e-01 7.12141752e-01 7.17689753e-01
-3.32488567e-01 -7.02103734e-01 -6.26560152e-01 -3.19823444e-01
-3.86979103e-01 -7.58516371e-01 7.05524147e-01 6.25082433e-01
-1.06115603e+00 -6.29707531e-04 -2.85469115e-01 6.72645450e-01
1.34688532e+00 5.15981317e-01 8.88654709e-01 -9.89847541e-01
-7.72190392e-01 -1.40052706e-01 4.79820073e-01 -1.49617410e+00
-1.19973079e-03 -5.20275354e-01 2.49143079e-01 -1.65459847e+00
1.52187020e-01 -5.96640110e-01 -1.27539914e-02 7.91326016e-02
-2.53269762e-01 5.86650372e-01 -1.50240660e-01 9.99202859e-03
-3.52056861e-01 8.95191312e-01 1.59491682e+00 1.64800301e-01
-1.62061259e-01 -7.64728710e-02 -5.84472895e-01 1.02447844e+00
3.36196333e-01 -4.39671725e-01 -4.99652207e-01 -1.01119053e+00
4.59846169e-01 -1.57714918e-01 7.79164732e-01 -7.35731065e-01
-3.18120443e-03 -2.26409912e-01 5.18722355e-01 -8.76577914e-01
8.67150307e-01 -9.11508739e-01 5.25612056e-01 7.82231688e-02
7.08145201e-02 -3.66753846e-01 1.22212164e-01 6.47575021e-01
3.09633404e-01 -3.05477111e-03 8.40355694e-01 -4.55452979e-01
-1.49026439e-01 4.78476077e-01 1.62151784e-01 -1.17138185e-01
4.12276953e-01 -2.55130738e-01 -1.26279697e-01 -3.53855491e-01
-1.88238189e-01 -1.72775000e-01 1.01870418e+00 -1.21878743e-01
7.22051799e-01 -1.12296939e+00 -5.70385575e-01 3.19613427e-01
5.28825819e-03 4.57127839e-01 6.16632521e-01 9.03301835e-01
-7.73721218e-01 -1.77078336e-01 3.22334737e-01 -7.57358789e-01
-1.08812642e+00 1.24644369e-01 3.10692638e-01 -2.09219307e-01
-8.39839697e-01 9.99692261e-01 8.96383464e-01 -7.74485171e-01
-1.87665541e-02 -3.80769938e-01 2.74830401e-01 -4.54189837e-01
6.84862196e-01 6.55484796e-01 1.97407044e-02 -4.63079810e-01
-1.93051159e-01 1.04440689e+00 2.14303702e-01 -2.33743042e-01
1.58360338e+00 -5.63053310e-01 -4.01879966e-01 1.98167726e-01
1.31217861e+00 4.36121583e-01 -1.87464225e+00 -9.93223935e-02
-9.44357157e-01 -9.61921632e-01 6.27877355e-01 -8.58053267e-01
-1.31341457e+00 9.99791563e-01 3.42885226e-01 -1.96123913e-01
1.42823958e+00 -3.99724811e-01 8.68078649e-01 2.06910670e-02
5.87673545e-01 -8.13017488e-01 -2.65596688e-01 2.92224169e-01
8.14639926e-01 -9.21108544e-01 5.36684036e-01 -7.57641733e-01
-1.64601579e-01 1.16160166e+00 2.50897050e-01 8.18369240e-02
6.87853277e-01 4.37698215e-01 3.50097008e-02 -1.99627534e-01
-5.42698264e-01 9.12559703e-02 3.57343763e-01 6.39695108e-01
4.63612795e-01 -1.51869118e-01 3.23951036e-01 -7.83436894e-02
1.37662500e-01 -1.92682426e-02 2.98021585e-01 6.03544354e-01
8.37384239e-02 -8.62000525e-01 -6.17238581e-01 1.43282460e-02
-1.56449661e-01 -3.64357352e-01 1.42257825e-01 1.84477255e-01
1.07234597e-01 8.32087815e-01 4.27642986e-02 1.48195207e-01
2.39096522e-01 -4.56206262e-01 7.14290738e-01 -3.52176905e-01
-3.60135347e-01 4.23935741e-01 -1.61293805e-01 -1.13005006e+00
-5.89223444e-01 -6.27457261e-01 -1.06950593e+00 -2.33628824e-01
-3.39221418e-01 -3.48700047e-01 1.07067394e+00 7.04954922e-01
3.69274080e-01 8.64289224e-01 7.89313734e-01 -1.44296885e+00
-2.41656974e-01 -4.58804548e-01 -3.56594324e-01 2.31555030e-01
4.75121826e-01 -6.23230815e-01 -6.80995762e-01 -1.36597276e-01] | [9.704903602600098, -3.1075592041015625] |
0797a58f-9385-400b-ad37-0d9656feb6a6 | augmenting-pre-trained-language-models-with | 2204.04581 | null | https://arxiv.org/abs/2204.04581v3 | https://arxiv.org/pdf/2204.04581v3.pdf | Augmenting Pre-trained Language Models with QA-Memory for Open-Domain Question Answering | Retrieval augmented language models have recently become the standard for knowledge intensive tasks. Rather than relying purely on latent semantics within the parameters of large neural models, these methods enlist a semi-parametric memory to encode an index of knowledge for the model to retrieve over. Most prior work has employed text passages as the unit of knowledge, which has high coverage at the cost of interpretability, controllability, and efficiency. The opposite properties arise in other methods which have instead relied on knowledge base (KB) facts. At the same time, more recent work has demonstrated the effectiveness of storing and retrieving from an index of Q-A pairs derived from text \citep{lewis2021paq}. This approach yields a high coverage knowledge representation that maintains KB-like properties due to its representations being more atomic units of information. In this work we push this line of research further by proposing a question-answer augmented encoder-decoder model and accompanying pretraining strategy. This yields an end-to-end system that not only outperforms prior QA retrieval methods on single-hop QA tasks but also enables compositional reasoning, as demonstrated by strong performance on two multi-hop QA datasets. Together, these methods improve the ability to interpret and control the model while narrowing the performance gap with passage retrieval systems. | ['William Cohen', 'John Wieting', 'Michiel de Jong', 'Pat Verga', 'Wenhu Chen'] | 2022-04-10 | null | null | null | null | ['passage-retrieval'] | ['natural-language-processing'] | [ 1.79245234e-01 4.76311117e-01 -3.82640749e-01 -5.04681319e-02
-1.31181812e+00 -6.37032270e-01 8.25335801e-01 3.50683570e-01
-5.38564086e-01 9.01248336e-01 5.93526781e-01 -3.60096037e-01
-4.15159464e-01 -9.16197777e-01 -9.46545601e-01 -3.71510535e-01
2.21795380e-01 8.83243799e-01 4.42223758e-01 -6.73287332e-01
2.22881451e-01 1.43320620e-01 -1.30128920e+00 7.10268736e-01
9.51640546e-01 8.12627435e-01 1.79983437e-01 5.74109972e-01
-2.87141144e-01 1.26929545e+00 -3.95214766e-01 -7.72127926e-01
-9.08104330e-02 -3.65462124e-01 -1.31488562e+00 -5.91009855e-01
5.68925917e-01 -5.01444161e-01 -5.52513957e-01 5.28915823e-01
4.78730559e-01 2.31692448e-01 7.55209565e-01 -6.50992513e-01
-1.25650096e+00 8.95035505e-01 3.71404774e-02 2.34097958e-01
5.84195614e-01 -5.05666658e-02 1.44954097e+00 -7.60040879e-01
6.69212341e-01 1.22097003e+00 4.76635814e-01 4.08996761e-01
-1.27136159e+00 -2.15724155e-01 -3.35815176e-02 5.13328910e-01
-1.33439553e+00 -5.99341154e-01 4.38103348e-01 7.66653046e-02
1.71590555e+00 1.46743536e-01 6.04825437e-01 9.99065220e-01
1.03135400e-01 1.07244492e+00 7.35240877e-01 -9.10282731e-01
4.75622229e-02 2.15234458e-01 3.52286220e-01 8.14323962e-01
2.58084118e-01 -6.69901147e-02 -9.63479519e-01 -1.25000834e-01
4.97575521e-01 -3.78770232e-01 -4.96019661e-01 -4.43577081e-01
-1.16448069e+00 8.77293289e-01 3.77479345e-01 1.05227679e-01
-4.64115441e-01 3.12192857e-01 4.24113184e-01 5.48093200e-01
9.33690518e-02 7.33526826e-01 -5.14040828e-01 -1.49480179e-01
-8.99780154e-01 3.42973173e-01 1.02821529e+00 9.97911751e-01
5.52834690e-01 -2.65107691e-01 -5.24831116e-01 7.79178798e-01
4.00463760e-01 6.02982581e-01 5.57673693e-01 -1.01393509e+00
7.77170181e-01 6.69609964e-01 2.08965465e-01 -7.42647171e-01
-8.01329166e-02 -4.48589742e-01 -4.29841578e-01 -4.60648477e-01
1.14352509e-01 3.81118894e-01 -8.91618550e-01 1.82127595e+00
-9.47660729e-02 -2.52433300e-01 5.53290308e-01 6.32270098e-01
6.37130558e-01 7.69227684e-01 1.52685553e-01 2.24877764e-02
1.56291544e+00 -1.14468515e+00 -8.37506950e-01 -2.46054381e-01
8.80384028e-01 -4.24566507e-01 1.32311177e+00 2.68366724e-01
-1.45162892e+00 -2.79181570e-01 -1.14113057e+00 -7.58349001e-01
-6.53575599e-01 -1.21730223e-01 6.79046273e-01 6.01007700e-01
-1.31077719e+00 2.36708120e-01 -6.53578699e-01 -2.06634164e-01
3.76179397e-01 3.03404570e-01 -2.33367071e-01 -3.00330460e-01
-1.82419169e+00 1.34698188e+00 8.16569448e-01 1.13884080e-02
-7.04162717e-01 -8.57949674e-01 -8.71738195e-01 5.00705779e-01
5.02985239e-01 -1.35778737e+00 1.43226826e+00 -6.05417371e-01
-1.61073017e+00 5.13029456e-01 -1.84868753e-01 -8.40227664e-01
9.90006253e-02 -5.79337180e-01 -3.72773647e-01 6.06205940e-01
-4.29728664e-02 8.40610743e-01 6.88338220e-01 -1.02893281e+00
-4.23939556e-01 -1.64256498e-01 5.96399486e-01 5.27165353e-01
-4.64175791e-01 -1.80645481e-01 -8.11058104e-01 -3.49511355e-01
-1.26236007e-01 -7.50490665e-01 3.34545702e-01 -2.86717325e-01
-2.29230717e-01 -2.45942906e-01 2.28772730e-01 -7.38588810e-01
1.48320377e+00 -1.84271312e+00 3.76751304e-01 4.35720868e-02
3.74040306e-02 3.87779146e-01 -1.46255046e-01 9.69056606e-01
2.88630664e-01 6.11321703e-02 -2.85707742e-01 -2.77960151e-01
3.27660769e-01 2.89805919e-01 -9.39692378e-01 -8.39286223e-02
3.54594529e-01 1.53509724e+00 -8.75906110e-01 -6.98525786e-01
-1.12516172e-01 4.88456160e-01 -6.63450718e-01 2.95210071e-02
-5.85403740e-01 -1.46817997e-01 -3.98793548e-01 4.72355992e-01
4.10422273e-02 -5.19005895e-01 1.12066410e-01 -8.26626718e-02
3.45529705e-01 7.40956604e-01 -7.79828548e-01 2.16990089e+00
-5.77021718e-01 4.14066732e-01 -3.25144202e-01 -7.15662360e-01
3.93926233e-01 5.83869219e-01 -8.94949511e-02 -1.03748727e+00
-2.29144022e-01 2.02734470e-01 -2.13255197e-01 -5.38372219e-01
9.65659082e-01 -2.46066332e-01 -7.98246935e-02 5.42762518e-01
1.94868132e-01 -2.72589713e-01 2.57725954e-01 7.37289965e-01
1.03155661e+00 2.78693169e-01 9.46637765e-02 -8.86539966e-02
6.24454796e-01 2.67260253e-01 -1.67003021e-01 1.11076725e+00
2.66113698e-01 2.78178573e-01 1.81894124e-01 9.25211608e-02
-9.60138500e-01 -1.23235691e+00 -1.07840225e-01 1.17905343e+00
4.78523225e-02 -4.59207565e-01 -4.47994947e-01 -3.17856640e-01
8.84506255e-02 1.20480275e+00 -4.96277094e-01 -7.16250896e-01
-6.82045996e-01 -3.34622264e-01 1.00693023e+00 7.16904700e-01
4.60000008e-01 -1.13845348e+00 -5.01018465e-01 2.21834093e-01
-5.41298747e-01 -1.03194678e+00 -2.18352303e-01 1.68790087e-01
-9.49830234e-01 -6.73333168e-01 -5.83156705e-01 -6.38674974e-01
3.89108270e-01 9.86065865e-02 1.55370533e+00 2.20104139e-02
7.61618316e-02 8.57477903e-01 -4.86910433e-01 -2.94827729e-01
-3.70690137e-01 4.47604030e-01 -1.87416747e-01 -3.51041257e-01
3.42665404e-01 -3.89545500e-01 -5.65724909e-01 -2.57186174e-01
-1.29793692e+00 1.35591522e-01 9.16423440e-01 1.05907273e+00
5.24901569e-01 -1.63638040e-01 9.59643424e-01 -7.79513299e-01
9.42426264e-01 -5.84631324e-01 -2.88626343e-01 7.11092293e-01
-6.76960230e-01 6.11344218e-01 5.49683511e-01 -1.06257781e-01
-1.43789136e+00 -4.69450414e-01 1.50610628e-02 -1.98368147e-01
2.19426006e-01 1.00927353e+00 1.17257945e-01 2.26902530e-01
6.97036803e-01 5.65741181e-01 3.74053754e-02 -3.76432627e-01
1.04923308e+00 5.69772720e-01 4.52044129e-01 -8.62049878e-01
5.14491737e-01 3.89838248e-01 -2.29882345e-01 -4.98242915e-01
-1.17869925e+00 -5.44975936e-01 -6.17723048e-01 3.40375721e-01
7.05457389e-01 -1.09347069e+00 -5.75486839e-01 -8.07473734e-02
-9.78441715e-01 -2.67944455e-01 -5.07122219e-01 2.93919712e-01
-7.66838193e-01 5.52974403e-01 -8.19554865e-01 -6.88799322e-01
-6.11665964e-01 -8.21780682e-01 1.10266829e+00 -1.84493780e-01
-2.80945629e-01 -1.11644053e+00 1.24501988e-01 7.54487991e-01
5.51190913e-01 -4.97207701e-01 1.49717081e+00 -6.09286785e-01
-9.90835488e-01 -1.69310719e-01 -2.35187456e-01 2.65117168e-01
-3.14337164e-01 -5.97305119e-01 -8.98637235e-01 -1.97207078e-01
-2.57639259e-01 -9.38233972e-01 1.19421852e+00 -9.44578871e-02
7.84312785e-01 -3.78693283e-01 -1.44050539e-01 2.24875778e-01
1.44674373e+00 -1.32827431e-01 6.96520746e-01 4.89965707e-01
3.19654107e-01 5.34571767e-01 3.17995131e-01 -1.77118242e-01
7.50212550e-01 7.02668548e-01 1.31377429e-01 3.23619395e-01
-3.87003303e-01 -4.93163764e-01 4.93350834e-01 1.09681845e+00
1.28779233e-01 -3.72227520e-01 -1.03220427e+00 8.75974059e-01
-1.84156203e+00 -9.70923424e-01 3.52655500e-01 2.01677823e+00
1.43280602e+00 5.48129380e-02 -4.07469332e-01 -2.12493818e-02
-1.00914404e-01 -8.77595413e-03 -4.14252251e-01 -4.77652222e-01
-3.30211550e-01 5.21627545e-01 2.27527335e-01 6.33182883e-01
-6.45228982e-01 1.08475041e+00 6.21574831e+00 8.64864349e-01
-5.70538580e-01 -4.60278662e-03 2.67410353e-02 -2.49083400e-01
-5.65477073e-01 1.13154538e-02 -9.24749851e-01 1.11786202e-01
1.33370817e+00 -1.47375837e-01 5.37549794e-01 3.50807220e-01
-3.66410375e-01 -2.31011569e-01 -1.38075721e+00 5.95668375e-01
2.74268448e-01 -1.47020006e+00 7.77437031e-01 -2.45416015e-01
4.83467966e-01 2.99085695e-02 2.33810961e-01 7.56237566e-01
2.43967220e-01 -1.20736206e+00 7.87636817e-01 9.39241052e-01
6.79477930e-01 -5.13119221e-01 6.58754587e-01 5.58344483e-01
-7.03421533e-01 -2.18975902e-01 -3.44509035e-01 -3.90994828e-03
2.93829352e-01 3.16023827e-02 -9.04032886e-01 9.21428084e-01
3.89313936e-01 2.28819400e-01 -6.22583449e-01 5.85962057e-01
-3.73802871e-01 6.28796756e-01 -3.39787006e-01 -3.89473920e-04
3.11056703e-01 6.24266528e-02 2.28194654e-01 1.31171703e+00
1.60298478e-02 2.13036001e-01 -2.89398223e-01 9.26003575e-01
-2.66520768e-01 1.31558746e-01 -5.58369100e-01 -2.58164525e-01
5.89656532e-01 6.72701240e-01 -1.11034453e-01 -5.84471583e-01
-5.09527206e-01 9.59247470e-01 8.05742204e-01 5.21434724e-01
-5.43505251e-01 -2.36279711e-01 1.31788731e-01 -1.73942983e-01
5.30704141e-01 -3.67343545e-01 4.19463177e-04 -1.30182385e+00
2.56328136e-01 -9.63818252e-01 6.15742981e-01 -8.90163898e-01
-1.20692575e+00 3.99484962e-01 3.41186196e-01 -6.62783086e-01
-5.71839929e-01 -5.43185592e-01 1.89432248e-01 1.01837397e+00
-2.06453371e+00 -1.29063773e+00 2.61079222e-01 6.41509295e-01
4.34381247e-01 1.33409100e-02 1.31002462e+00 2.48139158e-01
-2.80142665e-01 6.23484075e-01 2.35578015e-01 1.08979605e-01
5.67234755e-01 -1.27860868e+00 5.41146053e-03 6.07690215e-01
4.17781383e-01 1.26819336e+00 3.48862708e-01 -4.47780281e-01
-1.89399087e+00 -6.29570663e-01 1.10583878e+00 -1.07584906e+00
6.88412130e-01 -1.95467189e-01 -1.12835479e+00 8.74324203e-01
5.39680660e-01 -4.35813099e-01 7.48003542e-01 3.70200366e-01
-6.29051745e-01 -7.66668236e-03 -6.18759513e-01 6.92071676e-01
8.19100082e-01 -1.05515468e+00 -1.36774993e+00 2.53214747e-01
1.06613719e+00 -4.20007080e-01 -1.00670648e+00 2.94882089e-01
5.82986534e-01 -6.75008059e-01 1.22448921e+00 -9.12660360e-01
4.65175807e-01 -2.47427315e-01 -1.78737774e-01 -1.08706880e+00
-2.46923268e-01 -1.97604701e-01 -8.16521287e-01 1.06176353e+00
7.43712485e-01 -6.07075870e-01 5.03680408e-01 7.74787903e-01
-1.70579448e-01 -8.92020404e-01 -9.13024902e-01 -7.67632544e-01
2.29594007e-01 -3.61214787e-01 5.00768065e-01 5.34138262e-01
2.14548051e-01 7.89172113e-01 6.14590198e-03 6.86736479e-02
2.98075765e-01 1.81553066e-01 3.87331009e-01 -8.31917942e-01
-4.37094361e-01 -4.16216046e-01 1.44691020e-02 -1.41933978e+00
2.96732813e-01 -1.23778963e+00 -1.16992794e-01 -1.74657726e+00
2.73769617e-01 -3.96308929e-01 -4.79064226e-01 6.37390137e-01
-2.75872618e-01 6.60795942e-02 1.11933388e-01 4.15826976e-01
-9.54448998e-01 8.24314773e-01 1.08047736e+00 -1.14041016e-01
-2.39550825e-02 -5.51156342e-01 -8.61385107e-01 2.50629961e-01
6.64876401e-01 -1.43459424e-01 -8.64769816e-01 -9.39650953e-01
7.70581603e-01 1.73493981e-01 4.44409788e-01 -7.16795921e-01
4.90329832e-01 3.43863040e-01 1.31243750e-01 -4.00835454e-01
8.36211979e-01 -8.50554705e-01 -2.36511618e-01 1.91920266e-01
-8.51933181e-01 9.73287448e-02 4.63826954e-01 8.02885175e-01
-4.41500425e-01 -2.87485570e-01 2.28962794e-01 -2.72830218e-01
-8.06645811e-01 -6.86643347e-02 -2.17599183e-01 3.06915492e-01
4.66813534e-01 -1.11009456e-01 -6.69389546e-01 -4.90662336e-01
-4.78646368e-01 3.10033858e-01 1.28556043e-01 4.17003989e-01
6.38026118e-01 -9.72932100e-01 -6.79166079e-01 2.08097454e-02
3.11338902e-01 -1.77410059e-02 2.12312609e-01 8.31096113e-01
-4.71525609e-01 1.31238055e+00 -1.05558991e-01 -3.20459813e-01
-7.19660401e-01 6.97105408e-01 3.06234032e-01 -6.07699513e-01
-7.08643794e-01 8.36017787e-01 -7.21558481e-02 -3.14807832e-01
1.94789827e-01 -2.41338536e-01 -2.36932158e-01 9.93467197e-02
4.13223922e-01 2.94683948e-02 3.27278107e-01 -3.44044626e-01
-1.08192079e-01 1.38423160e-01 -5.22780418e-01 -4.09834385e-01
1.03474712e+00 -2.46854842e-01 -2.65476465e-01 5.42377591e-01
1.02962232e+00 2.30093189e-02 -6.70642793e-01 -6.94652140e-01
2.50193417e-01 -1.01674072e-01 1.30295232e-01 -1.32833016e+00
-3.73975694e-01 9.04488862e-01 1.29863009e-01 5.57114631e-02
1.04629993e+00 1.69112816e-01 9.30211782e-01 1.01000142e+00
3.64221662e-01 -9.31173623e-01 2.15212092e-01 7.71412551e-01
9.78686154e-01 -9.31533158e-01 -1.12453409e-01 -1.14862971e-01
-6.88145638e-01 9.03174520e-01 2.99977481e-01 2.71216899e-01
2.12157845e-01 -1.15699306e-01 -4.52921912e-02 -4.78301406e-01
-1.07447934e+00 -3.09584975e-01 4.71550435e-01 3.44354898e-01
4.42720950e-01 -1.94228470e-01 -1.49135455e-01 5.13571382e-01
-3.73250037e-01 1.23858355e-01 1.01150237e-01 1.02947164e+00
-3.38864356e-01 -1.08680785e+00 -1.37160681e-02 4.67274427e-01
-3.93336654e-01 -7.44655550e-01 -2.70904601e-01 8.69429946e-01
-3.41766238e-01 8.35628986e-01 -1.53647989e-01 2.16958672e-01
3.21134418e-01 6.88777685e-01 6.70129657e-01 -6.11082315e-01
-5.91792941e-01 -5.18368781e-01 3.74082774e-01 -3.93320292e-01
-3.50030333e-01 -3.17586601e-01 -1.28278422e+00 -6.98399395e-02
-4.97546613e-01 4.77865547e-01 2.84290373e-01 9.47007358e-01
7.87707150e-01 4.50029790e-01 -1.46690339e-01 -2.71356888e-02
-1.00491071e+00 -9.86206412e-01 -2.87842512e-01 3.27377409e-01
3.80780011e-01 -4.82504576e-01 4.07509990e-02 9.96420458e-02] | [11.076493263244629, 8.038080215454102] |
4e137d55-c19f-4ab4-b1b0-9e0905869739 | key-mention-pairs-guided-document-level | null | null | https://aclanthology.org/2022.coling-1.165 | https://aclanthology.org/2022.coling-1.165.pdf | Key Mention Pairs Guided Document-Level Relation Extraction | Document-level Relation Extraction (DocRE) aims at extracting relations between entities in a given document. Since different mention pairs may express different relations or even no relation, it is crucial to identify key mention pairs responsible for the entity-level relation labels. However, most recent studies treat different mentions equally while predicting the relations between entities, leading to sub-optimal performance. To this end, we propose a novel DocRE model called Key Mention pairs Guided Relation Extractor (KMGRE) to directly model mention-level relations, containing two modules: a mention-level relation extractor and a key instance classifier. These two modules could be iteratively optimized with an EM-based algorithm to enhance each other. We also propose a new method to solve the multi-label problem in optimizing the mention-level relation extractor. Experimental results on two public DocRE datasets demonstrate that the proposed model is effective and outperforms previous state-of-the-art models. | ['Shengda Fan', 'Shasha Mo', 'Jianwei Niu', 'Feng Jiang'] | null | null | null | null | coling-2022-10 | ['document-level-relation-extraction'] | ['natural-language-processing'] | [-1.66272689e-02 5.87087929e-01 -5.40555418e-01 -3.31486672e-01
-8.43923390e-01 -5.31264305e-01 5.78315556e-01 5.65352619e-01
-3.37047428e-01 8.32346201e-01 3.18352249e-03 -2.45247290e-01
-3.27544898e-01 -1.11457682e+00 -3.48314047e-01 -3.71683121e-01
-9.71077904e-02 8.23935032e-01 3.03844243e-01 -1.05056524e-01
-5.97223639e-03 4.62070137e-01 -1.15832996e+00 4.05160666e-01
1.19684374e+00 9.12896693e-01 -4.51463722e-02 2.67924815e-01
-4.45365667e-01 1.10823154e+00 -4.20253009e-01 -9.46567833e-01
-2.33358070e-02 -2.13420197e-01 -1.27896059e+00 1.87189449e-02
2.66890619e-02 1.71415091e-01 -1.82858229e-01 1.06045771e+00
1.64025739e-01 5.20840622e-02 8.90845597e-01 -1.25168288e+00
-5.13519347e-01 1.00409317e+00 -6.94540441e-01 1.02830231e-01
3.46576542e-01 -7.00436771e-01 1.70072997e+00 -8.26569557e-01
7.76029885e-01 1.03538764e+00 4.17748839e-01 6.99136332e-02
-9.93292511e-01 -6.84353769e-01 3.65128875e-01 4.24414247e-01
-1.68854868e+00 -2.84313619e-01 6.76308572e-01 -4.33600277e-01
1.31187069e+00 3.21626693e-01 2.55606800e-01 3.17467958e-01
6.64747059e-02 5.86263478e-01 8.87616694e-01 -4.64441329e-01
-2.33014300e-01 2.87429035e-01 5.01793206e-01 6.30699873e-01
4.65668231e-01 -5.26497006e-01 -3.26133132e-01 -1.71656489e-01
3.34413856e-01 -2.55569667e-01 -2.96859473e-01 -1.39485717e-01
-9.25777733e-01 5.74382365e-01 4.97245938e-01 4.22248363e-01
-4.28745419e-01 -4.47428942e-01 2.69079298e-01 1.37524068e-01
7.21832275e-01 7.07642019e-01 -9.27601159e-01 4.05265540e-01
-5.28200150e-01 1.93154797e-01 9.89812434e-01 1.07537997e+00
8.66285801e-01 -9.42573845e-01 -3.76502991e-01 1.09061086e+00
4.49108720e-01 -1.69500038e-01 2.08584771e-01 -2.31611356e-01
9.38755035e-01 1.29517448e+00 3.86119336e-02 -1.30749404e+00
-5.83401263e-01 -6.94166005e-01 -8.99781168e-01 -5.23676157e-01
-3.24626081e-02 -1.37322500e-01 -6.39177620e-01 1.32581890e+00
6.67268455e-01 4.83873457e-01 3.24396104e-01 5.64895570e-01
1.34069335e+00 6.55642509e-01 3.29184622e-01 -4.09986347e-01
1.84534705e+00 -1.13992620e+00 -8.78796637e-01 -2.35200047e-01
1.02995431e+00 -7.18204200e-01 3.35842013e-01 -1.48481699e-02
-7.60808527e-01 -1.66842923e-01 -1.00269878e+00 -2.56545067e-01
-5.73706806e-01 5.44729054e-01 8.66677940e-01 3.59680057e-01
-2.58868933e-01 4.40121502e-01 -6.08765185e-01 -4.13185880e-02
7.62180462e-02 5.06409824e-01 -5.31727612e-01 1.27715722e-01
-1.60858107e+00 1.12034225e+00 1.09009850e+00 2.34739348e-01
5.61897224e-03 -4.22043771e-01 -1.04077518e+00 3.72407317e-01
9.33684349e-01 -4.26547199e-01 9.75065172e-01 -1.33916005e-01
-1.12245059e+00 9.76566613e-01 -3.36879700e-01 -2.79518694e-01
-2.52749175e-01 -2.74434298e-01 -8.47643137e-01 -6.36793077e-02
7.72173330e-02 1.27680600e-01 2.11653393e-02 -1.35497189e+00
-9.54572678e-01 -2.70488352e-01 2.53595322e-01 2.64200032e-01
-3.76462787e-01 1.88822314e-01 -8.04837227e-01 -4.18481350e-01
5.28432667e-01 -7.03376949e-01 -2.04772964e-01 -1.12890005e+00
-8.95041764e-01 -7.02232361e-01 3.78092766e-01 -6.81679010e-01
1.71754992e+00 -1.53274679e+00 -3.30809541e-02 4.79133993e-01
4.67407912e-01 3.95514846e-01 4.30951603e-02 5.15031159e-01
-3.54728907e-01 2.56660014e-01 -6.90659210e-02 -2.35099673e-01
-1.31609038e-01 2.85929930e-03 -6.55485168e-02 -2.24638656e-02
5.23522437e-01 1.05669868e+00 -1.01252890e+00 -9.07746077e-01
-1.74241826e-01 3.28805178e-01 -2.00120628e-01 3.33952218e-01
-1.26287863e-01 2.09252432e-01 -5.83463788e-01 6.32200181e-01
7.86502421e-01 -4.97770786e-01 8.13697159e-01 -6.06245697e-01
9.50004533e-02 8.14727008e-01 -1.46524084e+00 1.02671707e+00
-4.81065124e-01 2.82991286e-02 -3.94088626e-01 -9.65417087e-01
1.11850250e+00 5.24537623e-01 6.11266673e-01 -3.11360478e-01
1.83818296e-01 2.77621865e-01 4.70496602e-02 -6.53690338e-01
7.93038070e-01 7.56392106e-02 -4.71851341e-02 4.11520563e-02
1.86283886e-01 2.51461267e-01 4.96990919e-01 4.78717566e-01
1.23212361e+00 8.97867605e-02 7.09803402e-01 7.95305744e-02
9.35160875e-01 -1.02765180e-01 1.07319057e+00 2.87441224e-01
4.22708541e-01 1.53642371e-01 7.09649742e-01 -7.11908266e-02
-3.82237822e-01 -6.19508207e-01 -8.42666849e-02 7.61790037e-01
3.64393950e-01 -1.06650198e+00 -4.49287564e-01 -1.28319836e+00
-1.94029376e-01 6.30701125e-01 -4.37075704e-01 3.11238710e-02
-5.76106906e-01 -1.06813323e+00 4.64439958e-01 2.56971031e-01
4.24945384e-01 -7.20357597e-01 1.26823828e-01 4.45719004e-01
-6.24631941e-01 -1.69505560e+00 -2.09719300e-01 3.76259685e-01
-6.19246185e-01 -1.21974003e+00 -2.71816701e-01 -8.07965755e-01
6.76438272e-01 3.28381471e-02 1.40831482e+00 2.29444504e-01
2.01777115e-01 -3.60645771e-01 -7.26369143e-01 1.30526992e-02
-1.75834611e-01 7.33166516e-01 -3.10801983e-01 -2.83548739e-02
7.54395187e-01 -5.44861555e-01 -2.00015321e-01 2.55082339e-01
-5.71171761e-01 9.64885578e-02 8.81026268e-01 6.86854661e-01
6.78853810e-01 4.69365269e-01 7.57080495e-01 -1.66235769e+00
6.69856369e-01 -6.54883325e-01 -4.68974143e-01 9.02834415e-01
-1.05950463e+00 2.04059124e-01 5.13858020e-01 -2.02096328e-01
-1.25003660e+00 1.34971425e-01 -5.56727462e-02 3.33136231e-01
-9.00859833e-02 1.06238317e+00 -8.05263996e-01 3.41157883e-01
2.13732719e-01 -9.82746109e-02 -9.28109825e-01 -4.70307916e-01
4.48026180e-01 8.56875658e-01 5.31504750e-01 -7.41680086e-01
8.84938896e-01 -1.08023725e-01 1.17679350e-01 -2.35230505e-01
-1.29312491e+00 -8.81045043e-01 -9.60753679e-01 2.22442642e-01
7.16580808e-01 -1.00844586e+00 -6.81399286e-01 1.76084593e-01
-1.25510085e+00 3.20057809e-01 9.58060920e-02 5.35133064e-01
1.08643621e-01 2.54241556e-01 -7.76482999e-01 -7.12917566e-01
-4.19491172e-01 -9.03961182e-01 1.01595879e+00 4.82166976e-01
-8.05297419e-02 -9.62457478e-01 3.05244565e-01 5.29916346e-01
-3.71554166e-01 1.26480520e-01 9.90573585e-01 -1.06399763e+00
-6.49474382e-01 -2.22089529e-01 -5.61236441e-01 -7.92270452e-02
4.10174280e-01 5.20574860e-04 -7.29053319e-01 2.28581592e-01
-5.65562904e-01 -7.27205947e-02 9.34920490e-01 -2.05844209e-01
7.50435591e-01 -3.90909433e-01 -8.22432041e-01 3.75880569e-01
1.29923403e+00 2.93552428e-01 7.45853841e-01 4.60239619e-01
9.33907330e-01 6.87255800e-01 1.10327339e+00 1.28270403e-01
9.51164424e-01 9.29644108e-01 -4.07063849e-02 -8.18097219e-02
1.07062489e-01 -2.77544767e-01 -8.22149590e-02 9.93400633e-01
-2.50887245e-01 -3.94020289e-01 -8.92974079e-01 6.16048515e-01
-2.01537323e+00 -6.34973943e-01 -5.61833858e-01 1.75837648e+00
1.27557886e+00 1.98745802e-01 -7.38971755e-02 2.00560808e-01
8.20743263e-01 -2.80651450e-03 -8.76401290e-02 -6.52420744e-02
-5.41358441e-02 3.17493081e-01 3.89081120e-01 4.59694773e-01
-1.39325583e+00 1.37990201e+00 4.68609428e+00 1.07452345e+00
-5.37099183e-01 4.60343659e-02 4.44070101e-01 5.97637713e-01
-2.26465300e-01 4.20677394e-01 -1.32020366e+00 1.58511791e-02
7.36445844e-01 -8.58927444e-02 -1.92027435e-01 7.21677005e-01
-2.42890403e-01 -1.16751216e-01 -9.96963978e-01 7.37173796e-01
-9.79254544e-02 -1.11476171e+00 -1.49884462e-01 7.26890638e-02
7.90724456e-01 -4.62805182e-01 -5.33753097e-01 4.17441219e-01
5.21834552e-01 -8.44510555e-01 6.37094155e-02 4.36977029e-01
5.00188410e-01 -1.07128656e+00 1.11070728e+00 2.71815330e-01
-1.85190737e+00 2.73387074e-01 -2.22595677e-01 1.83890820e-01
2.60993391e-01 1.09287822e+00 -9.42754447e-01 1.35003340e+00
3.41383964e-01 7.49793828e-01 -5.73884308e-01 8.18932831e-01
-6.80320442e-01 4.55860049e-01 -2.68422365e-02 1.29456064e-02
-5.32968342e-02 -2.61321723e-01 2.83171386e-01 1.50833166e+00
1.44566568e-02 5.98275840e-01 2.13434294e-01 5.84833801e-01
-4.32784885e-01 5.58247745e-01 -9.55802798e-02 -6.15330227e-02
7.86471665e-01 1.82622063e+00 -8.91151369e-01 -3.29758495e-01
-4.99976069e-01 8.19699705e-01 7.97244310e-01 1.50206074e-01
-6.67183578e-01 -6.38897479e-01 4.83817369e-01 -2.19361499e-01
1.92743018e-01 -9.87852439e-02 -1.70547411e-01 -1.41500807e+00
9.02494416e-02 -7.72518456e-01 6.24156475e-01 -2.85883456e-01
-1.20250607e+00 7.80730724e-01 3.37050594e-02 -1.16481543e+00
-1.62950605e-01 -3.30840260e-01 -1.95712224e-01 7.70746529e-01
-1.72660124e+00 -1.33151674e+00 -6.71285093e-02 1.73939511e-01
1.01534069e-01 -3.59350555e-02 8.95923197e-01 6.84536457e-01
-1.04072666e+00 7.74494171e-01 -4.04549986e-01 5.33483505e-01
6.61528170e-01 -1.32041526e+00 3.78884405e-01 8.54934394e-01
5.50256312e-01 8.42066586e-01 4.07284737e-01 -8.91017735e-01
-7.89741993e-01 -1.18636131e+00 1.66685688e+00 -2.77528048e-01
5.26596010e-01 -2.40326524e-01 -9.76068676e-01 8.45710576e-01
3.70713510e-02 -1.26853943e-01 9.40387905e-01 8.22790563e-01
-5.05726874e-01 -1.34741142e-01 -9.60198939e-01 4.41854388e-01
1.01970589e+00 -3.99250031e-01 -8.06786299e-01 2.67534524e-01
6.48555100e-01 -5.53409159e-01 -1.28962028e+00 8.95111382e-01
3.06368798e-01 -5.42778611e-01 9.46821988e-01 -5.84319949e-01
5.53019047e-01 -5.17615736e-01 1.40723571e-01 -1.31504118e+00
-3.70984226e-01 -2.47498274e-01 -7.83887863e-01 2.07988238e+00
9.28348005e-01 -5.86371422e-01 5.70202053e-01 7.24101365e-01
2.28857994e-01 -1.01962364e+00 -4.74606127e-01 -7.77704477e-01
-3.17613691e-01 -1.54962972e-01 9.48326766e-01 1.30039752e+00
3.21966797e-01 1.04923022e+00 -2.74540812e-01 6.59636796e-01
2.81108230e-01 4.09440517e-01 6.39688730e-01 -1.48993552e+00
-3.10546786e-01 -2.02137664e-01 -3.31274122e-01 -1.16012466e+00
5.79009414e-01 -1.15211523e+00 -2.02014465e-02 -1.87146080e+00
5.40014982e-01 -9.46552277e-01 -4.31236416e-01 7.04586208e-01
-8.74037921e-01 -9.72245112e-02 -1.45890892e-01 2.40362242e-01
-7.52407610e-01 5.03426075e-01 9.83272791e-01 -1.65797815e-01
-3.11089188e-01 1.21500075e-01 -9.70690727e-01 6.63871229e-01
5.53988636e-01 -8.22509587e-01 -3.72283489e-01 4.62502204e-02
5.89792430e-01 -1.67079091e-01 -1.34976506e-01 -6.31843507e-01
3.97123337e-01 -3.05212945e-01 -6.20575659e-02 -7.56149054e-01
4.73731617e-03 -8.32903862e-01 1.32450745e-01 -1.14564933e-01
-3.57264429e-01 -2.96718359e-01 -1.61339939e-01 2.10599735e-01
-4.94030863e-01 -4.44505125e-01 3.82846832e-01 1.21730246e-01
-6.27152085e-01 3.79237145e-01 1.23810478e-01 1.78549543e-01
9.03514564e-01 4.12375540e-01 -5.04159927e-01 9.29651037e-02
-6.43640637e-01 4.68575686e-01 -1.29346505e-01 3.62660497e-01
3.61960649e-01 -1.39226151e+00 -9.35365498e-01 -4.36516374e-01
3.63151222e-01 2.03129247e-01 -5.01921168e-03 6.96210384e-01
-1.05897464e-01 5.55822194e-01 4.47209984e-01 -4.94043417e-02
-1.59191501e+00 4.71428722e-01 2.12174356e-01 -1.34716570e+00
-3.19458276e-01 1.08750582e+00 -4.78175581e-02 -7.62057960e-01
-2.09810808e-01 -1.82654142e-01 -8.52658927e-01 3.33033532e-01
4.42174524e-01 1.68268487e-01 4.26690251e-01 -8.41341794e-01
-6.28240168e-01 5.31786740e-01 -4.76444244e-01 2.01191783e-01
1.26083624e+00 -1.64028361e-01 -6.39775693e-01 2.18468666e-01
9.41988468e-01 3.13079268e-01 -5.39591134e-01 -5.76948464e-01
6.10325634e-01 -2.52616912e-01 -1.06247842e-01 -7.76476204e-01
-1.21108115e+00 2.99227387e-01 -8.43390003e-02 3.21582109e-01
1.14894581e+00 3.67799997e-01 9.69557762e-01 4.77715015e-01
3.24667901e-01 -9.04983819e-01 -4.26881522e-01 6.81690872e-01
4.07274932e-01 -1.17363989e+00 3.98804188e-01 -1.43656290e+00
-4.78356510e-01 8.33774865e-01 8.96587670e-01 2.78451353e-01
6.61937892e-01 4.63997692e-01 -2.66663674e-02 -3.53120327e-01
-7.46273994e-01 -7.10803509e-01 8.53819907e-01 3.46732855e-01
9.19366658e-01 1.95110664e-01 -8.72541428e-01 9.38937664e-01
-1.06888883e-01 -2.22382620e-01 7.96166956e-02 6.91872299e-01
-2.06848785e-01 -1.77508163e+00 2.60076788e-03 5.31992018e-01
-7.04209030e-01 -4.51022148e-01 -5.59647918e-01 5.23779035e-01
4.47491974e-01 1.30051434e+00 -3.54231447e-01 -7.70060956e-01
2.76475936e-01 1.21842854e-01 3.43965530e-01 -9.88874912e-01
-7.63888836e-01 -5.99486530e-02 6.76821709e-01 -1.25052348e-01
-7.25939453e-01 -4.39591318e-01 -1.50207770e+00 4.75648791e-02
-1.11499035e+00 6.03472710e-01 1.53046116e-01 1.36246097e+00
4.68918681e-01 8.00426662e-01 6.67766035e-01 -6.73546121e-02
1.42347679e-01 -1.09119773e+00 -6.96503460e-01 2.61701584e-01
-1.77204713e-01 -7.99031198e-01 5.61476760e-02 -1.19763613e-03] | [9.26065444946289, 8.632157325744629] |
0017d481-e63c-4d56-bc52-f699c55eb64a | enhancing-neural-rendering-methods-with-image | 2306.08904 | null | https://arxiv.org/abs/2306.08904v1 | https://arxiv.org/pdf/2306.08904v1.pdf | Enhancing Neural Rendering Methods with Image Augmentations | Faithfully reconstructing 3D geometry and generating novel views of scenes are critical tasks in 3D computer vision. Despite the widespread use of image augmentations across computer vision applications, their potential remains underexplored when learning neural rendering methods (NRMs) for 3D scenes. This paper presents a comprehensive analysis of the use of image augmentations in NRMs, where we explore different augmentation strategies. We found that introducing image augmentations during training presents challenges such as geometric and photometric inconsistencies for learning NRMs from images. Specifically, geometric inconsistencies arise from alterations in shapes, positions, and orientations from the augmentations, disrupting spatial cues necessary for accurate 3D reconstruction. On the other hand, photometric inconsistencies arise from changes in pixel intensities introduced by the augmentations, affecting the ability to capture the underlying 3D structures of the scene. We alleviate these issues by focusing on color manipulations and introducing learnable appearance embeddings that allow NRMs to explain away photometric variations. Our experiments demonstrate the benefits of incorporating augmentations when learning NRMs, including improved photometric quality and surface reconstruction, as well as enhanced robustness against data quality issues, such as reduced training data and image degradations. | ['Bernard Ghanem', 'Jesus Zarzar', 'Sara Rojas', 'Juan C. Pérez'] | 2023-06-15 | null | null | null | null | ['neural-rendering', '3d-reconstruction'] | ['computer-vision', 'computer-vision'] | [ 6.20760858e-01 3.06809068e-01 4.78475779e-01 -3.49589854e-01
-4.86218184e-01 -6.04110897e-01 6.65448189e-01 -1.50664732e-01
-1.12373568e-01 2.40763113e-01 2.61677414e-01 -1.42459080e-01
2.08204135e-01 -6.56939566e-01 -1.23709249e+00 -5.73366582e-01
3.91924903e-02 7.47581050e-02 -6.30896119e-03 -2.85989463e-01
2.93032676e-01 1.08183408e+00 -1.92500424e+00 7.43097812e-02
7.32532501e-01 7.54477859e-01 2.32129078e-02 5.49096406e-01
-2.84391701e-01 3.76305372e-01 -6.17874563e-01 -1.96474954e-01
6.99587226e-01 -6.35637566e-02 -2.86555201e-01 6.43037200e-01
9.86073315e-01 -7.39164650e-01 -4.11391646e-01 8.42882156e-01
3.72805029e-01 2.61480123e-01 6.74877822e-01 -1.21723628e+00
-1.03206074e+00 -4.45447415e-01 -7.09898353e-01 -2.06351563e-01
2.97495753e-01 3.37520868e-01 7.35995829e-01 -1.09108782e+00
6.60514355e-01 1.39581919e+00 6.47754788e-01 6.46976292e-01
-1.61039519e+00 -2.78141499e-01 4.44861680e-01 -9.55179483e-02
-1.04428828e+00 -6.36982083e-01 1.01573062e+00 -5.72511315e-01
8.59162152e-01 3.63637000e-01 6.57790184e-01 9.23557103e-01
-1.24229677e-03 5.63837171e-01 1.10708487e+00 -5.14987946e-01
1.49731994e-01 2.33787656e-01 -3.11497152e-01 6.23146594e-01
3.50418538e-01 3.21566135e-01 -5.65440059e-01 2.23377105e-02
1.40182471e+00 -5.41878529e-02 -4.61841911e-01 -7.76140690e-01
-9.09839153e-01 4.95573997e-01 6.08802974e-01 -2.47374445e-01
-3.42980683e-01 2.78725829e-02 -8.24390575e-02 -3.82352597e-03
7.56245434e-01 6.65822625e-01 -3.57353628e-01 3.08699518e-01
-3.64708245e-01 2.20995516e-01 1.77516699e-01 9.04267788e-01
9.76462007e-01 5.24781108e-01 -2.68097855e-02 8.65896106e-01
1.36074439e-01 5.21542013e-01 -5.31735306e-04 -1.37111926e+00
2.71522403e-01 6.81651652e-01 1.37365773e-01 -1.14470029e+00
-2.69570768e-01 -3.21730971e-01 -6.65641844e-01 9.14745986e-01
2.13549063e-01 1.92681968e-01 -1.15291977e+00 1.86025774e+00
5.18231750e-01 1.10949732e-01 -6.62851930e-02 1.03813207e+00
8.26804936e-01 4.97523040e-01 -2.85284787e-01 1.14114314e-01
8.90579164e-01 -5.65056324e-01 -5.69416046e-01 -4.13932890e-01
2.44793177e-01 -9.73074794e-01 1.32589269e+00 1.92547902e-01
-1.33734131e+00 -5.77250957e-01 -1.02276433e+00 -4.67321873e-01
-1.47144243e-01 -1.63050890e-01 5.19780695e-01 6.21208608e-01
-1.17806554e+00 5.36941051e-01 -6.50424540e-01 -9.72784087e-02
4.74914819e-01 1.28972948e-01 -4.33344036e-01 -4.38423127e-01
-5.65019071e-01 9.24750447e-01 -2.76158363e-01 2.24041045e-01
-5.71243763e-01 -1.02118945e+00 -1.10502124e+00 -1.72665954e-01
2.21228808e-01 -7.16624856e-01 9.49728608e-01 -9.85269010e-01
-1.36204362e+00 1.00059533e+00 -1.02287948e-01 8.97648856e-02
3.79106879e-01 -2.43366823e-01 -2.87544616e-02 -8.91441200e-03
-1.32244572e-01 8.71128440e-01 9.43427026e-01 -1.77916944e+00
-9.35080498e-02 -5.66575825e-01 1.82389185e-01 6.46486700e-01
-1.82342112e-01 -5.91404974e-01 -5.77575624e-01 -7.17502356e-01
3.74770910e-01 -8.95981669e-01 -3.91666085e-01 6.31796300e-01
-2.87578970e-01 4.32587177e-01 8.50528598e-01 -6.82155550e-01
2.57230967e-01 -2.28871703e+00 1.86432764e-01 1.70347959e-01
1.66256025e-01 1.01188511e-01 -5.58168054e-01 1.52859598e-01
-1.11534461e-01 8.63531455e-02 -2.07687348e-01 -6.32875919e-01
-1.87337831e-01 4.63529944e-01 -2.80151367e-01 4.19416487e-01
6.77614212e-01 6.81462646e-01 -5.56287289e-01 1.35036051e-01
5.84837854e-01 1.06432188e+00 -6.83885276e-01 4.09402341e-01
-4.35206950e-01 7.16104269e-01 8.33095703e-03 4.80423898e-01
9.14597571e-01 3.95922596e-03 -2.40621895e-01 -4.10593212e-01
-1.03756547e-01 3.54375809e-01 -1.11406410e+00 1.67264247e+00
-5.73657513e-01 7.90651917e-01 1.76857650e-01 -6.69708312e-01
8.90765488e-01 -3.81540917e-02 2.78488010e-01 -8.99424911e-01
-7.09543154e-02 -4.04791050e-02 -2.23470494e-01 -3.67701441e-01
6.17142260e-01 -9.16076824e-02 5.49583137e-01 4.61340398e-01
-3.47305566e-01 -6.28108263e-01 -3.50461990e-01 1.18332446e-01
6.59972906e-01 4.88960177e-01 -2.31720269e-01 6.90945089e-02
4.93952148e-02 4.09189286e-03 3.80731732e-01 4.64166671e-01
3.21565956e-01 1.02961147e+00 1.43758982e-01 -4.76690590e-01
-1.40550256e+00 -1.27010262e+00 4.75452095e-03 6.76211953e-01
8.60600471e-02 -4.47285511e-02 -4.52208370e-01 -1.74596027e-01
1.86133478e-02 6.53102458e-01 -6.86539531e-01 -2.41191104e-01
-5.96094668e-01 -5.41274011e-01 7.59094432e-02 4.88340437e-01
3.43530446e-01 -6.54452324e-01 -7.35326827e-01 -2.63649393e-02
-6.90698996e-02 -1.32788455e+00 -4.43056434e-01 -1.12261973e-01
-1.09755266e+00 -1.19224632e+00 -6.67924881e-01 -5.41137218e-01
1.10038483e+00 8.09637427e-01 1.07793987e+00 1.59337401e-01
-7.28532791e-01 8.26457560e-01 -8.44174474e-02 -4.94928837e-01
-3.97406280e-01 -5.92369497e-01 -1.16894595e-01 -5.51806018e-02
-2.88934708e-01 -7.42365301e-01 -6.69767201e-01 2.38552436e-01
-1.08646488e+00 4.57659423e-01 5.31058013e-01 5.43255866e-01
6.34285867e-01 -2.70622402e-01 1.07765337e-02 -8.35158706e-01
3.71584266e-01 1.05124205e-01 -7.14935482e-01 -5.46844974e-02
-4.32801545e-01 2.07332239e-01 3.64433855e-01 -3.84107471e-01
-1.26024532e+00 -8.63557309e-03 -1.83515146e-01 -7.24080980e-01
-3.00157458e-01 -1.98056381e-02 -9.77219045e-02 -2.44439140e-01
7.86312759e-01 3.97632085e-02 3.62280190e-01 -6.02051735e-01
4.44611758e-01 7.85417259e-02 5.25574982e-01 -5.31371951e-01
1.03708005e+00 7.54793525e-01 1.60585508e-01 -1.26579881e+00
-6.71896160e-01 -8.45390484e-02 -5.10008216e-01 -2.08139405e-01
6.09756470e-01 -9.92176831e-01 -2.62023807e-01 5.26991367e-01
-1.21143985e+00 -3.80814373e-01 -4.82259810e-01 2.59178489e-01
-3.81339282e-01 4.09184098e-01 -4.22207475e-01 -7.29421496e-01
-1.17406666e-01 -1.18604863e+00 1.08025801e+00 7.25837871e-02
-3.33845872e-03 -8.93277586e-01 -1.08230606e-01 4.31150168e-01
3.84744406e-01 6.32137716e-01 1.32624042e+00 2.88405925e-01
-9.37650561e-01 7.14436099e-02 -3.27069610e-01 5.05867898e-01
4.83203232e-01 1.18152857e-01 -1.43280983e+00 -1.81477815e-01
-1.43183991e-01 -2.19028853e-02 5.09254694e-01 5.17571628e-01
1.19539440e+00 -2.61793345e-01 1.94500253e-01 8.16350281e-01
1.34777689e+00 6.04438931e-02 6.79317236e-01 2.78755873e-01
9.79471326e-01 9.49200749e-01 1.90525249e-01 2.65195459e-01
2.77143180e-01 6.70048952e-01 6.85928404e-01 -6.48231804e-01
-5.87212861e-01 -2.54347444e-01 1.25313908e-01 2.70438433e-01
-9.39360708e-02 8.15771967e-02 -6.26018763e-01 2.40876511e-01
-1.39406180e+00 -3.78493339e-01 -1.23667307e-01 2.43920898e+00
6.29382849e-01 1.20064076e-02 -2.11126417e-01 1.28182948e-01
5.19940257e-01 1.09490603e-01 -8.82541358e-01 -3.73715490e-01
-3.84010464e-01 2.59873927e-01 3.06834370e-01 6.75242722e-01
-5.52464724e-01 7.01451540e-01 6.11003160e+00 6.77439272e-02
-1.05589914e+00 -4.07817096e-01 7.46302068e-01 -1.75371602e-01
-8.86215925e-01 -1.94626689e-01 -2.55048335e-01 -1.65092409e-01
4.39001732e-02 4.00236219e-01 6.53089166e-01 5.09898484e-01
2.44791895e-01 -2.34260727e-02 -1.12135112e+00 1.10477424e+00
3.72513592e-01 -1.42487252e+00 4.28862602e-01 1.19615160e-01
9.36215520e-01 5.64479828e-03 4.83385980e-01 -3.01307619e-01
2.06494957e-01 -1.01213694e+00 8.44365180e-01 3.71093541e-01
8.83001328e-01 -6.07365012e-01 2.92248040e-01 7.50340894e-02
-6.74296498e-01 1.81054369e-01 -2.46455058e-01 -2.45231554e-01
1.13112122e-01 6.51686668e-01 -7.37931192e-01 2.95662791e-01
6.98746443e-01 5.91134071e-01 -7.17357337e-01 7.99316227e-01
-3.02841842e-01 -1.26920305e-02 -2.97392637e-01 3.95672083e-01
4.19934690e-02 -4.49297369e-01 5.06898582e-01 5.68073273e-01
2.26584017e-01 1.39640704e-01 -1.35798156e-01 1.10231805e+00
-4.69232723e-02 -1.30027443e-01 -9.05041516e-01 1.45299494e-01
2.50210613e-01 9.31817651e-01 -4.97439772e-01 1.20443583e-01
-4.63327259e-01 1.04633081e+00 3.47632170e-01 7.19523072e-01
-3.68926555e-01 1.55741334e-01 1.10920668e+00 4.61090922e-01
9.10470411e-02 -5.34298360e-01 -7.09501207e-01 -1.02345729e+00
2.67501354e-01 -7.46932089e-01 -5.36785610e-02 -1.19817054e+00
-1.14179301e+00 3.74571770e-01 -2.00937539e-02 -1.07416606e+00
8.16645026e-02 -7.13862121e-01 -5.32831848e-01 9.17127013e-01
-1.76633143e+00 -9.74900663e-01 -5.23805797e-01 4.45214182e-01
5.82383454e-01 2.55725354e-01 7.98369050e-01 9.91849750e-02
-2.97706902e-01 3.12550247e-01 -1.56878561e-01 -2.88548380e-01
5.98968625e-01 -1.06639695e+00 8.06531608e-01 8.69270980e-01
2.16016889e-01 5.82918346e-01 6.35593534e-01 -2.94966042e-01
-1.59163594e+00 -1.01443541e+00 1.96779504e-01 -4.47662979e-01
-1.20110042e-01 -4.95728970e-01 -1.11105931e+00 5.95372736e-01
-1.88425966e-02 -7.95762390e-02 5.90972006e-01 1.66662913e-02
-7.60997236e-01 1.06867529e-01 -1.19041264e+00 1.00680590e+00
1.21322072e+00 -6.29503071e-01 -1.12715170e-01 1.03254080e-01
6.92550242e-01 -6.30836964e-01 -4.79865015e-01 4.55338299e-01
5.43961346e-01 -1.27164364e+00 1.32031631e+00 -3.96289140e-01
4.71243739e-01 -3.46545756e-01 -2.39636689e-01 -1.50679982e+00
-1.89122915e-01 -3.07643652e-01 6.97982982e-02 1.02467763e+00
2.25931227e-01 -5.64288735e-01 7.11913168e-01 1.03323269e+00
-4.24968928e-01 -5.07136464e-01 -6.52223170e-01 -5.23216844e-01
1.42822027e-01 -4.59962517e-01 5.11192203e-01 9.97065663e-01
-7.68451214e-01 1.68714330e-01 -3.16827238e-01 5.17168939e-01
7.20509052e-01 5.46992235e-02 1.06026328e+00 -1.10450077e+00
-1.65822685e-01 -3.42271745e-01 -3.24887723e-01 -1.16866148e+00
-1.56045020e-01 -4.85379159e-01 1.19557306e-01 -1.60614502e+00
-1.74680054e-01 -6.38829112e-01 1.93153158e-01 2.84653515e-01
-1.05989352e-01 6.07062280e-01 2.96735883e-01 8.86081904e-02
2.00815588e-01 6.97548568e-01 1.51370692e+00 3.58661413e-02
-2.97137976e-01 -1.95825741e-01 -7.07518756e-01 8.32942247e-01
7.73941636e-01 -1.52541790e-02 -6.94653213e-01 -1.12623727e+00
3.11196774e-01 -2.83996314e-01 6.39979780e-01 -6.90809131e-01
-2.83092380e-01 -1.05721049e-01 8.53161871e-01 -4.91148323e-01
7.16742516e-01 -9.10638273e-01 2.49703035e-01 1.19410329e-01
-2.86219001e-01 1.78154215e-01 5.85717022e-01 6.58138573e-01
2.46047229e-01 7.11343959e-02 8.33178282e-01 -2.88784564e-01
-5.96254706e-01 3.17745447e-01 -2.44287148e-01 -4.29516993e-02
6.88471258e-01 -5.55569410e-01 -1.17220014e-01 -4.48631316e-01
-5.39901972e-01 -7.15202391e-02 9.97233212e-01 6.51005626e-01
9.54657912e-01 -1.26499093e+00 -4.82433558e-01 7.42975056e-01
2.44035199e-01 5.14652312e-01 4.25362349e-01 2.77777314e-01
-6.27601147e-01 -1.48234650e-01 -3.60924214e-01 -7.27729261e-01
-1.27010214e+00 2.71712542e-01 3.49757642e-01 4.32263583e-01
-9.21805918e-01 7.09677517e-01 5.85344017e-01 -6.18128717e-01
5.42942405e-01 -4.44855422e-01 1.56932324e-01 -4.67800826e-01
3.04806054e-01 2.36403912e-01 2.90911973e-01 -4.59451288e-01
-5.40494211e-02 8.22354317e-01 2.91685313e-02 -1.51305556e-01
1.43151474e+00 -3.24281096e-01 2.20460877e-01 3.59039068e-01
1.08542478e+00 -1.07923865e-01 -1.74300826e+00 -3.37761551e-01
-5.46774268e-01 -8.72457325e-01 2.49304965e-01 -7.05524087e-01
-1.09619343e+00 1.04437816e+00 6.85315967e-01 -1.94233716e-01
1.07622015e+00 -1.65501788e-01 6.79545820e-01 2.24920601e-01
1.45146221e-01 -8.66585195e-01 4.00470823e-01 3.25750858e-01
1.11604965e+00 -1.21745396e+00 7.58342817e-02 -6.38657451e-01
-5.01758456e-01 1.07454395e+00 7.25409091e-01 -6.28936663e-02
3.23980063e-01 9.77657065e-02 2.99292654e-01 -2.08621383e-01
-3.90781015e-01 -1.07065342e-01 4.43604946e-01 1.04579341e+00
2.58236736e-01 -2.64989734e-01 4.55612689e-01 -3.04992676e-01
-9.23303664e-02 -7.37084925e-01 7.34043598e-01 9.51541483e-01
-6.88604116e-02 -9.07863617e-01 -4.79984224e-01 2.30107114e-01
1.06925420e-01 -1.63635463e-02 -5.02310455e-01 7.27380514e-01
4.60245237e-02 5.78883529e-01 3.35409880e-01 -1.16828218e-01
6.66536748e-01 -1.92765787e-01 9.30472434e-01 -7.85996437e-01
-1.91676095e-02 1.03551142e-01 -3.69069725e-02 -5.13653159e-01
-5.11578918e-01 -6.02837861e-01 -1.08505750e+00 -1.05702743e-01
-1.81075454e-01 -5.08383512e-01 9.32166159e-01 6.82623327e-01
4.82558548e-01 5.26614964e-01 6.91235960e-01 -1.20239258e+00
-1.79833561e-01 -5.16895592e-01 -4.66880113e-01 6.46329582e-01
6.94237292e-01 -8.27355206e-01 -4.93830323e-01 1.48075283e-01] | [9.313700675964355, -3.0905582904815674] |
3b1b1247-13f5-4443-b4c8-98dd303598b1 | wavlm-large-scale-self-supervised-pre | 2110.13900 | null | https://arxiv.org/abs/2110.13900v5 | https://arxiv.org/pdf/2110.13900v5.pdf | WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing | Self-supervised learning (SSL) achieves great success in speech recognition, while limited exploration has been attempted for other speech processing tasks. As speech signal contains multi-faceted information including speaker identity, paralinguistics, spoken content, etc., learning universal representations for all speech tasks is challenging. To tackle the problem, we propose a new pre-trained model, WavLM, to solve full-stack downstream speech tasks. WavLM jointly learns masked speech prediction and denoising in pre-training. By this means, WavLM does not only keep the speech content modeling capability by the masked speech prediction, but also improves the potential to non-ASR tasks by the speech denoising. In addition, WavLM employs gated relative position bias for the Transformer structure to better capture the sequence ordering of input speech. We also scale up the training dataset from 60k hours to 94k hours. WavLM Large achieves state-of-the-art performance on the SUPERB benchmark, and brings significant improvements for various speech processing tasks on their representative benchmarks. The code and pre-trained models are available at https://aka.ms/wavlm. | ['Jian Wu', 'Xiangzhan Yu', 'Michael Zeng', 'Furu Wei', 'Yao Qian', 'Yanmin Qian', 'Shuo Ren', 'Long Zhou', 'Xiong Xiao', 'Takuya Yoshioka', 'Naoyuki Kanda', 'Jinyu Li', 'Zhuo Chen', 'Shujie Liu', 'Yu Wu', 'Zhengyang Chen', 'Chengyi Wang', 'Sanyuan Chen'] | 2021-10-26 | null | null | null | null | ['speech-denoising'] | ['speech'] | [ 1.82988331e-01 1.55969873e-01 -1.84975818e-01 -6.76531792e-01
-1.15925300e+00 -4.25817639e-01 5.25624812e-01 -1.61949888e-01
-2.34577984e-01 3.30885261e-01 7.21902311e-01 -7.10295022e-01
2.32072324e-01 -8.46954733e-02 -5.96506894e-01 -6.47857070e-01
-8.19413457e-03 3.52050513e-01 6.82042316e-02 -3.35940659e-01
-1.67237222e-01 1.54819742e-01 -1.33258331e+00 6.99216604e-01
5.48780382e-01 8.99484277e-01 5.20040393e-01 8.95076752e-01
-2.21387044e-01 9.67554986e-01 -6.35710835e-01 -1.80116087e-01
7.89509267e-02 -2.84644276e-01 -9.58827436e-01 1.04422458e-01
3.28460246e-01 -2.02159971e-01 -7.98504531e-01 9.01237428e-01
8.01978827e-01 2.33784690e-01 3.00214142e-01 -1.09078288e+00
-5.87914765e-01 1.22094738e+00 -2.24119797e-01 5.42120099e-01
1.09790422e-01 2.25455746e-01 1.23318136e+00 -1.20570445e+00
6.95845038e-02 1.38354790e+00 3.57318699e-01 8.65836263e-01
-9.08417881e-01 -7.00743735e-01 3.10916752e-01 4.30999607e-01
-1.21162188e+00 -1.39025247e+00 6.21161103e-01 1.70372240e-02
1.24767947e+00 5.36211967e-01 1.37975171e-01 1.38609946e+00
-2.74850518e-01 1.24501193e+00 8.90972793e-01 -3.64668518e-01
1.99773729e-01 -1.55747324e-01 3.85437906e-01 5.57048857e-01
-4.95061755e-01 4.30393256e-02 -1.11921537e+00 5.03552742e-02
3.06148648e-01 -3.09089690e-01 -3.90142560e-01 2.39726394e-01
-1.22448123e+00 5.71347654e-01 -6.38463302e-03 2.42169693e-01
4.04896960e-02 -7.24441931e-02 6.49438679e-01 6.21194482e-01
6.69649005e-01 1.60053864e-01 -8.09875250e-01 -4.01320368e-01
-1.09220862e+00 -2.21348122e-01 7.39873767e-01 1.02719915e+00
5.61422527e-01 5.31848073e-01 -3.14175814e-01 1.30810618e+00
3.23297858e-01 4.00673509e-01 9.39626515e-01 -7.78888166e-01
9.19125915e-01 4.13112417e-02 -5.85400820e-01 -2.51509041e-01
-1.89613596e-01 -6.14694655e-01 -1.17162371e+00 -3.27748835e-01
3.18398625e-02 -1.35149896e-01 -1.12537384e+00 1.60402751e+00
1.86340332e-01 4.66883212e-01 4.30224478e-01 6.95927441e-01
1.23858106e+00 1.36568224e+00 -2.40013540e-01 -3.93614680e-01
1.23240232e+00 -1.54312599e+00 -8.47398221e-01 -6.90982938e-01
5.73877454e-01 -8.54322433e-01 1.23259437e+00 3.03964108e-01
-1.16805673e+00 -6.65930569e-01 -9.26105499e-01 -1.47985920e-01
-1.06097072e-01 2.27759391e-01 3.67252976e-01 6.48531199e-01
-1.27040040e+00 1.98729545e-01 -8.56102109e-01 -9.52015594e-02
3.69262099e-01 1.84651703e-01 -3.19071949e-01 -1.27206981e-01
-1.23426747e+00 5.96975923e-01 2.51928180e-01 1.12118833e-01
-1.16541183e+00 -7.33752310e-01 -9.78639424e-01 2.38967761e-01
3.99141341e-01 -1.00994252e-01 1.59647167e+00 -6.17318869e-01
-1.89554727e+00 6.27728581e-01 -6.52845919e-01 -7.39433944e-01
2.36677229e-01 -1.82562843e-01 -7.46400058e-01 -1.18599810e-01
-1.43727615e-01 5.39626300e-01 1.20409572e+00 -9.32463944e-01
-4.98737454e-01 -2.32495964e-01 -4.91184056e-01 1.91849276e-01
-5.08175910e-01 3.02658081e-01 -6.05624139e-01 -9.58376169e-01
2.56450921e-01 -7.46355951e-01 -1.35388300e-01 -6.67050779e-01
-5.05971074e-01 -5.69165111e-01 9.40207481e-01 -9.14432704e-01
1.37972224e+00 -2.53004336e+00 3.71923260e-02 -7.49791637e-02
-2.01890972e-02 7.00866640e-01 -4.89682317e-01 5.87044418e-01
-1.10272773e-01 -2.68926118e-02 -2.56364882e-01 -1.16765535e+00
7.22310692e-02 4.32793438e-01 -7.97036409e-01 2.69334793e-01
1.19984850e-01 8.83930445e-01 -6.30670667e-01 -2.74972707e-01
1.28433347e-01 4.08245921e-01 -3.43064636e-01 3.95055771e-01
-1.20672084e-01 4.56370711e-01 2.00355440e-01 5.38087606e-01
5.74665368e-01 -1.41316801e-01 2.05770433e-01 9.77647752e-02
2.91053206e-03 1.16160560e+00 -9.87156272e-01 1.85121369e+00
-5.82835197e-01 7.24153519e-01 3.37561995e-01 -1.13774300e+00
8.58867645e-01 6.87393963e-01 1.02272637e-01 -7.06633270e-01
-8.51101503e-02 2.14357242e-01 -1.22675300e-02 -1.94627866e-01
3.91202331e-01 9.90626812e-02 1.12086549e-01 3.58642578e-01
4.07115966e-01 -3.23563404e-02 -1.08797982e-01 3.53765279e-01
1.12180173e+00 -4.47552145e-01 1.95296139e-01 -1.49361536e-01
7.20853150e-01 -5.73772490e-01 6.70261621e-01 7.52080917e-01
-1.96150362e-01 6.00918651e-01 1.55350521e-01 -9.21950564e-02
-8.35202634e-01 -1.09788823e+00 2.93272063e-02 1.65738988e+00
-2.01898485e-01 -7.19240308e-01 -6.98808908e-01 -5.33065379e-01
-1.94117293e-01 5.35441101e-01 -9.51447636e-02 -3.34895104e-01
-7.33633697e-01 -4.74206090e-01 8.25701654e-01 4.91295815e-01
3.86625856e-01 -1.18480706e+00 3.79403591e-01 1.95688218e-01
-3.02700728e-01 -1.42533326e+00 -9.55843270e-01 4.46164191e-01
-6.07667148e-01 -3.49152625e-01 -5.65173328e-01 -1.22758114e+00
2.71778226e-01 4.67692882e-01 1.01246178e+00 2.63566971e-02
3.89067717e-02 2.35402510e-02 -6.43679321e-01 -1.39195532e-01
-8.30864489e-01 4.33183730e-01 3.74530107e-01 2.28753120e-01
-4.90509067e-03 -5.70807576e-01 -2.41623968e-01 2.77716458e-01
-5.46896517e-01 -5.51291667e-02 4.71225858e-01 9.63127792e-01
4.87816453e-01 1.47169054e-01 8.33049119e-01 -6.04173064e-01
3.97495657e-01 -4.09696758e-01 -2.72821993e-01 1.90339938e-01
-3.31753045e-01 5.53524643e-02 7.74933100e-01 -4.39005345e-01
-1.10641396e+00 -5.94679266e-02 -7.52041459e-01 -4.12374496e-01
-2.08932266e-01 3.53322983e-01 -4.88911122e-01 4.12399232e-01
2.19219208e-01 7.27233887e-01 5.68396486e-02 -8.31923902e-01
4.15815294e-01 1.16871154e+00 6.37559056e-01 -3.49442631e-01
6.05260134e-01 8.14426094e-02 -6.24352872e-01 -1.35815692e+00
-1.04135644e+00 -6.99724197e-01 -2.41063267e-01 2.54972965e-01
4.39303607e-01 -1.25163269e+00 -5.27840555e-01 6.01825595e-01
-1.25050426e+00 -5.89587510e-01 -1.77645668e-01 4.27622914e-01
-3.25762868e-01 5.16531169e-01 -9.08750653e-01 -9.55331802e-01
-6.62291348e-01 -1.32502759e+00 1.00636518e+00 -3.95032972e-01
-1.11722656e-01 -5.93824327e-01 -1.93843827e-01 7.01094210e-01
4.39631134e-01 -1.06550753e+00 7.04754055e-01 -1.00585091e+00
-5.04212797e-01 2.00440496e-01 1.05896376e-01 9.37943161e-01
2.66289681e-01 -4.83043104e-01 -1.33394623e+00 -5.93962014e-01
1.00955613e-01 -3.82982582e-01 1.22847235e+00 2.67117053e-01
1.35063791e+00 -5.42841792e-01 4.04695384e-02 8.19371045e-01
5.03101647e-01 1.85746849e-01 4.41215247e-01 -1.02136388e-01
8.11604440e-01 6.36355340e-01 2.47014806e-01 2.85391361e-01
4.07729298e-01 5.53246558e-01 1.78877607e-01 1.82382278e-02
-6.27832651e-01 -3.37163925e-01 7.96402812e-01 1.72581279e+00
5.47766745e-01 -4.72551435e-01 -8.70127738e-01 5.08778453e-01
-1.72899973e+00 -8.98358643e-01 1.37419179e-01 1.85681522e+00
9.27957594e-01 1.80877656e-01 1.36759698e-01 3.86641979e-01
7.17541277e-01 6.09339416e-01 -5.95302343e-01 -2.69921571e-01
-3.99297804e-01 3.44621062e-01 3.18512321e-01 7.10247636e-01
-1.13302028e+00 1.34845364e+00 6.23989344e+00 1.33620965e+00
-1.12715507e+00 4.39228058e-01 7.39887536e-01 -1.47183582e-01
-2.29658633e-01 -1.46198884e-01 -1.13354492e+00 5.22088289e-01
1.20207822e+00 -2.41852272e-02 7.30269074e-01 7.54095912e-01
2.60661095e-01 6.01690531e-01 -1.06974638e+00 1.28179479e+00
1.02231100e-01 -1.40463054e+00 1.43946826e-01 -1.33440837e-01
4.40157801e-01 5.76753199e-01 2.03712553e-01 5.08974135e-01
3.46569389e-01 -1.05897784e+00 8.43364000e-01 -1.12830028e-01
8.77007723e-01 -7.04274952e-01 5.10537863e-01 6.66103423e-01
-1.24512863e+00 -1.16362773e-01 -2.61839688e-01 -4.17858362e-02
2.28838861e-01 5.90677798e-01 -1.19675899e+00 3.60670358e-01
6.37540936e-01 7.23880351e-01 -3.29847157e-01 5.90948462e-01
-4.07811105e-01 1.37026906e+00 -2.59253085e-01 1.23299554e-01
2.21612170e-01 2.52984434e-01 5.50383866e-01 1.43196952e+00
1.94417208e-01 -1.42638475e-01 2.70429224e-01 1.71684906e-01
-2.71545261e-01 2.56816864e-01 -2.90471166e-01 -2.71037698e-01
8.53500545e-01 8.40460062e-01 -2.46253327e-01 -3.10841620e-01
-4.71264780e-01 1.02680171e+00 4.66887265e-01 4.23304915e-01
-5.18647730e-01 1.05239833e-02 1.18604255e+00 -6.64303079e-02
3.66404325e-01 -4.42335397e-01 -1.38993129e-01 -1.24832404e+00
-2.81670596e-03 -1.20679808e+00 3.01863998e-01 -4.98225808e-01
-1.26563656e+00 1.01798785e+00 -6.01478517e-01 -8.54570270e-01
-3.08310330e-01 -4.52858865e-01 -5.69582343e-01 8.61567676e-01
-1.57273698e+00 -1.13461006e+00 2.04781935e-01 6.34387970e-01
1.33652878e+00 -7.84886003e-01 8.13117266e-01 3.39370251e-01
-8.13484013e-01 8.80017042e-01 1.56394988e-01 3.26345384e-01
7.48492360e-01 -9.97654080e-01 1.17785895e+00 1.03247952e+00
5.44264257e-01 4.67383385e-01 4.52274770e-01 -3.64420563e-01
-1.52990425e+00 -1.19940460e+00 9.18259621e-01 -3.00971478e-01
7.50857770e-01 -7.77080476e-01 -1.08955503e+00 8.23562860e-01
3.37915689e-01 1.12140544e-01 7.24818051e-01 2.77766109e-01
-5.25360227e-01 -3.75407159e-01 -4.61010873e-01 5.31699359e-01
1.21414948e+00 -1.01121593e+00 -5.70924640e-01 2.72696465e-01
1.46530282e+00 -3.98227423e-01 -3.70388597e-01 1.87953949e-01
1.01655357e-01 -6.58293486e-01 1.03264987e+00 -7.08444178e-01
1.33928984e-01 -1.02117233e-01 -5.01124084e-01 -1.50859451e+00
-3.10191512e-01 -1.02258670e+00 -4.23406780e-01 1.60372055e+00
6.38204455e-01 -6.04908526e-01 7.49293983e-01 -5.99988475e-02
-7.35339761e-01 -7.67733514e-01 -1.28686166e+00 -1.01023829e+00
-5.27982749e-02 -9.11079288e-01 8.05911303e-01 8.20737422e-01
1.65831838e-02 6.54012561e-01 -7.96654880e-01 4.53407586e-01
5.09094238e-01 -1.08189553e-01 6.59599304e-01 -7.40819931e-01
-4.50558960e-01 -2.94256777e-01 -9.29838493e-02 -1.69420171e+00
6.79061890e-01 -1.34343207e+00 2.13056728e-01 -1.26193273e+00
-2.45022669e-01 -2.76737869e-01 -3.21664512e-01 6.24585569e-01
-1.09580308e-01 -1.25703514e-01 2.09849939e-01 1.05951980e-01
-6.29521668e-01 6.77206159e-01 9.35807526e-01 -2.86115021e-01
-2.07021520e-01 1.47740364e-01 -6.88215137e-01 5.31138599e-01
1.07878292e+00 -2.54166126e-01 -6.27345443e-01 -8.62919450e-01
-3.76548260e-01 1.64880082e-01 6.46491945e-02 -7.91860819e-01
5.77888608e-01 2.11519584e-01 -1.46863014e-01 -7.49472022e-01
7.58934677e-01 -3.87734741e-01 -3.48075718e-01 1.50940984e-01
-5.52078545e-01 -1.78717837e-01 3.11820190e-02 3.29163730e-01
-5.25730669e-01 -9.47210342e-02 6.59863412e-01 -6.77974075e-02
-8.71096730e-01 5.09979784e-01 -4.29501921e-01 2.10255414e-01
2.87335455e-01 2.10738242e-01 -4.01429176e-01 -6.26882434e-01
-7.19813228e-01 3.44073474e-01 -2.08912879e-01 6.77034140e-01
7.45219111e-01 -1.08936524e+00 -9.10008013e-01 6.82181180e-01
8.38771388e-02 1.03532657e-01 4.26989943e-01 4.61824149e-01
1.19632753e-02 4.44053620e-01 5.58265030e-01 -5.30366063e-01
-1.43148780e+00 4.23461318e-01 2.88803410e-02 -2.23376118e-02
-6.19054794e-01 1.31470478e+00 2.08929479e-01 -6.73063815e-01
7.62953222e-01 -3.85179222e-01 8.37310031e-02 -1.20583177e-01
7.98256814e-01 3.56445432e-01 3.66621763e-01 -7.80148268e-01
-5.27012527e-01 -8.47717747e-02 -3.65205824e-01 -1.85560808e-01
1.43042910e+00 -3.73134494e-01 5.99068403e-03 5.04460394e-01
1.39764893e+00 8.58707279e-02 -1.18949056e+00 -6.45931959e-01
-2.60197162e-03 -6.42359927e-02 3.59841377e-01 -6.97795808e-01
-1.02343798e+00 1.05530274e+00 1.54785529e-01 2.32623905e-01
9.77495551e-01 3.61968040e-01 1.22954094e+00 4.85584676e-01
1.58563226e-01 -1.00042200e+00 1.60114825e-01 1.19616711e+00
9.93972540e-01 -1.34881139e+00 -5.02908111e-01 -5.44634759e-01
-7.97655225e-01 7.27364957e-01 4.86886084e-01 3.24697495e-01
8.32830071e-01 5.81934988e-01 3.94462198e-01 2.99905717e-01
-1.12654448e+00 -2.56147861e-01 2.27849290e-01 4.19150800e-01
4.31018740e-01 2.15291440e-01 2.94016629e-01 8.56666625e-01
-6.49829268e-01 -5.89944482e-01 1.08165950e-01 6.26813352e-01
-5.02609372e-01 -1.23106849e+00 -3.68106425e-01 2.58171707e-01
-3.71381134e-01 -6.18929088e-01 -2.04576567e-01 -6.29320294e-02
-2.27004617e-01 1.32722318e+00 9.71445963e-02 -5.22615850e-01
2.45933965e-01 2.61748850e-01 1.12331072e-02 -7.88742125e-01
-4.13328856e-01 3.61377656e-01 1.73629254e-01 -3.21179092e-01
5.91904894e-02 -5.03609240e-01 -1.18292356e+00 -1.83991969e-01
-1.61942944e-01 3.44292134e-01 7.09387362e-01 9.02767718e-01
4.58977133e-01 6.58292413e-01 9.12013292e-01 -7.73100793e-01
-8.39574039e-01 -1.21042883e+00 -5.45978904e-01 5.57574667e-02
8.92935514e-01 -1.94193870e-01 -6.38969958e-01 -2.48334967e-02] | [14.540565490722656, 6.429957389831543] |
72aa417c-0dfc-4293-a703-e02fbb7144b8 | csagn-conversational-structure-aware-graph | 2109.11541 | null | https://arxiv.org/abs/2109.11541v2 | https://arxiv.org/pdf/2109.11541v2.pdf | CSAGN: Conversational Structure Aware Graph Network for Conversational Semantic Role Labeling | Conversational semantic role labeling (CSRL) is believed to be a crucial step towards dialogue understanding. However, it remains a major challenge for existing CSRL parser to handle conversational structural information. In this paper, we present a simple and effective architecture for CSRL which aims to address this problem. Our model is based on a conversational structure-aware graph network which explicitly encodes the speaker dependent information. We also propose a multi-task learning method to further improve the model. Experimental results on benchmark datasets show that our model with our proposed training objectives significantly outperforms previous baselines. | ['Linqi Song', 'Kun Xu', 'Han Wu'] | 2021-09-23 | null | https://aclanthology.org/2021.emnlp-main.177 | https://aclanthology.org/2021.emnlp-main.177.pdf | emnlp-2021-11 | ['semantic-role-labeling', 'dialogue-understanding'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.34811890e-01 6.07139587e-01 -2.02945963e-01 -8.05231035e-01
-8.88938963e-01 -6.86074495e-01 9.56072271e-01 -3.47211100e-02
-2.21784100e-01 7.73742557e-01 1.03234339e+00 -3.34436327e-01
3.30241233e-01 -4.35603857e-01 -2.52474636e-01 -2.92096436e-01
3.55385207e-02 4.95275080e-01 4.20135051e-01 -9.58664656e-01
3.09489548e-01 -1.16450571e-01 -9.49565887e-01 8.28387618e-01
7.38067389e-01 4.09015983e-01 3.00588846e-01 8.38815570e-01
-4.54606920e-01 1.64417911e+00 -8.54421437e-01 -2.77048349e-01
-2.83341050e-01 -8.78881156e-01 -1.65871847e+00 6.70380592e-02
3.99097353e-02 -2.08587766e-01 -2.22360313e-01 6.22283161e-01
3.31396729e-01 3.88681918e-01 1.91237137e-01 -1.19901371e+00
-2.62913495e-01 1.10146546e+00 -3.01240385e-02 2.95970380e-01
6.43631101e-01 -9.50062871e-02 1.59253860e+00 -4.08248514e-01
4.87242997e-01 1.81876719e+00 3.90346110e-01 1.10280025e+00
-8.28536451e-01 -1.80093825e-01 7.60839045e-01 3.32621634e-01
-6.52296960e-01 -5.85202515e-01 1.35002947e+00 7.59916455e-02
9.55293953e-01 2.66255975e-01 4.29878503e-01 1.29720902e+00
-2.68731534e-01 1.14147687e+00 1.14906979e+00 -5.37687778e-01
-1.25118047e-01 -1.74569458e-01 5.08028567e-01 6.99987233e-01
-3.64279002e-01 -6.03954613e-01 -8.79032731e-01 -1.29190370e-01
5.94042480e-01 -6.34187758e-01 -3.11912388e-01 -2.61130989e-01
-8.38059008e-01 1.11188161e+00 1.18194953e-01 5.48815370e-01
9.28272977e-02 1.55568078e-01 6.18428648e-01 5.01810014e-01
6.74160898e-01 5.70146918e-01 -3.93284947e-01 -5.81448972e-01
6.92727184e-03 9.33503509e-02 1.32395005e+00 5.39682388e-01
2.57064015e-01 -5.67561500e-02 -1.88389808e-01 1.40598691e+00
4.75900531e-01 -1.32500425e-01 2.66253024e-01 -1.25450563e+00
6.13436937e-01 9.00273085e-01 -1.46378756e-01 -7.80750334e-01
-5.48990726e-01 5.26021197e-02 -4.23620015e-01 -6.14554465e-01
3.93551141e-01 -2.16074614e-03 -1.42437622e-01 1.93459487e+00
3.75421882e-01 2.03283295e-01 4.78801399e-01 7.87472546e-01
1.35236108e+00 7.37979710e-01 3.21405560e-01 -1.48758769e-01
1.42926323e+00 -1.60063767e+00 -1.08945107e+00 -6.37709141e-01
6.96420729e-01 -4.26995307e-01 1.28073585e+00 -1.85624301e-01
-9.05900300e-01 -2.87834495e-01 -8.22277129e-01 -3.66089612e-01
5.72921485e-02 -1.69266388e-01 9.69772756e-01 7.56625414e-01
-1.10397100e+00 6.73259571e-02 -5.22156656e-01 -4.19564515e-01
5.13701104e-02 3.21512818e-02 -2.10202470e-01 -3.22027057e-02
-1.68786955e+00 9.03313458e-01 4.32962954e-01 2.94166535e-01
-7.14245141e-01 -1.19479068e-01 -1.41484129e+00 -1.36087239e-01
7.64701426e-01 -4.80420470e-01 1.83257687e+00 -6.06231451e-01
-2.15204859e+00 9.06912327e-01 -5.38605750e-01 -5.59604764e-01
4.75823693e-02 -1.10200278e-01 -1.21375479e-01 3.32059979e-01
1.34045154e-01 3.93792659e-01 4.26397562e-01 -1.44533110e+00
-5.72979689e-01 -2.04911634e-01 9.14718866e-01 8.51513326e-01
-1.59969404e-01 3.06101292e-01 -5.31756639e-01 -3.80722880e-01
-1.22364111e-01 -9.37593162e-01 -3.58500063e-01 -6.29943609e-01
-2.88844436e-01 -9.42080975e-01 6.06602728e-01 -6.81719124e-01
1.21457636e+00 -1.82185626e+00 2.84244865e-01 -4.32549745e-01
1.91217661e-01 2.56164879e-01 -3.18242699e-01 7.96375573e-01
2.30393648e-01 4.77090180e-02 -5.64659119e-01 -5.31885028e-01
-1.24737576e-01 4.98913974e-01 -2.07305685e-01 5.89522012e-02
1.87192768e-01 9.38681841e-01 -1.09313822e+00 -4.18228209e-01
2.39658296e-01 3.25469851e-01 -5.13177752e-01 7.13806391e-01
-4.68304187e-01 8.68466139e-01 -5.07983088e-01 2.88259864e-01
2.84861147e-01 -2.95591891e-01 9.28902328e-01 5.92709631e-02
1.19155414e-01 1.15873694e+00 -6.77336872e-01 2.01346254e+00
-6.44624650e-01 3.71053606e-01 4.45724100e-01 -1.32093835e+00
7.85554528e-01 4.04103875e-01 1.70134887e-01 -5.08781016e-01
-4.88066711e-02 -3.17692697e-01 2.44395599e-01 -4.08952266e-01
4.03295487e-01 -2.05121309e-01 -5.75461864e-01 7.05714762e-01
-2.08924152e-02 -2.89889991e-01 1.45640478e-01 5.41938484e-01
1.02081478e+00 3.31128053e-02 5.27845681e-01 -2.21041530e-01
1.25247562e+00 -5.58391251e-02 6.76664472e-01 6.72387898e-01
-6.27941668e-01 1.38989463e-01 1.05003226e+00 -3.88705045e-01
-2.77289659e-01 -5.15695035e-01 5.33972681e-01 1.57379174e+00
2.48681381e-01 -6.87029183e-01 -7.60666013e-01 -1.14651322e+00
-4.42302346e-01 8.48831713e-01 -4.32625532e-01 2.93696877e-02
-1.17500615e+00 -6.32245362e-01 6.03289843e-01 3.38896155e-01
9.03366089e-01 -1.14509523e+00 6.29641232e-04 2.42941007e-01
-7.68203199e-01 -1.62868559e+00 -4.39264297e-01 -2.79779583e-01
-5.52837193e-01 -1.41104770e+00 3.17914858e-02 -9.13099825e-01
2.67028540e-01 6.92935705e-01 1.20056307e+00 2.41526291e-01
2.54467070e-01 5.53367734e-01 -7.09719241e-01 -1.46094069e-01
-7.59082258e-01 3.37001473e-01 -6.47519767e-01 1.12590149e-01
2.08811894e-01 -4.43492442e-01 -4.32106227e-01 2.02707708e-01
-4.41552430e-01 3.04379791e-01 1.63980737e-01 8.53572547e-01
-1.20517313e-01 -1.63652048e-01 9.14721012e-01 -1.47810614e+00
1.19759333e+00 -2.47938320e-01 -6.81813583e-02 3.19459498e-01
-1.25357404e-01 2.21429661e-01 4.56200808e-01 2.97868222e-01
-1.57516456e+00 -9.45818275e-02 -5.12773097e-01 5.13904929e-01
-1.79426819e-01 4.67596591e-01 -5.73288441e-01 4.18773741e-02
9.27807912e-02 2.52859801e-01 1.97179765e-01 -5.40999830e-01
5.72259247e-01 6.61457002e-01 2.61721998e-01 -9.14523602e-01
3.15746456e-01 4.04369354e-01 -3.12148660e-01 -1.20454097e+00
-1.52629054e+00 -7.56287515e-01 -5.58431685e-01 -3.12876880e-01
9.04456913e-01 -1.03166723e+00 -9.56541955e-01 3.58992696e-01
-1.40950918e+00 -6.62490070e-01 2.79183090e-01 4.45117466e-02
-5.47033191e-01 6.93174601e-01 -9.06233430e-01 -9.99262035e-01
-4.05328274e-01 -8.92793059e-01 1.00813723e+00 -2.67950166e-02
-3.56681734e-01 -1.52196133e+00 1.18270390e-01 1.29569101e+00
3.11723948e-01 -3.51576544e-02 8.89748931e-01 -1.01931858e+00
-3.57225686e-01 3.46213251e-01 -1.99126676e-01 4.00597155e-01
4.03972208e-01 -5.93667269e-01 -9.92087901e-01 -6.28506467e-02
3.21471691e-01 -6.28348649e-01 8.37857902e-01 6.05886010e-03
1.07464385e+00 -3.26775998e-01 -6.67622909e-02 1.81348082e-02
7.15571404e-01 2.75794208e-01 4.27324712e-01 2.25719422e-01
9.23092186e-01 1.08925629e+00 6.66681230e-01 -3.13288230e-03
1.30024695e+00 6.26824975e-01 2.38096476e-01 1.76288754e-01
-2.42086098e-01 -4.32814956e-01 4.34802473e-01 1.34306669e+00
-5.26167005e-02 -3.06681484e-01 -8.70295763e-01 4.71563607e-01
-2.14179516e+00 -9.48161244e-01 -3.12652498e-01 1.41572416e+00
9.80403423e-01 1.45611107e-01 1.98641241e-01 -1.76755577e-01
8.16598535e-01 9.78413582e-01 -2.54388958e-01 -5.33328533e-01
-8.20178688e-02 -4.91318032e-02 -2.61411965e-01 1.08588445e+00
-1.31832647e+00 1.55074430e+00 6.62331200e+00 2.56899625e-01
-5.43384850e-01 3.17022353e-01 3.19118172e-01 3.95100921e-01
-3.06071073e-01 8.24767426e-02 -8.31550062e-01 1.06218286e-01
9.60711360e-01 -2.39877671e-01 3.25561404e-01 8.54969978e-01
2.62079149e-01 -1.04381412e-01 -8.68495166e-01 8.09441328e-01
5.46036959e-01 -1.23154330e+00 1.61550745e-01 -2.94457942e-01
2.21295610e-01 -3.23109418e-01 -6.10297143e-01 6.35734677e-01
7.02060938e-01 -9.28267777e-01 9.01032388e-02 -1.58385083e-01
1.57491878e-01 -6.64116800e-01 7.44184613e-01 4.54816282e-01
-1.34746432e+00 2.05413606e-02 -1.59182567e-02 -4.44251746e-01
2.92958379e-01 -1.06536500e-01 -1.12436306e+00 5.26089966e-01
2.09339514e-01 1.01049566e+00 -3.01515430e-01 2.15085670e-01
-1.01072896e+00 8.74531627e-01 2.89406866e-01 -2.88643986e-01
3.03104281e-01 -2.03186274e-01 6.05715036e-01 1.23945487e+00
-5.81276476e-01 3.67520690e-01 6.62336469e-01 3.83195341e-01
-3.49363685e-01 1.51604474e-01 -6.58662856e-01 -2.37023056e-01
4.93780345e-01 1.19469106e+00 -5.60569584e-01 -9.20599774e-02
-6.22392952e-01 1.10681915e+00 6.87824845e-01 -1.84071790e-02
-4.91459727e-01 -3.17893177e-02 8.69540215e-01 -3.46255749e-01
-1.25914752e-01 -4.08781081e-01 1.26801178e-01 -1.28904366e+00
-1.24464281e-01 -1.09432125e+00 7.84653664e-01 -2.88620263e-01
-1.19005549e+00 6.16670072e-01 -1.03088289e-01 -5.38398981e-01
-5.70077002e-01 -5.00433922e-01 -7.01912165e-01 7.15492785e-01
-1.89883101e+00 -1.64582002e+00 -1.34641215e-01 5.44202268e-01
1.30002141e+00 -5.82247525e-02 1.11088228e+00 -2.38888115e-01
-6.16599560e-01 2.57718503e-01 -7.01455474e-01 4.72029984e-01
7.14186192e-01 -1.54359567e+00 7.47114778e-01 7.34940588e-01
5.40338792e-02 6.79926097e-01 6.02729917e-01 -5.28532863e-01
-1.33619606e+00 -8.17266047e-01 1.15199435e+00 -6.06258214e-01
8.46069574e-01 -6.87371135e-01 -1.07731950e+00 8.69265616e-01
6.95363820e-01 -6.44224107e-01 1.17868900e+00 5.49751520e-01
-4.04653192e-01 1.97359830e-01 -7.12554812e-01 4.25803035e-01
1.27866995e+00 -8.52156937e-01 -1.32774007e+00 3.87709469e-01
1.08172011e+00 -5.26828647e-01 -6.00781798e-01 2.46669710e-01
4.29444574e-02 -8.72976661e-01 7.54542828e-01 -6.67908609e-01
4.51803595e-01 4.22771700e-04 2.89371628e-02 -1.52036548e+00
3.87598649e-02 -9.52101469e-01 -4.34678495e-02 1.20141864e+00
2.95930713e-01 -5.73042512e-01 6.69815719e-01 5.96423924e-01
-4.94714797e-01 -3.46260697e-01 -8.53517950e-01 -4.91795003e-01
-1.46307096e-01 -3.81998360e-01 2.49752149e-01 1.13636017e+00
4.50801790e-01 1.35690475e+00 -5.01710057e-01 1.38678238e-01
4.93018419e-01 3.55535179e-01 7.97758698e-01 -1.36652350e+00
-2.45054275e-01 -2.64083147e-01 3.12575072e-01 -1.42692804e+00
1.00589538e+00 -8.82253289e-01 2.28812099e-01 -2.14137268e+00
1.91407025e-01 -2.76917547e-01 -7.36513287e-02 4.84719485e-01
-5.77161372e-01 -2.44530156e-01 1.34856597e-01 1.89449824e-02
-1.21189773e+00 8.15969229e-01 1.33163965e+00 -5.89536503e-02
-2.52409995e-01 1.14587106e-01 -1.13809097e+00 8.67616296e-01
9.77921605e-01 -1.45600677e-01 -7.30957687e-01 -3.45648229e-01
-8.82455558e-02 2.94858575e-01 5.64694330e-02 -2.39528462e-01
2.77486682e-01 -3.45571637e-01 -6.11993551e-01 -4.09769654e-01
7.62582481e-01 -2.23014772e-01 -7.43702352e-01 2.59900630e-01
-7.57970273e-01 -3.66708308e-01 2.59980988e-02 7.68423438e-01
-4.87245947e-01 -2.10534856e-01 4.59403396e-01 -3.23614269e-01
-1.05394709e+00 -2.44641885e-01 -5.43232679e-01 4.29956734e-01
7.84409165e-01 4.24032420e-01 -4.67337459e-01 -1.04590023e+00
-5.89441955e-01 7.56511509e-01 -6.49264604e-02 8.97480965e-01
6.37290478e-01 -9.86158788e-01 -8.00161600e-01 -2.42557719e-01
2.97689557e-01 6.97145164e-02 3.12939316e-01 1.37901470e-01
-1.77152753e-01 7.42385626e-01 1.56880006e-01 -2.30369627e-01
-1.71853018e+00 -1.79729182e-02 3.19234788e-01 -6.43607378e-01
-4.87471014e-01 9.56583619e-01 1.65525109e-01 -9.96552706e-01
1.68759227e-01 3.23376060e-02 -9.01427329e-01 7.03393966e-02
7.38734663e-01 3.99789773e-02 -2.18903080e-01 -7.79296458e-01
-2.63540864e-01 1.49704620e-01 -1.93751901e-01 -1.48311809e-01
1.45815682e+00 -4.43229049e-01 -5.94848812e-01 8.68244410e-01
8.31667781e-01 -4.38282490e-02 -1.05973709e+00 -4.59132791e-01
5.25693417e-01 -1.65406480e-01 -1.69147491e-01 -6.98984802e-01
-5.07680714e-01 7.84906685e-01 -2.87391782e-01 4.91299182e-01
6.25435650e-01 3.56102526e-01 8.14341903e-01 6.18662000e-01
4.80021298e-01 -1.15600896e+00 4.49532360e-01 1.11617446e+00
1.16597450e+00 -1.48105872e+00 -1.84540287e-01 -1.16444337e+00
-1.05885184e+00 9.09498692e-01 9.29724097e-01 2.17550635e-01
3.26300204e-01 -1.32824197e-01 2.83664256e-01 -2.83634514e-01
-1.12858343e+00 -4.94982779e-01 -2.24008802e-02 4.35154706e-01
8.39133203e-01 2.37152129e-02 -5.67786694e-01 6.51729524e-01
-1.35633692e-01 -5.55959642e-01 5.66783965e-01 1.04058242e+00
-4.69926685e-01 -1.72066772e+00 9.52006355e-02 -1.31620556e-01
-4.55744863e-01 -1.50354981e-01 -1.07549846e+00 7.61424720e-01
-7.17602015e-01 1.51024103e+00 -1.26726642e-01 -1.26336321e-01
3.84956449e-01 4.40449268e-01 3.49053979e-01 -1.16115272e+00
-7.48744428e-01 -1.37022153e-01 1.15209210e+00 -7.17378318e-01
-1.02031493e+00 -4.72602010e-01 -1.37989795e+00 5.05879782e-02
1.81455389e-01 4.58731085e-01 5.35293996e-01 1.12509394e+00
1.73003048e-01 6.98498607e-01 8.47222030e-01 -2.92878121e-01
-4.08576131e-01 -1.11582458e+00 -1.85188591e-01 4.55298156e-01
2.02540886e-02 -4.54206318e-01 -3.69329363e-01 -1.43390641e-01] | [12.430427551269531, 8.051023483276367] |
524f3d8f-feb9-46c1-bc2c-2873ef81ff9e | extending-the-forward-forward-algorithm | 2307.04205 | null | https://arxiv.org/abs/2307.04205v1 | https://arxiv.org/pdf/2307.04205v1.pdf | Extending the Forward Forward Algorithm | The Forward Forward algorithm, proposed by Geoffrey Hinton in November 2022, is a novel method for training neural networks as an alternative to backpropagation. In this project, we replicate Hinton's experiments on the MNIST dataset, and subsequently extend the scope of the method with two significant contributions. First, we establish a baseline performance for the Forward Forward network on the IMDb movie reviews dataset. As far as we know, our results on this sentiment analysis task marks the first instance of the algorithm's extension beyond computer vision. Second, we introduce a novel pyramidal optimization strategy for the loss threshold - a hyperparameter specific to the Forward Forward method. Our pyramidal approach shows that a good thresholding strategy causes a difference of upto 8% in test error. 1 Lastly, we perform visualizations of the trained parameters and derived several significant insights, such as a notably larger (10-20x) mean and variance in the weights acquired by the Forward Forward network. | ['Advaith Sridhar', 'Jonah Kornberg', 'Ritu Gala', 'Saumya Gandhi'] | 2023-07-09 | null | null | null | null | ['sentiment-analysis'] | ['natural-language-processing'] | [-3.53236916e-03 2.49594480e-01 8.12986866e-02 -8.47450197e-01
-2.66193300e-01 -3.25614959e-01 7.31134236e-01 7.07544461e-02
-8.67030442e-01 4.17746097e-01 3.30746233e-01 -4.66493607e-01
-3.66309226e-01 -3.02444816e-01 -7.29971707e-01 -5.28014243e-01
-2.39993677e-01 1.66746214e-01 2.09988624e-01 -4.21389669e-01
8.21237266e-01 3.99055958e-01 -1.29685259e+00 6.13077402e-01
2.31368780e-01 1.26291525e+00 8.95416513e-02 5.37550867e-01
1.06245698e-02 1.18247831e+00 -4.44929868e-01 -7.32586324e-01
4.09872442e-01 2.86062267e-02 -1.11729360e+00 -4.38488603e-01
4.14366275e-01 -3.31597418e-01 -2.23679617e-01 9.40223992e-01
3.77352655e-01 4.46144462e-01 5.80769300e-01 -9.92676854e-01
-8.62319589e-01 1.02291501e+00 -3.84475023e-01 4.70203996e-01
-4.45844047e-02 1.00632198e-01 1.23623860e+00 -1.30881250e+00
5.58493316e-01 7.85850883e-01 1.31262565e+00 4.99536186e-01
-8.43306363e-01 -4.21618700e-01 5.00546932e-01 4.35810030e-01
-9.25407469e-01 -6.26579940e-01 6.66219234e-01 -5.34275770e-01
1.38670671e+00 4.18675020e-02 6.29428744e-01 7.23176420e-01
1.50865108e-01 1.04277873e+00 9.94860649e-01 -6.48544788e-01
2.40392223e-01 5.25843143e-01 5.30476987e-01 6.24736309e-01
2.40178499e-02 7.76620209e-02 -4.93589133e-01 2.37818301e-01
4.17137653e-01 -8.61427933e-02 -2.31162563e-01 -1.12677611e-01
-8.67159784e-01 1.00589180e+00 7.87970066e-01 2.72224575e-01
-3.24721336e-01 3.32396775e-01 3.61816674e-01 3.83836478e-01
5.83356082e-01 7.12101340e-01 -6.71751678e-01 -1.19716831e-01
-1.05461955e+00 8.03659633e-02 8.61724079e-01 6.27564490e-01
4.75220740e-01 -8.44794512e-02 1.05208628e-01 7.36745894e-01
6.48678958e-01 2.27764174e-02 7.10625648e-01 -1.09028685e+00
5.01901567e-01 2.00974151e-01 1.04968444e-01 -1.08552063e+00
-6.37505472e-01 -7.45891750e-01 -4.69167501e-01 2.10489452e-01
4.97272938e-01 -4.30601090e-01 -7.51299620e-01 1.40058815e+00
-7.66090751e-02 -3.05378318e-01 -1.79585412e-01 7.98386157e-01
8.29834759e-01 4.59851801e-01 -1.02618337e-01 7.19648749e-02
9.66220915e-01 -1.40717828e+00 -4.93568152e-01 -5.25633991e-01
7.00195789e-01 -5.57044923e-01 9.28322494e-01 6.96562469e-01
-1.28666818e+00 -4.03726250e-01 -1.20310950e+00 1.07160375e-01
-6.43017709e-01 2.02795655e-01 8.55011761e-01 5.02429962e-01
-1.26387060e+00 1.25988889e+00 -6.14137590e-01 -3.60770911e-01
3.81379813e-01 5.14172375e-01 -2.22507998e-01 2.07453579e-01
-1.15874088e+00 1.23693573e+00 3.36519659e-01 3.72448891e-01
-2.85736471e-01 -8.29557478e-01 -5.50267398e-01 1.64956361e-01
-9.05064791e-02 -4.62163329e-01 1.56740415e+00 -1.13304019e+00
-1.52221644e+00 8.17465246e-01 5.65992929e-02 -8.52110744e-01
3.66835296e-01 -4.44097400e-01 -1.90012187e-01 4.34670560e-02
-2.18440056e-01 7.88354397e-01 4.96235818e-01 -1.04390228e+00
-6.31941557e-01 -2.51248866e-01 1.83885038e-01 1.36487216e-01
-6.67219937e-01 2.25417823e-01 -2.67901480e-01 -6.03129745e-01
1.83898658e-01 -9.20012474e-01 -2.59762436e-01 1.46393580e-02
-1.68741271e-01 6.25835359e-02 2.16660067e-01 -5.85885227e-01
1.09158731e+00 -2.13831329e+00 -2.52536654e-01 2.11537123e-01
1.76573306e-01 7.64488503e-02 5.93398251e-02 3.87884974e-01
-5.31391919e-01 2.27534041e-01 -1.21756807e-01 -6.29908442e-01
2.38739684e-01 -1.07534766e-01 -3.57025504e-01 4.36749905e-01
-1.12085566e-01 8.15602899e-01 -6.55829430e-01 2.94595566e-02
-1.51949763e-01 4.17135030e-01 -5.65163255e-01 -9.61449519e-02
3.24663997e-01 -3.01313430e-01 1.66172653e-01 3.18267167e-01
4.10680890e-01 -4.62466747e-01 -1.55692816e-01 -3.59726995e-01
-2.42620841e-01 4.35126692e-01 -9.03640687e-01 1.43430793e+00
-2.64178425e-01 9.81410921e-01 -2.16976292e-02 -8.55148017e-01
7.84690440e-01 6.70302212e-02 2.75228769e-01 -5.50340652e-01
4.09126937e-01 1.71561077e-01 1.40970126e-02 -3.71754825e-01
6.65363729e-01 -4.16623116e-01 3.58539104e-01 4.59388107e-01
2.93613851e-01 1.51614383e-01 2.46169224e-01 -4.04206663e-02
1.01031923e+00 -6.56227171e-02 -2.16004718e-02 -2.47254848e-01
4.02164787e-01 8.06129873e-02 2.94804603e-01 8.43194723e-01
-4.11955982e-01 9.15143371e-01 2.98231751e-01 -7.32499778e-01
-7.61885226e-01 -6.19412899e-01 -9.83679444e-02 1.34470987e+00
-5.27010798e-01 -2.91239500e-01 -6.75350845e-01 -7.82423019e-01
-2.28066668e-02 8.43347371e-01 -9.99666035e-01 -1.19156413e-01
-4.56167936e-01 -1.12910211e+00 5.56707680e-01 7.59319305e-01
6.16128623e-01 -1.17442083e+00 -5.42885125e-01 -1.11945406e-01
1.74080551e-01 -9.02858377e-01 -2.35629633e-01 7.20188975e-01
-1.07721758e+00 -7.55168140e-01 -7.28555918e-01 -8.94094527e-01
5.37697673e-01 -1.74146965e-01 1.04952228e+00 2.16682944e-02
8.84383023e-02 1.77559912e-01 -2.25771904e-01 -6.12542450e-01
-1.37681097e-01 1.34213343e-01 2.21521333e-02 -2.76603490e-01
5.06246746e-01 -3.93094987e-01 -7.42238164e-01 -1.45160809e-01
-2.76814759e-01 -1.14640199e-01 4.97736394e-01 8.28594148e-01
2.16623649e-01 -3.90724950e-02 4.47213680e-01 -1.16872561e+00
7.94059932e-01 -3.37083995e-01 -4.75678772e-01 -1.66161120e-01
-1.22286248e+00 -5.38494484e-03 5.38074493e-01 -1.74047828e-01
-9.88974035e-01 -2.17101239e-02 -4.36428308e-01 -1.81257680e-01
-3.57039236e-02 7.53857851e-01 5.82736194e-01 -2.71598279e-01
7.73060918e-01 -5.05712107e-02 -1.09235175e-01 -5.35401881e-01
4.26476568e-01 4.48384941e-01 4.24162894e-01 -2.47178003e-02
4.03254241e-01 4.52119499e-01 -3.49763304e-01 -4.45539296e-01
-9.05681312e-01 -4.05769348e-01 -6.97237611e-01 -5.01053594e-02
6.49364471e-01 -6.34936333e-01 -7.28923202e-01 5.51916540e-01
-1.08456540e+00 -5.30623734e-01 -5.34909487e-01 6.82383895e-01
-4.85846400e-01 6.40068799e-02 -1.09610379e+00 -6.54958665e-01
-5.37133455e-01 -7.86877811e-01 1.34985834e-01 2.02294022e-01
-2.47970507e-01 -1.43167591e+00 8.41471478e-02 2.04118878e-01
7.31612146e-01 -3.83351505e-01 7.76284933e-01 -1.12701786e+00
1.29876539e-01 -1.93903238e-01 -1.37128502e-01 6.95952594e-01
-3.62887144e-01 1.10501446e-01 -1.23045099e+00 -1.78988263e-01
3.85083497e-01 -1.37996688e-01 1.14236569e+00 5.32940149e-01
8.31645548e-01 -3.40060711e-01 -8.92276540e-02 7.13808179e-01
1.35686994e+00 2.24182203e-01 5.07062137e-01 9.60592210e-01
4.56613749e-01 7.07621634e-01 3.84931356e-01 3.61775428e-01
3.33989918e-01 9.85748172e-02 3.35243732e-01 -4.55795117e-02
-2.11815596e-01 -1.52323380e-01 2.64056742e-01 9.89377618e-01
-6.94269314e-02 1.14084501e-02 -8.56227338e-01 5.97394168e-01
-1.66426849e+00 -1.00180614e+00 -1.85060188e-01 2.01807785e+00
6.44500673e-01 6.10502899e-01 -2.53425427e-02 2.86276609e-01
4.02896017e-01 5.16515225e-02 -4.60176885e-01 -7.59832263e-01
8.60335827e-02 4.65655252e-02 7.81115532e-01 5.42953014e-01
-1.08813548e+00 8.43288779e-01 8.06697464e+00 2.62229383e-01
-1.23566520e+00 -1.08209446e-01 7.52482176e-01 -3.77977788e-01
-5.74768186e-02 -7.51936212e-02 -9.30678725e-01 2.91963786e-01
1.27364492e+00 7.17520937e-02 3.85160178e-01 8.97376120e-01
8.17472264e-02 -7.10319309e-03 -1.15256131e+00 9.58642006e-01
1.69033840e-01 -1.55993581e+00 -3.18418384e-01 -2.21296817e-01
6.74513519e-01 6.16053045e-01 4.07864332e-01 4.71424341e-01
2.75692731e-01 -1.05525529e+00 9.43193913e-01 5.28028607e-01
8.01512375e-02 -8.18710864e-01 9.19132233e-01 3.31184089e-01
-4.96532232e-01 -6.53537512e-01 -5.23853719e-01 -4.21627849e-01
1.49805471e-02 7.11756349e-01 -7.92227507e-01 -1.31351709e-01
1.01590025e+00 8.73432994e-01 -7.60665715e-01 1.17450047e+00
-2.36673340e-01 8.56886804e-01 -1.05485000e-01 -1.87555492e-01
2.99615920e-01 9.62755457e-02 2.59006828e-01 1.48209774e+00
7.31094852e-02 -6.42058551e-02 -4.66312766e-01 7.52101660e-01
-1.53470397e-01 9.17761251e-02 -4.35621470e-01 -4.04216461e-02
2.92177081e-01 1.39907837e+00 -6.80752814e-01 -1.24934569e-01
-3.42467993e-01 6.71106815e-01 4.55797017e-01 4.47305232e-01
-6.20418966e-01 -8.11255932e-01 2.41809994e-01 1.09634157e-02
6.95411801e-01 8.94427020e-03 -8.44501376e-01 -8.21683228e-01
-1.05234660e-01 -6.86699092e-01 1.63164228e-01 -9.18854356e-01
-1.32099319e+00 6.63783610e-01 -2.34183729e-01 -7.62702107e-01
-3.75194341e-01 -1.07187450e+00 -5.79265237e-01 8.38119864e-01
-1.53347147e+00 -7.30966926e-01 -4.52931486e-02 2.03210294e-01
3.64585131e-01 -1.81361035e-01 7.49776006e-01 4.79742616e-01
-5.14727056e-01 6.91352665e-01 2.55517483e-01 3.27055067e-01
6.20437264e-01 -1.36598754e+00 7.15252697e-01 6.23696625e-01
1.15586683e-01 9.18232083e-01 8.81204844e-01 -1.25291973e-01
-1.05571878e+00 -6.17320180e-01 9.95442629e-01 -4.41509813e-01
7.90692627e-01 -1.05673231e-01 -8.12319994e-01 1.07319736e+00
4.01074320e-01 -1.20297641e-01 5.92224061e-01 3.31232995e-01
-2.21664563e-01 -1.05515406e-01 -1.27522922e+00 5.50701082e-01
6.71120405e-01 -3.20260465e-01 -8.17876220e-01 3.11239243e-01
6.02532327e-01 -2.93111295e-01 -8.40867400e-01 3.09974432e-01
7.78582871e-01 -1.40256929e+00 8.60767961e-01 -4.57258850e-01
6.47932351e-01 1.62284344e-01 -9.77194309e-02 -1.28575766e+00
-6.30670309e-01 -3.15293282e-01 -8.10337141e-02 1.10009360e+00
8.43558967e-01 -7.46493161e-01 1.03968692e+00 8.04490745e-01
-2.22830772e-01 -1.32146037e+00 -5.97980380e-01 -5.52897990e-01
4.17390436e-01 -7.19968021e-01 1.35844812e-01 8.13245595e-01
1.61456212e-01 2.96200544e-01 -7.99278393e-02 -2.72941113e-01
3.50406557e-01 -2.85656899e-01 2.66637862e-01 -1.12118053e+00
-5.02772987e-01 -6.72276616e-01 -8.96024480e-02 -1.27062321e+00
-1.04962431e-01 -9.70659256e-01 2.94331551e-01 -1.59375250e+00
7.73476735e-02 -3.44068974e-01 -7.05327988e-01 5.45512140e-01
2.53249198e-01 3.14920336e-01 4.98698503e-01 2.38746837e-01
-2.95931935e-01 2.39296600e-01 7.99668372e-01 -1.72363240e-02
-5.83322309e-02 1.20955735e-01 -1.11557531e+00 1.24502063e+00
7.95560896e-01 -6.33649468e-01 -4.20763582e-01 -8.03239167e-01
7.18112171e-01 -4.47579056e-01 1.70052201e-01 -1.10078490e+00
6.17807090e-01 1.92311317e-01 6.82335556e-01 -6.14829421e-01
4.59706962e-01 -6.49645567e-01 -4.42209274e-01 4.74789292e-01
-7.73357093e-01 3.88799906e-01 4.15381879e-01 2.48302743e-01
-1.28100097e-01 -7.13082790e-01 6.61485910e-01 -3.04557770e-01
-7.81600237e-01 -1.85847998e-01 -4.01946843e-01 1.16518237e-01
4.97113794e-01 -1.77558973e-01 -4.03448492e-01 -4.75472033e-01
-7.73406804e-01 2.64438745e-02 1.43084854e-01 2.30964258e-01
6.57394052e-01 -9.48102534e-01 -4.71629798e-01 1.20336331e-01
-3.73730779e-01 -4.96738344e-01 -1.07098036e-01 1.13878918e+00
-4.76034641e-01 4.74205285e-01 -7.39624202e-02 -2.76997030e-01
-1.19925070e+00 5.32658160e-01 5.45148253e-01 -1.38522580e-01
-5.35830140e-01 1.62196803e+00 -1.80458963e-01 -6.24076962e-01
4.41767782e-01 -4.00306016e-01 -5.74927151e-01 2.61603862e-01
6.85332477e-01 3.99737179e-01 4.44757283e-01 -3.15384269e-01
-5.63403070e-01 4.44833755e-01 -4.77128476e-01 -4.61557806e-01
1.80941772e+00 -2.15276733e-01 1.09924823e-02 7.63342977e-01
1.30425334e+00 -1.19288824e-01 -1.23734987e+00 6.65302761e-03
1.61132723e-01 -8.09931755e-02 3.36242318e-01 -1.10421085e+00
-1.14242697e+00 1.07080436e+00 7.08088219e-01 2.52952754e-01
1.05208981e+00 -3.20665956e-01 6.73310697e-01 8.54510128e-01
-4.60622758e-02 -1.32129776e+00 -1.39671564e-01 9.83505011e-01
7.79357851e-01 -1.13461971e+00 2.41599366e-01 4.83204201e-02
-7.05828369e-01 1.32822418e+00 4.34532136e-01 -4.05942202e-01
7.95797348e-01 2.68860549e-01 1.75885841e-01 -3.53085130e-01
-1.02664447e+00 2.24940807e-01 2.68772483e-01 3.99433881e-01
7.63665497e-01 -5.02747834e-01 -2.48098880e-01 7.55481243e-01
-5.36105573e-01 2.41487384e-01 5.99017859e-01 1.00973725e+00
-6.06554389e-01 -6.06977344e-01 -9.00080204e-02 3.94546896e-01
-6.65114105e-01 -5.93435347e-01 -3.52571875e-01 7.13161111e-01
-5.09105995e-02 9.63872015e-01 -1.60913616e-02 -6.30962312e-01
3.80308658e-01 2.30803847e-01 4.47866768e-01 -4.67435569e-01
-1.04367507e+00 -2.79676646e-01 -2.77363304e-02 -4.64890212e-01
-1.43124849e-01 -5.62400937e-01 -1.49389434e+00 -2.20632851e-01
-3.04025292e-01 1.68844074e-01 1.13585746e+00 9.74527478e-01
1.60629675e-01 5.59312940e-01 4.32427078e-01 -1.00839269e+00
-9.88318324e-01 -9.42393839e-01 -5.61015189e-01 1.93560317e-01
3.68125498e-01 -2.97935843e-01 -7.89857864e-01 -1.04354523e-01] | [8.1849946975708, 3.4648995399475098] |
c9acbaa2-f901-434b-90b3-c65be125c710 | dawn-of-the-transformer-era-in-speech-emotion | 2203.07378 | null | https://arxiv.org/abs/2203.07378v2 | https://arxiv.org/pdf/2203.07378v2.pdf | Dawn of the transformer era in speech emotion recognition: closing the valence gap | Recent advances in transformer-based architectures which are pre-trained in self-supervised manner have shown great promise in several machine learning tasks. In the audio domain, such architectures have also been successfully utilised in the field of speech emotion recognition (SER). However, existing works have not evaluated the influence of model size and pre-training data on downstream performance, and have shown limited attention to generalisation, robustness, fairness, and efficiency. The present contribution conducts a thorough analysis of these aspects on several pre-trained variants of wav2vec 2.0 and HuBERT that we fine-tuned on the dimensions arousal, dominance, and valence of MSP-Podcast, while additionally using IEMOCAP and MOSI to test cross-corpus generalisation. To the best of our knowledge, we obtain the top performance for valence prediction without use of explicit linguistic information, with a concordance correlation coefficient (CCC) of .638 on MSP-Podcast. Furthermore, our investigations reveal that transformer-based architectures are more robust to small perturbations compared to a CNN-based baseline and fair with respect to biological sex groups, but not towards individual speakers. Finally, we are the first to show that their extraordinary success on valence is based on implicit linguistic information learnt during fine-tuning of the transformer layers, which explains why they perform on-par with recent multimodal approaches that explicitly utilise textual information. Our findings collectively paint the following picture: transformer-based architectures constitute the new state-of-the-art in SER, but further advances are needed to mitigate remaining robustness and individual speaker issues. To make our findings reproducible, we release the best performing model to the community. | ['Björn W. Schuller', 'Felix Burkhardt', 'Florian Eyben', 'Maximilian Schmitt', 'Hagen Wierstorf', 'Andreas Triantafyllopoulos', 'Johannes Wagner'] | 2022-03-14 | null | null | null | null | ['cross-corpus'] | ['computer-vision'] | [ 5.71139976e-02 2.62987822e-01 2.57369488e-01 -6.08988464e-01
-5.37939966e-01 -5.09126663e-01 6.97535157e-01 4.42054898e-01
-6.68298423e-01 6.04072988e-01 3.64062935e-01 -5.46236662e-03
-1.56984061e-01 -4.16017115e-01 -4.49123949e-01 -6.17321134e-01
-3.40306669e-01 2.96614587e-01 -2.03261629e-01 -6.98144436e-01
-8.48105475e-02 3.54999870e-01 -1.70713639e+00 4.26337510e-01
3.42945099e-01 1.18271196e+00 -4.82516408e-01 6.06697679e-01
2.05179900e-01 3.16236407e-01 -8.11966240e-01 -9.41221058e-01
-9.56678465e-02 -3.52255046e-01 -7.98747599e-01 -2.89307266e-01
2.70413458e-01 1.26324311e-01 5.70957251e-02 4.09597635e-01
1.17330408e+00 1.64785936e-01 5.65072238e-01 -1.06697190e+00
-4.86272454e-01 7.12375641e-01 -1.53965680e-02 3.29398036e-01
1.11861914e-01 1.66417822e-01 1.13473272e+00 -6.77643478e-01
4.88493204e-01 1.11272395e+00 8.06199551e-01 8.03091109e-01
-1.37372553e+00 -8.12408090e-01 9.51309651e-02 2.03594863e-01
-1.05568624e+00 -8.40909243e-01 7.28540778e-01 -2.74241328e-01
1.33689153e+00 3.85506541e-01 7.98470020e-01 1.65617669e+00
3.96064036e-02 3.89152378e-01 1.25898051e+00 -1.63774699e-01
2.20009103e-01 4.57641870e-01 -2.14463204e-01 2.19283015e-01
-2.69793093e-01 2.05223426e-01 -9.38522339e-01 -3.58907692e-02
1.60259172e-01 -8.91791523e-01 -1.30927369e-01 -9.70046371e-02
-9.51897085e-01 9.05003309e-01 1.55404538e-01 6.50712371e-01
-2.56051391e-01 2.11356767e-02 1.18447733e+00 5.78636050e-01
7.14945793e-01 6.61151886e-01 -7.56210804e-01 -5.49543858e-01
-1.07374334e+00 8.12965166e-03 7.43468583e-01 2.65443027e-01
2.91918665e-01 3.50663543e-01 -1.79520950e-01 1.26412463e+00
9.50359926e-02 2.28906170e-01 6.38291419e-01 -7.11372674e-01
8.62627178e-02 2.76272148e-01 -6.28354073e-01 -9.40702021e-01
-5.91462076e-01 -6.97074056e-01 -6.67118073e-01 1.39838025e-01
2.79780298e-01 -2.64311492e-01 -4.12205070e-01 1.94121611e+00
8.32496360e-02 -4.94244918e-02 1.07145682e-01 8.14585507e-01
1.06712353e+00 2.46621490e-01 3.59961659e-01 -1.86519578e-01
1.32365251e+00 -3.92905325e-01 -5.47490835e-01 -2.01619476e-01
4.21210021e-01 -7.14961290e-01 1.07018018e+00 8.62485647e-01
-1.13315606e+00 -3.95360470e-01 -1.05543506e+00 2.36116759e-02
-7.29888976e-01 1.10426702e-01 7.71797121e-01 1.25260699e+00
-1.24764109e+00 6.35048151e-01 -5.55376947e-01 -5.46897590e-01
3.88028294e-01 6.29443944e-01 -6.02396250e-01 6.48013353e-01
-1.46922004e+00 1.09735310e+00 1.41091123e-01 3.37134987e-01
-7.69694507e-01 -6.88239932e-01 -7.55568027e-01 1.76041643e-03
-8.07635635e-02 -4.74743962e-01 9.30666864e-01 -1.14879990e+00
-1.91518784e+00 1.26194668e+00 1.88626140e-01 -5.88332772e-01
2.72732943e-01 1.26287058e-01 -5.68540633e-01 -5.76655054e-03
-5.43136120e-01 9.16351497e-01 5.99704921e-01 -1.00459719e+00
-1.40280217e-01 -2.58757532e-01 -8.46681893e-02 2.34669205e-02
-8.14368963e-01 1.15030073e-01 2.11530495e-02 -5.28323352e-01
-5.14395833e-01 -6.64105594e-01 1.17837615e-01 -2.77588338e-01
-1.33246660e-01 -1.68855175e-01 2.10955173e-01 -5.40023685e-01
1.19500124e+00 -2.22188926e+00 2.37036794e-01 2.56244808e-01
-1.02438845e-01 4.69553292e-01 -2.79886931e-01 5.73113263e-01
-3.43617588e-01 2.56608605e-01 -1.22398041e-01 -5.83430111e-01
3.37974817e-01 1.23021208e-01 -1.85144573e-01 5.18271565e-01
3.99482042e-01 7.65614510e-01 -5.54110646e-01 -3.15207392e-01
2.91791290e-01 9.26782191e-01 -7.06395805e-01 4.94891107e-02
1.53795853e-01 4.45838690e-01 1.96428463e-01 2.00820789e-01
3.90934885e-01 4.83631372e-01 7.49712512e-02 -2.07549840e-01
-2.61254042e-01 5.13709009e-01 -7.61164904e-01 1.54334068e+00
-6.66801214e-01 8.05907667e-01 2.17182279e-01 -1.13604867e+00
1.26160479e+00 4.97658461e-01 3.56258005e-01 -7.87214100e-01
6.25440180e-01 1.90518215e-01 4.18720126e-01 -4.33529049e-01
2.96326101e-01 -4.73273188e-01 -1.03195995e-01 1.24725521e-01
6.45745516e-01 -2.99780637e-01 5.83936237e-02 -8.71824399e-02
7.47807682e-01 -1.06641963e-01 5.66001795e-02 -2.09354892e-01
6.14623368e-01 -4.53438312e-01 2.03577533e-01 3.73296678e-01
-3.33528876e-01 6.94864750e-01 8.11533988e-01 -8.55585039e-02
-8.81324112e-01 -7.51753569e-01 -5.64328611e-01 1.43932891e+00
-6.76458657e-01 -5.72361588e-01 -8.63547802e-01 -2.88331270e-01
-3.33940268e-01 7.57424593e-01 -1.07052422e+00 -4.26388860e-01
-8.66358802e-02 -9.38185930e-01 1.22342098e+00 3.02831441e-01
1.16724238e-01 -1.22309673e+00 -6.49555087e-01 2.61290938e-01
-1.26179025e-01 -9.71300840e-01 8.76178369e-02 5.85088849e-01
-6.16899729e-01 -6.63028181e-01 -6.46122396e-01 -4.43638086e-01
-1.31613731e-01 -7.08944559e-01 1.10771239e+00 -1.42515466e-01
-1.61426470e-01 6.04507744e-01 -3.49001586e-01 -6.83504462e-01
-2.22109973e-01 3.89812589e-01 1.16679515e-03 3.93069014e-02
4.32806343e-01 -8.90717983e-01 -4.09583360e-01 1.31819487e-01
-8.62302005e-01 -4.62507993e-01 4.34776068e-01 9.61780787e-01
2.00730190e-01 -2.63256758e-01 8.36956143e-01 -6.99942708e-01
8.97973776e-01 -4.17069465e-01 1.57204956e-01 -1.58715695e-01
-5.16911507e-01 -1.75408319e-01 4.45378542e-01 -4.11249369e-01
-1.02558982e+00 -3.37464929e-01 -7.52389610e-01 -2.26394042e-01
-4.91317809e-01 4.75391269e-01 3.94546650e-02 -1.42664507e-01
7.75236964e-01 3.00028827e-02 6.67713955e-02 -2.66465366e-01
1.95827872e-01 7.34865725e-01 4.72933203e-01 -5.21476805e-01
2.55632788e-01 2.97903240e-01 -6.23958968e-02 -1.04258454e+00
-5.33411622e-01 -2.40483701e-01 -4.61271375e-01 -2.02913135e-01
8.10643017e-01 -8.71455014e-01 -1.02955496e+00 3.67667705e-01
-7.56831169e-01 -4.94353890e-01 -1.94024444e-01 3.75130922e-01
-5.22361815e-01 1.60473943e-01 -6.20330930e-01 -9.13588285e-01
-5.92045188e-01 -8.57654631e-01 9.99006271e-01 -1.64570227e-01
-7.14110494e-01 -1.18070543e+00 3.91151518e-01 4.21307027e-01
7.06229389e-01 2.84359515e-01 8.57342482e-01 -1.08217883e+00
5.50055563e-01 -4.46571372e-02 1.85720161e-01 6.04736149e-01
-2.79415607e-01 1.41186669e-01 -1.67901146e+00 -5.41966558e-02
-6.38624802e-02 -6.42104805e-01 8.29025626e-01 1.83339670e-01
9.68170345e-01 3.10612805e-02 3.62074107e-01 5.37263632e-01
8.79252672e-01 -2.14957833e-01 6.76556587e-01 4.76827949e-01
1.35064870e-01 9.52276111e-01 3.21602732e-01 6.48914814e-01
3.20059240e-01 6.26012325e-01 5.40829062e-01 -1.66892171e-01
-3.63686010e-02 1.33603945e-01 4.69227761e-01 6.86156690e-01
-9.05628577e-02 -1.84160888e-01 -6.84006512e-01 7.18198776e-01
-1.24267650e+00 -9.31599081e-01 -6.18696399e-02 2.13163733e+00
8.89282763e-01 2.51345813e-01 3.93625528e-01 5.05111456e-01
1.71161518e-01 2.28761569e-01 -8.03455710e-02 -1.29887712e+00
-4.52284604e-01 7.02220380e-01 7.07678422e-02 1.94273010e-01
-9.33918357e-01 7.01731205e-01 6.21235704e+00 7.05553830e-01
-1.55378246e+00 1.00539394e-01 6.93573117e-01 -5.59746981e-01
-2.30970308e-01 -5.17800033e-01 -4.11709666e-01 3.20553064e-01
1.50010920e+00 1.77704513e-01 5.41734874e-01 5.08239627e-01
3.53158861e-01 5.99247627e-02 -1.16982937e+00 8.90973330e-01
3.30396116e-01 -6.56674922e-01 -4.23530132e-01 9.87297017e-03
3.27852100e-01 2.03526825e-01 5.06301880e-01 6.30290389e-01
-2.93104291e-01 -1.34460366e+00 7.44415760e-01 2.52877146e-01
7.18018115e-01 -9.55441654e-01 1.00805771e+00 -1.05855174e-01
-6.69259489e-01 -1.03239920e-02 -2.76612878e-01 -1.17339820e-01
7.95153007e-02 4.99018431e-01 -8.56455088e-01 3.35081428e-01
9.32244837e-01 4.24590796e-01 -5.95948279e-01 6.98153734e-01
3.08706146e-02 8.49467993e-01 -2.81617463e-01 -1.12021394e-01
2.76858538e-01 2.68356949e-01 3.89489084e-01 1.71252978e+00
1.18805565e-01 -2.17824593e-01 -6.27943218e-01 6.22446179e-01
1.13740265e-01 6.19621277e-01 -3.75281960e-01 -2.50169277e-01
2.38028392e-02 1.47374618e+00 -4.32051629e-01 -8.43783095e-02
-3.20535749e-01 8.36643755e-01 3.64284337e-01 7.29168728e-02
-6.99746192e-01 -3.31270039e-01 8.42484355e-01 -1.36150330e-01
3.93221170e-01 1.76223785e-01 -6.35993958e-01 -8.43568504e-01
-2.18253598e-01 -1.03266394e+00 3.80306721e-01 -6.96333587e-01
-1.34628761e+00 9.14497674e-01 -6.99270144e-02 -7.24548042e-01
-2.99186289e-01 -7.77166128e-01 -5.57969749e-01 7.68998742e-01
-1.29709506e+00 -1.17177522e+00 1.25899926e-01 4.79760379e-01
1.90351903e-01 -1.23824790e-01 1.22560692e+00 3.46195608e-01
-5.04954100e-01 1.00456679e+00 -9.49516967e-02 -9.97736454e-02
1.02166545e+00 -1.08139217e+00 -1.93190090e-02 3.16979319e-01
6.82567284e-02 7.06312537e-01 9.80832517e-01 -6.02687672e-02
-8.73844206e-01 -6.23484671e-01 1.14298105e+00 -4.71785873e-01
6.72690928e-01 -7.71376193e-01 -7.96206117e-01 3.43489856e-01
6.31990671e-01 -2.58447498e-01 1.12334883e+00 6.97470009e-01
-4.69778746e-01 -2.49053031e-01 -1.18699050e+00 4.42732513e-01
7.35522330e-01 -5.06073356e-01 -4.27828640e-01 -6.51298389e-02
2.79473037e-01 -2.55958050e-01 -1.26947093e+00 3.08442593e-01
7.78212845e-01 -1.47784042e+00 8.85982335e-01 -4.23973173e-01
5.40480614e-01 2.99060613e-01 -4.40886877e-02 -1.63036096e+00
-9.08172429e-02 -3.46247792e-01 4.32400048e-01 1.77503943e+00
5.78719199e-01 -6.36023641e-01 4.35460210e-01 4.97981340e-01
-2.91639596e-01 -9.14211273e-01 -1.18022215e+00 -4.19112921e-01
4.24383491e-01 -8.27983141e-01 3.87411624e-01 9.25201774e-01
4.67056602e-01 4.37609702e-01 -4.66167212e-01 -3.16930503e-01
3.23192682e-03 -5.69457889e-01 6.48699045e-01 -1.19409430e+00
-3.10331255e-01 -7.48111188e-01 -5.61111808e-01 -3.44709605e-02
3.86519134e-01 -8.77167404e-01 -6.33921027e-02 -9.05396521e-01
-5.57231680e-02 -9.71193239e-02 -5.33366740e-01 5.60713112e-01
1.94112405e-01 6.79850459e-01 6.46644011e-02 -4.62136626e-01
-2.89445162e-01 7.96103597e-01 7.47510195e-01 7.22808912e-02
-1.72053158e-01 -2.59279519e-01 -9.38029110e-01 3.08819562e-01
9.54150021e-01 -2.80687988e-01 -4.05390054e-01 -2.46687740e-01
4.05737787e-01 -3.86944473e-01 2.99969107e-01 -9.57811892e-01
-1.27708122e-01 3.11590523e-01 3.91699612e-01 4.03565243e-02
8.03611457e-01 -6.64908171e-01 4.26404132e-03 1.00284800e-01
-6.28190279e-01 -2.03839093e-01 7.76237190e-01 5.29608205e-02
-2.67489046e-01 -9.66292694e-02 6.76452339e-01 4.92414087e-02
-3.78656000e-01 5.25926910e-02 -6.73859894e-01 8.12700689e-02
5.53416610e-01 -4.39239480e-02 -6.83400780e-02 -5.82557023e-01
-8.53007674e-01 -1.06978267e-01 1.27290100e-01 5.09967446e-01
1.92914620e-01 -9.96393979e-01 -8.49913657e-01 2.21370950e-01
3.05385202e-01 -7.20184863e-01 5.35054922e-01 1.28363132e+00
2.17557549e-02 5.82146704e-01 -4.42016840e-01 -4.31647301e-01
-1.41286802e+00 3.63804430e-01 3.97532225e-01 -4.22716551e-02
-1.04410565e-02 1.00966370e+00 -3.68015729e-02 -7.46710062e-01
2.93534487e-01 -7.29365870e-02 -5.57120681e-01 6.41799927e-01
2.84614146e-01 2.93321282e-01 5.82973599e-01 -8.71581495e-01
-4.60644066e-01 2.58053124e-01 1.79370075e-01 -5.01346946e-01
1.60261905e+00 -1.06910616e-01 -4.76052389e-02 7.57605314e-01
1.25038064e+00 1.30426824e-01 -8.07120800e-01 2.41067022e-01
-3.61755192e-01 8.67158324e-02 1.22002371e-01 -1.09727752e+00
-1.14866483e+00 1.37819374e+00 6.95081115e-01 1.01891316e-01
1.25234568e+00 -9.43263918e-02 3.34607065e-01 1.99172404e-02
-1.13178611e-01 -1.29323971e+00 -6.21617883e-02 5.23417473e-01
9.31032419e-01 -9.52787936e-01 -3.38052481e-01 -2.06761106e-04
-9.86485422e-01 1.11304402e+00 5.23471355e-01 4.92105894e-02
3.99890393e-01 1.86737552e-01 3.40200245e-01 -2.21463114e-01
-1.17379081e+00 -2.65313625e-01 3.45432729e-01 9.03169513e-01
1.03147316e+00 -1.08627230e-01 -5.25721252e-01 9.21620727e-01
-7.70934522e-01 -1.91014826e-01 2.20403954e-01 2.74475962e-01
-3.72214168e-02 -1.14899731e+00 -2.03652784e-01 2.52239019e-01
-9.95745301e-01 -2.44725734e-01 -7.72298396e-01 8.11891377e-01
2.08465695e-01 1.08312678e+00 4.00466146e-03 -5.79691291e-01
4.94189829e-01 5.38683712e-01 4.19172198e-01 -4.27140713e-01
-1.39839339e+00 5.09383753e-02 5.68779588e-01 -3.85810435e-01
-4.29532617e-01 -8.80687654e-01 -7.81751752e-01 -2.36712694e-01
-9.54323187e-02 1.14111871e-01 8.84954333e-01 8.68692875e-01
2.86278993e-01 6.75347269e-01 3.06778371e-01 -8.59943092e-01
-2.12791875e-01 -1.19576752e+00 -5.13020992e-01 3.55859488e-01
2.02666745e-01 -4.85738456e-01 -5.42938709e-01 -1.73562482e-01] | [13.66866683959961, 5.85616397857666] |
51404f40-deb7-4967-bfb8-dcf4f03d1eef | gesturegan-for-hand-gesture-to-gesture | 1808.04859 | null | https://arxiv.org/abs/1808.04859v2 | https://arxiv.org/pdf/1808.04859v2.pdf | GestureGAN for Hand Gesture-to-Gesture Translation in the Wild | Hand gesture-to-gesture translation in the wild is a challenging task since hand gestures can have arbitrary poses, sizes, locations and self-occlusions. Therefore, this task requires a high-level understanding of the mapping between the input source gesture and the output target gesture. To tackle this problem, we propose a novel hand Gesture Generative Adversarial Network (GestureGAN). GestureGAN consists of a single generator $G$ and a discriminator $D$, which takes as input a conditional hand image and a target hand skeleton image. GestureGAN utilizes the hand skeleton information explicitly, and learns the gesture-to-gesture mapping through two novel losses, the color loss and the cycle-consistency loss. The proposed color loss handles the issue of "channel pollution" while back-propagating the gradients. In addition, we present the Fr\'echet ResNet Distance (FRD) to evaluate the quality of generated images. Extensive experiments on two widely used benchmark datasets demonstrate that the proposed GestureGAN achieves state-of-the-art performance on the unconstrained hand gesture-to-gesture translation task. Meanwhile, the generated images are in high-quality and are photo-realistic, allowing them to be used as data augmentation to improve the performance of a hand gesture classifier. Our model and code are available at https://github.com/Ha0Tang/GestureGAN. | ['Yan Yan', 'Hao Tang', 'Wei Wang', 'Nicu Sebe', 'Dan Xu'] | 2018-08-14 | null | null | null | null | ['gesture-to-gesture-translation'] | ['computer-vision'] | [ 3.11845928e-01 -3.23789358e-01 -1.67692468e-01 -2.89544910e-01
-7.95947671e-01 -5.82086444e-01 5.31173468e-01 -9.91553962e-01
-3.54712546e-01 5.45885265e-01 1.96616836e-02 -1.38636634e-01
4.22915131e-01 -7.29757547e-01 -9.28664863e-01 -1.01418424e+00
2.73191899e-01 3.16370934e-01 3.87470014e-02 1.14897288e-01
-1.40171379e-01 4.70882237e-01 -9.90383744e-01 2.89104674e-02
5.72661400e-01 9.14611518e-01 4.93356436e-02 8.73864353e-01
-1.27536375e-02 5.59020519e-01 -6.17034733e-01 -5.33547342e-01
7.56876409e-01 -8.23134780e-01 -5.45565426e-01 -7.28097484e-02
5.55845857e-01 -8.22308779e-01 -9.21857774e-01 1.06794786e+00
8.84382188e-01 2.24901214e-01 6.54196143e-01 -1.45756006e+00
-5.94918728e-01 3.57884288e-01 -7.35143006e-01 -1.02159977e-01
1.06018640e-01 7.89885163e-01 7.40026474e-01 -8.84456575e-01
9.11157608e-01 1.30940354e+00 2.45717049e-01 9.80856836e-01
-9.53595519e-01 -1.26201022e+00 2.18548402e-01 4.87632900e-02
-1.48618090e+00 -1.74846724e-01 7.52547920e-01 -4.44085211e-01
4.88620311e-01 2.92062908e-01 6.91981852e-01 1.60995567e+00
-2.40759835e-01 8.58152270e-01 9.44255650e-01 -3.34384084e-01
2.37413310e-02 -4.77853537e-01 -3.34001929e-01 9.11826372e-01
-1.87366620e-01 4.43596333e-01 -4.37773347e-01 2.29944482e-01
1.31244957e+00 2.33106524e-01 -5.04315436e-01 -3.26895714e-01
-1.03544509e+00 5.85723758e-01 8.16389918e-01 2.09464114e-02
-2.37830311e-01 5.42906046e-01 5.67227192e-02 1.32136732e-01
1.61853984e-01 -2.71592826e-01 -1.05954461e-01 -1.77448213e-01
-6.80708528e-01 3.63583297e-01 5.57457447e-01 1.03591204e+00
2.73775965e-01 9.63859037e-02 -3.66125405e-01 5.75270414e-01
3.74991506e-01 1.04912055e+00 3.06698591e-01 -6.92283034e-01
9.87269402e-01 4.71425325e-01 -1.21333160e-01 -6.96051717e-01
1.59661490e-02 -1.50319785e-01 -9.74732399e-01 6.89007580e-01
7.53193796e-01 -2.15817764e-01 -1.40585470e+00 1.98104596e+00
3.45654428e-01 3.15324634e-01 -2.47517690e-01 1.42241263e+00
7.38845527e-01 5.90107441e-01 2.00899795e-01 2.03341439e-01
1.01436937e+00 -1.04916131e+00 -4.98846620e-01 -2.83938777e-02
-4.08639722e-02 -7.82700956e-01 1.42452526e+00 -3.16152303e-03
-1.05467284e+00 -4.36963052e-01 -9.03206587e-01 -1.57577842e-01
-1.52164856e-02 2.66442209e-01 3.50219250e-01 5.69422662e-01
-5.57914257e-01 3.07302505e-01 -1.21342909e+00 -1.04067415e-01
6.32359743e-01 4.27353621e-01 -6.67002797e-02 -3.08087707e-01
-7.92792022e-01 4.44899797e-01 1.25189036e-01 3.01978588e-01
-8.53098631e-01 -5.51288009e-01 -6.36822224e-01 -9.63804871e-02
1.55317143e-01 -6.75627291e-01 1.02477527e+00 -1.12912571e+00
-1.87203002e+00 6.38073742e-01 -6.62539825e-02 1.73224300e-01
1.25865245e+00 -1.14417829e-01 -8.71272907e-02 1.25910759e-01
-1.82177797e-01 7.34404385e-01 1.14586973e+00 -1.20374417e+00
-4.72252607e-01 -5.82696795e-01 -1.54826626e-01 3.79773498e-01
1.32074328e-02 -3.24185900e-02 -9.80741680e-01 -1.12318778e+00
8.19031745e-02 -1.18773949e+00 2.27603614e-01 3.39513838e-01
-6.86828792e-01 1.16951555e-01 9.36601162e-01 -9.42807555e-01
8.67229283e-01 -2.15169001e+00 4.73032147e-01 4.76854354e-01
-2.48179752e-02 3.46681654e-01 -5.29950202e-01 2.38462575e-02
8.85185450e-02 8.11925307e-02 -3.96992922e-01 -2.59708762e-01
-4.15696390e-02 1.61121994e-01 -3.07945520e-01 3.18370640e-01
6.28366768e-02 1.28427207e+00 -7.35700428e-01 -4.49680746e-01
2.79892564e-01 8.95874977e-01 -3.70293111e-01 5.30777037e-01
-3.57980490e-01 9.29391980e-01 -4.88394886e-01 7.22805679e-01
6.00218236e-01 -1.01997539e-01 2.15337887e-01 -9.58816856e-02
3.32329392e-01 -4.89855334e-02 -1.06423151e+00 1.91028130e+00
-3.78495455e-01 6.54370546e-01 -9.82932076e-02 -2.44901001e-01
6.41243339e-01 2.18179524e-01 3.21867228e-01 -7.25872338e-01
3.82726103e-01 2.34015793e-01 9.86002013e-02 -4.48264837e-01
-1.27454668e-01 -1.99373290e-02 2.75171220e-01 7.10805357e-01
-1.90242693e-01 -4.78080511e-02 -8.79625529e-02 -1.07246697e-01
9.11620021e-01 4.65433538e-01 -3.15374345e-01 2.85171002e-01
1.23905927e-01 -2.77825475e-01 3.63669008e-01 3.59121025e-01
1.44809529e-01 9.75607038e-01 2.70602345e-01 -2.82591462e-01
-1.12859607e+00 -1.23411787e+00 3.53917420e-01 9.91006434e-01
2.65823275e-01 3.25280726e-01 -9.78633344e-01 -9.20293510e-01
-1.49493083e-01 2.38469124e-01 -5.12908697e-01 8.78664777e-02
-1.09937692e+00 -5.64059198e-01 8.25375199e-01 9.07643199e-01
8.04809809e-01 -1.22268736e+00 -6.31331861e-01 -3.97649445e-02
-3.38553756e-01 -9.97196734e-01 -1.13930976e+00 -2.61536896e-01
-6.37332141e-01 -1.30229712e+00 -1.40272927e+00 -8.82140696e-01
9.54341054e-01 -1.26277789e-01 6.14306808e-01 2.41795585e-01
-5.06359041e-01 1.40460506e-01 -4.31326509e-01 -1.99344039e-01
-2.25924715e-01 1.74124673e-01 -2.15033069e-01 7.33444691e-02
1.49738818e-01 -5.46719193e-01 -9.40434217e-01 3.53519082e-01
-9.50607061e-01 -3.15428339e-02 6.47055507e-01 1.05198622e+00
6.66074038e-01 -3.54996592e-01 -1.22454483e-02 -4.07452673e-01
4.49615628e-01 1.07930690e-01 -5.79086602e-01 2.78178126e-01
-4.40834790e-01 1.28243953e-01 4.54310626e-01 -7.91620791e-01
-1.04604065e+00 3.47426653e-01 -2.26183310e-01 -7.01464176e-01
8.47955570e-02 -1.56903788e-01 -6.09430611e-01 -1.32210508e-01
5.02200425e-01 2.95192808e-01 6.29597157e-02 -4.89011079e-01
5.22077918e-01 5.86232185e-01 6.57372236e-01 -5.59969783e-01
1.09961116e+00 4.42480415e-01 -4.65885401e-02 -6.11218154e-01
-1.53087124e-01 -9.51275080e-02 -6.24640524e-01 -1.53680414e-01
8.63774359e-01 -6.18526578e-01 -6.64705992e-01 9.61729467e-01
-1.16280437e+00 -9.74227786e-01 -3.61645043e-01 4.65811551e-01
-4.81462091e-01 1.21469863e-01 -6.22829616e-01 -5.17056704e-01
-5.27373970e-01 -1.30352378e+00 1.01148391e+00 2.83551514e-01
1.20948955e-01 -6.63195372e-01 -7.41559044e-02 1.13414846e-01
2.26378143e-01 5.17509639e-01 8.79198194e-01 -9.79390442e-02
-1.02062249e+00 -1.43941462e-01 -4.73300368e-01 4.11388040e-01
1.20389111e-01 -8.80734026e-02 -7.89159060e-01 -4.51337278e-01
-4.44008142e-01 -2.97225982e-01 7.72313297e-01 3.25672507e-01
1.09974611e+00 -4.63791430e-01 -1.58718362e-01 1.04387975e+00
1.23001385e+00 3.39784116e-01 7.67922878e-01 -5.45240752e-02
1.17169011e+00 2.07790539e-01 2.93436259e-01 1.76859319e-01
1.70196339e-01 9.77928758e-01 3.54329944e-01 -2.61152864e-01
-8.35468173e-01 -5.90279937e-01 3.21127832e-01 4.57288146e-01
-5.70161402e-01 -5.11920035e-01 -7.65135050e-01 1.00628868e-01
-1.69769776e+00 -7.14616597e-01 1.09933056e-01 1.99676323e+00
8.24471653e-01 -2.60224789e-01 2.00638890e-01 9.05142054e-02
7.40397155e-01 1.96420133e-01 -9.83171701e-01 2.17126340e-01
-2.06085443e-02 7.00649798e-01 6.06426239e-01 5.05927444e-01
-9.65235591e-01 1.15306973e+00 5.03144741e+00 6.96049035e-01
-1.57271302e+00 2.46794388e-01 3.07070076e-01 -3.17505866e-01
-1.09835811e-01 -4.12436694e-01 -4.20974940e-01 5.77448606e-01
1.85276002e-01 3.64503294e-01 8.55263650e-01 4.89713043e-01
8.18181038e-02 3.44392776e-01 -9.34004962e-01 1.06681526e+00
8.59048292e-02 -9.46262360e-01 2.21359819e-01 1.71695575e-01
6.21529520e-01 -6.41932040e-02 3.16195697e-01 -1.21279970e-01
4.19556320e-01 -1.19301069e+00 8.20628524e-01 2.91871876e-01
1.60566640e+00 -5.43483377e-01 5.04527926e-01 -4.82115749e-04
-1.28618085e+00 1.84441566e-01 1.74849659e-01 2.24492133e-01
1.59192815e-01 -1.69566780e-01 -5.87773085e-01 2.40761802e-01
5.70577562e-01 2.33359858e-01 -2.64174700e-01 7.40750909e-01
-8.14976275e-01 4.47832197e-01 -4.06629473e-01 4.69991826e-02
1.19942345e-01 -1.86821390e-02 5.44705451e-01 1.08415103e+00
2.99315453e-01 4.46100712e-01 4.25231867e-02 9.83000398e-01
-3.92564923e-01 5.25816903e-02 -2.38852099e-01 -8.65775626e-03
4.39196408e-01 6.76634252e-01 -6.54287755e-01 -9.57447365e-02
-4.40975577e-02 1.57708204e+00 -1.26848761e-02 7.55254447e-01
-8.19733858e-01 -4.29784477e-01 8.66225958e-01 6.10313602e-02
2.59588569e-01 -2.23976955e-01 -3.71124864e-01 -1.26599121e+00
3.93520236e-01 -8.37930679e-01 1.44371226e-01 -5.66337287e-01
-1.00672996e+00 5.30512571e-01 -2.77920008e-01 -1.19612503e+00
-3.87559086e-01 -5.76379359e-01 -5.56456208e-01 1.02298951e+00
-1.37032878e+00 -1.48523819e+00 -8.24662924e-01 9.59019780e-01
5.88986874e-01 -1.59168050e-01 6.66502535e-01 4.67268109e-01
-4.72498536e-01 1.09614241e+00 -6.98290244e-02 7.92204797e-01
5.97127616e-01 -9.32770193e-01 6.50696218e-01 9.39831018e-01
2.52358854e-01 2.97265291e-01 2.50519216e-01 -5.99725485e-01
-1.39938378e+00 -1.21346438e+00 2.94147015e-01 -3.62343580e-01
1.47945240e-01 -2.91447371e-01 -6.09208345e-01 7.70849586e-01
-6.14662245e-02 2.43903875e-01 1.96110517e-01 -5.32176375e-01
-5.65406680e-01 6.17570952e-02 -1.18081534e+00 6.87845230e-01
1.56411505e+00 -3.94868553e-01 -8.95893872e-02 2.09912613e-01
3.60119313e-01 -9.68840241e-01 -4.46822584e-01 1.95771962e-01
1.05816460e+00 -5.86933553e-01 1.05144739e+00 -6.57191694e-01
5.96783161e-01 -1.87446371e-01 -1.51806980e-01 -1.12822485e+00
1.40366942e-01 -5.64705372e-01 4.90198564e-03 9.94614840e-01
1.99164137e-01 -4.21015531e-01 1.01720214e+00 6.59301102e-01
2.59962767e-01 -6.02514446e-01 -9.54499066e-01 -9.19129491e-01
2.66166717e-01 -2.55266994e-01 8.43553007e-01 5.87695003e-01
-4.36158508e-01 2.10667253e-02 -4.57999587e-01 5.72256595e-02
8.62277746e-01 1.34863049e-01 1.03401613e+00 -7.49143898e-01
-3.36105227e-01 -5.83600223e-01 -3.16044807e-01 -1.33171010e+00
1.69955090e-01 -9.75131094e-01 2.25933269e-01 -1.29960287e+00
1.28719002e-01 -6.29090190e-01 2.33867448e-02 6.35499656e-01
-1.62791744e-01 6.35964334e-01 5.90713203e-01 4.59897906e-01
6.13387823e-02 4.87615228e-01 1.68280911e+00 -4.51205313e-01
-4.17891055e-01 6.55577984e-04 -1.01671852e-01 5.93898058e-01
8.04953277e-01 -4.37982857e-01 -2.10391998e-01 -7.75705099e-01
-4.60330129e-01 9.82086137e-02 7.55587697e-01 -6.62632346e-01
7.53623620e-02 -3.59014809e-01 3.86634380e-01 -1.35649920e-01
4.15763378e-01 -8.81733537e-01 2.57489204e-01 6.49282336e-01
-3.21161211e-01 -5.16147278e-02 -1.81350976e-01 4.11901534e-01
-2.90091839e-02 2.00690001e-01 8.95703018e-01 -1.56265572e-02
-5.08714378e-01 6.71193779e-01 4.13406968e-01 1.46531001e-01
9.26090121e-01 -1.15431808e-01 -2.17341825e-01 -4.27901834e-01
-6.03155255e-01 -1.68731604e-02 5.21588802e-01 6.48279309e-01
8.32270265e-01 -1.43449867e+00 -7.77541459e-01 5.31592429e-01
-1.15671553e-01 2.81908631e-01 1.54596254e-01 5.57229578e-01
-7.99827933e-01 5.77977635e-02 -4.15396065e-01 -5.54601550e-01
-1.32132840e+00 9.85207334e-02 3.37810278e-01 -5.32258078e-02
-9.57076252e-01 1.07051611e+00 1.34489611e-01 -2.71484554e-01
6.93819225e-01 -4.14902180e-01 5.31039536e-01 -4.34907794e-01
5.42893231e-01 4.58108008e-01 -2.90993810e-01 -6.73552275e-01
-2.16627195e-01 8.53882909e-01 2.67155409e-01 -4.96558756e-01
1.07976818e+00 2.45760381e-01 2.70749956e-01 -1.17699280e-01
1.21688449e+00 -1.58280611e-01 -1.80205023e+00 -1.95628420e-01
-5.73369861e-01 -7.90817082e-01 -2.90832728e-01 -1.18992400e+00
-1.71383429e+00 1.10653973e+00 1.00019431e+00 -6.60254538e-01
1.12706017e+00 -1.34133860e-01 1.17142415e+00 1.06867798e-01
3.16504806e-01 -7.16565609e-01 2.90776581e-01 3.34017962e-01
1.12434089e+00 -9.65745211e-01 -3.21088374e-01 -3.81462604e-01
-5.30738294e-01 9.69846368e-01 4.79872912e-01 -7.70976990e-02
4.81295168e-01 3.66376847e-01 5.60394585e-01 1.07575640e-01
6.26298487e-02 -2.65881151e-01 2.62854546e-01 7.69142866e-01
1.10308133e-01 3.39088023e-01 -3.09394777e-01 3.85705054e-01
-3.41096312e-01 2.66144514e-01 -5.95200323e-02 7.45833993e-01
2.69011676e-01 -1.36634767e+00 -3.91448051e-01 1.15544148e-01
-2.87776470e-01 -1.52597968e-02 -5.33832371e-01 9.27369237e-01
1.02731451e-01 4.96005833e-01 -6.71725050e-02 -6.56031966e-01
4.51941699e-01 1.63661286e-01 8.30870509e-01 -2.86884069e-01
-4.98352051e-01 1.08139515e-01 -5.73206306e-01 -5.98515630e-01
-3.12521279e-01 -5.08615613e-01 -1.51308906e+00 -4.07575011e-01
-1.15811884e-01 -4.00428563e-01 8.49152684e-01 7.10662544e-01
2.29719236e-01 5.09815812e-01 4.68987525e-01 -1.08761024e+00
-7.47702599e-01 -9.50039923e-01 -4.48846728e-01 6.82222188e-01
2.71833420e-01 -6.13585770e-01 -1.99897766e-01 2.89329976e-01] | [11.702908515930176, -0.7249549031257629] |
452c1025-3c1b-4be7-a621-f605864438bc | learning-discrete-state-abstractions-with | 2003.04300 | null | https://arxiv.org/abs/2003.04300v3 | https://arxiv.org/pdf/2003.04300v3.pdf | Learning Discrete State Abstractions With Deep Variational Inference | Abstraction is crucial for effective sequential decision making in domains with large state spaces. In this work, we propose an information bottleneck method for learning approximate bisimulations, a type of state abstraction. We use a deep neural encoder to map states onto continuous embeddings. We map these embeddings onto a discrete representation using an action-conditioned hidden Markov model, which is trained end-to-end with the neural network. Our method is suited for environments with high-dimensional states and learns from a stream of experience collected by an agent acting in a Markov decision process. Through this learned discrete abstract model, we can efficiently plan for unseen goals in a multi-goal Reinforcement Learning setting. We test our method in simplified robotic manipulation domains with image states. We also compare it against previous model-based approaches to finding bisimulations in discrete grid-world-like environments. Source code is available at https://github.com/ondrejba/discrete_abstractions. | ['Jan-Willem van de Meent', 'Lawson L. S. Wong', 'Robert Platt', 'Ondrej Biza'] | 2020-03-09 | null | https://openreview.net/forum?id=oU7dRDX8SNA | https://openreview.net/pdf?id=oU7dRDX8SNA | pproximateinference-aabi-symposium-2021-1 | ['multi-goal-reinforcement-learning'] | ['methodology'] | [-7.71408379e-02 4.23085600e-01 -2.34982207e-01 -2.38781035e-01
-6.81603968e-01 -4.14624572e-01 8.50616455e-01 2.53682822e-01
-5.03582656e-01 7.23754108e-01 4.77609545e-01 -1.61004424e-01
-4.57690582e-02 -8.29055071e-01 -1.00361085e+00 -3.78240913e-01
-5.14446437e-01 1.13289821e+00 7.00032488e-02 -2.76691377e-01
3.55870932e-01 4.37994391e-01 -1.46070313e+00 1.71055540e-01
3.93663824e-01 6.21991456e-01 4.63092655e-01 1.15045178e+00
1.71408489e-01 1.16608989e+00 -2.86046445e-01 2.86942720e-01
4.71258014e-01 -4.89704430e-01 -1.17453074e+00 4.04379368e-02
-7.03532696e-02 -8.19264531e-01 -8.40762675e-01 8.00701261e-01
1.58378989e-01 5.20960331e-01 6.42212212e-01 -1.18073583e+00
-3.90273809e-01 5.16497910e-01 1.77328229e-01 -1.06483027e-01
5.60342312e-01 6.92371607e-01 8.73701215e-01 -4.77941364e-01
7.67069876e-01 1.56564713e+00 2.52369642e-01 7.88615584e-01
-1.42130864e+00 -2.33473703e-01 4.06809062e-01 2.27489725e-01
-1.06738508e+00 -4.38040495e-01 3.43089908e-01 -3.71045649e-01
1.47395468e+00 -4.10646647e-01 1.08121443e+00 1.17581642e+00
5.88515759e-01 1.03409028e+00 9.93232667e-01 -1.93015739e-01
8.97413552e-01 -4.70544815e-01 -2.80734032e-01 9.88898337e-01
-6.31148219e-02 5.86636066e-01 -6.64278686e-01 -2.53851473e-01
1.09934103e+00 5.07983267e-01 3.63175874e-03 -8.53932977e-01
-1.37744308e+00 8.39481413e-01 5.06245732e-01 -2.14197263e-01
-6.53898001e-01 7.57529318e-01 3.87345165e-01 6.04784727e-01
-7.64693841e-02 6.55105829e-01 -3.08575243e-01 -5.27995050e-01
-5.32832444e-01 6.87565923e-01 1.13268948e+00 1.10404420e+00
8.24273527e-01 2.14757212e-02 6.86001498e-03 1.32822409e-01
3.93947840e-01 3.51240367e-01 4.47508961e-01 -1.63577068e+00
3.15340102e-01 2.01913550e-01 7.07975984e-01 -1.51363298e-01
-3.72215092e-01 1.21682517e-01 -4.59543139e-01 8.81606996e-01
3.12532634e-01 -1.90462202e-01 -1.14734828e+00 1.71096039e+00
3.60533029e-01 3.22012544e-01 4.12174791e-01 8.93975377e-01
-2.09737659e-01 9.77401674e-01 -1.08366780e-01 1.23482063e-01
8.23527217e-01 -1.09586811e+00 -5.07555664e-01 -3.67243648e-01
6.03585005e-01 -1.38716370e-01 8.93853068e-01 6.22241080e-01
-1.32102442e+00 -4.55320984e-01 -1.11764312e+00 -1.28268301e-01
-2.63129562e-01 -5.27323127e-01 5.95898986e-01 -2.85222620e-01
-1.29053700e+00 1.17777109e+00 -1.81902337e+00 -3.98464054e-01
2.99254656e-01 2.98004657e-01 -2.35554174e-01 -1.69314682e-01
-8.89349520e-01 1.27115679e+00 6.45851731e-01 -4.58639897e-02
-2.20001841e+00 2.75104796e-03 -1.19829118e+00 -1.62816383e-02
3.85161340e-01 -8.67197752e-01 2.03413510e+00 -3.18850338e-01
-2.04695272e+00 3.35574687e-01 -1.12146638e-01 -8.70325625e-01
2.83107758e-01 -4.39965963e-01 1.77126110e-01 1.52059004e-01
1.67482093e-01 7.06966162e-01 7.87777662e-01 -1.08276975e+00
-6.52966201e-01 -4.38041449e-01 4.52959269e-01 6.11007333e-01
3.29422355e-01 -3.81418437e-01 1.37794828e-02 1.40976191e-01
1.32494897e-01 -1.20750558e+00 -9.25386250e-01 1.02934167e-01
-9.02840272e-02 -2.50765413e-01 4.88176376e-01 -4.06608701e-01
6.47946358e-01 -1.68562019e+00 7.90828168e-01 -1.69904619e-01
9.78149548e-02 -1.71000138e-01 -1.88281626e-01 1.11433434e+00
3.31701905e-01 -2.86214024e-01 -3.78647029e-01 -7.49575555e-01
4.75144506e-01 6.02121294e-01 -4.30791587e-01 5.12802958e-01
1.97512522e-01 8.68578851e-01 -1.39433074e+00 -1.64584130e-01
4.54330832e-01 1.10038795e-01 -7.36907482e-01 6.14827931e-01
-7.62148499e-01 5.89452446e-01 -6.02930129e-01 3.03977937e-01
1.26078054e-01 1.57075115e-02 3.57710749e-01 6.70495331e-01
-2.94736147e-01 6.30497515e-01 -1.13783431e+00 2.60933805e+00
-7.93675005e-01 3.44083965e-01 4.57310975e-02 -1.01922810e+00
6.77854538e-01 3.37805361e-01 2.69295931e-01 -3.86129051e-01
-3.53906751e-02 -3.29453200e-02 -2.57532448e-01 -3.29804718e-01
4.76753652e-01 -2.27122664e-01 -5.93687236e-01 6.32290542e-01
3.50695997e-01 -9.31489587e-01 1.12716943e-01 3.01751763e-01
1.47924590e+00 7.58840501e-01 4.75541353e-01 -4.77500148e-02
-2.08066896e-01 3.62285078e-01 4.73717541e-01 1.15601218e+00
-4.10429627e-01 2.86898226e-01 5.07356524e-01 -8.85508299e-01
-1.32720172e+00 -1.51647365e+00 3.60088646e-01 8.34343016e-01
2.90920734e-01 -6.87879682e-01 -4.15866315e-01 -3.50272566e-01
-1.80498306e-02 9.11692381e-01 -5.81156254e-01 -3.36340725e-01
-6.72983885e-01 2.50272483e-01 2.04759583e-01 5.45520127e-01
2.81459928e-01 -1.35970151e+00 -1.32718122e+00 5.21682382e-01
7.31958672e-02 -7.78662801e-01 -6.04556911e-02 6.03332818e-01
-1.00911558e+00 -8.69917512e-01 -4.63781297e-01 -7.78677166e-01
3.77045870e-01 -2.33526543e-01 1.08467329e+00 -3.41189474e-01
-2.43301585e-01 5.83373785e-01 -2.58230329e-01 -4.58535582e-01
-6.69184566e-01 -1.35356903e-01 3.54454696e-01 -6.42211080e-01
1.65709168e-01 -6.37465000e-01 -6.47303760e-01 -8.15097243e-02
-7.99476266e-01 4.40079898e-01 4.45116073e-01 9.39556599e-01
4.79220331e-01 -1.83474004e-01 -8.20779577e-02 -1.02327511e-01
6.81933999e-01 -4.74318981e-01 -9.38766837e-01 -2.36896530e-01
-1.00319244e-01 6.53405964e-01 6.96649075e-01 -2.68459260e-01
-1.01116014e+00 3.59722435e-01 9.34446976e-02 -5.51265061e-01
-5.88067293e-01 3.41512352e-01 3.80136281e-01 5.59230685e-01
7.43672192e-01 3.99383962e-01 2.57981867e-01 -1.68470234e-01
5.54771721e-01 3.87546450e-01 3.14678103e-01 -9.61183012e-01
4.97649282e-01 6.25482142e-01 3.20733190e-02 -4.83911425e-01
-6.01777077e-01 -1.93756491e-01 -6.61970198e-01 1.38355428e-02
8.40551078e-01 -9.45434272e-01 -8.61480892e-01 4.19040829e-01
-1.15443504e+00 -1.49039519e+00 -6.81969106e-01 5.76355040e-01
-1.57310271e+00 3.35725211e-02 -9.30759192e-01 -9.66828644e-01
1.45115942e-01 -1.25230575e+00 1.21836817e+00 1.31237924e-01
-3.52233350e-01 -7.58019447e-01 5.90270221e-01 -1.92704484e-01
2.15075627e-01 2.85064846e-01 4.88859922e-01 -3.06768805e-01
-9.04258609e-01 3.03411782e-02 3.94203097e-01 3.95366810e-02
-1.52207538e-01 -4.48013395e-01 -5.38339496e-01 -4.32267934e-01
-1.44907817e-01 -9.31818008e-01 7.71389365e-01 4.42380905e-01
1.02424026e+00 -4.47311252e-01 -3.30107123e-01 2.53908545e-01
1.42750406e+00 2.61977285e-01 4.80922103e-01 3.70198309e-01
1.93868682e-01 3.82930279e-01 8.28653336e-01 8.73713374e-01
5.63900352e-01 4.19634908e-01 8.85017514e-01 4.31420177e-01
2.41457224e-01 -6.56002700e-01 7.24693418e-01 3.02674264e-01
2.47084424e-02 5.94934970e-02 -1.16061425e+00 8.63793135e-01
-2.28059220e+00 -1.05362582e+00 6.68824792e-01 2.09257627e+00
8.13398659e-01 3.02136421e-01 9.31693241e-02 -3.93255234e-01
3.76393974e-01 1.24193944e-01 -8.66548777e-01 -7.86022723e-01
7.17298508e-01 4.30289149e-01 3.98872107e-01 8.77938986e-01
-9.27239418e-01 1.33591986e+00 5.99719667e+00 1.44180700e-01
-6.99861169e-01 7.54558817e-02 1.30749211e-01 -4.52191114e-01
8.53857175e-02 3.18283319e-01 -6.27744377e-01 1.38815477e-01
1.28000665e+00 -9.18023661e-02 1.00024378e+00 1.03688085e+00
2.74355829e-01 -3.47940981e-01 -1.62562311e+00 6.56540453e-01
-4.46520895e-01 -1.38755620e+00 -2.29221746e-01 1.11895353e-01
7.00814664e-01 4.75341946e-01 -1.24021778e-02 7.35710800e-01
1.47329485e+00 -1.04312778e+00 9.70339477e-01 3.46660048e-01
3.64817053e-01 -7.56156564e-01 2.29977369e-01 7.87170947e-01
-1.06548870e+00 -4.53121722e-01 -4.44855809e-01 -8.12255859e-01
5.74782670e-01 -1.28708258e-01 -1.02472746e+00 1.55603290e-01
5.04700780e-01 7.98707783e-01 2.94528723e-01 7.69255221e-01
-6.04822874e-01 1.98328435e-01 -5.15894711e-01 -2.25376084e-01
6.91940248e-01 -2.60682642e-01 5.50398827e-01 7.73854375e-01
2.39826649e-01 1.11460403e-01 6.30487442e-01 1.01423168e+00
4.01907206e-01 -7.90428877e-01 -1.12454855e+00 -1.07546039e-01
4.48728889e-01 6.18222415e-01 -4.74792004e-01 -7.00034678e-01
1.71520282e-02 1.36354446e+00 7.80506551e-01 4.65488881e-01
-7.47770846e-01 -3.58087033e-01 1.18681192e+00 -2.52455741e-01
3.45333338e-01 -9.01751697e-01 2.24902704e-01 -1.26486039e+00
-2.86597818e-01 -6.84157252e-01 2.12234065e-01 -8.80098581e-01
-6.20152116e-01 2.35907003e-01 3.10156852e-01 -1.07243741e+00
-7.40694463e-01 -4.70653594e-01 -5.89836776e-01 6.57379150e-01
-1.30563736e+00 -8.34254324e-01 -9.45493057e-02 5.14881551e-01
9.86990750e-01 9.45781469e-02 1.22437418e+00 -5.80599725e-01
-1.78180188e-01 -2.87209451e-01 1.70476422e-01 5.29473498e-02
2.91455537e-01 -1.52765012e+00 9.27362263e-01 7.45744705e-01
6.33375272e-02 6.59914017e-01 8.53612661e-01 -5.98832667e-01
-1.57405794e+00 -9.73116159e-01 3.23937088e-01 -5.34937143e-01
7.12718546e-01 -5.39062440e-01 -5.23094893e-01 1.38701832e+00
3.93924296e-01 -9.51948911e-02 1.45942554e-01 2.22336426e-02
-1.05465129e-01 4.05669481e-01 -1.05986166e+00 9.36207235e-01
1.35936785e+00 -5.54805100e-01 -9.25922155e-01 3.85383338e-01
8.62986624e-01 -7.77537644e-01 -8.00848961e-01 -1.82423219e-01
5.03633142e-01 -8.37132514e-01 8.18918169e-01 -1.01234996e+00
6.38420701e-01 -2.67215699e-01 -1.28229618e-01 -1.91915393e+00
-4.91275936e-01 -9.09111321e-01 -5.97511053e-01 9.85115170e-02
2.15436086e-01 -7.28011549e-01 7.79367805e-01 5.10898113e-01
-2.93058991e-01 -7.31323242e-01 -8.43663275e-01 -9.00227726e-01
2.31514350e-01 -2.62960136e-01 6.24196351e-01 3.08074147e-01
4.40271974e-01 -3.37856710e-02 -1.49985760e-01 4.16778028e-01
7.94325054e-01 2.73505419e-01 9.21014309e-01 -7.56968915e-01
-4.51895028e-01 -4.35413867e-02 -4.21109647e-01 -1.43786871e+00
3.50074232e-01 -6.59584939e-01 5.41769564e-01 -1.82575428e+00
-1.18653193e-01 -3.07794511e-01 -2.07855821e-01 5.16801000e-01
5.33956587e-01 -4.75588411e-01 3.62653047e-01 6.31651059e-02
-9.87802744e-01 9.89471257e-01 1.30381441e+00 -7.67329484e-02
-3.05282563e-01 -1.24669127e-01 -1.59830898e-01 7.24005759e-01
1.26394856e+00 -5.75777709e-01 -4.96146113e-01 -6.14227235e-01
4.32448126e-02 6.39202118e-01 4.48802352e-01 -1.31763721e+00
3.55273128e-01 -6.11571014e-01 1.76572576e-01 -4.39432293e-01
8.50272298e-01 -6.11826599e-01 -1.13889337e-01 1.09423137e+00
-7.14513302e-01 1.77130610e-01 7.00203627e-02 8.68550241e-01
7.67867342e-02 -1.10694557e-01 6.47321761e-01 -7.66765952e-01
-1.05084705e+00 4.27962631e-01 -9.01531875e-01 4.75357510e-02
1.12967694e+00 2.60434616e-02 2.93549821e-02 -6.13658845e-01
-1.15357280e+00 5.06588340e-01 7.83502281e-01 3.14493895e-01
7.90840864e-01 -1.18480778e+00 -4.68774319e-01 1.44675612e-01
6.68067932e-02 4.98521149e-01 -2.67695710e-02 3.40019643e-01
-6.88863516e-01 2.71446139e-01 -6.72832668e-01 -5.35619199e-01
-5.72571635e-01 5.52333295e-01 4.11744982e-01 -4.08020109e-01
-7.61940718e-01 6.70434117e-01 -7.91796446e-02 -7.39032865e-01
2.41533101e-01 -8.40475857e-01 3.73012483e-01 -5.78930318e-01
5.82474232e-01 2.67062247e-01 -4.69696343e-01 2.75594257e-02
-1.00268953e-01 -1.51461819e-02 -4.81384946e-03 -7.95995414e-01
1.42740011e+00 -7.42139593e-02 1.83854580e-01 6.97771847e-01
1.06357062e+00 -8.55572104e-01 -2.01184320e+00 -2.19916508e-01
-1.61295876e-01 -4.00934666e-01 -9.50068980e-03 -4.31137949e-01
-4.22622040e-02 9.60188687e-01 2.89753079e-01 6.41279221e-02
4.17650044e-01 1.05605617e-01 7.38540292e-01 9.54440653e-01
9.81558859e-01 -1.28514552e+00 5.61017275e-01 9.46043193e-01
9.87074435e-01 -1.20929646e+00 -3.42468292e-01 5.43500304e-01
-7.32955337e-01 9.84357119e-01 5.65066934e-01 -6.70878708e-01
4.33090180e-01 4.34889823e-01 -2.12931052e-01 -1.66118875e-01
-1.24403918e+00 -4.08207774e-01 -6.35754347e-01 8.74572754e-01
-3.17894220e-01 2.11547941e-01 5.88722706e-01 -7.27389082e-02
-1.85112640e-01 2.54933536e-01 8.13123047e-01 1.64855349e+00
-8.18910062e-01 -8.67666483e-01 -1.32477865e-01 1.45173877e-01
1.42089829e-01 2.06017375e-01 -4.14648503e-02 5.17889857e-01
-4.39604312e-01 6.96666420e-01 3.18588942e-01 -6.39656112e-02
2.00848743e-01 1.92843527e-01 9.00064170e-01 -1.11919606e+00
-4.39145900e-02 -5.16749024e-01 2.10900381e-02 -1.40335536e+00
7.88576081e-02 -8.68019462e-01 -1.65704250e+00 -3.13062400e-01
4.04947877e-01 1.12379044e-01 7.98778892e-01 7.81128466e-01
4.69850272e-01 4.39367771e-01 4.26938623e-01 -1.61939657e+00
-1.24337363e+00 -7.81952679e-01 -4.04690266e-01 3.33941936e-01
8.35879087e-01 -6.79871678e-01 -2.22736791e-01 -7.13759009e-03] | [4.412838935852051, 1.0712003707885742] |
cc58acd9-b150-42cd-9a9e-d106f1c058bb | identification-of-energy-management | 2306.08318 | null | https://arxiv.org/abs/2306.08318v1 | https://arxiv.org/pdf/2306.08318v1.pdf | Identification of Energy Management Configuration Concepts from a Set of Pareto-optimal Solutions | Optimizing building configurations for an efficient use of energy is increasingly receiving attention by current research and several methods have been developed to address this task. Selecting a suitable configuration based on multiple conflicting objectives, such as initial investment cost, recurring cost, robustness with respect to uncertainty of grid operation is, however, a difficult multi-criteria decision making problem. Concept identification can facilitate a decision maker by sorting configuration options into semantically meaningful groups (concepts), further introducing constraints to meet trade-off expectations for a selection of objectives. In this study, for a set of 20000 Pareto-optimal building energy management configurations, resulting from a many-objective evolutionary optimization, multiple concept identification iterations are conducted to provide a basis for making an informed investment decision. In a series of subsequent analysis steps, it is shown how the choice of description spaces, i.e., the partitioning of the features into sets for which consistent and non-overlapping concepts are required, impacts the type of information that can be extracted and that different setups of description spaces illuminate several different aspects of the configuration data - an important aspect that has not been addressed in previous work. | ['Sebastian Schmitt', 'Yaochu Jin', 'Qiqi Liu', 'Felix Lanfermann'] | 2023-06-14 | null | null | null | null | ['management', 'energy-management'] | ['miscellaneous', 'time-series'] | [ 9.73312482e-02 -5.31957805e-01 1.41115859e-01 -4.03640807e-01
-3.54338050e-01 -6.95662916e-01 1.89998358e-01 9.00264561e-01
-3.17483902e-01 7.15028703e-01 1.60933748e-01 -2.22970590e-01
-9.60792422e-01 -9.35396850e-01 2.00297479e-02 -7.91572452e-01
-4.43974845e-02 5.62025487e-01 -2.20046505e-01 -7.62800425e-02
5.46302080e-01 7.01269269e-01 -1.92184126e+00 -3.51981107e-05
9.33596671e-01 9.94955540e-01 4.33840305e-01 1.60294384e-01
-8.43567699e-02 -8.25808011e-03 -7.22253501e-01 2.01256752e-01
1.41865313e-01 -4.81816143e-01 -6.49464011e-01 2.71675050e-01
-6.60496294e-01 2.92280555e-01 1.03650618e+00 9.39922810e-01
6.49355054e-01 2.31777772e-01 7.83757985e-01 -1.11040950e+00
2.19836961e-02 5.51756322e-01 -1.02862746e-01 1.50948480e-01
2.55783767e-01 6.03247201e-03 1.08964396e+00 -4.86901879e-01
2.16429010e-01 9.01641607e-01 1.88947499e-01 -2.10593358e-01
-1.53572440e+00 -4.22950208e-01 1.55497953e-01 3.62025589e-01
-1.77293181e+00 -9.23715979e-02 9.30902839e-01 -5.17863631e-01
8.21001351e-01 6.84317887e-01 1.02764380e+00 1.66097909e-01
9.43139046e-02 7.44853094e-02 1.19811618e+00 -7.37468362e-01
9.11317229e-01 4.09632027e-01 -7.80813470e-02 -1.14454530e-01
7.07695663e-01 8.83101895e-02 -3.34402174e-02 -1.39337718e-01
-4.60505001e-02 -4.65602547e-01 -3.88261974e-01 -5.45791090e-01
-6.13243580e-01 1.03284681e+00 2.23169968e-01 8.00658703e-01
-4.50725079e-01 -2.59397686e-01 1.99774042e-01 1.15812905e-01
1.22706637e-01 9.31236684e-01 -3.73623639e-01 6.64485851e-03
-8.98095310e-01 7.60322213e-02 6.38430953e-01 5.55836499e-01
8.21628630e-01 -3.64523679e-02 -1.01156600e-01 5.21280169e-01
3.51778656e-01 -3.60225700e-02 2.52641886e-01 -4.49475437e-01
3.74271631e-01 9.41814065e-01 2.25040540e-01 -8.74200523e-01
-5.44327676e-01 -6.42491519e-01 -5.41572988e-01 4.51236099e-01
5.65930493e-02 -3.54430616e-01 -4.87681597e-01 1.43654048e+00
3.32172334e-01 -7.85883963e-01 -4.95169871e-02 8.36295068e-01
1.50813386e-01 5.63646436e-01 1.83703646e-01 -5.67291677e-01
1.29512775e+00 -1.16258353e-01 -6.30257607e-01 4.29539755e-02
2.47855753e-01 -7.12470472e-01 6.25344932e-01 2.63559401e-01
-8.24681401e-01 -2.97478318e-01 -1.33087146e+00 7.85413921e-01
-5.77438593e-01 9.74475145e-02 3.26711297e-01 9.34482098e-01
-8.28638315e-01 6.38683379e-01 -4.88435715e-01 -3.86005759e-01
-1.51475996e-01 4.99039322e-01 2.08236948e-01 1.68728694e-01
-1.09052765e+00 1.25045061e+00 9.46581125e-01 2.03349784e-01
-3.62523794e-01 -4.00648057e-01 -4.31783229e-01 5.05494356e-01
6.69164777e-01 -5.86303592e-01 5.93805730e-01 -7.89370477e-01
-1.27878225e+00 1.76574543e-01 2.67485261e-01 -1.87375784e-01
3.97669286e-01 4.11266595e-01 -5.82908392e-01 -4.94001247e-02
-5.08706346e-02 4.89388295e-02 5.01947820e-01 -1.34332204e+00
-8.12694073e-01 -4.38189119e-01 3.10815051e-02 5.06340623e-01
-2.03479677e-01 3.49656761e-01 -1.02900386e-01 -4.99296367e-01
-1.28663005e-02 -8.05781901e-01 -4.00599271e-01 -6.49317563e-01
-2.32430130e-01 -1.03783675e-01 3.82920921e-01 -3.59510392e-01
1.59231675e+00 -1.82827318e+00 5.79693198e-01 8.23480666e-01
-4.02296424e-01 -1.46126717e-01 4.36004341e-01 6.66399896e-01
-1.19245149e-01 4.24074143e-01 -5.49417317e-01 4.63056900e-02
1.66326895e-01 4.73490171e-02 4.39419985e-01 1.84456468e-01
1.90721318e-01 1.49141476e-01 -5.23578465e-01 -3.56340110e-01
5.39199054e-01 2.39853948e-01 -1.87258303e-01 9.61082056e-02
-2.24137738e-01 3.47580671e-01 -7.26027012e-01 7.28360772e-01
3.02647650e-01 -7.54371583e-02 4.13575888e-01 -1.76137894e-01
-5.09559155e-01 -9.18365866e-02 -1.61780298e+00 1.00528669e+00
-5.55490732e-01 2.81735033e-01 -2.93776230e-03 -1.17405176e+00
8.94083917e-01 3.15823793e-01 8.32905173e-01 -5.90800345e-01
3.92927229e-01 2.41308674e-01 2.69744039e-01 -2.33892892e-02
6.28211021e-01 -3.40370655e-01 -1.80738285e-01 5.29964447e-01
-2.61359662e-01 -3.60130966e-01 6.13126397e-01 -6.45370364e-01
5.22163987e-01 -1.90643713e-01 5.96426427e-01 -7.89788187e-01
5.83206415e-01 2.02471644e-01 7.02779651e-01 2.25098893e-01
1.75999328e-01 4.51818109e-01 1.88577950e-01 -1.18815742e-01
-7.54161894e-01 -4.90828812e-01 -4.11377430e-01 3.85527909e-01
1.18572220e-01 -2.83872604e-01 -5.94994724e-01 -1.24089859e-01
-5.91798080e-03 1.40376329e+00 -3.66364926e-01 -1.43903093e-02
-3.82290095e-01 -1.23712158e+00 -3.78191531e-01 8.02525803e-02
1.17968708e-01 -8.90963376e-01 -1.69093406e+00 5.27986586e-01
-5.95285334e-02 -4.52031821e-01 4.44596857e-02 6.93097651e-01
-8.07256460e-01 -1.18261385e+00 -4.56011802e-01 -1.21876471e-01
8.72128010e-01 1.96969435e-02 1.29334462e+00 1.85344473e-01
-4.53252435e-01 3.10681731e-01 -5.96348286e-01 -5.43285251e-01
-4.83000547e-01 4.74782698e-02 -2.10725650e-01 -9.12506357e-02
1.62629783e-01 -5.10202110e-01 -4.90826190e-01 4.92092133e-01
-9.00766313e-01 -7.17908815e-02 6.32698715e-01 4.59735394e-01
6.79433703e-01 1.19008076e+00 4.70425785e-01 -1.75998777e-01
8.75728965e-01 -4.33152527e-01 -9.90715981e-01 7.55426884e-01
-9.76296425e-01 4.82569009e-01 5.83682895e-01 -2.90971305e-02
-9.95687723e-01 -2.48938173e-01 1.58178046e-01 -4.04727040e-03
-1.13389485e-01 8.02605867e-01 -5.18694401e-01 -8.92440788e-03
1.19801730e-01 -3.25735547e-02 -2.87492841e-01 -5.03000081e-01
1.11288153e-01 3.40665340e-01 -7.76017010e-02 -8.17449868e-01
7.46715069e-01 -1.25429302e-01 1.95771158e-01 -5.89976132e-01
-2.70235240e-01 -4.39339489e-01 -7.14295089e-01 -3.40569556e-01
8.68426323e-01 -4.25856203e-01 -4.56308931e-01 1.66448373e-02
-5.97396731e-01 1.75265208e-01 -4.67541814e-01 3.64958912e-01
-4.50348258e-01 -3.16286534e-02 6.37780488e-01 -9.95035470e-01
-2.48311415e-01 -1.58327901e+00 2.08699346e-01 3.28608990e-01
-5.70823193e-01 -7.63461888e-01 -9.90213379e-02 2.15768039e-01
5.81327200e-01 6.46593869e-01 1.42638755e+00 -5.21927655e-01
-5.39357781e-01 7.96214119e-03 3.52640539e-01 2.48371929e-01
5.43174922e-01 1.52840123e-01 -3.98855448e-01 -5.10746777e-01
2.74234533e-01 1.80598885e-01 3.70778322e-01 5.47018230e-01
1.00311911e+00 -2.48672456e-01 -2.73991138e-01 1.46658853e-01
1.93011928e+00 8.33344579e-01 4.32022661e-01 6.81733072e-01
-8.93355086e-02 1.00913751e+00 9.13710356e-01 7.27927387e-01
1.11595973e-01 9.11813080e-01 5.99375546e-01 8.60731751e-02
4.10576642e-01 4.80986625e-01 -5.91222011e-02 2.67408699e-01
-1.86550632e-01 -3.39923114e-01 -8.42489541e-01 5.34070373e-01
-1.54807436e+00 -7.85099506e-01 4.99738872e-01 2.48755741e+00
4.78380412e-01 2.05232084e-01 3.23008925e-01 5.92116773e-01
8.80856872e-01 -9.67523269e-03 -4.30355608e-01 -5.39777100e-01
2.13325825e-02 3.58673893e-02 4.07352954e-01 2.16160119e-01
-7.00482845e-01 -4.67738770e-02 5.44277573e+00 5.89143574e-01
-9.68745172e-01 -4.41731960e-01 7.34434307e-01 -2.63825864e-01
-6.21388614e-01 2.03081906e-01 -4.33708906e-01 7.04266310e-01
8.19310367e-01 -7.58616924e-01 4.60239530e-01 4.48603660e-01
5.42032063e-01 -6.44022584e-01 -9.84549403e-01 5.31527340e-01
-2.24545166e-01 -1.18964911e+00 -2.27361575e-01 2.59850889e-01
7.47175813e-01 -5.26499391e-01 -1.24877945e-01 -2.34910190e-01
3.29113513e-01 -1.08753383e+00 1.06140256e+00 3.28239113e-01
2.84611613e-01 -1.33041084e+00 8.77476752e-01 2.33882234e-01
-1.40864193e+00 -6.56864882e-01 9.39659327e-02 1.84709042e-01
3.91059041e-01 6.07385933e-01 -5.14431059e-01 1.28110254e+00
8.15893888e-01 -1.49964988e-01 -4.74047691e-01 1.15759897e+00
2.23788712e-02 3.50037664e-01 -5.84582150e-01 -3.42231959e-01
1.78991467e-01 -5.83348036e-01 5.30241311e-01 8.11227560e-01
7.65860736e-01 1.89708382e-01 7.70058185e-02 1.23554683e+00
5.89494526e-01 4.14342761e-01 -3.93728279e-02 -5.00159711e-02
6.98426306e-01 1.15544236e+00 -1.22002232e+00 2.15086758e-01
-1.05128244e-01 1.16719887e-01 -2.20539391e-01 1.40779629e-01
-5.72661459e-01 -2.55773634e-01 5.58846712e-01 1.53090417e-01
4.66876328e-01 -2.14452907e-01 -4.74550605e-01 -5.82959771e-01
4.55104485e-02 -7.01990187e-01 5.41284442e-01 -3.89910668e-01
-8.36152315e-01 3.93665642e-01 6.25327289e-01 -1.16250420e+00
-3.83402884e-01 -1.71880618e-01 -9.03421223e-01 1.05978239e+00
-1.22715437e+00 -5.02207994e-01 -2.12304428e-01 3.96257192e-02
3.83058369e-01 -1.37368757e-02 8.54316831e-01 -9.37325731e-02
-6.87620044e-01 -1.95008200e-02 4.47220922e-01 -4.64348078e-01
-4.53634746e-02 -1.16081882e+00 -3.00349921e-01 8.70008767e-01
-1.41438678e-01 2.61206120e-01 1.01791298e+00 -4.83312428e-01
-1.00035787e+00 -7.49026120e-01 7.00044215e-01 2.67598331e-01
3.67208093e-01 6.24216311e-02 -6.29565656e-01 -2.40465850e-02
1.96796030e-01 -7.30107486e-01 8.73243928e-01 5.15239686e-02
5.48866510e-01 -1.29740715e-01 -1.47893608e+00 4.98626322e-01
3.87059987e-01 2.22531855e-02 -4.22766984e-01 -1.76175162e-01
2.05669075e-01 2.20576793e-01 -1.15853667e+00 4.68482226e-01
1.78779364e-01 -1.12467098e+00 9.05802131e-01 -7.53775761e-02
-2.17074484e-01 -5.75636744e-01 -3.45203489e-01 -1.60879850e+00
-4.88891244e-01 -3.14289212e-01 3.50870699e-01 1.44868875e+00
4.63585019e-01 -5.59388936e-01 4.00355548e-01 8.77166748e-01
2.26563998e-02 -9.85354722e-01 -1.11815739e+00 -7.15586305e-01
-4.02623788e-02 -1.48331746e-01 1.18851495e+00 5.26726544e-01
-2.33884782e-01 3.08401622e-02 2.90308982e-01 1.17130771e-01
5.78364193e-01 6.31899774e-01 1.49395347e-01 -1.29724634e+00
-3.96379344e-02 -8.45838070e-01 -3.15573484e-01 8.14719722e-02
-3.51125270e-01 -5.20190418e-01 -8.12896043e-02 -1.67601466e+00
4.77849413e-03 -8.03605616e-01 -4.76464033e-01 2.65338331e-01
-1.40170595e-02 -4.48791295e-01 4.85593766e-01 -9.96960234e-03
-5.92438579e-02 7.34046221e-01 4.72179085e-01 -1.81398794e-01
-4.75116521e-01 2.84768283e-01 -9.00535107e-01 5.77899754e-01
9.39656198e-01 -4.58341420e-01 -5.88824749e-01 -3.31076793e-02
2.68755227e-01 1.58759952e-01 -4.36909571e-02 -1.14982140e+00
1.54306805e-02 -7.08868086e-01 3.45826805e-01 -7.31131911e-01
1.24537967e-01 -1.33761621e+00 1.06587648e+00 5.11495173e-01
-7.72404075e-02 4.23003346e-01 2.69515932e-01 3.72059405e-01
-1.91437766e-01 -7.76779711e-01 6.53067946e-01 -1.43453240e-01
-9.39885616e-01 -1.60219699e-01 -5.00157893e-01 -2.12355971e-01
1.42715418e+00 -5.38976490e-01 3.84332299e-01 -1.69934660e-01
-7.70007908e-01 4.83421594e-01 6.69194520e-01 3.87219846e-01
1.32677972e-01 -1.11832643e+00 -7.53235459e-01 -2.93327630e-01
1.36364564e-01 -2.70825997e-02 1.59834892e-01 3.26058358e-01
-3.30554485e-01 5.50792813e-01 -2.57946044e-01 -4.49433118e-01
-1.28828514e+00 3.76228333e-01 3.58489454e-01 -5.38035393e-01
-3.38074595e-01 3.21046472e-01 -3.40140790e-01 1.79738566e-01
-6.32458255e-02 -4.06515628e-01 -6.17893457e-01 7.56404161e-01
5.92222810e-02 6.37912273e-01 4.38211530e-01 -6.37239814e-01
-5.43842435e-01 7.32409716e-01 5.55463672e-01 -1.07298858e-01
1.50402319e+00 -3.01059067e-01 -9.94078293e-02 3.15220475e-01
7.89041221e-01 -2.52931565e-01 -9.40283477e-01 2.26294115e-01
4.59008187e-01 -4.65754449e-01 1.52550578e-01 -1.01589620e+00
-1.04263175e+00 2.24998668e-01 6.50380254e-01 5.78606725e-01
1.67746961e+00 -2.52133489e-01 -2.01586083e-01 -7.86722675e-02
6.01213932e-01 -1.51595819e+00 -2.31682226e-01 -1.68093726e-01
1.09520912e+00 -8.81415129e-01 3.65624815e-01 -1.67661160e-01
-4.66081917e-01 1.16937518e+00 1.84424907e-01 4.78524238e-01
6.00503623e-01 1.72630310e-01 -4.10817206e-01 -1.86691537e-01
-4.04553384e-01 -2.84652978e-01 3.21072787e-01 2.75447279e-01
8.89343917e-02 3.98350567e-01 -6.69296026e-01 4.90548551e-01
-2.25149795e-01 -3.21544528e-01 2.53917068e-01 1.16119862e+00
-4.53558683e-01 -1.22422361e+00 -7.27491319e-01 4.16745573e-01
-1.00605950e-01 1.47145271e-01 -2.57924676e-01 8.29125941e-01
4.61640596e-01 1.28984869e+00 -3.11648864e-02 -1.07157253e-01
4.12708938e-01 -1.15139060e-01 2.40844563e-01 -5.98589718e-01
-7.21477568e-01 -4.46245894e-02 4.11377579e-01 9.61077288e-02
-4.39039111e-01 -9.76813614e-01 -1.11871505e+00 3.22353537e-03
-7.75910079e-01 6.58658624e-01 9.96229947e-01 1.13732255e+00
2.09444419e-01 7.95830667e-01 7.67740369e-01 -7.37449527e-01
-6.80713058e-01 -5.90922356e-01 -5.50145090e-01 1.38937294e-01
-2.12413624e-01 -9.19421971e-01 -4.90123689e-01 -4.37565178e-01] | [5.8413004875183105, 3.515735149383545] |
131dca57-0cc2-4a64-bca5-4c38c7cdcaaf | birds-of-a-feather-flock-together-category | 2204.02111 | null | https://arxiv.org/abs/2204.02111v1 | https://arxiv.org/pdf/2204.02111v1.pdf | Birds of A Feather Flock Together: Category-Divergence Guidance for Domain Adaptive Segmentation | Unsupervised domain adaptation (UDA) aims to enhance the generalization capability of a certain model from a source domain to a target domain. Present UDA models focus on alleviating the domain shift by minimizing the feature discrepancy between the source domain and the target domain but usually ignore the class confusion problem. In this work, we propose an Inter-class Separation and Intra-class Aggregation (ISIA) mechanism. It encourages the cross-domain representative consistency between the same categories and differentiation among diverse categories. In this way, the features belonging to the same categories are aligned together and the confusable categories are separated. By measuring the align complexity of each category, we design an Adaptive-weighted Instance Matching (AIM) strategy to further optimize the instance-level adaptation. Based on our proposed methods, we also raise a hierarchical unsupervised domain adaptation framework for cross-domain semantic segmentation task. Through performing the image-level, feature-level, category-level and instance-level alignment, our method achieves a stronger generalization performance of the model from the source domain to the target domain. In two typical cross-domain semantic segmentation tasks, i.e., GTA5 to Cityscapes and SYNTHIA to Cityscapes, our method achieves the state-of-the-art segmentation accuracy. We also build two cross-domain semantic segmentation datasets based on the publicly available data, i.e., remote sensing building segmentation and road segmentation, for domain adaptive segmentation. | ['Changhu Wang', 'Zehuan Yuan', 'Shuai Shao', 'Danpei Zhao', 'Bo Yuan'] | 2022-04-05 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 5.17931700e-01 -8.10873881e-02 -1.64550647e-01 -7.42087901e-01
-7.50170588e-01 -6.54132128e-01 4.33571666e-01 2.51780957e-01
-3.89845580e-01 4.69050646e-01 -3.96272503e-02 7.31077641e-02
-4.27697182e-01 -1.20027113e+00 -4.79595363e-01 -8.06985140e-01
4.30578560e-01 4.88102704e-01 4.95025724e-01 -3.61678861e-02
2.08224669e-01 3.27192068e-01 -1.55707562e+00 3.38849276e-01
1.67264891e+00 1.00546002e+00 4.24735069e-01 5.57171553e-03
-5.20578504e-01 -3.92925739e-02 -4.17998761e-01 9.66815650e-02
3.49581808e-01 -3.27214807e-01 -8.73642087e-01 6.20222211e-01
4.03557599e-01 1.79683536e-01 2.21785977e-01 1.22496665e+00
3.30869079e-01 2.00926006e-01 8.33688259e-01 -1.16457450e+00
-5.69926023e-01 4.41190183e-01 -7.13673472e-01 6.84979558e-03
-1.93826124e-01 6.10934012e-02 9.14189637e-01 -7.44736969e-01
5.11337698e-01 1.16907656e+00 5.92789710e-01 3.84846449e-01
-1.36225951e+00 -7.90466905e-01 6.65198505e-01 3.43453139e-02
-1.46902347e+00 -6.23029545e-02 9.75602150e-01 -6.27131820e-01
4.44727987e-01 5.95469512e-02 3.71882230e-01 6.66029930e-01
-4.53218162e-01 6.76593602e-01 1.29029214e+00 -1.75947547e-01
4.53148156e-01 2.62725860e-01 2.68449932e-01 1.47611216e-01
1.55564472e-01 -1.62927344e-01 -1.44952551e-01 1.28443018e-01
4.83377308e-01 1.00353763e-01 5.10433614e-02 -6.28937483e-01
-1.01429987e+00 8.54011953e-01 6.21092319e-01 4.31171775e-01
-2.81191140e-01 -6.31258607e-01 3.96662384e-01 4.51988988e-02
6.93616509e-01 4.30978745e-01 -7.99882829e-01 3.23220611e-01
-9.51279819e-01 1.99372426e-01 2.91669518e-01 1.01066720e+00
1.12051582e+00 -1.49827212e-01 -1.33781120e-01 1.40266037e+00
3.45635772e-01 4.65552866e-01 6.20098174e-01 -7.22872496e-01
6.23978078e-01 1.14185047e+00 -1.27306536e-01 -9.33248997e-01
-3.72212559e-01 -6.81331694e-01 -8.03163588e-01 3.24981771e-02
4.33476299e-01 -4.91833985e-02 -1.23778808e+00 1.86475694e+00
6.62188232e-01 1.62190363e-01 3.02822262e-01 8.60486269e-01
8.76350164e-01 6.07753098e-01 5.24821579e-01 -3.80137973e-02
1.30723500e+00 -8.78036797e-01 -2.84569025e-01 -4.52622920e-01
6.32881224e-01 -4.78522569e-01 1.30612016e+00 5.01955375e-02
-6.54137254e-01 -8.48148465e-01 -8.97681415e-01 2.50464201e-01
-7.44124472e-01 1.01379275e-01 3.54912817e-01 4.44488436e-01
-4.09032643e-01 4.27893341e-01 -4.87823069e-01 -6.14328384e-01
5.35778761e-01 1.86688200e-01 -3.00009906e-01 -1.68791004e-02
-1.26263714e+00 4.65809762e-01 8.03758919e-01 -3.18273216e-01
-5.27517736e-01 -8.78185689e-01 -8.47781539e-01 -1.80481840e-02
3.66952538e-01 -4.19359922e-01 8.15813780e-01 -1.28929174e+00
-1.24811935e+00 1.18194640e+00 -7.76582733e-02 -3.01417410e-01
3.08291972e-01 6.39774725e-02 -6.27698600e-01 -5.95360324e-02
5.64688444e-01 8.67953360e-01 6.05789721e-01 -1.44014823e+00
-1.08489788e+00 -7.11267591e-01 -1.54742315e-01 4.66213554e-01
-3.68037671e-01 -4.20195878e-01 -3.54377359e-01 -8.54184747e-01
6.06143475e-01 -8.29685450e-01 -2.21758574e-01 -2.70397067e-01
-2.98096269e-01 -2.93464601e-01 1.01994872e+00 -5.32842755e-01
1.23906887e+00 -2.38857532e+00 2.05435067e-01 4.29991901e-01
-4.79615740e-02 2.92122006e-01 -8.48668069e-02 -1.09053239e-01
-1.04308344e-01 6.81975782e-02 -1.00467372e+00 1.09051920e-01
-1.27871752e-01 3.69795978e-01 4.89232503e-02 1.35360971e-01
2.77987927e-01 4.54465479e-01 -8.37052405e-01 -7.24595547e-01
2.63035804e-01 -3.42205986e-02 -5.47832072e-01 1.44285560e-01
-3.01583052e-01 8.40535700e-01 -6.52893722e-01 6.63291276e-01
1.03153503e+00 -1.40014291e-01 6.31516576e-02 -1.26630098e-01
-9.55752060e-02 7.80119076e-02 -1.41432428e+00 1.64565194e+00
-5.54795027e-01 2.60699876e-02 -7.59350508e-02 -1.30707836e+00
1.25837779e+00 -1.24489233e-01 6.54017806e-01 -8.18152785e-01
-1.08729646e-01 4.54168886e-01 -2.66028009e-02 -3.02385241e-01
3.22767794e-01 -2.17917666e-01 -3.43602657e-01 5.14279259e-03
1.81225557e-02 -2.24826112e-01 9.01316851e-02 -1.86845109e-01
3.50578129e-01 1.86026514e-01 3.82216930e-01 -5.80548465e-01
7.69013703e-01 3.71641278e-01 8.14632535e-01 5.76891840e-01
-3.03215235e-01 6.68725073e-01 8.42956230e-02 -1.01018265e-01
-9.53552365e-01 -1.28810155e+00 -5.43582737e-01 1.18015790e+00
4.59821314e-01 2.44730994e-01 -1.00785375e+00 -9.24065411e-01
2.66299486e-01 6.80461347e-01 -5.02582371e-01 -2.95516700e-01
-2.72318423e-01 -8.44206393e-01 4.09609586e-01 5.75614929e-01
1.17466295e+00 -9.70136344e-01 -1.26325637e-01 3.15059423e-01
-3.22791070e-01 -1.06938529e+00 -4.42969441e-01 9.94139463e-02
-1.00249243e+00 -8.61465693e-01 -8.51910532e-01 -1.03576267e+00
7.12972224e-01 1.89566314e-01 9.53796685e-01 -2.74960637e-01
1.13186859e-01 2.61710640e-02 -4.05594915e-01 -2.01653644e-01
-2.59705573e-01 2.89753556e-01 -6.30167946e-02 3.02888572e-01
5.91444492e-01 -5.76419413e-01 -5.41716754e-01 7.85236061e-01
-8.77585471e-01 8.01857561e-03 4.52912927e-01 6.45979166e-01
1.01597297e+00 1.97977707e-01 7.22988188e-01 -1.01976609e+00
3.66194069e-01 -7.23698974e-01 -5.56171775e-01 2.32428044e-01
-6.88193500e-01 -9.42969844e-02 7.07933903e-01 -5.55581570e-01
-1.35954773e+00 2.43156999e-01 3.11986078e-02 -5.00362329e-02
-6.95830345e-01 4.97753531e-01 -7.97271073e-01 1.98334932e-01
7.59622753e-01 3.36187810e-01 -3.73002857e-01 -5.85201740e-01
4.07374918e-01 9.22589481e-01 6.07960045e-01 -7.90052235e-01
7.42925107e-01 5.27981579e-01 -3.35100740e-01 -6.63083851e-01
-9.28945124e-01 -8.91998827e-01 -1.03258598e+00 2.91018300e-02
1.07505190e+00 -9.94330347e-01 7.36090839e-02 6.74858928e-01
-7.39861548e-01 -3.30275983e-01 -4.71368700e-01 2.96634346e-01
-4.13408458e-01 4.22164142e-01 4.74865362e-02 -4.34175283e-01
-1.61894709e-01 -1.07823777e+00 1.22364688e+00 5.73343158e-01
-7.68757192e-03 -9.99539554e-01 -7.11087510e-02 3.12735528e-01
1.43980294e-01 3.02868247e-01 9.36309993e-01 -9.72036779e-01
-9.41476971e-02 1.47895008e-01 -6.34892106e-01 6.22839808e-01
4.76758778e-01 -1.93199575e-01 -8.33355844e-01 -8.15576389e-02
-2.78684407e-01 9.96495709e-02 8.45791459e-01 4.50044960e-01
1.38396657e+00 -8.81381333e-02 -4.13394123e-01 6.79092348e-01
1.45942974e+00 3.76952261e-01 4.95271236e-01 5.66930413e-01
8.21184933e-01 8.76122177e-01 1.01216340e+00 3.23798239e-01
4.38459992e-01 6.90912962e-01 1.59561366e-01 -3.52787107e-01
-4.29000743e-02 -2.91424692e-01 -1.01716124e-01 4.69372690e-01
1.77173346e-01 3.82441562e-03 -1.03672194e+00 9.41798210e-01
-1.76573193e+00 -6.32029712e-01 -2.19152391e-01 2.24366307e+00
8.45743895e-01 1.98681518e-01 2.13097662e-01 2.50555389e-02
1.03641880e+00 -6.99239299e-02 -8.36924970e-01 -2.11405411e-01
-1.90641209e-01 1.01398289e-01 6.09359920e-01 4.13741857e-01
-1.43455482e+00 1.21510601e+00 5.26029778e+00 1.20715082e+00
-1.00593865e+00 3.90456915e-02 6.57821059e-01 5.09058118e-01
-2.01470301e-01 -2.70489790e-02 -6.79593980e-01 7.03437626e-01
4.16625261e-01 -7.47038200e-02 2.08287224e-01 1.06870115e+00
2.35655963e-01 7.65839824e-03 -7.28591442e-01 7.29934394e-01
-2.95033604e-01 -9.14281964e-01 1.64055541e-01 4.99259979e-02
9.46781695e-01 -1.04797043e-01 -7.05660926e-03 3.87709767e-01
3.84469181e-01 -6.95042253e-01 5.16416967e-01 1.72156438e-01
8.08243215e-01 -7.89595604e-01 5.59790432e-01 4.10722166e-01
-1.47527206e+00 -1.61310732e-01 -3.86982679e-01 3.35363299e-01
-9.64075234e-03 6.21429563e-01 -4.14933890e-01 8.02678823e-01
8.30623925e-01 7.84697652e-01 -6.35867774e-01 8.41561854e-01
-9.21456069e-02 4.91584837e-01 -3.20270836e-01 5.20549417e-01
5.22835791e-01 -5.62735856e-01 5.23708820e-01 1.13607109e+00
2.33325124e-01 1.51508510e-01 4.95299160e-01 8.96142840e-01
1.64359763e-01 1.83853880e-01 -3.68160456e-01 3.02197456e-01
6.06070459e-01 1.05820549e+00 -7.33292222e-01 -5.84923148e-01
-1.94960400e-01 8.91123235e-01 9.16670486e-02 3.71382833e-01
-9.51651454e-01 -5.40258229e-01 8.12590718e-01 9.66405943e-02
2.25746989e-01 -2.74089947e-02 -8.47081363e-01 -9.40265596e-01
8.52800980e-02 -7.12782800e-01 8.39009821e-01 -5.30447841e-01
-1.49284065e+00 4.01878834e-01 2.56964505e-01 -1.46716666e+00
2.59849459e-01 -4.40135419e-01 -5.47097683e-01 8.91767323e-01
-1.66530418e+00 -1.21221983e+00 -5.51979721e-01 7.25560248e-01
6.44473016e-01 -1.26263425e-01 5.14460802e-01 5.15246034e-01
-3.91217709e-01 5.49292088e-01 2.96614587e-01 1.71212748e-01
7.19378233e-01 -1.17607653e+00 2.07157105e-01 8.19484413e-01
-2.83090591e-01 3.65843177e-01 4.17838901e-01 -6.66762650e-01
-3.57399702e-01 -1.57218003e+00 6.64023042e-01 -1.02722175e-01
5.78247666e-01 -1.07771046e-01 -1.34858692e+00 3.94718498e-01
-3.93946290e-01 -1.93161592e-01 5.38254380e-01 8.77183676e-02
-3.70305121e-01 -4.55229938e-01 -1.48972797e+00 4.14225996e-01
1.21773291e+00 -4.12182927e-01 -6.84034467e-01 3.24401915e-01
7.36435890e-01 -1.44205332e-01 -1.03019989e+00 6.40794218e-01
3.58372957e-01 -7.09109783e-01 1.01684248e+00 -5.63903511e-01
4.60791826e-01 -6.07092798e-01 -2.72094280e-01 -1.48174024e+00
-5.34116447e-01 1.47578776e-01 5.12504697e-01 1.59902573e+00
5.00384271e-01 -8.36853623e-01 7.35757709e-01 3.80979002e-01
-3.99708331e-01 -2.65053868e-01 -8.72090578e-01 -1.10560954e+00
4.78124410e-01 -2.95740694e-01 1.00864756e+00 1.15460956e+00
-3.58483672e-01 1.01246729e-01 1.37342617e-01 5.61922967e-01
6.78306043e-01 4.24818873e-01 6.91652656e-01 -1.54005992e+00
-5.70395403e-02 -6.66538656e-01 -2.07272336e-01 -9.08630788e-01
9.10024941e-02 -1.04277813e+00 1.63452715e-01 -1.62550294e+00
1.26453653e-01 -8.30191076e-01 -4.14872557e-01 4.62920845e-01
-2.65077829e-01 1.38216503e-02 9.81003866e-02 3.96268219e-01
-4.90319341e-01 4.47796822e-01 1.32856882e+00 -2.96971411e-01
-5.15015125e-01 4.19768458e-03 -8.96940291e-01 7.16144204e-01
9.72506046e-01 -5.11286080e-01 -3.21202397e-01 -4.58249837e-01
-4.86002922e-01 -4.32073832e-01 2.48319194e-01 -1.14248633e+00
-1.09967485e-01 -5.21291435e-01 1.41243324e-01 -5.02071679e-01
-1.77202567e-01 -1.04027879e+00 -4.19162512e-02 3.06479424e-01
-3.06829661e-01 -5.81023276e-01 1.92096785e-01 5.60342014e-01
-3.63499969e-01 -6.43999949e-02 1.29657090e+00 -1.26438498e-01
-1.27873230e+00 4.23034579e-01 -6.02785759e-02 3.19652438e-01
1.11893952e+00 -6.27352357e-01 -1.68147460e-01 2.16047987e-01
-9.55594957e-01 5.27410984e-01 5.87008238e-01 5.14879167e-01
1.73705116e-01 -1.36190879e+00 -8.23008239e-01 2.59545922e-01
7.14909732e-01 4.70160544e-01 5.23826241e-01 5.30029953e-01
-4.76579107e-02 8.96897167e-02 -2.79530317e-01 -1.00332785e+00
-1.13626730e+00 4.26212102e-01 4.73007768e-01 -4.03444618e-01
-3.52857322e-01 6.80055499e-01 7.70301044e-01 -1.05920517e+00
-1.49048448e-01 -2.71948159e-01 -4.62900072e-01 1.35310799e-01
1.25824735e-01 2.63397038e-01 2.69632350e-04 -8.59024704e-01
-4.79435086e-01 9.51396525e-01 2.70934641e-01 2.51355529e-01
1.08136356e+00 -2.93794841e-01 2.28551868e-02 2.93974727e-01
1.08716357e+00 -2.65051544e-01 -1.42547274e+00 -5.17033577e-01
2.75217116e-01 -4.16635633e-01 -2.25282729e-01 -1.02032685e+00
-1.08860421e+00 7.69466460e-01 8.37594271e-01 1.30903602e-01
1.38899767e+00 1.86956093e-01 7.13790894e-01 -1.17695905e-01
1.80206254e-01 -1.66660643e+00 -1.31692141e-01 5.38101017e-01
5.42693973e-01 -1.40212989e+00 -1.93572715e-01 -5.92568636e-01
-9.06343341e-01 7.02082992e-01 8.73444617e-01 -1.01520963e-01
5.20734489e-01 -2.38926426e-01 -6.39594160e-03 -8.65969732e-02
5.16948812e-02 -6.31411731e-01 4.84721392e-01 9.04955924e-01
1.17266178e-01 4.41145658e-01 -3.63998502e-01 8.44135523e-01
3.40936109e-02 -1.41405150e-01 -2.12916899e-02 5.76205432e-01
-6.89280808e-01 -1.02358389e+00 -5.39435685e-01 2.69505709e-01
4.20932397e-02 -1.57776922e-02 -3.40210140e-01 8.62733960e-01
5.85138977e-01 1.00732863e+00 2.41614878e-01 -3.18027616e-01
7.66745448e-01 1.56704664e-01 3.97449434e-02 -7.62041748e-01
-4.08271015e-01 1.06680952e-01 -1.27357334e-01 -1.99007154e-01
-6.57024026e-01 -6.47488832e-01 -1.50780666e+00 -2.36006025e-02
-2.65146673e-01 7.78881088e-02 4.29342985e-01 1.02648282e+00
4.71443653e-01 4.97727573e-01 8.36553097e-01 -4.78005439e-01
-2.76178926e-01 -8.54959428e-01 -7.20762432e-01 8.68394315e-01
-5.81923779e-03 -8.63770962e-01 -3.39889169e-01 1.74897872e-02] | [9.689699172973633, 1.412293791770935] |
a4a33756-3d96-409e-9a84-9cfb26875123 | registration-free-hybrid-learning-empowers | 2307.03425 | null | https://arxiv.org/abs/2307.03425v1 | https://arxiv.org/pdf/2307.03425v1.pdf | Registration-Free Hybrid Learning Empowers Simple Multimodal Imaging System for High-quality Fusion Detection | Multimodal fusion detection always places high demands on the imaging system and image pre-processing, while either a high-quality pre-registration system or image registration processing is costly. Unfortunately, the existing fusion methods are designed for registered source images, and the fusion of inhomogeneous features, which denotes a pair of features at the same spatial location that expresses different semantic information, cannot achieve satisfactory performance via these methods. As a result, we propose IA-VFDnet, a CNN-Transformer hybrid learning framework with a unified high-quality multimodal feature matching module (AKM) and a fusion module (WDAF), in which AKM and DWDAF work in synergy to perform high-quality infrared-aware visible fusion detection, which can be applied to smoke and wildfire detection. Furthermore, experiments on the M3FD dataset validate the superiority of the proposed method, with IA-VFDnet achieving the best detection performance than other state-of-the-art methods under conventional registered conditions. In addition, the first unregistered multimodal smoke and wildfire detection benchmark is openly available in this letter. | ['Yuanjie Gu', 'Shouyu Wang', 'Zekuan Yu', 'Haoran Dai', 'Yinghan Guan'] | 2023-07-07 | null | null | null | null | ['image-registration'] | ['computer-vision'] | [ 3.69834721e-01 -8.73297453e-01 2.42123559e-01 -2.16032416e-01
-8.47746313e-01 -3.00921530e-01 5.71334481e-01 -6.17201440e-02
-5.29420137e-01 9.80178863e-02 1.34260012e-02 -4.13871370e-02
-3.55596691e-01 -1.12458324e+00 -2.36533388e-01 -1.11057258e+00
2.35531896e-01 -1.39862686e-01 2.08508506e-01 -3.18766475e-01
-6.36596605e-02 4.33183789e-01 -1.77726054e+00 2.15939835e-01
1.01964319e+00 1.24934280e+00 6.35458171e-01 3.39567333e-01
-1.29097953e-01 3.14044505e-01 -2.08935484e-01 -2.74899036e-01
5.02600789e-01 -1.93178862e-01 -3.26488823e-01 1.74566228e-02
7.35780537e-01 -3.67410898e-01 -4.88920808e-01 1.39025331e+00
6.94700718e-01 3.79112922e-02 5.10719478e-01 -1.15430582e+00
-6.97314262e-01 3.75574231e-01 -7.30013251e-01 2.49011114e-01
-3.76249626e-02 2.91475266e-01 7.22624004e-01 -1.05628300e+00
1.28240883e-01 1.03372836e+00 7.79038370e-01 1.31088033e-01
-1.00956261e+00 -8.41834188e-01 -5.31228371e-02 7.29779974e-02
-1.62291074e+00 -3.18747818e-01 7.17310429e-01 -5.56129336e-01
3.85728359e-01 4.24918830e-01 7.07706213e-01 6.32143557e-01
1.67301059e-01 2.53429353e-01 1.23488748e+00 -9.89853367e-02
-1.43624365e-01 -2.91747272e-01 2.87022395e-03 7.76166141e-01
5.38631141e-01 4.78931755e-01 -3.52519095e-01 -1.09369643e-01
7.04260528e-01 6.90864861e-01 -6.15709126e-01 -2.68621021e-03
-1.36986971e+00 6.64424419e-01 9.02358532e-01 3.44948709e-01
-4.66886818e-01 -1.48858964e-01 1.80185676e-01 1.75902396e-01
5.22997439e-01 -1.42997131e-01 -2.14418799e-01 6.11350119e-01
-1.07443726e+00 2.05056235e-01 1.78930432e-01 5.28352082e-01
9.70778465e-01 -5.39698750e-02 -3.59441787e-01 8.29235077e-01
7.97972381e-01 1.11503649e+00 2.01147586e-01 -7.82199085e-01
4.98969078e-01 8.65350485e-01 -3.79725471e-02 -1.17766190e+00
-5.55789053e-01 -5.18522263e-01 -1.30731583e+00 4.09438133e-01
1.58187568e-01 -1.00368313e-01 -8.26615572e-01 1.48781550e+00
1.94232821e-01 5.55458963e-01 4.37647432e-01 1.34461844e+00
1.41846788e+00 5.46581388e-01 1.58205718e-01 -2.71304041e-01
1.51167631e+00 -6.92436337e-01 -5.93535841e-01 -1.43744573e-01
1.69277221e-01 -1.06228757e+00 6.67786837e-01 -8.96173865e-02
-5.44082463e-01 -8.07801604e-01 -8.90994787e-01 3.41259062e-01
-4.46928769e-01 4.22289103e-01 6.09302282e-01 3.65255445e-01
-7.51383662e-01 2.49675900e-01 -3.75400007e-01 -4.11957413e-01
4.51374382e-01 1.95951294e-02 -5.23591638e-01 -3.59088033e-01
-1.07064414e+00 7.32637942e-01 3.53178591e-01 8.92626166e-01
-9.09928799e-01 -6.45859241e-01 -7.08576918e-01 -1.41543180e-01
2.38313451e-01 -8.39672863e-01 8.48648131e-01 -4.87106562e-01
-9.37986851e-01 8.50612581e-01 7.61055723e-02 7.16283470e-02
3.50792378e-01 -1.45215154e-01 -8.47138345e-01 -1.49199842e-02
6.83102086e-02 3.22198838e-01 8.01792562e-01 -1.54096091e+00
-7.79384136e-01 -5.24015665e-01 -3.45399044e-02 3.40919256e-01
-1.99461728e-01 5.61639592e-02 3.35304849e-02 -5.56739151e-01
3.45236868e-01 -5.79160392e-01 -2.95660168e-01 9.37993377e-02
-1.82891086e-01 -1.76998284e-02 1.00300157e+00 -6.55844152e-01
1.07796323e+00 -2.33510494e+00 7.45346099e-02 2.50412673e-01
2.90356070e-01 5.97199857e-01 -3.52370590e-01 2.87564546e-01
5.41775078e-02 -6.97257444e-02 -6.19839787e-01 -1.17039777e-01
-1.51874110e-01 1.04296848e-01 -5.06924428e-02 6.58380747e-01
2.90856380e-02 7.66854346e-01 -8.83096039e-01 -6.17549181e-01
5.31746209e-01 6.21590674e-01 1.22275487e-01 3.66623819e-01
8.91432166e-02 4.90628302e-01 -4.37411278e-01 9.19057190e-01
1.15157104e+00 2.20205858e-02 -1.74084708e-01 -8.47949088e-01
-5.23278475e-01 -5.42392790e-01 -1.19167387e+00 1.64448237e+00
-3.78425837e-01 5.55777401e-02 3.90702754e-01 -7.14833558e-01
1.19459212e+00 3.54850680e-01 6.00464523e-01 -7.14381754e-01
3.51413459e-01 3.57702732e-01 -2.66511917e-01 -7.61190236e-01
2.24415660e-01 -1.29333943e-01 2.59614140e-01 2.35926751e-02
-1.09450504e-01 2.24781204e-02 -8.57150182e-02 -2.15577349e-01
7.53944576e-01 -5.29198386e-02 -1.90144945e-02 -5.62097467e-02
8.53802443e-01 -7.61429593e-02 6.25705302e-01 5.58442354e-01
-1.64814934e-01 6.45629346e-01 -3.60782355e-01 -3.04120123e-01
-6.86811090e-01 -1.12744617e+00 -2.40441173e-01 7.15251684e-01
6.91563427e-01 -2.55972832e-01 -2.57192045e-01 -5.57764173e-01
3.42310555e-02 2.05090135e-01 -2.80151784e-01 -1.15791269e-01
-2.97052860e-01 -1.11076784e+00 5.75186610e-01 3.79072726e-01
1.13526714e+00 -7.80655444e-01 -6.72845840e-01 9.74564906e-03
-4.14303780e-01 -9.36344683e-01 -3.67803931e-01 -7.76295662e-02
-7.53928542e-01 -1.36776114e+00 -6.37159586e-01 -8.13997030e-01
3.54776740e-01 1.08198512e+00 8.20678234e-01 3.63774389e-01
-3.21657747e-01 1.86812803e-01 -5.28775692e-01 -2.35601246e-01
-1.03629664e-01 -2.97486722e-01 -4.02143002e-01 3.62490624e-01
2.37947643e-01 -4.86748159e-01 -7.79258490e-01 2.43018657e-01
-1.09940565e+00 -5.50649269e-03 7.20273256e-01 8.67265940e-01
3.96739542e-01 1.92683473e-01 2.18302637e-01 -2.15702415e-01
1.97445646e-01 -3.75574023e-01 -7.11140037e-01 4.82704937e-01
-6.03106022e-01 -3.74222100e-01 1.06978841e-01 5.64007908e-02
-1.33024859e+00 3.23084533e-01 -3.46112661e-02 -5.81716478e-01
-3.34694296e-01 7.29131043e-01 -3.92211288e-01 -3.28892827e-01
6.09955251e-01 2.45931849e-01 1.26705453e-01 -7.47456849e-01
4.43139225e-01 8.07817817e-01 9.30123508e-01 -1.00284167e-01
1.29524922e+00 5.50362468e-01 1.36793926e-01 -6.18314862e-01
-7.75160193e-01 -7.56803870e-01 -7.06358373e-01 -5.48213243e-01
1.10697269e+00 -1.39000809e+00 -6.55735433e-01 9.40148056e-01
-1.22415233e+00 1.93355486e-01 1.50196403e-02 7.36281157e-01
3.96181718e-02 5.98342359e-01 -1.96196109e-01 -8.56344700e-01
-6.29007816e-01 -1.11834097e+00 1.33039892e+00 4.90692675e-01
7.00140178e-01 -6.47806823e-01 1.34246901e-01 6.13444626e-01
6.93279743e-01 3.14804703e-01 3.70418578e-01 -6.72367215e-02
-6.80304646e-01 -3.78999151e-02 -6.16340876e-01 4.22023982e-01
2.34860748e-01 1.64027333e-01 -1.07573831e+00 -1.56927481e-01
-2.71961391e-01 4.13159691e-02 1.37543309e+00 1.62053376e-01
7.47214258e-01 -2.22276687e-03 -3.09063226e-01 7.50639319e-01
1.75757110e+00 4.82535698e-02 6.68502271e-01 3.64662617e-01
9.10374224e-01 2.27100596e-01 7.01127768e-01 4.64347094e-01
4.78039324e-01 7.40713000e-01 1.07530093e+00 -6.68989241e-01
-1.91152856e-01 1.42583370e-01 2.99411863e-01 5.57734907e-01
-5.83680689e-01 -5.22975996e-02 -6.96092010e-01 2.63529718e-01
-2.04185009e+00 -1.30063617e+00 -5.51573694e-01 2.19250298e+00
4.25437003e-01 -6.02710962e-01 -1.64650038e-01 5.77868894e-02
9.85066831e-01 4.66255814e-01 -8.97970349e-02 6.08020306e-01
-6.00597560e-01 2.60416657e-01 7.01556444e-01 3.37237060e-01
-1.62276387e+00 5.18512011e-01 5.45702314e+00 6.98254943e-01
-1.12089205e+00 4.76733744e-01 -5.13140634e-02 3.75456750e-01
-2.09708139e-01 1.24464910e-02 -3.84147316e-01 3.65450799e-01
3.44092131e-01 2.31312588e-01 5.13190567e-01 5.10047495e-01
2.41488114e-01 3.92485484e-02 -2.73679078e-01 1.21848750e+00
6.07715584e-02 -1.13306987e+00 -1.99260697e-01 -5.25118411e-02
4.94517475e-01 3.15831959e-01 -2.49770507e-01 -1.42612770e-01
1.44381985e-01 -8.45400095e-01 5.89191377e-01 1.01442719e+00
5.28777719e-01 -4.79836226e-01 1.13872528e+00 2.59854198e-01
-1.84422791e+00 -1.86226144e-01 -5.04726827e-01 2.35792667e-01
2.34191701e-01 9.70248401e-01 2.23286618e-02 1.41735590e+00
8.28436494e-01 8.69678080e-01 -8.01548541e-01 1.40317571e+00
-6.64891899e-02 3.68089139e-01 -2.97157496e-01 5.99006414e-01
2.41383947e-02 -4.82266545e-01 4.39700484e-01 1.26609266e+00
7.54953980e-01 5.84196076e-02 7.26269186e-01 8.28485548e-01
2.39471629e-01 7.47593194e-02 -6.15357280e-01 2.44024485e-01
3.90654981e-01 1.85288453e+00 -3.62893909e-01 -1.71776503e-01
-5.72218478e-01 6.82963073e-01 -2.31777117e-01 2.88653255e-01
-8.68053019e-01 -9.72621888e-02 8.08849335e-01 -2.22936615e-01
2.00761318e-01 -3.30635667e-01 -1.22324917e-02 -1.27973628e+00
2.34216508e-02 -5.76607764e-01 5.24096191e-01 -9.66193020e-01
-1.65236950e+00 7.26254046e-01 -1.67130932e-01 -1.61847496e+00
4.34720635e-01 -5.17021775e-01 -8.55193317e-01 1.01960337e+00
-1.87634337e+00 -1.76793718e+00 -1.29181111e+00 8.09356689e-01
3.51449810e-02 -3.17195982e-01 8.20020854e-01 7.72113264e-01
-6.98187053e-01 5.88413551e-02 -2.57743895e-03 6.67215362e-02
6.81152225e-01 -7.66942799e-01 -2.42288426e-01 1.06302094e+00
1.53677166e-01 1.53646141e-01 2.20195800e-01 -5.66096187e-01
-1.40372205e+00 -1.59328282e+00 5.96058369e-01 5.94524853e-02
2.65682995e-01 1.66412696e-01 -9.72824514e-01 1.26352206e-01
3.08260530e-01 2.78570533e-01 5.32770991e-01 -3.23874772e-01
-4.81658876e-01 -5.66889584e-01 -9.61349308e-01 1.09935589e-01
1.11688900e+00 -5.32225966e-01 -4.15209770e-01 5.24116158e-01
4.46966499e-01 -2.37353474e-01 -1.12644112e+00 9.34396803e-01
4.42337900e-01 -1.13319111e+00 1.21058965e+00 -4.28215712e-02
2.21331432e-01 -1.15044641e+00 -6.90838575e-01 -1.19412446e+00
-6.21710896e-01 7.84869492e-02 2.45567366e-01 1.47870004e+00
4.73454259e-02 -6.17349148e-01 -4.77138758e-02 -1.20391428e-01
-1.87519744e-01 -2.00155169e-01 -8.84594262e-01 -7.12950647e-01
-4.66867447e-01 -3.22691351e-01 8.17695498e-01 9.12732601e-01
-5.55593252e-01 9.52977613e-02 -3.76667738e-01 7.95381844e-01
8.27281475e-01 4.89923894e-01 5.98066688e-01 -1.59792411e+00
5.30167893e-02 -5.36382318e-01 -4.15659606e-01 -4.66278613e-01
1.24292046e-01 -1.11676192e+00 2.18905136e-01 -1.75486958e+00
4.14658606e-01 -3.64157170e-01 -5.34324527e-01 6.62512422e-01
-2.61368781e-01 6.16831422e-01 3.53352606e-01 4.92926419e-01
-2.81720757e-01 8.20652187e-01 1.14312387e+00 -5.76072276e-01
1.06557056e-01 -9.28986073e-02 -5.41220427e-01 5.94899952e-01
8.17843974e-01 -2.54879147e-01 -5.23904637e-02 -4.81362939e-01
-2.28809625e-01 -8.39752033e-02 1.17321885e+00 -1.18333769e+00
4.43848729e-01 -3.42589170e-01 4.01740849e-01 -7.54160702e-01
2.26413995e-01 -1.20767212e+00 6.19632483e-01 4.15120959e-01
1.10926591e-01 1.17879756e-01 -9.74641889e-02 4.84438896e-01
-4.70585197e-01 -1.43026009e-01 7.98269570e-01 -8.10231790e-02
-9.37820077e-01 6.14346445e-01 -1.24682210e-01 -4.13345277e-01
9.29381013e-01 -9.16399527e-03 -8.15451026e-01 8.74713361e-02
-3.60091269e-01 5.13268113e-02 2.89359301e-01 4.89796042e-01
7.55420208e-01 -1.53768337e+00 -9.65381324e-01 2.55505383e-01
4.53918874e-01 1.26625180e-01 6.84374571e-01 1.15872288e+00
-3.19788843e-01 -2.21337438e-01 -1.92168370e-01 -7.10863888e-01
-1.44978642e+00 3.84130329e-01 5.77916443e-01 9.63761061e-02
-6.55024230e-01 4.08426821e-01 3.77416387e-02 -5.38196564e-01
-6.33419454e-02 -8.04282576e-02 -3.95619005e-01 1.57794014e-01
4.07981157e-01 3.21217805e-01 1.29896268e-01 -1.01484549e+00
-5.27908921e-01 1.05880630e+00 6.05029047e-01 2.54091859e-01
1.31753254e+00 -1.70024917e-01 -4.74147648e-01 4.91233282e-02
8.77976239e-01 -2.35717818e-01 -9.35537040e-01 -4.99241292e-01
-4.32801217e-01 -7.75367081e-01 5.52751780e-01 -8.00490379e-01
-1.74363780e+00 7.50859022e-01 1.27270079e+00 3.57381329e-02
1.51671362e+00 -1.37438491e-01 7.10507989e-01 2.42548391e-01
1.65205434e-01 -6.77102864e-01 -2.47677609e-01 2.80127615e-01
9.72775161e-01 -1.49426234e+00 -3.22203003e-02 -6.27049387e-01
-4.87189949e-01 1.20559657e+00 6.36403024e-01 1.82615787e-01
9.64398444e-01 1.89262629e-01 5.11773169e-01 -4.45332825e-01
-1.74915478e-01 -7.83237815e-01 4.43029493e-01 6.24723136e-01
3.32367688e-01 2.11362347e-01 -2.42064416e-01 4.15191621e-01
2.65423864e-01 -1.64263565e-02 -1.73396155e-01 7.37934053e-01
-7.05804825e-01 -8.96027267e-01 -7.16240823e-01 4.22363549e-01
9.21497941e-02 -1.23193428e-01 -1.60454616e-01 5.06633759e-01
8.00869286e-01 1.26728642e+00 -1.09850809e-01 -8.66800785e-01
5.75567961e-01 -1.38590127e-01 2.47178212e-01 -1.41353115e-01
-7.63516426e-01 2.13846147e-01 -2.29434252e-01 -4.92780477e-01
-1.15623856e+00 -5.76640606e-01 -9.35910642e-01 -3.94428968e-01
-6.43488884e-01 -1.77272916e-01 4.92006183e-01 9.38681364e-01
1.72789976e-01 5.35112381e-01 8.73398125e-01 -1.01000798e+00
-4.01254833e-01 -9.36918676e-01 -8.02441835e-01 4.42303479e-01
2.78787494e-01 -9.33293223e-01 -2.37536684e-01 -1.65922251e-02] | [10.230500221252441, -1.7408684492111206] |
7c3fe395-6bb6-441b-8398-ee1b17cfb466 | occlusion-robust-face-alignment-using-a | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Zhu_Occlusion-Robust_Face_Alignment_Using_a_Viewpoint-Invariant_Hierarchical_Network_Architecture_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Zhu_Occlusion-Robust_Face_Alignment_Using_a_Viewpoint-Invariant_Hierarchical_Network_Architecture_CVPR_2022_paper.pdf | Occlusion-Robust Face Alignment Using a Viewpoint-Invariant Hierarchical Network Architecture | The occlusion problem heavily degrades the localization performance of face alignment. Most current solutions for this problem focus on annotating new occlusion data, introducing boundary estimation, and stacking deeper models to improve the robustness of neural networks. However, the performance degradation of models remains under extreme occlusion (average occlusion of over 50%) because of missing a large amount of facial context information. We argue that exploring neural networks to model the facial hierarchies is a more promising method for dealing with extreme occlusion. Surprisingly, in recent studies, little effort has been devoted to representing the facial hierarchies using neural networks. This paper proposes a new network architecture called GlomFace to model the facial hierarchies against various occlusions, which draws inspiration from the viewpoint-invariant hierarchy of facial structure. Specifically, GlomFace is functionally divided into two modules: the part-whole hierarchical module and the whole-part hierarchical module. The former captures the part-whole hierarchical dependencies of facial parts to suppress multi-scale occlusion information, whereas the latter injects structural reasoning into neural networks by building the whole-part hierarchical relations among facial parts. As a result, GlomFace has a clear topological interpretation due to its correspondence to the facial hierarchies. Extensive experimental results indicate that the proposed GlomFace performs comparably to existing state-of-the-art methods, especially in cases of extreme occlusion. Models are available at https://github.com/zhuccly/GlomFace-Face-Alignment. | ['Yinzheng Gu', 'Xiaoqiang Li', 'Shaorong Xie', 'Xintong Wan', 'Congcong Zhu'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['robust-face-alignment', 'face-alignment'] | ['computer-vision', 'computer-vision'] | [-2.18837142e-01 4.29842085e-01 -3.20443898e-01 -6.94488823e-01
-1.64349955e-02 -1.75124317e-01 3.91332537e-01 -5.50375402e-01
2.12169960e-01 2.23861724e-01 3.58355910e-01 8.06469992e-02
-5.60465381e-02 -6.66627586e-01 -6.47194564e-01 -6.50948346e-01
2.26753652e-01 3.52287948e-01 5.56551442e-02 -4.38480675e-01
-1.92655236e-01 8.57662737e-01 -1.75263476e+00 5.18643439e-01
5.98462820e-01 1.10980558e+00 -2.20582008e-01 -2.00633168e-01
-3.06650072e-01 3.05737466e-01 -4.97696757e-01 -7.13434696e-01
3.71167094e-01 -1.24533214e-01 -7.06083894e-01 2.32776597e-01
1.16701412e+00 -3.09621602e-01 -4.49840695e-01 1.10073996e+00
3.09090227e-01 -2.75269359e-01 3.90328109e-01 -1.49150705e+00
-5.79848230e-01 3.25050622e-01 -7.89645553e-01 -1.96667999e-01
1.52252063e-01 -1.29569456e-01 9.32956696e-01 -1.14048791e+00
6.17651880e-01 1.61574185e+00 6.70463741e-01 6.48408294e-01
-1.08850753e+00 -9.36696768e-01 5.34415901e-01 2.08196104e-01
-1.66078854e+00 -7.20716536e-01 1.06474304e+00 -3.27513993e-01
6.04153097e-01 3.79481494e-01 7.72210479e-01 8.80998135e-01
1.64774165e-01 6.01407886e-01 8.98044705e-01 -1.98678985e-01
-1.02784544e-01 -1.58397362e-01 2.11705968e-01 1.05127239e+00
2.24788100e-01 -1.09909467e-01 -6.38237655e-01 -9.66424718e-02
8.85990858e-01 -5.61610572e-02 -2.25210398e-01 -4.87720490e-01
-4.43662286e-01 4.65604872e-01 8.64928603e-01 3.24470431e-01
-6.47401288e-02 9.91597697e-02 2.79953063e-01 -1.41040832e-01
5.70307970e-01 -1.81726925e-02 -4.90469545e-01 6.91933692e-01
-1.01069534e+00 1.59030974e-01 4.19516027e-01 8.99263978e-01
8.44978631e-01 1.85646266e-02 -3.74464951e-02 9.64382887e-01
5.68937361e-01 1.12281116e-02 6.69184700e-02 -1.19693244e+00
4.45874393e-01 9.90609467e-01 -4.37894672e-01 -1.45194137e+00
-6.44742787e-01 -5.43866575e-01 -1.16916239e+00 1.97850153e-01
3.60213310e-01 3.36434841e-01 -9.52327192e-01 2.03749895e+00
5.47448993e-01 1.26353979e-01 -5.25406182e-01 8.41726363e-01
1.25647569e+00 2.22236812e-01 4.57656756e-02 -3.96376327e-02
1.66643643e+00 -1.11105657e+00 -9.17123616e-01 -3.00059021e-01
3.86615604e-01 -6.64415717e-01 8.49324882e-01 2.20954850e-01
-9.89190996e-01 -8.76747727e-01 -8.59135091e-01 -2.74477869e-01
-4.11715984e-01 4.49123472e-01 7.15247095e-01 4.87279624e-01
-1.24597418e+00 3.38479578e-01 -5.88185430e-01 -3.32688212e-01
8.97832096e-01 6.53654456e-01 -8.50421369e-01 6.36720881e-02
-1.05091763e+00 6.50336623e-01 2.72194415e-01 8.66926432e-01
-5.33496261e-01 -4.25072640e-01 -1.02963257e+00 1.46604687e-01
2.58742541e-01 -8.10132205e-01 8.31372857e-01 -9.78831649e-01
-1.26364577e+00 9.97528911e-01 -5.36323547e-01 9.60182399e-02
3.76110554e-01 -3.03533912e-01 -2.54730374e-01 2.18918342e-02
-4.59277257e-02 1.19249809e+00 8.52233291e-01 -1.38332558e+00
-2.13202506e-01 -8.01034927e-01 1.30136132e-01 2.11535692e-02
-5.45429766e-01 1.01011030e-01 -8.33288550e-01 -6.39092624e-01
7.08291829e-01 -9.21514750e-01 1.21434055e-01 3.44049156e-01
-4.64188099e-01 -5.28790116e-01 1.12952936e+00 -7.96054423e-01
1.37208045e+00 -1.87572289e+00 1.34109080e-01 1.92122653e-01
3.84160668e-01 3.39100599e-01 -3.30359995e-01 2.04395816e-01
-5.84972978e-01 1.87410146e-01 3.09773553e-02 -5.66800892e-01
-8.91055986e-02 3.55228841e-01 -2.64415592e-01 4.65722263e-01
3.09897095e-01 1.05576122e+00 -2.51104832e-01 -6.33977950e-01
8.89450088e-02 7.63820052e-01 -6.22766972e-01 -1.13445476e-01
-1.89537257e-02 4.18507963e-01 -1.73412561e-01 1.04058897e+00
1.17738914e+00 -1.94222212e-01 3.72415960e-01 -7.21995056e-01
2.35040352e-01 -1.54183749e-02 -1.04901230e+00 1.75103509e+00
-1.31910145e-01 5.36462963e-01 3.11121166e-01 -7.10849166e-01
9.40603316e-01 2.55882502e-01 2.96818286e-01 -6.26433253e-01
3.49277914e-01 -5.78440279e-02 7.90246502e-02 -5.46151459e-01
4.24572341e-02 1.73426583e-01 2.37852484e-01 -1.13047287e-01
3.13736573e-02 2.02831402e-01 1.32393852e-01 1.23116657e-01
5.61373055e-01 5.73625505e-01 1.25562459e-01 -2.85846829e-01
7.97209084e-01 -6.29622161e-01 9.15944278e-01 -8.03210307e-03
-3.50475788e-01 6.02463305e-01 6.27116203e-01 -8.88180315e-01
-7.32626021e-01 -8.15012336e-01 -2.79569715e-01 1.10883904e+00
2.86750197e-01 -7.57787228e-01 -1.18963516e+00 -6.59884810e-01
4.79319692e-02 1.72519937e-01 -9.70018566e-01 -1.15744449e-01
-8.74453366e-01 -5.80703557e-01 5.80284357e-01 6.51535571e-01
8.24539900e-01 -9.51451600e-01 -1.41756684e-01 -2.51651049e-01
-3.29773366e-01 -1.23252726e+00 -3.39302450e-01 -3.99897784e-01
-8.61278713e-01 -1.08156621e+00 -3.74630481e-01 -8.41940701e-01
9.77630317e-01 2.57449389e-01 1.08126032e+00 5.40563524e-01
-3.08356196e-01 -2.00381979e-01 3.12149763e-01 -9.94879976e-02
7.88104236e-02 5.53309284e-02 1.53658718e-01 2.68576711e-01
2.46625856e-01 -7.96230018e-01 -6.38734937e-01 7.64851868e-01
-8.14388573e-01 2.54369527e-01 3.98724765e-01 6.40300333e-01
3.83057088e-01 9.00473073e-02 3.43632728e-01 -6.43498778e-01
1.79672703e-01 -5.69094792e-02 -5.18607378e-01 2.78906882e-01
-1.31297603e-01 -1.68116808e-01 4.91181552e-01 -2.83314049e-01
-1.12572443e+00 1.71269387e-01 -3.11931908e-01 -5.46830535e-01
-3.73011917e-01 1.38253774e-02 -9.13576305e-01 -2.75907993e-01
3.67055923e-01 -3.03205311e-01 6.42951503e-02 -5.19411027e-01
3.38327587e-01 2.54101962e-01 4.48198974e-01 -6.77546799e-01
7.28048921e-01 7.43303835e-01 3.08051497e-01 -8.17017198e-01
-1.06706321e+00 -8.56980905e-02 -1.13060844e+00 -4.86718148e-01
6.74038351e-01 -8.75413954e-01 -7.21446633e-01 5.20460367e-01
-1.37215364e+00 1.82265730e-03 1.16781168e-01 -6.20350949e-02
-2.73347616e-01 3.47236544e-01 -7.17362583e-01 -4.47397083e-01
-2.26292536e-01 -1.22512531e+00 1.29721987e+00 3.45550358e-01
-3.83807361e-01 -7.62633145e-01 -3.86042893e-01 7.26014853e-01
1.64940357e-01 3.57640028e-01 9.41321313e-01 -1.70416296e-01
-5.71552575e-01 -7.85899907e-02 -4.32827562e-01 2.44387910e-01
1.47954464e-01 2.98483014e-01 -1.20624280e+00 -7.29321241e-02
-6.03911541e-02 -1.34793252e-01 8.16511631e-01 3.61536741e-01
1.45132244e+00 -3.94265652e-01 -4.87608224e-01 8.43250334e-01
9.53055501e-01 -3.01597435e-02 9.49318528e-01 2.10047573e-01
1.06279469e+00 1.13865554e+00 4.57435817e-01 -7.02761412e-02
3.98012757e-01 9.65190351e-01 7.65701294e-01 -3.10236812e-01
-3.77805293e-01 -1.92345604e-01 4.88250479e-02 6.14417851e-01
-1.86761081e-01 6.63499981e-02 -8.81767452e-01 2.12237716e-01
-1.88978076e+00 -7.36869395e-01 -1.58763632e-01 1.73474598e+00
6.50441647e-01 -4.85124951e-03 -8.00880864e-02 -8.62787105e-03
8.60898793e-01 2.98540115e-01 -3.79009873e-01 -1.53047442e-01
-2.77904034e-01 7.13920817e-02 -1.46903053e-01 4.05960083e-01
-1.15763247e+00 1.24015975e+00 5.85183859e+00 9.82244849e-01
-1.02546859e+00 6.42858446e-02 7.92587936e-01 8.26648786e-04
6.99134544e-02 7.45370016e-02 -1.13317955e+00 2.26496056e-01
3.83713126e-01 6.12449348e-01 6.19480424e-02 8.37780952e-01
1.15781374e-01 1.48226485e-01 -9.49889958e-01 1.04428363e+00
1.69801742e-01 -1.15185285e+00 4.53986764e-01 3.36831212e-01
5.73370516e-01 -4.40500051e-01 -1.47182832e-03 -1.23291342e-02
-2.49510586e-01 -1.21095443e+00 7.40475535e-01 4.78624970e-01
7.31100738e-01 -6.53185487e-01 7.23634481e-01 2.01405808e-01
-1.43075418e+00 -5.14990017e-02 -4.28143948e-01 -7.87762851e-02
3.16535681e-02 5.81786394e-01 -2.69525886e-01 4.38364089e-01
9.42633450e-01 4.96531725e-01 -8.25794756e-01 7.26963341e-01
-5.48222899e-01 1.68830544e-01 -3.11067671e-01 7.13105083e-01
4.27642278e-02 -2.87674546e-01 2.20627531e-01 8.86477113e-01
8.63974690e-02 -1.91207826e-02 2.23493539e-02 8.69122088e-01
-2.80712962e-01 1.14908628e-01 -5.87504029e-01 5.13732195e-01
3.12685937e-01 1.57037437e+00 -7.60183930e-01 -2.45582834e-01
-2.48091653e-01 6.75792456e-01 6.92610443e-01 4.39661115e-01
-1.04026592e+00 6.19914904e-02 7.66705692e-01 3.79127502e-01
-7.46258767e-03 -1.10882707e-01 -4.37262982e-01 -9.46687222e-01
3.79706293e-01 -7.95149088e-01 3.05205017e-01 -9.27698731e-01
-9.23061371e-01 8.21770668e-01 1.01856507e-01 -8.86549890e-01
2.59689689e-01 -6.30578220e-01 -5.62164843e-01 6.32131636e-01
-1.22711945e+00 -1.65135884e+00 -6.54237926e-01 4.77627516e-01
5.00936091e-01 -2.39720121e-02 7.07918823e-01 3.72002155e-01
-1.08985019e+00 7.69464552e-01 -6.48562670e-01 3.38973790e-01
6.38930202e-01 -5.92542648e-01 3.52204710e-01 6.89522624e-01
1.57043248e-01 1.04668725e+00 3.43701839e-01 -6.61645293e-01
-9.58209455e-01 -1.19214129e+00 9.52324331e-01 -4.11439180e-01
3.38886946e-01 -6.63573086e-01 -1.21803856e+00 6.88953936e-01
1.86619759e-01 3.56218040e-01 5.56573689e-01 1.87372640e-01
-8.58787179e-01 -5.07362545e-01 -9.87101555e-01 8.67817223e-01
1.65295553e+00 -4.50637400e-01 -5.42142928e-01 2.34756723e-01
4.86044556e-01 -4.09357071e-01 -8.44123363e-01 8.38787913e-01
8.90459657e-01 -1.24133527e+00 9.32927489e-01 -4.93731320e-01
4.19124186e-01 -5.55486858e-01 -8.52771699e-02 -7.98937321e-01
-4.69336033e-01 -4.51650620e-01 -2.33463541e-01 1.41234708e+00
5.78075238e-02 -7.24854767e-01 9.83645678e-01 5.31826496e-01
4.52379743e-03 -1.11483681e+00 -1.03138804e+00 -5.81577718e-01
-9.23859775e-02 -9.50417295e-02 7.79670537e-01 9.01060760e-01
-3.97242963e-01 3.36959839e-01 -2.10405245e-01 2.90204436e-01
4.78013515e-01 1.37932032e-01 8.37201715e-01 -1.45303571e+00
2.24561259e-01 -5.82111478e-01 -4.73100185e-01 -9.77183878e-01
5.60817480e-01 -7.67396569e-01 -3.58112425e-01 -1.53570437e+00
9.67313945e-02 -2.40334019e-01 -7.99274147e-02 9.10708070e-01
5.80361709e-02 6.92102134e-01 1.97062388e-01 2.39828423e-01
-3.99053872e-01 6.93664372e-01 1.35447860e+00 -1.31073222e-01
-3.77566628e-02 -2.03779757e-01 -8.06702971e-01 1.33824122e+00
8.90417576e-01 -2.82973289e-01 -3.02769363e-01 -4.71327543e-01
1.29251808e-01 -3.31849754e-01 4.76070702e-01 -9.15789604e-01
2.51754373e-01 -6.25654310e-02 6.95431232e-01 -7.05343783e-01
6.71173632e-01 -9.38196838e-01 4.71755683e-01 2.68771917e-01
2.07873955e-02 1.18660666e-01 4.40274715e-01 9.03946534e-02
-4.47551459e-01 1.32339627e-01 8.21757674e-01 -3.86298932e-02
-6.02320433e-01 4.88262534e-01 1.73794106e-01 -3.42517227e-01
9.31618333e-01 -3.73000085e-01 -5.05610287e-01 -1.96743891e-01
-1.01452565e+00 4.65850309e-02 4.79688704e-01 7.15189695e-01
5.16916752e-01 -1.51896012e+00 -4.65383053e-01 4.97402459e-01
-1.01807667e-02 1.78494081e-01 4.25667226e-01 8.10958028e-01
-4.72780883e-01 5.75788319e-01 -3.44377965e-01 -7.69358337e-01
-1.69115829e+00 4.54115689e-01 5.29825032e-01 -6.66217804e-02
-3.83750826e-01 1.01893759e+00 1.00528455e+00 -4.95288432e-01
3.83933246e-01 -1.28896251e-01 -3.93878460e-01 2.37360463e-01
3.77227336e-01 1.90985739e-01 8.92674178e-02 -1.28147852e+00
-5.18052399e-01 1.21702576e+00 -1.28321052e-01 3.81080329e-01
1.04840195e+00 -5.57086654e-02 -7.32888699e-01 -7.23753497e-02
1.12988389e+00 -2.53437757e-02 -1.13247931e+00 -1.25418007e-01
-2.14371264e-01 -4.19851661e-01 -3.10911119e-01 -5.14237881e-01
-1.57722080e+00 1.08255160e+00 4.47697461e-01 -1.48306265e-01
1.38445508e+00 -1.27600657e-03 4.53212827e-01 5.15651107e-02
4.17072594e-01 -8.37895036e-01 4.76031005e-02 5.77935576e-01
1.38026536e+00 -7.96290815e-01 1.55156448e-01 -1.05048716e+00
-1.58534393e-01 1.11799967e+00 1.22558916e+00 1.01066101e-02
7.65721679e-01 1.14134297e-01 8.57596919e-02 -4.14775848e-01
-5.12113154e-01 1.00440020e-02 6.62378252e-01 2.30043277e-01
4.19611394e-01 -1.42250270e-01 -1.44218817e-01 3.48255903e-01
-3.94009173e-01 -2.30804473e-01 -1.63871571e-01 6.00302637e-01
-1.96866989e-01 -1.19385719e+00 -5.42579770e-01 3.85581926e-02
-3.57598066e-01 3.13651748e-02 -6.42005563e-01 1.04624927e+00
7.68921793e-01 6.71043575e-01 9.77146849e-02 -3.34412634e-01
4.01534408e-01 1.90693721e-01 5.59853554e-01 -4.30878758e-01
-4.60882962e-01 3.08143348e-01 -1.49464672e-02 -9.15382206e-01
-5.06433547e-01 -3.53027195e-01 -1.14461040e+00 -1.55262142e-01
-3.44095320e-01 -1.95312619e-01 5.37736535e-01 7.93097854e-01
6.42400622e-01 5.53789377e-01 2.83408642e-01 -1.03363943e+00
1.11034796e-01 -9.22185957e-01 -2.78559804e-01 4.06576306e-01
1.02100499e-01 -1.05481660e+00 -2.25225478e-01 -8.64424333e-02] | [13.420672416687012, 0.46458008885383606] |
aee85ce9-1c5b-4e4f-9815-c3f072d0abee | meta-rangeseg-lidar-sequence-semantic | 2202.13377 | null | https://arxiv.org/abs/2202.13377v3 | https://arxiv.org/pdf/2202.13377v3.pdf | Meta-RangeSeg: LiDAR Sequence Semantic Segmentation Using Multiple Feature Aggregation | LiDAR sensor is essential to the perception system in autonomous vehicles and intelligent robots. To fulfill the real-time requirements in real-world applications, it is necessary to efficiently segment the LiDAR scans. Most of previous approaches directly project 3D point cloud onto the 2D spherical range image so that they can make use of the efficient 2D convolutional operations for image segmentation. Although having achieved the encouraging results, the neighborhood information is not well-preserved in the spherical projection. Moreover, the temporal information is not taken into consideration in the single scan segmentation task. To tackle these problems, we propose a novel approach to semantic segmentation for LiDAR sequences named Meta-RangeSeg, where a new range residual image representation is introduced to capture the spatial-temporal information. Specifically, Meta-Kernel is employed to extract the meta features, which reduces the inconsistency between the 2D range image coordinates input and 3D Cartesian coordinates output. An efficient U-Net backbone is used to obtain the multi-scale features. Furthermore, Feature Aggregation Module (FAM) strengthens the role of range channel and aggregates features at different levels. We have conducted extensive experiments for performance evaluation on SemanticKITTI and SemanticPOSS. The promising results show that our proposed Meta-RangeSeg method is more efficient and effective than the existing approaches. Our full implementation is publicly available at https://github.com/songw-zju/Meta-RangeSeg . | ['Ruixiang Zhang', 'Jianke Zhu', 'Song Wang'] | 2022-02-27 | null | null | null | null | ['lidar-semantic-segmentation'] | ['computer-vision'] | [ 2.12549716e-01 -3.57426435e-01 6.56442642e-02 -6.97505832e-01
-4.78641093e-01 -2.18416914e-01 4.65541095e-01 -1.34165883e-01
-6.46575451e-01 2.79536068e-01 -4.63393420e-01 -2.44673491e-01
-2.07262069e-01 -1.29814434e+00 -8.43759596e-01 -6.30508184e-01
2.50065833e-01 5.14379263e-01 7.24358737e-01 -3.45343560e-01
4.50638235e-01 9.04808104e-01 -1.83195662e+00 -8.34258571e-02
1.08953619e+00 1.21483588e+00 6.33536160e-01 3.00219625e-01
-6.49742603e-01 -1.42645650e-02 -2.38685340e-01 1.20147362e-01
5.01329482e-01 6.43096864e-02 -3.78196746e-01 1.00861646e-01
2.40049779e-01 -3.97124290e-01 -1.45568877e-01 1.31581569e+00
4.06627804e-01 1.83871597e-01 3.99158865e-01 -1.27925205e+00
-1.63972855e-01 2.09448725e-01 -8.02279949e-01 -1.20882969e-02
1.65737197e-02 7.49780461e-02 5.45233011e-01 -9.50034082e-01
4.17439371e-01 1.23831868e+00 5.18946946e-01 4.23617661e-02
-7.65721262e-01 -8.44309032e-01 5.06799296e-02 4.94728446e-01
-1.50830185e+00 -9.55352262e-02 9.54546571e-01 -1.79431885e-01
6.03245139e-01 1.09320797e-01 6.57498896e-01 3.14496845e-01
6.46376237e-02 5.89479983e-01 8.53986025e-01 -8.51254240e-02
3.52668315e-02 -7.79671073e-02 2.31131658e-01 6.69321120e-01
7.85072967e-02 -4.23318036e-02 -1.22884348e-01 1.95023879e-01
9.25418854e-01 4.48089391e-01 -1.87392786e-01 -5.51807582e-01
-1.12518835e+00 8.19196105e-01 7.67585874e-01 3.10554206e-01
-3.86516988e-01 8.93797278e-02 2.03691229e-01 4.94502112e-02
3.20956200e-01 -1.47596747e-01 -4.05455172e-01 -1.67177115e-02
-7.20847011e-01 5.33847548e-02 1.92364007e-01 1.16695762e+00
1.29362702e+00 -1.86865225e-01 3.32467437e-01 8.51692915e-01
2.70943761e-01 6.75628483e-01 3.93119782e-01 -1.15168107e+00
5.14320493e-01 9.23307002e-01 -8.22261050e-02 -1.02032781e+00
-5.89895725e-01 -2.16719881e-01 -6.96126878e-01 2.19284967e-01
1.87857524e-01 2.04416603e-01 -1.05805087e+00 1.25428867e+00
6.03087962e-01 3.07500660e-01 -3.61333564e-02 9.42915738e-01
9.61703539e-01 8.83491874e-01 -2.12310180e-01 3.38649913e-03
1.35087526e+00 -8.41348767e-01 -4.01684225e-01 -1.28322572e-01
3.53753060e-01 -8.73134375e-01 1.03139591e+00 6.53218925e-02
-9.41063643e-01 -8.28737617e-01 -1.07498515e+00 -2.67064154e-01
-6.04938865e-01 1.36542767e-01 4.43892330e-01 2.93929189e-01
-7.03633130e-01 3.85566831e-01 -7.57846296e-01 -3.04083109e-01
5.48823655e-01 4.79109377e-01 -4.12578791e-01 -3.13030839e-01
-9.67818081e-01 5.74987710e-01 5.32732904e-01 4.00385499e-01
-2.33794734e-01 -5.05324543e-01 -9.12766933e-01 -4.95809317e-02
5.15291989e-01 -3.40044081e-01 1.07491159e+00 -3.90460968e-01
-1.15902579e+00 5.31279564e-01 -2.18542263e-01 -1.83769792e-01
5.59159756e-01 -3.42727154e-02 -5.30795567e-02 3.17113817e-01
3.09739530e-01 8.77543032e-01 5.36665440e-01 -1.20737922e+00
-9.53969181e-01 -6.31071270e-01 -5.70512377e-03 4.30652261e-01
-8.59655440e-02 -3.47749501e-01 -8.69562566e-01 -2.10556999e-01
7.00566947e-01 -8.55041862e-01 -3.62303793e-01 -6.45747781e-02
-2.21625596e-01 -2.80039608e-01 1.22973096e+00 -3.41127247e-01
6.53011262e-01 -2.21891618e+00 -1.68498278e-01 3.05642903e-01
-1.63125545e-01 3.77970248e-01 -9.58815275e-04 8.13542455e-02
1.36399850e-01 -9.88902058e-04 -5.07440865e-01 -1.88967124e-01
-2.15184733e-01 3.70426029e-01 4.14014943e-02 4.34100926e-01
6.14885800e-02 7.61772752e-01 -5.83113194e-01 -6.71364188e-01
7.18162298e-01 5.53125918e-01 -3.80396128e-01 -5.96975274e-02
-2.39367977e-01 4.71419930e-01 -8.45205784e-01 6.34333193e-01
1.36620581e+00 1.19600385e-01 -3.97105932e-01 -3.84946555e-01
-5.83529532e-01 1.07754897e-02 -1.21826088e+00 1.85310817e+00
-3.91252160e-01 2.66634196e-01 1.64563153e-02 -1.00526750e+00
1.00738311e+00 -1.14931382e-01 7.94767916e-01 -1.05143476e+00
2.83531755e-01 3.80664736e-01 -2.54373103e-01 -3.30788136e-01
5.82412601e-01 1.88603222e-01 -1.39023766e-01 3.04236431e-02
-3.38773012e-01 -5.49760759e-01 1.43139184e-01 2.57678470e-03
6.56207800e-01 2.33566180e-01 1.22942507e-01 5.77595690e-03
8.85832548e-01 3.89364570e-01 7.18959332e-01 3.22832644e-01
1.96327548e-02 6.55308783e-01 1.17668711e-01 -3.12698334e-01
-8.75376821e-01 -9.63879526e-01 -2.91340441e-01 5.19493401e-01
7.86930382e-01 -5.82658909e-02 -6.69553041e-01 -4.73944962e-01
6.46561980e-02 5.04773796e-01 -2.13211328e-01 1.91192534e-02
-7.99859762e-01 -4.18254256e-01 2.41914839e-01 5.84425628e-01
1.03470111e+00 -1.02925479e+00 -8.85373116e-01 2.10579425e-01
-1.85198456e-01 -1.37381351e+00 -4.46334034e-01 9.57169756e-02
-9.96829987e-01 -9.73722577e-01 -4.22950447e-01 -7.13556349e-01
6.28144085e-01 7.71170974e-01 4.43666458e-01 7.73518905e-02
-3.17450970e-01 -1.41237369e-02 -3.25072825e-01 -3.58294606e-01
2.10580975e-01 1.48703307e-01 -3.38080138e-01 -2.12694913e-01
5.59936643e-01 -7.00704753e-01 -5.80724597e-01 4.84245181e-01
-9.39999998e-01 1.93050325e-01 6.84865952e-01 5.36159813e-01
9.19169247e-01 2.67241329e-01 2.88052380e-01 -6.36390984e-01
1.62419900e-01 -2.55381465e-01 -9.12529826e-01 -1.81242540e-01
-3.65156949e-01 -1.37121305e-01 3.97058129e-01 -1.39696896e-01
-9.65643167e-01 3.62244070e-01 -4.41630244e-01 -5.46601176e-01
-4.00951117e-01 3.12952399e-01 -4.87682372e-01 -4.96644713e-02
6.19892813e-02 3.44580740e-01 1.08923644e-01 -5.29409111e-01
4.43522185e-01 8.29437673e-01 5.19627035e-01 -3.67712736e-01
9.42759454e-01 7.32013226e-01 5.54412119e-02 -1.04105175e+00
-5.15188992e-01 -7.59050906e-01 -8.53062928e-01 -1.04256563e-01
1.03488600e+00 -8.02951694e-01 -6.69649422e-01 4.50324774e-01
-1.29998291e+00 1.84900295e-02 -1.46303132e-01 5.35002947e-01
-5.89618564e-01 4.52968359e-01 -2.02489391e-01 -6.02583826e-01
-2.89325893e-01 -1.54388380e+00 1.14547634e+00 6.01786852e-01
4.46961045e-01 -4.14848685e-01 -3.38739306e-01 4.46487576e-01
2.21470803e-01 2.71351486e-01 8.77542317e-01 -3.56996238e-01
-1.01030457e+00 -2.04829555e-02 -6.93472326e-01 3.11666816e-01
1.02348611e-01 -8.36440176e-02 -7.81613111e-01 1.68627888e-01
-7.57965446e-02 1.16928622e-01 8.83483291e-01 4.16924179e-01
1.11459637e+00 3.62028182e-01 -3.44435841e-01 8.17648232e-01
1.50818467e+00 5.30882180e-01 6.37610376e-01 4.39514607e-01
8.88603508e-01 5.68960488e-01 1.07137799e+00 2.25270122e-01
7.21972346e-01 6.51746809e-01 6.81961417e-01 -1.42486677e-01
5.72208837e-02 1.98820494e-02 5.60617745e-02 7.23248959e-01
-7.97955841e-02 9.37217399e-02 -7.99700916e-01 5.54435074e-01
-1.96586239e+00 -7.68163800e-01 -3.61988246e-01 2.11898327e+00
3.82266998e-01 2.23264664e-01 -2.22920030e-01 1.63150117e-01
8.13698053e-01 2.96989009e-02 -6.34630680e-01 -3.04534644e-01
4.00997512e-02 1.23503417e-01 6.73545301e-01 3.87454689e-01
-1.07400930e+00 1.08318579e+00 4.37684298e+00 9.56417263e-01
-1.16023433e+00 1.26822233e-01 1.51453495e-01 5.66439815e-02
-1.77449465e-01 -1.14120722e-01 -8.17417264e-01 4.86323565e-01
4.30229723e-01 1.21361233e-01 1.06150143e-01 9.35776353e-01
3.51019233e-01 -3.27609718e-01 -6.03850901e-01 1.01737785e+00
-2.35733986e-01 -1.15321779e+00 -1.05341092e-01 1.42291412e-01
4.59608227e-01 4.19710934e-01 -1.02896251e-01 6.60668090e-02
-1.90960482e-01 -6.70566261e-01 8.28805089e-01 3.92240405e-01
7.61160493e-01 -1.15216601e+00 7.94839382e-01 6.29632533e-01
-1.59812152e+00 -5.12138344e-02 -6.78617835e-01 2.19959825e-01
3.14771265e-01 7.64634192e-01 -8.44987273e-01 8.69667411e-01
7.76238620e-01 6.61465108e-01 -4.53661650e-01 1.00859404e+00
-1.13432072e-01 1.43242711e-02 -6.95839167e-01 1.93541363e-01
4.86307114e-01 -6.06212020e-01 4.99940813e-01 8.96773160e-01
6.16223276e-01 1.45661116e-01 3.23001921e-01 8.25467944e-01
8.17208737e-02 2.53789216e-01 -6.97409153e-01 1.99804172e-01
5.03624439e-01 1.33556879e+00 -1.07952046e+00 -2.51971722e-01
-4.61452991e-01 7.11134315e-01 -9.85724851e-02 1.60271883e-01
-8.46700728e-01 -6.21741772e-01 5.45158088e-01 1.06405757e-01
5.07601798e-01 -5.93912661e-01 -4.40250367e-01 -8.72087717e-01
1.47965282e-01 -2.45196059e-01 1.09884724e-01 -8.55037332e-01
-6.80900037e-01 5.50843537e-01 1.51125148e-01 -1.32510567e+00
-7.99139589e-02 -3.80610675e-01 -4.81565356e-01 8.22501361e-01
-1.97242832e+00 -1.17757928e+00 -6.80281878e-01 8.10601056e-01
7.88917422e-01 2.29720220e-01 2.21647248e-01 3.99028599e-01
-4.71896619e-01 7.01772571e-02 -3.69992591e-02 -2.98320819e-02
3.79713118e-01 -9.29221570e-01 3.90261799e-01 8.14074993e-01
-1.48724914e-01 4.58729059e-01 3.53925318e-01 -6.96769714e-01
-1.26495802e+00 -1.16384304e+00 6.11683488e-01 6.93528727e-02
1.71336204e-01 -1.85233235e-01 -8.98368180e-01 4.16162461e-01
-9.37167332e-02 1.39234588e-01 1.77924812e-01 -5.44406474e-01
1.49816982e-02 -2.47988656e-01 -1.27471948e+00 3.50051463e-01
1.17659295e+00 -1.48530886e-01 -4.94276315e-01 -3.12710889e-02
1.00058794e+00 -5.00335276e-01 -6.82549357e-01 7.77064085e-01
4.56754297e-01 -1.12367570e+00 1.09552646e+00 2.11542875e-01
2.05979019e-01 -7.90200591e-01 -2.56494880e-01 -9.78297651e-01
1.38817176e-01 -1.08655673e-02 4.11246032e-01 1.09176779e+00
1.25342667e-01 -8.77447188e-01 7.69567251e-01 1.82343036e-01
-4.51201171e-01 -7.67321885e-01 -9.81755257e-01 -6.12475216e-01
-2.43293658e-01 -7.57683754e-01 9.66144204e-01 5.58807135e-01
-6.71583176e-01 1.83020085e-01 1.66740358e-01 4.56904024e-01
7.31243551e-01 6.02856457e-01 8.53935957e-01 -1.49717021e+00
3.56440425e-01 -4.72568840e-01 -5.85067570e-01 -1.17602170e+00
3.10779624e-02 -7.63193488e-01 1.89143196e-01 -1.78066945e+00
-1.03326686e-01 -7.77267396e-01 -4.98715118e-02 3.25358272e-01
7.56916851e-02 4.94779587e-01 3.14718693e-01 1.95991680e-01
-3.21349144e-01 6.36412024e-01 1.46783638e+00 -7.65203238e-02
-2.64469504e-01 2.03971311e-01 -4.23654377e-01 8.61653268e-01
9.88956809e-01 -3.79402310e-01 -5.80014408e-01 -5.17652035e-01
-1.01489142e-01 -7.54166991e-02 2.89181888e-01 -1.08487380e+00
4.57353652e-01 -2.85450995e-01 2.95265526e-01 -1.45064127e+00
6.27774954e-01 -1.05619824e+00 2.93690749e-02 2.69757003e-01
3.27321529e-01 -3.40146795e-02 5.26986308e-02 3.13600123e-01
-3.05405438e-01 -2.73523808e-01 9.01196182e-01 -3.12895626e-01
-1.09166551e+00 5.40304124e-01 -1.61028132e-02 -4.59584624e-01
1.05884242e+00 -6.18939638e-01 -7.16596991e-02 -3.61616760e-02
-4.12093461e-01 5.29510200e-01 6.17689312e-01 3.31024677e-01
8.90807033e-01 -1.33757615e+00 -3.65273654e-01 4.27519262e-01
1.12426862e-01 7.98520744e-01 4.83507901e-01 9.14438725e-01
-7.51400590e-01 3.53986531e-01 -2.56939828e-01 -1.00922203e+00
-1.19667399e+00 2.68297017e-01 1.21035136e-01 1.84882954e-01
-7.66015470e-01 6.19865894e-01 1.98275983e-01 -6.67994797e-01
3.09918933e-02 -5.67459702e-01 -3.36185187e-01 1.55584589e-01
3.20299059e-01 4.25147206e-01 1.87573850e-01 -9.22783315e-01
-4.07749712e-01 1.24628472e+00 4.24805433e-02 -5.40326759e-02
1.33899200e+00 -3.29309225e-01 -2.48677224e-01 2.31557235e-01
1.37531054e+00 -5.87378182e-02 -1.14829659e+00 -2.71704704e-01
-1.49767563e-01 -4.57499862e-01 1.04412913e-01 -1.82338119e-01
-1.30139601e+00 1.12475693e+00 6.63663983e-01 2.25097965e-02
1.20273900e+00 -1.25118405e-01 1.12165272e+00 3.35034579e-01
5.24852455e-01 -9.10162508e-01 -2.42000997e-01 6.96876884e-01
5.25805295e-01 -1.18071449e+00 2.45466847e-02 -8.09580743e-01
-3.88656288e-01 1.21900272e+00 7.08216012e-01 -1.49073943e-01
6.83274925e-01 2.18216002e-01 1.59019083e-01 -2.10737303e-01
-3.01570334e-02 -5.60137510e-01 1.34005085e-01 5.14586568e-01
-6.41385093e-02 -8.67847651e-02 -3.32746953e-01 3.61806303e-01
-3.02820146e-01 -1.41534477e-01 3.39204371e-01 1.03370035e+00
-8.51084828e-01 -1.06359029e+00 -5.01218021e-01 1.43625751e-01
6.36378154e-02 2.22130865e-01 3.27755064e-02 8.49745691e-01
5.59166431e-01 7.77585685e-01 2.71781474e-01 -4.70790118e-01
5.20118117e-01 -5.24942055e-02 2.56714672e-01 -4.93322700e-01
-5.59293106e-02 3.50432783e-01 -3.37941498e-01 -7.92438149e-01
-5.05115986e-01 -6.12698078e-01 -2.00907040e+00 -1.06874727e-01
-3.60404998e-01 1.19446509e-01 1.10829067e+00 8.74479055e-01
3.35285366e-01 4.68282282e-01 7.99059689e-01 -1.15115392e+00
-1.17559560e-01 -7.14570105e-01 -3.34092617e-01 5.76153658e-02
1.08502083e-01 -8.83884192e-01 -1.13385752e-01 -2.39490658e-01] | [8.23693561553955, -2.60425066947937] |
d41fe8a6-cbed-4825-8783-69d84118cca2 | orthogonal-coding-based-feature-generation | 2207.05957 | null | https://arxiv.org/abs/2207.05957v1 | https://arxiv.org/pdf/2207.05957v1.pdf | Orthogonal-Coding-Based Feature Generation for Transductive Open-Set Recognition via Dual-Space Consistent Sampling | Open-set recognition (OSR) aims to simultaneously detect unknown-class samples and classify known-class samples. Most of the existing OSR methods are inductive methods, which generally suffer from the domain shift problem that the learned model from the known-class domain might be unsuitable for the unknown-class domain. Addressing this problem, inspired by the success of transductive learning for alleviating the domain shift problem in many other visual tasks, we propose an Iterative Transductive OSR framework, called IT-OSR, which implements three explored modules iteratively, including a reliability sampling module, a feature generation module, and a baseline update module. Specifically, at each iteration, a dual-space consistent sampling approach is presented in the explored reliability sampling module for selecting some relatively more reliable ones from the test samples according to their pseudo labels assigned by a baseline method, which could be an arbitrary inductive OSR method. Then, a conditional dual-adversarial generative network under an orthogonal coding condition is designed in the feature generation module to generate discriminative sample features of both known and unknown classes according to the selected test samples with their pseudo labels. Finally, the baseline method is updated for sample re-prediction in the baseline update module by jointly utilizing the generated features, the selected test samples with pseudo labels, and the training samples. Extensive experimental results on both the standard-dataset and the cross-dataset settings demonstrate that the derived transductive methods, by introducing two typical inductive OSR methods into the proposed IT-OSR framework, achieve better performances than 15 state-of-the-art methods in most cases. | ['Qiulei Dong', 'Jiayin Sun'] | 2022-07-13 | null | null | null | null | ['open-set-learning'] | ['miscellaneous'] | [ 5.52505970e-01 2.40958594e-02 -3.55545700e-01 -1.49075389e-01
-1.20314062e+00 -4.43000138e-01 4.86387640e-01 -4.24045652e-01
1.12230778e-01 8.28224838e-01 -6.43125176e-02 6.89702779e-02
-1.06816635e-01 -7.32863128e-01 -6.26083970e-01 -9.80279088e-01
3.73497456e-01 6.32240176e-01 1.39654309e-01 -5.42204976e-02
1.17606834e-01 3.53067726e-01 -1.65524447e+00 3.43159288e-01
1.15100253e+00 1.25333238e+00 -3.33834477e-02 1.71160147e-01
1.10949697e-02 6.36585891e-01 -6.58570290e-01 -3.11391383e-01
3.86567295e-01 -6.50840640e-01 -4.16840732e-01 5.19049764e-01
1.97176486e-01 -1.66165262e-01 -4.12284940e-01 1.00936353e+00
6.22565866e-01 3.19758147e-01 8.55856419e-01 -1.36179030e+00
-1.13744426e+00 3.85155350e-01 -5.51915646e-01 -2.49635622e-01
5.98040104e-01 2.64104277e-01 7.13934422e-01 -1.27503788e+00
5.52920818e-01 1.29420185e+00 4.54575628e-01 8.04319859e-01
-1.45949686e+00 -9.18539643e-01 1.44461691e-01 1.72749266e-01
-1.54779685e+00 -4.28048551e-01 1.31955862e+00 -3.57088149e-01
2.17058539e-01 3.85496944e-01 4.78288323e-01 1.39374995e+00
-1.57634810e-01 9.74590003e-01 1.46171665e+00 -3.50194126e-01
3.59906077e-01 5.54549336e-01 1.91901587e-02 4.68874067e-01
-2.45229080e-02 4.44724947e-01 -3.32204014e-01 -2.38007739e-01
5.94359040e-01 2.52326250e-01 -4.39741313e-01 -6.40035987e-01
-1.08613122e+00 8.13823879e-01 6.20166898e-01 2.77670562e-01
-1.07085943e-01 -4.64747429e-01 2.07950324e-01 3.70866597e-01
6.24027669e-01 2.80287653e-01 -1.98968247e-01 5.67250848e-01
-7.39429951e-01 -3.26084867e-02 4.98479664e-01 1.08134091e+00
9.94890392e-01 2.51909852e-01 -5.11661649e-01 1.10230458e+00
3.49920690e-01 5.89936614e-01 8.13603222e-01 -3.32062036e-01
3.99553359e-01 8.22424531e-01 7.93913826e-02 -8.73150051e-01
1.95592672e-01 -8.09886336e-01 -8.90019417e-01 1.09554693e-01
1.04167089e-01 7.08221272e-02 -1.22046340e+00 1.48555660e+00
5.64929366e-01 6.17310822e-01 4.60417360e-01 8.53133976e-01
1.13595462e+00 6.79837525e-01 -2.82151788e-01 -3.90946150e-01
9.58166122e-01 -7.91310191e-01 -4.02891964e-01 -1.62865698e-01
3.81114006e-01 -6.75497055e-01 1.03752398e+00 3.51974666e-01
-4.08558846e-01 -7.86608577e-01 -1.25659406e+00 2.71476209e-01
-2.20545530e-01 5.16796649e-01 2.10197642e-01 5.47455311e-01
-5.95041215e-01 3.97819966e-01 -2.17538401e-01 -4.62923199e-02
6.21844411e-01 7.76671395e-02 -4.48231220e-01 -5.96608341e-01
-1.09323978e+00 5.18602788e-01 3.93843263e-01 1.45526931e-01
-1.38416493e+00 -6.77782655e-01 -9.28261399e-01 -1.52157590e-01
5.29041469e-01 -4.89555985e-01 7.25969434e-01 -1.37280154e+00
-1.62367344e+00 8.32327247e-01 1.30552024e-01 -3.21499333e-02
5.22015214e-01 1.91019848e-02 -6.04835629e-01 7.52455145e-02
1.85746282e-01 4.52659607e-01 1.40186250e+00 -1.70113182e+00
-2.87284017e-01 -5.10841489e-01 -2.12715104e-01 3.55824649e-01
-1.53002784e-01 -4.11596358e-01 -2.44054750e-01 -8.74758899e-01
3.52715254e-01 -9.29491699e-01 -1.88686758e-01 -1.05672404e-01
-5.68399251e-01 -2.01492026e-01 9.13774967e-01 -4.56223965e-01
8.97678077e-01 -2.45438170e+00 2.42268115e-01 3.42196137e-01
1.69822380e-01 3.97578299e-01 -2.93521821e-01 -2.56214151e-03
-4.85022455e-01 -2.99848080e-01 -3.61779541e-01 -2.08174139e-01
-1.85077414e-01 4.67571542e-02 -6.99940145e-01 5.33073068e-01
4.85174000e-01 6.87063396e-01 -1.21673298e+00 -4.07395780e-01
2.95146465e-01 3.63343388e-01 -4.64443341e-02 5.49174190e-01
-1.57180429e-01 7.20661998e-01 -5.98002076e-01 9.57201362e-01
7.37862766e-01 -1.43982051e-02 -1.19699519e-02 -1.73184305e-01
2.97503799e-01 -1.64631054e-01 -1.33408272e+00 1.41044390e+00
-1.82174757e-01 1.88724965e-01 -4.16853964e-01 -1.21649611e+00
1.47150373e+00 2.41470724e-01 2.54034549e-01 -4.49294806e-01
8.21532533e-02 5.03881514e-01 -1.46010011e-01 -2.17711136e-01
2.69152343e-01 -4.11298513e-01 4.48252037e-02 1.18918680e-01
3.11599880e-01 -1.19546637e-01 -1.64935023e-01 8.38453621e-02
9.01905835e-01 4.15751398e-01 3.00906450e-01 6.32353947e-02
8.40840518e-01 -1.63386077e-01 7.76883960e-01 5.89341521e-01
-3.16259295e-01 8.26359928e-01 2.18466520e-01 -3.74078989e-01
-7.94136167e-01 -1.22202659e+00 -3.68885607e-01 7.21212268e-01
3.88771117e-01 7.95322508e-02 -4.88456875e-01 -1.14475811e+00
2.28189235e-03 7.30798364e-01 -8.08201611e-01 -5.83435535e-01
-1.19781196e-01 -5.81867397e-01 2.78544784e-01 4.49439913e-01
6.15711749e-01 -9.83076692e-01 1.88701637e-02 7.60622993e-02
-1.30414754e-01 -6.82360232e-01 -3.83221000e-01 1.47550419e-01
-6.58541083e-01 -1.02397418e+00 -9.51325774e-01 -8.52195919e-01
9.18961883e-01 3.96792710e-01 6.80098176e-01 -2.19369084e-01
-1.14756942e-01 2.52777100e-01 -7.18246520e-01 -7.00356364e-02
-6.79430366e-01 -2.96175122e-01 1.58949289e-02 8.19142699e-01
3.44507784e-01 -3.43573540e-01 -4.52001095e-01 6.86524093e-01
-8.92500997e-01 -6.47425503e-02 7.31400788e-01 1.32370877e+00
8.03963482e-01 -3.21954265e-02 1.06194520e+00 -1.06625676e+00
1.95618361e-01 -6.72409415e-01 -3.39842975e-01 4.31256622e-01
-6.20888352e-01 -3.89772281e-02 9.61764514e-01 -9.27304268e-01
-1.05430937e+00 9.36903954e-02 6.35321289e-02 -9.14021134e-01
-1.80876046e-01 2.52006233e-01 -6.30276024e-01 -1.65772259e-01
9.49559569e-01 7.53597260e-01 3.55184823e-03 -1.45126820e-01
2.97747195e-01 8.43135655e-01 4.57797587e-01 -6.56754673e-01
1.24326742e+00 4.19188321e-01 -1.56041235e-01 -5.33817887e-01
-8.95904899e-01 -5.37261844e-01 -4.13740486e-01 -2.88855135e-01
4.98580545e-01 -1.03002608e+00 1.05342850e-01 7.26466775e-01
-6.88338578e-01 -1.27078565e-02 -6.08892024e-01 4.19850647e-01
-5.48182487e-01 2.64043450e-01 -4.59935404e-02 -7.35021114e-01
-1.64189368e-01 -1.37406194e+00 1.19787169e+00 3.03080946e-01
3.13502401e-01 -7.87811577e-01 -5.89806959e-03 4.66448784e-01
-5.34032956e-02 3.18704039e-01 7.28727818e-01 -1.07538557e+00
-4.76838827e-01 -4.90895033e-01 -8.95837471e-02 6.90321207e-01
3.21707428e-01 -3.42912465e-01 -1.16620970e+00 -5.11468887e-01
-1.24800131e-02 -7.10881293e-01 8.27415168e-01 8.30670893e-02
1.10677087e+00 -1.08705148e-01 -4.83148098e-01 5.62166095e-01
1.44685292e+00 4.11030769e-01 9.11567211e-01 -9.79683399e-02
7.49852836e-01 3.49384367e-01 1.01120806e+00 3.84023577e-01
-1.84485510e-01 5.19236088e-01 1.92871660e-01 -5.63500635e-02
-6.49855435e-02 -4.02662456e-01 5.38741112e-01 5.03531694e-01
2.16503233e-01 -1.60355136e-01 -4.37660843e-01 4.35783863e-01
-1.67879546e+00 -9.77871537e-01 3.30968261e-01 2.60202456e+00
9.56586659e-01 1.77222341e-02 -5.13022728e-02 3.67757469e-01
1.11193061e+00 1.46968246e-01 -8.87005210e-01 1.10034540e-01
-7.06409663e-02 1.86079368e-01 5.79221919e-02 -1.47252379e-03
-1.11902595e+00 6.60037220e-01 5.12240028e+00 1.29923534e+00
-1.02840257e+00 -3.87880132e-02 7.71412969e-01 2.53153145e-01
-4.99858946e-01 1.48587942e-01 -8.64189923e-01 4.46334273e-01
4.57937390e-01 -6.91130757e-03 5.26361346e-01 1.06204832e+00
-3.01021367e-01 4.01182957e-02 -1.33533847e+00 1.08394277e+00
5.13434470e-01 -9.65698123e-01 2.71097660e-01 -1.44160181e-01
8.58747303e-01 -4.97705132e-01 2.80377895e-01 7.45321929e-01
1.63139135e-01 -7.17039585e-01 4.95948374e-01 5.95863640e-01
1.26756263e+00 -7.54136741e-01 6.50213361e-01 3.50425154e-01
-1.10387921e+00 -3.08426082e-01 -4.40551698e-01 4.15340364e-01
-3.79671037e-01 6.81205451e-01 -7.88809657e-01 1.00779963e+00
3.81972700e-01 1.02558255e+00 -6.99121237e-01 8.24905992e-01
-2.24650294e-01 5.67915142e-01 4.81920578e-02 2.12296411e-01
-7.17810020e-02 -3.71660560e-01 9.02684987e-01 5.64253211e-01
2.63054073e-01 -1.48272008e-01 4.05702591e-01 9.17985797e-01
-1.03982069e-01 3.24690565e-02 -8.97045910e-01 2.06241995e-01
5.84463537e-01 1.25496197e+00 -5.64181924e-01 -4.07363504e-01
-3.08613867e-01 9.50984776e-01 3.38437408e-01 4.33108628e-01
-9.06762898e-01 -4.69122589e-01 1.22642159e-01 -2.66674105e-02
2.76522636e-01 5.15824676e-01 -1.18532434e-01 -1.39224076e+00
1.13519669e-01 -1.05866289e+00 4.44925785e-01 -9.34505463e-01
-1.61513591e+00 5.92207730e-01 -8.06366056e-02 -2.00039792e+00
-1.20308712e-01 -4.61826622e-01 -4.14348781e-01 8.57777953e-01
-1.43293881e+00 -1.34705424e+00 -3.26600641e-01 6.98615074e-01
6.61596060e-01 -5.24662673e-01 7.75060236e-01 1.74278095e-01
-6.22638404e-01 8.99156213e-01 4.91899014e-01 1.53837442e-01
8.27611566e-01 -1.19766617e+00 -3.33052427e-01 9.14379299e-01
4.26546931e-02 4.70235050e-01 3.04563642e-01 -6.52834296e-01
-1.24300838e+00 -1.57184052e+00 1.98015004e-01 -2.40762621e-01
4.22283828e-01 -3.64438623e-01 -1.02648902e+00 7.24942863e-01
-2.66598493e-01 4.54417557e-01 8.29761207e-01 -1.36093318e-01
-6.30437553e-01 -4.83465284e-01 -1.48162866e+00 4.49627310e-01
8.20058346e-01 -3.73285294e-01 -7.50343978e-01 2.33023316e-01
7.71270156e-01 -4.09507900e-01 -7.65472293e-01 5.66568851e-01
4.04227972e-01 -7.10591733e-01 1.01332533e+00 -2.78148115e-01
5.00619888e-01 -5.94922066e-01 -2.20968917e-01 -1.64873087e+00
-4.56404299e-01 -2.67596394e-01 -1.10961854e-01 1.64988256e+00
1.05409242e-01 -1.00284362e+00 4.38921750e-01 2.24744037e-01
-3.78569365e-01 -8.12569261e-01 -1.03362429e+00 -8.98745239e-01
-1.71359375e-01 -2.06240922e-01 5.33279359e-01 9.36154544e-01
-4.05369461e-01 4.87158686e-01 -3.81131828e-01 1.07170627e-01
7.40846932e-01 4.06986713e-01 8.39862049e-01 -1.19114864e+00
-4.91597742e-01 5.24568632e-02 -5.02087474e-01 -8.98749173e-01
2.99612701e-01 -1.02771437e+00 2.41125017e-01 -1.03249288e+00
2.94213295e-01 -6.94468141e-01 -4.65903044e-01 3.82930964e-01
-4.44417000e-01 3.32154512e-01 7.77406767e-02 3.23457330e-01
-4.57824171e-01 1.00976229e+00 1.47611773e+00 -3.46999824e-01
-3.33415836e-01 9.98622105e-02 -8.30250919e-01 4.26354319e-01
2.83325434e-01 -3.88218343e-01 -6.92631841e-01 1.88085645e-01
-3.33330214e-01 2.68595874e-01 5.38447499e-01 -1.21966362e+00
-3.39823246e-01 -1.90232605e-01 6.93610966e-01 -5.41118860e-01
2.32176855e-01 -9.21662450e-01 2.25520998e-01 2.76427001e-01
-3.50979269e-01 -7.75810182e-01 -1.27047047e-01 1.00306952e+00
-1.61830544e-01 -3.14845383e-01 1.06211305e+00 1.38069242e-01
-6.89971507e-01 4.02776480e-01 3.90869454e-02 9.31131616e-02
1.36881399e+00 -5.61461806e-01 -3.24873447e-01 5.46944030e-02
-7.42684960e-01 1.37111157e-01 3.12445670e-01 3.80244941e-01
9.93209720e-01 -1.70163310e+00 -8.08115363e-01 5.27352333e-01
6.14825308e-01 3.29181343e-01 3.08731019e-01 5.65122783e-01
1.08212255e-01 -3.74017060e-02 1.80072081e-03 -7.07120061e-01
-8.66971374e-01 1.01170504e+00 2.76601493e-01 -4.58621025e-01
-3.63807619e-01 6.40744328e-01 4.38589722e-01 -6.41266882e-01
-3.71332392e-02 3.19368355e-02 -3.39402258e-01 6.71766838e-03
4.73254979e-01 3.10481578e-01 -1.37082469e-02 -6.92232966e-01
-2.43410379e-01 4.79367644e-01 -7.16873258e-02 2.48565391e-01
1.09406424e+00 -1.78772174e-02 1.97467521e-01 6.61312759e-01
1.22199488e+00 -1.21372446e-01 -1.35789406e+00 -5.61091781e-01
-5.02193451e-01 -6.35341048e-01 -2.47661635e-01 -8.16387653e-01
-1.01514435e+00 5.42090535e-01 8.77186418e-01 -8.21967330e-03
1.41632116e+00 -4.58898656e-02 5.36755919e-01 3.17683183e-02
5.72427750e-01 -8.19147706e-01 3.99975210e-01 6.60403967e-02
1.08914459e+00 -1.42450559e+00 -7.54422694e-02 -5.03725886e-01
-7.38529682e-01 9.82721210e-01 8.76801312e-01 -3.74715745e-01
4.81548399e-01 -2.99124092e-01 -1.84294373e-01 2.50641376e-01
-5.55603325e-01 -1.57293081e-01 3.79852146e-01 9.58947480e-01
-1.11784995e-01 1.87222332e-01 -1.52606934e-01 6.57434940e-01
3.70008647e-01 1.52726442e-01 2.98233420e-01 6.49918437e-01
-1.42545611e-01 -9.85781312e-01 -7.56761551e-01 6.01968229e-01
8.53866041e-02 9.48813334e-02 -2.87297904e-01 5.21684527e-01
3.30050677e-01 8.69059741e-01 -1.00740336e-01 -8.75817895e-01
1.89549044e-01 7.06822425e-02 4.26003069e-01 -8.21620762e-01
-2.96441019e-01 3.24286632e-02 -1.39586255e-01 -1.74763560e-01
-4.10287231e-01 -7.13544488e-01 -1.00499296e+00 2.91326672e-01
-7.57615507e-01 1.10798493e-01 9.27340761e-02 9.89083409e-01
3.12453508e-01 4.56323624e-01 1.35941911e+00 -8.26783478e-01
-9.46323037e-01 -1.09690440e+00 -7.05610394e-01 6.09639168e-01
2.63040215e-01 -1.00748706e+00 -4.74545389e-01 1.07270718e-01] | [10.071480751037598, 2.9308316707611084] |
1c54a37b-d352-4b05-9507-765e5beb84eb | how-to-solve-fair-k-center-in-massive-data | 2002.07682 | null | https://arxiv.org/abs/2002.07682v2 | https://arxiv.org/pdf/2002.07682v2.pdf | How to Solve Fair $k$-Center in Massive Data Models | Fueled by massive data, important decision making is being automated with the help of algorithms, therefore, fairness in algorithms has become an especially important research topic. In this work, we design new streaming and distributed algorithms for the fair $k$-center problem that models fair data summarization. The streaming and distributed models of computation have an attractive feature of being able to handle massive data sets that do not fit into main memory. Our main contributions are: (a) the first distributed algorithm; which has provably constant approximation ratio and is extremely parallelizable, and (b) a two-pass streaming algorithm with a provable approximation guarantee matching the best known algorithm (which is not a streaming algorithm). Our algorithms have the advantages of being easy to implement in practice, being fast with linear running times, having very small working memory and communication, and outperforming existing algorithms on several real and synthetic data sets. To complement our distributed algorithm, we also give a hardness result for natural distributed algorithms, which holds for even the special case of $k$-center. | ['Sivaramakrishnan Natarajan Ramamoorthy', 'Sagar Kale', 'Ashish Chiplunkar'] | 2020-02-18 | null | null | null | null | ['data-summarization'] | ['miscellaneous'] | [-3.22423764e-02 3.71868664e-04 -3.22307110e-01 -3.93436551e-01
-8.69346142e-01 -4.46505040e-01 1.55929774e-01 8.37715328e-01
-5.62503219e-01 8.41986001e-01 1.68104693e-01 -2.15026751e-01
-5.85693300e-01 -1.13511944e+00 -5.95554233e-01 -7.10175335e-01
-4.93924707e-01 9.43735957e-01 5.54890752e-01 -1.69280186e-01
4.40101832e-01 3.50712627e-01 -1.73381889e+00 -4.16077450e-02
1.01939261e+00 1.35432124e+00 -1.69307306e-01 8.49157453e-01
-3.05236280e-01 1.09740961e+00 -4.08172816e-01 -5.44053555e-01
6.19271338e-01 -4.87481952e-01 -1.09025753e+00 -1.69723406e-01
3.25611979e-01 -4.97528285e-01 -2.33631611e-01 9.51611996e-01
6.65659904e-01 2.54747540e-01 2.12795451e-01 -1.55016935e+00
-1.39115512e-01 1.17338705e+00 -9.44142044e-01 1.79128021e-01
2.54337519e-01 -4.33503807e-01 1.29747200e+00 -2.34154642e-01
4.92787361e-01 1.09487987e+00 3.58525932e-01 3.08417827e-01
-9.37685311e-01 -5.75952768e-01 -6.56532347e-02 3.43610972e-01
-1.33780658e+00 -3.66692007e-01 4.46170330e-01 1.68796137e-01
5.70340097e-01 9.20414686e-01 6.56243205e-01 -3.07603955e-01
-1.97828397e-01 9.71947491e-01 9.81871128e-01 -3.79038393e-01
9.21480000e-01 -2.13566110e-01 2.56384909e-01 3.12810510e-01
6.53452158e-01 -4.07525480e-01 -5.47321558e-01 -8.41164470e-01
6.22266065e-03 1.92480996e-01 -1.35314956e-01 -2.35341817e-01
-9.53535795e-01 9.26293612e-01 1.58182859e-01 1.05724279e-02
-4.93544787e-01 4.75271046e-01 5.97333550e-01 4.79216188e-01
6.94273472e-01 -1.75609961e-01 -2.09561959e-01 -3.28147501e-01
-1.35054815e+00 8.60415876e-01 1.33866370e+00 1.03030443e+00
5.89681566e-01 -3.03185284e-01 2.91551147e-02 5.20689011e-01
-2.01202616e-01 7.14223206e-01 2.21046418e-01 -1.47696221e+00
3.84235173e-01 3.89270723e-01 3.44556987e-01 -9.70883965e-01
-1.80142462e-01 8.58629271e-02 -1.12917030e+00 -6.58830181e-02
3.24499428e-01 1.24832474e-01 7.12417960e-02 1.60339665e+00
6.63402200e-01 -3.05244058e-01 -5.95507920e-02 8.29194009e-01
2.42269456e-01 9.29154098e-01 -2.63467044e-01 -1.05913854e+00
1.04288912e+00 -8.14531207e-01 -7.21458852e-01 3.73431683e-01
4.80368763e-01 -5.89050889e-01 5.81076145e-01 5.74173331e-01
-1.84627271e+00 1.14848450e-01 -6.92487240e-01 -9.37836394e-02
5.54327928e-02 -9.04677153e-01 8.05954218e-01 6.58652306e-01
-1.18311870e+00 6.01755917e-01 -5.92250109e-01 -2.11253449e-01
6.20993137e-01 1.98520467e-01 9.75258052e-02 -3.01894993e-01
-6.91804409e-01 2.31413215e-01 2.70786405e-01 -4.38873708e-01
-5.01413882e-01 -8.47002506e-01 -2.98255533e-01 3.96515220e-01
6.25343621e-01 -8.96307588e-01 1.25366080e+00 -1.00063610e+00
-1.26928461e+00 7.17857897e-01 -1.83746114e-01 -7.66536415e-01
7.63338089e-01 2.59966344e-01 2.83469260e-01 4.93591309e-01
-7.96100870e-02 8.06676075e-02 3.68868947e-01 -9.23772752e-01
-1.01389790e+00 -7.05736160e-01 -6.98317215e-02 2.05777928e-01
-6.99504077e-01 4.66078043e-01 1.38997927e-01 -2.61304498e-01
-1.01942234e-01 -4.33290124e-01 -6.81739867e-01 2.68153161e-01
-5.61411567e-02 -6.09869540e-01 3.44654471e-01 -8.90517756e-02
1.55384862e+00 -1.92898738e+00 -1.37246728e-01 2.45901987e-01
6.19243443e-01 1.19914904e-01 1.89874157e-01 9.55706358e-01
5.87350726e-01 2.38752604e-01 -2.99184531e-01 -2.00590819e-01
3.53420734e-01 1.57500997e-01 -4.31753993e-01 6.70632958e-01
-7.33918965e-01 4.28170055e-01 -8.54373217e-01 -7.35967040e-01
-3.12998861e-01 -4.81689423e-01 -7.27134526e-01 3.42017174e-01
-4.20417041e-01 -5.21582365e-01 -3.54260534e-01 2.84288853e-01
1.07737708e+00 -1.72205269e-01 6.20768294e-02 4.82829660e-01
-3.33410501e-01 -1.76227480e-01 -1.66519260e+00 1.28297591e+00
-2.77009513e-02 9.76412371e-02 7.52041101e-01 -1.32844090e+00
7.98825562e-01 3.22634459e-01 7.55387127e-01 -4.64678556e-01
8.33951309e-02 6.15713418e-01 -4.56333756e-01 -2.80852318e-01
9.29279566e-01 -2.40712866e-01 -2.92171389e-01 1.32489228e+00
-6.09371781e-01 -3.77881229e-01 4.48335081e-01 7.93705523e-01
1.19444776e+00 -9.37928975e-01 1.58276513e-01 -4.61306840e-01
1.04452819e-01 -3.86208296e-02 6.41099632e-01 9.22409475e-01
-3.73070747e-01 2.46898279e-01 7.75695384e-01 -6.93327367e-01
-1.10810149e+00 -7.96065092e-01 1.93123907e-01 1.30706584e+00
4.92999583e-01 -6.41557574e-01 -7.98843324e-01 5.16491644e-02
1.80350661e-01 4.46330220e-01 -3.49879175e-01 2.25863963e-01
-2.19149858e-01 -5.24078846e-01 4.00907308e-01 3.90160650e-01
4.56942677e-01 -5.55527508e-01 -7.85237432e-01 1.25626415e-01
-2.86884516e-01 -9.69639838e-01 -6.70656562e-01 -1.33804813e-01
-9.39128697e-01 -9.84250307e-01 -5.71162403e-01 -4.67825323e-01
4.72914964e-01 6.59590960e-01 1.03153968e+00 2.78222769e-01
-2.39990234e-01 3.80592883e-01 -1.74769357e-01 -8.43439460e-01
-1.92114368e-01 -4.97260019e-02 -7.22642615e-02 -2.55580954e-02
1.34666413e-01 -4.91947323e-01 -7.95269191e-01 1.91651285e-01
-1.28682005e+00 -1.03170261e-01 -8.52258578e-02 6.20720983e-01
7.43901849e-01 4.19301331e-01 8.48365963e-01 -1.05688620e+00
7.23710299e-01 -6.75212502e-01 -6.48850977e-01 1.36744693e-01
-7.19713986e-01 -1.90380573e-01 1.04896951e+00 2.70660341e-01
-7.66897857e-01 -2.91086406e-01 3.43092903e-02 2.99558323e-02
2.59954393e-01 3.20051104e-01 -1.18602060e-01 2.99869478e-01
5.72130084e-01 8.88502076e-02 3.93034905e-01 -2.21134871e-01
5.17778754e-01 9.12288249e-01 2.59812862e-01 -8.11056733e-01
4.60793257e-01 9.36278880e-01 1.98982775e-01 -6.78897798e-01
-6.03557348e-01 -7.64811099e-01 -4.40285495e-03 -9.13276970e-02
1.54669713e-02 -6.41793549e-01 -1.00557041e+00 6.60196960e-01
-7.75665283e-01 -1.92543134e-01 -7.62583017e-01 9.70839188e-02
-9.30327117e-01 6.29897773e-01 -6.63399756e-01 -1.10065734e+00
-9.71776605e-01 -4.01390493e-01 5.69488406e-01 2.90192097e-01
2.51585305e-01 -7.06979156e-01 1.82607949e-01 4.45452511e-01
8.56179595e-01 3.30708504e-01 5.50701737e-01 -8.47802639e-01
-6.82071865e-01 -5.23891926e-01 -3.86164993e-01 1.06453359e-01
-3.24547976e-01 1.98043972e-01 -4.83568132e-01 -6.44508123e-01
-1.00003883e-01 -6.87775254e-01 7.26198256e-01 3.65664750e-01
1.70182323e+00 -6.17080212e-01 -5.43972626e-02 3.55647594e-01
1.50897157e+00 -5.34198821e-01 5.03610432e-01 4.09651212e-02
3.04060820e-02 5.31040251e-01 6.99522197e-01 1.24695361e+00
6.17926896e-01 8.10250193e-02 4.86054212e-01 2.79251695e-01
4.17195171e-01 -7.47848526e-02 -5.78867048e-02 1.11302507e+00
-2.97584891e-01 -3.48660976e-01 -5.06693840e-01 8.57584238e-01
-2.31860662e+00 -1.44343185e+00 -3.20074737e-01 2.66136098e+00
9.59319592e-01 -5.77777773e-02 7.71104693e-01 4.96326417e-01
7.14807868e-01 1.68435603e-01 -3.53521645e-01 -8.36943567e-01
-1.47879153e-01 4.05028790e-01 7.43774116e-01 1.49407804e-01
-5.29511631e-01 4.97896284e-01 6.05098486e+00 1.27697635e+00
-5.22836745e-01 3.76256295e-02 7.17292845e-01 -5.33531070e-01
-8.19949031e-01 -7.44142085e-02 -2.86002725e-01 5.29617608e-01
1.32453215e+00 -1.19735193e+00 5.76204717e-01 8.37725759e-01
3.12833011e-01 -3.58029872e-01 -8.46563816e-01 1.06923687e+00
-1.53022662e-01 -1.52701223e+00 -1.66120213e-02 1.13518134e-01
9.51571167e-01 -1.65746883e-01 -2.48171285e-01 -1.69986505e-02
5.47579050e-01 -7.73743808e-01 6.05241656e-01 1.87935591e-01
7.32176006e-01 -1.27461112e+00 7.33695090e-01 7.97022700e-01
-8.67431700e-01 -1.72733843e-01 -8.50480199e-01 -3.41374606e-01
2.48050570e-01 8.90069485e-01 -3.69576029e-02 5.99997520e-01
8.71144295e-01 6.55290857e-02 8.24142992e-02 1.43119156e+00
4.17413622e-01 7.92402685e-01 -1.00540483e+00 -5.79634488e-01
2.42170296e-03 6.77719191e-02 3.22350889e-01 1.00471103e+00
2.66050220e-01 5.45149207e-01 3.33972454e-01 3.48930538e-01
-4.82735634e-01 6.07632220e-01 -3.12933892e-01 1.54905125e-01
9.43843246e-01 1.24110281e+00 -7.33606040e-01 -4.67035830e-01
-2.29815990e-01 6.76662982e-01 3.04785162e-01 -2.67241716e-01
-5.77096939e-01 -9.84087944e-01 4.29399014e-01 2.90567935e-01
3.69030058e-01 -5.51970713e-02 -5.37604034e-01 -8.36569548e-01
2.01318443e-01 -7.00760663e-01 8.12296391e-01 -2.12433413e-01
-1.46300912e+00 3.04951251e-01 -4.15339880e-02 -6.97274148e-01
-1.66880980e-01 4.37226221e-02 -7.30661452e-01 5.83072424e-01
-1.41119075e+00 -7.20836103e-01 -4.28494722e-01 9.27479863e-01
1.17775708e-01 2.60942370e-01 6.86635315e-01 3.50618005e-01
-8.61754343e-02 7.59388566e-01 7.38587141e-01 -4.37768400e-01
5.94341934e-01 -1.20375717e+00 -1.31548226e-01 7.18253195e-01
-3.79254788e-01 4.42253612e-02 8.06334376e-01 -8.45459923e-02
-1.89026237e+00 -8.49673867e-01 1.11940920e+00 1.09247431e-01
7.58225381e-01 -1.20630026e-01 -6.64640486e-01 1.31035864e-01
1.18667744e-01 3.09222400e-01 9.86178398e-01 5.80972387e-03
-1.84443504e-01 -7.03228056e-01 -1.46804523e+00 1.10289373e-01
1.04955530e+00 -7.57089853e-02 -2.86416233e-01 4.92617667e-01
6.57610774e-01 -2.84722656e-01 -7.84585655e-01 -1.55388445e-01
4.00160849e-01 -1.12706280e+00 4.27134037e-01 -5.01576960e-01
3.97178233e-01 1.31492510e-01 -2.39443362e-01 -9.44306791e-01
-3.56244184e-02 -1.13731778e+00 -2.22970441e-01 1.17209840e+00
5.59516735e-02 -7.90594041e-01 8.81437421e-01 9.92710173e-01
1.53476238e-01 -8.43620002e-01 -1.16423082e+00 -8.45087945e-01
3.16000730e-01 -3.42466056e-01 7.82949030e-01 6.94396317e-01
4.94653523e-01 -5.29199280e-02 -3.24325532e-01 -2.58253127e-01
1.04811215e+00 8.93107057e-01 1.02943540e+00 -1.31737244e+00
-2.21354440e-01 -4.92601901e-01 -1.92936659e-01 -1.12525821e+00
-9.29246619e-02 -8.36658239e-01 -1.64369166e-01 -1.42686152e+00
6.07006490e-01 -8.00094306e-01 -3.82894576e-02 2.50677466e-01
-1.53230214e-02 -1.19300578e-02 1.98346272e-01 3.91390860e-01
-1.32049954e+00 5.18703103e-01 8.15115452e-01 1.63489114e-02
2.61852760e-02 3.21346253e-01 -9.31948662e-01 3.95030886e-01
7.44597733e-01 -4.60848361e-01 -3.29521567e-01 -3.54864299e-01
5.06026328e-01 4.14573550e-01 -1.39436096e-01 -6.33786201e-01
7.17758954e-01 -5.44376671e-01 -5.14626622e-01 -6.66011095e-01
-3.55330557e-01 -7.14017987e-01 -2.83223148e-02 5.91139615e-01
-5.54861069e-01 7.48290643e-02 -3.07414532e-01 7.71695852e-01
-2.88103849e-01 -2.06284389e-01 8.88584077e-01 -1.20240696e-01
-1.46743417e-01 8.94559383e-01 -3.06120545e-01 6.65576339e-01
1.40854263e+00 -1.04412936e-01 -5.36927581e-01 -7.08477974e-01
-3.39074396e-02 7.82370090e-01 6.56458914e-01 -3.54997158e-01
4.69502002e-01 -1.12902665e+00 -1.12268794e+00 -2.82100201e-01
-4.59151790e-02 4.06566113e-01 4.39012349e-01 6.52444184e-01
-9.63111401e-01 1.05085194e-01 -1.40907029e-02 -1.94634885e-01
-1.27387500e+00 9.16238487e-01 -1.43319607e-01 -7.24330023e-02
-4.29565430e-01 6.70606434e-01 -4.78732586e-01 -7.72247324e-03
4.48994875e-01 8.85420889e-02 5.98462462e-01 -1.37992157e-02
1.09555876e+00 9.95003223e-01 4.09106500e-02 -1.39587894e-01
-2.81691968e-01 1.60576329e-02 -2.88985167e-02 -7.01627955e-02
1.66106415e+00 -3.48718256e-01 -7.85357833e-01 3.44531029e-01
1.02596200e+00 1.65098429e-01 -6.56162858e-01 -5.23529470e-01
-2.52942502e-01 -7.99720109e-01 3.21426913e-02 -2.28318021e-01
-1.09347606e+00 5.29725790e-01 6.65790886e-02 1.03865838e+00
1.31463087e+00 3.87036949e-02 1.30350351e+00 3.67846370e-01
9.45678294e-01 -1.51628137e+00 -3.51799041e-01 1.72899142e-01
4.45320815e-01 -8.99184108e-01 3.51720899e-01 -2.09064141e-01
-2.33326182e-01 1.19198918e+00 2.53987789e-01 -1.89470619e-01
5.85515559e-01 4.03118044e-01 -4.10402238e-01 -9.04030353e-02
-1.29379725e+00 -2.21870020e-02 -5.73310971e-01 6.00926653e-02
5.15702739e-02 3.50679696e-01 -1.15832126e+00 7.68615484e-01
-1.60561189e-01 2.95801163e-01 8.53361487e-01 1.15209579e+00
-1.01443958e+00 -9.04854000e-01 -2.68137515e-01 6.58033669e-01
-7.94316769e-01 2.27922842e-01 -2.37760305e-01 1.36771560e-01
-4.82359558e-01 1.10011029e+00 3.10457259e-01 1.11179546e-01
3.75317745e-02 -3.97277325e-01 1.81817800e-01 -1.25236690e-01
-5.22154570e-01 -3.21564138e-01 -1.39677718e-01 -5.19474685e-01
-3.40518892e-01 -5.11401713e-01 -1.34392893e+00 -1.43623757e+00
-3.55905443e-01 8.69569838e-01 5.30065775e-01 6.24000132e-01
3.99547637e-01 -2.13980883e-01 1.33614767e+00 -4.38774645e-01
-1.25269866e+00 -4.70690995e-01 -1.13551688e+00 3.15651894e-01
-2.51896922e-02 3.00012380e-01 -5.20603240e-01 -2.28910401e-01] | [6.613731384277344, 4.905677795410156] |
f4e629ad-206d-472e-85f8-9b4bc105037b | self-ensembling-contrastive-learning-for-semi | 2105.12924 | null | https://arxiv.org/abs/2105.12924v2 | https://arxiv.org/pdf/2105.12924v2.pdf | Self-Ensembling Contrastive Learning for Semi-Supervised Medical Image Segmentation | Deep learning has demonstrated significant improvements in medical image segmentation using a sufficiently large amount of training data with manual labels. Acquiring well-representative labels requires expert knowledge and exhaustive labors. In this paper, we aim to boost the performance of semi-supervised learning for medical image segmentation with limited labels using a self-ensembling contrastive learning technique. To this end, we propose to train an encoder-decoder network at image-level with small amounts of labeled images, and more importantly, we learn latent representations directly at feature-level by imposing contrastive loss on unlabeled images. This method strengthens intra-class compactness and inter-class separability, so as to get a better pixel classifier. Moreover, we devise a student encoder for online learning and an exponential moving average version of it, called teacher encoder, to improve the performance iteratively in a self-ensembling manner. To construct contrastive samples with unlabeled images, two sampling strategies that exploit structure similarity across medical images and utilize pseudo-labels for construction, termed region-aware and anatomical-aware contrastive sampling, are investigated. We conduct extensive experiments on an MRI and a CT segmentation dataset and demonstrate that in a limited label setting, the proposed method achieves state-of-the-art performance. Moreover, the anatomical-aware strategy that prepares contrastive samples on-the-fly using pseudo-labels realizes better contrastive regularization on feature representations. | ['Shaoting Zhang', 'Qing Xia', 'Wenji Wang', 'Zhuowei Li', 'Jinxi Xiang'] | 2021-05-27 | null | null | null | null | ['semi-supervised-medical-image-segmentation'] | ['computer-vision'] | [ 5.63527763e-01 4.71182674e-01 -3.98085177e-01 -7.84205317e-01
-1.12227178e+00 -2.44503468e-01 1.75213113e-01 1.20933294e-01
-7.18312144e-01 6.90023124e-01 -1.99878439e-01 -1.50439620e-01
-2.26216074e-02 -6.01212561e-01 -7.52280712e-01 -9.41509664e-01
8.77431110e-02 5.28989255e-01 1.97756529e-01 3.40476662e-01
-8.48430470e-02 4.12485749e-01 -1.31003416e+00 1.75183520e-01
1.20837188e+00 1.07702422e+00 3.56172025e-01 2.87832588e-01
-1.16040014e-01 9.01385844e-01 -2.10082263e-01 -2.23786458e-01
4.14740980e-01 -6.76256359e-01 -9.70528364e-01 7.51041591e-01
3.11918914e-01 -3.95733386e-01 7.54599506e-03 1.23090971e+00
3.76227379e-01 -7.55453557e-02 9.33263838e-01 -6.32820189e-01
-5.90942800e-01 6.40470922e-01 -8.08989942e-01 1.37895405e-01
-3.80697310e-01 2.94541210e-01 8.05334032e-01 -7.04875410e-01
4.35283363e-01 7.99098849e-01 5.13215959e-01 6.40296936e-01
-1.33841085e+00 -5.90017140e-01 1.07998699e-01 -1.57553539e-01
-1.38265324e+00 -2.41980970e-01 1.05415452e+00 -4.67390954e-01
1.86763883e-01 9.35579017e-02 5.65350473e-01 7.44878829e-01
-1.92363504e-02 1.03970814e+00 1.34695518e+00 -3.71603340e-01
3.32219481e-01 3.61789405e-01 9.66142341e-02 1.16599429e+00
6.67260587e-02 -1.01304585e-02 1.24541029e-01 -2.25142166e-02
9.74592268e-01 2.50344157e-01 -3.81823152e-01 -5.71116149e-01
-1.03440094e+00 8.80984962e-01 6.92560315e-01 3.16920698e-01
-5.34605324e-01 -6.93256482e-02 4.71714854e-01 -1.32991537e-01
8.30402851e-01 2.38836780e-01 -3.82315546e-01 4.00565177e-01
-1.27701819e+00 -3.80964100e-01 3.93502951e-01 8.77166510e-01
1.00103736e+00 -2.80884467e-02 -2.43264124e-01 1.03750515e+00
1.20586112e-01 2.96177179e-01 8.03441405e-01 -7.04327047e-01
2.92187244e-01 5.71636856e-01 -1.88897014e-01 -5.55027187e-01
-5.04045129e-01 -8.74320924e-01 -1.02136612e+00 6.52850419e-02
4.20255005e-01 -1.28337815e-01 -1.27579105e+00 1.73748183e+00
5.46609461e-01 2.69553632e-01 -9.46245566e-02 9.47168171e-01
5.29732704e-01 4.14674729e-01 2.52048641e-01 -5.81003904e-01
1.27375829e+00 -1.36268234e+00 -6.79333091e-01 -3.39896157e-02
8.95468533e-01 -3.58785510e-01 1.38960135e+00 3.47903073e-01
-1.10352564e+00 -6.46078229e-01 -1.14490008e+00 9.60631371e-02
2.85185538e-02 4.95690912e-01 5.55259705e-01 6.01128101e-01
-6.56993985e-01 6.53491020e-01 -1.12918341e+00 2.33066648e-01
9.91958857e-01 4.08334196e-01 -2.81822026e-01 -7.20738024e-02
-8.00619125e-01 5.36769927e-01 4.94620234e-01 5.07400334e-02
-1.07959068e+00 -7.48028159e-01 -9.39407527e-01 -1.00103803e-01
5.96557856e-01 -5.64064741e-01 1.04781854e+00 -1.38215268e+00
-1.59958887e+00 1.16120553e+00 1.33812428e-01 -4.11191404e-01
7.36902118e-01 1.22781441e-01 -5.35810888e-02 6.57759190e-01
2.36722767e-01 9.07084167e-01 1.12096143e+00 -1.44418275e+00
-4.57825661e-01 -3.54867220e-01 -1.00655995e-01 1.69328734e-01
-3.97079349e-01 -3.92093211e-01 -2.71092892e-01 -6.36496246e-01
2.85236418e-01 -8.80821228e-01 -5.49907684e-01 1.81100622e-01
-5.04481971e-01 -3.10197268e-02 4.42809135e-01 -5.90243399e-01
8.84635627e-01 -2.21864009e+00 -8.40111151e-02 2.15664864e-01
4.14396495e-01 2.76383847e-01 -2.73152590e-02 -3.90417874e-01
-6.03371337e-02 -8.34130049e-02 -8.97505045e-01 -3.91364008e-01
-3.96937817e-01 2.78839648e-01 1.34170696e-01 8.04367840e-01
4.58021373e-01 8.86037946e-01 -1.03359306e+00 -1.02457106e+00
1.59475729e-01 2.52882183e-01 -7.38567173e-01 2.25924045e-01
-1.14351116e-01 8.78812969e-01 -5.45471072e-01 5.20686448e-01
7.59641349e-01 -6.69552803e-01 9.82927680e-02 -3.00646424e-01
1.45767525e-01 -1.57405660e-01 -8.46618354e-01 1.87795281e+00
-6.38922572e-01 4.44932580e-02 1.61778793e-01 -1.50060332e+00
7.48380482e-01 2.45084718e-01 6.20668888e-01 -5.26616573e-01
3.37943941e-01 5.18573642e-01 -1.47924453e-01 -5.91164708e-01
-3.23991962e-02 -4.46599096e-01 1.86359942e-01 4.18171912e-01
1.52659088e-01 -8.12720954e-02 1.53448164e-01 2.97926459e-02
8.08161438e-01 4.82741371e-02 1.88819855e-01 -4.66785461e-01
8.70705724e-01 -1.36443973e-01 6.26438439e-01 3.89187336e-01
-3.13241780e-01 6.80346012e-01 2.85743386e-01 -3.53463233e-01
-9.49757040e-01 -9.29740608e-01 -4.34811741e-01 8.06525648e-01
2.17081845e-01 1.57335326e-01 -9.80022788e-01 -1.30352890e+00
-3.65548253e-01 4.96716678e-01 -8.05705428e-01 -2.68782496e-01
-7.04447985e-01 -8.51206839e-01 2.41084054e-01 6.30087733e-01
6.62769616e-01 -1.02098596e+00 -7.45152175e-01 1.28458306e-01
-1.22634888e-01 -9.52000678e-01 -7.35210121e-01 5.68275630e-01
-1.10266984e+00 -9.68744636e-01 -1.21101213e+00 -1.19531727e+00
1.39542234e+00 1.01666022e-02 8.55255127e-01 7.39466539e-03
-3.86799902e-01 5.35275005e-02 -2.70789146e-01 1.08233683e-01
-3.69288027e-01 1.85054049e-01 -3.21453273e-01 2.64551401e-01
-1.46533921e-01 -5.20267487e-01 -9.52130020e-01 3.13808352e-01
-1.08073306e+00 2.10970059e-01 1.00026321e+00 1.25789940e+00
1.11489916e+00 -7.27452263e-02 6.38122737e-01 -1.35028398e+00
1.74465761e-01 -3.09754431e-01 -4.93031114e-01 3.23229015e-01
-6.28293633e-01 3.19719434e-01 9.23949778e-01 -5.90497494e-01
-1.13807607e+00 3.98748249e-01 -1.84795529e-01 -5.67742646e-01
-1.22798301e-01 3.58691186e-01 1.15316175e-02 -4.22959924e-02
6.02540493e-01 4.97376561e-01 1.36313066e-01 -3.32643569e-01
5.62281907e-01 7.11634874e-01 5.28714776e-01 -6.34820580e-01
5.57296157e-01 7.23556578e-01 -1.10550478e-01 -4.41897362e-01
-1.09266722e+00 -4.83625352e-01 -8.02531600e-01 -8.10009167e-02
9.57388699e-01 -7.53300905e-01 -3.76438916e-01 3.21289331e-01
-5.96622705e-01 -5.86779118e-01 -6.59233272e-01 7.26623654e-01
-6.84153378e-01 4.43492889e-01 -7.76668012e-01 -5.58916152e-01
-5.30506015e-01 -1.60015917e+00 1.07356513e+00 2.31712639e-01
2.93319225e-01 -8.90502155e-01 -1.96265742e-01 3.93601716e-01
1.53596982e-01 1.40526384e-01 7.78572917e-01 -7.27434456e-01
-4.73889172e-01 4.85751173e-03 -4.87607390e-01 7.91916370e-01
3.76816928e-01 -6.08487725e-01 -9.58218336e-01 -4.39006060e-01
1.47845834e-01 -6.93528771e-01 9.73957896e-01 6.04733765e-01
1.69659185e+00 -2.41270900e-01 -2.61737913e-01 8.75929177e-01
1.33165658e+00 7.78459162e-02 2.36245781e-01 -8.87384266e-02
7.88045347e-01 6.75333917e-01 5.83170533e-01 3.82496417e-01
1.53883517e-01 2.77796388e-01 2.10159019e-01 -3.61945391e-01
-3.03952932e-01 -1.39263526e-01 -2.24642947e-01 1.12484598e+00
1.75768852e-01 1.92937434e-01 -5.74811339e-01 5.16350031e-01
-1.44593716e+00 -4.47568893e-01 2.34075278e-01 2.01404285e+00
1.38082612e+00 1.31210700e-01 1.74624156e-02 1.28697604e-01
7.25488424e-01 -6.90616155e-03 -8.96428168e-01 1.95569053e-01
2.83975959e-01 2.54724801e-01 7.07697988e-01 4.05704707e-01
-1.32956755e+00 8.58974397e-01 5.34677362e+00 1.22186184e+00
-1.46219373e+00 4.01175410e-01 1.27163601e+00 1.88207105e-01
-2.58099377e-01 -2.09805921e-01 -5.04445851e-01 5.79964042e-01
7.02691913e-01 3.16853464e-01 2.41554230e-01 9.82839584e-01
3.60526480e-02 -4.64632250e-02 -1.01252961e+00 9.79014397e-01
6.92448094e-02 -1.23395765e+00 -1.01160496e-01 -4.53622378e-02
8.12686443e-01 -2.44587526e-01 5.55737019e-02 2.40318358e-01
-8.93574730e-02 -6.83292449e-01 7.17123151e-01 3.58586252e-01
1.10158026e+00 -7.01678813e-01 6.27820253e-01 4.12503302e-01
-9.83753741e-01 2.99805626e-02 -3.38750422e-01 4.81973469e-01
2.05902845e-01 7.82093167e-01 -6.36042535e-01 5.45363247e-01
3.26368302e-01 6.43458545e-01 -5.35720646e-01 8.97958398e-01
-2.21909240e-01 6.96826637e-01 -1.75875530e-01 1.68213263e-01
4.20673102e-01 -3.80648524e-01 7.51631334e-02 1.06819785e+00
-7.97583163e-02 1.93238243e-01 5.56224704e-01 9.27600980e-01
-2.21969634e-01 3.50643128e-01 -4.06166632e-03 2.30818018e-01
2.02689826e-01 1.44399476e+00 -1.25336421e+00 -6.10187113e-01
-2.06361637e-01 1.02652180e+00 4.08138037e-01 1.88256890e-01
-9.19616461e-01 -2.34919235e-01 -3.37182999e-01 1.34255260e-01
2.67819226e-01 1.27182618e-01 -4.01879460e-01 -1.10775232e+00
-1.94132298e-01 -6.42018437e-01 3.36802959e-01 -3.53688002e-01
-1.28632426e+00 7.97036767e-01 7.87176937e-02 -1.30841577e+00
-3.93703952e-03 -3.79879862e-01 -2.75086403e-01 5.78289747e-01
-1.74163318e+00 -1.15765011e+00 -3.23979974e-01 5.42491019e-01
5.92892468e-01 1.26152635e-01 4.91139084e-01 5.78019083e-01
-6.91048861e-01 8.02744329e-01 2.55520269e-02 2.46462420e-01
5.80062151e-01 -1.28315604e+00 -3.31808895e-01 5.22914529e-01
6.56377450e-02 4.32486057e-01 1.93616524e-01 -5.17454028e-01
-7.03709424e-01 -1.24417615e+00 1.91290423e-01 2.63903260e-01
3.64918411e-01 -1.93005860e-01 -9.19717968e-01 5.67994356e-01
7.83934351e-03 6.09496236e-01 6.38999343e-01 -2.72852749e-01
-2.51474958e-02 -2.04854682e-01 -1.53870463e+00 3.69215518e-01
8.65164101e-01 -4.24993366e-01 -4.98221099e-01 6.81181669e-01
8.03155601e-01 -3.74304354e-01 -8.83880675e-01 5.55572867e-01
2.58316517e-01 -6.27174973e-01 8.30837488e-01 -2.05129862e-01
4.44795668e-01 -1.72008201e-01 1.43457696e-01 -1.08312345e+00
-1.69482499e-01 -3.54081154e-01 2.61395633e-01 1.02183414e+00
4.50355023e-01 -6.15533471e-01 1.08320141e+00 4.40210074e-01
-5.43698251e-01 -1.32223713e+00 -7.42047489e-01 -7.45269418e-01
2.41725057e-01 -4.54536565e-02 2.25461811e-01 1.03170896e+00
-2.34326765e-01 2.67543793e-01 -2.24037245e-01 1.03402371e-02
8.81644368e-01 1.55279681e-01 2.16764286e-01 -8.72330427e-01
-4.11866605e-01 -3.20317864e-01 -1.18075773e-01 -1.19799542e+00
2.15148628e-01 -1.13709903e+00 2.73894072e-01 -1.26244342e+00
4.84397501e-01 -8.48186851e-01 -4.35629904e-01 5.54671109e-01
-3.41696709e-01 4.73299891e-01 -1.73171520e-01 4.00306582e-01
-6.99331343e-01 7.12619841e-01 1.78189206e+00 -3.85896683e-01
-9.94607136e-02 4.08749580e-02 -4.53359991e-01 6.92981720e-01
6.75174832e-01 -6.46926999e-01 -6.96819067e-01 -2.14803323e-01
-3.58240336e-01 4.81958687e-02 1.22123487e-01 -9.23128247e-01
-9.63363238e-03 9.36991125e-02 3.72512072e-01 -5.84473610e-01
1.01011872e-01 -7.04297781e-01 -4.14800555e-01 7.14664400e-01
-7.30324328e-01 -4.80072767e-01 -1.93184182e-01 5.56282103e-01
-2.01463997e-01 -6.45395815e-01 1.25113249e+00 -3.35480005e-01
-2.76405364e-01 5.74590325e-01 -8.67010430e-02 3.17810446e-01
1.04423940e+00 -3.15118171e-02 1.12789929e-01 4.91170958e-02
-9.10759628e-01 1.68873996e-01 2.40540892e-01 -3.02731395e-01
5.60558558e-01 -1.11626220e+00 -5.46322584e-01 3.18722904e-01
1.65696759e-02 3.05541903e-01 4.51211005e-01 1.17753005e+00
-6.50204599e-01 1.12165324e-01 -1.50163248e-01 -8.17086816e-01
-8.82886231e-01 7.73356318e-01 3.02418798e-01 -5.50393403e-01
-7.94553936e-01 1.05707467e+00 5.67433894e-01 -5.65633714e-01
2.43628353e-01 -5.56561589e-01 -1.78497776e-01 -5.67740463e-02
4.16268855e-01 6.89610243e-02 -1.60219725e-02 -4.73916560e-01
-2.24613070e-01 5.53988159e-01 -3.20369393e-01 7.07925111e-02
1.17534506e+00 -9.20608640e-02 1.72522202e-01 3.26269865e-01
1.67819476e+00 -8.99030641e-02 -1.74623525e+00 -3.55262101e-01
-3.44872065e-02 -2.92011678e-01 3.38810891e-01 -6.75794721e-01
-1.51594996e+00 9.19460714e-01 8.74305069e-01 -2.17211042e-02
1.28911114e+00 1.46566376e-01 9.84334528e-01 1.05317175e-01
2.11730912e-01 -9.99305069e-01 2.83321172e-01 -1.51648730e-01
4.13225770e-01 -1.53658795e+00 1.25303224e-01 -5.59293270e-01
-8.00753832e-01 8.77494812e-01 4.45261061e-01 -3.59984905e-01
7.60071874e-01 1.37951285e-01 8.45000800e-03 -1.30005836e-01
-3.13180894e-01 -2.87589043e-01 3.46181273e-01 2.86491841e-01
2.96687424e-01 2.16989785e-01 -5.14330328e-01 6.28268301e-01
2.01364443e-01 1.06703065e-01 2.26724669e-01 8.73607218e-01
-4.51551855e-01 -1.01588905e+00 -5.92198558e-02 6.83608830e-01
-5.28822839e-01 -7.37364367e-02 2.22638905e-01 5.42950988e-01
2.32386380e-01 4.96548206e-01 7.87471607e-03 -9.64674130e-02
-4.81171012e-02 -1.38310656e-01 6.69715226e-01 -7.65443623e-01
-4.72326338e-01 4.01481420e-01 -1.64948523e-01 -2.80034035e-01
-5.90772331e-01 -4.79046643e-01 -1.57965004e+00 3.58880401e-01
-6.93265676e-01 3.03097039e-01 4.63240445e-01 1.05747473e+00
2.72935703e-02 7.69293547e-01 9.17828918e-01 -7.68749177e-01
-9.34074938e-01 -9.48632598e-01 -7.24256575e-01 4.46040362e-01
1.92726344e-01 -6.23699069e-01 -3.86753976e-01 2.32481629e-01] | [14.689047813415527, -2.105760335922241] |
f5dfa716-1b40-4ed0-a7dc-018906c7bda2 | chosen-methods-of-improving-object | 2208.13591 | null | https://arxiv.org/abs/2208.13591v2 | https://arxiv.org/pdf/2208.13591v2.pdf | Chosen methods of improving small object recognition with weak recognizable features | Many object detection models struggle with several problematic aspects of small object detection including the low number of samples, lack of diversity and low features representation. Taking into account that GANs belong to generative models class, their initial objective is to learn to mimic any data distribution. Using the proper GAN model would enable augmenting low precision data increasing their amount and diversity. This solution could potentially result in improved object detection results. Additionally, incorporating GAN-based architecture inside deep learning model can increase accuracy of small objects recognition. In this work the GAN-based method with augmentation is presented to improve small object detection on VOC Pascal dataset. The method is compared with different popular augmentation strategies like object rotations, shifts etc. The experiments are based on FasterRCNN model. | ['Marcin Pietroń', 'Magdalena Stachoń'] | 2022-08-29 | null | null | null | null | ['small-object-detection'] | ['computer-vision'] | [ 3.78251731e-01 1.71630695e-01 1.49684295e-01 -2.28151262e-01
-3.28402221e-01 -3.95062417e-01 6.79919958e-01 -4.20396566e-01
-2.05545679e-01 9.49583232e-01 -4.55604792e-02 1.37192428e-01
1.64056391e-01 -1.02643383e+00 -7.46447146e-01 -7.58145928e-01
4.86963838e-01 5.85702479e-01 3.28855097e-01 -5.88145927e-02
1.63595781e-01 8.89740586e-01 -1.78234839e+00 3.76307607e-01
6.02581918e-01 8.51445079e-01 1.81432307e-01 9.84965742e-01
-2.15824962e-01 1.05591381e+00 -9.88046408e-01 -1.89474374e-01
7.14932084e-01 -5.36405444e-01 -3.97506863e-01 1.60594314e-01
7.77802110e-01 -5.22552073e-01 9.40561388e-03 8.37043405e-01
7.11304545e-01 6.59065247e-02 9.51647222e-01 -1.38731396e+00
-6.48313642e-01 2.93844104e-01 -8.40946078e-01 3.09826672e-01
6.41820952e-02 3.02393913e-01 3.39833051e-01 -7.65226007e-01
3.89243215e-01 1.20321906e+00 7.50021398e-01 8.75915706e-01
-1.13463688e+00 -6.59365058e-01 -2.50480622e-01 2.77357578e-01
-1.18922675e+00 -1.27146766e-01 6.48522854e-01 -1.98495582e-01
1.14659047e+00 4.08291280e-01 9.51992810e-01 1.00749040e+00
5.87102361e-02 8.38156879e-01 1.18495464e+00 -5.83500922e-01
1.10756852e-01 3.98599297e-01 4.69925441e-02 4.45788503e-01
5.95286012e-01 3.57813835e-01 -2.25903615e-01 2.15534821e-01
7.91542947e-01 6.38823509e-02 8.50500986e-02 -2.34082147e-01
-6.96829259e-01 9.31425154e-01 6.15467191e-01 2.42260918e-01
-5.07144570e-01 5.11645973e-01 3.89663577e-01 1.24823637e-01
2.55101193e-02 4.15929288e-01 -1.74000651e-01 -1.76502600e-01
-8.03864062e-01 3.09723467e-01 3.89795065e-01 1.08182955e+00
6.48202121e-01 7.85008013e-01 -3.82609695e-01 6.78758740e-01
3.53042006e-01 5.37041664e-01 7.26045191e-01 -5.05526841e-01
2.44786441e-01 8.75989318e-01 -1.19374029e-01 -7.26822495e-01
-2.61008650e-01 -8.69405568e-01 -7.30382681e-01 7.40912616e-01
2.90851504e-01 -1.06425583e-01 -1.49386299e+00 1.47275829e+00
4.32566255e-01 1.26368031e-01 1.38376251e-01 6.75733030e-01
9.10974383e-01 7.81342506e-01 3.34756859e-02 1.95456699e-01
1.27835059e+00 -9.93493259e-01 -5.99654257e-01 -1.85117185e-01
4.73278582e-01 -1.00714505e+00 7.62520611e-01 4.89804566e-01
-8.02484095e-01 -9.74301994e-01 -1.16134930e+00 1.15023784e-01
-5.28076351e-01 5.70895016e-01 6.33408427e-01 1.42000711e+00
-8.36471379e-01 -1.14051509e-03 -7.49449253e-01 -5.46353698e-01
9.08441424e-01 5.88130414e-01 6.92529231e-02 -8.08364972e-02
-4.13902998e-01 9.42108750e-01 7.72818983e-01 8.79025757e-02
-1.11740386e+00 -5.33294678e-01 -4.95656401e-01 -5.82657382e-02
9.03719887e-02 -7.44573295e-01 1.07391334e+00 -1.15108788e+00
-1.48898923e+00 5.85153043e-01 1.67448103e-01 -8.66891623e-01
7.12462723e-01 -3.04600388e-01 -2.23264731e-02 -4.06997532e-01
-3.20695728e-01 1.21716344e+00 1.06534243e+00 -1.23304558e+00
-6.25675380e-01 -3.60038847e-01 -1.12127863e-01 1.44922614e-01
-1.62203625e-01 -2.27330521e-01 1.46120951e-01 -7.10175276e-01
-1.85364559e-01 -1.01829100e+00 -2.27915555e-01 -2.97961235e-01
-4.29073304e-01 -9.33157951e-02 1.50862539e+00 -4.19138551e-01
5.90811551e-01 -1.90216672e+00 -2.59789735e-01 -5.31120300e-02
-1.89267248e-02 6.00151241e-01 -1.09986968e-01 2.65161693e-01
4.15005311e-02 -1.46084026e-01 9.85753909e-02 -1.93791643e-01
-2.65431643e-01 2.95600742e-01 -1.64011925e-01 3.24793249e-01
5.92218697e-01 1.09993517e+00 -3.53722781e-01 -4.01095510e-01
6.02970481e-01 6.68661892e-01 -7.04117000e-01 2.18819454e-01
-2.97195524e-01 3.70990068e-01 -2.01288044e-01 8.18695068e-01
1.01954865e+00 3.43873292e-01 -5.13884306e-01 -4.38145310e-01
1.75207764e-01 -2.58595407e-01 -1.32060695e+00 1.23479533e+00
-3.69244069e-01 8.89066160e-01 -4.33197737e-01 -9.97028887e-01
1.10937405e+00 5.67583255e-02 7.32195452e-02 -3.54776621e-01
3.91111493e-01 1.08534351e-01 5.89494944e-01 -1.77914053e-01
6.25602543e-01 8.35168641e-03 4.28596407e-01 1.62118047e-01
2.36196429e-01 -2.98475921e-01 1.60430729e-01 -1.03287943e-01
7.40685582e-01 2.76096612e-01 3.79893124e-01 -8.51024091e-02
3.82549316e-01 2.16933668e-01 1.63393483e-01 9.29639816e-01
7.62295127e-02 8.11073780e-01 1.34892210e-01 -5.14888108e-01
-1.35397840e+00 -8.47098649e-01 -2.52053022e-01 6.33835137e-01
-2.86704272e-01 9.29343775e-02 -7.14230955e-01 -7.17935920e-01
-2.71593750e-01 7.10915089e-01 -8.58480990e-01 -2.22278371e-01
-6.34765327e-01 -1.18847930e+00 5.79799414e-01 8.54452550e-01
8.93022239e-01 -1.19950211e+00 -1.11412740e+00 3.13511938e-02
3.82381916e-01 -8.13143611e-01 1.37170240e-01 4.01190460e-01
-1.00061655e+00 -9.40323472e-01 -7.78572023e-01 -6.33190215e-01
8.48699629e-01 8.72703455e-03 8.44431460e-01 -4.65466343e-02
-8.77514362e-01 3.33140045e-01 -5.70426881e-01 -1.18081748e+00
-5.55719852e-01 1.15301728e-01 -2.50697911e-01 -2.09021389e-01
6.61683440e-01 -2.41588160e-01 -6.41193926e-01 1.74707606e-01
-8.62284780e-01 -7.73395076e-02 9.15702999e-01 1.11424506e+00
4.67128426e-01 -3.50786671e-02 5.86640894e-01 -7.65088439e-01
3.96408677e-01 -2.16870978e-01 -6.83419764e-01 7.76651949e-02
-4.62954134e-01 -9.81713757e-02 4.39706773e-01 -6.73669338e-01
-1.05834377e+00 3.33585232e-01 -2.28939742e-01 -4.38655585e-01
-3.89191598e-01 -2.54046768e-01 -1.65399060e-01 -2.93541640e-01
9.01935279e-01 4.44931984e-01 -4.46726158e-02 -2.27368638e-01
3.44539106e-01 5.78768373e-01 1.89399198e-01 -2.37902582e-01
7.40310848e-01 2.60008305e-01 1.92557529e-01 -8.73688340e-01
-4.36189264e-01 -1.84617072e-01 -6.17298663e-01 -3.51020306e-01
7.34159172e-01 -7.14594245e-01 -3.42453331e-01 5.94849586e-01
-1.13484561e+00 -1.43515661e-01 -9.24153090e-01 4.15490299e-01
-4.77022588e-01 -5.33959083e-02 -2.09815994e-01 -1.10918498e+00
-5.99309862e-01 -1.01087415e+00 9.95287180e-01 5.63423693e-01
-2.03380119e-02 -6.04441047e-01 -8.55816603e-02 3.85200113e-01
7.61042118e-01 4.35373664e-01 4.15511370e-01 -7.10563660e-01
-7.75267959e-01 -5.52269340e-01 -2.07163602e-01 6.75895095e-01
2.83596158e-01 2.84764785e-02 -1.16430295e+00 -2.63795793e-01
1.76878944e-01 -4.75231349e-01 8.41503978e-01 5.21405756e-01
9.99701500e-01 -2.82649249e-01 -3.23275268e-01 4.77921098e-01
1.69120514e+00 4.84505862e-01 9.95087504e-01 2.97276825e-01
8.06865990e-01 1.87453553e-01 6.37111723e-01 3.07469308e-01
-3.24576110e-01 5.37072122e-01 6.84052050e-01 -3.62659730e-02
-7.60104775e-01 -5.00728935e-02 1.75522894e-01 2.04303578e-01
-1.84111610e-01 -5.61903059e-01 -7.20548987e-01 4.62135315e-01
-1.43046498e+00 -1.11583114e+00 -2.96802759e-01 2.07576394e+00
2.93036669e-01 1.58204108e-01 4.05804366e-01 3.63048017e-01
4.95325953e-01 -3.50051135e-01 -5.11354387e-01 -5.81515253e-01
-3.07155341e-01 2.80076206e-01 5.62126458e-01 1.57018542e-01
-8.57682168e-01 7.41185009e-01 6.41488409e+00 9.26639259e-01
-1.22027302e+00 2.91542023e-01 7.26000130e-01 -2.26580888e-01
2.44626045e-01 -2.86906004e-01 -1.30145323e+00 5.06434023e-01
6.33295059e-01 2.29617789e-01 3.89371999e-02 1.06059110e+00
-3.05147797e-01 -3.04757804e-01 -9.68643069e-01 9.04206216e-01
3.34808648e-01 -1.21880698e+00 3.76021802e-01 6.09057490e-03
1.05609477e+00 1.24603905e-01 2.20269009e-01 3.57741386e-01
2.65289426e-01 -1.27754223e+00 5.66142499e-01 3.55806857e-01
5.43403149e-01 -8.56648624e-01 1.17212939e+00 2.54893839e-01
-8.80383134e-01 -1.29794568e-01 -6.64600372e-01 6.05645590e-02
-2.26273034e-02 1.94188222e-01 -1.48856473e+00 2.94917345e-01
6.34332061e-01 2.33348697e-01 -9.03024554e-01 1.29292023e+00
3.56449932e-02 7.82669902e-01 -6.41414404e-01 -5.06931603e-01
1.58368111e-01 -1.02224201e-01 5.90703785e-01 1.27927732e+00
5.64197063e-01 -5.02723157e-02 -1.63981691e-01 9.50491786e-01
2.24490315e-01 1.09098956e-01 -8.60798419e-01 8.69830400e-02
2.97273636e-01 1.24458218e+00 -9.27876234e-01 -3.86830598e-01
-2.29516238e-01 8.01894963e-01 -5.88991269e-02 -3.23942937e-02
-9.06679153e-01 -1.48096487e-01 1.12190500e-01 3.64949197e-01
5.24281383e-01 4.65056002e-02 -3.88934523e-01 -5.55637002e-01
-1.13265850e-01 -9.58801091e-01 2.83981085e-01 -7.75423646e-01
-1.00571263e+00 7.44376004e-01 4.97790836e-02 -1.37959754e+00
-3.25979769e-01 -8.15295756e-01 -7.61333644e-01 6.35402441e-01
-1.02482796e+00 -1.77658916e+00 -8.00466359e-01 2.59523511e-01
9.82064426e-01 -7.09583640e-01 6.00194752e-01 3.41381162e-01
-3.33170682e-01 6.29400492e-01 -7.20416978e-02 -7.27089122e-02
2.16571182e-01 -1.21688807e+00 3.69598693e-03 9.83591020e-01
3.20560515e-01 2.21172005e-01 8.91732097e-01 -6.89108431e-01
-1.15684342e+00 -1.23012841e+00 -1.69997245e-01 -6.17669284e-01
-2.90445834e-02 -2.62405813e-01 -6.24389827e-01 7.47590363e-01
4.54785794e-01 3.07121724e-02 4.93048877e-01 -3.27977926e-01
1.92682892e-01 -1.66012675e-01 -1.57780790e+00 2.83677191e-01
7.15466440e-01 1.57107949e-01 -4.75281566e-01 3.06355089e-01
2.02026442e-01 -5.70376992e-01 -3.40010971e-01 5.14907002e-01
4.63543564e-01 -1.05235219e+00 8.92632782e-01 -5.05262792e-01
2.43807107e-01 -3.59122306e-01 -3.06684941e-01 -1.18040133e+00
-1.74579635e-01 -1.76573828e-01 -2.27919728e-01 1.48451602e+00
2.55732208e-01 -7.00077772e-01 1.32548928e+00 2.12902695e-01
3.54853682e-02 -6.64783716e-01 -9.29095566e-01 -9.93305743e-01
-1.03955433e-01 -1.92837462e-01 5.62066019e-01 5.24026453e-01
-8.49813879e-01 2.88811445e-01 -5.39947808e-01 -1.37396511e-02
6.43810153e-01 -1.61756888e-01 1.25939882e+00 -1.01084006e+00
-4.06501383e-01 -2.51611710e-01 -1.11710751e+00 -6.47681355e-01
-4.85834628e-01 -5.18938303e-01 -1.86506003e-01 -1.47167623e+00
2.92874396e-01 -6.27519190e-01 -3.77207994e-02 4.29134101e-01
1.47086103e-04 7.87118137e-01 2.52341479e-01 -8.77554864e-02
-9.29627046e-02 6.38353169e-01 1.17202091e+00 -2.46655375e-01
-1.14373185e-01 1.59754857e-01 -3.01224768e-01 4.30690795e-01
1.18359554e+00 -6.16146982e-01 -5.58282018e-01 -2.50641972e-01
-4.96744029e-02 -6.53146625e-01 4.85340446e-01 -1.61106193e+00
-2.00975202e-02 1.17668770e-01 9.44376171e-01 -9.09076691e-01
6.12025559e-01 -1.00942361e+00 4.76388097e-01 9.85573888e-01
1.09116159e-01 -3.51380594e-02 5.68761408e-01 4.61721659e-01
-1.63490236e-01 -6.12735093e-01 1.00726950e+00 -2.24168792e-01
-5.96603811e-01 -1.07599899e-01 -3.41654271e-01 -3.26528341e-01
1.55519259e+00 -7.95518994e-01 -2.47161463e-01 -8.17979127e-02
-5.24868190e-01 -3.12343210e-01 3.22936863e-01 5.44396460e-01
4.59911615e-01 -1.42826045e+00 -8.20767522e-01 2.95588195e-01
-7.01276436e-02 2.80498683e-01 9.63224694e-02 6.50176167e-01
-8.33971143e-01 3.83054584e-01 -7.70758212e-01 -8.56769919e-01
-1.50404549e+00 4.22756195e-01 5.03232360e-01 -1.55901626e-01
-3.03902805e-01 1.10120070e+00 2.20184430e-01 -2.66969383e-01
2.87522376e-03 -3.31227243e-01 -3.07165444e-01 -2.15566736e-02
3.41761619e-01 6.39534771e-01 6.13765009e-02 -3.75406533e-01
-2.17912793e-01 5.74723899e-01 -3.49454463e-01 2.03296721e-01
1.25059807e+00 2.57806271e-01 1.45373777e-01 1.80009633e-01
8.34943056e-01 -6.19728267e-02 -1.18246365e+00 2.18426183e-01
-3.17948312e-01 -6.94815099e-01 9.32021067e-03 -9.08039033e-01
-1.19850373e+00 8.83690655e-01 1.44124150e+00 6.15999401e-02
1.14925444e+00 -8.84403139e-02 2.64063984e-01 4.15118903e-01
2.20156834e-01 -9.49449480e-01 4.27109629e-01 1.29893914e-01
1.07438838e+00 -1.51322794e+00 2.09820628e-01 -2.95999855e-01
-4.78613853e-01 1.19213283e+00 1.27941954e+00 -3.67998749e-01
1.97354570e-01 6.01115823e-01 -1.55018149e-02 -2.22363099e-01
-4.00268465e-01 -1.40147731e-01 3.75575334e-01 1.05508828e+00
4.38947678e-01 -1.37194395e-01 -1.97310448e-01 1.32469073e-01
-3.56104553e-01 6.69447109e-02 6.11881912e-01 1.02009475e+00
-4.43160772e-01 -9.71726298e-01 -6.31987989e-01 9.01108503e-01
-4.75015908e-01 1.28516942e-01 -4.21680629e-01 9.49223995e-01
4.81113046e-01 6.71660662e-01 2.13912711e-01 -1.89058095e-01
1.97693989e-01 -1.61959529e-01 9.03937519e-01 -6.30133510e-01
-6.81160152e-01 -1.71282366e-01 -1.29507527e-01 -1.19191438e-01
-3.40182006e-01 -7.07934141e-01 -8.14481437e-01 -7.95022324e-02
-8.85903180e-01 -1.94589302e-01 8.57312441e-01 6.23188078e-01
1.47356570e-01 7.85713673e-01 1.68453366e-01 -8.66695762e-01
-4.71769512e-01 -1.44867003e+00 -2.29635760e-01 7.36920685e-02
1.53834283e-01 -5.61953723e-01 -1.66788530e-02 6.69318214e-02] | [8.542191505432129, -0.6365504264831543] |
5d8e32a8-a851-4217-a6ce-51bf7e714dc8 | all-in-one-multi-task-prompting-for-graph | 2307.01504 | null | https://arxiv.org/abs/2307.01504v1 | https://arxiv.org/pdf/2307.01504v1.pdf | All in One: Multi-task Prompting for Graph Neural Networks | Recently, ''pre-training and fine-tuning'' has been adopted as a standard workflow for many graph tasks since it can take general graph knowledge to relieve the lack of graph annotations from each application. However, graph tasks with node level, edge level, and graph level are far diversified, making the pre-training pretext often incompatible with these multiple tasks. This gap may even cause a ''negative transfer'' to the specific application, leading to poor results. Inspired by the prompt learning in natural language processing (NLP), which has presented significant effectiveness in leveraging prior knowledge for various NLP tasks, we study the prompting topic for graphs with the motivation of filling the gap between pre-trained models and various graph tasks. In this paper, we propose a novel multi-task prompting method for graph models. Specifically, we first unify the format of graph prompts and language prompts with the prompt token, token structure, and inserting pattern. In this way, the prompting idea from NLP can be seamlessly introduced to the graph area. Then, to further narrow the gap between various graph tasks and state-of-the-art pre-training strategies, we further study the task space of various graph applications and reformulate downstream problems to the graph-level task. Afterward, we introduce meta-learning to efficiently learn a better initialization for the multi-task prompt of graphs so that our prompting framework can be more reliable and general for different tasks. We conduct extensive experiments, results from which demonstrate the superiority of our method. | ['Jihong Guan', 'Bo Liu', 'Jia Li', 'Hong Cheng', 'Xiangguo Sun'] | 2023-07-04 | null | null | null | null | ['meta-learning'] | ['methodology'] | [ 2.70777255e-01 2.58627743e-01 -2.60326117e-01 -3.07653874e-01
-4.21518773e-01 -5.60202897e-01 5.87525249e-01 3.65360260e-01
-1.66507110e-01 2.85087109e-01 2.34327868e-01 -6.14810348e-01
-8.19262192e-02 -8.74109209e-01 -6.04376078e-01 -4.63269532e-01
7.29198307e-02 1.84836075e-01 3.11338395e-01 -2.20241651e-01
3.28940861e-02 1.40722945e-01 -9.17123377e-01 2.77973682e-01
1.12280285e+00 5.64527631e-01 6.64207399e-01 4.30883139e-01
-5.81098855e-01 7.79853582e-01 -5.77050805e-01 -4.59173232e-01
-1.62564386e-02 -1.60957202e-01 -1.05095565e+00 2.55513847e-01
1.38562486e-01 -8.37299898e-02 -3.81946564e-01 9.83238637e-01
4.22496289e-01 2.47838795e-01 3.95258248e-01 -1.55709314e+00
-8.41910958e-01 9.35885727e-01 -7.12586999e-01 -8.36354420e-02
4.67311770e-01 1.84566453e-01 1.10625601e+00 -6.95158184e-01
4.29178208e-01 1.19320917e+00 6.32239699e-01 3.73182148e-01
-7.52516687e-01 -6.14688098e-01 9.33546424e-01 9.70148966e-02
-1.21134436e+00 -7.52194524e-02 1.06265306e+00 -3.63449156e-01
1.02924824e+00 -4.23660353e-02 5.39242744e-01 1.19976914e+00
-1.28184482e-01 9.91999924e-01 6.12440467e-01 -3.99267286e-01
-3.04080576e-01 -2.81254768e-01 1.00941487e-01 9.05159175e-01
2.63214648e-01 -1.80125773e-01 -2.21240327e-01 1.21954396e-01
7.59376526e-01 2.13758260e-01 -3.29173297e-01 -1.18879393e-01
-1.16399121e+00 5.40023685e-01 6.41762137e-01 3.35212559e-01
-3.13136160e-01 1.93878338e-01 5.87344885e-01 3.31258506e-01
5.71419597e-01 2.95553744e-01 -6.19488597e-01 2.13846982e-01
-4.79550749e-01 2.08227746e-02 8.32361042e-01 1.28652024e+00
9.05302703e-01 -1.64484922e-02 -6.91249549e-01 8.95109832e-01
5.09430826e-01 9.22400728e-02 1.18112631e-01 -2.79226750e-01
8.39155436e-01 9.65760112e-01 -2.74214417e-01 -1.02685487e+00
-4.28808659e-01 -5.58855951e-01 -1.12925494e+00 -4.90601718e-01
4.00372177e-01 -2.60912329e-01 -1.07526827e+00 1.87340927e+00
3.53300333e-01 4.62480307e-01 -1.47371560e-01 6.71090186e-01
1.11745095e+00 6.45259380e-01 4.76281464e-01 -1.92395404e-01
1.56497848e+00 -1.38790917e+00 -6.08059704e-01 -5.29071689e-01
1.00327992e+00 -7.01757014e-01 1.65019095e+00 7.72462860e-02
-4.26975757e-01 -7.73603797e-01 -6.87673986e-01 -3.12475532e-01
-4.50571120e-01 -4.01628353e-02 9.64578867e-01 1.57023758e-01
-1.12084520e+00 4.94231999e-01 -5.48649549e-01 -4.82686847e-01
2.81173676e-01 1.15649037e-01 -2.83762187e-01 -4.93036091e-01
-1.36462283e+00 6.11350060e-01 5.38590312e-01 1.79234609e-01
-7.10441589e-01 -8.05576980e-01 -9.70336437e-01 3.28916460e-01
9.98335123e-01 -1.03329635e+00 1.15667844e+00 -5.48935354e-01
-1.34397364e+00 8.48733008e-01 -3.05952411e-02 6.43530488e-02
2.49836028e-01 -1.47365630e-01 -3.14988524e-01 -2.38684118e-01
1.92448512e-01 4.61515725e-01 8.48685086e-01 -1.15668499e+00
-5.23850560e-01 -1.00586802e-01 3.43768865e-01 3.81150275e-01
-3.89910161e-01 4.28052992e-02 -9.13356483e-01 -7.63239563e-01
-1.90393150e-01 -6.28342509e-01 -5.13068080e-01 -4.94858772e-01
-6.00888848e-01 -8.50891113e-01 7.09789217e-01 -4.33583260e-01
1.41035175e+00 -2.20154500e+00 2.03545280e-02 -1.54481633e-02
7.41687417e-01 3.03126067e-01 -5.85314572e-01 7.35954106e-01
-2.03567505e-01 3.42924178e-01 -8.87343287e-02 -4.96775299e-01
6.45028576e-02 4.45844889e-01 -1.66954428e-01 -1.52027039e-02
3.01955312e-01 1.26523888e+00 -1.40459991e+00 -6.45519912e-01
-1.67647805e-02 2.25004390e-01 -4.04468656e-01 5.65776765e-01
-4.87693518e-01 4.37097341e-01 -8.08344662e-01 6.95090950e-01
6.83639646e-01 -8.35622430e-01 1.75422639e-01 -2.09099084e-01
2.54750192e-01 3.92123938e-01 -9.95843172e-01 1.98021924e+00
-4.84811664e-01 -1.35270283e-01 2.98210174e-01 -1.15093672e+00
9.30670142e-01 2.67095119e-01 3.62116605e-01 -5.51008642e-01
-1.25864819e-01 -4.37874570e-02 4.31531630e-02 -6.32119656e-01
4.26682144e-01 -5.36138751e-02 -1.65449932e-01 3.67380321e-01
2.05687046e-01 -7.10778460e-02 4.27334428e-01 5.92608750e-01
1.26273489e+00 1.69058219e-01 3.04881662e-01 -4.42691445e-02
4.85175312e-01 -1.73949957e-01 4.64691013e-01 8.46331358e-01
-1.65615052e-01 3.57198089e-01 7.26941407e-01 -2.29878694e-01
-4.28328753e-01 -6.44961834e-01 5.30594230e-01 1.61720729e+00
1.50856808e-01 -9.31142449e-01 -4.44827110e-01 -1.30220747e+00
-3.82446274e-02 3.63320619e-01 -4.28187370e-01 -2.47767136e-01
-5.19206703e-01 -6.35549545e-01 3.52566689e-01 5.82478046e-01
4.52143729e-01 -1.30270886e+00 3.88892919e-01 2.34372213e-01
-2.32795358e-01 -1.51649714e+00 -7.77105391e-01 1.15321010e-01
-7.53698170e-01 -1.22133768e+00 -5.21694362e-01 -1.03912711e+00
9.09069896e-01 6.75726116e-01 1.55862999e+00 6.93262041e-01
2.07564101e-01 3.95639300e-01 -7.96058416e-01 -2.21833318e-01
-1.38619497e-01 4.63513821e-01 -3.24489325e-01 -1.93264797e-01
4.79925796e-02 -6.43866360e-01 -4.23355818e-01 1.29842967e-01
-9.35950458e-01 4.13520992e-01 7.78847754e-01 5.42690694e-01
4.04316455e-01 1.52259067e-01 5.91565192e-01 -1.46253026e+00
1.10155261e+00 -6.38928831e-01 -3.75594050e-01 5.14425755e-01
-7.27148771e-01 1.97272878e-02 1.02141571e+00 -4.29456770e-01
-9.07330453e-01 -1.15598008e-01 -1.70355767e-01 -3.68911088e-01
-1.10813297e-01 1.05357254e+00 -5.43618321e-01 -9.01030153e-02
3.70799661e-01 1.60243899e-01 -2.78011948e-01 -4.31092918e-01
6.62378252e-01 1.70903996e-01 4.72721189e-01 -1.10382617e+00
1.03544009e+00 -5.93175218e-02 -8.80809352e-02 -3.49745125e-01
-1.06283879e+00 -6.32959008e-01 -4.35230613e-01 -2.44903080e-02
6.26579702e-01 -8.69670570e-01 -6.00277603e-01 5.43194234e-01
-1.37657821e+00 -6.78806305e-01 -6.36273474e-02 9.20836404e-02
-2.45692860e-02 8.19096565e-01 -7.21529365e-01 -4.64769393e-01
-3.59919608e-01 -1.17686462e+00 1.19554174e+00 1.98826805e-01
8.82445872e-02 -1.44267178e+00 -1.06662445e-01 1.36216402e-01
3.38985592e-01 2.82197688e-02 1.11837792e+00 -9.41562712e-01
-5.77230752e-01 -3.20943221e-02 -7.16764092e-01 1.47656262e-01
3.80313784e-01 -1.23044521e-01 -6.50103152e-01 -3.03882986e-01
-1.59575447e-01 -3.54627192e-01 8.12671304e-01 1.03885889e-01
1.28515279e+00 -3.84328395e-01 -6.10442281e-01 6.71545506e-01
1.31978595e+00 -2.37642616e-01 4.23123091e-01 1.67041436e-01
1.38401890e+00 5.81322551e-01 6.24430776e-01 1.05580054e-01
9.24798429e-01 3.47951293e-01 4.56749380e-01 -4.64029104e-01
-1.36314124e-01 -7.86818862e-01 2.23345131e-01 1.02751565e+00
3.59460041e-02 -6.11983597e-01 -8.84712696e-01 4.20519441e-01
-2.07602096e+00 -4.96621251e-01 -2.25400805e-01 1.87249696e+00
8.13293278e-01 5.65519258e-02 1.17022008e-01 -1.61858022e-01
8.51730406e-01 5.42267501e-01 -2.76115716e-01 -2.63989531e-02
3.37662071e-01 -3.39328498e-02 2.43565023e-01 5.55731475e-01
-1.10273230e+00 1.46306586e+00 5.45345592e+00 9.66067314e-01
-1.06282866e+00 -9.78166983e-02 3.42083365e-01 7.35130489e-01
-4.91473287e-01 4.00143355e-01 -7.09408939e-01 4.30053890e-01
5.22112608e-01 -3.48082513e-01 5.79587162e-01 8.74151468e-01
1.16202705e-01 3.61792147e-01 -1.27260184e+00 8.75403464e-01
-1.65832102e-01 -1.03432369e+00 2.31595933e-01 -2.20193737e-03
4.54226494e-01 1.14193380e-01 -3.24848086e-01 9.25035119e-01
6.83917165e-01 -9.24988925e-01 1.97725683e-01 9.71438084e-03
6.64441049e-01 -1.89314723e-01 5.67560136e-01 6.39111042e-01
-1.71316481e+00 1.91117421e-01 -3.50108236e-01 -2.66495794e-01
2.37744004e-01 8.65547180e-01 -1.01964426e+00 1.24034822e+00
2.84316957e-01 9.58214760e-01 -6.31982148e-01 8.53715301e-01
-6.86676443e-01 4.85069871e-01 -2.08902135e-01 2.49103140e-02
2.92493105e-01 -2.40248486e-01 2.36841202e-01 1.31960809e+00
1.94962576e-01 2.48483434e-01 9.56814229e-01 6.48184597e-01
-4.31453913e-01 2.95411438e-01 -8.46483946e-01 -3.37714791e-01
5.22284389e-01 1.61569488e+00 -8.55893135e-01 -3.44602823e-01
-6.70967579e-01 8.38704228e-01 6.87251270e-01 7.02477872e-01
-7.07824111e-01 -4.40086573e-01 1.40065566e-01 1.00431517e-01
3.07727885e-02 -3.54119241e-01 -2.61044521e-02 -1.33251822e+00
8.00208002e-02 -7.65531838e-01 6.73520267e-01 -6.09655082e-01
-1.57474899e+00 5.68390846e-01 1.11330703e-01 -8.44917655e-01
-5.29061668e-02 -6.85720026e-01 -1.06855750e+00 8.71722639e-01
-1.83720112e+00 -1.54575670e+00 -5.80154061e-01 6.60588562e-01
3.90776634e-01 1.01064675e-01 5.35637021e-01 4.15634394e-01
-6.75306737e-01 6.06043279e-01 -7.10415006e-01 3.28738421e-01
8.67855847e-01 -1.42962098e+00 7.17012465e-01 1.12952876e+00
7.35694692e-02 6.56745076e-01 3.64248633e-01 -8.10392857e-01
-1.50342727e+00 -1.33260107e+00 9.36876535e-01 -3.04166913e-01
8.87988865e-01 -4.89492267e-01 -1.17383075e+00 8.60437512e-01
1.28912389e-01 2.38928467e-01 4.27155674e-01 7.74466932e-01
-4.20823008e-01 -1.17905147e-01 -5.52885413e-01 6.19515479e-01
1.51191258e+00 -5.13905287e-01 -3.49301517e-01 8.04409146e-01
1.28285742e+00 -5.65400183e-01 -7.21401513e-01 6.39295042e-01
-3.11562568e-01 -4.48872417e-01 9.09240544e-01 -7.10428119e-01
3.32912743e-01 -2.61204243e-01 3.20669502e-01 -1.57230270e+00
-6.70771360e-01 -9.11255062e-01 -1.32179961e-01 1.72241879e+00
3.43783885e-01 -5.95853508e-01 7.24313915e-01 3.43035400e-01
-6.41050339e-01 -8.39854658e-01 -1.94587454e-01 -5.72543800e-01
-2.36696213e-01 -5.35868526e-01 7.64135897e-01 1.16398406e+00
3.62830162e-02 9.90304708e-01 -3.48920316e-01 1.98452368e-01
2.34097078e-01 2.06019267e-01 1.03851867e+00 -1.31579340e+00
-3.88897806e-01 -4.88668293e-01 2.58374989e-01 -1.43428600e+00
4.63307947e-01 -1.41110444e+00 1.08674780e-01 -1.98012197e+00
2.39401326e-01 -7.03469455e-01 -4.53377992e-01 1.04650807e+00
-8.89238238e-01 -5.19973457e-01 1.31193668e-01 1.42845765e-01
-9.00119305e-01 5.28745294e-01 1.87965250e+00 -1.73472643e-01
-1.79677114e-01 7.42196739e-02 -1.19218242e+00 5.05775630e-01
5.06359518e-01 -4.97775257e-01 -8.73553097e-01 -7.79447258e-01
4.18504149e-01 1.73300877e-02 1.83406249e-01 -3.65552098e-01
4.76644367e-01 -3.64331126e-01 -1.70812637e-01 -2.39351079e-01
-2.53531486e-01 -6.54701233e-01 -3.17474186e-01 8.61895382e-02
6.56023808e-03 1.00447774e-01 6.11045770e-02 7.27498353e-01
-3.08526397e-01 -4.23472151e-02 2.91727751e-01 -3.48451704e-01
-9.31524754e-01 9.78799343e-01 2.29979038e-01 4.28479642e-01
7.25612521e-01 2.85562593e-02 -5.71766913e-01 -2.68913209e-01
-5.37796140e-01 8.92628908e-01 4.20447767e-01 5.45050144e-01
5.02582848e-01 -1.16237295e+00 -7.22112894e-01 8.76730233e-02
2.89377481e-01 6.73950553e-01 3.54085386e-01 7.98322380e-01
-1.66910112e-01 9.22742635e-02 2.27698833e-01 -4.95497227e-01
-8.66996050e-01 9.16921079e-01 -1.14273955e-03 -9.32204306e-01
-7.67017484e-01 8.86007726e-01 6.08180940e-01 -6.90462828e-01
2.75829911e-01 -5.10856569e-01 -2.13122457e-01 -2.03297600e-01
9.18014422e-02 -2.37836465e-01 4.75907139e-02 -9.18241963e-03
-3.31706762e-01 3.58185947e-01 -3.02522272e-01 4.92104590e-01
1.10856199e+00 -4.33971249e-02 -3.40619296e-01 3.91439199e-02
9.23495650e-01 1.28346503e-01 -1.08251321e+00 -4.53878015e-01
1.96953341e-01 -2.27420837e-01 -3.37511450e-01 -6.50966763e-01
-1.13711345e+00 8.24641049e-01 -3.65635395e-01 5.47205746e-01
1.17400038e+00 1.67467579e-01 8.69277894e-01 3.18858951e-01
1.76949188e-01 -7.36750305e-01 2.12836862e-01 7.35373259e-01
7.84025967e-01 -1.29430842e+00 -9.47770774e-02 -1.12409222e+00
-6.20236576e-01 8.50151837e-01 9.59383488e-01 4.61817347e-02
6.56274378e-01 1.83047295e-01 -3.05603482e-02 -4.04413611e-01
-8.07588100e-01 -2.90457457e-01 2.59042025e-01 8.07913244e-01
6.49073422e-01 2.96555161e-02 -2.58736730e-01 9.11772609e-01
7.22741336e-02 6.23974651e-02 1.32146776e-01 7.03649759e-01
-1.05992638e-01 -1.50367391e+00 1.76054522e-01 2.70629019e-01
-1.85715988e-01 -4.09479439e-01 -4.57898468e-01 8.94222856e-01
-1.97560340e-02 8.76548529e-01 -4.05666322e-01 -4.68029648e-01
4.42023128e-01 -9.07804221e-02 4.05410439e-01 -1.12945521e+00
-5.82388878e-01 -4.31380176e-04 2.41173238e-01 -4.41062868e-01
-1.34136721e-01 2.44887639e-02 -1.26991916e+00 -2.58811325e-01
-3.95056099e-01 4.11112279e-01 2.91444473e-02 1.17751503e+00
4.32540029e-01 9.16870415e-01 4.76710051e-01 -6.80462122e-01
-4.61514562e-01 -1.03916740e+00 -4.36545581e-01 5.03964245e-01
-1.59372151e-01 -5.07082701e-01 -1.09224372e-01 -1.33355275e-01] | [8.728549003601074, 7.515232563018799] |
8528bbf5-a3e0-425c-999c-1a9c238fd4c1 | 360-circ-high-resolution-depth-estimation-via | 2304.07967 | null | https://arxiv.org/abs/2304.07967v1 | https://arxiv.org/pdf/2304.07967v1.pdf | 360$^\circ$ High-Resolution Depth Estimation via Uncertainty-aware Structural Knowledge Transfer | Recently, omnidirectional images (ODIs) have become increasingly popular; however, their angular resolution tends to be lower than that of perspective images.This leads to degraded structural details such as edges, causing difficulty in learning 3D scene understanding tasks, especially monocular depth estimation. Existing methods typically leverage high-resolution (HR) ODI as the input, so as to recover the structural details via fully-supervised learning. However, the HR depth ground truth (GT) maps may be arduous or expensive to be collected due to resource-constrained devices in practice. Therefore, in this paper, we explore for the first time to estimate the HR omnidirectional depth directly from a low-resolution (LR) ODI, when no HR depth GT map is available. Our key idea is to transfer the scene structural knowledge from the readily available HR image modality and the corresponding LR depth maps to achieve the goal of HR depth estimation without extra inference cost. Specifically, we introduce ODI super-resolution (SR) as an auxiliary task and train both tasks collaboratively in a weakly supervised manner to boost the performance of HR depth estimation. The ODI SR task takes an LR ODI as the input to predict an HR image, enabling us to extract the scene structural knowledge via uncertainty estimation. Buttressed by this, a scene structural knowledge transfer (SSKT) module is proposed with two key components. First, we employ a cylindrical implicit interpolation function (CIIF) to learn cylindrical neural interpolation weights for feature up-sampling and share the parameters of CIIFs between the two tasks. Then, we propose a feature distillation (FD) loss that provides extra structural regularization to help the HR depth estimation task learn more scene structural knowledge. | ['Lin Wang', 'Hao Ai', 'Zidong Cao'] | 2023-04-17 | null | null | null | null | ['monocular-depth-estimation'] | ['computer-vision'] | [ 3.61466736e-01 1.75011605e-01 -2.05073729e-01 -5.05190969e-01
-8.58034492e-01 -2.72973031e-01 4.00227278e-01 -4.40890342e-01
-2.80472100e-01 8.11001837e-01 2.72179246e-01 -7.60355890e-02
-6.60663918e-02 -9.51174140e-01 -8.33868444e-01 -7.77333856e-01
3.70595843e-01 7.71541670e-02 4.92664315e-02 9.91993546e-02
1.56669796e-01 4.01447386e-01 -1.57138073e+00 8.36391225e-02
1.22468579e+00 1.07233858e+00 7.00748742e-01 4.05268967e-01
5.83247095e-02 1.06589580e+00 -1.18847853e-02 1.11408532e-02
2.96243787e-01 -1.72586367e-02 -7.13514507e-01 1.64326951e-01
4.83946443e-01 -8.19813192e-01 -7.88584292e-01 9.50212777e-01
3.90523493e-01 2.33693421e-01 4.44159299e-01 -7.97661960e-01
-3.76979887e-01 8.23058337e-02 -8.14373136e-01 -1.72955561e-02
4.01607394e-01 2.95495335e-02 7.95810759e-01 -1.01030755e+00
6.27285302e-01 1.11117792e+00 3.62926334e-01 3.77429962e-01
-1.07029974e+00 -5.95547557e-01 3.69370520e-01 2.21479833e-01
-1.49331200e+00 -5.05643189e-01 1.03295481e+00 -2.61069089e-01
6.10178053e-01 -1.51985645e-01 4.74630356e-01 9.35703397e-01
-6.28513619e-02 9.01554286e-01 1.14086044e+00 -1.51460215e-01
1.55326277e-01 2.31937751e-01 -2.92659193e-01 6.77752674e-01
8.23041350e-02 2.65620679e-01 -6.51833892e-01 4.75168526e-01
1.29494190e+00 1.66743323e-01 -7.21789002e-01 -4.16619062e-01
-8.92631054e-01 5.46533823e-01 8.12179446e-01 -8.02536868e-03
-2.62581825e-01 -9.36792269e-02 3.93949375e-02 2.34604433e-01
7.37712681e-01 2.58805186e-01 -5.04134953e-01 1.04909226e-01
-4.44730788e-01 -9.40473527e-02 2.87423819e-01 8.82400274e-01
1.18780923e+00 -3.08240429e-02 1.94893405e-01 9.34209287e-01
3.38640243e-01 4.24178451e-01 1.22633837e-01 -1.04164910e+00
7.73433149e-01 4.51172650e-01 2.47833118e-01 -7.97571421e-01
-2.95887083e-01 -5.18116951e-01 -8.97988200e-01 2.37489983e-01
3.25163394e-01 -2.94351997e-03 -8.63214254e-01 1.75706029e+00
5.78728199e-01 3.01297158e-01 1.46307170e-01 1.23095787e+00
8.57503533e-01 6.31648540e-01 -3.26685935e-01 1.00065181e-02
1.01065278e+00 -8.11407328e-01 -3.15539211e-01 -5.23258448e-01
4.58630919e-01 -5.58337033e-01 1.16070914e+00 4.16985422e-01
-7.85822749e-01 -6.04155362e-01 -1.05856454e+00 -5.07419944e-01
6.49051815e-02 2.71384746e-01 6.75189435e-01 1.96005136e-01
-7.16365159e-01 3.22596341e-01 -7.69646168e-01 7.47161284e-02
4.26908404e-01 1.00744121e-01 -4.89749253e-01 -8.41724455e-01
-1.09821212e+00 6.72043264e-01 3.49793077e-01 3.50099504e-01
-8.77803087e-01 -8.32134962e-01 -1.31187308e+00 -2.10182846e-01
6.36541605e-01 -8.57157946e-01 8.57914031e-01 -6.98951542e-01
-1.74245083e+00 7.31688857e-01 -2.04563916e-01 2.01676656e-02
5.72451353e-01 -3.09090853e-01 -7.15867132e-02 2.74401963e-01
1.81838840e-01 6.80534661e-01 9.15788651e-01 -1.54898059e+00
-7.77745485e-01 -6.55085862e-01 5.23091733e-01 7.30677545e-01
1.19609512e-01 -6.66236401e-01 -5.64248443e-01 -3.05040956e-01
6.39484107e-01 -6.00046933e-01 -9.26829353e-02 1.66901544e-01
-4.55669165e-01 1.61562786e-01 5.81206024e-01 -6.83644474e-01
5.97949445e-01 -2.19028354e+00 3.08247685e-01 -1.15834810e-01
2.82665908e-01 -1.99500531e-01 -7.64132440e-02 -2.18031421e-01
1.17834970e-01 -4.42983121e-01 -2.14667290e-01 -6.32206321e-01
-4.33011919e-01 3.69042248e-01 -2.45002195e-01 4.08839285e-01
1.29794583e-01 7.59878576e-01 -1.05479443e+00 -2.65637368e-01
6.55813932e-01 7.31656730e-01 -5.90993345e-01 5.62342584e-01
-1.71239316e-01 1.19100082e+00 -6.09134614e-01 5.16517997e-01
1.06893122e+00 -2.07253754e-01 -6.52293712e-02 -6.18547261e-01
-4.10725325e-01 3.66764188e-01 -1.13605833e+00 2.09246731e+00
-1.05282974e+00 3.50493431e-01 1.51067570e-01 -8.83806050e-01
9.38244045e-01 7.59728029e-02 3.54202062e-01 -9.47586715e-01
-2.01974958e-01 2.71593213e-01 -4.75625038e-01 -4.21666533e-01
3.41432661e-01 -2.31407210e-01 1.04594506e-01 3.96732502e-02
-1.41648337e-01 -5.46218157e-01 -5.31536043e-01 2.51516551e-02
6.88526869e-01 4.58101362e-01 2.31578246e-01 1.66953191e-01
6.86919928e-01 -4.13482368e-01 7.95367062e-01 3.92810464e-01
2.83341080e-01 8.90976369e-01 9.40898433e-02 -3.81613046e-01
-1.07470894e+00 -1.15965831e+00 -2.55248159e-01 5.03690779e-01
5.49656451e-01 1.04445666e-01 -3.62350762e-01 -4.79715466e-01
-1.16254188e-01 4.40133333e-01 -4.74868387e-01 -1.86761543e-01
-4.29528058e-01 -4.92643327e-01 -3.80483852e-03 5.90759635e-01
9.84673440e-01 -6.90078080e-01 -4.98618543e-01 1.68268476e-02
-4.36620772e-01 -1.50572574e+00 -4.21471179e-01 8.31421688e-02
-8.69779766e-01 -8.61140013e-01 -6.92884564e-01 -3.82048458e-01
6.37356579e-01 7.14058280e-01 8.94039810e-01 -3.41391623e-01
-2.49566451e-01 1.29781559e-01 -9.17173326e-02 -3.72508615e-02
2.47104332e-01 4.66703139e-02 -1.18049338e-01 2.23800048e-01
-2.59213876e-02 -8.14223945e-01 -8.78627837e-01 3.26132715e-01
-8.65739942e-01 6.06882513e-01 7.01292038e-01 8.98720622e-01
7.43599355e-01 3.04336309e-01 3.16430390e-01 -7.96400368e-01
-2.37522945e-01 -4.28707302e-01 -8.29161525e-01 -1.48280218e-01
-2.48205036e-01 1.06068522e-01 5.89469492e-01 -5.39532192e-02
-1.68782604e+00 8.88723060e-02 -2.62787163e-01 -6.61554098e-01
-1.36840150e-01 4.99682754e-01 -6.61674976e-01 -1.39961585e-01
4.02933985e-01 3.16179335e-01 -2.33925536e-01 -5.40946424e-01
2.72475958e-01 6.16829872e-01 5.23087204e-01 -4.86604124e-01
8.64095926e-01 8.74625564e-01 9.23992768e-02 -9.81654465e-01
-1.55171514e+00 -5.86510837e-01 -6.33974016e-01 3.35821882e-02
7.21530437e-01 -1.58829463e+00 -5.01025081e-01 5.73690772e-01
-8.96016002e-01 -5.26175976e-01 -1.92339748e-01 6.95027649e-01
-6.18336260e-01 4.69525188e-01 -5.37064672e-01 -8.47629845e-01
-4.30259705e-02 -9.51782167e-01 1.38316083e+00 3.63785982e-01
4.91781294e-01 -8.59026074e-01 -1.69976220e-01 7.15319157e-01
9.92533937e-02 1.35075882e-01 7.04834759e-01 5.13291955e-01
-1.16730785e+00 1.98158413e-01 -8.69454205e-01 3.80929321e-01
3.18659425e-01 -5.66787779e-01 -1.35011756e+00 -2.10716054e-01
2.95988053e-01 -4.42110330e-01 8.74462724e-01 5.27600527e-01
1.25832391e+00 -3.11321262e-02 -1.35123879e-01 1.20963645e+00
1.61972809e+00 -3.76477279e-02 7.21270740e-01 3.07730615e-01
1.22125459e+00 7.77865708e-01 9.48592126e-01 5.49976230e-01
8.05198014e-01 8.46987128e-01 4.57901955e-01 -1.58186376e-01
-2.57755041e-01 -6.58468366e-01 1.66072667e-01 5.13884425e-01
-1.91874638e-01 1.58914804e-01 -5.79829156e-01 3.41406673e-01
-1.71698368e+00 -5.88747680e-01 2.83474505e-01 2.46331596e+00
8.18344176e-01 8.18479210e-02 -3.80385816e-01 -3.95429134e-02
3.33205521e-01 2.62905270e-01 -9.62882042e-01 3.02863955e-01
-6.27765208e-02 -1.71267197e-01 5.01929045e-01 8.61517072e-01
-8.44258070e-01 9.06536222e-01 4.10842896e+00 7.07595766e-01
-1.27301824e+00 3.60697322e-02 5.94050646e-01 -1.11798802e-02
-6.49467349e-01 7.75376288e-03 -8.22734118e-01 3.32552046e-01
3.50665152e-01 1.91774800e-01 5.93207181e-01 7.74746895e-01
3.20862263e-01 -3.86397958e-01 -1.05095899e+00 1.43829501e+00
-1.25443295e-01 -9.11011517e-01 -1.58053532e-01 1.10679716e-01
8.59003186e-01 1.98673964e-01 2.15421282e-02 2.01923445e-01
8.41064528e-02 -9.72405791e-01 5.09488344e-01 6.44439697e-01
1.18263972e+00 -7.93350160e-01 6.70375645e-01 5.49014628e-01
-1.34112895e+00 -2.04402655e-01 -6.41295791e-01 -8.42893645e-02
3.04562777e-01 1.01733470e+00 -3.86973917e-01 1.01689601e+00
6.96137249e-01 1.06603861e+00 -1.64993927e-01 8.62999678e-01
-5.22730768e-01 -3.57341506e-02 -3.97764593e-01 8.10848534e-01
1.63605418e-02 -4.36229944e-01 4.60650682e-01 4.98371810e-01
3.16274077e-01 3.73907566e-01 2.27181092e-01 9.32946563e-01
-4.56708409e-02 -2.16192082e-01 -7.76111484e-01 4.48386163e-01
3.12408090e-01 1.19336188e+00 -2.52466887e-01 -7.69658887e-04
-6.24844432e-01 1.18870771e+00 4.68804032e-01 5.21132231e-01
-5.40711522e-01 -1.68029204e-01 7.63880610e-01 1.48255691e-01
1.11669183e-01 -3.61142993e-01 -3.71319979e-01 -1.70249784e+00
1.58190623e-01 -4.69394118e-01 1.74834684e-01 -1.02159905e+00
-1.14398766e+00 4.41928685e-01 -1.53516427e-01 -1.36813772e+00
-1.91507086e-01 -4.39195484e-01 -2.67536372e-01 1.14187002e+00
-2.31104374e+00 -1.21492243e+00 -1.00330317e+00 5.70452034e-01
5.87210238e-01 3.82860929e-01 2.91746438e-01 3.89110625e-01
-5.12311518e-01 3.88676941e-01 -3.42399417e-03 -1.18301086e-01
6.67685628e-01 -1.10042870e+00 7.72282109e-02 6.42828882e-01
-1.15243733e-01 3.36732209e-01 4.15982574e-01 -4.15336102e-01
-1.34692872e+00 -1.14472127e+00 4.73795146e-01 -3.58167320e-01
1.83096796e-01 -4.02124107e-01 -9.98749375e-01 5.83240151e-01
-6.26776516e-01 4.15122360e-01 2.30023209e-02 -5.51433936e-02
-3.67524505e-01 -3.55289072e-01 -1.10715914e+00 5.22330165e-01
1.37788427e+00 -1.01015007e+00 -1.59637019e-01 1.08332396e-01
8.69032383e-01 -6.76284194e-01 -9.30077434e-01 7.05562949e-01
4.61515069e-01 -1.12977386e+00 1.28363574e+00 1.41765192e-01
6.21188045e-01 -5.56144834e-01 -3.13091010e-01 -1.18645656e+00
6.08854145e-02 -1.70123935e-01 -2.16667488e-01 1.01834309e+00
1.48636565e-01 -8.02462339e-01 9.78628874e-01 5.74487567e-01
-3.11602563e-01 -7.02045143e-01 -8.28244269e-01 -6.19019449e-01
-2.98969358e-01 -3.38902265e-01 3.89221936e-01 8.54423642e-01
-3.85640264e-01 5.94626784e-01 -4.77872998e-01 5.38398266e-01
9.20519352e-01 3.54007065e-01 8.89270186e-01 -1.15964937e+00
-5.15954077e-01 1.38512790e-01 -1.91059127e-01 -1.76023376e+00
1.11453936e-01 -5.11368930e-01 1.71435997e-01 -1.43987107e+00
1.49104729e-01 -8.38970900e-01 -8.05396289e-02 1.08336426e-01
-2.38141730e-01 2.12089643e-01 -1.52980298e-01 3.04401994e-01
-2.98784584e-01 9.83879924e-01 1.60255587e+00 1.13715537e-01
-3.00294459e-01 1.21528342e-01 -5.91457009e-01 8.70335519e-01
4.01982963e-01 -1.42570823e-01 -9.23393369e-01 -6.25347316e-01
2.60664314e-01 4.51930225e-01 4.46906060e-01 -8.41818929e-01
9.99851972e-02 -1.51529729e-01 5.07797837e-01 -5.90279400e-01
6.92812741e-01 -7.43038476e-01 -7.03651109e-04 -1.99409071e-02
5.40544093e-02 -7.31109083e-01 -7.80669153e-02 7.27473497e-01
-3.83514673e-01 -8.98722634e-02 8.68388891e-01 -1.64414868e-01
-8.75639558e-01 7.84552515e-01 3.06996942e-01 -7.32985139e-02
5.68953574e-01 -3.42923075e-01 -9.97757167e-02 -4.77319747e-01
-4.42908347e-01 1.86165139e-01 8.08797956e-01 2.90889591e-01
9.00830090e-01 -1.06310666e+00 -5.15341461e-01 4.25594747e-01
3.96906197e-01 9.74876583e-01 7.25441933e-01 8.41104686e-01
-2.46721223e-01 4.59086031e-01 -9.00251046e-02 -6.99857354e-01
-8.22688043e-01 5.21552861e-01 4.11542475e-01 -2.28048235e-01
-1.05643153e+00 8.08110714e-01 9.50822115e-01 -6.57973468e-01
2.09206082e-02 -3.23111862e-01 -2.49578848e-01 -2.58384854e-01
5.28735995e-01 1.34046599e-01 -5.06309234e-02 -3.52756500e-01
-4.02562544e-02 9.41001236e-01 -1.64880097e-01 -4.37600687e-02
1.26431334e+00 -7.73422956e-01 1.20833173e-01 3.78592759e-01
1.32384086e+00 -1.04960509e-01 -2.04438543e+00 -6.79615557e-01
-3.05462211e-01 -9.84404802e-01 4.23705965e-01 -4.05722946e-01
-1.04409683e+00 9.57563698e-01 4.04393494e-01 -5.76384127e-01
1.25201952e+00 1.19690426e-01 7.79588997e-01 2.86976606e-01
8.08035493e-01 -8.01928341e-01 1.36924133e-01 4.74179238e-01
6.89250946e-01 -1.62087202e+00 -6.81872293e-02 -8.73885810e-01
-5.48764586e-01 9.97554302e-01 7.86469817e-01 1.06383055e-01
4.82539058e-01 -3.26425768e-04 -1.66452155e-01 -6.55100942e-02
-4.35983479e-01 -2.43918613e-01 3.61677468e-01 6.38780475e-01
1.31194934e-01 -1.62935466e-01 2.99445629e-01 2.43132979e-01
6.59257397e-02 1.02364436e-01 3.68272066e-01 5.26627541e-01
-3.94916356e-01 -8.00486505e-01 -9.18613933e-03 2.18365341e-01
-3.94733399e-02 -1.01758048e-01 3.36147606e-01 4.29060191e-01
8.74597952e-02 6.91038907e-01 4.27059084e-02 -2.36698449e-01
3.14944476e-01 -5.48807561e-01 7.37918019e-01 -7.54730463e-01
4.35986221e-01 -1.92788288e-01 1.84382591e-02 -7.46479213e-01
-3.43221247e-01 -4.79263335e-01 -1.13858390e+00 -1.28055975e-01
-1.26084507e-01 -1.46833211e-01 7.03190923e-01 1.07368767e+00
1.34755433e-01 3.70996773e-01 8.77667546e-01 -1.14628220e+00
-6.70115426e-02 -7.30220497e-01 -7.22751796e-01 1.97627589e-01
6.72197044e-01 -8.90311837e-01 -4.72824067e-01 -3.08935523e-01] | [8.939605712890625, -2.7149229049682617] |
dee579e1-eeca-4fea-8869-09be40dce852 | dialogue-strategy-adaptation-to-new-action | 2204.07082 | null | https://arxiv.org/abs/2204.07082v1 | https://arxiv.org/pdf/2204.07082v1.pdf | Dialogue Strategy Adaptation to New Action Sets Using Multi-dimensional Modelling | A major bottleneck for building statistical spoken dialogue systems for new domains and applications is the need for large amounts of training data. To address this problem, we adopt the multi-dimensional approach to dialogue management and evaluate its potential for transfer learning. Specifically, we exploit pre-trained task-independent policies to speed up training for an extended task-specific action set, in which the single summary action for requesting a slot is replaced by multiple slot-specific request actions. Policy optimisation and evaluation experiments using an agenda-based user simulator show that with limited training data, much better performance levels can be achieved when using the proposed multi-dimensional adaptation method. We confirm this improvement in a crowd-sourced human user evaluation of our spoken dialogue system, comparing partially trained policies. The multi-dimensional system (with adaptation on limited training data in the target scenario) outperforms the one-dimensional baseline (without adaptation on the same amount of training data) by 7% perceived success rate. | ['Rama Doddipatla', 'Svetlana Stoyanchev', 'Norbert Braunschweiler', 'Simon Keizer'] | 2022-04-14 | null | null | null | null | ['dialogue-management', 'spoken-dialogue-systems'] | ['natural-language-processing', 'speech'] | [ 1.22144178e-01 6.58126473e-01 1.16320066e-01 -4.77854103e-01
-1.00123990e+00 -6.79281175e-01 9.88577306e-01 -2.27521695e-02
-1.05565012e+00 1.21510005e+00 7.33892083e-01 -4.02327180e-01
2.42680937e-01 -1.81089401e-01 -7.23524690e-02 -5.73877156e-01
-6.21541440e-02 1.07814658e+00 4.04121637e-01 -7.41547823e-01
2.73556799e-01 1.38230711e-01 -1.35311186e+00 3.92626017e-01
6.49909377e-01 4.09164578e-01 5.30510902e-01 1.21347690e+00
-3.54351878e-01 5.49650908e-01 -1.18597472e+00 2.34539166e-01
1.53371394e-02 -6.28552675e-01 -1.33030415e+00 3.53079230e-01
7.25910291e-02 -7.06473529e-01 -5.80239184e-02 3.12933117e-01
1.09654200e+00 6.08421504e-01 3.49172860e-01 -8.49527597e-01
1.72789469e-01 2.76983708e-01 1.05959803e-01 1.77451462e-01
7.98287988e-01 2.98565269e-01 7.64582157e-01 -5.12302577e-01
4.21627313e-01 1.61750460e+00 1.64240941e-01 1.10934925e+00
-1.30617535e+00 1.01038821e-01 6.90073594e-02 -2.90332526e-01
-7.44027078e-01 -8.73829186e-01 2.21366212e-01 -8.40548947e-02
1.33062708e+00 3.76755953e-01 2.34360889e-01 1.17482924e+00
-2.99020439e-01 9.68001664e-01 1.29923093e+00 -8.33889306e-01
5.76951146e-01 6.74333692e-01 -1.01088144e-01 4.38838452e-01
-5.63067734e-01 -1.36274070e-01 -4.28533405e-01 -4.85407442e-01
6.58446014e-01 -8.40608656e-01 -2.15550616e-01 -2.76812673e-01
-1.14336526e+00 9.77832437e-01 -3.70412290e-01 3.89222652e-01
-4.45170492e-01 -6.28321946e-01 6.99246526e-01 6.73310876e-01
5.43931007e-01 8.53768587e-01 -9.36158717e-01 -9.10000980e-01
-4.08358812e-01 5.65208256e-01 1.60341811e+00 7.57395506e-01
3.77770215e-01 1.60460114e-01 -5.71355939e-01 1.36437798e+00
8.36322829e-02 3.98977220e-01 7.06259966e-01 -1.29385030e+00
6.44727647e-01 2.85678416e-01 5.66206515e-01 -1.60210446e-01
-7.30849504e-01 2.21584976e-01 -4.63878751e-01 9.75543261e-02
7.03511834e-01 -9.85957146e-01 -5.25469780e-01 1.57290840e+00
5.66528618e-01 -2.65241474e-01 6.18669987e-01 7.52638698e-01
4.72365886e-01 7.63001680e-01 1.71985358e-01 -5.11327028e-01
1.25449395e+00 -9.35152113e-01 -7.99265265e-01 -3.02559257e-01
1.13649666e+00 -6.33205652e-01 1.30034471e+00 3.45786512e-01
-1.17635691e+00 -2.91300863e-01 -6.75746143e-01 3.64343315e-01
-3.38702346e-03 -1.14976607e-01 2.62330204e-01 6.80310726e-01
-1.21342039e+00 2.41675079e-01 -4.48539823e-01 -7.90987015e-01
-4.00114715e-01 4.28155839e-01 -2.58030176e-01 1.58936471e-01
-1.40439034e+00 1.20780754e+00 4.27665710e-01 -3.93702805e-01
-6.73526585e-01 -2.17845604e-01 -7.74489880e-01 2.65898313e-02
5.09178340e-01 -3.64936709e-01 2.12990952e+00 -6.69407785e-01
-2.43160772e+00 5.92432380e-01 -3.35562900e-02 -4.06541109e-01
5.45775115e-01 -4.09821182e-01 -1.10474810e-01 7.44652227e-02
-1.19244732e-01 5.99877238e-01 4.51675951e-01 -1.28277612e+00
-9.79577124e-01 -1.45344045e-02 3.28906178e-01 9.39394772e-01
-4.90109384e-01 1.55863777e-01 -2.31190100e-01 -4.36779782e-02
-5.36051929e-01 -1.06827605e+00 -2.80642241e-01 -7.22370923e-01
-9.29060951e-02 -4.59463328e-01 6.92175746e-01 -5.66769063e-01
1.28991139e+00 -1.78715813e+00 3.88375551e-01 -7.66061693e-02
-2.71584004e-01 8.47385287e-01 -3.31869513e-01 7.00863242e-01
3.70347023e-01 -3.98882329e-02 -1.41618967e-01 -5.67339957e-01
8.44327584e-02 6.03703558e-01 1.25042588e-01 -1.54052109e-01
1.26945168e-01 3.92778456e-01 -9.09733713e-01 -4.79754418e-01
2.56203234e-01 -6.67948648e-03 -6.66590750e-01 8.38726699e-01
-5.76396883e-01 8.72322619e-01 -5.35544038e-01 -6.41291263e-04
1.81413144e-01 6.15279004e-02 4.45204228e-01 5.10202169e-01
-2.52633601e-01 6.46419764e-01 -1.17493558e+00 1.71787536e+00
-7.68566787e-01 4.21527267e-01 4.99504238e-01 -8.17700863e-01
8.22421968e-01 7.55373716e-01 3.22177291e-01 -8.68648708e-01
-2.14252904e-01 -1.11572661e-01 2.68852204e-01 -7.40127623e-01
7.13117480e-01 3.41981873e-02 -3.35617304e-01 9.10038650e-01
2.18990281e-01 -3.06813061e-01 2.85821911e-02 3.07532758e-01
1.13320386e+00 -1.23635322e-01 3.04748774e-01 -1.96604803e-01
6.84862733e-01 -4.45064418e-02 1.99917838e-01 9.45061982e-01
-5.14350832e-01 1.35129929e-01 7.31188834e-01 -2.43202895e-01
-1.15842164e+00 -3.93171966e-01 1.36240602e-01 1.96575093e+00
-5.21814644e-01 -1.28611073e-01 -1.03455591e+00 -8.05245936e-01
-1.88953206e-01 1.20771968e+00 -2.70656586e-01 3.74342538e-02
-8.24761629e-01 -6.54243410e-01 6.61906421e-01 5.18717729e-02
5.74982166e-01 -1.35969496e+00 -6.72752917e-01 5.40387213e-01
-3.05982143e-01 -1.14722502e+00 -3.56392771e-01 2.20226794e-01
-5.22711337e-01 -5.70454180e-01 -9.56978381e-01 -5.82282543e-01
1.76736712e-01 5.92342168e-02 1.12213981e+00 -3.46054435e-02
2.09818870e-01 8.38102818e-01 -4.21974957e-01 -3.69895965e-01
-1.16877985e+00 4.15763736e-01 3.59800071e-01 -3.99760306e-01
1.55592442e-01 -4.11123857e-02 -3.67402107e-01 5.82924902e-01
-5.57606161e-01 1.05323397e-01 5.02161145e-01 1.21334732e+00
-4.52814788e-01 -6.35317922e-01 1.03349257e+00 -7.70925701e-01
1.50237906e+00 -3.00034046e-01 -3.38959992e-01 2.97787666e-01
-4.91582334e-01 3.82236600e-01 2.61190325e-01 -4.23776835e-01
-1.44036090e+00 -5.95021695e-02 -2.62941658e-01 3.47680360e-01
-6.11928344e-01 1.91732764e-01 -2.17417292e-02 3.42557937e-01
8.58690023e-01 2.68555701e-01 5.87403297e-01 -4.82286036e-01
3.71505290e-01 1.21347773e+00 4.43871282e-02 -8.82629693e-01
1.12036327e-02 -2.64605492e-01 -5.72397113e-01 -1.25463510e+00
-5.11379063e-01 -7.29624033e-01 -7.73357749e-01 -1.73623711e-01
7.42649496e-01 -6.99464142e-01 -7.71219969e-01 3.83966506e-01
-1.22344673e+00 -1.05363071e+00 -7.92314187e-02 5.10099351e-01
-8.28465700e-01 3.14008117e-01 -5.28267980e-01 -1.32906866e+00
-2.13573903e-01 -9.81555521e-01 1.00273681e+00 1.82631880e-01
-5.45084000e-01 -1.26740670e+00 4.41093206e-01 2.09758982e-01
5.47831774e-01 -4.03685272e-01 9.03385937e-01 -1.52848864e+00
2.45767787e-01 -3.68107408e-02 1.85389996e-01 4.68288332e-01
3.36878419e-01 -5.33299863e-01 -1.02732515e+00 -5.15863180e-01
1.43032102e-02 -1.00523651e+00 1.06012352e-01 -4.08001430e-02
4.04350191e-01 -5.04250288e-01 6.96119806e-03 -4.23852980e-01
5.22796571e-01 2.40549967e-01 2.40922794e-01 5.23581445e-01
2.66004413e-01 9.30344701e-01 9.51228023e-01 8.77435565e-01
6.21435285e-01 9.79943395e-01 -9.90189984e-02 -6.98743239e-02
3.01274687e-01 7.04637542e-02 4.27719891e-01 7.72703409e-01
-1.05582434e-03 -4.63969052e-01 -9.13907886e-01 4.27920103e-01
-1.95300746e+00 -7.41274476e-01 4.44158822e-01 2.45716047e+00
1.16793370e+00 1.26430899e-01 7.06700146e-01 -8.74689221e-02
5.69675267e-01 8.11166391e-02 -4.21510905e-01 -8.54366004e-01
1.62770689e-01 -1.63661819e-02 8.36997479e-02 9.92230773e-01
-8.86867106e-01 1.16448450e+00 6.94388771e+00 3.23817968e-01
-8.93058240e-01 -6.53159618e-02 4.40325379e-01 -2.58692331e-03
1.22170888e-01 -3.09482723e-01 -9.12709475e-01 1.79241195e-01
1.57197523e+00 -2.04768747e-01 4.05110776e-01 6.30358875e-01
6.71498954e-01 -5.09806633e-01 -1.08435953e+00 3.92216086e-01
-7.30860233e-02 -9.37856317e-01 -2.39323288e-01 1.61045864e-01
4.07578647e-01 -1.53618455e-01 -2.34079286e-01 8.16593289e-01
6.85117483e-01 -6.32342637e-01 -1.38006270e-01 2.25231186e-01
5.41033268e-01 -6.04266465e-01 6.69550478e-01 9.91769850e-01
-3.17853719e-01 4.47991118e-02 -2.51141816e-01 -2.30022311e-01
2.73993462e-01 -4.59478408e-01 -1.87474358e+00 1.57037169e-01
2.37573698e-01 -2.31927112e-01 -5.35936467e-02 6.55323982e-01
2.67598331e-02 8.21550190e-01 -3.82139146e-01 -3.40308696e-01
4.62818116e-01 6.61403015e-02 6.31268084e-01 1.38131464e+00
-6.27349615e-02 4.37450290e-01 4.84826148e-01 -1.27534149e-03
2.11853579e-01 1.99685618e-01 -6.68874502e-01 1.45091966e-01
5.28782129e-01 1.10005510e+00 -6.89653158e-02 -5.10306776e-01
-3.88748765e-01 1.14493990e+00 3.12471300e-01 3.86377752e-01
-3.33196968e-01 -1.22706480e-01 5.47926784e-01 -2.79787928e-01
5.24128303e-02 -3.13960701e-01 1.98946223e-01 -6.75319076e-01
-4.34219122e-01 -1.34759784e+00 2.75554031e-01 -3.76999140e-01
-7.71752834e-01 6.02107584e-01 2.67964482e-01 -7.89027214e-01
-1.29238248e+00 -3.63329977e-01 -4.95226204e-01 1.14896965e+00
-1.14462924e+00 -5.50291300e-01 8.63345414e-02 5.08909762e-01
1.14874852e+00 -5.75833499e-01 1.64518476e+00 -2.43368298e-02
-3.97642314e-01 6.79792881e-01 1.40344307e-01 -8.44694898e-02
1.21900690e+00 -1.52216864e+00 3.26122135e-01 -3.16591188e-02
-3.63622874e-01 2.61121362e-01 9.64456797e-01 -4.87470031e-01
-1.13561118e+00 -5.21912932e-01 7.30562568e-01 -5.19872308e-01
4.63588059e-01 -4.68684316e-01 -1.26519775e+00 4.25639600e-01
6.53044522e-01 -6.13366663e-01 8.47163558e-01 3.91982436e-01
2.51314372e-01 4.96118724e-01 -1.28263307e+00 5.33981264e-01
4.86073285e-01 -3.11502635e-01 -8.66229832e-01 6.05423033e-01
8.53678465e-01 -6.07939899e-01 -1.00794256e+00 -2.02546408e-03
3.61268669e-01 -8.84217381e-01 5.66417217e-01 -9.76204455e-01
-1.25392348e-01 4.26035881e-01 2.72016358e-02 -1.69257915e+00
6.92401454e-02 -1.05356240e+00 -6.29197201e-03 1.11599481e+00
4.68266398e-01 -6.34064496e-01 7.17105031e-01 1.21495056e+00
-1.27453968e-01 -5.61526299e-01 -9.22219336e-01 -4.90733564e-01
3.15151215e-01 1.06462672e-01 2.76752084e-01 6.74633205e-01
4.37864333e-01 9.25571263e-01 -6.69422925e-01 -8.37098286e-02
9.22813416e-02 -5.81423521e-01 1.29140234e+00 -1.00488234e+00
-4.29250330e-01 -2.15520501e-01 2.07391053e-01 -1.52396417e+00
3.13396215e-01 -5.32092573e-03 3.93354744e-01 -1.37003660e+00
-4.08364922e-01 -5.21193385e-01 1.35949597e-01 3.51698637e-01
-2.66082615e-01 -7.46219456e-01 2.43665203e-01 1.03326388e-01
-8.62564564e-01 7.08069026e-01 1.22366905e+00 2.18518034e-01
-7.90615499e-01 4.82373267e-01 -3.45952302e-01 5.86535692e-01
9.28750575e-01 -1.43079832e-01 -5.76521933e-01 -2.08286718e-01
-4.85654265e-01 5.73191404e-01 -2.90961981e-01 -7.19535410e-01
3.20309788e-01 -2.64408439e-01 -8.73698965e-02 -7.78339058e-02
5.68003297e-01 -3.66646141e-01 -7.72923410e-01 3.33515197e-01
-8.96721900e-01 2.28642523e-02 6.55666769e-01 4.10557240e-01
4.18836661e-02 -2.64752179e-01 5.57228267e-01 -3.30345988e-01
-7.61413872e-01 -3.94261152e-01 -1.00635982e+00 2.29150221e-01
8.76548707e-01 -3.83212306e-02 -2.24399701e-01 -9.62169826e-01
-9.39249396e-01 6.22053623e-01 8.70970190e-02 4.77454275e-01
3.40948373e-01 -7.82432854e-01 -8.98942530e-01 4.41205613e-02
5.67339808e-02 -8.38796645e-02 2.21839891e-05 3.94792527e-01
-4.97877188e-02 7.67082453e-01 -2.45826364e-01 -6.48487747e-01
-1.50177646e+00 7.83526246e-03 3.69288325e-01 -4.29848254e-01
-3.68872613e-01 6.25346243e-01 -6.19004108e-02 -9.08317327e-01
6.02603376e-01 3.38006243e-02 -5.05536020e-01 7.40776435e-02
5.76835811e-01 3.48464757e-01 2.32227948e-02 -4.27688003e-01
3.16264965e-02 -2.65653044e-01 -3.85164738e-01 -9.34791148e-01
1.06709659e+00 -4.10930425e-01 3.83738816e-01 8.31777275e-01
7.88194776e-01 -2.42881939e-01 -1.44312596e+00 -6.03004456e-01
2.14887559e-01 -2.42629349e-01 -2.04056695e-01 -1.21367884e+00
4.04699668e-02 7.32450843e-01 5.24735391e-01 5.73652923e-01
5.33219576e-01 -1.61449596e-01 3.51872087e-01 1.02626991e+00
4.20875520e-01 -1.50926161e+00 3.69900405e-01 1.07162321e+00
1.09277380e+00 -1.45388222e+00 -1.81539759e-01 1.71368316e-01
-1.46205938e+00 8.91624391e-01 9.39160705e-01 3.83468717e-01
2.28947341e-01 1.34276496e-02 5.16125619e-01 7.97057599e-02
-1.23354852e+00 -2.46701658e-01 -9.33105946e-02 7.50893056e-01
6.20317817e-01 4.01199199e-02 -1.70680881e-01 2.59906729e-03
-2.25266032e-02 -4.13408697e-01 6.66667521e-01 1.22620106e+00
-7.62274384e-01 -1.45641875e+00 -3.45748395e-01 3.30765754e-01
-1.09287791e-01 3.11482232e-02 -5.99238813e-01 9.07588065e-01
-7.74712920e-01 1.20056605e+00 9.23515335e-02 -1.41917184e-01
7.80545413e-01 6.44204438e-01 3.36575001e-01 -1.02334678e+00
-8.25118363e-01 3.10509562e-01 8.90877008e-01 -3.47387493e-01
-3.60230982e-01 -8.11478913e-01 -1.22129369e+00 -1.21202692e-02
-2.54213333e-01 6.01241887e-01 7.96203792e-01 9.65452671e-01
5.22484541e-01 3.87365282e-01 9.16156888e-01 -9.48666751e-01
-1.30883908e+00 -1.54664063e+00 -2.62453645e-01 2.37435237e-01
4.23061371e-01 -4.44545478e-01 -2.72878110e-01 -2.47411400e-01] | [13.02657413482666, 8.017746925354004] |
4a20de99-af44-491e-9127-6a05db32d4f0 | learning-large-scale-topological-maps-using | 1706.03416 | null | http://arxiv.org/abs/1706.03416v2 | http://arxiv.org/pdf/1706.03416v2.pdf | Learning Large-Scale Topological Maps Using Sum-Product Networks | In order to perform complex actions in human environments, an autonomous
robot needs the ability to understand the environment, that is, to gather and
maintain spatial knowledge. Topological map is commonly used for representing
large scale, global maps such as floor plans. Although much work has been done
in topological map extraction, we have found little previous work on the
problem of learning the topological map using a probabilistic model. Learning a
topological map means learning the structure of the large-scale space and
dependency between places, for example, how the evidence of a group of places
influence the attributes of other places. This is an important step towards
planning complex actions in the environment. In this thesis, we consider the
problem of using probabilistic deep learning model to learn the topological
map, which is essentially a sparse undirected graph where nodes represent
places annotated with their semantic attributes (e.g. place category). We
propose to use a novel probabilistic deep model, Sum-Product Networks (SPNs),
due to their unique properties. We present two methods for learning topological
maps using SPNs: the place grid method and the template-based method. We
contribute an algorithm that builds SPNs for graphs using template models. Our
experiments evaluate the ability of our models to enable robots to infer
semantic attributes and detect maps with novel semantic attribute arrangements.
Our results demonstrate their understanding of the topological map structure
and spatial relations between places. | ['Kaiyu Zheng'] | 2017-06-11 | null | null | null | null | ['probabilistic-deep-learning'] | ['computer-vision'] | [-5.45293503e-02 4.51678663e-01 1.75238743e-01 -6.76428497e-01
3.12015209e-02 -5.57947040e-01 7.81953096e-01 5.41775346e-01
-1.20171100e-01 9.10916209e-01 4.50920969e-01 -1.42319441e-01
-8.15522373e-01 -1.42126894e+00 -9.98139203e-01 -4.42251801e-01
-7.41807818e-01 1.19599891e+00 6.11476004e-01 -3.93122584e-01
6.41359687e-01 6.44202173e-01 -1.64138007e+00 2.68128753e-01
6.06698573e-01 6.22641504e-01 6.42240107e-01 4.23842490e-01
-3.20869535e-01 8.44840348e-01 -3.13842386e-01 2.79553562e-01
5.85441515e-02 3.43213230e-02 -1.10130155e+00 -2.84195840e-01
7.13468716e-02 1.22491419e-02 -4.81247127e-01 8.79765391e-01
2.50313245e-03 4.67747211e-01 1.02149308e+00 -1.54248822e+00
-5.95785022e-01 8.94767404e-01 -8.25115964e-02 1.09951822e-02
2.99510837e-01 -2.83447593e-01 8.36274803e-01 -3.61566693e-01
6.25538468e-01 1.52359772e+00 7.23229766e-01 -2.25206122e-01
-1.20336175e+00 -2.47193381e-01 2.00332582e-01 6.38754368e-01
-1.60423326e+00 1.45743473e-03 8.05106103e-01 -6.83663070e-01
1.05134797e+00 -1.54476434e-01 5.57530820e-01 8.52445364e-01
5.08875191e-01 4.25297439e-01 1.28367853e+00 -3.04985285e-01
4.87083912e-01 -1.56584792e-02 -7.90755078e-02 8.56668949e-01
5.32529205e-02 -5.03416918e-02 -6.67883813e-01 9.73955542e-02
1.13835871e+00 1.79135799e-02 4.05583888e-01 -8.42239022e-01
-1.27579451e+00 4.98656392e-01 1.03321254e+00 5.66169977e-01
-4.98038173e-01 4.51913774e-01 6.83880448e-02 -8.37076753e-02
1.28049910e-01 5.50114274e-01 -4.92628396e-01 -7.08419979e-02
-5.45717359e-01 3.27770740e-01 8.93500388e-01 1.05523491e+00
1.32318091e+00 -4.60905969e-01 1.75731480e-01 6.00830495e-01
4.53248769e-01 4.07363147e-01 1.27096295e-01 -9.80008245e-01
4.60678458e-01 7.84727216e-01 1.06414020e-01 -1.79311967e+00
-7.42014229e-01 -5.41305058e-02 -6.97491467e-01 2.30380341e-01
1.75927207e-01 3.80662590e-01 -1.09734452e+00 1.82834840e+00
1.58274978e-01 1.98027194e-01 3.07833333e-03 5.22005916e-01
5.72222531e-01 7.22603679e-01 1.21679232e-01 7.32585490e-01
1.14139247e+00 -4.56239820e-01 -5.66222727e-01 -4.02087748e-01
6.19282842e-01 -1.42835140e-01 6.69118941e-01 -8.60877633e-02
-4.61478174e-01 -4.04351950e-01 -1.04667890e+00 -1.45447001e-01
-1.33874154e+00 -1.13299094e-01 9.06507671e-01 2.59565622e-01
-1.44251907e+00 8.65845919e-01 -1.00735629e+00 -7.12635040e-01
4.37471539e-01 5.58444381e-01 -7.70432770e-01 -3.29043306e-02
-1.24167597e+00 1.32588673e+00 1.05665445e+00 -1.84263457e-02
-8.17221820e-01 -1.36082023e-01 -1.29793310e+00 1.77360535e-01
2.44225517e-01 -5.94517767e-01 7.50246465e-01 -3.82728606e-01
-1.02818847e+00 4.85274285e-01 -1.34369597e-01 -3.88626605e-01
4.33397405e-02 1.51616126e-01 -7.99528211e-02 -2.16188669e-01
5.94312191e-01 9.47256923e-01 2.04514116e-01 -1.33109295e+00
-7.37580895e-01 -5.80221653e-01 3.20852071e-01 2.99781382e-01
1.29480302e-01 -5.32992780e-01 -1.41694592e-02 -1.39229119e-01
7.26915061e-01 -1.01886582e+00 -4.00857687e-01 -3.25776964e-01
-5.95715106e-01 -4.10377085e-01 6.58015132e-01 -5.98540187e-01
7.28896499e-01 -2.10424089e+00 2.58951075e-02 7.79731452e-01
1.58306912e-01 -5.59324503e-01 9.77293239e-04 6.98499084e-01
3.27271931e-02 1.96572363e-01 -4.33561116e-01 2.45333705e-02
4.67602342e-01 6.67858839e-01 -1.12705797e-01 3.18498671e-01
9.23282728e-02 7.13109970e-01 -1.33538866e+00 -4.55980510e-01
5.62273264e-01 4.26501811e-01 -3.38483691e-01 -1.79134652e-01
-1.42772987e-01 3.25864285e-01 -5.74627042e-01 3.16241115e-01
5.75499177e-01 2.98331287e-02 2.87445277e-01 -6.48676604e-03
-3.15270573e-01 5.82562923e-01 -1.44670796e+00 1.77995682e+00
-3.27910155e-01 8.73735428e-01 -3.97514760e-01 -9.78958964e-01
1.20001662e+00 -8.38419646e-02 4.08298612e-01 -6.73707724e-01
-7.21877441e-02 2.60241747e-01 -1.99161738e-01 -3.18293422e-01
6.68650508e-01 -3.49609554e-03 -3.82852107e-01 1.82691410e-01
2.11873904e-01 -2.63743252e-01 1.20404154e-01 2.49644309e-01
1.40794408e+00 3.28415245e-01 2.60218620e-01 -5.56875825e-01
1.94482878e-01 1.72191530e-01 3.21383387e-01 7.10907936e-01
-1.13500580e-02 1.27041146e-01 5.45228064e-01 -9.34330165e-01
-1.02975535e+00 -1.49186754e+00 1.13215581e-01 9.99610186e-01
3.53667319e-01 -4.25255835e-01 -4.55837488e-01 -4.62608039e-01
1.78890787e-02 7.15960324e-01 -9.01172519e-01 -7.81241292e-03
-6.00317895e-01 -2.65688390e-01 2.84755498e-01 6.69206738e-01
6.22282863e-01 -1.39435434e+00 -5.88705599e-01 1.93757653e-01
-3.38563591e-01 -8.44612598e-01 2.77675420e-01 5.77963233e-01
-7.66799748e-01 -1.19780409e+00 7.85776302e-02 -1.00815010e+00
8.60333920e-01 2.87010223e-02 1.10453629e+00 -3.07030290e-01
3.01476736e-02 4.02618080e-01 -3.48583549e-01 -6.21183872e-01
-6.36702329e-02 1.71023741e-01 2.51130968e-01 -3.58205169e-01
3.64585787e-01 -1.09258676e+00 -2.33420685e-01 4.34151411e-01
-6.83099627e-01 1.01418175e-01 8.26898754e-01 4.49783385e-01
6.77427232e-01 7.09888995e-01 1.62407011e-01 -5.71222365e-01
4.18854415e-01 -7.12091088e-01 -4.55428898e-01 1.90353006e-01
-6.46603853e-02 6.13462150e-01 1.95005536e-01 2.01036334e-01
-7.01316595e-01 4.28886741e-01 1.61372684e-02 1.10967852e-01
-4.92844224e-01 7.47121334e-01 -3.40549707e-01 2.75462884e-02
5.89438438e-01 1.32404670e-01 -2.13530019e-01 -2.44945124e-01
5.74301720e-01 3.10927898e-01 6.06340647e-01 -8.21126640e-01
6.66844785e-01 5.86448491e-01 5.68040550e-01 -6.94407582e-01
-4.83756274e-01 -5.39716363e-01 -1.44075358e+00 -1.29259586e-01
8.95593345e-01 -5.00494659e-01 -7.14283526e-01 5.04486971e-02
-1.20726752e+00 -4.10757124e-01 -7.67165646e-02 4.37264889e-01
-1.03639352e+00 -4.81799208e-02 -2.13039279e-01 -6.46567404e-01
5.08864284e-01 -8.81449342e-01 1.02309239e+00 4.74024191e-02
-3.25685352e-01 -1.30698717e+00 4.40021515e-01 -1.97462693e-01
3.39066744e-01 6.34771168e-01 1.00281179e+00 -7.46258497e-01
-1.01232493e+00 1.72233447e-01 -3.05963129e-01 -1.97122306e-01
3.41158330e-01 -3.93622696e-01 -7.69780815e-01 3.17564726e-01
-3.48408490e-01 1.51122525e-01 6.73159361e-01 4.99749482e-01
1.15617251e+00 -3.34951311e-01 -8.42496216e-01 2.78429657e-01
1.40217948e+00 2.44621009e-01 8.73027921e-01 7.65115976e-01
8.78791988e-01 1.03171027e+00 5.91156542e-01 1.80922642e-01
8.74662340e-01 5.35286725e-01 7.64599204e-01 2.19755173e-01
1.76526353e-01 -6.98555112e-01 -2.96578766e-03 4.81757224e-01
-1.83727741e-01 7.56296068e-02 -1.43964565e+00 8.41874480e-01
-2.20953918e+00 -9.65033889e-01 -2.39095867e-01 1.93632734e+00
3.14183116e-01 1.79866701e-01 -3.26269150e-01 7.54928067e-02
7.66278565e-01 1.09211922e-01 -1.51436433e-01 -2.43605569e-01
-3.19861583e-02 3.41087300e-03 8.06948304e-01 6.87575638e-01
-1.40994084e+00 1.22536385e+00 6.01927280e+00 5.25498927e-01
-5.69957197e-01 -1.80282325e-01 3.35447580e-01 6.71185493e-01
-1.76857904e-01 1.91517770e-01 -5.20897567e-01 1.68212757e-01
9.13242579e-01 2.79545218e-01 5.10085881e-01 1.03056407e+00
1.50424093e-01 -3.63192856e-01 -1.22900784e+00 9.36274230e-01
-1.21901803e-01 -1.45058692e+00 -1.94064416e-02 3.30555111e-01
5.62116802e-01 1.24705374e-01 -1.36314020e-01 3.99535865e-01
8.54799569e-01 -1.25891423e+00 8.72228920e-01 8.60458910e-01
2.98866808e-01 -7.45581567e-01 9.02676344e-01 4.88735110e-01
-1.44285274e+00 -8.62836465e-02 -5.21479845e-01 -3.70826781e-01
1.10101379e-01 5.06637096e-01 -1.51729310e+00 6.27674282e-01
8.70805740e-01 8.27574909e-01 -6.51663303e-01 1.21027601e+00
-5.34013689e-01 -1.57256667e-02 -5.58508575e-01 -2.55511314e-01
4.86153841e-01 -1.35506883e-01 4.73966509e-01 1.02532566e+00
4.89475191e-01 -3.54871601e-02 3.88282239e-01 1.09618390e+00
2.33398378e-01 -2.74002463e-01 -1.37615287e+00 1.90175235e-01
7.38195062e-01 9.39324975e-01 -1.26449180e+00 -2.02847809e-01
2.79039741e-01 5.83603084e-01 4.78683084e-01 9.14795548e-02
-5.02333283e-01 -5.47968924e-01 4.89799201e-01 2.38083214e-01
1.49976566e-01 -7.85643756e-01 -5.96760392e-01 -3.07076782e-01
1.65261002e-03 -1.45891562e-01 8.70705172e-02 -1.13075316e+00
-1.05700612e+00 3.78034770e-01 3.81795675e-01 -9.31884110e-01
-1.50765955e-01 -8.01179886e-01 -3.32176954e-01 9.78383183e-01
-1.28959620e+00 -1.10284448e+00 -2.92756170e-01 5.27529240e-01
2.85419524e-01 1.06310755e-01 8.90629709e-01 -2.20198944e-01
3.58160347e-01 -2.92860508e-01 7.17581064e-02 2.13699609e-01
2.75931031e-01 -1.66763377e+00 5.67861617e-01 4.90965515e-01
3.84130210e-01 6.52358413e-01 7.17837632e-01 -1.04839599e+00
-9.94052470e-01 -1.14832520e+00 1.12786341e+00 -1.01502204e+00
6.84098661e-01 -6.52022898e-01 -7.76140213e-01 8.95569026e-01
-1.22234605e-01 -2.48644784e-01 4.56612796e-01 4.51651514e-01
-2.58590966e-01 -3.01227327e-02 -1.15980971e+00 4.43030149e-01
1.30858397e+00 -6.43194199e-01 -9.66950536e-01 4.25376147e-01
6.24866784e-01 -1.84229553e-01 -6.94072723e-01 4.12448406e-01
3.81862164e-01 -9.59826410e-01 9.91404176e-01 -2.22960159e-01
1.40039220e-01 -7.11227119e-01 -4.59302098e-01 -1.76490927e+00
-9.36400831e-01 1.23930477e-01 3.23702097e-01 9.53934014e-01
4.79046643e-01 -6.88813448e-01 8.23203206e-01 4.92026150e-01
-4.24298227e-01 -2.70151079e-01 -9.91329908e-01 -8.76798451e-01
-2.76567996e-01 -4.87036526e-01 9.80418205e-01 1.20539868e+00
9.82661620e-02 1.48132235e-01 6.18742518e-02 6.84807658e-01
4.78537768e-01 -2.62394875e-01 7.07033455e-01 -1.65596616e+00
3.82440537e-01 -4.63131011e-01 -1.25279117e+00 -5.72863877e-01
2.50839561e-01 -9.39018369e-01 4.71366644e-01 -2.35658622e+00
-5.79721183e-02 -9.24996257e-01 -3.70631516e-01 7.14452863e-01
3.53282273e-01 -2.14687422e-01 -4.71914597e-02 1.00719586e-01
-8.70918453e-01 5.72392106e-01 7.45422482e-01 -3.53413731e-01
-2.23333240e-01 -3.45490277e-01 -2.96544880e-01 8.04027796e-01
1.03177106e+00 -6.90400302e-01 -5.29386997e-01 -4.86997426e-01
6.59663439e-01 -3.60653847e-01 5.04421532e-01 -1.41736937e+00
5.59523702e-01 -2.44615927e-01 5.74288368e-01 -8.84628952e-01
4.00314867e-01 -1.07429802e+00 3.01299930e-01 3.87286931e-01
-1.29562542e-01 3.67059499e-01 2.20492855e-01 7.57507920e-01
-3.08621556e-01 -7.18839234e-03 1.06346101e-01 -4.89208907e-01
-1.40408802e+00 8.97714719e-02 -3.80491674e-01 -4.28245753e-01
9.26877141e-01 -3.18645656e-01 -2.82090455e-01 -2.23619938e-01
-1.04005408e+00 3.29765350e-01 4.93015110e-01 5.21690845e-01
7.39225566e-01 -1.48772097e+00 -3.32831889e-01 6.51674420e-02
1.31992370e-01 2.36985803e-01 1.54742420e-01 5.12576997e-01
-8.77315938e-01 5.27038515e-01 -7.97654033e-01 -6.27239168e-01
-7.35054493e-01 6.22992516e-01 1.58075213e-01 -2.47991517e-01
-6.05969369e-01 7.91542768e-01 2.40480497e-01 -1.03596938e+00
-8.64098966e-03 -5.60468554e-01 -5.85062802e-01 -2.35908151e-01
2.23335415e-01 4.51385975e-01 -4.83574718e-03 -8.61723125e-01
-5.14661431e-01 3.78731012e-01 4.48786795e-01 -2.26203099e-01
1.59876478e+00 -1.39342561e-01 -6.88197732e-01 7.99061000e-01
8.25745344e-01 -3.11388254e-01 -9.59318876e-01 -5.15918508e-02
4.77219135e-01 -5.48442066e-01 -2.06225350e-01 -8.07725847e-01
-3.53046745e-01 6.83101594e-01 5.56771457e-01 3.56528759e-01
7.04232693e-01 4.09973532e-01 5.16680330e-02 8.59021008e-01
1.10054708e+00 -1.17499578e+00 -1.22830607e-01 1.09873545e+00
8.96154523e-01 -1.15884697e+00 -9.88622308e-02 -4.51707840e-01
-3.71388733e-01 1.01020873e+00 3.08723986e-01 -2.76337802e-01
8.94900680e-01 -9.37466603e-03 -4.29364592e-01 -6.97314560e-01
-2.80545682e-01 -3.55283529e-01 2.53744423e-01 1.15489709e+00
-9.89791192e-03 7.19668329e-01 3.60678464e-01 2.06351414e-01
-7.33247042e-01 -3.58283043e-01 4.26640421e-01 1.05018806e+00
-9.08265233e-01 -9.17643070e-01 -5.72991848e-01 4.23652023e-01
2.38489479e-01 1.80539638e-02 -6.08106196e-01 7.36468017e-01
4.64471579e-01 8.00490916e-01 3.49774003e-01 -6.45676732e-01
3.70998412e-01 5.39325885e-02 4.94601488e-01 -7.65582561e-01
1.53221488e-01 -7.96596825e-01 1.76678210e-01 -8.21640849e-01
-5.09724915e-01 -7.21699297e-01 -1.43968344e+00 -1.91496417e-01
3.23638529e-01 2.07820296e-01 1.25125015e+00 1.21298981e+00
3.81971329e-01 6.55830383e-01 1.82758063e-01 -1.18285465e+00
3.91639620e-01 -9.44140732e-01 -7.68652081e-01 2.69296736e-01
9.82041582e-02 -1.25856411e+00 4.16326821e-02 -5.30553535e-02] | [4.910096645355225, 0.41539883613586426] |
cd392296-345b-4426-a25d-3c1ef36dcbb9 | counterpoint-by-convolution | 1903.07227 | null | http://arxiv.org/abs/1903.07227v1 | http://arxiv.org/pdf/1903.07227v1.pdf | Counterpoint by Convolution | Machine learning models of music typically break up the task of composition
into a chronological process, composing a piece of music in a single pass from
beginning to end. On the contrary, human composers write music in a nonlinear
fashion, scribbling motifs here and there, often revisiting choices previously
made. In order to better approximate this process, we train a convolutional
neural network to complete partial musical scores, and explore the use of
blocked Gibbs sampling as an analogue to rewriting. Neither the model nor the
generative procedure are tied to a particular causal direction of composition.
Our model is an instance of orderless NADE (Uria et al., 2014), which allows
more direct ancestral sampling. However, we find that Gibbs sampling greatly
improves sample quality, which we demonstrate to be due to some conditional
distributions being poorly modeled. Moreover, we show that even the cheap
approximate blocked Gibbs procedure from Yao et al. (2014) yields better
samples than ancestral sampling, based on both log-likelihood and human
evaluation. | ['Cheng-Zhi Anna Huang', 'Tim Cooijmans', 'Adam Roberts', 'Douglas Eck', 'Aaron Courville'] | 2019-03-18 | null | null | null | null | ['music-modeling'] | ['music'] | [ 3.97687674e-01 2.58133650e-01 2.61938348e-02 -1.35382771e-01
-6.95532620e-01 -9.12446260e-01 9.17533696e-01 -1.94205835e-01
-2.23974273e-01 7.19892502e-01 4.66929018e-01 -9.21991095e-02
-4.70609926e-02 -9.44874704e-01 -9.89758611e-01 -4.45933193e-01
-2.80861650e-02 8.66650105e-01 -1.67354524e-01 -2.18029603e-01
2.11802751e-01 1.92577466e-01 -1.27808440e+00 2.72087902e-01
4.70155299e-01 3.64205033e-01 3.17169749e-03 8.60535502e-01
9.91666317e-02 6.67328477e-01 -3.91273707e-01 -8.07339311e-01
2.77931422e-01 -1.03975523e+00 -8.13944757e-01 -8.98807421e-02
5.16387820e-01 -3.93264502e-01 -1.33783847e-01 9.36585665e-01
2.49120608e-01 1.12079568e-01 7.87876189e-01 -7.45588720e-01
-6.37117803e-01 1.32619154e+00 -1.57429203e-01 -5.21758020e-01
2.21096575e-01 1.21087499e-01 1.44771659e+00 -6.50162935e-01
7.16249108e-01 9.65603590e-01 1.11800432e+00 4.47575033e-01
-1.65469253e+00 -4.31904376e-01 -1.92940161e-01 -1.17108710e-01
-1.18694842e+00 -4.44141597e-01 8.06933641e-01 -3.96879405e-01
5.14001667e-01 3.58678490e-01 1.14171660e+00 1.26701498e+00
-1.04178570e-01 9.42881346e-01 8.52318287e-01 -3.87025774e-01
3.39075595e-01 -5.19613564e-01 -3.52072775e-01 6.06236517e-01
1.27795227e-02 1.98160529e-01 -8.16948116e-01 -4.14902687e-01
8.15731764e-01 -1.63632095e-01 -1.50294006e-01 -4.11447078e-01
-1.23492694e+00 5.74531019e-01 3.65747899e-01 1.39042750e-01
-4.02170271e-01 5.76120019e-01 2.05083787e-01 1.70906618e-01
-4.58024368e-02 7.93282390e-01 -2.91566223e-01 -4.25753713e-01
-1.48408902e+00 6.02385938e-01 1.09766471e+00 6.53080642e-01
6.30586624e-01 -6.42016083e-02 -1.12278163e-02 6.14790618e-01
3.45275700e-02 4.27068472e-02 2.37758026e-01 -1.38509095e+00
-1.32518649e-01 -1.13423839e-01 1.46603554e-01 -3.63137335e-01
-5.26989363e-02 -5.83448887e-01 -7.09844947e-01 2.94383317e-01
9.50010180e-01 8.22376534e-02 -7.85524249e-01 2.15631104e+00
-1.09421298e-01 2.92419821e-01 -3.29932809e-01 8.66061151e-01
-4.67443392e-02 3.56231481e-01 -2.72658374e-02 -2.17388898e-01
1.13720262e+00 -7.83820629e-01 -3.98989499e-01 8.14559869e-03
1.69600040e-01 -8.61194134e-01 1.35975969e+00 8.93638968e-01
-1.55386364e+00 -2.86254078e-01 -1.05740333e+00 -3.03842157e-01
4.42063004e-01 -4.11722623e-03 7.90355086e-01 4.20106381e-01
-9.26370621e-01 1.35069180e+00 -9.88867462e-01 -1.67734087e-01
5.67155063e-01 2.82461215e-02 5.50404117e-02 2.45561212e-01
-9.35575902e-01 5.32088280e-01 7.34101012e-02 6.61719516e-02
-1.13123345e+00 -8.13390613e-01 -3.41453671e-01 5.95855080e-02
3.15606326e-01 -1.02124035e+00 1.70037925e+00 -1.26015699e+00
-1.95575368e+00 8.13665450e-01 -9.63255167e-02 -5.53680658e-01
8.64580274e-01 -3.90836835e-01 2.07179070e-01 -1.98695555e-01
-1.30417153e-01 3.59967381e-01 7.81514525e-01 -1.08235157e+00
-8.88075754e-02 -2.20443204e-01 -4.25325008e-03 4.29327562e-02
2.60001659e-01 -1.68840289e-01 -3.65327567e-01 -8.00639510e-01
1.80097222e-01 -1.06742752e+00 -1.54391289e-01 -9.56531167e-02
-5.56247115e-01 -1.40316024e-01 -1.99954748e-01 -8.31234813e-01
1.24744833e+00 -1.99078679e+00 4.86932337e-01 2.00340569e-01
7.39502162e-02 -2.35008687e-01 -2.46686116e-03 5.09170771e-01
1.17034167e-02 2.43903786e-01 -8.23295176e-01 -3.96333635e-01
4.13114518e-01 5.81120290e-02 -4.61114287e-01 2.49634400e-01
2.76731439e-02 1.17853153e+00 -9.42935705e-01 -2.75067557e-02
-3.19680423e-01 4.30784047e-01 -8.50797057e-01 -7.72685260e-02
-7.01373279e-01 4.34665114e-01 -3.86840627e-02 4.31726396e-01
2.29181722e-01 -2.28540093e-01 6.28361821e-01 -1.27164215e-01
-4.10904549e-02 9.65099335e-01 -1.05337477e+00 2.31191730e+00
-4.17582750e-01 6.43605471e-01 -1.03311203e-01 -6.78845286e-01
5.99271834e-01 4.18210953e-01 2.91679591e-01 -2.07663357e-01
-3.13867889e-02 3.42761368e-01 1.03397787e-01 -1.68920040e-01
6.83729231e-01 -6.09724939e-01 -1.00829892e-01 9.78836715e-01
-1.37746304e-01 -3.46757948e-01 1.62269801e-01 1.57408535e-01
1.17677653e+00 9.98282373e-01 3.51144552e-01 -8.50483403e-02
-1.88280866e-01 4.18545045e-02 5.98568797e-01 9.74129140e-01
1.90082580e-01 1.05520737e+00 7.23078787e-01 -4.43942457e-01
-1.65348005e+00 -1.57043040e+00 -5.76991169e-03 9.10431027e-01
-3.84293318e-01 -6.69367433e-01 -7.49876678e-01 -3.30367774e-01
-1.85830355e-01 1.03911579e+00 -4.81362641e-01 -1.00914538e-01
-8.64621699e-01 -6.78136170e-01 8.47512126e-01 3.53081852e-01
1.05574615e-01 -1.17782545e+00 -7.09691107e-01 4.04011339e-01
-2.08274096e-01 -2.54934877e-01 -5.13587177e-01 2.29065225e-01
-9.18172717e-01 -8.36455703e-01 -6.53735578e-01 -4.99628127e-01
1.70478359e-01 -5.82865298e-01 1.59542310e+00 1.13737294e-02
-2.28570074e-01 3.12476978e-02 3.24964188e-02 -1.38988316e-01
-7.09879339e-01 3.10155123e-01 -1.51802719e-01 -2.16531366e-01
2.25459471e-01 -1.31306911e+00 -5.25704384e-01 -1.19087167e-01
-8.76977861e-01 3.00816149e-01 6.37323499e-01 9.12343979e-01
4.79070663e-01 -3.75674486e-01 2.55796939e-01 -8.86799157e-01
5.70113599e-01 -2.78078914e-01 -4.16944325e-01 1.71229929e-01
-4.65300053e-01 2.93569505e-01 5.37871361e-01 -5.79596817e-01
-7.60026932e-01 -1.20748943e-02 -1.31993517e-01 -2.45062292e-01
-9.54875574e-02 4.66749609e-01 4.56697792e-02 6.66896760e-01
8.52050185e-01 2.44223729e-01 -8.21186602e-03 -7.40371346e-01
4.63802874e-01 1.58693001e-01 7.05853581e-01 -9.28047121e-01
7.03693867e-01 4.53816026e-01 -1.11816507e-02 -3.09701622e-01
-7.96072483e-01 3.45862150e-01 -5.19681633e-01 -1.67215262e-02
7.88597882e-01 -7.55637050e-01 -9.67864335e-01 1.96085110e-01
-1.03395176e+00 -7.74020553e-01 -7.53759444e-01 4.37155485e-01
-1.00235569e+00 2.05719516e-01 -8.80582035e-01 -7.97396600e-01
-4.37394865e-02 -7.71604538e-01 1.00449169e+00 -1.47396371e-01
-8.24198067e-01 -5.40689111e-01 3.99871469e-01 -3.86686921e-02
1.84909970e-01 2.84376144e-01 1.03360271e+00 -3.45126182e-01
-8.62560868e-01 4.61783446e-02 3.64662826e-01 1.34537786e-01
1.24042422e-01 1.45163849e-01 -8.74648213e-01 -4.33564521e-02
-4.84077223e-02 -3.83976310e-01 1.07533002e+00 1.41097412e-01
8.95142496e-01 -3.07358801e-01 1.18050791e-01 7.50791013e-01
1.38510895e+00 -9.17086005e-02 7.04846323e-01 2.60477245e-01
5.81366479e-01 4.23195392e-01 -7.45313019e-02 3.42558116e-01
1.27442479e-01 5.70151925e-01 2.12171301e-01 4.75417137e-01
-3.46793115e-01 -7.98656881e-01 4.66084689e-01 9.48297441e-01
-3.85484785e-01 5.98018840e-02 -8.78868163e-01 5.67300141e-01
-1.75744808e+00 -1.36218202e+00 2.79691983e-02 2.40211797e+00
1.40802860e+00 1.34884983e-01 3.32883149e-01 1.77548960e-01
4.93386835e-01 9.80915129e-02 -5.29493511e-01 -2.65279263e-01
-1.22806296e-01 6.52231514e-01 2.26273641e-01 6.51709437e-01
-6.67692780e-01 8.62387002e-01 6.99161243e+00 6.44615710e-01
-8.50522220e-01 -1.90109189e-03 4.61383581e-01 -5.62616765e-01
-8.35623741e-01 4.64869797e-01 -1.90138146e-01 4.50509757e-01
7.09406853e-01 5.59949726e-02 1.29204142e+00 4.59745318e-01
-5.61163537e-02 -4.79939859e-03 -1.44994521e+00 6.74394250e-01
-3.10871098e-02 -1.50202847e+00 2.87908912e-02 9.00963396e-02
6.98541284e-01 -3.45998667e-02 -8.50190893e-02 3.55295241e-02
8.50580275e-01 -1.20651317e+00 1.27752304e+00 8.91845286e-01
5.98511755e-01 -6.75594568e-01 -1.89786218e-03 3.67020100e-01
-7.58193433e-01 2.44734302e-01 -1.22238085e-01 -3.23564708e-01
4.62610453e-01 5.51173925e-01 -7.41012037e-01 4.09344226e-01
4.27649140e-01 4.71644342e-01 -3.10296178e-01 7.49229610e-01
-6.53298438e-01 9.85787153e-01 -5.16100764e-01 -3.79394516e-02
-1.15654863e-01 -5.63766897e-01 7.36657560e-01 7.79569447e-01
5.22565424e-01 -1.27801090e-01 -1.45992488e-01 1.42144704e+00
-2.57674068e-01 -2.35910177e-01 -3.67313147e-01 -2.08833605e-01
4.14073586e-01 9.11834061e-01 -5.89794159e-01 -2.49481812e-01
8.85027051e-02 1.13610089e+00 3.82498085e-01 4.74274486e-01
-6.26136005e-01 -2.71951351e-02 4.81995225e-01 2.36786112e-01
4.96167213e-01 -4.33617681e-01 -6.49241626e-01 -1.17400336e+00
2.82196701e-02 -1.07583237e+00 -9.28007513e-02 -9.19456422e-01
-1.23385143e+00 2.05856681e-01 -3.90688777e-01 -9.37396228e-01
-4.85287964e-01 -1.06145121e-01 -7.14791179e-01 8.64304602e-01
-8.79167855e-01 -1.00946057e+00 2.27069497e-01 -2.58695101e-03
3.99448156e-01 1.83030501e-01 8.65698755e-01 9.34154019e-02
-5.23899496e-02 4.33246732e-01 1.25751555e-01 1.44464478e-01
7.62350142e-01 -1.38739502e+00 6.87105715e-01 7.59653628e-01
7.60838628e-01 9.34478641e-01 9.59068239e-01 -6.21183693e-01
-1.21361160e+00 -5.25034070e-01 1.07913256e+00 -5.51200569e-01
6.94632649e-01 -4.48685795e-01 -7.17307389e-01 8.39438856e-01
3.90119374e-01 -6.68815374e-01 6.07170820e-01 4.65562463e-01
-5.93613625e-01 1.09133519e-01 -5.67519605e-01 1.04634857e+00
1.38144743e+00 -7.67900884e-01 -7.23460555e-01 9.09659863e-02
4.88103271e-01 -1.32882640e-01 -7.25118935e-01 2.57278442e-01
1.03385866e+00 -1.22612739e+00 7.30580270e-01 -6.19663537e-01
8.61737967e-01 -5.36782265e-01 -2.47123212e-01 -1.27004576e+00
-3.13429087e-01 -1.23279464e+00 2.98074745e-02 1.13952518e+00
4.57254469e-01 -1.47084042e-01 9.51748610e-01 3.34770828e-01
-1.64135069e-01 -5.67792833e-01 -6.64374053e-01 -6.14972174e-01
3.29558909e-01 -7.56091833e-01 8.58667612e-01 8.39242458e-01
-1.78117439e-01 3.19698662e-01 -3.75150084e-01 -2.37049803e-01
7.67203808e-01 3.62309635e-01 8.63081753e-01 -1.18554080e+00
-1.09480047e+00 -8.06909382e-01 1.88881442e-01 -1.13289011e+00
-1.03815295e-01 -1.23971641e+00 2.19250634e-01 -1.34688056e+00
2.58826137e-01 -4.80100274e-01 -1.44571826e-01 2.33588427e-01
-4.70595397e-02 5.03160536e-01 3.26229632e-01 4.13140863e-01
-1.93476275e-01 4.03340489e-01 1.18694723e+00 2.38439236e-02
-1.03275754e-01 -2.57462692e-02 -7.81741261e-01 7.91036248e-01
7.74841964e-01 -5.61130702e-01 -1.60073996e-01 -5.15075088e-01
1.02007401e+00 8.45817551e-02 6.07155800e-01 -8.66738796e-01
2.10128844e-01 -7.75313471e-03 3.44958276e-01 -3.63443881e-01
2.17426285e-01 -2.85466015e-01 8.14883471e-01 4.11162704e-01
-7.29269564e-01 -2.73139402e-02 -2.64024079e-01 4.65458423e-01
2.81997640e-02 -3.99986446e-01 5.86255431e-01 -3.82530093e-01
1.82947889e-02 6.18243814e-02 -1.83043063e-01 4.88705281e-03
2.71979511e-01 -8.19223002e-02 1.70845628e-01 -5.29482961e-01
-1.02780879e+00 -4.23684210e-01 9.35726702e-01 -6.26347363e-02
3.45258676e-02 -1.52163112e+00 -7.62922645e-01 1.42008305e-01
-3.46138269e-01 1.06164142e-01 -7.82540590e-02 8.75688076e-01
-7.36760080e-01 -2.93786168e-01 -4.19177935e-02 -3.95896584e-01
-6.15397513e-01 2.39049867e-01 2.79915124e-01 -2.39477918e-01
-7.07122684e-01 8.47919822e-01 -1.38221066e-02 -4.70574021e-01
1.91771626e-01 -1.93115830e-01 4.71022516e-01 3.03681511e-02
2.69511461e-01 2.54318595e-01 -2.71982878e-01 -2.68366877e-02
9.61878896e-03 4.13064867e-01 3.03358436e-01 -6.97737098e-01
1.40429568e+00 1.03205241e-01 -3.86402905e-01 8.49159956e-01
7.75186062e-01 3.76524985e-01 -1.55596042e+00 -9.31419805e-02
1.08621411e-01 -1.16656400e-01 -4.98506188e-01 -1.00484562e+00
-5.66067040e-01 7.15783119e-01 -1.54461013e-02 2.27990463e-01
8.54900777e-01 -1.33062722e-02 8.44422042e-01 3.64793479e-01
3.32308948e-01 -1.03165710e+00 -1.45777211e-01 5.26807666e-01
7.55835712e-01 -5.39526939e-01 4.94445041e-02 2.71375448e-01
-3.48293632e-01 1.12826777e+00 -1.23822577e-01 -4.93583977e-01
4.54204589e-01 3.06201398e-01 -2.98155010e-01 -7.84039870e-03
-7.76707649e-01 3.46972123e-02 5.25617003e-02 2.46855646e-01
6.39389694e-01 2.33656943e-01 -2.60818362e-01 6.81354225e-01
-9.94575381e-01 1.82763293e-01 4.49233651e-01 6.16112053e-01
-2.45338514e-01 -1.30608487e+00 -2.76779115e-01 2.82898426e-01
-5.33564448e-01 -4.30319101e-01 -5.57638109e-01 6.31408095e-01
1.54053435e-01 3.78597885e-01 3.05134743e-01 -2.21566007e-01
-6.56381845e-02 5.77884078e-01 9.59316671e-01 -5.00525177e-01
-6.15777671e-01 2.59123355e-01 3.16762030e-01 -3.84380430e-01
-3.03101003e-01 -9.49214876e-01 -1.07257724e+00 -5.50178885e-01
7.02942610e-02 8.90653282e-02 6.34587646e-01 8.75267744e-01
1.05264470e-01 6.46627307e-01 2.76619822e-01 -9.19632256e-01
-7.20708847e-01 -1.00274193e+00 -6.08043849e-01 5.25593281e-01
2.42784187e-01 -7.92584270e-02 -3.22361648e-01 3.15395087e-01] | [15.896590232849121, 5.562900066375732] |
01a2e095-63c8-46fd-847f-60e223718bfa | tapex-table-pre-training-via-learning-a | 2107.07653 | null | https://arxiv.org/abs/2107.07653v3 | https://arxiv.org/pdf/2107.07653v3.pdf | TAPEX: Table Pre-training via Learning a Neural SQL Executor | Recent progress in language model pre-training has achieved a great success via leveraging large-scale unstructured textual data. However, it is still a challenge to apply pre-training on structured tabular data due to the absence of large-scale high-quality tabular data. In this paper, we propose TAPEX to show that table pre-training can be achieved by learning a neural SQL executor over a synthetic corpus, which is obtained by automatically synthesizing executable SQL queries and their execution outputs. TAPEX addresses the data scarcity challenge via guiding the language model to mimic a SQL executor on the diverse, large-scale and high-quality synthetic corpus. We evaluate TAPEX on four benchmark datasets. Experimental results demonstrate that TAPEX outperforms previous table pre-training approaches by a large margin and achieves new state-of-the-art results on all of them. This includes the improvements on the weakly-supervised WikiSQL denotation accuracy to 89.5% (+2.3%), the WikiTableQuestions denotation accuracy to 57.5% (+4.8%), the SQA denotation accuracy to 74.5% (+3.5%), and the TabFact accuracy to 84.2% (+3.2%). To our knowledge, this is the first work to exploit table pre-training via synthetic executable programs and to achieve new state-of-the-art results on various downstream tasks. Our code can be found at https://github.com/microsoft/Table-Pretraining. | ['Weizhu Chen', 'Morteza Ziyadi', 'Jian-Guang Lou', 'Zeqi Lin', 'Jiaqi Guo', 'Bei Chen', 'Qian Liu'] | 2021-07-16 | tapex-table-pre-training-via-learning-a-1 | https://openreview.net/forum?id=O50443AsCP | https://openreview.net/pdf?id=O50443AsCP | iclr-2022-4 | ['table-based-fact-verification'] | ['natural-language-processing'] | [ 4.82842922e-02 2.55169779e-01 -5.27637720e-01 -5.25215089e-01
-1.47603846e+00 -6.56662941e-01 5.30988812e-01 2.91572124e-01
-2.35133469e-01 8.29192460e-01 1.12118125e-01 -6.72509611e-01
1.57589927e-01 -9.65113938e-01 -1.25234127e+00 9.70418900e-02
-1.37735337e-01 9.90278482e-01 -2.76906379e-02 -2.83183992e-01
1.02441296e-01 -1.06490552e-01 -1.48043764e+00 9.76231933e-01
1.15048528e+00 1.02822542e+00 -2.19283625e-01 4.65016156e-01
-3.92148018e-01 1.08039308e+00 -6.68430865e-01 -5.93886971e-01
2.07671478e-01 -1.56857029e-01 -8.32934201e-01 -3.10569406e-01
5.02982318e-01 -4.10419255e-01 9.15219188e-02 8.05551529e-01
1.70331538e-01 -3.62877160e-01 3.53833109e-01 -1.18198287e+00
-6.32625759e-01 1.16607475e+00 -3.03354621e-01 -3.71889174e-01
2.14910150e-01 2.51710802e-01 1.38045990e+00 -8.34382057e-01
7.17666984e-01 9.47888434e-01 7.71243155e-01 4.81305450e-01
-1.41139686e+00 -6.41961098e-01 -3.64827752e-01 -1.57887608e-01
-1.13124907e+00 -3.97240847e-01 3.15617859e-01 -4.55036193e-01
1.45136416e+00 4.60042804e-01 1.53305888e-01 1.10651791e+00
1.75455675e-01 1.01404655e+00 1.17435789e+00 -2.86006361e-01
3.95456553e-02 3.97766620e-01 1.54538900e-01 7.34576821e-01
2.00121626e-01 -1.96983665e-01 -5.06541789e-01 -7.18333945e-02
1.53786168e-01 -3.99761915e-01 -2.11616326e-02 -3.07085335e-01
-1.32660639e+00 6.00359976e-01 2.23130301e-01 -1.51646044e-02
-2.41175517e-01 2.59176999e-01 8.00178587e-01 5.47634244e-01
2.75574267e-01 1.06778646e+00 -9.11714852e-01 -6.60465062e-01
-8.73320282e-01 3.80033076e-01 1.20679820e+00 1.14104879e+00
7.97226965e-01 2.15158746e-01 -1.11470133e-01 7.59841442e-01
-1.02321707e-01 7.93927312e-01 5.25618970e-01 -9.57496047e-01
1.10070646e+00 8.33116591e-01 2.11544652e-02 -4.11726624e-01
-1.88458711e-01 -3.32420766e-01 -5.87963045e-01 -2.06027068e-02
6.78423524e-01 -1.84402660e-01 -7.76608825e-01 1.54005098e+00
-4.54557464e-02 -4.78327066e-01 3.72307658e-01 4.59368080e-01
8.06939602e-01 9.17447031e-01 -2.09910527e-01 5.14787212e-02
1.27594018e+00 -9.19439137e-01 -4.51819479e-01 -3.04960459e-01
1.17521429e+00 -5.37284613e-01 1.90978944e+00 5.17098427e-01
-1.14761674e+00 -3.63221198e-01 -1.07444119e+00 -9.71934572e-02
-5.26223183e-01 2.01345026e-01 6.42458737e-01 4.83910948e-01
-8.00066531e-01 3.62548113e-01 -9.74363327e-01 -2.28969514e-01
4.70838785e-01 2.39099860e-01 -2.76275247e-01 -8.61959606e-02
-1.24093699e+00 6.04466915e-01 7.57819831e-01 -3.54405403e-01
-7.84035861e-01 -1.38929248e+00 -7.73799837e-01 1.84015453e-01
9.96200681e-01 -3.91129673e-01 1.57104886e+00 -5.15416145e-01
-1.27398086e+00 8.28606546e-01 1.36748001e-01 -6.75724566e-01
5.51961184e-01 -4.23030287e-01 -1.04370341e-01 -3.64936888e-01
1.71955496e-01 4.79232609e-01 -6.33592606e-02 -1.03480375e+00
-3.52087349e-01 -1.99606493e-01 -3.49557661e-02 1.73724105e-03
-3.56314629e-01 -1.68832690e-01 -7.11590946e-01 -4.16521013e-01
-4.69249159e-01 -7.98353910e-01 1.21311724e-01 -3.70966017e-01
-6.97673738e-01 -1.73220217e-01 3.14548671e-01 -7.04666555e-01
1.38483119e+00 -1.84204781e+00 -1.97723553e-01 1.16015412e-03
2.54463613e-01 3.07904780e-01 -1.05879977e-01 7.53488064e-01
-9.30381492e-02 4.06588256e-01 -3.48526746e-01 -1.72549218e-01
3.93071324e-01 5.98553568e-02 -5.95545590e-01 -1.92237556e-01
4.51192170e-01 1.15524733e+00 -7.08001733e-01 -3.21852863e-01
-2.24912986e-01 -1.92768946e-02 -8.09866190e-01 3.98072064e-01
-9.45834816e-01 -3.57930005e-01 -2.18651608e-01 6.91490412e-01
3.16707373e-01 -5.36561012e-01 3.18894088e-01 2.16842955e-03
-7.89765492e-02 9.20243204e-01 -9.06353056e-01 1.62139213e+00
-5.88910282e-01 4.57231134e-01 -3.88528913e-01 -8.02917778e-01
9.43384290e-01 1.00184090e-01 4.01941895e-01 -1.12124634e+00
-2.78633535e-01 5.16296268e-01 -8.35972726e-02 -4.39156890e-01
5.47127068e-01 2.73848027e-02 -5.28415501e-01 6.78112388e-01
-4.72666956e-02 -1.97435096e-01 5.04023194e-01 3.69583368e-01
1.26156592e+00 1.87623918e-01 2.87718296e-01 -2.34603480e-01
3.30435067e-01 4.84867096e-01 2.04804391e-01 7.86892712e-01
3.39989930e-01 5.67240953e-01 1.09287643e+00 -4.08041686e-01
-1.33010137e+00 -1.06423354e+00 -7.43408725e-02 1.08660793e+00
-6.32674515e-01 -9.72107708e-01 -9.41874444e-01 -8.10839117e-01
2.07372367e-01 1.08733869e+00 -7.30378449e-01 -7.69202933e-02
-7.38813758e-01 -5.87620497e-01 8.30414474e-01 7.52633512e-01
4.98734862e-01 -1.29651451e+00 -3.60810608e-01 2.45598793e-01
-2.93486208e-01 -1.16958499e+00 -3.88906509e-01 5.00076354e-01
-6.50994122e-01 -1.00481665e+00 -4.86497320e-02 -4.08849686e-01
2.53341526e-01 -2.66753167e-01 1.62052369e+00 6.67692721e-02
2.45650169e-02 -8.80521983e-02 -1.00728914e-01 -4.09684330e-01
-8.32480371e-01 5.95109224e-01 -2.06940547e-01 -4.51579630e-01
3.11264068e-01 -2.28441477e-01 1.09468989e-01 1.03954755e-01
-9.51343119e-01 4.03287649e-01 5.49743295e-01 9.15753782e-01
7.10620046e-01 -7.17079416e-02 5.85054278e-01 -1.57528758e+00
6.02629960e-01 -4.48345810e-01 -9.48728502e-01 4.47351187e-01
-1.06918824e+00 5.39974093e-01 1.19314981e+00 -8.27554241e-02
-8.81921530e-01 -2.93377757e-01 -1.27507329e-01 -1.34067431e-01
-4.43708040e-02 1.13008106e+00 -1.54116049e-01 7.55803764e-01
9.22957301e-01 2.97519684e-01 6.73933029e-02 -3.73486489e-01
3.27895731e-01 6.31464899e-01 6.26825333e-01 -9.44119573e-01
9.03237402e-01 3.48103493e-02 -2.84447283e-01 -1.71278059e-01
-1.05205715e+00 -1.68066442e-01 -3.69629562e-01 1.85849175e-01
4.95904088e-01 -8.77319157e-01 -7.02478349e-01 3.32745314e-01
-7.21696734e-01 -1.02251303e+00 -2.51037180e-01 3.50148045e-02
-7.02020168e-01 -1.65346414e-02 -5.97654164e-01 -4.34683055e-01
-6.55576468e-01 -1.19767201e+00 1.01223397e+00 -4.22246426e-01
-4.46741641e-01 -8.56199622e-01 1.81648210e-02 5.93518138e-01
4.05846715e-01 3.33767653e-01 1.18257904e+00 -9.42068398e-01
-7.30272532e-01 -1.19955815e-01 -2.91908532e-01 3.67835552e-01
2.81773265e-02 1.25927925e-01 -6.25879109e-01 -2.32660044e-02
-2.64217257e-01 -9.88215744e-01 6.30906105e-01 -2.54459321e-01
1.36901700e+00 -5.79507113e-01 -7.82147348e-02 6.45998120e-01
1.49144101e+00 9.15621072e-02 6.84828401e-01 6.48456931e-01
6.60671294e-01 5.00536740e-01 8.10009360e-01 4.24262702e-01
5.10888517e-01 5.46849608e-01 4.17526186e-01 2.27655590e-01
-1.53824324e-02 -5.42480171e-01 4.54675883e-01 7.85355330e-01
5.96743882e-01 -3.21423978e-01 -1.70158219e+00 5.93664229e-01
-1.77302659e+00 -7.72629499e-01 -1.34104908e-01 2.23104286e+00
1.55671740e+00 6.08387589e-01 1.31715417e-01 9.14897993e-02
6.62197843e-02 4.52987514e-02 -6.06246591e-01 -4.38256860e-01
7.06379935e-02 2.82017231e-01 3.30828607e-01 4.07310069e-01
-1.06940162e+00 1.00598669e+00 4.89989948e+00 7.54111588e-01
-1.21690881e+00 -2.93339580e-01 8.16792071e-01 -2.34091505e-01
-4.77715790e-01 -2.69183338e-01 -1.01915824e+00 5.40610075e-01
1.58757365e+00 -4.76184130e-01 6.87406480e-01 1.03703690e+00
-1.39903560e-01 1.46618992e-01 -1.42838264e+00 8.33770633e-01
-3.13348286e-02 -1.65502620e+00 1.97148863e-02 -7.59567507e-03
7.01693892e-01 4.01785970e-01 1.73081666e-01 9.95871007e-01
4.12586719e-01 -1.34302688e+00 7.92547882e-01 2.85893768e-01
1.16954505e+00 -7.02976823e-01 7.41027892e-01 5.21938562e-01
-9.33382571e-01 2.12584600e-01 -4.31842767e-02 7.18157366e-02
-3.10402095e-01 4.65062797e-01 -1.37491202e+00 4.28631991e-01
8.29556584e-01 7.04119623e-01 -9.53114748e-01 4.52832043e-01
-7.41900550e-03 9.11121607e-01 -3.54495794e-01 -4.43989307e-01
3.67944330e-01 4.01756391e-02 -9.12862048e-02 1.36638010e+00
5.87544851e-02 -2.61698157e-01 3.23961452e-02 1.13734448e+00
-6.74867034e-01 2.46254966e-01 -5.91223598e-01 -3.64016861e-01
3.65693152e-01 9.99081433e-01 -5.90428747e-02 -6.31524801e-01
-4.92355883e-01 3.58972877e-01 7.30997264e-01 1.44867226e-01
-1.03503513e+00 -5.63449442e-01 4.34613466e-01 2.89276928e-01
2.66259134e-01 -1.80655234e-02 -6.95891500e-01 -1.34571791e+00
5.86700857e-01 -1.55891931e+00 4.84026611e-01 -8.71474385e-01
-1.07537198e+00 7.04640269e-01 1.85168043e-01 -9.55082893e-01
-5.61939538e-01 -6.77390099e-01 -1.26189783e-01 8.49165976e-01
-1.29047453e+00 -8.15147340e-01 -2.85842270e-01 1.13841288e-01
3.23053151e-01 -3.58284235e-01 9.17828739e-01 2.76161790e-01
-5.42794526e-01 1.00751972e+00 4.03270930e-01 4.01633054e-01
8.75585914e-01 -1.71889448e+00 7.05764651e-01 5.55803120e-01
-7.17003122e-02 8.79887760e-01 5.30793786e-01 -6.28662765e-01
-1.86818552e+00 -1.47904503e+00 9.89622474e-01 -9.40967202e-01
1.03365576e+00 -8.43104064e-01 -1.31634259e+00 9.69622850e-01
2.54479527e-01 -2.02792749e-01 6.10065520e-01 4.15113419e-01
-9.03186262e-01 -5.54139495e-01 -8.66919577e-01 7.15788662e-01
6.16239727e-01 -5.94796062e-01 -4.20707434e-01 5.46942711e-01
8.24042857e-01 -7.47835696e-01 -1.27783430e+00 5.91211259e-01
5.86646318e-01 -7.60271192e-01 6.96254671e-01 -8.56140733e-01
1.08465934e+00 -4.44111861e-02 -3.89203668e-01 -1.13113809e+00
2.16781855e-01 -5.82116842e-01 -2.97168016e-01 1.37985265e+00
9.61651504e-01 -6.07463419e-01 1.05080140e+00 5.74686706e-01
-3.16911638e-01 -1.17403042e+00 -4.23247516e-01 -8.89043808e-01
4.42908257e-01 -6.98097289e-01 7.44595647e-01 9.37094450e-01
1.93033367e-01 5.78429937e-01 -1.98134974e-01 -1.27913028e-01
3.82978559e-01 3.77190083e-01 1.22097993e+00 -7.58750319e-01
-5.73114812e-01 -3.99844766e-01 2.49850482e-01 -8.96170080e-01
2.26118878e-01 -1.23029757e+00 4.53131013e-02 -1.40524268e+00
2.75595784e-01 -4.59549755e-01 3.89291272e-02 9.16173160e-01
-2.07522333e-01 -5.79906181e-02 1.86949447e-01 1.53486356e-01
-7.86338925e-01 4.66051430e-01 9.44660723e-01 -2.70204872e-01
-1.09025002e-01 -2.86582977e-01 -8.94203663e-01 1.72893643e-01
1.17835391e+00 -5.38788557e-01 -4.88201678e-01 -5.18333375e-01
5.53246737e-01 3.81788850e-01 -2.93434691e-02 -1.07955205e+00
-2.36465689e-02 -1.98419675e-01 -4.70557511e-02 -7.25017726e-01
4.52146567e-02 -4.16846156e-01 1.26953915e-01 4.71347839e-01
-6.49886727e-01 3.59951019e-01 6.57096088e-01 8.05330276e-02
-3.04713964e-01 -8.59924033e-02 5.56038737e-01 -1.78188577e-01
-4.35623735e-01 9.82088447e-02 -9.59162787e-02 7.53392160e-01
6.73287392e-01 3.02771837e-01 -7.68069804e-01 -3.04585218e-01
-1.12606645e-01 4.20394599e-01 4.55883861e-01 3.96194786e-01
3.51468295e-01 -1.08003056e+00 -8.26248527e-01 2.96952903e-01
4.84227359e-01 1.35687232e-01 -3.84733588e-01 5.48408210e-01
-8.93525600e-01 8.70442331e-01 -6.94143251e-02 -5.34880042e-01
-1.03474736e+00 5.66197872e-01 4.02373701e-01 -7.64430046e-01
-2.81920910e-01 4.19955820e-01 -6.28113374e-02 -1.04798663e+00
3.12145144e-01 -7.83741295e-01 4.51348335e-01 -3.41190606e-01
3.52130175e-01 2.00488567e-02 4.71628785e-01 1.11923680e-01
-2.54609585e-01 4.08988968e-02 -3.41444194e-01 -3.00040320e-02
1.40339506e+00 6.59259021e-01 -2.33026594e-01 6.78629637e-01
1.39299893e+00 1.23950876e-01 -9.50080872e-01 -2.98160553e-01
1.76038653e-01 -8.11529234e-02 -5.36346138e-01 -1.45793974e+00
-8.51389289e-01 9.35166121e-01 -1.95016727e-01 2.70405024e-01
9.82075036e-01 -5.84575087e-02 7.94228554e-01 9.97006238e-01
3.50313753e-01 -7.02251196e-01 2.73408264e-01 8.42947900e-01
1.05978084e+00 -1.32747447e+00 -7.80882463e-02 -1.91392720e-01
-7.15642989e-01 1.06900907e+00 1.06000996e+00 1.14449464e-01
3.34860176e-01 6.86664760e-01 1.35958806e-01 -1.14612862e-01
-1.46529627e+00 3.41673523e-01 1.71746626e-01 2.86171883e-01
7.45024085e-01 7.31373876e-02 1.01892516e-01 7.17660666e-01
-6.94463074e-01 5.39138280e-02 6.68469012e-01 7.71601617e-01
-4.12586890e-02 -1.24977636e+00 -1.33762345e-01 8.20205629e-01
-5.22934318e-01 -5.29296339e-01 -3.56483400e-01 1.17577887e+00
-4.74735767e-01 5.30782700e-01 -7.91051313e-02 -4.92782921e-01
5.08398235e-01 3.66853744e-01 1.80757567e-01 -8.92834246e-01
-8.69518518e-01 -1.86895177e-01 6.16341233e-01 -6.21150672e-01
3.11648548e-01 -6.64246976e-01 -1.42571008e+00 -6.44463360e-01
1.72452688e-01 3.36563706e-01 4.72156763e-01 3.75050187e-01
4.88266319e-01 6.43160403e-01 1.25275567e-01 1.22479901e-01
-1.01314652e+00 -8.22551608e-01 -1.75912276e-01 4.97861385e-01
1.37205362e-01 -1.28411502e-01 -1.30511940e-01 9.87193808e-02] | [9.896820068359375, 7.862509727478027] |
5589fe70-efb2-4f86-9ebf-4c35365a8b67 | interpreting-bert-architecture-predictions | 2111.07137 | null | https://arxiv.org/abs/2111.07137v1 | https://arxiv.org/pdf/2111.07137v1.pdf | Interpreting BERT architecture predictions for peptide presentation by MHC class I proteins | The major histocompatibility complex (MHC) class-I pathway supports the detection of cancer and viruses by the immune system. It presents parts of proteins (peptides) from inside a cell on its membrane surface enabling visiting immune cells that detect non-self peptides to terminate the cell. The ability to predict whether a peptide will get presented on MHC Class I molecules helps in designing vaccines so they can activate the immune system to destroy the invading disease protein. We designed a prediction model using a BERT-based architecture (ImmunoBERT) that takes as input a peptide and its surrounding regions (N and C-terminals) along with a set of MHC class I (MHC-I) molecules. We present a novel application of well known interpretability techniques, SHAP and LIME, to this domain and we use these results along with 3D structure visualizations and amino acid frequencies to understand and identify the most influential parts of the input amino acid sequences contributing to the output. In particular, we find that amino acids close to the peptides' N- and C-terminals are highly relevant. Additionally, some positions within the MHC proteins (in particular in the A, B and F pockets) are often assigned a high importance ranking - which confirms biological studies and the distances in the structure visualizations. | ['Ajitha Rajan', 'Javier Alfaro', 'David Goodlett', 'Bo Ren', 'Georges Bedran', 'Hans-Christof Gasser'] | 2021-11-13 | null | null | null | null | ['mhc-presentation-prediction'] | ['medical'] | [ 4.12664711e-01 3.82457636e-02 -3.98848683e-01 -2.45992154e-01
-3.73910926e-02 -8.57053280e-01 2.08674535e-01 7.97326148e-01
-2.03270584e-01 1.01543880e+00 1.97780758e-01 -5.81046045e-01
-1.28534600e-01 -7.40900517e-01 -4.34818119e-01 -9.62407529e-01
-4.92315888e-01 6.66822314e-01 2.57297754e-01 -4.59093601e-01
5.70249259e-01 8.03601146e-01 -1.56916213e+00 8.35316539e-01
8.30418706e-01 4.43301588e-01 3.20462435e-01 4.36930567e-01
-6.78550541e-01 4.77033667e-02 -6.92039490e-01 3.35625261e-02
-1.17766872e-01 -3.91769379e-01 -3.38398546e-01 -7.76151121e-01
-1.34574518e-01 3.09715182e-01 7.31759250e-01 8.89716625e-01
2.95430720e-01 -3.58131558e-01 1.08445001e+00 -9.04983997e-01
-2.73986220e-01 -1.25563413e-01 -5.54657340e-01 -1.65000111e-02
7.97997892e-01 7.06964806e-02 7.75588036e-01 -1.05199516e+00
1.53119254e+00 1.46968758e+00 5.94756007e-01 4.90780354e-01
-1.14153159e+00 -2.16078356e-01 3.68590727e-02 2.05117226e-01
-1.05617571e+00 3.71349722e-01 4.51998889e-01 -7.53520906e-01
1.48646462e+00 8.22602034e-01 8.65235686e-01 7.43112087e-01
1.05434144e+00 5.05866826e-01 1.31198812e+00 -6.01075530e-01
4.23538625e-01 4.03792858e-02 4.40791667e-01 5.42840660e-01
-1.04518972e-01 3.66985172e-01 -6.30399346e-01 -8.60118210e-01
2.55350292e-01 2.28411332e-01 -8.63776449e-03 -8.13619077e-01
-1.04435408e+00 8.69800329e-01 3.86498213e-01 1.14176407e-01
-5.44193447e-01 -6.36974454e-01 3.42637479e-01 1.76448971e-01
2.19205335e-01 2.46086091e-01 -5.84147811e-01 1.69592038e-01
-3.74865383e-01 3.81032437e-01 7.26519763e-01 1.88366994e-01
8.36073816e-01 -8.52752209e-01 6.97241165e-03 6.57851815e-01
3.21577817e-01 4.11868304e-01 9.30652022e-02 5.79930916e-02
-2.01609820e-01 1.08953476e+00 1.04229443e-01 -8.63251925e-01
-8.88556838e-01 -2.91814238e-01 -4.69103307e-01 9.37602997e-01
3.82966548e-01 1.58033207e-01 -8.34860325e-01 1.46497333e+00
7.54663765e-01 -6.03936851e-01 1.88631326e-01 1.14344060e+00
4.75561649e-01 7.08735347e-01 6.05931103e-01 -2.71295786e-01
1.80243325e+00 -4.89100426e-01 -5.47789991e-01 2.33845651e-01
7.41620362e-01 -9.15186048e-01 9.02682304e-01 2.22456858e-01
-4.80865479e-01 -4.26504850e-01 -1.21757305e+00 2.61921674e-01
-9.21219170e-01 -2.85983831e-01 5.40450335e-01 2.89325655e-01
-6.53675377e-01 7.22081006e-01 -6.09415710e-01 -6.60803854e-01
6.55124485e-02 3.32468033e-01 -5.07533014e-01 2.46168226e-01
-1.43615818e+00 1.29710662e+00 4.19921994e-01 -4.76871639e-01
-4.93120193e-01 -8.60230923e-01 -4.97537076e-01 -2.97344476e-01
-2.41633177e-01 -6.91343248e-01 3.08266819e-01 -1.13313901e+00
-1.05843401e+00 1.13237274e+00 -1.10506974e-01 -1.60452306e-01
2.03258842e-01 -9.54337511e-03 -4.64424938e-01 7.59042874e-02
-2.94462204e-01 9.50754225e-01 2.80163318e-01 -1.00383019e+00
-6.93810940e-01 -7.99881220e-01 -3.26460540e-01 2.59485424e-01
4.25453961e-01 2.92554051e-01 2.23933414e-01 -8.26270998e-01
1.51575163e-01 -7.97859848e-01 -4.87377435e-01 2.61143595e-01
-4.64821935e-01 -1.83041826e-01 7.79210210e-01 -6.07743263e-01
9.23102140e-01 -1.77163589e+00 2.77062833e-01 8.48648548e-01
1.63582284e-02 3.44804585e-01 -1.39174052e-02 1.16069281e+00
-6.17250204e-01 7.16974288e-02 -7.31590912e-02 9.46841598e-01
-1.79668859e-01 6.34985343e-02 -2.29436442e-01 4.95111555e-01
3.53956640e-01 6.63669884e-01 -7.01420128e-01 -1.97396651e-01
1.76107809e-01 8.22110295e-01 -4.11618203e-01 1.30795240e-01
-5.71800411e-01 3.75629872e-01 -3.87864351e-01 6.91646278e-01
1.00375128e+00 -3.37438360e-02 8.85278761e-01 -1.63950726e-01
-3.19586128e-01 3.11231524e-01 -7.45994747e-01 9.82229769e-01
4.37593341e-01 2.44446248e-01 -2.17581447e-03 -3.40204686e-01
1.10737133e+00 -2.53923200e-02 1.29317105e-01 -9.09335375e-01
-4.35535640e-01 8.55061486e-02 2.49179393e-01 -2.42170990e-01
-2.39993513e-01 -2.58815765e-01 2.65295029e-01 7.52812326e-01
-5.06224394e-01 8.38569343e-01 8.70108828e-02 4.15920079e-01
8.69136214e-01 4.95558113e-01 7.66023159e-01 -8.04193497e-01
6.90871477e-01 5.03992856e-01 5.44326127e-01 1.86249837e-01
1.26816913e-01 2.80531589e-02 7.59476781e-01 -1.06187117e+00
-1.21977389e+00 -1.11487734e+00 -2.59728670e-01 1.31157577e+00
3.15280654e-03 -3.08069199e-01 -8.90558898e-01 -5.73970437e-01
3.43199909e-01 4.48712170e-01 -1.02432036e+00 1.41178116e-01
-4.36504215e-01 -6.56251729e-01 3.93491872e-02 1.34994254e-01
-4.99957472e-01 -1.25050128e+00 -9.72122312e-01 3.17561656e-01
2.86231339e-01 2.04874814e-01 6.96854442e-02 6.35163903e-01
-8.31168532e-01 -1.20576513e+00 -7.56265104e-01 -7.70977020e-01
9.50186074e-01 -3.50470096e-02 1.02586138e+00 1.81446999e-01
-8.20831299e-01 -3.45729768e-01 -1.87529013e-01 -6.94111288e-01
-6.52951956e-01 -3.25287640e-01 1.66926771e-01 -4.66716528e-01
7.65369892e-01 -3.57110441e-01 -7.25567818e-01 5.20563066e-01
-9.20147181e-01 4.23215628e-01 6.10560238e-01 8.27932298e-01
1.00581110e+00 -5.54156780e-01 3.64156216e-01 -1.18024409e+00
3.97501498e-01 -4.38577235e-01 -4.06976551e-01 5.77828646e-01
-3.28779787e-01 4.07906085e-01 4.28785294e-01 -3.18032205e-01
-7.36198604e-01 3.43395829e-01 -1.95722148e-01 1.78618178e-01
-2.65430093e-01 4.38851267e-01 -3.95710051e-01 3.23808305e-02
7.11013436e-01 2.14139625e-01 6.47210300e-01 -5.16605616e-01
1.15785941e-01 2.96002954e-01 5.31421117e-02 -3.00729007e-01
4.53640610e-01 4.60400552e-01 1.92864269e-01 -7.26889551e-01
-1.59748986e-01 -3.33103687e-01 -7.71741331e-01 -1.85046598e-01
7.09964633e-01 -3.65636379e-01 -1.09306777e+00 1.39954472e-02
-1.02758014e+00 3.08946297e-02 2.87371010e-01 2.33733475e-01
-3.89213055e-01 5.46313047e-01 -4.99900162e-01 -6.27141416e-01
-5.09611726e-01 -9.57729399e-01 7.72801816e-01 1.96064934e-01
-7.25536466e-01 -1.00488532e+00 7.13683605e-01 -1.99289501e-01
3.74610662e-01 3.33655983e-01 1.93049324e+00 -9.82500315e-01
-2.09984079e-01 4.35054541e-01 4.63321107e-03 -5.08114100e-01
-1.24275915e-01 3.60755682e-01 -7.24664211e-01 -2.19068512e-01
-4.73510116e-01 1.07827581e-01 6.71045005e-01 2.78353959e-01
3.76739025e-01 -2.77280599e-01 -1.06784832e+00 3.86685282e-01
1.34300828e+00 6.01480067e-01 7.20886588e-01 4.99847889e-01
6.83920532e-02 1.44792008e+00 1.24532902e+00 2.87156671e-01
-5.37467778e-01 9.30258512e-01 4.98212725e-01 -6.43616915e-01
3.72246742e-01 -9.33434069e-02 3.34949285e-01 2.03518078e-01
1.27968952e-01 -2.54215486e-02 -9.63299990e-01 -3.15166004e-02
-1.59112680e+00 -7.44373500e-01 -1.97420657e-01 2.29732633e+00
1.07146704e+00 1.48287147e-01 2.89781272e-01 -1.95144877e-01
6.36974454e-01 -3.08765382e-01 -6.72379136e-01 -7.20215082e-01
-4.48584527e-01 3.75380844e-01 3.61067116e-01 7.59441614e-01
-9.18354869e-01 4.67781425e-01 6.99053335e+00 8.89369071e-01
-1.12625003e+00 -6.56182885e-01 6.41482830e-01 -5.33807352e-02
-7.67331898e-01 2.41042897e-01 -6.55846894e-01 3.94663155e-01
8.28736901e-01 1.09357610e-01 -3.90421785e-02 7.64022648e-01
1.71142206e-01 -1.48464501e-01 -1.04461527e+00 6.71458542e-01
-4.77916390e-01 -1.71296144e+00 3.39065969e-01 3.20841819e-01
1.30855858e-01 -2.09805191e-01 -7.21196234e-02 -2.55955011e-01
8.93599764e-02 -1.13625264e+00 3.82084638e-01 6.22549832e-01
7.19929278e-01 -1.08076060e+00 5.46683371e-01 2.00330853e-01
-8.03895652e-01 3.60368252e-01 -6.18414700e-01 7.72089465e-03
8.44914757e-04 6.32768989e-01 -1.24967968e+00 1.57669231e-01
4.49563861e-01 2.31936604e-01 -4.88969088e-01 6.43176675e-01
-7.13726645e-03 1.22340992e-01 -1.12755284e-01 -3.11874866e-01
1.03641279e-01 -3.87102842e-01 6.74855769e-01 1.38647032e+00
-1.56862605e-02 2.34883472e-01 9.99501422e-02 6.08534038e-01
6.26204193e-01 6.46070242e-01 -6.80007756e-01 1.25752762e-01
1.97713196e-01 1.29890525e+00 -8.22545707e-01 -2.95750767e-01
-1.24826387e-01 7.33752847e-01 2.53966331e-01 3.19776833e-01
-3.95321339e-01 -4.43185180e-01 1.39417100e+00 3.50698441e-01
1.19555376e-01 2.54632860e-01 1.91390216e-01 -1.78451315e-01
-4.49328780e-01 -1.17374694e+00 7.18813539e-01 -7.27230549e-01
-1.24429846e+00 5.61647058e-01 -5.58735549e-01 -1.09474492e+00
-2.11468086e-01 -1.11390162e+00 -4.06268388e-01 1.31145036e+00
-1.23583364e+00 -9.15905714e-01 3.55823129e-01 2.09565803e-01
7.23376218e-03 -1.62175879e-01 1.41468322e+00 -2.33144820e-01
-1.49691090e-01 2.73196429e-01 4.90030795e-01 -4.38485622e-01
8.29540074e-01 -1.23951447e+00 5.36820054e-01 2.12284382e-02
-2.94640839e-01 1.13096726e+00 9.98816192e-01 -1.29487145e+00
-1.26139700e+00 -6.01553679e-01 9.89273667e-01 -4.40906912e-01
3.24525833e-01 -5.59022307e-01 -1.17344487e+00 8.05378407e-02
1.62765175e-01 -6.72697842e-01 1.60491633e+00 5.73620498e-02
-4.18932408e-01 4.70227689e-01 -1.23742890e+00 6.78538024e-01
5.99415362e-01 -2.88250417e-01 -5.19719899e-01 5.07579327e-01
4.25998241e-01 -1.43051982e-01 -7.17219770e-01 3.36159110e-01
1.05955470e+00 -1.18455231e+00 1.11833155e+00 -1.25821483e+00
-3.34409364e-02 -5.67598462e-01 1.00286804e-01 -1.09673595e+00
-4.04754251e-01 -3.63649905e-01 2.97196865e-01 3.81361663e-01
6.61571860e-01 -6.13774121e-01 1.00806355e+00 6.48200065e-02
2.33880147e-01 -8.72505248e-01 -9.73183334e-01 -2.59476960e-01
-6.09374493e-02 2.24759504e-01 5.17954946e-01 7.94530213e-01
5.05499542e-01 1.78112701e-01 -3.00741419e-02 -1.31955668e-01
5.17123878e-01 5.47693312e-01 5.22241235e-01 -1.39155018e+00
-8.14879686e-02 -3.82191539e-01 -3.90478432e-01 -4.17616278e-01
-5.19631326e-01 -1.00211954e+00 -4.88434315e-01 -1.42822409e+00
4.07400608e-01 -5.35768986e-01 -5.51847160e-01 4.34923053e-01
-8.50225762e-02 -1.05631717e-01 2.96439938e-02 3.65010530e-01
-2.59275645e-01 -7.40671158e-02 9.54168499e-01 -2.73726936e-02
-1.25434920e-01 -2.63676941e-01 -3.48730266e-01 6.53296113e-01
6.04136527e-01 -5.56215525e-01 -7.35443691e-03 3.67593586e-01
5.01322091e-01 -8.25532898e-02 -2.62358468e-02 -4.88124341e-01
-1.86470628e-01 -4.72070932e-01 9.25273597e-01 -1.19150209e+00
1.06435142e-01 -7.52939880e-01 8.08341205e-01 1.42596030e+00
-2.74010390e-01 3.66447806e-01 3.04152519e-01 4.80205208e-01
1.15261167e-01 8.50708112e-02 7.97512412e-01 -8.29051509e-02
-6.03659272e-01 -3.86480421e-01 -1.01046550e+00 -5.90145111e-01
1.25465798e+00 -5.34634471e-01 -6.03950560e-01 1.37724876e-01
-9.60464835e-01 -1.42726526e-02 1.05710304e+00 3.58112425e-01
6.79978251e-01 -1.03509128e+00 -5.84428787e-01 7.39972353e-01
4.52578217e-01 -1.00848937e+00 3.61411035e-01 5.77520490e-01
-1.13983297e+00 7.76525915e-01 -1.04687619e+00 -7.67667651e-01
-1.79611301e+00 1.01450312e+00 2.32570663e-01 -3.19030106e-01
-3.71167094e-01 6.41309619e-01 7.50525236e-01 -6.33872688e-01
1.11688666e-01 1.78072359e-02 -6.39252841e-01 1.09466843e-01
9.46283281e-01 1.70225799e-01 -3.35890912e-02 -6.03799999e-01
-1.03543937e+00 6.48210108e-01 -3.90294254e-01 3.30979079e-01
1.16222227e+00 3.56901139e-01 -4.16443974e-01 -2.82664932e-02
9.21767056e-01 2.57140189e-01 -8.76181424e-01 -7.81430230e-02
4.36872214e-01 -4.00934339e-01 -8.87851894e-01 -1.51747024e+00
-3.51603240e-01 9.07564878e-01 1.19840074e+00 -1.18937798e-01
8.41062903e-01 -3.01632192e-02 4.26677495e-01 1.35648236e-01
4.92087692e-01 -1.03540635e+00 -2.54142582e-01 2.44666338e-01
7.52802789e-01 -7.46412754e-01 5.34637347e-02 -4.66930658e-01
-8.63972247e-01 1.24564016e+00 2.59600967e-01 3.33110213e-01
2.52735943e-01 4.54288930e-01 7.96556473e-01 -5.70478261e-01
-1.28855431e+00 2.75977731e-01 3.26747328e-01 9.84895349e-01
7.67440200e-01 3.03685337e-01 -9.14379537e-01 2.43875712e-01
2.86428601e-01 -6.11504257e-01 -2.34074503e-01 1.04283893e+00
-9.64360058e-01 -1.75068629e+00 -5.88148236e-01 3.99413966e-02
-3.39148521e-01 -4.70607467e-02 -1.11437428e+00 6.40842617e-01
4.01823148e-02 3.66120845e-01 1.69595778e-01 -1.39497593e-01
2.15814248e-01 3.25960129e-01 6.85059726e-02 -2.57719278e-01
-9.24258113e-01 4.47465926e-01 -5.81953004e-02 -6.02547646e-01
-2.28939980e-01 -3.67440134e-01 -1.81166613e+00 -2.21823081e-01
7.69877993e-03 5.29912353e-01 9.39780891e-01 3.23127270e-01
7.99500644e-01 1.44138053e-01 3.52072328e-01 -3.87529969e-01
-1.61605537e-01 -6.62669599e-01 -6.44869089e-01 4.73003834e-01
-1.02671459e-01 -6.11092746e-01 8.06266814e-02 -2.22256824e-01] | [4.74350643157959, 5.409682750701904] |
b5e93ccf-b1cb-457f-998a-035f40b7442a | the-truth-and-nothing-but-the-truth | 1903.04484 | null | http://arxiv.org/abs/1903.04484v1 | http://arxiv.org/pdf/1903.04484v1.pdf | The Truth and Nothing but the Truth: Multimodal Analysis for Deception Detection | We propose a data-driven method for automatic deception detection in
real-life trial data using visual and verbal cues. Using OpenFace with facial
action unit recognition, we analyze the movement of facial features of the
witness when posed with questions and the acoustic patterns using OpenSmile. We
then perform a lexical analysis on the spoken words, emphasizing the use of
pauses and utterance breaks, feeding that to a Support Vector Machine to test
deceit or truth prediction. We then try out a method to incorporate
utterance-based fusion of visual and lexical analysis, using string based
matching. | ['Sairam Tabibu', 'Rajiv Bajpai', 'Mimansa Jaiswal'] | 2019-03-11 | null | null | null | null | ['facial-action-unit-detection', 'deception-detection', 'lexical-analysis'] | ['computer-vision', 'miscellaneous', 'natural-language-processing'] | [ 9.51646641e-02 -4.78806235e-02 2.66887303e-02 -9.36055183e-01
-9.20628071e-01 -5.18480718e-01 4.78590488e-01 7.41139054e-02
-3.35844189e-01 2.43542299e-01 4.25231963e-01 -1.38213737e-02
1.92666307e-01 -7.99002200e-02 -2.28377879e-01 -4.54594761e-01
-1.71027556e-01 -2.37225682e-01 -1.89996734e-01 -1.90662786e-01
4.43976402e-01 6.29870534e-01 -1.87023616e+00 6.43934965e-01
1.32565275e-01 1.37432182e+00 -8.35232079e-01 8.09962094e-01
3.18399407e-02 6.93130791e-01 -6.91380441e-01 -6.01148844e-01
9.34550464e-02 -3.81817967e-01 -5.17808795e-01 2.35227585e-01
7.29422271e-01 -5.74267268e-01 -1.78171024e-01 1.08013034e+00
6.71517909e-01 2.18633980e-01 5.56227863e-01 -1.37913346e+00
-3.71452957e-01 2.24432379e-01 -4.23063606e-01 3.29389811e-01
1.38846529e+00 3.92598547e-02 5.11844397e-01 -1.26992798e+00
6.48350120e-01 1.39891946e+00 6.16825223e-01 8.66364002e-01
-9.25625026e-01 -8.69571388e-01 -3.13474864e-01 7.49885023e-01
-1.61624539e+00 -1.23363948e+00 1.07169652e+00 -4.36342537e-01
9.25541878e-01 4.87796456e-01 5.58937013e-01 1.56733096e+00
3.43035012e-02 5.79668105e-01 9.82943833e-01 -5.82324564e-01
2.47567236e-01 2.48654112e-01 3.20926696e-01 1.01615238e+00
-6.24423742e-01 2.83680558e-01 -1.07765114e+00 -6.83271289e-01
-6.52107969e-02 -9.65890959e-02 -6.48350939e-02 1.61528960e-01
-5.38011432e-01 8.85673285e-01 -1.84728026e-01 2.50900537e-01
-5.65457582e-01 -3.70924547e-02 7.34770060e-01 4.73016590e-01
5.37247717e-01 -3.24859619e-01 2.24938542e-01 -3.26734513e-01
-1.22135448e+00 -7.65225012e-03 8.51676762e-01 2.32735902e-01
4.11875159e-01 1.84718370e-01 1.47283003e-01 7.68027306e-01
5.78448176e-01 5.04865766e-01 7.84553468e-01 -8.74899447e-01
-9.93731394e-02 3.31309199e-01 -2.94062257e-01 -1.41306472e+00
-3.91790301e-01 8.11165154e-01 -1.52556792e-01 5.10579765e-01
1.32498503e-01 -1.54848829e-01 -7.63137102e-01 1.39699423e+00
5.95594645e-01 2.11347342e-01 -7.30266571e-02 1.00556552e+00
1.18064582e+00 2.94568568e-01 5.59480973e-02 -5.95795631e-01
1.49552131e+00 -3.83251995e-01 -1.26997125e+00 2.68384397e-01
5.21242142e-01 -5.64754844e-01 6.52329028e-01 7.35637903e-01
-8.56859148e-01 -4.44589555e-01 -1.05312479e+00 1.96014568e-01
-4.00055051e-01 -8.73774663e-02 1.86494067e-01 1.12340450e+00
-6.77891791e-01 4.28849459e-01 -5.01755834e-01 -4.06610519e-01
3.59005332e-01 3.03595573e-01 -9.24746633e-01 3.40121955e-01
-1.08852744e+00 8.64336193e-01 -1.20078795e-01 2.54120260e-01
-8.22734177e-01 -1.01834990e-01 -1.30817425e+00 -4.05471534e-01
1.65990457e-01 1.32115513e-01 9.66543734e-01 -1.38341975e+00
-1.90976107e+00 1.21090460e+00 -4.27693605e-01 -3.15972209e-01
1.45409122e-01 1.19237170e-01 -1.06720674e+00 6.01771176e-01
-4.90664780e-01 4.34448838e-01 1.66745496e+00 -7.58272350e-01
-6.12730607e-02 -8.88679504e-01 -3.42703283e-01 -1.12836426e-02
-2.12163419e-01 6.67881906e-01 3.17041188e-01 -3.74697506e-01
9.28804502e-02 -5.59946835e-01 5.04204571e-01 1.20238245e-01
-2.19110936e-01 -5.80454171e-01 1.04221225e+00 -9.70215619e-01
1.25142121e+00 -2.63073158e+00 -8.61535370e-02 4.86201882e-01
1.80144161e-01 2.73489356e-01 -2.09529847e-01 3.39589179e-01
-4.21773344e-01 -8.85315612e-02 2.51454175e-01 -6.86009467e-01
-4.56579104e-02 1.12991348e-01 -2.89635658e-01 9.35567141e-01
-1.47491112e-01 6.69705927e-01 -3.02309334e-01 -6.74089909e-01
3.14049214e-01 4.12518591e-01 -3.86859864e-01 3.23899150e-01
3.23474616e-01 -1.79474458e-01 -1.09845087e-01 1.12869430e+00
5.26321292e-01 8.76183033e-01 -1.55144528e-01 -1.68808982e-01
-4.00371961e-02 -1.94090039e-01 -1.07727575e+00 1.60734987e+00
2.29280461e-02 9.26594257e-01 5.67304432e-01 -8.28370690e-01
1.17110455e+00 5.10188222e-01 -1.23318210e-02 -5.71638584e-01
7.19436526e-01 -2.86028218e-02 -2.86833614e-01 -1.35135710e+00
1.06684364e-01 -3.97939801e-01 -2.15846509e-01 3.50725472e-01
2.73293793e-01 7.64697492e-02 -4.83977437e-01 1.63709018e-02
9.02290583e-01 1.41263697e-02 3.52887690e-01 2.43478790e-01
6.12527788e-01 -4.13505912e-01 1.35985643e-01 1.88476115e-01
-7.03166425e-01 5.83019674e-01 4.04929280e-01 -6.21366858e-01
-2.70378590e-01 -7.42570937e-01 -2.89541725e-02 1.19254553e+00
-2.96066225e-01 -3.04702848e-01 -1.05900729e+00 -6.77029967e-01
1.65491417e-01 8.18166673e-01 -1.04596806e+00 -4.02663857e-01
1.27219427e-02 -1.37616947e-01 8.35084081e-01 7.00311437e-02
-2.03967914e-01 -1.43010116e+00 -7.51393855e-01 -6.77030161e-02
4.01538238e-02 -9.67528105e-01 -4.98911142e-01 -5.24522066e-02
-1.32211834e-01 -1.06814420e+00 -6.57546595e-02 -4.63413984e-01
1.76018775e-01 -2.74839908e-01 3.18248540e-01 2.22173370e-02
-9.27423477e-01 9.42074716e-01 -5.19263685e-01 -5.71768999e-01
-4.58806574e-01 -9.64001656e-01 6.35957956e-01 8.69803429e-01
8.83620918e-01 -5.12811542e-01 -3.39368373e-01 1.47973523e-01
-6.37899280e-01 -7.47125924e-01 -1.47337601e-01 5.46338737e-01
7.13883415e-02 -7.26006866e-01 3.65627199e-01 6.46658847e-03
1.07282138e+00 -4.52550411e-01 -9.36104134e-02 1.93630278e-01
-1.12638570e-01 -3.65139842e-01 -1.99456085e-02 -7.03788638e-01
-6.35655701e-01 1.62555233e-01 -2.85412312e-01 -1.03578401e+00
-5.80484271e-01 2.39571437e-01 -1.42332003e-01 -3.01298231e-01
8.66402268e-01 2.55270451e-01 6.41147733e-01 -1.95598379e-01
2.64125675e-01 1.33694506e+00 7.56034315e-01 -2.49659382e-02
1.98190138e-01 4.87755239e-01 -5.34283280e-01 -1.33186948e+00
-4.10663843e-01 -4.00110304e-01 -7.29975224e-01 -8.27453613e-01
9.22385633e-01 -4.20875192e-01 -1.44540250e+00 4.39236462e-01
-1.35423267e+00 3.19687933e-01 -2.03997687e-01 4.46231484e-01
-5.20182908e-01 8.55075300e-01 -2.86083847e-01 -1.49623954e+00
-5.35619140e-01 -6.96586668e-01 1.02071357e+00 -8.57319310e-02
-6.28814101e-01 -5.83103716e-01 2.75493771e-01 7.81974018e-01
-1.66017264e-01 4.08944488e-01 3.26684684e-01 -1.30820513e+00
5.40948451e-01 -6.58999324e-01 1.99276298e-01 6.90101445e-01
-7.53860269e-03 3.44453633e-01 -1.25565600e+00 1.48542911e-01
3.07505608e-01 -7.99347043e-01 4.28100616e-01 1.18955187e-01
9.92856801e-01 -4.11649525e-01 -8.26007128e-02 4.53665048e-01
8.17075968e-01 3.14357281e-01 5.71692944e-01 -3.93737018e-01
2.11140588e-01 1.03425944e+00 3.76066118e-01 7.94505715e-01
6.05182312e-02 8.29503238e-01 6.03446662e-01 3.36613774e-01
1.61071151e-01 -2.00313285e-01 6.47520006e-01 3.55326980e-01
7.23987296e-02 4.61214781e-02 -8.28491211e-01 3.89467895e-01
-1.54368818e+00 -1.36530149e+00 -2.56577600e-02 1.81376910e+00
6.62701845e-01 -2.45254308e-01 2.82742292e-01 3.90749395e-01
5.33043385e-01 4.54735868e-02 -2.68249303e-01 -1.25922036e+00
5.92145249e-02 1.38574064e-01 -2.21525371e-01 7.36046135e-01
-9.35949326e-01 6.81083143e-01 7.16344976e+00 8.14019024e-01
-1.26698518e+00 9.22548026e-02 2.78739840e-01 -4.85610247e-01
-6.39512623e-03 -5.31627536e-01 -4.62186098e-01 4.36955899e-01
1.26847243e+00 9.34314728e-02 3.95589858e-01 6.46483600e-01
4.55713123e-01 -3.69845599e-01 -1.25573885e+00 1.49989355e+00
1.11822522e+00 -8.82133424e-01 -2.41989493e-01 -1.55515522e-01
-2.80071139e-01 -4.24588591e-01 9.64505039e-03 1.71156880e-02
-4.59765255e-01 -1.29973352e+00 7.82668471e-01 8.42900217e-01
7.29744792e-01 -6.58726215e-01 3.77505511e-01 3.94141376e-01
-4.83280420e-01 -1.26708925e-01 2.76254881e-02 -1.03092507e-01
6.62322268e-02 -2.05441177e-01 -8.43235850e-01 2.38826964e-02
5.31973481e-01 1.62361696e-01 -1.90874785e-01 3.90616655e-01
-9.62870102e-03 4.97510672e-01 -3.61753076e-01 -2.96994001e-01
-7.84287080e-02 1.34047300e-01 8.33494663e-01 1.44881284e+00
1.48839220e-01 5.11335135e-01 -1.16864584e-01 4.68887925e-01
3.25485528e-01 5.30227125e-01 -1.04707503e+00 1.04516976e-01
2.43131563e-01 1.35461831e+00 -3.03189456e-02 -9.97036994e-02
-3.09269726e-01 1.30035448e+00 -7.09313378e-02 9.73921195e-02
-6.01530790e-01 -4.05326962e-01 7.40355492e-01 9.70712975e-02
4.03537676e-02 -2.60768421e-02 1.78759947e-01 -9.66244996e-01
3.29795852e-03 -1.09146309e+00 5.12150228e-01 -9.50892746e-01
-1.06432903e+00 7.22608089e-01 3.84267457e-02 -8.80593002e-01
-5.30119956e-01 -5.18121362e-01 -7.90419579e-01 6.34873629e-01
-8.66448820e-01 -1.02775931e+00 -9.37685370e-02 1.12500632e+00
6.51108861e-01 -4.90786910e-01 1.19380200e+00 -1.66164022e-02
-5.32850564e-01 9.08763349e-01 -6.52024865e-01 2.99361199e-01
7.17937529e-01 -5.23703396e-01 -3.02509665e-01 3.11311692e-01
3.95887554e-01 3.18368495e-01 9.18010473e-01 -4.56612796e-01
-1.35622883e+00 -2.64722258e-01 1.03262568e+00 -4.08378661e-01
7.19182491e-01 -5.82519650e-01 -8.80114436e-01 2.90556490e-01
2.76691854e-01 2.33466893e-01 1.37061858e+00 -1.93982810e-01
-6.80618167e-01 -1.05905041e-01 -1.65012014e+00 2.17960924e-01
6.28980339e-01 -9.13897276e-01 -1.10431790e+00 1.64342180e-01
2.23177031e-01 -1.48488209e-01 -5.34051597e-01 1.80854753e-01
1.06990004e+00 -1.00000536e+00 6.10755265e-01 -8.63133073e-01
3.18995379e-02 1.66618720e-01 -4.77186918e-01 -1.05884326e+00
3.51402253e-01 -8.95309448e-01 1.17163293e-01 1.18825781e+00
1.48826823e-01 -3.49378079e-01 6.78539157e-01 7.62169182e-01
4.24047440e-01 -2.69641221e-01 -1.53292727e+00 -2.85101891e-01
-4.86590654e-01 -8.15989971e-01 -2.37336364e-02 9.70485330e-01
9.73104775e-01 1.37579650e-01 -6.43112063e-01 4.43492085e-02
5.40204942e-01 -2.21465498e-01 3.78363252e-01 -1.02535117e+00
2.25863352e-01 -8.63543749e-02 -1.05191898e+00 -1.55764207e-01
6.36555672e-01 -6.59069359e-01 -4.00361679e-02 -5.42971730e-01
-1.47387221e-01 6.38290823e-01 4.53584455e-02 6.42699122e-01
3.81141871e-01 6.37109697e-01 2.97271758e-02 -3.01795483e-01
-4.60662872e-01 6.42150640e-01 3.88783634e-01 2.78440397e-02
-8.58176127e-02 -1.32682949e-01 -2.49137864e-01 8.75622630e-01
3.93485367e-01 -4.25756007e-01 1.20619632e-01 2.47010902e-01
-9.63574573e-02 6.23371363e-01 6.19821429e-01 -7.05876648e-01
4.16252553e-01 2.44852751e-02 3.25985193e-01 -5.34004033e-01
8.75617027e-01 -8.83597493e-01 -2.12539166e-01 2.75244236e-01
-4.23784971e-01 -1.69878490e-02 3.59470606e-01 5.96301198e-01
-3.59611601e-01 -3.33316505e-01 7.13955343e-01 2.36111045e-01
-5.84388554e-01 -1.97455008e-02 -8.74293089e-01 -5.11987567e-01
1.33614171e+00 -6.26015007e-01 2.87225813e-01 -8.13319266e-01
-1.51205683e+00 1.78793855e-02 -1.44885823e-01 6.55487299e-01
1.25309741e+00 -1.39687717e+00 -8.38799357e-01 8.55738759e-01
2.07331762e-01 -1.23900020e+00 3.92327100e-01 8.70803237e-01
-8.26226100e-02 -1.85843036e-02 -3.86696488e-01 -3.88334751e-01
-2.30273890e+00 5.65243423e-01 4.76492852e-01 9.02105451e-01
-1.69362649e-01 9.97827291e-01 -4.37738389e-01 3.12208235e-02
5.65678775e-01 1.70236215e-01 -5.63360274e-01 7.78418362e-01
9.98653829e-01 2.64726251e-01 2.01478124e-01 -1.02883089e+00
-7.92891562e-01 4.50393528e-01 2.64902323e-01 -3.74278903e-01
1.01045418e+00 -2.02525049e-01 -1.07254535e-01 6.24720752e-01
1.49180436e+00 1.18119292e-01 -6.75186694e-01 -1.51774421e-01
-1.45650268e-01 -5.14578462e-01 3.46185006e-02 -6.40269637e-01
-6.29217148e-01 1.04035664e+00 9.35899317e-01 3.43392938e-01
1.00811088e+00 2.80759148e-02 3.20931971e-01 2.21921787e-01
5.46386391e-02 -1.31741416e+00 2.03087673e-01 5.48453443e-02
1.31012404e+00 -1.33997774e+00 -1.03509530e-01 -1.03343390e-01
-8.12165976e-01 1.50558686e+00 1.73246890e-01 -1.15546674e-01
1.07612789e+00 3.57830048e-01 5.02106130e-01 -4.89779830e-01
-8.71206820e-01 -3.04733645e-02 5.04356563e-01 7.59526312e-01
-4.35408726e-02 2.87040863e-02 -4.19071406e-01 8.22995722e-01
-2.42055133e-01 -1.16411649e-01 2.70581096e-01 6.08321905e-01
-5.97535968e-01 -7.17407644e-01 -5.97449958e-01 7.31949322e-03
-6.88389599e-01 -2.24049650e-02 -9.85097885e-01 4.00067627e-01
1.65314659e-01 1.27402937e+00 1.58745199e-01 -7.98832774e-01
5.02996206e-01 8.20130348e-01 5.70148885e-01 -4.39925373e-01
-7.77157366e-01 -4.78966627e-03 2.00444147e-01 -9.73578811e-01
-4.25167084e-01 -1.17166424e+00 -9.01043594e-01 -2.70431131e-01
-1.70138657e-01 -1.15272082e-01 1.03799057e+00 9.94734943e-01
7.57018775e-02 -3.40303928e-01 9.64172781e-01 -8.14066708e-01
-5.93333066e-01 -1.10617590e+00 -6.08602226e-01 4.54689652e-01
7.62552440e-01 -3.49299639e-01 -8.25713873e-01 6.56560585e-02] | [13.336318016052246, 2.1722841262817383] |
49e8efc0-4b6e-4ccb-8d14-fdc3f732cf2b | seed-guided-topic-discovery-with-out-of | 2205.01845 | null | https://arxiv.org/abs/2205.01845v1 | https://arxiv.org/pdf/2205.01845v1.pdf | Seed-Guided Topic Discovery with Out-of-Vocabulary Seeds | Discovering latent topics from text corpora has been studied for decades. Many existing topic models adopt a fully unsupervised setting, and their discovered topics may not cater to users' particular interests due to their inability of leveraging user guidance. Although there exist seed-guided topic discovery approaches that leverage user-provided seeds to discover topic-representative terms, they are less concerned with two factors: (1) the existence of out-of-vocabulary seeds and (2) the power of pre-trained language models (PLMs). In this paper, we generalize the task of seed-guided topic discovery to allow out-of-vocabulary seeds. We propose a novel framework, named SeeTopic, wherein the general knowledge of PLMs and the local semantics learned from the input corpus can mutually benefit each other. Experiments on three real datasets from different domains demonstrate the effectiveness of SeeTopic in terms of topic coherence, accuracy, and diversity. | ['Jiawei Han', 'Sheng Wang', 'Xuan Wang', 'Yu Meng', 'Yu Zhang'] | 2022-05-04 | null | https://aclanthology.org/2022.naacl-main.21 | https://aclanthology.org/2022.naacl-main.21.pdf | naacl-2022-7 | ['general-knowledge', 'topic-models'] | ['miscellaneous', 'natural-language-processing'] | [-6.90441430e-02 2.52306968e-01 -7.22398400e-01 -3.70457500e-01
-8.24064553e-01 -5.41697860e-01 9.63475049e-01 3.38860363e-01
-6.74558133e-02 6.66002631e-01 5.13037264e-01 -3.12770382e-02
6.21757507e-02 -8.43618810e-01 -4.16568279e-01 -5.70779562e-01
-1.25789136e-01 5.49504757e-01 6.99332654e-01 -1.89178571e-01
2.56654710e-01 -3.43397588e-01 -1.44467700e+00 1.69369444e-01
1.27500105e+00 4.84102577e-01 5.32121480e-01 -1.10121304e-02
-5.34077883e-01 1.50785431e-01 -5.62639296e-01 1.31720349e-01
-2.38476563e-02 -2.45971248e-01 -4.11301523e-01 2.71869183e-01
-7.30724111e-02 -1.26545474e-01 -5.97565696e-02 1.03253162e+00
7.70420656e-02 2.17410773e-01 5.87695122e-01 -1.25302887e+00
-4.93070185e-01 1.11354721e+00 -6.53295040e-01 9.17721838e-02
1.19260177e-01 -2.49214381e-01 1.30685890e+00 -8.21343899e-01
6.68732107e-01 1.16724575e+00 2.11604372e-01 2.70413846e-01
-1.04446769e+00 -8.91015291e-01 8.16487253e-01 -2.95768321e-01
-1.45906448e+00 -1.90878600e-01 9.61533248e-01 -5.15270770e-01
4.43774939e-01 2.17866129e-03 4.45635796e-01 1.08915460e+00
-2.38437966e-01 1.09274113e+00 8.48971009e-01 -3.34223986e-01
5.26931047e-01 6.50779903e-01 4.31978941e-01 1.66251376e-01
3.73306364e-01 -3.65258485e-01 -8.79579425e-01 -8.11403334e-01
7.41714418e-01 1.30227968e-01 -3.58264655e-01 -7.07418978e-01
-1.14285839e+00 1.21540046e+00 7.63097927e-02 5.65806031e-01
-5.46775460e-01 -2.70907998e-01 2.44919166e-01 -1.52855650e-01
9.56678271e-01 6.19905829e-01 -4.62538004e-01 1.60906408e-02
-1.28540874e+00 3.19272369e-01 7.82810628e-01 1.20362949e+00
8.68201494e-01 -2.76597589e-01 -1.36438996e-01 6.96713209e-01
5.00050008e-01 3.21277380e-01 9.08962131e-01 -1.25991091e-01
4.28697854e-01 7.11900830e-01 2.68375874e-01 -8.39350581e-01
-3.63240987e-02 -5.35425365e-01 -2.92704821e-01 -6.12004519e-01
8.57175812e-02 -1.88359186e-01 -9.21931624e-01 1.91612625e+00
5.34782290e-01 3.95237982e-01 1.10912748e-01 7.75184691e-01
6.04337811e-01 8.37267935e-01 2.68084615e-01 -2.73170084e-01
1.26485646e+00 -8.94352734e-01 -6.60607874e-01 -4.41249549e-01
4.81059402e-01 -5.36408067e-01 1.17479384e+00 2.00641900e-01
-7.18836665e-01 -1.07625663e-01 -8.04803014e-01 3.15794885e-01
-3.35290909e-01 9.69873369e-03 7.88780808e-01 5.05597830e-01
-8.30781281e-01 1.07272431e-01 -8.87249708e-01 -5.63154578e-01
3.39863420e-01 1.16715066e-01 1.91451430e-01 1.22325264e-01
-1.40884268e+00 1.97331846e-01 6.06444836e-01 -4.22896445e-01
-1.04468989e+00 -6.96444809e-01 -4.91768897e-01 2.91737378e-01
9.50386167e-01 -4.50470775e-01 1.29442787e+00 -7.88233459e-01
-1.06750917e+00 3.79156768e-01 -6.09304607e-01 -5.14086008e-01
1.87545702e-01 -5.03193259e-01 -2.34412938e-01 1.73095524e-01
6.34501994e-01 4.87665594e-01 9.16282535e-01 -1.35150778e+00
-9.31627810e-01 -1.73188090e-01 4.21210043e-02 3.55742842e-01
-7.98127353e-01 -1.54528514e-01 -6.24262989e-01 -7.75094986e-01
3.32627326e-01 -6.38446093e-01 -3.57632160e-01 -3.97795379e-01
-6.43674374e-01 -7.39466369e-01 1.05315018e+00 -1.16670355e-01
1.42740738e+00 -2.09123778e+00 -2.96184719e-01 2.84315765e-01
3.50468367e-01 -9.96974111e-02 1.21413045e-01 6.04608476e-01
2.20995650e-01 3.66462082e-01 1.63434550e-01 -2.76383936e-01
-8.48321989e-02 -4.57805097e-02 -1.00948572e+00 1.50215775e-01
-1.02438882e-01 6.56790912e-01 -1.24170554e+00 -6.69103265e-01
-1.93851650e-01 2.89124459e-01 -4.21407610e-01 1.90698221e-01
-8.23257506e-01 1.39486581e-01 -1.23296905e+00 4.81731117e-01
4.11030620e-01 -6.99316978e-01 2.96507329e-01 2.60720760e-01
-2.83200163e-02 6.43139541e-01 -1.08796024e+00 1.59817636e+00
-1.71306670e-01 3.89683694e-01 -1.51050583e-01 -6.20698154e-01
8.68478358e-01 7.12016225e-01 7.04516947e-01 -1.47159740e-01
-2.26646215e-01 1.59075812e-01 -4.13236290e-01 -7.78510869e-02
7.83090174e-01 -9.63648334e-02 -1.60376113e-02 1.01869357e+00
3.12263321e-04 2.40626290e-01 1.85598060e-01 7.85793304e-01
7.01467395e-01 -1.68879062e-01 4.53780800e-01 -5.37508547e-01
-8.21434706e-02 2.37360910e-01 5.37365973e-01 1.03084600e+00
-8.81355815e-03 4.51274037e-01 4.10856128e-01 1.52975554e-02
-8.14102769e-01 -1.06299520e+00 -1.67594790e-01 1.46969998e+00
3.70036244e-01 -7.49215543e-01 -4.96420622e-01 -7.71355152e-01
-1.51631951e-01 8.66209149e-01 -6.11954391e-01 7.16786385e-02
-2.83129841e-01 -5.58730483e-01 1.35947838e-01 2.08455577e-01
1.19467750e-01 -7.50970423e-01 -5.65575480e-01 4.27955091e-01
-4.42450702e-01 -1.11408353e+00 -6.55582547e-01 -1.10106375e-02
-1.04660165e+00 -8.04439843e-01 -8.51773202e-01 -4.76427525e-01
7.77443111e-01 8.86243880e-01 1.05251932e+00 -9.09737200e-02
4.12954330e-01 2.74645507e-01 -5.75359881e-01 -6.96250141e-01
-1.40225828e-01 5.86571872e-01 1.60872161e-01 1.53882271e-02
7.11989403e-01 -6.80066049e-01 -5.04214764e-01 3.88109595e-01
-1.00638843e+00 1.02145292e-01 3.96441132e-01 6.26428127e-01
2.11139053e-01 4.46556538e-01 8.78266871e-01 -1.22673202e+00
8.69744778e-01 -8.83201897e-01 -4.08288300e-01 2.74772972e-01
-8.33642781e-01 3.30140665e-02 2.20537588e-01 -7.67720759e-01
-1.13121903e+00 -3.23768735e-01 6.41411006e-01 -3.50323737e-01
-1.41590878e-01 7.54923582e-01 -1.34101972e-01 7.55327702e-01
6.54744506e-01 4.29942995e-01 -4.69515711e-01 -6.05200231e-01
5.17350972e-01 5.88098347e-01 1.03662968e-01 -7.77123988e-01
8.52236986e-01 7.03444481e-01 -7.99117625e-01 -8.73500645e-01
-1.10529637e+00 -1.02068388e+00 -3.19262147e-01 4.71571051e-02
5.02790570e-01 -1.38892412e+00 1.39892414e-01 1.00759231e-01
-1.06221437e+00 -5.74654266e-02 -3.19534063e-01 5.76273799e-01
-1.01913206e-01 2.53949732e-01 -1.44512028e-01 -9.35536742e-01
-4.39125508e-01 -9.14411485e-01 1.16621816e+00 4.33091104e-01
-3.24861050e-01 -1.18130553e+00 3.80031802e-02 -1.06563158e-01
4.07580048e-01 -1.68184325e-01 9.53022003e-01 -1.33891666e+00
-8.09130192e-01 -2.53169060e-01 -5.16453646e-02 -2.89760739e-01
5.42039633e-01 -3.72505754e-01 -1.12308359e+00 -3.60673219e-01
3.32284421e-02 -2.04751119e-01 8.28603148e-01 5.01405418e-01
6.83321238e-01 -5.31489730e-01 -9.76443708e-01 1.03392914e-01
1.09733582e+00 -2.03214679e-02 9.10886601e-02 4.58301574e-01
3.47600341e-01 7.14059591e-01 7.69628167e-01 6.76981628e-01
5.06368816e-01 6.01279378e-01 1.57849967e-01 -1.40300691e-01
4.42619205e-01 -6.87496424e-01 2.54909873e-01 7.00031519e-01
2.51749307e-01 -3.96879911e-01 -8.65408123e-01 1.09005201e+00
-1.88891637e+00 -7.45986044e-01 2.25094602e-01 2.08814955e+00
1.14234972e+00 3.70891064e-01 2.98823446e-01 -3.31577659e-01
8.48914146e-01 3.29704046e-01 -5.62146544e-01 2.92125463e-01
-1.96500029e-02 -2.62024969e-01 1.34117737e-01 2.11537287e-01
-1.21517050e+00 1.18081892e+00 5.76855278e+00 1.00887179e+00
-1.10361421e+00 2.70964652e-01 4.42591786e-01 -1.02416329e-01
-7.00875521e-01 3.87569070e-01 -1.26748097e+00 5.18454313e-01
6.66289449e-01 -7.04797566e-01 -1.76758960e-01 1.31106663e+00
2.79180914e-01 -1.47413269e-01 -9.37974334e-01 3.21707100e-01
9.57836211e-02 -1.07824874e+00 1.96963832e-01 3.29657048e-01
9.67835426e-01 1.20928027e-01 1.86777283e-02 3.43678206e-01
6.49132729e-01 -6.08661234e-01 7.04288185e-01 1.30466744e-01
2.84759402e-01 -3.91917020e-01 3.76709551e-01 7.10320115e-01
-1.20060968e+00 1.86866656e-01 -4.55860078e-01 2.60226518e-01
3.68256420e-01 9.83194292e-01 -1.36221492e+00 4.02656347e-01
5.23666084e-01 6.35677576e-01 -3.20183486e-01 8.99565339e-01
-3.63583654e-01 1.23181856e+00 -5.47716796e-01 -2.16000065e-01
6.11398160e-01 2.82569230e-03 7.88198352e-01 1.01005948e+00
3.53389233e-01 -5.50469086e-02 5.92132092e-01 9.73734975e-01
9.65643525e-02 4.77160037e-01 -4.64889050e-01 -3.01904619e-01
8.29188764e-01 1.15310299e+00 -1.08336067e+00 -5.85717916e-01
-4.22626019e-01 3.80864233e-01 -3.38873677e-02 4.94993895e-01
-4.55849469e-01 -9.90406796e-02 5.88860512e-01 4.21282858e-01
5.45388639e-01 -2.58424103e-01 -3.23213398e-01 -1.37596428e+00
-1.99842025e-02 -6.63477659e-01 3.67695242e-01 -4.76387173e-01
-1.26587546e+00 5.27709723e-01 4.26124364e-01 -1.03806961e+00
-3.12755257e-01 1.34901300e-01 -8.02321494e-01 7.68610358e-01
-1.43707454e+00 -1.18234622e+00 -3.06731034e-02 3.74486983e-01
8.56944919e-01 -1.68067500e-01 6.18542373e-01 -2.08269387e-01
-2.50532359e-01 2.75434762e-01 2.92867661e-01 -1.26359791e-01
7.47023940e-01 -1.12316525e+00 4.95203346e-01 8.93196285e-01
3.86846930e-01 1.17408502e+00 9.09675002e-01 -1.00496662e+00
-9.36917245e-01 -1.10459995e+00 1.01567733e+00 -3.76110554e-01
7.35339165e-01 -6.62591994e-01 -1.17044139e+00 6.78049326e-01
1.70650825e-01 -5.89225292e-01 7.90316105e-01 7.88355291e-01
-4.28869814e-01 3.01080823e-01 -6.38815045e-01 6.80668354e-01
6.79263711e-01 -3.95733029e-01 -6.80648208e-01 4.28824276e-01
1.11774945e+00 -6.42810389e-02 -3.87819082e-01 9.81434062e-02
3.00111175e-01 -3.92741770e-01 8.12494278e-01 -5.95218301e-01
2.00864866e-01 -1.73825964e-01 -5.35411872e-02 -1.15283549e+00
-2.78101787e-02 -6.50773942e-01 -2.06880540e-01 1.56389380e+00
4.96376932e-01 -5.36299825e-01 8.66236746e-01 5.82889676e-01
1.40770033e-01 -7.12240100e-01 -5.91791868e-01 -7.06597388e-01
2.16707624e-02 -3.79244059e-01 8.28154027e-01 1.05796671e+00
2.73495585e-01 6.73216939e-01 -3.39136720e-01 3.49669337e-01
4.86682028e-01 5.04822910e-01 8.91902924e-01 -1.51137567e+00
-1.63473994e-01 -2.18701497e-01 1.72856227e-01 -1.40119648e+00
-8.00538734e-02 -6.04067147e-01 3.20353985e-01 -1.51510966e+00
5.65196872e-01 -8.21563244e-01 -3.17598134e-01 5.27505040e-01
-5.63484788e-01 -5.39272428e-01 -1.37455374e-01 7.71842420e-01
-9.31350410e-01 6.70509517e-01 8.86272013e-01 1.70216620e-01
-6.85695291e-01 1.43546730e-01 -1.19272912e+00 8.78936291e-01
6.86545491e-01 -7.81215906e-01 -9.94223833e-01 -2.41433010e-01
1.18247405e-01 -3.11057419e-01 3.52467224e-02 -6.13555729e-01
3.92374009e-01 -4.80176061e-01 -2.05506548e-01 -7.73077071e-01
1.50402233e-01 -3.65779579e-01 -3.05824727e-01 -1.43696249e-01
-5.62215149e-01 -4.17034388e-01 -9.08635184e-02 9.53686714e-01
-2.60219544e-01 -1.67620987e-01 2.96019971e-01 -2.35579297e-01
-6.53232694e-01 4.04248983e-01 -4.35371846e-01 3.62705052e-01
7.95265317e-01 1.14336368e-02 -3.29277277e-01 -6.61168814e-01
-2.87076741e-01 2.93621540e-01 4.75013852e-01 5.68802416e-01
2.87765175e-01 -1.03458822e+00 -5.82410216e-01 6.82778507e-02
5.07010937e-01 1.98633015e-01 2.86691934e-02 4.73221064e-01
4.98769760e-01 8.50065649e-01 5.89527905e-01 -7.67942727e-01
-8.88889968e-01 4.95408535e-01 -2.48027042e-01 -5.81592560e-01
-7.60413826e-01 7.19941735e-01 1.01730430e+00 -1.20317228e-01
2.64906704e-01 -1.59902215e-01 -1.78678200e-01 7.20510930e-02
5.15280187e-01 -2.88480580e-01 -3.27039003e-01 -3.27841043e-01
-9.21285897e-02 7.93942809e-02 -5.66401243e-01 -3.43381822e-01
1.30260968e+00 -3.63227069e-01 1.75896585e-01 5.89689434e-01
6.49435282e-01 3.72408815e-02 -1.12599730e+00 -9.99520540e-01
4.99232560e-01 -3.97367865e-01 1.73831150e-01 -4.86602008e-01
-6.45019054e-01 7.01226711e-01 1.60174817e-01 4.90212917e-01
8.77682745e-01 3.77888113e-01 8.24495196e-01 2.68712789e-01
6.05596721e-01 -1.00456572e+00 2.33577505e-01 3.06642443e-01
4.36680436e-01 -1.13343620e+00 6.23493195e-02 -6.88837230e-01
-6.45275056e-01 7.13572145e-01 5.00931203e-01 1.89710796e-01
8.82333457e-01 -7.19144791e-02 1.03126645e-01 -3.60981345e-01
-1.07186818e+00 -2.99351692e-01 4.51314986e-01 4.33730155e-01
5.80958903e-01 3.43929231e-02 -3.07179898e-01 1.03256500e+00
-1.86526552e-01 -1.15693159e-01 3.28266710e-01 1.07326508e+00
-1.02710927e+00 -1.19820821e+00 -2.66038507e-01 5.28260887e-01
-4.03365016e-01 -1.59846500e-01 -2.96321452e-01 7.21521795e-01
-1.28289163e-01 9.45726275e-01 -1.58086315e-01 -6.32374063e-02
-2.20819339e-01 2.05284446e-01 -3.89328569e-01 -1.25441563e+00
-1.53626740e-01 6.96609914e-01 -1.48180217e-01 -1.92772225e-01
-3.17578107e-01 -8.52856159e-01 -9.47600842e-01 1.78992406e-01
-8.48523378e-01 8.33112121e-01 5.78658104e-01 1.06078780e+00
5.42177975e-01 2.09306300e-01 5.71273208e-01 -2.71641016e-01
-5.23117006e-01 -1.27425385e+00 -8.96598518e-01 2.22096860e-01
5.46090789e-02 -8.29909384e-01 -2.55379051e-01 1.25267565e-01] | [10.404695510864258, 6.926445484161377] |
f4d4ab82-4de0-468e-a173-078139d313dc | investigating-evaluation-of-open-domain | 1907.10568 | null | https://arxiv.org/abs/1907.10568v2 | https://arxiv.org/pdf/1907.10568v2.pdf | Investigating Evaluation of Open-Domain Dialogue Systems With Human Generated Multiple References | The aim of this paper is to mitigate the shortcomings of automatic evaluation of open-domain dialog systems through multi-reference evaluation. Existing metrics have been shown to correlate poorly with human judgement, particularly in open-domain dialog. One alternative is to collect human annotations for evaluation, which can be expensive and time consuming. To demonstrate the effectiveness of multi-reference evaluation, we augment the test set of DailyDialog with multiple references. A series of experiments show that the use of multiple references results in improved correlation between several automatic metrics and human judgement for both the quality and the diversity of system output. | ['Prakhar Gupta', 'Amy Pavel', 'Tiancheng Zhao', 'Shikib Mehri', 'Maxine Eskenazi', 'Jeffrey P. Bigham'] | 2019-07-24 | investigating-evaluation-of-open-domain-1 | https://aclanthology.org/W19-5944 | https://aclanthology.org/W19-5944.pdf | ws-2019-9 | ['dialogue-evaluation', 'open-domain-dialog'] | ['natural-language-processing', 'natural-language-processing'] | [-2.80027658e-01 2.52071321e-01 2.90702820e-01 -6.56785429e-01
-9.90134537e-01 -1.12747169e+00 8.89581800e-01 2.63342261e-01
-7.19382107e-01 1.10500085e+00 3.91116917e-01 -1.62433669e-01
-6.73952848e-02 -1.98598549e-01 3.13571483e-01 -1.06901929e-01
3.35677296e-01 7.87253439e-01 4.60554570e-01 -7.09723890e-01
2.77680397e-01 6.89730644e-02 -1.11340952e+00 1.93383023e-01
8.84394944e-01 4.95654285e-01 -4.38897386e-02 8.23062599e-01
-1.77332893e-01 6.34324729e-01 -1.24132085e+00 -6.85752749e-01
-1.58222467e-02 -6.54213488e-01 -1.11962306e+00 1.80721253e-01
8.35923031e-02 -5.20846367e-01 1.49158388e-01 7.88603663e-01
7.11414456e-01 4.41855311e-01 6.55386508e-01 -1.02508032e+00
-2.90108740e-01 4.01487648e-01 1.69770688e-01 2.41632491e-01
7.75751889e-01 2.19607159e-01 9.88769889e-01 -6.14883304e-01
6.32377863e-01 1.49358618e+00 5.56555331e-01 5.53211808e-01
-1.51017964e+00 -3.30640614e-01 -1.55652851e-01 -2.98533112e-01
-1.09174216e+00 -7.89945364e-01 3.08855206e-01 -5.84182680e-01
9.04324353e-01 1.02436617e-01 -2.65764655e-03 9.27715659e-01
-1.40747368e-01 1.72573224e-01 1.30774581e+00 -5.61267734e-01
2.42631659e-01 8.60364377e-01 5.69441378e-01 4.75499481e-01
6.62530735e-02 8.55267886e-03 -2.89591014e-01 -4.95966554e-01
4.05647814e-01 -5.76209247e-01 -2.87480801e-01 -1.97300434e-01
-9.21150565e-01 1.09893656e+00 1.15107484e-01 6.18247747e-01
-5.57446331e-02 -6.30704761e-01 5.57734072e-01 6.69423401e-01
5.40360332e-01 9.62099195e-01 -3.74160260e-01 -4.71911609e-01
-4.72666740e-01 4.69899088e-01 1.40611565e+00 3.02818954e-01
4.60405797e-01 -3.05218369e-01 -3.82262915e-01 1.32076263e+00
3.43135357e-01 3.52126449e-01 6.86928213e-01 -1.16940773e+00
5.05341887e-01 6.93193495e-01 6.49965644e-01 -7.93236494e-01
-5.04367530e-01 1.53057873e-01 -9.94229689e-03 3.31996590e-01
9.18016672e-01 -5.85077167e-01 -3.25366080e-01 1.75066209e+00
2.33854324e-01 -9.57538486e-01 3.33632857e-01 8.95101190e-01
8.65841031e-01 1.91572279e-01 2.89548546e-01 -2.39315346e-01
1.05069053e+00 -7.72905648e-01 -9.88129079e-01 2.52290946e-02
1.02982378e+00 -9.32528436e-01 1.22657120e+00 4.26824316e-02
-9.61647809e-01 -5.05420864e-01 -9.27957833e-01 1.55445319e-02
-3.19300830e-01 -1.16191834e-01 1.35115206e-01 7.88672984e-01
-9.24037218e-01 3.97954166e-01 -2.32266098e-01 -5.27864635e-01
-4.40614879e-01 3.23139191e-01 -2.69523263e-01 3.17176759e-01
-1.41639555e+00 1.53960240e+00 1.67613029e-01 -4.21941340e-01
-1.95893094e-01 -6.01539090e-02 -8.34268808e-01 2.68844403e-02
2.36536309e-01 -1.86309397e-01 1.90485239e+00 -6.77719474e-01
-1.66711271e+00 8.01108420e-01 -5.25486376e-03 -2.54796028e-01
5.91666222e-01 -1.91122219e-01 -3.45211387e-01 -3.10355537e-02
2.20935568e-01 4.82555449e-01 1.63207218e-01 -1.16945517e+00
-3.98311287e-01 -2.49458626e-01 4.26306397e-01 5.34729540e-01
-3.33493859e-01 3.16780895e-01 -6.08313270e-02 -2.36491621e-01
-3.83419871e-01 -8.79797697e-01 -1.98994413e-01 -6.68147683e-01
-1.89588573e-02 -5.49219251e-01 6.55062497e-01 -6.84408844e-01
1.32725930e+00 -2.10585189e+00 -2.97502708e-02 1.72199532e-02
3.03176552e-01 5.26704431e-01 2.15913102e-01 4.51073974e-01
2.12160185e-01 8.96278396e-02 -4.00898047e-03 -2.93040544e-01
3.07262033e-01 2.14263991e-01 3.12778950e-02 9.44271870e-03
4.99030240e-02 6.93452179e-01 -8.56898367e-01 -7.25755334e-01
3.70058745e-01 6.16639666e-02 -3.98561209e-01 4.85241711e-01
-3.47938031e-01 6.23335421e-01 -2.69683003e-01 9.45402756e-02
1.00098029e-01 -2.46265829e-01 3.24139059e-01 2.69747794e-01
-7.94195086e-02 6.73578322e-01 -9.01436746e-01 1.32364583e+00
-4.13862675e-01 6.19273484e-01 2.38751665e-01 -1.87378690e-01
1.01627433e+00 6.45781934e-01 -5.39934635e-03 -8.15392137e-01
3.08415413e-01 2.71281302e-01 4.36898619e-01 -4.53100115e-01
6.92972362e-01 -3.48962218e-01 -3.29002351e-01 9.13108408e-01
1.02857023e-01 -3.03050607e-01 4.54992652e-01 4.32086349e-01
8.30031991e-01 -2.61083215e-01 3.46011788e-01 -2.70775735e-01
7.35220075e-01 1.26132965e-01 9.65376347e-02 5.53394675e-01
-5.21579206e-01 2.51219064e-01 5.69693863e-01 -1.56439617e-02
-7.63248086e-01 -8.11244607e-01 -4.46311422e-02 1.33408952e+00
-8.00441727e-02 -2.66567737e-01 -9.96767282e-01 -8.32389057e-01
-1.09482020e-01 8.40741754e-01 -1.67669088e-01 8.82146060e-02
-3.13186258e-01 -4.54890221e-01 5.72167516e-01 3.52303892e-01
4.79695052e-01 -8.47402394e-01 -6.21066153e-01 3.09526652e-01
-5.86516976e-01 -1.12514520e+00 -3.68360460e-01 7.92798325e-02
-6.84083939e-01 -9.20417249e-01 -6.89882159e-01 -4.66804475e-01
2.15776727e-01 6.76211901e-03 1.23894525e+00 2.09434539e-01
2.68363088e-01 4.91232991e-01 -3.61526370e-01 -1.95440590e-01
-8.41145575e-01 1.37434766e-01 3.74203771e-02 -4.81078565e-01
5.58365643e-01 -7.02420399e-02 -1.85023412e-01 7.56027400e-01
-5.59896827e-01 -5.33717573e-01 1.43310547e-01 1.03915346e+00
-3.08043599e-01 -4.59269911e-01 8.64356101e-01 -8.47756684e-01
1.63087082e+00 -2.63029933e-01 -5.77066064e-01 3.51847649e-01
-9.25242782e-01 1.78775966e-01 2.81780899e-01 -2.12886274e-01
-1.30564475e+00 -2.06075579e-01 -1.51345864e-01 9.46791992e-02
-2.99412221e-01 3.98933589e-01 1.22907355e-01 -1.20433681e-01
1.01171446e+00 -4.79340881e-01 2.96235651e-01 -5.03879964e-01
2.00712904e-01 1.03293049e+00 3.36074442e-01 -5.30194700e-01
3.70866030e-01 -3.23980510e-01 -5.63429832e-01 -7.74444759e-01
-7.90065765e-01 -6.03274405e-01 -6.29376471e-01 -4.04935747e-01
8.90099466e-01 -7.27355778e-01 -5.26951253e-01 1.41356504e-02
-1.11196876e+00 -6.68250322e-01 4.23914641e-02 4.90955234e-01
-2.63276249e-01 4.71302092e-01 -5.68235636e-01 -9.46017861e-01
-1.60544127e-01 -1.21108735e+00 8.26706290e-01 1.60441145e-01
-1.15242207e+00 -1.37748528e+00 4.37816918e-01 5.34286439e-01
3.18665862e-01 -6.37889430e-02 6.11713827e-01 -1.30872166e+00
1.01320677e-01 -4.16305125e-01 -5.86257949e-02 3.24301034e-01
1.32899150e-01 -2.02829778e-01 -1.01393163e+00 -7.30275288e-02
2.19082296e-01 -8.57685685e-01 2.54387736e-01 -2.00682476e-01
1.06978208e-01 -5.94178103e-02 -5.84218502e-02 -3.99192244e-01
9.35098290e-01 3.28039944e-01 2.23104209e-01 3.85600150e-01
1.43151373e-01 8.42186630e-01 8.86271298e-01 3.63514930e-01
4.63417053e-01 9.60489094e-01 -3.45692009e-01 1.77656367e-01
4.11716430e-03 7.22692013e-02 2.04695433e-01 6.79356873e-01
2.35906467e-01 -3.60853314e-01 -9.76476431e-01 5.16423762e-01
-1.63706970e+00 -7.14711785e-01 -1.67770475e-01 2.04399252e+00
7.66596079e-01 2.28627369e-01 5.54957569e-01 2.48110458e-01
7.66781390e-01 -3.98490840e-04 3.77925229e-04 -7.44941831e-01
1.15390159e-01 7.87841082e-02 -8.63540731e-03 1.02046371e+00
-6.87795818e-01 6.12273932e-01 7.24084520e+00 7.14603588e-02
-5.07826030e-01 3.70702237e-01 3.25529367e-01 6.56795278e-02
9.77272540e-02 -1.75470039e-01 -6.90480649e-01 5.16444147e-01
1.28207886e+00 -4.35360581e-01 1.62503064e-01 6.34500861e-01
1.27451271e-01 -4.58584011e-01 -1.25313437e+00 5.41709483e-01
2.68164203e-02 -5.70052207e-01 -5.69537222e-01 2.34059207e-02
5.33496916e-01 -2.38003969e-01 -3.81642848e-01 7.08511293e-01
6.64480984e-01 -6.37562573e-01 2.22870186e-01 2.05916073e-02
4.18919384e-01 -4.13747460e-01 9.70561981e-01 4.25382197e-01
-5.24638534e-01 1.23458780e-01 -2.63140276e-02 -2.59890258e-01
2.85901070e-01 -6.50418783e-03 -1.34146929e+00 7.38141984e-02
2.89762765e-01 -1.94682822e-01 -5.21765172e-01 6.64364457e-01
-2.49949545e-01 1.35667637e-01 -2.39882186e-01 -3.77895683e-01
2.84407496e-01 1.17992959e-03 3.96782309e-01 1.27121425e+00
-2.06319645e-01 3.27509701e-01 4.22544658e-01 4.78415519e-01
1.33320034e-01 1.87105224e-01 -7.91500211e-01 -1.64743796e-01
5.28688610e-01 1.04542053e+00 -4.31503445e-01 -4.46154207e-01
-2.12082803e-01 8.24814677e-01 3.75182211e-01 6.89517483e-02
-5.71326435e-01 -5.16084850e-01 2.84504116e-01 -8.31046328e-02
-8.87551382e-02 -2.97315210e-01 -4.10368472e-01 -7.68230498e-01
-1.02420725e-01 -1.00856829e+00 5.91611028e-01 -4.67990011e-01
-1.13359261e+00 8.03143561e-01 8.62414166e-02 -9.23847854e-01
-8.79777372e-01 -4.30312335e-01 -4.30499226e-01 1.12530315e+00
-8.29958141e-01 -3.20725769e-01 -4.54247385e-01 4.60199714e-01
6.52212858e-01 -1.61369294e-01 1.12456429e+00 3.25961620e-01
-2.83589035e-01 6.97612643e-01 -6.64243773e-02 1.01443850e-01
1.21628189e+00 -1.40783429e+00 1.37646243e-01 2.73000330e-01
-1.88207716e-01 6.99475884e-01 9.34965372e-01 -6.12671435e-01
-5.02497077e-01 -5.02626002e-01 1.08377182e+00 -8.98050368e-01
7.90517628e-01 -2.70690937e-02 -9.33614731e-01 3.79549086e-01
5.56238174e-01 -6.48353696e-01 8.03801179e-01 4.04900968e-01
-2.83930421e-01 2.47785404e-01 -1.27798152e+00 5.47231138e-01
4.98876184e-01 -7.28257358e-01 -1.03574407e+00 4.64691043e-01
5.33578992e-01 -4.51485455e-01 -1.04181719e+00 2.26828516e-01
4.33091611e-01 -9.47729826e-01 5.48464835e-01 -3.77858669e-01
1.93113670e-01 3.53787504e-02 -1.55377686e-01 -1.47907984e+00
-5.07009439e-02 -5.48242271e-01 2.58587122e-01 1.45582736e+00
6.31014407e-01 -6.86788738e-01 2.57005155e-01 1.24888229e+00
8.95700753e-02 -1.68912277e-01 -6.01685822e-01 -6.85617566e-01
5.72901331e-02 1.03758834e-01 1.57478854e-01 7.84694910e-01
7.54366338e-01 1.12082422e+00 -6.98490813e-02 -2.56165594e-01
2.51787722e-01 -1.82236969e-01 9.76729810e-01 -1.45450771e+00
-1.41604602e-01 -3.27667415e-01 -1.94655895e-01 -7.72103369e-01
1.58037752e-01 -3.63382071e-01 3.16077560e-01 -1.27010357e+00
-2.25458741e-02 -3.78258020e-01 2.86649470e-03 7.26471394e-02
-3.20013613e-01 1.63308457e-01 3.14482361e-01 3.26267064e-01
-8.57004285e-01 3.27309638e-01 9.63166535e-01 3.08582485e-01
-6.22783422e-01 -2.54071155e-03 -6.69568479e-01 5.96443534e-01
9.19086218e-01 -3.21834892e-01 -4.71001416e-01 -3.61618370e-01
-3.82705718e-01 3.32065016e-01 6.34340495e-02 -9.18589532e-01
7.97997639e-02 1.63909629e-01 8.50225762e-02 -3.37360054e-01
5.65715730e-01 -5.75901866e-01 -2.72816598e-01 2.88218647e-01
-6.53414786e-01 3.83853734e-01 1.91297084e-01 3.29903156e-01
-3.74501765e-01 -5.87464511e-01 7.59927154e-01 -3.47120166e-01
-4.73933160e-01 -6.30260050e-01 -4.42175895e-01 3.19459617e-01
8.77233386e-01 1.56567059e-02 -4.96020228e-01 -7.42553532e-01
-7.37485826e-01 3.62285316e-01 5.11510313e-01 3.11618686e-01
6.55688196e-02 -8.99506152e-01 -3.62763673e-01 -2.41070747e-01
9.12603363e-02 -4.81279045e-01 -3.30372214e-01 7.61178017e-01
-1.92778587e-01 8.83921921e-01 -1.64676145e-01 -5.51480591e-01
-1.51442766e+00 1.43729731e-01 3.08997691e-01 -5.40576279e-01
-1.90835223e-01 5.98640680e-01 -1.49774745e-01 -5.89590967e-01
4.42505211e-01 1.65816963e-01 -4.00268584e-01 2.22349927e-01
4.78087127e-01 3.82470191e-01 1.90577582e-01 -6.82637036e-01
-1.41604573e-01 -6.95401728e-02 -2.72102028e-01 -9.47168708e-01
8.09179306e-01 -4.70574945e-01 1.62957832e-01 8.09183300e-01
8.98185372e-01 -7.95178413e-02 -6.45818651e-01 -2.20238149e-01
4.25542384e-01 -4.74440902e-01 -2.67583728e-01 -1.14474082e+00
-1.08117454e-01 6.08780861e-01 6.83093190e-01 6.17902517e-01
4.97757137e-01 -1.89788014e-01 5.12911856e-01 7.39232898e-01
3.24291736e-01 -1.32863200e+00 2.35474780e-01 7.86487877e-01
8.10727060e-01 -1.54454279e+00 -1.44486681e-01 -2.66001642e-01
-1.03565037e+00 8.86000574e-01 9.33952212e-01 3.19603533e-01
3.25891763e-01 -2.87839063e-02 6.68968499e-01 -3.52591485e-01
-7.26494968e-01 -1.73457459e-01 -1.77325010e-02 4.36338037e-01
8.76714408e-01 4.12333496e-02 -7.66446054e-01 3.12308937e-01
-2.52885401e-01 -1.62612930e-01 4.75103050e-01 1.05651402e+00
-4.42454517e-01 -1.23711121e+00 -5.06849527e-01 1.98771194e-01
-4.73063737e-01 3.10557485e-01 -1.03692198e+00 7.37413406e-01
-4.34575438e-01 1.63799691e+00 -2.05179453e-01 -4.07438129e-01
5.17553508e-01 7.12736368e-01 2.27056503e-01 -8.39583039e-01
-1.06369460e+00 -2.83105969e-02 8.38876307e-01 -2.00428903e-01
-5.09271860e-01 -6.15440011e-01 -9.09083903e-01 -2.41973624e-01
-6.58556104e-01 6.11881793e-01 6.28745198e-01 1.06709516e+00
-1.92395840e-02 3.38081181e-01 5.94922006e-01 -5.00416100e-01
-1.20882618e+00 -1.49551189e+00 -2.81412572e-01 7.14692056e-01
2.21555993e-01 -9.30122614e-01 -4.92602527e-01 -9.10023302e-02] | [12.894831657409668, 8.035649299621582] |
c786bc1a-f17b-4b14-96d6-a1d7240d4ff8 | memsac-memory-augmented-sample-consistency | 2207.12389 | null | https://arxiv.org/abs/2207.12389v1 | https://arxiv.org/pdf/2207.12389v1.pdf | MemSAC: Memory Augmented Sample Consistency for Large Scale Domain Adaptation | Practical real world datasets with plentiful categories introduce new challenges for unsupervised domain adaptation like small inter-class discriminability, that existing approaches relying on domain invariance alone cannot handle sufficiently well. In this work we propose MemSAC, which exploits sample level similarity across source and target domains to achieve discriminative transfer, along with architectures that scale to a large number of categories. For this purpose, we first introduce a memory augmented approach to efficiently extract pairwise similarity relations between labeled source and unlabeled target domain instances, suited to handle an arbitrary number of classes. Next, we propose and theoretically justify a novel variant of the contrastive loss to promote local consistency among within-class cross domain samples while enforcing separation between classes, thus preserving discriminative transfer from source to target. We validate the advantages of MemSAC with significant improvements over previous state-of-the-art on multiple challenging transfer tasks designed for large-scale adaptation, such as DomainNet with 345 classes and fine-grained adaptation on Caltech-UCSD birds dataset with 200 classes. We also provide in-depth analysis and insights into the effectiveness of MemSAC. | ['Manmohan Chandraker', 'Astuti Sharma', 'Tarun Kalluri'] | 2022-07-25 | null | null | null | null | ['fine-grained-visual-recognition'] | ['computer-vision'] | [ 3.14824522e-01 -2.15564221e-01 -2.81384587e-01 -7.77698040e-01
-8.68644118e-01 -1.01965892e+00 6.36131883e-01 1.48088112e-01
-6.08932197e-01 8.21149290e-01 3.66572365e-02 1.36931226e-01
-3.21985006e-01 -3.39941174e-01 -6.67726994e-01 -5.79808891e-01
-1.35387212e-01 9.10073936e-01 4.26290929e-01 -1.75207496e-01
-9.04377475e-02 4.26394969e-01 -1.50921082e+00 5.35707474e-01
9.01018322e-01 1.07741582e+00 -3.37627493e-02 1.82413444e-01
1.74223542e-01 3.94394130e-01 -4.67760414e-01 -3.15903932e-01
4.52230126e-01 -3.87092173e-01 -9.88532901e-01 1.10324845e-01
1.22475505e+00 -2.66106546e-01 -1.00592434e-01 8.76945853e-01
4.12807643e-01 4.07590836e-01 1.10339463e+00 -1.43408310e+00
-8.67146134e-01 4.63893205e-01 -4.30927485e-01 3.13627571e-01
-1.41011730e-01 5.76865934e-02 1.07455266e+00 -8.95237148e-01
7.00105727e-01 1.25806320e+00 7.92369545e-01 7.90855706e-01
-1.61000526e+00 -9.32598770e-01 4.19636309e-01 4.10095751e-01
-1.26761901e+00 -5.45866847e-01 6.94779515e-01 -6.27795875e-01
1.05842483e+00 -1.13739774e-01 2.45041087e-01 1.36859131e+00
-1.33790746e-01 7.67300308e-01 1.17926872e+00 -3.46561730e-01
2.62953728e-01 3.15166175e-01 3.49046171e-01 1.93859294e-01
1.80689335e-01 3.53765219e-01 -5.00798047e-01 -1.52277753e-01
5.37926137e-01 -7.28459731e-02 1.47157639e-01 -1.17116058e+00
-1.28314316e+00 9.83499527e-01 7.18179345e-01 2.47653693e-01
4.52523790e-02 -3.12272489e-01 6.54405892e-01 6.05697811e-01
5.42040884e-01 5.69750249e-01 -8.77847970e-01 2.65484929e-01
-8.88409078e-01 2.51372367e-01 6.02339625e-01 1.27181208e+00
7.52006710e-01 2.94924434e-02 -6.68666437e-02 1.21753883e+00
1.64345442e-03 5.45487583e-01 6.47676826e-01 -7.98293412e-01
3.62143368e-01 4.28657591e-01 -2.08071202e-01 -5.61360061e-01
-3.72849315e-01 -6.36347771e-01 -8.24318469e-01 2.37524480e-01
5.82395971e-01 1.00600027e-01 -9.76189554e-01 2.13668990e+00
3.61635655e-01 1.85449600e-01 2.58088142e-01 9.50092256e-01
5.85574329e-01 2.04305038e-01 3.78422201e-01 2.03755647e-01
1.13639557e+00 -1.02727866e+00 -1.34046704e-01 -4.96216089e-01
4.99411583e-01 -5.42247891e-01 1.21316969e+00 2.77280182e-01
-7.16259360e-01 -7.91178048e-01 -1.05748856e+00 -1.81892470e-01
-6.36516213e-01 2.54345775e-01 5.16992092e-01 3.44108731e-01
-8.66960108e-01 4.55820084e-01 -7.22470820e-01 -7.10263550e-01
7.38020539e-01 5.19719124e-01 -6.38389468e-01 -1.87317684e-01
-1.10224617e+00 1.02804697e+00 7.11558521e-01 -6.08822882e-01
-1.06499231e+00 -1.04646730e+00 -8.29319596e-01 6.00421391e-02
3.64722386e-02 -5.95554769e-01 1.28422761e+00 -1.47150552e+00
-1.28978384e+00 1.23659384e+00 1.68889657e-01 -7.45347321e-01
2.99927384e-01 -3.04127842e-01 -5.22929847e-01 4.81134318e-02
3.47951919e-01 1.14441168e+00 7.35552907e-01 -1.11433637e+00
-8.10662627e-01 -5.64978302e-01 -1.14579722e-01 3.17005306e-01
-7.43299782e-01 -2.02542439e-01 8.99604559e-02 -8.31819952e-01
-1.30631596e-01 -9.89922941e-01 -2.82387175e-02 1.47680283e-01
3.44560109e-02 -2.83083022e-01 8.95126462e-01 -3.38612229e-01
5.95716000e-01 -2.24126196e+00 3.54912728e-01 3.38567756e-02
4.25920375e-02 4.46388721e-01 -5.48198104e-01 1.58308670e-01
-2.00552329e-01 -5.21301806e-01 -6.20535016e-01 -2.35579178e-01
1.92210153e-01 3.42446387e-01 -5.06761253e-01 5.03110766e-01
4.17894810e-01 6.74676895e-01 -7.58660316e-01 -2.93862313e-01
8.66844282e-02 3.15751135e-01 -7.47020185e-01 2.38046080e-01
-1.16263390e-01 4.57536757e-01 -2.61889756e-01 4.96655613e-01
8.84567738e-01 -2.53957212e-01 1.89934164e-01 -7.74543956e-02
4.15779591e-01 2.44265318e-01 -9.27774429e-01 1.98236918e+00
-4.09293234e-01 5.80654502e-01 2.42355615e-01 -1.33844149e+00
9.53978956e-01 -6.86148778e-02 3.33671093e-01 -6.30765378e-01
-1.19054794e-01 3.25670958e-01 -1.65700354e-02 1.92470148e-01
3.31535190e-01 -4.61849988e-01 -2.99501628e-01 1.65758535e-01
7.95018971e-01 2.55901795e-02 7.80187696e-02 1.33955866e-01
8.72313201e-01 1.93133712e-01 5.63436568e-01 -6.81673586e-01
5.29025853e-01 2.50640601e-01 4.35108811e-01 5.45641363e-01
-4.92538720e-01 5.80727577e-01 -1.42255556e-02 -2.81776249e-01
-1.04403329e+00 -1.58108389e+00 -3.93069327e-01 1.69720328e+00
2.24621311e-01 1.39863521e-01 -6.27039492e-01 -1.25413322e+00
4.78845865e-01 6.00460052e-01 -7.78701782e-01 -4.30543482e-01
-3.25981885e-01 -4.98271614e-01 6.47405267e-01 9.07466054e-01
5.51344991e-01 -8.60927165e-01 -2.74208307e-01 1.67473286e-01
-5.89092486e-02 -1.25604904e+00 -5.59824824e-01 5.60656846e-01
-1.01378465e+00 -9.91961360e-01 -7.46950984e-01 -1.11764061e+00
5.89577377e-01 2.60196984e-01 1.39752972e+00 -5.93364775e-01
-2.15152755e-01 3.74530941e-01 -3.09807062e-01 -2.06746981e-01
-2.62712628e-01 3.73470306e-01 3.18343103e-01 -1.38528585e-01
8.10200393e-01 -7.58178830e-01 -3.38689476e-01 7.28413582e-01
-8.12381625e-01 -3.34486037e-01 7.57412076e-01 1.15863633e+00
6.34822428e-01 -3.17222685e-01 8.35823476e-01 -1.04355156e+00
2.56660998e-01 -6.55786395e-01 -4.82059747e-01 3.16169530e-01
-5.22962570e-01 6.44241124e-02 9.40925896e-01 -7.13134587e-01
-1.12538016e+00 2.84249634e-01 1.57412872e-01 -5.49847484e-01
-6.11558378e-01 1.04005389e-01 -2.86900610e-01 -1.27342701e-01
1.25380504e+00 3.04879725e-01 9.45293810e-03 -5.55725098e-01
5.54907143e-01 6.08237803e-01 6.16305888e-01 -7.94965088e-01
8.50521088e-01 6.41132236e-01 -3.14985484e-01 -5.93618691e-01
-9.55490053e-01 -8.33449841e-01 -1.33904755e+00 3.13114434e-01
6.73556745e-01 -1.22904181e+00 3.20482044e-03 4.31788266e-01
-7.60408282e-01 -6.23673856e-01 -5.85758150e-01 5.14493644e-01
-7.19072163e-01 1.61186263e-01 -4.50941026e-01 -7.27236792e-02
-1.30222395e-01 -6.87649429e-01 1.04829311e+00 -5.33448569e-02
-2.13085905e-01 -1.24116695e+00 2.86059439e-01 2.56616145e-01
4.49707568e-01 1.11924924e-01 7.40251005e-01 -1.27858984e+00
-1.53545350e-01 3.82818915e-02 -3.72462958e-01 6.46483898e-01
1.37322828e-01 -4.99799311e-01 -1.00641656e+00 -8.11852396e-01
-4.29566354e-01 -7.70404458e-01 1.01017606e+00 1.63339168e-01
9.99745250e-01 -1.62278667e-01 -4.89669293e-01 7.89018691e-01
1.30671227e+00 -4.29311134e-02 2.88229406e-01 3.56834799e-01
6.82145476e-01 6.94616854e-01 8.53784561e-01 2.82111079e-01
2.49240041e-01 7.98391223e-01 2.69951466e-02 -1.59619406e-01
-3.45614374e-01 -2.53722161e-01 3.73425066e-01 5.43935955e-01
3.73643011e-01 -5.66783734e-02 -8.58002722e-01 8.69576752e-01
-1.68423080e+00 -8.11845362e-01 4.23184007e-01 2.11300635e+00
8.50215018e-01 4.23017554e-02 3.77353609e-01 -2.71539897e-01
6.85262680e-01 -1.36088118e-01 -9.07073975e-01 -1.73905328e-01
-2.68220037e-01 3.71224463e-01 5.55921078e-01 2.90556908e-01
-1.52784431e+00 1.16703963e+00 6.29027891e+00 1.07920957e+00
-1.07881939e+00 2.73762226e-01 3.95625770e-01 -4.11025025e-02
3.16871628e-02 -2.18041658e-01 -9.13350224e-01 2.08055422e-01
8.47976327e-01 -1.36559516e-01 2.08559141e-01 1.34226286e+00
-5.50189078e-01 3.35875273e-01 -1.53537393e+00 7.68291891e-01
1.77242517e-01 -9.19652343e-01 1.82612553e-01 -1.15778662e-01
9.24463928e-01 2.39324957e-01 2.72605479e-01 6.19025528e-01
7.10911453e-01 -8.42141151e-01 6.06073439e-01 -2.04809025e-01
9.38036084e-01 -6.57931507e-01 3.90717983e-01 2.25099966e-01
-1.13287759e+00 -2.09177122e-01 -8.02278817e-01 2.94429492e-02
-3.29293013e-01 4.29627419e-01 -9.74058449e-01 4.66524065e-01
8.41922164e-01 1.18623924e+00 -7.19303668e-01 8.52930129e-01
3.80991958e-02 5.63372672e-01 -4.21286792e-01 3.19236487e-01
2.51927167e-01 7.12988898e-02 4.35701638e-01 1.50375664e+00
1.88513115e-01 -1.16643623e-01 2.98055559e-01 6.45385861e-01
-9.37317759e-02 3.82063687e-02 -6.44325852e-01 1.92239210e-01
6.26840115e-01 1.12658656e+00 -5.29230654e-01 -4.57313597e-01
-3.47804368e-01 1.16357398e+00 7.31502652e-01 4.46479380e-01
-8.15171301e-01 -2.05651119e-01 1.08697045e+00 1.18883722e-01
5.04298627e-01 -2.45525703e-01 -4.07634735e-01 -1.37123978e+00
-8.83442685e-02 -1.00012553e+00 9.65590358e-01 -2.58079380e-01
-2.05371070e+00 5.74921310e-01 1.95392445e-01 -1.57409227e+00
-1.57847717e-01 -8.64171028e-01 -1.94201693e-01 7.03277886e-01
-1.67602551e+00 -1.54559386e+00 -2.51196235e-01 1.05427766e+00
5.77474952e-01 -5.50605297e-01 9.90903974e-01 3.05689603e-01
-3.24516483e-02 1.05961454e+00 6.22193754e-01 1.88341632e-01
1.44581378e+00 -1.28136325e+00 3.42814028e-01 7.54604638e-01
2.49803305e-01 5.21602750e-01 3.92125845e-01 -2.86182821e-01
-8.09234619e-01 -1.42773461e+00 5.62232792e-01 -6.92380488e-01
6.09167099e-01 -5.93736768e-01 -1.06521201e+00 9.56996799e-01
9.27331820e-02 2.70725399e-01 8.32863271e-01 4.30826038e-01
-1.09868371e+00 -4.55842286e-01 -1.39650929e+00 2.08220407e-01
1.25343299e+00 -5.39674461e-01 -8.94884229e-01 2.46688619e-01
4.77728307e-01 -1.68246850e-01 -8.65980268e-01 5.78249812e-01
4.88247842e-01 -8.64764392e-01 1.23261917e+00 -8.97701263e-01
3.01839858e-01 -1.66443095e-01 -3.70555460e-01 -1.62304664e+00
-6.71288192e-01 1.90721992e-02 2.49582753e-01 1.15107512e+00
3.45461577e-01 -7.99830973e-01 7.49046147e-01 2.43442923e-01
-2.86527336e-01 -7.84651041e-02 -1.05035007e+00 -1.29770589e+00
8.13772023e-01 2.12272611e-02 3.22605073e-01 1.36307979e+00
-2.94506811e-02 4.59518582e-01 -1.63539812e-01 1.06379844e-01
8.80759239e-01 3.10703486e-01 6.80915177e-01 -1.43536282e+00
-3.82141113e-01 -4.01816219e-01 -6.13096774e-01 -1.08273876e+00
4.86963689e-01 -1.14318144e+00 1.00810833e-01 -9.76971686e-01
3.00361335e-01 -5.93642890e-01 -6.25898182e-01 6.72860026e-01
-4.42113951e-02 4.75952953e-01 1.07197590e-01 3.64120960e-01
-9.24531162e-01 7.12269127e-01 9.44874346e-01 -2.76661426e-01
-2.98073906e-02 -2.27079466e-01 -7.89687514e-01 4.19503003e-01
6.44975662e-01 -4.28842157e-01 -7.51595795e-01 -6.12851799e-01
-7.18422234e-01 -4.54981327e-01 4.73933071e-01 -1.35361981e+00
2.52079275e-02 -1.38654962e-01 6.67123258e-01 -1.66961297e-01
2.76953608e-01 -9.72219050e-01 -2.11899608e-01 2.32076675e-01
-7.39577591e-01 -3.24837178e-01 6.13319278e-01 6.63029969e-01
-4.53959316e-01 3.21625620e-01 1.44298303e+00 1.57693595e-01
-1.25890422e+00 3.97880137e-01 9.32581536e-03 5.03538072e-01
1.25719678e+00 -8.80316347e-02 -3.44606608e-01 6.05322197e-02
-9.73340333e-01 2.87971824e-01 6.12650454e-01 6.42169416e-01
4.17082638e-01 -1.55888999e+00 -8.80904257e-01 3.68160784e-01
6.13236785e-01 -1.75951928e-01 3.71390641e-01 4.11793381e-01
-3.75985801e-02 4.49644506e-01 -9.14569795e-01 -8.73531640e-01
-1.35089171e+00 8.15293431e-01 3.51060331e-01 -3.09350550e-01
-3.04688066e-01 1.12496006e+00 7.93586433e-01 -9.97230232e-01
8.79327059e-02 -1.77924067e-01 -8.06087181e-02 1.59767523e-01
3.34047467e-01 1.00578941e-01 1.69158176e-01 -7.54933059e-01
-7.47028470e-01 6.49711668e-01 -4.04569626e-01 3.39412764e-02
1.30592036e+00 -1.59058958e-01 3.42781454e-01 1.55409440e-01
1.41685867e+00 -4.66291219e-01 -1.73552024e+00 -6.42382562e-01
7.49839544e-02 -4.06408370e-01 -3.75760883e-01 -1.29180396e+00
-8.39148700e-01 1.12568152e+00 8.69330704e-01 -2.29651734e-01
1.32758713e+00 1.78220421e-01 6.20395958e-01 3.94212276e-01
4.13335234e-01 -1.19251800e+00 1.97806254e-01 6.59376144e-01
8.35976779e-01 -1.45529664e+00 -3.07660520e-01 -3.36665362e-01
-8.56328309e-01 9.02430236e-01 9.87066209e-01 -3.92906755e-01
5.65060198e-01 2.04987135e-02 1.05438210e-01 1.71165496e-01
-7.23972499e-01 -2.86236972e-01 5.11797667e-01 1.16426337e+00
3.61443579e-01 1.84362814e-01 1.37938485e-01 6.32510841e-01
-3.97509001e-02 -9.18410346e-02 -1.12715527e-01 7.78275609e-01
-4.54498231e-01 -1.25341475e+00 -1.60903722e-01 2.05922455e-01
-1.47243112e-01 -1.01708643e-01 -6.16342664e-01 9.33196187e-01
1.80799320e-01 5.65679908e-01 2.86575049e-01 -3.13793927e-01
5.11098623e-01 2.17717096e-01 6.66378677e-01 -7.35241771e-01
-4.30243462e-01 -1.71921730e-01 -4.63778488e-02 -4.14811552e-01
-5.67234993e-01 -8.27771842e-01 -1.00284338e+00 -8.77291411e-02
1.21490322e-01 9.97284651e-02 1.18777581e-01 7.60707736e-01
5.71735084e-01 1.48046955e-01 4.50846821e-01 -7.68774986e-01
-9.87147987e-01 -1.03309810e+00 -7.41141915e-01 8.92528832e-01
3.64788592e-01 -1.03631747e+00 -3.21914405e-01 2.96240121e-01] | [10.235920906066895, 2.878852605819702] |
dc4fd0f4-b60a-4a77-8589-a47665a4f26c | behavior-contrastive-learning-for | 2305.04477 | null | https://arxiv.org/abs/2305.04477v1 | https://arxiv.org/pdf/2305.04477v1.pdf | Behavior Contrastive Learning for Unsupervised Skill Discovery | In reinforcement learning, unsupervised skill discovery aims to learn diverse skills without extrinsic rewards. Previous methods discover skills by maximizing the mutual information (MI) between states and skills. However, such an MI objective tends to learn simple and static skills and may hinder exploration. In this paper, we propose a novel unsupervised skill discovery method through contrastive learning among behaviors, which makes the agent produce similar behaviors for the same skill and diverse behaviors for different skills. Under mild assumptions, our objective maximizes the MI between different behaviors based on the same skill, which serves as an upper bound of the previous MI objective. Meanwhile, our method implicitly increases the state entropy to obtain better state coverage. We evaluate our method on challenging mazes and continuous control tasks. The results show that our method generates diverse and far-reaching skills, and also obtains competitive performance in downstream tasks compared to the state-of-the-art methods. | ['Xuelong Li', 'Peng Liu', 'Zhen Wang', 'Bin Zhao', 'Siyuan Li', 'Hongyi Guo', 'Chenjia Bai', 'Rushuai Yang'] | 2023-05-08 | null | null | null | null | ['continuous-control'] | ['playing-games'] | [ 2.39038944e-01 2.38293037e-01 -3.15683424e-01 -1.01583060e-02
-3.61891866e-01 -5.86836696e-01 3.11990678e-01 1.04631737e-01
-6.63258791e-01 1.22812223e+00 -2.77941935e-02 -1.76659048e-01
-4.72612113e-01 -6.89034939e-01 -6.61527693e-01 -8.05898368e-01
-2.43342176e-01 3.76241416e-01 3.34060431e-01 -3.22664350e-01
3.58398348e-01 -2.77943723e-02 -1.43284786e+00 -4.20141965e-01
1.48669970e+00 5.89227796e-01 6.72473192e-01 4.01661813e-01
9.93002579e-02 1.19634402e+00 -4.71003234e-01 3.58089134e-02
3.93125504e-01 -8.80476713e-01 -8.39489639e-01 4.72408198e-02
-1.96273491e-01 -2.43704811e-01 -4.38610762e-01 1.26785886e+00
1.51588917e-01 4.64143187e-01 4.38915431e-01 -1.03059399e+00
-2.28153110e-01 9.98496830e-01 -4.84794080e-01 2.96599954e-01
2.26933494e-01 5.14431119e-01 9.15585279e-01 1.15238674e-01
3.93889695e-01 8.76926720e-01 6.17936999e-02 8.87188196e-01
-1.48455811e+00 -9.61302936e-01 6.05685174e-01 -1.11852162e-01
-9.82714236e-01 -1.70825198e-01 5.34925401e-01 -2.41288319e-01
5.96122563e-01 -1.81615919e-01 1.06099093e+00 9.87786233e-01
1.66534975e-01 9.67303157e-01 1.56006610e+00 -3.36395577e-02
5.18331468e-01 -1.31258234e-01 -1.29610404e-01 1.01757002e+00
4.59331810e-01 7.24658608e-01 -6.89541578e-01 1.66498646e-01
9.58997309e-01 1.05372101e-01 -1.39269948e-01 -5.42698681e-01
-1.31762064e+00 5.16302228e-01 2.20955655e-01 2.31495619e-01
-3.90181541e-01 3.52532685e-01 -6.34210631e-02 6.83199346e-01
-1.01306871e-01 1.00992429e+00 -5.41134536e-01 -7.39630759e-01
-6.29964650e-01 3.44812393e-01 7.41389513e-01 9.27453697e-01
9.79793608e-01 1.95363432e-01 -3.78489763e-01 3.62059623e-01
-1.12383133e-02 5.59685171e-01 4.74573702e-01 -1.21261346e+00
5.91692805e-01 6.96653962e-01 3.45347434e-01 -5.05489826e-01
-3.70623320e-01 -7.08201408e-01 -5.24377644e-01 3.56036454e-01
5.83163023e-01 -7.18208492e-01 -9.16017473e-01 2.18899274e+00
2.67302431e-02 4.72202092e-01 3.25342089e-01 8.33318949e-01
2.71109402e-01 5.05374372e-01 2.22816393e-02 -5.15720844e-01
4.97041225e-01 -9.66623843e-01 -7.40195632e-01 -5.81851065e-01
5.17979622e-01 4.77926806e-02 1.15663314e+00 3.16506267e-01
-1.28739262e+00 -4.12452519e-01 -8.89550149e-01 8.02231491e-01
3.15950632e-01 -1.90776944e-01 6.76340520e-01 3.65976602e-01
-8.72310758e-01 9.94665146e-01 -1.10280919e+00 -1.22093253e-01
4.72652018e-01 5.06808877e-01 4.80108634e-02 3.78209889e-01
-9.95009720e-01 8.13870609e-01 6.52668417e-01 -5.42230904e-01
-1.61591208e+00 -4.70712692e-01 -7.29084253e-01 3.29042017e-01
1.09126985e+00 -3.91035885e-01 1.47338319e+00 -8.73632491e-01
-2.20197701e+00 4.36480939e-01 2.99736056e-02 -5.46890676e-01
3.35354537e-01 -1.74486175e-01 -7.59055689e-02 6.97933324e-03
1.04137629e-01 6.58267498e-01 7.65645146e-01 -1.20927215e+00
-9.37188268e-01 -3.78839560e-02 5.84207654e-01 5.27420342e-01
-4.09694552e-01 -6.78370237e-01 -2.37379059e-01 -3.07474464e-01
9.68778655e-02 -9.76621866e-01 -6.49630308e-01 -5.97876012e-01
-3.55998188e-01 -2.10091576e-01 7.86695704e-02 -1.49620920e-01
1.30290580e+00 -2.04344130e+00 7.03685701e-01 3.52506459e-01
4.19916630e-01 -1.38160035e-01 -1.25725314e-01 2.55212098e-01
5.16040325e-01 -2.37189487e-01 -3.67451310e-01 2.40281802e-02
1.65799167e-03 3.70835185e-01 -1.37676015e-01 1.70426324e-01
-5.30287204e-03 9.82528687e-01 -1.42173982e+00 -4.47228849e-01
-1.85249954e-01 -5.64692378e-01 -6.13822937e-01 4.01161611e-01
-3.42564225e-01 9.89531636e-01 -9.04664099e-01 4.15624589e-01
1.39952242e-01 -5.14517963e-01 5.99144757e-01 7.07487524e-01
-1.23071648e-01 3.40909630e-01 -9.71780300e-01 1.81769323e+00
-4.19138730e-01 1.10806935e-01 -2.83566892e-01 -1.21849442e+00
9.86654580e-01 -1.65857282e-02 5.16458035e-01 -9.12960708e-01
1.83875203e-01 1.55979306e-01 5.80745935e-01 -3.37753236e-01
2.68947303e-01 -9.81069282e-02 -3.03971648e-01 5.39686263e-01
1.77755579e-01 -3.09116662e-01 5.32435715e-01 2.25422695e-01
1.22182226e+00 4.86054808e-01 4.11352694e-01 -5.44561386e-01
2.82651871e-01 5.34488335e-02 8.77360940e-01 1.18468976e+00
-3.05210352e-01 -1.65513545e-01 9.37189460e-01 1.34762004e-01
-6.03862524e-01 -1.28056633e+00 5.23670852e-01 9.39499259e-01
7.93863237e-01 -2.11760789e-01 -6.05676532e-01 -9.15283680e-01
1.82575826e-02 5.69222689e-01 -5.64077497e-01 -4.62116420e-01
-5.30497313e-01 -1.90340191e-01 3.72895867e-01 4.87375647e-01
7.36436963e-01 -1.33596730e+00 -9.86655295e-01 1.69566140e-01
5.47816453e-04 -6.44396424e-01 -5.73422968e-01 5.50234199e-01
-1.12226069e+00 -1.05670047e+00 -7.71956205e-01 -8.79422009e-01
7.68714488e-01 2.11650711e-02 9.64547992e-01 4.45672013e-02
2.38121703e-01 1.99297175e-01 -2.06453979e-01 -1.93698168e-01
-2.04228327e-01 3.10437232e-01 4.12313789e-01 -4.60991651e-01
2.02015638e-01 -7.07528234e-01 -6.41870618e-01 1.64330855e-01
-5.96425414e-01 2.49661833e-01 7.92650342e-01 1.03728104e+00
6.12200379e-01 1.99893296e-01 7.72317231e-01 -6.47907138e-01
9.24193978e-01 -4.80625838e-01 -9.32371438e-01 1.31606787e-01
-9.90082264e-01 6.83014989e-01 8.49992514e-01 -8.88753116e-01
-1.17778766e+00 4.57011200e-02 4.17870164e-01 -2.05739200e-01
-1.78127307e-02 5.48862100e-01 5.86190373e-02 8.59645456e-02
5.94012022e-01 7.30759025e-01 1.76058009e-01 -3.09281856e-01
2.04937346e-02 5.46166562e-02 2.65152663e-01 -8.98228586e-01
8.95362675e-01 2.13398248e-01 1.92081109e-01 -2.38485396e-01
-9.92932975e-01 -9.45309326e-02 -1.72168836e-01 -2.12858215e-01
4.80268806e-01 -6.99533939e-01 -1.24835336e+00 3.64392966e-01
-3.06360930e-01 -1.02802479e+00 -4.76008236e-01 6.22809291e-01
-9.54084814e-01 1.95915520e-01 -5.52655041e-01 -9.74171877e-01
4.17063907e-02 -1.24989116e+00 4.19493616e-01 7.74221539e-01
-7.44094402e-02 -7.32243657e-01 2.85141408e-01 -1.35526195e-01
2.79048294e-01 -5.10596158e-03 5.62674940e-01 -3.07470441e-01
-7.08671808e-01 4.73287255e-01 1.81807891e-01 -4.45242710e-02
3.25040996e-01 -6.84426188e-01 -1.82171836e-01 -5.76730549e-01
-3.20758313e-01 -7.71268845e-01 9.04621601e-01 2.81194478e-01
1.23091722e+00 -3.61297518e-01 -3.57195467e-01 4.03389990e-01
1.07260466e+00 6.86804235e-01 4.21612680e-01 5.06248295e-01
3.39952826e-01 5.01886249e-01 9.55735266e-01 5.81213892e-01
1.06117375e-01 3.43832254e-01 4.60562825e-01 3.47404242e-01
4.60559636e-01 -5.90664983e-01 5.11189163e-01 5.79718173e-01
-2.61283457e-01 6.01967610e-02 -6.67409003e-01 6.39331281e-01
-2.11605120e+00 -1.02559185e+00 5.63362598e-01 2.32574654e+00
1.28396595e+00 5.55376470e-01 4.19011056e-01 -2.22183451e-01
2.54787356e-01 -3.51886116e-02 -1.36261165e+00 -4.61436585e-02
3.02729428e-01 4.03254658e-01 4.63054270e-01 3.64068002e-01
-7.48529971e-01 1.29470241e+00 6.19191599e+00 7.96848357e-01
-8.06049883e-01 -8.64843726e-02 4.08102363e-01 -4.99142945e-01
-1.97882622e-01 5.96468411e-02 -6.15189552e-01 6.40602171e-01
5.10225594e-01 -5.14061332e-01 8.80452573e-01 8.02777708e-01
1.57893524e-01 -3.21644634e-01 -1.07443631e+00 6.39291465e-01
-5.66918612e-01 -1.06805086e+00 -4.48858649e-01 1.17857032e-01
1.17282486e+00 -1.80733487e-01 1.44683838e-01 7.18721390e-01
9.82329071e-01 -9.65828717e-01 5.79055727e-01 6.43235862e-01
7.63792753e-01 -1.11802471e+00 4.95144010e-01 7.60305941e-01
-9.91289556e-01 -3.35885704e-01 -2.01097086e-01 -4.35753047e-01
2.13813409e-02 9.95001644e-02 -2.72506684e-01 1.44547805e-01
4.97755527e-01 7.69838095e-01 -1.96181595e-01 8.50927174e-01
-6.49676740e-01 6.73466921e-01 -1.62582740e-01 -7.56795943e-01
5.40796578e-01 -5.25231540e-01 5.87090850e-01 5.70093632e-01
1.55943602e-01 1.53984129e-01 6.12322867e-01 1.06821597e+00
3.62291187e-02 -1.84540346e-01 -6.02165163e-01 -5.28388560e-01
5.27763605e-01 9.20757949e-01 -6.80687904e-01 -3.51245642e-01
7.21637830e-02 8.77239525e-01 7.52660334e-01 4.39235151e-01
-6.76881075e-01 -3.40999573e-01 9.06176567e-01 -2.54597157e-01
1.17168017e-01 -3.20311695e-01 -3.25759858e-01 -1.16771770e+00
-8.89835432e-02 -8.79100502e-01 2.29224578e-01 -2.00811237e-01
-8.73399854e-01 3.34290266e-01 1.13944747e-02 -1.38684857e+00
-4.12093580e-01 -2.44401455e-01 -5.37226915e-01 6.07752383e-01
-1.44596744e+00 -1.63841665e-01 -2.33697206e-01 4.77241933e-01
3.92021686e-01 -5.26598394e-01 4.88916993e-01 -2.85532624e-01
-5.40938079e-01 6.49577558e-01 2.34020114e-01 -1.99191004e-01
3.38237882e-01 -1.44011140e+00 8.26643780e-02 7.29312241e-01
-9.48783010e-02 6.88276291e-01 6.85197413e-01 -8.91214371e-01
-1.25944269e+00 -7.02664971e-01 1.66774899e-01 5.12010865e-02
7.58756220e-01 4.31588143e-02 -5.54973841e-01 5.30231595e-01
-1.37833813e-02 -6.51056409e-01 2.74465382e-01 2.51951396e-01
1.51186332e-01 5.72819356e-03 -8.83261561e-01 1.15125394e+00
1.55601943e+00 -6.54301792e-02 -5.91741860e-01 4.44447510e-02
8.63615572e-01 -5.14890075e-01 -4.94093090e-01 3.01013112e-01
4.56662804e-01 -7.71395326e-01 6.22501612e-01 -9.02440488e-01
8.24625373e-01 -1.95958585e-01 3.88659835e-01 -1.80811703e+00
-4.71906483e-01 -8.81080568e-01 -5.94608486e-01 5.31784534e-01
6.15409553e-01 -7.07464755e-01 1.02338207e+00 1.47273555e-01
-2.64397897e-02 -9.63560760e-01 -6.29742146e-01 -1.25326502e+00
1.80413753e-01 1.25673220e-01 6.52780771e-01 7.12192416e-01
5.56201160e-01 2.38964453e-01 -6.61846876e-01 -4.30124477e-02
7.73441374e-01 4.32971150e-01 6.79527223e-01 -1.17686987e+00
-7.96711206e-01 -5.77386618e-01 1.44073501e-01 -1.64494336e+00
3.65577579e-01 -6.94468141e-01 3.55639994e-01 -1.31653726e+00
5.02540410e-01 -5.71689904e-01 -5.53739011e-01 5.25757551e-01
-5.10278165e-01 -3.53607416e-01 2.55479544e-01 1.28471062e-01
-1.03954494e+00 7.22721994e-01 1.92239952e+00 -1.19990304e-01
-7.43124485e-01 1.61796406e-01 -1.01627326e+00 5.64616680e-01
1.25525641e+00 -5.15847385e-01 -7.13935435e-01 -1.18722893e-01
3.41591358e-01 3.99399489e-01 -1.21060863e-01 -1.07925439e+00
1.62705302e-01 -8.29793334e-01 1.01467408e-01 -2.32379243e-01
2.12723151e-01 -5.81174731e-01 -2.41077304e-01 1.16186273e+00
-6.66919827e-01 -2.56694406e-01 3.02793905e-02 9.04965580e-01
1.40566811e-01 -3.45988363e-01 6.07318401e-01 -3.24378848e-01
-7.36175537e-01 3.78656626e-01 -5.87278366e-01 3.10575396e-01
1.17606843e+00 -1.17323197e-01 -2.07679093e-01 -6.44370198e-01
-7.33379662e-01 8.81199121e-01 3.84871095e-01 3.40435326e-01
6.14511490e-01 -1.13587141e+00 -5.66428781e-01 3.23861726e-02
1.23781182e-01 2.36787479e-02 7.78610408e-02 8.14335942e-01
-1.07092142e-01 1.99722290e-01 -5.02158046e-01 -3.10325801e-01
-9.66065824e-01 4.87554193e-01 3.23175341e-01 -7.13946521e-01
-5.38647473e-01 6.44916892e-01 2.96878159e-01 -4.67424333e-01
2.69167721e-01 -2.14343473e-01 -3.69583964e-01 -4.56598252e-01
2.87932336e-01 3.75430346e-01 -7.20813274e-01 2.01956838e-01
-1.54092133e-01 2.20901847e-01 -1.56048343e-01 -3.38686019e-01
1.12250876e+00 -3.62003148e-02 3.58978242e-01 4.40242469e-01
4.82202142e-01 -2.29973882e-01 -1.94742906e+00 -4.08988237e-01
6.90201968e-02 -4.56775457e-01 -1.50596485e-01 -8.41693759e-01
-9.68495965e-01 5.04949868e-01 4.32125211e-01 1.46991923e-01
1.01724386e+00 3.25384289e-02 6.31495237e-01 8.28312457e-01
8.99841309e-01 -1.41142273e+00 6.74209118e-01 8.28833103e-01
2.11127773e-01 -1.32802629e+00 -1.50988132e-01 -1.65793821e-02
-1.03057587e+00 6.29963756e-01 1.22805083e+00 -2.81920463e-01
1.02337532e-01 2.28097647e-01 -2.95990884e-01 -1.12127408e-01
-7.58262634e-01 -7.54905760e-01 -1.25142820e-02 5.15237808e-01
3.81279252e-02 1.86780661e-01 -5.51659465e-01 5.46391249e-01
-1.37655675e-01 9.86069068e-02 3.13594073e-01 1.23995662e+00
-7.41118610e-01 -1.11520910e+00 1.46448731e-01 7.68782496e-01
-3.41890335e-01 -6.84960112e-02 -2.06537440e-01 5.69313467e-01
-2.32780546e-01 9.22564983e-01 -5.85364439e-02 -5.83089650e-01
1.25816360e-01 -2.15606347e-01 7.98227608e-01 -7.48000622e-01
-3.59493703e-01 -1.66409999e-01 -1.03359513e-01 -7.24761367e-01
-2.10358754e-01 -6.98757827e-01 -1.53966391e+00 -1.86398432e-01
-2.01045841e-01 3.77681226e-01 -5.70182018e-02 1.16377676e+00
2.15595156e-01 9.16594982e-01 8.14522207e-01 -4.05864507e-01
-9.50982749e-01 -9.47565258e-01 -7.91164935e-01 3.93663526e-01
2.49029353e-01 -1.02924073e+00 -1.86464876e-01 -2.66092747e-01] | [4.051044940948486, 1.779664158821106] |
bcb9c68e-f2ae-467b-bb8d-1d486a2e6c43 | multi-xscience-a-large-scale-dataset-for | 2010.14235 | null | https://arxiv.org/abs/2010.14235v1 | https://arxiv.org/pdf/2010.14235v1.pdf | Multi-XScience: A Large-scale Dataset for Extreme Multi-document Summarization of Scientific Articles | Multi-document summarization is a challenging task for which there exists little large-scale datasets. We propose Multi-XScience, a large-scale multi-document summarization dataset created from scientific articles. Multi-XScience introduces a challenging multi-document summarization task: writing the related-work section of a paper based on its abstract and the articles it references. Our work is inspired by extreme summarization, a dataset construction protocol that favours abstractive modeling approaches. Descriptive statistics and empirical results---using several state-of-the-art models trained on the Multi-XScience dataset---reveal that Multi-XScience is well suited for abstractive models. | ['Laurent Charlin', 'Yue Dong', 'Yao Lu'] | 2020-10-27 | null | https://aclanthology.org/2020.emnlp-main.648 | https://aclanthology.org/2020.emnlp-main.648.pdf | emnlp-2020-11 | ['extreme-summarization'] | ['natural-language-processing'] | [ 3.50358844e-01 1.89355791e-01 -7.43393242e-01 6.38091490e-02
-1.34283125e+00 -4.59525526e-01 8.12375724e-01 8.57693613e-01
-3.69384885e-02 1.15521657e+00 1.13152611e+00 -1.46026956e-03
-3.29341114e-01 -3.88636738e-01 -8.27719688e-01 -2.41330341e-01
2.44281366e-01 4.29020077e-01 -1.67803526e-01 -2.78502971e-01
1.23663878e+00 5.97279146e-02 -1.42877972e+00 7.20934987e-01
1.38475347e+00 3.75499845e-01 1.26019880e-01 1.17232633e+00
-4.88577008e-01 6.71489716e-01 -1.33648503e+00 -3.65179688e-01
-1.92363128e-01 -6.93059325e-01 -9.82669592e-01 -3.19760442e-01
1.14965272e+00 1.33403391e-01 -7.87667707e-02 6.44179523e-01
6.99400127e-01 -1.08519517e-01 9.49012101e-01 -1.11737597e+00
-6.90841198e-01 1.15192997e+00 -8.71045649e-01 5.04870594e-01
6.89562321e-01 -1.94210932e-01 1.26474392e+00 -3.89946610e-01
1.00422323e+00 1.25175595e+00 5.99472821e-01 3.41244221e-01
-8.93091381e-01 -2.60648519e-01 -8.43664333e-02 -2.96529457e-02
-7.64962196e-01 -4.89654928e-01 9.27844882e-01 -2.34843493e-01
1.39922750e+00 9.02738333e-01 6.85765386e-01 1.42318380e+00
8.21858704e-01 1.05299771e+00 7.15455711e-01 -5.81012189e-01
1.77400082e-01 -4.48711276e-01 7.22167552e-01 2.53681242e-01
9.66772914e-01 -9.63976622e-01 -8.73776078e-01 -3.58816117e-01
-5.09677157e-02 -8.35198984e-02 -1.88251100e-02 5.17057240e-01
-1.58944464e+00 5.50181985e-01 -2.62962490e-01 5.16221285e-01
-7.22328961e-01 1.40847430e-01 9.90064144e-01 2.64187962e-01
8.31898868e-01 1.01175368e+00 -4.51528490e-01 -4.81098622e-01
-1.76932442e+00 7.79206336e-01 1.14344406e+00 1.17940760e+00
1.69751927e-01 2.61868656e-01 -7.70444334e-01 8.10173452e-01
-1.06749155e-01 4.55385059e-01 7.35047162e-01 -9.85278130e-01
9.79981899e-01 9.52811956e-01 -2.40917400e-01 -9.37616825e-01
-3.57136309e-01 -6.09456539e-01 -1.01073015e+00 -6.73647523e-01
-5.77399909e-01 -1.25397414e-01 -4.96020496e-01 9.41592872e-01
-1.37156367e-01 -2.34204426e-01 4.04640049e-01 5.15065826e-02
1.88468146e+00 1.04361141e+00 -7.78841302e-02 -1.04143202e+00
1.31987655e+00 -1.20231700e+00 -1.12993622e+00 -2.15053856e-02
4.42884982e-01 -7.00715423e-01 8.07012856e-01 6.01030290e-01
-1.63553560e+00 -3.89745265e-01 -1.35490727e+00 -2.13108763e-01
-5.16625166e-01 4.06211406e-01 2.04766691e-01 1.02245525e-01
-8.96785855e-01 6.54109716e-01 -3.48883450e-01 -7.18406618e-01
5.58145583e-01 -1.86856210e-01 -2.17435747e-01 2.59634227e-01
-6.20066404e-01 1.02769017e+00 6.98779941e-01 -6.18983924e-01
-3.90689313e-01 -1.14906323e+00 -4.14603978e-01 1.47849694e-01
5.14860988e-01 -1.25403857e+00 1.36404240e+00 3.51471044e-02
-1.31938088e+00 7.60533690e-01 -3.81987363e-01 -5.86828887e-01
3.55091274e-01 -7.05347121e-01 -3.94595355e-01 5.70904732e-01
3.51280451e-01 -1.36036361e-02 7.88035274e-01 -1.40752637e+00
-4.75414753e-01 -4.25881565e-01 -3.74145329e-01 1.45883590e-01
-5.11657059e-01 3.49633574e-01 -5.52518189e-01 -1.20502949e+00
-4.41139162e-01 -5.87238491e-01 -1.84936404e-01 -1.07865858e+00
-1.39854276e+00 -4.75791931e-01 7.93096721e-01 -7.01736331e-01
2.02088380e+00 -1.34773862e+00 3.39233637e-01 -5.48706412e-01
4.49174911e-01 2.83324391e-01 -9.62827131e-02 1.16603065e+00
1.12908915e-01 5.99539042e-01 -3.79959196e-02 -5.24486899e-01
-2.22820807e-02 -1.65797248e-01 -7.10201144e-01 4.57489714e-02
-2.08477095e-01 1.08715343e+00 -8.32603157e-01 -9.50674891e-01
-1.47778317e-01 -2.33903781e-01 -5.82351349e-02 -1.31177222e-02
-4.72322196e-01 -5.78957498e-02 -8.63729775e-01 6.77419186e-01
4.64170456e-01 -1.36835799e-01 -3.26343060e-01 -2.24614054e-01
-4.77149099e-01 2.94602569e-02 -6.23366535e-01 2.01143837e+00
-1.03963211e-01 6.83468640e-01 -3.65131706e-01 -9.93698359e-01
6.86629891e-01 1.50399163e-01 6.02229238e-01 -3.24079305e-01
1.12450138e-01 3.44943851e-01 -5.55894494e-01 -6.72791004e-01
1.31299269e+00 4.34402943e-01 -5.02450109e-01 5.21027505e-01
1.10233523e-01 -7.74177790e-01 1.16402900e+00 1.03637886e+00
1.23073471e+00 -1.61342815e-01 8.31109703e-01 -4.71267104e-01
3.25368822e-01 4.05211687e-01 3.19894940e-01 1.20687318e+00
4.53837901e-01 8.62443984e-01 6.44587815e-01 -1.40186414e-01
-1.05069280e+00 -7.25525796e-01 5.57308793e-02 8.30862641e-01
-2.99399346e-01 -1.29091561e+00 -8.64261270e-01 -5.66400349e-01
1.86871514e-01 1.24372602e+00 -7.54162729e-01 -1.37621552e-01
-4.59347636e-01 -8.79855335e-01 7.54228711e-01 2.00685576e-01
3.47215652e-01 -1.07415068e+00 -6.95254624e-01 2.29274645e-01
-2.38426641e-01 -7.79587746e-01 -4.58694339e-01 -7.51352906e-02
-1.15839505e+00 -1.01411760e+00 -8.85626554e-01 -5.04256308e-01
2.79754966e-01 2.36101970e-01 1.31345809e+00 -2.76206434e-01
-3.74606520e-01 5.73626935e-01 -5.47311425e-01 -1.08028805e+00
-9.31751132e-01 4.91917908e-01 -1.03536636e-01 -6.53336227e-01
-3.38430852e-02 -3.19598824e-01 -1.62422895e-01 -6.66475594e-01
-1.08105493e+00 1.60580710e-01 8.79350185e-01 2.93132514e-01
5.21166801e-01 -1.50132284e-01 1.05796123e+00 -1.09973848e+00
1.39984071e+00 -4.22769397e-01 2.65082657e-01 6.73556447e-01
-6.54837251e-01 -6.31942675e-02 6.21674895e-01 -1.82653472e-01
-9.43963170e-01 -8.77893388e-01 2.57118553e-01 -8.11717007e-03
2.18925804e-01 1.06679094e+00 8.46583471e-02 5.89224339e-01
7.76770592e-01 4.61314023e-01 -1.89573377e-01 -7.37734675e-01
4.82999414e-01 9.16831315e-01 7.69324303e-01 -5.27884483e-01
3.52867395e-01 1.31060272e-01 2.58737691e-02 -1.54328203e+00
-1.13576031e+00 -5.16049862e-01 -3.79920453e-01 -1.92132518e-01
6.90902412e-01 -8.11246395e-01 -3.26241046e-01 4.39444959e-01
-1.46039045e+00 2.99276531e-01 -5.50382018e-01 1.12685829e-01
-3.00898224e-01 5.97091377e-01 -3.80298674e-01 -4.48730826e-01
-1.37151277e+00 -4.96833593e-01 1.24888003e+00 4.73672271e-01
-8.85116816e-01 -8.64174664e-01 4.05928969e-01 5.81550717e-01
4.34646934e-01 6.73680902e-01 7.42079914e-01 -9.97067451e-01
1.32514730e-01 -4.21609968e-01 1.11107707e-01 1.54986739e-01
-1.75935514e-02 8.55355442e-01 -3.53062570e-01 -2.76490033e-01
-7.04673901e-02 -3.45939308e-01 1.33306348e+00 1.04846621e+00
1.41341233e+00 -9.37434614e-01 -6.14493251e-01 -1.62177011e-02
1.27999032e+00 -2.09923148e-01 4.39044863e-01 6.49089277e-01
8.84670913e-01 2.62603968e-01 5.23485899e-01 9.63362992e-01
7.21262455e-01 2.04809189e-01 -7.03281760e-02 3.58392775e-01
-2.35927910e-01 -1.72681898e-01 1.50825232e-01 1.64962673e+00
-9.18011516e-02 -8.39111805e-01 -7.40283728e-01 6.74883306e-01
-2.06589413e+00 -1.28408325e+00 -7.91306198e-01 1.48750365e+00
1.16345239e+00 1.63130760e-01 1.41490623e-01 1.60956159e-02
4.48927194e-01 6.52694404e-01 -3.79221439e-01 -8.75310242e-01
-6.26871705e-01 6.78737611e-02 1.34540260e-01 -2.93307435e-02
-9.00557578e-01 6.04385674e-01 6.96159935e+00 1.11417043e+00
-8.35192025e-01 -2.06881791e-01 1.90694317e-01 -4.78542119e-01
-3.81596088e-01 2.14092061e-02 -9.61441040e-01 5.68228900e-01
1.28154266e+00 -1.23297060e+00 -5.07351220e-01 6.85614765e-01
4.63978410e-01 -4.46857721e-01 -9.87476051e-01 8.34639966e-01
7.99029410e-01 -2.11971664e+00 8.86794150e-01 -9.09679979e-02
1.14522898e+00 4.39274944e-02 -2.61824459e-01 2.54516631e-01
1.12648681e-01 -5.35509825e-01 8.77858639e-01 8.62820566e-01
6.38742507e-01 -4.88980442e-01 5.56043148e-01 4.00984615e-01
-6.67238712e-01 -1.40166115e-02 -2.54857719e-01 2.81588107e-01
1.84230775e-01 9.30285454e-01 -2.86317617e-01 1.32544708e+00
5.92295587e-01 1.36586857e+00 -9.64533985e-01 1.05923796e+00
1.18859142e-01 8.23150218e-01 1.89376295e-01 -4.39078361e-01
-1.07077040e-01 -1.81118175e-01 1.15551412e+00 1.92143071e+00
4.59964305e-01 -2.64295395e-02 2.33031474e-02 5.58102965e-01
-4.72336680e-01 3.25445056e-01 -5.68800449e-01 -5.45445263e-01
4.52832699e-01 1.19473839e+00 -4.41033661e-01 -9.85310793e-01
-4.85366657e-02 6.71269536e-01 -1.65256962e-01 1.15401730e-01
-4.03972626e-01 -8.15239191e-01 -1.36306316e-01 -1.47690535e-01
1.38616726e-01 4.56372462e-02 -9.50065613e-01 -1.33212495e+00
1.74966261e-01 -1.05257308e+00 4.68377471e-01 -9.71641958e-01
-1.05918705e+00 4.39655334e-01 4.86640155e-01 -1.25670493e+00
-1.28184199e-01 2.67632246e-01 -1.23002052e+00 3.20746005e-01
-1.16238225e+00 -1.52596605e+00 -1.68930858e-01 -1.82783514e-01
1.11842334e+00 -6.08811021e-01 6.75285220e-01 -3.21039736e-01
-8.98461819e-01 2.48625651e-01 7.42678642e-01 -5.52970171e-01
7.92886257e-01 -1.47363043e+00 4.05380458e-01 8.12945068e-01
-1.67810678e-01 9.00589526e-01 1.26437843e+00 -1.10468900e+00
-1.77289224e+00 -1.00956810e+00 1.31691492e+00 -6.69625938e-01
7.16617823e-01 3.13571990e-01 -8.60953510e-01 3.36614698e-01
1.06086159e+00 -9.64835465e-01 9.42236304e-01 3.80788185e-02
9.30388197e-02 -1.28957510e-01 -7.88266122e-01 4.45730120e-01
7.52071619e-01 -2.78108623e-02 -1.52837694e+00 5.47443151e-01
8.06222618e-01 -2.95770526e-01 -1.02491701e+00 2.01637223e-01
3.94919515e-01 -4.61917549e-01 7.51787007e-01 -5.92771709e-01
1.45601833e+00 1.24661811e-01 1.70476988e-01 -1.78708076e+00
-5.45225963e-02 -8.16827953e-01 -7.74326324e-01 1.64001524e+00
3.87857109e-01 -2.02943325e-01 4.92552936e-01 7.12361708e-02
-7.04400718e-01 -6.14023626e-01 -6.06609106e-01 -7.44573891e-01
3.33180726e-01 -3.27757783e-02 3.44619006e-01 8.59559059e-01
3.11302960e-01 8.70490730e-01 -1.24775298e-01 -6.07649565e-01
8.17782581e-01 5.04459620e-01 1.23204279e+00 -1.57170475e+00
3.09950858e-01 -9.10575449e-01 2.11517558e-01 -5.27711391e-01
2.44603127e-01 -8.23463082e-01 -4.87923443e-01 -2.57456350e+00
7.50529945e-01 6.95680559e-01 2.87666321e-01 3.76557320e-01
-5.25638044e-01 -3.56100559e-01 1.33723497e-01 5.36922216e-01
-1.30759883e+00 7.97903240e-01 1.25126302e+00 -5.75955629e-01
-2.55435169e-01 -1.40933067e-01 -1.39570940e+00 4.05674905e-01
5.75634956e-01 -4.81642872e-01 -2.32512072e-01 -1.17916323e-01
2.41274446e-01 1.07144780e-01 -1.35735646e-01 -9.70539391e-01
3.08435172e-01 -1.98505238e-01 1.35417700e-01 -1.44998825e+00
-1.45344898e-01 1.35599539e-01 -3.24092023e-02 4.36433762e-01
-7.62080908e-01 3.70675981e-01 5.09556413e-01 5.36002696e-01
-3.00410956e-01 -3.62205923e-01 2.07152635e-01 -3.63461882e-01
-1.82373002e-01 -5.81912585e-02 -4.90149468e-01 6.04820013e-01
8.29905391e-01 -2.01083407e-01 -1.00484359e+00 -9.95766968e-02
-1.72962900e-02 4.16676164e-01 4.04107392e-01 7.50147164e-01
7.12142289e-01 -8.16939056e-01 -1.53245866e+00 -4.10495907e-01
2.75154769e-01 3.10345180e-02 2.71243811e-01 6.20761335e-01
-6.33325696e-01 7.41144657e-01 -2.20793918e-01 -2.47415602e-01
-1.46302247e+00 2.98733681e-01 -6.66698575e-01 -6.48363233e-01
-5.95092595e-01 2.68143624e-01 -5.46690345e-01 1.52066229e-02
-1.29317075e-01 -4.99184251e-01 -6.75694168e-01 6.66575432e-01
7.59542823e-01 1.01975143e+00 3.43580484e-01 -4.82216269e-01
-1.00913644e-01 4.83469605e-01 -5.11103809e-01 6.13110140e-02
1.97136772e+00 -1.00737229e-01 -7.40913391e-01 8.33190262e-01
1.22175121e+00 1.85274884e-01 -3.15899611e-01 -4.95762639e-02
2.19519019e-01 -1.90187397e-03 2.18527138e-01 -9.67163086e-01
-3.12855542e-01 2.12097704e-01 -4.59690303e-01 5.98379135e-01
7.09024489e-01 -4.98659015e-02 7.37206995e-01 5.89828610e-01
-2.31873274e-01 -1.59604168e+00 3.90603721e-01 6.21635556e-01
1.66781974e+00 -1.06668663e+00 1.17207909e+00 1.53107241e-01
-8.76939058e-01 1.19645405e+00 3.25891554e-01 4.57566828e-02
2.77846366e-01 -5.34502640e-02 -3.59398156e-01 -5.88340402e-01
-1.10607111e+00 5.66870689e-01 4.81219471e-01 3.18762287e-02
3.34334731e-01 -4.85406071e-02 -8.81210864e-01 8.99096668e-01
-5.61386287e-01 9.21800062e-02 1.17133176e+00 1.34498417e+00
-4.07534838e-01 -6.56249046e-01 -4.20288146e-01 1.31466103e+00
-7.54234374e-01 3.06982920e-03 -1.06969130e+00 5.71052492e-01
-6.47939861e-01 1.17655075e+00 -2.64211714e-01 -5.58661781e-02
5.82325339e-01 -2.03873679e-01 3.99335384e-01 -7.58608043e-01
-1.01954567e+00 -7.69600878e-03 2.65037328e-01 -4.30918261e-02
-6.95959270e-01 -9.49291766e-01 -1.11941040e+00 -6.01333499e-01
-1.59856468e-01 4.42729563e-01 8.60094011e-01 8.16026986e-01
7.48602092e-01 1.06722796e+00 4.14611369e-01 -1.05262971e+00
-5.42786837e-01 -1.42081022e+00 -6.42220438e-01 7.75332674e-02
4.96033490e-01 -1.99380517e-03 -2.94273794e-01 3.33910316e-01] | [12.435452461242676, 9.523080825805664] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.