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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
ef122c07-c6c3-46f2-9d97-5c7cdc399de8 | exploring-the-behavior-of-classic-reg | null | null | https://aclanthology.org/W17-3507 | https://aclanthology.org/W17-3507.pdf | Exploring the Behavior of Classic REG Algorithms in the Description of Characters in 3D Images | Describing people and characters can be very useful in different contexts, such as computational narrative or image description for the visually impaired. However, a review of the existing literature shows that the automatic generation of people descriptions has not received much attention. Our work focuses on the description of people in snapshots from a 3D environment. First, we have conducted a survey to identify the way in which people describe other people under different conditions. We have used the information extracted from this survey to design several Referring Expression Generation algorithms which produce similar results. We have evaluated these algorithms with users in order to identify which ones generate the best description for specific characters in different situations. The evaluation has shown that, in order to generate good descriptions, a combination of different algorithms has to be used depending on the features and situation of the person to be described. | ["Teresa Rodr{\\'\\i}guez", "Adri{\\'a}n Rabad{\\'a}n", "Raquel Herv{\\'a}s", "Gonzalo M{\\'e}ndez", 'Susana Bautista'] | 2017-09-01 | null | null | null | ws-2017-9 | ['referring-expression-generation'] | ['computer-vision'] | [-1.55843616e-01 -1.10438310e-01 1.26143977e-01 -4.34454352e-01
-1.58589706e-01 -4.89648730e-01 9.69947219e-01 2.38994777e-01
-3.57999295e-01 7.46495128e-01 8.06406915e-01 2.02694148e-01
2.10141484e-02 -4.92127120e-01 1.88311949e-01 -3.98295641e-01
3.70771915e-01 8.36516976e-01 3.65490615e-01 -3.76502007e-01
4.62723523e-01 7.35968471e-01 -1.93311381e+00 4.37775493e-01
4.22725141e-01 4.86880094e-02 3.20920527e-01 6.76289797e-01
-5.31052649e-01 6.20677769e-01 -1.06072855e+00 -3.75340790e-01
-1.50690287e-01 -7.21021414e-01 -8.40395033e-01 6.01191223e-01
2.58372068e-01 -3.15940052e-01 -5.38653992e-02 8.65230143e-01
1.02692258e+00 1.93804950e-01 9.27475691e-01 -1.05622280e+00
-3.98881614e-01 4.05754685e-01 1.25284716e-01 -7.99968019e-02
1.26497197e+00 1.23300724e-01 3.42495471e-01 -4.32542711e-01
9.23402905e-01 1.52480268e+00 3.51575136e-01 8.20398152e-01
-1.06297970e+00 -2.91599572e-01 -7.26410225e-02 2.77451754e-01
-1.66505861e+00 -4.05447513e-01 6.50834262e-01 -7.69641578e-01
7.53109574e-01 2.65847266e-01 1.06377864e+00 1.15169144e+00
-2.87061661e-01 5.15592277e-01 1.23501313e+00 -9.70925152e-01
2.54699230e-01 1.01746297e+00 1.59842938e-01 4.00085539e-01
3.18102658e-01 -3.87476832e-01 -3.84005219e-01 -1.88137844e-01
7.09547222e-01 -7.12711155e-01 -2.94664174e-01 -3.65842313e-01
-1.02726936e+00 7.24083602e-01 -2.19253212e-01 8.73346984e-01
-3.86608750e-01 3.42030115e-02 6.44864976e-01 -1.63009807e-01
1.63812414e-01 5.62888503e-01 2.46219844e-01 -3.48184705e-01
-6.11281276e-01 7.18233585e-01 1.10539162e+00 9.86226320e-01
2.81559139e-01 -3.39077055e-01 -4.63883728e-01 9.22594309e-01
1.03600889e-01 2.30732456e-01 2.11045146e-01 -6.44563258e-01
1.28257781e-01 7.27271736e-01 5.71518958e-01 -1.13184524e+00
-3.87807369e-01 1.39301633e-02 -2.62283593e-01 5.24619341e-01
3.99250180e-01 -2.94407785e-01 -6.68165982e-01 1.39499593e+00
-6.39399812e-02 -5.97553670e-01 2.33812734e-01 1.03943610e+00
1.25954533e+00 5.42101502e-01 2.60331988e-01 -2.17023909e-01
1.51705194e+00 -3.25057209e-01 -1.11537385e+00 1.42073734e-02
6.66786313e-01 -1.03922951e+00 1.14566851e+00 2.22166032e-01
-1.11655593e+00 -4.48865473e-01 -6.93710566e-01 1.77113786e-01
-4.77482289e-01 3.41605395e-01 2.39433616e-01 8.15613151e-01
-1.09465706e+00 2.35186681e-01 -2.35374406e-01 -1.29352903e+00
1.19869113e-02 1.12976998e-01 -3.53515029e-01 9.44991782e-02
-9.75248098e-01 1.38930142e+00 4.68151480e-01 -2.53369957e-01
-3.43958437e-01 7.92430267e-02 -4.95041817e-01 -2.77349681e-01
9.84495431e-02 -9.43628192e-01 1.08167124e+00 -9.86403227e-01
-1.14555585e+00 1.33003557e+00 -3.39759409e-01 -6.56019077e-02
8.12053859e-01 1.79440179e-03 -3.12882632e-01 1.67379990e-01
2.87561744e-01 6.63769782e-01 1.03018694e-01 -1.72052002e+00
-6.26567602e-01 -1.53598964e-01 3.99692982e-01 5.60243726e-01
-1.09039284e-01 6.41407430e-01 -5.24697423e-01 -4.88088816e-01
-2.34472662e-01 -8.31953943e-01 -1.00319304e-01 -4.80027869e-02
-4.08910304e-01 -3.59027088e-01 5.11757076e-01 -5.42734385e-01
1.19965243e+00 -1.95190823e+00 1.68031171e-01 1.43413201e-01
-6.16577566e-02 3.07258785e-01 4.46968764e-01 1.00065923e+00
2.38636762e-01 2.94401288e-01 8.61726552e-02 -4.67360854e-01
-2.31617633e-02 2.83630848e-01 2.14269355e-01 1.36296839e-01
-1.26383930e-01 1.25409648e-01 -7.81040728e-01 -9.01810527e-01
3.22504789e-01 8.18468690e-01 -8.50237012e-02 2.23816767e-01
-1.18894152e-01 5.35205901e-01 -6.56680346e-01 2.51354605e-01
2.83945858e-01 3.12226623e-01 -5.69891781e-02 -1.22290038e-01
-4.69460547e-01 -8.52962732e-02 -1.31106329e+00 1.28237295e+00
-3.41943115e-01 8.85574877e-01 -5.70514858e-01 -4.75970268e-01
1.09438491e+00 8.43534708e-01 1.71872079e-01 -4.16130066e-01
3.46612185e-01 2.56291181e-01 -1.54037282e-01 -1.30236006e+00
4.69596624e-01 -2.40974516e-01 -9.94363874e-02 3.80441129e-01
-3.56933266e-01 -2.55881250e-01 7.15066075e-01 1.60708139e-03
6.05173051e-01 1.23867221e-01 6.27463162e-01 -7.42302537e-02
7.78018534e-01 3.95585209e-01 -2.24537984e-03 7.58874536e-01
-1.37028277e-01 6.79731727e-01 5.44698894e-01 -6.07343733e-01
-1.09989381e+00 -5.92310071e-01 4.37375940e-02 5.14663637e-01
2.02327177e-01 -4.71253037e-01 -9.84937429e-01 -1.44490138e-01
-6.09933555e-01 1.03058064e+00 -2.70516455e-01 2.38179013e-01
-5.19979239e-01 -5.03577411e-01 3.86932880e-01 1.07281007e-01
5.25256991e-01 -1.35202193e+00 -1.21566832e+00 1.92628518e-01
-3.48218918e-01 -1.22001815e+00 8.47584661e-03 -3.45226228e-01
-4.17786658e-01 -8.04606855e-01 -9.98756707e-01 -7.83662856e-01
8.08214843e-01 6.55446276e-02 9.55042779e-01 1.14583462e-01
-2.67782122e-01 6.25848770e-01 -8.38965416e-01 -6.91558897e-01
-8.78205359e-01 1.13879498e-02 -9.43183079e-02 -2.22791076e-01
4.74990577e-01 -2.72461951e-01 -9.72808152e-02 2.54086941e-01
-7.78523088e-01 3.43807817e-01 2.26383239e-01 2.57591039e-01
-8.18018802e-03 1.44165218e-01 1.33778557e-01 -7.36587524e-01
1.17912328e+00 -1.00239791e-01 -1.69501349e-01 2.62350708e-01
9.15863086e-03 1.60169810e-01 2.64961749e-01 -5.88924050e-01
-9.67787206e-01 1.52008191e-01 -2.05239818e-01 2.21109584e-01
-7.94592500e-01 1.47100419e-01 -2.35547528e-01 -5.13606630e-02
8.45696807e-01 1.45838276e-01 3.35826585e-03 -3.24396640e-01
6.92686588e-02 8.83791089e-01 2.77159035e-01 -3.97037357e-01
5.65591633e-01 2.90482968e-01 -1.62259579e-01 -9.40045059e-01
-4.56061631e-01 -5.17712235e-01 -6.94396675e-01 -8.77333403e-01
1.09691143e+00 -5.34212351e-01 -5.09182632e-01 4.96275276e-01
-1.71620238e+00 -1.01630889e-01 -1.49829775e-01 5.61158717e-01
-8.38151395e-01 6.68764338e-02 6.18494451e-02 -1.21629667e+00
-4.98396158e-02 -1.16578162e+00 7.90279865e-01 5.62179744e-01
-8.71877789e-01 -9.37859893e-01 9.02544335e-03 2.41483554e-01
4.72398877e-01 5.22572160e-01 9.77851093e-01 -5.16697764e-01
-2.75260329e-01 -2.73255259e-01 -8.25673938e-02 -5.70383146e-02
3.72657925e-01 6.05308190e-02 -7.46816218e-01 3.22609305e-01
-3.65509391e-01 -3.65456566e-02 2.34804720e-01 2.02341527e-01
5.46719611e-01 -2.79564530e-01 -4.51369137e-01 3.22182402e-02
1.49672735e+00 5.31507671e-01 8.88450861e-01 5.72421193e-01
3.57095420e-01 1.09996700e+00 5.89119852e-01 6.49615526e-01
3.89027953e-01 1.07206595e+00 1.13174506e-01 -8.34694877e-02
-4.13310289e-01 1.20005384e-01 1.63278118e-01 -6.47806823e-02
-6.48985147e-01 -5.04834354e-01 -1.01132751e+00 6.49104238e-01
-1.79829729e+00 -1.33664453e+00 -2.98644781e-01 2.00991035e+00
3.42196852e-01 -6.75569847e-02 5.92419863e-01 4.07962412e-01
9.07714307e-01 -4.84029315e-02 3.46294075e-01 -7.23928154e-01
-1.83946341e-01 -2.35761702e-01 1.63839549e-01 4.26787764e-01
-6.56183302e-01 8.19792151e-01 6.92394924e+00 2.69459963e-01
-1.00425243e+00 -1.96279034e-01 9.13286284e-02 1.43921062e-01
-2.36330256e-01 -3.66995372e-02 -8.36493850e-01 3.19372565e-01
5.48062503e-01 -4.78949219e-01 1.25799021e-02 4.52243686e-01
7.99171209e-01 -4.84896213e-01 -8.65045607e-01 1.03672612e+00
5.25862515e-01 -1.00698209e+00 4.16946828e-01 -1.47168681e-01
4.20534849e-01 -8.12620580e-01 -3.99187863e-01 -2.61144191e-01
-1.16734892e-01 -9.59030807e-01 9.14610326e-01 9.94093657e-01
4.43773568e-01 -7.50083387e-01 1.11739182e+00 3.68314117e-01
-8.08711290e-01 1.29742339e-01 -5.86468540e-02 -1.94450051e-01
7.22080946e-01 1.09221093e-01 -1.35734308e+00 2.64422625e-01
5.07975698e-01 5.85017800e-02 -5.13913453e-01 1.40679467e+00
-1.92818716e-01 4.45416085e-02 -8.45208243e-02 -7.78511345e-01
9.44490582e-02 -8.85977969e-02 9.04779315e-01 1.61308551e+00
5.42497396e-01 2.12190852e-01 3.36631089e-02 9.18106735e-01
7.72594213e-01 6.72365308e-01 -9.23024654e-01 2.49713697e-02
3.56999874e-01 7.73683846e-01 -7.69837737e-01 -3.45642954e-01
-2.95799732e-01 9.52116966e-01 -2.07752600e-01 2.19102636e-01
-6.52015030e-01 -2.99237221e-01 4.04276997e-01 8.65049422e-01
-1.30974442e-01 -2.60337025e-01 -6.72926232e-02 -4.91299242e-01
-1.46264970e-01 -7.01797068e-01 1.41366482e-01 -1.49939978e+00
-9.24913168e-01 9.85280097e-01 8.78550947e-01 -1.21242070e+00
-6.78659320e-01 -4.28000659e-01 -6.12182975e-01 1.07472718e+00
-7.71011651e-01 -1.17121994e+00 -6.34265900e-01 3.14160347e-01
6.78201318e-01 -4.29153115e-01 9.57191765e-01 2.00210243e-01
-3.98230344e-01 3.08277700e-02 -6.66104794e-01 -2.82417759e-02
6.78627431e-01 -1.01758027e+00 5.32670459e-03 6.31940067e-01
-1.75466150e-01 4.89680529e-01 1.43792379e+00 -6.75634742e-01
-5.18084168e-01 -4.74157333e-01 1.49971485e+00 -2.52332658e-01
1.63728669e-01 -3.34356762e-02 -7.14297593e-01 2.26145700e-01
3.47005337e-01 -5.49992859e-01 5.90010822e-01 -3.12765688e-01
1.34534165e-01 1.98503554e-01 -1.23928595e+00 9.29985285e-01
1.04415166e+00 -3.26566309e-01 -6.28536105e-01 3.79843414e-01
1.71020642e-01 -3.08701843e-01 -4.48836029e-01 -9.40129310e-02
4.94746536e-01 -1.35836887e+00 7.82584071e-01 -3.57966870e-01
2.18244776e-01 -4.80098605e-01 1.03817647e-02 -1.26262951e+00
-1.93380579e-01 -4.54891860e-01 5.94540715e-01 1.54932988e+00
3.54632050e-01 -5.10519445e-01 4.41222459e-01 8.25473487e-01
5.32106124e-02 -3.03564936e-01 -4.38148081e-01 -5.54454565e-01
-4.88577306e-01 -2.65281737e-01 4.60785389e-01 4.69067097e-01
-2.34490484e-02 3.16947877e-01 -3.93925905e-01 -5.86418211e-02
2.08158135e-01 -3.57767731e-01 1.08935428e+00 -1.46294594e+00
1.99688375e-01 -5.43285191e-01 -9.46757376e-01 -3.83025020e-01
5.25381304e-02 -4.28625613e-01 -9.97794047e-03 -2.22008753e+00
4.33217824e-01 -3.03845555e-01 6.79450691e-01 1.68338746e-01
9.10688713e-02 9.65188667e-02 3.72414500e-01 1.26205117e-01
-5.87827154e-02 -8.04569200e-03 1.16193497e+00 2.24537477e-01
-4.39976007e-01 1.11261174e-01 -7.18870103e-01 8.36315095e-01
8.35016966e-01 -4.13479209e-01 -3.58252645e-01 -1.93459615e-01
3.28128427e-01 -1.58522055e-01 4.65370178e-01 -1.21600151e+00
4.34032567e-02 -2.53051341e-01 1.81229129e-01 -5.60168564e-01
5.31351388e-01 -7.28805244e-01 6.49225473e-01 4.91191983e-01
-3.09989333e-01 2.09535792e-01 3.37919779e-02 -3.22508602e-03
-1.52711213e-01 -9.51340437e-01 7.58447051e-01 -5.66052079e-01
-1.05516839e+00 -3.76194358e-01 -7.98263013e-01 -2.42967129e-01
1.19008756e+00 -8.11002433e-01 -2.41677724e-02 -9.20412302e-01
-7.84560680e-01 -7.18510374e-02 7.20259309e-01 1.86537847e-01
5.05345404e-01 -1.34687340e+00 -9.00661111e-01 -3.62560123e-01
4.63159084e-01 -3.30741107e-01 7.24587813e-02 3.32718790e-01
-9.84198570e-01 2.62200445e-01 -6.01749003e-01 -3.50791752e-01
-1.82318926e+00 3.61037016e-01 3.56630832e-01 5.86859807e-02
-7.27567434e-01 3.08924407e-01 -2.73362666e-01 2.48149365e-01
3.12185168e-01 6.76620454e-02 -1.04046166e+00 4.13108319e-01
7.02680528e-01 4.03322846e-01 -3.33656579e-01 -1.26342607e+00
-2.48859271e-01 9.83895957e-01 3.86971951e-01 -6.03631258e-01
1.06363881e+00 -3.34730625e-01 4.13689427e-02 7.07049012e-01
7.59161115e-01 1.30715266e-01 -5.50358295e-01 1.73302829e-01
-1.13886148e-01 -6.62714005e-01 -3.44282240e-01 -5.61310649e-01
-4.73257542e-01 6.00978792e-01 6.26963258e-01 5.23251235e-01
1.17989790e+00 2.01717585e-01 2.25576416e-01 1.91705540e-01
6.17339969e-01 -1.08049917e+00 6.99670911e-02 2.74713099e-01
1.36072791e+00 -8.49046350e-01 9.00225118e-02 -7.02500463e-01
-1.09513271e+00 1.32971966e+00 4.36043799e-01 1.80147767e-01
1.98387653e-01 1.74712479e-01 3.16444248e-01 -2.02353522e-01
-3.97192329e-01 -7.51049995e-01 1.34951457e-01 1.14815068e+00
6.03593290e-01 5.35145737e-02 -9.87121224e-01 2.93764383e-01
-2.92245507e-01 2.56261557e-01 1.00633359e+00 7.21482158e-01
-4.83370870e-01 -1.32514477e+00 -7.95464277e-01 1.07036278e-01
-2.23199040e-01 4.83974844e-01 -9.39989865e-01 1.12423229e+00
4.99176502e-01 9.35727596e-01 -6.15684390e-02 -4.18316163e-02
9.03017938e-01 1.14875853e-01 5.94485581e-01 -8.04066777e-01
-6.22169435e-01 -1.51318595e-01 8.74020576e-01 -7.96649605e-02
-9.54493284e-01 -9.12683010e-01 -1.20424628e+00 -2.06871003e-01
6.53012618e-02 9.82303265e-03 7.84377933e-01 1.05739224e+00
-2.49174550e-01 2.79112160e-01 1.97759688e-01 -8.96780670e-01
1.25718892e-01 -1.00823081e+00 -4.13738906e-01 7.28778660e-01
-1.18420966e-01 -7.87504435e-01 -2.67900169e-01 3.46102715e-01] | [10.82009506225586, 0.8035796880722046] |
4b571e61-bbf3-45d6-a606-d85f3d6ecb48 | the-many-moods-of-emotion | 1810.13197 | null | http://arxiv.org/abs/1810.13197v1 | http://arxiv.org/pdf/1810.13197v1.pdf | The Many Moods of Emotion | This paper presents a novel approach to the facial expression generation
problem. Building upon the assumption of the psychological community that
emotion is intrinsically continuous, we first design our own continuous emotion
representation with a 3-dimensional latent space issued from a neural network
trained on discrete emotion classification. The so-obtained representation can
be used to annotate large in the wild datasets and later used to trained a
Generative Adversarial Network. We first show that our model is able to map
back to discrete emotion classes with a objectively and subjectively better
quality of the images than usual discrete approaches. But also that we are able
to pave the larger space of possible facial expressions, generating the many
moods of emotion. Moreover, two axis in this space may be found to generate
similar expression changes as in traditional continuous representations such as
arousal-valence. Finally we show from visual interpretation, that the third
remaining dimension is highly related to the well-known dominance dimension
from psychology. | ['Frédéric Jurie', 'Stéphane Pateux', 'Valentin Vielzeuf', 'Corentin Kervadec'] | 2018-10-31 | null | null | null | null | ['facial-expression-generation'] | ['computer-vision'] | [ 4.82280105e-01 6.59452498e-01 2.88745552e-01 -6.40014470e-01
-4.13363457e-01 -7.02400804e-01 6.55727565e-01 -3.24492723e-01
-9.00542066e-02 1.03819132e+00 1.01494804e-01 1.34307638e-01
1.48274630e-01 -8.66137981e-01 -6.90976560e-01 -8.27323496e-01
-1.52140275e-01 2.13451520e-01 -3.91879529e-01 -5.22978842e-01
-2.82677591e-01 8.54726732e-01 -1.86564386e+00 3.21193427e-01
3.83404881e-01 1.26097333e+00 -7.33815491e-01 6.12524033e-01
-1.34945586e-01 6.51871979e-01 -8.18048716e-01 -7.90295660e-01
2.90583670e-01 -6.69656813e-01 -7.52836466e-01 3.25244546e-01
3.47655356e-01 -2.62504239e-02 2.45987266e-01 1.17338693e+00
2.56639272e-01 2.16390826e-02 8.76931071e-01 -1.63914633e+00
-7.25823164e-01 8.65346417e-02 -1.99067086e-01 -6.13479316e-01
6.20751739e-01 -1.60115272e-01 8.31906736e-01 -8.01773787e-01
1.05888784e+00 1.33873379e+00 5.17995954e-01 8.97431076e-01
-1.54512727e+00 -5.85707068e-01 -6.57875240e-02 -1.12206191e-01
-1.31312728e+00 -2.33153656e-01 1.17904878e+00 -2.48607472e-01
3.94387662e-01 5.79065859e-01 8.33429396e-01 1.66012108e+00
-8.85667950e-02 4.14000750e-01 1.50470507e+00 -5.67592382e-01
3.90719205e-01 4.56493527e-01 -3.44713151e-01 6.22869253e-01
-4.01064843e-01 1.87677428e-01 -1.48825467e-01 -2.79185891e-01
6.28311753e-01 -1.79741576e-01 -9.33107659e-02 -4.73160505e-01
-6.62278056e-01 1.00880790e+00 3.99107665e-01 4.02376205e-01
-6.24631941e-01 3.11098695e-01 4.61177111e-01 5.22355855e-01
6.18249953e-01 6.29896760e-01 -2.36956313e-01 -7.45148361e-02
-7.33651757e-01 3.41885954e-01 7.84855604e-01 5.77067971e-01
9.25604224e-01 3.09046179e-01 8.51131603e-03 7.11491644e-01
9.67061669e-02 2.24265501e-01 4.40025806e-01 -1.13243318e+00
-2.96531558e-01 3.91647100e-01 2.94684529e-01 -1.16950369e+00
-3.43876868e-01 -2.48161778e-02 -7.72684395e-01 8.27816486e-01
3.34333152e-01 -4.72496271e-01 -7.43063390e-01 2.26432276e+00
1.62811100e-01 -2.60435026e-02 2.37725914e-01 7.00794995e-01
3.73736233e-01 7.51880944e-01 1.93779960e-01 -4.08060461e-01
1.18046427e+00 -3.71823698e-01 -9.81862843e-01 1.80491611e-01
2.45615110e-01 -4.78252888e-01 9.87807691e-01 5.97788870e-01
-1.04648709e+00 -5.00464082e-01 -1.14396179e+00 1.90133274e-01
-7.58849144e-01 2.35537440e-02 1.12941527e+00 9.12584424e-01
-1.37355566e+00 5.69448054e-01 -4.22187448e-01 -2.23716468e-01
1.86978683e-01 3.52228642e-01 -6.84306562e-01 4.40214723e-01
-1.47408617e+00 1.01791847e+00 1.40231326e-01 2.89746493e-01
-5.06932557e-01 -1.52912408e-01 -8.75940382e-01 -3.52253988e-02
-8.78901631e-02 -5.14310598e-01 1.00145710e+00 -1.89546680e+00
-1.85609746e+00 1.20058250e+00 -4.17074710e-02 -2.21537516e-01
4.95116383e-01 2.52717912e-01 -6.31084621e-01 2.75688171e-01
-2.48790294e-01 9.98851776e-01 1.18155742e+00 -1.63731325e+00
7.21540079e-02 -4.41072911e-01 2.72209309e-02 -1.55535927e-02
-3.50143552e-01 -4.32564728e-02 2.05667526e-01 -6.61364555e-01
-9.43478197e-02 -1.02248788e+00 -3.47171932e-01 1.61846861e-01
-2.85730094e-01 -1.32807955e-01 7.26531148e-01 -2.75402814e-01
7.52034366e-01 -2.10141110e+00 5.36980808e-01 4.39438552e-01
1.24067619e-01 -5.98024279e-02 -1.69793025e-01 3.75577986e-01
-8.76163542e-01 2.94399738e-01 -1.97430119e-01 -4.86490011e-01
3.32455784e-01 4.20335352e-01 -5.26936948e-01 5.24619102e-01
6.03143930e-01 9.58055139e-01 -6.23528719e-01 -2.99782783e-01
-5.44386953e-02 5.86022258e-01 -5.15513361e-01 3.50862622e-01
-2.94966131e-01 4.09251302e-01 -2.77158856e-01 4.51474011e-01
6.25818551e-01 2.22858086e-01 1.04738854e-01 2.60921307e-02
1.84641957e-01 -3.45667928e-01 -8.78691912e-01 1.53052390e+00
-5.74775755e-01 6.14369333e-01 1.47849232e-01 -1.11109114e+00
1.30997038e+00 5.47522306e-01 5.43505609e-01 -4.01846826e-01
4.14326698e-01 -9.25639495e-02 -3.19059283e-01 -3.20485651e-01
4.52108890e-01 -8.68015826e-01 -5.89635789e-01 4.16497946e-01
2.68042505e-01 -4.43561047e-01 -2.67038465e-01 -1.70789182e-01
8.31511080e-01 1.94455668e-01 1.46077186e-01 -3.48256715e-02
4.32973385e-01 -3.22390527e-01 2.38125235e-01 3.00318450e-01
-1.75530508e-01 4.63171184e-01 1.08176625e+00 -5.81979156e-01
-1.05256259e+00 -1.03018284e+00 -2.51182526e-01 1.00381827e+00
-2.80087560e-01 -6.13267533e-02 -9.43431735e-01 -6.45769835e-01
-2.60167956e-01 6.51403546e-01 -1.26216102e+00 -3.15861583e-01
-6.98824376e-02 -3.26752990e-01 7.65886188e-01 2.06660241e-01
-7.92905018e-02 -1.47309959e+00 -4.82524723e-01 7.45369215e-03
6.32601678e-02 -8.77785027e-01 2.91254550e-01 4.29782927e-01
-4.43719417e-01 -4.74944621e-01 -7.61166692e-01 -5.09927154e-01
6.24726951e-01 -7.27001190e-01 1.23822331e+00 -1.87736869e-01
-3.53152066e-01 5.16944945e-01 -5.72276831e-01 -5.64268172e-01
-7.46731579e-01 -4.67025518e-01 1.52954862e-01 5.74729860e-01
3.95291448e-01 -9.09361959e-01 -2.54996777e-01 -1.48854971e-01
-1.32215607e+00 -1.65749565e-01 1.52268350e-01 7.78955877e-01
4.36853170e-01 5.67727350e-03 8.78316224e-01 -9.73624170e-01
8.13872397e-01 -4.39144403e-01 -2.78737217e-01 3.41226049e-02
-2.60068595e-01 1.24559917e-01 7.69558072e-01 -6.12856925e-01
-9.34380829e-01 4.43864435e-01 -5.29973805e-01 -5.55353999e-01
-6.46918178e-01 9.35141519e-02 -3.73982638e-01 -1.16848126e-01
8.20037842e-01 6.50328323e-02 2.17074081e-01 -5.73010184e-02
1.02276647e+00 3.83070409e-01 5.04851341e-01 -5.16570389e-01
6.96915567e-01 5.64364135e-01 2.40294233e-01 -7.30410635e-01
-4.53042805e-01 1.62272394e-01 -6.57662928e-01 -3.25481236e-01
1.11998498e+00 -5.85332990e-01 -6.57048821e-01 1.98934287e-01
-1.36660719e+00 -9.09703448e-02 -6.92885637e-01 -5.59050255e-02
-1.12245619e+00 1.03880435e-01 -5.29277682e-01 -9.83889282e-01
6.66011795e-02 -9.48026061e-01 1.11689448e+00 3.20146307e-02
-5.54942310e-01 -1.09563947e+00 1.46345899e-01 -3.70234489e-01
1.91768691e-01 9.80435491e-01 9.93711412e-01 -4.40358520e-01
2.05476329e-01 -4.13240373e-01 1.34523124e-01 6.11263096e-01
2.87968013e-02 2.88350046e-01 -1.18204141e+00 7.93831423e-02
3.12951118e-01 -7.54846454e-01 4.84970510e-01 7.41505623e-02
1.18564153e+00 -3.62056702e-01 1.30402923e-01 6.30697906e-01
1.48609531e+00 1.45375162e-01 1.10592651e+00 -9.13582668e-02
2.69662976e-01 1.01495743e+00 4.43247706e-01 3.82626742e-01
-1.81009829e-01 8.63184929e-01 6.26969159e-01 -6.43330514e-01
3.61208797e-01 -1.41709760e-01 4.34043556e-01 2.28372604e-01
-4.37225960e-02 -7.28414506e-02 -4.39276159e-01 2.06776276e-01
-1.55052221e+00 -1.27112818e+00 1.04308680e-01 1.67262971e+00
9.16834593e-01 -8.41414630e-02 2.20989808e-01 2.76500374e-01
6.02967978e-01 2.34934110e-02 -2.62080401e-01 -1.32576048e+00
-2.79557258e-01 7.97653377e-01 1.18398435e-01 4.85473752e-01
-9.88378882e-01 9.28022623e-01 7.19301367e+00 5.39499104e-01
-1.26366520e+00 -5.15680350e-02 7.15142310e-01 1.66457966e-02
-5.30829072e-01 -1.55355722e-01 -2.34817281e-01 1.33300319e-01
1.02247667e+00 -1.59384295e-01 3.56410474e-01 1.05086875e+00
2.45757867e-02 1.46578342e-01 -1.17210674e+00 8.96364570e-01
2.63880312e-01 -8.63190830e-01 1.78776413e-01 1.09027982e-01
7.44307637e-01 -7.71510541e-01 4.08709705e-01 3.80621195e-01
9.17049199e-02 -1.53960276e+00 6.78823173e-01 7.52487123e-01
1.13744009e+00 -1.05364907e+00 3.97590190e-01 1.76329359e-01
-6.82079792e-01 1.14694141e-01 -3.24062496e-01 -1.73344746e-01
4.38519791e-02 1.78286821e-01 -4.69057947e-01 3.30100834e-01
1.06994748e-01 2.46966004e-01 -4.86686140e-01 3.12405109e-01
-2.51887888e-01 2.69301444e-01 -2.61027992e-01 -9.26365107e-02
3.46340150e-01 -3.15035462e-01 3.19518626e-01 1.17066133e+00
4.07881677e-01 1.31083801e-01 -2.41176814e-01 1.07052016e+00
-5.72589487e-02 1.33517668e-01 -1.21401238e+00 -7.91939348e-02
-7.13069364e-02 1.46287346e+00 -6.41748488e-01 -3.68838429e-01
-1.90004706e-01 1.55204010e+00 2.99773097e-01 5.06123185e-01
-1.03080094e+00 -4.20257598e-01 7.02980340e-01 -1.48673162e-01
2.58083820e-01 2.37786189e-01 -9.59502757e-02 -1.06625366e+00
-1.42690748e-01 -9.29552972e-01 -1.41141087e-01 -1.14193404e+00
-1.37552607e+00 1.16605580e+00 -1.38586923e-01 -1.14350498e+00
-7.15349138e-01 -8.46287787e-01 -5.21717370e-01 1.05102122e+00
-1.08150685e+00 -1.19886732e+00 -9.79569256e-02 7.96917856e-01
-1.66732632e-02 -4.20405343e-02 1.57698715e+00 -3.84346507e-02
-1.71649545e-01 5.50099432e-01 -2.20175490e-01 1.40418429e-02
5.29181838e-01 -1.44414771e+00 2.42259400e-03 5.13098657e-01
3.38746816e-01 4.79019403e-01 9.59539831e-01 -1.31516457e-01
-9.78625238e-01 -8.66590202e-01 6.48465693e-01 -4.21244413e-01
6.15720153e-01 -6.24995887e-01 -7.67979860e-01 5.90128720e-01
3.18036586e-01 2.06355318e-01 8.99597466e-01 1.52086169e-01
-5.60200751e-01 1.19553775e-01 -1.26454544e+00 5.73575854e-01
5.30796766e-01 -6.72687173e-01 -6.54723167e-01 1.06398746e-01
6.17690086e-01 -1.30030006e-01 -8.53469908e-01 2.52823263e-01
6.74236834e-01 -1.23539352e+00 8.64214003e-01 -1.02473021e+00
7.67029285e-01 5.85599430e-02 -1.73509628e-01 -1.49877822e+00
-1.22462280e-01 -6.34253383e-01 2.41229102e-01 1.13791442e+00
3.75559211e-01 -4.95388389e-01 8.82693350e-01 6.00372255e-01
2.34963223e-01 -8.25736582e-01 -1.08732188e+00 -5.39371490e-01
3.79792869e-01 -4.89117205e-01 6.17644072e-01 1.09674108e+00
8.04633573e-02 4.05638039e-01 -5.59548497e-01 -1.24912463e-01
3.21190625e-01 1.63385972e-01 7.26005971e-01 -1.20812726e+00
-3.94815236e-01 -4.54346120e-01 -8.02125156e-01 -5.44517100e-01
6.98707879e-01 -7.90006578e-01 -1.04960566e-02 -9.49748933e-01
-1.97844774e-01 -2.18213111e-01 -2.79679060e-01 5.59802890e-01
2.37516165e-01 7.72713780e-01 1.30461127e-01 -3.08612674e-01
-2.67635733e-01 7.24583268e-01 1.26226985e+00 2.37379089e-01
-8.53282884e-02 -1.16725489e-01 -7.53527462e-01 8.72003078e-01
5.84780395e-01 -3.65765959e-01 -3.75751406e-01 3.47964972e-01
5.12832880e-01 3.51529062e-01 6.23069227e-01 -6.47321165e-01
-5.39085090e-01 -1.79546624e-01 6.93148971e-01 -7.45598897e-02
7.85760641e-01 -1.08919775e+00 4.15355712e-01 3.79639119e-02
-4.87103939e-01 -1.80901065e-01 1.73180804e-01 2.19999522e-01
-4.28989559e-01 -2.60981739e-01 8.63585293e-01 -2.70219505e-01
-5.40380597e-01 1.00419484e-01 -4.74287808e-01 -3.25470597e-01
1.13392889e+00 -2.89381266e-01 1.44701421e-01 -8.50805283e-01
-1.39002848e+00 -4.49434996e-01 5.58408260e-01 5.09961426e-01
4.17393893e-01 -1.77538383e+00 -5.34699619e-01 3.30908686e-01
1.52130961e-01 -6.54045045e-01 1.02591582e-01 3.85871202e-01
-3.39776099e-01 -8.75282958e-02 -7.15741336e-01 -3.66326272e-01
-1.02756631e+00 7.28830218e-01 3.99560422e-01 6.86947210e-03
-2.06758648e-01 7.35878408e-01 1.70104921e-01 -2.85384566e-01
-2.39861067e-02 -4.29141633e-02 -2.16687679e-01 4.12360460e-01
3.80474359e-01 -2.33729959e-01 -1.29223108e-01 -9.54747200e-01
-1.61036596e-01 3.53622675e-01 5.56393087e-01 -4.79305357e-01
1.31633437e+00 9.79816020e-02 -3.78516614e-01 6.76881671e-01
1.64262450e+00 2.08435059e-01 -1.11294031e+00 3.89506459e-01
-2.94486821e-01 -1.75447702e-01 -3.28828573e-01 -7.55780101e-01
-7.96437442e-01 9.61216331e-01 6.68301702e-01 8.08921218e-01
1.51476634e+00 -7.74620697e-02 2.19379187e-01 1.07918471e-01
2.54328460e-01 -8.66429090e-01 1.49426773e-01 1.60784647e-01
1.17886937e+00 -8.74969304e-01 -4.01632011e-01 -2.64312476e-01
-8.58475447e-01 1.28088641e+00 2.82234341e-01 -2.80709088e-01
3.94388944e-01 4.62505758e-01 4.21083629e-01 -2.24643692e-01
-6.98025167e-01 -1.91223443e-01 2.28330135e-01 8.33091438e-01
5.12422323e-01 1.49672210e-01 -3.55147600e-01 6.51070952e-01
-5.29826522e-01 -1.06729165e-01 6.32759631e-01 4.13787693e-01
-1.14216395e-01 -1.46624434e+00 -4.31141853e-01 -3.87054458e-02
-6.09720111e-01 2.69451112e-01 -7.63849735e-01 8.40488791e-01
5.16624808e-01 5.20213723e-01 1.01207733e-01 -3.88891757e-01
2.80277401e-01 5.66054225e-01 7.13870227e-01 -4.98795718e-01
-4.08655792e-01 -2.54294537e-02 4.93105277e-02 -7.27264106e-01
-5.45547783e-01 -4.06944066e-01 -1.03182793e+00 3.43236923e-02
1.52300820e-02 3.39970022e-01 8.36950123e-01 6.08250916e-01
4.60414290e-02 3.98276567e-01 9.65461314e-01 -1.08795452e+00
-3.63213509e-01 -8.76772702e-01 -8.81979883e-01 9.99965608e-01
3.05179566e-01 -6.86373591e-01 -5.06800592e-01 3.01030368e-01] | [13.405956268310547, 1.6431485414505005] |
53123205-7954-4399-a685-e968e18c2f98 | deep-convolutional-encoder-decoders-with | 1901.09197 | null | http://arxiv.org/abs/1901.09197v2 | http://arxiv.org/pdf/1901.09197v2.pdf | Deep Convolutional Encoder-Decoders with Aggregated Multi-Resolution Skip Connections for Skin Lesion Segmentation | The prevalence of skin melanoma is rapidly increasing as well as the recorded
death cases of its patients. Automatic image segmentation tools play an
important role in providing standardized computer-assisted analysis for skin
melanoma patients. Current state-of-the-art segmentation methods are based on
fully convolutional neural networks, which utilize an encoder-decoder approach.
However, these methods produce coarse segmentation masks due to the loss of
location information during the encoding layers. Inspired by Pyramid Scene
Parsing Network (PSP-Net), we propose an encoder-decoder model that utilizes
pyramid pooling modules in the deep skip connections which aggregate the global
context and compensate for the lost spatial information. We trained and
validated our approach using ISIC 2018: Skin Lesion Analysis Towards Melanoma
Detection grand challenge dataset. Our approach showed a validation accuracy
with a Jaccard index of 0.837, which outperforms U-Net. We believe that with
this reported reliable accuracy, this method can be introduced for clinical
practice. | ['Karim Amer', 'Ahmed H. Shahin', 'Mustafa A. Elattar'] | 2019-01-26 | null | null | null | null | ['skin-lesion-segmentation'] | ['medical'] | [ 8.31747174e-01 2.35124156e-01 -1.57228276e-01 -1.38506532e-01
-1.01689625e+00 -4.23215419e-01 3.36372823e-01 4.38582987e-01
-8.37445319e-01 6.11220062e-01 1.15229882e-01 -3.81262302e-01
4.94308136e-02 -7.72157490e-01 -3.88586402e-01 -7.57632434e-01
2.27006748e-01 1.06213903e-02 4.81602341e-01 -5.23838326e-02
3.14888358e-01 3.64794195e-01 -1.09261382e+00 7.14643121e-01
1.22149932e+00 1.05101168e+00 2.37299070e-01 9.74155426e-01
-3.29061925e-01 6.64184153e-01 -5.28738439e-01 -6.54898107e-01
-5.83346421e-03 -4.26616162e-01 -8.24197829e-01 -9.53319594e-02
3.96953553e-01 -3.57530564e-01 -1.32774919e-01 9.08976436e-01
4.05633092e-01 -6.23093665e-01 5.40538371e-01 -5.15845418e-01
-2.19108060e-01 4.22097266e-01 -8.27745855e-01 3.56785804e-02
7.81159326e-02 3.04129511e-01 6.02360964e-01 -3.45468223e-01
8.63728762e-01 4.28728759e-01 1.24246788e+00 6.97625577e-01
-8.76252830e-01 -2.11795390e-01 -1.41839996e-01 4.89241481e-02
-1.31986213e+00 -1.35702109e-02 4.08016592e-01 -4.52794969e-01
9.46307123e-01 5.52557170e-01 8.41075718e-01 9.15814936e-01
4.17028308e-01 6.54971957e-01 1.10952353e+00 -4.44458812e-01
1.23217694e-01 -1.42797157e-01 1.36793102e-03 8.05422187e-01
2.32757807e-01 -3.33104491e-01 -3.32422346e-01 2.47403562e-01
7.68018723e-01 8.16856772e-02 -8.75774473e-02 1.85881376e-01
-1.00597441e+00 6.43322468e-01 8.30700219e-01 1.35834143e-01
-4.97171700e-01 3.27715307e-01 5.20096123e-01 -2.77920961e-01
5.81942022e-01 1.25230074e-01 -2.10426137e-01 -7.45616406e-02
-1.18809676e+00 -1.57471135e-01 4.31154490e-01 4.20244634e-01
2.55186319e-01 -4.78389412e-01 -4.09544230e-01 6.61132991e-01
4.01313081e-02 8.67318437e-02 6.01498663e-01 -5.82593083e-01
2.90254951e-01 1.10216165e+00 -3.41794491e-01 -5.15777349e-01
-8.72585535e-01 -5.72812736e-01 -1.11609268e+00 1.52182072e-01
7.21293330e-01 -3.19185644e-01 -1.31252754e+00 1.28154254e+00
1.21731482e-01 4.88023497e-02 -1.01764187e-01 7.91643322e-01
6.97383404e-01 -1.41396523e-01 4.89332557e-01 2.37011254e-01
1.41026652e+00 -9.63356078e-01 -4.86206561e-01 3.92947532e-02
1.03624976e+00 -7.20233560e-01 7.49366939e-01 2.65908182e-01
-9.87559915e-01 -1.30410433e-01 -8.26505661e-01 -1.59959450e-01
-4.99657363e-01 4.52608913e-01 6.93827212e-01 9.79257941e-01
-1.33091867e+00 5.21416545e-01 -1.04309237e+00 -8.13869119e-01
1.02187300e+00 4.78682071e-01 -4.65855390e-01 2.02337913e-02
-8.09159696e-01 6.97862148e-01 2.40714744e-01 3.76904786e-01
-6.17169678e-01 -8.60051632e-01 -7.37835288e-01 -3.06471586e-01
1.24722002e-02 -7.36617267e-01 1.00548661e+00 -1.08023250e+00
-1.36906826e+00 1.03800714e+00 -3.34883153e-01 -7.99798787e-01
6.93960607e-01 8.53367499e-04 -5.65257855e-02 6.17134809e-01
-1.02135958e-02 1.03478098e+00 4.99687761e-01 -7.13272095e-01
-8.38544130e-01 -4.04838890e-01 -4.90699634e-02 2.40764514e-01
-3.46953034e-01 -1.70011029e-01 -5.51324308e-01 -4.59689707e-01
-8.29082280e-02 -6.25811934e-01 -7.39993930e-01 4.94641602e-01
-6.77235246e-01 2.61090159e-01 5.60919225e-01 -1.23933196e+00
1.21407104e+00 -1.87676585e+00 -7.64622493e-03 3.11131507e-01
3.67246807e-01 4.54248488e-01 -2.13124499e-01 3.02175343e-01
1.25775829e-01 4.80119795e-01 -7.36923754e-01 -4.42809582e-01
-4.40072984e-01 -2.94987857e-01 1.95350081e-01 4.67591971e-01
5.51204801e-01 1.08423030e+00 -6.28780425e-01 -6.01646304e-01
5.15199542e-01 7.52972245e-01 -5.83613634e-01 -4.46611978e-02
-2.55904108e-01 4.87239808e-01 -1.31796777e-01 8.95210624e-01
7.33983219e-01 -2.76954740e-01 2.59083927e-01 -1.45504683e-01
-1.02452800e-01 -1.76187202e-01 -4.92392123e-01 2.19512272e+00
-2.90796548e-01 7.64617383e-01 6.90496340e-02 -5.06118476e-01
3.90843511e-01 6.89603165e-02 6.50585651e-01 -6.69576764e-01
9.50415060e-02 1.41483054e-01 1.02438601e-02 -6.49595320e-01
3.44433039e-01 7.17679635e-02 2.07417607e-01 -2.60334313e-01
-2.16711432e-01 5.33893704e-02 2.66035616e-01 2.27193147e-01
1.33140862e+00 3.00533235e-01 3.13527465e-01 -2.51630656e-02
6.18420482e-01 4.75132078e-01 2.37613440e-01 4.81521279e-01
-4.48176831e-01 1.12416601e+00 8.18179250e-01 -4.93397266e-01
-8.17911625e-01 -8.36347818e-01 -3.16470236e-01 4.21612471e-01
-5.52909374e-02 -3.21664333e-01 -1.29841542e+00 -8.05139840e-01
-4.06823963e-01 2.42851570e-01 -9.49997008e-01 2.96719372e-01
-3.14726830e-01 -1.08285630e+00 8.72386694e-01 5.34147382e-01
6.69415832e-01 -8.22437704e-01 -8.05536211e-01 3.66339594e-01
-6.45682067e-02 -1.17394364e+00 -1.25475973e-01 1.11382194e-02
-6.70372546e-01 -1.60095179e+00 -1.35223544e+00 -6.86440527e-01
9.79079247e-01 -1.61362410e-01 6.91885769e-01 8.33497476e-03
-1.09928262e+00 -7.95806665e-03 -2.27666229e-01 -2.57221848e-01
-2.09220573e-01 4.79671538e-01 -7.50600934e-01 1.81564894e-02
4.75288421e-01 -8.52803215e-02 -1.14625752e+00 -4.89449538e-02
-1.05431509e+00 4.75673586e-01 9.52107906e-01 7.05108345e-01
8.29724371e-01 -1.34279370e-01 2.69523829e-01 -1.11999428e+00
5.09493947e-01 -2.69037247e-01 -4.21622187e-01 2.83436269e-01
-1.61908641e-01 -3.41161370e-01 5.83917856e-01 1.63115650e-01
-1.06159055e+00 5.48000395e-01 -6.35845900e-01 1.98652342e-01
-4.98205781e-01 5.13723016e-01 2.29598761e-01 -4.88793343e-01
5.68654478e-01 2.04809621e-01 2.25508481e-01 -1.83817565e-01
1.43470600e-01 6.44358516e-01 4.62033421e-01 1.61650509e-01
1.58333629e-01 8.10856700e-01 3.56577694e-01 -8.90587270e-01
-7.01150000e-01 -5.92503011e-01 -8.63841116e-01 -2.43311092e-01
1.32884884e+00 -8.21355999e-01 -5.87272465e-01 9.63871658e-01
-1.16492796e+00 -5.96928477e-01 -1.77269742e-01 2.84147914e-02
-2.63827473e-01 4.82286990e-01 -6.78709090e-01 -6.42560959e-01
-5.09262502e-01 -1.31422973e+00 1.46942389e+00 5.00283003e-01
-2.81759679e-01 -1.14021409e+00 4.56809551e-02 5.50732434e-01
6.86058939e-01 6.13577068e-01 6.61089122e-01 -3.88041943e-01
-4.63314831e-01 -4.82295185e-01 -6.24441803e-01 1.91318870e-01
9.90364179e-02 1.74524084e-01 -1.05880070e+00 3.54934521e-02
-6.99784458e-01 -9.50011015e-02 1.24972188e+00 8.59725237e-01
1.54568279e+00 2.40681730e-02 -4.70271260e-01 8.19419026e-01
1.79053354e+00 -1.25626743e-01 8.63131702e-01 1.28646195e-01
8.80129278e-01 6.66227639e-01 9.06943753e-02 2.22227648e-01
4.78403866e-01 3.29758555e-01 6.94922268e-01 -5.68763852e-01
-6.85870290e-01 -2.25822613e-01 -1.10966042e-01 2.60439813e-01
-2.08548188e-01 -1.00504972e-01 -1.08419323e+00 7.89377511e-01
-1.46969569e+00 -5.86493850e-01 -4.92380083e-01 2.07039976e+00
8.03565323e-01 1.76114291e-01 5.11469506e-02 6.84611350e-02
6.47257149e-01 -5.08340374e-02 -4.38282669e-01 -3.99208069e-01
-2.56043047e-01 4.73943710e-01 8.19228351e-01 4.51346427e-01
-1.23241353e+00 9.70967710e-01 6.18122959e+00 9.59518969e-01
-1.35909247e+00 3.39498408e-02 1.04845750e+00 2.04714425e-02
-7.81178102e-03 -2.60114461e-01 -5.98774314e-01 5.06169140e-01
9.99530673e-01 6.87683403e-01 -1.45141467e-01 3.51610363e-01
1.69185519e-01 -5.80820799e-01 -5.61344862e-01 7.78550625e-01
6.62261918e-02 -1.71203518e+00 1.06342599e-01 3.14882755e-01
6.22657716e-01 2.09297668e-02 1.73876613e-01 -3.17951888e-01
-1.92256704e-01 -1.36127543e+00 6.60599694e-02 8.94530296e-01
1.27308452e+00 -5.97376287e-01 1.16746008e+00 -3.27440128e-02
-9.08465505e-01 2.44478554e-01 -1.42283618e-01 6.62480742e-02
1.69262186e-01 6.54353380e-01 -1.01577258e+00 5.40526628e-01
2.82155275e-01 6.91624999e-01 -9.85455155e-01 1.49924326e+00
-7.61739537e-02 6.75237596e-01 -2.87424505e-01 -3.77634652e-02
3.03578496e-01 -3.17463204e-02 1.52237624e-01 1.33954597e+00
2.10958272e-01 -3.86428088e-01 -3.39447349e-01 5.63300014e-01
-2.14324966e-02 1.94183007e-01 -2.50713795e-01 -6.90191910e-02
-1.25966251e-01 1.39777958e+00 -1.06949282e+00 9.17997584e-02
-3.66148472e-01 1.28595090e+00 1.19644605e-01 2.03942746e-01
-7.01635063e-01 -5.54480016e-01 3.97029936e-01 3.07530314e-01
1.22928746e-01 1.29413038e-01 -7.89540410e-01 -7.43919551e-01
-4.01365831e-02 -4.70241845e-01 3.58079851e-01 -4.19581234e-01
-1.02017200e+00 7.27078021e-01 -5.29981017e-01 -9.55659568e-01
1.22414127e-01 -9.32005703e-01 -7.37482548e-01 8.56629074e-01
-1.92059803e+00 -1.57717144e+00 -6.21243119e-01 3.38260740e-01
4.97695714e-01 -6.54303208e-02 1.01698303e+00 -3.32939215e-02
-9.35803235e-01 7.04833210e-01 -7.82704502e-02 3.13413322e-01
5.41587651e-01 -1.33053184e+00 2.42736623e-01 8.36708248e-01
-3.56365144e-01 2.28898808e-01 -5.80689684e-03 -6.46747589e-01
-9.31352735e-01 -1.25306332e+00 8.18523407e-01 -2.81986028e-01
6.11565769e-01 -1.07059419e-01 -4.68524069e-01 2.40865558e-01
4.94103163e-01 -6.84941411e-02 1.08852196e+00 -2.45095342e-01
-9.74358395e-02 -1.53267123e-02 -1.46171343e+00 6.61479115e-01
6.69725597e-01 -3.62002701e-01 4.97951470e-02 3.98319572e-01
4.43212777e-01 -6.51855171e-01 -7.91609526e-01 4.14836586e-01
6.07402682e-01 -1.18450427e+00 7.46949971e-01 -1.94984794e-01
9.00819063e-01 -2.66255531e-02 1.01762794e-01 -9.79283333e-01
-2.42722854e-02 -2.57236689e-01 2.29501188e-01 7.93380022e-01
6.04045272e-01 -4.97460961e-01 1.31418395e+00 5.40698886e-01
-1.32847577e-01 -1.18526089e+00 -1.03712940e+00 -6.17690459e-02
1.96861371e-01 -4.03423995e-01 2.71507770e-01 4.46502328e-01
-1.82382956e-01 -2.91065395e-01 7.88407624e-02 9.04766936e-03
6.80718541e-01 -5.19336760e-01 1.48693770e-01 -9.55595732e-01
7.03957528e-02 -7.59703457e-01 -6.87677085e-01 -5.16179562e-01
-2.88276821e-01 -8.79090071e-01 -3.32490146e-01 -1.94256973e+00
3.06306273e-01 -2.41862893e-01 -3.38889807e-01 5.06343663e-01
-1.88810185e-01 8.13741982e-01 -9.61641222e-02 4.52871360e-02
-5.16954839e-01 -6.46922886e-02 1.21638465e+00 -1.63360000e-01
-1.26941159e-01 -9.36846584e-02 -7.00715721e-01 7.81378090e-01
1.00521958e+00 -5.59734814e-02 -1.28694430e-01 -5.31113148e-01
5.56059927e-02 -2.60086387e-01 6.87904298e-01 -1.28975654e+00
4.92407501e-01 2.73011122e-02 7.46290743e-01 -5.99284232e-01
2.40674168e-01 -7.85021067e-01 -1.11236773e-01 8.68571520e-01
-3.60785514e-01 -4.19592798e-01 3.75054449e-01 4.83254611e-01
-3.26468527e-01 -1.40596122e-01 7.01501191e-01 -7.67255947e-02
-6.86256468e-01 3.40412319e-01 -3.84205520e-01 -2.68200606e-01
1.33176160e+00 -4.37450945e-01 -5.99946797e-01 3.86144184e-02
-6.69715524e-01 -5.17506637e-02 5.24492502e-01 -1.63000494e-01
6.30852997e-01 -6.97088778e-01 -8.21593523e-01 1.49895251e-01
1.82065412e-01 2.20315218e-01 7.47115672e-01 1.22812915e+00
-1.39653003e+00 5.99107623e-01 -3.03112656e-01 -6.37884974e-01
-1.20882082e+00 7.55720660e-02 5.63318312e-01 -5.00947356e-01
-5.70743799e-01 1.00706184e+00 -1.00774013e-01 -2.15366945e-01
2.41971478e-01 -4.30432677e-01 -2.59056926e-01 4.48552668e-02
4.95955706e-01 1.95338950e-01 2.45975837e-01 -4.37872976e-01
-4.37686473e-01 6.83852911e-01 -3.10362846e-01 6.73636720e-02
1.11425173e+00 1.56713575e-01 -1.67076588e-01 -2.25658521e-01
1.02156639e+00 -1.61816865e-01 -1.36434412e+00 2.45636284e-01
-6.60338774e-02 -2.17409670e-01 1.48266882e-01 -1.29433000e+00
-1.07431066e+00 1.09171391e+00 9.45841849e-01 -4.59334478e-02
1.45725715e+00 -3.98008823e-01 9.08434272e-01 -1.65438876e-01
6.49818629e-02 -9.23492610e-01 -3.95911813e-01 6.04995824e-02
5.07537782e-01 -1.39016068e+00 -1.37001246e-01 -7.43177176e-01
-6.93479359e-01 1.16613066e+00 4.73172039e-01 -1.06372565e-01
4.66858864e-01 4.14474696e-01 2.81346679e-01 4.98600900e-02
-3.63254488e-01 -4.67229396e-01 2.98009515e-01 9.23635244e-01
7.41455376e-01 1.95251480e-01 -6.15503788e-01 5.59002876e-01
1.26909643e-01 1.73930645e-01 4.41924721e-01 6.99836433e-01
-1.64286345e-01 -1.21402848e+00 -7.37932092e-03 7.30578184e-01
-6.30329907e-01 -3.80714774e-01 -6.86598122e-01 7.08643317e-01
2.65708506e-01 6.30684197e-01 1.38977110e-01 -2.02416033e-01
1.39097884e-01 -1.75606515e-02 3.86408657e-01 -3.69387209e-01
-8.89642656e-01 5.43246493e-02 -1.33548109e-02 -6.09105289e-01
-4.61719543e-01 -3.56992483e-01 -1.16606700e+00 1.47519812e-01
2.10729271e-01 -4.07433152e-01 1.11116767e+00 7.37983525e-01
4.74949390e-01 9.35826778e-01 1.27877653e-01 -4.15092587e-01
-1.44665539e-02 -8.79528105e-01 -5.00987589e-01 1.71432555e-01
2.32746482e-01 -1.73385860e-03 7.33616427e-02 1.32020935e-01] | [15.619219779968262, -2.9417061805725098] |
d55d33cb-0868-4370-9eb8-2320aa5557a7 | rapping-singing-voice-synthesis-based-on | 2111.09146 | null | https://arxiv.org/abs/2111.09146v1 | https://arxiv.org/pdf/2111.09146v1.pdf | Rapping-Singing Voice Synthesis based on Phoneme-level Prosody Control | In this paper, a text-to-rapping/singing system is introduced, which can be adapted to any speaker's voice. It utilizes a Tacotron-based multispeaker acoustic model trained on read-only speech data and which provides prosody control at the phoneme level. Dataset augmentation and additional prosody manipulation based on traditional DSP algorithms are also investigated. The neural TTS model is fine-tuned to an unseen speaker's limited recordings, allowing rapping/singing synthesis with the target's speaker voice. The detailed pipeline of the system is described, which includes the extraction of the target pitch and duration values from an a capella song and their conversion into target speaker's valid range of notes before synthesis. An additional stage of prosodic manipulation of the output via WSOLA is also investigated for better matching the target duration values. The synthesized utterances can be mixed with an instrumental accompaniment track to produce a complete song. The proposed system is evaluated via subjective listening tests as well as in comparison to an available alternate system which also aims to produce synthetic singing voice from read-only training data. Results show that the proposed approach can produce high quality rapping/singing voice with increased naturalness. | ['Aimilios Chalamandaris', 'Pirros Tsiakoulis', 'Hyoungmin Park', 'June Sig Sung', 'Georgia Maniati', 'Georgios Vamvoukakis', 'Panos Kakoulidis', 'Myrsini Christidou', 'Alexandra Vioni', 'Nikolaos Ellinas', 'Konstantinos Markopoulos'] | 2021-11-17 | null | null | null | null | ['singing-voice-synthesis'] | ['speech'] | [ 3.35428804e-01 1.40333503e-01 2.38162935e-01 -1.24583311e-01
-1.03977513e+00 -8.12271476e-01 1.81679100e-01 -2.20516160e-01
-1.46768197e-01 4.31059211e-01 3.04452121e-01 -3.24537568e-02
9.69032720e-02 -2.59713858e-01 -3.30558330e-01 -8.18699777e-01
2.99113423e-01 3.76072168e-01 1.92398801e-01 -4.18871790e-01
3.93605605e-02 4.20161575e-01 -1.97260511e+00 4.29912984e-01
4.81114715e-01 8.20266366e-01 6.79992318e-01 1.36851871e+00
1.61693559e-03 3.91973764e-01 -1.13177884e+00 2.64491946e-01
3.61395180e-01 -7.41776645e-01 -3.24100256e-01 -3.91670093e-02
4.10732508e-01 -3.62890840e-01 4.72519845e-02 6.94626570e-01
1.04076636e+00 3.25753033e-01 4.29892153e-01 -6.08624637e-01
-1.33824363e-01 1.25474083e+00 8.45963806e-02 1.54588118e-01
3.79899591e-01 2.22140595e-01 1.00249672e+00 -8.75603378e-01
1.89754888e-01 1.19463778e+00 5.33240199e-01 6.77510560e-01
-1.36264873e+00 -7.73244143e-01 -7.71514475e-01 -9.06563327e-02
-1.22804546e+00 -1.01408839e+00 1.00442755e+00 -2.88349390e-01
1.01674783e+00 8.60302985e-01 6.38936043e-01 8.53297710e-01
-5.02137780e-01 4.93677050e-01 8.59838188e-01 -9.62632895e-01
2.10637286e-01 1.70453787e-01 -6.02525920e-02 -5.18674813e-02
-9.85309958e-01 4.13809836e-01 -9.06460226e-01 9.17908326e-02
5.90947568e-01 -8.64886522e-01 -6.23578608e-01 4.44013953e-01
-1.12400961e+00 6.73310280e-01 -8.65792930e-02 6.03658438e-01
-4.03491288e-01 -7.60736242e-02 5.98366737e-01 3.85094911e-01
1.13013051e-01 6.60940826e-01 -3.09510052e-01 -4.78822857e-01
-1.59473658e+00 3.09347689e-01 1.04839146e+00 9.39530611e-01
1.14445992e-01 9.20499146e-01 -4.88438487e-01 1.43569458e+00
2.79299557e-01 5.63576341e-01 8.01310599e-01 -1.01325858e+00
3.34378660e-01 -9.86140445e-02 -2.14948244e-02 -2.78490752e-01
-2.68871725e-01 -3.70530486e-01 -2.76744425e-01 1.75633773e-01
1.93789437e-01 -3.13527375e-01 -7.53119469e-01 1.48380613e+00
2.73472309e-01 1.92781389e-01 3.03140074e-01 9.21679676e-01
1.02267098e+00 1.18614507e+00 -5.47033906e-01 -8.25984418e-01
1.17340672e+00 -1.23680866e+00 -1.27174008e+00 6.22051023e-02
-1.79003537e-01 -1.26488185e+00 1.33309066e+00 6.00413978e-01
-1.35287559e+00 -1.14881408e+00 -9.79931831e-01 -9.20995250e-02
8.73810947e-02 5.07244349e-01 -1.94876596e-01 1.11243641e+00
-1.02177906e+00 5.85378826e-01 -4.62597042e-01 5.85070858e-03
-4.06942606e-01 4.42680508e-01 -2.06100382e-03 8.45283985e-01
-1.12949431e+00 5.56003153e-01 5.47698855e-01 1.51431084e-01
-9.11153734e-01 -9.01152015e-01 -7.35204160e-01 3.50536667e-02
1.01411298e-01 -1.29700378e-01 1.76043010e+00 -8.05790186e-01
-2.58684564e+00 3.88992757e-01 2.57719476e-02 -6.34435117e-01
2.83281833e-01 -3.34627241e-01 -7.08440602e-01 3.72792810e-01
-2.94681132e-01 3.48808169e-01 1.34404445e+00 -9.68790948e-01
-5.71410537e-01 1.65946230e-01 -6.50750041e-01 4.99792278e-01
-2.63228506e-01 2.64939457e-01 -3.68721992e-01 -1.02690065e+00
-2.79922009e-01 -6.90124333e-01 3.42728436e-01 -5.44809997e-01
-6.56755328e-01 2.92176694e-01 1.03266251e+00 -9.79385197e-01
1.55002415e+00 -2.31314731e+00 2.52647877e-01 5.85312173e-02
-4.89403903e-01 6.60538316e-01 -1.48931623e-01 5.95462561e-01
2.09286977e-02 -3.17805409e-01 -3.09856862e-01 -5.46021163e-01
-1.71052337e-01 -1.36057228e-01 -6.22026742e-01 1.44362673e-01
-2.31244345e-03 4.72694337e-01 -4.93945777e-01 -1.63076326e-01
5.58490992e-01 5.64309001e-01 -4.80205715e-01 7.35843003e-01
-2.44740352e-01 7.46149719e-01 2.42663607e-01 3.30338180e-01
3.94616485e-01 1.11456490e+00 -7.77291954e-02 -1.98144183e-01
-6.34294033e-01 6.69904828e-01 -1.30257106e+00 1.47196901e+00
-8.34022760e-01 5.29738545e-01 7.41748691e-01 -5.74075162e-01
1.49193871e+00 7.89161325e-01 1.48889227e-02 -2.03818575e-01
1.81565613e-01 6.46577179e-01 4.41611737e-01 -7.01431990e-01
7.05682635e-01 -4.49039638e-01 3.07413936e-01 2.68198460e-01
2.38385335e-01 -1.10982800e+00 6.38135448e-02 -4.76043403e-01
4.95933592e-01 2.43868738e-01 2.24692300e-01 -3.75079811e-01
8.10254753e-01 -1.03634529e-01 2.78474838e-01 3.70798379e-01
1.79771632e-01 9.39422846e-01 -1.58555761e-01 3.90828878e-01
-1.09990788e+00 -9.48161721e-01 -1.47444487e-01 1.18934536e+00
-5.05483210e-01 -2.78609931e-01 -1.08567631e+00 1.83247849e-01
-2.98657328e-01 1.22999811e+00 -8.67347643e-02 1.95613220e-01
-9.31714952e-01 -8.79925191e-02 8.70389163e-01 9.51652527e-02
6.55805171e-02 -1.67077816e+00 -3.97416323e-01 6.90041363e-01
-1.09821819e-01 -9.19931054e-01 -1.02572143e+00 3.96880478e-01
-5.80117047e-01 -5.23455918e-01 -9.50061023e-01 -8.79290938e-01
-1.85982302e-01 -1.82309806e-01 4.76269066e-01 -4.15390879e-01
-1.24724947e-01 1.18145257e-01 -5.55176973e-01 -7.02452183e-01
-1.38529658e+00 1.61416456e-02 5.61685443e-01 2.40885243e-01
-8.92539695e-02 -7.66538203e-01 -3.45051974e-01 3.48638326e-01
-7.45324552e-01 -4.27504107e-02 1.81990430e-01 7.07280874e-01
6.50562763e-01 8.81502852e-02 1.04701424e+00 -4.23549831e-01
9.58781660e-01 1.32274227e-02 -6.04313672e-01 -2.90046811e-01
-6.53280551e-03 -3.57451677e-01 1.26152623e+00 -1.05715847e+00
-1.12458348e+00 4.66272756e-02 -6.54469669e-01 -7.10378051e-01
-1.96249843e-01 1.83043107e-01 -4.29161429e-01 3.62514585e-01
7.82593608e-01 3.64712864e-01 2.75154769e-01 -7.31971145e-01
3.58041137e-01 1.55649698e+00 1.04028440e+00 -1.41799256e-01
8.18689287e-01 -2.11079344e-01 -4.66042906e-01 -1.64358854e+00
-4.51395661e-01 -5.48923790e-01 -6.04362667e-01 -2.58311242e-01
5.69233775e-01 -6.42573893e-01 -6.62623227e-01 7.04963923e-01
-1.09694612e+00 -5.76109231e-01 -7.50178456e-01 9.01144385e-01
-1.00859129e+00 2.12470010e-01 -7.19382286e-01 -1.17690563e+00
-8.58725727e-01 -1.23319721e+00 8.77267241e-01 3.56007189e-01
-4.69401330e-01 -5.16898215e-01 2.67909646e-01 5.41295826e-01
6.31219327e-01 -2.40763396e-01 5.57594299e-01 -7.79479265e-01
1.36650532e-01 2.30748430e-02 7.28977859e-01 1.02830243e+00
4.21512395e-01 1.98507383e-01 -1.53356671e+00 -1.22334421e-01
4.62601095e-01 -2.68732935e-01 4.08473730e-01 6.41507685e-01
4.07187998e-01 -4.85591352e-01 3.15331072e-01 4.29031551e-01
7.89062381e-01 4.95913595e-01 4.38131988e-01 -2.08349958e-01
3.95076871e-01 7.96960592e-01 1.01438737e+00 3.53814989e-01
-4.05098438e-01 8.07137787e-01 1.42850012e-01 2.09348783e-01
-6.90583229e-01 -2.98266709e-01 7.44977474e-01 1.43807495e+00
6.60345107e-02 -1.51114330e-01 -3.58339787e-01 4.52678323e-01
-1.05566454e+00 -1.04473364e+00 -2.86699444e-01 2.31718516e+00
1.21255982e+00 3.36479098e-02 3.97418529e-01 7.61414468e-01
1.08190691e+00 2.10431203e-01 -4.84093159e-01 -1.17662024e+00
-1.23550175e-02 7.52997577e-01 1.13169499e-01 9.24227774e-01
-7.99251735e-01 9.63822424e-01 5.92683077e+00 1.16183591e+00
-1.56746352e+00 -1.23278722e-01 -1.00990318e-01 -4.08980221e-01
-1.12854324e-01 -4.34111208e-01 -8.56104672e-01 3.25632274e-01
1.46388900e+00 -2.65141129e-01 8.98778737e-01 6.02388144e-01
1.01603019e+00 2.02565163e-01 -1.10578096e+00 9.27238166e-01
6.32392913e-02 -1.18377042e+00 -1.12230867e-01 -3.13964427e-01
3.08502823e-01 -3.44578326e-01 9.21071097e-02 3.27489108e-01
-6.37243509e-01 -9.96488512e-01 1.20500517e+00 3.56083572e-01
1.13658679e+00 -8.49086106e-01 2.95540154e-01 6.41506076e-01
-1.19946170e+00 -1.46834590e-02 -3.61486785e-02 6.92812726e-02
4.69309598e-01 1.21481515e-01 -1.44899058e+00 3.52385938e-01
2.64035434e-01 1.79501429e-01 2.20350493e-02 9.04789627e-01
-1.46638542e-01 1.28909171e+00 -5.19134045e-01 -1.35895154e-02
-1.82069004e-01 -1.68117151e-01 1.14222753e+00 1.51202536e+00
5.66164792e-01 6.41135126e-02 -2.10600302e-01 9.43511188e-01
1.52137965e-01 5.44160008e-01 -2.53642797e-01 -3.50694388e-01
7.63647437e-01 1.26940262e+00 -1.35182694e-01 -1.06309755e-02
1.28235608e-01 8.48880351e-01 -5.05995095e-01 1.59473717e-01
-6.55945897e-01 -6.18646204e-01 2.70788938e-01 1.08824685e-01
4.43880290e-01 3.41992825e-02 1.49533376e-01 -3.66389781e-01
-1.63495868e-01 -1.09362364e+00 -1.80300623e-01 -8.24381292e-01
-7.48990238e-01 9.36598659e-01 -2.39416212e-01 -1.39358783e+00
-6.63632214e-01 -2.98392951e-01 -1.03253675e+00 1.35586309e+00
-1.08581781e+00 -1.03999686e+00 2.18909413e-01 3.10636014e-01
1.24559224e+00 -4.92322594e-01 1.18447661e+00 1.35151476e-01
-2.79941499e-01 4.87842917e-01 6.53822348e-02 -3.18446577e-01
7.69906044e-01 -1.23929751e+00 1.40323848e-01 6.51414156e-01
9.47615504e-02 3.16184342e-01 1.21693695e+00 -2.96860605e-01
-1.18438256e+00 -9.58633125e-01 6.69448376e-01 2.23040238e-01
5.85281551e-01 -4.45833743e-01 -8.91046286e-01 1.60463423e-01
3.90326589e-01 -4.41113561e-01 8.67425203e-01 -5.74265599e-01
8.40288922e-02 -3.00876230e-01 -1.07790577e+00 2.88941115e-01
2.51510918e-01 -5.50209641e-01 -9.81450021e-01 -5.08949114e-03
9.68226969e-01 -5.75898290e-01 -8.65457296e-01 1.70310453e-01
5.23227751e-01 -9.62500572e-01 5.57343364e-01 1.66885890e-02
1.49407506e-01 -5.63991964e-01 -1.72067821e-01 -1.52311277e+00
1.87815160e-01 -1.51324320e+00 -1.86965615e-02 1.83466911e+00
5.00101507e-01 -2.92542018e-02 1.97605357e-01 -2.81094741e-02
-4.92051274e-01 -5.74073307e-02 -8.67955148e-01 -8.86188447e-01
-5.71328104e-02 -5.93189836e-01 3.16454262e-01 5.52377760e-01
1.58061832e-01 7.08380222e-01 -6.41295671e-01 3.31447035e-01
3.05957109e-01 -1.55736776e-02 7.21978545e-01 -7.72579253e-01
-7.00617850e-01 -3.06199014e-01 1.75622344e-01 -7.57335007e-01
-8.15686733e-02 -7.82259643e-01 4.75815088e-01 -9.87775862e-01
-6.79379523e-01 -1.31454438e-01 7.24484622e-02 -8.96218885e-03
8.28026757e-02 2.31796220e-01 4.78803366e-01 -5.41169289e-03
5.16139567e-01 6.32762909e-01 1.24901259e+00 2.35951934e-02
-1.05703425e+00 7.62590468e-01 -5.90528138e-02 6.79102838e-01
8.68464291e-01 -2.22751245e-01 -3.95465016e-01 2.13346392e-01
-5.87445736e-01 7.47193515e-01 -1.28749564e-01 -1.17275560e+00
1.89206019e-01 2.59783804e-01 -2.96797425e-01 -7.77251184e-01
9.56409216e-01 -6.46004617e-01 3.50810170e-01 2.87643194e-01
-6.67321920e-01 -4.11725193e-01 5.65599501e-01 1.15573235e-01
-4.58756328e-01 -5.69292247e-01 1.16020155e+00 1.88146040e-01
1.08518619e-02 -4.22053695e-01 -6.04129255e-01 -3.10445607e-01
7.36747503e-01 -2.86100984e-01 1.58087075e-01 -5.37492931e-01
-1.06392372e+00 -1.55576348e-01 -1.47707984e-01 3.71542841e-01
5.26796222e-01 -1.06287467e+00 -9.80750084e-01 3.18417192e-01
-2.11847231e-01 -1.77431852e-02 5.58126152e-01 5.98593891e-01
-5.19550264e-01 4.58304644e-01 -1.18431345e-01 -4.39923972e-01
-1.69986200e+00 4.50434446e-01 5.07023215e-01 3.17494482e-01
-5.01740754e-01 6.74858689e-01 -1.16521582e-01 -5.57259023e-01
5.28007388e-01 -2.76275396e-01 -5.48056424e-01 2.75885344e-01
5.81908286e-01 6.24208570e-01 1.41622452e-02 -7.98773706e-01
-6.48858696e-02 3.01812142e-01 3.05051297e-01 -7.59900153e-01
1.24366415e+00 -1.97095200e-01 3.94592844e-02 1.04094994e+00
9.84163344e-01 8.82063508e-01 -9.62173223e-01 2.04439491e-01
-3.22699815e-01 -1.06783621e-01 1.93702146e-01 -9.47682440e-01
-6.40978515e-01 9.19306934e-01 3.53156865e-01 3.24828804e-01
1.18634737e+00 -2.71249563e-01 8.45947921e-01 4.97403070e-02
-1.02159180e-01 -1.31814790e+00 -1.19103596e-01 4.09613729e-01
1.45964301e+00 -6.28940642e-01 -5.11019289e-01 -3.70427459e-01
-7.94077396e-01 1.55275106e+00 2.75483280e-01 3.98230404e-02
5.11633515e-01 6.12066865e-01 4.64992285e-01 5.29607236e-01
-5.07933021e-01 -4.72909473e-02 2.41692081e-01 6.21754587e-01
5.44288635e-01 5.81463650e-02 -2.72106886e-01 6.92960620e-01
-1.14074218e+00 -1.00422323e-01 6.73899114e-01 9.39568579e-02
-5.87249339e-01 -1.11028302e+00 -8.68002653e-01 1.35253206e-01
-7.56321967e-01 -4.24186438e-01 -3.99039626e-01 4.45718437e-01
-2.16174293e-02 1.33371532e+00 1.48549706e-01 -4.04870093e-01
5.05110741e-01 3.44764888e-01 2.86094397e-01 -9.60163236e-01
-1.15628183e+00 1.10737729e+00 2.78765529e-01 -9.96105075e-02
-2.24723876e-01 -6.40153766e-01 -1.43612254e+00 2.66280025e-01
-5.52274466e-01 4.04152095e-01 9.31249082e-01 6.40132189e-01
-1.55953601e-01 8.34164202e-01 9.84521151e-01 -1.09042978e+00
-6.89054072e-01 -1.47314119e+00 -9.12410438e-01 -1.63421333e-01
6.73416495e-01 1.34419635e-01 -7.12233663e-01 3.42359692e-01] | [15.431940078735352, 6.209712028503418] |
2deb479a-a461-455a-b555-938545277922 | weakly-supervised-text-to-sql-parsing-through | null | null | https://openreview.net/forum?id=T4mIFZTlEF | https://openreview.net/pdf?id=T4mIFZTlEF | Weakly Supervised Text-to-SQL Parsing through Question Decomposition | Text-to-SQL parsers are crucial in enabling non-experts to effortlessly query relational data. Training such parsers, by contrast, generally requires expert annotation of natural language (NL) utterances paired with corresponding SQL queries.In this work, we propose a weak supervision approach for training text-to-SQL parsers. We take advantage of the recently proposed question meaning representation called QDMR, an intermediate between NL and formal query languages.We show that given questions, their QDMR structures (annotated by non-experts or automatically predicted), and the answers, we can automatically synthesize SQL queries that are then used to train text-to-SQL models. Extensive experiments test our approach on five benchmark datasets. The results show that our models perform competitively with those trained on annotated NL-SQL data.Overall, we effectively train text-to-SQL parsers, using zero SQL annotations. | ['Anonymous'] | 2022-01-16 | null | null | null | acl-arr-january-2022-1 | ['text-to-sql'] | ['computer-code'] | [ 1.26888052e-01 7.59469509e-01 -1.30014747e-01 -9.42041457e-01
-1.44639814e+00 -7.60807753e-01 4.85380530e-01 4.20253068e-01
-2.75677025e-01 3.52003664e-01 2.89063305e-01 -8.08591664e-01
2.03116626e-01 -1.13615537e+00 -1.03115940e+00 4.07026321e-01
4.31747347e-01 9.99251544e-01 6.66423678e-01 -3.96591276e-01
1.30017087e-01 2.44618982e-01 -1.28615093e+00 8.19595397e-01
8.77817929e-01 1.01215684e+00 1.28770486e-01 8.84504378e-01
-7.50699878e-01 1.74211383e+00 -5.19038022e-01 -8.77196252e-01
-1.09075524e-01 -1.73572347e-01 -1.42196000e+00 -2.24688217e-01
3.73145193e-01 -4.42222059e-01 -8.46163854e-02 7.02863753e-01
4.31012772e-02 -1.89185306e-01 1.49836674e-01 -7.63699532e-01
-7.18750656e-01 8.41669440e-01 2.70317018e-01 -2.92135805e-01
9.82745469e-01 2.18194886e-03 1.62854075e+00 -9.37893212e-01
7.24995375e-01 1.40955853e+00 3.76157671e-01 5.90572894e-01
-1.37175643e+00 1.32168949e-01 -2.07698718e-01 -2.02281058e-01
-1.00655210e+00 -5.46864748e-01 5.19065857e-01 -2.69986331e-01
1.45671606e+00 4.59599555e-01 -1.55633584e-01 7.15056777e-01
-2.50510704e-02 8.34807754e-01 7.36735702e-01 -6.04956985e-01
1.69015139e-01 4.80637342e-01 3.95237148e-01 8.88123155e-01
-1.36347324e-01 -2.27661222e-01 -5.55979311e-01 -3.64548355e-01
4.27493900e-01 -2.70159811e-01 1.10722527e-01 -2.47412130e-01
-8.65569234e-01 1.00509048e+00 1.35082155e-01 7.65882581e-02
-2.23492444e-01 4.30004783e-02 5.05211830e-01 7.06276119e-01
2.18359217e-01 7.74048686e-01 -8.67632449e-01 -4.66086864e-02
-4.18983430e-01 2.98493505e-01 1.44242406e+00 1.31611621e+00
9.45703387e-01 -3.89247030e-01 -1.78777054e-01 7.56162107e-01
2.82685369e-01 5.69789290e-01 1.84936613e-01 -1.26450229e+00
8.55847776e-01 9.88219976e-01 1.44053325e-01 -6.67649090e-01
-1.84374303e-01 1.50911927e-01 -2.09207460e-01 -4.98442858e-01
2.68376589e-01 -9.57973022e-03 -4.16375965e-01 1.41599965e+00
2.73028642e-01 -6.23036265e-01 6.47047102e-01 5.14615417e-01
9.41136777e-01 6.34225965e-01 1.87482849e-01 -1.91657096e-02
1.49665356e+00 -7.32146561e-01 -5.77974737e-01 -2.85386264e-01
9.45156753e-01 -5.27176499e-01 1.69440663e+00 2.64701277e-01
-1.14662158e+00 -7.00275838e-01 -5.16821921e-01 -5.12907088e-01
-2.81562358e-01 2.15729475e-01 4.24812645e-01 4.88838285e-01
-1.04901254e+00 1.39250249e-01 -9.18869376e-01 -3.74443859e-01
8.58957171e-02 3.24675471e-01 -2.45295659e-01 -1.05826162e-01
-1.03316760e+00 7.71905303e-01 4.83319521e-01 -2.54497707e-01
-7.95417309e-01 -6.43660069e-01 -1.01182163e+00 9.77034867e-02
9.49950933e-01 -5.29175222e-01 1.96143651e+00 -4.13166314e-01
-1.64808595e+00 9.70313489e-01 -5.23887396e-01 -6.26105547e-01
1.04024835e-01 -4.55445498e-01 -1.48507848e-01 3.48545909e-01
3.17124277e-01 5.85570097e-01 2.21102014e-01 -1.08640957e+00
-5.43074787e-01 -3.43753159e-01 5.34130216e-01 -2.40214065e-01
-5.41854203e-02 4.64390278e-01 -4.73757684e-01 -5.45606203e-02
1.81461796e-01 -4.41530019e-01 -2.73137033e-01 -2.84944803e-01
-7.26053178e-01 -7.54036307e-01 4.64771569e-01 -5.70276797e-01
1.02437091e+00 -1.86155498e+00 -2.92645723e-01 9.00853053e-02
1.28004506e-01 -1.12567283e-02 -1.87690675e-01 7.44537830e-01
7.78986067e-02 2.65283376e-01 -1.60703227e-01 -4.94603552e-02
3.17457139e-01 6.54450953e-01 -9.53108668e-01 -4.71009642e-01
9.84700203e-01 1.17744350e+00 -7.45160878e-01 -8.08267951e-01
-6.65135086e-02 -1.41108856e-01 -8.12693596e-01 1.13507783e+00
-1.15041745e+00 2.73934782e-01 -8.45075786e-01 5.76513827e-01
2.04727963e-01 -3.40441912e-01 4.19661641e-01 -6.71602339e-02
1.84490725e-01 9.89868641e-01 -8.60221326e-01 1.49744141e+00
-6.42112374e-01 2.23898530e-01 -2.41107002e-01 -9.93978858e-01
1.25923049e+00 4.17710602e-01 -1.68803837e-02 -9.04938519e-01
-3.54555070e-01 4.09702808e-01 -5.20411611e-01 -9.84344482e-01
4.24368680e-01 -1.37563586e-01 -6.08341873e-01 4.61404443e-01
3.89849097e-01 -6.15521789e-01 3.47151756e-01 4.09718335e-01
1.30674815e+00 2.43936390e-01 2.70125926e-01 7.02292249e-02
8.42126250e-01 3.57285291e-01 2.29911894e-01 9.49496686e-01
4.33900267e-01 2.89291620e-01 1.09697640e+00 -5.22863030e-01
-7.88340271e-01 -1.18542993e+00 1.26741588e-01 1.55534589e+00
-3.98925185e-01 -6.58055425e-01 -6.68435454e-01 -1.05793691e+00
-3.39764774e-01 1.09941339e+00 -8.99661556e-02 1.21610336e-01
-8.56109381e-01 1.58421248e-01 7.05716014e-01 6.78572595e-01
3.80705521e-02 -1.29853392e+00 -7.57449687e-01 4.45203692e-01
-1.50335684e-01 -1.62849462e+00 -9.42107663e-02 4.87958968e-01
-9.93589461e-01 -1.17212045e+00 1.89107135e-01 -8.11271310e-01
4.73949492e-01 -2.62903720e-01 1.80913353e+00 2.02133268e-01
1.31020471e-01 6.06820881e-01 -4.84320968e-01 -4.14807022e-01
-9.76283252e-01 3.01382184e-01 -6.49634123e-01 -3.16130847e-01
5.88163316e-01 -1.65250778e-01 -6.27499968e-02 2.50448078e-01
-1.30777013e+00 -5.35779595e-02 6.21061146e-01 5.50525546e-01
7.57511973e-01 -3.17915738e-01 6.86202824e-01 -1.62113559e+00
6.71924055e-01 -2.39701316e-01 -8.88735235e-01 7.86938548e-01
-3.52070868e-01 7.47896194e-01 9.95020449e-01 1.31093383e-01
-1.24465656e+00 2.33467937e-01 -4.97122467e-01 8.45190883e-02
-4.32539970e-01 7.81764269e-01 -4.55066115e-01 3.64101708e-01
8.02299321e-01 1.84964202e-02 -3.77120316e-01 -4.58037674e-01
7.20438302e-01 6.76054358e-01 7.35677421e-01 -1.18882644e+00
7.96108902e-01 4.56037261e-02 -2.52372742e-01 -4.84844595e-01
-1.17866921e+00 -3.40968102e-01 -5.51472545e-01 4.71080095e-01
9.57026243e-01 -7.29143977e-01 -7.36231923e-01 -2.56778598e-01
-1.49016595e+00 -4.00259137e-01 -6.65289521e-01 2.37988271e-02
-6.67121053e-01 1.96013525e-01 -6.79061294e-01 -9.20071959e-01
-2.54376382e-01 -9.14761126e-01 1.30611897e+00 4.44054641e-02
-3.12646657e-01 -8.54533732e-01 2.52066523e-01 5.81597507e-01
2.98752278e-01 -8.50828141e-02 1.52225065e+00 -1.41549897e+00
-8.34778726e-01 -3.04614097e-01 -3.31802607e-01 3.97590935e-01
-8.00680071e-02 -8.07703882e-02 -8.69587958e-01 4.27163035e-01
1.77052006e-01 -1.11656368e+00 3.20788562e-01 -3.93715113e-01
1.23898315e+00 -6.08295202e-01 5.81079274e-02 5.39841549e-03
1.46467996e+00 -4.59766239e-02 5.58474779e-01 -9.20753777e-02
3.21863651e-01 1.05283999e+00 7.22941220e-01 1.86929896e-01
9.40848410e-01 3.61577600e-01 3.74179274e-01 1.36856228e-01
3.60895365e-01 -8.01557362e-01 2.58036971e-01 8.69230330e-01
5.81162095e-01 -6.75289258e-02 -1.14373147e+00 5.12938857e-01
-1.51145256e+00 -5.45871913e-01 -3.21467847e-01 1.82844424e+00
1.39415753e+00 2.28096694e-01 -8.37744474e-02 -2.88827151e-01
1.46827951e-01 -3.68843041e-02 -4.26730841e-01 -7.31425643e-01
2.97277775e-02 9.26611900e-01 5.40229306e-02 6.22697234e-01
-8.91875982e-01 1.15458345e+00 5.91075277e+00 2.66695499e-01
-6.95694447e-01 1.83417033e-02 4.79203701e-01 5.59834301e-01
-7.52778590e-01 3.86889011e-01 -8.95429254e-01 -2.97627270e-01
1.52417004e+00 2.52257362e-02 3.40771377e-01 1.03783667e+00
-1.60797443e-02 -4.11442369e-02 -1.69160068e+00 3.68984491e-01
-1.36896059e-01 -1.26250267e+00 2.65638292e-01 -5.78222752e-01
3.19856733e-01 1.88760702e-02 -3.96751642e-01 7.41215050e-01
6.91351235e-01 -1.03203225e+00 5.11147559e-01 6.79041266e-01
5.63110590e-01 -3.98937255e-01 8.08617175e-01 5.91834247e-01
-1.14484918e+00 -3.44077051e-02 -5.56895256e-01 2.03142256e-01
-7.52910152e-02 2.90257931e-01 -1.11558378e+00 8.42408419e-01
5.30471742e-01 2.23147109e-01 -9.98345017e-01 1.84904978e-01
-4.93655503e-01 5.84847093e-01 -2.32268393e-01 -1.99121729e-01
3.41307209e-03 4.15799916e-02 -1.78760722e-01 1.13272214e+00
3.16230915e-02 1.97525874e-01 3.23501468e-01 1.16237164e+00
-3.02452326e-01 2.59744018e-01 -6.68329418e-01 -3.48171890e-01
3.65298986e-01 9.98807788e-01 -1.42147958e-01 -5.98726511e-01
-8.59119713e-01 6.58997238e-01 5.44027209e-01 2.88939416e-01
-4.14474368e-01 -5.97929239e-01 -4.85388422e-03 2.66622156e-01
3.02754670e-01 1.79106351e-02 -1.91756904e-01 -1.19486821e+00
5.21275461e-01 -1.41138780e+00 5.79548180e-01 -9.89270091e-01
-1.45831168e+00 6.70642734e-01 8.61121248e-03 -6.06615603e-01
-7.20615506e-01 -6.60867095e-01 -3.98667037e-01 7.80793548e-01
-1.59552085e+00 -1.04976594e+00 -8.65502879e-02 4.40165043e-01
3.99020761e-01 1.01974703e-01 1.17138791e+00 -9.56728309e-03
-1.32069558e-01 2.96784788e-01 -5.92435479e-01 5.38856804e-01
4.86030042e-01 -1.50502086e+00 5.74488938e-01 5.26846409e-01
4.48865086e-01 9.37478840e-01 4.44386959e-01 -4.95249480e-01
-1.86505044e+00 -1.27631330e+00 1.40234673e+00 -9.31457520e-01
8.27168882e-01 -4.14056540e-01 -1.29546332e+00 1.04682696e+00
3.19454163e-01 4.67535071e-02 8.44190240e-01 2.88219657e-02
-5.82291842e-01 -2.65438735e-01 -9.28813756e-01 2.55898356e-01
5.65840781e-01 -1.19336057e+00 -1.19241929e+00 5.63468039e-01
1.50969160e+00 -3.79943341e-01 -1.14638948e+00 4.31940138e-01
1.58262596e-01 -1.00244033e+00 7.45164812e-01 -1.13714266e+00
5.38810194e-01 -1.89932108e-01 -6.13986909e-01 -4.42040205e-01
6.45355165e-01 -5.66753089e-01 -1.40802627e-02 1.23838580e+00
7.11170077e-01 -5.30389965e-01 8.76253068e-01 1.05137026e+00
-1.35560647e-01 -7.54920483e-01 -6.53502762e-01 -4.15470392e-01
4.19043973e-02 -7.69082725e-01 6.51688814e-01 4.80388790e-01
-9.61254176e-04 8.23427796e-01 3.03754449e-01 2.15386331e-01
1.75440699e-01 2.56835431e-01 1.08597088e+00 -1.27072060e+00
-6.55844331e-01 1.36647806e-01 1.09992743e-01 -1.51112711e+00
3.44279945e-01 -9.21365678e-01 2.98411250e-01 -1.63257241e+00
4.25500721e-02 -3.47778857e-01 2.82163054e-01 6.61206186e-01
-2.31261998e-01 -5.22231281e-01 -1.75008401e-02 -2.63485126e-02
-9.44435656e-01 2.95641720e-01 8.26067567e-01 8.26129541e-02
-1.56869486e-01 2.00518683e-01 -7.86164582e-01 5.26128232e-01
6.26830995e-01 -7.46529400e-01 -5.47098100e-01 -7.36125767e-01
7.27292061e-01 6.78910732e-01 3.56145144e-01 -6.59495354e-01
3.72492403e-01 -1.68858886e-01 -2.60541856e-01 -7.78030455e-01
-3.15901190e-02 -6.75265610e-01 -3.51892859e-01 2.23419935e-01
-8.84504199e-01 3.13776523e-01 -7.09470436e-02 3.43569785e-01
-6.56998754e-01 -5.72806418e-01 3.12608033e-01 -3.04469019e-01
-2.29391500e-01 -4.82324734e-02 -2.86435723e-01 8.29712749e-01
4.48895067e-01 4.57323253e-01 -3.91438216e-01 -5.41853011e-01
-5.66899896e-01 3.99338186e-01 1.66953325e-01 1.00100748e-01
7.89467454e-01 -8.46298397e-01 -5.30549705e-01 2.43046001e-01
4.34814960e-01 3.85149091e-01 -4.19847071e-01 3.01866680e-01
-6.76012576e-01 1.08942771e+00 3.87633830e-01 -6.52399719e-01
-7.32551932e-01 5.51122546e-01 2.30873525e-01 -7.49982715e-01
-4.41485383e-02 7.04501152e-01 3.68127041e-02 -1.31752753e+00
3.12482297e-01 -9.32126820e-01 -4.04882245e-03 -4.83041644e-01
1.91755846e-01 -2.51449466e-01 1.99158236e-01 -6.55924007e-02
-2.04325289e-01 4.21784461e-01 8.05373192e-02 -3.05553794e-01
1.34177673e+00 2.24067807e-01 -5.32381594e-01 4.58379149e-01
1.17887402e+00 1.75750792e-01 -7.15334058e-01 -7.42220759e-01
1.07292283e+00 -1.46177217e-01 -7.55362093e-01 -6.35392249e-01
-5.04252672e-01 1.12609410e+00 -8.49109814e-02 6.42935991e-01
1.12497103e+00 6.00589097e-01 8.70891631e-01 1.40118277e+00
4.06093210e-01 -7.44349301e-01 5.22439241e-01 6.99237943e-01
7.34504998e-01 -1.25125802e+00 -4.89814043e-01 -4.70143050e-01
-6.85529351e-01 1.30867195e+00 8.28408837e-01 -3.22792446e-03
1.70126185e-01 3.27893525e-01 2.63120413e-01 -3.76853406e-01
-1.26993048e+00 -1.93307236e-01 1.07668780e-01 5.42483032e-01
7.06948280e-01 -2.03996912e-01 2.11927325e-01 7.57526517e-01
-4.11626011e-01 7.39648193e-02 4.14884061e-01 1.27066195e+00
-6.48350477e-01 -1.77831125e+00 -1.45822689e-01 5.23957491e-01
-5.98978162e-01 -2.89236337e-01 -6.30981266e-01 7.05767035e-01
-4.18724447e-01 1.33968568e+00 -4.39526066e-02 -1.18244983e-01
8.27463269e-01 5.30703783e-01 2.66075760e-01 -1.26441443e+00
-9.14632320e-01 -3.50452363e-01 6.29777670e-01 -7.17903018e-01
-2.26398155e-01 -2.98576176e-01 -1.53635657e+00 1.53478667e-01
-1.16884366e-01 5.84492683e-01 4.10578400e-01 9.10434186e-01
3.93704027e-01 2.16329232e-01 6.35731459e-01 3.69124323e-01
-1.13424563e+00 -7.35704124e-01 1.14970326e-01 2.09885016e-01
1.37404829e-01 2.84937233e-01 1.54641435e-01 3.42498034e-01] | [10.017335891723633, 7.848401069641113] |
61172307-e941-481f-bdff-7c2483f3349f | multi-layer-content-interaction-through | 2001.05840 | null | https://arxiv.org/abs/2001.05840v2 | https://arxiv.org/pdf/2001.05840v2.pdf | Multi-Layer Content Interaction Through Quaternion Product For Visual Question Answering | Multi-modality fusion technologies have greatly improved the performance of neural network-based Video Description/Caption, Visual Question Answering (VQA) and Audio Visual Scene-aware Dialog (AVSD) over the recent years. Most previous approaches only explore the last layers of multiple layer feature fusion while omitting the importance of intermediate layers. To solve the issue for the intermediate layers, we propose an efficient Quaternion Block Network (QBN) to learn interaction not only for the last layer but also for all intermediate layers simultaneously. In our proposed QBN, we use the holistic text features to guide the update of visual features. In the meantime, Hamilton quaternion products can efficiently perform information flow from higher layers to lower layers for both visual and text modalities. The evaluation results show our QBN improved the performance on VQA 2.0, even though using surpass large scale BERT or visual BERT pre-trained models. Extensive ablation study has been carried out to testify the influence of each proposed module in this study. | ['Peng Gao', 'Songxiang Liu', 'Shijie Geng', 'Lei Shi', 'Sen Su', 'Kai Shuang', 'Chiori Hori'] | 2020-01-03 | null | null | null | null | ['video-description'] | ['computer-vision'] | [-3.07751894e-01 -1.25900283e-01 1.88866481e-01 -3.75258148e-01
-5.27498841e-01 -3.81090581e-01 6.85394049e-01 4.31718007e-02
-6.40846312e-01 5.86422145e-01 4.06087995e-01 -5.68670919e-03
2.76795357e-01 -3.86921704e-01 -5.74798167e-01 -4.28592950e-01
1.33006200e-01 2.31204614e-01 3.90576273e-01 -6.03308737e-01
1.09131359e-01 4.12562370e-01 -1.41894352e+00 5.46384275e-01
5.34631670e-01 1.15341496e+00 2.75344521e-01 9.56836879e-01
-3.54199469e-01 1.32794273e+00 -5.82783818e-01 -6.17335439e-01
-1.07792027e-01 -1.99770957e-01 -8.29392970e-01 -6.21316880e-02
4.69972849e-01 -7.25862801e-01 -5.57851672e-01 9.03977096e-01
7.19045222e-01 1.11292444e-01 4.61921871e-01 -1.78990340e+00
-5.19455194e-01 4.62921292e-01 -5.60254931e-01 1.14340827e-01
5.32697082e-01 2.77053773e-01 1.17640877e+00 -9.40167308e-01
5.51907599e-01 1.51677370e+00 4.47827727e-01 4.83854055e-01
-6.92170501e-01 -6.33401453e-01 3.48235071e-01 9.20476079e-01
-1.52076542e+00 -4.28342342e-01 9.72087204e-01 -3.41798544e-01
9.50954080e-01 2.06474558e-01 8.04045737e-01 8.28706026e-01
9.11521241e-02 1.18564987e+00 6.15878046e-01 4.90283892e-02
-1.03987157e-01 1.83464587e-01 -7.06105307e-02 1.00397646e+00
-1.96055055e-01 -2.81893015e-01 -8.73804927e-01 1.38105333e-01
7.77728796e-01 -1.80425808e-01 -2.19813213e-01 -4.19135690e-01
-1.34142029e+00 7.64929295e-01 7.32917249e-01 2.05270767e-01
-4.26370353e-01 4.17543232e-01 6.99903488e-01 1.24536581e-01
6.33468553e-02 1.51608109e-01 -4.03524518e-01 -1.70021519e-01
-7.68378854e-01 1.16271086e-01 6.65032744e-01 9.09858048e-01
8.71579528e-01 1.92760244e-01 -5.18992364e-01 5.23927391e-01
5.04804134e-01 4.59279180e-01 1.70564502e-01 -1.05736458e+00
7.14539587e-01 5.94866157e-01 1.58725113e-01 -1.06119728e+00
-5.20915329e-01 9.16727353e-03 -1.03177631e+00 -5.94404005e-02
1.38921961e-01 -2.68433243e-01 -8.75252903e-01 1.56198812e+00
3.29081416e-01 5.92996031e-02 1.45145655e-01 1.22009075e+00
1.44434249e+00 9.24504697e-01 3.17898810e-01 -1.34275153e-01
1.28459239e+00 -1.14767313e+00 -1.04998887e+00 1.94126517e-02
4.34826493e-01 -7.01230884e-01 9.79229748e-01 3.49822700e-01
-1.17661500e+00 -8.99212539e-01 -1.14226198e+00 -4.34679270e-01
-3.85340869e-01 2.95856804e-01 8.68023157e-01 3.47577959e-01
-1.26246274e+00 1.89873382e-01 -6.74388707e-01 -1.77224904e-01
1.69418484e-01 5.45417547e-01 -5.50870419e-01 1.54049814e-01
-1.48487663e+00 8.98926377e-01 3.92880052e-01 5.03114343e-01
-1.02450275e+00 -3.57267499e-01 -9.81002152e-01 2.53733307e-01
2.45275021e-01 -7.24881530e-01 1.33918655e+00 -9.24469829e-01
-1.71950948e+00 3.04637551e-01 -1.72615051e-01 -4.69586998e-01
4.22455251e-01 -3.19365859e-01 -1.39040768e-01 5.77388048e-01
-3.42613697e-01 1.26887798e+00 9.17180896e-01 -1.14609480e+00
-6.59339845e-01 -6.86330423e-02 5.51262677e-01 6.04459286e-01
-3.65438104e-01 -1.86015263e-01 -9.01114166e-01 -2.37099752e-01
-7.68529922e-02 -6.47033691e-01 3.77008431e-02 3.02272856e-01
-9.10554454e-02 -3.39463204e-01 8.79126191e-01 -8.64158511e-01
1.14239156e+00 -1.98345447e+00 3.52004200e-01 -2.00716957e-01
2.46700689e-01 3.53602141e-01 -3.43894899e-01 5.44126630e-01
1.75265327e-01 -2.40353569e-01 2.08689094e-01 -6.34576082e-01
1.13611780e-01 2.17954427e-01 -1.94110841e-01 4.09803152e-01
3.23153675e-01 1.12533951e+00 -6.00574672e-01 -8.54477763e-01
5.30008316e-01 7.71471500e-01 -7.85652697e-01 2.23196387e-01
-3.13436478e-01 3.01655829e-01 -2.12017044e-01 4.72980917e-01
8.32580566e-01 -3.39228779e-01 -1.44125775e-01 -9.30722296e-01
-1.26567677e-01 -7.61008486e-02 -1.23115170e+00 2.01725578e+00
-5.19994915e-01 8.40146482e-01 2.41696924e-01 -6.36243582e-01
4.88105267e-01 5.66917419e-01 5.50403118e-01 -8.04609120e-01
2.63147563e-01 -3.77581030e-01 -1.63635537e-01 -6.05837166e-01
8.88865769e-01 6.33524731e-02 1.13224871e-02 -4.08735797e-02
4.96828914e-01 -1.02018431e-01 6.86633959e-02 5.90932012e-01
5.13475299e-01 1.95225090e-01 4.26358789e-01 3.93986374e-01
9.91378784e-01 -2.93751024e-02 2.57650018e-01 5.35793602e-01
-5.44757128e-01 6.11186922e-01 4.40857559e-01 -5.53777181e-02
-8.62673283e-01 -9.46138859e-01 3.89145792e-01 1.06635416e+00
4.33899105e-01 -6.56408966e-01 -3.76389056e-01 -5.86166024e-01
-2.02786326e-01 4.48724836e-01 -4.13804799e-01 -1.03366554e-01
-2.96608865e-01 -2.70187199e-01 6.51767194e-01 5.41730940e-01
1.07098770e+00 -9.50403690e-01 -4.33480024e-01 1.50408924e-01
-5.39980173e-01 -1.45034969e+00 -4.40233409e-01 -3.33169371e-01
-4.17921126e-01 -7.63335168e-01 -8.85041595e-01 -5.54575205e-01
2.48804808e-01 2.61502057e-01 6.81887746e-01 -5.71653508e-02
1.52350143e-01 6.96722209e-01 -4.16071415e-01 -3.43203425e-01
-2.11114526e-01 -2.03261785e-02 -1.42906174e-01 1.74787536e-01
-2.98697948e-02 -1.50918826e-01 -5.58058679e-01 2.47067556e-01
-8.41029286e-01 5.22329688e-01 5.69385111e-01 7.24277914e-01
1.68931141e-01 -4.45768416e-01 4.41049814e-01 -1.71426862e-01
4.97884005e-01 -6.63867891e-02 -5.34253776e-01 5.48185825e-01
-7.24312961e-02 1.71537504e-01 4.96822149e-01 -4.30207849e-01
-1.13059127e+00 8.46324638e-02 -3.88959169e-01 -5.15132189e-01
1.35939375e-01 5.69394886e-01 -3.91630977e-01 -3.14290911e-01
2.31106848e-01 4.14974801e-02 -1.22710794e-01 -1.44719929e-01
6.44843221e-01 7.64612436e-01 4.39136833e-01 -3.60223204e-02
7.37477660e-01 5.89418292e-01 6.68157861e-02 -8.10218811e-01
-6.13354087e-01 -6.41662002e-01 -5.92906415e-01 -6.20396495e-01
1.16309690e+00 -1.25289977e+00 -1.50285208e+00 4.39143658e-01
-1.51323533e+00 1.10861890e-01 8.79043713e-02 5.90991914e-01
-3.80932719e-01 6.59662664e-01 -5.02552569e-01 -7.65248120e-01
-3.99414182e-01 -1.24618542e+00 1.19589710e+00 3.15617710e-01
1.62433177e-01 -8.32852244e-01 -2.46631607e-01 6.16815090e-01
3.11061919e-01 -1.82498187e-01 4.33911234e-01 -2.68919289e-01
-8.20101976e-01 -1.07553499e-02 -6.04393959e-01 4.67997134e-01
-1.12498999e-01 -5.33813089e-02 -1.08153296e+00 -2.65454412e-01
-2.14049891e-01 -4.73965168e-01 9.62455034e-01 1.21896148e-01
6.94111943e-01 -2.11585566e-01 1.11023419e-01 5.57566643e-01
1.19068670e+00 1.39085412e-01 5.85163891e-01 1.12187877e-01
1.02905130e+00 2.93273956e-01 6.58992589e-01 5.20440638e-01
9.23944175e-01 6.35161340e-01 7.19639182e-01 -2.05657482e-01
-2.25212276e-01 -6.38016015e-02 5.30797660e-01 1.09055853e+00
1.40318591e-02 -3.66682023e-01 -7.64193475e-01 4.34465051e-01
-1.85182762e+00 -8.39724779e-01 1.55172005e-01 1.76101935e+00
7.12597668e-01 -3.72817107e-02 -1.42414123e-01 -8.57638419e-02
4.10309941e-01 4.51080590e-01 -3.63353610e-01 -3.64053637e-01
-1.98284864e-01 -2.35199764e-01 2.97116190e-01 7.65109003e-01
-1.20720196e+00 1.26851082e+00 5.67270708e+00 7.32896745e-01
-1.29794872e+00 9.17038471e-02 1.04809768e-01 -1.14357816e-02
-1.88175023e-01 -1.25692319e-02 -5.99841416e-01 -1.20417342e-01
6.02584422e-01 -1.57318041e-02 3.65402520e-01 4.65060711e-01
1.54810220e-01 -2.76168108e-01 -7.99259663e-01 1.36697531e+00
2.65583932e-01 -1.41676986e+00 4.92391646e-01 -4.31322008e-01
3.43312174e-01 5.17078713e-02 7.14576468e-02 5.29840589e-01
-2.20925529e-02 -8.02232146e-01 7.02896833e-01 7.90532827e-01
6.97684884e-01 -8.16440701e-01 8.14761102e-01 9.75716263e-02
-1.32321596e+00 3.13576427e-03 -2.69246161e-01 2.58198436e-02
3.48678887e-01 -9.25235376e-02 -1.17699838e+00 8.27069700e-01
4.97258097e-01 6.47838414e-01 -8.14518929e-01 9.60041523e-01
-1.29945204e-01 3.36116582e-01 -2.64052927e-01 -1.69628680e-01
4.46945131e-01 1.30197152e-01 5.62528074e-01 9.87068892e-01
8.34064633e-02 -1.06676765e-01 -1.37934268e-01 2.87985682e-01
1.00140050e-01 1.89054742e-01 -2.32953742e-01 -1.56424001e-01
-9.14160982e-02 1.28189743e+00 -2.09786355e-01 -4.39701945e-01
-6.51459455e-01 1.25415504e+00 2.02368632e-01 6.31384552e-01
-1.07250440e+00 -3.03426683e-01 6.29978359e-01 -4.10009414e-01
4.74381089e-01 -6.11782432e-01 9.69698280e-02 -1.47812390e+00
-2.65745997e-01 -6.71892464e-01 3.79837900e-01 -1.28581440e+00
-7.41950989e-01 5.16818583e-01 1.46604240e-01 -1.13152850e+00
-2.98242897e-01 -6.11271620e-01 -9.96534824e-02 7.74032235e-01
-1.63572526e+00 -1.38755012e+00 -5.59180915e-01 8.97579610e-01
3.41850549e-01 -8.54256079e-02 5.99280596e-01 4.21752483e-01
-4.42548960e-01 4.93157208e-01 -2.18123615e-01 1.40704334e-01
8.49967599e-01 -9.89405334e-01 1.24909393e-01 6.77229345e-01
3.75334889e-01 2.57218599e-01 7.36081302e-01 -4.36403066e-01
-1.46742117e+00 -7.42412388e-01 7.62869358e-01 -1.12513162e-01
6.07999623e-01 -4.31864321e-01 -7.24774897e-01 5.26530206e-01
5.75301230e-01 -2.48372957e-01 2.01748520e-01 -2.11983368e-01
-1.81874946e-01 -4.11366522e-01 -7.68680751e-01 7.24648774e-01
5.81255257e-01 -8.28228116e-01 -3.09971422e-01 1.40203238e-01
9.12222087e-01 -4.87932533e-01 -7.13183939e-01 3.76663506e-01
5.67983687e-01 -9.76693213e-01 9.83790576e-01 -4.42837536e-01
1.45333484e-01 -5.42218506e-01 -2.43796006e-01 -9.22610462e-01
5.09025296e-03 -6.09007895e-01 -2.57621497e-01 1.23511207e+00
1.65080264e-01 -2.23361298e-01 6.81761563e-01 2.48005465e-01
8.00343975e-02 -3.50483209e-01 -1.01492023e+00 1.49980793e-02
-4.17057961e-01 -5.64880729e-01 5.62357724e-01 6.41814947e-01
-1.62461162e-01 7.55905569e-01 -8.16386640e-01 2.60074139e-01
2.24306390e-01 -7.19717294e-02 8.93353343e-01 -9.05712306e-01
-1.28190354e-01 -8.96648392e-02 -7.33016968e-01 -1.34618461e+00
1.12289019e-01 -6.46052480e-01 -9.31610018e-02 -1.88579261e+00
1.50535554e-01 3.05023700e-01 -2.28683144e-01 4.68755782e-01
-2.33005196e-01 2.26296395e-01 7.48406112e-01 -1.02545671e-01
-1.09093416e+00 1.04174280e+00 1.70557737e+00 -3.26670200e-01
2.91221472e-03 -3.01107526e-01 -4.20026213e-01 6.00112259e-01
5.82062066e-01 3.70685160e-02 -6.36007428e-01 -6.10305548e-01
3.44217330e-01 4.05980945e-01 8.17759395e-01 -1.12416446e+00
5.78734338e-01 -8.13164413e-02 4.19769436e-01 -1.01797867e+00
9.60485935e-01 -9.91452992e-01 -1.18519150e-01 1.14494622e-01
-2.33741298e-01 2.95466423e-01 3.96930367e-01 3.55045289e-01
-6.22236371e-01 9.22078043e-02 4.66251791e-01 8.30182359e-02
-1.19872236e+00 3.81372482e-01 -3.99695307e-01 -3.58618200e-01
8.00480425e-01 -8.85722339e-02 -3.50523859e-01 -1.05691504e+00
-7.44284987e-01 5.38132131e-01 1.14401706e-01 4.63601202e-01
8.63125563e-01 -1.30941141e+00 -4.84903067e-01 -3.36050913e-02
1.08553834e-01 -2.33536437e-01 6.29780769e-01 9.65676606e-01
-7.16249704e-01 7.50591338e-01 -4.68344092e-01 -6.89171731e-01
-1.33765709e+00 4.23950255e-01 4.04607385e-01 -2.20408574e-01
-1.80536479e-01 9.17237937e-01 2.64278501e-01 -1.94082022e-01
4.99983788e-01 -2.26002678e-01 -6.83727384e-01 4.81384784e-01
2.94935733e-01 1.99793875e-01 -1.91627532e-01 -1.01892531e+00
-4.62625593e-01 4.04870152e-01 -1.07382402e-01 -5.33365428e-01
8.78819942e-01 -4.59483236e-01 6.66804165e-02 5.57297885e-01
1.39212513e+00 -3.82918715e-01 -1.37119651e+00 -1.27766669e-01
-5.15549183e-01 -4.84610796e-02 2.94319987e-01 -6.20530725e-01
-1.08829415e+00 1.32572329e+00 6.19505346e-01 -2.19519570e-01
1.22509158e+00 -2.39635751e-01 6.49934351e-01 6.65628135e-01
7.85796642e-02 -1.07980490e+00 4.13267791e-01 7.40065932e-01
1.21498096e+00 -1.26871645e+00 1.98438197e-01 -2.02003077e-01
-1.12909472e+00 1.09077406e+00 8.31217229e-01 2.27566406e-01
4.32269305e-01 -9.48949307e-02 2.92538017e-01 -5.64257428e-02
-9.98954535e-01 -4.65506405e-01 4.92727071e-01 2.43986830e-01
1.71060249e-01 -3.45004112e-01 -8.99908394e-02 2.51757026e-01
-1.60618965e-02 -4.42868210e-02 4.17647421e-01 7.94626474e-01
-2.12006390e-01 -7.37652361e-01 -2.20478088e-01 8.41742605e-02
-9.62711275e-02 -2.23278597e-01 -2.90462613e-01 1.07728803e+00
2.22211424e-02 9.65532303e-01 1.09008260e-01 -5.65571785e-01
3.55822504e-01 -9.07293335e-02 5.70888460e-01 -5.99308200e-02
-8.15078795e-01 1.57645643e-01 1.37767270e-01 -6.59016013e-01
-8.23552907e-01 -4.79961693e-01 -1.37170231e+00 -1.73549578e-01
-3.96093577e-01 1.47694066e-01 8.37774158e-01 8.85673523e-01
2.29290068e-01 5.72605789e-01 3.40010405e-01 -8.93270910e-01
-3.65623474e-01 -1.09534192e+00 -3.98702443e-01 1.57673851e-01
6.52016699e-01 -6.74202621e-01 -1.65097192e-01 -1.98579766e-02] | [10.579328536987305, 1.1653881072998047] |
d61a235c-8887-4ea9-860f-16aca9618dab | protein-sequence-design-with-batch-bayesian | 2303.10429 | null | https://arxiv.org/abs/2303.10429v1 | https://arxiv.org/pdf/2303.10429v1.pdf | Protein Sequence Design with Batch Bayesian Optimisation | Protein sequence design is a challenging problem in protein engineering, which aims to discover novel proteins with useful biological functions. Directed evolution is a widely-used approach for protein sequence design, which mimics the evolution cycle in a laboratory environment and conducts an iterative protocol. However, the burden of laboratory experiments can be reduced by using machine learning approaches to build a surrogate model of the protein landscape and conducting in-silico population selection through model-based fitness prediction. In this paper, we propose a new method based on Batch Bayesian Optimization (Batch BO), a well-established optimization method, for protein sequence design. By incorporating Batch BO into the directed evolution process, our method is able to make more informed decisions about which sequences to select for artificial evolution, leading to improved performance and faster convergence. We evaluate our method on a suite of in-silico protein sequence design tasks and demonstrate substantial improvement over baseline algorithms. | ['Chuanjiao Zong'] | 2023-03-18 | null | null | null | null | ['protein-design', 'bayesian-optimisation'] | ['medical', 'methodology'] | [ 6.47000670e-01 -2.98233837e-01 6.93211183e-02 -2.74876922e-01
-5.39828241e-01 -6.34160221e-01 2.21600741e-01 2.09492326e-01
-5.37273288e-01 1.10167086e+00 -1.89678609e-01 -6.96315169e-01
6.37322068e-02 -4.06316191e-01 -9.23456192e-01 -8.81222069e-01
1.56594679e-01 7.11057067e-01 2.11432800e-01 -2.44514093e-01
4.57972318e-01 6.23329639e-01 -1.30530226e+00 4.92049828e-02
1.05179274e+00 3.37325990e-01 5.06429017e-01 7.51939654e-01
1.71726141e-02 4.02438253e-01 -5.21850109e-01 -3.48965317e-01
-5.79046039e-03 -1.00430226e+00 -6.50683343e-01 2.48065293e-02
-4.02120858e-01 2.38248155e-01 2.99531311e-01 8.63971949e-01
8.85420561e-01 2.00095862e-01 5.91863990e-01 -4.21173602e-01
-4.95648712e-01 2.87824124e-01 -4.06905919e-01 9.77418572e-02
3.18044156e-01 6.73472226e-01 1.00884974e+00 -7.93693244e-01
6.90012038e-01 1.12339246e+00 6.11332357e-01 6.20249927e-01
-1.92964709e+00 -3.33637834e-01 -4.88775112e-02 3.14871132e-01
-1.23095751e+00 -2.86283642e-01 6.52285516e-01 -3.97070855e-01
1.20765352e+00 2.49013066e-01 9.04572904e-01 8.52105260e-01
4.11035597e-01 6.22145236e-01 8.67066443e-01 -4.50605750e-01
8.42763305e-01 -2.68246353e-01 -1.79723412e-01 6.54336989e-01
7.78784230e-02 3.96521688e-01 -6.29543185e-01 -9.32720900e-01
1.70183703e-01 -1.53941974e-01 -4.08812165e-01 -7.55702138e-01
-9.48987246e-01 9.43281591e-01 9.18263197e-02 -1.88413784e-01
-7.75875449e-01 -3.89415100e-02 8.44540223e-02 2.50388198e-02
1.78315327e-01 9.90856349e-01 -7.93245256e-01 -3.25653970e-01
-7.26050436e-01 5.85705340e-01 1.04000890e+00 4.61350590e-01
6.75514162e-01 -2.39602253e-01 -1.49021238e-01 8.45225453e-01
3.73461694e-01 1.82329252e-01 4.13598627e-01 -7.68426895e-01
-3.85694772e-01 5.10288954e-01 3.21956158e-01 -4.69736308e-01
-3.04829717e-01 -3.70043725e-01 -1.29315019e-01 3.83077294e-01
3.13313156e-01 -1.62897542e-01 -9.26796913e-01 1.68005168e+00
6.13456011e-01 -7.62776881e-02 1.49635762e-01 6.90992057e-01
1.19907334e-01 5.94740808e-01 1.63164467e-01 -5.16992152e-01
1.34831643e+00 -8.35366845e-01 -4.49824929e-01 6.44577369e-02
4.88976270e-01 -5.60175180e-01 8.80312085e-01 6.48640156e-01
-9.26838934e-01 -1.58363476e-01 -1.28681958e+00 3.04626793e-01
-1.22363634e-01 -3.50987524e-01 6.86744034e-01 8.05369258e-01
-7.88323820e-01 9.82659638e-01 -9.85990405e-01 -2.90285945e-01
4.30807471e-01 6.88765824e-01 -3.16320732e-03 8.39788690e-02
-8.00535917e-01 9.49976146e-01 6.42883956e-01 -2.86878571e-02
-8.02722752e-01 -7.57330358e-01 -5.49484789e-01 4.88373935e-02
5.47726214e-01 -8.84880960e-01 1.14461541e+00 -6.84524834e-01
-2.03410244e+00 5.18407166e-01 -5.19491315e-01 -5.43732584e-01
1.85181767e-01 -2.19402183e-02 1.25428423e-01 -1.63876206e-01
-5.39796591e-01 5.67029834e-01 5.30969083e-01 -8.46040487e-01
-3.38873804e-01 -3.52877289e-01 -3.64971608e-01 1.60670206e-01
1.30146131e-01 2.71571994e-01 -1.40516520e-01 -5.39301336e-01
-1.20771229e-01 -1.20316148e+00 -8.31236959e-01 -1.99361831e-01
1.50979133e-02 -1.96063548e-01 3.73402387e-01 -5.98725021e-01
1.20088208e+00 -1.65849316e+00 6.38722062e-01 4.00532156e-01
1.23492755e-01 4.20679718e-01 -8.54059588e-03 4.57430005e-01
9.43927020e-02 -7.10701272e-02 -5.99077404e-01 3.74331683e-01
-1.91796988e-01 8.45075473e-02 -2.61197705e-02 1.75282076e-01
3.65937829e-01 1.00000143e+00 -8.86279643e-01 -5.64874560e-02
-9.49982405e-02 4.21593815e-01 -8.74422550e-01 4.85309124e-01
-8.00677240e-01 7.72331297e-01 -4.69712555e-01 6.59176886e-01
4.01952326e-01 -5.33175647e-01 8.24122906e-01 1.45823166e-01
-1.72567070e-02 8.51282328e-02 -6.73015594e-01 1.58513057e+00
-8.55426118e-02 7.80736208e-02 -2.91320920e-01 -9.20019984e-01
1.08102572e+00 6.72608241e-02 5.03845274e-01 -2.38994971e-01
1.98578864e-01 2.27930769e-01 4.58267391e-01 -3.37685734e-01
1.74870901e-02 -1.26016855e-01 2.93091625e-01 5.55345893e-01
-2.18184724e-01 -1.64961398e-01 2.99926132e-01 -3.53705257e-01
1.37768972e+00 7.14864075e-01 5.31636894e-01 -3.09937239e-01
4.64658797e-01 4.58042920e-01 9.56138492e-01 5.87007642e-01
-1.25610664e-01 3.59034538e-01 3.79393518e-01 -6.24563873e-01
-1.29659534e+00 -6.67521596e-01 1.60350055e-02 1.06400001e+00
-9.18947011e-02 -3.87652248e-01 -8.39502513e-01 -6.22135580e-01
9.30674970e-02 5.98698497e-01 -3.15384746e-01 -5.70553422e-01
-8.23545158e-01 -1.45350027e+00 3.08931857e-01 1.89714372e-01
-1.69371441e-01 -1.18860912e+00 -7.98306644e-01 7.98555076e-01
1.28933107e-02 -4.73654717e-01 -4.73759264e-01 5.29782236e-01
-8.81994367e-01 -8.84286284e-01 -9.21309590e-01 -7.85910428e-01
5.86951256e-01 -7.53442198e-02 1.06326115e+00 1.59609959e-01
-7.04586387e-01 -3.83827835e-01 -2.28540495e-01 -5.55752993e-01
-7.10382938e-01 -2.06649527e-02 -7.84326438e-03 -2.43552610e-01
8.24763000e-01 -6.51467025e-01 -8.51328135e-01 4.45317715e-01
-7.14199662e-01 -4.56819907e-02 5.70246816e-01 1.51535964e+00
8.56143355e-01 -1.45984799e-01 5.66367626e-01 -9.13024247e-01
8.22161317e-01 -1.44084081e-01 -1.00947094e+00 5.89595973e-01
-9.53790069e-01 7.23737001e-01 4.82999474e-01 -6.53324366e-01
-9.47808444e-01 4.49720681e-01 -3.49675387e-01 8.75576362e-02
1.18929036e-01 6.50528789e-01 -2.91346341e-01 -4.49262917e-01
7.61213243e-01 4.09592479e-01 3.23680043e-01 -6.18128836e-01
1.85598210e-01 4.73306060e-01 5.23670241e-02 -8.08736801e-01
3.39897513e-01 7.00759813e-02 1.70383118e-02 -7.23447263e-01
-1.88250750e-01 -2.73805261e-01 -4.25131261e-01 1.19258292e-01
6.71683371e-01 -5.50349712e-01 -1.16750276e+00 2.89114326e-01
-8.42163265e-01 -4.46583956e-01 1.22517474e-01 3.94881964e-01
-8.07541966e-01 6.39421403e-01 -2.99105555e-01 -7.62031257e-01
-4.43708718e-01 -1.52193797e+00 9.20627654e-01 2.36373484e-01
-5.48868835e-01 -6.35310709e-01 4.90316480e-01 3.93440247e-01
4.13031220e-01 3.08540195e-01 1.21459174e+00 -6.70262516e-01
-7.18576193e-01 1.61387771e-01 3.71130764e-01 4.19636704e-02
-1.40453000e-02 6.58654869e-02 -3.43245625e-01 -4.60122645e-01
-1.20153919e-01 -3.84785473e-01 8.02285373e-01 3.77368271e-01
1.00508964e+00 5.97782549e-04 -5.21522820e-01 7.67703235e-01
1.24172568e+00 7.32865393e-01 5.33165514e-01 5.37960410e-01
2.56332427e-01 5.79300582e-01 7.97070861e-01 5.82170367e-01
2.94136386e-02 9.00942028e-01 -2.18692459e-02 4.08790931e-02
3.85372519e-01 -1.08334407e-01 2.74201423e-01 2.70261973e-01
-1.40559524e-01 -3.82550657e-01 -9.28664565e-01 1.37388855e-01
-2.04851007e+00 -8.38684499e-01 2.79814273e-01 2.24476457e+00
1.40389574e+00 -5.09323366e-02 4.69663292e-01 -2.78125554e-01
6.17742121e-01 -4.65492457e-01 -1.18281174e+00 -4.47763860e-01
-2.91919321e-01 5.59976935e-01 4.92588937e-01 2.59231269e-01
-7.18202889e-01 9.78272915e-01 7.41832495e+00 7.13030994e-01
-9.97851372e-01 -3.82257611e-01 6.83073521e-01 -7.77897611e-02
-1.52441993e-01 2.89634228e-01 -8.44265103e-01 4.62399811e-01
1.18068206e+00 -1.97624832e-01 6.70378864e-01 5.45316160e-01
4.10920024e-01 5.00676397e-04 -8.84166062e-01 5.28395355e-01
-2.40778923e-01 -1.62477982e+00 -1.94255218e-01 2.57676601e-01
7.27646053e-01 -1.40837163e-01 -2.07837641e-01 -1.47523150e-01
7.42664218e-01 -9.96613383e-01 2.73272216e-01 3.94880176e-01
1.65947258e-01 -1.04098451e+00 3.84010315e-01 3.29403311e-01
-6.24057233e-01 -3.56251113e-02 -4.48196083e-01 3.40796113e-01
4.19400036e-01 5.00735283e-01 -9.71917570e-01 1.78353995e-01
6.34265780e-01 3.58041614e-01 -2.51199871e-01 1.11613941e+00
-1.56513274e-01 7.27735758e-01 -2.71326810e-01 -5.49415946e-01
-1.37949228e-01 -5.97757876e-01 5.78745782e-01 1.02094114e+00
1.27355516e-01 2.34960824e-01 2.41804719e-01 1.13052666e+00
1.97084486e-01 2.39604458e-01 -1.39916956e-01 -4.73353684e-01
2.91151583e-01 8.31431210e-01 -6.50148571e-01 -1.07781202e-01
7.11418390e-02 9.56228793e-01 5.05448401e-01 4.31872427e-01
-9.09705579e-01 -6.48911715e-01 7.51814544e-01 -2.50804424e-01
7.77042270e-01 -2.82802731e-01 1.54894823e-03 -8.81626844e-01
-2.27974534e-01 -1.46403885e+00 2.16801509e-01 -4.36393768e-01
-9.53031003e-01 3.53619576e-01 -4.18917209e-01 -5.54538906e-01
-2.63949096e-01 -7.50943184e-01 -2.82937467e-01 1.06838548e+00
-1.29603589e+00 -4.68605757e-01 2.64127523e-01 -9.62170660e-02
4.70612764e-01 -2.33731776e-01 6.75883412e-01 -8.19416642e-02
-6.91431999e-01 5.37185729e-01 5.82982957e-01 -7.34878480e-01
5.84040940e-01 -1.21512043e+00 6.54869020e-01 5.61265051e-01
-2.68515229e-01 8.75018358e-01 1.16801012e+00 -9.68202353e-01
-1.57675290e+00 -7.71794975e-01 6.05856538e-01 -3.01892519e-01
4.37298208e-01 -4.45258707e-01 -1.21968031e+00 1.44901365e-01
-3.53696309e-02 -5.75445354e-01 9.71475184e-01 3.53476149e-03
1.74689032e-02 2.82029599e-01 -1.15288949e+00 8.09027910e-01
1.15455103e+00 -1.02970690e-01 -3.15262347e-01 4.07874942e-01
8.13798428e-01 -2.46918485e-01 -1.01511705e+00 5.17329752e-01
7.91608095e-01 -5.98510087e-01 1.03227079e+00 -8.93451691e-01
3.98974568e-02 -6.47956610e-01 8.07608739e-02 -1.30386388e+00
-6.24766886e-01 -1.13989699e+00 -2.00571746e-01 7.71423757e-01
9.03041124e-01 -6.63734317e-01 1.06630826e+00 5.79238117e-01
1.62890106e-01 -1.01493776e+00 -4.07950252e-01 -7.41084158e-01
-1.35263890e-01 3.61775309e-01 8.88136327e-01 3.48649323e-01
7.11273551e-02 3.37885916e-01 -3.87907833e-01 -1.11714445e-01
4.73426104e-01 1.24998875e-01 7.20303178e-01 -1.02419186e+00
-1.00007617e+00 -5.24705589e-01 -2.08691597e-01 -1.21385264e+00
5.46681732e-02 -6.06894374e-01 4.69721407e-01 -1.04765940e+00
3.11209917e-01 -6.95349872e-02 -3.15676779e-01 1.98837787e-01
-6.77796006e-01 -2.71875747e-02 -2.53795147e-01 5.70930056e-02
-3.56942207e-01 6.92577064e-01 9.66028869e-01 -3.72006111e-02
-4.66211736e-01 9.20771360e-02 -7.28140295e-01 3.36330950e-01
7.84259200e-01 -6.01565778e-01 -2.22489759e-01 1.96246147e-01
2.25693122e-01 -1.40179724e-01 -1.52505577e-01 -5.70357144e-01
-4.01949044e-03 -4.47121292e-01 3.39023381e-01 -5.10378838e-01
2.80704349e-01 -4.88763392e-01 5.45824766e-01 9.34014082e-01
-4.45091873e-01 2.40631714e-01 9.81460065e-02 8.45679939e-01
1.29208624e-01 -2.39285722e-01 8.46261859e-01 -2.10570127e-01
-4.49680567e-01 1.62213847e-01 -6.76177502e-01 -3.27612609e-01
1.08688807e+00 -3.70458156e-01 4.75381464e-02 5.17439879e-02
-8.16450000e-01 2.70482868e-01 8.44520330e-01 1.44835323e-01
7.17967570e-01 -7.68531740e-01 -7.68649042e-01 2.81751573e-01
1.94027007e-01 -4.30124938e-01 -1.58643633e-01 7.47564256e-01
-9.24257040e-01 3.70661914e-01 2.78599355e-02 -7.28954852e-01
-1.55695283e+00 7.01585770e-01 3.80095214e-01 -3.11634839e-01
-4.01973188e-01 8.77205014e-01 2.32824255e-02 -4.66486007e-01
4.45195176e-02 3.43211293e-01 4.58816290e-02 -5.83795726e-01
5.41104794e-01 1.15618244e-01 1.37520254e-01 -2.23500714e-01
-3.96781713e-01 4.07688379e-01 -2.22417757e-01 -9.04714391e-02
1.52024746e+00 5.57225682e-02 -1.86481163e-01 -2.07193960e-02
8.11363280e-01 -3.60474259e-01 -1.46909428e+00 -2.23163843e-01
5.88499486e-01 -4.26447064e-01 -1.30693674e-01 -1.12655520e+00
-3.59431982e-01 3.15492243e-01 6.82446659e-01 -4.99327570e-01
1.06344938e+00 -1.87690735e-01 8.03184152e-01 8.62610221e-01
5.16290128e-01 -9.32149470e-01 -6.28970265e-02 4.88531709e-01
5.52207053e-01 -1.08471203e+00 4.89934944e-02 -1.48915827e-01
-5.22162318e-01 1.00831795e+00 2.96155065e-01 1.65037677e-01
3.77346069e-01 1.41147062e-01 -2.17280257e-02 -1.77693486e-01
-1.13603973e+00 -6.08045235e-02 1.71249777e-01 6.22512043e-01
5.26859581e-01 -1.71372101e-01 -6.86909199e-01 2.38883555e-01
1.01148382e-01 1.32097945e-01 8.34283978e-02 1.43281043e+00
-7.23342836e-01 -1.83089423e+00 -2.86817491e-01 2.41702437e-01
-5.32757938e-01 -1.63195923e-01 -6.89970076e-01 -1.39090950e-02
-2.32115850e-01 9.15395558e-01 -5.52842557e-01 -1.72724694e-01
2.64157593e-01 2.39979953e-01 6.79732800e-01 -4.37978923e-01
-6.45573199e-01 2.91000873e-01 1.35115340e-01 -3.31508785e-01
-1.64615437e-01 -8.23018610e-01 -1.19804764e+00 -1.98742077e-01
-8.55307162e-01 6.38767958e-01 8.36630821e-01 7.79144764e-01
1.01280344e+00 4.33827102e-01 5.57629228e-01 -6.19297862e-01
-8.36749315e-01 -4.88199979e-01 -2.63822913e-01 2.87993997e-01
-3.13228182e-02 -5.72068930e-01 4.85268235e-02 2.11779997e-01] | [4.7366623878479, 5.560551166534424] |
1fef0604-ed84-4823-8190-fd318c4da128 | on-the-ideal-number-of-groups-for-isometric | 2302.03193 | null | https://arxiv.org/abs/2302.03193v1 | https://arxiv.org/pdf/2302.03193v1.pdf | On the Ideal Number of Groups for Isometric Gradient Propagation | Recently, various normalization layers have been proposed to stabilize the training of deep neural networks. Among them, group normalization is a generalization of layer normalization and instance normalization by allowing a degree of freedom in the number of groups it uses. However, to determine the optimal number of groups, trial-and-error-based hyperparameter tuning is required, and such experiments are time-consuming. In this study, we discuss a reasonable method for setting the number of groups. First, we find that the number of groups influences the gradient behavior of the group normalization layer. Based on this observation, we derive the ideal number of groups, which calibrates the gradient scale to facilitate gradient descent optimization. Our proposed number of groups is theoretically grounded, architecture-aware, and can provide a proper value in a layer-wise manner for all layers. The proposed method exhibited improved performance over existing methods in numerous neural network architectures, tasks, and datasets. | ['Sang Woo Kim', 'Hyeonah Jang', 'Hyeyeon Choi', 'Bum Jun Kim'] | 2023-02-07 | null | null | null | null | ['panoptic-segmentation'] | ['computer-vision'] | [-5.58705628e-02 -2.71000445e-01 -3.78556997e-01 -7.68777013e-01
1.90116554e-01 -3.17275375e-01 3.87743175e-01 5.53062856e-02
-8.13149512e-01 4.74105060e-01 -3.68178375e-02 -3.87998402e-01
-1.57126591e-01 -8.04233015e-01 -4.63913262e-01 -9.41581368e-01
5.84523827e-02 -1.56982616e-01 3.23472500e-01 -3.39942604e-01
4.38506275e-01 5.94171882e-01 -1.39258182e+00 7.06586391e-02
9.69873905e-01 1.13840640e+00 2.13869721e-01 1.68519586e-01
-1.85309917e-01 4.14371908e-01 -7.30848253e-01 -1.61276162e-01
3.59671116e-01 -6.02903962e-01 -3.19875687e-01 -9.49782208e-02
2.21104607e-01 -1.81856945e-01 1.01160303e-01 1.14946854e+00
6.29830241e-01 3.84902954e-01 6.86344087e-01 -1.07242405e+00
-2.81831503e-01 9.37035799e-01 -4.91457939e-01 2.88206935e-01
-4.60495651e-01 -3.77830565e-02 8.36823881e-01 -7.05758989e-01
1.82742119e-01 1.09408176e+00 5.03907084e-01 5.42012811e-01
-9.36930120e-01 -1.14415050e+00 5.81110179e-01 -4.88891602e-02
-1.43987060e+00 -4.27088112e-01 1.00617480e+00 -4.25437808e-01
7.10333288e-01 2.27366555e-02 5.58586717e-01 6.65059566e-01
2.01340869e-01 3.35551411e-01 6.10780597e-01 -7.99285233e-01
4.62331593e-01 2.07999200e-01 2.40993425e-01 5.55695117e-01
6.65466368e-01 -4.27141309e-01 -3.54223251e-01 1.40634198e-02
9.49666739e-01 -5.92377596e-02 -1.28140315e-01 -4.09328043e-01
-9.91088212e-01 8.00691962e-01 7.00272024e-01 4.90557373e-01
-2.49902636e-01 1.75578460e-01 4.92052048e-01 1.75638601e-01
2.43852749e-01 6.31181538e-01 -4.51754212e-01 4.51172858e-01
-8.38213921e-01 1.57191232e-01 5.29869139e-01 5.06413758e-01
8.11540008e-01 2.55912215e-01 -2.33271450e-01 1.10231543e+00
4.06384915e-01 -9.84219238e-02 9.24747229e-01 -4.44695741e-01
8.50754261e-01 9.13758278e-01 -1.45361394e-01 -1.09736669e+00
-6.71606541e-01 -7.17748821e-01 -9.96902883e-01 2.39725292e-01
4.97690856e-01 -5.16913772e-01 -9.14192080e-01 1.94702685e+00
2.82323331e-01 -7.82233030e-02 -9.25410539e-02 1.07275641e+00
5.88437021e-01 2.35453069e-01 4.05376971e-01 -1.26001984e-01
1.22076333e+00 -8.80376637e-01 -8.38375330e-01 -3.83390218e-01
6.83440030e-01 -6.57974899e-01 1.32796633e+00 2.16303140e-01
-6.96768463e-01 -5.98781049e-01 -1.64795804e+00 3.47546339e-01
-3.97237599e-01 5.91305017e-01 8.78193796e-01 9.43467736e-01
-9.47160602e-01 7.27976501e-01 -8.11702788e-01 -3.37761730e-01
1.92032903e-01 7.19724953e-01 -4.66592237e-02 4.74476129e-01
-1.25142491e+00 5.95946372e-01 9.25845325e-01 5.78324616e-01
-1.41896114e-01 -3.67258072e-01 -3.81597638e-01 5.43413758e-02
6.80889264e-02 -5.82849026e-01 8.53349984e-01 -1.08551323e+00
-1.73763204e+00 5.04943669e-01 4.21956144e-02 -3.41112196e-01
5.00815094e-01 8.84729251e-03 -4.57474083e-01 -1.94704488e-01
-4.16265160e-01 5.80609262e-01 9.79314148e-01 -1.08555782e+00
-4.75888222e-01 -3.46733093e-01 8.16108659e-02 3.16337168e-01
-1.07634401e+00 1.36524901e-01 -5.31240344e-01 -4.91916031e-01
6.57434821e-01 -7.41221726e-01 -4.78653044e-01 -2.37922460e-01
-2.28346676e-01 -1.58077195e-01 3.04732382e-01 -2.43075445e-01
1.59568775e+00 -2.15056062e+00 -2.94873148e-01 4.81678188e-01
2.30583027e-01 3.26306492e-01 4.44780290e-02 -7.18637090e-03
-2.32041106e-01 4.37982976e-01 2.42799893e-01 -1.66982189e-01
-1.20551437e-01 1.51835280e-02 -3.80220041e-02 4.29598957e-01
1.72748327e-01 5.37059426e-01 -5.82200885e-01 -4.44783986e-01
1.41903237e-01 3.57806116e-01 -8.09608161e-01 1.77224547e-01
1.66806549e-01 2.40044743e-01 -8.42539191e-01 2.37586603e-01
7.84803927e-01 -2.61258423e-01 3.22001994e-01 -5.14579535e-01
-2.02573031e-01 3.60163957e-01 -1.56149197e+00 1.06844878e+00
-2.31647551e-01 4.51165050e-01 -4.36437987e-02 -9.23011422e-01
1.38640916e+00 1.08274020e-01 4.12375867e-01 -3.12297583e-01
8.14450979e-01 1.53183058e-01 3.35874766e-01 -1.76488221e-01
5.73487222e-01 3.91204841e-03 3.25374812e-01 3.33710313e-01
-1.05516851e-01 3.69502932e-01 3.04238617e-01 -3.33113223e-01
5.29308617e-01 -2.26795256e-01 3.86070788e-01 -4.70910221e-01
6.85127616e-01 -4.18387502e-01 7.78033972e-01 6.91315532e-01
-1.78503290e-01 4.98492181e-01 5.46171188e-01 -5.47364652e-01
-1.03411806e+00 -4.68168706e-01 -2.75152415e-01 1.28850055e+00
1.13651991e-01 -3.53047490e-01 -1.03938019e+00 -2.56787241e-01
-2.40511745e-01 3.03498179e-01 -6.94520175e-01 -3.22019309e-01
-7.71598756e-01 -1.36489129e+00 5.94770372e-01 5.10963976e-01
7.89893806e-01 -9.93181705e-01 -4.04656470e-01 1.61633000e-01
2.35125989e-01 -9.35998440e-01 -2.43036643e-01 4.37603921e-01
-1.21624458e+00 -8.46130311e-01 -3.10238034e-01 -8.88527930e-01
1.18086255e+00 1.58256233e-01 8.27657282e-01 4.57321137e-01
3.22876096e-01 -4.33495224e-01 -2.26673022e-01 -5.63394725e-01
-1.67431802e-01 7.53229141e-01 2.08424926e-01 9.49437320e-02
1.92042097e-01 -6.63820386e-01 -6.98878169e-01 4.92799252e-01
-9.39395010e-01 5.24767339e-02 6.47504568e-01 6.44628167e-01
5.16246974e-01 1.26457825e-01 6.60736024e-01 -8.39729667e-01
1.14010584e+00 -5.90781728e-03 -7.10946143e-01 3.77988368e-01
-7.86382496e-01 3.33482921e-01 1.02003050e+00 -7.77580380e-01
-9.01791513e-01 -2.05434754e-01 -5.17335385e-02 -2.34868571e-01
8.23202953e-02 4.79067624e-01 -3.73992264e-01 -3.98147106e-01
8.57010484e-01 -2.77751982e-01 -9.29173827e-02 -3.67613852e-01
2.74637878e-01 5.19625306e-01 2.14300230e-01 -6.51487231e-01
6.54482305e-01 -4.61030528e-02 -1.10174101e-02 -5.34903407e-01
-6.12132251e-01 -2.12510303e-01 -8.58015120e-01 -1.83968991e-01
6.71411633e-01 -7.08730936e-01 -6.72706306e-01 5.63676059e-01
-8.77154827e-01 -2.50598162e-01 1.95146918e-01 6.63252532e-01
8.00122470e-02 -6.16931319e-02 -3.54446113e-01 -6.48619711e-01
-5.11191130e-01 -1.14605749e+00 2.18953893e-01 5.09972990e-01
-7.44808186e-03 -9.56295252e-01 -3.64125609e-01 -2.98363939e-02
5.23342907e-01 2.40072280e-01 8.40110838e-01 -7.88290560e-01
-3.09373856e-01 -3.17332596e-01 -8.93123448e-02 6.05821073e-01
1.85059711e-01 2.45676219e-01 -9.06943440e-01 -3.56512874e-01
2.03298196e-01 3.42922173e-02 7.38786399e-01 4.72583801e-01
1.41651666e+00 -3.82389992e-01 -2.46008545e-01 1.02609515e+00
1.28806543e+00 3.07188690e-01 5.80577254e-01 8.40878844e-01
8.45871568e-01 2.42082477e-01 2.48383224e-01 5.34183383e-01
9.53549296e-02 3.96041304e-01 3.77340108e-01 -3.26185226e-01
1.35314435e-01 2.88490914e-02 -6.96428791e-02 9.48179305e-01
-3.17243040e-01 -1.88435107e-01 -7.51288354e-01 3.56562212e-02
-1.64882231e+00 -6.08654737e-01 1.97313353e-01 2.46594405e+00
8.61917913e-01 4.39450264e-01 3.44204977e-02 2.71104544e-01
8.63346219e-01 2.52283067e-01 -6.74146175e-01 -3.45057160e-01
-1.17694221e-01 -2.50750899e-01 8.44813049e-01 1.84671581e-01
-1.05316067e+00 9.93746758e-01 6.70105886e+00 6.31430566e-01
-1.58353269e+00 -2.76220918e-01 8.23341668e-01 -8.80373567e-02
-1.24199569e-01 -2.01497301e-01 -1.24527764e+00 3.93520623e-01
7.09381998e-01 -1.49757147e-01 4.00899708e-01 8.38427365e-01
3.65872711e-01 2.97458559e-01 -8.45641911e-01 8.09292078e-01
-1.53940633e-01 -1.01159680e+00 9.00091305e-02 -8.12635720e-02
8.46058071e-01 -9.82294232e-02 -5.51445223e-02 1.56556964e-01
-3.72288488e-02 -7.89775670e-01 7.25134313e-01 1.20184831e-01
2.02261940e-01 -9.06740308e-01 7.70833313e-01 2.51861423e-01
-1.06085622e+00 -3.07586491e-01 -6.20656550e-01 -1.92974418e-01
-2.19980806e-01 7.57782042e-01 -8.83380651e-01 2.19287753e-01
7.02090681e-01 2.18784094e-01 -8.30914199e-01 1.08062005e+00
-3.19680065e-01 7.48715162e-01 -4.84284252e-01 -2.91040540e-01
1.70003459e-01 -4.44130093e-01 2.24696100e-02 1.07149518e+00
2.73622453e-01 -3.83839719e-02 1.05417863e-01 6.25319123e-01
-2.20971093e-01 5.04445612e-01 9.22139436e-02 2.74300855e-02
8.63873482e-01 1.40826321e+00 -1.01577926e+00 -7.31637850e-02
-2.74005476e-02 2.23627687e-01 4.99241471e-01 4.42866594e-01
-8.82830143e-01 -5.64883947e-01 6.26123309e-01 1.95421591e-01
1.68708041e-01 -3.73716146e-01 -7.88056910e-01 -9.45858300e-01
1.57995492e-01 -6.16735637e-01 3.44530404e-01 -2.83043474e-01
-9.19110239e-01 8.70991290e-01 -4.66018319e-02 -1.26864946e+00
-1.89761832e-01 -7.03619182e-01 -7.42857397e-01 8.28871429e-01
-1.43445539e+00 -6.05438769e-01 -4.91213053e-01 3.60660493e-01
-2.05251649e-01 -3.24546069e-01 4.83166367e-01 4.01306570e-01
-1.14576960e+00 1.08800113e+00 -2.96823983e-03 3.10679495e-01
7.79216111e-01 -7.63465762e-01 2.22531512e-01 9.85723138e-01
-4.61081356e-01 1.07622826e+00 8.71376216e-01 -2.34313071e-01
-9.53358710e-01 -8.38554978e-01 6.51387691e-01 3.16669524e-01
5.54477096e-01 -4.81980443e-01 -1.08960748e+00 3.55547160e-01
-1.83261260e-01 -8.90656710e-02 6.45665884e-01 2.72111207e-01
-7.16514662e-02 -5.91329217e-01 -9.35698569e-01 1.01247489e+00
8.99992287e-01 -6.85299113e-02 -2.55899489e-01 1.82822689e-01
6.76109791e-01 -4.44955438e-01 -8.99685025e-01 5.35752535e-01
4.90943104e-01 -9.87607360e-01 9.67646778e-01 -3.97642344e-01
2.62026563e-02 -4.88083780e-01 2.48651244e-02 -1.21827555e+00
-5.21439731e-01 -4.31682825e-01 3.58189493e-01 1.41495025e+00
5.46733558e-01 -9.15641487e-01 9.97060180e-01 9.10372913e-01
-1.11002348e-01 -1.07239485e+00 -5.24570048e-01 -6.36925101e-01
4.52120672e-04 -2.50096738e-01 1.04193330e+00 8.45047474e-01
-4.25013572e-01 2.19413027e-01 -1.10457957e-01 1.22123852e-01
3.12041372e-01 -4.59228188e-01 6.55686915e-01 -1.36405742e+00
7.04224408e-02 -7.11166203e-01 -3.49655300e-01 -8.70924532e-01
1.60981324e-02 -5.55840135e-01 -1.75107211e-01 -1.37345564e+00
-3.55094016e-01 -7.69868255e-01 -7.68308401e-01 7.84092844e-01
-3.24502558e-01 6.40880466e-02 5.94026409e-02 3.90121609e-01
-3.10838461e-01 4.69658345e-01 1.23938096e+00 2.36014307e-01
-7.93027997e-01 6.02636077e-02 -9.93555784e-01 7.54418194e-01
1.32092535e+00 -2.99569637e-01 -5.65692186e-01 -4.78243053e-01
4.55734938e-01 -7.36821055e-01 -2.47777238e-01 -1.09431827e+00
3.41646940e-01 -5.21558225e-01 6.74861789e-01 -3.12674552e-01
-1.17148841e-02 -7.21531749e-01 -1.03161827e-01 2.49780014e-01
-4.21450734e-01 1.05059259e-01 1.75639108e-01 -4.21128161e-02
-1.66943908e-01 -5.81056952e-01 7.95242310e-01 6.90789968e-02
-6.46987438e-01 2.50142515e-01 -8.08537006e-02 -1.47586778e-01
6.38892889e-01 -5.07183611e-01 -1.81452274e-01 7.31348351e-04
-7.51785636e-02 1.56446695e-01 2.08961725e-01 4.07833189e-01
1.47651061e-01 -1.44876492e+00 -3.57940525e-01 5.34972012e-01
-1.22008501e-02 6.45935982e-02 -2.90816277e-02 6.16152048e-01
-5.83493412e-01 2.27811933e-01 -3.77077550e-01 -4.85359818e-01
-1.13883197e+00 1.91244051e-01 5.08524776e-01 -1.91795781e-01
-1.81407139e-01 6.97854161e-01 1.00300321e-02 -4.48877335e-01
4.73252892e-01 -5.78009129e-01 -6.81458771e-01 2.14126602e-01
6.71062529e-01 1.89873964e-01 3.65093589e-01 -3.27463090e-01
-3.86480987e-01 6.32899821e-01 -1.96331561e-01 2.13557586e-01
1.29247391e+00 -1.29470065e-01 -2.26130351e-01 2.89800912e-01
1.05992544e+00 -3.22891951e-01 -1.32914591e+00 -3.82653363e-02
-3.40349823e-02 -1.74639657e-01 2.31907934e-01 -3.21109116e-01
-1.34886456e+00 5.18800139e-01 6.33366466e-01 1.14586420e-01
1.43317366e+00 -5.79433680e-01 1.74720168e-01 6.14746273e-01
1.95706293e-01 -1.34837127e+00 -2.02123448e-01 6.23678446e-01
6.35463715e-01 -9.97613013e-01 2.37058103e-01 -3.28161657e-01
-4.12868023e-01 1.26413083e+00 9.91990864e-01 -9.51263905e-02
6.96789324e-01 2.03019977e-01 3.33329171e-01 1.59130171e-01
-4.67262954e-01 2.01526657e-01 4.84907031e-01 2.77734250e-01
7.78201520e-01 -1.81975737e-02 -7.66213953e-01 8.62345994e-01
-6.08327925e-01 -1.21443108e-01 1.45898074e-01 6.96583986e-01
-7.31359065e-01 -1.25301981e+00 -5.02102852e-01 3.96339267e-01
-4.80886161e-01 1.79351456e-02 -3.90576907e-02 8.13654602e-01
2.38657176e-01 8.54104042e-01 6.32102191e-02 -6.27035677e-01
4.46681947e-01 -4.64226082e-02 3.14878494e-01 -2.78716505e-01
-7.74190187e-01 -1.36846751e-01 -1.11225955e-01 -1.66408077e-01
-2.73906857e-01 -1.91445991e-01 -1.44711936e+00 -3.31053227e-01
-6.72487557e-01 7.55405799e-02 9.12924528e-01 1.00597072e+00
1.90969050e-01 7.08613038e-01 6.65637910e-01 -8.41041505e-01
-6.40218973e-01 -1.05302596e+00 -4.54931170e-01 3.24773967e-01
-8.61300603e-02 -7.47854114e-01 -5.63628316e-01 -1.76763728e-01] | [8.385810852050781, 3.4366824626922607] |
e83d27c6-55b8-4ed6-98be-dcc91cc3cc58 | country-level-arabic-dialect-identification-1 | null | null | https://aclanthology.org/2021.wanlp-1.32 | https://aclanthology.org/2021.wanlp-1.32.pdf | Country-level Arabic Dialect Identification using RNNs with and without Linguistic Features | This work investigates the value of augmenting recurrent neural networks with feature engineering for the Second Nuanced Arabic Dialect Identification (NADI) Subtask 1.2: Country-level DA identification. We compare the performance of a simple word-level LSTM using pretrained embeddings with one enhanced using feature embeddings for engineered linguistic features. Our results show that the addition of explicit features to the LSTM is detrimental to performance. We attribute this performance loss to the bivalency of some linguistic items in some text, ubiquity of topics, and participant mobility. | ['Gus Hahn-Powell', 'Reda Al-Bahrani', 'Mohammed AlShakhori1', 'Elsayed Issa'] | null | null | null | null | eacl-wanlp-2021-4 | ['dialect-identification'] | ['natural-language-processing'] | [-2.31874853e-01 -1.94621123e-02 9.41337347e-02 -4.67058331e-01
-4.46987987e-01 -5.98516524e-01 8.96598458e-01 -6.80155605e-02
-8.19542408e-01 6.13954008e-01 6.94619536e-01 -6.07185721e-01
6.01722812e-03 -5.65167129e-01 -1.70957386e-01 -3.25493544e-01
-3.27284008e-01 4.66919184e-01 -3.41670126e-01 -8.40423584e-01
-1.38966024e-01 7.91941285e-01 -1.14226067e+00 1.36279702e-01
8.95064354e-01 6.39350712e-01 -3.29977274e-01 4.32937950e-01
-1.34145990e-01 8.19362819e-01 -8.19656014e-01 -5.15329897e-01
8.78728181e-02 -2.12763876e-01 -8.70965004e-01 -3.03262293e-01
8.09433103e-01 -7.18613088e-01 -4.11805212e-01 7.20477819e-01
6.43709421e-01 2.49683425e-01 6.59740090e-01 -9.22575295e-01
-9.47788060e-01 1.01307213e+00 -2.24492639e-01 4.70961690e-01
1.38211727e-01 -2.28318885e-01 1.22222579e+00 -1.22596443e+00
8.39424610e-01 1.41245317e+00 1.18019521e+00 3.52406770e-01
-1.27017903e+00 -5.75593710e-01 2.70851403e-01 -3.09777795e-03
-1.41293621e+00 -8.03836346e-01 7.73318410e-01 -3.54217738e-01
1.28414512e+00 -1.18421488e-01 2.88684070e-01 1.20462978e+00
-5.85310087e-02 5.86354196e-01 1.14815259e+00 -6.45597935e-01
-3.14048797e-01 4.87079591e-01 6.65335894e-01 6.74161255e-01
1.63733497e-01 8.44322816e-02 -6.64795041e-01 -1.60223186e-01
4.34558213e-01 -4.36430931e-01 7.00772554e-02 5.24664186e-02
-1.06247020e+00 9.93955791e-01 4.40344810e-01 8.00301790e-01
-3.89891565e-01 -1.04131162e-01 8.63942504e-01 8.06839049e-01
6.62841260e-01 6.51285350e-01 -5.91068089e-01 -2.70951062e-01
-6.39561474e-01 1.08093970e-01 7.28205323e-01 3.89672756e-01
5.69988966e-01 6.51054740e-01 -5.87981157e-02 1.23372436e+00
7.86119848e-02 2.76224852e-01 9.01865065e-01 -6.52617455e-01
2.67351806e-01 4.03397202e-01 4.93094958e-02 -9.60308492e-01
-7.47921586e-01 -5.42446494e-01 -2.99695492e-01 6.52602762e-02
8.23014975e-01 -7.80876100e-01 -7.19792724e-01 2.07177043e+00
-6.68696836e-02 -5.37582040e-01 -6.70384541e-02 6.98824465e-01
6.52042091e-01 4.78894114e-01 2.91444153e-01 1.77530617e-01
1.08250332e+00 -7.73455143e-01 -8.51182163e-01 -2.57268578e-01
1.03662443e+00 -7.41635561e-01 1.22439766e+00 1.79347306e-01
-9.96633053e-01 -4.99723822e-01 -1.13968945e+00 -2.54519552e-01
-9.73409235e-01 3.89618099e-01 5.84725857e-01 1.29011011e+00
-1.42893445e+00 3.81594539e-01 -3.37215990e-01 -6.08028173e-01
1.42797455e-01 5.58232665e-01 -4.67336416e-01 2.94383094e-02
-1.58691406e+00 1.39033365e+00 1.95578918e-01 1.40903845e-01
-2.49097973e-01 -7.34267116e-01 -9.39694583e-01 -4.01969813e-02
-2.08412319e-01 7.21992226e-03 8.41664135e-01 -1.30674112e+00
-1.69929159e+00 8.08515429e-01 2.71395259e-02 -3.63036335e-01
1.38005063e-01 -3.76845032e-01 -7.98716605e-01 -1.88535899e-01
7.16874972e-02 7.75496244e-01 6.41222298e-01 -7.89827466e-01
-5.08935809e-01 -3.98049921e-01 6.81509823e-02 3.37831736e-01
-8.88569593e-01 3.55600923e-01 4.63834703e-01 -1.08518338e+00
-2.67624676e-01 -7.32718825e-01 1.85414538e-01 -5.72528243e-01
-2.77552269e-02 -4.79115635e-01 9.28083301e-01 -1.33422184e+00
1.22004712e+00 -2.20714450e+00 -8.88094082e-02 4.81061071e-01
1.75649092e-01 3.74058038e-01 -4.56624329e-01 3.89077038e-01
-4.84537371e-02 3.33174735e-01 4.40093249e-01 -3.14947158e-01
1.36828184e-01 -2.03496236e-02 -1.53935030e-01 3.47051948e-01
4.38827038e-01 1.07957745e+00 -5.26592433e-01 2.11329371e-01
-2.82659948e-01 6.38998926e-01 -3.61667663e-01 -3.42151970e-01
1.44288898e-01 -8.99652094e-02 3.60726446e-01 7.98785686e-01
4.79151517e-01 3.63989711e-01 2.86383897e-01 3.15461867e-02
-3.97580504e-01 9.36811030e-01 -8.51404727e-01 1.39005268e+00
-4.13738042e-01 9.12499607e-01 2.45675609e-01 -5.81769347e-01
9.27587748e-01 2.23255053e-01 1.29952073e-01 -7.74953008e-01
1.11652680e-01 5.33033073e-01 6.13590539e-01 1.19810745e-01
9.14527714e-01 -1.03933118e-01 -2.35343799e-01 8.68102491e-01
3.57638121e-01 4.50013876e-01 -1.89341322e-01 6.18953928e-02
7.27956593e-01 -8.80070925e-02 -1.10066622e-01 -8.24329734e-01
3.68629515e-01 -2.89250854e-02 3.84099305e-01 7.72802830e-01
-3.88151854e-01 2.21406758e-01 3.22722882e-01 -4.76659149e-01
-9.84825194e-01 -8.89447510e-01 -1.96123108e-01 1.87714267e+00
-9.04419065e-01 -4.81505603e-01 -5.39716423e-01 -6.65297270e-01
-7.60588869e-02 6.92102075e-01 -7.47708023e-01 -1.92740098e-01
-1.01206434e+00 -7.84837186e-01 1.23101938e+00 6.49911642e-01
3.76093507e-01 -7.57903516e-01 -5.02145410e-01 4.68950868e-01
4.54050787e-02 -7.86266446e-01 -6.78257644e-01 2.58643091e-01
-5.71608484e-01 -2.70204961e-01 -8.57042015e-01 -1.04712474e+00
2.70077139e-02 -1.26263767e-01 1.10731137e+00 3.62513326e-02
1.32452279e-01 1.93075538e-01 -3.74291480e-01 -3.90978754e-01
-4.46683705e-01 8.22345912e-01 4.16384190e-01 -1.50108710e-01
7.10867703e-01 -2.69172400e-01 1.30892873e-01 2.65390892e-03
-4.39937890e-01 -5.54087281e-01 1.46876439e-01 1.17355764e+00
-4.00331557e-01 -2.69058019e-01 1.10930824e+00 -8.28682005e-01
1.07606065e+00 -2.62042075e-01 8.57760571e-03 1.46074340e-01
-3.77646029e-01 -6.81729913e-02 2.83484071e-01 -6.89189076e-01
-1.01337039e+00 -2.73515731e-01 -3.42731297e-01 2.27993399e-01
-3.48333977e-02 8.45011353e-01 -8.79147276e-02 -1.52620807e-01
6.88838661e-01 -1.62154377e-01 4.05299157e-01 -3.68146539e-01
4.45730448e-01 7.55466580e-01 2.00485334e-01 -5.87065041e-01
2.75417745e-01 -1.71815187e-01 -6.59993947e-01 -1.10643637e+00
-2.92885780e-01 1.25080690e-01 -8.61549199e-01 -1.13498822e-01
5.41346669e-01 -1.01530230e+00 -3.90659928e-01 8.66722226e-01
-9.51221287e-01 -6.45636380e-01 -1.75328955e-01 4.50695246e-01
-9.28528309e-02 -7.66351894e-02 -1.01107371e+00 -5.43733776e-01
-2.71275431e-01 -9.73034084e-01 3.64819616e-01 -1.75449952e-01
-6.37666404e-01 -1.33516657e+00 1.02678634e-01 -8.62644613e-02
9.92405355e-01 6.22117221e-02 1.38426435e+00 -1.13694751e+00
2.79053241e-01 -1.23459347e-01 -1.28906250e-01 3.39087456e-01
3.76254588e-01 7.52314413e-03 -1.36030352e+00 -2.63474107e-01
-1.67288318e-01 -5.07935703e-01 7.24459112e-01 3.70623440e-01
2.33153492e-01 -3.41335952e-01 1.54514387e-01 2.79576987e-01
9.84075785e-01 1.28391728e-01 2.73185283e-01 4.86136407e-01
7.55766988e-01 8.92104626e-01 8.55516940e-02 1.15901455e-01
7.42423296e-01 6.89069152e-01 -3.70734364e-01 -1.63830921e-01
-2.56176680e-01 -1.20120704e-01 9.41307127e-01 1.02923763e+00
1.55311584e-01 1.47613049e-01 -1.31096375e+00 8.80011082e-01
-1.32373726e+00 -6.57464802e-01 1.01092264e-01 1.86601567e+00
7.76045620e-01 5.81767829e-03 8.15910399e-01 2.25693211e-01
6.07105732e-01 1.53268188e-01 -2.23943800e-01 -1.17352700e+00
-5.43619871e-01 2.13692456e-01 3.97650748e-01 8.12484622e-01
-8.54715824e-01 1.37471604e+00 7.18954229e+00 6.63088560e-01
-1.37897539e+00 4.03988421e-01 6.71040058e-01 -4.51602265e-02
-2.86071628e-01 -6.52868211e-01 -9.34169292e-01 2.44777009e-01
1.52066672e+00 -2.09410656e-02 4.63645160e-01 3.38431239e-01
-4.90577109e-02 1.71085387e-01 -1.05760694e+00 2.95904517e-01
2.46039346e-01 -1.11997235e+00 1.84809014e-01 2.37178326e-01
3.26381385e-01 2.93633252e-01 5.56229234e-01 5.05491674e-01
4.95100915e-01 -1.39079595e+00 8.18999410e-01 1.58460945e-01
8.96949708e-01 -1.05879843e+00 8.67114246e-01 -7.99009278e-02
-7.06869066e-01 -1.68736756e-01 -8.98037776e-02 -3.71748447e-01
5.54245757e-03 -6.06167726e-02 -9.55487490e-01 9.08416063e-02
5.52905619e-01 6.53793216e-01 -8.42153847e-01 4.12067920e-01
1.61861151e-01 8.51214230e-01 -5.49540162e-01 -2.24794429e-02
6.85571313e-01 -1.32642493e-01 4.50334519e-01 1.41270363e+00
8.16830844e-02 -3.68345886e-01 -2.47744635e-01 4.62338001e-01
-1.41236916e-01 3.11644554e-01 -9.15664077e-01 -2.80716121e-01
6.15719020e-01 8.72884810e-01 -3.74132991e-01 -2.17909351e-01
-4.52302754e-01 9.53275442e-01 7.16057360e-01 3.81089330e-01
-2.54950970e-01 -6.17902994e-01 1.00478017e+00 -7.86512420e-02
3.13080043e-01 -4.61493969e-01 -6.72267556e-01 -9.17226851e-01
-1.85976416e-01 -1.10212612e+00 5.44646263e-01 -3.25055659e-01
-1.37456143e+00 7.48895645e-01 -4.06325608e-01 -3.87540489e-01
-6.01577222e-01 -8.59938979e-01 -5.52631021e-01 1.25952816e+00
-1.21149194e+00 -1.55777192e+00 3.75591844e-01 5.51242352e-01
2.57335931e-01 -6.10028028e-01 1.09201622e+00 4.24930483e-01
-6.36131763e-01 1.22919643e+00 1.08751701e-02 4.35200453e-01
9.65862751e-01 -1.27106082e+00 7.31987059e-01 6.73905790e-01
1.63705841e-01 9.20826733e-01 3.28562498e-01 -4.11338449e-01
-8.73757124e-01 -8.37191641e-01 1.53974068e+00 -8.40160608e-01
9.00662720e-01 -7.82977879e-01 -8.86889815e-01 1.04934418e+00
5.47925949e-01 -5.73939800e-01 8.84706557e-01 8.38645399e-01
-4.99849975e-01 3.21683101e-02 -1.21476030e+00 8.24512959e-01
9.57771003e-01 -1.10473728e+00 -6.04650259e-01 -1.27331942e-01
8.40803087e-01 5.70694655e-02 -1.16460645e+00 1.51461542e-01
8.41094971e-01 -5.12023270e-01 9.80234623e-01 -7.08001733e-01
4.95887846e-02 4.47411090e-01 -1.68596655e-01 -1.62964809e+00
-4.85880703e-01 -4.32181299e-01 3.46725464e-01 1.52200639e+00
7.56632328e-01 -9.56622481e-01 4.32936937e-01 7.80207932e-01
-7.28732944e-02 -2.47448877e-01 -9.58364308e-01 -6.65695488e-01
9.23496127e-01 -1.47376105e-01 6.29074633e-01 1.49808633e+00
2.06941247e-01 6.01637602e-01 -2.74509460e-01 -4.04386111e-02
-7.34640658e-02 -5.76539695e-01 3.31267238e-01 -1.18917716e+00
1.73265114e-01 -7.61123359e-01 -3.74256283e-01 -4.05302644e-01
5.33977270e-01 -9.24422503e-01 -4.87562269e-01 -9.66536522e-01
-6.53670847e-01 -6.59252584e-01 -3.40809852e-01 5.07688403e-01
-4.26853895e-02 2.88444102e-01 4.41021115e-01 -4.45500948e-02
-4.86183204e-02 3.46522540e-01 4.06057030e-01 -2.01385133e-02
-6.65347755e-01 -3.63443226e-01 -7.92638302e-01 5.08474767e-01
1.14722502e+00 -1.52998015e-01 -5.88027649e-02 -9.19472456e-01
1.95622772e-01 -5.87799609e-01 7.20243976e-02 -7.96087742e-01
7.65569694e-03 4.17945057e-01 3.47240955e-01 -4.23495263e-01
4.72844690e-01 -4.66821343e-01 -3.54083568e-01 3.75386000e-01
-3.82557482e-01 6.97380066e-01 7.12873816e-01 -2.61870354e-01
-2.16836333e-01 -1.08519875e-01 4.74971324e-01 5.03335446e-02
-6.42714381e-01 -2.73784816e-01 -1.11519408e+00 3.61881182e-02
2.23788545e-01 -3.12223583e-01 -1.91881448e-01 -4.25134182e-01
-6.69402957e-01 -1.60703838e-01 3.74102443e-01 5.23455739e-01
1.37514800e-01 -1.42220235e+00 -8.99746656e-01 5.32070100e-01
-1.86356902e-01 -8.52934062e-01 -1.26971677e-01 7.44149387e-01
-4.11207646e-01 4.91286963e-01 -7.19636321e-01 -2.18415949e-02
-1.15636289e+00 -1.59651160e-01 4.50113386e-01 -1.38444975e-01
-2.92366624e-01 1.00358796e+00 -3.33467066e-01 -1.06192911e+00
2.75374889e-01 -2.60391142e-02 -3.74859303e-01 6.90543711e-01
4.70521212e-01 5.39853752e-01 2.50350446e-01 -8.94223571e-01
-5.28958261e-01 -8.29779208e-02 -7.25796998e-01 -5.68813384e-01
1.36665308e+00 -1.76912948e-01 -1.52456611e-01 7.56804347e-01
1.01970398e+00 3.89710188e-01 -7.07278609e-01 -4.53292191e-01
4.41675246e-01 5.75429350e-02 2.05732375e-01 -1.25397301e+00
-6.86521411e-01 9.39982474e-01 7.47571945e-01 4.36146781e-02
5.16312480e-01 -4.62837785e-01 8.12155783e-01 4.60249513e-01
3.55166085e-02 -1.44149196e+00 -3.02084625e-01 1.27543211e+00
7.09619701e-01 -9.08471167e-01 -3.74433190e-01 2.43705273e-01
-7.86686361e-01 1.10448694e+00 5.58082104e-01 -2.49359339e-01
7.86258936e-01 1.74737677e-01 4.56086725e-01 -1.37347177e-01
-4.00677562e-01 -1.02901883e-01 4.52962033e-02 7.13037193e-01
6.92879200e-01 1.61687315e-01 -3.66591334e-01 6.88498616e-01
-5.99292874e-01 -4.49547589e-01 6.03424370e-01 6.79793596e-01
-8.61912593e-02 -1.05579793e+00 -2.49456525e-01 5.38084447e-01
-5.23164511e-01 -5.03247023e-01 -7.69362926e-01 1.13828778e+00
5.54410294e-02 7.48416603e-01 5.98385692e-01 -4.66686547e-01
2.44846977e-02 7.46085942e-01 9.57945287e-02 -4.64939803e-01
-1.32756472e+00 -2.87463665e-01 6.93782687e-01 -4.07364517e-02
-1.76028535e-01 -9.99031961e-01 -7.65465498e-01 -5.44590592e-01
-2.11965919e-01 -1.92566663e-01 5.90714097e-01 9.32592988e-01
4.70010430e-01 1.46994993e-01 3.09214234e-01 -6.36518836e-01
-5.99327147e-01 -1.28693271e+00 -6.48658633e-01 4.73139212e-02
4.87045437e-01 -5.79924583e-01 -3.60555947e-02 -3.24773073e-01] | [10.27408218383789, 10.563895225524902] |
48d4e356-dd9f-46de-a212-4440725d55c1 | linguistically-informed-relation-extraction | 1910.03385 | null | https://arxiv.org/abs/1910.03385v1 | https://arxiv.org/pdf/1910.03385v1.pdf | Linguistically Informed Relation Extraction and Neural Architectures for Nested Named Entity Recognition in BioNLP-OST 2019 | Named Entity Recognition (NER) and Relation Extraction (RE) are essential tools in distilling knowledge from biomedical literature. This paper presents our findings from participating in BioNLP Shared Tasks 2019. We addressed Named Entity Recognition including nested entities extraction, Entity Normalization and Relation Extraction. Our proposed approach of Named Entities can be generalized to different languages and we have shown it's effectiveness for English and Spanish text. We investigated linguistic features, hybrid loss including ranking and Conditional Random Fields (CRF), multi-task objective and token-level ensembling strategy to improve NER. We employed dictionary based fuzzy and semantic search to perform Entity Normalization. Finally, our RE system employed Support Vector Machine (SVM) with linguistic features. Our NER submission (team:MIC-CIS) ranked first in BB-2019 norm+NER task with standard error rate (SER) of 0.7159 and showed competitive performance on PharmaCo NER task with F1-score of 0.8662. Our RE system ranked first in the SeeDev-binary Relation Extraction Task with F1-score of 0.3738. | ['Hinrich Schütze', 'Usama Yaseen', 'Pankaj Gupta'] | 2019-10-08 | linguistically-informed-relation-extraction-1 | https://aclanthology.org/D19-5720 | https://aclanthology.org/D19-5720.pdf | ws-2019-11 | ['binary-relation-extraction', 'nested-named-entity-recognition'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.01300649e-01 3.88240844e-01 -1.21654212e-01 -4.22208756e-01
-1.03288209e+00 -6.02218688e-01 3.98873210e-01 1.01118350e+00
-1.11845052e+00 1.61613667e+00 3.34584981e-01 -1.99360490e-01
-1.46216139e-01 -6.47784948e-01 -5.23654759e-01 -3.12121987e-01
-2.50736058e-01 6.61041975e-01 4.02445421e-02 -1.72121122e-01
2.47810096e-01 8.18759203e-01 -6.57701194e-01 5.03287971e-01
1.01631713e+00 5.39239526e-01 7.37505313e-03 7.44769812e-01
-1.24599695e-01 9.45191383e-01 -7.60872304e-01 -7.55573690e-01
-3.11506301e-01 -1.29077882e-01 -1.35286629e+00 -9.74544644e-01
-2.01454848e-01 4.98095185e-01 7.88888261e-02 9.89317060e-01
1.08553767e+00 2.66804844e-01 8.81967366e-01 -6.54538631e-01
-4.74750131e-01 8.15209746e-01 -4.19443429e-01 4.46257412e-01
4.82088655e-01 -4.79932696e-01 6.72533214e-01 -1.00955033e+00
1.18740380e+00 6.77229345e-01 1.00230563e+00 5.72252333e-01
-1.07370293e+00 -7.99360275e-01 -5.48592567e-01 4.83333468e-02
-1.65578485e+00 -5.42127073e-01 -1.82476297e-01 -3.66229832e-01
1.74229753e+00 1.06946558e-01 -3.60021298e-03 7.73906052e-01
4.52777058e-01 2.63152778e-01 1.09712648e+00 -5.13116419e-01
2.14803547e-01 4.22740221e-01 3.11680496e-01 7.27177024e-01
4.05761451e-01 -5.30120060e-02 -5.31554461e-01 -3.93755049e-01
3.12096953e-01 -6.39875531e-01 1.20145023e-01 5.99868000e-01
-1.33926141e+00 6.25926733e-01 1.81328580e-01 9.44497108e-01
-7.00005591e-01 -2.54185528e-01 6.26712978e-01 1.13657117e-01
1.83089241e-01 6.97812855e-01 -1.12404168e+00 3.46371159e-02
-9.61928606e-01 -1.88644454e-01 1.13638461e+00 1.02289057e+00
1.93616435e-01 -2.94317305e-01 -5.49522996e-01 1.04768884e+00
-2.70956159e-02 2.19216704e-01 7.01396286e-01 -3.38225186e-01
2.35817298e-01 3.37047666e-01 -2.57060714e-02 -7.47830749e-01
-1.01599693e+00 -2.41770700e-01 -1.13891852e+00 -5.10637224e-01
1.98178545e-01 -6.14961088e-01 -1.02624094e+00 1.76421595e+00
3.48141342e-01 6.52102679e-02 6.67783320e-01 3.48801494e-01
1.35504615e+00 2.24902362e-01 9.48586941e-01 -3.46266896e-01
1.82854342e+00 -5.57599306e-01 -9.88656580e-01 3.63617122e-01
8.71028543e-01 -1.12666249e+00 -8.83077830e-02 2.81076193e-01
-9.66018975e-01 -2.79602021e-01 -7.33951628e-01 -9.86470059e-02
-1.04968762e+00 4.23972189e-01 7.41579890e-01 8.32596183e-01
-8.85971367e-01 6.11204982e-01 -7.85005987e-01 -7.07629800e-01
4.64581966e-01 9.49253917e-01 -9.89752293e-01 4.18390363e-01
-1.56995404e+00 1.43491983e+00 9.66079891e-01 -9.26486328e-02
-4.97820199e-01 -8.10831964e-01 -7.56642222e-01 -2.59493142e-01
5.15816920e-02 -6.84508741e-01 7.34449387e-01 8.60813931e-02
-1.08115542e+00 1.15899122e+00 -2.39593089e-01 -7.91585565e-01
3.11555080e-02 -1.27084136e-01 -1.03218985e+00 7.75649771e-02
4.02623773e-01 7.31459737e-01 -4.04773414e-01 -5.56137919e-01
-6.78544700e-01 -3.88959140e-01 -5.44116735e-01 4.85193618e-02
4.64582369e-02 4.52591628e-01 1.59571782e-01 -6.91979766e-01
-1.70858234e-01 -5.21716118e-01 -5.25451481e-01 -8.48567903e-01
-6.24306321e-01 -4.41654801e-01 -1.68377489e-01 -9.92425084e-01
1.28509140e+00 -1.57857084e+00 -2.97053933e-01 2.79599935e-01
3.09859306e-01 3.00638944e-01 1.77203774e-01 4.65356410e-01
-5.94706655e-01 5.61671853e-01 -7.07247704e-02 2.48309910e-01
-2.36983597e-01 -7.99089447e-02 3.12552601e-01 3.52470607e-01
5.56020677e-01 9.94606793e-01 -7.96069562e-01 -8.96825850e-01
-1.45094216e-01 7.06140339e-01 -3.17415386e-01 3.52600999e-02
2.69007653e-01 3.63249213e-01 -5.16799986e-01 5.93729079e-01
5.64362705e-01 -1.11872375e-01 1.66347027e-01 -6.16552353e-01
-3.34731698e-01 2.31158316e-01 -1.27826488e+00 1.68123841e+00
-2.98908889e-01 7.37825632e-02 -2.77059704e-01 -9.33123767e-01
1.18816924e+00 6.30113959e-01 6.39816165e-01 -4.52123344e-01
2.22317398e-01 2.73140132e-01 -1.05976202e-01 -6.50524080e-01
3.96545827e-01 -4.47273076e-01 -1.57528743e-01 -2.26947412e-01
6.80084109e-01 3.98718625e-01 2.76874334e-01 1.43195331e-01
1.27156222e+00 1.39662996e-01 1.18773711e+00 -5.06183863e-01
8.29087973e-01 2.25658789e-01 7.14854479e-01 6.43339157e-01
-3.45457673e-01 2.32519954e-01 3.21852744e-01 -1.72768086e-01
-8.20449769e-01 -6.58438385e-01 -8.41907382e-01 8.88518214e-01
-4.51318145e-01 -4.45853710e-01 -5.57167113e-01 -7.50321150e-01
-3.37991804e-01 7.78576732e-01 -5.57725251e-01 1.02271013e-01
-6.08158648e-01 -1.37060213e+00 1.33543909e+00 2.01145858e-01
3.81274343e-01 -1.15531456e+00 -8.93649012e-02 5.88445842e-01
-1.78631604e-01 -1.31221330e+00 -3.03669959e-01 9.94912028e-01
-6.56981468e-01 -1.09383404e+00 -9.68480289e-01 -1.07125914e+00
3.50517958e-01 -1.01628339e+00 1.01961553e+00 -6.67963386e-01
-6.25826716e-01 -1.11970231e-01 -3.79912168e-01 -7.12251484e-01
-2.48625889e-01 3.22862089e-01 9.03593078e-02 -5.29784441e-01
8.38669658e-01 -3.39670151e-01 -4.87812549e-01 -2.05739159e-02
-6.14931464e-01 -3.48067611e-01 8.81055355e-01 9.95669663e-01
8.39306653e-01 -1.16783425e-01 1.10329783e+00 -1.19885898e+00
5.55099666e-01 -6.74947262e-01 -9.51280594e-02 5.93983650e-01
-7.59491920e-01 3.15197855e-01 3.79044026e-01 -3.16680938e-01
-1.08311999e+00 3.27757210e-01 -7.05364823e-01 5.88652313e-01
-6.15116298e-01 5.50383627e-01 -1.65492982e-01 1.03389360e-01
9.08209741e-01 9.24280193e-03 -7.31890440e-01 -5.46308041e-01
3.73612434e-01 9.77326393e-01 6.41949236e-01 -4.61065441e-01
3.39524820e-02 -8.15226436e-02 2.39730865e-01 -8.22975695e-01
-7.60497987e-01 -6.81130469e-01 -7.69644499e-01 6.25835359e-01
1.48133671e+00 -1.08378434e+00 -1.01152575e+00 1.77763343e-01
-1.27661383e+00 2.24624768e-01 -1.14553139e-01 8.94444108e-01
-1.28787473e-01 1.49963647e-01 -7.92139411e-01 -6.38915300e-01
-1.01144373e+00 -6.44935429e-01 7.47874081e-01 5.41609406e-01
-4.14160401e-01 -9.53212082e-01 4.48707849e-01 2.96413124e-01
1.75620958e-01 5.44527531e-01 8.61055195e-01 -1.70204151e+00
4.53307271e-01 -1.32982790e-01 -2.72464961e-01 -1.02607563e-01
1.33460417e-01 -5.22699416e-01 -8.72040272e-01 2.82834351e-01
-3.42042804e-01 -1.21045038e-01 7.82166600e-01 2.45101586e-01
6.00903869e-01 -2.25024536e-01 -6.47056460e-01 3.88936192e-01
1.47905183e+00 5.67505360e-01 6.89695597e-01 3.54794621e-01
4.48737442e-01 4.23401624e-01 5.40478647e-01 2.63641179e-01
3.89645815e-01 3.36104423e-01 -5.52111387e-01 -1.71373725e-01
-5.31529710e-02 -7.43789002e-02 -2.96326131e-02 5.57500064e-01
-1.83349758e-01 -1.69734478e-01 -1.16734457e+00 6.60804033e-01
-1.38064063e+00 -7.70488679e-01 -3.69960487e-01 1.82088792e+00
1.43266368e+00 1.61343871e-03 7.67827705e-02 -2.94999331e-01
9.43021417e-01 -6.42685235e-01 -1.99123040e-01 -6.56095743e-01
-6.20243430e-01 1.06996143e+00 7.76977599e-01 3.70832264e-01
-1.35190976e+00 1.05074906e+00 5.11688232e+00 1.06237137e+00
-6.15967095e-01 3.23158324e-01 5.55858195e-01 5.82824528e-01
3.77752960e-01 -1.57809079e-01 -1.31976676e+00 1.83762625e-01
1.54561484e+00 -2.87169605e-01 -2.26905018e-01 3.93434882e-01
-8.93845633e-02 -1.38771296e-01 -8.23653281e-01 7.98665941e-01
-6.32327050e-02 -1.35213542e+00 -2.26502836e-01 -3.56261581e-02
6.81998730e-01 4.24033582e-01 -6.62383914e-01 3.38487983e-01
5.09504080e-01 -1.41525733e+00 -5.09132352e-03 1.01218700e+00
8.70037377e-01 -7.88809836e-01 1.30439413e+00 5.34305237e-02
-1.19002938e+00 3.78450692e-01 -3.17088574e-01 5.61341763e-01
2.30873302e-01 9.05193269e-01 -1.15031719e+00 1.04744518e+00
4.99932289e-01 3.62107962e-01 -3.02058101e-01 1.10257208e+00
3.22485082e-02 3.26733172e-01 -5.36900699e-01 -2.01609939e-01
-1.94890946e-01 2.93728560e-01 4.17709976e-01 1.90146744e+00
4.55728322e-02 5.36253035e-01 1.19729955e-02 3.83775651e-01
-3.06009501e-01 9.63074684e-01 -2.20752850e-01 -1.47508606e-01
4.26709086e-01 1.37738228e+00 -9.41624522e-01 -4.64674652e-01
-1.11163810e-01 8.10479105e-01 3.59738767e-01 -1.08848557e-01
-6.82085752e-01 -1.03717852e+00 1.05626881e-01 -3.26558352e-01
3.30559462e-01 2.46766955e-01 -3.35041851e-01 -1.00580263e+00
-4.77873415e-01 -6.16703212e-01 8.48185062e-01 -4.06193227e-01
-1.43825471e+00 1.10524845e+00 -1.44609228e-01 -6.22477889e-01
1.94824114e-01 -5.72418392e-01 2.83427779e-02 1.09702694e+00
-1.21339750e+00 -1.20555675e+00 3.84598196e-01 3.39670390e-01
-9.15472209e-02 -3.69830012e-01 1.37503898e+00 9.30298626e-01
-7.16177881e-01 9.61460233e-01 8.47628191e-02 4.42563444e-01
1.14837015e+00 -1.30567217e+00 -1.67022850e-02 2.16619357e-01
4.58599553e-02 1.02937233e+00 3.01712483e-01 -1.02225745e+00
-5.99966586e-01 -1.18616188e+00 2.01580906e+00 -4.60669607e-01
5.60030341e-01 4.93480563e-02 -6.19045079e-01 3.89282912e-01
7.89316371e-02 -9.84032229e-02 1.27785456e+00 2.55115069e-02
-1.27255201e-01 2.26846263e-01 -1.74125433e+00 3.48684788e-02
6.92006290e-01 -3.22329044e-01 -7.74727881e-01 6.64579570e-01
5.09966791e-01 -4.70254928e-01 -1.90240669e+00 5.44428289e-01
4.86788869e-01 -1.96177855e-01 1.07855344e+00 -1.20089376e+00
9.06189755e-02 -3.87891203e-01 -2.53230512e-01 -7.46760666e-01
-2.53777206e-01 -4.98174876e-01 2.81340301e-01 1.75091743e+00
1.02914464e+00 -5.23911536e-01 4.35749322e-01 4.53581393e-01
5.98397627e-02 -7.35438287e-01 -9.08519685e-01 -4.34141755e-01
2.92443067e-01 -6.12085536e-02 4.40121889e-01 1.24882960e+00
1.90754414e-01 7.88161814e-01 -6.21844344e-02 7.70722926e-02
2.56731272e-01 -4.86893773e-01 -7.00785220e-02 -1.17752600e+00
2.95002721e-02 -8.39603245e-02 -4.82447386e-01 -1.99870080e-01
1.36001170e-01 -1.31551373e+00 -1.43934608e-01 -1.54484987e+00
5.50501108e-01 -4.05551046e-01 -7.56928802e-01 9.44349408e-01
-1.68139830e-01 2.23359868e-01 -2.90536284e-01 -1.53036609e-01
-6.43782556e-01 -8.63646269e-02 6.48804665e-01 1.98816866e-01
-2.60444164e-01 -2.71401554e-01 -8.64223421e-01 3.88190955e-01
8.78616273e-01 -1.13203371e+00 3.73189211e-01 2.62834609e-01
3.92712295e-01 1.65332586e-01 -1.76482782e-01 -6.67679250e-01
4.76444721e-01 4.78545353e-02 8.14760029e-01 -6.23554468e-01
-2.78539747e-01 -4.34118450e-01 4.50874388e-01 6.99367464e-01
-5.67361116e-01 -2.51229554e-02 3.03440690e-01 2.43772596e-01
-4.13554087e-02 -4.56821352e-01 7.16835380e-01 -1.83286816e-01
-4.32366163e-01 4.49863309e-03 -3.49410415e-01 3.24982852e-01
9.91427302e-01 8.56938586e-02 -2.54668415e-01 5.42836845e-01
-1.54264188e+00 2.15458095e-01 -2.63141155e-01 -2.61759199e-02
1.67611554e-01 -8.93253505e-01 -9.45889175e-01 -3.21494520e-01
8.22250769e-02 -4.02662545e-01 1.28583536e-01 9.98507738e-01
-6.59954429e-01 8.91506433e-01 -2.67149031e-01 3.57196070e-02
-1.43873990e+00 4.79311973e-01 2.67319858e-01 -8.97107005e-01
-1.46082282e-01 1.07274663e+00 -3.10724080e-01 -7.49337196e-01
4.54182588e-02 -9.16567594e-02 -8.91232312e-01 3.71301740e-01
4.82567519e-01 5.07376611e-01 4.81046736e-01 -8.00443411e-01
-1.05412638e+00 4.59728450e-01 -1.89924285e-01 -3.37696495e-03
1.51626289e+00 2.27309927e-01 -3.10262889e-01 1.04570739e-01
1.32033479e+00 2.38163307e-01 3.00245464e-01 -1.70513103e-03
7.30286539e-01 7.08587646e-01 -7.12376311e-02 -1.50421858e+00
-5.03847897e-01 3.06817412e-01 8.44940722e-01 -4.24411595e-01
8.81315708e-01 5.78668490e-02 6.23879910e-01 6.42326295e-01
2.61591882e-01 -1.02210629e+00 -7.72092223e-01 6.78222358e-01
3.85836512e-01 -1.22875583e+00 1.80918664e-01 -4.46803749e-01
-6.80868745e-01 1.01796031e+00 1.98577657e-01 2.09447816e-01
8.52838874e-01 6.81977510e-01 -1.79927036e-01 -3.33037645e-01
-4.99874622e-01 -3.24926674e-01 6.01717830e-01 7.72622645e-01
1.10080087e+00 4.97577414e-02 -1.13036299e+00 1.37078464e+00
-2.68369973e-01 4.10520196e-01 3.04514673e-02 8.26102793e-01
-9.32019651e-02 -1.31529164e+00 -3.60717275e-03 6.57489181e-01
-1.51336813e+00 -6.32669389e-01 -3.48782927e-01 5.56950331e-01
4.74778831e-01 9.73994792e-01 -5.13308287e-01 -1.30689979e-01
5.16850829e-01 3.18949789e-01 3.90283853e-01 -6.32418096e-01
-1.18920183e+00 4.97076772e-02 6.15088642e-01 -3.21696579e-01
-6.89479887e-01 -5.67914724e-01 -1.77761149e+00 1.35308638e-01
-6.61424577e-01 8.31398070e-01 7.27548540e-01 8.57708514e-01
5.32714009e-01 6.34824693e-01 1.12425521e-01 2.29892313e-01
-1.42211094e-03 -1.09667194e+00 -5.02121925e-01 6.53597265e-02
-6.33558184e-02 -3.36662084e-01 1.79518864e-01 4.13389355e-01] | [8.490625381469727, 8.757216453552246] |
347be2dc-4dda-4623-9420-c849f60a8f96 | multi-platform-version-of-starcraft-brood-war | 1801.02193 | null | http://arxiv.org/abs/1801.02193v1 | http://arxiv.org/pdf/1801.02193v1.pdf | Multi-platform Version of StarCraft: Brood War in a Docker Container: Technical Report | We present a dockerized version of a real-time strategy game StarCraft: Brood
War, commonly used as a domain for AI research, with a pre-installed collection
of AI developement tools supporting all the major types of StarCraft bots. This
provides a convenient way to deploy StarCraft AIs on numerous hosts at once and
across multiple platforms despite limited OS support of StarCraft. In this
technical report, we describe the design of our Docker images and present a few
use cases. | ['Michal Čertický', 'Jan Malý', 'Michal Šustr'] | 2018-01-07 | null | null | null | null | ['real-time-strategy-games'] | ['playing-games'] | [-6.79742396e-01 -4.05134916e-01 -3.02987665e-01 3.13444108e-01
3.10771555e-01 -1.31337512e+00 7.11341083e-01 -3.26499045e-01
-6.54021740e-01 7.51966000e-01 -2.94107914e-01 -6.15248263e-01
-5.56465127e-02 -5.47144771e-01 -1.30040616e-01 -2.51235992e-01
-4.28773403e-01 1.00312555e+00 1.03458452e+00 -1.23108065e+00
2.77856916e-01 8.63667130e-01 -1.46997023e+00 -2.22448468e-01
3.72159332e-01 6.56695068e-01 -1.64595261e-01 1.04367959e+00
3.16157013e-01 1.04345345e+00 -1.25928557e+00 -5.92320085e-01
8.40231717e-01 1.23253845e-01 -6.66056097e-01 -3.95836562e-01
2.08219305e-01 -5.41626513e-01 -4.90094960e-01 8.36314082e-01
6.37852669e-01 1.63750857e-01 -7.41427988e-02 -2.25891590e+00
3.41201514e-01 4.72519606e-01 -2.83648461e-01 8.12318563e-01
5.92934012e-01 8.64261687e-01 7.47999609e-01 7.33509474e-03
9.67654943e-01 7.12984324e-01 5.97077608e-01 4.75099236e-01
-7.46618986e-01 -1.05365169e+00 -3.07870597e-01 3.07752509e-02
-1.37171102e+00 -5.88935196e-01 3.99377942e-01 -2.81898230e-01
1.33337188e+00 1.50855184e-01 1.02661967e+00 1.37457430e+00
6.38468444e-01 4.57373232e-01 7.36887038e-01 1.86484933e-01
4.29366767e-01 -5.35364270e-01 9.41039249e-02 6.95416331e-01
6.89910889e-01 1.69361502e-01 -6.72487140e-01 -8.45384181e-01
1.15368843e+00 -2.23913133e-01 1.09093688e-01 -2.14701593e-01
-1.29789793e+00 5.51055968e-01 2.45325014e-01 4.08413649e-01
-7.36275494e-01 5.48841000e-01 6.43199623e-01 2.36160502e-01
-1.27495471e-02 7.27714777e-01 -3.03308964e-01 -1.12844682e+00
-6.76194727e-01 9.60126162e-01 1.24832308e+00 9.71710026e-01
3.34662259e-01 6.22358918e-01 3.95941257e-01 -5.29986359e-02
8.19051564e-02 1.66755959e-01 3.81490886e-01 -1.24099886e+00
-5.46562113e-02 3.46261591e-01 2.91407108e-01 -8.42005134e-01
-4.47450370e-01 -3.05984080e-01 2.55097806e-01 9.21844482e-01
2.91558266e-01 -7.24096537e-01 -6.81859493e-01 1.04254568e+00
6.37093484e-01 4.30745602e-01 1.72810569e-01 1.24765670e+00
8.19756687e-01 2.62848586e-01 -1.00629874e-01 3.20968539e-01
1.41079891e+00 -9.11727905e-01 -5.57428181e-01 -3.03536147e-01
2.76999235e-01 -6.08735800e-01 7.91263640e-01 5.16621768e-01
-1.21750927e+00 3.40221286e-01 -1.16182840e+00 4.81862992e-01
-4.99020457e-01 -7.03919053e-01 1.08585072e+00 7.29431748e-01
-1.28695953e+00 5.08664787e-01 -1.16929996e+00 -6.68642700e-01
1.06304489e-01 4.25257742e-01 -3.02869081e-01 2.98768699e-01
-7.14778423e-01 1.15230131e+00 5.52317441e-01 -7.82748580e-01
-1.47062051e+00 -3.50039303e-01 -3.42219621e-01 -1.18324794e-01
7.32729673e-01 -3.42768937e-01 1.54899204e+00 -4.19374764e-01
-1.61731553e+00 6.75045371e-01 1.01847994e+00 -5.27740777e-01
1.93231314e-01 -5.68090193e-02 -5.26486933e-01 1.05144672e-01
3.84135813e-01 1.88142180e-01 5.35344839e-01 -6.85220480e-01
-8.83257568e-01 -1.70273021e-01 8.23166490e-01 4.59775656e-01
1.03848830e-01 7.30795324e-01 -2.24082351e-01 -4.50007945e-01
-9.10325766e-01 -8.94937932e-01 -3.77726376e-01 -3.54466259e-01
1.10833727e-01 -9.88994315e-02 1.09081137e+00 -7.30946139e-02
7.78327644e-01 -2.09102011e+00 2.10387215e-01 -2.77798641e-02
6.30255163e-01 4.95361716e-01 -1.30140614e-02 8.22314262e-01
3.40561152e-01 -1.65558755e-01 3.76685947e-01 4.93229508e-01
2.53112257e-01 4.62162495e-01 -1.36984453e-01 4.98390049e-01
-4.32659388e-01 5.10081470e-01 -1.14838386e+00 -3.80011469e-01
1.75189544e-02 -2.85391533e-03 -5.64234614e-01 1.98239684e-01
-4.59514707e-01 7.56797791e-02 -4.98041421e-01 9.36241686e-01
2.78139740e-01 9.49811563e-02 3.36687207e-01 4.74439055e-01
-7.21060157e-01 2.50705600e-01 -9.61435258e-01 1.90339994e+00
1.62907362e-01 4.55911934e-01 8.53801787e-01 -3.97985131e-01
4.31784898e-01 4.99155372e-01 6.14986718e-01 -4.81579602e-02
6.53476417e-01 1.82421193e-01 2.96143174e-01 -2.46760733e-02
7.30137110e-01 1.35388225e-01 -3.86493266e-01 7.21951067e-01
4.41465795e-01 -3.70491326e-01 7.14772224e-01 4.36084002e-01
1.79823637e+00 6.11731075e-02 5.27573466e-01 -2.15009376e-01
-3.62170130e-01 1.24490798e+00 5.61039031e-01 5.93787193e-01
-7.86473274e-01 -4.15612578e-01 4.53537703e-01 -8.49189281e-01
-6.75435901e-01 -9.00158226e-01 3.93946767e-01 1.47803032e+00
3.24149847e-01 -1.19922388e+00 -7.48281002e-01 -7.88806081e-01
-1.23383522e-01 9.03960407e-01 -1.14681952e-01 1.94386482e-01
-4.43897784e-01 -2.96832383e-01 1.28071249e+00 6.41671717e-02
5.79694271e-01 -1.17596650e+00 -1.61135328e+00 2.20813528e-01
2.46545434e-01 -1.13137281e+00 -3.68538529e-01 8.72915089e-02
-1.75298989e-01 -1.41925418e+00 -1.30660996e-01 -2.54707575e-01
-2.55942225e-01 7.39751756e-01 1.14596105e+00 3.94448191e-01
-4.99129415e-01 6.93300784e-01 -5.55344164e-01 -6.41162276e-01
-2.20877007e-02 -1.77390664e-03 5.23839295e-01 -7.33105958e-01
2.96587974e-01 -8.13299954e-01 -1.77269623e-01 4.39034134e-01
-7.09268034e-01 -1.13372549e-01 7.67571852e-02 2.91504830e-01
-1.84224397e-01 1.33918330e-01 -2.68552333e-01 -2.20243603e-01
1.07871079e+00 -9.49351668e-01 -1.19420326e+00 -2.72273600e-01
-2.61702180e-01 -7.88519442e-01 4.02274549e-01 -2.25423217e-01
-4.48563546e-01 -1.75366402e-01 9.74904448e-02 -4.86425430e-01
-3.55100542e-01 4.81673300e-01 2.58513331e-01 -5.05018055e-01
1.04996777e+00 -1.44562721e-01 1.30224794e-01 -8.60579237e-02
1.72679290e-01 6.57016814e-01 6.13439441e-01 -6.56806171e-01
6.75560355e-01 4.08818185e-01 -2.54775494e-01 -7.71641433e-01
8.96733403e-02 -4.91052598e-01 -5.59616573e-02 -5.62429726e-01
7.41131365e-01 -6.16848826e-01 -1.30152202e+00 3.52029085e-01
-1.10833406e+00 -7.62992561e-01 -1.33389458e-01 2.04199210e-01
-5.35610259e-01 -1.93092868e-01 -3.75686795e-01 -3.70395750e-01
-4.38481808e-01 -1.15706348e+00 5.41185379e-01 6.30014896e-01
-3.41816574e-01 -6.36405289e-01 8.46044183e-01 2.70376265e-01
8.14130723e-01 4.38904554e-01 -2.91307122e-02 -1.01114392e+00
-4.47388291e-01 -3.21951091e-01 1.16335422e-01 -5.96206188e-01
-8.00423250e-02 3.18229437e-01 -2.32690737e-01 -4.13509458e-01
-3.95834833e-01 -2.24662557e-01 -2.97631025e-01 -5.34139536e-02
1.29238695e-01 -2.40014359e-01 -5.29338241e-01 5.89704931e-01
1.18678355e+00 6.09600663e-01 3.17830890e-01 9.73872304e-01
3.54546577e-01 -5.70993535e-02 9.54141200e-01 8.53747129e-01
2.05832765e-01 8.97324085e-01 9.81909811e-01 7.05957189e-02
3.37148398e-01 2.02091839e-02 5.90386391e-01 2.18495354e-01
-7.43262053e-01 -2.39833370e-01 -1.36509931e+00 4.96571124e-01
-2.02768421e+00 -1.16967368e+00 1.58906892e-01 1.71268296e+00
5.23205400e-01 1.63119033e-01 1.02038181e+00 -3.00871879e-01
3.77442330e-01 1.45536289e-01 -4.10208315e-01 -4.32613909e-01
3.26541007e-01 5.60396433e-01 9.26424444e-01 1.70852154e-01
-7.88570225e-01 1.79406893e+00 7.93091345e+00 6.63565993e-01
-1.21365607e+00 4.19816524e-01 -7.18157649e-01 -3.96818042e-01
3.52033883e-01 3.09604734e-01 -5.76829553e-01 3.69351774e-01
1.08720863e+00 -9.79855835e-01 1.01894593e+00 1.35576880e+00
-9.22884196e-02 -1.38767034e-01 -4.15735692e-01 7.68794477e-01
1.17371045e-02 -1.81369376e+00 -4.47988510e-01 3.20025414e-01
1.36182413e-01 7.70773172e-01 -4.05605167e-01 2.35836476e-01
1.70464313e+00 -1.02723718e+00 7.46077597e-01 -2.76622027e-01
7.81054318e-01 -8.55623424e-01 5.08085907e-01 4.08744127e-01
-9.29965317e-01 -5.60265919e-03 -2.20419347e-01 -6.13313854e-01
2.62527198e-01 -4.39144224e-01 -1.02422321e+00 3.01440716e-01
1.22355306e+00 4.35290247e-01 -2.00397655e-01 1.21233606e+00
-1.09056942e-01 4.54297990e-01 -5.87460697e-01 -1.91920280e-01
5.05839169e-01 -1.61908641e-01 1.18487275e+00 8.69865060e-01
2.20872499e-02 6.11978710e-01 5.87252021e-01 3.43164444e-01
3.74052912e-01 -2.85239607e-01 -1.03411674e+00 -4.65913653e-01
6.96900964e-01 1.55096507e+00 -9.71558988e-01 -1.86007440e-01
-7.07190633e-02 9.60507631e-01 2.63367593e-02 1.03736885e-01
-1.08109701e+00 -8.06245148e-01 1.66924953e+00 2.09201083e-01
6.10665977e-02 -9.98260021e-01 8.32769349e-02 -1.08381140e+00
-7.35481620e-01 -1.52399349e+00 4.47158277e-01 -9.11642611e-01
-8.09318602e-01 8.30522001e-01 1.68616310e-01 -1.01615429e+00
-4.69054401e-01 -5.17192304e-01 -1.05691576e+00 2.75538087e-01
-6.89494491e-01 -1.01608920e+00 -2.91034877e-01 1.02738643e+00
2.94896781e-01 -9.92061257e-01 8.72206986e-01 -4.50480394e-02
-7.37459540e-01 9.20081958e-02 -5.05489707e-01 1.15888424e-01
5.65260410e-01 -1.00940669e+00 8.70281279e-01 8.93447101e-01
1.15578547e-02 8.16705048e-01 1.16824794e+00 -9.92285192e-01
-1.89002860e+00 -3.80104899e-01 -1.65285394e-01 -4.24492568e-01
1.15040481e+00 -1.69379175e-01 -2.53719445e-02 1.21856391e+00
9.50140119e-01 -2.15858996e-01 5.89146554e-01 -1.91226363e-01
5.48009165e-02 -2.15452332e-02 -1.26333606e+00 1.07012630e+00
1.05277801e+00 9.93160009e-02 -5.85742474e-01 5.32045066e-01
6.90674961e-01 -8.17896545e-01 -6.54024184e-01 -3.00714672e-01
3.17305773e-01 -1.04474747e+00 6.19235933e-01 -8.10962379e-01
-1.09556668e-01 -7.24213481e-01 -1.57181680e-01 -1.66926038e+00
-3.28965843e-01 -1.48716331e+00 1.35940850e-01 7.13211238e-01
-1.85055807e-01 -8.32322299e-01 9.01185513e-01 5.70903182e-01
-3.11653972e-01 1.46465614e-01 -1.12978566e+00 -8.86015356e-01
-5.08858025e-01 -3.00881952e-01 8.85310829e-01 1.17920220e+00
7.53506064e-01 2.68272221e-01 -1.69790000e-01 1.06474131e-01
6.92247033e-01 -4.55164492e-01 1.40162110e+00 -1.07194281e+00
-4.88601148e-01 -2.49839634e-01 -9.61744487e-01 -4.93765295e-01
-1.32332832e-01 -6.20896518e-01 -1.26284137e-01 -1.13690746e+00
-4.45228726e-01 -3.95792574e-01 1.53915897e-01 9.11143422e-01
5.94058692e-01 3.04514378e-01 4.74955767e-01 3.60181242e-01
-6.95132136e-01 3.40929963e-02 6.18380785e-01 2.17960730e-01
8.05902258e-02 -1.49259329e-01 -6.23793781e-01 7.52131462e-01
8.44195783e-01 -4.34864730e-01 -4.26815718e-01 -3.69412452e-01
1.69795454e-01 9.01088789e-02 2.63489068e-01 -1.36805415e+00
7.36977875e-01 -9.16646421e-01 -2.99613625e-01 -1.95485264e-01
4.39391106e-01 -9.15026486e-01 4.24531966e-01 5.24757862e-01
5.65870643e-01 8.49759579e-01 5.50212622e-01 1.15683727e-01
2.21033767e-01 3.34393568e-02 4.74587500e-01 -3.21476668e-01
-1.12442029e+00 1.75880313e-01 -1.06010509e+00 4.32181396e-02
1.73638725e+00 -1.69061586e-01 -1.08145142e+00 -4.00377601e-01
-1.33485451e-01 3.83875161e-01 1.17894518e+00 3.09053510e-01
4.17101771e-01 -8.71496022e-01 -3.45670164e-01 1.57629456e-02
-9.82972011e-02 -6.34639204e-01 -2.63913780e-01 4.79442775e-01
-1.46540427e+00 3.08371577e-02 -1.35504270e+00 -2.73276716e-01
-1.50891864e+00 4.65055794e-01 3.28662336e-01 -5.16117085e-03
-8.12050462e-01 6.66128337e-01 -4.29245591e-01 -3.13909918e-01
4.16855924e-02 5.32233417e-01 -1.78496197e-01 -6.48690462e-01
8.42942655e-01 4.64442611e-01 -5.07949712e-03 -5.77683270e-01
-8.62793446e-01 -3.50984000e-02 1.39542162e-01 -7.64424741e-01
1.36776924e+00 4.16103393e-01 -2.24990547e-01 2.16512799e-01
-2.33955905e-01 3.96985896e-02 -1.06053436e+00 4.55773413e-01
-1.44601166e-01 -7.38680899e-01 2.44818971e-01 -8.77594769e-01
-9.65311527e-01 -4.30250634e-03 2.75005817e-01 1.86245307e-01
5.98965526e-01 -1.67599991e-01 7.99736023e-01 3.73769909e-01
1.25009239e+00 -8.39824855e-01 -8.02425668e-02 6.68430626e-01
7.18345106e-01 -4.67030704e-01 6.87760264e-02 -1.88657612e-01
-1.16025269e+00 8.80969942e-01 8.94935369e-01 -4.90547091e-01
3.76822382e-01 8.47874165e-01 3.16830069e-01 -9.18267071e-01
-1.06340945e+00 -3.86065394e-01 -7.18569160e-01 1.22578740e+00
-1.86395347e-02 1.37925427e-02 -4.02061224e-01 6.11839831e-01
-5.96177578e-01 9.29914713e-02 1.08588016e+00 1.74318945e+00
-5.76901138e-01 -9.78333175e-01 -5.60570419e-01 -1.20534748e-01
-4.44842398e-01 4.80255969e-02 -1.03276694e+00 8.84705663e-01
-2.89080627e-02 1.01744235e+00 1.66147754e-01 -7.40599930e-01
4.88762021e-01 -4.26966399e-01 1.92359552e-01 -7.07441747e-01
-1.35795200e+00 -3.77627492e-01 6.54918909e-01 -9.89859998e-01
1.42035261e-01 -2.39607900e-01 -1.21319437e+00 -1.35510099e+00
2.52282377e-02 4.41914022e-01 9.94210541e-01 6.36973798e-01
6.32292032e-01 3.18053812e-01 -4.75528277e-02 -1.23420215e+00
-6.40429705e-02 -5.87378383e-01 -7.99876928e-01 -1.90156147e-01
-9.57103446e-02 -1.00542200e+00 -8.66874605e-02 -3.79228413e-01] | [3.5756454467773438, 1.4515228271484375] |
149006f0-172f-466a-a6a3-9f26680ff2e9 | frame-level-speaker-embeddings-for-text | 1809.04437 | null | http://arxiv.org/abs/1809.04437v1 | http://arxiv.org/pdf/1809.04437v1.pdf | Frame-level speaker embeddings for text-independent speaker recognition and analysis of end-to-end model | In this paper, we propose a Convolutional Neural Network (CNN) based speaker
recognition model for extracting robust speaker embeddings. The embedding can
be extracted efficiently with linear activation in the embedding layer. To
understand how the speaker recognition model operates with text-independent
input, we modify the structure to extract frame-level speaker embeddings from
each hidden layer. We feed utterances from the TIMIT dataset to the trained
network and use several proxy tasks to study the networks ability to represent
speech input and differentiate voice identity. We found that the networks are
better at discriminating broad phonetic classes than individual phonemes. In
particular, frame-level embeddings that belong to the same phonetic classes are
similar (based on cosine distance) for the same speaker. The frame level
representation also allows us to analyze the networks at the frame level, and
has the potential for other analyses to improve speaker recognition. | ['James Glass', 'Suwon Shon', 'Hao Tang'] | 2018-09-12 | null | null | null | null | ['text-independent-speaker-recognition'] | ['speech'] | [ 1.37248844e-01 1.12136476e-01 -6.65718466e-02 -8.47091496e-01
-6.97137356e-01 -7.20216215e-01 4.71707910e-01 -1.43957153e-01
-3.99915993e-01 -5.23121320e-02 6.77536428e-01 -5.54352999e-01
2.03027233e-01 -4.43121791e-01 -4.19646233e-01 -6.35279119e-01
-1.96463391e-01 -8.98559403e-04 -2.57365823e-01 1.15383312e-01
-5.21515459e-02 7.55225778e-01 -1.58281553e+00 6.75171316e-01
2.18057141e-01 9.24995422e-01 -1.37798950e-01 9.93905902e-01
-1.14950866e-01 4.73466218e-01 -8.21927190e-01 -1.65682808e-02
-2.94216461e-02 -4.58018988e-01 -8.73895168e-01 -1.18039176e-01
7.83992231e-01 -3.76128316e-01 -4.02279437e-01 7.71831930e-01
6.26854122e-01 2.96029061e-01 6.59941852e-01 -1.02102494e+00
-7.87865222e-01 6.71824157e-01 3.44088793e-01 6.06067419e-01
3.18142831e-01 4.72746454e-02 1.04824829e+00 -1.12480617e+00
1.93255782e-01 1.73331225e+00 6.86277866e-01 6.92255139e-01
-1.28926837e+00 -5.20039380e-01 1.46548584e-01 2.99899131e-01
-1.25152636e+00 -1.21114218e+00 8.20017815e-01 -4.21795249e-01
1.16622353e+00 5.50116420e-01 2.65722275e-01 1.16294646e+00
-2.76216596e-01 7.46983290e-01 8.41084778e-01 -5.62460244e-01
1.74865246e-01 2.58031756e-01 3.58337432e-01 5.24153590e-01
-4.00271028e-01 3.33372146e-01 -7.54936755e-01 2.61790585e-02
5.81856906e-01 -3.06032836e-01 -3.20382267e-01 5.38825206e-02
-1.18382382e+00 9.80452061e-01 4.74872202e-01 5.60281098e-01
-1.69096306e-01 1.54969007e-01 5.02885342e-01 4.05613482e-01
3.60182732e-01 3.10906708e-01 -4.66984421e-01 -6.90127090e-02
-1.05615342e+00 -2.32056767e-01 8.69371116e-01 3.27023089e-01
4.64739650e-01 7.01602876e-01 -1.16145775e-01 1.01869953e+00
5.81359327e-01 2.54651815e-01 6.88665867e-01 -9.33199525e-01
1.07466832e-01 -3.52351665e-02 -3.73959333e-01 -7.01725066e-01
-2.44121507e-01 -3.02809775e-01 -5.12760401e-01 1.79442033e-01
5.31517267e-01 -6.68174922e-02 -9.25726533e-01 1.84290683e+00
-1.07588880e-02 1.93124011e-01 3.19988966e-01 7.78565347e-01
1.09965241e+00 8.02749813e-01 5.05017340e-02 2.32156157e-01
1.68378222e+00 -7.58981943e-01 -6.38018131e-01 -2.71002293e-01
6.11588359e-01 -6.63337469e-01 1.13801301e+00 -5.46007827e-02
-8.85603070e-01 -8.72555315e-01 -1.06203377e+00 -1.40910134e-01
-6.24614775e-01 2.45915830e-01 1.82149127e-01 9.57976758e-01
-1.25610232e+00 2.64227390e-01 -7.26652265e-01 -3.99447769e-01
2.60982692e-01 2.82064050e-01 -3.33890110e-01 1.94678843e-01
-1.27069664e+00 9.41948652e-01 2.10298896e-01 3.20842094e-03
-1.06551790e+00 -6.75879657e-01 -1.25850499e+00 4.13556844e-01
-5.14624834e-01 -2.28609994e-01 1.32254493e+00 -1.23972750e+00
-1.67094541e+00 7.05844223e-01 -7.56356239e-01 -5.19694746e-01
-1.09848723e-01 3.36905360e-01 -6.29715919e-01 2.15396971e-01
-2.90480763e-01 8.92490149e-01 1.02815855e+00 -8.20901573e-01
-4.38519150e-01 -2.97391891e-01 -6.88443258e-02 -1.15986556e-01
-6.40766144e-01 4.41578209e-01 2.12534755e-01 -6.46939456e-01
2.04076722e-01 -7.70029187e-01 4.30349737e-01 -1.02621257e-01
-3.05124730e-01 -4.79707330e-01 8.87602866e-01 -1.00424552e+00
9.34325278e-01 -2.51560211e+00 -4.41845832e-03 1.60439536e-02
-2.51179300e-02 2.20046148e-01 -3.65870416e-01 1.60434112e-01
-3.05453449e-01 3.19379240e-01 -7.03958496e-02 -5.36725342e-01
2.48305082e-01 4.26345021e-02 -4.00822133e-01 4.23979640e-01
4.93398100e-01 8.08044791e-01 -3.90891761e-01 -1.11426964e-01
1.19353272e-01 9.33876634e-01 -4.93434936e-01 1.21552758e-01
1.92570582e-01 1.49732813e-01 3.22825491e-01 4.68638480e-01
4.51420307e-01 5.45999587e-01 5.99370003e-02 -2.97964722e-01
-4.41495776e-02 1.02810299e+00 -9.55495238e-01 1.14982831e+00
-7.01464772e-01 1.44472504e+00 3.79753649e-01 -1.05861425e+00
9.49762225e-01 7.05472410e-01 -1.33536747e-02 -3.93179148e-01
4.02617455e-02 8.44354331e-02 4.17678088e-01 -2.98125714e-01
3.43130946e-01 -3.70527655e-01 9.87491310e-02 5.84864855e-01
3.98477346e-01 2.84511685e-01 -3.17858279e-01 -2.92598724e-01
7.24898458e-01 -6.09907746e-01 -3.02961975e-01 -3.53542030e-01
5.46549082e-01 -4.66317147e-01 2.90364653e-01 4.43578064e-01
-5.39028704e-01 5.74339390e-01 2.08013952e-01 -3.54572564e-01
-8.97761881e-01 -1.28120816e+00 -4.17745590e-01 1.60438788e+00
-5.75762451e-01 -8.87247548e-02 -7.30496049e-01 -4.22355384e-01
9.03552622e-02 8.43111157e-01 -5.66444337e-01 -2.56407768e-01
-5.75696170e-01 -1.40087500e-01 1.18740928e+00 9.04098690e-01
1.33561268e-01 -1.09609652e+00 -2.56023496e-01 8.80821943e-02
-7.67367631e-02 -1.07652211e+00 -6.93612099e-01 5.24104178e-01
-6.82175994e-01 -4.96425807e-01 -6.66385114e-01 -1.29986989e+00
4.47871327e-01 2.52221152e-02 8.53921115e-01 -1.32185310e-01
5.46323881e-02 3.27476591e-01 -1.07541949e-01 -5.04973054e-01
-7.54854262e-01 2.29657188e-01 4.10297841e-01 1.24179251e-01
7.63653338e-01 -2.46424317e-01 -1.46660909e-01 3.13935697e-01
-7.08603799e-01 -5.15706480e-01 9.30235460e-02 6.54590428e-01
2.75177341e-02 2.27326853e-03 8.03475142e-01 -8.55217651e-02
7.15469480e-01 -1.62811697e-01 -1.86484143e-01 -8.73082355e-02
8.22826009e-03 2.14180574e-01 5.31399012e-01 -5.31153262e-01
-7.09851086e-01 3.78792249e-02 -5.22853136e-01 -2.97599822e-01
-6.61508620e-01 4.40618247e-01 -3.40099514e-01 1.62678853e-01
5.65308392e-01 4.08153057e-01 3.20667565e-01 -5.58926046e-01
4.48472142e-01 1.16687381e+00 5.81612945e-01 -2.50822902e-01
5.48131168e-01 2.71472245e-01 -7.31902540e-01 -1.43336523e+00
-4.31265175e-01 -3.26857865e-01 -7.81153679e-01 4.94952686e-02
1.04221904e+00 -8.82241547e-01 -7.67825663e-01 2.91995853e-01
-1.27051079e+00 -2.70504266e-01 -2.68234044e-01 8.12771559e-01
-3.05711001e-01 8.00703838e-02 -7.92556405e-01 -9.34069335e-01
-9.62468386e-02 -1.24609470e+00 9.12886977e-01 5.86864464e-02
-4.87978816e-01 -1.14587259e+00 2.60654353e-02 1.47222936e-01
5.94695270e-01 -3.69631886e-01 1.04245853e+00 -1.19578218e+00
-1.65642545e-01 -8.91057774e-02 2.60840822e-02 7.74586320e-01
3.12307149e-01 2.43286073e-01 -1.79717517e+00 -3.52461606e-01
1.58237860e-01 -1.84457954e-02 1.06899250e+00 4.69712645e-01
1.02380216e+00 -5.35139024e-01 1.07627548e-02 6.91425323e-01
6.83225036e-01 2.34460697e-01 5.68724334e-01 2.07786411e-01
5.14618635e-01 8.92933249e-01 -4.86668080e-01 -2.94875681e-01
2.23998979e-01 5.59223235e-01 -5.72391115e-02 -4.34941053e-02
-3.64849061e-01 -1.34406313e-01 9.53342259e-01 1.14469624e+00
5.26642084e-01 -2.97414288e-02 -8.94379556e-01 8.11796606e-01
-1.15855873e+00 -1.22489321e+00 3.28768641e-01 2.02009010e+00
7.20054090e-01 1.59122452e-01 3.29884142e-01 3.71270150e-01
8.02340329e-01 3.79019767e-01 -3.20863605e-01 -1.11488831e+00
-1.74557000e-01 3.04187715e-01 1.81745517e-03 7.86644518e-01
-1.01522684e+00 8.03586721e-01 7.51105738e+00 3.48904490e-01
-1.51171124e+00 -4.74857800e-02 5.43023229e-01 -1.28080964e-01
-1.95141420e-01 -3.23916912e-01 -8.73847961e-01 3.32385451e-01
1.69620907e+00 4.46524583e-02 4.70998377e-01 8.00616264e-01
1.57406628e-01 5.56190729e-01 -1.64722919e+00 7.77468026e-01
1.59174189e-01 -1.24530184e+00 6.57571629e-02 1.92764640e-01
1.12606436e-01 8.44819844e-03 4.24556613e-01 3.86329919e-01
7.95659497e-02 -1.45409167e+00 8.32328081e-01 2.47213602e-01
6.56329870e-01 -7.72005320e-01 6.77712977e-01 1.42506510e-01
-1.26055193e+00 -8.33322480e-02 -4.49598044e-01 -7.05926642e-02
-6.29421324e-02 6.06329702e-02 -1.41137087e+00 -7.91155100e-02
6.37614965e-01 4.20209080e-01 -4.49980527e-01 6.60974920e-01
2.59975642e-02 9.70064104e-01 -3.48106235e-01 2.15403456e-02
1.54560179e-01 2.50612825e-01 4.69610423e-01 1.51971734e+00
1.59510061e-01 -3.12441766e-01 -2.07659051e-01 9.53794718e-01
-1.66894734e-01 -1.93563461e-01 -5.17312407e-01 -4.11867589e-01
7.53881037e-01 1.03258717e+00 -3.05243552e-01 -3.95608604e-01
-3.08971763e-01 6.99476540e-01 1.32801339e-01 5.63033521e-01
-4.37216014e-01 -5.66526651e-01 1.46083224e+00 -1.23171031e-01
5.66275954e-01 -5.58491170e-01 -2.68214703e-01 -8.89400899e-01
-2.11928129e-01 -9.47113752e-01 9.62927863e-02 -6.44978106e-01
-1.25652659e+00 7.26124346e-01 -2.81836540e-01 -6.90703332e-01
-5.25779128e-01 -9.11287606e-01 -9.85897481e-01 1.23702931e+00
-1.44021356e+00 -8.08112502e-01 1.42306164e-01 3.85907143e-01
6.86684012e-01 -3.94562215e-01 1.16514957e+00 1.14751339e-01
-6.57258868e-01 1.03690732e+00 1.69412866e-01 8.70014787e-01
5.99644125e-01 -1.12849998e+00 7.19288468e-01 7.29730129e-01
7.20838487e-01 1.01760793e+00 3.25871021e-01 3.26780155e-02
-1.12426198e+00 -9.26656485e-01 1.12997210e+00 -5.94739139e-01
3.92266870e-01 -6.88572943e-01 -1.13713384e+00 7.68846214e-01
4.19691086e-01 -1.64819226e-01 1.18074191e+00 3.15611422e-01
-7.70171046e-01 -7.70640448e-02 -1.05552745e+00 3.28981280e-01
4.85849112e-01 -1.59527147e+00 -1.00055647e+00 -5.98480590e-02
8.36518407e-01 7.95516446e-02 -7.49468982e-01 1.60269048e-02
7.55283415e-01 -8.02097082e-01 1.16188443e+00 -9.98461366e-01
1.26835201e-02 -1.84142470e-01 -5.46211183e-01 -1.45974874e+00
-3.98179621e-01 -4.03419361e-02 2.68280273e-03 1.46149445e+00
7.31829107e-01 -7.08823383e-01 6.23414755e-01 5.16623437e-01
-1.65757224e-01 -3.14620286e-01 -1.22601438e+00 -1.01386631e+00
3.55715543e-01 -6.65738106e-01 8.04502606e-01 1.06187832e+00
-4.81289364e-02 3.54270399e-01 1.96318567e-01 4.95720536e-01
1.98861897e-01 -2.80502915e-01 3.85173440e-01 -1.22362995e+00
-6.67840466e-02 -6.84088349e-01 -5.91969669e-01 -9.28430676e-01
6.10142529e-01 -1.22440743e+00 1.76792234e-01 -1.29989874e+00
-4.12379652e-01 -3.59950103e-02 -4.67865974e-01 5.17012417e-01
1.00540183e-01 1.98123485e-01 2.13193759e-01 1.02643766e-01
1.85094789e-01 3.38681877e-01 4.86883730e-01 -3.92690301e-01
-1.02478959e-01 -9.17093456e-02 -5.44530630e-01 7.31396258e-01
9.46431816e-01 -2.78752714e-01 -3.03931069e-02 -7.41204381e-01
-5.16173244e-01 -3.00234914e-01 4.73991781e-01 -9.48583364e-01
1.31763309e-01 3.53969693e-01 6.39535129e-01 -3.28615308e-01
5.41779816e-01 -6.35603666e-01 -2.95908719e-01 4.25795048e-01
-8.98442805e-01 -1.35320544e-01 4.67854470e-01 1.03965402e-01
-3.81092250e-01 -3.05197746e-01 8.99490893e-01 1.49591148e-01
-5.04219115e-01 -1.51106253e-01 -8.31543863e-01 -2.70457834e-01
2.37539247e-01 -3.09402645e-01 -1.97715849e-01 -5.05523503e-01
-9.05329883e-01 -4.05349821e-01 2.09336832e-01 6.36372685e-01
5.36033869e-01 -1.54126441e+00 -7.65443861e-01 6.89296305e-01
5.11833839e-02 -6.49857044e-01 -1.32102698e-01 4.76469487e-01
-2.07461163e-01 6.78259313e-01 -1.62277728e-01 -7.78236628e-01
-1.29967284e+00 3.36952358e-01 7.56680965e-01 5.59552133e-01
-6.12107366e-02 1.09283745e+00 2.09512293e-01 -8.08053672e-01
5.43641865e-01 -7.23292768e-01 -1.22154392e-01 4.04061705e-01
8.11973929e-01 2.00542673e-01 2.00074568e-01 -1.09217668e+00
-6.92570090e-01 2.20040828e-01 1.00882500e-02 -3.40361625e-01
1.25775218e+00 -4.49052528e-02 1.62707970e-01 7.12581217e-01
1.76951444e+00 8.81433189e-02 -1.02194774e+00 -2.75605321e-01
2.80890651e-02 -8.76676515e-02 3.19079727e-01 -5.50301373e-01
-8.97866547e-01 1.38928628e+00 8.44386101e-01 6.26787424e-01
5.45340538e-01 4.08732891e-02 6.91092014e-01 3.36755961e-01
-4.37938958e-01 -9.38733816e-01 -1.23995073e-01 7.94100761e-01
7.17483699e-01 -1.08055806e+00 -6.22016847e-01 9.74069089e-02
-4.20410335e-01 1.45560956e+00 2.65247136e-01 6.46083131e-02
7.77858734e-01 2.59594709e-01 3.98819864e-01 9.43284389e-03
-5.77962399e-01 -2.06792921e-01 6.29061759e-01 8.20767701e-01
8.72155190e-01 1.49669200e-01 4.84015673e-01 4.71198380e-01
-6.95944428e-01 -7.17830598e-01 2.51070648e-01 5.24933457e-01
-7.75257885e-01 -8.86821508e-01 -6.22843862e-01 2.04817265e-01
-3.30304354e-01 -3.15912426e-01 -7.09157646e-01 3.15948516e-01
-2.57665753e-01 1.20047104e+00 4.54576313e-01 -5.14036298e-01
2.29998603e-01 9.54902887e-01 1.28551781e-01 -7.09633768e-01
-6.33212447e-01 -4.04477604e-02 1.45435706e-01 -2.38764629e-01
-2.19155133e-01 -6.44680858e-01 -1.14985371e+00 -1.35909289e-01
-2.69034445e-01 1.67177737e-01 1.05238938e+00 8.74897718e-01
2.82629669e-01 6.11156702e-01 7.81998932e-01 -9.15186346e-01
-4.85405326e-01 -1.15187478e+00 -3.62037808e-01 1.34668842e-01
9.15394068e-01 -2.08806589e-01 -7.73253500e-01 3.00221741e-01] | [14.370903015136719, 6.198997497558594] |
aed5ef90-1ccb-4d16-891c-725d03c3ed38 | tweettaglish-a-dataset-for-investigating | null | null | https://aclanthology.org/2022.lrec-1.225 | https://aclanthology.org/2022.lrec-1.225.pdf | TweetTaglish: A Dataset for Investigating Tagalog-English Code-Switching | Deploying recent natural language processing innovations to low-resource settings allows for state-of-the-art research findings and applications to be accessed across cultural and linguistic borders. One low-resource setting of increasing interest is code-switching, the phenomenon of combining, swapping, or alternating the use of two or more languages in continuous dialogue. In this paper, we introduce a large dataset (20k+ instances) to facilitate investigation of Tagalog-English code-switching, which has become a popular mode of discourse in Philippine culture. Tagalog is an Austronesian language and former official language of the Philippines spoken by over 23 million people worldwide, but it and Tagalog-English are under-represented in NLP research and practice. We describe our methods for data collection, as well as our labeling procedures. We analyze our resulting dataset, and finally conclude by providing results from a proof-of-concept regression task to establish dataset validity, achieving a strong performance benchmark (R2=0.797-0.909; RMSE=0.068-0.057). | ['Natalie Parde', 'Ankit Aich', 'Megan Herrera'] | null | null | null | null | lrec-2022-6 | ['culture'] | ['speech'] | [-3.10064852e-01 2.05934001e-03 -5.25087893e-01 -1.48771241e-01
-8.26039612e-01 -9.78277504e-01 7.77320683e-01 3.54707539e-01
-5.75892746e-01 7.09938467e-01 7.65022993e-01 -5.15787244e-01
1.37092486e-01 -2.50115097e-01 -3.80037695e-01 -2.03682855e-01
-2.04711810e-01 1.56457543e-01 -3.16666394e-01 -3.13577056e-01
5.79869211e-01 -5.69409840e-02 -1.12583184e+00 2.74185568e-01
1.30097711e+00 1.91621751e-01 3.31639051e-01 4.74967867e-01
-2.32231066e-01 9.41120803e-01 -6.27688825e-01 -5.63180208e-01
1.95306670e-02 -3.09811592e-01 -1.00540519e+00 -2.20867544e-01
3.10640126e-01 -7.55618140e-02 -9.04823747e-03 9.48925853e-01
2.72329867e-01 -5.38215451e-02 3.94356757e-01 -8.98042262e-01
-9.82390702e-01 9.22945976e-01 -4.66150910e-01 3.24443020e-02
9.12483990e-01 -2.15073824e-02 1.13476384e+00 -8.11285973e-01
1.09263718e+00 1.17136180e+00 8.01048338e-01 4.10174221e-01
-1.11878204e+00 -7.44843006e-01 -1.34337861e-02 -3.11014056e-01
-1.51445520e+00 -7.17327893e-01 3.64306003e-01 -9.34504807e-01
1.18171239e+00 1.25402406e-01 8.57502759e-01 1.22141790e+00
1.13039926e-01 6.69111013e-01 1.21766877e+00 -5.37257910e-01
-1.09540597e-01 5.17875850e-01 -8.43351148e-03 8.18686604e-01
1.86769441e-01 -5.14342129e-01 -7.22768128e-01 -4.92130995e-01
3.43404800e-01 -2.74271399e-01 -1.16402738e-01 3.28850776e-01
-1.35300922e+00 1.15268302e+00 3.37320231e-02 5.62704146e-01
-4.92452607e-02 -1.51049972e-01 4.62160915e-01 4.36931700e-01
6.49134517e-01 6.14955246e-01 -5.42913616e-01 -1.03284514e+00
-9.05514717e-01 2.12270632e-01 1.18936574e+00 1.12926960e+00
2.85697699e-01 -2.84857392e-01 2.57987916e-01 1.54920578e+00
4.11125898e-01 5.59806585e-01 6.11746967e-01 -1.07479548e+00
8.61483693e-01 6.13861084e-01 6.46232292e-02 -1.19638228e+00
-4.66323584e-01 -1.05347551e-01 -4.43987757e-01 -7.48521864e-01
5.38396716e-01 -5.71550071e-01 2.40557361e-03 1.82347298e+00
-3.30482572e-02 -3.99784535e-01 8.73283893e-02 5.49516559e-01
4.85949636e-01 6.45807147e-01 4.09733355e-02 -4.78847980e-01
1.45485818e+00 -8.98434222e-01 -5.12958944e-01 -3.27068150e-01
1.01234484e+00 -9.46681440e-01 1.46688187e+00 3.86338025e-01
-9.05213475e-01 -3.85221541e-02 -5.22322059e-01 -2.12891027e-01
-4.77746040e-01 1.64385378e-01 7.43422031e-01 7.56215572e-01
-9.53094661e-01 -1.09328441e-01 -5.95036805e-01 -9.46786702e-01
2.65394330e-01 -1.68668583e-01 -4.18102831e-01 -8.72016251e-02
-1.07439244e+00 9.04189706e-01 2.18662649e-01 -1.17230482e-01
-3.02706420e-01 -7.31346846e-01 -8.79634380e-01 -4.07992274e-01
4.10409749e-01 1.42714992e-01 1.23238873e+00 -9.91314411e-01
-1.49862552e+00 1.24561810e+00 -1.99427456e-01 -8.39598104e-02
3.36220443e-01 -3.53439003e-01 -7.06660151e-01 -1.17167175e-01
6.06251180e-01 4.09054101e-01 2.82896876e-01 -9.50965762e-01
-6.62258565e-01 -8.24723616e-02 1.67406704e-02 1.96651295e-01
-4.30175930e-01 7.46137679e-01 -3.86387050e-01 -7.30442286e-01
-1.72868013e-01 -1.13211775e+00 1.68314517e-01 -3.15519512e-01
-3.98138225e-01 -1.34737864e-01 1.39870450e-01 -1.08360958e+00
1.81308877e+00 -2.30547190e+00 -7.84813054e-03 1.23559892e-01
2.53463268e-01 5.31475134e-02 4.92368303e-02 8.85329068e-01
2.88100392e-01 5.98094165e-01 -4.26601589e-01 -1.22191906e-01
-5.64225316e-02 2.89424043e-02 -5.73545620e-02 5.41630745e-01
2.16541607e-02 6.70114219e-01 -1.12908232e+00 -4.19506997e-01
-2.20492497e-01 9.84772667e-02 -6.83111191e-01 -1.53662458e-01
1.07165657e-01 1.95689872e-01 -2.08425120e-01 8.45285594e-01
2.76448458e-01 -1.33351266e-01 5.61384737e-01 5.50351560e-01
-7.76845753e-01 4.80284691e-01 -6.69960082e-01 1.73522270e+00
-4.94144350e-01 1.10904086e+00 3.21801901e-01 -3.87151629e-01
8.16037118e-01 3.25674079e-02 3.01176846e-01 -6.33842647e-01
-5.10443933e-02 3.67656291e-01 2.39711642e-01 -8.51893425e-01
8.96558285e-01 1.02689594e-01 -7.50353515e-01 5.27008414e-01
-2.34750167e-01 -1.29673153e-01 4.21196491e-01 3.41876417e-01
9.32553232e-01 -2.69150943e-01 7.54292309e-01 -7.74252653e-01
3.86120409e-01 4.30944830e-01 6.17259264e-01 5.83647549e-01
-2.99599737e-01 1.61970288e-01 8.80688369e-01 -1.76931638e-02
-1.08419013e+00 -6.40869021e-01 -3.28896165e-01 1.41326416e+00
-4.48851138e-01 -6.38200700e-01 -7.09255159e-01 -2.44013309e-01
1.06023163e-01 9.34651554e-01 -4.19767380e-01 1.79437250e-01
-4.69921410e-01 -6.62373185e-01 9.46259558e-01 -1.45258725e-01
6.42382264e-01 -7.98794270e-01 -5.86343765e-01 2.23404497e-01
-6.69952512e-01 -1.16854894e+00 -5.59314013e-01 -1.85931474e-01
-3.05471390e-01 -1.07839978e+00 -4.95561123e-01 -8.35704863e-01
4.18431878e-01 2.29721531e-01 1.17567408e+00 1.26516655e-01
-3.20779495e-02 3.78808588e-01 -6.31697834e-01 -1.88009441e-01
-7.09142506e-01 4.48460907e-01 1.96685866e-01 -3.43633622e-01
4.86819267e-01 3.02198622e-02 -6.72174692e-02 5.92490733e-02
-7.25604177e-01 7.21344277e-02 1.27544656e-01 6.50346279e-01
-3.10345352e-01 -3.68884921e-01 4.79223937e-01 -1.17745554e+00
1.19483674e+00 -9.82719481e-01 -5.35121620e-01 3.80078793e-01
-4.15119380e-01 -3.88093203e-01 6.60144746e-01 -3.59228879e-01
-1.12928629e+00 -3.34681273e-01 7.23304376e-02 4.60366815e-01
8.36523622e-02 1.24928701e+00 4.12663579e-01 2.81841934e-01
6.21840417e-01 -9.78816077e-02 7.50334784e-02 -2.77623266e-01
3.81526977e-01 1.32224524e+00 3.29890013e-01 -7.71020770e-01
3.43792111e-01 2.48639628e-01 -9.00987148e-01 -1.47505724e+00
-4.82243031e-01 -5.98672569e-01 -6.28152668e-01 -3.05496275e-01
8.74696016e-01 -1.15411758e+00 -7.14814663e-01 5.67587495e-01
-1.08844280e+00 -6.43181264e-01 2.88082570e-01 4.99071985e-01
-7.77037740e-02 1.93064094e-01 -7.70775616e-01 -8.07285130e-01
1.33759752e-01 -8.47503781e-01 7.47215986e-01 7.10789114e-02
-6.88814282e-01 -1.16711628e+00 4.26093757e-01 5.33708096e-01
4.35218424e-01 8.50955173e-02 1.03814757e+00 -7.71806657e-01
1.09791301e-01 1.70046702e-01 -1.70442671e-01 -6.83107600e-02
2.15595767e-01 5.00773191e-01 -5.49016178e-01 -3.67761225e-01
-5.01835793e-02 -5.43066025e-01 1.92614883e-01 2.98805106e-02
6.19689882e-01 -4.86247569e-01 1.69464760e-02 3.44394264e-03
1.35426939e+00 2.14144588e-01 2.14245304e-01 5.11143744e-01
5.36451519e-01 7.77424812e-01 4.19425309e-01 8.02459717e-01
8.13833117e-01 6.23542190e-01 -4.71241385e-01 4.37820762e-01
3.41419578e-01 -2.85829455e-01 7.53866196e-01 1.55517817e+00
2.46801555e-01 -3.74619290e-02 -1.71552575e+00 6.94337547e-01
-1.59167945e+00 -6.72113299e-01 -2.04018518e-01 2.16454458e+00
1.20794487e+00 -5.75453453e-02 2.75466710e-01 -4.99243021e-01
6.41830385e-01 -3.57077196e-02 -1.26407057e-01 -6.34180784e-01
-2.15435728e-01 -3.10549378e-01 3.80917490e-01 6.76325023e-01
-6.74208939e-01 1.12007606e+00 6.42145395e+00 5.26875496e-01
-1.02994490e+00 1.84910953e-01 4.34468627e-01 -6.44765869e-02
-2.91896433e-01 2.93998737e-02 -4.90758926e-01 7.29726553e-01
1.23768687e+00 -3.69694829e-01 9.54048574e-01 4.55899894e-01
4.50215667e-01 -4.58390653e-01 -8.59258771e-01 9.51932728e-01
3.48868400e-01 -1.19448280e+00 -2.90019065e-01 2.77181007e-02
8.65061998e-01 4.01395977e-01 -9.57280621e-02 4.90696758e-01
6.17988288e-01 -8.55807722e-01 9.30256128e-01 2.94901669e-01
8.95824134e-01 -3.26919317e-01 5.42080462e-01 5.28341532e-01
-9.48172867e-01 -4.01790887e-01 -2.02685446e-01 -3.60611022e-01
1.79138649e-02 6.15982533e-01 -6.26904726e-01 2.41920009e-01
7.95263290e-01 8.93916309e-01 -6.21916354e-01 5.31500101e-01
-1.34155855e-01 8.10966671e-01 -2.79262692e-01 -3.09366047e-01
4.21864569e-01 -3.70119750e-01 3.98028493e-01 1.65141213e+00
1.53742656e-01 1.15320146e-01 1.65017545e-01 8.16603482e-01
-3.55785668e-01 3.25641543e-01 -8.65441084e-01 -6.00072205e-01
1.03032184e+00 1.00135255e+00 -6.13213003e-01 -1.90723166e-01
-9.67828393e-01 6.80467486e-01 6.80543959e-01 3.54746312e-01
-7.16329277e-01 -4.63625908e-01 5.82724810e-01 -2.21290126e-01
-1.15651630e-01 -6.35410130e-01 -5.04813865e-02 -1.34514856e+00
-1.13447368e-01 -1.26508594e+00 2.31097162e-01 -5.73917449e-01
-1.39396775e+00 2.24873111e-01 7.39331916e-02 -7.95775831e-01
-2.43992105e-01 -4.49772567e-01 -5.09278513e-02 7.01582968e-01
-1.05218244e+00 -7.87508249e-01 -8.92396867e-02 3.44681352e-01
5.34003377e-01 -4.02134478e-01 7.57132113e-01 3.23764771e-01
-7.93176949e-01 5.73773861e-01 5.83368599e-01 4.51738000e-01
7.82975018e-01 -8.13131392e-01 2.11741567e-01 6.90999985e-01
-9.76309702e-02 1.23842692e+00 3.44346076e-01 -9.31898475e-01
-1.45758569e+00 -6.93552256e-01 1.39124453e+00 -4.70824778e-01
1.22717595e+00 -7.85643756e-01 -7.39784598e-01 6.00739419e-01
5.07050991e-01 -5.46578765e-01 1.03384864e+00 3.89355004e-01
-4.32845592e-01 2.62673080e-01 -1.24169409e+00 7.84545243e-01
8.83845687e-01 -9.96523142e-01 -6.65643454e-01 2.40952045e-01
6.73587620e-01 -3.27887088e-01 -1.14770985e+00 -4.85474080e-01
7.75548220e-01 -6.36480153e-01 1.50873244e-01 -5.01424611e-01
6.79259300e-01 9.56193134e-02 -3.34468514e-01 -1.22707105e+00
-2.49908417e-01 -1.01258647e+00 4.40810502e-01 1.48467791e+00
6.92406476e-01 -7.75698364e-01 1.45997345e-01 9.23588216e-01
-1.26092225e-01 -3.21175009e-01 -9.89541948e-01 -5.35704136e-01
4.73181367e-01 -3.65890801e-01 2.54032642e-01 1.66804361e+00
7.33735502e-01 2.81207293e-01 -1.46217123e-01 -3.50414008e-01
8.61076489e-02 -1.77918762e-01 7.54403055e-01 -1.03332031e+00
-7.17284158e-02 -5.47557175e-01 1.51936278e-01 -8.36484671e-01
3.41526330e-01 -1.06990671e+00 6.23783469e-02 -1.07995594e+00
4.72761899e-01 -4.71281052e-01 4.02446777e-01 4.91305649e-01
9.33713391e-02 -6.92958087e-02 3.38978142e-01 5.23793340e-01
-5.52800715e-01 3.72516662e-01 6.84366107e-01 3.40811424e-02
-5.06827116e-01 -4.68798012e-01 -1.03306592e+00 5.63186705e-01
8.43278527e-01 -4.82763052e-01 -1.94338173e-01 -6.18728578e-01
6.65491104e-01 2.77463570e-02 -1.91909179e-01 -4.71633732e-01
1.51086837e-01 -3.42761844e-01 -2.61129320e-01 8.62822309e-02
-2.79880732e-01 -6.94877088e-01 8.73939320e-03 3.92750949e-01
-6.01662457e-01 4.79553550e-01 2.87594676e-01 1.16550259e-01
-1.28747523e-01 -3.67754877e-01 4.76274520e-01 -2.51242131e-01
-4.96022820e-01 -2.79429853e-01 -1.06149757e+00 7.69004405e-01
8.96153152e-01 -9.54867899e-02 -5.63733578e-01 -2.86705047e-01
-9.67644155e-02 2.99884140e-01 5.93218982e-01 6.49627268e-01
2.02400818e-01 -1.05062616e+00 -9.21703160e-01 1.27894029e-01
3.42869371e-01 -5.03613830e-01 -2.56640673e-01 8.11340511e-01
-8.76215577e-01 5.30683935e-01 -1.20759293e-01 -2.10454032e-01
-9.70408976e-01 1.17307991e-01 7.20828027e-02 3.74815464e-02
-3.36462438e-01 4.85481769e-01 -4.73725259e-01 -9.22942460e-01
-3.93684059e-02 -3.75343919e-01 -1.78134903e-01 3.65512818e-01
2.28627607e-01 5.23754478e-01 -3.79846454e-01 -8.83839130e-01
-4.93370295e-01 1.33811891e-01 -1.64869484e-02 -3.58808547e-01
1.23773110e+00 -4.32441115e-01 -6.87031686e-01 1.05728662e+00
1.27131808e+00 5.84709883e-01 -6.53925240e-01 -5.60902134e-02
3.34459513e-01 -5.88184893e-01 -3.59537423e-01 -7.92633712e-01
-5.11919260e-01 2.29162499e-01 -1.21479575e-02 4.48982149e-01
5.00267267e-01 7.34219775e-02 3.38118285e-01 4.79560405e-01
5.30990481e-01 -1.51870334e+00 -3.10118914e-01 8.82499635e-01
9.49733436e-01 -1.20278144e+00 9.06122178e-02 -2.12828442e-01
-9.16958153e-01 8.12462687e-01 4.51389968e-01 3.91755283e-01
8.24891627e-01 1.66078389e-01 9.46306661e-02 -1.98218659e-01
-8.02475929e-01 3.62033039e-01 -1.68417439e-01 3.38497549e-01
1.10457265e+00 3.27306181e-01 -7.88248420e-01 4.78415608e-01
-6.15573049e-01 -2.64118880e-01 8.52989435e-01 1.00461209e+00
-1.15644753e-01 -9.32645857e-01 -2.95530021e-01 7.81895697e-01
-5.48420608e-01 -4.68093932e-01 -7.43685305e-01 1.20152903e+00
-2.06244022e-01 1.39686930e+00 2.26458609e-01 -3.03374827e-01
6.15690798e-02 1.51953831e-01 -1.06982149e-01 -8.10142279e-01
-9.37448442e-01 -1.02338046e-01 4.42987591e-01 -3.48843008e-01
-6.03882194e-01 -1.17715919e+00 -1.17359948e+00 -8.00711811e-01
-2.84638941e-01 3.16391408e-01 6.20554984e-01 7.82120764e-01
4.23662663e-01 -1.12680845e-01 4.73249912e-01 -1.80625573e-01
-1.15887754e-01 -1.14057887e+00 -6.80345893e-01 2.09149584e-01
2.37369895e-01 -4.43155199e-01 -2.77907461e-01 -3.38966399e-02] | [9.437313079833984, 10.238812446594238] |
211b18ad-78dd-4cb8-afbc-299d10dfa180 | disentangling-prosody-representations-with | 2212.06972 | null | https://arxiv.org/abs/2212.06972v1 | https://arxiv.org/pdf/2212.06972v1.pdf | Disentangling Prosody Representations with Unsupervised Speech Reconstruction | Human speech can be characterized by different components, including semantic content, speaker identity and prosodic information. Significant progress has been made in disentangling representations for semantic content and speaker identity in Automatic Speech Recognition (ASR) and speaker verification tasks respectively. However, it is still an open challenging research question to extract prosodic information because of the intrinsic association of different attributes, such as timbre and rhythm, and because of the need for unsupervised training schemes to achieve robust large-scale and speaker-independent ASR. The aim of this paper is to address the disentanglement of emotional prosody from speech based on unsupervised reconstruction. Specifically, we identify, design, implement and integrate three crucial components in our proposed speech reconstruction model Prosody2Vec: (1) a unit encoder that transforms speech signals into discrete units for semantic content, (2) a pretrained speaker verification model to generate speaker identity embeddings, and (3) a trainable prosody encoder to learn prosody representations. We first pretrain the Prosody2Vec representations on unlabelled emotional speech corpora, then fine-tune the model on specific datasets to perform Speech Emotion Recognition (SER) and Emotional Voice Conversion (EVC) tasks. Both objective and subjective evaluations on the EVC task suggest that Prosody2Vec effectively captures general prosodic features that can be smoothly transferred to other emotional speech. In addition, our SER experiments on the IEMOCAP dataset reveal that the prosody features learned by Prosody2Vec are complementary and beneficial for the performance of widely used speech pretraining models and surpass the state-of-the-art methods when combining Prosody2Vec with HuBERT representations. Some audio samples can be found on our demo website. | ['Stefan Wermter', 'Fuji Ren', 'Theresa Pekarek-Rosin', 'Cornelius Weber', 'Taihao Li', 'Leyuan Qu'] | 2022-12-14 | null | null | null | null | ['voice-conversion', 'voice-conversion', 'speech-emotion-recognition', 'speaker-verification'] | ['audio', 'speech', 'speech', 'speech'] | [ 8.79211724e-02 2.60021091e-01 -9.02036130e-02 -5.52189887e-01
-7.54515886e-01 -5.00490367e-01 2.64833391e-01 -1.00959152e-01
-3.12662631e-01 5.42943954e-01 8.29053879e-01 -1.18522428e-01
2.86231548e-01 -1.91250458e-01 -3.38285565e-01 -6.29925549e-01
9.75947082e-02 2.64289677e-01 -3.69781584e-01 -4.46678340e-01
-4.15963620e-01 4.13715243e-01 -1.66653132e+00 1.28398910e-01
5.94828963e-01 1.19015110e+00 5.60169183e-02 7.00232267e-01
-2.31215253e-01 6.91703320e-01 -7.16695905e-01 -2.73777217e-01
-2.27221787e-01 -4.61086065e-01 -7.65997291e-01 1.50771499e-01
-3.19622397e-01 -1.62475684e-03 -1.62013203e-01 1.00063562e+00
7.95576274e-01 4.35586959e-01 6.77407980e-01 -9.20554698e-01
-9.44921732e-01 9.94900942e-01 -4.51291949e-02 1.14978090e-01
3.58879209e-01 -1.83517888e-01 1.26710749e+00 -1.01965249e+00
4.89860624e-01 1.16529465e+00 4.98337984e-01 8.98345232e-01
-1.30919564e+00 -7.34397471e-01 -1.42170087e-01 1.75134227e-01
-1.33114696e+00 -9.93364871e-01 1.23494208e+00 -4.51332271e-01
1.02866864e+00 3.82231921e-01 2.90867686e-01 1.70391357e+00
-4.19447064e-01 7.56970286e-01 7.78614819e-01 -3.88570011e-01
1.51587933e-01 7.47422993e-01 3.49907279e-01 1.31785899e-01
-6.39661312e-01 4.32016402e-01 -4.42310125e-01 8.49998072e-02
4.62083399e-01 -5.29129744e-01 -6.47449732e-01 1.03485845e-01
-8.01577270e-01 9.84526694e-01 2.07754180e-01 6.16083324e-01
-4.63033468e-01 -1.52658939e-01 7.61329591e-01 4.60076153e-01
4.26868170e-01 3.94779652e-01 -4.51605380e-01 -2.21358642e-01
-6.87650204e-01 -4.41466063e-01 7.44939446e-01 5.66021264e-01
4.58855897e-01 7.85574555e-01 2.56124046e-03 1.52271473e+00
2.88686335e-01 3.56413335e-01 1.02639472e+00 -5.72946310e-01
2.77485605e-02 1.38117298e-01 -2.97172457e-01 -6.25104249e-01
-1.83788881e-01 -2.40701661e-01 -7.97564030e-01 -5.21375053e-02
-2.32569888e-01 -2.94649661e-01 -5.39115965e-01 2.05543113e+00
1.47558495e-01 2.26813257e-01 6.38143420e-01 1.04339433e+00
1.34215593e+00 1.03294921e+00 1.45526990e-01 -3.12173367e-01
1.55544901e+00 -8.78779590e-01 -1.06907654e+00 -9.61235538e-02
1.67777970e-01 -6.78642094e-01 1.14507544e+00 1.21210732e-01
-1.02060223e+00 -8.90072346e-01 -1.16487718e+00 -1.23618886e-01
-4.90910441e-01 3.62320483e-01 2.05129772e-01 9.03807878e-01
-9.07921433e-01 2.19211564e-03 -5.89283586e-01 -1.18806332e-01
7.06700385e-02 2.93794572e-01 -6.60768151e-01 5.62263489e-01
-1.52321720e+00 7.81004131e-01 5.66497087e-01 -7.98082426e-02
-7.60060966e-01 -6.77210569e-01 -1.41480386e+00 3.75668138e-01
-2.08192412e-02 -1.93618670e-01 1.10163653e+00 -1.28804243e+00
-2.02908015e+00 8.37681055e-01 -6.22632578e-02 -5.02840221e-01
-1.81334361e-01 1.04928091e-01 -9.55442905e-01 1.62664428e-01
-3.63981456e-01 7.02294230e-01 8.64722848e-01 -1.23451519e+00
-1.61324158e-01 -1.70474932e-01 -5.58781445e-01 2.28903890e-01
-5.99457681e-01 5.15280902e-01 -1.58580482e-01 -7.70274043e-01
-2.86477298e-01 -6.33422732e-01 1.93480894e-01 -6.07223034e-01
-3.37239981e-01 -2.80091047e-01 6.89755261e-01 -1.06117141e+00
1.16398537e+00 -2.55043197e+00 3.86275113e-01 -1.06265716e-01
-2.70411849e-01 3.36929232e-01 -2.95341522e-01 1.42320961e-01
-5.67607760e-01 1.58574015e-01 -1.44923478e-01 -6.58869803e-01
2.87406743e-01 1.83907852e-01 -6.22988284e-01 1.76455230e-01
5.92075944e-01 7.70896375e-01 -4.54465240e-01 -1.83546722e-01
3.66806626e-01 1.09981382e+00 -4.48519826e-01 3.77211303e-01
8.46692473e-02 5.56116581e-01 -2.20793225e-02 4.67057496e-01
3.58384371e-01 4.69876736e-01 1.82415292e-01 -1.09281145e-01
-1.60881542e-02 8.36533129e-01 -1.07948601e+00 1.32823813e+00
-7.22194195e-01 6.82307601e-01 5.39038599e-01 -1.06822455e+00
1.32715535e+00 1.02841902e+00 3.08837175e-01 -5.05852282e-01
5.45586228e-01 1.10701181e-01 -7.17668086e-02 -3.34963351e-01
5.81464529e-01 -6.73074663e-01 -4.04204011e-01 1.04632951e-01
6.51268184e-01 -6.60093352e-02 -3.80854428e-01 -3.67593795e-01
5.78276217e-01 -4.36278850e-01 3.91552091e-01 -9.96529609e-02
9.81715381e-01 -5.05266547e-01 7.24460661e-01 -8.62427726e-02
-4.44317251e-01 6.28187001e-01 1.51522771e-01 1.18441142e-01
-7.21312106e-01 -1.12925124e+00 -3.20950300e-01 1.29066837e+00
-1.75984949e-01 -1.66405156e-01 -6.72442317e-01 -4.25791770e-01
-2.69792050e-01 1.12370002e+00 -4.56280231e-01 -1.64600044e-01
-4.28869009e-01 -3.29079777e-01 8.98674428e-01 6.81076705e-01
-7.48884827e-02 -1.55543029e+00 -1.32418731e-02 2.62899786e-01
-2.43215382e-01 -1.28395939e+00 -6.85531497e-01 4.83540773e-01
-1.77377939e-01 -5.65231860e-01 -4.49491471e-01 -1.14126706e+00
8.18580016e-02 -2.49393687e-01 7.73765624e-01 -3.60597730e-01
-7.23812357e-02 4.58237559e-01 -5.61369956e-01 -3.68238598e-01
-8.68346810e-01 -1.14623763e-01 3.97162259e-01 3.81806731e-01
3.91987950e-01 -6.33453429e-01 -3.69576877e-03 3.48045886e-01
-8.04034710e-01 -3.72931510e-01 1.40711784e-01 1.05017567e+00
6.74330533e-01 1.88745260e-01 1.20419693e+00 -4.85591352e-01
7.96399176e-01 -3.65508378e-01 -2.39179552e-01 -1.01112891e-02
-7.31570050e-02 -1.53322563e-01 7.20687270e-01 -5.75357378e-01
-1.34612930e+00 -6.76611289e-02 -8.98327708e-01 -7.21819282e-01
-5.12394905e-01 4.90490228e-01 -6.25801504e-01 4.93522912e-01
4.23064321e-01 3.99492472e-01 7.81887993e-02 -3.93570065e-01
5.26451707e-01 1.26099408e+00 6.94519460e-01 -4.65858012e-01
4.27306175e-01 -4.92080823e-02 -9.31378245e-01 -1.63459003e+00
-4.11542207e-01 -6.00812197e-01 -3.32283288e-01 6.16334826e-02
1.18279660e+00 -1.05025136e+00 -6.88927114e-01 8.35047588e-02
-1.20092583e+00 -1.45952806e-01 -5.12453973e-01 7.01509237e-01
-6.70418680e-01 3.13173920e-01 -8.25103581e-01 -1.12695968e+00
-5.66980898e-01 -1.23643780e+00 8.74661922e-01 8.74310508e-02
-5.27903914e-01 -1.02846134e+00 1.90078869e-01 6.70969188e-01
4.60948497e-01 -1.81719318e-01 7.81837285e-01 -1.09719944e+00
7.66961798e-02 -4.28961627e-02 8.25793371e-02 1.01382601e+00
2.27019995e-01 -1.44694075e-01 -1.56742442e+00 -5.80868609e-02
3.78878325e-01 -5.67762196e-01 7.98639596e-01 3.03500682e-01
8.69342327e-01 -4.18197900e-01 2.29304120e-01 6.32672310e-01
8.96205485e-01 4.17834759e-01 5.66171288e-01 -2.89135754e-01
4.28900093e-01 8.90662551e-01 1.50070742e-01 1.51783958e-01
2.07968265e-01 7.11693704e-01 -1.27038239e-02 7.00706318e-02
-2.89216995e-01 -2.89268345e-01 9.11990106e-01 1.58867633e+00
2.14280754e-01 -9.52406377e-02 -5.95345855e-01 6.16668105e-01
-1.37358415e+00 -1.04792309e+00 4.89745378e-01 1.97911930e+00
1.01769137e+00 -1.57638028e-01 6.99792281e-02 3.19964141e-01
7.48793483e-01 3.21747094e-01 -3.23732078e-01 -7.55076349e-01
-3.16296160e-01 3.92602235e-01 -2.09011585e-01 6.77871346e-01
-1.16174793e+00 1.14492059e+00 5.29227400e+00 7.47455478e-01
-1.36057258e+00 2.75928825e-01 4.35334653e-01 9.18587074e-02
-3.88992310e-01 -5.04094899e-01 -6.75575733e-01 2.51204908e-01
1.33759737e+00 -1.40581563e-01 5.09460032e-01 1.02965927e+00
1.25768021e-01 8.23726058e-01 -1.17503870e+00 1.20090270e+00
3.48542124e-01 -9.68488693e-01 -2.03957036e-01 -1.21320933e-01
2.64226377e-01 2.60834694e-02 1.23169616e-01 8.35427761e-01
4.61465567e-02 -1.23314404e+00 6.34137809e-01 3.74436602e-02
9.41770017e-01 -8.31474900e-01 8.40409219e-01 1.52156219e-01
-1.19262135e+00 -2.44221743e-03 -2.08892867e-01 3.33433360e-01
3.59090388e-01 2.95481849e-02 -1.07105756e+00 4.26732719e-01
4.97769624e-01 4.95024860e-01 1.44085407e-01 3.35905403e-01
-3.80526781e-01 8.88289869e-01 -7.96761811e-02 3.79280075e-02
-7.53645599e-02 -1.21625833e-01 6.90835774e-01 1.37368524e+00
2.13415086e-01 2.91518159e-02 -1.06602132e-01 1.12444556e+00
-2.70076007e-01 4.92706656e-01 -4.40270692e-01 -6.61751568e-01
5.99582136e-01 1.17442930e+00 -1.97387382e-01 -1.51789412e-01
-3.70385796e-01 8.83954823e-01 2.61027157e-01 4.17862296e-01
-8.13156664e-01 -3.39359075e-01 1.26686943e+00 -3.94670784e-01
4.79887336e-01 -1.00511581e-01 -1.27010688e-01 -1.18278122e+00
-3.12352717e-01 -8.78921151e-01 1.53551295e-01 -6.98810637e-01
-1.45991814e+00 1.17410994e+00 -5.43555081e-01 -8.88469636e-01
-4.55582380e-01 -5.59634149e-01 -7.72233605e-01 1.03677189e+00
-1.64273071e+00 -1.07168102e+00 9.50876027e-02 5.94709396e-01
9.97905612e-01 -4.87361073e-01 1.27917516e+00 2.70925999e-01
-7.77569473e-01 8.33224356e-01 -1.88520283e-01 4.30943370e-01
5.20552039e-01 -1.10324359e+00 3.92103344e-02 4.62536246e-01
3.47304255e-01 3.89415264e-01 4.97832119e-01 -1.06921032e-01
-1.06670070e+00 -9.87136781e-01 9.13588583e-01 -2.80986816e-01
7.70216823e-01 -6.49530709e-01 -1.22574890e+00 6.69600368e-01
4.48834240e-01 -2.17661783e-01 1.30700791e+00 3.10348541e-01
-4.17816937e-01 -1.80281531e-02 -9.26599801e-01 4.25937206e-01
4.10162568e-01 -9.63583052e-01 -1.14635813e+00 -3.31360728e-01
1.21058798e+00 9.01536644e-02 -9.28985357e-01 2.49392286e-01
4.06557620e-01 -7.76167274e-01 9.68427122e-01 -5.60769379e-01
1.81674004e-01 3.32190208e-02 -7.04237342e-01 -1.43805528e+00
1.73516721e-02 -5.18678010e-01 1.93513393e-01 1.87274384e+00
5.65342724e-01 -5.89429200e-01 3.01012605e-01 3.04925889e-01
-4.25391287e-01 -3.56437981e-01 -1.00114512e+00 -7.06946552e-01
5.12157530e-02 -7.62756348e-01 5.81028044e-01 1.18040609e+00
3.59503895e-01 9.39032793e-01 -2.90173113e-01 3.42376262e-01
1.04961112e-01 -9.81372297e-02 4.71413970e-01 -1.19132686e+00
-4.14581656e-01 -6.71537459e-01 -4.62888181e-01 -8.16914201e-01
9.21149135e-01 -1.17038226e+00 2.55036443e-01 -1.01490784e+00
-2.63504475e-01 -1.21263348e-01 -6.16802454e-01 3.74444902e-01
8.24682415e-02 -1.67374641e-01 5.25630005e-02 -2.00098768e-01
5.43636344e-02 1.27286685e+00 6.53469086e-01 -2.52674669e-01
-6.00179851e-01 -7.75934756e-02 -6.86311662e-01 6.33581519e-01
7.83644199e-01 -1.09342389e-01 -4.53878105e-01 2.66878083e-02
-4.13913816e-01 6.04371190e-01 1.22416681e-02 -6.75081432e-01
6.15139455e-02 1.98897570e-01 5.93077950e-02 -3.19067121e-01
9.37027156e-01 -8.64430070e-01 -1.20072067e-01 -1.34449422e-01
-5.21878302e-01 -4.98297423e-01 4.21595156e-01 3.49532396e-01
-7.91457176e-01 -2.65632033e-01 8.58546615e-01 2.86046863e-01
-6.53555632e-01 -2.33246107e-02 -5.27761936e-01 -6.74210340e-02
6.43902421e-01 -1.13091767e-01 -6.25774339e-02 -5.03102183e-01
-1.04775131e+00 -5.48358560e-02 1.29812341e-02 7.94953465e-01
7.37890124e-01 -1.40524662e+00 -7.40315259e-01 7.39068925e-01
5.55772297e-02 -4.96355593e-01 4.79174107e-01 6.47537231e-01
1.15015239e-01 3.68704975e-01 -1.52215108e-01 -3.41803551e-01
-1.41633844e+00 6.71575427e-01 3.24097902e-01 2.15862647e-01
-2.74319082e-01 9.02564943e-01 4.63046163e-01 -8.20806921e-01
4.87193942e-01 -2.53527135e-01 -4.68219846e-01 4.32031661e-01
4.97790188e-01 -4.93980153e-03 -7.99587071e-02 -1.28849721e+00
-4.69920397e-01 2.79482454e-01 1.54221922e-01 -2.62374192e-01
1.45523596e+00 -2.70473123e-01 8.92734900e-02 5.89614093e-01
1.50387263e+00 3.27359110e-01 -7.71958530e-01 -1.45171314e-01
-1.48138687e-01 3.41760367e-01 2.73414046e-01 -7.28660524e-01
-9.58556592e-01 1.15103114e+00 4.23451215e-01 2.99938798e-01
1.12933052e+00 2.01020509e-01 1.00842583e+00 -2.13583633e-02
-9.02175158e-02 -1.21571648e+00 -2.51289122e-02 6.88530385e-01
1.15827692e+00 -1.20313871e+00 -9.04182374e-01 -4.12271172e-01
-1.15308022e+00 1.00610256e+00 3.58736336e-01 1.07261352e-01
8.95158291e-01 3.20579827e-01 4.00553465e-01 6.30898178e-02
-7.32615888e-01 -3.80902261e-01 4.81531948e-01 6.22210026e-01
6.60224676e-01 3.33222926e-01 9.87755582e-02 1.55835247e+00
-6.90825641e-01 -5.19006848e-01 2.48304546e-01 1.14788398e-01
-4.40151185e-01 -1.05383837e+00 -2.95107007e-01 -3.00185531e-01
-4.08463567e-01 -7.43492246e-02 -4.98916328e-01 4.68447447e-01
-1.57919213e-01 1.09576714e+00 2.19022647e-01 -5.57267129e-01
4.48850840e-01 6.79786086e-01 -9.70398933e-02 -7.24605858e-01
-5.92327535e-01 4.36553687e-01 4.45981085e-01 -1.78966984e-01
-1.98030278e-01 -4.03595835e-01 -1.35083139e+00 3.78341585e-01
-3.19185972e-01 3.58117104e-01 8.90693903e-01 7.24754572e-01
2.50768691e-01 8.31973374e-01 7.97468066e-01 -7.85347700e-01
-5.95820189e-01 -1.16563666e+00 -7.60107994e-01 5.18444955e-01
4.51675951e-01 -5.16455889e-01 -6.72849298e-01 1.20687641e-01] | [14.007193565368652, 6.075856685638428] |
49a6bf0c-d148-4f7c-a8dd-7968a8ed3820 | complete-end-to-end-low-cost-solution-to-a-3d | 1709.02247 | null | http://arxiv.org/abs/1709.02247v1 | http://arxiv.org/pdf/1709.02247v1.pdf | Complete End-To-End Low Cost Solution To a 3D Scanning System with Integrated Turntable | 3D reconstruction is a technique used in computer vision which has a wide
range of applications in areas like object recognition, city modelling, virtual
reality, physical simulations, video games and special effects. Previously, to
perform a 3D reconstruction, specialized hardwares were required. Such systems
were often very expensive and was only available for industrial or research
purpose. With the rise of the availability of high-quality low cost 3D sensors,
it is now possible to design inexpensive complete 3D scanning systems. The
objective of this work was to design an acquisition and processing system that
can perform 3D scanning and reconstruction of objects seamlessly. In addition,
the goal of this work also included making the 3D scanning process fully
automated by building and integrating a turntable alongside the software. This
means the user can perform a full 3D scan only by a press of a few buttons from
our dedicated graphical user interface. Three main steps were followed to go
from acquisition of point clouds to the finished reconstructed 3D model. First,
our system acquires point cloud data of a person/object using inexpensive
camera sensor. Second, align and convert the acquired point cloud data into a
watertight mesh of good quality. Third, export the reconstructed model to a 3D
printer to obtain a proper 3D print of the model. | ['Usama Pervaiz', 'Vu Hoang Minh', 'Yeman Brhane Hagos', 'Tajwar Abrar Aleef', 'Saed Khawaldeh'] | 2017-09-03 | null | null | null | null | ['physical-simulations'] | ['miscellaneous'] | [ 1.10126689e-01 -3.15762699e-01 5.82663953e-01 -3.69711161e-01
-2.09449362e-02 -3.42786580e-01 5.48457801e-01 9.73893926e-02
-3.58433843e-01 8.91120732e-02 -5.62722445e-01 -4.48850125e-01
3.30232945e-03 -1.03318655e+00 -4.28397387e-01 -6.80242255e-02
1.61757991e-01 9.87389266e-01 6.70343399e-01 -2.70214468e-01
4.17886764e-01 1.28575170e+00 -1.99803340e+00 -2.78720438e-01
5.99653363e-01 9.90252256e-01 7.00505316e-01 6.59960270e-01
-5.45100868e-01 -2.13858128e-01 -2.40464211e-01 2.01333240e-02
5.35153508e-01 -1.11747950e-01 -2.60731250e-01 4.88645673e-01
3.49576622e-02 -4.85155165e-01 3.71039629e-01 8.52086723e-01
3.22190791e-01 4.73648123e-02 4.38214362e-01 -9.19214547e-01
8.86609033e-02 -3.20055127e-01 -5.93864858e-01 -4.95537817e-01
8.11595559e-01 -6.17422685e-02 1.78250581e-01 -1.00700760e+00
7.14613855e-01 1.11541808e+00 6.33448362e-01 2.10257351e-01
-1.04613769e+00 -5.48023343e-01 -4.76123452e-01 1.22599704e-02
-1.36169291e+00 -1.35475382e-01 9.09402907e-01 -5.23814261e-01
7.66535103e-01 4.68157917e-01 1.18473351e+00 5.37514210e-01
-5.54482304e-02 4.11573425e-02 1.25483775e+00 -8.25542986e-01
2.81108946e-01 6.04099631e-01 8.40431005e-02 4.69557494e-01
1.99323520e-01 -3.29613276e-02 -1.08664958e-02 2.20359936e-02
1.38453150e+00 3.61473918e-01 -1.21485882e-01 -5.39250195e-01
-8.16431999e-01 4.44364667e-01 1.83879137e-01 4.29987520e-01
-5.36012232e-01 -1.49923325e-01 -1.16698064e-01 1.22687131e-01
1.92125931e-01 -1.01369105e-01 -3.37045372e-01 -2.62070656e-01
-6.74478471e-01 4.44780514e-02 5.91629088e-01 9.23270702e-01
8.19588184e-01 -1.63062781e-01 8.03686857e-01 5.55721104e-01
6.30644798e-01 7.19698310e-01 1.20253645e-01 -7.27515936e-01
3.06328565e-01 9.29397523e-01 2.38402098e-01 -9.33405519e-01
-4.06712681e-01 5.88974170e-02 -6.34316087e-01 9.72549260e-01
2.13350981e-01 3.35145831e-01 -7.59376466e-01 6.47067249e-01
8.94598424e-01 -7.11653456e-02 -3.46266598e-01 1.01336765e+00
7.46854365e-01 6.46441400e-01 -4.27698761e-01 -9.76252835e-03
1.39539087e+00 -6.95292428e-02 -4.87988710e-01 8.13595727e-02
3.22841108e-01 -1.12565446e+00 1.23299181e+00 6.34439886e-01
-9.08746123e-01 -7.33349919e-01 -9.37801361e-01 -5.40523380e-02
-3.73077691e-01 -1.67506412e-02 3.45344633e-01 6.93351388e-01
-7.25881040e-01 6.35607243e-01 -8.00357580e-01 -6.08428359e-01
7.67033026e-02 3.00083846e-01 -5.51300883e-01 -9.69250035e-03
-6.01886690e-01 1.18802381e+00 1.21852569e-01 1.44078761e-01
-1.42771274e-01 -3.42460662e-01 -4.67341602e-01 -2.38987848e-01
1.71292379e-01 -7.97261417e-01 9.72353756e-01 -7.59778142e-01
-1.82705331e+00 1.14708292e+00 6.60506710e-02 4.64876741e-02
6.72146082e-01 -1.71146914e-01 -2.80500144e-01 -3.04988027e-02
-2.14210264e-02 1.00941181e-01 7.00218856e-01 -1.24104333e+00
-3.60816121e-01 -7.67107904e-01 -1.35678545e-01 2.31462538e-01
1.62045255e-01 -8.52107443e-03 -6.48747563e-01 -8.47540889e-03
5.65387785e-01 -6.66559279e-01 -2.10800856e-01 3.43180537e-01
-1.36146724e-01 3.78408022e-02 1.16776156e+00 -6.59832895e-01
4.67599809e-01 -2.05678773e+00 -1.14780508e-01 4.77771133e-01
-2.32397765e-01 3.69502008e-01 4.99950618e-01 4.94260550e-01
-3.24474163e-02 -2.01942682e-01 -8.75024125e-02 -5.11800468e-01
-1.85149625e-01 3.15234400e-02 1.24890782e-01 4.50601250e-01
-2.83477962e-01 2.62267828e-01 -2.00264335e-01 -4.99626696e-01
8.57928753e-01 8.45501781e-01 -1.79221645e-01 9.51355472e-02
-6.43322021e-02 5.99950373e-01 -6.01907253e-01 8.41000080e-01
8.91773462e-01 1.85418859e-01 -1.44763201e-01 7.74274841e-02
-6.42311335e-01 1.57771166e-02 -1.83459759e+00 1.63089013e+00
-5.07084489e-01 3.65759432e-01 4.06417817e-01 -7.19615936e-01
1.31806445e+00 4.49366122e-01 4.74702775e-01 -7.42823184e-01
4.15587574e-01 3.53276938e-01 -5.74769318e-01 -5.86926460e-01
5.41102648e-01 -2.57444471e-01 2.27545694e-01 4.68561381e-01
-5.57042897e-01 -8.06119442e-01 -4.31203812e-01 -1.94553062e-01
4.21857715e-01 4.79053766e-01 3.76358777e-01 1.26311302e-01
5.57475448e-01 3.41817230e-01 3.07072103e-02 2.32257709e-01
3.49732786e-01 6.66579127e-01 -3.27052176e-02 -4.84464616e-01
-1.40033519e+00 -7.94051349e-01 -2.84140974e-01 2.06425592e-01
2.52168089e-01 3.00893764e-04 -6.02084637e-01 5.36994971e-02
-3.18617411e-02 4.99283373e-01 -1.48773074e-01 4.82952356e-01
-5.48952758e-01 -8.36366266e-02 -1.70404837e-01 1.02004059e-01
5.92695773e-01 -8.09965074e-01 -1.28092813e+00 1.77973479e-01
4.59931701e-01 -7.69850373e-01 1.38382539e-01 -2.25035995e-01
-1.40552568e+00 -1.02065444e+00 -5.60796380e-01 -5.84342420e-01
7.73714364e-01 4.61778373e-01 8.02801132e-01 4.06794429e-01
-4.37953711e-01 4.96623456e-01 -4.99522388e-01 -7.59981453e-01
-2.81917602e-01 -2.63341010e-01 7.95724541e-02 -2.36629680e-01
9.12935808e-02 -7.88059950e-01 -4.01178390e-01 3.77557844e-01
-7.98254848e-01 3.50259990e-01 3.75779122e-01 -9.22290329e-03
8.76396358e-01 1.55224547e-01 -3.07880866e-04 -5.08972645e-01
3.46079946e-01 5.30161262e-02 -1.18310869e+00 -1.96670070e-01
-3.19729656e-01 -4.14550155e-01 4.90200132e-01 -1.57019183e-01
-7.55924046e-01 4.22217280e-01 -5.53640425e-01 -4.58714038e-01
-5.51875770e-01 2.04518184e-01 -3.27580661e-01 -7.90233910e-03
5.36172211e-01 2.11651072e-01 5.18369138e-01 -9.05792713e-01
1.33504700e-02 1.01268578e+00 3.14388275e-01 7.47662783e-02
1.03197551e+00 6.29831016e-01 3.33288282e-01 -1.28501832e+00
1.87959522e-02 -5.50993621e-01 -1.02330947e+00 -6.55795693e-01
7.83799231e-01 -4.09349501e-01 -7.33802915e-01 4.71748620e-01
-1.21162236e+00 -4.97056432e-02 -2.15807885e-01 6.37459040e-01
-3.10347915e-01 3.11959118e-01 4.96995859e-02 -9.12311971e-01
-2.48017177e-01 -1.14384449e+00 8.94865334e-01 3.38203073e-01
-9.88951847e-02 -7.07098544e-01 1.27778957e-02 2.97343463e-01
1.97159886e-01 4.14715201e-01 4.30626601e-01 1.94012955e-01
-8.30536127e-01 -6.65144265e-01 4.97270115e-02 3.09256930e-02
1.32887959e-01 3.77421021e-01 -7.51401424e-01 2.00084493e-01
1.52559429e-01 2.67709225e-01 1.14678510e-01 2.26617470e-01
7.69556463e-01 3.77744734e-01 -5.27070582e-01 5.91079831e-01
1.73437643e+00 3.97910476e-01 7.87598431e-01 6.07663333e-01
4.87488717e-01 5.77949882e-01 8.21286857e-01 3.05044621e-01
2.23274261e-01 1.08143711e+00 6.52209699e-01 2.80450340e-02
4.47196811e-02 -1.60718963e-01 1.85932443e-02 6.57860219e-01
-5.87778389e-01 4.51475888e-01 -1.02631688e+00 -5.68133816e-02
-1.32235849e+00 -8.08681726e-01 -8.94749880e-01 2.66579843e+00
2.09067106e-01 1.32378280e-01 1.81926101e-01 5.76420665e-01
4.73466337e-01 -5.31524062e-01 -1.43732324e-01 -6.14564240e-01
4.66648072e-01 3.48011583e-01 5.31571448e-01 6.30066752e-01
-6.84933722e-01 5.48076332e-01 5.34528589e+00 2.40472928e-01
-1.33249021e+00 1.21761141e-02 -3.28439802e-01 5.98716289e-02
-2.73132086e-01 8.66356269e-02 -7.60719121e-01 3.53544235e-01
5.76273143e-01 1.23355664e-01 2.65708506e-01 1.03753006e+00
6.41651690e-01 -6.57909930e-01 -6.96780026e-01 1.21676064e+00
-1.76251873e-01 -1.18261266e+00 -1.59665391e-01 3.18234593e-01
2.44690612e-01 -1.72804564e-01 -3.47303271e-01 -1.81065425e-01
-2.45811284e-01 -8.28409612e-01 8.63822639e-01 7.27246821e-01
7.69540310e-01 -6.05712712e-01 5.38226128e-01 9.30931568e-01
-1.04585433e+00 3.80040735e-01 -3.54230732e-01 -3.44032049e-01
5.78800619e-01 6.99950695e-01 -1.15347648e+00 5.63917816e-01
7.65339553e-01 2.77971290e-02 -2.04954311e-01 1.31209457e+00
1.14521123e-01 3.76369841e-02 -7.05046058e-01 -1.79522350e-01
-1.78202838e-01 -7.54982650e-01 5.32839596e-01 7.90406764e-01
6.28691256e-01 4.08365071e-01 -1.40298218e-01 7.30901957e-01
4.13782805e-01 1.81281403e-01 -8.21703076e-01 4.11474705e-01
3.18153143e-01 1.11287248e+00 -1.00886285e+00 -1.11612692e-01
-2.61086702e-01 9.56606388e-01 -1.65607512e-01 -2.20332623e-01
-7.06778884e-01 -5.24510264e-01 2.40608633e-01 7.18347907e-01
2.23356709e-01 -8.47837389e-01 -5.85576475e-01 -7.79726207e-01
1.04639538e-01 -3.84280592e-01 -2.38408819e-01 -1.14126647e+00
-4.89456803e-01 5.23347318e-01 1.19867176e-01 -1.51434696e+00
-2.23152459e-01 -5.15143812e-01 -4.91580963e-01 1.18471742e+00
-1.04069030e+00 -9.89681125e-01 -6.94160938e-01 5.82174003e-01
5.23299575e-01 1.20328367e-01 9.85881507e-01 3.63808751e-01
-2.20803127e-01 -3.84855598e-01 -1.38083892e-02 -3.22829694e-01
2.43205264e-01 -9.49888110e-01 2.48067006e-01 6.07271731e-01
4.49546687e-02 3.70484263e-01 7.57022500e-01 -8.29014659e-01
-1.59406912e+00 -4.29700881e-01 7.88625956e-01 -3.42780620e-01
1.88509040e-02 -4.26921010e-01 -8.42208982e-01 4.82838780e-01
-2.39342138e-01 -1.20779641e-01 2.07842767e-01 -2.67435819e-01
2.71032095e-01 -2.03704655e-01 -1.42455494e+00 4.07079488e-01
7.66473472e-01 -2.36154512e-01 -6.20078504e-01 5.79728745e-02
1.61248147e-01 -6.23517036e-01 -6.65333211e-01 -6.96853772e-02
8.00242186e-01 -1.34462249e+00 9.71970737e-01 2.27120176e-01
-1.10799614e-02 -5.92560768e-01 2.75516491e-02 -9.82475460e-01
-3.91887315e-02 -3.46765041e-01 4.46797282e-01 8.61610234e-01
1.09114520e-01 -7.66075671e-01 7.87564635e-01 8.55185390e-01
-1.46902308e-01 -2.95444489e-01 -9.81658340e-01 -4.44188297e-01
-5.42775869e-01 -8.09180737e-01 7.39087343e-01 6.63518310e-01
-3.00569206e-01 -6.12961687e-03 3.52336979e-03 4.50642407e-01
7.11174190e-01 3.00326973e-01 1.21230769e+00 -1.82801914e+00
3.90806757e-02 -2.23658249e-01 -5.96690238e-01 -7.70462930e-01
-4.72899228e-01 -5.73437274e-01 -4.11417454e-01 -2.00246644e+00
-3.38143259e-01 -8.33224475e-01 7.60915101e-01 -1.04539193e-01
5.53132176e-01 1.47708356e-01 2.21950710e-01 4.87301081e-01
1.72413513e-01 5.47010973e-02 1.29202914e+00 5.60103953e-01
-5.52733898e-01 4.18850839e-01 -1.62144750e-01 8.51622462e-01
7.20461071e-01 -3.52610528e-01 -2.39082173e-01 -4.63165879e-01
1.55799091e-01 7.86206871e-02 5.15558064e-01 -9.89612877e-01
1.73771709e-01 -8.05766732e-02 6.00444496e-01 -1.12820399e+00
7.90058792e-01 -1.66180003e+00 9.44368303e-01 4.34646368e-01
6.56020045e-01 -4.73669805e-02 -3.22405659e-02 9.48587283e-02
1.44764155e-01 -4.94145781e-01 7.60170102e-01 -5.24642825e-01
-5.46648979e-01 1.21571505e-02 -2.22706959e-01 -8.82116020e-01
1.45191169e+00 -1.09957087e+00 3.55330676e-01 -1.61430299e-01
-7.21629202e-01 -1.94708362e-01 9.46903050e-01 2.31695548e-02
7.92755365e-01 -1.20247769e+00 -2.08423004e-01 6.13575280e-01
-3.82767797e-01 5.96234918e-01 2.84694612e-01 5.84774733e-01
-1.22827518e+00 5.15422404e-01 -3.99702489e-01 -8.21388483e-01
-1.63061202e+00 3.03069532e-01 3.26118141e-01 2.94251680e-01
-1.08978760e+00 3.08658153e-01 -7.02711463e-01 -5.01510143e-01
-2.27266457e-02 -5.69045365e-01 -3.89764369e-01 -6.15304150e-02
5.51915169e-01 5.58578730e-01 3.38716924e-01 -7.56347775e-01
-3.86877120e-01 1.25353265e+00 5.85671544e-01 -4.51626003e-01
1.63250709e+00 -1.34259194e-01 9.62868631e-02 4.50211465e-01
8.65233541e-01 1.47940174e-01 -9.92835999e-01 2.30458215e-01
-3.20290148e-01 -8.51841569e-01 1.33376554e-01 -3.52743328e-01
-6.84357166e-01 8.17544580e-01 6.63974762e-01 5.12666404e-01
1.04093289e+00 1.15802675e-01 4.73874867e-01 4.03522663e-02
9.26902950e-01 -1.05447698e+00 -4.32758063e-01 3.47432286e-01
8.60787868e-01 -8.84124219e-01 2.61581540e-01 -6.29419148e-01
-4.13677245e-01 1.28209126e+00 1.52882069e-01 -6.46398067e-02
7.68019855e-01 3.71740669e-01 -1.24426177e-02 -4.60872293e-01
-6.75419345e-02 -2.41102293e-01 3.93192805e-02 8.66207480e-01
8.72558877e-02 1.51276499e-01 -3.78467113e-01 2.27801308e-01
-3.43187392e-01 4.51227516e-01 5.86086988e-01 1.14987373e+00
-8.11093152e-01 -1.25082517e+00 -1.13531494e+00 2.74998307e-01
-1.09175012e-01 5.67913473e-01 -1.78270325e-01 1.01843262e+00
4.11681682e-02 5.38340151e-01 3.61603379e-01 -1.79519221e-01
9.41099405e-01 1.18761677e-02 5.28070331e-01 -4.98563468e-01
-4.29176033e-01 1.78446785e-01 -1.11224763e-01 -5.16493320e-01
-2.98939645e-01 -7.69829452e-01 -1.27090096e+00 -3.39071453e-01
-2.96274871e-01 -1.20426528e-01 1.61241794e+00 5.68719208e-01
2.46790633e-01 -5.91653399e-02 6.32407784e-01 -1.22136223e+00
-2.20949724e-01 -8.31929743e-01 -8.01118910e-01 1.81456074e-01
-1.38512254e-01 -7.24394858e-01 -1.79451630e-01 3.06665394e-02] | [8.334582328796387, -2.7179653644561768] |
a587bca7-29b2-46e7-b4d2-677d776583f3 | auxiliary-interference-speaker-loss-for | 1906.10876 | null | https://arxiv.org/abs/1906.10876v1 | https://arxiv.org/pdf/1906.10876v1.pdf | Auxiliary Interference Speaker Loss for Target-Speaker Speech Recognition | In this paper, we propose a novel auxiliary loss function for target-speaker automatic speech recognition (ASR). Our method automatically extracts and transcribes target speaker's utterances from a monaural mixture of multiple speakers speech given a short sample of the target speaker. The proposed auxiliary loss function attempts to additionally maximize interference speaker ASR accuracy during training. This will regularize the network to achieve a better representation for speaker separation, thus achieving better accuracy on the target-speaker ASR. We evaluated our proposed method using two-speaker-mixed speech in various signal-to-interference-ratio conditions. We first built a strong target-speaker ASR baseline based on the state-of-the-art lattice-free maximum mutual information. This baseline achieved a word error rate (WER) of 18.06% on the test set while a normal ASR trained with clean data produced a completely corrupted result (WER of 84.71%). Then, our proposed loss further reduced the WER by 6.6% relative to this strong baseline, achieving a WER of 16.87%. In addition to the accuracy improvement, we also showed that the auxiliary output branch for the proposed loss can even be used for a secondary ASR for interference speakers' speech. | ['Ryoichi Takashima', 'Shota Horiguchi', 'Kenji Nagamatsu', 'Naoyuki Kanda', 'Yusuke Fujita', 'Shinji Watanabe'] | 2019-06-26 | null | null | null | null | ['speaker-separation'] | ['speech'] | [ 4.11259562e-01 4.12959576e-01 3.24679792e-01 -3.14546406e-01
-1.75621736e+00 -2.32483968e-01 3.32345843e-01 -1.56488523e-01
-4.43986028e-01 5.06764770e-01 2.83948123e-01 -3.19423586e-01
3.52265328e-01 -1.03989705e-01 -5.74268043e-01 -9.96817231e-01
7.92006403e-02 3.17241400e-01 -5.79099096e-02 -1.73860341e-01
-2.79427409e-01 4.00518298e-01 -1.62911820e+00 3.08862656e-01
7.84787238e-01 9.26771462e-01 4.44132090e-01 9.13362086e-01
2.31632203e-01 4.80320156e-01 -1.32050443e+00 -1.04418665e-01
2.05634058e-01 -5.46364665e-01 -3.84628713e-01 6.67485446e-02
4.48131174e-01 9.26251262e-02 -5.10315657e-01 9.78564024e-01
9.74845409e-01 3.52702975e-01 4.80028808e-01 -7.55075037e-01
-1.09231010e-01 1.11292791e+00 -3.95590544e-01 1.59672081e-01
3.03235471e-01 4.21453975e-02 1.01205492e+00 -8.92508268e-01
-2.02129051e-01 1.35381806e+00 1.86936617e-01 8.88694465e-01
-1.38193321e+00 -1.06948352e+00 2.56167017e-02 -2.83773504e-02
-1.68886292e+00 -1.23335326e+00 8.86092305e-01 7.76637811e-03
1.06015837e+00 6.41375363e-01 2.38784075e-01 1.29790783e+00
-4.48742539e-01 9.03785348e-01 9.88739371e-01 -6.29899621e-01
4.34029959e-02 2.86626160e-01 2.40582675e-01 2.20709980e-01
-2.40018606e-01 3.18026096e-01 -7.09448278e-01 -4.13796082e-02
1.91079199e-01 -8.40112686e-01 -5.85732460e-01 4.86150324e-01
-9.33161139e-01 5.14903188e-01 6.06392473e-02 6.17869973e-01
-1.66815490e-01 -1.72820985e-02 1.06506765e-01 3.19974184e-01
5.36674619e-01 2.81103969e-01 -3.01390231e-01 -1.76337153e-01
-1.17377603e+00 -1.72479376e-02 8.96859050e-01 5.77632427e-01
3.82013619e-01 7.44878173e-01 -3.55853558e-01 1.46373820e+00
6.49188757e-01 1.00203121e+00 4.86403883e-01 -6.09024107e-01
5.42905927e-01 -2.53295124e-01 -1.47391796e-01 -3.94762754e-01
-4.91345115e-02 -1.14149594e+00 -7.95897484e-01 7.21673444e-02
3.93580556e-01 -1.78439990e-01 -1.08034241e+00 2.03550911e+00
7.13855475e-02 3.18413585e-01 3.90356034e-01 9.89818931e-01
6.28878474e-01 1.01734877e+00 -3.58476579e-01 -6.52980447e-01
1.08704090e+00 -9.38673854e-01 -9.80061650e-01 -3.00456107e-01
2.72845209e-01 -1.13225818e+00 1.14933419e+00 5.48367739e-01
-1.22773409e+00 -6.45672619e-01 -1.12594330e+00 5.00069261e-01
2.10546225e-01 2.07132652e-01 -1.97636530e-01 9.69196856e-01
-1.18858516e+00 7.20251203e-02 -6.81365490e-01 1.10111326e-01
-9.68652517e-02 3.42270792e-01 -3.86444367e-02 2.32448578e-01
-1.30750489e+00 8.81746352e-01 1.43341720e-04 3.21576223e-02
-1.28109837e+00 -6.77980483e-01 -9.20488000e-01 2.71043628e-01
1.41800717e-01 -1.68731004e-01 1.57629883e+00 -9.75041747e-01
-2.03557825e+00 8.88308525e-01 -6.40169740e-01 -6.54985547e-01
1.75282612e-01 -3.43929768e-01 -9.50444818e-01 -4.37656865e-02
-3.35286409e-01 2.62920022e-01 9.93214488e-01 -1.39206529e+00
-3.93678486e-01 -3.32321405e-01 -4.64667648e-01 2.94311255e-01
-1.54916570e-01 2.41062403e-01 -1.52169973e-01 -8.22203159e-01
1.47102714e-01 -8.16136241e-01 3.72200049e-02 -8.20112228e-01
-4.79836822e-01 -1.55144721e-01 6.81554019e-01 -9.69186127e-01
1.32606888e+00 -2.34445333e+00 -2.75833923e-02 3.40023071e-01
-2.49089375e-01 5.84640384e-01 -3.46703351e-01 1.10085122e-03
-2.52457827e-01 -2.34359771e-01 -3.50488961e-01 -8.05850625e-01
-9.53332633e-02 -1.87757000e-01 -2.92350829e-01 4.13721412e-01
6.64172843e-02 2.58797884e-01 -5.21397352e-01 -1.38636172e-01
2.39732921e-01 9.05357957e-01 -4.21193808e-01 5.73609233e-01
2.22996891e-01 3.84363532e-01 3.67733717e-01 4.26835358e-01
4.80209172e-01 5.04956126e-01 2.16831751e-02 5.73536009e-03
7.98266605e-02 9.20166850e-01 -1.23819470e+00 1.48773968e+00
-8.26896250e-01 7.58700252e-01 6.99781537e-01 -9.39724803e-01
1.24857163e+00 6.45812988e-01 1.23068318e-01 -6.86796308e-01
1.41705692e-01 4.02199745e-01 5.86431563e-01 8.86806995e-02
1.86921716e-01 -2.89684832e-01 1.03395358e-01 2.06768930e-01
1.64062619e-01 -2.37357646e-01 -2.74378538e-01 3.73219401e-02
8.45214367e-01 -4.65628445e-01 6.34770021e-02 -3.72118443e-01
9.61093307e-01 -8.49503994e-01 3.52418572e-01 7.52299428e-01
-2.43261829e-01 6.65878356e-01 -2.06721529e-01 3.98763448e-01
-7.14913368e-01 -1.36358559e+00 -2.74088502e-01 9.23176050e-01
-3.58402371e-01 -1.78469822e-01 -9.12457585e-01 -2.49813244e-01
-4.13781762e-01 1.37325752e+00 1.42532021e-01 -1.85388997e-01
-7.48991728e-01 -5.53094029e-01 9.52957273e-01 6.20636344e-02
1.78586185e-01 -8.28649282e-01 2.64684200e-01 2.19409361e-01
-4.75473523e-01 -1.10963440e+00 -8.31819415e-01 3.80855709e-01
-3.55528325e-01 -4.91243839e-01 -8.23413730e-01 -7.86827862e-01
3.78972709e-01 1.22121803e-01 8.85963798e-01 -3.89222264e-01
1.62143439e-01 2.81862438e-01 -1.44161865e-01 -4.07655656e-01
-1.18097949e+00 -5.71671650e-02 5.88743925e-01 3.14154953e-01
2.20649272e-01 -6.97817504e-01 -2.07802415e-01 4.56773460e-01
-6.36014283e-01 -3.09652984e-01 3.82977962e-01 9.23882365e-01
2.82899231e-01 -4.19425219e-03 8.87883127e-01 -9.43427980e-02
5.53768992e-01 -2.92160846e-02 -6.33260250e-01 -7.96694867e-03
-3.41422528e-01 1.94805026e-01 6.63677871e-01 -6.83538556e-01
-1.26096654e+00 -5.83799928e-02 -8.22029352e-01 -3.53722990e-01
-3.37674141e-01 7.94152319e-02 -6.71223879e-01 2.54934192e-01
7.70590663e-01 5.19983232e-01 2.39722598e-02 -5.55891275e-01
1.37655258e-01 1.42253685e+00 5.28523326e-01 -4.81620967e-01
7.22376227e-01 -2.24738568e-01 -4.20257688e-01 -1.34422851e+00
-7.46488094e-01 -7.26895750e-01 -8.03017914e-02 8.90393183e-03
3.63442600e-01 -1.12662530e+00 -6.70061767e-01 6.64981067e-01
-1.16636598e+00 -3.55514675e-01 -1.59438267e-01 9.10147786e-01
-4.51749176e-01 3.61099601e-01 -5.58024526e-01 -1.48329234e+00
-5.51848531e-01 -1.27527404e+00 9.89469290e-01 -2.12763533e-01
-2.46339023e-01 -6.80608869e-01 9.74814743e-02 5.23736119e-01
5.99530697e-01 -4.67594773e-01 3.35546017e-01 -1.03003907e+00
-1.17629699e-01 3.39177763e-03 3.62560630e-01 9.81055558e-01
4.37360138e-01 -3.39209884e-01 -1.66283011e+00 -4.89908457e-01
4.50868607e-01 -1.17467470e-01 8.97229433e-01 4.27587211e-01
8.15592527e-01 -5.20872593e-01 -6.00296110e-02 3.72631162e-01
8.90708923e-01 5.82337737e-01 6.61507487e-01 -1.99728578e-01
5.66921830e-01 4.78419006e-01 3.75784129e-01 7.32382759e-02
-2.26696432e-01 1.01545393e+00 5.57191409e-02 -1.01653829e-01
-5.50429344e-01 -4.02333438e-02 9.81896996e-01 1.47671962e+00
3.68582368e-01 -5.45399547e-01 -8.35736752e-01 6.41301870e-01
-1.28445578e+00 -8.81793618e-01 2.88117807e-02 2.60468650e+00
1.05626762e+00 2.58944035e-01 1.51798338e-01 5.51810384e-01
9.33005691e-01 2.10119247e-01 -2.20199645e-01 -5.04996896e-01
-3.26641381e-01 4.85370636e-01 2.54759967e-01 1.11254513e+00
-8.07656944e-01 8.11180115e-01 6.08614826e+00 1.14221859e+00
-1.32338715e+00 1.57032833e-01 4.77227151e-01 -3.92805636e-01
-1.95081174e-01 -4.98974651e-01 -1.08723962e+00 3.86455834e-01
1.65216458e+00 -2.38110632e-01 6.06570065e-01 5.92751503e-01
4.29819018e-01 1.34561971e-01 -1.15890133e+00 1.28575563e+00
3.31081688e-01 -6.10400558e-01 -2.25311127e-02 3.59858014e-02
3.63222450e-01 8.92434567e-02 1.13423109e-01 3.96147043e-01
1.82530023e-02 -1.12780869e+00 7.50266492e-01 8.98839720e-03
8.62510562e-01 -9.67453003e-01 5.39247990e-01 5.78619719e-01
-1.00211251e+00 1.55600250e-01 1.76109113e-02 4.26031388e-02
2.07172960e-01 8.83498192e-01 -1.38907361e+00 3.68227988e-01
2.96635479e-01 -2.65233144e-02 -4.15444598e-02 7.68691540e-01
-3.20173025e-01 1.28150547e+00 -6.60358787e-01 4.97181378e-02
-2.28896122e-02 2.40216181e-01 1.26600862e+00 1.61123848e+00
3.18304658e-01 -1.03964224e-01 -5.48711000e-03 6.36758566e-01
-2.27439731e-01 1.11177437e-01 -3.98315758e-01 1.37737527e-01
6.55612111e-01 8.65220666e-01 -7.47445524e-02 -2.86080092e-01
-3.00762486e-02 9.06370640e-01 1.19213156e-01 5.56138873e-01
-9.22517419e-01 -5.44984877e-01 9.03262198e-01 1.32129237e-01
1.55987471e-01 -1.61708295e-01 -6.28139824e-02 -1.16191864e+00
5.53396158e-03 -1.17742014e+00 -9.31011736e-02 -4.14809823e-01
-1.12663639e+00 8.98331523e-01 -2.11524904e-01 -9.62758243e-01
-3.82901967e-01 -3.59652430e-01 -5.30132651e-01 1.43666089e+00
-1.23573017e+00 -6.69541121e-01 2.95876563e-01 4.80270952e-01
8.68465126e-01 -4.35425639e-01 1.01507354e+00 5.64738691e-01
-4.87417459e-01 1.30690730e+00 6.52125254e-02 4.10722978e-02
7.46810555e-01 -1.16566205e+00 4.98914659e-01 1.05842614e+00
3.27081472e-01 5.59865415e-01 8.84569347e-01 -9.82131511e-02
-9.51961994e-01 -1.04476571e+00 9.20321941e-01 -1.94217786e-01
4.09349024e-01 -7.07263947e-01 -1.00467062e+00 4.91246462e-01
3.53929490e-01 -4.22014117e-01 7.72129893e-01 1.52201086e-01
-3.31204176e-01 -4.54080641e-01 -1.09053421e+00 5.51437974e-01
7.11404562e-01 -6.19156241e-01 -8.75825405e-01 1.31328359e-01
1.08629191e+00 -3.17549527e-01 -3.89203846e-01 3.58963251e-01
3.94585073e-01 -6.22800946e-01 8.76195669e-01 -1.48689121e-01
-2.54988879e-01 -4.07625198e-01 -5.52381754e-01 -1.71603346e+00
-1.70793355e-01 -9.74751234e-01 -1.93685964e-01 1.49337506e+00
6.96954966e-01 -7.18270779e-01 2.73558617e-01 2.02229917e-01
-4.91216958e-01 -4.25530732e-01 -1.06505656e+00 -1.09050608e+00
-3.34855877e-02 -8.88860881e-01 1.36480242e-01 5.73412538e-01
-4.36425582e-02 7.07245350e-01 -4.14398760e-01 5.36497653e-01
9.28082228e-01 -5.08867264e-01 5.68076730e-01 -7.87625909e-01
-5.18099427e-01 -4.03110564e-01 -9.71167833e-02 -1.30230498e+00
4.27857876e-01 -1.03103745e+00 5.21124840e-01 -1.12493217e+00
-2.26595521e-01 -1.54030859e-01 -5.90238333e-01 1.95350543e-01
-2.43861213e-01 6.74939901e-02 2.20477924e-01 -2.31874302e-01
-1.71814978e-01 7.22557843e-01 8.88219953e-01 -3.88307214e-01
-5.29965878e-01 4.70669031e-01 -6.54841661e-01 5.70955753e-01
9.33446527e-01 -5.72095394e-01 -1.50172353e-01 -5.33749200e-02
-6.16706371e-01 1.59281611e-01 -1.20792553e-01 -1.10404992e+00
3.09422463e-02 3.11701357e-01 2.59940308e-02 -5.78739941e-01
6.85723603e-01 -6.41388416e-01 5.22727147e-02 3.49302411e-01
-5.51998258e-01 -8.04852009e-01 4.92122382e-01 1.50378197e-01
-4.59997505e-01 -7.67016932e-02 1.15617228e+00 2.55079389e-01
8.27609226e-02 -3.53519171e-01 -5.39122343e-01 -2.94336714e-02
5.55120349e-01 7.99308792e-02 6.75362870e-02 -6.45638466e-01
-7.74973869e-01 -8.24904442e-02 -1.46477416e-01 4.06872064e-01
6.07592225e-01 -1.20980096e+00 -1.19015336e+00 3.72550845e-01
-2.12381825e-01 -1.74760163e-01 1.29326046e-01 6.73663974e-01
1.54165268e-01 6.09084785e-01 3.80640268e-01 -8.49248528e-01
-1.72240639e+00 1.67716652e-01 6.15342081e-01 -1.51540279e-01
-1.79978415e-01 9.77462709e-01 4.36506152e-01 -5.04902482e-01
7.27969527e-01 -1.75760359e-01 5.04602939e-02 -2.70480633e-01
6.80660307e-01 4.80537891e-01 4.00434911e-01 -9.15236115e-01
-4.74594414e-01 2.52591729e-01 9.50274169e-02 -7.37821460e-01
9.16933060e-01 -2.04129547e-01 2.55956829e-01 6.75843716e-01
1.41319013e+00 6.66027606e-01 -7.35504687e-01 -2.96151996e-01
-3.50323796e-01 -2.45156795e-01 2.86574244e-01 -1.00987804e+00
-7.15003252e-01 9.58013177e-01 7.22299099e-01 1.74202964e-01
1.22735703e+00 -3.11944820e-03 9.17987287e-01 3.08056414e-01
1.42412320e-01 -8.56616378e-01 -2.42656618e-02 7.24185109e-01
1.08623350e+00 -1.06333232e+00 -6.07662916e-01 -2.37304062e-01
-4.78568137e-01 8.27333689e-01 4.81714755e-01 2.70000160e-01
5.27816415e-01 5.14125228e-01 4.47495401e-01 3.89309824e-01
-6.81147873e-01 -3.22752148e-01 6.36071563e-01 5.45999587e-01
6.99496150e-01 2.55359590e-01 2.14884188e-02 7.09787011e-01
-6.47759259e-01 -7.26982117e-01 2.83910573e-01 2.82923222e-01
-6.05627179e-01 -9.40075994e-01 -8.73188734e-01 5.09399921e-03
-7.87198186e-01 -3.37667912e-01 -3.59511912e-01 2.04009980e-01
-4.51611847e-01 1.52404475e+00 -3.46195791e-03 -3.44137430e-01
5.87325096e-01 4.25313473e-01 3.60844135e-01 -7.25404978e-01
-5.52198589e-01 7.68112421e-01 2.58538336e-01 -2.60406792e-01
-1.99894503e-01 -4.25100327e-01 -1.15455103e+00 1.15591548e-01
-5.90976179e-01 4.99378264e-01 9.41260934e-01 7.13736653e-01
2.13424396e-02 9.74362671e-01 1.02062786e+00 -7.04063654e-01
-7.31567860e-01 -1.43514991e+00 -5.85023403e-01 1.71741128e-01
7.83995450e-01 -2.30756193e-01 -1.03318214e+00 -8.11520219e-02] | [14.802361488342285, 6.1045823097229] |
c816baf1-2a77-4b51-a14e-c0b0d869812d | orientation-shared-convolution-representation | 2212.13166 | null | https://arxiv.org/abs/2212.13166v1 | https://arxiv.org/pdf/2212.13166v1.pdf | Orientation-Shared Convolution Representation for CT Metal Artifact Learning | During X-ray computed tomography (CT) scanning, metallic implants carrying with patients often lead to adverse artifacts in the captured CT images and then impair the clinical treatment. Against this metal artifact reduction (MAR) task, the existing deep-learning-based methods have gained promising reconstruction performance. Nevertheless, there is still some room for further improvement of MAR performance and generalization ability, since some important prior knowledge underlying this specific task has not been fully exploited. Hereby, in this paper, we carefully analyze the characteristics of metal artifacts and propose an orientation-shared convolution representation strategy to adapt the physical prior structures of artifacts, i.e., rotationally symmetrical streaking patterns. The proposed method rationally adopts Fourier-series-expansion-based filter parametrization in artifact modeling, which can better separate artifacts from anatomical tissues and boost the model generalizability. Comprehensive experiments executed on synthesized and clinical datasets show the superiority of our method in detail preservation beyond the current representative MAR methods. Code will be available at \url{https://github.com/hongwang01/OSCNet} | ['Yefeng Zheng', 'Deyu Meng', 'Yawen Huang', 'Yuexiang Li', 'Qi Xie', 'Hong Wang'] | 2022-12-26 | null | null | null | null | ['metal-artifact-reduction'] | ['medical'] | [ 2.08588019e-01 -2.37851098e-01 1.97133124e-01 -2.13347584e-01
-8.65296245e-01 7.75960386e-02 1.39647901e-01 -2.11842746e-01
-8.02703649e-02 6.09081328e-01 2.99168587e-01 -1.13551833e-01
-4.41335738e-01 -4.07325894e-01 -4.91524160e-01 -9.26631153e-01
-6.54474692e-03 -3.85706825e-03 3.20445150e-01 1.82333082e-01
-4.79886383e-02 4.72410232e-01 -8.78503859e-01 4.65716094e-01
1.03821325e+00 1.16772687e+00 4.15891260e-01 -1.01932967e-02
3.37861747e-01 8.79791915e-01 -3.03338557e-01 -1.77112639e-01
3.13514989e-04 -4.45468158e-01 -5.93362749e-01 -7.83932731e-02
-9.40964669e-02 -3.73950303e-01 -7.08869576e-01 1.18618143e+00
7.47388542e-01 6.03459887e-02 6.29211903e-01 -6.06515527e-01
-7.14420676e-01 5.99893391e-01 -6.56862557e-01 4.07661229e-01
-8.88185129e-02 2.47056663e-01 2.78433233e-01 -1.06883383e+00
3.08126211e-01 5.44552386e-01 9.76989210e-01 5.56860745e-01
-8.06023180e-01 -7.74538219e-01 -1.03104770e-01 3.34594101e-01
-1.31817985e+00 -1.43465251e-01 1.14093685e+00 -4.46274996e-01
4.12353456e-01 4.99307990e-01 5.20672679e-01 1.32658851e+00
5.04227102e-01 6.79538131e-01 1.03064370e+00 -1.61617696e-01
1.19868241e-01 -2.60219108e-02 3.97379845e-02 5.32947600e-01
4.90281522e-01 4.10828292e-02 -2.12977573e-01 -3.33503753e-01
1.24596679e+00 2.62881339e-01 -9.68727410e-01 -2.21630290e-01
-1.15180171e+00 4.24643219e-01 6.84672654e-01 7.05107927e-01
-6.88135564e-01 2.02172399e-01 4.52124536e-01 -3.24045688e-01
4.49573129e-01 2.43836507e-01 -2.68481106e-01 8.64922404e-02
-7.86579013e-01 3.40337008e-02 6.93081915e-02 8.47758889e-01
1.69103757e-01 2.83168226e-01 -3.65960598e-01 1.00289321e+00
3.97801548e-01 3.20448637e-01 8.24561596e-01 -4.62147534e-01
2.28823766e-01 3.23662817e-01 -1.38819115e-02 -8.39900315e-01
-5.49150884e-01 -9.62077260e-01 -1.25831664e+00 3.82810533e-02
2.40726262e-01 -1.62664339e-01 -9.98875797e-01 1.31127131e+00
2.70761728e-01 6.06843174e-01 -3.90220612e-01 1.35293257e+00
9.14463341e-01 2.14137390e-01 1.86490759e-01 -3.46095085e-01
1.51747739e+00 -7.19369829e-01 -8.97790194e-01 -1.93498969e-01
4.06846136e-01 -9.82903183e-01 1.08958399e+00 4.77238625e-01
-1.23344600e+00 -3.89093965e-01 -1.18302441e+00 2.36179277e-01
3.69289577e-01 2.94488579e-01 7.86492050e-01 6.05668604e-01
-4.96627629e-01 7.34262824e-01 -1.33991373e+00 1.55987516e-01
6.94388747e-01 2.48779893e-01 9.95741338e-02 -1.68082118e-01
-1.19422519e+00 8.17832828e-01 -2.28738174e-01 7.00219691e-01
-7.60875821e-01 -1.02102029e+00 -4.72220838e-01 -1.72401190e-01
2.00944707e-01 -7.55985916e-01 1.40570545e+00 -7.64596462e-01
-1.58288813e+00 4.06279027e-01 1.66583389e-01 -8.34345296e-02
6.91310227e-01 -2.73713171e-01 -6.70247614e-01 2.17316970e-01
-1.23621501e-01 -2.04358041e-01 7.23537803e-01 -1.42691088e+00
3.88178267e-02 -4.22961622e-01 -3.36171001e-01 -1.15168989e-01
-3.76509130e-01 1.26044676e-01 -5.08715749e-01 -1.19129741e+00
6.05595112e-01 -8.71219218e-01 -4.98707950e-01 1.37491167e-01
-2.79219925e-01 2.58165687e-01 1.95034012e-01 -7.56606817e-01
1.51984537e+00 -2.16529608e+00 -1.91274628e-01 2.10183874e-01
2.33472064e-01 2.27669731e-01 2.96958506e-01 2.59897619e-01
-3.77748400e-01 -2.23797217e-01 -5.55262506e-01 2.38497276e-02
-3.98498893e-01 -1.45284921e-01 -2.47736108e-02 8.10842156e-01
-6.14122786e-02 8.32490385e-01 -6.41939461e-01 -2.82112002e-01
1.83204561e-01 7.62598753e-01 -4.81415480e-01 4.51927148e-02
1.74199626e-01 9.57602739e-01 -9.86983716e-01 5.85234463e-01
1.23662078e+00 -5.28964281e-01 2.67733485e-02 -6.79909766e-01
3.40998024e-02 1.62226290e-01 -1.05080295e+00 1.89326000e+00
-6.45519912e-01 1.90954149e-01 1.05951495e-01 -7.19562471e-01
5.41087270e-01 5.43951571e-01 9.63775098e-01 -7.31552601e-01
4.29770887e-01 5.42899370e-01 1.35745019e-01 -8.55509043e-01
-1.41019806e-01 -4.50598776e-01 4.27088290e-01 1.56616136e-01
-4.22080994e-01 1.62841588e-01 -6.74058139e-01 -2.48196080e-01
1.04775190e+00 1.70507170e-02 3.75628807e-02 -4.85899657e-01
5.80564976e-01 -1.61690667e-01 5.38070917e-01 5.45710683e-01
-1.14361376e-01 1.05021131e+00 -2.06192851e-01 -5.30934036e-01
-7.33379066e-01 -1.12194347e+00 -7.40890443e-01 2.47450769e-01
3.92445594e-01 -1.11732326e-01 -6.90895736e-01 -3.72847557e-01
-2.51777917e-01 5.57990551e-01 -5.52481532e-01 -3.61698508e-01
-9.23478365e-01 -1.19757330e+00 4.78903204e-01 7.85393655e-01
4.35235560e-01 -9.31710601e-01 -6.57283604e-01 3.54225039e-01
-3.42423886e-01 -9.73168910e-01 -4.28278059e-01 1.07154332e-01
-1.21230829e+00 -9.94966984e-01 -1.10864127e+00 -6.36453569e-01
6.55271471e-01 1.14771880e-01 7.71237731e-01 3.63099962e-01
-5.82173169e-01 8.54023620e-02 -4.16157067e-01 -3.08533758e-01
-2.36406401e-01 -4.42897499e-01 -1.59596413e-01 5.52506037e-02
1.55621991e-01 -7.74787426e-01 -1.25194597e+00 4.36460286e-01
-1.08417428e+00 1.21467955e-01 8.08500826e-01 8.49331737e-01
4.80391711e-01 5.65987714e-02 4.02864337e-01 -8.71920109e-01
4.76593912e-01 -4.06817794e-01 -2.95762807e-01 1.93749778e-02
-2.95722425e-01 -1.44831240e-01 5.88083208e-01 -5.05877793e-01
-1.44182014e+00 -9.86415148e-02 -4.03861642e-01 -4.48008925e-01
-1.38620615e-01 5.32612205e-01 -1.03743464e-01 -2.56375819e-01
7.40611434e-01 3.33193570e-01 -2.73192286e-01 -6.60212278e-01
-2.68427104e-01 5.43030322e-01 3.13886881e-01 -5.83684802e-01
6.20156527e-01 7.22149491e-01 2.17800438e-02 -6.83944643e-01
-5.36746323e-01 -2.98929513e-01 -2.35884175e-01 -1.95022181e-01
5.85982800e-01 -8.97700012e-01 -5.18054545e-01 6.85754418e-01
-1.18518341e+00 -1.43679962e-01 1.17308367e-02 9.85955238e-01
-3.88499528e-01 7.13456810e-01 -9.39645827e-01 -6.02664292e-01
-4.63089049e-01 -1.48941875e+00 7.87205637e-01 5.56122772e-02
-4.06898595e-02 -6.79002345e-01 -4.50379886e-02 2.23049909e-01
7.09129274e-01 2.02868894e-01 9.81227994e-01 -3.23199928e-01
-6.41814232e-01 -1.30226135e-01 -2.35093802e-01 3.64251763e-01
2.46300921e-01 -4.26021487e-01 -1.03197670e+00 -2.31832862e-01
7.30200708e-01 1.00003280e-01 6.47898376e-01 7.64760494e-01
1.72280633e+00 7.75795132e-02 -3.89691085e-01 8.53059530e-01
1.49908066e+00 4.53105420e-01 9.21523273e-01 2.67129719e-01
5.71061492e-01 1.44069046e-01 2.67073482e-01 8.28447461e-01
-1.25937000e-01 6.80510938e-01 3.76725525e-01 -1.71257317e-01
-3.93308818e-01 1.21692747e-01 -2.16596380e-01 8.24593306e-01
-4.07376915e-01 -3.93531658e-02 -9.12901580e-01 2.64680624e-01
-1.60752165e+00 -5.58445394e-01 -5.21611035e-01 2.02214265e+00
7.33142495e-01 -2.81651728e-02 -4.35246915e-01 2.22820602e-02
6.24362767e-01 -3.18615586e-02 -5.31768084e-01 3.77862841e-01
1.13532163e-01 5.49699903e-01 4.95585471e-01 1.93787873e-01
-9.23883379e-01 2.28849232e-01 5.53726673e+00 9.89616573e-01
-1.28695738e+00 5.23674905e-01 4.30345923e-01 2.88630575e-02
-4.79931891e-01 -2.48941705e-01 -1.87617436e-01 7.00925350e-01
3.67822587e-01 1.16058700e-01 1.90823421e-01 5.99615455e-01
4.82203990e-01 7.74102435e-02 -7.58781552e-01 1.10206676e+00
-1.55946687e-01 -1.15729809e+00 -9.64189023e-02 -1.12775631e-01
4.85468537e-01 -1.03200540e-01 2.05313474e-01 -7.23579302e-02
-3.16049725e-01 -8.43150437e-01 6.05599344e-01 6.08350575e-01
7.55651593e-01 -3.68600935e-01 7.74448812e-01 -3.10968747e-03
-9.02620137e-01 3.06091923e-02 -3.27321827e-01 2.27983654e-01
2.36936554e-01 1.00016856e+00 -3.75828922e-01 9.86405909e-01
8.42885613e-01 5.52972853e-01 -3.95185590e-01 1.44872665e+00
-1.12322606e-01 6.67904437e-01 -1.94127187e-01 2.53854871e-01
-5.13581224e-02 -6.69634417e-02 5.98209977e-01 1.01390171e+00
5.53426445e-01 5.47713339e-01 -2.34656870e-01 8.60840380e-01
1.88637629e-01 1.13250457e-01 -8.31879228e-02 5.40745974e-01
1.02412619e-01 1.10381556e+00 -7.27940083e-01 -1.01121329e-01
-6.77848220e-01 8.88330162e-01 -1.97008669e-01 5.82657397e-01
-1.25009227e+00 -1.49486765e-01 3.66551965e-01 5.34195840e-01
1.22255579e-01 -1.70918614e-01 -6.45023644e-01 -1.34357357e+00
4.04360145e-01 -7.59201169e-01 2.31794685e-01 -8.35853040e-01
-1.42384017e+00 9.09514844e-01 -6.14854731e-02 -1.62594604e+00
3.63868982e-01 -6.36303842e-01 -6.88042223e-01 7.40669608e-01
-1.33018959e+00 -1.15820956e+00 -6.29236937e-01 6.02599800e-01
4.51187968e-01 2.48120174e-01 6.62051558e-01 8.55140746e-01
-4.85999227e-01 5.07982492e-01 2.69981712e-01 -3.50532345e-02
7.21320331e-01 -6.70316696e-01 -1.88018516e-01 7.06130087e-01
-3.23242605e-01 7.13011026e-01 7.05875218e-01 -7.04227388e-01
-1.38236356e+00 -9.34123278e-01 3.65295485e-02 -1.78869441e-01
7.16518104e-01 5.26298620e-02 -1.15491199e+00 5.88106811e-01
2.41587907e-01 2.59627819e-01 6.93158090e-01 -3.11836869e-01
-3.21495906e-02 -6.59279898e-02 -1.01800776e+00 6.57566309e-01
1.05798888e+00 -1.08968310e-01 -3.49828541e-01 3.85161281e-01
2.89113224e-01 -6.87636971e-01 -8.97642493e-01 8.42692077e-01
5.66273630e-01 -9.29190397e-01 1.14085972e+00 -3.09167385e-01
6.23977005e-01 -2.60317057e-01 7.68468007e-02 -1.14663672e+00
-6.89015269e-01 -3.01684350e-01 9.69525352e-02 7.24223077e-01
1.81166738e-01 -7.14153945e-01 7.78678417e-01 6.23535931e-01
-7.43059635e-01 -1.00435686e+00 -1.09556425e+00 -6.74887478e-01
2.66156971e-01 -5.61845660e-01 4.95898515e-01 9.44895208e-01
-8.18762556e-02 -3.32879484e-01 -3.05684268e-01 6.32538497e-01
6.87844455e-01 -2.80518141e-02 1.53009251e-01 -8.94776046e-01
-5.16712308e-01 -4.57519770e-01 -1.75272167e-01 -9.82357502e-01
-3.43999147e-01 -8.20102036e-01 -2.45854110e-02 -1.36839998e+00
2.53768772e-01 -6.23871386e-01 -7.19075084e-01 1.34209394e-01
-3.02643478e-01 2.88470477e-01 -2.36690581e-01 3.91017765e-01
-1.42503053e-01 7.37030029e-01 1.77128744e+00 -9.57741439e-02
7.01985881e-02 1.57414958e-01 -5.35669744e-01 8.78094077e-01
6.62088335e-01 -4.74826574e-01 -2.96784431e-01 -7.45027304e-01
-5.54878972e-02 2.12963372e-01 5.86651802e-01 -1.28860080e+00
1.70538917e-01 -4.89177667e-02 6.96741045e-01 -3.93639058e-01
2.77311176e-01 -1.14693367e+00 7.05877304e-01 7.13569403e-01
-7.61018172e-02 -1.74286410e-01 4.31008250e-01 5.04398406e-01
-1.73370779e-01 -4.19069201e-01 1.02265692e+00 -2.30481774e-01
-1.71154305e-01 4.83441323e-01 -4.60666806e-01 -1.81618452e-01
7.27078319e-01 -1.88806921e-01 1.26588729e-03 3.49552147e-02
-7.78861225e-01 -2.40397796e-01 2.88656980e-01 1.37468040e-01
8.49031687e-01 -1.47819126e+00 -6.63442791e-01 2.20445126e-01
1.36163801e-01 -3.95983877e-03 1.11690640e+00 1.39178026e+00
-6.39131725e-01 1.60458714e-01 -2.39942316e-02 -5.29087543e-01
-8.21583271e-01 5.43200076e-01 6.26308084e-01 -2.83422023e-01
-9.94989514e-01 8.37566912e-01 5.44221699e-01 1.49931848e-01
-2.37627681e-02 -6.05904102e-01 8.02065283e-02 -5.94959676e-01
4.88213480e-01 2.78904825e-01 5.02655923e-01 -3.46382111e-01
-5.69407821e-01 6.73357844e-01 -1.69656411e-01 3.01359802e-01
1.40257120e+00 -4.63117547e-02 1.28611878e-01 -1.48342103e-02
1.00708413e+00 6.26768246e-02 -1.00896978e+00 -2.57904470e-01
-1.07942469e-01 -6.29808426e-01 2.40098193e-01 -8.60391319e-01
-1.46997011e+00 8.33893239e-01 8.70062411e-01 -1.23648979e-01
1.39516306e+00 -9.50936303e-02 9.64492381e-01 -2.77544558e-01
4.79665607e-01 -5.67753732e-01 2.60149926e-01 -1.18999496e-01
1.06833208e+00 -9.82186675e-01 2.66988039e-01 -8.66594791e-01
-4.99645263e-01 1.08938110e+00 5.41391730e-01 -2.28950679e-01
9.50547636e-01 3.87204707e-01 1.11987390e-01 -4.33158904e-01
-1.63666770e-01 2.77065426e-01 2.08967939e-01 3.84589940e-01
6.71308935e-01 -3.62990648e-02 -5.57145894e-01 1.30514634e+00
3.77225280e-01 2.96978951e-01 3.74391884e-01 9.15986836e-01
-9.86240134e-02 -8.96452963e-01 -4.76337075e-01 3.39022279e-01
-7.33308673e-01 -3.00564736e-01 3.71832073e-01 5.98056853e-01
5.79675138e-02 6.06956005e-01 -5.03026545e-01 -3.92436311e-02
6.24926686e-01 -2.53486395e-01 7.47665167e-01 -4.32100892e-01
-6.49388194e-01 4.71537828e-01 -2.84526169e-01 -5.63102901e-01
-3.44846606e-01 -4.88086373e-01 -1.28696442e+00 1.05228201e-01
-5.36442459e-01 -2.75136754e-02 4.91864264e-01 6.34854138e-01
2.94207215e-01 1.01290345e+00 4.89145696e-01 -7.59756684e-01
-6.70872629e-01 -1.01945198e+00 -7.06013560e-01 4.28638339e-01
1.19061187e-01 -9.08606291e-01 -4.05573666e-01 -1.34070486e-01] | [13.513352394104004, -2.585829257965088] |
22e95806-2036-4be2-844a-537d993b5bbd | end-to-end-evaluation-of-a-spoken-dialogue | 2211.03511 | null | https://arxiv.org/abs/2211.03511v1 | https://arxiv.org/pdf/2211.03511v1.pdf | End-to-End Evaluation of a Spoken Dialogue System for Learning Basic Mathematics | The advances in language-based Artificial Intelligence (AI) technologies applied to build educational applications can present AI for social-good opportunities with a broader positive impact. Across many disciplines, enhancing the quality of mathematics education is crucial in building critical thinking and problem-solving skills at younger ages. Conversational AI systems have started maturing to a point where they could play a significant role in helping students learn fundamental math concepts. This work presents a task-oriented Spoken Dialogue System (SDS) built to support play-based learning of basic math concepts for early childhood education. The system has been evaluated via real-world deployments at school while the students are practicing early math concepts with multimodal interactions. We discuss our efforts to improve the SDS pipeline built for math learning, for which we explore utilizing MathBERT representations for potential enhancement to the Natural Language Understanding (NLU) module. We perform an end-to-end evaluation using real-world deployment outputs from the Automatic Speech Recognition (ASR), Intent Recognition, and Dialogue Manager (DM) components to understand how error propagation affects the overall performance in real-world scenarios. | ['Lama Nachman', 'Roddy Fuentes Alba', 'Saurav Sahay', 'Eda Okur'] | 2022-11-07 | null | null | null | null | ['intent-recognition'] | ['natural-language-processing'] | [ 1.66972354e-01 5.60502172e-01 2.82966226e-01 -7.24748611e-01
-8.37440968e-01 -5.88440418e-01 7.48492777e-01 7.00158775e-01
-1.13210402e-01 1.93275675e-01 5.31737983e-01 -6.84168518e-01
-2.04544768e-01 -9.69592750e-01 -2.80528069e-01 9.15745050e-02
-8.02505538e-02 6.93225741e-01 3.44484657e-01 -1.03417420e+00
3.30037594e-01 3.28780860e-01 -1.80651510e+00 8.24414253e-01
1.06450236e+00 2.70336181e-01 3.18095051e-02 1.37682068e+00
-4.68156815e-01 1.73006403e+00 -9.60400045e-01 -3.11822087e-01
-2.33639345e-01 -2.25028917e-01 -1.37960243e+00 -3.51395547e-01
6.86820447e-01 -5.90145230e-01 -3.69574100e-01 5.13194203e-01
5.57155132e-01 7.71050096e-01 1.58737883e-01 -1.27148569e+00
-3.91680449e-01 1.12448478e+00 2.31785819e-01 1.07732907e-01
9.81898010e-01 3.68151635e-01 7.40090787e-01 -4.58843082e-01
1.26817614e-01 1.47061563e+00 2.57114887e-01 8.94842684e-01
-7.44762599e-01 -6.36764109e-01 -2.25463629e-01 2.91206360e-01
-8.65565419e-01 -6.71950936e-01 4.82393682e-01 -5.05822718e-01
1.22938836e+00 3.51851791e-01 8.99921715e-01 8.23600650e-01
-2.85542697e-01 1.33043885e+00 7.80801475e-01 -6.37912452e-01
1.76149160e-01 2.86262214e-01 7.48012424e-01 9.44237888e-01
-6.39241278e-01 -3.55596066e-01 -8.59173834e-01 2.55909413e-01
1.63932443e-01 -4.29590374e-01 8.85083005e-02 3.34752560e-01
-1.16087031e+00 9.05225694e-01 1.22519415e-02 4.79726404e-01
-2.09139157e-02 3.14408587e-03 4.10852671e-01 6.55533910e-01
4.78191316e-01 8.88502479e-01 -3.80653620e-01 -1.26823163e+00
-7.01134861e-01 3.62455815e-01 1.42799675e+00 9.51227844e-01
1.78676441e-01 5.99953867e-02 -4.20626789e-01 1.22511983e+00
5.59186816e-01 3.00519049e-01 4.48388100e-01 -9.56837118e-01
4.95698392e-01 1.12027228e+00 -5.36687851e-01 -4.64796484e-01
-2.60598600e-01 -8.61565582e-03 -2.11722061e-01 -4.75284457e-03
5.78481972e-01 -3.30276489e-01 -5.85024536e-01 1.33266139e+00
4.49224532e-01 3.76225948e-01 5.63355267e-01 4.80875492e-01
1.86929703e+00 9.87974703e-01 3.47400874e-01 5.27208507e-01
1.39059055e+00 -1.01519918e+00 -6.75538182e-01 -2.43915781e-01
1.00163686e+00 -9.63946044e-01 1.03265154e+00 6.51243627e-01
-1.34462464e+00 -3.31918925e-01 -1.00518906e+00 -4.38921869e-01
-4.62927610e-01 -8.87250751e-02 7.01726019e-01 1.14877617e+00
-1.25924110e+00 2.26557046e-01 -7.87259758e-01 -4.58685070e-01
6.28668517e-02 2.55575866e-01 -1.89904019e-01 -2.28376426e-02
-1.20628130e+00 7.60102570e-01 -1.29060343e-01 -4.20015872e-01
-1.02227294e+00 -1.23799932e+00 -9.82657552e-01 3.17313403e-01
1.30268261e-01 4.49470095e-02 1.93227601e+00 -3.79678726e-01
-2.30535555e+00 9.10980403e-01 3.09797615e-01 -3.29706371e-01
2.06806883e-01 -3.64667833e-01 -4.75415820e-03 1.80940732e-01
-2.00823382e-01 9.21858788e-01 -2.04623237e-01 -5.19680858e-01
-6.92954719e-01 -1.11603990e-01 4.48336661e-01 5.26286304e-01
-4.45254266e-01 2.01003715e-01 2.48170290e-02 -3.33533525e-01
9.35132578e-02 -6.23740792e-01 1.14341177e-01 -3.57821465e-01
6.00497723e-02 -8.56574714e-01 5.08001745e-01 -7.87218869e-01
1.13636148e+00 -1.84792590e+00 -1.97558299e-01 7.32980967e-02
5.22799268e-02 6.13691628e-01 -1.73467442e-01 9.26163137e-01
8.18272680e-02 -7.78732309e-03 3.78034800e-01 -3.07200223e-01
1.89702243e-01 -1.31165907e-01 -1.82655782e-01 -2.06178024e-01
4.80097272e-02 5.50366819e-01 -1.19381118e+00 4.48180363e-02
6.35128319e-01 2.53600031e-01 -6.28281415e-01 9.71673012e-01
-5.19970596e-01 2.81719863e-01 -4.11637545e-01 3.72705877e-01
1.10936292e-01 1.94234639e-01 -2.87026674e-01 7.12003410e-01
-3.78381789e-01 9.41407681e-01 -1.05606997e+00 1.95442331e+00
-7.23498285e-01 1.07673573e+00 3.08247149e-01 -1.05789983e+00
1.04947627e+00 6.31556988e-01 3.67923796e-01 -6.71705306e-01
-5.38207628e-02 -2.62980551e-01 4.01253849e-01 -8.15722287e-01
5.45901418e-01 3.82044464e-01 6.87195733e-02 7.42083132e-01
2.07187578e-01 -8.36531043e-01 7.92788193e-02 6.62770927e-01
1.47830367e+00 -1.94892809e-01 -4.53020409e-02 -2.64765590e-01
7.10020065e-01 1.82828307e-01 -3.42872828e-01 7.48127639e-01
-3.02603096e-01 8.18898305e-02 1.23645373e-01 -3.13135624e-01
-4.82290536e-01 -8.48264217e-01 2.24969715e-01 1.81113112e+00
-6.42719686e-01 -4.67432559e-01 -9.79008794e-01 -2.92624176e-01
-4.06817675e-01 1.23713303e+00 1.05723208e-02 -4.11688387e-01
-2.66549796e-01 1.69524506e-01 1.05562174e+00 6.16594478e-02
6.73306465e-01 -9.85073626e-01 -6.00962818e-01 5.08677721e-01
-2.34381650e-02 -1.26121211e+00 1.20371893e-01 -3.13426644e-01
-3.12977254e-01 -7.46295750e-01 -4.40180153e-01 -7.79920220e-01
1.67065814e-01 3.66071880e-01 1.16617203e+00 3.49294990e-01
-4.97084916e-01 1.37033105e+00 -6.15121424e-01 -8.60580325e-01
-1.05859697e+00 5.52377999e-02 2.01249234e-02 -7.24745929e-01
7.86481738e-01 -4.29288298e-01 -2.94182539e-01 -6.29763454e-02
-5.33435464e-01 6.42106175e-01 2.25680601e-02 3.55381489e-01
-7.41955340e-01 -1.48961499e-01 5.20011604e-01 -6.20971262e-01
9.87527668e-01 -5.28859735e-01 -2.96795338e-01 3.52943212e-01
-2.05405816e-01 -1.28322437e-01 2.26759121e-01 -3.46168786e-01
-1.30163503e+00 -3.47557396e-01 -7.85260201e-01 1.29897788e-01
-7.23380983e-01 5.86141706e-01 1.17007099e-01 -1.18440919e-01
5.44808209e-01 1.07020177e-01 1.55501375e-02 -1.76735312e-01
5.21820366e-01 1.30850899e+00 4.20798361e-01 -1.27749550e+00
1.37132123e-01 -4.33923095e-01 -3.80167246e-01 -1.63996017e+00
-6.35218501e-01 -6.57474458e-01 -3.01126063e-01 -8.07377219e-01
7.91125834e-01 -1.13297927e+00 -1.54241860e+00 4.20876890e-01
-1.08337855e+00 -8.97221565e-01 -8.53438973e-02 6.41036034e-01
-2.84348547e-01 -1.21221893e-01 -7.91280270e-01 -1.14610779e+00
-4.28594410e-01 -1.42692399e+00 7.50595272e-01 7.74705350e-01
-5.55308163e-01 -9.54150736e-01 2.42570832e-01 1.50565529e+00
3.95411283e-01 -4.58847195e-01 9.70925331e-01 -1.40034962e+00
-4.51728314e-01 6.95273802e-02 2.33551353e-01 2.83930629e-01
-4.67160076e-01 1.96004599e-01 -1.22885287e+00 2.98892241e-02
-3.59721124e-01 -7.63299525e-01 -5.12715168e-02 -1.65595621e-01
6.25096977e-01 -2.96110898e-01 1.99939862e-01 -2.17747927e-01
6.25755429e-01 6.20430470e-01 2.51660496e-01 4.68006469e-02
5.58337688e-01 1.17973423e+00 4.96184081e-01 3.72777313e-01
8.56885791e-01 4.06830937e-01 -2.46243134e-01 4.14169937e-01
-2.38499880e-01 -1.84912682e-01 6.63931191e-01 1.18649292e+00
4.70262825e-01 -2.80216426e-01 -1.75290322e+00 8.17732215e-01
-1.59657586e+00 -9.16743577e-01 -2.14702100e-01 1.94631970e+00
1.20630670e+00 -2.64133792e-02 1.22601643e-01 -7.55278617e-02
9.90286916e-02 -2.35915184e-01 7.26981983e-02 -1.12760448e+00
8.15520406e-01 6.80335343e-01 -1.88209906e-01 9.61454511e-01
-5.72382092e-01 1.24863935e+00 5.19777775e+00 7.10541129e-01
-9.51173246e-01 -6.41623139e-02 6.99961543e-01 1.33326605e-01
-2.16445476e-01 -3.83601815e-01 -8.19370627e-01 -1.22753466e-02
1.60917652e+00 -2.33472094e-01 5.73579669e-01 8.00185561e-01
2.59149879e-01 -3.92026663e-01 -1.44150579e+00 5.40293217e-01
-3.10242083e-02 -1.42969823e+00 -2.79593021e-01 -4.75140631e-01
2.78927624e-01 5.58311902e-02 -1.42314166e-01 8.49471927e-01
1.02533746e+00 -1.13099802e+00 1.83662996e-01 3.59705180e-01
2.94796944e-01 -9.58841622e-01 3.15222740e-01 6.98536932e-01
-8.57605577e-01 8.79968852e-02 1.57170936e-01 -6.11816883e-01
-4.72405583e-01 -1.11183636e-01 -1.85094643e+00 -5.66845797e-02
5.16789496e-01 2.36013278e-01 -4.52590257e-01 7.22509921e-01
-1.61028117e-01 1.12794960e+00 -4.43756729e-01 -8.28176141e-01
2.77328342e-01 -1.46108657e-01 6.12387359e-01 1.27809548e+00
5.06429709e-02 1.02078903e+00 4.05873567e-01 6.95064604e-01
1.03833601e-01 2.81316072e-01 -6.73555672e-01 -5.89276850e-01
7.23956883e-01 1.25719237e+00 -6.17851675e-01 -1.62902340e-01
-5.35808623e-01 6.04536176e-01 1.86604023e-01 -2.79403897e-03
-2.15761915e-01 -4.48033422e-01 9.89009500e-01 -7.18758628e-02
-6.54656887e-01 -3.71939898e-01 -1.99035093e-01 -7.32494414e-01
-5.37657201e-01 -1.36623216e+00 1.93800017e-01 -7.27044284e-01
-9.29319084e-01 2.07579751e-02 6.52819127e-02 -5.68564832e-01
-5.77931881e-01 -4.02749866e-01 -1.03937745e+00 7.60315478e-01
-5.94420135e-01 -1.04504466e+00 -2.79557437e-01 2.34370872e-01
1.22714782e+00 -5.91682315e-01 1.12577069e+00 2.14979857e-01
-6.00860417e-01 6.56227112e-01 -3.96972567e-01 3.55242521e-01
6.12305522e-01 -1.50346875e+00 3.39101940e-01 7.23174989e-01
2.06002131e-01 6.67273581e-01 8.36033225e-01 -5.62861800e-01
-1.86739266e+00 -4.76312339e-01 7.71623492e-01 -5.92224956e-01
6.40766561e-01 -5.09555936e-01 -7.00914681e-01 5.64050019e-01
7.73509622e-01 -9.66956973e-01 1.20576882e+00 2.50841528e-01
6.18853830e-02 2.00824663e-01 -1.10440266e+00 5.79200864e-01
4.98269379e-01 -5.74423671e-01 -9.48681414e-01 4.57530022e-01
1.09922719e+00 -8.34091604e-01 -1.25665200e+00 1.89228147e-01
2.75061250e-01 -6.77383423e-01 1.01160443e+00 -7.80477524e-01
9.50313985e-01 3.26509148e-01 1.91116080e-01 -1.38085139e+00
3.51617962e-01 -8.80489945e-01 7.27907643e-02 1.79244804e+00
2.50751376e-01 -1.48946047e-01 9.08650875e-01 1.44224632e+00
-3.89994144e-01 -5.59930265e-01 -6.42120481e-01 9.00922418e-02
3.20903569e-01 -8.37370992e-01 5.11764586e-01 1.18925834e+00
8.50355685e-01 7.24808753e-01 3.67484719e-01 2.90719986e-01
1.31315142e-01 -4.85579312e-01 1.09193802e+00 -1.05555332e+00
-2.06792112e-02 -6.54137492e-01 -5.75503051e-01 -9.07043874e-01
1.20128118e-01 -7.53134608e-01 1.46807536e-01 -1.45167625e+00
-4.20824856e-01 -4.12425250e-01 4.19774115e-01 6.33213460e-01
2.66491231e-02 -4.90373403e-01 2.73854256e-01 -6.43122137e-01
-6.03421867e-01 3.35506707e-01 9.12580788e-01 -2.81227022e-01
-3.67941797e-01 1.69566050e-01 -4.88480508e-01 6.61404610e-01
6.01194501e-01 2.98341122e-02 -8.15925658e-01 -3.20366055e-01
2.00944498e-01 3.59669209e-01 -1.79621145e-01 -1.36993539e+00
8.80197406e-01 -2.03430310e-01 -5.97427413e-02 -4.06694800e-01
2.70572722e-01 -6.03514493e-01 -7.44056702e-01 2.52536684e-01
-1.07436776e+00 -2.95989096e-01 6.30866170e-01 -3.74117643e-01
-1.45569235e-01 -4.75836486e-01 5.84236026e-01 4.16841991e-02
-7.85255075e-01 -2.39624575e-01 -1.09380615e+00 1.71467856e-01
1.13210046e+00 2.57655352e-01 -6.56943262e-01 -9.65016782e-01
-6.34728312e-01 8.80862534e-01 -3.41153890e-01 8.11314642e-01
7.95733988e-01 -7.42796481e-01 -9.27511752e-01 3.85309070e-01
1.41933203e-01 1.42436847e-01 2.28010729e-01 2.00440571e-01
-1.02868867e+00 6.11998260e-01 -1.23427175e-01 -5.88084579e-01
-1.85135937e+00 -2.41637945e-01 3.00267160e-01 -2.65024453e-02
-4.44652766e-01 1.59083784e+00 -3.87649119e-01 -1.12309718e+00
9.45550561e-01 -4.91365761e-01 -6.42729640e-01 9.01802182e-02
1.13638902e+00 7.54903138e-01 2.34793112e-01 -7.41253421e-02
1.38419136e-01 -2.97266811e-01 -1.28082290e-01 -5.26781678e-01
1.46443212e+00 2.34289300e-02 4.72748987e-02 5.52364171e-01
7.54301727e-01 -2.76683774e-02 -3.76708418e-01 -1.65623918e-01
-9.18118842e-03 4.02458906e-02 1.45513996e-01 -1.28553045e+00
-5.46458721e-01 1.35293090e+00 7.24727511e-01 3.29252958e-01
4.90449011e-01 -7.97695071e-02 6.46254122e-01 1.05283487e+00
7.73101673e-02 -9.93229747e-01 2.07559079e-01 9.43728805e-01
6.92153811e-01 -1.35850084e+00 -1.26038745e-01 -3.24599534e-01
-6.01443172e-01 1.52722013e+00 1.24761522e+00 3.50022405e-01
3.75598222e-01 3.72058690e-01 2.55583942e-01 -4.15584505e-01
-1.19130516e+00 -6.59983382e-02 3.14420134e-01 2.56747007e-01
1.32978380e+00 3.23368996e-01 7.19606057e-02 5.77295423e-01
-6.85620964e-01 -2.10742921e-01 8.84656370e-01 1.08332944e+00
-6.97965205e-01 -9.59131658e-01 -3.82089376e-01 1.38260990e-01
-3.10591996e-01 -3.82327974e-01 -8.84712040e-01 5.28455377e-01
-2.84115970e-01 1.64118755e+00 1.74500704e-01 -4.58565921e-01
3.08914393e-01 4.75956291e-01 3.23842257e-01 -1.30946159e+00
-1.21098590e+00 -6.36565328e-01 6.37766957e-01 -6.16014779e-01
-1.05310455e-02 -4.63681251e-01 -1.52744102e+00 -5.66380799e-01
2.12316029e-02 4.74744081e-01 7.67735064e-01 1.05604148e+00
4.96534482e-02 7.43811429e-01 2.33041555e-01 -2.42958263e-01
-3.82477701e-01 -1.12650120e+00 2.73265421e-01 -2.36932620e-01
1.73893929e-01 1.20926440e-01 1.01219006e-01 -1.65225372e-01] | [12.44359016418457, 8.037089347839355] |
56b57f9c-9c46-4174-81b4-523e14c31ccd | pagenet-towards-end-to-end-weakly-supervised | 2207.14807 | null | https://arxiv.org/abs/2207.14807v1 | https://arxiv.org/pdf/2207.14807v1.pdf | PageNet: Towards End-to-End Weakly Supervised Page-Level Handwritten Chinese Text Recognition | Handwritten Chinese text recognition (HCTR) has been an active research topic for decades. However, most previous studies solely focus on the recognition of cropped text line images, ignoring the error caused by text line detection in real-world applications. Although some approaches aimed at page-level text recognition have been proposed in recent years, they either are limited to simple layouts or require very detailed annotations including expensive line-level and even character-level bounding boxes. To this end, we propose PageNet for end-to-end weakly supervised page-level HCTR. PageNet detects and recognizes characters and predicts the reading order between them, which is more robust and flexible when dealing with complex layouts including multi-directional and curved text lines. Utilizing the proposed weakly supervised learning framework, PageNet requires only transcripts to be annotated for real data; however, it can still output detection and recognition results at both the character and line levels, avoiding the labor and cost of labeling bounding boxes of characters and text lines. Extensive experiments conducted on five datasets demonstrate the superiority of PageNet over existing weakly supervised and fully supervised page-level methods. These experimental results may spark further research beyond the realms of existing methods based on connectionist temporal classification or attention. The source code is available at https://github.com/shannanyinxiang/PageNet. | ['Songxuan Lai', 'Canjie Luo', 'Yuliang Liu', 'Lianwen Jin', 'Dezhi Peng'] | 2022-07-29 | null | null | null | null | ['handwritten-chinese-text-recognition', 'line-detection', 'handwritten-chinese-text-recognition'] | ['computer-vision', 'computer-vision', 'natural-language-processing'] | [ 3.91022742e-01 -4.92858768e-01 -4.45689559e-01 -2.74832428e-01
-6.69888079e-01 -5.73485553e-01 4.41607207e-01 -8.07331353e-02
-1.84806749e-01 4.75821912e-01 1.04450598e-01 -4.98699307e-01
3.36158574e-01 -5.16137779e-01 -4.77890223e-01 -7.54007697e-01
2.44937062e-01 2.41982684e-01 5.35750866e-01 3.05585414e-02
4.71500337e-01 3.11443001e-01 -9.94940996e-01 5.84039211e-01
9.77415979e-01 8.81496727e-01 3.21966022e-01 8.43565106e-01
-5.19836962e-01 7.53301382e-01 -7.27346361e-01 -2.30975658e-01
-1.24857217e-01 -6.16411865e-01 -4.83282328e-01 5.49806356e-01
4.54485208e-01 -3.22931230e-01 -4.41023499e-01 1.04915535e+00
5.07717669e-01 -2.12099984e-01 5.59161365e-01 -9.16200280e-01
-8.65354002e-01 8.19748640e-01 -1.04050291e+00 -2.08537355e-02
3.06618154e-01 1.24884620e-01 9.55918729e-01 -1.17181277e+00
3.08844000e-01 8.40489507e-01 6.08097136e-01 3.11616361e-01
-6.29566848e-01 -4.97618467e-01 4.76026833e-01 2.97595620e-01
-1.21684813e+00 -1.21844254e-01 7.71628737e-01 -2.60933399e-01
8.98531616e-01 2.89585948e-01 4.00529891e-01 1.25261259e+00
-5.64809814e-02 1.54166758e+00 1.02992356e+00 -6.54683888e-01
-2.66675707e-02 -1.18875936e-01 5.47028661e-01 8.77225161e-01
1.85678691e-01 -4.99728322e-01 -6.15368545e-01 2.97816485e-01
8.25152338e-01 6.77204551e-03 -5.60533166e-01 -1.96459696e-01
-1.33740151e+00 5.25777221e-01 1.58778906e-01 5.35604477e-01
-5.21340966e-02 -1.65754676e-01 6.01763427e-01 -1.73796952e-01
4.51894552e-01 -1.30930059e-02 -4.11262035e-01 -3.86604905e-01
-1.24901152e+00 -3.56441855e-01 7.89972425e-01 1.32248724e+00
4.31036204e-01 2.46752858e-01 -2.13413417e-01 1.07835615e+00
1.81543007e-01 5.09984553e-01 6.54330909e-01 -3.74478772e-02
9.98988748e-01 6.36681616e-01 -4.21363954e-03 -9.97995317e-01
-3.94869268e-01 -3.72416973e-01 -1.05866337e+00 -2.53927022e-01
5.61598301e-01 -2.75806427e-01 -1.06175673e+00 7.69525230e-01
-1.60440192e-01 -1.71551675e-01 -1.40723377e-01 9.68651295e-01
5.89915156e-01 9.33299839e-01 -2.62946904e-01 -1.16691284e-01
1.48592329e+00 -1.60847366e+00 -9.91784811e-01 -4.45440918e-01
7.74637163e-01 -9.96950805e-01 1.57669294e+00 4.80211347e-01
-8.25705349e-01 -3.58025223e-01 -1.22527599e+00 -1.90985739e-01
-3.10003549e-01 9.71209347e-01 5.04088640e-01 5.32609940e-01
-6.32813871e-01 1.29538774e-01 -8.53011072e-01 -5.21075964e-01
4.22434479e-01 -2.79807430e-02 1.55530479e-02 -1.82465062e-01
-9.77157772e-01 4.32864219e-01 2.88940430e-01 7.09132910e-01
-3.33499074e-01 -1.75232291e-01 -6.84419334e-01 1.70301795e-01
5.39604187e-01 9.10249725e-02 1.33414042e+00 -1.09612513e+00
-1.56609178e+00 4.63016003e-01 -2.46784851e-01 -1.74914494e-01
1.06466043e+00 -5.50280333e-01 -5.71556687e-01 1.85617611e-01
-1.49419188e-01 3.90077114e-01 7.57548511e-01 -1.00433803e+00
-6.27500534e-01 -1.89678922e-01 -5.96401036e-01 3.69043231e-01
-7.56381035e-01 -6.77500591e-02 -1.13762498e+00 -9.50244009e-01
3.55949163e-01 -7.12146401e-01 1.31846741e-01 1.52272135e-01
-8.57459128e-01 -2.64300406e-01 1.36778438e+00 -8.90243769e-01
1.20767486e+00 -2.07621193e+00 -5.08728802e-01 -1.07080661e-01
-2.05255747e-01 4.51471388e-01 -1.57273218e-01 4.43386793e-01
1.11918822e-01 2.26228938e-01 -2.10447267e-01 -1.29115835e-01
-7.31185898e-02 -2.32696533e-01 -5.21170020e-01 5.11827230e-01
1.83189020e-01 9.51952338e-01 -5.73669672e-01 -5.86329997e-01
1.63399562e-01 2.60900825e-01 2.05674762e-04 -3.37886140e-02
-4.28318381e-01 1.23663293e-02 -4.88272369e-01 8.53842258e-01
7.25974560e-01 -4.48979199e-01 2.39117503e-01 -2.17936113e-01
-2.67603219e-01 5.78329861e-02 -9.10209894e-01 1.36918354e+00
-1.23033211e-01 1.14404953e+00 -3.64510268e-01 -7.48713434e-01
1.21723032e+00 1.83457971e-01 1.14159673e-01 -9.19867992e-01
1.16261177e-01 1.85931191e-01 -7.86942989e-02 -5.49451232e-01
7.70909667e-01 4.53001112e-01 -2.00471133e-01 3.84643555e-01
-4.38220084e-01 2.10771874e-01 2.88729846e-01 9.27883387e-02
8.09105635e-01 4.10704017e-01 9.02037024e-02 -4.97780181e-02
4.93483335e-01 1.48015484e-01 5.08950949e-01 7.20247686e-01
-3.24686855e-01 8.37556005e-01 4.68635976e-01 -2.67454833e-01
-1.22167313e+00 -6.92920864e-01 -8.89706016e-02 1.11869609e+00
3.43397230e-01 -2.94565082e-01 -8.02390575e-01 -7.31409788e-01
-3.85964334e-01 5.35949588e-01 -3.49531323e-01 3.88651639e-01
-8.05462122e-01 -6.38838232e-01 9.07269299e-01 8.42242479e-01
1.10549474e+00 -1.04819191e+00 -2.62537092e-01 9.58528295e-02
-3.45130384e-01 -1.44139719e+00 -9.48684633e-01 2.14223057e-01
-8.85684431e-01 -9.00125206e-01 -1.26418650e+00 -1.30678833e+00
9.25251007e-01 2.84926325e-01 5.73106229e-01 1.64211556e-01
-3.70777875e-01 1.22517496e-01 -5.58711767e-01 -2.06616104e-01
5.45332544e-02 3.56766552e-01 -2.43163541e-01 4.82950695e-02
4.11628783e-01 2.05485120e-01 -5.22900879e-01 6.01060212e-01
-7.02011168e-01 4.61173505e-01 8.93370569e-01 9.46915746e-01
3.29108834e-01 1.41879514e-01 1.43683478e-01 -1.03449810e+00
5.98145545e-01 1.50642633e-01 -7.39398599e-01 5.98809540e-01
-4.60561275e-01 -2.29136676e-01 8.78039122e-01 -6.05460823e-01
-1.37781799e+00 1.32655367e-01 3.73744103e-03 -6.67252019e-02
-2.80884266e-01 5.16308486e-01 -2.45683253e-01 1.75498620e-01
3.15272450e-01 7.79738843e-01 -5.43909609e-01 -3.90145570e-01
1.77011698e-01 9.59839761e-01 5.70838153e-01 -4.57179040e-01
7.55018413e-01 3.04104239e-01 -5.04198074e-01 -1.20707870e+00
-7.41288722e-01 -3.93040806e-01 -9.15770948e-01 -2.78780431e-01
7.47558713e-01 -6.43010020e-01 -5.86532235e-01 1.05048215e+00
-1.00125062e+00 -6.88744426e-01 3.54652941e-01 3.31053704e-01
-2.21820667e-01 9.40092921e-01 -9.87401426e-01 -7.78828144e-01
-3.11225265e-01 -8.77730489e-01 9.76588309e-01 3.53349537e-01
-3.88279781e-02 -1.02385151e+00 -3.47121626e-01 4.18829173e-01
1.34602427e-01 -1.96518406e-01 9.38740551e-01 -4.79963899e-01
-6.80226445e-01 -4.76531059e-01 -5.44828236e-01 1.37491748e-01
1.41887128e-01 3.85936201e-01 -7.52566397e-01 -1.04217075e-01
-4.56505299e-01 -4.12761003e-01 9.00467455e-01 8.89555886e-02
1.32374060e+00 -4.12802190e-01 -2.88362682e-01 4.60951746e-01
1.33211648e+00 3.56225699e-01 6.87948644e-01 4.78074402e-01
1.03767872e+00 3.95877331e-01 8.09548974e-01 4.61112112e-01
2.07558140e-01 4.54519868e-01 -1.11019380e-01 -4.41730380e-01
-8.18077698e-02 -4.65173781e-01 5.09386837e-01 1.05574620e+00
3.19830626e-01 -7.69545138e-01 -1.24184477e+00 3.98681879e-01
-1.95074332e+00 -7.30885088e-01 -4.81797099e-01 1.67199779e+00
7.60845542e-01 4.30091560e-01 -1.60985544e-01 2.61921644e-01
1.07439542e+00 3.38953584e-01 -7.71550894e-01 -3.66626352e-01
-5.24713278e-01 -5.31186104e-01 6.65161490e-01 1.03017226e-01
-1.21185088e+00 1.36186934e+00 5.65572405e+00 9.96392429e-01
-1.43878436e+00 -3.99635285e-01 8.69932890e-01 2.70575017e-01
1.13488227e-01 -1.47051319e-01 -1.03277802e+00 4.38278019e-01
3.72482479e-01 3.08694929e-01 3.73752676e-02 7.86855102e-01
4.44937795e-01 -2.78220356e-01 -9.05756891e-01 9.99561071e-01
3.09733272e-01 -1.14670873e+00 1.32885709e-01 -1.95649192e-01
8.94970059e-01 -2.62326479e-01 1.58023790e-01 1.37686908e-01
9.26396474e-02 -8.42212439e-01 6.81227684e-01 3.12345684e-01
7.66258419e-01 -4.96670276e-01 6.68508649e-01 4.73489046e-01
-1.33317745e+00 9.27323103e-02 -3.73867542e-01 2.81487126e-03
-2.77377889e-02 5.24025619e-01 -8.94775569e-01 2.88988590e-01
5.60338557e-01 8.22285056e-01 -7.21609950e-01 1.08415520e+00
-6.09446466e-01 1.16010463e+00 -4.19487879e-02 -5.20554900e-01
5.62987268e-01 -1.12649217e-01 -3.24382372e-02 1.64910996e+00
2.34546751e-01 -3.53373848e-02 3.12194645e-01 4.61608708e-01
-1.22151919e-01 4.27620500e-01 -2.07715899e-01 -3.16176444e-01
3.01301450e-01 1.30393064e+00 -1.38129795e+00 -3.24203163e-01
-5.78955114e-01 1.38707566e+00 1.79051414e-01 6.04482830e-01
-8.63680422e-01 -8.24856162e-01 -2.89392304e-02 -2.62359411e-01
5.68715215e-01 -5.14703751e-01 -7.37314701e-01 -1.38185561e+00
3.28654140e-01 -8.21757734e-01 2.75273502e-01 -9.01116848e-01
-1.13549078e+00 4.87848490e-01 -6.72061086e-01 -1.27975416e+00
1.92390859e-01 -8.07563007e-01 -7.87863135e-01 6.24365330e-01
-1.37525845e+00 -1.17335260e+00 -5.39645195e-01 4.09771442e-01
1.11287451e+00 6.14686199e-02 4.21663076e-01 -7.63041899e-03
-1.02077663e+00 8.31707478e-01 4.68905121e-01 9.64304149e-01
8.47025394e-01 -1.14734697e+00 5.41811824e-01 1.15244305e+00
2.26231515e-01 3.25399131e-01 3.98914993e-01 -7.99693584e-01
-1.43410051e+00 -1.14107919e+00 8.51784468e-01 -6.21276535e-02
6.39950335e-01 -7.77725220e-01 -1.11440408e+00 4.87322301e-01
3.56582224e-01 -1.57479763e-01 4.52227384e-01 -5.43599464e-02
-3.45369399e-01 5.92422709e-02 -5.52261293e-01 9.70954120e-01
9.22413766e-01 -4.93713498e-01 -2.31247723e-01 5.09057403e-01
3.21546257e-01 -4.71104771e-01 -5.33267915e-01 2.02108636e-01
4.61236626e-01 -7.27965891e-01 2.92215288e-01 -1.30197719e-01
4.72134143e-01 -4.25542295e-01 3.21986735e-01 -7.99228609e-01
-2.69295692e-01 -2.97673255e-01 1.33368850e-01 1.42282999e+00
6.82150960e-01 -3.61079723e-01 1.05207276e+00 2.79098153e-01
-1.86201453e-01 -6.76326990e-01 -3.50841552e-01 -7.47495353e-01
-2.34689303e-02 -4.62695390e-01 1.57140568e-01 8.78241479e-01
1.98974162e-01 2.85296619e-01 -6.24385655e-01 2.74519175e-01
3.79922807e-01 3.65615994e-01 5.07750630e-01 -6.28098011e-01
-9.49457362e-02 -6.79160178e-01 -1.01271652e-01 -1.63239706e+00
1.20110987e-02 -6.01564586e-01 6.28526688e-01 -1.67465401e+00
1.83986619e-01 -2.21141115e-01 1.79011568e-01 7.57919550e-01
-1.85319513e-01 3.92794043e-01 2.50500381e-01 4.28132236e-01
-8.13701987e-01 5.30758917e-01 1.34918594e+00 -3.98834348e-01
-1.10233471e-01 -1.08821742e-01 -2.97172368e-01 9.03241694e-01
9.52753842e-01 2.89815627e-02 -1.72297373e-01 -6.66415572e-01
-5.50230294e-02 5.90715511e-03 -2.31609792e-02 -8.77058864e-01
7.39578605e-01 -1.42960008e-02 8.69585097e-01 -1.22614539e+00
-1.03056990e-01 -4.78451073e-01 -7.17962563e-01 3.54696542e-01
-5.68238974e-01 4.18421999e-02 2.12087691e-01 5.86452007e-01
-2.91217446e-01 -2.40543604e-01 6.73863113e-01 7.23987445e-02
-1.02953517e+00 2.97367126e-01 -7.10076988e-01 -9.08779576e-02
9.74157095e-01 -4.85179752e-01 -5.17704427e-01 -3.84256423e-01
-2.95846850e-01 2.60025978e-01 4.34891909e-01 6.38774216e-01
7.52536058e-01 -9.13608491e-01 -5.82400918e-01 4.54510003e-03
2.67124176e-01 -3.84104401e-02 2.36100182e-01 6.85340941e-01
-8.99319649e-01 8.31361830e-01 7.85472691e-02 -6.84091508e-01
-1.34581017e+00 4.23294067e-01 8.32651630e-02 -1.77579924e-01
-7.93690622e-01 5.59965909e-01 1.77799612e-01 -1.55081674e-01
6.53138280e-01 -3.59776109e-01 -1.63574308e-01 1.88840460e-02
4.98254657e-01 2.28077561e-01 -3.80458049e-02 -3.45363319e-01
-1.63902044e-01 7.73136616e-01 -4.59284335e-01 1.87822897e-02
1.07609057e+00 -1.04459956e-01 1.25185058e-01 6.22907758e-01
9.61102307e-01 -4.57684174e-02 -1.54815960e+00 -2.83621997e-01
3.84089023e-01 -2.96771049e-01 -1.92484856e-01 -9.38849807e-01
-9.60052729e-01 1.26289046e+00 3.74245882e-01 -8.17958862e-02
1.00429535e+00 -4.04222548e-01 7.83588648e-01 6.99637532e-01
1.27247363e-01 -1.36580706e+00 3.87701094e-01 7.25659728e-01
6.59554362e-01 -1.17750001e+00 -1.27762944e-01 -4.85917687e-01
-8.62074196e-01 1.64867330e+00 1.00565004e+00 1.27000451e-01
2.40362912e-01 5.52538693e-01 3.46261889e-01 2.64259011e-01
-6.31306052e-01 -4.32843342e-03 1.40622765e-01 3.47273141e-01
7.73296833e-01 -3.29787359e-02 -2.93527007e-01 4.24571395e-01
1.03275597e-01 -1.79898977e-01 6.47705019e-01 1.03089941e+00
-4.85117555e-01 -9.25384104e-01 -4.78070527e-01 4.72595990e-01
-3.47680658e-01 -4.03210729e-01 -6.28328562e-01 6.81972742e-01
-3.13888490e-01 7.90562093e-01 1.02771968e-01 -1.75246313e-01
1.15999050e-01 6.45474670e-03 1.73402891e-01 -3.34425956e-01
-2.69149959e-01 5.62231898e-01 5.50396787e-03 4.38442966e-03
-3.58664803e-02 -7.55616903e-01 -1.32280242e+00 -4.03356627e-02
-6.42559886e-01 1.40246861e-02 6.02073610e-01 7.47051239e-01
7.71513954e-02 4.78490174e-01 5.13090253e-01 -6.13141954e-01
-4.53763127e-01 -1.04348695e+00 -5.68763018e-01 4.59171683e-02
7.82086402e-02 3.26753817e-02 -5.96828386e-02 2.81485170e-01] | [11.942553520202637, 2.257850408554077] |
1b662b06-07d7-471f-a82b-afbeff508ef8 | prompt-learning-for-action-recognition | 2305.12437 | null | https://arxiv.org/abs/2305.12437v1 | https://arxiv.org/pdf/2305.12437v1.pdf | Prompt Learning for Action Recognition | We present a new general learning approach for action recognition, Prompt Learning for Action Recognition (PLAR), which leverages the strengths of prompt learning to guide the learning process. Our approach is designed to predict the action label by helping the models focus on the descriptions or instructions associated with actions in the input videos. Our formulation uses various prompts, including optical flow, large vision models, and learnable prompts to improve the recognition performance. Moreover, we propose a learnable prompt method that learns to dynamically generate prompts from a pool of prompt experts under different inputs. By sharing the same objective, our proposed PLAR can optimize prompts that guide the model's predictions while explicitly learning input-invariant (prompt experts pool) and input-specific (data-dependent) prompt knowledge. We evaluate our approach on datasets consisting of both ground camera videos and aerial videos, and scenes with single-agent and multi-agent actions. In practice, we observe a 3.17-10.2% accuracy improvement on the aerial multi-agent dataset, Okutamam and 0.8-2.6% improvement on the ground camera single-agent dataset, Something Something V2. We plan to release our code on the WWW. | ['Dinesh Manocha', 'Tianrui Guan', 'Ruiqi Xian', 'Xijun Wang'] | 2023-05-21 | null | null | null | null | ['action-recognition-in-videos'] | ['computer-vision'] | [ 4.45891649e-01 3.25972028e-02 -4.23785895e-01 -3.79040897e-01
-9.95101810e-01 -7.17084289e-01 6.91612601e-01 -3.09942275e-01
-4.05317754e-01 4.93770212e-01 6.82412326e-01 -5.01130102e-03
-1.31775111e-01 -2.86358982e-01 -6.76617444e-01 -5.60911357e-01
-5.39995693e-02 3.85136902e-01 3.22359622e-01 1.42805595e-02
4.34640437e-01 5.15795887e-01 -1.30260742e+00 8.06334376e-01
5.38461685e-01 8.68018448e-01 3.73890072e-01 1.28355694e+00
2.11732417e-01 1.97590756e+00 -3.71300280e-01 3.13281342e-02
5.88225007e-01 -3.61075193e-01 -1.10513067e+00 5.77428579e-01
8.59655738e-01 -1.13855350e+00 -7.79719830e-01 5.75607240e-01
2.59913445e-01 4.93586242e-01 6.25725567e-01 -1.51000190e+00
-6.70626104e-01 3.67677420e-01 -2.68141955e-01 2.25847721e-01
7.77891994e-01 1.05115747e+00 1.01045263e+00 -6.37513041e-01
6.06663525e-01 1.38634229e+00 1.63807571e-01 1.04823923e+00
-7.72971332e-01 -3.53083491e-01 6.60001874e-01 5.33568144e-01
-7.19920337e-01 -6.59672916e-01 7.22635865e-01 -8.23526502e-01
1.02477276e+00 -1.03360275e-03 3.14666331e-01 1.41971219e+00
-2.62230150e-02 1.34585631e+00 1.04039896e+00 -1.09677523e-01
1.34640500e-01 -3.64802480e-01 8.84683654e-02 8.95760059e-01
-2.98422396e-01 6.02148414e-01 -7.45192647e-01 -4.99410182e-02
9.24281597e-01 1.53928250e-01 -4.30739850e-01 -6.18838109e-02
-1.40944338e+00 6.58373892e-01 1.51316613e-01 -3.76782864e-01
-6.56629980e-01 4.20026600e-01 1.73781201e-01 2.62759894e-01
-7.17861652e-02 6.45788193e-01 -6.98289573e-01 -3.68249029e-01
-4.76877451e-01 4.04668331e-01 6.21717155e-01 9.34372544e-01
8.67027581e-01 2.66630530e-01 -7.28026211e-01 3.16360444e-01
4.28188235e-01 8.20504010e-01 4.24191684e-01 -1.74543500e+00
6.52244627e-01 6.58756316e-01 5.66505492e-01 -9.56070185e-01
-2.43993685e-01 2.17649534e-01 -4.40257341e-01 2.91273862e-01
5.34171581e-01 -5.03544748e-01 -1.10729671e+00 1.57651603e+00
2.32185245e-01 7.37661779e-01 4.43514317e-01 1.00101483e+00
9.17770386e-01 8.74051750e-01 3.41913521e-01 -6.91447556e-02
8.38934004e-01 -1.62866402e+00 -4.33021575e-01 -4.45880383e-01
6.89085960e-01 -5.95446289e-01 1.13797438e+00 3.96521598e-01
-9.03251708e-01 -8.72539937e-01 -4.76037860e-01 -8.13936163e-03
1.37223721e-01 5.43395996e-01 6.20893776e-01 -1.86805993e-01
-1.09255552e+00 5.23869693e-01 -9.49808598e-01 -2.34858721e-01
3.66727859e-01 2.84431934e-01 -3.41593713e-01 -3.58706594e-01
-5.54753006e-01 6.16757929e-01 1.60116702e-01 -2.89937913e-01
-1.73676956e+00 -6.92451656e-01 -7.97866702e-01 -1.23000041e-01
8.59335184e-01 -6.00620449e-01 1.79366410e+00 -1.21967185e+00
-1.91067147e+00 4.79817241e-01 2.64287349e-02 -4.97588962e-01
5.23531556e-01 -6.67124271e-01 -1.25725314e-01 6.27323568e-01
1.95918838e-03 1.08580852e+00 9.82305586e-01 -1.09887099e+00
-1.07060671e+00 1.62711944e-02 8.35369527e-01 2.94504762e-01
2.78664082e-02 1.58639908e-01 -1.34167165e-01 -4.34987664e-01
-5.28006971e-01 -1.10182929e+00 -5.67160547e-01 1.62032731e-02
-1.50301516e-01 -3.17478567e-01 7.98172951e-01 -5.30728519e-01
7.62012720e-01 -2.07379746e+00 1.88825712e-01 -4.64513063e-01
1.15181230e-01 4.29919034e-01 -9.14065540e-01 4.03306812e-01
7.06553757e-02 -3.89277160e-01 1.57574251e-01 -1.28165931e-01
-1.03225119e-01 2.98284978e-01 -6.77248299e-01 2.65601516e-01
4.78340685e-01 8.74233365e-01 -1.26864350e+00 -2.79760301e-01
4.05168772e-01 1.02690622e-01 -7.82727897e-01 9.43415523e-01
-6.75913036e-01 7.66264439e-01 -7.13151097e-01 7.64165401e-01
-5.93507141e-02 -4.42318916e-01 -4.65918938e-03 -1.91963211e-01
4.49425802e-02 8.70037600e-02 -1.27726877e+00 1.53417456e+00
-2.41752714e-01 5.54894567e-01 -9.19506997e-02 -7.93092847e-01
8.26895356e-01 4.77787137e-01 7.38153279e-01 -4.85039115e-01
-2.01339424e-01 -6.07485995e-02 -9.47620720e-02 -1.10413718e+00
4.81557012e-01 4.53866184e-01 1.77709088e-01 5.32816947e-01
2.54827976e-01 1.70077145e-01 2.60776162e-01 3.02644789e-01
1.50062501e+00 5.09892285e-01 4.06403840e-01 1.51188403e-01
5.87629795e-01 2.96799749e-01 6.23663366e-01 9.51129496e-01
-5.99935114e-01 3.68007213e-01 2.55212545e-01 -9.75846946e-01
-5.38632572e-01 -5.88595450e-01 8.48610818e-01 1.34164548e+00
1.10165395e-01 -4.73801136e-01 -3.93116266e-01 -1.17470789e+00
-1.05950505e-01 4.81697500e-01 -5.64595103e-01 7.73566077e-03
-7.95495093e-01 -4.39789109e-02 2.52370983e-01 9.02185798e-01
4.31783497e-01 -1.60012376e+00 -1.01910245e+00 8.25402066e-02
-2.07342654e-01 -1.41397119e+00 -7.53498435e-01 -2.06670947e-02
-8.08827400e-01 -1.55868077e+00 -4.93112355e-01 -4.09290701e-01
7.41947711e-01 4.95942980e-01 1.05632401e+00 8.44066739e-02
2.77547911e-02 1.37093306e+00 -6.17047012e-01 8.64622369e-02
-4.01227832e-01 -2.21594930e-01 1.53289706e-01 3.35538566e-01
1.56797886e-01 -1.65915713e-01 -5.76096773e-01 4.04747188e-01
-7.18180418e-01 2.47890681e-01 7.47085690e-01 6.89959526e-01
4.97465938e-01 -5.03490746e-01 3.17394465e-01 -6.36120617e-01
3.02960813e-01 -3.35746020e-01 -6.45449042e-01 3.36629122e-01
-2.50995576e-01 2.26568639e-01 7.86583483e-01 -6.92788184e-01
-1.05379903e+00 6.87291861e-01 1.81495488e-01 -7.55875051e-01
-8.16392064e-01 1.64238378e-01 4.50819964e-04 -1.37304254e-02
6.68697655e-01 1.72585964e-01 -3.80754441e-01 -2.85826862e-01
5.49898088e-01 5.12814343e-01 6.81790292e-01 -8.23637843e-01
6.74538672e-01 3.10483456e-01 -1.40995815e-01 -5.10636389e-01
-1.02315009e+00 -5.93719721e-01 -4.88683730e-01 -5.71606040e-01
9.82192159e-01 -1.16883779e+00 -9.21659052e-01 5.99494457e-01
-1.36475182e+00 -1.23596752e+00 -3.47834349e-01 5.79715252e-01
-1.03714693e+00 1.48276687e-01 -5.16256154e-01 -6.38806760e-01
-1.99849024e-01 -1.34584308e+00 1.12569332e+00 4.67572540e-01
-2.12768078e-01 -8.67044091e-01 2.60035932e-01 5.04618704e-01
1.62624449e-01 9.86137018e-02 2.92700380e-01 -7.71929443e-01
-1.18254197e+00 1.69277981e-01 -4.37816046e-02 4.44709301e-01
3.74076158e-01 9.32674184e-02 -7.57676125e-01 -3.09731275e-01
-3.82218212e-01 -8.43327403e-01 6.84410751e-01 3.15016896e-01
1.10181510e+00 -6.95069373e-01 -1.73523530e-01 5.50462246e-01
1.22837114e+00 5.59341669e-01 5.63872695e-01 9.72008556e-02
8.38612258e-01 5.33944964e-01 9.56593812e-01 5.66056907e-01
5.54677129e-01 6.67622089e-01 4.85078752e-01 1.86015502e-01
-3.94850075e-01 -4.48560476e-01 1.02568626e+00 3.12098116e-01
-3.36339980e-01 -2.68121779e-01 -7.41052806e-01 4.53195125e-01
-2.21503878e+00 -1.13319826e+00 -6.87834749e-04 1.66134071e+00
6.73903823e-01 -2.22424462e-01 3.15162957e-01 -4.89192784e-01
4.30286765e-01 4.24687088e-01 -8.03737223e-01 1.87447015e-02
2.78618753e-01 -1.77903101e-01 3.05403411e-01 8.06986749e-01
-1.29529369e+00 1.14538050e+00 6.31166983e+00 2.78645873e-01
-1.08279550e+00 -2.60104239e-01 4.07902658e-01 -4.31247950e-02
8.06147531e-02 1.24897867e-01 -1.02458572e+00 3.48941773e-01
8.07067096e-01 1.23072276e-02 5.94743669e-01 1.07521677e+00
5.55547833e-01 -4.19275910e-02 -1.52724612e+00 1.21678126e+00
2.17623636e-01 -1.43778980e+00 1.55420840e-01 -1.21106490e-01
8.35808814e-01 1.37400078e-02 -1.52188703e-01 4.89856392e-01
8.14873338e-01 -7.72910297e-01 5.03719628e-01 6.31949902e-01
6.02274835e-01 -1.38850555e-01 3.14968944e-01 3.37228328e-01
-1.07210767e+00 -5.36572158e-01 -9.64022055e-02 -4.45922971e-01
2.75742173e-01 -2.53430814e-01 -8.25792611e-01 2.83971667e-01
3.60531777e-01 1.22283351e+00 -6.41123414e-01 9.25581038e-01
-6.01172268e-01 6.75663650e-01 1.49948359e-01 1.59969792e-01
5.38698256e-01 -2.30829455e-02 4.64849353e-01 1.02798998e+00
5.56820482e-02 5.53092301e-01 1.01383662e+00 3.22140545e-01
5.29037081e-02 -1.72380775e-01 -6.25440657e-01 -1.34615630e-01
2.08354369e-01 1.27072942e+00 -2.74912477e-01 -5.86849034e-01
-5.68340957e-01 9.87859011e-01 2.71875590e-01 5.75926781e-01
-7.54451334e-01 1.64417863e-01 7.30523586e-01 -1.04624964e-01
2.81373054e-01 -2.51446098e-01 3.42008442e-01 -1.23664188e+00
-2.80901641e-01 -1.41885698e+00 4.87303227e-01 -1.00940406e+00
-1.10668671e+00 5.25896192e-01 1.20188266e-01 -1.35775685e+00
-5.79025090e-01 -9.61728215e-01 -6.42809451e-01 3.50081533e-01
-1.60240400e+00 -1.04800141e+00 -5.67247689e-01 8.36448312e-01
1.15743697e+00 -5.11847734e-01 7.91198015e-01 4.09412459e-02
-4.78408992e-01 1.63911313e-01 -5.86749792e-01 3.58343810e-01
8.52895379e-01 -1.24242818e+00 2.03490689e-01 8.75369608e-01
5.07967293e-01 9.22355205e-02 4.31615353e-01 -5.60613513e-01
-1.53235006e+00 -1.29661942e+00 4.61572856e-01 -7.41819203e-01
6.33439541e-01 2.48370051e-01 -6.19436622e-01 1.10457540e+00
4.32436824e-01 2.64942616e-01 5.40837288e-01 -2.24800408e-01
-3.04431409e-01 -8.08180645e-02 -7.47864366e-01 6.58730626e-01
1.18134785e+00 -4.25140768e-01 -7.05756605e-01 5.52136838e-01
7.80956328e-01 -6.52664542e-01 -5.54756105e-01 3.44354808e-01
2.93887377e-01 -7.86438704e-01 8.13609719e-01 -1.29539156e+00
6.86834037e-01 -4.95366991e-01 -3.79334271e-01 -1.28149426e+00
-5.23069739e-01 -9.33822274e-01 -5.46140730e-01 9.49333608e-01
9.06391442e-02 -1.99872807e-01 7.46834695e-01 7.97210515e-01
-3.04528803e-01 -6.27309561e-01 -3.56744111e-01 -6.79051399e-01
-4.67659533e-01 -2.37340912e-01 3.17389280e-01 6.84289634e-01
-3.49731684e-01 4.24261481e-01 -7.90089130e-01 3.26130480e-01
3.07221383e-01 2.10285634e-01 1.26618111e+00 -8.41334879e-01
-7.99202621e-01 4.50698622e-02 -1.80145144e-01 -1.64543951e+00
4.43862557e-01 -5.95233202e-01 2.85792261e-01 -1.48196387e+00
2.41811812e-01 -1.84461549e-01 -2.64969617e-01 1.02542579e+00
-3.56160760e-01 -3.11546504e-01 6.52904987e-01 1.99106351e-01
-1.22005117e+00 4.94259626e-01 1.43495464e+00 -3.06659728e-01
-3.08701247e-01 9.44249630e-02 -5.30448198e-01 8.56707156e-01
6.67071998e-01 -3.84033948e-01 -6.47744775e-01 -8.01578224e-01
-3.38475406e-01 3.40977848e-01 6.78853393e-01 -1.03380919e+00
3.15786153e-01 -9.20421004e-01 1.09001197e-01 -3.22201103e-01
2.88337469e-01 -7.91962624e-01 -2.37766385e-01 4.12534803e-01
-8.11864257e-01 1.07260436e-01 1.44817969e-02 6.45147681e-01
-6.16702996e-03 -1.26663581e-01 5.34571111e-01 -3.65901977e-01
-1.41980290e+00 6.84233129e-01 -3.81568611e-01 3.06842983e-01
1.11532593e+00 1.53111741e-01 -6.35527372e-01 -6.25868559e-01
-5.45931637e-01 4.99314219e-01 1.87684938e-01 5.79423964e-01
8.28610718e-01 -1.23539722e+00 -8.68921936e-01 -6.67687282e-02
1.54731423e-01 -6.68382645e-02 2.75458902e-01 5.56216657e-01
-1.55655056e-01 4.46006656e-01 -3.35648268e-01 -6.14979565e-01
-1.11888039e+00 5.81958115e-01 3.68241727e-01 -3.71511847e-01
-6.60946131e-01 8.05519164e-01 3.14045936e-01 -2.64499187e-01
4.73690867e-01 -3.99999499e-01 -1.70037806e-01 -2.00648934e-01
8.82711112e-01 5.15182734e-01 -6.12849474e-01 -4.53176707e-01
-2.58604854e-01 5.82073987e-01 -2.35053957e-01 2.77102459e-04
1.38324153e+00 1.83204100e-01 4.53949541e-01 1.71370551e-01
8.32012832e-01 -2.72216678e-01 -2.32455182e+00 -2.46675506e-01
-8.74206573e-02 -6.41874254e-01 -2.40686521e-01 -1.09000802e+00
-1.06578732e+00 6.80288315e-01 4.84056771e-01 -2.52437323e-01
1.14750087e+00 -1.55636650e-02 6.09857082e-01 7.82271028e-01
4.22350377e-01 -9.65846956e-01 1.03119576e+00 6.28811896e-01
1.04862607e+00 -1.49924946e+00 -2.30703339e-01 -4.89674369e-03
-1.11653006e+00 1.15542161e+00 1.20744371e+00 -3.80097657e-01
4.20666039e-02 2.33085662e-01 4.89359915e-01 -1.04192808e-01
-1.22065270e+00 -2.56263614e-01 2.66199708e-01 8.32206428e-01
5.42822946e-03 -6.28113076e-02 2.81477809e-01 5.19096553e-01
5.48973024e-01 3.49350750e-01 8.29786241e-01 8.88829112e-01
-4.21324313e-01 -1.11170805e+00 -1.71999663e-01 1.72491610e-01
-8.68203714e-02 1.53854862e-01 -4.51060742e-01 5.05553246e-01
-1.08829014e-01 1.07235157e+00 -8.01411867e-02 -5.84330678e-01
4.62758988e-01 5.21448953e-03 5.12347043e-01 -7.60520697e-01
-3.37739855e-01 -2.62090474e-01 9.30891111e-02 -1.26630676e+00
-8.66123796e-01 -7.70437777e-01 -1.24616289e+00 1.65797621e-01
2.72064656e-01 -2.25784466e-01 -9.38296542e-02 1.09815121e+00
5.39509475e-01 5.34306526e-01 7.10114717e-01 -8.73715222e-01
-8.34904432e-01 -8.05389285e-01 -8.69938731e-02 5.50699532e-01
6.14701450e-01 -7.10673451e-01 -2.84334689e-01 6.88312232e-01] | [8.499565124511719, 0.6555520296096802] |
43cab4e2-0e9a-416b-b2f9-4cc862c0e690 | bts-net-bi-directional-transfer-and-selection | 2104.01784 | null | https://arxiv.org/abs/2104.01784v1 | https://arxiv.org/pdf/2104.01784v1.pdf | BTS-Net: Bi-directional Transfer-and-Selection Network For RGB-D Salient Object Detection | Depth information has been proved beneficial in RGB-D salient object detection (SOD). However, depth maps obtained often suffer from low quality and inaccuracy. Most existing RGB-D SOD models have no cross-modal interactions or only have unidirectional interactions from depth to RGB in their encoder stages, which may lead to inaccurate encoder features when facing low quality depth. To address this limitation, we propose to conduct progressive bi-directional interactions as early in the encoder stage, yielding a novel bi-directional transfer-and-selection network named BTS-Net, which adopts a set of bi-directional transfer-and-selection (BTS) modules to purify features during encoding. Based on the resulting robust encoder features, we also design an effective light-weight group decoder to achieve accurate final saliency prediction. Comprehensive experiments on six widely used datasets demonstrate that BTS-Net surpasses 16 latest state-of-the-art approaches in terms of four key metrics. | ['Qijun Zhao', 'Keren Fu', 'Yao Jiang', 'Wenbo Zhang'] | 2021-04-05 | null | null | null | null | ['rgb-d-salient-object-detection'] | ['computer-vision'] | [ 4.67487901e-01 4.64347750e-02 -3.31184000e-01 -4.77591991e-01
-4.35432792e-01 2.57658828e-02 5.62904656e-01 -1.63162604e-01
-1.59000218e-01 5.38888752e-01 5.02183855e-01 5.70613099e-03
5.93358837e-02 -7.93410540e-01 -7.06427932e-01 -5.06818295e-01
1.93615004e-01 -3.22635442e-01 8.89616489e-01 -3.82709593e-01
3.59561443e-01 1.89441070e-01 -1.75092041e+00 4.36860085e-01
1.03658450e+00 1.32927132e+00 6.39776289e-01 2.44040340e-01
-1.47782834e-04 1.08182776e+00 -1.91876978e-01 -2.68609554e-01
3.54131997e-01 -3.36547643e-01 -6.31068766e-01 -9.51599255e-02
2.58022636e-01 -8.86198223e-01 -6.98756933e-01 1.11237240e+00
7.75751650e-01 -2.15052858e-01 2.37866744e-01 -1.22410607e+00
-8.54622424e-01 4.37601477e-01 -8.19779694e-01 3.63746613e-01
5.39954603e-01 2.41495296e-01 9.41868842e-01 -9.96184528e-01
6.48410022e-01 1.22226834e+00 4.74504411e-01 6.56874597e-01
-9.23699319e-01 -6.77111268e-01 2.57397115e-01 5.34684718e-01
-1.11247277e+00 -4.38615382e-01 1.17009532e+00 -5.63700823e-03
8.80014658e-01 8.48071501e-02 9.20868754e-01 9.20443892e-01
1.23832040e-01 1.40671933e+00 9.91305232e-01 -8.59574452e-02
1.67220592e-01 -1.57350630e-01 -3.26851815e-01 8.17842245e-01
1.79919437e-01 1.69388875e-01 -1.23514938e+00 4.90708560e-01
1.00998616e+00 -6.64366707e-02 -3.91687214e-01 -5.74626327e-01
-1.42557347e+00 5.23506880e-01 1.09774947e+00 8.96971449e-02
-4.48379755e-01 1.27609953e-01 2.75529146e-01 1.16307624e-02
4.13005024e-01 1.99244961e-01 -1.02100022e-01 -1.11697994e-01
-8.65911722e-01 6.34990856e-02 -5.48045672e-02 1.09506071e+00
8.30592930e-01 6.52134269e-02 -3.27898294e-01 6.51843905e-01
3.12185317e-01 2.65490174e-01 4.13679391e-01 -6.96818888e-01
7.58692205e-01 8.66966665e-01 1.54622942e-01 -1.08986235e+00
-4.98070002e-01 -5.04955947e-01 -8.35279882e-01 2.71985054e-01
-1.56916291e-01 2.49127492e-01 -9.78664160e-01 1.52875972e+00
4.11289543e-01 1.38993129e-01 -1.43846339e-02 1.43049097e+00
1.19275188e+00 5.40061831e-01 4.07293141e-02 4.53771539e-02
8.49126518e-01 -9.86582577e-01 -6.95785463e-01 -5.39474964e-01
3.46082628e-01 -4.35554475e-01 1.14899373e+00 2.01613173e-01
-1.24113572e+00 -6.43895864e-01 -1.41531026e+00 -6.99969888e-01
-6.74778074e-02 1.28997445e-01 8.91482234e-01 2.37989768e-01
-1.07051027e+00 5.02947748e-01 -8.21035683e-01 8.90349299e-02
7.89478958e-01 4.55043763e-01 -2.72332877e-01 -2.23317862e-01
-1.33231676e+00 9.20900524e-01 1.65832311e-01 3.62919539e-01
-9.59472835e-01 -6.39172971e-01 -8.95587265e-01 -2.27649868e-01
6.00385070e-02 -7.00442612e-01 1.15416944e+00 -7.63307750e-01
-1.42496037e+00 7.76931345e-01 -3.61452311e-01 -2.85357505e-01
4.31997597e-01 -2.82011777e-01 -2.01756164e-01 4.25350666e-01
1.66946739e-01 1.14014375e+00 8.74825656e-01 -1.28761280e+00
-1.00419009e+00 -3.88789147e-01 3.71935219e-01 5.89192390e-01
-6.14819705e-01 -2.44112730e-01 -4.00998265e-01 -6.01526439e-01
6.98713124e-01 -3.95561337e-01 5.82576692e-02 3.78419191e-01
-4.66936558e-01 8.44825730e-02 7.22244501e-01 -5.52331209e-01
1.30564415e+00 -2.13733172e+00 4.17582989e-01 -4.13257569e-01
5.47753096e-01 2.18189761e-01 9.02927667e-02 -7.19157308e-02
2.17367336e-01 -2.93770164e-01 -1.84786111e-01 -5.28367341e-01
-6.61578625e-02 1.50992647e-01 -2.35147238e-01 4.51695770e-01
4.83475685e-01 9.07019436e-01 -1.39117014e+00 -6.96375072e-01
5.77282071e-01 4.50553149e-01 -4.81110603e-01 2.08773941e-01
-3.36715467e-02 2.41949975e-01 -6.64863408e-01 1.01363337e+00
7.67348468e-01 -2.67422497e-01 -3.98536712e-01 -5.03930688e-01
-2.70756304e-01 7.32860088e-01 -7.71702707e-01 2.13186359e+00
-2.99510241e-01 6.84904337e-01 -2.61905134e-01 -7.43524075e-01
9.17562246e-01 -7.53521323e-02 4.56273168e-01 -1.16697550e+00
2.01933712e-01 4.37829852e-01 -1.42035410e-01 -2.77327091e-01
7.17218697e-01 8.08391646e-02 2.05904186e-01 7.56041333e-02
-1.07726902e-02 -1.20525740e-01 -1.32002041e-01 1.79488838e-01
1.01566136e+00 2.79271305e-01 7.11213574e-02 4.65229116e-02
4.11269873e-01 -1.90228552e-01 7.00599134e-01 3.30026507e-01
-6.17477655e-01 9.85962868e-01 4.68579561e-01 -2.57432371e-01
-8.84233356e-01 -1.12409413e+00 -5.22546358e-02 7.03143954e-01
9.25356328e-01 -1.93469122e-01 -4.27867174e-01 -6.15577102e-01
-1.73410028e-01 4.82565343e-01 -6.56022072e-01 -7.10375845e-01
-2.71940798e-01 -6.62312269e-01 2.48843759e-01 6.34981811e-01
1.08439255e+00 -9.78517413e-01 -1.07096124e+00 3.03709924e-01
-3.65202069e-01 -1.07951999e+00 -1.83059454e-01 3.18260461e-01
-1.06366253e+00 -6.94016337e-01 -8.48712802e-01 -8.19849432e-01
5.70643246e-01 7.53168046e-01 8.57034981e-01 -8.89817476e-02
1.52457803e-02 -2.27519631e-01 -5.06357491e-01 -3.91032845e-01
1.94307998e-01 2.99164593e-01 -2.10223198e-01 -6.56809509e-02
4.29602861e-01 -5.90926826e-01 -1.16466522e+00 2.76528627e-01
-8.53722036e-01 7.52111912e-01 8.02997530e-01 7.71024883e-01
5.35183311e-01 -2.40124747e-01 6.75599694e-01 -1.07974067e-01
3.82723399e-02 -3.01949531e-01 -3.92922103e-01 2.17635185e-02
-4.47606206e-01 -1.65061187e-02 2.53643364e-01 -9.91799012e-02
-9.63524222e-01 1.36237890e-01 -2.16292500e-01 -5.36292851e-01
2.38888919e-01 1.35454968e-01 -3.79850447e-01 -3.40878576e-01
4.09968048e-01 4.71804291e-01 -2.73331434e-01 -3.35395843e-01
8.31127986e-02 7.19660103e-01 5.12462139e-01 9.43425074e-02
7.47244120e-01 6.20495737e-01 -5.86046313e-04 -3.48398924e-01
-1.06027269e+00 -2.55634934e-01 -5.97535431e-01 -4.21468347e-01
6.26317084e-01 -1.33960772e+00 -4.97190714e-01 8.42737615e-01
-1.02865386e+00 -3.28308612e-01 -1.81704357e-01 3.76847267e-01
-4.69553351e-01 1.17658488e-01 -5.44068158e-01 -5.44638634e-01
-3.22102368e-01 -1.24792492e+00 1.33517468e+00 3.99318159e-01
1.65878862e-01 -4.28578228e-01 -5.35453618e-01 1.75986230e-01
3.69862646e-01 1.10487834e-01 6.59267783e-01 2.68279523e-01
-9.61009741e-01 1.27466679e-01 -7.64729321e-01 1.46062702e-01
3.20230275e-01 -2.99477696e-01 -1.04615366e+00 -1.98378298e-03
-1.31547138e-01 -3.46295863e-01 1.03478885e+00 3.51339042e-01
1.19111991e+00 7.55288750e-02 -2.60585010e-01 8.63821149e-01
1.44672322e+00 -3.85467857e-02 8.02471757e-01 7.57884204e-01
1.01315665e+00 3.51878852e-01 9.64822292e-01 5.97275972e-01
7.81427264e-01 5.95160186e-01 9.83440399e-01 -2.53654152e-01
-4.02610362e-01 -4.54499394e-01 3.99909437e-01 7.63531744e-01
-3.97089869e-02 -5.34765534e-02 -6.39235735e-01 7.61606514e-01
-1.71928990e+00 -8.07026386e-01 -5.77797443e-02 1.96696448e+00
1.17243695e+00 5.75515449e-01 1.14406236e-02 4.35267180e-01
5.41356325e-01 5.23885429e-01 -7.64289021e-01 -6.06690906e-02
-4.40866321e-01 -1.41816780e-01 6.27538979e-01 2.41796955e-01
-1.07773113e+00 9.50943947e-01 5.91490746e+00 7.31814802e-01
-1.32176721e+00 8.26200321e-02 5.19622147e-01 -4.92799759e-01
-6.81562781e-01 -9.73572582e-02 -8.66148829e-01 6.08143508e-01
1.83138728e-01 3.11189033e-02 1.51454747e-01 8.55489314e-01
1.25108227e-01 -5.09511828e-01 -9.22054410e-01 1.10523927e+00
4.98288684e-02 -1.33823800e+00 1.40958682e-01 -2.13821977e-01
8.78860235e-01 1.21293992e-01 2.32071817e-01 -3.56714800e-02
-7.31585026e-02 -5.83911121e-01 1.16996932e+00 3.17322999e-01
9.05906796e-01 -7.34691262e-01 6.25630915e-01 1.88860744e-02
-1.17631269e+00 -2.29909122e-01 -6.84655190e-01 -3.32207568e-02
3.47706139e-01 8.51093292e-01 -4.75916922e-01 5.63067138e-01
8.25253844e-01 1.39841640e+00 -6.17280006e-01 1.18973231e+00
-4.26045507e-01 1.05343014e-01 -3.17102015e-01 -7.02771395e-02
4.02607918e-01 2.71163672e-01 4.48408425e-01 7.27223575e-01
4.38912898e-01 1.07620671e-01 -3.01258773e-01 7.76851296e-01
7.77689368e-03 -3.42231542e-01 -3.26036841e-01 3.31087977e-01
5.11373460e-01 9.20674980e-01 -6.27794266e-01 -2.10711434e-01
-4.98983145e-01 1.19809341e+00 2.91395009e-01 1.11583419e-01
-1.02534842e+00 -4.61895823e-01 7.44483650e-01 2.72388637e-01
5.01860738e-01 -8.34383816e-02 -6.86688721e-01 -1.11019456e+00
1.63933858e-01 -4.54028398e-01 -7.19176158e-02 -1.27574492e+00
-7.18241513e-01 5.72744668e-01 -1.75905436e-01 -1.68496966e+00
2.59736747e-01 -4.05743569e-01 -1.47715271e-01 7.36451268e-01
-2.16032410e+00 -1.28702605e+00 -7.47660875e-01 5.94430447e-01
5.34634769e-01 1.87025219e-01 1.85894892e-01 5.57724535e-01
-4.76608068e-01 5.44899583e-01 -1.98205173e-01 -3.13079745e-01
5.55137753e-01 -1.04934657e+00 4.26661223e-01 1.03231680e+00
-3.15977216e-01 3.16425353e-01 5.59486330e-01 -6.29395962e-01
-1.52571034e+00 -1.12162650e+00 8.19226444e-01 3.52994422e-03
2.88196981e-01 -3.34907740e-01 -6.07909441e-01 4.15182650e-01
-1.10159673e-01 1.52990371e-01 -3.51360850e-02 -3.84240240e-01
-2.88655460e-02 -4.70190048e-01 -9.49968398e-01 7.40139127e-01
1.60416591e+00 -6.26664400e-01 -5.60026169e-01 -4.51551005e-02
9.23319042e-01 -7.06799150e-01 -4.57307696e-01 5.88755429e-01
6.51786864e-01 -1.50828445e+00 1.20020676e+00 1.65294856e-01
1.06959057e+00 -5.98013520e-01 -1.72232762e-01 -1.02209163e+00
-3.94341469e-01 -3.32210690e-01 -4.07363117e-01 1.12590587e+00
1.15317978e-01 -2.77980596e-01 8.19073081e-01 2.77113706e-01
-5.14183462e-01 -1.11499488e+00 -1.00959766e+00 -3.55472594e-01
-6.78841949e-01 -4.25106078e-01 7.71011770e-01 5.71135104e-01
3.39672118e-02 1.71958491e-01 -4.73452449e-01 1.16298281e-01
6.56106412e-01 2.10882470e-01 4.53058124e-01 -9.15335774e-01
9.52763185e-02 -4.53684986e-01 -7.01506257e-01 -1.64774668e+00
-2.29858637e-01 -5.36946774e-01 3.78245533e-01 -1.87417448e+00
7.77699649e-02 -7.99577594e-01 -5.15313327e-01 5.30701816e-01
-3.49687546e-01 5.42385519e-01 -3.68402377e-02 1.56753987e-01
-7.40189612e-01 1.09712553e+00 1.77258456e+00 -6.14289641e-02
-1.46680564e-01 -3.13577145e-01 -7.53826201e-01 6.01200342e-01
5.57526708e-01 -1.84952527e-01 -6.35058999e-01 -7.37474322e-01
2.82742798e-01 -1.98560268e-01 4.77815449e-01 -1.28368175e+00
2.45139599e-01 -2.36202806e-01 6.67419374e-01 -9.93535936e-01
5.73070109e-01 -5.90217471e-01 -2.88683891e-01 3.87755692e-01
-1.54382929e-01 -1.49914086e-01 2.12038048e-02 3.71020079e-01
-4.78643715e-01 2.40209457e-02 6.93713605e-01 9.11375582e-02
-1.32165122e+00 4.42400962e-01 9.27508026e-02 -1.29882008e-01
9.66973245e-01 -7.66753197e-01 -4.68879163e-01 -2.46215940e-01
-1.38562918e-01 1.55191198e-01 7.03250647e-01 4.97806877e-01
1.00349343e+00 -1.55127048e+00 -3.42198759e-01 4.22912270e-01
2.09099054e-01 5.13185620e-01 3.91716033e-01 9.45600629e-01
-5.45002460e-01 4.39538211e-01 -7.11207926e-01 -7.33949065e-01
-8.20193708e-01 4.52424437e-01 1.29718199e-01 2.12330729e-01
-7.11830974e-01 1.26169670e+00 3.58296245e-01 2.52123117e-01
2.77331322e-01 -4.39234734e-01 -1.63668886e-01 1.71945114e-02
5.24285853e-01 1.58548713e-01 5.07784523e-02 -7.36205399e-01
-5.45226216e-01 5.46901286e-01 9.41547751e-03 6.11117668e-02
1.38288915e+00 -5.57174027e-01 1.57334223e-01 4.53224391e-01
1.11006582e+00 -5.06804347e-01 -1.92986166e+00 -2.65897900e-01
-3.43562484e-01 -8.95693719e-01 5.50520658e-01 -3.88396263e-01
-1.29043686e+00 9.80544746e-01 6.19690537e-01 -1.59141585e-01
1.55958664e+00 -2.19258234e-01 1.13649154e+00 -7.50827268e-02
6.23093843e-01 -1.10806608e+00 2.38520592e-01 3.34202647e-01
8.79299819e-01 -1.44304359e+00 1.44654065e-01 -6.56506836e-01
-7.30616033e-01 7.16068387e-01 9.13554966e-01 -9.84555557e-02
4.06566083e-01 -1.35042248e-02 -2.50993162e-01 -1.19283341e-01
-7.15648293e-01 -4.12141144e-01 2.68797934e-01 6.69896066e-01
2.48765856e-01 -2.64438391e-01 -1.89722344e-01 3.70917499e-01
-1.23908289e-01 2.17172340e-01 3.64497721e-01 1.11918581e+00
-5.11544228e-01 -6.87343478e-01 4.32584845e-02 4.36044693e-01
-1.54039443e-01 -2.18375057e-01 -1.29419744e-01 4.99957502e-01
2.30688870e-01 7.30795681e-01 -8.76457542e-02 -7.55679727e-01
3.01678240e-01 -4.67944175e-01 5.96453190e-01 -3.94142270e-01
-3.29608560e-01 -1.65571734e-01 -4.65145297e-02 -8.05046022e-01
-7.68322885e-01 -6.31182134e-01 -1.24980950e+00 -2.38884985e-01
-4.91286367e-01 -4.33288306e-01 4.67419237e-01 8.67512226e-01
4.47018147e-01 7.89720833e-01 7.97107399e-01 -1.27703893e+00
-1.76717743e-01 -8.29978108e-01 -5.02332687e-01 2.37482086e-01
6.87215507e-01 -1.14819932e+00 -1.78769112e-01 -7.22007155e-02] | [9.674763679504395, -0.8040622472763062] |
2506d9b8-eae5-4733-92a3-f32d1a4e542f | s3t-self-supervised-pre-training-with-swin | 2202.10139 | null | https://arxiv.org/abs/2202.10139v1 | https://arxiv.org/pdf/2202.10139v1.pdf | S3T: Self-Supervised Pre-training with Swin Transformer for Music Classification | In this paper, we propose S3T, a self-supervised pre-training method with Swin Transformer for music classification, aiming to learn meaningful music representations from massive easily accessible unlabeled music data. S3T introduces a momentum-based paradigm, MoCo, with Swin Transformer as its feature extractor to music time-frequency domain. For better music representations learning, S3T contributes a music data augmentation pipeline and two specially designed pre-processors. To our knowledge, S3T is the first method combining the Swin Transformer with a self-supervised learning method for music classification. We evaluate S3T on music genre classification and music tagging tasks with linear classifiers trained on learned representations. Experimental results show that S3T outperforms the previous self-supervised method (CLMR) by 12.5 percents top-1 accuracy and 4.8 percents PR-AUC on two tasks respectively, and also surpasses the task-specific state-of-the-art supervised methods. Besides, S3T shows advances in label efficiency using only 10% labeled data exceeding CLMR on both tasks with 100% labeled data. | ['Kejun Zhang', 'Zejun Ma', 'Belei Zhu', 'Chen Zhang', 'Hang Zhao'] | 2022-02-21 | null | null | null | null | ['genre-classification', 'music-classification'] | ['computer-vision', 'music'] | [ 3.63842815e-01 -5.70189990e-02 -7.31279969e-01 -4.87260297e-02
-1.26963532e+00 -7.46746361e-01 2.44724154e-01 -2.10642305e-04
-2.84376472e-01 4.44970161e-01 1.24810547e-01 1.52293190e-01
-3.63891721e-01 -3.35835904e-01 -4.70521480e-01 -4.30875778e-01
-2.92562842e-01 7.89409816e-01 -4.14201021e-02 3.91811617e-02
2.70237476e-01 -3.37605059e-01 -1.47355199e+00 7.76457489e-01
6.83825493e-01 1.34438515e+00 2.13862695e-02 4.90170658e-01
-3.91796231e-02 9.30997133e-01 -2.74195641e-01 -1.51402682e-01
3.74986678e-01 -5.13418972e-01 -8.47192109e-01 -2.54907757e-01
5.71318507e-01 5.38750529e-01 1.05757393e-01 7.54009962e-01
6.26505733e-01 1.86937883e-01 5.79457462e-01 -1.09898746e+00
-5.56184888e-01 1.55162144e+00 -6.12617433e-01 1.14882976e-01
1.31481946e-01 -1.61662847e-01 1.52445424e+00 -1.01348650e+00
3.26076925e-01 8.45149875e-01 1.06682420e+00 5.37606895e-01
-1.27341056e+00 -1.24760962e+00 -5.93376011e-02 2.85068303e-01
-1.44564533e+00 -5.29499233e-01 8.40588927e-01 -4.15816724e-01
9.41150606e-01 4.08821076e-01 6.48555756e-01 1.09229398e+00
-4.83280301e-01 1.06995571e+00 1.18707204e+00 -5.48901498e-01
3.43372762e-01 -6.04750179e-02 1.36632144e-01 4.11827713e-01
-2.78798521e-01 -1.22888982e-02 -1.40434897e+00 -2.25802109e-01
5.52648187e-01 -3.75229806e-01 1.53420776e-01 -5.19914143e-02
-1.70463157e+00 4.10497516e-01 3.07667971e-01 5.83888352e-01
-1.62481576e-01 3.40600640e-01 7.00399935e-01 4.94608164e-01
7.53959715e-01 8.88432145e-01 -8.32445145e-01 -5.09351671e-01
-1.38582230e+00 4.46934141e-02 5.39704025e-01 8.58006537e-01
1.87805936e-01 4.89168376e-01 -4.48589981e-01 1.08471203e+00
-3.25298458e-02 4.34895933e-01 9.71069276e-01 -7.59362400e-01
4.73903924e-01 6.72200739e-01 -3.55652392e-01 -3.44902039e-01
-4.78418797e-01 -1.45863807e+00 -8.25603485e-01 -1.29691124e-01
3.11974734e-01 1.91501662e-01 -4.90903705e-01 1.76001453e+00
-6.01909012e-02 8.20106328e-01 6.28385395e-02 8.30520153e-01
8.95860076e-01 3.43291849e-01 -1.47158876e-02 -4.40327168e-01
1.16464567e+00 -1.49494493e+00 -5.08978844e-01 -8.77556354e-02
4.64067906e-01 -1.03727508e+00 1.39785159e+00 1.09633970e+00
-8.81790638e-01 -8.72695744e-01 -9.94620025e-01 1.81432664e-01
8.17098934e-03 8.54135513e-01 1.00963581e+00 5.51914155e-01
-5.36850095e-01 1.13183725e+00 -6.16851151e-01 -1.77138939e-01
6.62583888e-01 5.21717727e-01 -1.78846180e-01 4.10290271e-01
-6.94565654e-01 2.12239206e-01 6.97163105e-01 -3.90048712e-01
-9.04839933e-01 -1.07733226e+00 -2.23603517e-01 -3.45084667e-02
3.80498022e-01 -4.62902129e-01 1.47106779e+00 -1.11290550e+00
-1.76026380e+00 9.30267811e-01 5.33451065e-02 -6.79167211e-01
1.34418115e-01 -5.97759008e-01 -5.71674109e-01 -1.56886742e-01
2.58844405e-01 6.71370983e-01 9.24167335e-01 -7.36123681e-01
-5.55006564e-01 -2.06641495e-01 -5.05768239e-01 2.58886129e-01
-7.20173120e-01 -8.55208263e-02 -3.22086126e-01 -1.22055459e+00
4.54757810e-01 -1.17723298e+00 -5.22082038e-02 -5.68778157e-01
-4.91759062e-01 -4.82988685e-01 3.98214698e-01 -2.78215647e-01
1.35907769e+00 -2.34315157e+00 4.88638841e-02 -3.00697926e-02
-1.68369301e-02 1.74349815e-01 -4.45509553e-01 2.41682112e-01
-3.08171898e-01 -1.82915330e-01 -6.18173555e-02 -5.43894827e-01
2.30389208e-01 -6.27668425e-02 -5.56971252e-01 1.52731121e-01
-4.78536403e-03 1.04789340e+00 -1.03478932e+00 -2.96043903e-01
-6.91061988e-02 2.02332973e-01 -8.05029809e-01 -9.83889997e-02
-4.54112440e-01 8.34790587e-01 -1.45956770e-01 8.27990055e-01
2.65260190e-01 -2.97227651e-01 1.49414435e-01 -1.02046177e-01
-1.03831403e-01 9.24126565e-01 -1.12684214e+00 2.55585384e+00
-2.81661779e-01 3.15831214e-01 -6.06793046e-01 -9.36608911e-01
1.07467949e+00 5.12912333e-01 8.82976592e-01 -5.19690275e-01
7.83944875e-02 5.50505996e-01 -6.18886836e-02 -2.89023221e-02
3.48330766e-01 -2.98306018e-01 -3.33256960e-01 6.45109713e-01
5.76128304e-01 2.09185332e-01 1.63245991e-01 2.37282515e-02
1.04532623e+00 4.42203373e-01 2.33961895e-01 -3.15873444e-01
2.32340395e-01 6.54361844e-02 8.68233979e-01 7.85105586e-01
1.96908072e-01 6.74355924e-01 8.91339555e-02 -3.11537087e-01
-6.53066397e-01 -8.57565224e-01 -2.55075306e-01 1.75390780e+00
-2.95975924e-01 -1.02971601e+00 -5.11124492e-01 -7.36931086e-01
6.47299960e-02 4.02734816e-01 -4.93134350e-01 -8.66302326e-02
-2.40343168e-01 -7.41501868e-01 8.87861133e-01 5.56628287e-01
3.96745384e-01 -1.38194442e+00 -5.24180681e-02 3.55793625e-01
-1.44815236e-01 -8.53001595e-01 -5.29259801e-01 6.58274710e-01
-1.29697883e+00 -9.73770559e-01 -4.84685928e-01 -9.26551104e-01
8.10057223e-02 1.27372101e-01 1.21041644e+00 -5.39620757e-01
-1.62891150e-01 9.70151573e-02 -7.39749253e-01 -5.84597945e-01
-2.47952305e-02 7.54871368e-01 4.66094255e-01 1.71505705e-01
3.72430116e-01 -1.28822947e+00 -3.41434032e-01 2.60948151e-01
-3.49911839e-01 2.71940738e-01 5.90144575e-01 7.91169286e-01
9.60060656e-01 -3.24803367e-02 1.06489372e+00 -8.87446165e-01
1.85813844e-01 -3.71748120e-01 -1.98457792e-01 -1.74656630e-01
-9.23486888e-01 1.69417933e-01 5.03413856e-01 -8.90249252e-01
-5.54180920e-01 3.63136113e-01 3.84485978e-03 -6.74796104e-01
6.20342698e-03 4.95324880e-01 2.12262928e-01 2.97578424e-01
1.17205393e+00 1.26176268e-01 -5.41330040e-01 -1.20872748e+00
3.30208570e-01 7.79181898e-01 9.55135882e-01 -6.79688811e-01
9.60160017e-01 2.39185929e-01 -1.16732255e-01 -2.59543091e-01
-1.69588029e+00 -8.43936741e-01 -6.84078455e-01 -6.91531897e-02
4.58483845e-01 -1.22062230e+00 -7.14790404e-01 1.90202549e-01
-4.85705942e-01 -2.57326603e-01 -9.99860823e-01 8.99612486e-01
-8.58614683e-01 -3.52786854e-02 -6.38921559e-01 -9.89979208e-01
-8.46639633e-01 -3.63377720e-01 1.20519412e+00 -1.70929227e-02
-5.15416622e-01 -4.54243571e-01 4.96445179e-01 7.71105528e-01
3.44631106e-01 -1.54222175e-01 6.09078825e-01 -9.68059421e-01
-2.48656869e-01 -3.93004157e-02 1.21228032e-01 4.84121382e-01
2.41152495e-02 -8.60569537e-01 -1.62555563e+00 -2.79997498e-01
-4.45352226e-01 -8.15475106e-01 1.18871903e+00 2.15326205e-01
1.28329849e+00 -1.29748657e-01 9.25954729e-02 9.32316303e-01
1.00470781e+00 -1.07971050e-01 1.65656164e-01 5.53058565e-01
7.23693967e-01 -1.19793199e-01 7.34100282e-01 5.03224671e-01
9.29127634e-02 9.40669358e-01 2.15151399e-01 5.98009787e-02
-3.89412284e-01 -7.50313222e-01 7.20240355e-01 1.59039772e+00
-4.21994537e-01 3.58215660e-01 -5.85265338e-01 2.82115787e-01
-2.13368034e+00 -8.14760804e-01 -1.67866394e-01 2.26640821e+00
1.11220908e+00 1.12800062e-01 4.35340554e-01 8.60383689e-01
2.63253629e-01 -6.78132847e-03 -7.75876939e-01 2.42967099e-01
-2.62732416e-01 7.03750849e-01 2.17797205e-01 -1.50574937e-01
-1.55499458e+00 1.29621947e+00 5.98464251e+00 1.38117576e+00
-9.90295768e-01 4.47866440e-01 5.91665581e-02 -5.48667550e-01
1.88672706e-01 1.83791101e-01 -5.67307651e-01 1.91646412e-01
8.87230933e-01 6.25858009e-02 7.28933454e-01 1.05360878e+00
1.05066216e-02 5.28738439e-01 -1.03295767e+00 1.49850595e+00
7.06102625e-02 -1.17125607e+00 -3.13578323e-02 -6.07734025e-02
9.06634748e-01 3.45706075e-01 2.02102959e-01 8.62832487e-01
2.00012350e-03 -9.84452546e-01 9.57241058e-01 4.04958367e-01
1.04827607e+00 -8.17700922e-01 5.69818258e-01 3.47291172e-01
-1.50867438e+00 -1.27600268e-01 -2.46469751e-01 -3.63052636e-01
-1.32274762e-01 6.56452239e-01 -7.88553476e-01 6.55499816e-01
6.80202842e-01 1.55488253e+00 -7.28071868e-01 1.07727182e+00
-1.01257883e-01 1.37550414e+00 -1.67578682e-01 3.60095322e-01
-9.26339105e-02 -1.56549495e-02 6.50127470e-01 1.26730621e+00
4.09572214e-01 -2.62264729e-01 4.37809139e-01 6.02311015e-01
-2.27108523e-01 6.34961963e-01 -3.62334289e-02 -3.63559604e-01
2.76685655e-01 1.35398090e+00 -6.11371398e-01 -4.27618653e-01
2.54832834e-01 1.03442526e+00 1.45762518e-01 -1.12302557e-01
-5.86810648e-01 -1.03315018e-01 3.24834883e-01 8.73973370e-02
2.33553365e-01 8.18713978e-02 -4.85293776e-01 -1.48403585e+00
-1.11269251e-01 -9.93132949e-01 4.73456711e-01 -7.42789507e-01
-1.47869861e+00 6.14828467e-01 -5.44202685e-01 -1.91295648e+00
-6.41741678e-02 -4.62631911e-01 -3.98284316e-01 2.97144890e-01
-1.39398026e+00 -1.33664548e+00 9.77130681e-02 6.02722585e-01
6.98210239e-01 -1.03895736e+00 1.29835165e+00 5.44794202e-01
-4.03470814e-01 8.28486860e-01 6.99482709e-02 1.14968322e-01
9.79345918e-01 -1.37892175e+00 2.82672375e-01 4.68754284e-02
1.28975344e+00 2.93425858e-01 1.29878417e-01 -3.62077117e-01
-1.28322768e+00 -1.14793146e+00 8.69494736e-01 -5.54922283e-01
7.14960158e-01 -3.13248903e-01 -5.65897226e-01 5.22675037e-01
-1.01866543e-01 3.02216075e-02 1.27579331e+00 8.18620443e-01
-8.11347425e-01 -3.28624576e-01 -4.53251421e-01 7.97345638e-02
1.53666615e+00 -6.69700742e-01 -7.23025501e-01 4.34145182e-01
4.96066004e-01 -2.28395700e-01 -8.83072734e-01 7.24089503e-01
8.27890754e-01 -6.01006448e-01 9.54018593e-01 -7.47371912e-01
1.39750466e-01 -5.71365178e-01 -3.58906209e-01 -1.07008445e+00
-4.21663195e-01 -8.99210751e-01 -3.33099991e-01 1.27596915e+00
6.13432646e-01 1.02454573e-01 1.06076956e+00 -6.45473063e-01
-3.53219509e-01 -4.31694120e-01 -1.09137797e+00 -1.20924711e+00
-1.80140272e-01 -1.13921177e+00 2.89933115e-01 1.41192663e+00
4.92387116e-01 9.06610012e-01 -7.53551126e-01 -2.74317890e-01
6.70035481e-01 5.01080990e-01 8.25122297e-01 -1.71736443e+00
-8.45037043e-01 -4.49990481e-01 -4.79623973e-01 -8.72104645e-01
8.72425288e-02 -1.75590348e+00 -1.69960886e-01 -9.70879078e-01
5.78920305e-01 -6.89043403e-01 -1.13102555e+00 1.02072847e+00
7.86223710e-02 9.48812366e-01 3.84984314e-01 5.25922179e-01
-1.22672284e+00 5.45455933e-01 9.22758579e-01 -3.33753049e-01
-6.04504645e-01 2.15845019e-01 -8.15511644e-01 8.31918836e-01
1.01455760e+00 -8.53945792e-01 -5.48807323e-01 -1.73150990e-02
4.20317888e-01 -3.44513267e-01 2.96804421e-02 -1.56604135e+00
2.57640313e-02 1.12981506e-01 2.39842236e-01 -6.81063056e-01
4.42957968e-01 -3.30186546e-01 1.29015744e-01 1.46076575e-01
-6.73433602e-01 -4.10252571e-01 2.73003191e-01 4.56407815e-01
-2.78826773e-01 -1.20707043e-02 5.03571630e-01 7.95029476e-02
-3.53192925e-01 1.08512484e-01 1.75273623e-02 2.17590675e-01
3.70267212e-01 2.06812993e-01 5.43612614e-02 -1.62601098e-01
-1.11914599e+00 -1.57952741e-01 -1.52299598e-01 6.84024870e-01
2.58284956e-01 -1.71377361e+00 -8.34376454e-01 1.66293845e-01
4.24224734e-01 -3.66218209e-01 1.12529188e-01 9.16915894e-01
1.78227365e-01 4.60468680e-01 -4.43557724e-02 -7.37281561e-01
-1.22622824e+00 3.67601395e-01 -8.54239911e-02 -5.82983613e-01
-7.72444308e-01 9.25625384e-01 -2.11336046e-01 -8.09046626e-01
5.61735213e-01 -3.60908329e-01 -2.66605943e-01 1.60619095e-01
5.09403288e-01 1.83325633e-01 9.37409177e-02 -4.34889734e-01
-2.26258740e-01 7.62580335e-01 1.94104671e-01 -2.14824393e-01
1.56271803e+00 5.42988956e-01 5.64342253e-02 1.08979917e+00
6.13601744e-01 3.23678672e-01 -9.81732428e-01 -4.69389737e-01
3.98947150e-01 -1.73977315e-01 1.90143377e-01 -1.22136402e+00
-1.02166855e+00 6.08884752e-01 7.50321031e-01 -1.73867092e-01
1.19605243e+00 1.35371014e-01 7.97912776e-01 5.21661162e-01
5.50374269e-01 -1.07216334e+00 3.12652320e-01 5.53470671e-01
8.21894288e-01 -9.28197026e-01 -3.74940149e-02 -1.70465991e-01
-7.08510041e-01 8.38446259e-01 4.09846365e-01 -2.24772468e-01
5.57052493e-01 7.32910037e-02 1.27127707e-01 -2.70279571e-02
-9.69720483e-01 -4.54385668e-01 1.02068806e+00 2.62512237e-01
7.22264290e-01 2.53867209e-01 -1.34162009e-01 1.50048351e+00
-7.97944844e-01 3.23067963e-01 -2.14771420e-01 5.08415639e-01
-3.62334698e-01 -1.53214312e+00 -2.45094061e-01 4.80969578e-01
-3.68625671e-01 -2.95419753e-01 -5.29116631e-01 2.45177820e-01
6.21145785e-01 8.59750152e-01 -2.36138731e-01 -9.28727865e-01
2.76592106e-01 4.82566416e-01 4.59279060e-01 -9.33348656e-01
-1.07131135e+00 6.37986720e-01 1.66393705e-02 -4.34202611e-01
-6.11986101e-01 -5.46009421e-01 -1.19406486e+00 3.54997009e-01
-4.23774034e-01 4.63341683e-01 5.62824249e-01 7.50826061e-01
4.28349853e-01 7.37188816e-01 6.60197079e-01 -9.61801291e-01
-3.49369168e-01 -1.43531811e+00 -9.76058841e-01 2.98246711e-01
5.41530550e-03 -6.69936299e-01 -2.33922884e-01 4.74731550e-02] | [15.75326919555664, 5.237300872802734] |
f030836c-23fc-476b-ad52-e84e91d6e4ef | the-undesirable-dependence-on-frequency-of | 2301.00792 | null | https://arxiv.org/abs/2301.00792v1 | https://arxiv.org/pdf/2301.00792v1.pdf | The Undesirable Dependence on Frequency of Gender Bias Metrics Based on Word Embeddings | Numerous works use word embedding-based metrics to quantify societal biases and stereotypes in texts. Recent studies have found that word embeddings can capture semantic similarity but may be affected by word frequency. In this work we study the effect of frequency when measuring female vs. male gender bias with word embedding-based bias quantification methods. We find that Skip-gram with negative sampling and GloVe tend to detect male bias in high frequency words, while GloVe tends to return female bias in low frequency words. We show these behaviors still exist when words are randomly shuffled. This proves that the frequency-based effect observed in unshuffled corpora stems from properties of the metric rather than from word associations. The effect is spurious and problematic since bias metrics should depend exclusively on word co-occurrences and not individual word frequencies. Finally, we compare these results with the ones obtained with an alternative metric based on Pointwise Mutual Information. We find that this metric does not show a clear dependence on frequency, even though it is slightly skewed towards male bias across all frequencies. | ['Edgar Altszyler', 'Diego Fernandez Slezak', 'Germán Rosati', 'Francisco Valentini'] | 2023-01-02 | null | null | null | null | ['semantic-textual-similarity'] | ['natural-language-processing'] | [-1.56491458e-01 -2.08039925e-01 -5.22378504e-01 -4.60663050e-01
1.05728686e-01 -6.91788793e-01 1.07369208e+00 8.93074274e-01
-1.17319894e+00 6.52729511e-01 7.21765697e-01 -2.88446337e-01
-8.52192938e-02 -1.17676044e+00 -1.58084124e-01 -6.24628007e-01
-1.24332316e-01 1.75507545e-01 -8.85889307e-03 -4.55349535e-01
5.94440818e-01 1.69128925e-01 -1.71276391e+00 -1.45882785e-01
7.28648543e-01 2.20976919e-01 -2.95416564e-01 4.97065157e-01
-3.19678605e-01 2.18401089e-01 -6.58855975e-01 -6.49613857e-01
4.94832685e-03 -4.05694932e-01 -6.43813729e-01 -6.36431634e-01
5.84589958e-01 1.34365037e-01 4.69936877e-02 1.40387559e+00
6.40946865e-01 -1.48327529e-01 1.02196813e+00 -1.07313776e+00
-7.12557316e-01 8.47006440e-01 -5.58953762e-01 4.95745122e-01
4.32806224e-01 -7.83116519e-02 1.43864083e+00 -7.83199191e-01
6.82854176e-01 1.59735942e+00 8.61671627e-01 3.09480458e-01
-1.53880477e+00 -7.99984694e-01 5.89501038e-02 7.51725584e-02
-1.10664070e+00 -4.15256387e-03 6.74984872e-01 -6.76441073e-01
5.57794392e-01 5.42900324e-01 6.82465017e-01 1.34016740e+00
3.50653410e-01 5.46469651e-02 1.65908110e+00 -4.42133695e-01
5.75885139e-02 4.61638898e-01 6.25245869e-01 3.22839290e-01
1.23873138e+00 2.30841577e-01 -3.98210585e-01 -4.86096591e-01
2.39874110e-01 2.76747290e-02 -1.41157091e-01 -3.24615985e-01
-1.46447980e+00 1.42746568e+00 3.02323073e-01 9.79313374e-01
-1.68226302e-01 3.11622649e-01 6.89019680e-01 4.21043783e-01
6.83662534e-01 7.00977683e-01 -4.33787227e-01 -2.89932430e-01
-7.16652751e-01 7.99373627e-01 6.34029746e-01 1.01621568e-01
8.27307820e-01 -1.94287807e-01 -3.29428494e-01 1.00295031e+00
2.05402225e-01 7.51964748e-01 1.01686800e+00 -2.91301459e-01
2.63635311e-02 5.21867633e-01 6.25094995e-02 -1.72987258e+00
-6.41108572e-01 -1.98466167e-01 -4.83443767e-01 -4.53471877e-02
6.81944668e-01 -5.10994606e-02 -4.55252409e-01 2.11825871e+00
2.19006583e-01 -6.37885749e-01 -2.86442667e-01 8.22721779e-01
7.50033140e-01 2.31454626e-01 4.10861105e-01 -7.78739080e-02
1.73377395e+00 -1.15816705e-02 -9.90208805e-01 -2.11152285e-01
8.48259509e-01 -7.81265318e-01 1.42871320e+00 -7.58022740e-02
-6.04683816e-01 -3.41827393e-01 -1.03632331e+00 3.80353630e-03
-1.01139450e+00 -4.80547071e-01 6.11581087e-01 1.24948192e+00
-6.53162181e-01 8.30461740e-01 -1.36167765e-01 -7.13482320e-01
5.04493639e-02 6.89508095e-02 -2.68500328e-01 2.74292856e-01
-1.41705787e+00 1.17637634e+00 2.53199041e-01 -6.70991302e-01
1.00357868e-01 -8.33750486e-01 -7.95167804e-01 -1.32976621e-01
-1.40827715e-01 -3.56568366e-01 8.19541454e-01 -1.07715869e+00
-7.91713059e-01 1.22421622e+00 -1.44887552e-01 -3.60728830e-01
4.11942273e-01 -2.28698947e-03 -6.90980494e-01 -3.75510037e-01
3.30118775e-01 4.32418764e-01 5.67129314e-01 -1.17079341e+00
-3.39305162e-01 -5.79897761e-01 -4.62843366e-02 -2.28043094e-01
-8.14072907e-01 4.29397961e-03 7.68261850e-01 -9.17258739e-01
-1.63795322e-01 -7.92023122e-01 9.09399092e-02 -2.67369866e-01
-6.36396483e-02 -4.76373166e-01 3.04627955e-01 -3.03114474e-01
1.63734245e+00 -2.01158905e+00 -1.30805388e-01 3.27033162e-01
1.83581725e-01 1.97195448e-02 -3.74478921e-02 7.13787913e-01
-3.98953080e-01 4.99088526e-01 -2.73667783e-01 2.01952249e-01
3.84209067e-01 3.21262121e-01 -2.54594266e-01 8.53551209e-01
1.37760639e-01 6.29034400e-01 -1.13228941e+00 -5.24379134e-01
2.38643009e-02 2.84090102e-01 -8.31352115e-01 -3.39610457e-01
2.80198604e-01 -3.75143409e-01 7.39249885e-02 2.35116675e-01
7.84759402e-01 2.46628374e-01 3.90809238e-01 -2.95658916e-01
-4.26374376e-01 5.88912010e-01 -8.87461007e-01 1.04397571e+00
-6.12891197e-01 8.64478648e-01 -3.74257594e-01 -9.28104639e-01
1.02774870e+00 -7.77935833e-02 -9.34521854e-02 -7.80006588e-01
3.24167252e-01 4.83280331e-01 6.34131789e-01 -3.01971763e-01
1.00717854e+00 -7.65436113e-01 -3.20832789e-01 5.15817463e-01
8.22340790e-03 4.23264690e-02 3.00232738e-01 -4.73534986e-02
8.18240225e-01 -5.53962588e-01 5.33939004e-01 -1.13137984e+00
4.57560450e-01 -3.00460190e-01 2.78164476e-01 5.72837651e-01
-1.17964216e-01 3.90438855e-01 9.71563995e-01 -2.02136084e-01
-9.74560797e-01 -1.05660021e+00 -8.11500609e-01 1.21733665e+00
1.19309120e-01 -6.47969306e-01 -4.91816223e-01 -7.43449509e-01
5.14146686e-01 1.07264817e+00 -1.21652806e+00 -3.67660522e-01
-2.72136152e-01 -1.29722357e+00 4.57796723e-01 2.28779256e-01
-2.37589031e-01 -6.21859848e-01 -6.37888253e-01 -1.99424326e-01
1.11227326e-01 -6.03898942e-01 -2.86939323e-01 8.08050781e-02
-6.82240725e-01 -9.80115533e-01 -7.12679923e-01 -2.61999041e-01
4.99224126e-01 -7.26232380e-02 1.41190803e+00 1.47592843e-01
-7.14332610e-02 2.17353627e-01 -5.46775579e-01 -7.24799693e-01
-4.23921734e-01 1.03173606e-01 3.49345744e-01 -1.55466348e-01
1.01133573e+00 -6.04559958e-01 -6.83164179e-01 1.69460997e-01
-1.03110671e+00 -7.91446209e-01 3.43532681e-01 8.10817063e-01
-2.63528407e-01 -5.56169689e-01 7.46686399e-01 -1.01577854e+00
1.05847931e+00 -6.30064011e-01 -5.74288219e-02 -4.13690001e-01
-1.01744807e+00 2.65558183e-01 2.78315485e-01 -5.79938471e-01
-3.19077611e-01 -8.64055037e-01 1.44264065e-02 3.12983662e-01
2.47568879e-02 1.81538150e-01 2.81170189e-01 3.16053718e-01
8.99170280e-01 -2.14421496e-01 1.09727293e-01 -3.56148392e-01
2.87308872e-01 8.38283420e-01 -1.43714488e-01 -3.78276169e-01
6.29047930e-01 5.59634268e-01 -2.21404910e-01 -9.42971468e-01
-6.40174150e-01 -3.17311466e-01 -1.80966496e-01 2.93564089e-02
1.06539762e+00 -5.90421379e-01 -7.19977558e-01 -1.53185040e-01
-1.07624280e+00 1.43224433e-01 -4.35860842e-01 5.76429069e-01
-8.70800093e-02 3.42285305e-01 -3.01518679e-01 -8.45315158e-01
-6.69427663e-02 -7.96688557e-01 8.19546521e-01 -1.20914206e-01
-1.05210066e+00 -1.36264801e+00 5.32507598e-01 -1.45261645e-01
4.87065524e-01 3.26886296e-01 1.08406246e+00 -9.30573225e-01
6.59289002e-01 -3.45193356e-01 -3.34039092e-01 2.32318655e-01
4.35785443e-01 -6.83823600e-02 -9.29747880e-01 -2.64051825e-01
-3.19963217e-01 1.22990340e-01 1.00513613e+00 5.18938750e-02
8.91639590e-01 -4.13031131e-01 -1.86375737e-01 -8.06238279e-02
1.64358842e+00 -3.27251345e-01 5.10209441e-01 4.06958401e-01
6.95127726e-01 9.29384053e-01 2.96467155e-01 2.97968835e-01
1.67211741e-01 5.90686738e-01 1.29988804e-01 1.20427608e-01
1.13811836e-01 -3.11120450e-01 5.42895257e-01 6.74113512e-01
1.06748588e-01 6.60001388e-06 -9.19105589e-01 9.02257383e-01
-1.27923131e+00 -1.09207976e+00 -5.44639945e-01 2.51566267e+00
9.06362951e-01 1.98342726e-01 6.36333227e-01 5.60228229e-01
6.85526907e-01 5.91366172e-01 2.40376621e-01 -1.19328296e+00
-1.18699297e-01 3.27205271e-01 8.89722705e-01 6.93877041e-01
-6.28845096e-01 5.14779866e-01 6.72755861e+00 7.03781486e-01
-1.20683086e+00 1.73893496e-01 3.37655425e-01 -1.40104771e-01
-9.54174876e-01 -1.48016632e-01 -4.57064241e-01 7.70758867e-01
9.68202770e-01 -4.35316116e-01 -3.55330884e-01 4.47397470e-01
-2.62098815e-02 -2.19167471e-01 -1.02385330e+00 9.61473346e-01
1.25971869e-01 -6.17357492e-01 1.36800885e-01 4.19934183e-01
5.46636343e-01 -1.85290188e-01 1.43245459e-01 5.86764067e-02
7.65221938e-03 -1.26634431e+00 7.59073496e-01 2.26928025e-01
5.33331931e-01 -1.00561261e+00 1.02219296e+00 -8.57539549e-02
-4.33060646e-01 -7.46402740e-02 -5.25277734e-01 -5.78955710e-01
-1.23325475e-01 1.28879738e+00 -6.29348338e-01 1.40156418e-01
5.03335774e-01 5.58951855e-01 -6.10031605e-01 3.69676352e-01
-1.77731365e-02 6.27580404e-01 -1.37404278e-01 -6.94599390e-01
1.06717817e-01 -3.30518246e-01 6.32950246e-01 1.72166979e+00
3.94021332e-01 -5.33700049e-01 -5.65122545e-01 9.10253584e-01
2.01431751e-01 5.47271132e-01 -1.00012314e+00 -3.37943912e-01
3.42949301e-01 1.31985021e+00 -7.82545567e-01 -2.78825492e-01
-6.36142373e-01 6.64975584e-01 5.30584482e-03 -1.04692876e-01
-7.36012101e-01 -5.66053867e-01 1.24292219e+00 5.56527555e-01
2.08696216e-01 -1.33218959e-01 -5.09101450e-01 -9.10461426e-01
-1.78406239e-01 -7.40670323e-01 2.25267932e-01 -6.37575611e-02
-1.60388243e+00 1.57957539e-01 2.53311008e-01 -9.00478840e-01
-8.35194439e-02 -8.98791552e-01 -4.10965502e-01 8.87056708e-01
-1.20663476e+00 -3.38772357e-01 1.38729736e-01 2.84853484e-02
-2.66153216e-02 2.99406409e-01 9.03633714e-01 4.28878546e-01
-1.96069613e-01 6.55785203e-01 -1.33306399e-01 1.04665600e-01
9.96324897e-01 -1.53786731e+00 1.62861943e-01 3.73307586e-01
1.08138740e-01 8.95317018e-01 1.34856474e+00 -4.91085023e-01
-8.10774088e-01 -6.28329396e-01 1.66411757e+00 -7.80695498e-01
8.22371960e-01 -5.50478578e-01 -6.10675275e-01 2.74812698e-01
3.78960878e-01 -2.18849972e-01 1.10229075e+00 6.12078130e-01
-7.95952141e-01 -7.42008863e-03 -1.21262157e+00 8.41208637e-01
1.11785400e+00 -4.56570536e-01 -8.59779596e-01 1.98607862e-01
6.00790679e-01 3.98970991e-01 -8.38513911e-01 3.87659878e-01
7.44171262e-01 -1.26399267e+00 8.72614980e-01 -5.80108345e-01
6.09109044e-01 8.42397008e-03 -3.60321403e-01 -1.58002794e+00
-4.54660416e-01 -1.04041351e-02 5.43336987e-01 1.33989334e+00
5.83888352e-01 -1.07541859e+00 2.05863297e-01 6.48686737e-02
5.61479390e-01 -4.82555360e-01 -8.46578598e-01 -8.99809480e-01
7.64779210e-01 -2.85614192e-01 7.40804434e-01 1.47781622e+00
2.83042789e-01 4.87814158e-01 -2.39360379e-03 -4.33812499e-01
4.61064547e-01 -5.00415228e-02 5.30221760e-01 -1.40092897e+00
1.16697662e-01 -8.34026039e-01 -7.42306292e-01 -2.18250275e-01
1.76891595e-01 -1.03188324e+00 -3.34291279e-01 -9.56391394e-01
2.67749906e-01 -1.95700869e-01 -2.69059688e-01 -8.67679939e-02
-3.83199722e-01 4.60440546e-01 9.81869474e-02 -2.90560663e-01
1.00744143e-02 3.93617213e-01 1.00870752e+00 6.89942837e-02
1.31720990e-01 -5.49390852e-01 -9.73799407e-01 8.68055165e-01
8.90403450e-01 -6.34125888e-01 -4.79925238e-02 -1.59817666e-01
9.06731367e-01 -9.41857517e-01 4.58500713e-01 -7.60288894e-01
-5.68665862e-01 -1.04889154e-01 2.50840396e-01 3.58308032e-02
-3.46705504e-02 -7.34329581e-01 -3.72569680e-01 7.63352752e-01
-4.49239939e-01 6.70971453e-01 1.02653548e-01 3.09349537e-01
-1.00938298e-01 -4.09918100e-01 6.20005548e-01 2.63374904e-03
-2.56081164e-01 -3.26069832e-01 -6.96795106e-01 1.59276098e-01
6.02069438e-01 -2.14538589e-01 -2.22127333e-01 -2.09179297e-01
-1.74485013e-01 -2.70237356e-01 6.99588597e-01 7.43284345e-01
-1.81213603e-03 -1.63839674e+00 -8.89246047e-01 -2.47779358e-02
4.21593517e-01 -1.04813707e+00 -1.26675025e-01 1.00146639e+00
-3.55564415e-01 2.06274167e-01 -9.05157700e-02 -3.90968889e-01
-1.34432304e+00 6.21129334e-01 1.67538393e-02 -1.01068921e-01
7.53778219e-02 6.19028747e-01 5.10736965e-02 -6.87673390e-01
-2.67746866e-01 -6.30053043e-01 -3.63064110e-01 1.01253247e+00
4.72964257e-01 5.82925797e-01 -1.13521419e-01 -8.21259141e-01
-7.30761409e-01 7.65267968e-01 1.29345059e-01 -3.63025010e-01
1.07442057e+00 3.81219340e-03 -3.20832014e-01 1.13780725e+00
1.56976032e+00 6.90178156e-01 1.31698728e-01 1.02085993e-01
1.82460770e-01 -6.60725296e-01 -6.54842407e-02 -4.64396536e-01
-6.44488573e-01 7.65720308e-01 8.94813418e-01 7.96404123e-01
4.36047554e-01 -1.39798164e-01 5.31319261e-01 -1.58267155e-01
8.52257758e-02 -1.37990952e+00 -1.51003957e-01 3.99180621e-01
8.08291316e-01 -1.06967175e+00 3.31162661e-01 -2.25862265e-01
-3.23057264e-01 7.69204378e-01 2.18601793e-01 -4.71673399e-01
7.14856267e-01 -7.54460394e-02 5.84209710e-02 -2.35733792e-01
-3.07120472e-01 -4.85511243e-01 2.38646045e-01 4.40942913e-01
1.17891693e+00 4.04085845e-01 -1.61435676e+00 4.30621058e-01
-9.79656637e-01 -5.78256249e-01 4.34875011e-01 5.43401182e-01
-2.64893562e-01 -1.41683304e+00 -4.06519890e-01 5.90288699e-01
-7.13592112e-01 -1.95896029e-01 -7.82763302e-01 1.06005621e+00
3.37498754e-01 7.27109432e-01 6.48368359e-01 -5.46204031e-01
3.11434627e-01 3.13521981e-01 6.17498040e-01 -4.52322960e-01
-8.45578372e-01 -6.21450543e-01 3.31230134e-01 -3.70426118e-01
-5.53299010e-01 -7.78473258e-01 -8.45194578e-01 -6.50160491e-01
-2.35359460e-01 2.36746296e-01 6.22310102e-01 5.75387120e-01
1.30293384e-01 2.62070358e-01 3.54841411e-01 -4.91505325e-01
-5.45621753e-01 -1.28126478e+00 -6.71272159e-01 8.54751945e-01
4.52248007e-01 -8.89289141e-01 -8.25345278e-01 -6.74567282e-01] | [9.368722915649414, 10.130059242248535] |
796af7ae-e53f-48c9-97a4-b8087f51bee4 | a-multi-task-network-to-detect-junctions-in | 1806.03175 | null | https://arxiv.org/abs/1806.03175v1 | https://arxiv.org/pdf/1806.03175v1.pdf | A Multi-task Network to Detect Junctions in Retinal Vasculature | Junctions in the retinal vasculature are key points to be able to extract its topology, but they vary in appearance, depending on vessel density, width and branching/crossing angles. The complexity of junction patterns is usually accompanied by a scarcity of labels, which discourages the usage of very deep networks for their detection. We propose a multi-task network, generating labels for vessel interior, centerline, edges and junction patterns, to provide additional information to facilitate junction detection. After the initial detection of potential junctions in junction-selective probability maps, candidate locations are re-examined in centerline probability maps to verify if they connect at least 3 branches. The experiments on the DRIVE and IOSTAR showed that our method outperformed a recent study in which a popular deep network was trained as a classifier to find junctions. Moreover, the proposed approach is applicable to unseen datasets with the same degree of success, after training it only once. | [] | 2018-06-06 | null | null | null | null | ['junction-detection'] | ['computer-vision'] | [-2.99623981e-03 -3.50909494e-03 -8.87870789e-02 -3.38353813e-01
-1.51320353e-01 -8.41492176e-01 6.90398812e-01 5.14230847e-01
-5.64754069e-01 7.64435887e-01 -1.49159417e-01 -5.39371133e-01
-2.16886327e-01 -8.70591938e-01 -4.92082924e-01 -6.40421450e-01
-2.02530533e-01 1.91515580e-01 7.32162714e-01 1.53769031e-01
4.27697808e-01 9.84103203e-01 -1.67144692e+00 3.38934399e-02
8.38222682e-01 1.04327726e+00 3.74558382e-02 5.96625388e-01
-4.98957127e-01 3.51470709e-01 -5.42047441e-01 -3.97092998e-01
3.70512635e-01 -1.81429312e-01 -6.86989427e-01 -6.77476525e-02
8.62450004e-01 -7.91351944e-02 3.87453772e-02 9.57314909e-01
6.23392582e-01 -2.77441740e-01 9.64053571e-01 -8.51351857e-01
6.29180169e-04 2.26864010e-01 -5.00240803e-01 6.82881951e-01
4.76582199e-02 1.26533881e-01 1.13138855e+00 -5.73960483e-01
8.76004755e-01 6.25308216e-01 6.29849374e-01 2.16703996e-01
-1.28504264e+00 -2.85949916e-01 -2.10879669e-02 1.26755953e-01
-1.30024827e+00 -3.52525771e-01 5.36660194e-01 -7.47305453e-01
5.95851183e-01 1.04983293e-01 8.31066132e-01 9.13784981e-01
-2.45046124e-01 3.36388916e-01 1.44887865e+00 -4.89771694e-01
1.92987457e-01 2.49138951e-01 1.72993690e-01 8.17188025e-01
4.17090744e-01 1.32993355e-01 -1.15859844e-01 -7.63398930e-02
1.14012957e+00 -4.15408790e-01 -2.28961930e-01 -4.39365298e-01
-8.47703278e-01 6.41391814e-01 6.24461591e-01 5.73921561e-01
-2.63800144e-01 -2.78123081e-01 3.04738253e-01 1.23349823e-01
1.13637313e-01 5.13626397e-01 -1.86094671e-01 -5.83785307e-03
-8.66533995e-01 1.50793698e-02 8.17976892e-01 5.80183148e-01
7.04104125e-01 -4.96699065e-01 -1.80837885e-01 9.22997653e-01
4.75864410e-02 -9.44977030e-02 2.47220546e-01 -6.81080639e-01
8.74230489e-02 1.01745546e+00 1.30053144e-02 -8.82257223e-01
-8.90735209e-01 -6.12182021e-01 -8.65315557e-01 7.51321018e-01
1.15183604e+00 -9.81897488e-02 -9.96383250e-01 1.21399176e+00
3.87305915e-01 2.51435846e-01 -3.12951088e-01 1.08096671e+00
9.94453728e-01 1.89715430e-01 -3.96987051e-02 2.32142001e-01
1.39208329e+00 -7.01047599e-01 -1.59365922e-01 6.13368396e-03
6.94851696e-01 -1.07096756e+00 7.45528758e-01 4.40107942e-01
-7.04930663e-01 -4.59039360e-01 -8.58430684e-01 -1.92888137e-02
-5.42327702e-01 6.94429636e-01 6.95065379e-01 6.18765891e-01
-1.13026321e+00 6.51246905e-01 -5.46496093e-01 -3.93881232e-01
6.93916261e-01 2.55786836e-01 -5.03719747e-01 2.97556221e-01
-6.63484454e-01 9.94781017e-01 1.94581643e-01 2.44711712e-01
-5.46768248e-01 -5.41231692e-01 -5.62173367e-01 -9.76598542e-03
8.42498392e-02 -7.19794691e-01 7.34554470e-01 -5.29449522e-01
-1.42806232e+00 1.10795105e+00 -2.65318185e-01 -6.06291294e-01
8.51371944e-01 9.45073590e-02 -3.81197035e-01 4.64404225e-01
-4.86225970e-02 6.66356385e-01 7.32546449e-01 -9.98578191e-01
-1.18941927e+00 -5.56321219e-02 3.78785908e-01 -2.91948020e-01
3.52363586e-02 -1.86391324e-01 -3.92369479e-01 -4.55061704e-01
1.62181199e-01 -6.70118451e-01 -3.03468496e-01 4.88114297e-01
-8.21695507e-01 -4.10373539e-01 2.37996817e-01 -5.20383358e-01
9.67415333e-01 -2.05449939e+00 -1.16698653e-01 5.72486222e-01
5.35944760e-01 4.12145555e-01 -1.53713748e-01 6.99182674e-02
2.30920017e-02 3.05236608e-01 -7.70722702e-02 -9.68127474e-02
-4.15494144e-01 -9.00675505e-02 1.06125474e-01 3.35207909e-01
4.69886124e-01 4.78858054e-01 -7.57507265e-01 -6.35055065e-01
5.04749298e-01 5.60409963e-01 -2.18821347e-01 -1.08063899e-01
1.81285627e-02 5.37752450e-01 -4.15276349e-01 5.01528859e-01
6.09844744e-01 -3.50950718e-01 -1.46055534e-01 -3.67541254e-01
-4.49834943e-01 1.22233026e-01 -1.25277507e+00 1.19857132e+00
-6.84407890e-01 9.56259727e-01 -2.57397830e-01 -9.89949882e-01
1.21569955e+00 2.33889610e-01 3.08341801e-01 -5.33011556e-01
1.47787914e-01 4.84446347e-01 4.03737813e-01 -4.38398838e-01
-9.43248644e-02 3.44194949e-01 5.97817838e-01 1.13261482e-02
9.83167514e-02 1.78119645e-01 6.68767631e-01 -1.83090761e-01
8.57347012e-01 1.46660373e-01 2.22930819e-01 -1.92718714e-01
7.99565077e-01 -3.52105722e-02 4.75579411e-01 6.88785791e-01
-5.50370924e-02 7.79287279e-01 9.37170267e-01 -8.76865208e-01
-9.34012234e-01 -9.86230135e-01 -7.78716445e-01 3.03707391e-01
2.52125084e-01 -2.37952322e-01 -4.27149415e-01 -7.87244797e-01
-1.63909957e-01 2.89430290e-01 -5.64468443e-01 2.96277195e-01
-5.24551153e-01 -6.54388249e-01 2.65694678e-01 1.40811384e-01
4.38694149e-01 -7.96154439e-01 -8.41617346e-01 1.93831697e-01
1.42564207e-01 -1.23756981e+00 1.42905265e-01 7.86334425e-02
-8.76336038e-01 -1.42519581e+00 -9.29916799e-01 -8.77449632e-01
1.00461316e+00 -9.01992172e-02 7.99507260e-01 1.83088914e-01
-8.19490194e-01 -2.84754485e-01 -2.27553502e-01 -1.66577235e-01
-4.15845066e-01 1.72681719e-01 -4.95763659e-01 9.37994048e-02
2.32120395e-01 -7.03464806e-01 -8.23387027e-01 6.37804687e-01
-3.81987214e-01 -2.41730258e-01 8.51787508e-01 7.80170918e-01
6.09125316e-01 4.60038781e-02 4.84183729e-01 -8.80682707e-01
3.75433058e-01 -1.09507754e-01 -1.01204073e+00 1.39166772e-01
-4.55104440e-01 -1.17738269e-01 6.64081454e-01 -2.08701923e-01
-6.79438889e-01 5.20945042e-02 -2.83073992e-01 8.97447616e-02
-7.32241809e-01 3.27279091e-01 3.53597701e-01 -5.48807979e-01
9.42030311e-01 2.16474058e-03 9.23968554e-02 -4.96719688e-01
1.77951708e-01 3.20414454e-01 2.74926811e-01 -3.58185880e-02
5.47458887e-01 5.65400481e-01 4.45887953e-01 -8.83790851e-01
-6.45037532e-01 -6.05204582e-01 -8.72091174e-01 -2.24109873e-01
6.83151722e-01 -4.42802668e-01 -5.51992118e-01 2.86829680e-01
-1.09433877e+00 3.41740437e-03 -1.67146221e-01 6.75338507e-01
-1.60360381e-01 4.38448519e-01 -4.10191894e-01 -5.96050620e-01
2.58480310e-02 -1.23609352e+00 7.57055998e-01 5.96378744e-01
-9.39320493e-03 -1.09565663e+00 -1.82720441e-02 2.09252760e-01
4.81090993e-01 3.31439674e-01 1.08174682e+00 -6.61262095e-01
-7.85403371e-01 -5.12759030e-01 -4.67882514e-01 2.73917377e-01
1.78007707e-01 5.73074520e-01 -8.84795547e-01 1.87386349e-01
-7.40182817e-01 -1.79305077e-02 1.04323971e+00 7.00538754e-01
1.06747663e+00 1.28614008e-01 -5.00347078e-01 5.58484972e-01
1.32596910e+00 -1.74612533e-02 5.27794600e-01 4.83302057e-01
2.15711877e-01 7.28320301e-01 1.81503728e-01 3.27982187e-01
-3.19871157e-02 5.58050275e-01 6.45164371e-01 -7.61773348e-01
-4.38903093e-01 1.56047851e-01 -2.19779551e-01 -8.25421587e-02
-4.16961670e-01 -2.43663229e-02 -7.91385710e-01 6.19738877e-01
-1.37559521e+00 -5.01510024e-01 -5.45151591e-01 2.47103333e+00
5.05452514e-01 4.84781086e-01 4.55487102e-01 -1.55110031e-01
8.37910473e-01 -1.00774929e-01 -3.26856375e-01 -3.11676651e-01
-2.43357003e-01 3.15332323e-01 5.31674922e-01 4.19836998e-01
-1.29278767e+00 8.68459165e-01 6.54944038e+00 6.84222937e-01
-1.44120002e+00 -4.28682148e-01 8.65916252e-01 1.42240509e-01
-1.92402769e-02 1.18596986e-01 -1.05545092e+00 3.95220280e-01
3.54866773e-01 3.19397926e-01 5.19009978e-02 5.37953496e-01
8.32359716e-02 -4.62398291e-01 -7.16459036e-01 7.47098744e-01
-3.55905056e-01 -1.61107206e+00 -6.60101175e-02 9.84725356e-02
4.52897549e-01 1.40808418e-01 -1.93867788e-01 -2.15151772e-01
5.46659976e-02 -8.93320441e-01 -1.15740942e-02 5.90657175e-01
5.76839805e-01 -4.06157792e-01 1.04567719e+00 9.75404903e-02
-8.21411133e-01 4.28267643e-02 -5.44683695e-01 1.35987341e-01
3.50735754e-01 9.35817719e-01 -1.25389171e+00 4.57837731e-01
5.30901849e-01 5.43707192e-01 -8.47417057e-01 2.02381730e+00
-5.06592631e-01 5.54073453e-01 -7.25182176e-01 -1.32458016e-01
2.43395954e-01 -5.47513127e-01 6.99899375e-01 1.05059934e+00
4.23484504e-01 -7.14620829e-01 -1.44027814e-01 1.01494241e+00
3.59627217e-01 5.14143229e-01 -5.55778325e-01 2.59660929e-01
4.92090166e-01 1.67730546e+00 -1.20359576e+00 -6.89389408e-02
-5.13729751e-01 4.55499470e-01 3.38245004e-01 2.07873285e-01
-5.68567991e-01 -5.25699914e-01 2.27822170e-01 3.68407637e-01
3.16383153e-01 -7.82349519e-03 -2.20686957e-01 -7.26890445e-01
1.43543050e-01 -1.45456895e-01 4.13381219e-01 -3.66809011e-01
-1.36136127e+00 8.47841322e-01 -3.36672395e-01 -1.39023137e+00
-5.92548808e-04 -7.99603939e-01 -1.07200634e+00 9.22320366e-01
-1.92366219e+00 -1.06009400e+00 -4.20918554e-01 3.15633535e-01
9.45160910e-02 -2.50392646e-01 8.30002844e-01 3.70135993e-01
-6.40659750e-01 4.24074173e-01 -1.82942554e-01 2.31331408e-01
6.49909198e-01 -1.46957445e+00 2.64989197e-01 8.48150373e-01
2.97921330e-01 4.92616355e-01 4.47907031e-01 -3.81035388e-01
-4.92317915e-01 -8.55252147e-01 8.28474402e-01 -8.92619640e-02
7.20882118e-01 -1.61216855e-01 -6.96199179e-01 2.35041231e-01
3.96588147e-02 4.23605561e-01 5.77144802e-01 3.47606182e-01
-2.18566954e-01 -2.13867277e-01 -9.77041960e-01 5.81087232e-01
9.42698717e-01 -6.12009615e-02 -3.42617899e-01 4.62791413e-01
-4.19580862e-02 -1.70916662e-01 -9.85343158e-01 3.85778725e-01
4.94763017e-01 -1.43304312e+00 8.64251852e-01 -2.86725521e-01
3.47113460e-01 -4.56600785e-01 6.39813304e-01 -1.16874743e+00
-2.42314398e-01 -5.00064850e-01 2.90192813e-01 1.17638731e+00
8.58750820e-01 -1.00321651e+00 9.11249697e-01 3.12457476e-02
-1.38770998e-01 -7.40578592e-01 -1.20693374e+00 -5.40896177e-01
-1.60536095e-01 1.75484404e-01 2.17976540e-01 5.66414595e-01
-3.12323481e-01 2.54253447e-01 6.39766082e-02 2.17064932e-01
4.03984040e-01 4.34195101e-01 8.29231143e-01 -1.66735768e+00
5.37757985e-02 -1.02536702e+00 -9.12650585e-01 -9.64532793e-01
-1.22905977e-01 -9.72197592e-01 -4.59100515e-01 -1.61744416e+00
-3.90846819e-01 -9.44646418e-01 -2.17123285e-01 3.38468552e-01
-3.86229940e-02 3.69828641e-01 -2.08261043e-01 7.85260350e-02
1.97103694e-02 3.21752243e-02 1.47353256e+00 1.82172909e-01
-3.63107711e-01 4.81130928e-01 -3.34798068e-01 8.68429422e-01
9.13317859e-01 -2.04786226e-01 -2.43651971e-01 -2.97605470e-02
1.73670664e-01 -1.65007263e-01 5.42479396e-01 -1.21199000e+00
2.46546283e-01 3.52414340e-01 4.34149921e-01 -6.30212545e-01
1.83703929e-01 -7.43891537e-01 -3.14896107e-01 3.05473834e-01
-3.20390701e-01 -4.01987344e-01 7.08823055e-02 3.81270856e-01
-3.82951647e-01 -5.53381324e-01 7.85898387e-01 -4.82972749e-02
-8.05979788e-01 2.67375648e-01 -5.04169643e-01 -1.07466809e-01
1.05601716e+00 -5.54927945e-01 -5.37302673e-01 9.69120711e-02
-1.06027126e+00 9.33053121e-02 4.69113946e-01 1.41547307e-01
4.50604081e-01 -6.81249142e-01 -7.02185690e-01 3.10861647e-01
3.02608401e-01 1.72828421e-01 2.98284322e-01 1.26396489e+00
-9.13380384e-01 1.88981771e-01 -4.67018247e-01 -9.27474737e-01
-1.29092383e+00 3.03079307e-01 6.06240153e-01 -1.97058488e-02
-1.02462971e+00 9.04610276e-01 -4.14006151e-02 -5.31654060e-02
3.06668043e-01 -6.57948852e-01 -8.86990130e-01 1.81964666e-01
3.56704831e-01 2.69779772e-01 9.34232771e-02 -3.34005892e-01
-1.27695009e-01 9.83890295e-01 -9.81418192e-02 3.48374724e-01
1.17493176e+00 9.92452912e-03 -2.03195393e-01 2.31900066e-01
8.41679931e-01 1.89745933e-01 -1.14880168e+00 -1.42586097e-01
8.25668126e-02 -5.30998945e-01 1.28133759e-01 -8.24344516e-01
-1.29451144e+00 9.70716953e-01 7.43749678e-01 4.35780942e-01
8.76353741e-01 -1.68692887e-01 5.32015681e-01 1.29146963e-01
2.46296942e-01 -6.28127456e-01 -3.86718571e-01 7.77005926e-02
5.38102686e-01 -1.28474116e+00 -1.75705269e-01 -8.62537801e-01
-2.87787408e-01 1.56619859e+00 6.21232629e-01 -2.52415508e-01
8.86544526e-01 9.71744210e-02 2.73319542e-01 -2.56986827e-01
-3.76300216e-01 -6.10685527e-01 4.59856510e-01 7.98524737e-01
4.71945971e-01 -1.85029432e-01 -4.78548497e-01 -2.54478544e-01
-1.96631730e-01 8.88985991e-02 7.23314941e-01 3.85955483e-01
-3.29658538e-01 -1.19268370e+00 6.12650476e-02 6.99423492e-01
-4.63882864e-01 1.26361577e-02 -3.60575080e-01 1.02045560e+00
3.32682014e-01 7.24339843e-01 3.47353637e-01 9.85754356e-02
4.35476422e-01 -1.12869099e-01 3.21214944e-01 -4.32841003e-01
-4.82343107e-01 1.48639411e-01 4.35882658e-01 -4.88458127e-01
-3.75922590e-01 -3.93187732e-01 -9.93940294e-01 3.49836141e-01
-3.17673832e-01 2.42792070e-02 7.38880157e-01 8.04966748e-01
4.40341353e-01 3.62810194e-01 5.56999266e-01 -3.55618298e-01
-1.63221732e-01 -8.22544336e-01 -6.08954251e-01 4.95334059e-01
5.15839279e-01 -8.66607666e-01 -5.69672227e-01 5.10857813e-02] | [15.769947052001953, -3.9341750144958496] |
3a451474-dd79-43f4-907c-02e00a26e73b | language-modeling-via-stochastic-processes-1 | 2203.11370 | null | https://arxiv.org/abs/2203.11370v2 | https://arxiv.org/pdf/2203.11370v2.pdf | Language modeling via stochastic processes | Modern language models can generate high-quality short texts. However, they often meander or are incoherent when generating longer texts. These issues arise from the next-token-only language modeling objective. Recent work in self-supervised learning suggests that models can learn good latent representations via contrastive learning, which can be effective for discriminative tasks. Our work analyzes the application of contrastive representations for generative tasks, like long text generation. We propose one approach for leveraging constrastive representations, which we call Time Control (TC). TC first learns a contrastive representation of the target text domain, then generates text by decoding from these representations. Compared to domain-specific methods and fine-tuning GPT2 across a variety of text domains, TC performs competitively to methods specific for learning sentence representations on discourse coherence. On long text generation settings, TC preserves the text structure both in terms of ordering (up to $+15\%$ better) and text length consistency (up to $+90\%$ better). | ['Tatsunori Hashimoto', 'Noah Goodman', 'Esin Durmus', 'Rose E Wang'] | 2022-03-21 | language-modeling-via-stochastic-processes | https://openreview.net/forum?id=pMQwKL1yctf | https://openreview.net/pdf?id=pMQwKL1yctf | iclr-2022-4 | ['text-infilling'] | ['natural-language-processing'] | [ 3.66859704e-01 7.05463111e-01 -4.06021327e-01 -3.50145757e-01
-1.14504147e+00 -4.70040977e-01 1.23124552e+00 8.35911930e-02
-1.67582750e-01 1.03736448e+00 1.00507545e+00 -2.16453195e-01
6.76440001e-02 -8.88717711e-01 -6.24341547e-01 -4.80512619e-01
1.21217281e-01 7.90762007e-01 -4.09646153e-01 -4.77329433e-01
3.81066799e-01 -1.17874272e-01 -1.38810706e+00 6.38454616e-01
1.07485914e+00 1.92031458e-01 2.50642955e-01 6.84901893e-01
-5.60866654e-01 1.02296162e+00 -9.55851018e-01 -1.98639274e-01
-1.74044043e-01 -8.29452574e-01 -1.17017055e+00 -1.32777179e-02
3.85455340e-01 -2.13129550e-01 -2.76584148e-01 5.70566058e-01
5.54469287e-01 2.68969804e-01 1.02306390e+00 -9.80959415e-01
-8.95869195e-01 1.39947903e+00 -1.88964367e-01 2.17025325e-01
4.21669781e-01 1.50272682e-01 1.36540198e+00 -6.94491744e-01
8.93514156e-01 1.46163785e+00 4.33005363e-01 7.56039023e-01
-1.53015280e+00 -4.90661412e-01 3.27376842e-01 -2.78500021e-01
-8.04061592e-01 -7.49997139e-01 7.23806560e-01 -4.72608060e-01
1.46015036e+00 1.90200463e-01 3.64211231e-01 1.52681398e+00
2.84162819e-01 1.01737130e+00 9.68202651e-01 -5.70744991e-01
1.26619220e-01 -1.53766736e-01 1.58862606e-01 3.14459682e-01
1.12804368e-01 2.30264962e-02 -7.47638524e-01 -8.62785801e-02
4.77168828e-01 -4.10961986e-01 -6.79853559e-02 1.17253311e-01
-1.34737170e+00 1.12879896e+00 2.67380942e-02 5.51826656e-01
-2.09648699e-01 3.45227927e-01 4.13870841e-01 4.44718510e-01
9.50273514e-01 9.30810034e-01 -2.28129178e-01 -4.40379381e-01
-1.30001783e+00 6.93093777e-01 8.86371911e-01 1.09833145e+00
6.79390073e-01 4.81200725e-01 -7.68332899e-01 9.27392483e-01
1.24691539e-01 3.99136782e-01 9.83295143e-01 -7.33524024e-01
9.25629020e-01 3.36968631e-01 6.82985485e-02 -7.27887750e-01
-3.26321334e-01 -3.55437875e-01 -8.68935049e-01 -2.27223977e-01
1.85307056e-01 -3.19492102e-01 -6.79621518e-01 1.98926735e+00
-3.31318468e-01 -2.16509551e-01 2.53776431e-01 3.20769608e-01
7.85655618e-01 1.15350902e+00 1.76891804e-01 -5.39022446e-01
8.95643592e-01 -1.05395758e+00 -8.49398434e-01 -6.36709690e-01
1.04240918e+00 -8.92551184e-01 1.20312917e+00 9.44043919e-02
-1.42161584e+00 -5.38316965e-01 -9.02304590e-01 -3.02762628e-01
-8.62124488e-02 1.13448352e-01 4.75984901e-01 4.84006792e-01
-1.12652075e+00 8.37449670e-01 -7.54367352e-01 -2.16957837e-01
2.89025277e-01 2.20681261e-02 -3.36425602e-02 4.71943803e-02
-1.21560776e+00 9.16270256e-01 6.01014376e-01 -3.36252898e-01
-7.40477085e-01 -7.23515391e-01 -1.01878858e+00 9.86175090e-02
1.05967715e-01 -8.38995993e-01 1.42424548e+00 -9.31321919e-01
-1.68773460e+00 8.21381807e-01 -3.67452681e-01 -6.20406747e-01
6.00238264e-01 -5.94549835e-01 -1.91078886e-01 -2.09758699e-01
4.17284250e-01 8.20887566e-01 8.62931669e-01 -9.86110270e-01
-2.27270260e-01 7.76523948e-02 -3.98173593e-02 3.97830427e-01
-3.56927305e-01 -2.62897938e-01 2.56341010e-01 -9.81953025e-01
-1.74522951e-01 -8.20255578e-01 -2.59459108e-01 -5.58857024e-01
-5.74834943e-01 -7.21187055e-01 5.57757318e-01 -6.61667883e-01
1.37801468e+00 -1.65347624e+00 4.72674757e-01 -3.04439664e-01
2.10252941e-01 6.11472093e-02 -4.96780038e-01 7.85331309e-01
-6.63206205e-02 4.23124403e-01 -1.77886054e-01 -6.44470215e-01
1.65248230e-01 2.61230618e-01 -6.86341047e-01 8.17638412e-02
4.86360043e-01 1.22771382e+00 -1.04477191e+00 -4.45552438e-01
-7.98273757e-02 5.18750884e-02 -6.39685690e-01 4.34600830e-01
-7.98001826e-01 1.58583358e-01 -3.51235032e-01 2.00805571e-02
1.79982558e-01 -4.58422869e-01 5.16626656e-01 3.13016236e-01
-1.10623084e-01 9.36356246e-01 -7.97380865e-01 1.94548142e+00
-7.30989814e-01 8.23174894e-01 -5.49153626e-01 -9.24342692e-01
1.12718499e+00 4.69065994e-01 1.52291954e-01 -7.24293232e-01
-1.34795949e-01 7.03656822e-02 -1.51230410e-01 -4.50150728e-01
1.06824315e+00 -2.60367334e-01 -2.47507557e-01 1.16751873e+00
2.40413427e-01 -5.64935625e-01 4.69683409e-01 4.46527421e-01
8.36631238e-01 3.55006784e-01 3.50482762e-01 -5.16329348e-01
-7.97872990e-02 -9.46935266e-02 2.37687558e-01 7.75335729e-01
3.60029727e-01 5.69474816e-01 6.67027533e-01 -2.77775973e-01
-1.30153918e+00 -8.62275720e-01 2.28487402e-01 1.49836695e+00
-3.13819855e-01 -7.93770254e-01 -6.96601987e-01 -5.76986969e-01
-1.94938719e-01 1.29941118e+00 -5.47203839e-01 -3.12221795e-01
-9.83697355e-01 -6.83885038e-01 5.54909289e-01 5.97716212e-01
1.25445858e-01 -1.25758445e+00 -3.85618448e-01 4.52471048e-01
-5.28810441e-01 -6.90061748e-01 -5.23987591e-01 1.25368625e-01
-1.05643177e+00 -4.75413054e-01 -7.13940501e-01 -6.59780920e-01
5.11208773e-01 1.16093196e-01 1.53266811e+00 -2.23415322e-03
1.44486025e-01 -1.17145814e-02 -4.36102986e-01 -3.03867340e-01
-9.22520101e-01 6.46974087e-01 -1.87006280e-01 -4.93519455e-01
5.93005167e-03 -5.62025011e-01 -2.08982334e-01 -3.40461791e-01
-9.02008593e-01 4.59989965e-01 4.79502708e-01 1.14050806e+00
2.08340660e-01 -3.94228458e-01 8.88044894e-01 -1.20627427e+00
1.26762187e+00 -6.60781443e-01 -1.93367139e-01 1.86289176e-01
-7.87736535e-01 5.86202800e-01 7.24518061e-01 -4.45700407e-01
-1.22325730e+00 -5.16627014e-01 -4.88792211e-02 1.23877391e-01
2.69946922e-02 6.17589474e-01 1.01633400e-01 9.98600304e-01
1.07939959e+00 3.82673502e-01 1.11399982e-02 -2.36903220e-01
7.59968877e-01 4.06191081e-01 1.62540704e-01 -8.73840332e-01
6.90315068e-01 2.94785947e-02 -4.11691695e-01 -7.30857372e-01
-9.67354536e-01 1.06171921e-01 -5.78605771e-01 1.20975927e-01
6.97663963e-01 -8.96654189e-01 4.67984602e-02 1.75455526e-01
-1.32247233e+00 -7.18358994e-01 -5.92507601e-01 1.71166554e-01
-7.98686326e-01 2.71403342e-01 -6.91868365e-01 -7.07547069e-01
-6.46737158e-01 -8.11520517e-01 1.27049279e+00 8.32974985e-02
-8.63303781e-01 -1.23743105e+00 3.43186527e-01 2.68282324e-01
6.09455109e-01 2.64227271e-01 1.23341513e+00 -6.69455051e-01
-4.12423462e-01 2.07172409e-01 1.29627004e-01 1.92265153e-01
2.22753182e-01 -1.37874428e-02 -9.59181309e-01 -3.46617937e-01
-2.93646097e-01 -5.85672021e-01 1.06587780e+00 4.13383871e-01
9.65452373e-01 -8.46592486e-01 -2.70588368e-01 2.57449895e-01
8.74135554e-01 -8.64165202e-02 7.44769216e-01 1.55807421e-01
5.85623622e-01 7.13213623e-01 4.88459945e-01 5.50632775e-01
4.21215981e-01 5.49255013e-01 -3.67154516e-02 -1.39142182e-02
-3.15362662e-01 -5.17094314e-01 7.02839434e-01 1.00324774e+00
4.70805205e-02 -4.87499595e-01 -9.11985099e-01 6.56888008e-01
-1.84067023e+00 -1.29673612e+00 3.62904673e-03 1.76354992e+00
1.44942713e+00 3.36066872e-01 5.56388944e-02 -1.69682931e-02
5.32453179e-01 6.91911459e-01 -2.21654207e-01 -6.60683930e-01
-2.46054173e-01 5.15231550e-01 -1.50045425e-01 5.93621254e-01
-8.14764440e-01 1.19690251e+00 6.99481153e+00 9.38623011e-01
-1.08450425e+00 1.08097531e-01 7.05756247e-01 -2.83982664e-01
-9.58330810e-01 1.32395774e-01 -8.33782852e-01 4.21700001e-01
1.24046683e+00 -6.61954284e-01 1.85731500e-01 7.82080710e-01
3.34239870e-01 1.48761049e-01 -1.48262918e+00 7.05042839e-01
2.66916037e-01 -1.92398500e+00 4.11312848e-01 -9.46279466e-02
1.20705903e+00 -2.91385353e-01 1.77919921e-02 6.40589058e-01
6.63634896e-01 -1.45647693e+00 9.45464313e-01 4.25773799e-01
8.50853860e-01 -7.24826813e-01 5.17902315e-01 5.77659667e-01
-8.43976319e-01 1.81418747e-01 -3.50714058e-01 -2.87772119e-01
2.91691422e-01 5.93895435e-01 -1.23787081e+00 4.88921255e-01
8.59591812e-02 9.40880299e-01 -5.95052421e-01 1.63476437e-01
-4.16482151e-01 7.26170421e-01 1.66720465e-01 -2.30909720e-01
2.66519457e-01 5.01460135e-02 4.65079397e-01 1.49368024e+00
3.76644373e-01 -2.68098861e-01 2.89100200e-01 1.26552808e+00
-2.83567667e-01 1.56415880e-01 -7.79910862e-01 -5.27539611e-01
5.80870092e-01 8.06734860e-01 -3.04815471e-01 -5.52378297e-01
9.82394218e-02 8.51703107e-01 4.63253617e-01 2.69143492e-01
-5.17368197e-01 -1.47353709e-01 3.05681467e-01 9.85225588e-02
-1.65960956e-02 -4.19091612e-01 -5.01886845e-01 -1.22329986e+00
-1.07758477e-01 -9.90125179e-01 2.51841068e-01 -6.46372497e-01
-1.26002002e+00 7.90826261e-01 2.15554655e-01 -1.08664107e+00
-1.15262365e+00 -8.35307464e-02 -7.92628765e-01 9.33330476e-01
-1.40686011e+00 -1.29630172e+00 1.39064118e-02 2.75141716e-01
1.04178572e+00 -4.16303009e-01 1.09140146e+00 -3.67705286e-01
-3.05927038e-01 5.77906311e-01 2.14607567e-01 -4.48559038e-03
7.49767423e-01 -1.40113497e+00 7.70389080e-01 7.60960758e-01
3.18806231e-01 7.14623928e-01 7.68489718e-01 -6.86294913e-01
-9.75020409e-01 -1.25694656e+00 1.50603878e+00 -5.56724548e-01
5.02435684e-01 -4.30954009e-01 -8.04344952e-01 8.30899537e-01
7.88932562e-01 -5.69397986e-01 7.36091435e-01 3.99004072e-01
-4.96124268e-01 2.12585941e-01 -6.88956439e-01 8.34228098e-01
1.06671119e+00 -5.56437433e-01 -1.01638198e+00 6.65296078e-01
9.92826641e-01 -3.78524154e-01 -7.57255197e-01 -6.80758357e-02
1.64024323e-01 -6.90399051e-01 6.14868164e-01 -8.12896192e-01
1.13464963e+00 2.27292627e-01 1.17298789e-01 -1.64643586e+00
-4.45520341e-01 -9.02226627e-01 -2.60008752e-01 1.39899588e+00
5.81788063e-01 -3.63574952e-01 5.30998647e-01 1.97769001e-01
-4.07054961e-01 -5.21057129e-01 -7.52788186e-01 -8.03991020e-01
8.55069101e-01 -1.26438543e-01 5.54346561e-01 1.02300251e+00
1.66148096e-01 9.20401096e-01 -4.83612835e-01 -4.99715954e-01
2.95858979e-01 3.93240690e-01 7.62824953e-01 -1.06574845e+00
-3.54644418e-01 -7.00512767e-01 3.67078692e-01 -1.14038074e+00
5.96885562e-01 -1.25891721e+00 1.48660377e-01 -1.72466648e+00
1.76439181e-01 -4.53084618e-01 2.90920049e-01 3.96571249e-01
-3.26022804e-01 -2.83225566e-01 1.30281165e-01 3.14243078e-01
-4.24385399e-01 6.87686622e-01 1.20403123e+00 -3.00915569e-01
-3.26965272e-01 -2.16963395e-01 -9.82223511e-01 3.25242996e-01
9.41789091e-01 -4.44591314e-01 -6.31735682e-01 -7.69762993e-01
4.48336989e-01 1.67504191e-01 -6.27645524e-03 -6.23809874e-01
-9.50071365e-02 -5.01181126e-01 1.81051224e-01 -4.55325633e-01
2.15161026e-01 2.43238315e-01 4.12292480e-02 3.49977851e-01
-1.09577072e+00 8.70445892e-02 4.91624810e-02 3.28954071e-01
-2.57181704e-01 -3.34732115e-01 6.74719095e-01 -3.95692766e-01
-3.00891697e-01 -5.86736714e-04 -5.43634057e-01 5.90854645e-01
5.99015713e-01 -7.95942992e-02 -7.21803606e-01 -6.08314157e-01
-4.53654677e-01 1.09080739e-01 2.33976141e-01 6.83560967e-01
4.97156829e-01 -1.40898430e+00 -1.24955535e+00 6.05321787e-02
6.93701208e-02 8.14881772e-02 1.06972255e-01 3.85736138e-01
-1.85766935e-01 6.06475770e-01 -4.64646518e-02 -5.55749297e-01
-8.97210181e-01 1.56438306e-01 9.04190019e-02 -8.09029758e-01
-6.03775680e-01 8.87497067e-01 9.39473808e-02 -2.80294329e-01
-1.00061305e-01 -3.05008262e-01 -1.72167912e-01 2.91629821e-01
4.90259796e-01 2.21870780e-01 -1.56160519e-02 -3.71620893e-01
7.61639327e-02 9.38871801e-02 -4.84923691e-01 -3.65430683e-01
1.43400872e+00 1.26268007e-02 -1.33686453e-01 5.71540415e-01
9.86501575e-01 4.35639992e-02 -1.17113578e+00 -2.57295132e-01
3.08510661e-01 -1.97587013e-01 -2.87329018e-01 -6.43491328e-01
-5.07888258e-01 9.30627823e-01 -1.21349785e-02 3.49342674e-01
5.89683294e-01 1.46265015e-01 6.90902591e-01 3.50548744e-01
4.13594134e-02 -1.24033463e+00 8.35125446e-01 9.88151133e-01
1.25170493e+00 -9.82260227e-01 -9.73643810e-02 6.92009926e-02
-8.65479827e-01 1.17994666e+00 4.70116287e-01 -2.26492852e-01
1.30943075e-01 2.49037892e-01 -1.55446902e-01 6.31457716e-02
-1.31593239e+00 3.50113362e-02 2.70038903e-01 6.58935130e-01
1.08309436e+00 2.35166758e-01 -4.19012666e-01 5.28033197e-01
-8.37052584e-01 -3.45343053e-01 7.69615948e-01 8.20265710e-01
-5.31384110e-01 -1.46752679e+00 -9.71821696e-02 5.53997815e-01
-2.48488501e-01 -4.51242983e-01 -3.48810196e-01 6.06394470e-01
-1.59610286e-01 1.06182754e+00 2.41555691e-01 -5.31357378e-02
-7.57016838e-02 4.84378219e-01 4.49107260e-01 -1.12078249e+00
-7.88768888e-01 1.78462818e-01 5.48000276e-01 -2.83424646e-01
-4.17205989e-01 -6.96080267e-01 -1.25606179e+00 -4.10280704e-01
-1.13952309e-01 2.06468791e-01 1.99727327e-01 1.10757375e+00
3.39991957e-01 6.39290571e-01 5.73093772e-01 -8.18704247e-01
-7.93638706e-01 -1.30109811e+00 -2.80484796e-01 5.76633275e-01
2.19499841e-01 -2.98897654e-01 -1.50913224e-01 3.44664842e-01] | [11.71378231048584, 8.965578079223633] |
f7c9bff7-ed2c-45e6-a7da-cc38ffb7c9a9 | an-automated-vulnerability-detection | 2301.08824 | null | https://arxiv.org/abs/2301.08824v1 | https://arxiv.org/pdf/2301.08824v1.pdf | An Automated Vulnerability Detection Framework for Smart Contracts | With the increase of the adoption of blockchain technology in providing decentralized solutions to various problems, smart contracts have become more popular to the point that billions of US Dollars are currently exchanged every day through such technology. Meanwhile, various vulnerabilities in smart contracts have been exploited by attackers to steal cryptocurrencies worth millions of dollars. The automatic detection of smart contract vulnerabilities therefore is an essential research problem. Existing solutions to this problem particularly rely on human experts to define features or different rules to detect vulnerabilities. However, this often causes many vulnerabilities to be ignored, and they are inefficient in detecting new vulnerabilities. In this study, to overcome such challenges, we propose a framework to automatically detect vulnerabilities in smart contracts on the blockchain. More specifically, first, we utilize novel feature vector generation techniques from bytecode of smart contract since the source code of smart contracts are rarely available in public. Next, the collected vectors are fed into our novel metric learning-based deep neural network(DNN) to get the detection result. We conduct comprehensive experiments on large-scale benchmarks, and the quantitative results demonstrate the effectiveness and efficiency of our approach. | ['Bhavani Thuraisingham', 'Latifur Khan', 'Zhouxiang Wu', 'Xiaodi Li', 'Sadaf MD Halim', 'Zhuoyi Wang', 'Chen Zhao', 'Feng Mi'] | 2023-01-20 | null | null | null | null | ['metric-learning', 'metric-learning', 'vulnerability-detection'] | ['computer-vision', 'methodology', 'miscellaneous'] | [-7.45597407e-02 -3.94769847e-01 -1.68306157e-01 -2.76746452e-01
-6.90271854e-01 -9.48139429e-01 6.19530976e-01 -4.59635854e-02
-2.37499207e-01 4.84825939e-01 3.13424379e-01 -9.12014306e-01
3.29684019e-01 -1.01036775e+00 -5.15989602e-01 -7.52286017e-01
9.86992493e-02 1.28722295e-01 1.89910099e-01 -2.63032079e-01
5.12555063e-01 6.83786124e-02 -9.05370712e-01 2.34559342e-01
8.30587149e-01 8.80591214e-01 -2.23144412e-01 1.45721793e-01
-2.68024743e-01 7.57705867e-01 -6.43586636e-01 -6.06835246e-01
6.20951176e-01 -2.62067080e-01 -4.80765730e-01 -7.15111673e-01
-8.09522048e-02 -9.07391965e-01 -6.35930657e-01 1.75366867e+00
2.89351881e-01 -5.86504698e-01 9.91172343e-02 -1.25434232e+00
-5.03718674e-01 1.34309363e+00 -7.18464196e-01 2.19318364e-02
2.02987511e-02 4.74615246e-01 1.21062255e+00 -4.49282289e-01
5.84357560e-01 1.15850329e+00 4.69681829e-01 4.26020712e-01
-8.67517650e-01 -1.16219103e+00 -2.65887856e-01 3.13704908e-01
-8.78646791e-01 -2.70643741e-01 1.08017981e+00 -2.59815156e-01
8.11611176e-01 5.25199808e-02 5.42066693e-01 1.15571690e+00
4.05676007e-01 8.55910659e-01 1.09103382e+00 5.73171750e-02
1.32923499e-01 -7.77320117e-02 1.61337867e-01 1.77336231e-01
7.44084060e-01 5.17106235e-01 2.03460976e-01 -5.85648179e-01
5.59083998e-01 3.96834850e-01 -1.51512176e-01 -2.65001923e-01
-1.35708606e+00 1.22047138e+00 2.96770900e-01 6.04624689e-01
-1.64987370e-01 4.12892789e-01 1.13732326e+00 7.28417993e-01
1.27215777e-02 4.81874466e-01 -6.70723677e-01 -5.44791460e-01
-5.54509163e-01 3.58080417e-01 9.39733326e-01 5.39324284e-01
6.87576771e-01 -1.70871720e-01 2.09596798e-01 1.60594955e-01
4.70818728e-01 4.47256655e-01 5.12206078e-01 -7.96879768e-01
9.82292414e-01 1.01043630e+00 1.34598082e-02 -1.28527200e+00
-9.40743685e-02 -3.95211697e-01 -8.35863113e-01 1.67109802e-01
3.75902414e-01 -2.33228132e-01 -5.34360588e-01 1.35455406e+00
3.46395373e-01 -5.73223121e-02 2.83489488e-02 7.66690910e-01
1.60981491e-01 6.83592319e-01 -2.79067963e-01 -2.65997589e-01
1.08670747e+00 -6.95560694e-01 -5.81173539e-01 3.06344569e-01
9.39302444e-01 -6.78051949e-01 7.82345831e-01 3.96017700e-01
-3.76389861e-01 1.47922430e-02 -1.05114472e+00 2.82316148e-01
-2.41083637e-01 -5.83743691e-01 8.36703241e-01 1.01220572e+00
-3.50152284e-01 6.39064789e-01 -8.27000320e-01 2.77738512e-01
6.40403271e-01 3.13791484e-01 -1.57913506e-01 1.18748374e-01
-1.46150899e+00 5.21455586e-01 6.71605766e-01 1.95037708e-01
-1.11403883e+00 -3.14213902e-01 -7.06741810e-01 3.24243426e-01
4.34391290e-01 -3.92161397e-04 1.24395192e+00 -8.41758251e-01
-1.20483649e+00 2.30471298e-01 4.84090865e-01 -4.70212430e-01
6.27513826e-01 -1.99187063e-02 -4.55224305e-01 -1.29299298e-01
1.21286176e-02 -5.86971827e-02 6.11940205e-01 -9.65224683e-01
-8.02457452e-01 -2.34655812e-01 3.97456914e-01 -4.84631449e-01
-4.76595402e-01 3.80774736e-01 1.89669386e-01 -7.73066998e-01
2.99781701e-03 -9.35819805e-01 -2.27762341e-01 -6.44347548e-01
-3.74656379e-01 -4.46747690e-01 1.12243605e+00 -8.21627915e-01
1.29944503e+00 -2.08813143e+00 -2.50635684e-01 5.33039033e-01
3.92894119e-01 6.09638155e-01 -2.09164172e-01 5.94729960e-01
7.22882971e-02 2.81853229e-01 -3.91916305e-01 4.28237796e-01
2.99027175e-01 -2.60878615e-02 -7.20632195e-01 4.80831921e-01
-3.77191454e-01 1.18726718e+00 -1.06189156e+00 -2.79594004e-01
1.63385272e-02 2.67076731e-01 -5.09690642e-01 1.75795123e-01
-4.35848862e-01 4.39304441e-01 -9.73551273e-01 9.46950138e-01
9.52190042e-01 -1.81541577e-01 4.95986611e-01 -2.69726124e-02
-6.19554147e-02 4.64545697e-01 -8.62224519e-01 1.50121510e+00
-2.38445491e-01 2.82647073e-01 -9.66410190e-02 -1.04492748e+00
8.68513346e-01 4.00626332e-01 5.51422894e-01 -7.83156216e-01
4.78784114e-01 5.76739490e-01 3.09020877e-01 -4.90979880e-01
6.75285235e-02 1.33157477e-01 -2.42413506e-01 1.14693272e+00
-6.62414789e-01 2.23307148e-01 9.90959108e-02 1.88793465e-01
1.48007476e+00 -5.63700870e-03 1.50942355e-01 7.37740025e-02
6.38922632e-01 -3.98098305e-02 1.07458520e+00 4.84983742e-01
-3.39698821e-01 -3.95005830e-02 9.08631444e-01 -1.07408655e+00
-1.12485611e+00 -5.80333114e-01 2.09873319e-01 5.93299747e-01
3.81475687e-02 -3.82975340e-01 -8.19420338e-01 -1.35773337e+00
2.62782991e-01 3.74323249e-01 -3.29981148e-01 -1.29905179e-01
-9.34874356e-01 -6.24721467e-01 6.10901296e-01 4.47909862e-01
8.23384583e-01 -1.34778190e+00 -6.37195826e-01 5.03712595e-01
-1.87899873e-01 -7.61185229e-01 -4.79170412e-01 -3.20148021e-01
-8.04002345e-01 -1.53783548e+00 -4.45019215e-01 -4.85802710e-01
6.57850146e-01 3.09308648e-01 6.51644647e-01 5.35103977e-01
-5.67660555e-02 -4.99318302e-01 -4.56619233e-01 5.98763376e-02
-7.01552033e-01 1.94611162e-01 -4.27932255e-02 -7.36460015e-02
7.20106721e-01 -7.07548380e-01 -8.31586540e-01 2.28303924e-01
-1.05212057e+00 -2.81629562e-01 8.48591149e-01 8.49715233e-01
-1.25152633e-01 2.01539174e-01 6.58372104e-01 -9.29548144e-01
8.84230077e-01 -5.09128153e-01 -1.24065483e+00 2.76930362e-01
-9.23699677e-01 1.15664110e-01 7.46371090e-01 -3.14769298e-01
-8.03205729e-01 -1.76284343e-01 -1.05252624e-01 -3.29248548e-01
2.59680390e-01 6.08159363e-01 -3.97846133e-01 -2.69722372e-01
1.57618150e-01 2.90552378e-01 -1.59599409e-01 -5.18277049e-01
3.35344851e-01 9.70295191e-01 9.46968198e-02 -5.11735916e-01
1.23016250e+00 3.37504715e-01 -3.01424503e-01 2.72366673e-01
-2.62395442e-01 -3.77894901e-02 -2.49373894e-02 2.40313962e-01
4.72326130e-01 -3.25929523e-01 -1.22701633e+00 7.85399318e-01
-1.41188061e+00 1.09048061e-01 3.28361005e-01 3.67461860e-01
-1.54959723e-01 8.03392828e-01 -7.80626178e-01 -5.13326883e-01
-7.88046479e-01 -1.62629950e+00 5.72592080e-01 1.46998882e-01
3.21869589e-02 -5.47879457e-01 6.20298684e-01 4.04858738e-01
6.30561531e-01 2.57348567e-01 1.20557570e+00 -8.33969951e-01
-8.80794942e-01 -3.14094514e-01 -4.98846531e-01 4.44705933e-01
5.17812967e-01 -2.79422700e-01 -6.43938243e-01 -6.71473265e-01
2.19136953e-01 -9.13998634e-02 7.11514235e-01 -2.27306381e-01
1.08313632e+00 -7.76210070e-01 -4.25098777e-01 6.69539154e-01
1.37760913e+00 5.74384809e-01 5.72030365e-01 7.57457137e-01
6.97139621e-01 4.24539268e-01 4.17563409e-01 3.94934475e-01
2.48962432e-01 4.08894271e-01 7.57416844e-01 3.98209900e-01
5.43043971e-01 -3.77547234e-01 3.46176744e-01 8.94032478e-01
-9.98667628e-02 3.06089759e-01 -1.00330853e+00 5.62292635e-01
-1.81035781e+00 -1.01014686e+00 1.57930210e-01 1.95877123e+00
8.99558127e-01 2.01257840e-01 -1.14138365e-01 2.23437652e-01
7.87603140e-01 2.06697568e-01 -6.09919846e-01 -2.35686481e-01
2.49192461e-01 6.17115945e-02 6.37977064e-01 6.02097027e-02
-1.12842858e+00 7.61940837e-01 5.23748732e+00 6.24561787e-01
-1.20246387e+00 2.54863858e-01 5.09243309e-01 1.05746716e-01
-8.39089751e-01 5.30321360e-01 -4.73966062e-01 1.04777181e+00
5.32243431e-01 -3.80085796e-01 7.01356649e-01 8.93770576e-01
-1.02722896e-02 4.62647617e-01 -8.60780001e-01 9.53690588e-01
-4.49757397e-01 -1.54779267e+00 -1.97033718e-01 2.80420482e-01
8.66353810e-01 1.32518038e-01 2.79398263e-03 4.23194975e-01
8.02422523e-01 -7.95845985e-01 4.11087573e-01 -1.70307502e-01
5.37129939e-01 -9.27015960e-01 1.13570547e+00 1.54327348e-01
-1.01149952e+00 -4.79904860e-01 -4.11272526e-01 1.23679079e-01
-3.11551522e-03 6.50463223e-01 -6.73842311e-01 5.61794519e-01
5.82340777e-01 4.65428442e-01 -2.01884910e-01 8.31051767e-01
-6.27330363e-01 6.77612424e-01 -1.32510349e-01 -2.59096473e-01
5.55111229e-01 -2.30820894e-01 3.56885642e-01 8.97389889e-01
4.51300383e-01 -1.44476533e-01 -9.67797637e-03 9.46229875e-01
-3.77490520e-01 -6.57588616e-02 -6.59330964e-01 -3.94395173e-01
6.18939102e-01 1.26504135e+00 -4.52371240e-01 -2.92838037e-01
-7.12760866e-01 4.57815319e-01 1.08584508e-01 2.99779534e-01
-7.96840131e-01 -8.68488669e-01 7.33068287e-01 -1.95140064e-01
3.84009272e-01 -2.27671504e-01 -1.61636218e-01 -1.43446660e+00
2.55422264e-01 -1.51266646e+00 3.14471811e-01 -4.08043899e-02
-1.09481871e+00 3.76851737e-01 -3.83986861e-01 -1.34363306e+00
-2.60256112e-01 -2.61847973e-01 -8.67581308e-01 5.52346528e-01
-1.57697880e+00 -1.04575479e+00 1.34691060e-01 4.04372215e-01
1.06753550e-01 -4.82181907e-01 6.03638113e-01 3.86480987e-01
-4.34505522e-01 6.37051642e-01 5.04659534e-01 7.93843150e-01
3.93415779e-01 -8.23787928e-01 1.08328664e+00 8.93154204e-01
1.07334800e-01 1.04181731e+00 5.06823540e-01 -9.40133810e-01
-1.88249266e+00 -8.50637913e-01 8.39341342e-01 -1.93377331e-01
1.04570949e+00 -4.69057351e-01 -8.85798097e-01 7.47406125e-01
2.19520628e-01 -9.28364545e-02 2.63094515e-01 -2.65879929e-01
-8.60045731e-01 -4.03029263e-01 -1.22480071e+00 3.53992432e-01
7.73499072e-01 -5.41510522e-01 -6.65897489e-01 1.03731826e-01
8.85488033e-01 -5.67244031e-02 -5.85831940e-01 4.03159589e-01
8.73759210e-01 -1.01651204e+00 4.43198860e-01 -6.07225478e-01
3.86606485e-01 -3.37062091e-01 1.72621652e-03 -1.13991976e+00
1.70852035e-01 -1.01693916e+00 -2.44499907e-01 1.21602011e+00
1.74183398e-01 -1.18239629e+00 9.75333393e-01 3.85853171e-01
2.60831624e-01 -5.61522901e-01 -1.17726493e+00 -7.31002986e-01
2.07077116e-01 -9.63797867e-02 1.40073872e+00 1.36428499e+00
3.19220155e-01 -4.29344743e-01 -2.51247883e-01 -1.09882981e-01
7.66613066e-01 6.41326308e-01 8.65120709e-01 -1.02550864e+00
-2.62494951e-01 -4.72873509e-01 -5.35024226e-01 -8.08253229e-01
1.00930087e-01 -7.73941815e-01 -2.87834853e-01 -1.14989471e+00
4.00179476e-01 -6.46372616e-01 -7.21824110e-01 6.64002001e-01
6.31562248e-02 -2.06012085e-01 1.15297459e-01 4.59443182e-01
-2.29074284e-01 4.66316313e-01 1.06311333e+00 -7.01020300e-01
1.09756120e-01 -2.07284149e-02 -7.66618192e-01 5.38864791e-01
8.23864162e-01 -8.33093822e-01 -6.24800995e-02 -6.64988995e-01
7.44406521e-01 5.58103137e-02 1.01119921e-01 -5.48454463e-01
1.07457809e-01 -3.96153301e-01 -2.46152043e-01 -5.64726472e-01
-4.97069150e-01 -9.40698326e-01 1.32846966e-01 9.50227916e-01
6.45803884e-02 2.28307515e-01 -3.97833228e-01 4.39838946e-01
-3.69180053e-01 -1.80938423e-01 3.15420955e-01 -2.52140254e-01
-4.72354770e-01 4.06160980e-01 -6.43452723e-03 -2.75478870e-01
1.04500771e+00 1.61668941e-01 -7.09929585e-01 -1.79636598e-01
8.34134817e-02 4.26786602e-01 6.88345253e-01 6.86939478e-01
5.11263311e-01 -1.49128616e+00 -7.59127855e-01 3.69524479e-01
-1.66759081e-02 -1.86915010e-01 -5.68975769e-02 7.20650136e-01
-8.41355741e-01 5.46706736e-01 -2.48892784e-01 -1.45697385e-01
-1.08760107e+00 7.84235358e-01 1.49684444e-01 -4.90719169e-01
-6.52750134e-01 2.82859564e-01 5.61434291e-02 -5.38037598e-01
1.18279971e-01 -3.58476490e-01 -8.28799307e-02 -2.06050113e-01
6.71991527e-01 3.23145688e-01 -1.65004879e-01 -3.18392307e-01
-3.43637705e-01 4.59413022e-01 -4.68522370e-01 5.66460341e-02
1.48138034e+00 4.46540326e-01 -5.74128509e-01 -3.75738919e-01
1.27191436e+00 1.71693936e-01 -9.78247583e-01 -1.96014658e-01
3.24915379e-01 -7.79352248e-01 -2.13406324e-01 -5.46620429e-01
-1.49775016e+00 9.93260622e-01 4.85834897e-01 3.77878636e-01
6.82206392e-01 -2.37291768e-01 1.52269685e+00 7.78746486e-01
9.03842330e-01 -9.17368412e-01 -3.31368446e-01 3.95012796e-01
3.90527040e-01 -1.07085288e+00 -3.12831849e-01 6.30422384e-02
-2.15558663e-01 1.09967518e+00 2.05697849e-01 -8.96808803e-02
4.41304117e-01 2.45247826e-01 3.00181627e-01 -1.06713757e-01
-4.01168019e-01 5.06175160e-01 -5.69419324e-01 3.92610371e-01
8.68316591e-02 2.19042793e-01 -6.71550810e-01 5.47094882e-01
1.82539951e-02 5.48580103e-03 4.80022281e-01 1.03095806e+00
-4.38782752e-01 -1.82121825e+00 -3.35365385e-01 3.45067620e-01
-7.98241913e-01 -1.44382164e-01 -1.53829545e-01 4.45030183e-01
-7.25226924e-02 9.13475275e-01 -4.21527177e-01 -6.66071236e-01
-3.37008722e-02 -2.44101346e-01 2.78606210e-02 -2.11638883e-01
-7.64680624e-01 -1.77758843e-01 -1.33478835e-01 -6.47318959e-01
-9.30882394e-02 -5.45000553e-01 -1.15813923e+00 -6.43885612e-01
-4.45315778e-01 5.37503123e-01 7.54080117e-01 7.80942619e-01
1.51106551e-01 -1.05804525e-01 1.29765666e+00 -1.69853300e-01
-1.30060673e+00 -9.57925379e-01 -4.10053968e-01 5.71327090e-01
2.95264781e-01 -4.26171303e-01 -4.98696715e-01 -4.05015171e-01] | [6.800260543823242, 7.26087760925293] |
0aaeca70-5d6d-4ef7-9fe0-d281983e8cd0 | comparison-of-forecasting-methods-of-house | 2208.07217 | null | https://arxiv.org/abs/2208.07217v1 | https://arxiv.org/pdf/2208.07217v1.pdf | Comparison of Forecasting Methods of House Electricity Consumption for Honda Smart Home | The electricity consumption of buildings composes a major part of the city's energy consumption. Electricity consumption forecasting enables the development of home energy management systems resulting in the future design of more sustainable houses and a decrease in total energy consumption. Energy performance in buildings is influenced by many factors like ambient temperature, humidity, and a variety of electrical devices. Therefore, multivariate prediction methods are preferred rather than univariate. The Honda Smart Home US data set was selected to compare three methods for minimizing forecasting errors, MAE and RMSE: Artificial Neural Networks, Support Vector Regression, and Fuzzy Rule-Based Systems for Regression by constructing many models for each method on a multivariate data set in different time terms. The comparison shows that SVR is a superior method over the alternatives. | ['Mehmet Bodur', 'Farshad Ahmadi Asl'] | 2022-08-11 | null | null | null | null | ['total-energy', 'energy-management'] | ['miscellaneous', 'time-series'] | [-3.73391032e-01 -4.52630490e-01 -1.78249732e-01 -4.55862194e-01
-1.23198241e-01 -3.52292508e-02 3.70726377e-01 7.15958849e-02
-3.28338295e-02 9.56536114e-01 3.51008624e-01 -4.23731059e-01
-2.39012331e-01 -1.27482820e+00 3.42842400e-01 -8.76090646e-01
2.64497697e-01 1.21222638e-01 -3.34544003e-01 -3.91260147e-01
1.86062798e-01 4.87945139e-01 -1.55875039e+00 -4.60619301e-01
1.02215087e+00 1.23150289e+00 1.57855690e-01 2.08683431e-01
3.99970651e-01 7.10851371e-01 -2.78828353e-01 2.32805848e-01
5.43524139e-02 -4.06329542e-01 -2.13900208e-01 -1.49066761e-01
-8.82080257e-01 -3.32853317e-01 2.81155258e-01 5.76145887e-01
4.63842541e-01 4.44217443e-01 8.84937227e-01 -1.29194140e+00
-4.82851535e-01 4.61814672e-01 -1.23463692e-02 2.29972564e-02
2.83374041e-01 -2.06148356e-01 5.79837680e-01 -5.73415160e-01
-2.66769886e-01 5.79062879e-01 4.53286678e-01 -1.05833314e-01
-1.20688295e+00 -3.95883292e-01 -2.82769322e-01 5.71315646e-01
-1.62432921e+00 -3.96313995e-01 8.90169382e-01 -4.26136583e-01
1.54374552e+00 7.09591269e-01 9.41932917e-01 3.65857214e-01
3.09047967e-01 -5.52344546e-02 1.35746729e+00 -5.70253909e-01
6.14482522e-01 6.33295536e-01 -3.60103101e-02 2.42588729e-01
4.61344540e-01 3.58463973e-01 4.77212757e-01 -2.16660917e-01
2.37240016e-01 4.45685476e-01 -2.58332700e-01 1.54799104e-01
-7.04340756e-01 9.73553777e-01 3.70896637e-01 7.43488133e-01
-5.99519730e-01 -1.03504263e-01 -1.94677971e-02 1.34752318e-01
2.18004018e-01 2.04035759e-01 -6.40322089e-01 4.54371758e-02
-9.35049295e-01 -1.35940149e-01 8.49082410e-01 6.29186034e-01
5.70223451e-01 5.25827825e-01 1.98600039e-01 6.90463364e-01
6.52033031e-01 9.24106121e-01 6.70090079e-01 -9.87895250e-01
-8.69507063e-03 7.23459840e-01 2.45421171e-01 -1.08491945e+00
-5.43783367e-01 -2.32770354e-01 -1.29522479e+00 1.55946314e-01
-7.12186098e-02 -1.97866008e-01 -4.52666938e-01 1.13713062e+00
5.25402650e-03 -4.42805707e-01 -7.56455958e-02 3.91178906e-01
3.20531309e-01 9.05786753e-01 1.54776409e-01 -8.57985973e-01
9.37537789e-01 -5.53404927e-01 -8.48642945e-01 3.20144475e-01
2.44391352e-01 -3.48773390e-01 2.65661687e-01 1.49993777e-01
-7.85202026e-01 -4.10352379e-01 -8.99932265e-01 4.06310797e-01
-9.72181082e-01 1.24231771e-01 3.41554970e-01 8.04073751e-01
-8.27480376e-01 6.79821789e-01 -8.49881530e-01 -4.18913931e-01
3.80089805e-02 4.51578707e-01 3.24486911e-01 5.92810750e-01
-1.07506883e+00 1.50060952e+00 4.26802784e-01 1.59480691e-01
2.29789883e-01 -3.67826313e-01 -7.68166840e-01 2.25444496e-01
-3.64196569e-01 -6.20998204e-01 9.67007995e-01 -4.26990896e-01
-1.39782190e+00 1.02100195e-02 -3.12236935e-01 -3.73169154e-01
2.59053707e-01 3.29752326e-01 -9.74506497e-01 -4.83008623e-01
-1.12648919e-01 -1.61232382e-01 3.80870581e-01 -8.67869794e-01
-8.37038577e-01 -4.80303526e-01 -7.54871786e-01 -2.19300047e-01
-5.57944700e-02 -2.52581775e-01 6.75152659e-01 -3.41439694e-01
8.65493864e-02 -8.12247157e-01 -4.26769644e-01 -8.89196515e-01
-1.10081226e-01 -5.24961531e-01 6.99272990e-01 -1.19208694e+00
1.71673214e+00 -1.39871776e+00 -3.18755865e-01 8.10169160e-01
-3.52374375e-01 -2.28055179e-01 8.25608492e-01 4.32332158e-01
-1.28095865e-01 2.42715642e-01 -1.95915580e-01 2.73828417e-01
1.92260921e-01 2.18586802e-01 2.84439269e-02 2.05519393e-01
-1.38670936e-01 5.87410033e-01 -7.12611914e-01 -1.47731498e-01
9.28111911e-01 8.18723142e-01 1.54853716e-01 4.19316813e-02
2.75667161e-01 1.92060918e-01 -5.68103135e-01 8.16242874e-01
2.34770820e-01 -2.02059507e-01 2.18842059e-01 -3.18942040e-01
-6.08425021e-01 2.54648566e-01 -1.14991987e+00 7.11294591e-01
-6.68451905e-01 4.30729538e-01 -6.11719608e-01 -9.92690325e-01
1.20795262e+00 5.84862769e-01 8.28423619e-01 -1.03701937e+00
3.33402753e-01 4.53467101e-01 -2.36747861e-01 -5.49420118e-01
2.83030927e-01 -1.71403959e-02 1.89770654e-01 1.30092934e-01
-5.67521751e-01 -2.78511588e-02 2.93857068e-01 -5.05250454e-01
6.63844049e-01 -1.42461658e-01 8.07963789e-01 -5.25576651e-01
5.89307964e-01 -2.90796459e-01 6.48001671e-01 -2.79182136e-01
-1.51101038e-01 -6.43770024e-02 -2.33940482e-01 -5.12167513e-01
-1.16038382e+00 -6.74033344e-01 -5.14630914e-01 5.86840808e-01
-2.73732990e-01 2.89877527e-04 -3.26499760e-01 1.97269116e-02
7.80274868e-02 1.43245733e+00 -2.03130811e-01 2.40534544e-02
-4.58703816e-01 -1.04327881e+00 -4.15944725e-01 5.64963222e-01
7.23883808e-01 -6.73830867e-01 -1.14128590e+00 4.56186950e-01
-3.38869959e-01 -6.54277802e-01 7.23291561e-02 2.76861519e-01
-8.24516118e-01 -8.66617560e-01 -3.91789883e-01 -4.70962673e-01
6.31044388e-01 3.79336327e-02 1.29898822e+00 1.74511015e-01
-3.38213108e-02 1.37529582e-01 -1.50692493e-01 -3.47189665e-01
-2.00620279e-01 4.92294841e-02 3.06451470e-01 -5.14864981e-01
5.83927274e-01 -8.97320271e-01 -8.13736200e-01 3.30020875e-01
-1.40398085e-01 1.82409324e-02 2.05108136e-01 4.52028483e-01
4.53180730e-01 1.04633987e+00 8.28550100e-01 -3.93232584e-01
4.89408165e-01 -8.07278335e-01 -8.78670394e-01 4.32679743e-01
-1.57471704e+00 -5.00355288e-02 6.86616600e-01 1.34784341e-01
-1.10143209e+00 8.54173899e-02 4.68075983e-02 4.08225387e-01
-1.12586290e-01 2.14756444e-01 -2.77248502e-01 1.94188863e-01
2.50549912e-02 1.84170052e-01 -2.75436550e-01 -5.53772986e-01
-2.67104536e-01 6.51866913e-01 2.92887390e-01 1.30010974e-02
1.01306856e+00 -1.94915995e-01 2.61396587e-01 -7.53610551e-01
1.06481567e-01 -3.01178932e-01 -5.50361097e-01 -4.64739054e-01
8.55938017e-01 -8.68918896e-01 -8.85776162e-01 2.33140483e-01
-7.30889499e-01 -1.81932166e-01 2.74099037e-02 4.90691394e-01
-1.85962439e-01 -3.71360630e-01 -4.74452339e-02 -1.26254439e+00
-5.93912601e-01 -9.74030852e-01 4.94407080e-02 5.06169379e-01
-6.07629359e-01 -1.15462518e+00 -1.69778410e-02 1.75390661e-01
8.79170597e-01 8.66552293e-01 1.14031351e+00 -2.32365280e-01
-2.95207649e-01 -3.36628079e-01 1.42003283e-01 4.82171863e-01
6.31167293e-01 4.92982268e-01 -6.35266840e-01 -1.51383430e-01
2.40801126e-01 4.47862685e-01 2.33874649e-01 5.86581409e-01
9.69838142e-01 -8.02369356e-01 -3.17110687e-01 1.47261366e-01
2.03434253e+00 1.04469454e+00 7.90640056e-01 4.91995007e-01
2.39228413e-01 4.30829018e-01 1.85275391e-01 7.92400658e-01
7.78351605e-01 3.83580983e-01 1.79252729e-01 9.51330811e-02
6.95874929e-01 6.59038574e-02 2.05677390e-01 9.67656195e-01
-7.02839255e-01 -1.72409024e-02 -8.18738341e-01 4.38581467e-01
-1.75492561e+00 -1.38070774e+00 -3.93837005e-01 2.11713099e+00
3.33284140e-01 -2.88028240e-01 4.18405920e-01 8.34933221e-01
3.23664486e-01 -1.66757599e-01 -4.29021418e-01 -4.89899278e-01
7.61902183e-02 1.71258971e-01 7.70880699e-01 1.88933864e-01
-7.26033509e-01 -2.26210818e-01 6.61736679e+00 6.18166253e-02
-9.89870012e-01 -1.48804083e-01 8.46528947e-01 2.64232960e-02
-4.33538169e-01 -2.47215122e-01 -5.00247359e-01 9.84217644e-01
1.34796810e+00 -5.66163898e-01 8.02172720e-01 1.02441728e+00
8.07122767e-01 -5.08022308e-01 -6.65673256e-01 9.08714652e-01
-5.26978970e-01 -9.26497400e-01 -5.86811483e-01 2.76190639e-01
1.02814043e+00 -2.48365790e-01 -3.84178519e-01 2.05182016e-01
3.26686531e-01 -8.08594882e-01 3.30402613e-01 9.72257078e-01
2.90473521e-01 -9.34064567e-01 1.01212621e+00 2.74344891e-01
-1.56070995e+00 -2.96737075e-01 -1.93296112e-02 -2.37525761e-01
2.88784325e-01 6.84874833e-01 -4.22276288e-01 5.06386697e-01
9.88564014e-01 4.00296241e-01 -3.77627701e-01 6.53664768e-01
-9.80581120e-02 6.27029240e-01 -6.23724341e-01 -2.67695040e-01
-4.52348173e-01 -7.76775360e-01 -1.41848490e-01 6.39454007e-01
7.79599249e-01 6.42008841e-01 -2.15529457e-01 7.95222998e-01
4.57766294e-01 1.02134727e-01 -4.12990510e-01 2.35499576e-01
8.65748823e-01 1.06960738e+00 -4.14808482e-01 -2.97258317e-01
-5.51187694e-01 4.23463583e-01 -3.48470986e-01 4.24898267e-01
-7.50575662e-01 -2.91935951e-01 5.96248090e-01 2.40293294e-01
1.57488242e-01 -1.56731352e-01 -7.60760725e-01 -7.32275367e-01
-2.76338309e-01 -2.92147726e-01 1.83837429e-01 -6.41172945e-01
-9.57294583e-01 3.42396319e-01 5.62915690e-02 -8.76179695e-01
-7.84508944e-01 -2.99757898e-01 -9.17011917e-01 1.09086227e+00
-1.45890141e+00 -5.77807009e-01 -4.78490472e-01 5.11756241e-01
3.19593340e-01 -2.44144812e-01 1.30576468e+00 4.32088643e-01
-1.03032613e+00 -5.05487509e-02 6.89454257e-01 -2.28977278e-01
-1.43967867e-01 -1.31798732e+00 -1.60822809e-01 6.50093019e-01
-7.13962197e-01 1.66214223e-03 7.40287721e-01 -5.05327463e-01
-1.02149284e+00 -8.92350972e-01 1.31606686e+00 1.33524733e-02
2.73590446e-01 3.22408259e-01 -5.58054984e-01 4.90291774e-01
3.51544917e-01 -5.68326712e-01 9.05944645e-01 -1.80119023e-01
4.47860628e-01 -6.45499825e-01 -1.68662298e+00 2.26719946e-01
2.08569854e-01 -1.90294176e-01 -2.63378531e-01 -1.21311128e-01
1.60261035e-01 5.00679195e-01 -1.63817060e+00 6.10860825e-01
8.79352510e-01 -1.08169067e+00 8.04587483e-01 2.40927413e-01
7.88678080e-02 -3.61125112e-01 -5.04284680e-01 -1.32750595e+00
-8.05194259e-01 -2.66138941e-01 -3.32656562e-01 1.54279673e+00
3.75518799e-01 -1.09792268e+00 3.98114324e-01 1.50550938e+00
1.97240949e-01 -7.77240574e-01 -8.52812290e-01 -5.67082167e-01
-3.11112165e-01 -1.79300681e-01 1.39193976e+00 9.01457310e-01
3.87068698e-03 2.51615137e-01 -1.08305514e-02 -4.46321033e-02
8.89775574e-01 -6.11628108e-02 3.28382075e-01 -1.46590436e+00
3.28895420e-01 -6.53889656e-01 -1.84596509e-01 -2.00153533e-02
-1.39432579e-01 -3.14075649e-01 -1.19417630e-01 -1.89187396e+00
-1.01674832e-01 -3.34987879e-01 -4.31949794e-01 5.21366715e-01
2.11407661e-01 -1.21737398e-01 3.62750236e-03 2.03077435e-01
5.45848906e-01 7.80573070e-01 4.79694486e-01 -1.53193623e-01
-4.80569601e-01 6.03525758e-01 -4.37411696e-01 7.45609701e-01
1.28199863e+00 -7.39562139e-02 -4.56194490e-01 7.75426775e-02
1.50733501e-01 1.07326455e-01 1.60057284e-02 -1.21637821e+00
-6.37529120e-02 -6.94824815e-01 7.86325812e-01 -9.76365149e-01
-1.22521572e-01 -1.67098951e+00 1.22940648e+00 7.37883925e-01
2.95238018e-01 3.00891966e-01 -3.75571638e-01 -2.44160891e-02
4.41120453e-02 1.31011590e-01 6.80488110e-01 1.93983153e-01
-6.38540268e-01 -7.41827860e-02 -6.28439188e-01 -9.54544365e-01
1.11022007e+00 -5.26664913e-01 -1.49033651e-01 -2.00808987e-01
-6.88023925e-01 4.06376421e-01 3.82702410e-01 1.38957784e-01
2.49014929e-01 -1.71111500e+00 -4.06490833e-01 2.14130446e-01
-4.45160508e-01 -2.67827481e-01 -5.26170023e-02 6.88851118e-01
-3.70112598e-01 5.93997896e-01 -2.38321692e-01 -1.10259429e-01
-9.80905890e-01 2.54543990e-01 5.17954886e-01 -3.01212490e-01
-6.44964725e-02 -5.98923713e-02 -7.50169992e-01 7.38666728e-02
-5.41839600e-02 -5.73693573e-01 -5.33275545e-01 1.95500761e-01
3.36286306e-01 1.44428217e+00 8.93615559e-02 -8.08718145e-01
-4.75434452e-01 5.81429183e-01 8.19804966e-01 3.23527664e-01
1.61413646e+00 -5.45542181e-01 -3.55310619e-01 8.15534949e-01
1.21827400e+00 -5.73052943e-01 -5.26026785e-01 1.73057765e-01
2.51063794e-01 -1.21895090e-01 4.12717760e-01 -1.01212490e+00
-1.04345620e+00 7.67742889e-03 9.78866696e-01 7.75678217e-01
1.84869528e+00 -6.13609433e-01 9.99776244e-01 3.48746121e-01
5.73307574e-01 -1.58018577e+00 -7.28281677e-01 -2.35380251e-02
7.97874749e-01 -1.48966873e+00 3.41656864e-01 -1.03835121e-01
-3.42691720e-01 1.20880377e+00 1.48170307e-01 3.66454874e-03
1.34378517e+00 3.21820416e-02 -3.57964128e-01 3.67282122e-01
-5.26612520e-01 -2.04901814e-01 1.37317270e-01 4.63523656e-01
5.50231755e-01 6.86833739e-01 -5.08701324e-01 7.36746788e-01
-5.22132277e-01 1.29775330e-01 1.84492636e-02 7.57269323e-01
-5.03660738e-01 -7.87656844e-01 -5.08308709e-01 7.78106213e-01
-1.81784749e-01 1.97820559e-01 3.42528045e-01 7.15187848e-01
3.06794673e-01 1.60266113e+00 1.75480798e-01 -4.72603410e-01
6.22322619e-01 1.80186898e-01 5.35901301e-02 1.44961625e-01
-4.03936177e-01 -3.38384099e-02 4.26578633e-02 -1.87139526e-01
-7.37253428e-01 -7.69134343e-01 -1.31435442e+00 -1.09049809e+00
-4.38460588e-01 2.89440036e-01 1.06884003e+00 1.09514093e+00
-8.12440366e-02 5.84700882e-01 1.54251564e+00 -7.98334181e-01
-3.50376606e-01 -1.02230871e+00 -9.47336555e-01 1.51268050e-01
8.45147222e-02 -4.49758559e-01 -6.23073339e-01 1.13446683e-01] | [5.988003730773926, 2.646315813064575] |
26ada930-9fc7-4e88-b7bf-6c61d26c96a1 | accented-speech-recognition-inspired-by-human | 2104.04627 | null | https://arxiv.org/abs/2104.04627v1 | https://arxiv.org/pdf/2104.04627v1.pdf | Accented Speech Recognition Inspired by Human Perception | While improvements have been made in automatic speech recognition performance over the last several years, machines continue to have significantly lower performance on accented speech than humans. In addition, the most significant improvements on accented speech primarily arise by overwhelming the problem with hundreds or even thousands of hours of data. Humans typically require much less data to adapt to a new accent. This paper explores methods that are inspired by human perception to evaluate possible performance improvements for recognition of accented speech, with a specific focus on recognizing speech with a novel accent relative to that of the training data. Our experiments are run on small, accessible datasets that are available to the research community. We explore four methodologies: pre-exposure to multiple accents, grapheme and phoneme-based pronunciations, dropout (to improve generalization to a novel accent), and the identification of the layers in the neural network that can specifically be associated with accent modeling. Our results indicate that methods based on human perception are promising in reducing WER and understanding how accented speech is modeled in neural networks for novel accents. | ['Michael Picheny', 'Amber Wang', 'Elizabeth Combs', 'Xiangyun Chu'] | 2021-04-09 | null | null | null | null | ['accented-speech-recognition'] | ['speech'] | [ 3.88582885e-01 3.11552644e-01 1.72758549e-01 -9.81156409e-01
-5.76724172e-01 -7.19769001e-01 3.27095836e-01 3.67503933e-04
-6.51024818e-01 6.48771942e-01 4.31530446e-01 -3.87641728e-01
2.75509387e-01 -3.07031810e-01 -6.09431207e-01 -4.95416015e-01
1.28529951e-01 7.02839673e-01 -6.58443719e-02 -4.20228601e-01
3.39020900e-02 6.03508651e-01 -1.63956368e+00 2.60003179e-01
7.44289815e-01 4.78535980e-01 4.50181395e-01 7.42509127e-01
-6.89826766e-03 5.55920422e-01 -1.17964017e+00 -2.24996731e-01
-5.17929113e-03 -4.19679224e-01 -9.86788213e-01 1.89957798e-01
7.56993413e-01 8.57579857e-02 4.46297564e-02 1.11441100e+00
6.68128133e-01 5.08106887e-01 2.56635636e-01 -5.59408724e-01
-6.55683219e-01 1.04111230e+00 6.95014074e-02 5.12713969e-01
2.17867851e-01 1.84046149e-01 6.36661828e-01 -8.56010854e-01
2.97773063e-01 1.43950450e+00 5.17964184e-01 7.59720504e-01
-1.19330657e+00 -5.36320865e-01 3.56959999e-01 1.34940356e-01
-1.39136851e+00 -9.18319941e-01 7.10437059e-01 -1.31220922e-01
1.42298722e+00 5.90286613e-01 2.80579507e-01 1.05898774e+00
-1.24309339e-01 6.47446513e-01 1.20938063e+00 -6.17406964e-01
2.97688961e-01 3.17638874e-01 2.51196444e-01 2.02063054e-01
-3.89757872e-01 3.10180873e-01 -7.16275811e-01 2.71445423e-01
3.47211391e-01 -7.32020617e-01 -2.77285755e-01 1.93586320e-01
-1.08156407e+00 6.18282855e-01 3.33451807e-01 3.50714773e-01
-5.66115975e-01 -4.42509234e-01 3.75698775e-01 3.63481969e-01
6.10016704e-01 8.26055288e-01 -9.37384486e-01 -3.50851923e-01
-9.00494814e-01 7.92048573e-02 9.55581427e-01 8.40500236e-01
7.25329757e-01 5.67938864e-01 1.89592153e-01 1.25310218e+00
4.36911732e-02 4.31655586e-01 7.01245129e-01 -7.12580979e-01
3.85607064e-01 1.41568884e-01 -2.66294360e-01 -3.44382733e-01
-4.52353895e-01 -5.79305172e-01 -5.74230134e-01 2.76696682e-01
4.98179197e-01 -5.11160851e-01 -1.33307755e+00 1.89514506e+00
1.29288897e-01 -7.12839589e-02 3.18603635e-01 6.94027066e-01
3.77503216e-01 7.70084143e-01 3.47072005e-01 -2.51072049e-01
1.23957801e+00 -1.05558383e+00 -8.12460303e-01 -7.59141505e-01
5.27624607e-01 -1.06619585e+00 1.29503953e+00 4.78278369e-01
-1.18869603e+00 -9.16498721e-01 -1.02286041e+00 1.66351214e-01
-8.28340232e-01 -4.25771952e-01 2.62010187e-01 1.08112931e+00
-1.29419374e+00 2.36794695e-01 -6.18111253e-01 -3.92606229e-01
-3.01125422e-02 5.85333228e-01 -1.01396888e-01 1.77603006e-01
-1.39806080e+00 1.30421281e+00 6.94354653e-01 2.36236781e-01
-4.34916645e-01 -8.96491349e-01 -9.31853533e-01 1.38082147e-01
-5.83617054e-02 -3.28959197e-01 1.54217374e+00 -1.63617289e+00
-1.63334334e+00 7.27502704e-01 -2.59824753e-01 -5.77235878e-01
-8.42113048e-02 -9.78701487e-02 -7.85352767e-01 -4.01196659e-01
-3.71119082e-01 8.34330022e-01 7.06523478e-01 -1.00779068e+00
-8.25374186e-01 -4.47767824e-01 -2.16579616e-01 4.82012868e-01
-2.65738457e-01 3.78654510e-01 1.42551079e-01 -8.27136815e-01
-5.76948971e-02 -1.04287267e+00 -8.93228874e-02 -1.08618605e+00
-3.88055384e-01 -5.54495811e-01 6.61512733e-01 -9.09711778e-01
9.88657951e-01 -2.30793118e+00 1.15895845e-01 1.35371536e-01
-3.38742137e-01 7.00275242e-01 -1.46762714e-01 1.17165484e-01
-3.50051701e-01 1.88209459e-01 -6.65064827e-02 -2.91472346e-01
1.39759481e-02 3.07422698e-01 -3.37642699e-01 -6.19238131e-02
2.59482771e-01 3.96852106e-01 -7.26240158e-01 4.25506048e-02
2.60248959e-01 5.87533832e-01 -3.26506585e-01 3.78417611e-01
-8.69992077e-02 4.73622382e-01 2.10246563e-01 3.11043769e-01
7.76685297e-01 5.40559113e-01 2.87013292e-01 3.41385454e-02
-2.87210584e-01 8.84795666e-01 -1.02055311e+00 1.30941927e+00
-6.24940515e-01 5.76974034e-01 3.73739541e-01 -7.93899477e-01
1.09336746e+00 5.98560989e-01 -3.41762960e-01 -5.51891685e-01
-6.54200464e-02 2.33954415e-01 5.09282053e-01 -1.25617370e-01
5.59567571e-01 -4.42457765e-01 4.98906747e-02 5.94952144e-02
2.16731444e-01 -1.93669051e-01 1.33178487e-01 -4.28239822e-01
6.60429776e-01 -4.24824834e-01 2.00426325e-01 -5.12250781e-01
4.02550280e-01 -1.26068935e-01 6.13683224e-01 8.76767457e-01
-3.92945796e-01 4.97357607e-01 -6.88759312e-02 -5.01926363e-01
-9.52355742e-01 -1.21308649e+00 -1.15485825e-01 1.59063840e+00
-5.08326471e-01 2.58182213e-02 -1.24121463e+00 -4.21630412e-01
-3.90954316e-01 1.24207306e+00 -2.51289248e-01 -1.00219540e-01
-9.52809930e-01 -6.65359616e-01 6.87328517e-01 7.74000585e-01
3.84115338e-01 -1.60108888e+00 -1.07058719e-01 3.00317317e-01
6.08279370e-02 -9.60541427e-01 -6.65685058e-01 6.56700313e-01
-5.75698555e-01 -2.97327965e-01 -5.95184445e-01 -1.26068199e+00
5.98724484e-01 -2.39830300e-01 1.19086421e+00 -7.20102638e-02
1.04777284e-01 2.22522333e-01 -2.93194354e-01 -7.21024215e-01
-9.57991362e-01 7.17831731e-01 3.89696240e-01 -4.17330638e-02
8.05737734e-01 -3.98499399e-01 -3.65133956e-02 1.98411137e-01
-7.19777584e-01 -1.52117178e-01 7.56814241e-01 7.70546019e-01
3.54712486e-01 2.68933438e-02 9.06553209e-01 -1.01252091e+00
6.27771080e-01 -3.13511699e-01 -4.16033566e-01 6.79194853e-02
-5.87502420e-01 1.09666828e-02 8.05875719e-01 -4.73367721e-01
-1.31876922e+00 3.62934262e-01 -8.12283754e-01 -1.15022667e-01
-1.01989794e+00 6.17527306e-01 -4.17348683e-01 2.05384016e-01
7.39815593e-01 8.21391791e-02 1.48884831e-02 -5.92349112e-01
3.58545691e-01 9.03013706e-01 6.20211840e-01 -3.19791287e-01
5.60915112e-01 -4.79983240e-01 -7.30104327e-01 -1.38016963e+00
-8.94487441e-01 -3.05352777e-01 -7.83326447e-01 -4.78919707e-02
8.09931636e-01 -6.56222701e-01 -3.59885216e-01 8.19872200e-01
-1.13357162e+00 -3.88194323e-01 -3.14463943e-01 5.54041326e-01
-2.72640854e-01 2.99247839e-02 -4.99419391e-01 -8.50230098e-01
-2.11272418e-01 -1.32050073e+00 5.16627371e-01 5.42100906e-01
-6.54972732e-01 -1.15082812e+00 9.77963358e-02 5.42913198e-01
7.51335561e-01 -4.87983674e-01 9.53468084e-01 -1.16670728e+00
-2.53824741e-01 4.11672443e-02 4.70826805e-01 9.83189642e-01
3.36351693e-01 -1.74347237e-01 -1.47015858e+00 -2.95312852e-01
1.15087703e-01 -3.58758867e-01 5.90949833e-01 4.32489872e-01
9.16919351e-01 -3.78025740e-01 3.67159434e-02 4.03686106e-01
6.21840537e-01 6.66235805e-01 6.71504319e-01 3.33054721e-01
4.97763216e-01 8.45749140e-01 3.19293648e-01 -3.24892133e-01
2.56960690e-01 6.00998819e-01 -6.43944219e-02 -4.03279871e-01
-3.81554812e-01 -1.22114211e-01 5.20291805e-01 1.34563816e+00
5.47508411e-02 -3.18153828e-01 -9.86061513e-01 7.73703277e-01
-1.08461905e+00 -7.87030578e-01 2.84161508e-01 2.28134704e+00
1.16937768e+00 5.10417223e-01 1.30776763e-01 2.67154574e-02
1.04518509e+00 1.76948637e-01 -4.43583429e-01 -1.09385550e+00
-3.83272082e-01 6.76465929e-01 3.04074794e-01 1.10179818e+00
-1.16176212e+00 1.31249523e+00 7.37705755e+00 4.79447186e-01
-1.22140419e+00 -2.89758831e-01 8.54970574e-01 1.94116622e-01
8.63775015e-02 -1.62781164e-01 -1.01408219e+00 2.75290132e-01
1.59654140e+00 -9.61577985e-03 7.15824544e-01 8.68037760e-01
9.68915448e-02 2.43040115e-01 -1.12281382e+00 5.21632731e-01
1.07590221e-01 -1.01412427e+00 1.32818475e-01 -2.44102865e-01
6.17624819e-01 2.66807616e-01 3.64066333e-01 3.41816813e-01
1.90739870e-01 -1.20852304e+00 6.72301888e-01 2.70675510e-01
3.32001686e-01 -1.01864183e+00 8.80258620e-01 1.22260891e-01
-6.78252101e-01 3.27996910e-01 -2.52838194e-01 -2.64046282e-01
-1.33337691e-01 1.93128437e-01 -1.55957234e+00 -6.13408573e-02
5.83893239e-01 2.27138377e-03 -6.75239682e-01 7.22441196e-01
-2.68015951e-01 1.19692945e+00 -3.55054647e-01 -2.35691726e-01
8.06253776e-02 8.56436715e-02 5.21442711e-01 1.55911410e+00
-6.50826022e-02 -1.30036578e-01 1.91843599e-01 4.61378694e-01
4.22717482e-02 1.22566126e-01 -4.31496620e-01 -1.13377683e-01
7.04216421e-01 8.87883544e-01 -4.22408283e-01 -3.24010760e-01
-4.19361502e-01 8.68392527e-01 5.30487359e-01 5.76493442e-01
-2.86817938e-01 -6.02232635e-01 9.70316470e-01 -1.37443200e-01
3.67102116e-01 -2.02626854e-01 -3.44805717e-01 -5.33813894e-01
-8.84704962e-02 -1.37887859e+00 1.73806936e-01 -6.98174536e-01
-1.27096605e+00 1.09963644e+00 -2.25867197e-01 -3.33413661e-01
-7.66292810e-01 -7.69190788e-01 -5.28599322e-01 1.40498459e+00
-1.27077639e+00 -7.65013754e-01 3.00217867e-01 2.61503309e-01
9.70279992e-01 -2.88577914e-01 1.29327273e+00 1.89861655e-01
-5.15711665e-01 8.69797111e-01 -1.41773731e-01 2.43885472e-01
7.91073561e-01 -1.51226830e+00 9.62120473e-01 9.46315825e-01
2.36772731e-01 6.77189469e-01 9.91810620e-01 -4.13138807e-01
-8.42345893e-01 -8.40372741e-01 1.31970727e+00 -5.54702282e-01
4.60616201e-01 -6.52800083e-01 -1.27800179e+00 1.03166616e+00
6.05237126e-01 -5.63506842e-01 6.97315991e-01 7.14334488e-01
-1.58321083e-01 -2.37630323e-01 -7.89513469e-01 6.43502295e-01
6.37843847e-01 -6.79661393e-01 -9.92919147e-01 1.51882574e-01
7.43369162e-01 -4.87514287e-01 -7.02356160e-01 2.56868482e-01
8.91719833e-02 -7.57071555e-01 7.21793950e-01 -8.21598947e-01
-3.61501336e-01 -3.17119867e-01 -1.72586590e-01 -2.08925843e+00
-4.16122049e-01 -7.61989415e-01 4.34900194e-01 1.38434362e+00
9.84224617e-01 -6.30943120e-01 6.62968099e-01 6.85204923e-01
-6.38961911e-01 -3.37631017e-01 -9.59456921e-01 -7.82721102e-01
3.74249697e-01 -2.05031380e-01 6.71708941e-01 1.05048811e+00
-2.07359856e-03 6.33823395e-01 -1.58694401e-01 4.14240509e-01
1.10425197e-01 -4.67555225e-01 5.15817523e-01 -9.99652863e-01
-1.22950114e-01 -4.54564780e-01 -6.06543958e-01 -8.98863018e-01
3.59600157e-01 -8.39908361e-01 5.30328691e-01 -9.23069894e-01
-4.32216167e-01 -2.33465582e-01 -4.85319346e-01 5.24746478e-01
-4.18019146e-01 -2.42670290e-02 1.48169726e-01 -4.33106780e-01
-1.01445168e-01 2.14441657e-01 7.43764877e-01 -1.14305094e-01
-3.45282465e-01 2.25547403e-01 -6.53385937e-01 8.40949595e-01
1.24923646e+00 -3.08018178e-01 -3.98879290e-01 -6.13207877e-01
-2.22053349e-01 -2.69498110e-01 -1.49977222e-01 -1.10967875e+00
4.19506311e-01 -7.82812759e-02 3.67778808e-01 -3.77215296e-01
4.58511204e-01 -3.60367686e-01 -8.69276002e-03 7.63638616e-02
-6.20265424e-01 3.07730854e-01 6.51843846e-01 -5.20135388e-02
-4.19441015e-01 -3.79216403e-01 1.07202935e+00 -1.26846865e-01
-9.36714053e-01 -3.12104613e-01 -8.85542333e-01 2.23369241e-01
4.41572458e-01 9.65313911e-02 -3.48059505e-01 -2.76884496e-01
-1.16260087e+00 -1.19698815e-01 2.40974858e-01 7.38011301e-01
2.91248798e-01 -9.70732450e-01 -9.07860458e-01 5.59046328e-01
-1.83365539e-01 -3.16632152e-01 1.96087182e-01 3.81725609e-01
-3.33224177e-01 4.67551649e-01 -2.04433084e-01 -2.81770676e-01
-1.36015713e+00 5.06535888e-01 6.80574358e-01 1.02598615e-01
-1.80440202e-01 1.19448984e+00 2.94608235e-01 -9.83428121e-01
5.81584752e-01 -3.39547157e-01 -3.30174230e-02 -1.37932956e-01
5.87778330e-01 1.21365212e-01 4.39301968e-01 -7.39339352e-01
-3.97262007e-01 -1.16226785e-01 -6.59213126e-01 -2.08595842e-01
1.03882921e+00 -1.03804141e-01 -4.48502833e-03 8.01256418e-01
8.90265822e-01 1.32616088e-01 -1.16321623e+00 9.03109908e-02
2.11482525e-01 1.44843748e-02 1.42298833e-01 -1.37744486e+00
-7.88588047e-01 7.67091990e-01 8.30756307e-01 5.28175414e-01
1.06129348e+00 3.83909456e-02 5.98775387e-01 6.36707127e-01
-9.65531990e-02 -1.48986757e+00 -4.72754836e-01 9.68220115e-01
7.94242382e-01 -9.52760577e-01 -6.10930681e-01 -3.09808195e-01
-5.96562386e-01 1.02910995e+00 7.34527588e-01 1.73344344e-01
7.24612117e-01 3.35937172e-01 8.52623284e-01 1.81052566e-01
-5.86833239e-01 -7.86266178e-02 2.37293899e-01 9.48627412e-01
8.65924239e-01 3.95147741e-01 1.18226662e-01 2.76242077e-01
-9.31404889e-01 -6.54425323e-01 5.89977562e-01 7.97798932e-01
-6.47348404e-01 -1.19608319e+00 -5.82904816e-01 2.47049525e-01
-3.56584728e-01 -4.90445077e-01 -7.01232195e-01 5.62969625e-01
1.38063133e-01 1.17224038e+00 2.81819165e-01 -2.89528042e-01
6.45516813e-01 7.94584394e-01 2.11374328e-01 -8.44858408e-01
-7.84370720e-01 -2.94547342e-02 4.44404036e-01 -6.31986558e-02
-2.42279589e-01 -9.46053386e-01 -1.16025817e+00 -1.46440670e-01
-1.63752422e-01 3.36436272e-01 8.48221302e-01 1.11607480e+00
2.72766113e-01 6.27952635e-01 4.76559520e-01 -6.74565315e-01
-5.78354955e-01 -1.28259814e+00 -7.10417271e-01 3.10142279e-01
4.40704197e-01 -1.97642237e-01 -5.75730264e-01 3.93313020e-01] | [14.354582786560059, 6.697058200836182] |
11417a43-3d1c-4454-b087-ff16d436baa2 | no-surprises-training-robust-lung-nodule | 2003.03824 | null | https://arxiv.org/abs/2003.03824v2 | https://arxiv.org/pdf/2003.03824v2.pdf | No Surprises: Training Robust Lung Nodule Detection for Low-Dose CT Scans by Augmenting with Adversarial Attacks | Detecting malignant pulmonary nodules at an early stage can allow medical interventions which may increase the survival rate of lung cancer patients. Using computer vision techniques to detect nodules can improve the sensitivity and the speed of interpreting chest CT for lung cancer screening. Many studies have used CNNs to detect nodule candidates. Though such approaches have been shown to outperform the conventional image processing based methods regarding the detection accuracy, CNNs are also known to be limited to generalize on under-represented samples in the training set and prone to imperceptible noise perturbations. Such limitations can not be easily addressed by scaling up the dataset or the models. In this work, we propose to add adversarial synthetic nodules and adversarial attack samples to the training data to improve the generalization and the robustness of the lung nodule detection systems. To generate hard examples of nodules from a differentiable nodule synthesizer, we use projected gradient descent (PGD) to search the latent code within a bounded neighbourhood that would generate nodules to decrease the detector response. To make the network more robust to unanticipated noise perturbations, we use PGD to search for noise patterns that can trigger the network to give over-confident mistakes. By evaluating on two different benchmark datasets containing consensus annotations from three radiologists, we show that the proposed techniques can improve the detection performance on real CT data. To understand the limitations of both the conventional networks and the proposed augmented networks, we also perform stress-tests on the false positive reduction networks by feeding different types of artificially produced patches. We show that the augmented networks are more robust to both under-represented nodules as well as resistant to noise perturbations. | ['Si-Qi Liu', 'Bogdan Georgescu', 'Sasa Grbic', 'Arnaud Arindra Adiyoso Setio', 'Eli Gibson', 'Florin C. Ghesu', 'Dorin Comaniciu'] | 2020-03-08 | null | null | null | null | ['lung-nodule-detection'] | ['medical'] | [ 5.61094999e-01 5.68393409e-01 1.82627514e-01 2.41806498e-03
-6.20667815e-01 -6.41399324e-01 3.72280777e-01 -2.35086098e-01
-2.99371690e-01 4.73254800e-01 -1.97734371e-01 -5.23651958e-01
2.21515536e-01 -9.32066560e-01 -9.05005813e-01 -7.86888003e-01
2.24999934e-02 1.86540172e-01 6.55948877e-01 -1.24122500e-02
-2.30898082e-01 7.17419922e-01 -1.22313547e+00 6.87830031e-01
6.47879541e-01 7.53574133e-01 5.95581671e-03 9.24958766e-01
4.07699287e-01 7.30764747e-01 -6.41091585e-01 -1.30480409e-01
6.72139823e-01 -5.23355126e-01 -4.53091353e-01 6.48918673e-02
2.18585372e-01 -3.98427844e-01 -4.63297635e-01 1.26425779e+00
6.64871275e-01 -2.91967869e-01 6.35786176e-01 -8.54842126e-01
-4.33742553e-01 6.00777864e-01 -2.36789867e-01 3.22029829e-01
3.59191000e-02 5.48128605e-01 3.91934514e-01 -7.14445591e-01
3.49186599e-01 1.00295281e+00 8.79630148e-01 9.81959939e-01
-9.91358340e-01 -5.97001076e-01 -4.56781477e-01 -2.28776217e-01
-1.10917771e+00 -2.03175947e-01 5.56382895e-01 -2.45906428e-01
4.57272917e-01 7.26817727e-01 5.54798603e-01 1.37700462e+00
2.19012722e-01 4.19090033e-01 1.03321517e+00 -4.86161500e-01
1.45583168e-01 4.77896363e-01 -4.84962702e-01 8.82476211e-01
7.21134901e-01 5.54610848e-01 3.48431945e-01 -3.01522732e-01
1.07745707e+00 -1.90931428e-02 -4.92659181e-01 -1.41445652e-01
-1.35596287e+00 8.52187276e-01 1.00567627e+00 5.13351083e-01
-3.64565670e-01 2.41266236e-01 3.34409833e-01 -8.49221349e-02
5.78967854e-02 9.63821709e-01 -9.77851525e-02 5.20106077e-01
-6.25643969e-01 -1.59876764e-01 8.60285759e-01 5.04962862e-01
2.19565347e-01 3.18057597e-01 -7.66909420e-01 5.58057249e-01
7.09498897e-02 5.12411833e-01 9.21049595e-01 -5.84340096e-01
1.83853388e-01 6.14707530e-01 2.42824089e-02 -1.03498304e+00
-3.20160061e-01 -7.11969197e-01 -1.04209006e+00 3.07647049e-01
5.39349735e-01 -3.52205515e-01 -1.31661284e+00 1.46193814e+00
1.73599884e-01 4.69392210e-01 2.10386857e-01 8.85970950e-01
8.73039305e-01 3.53493482e-01 -1.02519087e-01 3.00176460e-02
1.29118431e+00 -8.81376088e-01 -3.12220275e-01 -2.33618692e-01
7.10312903e-01 -6.79286003e-01 1.07585692e+00 3.20519060e-02
-8.34131300e-01 -6.16973221e-01 -1.23663044e+00 5.38883865e-01
-5.05232438e-02 3.68144393e-01 4.15671663e-03 1.05676150e+00
-7.94992208e-01 5.39884031e-01 -1.23311043e+00 -4.27894413e-01
6.47857308e-01 5.66009879e-01 -1.46178648e-01 7.99537003e-02
-1.11245668e+00 8.10613692e-01 4.44529623e-01 1.23936243e-01
-1.19612765e+00 -7.19248533e-01 -4.10841674e-01 1.58936113e-01
3.47154140e-01 -7.57063866e-01 1.22929847e+00 -1.15354991e+00
-1.35798991e+00 6.41300023e-01 3.89180332e-01 -8.04249823e-01
8.61800671e-01 2.39136472e-01 -2.16816589e-01 2.60184079e-01
-2.42093936e-01 6.79186344e-01 9.04622316e-01 -1.08113658e+00
-3.05735677e-01 9.65859443e-02 -2.67903268e-01 7.05321431e-02
-4.35844928e-01 -3.39294195e-01 -1.17498599e-01 -8.18717539e-01
7.87192285e-02 -1.29404545e+00 -6.16729021e-01 2.48665050e-01
-8.28454554e-01 4.30461913e-01 1.03455675e+00 -4.17762250e-01
6.16403639e-01 -2.10935950e+00 -6.08991206e-01 4.67024893e-01
1.55028924e-01 6.50961339e-01 -1.13681331e-01 -1.69132143e-01
-1.07712992e-01 5.55402875e-01 -1.86504111e-01 3.79496664e-01
-5.35281837e-01 3.25441748e-01 3.27817127e-02 3.91825408e-01
5.05146742e-01 1.05901718e+00 -7.42902637e-01 -3.65391999e-01
1.75034642e-01 4.82883424e-01 -5.07060051e-01 2.93545514e-01
-1.52711451e-01 5.44357121e-01 -5.92906892e-01 3.73905987e-01
3.78469229e-01 -3.30408841e-01 -1.19525552e-01 -2.65208602e-01
4.76942688e-01 -6.08586334e-02 -9.94293511e-01 6.90194249e-01
-2.97209799e-01 5.38644552e-01 -2.45047525e-01 -7.93827176e-01
8.39142799e-01 7.09087372e-01 3.00145417e-01 -2.65547812e-01
2.50970364e-01 4.57885355e-01 7.01503515e-01 -5.76170623e-01
-1.71483248e-01 -1.51061013e-01 3.63073677e-01 1.99292272e-01
-3.17608744e-01 -2.19388336e-01 -2.29110569e-01 -1.96448758e-01
1.44557559e+00 -6.46684408e-01 4.01723206e-01 -1.50715649e-01
6.94761336e-01 1.42399549e-01 3.05322409e-01 1.03316975e+00
-2.42370397e-01 8.02947998e-01 3.89138758e-01 -3.44806254e-01
-1.25543225e+00 -9.73475456e-01 -1.13881245e-01 2.83444554e-01
-1.85868680e-01 6.01137161e-01 -7.61305630e-01 -1.02053833e+00
-2.76905358e-01 5.28297126e-01 -7.14541554e-01 -4.53439742e-01
-6.97509646e-01 -9.07107711e-01 1.18284297e+00 5.93549669e-01
7.91723013e-01 -1.23374641e+00 -8.94896030e-01 2.27609668e-02
7.33828694e-02 -1.17200565e+00 -2.13184431e-01 3.05946410e-01
-8.53981733e-01 -1.25435388e+00 -8.90429854e-01 -8.98340940e-01
1.17529464e+00 1.74954310e-02 7.93593526e-01 4.69463557e-01
-6.64055645e-01 6.30908385e-02 -2.94859856e-01 -5.04591286e-01
-1.34756792e+00 2.65252553e-02 -9.97170135e-02 -1.26235873e-01
-5.07065021e-02 -2.75537908e-01 -6.87735915e-01 5.24781704e-01
-1.10900724e+00 -5.54993264e-02 1.00480700e+00 1.19482350e+00
5.71688652e-01 1.88450009e-01 4.14584428e-01 -1.20123780e+00
6.02300346e-01 -3.81801605e-01 -4.32533413e-01 9.27648917e-02
-3.87216955e-01 1.90799966e-01 9.66209829e-01 -9.04722214e-01
-8.17839563e-01 4.42505121e-01 -2.27454752e-02 -6.00027859e-01
-2.33853057e-01 2.96277881e-01 2.28426799e-01 -4.68239456e-01
1.46744084e+00 9.74720865e-02 7.99383074e-02 1.23347878e-01
8.36432278e-02 4.11805719e-01 5.00559807e-01 -2.93547325e-02
1.33300197e+00 5.39133012e-01 2.90045381e-01 -6.97403610e-01
-7.47023702e-01 -2.35000372e-01 -2.70529151e-01 -1.07593320e-01
8.15395772e-01 -6.26171350e-01 -1.59909636e-01 6.96687177e-02
-8.76890302e-01 -1.39693603e-01 -4.50841427e-01 5.39210260e-01
-3.13035429e-01 3.27981383e-01 -6.48562968e-01 -4.67332929e-01
-4.99576628e-01 -1.23215044e+00 5.79773784e-01 3.45593512e-01
-1.64805874e-02 -7.85770595e-01 -1.43044487e-01 -3.24165560e-02
6.30735874e-01 5.46605289e-01 9.79665339e-01 -1.26982474e+00
-4.87956941e-01 -6.85460508e-01 -1.20163769e-01 6.65488839e-01
3.61382991e-01 8.51522908e-02 -1.02390683e+00 -3.70361984e-01
3.04676443e-01 -2.28394330e-01 6.86331928e-01 3.64681751e-01
1.20090175e+00 -6.47129714e-01 -6.10302806e-01 5.78373969e-01
1.27874923e+00 2.44557589e-01 8.78888309e-01 2.49333754e-01
4.93819892e-01 1.96260586e-01 3.00429702e-01 2.77932677e-02
-8.39589000e-01 1.94742456e-01 7.80557930e-01 -4.34831798e-01
-3.51460099e-01 -1.54779777e-01 1.26222238e-01 2.82801241e-01
-5.04614972e-02 -4.34725553e-01 -8.84313345e-01 4.97841984e-01
-1.30078685e+00 -8.68731320e-01 -1.94915801e-01 2.09765100e+00
7.15242088e-01 3.70820224e-01 -1.66920081e-01 2.90942669e-01
9.19858932e-01 -3.01546246e-01 -5.69103837e-01 -1.25647098e-01
9.82043669e-02 4.31874901e-01 8.13284814e-01 1.29325658e-01
-1.24499023e+00 5.10062635e-01 6.03919554e+00 5.80969870e-01
-1.46665776e+00 -2.16064490e-02 9.12636697e-01 1.60431743e-01
2.37440672e-02 -3.86777937e-01 -2.83373535e-01 1.32397428e-01
7.00561702e-01 7.36703724e-02 9.99533162e-02 1.06904292e+00
1.40178308e-01 1.41610563e-01 -1.17369246e+00 5.07119894e-01
-1.94129929e-01 -1.26177394e+00 8.76818225e-02 -1.14586554e-01
1.02944696e+00 8.91108066e-03 3.51523072e-01 1.25881791e-01
1.93737596e-01 -1.36256266e+00 1.19648054e-01 2.71058798e-01
7.14929163e-01 -4.03392732e-01 1.26678956e+00 5.82761705e-01
-7.48627365e-01 -1.27682686e-01 -4.36350465e-01 4.48804826e-01
-4.41232800e-01 3.85922283e-01 -2.00334120e+00 2.11721241e-01
4.36842829e-01 -2.15478297e-02 -1.02251065e+00 1.22342563e+00
-2.02088535e-01 8.37204576e-01 -5.84667802e-01 -3.25069010e-01
1.86630562e-01 4.94786412e-01 6.74116135e-01 1.07320333e+00
7.26475000e-01 -1.66421503e-01 2.96018757e-02 1.11616743e+00
-6.10317253e-02 -2.25355569e-02 -7.20028877e-01 -5.83488345e-02
4.80399847e-01 1.27218294e+00 -1.04414642e+00 -2.50923723e-01
-1.30686358e-01 8.81581008e-01 -2.77225554e-01 2.67829537e-01
-9.77646172e-01 -2.25275248e-01 -8.56629685e-02 4.42509532e-01
3.22291106e-01 3.38797510e-01 -3.38577211e-01 -7.12183297e-01
-4.66032401e-02 -1.19716620e+00 1.94741905e-01 -6.19356930e-01
-1.12354875e+00 8.83962691e-01 -2.93570518e-01 -1.57336617e+00
-3.02026540e-01 -6.94179535e-01 -8.47700059e-01 7.56437421e-01
-1.04463959e+00 -9.33038950e-01 -6.60453141e-01 5.45448780e-01
3.61285567e-01 -1.68615103e-01 8.19957256e-01 1.55196544e-02
-3.10346007e-01 7.41009533e-01 -2.39064157e-01 4.90392029e-01
5.68603218e-01 -1.02876914e+00 2.65004456e-01 1.02478409e+00
5.97022176e-02 2.09537297e-01 7.41907537e-01 -6.69733584e-01
-8.49362373e-01 -1.59005296e+00 9.11727622e-02 -3.16656590e-01
3.88739407e-01 -7.68834725e-02 -1.02590597e+00 5.69297552e-01
-2.48851657e-01 5.39089978e-01 1.57608181e-01 -8.81684124e-01
-1.05832204e-01 1.56452164e-01 -1.55906177e+00 8.25055778e-01
4.73167002e-01 -1.67740688e-01 -3.66092145e-01 5.60731769e-01
6.76888585e-01 -5.69042921e-01 -4.97471929e-01 8.25207472e-01
2.54874736e-01 -9.01146293e-01 1.02546000e+00 -4.91592973e-01
4.21743274e-01 -3.00854951e-01 1.50110990e-01 -1.20773482e+00
-2.98752010e-01 -2.41795346e-01 2.41105214e-01 5.76358974e-01
7.63247669e-01 -6.02018416e-01 1.28574538e+00 2.91897506e-01
-1.04879998e-01 -8.65825891e-01 -8.06143880e-01 -7.39559233e-01
6.36369288e-02 -5.60357273e-02 2.43464693e-01 6.71716511e-01
-4.74202514e-01 -2.34282568e-01 -8.08115527e-02 5.81751287e-01
1.49792820e-01 -6.21354163e-01 5.30341804e-01 -9.72938120e-01
-6.12954259e-01 -3.08022052e-01 -6.02412760e-01 -4.75788742e-01
-4.10553634e-01 -9.04118955e-01 3.53789181e-01 -1.22724068e+00
1.12429366e-01 -3.22454751e-01 -2.32293829e-01 4.48500454e-01
-3.03486049e-01 4.89021063e-01 -2.36594658e-02 2.48015225e-01
1.67932034e-01 -4.10564840e-02 1.59525537e+00 -1.94864005e-01
-1.43234879e-01 5.57701409e-01 -5.01916051e-01 8.55356932e-01
9.27168369e-01 -8.14306676e-01 -3.71167541e-01 1.14851454e-02
1.58310980e-02 1.11778177e-01 6.94034576e-01 -1.42605615e+00
1.03989907e-01 7.90320188e-02 6.09696031e-01 -2.93605655e-01
9.08803791e-02 -9.80111003e-01 1.82674497e-01 1.32507014e+00
-4.38621521e-01 -3.06349665e-01 2.08214939e-01 7.35892653e-01
-5.15982471e-02 -5.01212060e-01 1.11248589e+00 -3.87738913e-01
1.54139977e-02 1.12663627e-01 -6.46113575e-01 -2.57068217e-01
1.21420777e+00 -4.03879493e-01 -2.20196769e-01 -3.63425612e-01
-6.91073835e-01 -1.68026254e-01 2.55541354e-01 3.46032679e-02
6.01266265e-01 -1.12263286e+00 -6.83071733e-01 4.08643484e-01
-7.78759867e-02 1.79275960e-01 -6.94609433e-02 8.30906153e-01
-9.96326447e-01 3.23745698e-01 -1.61010981e-01 -7.17148602e-01
-1.25213456e+00 7.72010863e-01 1.09679854e+00 -5.18712938e-01
-5.47751665e-01 9.40592706e-01 1.22792117e-01 -3.74946207e-01
3.22737098e-01 -6.91818833e-01 -5.04327044e-02 -8.61963511e-01
4.85611334e-02 7.92889148e-02 1.68634012e-01 -1.23826019e-03
-1.40848175e-01 1.43162087e-01 -3.55987623e-02 9.47374105e-02
7.80692279e-01 5.32415628e-01 2.65305459e-01 4.37049083e-02
8.77989590e-01 1.65668782e-02 -9.41468239e-01 2.88643595e-03
4.19145599e-02 -1.37334228e-01 -1.47112161e-01 -9.04539287e-01
-1.25847459e+00 7.42640138e-01 1.06479776e+00 5.30946314e-01
1.12733579e+00 -1.95913211e-01 5.30088603e-01 6.52236938e-01
-2.38520220e-01 -4.91995394e-01 4.65340167e-01 6.05646819e-02
7.69336164e-01 -1.25812984e+00 -1.00910425e-01 -7.56562650e-01
-5.30859888e-01 1.30893576e+00 8.89724851e-01 -4.83816862e-01
6.28669620e-01 4.67240185e-01 2.98878700e-01 -1.03086919e-01
-6.01608336e-01 1.38258725e-01 4.15454507e-01 6.53440833e-01
2.73899108e-01 3.73445712e-02 -3.24319378e-02 5.31199276e-01
-1.89585432e-01 -7.13908300e-02 8.19997072e-01 7.59934247e-01
-5.07178664e-01 -7.60358155e-01 -7.75966644e-01 8.16735327e-01
-7.70895422e-01 8.77567679e-02 -5.60663760e-01 1.06661749e+00
1.58714145e-01 6.95711017e-01 -1.21627420e-01 -3.11043173e-01
3.33657533e-01 -5.99854102e-04 2.56112605e-01 -8.55644464e-01
-8.50359917e-01 3.90508212e-02 -1.65139325e-02 -1.33377403e-01
-1.78372115e-01 -2.14865863e-01 -1.19232333e+00 2.50532418e-01
-6.28797948e-01 -1.53654933e-01 2.64294118e-01 6.97410047e-01
5.49998656e-02 1.03354919e+00 6.31020129e-01 -5.40577590e-01
-1.18589175e+00 -1.10900259e+00 -1.02401413e-01 4.35375899e-01
2.37424955e-01 -2.27718666e-01 -6.63624763e-01 -2.86444649e-02] | [15.190533638000488, -2.115506410598755] |
65073517-0ac7-472e-bdfa-5144e7c259c9 | multi-chart-generative-surface-modeling | 1806.02143 | null | http://arxiv.org/abs/1806.02143v3 | http://arxiv.org/pdf/1806.02143v3.pdf | Multi-chart Generative Surface Modeling | This paper introduces a 3D shape generative model based on deep neural
networks. A new image-like (i.e., tensor) data representation for genus-zero 3D
shapes is devised. It is based on the observation that complicated shapes can
be well represented by multiple parameterizations (charts), each focusing on a
different part of the shape. The new tensor data representation is used as
input to Generative Adversarial Networks for the task of 3D shape generation.
The 3D shape tensor representation is based on a multi-chart structure that
enjoys a shape covering property and scale-translation rigidity.
Scale-translation rigidity facilitates high quality 3D shape learning and
guarantees unique reconstruction. The multi-chart structure uses as input a
dataset of 3D shapes (with arbitrary connectivity) and a sparse correspondence
between them. The output of our algorithm is a generative model that learns the
shape distribution and is able to generate novel shapes, interpolate shapes,
and explore the generated shape space. The effectiveness of the method is
demonstrated for the task of anatomic shape generation including human body and
bone (teeth) shape generation. | ['Heli Ben-Hamu', 'Gal Avineri', 'Yaron Lipman', 'Haggai Maron', 'Itay Kezurer'] | 2018-06-06 | null | null | null | null | ['3d-shape-generation'] | ['computer-vision'] | [-6.37603402e-02 5.47141492e-01 7.92271942e-02 -3.97648402e-02
-3.72923434e-01 -7.36168683e-01 5.86266577e-01 -3.09086800e-01
3.30533028e-01 2.74377435e-01 3.19409698e-01 -3.15930516e-01
-2.32747812e-02 -1.19797146e+00 -8.39995682e-01 -8.52340877e-01
-1.31674260e-02 1.00686979e+00 -1.91473886e-01 -3.55752289e-01
-2.69684941e-02 1.10705030e+00 -1.07057571e+00 2.02561006e-01
5.42098105e-01 7.68006802e-01 -3.03357720e-01 5.82474232e-01
-2.74523884e-01 4.05353941e-02 -3.69706988e-01 -4.58324015e-01
4.41052556e-01 -1.68202400e-01 -8.06210518e-01 4.72903192e-01
2.02274472e-01 -1.44060269e-01 -2.34996960e-01 6.08101428e-01
4.26163614e-01 -2.42714718e-01 1.26379180e+00 -1.11688471e+00
-1.13255334e+00 3.91567618e-01 -1.05554432e-01 -5.08628666e-01
2.01231211e-01 1.08029470e-01 7.82024682e-01 -1.09744823e+00
1.04666841e+00 1.43933666e+00 6.93408549e-01 7.28223741e-01
-1.70060897e+00 -1.37278363e-01 -4.56350088e-01 -5.55618584e-01
-1.18821931e+00 1.07515447e-01 1.24952269e+00 -7.26152658e-01
4.16440606e-01 3.04906368e-01 1.13408422e+00 1.17233193e+00
6.21837914e-01 7.47804523e-01 8.69370103e-01 -2.89628774e-01
4.61175859e-01 -2.72889167e-01 -5.29068768e-01 8.20173025e-01
2.19705272e-02 1.62923902e-01 4.20048833e-02 -3.71919185e-01
1.50306046e+00 4.69197594e-02 -2.18863715e-03 -9.00657237e-01
-1.31932056e+00 8.47898602e-01 8.20431173e-01 5.69186270e-01
-3.85160416e-01 2.93905824e-01 1.48324639e-01 2.39519805e-01
3.71247709e-01 3.91039312e-01 -3.05873100e-02 2.38179088e-01
-5.92130780e-01 5.57973802e-01 6.67811632e-01 9.27006423e-01
6.11351550e-01 5.21630943e-01 -1.97102159e-01 6.99095309e-01
3.47104788e-01 8.18416774e-01 4.37688351e-01 -8.49806428e-01
-4.61586826e-02 8.84322107e-01 -8.71340632e-02 -1.17447174e+00
-4.34777856e-01 -3.73795241e-01 -1.33664215e+00 3.57442886e-01
2.86539614e-01 2.16024727e-01 -1.26595938e+00 1.67912471e+00
5.20163596e-01 -1.77303433e-01 -1.40422568e-01 7.50136435e-01
9.60980833e-01 4.11042333e-01 -1.89891145e-01 4.36225504e-01
1.13485289e+00 -1.47460446e-01 -2.48641014e-01 4.29503173e-01
3.07533026e-01 -7.03120530e-01 8.52135241e-01 -8.89445916e-02
-1.36994648e+00 -5.08049130e-01 -8.87120247e-01 -1.64482534e-01
-3.48527789e-01 -6.87938854e-02 5.26388347e-01 6.46177530e-01
-1.25981569e+00 5.69983184e-01 -8.39316905e-01 -1.60047695e-01
6.48745298e-01 2.54513770e-01 -5.59692085e-01 3.98565978e-02
-7.83174098e-01 6.69051766e-01 5.06335720e-02 -2.82638352e-02
-8.93655360e-01 -7.14582801e-01 -1.02116096e+00 -1.31241143e-01
-3.51747125e-01 -1.45313168e+00 8.01114738e-01 -6.55553162e-01
-1.57728362e+00 1.25600421e+00 2.47404248e-01 -1.11087129e-01
6.58346176e-01 4.63658482e-01 4.30823565e-02 1.94543004e-01
5.31603806e-02 8.02743018e-01 1.37332654e+00 -1.59653234e+00
3.68307799e-01 -4.17073041e-01 -1.32739440e-01 -8.07204321e-02
2.04439104e-01 -6.35520637e-01 -1.25892341e-01 -1.13826692e+00
5.96050084e-01 -9.63988245e-01 -5.03338516e-01 1.75184995e-01
-7.81687796e-01 6.33777753e-02 6.46611691e-01 -4.46338952e-01
4.72582281e-01 -1.96114016e+00 6.68750465e-01 5.79992890e-01
3.94455910e-01 -1.42896339e-01 -3.53056401e-01 6.69007838e-01
-3.20893705e-01 3.83985102e-01 -6.16325319e-01 -3.41752529e-01
-6.98801875e-02 4.01193142e-01 -3.31217289e-01 3.05559725e-01
5.86453855e-01 1.28094649e+00 -8.02758992e-01 -2.40275830e-01
1.84432596e-01 7.62339592e-01 -8.53576541e-01 1.92699105e-01
-4.03486520e-01 8.10349405e-01 -6.87105238e-01 7.93431997e-01
6.51035964e-01 -1.11830227e-01 -1.12471677e-01 -3.97403061e-01
2.14839041e-01 -1.26080140e-01 -8.70192051e-01 1.73202550e+00
-5.29353023e-01 -7.54732490e-02 -9.04404446e-02 -8.50242436e-01
1.29273772e+00 4.73962903e-01 8.18042278e-01 -1.70703664e-01
3.84008020e-01 4.79620099e-01 -1.82444349e-01 -2.62728453e-01
2.49662593e-01 -5.65202117e-01 -1.88972443e-01 7.21577644e-01
-1.88892633e-02 -8.85870576e-01 -2.75815487e-01 8.46399739e-02
8.47929657e-01 2.26940915e-01 -8.53542536e-02 -3.66764098e-01
4.40537393e-01 -2.11235732e-01 -1.09044351e-02 1.01879194e-01
3.87898058e-01 8.50700080e-01 5.07320166e-01 -1.00766420e+00
-1.76008201e+00 -1.53195202e+00 -2.58440614e-01 2.09945083e-01
-2.64013320e-01 1.95812836e-01 -6.86327875e-01 -3.72639894e-01
2.92905152e-01 5.09478748e-01 -1.09920311e+00 -2.89929450e-01
-7.74346709e-01 -3.22436631e-01 4.25187379e-01 3.32592607e-01
7.01166317e-03 -1.30305266e+00 -4.34193254e-01 3.08263391e-01
-9.19658840e-02 -8.25280845e-01 -6.43657506e-01 -1.85605913e-01
-1.29045129e+00 -9.63306189e-01 -1.10672796e+00 -9.68177974e-01
1.30357659e+00 -2.13619009e-01 1.16750383e+00 5.25938682e-02
-3.70220900e-01 7.19728410e-01 -7.99938962e-02 -1.79684624e-01
-1.15625525e+00 2.60795932e-03 4.32676896e-02 4.89024073e-01
-2.94109374e-01 -1.16086400e+00 -4.53062177e-01 1.45567939e-01
-1.42605960e+00 2.75310159e-01 5.05573988e-01 1.07030511e+00
1.05606341e+00 -3.26578379e-01 6.55096233e-01 -5.28431892e-01
6.48126721e-01 -3.23413104e-01 -2.26358056e-01 -1.80308986e-02
2.58634631e-02 5.37468553e-01 5.85175931e-01 -5.49119771e-01
-6.20693624e-01 2.55645037e-01 -4.06700224e-01 -7.79546499e-01
-1.85262710e-01 1.17169864e-01 -8.65079686e-02 -4.25211526e-02
7.33847558e-01 5.39546788e-01 5.22977829e-01 -6.30395770e-01
8.74039531e-01 1.02708705e-01 5.35603762e-01 -9.12562251e-01
1.35213566e+00 7.09204912e-01 5.11758089e-01 -8.96938264e-01
-2.84382820e-01 2.22704530e-01 -1.15750515e+00 -2.38429815e-01
8.72727990e-01 -4.59220290e-01 -5.32436669e-01 4.16062862e-01
-1.17013931e+00 -1.90570474e-01 -7.45620668e-01 -2.84315981e-02
-1.15420592e+00 2.28482947e-01 -4.65960026e-01 -4.68134880e-01
-6.97325766e-01 -1.12660801e+00 1.40816319e+00 -3.23703170e-01
-1.41507044e-01 -1.08715689e+00 1.59484044e-01 -3.83711345e-02
4.42143708e-01 1.05680716e+00 1.60011160e+00 -4.03601453e-02
-6.90053344e-01 -3.88099015e-01 1.90814316e-01 2.55573303e-01
2.42447659e-01 4.83969375e-02 -5.67939162e-01 -2.10844174e-01
-8.01768601e-02 -1.79003507e-01 5.09081483e-01 5.52718103e-01
1.03364980e+00 -4.82375950e-01 -1.68293387e-01 8.48183453e-01
1.29992127e+00 -9.09841582e-02 6.73689604e-01 -1.61514714e-01
8.54303837e-01 4.52222019e-01 -3.10911030e-01 2.08249032e-01
4.07410502e-01 6.67497456e-01 6.46588564e-01 -1.86648786e-01
-2.89618701e-01 -6.21772110e-01 -2.58765202e-02 8.96447420e-01
-3.15746427e-01 3.66899431e-01 -7.89628446e-01 6.24471605e-01
-1.24460781e+00 -6.85253918e-01 2.12014336e-02 2.15917301e+00
8.06242526e-01 2.40585823e-02 4.17849153e-01 2.62098312e-01
4.68324780e-01 -1.09194033e-01 -5.50230384e-01 -5.40533900e-01
-6.65240213e-02 5.47734439e-01 4.68214750e-02 3.53887320e-01
-6.80341899e-01 6.58548594e-01 6.46977854e+00 5.40604234e-01
-1.25355327e+00 -3.56003135e-01 6.19399488e-01 3.69136930e-01
-9.69859242e-01 -4.71851498e-01 -1.62092075e-01 8.08979794e-02
3.39478135e-01 -2.40974888e-01 3.28808904e-01 7.53380835e-01
-7.76750371e-02 6.50090873e-01 -9.55317676e-01 9.13261473e-01
-1.34860396e-01 -1.68138623e+00 7.31649756e-01 1.64303556e-01
9.04119909e-01 -2.27238163e-01 3.00468177e-01 -1.99404478e-01
2.47367769e-01 -1.18470097e+00 9.25247788e-01 6.71328366e-01
1.24335170e+00 -7.62855947e-01 2.76675314e-01 3.08995128e-01
-9.96802211e-01 2.99318194e-01 -3.51358265e-01 6.39756501e-01
1.87562376e-01 6.09750152e-01 -1.00312591e+00 6.35634482e-01
1.73934698e-01 5.13010144e-01 -2.44851127e-01 9.18721855e-01
-1.06247976e-01 1.57932654e-01 -4.44479555e-01 3.65152955e-02
2.39932433e-01 -3.89016628e-01 9.28726494e-01 7.34140933e-01
5.79930961e-01 4.47580963e-02 1.48859248e-01 1.47921658e+00
-2.32644349e-01 2.19709292e-01 -1.13843536e+00 -9.95900556e-02
2.05019966e-01 1.06755877e+00 -7.61978626e-01 1.80846509e-02
1.50895655e-01 6.61775351e-01 7.06279129e-02 3.01278800e-01
-2.99844593e-01 1.28855810e-01 4.78142768e-01 4.81513470e-01
3.87312293e-01 -5.84312201e-01 -6.62113726e-01 -9.31988001e-01
-1.50120720e-01 -6.06472611e-01 -4.99805845e-02 -8.54630888e-01
-1.38386023e+00 9.22737956e-01 -1.58053517e-01 -1.43388355e+00
-3.12447190e-01 -5.95056295e-01 -5.08403838e-01 8.25675249e-01
-9.17379618e-01 -1.66217995e+00 5.43851331e-02 6.63919628e-01
6.78985640e-02 -3.04001302e-01 1.15997231e+00 -3.07238922e-02
1.09377004e-01 4.48366702e-01 -1.18675627e-01 2.71590918e-01
-1.43453842e-02 -1.33045304e+00 7.84199595e-01 2.76538342e-01
2.56019205e-01 4.54355359e-01 3.49016875e-01 -6.35487795e-01
-1.68434465e+00 -1.24385297e+00 6.13640368e-01 -7.93274760e-01
3.84662330e-01 -1.98356137e-01 -6.82019174e-01 6.34962201e-01
-3.13664585e-01 4.88341525e-02 5.12724102e-01 -4.38944310e-01
-5.14766276e-01 2.00889587e-01 -1.44404244e+00 7.02284336e-01
9.12519395e-01 -3.00093442e-01 -6.26802325e-01 3.29025418e-01
6.39472187e-01 -6.01749599e-01 -1.29714191e+00 1.71635821e-01
5.98480642e-01 -7.40051806e-01 1.32888889e+00 -7.23411918e-01
7.00887024e-01 -2.43512481e-01 1.00006312e-02 -1.48901582e+00
-3.68748426e-01 -7.19652116e-01 -1.16546683e-01 4.56680715e-01
8.17289054e-02 -3.53814423e-01 8.15297008e-01 1.83502704e-01
-3.76631945e-01 -1.05792475e+00 -1.18747425e+00 -8.02849591e-01
5.62586784e-01 -3.56057465e-01 1.05165946e+00 7.49226272e-01
-5.64866900e-01 1.52495071e-01 -2.26198047e-01 -1.74101487e-01
7.76506662e-01 5.74283123e-01 8.01279545e-01 -1.22075868e+00
-1.01243056e-01 -6.13344610e-01 -5.77373803e-01 -9.01306450e-01
-4.63894531e-02 -1.39457309e+00 -4.08403546e-01 -1.46419668e+00
-3.08738619e-01 -6.43462062e-01 3.45967472e-01 5.61671257e-01
4.26296383e-01 4.73066211e-01 1.92533389e-01 1.67227075e-01
5.00183463e-01 9.24571812e-01 2.15432596e+00 -2.91759640e-01
-3.23959619e-01 2.90946931e-01 -4.42520469e-01 5.56735575e-01
5.85826874e-01 -2.49822140e-01 -1.64927021e-01 -4.04275656e-01
1.78153679e-01 3.23136598e-01 5.74973345e-01 -7.13445663e-01
-4.10985142e-01 1.00135528e-01 4.89417374e-01 -7.49642670e-01
3.69567037e-01 -9.34642732e-01 7.04829454e-01 7.10840881e-01
-1.46435320e-01 1.35700265e-02 5.80121018e-03 3.61705244e-01
2.27443893e-02 9.89268720e-02 1.01646709e+00 -3.27024609e-01
1.40167952e-01 7.38999486e-01 8.09270740e-02 -6.54793754e-02
6.91008270e-01 -4.56434458e-01 3.12988281e-01 -2.83563852e-01
-1.33738649e+00 -4.00036663e-01 7.14841723e-01 4.30320024e-01
9.44863975e-01 -2.16919780e+00 -1.04016948e+00 7.38730967e-01
6.82978109e-02 3.75942141e-01 1.28153831e-01 3.96740764e-01
-7.73407876e-01 1.69833377e-01 -5.65916896e-01 -8.68350804e-01
-5.53422153e-01 5.16489208e-01 7.25551426e-01 -2.36850411e-01
-9.61961985e-01 5.93537867e-01 9.59750488e-02 -7.91458309e-01
-2.45766655e-01 -6.06001019e-01 2.10876912e-02 -3.79834086e-01
1.34875357e-01 -2.78257299e-02 2.97234878e-02 -8.06413949e-01
-3.62471268e-02 9.59644735e-01 5.54336965e-01 -1.91452131e-02
1.65402782e+00 4.48667258e-01 -3.70542049e-01 2.42866829e-01
1.27094829e+00 -3.42988186e-02 -9.87143636e-01 -2.17463210e-01
-4.31277961e-01 -3.81198347e-01 -3.54745120e-01 -5.13283432e-01
-1.35416996e+00 7.39690125e-01 2.41230667e-01 4.49932247e-01
8.13362420e-01 1.59245417e-01 9.07508731e-01 1.27367452e-01
6.06810987e-01 -4.19843674e-01 3.64924878e-01 5.26585817e-01
1.80850124e+00 -5.29112935e-01 -2.71297336e-01 -3.75413746e-01
-2.51437753e-01 1.33479559e+00 -3.54931764e-02 -5.23829639e-01
9.98305023e-01 7.80698657e-02 -7.19450042e-02 -5.22049546e-01
-2.93574244e-01 8.07520449e-02 6.89012945e-01 1.07709610e+00
1.20015763e-01 3.34009707e-01 -1.05260454e-01 2.66298980e-01
-5.96596181e-01 -2.63851911e-01 4.85557467e-01 3.59981805e-01
-2.81015188e-02 -1.39876890e+00 -6.08137846e-01 2.75718093e-01
-5.67525439e-02 2.78928339e-01 -3.25914383e-01 5.97653806e-01
6.13144562e-02 2.16877252e-01 -4.37170900e-02 -4.15693104e-01
4.31653202e-01 1.21149570e-01 7.13785887e-01 -4.95925367e-01
-3.68376732e-01 7.27721825e-02 -3.24412286e-01 -4.12149876e-01
-1.95713252e-01 -4.81025368e-01 -1.17391610e+00 -2.50626028e-01
3.39785248e-01 -1.61495060e-01 7.93349385e-01 5.27558386e-01
4.77859825e-01 3.45438123e-01 9.95315731e-01 -1.25711322e+00
-5.66347539e-01 -6.33540928e-01 -6.75659955e-01 9.08398628e-01
3.42170268e-01 -7.58634150e-01 -9.38623548e-02 2.79138178e-01] | [8.827613830566406, -3.65655517578125] |
3f0ac7c1-43f3-476a-9174-57d02f4e0021 | read-watch-and-move-reinforcement-learning | 1901.06829 | null | http://arxiv.org/abs/1901.06829v1 | http://arxiv.org/pdf/1901.06829v1.pdf | Read, Watch, and Move: Reinforcement Learning for Temporally Grounding Natural Language Descriptions in Videos | The task of video grounding, which temporally localizes a natural language
description in a video, plays an important role in understanding videos.
Existing studies have adopted strategies of sliding window over the entire
video or exhaustively ranking all possible clip-sentence pairs in a
pre-segmented video, which inevitably suffer from exhaustively enumerated
candidates. To alleviate this problem, we formulate this task as a problem of
sequential decision making by learning an agent which regulates the temporal
grounding boundaries progressively based on its policy. Specifically, we
propose a reinforcement learning based framework improved by multi-task
learning and it shows steady performance gains by considering additional
supervised boundary information during training. Our proposed framework
achieves state-of-the-art performance on ActivityNet'18 DenseCaption dataset
and Charades-STA dataset while observing only 10 or less clips per video. | ['Xiao Liu', 'Dongliang He', 'Shilei Wen', 'Fu Li', 'Jizhou Huang', 'Xiang Zhao'] | 2019-01-21 | null | null | null | null | ['video-grounding'] | ['computer-vision'] | [ 2.73393631e-01 -6.04973622e-02 -5.54128289e-01 -3.35367769e-01
-1.15161347e+00 -5.23334444e-01 4.80485588e-01 -2.55484749e-02
-6.17236614e-01 7.77048290e-01 4.32785302e-01 -9.41108316e-02
-1.98795702e-02 -2.28303775e-01 -9.35660601e-01 -4.38042879e-01
-4.82984453e-01 2.15295181e-01 6.88925087e-01 1.36304051e-01
2.47265503e-01 -1.88335404e-01 -1.26467383e+00 7.69036233e-01
5.64342141e-01 1.13383651e+00 5.87149680e-01 6.98659778e-01
5.45649007e-02 1.37454808e+00 -3.90528738e-01 -2.21777514e-01
2.10532874e-01 -5.71837068e-01 -1.14169741e+00 4.62170660e-01
6.34894788e-01 -4.28026021e-01 -4.55065787e-01 1.11178219e+00
1.67730123e-01 4.07273740e-01 2.97619045e-01 -1.27698267e+00
-2.10013524e-01 7.64799953e-01 -6.56883836e-01 6.80328846e-01
7.43239820e-01 -8.83201510e-02 1.21787202e+00 -6.69029057e-01
6.60445333e-01 8.81098628e-01 3.81440580e-01 6.07058048e-01
-8.53232980e-01 -6.59352541e-01 6.15503013e-01 7.30298519e-01
-1.56510127e+00 -5.31152904e-01 7.47172654e-01 -5.94870865e-01
1.00376129e+00 -5.28985187e-02 5.79745293e-01 1.04625976e+00
2.77287692e-01 1.03052580e+00 7.89125144e-01 -1.95802525e-01
1.78192198e-01 -3.10780674e-01 -2.59535819e-01 8.94261479e-01
-2.72619218e-01 -2.86440462e-01 -9.40787017e-01 1.39692336e-01
7.21613824e-01 -1.42664313e-01 -3.10317218e-01 -2.47326463e-01
-1.41000867e+00 6.28081799e-01 -5.72049059e-02 2.70950049e-01
-3.79794180e-01 2.72688329e-01 6.71011925e-01 2.96475768e-01
2.86931276e-01 3.38327557e-01 -4.17208910e-01 -3.59274119e-01
-1.27180362e+00 2.64271259e-01 4.46627110e-01 1.00754786e+00
6.82730198e-01 -1.16923198e-01 -3.52180392e-01 5.14355421e-01
1.85064659e-01 -8.36880282e-02 5.08803308e-01 -1.27308142e+00
7.86014557e-01 3.13549280e-01 2.20184028e-01 -8.39538217e-01
-1.29504830e-01 7.66891101e-03 -5.18288374e-01 -1.33551106e-01
2.03144848e-01 -3.94048780e-01 -7.26677120e-01 1.76243806e+00
9.68542323e-02 8.73439133e-01 2.68157385e-02 9.82872725e-01
6.35302544e-01 9.26313519e-01 3.49433362e-01 -5.91173470e-01
1.21246302e+00 -1.36627221e+00 -7.97152936e-01 -1.33307710e-01
3.96267951e-01 -3.98966640e-01 8.57862592e-01 6.10162377e-01
-1.10322046e+00 -8.02575409e-01 -1.08182395e+00 2.45797515e-01
-3.58350463e-02 6.95919916e-02 4.29415762e-01 3.63684632e-02
-1.08013499e+00 5.74828029e-01 -9.07002151e-01 -3.26517254e-01
4.08112586e-01 3.30601275e-01 -4.05474514e-01 1.09199263e-01
-1.28112531e+00 5.53833842e-01 7.36429751e-01 1.29735339e-02
-1.47881246e+00 -4.87162650e-01 -8.52561057e-01 -1.09156901e-02
9.38973725e-01 -4.66946363e-01 1.39508057e+00 -1.26989388e+00
-1.53701460e+00 8.16390872e-01 -7.82384276e-02 -9.15093184e-01
5.61451852e-01 -4.86259282e-01 -5.85614324e-01 7.14733064e-01
1.28987998e-01 9.18084800e-01 9.11979556e-01 -9.00950134e-01
-1.06462276e+00 7.95511678e-02 5.22330165e-01 4.40717429e-01
-1.87961549e-01 3.68162960e-01 -7.74945021e-01 -5.97190619e-01
-3.33552122e-01 -7.27008581e-01 -2.60671854e-01 -2.29897842e-01
-1.91881627e-01 -4.60795671e-01 7.07988977e-01 -7.97298133e-01
1.59688294e+00 -1.95024395e+00 3.24376345e-01 -3.36120307e-01
7.85579830e-02 1.46343142e-01 -1.46534696e-01 2.53991574e-01
1.48481891e-01 6.54784366e-02 1.17928900e-01 -2.84693629e-01
-3.16723466e-01 8.35582390e-02 -2.28160784e-01 5.33623278e-01
1.78609237e-01 5.29815793e-01 -1.08784080e+00 -1.08711720e+00
2.21507385e-01 7.11766630e-02 -6.54201806e-01 5.46027780e-01
-5.67797959e-01 4.49775249e-01 -5.53576887e-01 5.44490933e-01
1.59345552e-01 -4.51088309e-01 2.07182571e-01 -1.72921628e-01
-2.17296571e-01 -3.41213890e-03 -1.14672220e+00 2.56043816e+00
-1.86751351e-01 6.80446804e-01 -1.03773363e-02 -1.28081465e+00
5.27734041e-01 6.39130235e-01 9.24119294e-01 -3.85032564e-01
-9.08101276e-02 -2.79021710e-01 -4.44493629e-02 -9.30597305e-01
3.39102596e-01 -1.14829494e-02 -1.99865118e-01 2.01898545e-01
3.31202149e-01 4.42533314e-01 4.10107106e-01 2.08926082e-01
1.30424058e+00 6.35113835e-01 3.68827492e-01 -1.89411327e-01
6.32040501e-01 8.13703686e-02 7.64185786e-01 7.87617087e-01
-6.65593088e-01 6.08302057e-01 4.96630907e-01 -4.19752300e-01
-8.30912888e-01 -9.73638415e-01 2.78331429e-01 1.35466874e+00
5.39973140e-01 -6.47867680e-01 -8.11775327e-01 -8.14775765e-01
-4.70440269e-01 3.17845434e-01 -6.46052420e-01 5.94104528e-02
-7.60800242e-01 -3.56419653e-01 3.87580246e-01 4.28846210e-01
8.66926074e-01 -1.24404359e+00 -9.48163688e-01 3.35655600e-01
-5.41050494e-01 -1.63465333e+00 -7.32277095e-01 7.60358945e-02
-6.10169113e-01 -1.12409914e+00 -5.94587982e-01 -8.79154742e-01
4.49491173e-01 1.80783391e-01 1.15972424e+00 -1.74603000e-01
-7.52570527e-03 2.88145304e-01 -5.89816928e-01 2.04764009e-01
-1.78288877e-01 9.73004401e-02 -6.73589110e-02 2.44266674e-01
2.17636988e-01 -3.69574279e-01 -7.60254383e-01 2.60789335e-01
-7.43330300e-01 3.32158983e-01 5.36864579e-01 6.98555946e-01
6.21431947e-01 1.84237212e-01 6.46696150e-01 -6.83839560e-01
3.27747285e-01 -5.84669948e-01 -4.02414143e-01 5.57790279e-01
-1.09273180e-01 -1.98911503e-02 8.05471301e-01 -3.67597520e-01
-1.15106201e+00 2.45373517e-01 1.98881239e-01 -7.72555828e-01
-2.34605521e-01 4.67255175e-01 -4.92783040e-02 2.32214764e-01
2.80048758e-01 2.51129597e-01 -3.98676455e-01 -1.11632936e-01
1.81702048e-01 3.95499676e-01 7.53225625e-01 -6.40495777e-01
4.90732610e-01 3.27624261e-01 -3.24003428e-01 -6.40372455e-01
-1.08741212e+00 -9.00820792e-01 -6.92640185e-01 -6.82146072e-01
1.34608865e+00 -1.30240369e+00 -7.82020211e-01 6.01955093e-02
-1.12957561e+00 -4.49482143e-01 1.67404816e-01 5.53711414e-01
-8.55762959e-01 3.71043116e-01 -5.19530594e-01 -5.29702067e-01
-6.37353733e-02 -1.03363097e+00 9.66008961e-01 2.88526595e-01
-3.12964261e-01 -8.32121730e-01 1.69130772e-01 4.53004241e-01
-1.24925248e-01 3.63516122e-01 4.90513742e-01 -6.35009706e-01
-9.28052604e-01 7.31300935e-02 -5.61633967e-02 2.96169430e-01
1.31131053e-01 -9.07915980e-02 -6.59629643e-01 -3.06223541e-01
-2.09273845e-01 -6.51861191e-01 9.48299944e-01 4.38846439e-01
1.48247242e+00 -2.72926599e-01 -2.43060976e-01 4.37780827e-01
1.62176228e+00 4.49451596e-01 4.67250675e-01 3.19132954e-01
5.76677859e-01 3.96865398e-01 1.01376867e+00 7.04496086e-01
2.41574198e-01 6.58896387e-01 4.58892763e-01 7.22952038e-02
1.12237846e-02 -3.52694750e-01 5.69986522e-01 6.18349433e-01
-8.99946615e-02 -3.86506528e-01 -6.90431178e-01 6.20529652e-01
-2.28918457e+00 -1.46305239e+00 6.67661548e-01 1.92000365e+00
7.47009993e-01 4.29721653e-01 4.31783944e-01 -2.60872036e-01
8.93090367e-01 5.88740826e-01 -5.11149347e-01 -4.53066155e-02
8.14111754e-02 -1.62712485e-01 3.62149000e-01 4.68981147e-01
-1.48879492e+00 1.15425670e+00 6.41262388e+00 8.29795301e-01
-1.01524925e+00 2.22346395e-01 6.67817712e-01 -2.69610822e-01
1.33840233e-01 -4.26813886e-02 -7.28509068e-01 4.32858467e-01
8.24776769e-01 -1.69404522e-01 4.15734738e-01 6.46877408e-01
6.80785000e-01 -5.14479160e-01 -1.28374302e+00 1.07293797e+00
2.78070122e-01 -1.48244119e+00 1.15882963e-01 -4.18679327e-01
8.76348972e-01 5.68322241e-02 -2.43311927e-01 3.24813843e-01
1.37893558e-01 -8.44630897e-01 8.75241041e-01 3.80878299e-01
7.22738802e-01 -5.61368704e-01 3.84063035e-01 3.76697272e-01
-1.39567816e+00 -1.47791311e-01 -3.57427895e-02 -2.18212724e-01
3.81779581e-01 -3.89247993e-03 -6.72194839e-01 4.00604784e-01
6.88557565e-01 9.35274243e-01 -3.55546147e-01 1.19183660e+00
-4.22516055e-02 6.94910169e-01 5.53688779e-02 6.27344772e-02
5.54060698e-01 1.63255911e-03 4.13654387e-01 1.35553145e+00
1.55691415e-01 2.24980518e-01 8.30048501e-01 3.16923767e-01
-1.36746272e-01 -7.26918057e-02 -4.45565164e-01 3.85589525e-02
2.65207559e-01 9.97370243e-01 -1.06148839e+00 -5.88041008e-01
-7.33082235e-01 1.12796938e+00 2.72764087e-01 4.41235632e-01
-1.30698037e+00 -1.75398320e-01 4.80121851e-01 9.72559303e-02
4.70915854e-01 -2.06464499e-01 2.84169525e-01 -1.31384349e+00
5.98133616e-02 -7.63700187e-01 6.70537829e-01 -7.57886529e-01
-7.05399036e-01 7.17088223e-01 2.85943657e-01 -1.43057561e+00
-3.00009668e-01 -2.92370647e-01 -6.19805336e-01 1.77228928e-01
-1.42556965e+00 -8.08811545e-01 -2.92892456e-01 7.78995156e-01
1.42347610e+00 -1.82873890e-01 4.23871368e-01 4.78595287e-01
-6.88239872e-01 2.70226479e-01 -1.87032431e-01 1.97385877e-01
7.51280129e-01 -1.21755302e+00 -5.45644686e-02 1.07695937e+00
4.33952421e-01 1.46272838e-01 8.18676651e-01 -6.28875911e-01
-1.08816457e+00 -1.10635698e+00 6.81911469e-01 -1.47328824e-01
8.47643554e-01 -2.68510491e-01 -6.81934893e-01 7.78739393e-01
6.85342014e-01 1.16870098e-01 5.69454253e-01 -1.99467152e-01
7.68613666e-02 -2.35515416e-01 -6.56861186e-01 5.22441685e-01
1.22984016e+00 -5.29119551e-01 -7.12178469e-01 4.91155088e-01
7.84714818e-01 -5.77777386e-01 -7.58205891e-01 2.93031365e-01
4.40023363e-01 -9.81751859e-01 6.41001463e-01 -7.19504952e-01
6.96893811e-01 -4.49872702e-01 -7.46311843e-02 -9.42023158e-01
-2.36368477e-01 -1.06771016e+00 -1.60455793e-01 1.11066270e+00
3.40020597e-01 3.41094375e-01 1.04548514e+00 2.58570760e-01
-3.18526179e-01 -8.13344061e-01 -9.79546607e-01 -5.50822914e-01
-5.28270066e-01 -3.48986119e-01 6.37699962e-02 8.17747414e-01
3.29418451e-01 3.05146366e-01 -7.79917955e-01 2.25363433e-01
4.51128632e-01 6.95755482e-02 4.78399456e-01 -5.98137796e-01
-3.56351882e-01 -1.30568773e-01 -3.13360780e-01 -1.43052316e+00
5.53187251e-01 -6.14537954e-01 4.53197628e-01 -1.44416809e+00
4.22980160e-01 9.07959044e-02 -6.32243156e-01 3.49584073e-01
-5.89754768e-02 9.33002010e-02 1.80060774e-01 1.84052378e-01
-1.81354010e+00 3.27969313e-01 1.16409123e+00 -9.92959663e-02
-1.74299523e-01 -7.10278377e-02 -3.41326386e-01 8.07934999e-01
5.52167237e-01 -4.00248289e-01 -7.33711541e-01 -5.66738665e-01
4.09693122e-02 6.60369396e-01 2.02808037e-01 -1.27545345e+00
5.11865199e-01 -4.43804502e-01 1.94098592e-01 -6.26318395e-01
2.02669278e-01 -8.14162850e-01 -5.32232458e-03 2.97505975e-01
-7.92293906e-01 2.85291404e-01 4.86743115e-02 6.79037571e-01
-6.01756454e-01 -1.68011591e-01 5.82149982e-01 -3.42456490e-01
-1.50196016e+00 5.37837386e-01 -5.52389026e-01 2.18431696e-01
1.40589297e+00 -2.35860854e-01 2.39747670e-02 -2.86883801e-01
-8.86057019e-01 5.60418010e-01 3.90544124e-02 5.89179516e-01
6.59669161e-01 -1.20288217e+00 -5.83866000e-01 -2.38917857e-01
8.76904093e-03 -6.46342337e-02 4.02179807e-01 5.60628414e-01
-4.74540353e-01 4.24194485e-01 -3.35012704e-01 -6.93496466e-01
-1.25940228e+00 5.49671769e-01 2.63405472e-01 -4.26257074e-01
-5.27696729e-01 9.13042724e-01 1.94627330e-01 6.25101984e-01
4.70746219e-01 -3.97357047e-01 -4.08479333e-01 8.87531862e-02
4.53364968e-01 1.88008100e-01 -3.12737733e-01 -5.88450372e-01
-3.49466741e-01 5.35212219e-01 -2.44812533e-01 -1.62912965e-01
1.00327826e+00 -3.63171726e-01 3.23572844e-01 5.83296180e-01
1.21908092e+00 -3.68445247e-01 -1.75501537e+00 -1.32508710e-01
1.44446060e-01 -3.11458439e-01 -2.07444616e-02 -5.58719993e-01
-7.63313174e-01 5.71209788e-01 3.10360670e-01 3.15870866e-02
1.22581387e+00 2.18041018e-01 7.93664098e-01 3.56511801e-01
4.66455728e-01 -1.34537339e+00 6.45771980e-01 2.43240967e-01
6.70512319e-01 -1.47229731e+00 -6.40412867e-02 -2.35290274e-01
-8.22412670e-01 9.92263079e-01 9.74118531e-01 -2.77945101e-01
4.22586620e-01 5.18762618e-02 -1.43607765e-01 6.95054382e-02
-1.11780715e+00 -1.02013521e-01 1.56584620e-01 2.94116437e-01
2.66017377e-01 -2.37396196e-01 -1.28822222e-01 4.95433539e-01
4.09281790e-01 3.11607897e-01 3.61413509e-01 7.47569025e-01
-6.59773052e-01 -7.90541589e-01 -5.50192185e-02 3.67787153e-01
-7.08068013e-01 5.51387183e-02 9.17416885e-02 6.84712112e-01
2.50783265e-01 7.97504783e-01 2.61073828e-01 -4.29396868e-01
6.17217124e-02 -1.05883986e-01 6.43040061e-01 -6.88867748e-01
-4.34131622e-01 3.20880711e-01 1.96075648e-01 -7.62737632e-01
-9.97342050e-01 -9.26617384e-01 -1.24274552e+00 1.39802426e-01
-7.22692832e-02 4.18413073e-01 3.65810916e-02 1.17850208e+00
1.41223356e-01 6.64794743e-01 6.24326229e-01 -6.93618119e-01
-3.59444559e-01 -8.00500214e-01 -4.67249155e-01 3.67070913e-01
4.05668080e-01 -6.17721200e-01 -3.37784104e-02 6.45323455e-01] | [8.930583000183105, 0.5056243538856506] |
8de22a7e-ade5-43bb-9af4-b6a975a1b6e4 | experimentally-realized-memristive-memory | 2204.07429 | null | https://arxiv.org/abs/2204.07429v1 | https://arxiv.org/pdf/2204.07429v1.pdf | Experimentally realized memristive memory augmented neural network | Lifelong on-device learning is a key challenge for machine intelligence, and this requires learning from few, often single, samples. Memory augmented neural network has been proposed to achieve the goal, but the memory module has to be stored in an off-chip memory due to its size. Therefore the practical use has been heavily limited. Previous works on emerging memory-based implementation have difficulties in scaling up because different modules with various structures are difficult to integrate on the same chip and the small sense margin of the content addressable memory for the memory module heavily limited the degree of mismatch calculation. In this work, we implement the entire memory augmented neural network architecture in a fully integrated memristive crossbar platform and achieve an accuracy that closely matches standard software on digital hardware for the Omniglot dataset. The successful demonstration is supported by implementing new functions in crossbars in addition to widely reported matrix multiplications. For example, the locality-sensitive hashing operation is implemented in crossbar arrays by exploiting the intrinsic stochasticity of memristor devices. Besides, the content-addressable memory module is realized in crossbars, which also supports the degree of mismatches. Simulations based on experimentally validated models show such an implementation can be efficiently scaled up for one-shot learning on the Mini-ImageNet dataset. The successful demonstration paves the way for practical on-device lifelong learning and opens possibilities for novel attention-based algorithms not possible in conventional hardware. | ['Can Li', 'John Paul Strachan', 'Catherine E. Graves', 'Xia Sheng', 'X. Sharon Hu', 'Michael Neimier', 'Ann Franchesca Laguna', 'Arman Kazemi', 'Yahui Zhao', 'Bo Wen', 'Ruibin Mao'] | 2022-04-15 | null | null | null | null | ['one-shot-learning'] | ['methodology'] | [ 8.16505551e-02 -1.28896654e-01 -2.49198020e-01 -3.88459153e-02
-2.71276653e-01 -2.14158627e-03 1.87599674e-01 1.70244545e-01
-9.27827597e-01 8.06817949e-01 -4.49986160e-01 -1.24398075e-01
-1.12214945e-01 -1.02107692e+00 -1.14250195e+00 -1.14863575e+00
8.31624195e-02 5.16028047e-01 6.94383442e-01 -2.77964264e-01
2.86791265e-01 3.55708838e-01 -2.00379753e+00 2.47565463e-01
5.04553437e-01 1.07767951e+00 4.85362798e-01 2.96155602e-01
-2.84171999e-01 6.20864928e-01 -5.08889794e-01 1.51169114e-02
7.24390671e-02 -1.76344573e-01 -1.81077778e-01 -7.03123033e-01
5.64780056e-01 -4.02779847e-01 -6.88999534e-01 8.86861980e-01
6.53031468e-01 -1.92719549e-02 5.18988431e-01 -9.17394698e-01
-4.47336406e-01 1.02309299e+00 -2.35007063e-01 4.30322051e-01
-3.55434865e-01 3.24066639e-01 5.62758148e-01 -8.22135568e-01
3.50743860e-01 8.41236889e-01 5.39862216e-01 7.79630601e-01
-1.12619555e+00 -8.29988658e-01 -4.30983871e-01 3.97201270e-01
-1.36014116e+00 -2.18246698e-01 7.47688949e-01 -1.58899240e-02
1.55108881e+00 5.73364198e-02 1.02216220e+00 1.04604149e+00
7.33083785e-01 5.79125345e-01 1.12998521e+00 -3.33277255e-01
7.56158650e-01 2.12746799e-01 7.74308145e-01 6.59432113e-01
7.14313030e-01 1.70503750e-01 -7.53542840e-01 1.24916650e-01
5.77057362e-01 4.11972195e-01 1.90456696e-02 -2.39720508e-01
-7.71404028e-01 6.99432969e-01 9.73312616e-01 5.03979921e-01
5.58536761e-02 6.41301632e-01 6.78302526e-01 1.77093238e-01
-2.00876176e-01 2.89818197e-01 2.65320949e-02 -5.73919751e-02
-1.14123833e+00 -1.87632561e-01 6.39506102e-01 6.73222184e-01
7.68305719e-01 5.62845230e-01 1.73765630e-01 5.96636951e-01
1.57867417e-01 1.02918935e+00 9.58464861e-01 -3.71528953e-01
3.86332497e-02 7.97806025e-01 -4.89671230e-01 -6.90785408e-01
-7.94862747e-01 -5.03904700e-01 -1.15412772e+00 3.04112285e-01
2.03771800e-01 3.16131890e-01 -9.99235570e-01 1.65151668e+00
-2.76372563e-02 1.34811684e-01 5.08772284e-02 8.66417885e-01
8.58300745e-01 1.03572631e+00 -4.71545570e-02 -9.95454937e-02
1.37282073e+00 -8.42383921e-01 -6.77908242e-01 -1.69449344e-01
6.70435727e-01 -5.85493967e-02 1.15976584e+00 3.21342796e-01
-1.00927520e+00 -8.41201186e-01 -1.85819292e+00 -1.44060433e-01
-6.93360627e-01 -1.67336851e-01 4.54055637e-01 9.73986149e-01
-1.03653109e+00 5.50552845e-01 -1.07632494e+00 -2.62181759e-01
2.97352701e-01 1.07723522e+00 -2.44805276e-01 2.41430327e-01
-1.26968896e+00 7.97786415e-01 5.44812083e-01 3.45945060e-02
-8.34974349e-01 -4.68139440e-01 -5.50872266e-01 3.03974003e-01
-1.95353001e-01 -5.30196905e-01 6.38265312e-01 -7.02119291e-01
-1.68580544e+00 6.88696742e-01 3.20744455e-01 -1.07615292e+00
-2.89474249e-01 1.51954368e-01 -3.55027586e-01 3.01653575e-02
-5.38205922e-01 9.16528642e-01 9.83698368e-01 -5.41490674e-01
-1.86025530e-01 -5.58980763e-01 -2.56114900e-01 -3.46386105e-01
-1.28299284e+00 -6.00404620e-01 9.09635574e-02 -4.36255395e-01
1.35217115e-01 -1.09545624e+00 9.19955000e-02 -4.32884768e-02
7.03306571e-02 9.04263929e-02 1.14476323e+00 4.17510457e-02
1.19181120e+00 -2.37895131e+00 -1.29957765e-01 1.31899789e-01
-7.46839633e-03 5.12583315e-01 7.59765580e-02 2.22403035e-01
3.26991588e-01 -4.89048123e-01 -2.16537297e-01 3.67484540e-02
-2.33249679e-01 1.88544914e-01 -6.50050104e-01 4.47195321e-01
-1.12007089e-01 9.16554034e-01 -4.42380100e-01 -1.72102183e-01
1.42382324e-01 5.59562624e-01 -5.87801039e-01 -1.20152742e-01
-3.51463743e-02 -1.23248048e-01 7.33606890e-02 4.43010420e-01
7.18851686e-01 -3.21921825e-01 2.98137397e-01 -2.56428838e-01
-2.03603134e-01 2.67036349e-01 -8.85139465e-01 1.83438241e+00
-5.25532544e-01 5.59236586e-01 -2.50651717e-01 -9.38428521e-01
1.38258791e+00 -3.98464054e-02 1.00631557e-01 -1.31006801e+00
1.95944145e-01 6.73992693e-01 3.09770554e-01 -5.26788048e-02
6.97236598e-01 -3.61905433e-02 -1.69226274e-01 6.14660919e-01
3.97640437e-01 5.84220625e-02 -1.16815180e-01 -4.02729586e-03
1.08823752e+00 -3.16329330e-01 -9.42279771e-02 -7.34960616e-01
4.86750752e-01 -2.90500727e-02 2.57050008e-01 9.02378142e-01
-1.26963139e-01 2.83680439e-01 4.24017310e-02 -6.86185241e-01
-1.20941389e+00 -1.13455689e+00 -4.04199868e-01 8.07546496e-01
3.46792847e-01 -1.59601763e-01 -7.91577101e-01 -1.47128582e-01
-2.09092665e-02 3.30920726e-01 -3.70874375e-01 -7.16855764e-01
-8.69527459e-01 -9.78815794e-01 7.72659957e-01 7.65192747e-01
7.44429708e-01 -1.01100862e+00 -1.27143276e+00 3.88410181e-01
7.20901608e-01 -8.53150666e-01 -6.79752827e-02 8.97508264e-01
-1.34306681e+00 -3.98808122e-01 -3.57298881e-01 -9.43698406e-01
5.24593592e-01 1.17452212e-01 6.89283788e-01 4.36857827e-02
-6.26604855e-01 -4.13390510e-02 3.49510372e-01 4.35268953e-02
-8.95369500e-02 5.51936448e-01 5.98275125e-01 -2.19268054e-01
5.03870308e-01 -7.67949760e-01 -8.68070543e-01 8.26489180e-02
-9.25717473e-01 -1.79339439e-01 6.86002254e-01 1.42026103e+00
5.58150589e-01 -6.56937137e-02 9.95527625e-01 -6.55386269e-01
-1.54713597e-02 -2.81442642e-01 -8.62910867e-01 -5.12994900e-02
-7.57166684e-01 4.16501224e-01 8.21417928e-01 -6.41745269e-01
-6.74413860e-01 1.39769956e-01 -2.59829670e-01 -2.04966277e-01
4.10889059e-01 1.72813118e-01 -5.01470193e-02 -2.88042039e-01
7.31200755e-01 4.53999102e-01 2.56785482e-01 -5.16516296e-03
1.87069569e-02 7.32283235e-01 3.10390115e-01 -2.73616910e-01
2.96901792e-01 5.60915291e-01 2.58288980e-01 -1.00006938e+00
-4.35205363e-02 1.63701385e-01 -2.37602949e-01 -1.06362499e-01
4.65125501e-01 -1.06671727e+00 -1.05005240e+00 5.88732719e-01
-7.91270196e-01 -5.43312490e-01 -4.66545194e-01 5.19578159e-01
-4.15930271e-01 -2.33139619e-01 -1.04633522e+00 -7.99764335e-01
-8.17505538e-01 -1.32000983e+00 5.06933033e-01 5.23534954e-01
-4.68137860e-02 -7.83998251e-01 -3.79952043e-02 1.60965994e-02
8.12094450e-01 -4.80042756e-01 1.19902194e+00 -4.58868653e-01
-9.77271736e-01 -7.43746907e-02 -1.04654923e-01 2.10071996e-01
-4.01969731e-01 -1.99335381e-01 -1.22164989e+00 -5.78920364e-01
3.24833125e-01 -6.52568519e-01 1.41175663e+00 3.30310524e-01
9.24504220e-01 6.54588193e-02 -4.07576978e-01 3.67660403e-01
1.60191929e+00 2.98226476e-01 8.79998982e-01 2.49734253e-01
5.48120737e-01 1.09869681e-01 2.17501953e-01 1.76161095e-01
-5.90826608e-02 7.32012272e-01 4.39987749e-01 2.73001105e-01
-2.26462349e-01 -1.55882865e-01 6.20348334e-01 1.31713784e+00
6.25695944e-01 -1.71215266e-01 -7.59073377e-01 3.88716310e-01
-1.71052957e+00 -8.68781865e-01 -1.43425480e-01 2.46061873e+00
7.60507822e-01 4.78367537e-01 -2.36738428e-01 3.20886880e-01
5.11152208e-01 9.15296078e-02 -8.06370556e-01 -6.50056422e-01
-4.16415930e-01 4.74297792e-01 6.15228832e-01 1.52274922e-01
-7.10791051e-01 6.59313381e-01 5.92243242e+00 1.20490742e+00
-1.74132550e+00 4.68567163e-01 5.47321141e-01 -6.03285193e-01
-1.33832172e-01 -1.53364643e-01 -1.32766247e+00 7.71557152e-01
1.58281565e+00 3.56701761e-01 1.83908314e-01 8.85422766e-01
-5.40765762e-01 -4.14478511e-01 -1.01192248e+00 1.23473728e+00
6.10433817e-02 -1.57388318e+00 7.68015906e-02 3.30317654e-02
6.43026471e-01 8.91842544e-02 6.48627937e-01 4.08564031e-01
-6.29232824e-01 -7.35126913e-01 5.62230349e-01 4.39601690e-01
8.80704820e-01 -9.21184540e-01 6.94922745e-01 1.82170182e-01
-7.14619458e-01 -6.36590362e-01 -7.62451828e-01 -2.41243258e-01
-1.62011534e-01 7.58683681e-01 -6.40658975e-01 -4.87777442e-01
7.22519577e-01 2.02917278e-01 -5.42474568e-01 7.67212629e-01
4.70979154e-01 5.78642249e-01 -5.47374070e-01 -7.13456273e-01
9.25011635e-02 -2.87572737e-03 2.00028434e-01 1.06857264e+00
4.80153561e-01 -3.77834499e-01 -3.98081213e-01 9.59166527e-01
-3.00673664e-01 -1.66939989e-01 -7.68636644e-01 4.99103330e-02
5.65782607e-01 1.31146288e+00 -9.33067918e-01 -3.21362197e-01
-1.46387890e-01 8.54214668e-01 3.79965961e-01 -2.41510496e-01
-8.52993250e-01 -5.58828473e-01 3.61691564e-01 4.31538165e-01
3.13560516e-01 -4.28505331e-01 -6.16767108e-01 -7.58781433e-01
-1.51264817e-01 -4.66282368e-01 1.74020439e-01 -3.98620933e-01
-7.94301748e-01 7.36299813e-01 -4.88378048e-01 -9.98639822e-01
2.30059400e-02 -8.83319438e-01 -3.93981606e-01 3.52964252e-01
-1.03212500e+00 -7.89029598e-01 -3.43303055e-01 3.39969069e-01
3.08506876e-01 -5.16747117e-01 1.02824402e+00 5.01397192e-01
-8.29056799e-01 9.56231058e-01 5.03010035e-01 -2.68445432e-01
5.47787666e-01 -5.96536636e-01 3.54836911e-01 4.57638443e-01
1.36534467e-01 8.11834037e-01 3.00412953e-01 -5.54865062e-01
-2.20295835e+00 -8.73474896e-01 2.79816955e-01 3.02173682e-02
6.00375891e-01 -9.48447287e-01 -1.31193733e+00 1.49819121e-01
2.90868372e-01 1.53789923e-01 5.41998923e-01 -3.29669625e-01
-4.06669378e-01 -8.38301420e-01 -1.14746392e+00 4.60695207e-01
9.18065250e-01 -7.42915094e-01 -1.96586058e-01 1.17410354e-01
6.02380514e-01 -1.95468873e-01 -5.21400869e-01 4.80335027e-01
6.64468169e-01 -1.03725529e+00 6.64671063e-01 -1.81372445e-02
7.75606111e-02 -3.11572164e-01 -2.20003054e-01 -7.49252677e-01
-1.46989390e-01 -8.09385404e-02 -6.11401856e-01 9.55142081e-01
3.36454004e-01 -8.70112658e-01 1.02754247e+00 3.78869355e-01
-7.91633353e-02 -9.13517594e-01 -1.39649618e+00 -1.01846015e+00
1.70793233e-03 -2.91210376e-02 3.30940336e-01 2.25002751e-01
2.00926885e-01 4.57015723e-01 -1.49801359e-01 -1.89783782e-01
6.70723915e-01 -3.53793949e-02 1.95213258e-01 -1.05467367e+00
-3.77315819e-01 -2.84145802e-01 -8.05077791e-01 -9.39523160e-01
3.05877209e-01 -8.73570681e-01 -1.52698040e-01 -7.10335374e-01
3.11942399e-01 -9.08786237e-01 -5.32011807e-01 2.41578877e-01
3.45304459e-01 7.37469673e-01 1.35091366e-02 1.97907642e-01
-4.88916397e-01 7.42275059e-01 6.18032932e-01 -5.25383711e-01
-1.56854898e-01 -5.46134949e-01 -3.19010019e-02 2.54598111e-01
7.49696374e-01 -6.54381514e-01 -4.95610923e-01 -4.15198207e-01
4.57563519e-01 -1.20836839e-01 3.09742987e-01 -1.91549420e+00
9.02341664e-01 5.92879772e-01 3.36380631e-01 -5.56235373e-01
8.16164315e-01 -8.97086143e-01 3.36375505e-01 1.06887341e+00
-2.18237638e-01 1.24878719e-01 5.15506089e-01 4.32506770e-01
1.58032086e-02 -5.41694224e-01 9.39786732e-01 3.02949190e-01
-7.18050539e-01 7.74245262e-02 -5.21027446e-01 -3.26666772e-01
9.55841601e-01 -4.17774886e-01 -7.34771430e-01 2.06587762e-01
-3.82661551e-01 -3.88210177e-01 4.12032008e-01 1.65092736e-01
9.14676189e-01 -1.29955769e+00 -6.66567981e-02 7.23004818e-01
-2.15222329e-01 -3.95182967e-01 7.69542038e-01 8.26107144e-01
-4.90762085e-01 6.55637741e-01 -8.22018266e-01 -8.79005194e-01
-8.63223910e-01 8.76208901e-01 3.53109419e-01 6.45961687e-02
-5.28907001e-01 6.01330996e-01 -1.34511873e-01 -4.44782898e-03
1.80394933e-01 -2.24540785e-01 1.25074536e-01 8.80157202e-02
7.39888847e-01 3.57000768e-01 4.71295118e-01 -1.78100646e-01
-3.60073298e-01 6.47896886e-01 -3.39892536e-01 -7.45764002e-02
1.27401769e+00 1.26184076e-01 -3.09507996e-01 9.22457218e-01
1.24346352e+00 -4.20597464e-01 -9.96753991e-01 -1.41332805e-01
-1.74017072e-01 2.18091860e-01 1.18326768e-01 -2.85427272e-01
-1.18544233e+00 1.24480319e+00 1.30433679e+00 -1.26203299e-01
9.77871716e-01 -5.05772829e-01 1.12333250e+00 9.38078582e-01
9.62764144e-01 -1.41480291e+00 4.00972247e-01 6.91376269e-01
4.65777725e-01 -1.04660034e+00 -1.06157638e-01 2.36261636e-01
3.65445949e-03 1.28173530e+00 1.06209290e+00 -3.73605728e-01
8.22876096e-01 8.38761389e-01 -2.48387575e-01 -9.42101181e-02
-8.66658509e-01 2.58232504e-01 -1.15051925e-01 3.17512840e-01
1.94509700e-01 -7.15682730e-02 -2.84398943e-01 4.63722765e-01
1.78994462e-01 -1.00258164e-01 5.74634671e-01 8.74319255e-01
-7.56355822e-01 -8.28922808e-01 -1.74929142e-01 5.21285892e-01
-8.77101123e-02 -1.85029104e-01 2.51840830e-01 4.71034110e-01
-6.27536476e-02 3.03691387e-01 5.67999065e-01 -6.79958403e-01
-1.33453920e-01 2.14267597e-01 6.32706881e-01 -2.52709359e-01
-8.28389287e-01 -4.21167523e-01 -5.71198583e-01 -2.69902468e-01
8.03118274e-02 -2.26596761e-02 -1.75797796e+00 -3.86089325e-01
-4.26745266e-01 -6.64370805e-02 1.03404737e+00 4.78600770e-01
6.70550168e-01 5.04848003e-01 3.83564085e-01 -7.95122802e-01
-6.37239337e-01 -7.52035737e-01 -7.98702359e-01 -9.05375779e-02
1.21702895e-01 -7.12130725e-01 -1.73075601e-01 -6.90142751e-01] | [8.25271987915039, 2.547013521194458] |
8675eae2-1826-444b-a254-a0950c53070e | spmoe-generate-multiple-pattern-aware-outputs | 2108.07535 | null | https://arxiv.org/abs/2108.07535v2 | https://arxiv.org/pdf/2108.07535v2.pdf | SPMoE: Generate Multiple Pattern-Aware Outputs with Sparse Pattern Mixture of Experts | Many generation tasks follow a one-to-many mapping relationship: each input could be associated with multiple outputs. Existing methods like Conditional Variational AutoEncoder(CVAE) employ a latent variable to model this one-to-many relationship. However, this high-dimensional and dense latent variable lacks explainability and usually leads to poor and uncontrollable generations. In this paper, we innovatively introduce the linguistic concept of pattern to decompose the one-to-many mapping into multiple one-to-one mappings and further propose a model named Sparse Pattern Mixture of Experts(SPMoE). Each one-to-one mapping is associated with a conditional generation pattern and is modeled with an expert in SPMoE. To ensure each language pattern can be exclusively handled with an expert model for better explainability and diversity, a sparse mechanism is employed to coordinate all the expert models in SPMoE. We assess the performance of our SPMoE on the paraphrase generation task and the experiment results prove that SPMoE can achieve a good balance in terms of quality, pattern-level diversity, and corpus-level diversity. | ['Haiqing Chen', 'Wei Zhou', 'Ji Zhang', 'Zhongzhou Zhao', 'Xuming Lin', 'Xintong Bao', 'Shaobo Cui'] | 2021-08-17 | null | null | null | null | ['paraphrase-generation', 'paraphrase-generation'] | ['computer-code', 'natural-language-processing'] | [-1.25344366e-01 1.37517869e-01 -8.15866962e-02 -2.96447277e-01
-4.64491695e-01 -3.75440091e-01 7.63518631e-01 -5.51132381e-01
1.59480527e-01 7.83196449e-01 4.13110793e-01 1.05562165e-01
-1.26197472e-01 -8.97837520e-01 -7.52920270e-01 -7.16625571e-01
5.83792210e-01 8.36370587e-01 -1.36066601e-01 -2.43402645e-01
-6.81466833e-02 -1.26933023e-01 -1.56870973e+00 6.14378989e-01
1.41795349e+00 5.20274162e-01 6.29991829e-01 2.67311513e-01
-4.45071131e-01 6.93707585e-01 -7.24052250e-01 -5.34047246e-01
3.79568823e-02 -8.08097482e-01 -4.19847220e-01 -1.98314749e-02
-1.38150245e-01 9.44804102e-02 -1.33369520e-01 1.12232709e+00
3.89214844e-01 9.60424766e-02 1.06406868e+00 -1.41378403e+00
-1.19791985e+00 9.45732117e-01 -5.11371195e-01 -1.46981671e-01
2.07994148e-01 1.14281781e-01 9.78950500e-01 -1.08420730e+00
5.62232852e-01 1.47131276e+00 3.37944359e-01 5.63733160e-01
-1.33469176e+00 -7.35428512e-01 2.82532781e-01 4.93992046e-02
-1.42421627e+00 -2.19961986e-01 8.35115850e-01 -5.46920419e-01
8.95745635e-01 2.99136732e-02 5.95944405e-01 1.39163506e+00
4.95733351e-01 8.60250890e-01 8.06420326e-01 -3.59086573e-01
1.56226113e-01 5.11229217e-01 3.92646668e-03 6.25752568e-01
7.94860646e-02 -1.61408670e-02 -7.12990761e-01 1.58310414e-03
9.84121740e-01 1.40856951e-03 -3.77711594e-01 -7.21801668e-02
-1.16467309e+00 9.49066818e-01 2.49963045e-01 4.38982546e-01
-4.48364109e-01 2.38626986e-03 -1.10845014e-01 3.54245335e-01
2.03050599e-01 4.52160805e-01 -2.30592519e-01 -1.19138494e-01
-1.10965562e+00 1.89504191e-01 8.12527359e-01 1.28456116e+00
8.92297089e-01 3.49420130e-01 -4.01214778e-01 9.83192027e-01
3.91592085e-01 3.81850481e-01 9.49270368e-01 -8.34371269e-01
4.90711659e-01 7.53972888e-01 4.11172844e-02 -1.13525355e+00
1.97028695e-03 -7.14588583e-01 -1.25377727e+00 -1.92896812e-03
-2.33327523e-01 -1.24998368e-01 -7.86951900e-01 2.05255580e+00
-2.81613432e-02 6.49918523e-03 1.23057641e-01 9.33345854e-01
7.00488150e-01 1.15923810e+00 -2.40985882e-02 -3.75864089e-01
1.06651998e+00 -1.30332971e+00 -8.47244680e-01 -4.18663323e-01
1.24358185e-01 -7.38198280e-01 1.20077097e+00 4.40436840e-01
-1.16091096e+00 -9.19123530e-01 -8.92209411e-01 2.01194882e-02
-1.32972300e-01 3.72652769e-01 2.69451499e-01 3.16969931e-01
-7.99071908e-01 5.18572509e-01 -5.57479739e-01 9.81048197e-02
-3.37358192e-02 2.06434637e-01 -2.89640963e-01 8.19017738e-02
-1.53204834e+00 8.82131279e-01 4.55856591e-01 4.18326333e-02
-8.19092631e-01 -5.69284976e-01 -6.32581711e-01 3.02315772e-01
1.43231601e-01 -1.14974725e+00 8.54945183e-01 -9.56805110e-01
-1.44715989e+00 4.68559474e-01 -5.11565387e-01 -8.67338106e-02
4.21627074e-01 1.11797459e-01 -6.21919990e-01 -1.51503786e-01
4.07226294e-01 6.63444400e-01 1.00115037e+00 -1.44208503e+00
-4.41563070e-01 2.62607597e-02 -1.65915012e-01 3.10879529e-01
-4.10339832e-01 -1.64939672e-01 -4.26009238e-01 -9.74407554e-01
1.46675140e-01 -7.50641167e-01 7.10847899e-02 -3.42196405e-01
-2.90862560e-01 -3.65838379e-01 3.24450344e-01 -6.11724198e-01
1.56206787e+00 -2.04668283e+00 8.62469554e-01 -2.71644928e-02
2.82798499e-01 -3.02091241e-03 -1.37465909e-01 4.65274125e-01
2.93027572e-02 2.19350100e-01 -2.50122935e-01 -6.21929169e-01
2.73143589e-01 4.94105875e-01 -3.61206859e-01 -1.18617401e-01
2.73569465e-01 1.01843381e+00 -8.70084882e-01 -6.32679641e-01
-1.19208641e-01 3.59210014e-01 -6.82823002e-01 4.64413792e-01
-3.27162594e-01 2.07173824e-01 -4.90400374e-01 5.16299367e-01
5.55285394e-01 -5.92004180e-01 5.47554754e-02 -3.76225710e-02
7.75359347e-02 -1.67665333e-01 -1.14896607e+00 1.73282349e+00
-6.13781035e-01 2.41125956e-01 -6.28096610e-02 -6.75113142e-01
1.14512801e+00 4.27214354e-01 6.26666769e-02 -3.65845621e-01
-6.72098398e-02 3.56405228e-01 2.97117680e-02 -4.02315140e-01
5.09683430e-01 -3.89595747e-01 -1.94295004e-01 6.15246892e-01
3.06220204e-01 1.23878263e-01 1.72222167e-01 2.81143248e-01
6.17893338e-01 2.22658113e-01 1.06862389e-01 -5.15714109e-01
3.83238733e-01 -1.56721383e-01 1.10062873e+00 6.55815721e-01
1.27317831e-01 6.38670266e-01 5.04742920e-01 -2.91974783e-01
-1.11098945e+00 -1.14203286e+00 1.80616885e-01 7.55988657e-01
3.15107197e-01 -3.36438447e-01 -7.43955910e-01 -2.21538708e-01
-2.19541773e-01 1.11774755e+00 -6.71619058e-01 -3.73107284e-01
-3.73527199e-01 -4.69783455e-01 5.06667852e-01 4.58662242e-01
5.46825767e-01 -1.30116487e+00 -1.71778724e-01 5.37356973e-01
-3.94488394e-01 -5.49282312e-01 -6.86684489e-01 5.46586001e-03
-4.05328304e-01 -6.32682145e-01 -7.63106465e-01 -9.37507987e-01
7.04111993e-01 -1.07327513e-01 1.29282463e+00 -1.99160740e-01
1.94224223e-01 -1.50087595e-01 -4.53774512e-01 -3.19113761e-01
-7.17928350e-01 1.63257450e-01 3.40412378e-01 1.32647648e-01
4.05429035e-01 -8.69970024e-01 -4.28457916e-01 3.42080325e-01
-9.25641894e-01 4.25752342e-01 8.19528461e-01 1.01754737e+00
6.10004425e-01 1.61656395e-01 6.72855198e-01 -6.10861182e-01
1.08495080e+00 -7.46742666e-01 -1.85546473e-01 6.15962565e-01
-5.98178566e-01 3.58030170e-01 9.74441111e-01 -7.50572860e-01
-1.21977043e+00 -1.93118826e-01 2.00180620e-01 -8.92007649e-01
-9.72623974e-02 7.30587840e-01 -5.26996315e-01 5.77304542e-01
5.79790950e-01 6.58335686e-01 5.80218546e-02 -3.94259661e-01
4.31685984e-01 6.26068413e-01 3.86151373e-01 -7.15084434e-01
7.94133782e-01 -1.48558334e-01 -3.60657960e-01 -2.83895046e-01
-6.77667320e-01 1.75100401e-01 -2.78919756e-01 -1.63288698e-01
9.50354874e-01 -1.07440197e+00 -1.61161765e-01 4.27952796e-01
-1.41146505e+00 -1.23795345e-01 -2.36831456e-01 2.86041319e-01
-4.08109754e-01 5.56510761e-02 -5.51815569e-01 -5.89710653e-01
-2.28502959e-01 -1.32916951e+00 8.41300666e-01 4.16875690e-01
-4.51457441e-01 -9.54077303e-01 -2.78038252e-02 1.51552916e-01
5.58908403e-01 -1.78728595e-01 1.10221946e+00 -4.84543234e-01
-5.89310944e-01 1.01688340e-01 4.52979729e-02 2.24496871e-01
1.03974782e-01 3.08107827e-02 -5.70869088e-01 -6.38686493e-02
3.92123014e-01 -2.59667218e-01 7.79149055e-01 2.56653816e-01
9.51038182e-01 -4.64798450e-01 -2.86865056e-01 4.90334928e-01
1.22604215e+00 1.22390091e-01 5.68140566e-01 1.47086978e-02
8.34453046e-01 6.93489611e-01 3.90484601e-01 3.34840596e-01
5.19851208e-01 5.53782880e-01 5.63729927e-02 6.42152950e-02
-1.15336031e-02 -6.97803378e-01 4.19157475e-01 1.37752259e+00
5.77930100e-02 -5.68402231e-01 -5.67146242e-01 5.85138857e-01
-2.05539203e+00 -1.05361116e+00 2.96763182e-02 1.63663208e+00
9.48032200e-01 2.37413822e-03 -1.36230126e-01 -1.62163213e-01
1.06340921e+00 3.69842462e-02 -4.67786342e-01 -3.48297447e-01
-4.58673894e-01 -1.98264673e-01 -3.18549842e-01 5.04900992e-01
-4.98804122e-01 9.99941170e-01 6.10504580e+00 1.31051624e+00
-8.19902956e-01 1.00928113e-01 5.38819492e-01 -1.87349677e-01
-1.05266571e+00 5.00620008e-02 -8.78384113e-01 1.05248749e+00
4.78239566e-01 -3.65546465e-01 6.56396449e-01 7.69239128e-01
1.01755105e-01 2.75898933e-01 -1.08125186e+00 9.39795911e-01
1.30864516e-01 -1.33136845e+00 6.35939062e-01 1.28459349e-01
1.03526545e+00 -6.11137033e-01 4.08208035e-02 4.63654369e-01
2.40303174e-01 -1.11169040e+00 9.43273664e-01 8.89308810e-01
6.05447829e-01 -8.42770100e-01 4.21863288e-01 9.19259369e-01
-1.20582497e+00 -3.45608518e-02 -6.21657252e-01 -4.54999739e-03
4.31232721e-01 6.02640569e-01 -9.93668810e-02 7.76171565e-01
4.19569224e-01 5.36316335e-01 -1.91671163e-01 4.04265255e-01
-4.58097070e-01 4.30603921e-01 -6.70777708e-02 -2.19466954e-01
1.73984975e-01 -5.52109599e-01 7.42014050e-01 9.09275234e-01
7.63884962e-01 6.46642372e-02 1.68417886e-01 1.66307902e+00
6.69177668e-03 -3.14526148e-02 -4.57665771e-01 -6.19796850e-02
8.95515323e-01 8.13151896e-01 -2.17188507e-01 -5.00024915e-01
-1.97624937e-01 1.10116363e+00 3.73220384e-01 4.46893752e-01
-9.74916160e-01 -3.12744945e-01 3.46239150e-01 -2.88660377e-02
3.93920392e-01 6.46244735e-02 -3.85187358e-01 -1.37970805e+00
2.27880582e-01 -1.08055103e+00 4.67012413e-02 -9.90496099e-01
-1.72797751e+00 1.03882277e+00 -5.90367392e-02 -1.25392616e+00
-5.13280809e-01 -2.04091683e-01 -8.22513044e-01 1.28035712e+00
-1.07997632e+00 -1.32803321e+00 -1.73377052e-01 6.61399961e-01
6.92746162e-01 -8.71370018e-01 9.00768936e-01 4.88316827e-02
-6.73206985e-01 6.84861958e-01 -5.87825477e-03 -2.04003200e-01
4.74638879e-01 -1.08919466e+00 1.70785218e-01 1.01563454e+00
2.34715730e-01 9.62881088e-01 7.76532829e-01 -7.98110425e-01
-1.13511312e+00 -1.02184737e+00 1.18525946e+00 -3.50217998e-01
4.73725319e-01 -2.05461428e-01 -9.26284790e-01 5.69180131e-01
4.76507872e-01 -5.71849048e-01 6.26069605e-01 1.62987605e-01
-8.45375210e-02 1.12224244e-01 -8.97204101e-01 6.67977631e-01
8.94529641e-01 -4.63893741e-01 -9.35334742e-01 1.31564751e-01
9.47882950e-01 -2.03868315e-01 -6.42652094e-01 2.10085228e-01
3.25385571e-01 -9.28035319e-01 4.71901029e-01 -4.06631529e-01
1.15942013e+00 -4.96868342e-01 -1.09838329e-01 -1.64913046e+00
-7.96221375e-01 -5.62076509e-01 -3.64457935e-01 1.43229926e+00
5.92499852e-01 -6.27607405e-01 6.05857551e-01 6.35049760e-01
-3.13896328e-01 -8.45941246e-01 -8.03065002e-01 -8.56215596e-01
3.09792787e-01 9.60845277e-02 1.00011694e+00 1.04249775e+00
-9.75913480e-02 5.46003401e-01 -8.30562890e-01 6.08672202e-02
5.03895402e-01 5.11897504e-01 5.29378891e-01 -1.02934134e+00
-9.15782094e-01 -5.45125604e-01 1.43866673e-01 -1.25224900e+00
2.53470093e-01 -9.47477221e-01 7.09948763e-02 -1.59626293e+00
4.16192293e-01 -3.17029804e-01 -2.08882928e-01 2.55322665e-01
-4.17791128e-01 -1.46455526e-01 9.25938319e-03 5.47925353e-01
-2.88740844e-01 9.86151993e-01 1.32942939e+00 -1.12569995e-01
-3.15989643e-01 -2.60259420e-01 -8.74250948e-01 4.79243755e-01
6.71847701e-01 -5.77118456e-01 -7.41416097e-01 -8.36529315e-01
2.98603594e-01 2.97885239e-01 1.05424322e-01 -8.68252873e-01
5.71659923e-01 -1.97627038e-01 3.68476480e-01 -5.07696211e-01
4.25316751e-01 -5.67164958e-01 8.33599746e-01 3.00654143e-01
-2.51907796e-01 1.66966002e-02 -1.90739721e-01 6.46904647e-01
-6.19706094e-01 -2.53336728e-01 4.02694553e-01 -3.24358374e-01
-4.82135713e-01 2.45394021e-01 -3.89158189e-01 -6.54707775e-02
9.92575109e-01 -2.37271041e-01 -2.67293960e-01 -4.69363362e-01
-7.34055519e-01 5.55279255e-01 3.81812304e-01 6.15406454e-01
5.87726593e-01 -1.68875384e+00 -9.29383814e-01 4.77902591e-01
1.25415549e-02 4.11550924e-02 4.60042208e-01 2.58060902e-01
-2.62083530e-01 2.68646210e-01 -3.33619833e-01 -4.50866848e-01
-7.89093316e-01 5.44279635e-01 3.76700938e-01 -5.55051446e-01
-3.09763044e-01 1.23421097e+00 4.26014125e-01 -4.93522435e-01
-1.25839347e-02 1.18243732e-01 -1.38590768e-01 4.93777469e-02
2.27574795e-01 1.38055816e-01 -5.30697227e-01 -5.40685594e-01
-1.36091828e-01 5.12887120e-01 -6.93256333e-02 -2.08353356e-01
1.19438112e+00 -2.69869603e-02 -2.60510445e-01 5.74893355e-01
8.01652074e-01 -1.19510628e-01 -1.10874653e+00 -1.39142841e-01
-5.48310637e-01 -3.01294804e-01 -1.24958262e-01 -7.32428551e-01
-9.50620532e-01 9.70140159e-01 2.50502396e-02 2.22777963e-01
9.82300937e-01 -1.25715658e-01 7.76076615e-01 1.21638728e-02
4.98272568e-01 -1.13897359e+00 3.09462070e-01 4.14099783e-01
1.26700509e+00 -9.27916944e-01 -3.79387110e-01 -2.52723545e-01
-1.17311251e+00 7.07413852e-01 8.56925011e-01 -1.31850690e-01
5.34734190e-01 1.43518075e-01 -8.22918341e-02 -4.18860950e-02
-1.16570961e+00 2.68361121e-01 4.04396057e-01 3.95652443e-01
2.02332467e-01 2.09363431e-01 -2.17541352e-01 1.21356905e+00
-4.26920176e-01 -9.31633562e-02 1.29211113e-01 4.49804157e-01
-4.02270526e-01 -1.49404657e+00 -9.72032547e-02 4.10448283e-01
3.72218527e-02 -2.17776060e-01 -3.31080437e-01 4.84669030e-01
3.44505161e-01 9.93156075e-01 9.95705277e-02 -5.90097308e-01
1.25720337e-01 1.25075340e-01 3.39428872e-01 -6.55605018e-01
-5.41592181e-01 2.73848146e-01 -2.30853170e-01 -3.36426198e-01
-2.08069474e-01 -4.03973103e-01 -1.02253520e+00 -3.39249760e-01
-2.98473507e-01 3.17146927e-01 2.57970631e-01 9.36023593e-01
5.85651517e-01 7.35090911e-01 6.26912773e-01 -4.22308773e-01
-6.74884737e-01 -1.10093248e+00 -7.86849380e-01 4.12577093e-01
-5.68897910e-02 -6.57972217e-01 -4.57704246e-01 4.42037210e-02] | [11.766551971435547, 9.136078834533691] |
4aba614b-6b24-4fda-a2ec-8e02b9bc1a0c | read-attend-and-code-pushing-the-limits-of | 2107.10650 | null | https://arxiv.org/abs/2107.10650v1 | https://arxiv.org/pdf/2107.10650v1.pdf | Read, Attend, and Code: Pushing the Limits of Medical Codes Prediction from Clinical Notes by Machines | Prediction of medical codes from clinical notes is both a practical and essential need for every healthcare delivery organization within current medical systems. Automating annotation will save significant time and excessive effort spent by human coders today. However, the biggest challenge is directly identifying appropriate medical codes out of several thousands of high-dimensional codes from unstructured free-text clinical notes. In the past three years, with Convolutional Neural Networks (CNN) and Long Short-Term Memory (LTSM) networks, there have been vast improvements in tackling the most challenging benchmark of the MIMIC-III-full-label inpatient clinical notes dataset. This progress raises the fundamental question of how far automated machine learning (ML) systems are from human coders' working performance. We assessed the baseline of human coders' performance on the same subsampled testing set. We also present our Read, Attend, and Code (RAC) model for learning the medical code assignment mappings. By connecting convolved embeddings with self-attention and code-title guided attention modules, combined with sentence permutation-based data augmentations and stochastic weight averaging training, RAC establishes a new state of the art (SOTA), considerably outperforming the current best Macro-F1 by 18.7%, and reaches past the human-level coding baseline. This new milestone marks a meaningful step toward fully autonomous medical coding (AMC) in machines reaching parity with human coders' performance in medical code prediction. | ['Varun Ganapathi', 'Byung-Hak Kim'] | 2021-07-10 | null | null | null | null | ['multi-label-classification-of-biomedical', 'medical-code-prediction'] | ['medical', 'medical'] | [ 4.51587558e-01 5.54992676e-01 -1.74807698e-01 -5.09636402e-01
-1.27086961e+00 -2.24286020e-01 6.44471645e-02 9.06607926e-01
-4.67587888e-01 4.27958548e-01 5.96538901e-01 -6.47842467e-01
-2.10897401e-01 -3.43581796e-01 -2.40154102e-01 -2.43277088e-01
-2.45982707e-01 9.91467535e-01 -3.97324979e-01 -2.81878300e-02
6.41544303e-03 4.45724465e-02 -9.86473501e-01 8.04859936e-01
9.31366205e-01 9.18486714e-01 4.66515534e-02 1.07584453e+00
-2.24256247e-01 1.41121328e+00 -1.36160105e-01 -6.00217283e-01
-7.04726875e-02 -6.02729499e-01 -1.11872196e+00 -4.29286391e-01
2.38571689e-01 -3.43395144e-01 -2.93308377e-01 8.31868112e-01
7.66629815e-01 -3.46267760e-01 6.73486590e-01 -4.98626977e-01
-1.30810833e+00 7.23972738e-01 -8.00748989e-02 1.91050649e-01
4.84867036e-01 1.63187250e-01 1.12628651e+00 -8.22996616e-01
5.52470982e-01 6.62571311e-01 1.45899498e+00 8.52188945e-01
-1.19149232e+00 -3.64278167e-01 -4.31548566e-01 -1.46550640e-01
-1.38922334e+00 -4.48201537e-01 -2.83498764e-02 -1.07301581e+00
1.59076715e+00 2.93089032e-01 4.27680284e-01 1.03827715e+00
2.99735636e-01 5.02197921e-01 4.06308532e-01 -4.03894335e-01
1.71868466e-02 1.71643779e-01 2.76627094e-01 1.09335005e+00
2.14767352e-01 -7.38909245e-02 7.46404473e-03 -5.84702253e-01
4.27783430e-01 6.04962349e-01 -4.10692751e-01 -1.95009291e-01
-1.77884638e+00 9.66214120e-01 3.72132272e-01 5.10655999e-01
-2.98864841e-01 1.08087555e-01 7.61564195e-01 2.10259527e-01
3.75543147e-01 1.06695712e+00 -8.31646621e-01 -6.24735177e-01
-1.00240338e+00 7.54344137e-03 9.36346829e-01 9.60958183e-01
2.62506157e-02 -3.74098778e-01 -5.75611353e-01 1.03399396e+00
-5.58164604e-02 2.98198819e-01 9.47776496e-01 -4.09202337e-01
6.61568463e-01 7.85028040e-01 -1.07002392e-01 -8.71006668e-01
-8.68298650e-01 -6.31111860e-01 -1.18788064e+00 -3.14635664e-01
2.05590591e-01 -3.35129052e-01 -9.00574565e-01 1.32246304e+00
-3.58585745e-01 -3.14661324e-01 8.19536299e-02 3.72216880e-01
8.10807168e-01 3.02135706e-01 2.02737600e-01 7.99183324e-02
1.46409655e+00 -8.67094100e-01 -6.98565364e-01 -5.49280234e-02
1.41149211e+00 -5.56302607e-01 7.04482377e-01 6.32372871e-02
-1.00599694e+00 -5.84001541e-01 -6.85011864e-01 -2.92770982e-01
-2.61087805e-01 2.65078902e-01 5.26736856e-01 4.96709019e-01
-1.20087898e+00 5.15456080e-01 -9.39460218e-01 -4.66215223e-01
7.72576511e-01 6.61109686e-01 -4.40602541e-01 -1.92956328e-01
-1.06714499e+00 1.09407520e+00 1.52380571e-01 -1.92877829e-01
-5.55413127e-01 -1.20640659e+00 -9.60934758e-01 3.54643464e-01
-2.94276655e-01 -7.92652607e-01 1.28559101e+00 -8.11641634e-01
-6.74622834e-01 1.43828738e+00 3.65177765e-02 -7.75523067e-01
5.23648143e-01 -1.21670865e-01 -6.44701958e-01 -2.08815783e-02
1.22861445e-01 7.36746848e-01 4.53069769e-02 -6.79997146e-01
-3.48457456e-01 -1.89328298e-01 -3.89566332e-01 -1.67158648e-01
-5.93638480e-01 3.40332650e-02 8.10007378e-02 -6.65894210e-01
-4.43460792e-02 -1.07259572e+00 -3.25116932e-01 -1.13855652e-03
-3.85502398e-01 -1.72746450e-01 -8.28838721e-02 -8.97264361e-01
1.63428175e+00 -2.46281815e+00 -1.91358656e-01 -4.50923555e-02
5.85988104e-01 3.46877754e-01 -1.26692191e-01 5.93651772e-01
-4.99765158e-01 2.16373339e-01 -5.95808029e-01 -3.30149025e-01
-1.89849570e-01 1.72912236e-02 -8.65725651e-02 2.95707971e-01
3.80688667e-01 1.16081989e+00 -1.03356326e+00 -4.85788852e-01
-1.78108051e-01 3.77797574e-01 -1.10926151e+00 4.02464092e-01
2.60205805e-01 3.31648916e-01 -1.72199741e-01 5.56362867e-01
1.95043072e-01 -8.98417771e-01 4.00475077e-02 2.12742612e-01
1.75472885e-01 2.13459745e-01 -3.78737658e-01 2.05217242e+00
-5.31707108e-01 4.98312294e-01 -2.54934967e-01 -9.58856881e-01
8.77068520e-01 5.79763949e-01 9.29045141e-01 -2.74325699e-01
1.30753472e-01 4.74823833e-01 4.60068017e-01 -1.13583374e+00
1.43470302e-01 -2.45019883e-01 -1.52322501e-01 2.90763021e-01
1.16135858e-01 1.17701128e-01 -2.51315325e-01 8.19549561e-02
1.91621900e+00 -4.49716955e-01 6.02813005e-01 -5.28929174e-01
3.20261985e-01 2.49466181e-01 3.50052476e-01 7.52125382e-01
-4.08060610e-01 1.14140797e+00 5.45623243e-01 -1.09734571e+00
-1.04232252e+00 -7.94851661e-01 -5.59990406e-01 9.06153619e-01
-6.71868920e-01 -2.40790531e-01 -6.66596532e-01 -8.34690452e-01
3.02937359e-01 4.92811441e-01 -1.03895330e+00 -2.43466631e-01
-4.94788587e-01 -7.94629693e-01 8.94117892e-01 7.02287674e-01
2.33120322e-02 -1.02674210e+00 -7.99682915e-01 4.94919389e-01
-2.12640345e-01 -9.62926984e-01 -7.08807826e-01 5.66626668e-01
-6.42126620e-01 -1.27326524e+00 -9.70489740e-01 -1.03853226e+00
8.27518940e-01 -4.08803076e-01 1.54777360e+00 4.44182813e-01
-7.70932853e-01 1.68269485e-01 -4.39558148e-01 -4.49519992e-01
-7.80373693e-01 5.02900243e-01 -1.23704180e-01 -2.45196193e-01
7.72444308e-01 -1.11465536e-01 -7.02559352e-01 -3.78597319e-01
-6.51186645e-01 1.44101039e-01 7.34668911e-01 1.13654113e+00
2.49329999e-01 -8.10388088e-01 6.84059143e-01 -1.22299302e+00
6.92848921e-01 -7.39471436e-01 -3.86564690e-03 2.14306250e-01
-8.74799311e-01 1.68156818e-01 7.20227659e-01 7.90812969e-02
-4.38195735e-01 2.01199725e-01 -6.26979589e-01 -1.08627342e-01
-1.64067104e-01 3.94643486e-01 6.21885538e-01 3.10807973e-01
9.56448913e-01 3.09001785e-02 2.52726346e-01 -2.57105678e-01
-2.58770436e-01 1.16524994e+00 3.66909355e-01 -1.36564955e-01
1.97553471e-01 1.72885105e-01 -3.52218956e-01 -7.59306252e-02
-1.12098253e+00 -6.09758377e-01 -7.39688516e-01 3.61562937e-01
1.50131977e+00 -9.28250372e-01 -7.38084555e-01 -1.21726587e-01
-1.25107193e+00 -2.80891061e-01 -3.61832023e-01 5.23591340e-01
-3.84249687e-01 -5.50153293e-02 -8.26112151e-01 -3.22061837e-01
-5.27365983e-01 -1.25989628e+00 1.08316267e+00 -4.26473439e-01
-9.84356403e-01 -1.16589630e+00 4.36590999e-01 4.20283973e-01
6.59446716e-01 4.68455583e-01 1.27417684e+00 -1.30030012e+00
4.04759943e-02 -4.23419207e-01 -4.23309207e-01 4.47948873e-01
3.35762084e-01 -4.38726872e-01 -9.53911960e-01 -1.09986983e-01
-2.40662873e-01 -4.29110855e-01 9.83317733e-01 4.07768041e-01
1.50063491e+00 -2.04630077e-01 -4.70801622e-01 8.74055028e-01
1.38384843e+00 3.15197736e-01 1.95937946e-01 2.68370677e-02
8.75357211e-01 3.15754414e-01 -8.55016187e-02 4.53818321e-01
6.12115622e-01 3.18974137e-01 2.46806920e-01 -2.91945249e-01
-4.85049747e-02 -9.51399505e-02 -1.31460249e-01 1.26384377e+00
8.67614150e-02 2.37866595e-01 -1.81014514e+00 7.37711906e-01
-1.62329316e+00 -8.56172085e-01 -1.73690423e-01 1.96771657e+00
1.11882389e+00 -1.56475618e-01 -4.17156368e-01 -1.09081604e-01
6.23103380e-01 -3.83223236e-01 -5.86714447e-01 -8.41337204e-01
2.87530571e-01 1.61114633e-01 5.68639398e-01 2.71631092e-01
-1.28052974e+00 1.22711390e-01 6.44030237e+00 3.88941169e-01
-6.47554755e-01 3.77301097e-01 1.17227256e+00 -1.53280119e-03
-1.28669009e-01 -7.12642431e-01 -5.18612981e-01 6.56702459e-01
1.49940908e+00 2.68659920e-01 3.40751916e-01 1.01888561e+00
-1.03963315e-01 3.44959944e-01 -1.63065660e+00 1.28023207e+00
3.73593956e-01 -1.61832941e+00 -1.37329876e-01 2.16121182e-01
1.11124349e+00 5.61796188e-01 1.20585030e-02 5.77232599e-01
5.02120674e-01 -1.37624872e+00 1.94709584e-01 5.67240477e-01
1.43996215e+00 -4.82617736e-01 1.38504863e+00 1.13614917e-01
-7.77242661e-01 -6.81186199e-01 -1.46096811e-01 -4.75247530e-03
-1.00517616e-01 6.17078602e-01 -1.03980684e+00 1.34859905e-01
4.94501621e-01 7.92454004e-01 -5.85254610e-01 8.38437080e-01
5.20119309e-01 4.38560635e-01 4.31700826e-01 1.24873646e-01
3.22501808e-01 4.78833437e-01 -1.51259467e-01 1.73653007e+00
4.29320335e-01 2.75331438e-01 1.38574913e-01 6.63515568e-01
-4.61539328e-01 1.08840197e-01 -6.83991134e-01 -1.92122236e-01
1.57561287e-01 9.69788730e-01 -6.01378560e-01 -5.67759216e-01
-5.75828314e-01 9.69257712e-01 4.15221661e-01 -1.66389525e-01
-8.16782176e-01 -5.60299397e-01 6.89995050e-01 2.87548061e-02
3.12040970e-02 2.96629965e-01 -5.79926968e-01 -1.00320792e+00
-3.06283206e-01 -1.07099116e+00 5.63391447e-01 -4.00712937e-01
-1.35613203e+00 8.79231393e-01 -7.42430925e-01 -1.43999350e+00
-2.74502665e-01 -6.49920762e-01 -3.74621563e-02 6.92330182e-01
-1.36660516e+00 -7.72897482e-01 -3.69657546e-01 3.17689508e-01
5.71663737e-01 -3.56002092e-01 1.69470000e+00 6.71032965e-01
-1.53534621e-01 8.19105387e-01 3.82530123e-01 8.86578918e-01
7.09619284e-01 -1.27660525e+00 4.33232874e-01 2.26243868e-01
-9.31377634e-02 6.11279130e-01 2.10414395e-01 -3.35001081e-01
-1.07759345e+00 -1.36683607e+00 1.35357618e+00 -1.02954662e+00
5.64334214e-01 -3.06515783e-01 -9.06739533e-01 7.09271908e-01
2.58768331e-02 3.33058536e-01 1.24624050e+00 4.57161525e-03
-2.69174039e-01 9.80010256e-02 -1.14952135e+00 5.06737903e-02
1.12977540e+00 -7.02076614e-01 -6.24351203e-01 6.36861503e-01
9.44011927e-01 -5.29641390e-01 -1.17153370e+00 4.89197314e-01
4.62056667e-01 -8.76298547e-01 8.03007722e-01 -8.47706020e-01
1.12850821e+00 3.07604939e-01 1.40839051e-02 -1.24622321e+00
-5.84857166e-01 -3.18074107e-01 2.82977909e-01 7.15870678e-01
6.70481980e-01 -4.13401544e-01 4.72158134e-01 9.50417280e-01
-6.22449994e-01 -1.24116695e+00 -7.86063433e-01 -3.58057976e-01
3.14430028e-01 -2.07621112e-01 5.08538783e-01 1.34822643e+00
5.10151148e-01 -6.13281615e-02 -1.59797177e-01 -1.98610857e-01
9.58887301e-03 -1.15784831e-01 9.66361091e-02 -1.36753309e+00
-4.38264787e-01 -5.29675424e-01 -5.97675383e-01 -4.26910818e-01
-3.12764496e-02 -1.56403112e+00 1.68978468e-01 -1.93908191e+00
7.07447231e-01 -4.27450120e-01 -6.58310473e-01 6.72034442e-01
-2.73893416e-01 1.19345114e-01 4.44696769e-02 2.82511413e-01
-6.52377844e-01 -5.44511117e-02 9.16543186e-01 -2.84376532e-01
3.17374319e-02 -1.61202475e-01 -8.89187753e-01 5.53805411e-01
7.09139168e-01 -7.57977605e-01 -3.14398222e-02 -8.03511262e-01
5.23128092e-01 3.08859855e-01 4.25386019e-02 -1.20152700e+00
1.82230204e-01 3.98222893e-01 3.96212161e-01 4.67179902e-02
-2.08695531e-01 -9.58042324e-01 -2.04309240e-01 1.16830778e+00
-1.09519315e+00 5.60271680e-01 2.04842731e-01 3.73712927e-01
-2.59554386e-01 -2.95998722e-01 7.05867290e-01 -3.35032880e-01
-1.12425320e-01 3.21753740e-01 -4.27347243e-01 4.75627840e-01
7.64666915e-01 -1.19235277e-01 -2.23658569e-02 -4.59398143e-03
-9.69326198e-01 1.27679035e-01 6.57512620e-02 4.00451809e-01
5.62820554e-01 -1.13020027e+00 -1.00813150e+00 4.17866528e-01
6.67496562e-01 -2.06501573e-01 3.71460080e-01 1.09451509e+00
-8.27174842e-01 8.19020510e-01 -1.76808387e-02 -6.06271207e-01
-1.07411516e+00 6.18630588e-01 2.78458059e-01 -5.96855581e-01
-6.67010665e-01 1.16703963e+00 -1.34363547e-01 -7.54107773e-01
1.86982468e-01 -8.89834702e-01 -2.34598890e-01 2.41386984e-03
7.09836781e-01 7.44825527e-02 3.45461488e-01 -2.54216284e-01
-4.56399590e-01 4.17814612e-01 9.20048356e-03 4.89119798e-01
1.54787731e+00 2.25404605e-01 -1.84355170e-01 4.40827936e-01
1.63735580e+00 -1.75700247e-01 -6.27143860e-01 9.09711272e-02
3.00965816e-01 -7.05219582e-02 -2.89043814e-01 -9.99706924e-01
-7.75914729e-01 1.09111881e+00 8.33370447e-01 2.74592638e-01
6.80966139e-01 4.06771824e-02 9.27179813e-01 5.03108084e-01
4.12062407e-02 -8.82611215e-01 -8.69900212e-02 5.61183631e-01
7.26149559e-01 -1.65485084e+00 -5.15052915e-01 3.40538681e-01
-8.53179693e-01 1.24540019e+00 2.88902670e-01 -1.01115964e-02
8.58935773e-01 4.29225653e-01 2.93107182e-01 -2.90576339e-01
-8.38111222e-01 1.94160387e-01 1.93091720e-01 4.65554804e-01
1.15056062e+00 2.65563965e-01 7.02541471e-02 7.07289755e-01
-1.41047999e-01 1.08954675e-01 4.51129436e-01 5.22271693e-01
-3.71190518e-01 -7.90846884e-01 -1.04709499e-01 1.12641811e+00
-8.98115575e-01 -6.53296769e-01 1.08026139e-01 4.18976396e-01
3.95485818e-01 7.23551273e-01 2.87628710e-01 -3.97614062e-01
3.12600136e-01 5.77010155e-01 -1.62465513e-01 -1.20484316e+00
-1.11694312e+00 -6.30362391e-01 -1.06139658e-02 -4.60654527e-01
-1.25062257e-01 -6.63824260e-01 -1.54034591e+00 -1.18926816e-01
6.17393926e-02 1.90933257e-01 2.87998050e-01 5.67805529e-01
8.35396051e-01 1.02050221e+00 2.04440206e-01 -2.91277438e-01
-7.45459020e-01 -1.02944660e+00 -1.31564528e-01 6.00798965e-01
8.13355446e-01 -5.57361878e-02 -1.42875090e-01 2.95756996e-01] | [8.01669692993164, 6.812239646911621] |
889bd405-f6cb-41ee-b6fc-97680d4a7be1 | generative-ai-meets-3d-a-survey-on-text-to-3d | 2305.06131 | null | https://arxiv.org/abs/2305.06131v2 | https://arxiv.org/pdf/2305.06131v2.pdf | Generative AI meets 3D: A Survey on Text-to-3D in AIGC Era | Generative AI (AIGC, a.k.a. AI generated content) has made remarkable progress in the past few years, among which text-guided content generation is the most practical one since it enables the interaction between human instruction and AIGC. Due to the development in text-to-image as well 3D modeling technologies (like NeRF), text-to-3D has become a newly emerging yet highly active research field. Our work conducts the first yet comprehensive survey on text-to-3D to help readers interested in this direction quickly catch up with its fast development. First, we introduce 3D data representations, including both Euclidean data and non-Euclidean data. On top of that, we introduce various foundation technologies as well as summarize how recent works combine those foundation technologies to realize satisfactory text-to-3D. Moreover, we summarize how text-to-3D technology is used in various applications, including avatar generation, texture generation, shape transformation, and scene generation. | ['Choong Seon Hong', 'Sung-Ho Bae', 'Yang Yang', 'Francois Rameau', 'Lik-Hang Lee', 'Atish Waghwase', 'Chaoning Zhang', 'Chenghao Li'] | 2023-05-10 | null | null | null | null | ['texture-synthesis', 'scene-generation', 'text-to-3d'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.83607534e-01 2.43828207e-01 2.18814224e-01 -1.64547205e-01
-4.92595851e-01 -3.89730364e-01 8.36514533e-01 -2.15386555e-01
1.89106002e-01 5.97216666e-01 4.54953074e-01 -1.13948591e-01
1.54781893e-01 -1.16614699e+00 -8.93455684e-01 -6.89659059e-01
4.69054550e-01 6.59563899e-01 2.65646875e-01 -7.55015433e-01
4.24444050e-01 5.97953081e-01 -2.07087541e+00 1.64039284e-01
1.18569255e+00 7.71182477e-01 1.97660238e-01 6.61048114e-01
-7.98551261e-01 6.50850236e-01 -7.67493546e-01 -6.16787851e-01
-7.32175913e-03 -5.88694572e-01 -5.58823347e-01 6.52168617e-02
6.09032691e-01 -2.35509843e-01 -4.07744050e-01 8.99483562e-01
8.06934178e-01 3.37349892e-01 9.60839987e-01 -1.37094140e+00
-1.06674135e+00 6.13247931e-01 -4.06804055e-01 -5.94680667e-01
8.29458535e-01 -3.37920070e-01 2.71183282e-01 -1.23390925e+00
9.62157786e-01 1.19004679e+00 3.55836600e-01 6.60788715e-01
-6.21184349e-01 -3.15628797e-01 -6.26598921e-05 2.91511506e-01
-1.14371359e+00 5.45682348e-02 1.19418764e+00 -3.99901450e-01
6.66251123e-01 3.75521481e-01 1.29459381e+00 1.18897557e+00
5.02029240e-01 1.35789502e+00 1.06121886e+00 -1.07234573e+00
5.96584938e-02 -2.80510578e-02 -2.12293953e-01 7.32504070e-01
-9.13256705e-02 -1.89665198e-01 -7.15687811e-01 3.24392021e-01
1.18667018e+00 -2.22149074e-01 -1.59437060e-02 -5.88767827e-01
-1.45979953e+00 8.30729723e-01 9.32237208e-02 2.26407900e-01
-4.16289829e-02 2.96286821e-01 2.58211941e-01 1.33465737e-01
6.12489223e-01 3.63036603e-01 1.17992848e-01 -6.88763559e-01
-5.91486037e-01 6.79265738e-01 4.47854161e-01 1.79654896e+00
5.01671255e-01 6.71720028e-01 -1.69764712e-01 8.70174050e-01
8.58559832e-02 8.13315988e-01 7.14668632e-01 -7.65164196e-01
3.24419796e-01 6.27521634e-01 -2.21819773e-01 -1.07728517e+00
-1.71194926e-01 2.76218858e-02 -8.13423336e-01 2.50960171e-01
8.42259377e-02 -1.93421751e-01 -9.56837296e-01 1.14247000e+00
6.37457252e-01 -9.18799266e-02 2.17270218e-02 6.21686339e-01
1.59649658e+00 8.81848872e-01 -2.63653278e-01 2.09580883e-01
1.29627144e+00 -1.00036919e+00 -1.02942562e+00 2.94496119e-01
6.62870467e-01 -1.19066346e+00 1.12696016e+00 3.77914071e-01
-1.27828515e+00 -7.55424678e-01 -8.43894422e-01 -4.81038988e-01
-8.62236261e-01 1.55750392e-02 7.77927697e-01 6.95869088e-01
-1.08311951e+00 1.90295219e-01 -5.10050356e-01 -4.72897649e-01
1.19058527e-01 6.15818892e-04 -2.79751331e-01 -8.14665854e-02
-1.20984948e+00 8.87276769e-01 3.56154382e-01 -1.66349903e-01
-5.65258503e-01 -6.52150989e-01 -9.90720809e-01 -5.61679721e-01
3.41770202e-01 -9.71641660e-01 1.17189360e+00 -3.07647347e-01
-2.06282353e+00 9.86016035e-01 -3.15877632e-03 8.95744413e-02
5.72887838e-01 -5.57517052e-01 -2.27831617e-01 -2.74853408e-02
-8.81397277e-02 8.43221247e-01 6.33722067e-01 -1.25576758e+00
-3.24037731e-01 -2.26257905e-01 -2.92687453e-02 7.73515284e-01
-2.30509445e-01 -1.67731062e-01 -5.28041661e-01 -1.05109942e+00
2.74793327e-01 -7.36674070e-01 -1.48785040e-01 -1.99514721e-03
-4.97859716e-01 -4.44718570e-01 1.12205005e+00 -2.40028992e-01
8.60153913e-01 -1.84599042e+00 2.09850788e-01 -2.34661009e-02
2.23573580e-01 1.89248174e-01 9.10819545e-02 8.87484312e-01
4.32337523e-02 -1.03700981e-01 1.15136012e-01 -2.01012865e-01
2.00752318e-01 -1.64959468e-02 -1.82932153e-01 -7.89087266e-02
3.83413136e-02 1.17840469e+00 -9.45212066e-01 -7.42936134e-01
6.59671366e-01 7.28191257e-01 -6.26229107e-01 1.88695252e-01
-3.21352065e-01 5.09870410e-01 -8.52578521e-01 4.87503648e-01
5.93529105e-01 3.34661096e-01 -4.48077768e-01 -1.58086672e-01
-5.06644487e-01 -9.93304402e-02 -1.04817557e+00 2.00184464e+00
-3.40790689e-01 7.51050770e-01 -3.92574430e-01 -9.02274370e-01
1.50514531e+00 4.13481474e-01 6.25424743e-01 -6.78155243e-01
2.76137680e-01 2.96286911e-01 -4.26487535e-01 -6.35268211e-01
1.20528615e+00 2.17490315e-01 -2.74320066e-01 3.26001495e-01
1.46259507e-02 -1.19509351e+00 1.53899610e-01 3.80857468e-01
3.37356448e-01 9.40761268e-01 3.20575446e-01 -3.10169786e-01
3.86207432e-01 3.91299516e-01 -2.67510097e-02 3.76935661e-01
3.78425717e-01 8.69846404e-01 1.33106783e-01 -4.88094985e-01
-1.14846718e+00 -8.48879039e-01 -5.63527085e-02 8.57577741e-01
3.47731203e-01 -4.07738954e-01 -1.10962200e+00 -4.76628423e-01
-2.60339230e-01 9.86346900e-01 -5.88335216e-01 -3.57136317e-03
-6.06848300e-01 -2.97454178e-01 3.29223633e-01 4.12472337e-01
7.63136983e-01 -1.26458657e+00 -4.78200376e-01 2.83894598e-01
-1.42903283e-01 -9.04362738e-01 -5.26827216e-01 -2.37541929e-01
-1.00073373e+00 -5.50080657e-01 -1.32437658e+00 -1.16715991e+00
8.65557790e-01 6.07296765e-01 1.26765609e+00 -3.78047489e-03
-3.36851686e-01 7.37306297e-01 -9.10864055e-01 -9.79824185e-01
-6.22826576e-01 -6.38521165e-02 -1.24505170e-01 -4.59971666e-01
2.54429698e-01 -4.96374726e-01 -4.07546133e-01 1.77014127e-01
-1.03881013e+00 9.07082915e-01 3.50451142e-01 5.35152376e-01
7.71186650e-01 -1.39574364e-01 3.95621389e-01 -8.99756730e-01
6.47378922e-01 -2.11767592e-02 -3.55064571e-01 2.21173465e-01
-1.13360725e-01 -8.00884143e-02 6.54609323e-01 -4.17522609e-01
-1.29484355e+00 -1.37214154e-01 -4.50541794e-01 -1.24585740e-01
-4.56794620e-01 3.81382942e-01 -3.81377280e-01 7.31029175e-03
6.11690998e-01 5.81769645e-01 -1.33773103e-01 -3.68934989e-01
9.23682094e-01 7.38867044e-01 5.98669708e-01 -9.20954287e-01
9.75220025e-01 1.45359442e-01 1.30352005e-01 -1.16352761e+00
-5.50202608e-01 5.80009557e-02 -1.00511801e+00 -8.16840887e-01
1.04219949e+00 -5.80062509e-01 -5.13926148e-01 8.15727592e-01
-1.28195977e+00 -3.08584958e-01 -5.98975778e-01 4.41189766e-01
-9.31506038e-01 3.90244685e-02 -4.45589006e-01 -5.16800880e-01
-3.27715099e-01 -1.32626748e+00 1.19339943e+00 5.77298462e-01
1.09741822e-01 -1.21846402e+00 4.24922183e-02 4.23161864e-01
3.79180163e-01 6.38450742e-01 1.15489268e+00 -1.45097554e-01
-5.10901690e-01 -2.54232734e-01 7.48878121e-02 -1.80187285e-01
1.32631361e-01 2.87688494e-01 -6.84029341e-01 3.41208041e-01
-2.74808824e-01 -2.84562290e-01 7.46260509e-02 2.93129712e-01
1.25714612e+00 -2.00900733e-02 -2.38348544e-01 5.57375968e-01
1.10051167e+00 5.70944071e-01 8.69430959e-01 3.39921743e-01
1.09650421e+00 5.57688594e-01 7.93658137e-01 2.38194168e-01
6.43969297e-01 1.03547263e+00 2.28308812e-01 -1.79947048e-01
-6.76915586e-01 -7.57507622e-01 1.76477566e-01 1.70080996e+00
-5.27756035e-01 -1.89086318e-01 -6.66389823e-01 1.88702762e-01
-1.55360377e+00 -9.55398202e-01 -6.09435618e-01 2.04157948e+00
8.98678839e-01 -1.86098367e-01 1.68046858e-02 1.11298196e-01
4.63888139e-01 -1.27965003e-01 -3.40631932e-01 -6.81734383e-01
-2.38084063e-01 3.11028570e-01 -2.71300301e-02 2.30732635e-01
-6.91311896e-01 1.32105386e+00 6.14096308e+00 1.27234662e+00
-9.39706266e-01 -3.62256348e-01 5.76295853e-01 5.21978259e-01
-6.96467400e-01 -4.07698989e-01 -9.02025163e-01 1.55131623e-01
3.57335120e-01 -5.03433287e-01 3.45567148e-03 1.08002543e+00
1.62638724e-01 -1.42608956e-01 -7.58016944e-01 1.10443401e+00
4.70541537e-01 -1.54461312e+00 6.17892504e-01 -1.10153824e-01
1.36521864e+00 -7.58517981e-01 5.69593944e-02 3.15280288e-01
3.86700511e-01 -7.86762059e-01 1.01909280e+00 7.34769225e-01
1.16822386e+00 -9.89514530e-01 4.92778599e-01 3.19214612e-01
-1.29034567e+00 8.21551800e-01 -5.54494500e-01 2.28668287e-01
1.64851263e-01 6.58654034e-01 -6.84329569e-01 9.18584228e-01
3.95743251e-01 7.11462080e-01 -3.47630054e-01 1.04216504e+00
-2.13101730e-01 2.48589113e-01 4.35713418e-02 -5.94107151e-01
1.46228865e-01 -5.61890244e-01 5.09264648e-01 1.06789899e+00
8.29407752e-01 3.36852372e-01 7.65149295e-02 9.09717977e-01
1.01815462e-01 6.17630601e-01 -1.11776805e+00 7.24835768e-02
2.89383531e-01 1.03589177e+00 -8.51105869e-01 -5.21833897e-01
-3.28795135e-01 1.16025615e+00 -2.85493378e-02 1.45812809e-01
-8.44049573e-01 -8.24611664e-01 2.41142124e-01 -6.49217982e-03
-2.82227337e-01 -6.94237232e-01 -3.40497822e-01 -9.28425252e-01
-4.82562989e-01 -9.64518905e-01 -3.19764048e-01 -1.31081367e+00
-9.55112934e-01 7.13203549e-01 4.91843581e-01 -1.59204280e+00
-3.51757586e-01 -5.81764340e-01 -5.94223976e-01 5.88573456e-01
-1.01408160e+00 -1.19476116e+00 -5.73831081e-01 5.82999110e-01
9.07779753e-01 -2.29579419e-01 9.58884120e-01 8.35921392e-02
-4.10064846e-01 5.66812515e-01 2.17210382e-01 3.70593481e-02
5.17748654e-01 -1.28980887e+00 7.91740298e-01 5.32154560e-01
3.09585631e-01 3.27836841e-01 5.19777179e-01 -7.42385626e-01
-1.72020590e+00 -7.40904272e-01 6.28650784e-01 -3.93172204e-01
2.55693197e-01 -4.79662389e-01 -5.11410236e-01 4.31662649e-01
2.87086129e-01 -5.09217322e-01 4.24170971e-01 -5.08580744e-01
2.39969730e-01 1.67108864e-01 -8.84481966e-01 1.25601244e+00
1.34016216e+00 2.50690188e-02 -5.07947505e-01 3.57615322e-01
9.69807088e-01 -1.31899476e+00 -1.00410116e+00 1.10821255e-01
4.26727086e-01 -1.02292717e+00 8.48451138e-01 -2.68800944e-01
6.60143018e-01 -1.98300242e-01 -5.66851050e-02 -1.54085994e+00
-3.17464396e-02 -9.68679130e-01 1.78575307e-01 1.15420091e+00
4.76720519e-02 -3.65589947e-01 1.05871594e+00 5.02092004e-01
-9.62063432e-01 -7.69369185e-01 -4.42258477e-01 -5.40953696e-01
2.86347300e-01 -6.23498142e-01 9.35678005e-01 8.54300380e-01
-4.85824570e-02 4.96515557e-02 -4.00770932e-01 -6.01221442e-01
4.97941434e-01 3.21368515e-01 1.32828701e+00 -1.11199844e+00
2.96825409e-01 -4.73519027e-01 -5.70603848e-01 -1.73729634e+00
-2.40439549e-01 -9.55112875e-01 1.02250613e-01 -1.91282701e+00
-7.13106319e-02 -3.87720466e-01 7.26030469e-01 1.13225684e-01
-1.55235440e-01 4.07213837e-01 2.79780537e-01 -1.30286142e-01
-1.09213725e-01 1.06456494e+00 2.38627315e+00 1.37661114e-01
-3.50503325e-01 -1.33629575e-01 -3.97447079e-01 6.82252586e-01
7.14471161e-01 7.50220940e-03 -5.69273353e-01 -4.88713741e-01
1.54487953e-01 5.59817702e-02 -1.23199299e-01 -1.16391492e+00
-6.83243573e-03 -4.05555159e-01 3.92866343e-01 -1.11816144e+00
4.99747574e-01 -6.95396245e-01 1.92545295e-01 8.37205499e-02
-2.48813644e-01 3.06751072e-01 2.47464553e-01 4.19139629e-03
-1.91610366e-01 -3.37745041e-01 4.07727867e-01 -1.78800046e-01
-8.19970429e-01 2.28767380e-01 -4.84785140e-01 5.05644828e-02
1.19138741e+00 -7.47406125e-01 -3.11502993e-01 -8.19987655e-01
-4.46742177e-01 -2.34693766e-01 3.03440303e-01 4.97925401e-01
9.45901513e-01 -1.84875488e+00 -7.26781189e-01 3.97914916e-01
9.02611315e-02 3.39852303e-01 4.67853189e-01 3.69979143e-01
-1.11886132e+00 4.94883180e-01 -2.89041758e-01 -5.79741478e-01
-1.24122226e+00 2.16142476e-01 -9.27361622e-02 1.85174555e-01
-9.65361357e-01 6.73665106e-01 1.05044000e-01 -5.44267595e-01
1.76801875e-01 -2.85409272e-01 -3.54582936e-01 -2.64623433e-01
4.72050428e-01 5.54934919e-01 1.21679921e-02 -8.04514050e-01
2.98084527e-01 1.11326075e+00 1.84960097e-01 -1.27765298e-01
1.12464941e+00 1.60449240e-02 1.84029028e-01 6.18983567e-01
8.60876441e-01 2.45903239e-01 -9.96131122e-01 4.57483046e-02
-5.42806983e-01 -4.15078729e-01 -1.10598050e-01 -4.21083748e-01
-1.08941352e+00 1.13086462e+00 1.30317241e-01 3.64411205e-01
7.93035150e-01 -1.18389688e-01 1.04259050e+00 1.61856562e-01
6.67242825e-01 -1.10868454e+00 5.11092782e-01 7.67526686e-01
1.35306191e+00 -8.47974241e-01 9.55443606e-02 -7.19859481e-01
-6.80073500e-01 1.45521080e+00 8.27794492e-01 2.89683267e-02
6.23869240e-01 1.57097861e-01 2.36266553e-01 -3.22790891e-01
-3.31514835e-01 -1.76037595e-01 6.64400995e-01 1.22499073e+00
9.05082226e-01 1.83320612e-01 -1.98047325e-01 -3.20323408e-02
-1.06080496e+00 -1.10586762e-01 7.75548279e-01 8.37380528e-01
-4.24579144e-01 -1.43052530e+00 -7.11223900e-01 3.27422202e-01
1.00130945e-01 -4.18251418e-02 -3.14719141e-01 1.20155561e+00
-6.52243719e-02 7.81470299e-01 8.22948366e-02 -4.88108784e-01
6.43669367e-01 -1.40620908e-02 7.70336747e-01 -6.27954543e-01
-3.06786448e-01 9.58201359e-04 -9.90992263e-02 -3.43324006e-01
-3.82010072e-01 -4.38654006e-01 -1.48546576e+00 -6.37276947e-01
-2.21878231e-01 1.16875753e-01 1.03647435e+00 5.74720263e-01
1.96678385e-01 5.31372905e-01 5.22856355e-01 -1.24170303e+00
2.31903076e-01 -9.50790346e-01 -5.53603232e-01 3.92112315e-01
-5.40564179e-01 -7.76259542e-01 2.25785807e-01 2.34760195e-01] | [11.437715530395508, -0.41122424602508545] |
dcf6af36-8aab-4f76-8754-3707dd594a2a | interpretable-spectrum-transformation-attacks | 2302.10686 | null | https://arxiv.org/abs/2302.10686v1 | https://arxiv.org/pdf/2302.10686v1.pdf | Interpretable Spectrum Transformation Attacks to Speaker Recognition | The success of adversarial attacks to speaker recognition is mainly in white-box scenarios. When applying the adversarial voices that are generated by attacking white-box surrogate models to black-box victim models, i.e. \textit{transfer-based} black-box attacks, the transferability of the adversarial voices is not only far from satisfactory, but also lacks interpretable basis. To address these issues, in this paper, we propose a general framework, named spectral transformation attack based on modified discrete cosine transform (STA-MDCT), to improve the transferability of the adversarial voices to a black-box victim model. Specifically, we first apply MDCT to the input voice. Then, we slightly modify the energy of different frequency bands for capturing the salient regions of the adversarial noise in the time-frequency domain that are critical to a successful attack. Unlike existing approaches that operate voices in the time domain, the proposed framework operates voices in the time-frequency domain, which improves the interpretability, transferability, and imperceptibility of the attack. Moreover, it can be implemented with any gradient-based attackers. To utilize the advantage of model ensembling, we not only implement STA-MDCT with a single white-box surrogate model, but also with an ensemble of surrogate models. Finally, we visualize the saliency maps of adversarial voices by the class activation maps (CAM), which offers an interpretable basis to transfer-based attacks in speaker recognition for the first time. Extensive comparison results with five representative attackers show that the CAM visualization clearly explains the effectiveness of STA-MDCT, and the weaknesses of the comparison methods; the proposed method outperforms the comparison methods by a large margin. | ['Xiao-Lei Zhang', 'Hong Luo', 'Jiadi Yao'] | 2023-02-21 | null | null | null | null | ['speaker-recognition'] | ['speech'] | [ 2.22540811e-01 5.44128977e-02 2.92529196e-01 7.58641064e-02
-6.92691386e-01 -9.31706727e-01 6.37008071e-01 -6.16143346e-01
3.78094055e-03 2.77409613e-01 3.61152619e-01 -4.40675259e-01
-2.10987162e-02 -5.86105406e-01 -3.84458452e-01 -7.30996370e-01
-7.20546842e-02 -2.61193514e-01 1.08264618e-01 -4.89600748e-01
-1.04903698e-01 6.90046072e-01 -1.06704605e+00 3.50725889e-01
8.18977058e-01 7.72872150e-01 -1.41431600e-01 7.10814357e-01
1.05967307e-02 3.87801200e-01 -1.26199841e+00 -5.74962795e-01
2.92049646e-01 -6.41292810e-01 -3.98155004e-01 -4.69942719e-01
1.68126792e-01 -2.41360754e-01 -5.80552518e-01 1.31039655e+00
7.34085619e-01 1.61942989e-01 5.69174886e-01 -1.37149096e+00
-6.89218819e-01 6.22747600e-01 -2.87434042e-01 1.60867915e-01
5.25141180e-01 3.89477462e-01 5.87771714e-01 -7.95595109e-01
2.60150373e-01 1.83195281e+00 6.34016752e-01 1.06864727e+00
-9.80348229e-01 -1.09736717e+00 3.56844366e-01 2.48875037e-01
-1.17784595e+00 -5.47907948e-01 1.16913414e+00 -1.96764216e-01
5.44160187e-01 1.01240051e+00 3.71131778e-01 1.50983679e+00
-2.98967026e-02 5.03437102e-01 1.11774504e+00 -3.41855556e-01
8.07596147e-02 2.53045887e-01 -3.25315326e-01 3.75840724e-01
-1.97873622e-01 7.23816872e-01 -4.03727561e-01 -3.38611364e-01
5.26108921e-01 -1.21789791e-01 -6.68533504e-01 5.79409339e-02
-1.13282311e+00 6.52016342e-01 7.33325183e-01 3.98512483e-01
2.23158970e-02 5.78105822e-02 2.70993561e-01 2.67488390e-01
3.88243020e-01 5.67833543e-01 7.00079352e-02 9.22433659e-02
-7.36082315e-01 -8.67463052e-02 7.86889374e-01 5.29823840e-01
1.82510838e-01 9.23052609e-01 -2.52265990e-01 4.56490427e-01
4.78883415e-01 9.10759568e-01 6.30324543e-01 -3.98068249e-01
4.61444795e-01 1.86643913e-01 -6.76794350e-02 -1.01526403e+00
-1.45433351e-01 -5.59441686e-01 -7.28009760e-01 7.43942022e-01
4.36517805e-01 -4.18011814e-01 -8.08577597e-01 1.95562553e+00
3.58038664e-01 3.71931553e-01 3.60800594e-01 9.36044037e-01
8.39635730e-01 7.38827229e-01 1.41511373e-02 -7.65913054e-02
1.21994996e+00 -9.73340809e-01 -9.74842906e-01 -2.47080728e-01
5.12183607e-02 -8.69531453e-01 1.30415785e+00 1.86300576e-01
-7.26285636e-01 -7.32387722e-01 -1.28581107e+00 5.63406706e-01
-5.42287886e-01 -6.20556548e-02 2.39868183e-02 1.30259836e+00
-7.04697847e-01 5.16003489e-01 -5.65198302e-01 -7.05912337e-02
1.78137407e-01 2.36089170e-01 -3.07001203e-01 4.97188628e-01
-1.59635675e+00 1.10807931e+00 8.18208158e-02 1.87489584e-01
-1.30660474e+00 -6.63920820e-01 -6.79263234e-01 2.30526879e-01
9.99240391e-03 -2.78523654e-01 9.06528473e-01 -1.09865010e+00
-1.80272031e+00 2.04084247e-01 1.55173494e-02 -3.94327223e-01
7.07727075e-01 -1.46008819e-01 -1.02174163e+00 2.15259030e-01
-4.29504424e-01 2.67591029e-01 1.53683710e+00 -1.48267329e+00
-4.19632904e-02 -1.10236011e-01 1.31452218e-01 3.92090753e-02
-7.73752868e-01 2.69028157e-01 9.41476449e-02 -1.18019772e+00
-1.22562803e-01 -7.79969811e-01 4.29975428e-03 -3.29744332e-02
-6.67700648e-01 4.64690298e-01 1.37326169e+00 -9.09072936e-01
1.44277406e+00 -2.69247150e+00 -3.41743939e-02 3.46083254e-01
2.01620594e-01 6.68605685e-01 -2.38017619e-01 4.59977061e-01
-5.90779781e-01 4.18347418e-01 -2.81225324e-01 1.40382778e-02
1.56120971e-01 -2.47224644e-01 -9.09357011e-01 3.65226686e-01
2.57697046e-01 7.59769440e-01 -6.80845499e-01 -6.80131763e-02
2.15558514e-01 6.73540235e-01 -3.83720398e-01 2.65052587e-01
2.72174180e-01 6.63025856e-01 -4.32485014e-01 3.19919974e-01
7.35030711e-01 6.00815713e-01 -1.41555369e-01 -3.60160530e-01
1.19725972e-01 5.43255173e-02 -1.06363285e+00 9.75670516e-01
-5.49771130e-01 7.31310606e-01 2.80820698e-01 -5.17622590e-01
1.11498153e+00 6.71196222e-01 -5.83956651e-02 -2.54052311e-01
-1.11152053e-01 2.11111844e-01 3.11918557e-01 -2.87860513e-01
-1.39677733e-01 -3.81382614e-01 2.05152724e-02 3.05655092e-01
-1.09762713e-01 -2.36484379e-01 -5.11429548e-01 4.06860225e-02
9.79388952e-01 -1.13829695e-01 1.62082225e-01 -8.67268369e-02
8.00471604e-01 -5.75833440e-01 3.37223381e-01 5.16394317e-01
-3.22528988e-01 5.41724682e-01 3.48269135e-01 -1.75104573e-01
-7.45293558e-01 -1.35192168e+00 7.18585327e-02 8.92075956e-01
2.55224526e-01 -2.01358169e-01 -1.07022476e+00 -8.70724380e-01
-1.21110700e-01 9.46233332e-01 -5.96249938e-01 -6.83158159e-01
-6.11254334e-01 -4.42639410e-01 1.22616124e+00 3.61617655e-01
7.52480030e-01 -1.06885743e+00 -2.92362928e-01 2.28490005e-03
-1.06383778e-01 -9.42465842e-01 -8.37174654e-01 -3.91117930e-02
-4.82817292e-01 -7.48522758e-01 -7.22563863e-01 -5.84755182e-01
6.33260131e-01 1.00174122e-01 5.48096359e-01 -1.42261937e-01
-7.27546066e-02 1.64164543e-01 -2.61596650e-01 -6.78623974e-01
-7.54445076e-01 -4.27079588e-01 3.41583848e-01 3.07260126e-01
-1.47313625e-01 -5.81670702e-01 -4.40711766e-01 6.17891908e-01
-8.74249995e-01 -2.61853844e-01 3.77645314e-01 9.59348857e-01
-9.22160894e-02 1.36968374e-01 8.39628756e-01 -4.20060426e-01
9.15492833e-01 -1.37196586e-01 -2.60585278e-01 2.49653623e-01
-1.92154229e-01 -1.80202246e-01 1.06255305e+00 -9.17087257e-01
-1.14941001e+00 -2.15001896e-01 -4.00252610e-01 -7.63238728e-01
-4.17563645e-03 7.70437717e-02 -7.21543431e-01 -4.01380569e-01
1.00143516e+00 2.35551670e-01 5.12775220e-02 -4.04595166e-01
5.15070558e-01 6.80243254e-01 6.03665352e-01 -3.43714684e-01
1.59707201e+00 3.79088402e-01 -3.39065850e-01 -7.30226636e-01
-3.68466288e-01 1.93378344e-01 -2.44777352e-01 -3.37423474e-01
7.21468031e-01 -4.62723047e-01 -7.70083189e-01 4.84103352e-01
-1.20475328e+00 -6.55817986e-02 -4.57436085e-01 4.76775825e-01
-1.74134001e-01 4.37562853e-01 -4.33264554e-01 -1.08358490e+00
-3.91321421e-01 -1.20985293e+00 7.20351040e-01 1.73617721e-01
-2.69062042e-01 -1.07621753e+00 -2.05984995e-01 2.07050905e-01
6.63754404e-01 4.13279414e-01 9.87655222e-01 -1.08741617e+00
-7.84168914e-02 -3.99960369e-01 2.30333075e-01 6.12789631e-01
2.87912160e-01 2.73513738e-02 -1.62546003e+00 -4.13024426e-01
5.20003140e-01 1.67353541e-01 4.52894539e-01 6.40509129e-02
9.72411215e-01 -7.97923863e-01 -2.33154863e-01 6.10472143e-01
7.11861551e-01 5.11367083e-01 7.06302881e-01 1.98969036e-01
6.28457248e-01 6.87166810e-01 2.55366057e-01 2.88939010e-02
-3.26783389e-01 8.24812889e-01 6.60142839e-01 -5.95024109e-01
-3.28431845e-01 -4.94714230e-01 9.88584697e-01 7.66763747e-01
1.93304181e-01 -2.44698137e-01 -6.32022977e-01 8.35709572e-02
-1.24127805e+00 -1.24095690e+00 3.35141420e-01 2.18135190e+00
7.05218971e-01 2.80289531e-01 5.32541052e-02 4.80282366e-01
1.08005118e+00 4.13629144e-01 -5.77197552e-01 -6.21576607e-01
-1.17100120e-01 1.92305729e-01 3.51911522e-02 6.54387832e-01
-1.08683074e+00 7.68143833e-01 6.46887970e+00 1.15361798e+00
-1.48169649e+00 1.71949789e-01 3.69862765e-01 -1.00952603e-01
-4.69908267e-01 -1.42500564e-01 -3.24853480e-01 4.65446949e-01
8.57573748e-01 -5.20966887e-01 7.74890423e-01 8.26966882e-01
3.86349142e-01 8.12821388e-01 -1.08380687e+00 7.90743589e-01
-2.51218788e-02 -7.94440925e-01 2.68166691e-01 3.02347727e-03
3.64912391e-01 -7.38605618e-01 6.91919744e-01 2.79141486e-01
8.86107534e-02 -1.31053603e+00 9.25316513e-01 1.63373441e-01
9.99699712e-01 -7.08427608e-01 4.83110279e-01 2.58914232e-01
-1.26319718e+00 -2.03359902e-01 -5.53183518e-02 1.63897291e-01
-1.04583606e-01 1.79862633e-01 -8.51485848e-01 5.75902998e-01
4.15208906e-01 5.77821024e-02 -4.42271024e-01 7.54045427e-01
-4.29325134e-01 8.04135740e-01 -1.32868022e-01 2.52254754e-02
8.33718851e-02 1.06670380e-01 1.12511826e+00 1.20484102e+00
4.09538627e-01 -1.83356687e-01 -1.54755220e-01 1.09231555e+00
3.92890833e-02 -1.63187347e-02 -7.99641907e-01 -8.73870403e-03
6.98097229e-01 1.25004983e+00 -2.42570475e-01 -1.04897350e-01
-5.56854941e-02 9.29038346e-01 -3.61589670e-01 7.11467445e-01
-1.14536655e+00 -8.48208249e-01 7.42988825e-01 -8.37520733e-02
4.46627177e-02 3.00752576e-02 -3.11233312e-01 -8.98419023e-01
2.10651234e-02 -1.46657789e+00 1.13927178e-01 -7.69357026e-01
-1.24707747e+00 1.12735963e+00 -1.24070525e-01 -1.65329242e+00
-1.46006361e-01 -4.40069765e-01 -1.28658080e+00 1.25251234e+00
-1.18492305e+00 -1.17700422e+00 -2.02045858e-01 8.53720784e-01
4.22637850e-01 -4.53379780e-01 1.06982839e+00 9.79124680e-02
-5.99596441e-01 1.06094253e+00 -8.85445327e-02 2.33772621e-01
5.94071805e-01 -1.03036547e+00 7.82964885e-01 1.15939772e+00
2.76274234e-01 7.85047889e-01 9.16139424e-01 -3.23442250e-01
-1.07663071e+00 -1.05193424e+00 2.20486581e-01 -3.59709114e-01
6.57388210e-01 -5.56498349e-01 -9.50395882e-01 3.26939553e-01
3.23100626e-01 -1.09337792e-01 6.82225347e-01 -3.19056571e-01
-6.19423985e-01 -3.29923093e-01 -1.47442317e+00 9.59188640e-01
6.03702664e-01 -9.05617177e-01 -7.63549805e-01 1.37272149e-01
8.95891905e-01 -2.54005402e-01 -5.73592007e-01 2.97785789e-01
4.32944000e-01 -7.48632312e-01 1.26439643e+00 -5.91417730e-01
3.67668867e-02 -5.38319409e-01 1.45665677e-02 -1.63413525e+00
-1.73414439e-01 -1.09507990e+00 -5.80419712e-02 1.46025431e+00
4.77812678e-01 -9.23470438e-01 2.86159158e-01 2.66435921e-01
-6.04991466e-02 -3.75915170e-01 -1.32804024e+00 -9.67434764e-01
2.15122893e-01 -4.10942763e-01 9.31540251e-01 8.89388084e-01
1.40710846e-01 -9.32211056e-03 -4.39195722e-01 4.95162845e-01
4.36702520e-01 -3.82210344e-01 6.87062919e-01 -8.22152972e-01
-3.21413577e-01 -5.18245876e-01 -4.30883080e-01 -6.71507716e-01
1.15534730e-01 -7.36725390e-01 -7.47046322e-02 -9.60068882e-01
-3.91757280e-01 -2.77877860e-02 -3.48726600e-01 4.59591597e-01
-4.59373504e-01 2.09049284e-01 5.27402341e-01 1.75911248e-01
2.37692788e-01 7.14270949e-01 1.33242142e+00 -3.99928242e-01
-1.45148709e-01 1.77640930e-01 -6.56025887e-01 9.41819131e-01
6.93756163e-01 -3.82053673e-01 -4.03537303e-01 -9.46955010e-02
-5.53242803e-01 5.13272360e-02 5.52977979e-01 -1.18978012e+00
1.97634138e-02 -3.31999036e-03 4.37466174e-01 8.43800157e-02
6.14806712e-01 -8.60093474e-01 2.29563713e-01 7.82017469e-01
-3.75183284e-01 -6.75198585e-02 4.89387900e-01 4.87248451e-01
-3.69396359e-01 -2.06343517e-01 1.02100515e+00 1.80871904e-01
-3.03649038e-01 -1.14163500e-03 -3.30225289e-01 -1.03581697e-01
1.05764079e+00 -2.62042969e-01 -5.16503453e-01 -7.58992195e-01
-8.37006807e-01 -3.34907979e-01 6.90116510e-02 5.23407102e-01
7.24600852e-01 -1.52241683e+00 -6.95221484e-01 4.61901158e-01
-2.33199894e-01 -7.28946984e-01 3.02585095e-01 5.41352510e-01
-2.20054761e-01 2.42775261e-01 -2.17431873e-01 -2.75872201e-01
-1.37072265e+00 7.75714457e-01 6.57427549e-01 -4.72526178e-02
-4.59540784e-01 7.88471162e-01 7.54006624e-01 -4.05273795e-01
2.86602139e-01 -3.08653098e-02 -3.29843938e-01 -1.81102082e-01
6.13424301e-01 3.50041330e-01 -8.39096308e-02 -6.69076025e-01
-5.28676033e-01 5.26210308e-01 2.71676004e-01 -4.69095796e-01
8.46022606e-01 2.22034544e-01 1.81254879e-01 1.19047493e-01
9.87669826e-01 6.80571914e-01 -1.17450368e+00 1.00962840e-01
-3.08599025e-01 -6.00342751e-01 -1.41364783e-01 -9.66250181e-01
-1.15785992e+00 1.19943798e+00 8.38067293e-01 5.38994253e-01
1.27773452e+00 -5.35545886e-01 4.94408786e-01 3.36961299e-02
3.59652638e-02 -5.34345388e-01 2.74900049e-01 1.59351572e-01
1.43699110e+00 -5.68282366e-01 -3.59011352e-01 -5.05357027e-01
-8.24495077e-01 1.18350601e+00 5.85539222e-01 9.20022354e-02
4.98548090e-01 3.73444736e-01 5.16409576e-01 3.04524630e-01
-4.87897873e-01 1.17831096e-01 5.13148606e-01 9.40899670e-01
8.23067054e-02 -4.80559282e-02 1.50496960e-01 8.00021648e-01
-6.06344640e-01 -7.28201628e-01 2.04289392e-01 4.19739783e-01
-2.09729612e-01 -1.03208065e+00 -1.09144711e+00 -1.14341550e-01
-4.59409088e-01 -2.06327781e-01 -6.62428916e-01 6.09679103e-01
-5.13407178e-02 1.36182821e+00 -4.85543549e-01 -9.69503164e-01
5.14231801e-01 1.80722192e-01 -3.70034426e-02 -2.70250142e-01
-1.03418994e+00 1.96212143e-01 -5.44285215e-02 -2.35522851e-01
1.01157159e-01 -1.77239940e-01 -1.03996396e+00 -4.18926299e-01
-5.13999283e-01 2.33507872e-01 6.21857703e-01 7.07282186e-01
2.14863002e-01 9.77664769e-01 1.17619038e+00 -8.72432709e-01
-9.72618759e-01 -1.08474731e+00 -3.30468178e-01 3.90261412e-01
4.72242653e-01 -5.16392231e-01 -9.50054348e-01 -6.91994280e-02] | [13.997596740722656, 5.824460506439209] |
ba15518d-d128-49ff-b7cf-617a6eaab5cd | dialogpt-large-scale-generative-pre-training | 1911.00536 | null | https://arxiv.org/abs/1911.00536v3 | https://arxiv.org/pdf/1911.00536v3.pdf | DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation | We present a large, tunable neural conversational response generation model, DialoGPT (dialogue generative pre-trained transformer). Trained on 147M conversation-like exchanges extracted from Reddit comment chains over a period spanning from 2005 through 2017, DialoGPT extends the Hugging Face PyTorch transformer to attain a performance close to human both in terms of automatic and human evaluation in single-turn dialogue settings. We show that conversational systems that leverage DialoGPT generate more relevant, contentful and context-consistent responses than strong baseline systems. The pre-trained model and training pipeline are publicly released to facilitate research into neural response generation and the development of more intelligent open-domain dialogue systems. | ['Chris Brockett', 'Yen-Chun Chen', 'Xiang Gao', 'Yizhe Zhang', 'Jingjing Liu', 'Jianfeng Gao', 'Siqi Sun', 'Michel Galley', 'Bill Dolan'] | 2019-11-01 | null | null | null | null | ['conversational-response-generation'] | ['natural-language-processing'] | [ 2.91129529e-01 6.93006516e-01 4.58611213e-02 -7.45345950e-01
-1.20483184e+00 -9.00204718e-01 1.15469193e+00 -3.17532033e-01
-1.49687812e-01 1.24689209e+00 1.02182913e+00 -3.47370058e-01
5.00485718e-01 -6.25597537e-01 -1.97869927e-01 -1.26686454e-01
2.65579551e-01 1.16419291e+00 -2.35700428e-01 -1.00033021e+00
6.02019727e-02 -3.35997611e-01 -6.46443546e-01 9.83983994e-01
8.02231133e-01 5.25141478e-01 -2.18212217e-01 1.22762573e+00
-1.72532108e-02 1.19630241e+00 -1.11761677e+00 -9.39836681e-01
5.45719452e-02 -9.50360298e-01 -1.56780088e+00 -1.66820556e-01
2.55665332e-01 -5.25346398e-01 -2.98494458e-01 3.41584116e-01
8.61304939e-01 5.68489015e-01 5.99530578e-01 -8.33827734e-01
-9.82900739e-01 1.37625802e+00 2.24675566e-01 -2.36346256e-02
7.51059353e-01 7.10820317e-01 1.26370239e+00 -8.19434166e-01
9.47206795e-01 1.67863846e+00 4.60386962e-01 1.29359090e+00
-1.43994606e+00 -5.14928997e-01 -6.03537671e-02 -4.22020078e-01
-3.84445578e-01 -6.85346305e-01 4.86677647e-01 -2.01691076e-01
1.34320199e+00 4.84579504e-01 2.76245952e-01 2.05731630e+00
-6.33403584e-02 9.18395102e-01 1.00589097e+00 -8.86389613e-02
-1.70888454e-01 1.19638786e-01 1.39222562e-01 1.90895155e-01
-7.36286104e-01 -1.87181324e-01 -6.04927540e-01 -4.64193374e-01
4.24586177e-01 -7.61115134e-01 -3.64037782e-01 5.79223514e-01
-1.25890732e+00 1.25529253e+00 3.92661124e-01 9.19482261e-02
-4.47037965e-01 2.19273511e-02 7.39298940e-01 8.79338026e-01
9.05255735e-01 1.11817610e+00 -4.58227038e-01 -7.41374016e-01
-3.40216458e-01 7.55611598e-01 1.58159959e+00 1.06139648e+00
3.19195092e-01 1.50820524e-01 -8.85154486e-01 1.43418264e+00
-2.04174295e-01 3.68399203e-01 5.63009381e-01 -1.21913624e+00
8.69460464e-01 5.56100607e-01 2.01982439e-01 -4.48897302e-01
-3.91053140e-01 6.01288080e-02 -8.29110920e-01 -4.94454354e-01
5.07842362e-01 -1.09334004e+00 -2.69734800e-01 1.70895135e+00
3.06921136e-02 -7.57899880e-01 4.98709589e-01 9.36997890e-01
1.36256099e+00 9.56376851e-01 -9.47180614e-02 -6.07977174e-02
1.21983731e+00 -1.35073125e+00 -5.02741039e-01 -2.39334881e-01
6.00784421e-01 -7.59436607e-01 1.29263437e+00 1.89778388e-01
-1.51443493e+00 -4.46007043e-01 -4.33368802e-01 -3.52168322e-01
1.03789426e-01 7.59276189e-03 5.57098031e-01 3.46892536e-01
-1.20876253e+00 1.68169469e-01 -7.84105808e-02 -3.66515249e-01
-1.14436418e-01 1.87608134e-02 -1.82874933e-01 3.41005951e-01
-1.67640805e+00 1.05276036e+00 -3.32948193e-02 -3.00883073e-02
-1.00899267e+00 -4.44011480e-01 -6.29262805e-01 -6.61562830e-02
1.96208701e-01 -8.64933372e-01 2.31314421e+00 -9.50294435e-01
-2.19687533e+00 9.29581225e-01 -2.95154899e-02 -7.97342479e-01
7.37976909e-01 -2.72396654e-01 -7.53896534e-02 4.72687222e-02
-1.98121324e-01 9.17491794e-01 3.73349458e-01 -7.86027312e-01
-2.35621661e-01 2.95998931e-01 3.74959797e-01 5.62954605e-01
-7.57442191e-02 3.40939373e-01 1.30814835e-01 -4.28131402e-01
-8.10571015e-01 -1.14757371e+00 -2.68382728e-01 -8.26704860e-01
-6.99494421e-01 -8.38577688e-01 2.77210295e-01 -6.77830100e-01
7.59951353e-01 -1.45266569e+00 2.04342350e-01 -3.35880160e-01
1.23084761e-01 2.55891532e-01 -5.25854230e-01 1.19399905e+00
2.10016668e-01 1.84981674e-02 -1.85778067e-01 -4.77156162e-01
2.00743124e-01 -1.76765114e-01 -7.80501246e-01 -2.86817968e-01
2.70843536e-01 1.21308756e+00 -1.06123829e+00 -8.27267244e-02
-1.15667619e-01 6.65145814e-02 -7.26025045e-01 8.95258963e-01
-8.57089341e-01 8.77008557e-01 -3.55232924e-01 1.03988327e-01
3.40778455e-02 -3.57141048e-01 3.03310722e-01 5.32447219e-01
2.70124115e-02 1.17739463e+00 -9.50373523e-03 1.68018281e+00
-7.03857064e-01 8.48596394e-01 9.88106653e-02 -1.62688658e-01
1.21811664e+00 7.43809998e-01 2.27744859e-02 -5.39936662e-01
1.07344270e-01 -2.88874395e-02 1.67761907e-01 -3.53275329e-01
1.22029293e+00 -2.54936479e-02 -6.37079060e-01 1.32501125e+00
2.47113585e-01 -3.99755508e-01 1.54863268e-01 8.02113235e-01
1.20808637e+00 -2.65206248e-01 1.39871031e-01 -1.57700524e-01
4.94291574e-01 9.22137871e-02 9.86788720e-02 9.38976288e-01
-3.98401022e-02 3.95308018e-01 8.29377294e-01 -3.65102112e-01
-9.68762636e-01 -7.67396867e-01 2.58444875e-01 1.80362570e+00
-5.63055694e-01 -2.48045608e-01 -9.03130889e-01 -7.78622508e-01
-1.52851880e-01 8.53182137e-01 -4.81563866e-01 1.56430714e-02
-8.07757080e-01 -5.28624058e-01 1.06014407e+00 3.46017897e-01
6.28840685e-01 -1.67975640e+00 -1.84902810e-02 3.89349163e-01
-8.79430413e-01 -9.94249761e-01 -8.71074080e-01 -2.86640912e-01
-4.59585398e-01 -7.56674051e-01 -7.80592680e-01 -6.72978461e-01
1.93673491e-01 -1.08687460e-01 1.64902139e+00 3.95657495e-02
1.65244833e-01 2.84179360e-01 -5.19211471e-01 -1.75236702e-01
-1.31400633e+00 6.33826315e-01 -2.65717059e-01 -3.05083007e-01
2.12586224e-01 -3.17964911e-01 -6.27127111e-01 3.59621346e-01
-2.47117192e-01 3.94025028e-01 2.19728410e-01 1.25405562e+00
-5.54829240e-01 -1.35529137e+00 1.24777043e+00 -1.24834740e+00
1.86701190e+00 -6.39285982e-01 -1.91142321e-01 2.21296296e-01
-4.59169030e-01 -7.22104013e-02 7.25990891e-01 -3.81864548e-01
-1.68394113e+00 -5.15632987e-01 -4.28262711e-01 4.08259183e-01
-6.80550262e-02 1.36849061e-01 2.99462020e-01 4.35474396e-01
1.23319757e+00 2.04606608e-01 8.47304687e-02 -3.03905487e-01
8.38025331e-01 1.05938923e+00 7.23581672e-01 -8.91338706e-01
2.81757325e-01 -2.94625551e-01 -9.02547956e-01 -4.75205511e-01
-5.51674902e-01 -1.60783693e-01 -5.79646677e-02 -3.88046414e-01
8.09641838e-01 -8.09665203e-01 -1.22145188e+00 4.53367233e-01
-1.58690369e+00 -1.11134887e+00 -1.68872681e-02 -9.89289060e-02
-6.67381644e-01 -5.85174225e-02 -1.50889087e+00 -8.97654235e-01
-1.15744436e+00 -7.74942398e-01 8.83743525e-01 1.88848585e-01
-1.06286883e+00 -1.09685516e+00 6.63948536e-01 7.88271785e-01
7.48601615e-01 -6.05751574e-02 7.64663279e-01 -1.23966813e+00
-3.88879031e-01 8.84757936e-02 -6.77110925e-02 3.11088055e-01
-8.50115195e-02 -4.81989458e-02 -1.03722334e+00 -9.81850550e-02
-2.26372182e-01 -1.23749959e+00 6.62121654e-01 -9.06187370e-02
4.74147379e-01 -1.00482941e+00 6.27539456e-02 -1.50990183e-03
3.63779068e-01 1.03486188e-01 4.63719755e-01 -2.08103955e-01
3.19120437e-01 1.07578552e+00 3.35049182e-01 5.00256717e-01
6.72907412e-01 5.98185062e-01 -1.85858607e-02 1.43302996e-02
-4.97952402e-02 -5.17808318e-01 5.32917738e-01 1.01465404e+00
4.51982021e-02 -7.94553041e-01 -7.37934470e-01 6.28490627e-01
-1.98336720e+00 -1.26695573e+00 -1.35756016e-01 1.63854825e+00
1.64241052e+00 -5.24736233e-02 5.01563966e-01 -7.54378498e-01
5.96549988e-01 3.50945443e-01 -4.29655224e-01 -1.24480855e+00
-2.11438894e-01 2.48681381e-01 -1.79499224e-01 8.78247917e-01
-5.38311541e-01 1.12219858e+00 6.98314953e+00 2.81379491e-01
-8.48217666e-01 2.32271627e-01 8.30448866e-01 -1.79844648e-01
-5.95141292e-01 2.98526753e-02 -6.64541066e-01 4.68176752e-01
1.37739646e+00 -4.46478814e-01 6.27688885e-01 6.52666509e-01
1.72899663e-01 2.50184178e-01 -1.22507811e+00 3.56890529e-01
5.89247374e-03 -1.47815812e+00 -4.07123007e-02 -7.51067623e-02
9.51201320e-01 1.82318166e-01 -1.13631412e-02 9.77407336e-01
1.49902165e+00 -1.06955922e+00 2.67164916e-01 3.97662044e-01
6.66701734e-01 -4.59752023e-01 5.85743725e-01 4.21022773e-01
-3.56701702e-01 1.22580091e-02 -4.24034297e-02 -4.21315163e-01
5.86931825e-01 1.03327416e-01 -1.84274971e+00 2.45311439e-01
2.53689457e-02 4.41428959e-01 -1.20168999e-01 2.72039652e-01
-5.00296056e-01 9.20760453e-01 1.29697204e-01 -5.07294774e-01
3.69375914e-01 -8.51840451e-02 6.68005526e-01 1.59600627e+00
-3.07935208e-01 3.21193218e-01 1.38148785e-01 1.01397943e+00
-7.58837879e-01 2.11846195e-02 -5.67108154e-01 -2.25471407e-01
6.87572002e-01 1.50475085e+00 1.20657101e-01 -5.34119546e-01
-6.23956919e-02 9.44198191e-01 5.10720253e-01 3.99998933e-01
-4.88567263e-01 -2.11037815e-01 7.42916167e-01 -2.73926288e-01
-3.63953114e-01 3.05094477e-02 -2.53145337e-01 -1.01754153e+00
-3.60089093e-01 -1.47279298e+00 4.28171396e-01 -7.70263016e-01
-1.83930767e+00 1.06417656e+00 -1.01182759e-01 -6.58239245e-01
-1.40239096e+00 -2.02709064e-01 -9.94177043e-01 1.11291552e+00
-7.00733185e-01 -1.10450912e+00 -1.10550866e-01 3.72774124e-01
1.10843074e+00 -2.73795396e-01 1.28419042e+00 -1.90928936e-01
-3.01009417e-01 8.54123950e-01 -3.29315752e-01 5.44559419e-01
1.22651386e+00 -1.33089972e+00 1.29129207e+00 1.74297139e-01
-2.79438615e-01 8.26866210e-01 7.39931941e-01 -5.86535811e-01
-9.14517760e-01 -8.27558339e-01 1.06776404e+00 -9.98953164e-01
5.69184661e-01 -6.92822874e-01 -7.71147013e-01 7.96335101e-01
1.24362111e+00 -9.93302405e-01 7.51901090e-01 5.21948516e-01
-2.35888302e-01 3.71124923e-01 -9.48402703e-01 8.58504772e-01
8.52347314e-01 -8.42619658e-01 -8.01750898e-01 6.75624073e-01
1.14554465e+00 -6.76186085e-01 -9.71810341e-01 5.29459938e-02
5.09028614e-01 -7.42034256e-01 5.22018075e-01 -9.33173776e-01
8.28325987e-01 6.30590618e-01 3.27488989e-01 -1.72978687e+00
-1.80372372e-02 -1.62304866e+00 1.27988219e-01 1.25174654e+00
9.15893435e-01 -7.61728168e-01 4.42264080e-01 9.33703542e-01
-3.87970597e-01 -5.77030420e-01 -6.51991606e-01 -2.06285164e-01
5.38720548e-01 1.68244183e-01 6.68636262e-01 9.90341723e-01
7.93160439e-01 1.30521369e+00 -6.88613534e-01 -8.90238643e-01
-6.39293343e-02 1.80205911e-01 1.43860877e+00 -8.85650873e-01
-5.48539698e-01 -5.33111870e-01 5.86822391e-01 -1.57396865e+00
3.47433239e-01 -8.46679389e-01 3.94113004e-01 -1.44196475e+00
1.81892976e-01 -2.96317548e-01 5.89666545e-01 5.56073189e-01
-2.25690633e-01 2.04824377e-02 1.06688276e-01 1.52015030e-01
-5.27869880e-01 8.33488286e-01 1.42720819e+00 -1.29472718e-01
-2.89680272e-01 2.73931056e-01 -9.83603299e-01 1.95727065e-01
7.52482772e-01 -2.89104789e-01 -4.19489086e-01 -4.88355726e-01
1.61685526e-01 7.38110185e-01 7.24391118e-02 -2.15101480e-01
1.82754219e-01 -2.22718209e-01 -1.55647367e-01 -3.38166445e-01
6.26466155e-01 4.05951083e-01 -3.57638806e-01 1.03819788e-01
-1.48343766e+00 2.35797688e-01 2.05588583e-02 2.29819179e-01
3.43018249e-02 2.83295941e-02 3.85793179e-01 -4.31749016e-01
-2.08649095e-02 -4.51670997e-02 -7.39858806e-01 5.87963343e-01
3.57274979e-01 3.11401159e-01 -1.12252522e+00 -1.41429830e+00
-4.60604548e-01 5.79860687e-01 2.24855378e-01 8.15566719e-01
3.62709641e-01 -9.85688329e-01 -1.37763858e+00 -2.55712152e-01
1.17506713e-01 -1.73241928e-01 2.28458568e-01 4.02868450e-01
-1.76647797e-01 7.39802599e-01 -9.66536552e-02 -1.93161994e-01
-1.01204979e+00 -1.62567645e-01 4.44649160e-01 -6.28692269e-01
-6.21742368e-01 1.14531291e+00 1.74166203e-01 -1.06257153e+00
7.53671825e-02 -7.67603219e-02 -1.25847638e-01 -1.66198105e-01
6.30093813e-01 2.28165165e-01 -1.91605389e-01 -3.13079506e-01
2.33535498e-01 -6.97630584e-01 -3.54859889e-01 -8.17685902e-01
8.62243891e-01 4.17842492e-02 -1.41678631e-01 4.00830656e-01
8.18651497e-01 -1.89893804e-02 -1.19921505e+00 -3.70724410e-01
-1.83651164e-01 -1.41376883e-01 -8.40556324e-01 -1.42413068e+00
-4.16994095e-01 5.67602158e-01 -3.38678420e-01 4.60811198e-01
4.91278946e-01 -1.11841045e-01 1.26442564e+00 8.52134168e-01
2.21218809e-01 -1.04994190e+00 5.33205390e-01 1.18677545e+00
1.39607000e+00 -1.17665398e+00 -5.25396109e-01 9.68528688e-02
-1.37865067e+00 9.18583393e-01 1.10331690e+00 -1.12517901e-01
-1.23789400e-01 1.83381152e-03 4.43588436e-01 -5.14280871e-02
-1.86334407e+00 3.02132964e-01 1.87662497e-01 4.49696064e-01
1.04831135e+00 2.04990387e-01 -3.59416276e-01 6.60313845e-01
-8.59756827e-01 -3.62490207e-01 6.98255718e-01 3.60538840e-01
-2.09398702e-01 -1.20877445e+00 1.24649458e-01 2.69052953e-01
-3.07204455e-01 -2.84114093e-01 -1.44784153e+00 3.93061787e-01
-8.94571960e-01 1.50115621e+00 6.51875231e-03 -5.20900726e-01
3.71111363e-01 3.72870356e-01 1.79996222e-01 -9.31903183e-01
-1.72559917e+00 -3.94425422e-01 1.27979553e+00 -3.10308516e-01
-3.60791124e-02 -3.84633064e-01 -1.06609488e+00 -6.61672354e-01
-1.90624237e-01 4.52678025e-01 2.66266704e-01 7.47030079e-01
4.22733396e-01 2.52932161e-01 8.23974192e-01 -5.81644714e-01
-1.06954825e+00 -1.92222428e+00 3.57754856e-01 5.30349970e-01
1.75882578e-01 1.89873502e-01 -4.81183827e-02 -1.39172748e-01] | [12.676294326782227, 8.215250015258789] |
d43beb1f-7630-4f5b-9cb3-24bf72c900a9 | lethal-dose-conjecture-on-data-poisoning | 2208.03309 | null | https://arxiv.org/abs/2208.03309v3 | https://arxiv.org/pdf/2208.03309v3.pdf | Lethal Dose Conjecture on Data Poisoning | Data poisoning considers an adversary that distorts the training set of machine learning algorithms for malicious purposes. In this work, we bring to light one conjecture regarding the fundamentals of data poisoning, which we call the Lethal Dose Conjecture. The conjecture states: If $n$ clean training samples are needed for accurate predictions, then in a size-$N$ training set, only $\Theta(N/n)$ poisoned samples can be tolerated while ensuring accuracy. Theoretically, we verify this conjecture in multiple cases. We also offer a more general perspective of this conjecture through distribution discrimination. Deep Partition Aggregation (DPA) and its extension, Finite Aggregation (FA) are recent approaches for provable defenses against data poisoning, where they predict through the majority vote of many base models trained from different subsets of training set using a given learner. The conjecture implies that both DPA and FA are (asymptotically) optimal -- if we have the most data-efficient learner, they can turn it into one of the most robust defenses against data poisoning. This outlines a practical approach to developing stronger defenses against poisoning via finding data-efficient learners. Empirically, as a proof of concept, we show that by simply using different data augmentations for base learners, we can respectively double and triple the certified robustness of DPA on CIFAR-10 and GTSRB without sacrificing accuracy. | ['Soheil Feizi', 'Alexander Levine', 'Wenxiao Wang'] | 2022-08-05 | null | null | null | null | ['data-poisoning'] | ['adversarial'] | [-5.86996647e-03 -3.80921029e-02 -2.14652434e-01 7.36414939e-02
-1.14834797e+00 -1.13048744e+00 2.39128441e-01 4.49757963e-01
-6.75125480e-01 1.02866304e+00 -3.24892551e-01 -6.91006243e-01
-1.89204678e-01 -1.05550945e+00 -1.00598824e+00 -1.25128925e+00
-2.40367115e-01 6.09541833e-01 2.03831151e-01 -1.44747078e-01
1.58003300e-01 6.52750254e-01 -1.07829940e+00 4.41316903e-01
7.04632759e-01 7.63527334e-01 -6.62340045e-01 6.68189526e-01
4.23300475e-01 9.43580449e-01 -1.03814554e+00 -7.94737339e-01
7.73898900e-01 -4.91764635e-01 -9.85336065e-01 -2.96171695e-01
3.51280749e-01 -4.47580516e-01 -5.15540779e-01 1.31127119e+00
4.69027728e-01 -3.83092940e-01 5.98888814e-01 -1.63068914e+00
-3.53747129e-01 1.04996741e+00 -5.98235726e-01 2.86076546e-01
1.00101866e-01 5.36793888e-01 9.01445270e-01 -6.54941127e-02
1.72868315e-02 1.13698542e+00 6.88957512e-01 1.09991300e+00
-1.26872873e+00 -1.15366626e+00 -1.58115134e-01 5.69555163e-02
-1.31003416e+00 -1.60343036e-01 4.86599475e-01 -9.52108130e-02
6.68478668e-01 7.82498837e-01 3.20301920e-01 1.28030396e+00
1.66969255e-01 7.84817696e-01 1.28810441e+00 -9.16600674e-02
5.47092438e-01 1.66896239e-01 4.74182397e-01 5.46969295e-01
1.17297804e+00 3.77195239e-01 -5.45553446e-01 -1.04999506e+00
1.85840353e-02 -1.57305136e-01 -5.00579059e-01 -1.85908929e-01
-4.32687432e-01 1.10279906e+00 3.05309713e-01 5.41570298e-02
-4.37823720e-02 3.48018587e-01 4.72098053e-01 4.37909901e-01
2.25119725e-01 6.32703245e-01 -5.81288576e-01 4.39340115e-01
-4.23354685e-01 6.51721656e-01 9.66982424e-01 5.07274926e-01
3.00387114e-01 1.18014984e-01 1.01681471e-01 -1.26023278e-01
2.01330557e-01 9.25639629e-01 3.78878355e-01 -6.91600919e-01
4.77723390e-01 2.29847819e-01 2.10967854e-01 -5.90243459e-01
-3.09306741e-01 -4.41266447e-01 -7.51573384e-01 4.74313527e-01
8.63833666e-01 -2.66902417e-01 -4.96452391e-01 2.11868739e+00
5.19522786e-01 7.06197768e-02 3.43743026e-01 3.97338092e-01
2.15248957e-01 2.87155747e-01 4.69663769e-01 -4.50503290e-01
1.31721401e+00 -8.62754940e-04 -2.23239541e-01 2.70472139e-01
1.05576313e+00 9.96272936e-02 9.90214586e-01 9.84824657e-01
-9.43778038e-01 7.66748488e-02 -1.30189121e+00 3.97369444e-01
-4.88018900e-01 -8.03358555e-01 5.57180703e-01 1.42946482e+00
-6.57486737e-01 5.99490285e-01 -7.03131437e-01 8.54257718e-02
9.48582411e-01 8.50656390e-01 -3.32950443e-01 6.58556959e-03
-1.31928456e+00 6.20203495e-01 2.63913810e-01 -7.41509378e-01
-1.55832136e+00 -8.40051770e-01 -3.28011215e-01 -1.44966036e-01
1.91671744e-01 -6.95703387e-01 1.10707605e+00 -3.46595228e-01
-7.08355665e-01 7.08917975e-01 4.14106190e-01 -1.29490340e+00
6.60926640e-01 -1.45309180e-01 -3.93582471e-02 3.65224540e-01
-2.98119575e-01 3.53332907e-01 7.38310814e-01 -1.48958123e+00
-8.04529130e-01 -7.46960759e-01 4.49545443e-01 -7.41434917e-02
-6.01136446e-01 1.01222880e-02 7.07763851e-01 -2.68353641e-01
-3.45595688e-01 -7.57932127e-01 -4.51472342e-01 -1.53231695e-01
-6.90872848e-01 -4.20495182e-01 1.16656888e+00 -9.68904495e-02
9.21253562e-01 -2.02203512e+00 -2.89817989e-01 3.19819123e-01
6.83185577e-01 5.47705352e-01 -4.76057045e-02 1.23722307e-01
-9.56771523e-02 6.34090483e-01 -5.41496456e-01 -4.80687134e-02
1.24149071e-02 2.02830747e-01 -9.79878366e-01 1.05930376e+00
-2.68336505e-01 6.63260043e-01 -7.80861616e-01 5.66690881e-03
-2.26487339e-01 2.85402477e-01 -6.34694695e-01 -4.71280254e-02
-4.32131141e-01 3.73418599e-01 -4.11221296e-01 3.95224035e-01
9.62479830e-01 5.21844402e-02 1.35544404e-01 2.69044843e-02
4.70545709e-01 2.53941447e-01 -8.94503593e-01 7.87893474e-01
2.10398942e-01 2.15914007e-02 -7.45036602e-02 -9.99650240e-01
5.00670731e-01 3.76502723e-01 5.65128684e-01 -5.59025526e-01
4.32865709e-01 3.39118242e-01 1.07313924e-01 -2.82393575e-01
-7.77337030e-02 -4.94133770e-01 -4.67705429e-01 8.73647213e-01
-2.78650343e-01 4.25106511e-02 -1.25339627e-01 1.95527449e-01
1.58804071e+00 -6.90803885e-01 8.82775038e-02 -5.70060551e-01
4.29814070e-01 1.45328432e-01 4.71321374e-01 1.20339632e+00
-4.68238384e-01 1.12191558e-01 6.63932145e-01 -5.69271624e-01
-9.91105556e-01 -1.12092793e+00 -1.06362157e-01 8.96595418e-01
1.47690803e-01 -2.49274060e-01 -1.37503338e+00 -1.48067725e+00
1.86838791e-01 8.00047517e-01 -7.41396904e-01 -4.11585897e-01
-3.88230115e-01 -1.36811006e+00 1.58547628e+00 2.33632892e-01
6.19602859e-01 -6.57536685e-01 -8.45383227e-01 -3.25346351e-01
3.47737269e-03 -6.35898948e-01 -1.29750147e-01 6.63110316e-01
-8.56233120e-01 -1.81200838e+00 -7.38840103e-02 -1.46946773e-01
4.91899461e-01 3.90979588e-01 8.43872786e-01 4.78499502e-01
-8.88283178e-03 2.91709721e-01 -1.84394389e-01 -7.89766848e-01
-7.96233594e-01 2.20489763e-02 5.75003564e-01 -4.09260690e-01
6.51892066e-01 -6.61639154e-01 -4.16099131e-01 1.15639582e-01
-1.34101844e+00 -7.51761973e-01 3.07594836e-01 3.74897629e-01
4.61176634e-01 4.23410088e-01 7.12529659e-01 -1.30365610e+00
6.68805897e-01 -7.19375849e-01 -6.18789375e-01 2.72725940e-01
-5.08009434e-01 1.75311089e-01 1.11846900e+00 -5.17663956e-01
-2.04842195e-01 1.04804691e-02 -4.58749056e-01 -4.64214832e-01
-2.39248022e-01 -1.55504480e-01 -5.85766971e-01 -2.92655289e-01
1.43229997e+00 3.37148190e-01 -1.06526911e-01 -3.56707513e-01
3.07049364e-01 5.39004982e-01 4.66250092e-01 -8.64672542e-01
1.13171506e+00 6.54781342e-01 3.38027060e-01 -6.48551762e-01
-9.80742395e-01 -1.12631857e-01 -2.75968432e-01 4.11235929e-01
5.43110132e-01 -6.18133485e-01 -1.20837200e+00 6.11498058e-01
-9.76740539e-01 -4.52966213e-01 -6.11602664e-01 -5.58028137e-03
-4.39792514e-01 6.04692400e-01 -5.24719000e-01 -9.67440844e-01
-5.77392042e-01 -1.00317562e+00 3.73669595e-01 -9.24527124e-02
1.47229303e-02 -8.65152061e-01 2.47140341e-02 2.13504866e-01
9.36337784e-02 6.52738154e-01 1.15644050e+00 -1.70326328e+00
-1.89767331e-01 -3.25068951e-01 2.75847346e-01 4.04958725e-01
-1.28262699e-01 -4.60986018e-01 -1.30252826e+00 -7.20380604e-01
7.01692998e-01 -7.38438845e-01 9.02667344e-01 -1.88320335e-02
1.68044865e+00 -1.14131284e+00 -2.49591246e-01 5.27753532e-01
1.45274365e+00 3.24875146e-01 7.11697698e-01 1.65593356e-01
5.04950523e-01 2.18317807e-01 1.95842415e-01 5.65811813e-01
-2.89289076e-02 6.98704720e-02 7.53129601e-01 2.62728870e-01
2.45621279e-01 -2.50109911e-01 4.57379133e-01 1.96269210e-02
2.65278935e-01 -5.41449487e-01 -7.17345119e-01 2.23183826e-01
-1.13311458e+00 -1.10785174e+00 -3.58168930e-01 2.60443711e+00
1.23813653e+00 2.36523777e-01 6.09364927e-01 7.94797361e-01
3.50189179e-01 -6.49838075e-02 -8.56589198e-01 -2.06986427e-01
-2.86635011e-01 6.73760831e-01 1.02014267e+00 5.36416829e-01
-1.32015204e+00 5.55559754e-01 6.80923510e+00 1.20441747e+00
-8.87788951e-01 4.26680833e-01 1.07812572e+00 -1.99163720e-01
-4.85288829e-01 -2.71342337e-01 -9.33645546e-01 3.99624556e-01
1.37152946e+00 -2.77920902e-01 2.64432311e-01 9.66222227e-01
-3.23772252e-01 -4.16556820e-02 -1.27153194e+00 6.38398945e-01
6.16346346e-03 -1.25714445e+00 2.40943223e-01 4.91311371e-01
4.89041895e-01 -9.22088102e-02 5.01764655e-01 2.47229949e-01
8.38232696e-01 -1.23570001e+00 5.76498210e-01 -2.15501904e-01
6.86457932e-01 -1.28345096e+00 5.44463038e-01 8.21571171e-01
-4.98436958e-01 -5.46587288e-01 -5.72321653e-01 2.42995799e-01
-5.98710954e-01 4.12271082e-01 -5.83044350e-01 2.83667356e-01
6.24131262e-01 -1.74330011e-01 -5.42272449e-01 8.84151459e-01
-2.59824038e-01 9.18572068e-01 -6.94301784e-01 7.36817047e-02
1.82787254e-01 4.49928015e-01 5.05811453e-01 9.84268427e-01
-1.24746829e-01 5.78559101e-01 8.98516700e-02 5.34738243e-01
-5.09202361e-01 -3.59396458e-01 -7.90875852e-01 3.26021254e-01
7.36755490e-01 7.25661814e-01 -5.04368722e-01 -2.34011948e-01
3.24572653e-01 5.73572993e-01 1.73480377e-01 -8.16100016e-02
-1.02969933e+00 -7.22137615e-02 8.91252100e-01 1.59399956e-01
-1.05272666e-01 3.89006957e-02 -7.05149949e-01 -8.81485105e-01
-4.88570511e-01 -1.46439862e+00 8.86298478e-01 1.11696571e-01
-1.56629908e+00 4.41182971e-01 -5.27939526e-03 -9.72045004e-01
2.04679757e-01 -6.66603446e-01 -4.72195864e-01 4.91070688e-01
-1.07253039e+00 -8.50678742e-01 2.60433763e-01 9.07651365e-01
1.36029735e-01 -8.84153023e-02 1.06399620e+00 -6.74862191e-02
-5.28391540e-01 1.28708827e+00 9.24906954e-02 1.21724084e-01
3.09473991e-01 -1.24928367e+00 2.80798316e-01 9.87814724e-01
3.80276918e-01 4.85834837e-01 9.69184041e-01 -6.22548103e-01
-1.39473915e+00 -1.27265167e+00 4.93454754e-01 -9.41506386e-01
3.65527511e-01 -2.54937351e-01 -9.98615384e-01 3.83287400e-01
-7.55538000e-03 4.73258346e-02 1.20062280e+00 -2.84359157e-01
-1.03417790e+00 -1.40928522e-01 -1.96192789e+00 2.05864176e-01
7.27450669e-01 -4.31143284e-01 -3.76845211e-01 5.70193648e-01
7.04009891e-01 -2.11495712e-01 -5.47747850e-01 3.15200746e-01
2.04018265e-01 -1.03969347e+00 9.40329254e-01 -1.12680340e+00
3.49852256e-02 -2.29289338e-01 -4.59708512e-01 -8.90837848e-01
7.70975277e-02 -7.34824300e-01 -5.21208465e-01 7.34987199e-01
3.25776190e-01 -7.66351938e-01 1.00065339e+00 6.76132441e-01
3.09776485e-01 -7.94857621e-01 -1.14299977e+00 -9.46897209e-01
1.03277826e+00 -5.14324546e-01 8.78301322e-01 9.75616813e-01
2.01257225e-02 -1.55182868e-01 -3.44137281e-01 5.51578879e-01
1.18461537e+00 -4.85835463e-01 7.22862720e-01 -1.15655839e+00
-5.45293152e-01 -2.49667048e-01 -2.66869366e-01 -8.18974674e-01
9.90443379e-02 -7.99246609e-01 -9.64046046e-02 -7.37735927e-01
3.90536398e-01 -7.48945832e-01 -4.54207152e-01 9.08883214e-01
-4.14746590e-02 6.33188725e-01 1.67464837e-01 3.21632296e-01
-6.17880940e-01 1.71819665e-02 6.49254203e-01 -2.87289023e-01
1.00974828e-01 2.07016319e-01 -1.36659956e+00 6.25293970e-01
1.21536517e+00 -1.10574591e+00 -4.89335418e-01 -1.15030855e-01
1.27339229e-01 -1.92543805e-01 4.67541933e-01 -1.26233530e+00
2.64566004e-01 -3.29202384e-01 2.81941324e-01 -3.39735448e-01
-8.61486122e-02 -8.29179347e-01 2.64101718e-02 1.29497790e+00
-6.19234383e-01 -8.59418809e-02 1.40739456e-01 7.29350328e-01
6.09023631e-01 -3.33499759e-01 1.22871172e+00 -1.41980693e-01
1.93885863e-01 6.82215631e-01 -4.02301043e-01 4.02054340e-01
1.39122081e+00 2.60463327e-01 -9.30263042e-01 -1.16618715e-01
-2.47462958e-01 -2.67006345e-02 5.22740841e-01 -2.99715042e-01
5.18848419e-01 -8.91901791e-01 -7.77300179e-01 1.28122702e-01
-2.04159126e-01 4.58843596e-02 -7.87650570e-02 6.21395171e-01
-4.81139272e-01 2.36382112e-01 1.00380063e-01 -2.76996136e-01
-1.39323783e+00 1.00215137e+00 7.39568770e-01 -3.70552331e-01
-2.06857041e-01 1.03858256e+00 1.73544779e-01 6.02908507e-02
4.26991969e-01 4.12233546e-02 2.54881650e-01 -3.75790328e-01
1.04751801e+00 3.59475911e-01 1.50109539e-02 -3.23542923e-01
-2.62587756e-01 -1.46126598e-01 -2.30807543e-01 -2.15656254e-02
1.11911643e+00 4.62601542e-01 -1.88514031e-02 -1.06185980e-01
9.96798754e-01 3.57704878e-01 -1.08267260e+00 4.36929613e-02
-4.53400575e-02 -3.99238855e-01 -4.25144613e-01 -8.92501056e-01
-1.03279245e+00 9.17778134e-01 5.89428902e-01 7.56444395e-01
1.17325377e+00 3.54829542e-02 8.50182652e-01 6.95451677e-01
6.89925730e-01 -4.52241957e-01 3.27125788e-02 1.65725514e-01
4.23454404e-01 -6.63013875e-01 1.14365309e-01 -2.13459104e-01
-4.13022190e-01 7.92832971e-01 5.81891477e-01 -3.74422312e-01
6.72313094e-01 5.51544428e-01 -2.35179737e-01 -1.46144822e-01
-8.67641747e-01 1.91772744e-01 -3.51462394e-01 9.60574389e-01
-2.26083606e-01 3.89654189e-02 -2.98881710e-01 1.04333353e+00
-2.57866889e-01 -5.02953649e-01 4.88784730e-01 6.89994097e-01
-9.52746689e-01 -1.35735822e+00 -5.71618199e-01 4.04246479e-01
-1.01627171e+00 6.18849024e-02 -6.53166652e-01 7.60335147e-01
4.21170354e-01 1.20885706e+00 -3.03750575e-01 -8.10885131e-01
8.28964710e-02 4.82675768e-02 4.77349699e-01 -2.94099480e-01
-9.71496582e-01 -5.31217515e-01 -2.65025496e-01 -3.76648009e-01
-1.09617904e-01 -4.90621895e-01 -1.25133336e+00 -1.01412344e+00
-2.89528340e-01 6.09239995e-01 2.08570361e-01 1.11166584e+00
4.41475511e-02 -1.43942788e-01 1.06912959e+00 -6.72916844e-02
-1.43959403e+00 -5.86643457e-01 -7.30595648e-01 3.42421055e-01
2.94809997e-01 -1.43310457e-01 -8.47992718e-01 -2.02031210e-01] | [5.798232078552246, 7.577357292175293] |
acb67957-8c2e-4de8-a9cd-e922f1daa237 | aligning-latent-and-image-spaces-to-connect | 2104.06954 | null | https://arxiv.org/abs/2104.06954v1 | https://arxiv.org/pdf/2104.06954v1.pdf | Aligning Latent and Image Spaces to Connect the Unconnectable | In this work, we develop a method to generate infinite high-resolution images with diverse and complex content. It is based on a perfectly equivariant generator with synchronous interpolations in the image and latent spaces. Latent codes, when sampled, are positioned on the coordinate grid, and each pixel is computed from an interpolation of the nearby style codes. We modify the AdaIN mechanism to work in such a setup and train the generator in an adversarial setting to produce images positioned between any two latent vectors. At test time, this allows for generating complex and diverse infinite images and connecting any two unrelated scenes into a single arbitrarily large panorama. Apart from that, we introduce LHQ: a new dataset of \lhqsize high-resolution nature landscapes. We test the approach on LHQ, LSUN Tower and LSUN Bridge and outperform the baselines by at least 4 times in terms of quality and diversity of the produced infinite images. The project page is located at https://universome.github.io/alis. | ['Mohamed Elhoseiny', 'Grigorii Sotnikov', 'Ivan Skorokhodov'] | 2021-04-14 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Skorokhodov_Aligning_Latent_and_Image_Spaces_To_Connect_the_Unconnectable_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Skorokhodov_Aligning_Latent_and_Image_Spaces_To_Connect_the_Unconnectable_ICCV_2021_paper.pdf | iccv-2021-1 | ['infinite-image-generation'] | ['computer-vision'] | [ 4.85955328e-01 1.01711705e-01 2.49402091e-01 -8.13158154e-02
-1.19581366e+00 -1.04887056e+00 8.41709077e-01 -6.42085373e-01
-1.07967764e-01 8.79387856e-01 3.19061399e-01 8.88004899e-02
2.57319063e-01 -1.05181336e+00 -1.02317119e+00 -7.75593102e-01
5.52560352e-02 3.86642784e-01 2.56145671e-02 -3.86258453e-01
6.40596971e-02 3.69348109e-01 -1.53258240e+00 3.41115177e-01
8.92173946e-01 6.25541866e-01 3.33470941e-01 1.14029932e+00
2.88338721e-01 5.78070104e-01 -5.24209499e-01 -6.77266002e-01
7.18483269e-01 -7.68646717e-01 -6.49560094e-01 8.03126693e-02
8.22107255e-01 -5.07112026e-01 -1.79820195e-01 9.61359978e-01
5.86979508e-01 -9.92104113e-02 6.89325988e-01 -1.16304767e+00
-9.29785967e-01 4.61922705e-01 -5.62778890e-01 -1.74459234e-01
4.85516161e-01 4.12371308e-01 9.92042601e-01 -8.94901574e-01
9.70419526e-01 1.25641823e+00 7.27285683e-01 5.13312161e-01
-1.78639984e+00 -5.24202287e-01 -2.95068413e-01 -3.04963529e-01
-1.37915730e+00 -7.13093102e-01 5.33661544e-01 -5.04603982e-01
5.30092418e-01 5.34993529e-01 6.02577031e-01 1.61153650e+00
-1.18808456e-01 3.78329962e-01 1.22266829e+00 -4.73651856e-01
2.14638770e-01 -6.48552850e-02 -5.02288282e-01 5.59453666e-01
-8.30433220e-02 2.55062759e-01 -3.73579085e-01 -2.57958680e-01
1.27305245e+00 -2.93626398e-01 -3.27393889e-01 -3.37814867e-01
-1.61016774e+00 9.44446087e-01 6.26733363e-01 6.80313446e-03
-3.45502406e-01 3.18100989e-01 -1.04953460e-01 1.18068941e-01
4.26305622e-01 6.99520230e-01 -7.73474947e-03 1.52185813e-01
-1.02164829e+00 4.47604209e-01 5.58454037e-01 1.02900779e+00
8.33604991e-01 9.35947746e-02 -2.29203060e-01 7.76536286e-01
-2.85904914e-01 8.02002728e-01 1.62256569e-01 -1.64671266e+00
3.84365946e-01 4.99008186e-02 2.81584352e-01 -7.67476022e-01
8.70088711e-02 -1.98092133e-01 -1.12930858e+00 6.14961863e-01
3.73488218e-01 -1.85902238e-01 -9.18982744e-01 1.80890346e+00
1.24682009e-01 2.17840582e-01 6.47950023e-02 8.78219247e-01
5.25766373e-01 9.39409852e-01 -2.44915932e-01 1.33438870e-01
1.15663564e+00 -9.94124889e-01 -4.85917032e-01 -1.17694540e-03
2.29316488e-01 -9.45658326e-01 1.49346113e+00 2.86816895e-01
-1.26034153e+00 -7.00021029e-01 -1.03953099e+00 -2.74962187e-01
-3.43188077e-01 1.47316858e-01 3.62160176e-01 3.05179238e-01
-1.31157207e+00 6.24096751e-01 -4.76319730e-01 -2.83861309e-01
2.20137507e-01 -1.26822889e-01 -4.92000192e-01 2.15030909e-02
-1.14396417e+00 5.35633802e-01 3.51704240e-01 -9.38444808e-02
-1.05344796e+00 -6.56478941e-01 -9.13483441e-01 -3.13797563e-01
-1.76409632e-02 -1.01800036e+00 9.12218571e-01 -1.18858516e+00
-1.49372685e+00 1.06148601e+00 1.27564045e-02 -3.28076571e-01
1.08882523e+00 -1.03391692e-01 -2.34357208e-01 9.52949002e-02
4.49078768e-01 1.21339369e+00 1.04781449e+00 -1.48005223e+00
-3.44983876e-01 2.48897783e-02 1.40881509e-01 3.19101661e-01
1.07915707e-01 -2.66496241e-01 -4.38356519e-01 -8.57885540e-01
-5.67158237e-02 -1.05589211e+00 -2.93342233e-01 3.12944055e-01
-5.55275679e-01 4.13691670e-01 4.65908110e-01 -7.64857292e-01
7.04168141e-01 -2.34407759e+00 4.33072478e-01 3.64096202e-02
1.24142580e-01 -2.33407319e-01 -4.07415450e-01 3.47435176e-01
-1.62712872e-01 2.75747329e-01 -6.11549079e-01 -3.70894879e-01
2.57845987e-02 2.46444196e-01 -7.02437699e-01 4.19540465e-01
2.79570192e-01 1.09215128e+00 -9.47448850e-01 -2.76678771e-01
1.74605489e-01 5.33726454e-01 -5.54962933e-01 3.66339654e-01
-5.56383133e-01 7.70418227e-01 -5.20141385e-02 4.16608661e-01
8.30838144e-01 -2.28034958e-01 -1.62894696e-01 4.60719177e-03
-2.60470212e-01 -2.44192675e-01 -1.13552940e+00 2.07370591e+00
-4.15489733e-01 7.24708498e-01 -1.47512034e-01 -3.82485151e-01
8.56325984e-01 2.52564758e-01 7.98821673e-02 -6.76926911e-01
-4.59359705e-01 2.17355937e-01 -5.80244064e-01 -2.52309740e-01
6.36205852e-01 -5.57104452e-03 -3.31197739e-01 3.97714853e-01
-1.59653276e-02 -4.73464161e-01 1.98442146e-01 1.77975148e-01
9.05092061e-01 5.04644096e-01 8.99528563e-02 -7.55182505e-02
2.79290229e-01 3.54550802e-03 2.23528579e-01 8.14081132e-01
1.50301009e-01 1.40987694e+00 6.52262986e-01 -3.84114534e-01
-1.90565431e+00 -1.74833214e+00 -2.37375781e-01 8.65032673e-01
-1.91599727e-02 -2.29241356e-01 -8.12252164e-01 -3.97630841e-01
-3.12815398e-01 8.07304084e-01 -8.07611108e-01 1.97023585e-01
-5.37685752e-01 -4.38180178e-01 7.07585216e-01 2.32742310e-01
7.75037706e-01 -1.21671939e+00 -4.25281465e-01 -1.61806181e-01
-5.10364830e-01 -9.52287853e-01 -4.48090643e-01 -5.20991795e-02
-3.99090588e-01 -7.48653769e-01 -1.16397786e+00 -6.53199911e-01
6.84946775e-01 3.11730579e-02 1.55270851e+00 -3.27176273e-01
-3.59993219e-01 -4.23154235e-03 -3.23292226e-01 5.33618927e-02
-6.31215990e-01 1.44673884e-01 -2.41951615e-01 -7.33434281e-04
-3.86246771e-01 -6.69435263e-01 -6.74695611e-01 2.85736382e-01
-1.21332526e+00 5.45877755e-01 5.45968115e-01 9.54537928e-01
8.13585460e-01 -6.28504530e-02 2.76998281e-01 -8.16883624e-01
3.88117522e-01 -4.73677069e-01 -7.86338031e-01 1.31286249e-01
2.76909545e-02 2.04920784e-01 8.49817693e-01 -3.50087196e-01
-9.95423019e-01 1.80521443e-01 -2.39946410e-01 -4.17456418e-01
-3.25016916e-01 -1.12371705e-01 -2.56803393e-01 2.08099589e-01
1.14967811e+00 3.01329195e-01 -3.44193965e-01 -2.92534351e-01
1.04820406e+00 3.91162723e-01 1.02895367e+00 -5.51251173e-01
1.21849120e+00 6.83832943e-01 -2.11642236e-01 -7.10051715e-01
-6.27649546e-01 1.01893999e-01 -8.31427753e-01 -7.03726411e-02
1.17354667e+00 -1.10333872e+00 -7.09682107e-02 5.18189609e-01
-1.08698809e+00 -7.15492845e-01 -7.21701741e-01 6.01177141e-02
-8.81342590e-01 4.48932089e-02 -4.61650789e-01 -3.44722658e-01
-1.52090251e-01 -1.07207668e+00 1.44572079e+00 7.50276446e-02
-2.51889735e-01 -8.13227892e-01 3.21379721e-01 5.52253723e-02
4.01728034e-01 8.39230239e-01 6.80069625e-01 1.16678521e-01
-8.03299189e-01 4.28101011e-02 -1.67810738e-01 3.29133749e-01
-2.96645239e-02 1.92880183e-01 -1.06952250e+00 -3.38536918e-01
-2.85316169e-01 -6.72227621e-01 8.09927344e-01 2.30063856e-01
9.10670400e-01 -3.95913839e-01 2.23940630e-02 1.17168701e+00
1.62887704e+00 -3.31671268e-01 1.10382664e+00 3.91499460e-01
6.44221723e-01 4.66927052e-01 1.44640341e-01 4.41210359e-01
1.30632713e-01 7.54181325e-01 3.88177484e-01 -3.92977893e-01
-3.17949146e-01 -4.87180889e-01 5.25703311e-01 3.34983855e-01
-1.37275800e-01 -1.81896701e-01 -6.43967986e-01 5.07663190e-01
-1.50002909e+00 -1.35415292e+00 -9.36440378e-02 2.16708922e+00
9.61199999e-01 -1.23224758e-01 6.84804320e-02 -1.96800813e-01
7.43884027e-01 4.72468227e-01 -5.76120436e-01 -1.88942626e-01
-6.87610269e-01 2.26529762e-01 5.25977254e-01 6.20706439e-01
-1.00214994e+00 1.01525939e+00 6.47151899e+00 8.72214079e-01
-1.05868673e+00 3.33683044e-02 9.18727756e-01 -2.15424746e-01
-6.97447717e-01 4.05901447e-02 -4.85437244e-01 4.85970378e-01
8.00728798e-01 -4.39607129e-02 7.96429992e-01 6.35276318e-01
2.43431646e-02 -5.72899580e-02 -7.77796745e-01 8.06856751e-01
-3.05437427e-02 -1.45060134e+00 1.09144345e-01 7.66476095e-02
1.23386633e+00 2.55592950e-02 5.00741184e-01 2.17539500e-02
6.46343350e-01 -1.30255187e+00 9.54562247e-01 6.28714204e-01
1.39645314e+00 -6.49580002e-01 2.38473430e-01 2.26181820e-01
-1.01009345e+00 2.86594957e-01 -5.33438504e-01 1.87469736e-01
2.66786158e-01 3.96762371e-01 -4.25084114e-01 5.35875142e-01
8.08117032e-01 5.79305828e-01 -7.67094016e-01 5.17297268e-01
-4.53604460e-01 3.55312616e-01 -3.28797400e-01 6.85317516e-01
1.15891553e-01 -5.75553417e-01 4.68813360e-01 9.84026194e-01
6.12861693e-01 -5.26116081e-02 -9.89753306e-02 1.35885429e+00
-1.81566462e-01 -1.62257627e-01 -1.12795198e+00 3.35594296e-01
3.52798790e-01 1.31360328e+00 -6.12570643e-01 -3.15120488e-01
-1.43295810e-01 1.66219485e+00 2.86221474e-01 4.62220699e-01
-1.15184677e+00 -2.64496565e-01 5.24869680e-01 1.89232707e-01
3.34932148e-01 -1.65201247e-01 -2.81408578e-01 -1.29949319e+00
2.21204966e-01 -9.80702341e-01 1.66735098e-01 -1.33178830e+00
-1.28142369e+00 9.77864027e-01 -5.03718853e-02 -1.47140491e+00
-4.16242123e-01 -9.95663404e-02 -4.84709799e-01 1.20604181e+00
-1.24405611e+00 -1.31078959e+00 -6.13505840e-01 6.46899045e-01
4.55541104e-01 -3.38227339e-02 1.12663972e+00 2.92335805e-02
-3.16777885e-01 4.68354493e-01 4.08297181e-01 2.12465525e-01
7.79499888e-01 -1.37561929e+00 9.62593138e-01 1.04870021e+00
2.29382932e-01 2.72367060e-01 6.57548785e-01 -5.18850088e-01
-8.77927244e-01 -1.33822691e+00 5.54998517e-01 -7.54115701e-01
4.57618952e-01 -7.44748592e-01 -7.57772803e-01 8.24523509e-01
6.04532480e-01 1.42536208e-01 3.57851207e-01 -4.67661768e-01
-5.36520720e-01 1.69881448e-01 -1.09314835e+00 1.01896882e+00
1.25138772e+00 -5.71589172e-01 -2.13590145e-01 4.25283641e-01
8.49748492e-01 -5.47570229e-01 -8.81314397e-01 2.02849522e-01
5.43075204e-01 -1.29643917e+00 1.24048805e+00 -2.05446631e-01
1.02440381e+00 -4.97825593e-01 -3.49470317e-01 -1.33556783e+00
-5.30596495e-01 -8.12927306e-01 3.06088448e-01 1.22364593e+00
4.24108267e-01 -5.28658330e-01 6.24862790e-01 1.41784742e-01
1.48573250e-01 -3.65602940e-01 -7.07581043e-01 -7.11058497e-01
3.15532088e-01 -2.96406038e-02 6.79587185e-01 8.17191482e-01
-7.96999753e-01 4.42717254e-01 -6.21384680e-01 1.53443038e-01
8.92474234e-01 2.60112047e-01 1.03679872e+00 -8.22670758e-01
-5.27893782e-01 -3.24686855e-01 -1.26909435e-01 -8.86179209e-01
8.85057151e-02 -8.75215352e-01 -2.04527956e-02 -1.34900105e+00
9.44305733e-02 -5.10731339e-01 2.29634464e-01 2.03069493e-01
-1.15703769e-01 8.21666360e-01 2.08864331e-01 3.94912422e-01
-3.42926234e-01 5.15879571e-01 1.31237042e+00 1.40854586e-02
3.37529629e-02 -3.56007278e-01 -5.86397350e-01 5.21516681e-01
8.78449082e-01 -2.91197568e-01 -3.89658272e-01 -6.42084479e-01
2.26008773e-01 1.54795066e-01 7.42178619e-01 -1.17792678e+00
-2.80652136e-01 -5.71052963e-03 8.31825018e-01 -3.17374140e-01
4.67851311e-01 -5.52445889e-01 8.33791554e-01 1.21621817e-01
-5.88728070e-01 2.40968853e-01 -3.23149376e-02 3.50466251e-01
5.70671223e-02 2.50085536e-02 1.04140615e+00 -3.57594639e-01
-6.24116063e-01 2.30161980e-01 -4.74865846e-02 2.18806297e-01
1.01395333e+00 -1.51936620e-01 -4.53730643e-01 -5.37126124e-01
-6.99147522e-01 -6.36020303e-02 1.17998874e+00 3.83298457e-01
5.13280690e-01 -1.66699743e+00 -1.11419570e+00 5.69613039e-01
2.09287241e-01 2.35867932e-01 2.37628683e-01 3.31316054e-01
-9.91197288e-01 -6.86995685e-02 -5.80310404e-01 -6.48621678e-01
-8.25451791e-01 4.03963417e-01 3.78972411e-01 -2.45156601e-01
-7.98021197e-01 7.45519817e-01 5.02737999e-01 -6.23225629e-01
-7.48765469e-02 8.32655504e-02 4.69292812e-02 -1.01930745e-01
4.79126334e-01 1.04883097e-01 -4.71940160e-01 -8.31465304e-01
1.74701452e-01 7.53355563e-01 4.25363839e-01 -6.80346429e-01
1.26979566e+00 -1.01864241e-01 -1.67941093e-01 3.80568743e-01
1.24145973e+00 3.78176868e-01 -1.66892087e+00 5.90880513e-02
-3.62155199e-01 -6.77920282e-01 -4.33765292e-01 -6.32212639e-01
-9.63909149e-01 5.76553524e-01 6.90137744e-01 1.80337116e-01
1.18289053e+00 5.63393310e-02 6.38971508e-01 -2.81911530e-02
3.40952933e-01 -5.75814843e-01 2.10639581e-01 3.65373939e-01
1.23796797e+00 -1.06437147e+00 -1.74991116e-01 -4.86540124e-02
-7.79680371e-01 9.00322497e-01 3.83306652e-01 -5.31809866e-01
2.25036398e-01 3.65672469e-01 3.45952988e-01 5.24991155e-02
-6.14856958e-01 -1.10636212e-01 -9.67639163e-02 8.29528511e-01
3.03947121e-01 1.19453937e-01 1.72580987e-01 -3.14034857e-02
-7.31126904e-01 -1.30267859e-01 6.35411024e-01 5.21121800e-01
-1.00042433e-01 -1.08865905e+00 -6.88126683e-01 1.43466160e-01
-1.81640223e-01 -2.33902514e-01 -2.87946880e-01 5.79083383e-01
3.56410623e-01 5.53609073e-01 2.19641030e-01 -3.04157138e-01
3.10696900e-01 -1.81949526e-01 3.92329037e-01 -3.00448716e-01
-2.48548359e-01 6.16439711e-03 1.94642674e-02 -6.47958040e-01
-1.52086034e-01 -6.43749118e-01 -8.56214941e-01 -4.00388598e-01
8.49065259e-02 -1.13043096e-02 3.09792608e-01 2.14261055e-01
3.06034714e-01 4.23038632e-01 8.64431500e-01 -1.24051762e+00
-3.63461882e-01 -8.22075784e-01 -5.41262031e-01 7.35412717e-01
3.78498048e-01 -2.48799145e-01 -4.13095325e-01 5.76857865e-01] | [11.575419425964355, -0.4744025766849518] |
7dde48f4-8ccb-4fa3-a166-9ff2b393de65 | exemplar-based-image-colorization-with-a | 2209.05775 | null | https://arxiv.org/abs/2209.05775v1 | https://arxiv.org/pdf/2209.05775v1.pdf | Exemplar-Based Image Colorization with A Learning Framework | Image learning and colorization are hot spots in multimedia domain. Inspired by the learning capability of humans, in this paper, we propose an automatic colorization method with a learning framework. This method can be viewed as a hybrid of exemplar-based and learning-based method, and it decouples the colorization process and learning process so as to generate various color styles for the same gray image. The matching process in the exemplar-based colorization method can be regarded as a parameterized function, and we employ a large amount of color images as the training samples to fit the parameters. During the training process, the color images are the ground truths, and we learn the optimal parameters for the matching process by minimizing the errors in terms of the parameters for the matching function. To deal with images with various compositions, a global feature is introduced, which can be used to classify the images with respect to their compositions, and then learn the optimal matching parameters for each image category individually. What's more, a spatial consistency based post-processing is design to smooth the extracted color information from the reference image to remove matching errors. Extensive experiments are conducted to verify the effectiveness of the method, and it achieves comparable performance against the state-of-the-art colorization algorithms. | ['Yong liu', 'Jie Ren', 'Jiandang Yang', 'Zhenfeng Xue'] | 2022-09-13 | null | null | null | null | ['colorization'] | ['computer-vision'] | [-7.27241626e-03 -5.15657306e-01 -8.71903375e-02 -2.98099548e-01
-5.00341415e-01 -3.14550132e-01 2.35225275e-01 -2.08804116e-01
-5.01571476e-01 2.95809269e-01 -1.41629025e-01 4.04307768e-02
-3.02819051e-02 -8.33172321e-01 -5.49602211e-01 -1.09416771e+00
4.60309505e-01 7.35710636e-02 2.28729248e-01 -2.31380209e-01
4.94365960e-01 3.93410057e-01 -1.21380448e+00 1.41459122e-01
1.16919529e+00 8.93042147e-01 2.21157238e-01 1.84514388e-01
-5.72708726e-01 2.79618829e-01 -5.47268033e-01 -1.14966810e-01
2.55294830e-01 -6.25363290e-01 -3.81492674e-01 6.68494225e-01
2.49425337e-01 -1.04798384e-01 -1.56851232e-01 1.43598306e+00
2.46670082e-01 1.80220455e-01 7.33657062e-01 -1.29108286e+00
-9.04956877e-01 1.97329000e-01 -9.63715672e-01 -3.33432257e-01
3.72222997e-02 -3.64506431e-02 5.65850139e-01 -9.98438835e-01
4.34219152e-01 1.49622774e+00 3.24740767e-01 5.43525040e-01
-1.16554999e+00 -9.64448690e-01 4.49981242e-01 3.21105748e-01
-1.59113550e+00 -1.14113808e-01 1.15117359e+00 -3.74284238e-01
-2.77202129e-01 9.74187031e-02 9.91972625e-01 4.42825258e-01
3.18599213e-03 1.02697062e+00 1.24786448e+00 -6.81920648e-01
3.45524222e-01 3.73262763e-01 -1.85240269e-01 9.92925644e-01
3.21715951e-01 2.85018589e-02 -1.24063611e-01 1.10182121e-01
9.87983227e-01 4.99499142e-01 -4.32502121e-01 -5.87424219e-01
-1.23505080e+00 6.60749853e-01 8.03471565e-01 2.58062154e-01
-1.02632977e-01 -2.61873845e-02 1.78129509e-01 -3.68985673e-03
2.37206414e-01 1.79659516e-01 -1.71035767e-01 5.00271440e-01
-7.88306117e-01 1.24516934e-02 3.90086204e-01 1.15307164e+00
1.35079706e+00 1.38136417e-01 -1.92513168e-01 1.10833502e+00
4.97187734e-01 5.77611089e-01 6.03029728e-01 -6.98241591e-01
3.64786923e-01 9.67488229e-01 1.97318822e-01 -1.14791119e+00
-1.21818811e-01 -1.62887461e-02 -1.02859712e+00 4.11008090e-01
2.63975471e-01 -3.08612823e-01 -1.01819873e+00 1.44619834e+00
4.82140988e-01 4.68616635e-01 4.98832874e-02 1.15518248e+00
4.57452327e-01 1.00576365e+00 -4.43415940e-02 -4.53587294e-01
1.10180306e+00 -1.03569853e+00 -7.53280997e-01 -1.21189430e-01
9.47881043e-02 -9.27193105e-01 1.24183035e+00 5.43225825e-01
-7.17804790e-01 -8.65037024e-01 -1.36499560e+00 3.37616742e-01
-4.31804180e-01 3.57517153e-01 5.91388285e-01 5.32058597e-01
-7.11372018e-01 3.71318728e-01 -4.76343364e-01 -2.14926213e-01
1.98338151e-01 -5.52864224e-02 -6.80227578e-02 -2.33475283e-01
-1.08932817e+00 5.67614496e-01 8.22570682e-01 4.32872683e-01
-5.68132937e-01 -3.83687228e-01 -5.22001803e-01 -2.56751746e-01
4.38139498e-01 -3.88178736e-01 7.72946537e-01 -1.51974356e+00
-1.49143314e+00 6.60737932e-01 6.92550316e-02 3.09056491e-01
5.05137026e-01 -6.41463473e-02 -7.25850523e-01 1.10447764e-01
-4.40927185e-02 5.88900685e-01 1.19457662e+00 -1.88474596e+00
-1.04525483e+00 -4.67691682e-02 -2.20718712e-01 2.48210222e-01
-4.76965845e-01 -2.49971434e-01 -1.17160594e+00 -8.32134545e-01
4.57173526e-01 -8.80404353e-01 -2.23692447e-01 1.88434869e-01
-2.57075489e-01 7.91274831e-02 9.68561471e-01 -5.31297863e-01
1.17510486e+00 -2.30089903e+00 2.22981885e-01 6.21010065e-01
-8.32599998e-02 1.76062495e-01 -2.39228666e-01 1.48440138e-01
2.58557429e-03 -1.41493557e-02 -4.61472154e-01 1.39577478e-01
-2.22905055e-01 1.45002112e-01 2.64918786e-02 3.59839827e-01
1.09281726e-01 5.27077496e-01 -9.31497872e-01 -9.41117048e-01
3.70254755e-01 3.59245270e-01 -2.59367555e-01 4.32210088e-01
-1.95858136e-01 3.45996857e-01 -7.26574123e-01 7.49057293e-01
9.57123518e-01 1.45782441e-01 2.55157184e-02 -6.55002356e-01
-1.19019076e-01 -8.72039080e-01 -1.70229745e+00 1.59485567e+00
-3.39461446e-01 8.61233994e-02 7.86714703e-02 -9.63426650e-01
1.24157965e+00 -1.56628415e-01 4.62406784e-01 -5.65785766e-01
2.63525397e-01 1.82055607e-01 -1.73605889e-01 -6.27830625e-01
2.22706497e-01 -1.85744390e-01 3.57254408e-02 3.04617614e-01
-2.61858523e-01 -2.85482973e-01 2.97924817e-01 -2.20862627e-02
1.01755075e-01 2.94822514e-01 -3.25911976e-02 -2.45436072e-01
1.03592050e+00 3.95515300e-02 9.08576250e-01 1.92165032e-01
2.07539394e-01 5.15782416e-01 9.60101411e-02 -4.60216016e-01
-9.62762296e-01 -9.77893770e-01 1.18969167e-02 7.97807753e-01
8.48324060e-01 6.98251128e-02 -9.80476677e-01 -7.34363258e-01
-1.01750411e-01 4.85886812e-01 -6.39123976e-01 -5.00544727e-01
-6.37190104e-01 -8.17288160e-01 3.99280619e-03 4.29215163e-01
9.67352986e-01 -1.05122101e+00 -3.35264057e-02 9.95609239e-02
-7.29117170e-02 -6.50247514e-01 -8.05082917e-01 -2.85568595e-01
-8.01361799e-01 -1.28813088e+00 -8.99852574e-01 -1.18337858e+00
1.20735276e+00 5.73964894e-01 6.31082356e-01 4.27942157e-01
-2.65597939e-01 2.56587207e-01 -4.36243623e-01 -2.86473811e-01
-3.35553110e-01 -3.55770886e-01 -1.49912760e-01 6.75610006e-01
2.08076134e-01 -2.04534903e-01 -8.50667059e-01 4.13031876e-01
-1.29583263e+00 3.69228005e-01 1.02906299e+00 9.67880309e-01
9.78721321e-01 2.76569277e-01 2.76634544e-01 -9.97921824e-01
6.10130072e-01 -9.64692384e-02 -8.00685644e-01 7.54305303e-01
-7.08552480e-01 1.28624529e-01 7.89414048e-01 -7.27866054e-01
-1.24264598e+00 2.41501138e-01 3.10707867e-01 -5.34526467e-01
-3.76084819e-02 3.33715528e-01 -4.98639286e-01 -4.01938915e-01
1.78544611e-01 4.65543002e-01 8.05671960e-02 -3.99991035e-01
7.79551446e-01 5.95610142e-01 5.89121878e-01 -7.48743117e-01
1.15744460e+00 4.32890117e-01 -6.09148368e-02 -5.32114565e-01
-6.62146330e-01 -3.15162599e-01 -7.51417935e-01 -5.27937710e-01
9.76579845e-01 -6.55017257e-01 -5.95908105e-01 6.91740990e-01
-8.25244248e-01 -1.21856019e-01 8.68853107e-02 4.33368951e-01
-3.62273097e-01 4.56247479e-01 -3.87667120e-01 -6.41340673e-01
-2.21415639e-01 -1.00636137e+00 8.94355416e-01 8.88450265e-01
5.65498173e-01 -9.60195780e-01 -1.78271756e-01 -1.99651957e-01
1.48760736e-01 2.55225152e-01 1.23739159e+00 -4.36545089e-02
-6.76592290e-01 -2.49183863e-01 -4.42662805e-01 5.19353628e-01
3.60898793e-01 4.24852848e-01 -4.68528569e-01 -4.42441761e-01
-9.70214605e-02 -5.08771054e-02 7.00696528e-01 9.36434567e-02
1.32089102e+00 -2.63237327e-01 -2.67942429e-01 8.29813063e-01
1.82509220e+00 5.85583329e-01 6.39177084e-01 5.69309711e-01
9.41174626e-01 4.75154489e-01 9.10161138e-01 3.46050829e-01
1.34021714e-02 4.68393058e-01 1.89491346e-01 -5.91885805e-01
-9.85767469e-02 -4.22444433e-01 -3.57276872e-02 8.65649939e-01
2.42360998e-02 3.03734273e-01 -4.60951179e-01 6.06450997e-02
-1.88566971e+00 -7.46532321e-01 1.77735060e-01 2.31347322e+00
8.09111595e-01 1.09973233e-02 1.15431018e-01 2.85680238e-02
1.14863861e+00 3.09797022e-02 -6.82742000e-01 1.33860976e-01
3.02624237e-02 -1.30189195e-01 3.52113008e-01 2.46186689e-01
-9.87628400e-01 9.45028365e-01 5.73395729e+00 1.02245855e+00
-1.40941775e+00 -3.67442757e-01 6.89834058e-01 5.00474095e-01
-2.43061215e-01 1.72733024e-01 -4.88531202e-01 7.14619637e-01
-1.38397977e-01 -1.51710212e-01 7.06630290e-01 7.59290159e-01
2.98054367e-01 5.63150831e-03 -7.55294800e-01 1.18342352e+00
2.49558240e-01 -9.99983132e-01 4.18408215e-01 -4.49169159e-01
9.70000923e-01 -8.06681097e-01 1.60494164e-01 1.32118329e-01
1.00581594e-01 -6.08763933e-01 7.55610168e-01 9.48217571e-01
7.24057794e-01 -8.53505552e-01 4.69605833e-01 1.28235668e-01
-1.48857319e+00 -1.55209288e-01 -6.37923479e-01 5.19933581e-01
-1.43994153e-01 3.43444407e-01 -2.02664912e-01 6.89684749e-01
6.26874328e-01 7.72864580e-01 -9.12870765e-01 1.32402003e+00
-4.43094820e-01 2.76665866e-01 4.01390009e-02 -1.63095877e-01
2.01932088e-01 -1.00943863e+00 -7.64012039e-02 1.11790240e+00
3.30165088e-01 2.54781246e-01 4.74337876e-01 9.12667453e-01
3.20514962e-02 5.12402117e-01 6.49137795e-02 2.22044036e-01
7.18179047e-01 1.58445394e+00 -8.40946853e-01 -5.06823242e-01
-3.64872396e-01 1.12184370e+00 1.96138903e-01 7.68001676e-01
-7.54855514e-01 -7.28995264e-01 5.92241026e-02 -2.10724235e-01
1.25252411e-01 -1.34967133e-01 -1.02978230e-01 -1.06845212e+00
-1.10182263e-01 -8.36533904e-01 4.00119960e-01 -9.42984164e-01
-1.35023534e+00 3.16358566e-01 3.46309207e-02 -1.77047932e+00
2.61747032e-01 -6.59449816e-01 -1.08600295e+00 7.51917243e-01
-1.56224608e+00 -1.21906078e+00 -7.97578335e-01 7.67331421e-01
3.08574170e-01 -2.39909500e-01 4.53154922e-01 2.08277166e-01
-7.80415952e-01 4.77687836e-01 5.09605885e-01 2.80573457e-01
8.97831798e-01 -1.15966856e+00 -2.01779172e-01 8.67823482e-01
7.37458467e-02 5.12439072e-01 4.09938067e-01 -3.71590465e-01
-1.52010417e+00 -1.20041549e+00 -3.72964256e-02 3.78815085e-01
3.98085237e-01 -8.05234686e-02 -1.00752223e+00 1.72586933e-01
-1.49742868e-02 -7.40133673e-02 3.55143279e-01 -2.12173551e-01
-2.87163466e-01 -7.58767903e-01 -9.02906954e-01 8.21156383e-01
5.89500904e-01 -1.12584613e-01 -5.75407505e-01 2.99248725e-01
5.17230928e-01 -2.98083633e-01 -6.28207803e-01 2.61859179e-01
4.35647011e-01 -7.99660385e-01 9.25844371e-01 -2.47890696e-01
1.84601992e-01 -8.82805705e-01 2.41499126e-01 -1.42670441e+00
-5.69633424e-01 -1.18562587e-01 4.55592126e-01 1.51206398e+00
1.35835633e-01 -5.11961579e-01 6.70394599e-01 4.06208694e-01
-9.14990976e-02 -7.05610573e-01 -3.82815093e-01 -6.10235929e-01
-1.49281416e-02 2.53386617e-01 7.63330936e-01 9.27569151e-01
-3.12245578e-01 7.09819868e-02 -2.87848681e-01 2.16344193e-01
7.86644161e-01 4.59367633e-01 8.61075044e-01 -9.71400023e-01
-8.95555839e-02 -7.08603740e-01 -1.24977618e-01 -8.05690527e-01
8.45866054e-02 -5.24695873e-01 2.32520729e-01 -1.57219911e+00
2.26609409e-01 -7.17342257e-01 -4.97528672e-01 2.68050134e-01
-5.84054112e-01 2.02768613e-02 1.93455145e-01 3.95189643e-01
-6.01874232e-01 6.75191581e-01 1.61018479e+00 -5.28428912e-01
-3.12891722e-01 -4.56126854e-02 -7.90158391e-01 7.48193681e-01
7.23564267e-01 -9.27619785e-02 -4.40656155e-01 -3.24307531e-01
-2.17351601e-01 -5.99231794e-02 5.87043762e-02 -1.07609236e+00
2.51235634e-01 -6.91906750e-01 7.94757962e-01 -4.57134038e-01
4.54316474e-02 -1.13853788e+00 1.44593462e-01 5.35255373e-01
-2.52595782e-01 -9.41274241e-02 -2.43442878e-02 6.02069736e-01
-2.83856779e-01 -3.24809432e-01 1.11669278e+00 -2.60673940e-01
-1.12004697e+00 4.87509906e-01 1.01463594e-01 4.34377454e-02
1.14655650e+00 -3.61638308e-01 -1.19279467e-01 -9.67367366e-02
-2.93563455e-01 3.57661992e-01 6.00010037e-01 4.16208535e-01
8.87716115e-01 -1.78358519e+00 -6.29656672e-01 3.74616146e-01
3.48651141e-01 -1.70706194e-02 3.47332358e-01 3.90486538e-01
-7.96196342e-01 -3.18279117e-01 -3.48013341e-01 -4.55017626e-01
-9.62803721e-01 9.54654098e-01 4.69442368e-01 3.62949371e-01
-3.53331506e-01 4.78615373e-01 2.27103055e-01 -5.52256294e-02
2.20215723e-01 -1.00393817e-01 -3.52397054e-01 -6.69637173e-02
5.36889732e-01 3.10384691e-01 -2.70744294e-01 -4.86000627e-01
-5.48724756e-02 1.27073264e+00 -1.31787017e-01 -1.48258016e-01
1.04040158e+00 -3.20600063e-01 -3.40593338e-01 3.98934215e-01
1.34733081e+00 -1.15605202e-02 -1.32412064e+00 -5.13707042e-01
-1.72350302e-01 -8.11233521e-01 -1.14564084e-01 -6.35329783e-01
-1.47970998e+00 7.32424498e-01 9.57183063e-01 3.60699967e-02
1.40118408e+00 -3.44586015e-01 7.57260442e-01 1.78537190e-01
1.68804064e-01 -1.50273943e+00 4.38964248e-01 -7.79791772e-02
6.83143079e-01 -1.10728061e+00 1.74361795e-01 -4.54929560e-01
-7.38983989e-01 1.36001265e+00 1.03710175e+00 -3.95093113e-01
5.16936064e-01 -2.11129323e-01 4.05944347e-01 6.77053109e-02
-1.89338401e-02 -6.91279545e-02 5.20406425e-01 5.31842232e-01
1.57919183e-01 6.52728826e-02 -4.78884757e-01 4.38132226e-01
2.13387474e-01 -2.70016134e-01 1.73529357e-01 6.96170509e-01
-8.19188356e-01 -1.11561322e+00 -7.19967365e-01 1.82830676e-01
2.13236585e-01 1.81593388e-01 -1.32698327e-01 7.16351330e-01
3.77967983e-01 8.33550334e-01 -1.55503720e-01 -5.99337280e-01
5.24943948e-01 -1.32694572e-01 3.60990047e-01 -3.48191261e-01
-3.08670793e-02 3.75295579e-01 -6.57090068e-01 -2.74712414e-01
-4.92061883e-01 -3.65563244e-01 -1.46509385e+00 -1.35404229e-01
-3.92984778e-01 3.11877191e-01 5.08427858e-01 7.79115319e-01
-2.03119799e-01 5.04866302e-01 1.21994972e+00 -8.28556180e-01
-3.09782475e-01 -5.25692821e-01 -8.23060274e-01 8.48635435e-01
-2.99346447e-02 -5.44318676e-01 -2.32059881e-01 3.65080595e-01] | [11.074955940246582, -1.2417163848876953] |
a8662e4c-e693-4c6d-9297-9cb60f0e2300 | skin-cancer-detection-and-tracking-using-data | 1612.01074 | null | http://arxiv.org/abs/1612.01074v1 | http://arxiv.org/pdf/1612.01074v1.pdf | Skin Cancer Detection and Tracking using Data Synthesis and Deep Learning | Dense object detection and temporal tracking are needed across applications
domains ranging from people-tracking to analysis of satellite imagery over
time. The detection and tracking of malignant skin cancers and benign moles
poses a particularly challenging problem due to the general uniformity of large
skin patches, the fact that skin lesions vary little in their appearance, and
the relatively small amount of data available. Here we introduce a novel data
synthesis technique that merges images of individual skin lesions with
full-body images and heavily augments them to generate significant amounts of
data. We build a convolutional neural network (CNN) based system, trained on
this synthetic data, and demonstrate superior performance to traditional
detection and tracking techniques. Additionally, we compare our system to
humans trained with simple criteria. Our system is intended for potential
clinical use to augment the capabilities of healthcare providers. While
domain-specific, we believe the methods invoked in this work will be useful in
applying CNNs across domains that suffer from limited data availability. | ['Sebastian Thrun', 'Justin Ko', 'Brett Kuprel', 'Andre Esteva', 'Rob Novoa', 'Yunzhu Li'] | 2016-12-04 | null | null | null | null | ['dense-object-detection'] | ['computer-vision'] | [ 5.65169275e-01 1.13504894e-01 -3.17269325e-01 -1.82207316e-01
-5.30655682e-01 -5.98901153e-01 5.88719070e-01 7.31773898e-02
-4.21414226e-01 6.39826000e-01 -1.16423629e-01 -4.39696997e-01
1.36441812e-01 -7.90186107e-01 -5.62895358e-01 -5.90488553e-01
-4.16110605e-01 2.03864262e-01 4.41444308e-01 -1.54898033e-01
-5.55870354e-01 8.33681881e-01 -1.39256942e+00 9.80647951e-02
6.85399830e-01 7.99524665e-01 -1.07974960e-02 9.21778083e-01
1.37698472e-01 5.05222797e-01 -3.44050676e-01 -2.91943997e-01
4.11600024e-01 -3.42195749e-01 -4.67967778e-01 3.17288369e-01
1.02715421e+00 -6.61680281e-01 -3.34453613e-01 1.02748084e+00
4.52793896e-01 -1.80863157e-01 3.72273982e-01 -1.09400856e+00
-3.38542968e-01 2.98665445e-02 -7.22632289e-01 2.86149442e-01
2.95507282e-01 2.64906883e-01 3.12878132e-01 -3.69660288e-01
9.54526842e-01 9.34637725e-01 1.26819861e+00 1.18759966e+00
-1.07710445e+00 -6.28668368e-01 -8.83955881e-02 -3.49219441e-01
-1.32492459e+00 -4.54973280e-01 1.59674361e-01 -5.00852287e-01
5.81996083e-01 5.66663027e-01 8.84236693e-01 1.11470211e+00
9.41331312e-02 7.44651854e-01 7.86535084e-01 -3.44284326e-01
9.43575948e-02 2.14909744e-02 -3.50971043e-01 9.57988024e-01
5.16646743e-01 2.33915702e-01 -1.30910486e-01 -1.72405124e-01
1.09059405e+00 2.82278091e-01 -8.01213235e-02 -3.20355356e-01
-1.18421519e+00 4.82135355e-01 5.99432766e-01 2.38453478e-01
-3.94170374e-01 3.40319425e-01 1.85705438e-01 -1.01184212e-01
4.36846405e-01 1.36858359e-01 -2.16650218e-01 3.00999910e-01
-1.30219531e+00 3.59133780e-01 5.64838946e-01 9.02628899e-01
2.09252074e-01 1.21322237e-01 -2.13726833e-01 4.99168754e-01
1.54848874e-01 5.37274837e-01 4.05750811e-01 -8.47987592e-01
-1.46006495e-01 5.76639771e-01 1.81214899e-01 -7.49586284e-01
-6.98264956e-01 -3.36480975e-01 -7.44493961e-01 4.41909462e-01
6.18963242e-01 -3.46229345e-01 -1.46132350e+00 1.71674502e+00
7.17989087e-01 1.16387829e-01 -4.20899875e-02 7.36646950e-01
9.07032728e-01 7.65724108e-02 4.91609395e-01 2.73687746e-02
1.48931050e+00 -6.68707788e-01 -6.23459816e-01 -3.50086629e-01
5.87232530e-01 -4.34843719e-01 4.33111608e-01 8.23186338e-02
-1.05491531e+00 -1.22175343e-01 -9.19532180e-01 6.48581088e-02
-4.22176450e-01 1.85974434e-01 9.10986364e-01 8.51436794e-01
-1.10868096e+00 4.84419107e-01 -1.28509998e+00 -1.13099980e+00
8.58330131e-01 4.26258475e-01 -4.74954307e-01 -1.84657082e-01
-7.58441806e-01 8.72056961e-01 1.13650680e-01 2.29558628e-02
-9.27722037e-01 -7.89675772e-01 -9.45346951e-01 -3.94865036e-01
1.75964713e-01 -7.57059276e-01 1.45119143e+00 -1.05531979e+00
-7.10925102e-01 9.97658551e-01 5.01518361e-02 -6.23226643e-01
9.83718455e-01 1.99129567e-01 -3.57591450e-01 1.52076110e-01
5.55738322e-02 1.00025356e+00 6.54268026e-01 -8.16125095e-01
-7.53273129e-01 -3.31794947e-01 -1.28705680e-01 1.06750965e-01
-3.69321704e-01 1.00073926e-01 -6.86652541e-01 -6.01413488e-01
-2.15484783e-01 -1.02800560e+00 -6.79381073e-01 1.00731826e+00
-4.07482177e-01 4.05484885e-01 9.91763711e-01 -7.96526313e-01
7.18073487e-01 -2.08475542e+00 -3.65300506e-01 -4.13510576e-03
2.95881718e-01 4.82706904e-01 -1.09778076e-01 1.65714160e-01
8.65443200e-02 1.06728390e-01 -2.94109404e-01 -1.93670914e-01
-3.91041458e-01 1.11867383e-01 2.54130512e-01 6.98923707e-01
3.81226778e-01 9.65926409e-01 -1.06112134e+00 -7.28083789e-01
2.17546925e-01 6.05333269e-01 -8.59941691e-02 -9.70290750e-02
-4.20874953e-01 4.64453220e-01 -3.70516837e-01 1.22489858e+00
7.38586843e-01 -4.66451854e-01 2.27591902e-01 -1.62191465e-01
-5.30370362e-02 -2.81867057e-01 -9.28686261e-01 1.55783677e+00
-1.37743890e-01 6.65567040e-01 4.53766584e-01 -2.93400258e-01
3.23062271e-01 3.54645789e-01 7.40581930e-01 -5.56828082e-01
1.15460195e-01 5.04361987e-02 2.39021420e-01 -5.68945527e-01
5.41977882e-01 -2.30579257e-01 2.66404182e-01 2.64934748e-01
-1.92851529e-01 3.92993260e-03 4.31731582e-01 1.13709562e-01
1.42581916e+00 1.01357140e-01 5.65003872e-01 -6.49550706e-02
9.12661664e-03 5.67114174e-01 4.93538260e-01 6.68953538e-01
-5.08867562e-01 8.12848985e-01 5.71243912e-02 -6.97974145e-01
-1.23577833e+00 -1.02059484e+00 -4.72949445e-01 8.21442604e-01
1.11171221e-02 9.51481089e-02 -7.05666482e-01 -6.69554293e-01
3.09577137e-01 1.66601449e-01 -1.05597591e+00 1.89383030e-01
-3.52886379e-01 -9.39043760e-01 9.12750542e-01 7.66026974e-01
3.60419482e-01 -9.88475680e-01 -8.37187946e-01 2.85207719e-01
2.84780890e-01 -9.84694719e-01 -3.80757570e-01 -7.93531612e-02
-7.55259693e-01 -1.24785900e+00 -1.17813075e+00 -7.10374415e-01
1.02936900e+00 2.35437185e-01 8.42990160e-01 4.05930400e-01
-1.23345208e+00 3.44418138e-01 -2.04294492e-02 -7.15905309e-01
-5.43907940e-01 -1.29410103e-01 8.40167925e-02 -1.81208417e-01
2.30824992e-01 -7.87293687e-02 -7.60386884e-01 9.94951800e-02
-1.17432439e+00 1.43360510e-01 6.69401407e-01 5.98975003e-01
5.67795157e-01 -1.17732719e-01 1.73008278e-01 -1.01540399e+00
4.01845038e-01 -5.60535312e-01 -5.99449813e-01 2.79112756e-01
-1.53127983e-01 -4.07347143e-01 4.10184897e-02 -6.86135173e-01
-1.00388873e+00 5.63955963e-01 4.34030220e-02 -4.63325262e-01
-1.79908454e-01 1.28789395e-01 3.91752213e-01 -4.15693045e-01
1.05443847e+00 8.10939372e-02 3.27515185e-01 -2.23759681e-01
2.00485066e-01 3.83772314e-01 7.88401067e-01 1.11803494e-03
7.99340308e-01 1.03143203e+00 1.00414179e-01 -8.58410478e-01
-6.55531585e-01 -4.51402366e-01 -3.66997331e-01 -2.25805908e-01
9.06759918e-01 -9.67714131e-01 -4.88184601e-01 5.09682953e-01
-8.45845699e-01 -5.54706633e-01 -3.44014376e-01 2.62729019e-01
-1.20654270e-01 2.10119531e-01 -6.53805733e-01 -8.93442750e-01
-5.42297542e-01 -7.52745807e-01 1.30874622e+00 6.17904007e-01
-2.91901439e-01 -1.11728358e+00 6.54485598e-02 -3.83777767e-02
5.80413997e-01 7.70420551e-01 2.12684348e-01 -3.81648958e-01
-6.29744947e-01 -5.54387927e-01 -4.58034366e-01 -1.38021469e-01
5.74887216e-01 3.35661173e-01 -1.08497393e+00 -5.50288081e-01
-6.63061559e-01 -2.17186481e-01 8.32851291e-01 7.48556674e-01
1.12381661e+00 -1.17387429e-01 -1.05177534e+00 6.14341140e-01
1.32024336e+00 1.44874170e-01 3.97460312e-01 2.88527995e-01
5.40512800e-01 6.02971673e-01 5.05578220e-01 2.24859253e-01
1.96118608e-01 4.55512494e-01 5.67197204e-01 -8.10432851e-01
-4.25220698e-01 2.88555343e-02 9.80491564e-03 -6.81154653e-02
-4.36691605e-02 -1.31596714e-01 -1.11665332e+00 1.04698801e+00
-1.67754626e+00 -9.49174941e-01 -8.55785459e-02 2.13493180e+00
6.97929919e-01 -1.34802580e-01 2.82376885e-01 -4.17999268e-01
7.58397698e-01 -1.31361321e-01 -7.42182672e-01 7.51510859e-02
1.83519036e-01 1.90694183e-01 1.02078795e+00 1.71597660e-01
-1.47899795e+00 5.57312429e-01 7.40242577e+00 4.65504438e-01
-1.32843637e+00 -1.46086700e-02 5.48168421e-01 -1.76008791e-01
5.42691946e-02 -4.82375979e-01 -7.88796544e-01 2.82531202e-01
6.62650287e-01 -1.85646683e-01 3.72498930e-02 7.55059302e-01
2.71677803e-02 -1.50601476e-01 -1.16984284e+00 6.16791606e-01
-4.43105288e-02 -1.41245139e+00 -4.00532275e-01 1.95879430e-01
8.16258311e-01 2.23125383e-01 2.32978284e-01 -1.05243430e-01
6.12971663e-01 -1.24883294e+00 3.23421448e-01 5.77647328e-01
1.09052169e+00 -2.40961894e-01 5.68790913e-01 2.80894250e-01
-1.10981703e+00 1.16916165e-01 -1.00280009e-01 3.17151338e-01
-1.25953443e-02 3.46968532e-01 -1.45668185e+00 1.71361908e-01
7.04884827e-01 4.59203452e-01 -7.62594879e-01 1.52614880e+00
2.64416665e-01 2.16378629e-01 -6.11364603e-01 -1.19573057e-01
1.80639848e-02 3.48572105e-01 2.99472779e-01 1.46926630e+00
4.32398617e-01 -1.32753447e-01 1.79103374e-01 8.69947851e-01
-6.26884997e-02 -6.62548989e-02 -9.19261336e-01 -2.05589339e-01
3.68274599e-01 1.60898209e+00 -8.89017940e-01 -2.96339631e-01
-5.15325904e-01 8.50414753e-01 -1.07774645e-01 6.63175434e-02
-6.48140073e-01 -1.38992593e-01 7.17093527e-01 4.05173182e-01
2.71098495e-01 1.01797476e-01 -1.40578687e-01 -8.80173683e-01
-4.34092768e-02 -7.03885734e-01 5.02131701e-01 -6.30932331e-01
-1.22315526e+00 6.20498478e-01 -1.16772979e-01 -1.47698426e+00
-3.21997136e-01 -5.55111408e-01 -7.06787169e-01 6.53413653e-01
-1.48013449e+00 -1.67829692e+00 -7.88844526e-01 3.45276266e-01
3.00224870e-01 1.05280839e-01 8.84529650e-01 4.15173978e-01
-5.41787505e-01 5.90410948e-01 1.30662546e-02 3.42151791e-01
4.91657436e-01 -1.22682214e+00 7.40603983e-01 8.18826020e-01
-3.45868319e-01 6.41551197e-01 5.73936880e-01 -9.61590052e-01
-1.28367090e+00 -1.53754771e+00 3.71779203e-01 -5.41961968e-01
4.85115975e-01 -2.55200565e-01 -6.94770157e-01 8.65160823e-01
6.76469579e-02 5.57503521e-01 6.12432361e-01 -1.94996193e-01
-1.34548182e-02 1.37946792e-02 -1.64614820e+00 7.31134653e-01
7.27563679e-01 -1.56056881e-01 -3.49938422e-02 6.58816576e-01
3.64073217e-01 -9.21096563e-01 -8.47865045e-01 5.11672735e-01
8.81699085e-01 -5.77320218e-01 1.00522029e+00 -5.78308642e-01
2.45213181e-01 -3.09292167e-01 9.45012793e-02 -9.75651026e-01
-1.13261722e-01 -3.58116359e-01 2.26833317e-02 7.74803996e-01
4.42276090e-01 -2.88465828e-01 1.39559066e+00 8.33493352e-01
3.11899364e-01 -5.21723151e-01 -8.06175292e-01 -7.23060966e-01
-1.05289500e-02 -1.55801952e-01 3.88293296e-01 1.02603281e+00
-3.42698753e-01 -4.53976244e-01 -2.84787595e-01 3.14108700e-01
8.99453282e-01 -1.09985016e-01 7.57912934e-01 -1.07389224e+00
-1.53694987e-01 -8.01579580e-02 -5.81247985e-01 -3.90901953e-01
-5.72448790e-01 -6.24527216e-01 1.11010812e-01 -1.57157183e+00
3.47105831e-01 -5.93160927e-01 -5.13490997e-02 8.39680910e-01
-1.13225944e-01 7.82802105e-01 4.71885763e-02 -6.20363615e-02
-4.61060882e-01 -1.47217438e-01 1.41499162e+00 -3.25585246e-01
-1.15361162e-01 6.73715100e-02 -5.47166348e-01 6.76170468e-01
5.39321959e-01 -4.25296009e-01 -1.02347933e-01 -2.41315097e-01
-7.91110918e-02 2.75599621e-02 6.15830958e-01 -1.20501792e+00
4.23636407e-01 -2.63414204e-01 9.00905371e-01 -4.46178287e-01
4.58298296e-01 -7.61995316e-01 5.07445753e-01 9.11657929e-01
-2.35924334e-03 -3.45074564e-01 6.39346182e-01 4.88900244e-01
1.24841459e-01 -8.97185951e-02 9.76459622e-01 -5.73222697e-01
-7.36929953e-01 5.54080188e-01 -3.11561465e-01 -3.25365841e-01
1.32981086e+00 -4.39661771e-01 -3.05855483e-01 -3.70108753e-01
-9.03598785e-01 2.21893385e-01 8.19051147e-01 2.61597276e-01
3.17066878e-01 -1.40033531e+00 -8.01407099e-01 1.43718615e-01
2.93755561e-01 1.28889605e-01 3.14403832e-01 8.76154065e-01
-9.78689253e-01 2.23873794e-01 -4.34218168e-01 -7.99243212e-01
-1.67225802e+00 3.78364116e-01 6.61787033e-01 -6.55454770e-02
-7.26604223e-01 7.57703781e-01 2.09886074e-01 -4.72206235e-01
2.78409302e-01 -3.10684323e-01 4.69549298e-02 -9.34369862e-02
8.54922116e-01 1.00751668e-01 1.82829648e-02 -3.86976421e-01
-4.37534064e-01 1.90002248e-01 -3.16497713e-01 9.88417715e-02
1.34019256e+00 2.42195874e-01 2.55566567e-01 -2.27368295e-01
7.27972806e-01 -6.51743859e-02 -1.44296074e+00 -1.91132993e-01
-2.35004589e-01 -3.96246642e-01 -3.80945355e-02 -8.42452943e-01
-1.11151135e+00 4.77250457e-01 1.05811143e+00 2.71281809e-01
1.15907514e+00 2.88279466e-02 5.73978305e-01 1.76471636e-01
1.97263807e-01 -7.10871220e-01 -2.04934821e-01 -1.77666813e-01
4.90237623e-01 -1.39160252e+00 4.34538424e-01 -4.66628939e-01
-3.65326643e-01 1.02735865e+00 7.57983804e-01 9.16224197e-02
2.59396970e-01 7.25935996e-01 3.91031146e-01 -3.23160440e-01
-5.64269960e-01 -4.74794179e-01 2.90342093e-01 9.62309599e-01
3.72614890e-01 7.03831539e-02 -1.83793858e-01 -9.44017917e-02
1.56468675e-01 3.93627465e-01 5.80458105e-01 1.32818127e+00
-1.93624929e-01 -1.04452622e+00 -5.27499318e-01 7.44056344e-01
-7.33623445e-01 1.02677204e-01 -5.39721131e-01 1.17408454e+00
3.02058786e-01 3.95446420e-01 2.43725076e-01 9.37616006e-02
1.19277962e-01 -2.79662997e-01 5.99054635e-01 -6.86841607e-01
-5.19823670e-01 1.50787681e-02 2.29340985e-01 -4.61599946e-01
-5.37539005e-01 -9.30923641e-01 -1.03894103e+00 -9.94210616e-02
-2.57456124e-01 -6.24912739e-01 8.99717569e-01 6.34732485e-01
7.92930126e-02 6.54766202e-01 1.25375301e-01 -7.39094317e-01
-1.62929073e-01 -8.88597846e-01 -6.33097470e-01 3.78307134e-01
7.15073287e-01 -4.12654787e-01 1.81320146e-01 3.64274025e-01] | [15.363951683044434, -2.7570641040802] |
4388f164-6344-4abd-a3ad-7b2bb4fb601c | countering-language-drift-via-grounding | null | null | https://openreview.net/forum?id=BkMn9jAcYQ | https://openreview.net/pdf?id=BkMn9jAcYQ | Countering Language Drift via Grounding | While reinforcement learning (RL) shows a lot of promise for natural language processing—e.g. when fine-tuning natural language systems for optimizing a certain objective—there has been little investigation into potential language drift: when an external reward is used to train a system, the agents’ communication protocol may easily and radically diverge from natural language. By re-casting translation as a communication game, we show that language drift indeed happens when pre-trained agents are fine-tuned with policy gradient methods. We contend that simply adding a "naturalness" constraint to the reward, e.g. by using language model log likelihood, does not fully address the issue, and argue that (perceptual) grounding is required. That is, while language model constraints impose syntactic conformity, they do not lead to semantic correspondence. Our experiments show that grounded models give the best communication performance, while retaining English syntax along with the ability to convey the intended semantics. | ['Douwe Kiela', 'Kyunghyun Cho', 'Jason Lee'] | 2018-09-27 | null | null | null | null | ['policy-gradient-methods'] | ['methodology'] | [ 1.09305054e-01 4.09977704e-01 -3.22663724e-01 -3.97783488e-01
-7.19154894e-01 -8.43794525e-01 1.01609051e+00 2.07642049e-01
-8.92401934e-01 9.63632166e-01 4.72214997e-01 -7.45347321e-01
2.59122252e-01 -6.28926694e-01 -7.95629263e-01 -4.97561753e-01
-1.15093999e-01 6.49883807e-01 5.30974604e-02 -5.77625215e-01
3.95510256e-01 3.34209085e-01 -9.54139769e-01 -3.40678282e-02
8.36794615e-01 2.26187930e-01 2.45119154e-01 7.33258426e-01
-2.18695179e-01 1.02200723e+00 -7.43095875e-01 -2.07452387e-01
2.04336390e-01 -7.87642777e-01 -1.24515820e+00 -1.25193326e-02
-7.46946409e-02 -4.55496401e-01 -1.63960047e-02 1.21725023e+00
2.63096988e-01 -7.45740384e-02 4.27407622e-01 -1.14761925e+00
-6.87618136e-01 1.09798276e+00 -3.69092524e-01 -2.40450595e-02
4.30227339e-01 4.51099187e-01 1.24457538e+00 -2.34471038e-01
7.12397993e-01 1.72529113e+00 3.14474642e-01 9.07097042e-01
-1.43502772e+00 -4.26203579e-01 5.28820276e-01 -4.24761742e-01
-7.27270722e-01 -4.54765648e-01 5.55422187e-01 -3.43322158e-01
1.06203842e+00 4.17674072e-02 5.70946217e-01 1.22956347e+00
5.06709039e-01 7.49077380e-01 1.33036852e+00 -4.39205825e-01
4.87651110e-01 2.40063503e-01 -3.41037780e-01 7.74941802e-01
2.49279842e-01 4.69622821e-01 -6.26160741e-01 -3.73939902e-01
5.54027259e-01 -5.64941347e-01 -5.44922575e-02 -3.90833497e-01
-1.29185009e+00 1.02583122e+00 3.81657720e-01 3.68292451e-01
-3.31105351e-01 6.41429365e-01 4.66415793e-01 8.03593457e-01
1.08685575e-01 1.09601867e+00 -6.26119852e-01 -4.14658844e-01
-5.65797150e-01 4.54132408e-01 8.48442972e-01 8.02232146e-01
7.38431454e-01 1.44986197e-01 1.04515754e-01 5.35077691e-01
4.48413014e-01 4.33535397e-01 6.77707493e-01 -1.28002250e+00
2.79989988e-01 2.75591344e-01 3.88720810e-01 -5.75603366e-01
-5.10276318e-01 1.06741376e-01 -1.56032145e-01 4.67617065e-01
6.66976273e-01 -6.34835184e-01 -5.96078873e-01 2.34295678e+00
-6.92194775e-02 -3.91444176e-01 5.48358738e-01 1.02812946e+00
5.96616454e-02 6.51185513e-01 4.13597971e-01 -4.10929918e-01
1.20959687e+00 -5.93545973e-01 -3.08033645e-01 -7.18185425e-01
8.04021537e-01 -6.71252131e-01 1.47413683e+00 1.71816319e-01
-1.31775033e+00 -4.82399203e-02 -1.08164549e+00 2.10349962e-01
8.48334283e-02 -7.48308480e-01 7.71801293e-01 4.73608226e-01
-1.10663843e+00 5.65230846e-01 -9.38622177e-01 -7.17373848e-01
-9.75202993e-02 2.89030284e-01 5.18223271e-02 3.67924958e-01
-1.29299188e+00 1.08985353e+00 2.14984536e-01 -2.05776080e-01
-1.07858086e+00 3.48198898e-02 -5.66608906e-01 -1.40606433e-01
4.33815628e-01 -7.04719603e-01 1.70908475e+00 -1.60045218e+00
-1.98683059e+00 7.83615589e-01 2.41586547e-02 -6.18952274e-01
5.63137472e-01 -6.78239390e-02 -1.56013072e-01 -6.98226541e-02
1.77373767e-01 9.06772792e-01 7.89118528e-01 -1.23288608e+00
-6.47192001e-01 -2.24771678e-01 4.45365548e-01 6.20906889e-01
-1.06420055e-01 3.60712528e-01 1.67937756e-01 -4.43471432e-01
-6.84492588e-02 -1.19452381e+00 -3.67138177e-01 -4.12231982e-01
-1.39628828e-01 -3.95345002e-01 2.79011011e-01 -1.38168171e-01
8.73440564e-01 -2.00779295e+00 2.42801700e-02 2.35201120e-01
8.90064240e-02 -1.00124769e-01 -3.48101914e-01 5.49609959e-01
3.50903273e-01 4.78717834e-01 -3.35623831e-01 2.53351592e-02
2.62656212e-01 4.09230590e-01 -4.46019351e-01 4.61885363e-01
3.32079381e-01 8.73775661e-01 -1.16445339e+00 -2.92776883e-01
-3.51193070e-01 3.80980484e-02 -9.26566899e-01 3.42027657e-02
-7.30761290e-01 4.44844633e-01 -6.76486313e-01 1.22469008e-01
3.41531299e-02 -1.74471781e-01 4.07723546e-01 7.48880982e-01
-2.84576565e-01 8.53555262e-01 -9.29510832e-01 1.72872686e+00
-4.01229948e-01 4.66272622e-01 1.76849052e-01 -7.47894466e-01
7.25628734e-01 1.63764939e-01 2.61482388e-01 -8.83137047e-01
2.06565652e-02 4.05085444e-01 5.99542201e-01 -5.29323995e-01
4.68037397e-01 -5.67257822e-01 -3.51789206e-01 9.40883815e-01
-2.53735781e-01 -4.33282405e-01 -3.34885367e-03 1.68749705e-01
1.03854001e+00 4.28089797e-01 -7.28933066e-02 -6.08689427e-01
1.52184904e-01 5.94584107e-01 6.66758120e-01 1.02314556e+00
-3.02172422e-01 5.91008104e-02 6.72399282e-01 -1.64104924e-01
-1.22579038e+00 -8.84852707e-01 2.41533771e-01 1.43291450e+00
1.28545210e-01 -2.82945514e-01 -9.93942678e-01 -6.22826874e-01
-6.06527068e-02 9.31529582e-01 -2.70258158e-01 -3.40224117e-01
-7.01105714e-01 -7.49667048e-01 7.10054994e-01 2.50200123e-01
3.22417498e-01 -1.33681262e+00 -1.12263763e+00 4.63736296e-01
5.29763214e-02 -7.02147007e-01 -5.54886520e-01 4.14964408e-01
-8.43716204e-01 -6.42877936e-01 -3.93507212e-01 -6.90523624e-01
5.44485986e-01 2.61869989e-02 1.13123953e+00 2.20624015e-01
2.42954418e-01 4.71187919e-01 -1.21744394e-01 -3.42321485e-01
-1.00785828e+00 1.76509351e-01 2.26474643e-01 -5.42091787e-01
4.99399960e-01 -2.97605574e-01 -4.18624789e-01 3.39495391e-02
-7.65125453e-01 -2.19119668e-01 4.87774342e-01 1.00135767e+00
-1.03003316e-01 -2.36334890e-01 7.02154398e-01 -8.64233255e-01
1.49042892e+00 -2.21146107e-01 -7.89459169e-01 9.95006338e-02
-9.90723908e-01 6.31587923e-01 6.40362561e-01 -3.98459077e-01
-9.74621952e-01 -1.93298444e-01 1.40106514e-01 2.58051336e-01
-2.29776323e-01 4.15498197e-01 2.36281008e-01 2.75191963e-01
1.03892064e+00 6.41851798e-02 2.56637096e-01 -5.44710122e-02
4.48019505e-01 4.97706652e-01 1.76327735e-01 -1.12686229e+00
6.58153594e-01 2.35350505e-01 -4.55532193e-01 -6.73220396e-01
-4.27768499e-01 4.59195971e-02 -9.59595144e-02 2.00274199e-01
7.91607261e-01 -8.01161528e-01 -8.20995212e-01 -3.53400875e-03
-1.09496093e+00 -8.76463413e-01 -3.57202351e-01 6.93769157e-01
-9.98984337e-01 4.37657945e-02 -7.28511274e-01 -8.05125773e-01
-4.40030396e-02 -1.32751334e+00 6.96870387e-01 3.71627547e-02
-7.02796519e-01 -8.16883504e-01 1.82746530e-01 -1.23381272e-01
6.02936268e-01 -2.11063281e-01 1.13639224e+00 -6.60939932e-01
-4.26751137e-01 4.50012416e-01 1.14934616e-01 -1.94819681e-02
1.18992105e-01 -3.65166254e-02 -8.15101862e-01 -4.01565462e-01
7.33619705e-02 -6.10567927e-01 5.06296694e-01 2.63558567e-01
2.42679849e-01 -6.19873405e-01 -1.37668913e-02 2.10949659e-01
1.15463006e+00 3.99060279e-01 1.29254356e-01 7.24452078e-01
2.97694921e-01 8.64294291e-01 2.24543437e-01 3.50559592e-01
5.93814552e-01 4.21339303e-01 8.13930929e-02 1.27532989e-01
2.44336575e-01 -5.85556507e-01 7.94492424e-01 4.36546862e-01
4.39515024e-01 -1.28488600e-01 -1.05719078e+00 2.87351370e-01
-1.89254177e+00 -9.14365768e-01 4.62835699e-01 2.14040756e+00
1.10013640e+00 5.70496798e-01 3.46506298e-01 -4.79521215e-01
5.09292066e-01 1.59356162e-01 -8.43852043e-01 -1.07925677e+00
-7.25473985e-02 -1.25741199e-01 6.37129426e-01 1.17959332e+00
-3.79941642e-01 1.54181111e+00 6.80850363e+00 6.00446574e-02
-1.37039959e+00 -5.18038198e-02 4.54575300e-01 1.16096273e-01
-7.52843320e-01 3.70270431e-01 -5.39559305e-01 2.29669198e-01
1.00981808e+00 -4.43652987e-01 9.03414547e-01 4.50422317e-01
4.90803272e-01 -1.85473204e-01 -1.28112364e+00 4.39014673e-01
-2.16914341e-01 -9.09721613e-01 -1.97195321e-01 7.11980611e-02
4.44233835e-01 3.61783594e-01 -6.06326163e-02 4.29782540e-01
1.18706799e+00 -1.10798120e+00 1.07523239e+00 6.53585643e-02
3.04853052e-01 -7.57327616e-01 2.69964159e-01 7.73270965e-01
-3.63183200e-01 -3.71172428e-02 -4.38797772e-01 -3.78589898e-01
6.28153682e-02 -1.00848444e-01 -1.02367592e+00 -1.83447570e-01
3.78657460e-01 3.15590575e-02 -3.84364337e-01 4.53959823e-01
-5.07024288e-01 7.32503176e-01 -4.18603957e-01 -5.10841608e-01
8.11671138e-01 -2.16576815e-01 6.60442472e-01 1.10900521e+00
-1.18645914e-01 -1.21891551e-01 3.09432596e-01 9.35629129e-01
5.07190004e-02 1.13823295e-01 -7.94368923e-01 -2.60061502e-01
4.34537351e-01 5.72816193e-01 -6.86715186e-01 -2.71227509e-01
-4.48650390e-01 1.04076791e+00 2.63752311e-01 5.40895462e-01
-4.58354205e-01 -1.13790676e-01 6.93097770e-01 -1.41345218e-01
-7.98352137e-02 -4.43259567e-01 -2.99528867e-01 -1.19081759e+00
-1.32089198e-01 -1.28290939e+00 7.45394528e-02 -6.59991682e-01
-9.47918117e-01 4.64465082e-01 -1.01867683e-01 -5.58142424e-01
-9.33798194e-01 -3.33429188e-01 -3.15565556e-01 7.80448437e-01
-1.45217896e+00 -5.93706608e-01 5.45259833e-01 5.15183330e-01
5.90117693e-01 -1.23495236e-01 8.40604722e-01 -3.22964370e-01
-2.23999888e-01 4.56168175e-01 -1.34758532e-01 -1.31096495e-02
7.20803320e-01 -1.46865368e+00 5.94312191e-01 6.86023057e-01
3.93665314e-01 9.17372882e-01 1.18669260e+00 -4.63578761e-01
-1.59090090e+00 -5.27871609e-01 9.98557389e-01 -4.48291510e-01
8.87647629e-01 -3.78846169e-01 -7.56513953e-01 7.88905144e-01
5.45675397e-01 -3.69304627e-01 2.55059391e-01 1.79185435e-01
-4.00442451e-01 1.98700428e-01 -1.06778038e+00 1.13443816e+00
8.63246083e-01 -6.43761337e-01 -7.77814388e-01 2.68460035e-01
1.02039492e+00 -2.02707708e-01 -3.88535082e-01 -8.10644180e-02
4.46744740e-01 -7.07535267e-01 3.91514480e-01 -9.85741675e-01
1.86406091e-01 -3.33698899e-01 -2.72208273e-01 -1.55375409e+00
-2.10317567e-01 -1.07596529e+00 4.06236589e-01 1.02555704e+00
8.74130547e-01 -7.79942632e-01 8.08825791e-01 8.31997693e-01
1.49026245e-01 -2.66859889e-01 -5.86593270e-01 -8.65376711e-01
6.69307292e-01 -4.44296718e-01 6.28908992e-01 1.00217140e+00
5.61947405e-01 8.12060237e-01 -1.95015401e-01 1.44667579e-02
3.73093635e-01 -1.08668789e-01 6.81515038e-01 -9.10169482e-01
-5.73732615e-01 -8.75795186e-01 1.23492032e-01 -1.14002645e+00
3.78235877e-01 -8.29209924e-01 3.48395765e-01 -1.28531694e+00
-1.46408901e-01 -6.33142054e-01 -1.85661986e-01 4.51232642e-01
2.21425876e-01 -4.21569169e-01 3.74957561e-01 3.37709099e-01
-4.49557662e-01 3.76709372e-01 1.18859971e+00 -5.84082492e-02
-4.66970861e-01 -1.48646176e-01 -1.20234323e+00 7.93186605e-01
9.88713920e-01 -5.16050577e-01 -5.46673894e-01 -7.65592456e-01
6.37971401e-01 1.25953972e-01 1.13618515e-01 -4.46702391e-01
2.79772997e-01 -6.65053070e-01 -7.42826536e-02 4.96229559e-01
-1.03012890e-01 -5.85192442e-01 -3.74142885e-01 8.37035477e-01
-1.00226033e+00 5.07270515e-01 1.80591568e-01 4.07293856e-01
9.78759304e-02 -3.06512803e-01 7.75369406e-01 -4.29515213e-01
-5.84282875e-01 -1.34863064e-01 -8.63269627e-01 4.20341164e-01
6.96755767e-01 5.42618670e-02 -2.19469666e-01 -5.77679515e-01
-4.42031950e-01 4.36034024e-01 8.70000720e-01 5.83571434e-01
2.38550812e-01 -9.74469185e-01 -7.87942290e-01 1.22282989e-01
1.04043469e-01 -2.14352041e-01 -6.34694993e-01 3.91660273e-01
-3.46829176e-01 3.65252227e-01 -1.10203801e-02 -4.97440070e-01
-6.43479466e-01 4.15029377e-01 4.99910891e-01 -1.69788394e-02
-6.55916929e-01 8.02769899e-01 1.83270410e-01 -4.98103023e-01
3.39166790e-01 -4.74986196e-01 1.98338851e-01 -2.20498785e-01
2.42506191e-01 -2.84969717e-01 -2.15679973e-01 -2.83904642e-01
-3.44135404e-01 2.68036664e-01 -2.60663033e-01 -8.85116100e-01
1.11563540e+00 -2.95898348e-01 4.02085111e-02 5.69308877e-01
7.69164264e-01 1.38296917e-01 -1.53899324e+00 -3.22898507e-01
4.88350540e-01 -1.67381719e-01 -8.79126564e-02 -8.77286792e-01
-4.09131557e-01 7.72249103e-01 3.76569569e-01 5.12178898e-01
6.37470603e-01 8.53525102e-03 5.43222964e-01 8.39717269e-01
6.73265100e-01 -1.40377402e+00 1.26065210e-01 6.89852357e-01
6.90523148e-01 -1.22737694e+00 -2.35436022e-01 5.27596593e-01
-9.50478315e-01 7.89986074e-01 5.51832378e-01 -2.76977271e-01
1.56493872e-01 3.59237343e-01 3.29267085e-01 -1.23581953e-01
-1.15336347e+00 -1.66951448e-01 -5.36526263e-01 5.12656391e-01
7.64313161e-01 2.68377692e-01 -4.96677190e-01 -1.89301163e-01
-7.34901130e-01 -2.24626049e-01 6.40290618e-01 9.49493170e-01
-9.01286781e-01 -1.29138863e+00 -3.22961539e-01 4.62802276e-02
-6.24255717e-01 -1.88832358e-01 -7.23937333e-01 7.90804982e-01
-3.27817261e-01 9.74777460e-01 5.48723303e-02 -4.54741903e-02
2.51239985e-01 7.20919222e-02 5.27599931e-01 -7.43584394e-01
-7.87321568e-01 2.72295207e-01 1.53099194e-01 -4.91011500e-01
-2.48620227e-01 -7.21583009e-01 -1.67084372e+00 -4.72445995e-01
-2.68640462e-02 4.93627518e-01 6.06216013e-01 9.82395232e-01
1.68124288e-01 1.94809690e-01 5.45729041e-01 -2.56828338e-01
-1.04085684e+00 -6.68435574e-01 -2.76567221e-01 3.08008820e-01
7.13034093e-01 -1.50384039e-01 -3.03994447e-01 -1.37053475e-01] | [4.053663730621338, 1.5808204412460327] |
0e6c682f-8a66-440e-b6af-3c4fdbce5a16 | inhomogeneous-hypergraph-clustering-with | 1709.01249 | null | http://arxiv.org/abs/1709.01249v4 | http://arxiv.org/pdf/1709.01249v4.pdf | Inhomogeneous Hypergraph Clustering with Applications | Hypergraph partitioning is an important problem in machine learning, computer
vision and network analytics. A widely used method for hypergraph partitioning
relies on minimizing a normalized sum of the costs of partitioning hyperedges
across clusters. Algorithmic solutions based on this approach assume that
different partitions of a hyperedge incur the same cost. However, this
assumption fails to leverage the fact that different subsets of vertices within
the same hyperedge may have different structural importance. We hence propose a
new hypergraph clustering technique, termed inhomogeneous hypergraph
partitioning, which assigns different costs to different hyperedge cuts. We
prove that inhomogeneous partitioning produces a quadratic approximation to the
optimal solution if the inhomogeneous costs satisfy submodularity constraints.
Moreover, we demonstrate that inhomogenous partitioning offers significant
performance improvements in applications such as structure learning of
rankings, subspace segmentation and motif clustering. | ['Olgica Milenkovic', 'Pan Li'] | 2017-09-05 | inhomogeneous-hypergraph-clustering-with-1 | http://papers.nips.cc/paper/6825-inhomogeneous-hypergraph-clustering-with-applications | http://papers.nips.cc/paper/6825-inhomogeneous-hypergraph-clustering-with-applications.pdf | neurips-2017-12 | ['hypergraph-partitioning'] | ['graphs'] | [ 7.16773868e-02 2.55991936e-01 -3.74451429e-01 -2.11357042e-01
-2.95355111e-01 -1.05554926e+00 -1.32596821e-01 4.25463408e-01
-8.35593268e-02 4.15133655e-01 -5.53037524e-02 -1.04193576e-01
-6.24493301e-01 -9.88080502e-01 -5.79311848e-01 -7.39192188e-01
-1.45839438e-01 9.66896594e-01 1.46727145e-01 2.35149384e-01
2.53834337e-01 5.24286866e-01 -1.13244975e+00 3.23251933e-01
9.21770155e-01 3.35401624e-01 -1.12464651e-01 5.12318671e-01
-1.58870488e-01 1.02290614e-02 -2.44311526e-01 -2.56972194e-01
6.29042447e-01 -1.78546607e-01 -8.67045820e-01 4.61228430e-01
4.08025742e-01 2.23303795e-01 -1.36471823e-01 1.30717087e+00
1.63552910e-01 1.70041218e-01 9.13939416e-01 -1.54538131e+00
-2.60236233e-01 6.95672810e-01 -1.18419480e+00 -7.66496286e-02
1.12887360e-02 -3.95572871e-01 1.44620061e+00 -2.78449237e-01
6.79086149e-01 9.30795312e-01 3.64743918e-01 3.59031782e-02
-1.87202072e+00 -4.02430415e-01 3.16504508e-01 1.97550133e-01
-1.65593541e+00 1.60430521e-01 7.09181786e-01 -6.16952956e-01
4.97102588e-01 7.60951996e-01 6.54484093e-01 1.51653290e-01
-1.47093937e-01 5.34793139e-01 1.04899168e+00 -2.49975801e-01
3.10372263e-01 3.22345793e-01 6.53362393e-01 6.34796143e-01
7.56278336e-01 -5.62998116e-01 -4.65119146e-02 -3.67431670e-01
3.23590875e-01 3.55703384e-02 -3.87281567e-01 -8.93029213e-01
-9.73296285e-01 9.43948984e-01 3.44520450e-01 -9.37258732e-03
-1.15826197e-01 3.77302282e-02 3.26561898e-01 2.68102288e-01
5.92686981e-02 5.02216756e-01 -3.16836298e-01 3.44492465e-01
-8.67202640e-01 6.80557340e-02 9.88164902e-01 1.11302066e+00
1.00664663e+00 -4.95796978e-01 1.74350902e-01 7.20128119e-01
1.51340529e-01 2.65302002e-01 -1.29633859e-01 -1.13787866e+00
2.46674582e-01 1.00160062e+00 -5.75784333e-02 -1.40667856e+00
-5.28810561e-01 -1.42467663e-01 -9.21295047e-01 -1.46704882e-01
3.25453490e-01 5.87911792e-02 -8.63697231e-01 1.72098720e+00
5.44547975e-01 -7.10091293e-02 -3.49216342e-01 8.72515976e-01
4.23962951e-01 5.23869395e-01 -2.57606089e-01 -4.23054039e-01
1.06941998e+00 -5.89388847e-01 -3.53985608e-01 2.11235046e-01
4.58548903e-01 -5.77458501e-01 7.71064460e-01 4.10701782e-01
-9.97754931e-01 -8.60295519e-02 -7.37484813e-01 2.35432342e-01
-3.62682566e-02 -3.04498702e-01 4.59484220e-01 7.68023849e-01
-9.77413774e-01 3.30019355e-01 -6.15414679e-01 -4.05037165e-01
6.83526509e-03 7.43453145e-01 -2.91142136e-01 -1.54744998e-01
-5.83329856e-01 3.53933811e-01 5.75907648e-01 -2.00470880e-01
-3.38194042e-01 -5.45551479e-01 -5.37605226e-01 3.89785290e-01
6.49224460e-01 -6.11418366e-01 5.97968459e-01 -9.07492697e-01
-9.29506958e-01 1.00438070e+00 -1.49374872e-01 -1.99422896e-01
3.39939505e-01 2.25716338e-01 1.46471886e-02 1.73566252e-01
1.56527609e-02 3.31532151e-01 4.66325879e-01 -1.41911876e+00
-4.49066639e-01 -6.09748304e-01 2.26324096e-01 3.18018109e-01
-2.89843529e-01 -3.60444248e-01 -8.25881839e-01 -1.87674522e-01
4.81088668e-01 -1.49002802e+00 -4.25174981e-01 -6.71058953e-01
-1.02758813e+00 -9.54155773e-02 5.15353978e-01 -1.07659891e-01
1.36567628e+00 -1.88852966e+00 6.22631252e-01 1.07899213e+00
7.63063014e-01 -2.50975132e-01 4.60852347e-02 6.51839018e-01
-1.03468141e-02 2.53943413e-01 -3.33977401e-01 3.99268121e-01
7.71749839e-02 1.02222614e-01 -4.44252677e-02 7.39942133e-01
-4.78355080e-01 5.02693295e-01 -7.50023305e-01 -3.33659917e-01
5.90312034e-02 -7.70901665e-02 -8.84547651e-01 -1.71480089e-01
-1.65218472e-01 8.81955698e-02 -3.15155387e-01 1.14829466e-01
1.10335290e+00 -5.05724192e-01 1.08724368e+00 -1.89467579e-01
2.34717280e-02 -2.18496934e-01 -1.62689066e+00 1.15472305e+00
1.37133390e-01 6.39360011e-01 2.05152452e-01 -1.25072360e+00
4.58468825e-01 -7.84910808e-04 1.01699102e+00 -1.10884324e-01
1.21163502e-02 -2.60068709e-03 2.19403476e-01 -3.18215162e-01
4.44277525e-01 3.54287103e-02 -1.06378742e-01 7.02987134e-01
-3.83532614e-01 2.03310505e-01 5.38958132e-01 6.33885622e-01
1.28009355e+00 -5.32916009e-01 7.73669928e-02 -5.72484374e-01
2.00532168e-01 2.30875835e-01 7.40194798e-01 5.99818110e-01
6.74442649e-02 6.07244849e-01 1.01439440e+00 -8.53448138e-02
-1.18270564e+00 -1.06602955e+00 -1.38432562e-01 8.54979515e-01
5.05351663e-01 -5.54966569e-01 -1.08853519e+00 -6.61815822e-01
2.11846739e-01 2.86701750e-02 -4.49077576e-01 -1.36920586e-02
-3.55785787e-01 -1.02163219e+00 9.08107385e-02 2.37222567e-01
-3.62727717e-02 -4.41999346e-01 -4.27991837e-01 -9.07190293e-02
-2.42444396e-01 -1.00669086e+00 -7.53094912e-01 1.91400364e-01
-8.76977563e-01 -1.46388507e+00 -4.39486980e-01 -7.18863010e-01
1.11701488e+00 8.19146514e-01 9.72447276e-01 2.30923787e-01
-1.88061237e-01 5.27389586e-01 -2.83036768e-01 1.62734956e-01
1.26582742e-01 4.67378289e-01 1.04658671e-01 1.19586609e-01
3.68726969e-01 -4.01767701e-01 -5.27457118e-01 3.62313867e-01
-1.07127059e+00 -5.24984635e-02 3.24976146e-01 5.77023268e-01
9.68815029e-01 4.71554667e-01 2.55122155e-01 -1.62041724e+00
4.40167964e-01 -7.45049953e-01 -7.79066026e-01 4.12544012e-01
-6.72726691e-01 2.69998044e-01 5.71307003e-01 -2.12708160e-01
-5.47332764e-01 1.89274982e-01 6.02289021e-01 -4.43044305e-01
3.21749821e-02 6.16456628e-01 -5.52196503e-01 -1.03514329e-01
1.48871914e-01 -1.32241607e-01 -3.52307528e-01 -2.27982059e-01
5.55444956e-01 4.33549404e-01 1.20415151e-01 -4.08184111e-01
8.64823103e-01 7.60177433e-01 4.80140001e-01 -1.03596509e+00
-4.68816161e-01 -9.33879137e-01 -7.58172929e-01 -1.89307496e-01
7.52515614e-01 -4.94239867e-01 -1.07551193e+00 -7.70093948e-02
-7.62335181e-01 -2.36967448e-02 -3.46942954e-02 3.37596267e-01
-4.40129519e-01 7.25450814e-01 -3.51971865e-01 -5.95881999e-01
2.11368218e-01 -9.45670187e-01 6.38517976e-01 1.93603545e-01
-2.33462781e-01 -7.06350803e-01 3.72003585e-01 4.27993298e-01
-2.96172172e-01 4.20942962e-01 1.44660223e+00 -6.02409244e-01
-9.36550140e-01 -1.06594019e-01 -3.65611881e-01 -1.99918598e-01
-4.93424945e-03 1.90868005e-01 -4.00184125e-01 -4.98928934e-01
-5.51521897e-01 7.70349230e-04 9.38581705e-01 6.97134793e-01
1.09512758e+00 -4.49513108e-01 -5.97027659e-01 7.04860866e-01
1.68925369e+00 2.01008841e-01 4.59507883e-01 1.94514975e-01
1.12693167e+00 1.14416325e+00 2.29757741e-01 2.93455273e-01
3.53632838e-01 7.09110498e-01 2.33491793e-01 -1.04944669e-01
5.11734784e-01 1.48455769e-01 -5.78882508e-02 8.88227403e-01
1.06793121e-02 -4.89172548e-01 -8.18565786e-01 6.18704855e-01
-2.01171517e+00 -9.32201326e-01 -8.44686031e-01 2.63462496e+00
5.24515986e-01 6.81209937e-03 3.49809289e-01 2.08209351e-01
1.06194913e+00 -2.13312387e-01 -5.74582577e-01 -6.05124295e-01
-1.35784060e-01 -5.63117974e-02 8.18052888e-01 6.12938285e-01
-1.00862098e+00 7.82154083e-01 5.94914532e+00 5.36955059e-01
-5.46400428e-01 -1.59553558e-01 5.57951152e-01 -2.71204621e-01
-5.18655121e-01 1.96328759e-01 -5.40661812e-01 5.08409739e-01
5.02134204e-01 -4.21192527e-01 6.02698684e-01 7.03224540e-01
1.04910553e-01 -2.11662218e-01 -9.58375335e-01 6.60363317e-01
-8.80975053e-02 -9.53606784e-01 -5.11852698e-03 6.34059489e-01
1.28177249e+00 -2.22431943e-01 5.76181598e-02 -2.14980379e-01
6.20786071e-01 -8.54272485e-01 2.77068138e-01 1.05962768e-01
6.36973321e-01 -1.19078374e+00 2.84615129e-01 1.12735473e-01
-1.31219375e+00 1.54466685e-02 -5.57260573e-01 3.09914291e-01
6.71909899e-02 7.58078396e-01 -8.27355385e-01 5.49999237e-01
3.99638832e-01 1.56171724e-01 -2.57760823e-01 1.37998128e+00
2.78916210e-02 4.77194399e-01 -4.16946977e-01 3.92536223e-01
7.43196979e-02 -8.90327394e-01 7.10442185e-01 9.97179091e-01
-1.57112464e-01 1.76106587e-01 5.60068488e-01 7.29319155e-01
-2.29880542e-01 3.47290188e-01 -7.29144216e-01 -9.30386409e-02
4.25579011e-01 1.33036911e+00 -1.50001585e+00 -5.42483814e-02
-4.75444257e-01 8.81343424e-01 4.39587682e-01 5.19370317e-01
-7.34533310e-01 -3.63403291e-01 7.04719424e-01 2.96203375e-01
3.77445996e-01 -3.52378517e-01 -6.28757536e-01 -9.14898515e-01
-8.28164965e-02 -5.75031698e-01 7.31010079e-01 -1.81757867e-01
-9.85820770e-01 3.28108966e-01 4.40008305e-02 -9.63442504e-01
1.32595271e-01 -3.97792339e-01 -4.14037436e-01 3.37859958e-01
-1.04839718e+00 -7.74834156e-01 -3.25433820e-01 2.70478398e-01
1.62173375e-01 2.90769875e-01 3.91080916e-01 2.30009615e-01
-7.64512777e-01 2.77165949e-01 5.16749084e-01 -5.96007891e-02
5.64967513e-01 -1.61036086e+00 -2.23357767e-01 7.25455999e-01
3.13618898e-01 6.91106021e-01 5.32754064e-01 -7.48889625e-01
-1.46285069e+00 -1.09385204e+00 6.07334733e-01 -2.60615081e-01
3.18991870e-01 -2.92929500e-01 -8.95023942e-01 6.46101892e-01
2.46472716e-01 -3.86725932e-01 1.02182066e+00 4.38866913e-01
-3.51037562e-01 5.39286397e-02 -1.04026830e+00 5.81606925e-01
8.87704909e-01 -2.74860322e-01 7.11271120e-03 3.57940853e-01
5.11983812e-01 3.35027799e-02 -8.26207280e-01 4.16977912e-01
6.32806063e-01 -8.62299860e-01 8.85758638e-01 -9.13697422e-01
3.58334482e-01 -5.05040288e-01 -1.26458794e-01 -1.23385334e+00
-6.49535120e-01 -4.68262553e-01 4.06821631e-02 1.03358603e+00
4.52258199e-01 -3.96533191e-01 1.14373934e+00 8.05547476e-01
1.96628705e-01 -6.71681881e-01 -4.81670231e-01 -8.44263613e-01
-4.01017033e-02 -1.38000241e-02 3.71830374e-01 1.33560991e+00
3.19769800e-01 4.89375353e-01 -5.13496697e-02 3.75038564e-01
9.89269257e-01 5.97004652e-01 7.73996413e-01 -1.65492797e+00
-3.99002641e-01 -6.86307430e-01 -4.46537912e-01 -6.82925403e-01
3.92688215e-01 -1.16585124e+00 -2.13978603e-01 -1.67561221e+00
9.59106565e-01 -6.72463059e-01 -6.75048679e-02 2.63601076e-03
-1.94062158e-01 4.37445253e-01 1.46206677e-01 3.76249611e-01
-7.66557097e-01 5.58695458e-02 9.58910286e-01 -2.65746564e-01
-6.59533858e-01 -2.21589040e-02 -6.23422623e-01 6.53161287e-01
6.94185495e-01 -5.48003614e-01 -6.48087740e-01 -3.17172021e-01
5.25182724e-01 -2.74593756e-02 -8.38747099e-02 -4.96745706e-01
2.62291819e-01 -3.98553103e-01 3.77651677e-02 -5.78025341e-01
-4.26648855e-02 -9.93049860e-01 6.45037591e-01 3.62001121e-01
-3.65760446e-01 2.91740103e-03 4.75981906e-02 9.41781700e-01
3.65597419e-02 -2.54649341e-01 9.95117843e-01 -7.86995664e-02
-3.54519308e-01 3.20089638e-01 -4.41165268e-01 2.12570518e-01
1.46090984e+00 -3.97796720e-01 -1.86697900e-01 -1.93705216e-01
-7.04550385e-01 5.23954630e-01 7.66335964e-01 2.51086876e-02
3.62667173e-01 -1.09342909e+00 -4.93556201e-01 -1.67532369e-01
4.97547202e-02 -5.30149750e-02 2.82779574e-01 1.04852641e+00
-4.94514614e-01 5.72558343e-01 3.48679372e-03 -6.79409444e-01
-1.75612795e+00 6.96440339e-01 3.28146964e-02 -2.87443399e-01
-5.06563246e-01 6.76298320e-01 9.10245657e-01 -3.11059386e-01
-1.55305211e-02 3.22696827e-02 -1.52178749e-01 2.11701542e-01
-1.45534769e-01 8.82928073e-01 -4.25883420e-02 -7.46655464e-01
-2.50741541e-01 6.13110840e-01 -2.58357793e-01 -2.19426423e-01
1.07225561e+00 -2.66451389e-01 -3.91648263e-01 3.26283425e-01
1.08836687e+00 2.45728180e-01 -8.38620663e-01 7.11823069e-03
3.45361173e-01 -6.69019461e-01 -1.74244955e-01 -2.64066488e-01
-1.22228026e+00 4.57764417e-01 7.52376989e-02 7.96639562e-01
1.17433524e+00 1.68561235e-01 6.66467905e-01 3.24246347e-01
3.47393572e-01 -1.26642919e+00 -2.03491047e-01 2.69007444e-01
2.82156020e-01 -8.61212909e-01 1.46940246e-01 -9.07324672e-01
-6.73305690e-01 7.93546379e-01 6.42903566e-01 -1.91746548e-01
6.04818821e-01 8.37049112e-02 -5.08709192e-01 -4.82154101e-01
-5.17762244e-01 -4.59232897e-01 3.95927995e-01 3.07478428e-01
2.85446674e-01 5.55937767e-01 -7.25489676e-01 2.66322523e-01
-2.46062446e-02 -7.04003930e-01 6.62495673e-01 6.00190997e-01
-5.10549903e-01 -1.16523409e+00 -3.16611230e-01 7.02919066e-01
-3.62987399e-01 -5.28621338e-02 -8.87643158e-01 5.15277207e-01
7.85517134e-03 7.22272456e-01 1.29597396e-01 -4.34415281e-01
1.71060458e-01 -2.15828031e-01 5.64530611e-01 -8.49024713e-01
-2.73819536e-01 3.74558061e-01 -2.01238737e-01 -2.93060720e-01
-2.48415351e-01 -5.48219621e-01 -1.49536753e+00 -5.38968682e-01
-5.41217804e-01 4.08611953e-01 3.29984307e-01 6.53885543e-01
3.74845445e-01 3.95290107e-01 9.29489255e-01 -2.72747815e-01
-7.67227784e-02 -4.51137483e-01 -1.12254941e+00 4.59282666e-01
-2.33703516e-02 -4.58111107e-01 -3.65698963e-01 1.27213836e-01] | [7.093282699584961, 5.148623466491699] |
cef9d3d0-d436-4c8c-95ad-81cbff0d5dbb | image-storage-on-synthetic-dna-using-1 | 2306.12882 | null | https://arxiv.org/abs/2306.12882v1 | https://arxiv.org/pdf/2306.12882v1.pdf | Image storage on synthetic DNA using compressive autoencoders and DNA-adapted entropy coders | Over the past years, the ever-growing trend on data storage demand, more specifically for "cold" data (rarely accessed data), has motivated research for alternative systems of data storage. Because of its biochemical characteristics, synthetic DNA molecules are now considered as serious candidates for this new kind of storage. This paper presents some results on lossy image compression methods based on convolutional autoencoders adapted to DNA data storage, with synthetic DNA-adapted entropic and fixed-length codes. The model architectures presented here have been designed to efficiently compress images, encode them into a quaternary code, and finally store them into synthetic DNA molecules. This work also aims at making the compression models better fit the problematics that we encounter when storing data into DNA, namely the fact that the DNA writing, storing and reading methods are error prone processes. The main take aways of this kind of compressive autoencoder are our latent space quantization and the different DNA adapted entropy coders used to encode the quantized latent space, which are an improvement over the fixed length DNA adapted coders that were previously used. | ['Marc Antonini', 'Melpomeni Dimopoulou', 'Eva Gil San Antonio', 'Xavier Pic'] | 2023-06-22 | null | null | null | null | ['image-compression', 'quantization'] | ['computer-vision', 'methodology'] | [ 4.50764507e-01 5.32029718e-02 -8.90929776e-04 -1.23841681e-01
-6.60991371e-02 -1.19630210e-01 7.63097525e-01 4.66322809e-01
-7.20224440e-01 8.20841908e-01 5.08662939e-01 -1.86206289e-02
2.34781802e-02 -1.06626236e+00 -9.10376251e-01 -1.13395059e+00
-5.50820902e-02 4.69153136e-01 -4.12953980e-02 -2.49758303e-01
5.60775757e-01 6.98242843e-01 -2.33371067e+00 6.00692213e-01
5.67790687e-01 9.01703596e-01 6.81849897e-01 1.02976441e+00
-2.47923866e-01 9.55689430e-01 -5.66358864e-01 -4.14426535e-01
-4.94024903e-02 -4.92337912e-01 -7.48421907e-01 -2.09004238e-01
-1.39592975e-01 -4.46034849e-01 -8.27710927e-01 8.69436979e-01
4.15556490e-01 1.81374922e-02 9.29074645e-01 -5.03613234e-01
-1.17397761e+00 6.60259843e-01 5.28594315e-01 2.33606815e-01
2.39260927e-01 9.13214907e-02 3.94331753e-01 -6.76473379e-01
8.54680479e-01 1.13010418e+00 5.25932610e-01 7.05190182e-01
-9.80662405e-01 1.35357827e-01 -9.35987711e-01 5.75616598e-01
-1.42740929e+00 -6.31267428e-01 3.57207775e-01 -4.47438508e-01
1.63954949e+00 5.56162417e-01 8.74742270e-01 1.34388220e+00
7.99693942e-01 5.60942471e-01 6.51538193e-01 -6.05373144e-01
5.96286297e-01 2.44035020e-01 -2.06490681e-01 3.28226268e-01
5.03898144e-01 2.87360936e-01 -4.50585842e-01 2.18366355e-01
4.98138487e-01 1.97537780e-01 -2.23994538e-01 -1.99303344e-01
-8.96161139e-01 8.75752032e-01 2.14009136e-01 8.89970243e-01
-4.09523994e-01 1.15723088e-01 5.61366141e-01 2.86940753e-01
1.35792792e-01 3.05030197e-01 9.31471363e-02 -4.55736458e-01
-1.27211714e+00 2.54318565e-01 8.51174891e-01 8.85346293e-01
6.01066172e-01 1.96683273e-01 -1.24638021e-01 6.36580229e-01
1.95636570e-01 4.77086276e-01 1.30558598e+00 -7.09795773e-01
1.42091647e-01 4.20124888e-01 -1.54377967e-01 -1.22579896e+00
-1.54657349e-01 -2.59783268e-01 -1.24200284e+00 -1.98860094e-01
-1.49848953e-01 6.05766773e-01 -9.35714185e-01 1.21569514e+00
-2.20818788e-01 -2.80582428e-01 5.94230533e-01 7.00414121e-01
4.63478148e-01 1.25670183e+00 -2.01690391e-01 -3.32695574e-01
1.11170506e+00 -6.86764777e-01 -1.23549187e+00 3.67029309e-01
7.61063457e-01 -5.99103570e-01 6.51493430e-01 3.43085557e-01
-1.35011828e+00 -6.94960594e-01 -1.40860581e+00 -5.69976270e-01
-9.44274366e-01 -5.74757457e-02 2.87430733e-01 8.59249413e-01
-1.31060970e+00 1.15351999e+00 -8.00214708e-01 -2.92512029e-01
1.50941551e-01 2.15380043e-01 -3.36503655e-01 3.94503772e-03
-1.28083980e+00 1.12107313e+00 1.04342413e+00 9.27596446e-03
-7.85173178e-01 -2.01998666e-01 -6.66785061e-01 5.70856035e-01
-3.83334845e-01 -3.20400715e-01 6.59775078e-01 -6.39078915e-01
-1.51212442e+00 8.32446754e-01 9.05522481e-02 -9.82715905e-01
3.14925872e-02 1.24799743e-01 -5.36434114e-01 2.59844303e-01
-5.46601474e-01 5.28412759e-01 6.48498118e-01 -7.04007924e-01
2.14410558e-01 -1.59362867e-01 -6.53642416e-01 -1.87839374e-01
-7.42413402e-01 -3.26545864e-01 5.22988811e-02 -7.90502846e-01
-2.52408504e-01 -7.73342073e-01 1.56524375e-01 -1.13466822e-01
-1.06594376e-02 4.21090275e-02 7.97506630e-01 -7.89084852e-01
1.40101278e+00 -2.19724822e+00 7.60274649e-01 6.65741339e-02
-1.61693230e-01 7.69939244e-01 -7.49683753e-02 8.83557677e-01
-1.89252138e-01 7.70040601e-02 -4.13432658e-01 -2.49405637e-01
-1.34135231e-01 5.90934992e-01 -3.53040874e-01 2.85601765e-01
9.14252624e-02 8.46563935e-01 -5.39115131e-01 -3.71181309e-01
6.99406564e-02 7.58688569e-01 -5.22817850e-01 3.48944783e-01
-3.38748068e-01 -1.13807492e-01 1.59421619e-02 3.36481214e-01
7.48970270e-01 -1.81512192e-01 8.26456323e-02 -1.84486210e-01
-3.75933975e-01 1.92328319e-01 -6.67652607e-01 1.79692984e+00
-3.22966836e-02 7.44869709e-01 -3.58395368e-01 -1.10394430e+00
1.07054639e+00 1.66248441e-01 3.68847489e-01 -9.84837174e-01
1.37702480e-01 3.56946617e-01 -1.79403991e-01 -8.54262769e-01
1.11042237e+00 -8.53490978e-02 3.09446275e-01 2.42894843e-01
3.11873585e-01 -6.78986385e-02 4.09649491e-01 1.72103912e-01
9.08332586e-01 -4.98970188e-02 2.24020794e-01 -2.69019008e-01
6.67574286e-01 -2.03436926e-01 4.03037518e-02 4.66965765e-01
1.11066215e-02 6.48311377e-01 1.79745525e-01 -6.54682636e-01
-2.09432721e+00 -6.82377517e-01 -1.45832539e-01 4.34816420e-01
-2.04191700e-01 -5.49774110e-01 -9.65364993e-01 3.58564198e-01
-1.44538075e-01 5.82802057e-01 -4.41295654e-01 -6.30457699e-01
-5.82651138e-01 -6.36027813e-01 9.88678396e-01 2.70863116e-01
3.32495958e-01 -1.16454387e+00 -9.78012145e-01 4.35593069e-01
-2.92935759e-01 -5.53205669e-01 9.52881798e-02 4.96380240e-01
-9.57333624e-01 -4.97675359e-01 -1.01235259e+00 -5.12386799e-01
2.82782525e-01 -1.43051073e-01 7.65291393e-01 2.99513370e-01
-5.50564826e-01 -2.44899541e-02 -8.22931468e-01 -3.16859007e-01
-1.09305453e+00 4.74299081e-02 1.41796649e-01 -1.24239132e-01
4.35754418e-01 -5.56957603e-01 -6.56322360e-01 -5.04394412e-01
-1.62808955e+00 -3.06658796e-04 9.04710770e-01 1.01388049e+00
4.98618156e-01 5.16377278e-02 8.07858407e-02 -4.97266740e-01
6.20408475e-01 -5.91833949e-01 -2.53770977e-01 2.59181738e-01
-7.17837214e-01 6.78227305e-01 8.14113736e-01 -1.54412284e-01
-8.10197592e-01 -1.92187637e-01 -5.41718185e-01 -3.46460372e-01
-1.34587571e-01 5.71453989e-01 1.23245344e-01 4.94747572e-02
8.65544915e-01 1.02798676e+00 4.36874539e-01 -4.95790601e-01
1.55164286e-01 1.10387087e+00 4.44276571e-01 -1.23531237e-01
2.14032650e-01 3.15525651e-01 1.41379640e-01 -1.06907713e+00
2.75768582e-02 -1.13792874e-01 -5.27445853e-01 3.54534341e-03
9.49118614e-01 -6.15229428e-01 -6.32807851e-01 5.70126295e-01
-1.24607456e+00 9.60897058e-02 -5.61122596e-01 4.05507118e-01
-9.49765205e-01 7.83527195e-01 -1.10683155e+00 -7.31555641e-01
-3.92743617e-01 -1.24198651e+00 9.03266251e-01 1.19756117e-01
5.72347976e-02 -7.21798956e-01 4.09604788e-01 7.97840580e-03
9.34253335e-01 2.69070454e-02 1.08796716e+00 -2.49210641e-01
-6.92049861e-01 -9.64225456e-02 1.23102173e-01 5.65007269e-01
-1.87032640e-01 -2.30239742e-02 -8.74303401e-01 -5.71380913e-01
2.99466521e-01 -5.28159201e-01 1.21920216e+00 1.31686941e-01
1.46456635e+00 -7.28672326e-01 -1.61203116e-01 6.83881402e-01
1.53453720e+00 4.25084174e-01 1.38330305e+00 3.94574285e-01
3.92887890e-02 3.34993511e-01 2.37531230e-01 6.29899442e-01
-1.25578448e-01 6.05662405e-01 4.90961701e-01 5.10396600e-01
-1.89166605e-01 -3.25823992e-01 4.32041615e-01 1.66146636e+00
-2.05646813e-01 -6.83560729e-01 -6.44690275e-01 3.85013819e-01
-1.45168591e+00 -1.33325577e+00 2.81354159e-01 2.01133156e+00
1.03833318e+00 -3.19984674e-01 -2.69133031e-01 5.92531800e-01
3.08773220e-01 5.93432412e-02 -4.16311443e-01 -1.08638132e+00
-5.85345864e-01 4.32745218e-01 5.99070549e-01 2.09914640e-01
-9.20316100e-01 3.98000866e-01 6.79022169e+00 1.02963519e+00
-1.37151206e+00 -1.28627315e-01 4.58017528e-01 -5.51049374e-02
-4.17909682e-01 -2.39210919e-01 -6.52709901e-01 1.11205173e+00
2.00633168e+00 1.64327733e-02 5.09736776e-01 7.51477540e-01
-2.61624753e-01 4.90313545e-02 -9.46001291e-01 1.10342193e+00
3.01185250e-01 -1.95177770e+00 5.87362289e-01 1.74972191e-01
3.86143982e-01 -2.44839102e-01 2.40541235e-01 2.34443247e-02
-4.93079931e-01 -1.42291915e+00 8.76275301e-01 1.23001683e+00
8.36182535e-01 -6.33461475e-01 9.47077632e-01 5.31911016e-01
-6.22144163e-01 -2.97464877e-01 -1.09876990e+00 -1.08737476e-01
-1.84352808e-02 8.24577034e-01 -5.35433829e-01 4.45621252e-01
7.00656176e-01 6.25258267e-01 -4.37552840e-01 7.50610888e-01
6.37501657e-01 1.99372128e-01 -1.74791917e-01 -4.71430868e-01
1.46101326e-01 -1.24945268e-01 2.44212225e-01 1.48764300e+00
8.70372593e-01 6.08535036e-02 -6.66423261e-01 9.93293822e-01
1.03353292e-01 -8.22174996e-02 -1.05453241e+00 -7.09165394e-01
2.90587217e-01 4.79712963e-01 -4.53356028e-01 -5.13449728e-01
-1.19786344e-01 1.21587837e+00 1.55448824e-01 -3.37971970e-02
-4.96838570e-01 -6.55130625e-01 2.64783084e-01 4.85267043e-02
5.03522813e-01 -4.44863051e-01 8.09919834e-02 -1.20674872e+00
-5.05972765e-02 -8.00133228e-01 -1.12817869e-01 -8.22933376e-01
-7.59870172e-01 5.02601087e-01 -2.23470345e-01 -1.12515068e+00
-4.34704155e-01 -7.59844303e-01 2.45958492e-01 6.95810735e-01
-1.32604814e+00 -5.96180201e-01 -1.90307852e-02 2.91020036e-01
3.69436771e-01 -4.75693017e-01 1.34736562e+00 5.70724130e-01
-1.11743137e-01 6.50413871e-01 1.07926524e+00 -2.48182595e-01
1.87648371e-01 -7.42719710e-01 2.25056157e-01 6.53926611e-01
-2.97757238e-02 8.30642760e-01 8.79357994e-01 -6.36330724e-01
-1.79376483e+00 -8.17494273e-01 1.14377511e+00 -8.99817199e-02
9.33695659e-02 -3.76111031e-01 -1.14270234e+00 3.26920331e-01
2.62912899e-01 -5.05163789e-01 8.14549804e-01 -7.39388585e-01
-3.55041981e-01 9.21645835e-02 -1.24534917e+00 2.39977568e-01
7.08841562e-01 -7.60451853e-01 -5.72076142e-01 2.81303048e-01
8.26148212e-01 -2.71974683e-01 -1.11728203e+00 1.48985311e-01
6.02549851e-01 -1.15443826e+00 1.10639381e+00 -3.20267677e-01
9.84438479e-01 -1.28674269e-01 -4.32722539e-01 -1.11904228e+00
-4.92942154e-01 -1.50309995e-01 -5.65684795e-01 7.61630833e-01
-1.55724406e-01 -1.10274673e-01 8.63851428e-01 2.20170781e-01
-1.18398927e-01 -6.47239983e-01 -1.14984882e+00 -9.44936633e-01
3.97230424e-02 3.44290346e-01 6.30140960e-01 7.19355643e-01
2.20513806e-01 -3.21678042e-01 -7.47552276e-01 -4.98130411e-01
3.54494363e-01 -1.95819080e-01 1.78654030e-01 -9.61441576e-01
-3.71077091e-01 -3.44499081e-01 -9.98982370e-01 -1.05608189e+00
-2.23561063e-01 -1.13773572e+00 -6.52341917e-02 -8.57059836e-01
2.56046921e-01 -3.13535184e-01 -1.74243540e-01 -2.18855450e-03
4.81880039e-01 5.68833165e-02 2.17202604e-01 4.36452836e-01
-1.90873429e-01 8.89137983e-01 7.66971469e-01 -4.42315906e-01
3.67165178e-01 -7.04207659e-01 -1.83782354e-01 -9.28436890e-02
6.20000243e-01 -3.36657614e-01 -2.43367836e-01 -5.48947036e-01
5.75522661e-01 1.87169239e-01 1.23984002e-01 -1.49540544e+00
4.67862189e-01 7.29629919e-02 4.98039633e-01 -6.74268126e-01
6.00689828e-01 -8.04104090e-01 7.32064188e-01 9.30800855e-01
-6.53813899e-01 1.54909521e-01 1.82681009e-02 5.93629062e-01
-4.91373032e-01 -7.60015428e-01 8.61243904e-01 -4.10052299e-01
-4.29852903e-01 -5.79853840e-02 -9.99685287e-01 -7.55745173e-01
9.30724263e-01 -5.33703387e-01 -3.18628401e-01 -1.39175206e-01
-7.42091894e-01 -5.62887967e-01 8.24501395e-01 2.55718976e-01
1.01688015e+00 -1.15040040e+00 -4.28998500e-01 5.88728011e-01
-3.78501862e-02 -2.93430865e-01 5.39488673e-01 1.77755862e-01
-1.29476702e+00 1.11663556e+00 -9.00978565e-01 -3.89699012e-01
-1.03735673e+00 1.16571283e+00 7.27628767e-02 -1.03900827e-01
-5.09325802e-01 6.76093578e-01 -6.36046171e-01 1.03168070e-01
1.78782001e-01 -3.94568115e-01 -2.21366808e-01 -1.25428021e-01
8.24914098e-01 5.55783689e-01 4.43986386e-01 -6.22137666e-01
3.99235822e-02 3.74669552e-01 -1.04162097e-01 2.51752824e-01
1.56153202e+00 -1.18631527e-01 -4.81169641e-01 5.61746955e-01
1.41211724e+00 -5.72664797e-01 -5.88250875e-01 2.01062068e-01
-5.70510551e-02 -3.86407495e-01 -1.83799401e-01 -4.19408381e-01
-6.57593071e-01 1.23355830e+00 8.80110085e-01 4.37302083e-01
1.00514615e+00 -3.76738846e-01 9.43156242e-01 6.87566757e-01
3.14146489e-01 -1.36422229e+00 8.05341676e-02 5.76498091e-01
8.79374921e-01 -6.33383751e-01 -4.72791083e-02 9.41572115e-02
-6.46959171e-02 1.51654243e+00 -1.96312845e-01 5.55723300e-03
3.62823278e-01 4.11698222e-01 -5.56634665e-01 -2.91566700e-02
-8.51692259e-01 3.16608995e-01 -3.35999221e-01 6.41746998e-01
5.02084374e-01 8.27339143e-02 -6.33924305e-01 -1.75600387e-02
-3.28551471e-01 4.40509200e-01 8.85143340e-01 1.11171150e+00
-7.58855164e-01 -1.31281030e+00 -3.86505127e-01 4.80424464e-01
-2.77004927e-01 2.60703955e-02 -8.02787319e-02 1.42801493e-01
3.18318069e-01 3.39650780e-01 1.41595632e-01 -5.29003382e-01
-1.74324557e-01 4.03506428e-01 5.44871449e-01 -1.89125519e-02
-3.32240671e-01 -5.02516985e-01 -4.47827667e-01 -5.37003100e-01
-5.42185843e-01 -5.23305476e-01 -1.11591601e+00 -7.59291232e-01
-9.76471677e-02 2.35869318e-01 1.09011745e+00 5.77038169e-01
8.24640810e-01 4.49691713e-01 2.56988317e-01 -9.29864287e-01
-6.39530659e-01 -8.83572936e-01 -6.30803406e-01 6.70402646e-01
5.38717389e-01 -3.81170928e-01 -1.18402377e-01 4.72459286e-01] | [11.437989234924316, -1.6727232933044434] |
be3fc4ba-780b-43dc-b2ab-f989ec204a42 | resetting-the-baseline-ct-based-covid-19 | 2108.05649 | null | https://arxiv.org/abs/2108.05649v1 | https://arxiv.org/pdf/2108.05649v1.pdf | Resetting the baseline: CT-based COVID-19 diagnosis with Deep Transfer Learning is not as accurate as widely thought | Deep learning is gaining instant popularity in computer aided diagnosis of COVID-19. Due to the high sensitivity of Computed Tomography (CT) to this disease, CT-based COVID-19 detection with visual models is currently at the forefront of medical imaging research. Outcomes published in this direction are frequently claiming highly accurate detection under deep transfer learning. This is leading medical technologists to believe that deep transfer learning is the mainstream solution for the problem. However, our critical analysis of the literature reveals an alarming performance disparity between different published results. Hence, we conduct a systematic thorough investigation to analyze the effectiveness of deep transfer learning for COVID-19 detection with CT images. Exploring 14 state-of-the-art visual models with over 200 model training sessions, we conclusively establish that the published literature is frequently overestimating transfer learning performance for the problem, even in the prestigious scientific sources. The roots of overestimation trace back to inappropriate data curation. We also provide case studies that consider more realistic scenarios, and establish transparent baselines for the problem. We hope that our reproducible investigation will help in curbing hype-driven claims for the critical problem of COVID-19 diagnosis, and pave the way for a more transparent performance evaluation of techniques for CT-based COVID-19 detection. | ['Naveed Akhtar', 'Syed M. S. Islam', 'Fouzia Altaf'] | 2021-08-12 | null | null | null | null | ['covid-19-detection'] | ['medical'] | [-2.66639031e-02 -1.09415673e-01 -4.45492625e-01 -1.63344797e-02
-1.28860176e+00 -5.24517000e-01 2.03665659e-01 2.97580212e-01
-5.65941989e-01 5.05762219e-01 2.64595568e-01 -1.13431096e+00
-3.40543315e-02 -4.58685130e-01 -6.81282640e-01 -6.53442681e-01
-2.32607275e-01 7.76742518e-01 -1.89034399e-02 1.80819407e-01
6.33354783e-02 5.03511608e-01 -6.11103296e-01 3.42746228e-01
3.81685853e-01 7.09762156e-01 2.06381768e-01 7.46557236e-01
2.40950331e-01 9.61376309e-01 -4.22337174e-01 -4.40027028e-01
-1.27550021e-01 -4.38205034e-01 -8.25986266e-01 -3.19802850e-01
4.97481495e-01 -6.88759029e-01 -1.43804803e-01 6.02103829e-01
8.10521066e-01 -7.03512132e-01 1.01414418e+00 -9.87278402e-01
-7.17872560e-01 1.73530057e-01 -8.69827628e-01 8.12329710e-01
-1.42811000e-01 8.03732634e-01 6.69354558e-01 -8.31513226e-01
7.43107021e-01 8.60318661e-01 1.01256204e+00 5.63380778e-01
-8.70506525e-01 -8.99477363e-01 -4.20691848e-01 4.25355613e-01
-1.12142956e+00 4.06525694e-02 2.94435531e-01 -9.77148950e-01
8.21355999e-01 2.00895071e-01 1.04609132e+00 1.42166328e+00
8.77252698e-01 5.51218987e-01 1.06042480e+00 -2.03150198e-01
-7.90456980e-02 1.06304072e-01 5.29929027e-02 8.17167580e-01
8.18615079e-01 3.99771124e-01 -2.69680060e-02 -2.76629895e-01
8.72930706e-01 -1.30110219e-01 -2.59873182e-01 -3.47688287e-01
-1.01941800e+00 1.08972931e+00 6.30815804e-01 3.11361760e-01
-1.97495103e-01 3.62502128e-01 9.46019292e-01 2.07057912e-02
4.23687428e-01 3.94803077e-01 1.43785039e-02 -4.50063236e-02
-8.59042466e-01 1.75158933e-01 2.09031358e-01 3.03868234e-01
-1.10260129e-01 5.11862747e-02 -6.36885315e-02 5.41303277e-01
2.09592551e-01 7.80981183e-01 3.42515081e-01 -7.43531764e-01
1.06967337e-01 1.22822709e-01 5.03864214e-02 -9.04190004e-01
-6.54502869e-01 -7.45230913e-01 -8.48829508e-01 5.58952093e-01
4.44836229e-01 -2.30532780e-01 -1.09947133e+00 1.35167396e+00
-1.50237102e-02 7.02620745e-02 -3.21347058e-01 1.07548928e+00
7.63784349e-01 7.93125108e-02 5.20266771e-01 -1.16839772e-02
1.61875618e+00 -4.84061033e-01 -4.48992521e-01 -3.13127227e-02
9.73516226e-01 -7.49563634e-01 9.41203177e-01 3.98345143e-01
-8.43375862e-01 -1.55616954e-01 -1.21436393e+00 -2.58149449e-02
-1.98285077e-02 -5.68157993e-02 7.66590118e-01 8.79485130e-01
-7.79022217e-01 1.32873356e-01 -1.17323732e+00 -5.01436412e-01
9.48318124e-01 1.71559036e-01 -1.71949580e-01 -1.50122508e-01
-9.32995319e-01 1.48603547e+00 -2.51892488e-02 3.51483300e-02
-1.39829433e+00 -1.12316585e+00 -4.39343244e-01 -1.10516220e-01
2.81140625e-01 -1.18697941e+00 1.35902953e+00 -7.04371750e-01
-5.24769425e-01 1.35729659e+00 1.96682960e-01 -5.96979618e-01
8.28859389e-01 8.31638873e-02 -2.58598000e-01 3.35233867e-01
2.37616688e-01 3.58875513e-01 8.27794671e-01 -1.44787717e+00
-5.06578982e-01 -3.37832242e-01 -3.58286470e-01 -8.02347884e-02
-2.82153320e-02 1.35589689e-01 -5.51276244e-02 -5.14558494e-01
-3.43378872e-01 -9.77243841e-01 -2.12648749e-01 3.40959996e-01
-1.49237379e-01 -2.16983765e-01 7.62311757e-01 -6.11518443e-01
9.87048686e-01 -2.04957294e+00 -5.24350286e-01 -1.48521224e-02
1.04666519e+00 4.14520204e-01 2.86693603e-01 6.72672391e-02
-3.18394631e-01 3.63636196e-01 -2.34808266e-01 -1.32969052e-01
-5.51719666e-01 2.53802892e-02 -1.75516367e-01 9.42155957e-01
2.06263795e-01 1.35546422e+00 -1.09886789e+00 -9.09229398e-01
3.51222873e-01 4.15292740e-01 -4.57138628e-01 -5.22413924e-02
1.57384172e-01 5.59251130e-01 -4.39276576e-01 6.97006345e-01
6.02105498e-01 -1.01498020e+00 6.92095309e-02 -1.40208572e-01
-3.50763276e-02 -9.10661519e-02 -2.02159867e-01 1.25387609e+00
-3.43005121e-01 1.02154744e+00 2.26234831e-02 -8.27760518e-01
2.85138860e-02 4.19888467e-01 8.15587521e-01 -5.44171333e-01
4.92988259e-01 3.87313277e-01 5.76821387e-01 -7.58140802e-01
9.69425887e-02 -8.55249047e-01 3.04959595e-01 4.48063225e-01
-2.61158109e-01 -2.72709548e-01 -4.88907307e-01 2.20536754e-01
9.94902194e-01 -4.19346392e-01 1.79893389e-01 -4.02626954e-02
-1.76940367e-01 4.30268079e-01 8.26008469e-02 9.99590576e-01
-5.50510049e-01 7.82887399e-01 4.39671040e-01 -5.17098188e-01
-1.10398328e+00 -1.28779542e+00 -5.98606884e-01 2.68249005e-01
-2.68717498e-01 -6.22914620e-02 -4.13860381e-01 -6.95335388e-01
6.53080493e-02 4.70279008e-01 -1.06050324e+00 -1.84027031e-01
-5.02172947e-01 -9.47308540e-01 7.81721950e-01 7.47391045e-01
1.00612007e-01 -8.69386911e-01 -1.21900177e+00 1.94589607e-03
-2.14302585e-01 -9.45569158e-01 1.00686483e-01 7.31125996e-02
-7.29568064e-01 -1.42554486e+00 -1.37034488e+00 -7.38646328e-01
3.04323018e-01 2.51744568e-01 1.05234611e+00 4.70475554e-01
-7.04665720e-01 5.66359043e-01 -1.33126825e-01 -8.15845132e-01
-7.90932059e-01 -1.52267545e-01 -2.41745532e-01 -6.59587502e-01
4.15992886e-01 -1.14334345e-01 -8.96296978e-01 -8.76605511e-03
-6.39552176e-01 4.32961136e-02 7.81223536e-01 8.96237671e-01
3.26204956e-01 -5.88780403e-01 3.79745096e-01 -1.01289308e+00
5.20917833e-01 -8.58631372e-01 -3.29414994e-01 7.45681748e-02
-9.13973689e-01 -3.86428088e-01 -1.00777999e-01 -2.50301182e-01
-8.32272410e-01 -5.11146903e-01 4.78902310e-02 -7.38727987e-01
7.82740712e-02 7.51346111e-01 1.07171607e+00 -1.03094883e-01
1.16641569e+00 -4.80598509e-02 1.60599038e-01 -9.60309282e-02
-1.08711377e-01 4.45150048e-01 4.77620751e-01 -3.03340435e-01
6.69225097e-01 9.45137739e-01 1.47428930e-01 -8.75692964e-01
-7.07774878e-01 -3.82111937e-01 4.15161140e-02 -3.32574785e-01
1.25734961e+00 -8.79413724e-01 -7.78907597e-01 2.49348387e-01
-1.11730111e+00 -3.22270989e-01 -9.91334915e-02 5.83989620e-01
-4.69893217e-01 4.02386725e-01 -7.77377129e-01 -5.60988843e-01
-5.84120750e-01 -1.55036092e+00 8.74752939e-01 -2.44246230e-01
-5.09556770e-01 -1.01757216e+00 3.49939197e-01 3.08857203e-01
5.66016912e-01 5.46718419e-01 1.31199837e+00 -3.36503744e-01
-4.08069164e-01 -3.09484839e-01 -6.69056118e-01 8.18849877e-02
2.26150081e-01 1.09037884e-01 -1.07307255e+00 -2.59112626e-01
-9.87909082e-03 -5.19242287e-01 9.56732094e-01 9.90897536e-01
1.26203096e+00 3.69757563e-01 -6.37536287e-01 4.91882503e-01
1.43294203e+00 2.46598408e-01 4.13871646e-01 3.11590612e-01
7.77158201e-01 2.88145483e-01 1.39607266e-01 2.66693503e-01
1.84830070e-01 4.27470505e-01 6.25950336e-01 -6.66346252e-01
-5.41458845e-01 -1.01410076e-01 -3.54225516e-01 3.46692443e-01
-3.03763121e-01 -2.08527803e-01 -1.60800636e+00 6.93954766e-01
-1.25267780e+00 -8.90971065e-01 -3.99867266e-01 1.94625521e+00
4.51804072e-01 3.78541082e-01 1.17595591e-01 -2.45951042e-01
3.98070008e-01 -9.14074332e-02 -5.39954603e-01 -2.95381099e-01
9.55002159e-02 3.57816190e-01 5.68802238e-01 1.98572472e-01
-9.33951974e-01 5.93371451e-01 7.33614826e+00 6.84501410e-01
-1.60550654e+00 3.82265061e-01 8.36470127e-01 -8.34889188e-02
-1.64366245e-01 -3.71042937e-01 -2.54217416e-01 3.36496204e-01
8.68529141e-01 -3.03628951e-01 -3.05930614e-01 7.98321009e-01
4.74413157e-01 -3.45602095e-01 -1.07969940e+00 1.03649354e+00
1.05906926e-01 -1.79182482e+00 -4.85646948e-02 2.59158462e-01
5.08792818e-01 4.31501925e-01 6.79695129e-01 2.52982885e-01
1.54881820e-01 -1.31092358e+00 4.30241168e-01 9.13702250e-02
1.34826398e+00 -2.69672871e-01 8.89028311e-01 -2.12824702e-01
-5.92005670e-01 1.93535715e-01 -7.53355846e-02 2.47758821e-01
2.70222127e-01 2.52571017e-01 -1.53234577e+00 2.42011905e-01
7.68148303e-01 6.25281990e-01 -4.36055303e-01 1.24895835e+00
1.49585446e-02 1.06595230e+00 -2.68098451e-02 1.14350036e-01
3.52051854e-01 4.38649356e-01 4.79233652e-01 1.36506379e+00
7.39835426e-02 1.64286926e-01 -3.13408941e-01 8.10255766e-01
4.60105799e-02 2.91509442e-02 -7.37062037e-01 -6.13980033e-02
-8.40551779e-02 9.24676657e-01 -8.13810647e-01 -4.51494873e-01
-5.19107699e-01 5.79839826e-01 1.22522332e-01 2.81580329e-01
-1.02272487e+00 4.00393188e-01 3.74419034e-01 4.28804308e-01
6.34664595e-02 -6.57691136e-02 -8.42469096e-01 -7.22319365e-01
-3.51777107e-01 -8.72243106e-01 6.85092211e-01 -1.04232657e+00
-1.30855870e+00 3.28000575e-01 5.72485663e-02 -1.36105573e+00
-8.68281573e-02 -9.57091570e-01 -4.70397651e-01 7.80715525e-01
-1.52506483e+00 -1.08754981e+00 -2.70637095e-01 3.73159289e-01
4.00848895e-01 1.51552185e-01 9.41968381e-01 1.31723955e-01
-1.21704072e-01 6.56818509e-01 -1.44218743e-01 2.62880176e-01
9.35275257e-01 -1.00034261e+00 7.94590861e-02 3.22914481e-01
-3.06031197e-01 5.58450937e-01 6.96772873e-01 -7.38640606e-01
-1.07294607e+00 -7.02325284e-01 3.74992460e-01 -9.44946170e-01
7.29600549e-01 1.11165695e-01 -8.35078895e-01 8.63169372e-01
2.89072901e-01 -7.05946982e-02 8.43779802e-01 -1.17344959e-02
-4.14967030e-01 3.27119917e-01 -1.09698856e+00 5.49308956e-01
4.82141852e-01 -4.25584435e-01 -6.78936601e-01 4.50119466e-01
5.00859320e-01 -5.61485052e-01 -6.91653848e-01 4.49281782e-01
9.20271754e-01 -8.06060135e-01 8.59674156e-01 -8.05125356e-01
1.04516888e+00 2.71850199e-01 1.81499064e-01 -1.12966681e+00
-3.26075315e-01 7.43384473e-03 4.23206300e-01 4.85060550e-02
4.35712814e-01 -3.98732364e-01 1.11166358e+00 2.74835944e-01
-2.35383660e-01 -8.44782591e-01 -1.00859475e+00 -4.54145342e-01
7.21435785e-01 -6.87832713e-01 -2.47693837e-01 1.31631577e+00
-9.36056301e-02 1.07373372e-01 -3.23758662e-01 1.76851973e-01
7.00826287e-01 -1.82137683e-01 5.05839050e-01 -9.51693773e-01
-3.35833490e-01 -5.38778245e-01 -3.81228060e-01 -2.85280436e-01
-3.75133425e-01 -8.85924160e-01 -3.79988581e-01 -1.62575185e+00
7.16437876e-01 -4.91082698e-01 -3.88193130e-01 3.69278640e-01
-1.40529126e-01 5.39347649e-01 3.55094112e-02 1.62909627e-01
6.54092804e-02 -2.67047491e-02 1.63743520e+00 -3.00878912e-01
2.88506031e-01 -2.49790296e-01 -6.14299119e-01 6.86337829e-01
7.37876773e-01 -6.92873478e-01 -2.55614579e-01 -4.70254719e-01
3.30124527e-01 2.45243177e-01 8.03148270e-01 -9.13929403e-01
-7.77268559e-02 -1.04756720e-01 5.93379319e-01 -6.17536366e-01
2.63753593e-01 -8.68336678e-01 -7.47820139e-02 1.18896723e+00
-1.12323940e-01 1.33183599e-01 6.91485584e-01 3.91207576e-01
1.50491983e-01 7.53908455e-02 9.40906525e-01 -3.67546469e-01
-4.57306743e-01 3.43006045e-01 -7.27141559e-01 3.80965531e-01
8.57652068e-01 -1.74892738e-01 -4.94894981e-01 -5.64210534e-01
-7.49871671e-01 1.16211787e-01 3.06024104e-01 2.36295357e-01
6.86899364e-01 -9.11426902e-01 -1.16687429e+00 -1.26000658e-01
4.00102645e-01 -2.46010140e-01 5.38233638e-01 1.39694774e+00
-8.39540422e-01 4.84412551e-01 -1.96251363e-01 -9.22476411e-01
-1.21323121e+00 7.67742574e-01 6.85609043e-01 -2.84249932e-01
-7.71606386e-01 7.75462270e-01 5.21426976e-01 3.83166403e-01
-3.93412337e-02 -3.38149011e-01 2.81745136e-01 -8.89692828e-02
3.56027722e-01 2.69984961e-01 3.08537960e-01 -2.59698421e-01
-5.11687458e-01 3.86789143e-01 -4.71668959e-01 -1.38469934e-01
1.13341737e+00 3.51035088e-01 3.86164129e-01 3.90650272e-01
1.14740932e+00 -1.31387234e-01 -7.82951415e-01 1.93721384e-01
-3.79267365e-01 -1.85802534e-01 1.36943564e-01 -1.04113626e+00
-8.65494728e-01 1.11656702e+00 1.08221197e+00 -2.05161989e-01
6.83298051e-01 2.54077911e-01 4.91364896e-01 -9.84915271e-02
2.02739760e-01 -5.25293112e-01 2.88425535e-01 5.39860353e-02
9.18643117e-01 -1.49022925e+00 3.21195066e-01 -2.19620124e-01
-6.10722244e-01 7.66802728e-01 4.86527979e-01 -1.92958698e-01
5.75732172e-01 4.10655230e-01 5.10988057e-01 -6.84858799e-01
-4.56512868e-01 1.45694211e-01 -9.67292339e-02 6.95249319e-01
5.34059107e-01 3.16740334e-01 -4.51966256e-01 1.56305194e-01
-6.28556907e-02 2.35246390e-01 5.12577176e-01 8.60463381e-01
-2.97612518e-01 -7.93043733e-01 -5.30975461e-01 7.34786332e-01
-6.73654139e-01 -2.40382105e-01 -1.25868917e-01 1.29766333e+00
1.05200030e-01 5.89480639e-01 -5.02320193e-02 -7.04472139e-02
1.03566058e-01 -1.29723996e-01 7.17783630e-01 -5.61266601e-01
-5.64724743e-01 1.68672502e-01 -1.89337879e-02 -2.88385898e-01
-2.02811763e-01 -5.33283412e-01 -1.08476746e+00 -2.99684107e-01
-1.19665265e-01 -1.94982201e-01 5.27050018e-01 8.26581538e-01
-2.54793875e-02 6.77608490e-01 1.02629244e-01 -3.75574887e-01
-6.11635447e-01 -8.06748152e-01 -3.08871925e-01 2.54554033e-01
7.38655686e-01 -7.78124452e-01 -2.78606087e-01 -6.31863624e-02] | [15.208579063415527, -1.9789507389068604] |
1bef2a2a-a4c7-4b86-9837-e4b3e1628a9e | an-experimental-study-in-real-time-facial | null | null | https://www.opastpublishers.com/peer-review/an-experimental-study-in-realtime-facial-emotion-recognition-on-new-3rl-dataset-5362.html | https://www.opastpublishers.com/open-access-articles/an-experimental-study-in-realtime-facial-emotion-recognition-on-new-3rl-dataset.pdf | An experimental study in Real-time Facial Emotion Recognition on new 3RL dataset | Although real-time facial emotion recognition is a hot topic research domain in the field of human-computer interaction, state-of- the-art available datasets still suffer from various problems, such as some unrelated photos such as document photos, unbalanced numbers of photos in each class, and misleading images that can negatively affect correct classification. The 3RL dataset was created, which contains approximately 24K images and will be publicly available, to overcome previously available dataset prob- lems. The 3RL dataset is labelled with five basic emotions: happiness, fear, sadness, disgust, and anger. Moreover, we compared the 3RL dataset with other famous state-of-the-art datasets (FER dataset, CK+ dataset), and we applied the most commonly used algorithms in previous works, SVM and CNN. The results show a noticeable improvement in generalization on the 3RL dataset. Experiments have shown an accuracy of up to 91.4% on 3RL dataset using CNN where results on FER2013, CK+ are, respectively (approximately from 60% to 85%). | ['Rahmeh Abou Zafra; Lana Ahmad Abdullah;Rouaa Alaraj; Rasha Albezreh;Tarek Barhoum; Khloud Al Jallad'] | 2023-04-02 | null | null | null | journal-of-current-trends-in-computer-science | ['facial-emotion-recognition'] | ['computer-vision'] | [-1.00501262e-01 -2.96721850e-02 -4.76676114e-02 -6.54924929e-01
-1.21408939e-01 -1.80969745e-01 4.01901633e-01 -3.14180180e-02
-5.10239840e-01 9.40232217e-01 -1.43631518e-01 3.65251571e-01
2.16201410e-01 -4.27709371e-01 -3.31994623e-01 -7.69311130e-01
-7.62464628e-02 -1.13286734e-01 -1.95474565e-01 -4.62505817e-01
2.27764159e-01 6.06356144e-01 -1.95585573e+00 5.52390933e-01
4.52060640e-01 1.81619775e+00 -7.70829737e-01 3.58600646e-01
1.32810652e-01 1.00494969e+00 -7.86690891e-01 -9.41486418e-01
6.91072047e-02 -9.52289328e-02 -7.31376469e-01 2.13884711e-01
3.43704909e-01 -8.52076933e-02 -1.65405408e-01 1.02554893e+00
6.16192818e-01 1.88813433e-02 6.86126769e-01 -2.08216333e+00
-8.17578375e-01 9.88363326e-02 -8.67334366e-01 -2.17517108e-01
4.20318455e-01 -2.16561824e-01 5.36332726e-01 -8.92196953e-01
6.45254493e-01 1.34751832e+00 7.14334726e-01 7.52083540e-01
-5.92905283e-01 -1.21928263e+00 -1.24091439e-01 5.59555411e-01
-1.59694874e+00 -4.28489119e-01 9.37156379e-01 -4.12960023e-01
7.97670901e-01 2.98685789e-01 6.36562824e-01 1.31125343e+00
1.13442600e-01 8.38492632e-01 1.65556216e+00 -2.86305577e-01
1.14219248e-01 6.09520733e-01 1.38204619e-01 4.95425075e-01
-2.86332071e-01 -1.36122093e-01 -5.66486418e-01 -8.89602005e-02
1.96701020e-01 -1.04041681e-01 -2.77027100e-01 -1.09372005e-01
-6.77127004e-01 6.89467549e-01 6.47063434e-01 3.39608043e-01
-2.88098365e-01 -4.66154754e-01 8.14426303e-01 4.48335260e-01
6.14429891e-01 7.11447746e-02 -4.99804646e-01 -1.93837002e-01
-6.31907821e-01 9.75750163e-02 6.26936913e-01 8.55169475e-01
8.14903319e-01 -3.86547074e-02 1.55621797e-01 1.03178978e+00
1.30843660e-02 3.48233193e-01 7.78050959e-01 -5.10573685e-01
4.22606617e-02 7.81524658e-01 9.32912230e-02 -1.56739295e+00
-4.59030509e-01 -5.67016043e-02 -1.26945066e+00 2.66611725e-01
5.27249984e-02 -1.57211393e-01 -8.37379754e-01 1.64703727e+00
1.45469636e-01 -3.90564254e-03 4.68502313e-01 9.90555108e-01
1.44922066e+00 6.12244487e-01 2.04235137e-01 -3.29211295e-01
1.00656533e+00 -1.25523484e+00 -1.00848126e+00 -3.77841592e-02
3.84412020e-01 -8.31551075e-01 9.88921821e-01 7.90502012e-01
-7.38093734e-01 -6.73198879e-01 -9.16445732e-01 1.49746746e-01
-7.99437344e-01 5.45890093e-01 7.97473729e-01 7.96681345e-01
-1.04431415e+00 5.09836257e-01 -5.78580461e-02 -7.09255815e-01
7.60271966e-01 3.60752583e-01 -1.08668804e+00 -3.59388366e-02
-1.13645732e+00 9.56602216e-01 3.37278664e-01 5.11540949e-01
-5.42457759e-01 -2.55483508e-01 -7.22274899e-01 -1.73349917e-01
1.35616645e-01 3.45472515e-01 9.72407222e-01 -1.82174373e+00
-1.44563270e+00 1.39167118e+00 1.37017906e-01 -1.29984394e-01
4.40791428e-01 -2.73288906e-01 -1.02024436e+00 1.19130380e-01
-2.82857001e-01 8.91402423e-01 7.72309065e-01 -1.08165658e+00
-2.35155046e-01 -5.21791935e-01 -1.23557031e-01 -3.15554366e-02
-5.98861217e-01 1.73359826e-01 -1.15810454e-01 -4.20370728e-01
-2.13147074e-01 -9.60572898e-01 1.99907005e-01 -3.40095386e-02
-3.74597073e-01 -3.23703021e-01 1.11978948e+00 -6.07265651e-01
9.40809250e-01 -2.30431342e+00 -1.98220104e-01 -8.52931812e-02
-1.05424359e-01 5.45954168e-01 2.07791533e-02 3.01029652e-01
-7.23869801e-01 1.72122419e-01 -3.45703028e-02 -1.99920490e-01
-1.10249102e-01 7.90027082e-02 -1.53246820e-02 4.37606394e-01
5.60121179e-01 6.94628596e-01 -4.90801096e-01 -5.95696926e-01
1.95899233e-01 5.74236155e-01 -8.03814530e-02 3.03116947e-01
3.61331552e-01 9.16489214e-02 -1.19890466e-01 1.13096344e+00
1.00799823e+00 1.57366663e-01 -1.42404959e-01 -5.93517125e-01
1.67286143e-01 -6.79798722e-01 -9.58132625e-01 1.16333640e+00
-1.67405397e-01 7.95460820e-01 -1.45750299e-01 -1.14397967e+00
1.35638845e+00 4.80599672e-01 3.63024861e-01 -6.80357516e-01
7.10766673e-01 -3.97782959e-02 -2.06906289e-01 -9.75654483e-01
5.90247869e-01 -1.64886296e-01 -9.59966108e-02 -2.58679111e-02
2.62254894e-01 2.97984689e-01 8.86587799e-02 -4.00175937e-02
6.61205709e-01 -4.98572700e-02 3.60883892e-01 -3.56436744e-02
6.68702245e-01 -7.00962394e-02 5.45243263e-01 4.69901189e-02
-8.54971826e-01 5.33917844e-01 7.04968572e-01 -6.86577499e-01
-7.23084152e-01 -5.49082100e-01 -1.87491953e-01 8.98559034e-01
6.34049699e-02 -2.43338495e-01 -7.34574020e-01 -7.80466020e-01
-1.67497501e-01 3.02516788e-01 -9.96847928e-01 -3.03784639e-01
-1.55526362e-02 -9.45761621e-01 7.11169302e-01 4.40659136e-01
1.11210620e+00 -1.48711300e+00 -5.62724829e-01 -1.99737653e-01
-2.42179602e-01 -1.39115226e+00 3.76885951e-01 8.73327069e-03
-4.13228154e-01 -1.28403807e+00 -8.29539657e-01 -7.08716571e-01
8.24762881e-01 -1.11922383e-01 1.17124629e+00 1.13112018e-01
-6.37204826e-01 4.95566353e-02 -6.33794785e-01 -5.71426094e-01
1.18367285e-01 -3.45754743e-01 -2.78425869e-02 4.01393533e-01
7.36587763e-01 -2.41626725e-01 -3.45470965e-01 4.48752582e-01
-7.08952069e-01 -1.07688025e-01 5.74662447e-01 8.92645478e-01
1.30187958e-01 1.34187296e-01 8.55770111e-01 -8.07715595e-01
3.81321400e-01 -3.87855321e-01 -1.32978171e-01 2.42993340e-01
-3.15212190e-01 -6.44153535e-01 4.29086149e-01 -6.24771059e-01
-1.13342094e+00 4.00181301e-02 -2.31648311e-01 -5.14857411e-01
-6.31184459e-01 1.83202520e-01 -1.89806476e-01 -3.23167562e-01
5.24316669e-01 -2.80080047e-02 -5.35231158e-02 -1.09709717e-01
-2.37392336e-02 1.21282542e+00 4.89943713e-01 -1.60940856e-01
2.23115116e-01 2.46984854e-01 -1.23123921e-01 -8.39217365e-01
-8.81701767e-01 -3.64750654e-01 -5.04694521e-01 -6.63596809e-01
6.79525912e-01 -1.13201487e+00 -9.70627606e-01 1.13810492e+00
-8.17611933e-01 1.32094547e-01 1.56035587e-01 1.92346662e-01
-2.87089586e-01 5.12073264e-02 -5.86631477e-01 -1.13514280e+00
-4.28734988e-01 -1.06660509e+00 9.57629979e-01 6.20558918e-01
-2.09562853e-01 -6.35196209e-01 -4.16722298e-01 4.89288688e-01
4.04599816e-01 8.83454204e-01 7.13431120e-01 -5.42373002e-01
2.59845078e-01 -5.54002583e-01 -5.07696271e-01 7.38679945e-01
-7.82087296e-02 4.49813515e-01 -1.44522679e+00 -1.35039493e-01
-1.09763540e-01 -1.22864509e+00 6.53734744e-01 -1.80978611e-01
1.39789295e+00 -2.09673390e-01 -1.61769897e-01 1.54154301e-01
1.39045334e+00 4.23832089e-01 1.10380387e+00 2.73232549e-01
2.42430016e-01 7.42322683e-01 9.76151288e-01 5.72300792e-01
2.66995341e-01 4.09312427e-01 6.35942578e-01 -4.15305555e-01
3.29095006e-01 1.05748057e-01 2.05858797e-01 5.38519323e-01
-1.46131217e-01 -2.42651731e-01 -7.13219345e-01 3.75714898e-01
-1.81407964e+00 -9.31781948e-01 -1.66663170e-01 1.82925093e+00
7.13025630e-01 -9.83121172e-02 1.86840698e-01 6.39000297e-01
7.65600383e-01 8.84972140e-02 -3.40688974e-01 -7.56844044e-01
-4.98056829e-01 2.68623829e-01 1.29063204e-01 -1.55303299e-01
-1.38812590e+00 1.12326479e+00 6.06272984e+00 7.87953794e-01
-1.72351849e+00 -1.55877188e-01 1.29939115e+00 -7.19829947e-02
6.29577875e-01 -5.89886546e-01 -6.33590460e-01 3.54984522e-01
7.38736868e-01 4.18999828e-02 1.42769560e-01 1.28785074e+00
-2.68329471e-01 -3.70647758e-01 -8.09280038e-01 1.68273199e+00
6.45905793e-01 -7.25298524e-01 -1.32556051e-01 -4.57272768e-01
7.29775012e-01 -4.12655652e-01 -5.15641039e-03 7.50043631e-01
-2.25812659e-01 -1.54623067e+00 5.24117589e-01 4.75800335e-01
7.99056232e-01 -1.13565361e+00 1.30303192e+00 1.02472678e-01
-7.59481847e-01 -6.04654066e-02 -5.87844193e-01 -2.74767250e-01
-3.25370312e-01 6.11408174e-01 -4.19651210e-01 5.29719114e-01
1.41940820e+00 1.06783128e+00 -8.51587594e-01 7.99389005e-01
1.09553523e-02 2.20048949e-01 -3.79878096e-02 -4.08830822e-01
7.43150264e-02 1.05719887e-01 -1.79065630e-01 1.17191267e+00
7.28218257e-02 2.44780660e-01 -1.26983300e-01 3.48513395e-01
-2.03159630e-01 4.68635321e-01 -5.29056013e-01 -1.57550320e-01
-1.54584765e-01 1.88543534e+00 -5.22152483e-01 -3.17371041e-01
-2.40298480e-01 1.09695160e+00 3.22597861e-01 3.10541987e-01
-9.61991668e-01 -5.76878607e-01 5.33982038e-01 -4.30671483e-01
-7.02666165e-03 3.73402566e-01 1.23693824e-01 -1.07446027e+00
1.19011410e-01 -1.04699862e+00 3.31518024e-01 -1.35035300e+00
-1.43603516e+00 1.08586943e+00 -2.63879240e-01 -1.05285180e+00
1.40727028e-01 -9.75777864e-01 -3.70168775e-01 5.23014128e-01
-1.46307313e+00 -1.17751253e+00 -8.81180644e-01 7.87247360e-01
2.63222814e-01 -3.93231601e-01 1.00733197e+00 5.32078147e-01
-8.33590567e-01 6.49520338e-01 -2.36928135e-01 3.57454002e-01
1.05316758e+00 -8.39903951e-01 -4.00349021e-01 1.31329089e-01
-2.97355205e-01 1.35695949e-01 4.56145406e-01 -1.69139355e-01
-1.14464152e+00 -9.31955338e-01 7.65982687e-01 -9.31000561e-02
2.92042583e-01 -3.17811102e-01 -6.56968772e-01 5.37198007e-01
4.60794449e-01 3.28180820e-01 9.20428813e-01 -2.63169557e-02
-4.00076717e-01 -2.33721063e-01 -1.67632115e+00 5.05434096e-01
6.41769052e-01 -2.05788106e-01 -2.26192355e-01 1.41189262e-01
5.03915884e-02 -4.05338019e-01 -8.10764253e-01 7.59527624e-01
7.76782990e-01 -1.50985444e+00 5.81029654e-01 -7.56806493e-01
8.02129626e-01 3.22692394e-02 -3.43697548e-01 -1.37281132e+00
4.05566767e-03 -1.47095859e-01 1.75999194e-01 1.60839999e+00
7.83784762e-02 -3.83551478e-01 7.56971359e-01 6.47799015e-01
3.26395661e-01 -1.12084627e+00 -6.97771311e-01 -5.20419359e-01
-2.33470261e-01 -2.71719575e-01 4.75476235e-01 1.25603354e+00
7.53349587e-02 3.53149474e-01 -6.14187121e-01 -2.76425064e-01
1.78837687e-01 -4.47232276e-02 9.61266458e-01 -1.25552416e+00
5.11193275e-01 -3.20517540e-01 -9.10142601e-01 -1.13695137e-01
3.62891704e-01 -4.97959197e-01 -2.17373997e-01 -1.11476350e+00
2.51240432e-01 -2.34143540e-01 -4.54583883e-01 9.06809330e-01
9.77613032e-02 8.32307935e-01 1.56792954e-01 -3.61845702e-01
-6.23216331e-01 7.38621414e-01 1.07748556e+00 -6.42266870e-02
2.04563469e-01 -4.70515400e-01 -7.17384398e-01 9.80868697e-01
7.32287169e-01 -1.31700054e-01 -9.82352793e-02 1.38043523e-01
1.10400707e-01 -9.71086323e-02 4.30531412e-01 -1.21487582e+00
-1.93091720e-01 -7.31310099e-02 9.82675791e-01 -5.84820628e-01
7.33034611e-01 -9.86963987e-01 2.62047499e-01 8.75556543e-02
-3.12559843e-01 1.17390908e-01 4.07271922e-01 2.87744515e-02
-7.71536469e-01 -1.42965326e-02 1.16461337e+00 -5.92616089e-02
-9.81571972e-01 2.62629300e-01 1.86539069e-02 2.69125812e-02
1.46256125e+00 -1.76626578e-01 -4.05505806e-01 -6.46510780e-01
-7.48557925e-01 -5.87375797e-02 2.13139400e-01 7.46977508e-01
6.51430428e-01 -1.58388293e+00 -4.95148331e-01 1.54587388e-01
4.12290275e-01 -4.90862846e-01 4.56075579e-01 8.43650937e-01
-3.61845404e-01 1.24936543e-01 -9.19008255e-01 -4.49294358e-01
-1.72269976e+00 5.46249628e-01 2.31382713e-01 1.36106208e-01
-7.59764016e-02 9.13593471e-01 -7.07365349e-02 -3.79256159e-01
3.53276074e-01 9.44857299e-02 -4.64397281e-01 3.55161279e-01
5.90270996e-01 3.53380919e-01 1.26382813e-01 -1.13330567e+00
-6.22575343e-01 6.87456369e-01 -3.11703030e-02 3.53342980e-01
1.27409554e+00 1.14333786e-01 -2.42653176e-01 5.04312515e-01
1.46811175e+00 -3.97523493e-01 -6.46030486e-01 9.93527621e-02
-2.76795715e-01 -5.50759494e-01 -2.47922223e-02 -1.21849835e+00
-1.45870829e+00 1.03539717e+00 1.08691180e+00 7.32145756e-02
1.60853446e+00 -4.00168896e-01 7.19900191e-01 4.58389878e-01
2.74168730e-01 -1.19614089e+00 2.64854282e-01 3.49380076e-01
1.14894783e+00 -1.68197620e+00 -6.36332408e-02 -4.05075490e-01
-1.09356201e+00 1.24405873e+00 1.07377148e+00 -6.17759079e-02
7.26460278e-01 4.65574525e-02 4.51485306e-01 1.36908388e-03
-6.72919631e-01 8.92989784e-02 1.03703007e-01 5.06684244e-01
5.73227823e-01 -8.16769004e-02 -2.17791304e-01 9.35352623e-01
-4.06136841e-01 4.57226485e-01 4.20236945e-01 8.35120082e-01
-9.98219326e-02 -6.94478512e-01 -3.98852497e-01 2.48401940e-01
-8.26967537e-01 2.26571977e-01 -8.02064240e-01 1.14526463e+00
4.10795867e-01 1.04615724e+00 4.24418226e-02 -5.72448194e-01
3.62407655e-01 1.45495489e-01 3.07668746e-01 1.27078533e-01
-6.52388632e-01 -2.60609955e-01 1.87849879e-01 -5.07409334e-01
-7.34363139e-01 -2.11539805e-01 -9.05511439e-01 -3.83869737e-01
-2.19981492e-01 -1.58825517e-02 6.48671806e-01 7.19714880e-01
1.90389618e-01 1.96795523e-01 7.23308682e-01 -9.23181832e-01
2.15155631e-02 -1.08165503e+00 -6.55267596e-01 9.15693402e-01
5.89144975e-02 -8.87212634e-01 -4.11177099e-01 2.32511222e-01] | [13.612936973571777, 1.8990765810012817] |
03843849-3147-4ceb-a7ea-9b7cea8fa66a | differentiable-multi-target-causal-bayesian | 2302.10607 | null | https://arxiv.org/abs/2302.10607v2 | https://arxiv.org/pdf/2302.10607v2.pdf | Differentiable Multi-Target Causal Bayesian Experimental Design | We introduce a gradient-based approach for the problem of Bayesian optimal experimental design to learn causal models in a batch setting -- a critical component for causal discovery from finite data where interventions can be costly or risky. Existing methods rely on greedy approximations to construct a batch of experiments while using black-box methods to optimize over a single target-state pair to intervene with. In this work, we completely dispose of the black-box optimization techniques and greedy heuristics and instead propose a conceptually simple end-to-end gradient-based optimization procedure to acquire a set of optimal intervention target-state pairs. Such a procedure enables parameterization of the design space to efficiently optimize over a batch of multi-target-state interventions, a setting which has hitherto not been explored due to its complexity. We demonstrate that our proposed method outperforms baselines and existing acquisition strategies in both single-target and multi-target settings across a number of synthetic datasets. | ['Stefan Bauer', 'Adam Foster', 'Yarin Gal', 'Andrew Jesson', 'Desi R. Ivanova', 'Panagiotis Tigas', 'Yashas Annadani'] | 2023-02-21 | null | null | null | null | ['causal-discovery', 'experimental-design'] | ['knowledge-base', 'methodology'] | [ 5.10840595e-01 1.18804149e-01 -6.16274655e-01 -4.54257697e-01
-9.98750687e-01 -4.57917064e-01 6.62606776e-01 1.81270853e-01
-5.70082188e-01 9.73428786e-01 1.46124318e-01 -9.20273244e-01
-5.17579973e-01 -4.09511745e-01 -1.10346806e+00 -5.05380392e-01
-4.78628904e-01 6.51615500e-01 -9.17432383e-02 2.19200253e-01
5.01318336e-01 1.96555853e-01 -1.16989589e+00 -6.92798868e-02
6.47138774e-01 3.71933013e-01 4.95535471e-02 8.67492199e-01
4.89266872e-01 4.27944243e-01 -7.68947452e-02 -1.37351215e-01
3.45882356e-01 -8.32763672e-01 -6.69636965e-01 -9.42951590e-02
1.89905211e-01 -5.11476457e-01 -1.70305178e-01 7.94102848e-01
8.49989653e-01 1.04123041e-01 6.47745073e-01 -1.09548461e+00
-2.99859289e-02 8.99784923e-01 -6.46554530e-01 3.17601055e-01
2.49603838e-01 6.18832111e-01 1.26523077e+00 -5.44381320e-01
5.24088919e-01 1.63101757e+00 3.92422289e-01 2.33154222e-01
-1.84342563e+00 -5.59961200e-01 3.23857576e-01 -6.23434521e-02
-9.30155277e-01 -5.12917221e-01 6.29592121e-01 -5.84489286e-01
6.58447444e-01 2.68392917e-02 6.34886086e-01 1.26914859e+00
2.19461411e-01 6.56747878e-01 1.42490184e+00 -4.96048003e-01
7.32628942e-01 -4.17348325e-01 1.00365408e-01 8.65078092e-01
3.21559161e-01 7.63906777e-01 -6.61454797e-01 -5.36512673e-01
7.77213991e-01 -4.64057326e-02 -1.36394441e-01 -5.13588011e-01
-1.30304551e+00 1.08259809e+00 1.46588609e-01 -2.18652502e-01
-6.55191183e-01 6.08498931e-01 3.68767709e-01 2.84746647e-01
2.52271652e-01 7.64360905e-01 -7.14907587e-01 -2.32101306e-01
-8.62282932e-01 6.96763277e-01 7.78645873e-01 5.98971546e-01
5.64718544e-01 -2.20274165e-01 -7.15122581e-01 3.25966358e-01
2.25471839e-01 4.66974080e-01 -1.42389670e-01 -8.36852610e-01
4.68050927e-01 1.91850156e-01 9.26758111e-01 -7.34280586e-01
-4.81007040e-01 -2.05802679e-01 -4.89758790e-01 -2.17898823e-02
8.27156186e-01 -6.26474082e-01 -8.87705207e-01 1.97713673e+00
8.11229467e-01 4.15720195e-01 -4.42002058e-01 8.41088474e-01
-3.88808809e-02 3.29555422e-01 2.05569997e-01 -6.84452415e-01
1.15076542e+00 -6.87873304e-01 -6.15775108e-01 -3.97644609e-01
6.04983807e-01 -4.59352732e-01 1.43168676e+00 3.34909797e-01
-1.05406916e+00 -9.25044250e-03 -8.61185789e-01 4.05622482e-01
-3.02097481e-02 8.50600377e-02 8.79405022e-01 7.37003505e-01
-6.08565748e-01 9.89161015e-01 -8.21438313e-01 -2.97697425e-01
6.57099366e-01 4.55187738e-01 1.42027512e-01 -1.10031411e-01
-9.31187332e-01 6.60418630e-01 3.99359912e-01 -1.84181426e-02
-1.78840613e+00 -1.19320869e+00 -4.56048846e-01 1.17056884e-01
1.16315603e+00 -9.31859672e-01 1.40877533e+00 -3.88987273e-01
-1.76703012e+00 2.79725045e-01 -2.41475597e-01 -5.29177785e-01
3.98675978e-01 -4.47882980e-01 3.26733947e-01 -1.56280935e-01
-3.56953740e-02 3.92832339e-01 9.26886916e-01 -1.02212512e+00
-4.35977668e-01 -2.82962292e-01 1.65444300e-01 -5.20974509e-02
6.67284206e-02 3.23962301e-01 -1.11289583e-02 -4.77528930e-01
-2.56123155e-01 -1.03128254e+00 -7.99348652e-01 -4.24853176e-01
-8.08671653e-01 2.30770465e-02 1.72140867e-01 -2.79618919e-01
1.38696933e+00 -1.50506723e+00 2.71475047e-01 1.59942463e-01
1.28731251e-01 -6.85539991e-02 -1.29890874e-01 6.26160085e-01
-2.49377787e-01 2.79860020e-01 -1.85516253e-01 -2.29653716e-01
1.76290065e-01 -1.92212015e-01 -2.60293603e-01 6.90637648e-01
1.04121715e-01 1.03969109e+00 -1.20322788e+00 -4.35255438e-01
1.15103491e-01 -1.92885846e-01 -1.03491902e+00 5.69635749e-01
-7.40004539e-01 6.95192516e-01 -7.49803960e-01 4.28519994e-01
2.91494876e-01 -2.48041809e-01 5.16908228e-01 1.20717272e-01
-1.52585149e-01 3.75023633e-01 -1.25849986e+00 1.46969056e+00
-3.31492722e-01 3.00672092e-02 -3.90662737e-02 -1.49298573e+00
3.70613694e-01 2.36739412e-01 6.23046219e-01 -1.30788550e-01
3.77220422e-01 2.08155867e-02 9.04995948e-02 -5.13530850e-01
-1.34700343e-01 -3.54721397e-01 -3.64559561e-01 6.25750422e-01
-6.48154691e-02 -7.80181959e-02 1.57329097e-01 3.59748900e-02
1.54431331e+00 1.48636103e-01 7.06518710e-01 -4.59873945e-01
-4.10026349e-02 -1.55304568e-02 5.97616553e-01 1.43270457e+00
-8.45347494e-02 8.03440064e-02 1.01348686e+00 -4.20983106e-01
-1.14927459e+00 -8.28638852e-01 1.67076856e-01 1.07629192e+00
-3.94703120e-01 -3.90130132e-01 -6.69406056e-01 -7.74887323e-01
1.54985175e-01 8.53136897e-01 -8.36022437e-01 -1.07352711e-01
-5.08533478e-01 -1.05798423e+00 1.70215085e-01 1.11183286e-01
1.06989458e-01 -7.78074324e-01 -8.37029159e-01 3.85096043e-01
2.08167434e-01 -7.06966400e-01 -6.43195808e-01 4.41297293e-01
-8.97723615e-01 -1.23947835e+00 -3.54281932e-01 -3.00648976e-02
6.05035126e-01 -7.71419927e-02 1.15406895e+00 -3.01022232e-01
-4.37456191e-01 2.85543680e-01 1.42467171e-01 -5.70533633e-01
-4.19466615e-01 -2.77262330e-01 6.67467341e-02 -6.08246699e-02
1.46278024e-01 -6.07602358e-01 -9.85354781e-01 2.12649494e-01
-4.86560404e-01 1.29065802e-02 6.28342867e-01 1.27596533e+00
3.78272265e-01 -1.62679270e-01 6.75561011e-01 -1.11842871e+00
7.97765613e-01 -5.49931049e-01 -1.31953931e+00 2.78401166e-01
-8.46494257e-01 4.97554004e-01 5.44114172e-01 -6.89577520e-01
-9.38225865e-01 2.02428967e-01 5.07301450e-01 -2.44373739e-01
-6.86081946e-02 7.83824027e-01 -1.48246363e-01 2.76024491e-01
7.79546797e-01 -3.68677348e-01 -2.50324584e-03 -3.86629909e-01
6.25171304e-01 3.41668040e-01 7.05997720e-02 -1.10983825e+00
4.32604522e-01 8.52090418e-02 3.74781191e-01 -3.18625391e-01
-1.02169943e+00 -2.30364770e-01 -3.09056371e-01 -1.40024439e-01
6.09815001e-01 -7.32528389e-01 -1.23631406e+00 2.15854749e-01
-8.94641578e-01 -9.51358974e-01 -2.05359399e-01 6.95833206e-01
-7.34574378e-01 7.07654878e-02 -3.23805839e-01 -8.61099780e-01
-1.44164423e-02 -1.20765710e+00 1.19258130e+00 1.06106125e-01
-3.27604353e-01 -8.37948382e-01 4.07382041e-01 8.06092173e-02
6.85624927e-02 3.09324473e-01 1.10272348e+00 -4.75131720e-01
-5.94476104e-01 -1.48532674e-01 9.96007770e-02 -2.40003973e-01
9.47626494e-03 -7.14728460e-02 -5.94002604e-01 -4.87363160e-01
-1.70171887e-01 -5.12482345e-01 6.79253221e-01 1.10177481e+00
1.21232235e+00 -5.47203422e-01 -5.90846539e-01 2.52409071e-01
1.24020672e+00 2.13068470e-01 2.96297252e-01 1.15683652e-01
3.16838235e-01 3.57686341e-01 9.86211836e-01 8.58803451e-01
1.62860498e-01 7.58885562e-01 2.96197295e-01 -9.74571630e-02
4.86916631e-01 -6.07123017e-01 2.31600419e-01 -1.73306376e-01
2.42465347e-01 -2.45468214e-01 -8.27495456e-01 7.01457381e-01
-2.08084369e+00 -8.52723420e-01 2.20008254e-01 2.66745973e+00
1.31302202e+00 2.16322735e-01 5.90471089e-01 -3.23543698e-01
6.26744688e-01 -8.38671252e-02 -7.87934780e-01 -2.16604427e-01
5.07865846e-01 2.37388417e-01 7.29878545e-01 6.41066253e-01
-1.08944595e+00 9.66835439e-01 7.50624943e+00 7.35359251e-01
-9.33872342e-01 6.72915429e-02 8.40857387e-01 -3.35339844e-01
-2.14764282e-01 5.95721900e-01 -9.80665863e-01 4.10956740e-01
1.37389088e+00 -3.09462339e-01 6.88766956e-01 5.35523176e-01
9.92258668e-01 -4.71834868e-01 -1.64822483e+00 5.41681230e-01
-7.38089144e-01 -1.28576803e+00 -4.38168108e-01 1.51320785e-01
7.34466374e-01 -3.63809586e-01 -4.30272594e-02 1.48810238e-01
1.05085111e+00 -1.13495421e+00 5.46545267e-01 2.71623820e-01
4.81887728e-01 -5.28088450e-01 1.37158811e-01 3.69662672e-01
-4.76953089e-01 -4.25797433e-01 -2.14112923e-01 -2.01574102e-01
3.00143301e-01 1.08211935e+00 -1.16833842e+00 1.96558341e-01
4.26938474e-01 2.66825825e-01 -1.63364828e-01 9.81349111e-01
-3.59979749e-01 1.24434125e+00 -5.36696196e-01 -2.59279341e-01
1.89053938e-01 -1.64853305e-01 6.03470981e-01 1.00085258e+00
1.56393498e-01 1.63533792e-01 3.80168229e-01 1.08547640e+00
3.15253101e-02 -6.03126362e-02 -6.65895820e-01 -4.24165308e-01
7.14266896e-01 8.36722314e-01 -6.91245019e-01 -1.88729018e-01
-1.31832227e-01 3.52965027e-01 4.04928386e-01 4.43063945e-01
-9.95304763e-01 -3.51634659e-02 4.22101974e-01 -4.00851667e-02
4.09177482e-01 -2.26380631e-01 -4.52644855e-01 -7.92004883e-01
-3.74584168e-01 -1.21850598e+00 5.72569966e-01 -2.87139803e-01
-8.88140619e-01 -3.46043825e-01 7.04290509e-01 -6.56658351e-01
-3.80574822e-01 -3.36318254e-01 -6.76460803e-01 7.12282419e-01
-9.05460238e-01 -7.07944870e-01 4.02986705e-01 3.54889244e-01
3.69840145e-01 1.64186195e-01 4.75819290e-01 -3.82651202e-02
-9.32094216e-01 5.02979338e-01 1.00965030e-01 -4.83826488e-01
6.46666944e-01 -1.34924889e+00 1.14944957e-01 9.21401501e-01
-1.75938234e-01 8.38349164e-01 1.27440500e+00 -8.85336995e-01
-1.53658307e+00 -8.08105469e-01 4.51437294e-01 -1.60776049e-01
7.81192183e-01 -5.39291143e-01 -3.85494053e-01 7.32828856e-01
1.33070603e-01 -2.33049050e-01 4.26560253e-01 6.60129905e-01
4.36102040e-02 -8.57074410e-02 -9.22060609e-01 1.14769387e+00
1.07152200e+00 -1.03117250e-01 -2.06126660e-01 6.83021247e-01
7.52460599e-01 -3.21527451e-01 -8.11362565e-01 3.22170049e-01
3.86570901e-01 -6.25959694e-01 9.80593562e-01 -1.14547122e+00
5.07672608e-01 -2.17104986e-01 2.51627594e-01 -1.57838047e+00
-1.10552020e-01 -1.45443392e+00 1.10865189e-02 9.58857536e-01
4.91600037e-01 -5.48213065e-01 5.88164270e-01 5.60467720e-01
1.75655380e-01 -8.87684822e-01 -8.58752966e-01 -5.17296672e-01
5.18424320e-04 -3.26849133e-01 4.59286481e-01 5.33716023e-01
-4.94956523e-02 5.69596767e-01 -6.88514829e-01 2.40612790e-01
1.05008113e+00 1.87007964e-01 1.05054796e+00 -7.53581107e-01
-7.62852371e-01 -2.80183434e-01 2.72664130e-01 -1.05296361e+00
8.89516100e-02 -2.96423227e-01 4.26506788e-01 -9.35389042e-01
4.87190962e-01 -4.21749800e-01 -2.08225310e-01 5.53872526e-01
-7.27314234e-01 -6.78473175e-01 -2.42378518e-01 -2.31421709e-01
-4.96945441e-01 5.69113195e-01 1.05359960e+00 5.40429056e-02
-5.23367822e-01 1.28410235e-01 -9.07464921e-01 3.94470930e-01
6.20182037e-01 -9.14633632e-01 -7.84729302e-01 1.58420756e-01
2.24381089e-01 5.56856871e-01 5.25188506e-01 -3.34331363e-01
-7.59738609e-02 -9.19183552e-01 -6.33942545e-04 -4.01224256e-01
-1.63796544e-01 -3.58259022e-01 2.29569167e-01 6.00044966e-01
-7.97195375e-01 -1.55452490e-01 1.45306274e-01 9.67135847e-01
4.24141198e-01 -2.09125176e-01 7.80254245e-01 -1.44724175e-01
-9.24865156e-02 2.54552364e-01 -4.06151980e-01 3.11135292e-01
1.00388074e+00 3.84406894e-01 -2.26675108e-01 -3.47284138e-01
-5.99538326e-01 4.82672095e-01 1.85055360e-01 8.13697726e-02
3.62333179e-01 -9.71149564e-01 -6.62228763e-01 -3.06325227e-01
-2.65856683e-01 -1.81784749e-01 7.35976771e-02 1.04775405e+00
-9.14962366e-02 5.37825942e-01 1.96197495e-01 -5.96741617e-01
-1.05448973e+00 9.34655488e-01 3.36181849e-01 -8.26234818e-01
-4.00479585e-01 7.58864462e-01 3.20119023e-01 -2.70959973e-01
1.16909549e-01 -2.79786885e-01 3.32031459e-01 -2.78681278e-01
3.59897554e-01 3.10639292e-01 -3.57371658e-01 3.36991489e-01
-1.61722854e-01 1.84656873e-01 -7.60750324e-02 -4.13901329e-01
1.39807034e+00 6.29004370e-03 -3.68238203e-02 4.44790274e-01
9.45759535e-01 -2.29074210e-01 -1.50648892e+00 -1.16508603e-01
2.50205874e-01 -6.82587564e-01 3.24338317e-01 -9.10631895e-01
-4.62490469e-01 4.81866717e-01 5.11098325e-01 -1.06872611e-01
8.69992554e-01 -1.51087373e-01 1.38790771e-01 5.14143944e-01
3.05914909e-01 -1.02541161e+00 2.90436178e-01 2.01632399e-02
7.84254551e-01 -1.24322832e+00 3.28906983e-01 -2.40515366e-01
-2.75511682e-01 5.97455382e-01 2.89855301e-01 9.04180575e-03
7.10219681e-01 2.29866534e-01 -5.66969573e-01 -3.90224129e-01
-1.05712318e+00 -2.28358246e-02 3.54974670e-03 3.75506371e-01
3.47572416e-01 2.15862289e-01 -5.82718074e-01 3.99021238e-01
1.89601839e-01 3.90924275e-01 4.34420347e-01 1.20815802e+00
-2.93929368e-01 -1.28078461e+00 -4.51763541e-01 7.37022758e-01
-5.74870884e-01 -1.64087996e-01 -1.40695021e-01 7.61656940e-01
-4.25216526e-01 1.16086495e+00 -4.06848133e-01 -1.30036294e-01
3.57299656e-01 8.17337185e-02 6.90008998e-01 -4.66321051e-01
-4.36254054e-01 3.38115513e-01 3.19563687e-01 -8.23772967e-01
-3.56238961e-01 -9.23774183e-01 -7.11803973e-01 -2.97948718e-01
-4.96203274e-01 1.12051897e-01 3.72286379e-01 1.04161572e+00
4.63291377e-01 4.97451931e-01 8.80475402e-01 -8.53853047e-01
-1.18883431e+00 -9.10881162e-01 -4.41250861e-01 1.27911597e-01
3.15626383e-01 -1.02737749e+00 -2.74660438e-01 -5.48654459e-02] | [7.7291035652160645, 5.227616786956787] |
7e8811a4-74d5-4a3b-80d6-7e346543ae96 | low-confidence-samples-mining-for-semi | 2306.16201 | null | https://arxiv.org/abs/2306.16201v1 | https://arxiv.org/pdf/2306.16201v1.pdf | Low-Confidence Samples Mining for Semi-supervised Object Detection | Reliable pseudo-labels from unlabeled data play a key role in semi-supervised object detection (SSOD). However, the state-of-the-art SSOD methods all rely on pseudo-labels with high confidence, which ignore valuable pseudo-labels with lower confidence. Additionally, the insufficient excavation for unlabeled data results in an excessively low recall rate thus hurting the network training. In this paper, we propose a novel Low-confidence Samples Mining (LSM) method to utilize low-confidence pseudo-labels efficiently. Specifically, we develop an additional pseudo information mining (PIM) branch on account of low-resolution feature maps to extract reliable large-area instances, the IoUs of which are higher than small-area ones. Owing to the complementary predictions between PIM and the main branch, we further design self-distillation (SD) to compensate for both in a mutually-learning manner. Meanwhile, the extensibility of the above approaches enables our LSM to apply to Faster-RCNN and Deformable-DETR respectively. On the MS-COCO benchmark, our method achieves 3.54% mAP improvement over state-of-the-art methods under 5% labeling ratios. | ['Bin Wang', 'Tianxiang Pan', 'Fangyuan Zhang', 'Guandu Liu'] | 2023-06-28 | null | null | null | null | ['semi-supervised-object-detection'] | ['computer-vision'] | [ 1.71026379e-01 4.32930619e-01 -4.61148024e-01 -4.47865307e-01
-9.58714724e-01 -1.98918834e-01 3.48615915e-01 -1.03903213e-03
-4.69827026e-01 9.76180851e-01 -3.15670848e-01 -2.07797438e-01
-7.26480931e-02 -8.44847143e-01 -8.38898838e-01 -9.12157297e-01
1.84829667e-01 3.99897397e-01 7.33552158e-01 1.45912692e-01
5.17068096e-02 4.07437921e-01 -1.76017916e+00 2.28614807e-01
1.10096002e+00 1.26305413e+00 5.21361172e-01 -3.61963324e-02
-3.10989112e-01 6.65400922e-01 -3.03754359e-01 -1.00584134e-01
2.96451509e-01 -5.47266714e-02 -5.93435407e-01 5.52022569e-02
2.29462519e-01 -3.29913229e-01 -1.14911579e-01 1.10454047e+00
2.96184599e-01 -2.41162226e-01 9.09370601e-01 -1.08185935e+00
-4.59052801e-01 6.58924103e-01 -9.15690958e-01 -5.95214358e-03
-4.11594003e-01 -1.87586352e-01 9.27160680e-01 -1.29771423e+00
3.87747079e-01 9.64842081e-01 8.40520322e-01 3.09719592e-01
-1.01841700e+00 -8.76572549e-01 4.08680975e-01 8.06077495e-02
-1.58562684e+00 -2.94290185e-01 7.49822915e-01 -2.06759244e-01
4.77709651e-01 9.97829586e-02 3.00284833e-01 7.52428710e-01
-1.24431513e-01 1.03096628e+00 1.18129897e+00 -4.49959248e-01
2.45756403e-01 4.00632650e-01 1.45905778e-01 7.36641169e-01
7.02480495e-01 3.50502171e-02 -3.70008737e-01 1.50539249e-01
5.38545191e-01 2.59556621e-01 3.86260189e-02 -2.62066603e-01
-1.04555142e+00 6.35621130e-01 5.41129172e-01 3.73050608e-02
-1.95309818e-01 -2.79667139e-01 2.02102661e-01 -1.40130326e-01
5.70033193e-01 1.74233109e-01 -5.99911511e-01 4.15156931e-01
-1.02012038e+00 -1.99908763e-01 5.04617333e-01 1.19651723e+00
9.99094605e-01 4.66592982e-02 -1.63735747e-01 8.07065010e-01
4.18409914e-01 4.13139939e-01 2.75211215e-01 -5.46462238e-01
5.96135616e-01 9.25532997e-01 1.79509491e-01 -8.04838598e-01
-4.75357801e-01 -7.81706274e-01 -8.54855895e-01 1.78284883e-01
3.87440979e-01 -6.93150535e-02 -1.23363173e+00 1.53774822e+00
5.05181074e-01 2.07637042e-01 2.22222768e-02 9.00483906e-01
6.23537838e-01 5.68205833e-01 5.81369251e-02 -5.03398418e-01
1.08130884e+00 -1.10934985e+00 -4.69920367e-01 -3.50844502e-01
6.98917329e-01 -3.98076743e-01 9.70591307e-01 4.56151426e-01
-4.20638114e-01 -6.46225512e-01 -1.32703507e+00 3.28882366e-01
-3.70336741e-01 6.31361544e-01 5.41512430e-01 4.35692132e-01
-4.06663775e-01 5.94644129e-01 -8.19522142e-01 -1.20289758e-01
7.87937224e-01 2.38450035e-01 -2.68670857e-01 -1.38374582e-01
-9.74091649e-01 7.23871589e-01 7.59153068e-01 4.37395185e-01
-9.17175412e-01 -3.90166551e-01 -5.97308815e-01 -2.65742660e-01
9.17109132e-01 1.41250923e-01 9.02948737e-01 -6.27685666e-01
-8.52381766e-01 7.28499472e-01 5.88421486e-02 -3.46358627e-01
6.70313120e-01 -3.60336840e-01 -5.98159909e-01 1.87351644e-01
3.90717745e-01 7.40523040e-01 7.04756439e-01 -1.55069017e+00
-1.02090263e+00 -4.81987178e-01 -2.38395348e-01 2.19020799e-01
-6.66529238e-01 -5.49804926e-01 -5.92839181e-01 -6.09978616e-01
5.34736991e-01 -8.62501204e-01 -3.09129149e-01 2.57197917e-01
-6.42408550e-01 -3.68058801e-01 9.94608104e-01 -3.68109226e-01
1.33745360e+00 -2.12235951e+00 -2.80166119e-01 1.34402826e-01
2.68986315e-01 4.35829669e-01 1.67611480e-01 -1.02499224e-01
2.16743574e-01 7.04431534e-03 -3.95815939e-01 -3.20531607e-01
-2.96395093e-01 3.94441456e-01 -2.16394767e-01 3.95949960e-01
5.61103940e-01 6.60941899e-01 -9.45902407e-01 -9.87561345e-01
1.60088807e-01 2.28983387e-01 -4.35725413e-02 1.52420014e-01
-2.10139886e-01 1.70858905e-01 -7.86957800e-01 1.04990625e+00
8.34864497e-01 -4.12827373e-01 1.49463132e-01 -4.17913586e-01
-2.76794314e-01 -1.84589643e-02 -1.51444602e+00 1.51048303e+00
-2.33549163e-01 5.15453480e-02 -6.07050098e-02 -1.06276679e+00
1.13808608e+00 -9.71170980e-03 2.89521515e-01 -3.43953729e-01
3.03098420e-03 5.69450378e-01 -3.78287196e-01 -3.27590823e-01
4.35639918e-01 -6.33508852e-03 1.50828212e-01 3.74540687e-01
-1.06877640e-01 2.86556184e-01 1.73681393e-01 2.33359784e-01
7.43145108e-01 4.27046984e-01 2.39951000e-01 -4.30014461e-01
4.62802142e-01 9.38565806e-02 8.94161820e-01 7.35998809e-01
-2.34276757e-01 7.25108266e-01 2.47100443e-01 -3.14331114e-01
-7.97769129e-01 -9.84947801e-01 -4.41389561e-01 8.53118896e-01
5.29312015e-01 -7.67913610e-02 -6.03409946e-01 -1.22382963e+00
1.51301339e-01 5.84538341e-01 -6.70996666e-01 -1.40208870e-01
-3.47529382e-01 -1.05390680e+00 4.97164220e-01 9.42569017e-01
6.89240754e-01 -9.99239743e-01 -4.44644630e-01 3.25342029e-01
-4.78867032e-02 -9.99278188e-01 -1.68033317e-01 7.52759099e-01
-9.63417828e-01 -1.00483298e+00 -5.62337399e-01 -9.18639183e-01
1.01951826e+00 5.48221648e-01 8.56219649e-01 1.00447364e-01
-1.49416924e-01 -4.86893207e-01 -4.87379670e-01 -4.55626070e-01
-2.00102136e-01 2.55418658e-01 3.68050039e-01 1.02838259e-02
6.49828792e-01 -3.88881564e-01 -5.15913904e-01 7.36459672e-01
-7.09582806e-01 2.31442899e-01 9.93862331e-01 9.76666391e-01
1.22248912e+00 2.82514304e-01 1.09241951e+00 -1.14081669e+00
-1.53013423e-01 -6.59479618e-01 -6.52330816e-01 4.64920968e-01
-1.31599712e+00 1.77393839e-01 5.98221838e-01 -5.85213065e-01
-1.26630557e+00 4.30820376e-01 7.10887536e-02 -4.01727200e-01
-1.16260551e-01 3.15082550e-01 -3.05359006e-01 1.21110842e-01
7.66837895e-01 9.98784602e-02 -1.88285396e-01 -7.08609641e-01
1.68521494e-01 9.52545106e-01 6.35446668e-01 -5.55602908e-01
8.98454428e-01 6.54480457e-01 -3.09126265e-02 -4.45063353e-01
-1.49085522e+00 -5.72230399e-01 -6.95444763e-01 -1.32362798e-01
5.93760073e-01 -1.12840009e+00 -1.65854201e-01 4.88969773e-01
-7.58491755e-01 -2.61167847e-02 -2.51980066e-01 5.50015569e-01
-1.08436625e-02 3.59237969e-01 -5.11754215e-01 -1.16190445e+00
-4.28775758e-01 -9.75549519e-01 1.09238327e+00 5.26719570e-01
4.68030721e-01 -2.35969499e-01 -4.84032154e-01 3.34307373e-01
5.72530739e-02 1.49507761e-01 4.14010763e-01 -7.66700268e-01
-5.56680799e-01 -1.64698362e-01 -7.55833089e-01 5.74826360e-01
1.62005514e-01 -1.12436973e-01 -1.26771891e+00 -1.69612318e-01
-1.57635108e-01 -6.45581007e-01 1.13428736e+00 1.53705135e-01
1.36848629e+00 -5.53041101e-02 -6.40969336e-01 3.67532581e-01
1.30809677e+00 -2.11530663e-02 4.85664368e-01 3.64046991e-01
8.36838782e-01 7.17139840e-01 1.36670470e+00 4.86488402e-01
4.15602148e-01 4.25417453e-01 6.32025898e-01 -1.37422353e-01
5.73199301e-04 -4.32105273e-01 8.66921395e-02 6.83294177e-01
-1.43579964e-03 3.59474793e-02 -7.33614683e-01 5.96231639e-01
-1.95865488e+00 -6.00497782e-01 -4.05253261e-01 2.05486941e+00
1.11064970e+00 7.17742145e-01 -7.94419495e-04 3.83189648e-01
1.02360106e+00 -3.81548516e-02 -7.82398105e-01 6.07477307e-01
-2.25250125e-01 -2.24753276e-01 8.12252283e-01 -5.19819371e-02
-1.38541162e+00 7.07950830e-01 5.21967936e+00 1.34391129e+00
-7.63102949e-01 1.37989387e-01 8.31829071e-01 1.64981067e-01
-1.45882666e-01 -1.37212008e-01 -1.38923299e+00 5.34788013e-01
5.67936838e-01 3.68767798e-01 2.94418819e-02 1.41724849e+00
-6.82604760e-02 -3.31776470e-01 -8.53688359e-01 7.80807018e-01
-8.29388350e-02 -1.20802307e+00 -2.85119534e-01 -3.83508168e-02
7.89225817e-01 2.19596639e-01 -3.04829031e-01 4.17373478e-01
2.49888189e-02 -6.50264323e-01 9.14767981e-01 2.27688745e-01
1.19137800e+00 -7.39942968e-01 9.11897242e-01 6.57721519e-01
-1.48748541e+00 -1.32925376e-01 -6.05074584e-01 1.58672616e-01
9.09764320e-02 1.12179422e+00 -6.86757147e-01 7.27371991e-01
8.14638555e-01 6.42252088e-01 -6.33497655e-01 8.19258034e-01
-4.20319378e-01 7.89710402e-01 -6.57028675e-01 1.11027081e-02
1.08418666e-01 -3.60438302e-02 1.73706800e-01 1.17069924e+00
2.51297563e-01 -1.23600420e-02 3.12653363e-01 8.20258498e-01
-1.00364476e-01 -1.79913547e-02 -3.42604458e-01 1.05268337e-01
7.59638727e-01 1.50725436e+00 -1.07225895e+00 -3.77519220e-01
-3.72138768e-01 5.64375937e-01 3.69312435e-01 4.42254432e-02
-7.86486626e-01 -4.56572801e-01 -4.59753387e-02 2.31834412e-01
3.65987420e-01 -1.40976449e-02 -5.27009666e-01 -1.07323086e+00
3.00653517e-01 -3.60265017e-01 3.83858025e-01 -4.10098314e-01
-1.45000041e+00 6.46088958e-01 9.85320751e-03 -1.60854483e+00
1.42097160e-01 -5.58206260e-01 -3.29998702e-01 6.83049917e-01
-1.85314119e+00 -1.25085950e+00 -5.50458431e-01 1.88239664e-01
5.66431999e-01 -1.38034821e-01 6.38413429e-01 4.24823999e-01
-7.26374507e-01 4.82200801e-01 4.81497571e-02 4.52150069e-02
7.13068128e-01 -1.21876121e+00 1.96137294e-01 9.44320083e-01
3.70736495e-02 3.64089102e-01 3.25250030e-01 -7.78668225e-01
-9.60172415e-01 -1.47353399e+00 5.00002086e-01 -3.00096780e-01
4.23634768e-01 -3.45362544e-01 -1.01924670e+00 3.01966876e-01
-6.25826895e-01 4.47621316e-01 3.62886369e-01 -6.73036128e-02
-4.60696399e-01 -4.25509959e-01 -1.23954892e+00 2.24398196e-01
1.11317158e+00 -2.91853219e-01 -5.22267878e-01 2.17034474e-01
8.91159534e-01 -1.37939677e-01 -6.32222176e-01 8.83601725e-01
5.28547823e-01 -7.64144659e-01 7.59306848e-01 -1.29013151e-01
3.86444598e-01 -9.18344259e-01 -1.58870876e-01 -7.37985432e-01
-1.37762561e-01 -7.31983334e-02 -4.94093359e-01 1.47605026e+00
5.81424356e-01 -3.50660533e-01 1.09343147e+00 2.91566283e-01
-2.72983789e-01 -9.96208727e-01 -7.72297978e-01 -9.43384349e-01
-4.60349560e-01 -4.44859207e-01 4.11714166e-01 9.39224601e-01
-3.44100893e-01 3.02675724e-01 -4.57848966e-01 2.87309915e-01
8.12024593e-01 2.61299908e-01 6.07093871e-01 -1.61980081e+00
-3.45530927e-01 6.99370950e-02 1.39475465e-02 -1.25907969e+00
-2.10694641e-01 -6.88384593e-01 4.45035875e-01 -1.25200951e+00
3.46564710e-01 -1.10185707e+00 -6.25579417e-01 8.44413757e-01
-4.12855953e-01 6.25623047e-01 -2.38930807e-01 5.10065913e-01
-7.88847148e-01 6.03045583e-01 1.00798941e+00 1.85162779e-02
-1.57416075e-01 3.43692042e-02 -7.48172462e-01 7.39873290e-01
6.49282217e-01 -7.98512995e-01 -4.11942780e-01 -2.06419989e-01
3.40324491e-02 -3.15846682e-01 7.32359663e-02 -1.00717342e+00
1.12000972e-01 -3.19494188e-01 5.07125854e-01 -1.15560591e+00
3.01749781e-02 -8.32151055e-01 -8.81955922e-02 4.34307277e-01
3.91815975e-02 -5.03057301e-01 -3.26993428e-02 9.05711830e-01
-8.88071507e-02 -4.51002687e-01 8.92967939e-01 3.56547758e-02
-8.41165423e-01 2.70542353e-01 1.29785448e-01 -1.14300258e-01
1.15311444e+00 -2.51219511e-01 -2.94452280e-01 1.97241709e-01
-4.54794765e-01 4.71932828e-01 3.13518971e-01 3.01686704e-01
4.82566386e-01 -1.28136301e+00 -5.94043195e-01 7.28544444e-02
4.50019926e-01 9.00211453e-01 1.54262319e-01 7.90141344e-01
-2.85439044e-01 9.20546651e-02 5.80491312e-02 -7.03779042e-01
-7.95633793e-01 5.52224994e-01 -9.02212188e-02 -2.88296342e-01
-5.55698693e-01 7.62320876e-01 1.57910034e-01 -4.35671091e-01
3.78650725e-01 4.89362106e-02 -2.69106984e-01 2.28520244e-01
5.44623852e-01 4.31239694e-01 1.69082329e-01 -3.34759831e-01
-4.05477762e-01 3.47389102e-01 -3.35932344e-01 2.80805230e-01
1.46861637e+00 -1.12501867e-01 2.12807104e-01 4.57674056e-01
8.02880108e-01 -1.60784826e-01 -1.56450057e+00 -4.54646379e-01
1.84785903e-01 -3.75374496e-01 1.12629205e-01 -7.28607953e-01
-1.09641433e+00 6.59622431e-01 6.08142674e-01 -1.22924028e-02
1.01772249e+00 1.04145423e-01 5.01562834e-01 5.98910093e-01
6.37653768e-01 -1.51563954e+00 1.26364514e-01 9.25321355e-02
3.74843419e-01 -1.61403441e+00 3.17528516e-01 -6.36035621e-01
-6.67135298e-01 9.62392271e-01 1.05329168e+00 1.36610284e-01
5.42819977e-01 3.32564294e-01 -1.56821758e-01 -3.92281450e-02
-4.82649684e-01 -2.64639139e-01 1.77398950e-01 5.04254997e-01
-2.39941794e-02 5.72886392e-02 -3.53001952e-01 8.82628143e-01
5.22868931e-01 9.41268727e-02 2.88040042e-01 1.02600968e+00
-7.86613584e-01 -8.35019708e-01 -2.82544822e-01 9.15495217e-01
-3.60751033e-01 5.32864146e-02 -7.23844301e-03 7.11138844e-01
3.68880093e-01 8.93976331e-01 -1.97529376e-01 -6.35310888e-01
1.77734345e-01 -1.79127619e-01 -5.57055362e-02 -7.42357194e-01
-4.90008257e-02 2.56072551e-01 7.61253908e-02 -2.65469283e-01
-5.48270643e-01 -5.72432697e-01 -1.56224334e+00 1.45014122e-01
-1.17433286e+00 1.52291819e-01 5.55833399e-01 9.09640253e-01
2.05858856e-01 3.49806309e-01 7.17725039e-01 -7.47380793e-01
-8.42314124e-01 -1.22635365e+00 -7.99710333e-01 1.33725464e-01
-3.72101441e-02 -1.07229066e+00 -4.32188511e-01 -1.16978288e-01] | [9.168694496154785, 1.269552230834961] |
ed45a653-1c73-4628-abbf-15639077bb36 | dear-sir-or-madam-may-i-introduce-the-gyafc | 1803.06535 | null | http://arxiv.org/abs/1803.06535v2 | http://arxiv.org/pdf/1803.06535v2.pdf | Dear Sir or Madam, May I introduce the GYAFC Dataset: Corpus, Benchmarks and Metrics for Formality Style Transfer | Style transfer is the task of automatically transforming a piece of text in
one particular style into another. A major barrier to progress in this field
has been a lack of training and evaluation datasets, as well as benchmarks and
automatic metrics. In this work, we create the largest corpus for a particular
stylistic transfer (formality) and show that techniques from the machine
translation community can serve as strong baselines for future work. We also
discuss challenges of using automatic metrics. | ['Sudha Rao', 'Joel Tetreault'] | 2018-03-17 | dear-sir-or-madam-may-i-introduce-the-gyafc-1 | https://aclanthology.org/N18-1012 | https://aclanthology.org/N18-1012.pdf | naacl-2018-6 | ['formality-style-transfer'] | ['natural-language-processing'] | [ 5.81294358e-01 1.68620735e-01 -3.87348324e-01 -4.86922890e-01
-1.26684439e+00 -9.92668629e-01 1.14479077e+00 -1.95184097e-01
-4.04084951e-01 1.28540742e+00 4.00533020e-01 -4.86120582e-01
3.91044259e-01 -3.47201735e-01 -6.81085944e-01 -1.54640079e-01
4.58543807e-01 8.72511268e-01 1.98438078e-01 -4.72772717e-01
5.32318652e-01 4.74657983e-01 -9.15791452e-01 1.34268746e-01
9.95043159e-01 4.15322959e-01 -1.50467262e-01 4.31698263e-01
-3.61983448e-01 3.80522102e-01 -8.50469410e-01 -9.97035921e-01
1.24801718e-01 -8.47835481e-01 -1.39814758e+00 -1.14850610e-01
7.51605570e-01 -1.21459262e-02 8.08901191e-02 8.84699404e-01
5.57921052e-01 -5.18929511e-02 9.58257616e-01 -1.11599874e+00
-9.85345066e-01 5.62206328e-01 -3.82146925e-01 1.52759761e-01
2.79106289e-01 1.86735004e-01 9.71404195e-01 -8.59972298e-01
9.96278644e-01 1.29490626e+00 4.90406245e-01 9.53447402e-01
-1.47655571e+00 -5.97038507e-01 -2.21472502e-01 -8.43678862e-02
-9.21468556e-01 -6.42910123e-01 6.22982681e-01 -4.91121173e-01
9.55128312e-01 2.24766180e-01 1.56995952e-01 1.42424858e+00
1.45263970e-01 7.80894399e-01 1.54062533e+00 -8.74178946e-01
-1.09842882e-01 4.36149001e-01 -1.12670638e-01 3.89034837e-01
1.60655975e-01 1.98937967e-01 -6.02103591e-01 6.89490363e-02
7.42204845e-01 -8.46475184e-01 -7.18895718e-02 -3.09888244e-01
-1.29262865e+00 8.05098534e-01 5.59570752e-02 7.08658338e-01
2.86574394e-01 1.50301933e-01 5.84417284e-01 7.45112896e-01
7.93536425e-01 1.20879030e+00 -4.74405825e-01 -7.02211559e-01
-1.13929987e+00 3.34927946e-01 1.00448775e+00 1.25785863e+00
6.72104895e-01 -8.83793011e-02 -2.29361519e-01 8.94767284e-01
-3.06681216e-01 4.76294369e-01 2.59243727e-01 -1.08297622e+00
7.19016552e-01 1.75118148e-01 3.55856717e-01 -5.06372035e-01
7.45019838e-02 -1.23892561e-01 -2.00681925e-01 1.20381944e-01
5.41205227e-01 -3.20922047e-01 -5.32608509e-01 1.82786500e+00
-7.05808103e-02 -3.42051983e-01 -1.07544981e-01 4.68907952e-01
4.17780697e-01 4.51990426e-01 2.36340500e-02 -1.67818088e-02
1.02773809e+00 -1.04866838e+00 -5.87035418e-01 -3.44421297e-01
7.50197887e-01 -1.49831748e+00 1.51372182e+00 2.75573939e-01
-1.35850680e+00 -4.35578287e-01 -8.82363975e-01 -2.84148246e-01
-3.04491252e-01 1.08077191e-01 5.60414314e-01 7.21813262e-01
-1.30879855e+00 8.06651890e-01 -6.33739173e-01 -8.69771600e-01
3.51616949e-01 1.56041607e-01 -2.88924128e-01 7.19136447e-02
-1.04684007e+00 1.40980744e+00 1.71084687e-01 -4.01902288e-01
-4.35402215e-01 -7.35316813e-01 -6.33839846e-01 -4.27103579e-01
4.82194796e-02 -7.40503371e-01 1.68108916e+00 -1.34222639e+00
-1.76582694e+00 1.19449353e+00 -2.04354689e-01 -2.84244835e-01
7.06658602e-01 -4.49691892e-01 -4.95128542e-01 -1.73554346e-01
3.33877265e-01 9.42889214e-01 6.02535009e-01 -1.22816026e+00
-7.40663230e-01 -7.00105801e-02 -4.68883477e-02 1.20911628e-01
-5.89323342e-01 5.63266337e-01 -3.16905320e-01 -1.13559234e+00
-5.06401956e-01 -1.21281064e+00 1.22732811e-01 -2.83022195e-01
-2.81907886e-01 -3.23436439e-01 5.66605091e-01 -7.85988688e-01
1.03665912e+00 -1.82067966e+00 4.62745816e-01 -3.27316046e-01
-3.48322123e-01 2.16404766e-01 -2.00203434e-01 6.26243055e-01
1.33381128e-01 4.78443950e-01 -3.33654374e-01 -5.87843359e-01
-5.72028160e-02 2.02976599e-01 -5.81234932e-01 6.74269050e-02
6.30713820e-01 1.00952518e+00 -1.07168865e+00 -5.24814785e-01
-1.93198487e-01 2.37154901e-01 -4.36560452e-01 1.64622530e-01
-1.92455441e-01 6.26153886e-01 -2.02456415e-01 3.44966143e-01
3.55802000e-01 1.29424036e-01 9.12225898e-03 8.12929496e-02
-1.24636754e-01 1.00173795e+00 -6.28549635e-01 2.12540102e+00
-8.41319323e-01 8.07776928e-01 -2.05116764e-01 -4.80300754e-01
8.66008937e-01 3.43995720e-01 3.54202390e-02 -4.20283824e-01
1.13344252e-01 6.00983977e-01 -7.65305609e-02 -1.23872206e-01
7.57481992e-01 -4.76253599e-01 -2.01469868e-01 1.02454388e+00
4.12927330e-01 -7.73579538e-01 5.13261378e-01 4.77185659e-02
8.46231103e-01 6.39714360e-01 1.04079701e-01 -7.48354137e-01
4.42972958e-01 3.35872293e-01 3.45033348e-01 4.52102065e-01
-8.83525908e-02 6.60158455e-01 2.44931683e-01 -2.50658244e-01
-1.48770583e+00 -1.05885446e+00 -1.85660303e-01 1.17059398e+00
-3.21242183e-01 -3.58711004e-01 -1.01102293e+00 -1.07237041e+00
-1.72312915e-01 1.02840817e+00 -7.78712928e-01 2.32905578e-02
-9.63161409e-01 -4.50359344e-01 9.54189718e-01 7.12624431e-01
1.26607746e-01 -1.12211251e+00 -7.73927122e-02 2.25126952e-01
-4.02794451e-01 -1.01000953e+00 -9.18734133e-01 7.61544332e-02
-1.30011666e+00 -7.06647992e-01 -7.78750956e-01 -9.94223952e-01
5.14903724e-01 1.82687476e-01 1.78742039e+00 -1.38965383e-01
1.57552883e-01 9.55366343e-02 -3.64486784e-01 -7.29632258e-01
-1.02406538e+00 7.53367484e-01 -3.40630971e-02 -5.21363556e-01
5.66548884e-01 -5.29519677e-01 -3.16402346e-01 4.56837565e-01
-8.23084891e-01 1.21474735e-01 5.10257781e-01 7.64049947e-01
1.59477338e-01 -6.71855032e-01 6.60000205e-01 -1.16712177e+00
1.00341976e+00 -4.46342453e-02 -3.31133634e-01 3.17418516e-01
-7.57053792e-01 2.68252552e-01 5.65374434e-01 -1.02259010e-01
-1.12694693e+00 -2.89580852e-01 -2.66018603e-02 2.09880099e-01
-3.59272987e-01 6.82900101e-02 -3.65054086e-02 -2.44913232e-02
1.06491315e+00 1.95685718e-02 3.86004485e-02 -7.38167942e-01
6.69748008e-01 6.61546528e-01 4.18684602e-01 -1.16296363e+00
1.02299273e+00 2.32277885e-01 -1.85024276e-01 -5.48931062e-01
-8.97050440e-01 -1.65633023e-01 -1.00989091e+00 3.14772606e-01
5.10326445e-01 -6.47005022e-01 4.39215958e-01 1.06573388e-01
-1.33983350e+00 -5.43878436e-01 -4.54723239e-01 3.05559427e-01
-8.61045182e-01 3.64600807e-01 -8.23860645e-01 -1.78833246e-01
-5.43238401e-01 -1.09229088e+00 1.07797778e+00 1.37675861e-02
-9.49710667e-01 -1.32731688e+00 4.94108319e-01 3.79343152e-01
5.34893751e-01 3.92924771e-02 8.64480913e-01 -5.44919133e-01
-1.29587159e-01 -5.21657318e-02 -9.47541445e-02 4.93871003e-01
5.29484332e-01 1.86169416e-01 -8.39637578e-01 -3.34670484e-01
-1.71304476e-02 -7.56445050e-01 5.24745107e-01 -2.51730215e-02
4.51707155e-01 -2.12668970e-01 -2.26361498e-01 4.54269290e-01
1.17085934e+00 1.74164310e-01 5.96337140e-01 6.26041949e-01
5.58012187e-01 6.68051839e-01 5.03561258e-01 -3.18564504e-01
3.32362831e-01 7.51284897e-01 -5.26159108e-01 -1.69617400e-01
-4.96550143e-01 -3.79591674e-01 5.06568491e-01 1.12458050e+00
-2.30116948e-01 -1.45539746e-01 -8.17863166e-01 7.30283856e-01
-1.43324053e+00 -8.77424359e-01 1.61319211e-01 2.04918861e+00
1.52319801e+00 2.73482859e-01 4.04632419e-01 -1.58617366e-02
7.29027629e-01 -2.70400673e-01 -1.54744223e-01 -9.47201490e-01
-2.11461976e-01 6.17288172e-01 3.10401708e-01 6.66949034e-01
-9.93702888e-01 1.52120340e+00 7.69649696e+00 9.11104560e-01
-1.10208702e+00 1.53382316e-01 6.68818474e-01 2.14691665e-02
-5.78547835e-01 1.93536356e-01 -8.02852869e-01 3.24724555e-01
9.97612178e-01 -2.68962234e-01 6.26505196e-01 5.57704628e-01
3.29303220e-02 1.90439895e-01 -1.59520841e+00 4.82244045e-01
2.89819956e-01 -1.11519563e+00 1.51661679e-01 5.81781119e-02
1.08616173e+00 1.20739825e-01 7.81667978e-02 2.85483509e-01
6.84679091e-01 -9.39967692e-01 8.57347429e-01 -5.44968806e-03
1.13881111e+00 -5.68772554e-01 3.51563632e-01 -1.07691422e-01
-6.39856339e-01 7.22259820e-01 -2.28840768e-01 -7.88599029e-02
7.72310868e-02 2.01335862e-01 -8.80802572e-01 6.25115573e-01
3.95901084e-01 8.45490813e-01 -7.82162368e-01 6.83699250e-01
-5.40288985e-01 1.08034194e+00 -1.44678727e-01 -1.20726325e-01
3.04005921e-01 -4.40353841e-01 4.89476234e-01 1.63763881e+00
2.91917264e-01 -4.83466864e-01 -1.09933406e-01 8.37379158e-01
-2.64607817e-01 4.61410046e-01 -9.13725197e-01 -2.14395404e-01
3.11255842e-01 1.23200810e+00 -5.52302241e-01 -2.57527232e-01
-6.26959860e-01 1.20511460e+00 6.97124422e-01 3.36233497e-01
-5.68832457e-01 -4.65895742e-01 6.02187634e-01 -2.13772543e-02
5.85182086e-02 -3.42193604e-01 -9.62088168e-01 -1.27711523e+00
-1.28193153e-02 -1.31864381e+00 2.76477814e-01 -4.96452570e-01
-1.48245549e+00 6.61633015e-01 -2.99961064e-02 -1.25167894e+00
-4.51589286e-01 -5.71149647e-01 -6.52928174e-01 1.17780530e+00
-1.31749952e+00 -1.16182840e+00 8.87695476e-02 1.15338020e-01
8.19192171e-01 -3.28571767e-01 9.26915944e-01 5.27064539e-02
-2.75663376e-01 9.83604729e-01 2.44737700e-01 3.08720563e-02
1.45106542e+00 -1.46762133e+00 1.23418665e+00 7.99684405e-01
4.47455227e-01 6.69174254e-01 1.01442516e+00 -5.03395617e-01
-1.06718624e+00 -8.77887011e-01 1.21506965e+00 -1.09319997e+00
9.62141991e-01 -3.77623022e-01 -8.21859181e-01 9.24454570e-01
6.85881615e-01 -7.02701509e-01 8.52201819e-01 4.23001796e-01
-4.45644498e-01 2.65266776e-01 -9.35987890e-01 1.01770651e+00
1.20570827e+00 -6.50878966e-01 -8.68635952e-01 2.90473908e-01
5.48537016e-01 -3.26217532e-01 -1.01009333e+00 1.21466920e-01
4.55460072e-01 -6.44759774e-01 5.16108632e-01 -8.42130125e-01
8.57968926e-01 3.88223561e-03 8.12619403e-02 -1.93567777e+00
-2.87443757e-01 -1.03780532e+00 3.23010236e-01 1.78465986e+00
8.10501754e-01 -4.24638957e-01 6.34033263e-01 6.72919393e-01
-1.76376104e-01 -5.21622777e-01 -6.73496723e-01 -1.12634325e+00
1.13483369e+00 -1.63839072e-01 5.72347403e-01 9.52462435e-01
8.95016640e-02 9.72683012e-01 -2.90871322e-01 -7.81077385e-01
3.32603365e-01 1.09774992e-01 1.12516654e+00 -1.08425379e+00
-1.54185221e-01 -8.48311245e-01 -3.71819474e-02 -8.03986967e-01
2.78843403e-01 -1.15985644e+00 -1.09052345e-01 -1.59577942e+00
2.00420901e-01 -4.88377929e-01 -8.25152919e-02 2.36431614e-01
-3.55369449e-01 5.15434682e-01 2.32383355e-01 2.85525084e-01
-2.99249560e-01 4.32311237e-01 1.47934866e+00 -1.15562990e-01
-1.93518370e-01 -3.19934815e-01 -9.56957042e-01 5.03888667e-01
1.14398038e+00 -5.75930595e-01 -3.30723792e-01 -8.78424287e-01
-1.93447825e-02 -4.36494857e-01 -2.66748399e-01 -6.86487615e-01
-3.19584250e-01 -3.42545539e-01 1.55103028e-01 -3.81819606e-02
6.96732327e-02 -3.10426503e-01 -2.35093564e-01 1.55256197e-01
-5.66676080e-01 4.47016656e-01 4.68438953e-01 1.09843656e-01
-2.43711904e-01 -3.52259815e-01 7.71041572e-01 -7.28962049e-02
-3.45038474e-01 5.52187627e-03 -2.24675626e-01 8.69177699e-01
7.58081019e-01 -8.00660287e-04 -2.99701899e-01 -2.70778537e-01
-2.05773622e-01 -7.54670873e-02 1.09416795e+00 6.79140031e-01
3.82228731e-03 -1.64268351e+00 -9.53609049e-01 -2.47134507e-01
3.35952342e-01 -5.23443639e-01 -6.01529896e-01 4.89933819e-01
-5.88096201e-01 5.40462315e-01 -5.01851678e-01 -2.87137419e-01
-9.50208306e-01 4.96659577e-01 1.29077524e-01 -3.48050326e-01
-3.47232580e-01 6.39148235e-01 2.31363438e-02 -5.56515157e-01
-1.63609684e-01 -2.51389835e-02 1.20467067e-01 -1.95586056e-01
4.81014997e-01 3.10668409e-01 2.60757059e-01 -7.33732820e-01
-1.78291306e-01 4.09436077e-01 -2.26914749e-01 -7.28728950e-01
1.20483887e+00 -5.31934276e-02 -2.29793027e-01 6.68659627e-01
1.13259172e+00 1.37787074e-01 -1.00470471e+00 -3.12646121e-01
4.62144166e-01 -4.44745153e-01 -2.80802667e-01 -9.12032008e-01
-4.70021188e-01 9.85757291e-01 1.70543671e-01 -4.65679727e-02
8.07720244e-01 -8.15260187e-02 9.26667213e-01 3.54090095e-01
5.00429749e-01 -1.53465486e+00 -4.73952144e-02 7.38471329e-01
1.12923610e+00 -1.03657746e+00 -1.65239528e-01 -3.81207079e-01
-7.83138216e-01 1.14128876e+00 4.65578973e-01 -2.39284903e-01
3.99883509e-01 2.78470069e-01 4.96800035e-01 2.15867117e-01
-5.53451061e-01 -1.08578518e-01 5.46944439e-01 6.57806814e-01
1.25852275e+00 2.50896197e-02 -8.51386726e-01 1.79053038e-01
-6.28016472e-01 1.31882936e-01 3.94491881e-01 9.90801930e-01
-1.53957829e-01 -2.06113744e+00 -1.90609246e-01 3.58350635e-01
-6.91313326e-01 -2.98662931e-01 -1.08776450e+00 7.15948820e-01
-2.88386554e-01 1.10417056e+00 -1.98932871e-01 -3.42494637e-01
4.21874344e-01 3.82428259e-01 1.05628514e+00 -9.65377152e-01
-8.19328129e-01 -2.13547871e-02 4.27186370e-01 -1.19152330e-01
-3.03456277e-01 -8.80612254e-01 -5.83389580e-01 -4.32389081e-01
-5.35876192e-02 3.17474753e-01 5.38604677e-01 9.35431957e-01
1.65996343e-01 1.24962129e-01 5.19619882e-01 -8.15720737e-01
-6.64752185e-01 -1.21502244e+00 -2.11210012e-01 8.59376311e-01
-1.67259827e-01 -5.58799624e-01 -1.63819760e-01 4.25051987e-01] | [11.501630783081055, 9.752184867858887] |
69bbe372-ac55-478e-bc55-87000a9a66e3 | adapter-tst-a-parameter-efficient-method-for | 2305.05945 | null | https://arxiv.org/abs/2305.05945v1 | https://arxiv.org/pdf/2305.05945v1.pdf | Adapter-TST: A Parameter Efficient Method for Multiple-Attribute Text Style Transfer | Adapting a large language model for multiple-attribute text style transfer via fine-tuning can be challenging due to the significant amount of computational resources and labeled data required for the specific task. In this paper, we address this challenge by introducing AdapterTST, a framework that freezes the pre-trained model's original parameters and enables the development of a multiple-attribute text style transfer model. Using BART as the backbone model, Adapter-TST utilizes different neural adapters to capture different attribute information, like a plug-in connected to BART. Our method allows control over multiple attributes, like sentiment, tense, voice, etc., and configures the adapters' architecture to generate multiple outputs respected to attributes or compositional editing on the same sentence. We evaluate the proposed model on both traditional sentiment transfer and multiple-attribute transfer tasks. The experiment results demonstrate that Adapter-TST outperforms all the state-of-the-art baselines with significantly lesser computational resources. We have also empirically shown that each adapter is able to capture specific stylistic attributes effectively and can be configured to perform compositional editing. | ['Nancy F. Chen', 'Roy Ka-Wei Lee', 'Zhiqiang Hu'] | 2023-05-10 | null | null | null | null | ['style-transfer', 'text-style-transfoer'] | ['computer-vision', 'natural-language-processing'] | [ 3.12547743e-01 1.51325926e-01 7.23213330e-02 -7.71895409e-01
-9.03092384e-01 -9.22772348e-01 5.85841060e-01 -2.81774372e-01
-4.32762563e-01 7.74226844e-01 1.39017150e-01 -3.62913162e-01
3.82071823e-01 -7.13565826e-01 -7.55125463e-01 -3.85669023e-01
5.71114242e-01 9.33939755e-01 -6.35936633e-02 -5.70200562e-01
-1.51417358e-02 1.62975281e-01 -1.13429582e+00 6.66909873e-01
9.34874952e-01 1.00830901e+00 1.96581334e-01 5.84881246e-01
-7.02556610e-01 4.74149942e-01 -8.75555873e-01 -8.41235578e-01
1.34624466e-01 -4.69424784e-01 -9.33109820e-01 -2.81788021e-01
5.94962418e-01 -9.86841395e-02 4.11030233e-01 6.57609582e-01
6.99388087e-01 3.27153616e-02 9.11315501e-01 -1.24376106e+00
-7.77201772e-01 8.71159554e-01 -8.11005011e-02 -3.39070737e-01
1.37862697e-01 1.04036547e-01 9.93589282e-01 -9.60771978e-01
4.16238040e-01 1.55461633e+00 6.91663384e-01 9.91626441e-01
-1.51930678e+00 -8.61110270e-01 1.77404001e-01 -1.95523888e-01
-9.56658423e-01 -5.38896084e-01 6.99412465e-01 -2.79494226e-01
7.92743385e-01 2.60926574e-01 4.49905485e-01 1.61900413e+00
-2.53797006e-02 8.24799240e-01 1.24282539e+00 -5.72654843e-01
6.26278520e-02 5.75566471e-01 -2.36112386e-01 3.99950236e-01
-3.69390965e-01 -3.49358559e-01 -5.24246573e-01 5.36547825e-02
5.96485794e-01 -4.52333272e-01 6.78439438e-02 -4.98085953e-02
-1.32708836e+00 8.50838542e-01 7.82290846e-02 1.73388988e-01
1.00770593e-03 1.76785737e-01 8.09671283e-01 8.50918412e-01
5.54236352e-01 7.54046977e-01 -1.04810119e+00 -2.03594670e-01
-5.96208632e-01 -1.56063689e-02 8.83534551e-01 1.42998588e+00
8.53733420e-01 1.24340571e-01 -4.90260065e-01 1.10444760e+00
-1.18775748e-01 5.58874428e-01 6.01848245e-01 -7.70677984e-01
6.91950023e-01 5.76364636e-01 -4.40714881e-02 -2.47816220e-01
-1.75857306e-01 -3.76607955e-01 -7.65734553e-01 1.41459882e-01
3.35380346e-01 -4.74638194e-01 -7.02576578e-01 2.08016253e+00
2.11252540e-01 -1.26744851e-01 1.57862395e-01 4.60237712e-01
6.70566678e-01 5.99469185e-01 3.10990810e-01 1.96584806e-01
1.36021483e+00 -1.09026790e+00 -6.94834709e-01 -3.99309218e-01
8.19030643e-01 -1.01425719e+00 2.05832505e+00 1.41850337e-01
-1.00328922e+00 -7.02484429e-01 -8.52877736e-01 -2.45386526e-01
-4.43016231e-01 3.35056126e-01 4.35583115e-01 6.27016723e-01
-1.07962966e+00 5.77314556e-01 -4.22733486e-01 -4.58753079e-01
2.69085228e-01 4.81949657e-01 -3.67700785e-01 2.99876213e-01
-1.30450130e+00 9.78008449e-01 3.43412608e-01 -3.40425283e-01
-4.79755312e-01 -8.97700131e-01 -7.52620697e-01 1.24806948e-01
9.84482914e-02 -1.29440880e+00 1.51350188e+00 -1.54109085e+00
-2.22355628e+00 1.01349235e+00 -1.70566607e-02 -7.52519369e-02
5.40736854e-01 -4.26569521e-01 -2.94526398e-01 -2.85078585e-01
6.11653738e-02 8.53697658e-01 1.00885904e+00 -1.05718005e+00
-4.55981106e-01 -5.29784858e-02 1.65963814e-01 3.08730513e-01
-8.52637947e-01 2.70484686e-01 -3.52156907e-01 -1.07524085e+00
-7.65372574e-01 -1.18427956e+00 7.68654570e-02 -3.12160760e-01
-3.99528980e-01 -1.32303908e-01 7.24957645e-01 -2.81880468e-01
1.01322258e+00 -2.08219504e+00 4.75132525e-01 -4.66863215e-02
-3.94072741e-01 2.88529456e-01 -4.60190922e-01 3.79532367e-01
1.93300601e-02 2.26986438e-01 -2.57434249e-01 -9.19794500e-01
1.71540365e-01 3.66929024e-01 -2.90673256e-01 -2.57672489e-01
3.68950039e-01 9.05411839e-01 -5.67100883e-01 -5.30539930e-01
1.05280593e-01 4.33282703e-01 -6.77548230e-01 5.98352015e-01
-4.63693827e-01 7.28116095e-01 -4.35205907e-01 2.15996593e-01
4.35146779e-01 -4.53480706e-02 7.83377215e-02 -1.96905002e-01
1.35675743e-01 5.53881288e-01 -1.03318048e+00 2.03122902e+00
-1.20225596e+00 2.59024560e-01 8.69259983e-02 -5.64413488e-01
1.14872932e+00 4.05543476e-01 -1.14236884e-01 -4.51205522e-01
2.14814112e-01 3.32674116e-01 -2.71888942e-01 -2.24165916e-01
5.38277984e-01 -4.24808711e-01 -5.75932622e-01 8.58300328e-01
4.53147084e-01 -4.85447258e-01 -5.32531217e-02 1.13915183e-01
7.25496829e-01 3.95773917e-01 1.65460899e-01 -4.45268065e-01
7.94902682e-01 -3.34587902e-01 4.23450798e-01 4.86048281e-01
3.36935252e-01 4.62224901e-01 3.04050446e-01 -2.10569307e-01
-1.09476304e+00 -9.52879429e-01 1.70517653e-01 1.81402314e+00
-4.11777079e-01 -4.50263947e-01 -1.08673251e+00 -8.91600788e-01
-7.65224025e-02 1.09756410e+00 -7.54996955e-01 -2.08534211e-01
-7.02529073e-01 -4.55037683e-01 7.38639235e-01 6.51973724e-01
4.27603960e-01 -1.34381330e+00 -2.37481192e-01 1.78199321e-01
-3.80132467e-01 -1.00779080e+00 -9.44006264e-01 1.29217580e-01
-8.70445549e-01 -4.88794744e-01 -4.22632247e-01 -8.49878311e-01
5.11946738e-01 -2.47355938e-01 1.47901404e+00 -1.76570103e-01
2.95208544e-01 2.25517914e-01 -4.23790365e-01 -6.97100163e-01
-9.16862071e-01 9.31708038e-01 -7.75659755e-02 2.43077874e-01
1.83210522e-01 -7.82348216e-01 -3.86527747e-01 3.93328398e-01
-9.84463692e-01 2.66742110e-01 6.63633883e-01 9.16468799e-01
3.46587330e-01 -7.36009061e-01 9.44462776e-01 -1.41725039e+00
8.99099886e-01 -2.24656627e-01 -2.79643506e-01 3.17805827e-01
-4.16232914e-01 4.54542011e-01 1.21475625e+00 -6.71050012e-01
-1.34599221e+00 2.88336743e-02 -1.97516590e-01 -1.84243843e-01
-2.61060238e-01 2.64190674e-01 -5.61302483e-01 2.17852011e-01
6.29982650e-01 1.37614226e-02 4.58662473e-02 -7.64269173e-01
7.40813911e-01 1.03170943e+00 3.60603303e-01 -1.02692068e+00
7.35030293e-01 1.28651574e-01 -3.19466650e-01 -2.76930362e-01
-1.14690185e+00 -4.60863151e-02 -8.57290566e-01 8.88375565e-02
6.34959519e-01 -8.09622467e-01 -4.77546215e-01 6.54838860e-01
-1.29339600e+00 -7.52972484e-01 -4.38151836e-01 1.15964875e-01
-7.59731710e-01 -1.54935047e-01 -7.57512629e-01 -1.94500685e-01
-8.85385692e-01 -1.14703500e+00 1.17124653e+00 -1.50049523e-01
-6.93874180e-01 -1.20437181e+00 5.58365881e-02 2.45103285e-01
8.89987409e-01 -7.68033117e-02 1.24680805e+00 -7.65972018e-01
-8.85542482e-02 1.07404657e-01 5.48475422e-02 6.70013845e-01
4.30125237e-01 -1.26564056e-01 -1.21475518e+00 -2.82545805e-01
-7.39047676e-02 -6.50931001e-01 4.57721204e-01 -9.44220498e-02
1.13859928e+00 -5.14809966e-01 7.55724534e-02 8.60796094e-01
9.40936983e-01 -1.28600568e-01 4.96508658e-01 5.37255943e-01
8.48513007e-01 6.21562183e-01 5.35087883e-01 1.50184795e-01
4.03113842e-01 9.07230914e-01 1.28555313e-01 -2.45875925e-01
-1.75705940e-01 -2.83016473e-01 6.07377172e-01 1.19726205e+00
5.92199154e-02 -5.85071780e-02 -4.33212340e-01 4.16215420e-01
-1.59114897e+00 -6.94414556e-01 2.33817741e-01 2.09209466e+00
1.54226243e+00 4.34552543e-02 1.23969711e-01 -1.40495062e-01
6.29015923e-01 -1.26919955e-01 -5.07150948e-01 -8.97262990e-01
-1.45620316e-01 5.77770710e-01 1.84362963e-01 6.20818257e-01
-9.23044503e-01 1.52416158e+00 6.38364315e+00 8.98626864e-01
-1.23771310e+00 1.73201829e-01 4.27641600e-01 -1.13293208e-01
-5.60379088e-01 -1.18324742e-01 -8.32577586e-01 4.24019963e-01
1.02278781e+00 -1.22751899e-01 6.26957893e-01 6.84456646e-01
4.10193577e-02 5.73998570e-01 -1.46761405e+00 5.11683583e-01
9.05368328e-02 -9.94459450e-01 6.27924502e-01 -4.00030941e-01
5.63993275e-01 -2.42325887e-01 2.36946315e-01 5.51673114e-01
4.23523217e-01 -1.02398849e+00 1.01619351e+00 2.79543638e-01
1.46766913e+00 -7.25095749e-01 5.65431714e-01 1.41069487e-01
-9.22248304e-01 1.12454928e-01 -1.12211242e-01 -5.49836084e-02
8.49395990e-02 2.15566665e-01 -1.01749527e+00 3.49927276e-01
5.54751873e-01 5.43148935e-01 -6.58309937e-01 2.53477544e-01
-5.19141972e-01 7.81254172e-01 -1.32155612e-01 8.04687105e-03
9.40801054e-02 -3.40918273e-01 3.35997283e-01 1.56682968e+00
4.36453611e-01 -3.50156635e-01 -1.13824181e-01 7.95583606e-01
-3.25808465e-01 5.09518623e-01 -3.96408856e-01 1.35772601e-01
6.85962081e-01 1.37432373e+00 -9.94871855e-02 -4.63472426e-01
-3.50454122e-01 1.33750916e+00 6.45388663e-01 2.78778374e-01
-7.61840284e-01 -5.69459021e-01 8.15722108e-01 2.49656774e-02
2.60586768e-01 -7.34544545e-02 -5.02118468e-01 -1.17594612e+00
-9.12031755e-02 -1.23028195e+00 3.96975547e-01 -8.98562372e-01
-1.50853372e+00 9.66281354e-01 -1.04187019e-01 -9.90194619e-01
-3.93693060e-01 -6.45242274e-01 -6.49507821e-01 1.12158799e+00
-1.31503594e+00 -1.66660464e+00 -1.55662909e-01 8.99678290e-01
6.32124245e-01 -4.76573855e-01 1.40648532e+00 2.62373716e-01
-4.10203874e-01 1.21546590e+00 -3.78679931e-02 1.44573703e-01
1.41391027e+00 -1.56428015e+00 7.69008338e-01 3.17274839e-01
-1.50406152e-01 6.35054469e-01 6.89929843e-01 -3.30690682e-01
-1.09583819e+00 -1.29494250e+00 8.66646469e-01 -6.87138438e-01
7.66163945e-01 -7.81942070e-01 -8.73732150e-01 1.12467265e+00
5.74377596e-01 -3.66154462e-01 9.65701938e-01 3.02552968e-01
-5.82506359e-01 -3.77977937e-01 -9.82113481e-01 7.35377192e-01
9.49568748e-01 -5.00796676e-01 -6.62969291e-01 -1.03286635e-02
1.01665998e+00 -2.78480560e-01 -1.00172615e+00 1.02710500e-01
4.88833457e-01 -5.67243934e-01 7.27604449e-01 -8.58509183e-01
4.28792059e-01 -2.88395602e-02 -1.28865438e-02 -2.01077056e+00
-2.05130681e-01 -8.32385182e-01 3.71648759e-01 1.83707702e+00
7.74838924e-01 -7.16515839e-01 3.20404321e-01 4.49987143e-01
-4.02056962e-01 -3.85066897e-01 -7.61091173e-01 -6.91363335e-01
5.13681591e-01 -2.94606179e-01 1.01779711e+00 1.01539874e+00
-1.63414583e-01 1.05441201e+00 -4.40042496e-01 -3.00640255e-01
2.01758012e-01 1.99675232e-01 1.06345725e+00 -1.17933762e+00
-5.16070545e-01 -3.97886336e-01 1.95458785e-01 -8.44624162e-01
5.10585189e-01 -1.06506813e+00 -2.68191367e-01 -1.15775132e+00
-6.62413687e-02 -8.80804598e-01 -1.78631753e-01 8.16882193e-01
-3.05294126e-01 1.98676929e-01 3.58476013e-01 2.66714003e-02
-3.10837001e-01 7.91012466e-01 1.35044396e+00 -1.68922678e-01
-2.70075738e-01 8.32890123e-02 -9.01000619e-01 5.82916558e-01
1.15478492e+00 -5.88090956e-01 -5.32350600e-01 -7.82030463e-01
3.60925496e-01 -2.27506667e-01 -1.32077217e-01 -7.05425739e-01
-2.12554306e-01 -1.86788604e-01 1.42385766e-01 4.30581830e-02
4.24283981e-01 -8.29225421e-01 1.05680246e-02 -1.47930607e-01
-6.82678580e-01 2.66617209e-01 5.69271982e-01 7.15975538e-02
-1.30080178e-01 -1.69893712e-01 8.20382416e-01 -7.43972808e-02
-1.15832590e-01 7.75620341e-02 -3.15421402e-01 3.22683036e-01
7.39716649e-01 8.02988932e-02 -1.85904905e-01 -3.45070839e-01
-5.23636281e-01 1.01781331e-01 5.92852175e-01 5.90465367e-01
1.13405056e-01 -1.55757523e+00 -8.62002373e-01 2.34643266e-01
3.13826412e-01 -1.98984779e-02 -5.34247756e-02 3.50465149e-01
-1.52161583e-01 9.48634893e-02 -5.40080249e-01 -3.64639223e-01
-1.25239384e+00 5.41822970e-01 3.36993277e-01 -3.64187926e-01
-3.68171543e-01 7.07347691e-01 2.96125680e-01 -1.13112926e+00
-1.56336594e-02 -1.80574283e-01 -9.42294076e-02 1.17110610e-01
2.37617150e-01 9.00804847e-02 1.89344391e-01 -4.32117313e-01
-2.50042863e-02 6.93154037e-01 -2.75200695e-01 -3.28712136e-01
1.23797584e+00 -2.42426664e-01 -1.35231331e-01 7.59893417e-01
1.07497334e+00 1.58203796e-01 -1.21175468e+00 -3.64577144e-01
-2.19031259e-01 -1.28335774e-01 -3.54265869e-01 -1.12045991e+00
-9.34533298e-01 1.07091689e+00 9.24023911e-02 -1.68110475e-01
1.17309868e+00 -1.91846237e-01 9.74661708e-01 4.40672874e-01
2.61073679e-01 -1.22931123e+00 1.77736297e-01 7.75333345e-01
1.14574730e+00 -9.18911517e-01 -5.30745149e-01 -3.81465703e-01
-1.08943534e+00 1.04646194e+00 8.23786378e-01 1.44657612e-01
2.57348716e-01 4.50041801e-01 6.21109009e-01 3.01001549e-01
-8.41486752e-01 2.86516905e-01 2.83916801e-01 4.43565995e-01
8.07524145e-01 1.59528896e-01 2.83160377e-02 6.70855999e-01
-6.84289396e-01 -8.20130482e-02 4.29348946e-01 5.31098306e-01
-2.51738150e-02 -1.69532979e+00 -2.57213295e-01 3.65245134e-01
-4.82032090e-01 -3.69080812e-01 -5.76621115e-01 6.55763984e-01
-4.92055155e-02 7.38878369e-01 6.30152300e-02 -3.49443734e-01
5.46642244e-01 5.23408413e-01 4.81719702e-01 -8.25804830e-01
-1.10389304e+00 -1.40158370e-01 3.37302268e-01 -3.30178887e-01
-3.55923712e-01 -5.50693989e-01 -9.94167805e-01 -2.38928452e-01
-1.14963122e-01 1.43205881e-01 5.51562428e-01 9.25137937e-01
5.73724329e-01 5.51321149e-01 5.62374949e-01 -7.62418270e-01
-7.88751900e-01 -1.46131456e+00 -3.53137553e-01 6.95816576e-01
1.52733192e-01 -4.93775576e-01 -2.24810675e-01 3.51451129e-01] | [11.53559398651123, 9.57422924041748] |
4de2f1ad-f39c-4703-b574-b4dae0bfee27 | data-aware-neural-architecture-search | 2304.01821 | null | https://arxiv.org/abs/2304.01821v1 | https://arxiv.org/pdf/2304.01821v1.pdf | Data Aware Neural Architecture Search | Neural Architecture Search (NAS) is a popular tool for automatically generating Neural Network (NN) architectures. In early NAS works, these tools typically optimized NN architectures for a single metric, such as accuracy. However, in the case of resource constrained Machine Learning, one single metric is not enough to evaluate a NN architecture. For example, a NN model achieving a high accuracy is not useful if it does not fit inside the flash memory of a given system. Therefore, recent works on NAS for resource constrained systems have investigated various approaches to optimize for multiple metrics. In this paper, we propose that, on top of these approaches, it could be beneficial for NAS optimization of resource constrained systems to also consider input data granularity. We name such a system "Data Aware NAS", and we provide experimental evidence of its benefits by comparing it to traditional NAS. | ['Xenofon Fafoutis', 'Jan Madsen', 'Emil Njor'] | 2023-04-04 | null | null | null | null | ['architecture-search'] | ['methodology'] | [-3.21263582e-01 -2.91775435e-01 -3.59508276e-01 -4.18203920e-01
-3.25723588e-01 -5.15833676e-01 1.25858501e-01 1.72314774e-02
-3.28636974e-01 5.99349022e-01 -1.27242655e-01 -9.56608117e-01
-1.38536096e-01 -8.70756984e-01 -7.97385991e-01 -4.75495994e-01
2.64924139e-01 5.52969217e-01 3.51828516e-01 -1.39662270e-02
3.98777694e-01 1.05138838e+00 -1.83745730e+00 3.17051053e-01
4.44668680e-01 1.02008653e+00 -1.57452270e-01 7.60529757e-01
-5.07652998e-01 8.50725055e-01 -9.30108130e-01 -5.86353764e-02
3.79819185e-01 -2.34743282e-01 -7.86301196e-01 -8.16266954e-01
4.16133106e-01 -3.29362661e-01 -1.42603442e-01 9.63653564e-01
3.14735502e-01 -4.07814421e-02 4.81915861e-01 -1.35586989e+00
-3.34084809e-01 8.85730147e-01 9.51661915e-02 4.26121086e-01
-2.54045695e-01 2.79227167e-01 9.88197803e-01 -5.35327494e-01
1.99348092e-01 1.04575026e+00 5.17416894e-01 4.78386194e-01
-1.29405487e+00 -4.79313761e-01 1.03388548e-01 2.72104353e-01
-1.45176280e+00 -6.63607538e-01 6.87148035e-01 -1.51488602e-01
1.49939668e+00 6.28584266e-01 5.79534352e-01 8.40107620e-01
9.09993052e-02 4.25921500e-01 9.49666083e-01 -6.86129451e-01
8.49527180e-01 3.01399380e-01 5.08025110e-01 3.40581059e-01
5.67777216e-01 2.91372925e-01 -4.16603237e-01 -3.43775749e-01
5.77535093e-01 -2.91667491e-01 9.16600376e-02 -2.85299957e-01
-7.18406677e-01 7.47052014e-01 5.87015808e-01 5.19977987e-01
-3.59761238e-01 2.50782430e-01 3.97382349e-01 4.38212037e-01
6.54370338e-02 9.90850985e-01 -6.24172509e-01 -4.35652167e-01
-1.13507390e+00 1.53568789e-01 1.09636986e+00 8.70516360e-01
5.86427093e-01 5.58577955e-01 -2.38576308e-01 5.04851460e-01
1.65491864e-01 1.43745527e-01 5.62838674e-01 -1.00825667e+00
5.52180298e-02 7.75793254e-01 4.02261652e-02 -4.15318400e-01
-6.20681286e-01 -3.16168129e-01 -5.75482070e-01 4.66256708e-01
2.69187570e-01 -1.51706964e-01 -7.78585911e-01 1.55807889e+00
-7.09127681e-03 9.09947231e-02 2.54513454e-02 8.96232963e-01
4.03537691e-01 6.72918499e-01 -2.15552226e-01 -1.10878229e-01
8.78930211e-01 -9.37251568e-01 -3.93706828e-01 -1.34577705e-02
5.55886328e-01 -5.97551763e-01 1.35127163e+00 5.81152678e-01
-1.13862157e+00 -4.58566248e-01 -1.28426123e+00 4.02345538e-01
-5.29009163e-01 -1.66505516e-01 6.18842959e-01 1.13433409e+00
-1.32295203e+00 5.93991876e-01 -9.01719749e-01 -2.80132920e-01
-8.18837620e-03 6.78711414e-01 4.59931403e-01 3.16014677e-01
-8.92092347e-01 9.84428883e-01 4.99348730e-01 -2.52439588e-01
-6.92263067e-01 -5.88510454e-01 -1.08821154e-01 5.09312630e-01
2.78076977e-01 -6.66725874e-01 1.47122216e+00 -1.01013780e+00
-1.61298895e+00 2.51680344e-01 6.88975817e-03 -6.81816280e-01
-2.57296171e-02 -2.23184973e-01 -4.76998627e-01 -3.54546100e-01
-8.19345117e-01 3.91637385e-01 7.22376645e-01 -1.14740479e+00
-3.28627676e-01 -2.39861146e-01 3.42374742e-01 -1.74330696e-01
-8.99729311e-01 1.66745439e-01 -2.93013096e-01 -1.76170871e-01
-2.59617746e-01 -1.11526096e+00 -2.21560776e-01 -4.30283844e-01
-4.69328851e-01 -4.63764742e-02 6.13453984e-01 -2.28630334e-01
1.69311273e+00 -1.94941771e+00 -1.33624673e-01 7.24565685e-01
-1.58513300e-02 4.46932226e-01 -6.38082623e-02 -1.40406772e-01
-1.35445492e-02 4.81755465e-01 1.45637542e-01 -1.28562115e-02
1.01660594e-01 1.60562217e-01 -3.47072870e-01 -1.34445369e-01
5.36625348e-02 6.17618799e-01 -2.50242084e-01 -3.57085675e-01
7.56162554e-02 3.04248542e-01 -7.07875311e-01 3.86002302e-01
-5.35375118e-01 -1.57363981e-01 -3.43801975e-01 8.26294839e-01
1.78636029e-01 -3.32438022e-01 2.53447711e-01 -1.66243225e-01
-3.53531659e-01 2.73935407e-01 -1.11206436e+00 1.03179157e+00
-7.22381949e-01 5.36216617e-01 -1.37067035e-01 -6.93053961e-01
1.00186813e+00 1.13959052e-01 2.93965966e-01 -8.31574917e-01
1.07889667e-01 3.41478497e-01 2.81655282e-01 -1.40746415e-01
6.35407150e-01 3.90487611e-01 2.36354962e-01 6.51527524e-01
-3.98604631e-01 1.09175786e-01 -1.16303772e-01 -3.86036247e-01
1.49300003e+00 -1.97473362e-01 2.63094813e-01 -3.18755865e-01
3.03840220e-01 2.23213121e-01 3.24869275e-01 8.62925470e-01
-4.84524071e-02 5.35061359e-01 4.63396668e-01 -6.16103292e-01
-1.46721530e+00 -9.10168111e-01 -2.91218720e-02 1.03413999e+00
-3.47131670e-01 -4.00273502e-01 -1.11484909e+00 -5.55287957e-01
-2.38757282e-01 1.05331933e+00 -1.52440697e-01 -2.07170650e-01
-7.59136796e-01 -6.91676140e-01 7.25226521e-01 6.20097458e-01
1.41921565e-01 -1.04605150e+00 -1.10089529e+00 2.39607289e-01
4.43744242e-01 -6.49681091e-01 -2.13433653e-01 5.52812874e-01
-1.23196650e+00 -7.74887741e-01 -2.33623728e-01 -2.70103037e-01
4.02886957e-01 1.54807210e-01 1.50162339e+00 6.67711318e-01
-1.61095589e-01 7.54822567e-02 -2.08969548e-01 -4.25396204e-01
-6.09902143e-01 5.79959035e-01 4.12380904e-01 -2.95325786e-01
4.83010769e-01 -5.77990770e-01 -4.18582320e-01 2.09263742e-01
-9.94760156e-01 -1.59465671e-01 9.04041946e-01 6.99189603e-01
5.18472612e-01 1.90090518e-02 4.32988346e-01 -8.80590796e-01
6.84954166e-01 -5.61028957e-01 -9.05627728e-01 5.91674805e-01
-1.30022788e+00 4.68497694e-01 1.02336192e+00 -4.66089040e-01
-6.40190601e-01 2.73219328e-02 -1.76764011e-01 -7.80358970e-01
-4.34622049e-01 5.64083934e-01 -8.47642571e-02 -9.69240069e-02
1.06693363e+00 -5.10229282e-02 -9.45102125e-02 -5.32126904e-01
-2.21809745e-02 7.91742504e-01 2.68473834e-01 -8.99844170e-01
4.88129109e-01 -1.08717367e-01 2.71972176e-02 -6.52387977e-01
-3.96981984e-01 -1.54834643e-01 -4.81208086e-01 -5.66667095e-02
4.48673338e-01 -3.03841710e-01 -7.00580120e-01 -3.33763063e-02
-1.08069432e+00 -6.40549719e-01 -2.70309895e-01 2.21108153e-01
-5.24956167e-01 -2.91390151e-01 -4.02632862e-01 -9.05186117e-01
-5.52186966e-01 -1.37467337e+00 4.21765536e-01 5.24322152e-01
-5.72031021e-01 -8.22277904e-01 8.87495354e-02 -5.85524067e-02
1.17951298e+00 -2.08176374e-01 1.49305272e+00 -1.07821727e+00
-8.09966505e-01 -1.77033961e-01 -2.05107316e-01 3.54581237e-01
-2.29523912e-01 5.00477850e-01 -1.05337703e+00 -1.08417422e-01
-7.06211030e-02 -1.03591233e-01 3.25655103e-01 3.63255411e-01
1.51232970e+00 -6.21533155e-01 -1.27141967e-01 5.12037814e-01
1.66818559e+00 5.04942656e-01 6.68486059e-01 5.30659676e-01
5.30440211e-01 3.28349799e-01 3.86053950e-01 3.11090201e-01
1.97450072e-02 1.09435856e+00 6.17556214e-01 1.32594928e-01
1.14754331e-03 3.15845162e-01 6.24683976e-01 7.47057319e-01
-1.02352805e-01 -2.18000844e-01 -1.21642578e+00 3.82710546e-01
-1.69753921e+00 -5.97613275e-01 9.04587209e-02 2.43668079e+00
8.15410316e-01 4.71166283e-01 2.82690912e-01 1.49916679e-01
4.81095612e-01 -2.29081988e-01 -9.67192054e-01 -1.08166289e+00
2.88073323e-03 4.15918231e-01 4.33870912e-01 1.58548698e-01
-5.05185544e-01 7.18891442e-01 7.01339388e+00 6.71948314e-01
-1.30979681e+00 4.78618890e-02 7.36904085e-01 -4.46575165e-01
-3.91430765e-01 8.87805000e-02 -1.14149439e+00 5.15302658e-01
1.91539454e+00 -2.03363866e-01 7.64653087e-01 1.40511560e+00
8.94657746e-02 7.50493780e-02 -1.30858636e+00 6.39427960e-01
-8.87960196e-02 -1.71834362e+00 -3.27285081e-02 1.89078659e-01
4.88182783e-01 6.64856657e-02 -1.12602049e-02 5.45152903e-01
2.54737556e-01 -1.35381126e+00 7.28554666e-01 6.90476656e-01
6.40729070e-01 -1.09533453e+00 8.84728372e-01 3.48951548e-01
-8.35896671e-01 -2.39125997e-01 -2.60498315e-01 2.80788448e-02
-2.29350686e-01 5.54694235e-01 -9.31731343e-01 -1.58429574e-02
8.41410100e-01 -3.83143485e-01 -8.57517004e-01 1.12205172e+00
3.41707289e-01 8.71017337e-01 -4.67460871e-01 -4.82827723e-01
3.91138857e-03 1.35041341e-01 1.83717206e-01 9.95742261e-01
6.41781271e-01 -1.62197277e-03 -1.14411131e-01 1.03530443e+00
9.21339616e-02 6.04005978e-02 -6.36820793e-01 -9.35444757e-02
9.84182715e-01 1.08564234e+00 -6.16241753e-01 -2.46819124e-01
-3.35487574e-01 3.61487746e-01 3.50263000e-01 4.97821681e-02
-8.78657460e-01 -2.27933094e-01 8.07533264e-01 8.66113529e-02
-7.86907002e-02 -2.35615715e-01 -1.00361240e+00 -6.31476045e-01
-1.72081068e-01 -1.09777236e+00 3.57163638e-01 -8.10774326e-01
-9.36400294e-01 9.64976072e-01 -1.05944671e-01 -8.53024125e-01
-5.99611104e-01 -6.11292779e-01 -6.05553925e-01 7.70721972e-01
-1.05555665e+00 -7.20333993e-01 -2.88755834e-01 3.43503177e-01
4.39599782e-01 -6.00970387e-01 1.02492857e+00 1.89735234e-01
-8.55394602e-01 8.44635010e-01 -3.21221165e-02 -5.26483715e-01
4.41217124e-01 -1.23504186e+00 5.51207960e-01 8.28082800e-01
4.03535455e-01 9.15251136e-01 8.99820328e-01 -4.44345146e-01
-1.68703735e+00 -9.69499052e-01 6.13492906e-01 -4.17140454e-01
4.64670956e-01 4.78351600e-02 -1.28434050e+00 4.76913989e-01
1.17388688e-01 -2.67695755e-01 6.05853379e-01 3.94316852e-01
-3.74965429e-01 -2.89433837e-01 -9.92436469e-01 7.91610837e-01
7.60957956e-01 -2.90370345e-01 -4.39953879e-02 3.16510618e-01
9.15176570e-01 -3.33419114e-01 -1.03841245e+00 3.15378994e-01
4.61067826e-01 -1.28941798e+00 9.85307753e-01 -6.94782376e-01
1.48437664e-01 -3.41091156e-01 -4.05174345e-01 -1.20732546e+00
-1.95085004e-01 -2.94576049e-01 -8.59711409e-01 1.13002026e+00
5.86311996e-01 -4.36801493e-01 1.09710193e+00 1.01733363e+00
-3.61733079e-01 -8.53416085e-01 -8.26752603e-01 -1.11384439e+00
3.54356654e-02 -7.08621383e-01 1.33031094e+00 6.55379057e-01
-4.36211169e-01 1.57937005e-01 -1.20411836e-01 2.50258684e-01
3.08199435e-01 1.27610639e-01 7.49294877e-01 -1.18548155e+00
-4.84960973e-01 -8.79242241e-01 -5.36237732e-02 -4.59116757e-01
1.86170876e-01 -6.36073709e-01 -1.83228347e-02 -9.82178092e-01
-6.92673698e-02 -9.71838176e-01 -5.43999255e-01 6.05287075e-01
3.47793579e-01 -8.81176814e-02 2.14230925e-01 3.43004376e-01
-3.11801255e-01 5.87158091e-02 2.17255533e-01 2.48721354e-02
-3.46483439e-01 1.75944075e-01 -4.67353880e-01 5.09517670e-01
1.15180027e+00 -4.98693436e-01 -3.05412352e-01 -3.57758760e-01
3.68998349e-01 -6.99469075e-02 2.14150637e-01 -1.45812571e+00
5.43318152e-01 -5.04549980e-01 1.18037142e-01 -3.56401384e-01
2.67179370e-01 -1.03466964e+00 5.29326618e-01 5.13942122e-01
-2.93917269e-01 5.76237559e-01 2.90258765e-01 6.34678593e-03
-5.28340265e-02 -6.19574964e-01 8.05533171e-01 1.00223497e-01
-8.43322575e-01 3.96226272e-02 -2.05603808e-01 -3.32970291e-01
8.17136407e-01 -1.83625847e-01 -5.82756639e-01 -5.59824668e-02
-2.22215131e-01 -1.71097055e-01 8.01808357e-01 3.56770515e-01
6.71556234e-01 -1.18280268e+00 -2.39223525e-01 3.67302805e-01
1.68615095e-02 -2.63856709e-01 -4.50469740e-03 5.68499327e-01
-7.74088085e-01 8.04758608e-01 -3.56822371e-01 -4.40091163e-01
-1.37610817e+00 6.65578663e-01 3.88064295e-01 -2.44967565e-01
-1.10723712e-01 4.19602126e-01 -4.33449328e-01 -2.26095289e-01
4.56921130e-01 -2.99643397e-01 -9.84864533e-02 -2.75222272e-01
5.87972760e-01 4.21901762e-01 6.04717493e-01 -2.04350054e-02
-3.84395510e-01 8.84341821e-02 8.44967887e-02 -1.96655676e-01
1.36245418e+00 2.72158325e-01 -9.82257202e-02 7.55335987e-01
8.24286044e-01 -1.49665877e-01 -8.54884624e-01 -1.43775240e-01
4.42736179e-01 -1.89974576e-01 3.04309487e-01 -7.98880339e-01
-1.21717381e+00 7.83282816e-01 8.42854917e-01 5.83415687e-01
1.31998575e+00 -4.21238214e-01 5.64638138e-01 7.65149653e-01
5.23670554e-01 -1.24927688e+00 -1.69125292e-02 7.53118336e-01
6.49492919e-01 -9.41246510e-01 -9.85824391e-02 1.89516649e-01
-4.32520598e-01 1.44372582e+00 9.64373350e-01 1.70671064e-02
6.05150282e-01 8.10208499e-01 -3.94913889e-02 -6.18166402e-02
-1.19364750e+00 1.00901388e-01 2.71565288e-01 4.22288686e-01
4.26002771e-01 1.60328656e-01 1.31042883e-01 5.93093336e-01
-3.52477998e-01 1.89132109e-01 6.46863520e-01 9.29797828e-01
-5.61535895e-01 -1.15276611e+00 -5.29913306e-01 7.28141904e-01
-3.23026091e-01 -2.46970683e-01 -3.83608550e-01 6.91915452e-01
-2.48721004e-01 5.82564116e-01 3.19716960e-01 -7.73782790e-01
4.15576845e-01 4.40435708e-01 2.27432966e-01 -5.74895263e-01
-1.02922833e+00 -3.05483460e-01 2.78326362e-01 -6.08897567e-01
2.07821235e-01 -5.76989412e-01 -1.23343277e+00 -6.38733208e-01
-1.30915076e-01 3.80607955e-02 1.20044732e+00 7.19873130e-01
5.98164916e-01 6.73552990e-01 5.27276993e-01 -6.59019411e-01
-7.33004272e-01 -5.34392834e-01 -3.23752463e-01 -1.58113912e-01
-3.08541823e-02 -4.09927189e-01 -2.71044850e-01 -2.20891327e-01] | [8.413973808288574, 3.3572630882263184] |
fa2a38c1-657a-43f1-9927-906f67ba6a3d | a-perturbation-bound-on-the-subspace | 2206.14278 | null | https://arxiv.org/abs/2206.14278v1 | https://arxiv.org/pdf/2206.14278v1.pdf | A Perturbation Bound on the Subspace Estimator from Canonical Projections | This paper derives a perturbation bound on the optimal subspace estimator obtained from a subset of its canonical projections contaminated by noise. This fundamental result has important implications in matrix completion, subspace clustering, and related problems. | ['Daniel L. Pimentel-Alarcón', 'Karan Srivastava'] | 2022-06-28 | null | null | null | null | ['matrix-completion'] | ['methodology'] | [ 4.19788629e-01 -2.08993748e-01 -2.36655831e-01 -1.52725121e-02
-7.53341436e-01 -8.48932385e-01 3.72936189e-01 -6.42310262e-01
-2.04871878e-01 7.62659729e-01 4.76696551e-01 -2.34425545e-01
-4.31265652e-01 4.96208481e-02 -3.94463778e-01 -9.56144631e-01
-3.69945288e-01 3.33269626e-01 -3.34841311e-01 2.25108847e-01
1.03239261e-01 7.50552475e-01 -9.46334839e-01 -4.56698745e-01
7.38963604e-01 3.16556931e-01 1.65703028e-01 6.52294993e-01
5.19477963e-01 -1.07213892e-01 -3.31773758e-01 6.23471066e-02
7.35260248e-01 -6.32195532e-01 -2.68743068e-01 5.04778028e-01
3.80277485e-01 -1.37201965e-01 -6.62269294e-01 1.68249989e+00
5.26524007e-01 4.43776846e-01 9.86881435e-01 -1.10504663e+00
-1.88922629e-01 4.93420839e-01 -6.97521329e-01 2.82926053e-01
3.56844485e-01 -4.85719860e-01 6.50009513e-01 -1.32499754e+00
7.53934324e-01 1.28356898e+00 6.33267820e-01 4.67301637e-01
-1.87556863e+00 -4.99913990e-01 -2.54775852e-01 2.53611594e-01
-1.89989996e+00 -7.57314801e-01 8.27753603e-01 -6.14678681e-01
2.08972007e-01 6.34418070e-01 3.69966090e-01 1.18483078e+00
-1.57699227e-01 5.47256589e-01 1.16341734e+00 -5.31405449e-01
4.23426867e-01 1.07338473e-01 1.88068599e-01 6.30224049e-02
8.59756291e-01 2.24456862e-01 -3.66097391e-01 -6.38007641e-01
8.25259805e-01 -2.14011967e-01 -9.97137427e-01 -1.02450526e+00
-1.18459141e+00 9.25039947e-01 -1.57906502e-01 2.18814939e-01
-3.58111292e-01 -6.57720342e-02 2.69182529e-02 2.83484876e-01
1.46789039e-02 4.68360186e-01 1.40569329e-01 1.04081463e-02
-9.81938303e-01 9.01282430e-02 1.00845325e+00 1.14017928e+00
4.99726474e-01 4.33469504e-01 3.35425287e-01 6.71681881e-01
2.13321283e-01 1.18544626e+00 -6.47848547e-02 -1.34892869e+00
1.87871248e-01 -1.65151805e-01 2.61654526e-01 -9.00327384e-01
-1.98639870e-01 -6.31020427e-01 -1.07850122e+00 -1.54957801e-01
1.80391550e-01 -1.89335689e-01 -2.92292267e-01 1.63923573e+00
3.19592983e-01 5.09988606e-01 5.57224527e-02 1.18819118e+00
2.25955665e-01 5.56406677e-01 -6.12535775e-01 -9.78684127e-01
4.51938212e-01 -3.55398804e-01 -1.06793594e+00 -1.47683889e-01
3.05460133e-02 -8.58181715e-01 3.08960438e-01 5.21971583e-01
-6.34792030e-01 -1.72240436e-01 -1.06757414e+00 4.87015724e-01
2.72600889e-01 1.42796516e-01 3.29120159e-01 7.15020597e-01
-1.02895474e+00 5.16968787e-01 -7.40716279e-01 -7.02061236e-01
-2.17788607e-01 4.12874490e-01 -6.71270847e-01 -1.65576652e-01
-5.88903904e-01 6.81931376e-01 2.51371950e-01 1.39105409e-01
-6.33343041e-01 -1.43635780e-01 -7.21304476e-01 -3.02213043e-01
5.42980656e-02 -6.15553975e-01 7.77200997e-01 -4.41052943e-01
-1.06714356e+00 4.84625280e-01 -6.29166961e-01 -2.93296814e-01
4.36004490e-01 -2.82098114e-01 -6.19604111e-01 1.90685585e-01
3.43996942e-01 -2.45911643e-01 1.30897391e+00 -1.13107312e+00
6.46230280e-02 -9.30165052e-01 -8.47295225e-01 4.66992140e-01
-2.77762949e-01 9.37766302e-03 -4.83545065e-01 -6.12721324e-01
8.81429136e-01 -1.28898847e+00 -6.63442433e-01 -3.11949521e-01
-7.02689111e-01 2.91702807e-01 7.67434597e-01 -6.93638563e-01
1.21557570e+00 -2.28227782e+00 9.65688407e-01 8.04287434e-01
4.21139859e-02 -4.30218756e-01 -5.67083247e-02 6.27826393e-01
-4.79918718e-01 -1.40270963e-01 -2.44110182e-01 2.65473686e-02
-1.74722075e-01 -7.04973787e-02 -8.36953819e-01 1.18495178e+00
-5.69790721e-01 1.97388738e-01 -7.70555019e-01 -9.54308361e-02
3.50472957e-01 2.36542359e-01 -2.88726389e-01 -4.16426733e-03
6.99179232e-01 5.39590955e-01 -3.46267492e-01 4.19923693e-01
9.29356873e-01 -3.77110913e-02 5.72137713e-01 -4.38228339e-01
1.17190883e-01 -4.30859417e-01 -1.82428598e+00 1.55014944e+00
2.31042638e-01 8.64633024e-01 9.71279025e-01 -9.44661379e-01
5.18516839e-01 5.57872474e-01 7.49227881e-01 3.40191990e-01
9.58047286e-02 2.75211573e-01 1.12486772e-01 -6.95143417e-02
2.19268203e-01 -2.25785255e-01 6.27492964e-02 5.20169973e-01
-2.38303700e-03 -7.28833452e-02 7.14869797e-02 6.12763584e-01
9.78519797e-01 -3.56204271e-01 5.83486259e-01 -7.80694842e-01
6.51208401e-01 -2.54683137e-01 5.90501726e-01 8.72257411e-01
-2.58497655e-01 4.97374535e-01 -5.50614409e-02 5.97573109e-02
-1.07746220e+00 -1.48863852e+00 -6.40409768e-01 4.99026388e-01
3.72585833e-01 -3.00844043e-01 -6.58360362e-01 8.91582761e-03
1.16640821e-01 5.07576823e-01 -4.74722266e-01 -5.28097153e-02
-3.81759018e-01 -8.49005461e-01 1.74865827e-01 3.62036079e-01
3.78413498e-02 -1.27091154e-01 1.61754102e-01 9.10323337e-02
-3.55748057e-01 -1.23167729e+00 -6.64992034e-01 3.92465629e-02
-1.27067089e+00 -1.23028123e+00 -7.71458328e-01 -4.53313917e-01
1.01316524e+00 8.74205828e-01 5.27556896e-01 -5.37887573e-01
8.74813646e-03 7.65578210e-01 4.37110662e-02 6.07880168e-02
-2.78160334e-01 -3.14150810e-01 1.00791335e+00 1.71800867e-01
4.36568260e-01 -8.59790981e-01 -1.08361498e-01 5.70541024e-01
-6.09375834e-01 -3.05718720e-01 3.72048974e-01 9.48862374e-01
5.90733767e-01 1.07439905e-01 2.00293720e-01 -5.52589715e-01
6.59133613e-01 -4.15385842e-01 -7.38035023e-01 -2.71739624e-03
-5.23134887e-01 1.64349914e-01 4.86742467e-01 -3.98371458e-01
-8.24507773e-01 6.01290464e-01 6.13682628e-01 -7.78758526e-01
2.87644360e-02 1.80135205e-01 -4.91134554e-01 -3.03501129e-01
1.02287138e+00 5.93409657e-01 2.67384887e-01 -7.66596019e-01
6.00232780e-01 5.03692150e-01 8.47794771e-01 -5.87064087e-01
1.38385797e+00 7.78048813e-01 3.17261547e-01 -1.42079687e+00
-5.61674297e-01 -1.12406015e+00 -1.02313519e+00 -1.10671908e-01
4.14371401e-01 -1.01644421e+00 -2.57642895e-01 -1.55623600e-01
-9.88216221e-01 2.27440313e-01 -1.39124155e-01 8.64142656e-01
-7.64672935e-01 7.82895505e-01 -3.18856478e-01 -1.00484252e+00
7.50110252e-04 -7.96512306e-01 6.79821312e-01 2.68225335e-02
-3.59411091e-01 -9.62824225e-01 4.36022311e-01 3.40959392e-02
-9.23100784e-02 9.07112509e-02 3.39838564e-01 -5.46981215e-01
-4.59614336e-01 -3.88619930e-01 2.80293021e-02 2.71979243e-01
-9.83642638e-02 -3.04617137e-01 -6.57203674e-01 -8.28879774e-01
4.03730839e-01 4.20639127e-01 7.95785069e-01 7.28994489e-01
7.43608773e-01 -4.08237815e-01 -8.05579841e-01 1.01270509e+00
1.51965451e+00 -1.33794114e-01 7.80750737e-02 1.03236005e-01
6.34657264e-01 3.06149721e-01 1.84270859e-01 3.61624360e-01
-6.24832332e-01 6.35773361e-01 -2.40373164e-01 4.01283532e-01
2.08504558e-01 -1.46477167e-02 5.26798129e-01 1.15034521e+00
1.05825730e-01 2.23457128e-01 -5.58145344e-01 6.42160416e-01
-1.82252991e+00 -1.33107364e+00 -1.70879543e-01 2.41164804e+00
2.11010292e-01 -6.85006261e-01 1.99202627e-01 3.09083611e-01
1.20333755e+00 1.59149095e-01 -7.26729929e-01 -2.94416044e-02
-6.41631067e-01 -1.28079951e-01 8.77855182e-01 6.60995424e-01
-1.20054603e+00 4.63485807e-01 8.85389328e+00 6.54887319e-01
-5.35206318e-01 9.22250282e-03 -2.10979640e-01 -2.13579521e-01
-2.42278919e-01 1.08100429e-01 -5.44637620e-01 1.92951754e-01
7.41713643e-01 -9.80763137e-01 1.05678153e+00 1.11372352e+00
2.79437929e-01 1.77173167e-01 -1.22675705e+00 1.41279483e+00
2.74784118e-01 -7.98565984e-01 -1.99106082e-01 6.81543231e-01
1.08530104e+00 6.26032054e-02 7.27509111e-02 -2.14703158e-01
1.44860297e-01 -8.05416107e-01 9.08858925e-02 2.38478318e-01
1.07588863e+00 -5.82056224e-01 4.62668955e-01 3.84353012e-01
-9.84265924e-01 -9.24070925e-03 -6.75436914e-01 3.09751136e-03
3.24909151e-01 7.08083630e-01 -9.58499014e-01 4.12129581e-01
1.78222239e-01 9.19932783e-01 -2.96185732e-01 1.17370892e+00
1.04911722e-01 7.28185356e-01 -5.35359681e-01 2.89559931e-01
-3.54629129e-01 -1.09980452e+00 1.52559519e+00 1.08281755e+00
6.78557098e-01 4.68039453e-01 1.73016623e-01 6.71481252e-01
2.09432244e-01 1.49691448e-01 -1.21241546e+00 4.43938114e-02
1.04657578e+00 1.13033259e+00 -6.09567821e-01 -2.33107820e-01
-2.93858021e-01 1.14929831e+00 -1.22199357e-01 8.59412849e-01
-1.75474018e-01 -2.99326092e-01 8.20828497e-01 8.61568376e-02
2.65257746e-01 -4.02191937e-01 -4.17426705e-01 -1.47010100e+00
3.96879390e-02 -9.14450049e-01 2.71830976e-01 -2.19160900e-01
-1.22953093e+00 1.55437201e-01 9.19892117e-02 -1.54577315e+00
-5.18471837e-01 -6.02084875e-01 -5.87717354e-01 8.37070823e-01
-1.35666758e-01 -3.73930931e-01 1.86949726e-02 6.68638885e-01
-6.77160397e-02 -4.99838412e-01 8.83934319e-01 -5.56742288e-02
-6.91284180e-01 2.23104954e-01 1.09989095e+00 -1.74569860e-02
4.78021681e-01 -1.46210420e+00 4.03553434e-02 1.53960621e+00
2.37376437e-01 9.79751468e-01 1.18501031e+00 -4.50058222e-01
-1.87897766e+00 -8.74087572e-01 1.94786623e-01 -5.45170307e-01
7.15230286e-01 -3.11138798e-02 -6.69070780e-01 8.61212254e-01
3.58427688e-02 -1.70240834e-01 9.31882203e-01 3.80440950e-01
-4.06701446e-01 2.82914545e-02 -9.47239816e-01 6.27233624e-01
1.14829922e+00 -7.35604584e-01 -4.86269772e-01 4.62026000e-01
1.53553575e-01 -1.01388521e-01 -7.89140701e-01 5.74446142e-01
5.64842224e-01 -7.84411490e-01 1.25531626e+00 -5.55706620e-01
-5.80777049e-01 -4.59900677e-01 -6.63095891e-01 -1.34534526e+00
-6.81217790e-01 -1.01284397e+00 -3.67740482e-01 4.54767823e-01
-2.16532908e-02 -5.36969602e-01 9.20923591e-01 4.58277851e-01
2.30168074e-01 -3.79972346e-02 -1.28132355e+00 -1.34459794e+00
-7.56347552e-02 -2.57650524e-01 -3.29887830e-02 1.01297140e+00
3.08055073e-01 6.68456376e-01 -6.60978973e-01 3.05542439e-01
1.58066034e+00 2.77119040e-01 8.85562599e-01 -1.36936057e+00
-1.96083799e-01 -4.42555577e-01 -4.83922064e-01 -1.24273705e+00
4.36825812e-01 -8.94293725e-01 1.50130987e-01 -1.02238011e+00
5.00576913e-01 7.12704137e-02 -1.45414710e-01 -6.36077583e-01
-1.75263286e-01 1.13431916e-01 2.80643851e-01 4.85333592e-01
-4.68583643e-01 4.96873230e-01 8.21731567e-01 -4.44616005e-02
-7.65761435e-02 1.79041505e-01 -5.23332298e-01 7.34418988e-01
4.97932255e-01 -4.04646724e-01 -4.11163867e-01 4.51231562e-02
-2.27966756e-01 3.37319195e-01 4.79073301e-02 -1.03908396e+00
1.79027677e-01 -2.04292074e-01 4.10885870e-01 -6.93792641e-01
5.78128636e-01 -9.15634871e-01 7.01354682e-01 4.33641881e-01
6.01048879e-02 1.38795123e-01 -1.05764434e-01 1.24639988e+00
-1.00270091e-02 -3.47858101e-01 8.56386304e-01 1.12396561e-01
-4.94984686e-01 7.71749616e-02 -6.10560715e-01 6.94700107e-02
9.24578607e-01 -4.30461347e-01 1.01354465e-01 -7.50046611e-01
-1.05336285e+00 -7.72756264e-02 7.88829088e-01 -9.85092670e-02
5.62530696e-01 -1.64718878e+00 -7.42413819e-01 4.56379741e-01
-7.19127953e-02 -5.59497178e-01 1.28687963e-01 7.98140764e-01
-1.21893272e-01 7.04373062e-01 6.07081130e-02 -8.11419785e-01
-1.30861223e+00 7.09099650e-01 4.25307527e-02 4.16238636e-01
-4.40820932e-01 5.80324888e-01 2.55720671e-02 -3.35881591e-01
3.83665338e-02 5.15386105e-01 2.33563021e-01 -3.89939666e-01
6.20270610e-01 1.04045773e+00 -5.21843493e-01 -1.10522866e+00
-4.35101151e-01 5.70828319e-01 4.28070128e-01 -8.17940652e-01
7.68007159e-01 -3.72534424e-01 -5.29396474e-01 6.91266298e-01
1.28641677e+00 4.39049721e-01 -7.22061753e-01 -4.64829445e-01
1.19855054e-01 -7.70173848e-01 -4.92621725e-03 -9.99811739e-02
-4.79499280e-01 4.42192525e-01 4.90024209e-01 1.11320339e-01
1.15742052e+00 -5.12291975e-02 9.87213105e-02 7.37290502e-01
5.55265427e-01 -1.07044590e+00 -4.33752239e-01 3.30497324e-01
1.15359986e+00 -1.03546071e+00 4.47004408e-01 -6.64620280e-01
-5.95277026e-02 9.56176937e-01 1.77927881e-01 -3.84282827e-01
9.90012050e-01 -3.94648984e-02 -1.41271040e-01 1.53369263e-01
-4.85287160e-01 6.26135012e-03 5.43792009e-01 9.90059793e-01
1.11534581e-01 4.62040484e-01 -3.93092424e-01 2.80320138e-01
-2.00707987e-01 -6.13655806e-01 7.20144749e-01 2.23149598e-01
-4.01190519e-01 -8.40602577e-01 -1.06139302e+00 4.57367957e-01
-1.11595646e-01 -1.17249236e-01 -4.24310416e-01 4.42349613e-01
-6.85259521e-01 8.48085582e-01 -2.66165107e-01 -2.74457514e-01
-9.56429541e-02 6.79045990e-02 5.91190875e-01 -5.27397692e-01
5.71957111e-01 6.75202370e-01 -1.53889894e-01 -5.45214176e-01
-3.50302756e-01 -1.31208551e+00 -5.74081123e-01 -3.92301619e-01
-5.34627259e-01 6.90717220e-01 4.71054882e-01 4.23789710e-01
3.13390225e-01 -3.41039479e-01 9.04504776e-01 -7.47644424e-01
-1.18235731e+00 -1.13401425e+00 -1.15074480e+00 3.95730346e-01
3.29728872e-01 -4.87569541e-01 -8.95680726e-01 9.03612748e-02] | [7.543132305145264, 4.393241882324219] |
0a560a0c-8957-4158-9ed5-91d6aa851bcb | tg-vqa-ternary-game-of-video-question | 2305.10049 | null | https://arxiv.org/abs/2305.10049v2 | https://arxiv.org/pdf/2305.10049v2.pdf | TG-VQA: Ternary Game of Video Question Answering | Video question answering aims at answering a question about the video content by reasoning the alignment semantics within them. However, since relying heavily on human instructions, i.e., annotations or priors, current contrastive learning-based VideoQA methods remains challenging to perform fine-grained visual-linguistic alignments. In this work, we innovatively resort to game theory, which can simulate complicated relationships among multiple players with specific interaction strategies, e.g., video, question, and answer as ternary players, to achieve fine-grained alignment for VideoQA task. Specifically, we carefully design a VideoQA-specific interaction strategy to tailor the characteristics of VideoQA, which can mathematically generate the fine-grained visual-linguistic alignment label without label-intensive efforts. Our TG-VQA outperforms existing state-of-the-art by a large margin (more than 5%) on long-term and short-term VideoQA datasets, verifying its effectiveness and generalization ability. Thanks to the guidance of game-theoretic interaction, our model impressively convergences well on limited data (${10}^4 ~videos$), surpassing most of those pre-trained on large-scale data ($10^7~videos$). | ['Jie Chen', 'Chang Liu', 'Zhennan Wang', 'Kai Chen', 'Songyang Zhang', 'Zesen Cheng', 'Peng Jin', 'Hao Li'] | 2023-05-17 | null | null | null | null | ['video-question-answering'] | ['computer-vision'] | [ 2.19751429e-02 -1.72241285e-01 -2.61113849e-02 -2.13876203e-01
-9.34921980e-01 -8.39541554e-01 5.42861164e-01 -2.01269746e-01
-4.43351924e-01 4.05835569e-01 1.79096535e-01 -4.72278625e-01
7.68599659e-02 -6.52097344e-01 -9.51634884e-01 -3.98618758e-01
2.04655305e-01 5.34632862e-01 5.20725310e-01 -5.54713726e-01
-1.99148487e-02 -2.48036459e-01 -1.67419088e+00 4.46076035e-01
9.69568610e-01 1.10285938e+00 2.93116093e-01 7.69454837e-01
-2.91715294e-01 1.18615508e+00 -3.50561082e-01 -1.13370728e+00
3.81783128e-01 -7.05156326e-01 -9.19582844e-01 1.86592355e-01
6.74822807e-01 -5.88521898e-01 -3.80204201e-01 1.08818352e+00
1.79621592e-01 2.35488951e-01 5.54744840e-01 -1.39167893e+00
-7.71870852e-01 5.15198112e-01 -4.83762413e-01 2.47247368e-01
6.38846636e-01 6.46730304e-01 1.54058194e+00 -8.06452572e-01
6.58694327e-01 1.03301549e+00 3.30305398e-01 7.40988076e-01
-6.95062876e-01 -6.87453032e-01 4.10016835e-01 6.54016614e-01
-1.38531685e+00 -3.80379438e-01 7.62985170e-01 -4.68698114e-01
6.96473062e-01 3.68880570e-01 7.28468180e-01 1.13370419e+00
3.55164707e-02 9.37070310e-01 8.41916680e-01 -1.40419856e-01
1.74300432e-01 -3.92723501e-01 -1.70772910e-01 1.01373792e+00
-3.15023214e-01 -1.73121139e-01 -5.20249069e-01 2.01353431e-01
7.81945705e-01 -1.81736574e-01 -1.99954987e-01 -4.89261478e-01
-1.23260975e+00 8.43031406e-01 3.17421913e-01 4.64800596e-02
-3.27455044e-01 2.78844684e-01 4.62807000e-01 3.50269765e-01
2.41444379e-01 5.29662251e-01 -3.24654102e-01 -4.72642481e-01
-7.42757022e-01 3.67820472e-01 4.61770415e-01 1.19566190e+00
6.88100994e-01 1.45653803e-02 -5.13708055e-01 5.64406455e-01
3.13819200e-01 5.96175075e-01 1.49617210e-01 -1.40964806e+00
7.16570973e-01 5.85195243e-01 1.12405926e-01 -1.03229725e+00
-3.80251147e-02 -1.53593406e-01 -6.56484425e-01 -1.56125739e-01
7.13465095e-01 -7.43360445e-02 -9.49208438e-01 1.82753074e+00
2.17657164e-01 3.79303157e-01 -1.77190259e-01 1.21265030e+00
8.74118567e-01 6.92010343e-01 3.57903838e-01 -2.53586292e-01
1.60659885e+00 -1.24941707e+00 -5.85162759e-01 -2.80720383e-01
5.53059638e-01 -4.83979851e-01 1.85596299e+00 3.45045805e-01
-1.27628982e+00 -6.43798590e-01 -6.91297293e-01 -1.53566062e-01
9.62145701e-02 -2.26018116e-01 6.35129929e-01 5.98316491e-01
-1.15735364e+00 5.72946705e-02 -5.01809776e-01 -1.16030023e-01
5.34737945e-01 2.18631461e-01 -2.65147865e-01 -4.83932197e-01
-1.20820487e+00 3.91898841e-01 1.50979087e-01 -2.56557465e-02
-1.20145214e+00 -6.41746640e-01 -8.68015051e-01 5.56246191e-02
1.02984083e+00 -1.02702820e+00 1.42560363e+00 -1.23588312e+00
-1.63409209e+00 9.09791827e-01 -1.22621641e-01 -3.26584697e-01
5.39886057e-01 -1.75966099e-01 -1.97239757e-01 4.61645603e-01
6.64771050e-02 9.19283450e-01 8.95053267e-01 -1.11829233e+00
-8.21723402e-01 -8.22778419e-02 9.54954267e-01 5.17283022e-01
-3.17961276e-01 2.78553274e-02 -1.06438780e+00 -6.09749973e-01
-3.19171220e-01 -9.01349247e-01 -2.96451479e-01 -2.41089836e-02
1.03840366e-01 -3.01508844e-01 3.34182322e-01 -6.88335001e-01
1.11346638e+00 -2.13860464e+00 3.31442595e-01 -8.75534862e-02
5.54088354e-01 2.65896380e-01 -4.41766202e-01 1.89962357e-01
2.56335318e-01 1.09577641e-01 -7.63951614e-02 -2.09131479e-01
2.31971100e-01 2.83957928e-01 -1.57933503e-01 1.30167857e-01
1.96421877e-01 1.32025826e+00 -1.22773898e+00 -7.70388365e-01
1.23568930e-01 -3.31850089e-02 -1.04457057e+00 5.96938252e-01
-7.28379488e-01 3.48324746e-01 -4.91026402e-01 5.60320735e-01
2.60107994e-01 -5.40683508e-01 1.17852524e-01 -3.15396965e-01
3.26299161e-01 -3.61645892e-02 -7.58192420e-01 2.01982808e+00
-4.21959668e-01 4.61050987e-01 2.99368333e-02 -9.77344155e-01
6.02426112e-01 2.19102085e-01 3.94630492e-01 -1.12854266e+00
1.53512061e-01 3.72124836e-03 2.41199043e-02 -6.18295848e-01
4.78271902e-01 -9.56290513e-02 -3.04483473e-01 3.25940996e-01
2.07352310e-01 -1.77138865e-01 4.00616676e-01 5.64253211e-01
1.10676658e+00 2.78266609e-01 1.74558148e-01 -2.32481267e-02
5.45443356e-01 3.95286269e-02 3.65444034e-01 7.63853729e-01
-4.69473869e-01 6.31238759e-01 6.14292920e-01 -2.73154736e-01
-9.68246996e-01 -9.70504284e-01 6.71682298e-01 1.73892748e+00
5.59985518e-01 -6.27814472e-01 -1.06604242e+00 -8.81102085e-01
-5.05712211e-01 4.15412575e-01 -4.70695943e-01 -1.17800929e-01
-5.66164613e-01 -2.41464093e-01 4.50677961e-01 5.18803954e-01
6.24924779e-01 -1.15131605e+00 -4.51032966e-01 5.04285172e-02
-6.20165586e-01 -1.36471057e+00 -7.64943779e-01 -3.56833577e-01
-3.71052772e-01 -1.22506762e+00 -5.35181105e-01 -7.43242264e-01
3.38603020e-01 3.90375733e-01 1.57899749e+00 2.79664516e-01
3.13496411e-01 6.47657394e-01 -6.72878444e-01 7.46326745e-02
-2.25329295e-01 -9.13457945e-02 -2.67867535e-01 -1.96820349e-02
2.65978307e-01 -3.63325894e-01 -7.89128125e-01 4.73417550e-01
-9.37935352e-01 2.50774652e-01 6.15038991e-01 7.91629672e-01
5.40905356e-01 -9.51690823e-02 3.36568981e-01 -8.39375198e-01
6.00941837e-01 -4.08955604e-01 -4.32199180e-01 5.11338353e-01
-1.77252442e-01 -9.73025337e-02 7.54117727e-01 -5.18958986e-01
-9.38485444e-01 -1.62099943e-01 -2.31017336e-01 -8.11308265e-01
-1.74984172e-01 5.15803635e-01 -5.20000398e-01 1.21591628e-01
5.11537313e-01 2.59664804e-01 -1.59562185e-01 1.12903230e-01
8.60246778e-01 4.21806574e-01 7.97282934e-01 -8.48058105e-01
7.47399986e-01 2.81455636e-01 -4.21023935e-01 -3.31708252e-01
-8.80079746e-01 -3.29285324e-01 -4.39132929e-01 -5.45093417e-01
1.27559388e+00 -1.01487207e+00 -1.06468427e+00 3.87538999e-01
-1.01106381e+00 -6.16037607e-01 -1.85498744e-01 1.02277271e-01
-8.37507486e-01 4.45063204e-01 -7.27626979e-01 -4.63875800e-01
-1.61343619e-01 -1.28837097e+00 9.86684740e-01 2.84763258e-02
-5.38511351e-02 -7.83339560e-01 -1.77615315e-01 1.11834061e+00
1.80561274e-01 -1.56612799e-01 8.45633805e-01 -4.55167085e-01
-9.25244808e-01 2.43746564e-01 -4.00001645e-01 2.09670171e-01
-7.32554346e-02 3.34219150e-02 -5.43992281e-01 -1.93621591e-01
-1.37912318e-01 -5.44015825e-01 5.14289081e-01 1.32456094e-01
1.34372628e+00 -3.57242793e-01 1.04534835e-01 3.23655933e-01
1.05877924e+00 2.69704670e-01 7.81154990e-01 1.77883953e-01
9.83742952e-01 4.91920531e-01 9.38779771e-01 3.35799247e-01
8.56799006e-01 7.92011738e-01 8.56774509e-01 1.02767691e-01
-1.62285522e-01 -4.13819432e-01 3.70731622e-01 9.64084685e-01
-2.15987012e-01 -5.46897113e-01 -9.39367652e-01 5.82740009e-01
-2.02717614e+00 -1.04400146e+00 4.28665578e-02 1.71391070e+00
7.38752663e-01 2.81835228e-01 3.80028069e-01 -2.46159568e-01
5.02557755e-01 3.49135548e-01 -4.24587190e-01 -1.15027666e-01
7.01381490e-02 2.06935495e-01 1.51917353e-01 3.43130022e-01
-1.07380962e+00 1.32146323e+00 5.84275389e+00 1.27992392e+00
-8.66638601e-01 1.79077998e-01 6.44255340e-01 -3.61867212e-02
-6.81994200e-01 -1.79117754e-01 -2.56599307e-01 5.36658049e-01
7.34602869e-01 -3.67199555e-02 7.70187140e-01 5.25427878e-01
9.77254957e-02 2.19490439e-01 -1.03253460e+00 1.14287114e+00
1.88535079e-01 -1.46550941e+00 4.29447323e-01 -2.44188353e-01
6.45723581e-01 -2.87753552e-01 1.90995470e-01 7.42570162e-01
5.67693174e-01 -1.04972243e+00 9.66931343e-01 2.88268864e-01
9.50440645e-01 -6.15232706e-01 7.15790868e-01 2.79443711e-01
-1.32536149e+00 -1.08836710e-01 -1.41583055e-01 -2.46736303e-01
3.17059785e-01 1.53276771e-02 -4.10440773e-01 7.27597475e-01
8.46140087e-01 6.41982675e-01 -5.09821594e-01 6.13094389e-01
-2.84907103e-01 6.59695745e-01 5.23983799e-02 -1.83518883e-02
4.53299791e-01 -2.67287493e-01 1.29375070e-01 7.76304543e-01
1.95014447e-01 6.34029031e-01 3.11312497e-01 4.70026374e-01
-3.19901496e-01 1.91644549e-01 -2.99065351e-01 -1.55127391e-01
2.89108694e-01 1.11856079e+00 -5.76170027e-01 -3.20662677e-01
-7.18739450e-01 9.67834353e-01 4.05332476e-01 3.98951828e-01
-1.29138303e+00 7.56474584e-02 7.18925595e-01 9.87591818e-02
5.95213771e-01 -1.89536378e-01 1.54943824e-01 -1.27018249e+00
5.67642637e-02 -1.48056293e+00 5.68897724e-01 -1.08617544e+00
-1.25839341e+00 7.16330588e-01 -7.16914609e-02 -1.31816113e+00
-3.76018852e-01 -5.62078416e-01 -4.20011282e-01 2.38617063e-01
-1.03654778e+00 -1.27781975e+00 -4.57702965e-01 1.01803279e+00
7.12157965e-01 -1.68746725e-01 4.84135836e-01 4.62713301e-01
-3.67379457e-01 7.34229863e-01 -4.20707166e-01 2.78472006e-01
6.64963901e-01 -1.12193584e+00 3.24892074e-01 9.32825983e-01
5.64438522e-01 1.83957919e-01 7.22810805e-01 -3.05099308e-01
-1.41624343e+00 -9.09615695e-01 3.95945519e-01 -6.34174228e-01
9.46223438e-01 -5.32867432e-01 -8.04055035e-01 5.37286937e-01
3.85846734e-01 1.52388245e-01 6.11317694e-01 -5.60357086e-02
-5.94028115e-01 -2.02056289e-01 -8.04907143e-01 1.07251871e+00
1.59208572e+00 -8.22019875e-01 -6.12282217e-01 1.25506401e-01
1.13674593e+00 -6.44159019e-01 -7.39996552e-01 4.46183562e-01
4.50082719e-01 -9.41814661e-01 1.02117991e+00 -8.60374749e-01
8.32278013e-01 -3.47952366e-01 -3.11285943e-01 -9.89266753e-01
-2.87785828e-01 -6.64400399e-01 -1.57900333e-01 1.14716697e+00
2.04984739e-01 -2.08575074e-02 9.60426390e-01 6.08641863e-01
-2.32819051e-01 -7.05164254e-01 -8.38492393e-01 -4.49383110e-01
-5.15167639e-02 -6.56635761e-01 6.12609506e-01 8.73872101e-01
-1.69929862e-01 4.85486180e-01 -5.90617776e-01 1.22223511e-01
2.64274567e-01 1.82676330e-01 9.75392938e-01 -7.23316491e-01
-5.85877657e-01 -5.89905024e-01 -3.63330215e-01 -1.58703947e+00
3.91344815e-01 -4.85791802e-01 1.62772074e-01 -1.37577975e+00
2.56479830e-01 -3.57040167e-01 -2.69280434e-01 4.29326802e-01
-5.56580961e-01 5.42694390e-01 5.09174228e-01 6.17630072e-02
-1.48377287e+00 6.66337729e-01 1.54896367e+00 -3.68768036e-01
2.62083746e-02 -2.33028546e-01 -7.62344480e-01 5.23199856e-01
5.73149562e-01 -1.20473273e-01 -8.19629073e-01 -6.77466214e-01
5.06654620e-01 4.38397884e-01 3.57229263e-01 -7.88074791e-01
2.52645224e-01 -4.71705288e-01 -2.85967916e-01 -1.68873191e-01
3.74863356e-01 -7.53507376e-01 1.07557222e-01 1.09270878e-01
-3.92880946e-01 1.38815299e-01 5.50881587e-02 6.68307841e-01
-4.92419988e-01 -4.71857041e-02 2.62366712e-01 -1.73008054e-01
-1.21980143e+00 7.32815981e-01 -3.45405996e-01 5.67847669e-01
1.20169663e+00 -2.44135708e-01 -4.58669692e-01 -8.47108126e-01
-5.56775451e-01 5.25724232e-01 5.28019130e-01 4.99594629e-01
6.23378456e-01 -1.31644595e+00 -6.79294467e-01 -1.75637081e-01
3.45150441e-01 1.18469179e-01 5.72587669e-01 6.22677028e-01
-5.85440218e-01 1.96434915e-01 -3.87403607e-01 -6.53549194e-01
-1.33563972e+00 7.92185307e-01 2.36831173e-01 -4.26521152e-01
-2.43539855e-01 1.02094567e+00 6.16259396e-01 -1.39299080e-01
1.38681065e-02 -1.73135385e-01 -2.97584593e-01 -1.41623989e-01
3.06264132e-01 -1.81503277e-02 -2.25200102e-01 -6.22155249e-01
-3.11040312e-01 6.38734102e-01 -5.29541187e-02 -5.47026694e-02
9.37469542e-01 -3.10171098e-01 1.43260717e-01 5.35638779e-02
8.53216231e-01 5.35016917e-02 -1.60592723e+00 -1.62179008e-01
-2.61800706e-01 -5.29373765e-01 -3.31216097e-01 -5.44588923e-01
-1.24744260e+00 9.69604909e-01 2.39847586e-01 1.30983725e-01
1.27163005e+00 3.34010988e-01 7.65406251e-01 3.16188604e-01
5.94892740e-01 -7.69703329e-01 7.97161043e-01 5.01999080e-01
6.50804281e-01 -1.23003340e+00 -3.73207092e-01 -4.92112696e-01
-1.06762707e+00 6.45326972e-01 9.33326185e-01 3.97274550e-03
2.57185042e-01 -2.56909102e-01 2.75120288e-01 -2.57477492e-01
-8.79299283e-01 -3.63604695e-01 4.71945226e-01 5.84357023e-01
1.77529857e-01 9.29432884e-02 -1.36681914e-01 7.69570231e-01
-2.68089116e-01 -8.17152038e-02 3.97094309e-01 5.15224040e-01
-2.59938657e-01 -1.05578279e+00 1.29650235e-01 3.77569079e-01
-3.84376198e-01 -2.57030338e-01 -1.07284307e-01 8.40742230e-01
1.36755869e-01 1.10169756e+00 2.31991246e-01 -6.14729881e-01
3.30062747e-01 -2.86315858e-01 5.12404740e-01 -4.50230986e-01
-5.72463155e-01 -1.29948169e-01 1.00318067e-01 -9.15365934e-01
-7.08196580e-01 -4.80876327e-01 -1.16023350e+00 -5.21882296e-01
-6.71851411e-02 1.51244968e-01 -5.29872440e-02 1.29348373e+00
2.59995162e-01 3.81983995e-01 3.81675661e-01 -4.85845387e-01
-3.06842387e-01 -5.07813931e-01 -2.47070745e-01 8.10249627e-01
-2.89202016e-03 -6.19249701e-01 -1.24819554e-01 2.02201843e-01] | [10.392266273498535, 1.0364511013031006] |
3c242e1c-129c-40bc-bd46-a669299bd431 | action-and-intention-recognition-of | 1810.09805 | null | http://arxiv.org/abs/1810.09805v1 | http://arxiv.org/pdf/1810.09805v1.pdf | Action and intention recognition of pedestrians in urban traffic | Action and intention recognition of pedestrians in urban settings are
challenging problems for Advanced Driver Assistance Systems as well as future
autonomous vehicles to maintain smooth and safe traffic. This work investigates
a number of feature extraction methods in combination with several machine
learning algorithms to build knowledge on how to automatically detect the
action and intention of pedestrians in urban traffic. We focus on the motion
and head orientation to predict whether the pedestrian is about to cross the
street or not. The work is based on the Joint Attention for Autonomous Driving
(JAAD) dataset, which contains 346 videoclips of various traffic scenarios
captured with cameras mounted in the windshield of a car. An accuracy of 72%
for head orientation estimation and 85% for motion detection is obtained in our
experiments. | ['Fernando Alonso-Fernandez', 'Cristofer Englund', 'Boris Duran', 'Dimitrios Varytimidis'] | 2018-10-23 | null | null | null | null | ['motion-detection'] | ['computer-vision'] | [-1.37116343e-01 -1.15467258e-01 -4.18069601e-01 -6.12418413e-01
-5.38942993e-01 -4.14858535e-02 8.47767651e-01 -3.39242280e-01
-6.57135129e-01 4.39649433e-01 6.71603605e-02 -6.18804038e-01
3.65612149e-01 -5.27848125e-01 -4.83314127e-01 -7.84257710e-01
1.69795886e-01 6.28018156e-02 6.27751529e-01 -1.58424854e-01
3.34583133e-01 5.99065185e-01 -1.96143305e+00 4.13819626e-02
5.11900723e-01 9.12965059e-01 2.61485249e-01 1.16276228e+00
3.92502278e-01 1.00623679e+00 -1.33220091e-01 -4.15930510e-01
4.48067039e-02 -2.45681386e-02 -3.60332072e-01 3.44832540e-01
8.38025987e-01 -4.78461325e-01 -7.91372359e-01 8.01789522e-01
5.18108547e-01 1.13102339e-01 5.43237805e-01 -1.71079957e+00
1.65615916e-01 -2.18889341e-01 -3.58347684e-01 6.87277019e-01
2.92660326e-01 7.07848370e-01 5.74852288e-01 -7.85058081e-01
3.14137459e-01 1.36490595e+00 3.08477879e-01 4.06154186e-01
-5.88356018e-01 -5.78954637e-01 1.24039851e-01 1.22961843e+00
-1.35053515e+00 -1.06651163e+00 7.93588161e-01 -6.75705671e-01
8.81911457e-01 9.64741036e-02 6.53089046e-01 9.26556766e-01
5.37988722e-01 1.11714983e+00 6.84090137e-01 -7.79853910e-02
2.06847742e-01 1.36345103e-01 7.41469383e-01 6.36393368e-01
4.21910286e-01 5.34953833e-01 -4.28775281e-01 3.06017309e-01
6.30776063e-02 -3.68894249e-01 1.18432008e-01 -2.55257547e-01
-1.14887977e+00 1.07409191e+00 2.66295046e-01 -2.32832000e-01
-6.91447914e-01 1.09221146e-01 3.90856802e-01 -1.07173830e-01
1.83130261e-02 -3.89750928e-01 -5.29282028e-03 -3.38438839e-01
-4.32160169e-01 4.58720654e-01 4.75314081e-01 1.23118758e+00
8.39601219e-01 1.82291314e-01 -2.14733377e-01 4.43225950e-01
5.23091316e-01 1.06178832e+00 1.51801184e-01 -1.08563113e+00
4.70156223e-01 3.61869872e-01 1.67597011e-01 -1.05932546e+00
-5.44699728e-01 8.15395713e-02 -3.94992173e-01 4.33107585e-01
3.33886445e-01 -2.46035933e-01 -7.66476095e-01 1.09826720e+00
4.14785385e-01 3.87589544e-01 1.08582564e-02 9.37415600e-01
9.61244404e-01 4.27650362e-01 3.24987561e-01 7.77768297e-03
1.71848297e+00 -1.00748730e+00 -7.51056373e-01 -8.41971040e-01
7.53591061e-01 -7.00473309e-01 3.38499933e-01 2.75889248e-01
-8.25037181e-01 -9.56320107e-01 -1.07616365e+00 -9.61805731e-02
-4.29005921e-01 2.50983357e-01 3.24274570e-01 8.84644568e-01
-9.79249358e-01 -3.44362408e-01 -7.57583022e-01 -5.68378389e-01
4.13923681e-01 2.44630262e-01 -4.14085150e-01 -3.79174888e-01
-9.92380619e-01 1.38338089e+00 1.97514161e-01 1.60726845e-01
-7.47403800e-01 -1.74581274e-01 -1.24007535e+00 -5.09466648e-01
1.89318180e-01 -6.10842645e-01 1.32258654e+00 -5.38211644e-01
-1.19561207e+00 1.00650048e+00 -7.75387526e-01 -9.82481122e-01
4.83844101e-01 -3.19942176e-01 -8.32161844e-01 1.45689221e-02
4.49270338e-01 9.43992138e-01 8.11248899e-01 -7.78185308e-01
-1.46982372e+00 -4.80457515e-01 -3.01060379e-01 1.84738651e-01
3.70165586e-01 2.36388087e-01 -4.49511468e-01 2.03748569e-01
-2.19145969e-01 -1.27018511e+00 -1.95928544e-01 -2.36320138e-01
-4.36221838e-01 -4.73299742e-01 1.50355685e+00 -5.77040851e-01
1.04460263e+00 -1.91532815e+00 -4.02247012e-01 -1.56885669e-01
1.14376962e-01 5.66971958e-01 7.32016098e-03 -2.22060204e-01
2.20257029e-01 -5.51675737e-01 2.39415094e-01 -8.63932595e-02
-5.60633950e-02 3.00983131e-01 -1.34852350e-01 7.71687567e-01
2.23398030e-01 9.81511652e-01 -7.28161454e-01 -6.18605077e-01
9.50573564e-01 5.51938295e-01 -2.92297393e-01 2.02385157e-01
5.79528034e-01 3.66082191e-01 -4.56480235e-01 5.05263984e-01
8.46731484e-01 6.64655864e-01 -5.03884554e-01 -7.75913894e-02
-5.82618892e-01 4.56391484e-01 -1.02899575e+00 4.84387249e-01
-2.56427467e-01 1.35471690e+00 4.56701405e-02 -9.84891415e-01
6.84140027e-01 2.25214049e-01 2.45527163e-01 -9.26774502e-01
2.53837466e-01 -6.76800460e-02 2.70287961e-01 -9.83396530e-01
6.02233469e-01 3.22848886e-01 -1.24628946e-01 -1.43804803e-01
-4.10065353e-01 2.27524534e-01 5.14397144e-01 -2.55508184e-01
1.00232780e+00 -3.91158342e-01 4.02601004e-01 -1.95852503e-01
1.07459700e+00 8.83433521e-02 4.25142795e-01 7.72838235e-01
-1.11629307e+00 1.45559072e-01 2.05885321e-02 -7.04309642e-01
-8.18542898e-01 -8.70981693e-01 -1.16533905e-01 9.77724373e-01
1.79963514e-01 -3.79479257e-04 -7.23258376e-01 -4.46577877e-01
-1.44007012e-01 1.19491661e+00 -3.39113116e-01 -2.75712371e-01
-8.85475159e-01 -3.45853597e-01 2.36997604e-01 7.12743938e-01
8.63860905e-01 -8.60795259e-01 -8.91185164e-01 1.00787997e-01
-3.20068508e-01 -1.66135383e+00 -4.43082213e-01 -2.71502912e-01
-1.17860891e-01 -1.33934641e+00 -3.02700549e-01 -8.64800930e-01
3.57934535e-01 9.97163892e-01 7.84882784e-01 -1.54207736e-01
-1.63132772e-01 5.17822325e-01 6.00371733e-02 -8.22110236e-01
-3.48850101e-01 -1.29706383e-01 1.69060931e-01 5.11816680e-01
1.11489761e+00 1.32871434e-01 -8.82671356e-01 7.78392076e-01
5.82782738e-02 -1.67740732e-02 4.39430624e-01 3.66319060e-01
-4.03535627e-02 2.68962383e-01 1.64285630e-01 -3.71464193e-02
1.58800200e-01 -5.23464620e-01 -7.12759197e-01 -4.13328022e-01
-1.79991424e-01 -4.18983638e-01 1.96888551e-01 1.51265673e-02
-1.01059961e+00 3.69237840e-01 -5.70394218e-01 3.89148691e-03
-1.00355291e+00 -3.54580820e-01 -4.47125822e-01 -4.24637366e-03
2.62496740e-01 2.70276546e-01 7.99590200e-02 1.22548327e-01
3.26909959e-01 9.49276686e-01 8.04920435e-01 2.16112539e-01
6.37047887e-01 6.90715969e-01 9.46977437e-02 -1.61309588e+00
-5.34220338e-01 -1.03232968e+00 -9.75580633e-01 -7.71566927e-01
1.30505574e+00 -1.06835294e+00 -1.06562483e+00 7.32382655e-01
-1.13078332e+00 -7.61511130e-03 3.55104476e-01 9.62358952e-01
-7.44989574e-01 2.62998939e-01 -1.78933024e-01 -1.09282541e+00
3.75162102e-02 -1.44927692e+00 1.21727812e+00 2.21714064e-01
-1.67979434e-01 -8.78745615e-01 -2.62968600e-01 8.49579036e-01
3.33088964e-01 -1.58726513e-01 3.83446604e-01 -3.87239814e-01
-7.89140880e-01 -5.40099859e-01 -2.26477921e-01 1.28714204e-01
-1.38415411e-01 3.55820899e-04 -1.04074037e+00 4.37806174e-03
-1.33390650e-01 2.29502901e-01 9.53818858e-01 9.81996834e-01
3.36965054e-01 -1.44824192e-01 -6.12570226e-01 2.31357068e-01
7.58004248e-01 7.14809358e-01 1.06820059e+00 5.37244201e-01
7.04905510e-01 9.09859478e-01 8.67511749e-01 2.31406331e-01
8.90188813e-01 9.89298582e-01 4.52608049e-01 1.56368896e-01
-1.89038903e-01 7.24272504e-02 6.35756969e-01 2.00752378e-01
-5.79825835e-03 -2.19235063e-01 -9.57020998e-01 6.82791650e-01
-1.83297646e+00 -1.33026433e+00 -9.12477791e-01 2.15651226e+00
-1.52862057e-01 3.65314692e-01 5.69051564e-01 3.71009797e-01
1.00829494e+00 2.77919043e-02 -3.03000838e-01 -4.21092778e-01
1.80924311e-01 -6.13372922e-01 8.06232870e-01 8.13667834e-01
-1.61449409e+00 9.39394474e-01 6.61047077e+00 4.63162720e-01
-1.04091120e+00 -8.67065117e-02 6.02777898e-01 3.06849957e-01
5.35975993e-01 -2.20938087e-01 -1.68705869e+00 5.51497281e-01
1.43789005e+00 3.94689478e-02 -1.07044093e-01 1.06410146e+00
7.65514791e-01 -5.31963825e-01 -8.12760830e-01 9.46359217e-01
1.47333160e-01 -1.02391434e+00 -5.35981953e-01 1.47492930e-01
1.51431665e-01 1.63713381e-01 1.37034059e-01 4.99467552e-01
1.92228183e-01 -6.91995800e-01 6.80029392e-01 4.56244707e-01
9.28117707e-02 -9.55673695e-01 8.39111447e-01 5.87372124e-01
-1.44538915e+00 -2.36832112e-01 -1.36248887e-01 -2.84845352e-01
6.58323228e-01 1.33141771e-01 -1.21823013e+00 -1.80870667e-01
4.88895059e-01 9.02679265e-01 -8.02208722e-01 1.20953810e+00
-2.14833304e-01 6.36554420e-01 -1.56109706e-01 -1.69534504e-01
5.23436069e-01 -1.36182815e-01 8.64322960e-01 1.38063502e+00
2.21470073e-01 -5.80819212e-02 1.93457752e-01 2.94167727e-01
7.41873682e-01 -2.00096920e-01 -1.23398793e+00 4.65518743e-01
7.27168769e-02 1.10928738e+00 -4.52096641e-01 -4.19173837e-01
-7.12476790e-01 5.04273474e-01 -2.14299485e-01 3.67292166e-01
-1.03636408e+00 -2.53031820e-01 1.21933043e+00 4.39299375e-01
5.53936064e-01 -5.30200362e-01 -2.57398218e-01 -6.00503147e-01
-1.01757266e-01 -1.19363457e-01 2.07340047e-01 -8.06329310e-01
-6.37493312e-01 4.24599856e-01 1.20684125e-01 -1.33728826e+00
-4.52672362e-01 -8.15301120e-01 -9.04074490e-01 5.26742458e-01
-1.84855688e+00 -1.17734206e+00 -5.15903473e-01 4.96496916e-01
9.76008475e-01 -4.13684219e-01 7.19421655e-02 4.21883643e-01
-9.03775275e-01 3.32018614e-01 -3.19243520e-01 2.31459722e-01
3.07581514e-01 -8.35755944e-01 7.69661486e-01 1.05227602e+00
-4.91452664e-01 4.97829355e-02 1.11192179e+00 -4.17819291e-01
-1.50476515e+00 -1.42773342e+00 1.41677356e+00 -7.05727637e-01
2.97208220e-01 -1.13199912e-01 -5.69408596e-01 7.45101035e-01
4.25799221e-01 2.62608286e-02 2.31016666e-01 -3.56691092e-01
1.59612641e-01 -1.58161059e-01 -8.57899129e-01 7.79259205e-01
7.45443225e-01 -1.81703866e-01 -5.66278458e-01 3.06713581e-01
1.31874382e-01 -8.06314796e-02 -1.83765694e-01 3.16612244e-01
4.52792406e-01 -1.07848036e+00 1.28592098e+00 -6.15823150e-01
-2.42891669e-01 -4.52896327e-01 -2.88846612e-01 -8.22418809e-01
-6.04863644e-01 -2.36695692e-01 -1.43818468e-01 7.53796697e-01
-4.01148051e-02 -6.63541973e-01 7.29383767e-01 7.36717999e-01
-2.48370022e-01 -2.47289062e-01 -1.18966830e+00 -5.66910505e-01
-2.82282233e-01 -1.05962574e+00 3.16705294e-02 1.00518286e-01
-3.22093397e-01 7.42524505e-01 -4.37532514e-01 3.64296734e-01
7.98267186e-01 -4.60929781e-01 1.16828620e+00 -1.03002644e+00
7.61095047e-01 -6.35539770e-01 -1.27010190e+00 -1.31185949e+00
5.41580319e-01 -3.82037520e-01 3.47845078e-01 -1.37548006e+00
-2.66882833e-02 2.21881539e-01 1.66867673e-01 -1.98064987e-02
-1.36833087e-01 2.13788614e-01 -1.50520444e-01 -2.26863608e-01
-6.70282543e-01 4.60338295e-01 7.99703181e-01 -3.17817360e-01
1.10905282e-01 7.31142998e-01 -3.62464964e-01 1.03801048e+00
5.96413553e-01 -6.92889914e-02 -1.67803913e-01 5.69575578e-02
-5.79939067e-01 9.81840715e-02 8.25342238e-01 -1.31764185e+00
5.48597515e-01 -2.15337560e-01 1.47862896e-01 -1.33557045e+00
6.56182110e-01 -8.74527991e-01 -3.48469496e-01 7.03814328e-01
-5.09025306e-02 1.54885337e-01 2.61044443e-01 5.50060987e-01
-4.24496010e-02 1.01343177e-01 1.17503977e+00 2.09619865e-01
-1.57637548e+00 2.05091223e-01 -1.30860424e+00 -9.19214040e-02
1.61093116e+00 -4.21969116e-01 -3.99416476e-01 -7.33146489e-01
-3.85966659e-01 4.09499884e-01 -2.16497146e-02 8.56424034e-01
8.62995148e-01 -1.54188251e+00 -8.48904073e-01 7.37983823e-01
1.45589948e-01 -6.55782104e-01 2.43361756e-01 1.19662774e+00
-4.18015093e-01 1.34414148e+00 -2.54805416e-01 -1.05643046e+00
-1.52448690e+00 6.50020301e-01 3.92557770e-01 4.45241719e-01
-6.07741475e-01 4.36958879e-01 2.23043486e-01 5.64405434e-02
7.68324807e-02 -4.68396842e-02 -5.90340078e-01 -2.88592577e-02
8.80619824e-01 8.10244143e-01 1.16532549e-01 -1.73424447e+00
-6.69657230e-01 5.69742978e-01 -2.48786494e-01 2.91176345e-02
5.78266323e-01 -5.84883690e-01 5.90418041e-01 1.88630939e-01
1.33608210e+00 -4.35919404e-01 -1.31417704e+00 -4.30600680e-02
9.29465424e-03 -7.17711091e-01 3.57018083e-01 -5.32855690e-02
-7.78771639e-01 1.07502484e+00 9.15108860e-01 3.54722366e-02
5.52874804e-01 -1.37241455e-02 1.05336571e+00 4.25138682e-01
3.07818234e-01 -1.18636036e+00 -3.13326776e-01 6.87540770e-01
5.19974649e-01 -1.91709101e+00 -2.82422036e-01 -3.24615151e-01
-1.07367551e+00 9.88241315e-01 6.85012519e-01 1.55321047e-01
7.73114920e-01 2.96511929e-02 2.08184674e-01 -8.89853165e-02
-8.68602455e-01 -9.00139570e-01 4.94422555e-01 9.61811364e-01
1.30000710e-01 1.09304331e-01 5.58503158e-02 1.14684934e-02
-1.74124837e-01 -3.63409489e-01 2.84777999e-01 1.01016474e+00
-1.08443165e+00 -4.07949001e-01 -4.86468971e-01 3.33077312e-01
8.20822939e-02 4.08305466e-01 -1.24668375e-01 7.21934497e-01
2.21259773e-01 1.59861088e+00 2.92316347e-01 -4.40515071e-01
6.35353982e-01 2.14018822e-01 -5.73172122e-02 -9.87820625e-02
2.16718793e-01 -1.93432152e-01 4.88310367e-01 -5.36875665e-01
-3.90448004e-01 -1.17618597e+00 -9.06916380e-01 -4.57353234e-01
1.27120838e-02 -2.66045958e-01 6.28962040e-01 1.21297157e+00
2.02248827e-01 2.82389194e-01 7.07737744e-01 -1.15018058e+00
1.57992139e-01 -7.54169822e-01 -2.81407654e-01 1.59926236e-01
6.54910505e-01 -9.41342831e-01 -3.36677790e-01 1.23235427e-01] | [7.7363176345825195, -0.6535366773605347] |
761da8da-a5e3-496e-8798-e908c234216d | dc-shadownet-single-image-hard-and-soft-1 | 2207.10434 | null | https://arxiv.org/abs/2207.10434v1 | https://arxiv.org/pdf/2207.10434v1.pdf | DC-ShadowNet: Single-Image Hard and Soft Shadow Removal Using Unsupervised Domain-Classifier Guided Network | Shadow removal from a single image is generally still an open problem. Most existing learning-based methods use supervised learning and require a large number of paired images (shadow and corresponding non-shadow images) for training. A recent unsupervised method, Mask-ShadowGAN, addresses this limitation. However, it requires a binary mask to represent shadow regions, making it inapplicable to soft shadows. To address the problem, in this paper, we propose an unsupervised domain-classifier guided shadow removal network, DC-ShadowNet. Specifically, we propose to integrate a shadow/shadow-free domain classifier into a generator and its discriminator, enabling them to focus on shadow regions. To train our network, we introduce novel losses based on physics-based shadow-free chromaticity, shadow-robust perceptual features, and boundary smoothness. Moreover, we show that our unsupervised network can be used for test-time training that further improves the results. Our experiments show that all these novel components allow our method to handle soft shadows, and also to perform better on hard shadows both quantitatively and qualitatively than the existing state-of-the-art shadow removal methods. | ['Robby T. Tan', 'Aashish Sharma', 'Yeying Jin'] | 2022-07-21 | dc-shadownet-single-image-hard-and-soft | http://openaccess.thecvf.com//content/ICCV2021/html/Jin_DC-ShadowNet_Single-Image_Hard_and_Soft_Shadow_Removal_Using_Unsupervised_Domain-Classifier_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Jin_DC-ShadowNet_Single-Image_Hard_and_Soft_Shadow_Removal_Using_Unsupervised_Domain-Classifier_ICCV_2021_paper.pdf | iccv-2021-1 | ['shadow-removal', 'image-shadow-removal'] | ['computer-vision', 'computer-vision'] | [ 7.29804695e-01 6.90888464e-02 -3.21964324e-02 -3.88497651e-01
-2.97480017e-01 -4.45812374e-01 4.03361112e-01 -3.68405074e-01
-7.99928093e-04 9.16846752e-01 -1.74132153e-01 -4.08479065e-01
2.90427148e-01 -8.52596879e-01 -6.15867913e-01 -1.01881111e+00
2.23195970e-01 3.75434875e-01 1.02094138e+00 -2.09513694e-01
3.59223709e-02 5.56413770e-01 -1.39747560e+00 4.16256189e-02
1.45084417e+00 9.63458598e-01 6.77618980e-01 4.63226378e-01
-1.01533651e-01 5.98952293e-01 -8.54801536e-01 1.05020598e-01
3.80575359e-01 -7.66258836e-01 -2.03279570e-01 2.88572232e-03
4.07055914e-01 -5.18104196e-01 -2.98843920e-01 6.96830630e-01
5.31715930e-01 2.64757663e-01 8.32289279e-01 -1.29607630e+00
-4.84242290e-01 -5.79966120e-02 -3.09490472e-01 -3.69644552e-01
7.53590837e-02 1.51983842e-01 7.33084857e-01 -6.54004037e-01
4.05843198e-01 1.04459333e+00 7.45464265e-01 3.12967777e-01
-1.04417503e+00 -8.04749608e-01 3.55252028e-02 2.28226840e-01
-1.08948815e+00 -2.36669779e-01 1.16672540e+00 1.15421331e-02
1.39857352e-01 3.70036036e-01 6.41653836e-01 1.11080003e+00
2.08193630e-01 8.37442338e-01 1.76905751e+00 -5.87121844e-01
4.03461903e-01 3.17794770e-01 -4.41672534e-01 1.07376361e+00
2.52543181e-01 4.18715000e-01 -4.10179049e-01 7.97526166e-02
7.21295059e-01 9.48412344e-03 -6.06147170e-01 -6.10631824e-01
-7.83083975e-01 6.70340359e-01 8.03447485e-01 1.15583856e-02
1.03659309e-01 2.31127758e-02 -2.08662733e-01 -4.87113707e-02
5.20352602e-01 3.60085994e-01 -2.74532050e-01 4.96098012e-01
-1.15679312e+00 -1.11145079e-01 9.92662728e-01 7.37748563e-01
1.15505838e+00 1.14560969e-01 -2.34951854e-01 7.62410045e-01
2.44092420e-01 1.01448786e+00 1.05564475e-01 -8.00886452e-01
-2.40923502e-02 4.77493316e-01 9.18388069e-02 -1.00530684e+00
-3.88624370e-01 -2.98921019e-01 -9.30742264e-01 6.63419545e-01
2.43823528e-01 -3.54383737e-02 -1.39413512e+00 1.52413058e+00
2.24318057e-01 4.40732747e-01 1.01192459e-01 8.53798687e-01
7.39332557e-01 5.63101053e-01 -4.18269396e-01 -1.97639078e-01
8.67371082e-01 -1.16340089e+00 -6.89963758e-01 -3.69069904e-01
1.16452783e-01 -7.97507107e-01 1.32489491e+00 3.13566476e-01
-5.67040741e-01 -4.25312489e-01 -1.25005805e+00 2.41112188e-01
-5.75347483e-01 2.12774172e-01 6.43733084e-01 8.01789701e-01
-8.25252533e-01 5.86139977e-01 -4.63858426e-01 -2.70694375e-01
4.02704984e-01 2.15170890e-01 1.96457773e-01 -1.66242316e-01
-1.13513863e+00 7.96136022e-01 2.64649540e-01 -1.05691263e-02
-9.44482744e-01 -4.34073001e-01 -7.49646366e-01 6.10280819e-02
8.08190763e-01 -3.05473298e-01 8.84627521e-01 -1.16046488e+00
-1.77456236e+00 4.16162878e-01 -2.17801601e-01 -1.72170386e-01
4.63148504e-01 -5.01527973e-02 -3.59264314e-01 2.70473272e-01
-1.79718081e-02 3.55466127e-01 1.22486055e+00 -2.18058395e+00
-4.41099226e-01 1.83935598e-01 1.29430309e-01 3.11054498e-01
-4.44560707e-01 -4.93079185e-01 -5.62707722e-01 -8.14136088e-01
3.27861495e-02 -1.05424118e+00 -1.77398883e-02 2.76705623e-01
-7.55739152e-01 2.29262814e-01 1.33563662e+00 -6.15488350e-01
8.15646350e-01 -2.17050815e+00 -1.42456084e-01 5.06227076e-01
8.63235071e-02 3.99299294e-01 -1.10545866e-01 2.16638923e-01
4.47779834e-01 -1.68376878e-01 -9.55544293e-01 -4.42347646e-01
-5.44253998e-02 5.86325586e-01 -3.29219460e-01 3.53574902e-01
4.01523896e-03 7.46159971e-01 -8.23202312e-01 -7.51729012e-01
3.73227894e-01 3.57678205e-01 -1.56613648e-01 4.40141469e-01
-4.05258119e-01 4.73944545e-01 -2.46359527e-01 9.03777957e-01
1.05674684e+00 6.78388551e-02 2.09414244e-01 -1.86771959e-01
-3.21162231e-02 -8.13348442e-02 -9.61180985e-01 1.25163484e+00
-7.01050520e-01 9.16999340e-01 2.06043959e-01 -6.64722502e-01
1.01493657e+00 -1.63825855e-01 2.86476195e-01 -7.13443339e-01
3.82025242e-02 3.17125499e-01 -1.73498034e-01 -2.99088895e-01
3.55973244e-01 -1.26071110e-01 1.40146017e-01 5.34987628e-01
-3.26727927e-01 -4.64240134e-01 -1.45767391e-01 1.09872483e-01
1.13266337e+00 2.73284107e-01 -6.48194030e-02 -1.96885958e-01
4.85149443e-01 -2.39075899e-01 7.73170590e-01 8.60900104e-01
-7.43264193e-03 9.25119519e-01 2.54029125e-01 1.82235867e-01
-5.90584338e-01 -1.25166047e+00 5.90118431e-02 1.09259927e+00
8.20514858e-01 -7.04580545e-02 -7.47680664e-01 -1.00664067e+00
1.77371502e-01 7.05926597e-01 -5.09985566e-01 -1.75415233e-01
-6.81584060e-01 -6.11247599e-01 3.43366623e-01 4.40341800e-01
8.94087911e-01 -1.12907469e+00 -6.66055441e-01 -2.35616758e-01
-1.60774007e-01 -1.14935565e+00 -3.63813609e-01 4.42644119e-01
-5.25904894e-01 -1.18574822e+00 -9.08390820e-01 -7.69727707e-01
8.57319832e-01 6.81266069e-01 9.22092974e-01 4.00124133e-01
-2.89081424e-01 2.38258585e-01 -6.18712962e-01 -5.59046149e-01
-4.45260674e-01 -7.22732693e-02 -1.62606046e-01 2.80295461e-01
-4.24590468e-01 -7.56342173e-01 -7.85727859e-01 6.43615842e-01
-1.06767619e+00 3.15447599e-01 1.07915175e+00 7.97365665e-01
4.60016996e-01 2.71834463e-01 1.75224632e-01 -1.17232275e+00
3.15698922e-01 -1.04881898e-01 -7.26469994e-01 3.67475063e-01
-8.87140274e-01 -1.78118167e-03 8.44506681e-01 -2.75424927e-01
-1.53987455e+00 2.54657060e-01 2.98537433e-01 -3.52649391e-01
-1.21302247e-01 -6.25151396e-02 -5.23564816e-01 -6.97871447e-01
6.36063576e-01 4.37318563e-01 -2.40400776e-01 -3.19966257e-01
3.54835361e-01 4.01985586e-01 6.01654291e-01 -3.63485575e-01
1.54761958e+00 8.98816049e-01 1.28173962e-01 -9.64395702e-01
-9.69104171e-01 -4.18687761e-01 -6.33132398e-01 -3.40371102e-01
6.03711724e-01 -5.08840621e-01 -3.36825788e-01 4.90682721e-01
-9.57852125e-01 -1.16599488e+00 -1.73550338e-01 -2.35571805e-02
-3.94500077e-01 5.53588986e-01 -3.05013023e-02 -8.77942204e-01
1.99267380e-02 -8.36337626e-01 1.03620195e+00 3.76971841e-01
4.46849912e-01 -1.01883936e+00 7.70997675e-03 1.41400948e-01
5.43763161e-01 4.05621320e-01 8.34935963e-01 -7.10927993e-02
-9.16664600e-01 2.49707565e-01 -6.26499712e-01 7.34794796e-01
4.50490981e-01 -2.54245549e-01 -1.12823045e+00 -1.10540524e-01
-1.25457674e-01 -3.01908761e-01 1.43281436e+00 2.19076291e-01
1.41233265e+00 -9.45273042e-02 -5.58670998e-01 8.73950005e-01
1.42732131e+00 2.63933893e-02 8.32779765e-01 9.88455042e-02
9.33767140e-01 5.43067515e-01 7.97529519e-01 2.12392986e-01
2.27276355e-01 6.58601403e-01 4.49404269e-01 -8.83325696e-01
-7.53016174e-01 -5.98337017e-02 3.18105161e-01 3.91744435e-01
-1.06737942e-01 -5.92920721e-01 -6.33497894e-01 2.88812071e-01
-1.72440207e+00 -5.63565373e-01 -1.76019982e-01 2.04109502e+00
8.61054122e-01 1.93696722e-01 -3.42194974e-01 9.78147313e-02
5.32084107e-01 3.77198130e-01 -5.76048493e-01 6.60127178e-02
-4.99335140e-01 4.95781362e-01 8.13561201e-01 6.98141217e-01
-1.02907538e+00 1.30275118e+00 5.78245306e+00 8.81682277e-01
-1.07245851e+00 8.03550929e-02 2.68294901e-01 3.79882038e-01
-5.36507607e-01 2.77901947e-01 -4.43876326e-01 4.93238926e-01
3.04617375e-01 4.47472304e-01 5.22413194e-01 7.06163943e-01
2.11584270e-01 -8.52090061e-01 -6.85719132e-01 6.88891888e-01
4.03526068e-01 -9.47782993e-01 -3.27276528e-01 -1.42167611e-02
1.06427705e+00 -3.17055970e-01 -5.47284968e-02 2.19528973e-01
2.36494377e-01 -9.05407846e-01 5.24147272e-01 5.83965659e-01
9.04674470e-01 -3.89400631e-01 7.26738453e-01 1.95794821e-01
-1.26050854e+00 -1.74391102e-02 -2.64168501e-01 2.78657019e-01
1.44688450e-02 9.87392545e-01 -9.88075972e-01 7.09161222e-01
5.28021336e-01 3.71866971e-01 -6.78013623e-01 1.02941251e+00
-7.09773183e-01 7.13298380e-01 -3.11476231e-01 1.12891510e-01
-1.15481138e-01 -1.62264749e-01 4.17608231e-01 1.33974314e+00
1.17445551e-01 9.01997238e-02 5.46250045e-01 8.08634281e-01
-6.53751343e-02 -1.98650882e-01 -5.77322364e-01 3.31683844e-01
3.90429080e-01 1.36529648e+00 -1.22002077e+00 -3.72553080e-01
-8.61986652e-02 1.53567338e+00 -8.16388577e-02 6.54148936e-01
-1.04147017e+00 -6.23541713e-01 2.64749289e-01 2.95465551e-02
2.55748481e-01 -2.77040869e-01 -4.41095054e-01 -9.49194908e-01
-8.55566710e-02 -6.31655335e-01 -2.02229872e-01 -8.20646048e-01
-1.03959548e+00 3.29275697e-01 1.13083730e-02 -1.17552960e+00
2.53625005e-01 -6.36721492e-01 -8.92590225e-01 5.90160072e-01
-2.17317772e+00 -1.30884707e+00 -1.11312985e+00 7.37447202e-01
3.22888225e-01 -5.33477962e-02 7.10065067e-01 1.20352998e-01
-3.03957969e-01 4.91008192e-01 3.29882979e-01 -1.03265641e-03
1.27790558e+00 -1.48078811e+00 1.11579616e-02 8.35309029e-01
-1.84820220e-01 5.36239557e-02 6.41331017e-01 -8.42574477e-01
-1.16147351e+00 -1.23069036e+00 3.56450826e-01 -1.85462192e-01
2.09011719e-01 -6.78202987e-01 -9.36796367e-01 9.37796533e-02
7.10792914e-02 -3.63281965e-02 4.12738174e-01 -2.14676067e-01
-2.59487927e-01 -5.23813188e-01 -1.11947620e+00 6.10966086e-01
1.14108038e+00 -4.08099383e-01 -2.86923379e-01 6.00682616e-01
6.40108287e-01 -4.43239510e-01 -1.04801454e-01 7.13551283e-01
4.72012579e-01 -1.40897822e+00 8.44408929e-01 4.01700556e-01
2.19955400e-01 -4.63311523e-01 -3.35575789e-02 -1.35332704e+00
-2.63315868e-02 -6.33158267e-01 -2.55426884e-01 1.28095901e+00
2.28755802e-01 -9.13284719e-01 8.94256175e-01 -1.82833999e-01
-5.06055772e-01 -6.72281265e-01 -5.60788214e-01 -1.07532525e+00
-3.59140605e-01 -2.32201263e-01 4.08105224e-01 5.99153578e-01
-6.96953535e-01 4.27927934e-02 -4.40175951e-01 4.02372390e-01
7.24269867e-01 6.06602728e-01 9.88998294e-01 -1.31571448e+00
-4.49184179e-01 -2.43493214e-01 1.68052450e-01 -1.07024920e+00
2.53270090e-01 -6.38220131e-01 9.35595691e-01 -1.69192052e+00
1.72879964e-01 -1.04462254e+00 -1.66613013e-01 7.05020547e-01
-1.98307350e-01 6.27196014e-01 1.80520579e-01 3.11812043e-01
-5.07353604e-01 8.80992055e-01 1.49416244e+00 -1.58228457e-01
-4.88545835e-01 1.81505620e-01 -4.02763039e-01 7.51282811e-01
8.98937285e-01 -4.41290408e-01 -3.88798535e-01 -6.38011703e-03
-3.71012598e-01 -3.75387788e-01 6.15505159e-01 -1.34825361e+00
-3.52102108e-02 -5.35515666e-01 5.16002417e-01 -5.48522830e-01
6.77128375e-01 -7.89857209e-01 -3.80632997e-01 4.86177474e-01
1.34683490e-01 -8.97491217e-01 4.70105447e-02 7.31759965e-01
1.48347124e-01 -9.87343639e-02 9.13366199e-01 6.73674867e-02
-6.64295614e-01 4.16433215e-02 -2.88688272e-01 -2.09446728e-01
9.46931303e-01 -2.40998924e-01 -3.81936580e-01 -5.72663426e-01
-1.65649340e-01 5.24632707e-02 7.66869009e-01 -2.44163666e-02
8.08444977e-01 -1.01033247e+00 -3.89279544e-01 2.66225427e-01
1.96332484e-03 1.60107687e-01 -5.34594692e-02 7.63096094e-01
-5.72715342e-01 7.55572915e-02 -9.14403200e-02 -5.38906455e-01
-1.23584282e+00 1.33771241e-01 2.11372361e-01 -6.90578148e-02
-6.94876194e-01 6.46054626e-01 6.92663074e-01 -4.00784999e-01
3.47866893e-01 -2.52063632e-01 2.73163170e-01 -3.12446982e-01
-1.54697085e-02 2.30868846e-01 -1.12915918e-01 -2.67399609e-01
-2.43643627e-01 6.05132818e-01 4.59225059e-01 -1.02017418e-01
1.03560913e+00 2.29749195e-02 -1.13466702e-01 3.08494419e-01
7.68290043e-01 4.75881100e-01 -1.64202237e+00 -1.42708480e-01
-3.21660638e-01 -5.45880020e-01 7.40821436e-02 -1.11512661e+00
-1.08793104e+00 6.31904304e-01 7.85769999e-01 2.51713723e-01
1.53289890e+00 2.32114997e-02 1.09240365e+00 4.70236987e-01
1.48685083e-01 -1.22005689e+00 5.36737740e-01 5.66731870e-01
8.20658982e-01 -1.35028577e+00 2.27571577e-01 -9.03798699e-01
-3.96012485e-01 1.05973971e+00 5.99953473e-01 -1.14158273e-01
5.79919577e-01 3.44208032e-01 1.71317771e-01 7.26266801e-02
-1.53908401e-03 -6.46061063e-01 5.00189662e-01 9.65320528e-01
-1.76493615e-01 1.49920046e-01 -5.90373427e-02 1.79492310e-01
-6.32264167e-02 -3.15724581e-01 4.79001164e-01 1.04076469e+00
-7.52740026e-01 -1.31422603e+00 -6.69233859e-01 3.63972068e-01
3.39091212e-01 -1.80471882e-01 -8.71562243e-01 8.27068448e-01
5.45371950e-01 9.78128731e-01 -3.00158709e-01 -4.46485251e-01
1.46306185e-02 -2.31552422e-01 5.05949080e-01 -6.60169303e-01
-9.45268106e-03 5.91664053e-02 -1.18790865e-01 -5.09862542e-01
-4.35600132e-01 -2.20914036e-01 -1.43041646e+00 -8.60091746e-02
-5.98023891e-01 -2.22694665e-01 6.59400463e-01 1.05988717e+00
2.16198992e-02 7.28876412e-01 9.39836800e-01 -1.16975951e+00
3.11614927e-02 -7.70123661e-01 -6.30487859e-01 1.14515662e-01
4.83767658e-01 -1.04285049e+00 -5.21206856e-01 5.27912714e-02] | [10.844452857971191, -4.102987766265869] |
27c11294-32e1-44bd-a191-3a5fd37479d7 | table-filling-multi-task-recurrent-neural | null | null | https://aclanthology.org/C16-1239 | https://aclanthology.org/C16-1239.pdf | Table Filling Multi-Task Recurrent Neural Network for Joint Entity and Relation Extraction | This paper proposes a novel context-aware joint entity and word-level relation extraction approach through semantic composition of words, introducing a Table Filling Multi-Task Recurrent Neural Network (TF-MTRNN) model that reduces the entity recognition and relation classification tasks to a table-filling problem and models their interdependencies. The proposed neural network architecture is capable of modeling multiple relation instances without knowing the corresponding relation arguments in a sentence. The experimental results show that a simple approach of piggybacking candidate entities to model the label dependencies from relations to entities improves performance. We present state-of-the-art results with improvements of 2.0{\%} and 2.7{\%} for entity recognition and relation classification, respectively on CoNLL04 dataset. | ['Hinrich Sch{\\"u}tze', 'Bernt Andrassy', 'Pankaj Gupta'] | 2016-12-01 | table-filling-multi-task-recurrent-neural-1 | https://aclanthology.org/C16-1239 | https://aclanthology.org/C16-1239.pdf | coling-2016-12 | ['joint-entity-and-relation-extraction'] | ['natural-language-processing'] | [ 2.84384489e-01 5.71376920e-01 -3.61242533e-01 -5.28275609e-01
-7.59370923e-01 -3.22788984e-01 4.66349661e-01 7.94375539e-01
-6.40114009e-01 1.11096430e+00 1.32220134e-01 -7.25282609e-01
-1.76432312e-01 -1.18799412e+00 -6.61707878e-01 -1.73586130e-01
-1.70894176e-01 8.57824206e-01 7.98951983e-02 -4.61770773e-01
-1.53345212e-01 4.54788148e-01 -1.32060051e+00 7.53369451e-01
7.36871004e-01 1.03779685e+00 -1.07829437e-01 5.76431692e-01
-6.67512476e-01 1.28438246e+00 -8.53442490e-01 -6.92014039e-01
-1.65890813e-01 -1.47619098e-01 -1.38769019e+00 -3.01760852e-01
4.41132225e-02 2.13090703e-01 -2.94126093e-01 7.15792358e-01
2.15926543e-01 3.45806360e-01 6.65038168e-01 -9.26288307e-01
-8.89100492e-01 1.16511929e+00 -4.75708783e-01 3.13883603e-01
2.61120260e-01 -8.85149717e-01 1.40953064e+00 -7.51136422e-01
8.66936326e-01 1.19336665e+00 4.62747723e-01 2.15956271e-01
-8.72432888e-01 -5.66815853e-01 2.87757069e-01 5.13438284e-01
-1.48470509e+00 -2.46119156e-01 3.78876656e-01 -9.67346355e-02
2.23667550e+00 2.38648102e-01 -3.91911194e-02 7.61520863e-01
1.12714767e-01 4.84027773e-01 7.94094503e-01 -7.89232731e-01
-2.95965791e-01 2.09406480e-01 7.76413321e-01 5.24949849e-01
5.64122736e-01 -3.86005521e-01 -4.61584330e-01 -4.00312729e-02
4.59703922e-01 -4.41657335e-01 9.71408263e-02 1.72813833e-01
-8.13342810e-01 6.19303167e-01 5.16755939e-01 7.06349611e-01
-4.97864366e-01 -3.34052257e-02 6.78648531e-01 2.25151360e-01
7.05009937e-01 3.37793648e-01 -1.06153953e+00 2.77001500e-01
-3.32172900e-01 7.99055845e-02 1.05750024e+00 1.37450254e+00
6.83978021e-01 -9.71245840e-02 -4.07378733e-01 9.75923181e-01
1.32643893e-01 1.94117472e-01 2.86965042e-01 -1.84048414e-01
1.01281452e+00 1.01736546e+00 -2.84065474e-02 -1.01370561e+00
-5.93364596e-01 -6.32542312e-01 -7.68160403e-01 -4.69273776e-01
-1.56193133e-02 -3.08801711e-01 -9.55429077e-01 1.54314005e+00
3.34778368e-01 -5.84757701e-02 7.34786570e-01 1.74017265e-01
1.44740617e+00 8.46986830e-01 5.78853011e-01 -4.49352950e-01
1.83175969e+00 -1.09611666e+00 -1.30297446e+00 -3.37071985e-01
1.10509825e+00 -5.47903597e-01 4.21337515e-01 -8.25484246e-02
-9.25824404e-01 -4.84569073e-01 -1.03975499e+00 -3.85700792e-01
-1.05919623e+00 5.00899613e-01 9.94562566e-01 3.59825611e-01
-6.56630397e-01 5.98356128e-01 -5.52748263e-01 -2.34298065e-01
2.66213119e-01 7.54620254e-01 -6.85333550e-01 3.62285465e-01
-1.61770606e+00 1.36582518e+00 1.08213067e+00 5.15876651e-01
-1.99477032e-01 -3.03218156e-01 -1.19944191e+00 2.54009157e-01
7.75646329e-01 -3.20900768e-01 9.74988520e-01 -1.76035658e-01
-1.04775810e+00 9.37334001e-01 -5.73664784e-01 -9.33074355e-01
-3.20957571e-01 -5.80131054e-01 -9.42204893e-01 -2.77189881e-01
1.42963052e-01 2.19735861e-01 1.01274101e-03 -1.08799446e+00
-8.02778304e-01 -1.39265850e-01 -1.52731642e-01 3.01282555e-01
-1.39561921e-01 5.22409081e-01 -2.56939083e-01 -4.70237523e-01
1.36597410e-01 -5.89606762e-01 -2.09856480e-01 -9.18442070e-01
-6.79328740e-01 -7.60451853e-01 5.14360785e-01 -9.95367706e-01
1.55387795e+00 -1.58705652e+00 -1.18857855e-03 9.55913961e-02
-2.45889369e-02 4.79839653e-01 -2.16857996e-02 5.78470349e-01
-5.24359703e-01 3.07917804e-01 -8.72819498e-02 -3.37042868e-01
-2.67628044e-01 4.24265146e-01 -1.90794632e-01 -1.17062852e-01
6.13281667e-01 1.19609427e+00 -3.86260122e-01 -4.44434226e-01
-6.56961873e-02 5.62736273e-01 6.50394037e-02 1.87552482e-01
-1.54036328e-01 -1.60253197e-01 -3.05465430e-01 5.45187354e-01
5.42749822e-01 -3.16847742e-01 7.14609563e-01 -4.75242466e-01
1.59157559e-01 9.27390277e-01 -1.27637637e+00 1.34740090e+00
-7.14187026e-01 3.74773204e-01 -4.74688947e-01 -1.06485665e+00
1.30035496e+00 6.72051966e-01 1.59313723e-01 -5.24386942e-01
3.14849406e-01 2.29671672e-01 1.49837688e-01 -4.16459411e-01
5.64521492e-01 5.00164973e-03 -2.51797616e-01 -8.93297419e-03
4.92430955e-01 5.98197937e-01 4.27589118e-01 5.79504371e-02
1.00245976e+00 3.74398857e-01 7.33210742e-01 -1.11711107e-01
8.65185380e-01 -9.07543749e-02 8.03948522e-01 5.21592081e-01
4.20120627e-01 8.25963840e-02 5.63464344e-01 -6.40984356e-01
-7.83837378e-01 -6.46190047e-01 3.85761331e-03 1.15160775e+00
-2.16651917e-01 -5.44656813e-01 -4.53248143e-01 -8.37435305e-01
-1.66374788e-01 1.01700401e+00 -7.31319010e-01 -1.07657788e-02
-9.32294011e-01 -1.06660450e+00 5.51990092e-01 6.93154812e-01
5.23792446e-01 -1.41283548e+00 -2.05755472e-01 6.46975994e-01
-4.25512314e-01 -1.74035430e+00 2.10240930e-01 9.08955395e-01
-6.02783322e-01 -8.62883747e-01 -1.57962561e-01 -1.08279228e+00
4.49448615e-01 -3.30638647e-01 1.40726244e+00 -3.91439982e-02
-1.15203336e-02 -6.50292218e-01 -5.59414566e-01 -3.16311508e-01
-2.85857230e-01 5.63490033e-01 -1.98682532e-01 -1.24507509e-01
7.38166392e-01 -5.34586668e-01 2.51433045e-01 -1.00883789e-01
-4.41282809e-01 -9.28996280e-02 7.30592012e-01 7.39046812e-01
6.40664399e-01 2.13862304e-02 7.44430423e-01 -1.68309820e+00
6.08449519e-01 -4.30676967e-01 -3.50617707e-01 8.44811141e-01
-8.07503760e-01 4.21605974e-01 3.51802796e-01 -2.02059552e-01
-1.36286807e+00 1.12650797e-01 -1.33216783e-01 6.79725781e-02
-3.77662331e-01 7.07355440e-01 -3.13900411e-01 4.46565330e-01
4.92581159e-01 -1.77928686e-01 -8.20457816e-01 -5.43112457e-01
5.28779626e-01 5.74408650e-01 4.93814081e-01 -4.90287095e-01
4.39452738e-01 -1.86211511e-01 1.24969296e-01 -3.11417162e-01
-1.28887963e+00 -4.91051853e-01 -8.95016968e-01 3.83360386e-01
1.10638380e+00 -1.02732670e+00 -9.38955426e-01 5.68854548e-02
-1.61150968e+00 8.53417739e-02 -2.79425353e-01 3.62753749e-01
9.55786183e-03 9.67161823e-03 -1.04390013e+00 -8.19240570e-01
-7.81127810e-01 -6.40789509e-01 8.40991080e-01 2.65321553e-01
-3.13807786e-01 -9.17790830e-01 -1.34541526e-01 5.18802345e-01
1.06436923e-01 2.17621505e-01 1.35617316e+00 -1.42226744e+00
-4.51434612e-01 -2.90912628e-01 -5.14990687e-01 -9.77884457e-02
3.14667732e-01 -2.96137720e-01 -8.89358699e-01 3.20995271e-01
-4.32305753e-01 -1.99750587e-01 8.93298805e-01 3.58830392e-02
6.78696156e-01 -4.06491667e-01 -6.29689813e-01 1.95611849e-01
1.46412027e+00 6.40855134e-01 6.26585245e-01 3.68909508e-01
9.68487561e-01 9.04362381e-01 7.15111911e-01 2.68698130e-02
6.63918197e-01 6.60119653e-01 8.39702040e-02 -6.70368597e-02
-1.36654466e-01 -2.51020104e-01 -2.42573097e-01 7.83854127e-01
-2.14884296e-01 -7.74587512e-01 -1.03441107e+00 5.82788467e-01
-2.02134109e+00 -5.82782626e-01 -4.95089710e-01 1.69343376e+00
1.10119689e+00 3.55380058e-01 -2.92374104e-01 1.73977286e-01
1.00872493e+00 -1.53594371e-02 -1.36214951e-02 -7.72768736e-01
-4.10551459e-01 6.87871814e-01 5.89314878e-01 8.00610662e-01
-1.43495548e+00 1.62871861e+00 5.38119364e+00 8.77052903e-01
-4.36241895e-01 8.75103772e-02 7.46088147e-01 5.57822227e-01
-2.43567042e-02 1.21158578e-01 -1.45938742e+00 -3.17556888e-01
1.33130944e+00 7.88069963e-02 5.83404163e-03 6.33010864e-01
-3.51699680e-01 -2.39779409e-02 -9.33871925e-01 7.57800758e-01
-2.56565344e-02 -1.40662253e+00 3.28643620e-01 -2.32309923e-01
3.10452640e-01 -2.65817940e-01 -4.87974256e-01 6.26068830e-01
3.71528745e-01 -1.28089786e+00 2.25250095e-01 3.98832470e-01
6.06833339e-01 -1.04568863e+00 1.34746456e+00 5.19862734e-02
-1.58475745e+00 2.81273872e-01 -2.03252345e-01 -1.94289133e-01
3.20880353e-01 5.78321576e-01 -1.06221581e+00 1.11064363e+00
5.67909241e-01 4.75530565e-01 -6.41621113e-01 3.17447364e-01
-5.25012314e-01 3.27501237e-01 -2.48198777e-01 -2.59356320e-01
-1.13754116e-01 -2.89927516e-02 5.01301102e-02 1.54302633e+00
-8.55362937e-02 2.95533001e-01 -5.65105714e-02 6.04480088e-01
-3.85740370e-01 5.05953074e-01 -4.40346837e-01 -6.24568276e-02
4.77642655e-01 1.44666195e+00 -1.06072402e+00 -5.11976480e-01
-3.42474520e-01 8.79764080e-01 8.21132660e-01 2.70525217e-01
-5.98019719e-01 -7.91355550e-01 2.68426418e-01 -5.61643243e-01
6.63685024e-01 -7.08818808e-02 -4.34763700e-01 -9.84784842e-01
3.38556588e-01 -4.09255803e-01 5.98125458e-01 -4.93659675e-01
-9.46191549e-01 1.32933605e+00 7.56417289e-02 -6.13278031e-01
-5.02563179e-01 -4.93055940e-01 -2.31120616e-01 1.06077385e+00
-1.43555474e+00 -1.44418919e+00 1.42431825e-01 8.46097395e-02
3.21214885e-01 -1.65546760e-01 1.36040306e+00 5.54275393e-01
-8.11111927e-01 7.48153567e-01 -5.11867702e-01 7.15903640e-01
2.92680532e-01 -1.21381736e+00 8.60233963e-01 8.29355121e-01
6.18525088e-01 6.59667552e-01 4.99689579e-01 -8.47211957e-01
-8.53593707e-01 -1.10474765e+00 2.01928401e+00 -3.08405817e-01
5.49566209e-01 -5.55285454e-01 -1.03364885e+00 9.71988201e-01
4.60951865e-01 1.70273066e-01 7.15284884e-01 6.89267576e-01
-5.44549286e-01 -5.03606834e-02 -1.05755794e+00 3.98488551e-01
1.03855574e+00 -6.83036029e-01 -6.57737076e-01 4.46934342e-01
1.12827516e+00 -4.05335128e-01 -1.19710183e+00 9.80767965e-01
1.18334040e-01 -2.33619228e-01 8.63313317e-01 -8.75425577e-01
2.12121919e-01 -2.07125008e-01 -1.30953252e-01 -8.63494873e-01
-1.34944603e-01 -3.19458455e-01 -5.92321634e-01 1.69801843e+00
1.37585270e+00 -4.62655395e-01 8.57351422e-01 5.40966690e-01
1.16738126e-01 -1.02468014e+00 -8.48458827e-01 -4.84018266e-01
-2.17071190e-01 -2.79259413e-01 5.07081568e-01 1.06595421e+00
-4.45751548e-02 1.19238365e+00 -3.09916675e-01 2.90920109e-01
7.65361935e-02 2.44464576e-01 5.19071370e-02 -1.05509913e+00
-2.63903826e-01 -1.42486282e-02 -2.48500973e-01 -5.54269552e-01
5.24877787e-01 -9.33540821e-01 -1.93710670e-01 -1.87373865e+00
-2.10533440e-01 -4.98808801e-01 -5.37326336e-01 8.85986328e-01
-3.35055828e-01 -1.06908724e-01 3.08245141e-02 -2.43465424e-01
-5.22734761e-01 3.27397376e-01 6.04556143e-01 -7.48041412e-03
-2.98696607e-01 1.76497153e-03 -5.76919734e-01 3.89726013e-01
7.77439892e-01 -8.06913435e-01 -2.99977928e-01 -3.13811511e-01
4.99735504e-01 3.25961590e-01 -2.44908303e-01 -7.39376307e-01
2.75532842e-01 1.55276731e-01 3.13367695e-01 -8.24624479e-01
4.05753255e-01 -7.37990141e-01 7.09301159e-02 2.69100547e-01
-7.06496477e-01 1.89757854e-01 3.83524865e-01 3.50724399e-01
-4.59182829e-01 -3.43852460e-01 1.76377147e-01 -1.42637104e-01
-6.83059394e-01 -1.92716233e-02 -2.22606272e-01 3.50191519e-02
9.78790700e-01 2.20429838e-01 -4.03979957e-01 -4.56523821e-02
-1.15409851e+00 1.66568726e-01 -6.80123448e-01 5.47322512e-01
4.06996101e-01 -1.16691971e+00 -7.85480618e-01 -9.03741419e-02
5.70026785e-02 1.99633047e-01 -5.35368919e-04 1.15851767e-01
-4.18235719e-01 7.49488294e-01 7.46205002e-02 1.11305922e-01
-1.47014153e+00 5.34953654e-01 3.50021958e-01 -1.05134010e+00
-2.91897953e-01 1.27242124e+00 -3.20598364e-01 -6.97728992e-01
2.88644165e-01 -5.47103167e-01 -9.85135972e-01 2.58871496e-01
2.12741345e-01 1.07835367e-01 5.54596722e-01 -6.92570031e-01
-6.00066602e-01 2.62590617e-01 -5.48210919e-01 -5.85322641e-03
1.31420445e+00 1.40797049e-01 -3.98079067e-01 5.50430417e-01
1.03831148e+00 -2.33334959e-01 -2.35853538e-01 -4.98769611e-01
7.57372797e-01 3.22650135e-01 -2.76362419e-01 -1.14229715e+00
-7.95006216e-01 6.08591914e-01 2.06228450e-01 1.36094809e-01
9.51267660e-01 1.76442429e-01 9.00213897e-01 8.76526535e-01
2.18372270e-01 -9.22039509e-01 -6.06801271e-01 9.11565542e-01
6.42388880e-01 -9.98986483e-01 1.77330390e-01 -1.14140451e+00
-6.37627006e-01 1.03116751e+00 7.47725189e-01 -5.41634150e-02
6.78591132e-01 5.62112391e-01 -6.14722706e-02 -1.85172275e-01
-1.05879033e+00 -5.51179826e-01 4.33622420e-01 3.60847831e-01
9.23510253e-01 1.86572924e-01 -5.60020447e-01 7.31309175e-01
-8.94885883e-02 -3.94681007e-01 2.17121661e-01 9.07121301e-01
-2.16073886e-01 -1.62858093e+00 2.10634813e-01 4.42274839e-01
-8.62667203e-01 -5.63747942e-01 -4.44941491e-01 7.30501294e-01
2.88511902e-01 1.00892794e+00 3.66937593e-02 -5.09537637e-01
6.35175169e-01 4.52133089e-01 1.52355045e-01 -9.89816725e-01
-1.13449681e+00 -5.72348237e-02 1.06680977e+00 -1.76845089e-01
-6.01548314e-01 -3.97396982e-01 -1.56583822e+00 3.04190189e-01
-7.61366248e-01 4.85991448e-01 4.72208798e-01 1.28974223e+00
4.99368429e-01 1.03682721e+00 1.86685666e-01 -4.59419414e-02
3.43960486e-02 -1.31236875e+00 -2.72601873e-01 1.72234163e-01
5.15164770e-02 -5.73444784e-01 1.83699608e-01 -5.82195669e-02] | [9.313231468200684, 8.763917922973633] |
dec73aa0-33c4-4736-bdbb-6a3a1e08257d | distant-domain-transfer-learning-for-medical | 2012.06346 | null | https://arxiv.org/abs/2012.06346v1 | https://arxiv.org/pdf/2012.06346v1.pdf | Distant Domain Transfer Learning for Medical Imaging | Medical image processing is one of the most important topics in the field of the Internet of Medical Things (IoMT). Recently, deep learning methods have carried out state-of-the-art performances on medical image tasks. However, conventional deep learning have two main drawbacks: 1) insufficient training data and 2) the domain mismatch between the training data and the testing data. In this paper, we propose a distant domain transfer learning (DDTL) method for medical image classification. Moreover, we apply our methods to a recent issue (Coronavirus diagnose). Several current studies indicate that lung Computed Tomography (CT) images can be used for a fast and accurate COVID-19 diagnosis. However, the well-labeled training data cannot be easily accessed due to the novelty of the disease and a number of privacy policies. Moreover, the proposed method has two components: Reduced-size Unet Segmentation model and Distant Feature Fusion (DFF) classification model. It is related to a not well-investigated but important transfer learning problem, termed Distant Domain Transfer Learning (DDTL). DDTL aims to make efficient transfers even when the domains or the tasks are entirely different. In this study, we develop a DDTL model for COVID-19 diagnose using unlabeled Office-31, Catech-256, and chest X-ray image data sets as the source data, and a small set of COVID-19 lung CT as the target data. The main contributions of this study: 1) the proposed method benefits from unlabeled data collected from distant domains which can be easily accessed, 2) it can effectively handle the distribution shift between the training data and the testing data, 3) it has achieved 96\% classification accuracy, which is 13\% higher classification accuracy than "non-transfer" algorithms, and 8\% higher than existing transfer and distant transfer algorithms. | ['Houbing Song', 'Jian Wang', 'Yongxin Liu', 'Meryl Liu', 'Shuteng Niu'] | 2020-12-10 | null | null | null | null | ['unet-segmentation'] | ['computer-vision'] | [ 1.68665320e-01 -2.76486456e-01 -3.37418646e-01 -3.56086552e-01
-9.00308430e-01 -2.64134645e-01 1.79289609e-01 -1.22885192e-02
-5.59388161e-01 9.18706477e-01 -1.32765859e-01 -4.89167571e-01
-2.14150071e-01 -8.07324886e-01 -5.46534777e-01 -8.39553714e-01
2.74582267e-01 9.54156339e-01 3.93253326e-01 9.30935070e-02
-2.81046212e-01 3.09903473e-01 -1.06755495e+00 5.29830635e-01
1.08034563e+00 1.24140632e+00 3.42648953e-01 3.02140862e-01
-1.07513063e-01 5.64488471e-01 -4.75864172e-01 -1.95796102e-01
2.44156346e-01 -6.77526295e-01 -1.00470877e+00 4.83606346e-02
-5.68927303e-02 -6.13109231e-01 -3.99904922e-02 9.40174341e-01
8.29939365e-01 -2.87216038e-01 1.04881442e+00 -1.24578822e+00
-6.99300289e-01 -9.38604958e-03 -5.74842930e-01 3.16149205e-01
-1.18435137e-01 -1.62792988e-02 1.81229979e-01 -5.49539447e-01
5.66237390e-01 8.64713728e-01 7.36023366e-01 7.71368504e-01
-6.52721703e-01 -1.06360137e+00 -3.85978103e-01 2.53561288e-01
-1.28702354e+00 1.79504454e-01 4.48464721e-01 -6.33266687e-01
2.40732744e-01 -7.54163116e-02 4.16227728e-01 1.19021034e+00
4.77998078e-01 6.77784443e-01 1.36363780e+00 -8.69204327e-02
1.40150681e-01 5.19828558e-01 -1.10286348e-01 6.66369140e-01
5.93379214e-02 2.93042627e-03 2.72421420e-01 -3.03972155e-01
8.91303241e-01 4.18863326e-01 -3.02846104e-01 -2.89502650e-01
-1.18641603e+00 9.48582709e-01 6.26453936e-01 7.10191309e-01
-2.27359578e-01 -4.78371918e-01 4.50032115e-01 3.97046983e-01
6.74230933e-01 -2.71783978e-01 -7.37962902e-01 1.98935464e-01
-4.82280850e-01 -1.51231602e-01 7.33674943e-01 7.91261256e-01
4.90784585e-01 -3.34166348e-01 -1.24193758e-01 6.90751553e-01
2.87291825e-01 6.94155931e-01 9.34724271e-01 -2.93575853e-01
7.07651615e-01 4.73581165e-01 -1.63937032e-01 -7.20592260e-01
-2.84366786e-01 -3.89664143e-01 -1.19886494e+00 -1.01926796e-01
3.97475183e-01 -4.60619658e-01 -9.50619280e-01 1.43132973e+00
4.50431436e-01 2.56569296e-01 2.54745781e-01 7.63456523e-01
9.97660458e-01 6.73406005e-01 3.43085289e-01 -1.67263299e-01
1.53960025e+00 -7.39040613e-01 -6.81679845e-01 1.07627504e-01
8.41396093e-01 -6.74984157e-01 6.99140608e-01 1.32385299e-01
-7.21823752e-01 -6.46098137e-01 -7.26010263e-01 1.47213697e-01
-4.07587081e-01 1.94851607e-01 2.37079203e-01 5.44308662e-01
-7.44529963e-01 3.40827495e-01 -6.08135641e-01 -4.98679519e-01
8.55782330e-01 4.15852755e-01 -3.97828847e-01 -5.33550322e-01
-1.26910520e+00 7.43472099e-01 3.24898452e-01 -4.27992880e-01
-7.26326585e-01 -6.32133305e-01 -4.19186562e-01 -5.87825328e-02
2.73882478e-01 -9.90147769e-01 1.08516657e+00 -1.01824546e+00
-1.18772960e+00 1.16609240e+00 1.42895728e-01 -2.46382460e-01
7.69175589e-01 1.22508593e-02 -6.24060214e-01 3.21550578e-01
3.06244582e-01 5.67164004e-01 8.64485323e-01 -1.07355702e+00
-5.79962075e-01 -6.50230646e-01 -6.62729442e-01 1.39369592e-01
-5.60340762e-01 -1.00288391e-01 -3.27607721e-01 -7.40396500e-01
-2.47539029e-01 -1.02873003e+00 -7.15099871e-02 2.07949013e-01
-4.62779887e-02 -4.42710370e-01 1.11517608e+00 -7.64258146e-01
8.43683600e-01 -2.50336456e+00 -1.99855551e-01 -5.44812009e-02
2.90754497e-01 8.44805717e-01 -5.21156117e-02 1.01314902e-01
-1.74021319e-01 1.73172519e-01 -3.13494414e-01 1.29248366e-01
-3.07218343e-01 2.04545781e-01 1.38071790e-01 3.22501600e-01
-3.83020230e-02 9.97908771e-01 -6.03406191e-01 -9.35860813e-01
1.24113813e-01 2.88578987e-01 -2.90827602e-01 3.69189411e-01
-8.56089406e-03 1.10606742e+00 -8.51380467e-01 4.78770643e-01
8.14162433e-01 -5.69704771e-01 -8.36356208e-02 -1.58092543e-01
3.75191718e-01 -3.93288374e-01 -7.21908629e-01 1.46316731e+00
-1.95988879e-01 -5.55347698e-03 1.11864265e-02 -1.31587529e+00
9.35452640e-01 8.03642571e-01 8.65082502e-01 -8.38438869e-01
5.52922249e-01 4.31172580e-01 7.61271920e-03 -9.19916511e-01
-5.50686598e-01 -6.33457780e-01 -1.91667490e-02 3.56502444e-01
-7.33959228e-02 -5.11825457e-02 -4.06079948e-01 -3.59307021e-01
8.78246605e-01 -3.94313395e-01 3.30254823e-01 -1.63624376e-01
5.27024746e-01 4.82219905e-02 7.91128159e-01 3.51204306e-01
-6.21645927e-01 5.80587387e-01 1.39560401e-01 -4.86338019e-01
-6.82478130e-01 -1.09459805e+00 -5.66910744e-01 6.91743135e-01
1.13069348e-01 3.42798948e-01 -8.65258157e-01 -1.14232993e+00
7.99566228e-03 2.58866370e-01 -5.11866689e-01 -1.71564907e-01
-4.60621744e-01 -7.11072743e-01 6.09674573e-01 6.75013185e-01
1.07422245e+00 -1.09696758e+00 -2.27671221e-01 7.57310092e-02
-4.95519191e-01 -1.05905962e+00 -4.48324293e-01 3.66861261e-02
-1.23571026e+00 -1.15163958e+00 -1.24726391e+00 -1.28620875e+00
6.38691008e-01 1.81482956e-01 7.90298343e-01 -7.21151829e-02
-2.61197299e-01 2.15523139e-01 -3.23716909e-01 -6.70015693e-01
-4.78376865e-01 1.44365221e-01 -1.70571551e-01 5.86775094e-02
6.68649971e-01 -2.21283570e-01 -7.38896966e-01 6.44115329e-01
-9.77724552e-01 -1.58705220e-01 9.44047868e-01 1.02380204e+00
7.12012112e-01 1.36713922e-01 9.28836405e-01 -1.22969055e+00
5.05620122e-01 -9.18928683e-01 -2.03482226e-01 2.53997803e-01
-6.95062518e-01 -3.41189653e-01 6.18695557e-01 -3.30541223e-01
-1.24184906e+00 -1.84626594e-01 -3.22993577e-01 -5.36360502e-01
-5.21129370e-01 3.04975003e-01 -1.14200503e-01 1.94120929e-01
6.90172195e-01 1.17632292e-01 2.82369584e-01 -4.78987426e-01
-1.31714135e-01 1.17446077e+00 1.71821818e-01 -4.25537586e-01
6.38766408e-01 5.35313308e-01 -1.34687036e-01 -5.12127340e-01
-7.43195415e-01 -5.95633447e-01 -6.69584811e-01 2.09232092e-01
1.38824880e+00 -1.01983178e+00 -3.67204279e-01 6.95915341e-01
-8.64359200e-01 -1.58326536e-01 -2.50600517e-01 9.04790223e-01
-3.95975024e-01 3.70361626e-01 -7.94935405e-01 -1.93320781e-01
-5.04020810e-01 -1.25648594e+00 9.67648923e-01 9.69331935e-02
1.19600676e-01 -1.17556930e+00 -6.50789812e-02 8.11152518e-01
4.69039559e-01 3.26402396e-01 1.32759678e+00 -1.00788164e+00
-3.51129472e-01 -1.70838237e-01 -4.41240549e-01 5.65021813e-01
4.88867760e-01 -6.40069425e-01 -7.10077226e-01 -5.90982616e-01
4.09681410e-01 -5.48815489e-01 6.04068935e-01 4.55988228e-01
1.38382900e+00 -6.57217801e-02 -7.22662330e-01 5.68295300e-01
1.25910735e+00 6.95927262e-01 5.83573103e-01 1.39040560e-01
5.57757676e-01 5.88660002e-01 7.93307722e-01 2.49240115e-01
2.12863773e-01 3.62308174e-01 2.38023147e-01 -6.55740857e-01
-1.28054813e-01 -8.81219357e-02 -1.04749687e-01 9.99315202e-01
1.02190204e-01 -2.33743891e-01 -1.06098568e+00 5.87772191e-01
-1.67238379e+00 -5.66613674e-01 -2.45699912e-01 1.97441411e+00
7.40159571e-01 -2.29698524e-01 1.05403317e-02 -1.15842864e-01
9.24137354e-01 -4.23295081e-01 -8.26046228e-01 -5.01505621e-02
3.65269870e-01 4.11696970e-01 2.81806678e-01 -1.66007057e-01
-1.23326123e+00 4.56792027e-01 5.18378830e+00 1.07096958e+00
-1.27310956e+00 5.23620903e-01 9.02092278e-01 4.22713548e-01
5.69617674e-02 -6.08055830e-01 -5.78463852e-01 7.13511705e-01
8.21631730e-01 9.41633582e-02 -1.35274529e-01 8.43401194e-01
-9.68971997e-02 3.00497919e-01 -9.87650037e-01 1.10614622e+00
3.57958488e-02 -9.39780831e-01 8.66489485e-02 2.95098364e-01
6.87748671e-01 3.00643355e-01 3.70219767e-01 4.68075752e-01
2.12813052e-03 -9.93792534e-01 -6.85410723e-02 1.74744159e-01
1.29637897e+00 -6.27281427e-01 1.22284448e+00 6.66146755e-01
-1.00877345e+00 -9.57120135e-02 -3.39962423e-01 3.32736373e-01
-1.70322239e-01 5.56787312e-01 -1.08790314e+00 8.21098566e-01
1.04405701e+00 7.01459229e-01 -2.92111218e-01 8.25431645e-01
1.64111957e-01 5.74430883e-01 -1.56472716e-02 7.66906962e-02
1.72230944e-01 -2.48871297e-01 2.62177307e-02 9.06875551e-01
4.90294188e-01 2.10537270e-01 3.58656704e-01 5.93374252e-01
-2.39616826e-01 2.75463849e-01 -8.87390316e-01 3.06205809e-01
2.29099140e-01 8.58562827e-01 -7.02111244e-01 -4.24541831e-01
-5.80191493e-01 1.00460482e+00 -8.40155929e-02 2.22858354e-01
-9.78288710e-01 -1.38545245e-01 1.96319863e-01 5.23348376e-02
4.90786582e-01 4.11440104e-01 -4.74184528e-02 -1.19964731e+00
-9.86309648e-02 -9.28195298e-01 7.48407960e-01 -5.35482645e-01
-1.90297425e+00 7.30337322e-01 7.74309263e-02 -1.66531873e+00
-1.03794605e-01 -6.48428977e-01 -3.55333328e-01 8.47930670e-01
-1.75964010e+00 -1.15235615e+00 -3.44846815e-01 1.26882350e+00
3.28773141e-01 -3.33720177e-01 9.69314218e-01 7.40836203e-01
-1.45077139e-01 5.12622416e-01 5.61881661e-01 5.11010230e-01
1.02656996e+00 -9.45269227e-01 -2.67967224e-01 1.58244148e-01
-3.59415352e-01 1.93840697e-01 -1.53274685e-01 -5.34787297e-01
-7.24999368e-01 -1.51681638e+00 7.27248669e-01 -3.13450426e-01
1.87100247e-01 6.19774126e-02 -9.79389012e-01 8.46469104e-01
7.92549625e-02 4.59728837e-01 1.13642836e+00 -4.38674837e-01
1.38536151e-02 -3.06109250e-01 -1.67507362e+00 5.18928468e-02
7.43934274e-01 -3.51194203e-01 -7.36596763e-01 5.10731637e-01
5.25483668e-01 -2.99629509e-01 -1.12313449e+00 6.48727357e-01
2.61491686e-01 -7.18230903e-01 8.48962069e-01 -4.78604287e-01
2.99374253e-01 -8.89700949e-02 -9.60961431e-02 -1.16497493e+00
-2.77582765e-01 1.84911698e-01 4.69398767e-01 1.14690232e+00
1.00313433e-01 -9.40895498e-01 7.30518579e-01 2.20266074e-01
1.50200129e-01 -7.48917103e-01 -1.11849141e+00 -7.44929135e-01
5.63164711e-01 9.89000034e-03 5.00535965e-01 1.30728948e+00
-3.73767346e-01 4.39252108e-01 -1.70787230e-01 -1.66659672e-02
3.97640616e-01 2.54223287e-01 4.71364677e-01 -1.48961651e+00
-2.99222946e-01 1.17533408e-01 -2.34772548e-01 -9.55740333e-01
-1.65001929e-01 -1.13470149e+00 -2.34196693e-01 -1.47280550e+00
4.84665453e-01 -7.76691973e-01 -4.10328448e-01 5.17657578e-01
-1.02053201e-02 1.80664748e-01 3.53608206e-02 4.90116894e-01
-4.19291019e-01 5.34114122e-01 1.79512179e+00 -2.45442286e-01
8.22086483e-02 4.39548910e-01 -5.08641183e-01 7.02453911e-01
9.04318273e-01 -7.27086127e-01 -6.29516423e-01 -5.06648004e-01
-5.47884703e-01 5.11078596e-01 3.12770784e-01 -9.51145172e-01
7.07985759e-02 9.02624130e-02 5.57405114e-01 -5.48208475e-01
1.72460941e-03 -1.30023742e+00 1.12262309e-01 1.04185379e+00
7.09719062e-02 -1.32028073e-01 6.32747114e-02 8.21652174e-01
-4.29928958e-01 -1.02368243e-01 9.23844874e-01 -3.73095065e-01
-5.25228381e-01 6.87567711e-01 -3.26592177e-01 1.55022100e-01
1.54860616e+00 -1.31026372e-01 -3.13825309e-01 -1.05443627e-01
-6.77559972e-01 1.19694538e-01 1.37586594e-01 3.00680876e-01
5.68231583e-01 -1.24648237e+00 -8.39482427e-01 2.79781193e-01
2.92434454e-01 2.33634561e-01 3.72954577e-01 1.00969493e+00
-5.53929746e-01 4.07191485e-01 -3.16327542e-01 -9.44936275e-01
-1.30087996e+00 8.56005192e-01 3.00153524e-01 -5.53046644e-01
-5.63837647e-01 5.15032709e-01 6.38966739e-01 -8.01818013e-01
-8.87543522e-03 -3.58279616e-01 -2.28515297e-01 -4.31293808e-02
1.98035210e-01 1.99790940e-01 2.76784509e-01 -3.93125176e-01
-4.17547435e-01 7.94995308e-01 -2.35882103e-01 4.07884508e-01
1.26052356e+00 -1.15172472e-02 4.03047446e-03 1.88070357e-01
1.50731981e+00 -4.25073147e-01 -8.56103420e-01 -4.37734544e-01
-4.02864218e-01 -2.53846526e-01 -2.21818313e-01 -9.36561823e-01
-1.21656609e+00 1.17195415e+00 1.22665548e+00 -4.21750285e-02
1.24496651e+00 1.46725029e-01 1.31362545e+00 1.79844722e-01
3.88530463e-01 -8.53065252e-01 2.58448124e-01 1.36114314e-01
4.69888151e-01 -1.46260607e+00 -3.52033377e-01 -5.01043797e-01
-6.73061430e-01 8.01363111e-01 6.64315283e-01 1.99446127e-01
9.82079923e-01 7.52038285e-02 4.00370926e-01 -1.64678767e-01
-2.47207239e-01 7.79747143e-02 1.64031032e-02 8.95534813e-01
2.66361713e-01 1.37485743e-01 -3.66070479e-01 7.02049136e-01
3.78571600e-01 4.33524162e-01 -1.86547846e-01 1.05445731e+00
-2.06406936e-01 -1.25072205e+00 -4.95870113e-01 5.73263228e-01
-7.85079181e-01 1.44891784e-01 -6.29691407e-02 9.10121083e-01
5.29353678e-01 8.73706460e-01 -1.35829270e-01 -2.44758382e-01
1.72069088e-01 -8.63246911e-04 2.02356771e-01 -5.62127113e-01
-3.84409666e-01 -3.42858513e-03 -4.20166910e-01 -9.28547159e-02
-4.86179113e-01 -4.19419557e-01 -1.26326597e+00 -2.30121821e-01
-3.21308225e-01 3.08349133e-01 3.85330319e-01 1.12657094e+00
5.38552105e-01 6.38101161e-01 6.05293214e-01 -1.74479842e-01
-6.47957683e-01 -8.34648192e-01 -7.69993424e-01 8.04103673e-01
1.65321931e-01 -5.02496302e-01 -1.08568363e-01 2.28866175e-01] | [14.789413452148438, -2.023601531982422] |
bd558347-9876-4b22-8187-b3590d1cb362 | three-dimensional-microstructural-image | 2204.01645 | null | https://arxiv.org/abs/2204.01645v1 | https://arxiv.org/pdf/2204.01645v1.pdf | Three-dimensional Microstructural Image Synthesis from 2D Backscattered Electron Image of Cement Paste | The microstructure is significant for exploring the physical properties of hardened cement paste. In general, the microstructures of hardened cement paste are obtained by microscopy. As a popular method, scanning electron microscopy (SEM) can acquire high-quality 2D images but fails to obtain 3D microstructures.Although several methods, such as microtomography (Micro-CT) and Focused Ion Beam Scanning Electron Microscopy (FIB-SEM), can acquire 3D microstructures, these fail to obtain high-quality 3D images or consume considerable cost. To address these issues, a method based on solid texture synthesis is proposed, synthesizing high-quality 3D microstructural image of hardened cement paste. This method includes 2D backscattered electron (BSE) image acquisition and 3D microstructure synthesis phases. In the approach, the synthesis model is based on solid texture synthesis, capturing microstructure information of the acquired 2D BSE image and generating high-quality 3D microstructures. In experiments, the method is verified on actual 3D Micro-CT images and 2D BSE images. Finally, qualitative experiments demonstrate that the 3D microstructures generated by our method have similar visual characteristics to the given 2D example. Furthermore, quantitative experiments prove that the synthetic 3D results are consistent with the actual instance in terms of porosity, particle size distribution, and grey scale co-occurrence matrix. | ['Bo Yang', 'Yuxuan Zhang', 'Qinfei Li', 'Pengkun Hou', 'Lin Wang', 'Xu Wu', 'Xin Zhao'] | 2022-04-04 | null | null | null | null | ['texture-synthesis'] | ['computer-vision'] | [ 3.1022993e-01 -5.8089662e-02 2.6908159e-01 6.1336942e-02
-3.2503435e-01 5.9592184e-02 3.4255552e-01 4.1120270e-01
-2.2252202e-01 5.0474238e-01 -3.6949432e-01 -2.6831970e-01
-4.1422290e-01 -1.2093476e+00 -5.0026971e-01 -8.5699397e-01
-7.3471524e-02 8.6464953e-01 6.5826166e-01 -9.4506674e-02
5.5844158e-01 7.5363213e-01 -1.5388341e+00 -6.7394413e-02
3.9832258e-01 1.0884984e+00 1.0315477e+00 4.8591697e-01
-1.4477643e-01 4.0397990e-01 -3.1069329e-01 6.9800436e-02
-3.3836007e-02 -3.5562709e-01 -6.7413777e-01 5.7138431e-01
-3.4010324e-01 -3.7256199e-01 4.4506010e-01 1.0189627e+00
-8.7400839e-02 -8.2892530e-02 8.6469173e-01 -6.5358478e-01
-7.4733710e-01 1.9478236e-01 -7.9457963e-01 -4.4715255e-02
7.2312850e-01 1.3610366e-01 7.2615451e-01 -1.1677376e+00
7.8878075e-01 1.2681446e+00 4.3508984e-02 3.9951235e-01
-9.6795797e-01 -1.4569342e-01 -5.0750101e-01 4.1621417e-01
-8.9404780e-01 3.5868011e-02 8.4939033e-01 -3.5673067e-01
5.1603729e-01 3.7524033e-01 8.5188687e-01 4.8475575e-01
1.0277319e+00 3.6796081e-01 1.8942823e+00 -7.6686215e-01
7.8203255e-01 -1.9677706e-01 4.7274556e-02 5.9912914e-01
5.3687006e-01 2.3800080e-01 6.5799579e-02 2.3973423e-01
9.6283495e-01 2.5199577e-01 -4.7110599e-01 -8.8749491e-02
-9.7267395e-01 4.4356862e-01 3.8775209e-01 6.7189962e-01
-6.1911625e-01 -2.9648763e-01 -4.6486564e-02 2.2242485e-01
4.0019181e-01 3.7406868e-01 -7.9715706e-02 -6.9498330e-02
-7.7810198e-01 9.6438073e-02 7.4021733e-01 7.0253456e-01
1.0521940e+00 -1.9629565e-01 6.4459711e-01 7.4211240e-01
7.6689571e-01 1.0094328e+00 4.8904151e-01 -8.2404602e-01
-3.2933185e-01 6.8845975e-01 -1.8731135e-01 -8.4046972e-01
-2.0515346e-03 5.2164823e-01 -7.9801834e-01 8.1963617e-01
7.7600278e-02 6.5942180e-01 -1.1256999e+00 5.8945853e-01
6.7124313e-01 -4.7180405e-01 -1.3580781e-01 8.2387733e-01
7.5059474e-01 3.8093692e-01 -2.7594289e-01 -4.4850582e-01
1.4459199e+00 -5.8631343e-01 -9.1711938e-01 -9.6563511e-02
6.9518827e-02 -7.7812022e-01 1.1969877e+00 3.6324054e-01
-1.2813382e+00 -2.0697774e-01 -1.2501569e+00 5.9969687e-01
-7.6509565e-02 -3.2433036e-01 1.6883107e-01 4.4507566e-01
-6.3694948e-01 7.9197210e-01 -1.1483480e+00 -3.4667918e-01
2.4386993e-01 1.8694478e-01 -4.8136476e-01 -2.2832063e-01
-6.1571395e-01 8.9581037e-01 2.3141479e-01 2.8082824e-01
-5.2963299e-01 -4.0554002e-01 -5.5547857e-01 -1.4944829e-01
1.5581246e-01 -5.3707588e-01 1.1260462e+00 -3.5671543e-02
-2.1228228e+00 1.0156977e+00 6.3600489e-03 1.9556829e-01
1.0097341e-01 3.8135465e-02 -2.4162188e-01 8.4103954e-01
3.7429792e-01 -9.4745718e-02 5.0074410e-01 -1.8883845e+00
-3.9995995e-01 -5.2545696e-01 -2.2454995e-01 -2.5884348e-01
3.1119317e-01 -1.4401597e-01 1.0008359e-01 -3.0061814e-01
7.2345001e-01 -4.5412952e-01 -3.5820985e-01 -2.7255508e-01
-5.5652958e-01 2.4997568e-02 1.0448977e+00 -3.9617392e-01
7.0752096e-01 -2.1436174e+00 -2.8207085e-01 5.7879663e-01
2.3787993e-01 -6.3017763e-02 3.6656824e-01 5.5635822e-01
5.2229252e-02 1.1023646e-01 -5.4037827e-01 -2.3906530e-01
-5.3306930e-03 3.5949281e-01 2.8132525e-01 5.8929121e-01
-1.7182352e-01 6.5874213e-01 -9.1278970e-01 -7.9165083e-01
5.6374741e-01 1.2235700e-01 -1.2712671e-01 2.0016100e-01
-1.0030792e-01 3.3097532e-01 -8.1365138e-01 1.0501226e+00
9.8060459e-01 -2.9240847e-01 2.6428759e-01 -3.4287259e-01
-3.2724714e-01 -2.8442034e-01 -8.1378788e-01 1.0476425e+00
-3.5448256e-01 -2.6010979e-02 5.9799629e-01 -8.1782931e-01
1.1688751e+00 4.5695677e-01 3.9291230e-01 -1.2604226e+00
8.6998723e-02 8.5001266e-01 -1.9770367e-01 -8.1632328e-01
2.3308367e-01 -1.0687941e+00 5.6624508e-01 7.1373451e-01
-5.0116992e-01 -1.0569333e+00 1.5765856e-01 -8.7369150e-03
7.1572155e-01 -7.1161717e-02 1.3742353e-01 -4.5876637e-01
4.4419754e-01 1.5507694e-02 -5.0906703e-02 3.5763353e-01
4.4444668e-01 6.4699781e-01 3.0265157e-03 -1.0599642e-01
-1.5107417e+00 -1.1547559e+00 -2.4297705e-01 -1.2690622e-01
5.9276563e-01 1.6382547e-01 -8.9479697e-01 7.3400438e-02
-1.6549335e-01 1.7570902e-01 -5.9516138e-01 2.3294556e-01
-6.3804203e-01 -5.9371734e-01 -3.9726099e-01 2.4895059e-01
5.3245616e-01 -1.3130282e+00 -1.0329331e+00 3.9931074e-01
1.9448715e-01 -9.9166280e-01 1.7282246e-01 -3.7278179e-02
-1.2863380e+00 -1.5687621e+00 -7.7497822e-01 -8.8406676e-01
6.2493628e-01 3.5356337e-01 1.0319846e+00 7.2908485e-01
-3.0616856e-01 5.7053471e-01 -8.2708728e-01 -1.5320572e-01
-8.7438464e-01 -4.2570958e-01 4.1271731e-02 -1.6845375e-01
-5.9872605e-02 -8.3746439e-01 -7.9693848e-01 2.5369245e-01
-1.3859730e+00 6.0773425e-02 7.7102536e-01 6.5350765e-01
1.2184039e+00 7.2440594e-01 3.8315043e-01 -9.6952641e-01
7.0148975e-01 -3.0198518e-02 -3.6716804e-01 2.5804958e-01
-1.0306146e+00 -3.8735980e-01 6.6864157e-01 -1.5683597e-01
-1.2864668e+00 -8.3848447e-01 -4.2805672e-01 -1.8473625e-01
-3.4686744e-01 8.9074612e-01 -2.2534149e-02 1.6956231e-01
3.6298850e-01 4.0195781e-01 5.9485102e-01 -7.9304886e-01
-3.6594388e-01 7.3660022e-01 5.2097827e-01 -5.1653099e-01
8.9389920e-01 8.0528289e-01 1.6118731e-01 -1.1326541e+00
-4.8121375e-01 -2.3614614e-01 -5.1423377e-01 -5.5144435e-01
9.0012091e-01 -2.9652005e-01 -9.5208436e-01 6.9207346e-01
-6.1970204e-01 -5.6007892e-01 -4.0258318e-01 9.5636487e-01
-6.4466208e-01 7.7667028e-01 -1.0201734e+00 -7.1689343e-01
-3.0892789e-01 -1.4097630e+00 9.8903531e-01 5.3510986e-02
-1.1461377e-01 -1.2172269e+00 1.7758600e-01 4.2142850e-01
6.6548854e-01 5.1096678e-01 1.4746571e+00 3.0601797e-01
-4.3541202e-01 -4.3223474e-02 1.6861083e-01 2.5089809e-01
5.5703002e-01 3.4473082e-01 -5.1500297e-01 -6.2859066e-02
8.7082642e-01 -4.0865436e-02 3.7355205e-01 7.4177569e-01
5.8200896e-01 1.5522101e-03 -4.7967103e-01 -1.8055474e-02
1.7945957e+00 6.7738521e-01 9.6376330e-01 9.4556266e-01
3.3936560e-01 5.5747455e-01 8.8535774e-01 2.3076107e-01
5.6624975e-02 4.9275050e-01 6.6901481e-01 -2.7062294e-01
-2.2375967e-01 9.3199134e-02 1.3891189e-01 1.0853759e+00
-6.4018530e-01 1.7885767e-02 -6.8343449e-01 3.0944863e-01
-1.0550684e+00 -8.8081473e-01 -6.9134367e-01 2.0930614e+00
7.1659034e-01 1.6951561e-01 -9.9830762e-02 8.0674642e-01
7.6954484e-01 -3.6938274e-01 -3.1631684e-01 -2.1534529e-01
-1.8265176e-01 9.2273682e-01 1.5583679e-01 8.0761290e-01
-1.5433531e-01 3.7461114e-01 6.3447242e+00 6.2999433e-01
-1.2838732e+00 -6.6463321e-02 2.2790395e-01 4.7940528e-01
-7.1964729e-01 7.7584736e-02 -1.4528216e-01 6.3824868e-01
4.2782566e-01 -8.0507956e-02 5.9139628e-02 4.1923675e-01
4.6375895e-01 -9.6848661e-01 -8.5542631e-01 7.1031553e-01
-6.4190125e-01 -1.5537119e+00 7.0637368e-02 4.4526950e-01
3.1394705e-01 -2.3087455e-01 -2.3829147e-01 -5.4081345e-01
5.5890635e-02 -9.4991159e-01 1.0140355e+00 4.2157197e-01
7.8143239e-01 -3.3026290e-01 1.0561786e+00 1.8931703e-01
-1.1858150e+00 3.0524862e-01 -2.1530789e-01 -1.1943933e-02
6.4890158e-01 1.3396449e+00 -6.4423007e-01 8.3960378e-01
1.1183319e+00 5.4424018e-01 -3.4843686e-01 9.5131451e-01
-2.0872734e-01 5.1950771e-01 -5.6156123e-01 -1.6127437e-01
1.5796708e-01 -7.1519774e-01 3.9553657e-01 6.2163061e-01
5.2997655e-01 3.1360525e-01 -2.7060631e-01 1.1118973e+00
5.3519815e-01 -2.3397435e-02 -2.8047249e-01 1.3997558e-01
6.3408417e-01 8.0025959e-01 -1.4524201e+00 -4.6763331e-01
-1.6579047e-01 4.7887757e-01 -3.5906315e-01 2.0599814e-01
-3.3903384e-01 -7.1699999e-02 -2.5169167e-01 4.3623465e-01
3.1517792e-01 -3.2220879e-01 -3.6567774e-01 -7.0335537e-01
-6.9487102e-02 -7.0514756e-01 -2.8993100e-01 -8.9989072e-01
-1.6381875e+00 3.3476594e-01 1.8039162e-01 -1.0144093e+00
1.5683572e-01 -7.6966512e-01 -7.7917439e-01 6.1919105e-01
-1.2935890e+00 -6.5744150e-01 -1.6827370e-01 4.3055460e-01
2.9037178e-01 3.5595176e-01 8.3189362e-01 1.7929196e-01
-4.1704047e-01 -4.7032714e-01 3.1423530e-01 -4.1234010e-01
1.8176931e-01 -1.3620836e+00 -1.8836169e-01 3.0729744e-01
-6.2712091e-01 2.0892707e-01 8.8127249e-01 -7.7516115e-01
-1.6135063e+00 -4.3963048e-01 6.2494767e-01 1.5439966e-01
4.4116604e-01 1.9025958e-01 -9.7920614e-01 3.3662078e-01
2.2664094e-01 -4.5982301e-01 6.0342735e-01 -4.3722934e-01
4.4600490e-01 3.8283888e-01 -1.4643204e+00 2.7403265e-01
4.5073497e-01 -2.2644587e-01 -7.3416728e-01 1.8829234e-01
-1.4764729e-01 -3.0984414e-01 -1.6111691e+00 5.2261055e-01
6.7173839e-01 -1.2592289e+00 8.9400727e-01 4.3350324e-01
1.0482333e+00 -3.7317079e-01 -2.8974280e-01 -1.0736500e+00
-3.6557922e-01 6.0493510e-02 3.4268996e-01 8.2554132e-01
2.9895461e-01 -7.2372317e-01 8.3030134e-01 3.2780555e-01
-3.8671508e-01 -1.0239093e+00 -7.7311391e-01 -1.0003735e+00
-1.2156995e-02 2.5355192e-03 6.7728174e-01 6.7143607e-01
1.0398370e-01 2.4078576e-01 3.7462732e-01 -6.4497777e-05
9.5647842e-01 6.0379195e-01 4.4943285e-01 -1.3457122e+00
-1.4661470e-01 -2.3160674e-01 -3.5354212e-01 -8.0182129e-01
-2.2137815e-01 -5.3818375e-01 7.6511413e-02 -2.0670283e+00
3.6906680e-01 -7.4316317e-01 4.7989953e-01 7.4759685e-02
6.9721848e-02 2.2757512e-01 -9.8429658e-02 7.7956003e-01
1.0744978e-01 6.2702149e-01 2.1383669e+00 -2.0394363e-01
-1.0836575e-01 -1.3428760e-01 -8.4744498e-02 4.6359631e-01
5.2434075e-01 -3.6696932e-01 -1.9364087e-01 -6.8156414e-02
1.6035402e-01 2.9159060e-01 2.9722068e-01 -8.0237472e-01
-1.4729067e-03 -2.8816640e-01 1.3267966e-01 -8.3291054e-01
3.3138573e-01 -1.1479849e+00 7.1265197e-01 8.3580226e-01
3.9940077e-01 -1.2601630e-01 -3.8441408e-01 3.7376562e-01
-7.1727204e-01 -5.2137405e-01 9.8162997e-01 -7.1322095e-01
-4.4348767e-01 1.3323568e-01 -7.0297593e-01 -5.3364348e-01
9.9515033e-01 -1.0602708e+00 -2.0838434e-01 5.6675699e-02
-1.0344245e+00 -1.9532703e-01 1.1009758e+00 -6.4479244e-01
1.0426319e+00 -1.4118648e+00 -3.0122095e-01 1.5743065e-01
-4.7963482e-01 4.8404446e-01 4.1271195e-01 9.9154299e-01
-1.2957667e+00 1.2924575e-02 -4.0273091e-01 -8.9998710e-01
-9.6642017e-01 3.6220467e-01 2.6920727e-01 -5.2361798e-01
-9.9226516e-01 3.4090614e-01 2.4617885e-01 -7.5875647e-02
-6.3356978e-01 -4.5834178e-01 -2.0310530e-01 -2.9654768e-01
3.4839338e-01 5.2448535e-01 4.0983400e-01 -4.2635694e-01
-3.0026075e-01 1.0694121e+00 2.2371943e-01 -1.0715397e-01
1.9881771e+00 -3.5811526e-01 -5.5932665e-01 5.2965838e-01
1.1156336e+00 4.6601616e-02 -1.0237218e+00 2.6049092e-01
-5.5523068e-02 -6.5499341e-01 -9.5338702e-02 -2.5160104e-01
-1.3731842e+00 7.4088442e-01 3.0518797e-01 7.5810450e-01
1.1645672e+00 2.6316115e-01 9.9546111e-01 5.5924837e-02
6.6185582e-01 -1.2031988e+00 4.2939878e-01 8.6231045e-02
7.1435696e-01 -8.0994481e-01 2.4872465e-01 -8.0717617e-01
-4.0369511e-02 1.2919078e+00 2.9235592e-01 -2.2298880e-01
1.0608457e+00 4.8469242e-01 8.6213462e-02 -1.0582118e+00
-4.2335790e-01 2.0089984e-01 -4.4306657e-01 5.5301839e-01
1.3239658e-01 1.8726644e-01 -4.8235142e-01 1.8424141e-03
-4.3570942e-01 1.1720268e-01 7.3959571e-01 1.5564536e+00
-5.6360358e-01 -1.1641355e+00 -7.9916817e-01 2.5964722e-01
-4.2680955e-01 2.2762582e-01 2.0011481e-02 1.0018144e+00
-4.6739978e-01 9.7253984e-01 9.9090531e-02 -2.8035420e-01
5.4945773e-01 -4.1735840e-01 9.8448348e-01 -7.1119356e-01
-3.8504358e-02 2.6586396e-01 -1.8458611e-01 -1.7754808e-01
-1.1876539e+00 -3.5996220e-01 -1.4259413e+00 -4.7228572e-01
-6.6740119e-01 3.1807047e-01 6.7472726e-01 1.1503602e+00
-4.3214881e-01 2.8484696e-01 9.4775063e-01 -1.0326864e+00
-6.5942414e-02 -9.9686062e-01 -1.5891160e+00 6.8606365e-01
1.5267472e-01 -7.9762995e-01 -6.6830957e-01 3.1933543e-01] | [12.914645195007324, -2.7731945514678955] |
a1c238c5-7bda-456d-aeec-1a90753db4c5 | a-large-scale-dataset-for-end-to-end-table | 2303.14884 | null | https://arxiv.org/abs/2303.14884v1 | https://arxiv.org/pdf/2303.14884v1.pdf | A large-scale dataset for end-to-end table recognition in the wild | Table recognition (TR) is one of the research hotspots in pattern recognition, which aims to extract information from tables in an image. Common table recognition tasks include table detection (TD), table structure recognition (TSR) and table content recognition (TCR). TD is to locate tables in the image, TCR recognizes text content, and TSR recognizes spatial ogical structure. Currently, the end-to-end TR in real scenarios, accomplishing the three sub-tasks simultaneously, is yet an unexplored research area. One major factor that inhibits researchers is the lack of a benchmark dataset. To this end, we propose a new large-scale dataset named Table Recognition Set (TabRecSet) with diverse table forms sourcing from multiple scenarios in the wild, providing complete annotation dedicated to end-to-end TR research. It is the largest and first bi-lingual dataset for end-to-end TR, with 38.1K tables in which 20.4K are in English\, and 17.7K are in Chinese. The samples have diverse forms, such as the border-complete and -incomplete table, regular and irregular table (rotated, distorted, etc.). The scenarios are multiple in the wild, varying from scanned to camera-taken images, documents to Excel tables, educational test papers to financial invoices. The annotations are complete, consisting of the table body spatial annotation, cell spatial logical annotation and text content for TD, TSR and TCR, respectively. The spatial annotation utilizes the polygon instead of the bounding box or quadrilateral adopted by most datasets. The polygon spatial annotation is more suitable for irregular tables that are common in wild scenarios. Additionally, we propose a visualized and interactive annotation tool named TableMe to improve the efficiency and quality of table annotation. | ['Zhenghui Gu', 'Shuangping Huang', 'Xinwu Liu', 'Lei Hu', 'Fan Yang'] | 2023-03-27 | null | null | null | null | ['table-recognition', 'table-annotation', 'table-detection', 'table-annotation'] | ['computer-vision', 'knowledge-base', 'miscellaneous', 'natural-language-processing'] | [ 1.68064889e-02 -1.61183193e-01 -9.33119059e-02 -2.49432072e-01
-8.34003210e-01 -1.14213097e+00 3.90273243e-01 2.27635443e-01
-7.14289770e-02 6.85862005e-01 3.56297195e-02 -3.57795209e-01
-1.72437340e-01 -8.84393871e-01 -6.65049374e-01 -4.12926972e-01
3.00660640e-01 7.19725728e-01 2.58639097e-01 -2.46432379e-01
5.10631680e-01 7.43402004e-01 -1.60262764e+00 7.48141468e-01
7.06413686e-01 1.40309298e+00 4.23389040e-02 5.24916112e-01
-7.17779636e-01 9.60021794e-01 -7.92725444e-01 -6.91506803e-01
2.18500122e-01 -1.50080666e-01 -6.38719976e-01 3.11226577e-01
4.26694930e-01 -1.57245263e-01 -1.65979683e-01 9.03653145e-01
3.88364255e-01 -2.51921892e-01 6.29746854e-01 -1.27460039e+00
-5.74714184e-01 7.15676785e-01 -7.72005856e-01 -8.48089252e-03
6.62947774e-01 -2.51623303e-01 7.99196541e-01 -1.24701357e+00
9.02220905e-01 1.28588629e+00 5.25234997e-01 -7.10026026e-02
-7.46697068e-01 -6.61163747e-01 3.98027115e-02 1.67774558e-01
-1.77918732e+00 -3.30596596e-01 6.20992601e-01 -5.27064443e-01
7.08960414e-01 6.90870404e-01 2.65142113e-01 7.19294786e-01
2.82322824e-01 8.42413366e-01 1.00337851e+00 -4.81932282e-01
1.95678733e-02 5.74339330e-01 8.71996582e-02 4.85430241e-01
4.83743846e-01 -8.03733766e-01 -6.84158087e-01 2.04879716e-01
8.01210761e-01 -2.74566049e-03 -1.52153805e-01 -3.56673867e-01
-1.48698497e+00 1.50601804e-01 -4.90692742e-02 2.63917744e-01
2.54800972e-02 -5.80203772e-01 5.84260881e-01 1.58903912e-01
2.55190209e-02 9.71425772e-02 -5.84580362e-01 -3.78114879e-01
-6.03391349e-01 2.20202208e-01 8.11507523e-01 1.66098893e+00
6.74912572e-01 -1.41856357e-01 -3.23568314e-01 9.57802713e-01
6.98314309e-02 6.64734185e-01 2.86211044e-01 -3.14868331e-01
1.21383381e+00 1.30248249e+00 1.17194772e-01 -1.26194751e+00
-3.43784273e-01 -5.12322634e-02 -1.18202579e+00 -2.15471253e-01
7.11660802e-01 2.44916245e-01 -6.67347848e-01 7.77259886e-01
4.05436218e-01 -6.06392384e-01 1.82559472e-02 6.04129493e-01
1.29097569e+00 7.60979295e-01 -4.91461754e-01 -4.79953773e-02
2.04424119e+00 -6.61683142e-01 -1.14070749e+00 -4.94572781e-02
6.94737077e-01 -1.09098518e+00 1.41565812e+00 8.56628180e-01
-9.24608767e-01 -6.30634308e-01 -1.01093900e+00 -3.44844371e-01
-1.13874948e+00 7.74788499e-01 1.55089572e-01 8.41526091e-01
-6.74628019e-01 -6.47559762e-02 -2.26489201e-01 -4.57246393e-01
3.49438339e-01 2.11351231e-01 -7.01906383e-01 -1.38370350e-01
-8.99386466e-01 5.51347017e-01 5.48395276e-01 5.08847654e-01
-1.85721591e-01 -4.62551743e-01 -8.67192030e-01 -6.78993687e-02
1.07126009e+00 2.97567118e-02 8.36004376e-01 -1.95331588e-01
-1.14146018e+00 1.14943516e+00 5.57303280e-02 4.17848676e-02
6.86457694e-01 -8.50206316e-02 -7.07051039e-01 -1.16775490e-01
1.83425173e-01 2.11655900e-01 4.18501705e-01 -1.20742059e+00
-6.64682150e-01 -5.81526637e-01 -2.88736850e-01 2.77674675e-01
-2.17288807e-01 1.75125152e-01 -1.09443295e+00 -9.72576380e-01
2.20860213e-01 -5.43666124e-01 3.53779435e-01 -5.00693172e-02
-8.34436297e-01 -3.04155685e-02 9.93044317e-01 -9.86718655e-01
1.78358257e+00 -2.20053434e+00 -2.92812854e-01 3.97079557e-01
1.55988947e-01 1.20517816e-02 4.62577075e-01 6.55820370e-01
-5.74457981e-02 3.31616104e-01 -1.98704600e-01 1.93655882e-02
1.54751092e-01 4.42680120e-02 -3.17929655e-01 2.24263877e-01
4.31970097e-02 6.88275933e-01 -2.02937603e-01 -9.27117825e-01
9.98452082e-02 1.12955064e-01 -3.93559039e-02 1.07182013e-02
-9.66666639e-02 1.42788619e-01 -4.18650776e-01 1.19807482e+00
9.60834026e-01 -1.48223624e-01 3.17815006e-01 -5.31055689e-01
-4.28421229e-01 4.65157889e-02 -2.19164395e+00 1.25515676e+00
-9.65102464e-02 2.92082816e-01 8.52422342e-02 -6.13417804e-01
1.38174021e+00 1.09942675e-01 1.84381962e-01 -8.89932811e-01
1.53772775e-02 3.49507898e-01 -3.16265285e-01 -5.26841700e-01
8.31623137e-01 7.24102259e-01 -3.43891025e-01 2.21680462e-01
-3.72014403e-01 3.01755052e-02 7.24676132e-01 2.09518582e-01
7.56661057e-01 1.55075952e-01 5.89122832e-01 -3.26762855e-01
9.45220292e-01 2.75748044e-01 3.31196725e-01 5.29359639e-01
2.08967194e-01 8.54603171e-01 9.37679052e-01 -6.30137086e-01
-1.04828715e+00 -9.47977364e-01 -2.45349303e-01 7.75969446e-01
2.40446046e-01 -6.38678253e-01 -6.91297114e-01 -6.49875760e-01
-4.71980907e-02 1.93649709e-01 -4.33351964e-01 4.88062978e-01
-7.77896404e-01 -4.10839617e-01 6.43522918e-01 5.08843660e-01
1.03818297e+00 -1.27613473e+00 -4.42855865e-01 1.25103980e-01
-3.37321222e-01 -1.45496607e+00 -5.97418189e-01 2.07297727e-01
-5.46244681e-01 -1.22650516e+00 -3.58282268e-01 -7.61543870e-01
6.91392004e-01 -5.06767295e-02 1.28637528e+00 -1.61031082e-01
-4.55389678e-01 1.21284775e-01 -3.11273187e-01 -5.00907123e-01
4.34311442e-02 1.06594622e-01 -1.82640538e-01 2.33147025e-01
2.50860095e-01 1.38185620e-01 -1.64291471e-01 9.30112720e-01
-1.05230880e+00 9.14804116e-02 5.46496153e-01 4.75188702e-01
1.12512112e+00 4.53338742e-01 5.14321849e-02 -1.18444192e+00
4.47080493e-01 -1.07281394e-01 -8.77029777e-01 7.44984984e-01
-3.47864717e-01 -2.64319088e-02 9.04704988e-01 -2.53183275e-01
-8.92283857e-01 4.39440459e-02 1.32487282e-01 -5.61758243e-02
-3.96078795e-01 4.44818616e-01 -9.62176979e-01 3.73347104e-01
4.15992975e-01 4.73723859e-01 -2.94812799e-01 -6.61577940e-01
-1.47527386e-03 8.64962041e-01 6.01743281e-01 -6.79431438e-01
8.68670642e-01 3.59273642e-01 -4.93695177e-02 -7.22239316e-01
-5.07555902e-01 -3.79388571e-01 -9.11659837e-01 -3.41147482e-01
8.70317400e-01 -7.66651392e-01 -1.00937545e+00 4.05813396e-01
-8.42539191e-01 -1.76747628e-02 -5.19454554e-02 1.33851126e-01
-2.95575976e-01 9.47973430e-02 -5.82100749e-01 -7.65288413e-01
-2.61729777e-01 -1.25839770e+00 1.23159039e+00 1.47605076e-01
-3.17286700e-02 -6.93422496e-01 -4.54224885e-01 4.81200337e-01
5.53442053e-02 2.64349937e-01 1.00606930e+00 -6.43245876e-01
-8.54914427e-01 -2.41775244e-01 -4.70479935e-01 -1.50600493e-01
2.43843541e-01 3.45317006e-01 -7.86646247e-01 1.28305376e-01
-3.22739094e-01 -8.55342820e-02 3.31897497e-01 -5.05601242e-02
1.44104445e+00 -3.02195877e-01 -3.11284870e-01 4.30823565e-01
1.44288647e+00 7.77880847e-01 1.06967521e+00 5.85816681e-01
9.76908565e-01 6.88141942e-01 1.18038774e+00 5.72939098e-01
5.46302617e-01 8.38638067e-01 9.20315236e-02 -6.41788356e-03
1.80206880e-01 -2.32089236e-01 3.69043618e-01 8.00553381e-01
1.28712952e-01 -5.27403057e-01 -1.29731929e+00 2.09548280e-01
-1.59917545e+00 -7.89577961e-01 -4.86713320e-01 2.29554391e+00
6.79797351e-01 5.00053048e-01 2.36718774e-01 7.98775315e-01
8.06305766e-01 -9.67094898e-02 -3.00636321e-01 -5.84830821e-01
-3.72793853e-01 -1.41891435e-01 4.87831622e-01 -3.79484966e-02
-1.18831956e+00 6.87323749e-01 5.21122599e+00 1.26152945e+00
-8.10624659e-01 -5.12482107e-01 1.03064620e+00 4.63057548e-01
-5.13041429e-02 -3.24001819e-01 -1.43372691e+00 3.10838729e-01
6.05855465e-01 1.82904527e-01 2.57006258e-01 7.84893930e-01
9.87528544e-03 -3.69048566e-01 -1.07188177e+00 1.49469411e+00
1.56872556e-01 -1.25071263e+00 2.45276481e-01 9.53500345e-02
2.13369787e-01 -8.84009063e-01 2.87849344e-02 2.62714267e-01
-2.06074774e-01 -1.17607546e+00 1.09561205e+00 5.53714097e-01
1.30496120e+00 -7.54340231e-01 8.49694371e-01 2.12123752e-01
-1.91948891e+00 2.47056693e-01 -2.35096321e-01 3.24906111e-01
-3.18492442e-01 4.63414967e-01 -6.92162931e-01 8.59772563e-01
1.04025328e+00 6.91427946e-01 -1.06462145e+00 6.85836196e-01
2.13978186e-01 1.94790855e-01 -3.19947958e-01 -2.70080455e-02
-1.33376524e-01 -5.16410172e-01 7.35413879e-02 1.33933973e+00
3.44517738e-01 -5.13572991e-02 2.40890086e-01 6.64430141e-01
-2.29594931e-01 5.40173292e-01 -7.40637660e-01 3.75943556e-02
6.07357621e-01 1.38886857e+00 -1.45905483e+00 -3.68850619e-01
-4.01154280e-01 6.24005556e-01 -8.63994807e-02 1.53924897e-01
-8.68002772e-01 -8.04670811e-01 1.32393733e-01 3.06157589e-01
6.03567243e-01 2.56291255e-02 -8.96890163e-01 -9.56658602e-01
7.05889106e-01 -1.22463012e+00 7.81648636e-01 -8.51985157e-01
-9.80440259e-01 6.90552950e-01 9.49708074e-02 -1.64274871e+00
1.76195771e-01 -8.95091236e-01 -6.59547374e-02 7.27303982e-01
-1.03142250e+00 -1.02861309e+00 -6.03624463e-01 8.09389114e-01
5.92268407e-01 -3.23008001e-01 5.48960567e-01 7.36444414e-01
-1.04707921e+00 8.75173390e-01 1.93937093e-01 5.60861349e-01
7.91231930e-01 -1.45377958e+00 1.62695512e-01 7.77361035e-01
1.71548948e-01 5.14025152e-01 2.81672895e-01 -7.07375050e-01
-1.94606483e+00 -8.77274752e-01 7.83926070e-01 -7.20917106e-01
3.72945458e-01 -1.07740939e+00 -9.44360316e-01 6.13476992e-01
-1.93325251e-01 2.11801052e-01 2.87008435e-01 -1.88100204e-01
-4.26384807e-01 -6.95087552e-01 -1.34612858e+00 6.12699509e-01
7.77038515e-01 -3.89541447e-01 1.30410939e-02 1.73252940e-01
2.23816112e-01 -9.89064813e-01 -1.18811560e+00 1.11426428e-01
6.77601099e-01 -8.95446599e-01 9.44556832e-01 1.76239669e-01
3.93238693e-01 -7.60982394e-01 -3.25023443e-01 -5.38092434e-01
1.55042589e-01 -3.89946520e-01 3.97004653e-03 1.77536285e+00
6.24931276e-01 -2.92550564e-01 9.50477660e-01 4.77106988e-01
-9.05881226e-02 -5.63915730e-01 -5.81178010e-01 -7.62606382e-01
-3.65600705e-01 -4.41090167e-01 1.02992558e+00 9.11831677e-01
-1.62462756e-01 1.18933983e-01 -2.58199722e-01 6.04537502e-02
2.68362492e-01 1.65677473e-01 1.06164122e+00 -1.18695593e+00
2.41265893e-01 -4.03396636e-01 -3.11852843e-01 -9.17865276e-01
-4.63438183e-01 -6.20096385e-01 -2.93627232e-01 -1.71952164e+00
-5.50631471e-02 -3.26882124e-01 2.09992915e-01 4.10942167e-01
1.46028355e-01 1.39866740e-01 2.52478331e-01 2.18360528e-01
-7.13475645e-01 1.16578536e-02 1.38023424e+00 -1.89842224e-01
-2.02539191e-01 -2.81620413e-01 -6.07303679e-01 4.51009989e-01
6.48689508e-01 -3.54228169e-01 -3.48559380e-01 -1.33884341e-01
4.29938018e-01 2.06422910e-01 -1.34169638e-01 -9.81276870e-01
2.51029789e-01 -2.33695567e-01 6.83303356e-01 -1.46463788e+00
-4.77734720e-03 -1.08837032e+00 3.12569350e-01 7.39711896e-02
7.61860311e-02 6.94174349e-01 3.51395190e-01 2.02439860e-01
-5.50839365e-01 8.96806270e-02 4.62467074e-01 -2.23696575e-01
-7.66352892e-01 2.18837529e-01 -3.41333330e-01 3.03589076e-01
1.02893114e+00 -8.04495871e-01 -3.79590124e-01 -4.90031987e-02
-4.52557921e-01 2.19237819e-01 2.10300177e-01 2.76682734e-01
7.00342357e-01 -1.32929385e+00 -4.85568553e-01 3.54415476e-01
6.58496797e-01 4.42265511e-01 1.82695851e-01 5.46727002e-01
-1.13766205e+00 6.09297216e-01 -2.50354260e-01 -6.13802969e-01
-1.28494883e+00 5.65390646e-01 1.21044298e-03 -5.59781849e-01
-6.38800144e-01 3.40843320e-01 8.05683061e-02 -6.41097188e-01
5.13916731e-01 -7.01675236e-01 -6.50646567e-01 3.75453323e-01
7.08555698e-01 4.18087751e-01 7.35424340e-01 -4.38406378e-01
-5.44319987e-01 7.91813433e-01 -2.42498562e-01 1.28401637e-01
8.73404026e-01 -1.19890451e-01 -1.79829091e-01 6.89926088e-01
7.70623863e-01 4.55943257e-01 -6.67776227e-01 -7.07481578e-02
4.20914829e-01 -5.13639569e-01 -6.97922289e-01 -1.07262766e+00
-9.27892566e-01 6.31134033e-01 4.11058962e-01 3.58714521e-01
1.18474591e+00 -3.62656802e-01 4.67645258e-01 5.53010523e-01
4.70504522e-01 -1.17762911e+00 -9.65161175e-02 5.87969661e-01
1.09133399e+00 -1.12245214e+00 1.60375327e-01 -8.45112979e-01
-9.31838632e-01 1.42044806e+00 8.38049650e-01 4.26795363e-01
6.04302526e-01 8.38877082e-01 5.94570152e-02 -2.19102845e-01
-6.11071646e-01 7.46915629e-03 4.00413841e-01 4.98915702e-01
6.39168561e-01 -1.88791864e-02 -1.08682111e-01 6.58731759e-01
-5.08026779e-01 -2.52678275e-01 6.06682479e-01 1.13492560e+00
-1.21741168e-01 -1.08759630e+00 -1.08803546e+00 6.48863673e-01
-5.49891651e-01 5.48993722e-02 -4.72748131e-01 1.09752214e+00
2.25371331e-01 8.46017361e-01 1.58957437e-01 -3.25457364e-01
8.14676225e-01 -3.69775742e-02 1.56016156e-01 -3.34808379e-01
-7.82531083e-01 6.19240366e-02 1.66710943e-01 -3.83027166e-01
1.47145361e-01 -5.65550864e-01 -1.18495727e+00 -5.47316968e-01
3.13127600e-02 3.26160640e-02 5.31602204e-01 5.47005594e-01
2.36653209e-01 4.35632139e-01 3.31161648e-01 -1.95534796e-01
4.82232608e-02 -8.10886502e-01 -8.40692043e-01 3.46774757e-01
-9.61672456e-04 -7.08698332e-01 6.75856844e-02 3.32806587e-01] | [11.69633674621582, 3.008216619491577] |
982dc15b-517d-425a-b2e0-309542d2eaa4 | adamsformer-for-spatial-action-localization | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Chi_AdamsFormer_for_Spatial_Action_Localization_in_the_Future_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Chi_AdamsFormer_for_Spatial_Action_Localization_in_the_Future_CVPR_2023_paper.pdf | AdamsFormer for Spatial Action Localization in the Future | Predicting future action locations is vital for applications like human-robot collaboration. While some computer vision tasks have made progress in predicting human actions, accurately localizing these actions in future frames remains an area with room for improvement. We introduce a new task called spatial action localization in the future (SALF), which aims to predict action locations in both observed and future frames. SALF is challenging because it requires understanding the underlying physics of video observations to predict future action locations accurately. To address SALF, we use the concept of NeuralODE, which models the latent dynamics of sequential data by solving ordinary differential equations (ODE) with neural networks. We propose a novel architecture, AdamsFormer, which extends observed frame features to future time horizons by modeling continuous temporal dynamics through ODE solving. Specifically, we employ the Adams method, a multi-step approach that efficiently uses information from previous steps without discarding it. Our extensive experiments on UCF101-24 and JHMDB-21 datasets demonstrate that our proposed model outperforms existing long-range temporal modeling methods by a significant margin in terms of frame-mAP. | ['Chiho Choi', 'Karthik Ramani', 'Yi Xu', 'Nakul Agarwal', 'Kwonjoon Lee', 'Hyung-gun Chi'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['action-localization'] | ['computer-vision'] | [ 2.38877952e-01 -2.48269200e-01 -4.80988681e-01 -3.34259182e-01
-4.18621421e-01 -1.13847122e-01 7.96732843e-01 -2.01956928e-01
-6.21912777e-01 7.80459583e-01 6.14667237e-01 -1.17991187e-01
-7.53245642e-03 -3.90980661e-01 -6.81238174e-01 -6.28205240e-01
-4.33467746e-01 7.01379105e-02 5.78718185e-01 8.00135955e-02
2.46485755e-01 5.86377859e-01 -1.34837186e+00 5.03189325e-01
5.09470940e-01 9.79758680e-01 3.37681979e-01 7.90813506e-01
2.56089658e-01 1.90795457e+00 -2.61547148e-01 1.47222832e-01
2.95294464e-01 -4.84107971e-01 -8.72330546e-01 1.88421294e-01
1.24374926e-01 -7.55556405e-01 -8.38114619e-01 5.06609619e-01
-1.21252593e-02 6.47725523e-01 5.14184594e-01 -1.58272958e+00
-5.64901292e-01 1.96004421e-01 -5.09579062e-01 3.67700398e-01
3.69318068e-01 4.25280899e-01 5.65638006e-01 -7.14608729e-01
6.29733980e-01 1.44860160e+00 8.15703392e-01 6.13952696e-01
-8.46854329e-01 -4.35290068e-01 5.42229056e-01 7.64280975e-01
-1.23819768e+00 -4.89057243e-01 6.24812543e-01 -5.78794360e-01
1.28177834e+00 -5.62079921e-02 5.07150769e-01 1.16376197e+00
5.82165897e-01 1.22774541e+00 6.32170916e-01 -1.84756324e-01
1.81817174e-01 -6.22917295e-01 -7.94175789e-02 7.18077302e-01
-5.26937068e-01 9.55851227e-02 -8.49148989e-01 -7.39659965e-02
9.29801345e-01 4.38758105e-01 -2.33209535e-01 -2.22368717e-01
-1.69746256e+00 4.98175502e-01 2.24138454e-01 3.80621571e-03
-8.51693392e-01 6.50812209e-01 3.92238647e-01 -2.93685938e-03
6.59062743e-01 5.06234653e-02 -5.26346862e-01 -6.89166009e-01
-7.71012187e-01 3.32370847e-01 4.55081433e-01 7.97917664e-01
5.43208838e-01 -8.41581672e-02 -4.64458495e-01 4.77142841e-01
2.78552622e-02 3.43858391e-01 4.88455325e-01 -1.65206027e+00
3.73896778e-01 4.53501701e-01 5.63314319e-01 -1.20149994e+00
-4.44605559e-01 1.73319146e-01 -7.39826798e-01 3.65409106e-02
5.29700577e-01 -2.62199223e-01 -7.01782227e-01 1.69578397e+00
3.70284766e-01 1.02661991e+00 -1.24090619e-03 8.92861187e-01
1.04654849e-01 1.00040197e+00 2.99604923e-01 -3.96866739e-01
8.25507522e-01 -1.29739785e+00 -7.57287502e-01 -2.01561511e-01
8.43487799e-01 -3.52748036e-01 5.91803312e-01 2.57724077e-01
-1.03473330e+00 -7.24746406e-01 -5.02651274e-01 -2.10246637e-01
-8.60821158e-02 4.24702972e-01 6.35513246e-01 -1.79942548e-01
-1.10507607e+00 7.58236349e-01 -1.51162040e+00 -3.91383499e-01
3.62781495e-01 1.60213456e-01 -4.21328932e-01 6.97810650e-02
-9.76958096e-01 7.95030951e-01 4.33165669e-01 1.93208516e-01
-1.06219804e+00 -6.23249948e-01 -9.31563139e-01 4.97784428e-02
6.91412389e-01 -4.14948106e-01 1.50490344e+00 -8.83024752e-01
-1.53956366e+00 8.02381411e-02 -6.98331416e-01 -1.05316567e+00
5.92529833e-01 -5.76083541e-01 -3.10486525e-01 2.50916839e-01
2.01598063e-01 8.80402327e-01 7.23952889e-01 -6.33229554e-01
-1.12495565e+00 -9.70045999e-02 2.55735189e-01 2.35125124e-01
-2.70377636e-01 7.43270293e-02 -5.28754473e-01 -7.67533839e-01
4.38359194e-02 -1.20837176e+00 -4.31511998e-01 3.82678270e-01
1.92559548e-02 -5.01556933e-01 1.12723470e+00 -7.80830204e-01
1.24844110e+00 -2.06419897e+00 1.62374735e-01 -3.39781404e-01
1.80674002e-01 1.12154052e-01 -1.25316486e-01 3.70959222e-01
-2.84822918e-02 -4.08223450e-01 5.84111959e-02 -4.54939544e-01
-1.81894809e-01 3.66449118e-01 -5.79400539e-01 6.14453375e-01
1.16443165e-01 9.69684660e-01 -1.06551981e+00 -5.12575090e-01
5.36594868e-01 4.70781833e-01 -4.13992763e-01 6.71699047e-02
-4.24524516e-01 5.75217605e-01 -5.47015727e-01 3.61809164e-01
2.46483222e-01 -4.24936533e-01 -8.41299631e-03 5.51861562e-02
-4.11594719e-01 1.34175029e-02 -9.91264939e-01 1.75815356e+00
-2.48274058e-01 8.71964753e-01 -3.43456417e-01 -1.11113966e+00
5.12005806e-01 3.75114888e-01 1.08103991e+00 -6.53234303e-01
-1.40850171e-01 -1.78265363e-01 -3.78096133e-01 -6.91360354e-01
7.01662898e-01 1.94029912e-01 -7.54850172e-03 3.75968039e-01
-1.26391575e-01 6.96592212e-01 3.25559944e-01 2.32814878e-01
1.53275204e+00 6.54950440e-01 3.42920840e-01 2.48178467e-01
5.54462969e-01 2.77755588e-01 9.26912606e-01 8.92464459e-01
-6.49993718e-01 3.63617778e-01 4.06070977e-01 -8.57061267e-01
-8.28593254e-01 -7.66869783e-01 5.01176059e-01 1.16125727e+00
2.07684368e-01 -4.21383470e-01 -5.96129358e-01 -6.64438903e-01
-1.65487647e-01 7.45752335e-01 -6.50225282e-01 -1.68473229e-01
-1.03092492e+00 -2.27126658e-01 2.64739633e-01 8.81137192e-01
7.24706233e-01 -1.28265715e+00 -1.00016844e+00 4.06419307e-01
-5.26208639e-01 -1.39944112e+00 -5.50551176e-01 -2.77338684e-01
-7.47932374e-01 -1.00228965e+00 -1.07918823e+00 -3.86704385e-01
3.80907714e-01 6.31927431e-01 8.25594366e-01 -1.89815119e-01
-9.63169485e-02 6.50368690e-01 -4.46994543e-01 -1.33940950e-01
-1.94425523e-01 -2.50454009e-01 2.93123037e-01 2.38220885e-01
5.62902689e-01 -3.52541715e-01 -6.98654354e-01 3.73917580e-01
-4.55204397e-01 4.57293838e-01 4.43863600e-01 5.76445043e-01
6.66054547e-01 2.02813029e-01 2.44771585e-01 -3.56241286e-01
2.22426176e-01 -6.80045724e-01 -4.05279726e-01 3.34323853e-01
-7.75619149e-02 -1.46858171e-01 6.82649016e-01 -5.17585695e-01
-1.46016002e+00 3.98944706e-01 2.20401794e-01 -7.27203667e-01
-3.54847491e-01 4.53046173e-01 2.84793496e-01 2.67537057e-01
3.78886402e-01 5.19058347e-01 -1.06609739e-01 -4.34405714e-01
1.45637855e-01 3.29521775e-01 8.07706833e-01 -3.86911511e-01
2.47950017e-01 8.78161371e-01 9.97874811e-02 -7.79106319e-01
-8.15889537e-01 -6.19365513e-01 -8.09631050e-01 -5.42272985e-01
1.09634781e+00 -1.06263781e+00 -1.05495715e+00 7.18173325e-01
-1.47048402e+00 -6.67894602e-01 -2.03770220e-01 7.74691045e-01
-9.14514542e-01 5.33080459e-01 -6.80610299e-01 -9.89811420e-01
2.63804138e-01 -1.01253486e+00 1.12489748e+00 7.60589018e-02
-3.97009045e-01 -9.29184973e-01 8.95711780e-02 1.89438209e-01
3.11228856e-02 3.36096346e-01 3.86503786e-01 -1.86956689e-01
-8.70217979e-01 -9.76411924e-02 -6.41522035e-02 2.08015412e-01
1.50305912e-01 -1.57053262e-01 -3.92075479e-01 -5.70838563e-02
7.06705227e-02 -9.21543837e-02 9.51834083e-01 8.70628953e-01
1.48207998e+00 -2.71720856e-01 -5.42248666e-01 4.36160535e-01
9.25005317e-01 5.66876352e-01 5.47885597e-01 3.70036185e-01
7.44642556e-01 4.64505166e-01 1.10993791e+00 7.84927905e-01
6.08338714e-01 7.26512730e-01 2.89169997e-01 6.18344210e-02
2.17508301e-02 -2.26884767e-01 6.81230068e-01 2.99890101e-01
-3.62257391e-01 -3.74182403e-01 -1.01944029e+00 6.93603098e-01
-2.61959958e+00 -1.38795352e+00 -1.81470662e-01 1.76777065e+00
2.58921862e-01 -1.42032336e-02 2.20399335e-01 -1.90560073e-01
5.95461011e-01 3.05858731e-01 -7.78277099e-01 8.93571451e-02
1.11752190e-01 -3.32073390e-01 5.39927721e-01 4.09828633e-01
-1.46912861e+00 1.04004443e+00 6.16260338e+00 7.76168406e-01
-8.96201730e-01 1.24236420e-01 7.30094612e-01 -3.13277304e-01
5.57010055e-01 3.66669297e-02 -6.76372826e-01 5.69572330e-01
1.01312029e+00 8.26485083e-03 3.04282486e-01 9.16965425e-01
1.00528717e+00 -4.55463976e-01 -1.10212588e+00 9.70412493e-01
-1.89479720e-02 -1.62030780e+00 -9.23131704e-02 -1.32840067e-01
7.32418001e-01 8.32502767e-02 -8.50951448e-02 3.91916931e-01
4.30256248e-01 -7.25603461e-01 7.13829279e-01 1.12889421e+00
3.26873839e-01 -6.00588679e-01 3.13785523e-01 7.34381676e-01
-1.38884306e+00 -4.47689891e-01 -2.66224056e-01 -5.49613893e-01
6.37220144e-01 1.66868672e-01 -6.32501781e-01 1.94100291e-01
8.11281800e-01 1.45159113e+00 -3.70217651e-01 8.66584301e-01
-5.23335487e-02 4.88078266e-01 -1.42220750e-01 2.39115894e-01
5.47240019e-01 -3.34099717e-02 4.53518689e-01 9.36037242e-01
5.81323147e-01 2.77523428e-01 4.93179888e-01 5.09148657e-01
3.70673686e-01 -4.06398535e-01 -6.10219896e-01 -1.15770847e-01
1.49973661e-01 7.60140657e-01 -6.18664384e-01 -4.75030243e-01
-6.58235848e-01 1.31675494e+00 2.62362927e-01 5.83172679e-01
-1.24117756e+00 1.47196397e-01 8.06515574e-01 -4.90953438e-02
2.53063053e-01 -7.46524751e-01 1.79497391e-01 -1.28505826e+00
8.74122530e-02 -5.38035750e-01 3.31659555e-01 -1.01374662e+00
-8.42773855e-01 1.67418346e-01 1.63410872e-01 -1.53574944e+00
-5.86630106e-01 -5.23649931e-01 -4.61738974e-01 4.96738523e-01
-1.24295640e+00 -1.20335054e+00 -2.75519252e-01 7.29553044e-01
1.06902540e+00 -1.33755445e-01 4.14967328e-01 7.38277286e-02
-4.83406305e-01 -7.00328946e-02 2.43835554e-01 6.38275817e-02
5.73253036e-01 -8.79513144e-01 5.23091078e-01 9.57220733e-01
2.05289900e-01 1.93320051e-01 5.93145013e-01 -8.95394742e-01
-1.21930623e+00 -1.34120643e+00 1.06888139e+00 -4.27955717e-01
7.56404340e-01 -1.67797913e-03 -7.98245072e-01 1.07209027e+00
-9.99637693e-02 2.02259257e-01 2.32892454e-01 -2.47123688e-01
2.54693300e-01 1.84013903e-01 -5.79402924e-01 8.21964443e-01
1.32693565e+00 -3.61095101e-01 -3.22984785e-01 4.90474582e-01
7.31629014e-01 -3.81067276e-01 -4.73986119e-01 3.55311334e-01
5.23490369e-01 -1.06492066e+00 1.08907890e+00 -6.31156266e-01
5.33523381e-01 -2.91596830e-01 -6.33284524e-02 -1.04742193e+00
-5.04563689e-01 -7.09580839e-01 -6.94540739e-01 6.89990580e-01
-7.37824515e-02 -1.93806052e-01 1.05767548e+00 9.91039574e-01
-7.69905448e-02 -7.78737068e-01 -9.71576810e-01 -8.70944798e-01
-3.24520886e-01 -7.92889535e-01 4.05126750e-01 6.93686724e-01
-1.29139408e-01 -2.04738736e-01 -1.01416183e+00 1.14554964e-01
2.42062151e-01 -1.93723962e-01 9.39443111e-01 -7.03114271e-01
-2.99088359e-01 -6.59903735e-02 -5.38375378e-01 -1.81455672e+00
5.64091206e-01 -2.12074131e-01 2.65319616e-01 -1.51951993e+00
2.03011453e-01 6.54551312e-02 -3.20117325e-01 5.19740641e-01
-7.99874868e-03 -1.48944575e-02 3.09604049e-01 4.22422558e-01
-1.17671692e+00 8.77954543e-01 1.11668253e+00 -4.80441861e-02
-1.25978678e-01 2.57113986e-02 6.46662265e-02 1.21455014e+00
5.88243186e-01 -3.87537301e-01 -5.42996705e-01 -6.35565460e-01
-2.67355472e-01 5.38622022e-01 7.52948940e-01 -1.53119731e+00
6.14489913e-01 -7.25389779e-01 3.40019733e-01 -8.85973215e-01
7.56756246e-01 -8.83471191e-01 7.97127113e-02 4.50200409e-01
-5.11262178e-01 2.53448218e-01 -7.31718019e-02 9.89627600e-01
-2.23264977e-01 1.22315273e-01 3.72857690e-01 -1.75720394e-01
-1.54573905e+00 5.14947653e-01 -7.80894816e-01 -2.90225357e-01
1.51064885e+00 -3.15444201e-01 -1.31040052e-01 -6.87373221e-01
-1.03088176e+00 4.32361990e-01 2.64912397e-01 5.49996138e-01
7.47594893e-01 -1.37258875e+00 -4.26124603e-01 -1.14161789e-01
-1.00425772e-01 -3.18889081e-01 5.85888803e-01 1.11853838e+00
-4.48008120e-01 5.33853114e-01 -4.88052368e-02 -6.68992400e-01
-1.33415842e+00 5.55813193e-01 2.60533661e-01 -2.85458416e-01
-8.81875753e-01 7.41923749e-01 4.40066278e-01 -7.55657861e-03
3.45914930e-01 -2.98026115e-01 -2.84089237e-01 -2.81240225e-01
7.65589535e-01 7.08864748e-01 -6.16379559e-01 -1.09856200e+00
-2.38262534e-01 1.23036563e-01 -2.98784785e-02 -9.74505022e-03
1.37218225e+00 -4.22304422e-01 7.90903941e-02 6.37989402e-01
1.15183616e+00 -8.03462803e-01 -2.06295753e+00 -4.59772110e-01
1.33531913e-01 -6.67192400e-01 3.48655991e-02 -4.95883763e-01
-8.59452605e-01 8.02030265e-01 4.78892684e-01 -1.83410600e-01
1.05340862e+00 -1.85115263e-01 1.16771269e+00 7.23713100e-01
5.80132365e-01 -1.25546467e+00 3.12569112e-01 9.64792728e-01
7.96996891e-01 -1.32229352e+00 -1.69266090e-01 -1.67248636e-01
-7.36654997e-01 1.07602894e+00 8.18335652e-01 -1.12053268e-01
6.42894745e-01 4.99942526e-03 -2.03186050e-01 3.26964632e-02
-9.32485521e-01 -6.48833364e-02 1.30090788e-01 4.29956555e-01
1.95533097e-01 -1.95517942e-01 -1.46412300e-02 2.18012333e-01
3.63642573e-01 5.64058602e-01 4.00572687e-01 1.13324225e+00
-3.33961040e-01 -7.60593474e-01 -2.60524690e-01 3.76397789e-01
-3.30246925e-01 2.48479247e-01 -1.03596389e-01 6.49874032e-01
1.22015014e-01 9.01534438e-01 3.48793417e-01 -3.75704110e-01
4.41355519e-02 6.15031309e-02 2.73886949e-01 -3.47157329e-01
3.07923146e-02 -2.10635588e-02 3.01978714e-03 -1.09174371e+00
-8.62418652e-01 -9.66385901e-01 -1.28911841e+00 -3.95061761e-01
1.96937263e-01 -2.54233330e-01 2.17875332e-01 1.31104732e+00
6.15928411e-01 6.17600799e-01 2.77969390e-01 -1.23481083e+00
-2.85765082e-01 -8.55504513e-01 -3.65265936e-01 4.08571303e-01
4.81795639e-01 -9.12702322e-01 2.91999504e-02 5.45175731e-01] | [8.138017654418945, 0.4040980935096741] |
23ace87b-b618-4eb7-81ae-277b49c17efe | learning-a-probabilistic-model-for | 1812.07460 | null | http://arxiv.org/abs/1812.07460v2 | http://arxiv.org/pdf/1812.07460v2.pdf | Learning a Probabilistic Model for Diffeomorphic Registration | We propose to learn a low-dimensional probabilistic deformation model from
data which can be used for registration and the analysis of deformations. The
latent variable model maps similar deformations close to each other in an
encoding space. It enables to compare deformations, generate normal or
pathological deformations for any new image or to transport deformations from
one image pair to any other image. Our unsupervised method is based on
variational inference. In particular, we use a conditional variational
autoencoder (CVAE) network and constrain transformations to be symmetric and
diffeomorphic by applying a differentiable exponentiation layer with a
symmetric loss function. We also present a formulation that includes spatial
regularization such as diffusion-based filters. Additionally, our framework
provides multi-scale velocity field estimations. We evaluated our method on 3-D
intra-subject registration using 334 cardiac cine-MRIs. On this dataset, our
method showed state-of-the-art performance with a mean DICE score of 81.2% and
a mean Hausdorff distance of 7.3mm using 32 latent dimensions compared to three
state-of-the-art methods while also demonstrating more regular deformation
fields. The average time per registration was 0.32s. Besides, we visualized the
learned latent space and show that the encoded deformations can be used to
transport deformations and to cluster diseases with a classification accuracy
of 83% after applying a linear projection. | ['Boris Mailhé', 'Hervé Delingette', 'Nicholas Ayache', 'Julian Krebs', 'Tommaso Mansi'] | 2018-12-18 | null | null | null | null | ['deformable-medical-image-registration', 'diffeomorphic-medical-image-registration'] | ['medical', 'medical'] | [ 2.75636986e-02 2.99800664e-01 2.79577613e-01 -3.23337615e-01
-8.30979943e-01 -3.99004787e-01 5.98040521e-01 8.15529898e-02
-5.35401642e-01 5.98342597e-01 2.60731816e-01 3.00018758e-01
-2.60371685e-01 -7.97945738e-01 -8.46249104e-01 -1.14780390e+00
-4.56734091e-01 7.63794303e-01 1.92672729e-01 1.76245198e-01
-9.22200922e-03 6.84163272e-01 -8.68763566e-01 8.02015439e-02
7.96258688e-01 5.49186289e-01 2.31238887e-01 4.33135778e-01
4.26756263e-01 1.51073337e-01 -3.31847697e-01 -3.33348095e-01
1.20778829e-01 -1.87326550e-01 -9.98788655e-01 3.10284346e-02
4.20974165e-01 -3.75080734e-01 -2.60025799e-01 9.70776498e-01
6.31115794e-01 1.73147038e-01 9.06757653e-01 -7.22160280e-01
-8.29203844e-01 3.68865073e-01 -6.73869625e-02 8.25999603e-02
-1.31319650e-02 4.81454507e-02 5.85713863e-01 -7.52007782e-01
1.27301764e+00 9.78728592e-01 6.99873209e-01 7.10569024e-01
-1.76913142e+00 -9.87128764e-02 -5.47199488e-01 -4.80696699e-03
-1.14412141e+00 -7.87777305e-02 7.63822436e-01 -1.05534661e+00
6.31262362e-01 2.70159036e-01 5.38608670e-01 1.10448933e+00
8.29109669e-01 2.84790784e-01 1.20524991e+00 -1.14547983e-01
1.31287903e-01 -2.33445361e-01 -3.93793523e-01 6.45354211e-01
-6.16174676e-02 -1.89908370e-02 -1.39211984e-02 -2.46422395e-01
1.14214957e+00 1.25756100e-01 -4.39068764e-01 -3.95335168e-01
-1.70706725e+00 1.04696906e+00 6.28287673e-01 6.92598343e-01
-8.19991648e-01 2.02457577e-01 2.16281816e-01 -2.15213925e-01
7.79186249e-01 4.34330791e-01 -4.88308258e-02 2.15060282e-02
-1.03676748e+00 1.88939303e-01 4.32889551e-01 2.37317279e-01
6.12056851e-01 -2.68209279e-02 -3.80529255e-01 5.90638936e-01
5.27679682e-01 4.03569430e-01 6.74173295e-01 -1.30760431e+00
2.27833077e-01 3.47188264e-01 -2.13177979e-01 -1.09455192e+00
-4.69299316e-01 -2.34901935e-01 -1.16364074e+00 2.91257709e-01
3.53510380e-01 -2.13489737e-02 -7.79318571e-01 1.71457469e+00
3.04052055e-01 3.02306980e-01 -1.59514546e-01 1.07949114e+00
4.41472054e-01 4.98949915e-01 1.16973063e-02 -3.47686291e-01
1.13980615e+00 -3.98955882e-01 -8.41207087e-01 3.36411417e-01
6.08788490e-01 -7.10489690e-01 8.28819752e-01 4.68244851e-02
-1.41747844e+00 -4.37873453e-01 -8.26210201e-01 -5.66485664e-03
4.01383080e-02 -1.11083395e-03 2.58156538e-01 3.16302747e-01
-1.23906553e+00 1.29172552e+00 -1.61445045e+00 -1.29301190e-01
4.62025672e-01 3.46625030e-01 -6.67202950e-01 3.24774086e-01
-1.07780540e+00 8.45310450e-01 5.57750948e-02 1.42770112e-01
-7.75722563e-01 -9.56946075e-01 -8.85639668e-01 -1.65759161e-01
-3.83623600e-01 -8.21027637e-01 4.16465253e-01 -3.88413370e-01
-1.74450421e+00 1.07151973e+00 1.00717001e-01 -5.53314984e-01
8.66674244e-01 -1.21521927e-01 -1.76056966e-01 4.78307128e-01
1.25369057e-01 7.39321709e-01 6.74981833e-01 -9.43393707e-01
4.99693930e-01 -4.62192714e-01 -3.43049258e-01 -5.33207506e-02
-3.34371924e-01 -2.88366410e-03 -2.48889118e-01 -9.96134520e-01
5.24518311e-01 -1.17739797e+00 -2.50363588e-01 2.71895528e-01
-4.12733138e-01 1.22901179e-01 6.87238574e-01 -1.10312486e+00
8.98132861e-01 -1.91685236e+00 9.18415904e-01 1.66056618e-01
5.18527985e-01 -1.14092328e-01 1.04166515e-01 1.44513980e-01
-2.38998607e-01 1.75012395e-01 -8.28618646e-01 -5.30341923e-01
-1.23629771e-01 2.74484456e-01 7.50052929e-03 1.00371754e+00
6.92473575e-02 1.06974638e+00 -8.16750526e-01 -4.69405353e-01
2.88348287e-01 1.12984431e+00 -7.80087411e-01 3.61204982e-01
1.97368354e-01 9.85917926e-01 -9.62696970e-02 1.26762882e-01
6.53939426e-01 -1.22182898e-01 2.88556844e-01 -5.03610134e-01
-9.48436093e-04 -1.45601109e-01 -9.31448996e-01 2.08179832e+00
-2.32982025e-01 6.05943441e-01 -4.19089533e-02 -1.13393092e+00
8.59573781e-01 4.64444816e-01 1.08433425e+00 -1.94233194e-01
2.12858856e-01 1.94548413e-01 -1.46486938e-01 -5.69606602e-01
-1.44382536e-01 -2.72825599e-01 2.18619689e-01 4.98708129e-01
2.62428492e-01 -7.29130134e-02 4.29794192e-03 -1.36167422e-01
7.84073174e-01 2.30938494e-01 -3.75428379e-01 -5.77831030e-01
5.60925066e-01 -5.89178503e-01 5.18742442e-01 2.16591865e-01
-1.08692817e-01 8.23405683e-01 4.42393720e-01 -6.22145176e-01
-1.10255742e+00 -1.51055658e+00 -4.72019732e-01 2.28805155e-01
-2.53901184e-01 -6.15989603e-02 -1.19836926e+00 -5.52260399e-01
-1.63776681e-01 3.46424282e-01 -7.28190303e-01 -1.15360329e-02
-8.07032287e-01 -9.22507346e-01 4.44529474e-01 5.18776059e-01
2.99774617e-01 -1.06333315e+00 -5.89814603e-01 3.23014796e-01
-3.69574249e-01 -1.06938338e+00 -4.33132380e-01 -3.30741674e-01
-1.39893043e+00 -8.68338168e-01 -1.19653273e+00 -8.02435458e-01
1.00568426e+00 -8.50165248e-01 9.19865727e-01 -1.56892821e-01
-4.70516175e-01 2.92148918e-01 4.15045023e-02 2.69082785e-01
-6.98183894e-01 -2.31456488e-01 3.83515626e-01 2.67642558e-01
-2.74799585e-01 -8.66504252e-01 -9.43142295e-01 2.43283212e-01
-9.83172536e-01 -1.16602108e-01 2.96010584e-01 8.22470367e-01
9.91268635e-01 -3.78723443e-01 5.23618422e-02 -6.08934343e-01
5.53768158e-01 -3.05336773e-01 -4.66495037e-01 8.14671516e-02
-5.88823557e-01 3.23907077e-01 1.62442505e-01 -5.08622229e-01
-6.71816468e-01 1.78794444e-01 -1.76373273e-01 -6.62272573e-01
-1.26394734e-01 2.74402767e-01 1.62862957e-01 -4.46868390e-02
7.18249381e-01 1.86785057e-01 3.82279634e-01 -4.75621074e-01
3.80565912e-01 2.56693631e-01 5.43380737e-01 -4.73385781e-01
7.66618490e-01 7.71090627e-01 2.25781024e-01 -7.23218322e-01
-3.20910998e-02 2.00028699e-02 -1.28652930e+00 -2.82522947e-01
1.62496853e+00 -4.74261105e-01 -7.75203586e-01 5.88111997e-01
-1.27670062e+00 -5.42724192e-01 -4.45783138e-01 9.63525474e-01
-8.05096149e-01 6.42433643e-01 -1.05211937e+00 -1.91236570e-01
-5.23747444e-01 -1.53679800e+00 1.06286895e+00 -3.24074417e-01
-1.99324816e-01 -1.51718950e+00 3.77821147e-01 1.82636857e-01
5.50926924e-01 9.19583201e-01 8.15804005e-01 -1.90207556e-01
-5.12654603e-01 -6.57116016e-03 2.62808055e-01 5.57942092e-01
2.14335278e-01 -1.92971870e-01 -7.35223114e-01 -3.64801943e-01
2.26369530e-01 1.60175741e-01 7.58708954e-01 8.92133117e-01
1.26126564e+00 -3.65970194e-01 -1.71608880e-01 8.58729064e-01
1.26559234e+00 -1.77694008e-01 7.08525121e-01 -7.68273473e-02
8.37406397e-01 7.00651050e-01 -7.54675567e-02 1.10824160e-01
1.64578348e-01 9.23417091e-01 3.98451209e-01 -1.33107349e-01
-1.94398090e-01 1.87562004e-01 2.49701947e-01 1.15481949e+00
-5.38465679e-01 2.94096023e-01 -1.02716136e+00 5.17773390e-01
-1.57734716e+00 -8.14176261e-01 -3.01628441e-01 2.26811194e+00
9.02609408e-01 -6.49205521e-02 -1.46111190e-01 -9.22904909e-02
6.93697989e-01 1.14715986e-01 -1.34556875e-01 -1.81722090e-01
5.19252606e-02 3.75088453e-01 3.38397443e-01 8.83770347e-01
-1.17713189e+00 6.26784861e-01 6.03418589e+00 1.96602762e-01
-1.51900482e+00 5.00015676e-01 4.88837481e-01 2.04457104e-01
-3.81336361e-01 -2.63365149e-01 -9.98141021e-02 5.30993700e-01
9.74947512e-01 1.35282055e-01 3.88877004e-01 3.71439397e-01
2.14778647e-01 3.80267799e-01 -9.35598552e-01 8.72631013e-01
-1.44509198e-02 -1.64710999e+00 -5.59465475e-02 2.49647766e-01
7.82253504e-01 1.88188791e-01 5.04916199e-02 -3.26224566e-01
-3.07643533e-01 -9.84707355e-01 3.66072536e-01 1.13087380e+00
1.01591384e+00 -4.47944045e-01 6.41105235e-01 7.58552998e-02
-9.09763038e-01 5.16099274e-01 -2.05686137e-01 3.67976069e-01
4.64420795e-01 6.66297853e-01 -6.18432522e-01 4.42632914e-01
6.20659351e-01 8.12451661e-01 -2.61812240e-01 7.30987310e-01
-2.35343933e-01 3.69631946e-01 -2.52406299e-01 4.90474015e-01
-7.47758374e-02 -5.25507212e-01 8.18855047e-01 1.14347231e+00
4.75452721e-01 -1.87139049e-01 2.55448110e-02 1.23192382e+00
-8.81046429e-02 1.39529645e-01 -3.79514962e-01 2.07343474e-01
-3.73085216e-02 1.22688174e+00 -7.76663661e-01 -2.01008350e-01
1.58844531e-01 1.41089356e+00 -5.65603226e-02 4.18981612e-01
-7.76299715e-01 -2.04060972e-01 4.62208062e-01 3.77298415e-01
1.21807724e-01 -4.98370528e-01 -1.84797756e-02 -1.30664527e+00
2.20639586e-01 -7.50061274e-02 3.56372334e-02 -5.88263392e-01
-1.29784024e+00 8.44112754e-01 1.76112488e-01 -1.14975059e+00
-3.87587428e-01 -5.34137249e-01 -5.53994179e-01 1.10733855e+00
-1.20168972e+00 -1.08021629e+00 -2.55613953e-01 5.73097229e-01
8.77554491e-02 -4.79249135e-02 1.13010275e+00 4.52986002e-01
-1.41377792e-01 2.81177223e-01 2.20308483e-01 4.38293636e-01
7.68274426e-01 -1.43896949e+00 2.79869735e-01 7.21326947e-01
1.28570229e-01 8.20024610e-01 4.68971908e-01 -7.21946120e-01
-1.09057367e+00 -1.09238410e+00 1.00902987e+00 -4.56816018e-01
5.25548697e-01 -3.08310419e-01 -1.12229812e+00 8.05031002e-01
8.36005211e-02 6.94831610e-01 6.08601213e-01 -3.92570078e-01
1.13023415e-01 9.89614800e-02 -1.39321053e+00 2.64737368e-01
7.43675411e-01 -6.40095711e-01 -5.48201501e-01 5.71597576e-01
5.52564085e-01 -7.10868239e-01 -1.71206176e+00 3.72070044e-01
5.92673659e-01 -8.34338784e-01 1.10946524e+00 -4.20545906e-01
4.69335526e-01 -2.42856637e-01 8.13955367e-02 -1.38985169e+00
-4.42367971e-01 -6.12033904e-01 -1.69773161e-01 8.49258900e-01
3.64964679e-02 -6.70928419e-01 5.81094563e-01 5.76455116e-01
-2.10100248e-01 -9.00122643e-01 -1.13898909e+00 -6.77852571e-01
4.98942703e-01 -3.71677876e-01 1.65893525e-01 1.18497860e+00
-3.93655151e-01 -3.34866256e-01 -8.71462598e-02 2.23585755e-01
9.62991893e-01 -1.84366405e-01 2.08711118e-01 -1.23994482e+00
-1.75680682e-01 -2.82379210e-01 -7.24620044e-01 -5.83559692e-01
3.94694656e-01 -1.44914520e+00 -1.62597552e-01 -1.43540609e+00
3.25918905e-02 -2.46366844e-01 -3.52477729e-01 3.84718090e-01
1.98466390e-01 5.14575601e-01 7.02889562e-02 4.90468830e-01
5.74994311e-02 5.79318285e-01 1.44858718e+00 -6.62765354e-02
-2.81712532e-01 -2.46193931e-01 -2.04233136e-02 6.58948660e-01
6.96023703e-01 -6.12875402e-01 -2.56114788e-02 -4.74859208e-01
-1.16374016e-01 1.99193075e-01 6.60530269e-01 -1.07447720e+00
4.74390797e-02 2.66469806e-01 4.39041317e-01 -1.20442241e-01
2.60230064e-01 -6.51575148e-01 3.53556931e-01 7.88228631e-01
-2.18052968e-01 -5.54608442e-02 -1.96299162e-02 2.89945275e-01
-1.79367483e-01 6.56687245e-02 8.41339171e-01 1.22841187e-01
-4.21310030e-02 5.72577178e-01 -2.37975106e-01 -3.02979480e-02
1.02033198e+00 5.48689403e-02 2.34257326e-01 -6.08502217e-02
-1.38334572e+00 -4.10648435e-01 4.06431913e-01 2.96283603e-01
6.82709515e-01 -1.62198305e+00 -9.85473216e-01 3.48148823e-01
-2.09671319e-01 -8.35037306e-02 5.59667289e-01 1.29670668e+00
-8.98689687e-01 9.94329825e-02 -5.73994100e-01 -1.14626241e+00
-1.09793854e+00 2.34622225e-01 5.90155721e-01 -3.98987770e-01
-9.44625974e-01 6.07069671e-01 -1.25237769e-02 -4.67202187e-01
-1.80840269e-01 -5.51711321e-01 -1.79209605e-01 -7.60902278e-03
1.75944269e-01 3.83914322e-01 4.26803418e-02 -9.66607332e-01
-5.06772697e-01 1.02253938e+00 4.49891686e-01 -2.12399915e-01
1.63033068e+00 1.15296945e-01 -3.73205751e-01 3.35477144e-01
1.53566563e+00 -2.92907488e-02 -1.51733530e+00 -4.73110378e-02
-1.98486879e-01 -1.88989401e-01 2.48990968e-01 -4.40376580e-01
-1.35095811e+00 1.04221284e+00 1.17367232e+00 1.08085806e-02
8.42436492e-01 6.43726662e-02 8.38248789e-01 -9.86469761e-02
8.42425376e-02 -7.54724324e-01 -7.11381529e-03 1.77254677e-01
1.26417017e+00 -1.07669353e+00 7.85968732e-03 -1.95150465e-01
-5.70841551e-01 1.34951901e+00 -1.70169890e-01 -5.21307290e-01
8.94432902e-01 1.94066823e-01 9.23995376e-02 -3.19261402e-01
-2.58934088e-02 3.98808271e-01 7.53836513e-01 6.12516403e-01
5.57859898e-01 2.68316060e-01 -5.34387290e-01 1.66416526e-01
-1.29913306e-02 -2.12494023e-02 2.63458580e-01 4.94667768e-01
1.59267738e-01 -1.29831743e+00 -2.28287220e-01 4.76094633e-02
-7.37472475e-01 1.78369001e-01 3.21985334e-01 4.91563439e-01
1.09547816e-01 2.38820761e-01 2.34305277e-01 -1.06290787e-01
1.59096852e-01 1.72887668e-01 6.01616502e-01 -4.57920700e-01
-4.66583431e-01 2.87768424e-01 -5.61149180e-01 -9.28342342e-01
-7.06922054e-01 -9.34922516e-01 -1.43344629e+00 3.36231180e-02
1.56033039e-01 -8.80271494e-02 9.28646624e-01 7.80535519e-01
4.00916994e-01 5.57585359e-01 4.35467422e-01 -1.13614190e+00
-3.42875779e-01 -9.48149502e-01 -5.69264531e-01 9.08438981e-01
2.67852247e-01 -6.67192698e-01 -2.59005964e-01 5.79693079e-01] | [14.037057876586914, -2.4571945667266846] |
e6794bb7-fb84-4bd7-8d0e-328791ee21d1 | dynamicgem-a-library-for-dynamic-graph | 1811.10734 | null | http://arxiv.org/abs/1811.10734v1 | http://arxiv.org/pdf/1811.10734v1.pdf | DynamicGEM: A Library for Dynamic Graph Embedding Methods | DynamicGEM is an open-source Python library for learning node representations
of dynamic graphs. It consists of state-of-the-art algorithms for defining
embeddings of nodes whose connections evolve over time. The library also
contains the evaluation framework for four downstream tasks on the network:
graph reconstruction, static and temporal link prediction, node classification,
and temporal visualization. We have implemented various metrics to evaluate the
state-of-the-art methods, and examples of evolving networks from various
domains. We have easy-to-use functions to call and evaluate the methods and
have extensive usage documentation. Furthermore, DynamicGEM provides a template
to add new algorithms with ease to facilitate further research on the topic. | ['Emilio Ferrara', 'Arquimedes Canedo', 'Palash Goyal', 'Ninareh Mehrabi', 'Sujit Rokka Chhetri'] | 2018-11-26 | null | null | null | null | ['dynamic-graph-embedding', 'graph-reconstruction'] | ['graphs', 'graphs'] | [-5.59521914e-01 1.79848313e-01 -3.59857231e-01 -2.40644500e-01
1.56755731e-01 -7.53995717e-01 7.20187545e-01 2.72662222e-01
1.07902050e-01 4.45425719e-01 4.91290316e-02 -6.80556476e-01
-3.92512798e-01 -1.11058021e+00 -2.28696570e-01 -4.62424129e-01
-1.14069831e+00 6.56178057e-01 6.33064508e-01 -4.19366837e-01
-2.78856903e-01 5.46767175e-01 -1.04928291e+00 3.04422458e-03
2.00240016e-01 4.98901814e-01 -2.08138004e-01 1.13132787e+00
-2.93311104e-02 5.77700078e-01 -3.94236922e-01 -7.48821318e-01
7.88171738e-02 -5.40911295e-02 -9.62480068e-01 -5.36478579e-01
1.29756734e-01 -1.42967850e-02 -1.18103468e+00 7.27907240e-01
5.24088681e-01 2.59199254e-02 6.09743714e-01 -1.79012573e+00
-7.19558716e-01 9.89711165e-01 -4.54380251e-02 1.00827491e+00
4.43308234e-01 4.09183353e-01 1.23700833e+00 -4.62569743e-01
1.28827810e+00 1.31832230e+00 9.68976378e-01 4.30774748e-01
-1.37240398e+00 -6.12269282e-01 3.22091430e-01 2.98261911e-01
-1.17964852e+00 -2.83079416e-01 6.97990239e-01 -7.46670604e-01
1.33825374e+00 3.85635704e-01 8.76003504e-01 1.47905457e+00
2.17117056e-01 4.58063841e-01 3.16890031e-01 9.89101157e-02
-1.89410061e-01 -5.13989516e-02 5.00255048e-01 1.04917634e+00
3.47383693e-02 1.33736730e-01 -2.29495063e-01 -4.77612287e-01
6.21730745e-01 -6.77338094e-02 2.01340355e-02 -4.31560516e-01
-1.05681586e+00 6.73825145e-01 6.86420798e-01 4.46809083e-01
-1.24913774e-01 4.59923208e-01 8.39843929e-01 5.83466172e-01
7.88417697e-01 2.91899979e-01 -6.64035022e-01 -3.56238574e-01
-5.98975062e-01 2.84912169e-01 1.20056224e+00 9.02644038e-01
4.64161545e-01 1.01412795e-01 -8.06901008e-02 6.57287300e-01
2.98008323e-01 -3.78438234e-01 7.39098936e-02 -7.43005633e-01
2.48158902e-01 7.37102807e-01 -5.25724769e-01 -1.23307145e+00
-6.41169012e-01 -6.64139152e-01 -5.67517936e-01 -1.38897896e-01
1.55200407e-01 -2.13487133e-01 -6.85001075e-01 1.60305536e+00
2.83424258e-01 6.00308418e-01 -1.56595707e-01 3.95200104e-01
1.35162795e+00 7.86687315e-01 1.63637877e-01 -4.83382540e-03
8.56541812e-01 -1.17174613e+00 -5.99140942e-01 8.19987804e-02
8.50239694e-01 -2.24515244e-01 7.77409852e-01 -2.80416042e-01
-8.68400633e-01 -1.44715488e-01 -1.04172802e+00 1.48118123e-01
-9.17716503e-01 -3.67118478e-01 1.05084980e+00 3.58567625e-01
-1.50699031e+00 1.28448188e+00 -1.22514355e+00 -9.70900834e-01
3.51890564e-01 3.17889810e-01 -5.73927104e-01 1.99849233e-01
-1.43641984e+00 9.10082936e-01 4.08787817e-01 -9.16439444e-02
-1.12818921e+00 -1.02022505e+00 -9.52760994e-01 2.24210575e-01
1.01993747e-01 -8.43471885e-01 9.76297498e-01 -5.15833139e-01
-1.01913369e+00 1.06152117e+00 1.90676212e-01 -3.68776321e-01
5.70130169e-01 2.68034726e-01 -9.43791151e-01 -3.99764441e-02
-1.70399621e-01 2.95604020e-01 4.61537153e-01 -8.42356503e-01
-3.77513617e-02 -6.41137958e-02 1.94092050e-01 -1.31216392e-01
-7.08638132e-01 -3.44641507e-02 -8.33189845e-01 -6.18171811e-01
-3.87517750e-01 -9.37837660e-01 -1.06618769e-01 3.17790538e-01
-5.35200536e-01 -4.51858670e-01 1.11967504e+00 -7.16944635e-01
1.80396211e+00 -1.92847562e+00 2.81844199e-01 2.44903505e-01
5.73892534e-01 7.98880905e-02 -3.70578736e-01 9.05663133e-01
-3.60834152e-01 5.99625587e-01 1.11448616e-01 -4.21731412e-01
-7.39305168e-02 1.77276745e-01 1.28204077e-01 4.51532274e-01
2.54920665e-02 9.58859563e-01 -1.31854367e+00 -4.10366565e-01
9.29438509e-03 5.02445340e-01 -1.73719063e-01 1.00376360e-01
-3.01456362e-01 9.07063484e-02 -2.65455782e-01 7.50248015e-01
3.21738333e-01 -5.55666864e-01 7.21466064e-01 -1.28329292e-01
1.86069548e-01 3.50009739e-01 -8.80270958e-01 1.55356634e+00
-1.56657875e-01 1.06187272e+00 1.34551510e-01 -6.31980956e-01
6.55933380e-01 1.01751141e-01 5.40217221e-01 1.09511688e-02
2.64777280e-02 -1.16037674e-01 9.06273723e-02 -5.88996530e-01
2.56352037e-01 6.95244431e-01 1.94054142e-01 6.20773852e-01
3.61290365e-01 3.50171357e-01 6.56663597e-01 7.57505476e-01
1.87820542e+00 -1.69014204e-02 1.04160771e-01 -2.41074875e-01
1.63004696e-01 -1.21542387e-01 1.86455727e-01 3.06601703e-01
-1.88165918e-01 -1.03857540e-01 9.65732515e-01 -6.11661494e-01
-8.70173633e-01 -1.18750072e+00 3.51177864e-02 1.54092550e+00
-1.24569856e-01 -1.09383118e+00 -1.69533238e-01 -8.25710177e-01
5.29836833e-01 4.43536520e-01 -1.01848602e+00 -1.25518531e-01
-3.71296674e-01 -5.90012789e-01 6.22973144e-01 6.79844379e-01
-9.82888639e-02 -9.39028323e-01 2.10233971e-01 1.21697903e-01
3.32219332e-01 -8.80019367e-01 -4.53509718e-01 -8.66106451e-02
-9.54383492e-01 -1.51019335e+00 -2.00331554e-01 -9.00128663e-01
5.92564881e-01 7.75685757e-02 1.47080922e+00 5.51630437e-01
-5.52653313e-01 6.65515482e-01 -2.27762565e-01 2.45796531e-01
-5.20076811e-01 5.75473011e-01 -5.58301546e-02 -5.25132239e-01
-1.69099629e-01 -1.00320697e+00 -4.57634151e-01 1.88024610e-01
-5.29472530e-01 -2.77507246e-01 -2.53780019e-02 4.92775083e-01
1.43511653e-01 9.01634395e-02 2.14862496e-01 -1.15948355e+00
1.11018693e+00 -1.03670764e+00 -5.47199965e-01 2.95703948e-01
-8.35282266e-01 5.49491085e-02 3.37181419e-01 -4.06439245e-01
-2.98950762e-01 -3.65196079e-01 -4.84501868e-02 -6.18322968e-01
4.75932986e-01 9.07949328e-01 2.59451985e-01 -1.03639245e-01
6.71069801e-01 -2.42957816e-01 1.20725423e-01 -4.82000053e-01
6.37490451e-01 1.15179695e-01 4.97408748e-01 -2.13167384e-01
9.83663678e-01 1.95866227e-01 1.32442653e-01 -6.36398971e-01
-2.78672427e-01 -2.71152735e-01 -6.56951666e-01 -4.64506418e-01
3.53761375e-01 -6.72758698e-01 -7.41058707e-01 3.63435090e-01
-9.09499466e-01 -7.13132560e-01 1.03647582e-01 -2.60986537e-01
4.72425595e-02 2.77342111e-01 -1.02255023e+00 -4.54020113e-01
-5.06761014e-01 -7.29330897e-01 4.25816566e-01 1.69085398e-01
-3.44915807e-01 -1.97324240e+00 4.20588970e-01 -4.54888612e-01
6.48790777e-01 7.26181507e-01 9.73818541e-01 -8.42735410e-01
-3.79964113e-01 -3.33314687e-01 2.32962053e-03 2.06895871e-04
-5.72305806e-02 9.14353311e-01 -6.88888013e-01 -6.88092113e-01
-1.00616384e+00 7.35189244e-02 9.98588085e-01 2.34026268e-01
1.23671985e+00 -3.94717306e-01 -1.28142774e+00 8.03297698e-01
1.23387325e+00 -2.38608435e-01 3.94757658e-01 2.57097363e-01
6.37877762e-01 5.41696906e-01 1.96539730e-01 3.51754785e-01
7.61509597e-01 6.05610907e-01 6.89525604e-01 3.39161232e-02
-1.87010095e-01 -3.65635216e-01 3.18121552e-01 8.29684794e-01
1.90467685e-01 -5.41912735e-01 -1.23735785e+00 7.60554433e-01
-1.99573541e+00 -1.06797290e+00 -4.65435505e-01 1.51658905e+00
6.53074503e-01 3.92774552e-01 4.64100122e-01 -1.14913329e-01
7.49756932e-01 6.46116972e-01 -5.59210598e-01 -4.00521934e-01
2.59798676e-01 -5.14168181e-02 4.00160134e-01 5.16587317e-01
-1.33917499e+00 1.03438497e+00 7.93585062e+00 3.26902658e-01
-1.07929218e+00 1.77314579e-01 4.06786591e-01 2.78569816e-04
-4.59046066e-01 2.92045057e-01 -5.31920552e-01 3.71568233e-01
1.36243200e+00 -7.16981292e-01 4.97534990e-01 1.00484443e+00
8.24150443e-02 4.65655774e-01 -1.41588116e+00 5.69744170e-01
-4.09432769e-01 -1.76374960e+00 -3.63628507e-01 -2.43546609e-02
4.44406301e-01 5.28510153e-01 -8.64909515e-02 4.31647331e-01
7.70668507e-01 -9.62249339e-01 1.14756972e-01 6.88554823e-01
8.95544827e-01 -4.18949366e-01 5.77359259e-01 -8.62754062e-02
-1.66685915e+00 -1.07068360e-01 -1.41351134e-01 -4.59884740e-02
2.08335176e-01 7.15120792e-01 -1.03785825e+00 5.82296669e-01
7.52277195e-01 1.51857710e+00 -1.17163718e+00 1.11135721e+00
-2.40571022e-01 7.13209867e-01 -3.71046156e-01 -2.11804479e-01
-8.23290572e-02 3.53370197e-02 8.84602964e-01 1.48906898e+00
3.56266834e-02 -4.31692272e-01 2.33038753e-01 6.84027016e-01
-2.18854994e-01 -1.61689401e-01 -9.24952447e-01 -4.87165213e-01
9.46471930e-01 1.61353004e+00 -6.81534052e-01 -1.70093566e-01
-1.18400492e-01 6.67008996e-01 6.28808320e-01 4.26815718e-01
-7.84927309e-01 -5.73436260e-01 1.04213727e+00 3.80308032e-01
1.60307020e-01 -5.82034111e-01 1.69441625e-01 -8.48119915e-01
-2.96323895e-01 -3.32501203e-01 9.46908832e-01 -7.03877270e-01
-1.66046369e+00 7.33557165e-01 1.54590145e-01 -7.32165515e-01
-3.91364723e-01 -6.61734641e-01 -1.16603708e+00 5.61193287e-01
-9.95839953e-01 -1.08938968e+00 -5.27379930e-01 4.64624643e-01
7.12970048e-02 -2.62486190e-01 1.06010759e+00 4.34092313e-01
-1.01888907e+00 6.02125406e-01 -3.79563235e-02 4.54403579e-01
5.08204877e-01 -1.31714654e+00 1.15539694e+00 7.13474929e-01
-4.59952652e-02 7.05826998e-01 7.79010355e-01 -8.86693180e-01
-1.54575002e+00 -1.20492959e+00 5.94192624e-01 -5.42872131e-01
1.53747821e+00 -5.87001562e-01 -7.30905652e-01 1.20592391e+00
3.48208785e-01 3.30857098e-01 5.86012065e-01 5.56865513e-01
-3.83378178e-01 -1.91486448e-01 -9.39505398e-01 6.60486460e-01
1.61815786e+00 -5.86143851e-01 1.44134113e-03 6.31284535e-01
8.82587433e-01 -2.93179303e-01 -1.33772790e+00 2.76381135e-01
5.19523084e-01 -5.27593434e-01 8.39838147e-01 -8.94831181e-01
9.84146222e-02 -1.23870917e-01 4.26932067e-01 -1.56370294e+00
-6.53719068e-01 -1.02918696e+00 -9.66101348e-01 1.36559510e+00
7.94504762e-01 -9.99164581e-01 8.01373243e-01 2.10553586e-01
-5.94633073e-02 -9.87015009e-01 -6.86595201e-01 -7.66032934e-01
-2.74499804e-01 -2.86875486e-01 4.95942384e-01 1.15962160e+00
2.50973761e-01 2.99764007e-01 6.95890933e-02 6.30610660e-02
2.22525343e-01 -1.89030826e-01 7.99773514e-01 -1.64101481e+00
-1.65007338e-01 -6.97001040e-01 -8.20852578e-01 -6.10523939e-01
5.93594730e-01 -1.57655978e+00 -7.56771505e-01 -1.93417954e+00
2.20668204e-02 -5.22002280e-01 -2.94434428e-01 8.04402113e-01
5.19455224e-02 -1.38416037e-01 -9.18502808e-02 2.66365498e-01
-5.71115196e-01 3.77696127e-01 9.53230798e-01 -2.34014660e-01
-2.13403329e-01 2.39233002e-02 -4.21479404e-01 2.12052733e-01
7.50500381e-01 -5.83235919e-01 -5.66280782e-01 -2.88730800e-01
2.90628940e-01 -2.15675607e-01 2.61624515e-01 -7.77448952e-01
3.48392516e-01 2.02709332e-01 2.21149847e-01 -5.60409963e-01
2.56101549e-01 -6.03214502e-01 6.11334026e-01 5.80601215e-01
-3.06379288e-01 7.71665633e-01 2.59762824e-01 7.54215479e-01
2.16923624e-01 1.13544866e-01 5.23052871e-01 -1.06078193e-01
-8.52163196e-01 7.46804416e-01 -1.51970372e-01 1.16907442e-02
1.30304277e+00 6.65658340e-02 -8.22583318e-01 -6.42592669e-01
-1.14366889e+00 8.11389089e-01 5.59067130e-01 8.72783244e-01
4.39123154e-01 -1.46300328e+00 -5.72146654e-01 -9.17826965e-02
8.14413428e-02 -5.88740587e-01 -1.72508016e-01 8.74904275e-01
-7.58091867e-01 8.12889040e-02 -1.24460176e-01 -4.53323692e-01
-1.40477943e+00 6.67336643e-01 6.73135459e-01 -6.00222349e-01
-7.29343116e-01 8.79948199e-01 -8.31164658e-01 -6.56835496e-01
3.16443294e-01 -1.15531646e-01 -2.13342145e-01 3.15294653e-01
1.73050165e-01 7.01074600e-01 -5.07585611e-03 -1.15317382e-01
-7.02754915e-01 3.86127792e-02 7.07099065e-02 2.47430876e-01
1.61623120e+00 1.92976132e-01 -4.25198823e-01 5.63598931e-01
1.39626634e+00 -3.20481211e-01 -9.73678172e-01 2.27129105e-02
1.03580505e-01 -1.73921898e-01 -1.15093410e-01 -6.96290970e-01
-1.41651380e+00 3.83759707e-01 5.40359557e-01 8.48147392e-01
6.45292044e-01 2.84588009e-01 4.64545608e-01 1.61525369e-01
1.56999022e-01 -7.59441435e-01 1.09509222e-01 6.70536280e-01
8.58030975e-01 -7.26413369e-01 3.63242656e-01 -6.56681836e-01
-2.64681906e-01 1.35911143e+00 6.88722670e-01 -1.18677124e-01
1.40416479e+00 3.71314973e-01 -1.86724693e-01 -6.15658045e-01
-1.42462659e+00 -8.43220428e-02 2.34991670e-01 7.80053198e-01
5.80871284e-01 1.13681778e-01 -1.04921274e-01 1.96807578e-01
-1.96404397e-01 -4.70600635e-01 5.00190496e-01 8.54816914e-01
1.56043917e-01 -1.31344318e+00 3.69161755e-01 8.03307354e-01
-1.76998034e-01 -5.96178509e-02 -6.38452947e-01 1.03684235e+00
-5.13256907e-01 7.16297925e-01 3.79493311e-02 -8.74528706e-01
2.97489136e-01 1.76992461e-01 2.85679400e-01 -7.00614512e-01
-7.00264990e-01 -6.24896824e-01 7.29710221e-01 -5.79306901e-01
5.83500136e-03 -5.45594990e-01 -1.07844758e+00 -7.72124112e-01
-1.60206854e-01 -8.20634142e-02 5.36279500e-01 2.54100055e-01
7.76557565e-01 8.33521783e-01 4.72689956e-01 -1.02333605e+00
-1.06235117e-01 -1.14669871e+00 -5.06840587e-01 2.66767830e-01
7.92986155e-02 -7.87738025e-01 -5.84970534e-01 -5.16872406e-01] | [7.089761257171631, 6.039525032043457] |
ce8a6e4d-d46c-4b18-90a6-c4689903f34c | a-survey-on-knowledge-enhanced-multimodal | 2211.12328 | null | https://arxiv.org/abs/2211.12328v2 | https://arxiv.org/pdf/2211.12328v2.pdf | A survey on knowledge-enhanced multimodal learning | Multimodal learning has been a field of increasing interest, aiming to combine various modalities in a single joint representation. Especially in the area of visiolinguistic (VL) learning multiple models and techniques have been developed, targeting a variety of tasks that involve images and text. VL models have reached unprecedented performances by extending the idea of Transformers, so that both modalities can learn from each other. Massive pre-training procedures enable VL models to acquire a certain level of real-world understanding, although many gaps can be identified: the limited comprehension of commonsense, factual, temporal and other everyday knowledge aspects questions the extendability of VL tasks. Knowledge graphs and other knowledge sources can fill those gaps by explicitly providing missing information, unlocking novel capabilities of VL models. In the same time, knowledge graphs enhance explainability, fairness and validity of decision making, issues of outermost importance for such complex implementations. The current survey aims to unify the fields of VL representation learning and knowledge graphs, and provides a taxonomy and analysis of knowledge-enhanced VL models. | ['Giorgos Stamou', 'Maria Lymperaiou'] | 2022-11-19 | null | null | null | null | ['vision-language-navigation', 'visual-reasoning', 'conditional-image-generation', 'factual-visual-question-answering', 'visual-dialogue', 'visual-storytelling', 'visual-dialogue', 'visual-commonsense-reasoning', 'visual-reasoning', 'visual-entailment'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'natural-language-processing', 'natural-language-processing', 'reasoning', 'reasoning', 'reasoning'] | [ 7.41592124e-02 2.61046916e-01 -5.15392065e-01 -1.32729694e-01
-2.23868787e-01 -6.80217743e-01 9.07337844e-01 3.75028640e-01
-4.21534002e-01 8.43389273e-01 2.82627672e-01 -3.31773520e-01
-4.16270047e-01 -7.48762786e-01 -5.00688374e-01 -5.82762897e-01
9.59643349e-02 3.10843796e-01 1.02056280e-01 -4.21096802e-01
2.93964893e-01 5.68538487e-01 -1.80577052e+00 7.14451730e-01
8.20989788e-01 7.61647999e-01 3.58107984e-01 4.25435275e-01
-6.24489605e-01 1.32470322e+00 -4.01921928e-01 -6.33227587e-01
-3.56886864e-01 -3.50204706e-01 -1.09296882e+00 -2.71595009e-02
3.71521741e-01 -2.53396463e-02 -4.27194357e-01 8.75279486e-01
3.19786221e-01 2.36480936e-01 6.56750143e-01 -1.35089695e+00
-1.04009247e+00 6.40341759e-01 -3.02724838e-01 3.79571989e-02
7.21590161e-01 -2.95643788e-02 1.06136096e+00 -6.96475863e-01
6.71059132e-01 1.46138823e+00 3.82011235e-01 5.85734129e-01
-1.13531089e+00 -2.73762822e-01 3.23400497e-01 9.98561203e-01
-1.17104995e+00 -2.22540081e-01 7.27050781e-01 -4.69563991e-01
9.18819904e-01 2.96522051e-01 8.58154953e-01 1.17927229e+00
-5.38276397e-02 1.16086924e+00 1.42348945e+00 -7.16845453e-01
-1.66241795e-01 4.95130479e-01 1.27756685e-01 7.57467091e-01
1.25494674e-01 5.69860004e-02 -1.05352736e+00 3.26461822e-01
7.98010886e-01 2.06835717e-02 -5.39688587e-01 -6.19400620e-01
-1.48086190e+00 7.57246256e-01 4.29396868e-01 6.72990561e-01
-1.90351188e-01 5.29595204e-02 4.94065136e-01 4.54953969e-01
4.12232019e-02 4.04687673e-01 -2.96903938e-01 -8.57791584e-03
-5.63354194e-01 -2.83446331e-02 7.14693606e-01 7.06019461e-01
8.78502846e-01 6.68026730e-02 -2.24423975e-01 7.23455608e-01
2.72689849e-01 6.27748489e-01 3.89045686e-01 -7.39474177e-01
4.05753613e-01 9.34738636e-01 -1.82326362e-01 -1.00398648e+00
-5.45383155e-01 -4.14775610e-01 -6.75370634e-01 3.15867811e-01
5.58563650e-01 2.41570428e-01 -7.28277802e-01 1.80027401e+00
1.92420006e-01 -2.28718087e-01 2.61434376e-01 8.06464672e-01
1.20006585e+00 5.41031182e-01 4.20306027e-01 -2.77534634e-01
1.41139376e+00 -7.53974199e-01 -9.91937697e-01 -1.61380425e-01
5.13893247e-01 -5.03126562e-01 1.05031312e+00 3.86827946e-01
-1.00236487e+00 -5.63890040e-01 -6.52194619e-01 -3.17022026e-01
-9.46298957e-01 -4.96431515e-02 9.63160336e-01 4.85273033e-01
-1.07804239e+00 3.51076037e-01 -2.36885145e-01 -3.60777646e-01
4.68106806e-01 1.88996971e-01 -6.58528864e-01 -3.66272300e-01
-1.45009470e+00 1.45282614e+00 7.62665510e-01 2.78989166e-01
-7.78546393e-01 -5.65230727e-01 -1.12150753e+00 5.39183281e-02
6.67019367e-01 -7.90998757e-01 8.07708979e-01 -1.12008274e+00
-1.15570998e+00 1.14273632e+00 -2.05475688e-01 -1.65016681e-01
4.80995804e-01 -1.31673887e-01 -5.87132096e-01 2.19888225e-01
-1.95428237e-01 4.50190425e-01 8.55902314e-01 -1.65236151e+00
-5.31086266e-01 -5.46526790e-01 4.31833744e-01 4.85429376e-01
-3.31091464e-01 -2.30049610e-01 -1.53465793e-01 -2.34073550e-01
2.80695781e-02 -4.13899064e-01 1.91295043e-01 -6.04167879e-02
-1.81301445e-01 -3.15695882e-01 7.93105960e-01 -7.04770505e-01
1.20658207e+00 -1.83778965e+00 6.55441642e-01 9.85461324e-02
5.48213065e-01 3.75567973e-01 -7.22549781e-02 6.80749714e-01
2.00813524e-02 2.81895027e-02 2.49350294e-02 -5.34140244e-02
1.77160218e-01 3.80738676e-01 -2.09420159e-01 1.10432394e-01
-6.16554730e-03 1.33883965e+00 -9.88084137e-01 -7.05733895e-01
5.37790358e-01 5.01198947e-01 -1.92751957e-03 8.16262960e-02
-2.88853735e-01 5.31948209e-01 -2.04942375e-01 5.62047541e-01
3.32957566e-01 -2.85131395e-01 4.19197619e-01 -3.29311311e-01
-5.58363087e-02 -2.13819936e-01 -1.05508006e+00 1.79084551e+00
-4.63956207e-01 9.36906576e-01 -8.41123536e-02 -1.14433217e+00
5.76754391e-01 3.66020888e-01 7.68035427e-02 -9.51382577e-01
1.92704663e-01 -8.76665264e-02 -4.55493741e-02 -9.76692796e-01
5.48900425e-01 -4.06209797e-01 2.06464022e-01 3.18637073e-01
2.90492088e-01 -6.35449886e-02 1.44862056e-01 3.76969069e-01
4.63753790e-01 4.67571378e-01 6.35587275e-01 1.06670000e-01
7.77649820e-01 1.81472432e-02 9.24415607e-03 5.70182681e-01
-1.03527308e-01 -2.07840465e-02 6.24778450e-01 -3.39406222e-01
-5.40513873e-01 -1.13462329e+00 -2.83194613e-02 1.29042017e+00
3.28609556e-01 -1.96085200e-01 -3.67552131e-01 -6.49959207e-01
-6.74242154e-02 9.11298275e-01 -7.91514158e-01 -1.86101541e-01
-3.13622922e-01 -3.88799399e-01 4.09163266e-01 4.78494138e-01
3.92021060e-01 -1.38794267e+00 -7.69834399e-01 -5.35891354e-02
-5.18653750e-01 -1.31524372e+00 2.93318003e-01 2.47999956e-03
-8.01826775e-01 -1.32665586e+00 -5.82811892e-01 -7.08086014e-01
5.31246245e-01 4.04780954e-01 1.26200807e+00 2.61911750e-01
-2.61622399e-01 9.54208374e-01 -6.09170973e-01 -6.58429682e-01
-4.63936359e-01 -3.45552057e-01 -1.54030204e-01 6.03415780e-02
3.32438558e-01 -4.78993058e-01 -6.70705140e-02 4.65483451e-03
-9.97930884e-01 2.80436993e-01 5.89748919e-01 8.67732167e-01
3.49329799e-01 -2.07336172e-01 7.03484714e-01 -6.55375838e-01
6.40189826e-01 -4.38800603e-01 -2.55190045e-01 8.00599754e-01
-3.21978599e-01 1.03802204e-01 2.55192608e-01 -5.06865501e-01
-1.40092957e+00 -3.63341093e-01 1.32828310e-01 -3.40787619e-01
-2.98436522e-01 8.08614731e-01 -2.69112647e-01 -2.68351972e-01
5.79684794e-01 4.89145219e-01 2.23591283e-01 -3.09860945e-01
9.72471833e-01 2.24496573e-01 4.56463158e-01 -7.29339480e-01
5.64109206e-01 3.09067428e-01 6.84578046e-02 -9.71498907e-01
-9.13851500e-01 -3.31045002e-01 -8.53416502e-01 -5.88921368e-01
8.01782548e-01 -7.28248060e-01 -1.09245586e+00 2.65400529e-01
-1.19243169e+00 -1.68797314e-01 -4.50407356e-01 5.54264605e-01
-5.86937249e-01 5.84407747e-01 -3.12625676e-01 -8.78731549e-01
9.91500020e-02 -8.46530020e-01 5.79094827e-01 3.51342440e-01
-9.48969945e-02 -1.47356367e+00 -1.73594475e-01 6.67962074e-01
3.30455929e-01 2.87390411e-01 1.35855401e+00 -6.21148407e-01
-6.60108685e-01 4.73441072e-02 -4.23411667e-01 2.60886520e-01
-2.72513982e-02 -3.58063281e-01 -1.24264336e+00 6.04564063e-02
-1.92756966e-01 -7.85336912e-01 8.90393734e-01 1.97661996e-01
9.75416422e-01 4.65894758e-04 -2.35945195e-01 1.20060906e-01
1.37138820e+00 1.57325998e-01 6.64656103e-01 2.12980375e-01
8.10803115e-01 1.07592273e+00 3.83270830e-01 2.02289343e-01
7.27351487e-01 4.81285661e-01 7.20442653e-01 -1.33960187e-01
-4.14068490e-01 -2.33064160e-01 1.70159474e-01 7.33032107e-01
-6.18936360e-01 -1.42528355e-01 -1.00048566e+00 5.57277322e-01
-1.96953738e+00 -1.35619843e+00 -1.39494970e-01 2.10912085e+00
7.67699182e-01 -3.04214537e-01 -1.80486962e-01 2.14319810e-01
5.40748060e-01 1.79342851e-01 -4.39402372e-01 -2.19362959e-01
-4.14493591e-01 -1.38128042e-01 -1.27015576e-01 5.26482522e-01
-6.85963213e-01 1.18889344e+00 6.79392099e+00 8.84413719e-01
-8.52549791e-01 -2.00762041e-02 3.93944867e-02 3.49078268e-01
-6.15670085e-01 -1.57993764e-01 -4.49774951e-01 -1.32763267e-01
6.10117674e-01 -1.59618691e-01 4.71895993e-01 4.37126786e-01
-3.26658674e-02 -4.47417885e-01 -1.17257214e+00 1.20744419e+00
4.61850882e-01 -1.39386857e+00 5.31852603e-01 -3.27361785e-02
4.75538850e-01 -4.63759154e-01 3.92905138e-02 4.78359431e-01
4.55099763e-03 -1.24422765e+00 6.81618929e-01 8.84322762e-01
6.91342056e-01 -6.25934541e-01 7.50715792e-01 4.54567283e-01
-1.12706113e+00 -1.13593839e-01 -1.92921251e-01 -2.42871329e-01
2.09855229e-01 2.38668263e-01 -4.74363565e-01 1.05423403e+00
4.27896857e-01 6.57066286e-01 -7.72576630e-01 9.90617812e-01
-5.01120448e-01 9.04881805e-02 2.07528859e-01 -9.98406634e-02
-2.98511628e-02 -1.32317722e-01 3.50426406e-01 1.14349818e+00
1.93694551e-02 1.41180977e-01 -5.21148220e-02 9.00185049e-01
1.45303728e-02 3.67194474e-01 -8.47075105e-01 -2.27463752e-01
1.93041876e-01 1.05573690e+00 -5.04413247e-01 -3.42922688e-01
-6.70353115e-01 6.95578456e-01 5.42263746e-01 4.41883892e-01
-7.24225223e-01 -7.37509057e-02 3.91243339e-01 -2.46300176e-01
-7.87260383e-02 -3.01207215e-01 -1.31057799e-01 -1.31372571e+00
-2.70963579e-01 -8.96289349e-01 5.20427108e-01 -1.06749690e+00
-1.23926330e+00 3.15332025e-01 1.79120272e-01 -1.01264358e+00
-8.96749869e-02 -9.00663257e-01 -1.44303739e-01 6.83657050e-01
-1.81150091e+00 -1.67191148e+00 -4.54205126e-01 1.11680520e+00
4.21533406e-01 -9.38858539e-02 9.39585924e-01 -4.83129099e-02
-1.50158450e-01 1.89822719e-01 -2.80359268e-01 -6.06103428e-02
6.35367155e-01 -1.21110308e+00 -4.78991568e-01 4.44341004e-01
4.45435226e-01 5.79086423e-01 5.31155348e-01 -4.76045758e-01
-1.54240179e+00 -4.08912778e-01 1.02908576e+00 -7.78137922e-01
6.73112929e-01 -1.67472642e-02 -1.18142700e+00 7.58055925e-01
3.36353391e-01 -2.96572268e-01 9.17189240e-01 4.62577015e-01
-6.79039478e-01 1.27252504e-01 -9.21057224e-01 7.45453298e-01
1.09973443e+00 -8.85002196e-01 -1.01135004e+00 2.71776110e-01
5.95182836e-01 -2.86643624e-01 -8.54551494e-01 6.19468316e-02
6.12535357e-01 -1.17337871e+00 1.03591490e+00 -8.13231111e-01
3.83744806e-01 -1.74871936e-01 -9.76562947e-02 -1.21988511e+00
-1.53451428e-01 -1.34664804e-01 -3.73025328e-01 1.14007139e+00
3.33662063e-01 -6.06077313e-01 3.65448713e-01 4.85737950e-01
5.44535555e-02 -3.88959974e-01 -6.74001992e-01 -6.34621799e-01
2.35851258e-02 -7.76626885e-01 3.21640521e-01 1.27672040e+00
4.97345328e-01 5.29803455e-01 -5.85479677e-01 -2.68772885e-04
6.23537183e-01 1.57434657e-01 5.39235651e-01 -1.52356970e+00
-1.41180102e-02 -5.60307264e-01 -4.79746878e-01 -7.75835395e-01
4.28980470e-01 -1.10930502e+00 -5.09148777e-01 -1.96861231e+00
3.49652082e-01 9.41796899e-02 -2.41776273e-01 7.79179871e-01
-5.36064766e-02 3.20703872e-02 5.63701451e-01 1.46419778e-01
-7.48312593e-01 5.32399118e-01 1.72507107e+00 -1.05139829e-01
-2.18362939e-02 -4.10675555e-01 -6.36887312e-01 7.76188254e-01
6.11544907e-01 6.06923252e-02 -7.43496835e-01 -4.83620882e-01
4.68598425e-01 2.16152966e-01 7.15933561e-01 -7.01069832e-01
3.46817970e-01 -3.71452987e-01 4.21524495e-01 -4.58974004e-01
4.94601250e-01 -1.07066703e+00 4.17720228e-02 2.95961559e-01
-3.54376704e-01 -1.25357166e-01 4.24549043e-01 7.58945584e-01
-4.79285955e-01 -1.77020177e-01 5.31165481e-01 -2.77028561e-01
-1.51383412e+00 -4.53459062e-02 -3.79767627e-01 -4.29362394e-02
1.12706041e+00 -4.32516754e-01 -5.36568642e-01 -5.34150124e-01
-1.09577453e+00 2.56838113e-01 1.49379387e-01 6.74015164e-01
7.71598041e-01 -1.10922265e+00 -4.31085944e-01 -5.73941469e-02
4.43109870e-01 -4.71312791e-01 6.86135232e-01 1.02058041e+00
-1.90811053e-01 5.81122220e-01 -5.45244277e-01 -4.29400742e-01
-1.27859223e+00 7.10478425e-01 3.09100240e-01 -2.07795903e-01
-4.87718642e-01 5.28948247e-01 2.55371481e-01 -3.67371291e-01
3.34491342e-01 5.08049242e-02 -8.38943005e-01 5.37140787e-01
6.56794667e-01 4.05019134e-01 -2.83911973e-01 -7.45102525e-01
-2.98279226e-01 6.95266783e-01 2.30868295e-01 -4.94957855e-03
1.00169206e+00 -3.54577988e-01 -3.75282675e-01 7.42540359e-01
6.34190261e-01 -1.09429181e-01 -7.47906029e-01 -5.56407273e-01
3.44679132e-02 -2.64955014e-01 -1.63790375e-01 -1.29157174e+00
-8.10474098e-01 1.27302086e+00 3.26858014e-01 2.76371479e-01
1.05048263e+00 2.89512098e-01 1.55554518e-01 4.56893623e-01
4.82922375e-01 -9.36708808e-01 2.30140954e-01 5.26782274e-01
9.89972830e-01 -1.46383047e+00 8.68599582e-03 -3.66802156e-01
-9.45309401e-01 1.45082128e+00 5.26960194e-01 6.64480984e-01
2.34880209e-01 -2.04868123e-01 1.43857375e-02 -4.23236877e-01
-5.10095835e-01 -7.48640716e-01 5.65797806e-01 1.03408837e+00
4.88920987e-01 8.21760073e-02 -2.73694277e-01 4.82652009e-01
1.26777843e-01 -5.42619042e-02 3.42974991e-01 7.98025966e-01
-5.18215299e-01 -1.08090591e+00 -4.12920833e-01 6.73814416e-02
-4.85390387e-02 -8.18360299e-02 -3.55812043e-01 1.20014155e+00
1.38850808e-01 9.79873598e-01 -3.69660974e-01 -2.31429994e-01
3.89287204e-01 3.30319405e-01 8.27186048e-01 -3.17419767e-01
-4.21924382e-01 -3.12402189e-01 2.11266860e-01 -5.78027546e-01
-8.84918332e-01 -3.19207370e-01 -1.11158872e+00 -3.73342603e-01
-1.61699936e-01 4.00648974e-02 6.74134076e-01 1.32864928e+00
-2.21173838e-02 6.43868744e-01 -5.11709489e-02 -8.38812947e-01
-2.65310615e-01 -6.29259467e-01 -6.45604730e-01 4.41531509e-01
2.74679601e-01 -7.85404086e-01 -2.90962696e-01 1.48179606e-01] | [10.65420913696289, 1.8525327444076538] |
ad6e3b1e-61dd-489d-b8ea-27764466d2d8 | handy-towards-a-high-fidelity-3d-hand-shape | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Potamias_Handy_Towards_a_High_Fidelity_3D_Hand_Shape_and_Appearance_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Potamias_Handy_Towards_a_High_Fidelity_3D_Hand_Shape_and_Appearance_CVPR_2023_paper.pdf | Handy: Towards a High Fidelity 3D Hand Shape and Appearance Model | Over the last few years, with the advent of virtual and augmented reality, an enormous amount of research has been focused on modeling, tracking and reconstructing human hands. Given their power to express human behavior, hands have been a very important, but challenging component of the human body. Currently, most of the state-of-the-art reconstruction and pose estimation methods rely on the low polygon MANO model. Apart from its low polygon count, MANO model was trained with only 31 adult subjects, which not only limits its expressive power but also imposes unnecessary shape reconstruction constraints on pose estimation methods. Moreover, hand appearance remains almost unexplored and neglected from the majority of hand reconstruction methods. In this work, we propose "Handy", a large-scale model of the human hand, modeling both shape and appearance composed of over 1200 subjects which we make publicly available for the benefit of the research community. In contrast to current models, our proposed hand model was trained on a dataset with large diversity in age, gender, and ethnicity, which tackles the limitations of MANO and accurately reconstructs out-of-distribution samples. In order to create a high quality texture model, we trained a powerful GAN, which preserves high frequency details and is able to generate high resolution hand textures. To showcase the capabilities of the proposed model, we built a synthetic dataset of textured hands and trained a hand pose estimation network to reconstruct both the shape and appearance from single images. As it is demonstrated in an extensive series of quantitative as well as qualitative experiments, our model proves to be robust against the state-of-the-art and realistically captures the 3D hand shape and pose along with a high frequency detailed texture even in adverse "in-the-wild" conditions. | ['Stefanos Zafeiriou', 'Vasileios Triantafyllou', 'Stylianos Moschoglou', 'Stylianos Ploumpis', 'Rolandos Alexandros Potamias'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['hand-pose-estimation'] | ['computer-vision'] | [-4.66803648e-02 1.30896300e-01 1.59287676e-02 1.32059321e-01
-2.26156861e-01 -3.08806866e-01 3.83885324e-01 -6.34965301e-01
2.09314916e-02 6.42839551e-01 1.95094392e-01 2.33343765e-01
3.34046707e-02 -7.69546449e-01 -6.08864248e-01 -5.65827549e-01
9.73865688e-02 9.22757447e-01 4.43801358e-02 -2.98857540e-01
-3.25724065e-01 6.61342621e-01 -1.82054806e+00 -2.13056058e-02
5.18673599e-01 1.02139235e+00 9.96318385e-02 3.05370122e-01
2.77252764e-01 3.80445927e-01 -5.84549904e-01 -7.25435078e-01
3.06785017e-01 -2.29004070e-01 -4.61969703e-01 2.52568990e-01
6.28861129e-01 -6.13900006e-01 -5.92464030e-01 5.92072427e-01
9.92489457e-01 -8.99589360e-02 4.76842791e-01 -7.04637289e-01
-5.08421123e-01 2.38939255e-01 -5.78478754e-01 -5.25771379e-01
6.55176401e-01 1.89205334e-01 7.10847378e-01 -7.80455887e-01
1.10876846e+00 1.19712555e+00 8.52760494e-01 8.77406418e-01
-1.26725984e+00 -6.83479786e-01 1.27152011e-01 -2.99281716e-01
-1.45773637e+00 -1.64037213e-01 1.08084500e+00 -4.15715933e-01
3.84051502e-01 4.84210312e-01 1.22023797e+00 1.90181661e+00
3.05168405e-02 9.13757145e-01 1.43479121e+00 -5.77704430e-01
-5.23340888e-02 1.07697345e-01 -4.06106651e-01 8.96166205e-01
-9.00198147e-02 1.98891953e-01 -5.50688863e-01 -1.81646779e-01
1.32468450e+00 3.98517326e-02 -3.58510166e-01 -5.11959970e-01
-1.35953772e+00 3.10680091e-01 2.65633643e-01 1.80202454e-01
-3.84808630e-01 1.64465949e-01 1.23400807e-01 -1.60868108e-01
5.30761361e-01 6.00905530e-03 -2.93899536e-01 -2.89444476e-01
-7.51634300e-01 5.57110190e-01 7.68871546e-01 8.11864495e-01
2.51805395e-01 1.68012515e-01 -2.63874799e-01 1.02603889e+00
1.36115924e-01 9.30658102e-01 1.88308716e-01 -6.33320630e-01
3.83577675e-01 5.17324388e-01 9.68013331e-02 -1.12827063e+00
-4.56086665e-01 -6.82235062e-01 -1.08868396e+00 4.01960403e-01
7.57996798e-01 1.26892282e-02 -1.06058180e+00 1.84305811e+00
5.64294398e-01 -2.35089853e-01 -7.47499943e-01 1.25250030e+00
6.14715874e-01 2.78077424e-02 -2.20092416e-01 1.26201034e-01
1.40334916e+00 -5.03018856e-01 -6.35837615e-01 -5.19856587e-02
-2.13620931e-01 -9.77330983e-01 1.33358920e+00 6.54494226e-01
-1.07868350e+00 -6.57738149e-01 -6.11109138e-01 1.55700102e-01
4.80402671e-02 4.33103323e-01 8.46412301e-01 8.74858618e-01
-7.80821025e-01 5.71011007e-01 -1.02171814e+00 -4.68540698e-01
5.99928498e-01 3.72156560e-01 -6.37326121e-01 -1.87252864e-01
-8.60391378e-01 7.50523925e-01 -2.15224370e-01 3.76197547e-01
-6.63420081e-01 -5.96121073e-01 -5.08605599e-01 -1.63658351e-01
4.99830067e-01 -9.12746370e-01 8.08744013e-01 -7.57235646e-01
-1.92414308e+00 8.66502941e-01 5.41053265e-02 2.33802751e-01
1.09253502e+00 -2.89623529e-01 -2.24413037e-01 5.73472269e-02
-3.08607012e-01 3.70796561e-01 1.03974032e+00 -1.51039445e+00
-7.24515915e-02 -7.98784375e-01 -5.02941981e-02 -7.20458329e-02
-1.98349640e-01 -5.80950826e-02 -8.31313729e-01 -1.20349789e+00
6.83665574e-02 -1.23029912e+00 7.62842223e-02 2.68681198e-01
-4.53328669e-01 3.32462549e-01 6.05620146e-01 -1.20066595e+00
1.04295802e+00 -1.97874331e+00 5.75555027e-01 4.11321312e-01
3.20082068e-01 1.54966399e-01 7.72985518e-02 2.84063578e-01
2.72698522e-01 -3.13022554e-01 -5.64758517e-02 -7.06267953e-01
2.99757063e-01 1.68955490e-01 -2.05833301e-01 5.14603794e-01
-1.70310989e-01 9.40185964e-01 -5.92854202e-01 -5.39165974e-01
3.73648435e-01 1.25044990e+00 -6.57795012e-01 2.33646616e-01
-2.27554455e-01 1.08859038e+00 -4.98619914e-01 1.04160786e+00
5.98007560e-01 -1.76264048e-02 4.30877000e-01 -2.91274846e-01
2.70681441e-01 -4.94542748e-01 -1.10643005e+00 1.83287930e+00
-6.55290365e-01 2.04267994e-01 2.79701799e-01 -4.38161165e-01
7.35326767e-01 5.48421323e-01 5.51262438e-01 -7.82064557e-01
4.76074904e-01 3.03414553e-01 -2.22006321e-01 -3.91586870e-01
1.34207428e-01 -1.85200065e-01 1.52376086e-01 2.90299654e-01
3.45696807e-02 -1.04930185e-01 -3.40192765e-01 -2.94403404e-01
5.85422158e-01 6.59977436e-01 -2.22796872e-02 2.03340091e-02
2.34725758e-01 -4.67205614e-01 1.70160338e-01 4.08117771e-01
7.00303316e-02 1.15057504e+00 1.88299745e-01 -5.04501224e-01
-1.14383578e+00 -1.05335045e+00 -2.00103864e-01 7.93795526e-01
-1.18741058e-01 -7.72937611e-02 -1.05960405e+00 -4.39368248e-01
2.74512023e-01 1.29712909e-03 -1.01559627e+00 3.17932785e-01
-6.77014887e-01 -6.05938733e-01 5.51103115e-01 5.86622238e-01
5.09382725e-01 -1.18608427e+00 -6.37603521e-01 5.82587421e-02
-2.33327180e-01 -1.04767776e+00 -4.78061706e-01 -4.62531239e-01
-4.89901513e-01 -1.11786473e+00 -1.34659922e+00 -5.74603498e-01
7.25656211e-01 -4.52374578e-01 1.07939029e+00 -3.09055261e-02
-6.43832326e-01 4.95919287e-01 -3.77222300e-01 -3.23321462e-01
-1.43727049e-01 1.18400678e-01 2.14689881e-01 3.96300614e-01
-2.88223922e-01 -9.86885548e-01 -7.94997096e-01 5.59559643e-01
-4.58444804e-01 6.05814680e-02 6.24957263e-01 9.39311922e-01
7.30474651e-01 -2.83292532e-01 2.88483977e-01 -8.00353825e-01
3.07139575e-01 1.86876431e-02 -3.73673290e-01 2.42855087e-01
-1.81506008e-01 -1.87798470e-01 5.95858574e-01 -8.88271153e-01
-1.22342932e+00 2.20383719e-01 -3.49330008e-01 -6.59101307e-01
-1.70487255e-01 1.06396586e-01 -3.14120978e-01 -2.10850775e-01
5.34502149e-01 3.21122557e-01 2.79081732e-01 -9.36810851e-01
1.61742955e-01 6.05973423e-01 5.15856326e-01 -1.00753105e+00
6.85091496e-01 7.40678549e-01 9.16475803e-02 -8.87875855e-01
-5.22184551e-01 2.44365986e-02 -7.68917978e-01 -4.27140266e-01
4.61137474e-01 -6.27537847e-01 -1.14524686e+00 8.73584032e-01
-8.80524039e-01 -6.38215065e-01 -2.88793683e-01 2.92738110e-01
-5.19312084e-01 2.85917699e-01 -4.92838979e-01 -1.09707296e+00
-3.84980112e-01 -1.09980822e+00 1.41255760e+00 -2.18899816e-01
-3.92613351e-01 -7.86649823e-01 -5.14104925e-02 6.23235285e-01
4.69984114e-01 7.17016578e-01 6.54836476e-01 2.94333458e-01
-3.50596100e-01 -4.23332870e-01 2.38415543e-02 1.72144726e-01
9.91614014e-02 -2.94024199e-01 -1.08334398e+00 -5.00054300e-01
-1.82336375e-01 -3.19049895e-01 3.48787457e-01 3.18832368e-01
1.16028237e+00 -1.76219761e-01 -1.65465370e-01 6.28147602e-01
1.11270881e+00 -3.06321174e-01 6.45946324e-01 -1.27826452e-01
1.06040227e+00 7.35545635e-01 2.59438068e-01 6.83436692e-01
2.54386753e-01 1.11660326e+00 3.98322403e-01 -1.47554263e-01
-7.01694429e-01 -4.33211803e-01 -6.48017898e-02 7.44261563e-01
-1.01391172e+00 -3.36506292e-02 -7.68138707e-01 2.69719988e-01
-1.52081919e+00 -7.93793499e-01 3.04657936e-01 2.27475452e+00
8.09234440e-01 -1.18427642e-01 5.19481778e-01 2.51060694e-01
4.18892741e-01 -4.97760549e-02 -3.73925567e-01 1.81661263e-01
-2.03674525e-01 7.31853843e-01 1.41155824e-01 2.19632909e-01
-8.62540603e-01 7.36367702e-01 5.80009174e+00 8.52873385e-01
-1.35794437e+00 1.48643434e-01 3.10370773e-01 -1.95159748e-01
-8.71520489e-02 -4.74213660e-01 -4.01617497e-01 5.19634068e-01
5.09300865e-02 5.80869198e-01 8.94246936e-01 8.02271426e-01
-7.07942396e-02 7.68945217e-02 -8.21973085e-01 1.16942942e+00
1.69980109e-01 -8.75723541e-01 1.08641349e-01 3.69480968e-01
6.54842019e-01 -4.29408699e-01 3.57382298e-01 4.13378216e-02
-1.67431444e-01 -1.23894453e+00 1.20960343e+00 7.08444893e-01
1.40602636e+00 -6.01791024e-01 6.32485151e-01 4.13336992e-01
-1.19862998e+00 8.79160687e-02 -1.89438313e-02 -8.96602031e-03
1.28976882e-01 5.16520023e-01 -4.30648506e-01 6.53478444e-01
6.91120327e-01 2.45360047e-01 -4.65423673e-01 5.34198165e-01
-9.85868648e-02 3.58573496e-01 -4.58608836e-01 1.47777468e-01
-4.47504371e-01 1.68931298e-02 3.67966712e-01 8.96914721e-01
2.62300462e-01 2.44209338e-02 3.02818179e-01 8.96675169e-01
4.05512750e-03 8.03091377e-02 -2.55093992e-01 4.40739579e-02
1.85933381e-01 9.85483348e-01 -6.87997520e-01 5.12529574e-02
-2.31529087e-01 1.16834056e+00 3.37610334e-01 3.01454216e-01
-6.57510519e-01 -7.02321678e-02 3.67008418e-01 7.32628942e-01
3.33019137e-01 -2.80547768e-01 -2.95753777e-01 -1.27975535e+00
4.27654117e-01 -1.13708115e+00 -1.30813688e-01 -5.46219409e-01
-1.32451439e+00 8.15520287e-01 -1.45711139e-01 -1.04307425e+00
-1.81916550e-01 -7.46994734e-01 -3.93063761e-02 1.09157157e+00
-9.16027009e-01 -2.01903915e+00 -6.29069924e-01 6.70465708e-01
2.07298636e-01 -1.46047711e-01 1.14609540e+00 5.57301581e-01
-4.07179028e-01 8.76677155e-01 -1.50586441e-01 1.57073811e-01
5.80886900e-01 -9.39955711e-01 3.57457727e-01 2.41235152e-01
1.06377676e-02 8.37848663e-01 7.13229477e-01 -7.43051708e-01
-1.72462189e+00 -6.04605198e-01 2.79932946e-01 -7.22995281e-01
2.32250482e-01 -6.82969093e-01 -5.82276344e-01 5.90563118e-01
-4.06516701e-01 3.34117502e-01 2.63065368e-01 1.90032765e-01
-4.50557888e-01 -1.65170655e-04 -1.25879943e+00 5.72334528e-01
1.37560606e+00 -5.74903607e-01 -2.17545450e-01 1.42724305e-01
-8.45542029e-02 -9.00331974e-01 -1.05997026e+00 5.34442306e-01
1.49705696e+00 -1.07035148e+00 1.13022470e+00 -2.04504699e-01
1.92953885e-01 4.61850092e-02 -1.45209789e-01 -1.02141833e+00
-5.21771610e-02 -6.60745919e-01 -4.56880212e-01 1.26905942e+00
-1.58186495e-01 -4.58851099e-01 1.24014032e+00 4.44156855e-01
3.19911152e-01 -1.10422611e+00 -9.59778786e-01 -7.51406193e-01
-2.39507314e-02 -3.13973486e-01 6.83519185e-01 7.69855678e-01
-3.52476358e-01 -1.22838743e-01 -8.06814313e-01 -1.08005621e-01
8.26522350e-01 1.29837841e-01 1.09150898e+00 -1.42936492e+00
-6.99951351e-01 -2.70395160e-01 -3.42043996e-01 -8.39195073e-01
1.78283721e-01 -5.53646684e-01 -2.38898680e-01 -1.11748576e+00
2.23551989e-01 -6.57314897e-01 1.68772787e-01 4.09270763e-01
4.60776873e-02 7.25854337e-01 1.26500249e-01 1.69807732e-01
1.27525687e-01 5.08132219e-01 1.77880585e+00 2.25705318e-02
-7.00649172e-02 1.81297958e-01 -2.23502517e-01 6.58184826e-01
3.43767852e-01 2.39638295e-02 -2.21678749e-01 -3.52781326e-01
1.55749843e-01 2.29901612e-01 6.83629155e-01 -9.13082540e-01
-2.65061498e-01 1.20309077e-01 6.29946172e-01 -2.08654106e-01
7.02707827e-01 -9.92643476e-01 7.31139302e-01 3.61994803e-01
1.91325083e-01 -2.38026783e-01 4.67637591e-02 3.58923852e-01
6.84765503e-02 4.90577221e-01 6.96642399e-01 -1.90982163e-01
-1.46769866e-01 5.25551260e-01 2.43113071e-01 -1.31678537e-01
6.52630031e-01 -4.00043994e-01 1.42550856e-01 -3.05360824e-01
-8.85659754e-01 -3.62035334e-01 7.83627987e-01 4.03737307e-01
3.99197310e-01 -1.27462244e+00 -6.45635724e-01 5.15437543e-01
-4.67706993e-02 1.18743159e-01 6.28524363e-01 8.42785239e-01
-7.16516495e-01 1.88920647e-01 -4.29675788e-01 -4.80901629e-01
-1.21630442e+00 4.81102049e-01 2.59367704e-01 -2.20087394e-01
-1.00817358e+00 5.98139644e-01 1.21500403e-01 -6.58407390e-01
4.96570021e-01 -1.18634701e-01 8.59093741e-02 -1.46543205e-01
3.01243037e-01 5.13576448e-01 -1.57231011e-03 -9.24326241e-01
-2.74930924e-01 1.04699671e+00 4.28851902e-01 -1.42626420e-01
1.28830647e+00 1.98935658e-01 4.21594549e-03 2.13416055e-01
7.51803160e-01 4.70918626e-01 -1.38131392e+00 -3.62591911e-03
-7.62386739e-01 -7.58944392e-01 -2.37132654e-01 -1.13951337e+00
-1.38879800e+00 9.84992862e-01 6.85334623e-01 -2.98847646e-01
9.52915668e-01 -4.96398956e-02 9.27898347e-01 -6.78053126e-02
1.05512571e+00 -9.54497755e-01 1.75595820e-01 1.54369026e-01
1.16980815e+00 -9.39869702e-01 -1.01706525e-02 -8.06185246e-01
-5.03333688e-01 8.49399269e-01 3.70554507e-01 4.95391116e-02
6.31629467e-01 3.70373815e-01 7.53667876e-02 -1.25508621e-01
1.26303717e-01 8.28306451e-02 3.92061889e-01 8.99026632e-01
3.48654598e-01 2.93170959e-01 -3.75550091e-02 7.97159493e-01
-4.45710391e-01 2.59895444e-01 -1.48128346e-01 8.05333316e-01
3.08180481e-01 -1.31355679e+00 -6.69834673e-01 2.02980623e-01
-6.19187355e-01 3.40091050e-01 -3.68328840e-01 1.02475798e+00
3.15542519e-01 3.28672618e-01 -3.76479238e-01 -6.09412730e-01
6.52945220e-01 -8.79290402e-02 1.27482283e+00 -3.64213705e-01
-5.79077601e-01 7.81503096e-02 -8.81191567e-02 -5.81401646e-01
-2.12625414e-01 -5.09065747e-01 -7.37699032e-01 -5.62485993e-01
-2.53088593e-01 -4.39552248e-01 6.82369173e-01 7.68705130e-01
2.73215234e-01 5.55817008e-01 2.72162288e-01 -1.49007130e+00
-6.73716366e-01 -9.76248264e-01 -1.03321528e+00 7.50919104e-01
1.17611676e-01 -1.20460451e+00 1.02153816e-03 -9.41600353e-02] | [7.044389724731445, -1.189358115196228] |
b02cfd71-096b-430b-a29f-18e91c8ee2f7 | coupled-oscillatory-recurrent-neural-network | 2010.00951 | null | https://arxiv.org/abs/2010.00951v2 | https://arxiv.org/pdf/2010.00951v2.pdf | Coupled Oscillatory Recurrent Neural Network (coRNN): An accurate and (gradient) stable architecture for learning long time dependencies | Circuits of biological neurons, such as in the functional parts of the brain can be modeled as networks of coupled oscillators. Inspired by the ability of these systems to express a rich set of outputs while keeping (gradients of) state variables bounded, we propose a novel architecture for recurrent neural networks. Our proposed RNN is based on a time-discretization of a system of second-order ordinary differential equations, modeling networks of controlled nonlinear oscillators. We prove precise bounds on the gradients of the hidden states, leading to the mitigation of the exploding and vanishing gradient problem for this RNN. Experiments show that the proposed RNN is comparable in performance to the state of the art on a variety of benchmarks, demonstrating the potential of this architecture to provide stable and accurate RNNs for processing complex sequential data. | ['Siddhartha Mishra', 'T. Konstantin Rusch'] | 2020-10-02 | null | https://openreview.net/forum?id=F3s69XzWOia | https://openreview.net/pdf?id=F3s69XzWOia | iclr-2021-1 | ['sequential-image-classification'] | ['computer-vision'] | [ 1.45414501e-01 2.19716489e-01 3.22381228e-01 -7.71033904e-03
3.67399126e-01 -5.48988461e-01 4.30169940e-01 -3.91968608e-01
-4.69589472e-01 5.42167962e-01 -7.82817900e-02 -2.48690978e-01
5.16201509e-03 -4.80137169e-01 -8.60515594e-01 -9.54037189e-01
-2.10365370e-01 7.84055814e-02 2.00453207e-01 -6.85750782e-01
1.57071948e-02 7.68558562e-01 -1.35842729e+00 -3.12503171e-03
3.88445467e-01 9.79120016e-01 4.53097513e-03 8.44434261e-01
2.19747961e-01 1.17131710e+00 -5.68188667e-01 4.12363797e-01
2.28526130e-01 -6.38783991e-01 -3.77996862e-01 -5.17553389e-01
1.89251639e-02 5.71975484e-02 -7.99473405e-01 8.63970995e-01
4.48807061e-01 1.95788801e-01 5.98892093e-01 -8.75016332e-01
-6.41273379e-01 8.61162126e-01 3.40345174e-01 3.54456246e-01
-2.26523012e-01 1.04549602e-01 8.02956223e-01 -7.71592319e-01
5.59753358e-01 1.05438209e+00 8.69699359e-01 9.51243460e-01
-1.73873532e+00 -6.15426421e-01 1.15183391e-01 -5.89782894e-01
-1.49969149e+00 -7.92044878e-01 3.86495262e-01 -2.64606267e-01
1.49999833e+00 1.09923016e-02 6.97204769e-01 1.08775210e+00
8.74620378e-01 2.56893903e-01 5.60029745e-01 -1.83959007e-01
5.14916003e-01 -3.84285003e-02 3.14249098e-01 8.00290227e-01
2.60782897e-01 4.21102308e-02 -6.65547550e-01 -5.02502136e-02
1.03278780e+00 3.39168191e-01 -2.76167154e-01 -8.22198018e-02
-1.16227460e+00 5.44029295e-01 7.49145746e-01 6.27036273e-01
-3.70062172e-01 9.36171532e-01 2.44912967e-01 4.56993341e-01
2.07453847e-01 7.02755690e-01 -3.74503314e-01 2.20254496e-01
-7.18378186e-01 1.66980341e-01 1.09814429e+00 9.56871927e-01
5.12003779e-01 4.24332827e-01 -1.62580401e-01 3.85017365e-01
3.40349436e-01 5.92115760e-01 7.46259570e-01 -9.28520322e-01
1.74578670e-02 6.14956379e-01 -3.80649939e-02 -8.32692087e-01
-8.03946733e-01 -6.49731576e-01 -1.52081800e+00 -4.73710746e-02
3.59743178e-01 -3.95667225e-01 -9.40604448e-01 2.02661753e+00
-2.29105562e-01 1.09450907e-01 2.87694186e-01 5.96198618e-01
3.21413398e-01 1.19682419e+00 -4.52046961e-01 -3.51856679e-01
9.43394780e-01 -6.99912667e-01 -8.67997050e-01 -2.33404770e-01
4.09823805e-01 1.38037009e-02 4.69943196e-01 1.22982651e-01
-1.27268016e+00 -4.09327477e-01 -9.91016865e-01 -1.79570049e-01
-3.37468892e-01 9.60198697e-03 3.13263565e-01 1.95320010e-01
-1.65993631e+00 1.03158092e+00 -1.30463815e+00 -1.74173757e-01
3.28390556e-03 8.28226745e-01 -2.37115063e-02 8.15930843e-01
-1.28358960e+00 5.97106159e-01 1.99129120e-01 9.08546865e-01
-9.74612832e-01 -9.31706846e-01 -5.70449710e-01 3.12655121e-01
-2.79917538e-01 -5.76326072e-01 1.12922215e+00 -7.67711759e-01
-1.87025523e+00 3.54148477e-01 -3.72564316e-01 -9.64715660e-01
3.16136122e-01 -8.14814940e-02 -1.20570272e-01 -2.55134962e-02
-3.75093341e-01 5.60136259e-01 8.01317573e-01 -5.34235358e-01
-1.77750774e-02 -1.57838479e-01 -2.69362867e-01 -3.22712898e-01
-6.25207543e-01 -2.84381926e-01 -3.07077095e-02 -3.96085203e-01
2.53249854e-01 -1.28113163e+00 -7.95693815e-01 -3.03981714e-02
-3.81576151e-01 -5.43654338e-02 4.83185500e-01 -2.05235854e-01
1.27902663e+00 -2.15986252e+00 4.77935493e-01 2.24970326e-01
3.51452053e-01 -3.15739177e-02 6.18140511e-02 4.96581703e-01
-1.42256813e-02 1.72566801e-01 -2.80596107e-01 -5.14907479e-01
-1.39642954e-01 3.55259389e-01 -7.97867596e-01 5.42967618e-01
5.19101679e-01 9.88018572e-01 -7.86344588e-01 1.24548331e-01
-3.77466321e-01 8.56058061e-01 -4.00659531e-01 2.35140294e-01
-1.94591433e-01 4.37882483e-01 -3.39077562e-01 5.68136163e-02
1.19695487e-02 -3.35525900e-01 1.73530996e-01 1.44019261e-01
-4.60779876e-01 5.42657793e-01 -1.06504798e+00 1.29679036e+00
-5.37559390e-01 1.07687438e+00 -6.01985790e-02 -9.17892277e-01
9.67079759e-01 5.48657596e-01 3.41109425e-01 -4.91316497e-01
3.12643349e-01 2.68623888e-01 3.01753432e-01 1.10916063e-01
1.15560949e-01 4.65136347e-03 -4.24125493e-02 4.75354940e-01
2.17248425e-01 -1.23004407e-01 3.13906491e-01 4.98174578e-02
1.31876683e+00 -2.04359949e-01 -1.69500932e-01 -7.89767385e-01
5.01717806e-01 -5.64143956e-01 6.78046644e-01 8.50507319e-01
1.13229454e-01 3.34156901e-01 9.11522627e-01 -4.18197930e-01
-1.21420074e+00 -8.20986569e-01 -2.61616051e-01 7.08140552e-01
-2.00771973e-01 -9.87675935e-02 -8.40246499e-01 4.60471869e-01
-1.69315889e-01 7.65153915e-02 -7.73925424e-01 -4.20168698e-01
-8.67268324e-01 -7.00236559e-01 1.05068338e+00 5.78175843e-01
2.58704782e-01 -1.22545052e+00 -1.10032940e+00 4.71834391e-01
3.45658749e-01 -1.16502798e+00 -4.17782992e-01 9.10582721e-01
-1.15042782e+00 -4.04229075e-01 -7.34185159e-01 -9.90075052e-01
7.25415826e-01 -3.60838652e-01 9.01111364e-01 -1.03679761e-01
-1.63457811e-01 -6.39995560e-03 3.07990313e-01 -2.37658516e-01
-6.10677660e-01 5.10392785e-01 6.22539282e-01 1.08093649e-01
-2.95560390e-01 -9.69673395e-01 -4.87417251e-01 1.51059106e-01
-1.13535380e+00 3.09786890e-02 4.13292855e-01 8.33426416e-01
5.58172703e-01 -1.46985307e-01 4.89417881e-01 -6.82859778e-01
7.05749393e-01 -3.74437958e-01 -1.09659576e+00 5.66173950e-03
-6.20511830e-01 9.78008926e-01 1.21417809e+00 -7.96053052e-01
-6.42945826e-01 2.85018504e-01 2.83795595e-01 -2.69470811e-01
4.50104326e-01 3.01080823e-01 5.27398229e-01 -1.70067221e-01
6.31707013e-01 3.68637800e-01 9.15286839e-02 -1.66305393e-01
8.72337371e-02 1.91982850e-01 5.78633547e-01 -2.05565378e-01
5.94803751e-01 4.69190538e-01 5.13167322e-01 -9.16159272e-01
-3.64032865e-01 -1.08857907e-01 -5.63516736e-01 -2.72711627e-02
4.92901921e-01 -8.18830192e-01 -1.29119313e+00 6.60421729e-01
-1.34007585e+00 -6.32287562e-01 -3.86266798e-01 2.02749103e-01
-5.42308211e-01 -3.85144025e-01 -1.28308105e+00 -1.04158795e+00
-5.42429030e-01 -8.80790114e-01 5.42765796e-01 5.85377455e-01
-2.40223445e-02 -9.66029882e-01 5.28117836e-01 -8.94168675e-01
7.99351335e-01 3.68899107e-01 6.75570130e-01 -3.58749300e-01
-6.13614559e-01 -1.94472060e-01 3.70862573e-01 4.02718544e-01
-2.32930884e-01 5.25334477e-01 -1.01747429e+00 -1.88198358e-01
3.64052087e-01 -9.54657644e-02 1.13824618e+00 6.04843616e-01
4.27123547e-01 -3.12487155e-01 -4.02146548e-01 6.41596019e-01
1.50008523e+00 9.15083215e-02 5.40850222e-01 -3.29010993e-01
4.95149732e-01 2.07755059e-01 -5.84206223e-01 3.76652658e-01
-8.05303901e-02 2.23035663e-01 2.71594822e-01 4.81982231e-02
5.06050229e-01 4.08199243e-02 9.33156550e-01 1.16276741e+00
-8.82847831e-02 -1.66924417e-01 -8.27846885e-01 5.65709889e-01
-1.92734671e+00 -9.31684792e-01 -1.94552496e-01 1.89655232e+00
8.67812037e-01 3.83275688e-01 -1.37019623e-02 7.42173120e-02
5.46023488e-01 3.48755382e-02 -9.15464580e-01 -8.08822453e-01
-3.50467950e-01 4.92363751e-01 6.98411047e-01 3.15922797e-01
-6.39339149e-01 7.01283276e-01 7.50275087e+00 -1.59921512e-01
-1.37230587e+00 -1.82996720e-01 4.99549538e-01 -3.15177500e-01
9.07203406e-02 -2.41641790e-01 -1.02490699e+00 2.96106011e-01
1.81370306e+00 -8.37991685e-02 8.17112386e-01 4.35744554e-01
3.76599997e-01 3.48082930e-01 -1.37089622e+00 5.77373981e-01
-4.04233187e-01 -1.46550703e+00 -2.48206824e-01 3.88466637e-03
9.93924797e-01 3.98988485e-01 2.55634487e-01 2.19514206e-01
2.25268260e-01 -1.19778228e+00 8.90466571e-01 9.55439925e-01
2.71359891e-01 -6.24795198e-01 4.66525108e-01 3.76033187e-01
-1.16528058e+00 -4.11899209e-01 -5.96592247e-01 -5.54440856e-01
-1.04604036e-01 7.16769516e-01 -3.73429269e-01 -3.16267967e-01
6.24332845e-01 8.34172487e-01 -3.06744903e-01 5.92969716e-01
1.11343853e-01 6.20230734e-01 -6.61802530e-01 -6.34004295e-01
3.35380793e-01 -9.65852365e-02 2.45901138e-01 1.25319552e+00
2.62341857e-01 5.85913621e-02 -5.38414538e-01 1.50270891e+00
-3.78709942e-01 -5.17163813e-01 -7.33289123e-01 -4.15860444e-01
2.76635796e-01 1.27115119e+00 -8.68266582e-01 -1.28653660e-01
3.02761961e-02 8.62109900e-01 5.35260320e-01 6.87440217e-01
-8.21696341e-01 -5.81407249e-01 8.97510648e-01 -1.27879694e-01
4.54000503e-01 -6.01502597e-01 -2.15393305e-01 -1.17626166e+00
1.22945920e-01 -5.49547911e-01 -2.23262787e-01 -4.82217520e-01
-7.21061885e-01 9.67174411e-01 -4.65225130e-01 -9.00881112e-01
-6.86000824e-01 -9.19129074e-01 -4.93378639e-01 8.09106886e-01
-1.18669248e+00 -1.84680238e-01 1.20609455e-01 4.96336728e-01
-3.19742896e-02 1.64913237e-01 9.37568128e-01 4.26324233e-02
-1.02373648e+00 3.26367170e-01 5.65849543e-01 2.77603179e-01
1.16942301e-01 -1.18424070e+00 8.37280095e-01 9.80006576e-01
4.74049561e-02 1.14137542e+00 8.96867394e-01 -3.69513482e-02
-1.74049699e+00 -9.72209334e-01 7.30187416e-01 -2.59923637e-01
7.90969193e-01 -1.00463641e+00 -9.42742288e-01 8.47037911e-01
1.04518615e-01 3.55835348e-01 5.39409295e-02 -4.04057950e-01
-2.18179911e-01 -3.43744904e-01 -6.19230568e-01 6.22713387e-01
9.18362498e-01 -6.59107208e-01 -1.13227919e-01 2.11922273e-01
8.12609136e-01 -4.75534976e-01 -7.44589865e-01 3.42598036e-02
7.97120094e-01 -8.41187358e-01 5.99590778e-01 -6.33474648e-01
5.39602160e-01 -1.95734590e-01 2.23914325e-01 -1.31666338e+00
-4.41750199e-01 -1.17241895e+00 -5.46219647e-01 7.44298339e-01
6.21580184e-01 -8.70185018e-01 4.90543187e-01 7.54699051e-01
-4.65871431e-02 -8.33104253e-01 -9.25864339e-01 -7.24834204e-01
1.20854549e-01 6.60928190e-02 1.10548511e-01 2.99753070e-01
1.81790918e-01 3.58803600e-01 -1.30231172e-01 4.02315050e-01
2.60210335e-01 -2.77793139e-01 1.24509141e-01 -1.26236939e+00
-1.99934721e-01 -7.03410804e-01 -5.63426077e-01 -1.17184854e+00
3.62035394e-01 -7.13614523e-01 4.61879283e-01 -1.00854206e+00
-2.84021884e-01 -7.09164068e-02 -7.75143445e-01 2.56435513e-01
3.30388695e-01 2.64939517e-01 -9.49381385e-03 1.85085908e-01
-2.35591844e-01 4.25095648e-01 7.48771608e-01 6.10859890e-04
-5.19586980e-01 6.20250963e-03 -3.27591568e-01 4.59455967e-01
6.25039518e-01 -7.40131319e-01 -2.34715775e-01 -4.39690560e-01
7.08822012e-01 4.47910875e-02 2.47768134e-01 -1.36620557e+00
7.68410444e-01 1.90967202e-01 2.90194064e-01 -1.49453595e-01
2.37184614e-01 -6.57835782e-01 1.24445379e-01 1.14792490e+00
-8.20147455e-01 3.97158772e-01 2.18747169e-01 3.92820120e-01
-1.46074668e-01 6.60696924e-02 7.89870918e-01 7.34930485e-02
-2.30212626e-03 7.37232715e-02 -9.63805854e-01 9.98719484e-02
5.46055079e-01 9.08292383e-02 -1.79991856e-01 -3.46313536e-01
-5.76846361e-01 2.04458341e-01 2.82907486e-03 8.14679414e-02
3.61730039e-01 -1.10613644e+00 -4.09022689e-01 4.90612686e-01
-4.29034263e-01 3.65656130e-02 -1.10206775e-01 7.94227839e-01
-6.74109221e-01 8.80366802e-01 -3.72099817e-01 -5.58334470e-01
-4.81435537e-01 2.72136122e-01 1.18350661e+00 -4.04660314e-01
-3.43497127e-01 7.00400889e-01 1.86893772e-02 -3.10249448e-01
2.68849760e-01 -1.41156101e+00 1.09224729e-01 -2.03773424e-01
4.61187482e-01 2.01612815e-01 1.26110055e-02 -1.69657916e-01
-4.89455551e-01 3.07327956e-01 1.32039443e-01 -1.93489999e-01
1.45425987e+00 2.01173946e-01 -5.81271291e-01 1.18485296e+00
1.16159785e+00 -5.45522332e-01 -1.17752481e+00 -1.24778554e-01
-3.68019333e-04 9.07251239e-01 1.60054918e-02 -2.44564489e-01
-1.03619826e+00 1.04535127e+00 5.86773098e-01 5.69261730e-01
1.08709013e+00 -3.86563390e-01 8.02648365e-01 1.25301802e+00
1.46348298e-01 -9.01067078e-01 -1.34906530e-01 1.11866677e+00
6.13211691e-01 -5.25958776e-01 -5.12105227e-01 2.81649351e-01
2.91663017e-02 1.46094120e+00 2.20265806e-01 -8.66447985e-01
8.09294522e-01 8.01890314e-01 -1.49208844e-01 1.41285837e-01
-1.40111208e+00 2.81248808e-01 9.66824740e-02 1.04092628e-01
7.11481273e-01 -2.74685860e-01 -4.96143550e-02 4.66076553e-01
2.79681403e-02 9.46035311e-02 8.24611425e-01 7.79591858e-01
-2.95462936e-01 -4.43510473e-01 5.65209426e-02 1.25336602e-01
-4.35646862e-01 -1.76459134e-01 -2.59508967e-01 5.36546409e-01
-4.43812579e-01 6.80060148e-01 4.65575993e-01 -6.62730262e-02
3.28613281e-01 1.81929916e-01 3.39524359e-01 -4.40729767e-01
-9.75093424e-01 -1.95601508e-01 -5.55001855e-01 -6.62798345e-01
-1.51497528e-01 -2.68508732e-01 -1.71861887e+00 -9.80358720e-02
-2.55289435e-01 2.22502146e-02 5.47753096e-01 9.50630069e-01
6.07139349e-01 9.39822495e-01 6.34152293e-01 -9.88481224e-01
-8.95258129e-01 -9.12519813e-01 -7.20564067e-01 -2.92083267e-02
9.90441859e-01 -8.43254328e-02 -5.03820658e-01 1.14456370e-01] | [7.755131721496582, 3.2860071659088135] |
5041e598-87e0-4384-8bfe-09076d77e3ce | covidx-computer-aided-diagnosis-of-covid-19 | 2012.13605 | null | https://arxiv.org/abs/2012.13605v1 | https://arxiv.org/pdf/2012.13605v1.pdf | COVIDX: Computer-aided diagnosis of Covid-19 and its severity prediction with raw digital chest X-ray images | Coronavirus disease (COVID-19) is a contagious infection caused by severe acute respiratory syndrome coronavirus-2 (SARS-COV-2) and it has infected and killed millions of people across the globe. In the absence of specific drugs or vaccines for the treatment of COVID-19 and the limitation of prevailing diagnostic techniques, there is a requirement for some alternate automatic screening systems that can be used by the physicians to quickly identify and isolate the infected patients. A chest X-ray (CXR) image can be used as an alternative modality to detect and diagnose the COVID-19. In this study, we present an automatic COVID-19 diagnostic and severity prediction (COVIDX) system that uses deep feature maps from CXR images to diagnose COVID-19 and its severity prediction. The proposed system uses a three-phase classification approach (healthy vs unhealthy, COVID-19 vs Pneumonia, and COVID-19 severity) using different shallow supervised classification algorithms. We evaluated COVIDX not only through 10-fold cross2 validation and by using an external validation dataset but also in real settings by involving an experienced radiologist. In all the evaluation settings, COVIDX outperforms all the existing stateof-the-art methods designed for this purpose. We made COVIDX easily accessible through a cloud-based webserver and python code available at https://sites.google.com/view/wajidarshad/software and https://github.com/wajidarshad/covidx, respectively. | ['Saiqa Andleeb', 'Syed Ali Abbas', 'Wajid Arshad Abbasi'] | 2020-12-25 | null | null | null | null | ['severity-prediction'] | ['computer-vision'] | [-1.11979902e-01 -7.34600246e-01 -2.32633632e-02 -8.46287459e-02
-3.34382236e-01 -8.12312126e-01 1.57732382e-01 4.42628741e-01
-2.64227092e-01 6.90246284e-01 -2.01640390e-02 -5.19936442e-01
-1.38509139e-01 -6.64415300e-01 -1.97627187e-01 -7.46178329e-01
-1.36736140e-01 9.31164742e-01 1.43340811e-01 2.80252159e-01
8.13598093e-03 7.56001294e-01 -1.21543181e+00 2.97993213e-01
9.36386347e-01 7.55338132e-01 7.39620984e-01 1.07343042e+00
2.89319038e-01 4.41451073e-01 -3.10938954e-01 2.11472049e-01
1.77340701e-01 -5.15691817e-01 -5.82183361e-01 -4.89885718e-01
-1.00136951e-01 -4.90050614e-01 2.69309610e-01 4.42827344e-01
6.22778296e-01 -2.71095008e-01 9.35447156e-01 -1.15794075e+00
-3.03067207e-01 -4.16916609e-01 -4.71939504e-01 7.75508344e-01
2.90714055e-01 3.85400802e-01 4.33803231e-01 -7.56382108e-01
6.67202711e-01 8.33439767e-01 7.70142615e-01 6.92080438e-01
-7.34188139e-01 -6.55864894e-01 -4.87868458e-01 2.91491687e-01
-1.23262918e+00 3.99109155e-01 2.03993842e-01 -1.05137062e+00
9.37696636e-01 4.58540976e-01 8.08316290e-01 1.01314890e+00
5.29528975e-01 3.78595173e-01 1.35487318e+00 1.50634885e-01
1.64006844e-01 5.90477586e-02 3.27939630e-01 7.32371986e-01
6.08598769e-01 8.90989229e-02 2.70201415e-01 -7.42721319e-01
6.18753314e-01 7.69381762e-01 -4.89181757e-01 -1.23910874e-01
-1.16994953e+00 9.86069322e-01 2.36159265e-01 1.50723696e-01
-7.27653861e-01 -3.62774223e-01 6.55838430e-01 1.55457072e-02
4.02797684e-02 3.15150470e-01 -8.11301649e-01 9.71742570e-02
-6.83645487e-01 1.13997102e-01 4.68879193e-01 2.34045669e-01
2.65718997e-01 -2.66907334e-01 -2.63571322e-01 6.60932660e-01
3.81286681e-01 1.09710753e+00 5.34412444e-01 -4.28420424e-01
5.45952842e-02 5.89626908e-01 2.90267378e-01 -6.71687782e-01
-6.37550592e-01 -3.61711413e-01 -9.47997272e-01 3.21397521e-02
-9.20170732e-03 -4.51587290e-01 -9.38996792e-01 1.35646641e+00
5.16564488e-01 3.96304846e-01 9.71846469e-03 1.08543062e+00
8.55408728e-01 7.28287935e-01 1.78545624e-01 -4.12227184e-01
1.78458345e+00 -8.10486495e-01 -3.93066168e-01 2.61586994e-01
8.23525786e-01 -7.44994342e-01 9.30152476e-01 2.83799231e-01
-4.31973398e-01 -1.38637796e-01 -8.88103426e-01 5.27726591e-01
-4.47134078e-01 2.34252349e-01 2.86118120e-01 5.46402514e-01
-8.10305953e-01 3.02505672e-01 -1.02676880e+00 -9.40006733e-01
4.30675060e-01 1.24352813e-01 -2.85342306e-01 -1.65842488e-01
-1.00922012e+00 1.03856242e+00 2.78458279e-02 -1.31846711e-01
-1.16691709e+00 -7.83368826e-01 -4.24536139e-01 -1.67371079e-01
-2.09779255e-02 -1.23237836e+00 8.63405645e-01 -2.62538284e-01
-7.73939133e-01 1.01578629e+00 -2.45244324e-01 -1.07528031e-01
4.41043705e-01 -4.71061200e-01 -4.54008609e-01 5.27487040e-01
5.45249414e-03 2.33544931e-02 5.10720968e-01 -1.01681173e+00
-4.14527953e-01 -6.14500761e-01 -4.40056443e-01 -4.46866415e-02
2.01304927e-01 4.29685891e-01 1.03225894e-01 -6.36096954e-01
-4.73315567e-01 -1.17188668e+00 -1.38760805e-01 -7.75594190e-02
-2.13756546e-01 -3.52092415e-01 1.21526706e+00 -7.19492733e-01
9.55515802e-01 -1.92965353e+00 -3.62393886e-01 5.45823947e-02
3.80207807e-01 1.14705575e+00 -5.70833981e-02 6.57827079e-01
2.01343969e-02 3.06023598e-01 -3.01282316e-01 1.11140966e-01
-5.17465591e-01 6.90806434e-02 5.40972613e-02 7.82899261e-01
2.86676973e-01 6.82379544e-01 -8.61739993e-01 -5.87290645e-01
3.14315945e-01 9.31639850e-01 -2.75181979e-01 6.78458154e-01
-7.47077614e-02 7.70795643e-01 -6.72158480e-01 7.47545004e-01
8.51688862e-01 -8.88560116e-01 1.54754341e-01 4.17592451e-02
2.76563819e-02 -1.82879925e-01 -7.54989743e-01 9.43058729e-01
-1.79484218e-01 2.31620371e-01 1.26116741e-02 -7.71754622e-01
6.35606229e-01 8.18471491e-01 4.87760097e-01 -1.85058206e-01
4.46962923e-01 2.44547606e-01 -1.50698051e-01 -1.00160515e+00
-4.33017343e-01 -2.00469732e-01 3.57447237e-01 8.08704853e-01
-4.31972384e-01 2.91869253e-01 4.61866222e-02 5.23511171e-02
1.27269387e+00 -4.02006924e-01 5.41822255e-01 -3.05636168e-01
6.39815986e-01 3.35704058e-01 6.45273149e-01 5.97718239e-01
-5.23512602e-01 7.17120171e-01 -5.92856556e-02 -4.96605933e-01
-8.00522208e-01 -1.41681063e+00 -4.87985313e-01 4.97577876e-01
-7.56638870e-02 3.28451209e-02 -5.64100087e-01 -6.49511576e-01
1.32784564e-02 3.95716548e-01 -5.77885389e-01 3.41850996e-01
-5.48499763e-01 -9.35365081e-01 4.58038092e-01 4.76683527e-01
2.94550687e-01 -1.13308144e+00 -1.18859792e+00 -5.21669835e-02
-2.64313728e-01 -6.67801857e-01 -2.68081784e-01 -1.16494000e-01
-8.46086264e-01 -1.49421656e+00 -9.77168322e-01 -6.75531685e-01
5.32348931e-01 3.87763023e-01 8.04415166e-01 6.55766308e-01
-8.50368857e-01 3.42602909e-01 -4.64530498e-01 -5.15110612e-01
-3.73964548e-01 -2.21861288e-01 2.29159310e-01 -3.27608109e-01
5.21713853e-01 -3.51769984e-01 -1.32340121e+00 2.66026646e-01
-6.99862540e-01 -3.58948624e-03 5.93335927e-01 4.61414903e-01
7.05278695e-01 -4.21017140e-01 7.55347431e-01 -9.34692860e-01
5.69719076e-01 -9.48212683e-01 -4.42023605e-01 1.74232841e-01
-8.67166638e-01 -5.04322469e-01 6.85149610e-01 -1.24667026e-01
-6.44499242e-01 -2.20593572e-01 -1.77455768e-01 -5.72211683e-01
-4.80461895e-01 2.21080557e-01 4.17822301e-01 4.68168378e-01
4.74925548e-01 7.17255026e-02 2.09444582e-01 -6.72986567e-01
-2.70953596e-01 1.06766653e+00 1.75112337e-01 1.06903262e-01
6.75152600e-01 5.69620192e-01 -1.49121270e-01 -7.85372019e-01
-6.63614571e-01 -9.32540953e-01 -3.31668973e-01 -1.47648260e-01
1.49993300e+00 -9.94199753e-01 -7.12173522e-01 5.58215082e-01
-1.09491527e+00 -1.16547473e-01 3.49395752e-01 8.54876518e-01
-1.72230303e-01 3.02087098e-01 -8.33112061e-01 -6.26392066e-01
-1.10887980e+00 -9.99468923e-01 7.76088357e-01 2.55817980e-01
-3.97003174e-01 -9.52535570e-01 7.30442643e-01 6.32731080e-01
6.33772731e-01 6.01385593e-01 1.05437768e+00 -1.02908742e+00
-3.96259248e-01 -2.27616087e-01 -4.13009226e-01 3.98894131e-01
3.41160297e-01 1.51121363e-01 -6.68464422e-01 -6.06653869e-01
1.21690743e-01 -1.75226703e-01 4.58313823e-01 4.70780641e-01
7.69610286e-01 -2.70134270e-01 -6.23475075e-01 5.61395168e-01
1.78039086e+00 5.50343394e-01 3.30571353e-01 1.02656893e-01
5.44256687e-01 1.70478567e-01 5.48836946e-01 6.06168151e-01
2.21239626e-01 1.98144510e-01 3.36239159e-01 -2.91685641e-01
1.13488898e-01 2.53580064e-01 -2.38137633e-01 8.37557375e-01
-3.16435635e-01 -4.85310823e-01 -1.24425030e+00 5.99867523e-01
-1.41584694e+00 -9.69627023e-01 -5.93694210e-01 1.93824065e+00
5.61507583e-01 -4.86584365e-01 1.68389142e-01 -1.52923882e-01
7.99940765e-01 -2.15891838e-01 -5.39007306e-01 -6.76979482e-01
1.60970807e-01 2.38684550e-01 -2.50828061e-02 1.30655542e-01
-9.71746385e-01 1.00209013e-01 5.48924255e+00 9.42429304e-02
-1.53110373e+00 3.92224073e-01 3.91957819e-01 9.73329246e-02
9.90703776e-02 -3.16620022e-01 -4.65720594e-01 6.16990685e-01
6.72098100e-01 1.52240604e-01 1.71852365e-01 8.38246346e-01
5.29142618e-01 2.47982442e-02 -6.52395964e-01 8.17299306e-01
5.57945594e-02 -1.32551885e+00 -1.91637084e-01 -1.09659918e-01
5.69882333e-01 6.62907362e-01 -3.57165903e-01 -6.12896010e-02
-1.21959604e-01 -8.62534761e-01 -2.07957432e-01 5.00956357e-01
1.01966810e+00 -2.65995383e-01 1.13134408e+00 2.83714861e-01
-1.16308033e+00 1.79041713e-01 -1.32315913e-02 2.81648695e-01
1.99276805e-01 5.27628660e-01 -1.17483902e+00 1.64525539e-01
9.35690999e-01 3.04277122e-01 -3.81255537e-01 1.18345594e+00
-2.13907734e-02 7.18274593e-01 -1.93204924e-01 -1.50909424e-01
-1.03720963e-01 -1.02378190e-01 6.18085384e-01 1.44078279e+00
3.42796624e-01 5.31526268e-01 4.27644141e-02 6.40110612e-01
3.91372442e-01 3.68530512e-01 -6.47969902e-01 -3.32099162e-02
4.20172632e-01 1.30395174e+00 -7.86828220e-01 -6.48305953e-01
-3.57406914e-01 8.32688749e-01 -1.30463794e-01 2.65651613e-01
-9.65405285e-01 -3.30335498e-01 7.74458408e-01 5.06646454e-01
6.18026018e-01 2.24012375e-01 -2.12098639e-02 -1.07288468e+00
-3.96393567e-01 -8.25833619e-01 7.50891745e-01 -9.41560626e-01
-1.40112066e+00 8.86783183e-01 2.98243295e-02 -1.19776917e+00
-2.05771908e-01 -5.14433801e-01 -9.42197144e-01 8.81144285e-01
-1.54409266e+00 -7.71811306e-01 -5.78601897e-01 5.78616679e-01
2.63358027e-01 2.59385146e-02 1.29079390e+00 1.24752432e-01
-5.37672520e-01 1.41977519e-01 3.41179401e-01 -1.48811899e-02
5.13313532e-01 -9.69572008e-01 -2.49633983e-01 6.23742402e-01
-8.42817545e-01 7.45676756e-01 5.80545843e-01 -9.15068150e-01
-1.05813479e+00 -1.36468458e+00 9.04688060e-01 -4.44452941e-01
2.40583688e-01 -1.18784755e-01 -9.11155939e-01 5.24832785e-01
3.59413564e-01 2.32642755e-01 1.09165657e+00 -5.97395360e-01
-2.42040470e-01 1.19877458e-01 -1.60463250e+00 1.02867998e-01
5.78330576e-01 -3.10926527e-01 -6.28383100e-01 5.65686345e-01
4.95076537e-01 4.88155670e-02 -1.04665041e+00 7.51934350e-01
5.88169336e-01 -9.72693443e-01 8.32861722e-01 -6.45901084e-01
2.08233744e-01 -5.07099628e-01 -2.88694408e-02 -8.93511415e-01
-2.75279880e-01 -1.90797701e-01 1.23067699e-01 5.67563117e-01
1.19493902e-01 -9.45369601e-01 4.78430271e-01 1.23861404e-02
1.67520002e-01 -1.12310028e+00 -4.05943692e-01 -5.77346265e-01
-1.36363938e-01 7.70422146e-02 4.82486725e-01 1.07448089e+00
-3.13950747e-01 2.57111609e-01 4.40790690e-02 4.06178951e-01
4.85544443e-01 5.80129206e-01 4.10255224e-01 -1.42196679e+00
-2.17721179e-01 1.68073829e-02 -2.37987250e-01 -1.41815037e-01
-5.04121184e-01 -8.48609805e-01 -3.17878425e-01 -1.95218301e+00
5.97130954e-01 -5.66703498e-01 -4.70299035e-01 3.77283305e-01
-1.99740410e-01 3.73537987e-01 1.94210023e-01 5.64959824e-01
-2.80027419e-01 5.70155084e-02 1.20073378e+00 2.69821346e-01
-1.17157288e-01 3.66047136e-02 -2.46160135e-01 6.48619533e-01
1.37441683e+00 -7.67858267e-01 -4.67445165e-01 -1.99357435e-01
1.45026166e-02 2.27906346e-01 4.73789096e-01 -9.09727871e-01
-3.31060588e-01 -3.24686050e-01 3.32973242e-01 -9.88123953e-01
8.63490477e-02 -7.17971444e-01 5.44056177e-01 1.32418001e+00
2.59105802e-01 4.99431431e-01 1.05674267e-02 4.22436237e-01
1.64754078e-01 -5.19773513e-02 9.85876739e-01 -1.38123527e-01
-2.41530895e-01 4.24009830e-01 -9.43043709e-01 2.95396775e-01
1.50573504e+00 7.84539804e-03 -8.08571398e-01 1.33055568e-01
-4.23129827e-01 1.95089534e-01 6.56981647e-01 1.78109288e-01
7.47799456e-01 -9.08449352e-01 -9.55807149e-01 2.75583953e-01
3.07440758e-01 -2.34062731e-01 5.26082218e-01 1.34805775e+00
-1.23182809e+00 6.12077713e-01 -2.34870419e-01 -7.16687202e-01
-1.67125678e+00 9.14729953e-01 3.81858647e-01 -3.28860998e-01
-5.91412842e-01 5.00548840e-01 3.37861031e-01 -2.78343499e-01
-6.17954833e-03 -1.15027830e-01 -3.86955440e-01 -1.98799044e-01
7.63877213e-01 3.33022177e-01 4.68677059e-02 -6.76335990e-01
-8.62800956e-01 5.52067339e-01 -4.29713614e-02 5.93851984e-01
1.35664141e+00 2.16370262e-02 -1.64753258e-01 2.13293627e-01
1.35078561e+00 -7.08747283e-02 -4.93641227e-01 1.51474684e-01
-4.37248439e-01 -2.77458936e-01 -2.88370192e-01 -1.01504183e+00
-9.53892887e-01 8.88204992e-01 1.25396502e+00 3.77469882e-02
1.13911498e+00 2.64563173e-01 1.12315619e+00 -7.82034826e-03
7.20836446e-02 -4.52920735e-01 -1.41449198e-01 6.54025301e-02
7.20414758e-01 -1.29335415e+00 -6.35053366e-02 -3.02048177e-01
-7.66545057e-01 6.21043563e-01 3.09992433e-01 -2.53828496e-01
1.02559566e+00 2.44922817e-01 7.15546846e-01 -5.95089793e-01
-1.02174127e+00 1.45265773e-01 -1.52077273e-01 7.68482208e-01
4.05644119e-01 5.16463518e-01 -6.93289042e-01 5.35606444e-01
2.41415620e-01 2.52861619e-01 2.80244201e-01 1.05433786e+00
-2.69342929e-01 -8.09019864e-01 -4.90510166e-01 9.70433891e-01
-7.93149769e-01 -7.12724403e-02 -1.62713721e-01 6.48082137e-01
3.57579529e-01 9.18499649e-01 -1.38350472e-01 -2.48710677e-01
-1.59544516e-02 2.59913579e-02 8.01043436e-02 -5.82628191e-01
-6.04794085e-01 -1.78722739e-01 -1.68283015e-01 -3.93911332e-01
-4.49669838e-01 -7.08503067e-01 -1.54466295e+00 -1.49132416e-01
-1.72608331e-01 2.29786053e-01 6.84933424e-01 5.76059759e-01
6.37371719e-01 1.98253214e-01 6.69754446e-01 -1.41207188e-01
-3.58813286e-01 -8.67110133e-01 -5.29131711e-01 4.05231476e-01
4.43663985e-01 -5.38366437e-01 -6.01655841e-01 2.67297844e-03] | [15.57456111907959, -1.6903070211410522] |
d4ef6117-4558-4638-8b90-1dea25d1dc83 | regen-zero-shot-text-classification-via | 2305.10703 | null | https://arxiv.org/abs/2305.10703v1 | https://arxiv.org/pdf/2305.10703v1.pdf | ReGen: Zero-Shot Text Classification via Training Data Generation with Progressive Dense Retrieval | With the development of large language models (LLMs), zero-shot learning has attracted much attention for various NLP tasks. Different from prior works that generate training data with billion-scale natural language generation (NLG) models, we propose a retrieval-enhanced framework to create training data from a general-domain unlabeled corpus. To realize this, we first conduct contrastive pretraining to learn an unsupervised dense retriever for extracting the most relevant documents using class-descriptive verbalizers. We then further propose two simple strategies, namely Verbalizer Augmentation with Demonstrations and Self-consistency Guided Filtering to improve the topic coverage of the dataset while removing noisy examples. Experiments on nine datasets demonstrate that REGEN achieves 4.3% gain over the strongest baselines and saves around 70% of the time compared to baselines using large NLG models. Besides, REGEN can be naturally integrated with recently proposed large language models to boost performance. | ['Chao Zhang', 'Jiaming Shen', 'Yu Meng', 'Rongzhi Zhang', 'Yuchen Zhuang', 'Yue Yu'] | 2023-05-18 | null | null | null | null | ['topic-coverage'] | ['natural-language-processing'] | [ 2.20766038e-01 6.68361545e-01 -4.61294115e-01 -1.75214142e-01
-1.36248529e+00 -2.62687415e-01 1.12942004e+00 4.21844795e-02
-5.63883722e-01 1.00566840e+00 8.80116582e-01 -9.14435759e-02
2.92467266e-01 -8.99949372e-01 -6.19981647e-01 -3.42100412e-01
3.75658423e-01 9.96745050e-01 7.39328489e-02 -3.72701466e-01
4.85139638e-02 -1.84447691e-01 -1.59448743e+00 2.47686192e-01
1.22793150e+00 4.90520090e-01 3.19577754e-01 3.21643710e-01
-5.76122582e-01 6.14711583e-01 -7.37471819e-01 -3.46336216e-01
2.51569748e-02 -5.19648731e-01 -8.61688614e-01 1.63909510e-01
1.98491395e-01 -4.95975435e-01 -4.27086800e-01 7.24995673e-01
8.23731124e-01 6.70805991e-01 8.70234609e-01 -8.56190085e-01
-1.22321928e+00 1.11283576e+00 -3.96857798e-01 -6.40220183e-04
3.02457511e-01 1.66602209e-01 1.21355307e+00 -1.47570050e+00
1.10472488e+00 1.44605863e+00 2.27928311e-02 9.47368383e-01
-1.29281497e+00 -6.28832638e-01 1.74002334e-01 6.53502792e-02
-1.33974695e+00 -6.47229373e-01 6.04110539e-01 -2.68048290e-02
1.14396954e+00 -1.09248459e-01 3.95025074e-01 1.53761637e+00
-4.95548666e-01 1.31656516e+00 6.62297130e-01 -8.09793830e-01
3.11510295e-01 3.39085585e-03 3.94444913e-01 5.33007860e-01
5.33631265e-01 -2.33341277e-01 -6.12721741e-01 -2.07354411e-01
4.41857725e-01 -5.06150573e-02 -2.33262226e-01 -2.46867567e-01
-1.18973041e+00 1.19702506e+00 2.84727573e-01 1.87334791e-01
-3.16232741e-01 1.97642162e-01 2.54350603e-01 -7.28259385e-02
7.23996162e-01 9.81439590e-01 -1.11087836e-01 4.54622358e-02
-1.16594326e+00 4.58957374e-01 7.02255726e-01 1.46572924e+00
8.64683449e-01 2.84441829e-01 -9.60763276e-01 1.01504612e+00
1.64014578e-01 4.16785508e-01 1.06800389e+00 -7.34196782e-01
6.07154131e-01 5.03150523e-01 1.22005455e-01 -3.29463661e-01
4.46229093e-02 -3.49779993e-01 -6.87757909e-01 -5.75435519e-01
-1.07392840e-01 -2.38458231e-01 -1.35529196e+00 1.73770940e+00
1.10489078e-01 3.10046494e-01 4.28809583e-01 6.59007907e-01
1.17295182e+00 1.02306497e+00 3.37321758e-01 -2.91144252e-01
1.26842165e+00 -1.19733500e+00 -1.00413918e+00 -5.10860801e-01
6.29536569e-01 -4.11970645e-01 1.56332743e+00 1.36848690e-03
-1.07111740e+00 -4.88509893e-01 -1.07463717e+00 -4.38283026e-01
-5.12444019e-01 4.99828346e-02 7.89499223e-01 2.44943321e-01
-8.63965511e-01 2.23981947e-01 -6.22627079e-01 -2.68103629e-01
4.30609882e-01 -1.13719307e-01 -1.75589025e-02 -3.17130625e-01
-1.74503720e+00 6.21433377e-01 7.31256425e-01 -4.56611723e-01
-1.27403021e+00 -7.76979148e-01 -1.16928554e+00 2.49628931e-01
6.41635478e-01 -8.19576144e-01 1.44133508e+00 -2.79011279e-01
-1.33358753e+00 5.84250331e-01 -4.58484173e-01 -7.21643448e-01
2.33542055e-01 -5.97322166e-01 -2.10623533e-01 1.79521531e-01
4.30868030e-01 1.08929181e+00 8.20989370e-01 -1.05199993e+00
-2.93180972e-01 -2.13325247e-02 -1.66875824e-01 3.81403148e-01
-6.52459919e-01 -1.93485156e-01 -9.03053761e-01 -9.50591445e-01
-1.48137704e-01 -5.88138521e-01 -4.45581049e-01 -5.30676842e-01
-4.49522853e-01 -7.90105641e-01 5.99533081e-01 -4.64157969e-01
1.29447091e+00 -1.91913140e+00 2.09865496e-02 -2.22798929e-01
2.39217356e-01 4.01472449e-01 -6.69926643e-01 5.26088953e-01
2.89914966e-01 2.36824244e-01 -1.71136424e-01 -6.07220829e-01
1.77409500e-01 1.05613567e-01 -1.04148209e+00 -2.06683904e-01
3.75412136e-01 1.47702610e+00 -1.23376215e+00 -5.37880242e-01
4.07800972e-02 3.96641403e-01 -6.27433240e-01 3.85616541e-01
-6.17886782e-01 -1.64969936e-01 -4.82225060e-01 4.78279442e-01
2.03235224e-01 -5.53134918e-01 1.48186712e-02 2.51268029e-01
4.98465925e-01 5.48576534e-01 -9.30853605e-01 1.99293327e+00
-4.68549430e-01 4.66427535e-01 -4.18453366e-01 -7.92007267e-01
9.59844768e-01 5.52220762e-01 4.88314666e-02 -6.68669999e-01
-9.15795565e-02 1.30771007e-02 -4.73514736e-01 -1.82860449e-01
1.05375338e+00 -2.20004201e-01 -5.02195418e-01 7.65596926e-01
7.20559657e-01 -4.16082382e-01 5.83711863e-01 9.87105668e-01
1.04593337e+00 1.18384771e-01 3.66384983e-01 -6.96060061e-02
8.99510607e-02 -5.34482859e-02 3.53471160e-01 9.56589162e-01
5.35062142e-03 7.68581271e-01 6.87886178e-02 -5.03718220e-02
-8.42609346e-01 -1.18107128e+00 3.11559767e-01 1.30793250e+00
1.04216233e-01 -6.45472705e-01 -5.90107143e-01 -5.50332725e-01
-2.27753147e-01 1.32174623e+00 -3.73606145e-01 -5.12867928e-01
-4.16562647e-01 -6.97904527e-01 6.88277602e-01 5.75522721e-01
4.60915953e-01 -1.31041944e+00 -2.50275970e-01 3.55582416e-01
-4.57224309e-01 -1.08178771e+00 -4.72220451e-01 1.24993026e-01
-7.00818539e-01 -6.05961561e-01 -9.97951329e-01 -8.85143101e-01
6.27117038e-01 4.17677611e-01 1.36626780e+00 -2.01960802e-01
-3.14648956e-01 2.33644724e-01 -5.79266548e-01 -5.28562069e-01
-2.35559925e-01 4.50749725e-01 1.60784483e-01 -3.83600503e-01
6.77262902e-01 -3.99790317e-01 -4.18251812e-01 -3.51551324e-01
-1.05315936e+00 1.33820668e-01 7.06577897e-01 1.03429198e+00
6.13692701e-01 -4.16714787e-01 1.12093735e+00 -1.08707368e+00
1.22343504e+00 -4.91501421e-01 -3.16797793e-01 4.55482513e-01
-9.48494971e-01 5.20675004e-01 5.12992918e-01 -5.88405013e-01
-1.51974547e+00 -1.94336057e-01 2.70194799e-01 -4.12850410e-01
-8.00757855e-02 5.37798643e-01 -1.27729535e-01 8.20150137e-01
9.29633915e-01 5.34548402e-01 -3.95756841e-01 -4.98042226e-01
1.13997030e+00 6.56779647e-01 5.10269105e-01 -7.08540320e-01
8.20614576e-01 2.88866371e-01 -8.03748131e-01 -8.33511829e-01
-1.21622336e+00 -6.52175725e-01 -2.48117462e-01 1.86650410e-01
7.13334501e-01 -1.31586909e+00 2.58809596e-01 -1.26083702e-01
-1.31329870e+00 -2.83029169e-01 -7.33376622e-01 4.04487938e-01
-4.06623214e-01 2.09909737e-01 -7.75360763e-01 -8.08266401e-01
-9.23291564e-01 -7.60809660e-01 1.34504080e+00 2.19678745e-01
-4.31248546e-01 -7.35019922e-01 3.31958294e-01 3.86834383e-01
3.52693766e-01 -3.47667754e-01 8.44372690e-01 -9.79490578e-01
-6.38271332e-01 -2.18671843e-01 -2.14200556e-01 1.47006791e-02
-3.53036751e-03 -5.56091964e-01 -9.27142620e-01 -2.34587416e-01
-2.15254292e-01 -1.05625331e+00 1.26586592e+00 2.00659156e-01
9.39339399e-01 -4.65266466e-01 -5.61348200e-01 4.17551428e-01
1.12064493e+00 -7.10076839e-02 5.91109753e-01 -2.53954269e-02
5.49433053e-01 4.83129501e-01 6.85466409e-01 4.73870248e-01
2.35500395e-01 3.63082856e-01 -1.56888798e-01 6.75943047e-02
-4.22555119e-01 -9.20094371e-01 3.00781727e-01 9.13116395e-01
1.47795081e-01 -4.53172177e-01 -7.19154716e-01 8.72444212e-01
-1.75003445e+00 -1.00975955e+00 5.35574496e-01 1.90102410e+00
1.35759377e+00 4.98603024e-02 -3.57134163e-01 -3.37979287e-01
6.00762308e-01 5.51188707e-01 -5.81674695e-01 9.63982865e-02
-1.25328124e-01 4.85870719e-01 7.23690093e-02 5.66097558e-01
-9.86629486e-01 1.68494046e+00 6.62280512e+00 1.21284723e+00
-6.57056928e-01 1.73186719e-01 5.38016319e-01 -5.03378451e-01
-7.89812207e-01 7.95363188e-02 -1.28074503e+00 2.50369936e-01
1.01737118e+00 -7.49862075e-01 2.13033333e-01 1.10857058e+00
1.05696961e-01 2.82529116e-01 -9.37362432e-01 7.49492109e-01
4.85923409e-01 -1.54585505e+00 7.51512110e-01 -9.33855623e-02
1.16554713e+00 6.34872317e-02 -1.88347310e-01 1.16125882e+00
8.86557996e-01 -9.93626416e-01 4.24681544e-01 3.11156601e-01
9.15784597e-01 -6.79473519e-01 4.48607147e-01 6.03646398e-01
-1.00118673e+00 2.90348917e-01 -6.67300522e-01 -3.82169262e-02
5.48938036e-01 5.66409051e-01 -1.16284895e+00 3.06361109e-01
2.23238960e-01 5.38525939e-01 -5.50519228e-01 6.43752873e-01
-6.41545057e-01 7.91758657e-01 -6.34815320e-02 -3.73108566e-01
3.36185604e-01 5.46539426e-02 5.07332921e-01 1.20115435e+00
2.07855999e-01 2.38099083e-01 2.03131288e-01 1.18473577e+00
-6.45121634e-01 2.15182915e-01 -7.84041941e-01 -4.35229570e-01
8.04239035e-01 1.16306698e+00 -4.29458290e-01 -1.00650692e+00
-2.54671752e-01 1.02175343e+00 6.05175555e-01 6.74047947e-01
-5.97704649e-01 -7.42978096e-01 2.22041935e-01 4.39806767e-02
2.45996356e-01 -1.06608182e-01 3.50618884e-02 -1.52850282e+00
-1.02748118e-01 -7.20158517e-01 3.39013875e-01 -7.08059192e-01
-1.20281041e+00 6.03591740e-01 8.18462968e-02 -1.07847214e+00
-8.36787105e-01 -1.10567100e-01 -5.29217124e-01 7.19244182e-01
-1.44683921e+00 -1.08058310e+00 -2.17203796e-02 3.10903817e-01
1.29715884e+00 -3.80609572e-01 1.00858128e+00 -9.70147923e-02
-2.96828091e-01 5.66743731e-01 -8.18692967e-02 1.81500077e-01
8.69133353e-01 -1.22691476e+00 7.27802932e-01 1.06648993e+00
5.32774508e-01 1.10410404e+00 5.57511926e-01 -8.80712152e-01
-1.11696172e+00 -1.32626283e+00 1.17771971e+00 -4.47304308e-01
5.92934728e-01 -6.28074050e-01 -8.92422616e-01 6.85999632e-01
5.75586677e-01 -2.30137095e-01 7.43305087e-01 1.72643870e-01
-5.45681238e-01 2.48164207e-01 -6.26932383e-01 8.49083126e-01
1.08517206e+00 -5.14656723e-01 -1.31348467e+00 4.55469638e-01
1.45484996e+00 -9.80440527e-02 -2.22342014e-01 2.62767196e-01
2.25862153e-02 -1.39089346e-01 1.09159863e+00 -9.19573426e-01
6.58478200e-01 -4.00855131e-02 7.48827606e-02 -1.41768289e+00
-2.51187772e-01 -7.92850733e-01 -7.09201813e-01 1.40187049e+00
6.01046741e-01 -6.33324012e-02 7.74560392e-01 6.77740395e-01
-2.16060817e-01 -5.41401446e-01 -6.40035808e-01 -8.31577063e-01
1.41801521e-01 -3.79398435e-01 4.29231703e-01 7.68029690e-01
4.01170135e-01 1.31068778e+00 -5.40004492e-01 -5.41661203e-01
5.59661567e-01 1.68082267e-01 8.32402706e-01 -1.09499633e+00
-2.17369780e-01 -1.75252572e-01 3.14265192e-01 -1.43881583e+00
4.80728507e-01 -1.20164347e+00 4.76838052e-01 -1.98527718e+00
4.69429225e-01 -1.60526074e-02 -1.37329981e-01 4.97157037e-01
-7.07623899e-01 5.07089607e-02 5.91083765e-02 1.43405735e-01
-1.11898756e+00 1.23966420e+00 1.13638008e+00 -3.44745696e-01
-3.74006480e-01 -3.22571665e-01 -1.07304490e+00 3.79225701e-01
4.93703455e-01 -1.71499655e-01 -9.13214386e-01 -5.53978145e-01
2.46488340e-02 -1.02356955e-01 -8.37706178e-02 -7.96611011e-01
2.03440577e-01 -6.03037663e-02 2.03410283e-01 -6.82393789e-01
4.86611038e-01 -1.35469705e-01 -5.82510412e-01 2.15399370e-01
-8.29192221e-01 -3.61455172e-01 1.71760507e-02 8.53057384e-01
-2.97004551e-01 -2.70513505e-01 4.15998787e-01 -3.41803074e-01
-9.39158857e-01 3.89582753e-01 -4.49166536e-01 5.53946733e-01
7.63936758e-01 3.07882041e-01 -4.75500494e-01 -6.60566986e-01
-4.69016373e-01 3.70442033e-01 1.42264262e-01 6.95587873e-01
8.75730634e-01 -1.50150168e+00 -8.41950417e-01 6.74892887e-02
5.68248093e-01 2.48226672e-01 1.00655213e-01 6.89472351e-03
1.25688193e-02 8.74029040e-01 3.98631573e-01 -4.11951952e-02
-6.32073700e-01 7.40781248e-01 -3.46767485e-01 -6.53326571e-01
-6.70674562e-01 9.32547510e-01 2.53586054e-01 -4.07158822e-01
2.73025632e-01 -1.27334744e-01 -2.50371814e-01 2.17506081e-01
9.47256148e-01 1.73440114e-01 -1.31094262e-01 -1.70362756e-01
1.15463007e-02 -1.56870499e-01 -4.62469369e-01 -7.14371562e-01
1.32175708e+00 -2.54456326e-02 2.46476501e-01 2.93249905e-01
1.02998757e+00 -6.11379892e-02 -9.41171408e-01 -6.46875262e-01
1.71431437e-01 -7.89089128e-02 -1.52564701e-02 -6.42210960e-01
-5.60680211e-01 9.01579857e-01 -8.72713886e-03 -1.59052193e-01
6.91915989e-01 3.82328451e-01 1.02569473e+00 7.91313946e-01
3.68005425e-01 -1.30962396e+00 5.51890075e-01 7.69302785e-01
8.48023772e-01 -1.33525050e+00 -1.27952918e-01 -1.90913588e-01
-8.69793177e-01 5.43194294e-01 9.24696565e-01 -1.80404440e-01
2.64933258e-01 8.94652307e-03 -1.71778545e-01 -2.18891412e-01
-1.11433291e+00 -5.07252812e-01 4.43267018e-01 5.93526542e-01
6.05336726e-01 -3.50871794e-02 -6.08186543e-01 9.84321296e-01
-2.03870818e-01 7.28568286e-02 3.72934610e-01 6.71081364e-01
-7.58308351e-01 -9.78218436e-01 8.56876448e-02 6.79263949e-01
-6.29063621e-02 -7.30786085e-01 -3.73633564e-01 5.52374423e-01
-4.62710172e-01 8.86052489e-01 4.20160666e-02 -1.19672744e-02
1.54651497e-02 5.89295268e-01 6.17652526e-03 -1.44643283e+00
-8.99783075e-02 2.65318245e-01 5.78681976e-02 -4.25418884e-01
-7.41183385e-02 -3.03841263e-01 -1.34517002e+00 3.02751213e-01
-3.96466911e-01 5.02321482e-01 3.37085426e-02 9.27082658e-01
6.25809610e-01 5.54408193e-01 1.21981837e-01 -6.68388963e-01
-7.82010794e-01 -1.45384729e+00 -4.13723618e-01 5.08719742e-01
-3.04180719e-02 -5.35891175e-01 -3.53728682e-01 1.22668140e-01] | [11.517375946044922, 8.587799072265625] |
3b79efad-9dc1-49ea-9aba-8a801489c9c2 | high-precision-machine-learning-based-indoor | 2303.03743 | null | https://arxiv.org/abs/2303.03743v1 | https://arxiv.org/pdf/2303.03743v1.pdf | High-Precision Machine-Learning Based Indoor Localization with Massive MIMO System | High-precision cellular-based localization is one of the key technologies for next-generation communication systems. In this paper, we investigate the potential of applying machine learning (ML) to a massive multiple-input multiple-output (MIMO) system to enhance localization accuracy. We analyze a new ML-based localization pipeline that has two parallel fully connected neural networks (FCNN). The first FCNN takes the instantaneous spatial covariance matrix to capture angular information, while the second FCNN takes the channel impulse responses to capture delay information. We fuse the estimated coordinates of these two FCNNs for further accuracy improvement. To test the localization algorithm, we performed an indoor measurement campaign with a massive MIMO testbed at 3.7GHz. In the measured scenario, the proposed pipeline can achieve centimeter-level accuracy by combining delay and angular information. | ['Fredrik Tufvesson', 'Liang Liu', 'Xuesong Cai', 'Michiel Sandra', 'Ilayda Yaman', 'Guoda Tian'] | 2023-03-07 | null | null | null | null | ['indoor-localization'] | ['computer-vision'] | [-3.63264501e-01 -1.09461263e-01 -1.92717955e-01 -7.95777440e-02
-1.03994238e+00 -5.76824009e-01 2.00854465e-01 1.86433136e-01
-2.87000656e-01 1.12635911e+00 -2.90102363e-01 -9.72006738e-01
-4.09052163e-01 -7.86963701e-01 -6.71733201e-01 -7.42861629e-01
-6.26463115e-01 2.10607409e-01 -1.81746230e-01 1.90142989e-01
-1.06663920e-01 8.87163639e-01 -4.28562760e-01 5.93811162e-02
5.67860425e-01 1.59334815e+00 -1.19779453e-01 1.09544098e+00
3.26460958e-01 4.22082305e-01 -1.00726378e+00 1.79754883e-01
2.88399786e-01 1.36369050e-01 -1.06953584e-01 -6.89712882e-01
2.13265151e-01 -3.54467869e-01 -6.57423317e-01 3.85101497e-01
9.58477497e-01 -3.87914300e-01 2.31775552e-01 -1.26551807e+00
8.29587579e-02 6.62951171e-01 -8.34600776e-02 8.94613862e-02
-1.10993665e-02 -3.74612093e-01 3.71448576e-01 -6.15811408e-01
4.92853448e-02 8.59082639e-01 1.33751547e+00 -4.75489527e-01
-9.84821022e-01 -9.45404708e-01 -4.53357339e-01 -1.48093358e-01
-1.91704202e+00 -4.78233427e-01 3.52470547e-01 -1.96316808e-01
6.15392864e-01 1.56886622e-01 5.02007544e-01 6.31670535e-01
6.19630277e-01 8.91115516e-02 8.33826721e-01 -5.31493425e-01
4.97016370e-01 7.87229761e-02 -3.33695829e-01 6.01200759e-01
4.36572403e-01 2.76220173e-01 -2.33891442e-01 -4.02275264e-01
1.06201065e+00 5.24440967e-03 -1.37513056e-02 -1.58361033e-01
-1.35653138e+00 3.72009009e-01 9.66993809e-01 7.12042511e-01
-3.20791960e-01 1.14958954e+00 -3.51941288e-01 2.34982580e-01
1.61429152e-01 7.06052661e-01 -6.95872307e-01 -1.25242442e-01
-1.30040073e+00 -2.15278536e-01 8.17318857e-01 1.19413781e+00
8.19267631e-01 2.51012027e-01 -2.91216612e-01 1.28354385e-01
5.16390860e-01 1.11271429e+00 1.09059727e-02 -9.21446323e-01
5.64971209e-01 -1.75431594e-02 2.26871759e-01 -1.14971542e+00
-1.22607660e+00 -1.61583388e+00 -1.21836853e+00 -3.97033930e-01
4.82791275e-01 -1.14784527e+00 -4.36859220e-01 1.44938588e+00
-1.55813724e-01 8.21432412e-01 1.48382187e-01 4.72506762e-01
2.79852450e-01 4.04272974e-01 -3.93782109e-01 9.25452039e-02
6.22025192e-01 -8.68403673e-01 -5.20806968e-01 -2.61772070e-02
1.04891491e+00 -6.65529430e-01 -2.34167695e-01 9.39405635e-02
-6.28892243e-01 -7.05659747e-01 -1.51642776e+00 6.23580217e-01
-3.91999871e-01 1.05808222e+00 6.97614968e-01 1.10507739e+00
-1.28626084e+00 5.53175628e-01 -8.46543789e-01 -4.20463294e-01
1.26027927e-01 8.62854362e-01 2.85524689e-02 1.44408122e-01
-1.22608364e+00 2.97526896e-01 1.25572234e-01 3.84654641e-01
-3.30416739e-01 -7.05729663e-01 -4.61592257e-01 3.03408593e-01
-2.65538841e-01 -4.57109511e-01 1.09089029e+00 -3.12101245e-01
-1.43675756e+00 -4.01182592e-01 -3.56839031e-01 -7.00449109e-01
1.14955358e-01 2.38483742e-01 -9.22855020e-01 9.21972916e-02
2.99955551e-02 3.28345448e-01 3.13123733e-01 -1.07037246e+00
-9.72981930e-01 -1.01336613e-01 -1.09650552e-01 -5.91270447e-01
-1.45632491e-01 -8.64290118e-01 -3.58435810e-01 -4.10712093e-01
5.75295568e-01 -1.32401884e+00 -4.68534321e-01 -3.85930926e-01
-4.68839318e-01 7.30141163e-01 6.30089462e-01 -5.67903697e-01
1.27630973e+00 -2.09936047e+00 -5.75333178e-01 9.21054006e-01
2.59498209e-01 1.28076263e-02 9.84415859e-02 6.29176915e-01
3.64922822e-01 7.94620439e-02 6.95223272e-01 -4.31934714e-01
-1.22973666e-01 -1.83461681e-01 5.29726921e-03 5.32142580e-01
-2.49805495e-01 1.02702618e+00 -6.85212612e-01 1.61874354e-01
1.46698937e-01 4.07025754e-01 -5.40983200e-01 -3.12898457e-01
3.86645168e-01 1.03634501e+00 -5.57360411e-01 7.05171108e-01
9.19782341e-01 -5.63697278e-01 3.22232664e-01 -6.03238046e-01
-4.21729773e-01 -1.54604644e-01 -1.08869827e+00 1.46613717e+00
-1.10426855e+00 1.10638571e+00 1.68042526e-01 -5.80554783e-01
8.05751324e-01 4.69582230e-01 8.07761252e-01 -8.30403090e-01
3.76584768e-01 5.59506476e-01 4.59459163e-02 -4.63259071e-02
1.36837617e-01 4.78792816e-01 -5.22642612e-01 3.46571147e-01
2.54674375e-01 2.68027276e-01 -4.08035457e-01 1.83989137e-01
1.62408245e+00 -4.35741514e-01 1.72170460e-01 -2.81788915e-01
7.27031827e-01 -2.78589576e-01 4.08301026e-01 1.16342545e+00
8.00662790e-04 2.34560013e-01 1.32693842e-01 -3.32141012e-01
-8.28952253e-01 -1.21608388e+00 7.96655864e-02 3.98144364e-01
2.27437496e-01 -2.72582918e-01 -2.81295002e-01 -2.94901729e-01
3.44542742e-01 3.31613928e-01 -7.53412023e-02 6.20882623e-02
-2.60408014e-01 -5.63598037e-01 1.32506573e+00 3.94188643e-01
9.41309810e-01 1.86608866e-01 5.38918637e-02 5.41807532e-01
-1.09052792e-01 -1.31294036e+00 6.32410273e-02 5.29909849e-01
-2.91356087e-01 -4.23047215e-01 -4.73334670e-01 -5.67590356e-01
5.01273930e-01 3.22169542e-01 5.45164406e-01 -2.40093749e-02
1.70979142e-01 3.29636544e-01 4.56785643e-03 -2.88284898e-01
5.45767993e-02 6.37637794e-01 4.98365581e-01 1.06066354e-01
-1.90871418e-01 -8.55262458e-01 -3.73637527e-01 3.84706587e-01
1.34678155e-01 -3.54154289e-01 1.15851462e+00 3.86579782e-01
2.39070669e-01 4.45248842e-01 6.46676540e-01 -3.03486645e-01
3.64403427e-01 -6.07967854e-01 -1.05078292e+00 1.89227745e-01
-2.71046907e-01 1.23488791e-01 7.34317064e-01 -2.06275016e-01
-5.63962638e-01 3.98032635e-01 -3.92780900e-01 1.08752079e-01
3.94433849e-02 8.67798746e-01 -1.79445475e-01 -1.03495789e+00
5.89887559e-01 -9.99439210e-02 -6.82199001e-01 -1.18455045e-01
2.22720012e-01 9.95585561e-01 5.56481540e-01 -3.38604987e-01
1.08283675e+00 4.61799890e-01 6.33388221e-01 -1.03053987e+00
-5.16103566e-01 -4.97273296e-01 -7.56078541e-01 -1.61676794e-01
2.52037257e-01 -1.48602986e+00 -1.19997060e+00 7.27572218e-02
-1.32448483e+00 -4.83162463e-01 5.40095925e-01 1.07029486e+00
-1.86629891e-01 -3.30681354e-01 -4.70130831e-01 -6.97239578e-01
-5.02199233e-02 -8.80056739e-01 1.02275896e+00 4.08054441e-01
1.21123724e-01 -1.18807983e+00 3.63267139e-02 -5.44942208e-02
1.08216345e+00 1.65237874e-01 5.08118033e-01 -4.66696560e-01
-1.02906632e+00 -7.53939211e-01 -4.19974566e-01 -2.23008201e-01
-1.71135873e-01 -6.03689313e-01 -9.27023470e-01 -5.57095110e-01
-5.08236766e-01 4.01546121e-01 2.59153873e-01 7.57370174e-01
8.57769012e-01 -1.04208723e-01 -9.39267755e-01 1.00016844e+00
1.56286764e+00 1.54696509e-01 5.80536962e-01 -1.14226798e-02
5.35827279e-01 -5.49726427e-01 2.78275609e-01 3.96348417e-01
3.45855534e-01 8.31953704e-01 2.95571119e-01 -3.39609206e-01
5.03651705e-03 -2.90709846e-02 2.36261114e-02 6.49585605e-01
2.10545093e-01 -6.18165672e-01 -8.50231171e-01 -1.41306147e-02
-1.87483346e+00 -7.01764643e-01 -2.86927193e-01 2.09446597e+00
-8.23726803e-02 1.93948746e-01 -4.39786285e-01 -6.53747693e-02
4.80808616e-01 -2.90811151e-01 -1.40392348e-01 8.02576393e-02
9.17013884e-02 1.56511873e-01 1.52811909e+00 7.57408082e-01
-1.26566660e+00 5.31258583e-01 5.42298603e+00 7.74249494e-01
-1.62856007e+00 3.42222035e-01 3.82091582e-01 1.56313255e-01
2.27956697e-01 9.30761453e-03 -9.05401111e-01 3.66152048e-01
1.52051377e+00 3.06861639e-01 2.59921759e-01 7.61642992e-01
3.93906355e-01 -1.57122821e-01 -9.41975713e-01 1.12520206e+00
-3.35269839e-01 -1.80831802e+00 -7.29135752e-01 4.74656254e-01
8.19756925e-01 2.94063926e-01 2.40819067e-01 5.72513700e-01
-4.59224023e-02 -8.51319969e-01 3.70774537e-01 6.93945229e-01
8.75948787e-01 -8.88599932e-01 1.43798316e+00 5.10252059e-01
-1.67979670e+00 -4.45483714e-01 -6.43665195e-02 -4.29340005e-01
2.15589792e-01 1.13401163e+00 -1.45156062e+00 7.95355022e-01
1.97239280e-01 2.21337453e-01 -7.04717219e-01 1.55502677e+00
3.11321020e-02 5.62230170e-01 -6.81583345e-01 -2.15306327e-01
3.62929791e-01 3.13289732e-01 1.45419553e-01 1.13182104e+00
1.12131321e+00 -1.76145464e-01 2.40811214e-01 3.61017972e-01
-9.44366530e-02 -4.11134005e-01 -3.43402326e-01 3.94858688e-01
1.08617580e+00 1.56890321e+00 -6.54024124e-01 -2.03033201e-02
-8.86592045e-02 6.99676633e-01 -1.07975163e-01 6.18776441e-01
-8.73662293e-01 -6.55216694e-01 3.62575531e-01 1.36972278e-01
1.73764914e-01 -1.01605475e+00 -3.76287609e-01 -6.20127201e-01
-4.10914510e-01 -1.04968980e-01 -4.94600743e-01 -5.73164403e-01
-5.78143418e-01 2.64135987e-01 -6.71139956e-01 -1.28968096e+00
-2.93205291e-01 -4.16823804e-01 -4.00827140e-01 9.43005681e-01
-1.25244534e+00 -1.44853234e+00 -4.64728028e-01 3.02895695e-01
-2.61694551e-01 -3.49053711e-01 9.98138785e-01 9.94655073e-01
-3.90216380e-01 7.95684993e-01 8.82657468e-01 3.90600026e-01
6.11477375e-01 -9.73866761e-01 3.40044737e-01 8.11742127e-01
3.70027572e-02 7.52531528e-01 2.44420826e-01 -7.18013465e-01
-1.33309340e+00 -1.53361690e+00 9.66368079e-01 -1.39417008e-01
4.32126313e-01 -6.07151926e-01 2.55216539e-01 7.27699399e-01
-1.41690418e-01 5.31780660e-01 7.05347955e-01 8.00446421e-02
2.07689688e-01 -5.71611524e-01 -1.03975022e+00 4.49945301e-01
4.64331299e-01 -3.37634355e-01 7.07830310e-01 3.39568108e-01
5.04426301e-01 -4.47033376e-01 -1.03513634e+00 4.18713838e-01
9.41401064e-01 -4.98040825e-01 1.23246026e+00 2.69034594e-01
-4.92314398e-01 -5.29996872e-01 -5.50307333e-01 -1.55996883e+00
-5.92261791e-01 -5.36330462e-01 -3.84658456e-01 9.96087909e-01
7.47328639e-01 -7.02779770e-01 1.05569375e+00 -2.57893741e-01
-6.51178462e-03 -4.39665705e-01 -1.07520461e+00 -9.43891704e-01
-2.68081337e-01 -5.20112216e-01 7.47495770e-01 4.26355422e-01
-1.00004092e-01 4.84734535e-01 -4.16224539e-01 1.26871967e+00
4.73003656e-01 -4.77981567e-01 1.02154362e+00 -1.29961848e+00
-5.02976477e-01 -1.29419237e-01 -6.18704617e-01 -1.42071629e+00
-2.35776722e-01 -6.65963888e-01 -1.14138477e-01 -1.47482932e+00
-8.51467788e-01 -1.15079665e+00 -3.69009823e-01 8.18294063e-02
6.89156830e-01 6.18292451e-01 -1.46993965e-01 -2.26739064e-01
-9.38376904e-01 1.60929703e-04 5.40620983e-01 -2.56737322e-01
-3.10799330e-01 8.06836188e-01 -3.14920187e-01 5.86925685e-01
1.25984693e+00 -2.75710851e-01 -1.37424693e-01 -5.45506358e-01
2.27281749e-01 4.79504228e-01 4.17338282e-01 -2.19606757e+00
8.13407540e-01 3.49878669e-01 1.27072775e+00 -7.99937725e-01
5.16961217e-01 -1.21884894e+00 4.34716731e-01 7.79513597e-01
1.36127919e-01 -1.76760226e-01 2.75369316e-01 5.62065840e-01
5.75384647e-02 5.84459364e-01 3.61803561e-01 5.41894495e-01
-3.63147318e-01 2.51748323e-01 -6.48453236e-01 -7.98242450e-01
9.30009723e-01 3.38528901e-01 -3.66689444e-01 -9.55509186e-01
-5.49757361e-01 2.36218572e-01 -1.22988723e-01 -1.17730789e-01
2.68715799e-01 -1.60378408e+00 -1.57122359e-01 2.93983221e-01
-8.29071552e-02 -7.09161043e-01 6.98189065e-02 1.18545103e+00
-6.98833168e-01 1.42799520e+00 5.46528772e-02 -7.67018259e-01
-8.69746268e-01 7.18929619e-02 7.33026028e-01 -3.52574766e-01
3.67334604e-01 7.15402901e-01 -7.54125476e-01 -7.84043074e-01
3.70113105e-01 -4.44741935e-01 1.79520309e-01 -4.93537456e-01
2.95381546e-01 4.96376216e-01 4.40442234e-01 -3.71050596e-01
-6.53191805e-01 5.95873237e-01 5.67511976e-01 -4.74758953e-01
7.95589447e-01 -4.39833313e-01 3.69279459e-02 2.81330571e-02
1.49143457e+00 6.06403530e-01 -8.11509073e-01 -1.69413000e-01
9.08982605e-02 1.79101024e-02 2.12980434e-01 -9.84091818e-01
-9.21672106e-01 7.41123855e-01 9.71457243e-01 1.34686828e-01
8.20512116e-01 -3.02743822e-01 6.22252405e-01 1.01840949e+00
1.16711390e+00 -8.51736248e-01 -2.03563422e-01 7.81609058e-01
-9.19254199e-02 -8.77484798e-01 3.48441489e-02 -2.48714268e-01
5.30973375e-01 1.24512351e+00 2.60764927e-01 4.53657582e-02
1.09932637e+00 7.50664413e-01 1.81995943e-01 1.55101523e-01
-1.66230500e-01 -1.33074686e-01 6.12189854e-03 7.49099255e-01
2.58000612e-01 1.51116595e-01 1.78916886e-01 8.09185028e-01
-2.43006676e-01 -1.99732095e-01 2.27021992e-01 8.20453286e-01
-6.16027653e-01 -1.11606324e+00 -7.22385228e-01 5.29630780e-01
-1.10158965e-01 -6.76605327e-04 -6.41999096e-02 6.08288765e-01
5.98037064e-01 1.23813713e+00 -3.28466147e-02 -9.28927660e-01
5.46322279e-02 -5.39000690e-01 3.72424960e-01 -2.94473827e-01
-2.25421861e-01 1.04972214e-01 1.55289192e-02 -6.51121438e-01
3.48791748e-01 -2.53584713e-01 -1.35623574e+00 -3.82581353e-01
-3.01334530e-01 3.34932148e-01 1.19541645e+00 1.09202957e+00
7.35982895e-01 1.12329853e+00 9.07614648e-01 -9.78215158e-01
2.88488041e-03 -6.74423873e-01 -5.61056793e-01 -9.02656376e-01
6.34622455e-01 -4.43009049e-01 -7.15627447e-02 -5.77488482e-01] | [6.37555456161499, 0.9720033407211304] |
33dea82c-0faf-408e-9503-4670e8c8fe70 | learning-contact-based-navigation-in-crowds | 2303.01455 | null | https://arxiv.org/abs/2303.01455v1 | https://arxiv.org/pdf/2303.01455v1.pdf | Learning Contact-based Navigation in Crowds | Navigation strategies that intentionally incorporate contact with humans (i.e. "contact-based" social navigation) in crowded environments are largely unexplored even though collision-free social navigation is a well studied problem. Traditional social navigation frameworks require the robot to stop suddenly or "freeze" whenever a collision is imminent. This paradigm poses two problems: 1) freezing while navigating a crowd may cause people to trip and fall over the robot, resulting in more harm than the collision itself, and 2) in very dense social environments where collisions are unavoidable, such a control scheme would render the robot unable to move and preclude the opportunity to study how humans incorporate robots into these environments. However, if robots are to be meaningfully included in crowded social spaces, such as busy streets, subways, stores, or other densely populated locales, there may not exist trajectories that can guarantee zero collisions. Thus, adoption of robots in these environments requires the development of minimally disruptive navigation plans that can safely plan for and respond to contacts. We propose a learning-based motion planner and control scheme to navigate dense social environments using safe contacts for an omnidirectional mobile robot. The planner is evaluated in simulation over 360 trials with crowd densities varying between 0.0 and 1.6 people per square meter. Our navigation scheme is able to use contact to safely navigate in crowds of higher density than has been previously reported, to our knowledge. | ['Luis Sentis', 'Junfeng Jiao', 'Kyle Morgenstein'] | 2023-03-02 | null | null | null | null | ['social-navigation'] | ['robots'] | [-4.57895510e-02 4.14240211e-01 3.83116126e-01 3.93700833e-03
-1.07798256e-01 -3.67939115e-01 5.03191352e-01 1.96380526e-01
-1.01337302e+00 1.07615709e+00 2.78585255e-02 -4.20448452e-01
-2.13429421e-01 -1.06842458e+00 -4.55539584e-01 -6.14655554e-01
-3.83840442e-01 8.07453990e-01 7.15090156e-01 -7.46203184e-01
2.25752532e-01 4.79925573e-01 -1.63652122e+00 -6.65477991e-01
1.07704902e+00 6.95435405e-02 5.59143782e-01 6.14558280e-01
-1.08455950e-02 4.21258509e-01 -4.15643841e-01 1.51018068e-01
2.72939473e-01 -1.71812311e-01 -5.78334272e-01 4.00977470e-02
-3.80991548e-01 -3.93126041e-01 -2.43601695e-01 5.74175715e-01
5.64061284e-01 6.56621695e-01 7.86120534e-01 -1.48155558e+00
7.98233449e-02 2.22256958e-01 -5.19844890e-01 -7.72372335e-02
7.01072693e-01 4.19683248e-01 3.25495660e-01 -4.91203427e-01
5.82097113e-01 1.27145290e+00 5.60510099e-01 4.53821450e-01
-6.90553784e-01 -3.22762489e-01 1.33919895e-01 -7.97395781e-02
-1.42636764e+00 -4.28353816e-01 4.12154384e-02 -2.99554825e-01
1.09077728e+00 1.76304281e-02 8.68775070e-01 8.45942914e-01
6.69939697e-01 1.78766564e-01 5.11132836e-01 -1.21725783e-01
7.30367899e-01 -1.36194289e-01 -3.77365381e-01 4.83984202e-01
8.95884395e-01 -1.90692961e-01 -2.43377000e-01 -2.10941032e-01
3.02528977e-01 2.60104313e-02 -9.64489877e-02 -5.22431791e-01
-1.10962152e+00 7.86806345e-01 6.11711800e-01 2.05474228e-01
-7.19540536e-01 1.72152027e-01 2.79908389e-01 1.04376145e-01
-5.56464456e-02 3.50450933e-01 9.14495811e-02 -4.18659508e-01
-4.26956445e-01 7.31752694e-01 9.99248743e-01 1.03880060e+00
6.30199194e-01 -4.58112247e-02 5.33230662e-01 5.31058431e-01
2.56772280e-01 8.39734375e-01 1.73945963e-01 -1.18713760e+00
5.28954089e-01 6.32193327e-01 8.57598305e-01 -1.31071031e+00
-8.69092822e-01 2.80513167e-01 -6.51525497e-01 5.61813414e-01
6.92777514e-01 -6.56272054e-01 -3.41273338e-01 1.44559991e+00
5.92191517e-01 -2.51992643e-01 2.56979078e-01 9.56736028e-01
3.19789350e-01 4.71773267e-01 -1.36393057e-02 -1.79095030e-01
1.02468681e+00 -5.08449674e-01 -5.21009684e-01 -5.38221240e-01
8.06783199e-01 -3.72009337e-01 9.91016507e-01 9.45701078e-02
-8.59535158e-01 6.55535683e-02 -1.02471542e+00 3.21778446e-01
-3.21657896e-01 -6.63336217e-01 2.80998707e-01 5.00029385e-01
-1.19309580e+00 3.16498548e-01 -1.00987387e+00 -1.07507312e+00
-4.62983176e-02 4.93637621e-01 -3.84660333e-01 -2.20478848e-01
-1.12768221e+00 1.18502200e+00 -3.07165682e-01 2.33732108e-02
-6.10131621e-01 1.36556804e-01 -9.36779261e-01 -2.67007828e-01
4.59099889e-01 -7.17988133e-01 1.04603398e+00 -3.39143574e-01
-1.27555835e+00 4.37784106e-01 -3.43883604e-01 -4.23360020e-01
8.97138298e-01 -2.13541687e-01 7.74893910e-02 -3.10568139e-02
7.22876787e-01 7.06391037e-01 1.34282291e-01 -1.45731378e+00
-8.03433061e-01 -1.87786594e-01 3.29862952e-01 8.09040129e-01
-4.77376729e-02 -2.85022259e-01 -1.34868994e-01 1.34549886e-01
1.40143558e-01 -1.47353506e+00 -9.56209183e-01 -7.08355382e-02
-2.85140306e-01 -1.51659787e-01 5.92685699e-01 7.98224062e-02
6.49772227e-01 -1.82716691e+00 -1.90760776e-01 3.30605030e-01
3.03808320e-02 -1.00073710e-01 9.77119207e-02 7.97597229e-01
9.04448628e-01 8.84430576e-03 -3.14866781e-01 -5.22590697e-01
-1.06648721e-01 5.18819392e-01 2.41587937e-01 6.41083062e-01
-4.63935107e-01 3.97834569e-01 -1.33077109e+00 -3.43099892e-01
1.13349333e-01 3.37008476e-01 -7.10974932e-01 -9.11431089e-02
3.39500636e-01 5.07018745e-01 -4.71452922e-01 4.47608739e-01
5.88715374e-01 3.41546237e-01 1.52165204e-01 9.11833644e-01
-7.26630986e-01 9.02895704e-02 -1.26543117e+00 8.86941135e-01
-4.53783870e-01 5.48113585e-01 4.17420059e-01 -4.64680880e-01
7.22482443e-01 2.22430360e-02 4.90744263e-01 -5.30697286e-01
8.19361880e-02 5.05933464e-01 -2.17621222e-01 -5.81392288e-01
1.08856428e+00 -5.08990623e-02 -3.17045361e-01 5.68365335e-01
-9.05885696e-01 -2.77803302e-01 3.95206437e-02 1.34963959e-01
1.52912903e+00 -1.44980118e-01 2.04223812e-01 -4.82690483e-01
1.90287545e-01 3.43759656e-01 6.76432490e-01 1.02328467e+00
-6.35541379e-01 2.67742008e-01 2.95157768e-02 -1.93793103e-01
-8.50330889e-01 -1.11446321e+00 2.56825417e-01 9.95700598e-01
1.09019387e+00 -1.41600836e-02 -6.85775876e-01 -4.95211892e-02
4.64645214e-02 6.09802663e-01 -2.68934786e-01 3.42626050e-02
-8.95772219e-01 -3.94166410e-01 3.49600792e-01 1.11892171e-01
6.22094810e-01 -1.10924125e+00 -1.29003596e+00 4.50381905e-01
-3.96836430e-01 -9.74852026e-01 -2.94905096e-01 4.44582961e-02
-3.53542268e-01 -1.13890707e+00 -6.01961434e-01 -7.31879532e-01
9.25397456e-01 1.02055740e+00 4.63196605e-01 2.81477779e-01
2.68129036e-02 6.99261725e-01 -4.42818642e-01 -3.36406976e-01
-2.15795308e-01 7.17781708e-02 5.61987281e-01 -4.54431385e-01
3.82401824e-01 -6.24620736e-01 -8.79423201e-01 6.93170249e-01
-3.91088635e-01 -1.89352378e-01 1.21815838e-01 2.97400594e-01
8.49555805e-03 3.11384857e-01 7.80267775e-01 -2.28115022e-01
7.20966697e-01 -1.00166893e+00 -3.79733384e-01 -3.74335766e-01
-1.61819860e-01 -4.47428823e-01 5.34397006e-01 -2.55004048e-01
-8.60150754e-01 1.35992453e-01 8.80763903e-02 4.63944852e-01
-3.47292870e-01 1.98374569e-01 -1.47013098e-01 -2.67485678e-02
7.33662486e-01 -1.48132026e-01 9.81008708e-02 1.69728771e-01
2.34813482e-01 8.74086857e-01 2.30777949e-01 -3.96935761e-01
7.48267651e-01 8.90316010e-01 1.53207779e-01 -1.42933774e+00
4.36509192e-01 -5.26345849e-01 -4.55943435e-01 -4.76280332e-01
8.09799612e-01 -8.01873505e-01 -9.35322642e-01 3.53143960e-01
-9.91822362e-01 -6.64890826e-01 1.99345872e-02 4.10220683e-01
-6.94782197e-01 4.38584954e-01 -2.89348483e-01 -1.33258843e+00
2.27516487e-01 -1.20270371e+00 6.56353652e-01 4.72215652e-01
-7.13271856e-01 -8.36783588e-01 -6.13753982e-02 1.24073587e-01
6.11515403e-01 3.59541088e-01 2.30080158e-01 -2.31489420e-01
-6.79815412e-01 5.86271957e-02 2.32848644e-01 -6.55512571e-01
2.75139421e-01 -2.99109519e-01 -2.55495310e-01 -3.62533689e-01
-1.73714221e-01 -9.53495875e-02 2.51517028e-01 3.20000499e-01
-2.60381341e-01 -3.67696971e-01 -9.54205632e-01 -1.01202317e-01
8.98979187e-01 5.64273715e-01 5.90618789e-01 8.55002105e-01
3.37499291e-01 1.20359993e+00 9.01790500e-01 6.27674639e-01
1.15045333e+00 6.11944616e-01 5.64855278e-01 1.39198333e-01
4.87415850e-01 -9.64466631e-02 5.23630559e-01 4.19770598e-01
-9.83227491e-02 -6.21466517e-01 -1.29621053e+00 9.70178068e-01
-2.03090906e+00 -7.03757405e-01 -4.25219446e-01 2.23531222e+00
2.27448300e-01 1.08885519e-01 1.29801422e-01 1.51531905e-01
8.52667809e-01 -2.89578080e-01 -4.55185682e-01 -3.17256302e-01
8.09643567e-02 -6.49770021e-01 9.36559916e-01 1.08385193e+00
-7.31798947e-01 9.20623243e-01 5.59356928e+00 1.61325544e-01
-6.47469938e-01 -5.99941202e-02 4.54773307e-02 -7.37705827e-02
-2.74220914e-01 2.07691595e-01 -7.92054296e-01 4.15274978e-01
7.34233618e-01 -9.06287730e-02 2.94613957e-01 7.41941690e-01
8.60788107e-01 -1.06865382e+00 -5.29070139e-01 5.71904838e-01
-3.13330859e-01 -5.16251862e-01 -3.38416100e-01 3.31383169e-01
3.86284649e-01 3.43181938e-02 -3.12401026e-01 1.77190572e-01
7.49953806e-01 -9.26084101e-01 7.85118341e-01 4.49535519e-01
2.55247712e-01 -9.65314150e-01 6.26321077e-01 9.14351761e-01
-1.28537095e+00 -2.06018865e-01 -4.23444331e-01 -6.18731797e-01
9.52764511e-01 3.80267173e-01 -1.03050792e+00 1.06240310e-01
7.38719404e-01 4.96167876e-02 -8.90459940e-02 1.22774553e+00
4.15525287e-02 -4.74990606e-02 -7.70234764e-01 -7.10944414e-01
5.86037934e-01 -3.33477378e-01 9.83470500e-01 7.70905733e-01
7.23035455e-01 3.18795413e-01 3.75551015e-01 1.91632643e-01
5.64592659e-01 3.29431370e-02 -1.33604193e+00 6.25348330e-01
7.86210775e-01 6.82900667e-01 -1.01711702e+00 1.27487972e-01
-9.65043828e-02 8.94737780e-01 4.08685626e-03 4.36263919e-01
-6.03828430e-01 -5.54431736e-01 7.39524603e-01 6.47999227e-01
-2.79208779e-01 -7.71106899e-01 -3.34781468e-01 -6.25696599e-01
-1.11403517e-01 -2.64524579e-01 -2.31712982e-01 -5.83407879e-01
-8.01432610e-01 4.61190552e-01 7.51499310e-02 -1.21566260e+00
-4.42400664e-01 -5.75250536e-02 -5.77529669e-01 4.41994846e-01
-1.08406293e+00 -6.67928338e-01 -4.58683014e-01 3.57644618e-01
4.84650820e-01 -1.46650776e-01 4.49696541e-01 1.53031766e-01
1.07018731e-03 1.77649185e-01 4.16764356e-02 -4.40228075e-01
4.27928388e-01 -9.55390513e-01 3.82259279e-01 7.23973930e-01
-8.85472953e-01 7.50777066e-01 9.66072738e-01 -1.18765366e+00
-1.30758595e+00 -9.65238333e-01 1.10736799e+00 -2.57500768e-01
2.54865974e-01 -3.47712785e-01 -5.84653139e-01 3.81683677e-01
-4.87136394e-02 -4.12073731e-01 4.74446505e-01 -2.74316162e-01
5.41113615e-01 3.77220303e-01 -1.45103025e+00 1.41477859e+00
1.42852652e+00 8.65395665e-02 -2.15481311e-01 2.25004032e-01
5.30290067e-01 -5.98298311e-02 -3.50292511e-02 2.67466784e-01
4.75458890e-01 -1.10798323e+00 7.52163470e-01 2.38082752e-01
-2.64238358e-01 -5.15731871e-01 -2.78712004e-01 -1.39364672e+00
6.18288256e-02 -8.36094737e-01 6.44352734e-01 8.93139720e-01
3.55180144e-01 -9.75639105e-01 7.36056209e-01 8.71905088e-01
-1.11932717e-01 -4.41828668e-01 -1.38457322e+00 -8.87852073e-01
3.91077816e-01 -2.88138896e-01 4.72368509e-01 4.89105105e-01
3.53828758e-01 1.19900383e-01 -3.35279614e-01 3.38808328e-01
5.55204809e-01 -7.81821072e-01 1.12741995e+00 -9.52313662e-01
3.92514765e-01 -3.23162228e-01 -3.54209781e-01 -1.08486938e+00
1.72408134e-01 -3.55155796e-01 7.13319898e-01 -2.08804822e+00
-3.13682139e-01 -1.10403061e+00 8.44846487e-01 2.28236377e-01
5.67741469e-02 -7.73450285e-02 1.39181808e-01 3.03343087e-01
-6.90162539e-01 6.87469900e-01 9.65240717e-01 1.68041468e-01
-6.82349086e-01 1.50681198e-01 -4.57673490e-01 8.95125091e-01
9.79444802e-01 -3.40764195e-01 -5.49622416e-01 -1.81459282e-02
3.31462950e-01 4.44456458e-01 2.66776755e-02 -1.34349442e+00
8.05778086e-01 -5.04771948e-01 -3.60391885e-01 -4.26305771e-01
4.60362226e-01 -8.86772037e-01 2.81200439e-01 8.43568623e-01
2.06382245e-01 1.73201531e-01 -6.51935115e-02 8.14776301e-01
2.70828635e-01 -3.00555110e-01 6.84713662e-01 -1.94774568e-01
-4.64392871e-01 -1.94832459e-01 -1.53932929e+00 -1.28928378e-01
1.47006273e+00 -6.71888292e-01 -1.81749552e-01 -8.98935080e-01
-4.59235519e-01 8.82273793e-01 1.04907203e+00 2.71202117e-01
6.38313591e-01 -9.53363836e-01 -4.40861166e-01 -2.22460106e-01
-7.55555481e-02 1.74393564e-01 2.53040671e-01 6.01618648e-01
-9.45664942e-01 2.14389548e-01 -8.01490396e-02 -4.22660500e-01
-9.98161316e-01 2.11937860e-01 3.67435515e-01 1.94945648e-01
-7.02537060e-01 4.00265753e-01 1.22187108e-01 -6.55584633e-01
1.40874162e-01 -1.21154323e-01 -2.84030497e-01 -9.01505053e-02
3.30024600e-01 7.60724604e-01 -3.93182248e-01 -1.02800810e+00
-5.72970450e-01 5.70849836e-01 4.29085195e-01 -5.16781867e-01
1.05427980e+00 -9.08430338e-01 9.65619087e-02 3.22637498e-01
4.98896390e-01 1.86069533e-01 -1.44258273e+00 1.72379360e-01
1.11568555e-01 -3.77407610e-01 -5.39286375e-01 2.11753175e-02
-2.47118816e-01 3.15962821e-01 3.34076464e-01 2.14549556e-01
4.45467830e-01 -2.29607850e-01 9.73481059e-01 8.65422249e-01
1.33456528e+00 -1.17456877e+00 1.26379088e-01 9.58169520e-01
6.97195470e-01 -1.09395266e+00 -2.25193098e-01 -3.16008747e-01
-7.28154600e-01 6.30148888e-01 7.30759799e-01 -1.85313568e-01
4.93221283e-01 4.21611041e-01 9.14995819e-02 3.35906595e-02
-3.70319396e-01 -4.65655208e-01 -7.63935268e-01 1.08424544e+00
-1.60548866e-01 2.18673348e-01 -4.88745093e-01 3.16041410e-01
-5.45826972e-01 -2.99384773e-01 1.25376225e+00 1.54535246e+00
-1.20533538e+00 -7.89256334e-01 -7.93311894e-01 2.23298550e-01
3.58275026e-02 3.91280055e-01 -1.04362249e-01 8.39373827e-01
2.05422953e-01 1.62626851e+00 2.31081858e-01 -2.31904984e-01
5.29938400e-01 -2.96810210e-01 -8.96823704e-02 -4.88832891e-01
-2.54019797e-01 -3.15627575e-01 3.68545562e-01 -3.27889085e-01
-2.95200855e-01 -9.95445251e-01 -1.98526740e+00 -6.23014092e-01
-3.65852416e-02 2.90180743e-01 5.40149391e-01 9.50718641e-01
1.76862344e-01 -3.31547074e-02 4.52380925e-01 -1.36970878e+00
-1.47030633e-02 -6.85905218e-01 -5.63079834e-01 -1.41392857e-01
3.36934924e-01 -1.16941011e+00 -5.32940269e-01 -3.71487945e-01] | [4.87291145324707, 1.1047884225845337] |
a9e5b38e-7120-47ca-8eb2-a193725c3aef | geometric-latent-diffusion-models-for-3d | 2305.01140 | null | https://arxiv.org/abs/2305.01140v1 | https://arxiv.org/pdf/2305.01140v1.pdf | Geometric Latent Diffusion Models for 3D Molecule Generation | Generative models, especially diffusion models (DMs), have achieved promising results for generating feature-rich geometries and advancing foundational science problems such as molecule design. Inspired by the recent huge success of Stable (latent) Diffusion models, we propose a novel and principled method for 3D molecule generation named Geometric Latent Diffusion Models (GeoLDM). GeoLDM is the first latent DM model for the molecular geometry domain, composed of autoencoders encoding structures into continuous latent codes and DMs operating in the latent space. Our key innovation is that for modeling the 3D molecular geometries, we capture its critical roto-translational equivariance constraints by building a point-structured latent space with both invariant scalars and equivariant tensors. Extensive experiments demonstrate that GeoLDM can consistently achieve better performance on multiple molecule generation benchmarks, with up to 7\% improvement for the valid percentage of large biomolecules. Results also demonstrate GeoLDM's higher capacity for controllable generation thanks to the latent modeling. Code is provided at \url{https://github.com/MinkaiXu/GeoLDM}. | ['Jure Leskovec', 'Stefano Ermon', 'Ron Dror', 'Alexander Powers', 'Minkai Xu'] | 2023-05-02 | null | null | null | null | ['3d-molecule-generation'] | ['medical'] | [-2.24804990e-02 1.02043673e-01 -2.14039817e-01 -7.30366409e-02
-5.88952959e-01 -6.76901340e-01 9.10323381e-01 -3.95096168e-02
2.79595554e-01 8.16395044e-01 5.94629467e-01 -4.20612037e-01
-2.45760828e-02 -1.02868629e+00 -9.48318064e-01 -1.03863192e+00
-1.14775114e-01 5.23928940e-01 -5.66058755e-01 -1.62528262e-01
1.93185300e-01 6.56612873e-01 -8.96292448e-01 1.41458780e-01
1.03810024e+00 5.07297277e-01 8.57588928e-03 4.87168491e-01
3.63652259e-02 5.06635249e-01 -2.99135327e-01 -4.09609139e-01
2.95866638e-01 -5.79854488e-01 -6.28337920e-01 -1.82729185e-01
2.30289191e-01 -1.05724387e-01 -6.43974304e-01 8.28239322e-01
7.05593050e-01 2.22924352e-01 1.22413933e+00 -1.03036976e+00
-1.25758231e+00 6.86183870e-01 -2.81940520e-01 -3.20227146e-01
3.94250661e-01 3.40068281e-01 9.48017478e-01 -1.17843807e+00
1.05103040e+00 1.46551001e+00 3.34066510e-01 7.72881508e-01
-1.59586000e+00 -5.81118524e-01 2.14947201e-03 -2.66886532e-01
-1.44009650e+00 -4.84810740e-01 9.29790974e-01 -7.34214664e-01
1.14278722e+00 1.71928436e-01 4.54106212e-01 1.50477040e+00
6.70087397e-01 6.08500719e-01 8.21798563e-01 -1.76952079e-01
4.10765588e-01 -3.86838466e-01 -2.66682148e-01 7.80259073e-01
3.52351069e-01 7.00635323e-03 -6.66485012e-01 -5.02985716e-01
1.14928627e+00 2.37985089e-01 -2.37029213e-02 -6.09039426e-01
-1.54143107e+00 1.17739058e+00 5.65380991e-01 5.45002930e-02
-5.78314066e-01 3.48297715e-01 -4.55688164e-02 -1.72584725e-04
5.77155709e-01 6.74788654e-01 -9.24164779e-04 -3.49422619e-02
-5.86121500e-01 5.70411205e-01 6.50410771e-01 1.20694125e+00
4.88577545e-01 3.28838974e-01 -2.05335319e-01 3.94306719e-01
5.47768056e-01 4.97399956e-01 3.66057873e-01 -8.45218182e-01
3.82724375e-01 5.39330959e-01 1.41977489e-01 -1.04818332e+00
-2.15108976e-01 -4.39003110e-01 -1.26168144e+00 -2.03952879e-01
-1.96424965e-02 -1.60036549e-01 -9.74914432e-01 1.68000007e+00
4.53067511e-01 -1.56594664e-01 2.78807998e-01 6.35210812e-01
9.00914907e-01 1.10539293e+00 2.31615156e-01 -2.59878725e-01
7.79583991e-01 -8.17606330e-01 -5.66922247e-01 4.48507279e-01
7.10744500e-01 -8.35087061e-01 4.79132533e-01 2.09838137e-01
-1.13394082e+00 -4.83848304e-01 -9.75091517e-01 -3.12437028e-01
-3.66668522e-01 3.15066357e-03 1.26173449e+00 3.36286098e-01
-9.80324388e-01 8.57866228e-01 -9.67831552e-01 -6.88531548e-02
3.21860790e-01 4.21172827e-01 -4.26836073e-01 -6.21752590e-02
-1.17643821e+00 2.73531973e-01 3.34240347e-01 8.85996222e-02
-1.37746191e+00 -7.76245296e-01 -8.83330345e-01 -2.45937020e-01
-1.21373594e-01 -1.17839456e+00 7.13874161e-01 -2.14336291e-01
-1.76200092e+00 4.48427290e-01 -3.08204532e-01 -3.59847307e-01
5.17859340e-01 -2.37409007e-02 -1.33031353e-01 1.09960455e-02
3.30378637e-02 1.09039450e+00 8.31552982e-01 -1.07402575e+00
3.61397803e-01 -1.65736005e-01 -1.97757214e-01 4.04356904e-02
-2.55185366e-01 -5.44463634e-01 -1.29901338e-02 -9.51137602e-01
3.40620220e-01 -1.17350507e+00 -6.64404094e-01 -3.92745972e-01
-7.94636965e-01 -3.09048772e-01 6.17658496e-01 -3.77557188e-01
1.04224551e+00 -1.53709090e+00 9.08276856e-01 1.74254864e-01
7.80510366e-01 -5.81420027e-02 -1.97423294e-01 9.59294260e-01
-4.20494407e-01 3.00666571e-01 -2.53545672e-01 -3.46966058e-01
2.31530845e-01 -1.25293165e-01 -5.40018797e-01 2.70218372e-01
1.24530718e-01 1.36858916e+00 -9.51988995e-01 1.14640623e-01
2.45345369e-01 9.12998378e-01 -8.16528738e-01 1.07447371e-01
-6.00051403e-01 7.04150975e-01 -5.84628463e-01 7.13194847e-01
5.44899464e-01 -6.89826488e-01 1.58174574e-01 -2.50460535e-01
-1.56677678e-01 1.64064139e-01 -7.53121316e-01 1.99662662e+00
2.17896253e-02 1.95717767e-01 -5.47982454e-01 -4.77605075e-01
1.11296833e+00 2.83964932e-01 8.07179391e-01 -3.51121277e-01
-8.97017494e-02 2.64054716e-01 -8.65968317e-02 2.05963571e-02
4.94285971e-01 -1.42440602e-01 -9.81549919e-02 4.39155698e-01
2.37876087e-01 -2.45031059e-01 8.45970809e-02 5.16632795e-01
7.62522340e-01 3.57633621e-01 -7.11334124e-02 -4.06868905e-01
2.29064777e-01 -2.00850040e-01 3.66047740e-01 5.54664373e-01
2.26530403e-01 3.88859957e-01 5.27409673e-01 -6.89353943e-01
-1.44932318e+00 -1.14217794e+00 -3.78116742e-02 4.56675470e-01
-5.00039347e-02 -7.64503598e-01 -7.76196420e-01 -3.15892607e-01
9.69100744e-02 4.19256717e-01 -6.14118457e-01 -4.76335883e-01
-3.42676371e-01 -1.15065920e+00 3.43319416e-01 2.72039056e-01
1.16182357e-01 -7.81508327e-01 2.69298494e-01 4.71299231e-01
1.29712805e-01 -6.64717615e-01 -5.02627432e-01 -8.18223432e-02
-1.00114667e+00 -5.73307514e-01 -1.12964940e+00 -4.66086775e-01
8.69175911e-01 2.24307001e-01 7.99418926e-01 -5.24506092e-01
-5.47128081e-01 8.44763070e-02 -1.36632398e-01 -2.74804562e-01
-5.96534014e-01 2.96576053e-01 4.96611506e-01 -1.44433305e-02
2.60475934e-01 -8.51857126e-01 -1.00465345e+00 5.39315306e-02
-8.78281653e-01 5.39783597e-01 5.17704546e-01 7.42082655e-01
8.97268355e-01 -1.65547103e-01 5.13433933e-01 -6.92328870e-01
7.29896009e-01 -6.01421475e-01 -5.81498921e-01 -1.52322456e-01
-6.77965105e-01 4.86541092e-01 4.02973950e-01 -2.82107711e-01
-8.51914644e-01 -8.27540234e-02 -3.37843031e-01 -4.99340922e-01
-1.08023241e-01 4.19953436e-01 -3.43809307e-01 -1.67329330e-02
6.37095332e-01 4.33542222e-01 -8.69616109e-04 -4.95324969e-01
6.72127724e-01 1.83760196e-01 -7.38263726e-02 -9.72081482e-01
8.49832654e-01 5.43761551e-01 3.78776520e-01 -8.26440215e-01
-4.99588132e-01 -1.03289507e-01 -7.66228139e-01 1.09028734e-01
1.03225195e+00 -1.01484561e+00 -7.77662575e-01 5.51257610e-01
-1.17657399e+00 -4.26078588e-01 -2.11251304e-01 5.12703300e-01
-7.03010440e-01 4.14221972e-01 -9.10405636e-01 -4.64572012e-01
-5.62794507e-01 -1.43183422e+00 1.22659206e+00 -9.55544878e-03
-2.77151138e-01 -1.23991072e+00 3.07386905e-01 1.79304630e-01
4.06345576e-01 7.84737110e-01 1.32517958e+00 -1.21096350e-01
-1.07497180e+00 -1.04654342e-01 2.95439482e-01 -1.14558943e-01
2.98397809e-01 9.83419865e-02 -7.28090763e-01 -4.14003581e-01
-2.95792133e-01 -1.13789812e-01 1.00143218e+00 6.00103140e-01
9.48507905e-01 -4.21907246e-01 -5.54655433e-01 9.11335707e-01
1.25594234e+00 1.80170655e-01 5.05181134e-01 -2.68686652e-01
1.24053395e+00 2.82098472e-01 5.70107996e-02 4.88983482e-01
2.36754641e-01 6.24601841e-01 2.77205765e-01 -1.86006367e-01
-8.63352120e-02 -8.81585240e-01 5.95015347e-01 1.37122357e+00
-3.72253805e-01 -3.59915227e-01 -7.78370380e-01 1.12004533e-01
-1.76045454e+00 -1.05152035e+00 -1.80059850e-01 2.03368735e+00
8.63353133e-01 -3.30310687e-02 1.93739578e-01 -3.17142725e-01
5.41120291e-01 3.11688811e-01 -9.69442427e-01 -2.30332077e-01
-2.99144477e-01 1.30210072e-01 4.05518591e-01 6.99518025e-01
-9.57828045e-01 1.17349327e+00 6.62497663e+00 8.94316435e-01
-1.16446865e+00 -4.22036760e-02 8.27356458e-01 -4.23801504e-02
-8.64868581e-01 -5.63911442e-03 -1.13354492e+00 4.39297199e-01
9.06805694e-01 -2.97070354e-01 1.78742051e-01 8.90539587e-01
4.00201231e-01 5.74833035e-01 -9.08836782e-01 9.04081166e-01
-1.44858986e-01 -2.01848269e+00 7.90585220e-01 5.58292150e-01
1.30532539e+00 -7.36600161e-02 4.76518124e-01 7.85602406e-02
4.54579502e-01 -1.25045919e+00 5.06747603e-01 9.25397396e-01
1.00426674e+00 -8.04504752e-01 7.78432488e-02 2.12225631e-01
-1.00938725e+00 4.52628434e-01 -6.35122478e-01 9.07437354e-02
1.32831275e-01 7.81415284e-01 -7.87015796e-01 4.52269703e-01
-1.93002187e-02 1.16310596e+00 -1.19304098e-01 6.29353285e-01
-1.33577868e-01 4.84538704e-01 -2.81852633e-02 -1.18236393e-01
2.57967085e-01 -7.49267936e-01 7.19523132e-01 9.63695049e-01
4.10345435e-01 6.35074545e-03 7.42119923e-02 1.40399885e+00
-3.22986811e-01 6.62041223e-03 -8.53053808e-01 -6.41821802e-01
4.10075784e-01 1.03047574e+00 -5.61231852e-01 -1.76786199e-01
1.40821889e-01 1.07589388e+00 4.21240600e-03 6.70162797e-01
-8.92998457e-01 -2.79236168e-01 8.42146218e-01 4.86862622e-02
1.78645402e-01 -9.48233545e-01 2.46192701e-02 -1.34673488e+00
-2.20364317e-01 -7.69586205e-01 -6.80733174e-02 -6.62074387e-01
-1.29220319e+00 4.78893280e-01 -2.25287706e-01 -9.67837155e-01
-8.00647289e-02 -8.07625771e-01 -2.79999197e-01 1.05449021e+00
-1.19597590e+00 -1.38655651e+00 -2.06589028e-02 3.03670555e-01
4.64280456e-01 -2.55254447e-01 1.16120362e+00 1.10909246e-01
-6.89046741e-01 4.03358519e-01 5.41732728e-01 -1.64268896e-01
5.86173415e-01 -1.32216752e+00 9.56093013e-01 4.39749599e-01
1.71434954e-01 1.35095000e+00 5.59743345e-01 -9.27333653e-01
-1.85249650e+00 -1.25093746e+00 5.95695436e-01 -8.13630104e-01
4.68871862e-01 -7.15164125e-01 -5.76165199e-01 5.28749347e-01
-4.53014709e-02 -4.93164271e-01 1.01756811e+00 -5.40072955e-02
-4.25632983e-01 3.60231608e-01 -6.57447696e-01 7.70643413e-01
1.29873967e+00 -6.09288335e-01 2.27575526e-02 9.07287002e-01
1.20543432e+00 -5.91971993e-01 -1.38964200e+00 3.14087033e-01
4.19644773e-01 -6.43577337e-01 1.26681387e+00 -9.27360177e-01
7.28856802e-01 -2.96245486e-01 -7.21474588e-02 -1.14327466e+00
-6.66874468e-01 -1.16526318e+00 -7.16069400e-01 8.94622147e-01
4.87142801e-01 -4.67744470e-01 9.24907804e-01 6.19904757e-01
-2.05909491e-01 -1.03103781e+00 -5.76837063e-01 -6.39391780e-01
4.77325052e-01 -1.28397923e-02 8.17034960e-01 1.04475188e+00
-2.72402704e-01 3.42396051e-01 -5.59350491e-01 -3.48957069e-02
7.13464320e-01 2.92719364e-01 7.22892106e-01 -9.72682536e-01
-3.24053377e-01 -4.62346673e-01 -3.66906404e-01 -1.52964354e+00
6.98271301e-03 -1.29680097e+00 -4.75385845e-01 -1.52101934e+00
1.58338681e-01 -3.86681974e-01 -1.83598489e-01 1.75789386e-01
4.47659232e-02 2.54422985e-02 -4.57381718e-02 5.05026460e-01
-5.37259579e-01 1.16062152e+00 1.58279848e+00 -2.03655779e-01
-3.43040437e-01 -2.88780212e-01 -7.86980212e-01 2.92054772e-01
7.29419112e-01 -2.41061598e-01 -4.16022778e-01 -3.59505355e-01
5.07750750e-01 6.79242685e-02 2.39284500e-01 -9.81428802e-01
4.22437675e-03 -4.07110661e-01 6.18091106e-01 -5.66927612e-01
4.51721132e-01 -9.46098268e-02 4.40978080e-01 4.75743592e-01
-4.52374965e-01 3.25012691e-02 3.78625765e-02 7.64162958e-01
1.69585925e-02 3.68833899e-01 5.00006020e-01 -2.62936652e-01
-1.82722956e-01 1.07926929e+00 -2.90647477e-01 -3.52266103e-01
8.97637367e-01 -1.19157191e-02 -2.55481094e-01 -3.00735146e-01
-8.80740702e-01 -1.53540298e-01 7.05680370e-01 4.36109304e-01
6.86978281e-01 -1.61326337e+00 -5.26408017e-01 2.31962442e-01
-1.63788609e-02 2.57346749e-01 4.59523290e-01 5.66897750e-01
-6.39512897e-01 9.65624452e-01 -4.36306223e-02 -5.21496058e-01
-7.26182461e-01 6.68112993e-01 3.18413824e-01 -2.77673192e-02
-4.13432896e-01 1.13718760e+00 4.49953824e-01 -3.85344058e-01
-2.01874882e-01 -2.91014969e-01 2.84782529e-01 -1.32400259e-01
3.18497032e-01 1.82626680e-01 -1.69250175e-01 -6.47682071e-01
6.26646951e-02 5.20660639e-01 -3.08244318e-01 1.36452213e-01
1.57164395e+00 2.56499797e-01 -2.05702558e-01 2.11093813e-01
1.25473654e+00 2.30873570e-01 -1.43490279e+00 -1.98999303e-03
-3.22146446e-01 -2.75642127e-02 -1.45567164e-01 -2.81679660e-01
-5.91634452e-01 1.09867263e+00 2.77482718e-01 -1.07091248e-01
3.47530484e-01 -8.02366361e-02 9.66403425e-01 4.69133765e-01
4.85653400e-01 -5.64048827e-01 4.01846379e-01 4.34006482e-01
1.06168902e+00 -9.21491683e-01 6.96793497e-02 -2.60829777e-01
-5.24773717e-01 1.03977728e+00 2.68284351e-01 -1.59536228e-01
5.04268050e-01 -1.85789943e-01 -2.38234460e-01 -4.58049953e-01
-7.72300720e-01 2.49436527e-01 3.45978588e-01 5.15821874e-01
7.91452229e-01 3.63443345e-01 1.97551958e-02 2.47916311e-01
-2.62708038e-01 -3.53931487e-01 2.93030530e-01 7.06628561e-01
-2.67016888e-01 -1.61792982e+00 -7.77741820e-02 1.89264476e-01
-2.62934491e-02 -3.19949299e-01 -5.43783128e-01 3.03185791e-01
3.46749052e-02 4.29277033e-01 -3.32036495e-01 -2.77690679e-01
-1.11782342e-01 1.30714580e-01 5.68704963e-01 -6.38854980e-01
1.60421565e-01 2.87567109e-01 -4.31673020e-01 -5.03742456e-01
-2.14423537e-01 -6.80649877e-01 -1.09971869e+00 -5.26047707e-01
-4.57774475e-02 5.62020600e-01 5.90654671e-01 3.56616735e-01
1.00893462e+00 4.56921667e-01 6.29843056e-01 -9.50155854e-01
-4.13892537e-01 -7.60507703e-01 -3.41802031e-01 3.64518493e-01
1.73741996e-01 -6.45160615e-01 1.45371873e-02 1.88916162e-01] | [5.057343006134033, 5.732901573181152] |
aa55b43d-6f60-48df-960a-68bf3c92cc0d | explainable-authorship-verification-in-social | 1910.08144 | null | https://arxiv.org/abs/1910.08144v2 | https://arxiv.org/pdf/1910.08144v2.pdf | Explainable Authorship Verification in Social Media via Attention-based Similarity Learning | Authorship verification is the task of analyzing the linguistic patterns of two or more texts to determine whether they were written by the same author or not. The analysis is traditionally performed by experts who consider linguistic features, which include spelling mistakes, grammatical inconsistencies, and stylistics for example. Machine learning algorithms, on the other hand, can be trained to accomplish the same, but have traditionally relied on so-called stylometric features. The disadvantage of such features is that their reliability is greatly diminished for short and topically varied social media texts. In this interdisciplinary work, we propose a substantial extension of a recently published hierarchical Siamese neural network approach, with which it is feasible to learn neural features and to visualize the decision-making process. For this purpose, a new large-scale corpus of short Amazon reviews for text comparison research is compiled and we show that the Siamese network topologies outperform state-of-the-art approaches that were built up on stylometric features. Our linguistic analysis of the internal attention weights of the network shows that the proposed method is indeed able to latch on to some traditional linguistic categories. | ['Robert M. Nickel', 'Dorothea Kolossa', 'Benedikt Boenninghoff', 'Steffen Hessler'] | 2019-10-17 | null | null | null | null | ['authorship-verification'] | ['natural-language-processing'] | [-8.62683132e-02 -5.40704988e-02 -7.12394789e-02 -2.46695966e-01
-2.87843674e-01 -5.80444753e-01 8.77651751e-01 7.83540726e-01
-7.02105463e-01 4.87334520e-01 1.28601477e-01 -2.91270941e-01
-2.23070249e-01 -5.66374362e-01 -2.54312724e-01 -4.82191801e-01
2.39115313e-01 7.56696224e-01 1.02096032e-02 -4.43010539e-01
8.52899730e-01 5.78587353e-01 -1.73492944e+00 -3.71719003e-02
9.32434082e-01 9.43962157e-01 -2.37244293e-01 3.80385041e-01
-6.22414708e-01 7.01757073e-01 -7.23777115e-01 -7.57113039e-01
-6.88640475e-02 -3.37281555e-01 -6.99360788e-01 -2.47194067e-01
7.72910237e-01 2.58836120e-01 8.66203681e-02 1.27736676e+00
2.25373805e-01 5.77229485e-02 9.71324742e-01 -1.11277139e+00
-8.84961486e-01 8.91105354e-01 -5.97644269e-01 3.43120039e-01
3.36472452e-01 -1.57882541e-01 1.51208663e+00 -8.30262423e-01
6.86815977e-01 1.18516731e+00 7.39517331e-01 1.72944605e-01
-1.19081748e+00 -4.33828950e-01 1.00888140e-01 2.55342096e-01
-1.04332352e+00 -2.01952562e-01 1.09048641e+00 -8.38419616e-01
5.47697186e-01 -3.65066230e-02 5.18949151e-01 1.01156163e+00
8.38434920e-02 4.99402612e-01 1.23771107e+00 -6.84305012e-01
9.93429497e-02 2.74372607e-01 5.93510211e-01 8.65182281e-01
4.36873138e-01 -3.15698713e-01 -6.27649665e-01 -1.37106767e-02
1.35239422e-01 -1.27548590e-01 -1.63300157e-01 -3.05270523e-01
-1.33570504e+00 9.81669784e-01 4.26579595e-01 1.08670557e+00
-1.65105283e-01 -1.35544285e-01 7.44959831e-01 3.24604154e-01
6.23762190e-01 9.39509571e-01 -1.06521688e-01 -4.68569174e-02
-1.39033175e+00 4.64389957e-02 1.05092132e+00 3.32039446e-01
7.97038138e-01 -1.22395493e-01 -7.87267759e-02 6.85485184e-01
1.55061916e-01 2.30180144e-01 9.27815020e-01 -6.75283909e-01
5.00770628e-01 8.88580978e-01 -4.36668210e-02 -1.59609377e+00
-5.72335660e-01 -3.35541666e-01 -9.37017322e-01 2.85791308e-01
1.00438118e+00 2.03564554e-01 -3.93592417e-01 1.34245491e+00
-5.18357903e-02 -4.81124401e-01 -1.03560559e-01 7.21256256e-01
7.59695113e-01 2.22161800e-01 -1.73796758e-01 8.96335691e-02
1.28443491e+00 -7.86378145e-01 -7.89442539e-01 2.02765137e-01
5.64121068e-01 -6.70847178e-01 1.32454574e+00 6.11790121e-01
-1.02340019e+00 -4.93817210e-01 -1.10313272e+00 -2.18881350e-02
-9.66866076e-01 3.83397996e-01 3.46922666e-01 7.66273737e-01
-9.06956136e-01 1.20573699e+00 -5.11298358e-01 -4.53999281e-01
3.09746921e-01 2.26717666e-01 -3.50899130e-01 4.98220861e-01
-1.20016551e+00 1.12030709e+00 1.81989238e-01 1.82951316e-01
-3.70661654e-02 -3.63923520e-01 -7.45838225e-01 2.42963567e-01
2.70137876e-01 -1.94227174e-01 9.43304598e-01 -1.38248312e+00
-1.71330976e+00 1.22418272e+00 -6.38281330e-02 -4.47752416e-01
7.66189218e-01 9.86730456e-02 -3.41737062e-01 1.42418787e-01
1.42389685e-01 5.57519719e-02 8.88695478e-01 -9.56677556e-01
-4.30684328e-01 -5.54287195e-01 -2.31260881e-01 -7.86192641e-02
-7.86415994e-01 1.63475141e-01 -6.68464378e-02 -7.17493773e-01
-1.29012555e-01 -8.31485748e-01 2.05732822e-01 -5.05635142e-02
-6.00529492e-01 -6.10892713e-01 4.77776349e-01 -7.47759581e-01
1.23726714e+00 -1.89088881e+00 3.74102861e-01 5.88533640e-01
5.67517102e-01 2.75246590e-01 6.85716197e-02 3.76279444e-01
6.44579828e-02 4.08001721e-01 -1.74527347e-01 -4.91384119e-01
2.85777599e-01 -2.82496840e-01 -1.09923020e-01 6.51435614e-01
8.34208727e-03 6.74053431e-01 -8.69636357e-01 -6.40603542e-01
-3.71390283e-02 2.10690036e-01 -1.46588475e-01 -1.57020092e-01
5.04705682e-02 2.44370326e-01 -1.24871142e-01 3.15045983e-01
2.57081211e-01 -3.03174198e-01 2.47324124e-01 -5.49454615e-02
-3.03559244e-01 3.70489001e-01 -9.14935648e-01 1.45333302e+00
-3.65090847e-01 1.13047612e+00 -1.02601066e-01 -1.14483666e+00
1.07141173e+00 -2.47018635e-02 1.01303019e-01 -6.03650033e-01
4.74924177e-01 5.05236864e-01 2.46506512e-01 -4.12762374e-01
7.20911860e-01 7.11961687e-02 -1.33272603e-01 7.55287290e-01
-3.58552374e-02 4.92872372e-02 7.25137532e-01 1.90190420e-01
6.88317239e-01 9.69152618e-03 4.26218241e-01 -5.44607937e-01
1.00901735e+00 -1.56752944e-01 1.89885899e-01 6.29397094e-01
-2.24574581e-01 4.69893336e-01 1.00971270e+00 -5.06987929e-01
-1.08878040e+00 -6.53355598e-01 -1.32812202e-01 1.17024958e+00
-2.72156924e-01 -2.83943981e-01 -9.02713120e-01 -7.48071849e-01
2.51619071e-01 5.11191964e-01 -1.05092692e+00 1.42271236e-01
-4.94037479e-01 -3.73953223e-01 6.30434155e-01 1.85311109e-01
1.53987601e-01 -1.18046200e+00 -7.17577398e-01 2.05510780e-02
1.27547920e-01 -8.59367788e-01 -3.62883002e-01 1.01761319e-01
-6.72156334e-01 -1.20551753e+00 -1.02385736e+00 -5.56658447e-01
4.93732810e-01 -2.29102343e-01 1.25265658e+00 2.89568275e-01
-6.97918683e-02 7.10956752e-02 -3.35845888e-01 -2.96452224e-01
-7.11017847e-01 5.98363996e-01 2.28121147e-01 2.90459454e-01
4.43425089e-01 -6.06075108e-01 -1.46153525e-01 8.23444650e-02
-7.67498493e-01 -3.56004238e-01 4.43300307e-01 9.44403410e-01
1.64983153e-01 -2.68506974e-01 5.50400734e-01 -1.15682495e+00
1.04504561e+00 -3.02231103e-01 -5.85478902e-01 4.13522631e-01
-8.08985114e-01 3.73698860e-01 1.11954582e+00 -3.97693485e-01
-6.06571853e-01 -3.07642281e-01 1.91087246e-01 -2.48639017e-01
-1.28935501e-01 7.83121467e-01 1.56164557e-01 -1.63368329e-01
6.12198353e-01 7.99425319e-02 1.60563484e-01 -5.04108846e-01
3.03331196e-01 6.23851836e-01 4.62479651e-01 -4.83676881e-01
7.63088465e-01 2.49689370e-01 1.62764311e-01 -7.95665920e-01
-7.75662720e-01 -2.84804016e-01 -1.04619110e+00 -3.14982235e-01
6.87888265e-01 -2.12688789e-01 -1.05499458e+00 5.01415789e-01
-1.12672555e+00 2.56962739e-02 -1.25005037e-01 1.50360033e-01
-2.96639174e-01 7.91833878e-01 -4.28104788e-01 -7.24923968e-01
-2.24236846e-01 -9.27241206e-01 7.26971090e-01 1.67968348e-01
-5.11273503e-01 -1.21847522e+00 2.11548761e-01 2.90230721e-01
5.05132735e-01 1.01514019e-01 1.21451402e+00 -1.10035694e+00
9.61531475e-02 -2.74538398e-01 -2.57415593e-01 1.93915442e-01
2.01898851e-02 5.08563697e-01 -8.23712349e-01 -7.61777237e-02
-2.75501847e-01 -2.32761398e-01 9.42564309e-01 1.28339291e-01
1.14233291e+00 -3.34757209e-01 -6.37878254e-02 1.39553174e-01
1.07444692e+00 -1.38408825e-01 3.28210562e-01 4.81625021e-01
7.23813593e-01 1.19127667e+00 2.09032372e-01 2.08053827e-01
2.86841571e-01 7.19988763e-01 2.44313464e-01 1.03340149e-01
1.81810334e-01 -1.38644531e-01 5.94026707e-02 1.02335930e+00
-2.37114280e-01 -2.25579455e-01 -1.12525570e+00 4.53045309e-01
-1.74552679e+00 -1.00632405e+00 -2.34928697e-01 2.10748768e+00
6.39191329e-01 3.68775308e-01 3.80660564e-01 6.73383653e-01
8.39773893e-01 2.41031766e-01 -1.97069943e-01 -7.00022340e-01
-2.35529080e-01 5.38832806e-02 2.12756857e-01 4.09521848e-01
-1.09982061e+00 7.60169685e-01 5.19229841e+00 7.15755045e-01
-1.21886849e+00 -7.56966770e-02 5.06300628e-01 4.70218062e-02
-1.60842001e-01 -4.12045479e-01 -5.79421461e-01 6.58294022e-01
9.72414017e-01 -4.09119390e-02 3.37057203e-01 5.86856544e-01
1.07447967e-01 -2.75803715e-01 -1.20301974e+00 1.02498722e+00
5.05416274e-01 -1.30542886e+00 6.76649213e-02 1.71482358e-02
5.29692531e-01 -2.78263301e-01 1.50304943e-01 -2.12674923e-02
-4.07777913e-02 -1.07695079e+00 8.91843557e-01 6.93240762e-01
5.80798864e-01 -6.62989914e-01 9.08112109e-01 1.35361746e-01
-8.94731879e-01 -8.35315436e-02 -6.56419098e-02 -3.60531360e-02
-7.48503655e-02 5.38557589e-01 -4.81339246e-01 3.19865257e-01
6.65344894e-01 9.73483741e-01 -8.97934437e-01 8.40139866e-01
-1.92783862e-01 3.77124935e-01 -1.07756972e-01 -5.95491648e-01
3.19088250e-01 -4.76667285e-01 6.10563040e-01 1.21792090e+00
2.48054057e-01 -7.36618221e-01 -2.91247219e-01 9.96386051e-01
-3.14348489e-01 6.58613980e-01 -7.35944986e-01 -3.63778949e-01
7.58748353e-02 1.42493439e+00 -1.04847193e+00 -3.83388460e-01
-3.15788716e-01 9.10583198e-01 6.74763441e-01 3.35115846e-03
-4.04162258e-01 -6.76613033e-01 2.54984766e-01 1.13035396e-01
1.53776452e-01 -1.55725643e-01 -4.81412023e-01 -1.20489299e+00
1.24751247e-01 -8.04030478e-01 1.76523402e-01 -4.61173445e-01
-1.64793861e+00 7.09521532e-01 -5.20454824e-01 -1.04458070e+00
-1.83519512e-01 -1.01008570e+00 -7.17107534e-01 9.40818012e-01
-1.42760682e+00 -8.59038234e-01 -2.11169779e-01 3.34287107e-01
2.68251985e-01 -6.45531654e-01 6.00844562e-01 2.45760560e-01
-7.59118676e-01 6.90693617e-01 3.97416562e-01 2.44587541e-01
7.23604620e-01 -1.53607941e+00 9.32312459e-02 6.32634819e-01
4.84173477e-01 6.38132989e-01 7.20537066e-01 -4.96394724e-01
-8.12092543e-01 -4.85355914e-01 1.54932666e+00 -4.79641795e-01
1.23041189e+00 -3.33228022e-01 -9.72292125e-01 1.88877404e-01
3.30380559e-01 -3.37472826e-01 6.80521429e-01 3.23186725e-01
-5.19311607e-01 -6.83321878e-02 -8.93210948e-01 5.91014922e-01
6.41940773e-01 -6.99295044e-01 -8.33789349e-01 1.69289529e-01
9.04723704e-02 1.18119739e-01 -7.98254967e-01 -1.29040733e-01
7.77703941e-01 -1.15213811e+00 5.53158939e-01 -7.15193689e-01
8.17322314e-01 -1.64617777e-01 3.24911535e-01 -1.28875887e+00
-5.84446341e-02 -4.75087136e-01 1.34026572e-01 1.34808731e+00
5.95658779e-01 -7.14808881e-01 7.49244332e-01 3.12603861e-01
1.53705716e-01 -5.85787833e-01 -9.63389456e-01 -6.08047009e-01
3.86809498e-01 4.36911499e-03 3.49855006e-01 1.11536837e+00
2.97222555e-01 4.89357442e-01 -1.24600485e-01 -5.70941567e-01
3.75560850e-01 2.28112623e-01 4.83451068e-01 -2.07770085e+00
-1.77850425e-02 -1.28257298e+00 -4.15660739e-01 -2.80717492e-01
6.91356599e-01 -1.12082326e+00 -2.61303961e-01 -1.18688798e+00
-1.48187745e-02 -3.65001470e-01 -2.76837200e-01 1.73365280e-01
-5.89704998e-02 3.71528566e-01 4.12003845e-01 4.09229815e-01
-5.09137392e-01 4.14613813e-01 1.02086592e+00 -3.04684967e-01
-9.41485092e-02 -3.58670542e-04 -5.54007351e-01 8.95926535e-01
7.60664761e-01 -3.84073704e-01 1.94070265e-01 -7.70256892e-02
8.12288284e-01 -4.60110337e-01 2.33278185e-01 -9.55412745e-01
3.39141458e-01 3.15552950e-01 2.54472107e-01 -3.66529256e-01
1.14726955e-02 -7.70657003e-01 -5.31629920e-01 4.99193251e-01
-6.22506142e-01 4.09129500e-01 -1.84104547e-01 5.57715476e-01
-4.52587605e-01 -5.08165061e-01 6.44931018e-01 -1.43328115e-01
-2.44331837e-01 -9.26299691e-02 -6.29389286e-01 1.37814134e-01
6.09186471e-01 -1.51954636e-01 -2.94937193e-01 -3.64181072e-01
-5.17988205e-01 -7.46691227e-02 5.44972241e-01 3.47183436e-01
1.74552754e-01 -1.02464032e+00 -6.05949283e-01 -1.28216848e-01
1.57827511e-01 -6.47640765e-01 -1.03289992e-01 1.11674416e+00
-6.29540324e-01 4.96152818e-01 -4.39007372e-01 -3.40978324e-01
-1.27815962e+00 6.32999182e-01 3.90318334e-01 -6.27682984e-01
-3.27973902e-01 4.19559389e-01 -4.23263550e-01 -4.49517339e-01
2.72219092e-01 -3.93488288e-01 -8.40060711e-01 9.36976612e-01
2.82973528e-01 6.27854943e-01 2.57440358e-01 -9.57494974e-01
-2.74923235e-01 8.64343643e-01 1.02355972e-01 -1.78529590e-01
1.22366297e+00 3.37949544e-02 -4.04119432e-01 9.11984861e-01
1.06404471e+00 3.43183130e-01 -5.74559152e-01 -2.95997143e-01
4.76736575e-01 -1.66973606e-01 -1.82816282e-01 -6.86201036e-01
-1.07972181e+00 1.33900416e+00 3.11466873e-01 7.72390008e-01
6.31723225e-01 -2.90962249e-01 4.41638023e-01 7.16034114e-01
1.08723976e-01 -1.44338000e+00 1.03204045e-02 6.93449736e-01
8.38713169e-01 -1.36393964e+00 -9.53880325e-02 7.64017605e-05
-5.78755558e-01 1.67641699e+00 3.06038439e-01 -2.50173241e-01
5.54658055e-01 -7.88154975e-02 1.18560813e-01 -2.60667920e-01
-6.90339804e-02 -2.08297014e-01 6.41830504e-01 2.48007372e-01
6.69772029e-01 -8.36346671e-02 -6.37692451e-01 5.00348926e-01
-4.80436713e-01 -2.83108592e-01 5.35264552e-01 4.06615138e-01
-7.91051760e-02 -1.11697233e+00 -3.69939446e-01 6.35637283e-01
-6.57258987e-01 -2.08238497e-01 -8.16311061e-01 7.84280121e-01
-4.17791680e-02 7.17891634e-01 1.33378327e-01 -1.96542263e-01
1.39220491e-01 3.31380546e-01 3.38057637e-01 -3.12869340e-01
-1.23599863e+00 -4.98220623e-01 9.21416506e-02 -3.35384965e-01
-6.93320036e-01 -9.04635131e-01 -7.66510010e-01 -3.63934040e-01
-1.27848849e-01 2.59867191e-01 8.21287096e-01 1.06635559e+00
-2.91196145e-02 4.50241596e-01 3.88679713e-01 -1.22674143e+00
-5.54110646e-01 -1.05921960e+00 -7.74680197e-01 7.06805468e-01
3.40460747e-01 -7.48338580e-01 -6.44307256e-01 -1.44288197e-01] | [9.604730606079102, 10.523531913757324] |
f5661891-6b97-4808-bca8-a8f3ab8e65db | 191013276 | 1910.13276 | null | https://arxiv.org/abs/1910.13276v2 | https://arxiv.org/pdf/1910.13276v2.pdf | a novel cross-lingual voice cloning approach with a few text-free samples | In this paper, we present a cross-lingual voice cloning approach. BN features obtained by SI-ASR model are used as a bridge across speakers and language boundaries. The relationships between text and BN features are modeled by the latent prosody model. The acoustic model learns the translation from BN features to acoustic features. The acoustic model is fine-tuned with a few samples of the target speaker to realize voice cloning. This system can generate speech of arbitrary utterance of target language in cross-lingual speakers' voice. We verify that with small amount of audio data, our proposed approach can well handle cross-lingual tasks. And in intra-lingual tasks, our proposed approach also performs better than baseline approach in naturalness and similarity. | ['Xinyong Zhou', 'Xiaorui Wang', 'Lei Xie', 'Hao Che'] | 2019-10-29 | null | null | null | null | ['voice-cloning'] | ['speech'] | [-5.03575169e-02 3.09591386e-02 -3.12711209e-01 -5.27022183e-01
-1.30189621e+00 -5.37432909e-01 4.92478997e-01 -5.90862095e-01
-1.35290980e-01 4.65521812e-01 6.82450056e-01 -4.84955348e-02
4.81987417e-01 -3.89063954e-01 -6.19530678e-01 -3.64980727e-01
2.95607209e-01 2.13937223e-01 -9.51489434e-03 -4.26963449e-01
-2.00672075e-01 2.43475348e-01 -1.35416245e+00 4.88512874e-01
4.76629555e-01 7.40202069e-01 4.18332726e-01 8.82508934e-01
-3.13650489e-01 2.90421605e-01 -6.56013310e-01 -3.23045731e-01
3.51744175e-01 -5.81852734e-01 -7.33725011e-01 2.90344725e-03
3.39916617e-01 2.09297258e-02 -1.49006978e-01 1.00157249e+00
7.28205979e-01 1.50944456e-01 5.51242292e-01 -9.69744265e-01
-8.45811129e-01 1.34898818e+00 -1.33721411e-01 8.19816589e-02
4.81540501e-01 -2.49570280e-01 1.16956854e+00 -1.09566295e+00
2.77348995e-01 1.82040524e+00 7.31034219e-01 8.85823548e-01
-1.34396100e+00 -9.73840952e-01 -3.68229412e-02 -2.70541050e-02
-1.58221352e+00 -1.07282579e+00 1.04420686e+00 -2.73534000e-01
9.21070457e-01 4.16617930e-01 3.49868566e-01 1.41442788e+00
3.26055549e-02 7.53907084e-01 8.12583566e-01 -8.54593873e-01
-1.11841718e-02 5.17359138e-01 -2.44338155e-01 2.19569683e-01
-5.56139886e-01 4.14017379e-01 -1.05504155e+00 -3.06611687e-01
4.93556082e-01 -5.73455811e-01 -1.51350915e-01 1.58338234e-01
-1.18193483e+00 9.84683812e-01 -1.99258089e-01 6.46870315e-01
-1.16228290e-01 2.64336076e-02 6.27560973e-01 5.48717499e-01
3.50343674e-01 1.53916702e-01 -5.50916255e-01 -2.65652269e-01
-9.08721328e-01 -6.70144781e-02 5.96746206e-01 1.23129380e+00
3.63482893e-01 8.01854968e-01 -1.06194399e-01 1.57544947e+00
5.52634120e-01 7.19207287e-01 1.27310908e+00 -8.04531634e-01
3.27900618e-01 -1.85718164e-01 -1.62751287e-01 -5.11995435e-01
9.45338681e-02 -3.10881436e-01 -5.09894371e-01 -3.56972843e-01
-7.24596307e-02 -5.90243451e-02 -6.17893159e-01 1.85449076e+00
2.33358666e-01 2.34454408e-01 4.19129878e-01 4.81766760e-01
7.15041637e-01 1.01043761e+00 -1.37723938e-01 -5.64223886e-01
1.38165987e+00 -1.27500820e+00 -1.10188568e+00 -3.49685811e-02
3.75807106e-01 -1.31466556e+00 1.66832340e+00 3.74351799e-01
-1.05754137e+00 -1.09011817e+00 -8.77454162e-01 -4.26657163e-02
-1.56016722e-01 4.19841349e-01 2.63664983e-02 1.01084542e+00
-1.03148389e+00 2.00713813e-01 -5.07973850e-01 -3.55269313e-01
-3.25264722e-01 2.53186017e-01 -3.16368669e-01 2.86560804e-01
-1.47019100e+00 4.72927511e-01 2.17309415e-01 -3.51153240e-02
-1.00726354e+00 -6.60017669e-01 -9.52348709e-01 -9.00362283e-02
-1.10992849e-01 -2.39411592e-01 1.69442356e+00 -8.94213140e-01
-2.31296730e+00 5.64160347e-01 -4.81648475e-01 -4.40759331e-01
1.83257848e-01 -3.19340885e-01 -8.91369760e-01 -2.59777695e-01
1.04488686e-01 6.36059225e-01 1.14243889e+00 -1.18544686e+00
-7.26296186e-01 -4.91144098e-02 -6.55161798e-01 3.14573288e-01
-4.24961567e-01 3.85178536e-01 -2.14618504e-01 -8.61430168e-01
1.03933573e-01 -9.30956185e-01 2.56839663e-01 -6.64574623e-01
-5.26912570e-01 -3.30296516e-01 8.72962177e-01 -6.57214761e-01
1.23546672e+00 -2.32228160e+00 -1.01475649e-01 -1.02874711e-01
-5.70935905e-01 9.07955766e-02 -1.27707317e-01 4.80383724e-01
4.32885922e-02 1.82983279e-01 8.54831338e-02 -6.55990720e-01
1.81068316e-01 2.25107938e-01 -8.70107293e-01 1.76098347e-01
-1.33727744e-01 5.77435136e-01 -4.10090387e-01 -4.62963015e-01
-2.56271660e-02 6.61225855e-01 -5.04961252e-01 5.01259685e-01
1.49923131e-01 3.81888181e-01 -1.51967129e-03 3.57033461e-01
3.97300631e-01 8.40088248e-01 -7.04725906e-02 -1.88468203e-01
-1.56478733e-01 9.60964262e-01 -1.08727908e+00 1.76409352e+00
-9.24114883e-01 3.91276091e-01 1.36699021e-01 -6.65591061e-01
1.34554803e+00 9.05581772e-01 -4.38789045e-03 -2.42446169e-01
6.32867664e-02 2.68366367e-01 1.75246581e-01 -3.58602464e-01
4.11151350e-01 -6.79810822e-01 -3.33121270e-01 2.44153708e-01
3.72991443e-01 -5.10135651e-01 -3.09366196e-01 -4.20427084e-01
4.27546114e-01 -3.70330438e-02 4.47304279e-01 -3.23563486e-01
7.46298611e-01 -3.25731456e-01 6.95055008e-01 6.56978190e-01
-1.89590916e-01 5.65204501e-01 -1.53442055e-01 3.87593135e-02
-9.34107780e-01 -1.17901778e+00 -3.56049985e-01 1.68587446e+00
-3.01925451e-01 -2.94671834e-01 -8.21981490e-01 -3.52722406e-01
-8.74732435e-02 1.09392369e+00 -2.59067625e-01 -1.47515669e-01
-6.24640644e-01 -8.97993222e-02 1.21718872e+00 3.37595105e-01
4.78324257e-02 -1.15880656e+00 4.26216215e-01 4.87987638e-01
-2.59870589e-01 -1.26388705e+00 -1.17978048e+00 6.19723275e-02
-6.49082601e-01 -1.81464225e-01 -4.67231691e-01 -1.28157663e+00
-4.88804765e-02 9.09845307e-02 8.25839996e-01 -5.81521749e-01
-4.34493385e-02 2.10603148e-01 -3.83698523e-01 -3.29613388e-01
-1.16124237e+00 1.87643483e-01 6.66018128e-01 2.73138851e-01
4.31734502e-01 -6.33159518e-01 5.10660745e-02 4.99898702e-01
-5.49472749e-01 -2.24712491e-01 2.27265611e-01 1.02829146e+00
5.46062112e-01 -2.18541105e-03 8.67547154e-01 -6.62267029e-01
9.97063637e-01 -1.75356284e-01 -4.42349970e-01 1.33913875e-01
-3.05899560e-01 9.36257690e-02 9.83206451e-01 -7.35238612e-01
-1.25798368e+00 1.10149734e-01 -4.10078764e-01 -3.58781219e-01
-3.30150843e-01 2.26293504e-01 -6.10503376e-01 3.73878509e-01
5.86094856e-01 4.88332987e-01 -1.97786003e-01 -8.41667831e-01
5.99578679e-01 1.41313267e+00 6.63034737e-01 -8.02892029e-01
6.92015231e-01 -9.98020917e-02 -7.05326796e-01 -1.08822632e+00
-5.53140223e-01 -4.84091252e-01 -8.68322492e-01 5.10189235e-02
4.94224548e-01 -9.43451047e-01 -6.19417548e-01 2.94969290e-01
-1.29939115e+00 -1.88926002e-03 -3.03141654e-01 9.37339962e-01
-6.28764749e-01 1.50902912e-01 -7.76340842e-01 -9.31282103e-01
-3.89535367e-01 -1.24529707e+00 1.20326614e+00 -2.41588891e-01
-3.70316803e-01 -9.66055751e-01 3.64247531e-01 3.64260107e-01
4.94484097e-01 -7.15074778e-01 8.92871201e-01 -8.94956470e-01
4.11236361e-02 -2.99076717e-02 4.59310353e-01 7.80825377e-01
6.97882771e-01 1.05039530e-01 -1.50537169e+00 -1.96854159e-01
3.56492251e-01 -2.71315664e-01 4.07938391e-01 3.22817564e-01
6.79476559e-01 -4.56770480e-01 -6.82565942e-02 4.26122040e-01
8.26748669e-01 4.59575117e-01 2.40869790e-01 -3.17016661e-01
5.73058605e-01 6.41349137e-01 4.47882235e-01 2.19569042e-01
2.78616399e-01 9.45323288e-01 -3.26875687e-01 9.23135281e-02
-3.91772479e-01 -5.41994393e-01 9.88805175e-01 2.06342173e+00
4.97536957e-01 -4.41977754e-02 -7.71018565e-01 6.42856419e-01
-1.29108429e+00 -8.85687828e-01 2.50276566e-01 2.15511250e+00
1.37221920e+00 1.68311119e-01 1.63196281e-01 5.56082316e-02
9.74321306e-01 1.88109390e-02 -2.29570478e-01 -8.86345088e-01
-1.02120087e-01 1.79525629e-01 8.84866416e-02 1.21369171e+00
-8.59478295e-01 1.41165888e+00 7.20729876e+00 1.23726594e+00
-1.48158717e+00 4.19278145e-01 2.01999485e-01 1.43536162e-02
-4.60159242e-01 -7.90120214e-02 -1.29606926e+00 3.29207778e-01
1.29875636e+00 -4.91278380e-01 6.28345132e-01 9.33806181e-01
4.72479552e-01 5.40344715e-01 -1.34307563e+00 9.99771416e-01
2.58464187e-01 -8.48119497e-01 2.11662918e-01 -1.90982133e-01
4.07927901e-01 1.48142288e-02 2.08190963e-01 5.87390721e-01
3.25288743e-01 -1.10908890e+00 8.80794883e-01 1.99250821e-02
9.38764751e-01 -6.72040343e-01 3.57071251e-01 3.50467294e-01
-1.44273543e+00 2.11724237e-01 -5.43197989e-01 2.40125090e-01
2.83815712e-01 1.58303410e-01 -1.38072157e+00 2.41250500e-01
4.15406793e-01 3.50110263e-01 -4.85046320e-02 3.98602277e-01
5.14822491e-02 1.26490283e+00 -2.50830680e-01 1.43207476e-01
-2.01370548e-02 -6.28053397e-02 7.03115463e-01 1.48455179e+00
5.08905292e-01 -4.92258847e-01 2.82608628e-01 8.32573295e-01
-1.05500743e-01 7.26795435e-01 -7.57126570e-01 -1.23990506e-01
8.39229941e-01 8.66131127e-01 -3.11878156e-02 -2.16207892e-01
-2.43748590e-01 9.98058796e-01 1.41474202e-01 2.98687339e-01
-6.90100253e-01 -3.20227742e-01 7.22148657e-01 -1.74045816e-01
2.11045697e-01 -2.17616558e-01 -3.36208902e-02 -9.52598751e-01
-1.05378725e-01 -9.90482867e-01 -6.94197565e-02 -5.97805679e-01
-1.45960522e+00 1.12921524e+00 -2.66202271e-01 -1.20876706e+00
-7.54323244e-01 -3.66353959e-01 -5.16771674e-01 1.07558751e+00
-1.34360194e+00 -1.34199321e+00 4.17849690e-01 6.83869958e-01
1.07605875e+00 -7.33287692e-01 1.27499616e+00 2.50653207e-01
-2.98854947e-01 9.97417331e-01 2.50283509e-01 1.50512233e-01
9.23161209e-01 -9.20095682e-01 5.95877409e-01 5.07333338e-01
6.16693199e-01 7.04681277e-01 7.34663010e-01 -4.70408857e-01
-1.05336750e+00 -1.09600723e+00 1.15362298e+00 -1.43128112e-01
6.55615568e-01 -8.14836919e-01 -9.38599467e-01 8.29734027e-01
4.76904958e-01 -5.51011227e-02 1.19104707e+00 2.04043537e-01
-5.08207381e-01 -4.01522666e-01 -8.91537488e-01 5.75850308e-01
8.27785730e-01 -1.05920911e+00 -9.51123357e-01 1.11475714e-01
1.40718818e+00 -1.76736519e-01 -9.22487199e-01 1.12643577e-01
6.39171243e-01 -5.52977264e-01 7.86314428e-01 -6.27771974e-01
-1.37190133e-01 -9.75912958e-02 -6.83350503e-01 -1.46621418e+00
-2.22954378e-01 -1.05067325e+00 5.83058357e-01 1.80152667e+00
6.14691615e-01 -5.33914506e-01 2.11574852e-01 -3.03057358e-02
-3.33430529e-01 -8.72092471e-02 -1.22039831e+00 -1.16689897e+00
3.48944366e-01 -8.08172166e-01 6.93523884e-01 9.55516458e-01
1.47539884e-01 6.56902611e-01 -6.96871221e-01 2.47388959e-01
2.77394921e-01 -6.61271662e-02 7.19653070e-01 -8.87769103e-01
-6.29426062e-01 -2.23133951e-01 -1.24928869e-01 -1.20518124e+00
6.54158771e-01 -1.01812541e+00 3.36071044e-01 -5.23624659e-01
-2.05691591e-01 -4.16112065e-01 -4.03169990e-01 3.66519421e-01
9.36260372e-02 1.49429440e-01 1.53748810e-01 1.20846376e-01
1.58572778e-01 9.68092740e-01 8.55745912e-01 -1.00295998e-01
-6.14956558e-01 2.61849314e-01 -3.64075929e-01 6.40543759e-01
9.27877069e-01 -9.00157809e-01 -4.61386532e-01 -1.79993629e-01
-5.62535882e-01 3.11455786e-01 -3.40017319e-01 -7.76186287e-01
1.01383090e-01 -2.14148924e-01 -1.76513821e-01 -4.75004703e-01
7.01077998e-01 -6.29229963e-01 1.15504161e-01 1.16378479e-01
-7.38411188e-01 -1.31893188e-01 2.86014467e-01 2.54008353e-01
-7.24399805e-01 -3.43824446e-01 7.61884391e-01 1.25231564e-01
-2.98605338e-02 -1.18319942e-02 -6.20677948e-01 1.54263794e-01
4.85210299e-01 -1.12825863e-01 1.87738732e-01 -5.40189147e-01
-8.61716986e-01 -3.88245493e-01 -8.62169191e-02 8.82620633e-01
3.86429906e-01 -1.81280255e+00 -8.22318614e-01 7.34924316e-01
1.27449647e-01 -5.12772977e-01 -3.29584666e-02 4.11435097e-01
2.91150548e-02 6.12735748e-01 8.75508226e-03 -6.95642650e-01
-1.47962391e+00 3.99807960e-01 4.06961769e-01 3.21287394e-01
-3.33479851e-01 1.01337826e+00 4.17947590e-01 -9.94714379e-01
4.96986538e-01 -3.91468316e-01 5.65442480e-02 -1.57383621e-01
3.82160693e-01 4.85572629e-02 -5.47847822e-02 -1.28616011e+00
-2.78581232e-01 5.23891807e-01 -7.22833425e-02 -7.41997004e-01
1.02486885e+00 -5.90040922e-01 9.77238119e-02 1.19527233e+00
1.38691878e+00 8.83684576e-01 -7.88155735e-01 -4.54281837e-01
1.79311894e-02 -2.00523511e-01 -1.46307293e-02 -4.62834537e-01
-6.14354432e-01 1.12355661e+00 5.18858254e-01 1.94381833e-01
7.50227034e-01 4.74493504e-02 9.76736844e-01 1.65700510e-01
1.85759455e-01 -1.19960439e+00 -4.25739549e-02 5.77593148e-01
1.25700283e+00 -1.05270016e+00 -6.83481812e-01 -4.67193604e-01
-9.15733278e-01 1.14143014e+00 4.44978178e-01 -2.99941935e-03
9.22336817e-01 5.25166810e-01 6.41272366e-01 4.64669079e-01
-9.14120793e-01 9.03957337e-02 4.48595762e-01 6.05427504e-01
7.95213401e-01 3.82694811e-01 2.48937253e-02 9.83728290e-01
-1.13323796e+00 -3.78644228e-01 3.88399184e-01 2.30284736e-01
-5.56748569e-01 -1.44323754e+00 -6.86984241e-01 -3.78877550e-01
-6.39311612e-01 -3.91655535e-01 -3.68644059e-01 5.47292292e-01
-6.33475184e-02 1.13898194e+00 -4.59521487e-02 -5.92477739e-01
2.90984690e-01 6.87773347e-01 1.67797551e-01 -8.32341254e-01
-5.45402884e-01 8.05756629e-01 1.45970598e-01 -2.55369276e-01
-1.56937256e-01 -6.75366044e-01 -1.24697733e+00 8.02307203e-02
-5.44430077e-01 4.24161196e-01 7.79373765e-01 8.25533628e-01
1.71081826e-01 3.70051771e-01 1.08518815e+00 -5.23762405e-01
-9.28812623e-01 -1.37744224e+00 -9.38684821e-01 2.09992141e-01
3.99935216e-01 -2.68675208e-01 -5.25714219e-01 4.88597333e-01] | [14.827778816223145, 6.6684112548828125] |
05fb44fb-1ec9-43ce-ba9b-078850277bea | cave-correcting-attribute-values-in-e | null | null | https://dl.acm.org/doi/abs/10.1145/3511808.3557161 | https://dl.acm.org/doi/pdf/10.1145/3511808.3557161 | CAVE: Correcting Attribute Values in E-commerce Profiles | Attribute value extraction from product profiles is essential for many applications such as product retrieval, comparison, and recommendation. While existing techniques focus mainly on the extraction task, none of them deals with the problem of correcting wrong attribute values. In this paper we propose CAVE, a novel system for attribute correction and enrichment using the Question Answering (QA) paradigm. CAVE learns information from both titles and attribute tables, using encoder and language models to correct attribute values. It also has the capability to enrich existing product descriptions with new attribute values extracted from titles. To the best of our knowledge, CAVE is the first system that allows users to experiment with a number of powerful QA models and compare their performances on attribute values correction using real-word datasets. | ['Johann Gamper', 'Mouna Kacimi', 'Kassem Sabeh'] | 2022-10-17 | null | null | null | acm-international-conference-on-information-3 | ['attribute-value-extraction'] | ['natural-language-processing'] | [ 3.49282503e-01 1.87756971e-01 -4.74115878e-01 -7.61409104e-01
-8.51141751e-01 -6.38764679e-01 3.32489640e-01 9.80245709e-01
-4.88465607e-01 7.76280761e-01 2.98002988e-01 -1.81314975e-01
-3.18678260e-01 -1.05304062e+00 -5.81203640e-01 -4.24552374e-02
3.24623525e-01 1.12226319e+00 6.72526807e-02 -7.53159761e-01
3.19530845e-01 3.47282290e-01 -1.88303161e+00 7.11126328e-01
9.22682643e-01 1.08756924e+00 -2.28841394e-01 3.15916419e-01
-7.87246466e-01 9.50809538e-01 -4.93914664e-01 -1.23805177e+00
1.13686495e-01 -3.43719646e-02 -1.14745128e+00 -3.05818081e-01
5.25512934e-01 -1.35505021e-01 1.55295253e-01 1.17130995e+00
3.56429964e-01 8.33506957e-02 6.40803039e-01 -1.37490535e+00
-1.09931278e+00 1.16540563e+00 -4.18586470e-02 -1.59682319e-01
6.29656136e-01 -2.10209116e-01 1.39299321e+00 -9.55546319e-01
6.31808400e-01 9.44583118e-01 6.99009240e-01 5.85142314e-01
-1.27304554e+00 -4.62689430e-01 -3.61635178e-01 4.42297012e-01
-1.41662884e+00 -2.77544826e-01 3.69811177e-01 -1.33552030e-01
1.28432012e+00 4.36387390e-01 1.99273050e-01 6.64530098e-01
-4.16590035e-01 7.29160190e-01 8.25491071e-01 -6.33454978e-01
1.65142640e-01 8.22024405e-01 3.35465521e-01 2.69653529e-01
3.57109129e-01 -2.20015332e-01 -4.76081014e-01 -1.14961021e-01
1.02739096e-01 -3.67776841e-01 6.34703040e-02 -3.65372539e-01
-1.02837074e+00 9.27365243e-01 -2.59029213e-02 1.42528400e-01
-5.45119941e-01 3.73175703e-02 4.95893598e-01 4.28119332e-01
2.14618132e-01 1.29030490e+00 -1.28814030e+00 -1.02208816e-02
-5.26864350e-01 6.13957107e-01 1.08230960e+00 1.51373208e+00
8.37482333e-01 -4.62226331e-01 -3.88857931e-01 6.58662796e-01
3.25159878e-01 6.23192549e-01 3.29856306e-01 -9.89889503e-01
3.25456738e-01 8.72757673e-01 3.79880786e-01 -6.63922906e-01
-2.72752553e-01 -4.88054574e-01 -2.96948016e-01 -1.53484002e-01
5.48276901e-01 3.07204098e-01 -7.57947624e-01 1.51691735e+00
3.21415246e-01 -4.23254460e-01 3.57343376e-01 5.45564234e-01
1.21309698e+00 2.91244477e-01 6.53349996e-01 -8.52449760e-02
1.79317236e+00 -6.25200272e-01 -1.07887065e+00 -1.84578836e-01
1.17493320e+00 -1.07728684e+00 1.27497113e+00 4.23560649e-01
-9.92915094e-01 -4.31551129e-01 -6.81784511e-01 -4.66739595e-01
-1.11472237e+00 1.67133585e-02 8.91189575e-01 7.99914062e-01
-4.33971077e-01 7.26838648e-01 1.26767144e-01 -2.91578352e-01
4.61988837e-01 3.40042800e-01 -3.34720820e-01 -1.38807222e-01
-1.81674778e+00 1.23475802e+00 4.64761913e-01 -4.74213272e-01
-2.77779371e-01 -1.20578587e+00 -1.09705091e+00 3.02290857e-01
5.05935907e-01 -9.27685678e-01 1.73677802e+00 -8.44565690e-01
-9.88406241e-01 5.75783193e-01 -1.53061256e-01 -4.53655809e-01
-1.15746498e-01 -4.54728067e-01 -7.97124743e-01 -4.43331152e-01
3.15646499e-01 6.12778008e-01 4.29775357e-01 -8.79488885e-01
-7.98416436e-01 -4.66668963e-01 1.04543865e-01 1.88571036e-01
-3.71854037e-01 1.64730355e-01 -7.56058767e-02 -4.92069811e-01
-1.42099112e-01 -4.93342102e-01 -5.97640797e-02 -2.62918890e-01
-2.16318578e-01 -5.24266124e-01 1.34395495e-01 -5.70457041e-01
1.43052459e+00 -1.86002636e+00 -1.41177088e-01 4.83552694e-01
-2.83105932e-02 2.77171552e-01 -4.91537228e-02 5.21806777e-01
-1.76327467e-01 2.80132324e-01 -1.02057114e-01 -4.37947325e-02
3.75789315e-01 3.48487526e-01 -3.51431400e-01 -2.68150687e-01
1.53684452e-01 1.16840231e+00 -1.08204591e+00 -5.64203322e-01
-4.83132638e-02 4.21791613e-01 -4.93884087e-01 -5.24433963e-02
-5.06988585e-01 -1.29632980e-01 -3.20563346e-01 7.31594622e-01
6.63169801e-01 -4.77656499e-02 9.61022526e-02 -5.05713224e-01
2.70408958e-01 6.25341535e-01 -1.35186815e+00 1.51090753e+00
-4.26474065e-01 2.22373024e-01 -6.87557817e-01 -3.47356915e-01
8.54327381e-01 2.34761328e-01 4.08442527e-01 -1.01307380e+00
1.90188974e-01 5.00512242e-01 -2.79793948e-01 -6.09595597e-01
1.11938107e+00 -2.16109887e-01 -2.96652764e-01 1.82585359e-01
1.95728675e-01 -3.54818255e-01 5.35814285e-01 3.94494474e-01
7.74702311e-01 1.60991728e-01 4.48404700e-01 9.25172213e-03
9.31368887e-01 3.70263070e-01 2.20650136e-01 5.96542895e-01
2.67608434e-01 8.11123252e-02 2.51291573e-01 -3.38902295e-01
-1.37582028e+00 -9.60458219e-01 -3.86380762e-01 1.33100414e+00
-3.48724425e-02 -1.02673447e+00 -6.44646347e-01 -8.21450949e-01
4.80957866e-01 1.46952271e+00 -6.15993142e-01 -3.31496805e-01
-1.79086208e-01 -3.36603820e-01 4.58019525e-01 7.16083229e-01
2.72705078e-01 -1.07598662e+00 -1.13531671e-01 3.89658064e-01
-3.62022281e-01 -9.86476541e-01 -2.72104472e-01 1.57738000e-01
-8.09245110e-01 -1.06278992e+00 5.09993732e-03 -6.65441751e-01
3.96255136e-01 -2.71948576e-01 1.75011945e+00 3.86953428e-02
8.13967139e-02 3.14588904e-01 -5.67007065e-01 -8.30520213e-01
-5.86089313e-01 4.05418754e-01 -1.71776652e-01 -1.98875487e-01
1.42241120e+00 -3.93330641e-02 -1.60443634e-01 2.58398235e-01
-8.45997214e-01 -2.57984549e-01 7.30132341e-01 7.15186477e-01
8.31907630e-01 2.26415977e-01 6.15802467e-01 -1.58244884e+00
7.65433311e-01 -4.06908005e-01 -4.03806567e-01 4.50543076e-01
-1.37241280e+00 5.49254358e-01 4.56524581e-01 -6.04338795e-02
-1.19620466e+00 3.40930820e-01 -6.48551643e-01 3.73239815e-01
-3.80436033e-01 6.22028708e-01 -4.60540235e-01 2.03395039e-01
9.24810767e-01 -2.51731873e-02 -2.51890332e-01 -8.57117474e-01
8.57946634e-01 7.32220352e-01 5.82681537e-01 -3.13536227e-01
5.26710749e-01 6.95002377e-02 8.11904445e-02 -2.07214653e-01
-1.25529361e+00 -8.07998657e-01 -7.94807613e-01 1.92364648e-01
5.53275883e-01 -5.95909238e-01 -9.16809082e-01 7.64262676e-02
-9.04400766e-01 2.78903216e-01 -9.36932504e-01 3.97017807e-01
-5.92319608e-01 2.97704607e-01 -4.56062198e-01 -4.94813591e-01
-4.64821190e-01 -5.57889760e-01 8.23334932e-01 1.01649612e-01
-5.72541595e-01 -6.81892693e-01 -6.97733983e-02 4.29129660e-01
4.93793637e-01 -2.54423946e-01 1.18297732e+00 -9.95713949e-01
-2.06537783e-01 -4.58370537e-01 -4.16296870e-02 3.57570291e-01
-4.79310900e-02 -4.46294278e-01 -9.12629843e-01 2.29548782e-01
-4.60284829e-01 -2.48876706e-01 7.40619600e-01 -9.70897675e-02
1.12963068e+00 -4.05351520e-01 -1.72424659e-01 2.59115815e-01
1.26662362e+00 3.18442918e-02 1.16961861e+00 5.86063147e-01
4.81295168e-01 8.93966675e-01 1.20863056e+00 3.99327904e-01
4.85314995e-01 7.39817083e-01 4.84221041e-01 1.25962183e-01
-1.02584608e-01 -4.05803621e-01 -1.96569204e-01 1.07255779e-01
2.17554152e-01 -1.42770424e-01 -7.07273304e-01 6.76033497e-01
-1.63104200e+00 -9.62650895e-01 -7.22503066e-01 2.23553514e+00
1.19097745e+00 9.08402652e-02 -9.97223631e-02 2.23966941e-01
3.19462031e-01 -7.30471671e-01 -2.85347939e-01 -6.22677624e-01
-1.91097841e-01 4.36133295e-01 7.05938160e-01 3.71258736e-01
-1.14844275e+00 1.00187719e+00 6.03266430e+00 6.63262486e-01
-3.02863847e-02 1.89846411e-01 1.86872885e-01 2.37679467e-01
-7.59338140e-01 3.34794484e-02 -1.07098842e+00 1.64874852e-01
1.05556476e+00 -1.86092779e-01 1.95454746e-01 1.01048374e+00
-4.73227203e-01 -1.06837414e-02 -1.14662731e+00 8.41729164e-01
6.53051659e-02 -1.02614427e+00 3.48642886e-01 -1.86605081e-01
3.64514709e-01 -5.88094831e-01 1.51485965e-01 6.51982069e-01
3.28570306e-01 -1.10215032e+00 6.18855715e-01 7.49487162e-01
7.30465710e-01 -1.09056723e+00 1.40351915e+00 5.01243137e-02
-8.67906570e-01 -1.45425573e-01 -6.01759672e-01 1.16175406e-01
1.13204382e-01 5.43634236e-01 -1.03072155e+00 4.43649620e-01
8.10426712e-01 4.63612497e-01 -9.69419837e-01 9.69771504e-01
-4.29879189e-01 2.59088039e-01 6.25469163e-02 -1.53774798e-01
-1.98334336e-01 1.49861977e-01 8.90062004e-02 1.17847991e+00
4.03908104e-01 1.37738496e-01 -4.73590583e-01 6.43850744e-01
-2.34761804e-01 5.98913193e-01 -4.48199242e-01 -1.91069037e-01
5.63715994e-01 1.19446909e+00 -4.23704833e-02 -6.66627526e-01
-6.79589272e-01 8.36369753e-01 1.41112760e-01 -9.38618928e-02
-6.49875045e-01 -9.62937534e-01 9.06563878e-01 3.13417643e-01
3.23582649e-01 1.43451855e-01 -6.26573741e-01 -9.57784057e-01
-1.31727427e-01 -1.00868034e+00 7.74857402e-01 -9.37800884e-01
-1.47133708e+00 4.96229976e-01 5.26354164e-02 -1.03034544e+00
-4.54276264e-01 -6.67315543e-01 4.12667811e-01 9.32777524e-01
-1.70255494e+00 -1.11019325e+00 -3.44910324e-01 6.63780332e-01
9.26376656e-02 -1.00865640e-01 1.12857962e+00 6.46309197e-01
1.46706045e-01 7.71107793e-01 4.34345342e-02 1.22667223e-01
1.05760992e+00 -1.60663247e+00 3.92860472e-01 2.32107028e-01
2.38567591e-01 8.38494182e-01 1.05595100e+00 -7.59245336e-01
-1.06397045e+00 -1.04002178e+00 1.91275024e+00 -1.09895587e+00
4.48253989e-01 -2.13823065e-01 -1.25724125e+00 6.08236074e-01
2.89433569e-01 -3.09232086e-01 1.12293756e+00 3.00245702e-01
-8.37737620e-01 -2.89255589e-01 -1.28079605e+00 2.08126128e-01
7.11764812e-01 -4.90496904e-01 -8.53242040e-01 4.22190577e-01
8.15184832e-01 -2.97208965e-01 -1.32855570e+00 3.80707115e-01
3.77798736e-01 -3.50231230e-01 1.18998456e+00 -1.18442750e+00
3.04259807e-01 -5.02933145e-01 -2.24101648e-01 -1.35358465e+00
-5.38230360e-01 2.26841196e-02 -2.90756106e-01 1.73012829e+00
1.07941902e+00 -2.80401558e-01 7.19702184e-01 1.06882572e+00
-1.39952242e-01 -3.79166484e-01 -3.75643492e-01 -6.28028631e-01
-8.25100765e-02 -5.24883211e-01 1.48659778e+00 9.55128610e-01
1.42240701e-02 6.58398449e-01 -2.73866028e-01 9.26420838e-02
4.93509591e-01 9.42359120e-02 5.53713620e-01 -1.48045290e+00
-2.40367264e-01 -1.36215225e-01 -2.69246638e-01 -5.15088677e-01
-5.75676374e-02 -1.06268847e+00 -1.37469128e-01 -1.65054536e+00
1.06095284e-01 -5.60018361e-01 -3.23933959e-01 7.61932135e-01
-2.42997184e-02 4.22083825e-01 -7.30331242e-02 -4.15567942e-02
-5.74008703e-01 2.71753371e-01 8.56385648e-01 -2.03805998e-01
-4.87760827e-02 8.03433061e-02 -1.06740916e+00 3.67201358e-01
1.28759015e+00 -8.40167403e-01 -4.10881877e-01 -3.60483080e-01
1.25745046e+00 -4.37311411e-01 1.93767369e-01 -6.46408737e-01
1.90052226e-01 -1.74849555e-01 4.94263858e-01 -5.00795662e-01
-2.99802572e-02 -9.80290115e-01 9.39742476e-02 1.38156310e-01
-5.91001809e-01 3.17472219e-01 2.51024872e-01 2.53357768e-01
-3.13490301e-01 -8.36147904e-01 5.11299908e-01 -2.71434456e-01
-1.21544123e+00 -5.16732736e-03 -2.60776520e-01 1.43610254e-01
7.61071742e-01 1.35242790e-01 -5.03310621e-01 -2.55997896e-01
-8.45434546e-01 2.60541022e-01 3.07834744e-01 5.02057612e-01
6.66486323e-01 -1.60945785e+00 -7.56241798e-01 2.34184474e-01
8.02716196e-01 -4.59534436e-01 -6.68684915e-02 5.72852671e-01
-3.32118005e-01 6.00188255e-01 -1.87744126e-01 4.33632061e-02
-1.25357962e+00 1.13462520e+00 2.23415315e-01 -2.73487628e-01
-1.35904893e-01 6.84535384e-01 -3.97738129e-01 -6.55157030e-01
9.72086266e-02 7.59400129e-02 -7.97328591e-01 4.05694842e-01
7.42139161e-01 2.37530440e-01 6.36316299e-01 -5.28280377e-01
-2.75803655e-01 2.47514546e-01 -3.33488971e-01 1.00298904e-01
1.12568212e+00 -2.87509024e-01 -2.62619287e-01 -2.08732467e-02
9.34539497e-01 5.37610240e-02 -3.95004898e-01 -5.98636270e-01
5.88653743e-01 -8.43525052e-01 1.07652545e-01 -1.39401639e+00
-7.40308046e-01 7.14619875e-01 5.56638539e-01 2.58872639e-02
1.26012325e+00 1.47228479e-01 9.26902473e-01 6.98938787e-01
3.88871521e-01 -1.28723431e+00 -1.18987739e-01 5.33330321e-01
6.59082949e-01 -1.09259331e+00 -1.04626790e-01 -6.89637065e-01
-1.00941861e+00 1.01219666e+00 6.31011546e-01 6.23551071e-01
4.64999825e-01 8.43188688e-02 7.00695589e-02 -3.52873564e-01
-5.69980085e-01 -7.85077572e-01 5.45085728e-01 8.81049395e-01
8.29980969e-01 -1.55909628e-01 -4.50659662e-01 8.82102549e-01
-2.54809797e-01 1.87174529e-01 6.11929655e-01 7.13000357e-01
-4.42594230e-01 -1.56490052e+00 -1.12887751e-02 7.30541587e-01
-5.14715195e-01 -6.53019309e-01 -5.44736981e-01 7.00622559e-01
4.36245978e-01 8.89645517e-01 -9.14296806e-02 -3.87974650e-01
8.06471944e-01 4.13703084e-01 4.74036902e-01 -7.47535408e-01
-9.75933552e-01 -5.41040003e-01 5.62146246e-01 -4.78736132e-01
-2.38120139e-01 -8.30788493e-01 -1.29428053e+00 -5.64101875e-01
-3.68614465e-01 5.61239958e-01 6.50520086e-01 4.42266375e-01
4.45646256e-01 3.12102944e-01 1.69131279e-01 3.87209445e-01
-7.08235443e-01 -9.74554420e-01 -7.46878624e-01 9.51083362e-01
-1.96173955e-02 -5.90106428e-01 1.36397798e-02 -1.44181056e-02] | [9.976996421813965, 6.310086250305176] |
3fefa0ac-d2f1-4eba-9a19-0c8ab66ff052 | zits-image-inpainting-by-improving-the | 2210.05950 | null | https://arxiv.org/abs/2210.05950v3 | https://arxiv.org/pdf/2210.05950v3.pdf | ZITS++: Image Inpainting by Improving the Incremental Transformer on Structural Priors | Image inpainting involves filling missing areas of a corrupted image. Despite impressive results have been achieved recently, restoring images with both vivid textures and reasonable structures remains a significant challenge. Previous methods have primarily addressed regular textures while disregarding holistic structures due to the limited receptive fields of Convolutional Neural Networks (CNNs). To this end, we study learning a Zero-initialized residual addition based Incremental Transformer on Structural priors (ZITS++), an improved model upon our conference work, ZITS. Specifically, given one corrupt image, we present the Transformer Structure Restorer (TSR) module to restore holistic structural priors at low image resolution, which are further upsampled by Simple Structure Upsampler (SSU) module to higher image resolution. To recover image texture details, we use the Fourier CNN Texture Restoration (FTR) module, which is strengthened by Fourier and large-kernel attention convolutions. Furthermore, to enhance the FTR, the upsampled structural priors from TSR are further processed by Structure Feature Encoder (SFE) and optimized with the Zero-initialized Residual Addition (ZeroRA) incrementally. Besides, a new masking positional encoding is proposed to encode the large irregular masks. Compared with ZITS, ZITS++ improves the FTR's stability and inpainting ability with several techniques. More importantly, we comprehensively explore the effects of various image priors for inpainting and investigate how to utilize them to address high-resolution image inpainting with extensive experiments. This investigation is orthogonal to most inpainting approaches and can thus significantly benefit the community. Codes and models will be released in https://github.com/ewrfcas/ZITS-PlusPlus. | ['Yanwei Fu', 'Qiaole Dong', 'Chenjie Cao'] | 2022-10-12 | null | null | null | null | ['image-inpainting'] | ['computer-vision'] | [ 9.17676032e-01 1.13715284e-01 7.71450922e-02 -7.31182992e-02
-7.44769931e-01 -1.10475458e-01 3.71839613e-01 -4.68474776e-01
-4.99808267e-02 8.09430361e-01 3.59945834e-01 -5.84256873e-02
-6.34441292e-03 -9.61885810e-01 -1.14422464e+00 -7.83591270e-01
2.64823020e-01 -2.70500124e-01 -6.68955371e-02 -4.65619534e-01
3.13247889e-01 5.48583746e-01 -1.61612713e+00 6.76317275e-01
1.05038965e+00 9.85126853e-01 5.88132620e-01 4.90083516e-01
-1.95681360e-02 9.87645149e-01 -4.53223974e-01 -1.38820022e-01
3.10586542e-01 -4.64920819e-01 -6.96114182e-01 1.74637944e-01
5.85086286e-01 -6.73873305e-01 -6.57561660e-01 9.54128563e-01
3.77738446e-01 1.67003170e-01 2.05173001e-01 -7.21442223e-01
-1.16855681e+00 4.29126859e-01 -9.19200540e-01 3.75872552e-02
2.94391721e-01 1.83549762e-01 5.40555179e-01 -1.01367176e+00
6.40667439e-01 1.33491254e+00 7.09692597e-01 5.52951396e-01
-1.43806207e+00 -5.41914582e-01 -3.13818268e-02 2.44929329e-01
-1.24270034e+00 -5.24199367e-01 1.17551255e+00 8.91049877e-02
8.78574252e-01 4.38809156e-01 4.36005890e-01 9.63299274e-01
3.24565172e-01 6.12302899e-01 1.27604806e+00 -3.50722551e-01
-5.85876666e-02 -4.35165972e-01 -3.63969237e-01 5.88523567e-01
-1.08007966e-02 2.81665862e-01 -6.21516347e-01 2.34557018e-01
1.30238533e+00 2.47714505e-01 -7.01596200e-01 1.45413324e-01
-1.05107963e+00 4.77315575e-01 5.13854742e-01 1.93754897e-01
-5.15263438e-01 3.11424166e-01 1.26377061e-01 4.11271125e-01
7.29828119e-01 4.60980028e-01 -2.87321001e-01 2.19730347e-01
-1.09631455e+00 2.14310780e-01 2.06477612e-01 6.79222405e-01
1.09074938e+00 3.41603428e-01 -3.64594549e-01 1.15469098e+00
-2.02145591e-01 3.12753439e-01 2.78562248e-01 -1.31555569e+00
3.74086887e-01 3.08052391e-01 6.19645230e-02 -1.03738952e+00
9.33214054e-02 -4.32995945e-01 -1.26298523e+00 3.80994201e-01
-4.72938828e-03 2.00643063e-01 -9.45406795e-01 1.52635670e+00
2.11707786e-01 3.37540507e-01 -4.91799302e-02 1.03679144e+00
9.05020416e-01 8.82636726e-01 -2.33646080e-01 -1.69236422e-01
1.33880031e+00 -1.17107308e+00 -9.29025769e-01 -2.65898287e-01
2.28352398e-01 -1.08861184e+00 1.28100312e+00 4.03732747e-01
-1.35084581e+00 -7.65742838e-01 -1.09581685e+00 -6.63206458e-01
5.75135760e-02 3.03692728e-01 4.95381981e-01 2.55441040e-01
-1.28497696e+00 1.08035171e+00 -8.65698218e-01 -5.89142926e-02
7.88440228e-01 1.02033146e-01 -4.63211447e-01 -6.22964025e-01
-1.07111907e+00 7.63212919e-01 1.04174912e-02 2.57725716e-01
-8.36302161e-01 -8.97173226e-01 -1.12695014e+00 6.31887987e-02
3.42085272e-01 -7.10883439e-01 9.01349127e-01 -1.05131173e+00
-1.67362630e+00 6.66703343e-01 -3.82989973e-01 -4.47834611e-01
3.37182552e-01 -2.08563462e-01 -2.53578842e-01 4.03366387e-01
6.67115375e-02 6.67795599e-01 1.27273393e+00 -1.46143878e+00
-8.42809454e-02 1.27263526e-02 -8.86220559e-02 2.67841071e-01
-1.32858664e-01 -1.26992315e-01 -3.18047881e-01 -1.28778732e+00
2.17345700e-01 -3.16091657e-01 -1.51259288e-01 1.50181741e-01
-3.47502619e-01 5.02863944e-01 1.11819315e+00 -1.24468756e+00
1.13795125e+00 -2.28044176e+00 2.11207271e-01 -2.05338687e-01
2.96035677e-01 2.09308878e-01 -5.42297363e-01 4.19751465e-01
-4.09127295e-01 -9.07386933e-03 -7.63160229e-01 -6.65028274e-01
-3.64568144e-01 4.34887469e-01 -6.32057965e-01 3.56285214e-01
6.46371365e-01 9.84544277e-01 -4.85884070e-01 -9.21725705e-02
4.86868978e-01 9.74567354e-01 -6.85776651e-01 2.36639932e-01
-9.65732783e-02 6.41396344e-01 -1.01050466e-01 8.23449850e-01
1.18415785e+00 -1.20163448e-01 -6.80437684e-02 -6.72918677e-01
-2.37646475e-01 1.15783498e-01 -9.26475823e-01 1.65544391e+00
-6.13056719e-01 5.13317049e-01 4.29257154e-01 -1.06128085e+00
1.03351176e+00 7.39734322e-02 3.81327778e-01 -1.09630537e+00
-3.71168926e-02 1.55634865e-01 -3.26911330e-01 -3.98314744e-01
7.51300573e-01 -1.98486760e-01 4.71428543e-01 3.06077510e-01
-2.40790099e-02 -1.17871620e-01 -5.39259985e-02 9.22547877e-02
1.10583317e+00 5.29901981e-01 -1.77296139e-02 -7.31864944e-02
4.01942790e-01 -3.73241186e-01 4.24025148e-01 5.83026409e-01
1.36269465e-01 1.27336574e+00 2.87943184e-01 -6.02190197e-01
-1.26180851e+00 -9.32539642e-01 -1.15737751e-01 7.18745351e-01
2.56396532e-01 -3.02526355e-01 -8.86047721e-01 -1.32445171e-01
-2.87709594e-01 5.28085947e-01 -8.41490686e-01 -1.91555530e-01
-8.82237613e-01 -6.99686825e-01 2.47831047e-01 3.31867218e-01
9.67367709e-01 -1.29508674e+00 -4.66665864e-01 2.58160263e-01
-5.57648242e-01 -9.51868415e-01 -7.58868098e-01 6.14769571e-02
-8.43054593e-01 -9.38844562e-01 -9.06767249e-01 -7.15460420e-01
7.39584088e-01 5.43348730e-01 9.46394086e-01 4.26561296e-01
-4.17165041e-01 8.15942511e-02 -4.56251621e-01 2.15240434e-01
-2.91834265e-01 -2.49480054e-01 -3.72732013e-01 2.67922521e-01
-4.32140529e-01 -9.53292012e-01 -8.09138477e-01 8.54104757e-02
-1.42978930e+00 5.60819566e-01 8.69217575e-01 9.93856907e-01
8.72574568e-01 2.32945770e-01 2.72477359e-01 -8.40657234e-01
4.57602859e-01 -2.21265644e-01 -2.34126061e-01 4.63576801e-02
-3.55449349e-01 3.51848602e-02 7.96541989e-01 -4.83300328e-01
-1.40354633e+00 -2.41258964e-01 -3.85005563e-01 -6.96207821e-01
-1.30686074e-01 2.02293813e-01 -3.11199546e-01 -3.40473771e-01
4.47612762e-01 5.59551418e-01 1.95156112e-01 -6.42381370e-01
2.73635387e-01 2.28652224e-01 7.37864494e-01 -7.36414909e-01
8.29747438e-01 7.20978498e-01 -1.41560137e-01 -7.62623489e-01
-7.05224335e-01 2.29419023e-01 -2.38596782e-01 -4.21648063e-02
5.74602544e-01 -9.02807057e-01 -4.51873958e-01 7.31508970e-01
-1.08455420e+00 -6.72192633e-01 -5.32580733e-01 1.00901313e-01
-5.88773847e-01 7.04503179e-01 -1.02562010e+00 -5.09006679e-01
-5.39529264e-01 -1.17560959e+00 1.24498463e+00 1.96686208e-01
9.13094804e-02 -6.47684097e-01 -4.11007077e-01 3.92588496e-01
7.98303962e-01 3.90623778e-01 8.88359249e-01 3.93427849e-01
-8.58987510e-01 3.95113558e-01 -4.85108554e-01 6.05992019e-01
3.37353647e-01 -2.38027573e-01 -9.44825709e-01 -3.61897290e-01
2.61986315e-01 -1.84799194e-01 1.34487510e+00 6.70417428e-01
1.51003850e+00 -4.36537355e-01 -2.70847566e-02 1.10288429e+00
1.51443994e+00 9.87593159e-02 1.33166897e+00 4.79968369e-01
7.65625596e-01 4.79344189e-01 4.21578199e-01 3.55898410e-01
1.27181351e-01 4.51705366e-01 4.75366384e-01 -6.15675986e-01
-7.86431909e-01 -2.69907445e-01 3.47193599e-01 5.99822402e-01
-2.93473899e-01 5.35306428e-03 -3.12086493e-01 3.76748592e-01
-1.53387594e+00 -9.79168713e-01 3.13775018e-02 1.96521008e+00
1.08801508e+00 -1.91023484e-01 -5.45067966e-01 1.38565436e-01
7.65466690e-01 4.81445789e-01 -5.40769994e-01 -2.99685866e-01
-5.89354217e-01 8.36131513e-01 3.93755615e-01 8.46537292e-01
-8.89973819e-01 1.01335049e+00 5.27562284e+00 1.28023112e+00
-1.15742183e+00 7.49635547e-02 1.09775019e+00 7.14905793e-03
-4.99801248e-01 2.15215087e-01 -3.45971853e-01 4.36835706e-01
3.82569581e-01 3.20245981e-01 1.00393987e+00 3.16770434e-01
3.66075367e-01 -2.64324427e-01 -5.95360637e-01 9.28839386e-01
3.98144126e-02 -1.62450516e+00 3.33160877e-01 -9.04368237e-02
7.97621250e-01 -4.21100646e-01 3.36872011e-01 1.66759655e-01
-5.85066825e-02 -1.20598054e+00 7.06226647e-01 7.13192284e-01
1.23729384e+00 -8.59085202e-01 4.48427230e-01 -5.88851906e-02
-1.13555610e+00 -4.75064404e-02 -4.96708423e-01 -9.79089960e-02
2.30122641e-01 8.97815287e-01 -7.43999630e-02 8.26640606e-01
8.85150850e-01 9.46539938e-01 -3.56043994e-01 5.82564831e-01
-2.67942578e-01 4.31701213e-01 -1.22820780e-01 8.28111172e-01
-1.39241710e-01 -3.19149524e-01 3.84932220e-01 8.28054130e-01
5.38438916e-01 2.66843498e-01 -1.45743445e-01 1.12272203e+00
-2.69178003e-01 -1.96904734e-01 -3.14327478e-01 2.21803471e-01
3.11196446e-01 1.33046341e+00 -6.38861656e-01 -3.30743879e-01
-2.29234427e-01 1.35113215e+00 2.39342362e-01 6.28926635e-01
-7.52687812e-01 -3.92914504e-01 7.12630749e-01 4.50511336e-01
3.87327373e-01 -3.45132090e-02 -4.66039807e-01 -1.31590569e+00
1.66621149e-01 -1.00210750e+00 -1.52455464e-01 -1.07553589e+00
-1.15831304e+00 6.83535993e-01 -3.00089926e-01 -1.25353456e+00
4.75005060e-01 -2.86636382e-01 -5.71557045e-01 1.06258821e+00
-1.98650515e+00 -1.37303805e+00 -3.79949540e-01 6.44662976e-01
6.11925244e-01 2.29925230e-01 6.32455587e-01 3.93127263e-01
-5.01792669e-01 4.44538265e-01 1.46635072e-02 -8.85022655e-02
8.59380603e-01 -7.05983639e-01 4.18298304e-01 9.99439120e-01
-3.95616919e-01 6.36797190e-01 6.21217191e-01 -8.62159073e-01
-1.44011843e+00 -1.38881624e+00 5.08451641e-01 9.22977254e-02
2.83500105e-01 -1.18371256e-01 -1.26189542e+00 5.67171454e-01
4.68540341e-01 1.27534792e-01 1.61643654e-01 -5.62537193e-01
-4.91894960e-01 -1.50604397e-01 -1.26010728e+00 7.80267894e-01
1.07521749e+00 -4.60613132e-01 -2.21015513e-01 2.05756035e-02
9.10366237e-01 -6.44827366e-01 -7.89988339e-01 5.67781925e-01
3.53745461e-01 -1.17873406e+00 1.18912196e+00 1.74215212e-01
1.09162438e+00 -5.28610766e-01 -2.73829401e-01 -1.19117868e+00
-3.97191286e-01 -7.42154837e-01 -1.62673295e-01 1.14293790e+00
-1.08034931e-01 -6.58062279e-01 6.08807385e-01 2.00152516e-01
-4.70881552e-01 -8.48663270e-01 -8.24802637e-01 -4.19851154e-01
-8.34603757e-02 -2.23788232e-01 5.79376400e-01 1.02340019e+00
-6.21500254e-01 -7.13510886e-02 -8.67377698e-01 1.76846445e-01
6.87173367e-01 2.36972496e-01 4.22813952e-01 -5.45880616e-01
-4.61180210e-01 -2.70969540e-01 1.57508209e-01 -1.06205392e+00
-1.27943695e-01 -6.97798371e-01 -1.81703586e-02 -1.52388251e+00
1.84603885e-01 -1.91600338e-01 -1.31483041e-02 6.65449858e-01
-2.19248876e-01 7.72090375e-01 1.21486694e-01 2.24481061e-01
-4.09747809e-02 9.15593743e-01 1.79961348e+00 -1.26259953e-01
-8.03157464e-02 -4.30120140e-01 -9.11609530e-01 4.31057692e-01
7.80095160e-01 -2.45105967e-01 -2.55541533e-01 -6.21500731e-01
-4.83206511e-02 2.33614311e-01 6.99166954e-01 -9.16325033e-01
-1.06250167e-01 -1.09125271e-01 7.36910701e-01 -4.64501143e-01
5.25590718e-01 -4.37473834e-01 3.52049172e-01 3.24538708e-01
-1.94667354e-01 -2.02176273e-01 3.62937003e-01 3.61357987e-01
-4.50258374e-01 8.86236653e-02 1.05068743e+00 -2.44494855e-01
-4.84343022e-01 4.35250074e-01 -2.92748719e-01 -3.49245250e-01
5.33797801e-01 -4.10424501e-01 -3.70937258e-01 -3.88003767e-01
-6.35558903e-01 -2.69212306e-01 7.13265538e-01 1.49843127e-01
1.04946685e+00 -1.31189275e+00 -6.57873929e-01 4.98734593e-01
-3.38281661e-01 2.35892922e-01 1.00566316e+00 8.02656293e-01
-7.38760710e-01 -1.60226896e-01 -4.62130368e-01 -1.98565036e-01
-9.68888819e-01 5.32006204e-01 2.55008757e-01 -2.75880426e-01
-9.41008210e-01 7.24128008e-01 5.48594296e-01 -2.83002585e-01
-9.02569667e-02 -3.27838808e-01 -2.52512861e-02 -3.59379590e-01
7.94205010e-01 1.54801294e-01 1.17173724e-01 -4.90396678e-01
1.54083237e-01 6.15266383e-01 -2.53700614e-01 1.36490554e-01
1.64807510e+00 -3.28905642e-01 -5.18945098e-01 -2.83715099e-01
1.10769224e+00 9.19707939e-02 -1.79376209e+00 -2.13125467e-01
-5.40133297e-01 -6.79219544e-01 1.01851985e-01 -6.20433927e-01
-1.43711889e+00 8.44477475e-01 5.60117841e-01 -2.90424019e-01
1.69082808e+00 -3.43453526e-01 1.13943017e+00 -1.47848859e-01
1.16898522e-01 -7.24423110e-01 2.95846790e-01 3.73776436e-01
1.27999961e+00 -9.00829136e-01 -7.32504390e-03 -6.27296746e-01
-3.27419996e-01 1.17882180e+00 5.82386076e-01 -5.05585611e-01
4.15543497e-01 6.49841845e-01 -1.51020303e-01 7.55218938e-02
-5.51227927e-01 -2.05757134e-02 4.04083654e-02 7.09397435e-01
3.84765506e-01 -1.89058855e-01 -2.53177881e-01 5.76333463e-01
-6.23728335e-02 4.84611094e-02 6.19621336e-01 8.18402410e-01
-3.14804584e-01 -1.18884492e+00 -6.45028234e-01 2.75890201e-01
-4.78257567e-01 -5.65485358e-01 5.81521317e-02 4.36092556e-01
2.35683307e-01 7.99623311e-01 -7.88466483e-02 -3.49031091e-01
1.81363702e-01 -3.37527484e-01 6.27511024e-01 -4.54061449e-01
-3.43921334e-01 3.62428844e-01 -2.17703953e-01 -8.10443997e-01
-3.65608960e-01 -2.98364908e-01 -1.00454259e+00 -5.32247424e-01
7.65645355e-02 -3.88621211e-01 2.93121845e-01 7.34006345e-01
5.62531471e-01 8.90549123e-01 4.33642566e-01 -1.29035783e+00
-3.61631438e-02 -9.24456894e-01 -6.16474867e-01 3.22088838e-01
4.69980508e-01 -4.74748015e-01 -2.65414715e-01 2.84746885e-01] | [11.229966163635254, -1.562385082244873] |
ea63d4af-e68c-4f88-ade6-06f84dfdfe64 | skip-attention-improving-vision-transformers | 2301.02240 | null | https://arxiv.org/abs/2301.02240v2 | https://arxiv.org/pdf/2301.02240v2.pdf | Skip-Attention: Improving Vision Transformers by Paying Less Attention | This work aims to improve the efficiency of vision transformers (ViT). While ViTs use computationally expensive self-attention operations in every layer, we identify that these operations are highly correlated across layers -- a key redundancy that causes unnecessary computations. Based on this observation, we propose SkipAt, a method to reuse self-attention computation from preceding layers to approximate attention at one or more subsequent layers. To ensure that reusing self-attention blocks across layers does not degrade the performance, we introduce a simple parametric function, which outperforms the baseline transformer's performance while running computationally faster. We show the effectiveness of our method in image classification and self-supervised learning on ImageNet-1K, semantic segmentation on ADE20K, image denoising on SIDD, and video denoising on DAVIS. We achieve improved throughput at the same-or-higher accuracy levels in all these tasks. | ['Amirhossein Habibian', 'Fatih Porikli', 'Yuki M. Asano', 'Amir Ghodrati', 'Shashanka Venkataramanan'] | 2023-01-05 | null | null | null | null | ['video-denoising'] | ['computer-vision'] | [ 9.61241424e-02 6.95897415e-02 1.63510829e-01 -4.00422692e-01
-8.07604134e-01 -2.63487309e-01 3.74357373e-01 -5.19783646e-02
-6.32495165e-01 2.05612361e-01 1.34555325e-01 -4.44486380e-01
3.38969648e-01 -6.72915101e-01 -1.14911175e+00 -4.25969213e-01
1.07334949e-01 5.64703830e-02 5.67530572e-01 1.32434219e-01
1.51819587e-01 4.62718695e-01 -1.54962087e+00 7.57014215e-01
6.90243304e-01 1.23457885e+00 4.06181127e-01 8.93037617e-01
-2.07645848e-01 1.28917944e+00 -5.21693528e-01 -7.92132318e-01
3.38293433e-01 -2.31439397e-02 -1.03785956e+00 5.53547889e-02
9.75175142e-01 -7.23243177e-01 -4.99341637e-01 1.10504436e+00
4.38175857e-01 -3.62134784e-01 5.41261494e-01 -1.16952085e+00
-5.03699422e-01 6.92386687e-01 -6.64070129e-01 1.79728165e-01
-3.03700179e-01 3.46711934e-01 1.04163039e+00 -7.99681425e-01
3.00180048e-01 1.36350703e+00 1.14619100e+00 3.68932217e-01
-1.25338399e+00 -6.12817228e-01 1.93921030e-01 3.88350070e-01
-1.07608378e+00 -6.52402282e-01 3.33583623e-01 -1.03213608e-01
1.27434433e+00 -4.40570116e-02 5.68090260e-01 8.81037354e-01
2.59799153e-01 1.04418027e+00 9.64215398e-01 -1.81370109e-01
1.58977419e-01 -1.92308754e-01 3.90728474e-01 9.51649964e-01
1.78654686e-01 -2.00542659e-01 -5.58819294e-01 1.19012140e-01
7.60391414e-01 -1.26049921e-01 -1.21232994e-01 8.95243022e-04
-1.03960156e+00 4.89169687e-01 6.43024921e-01 -8.48360732e-02
-4.52624887e-01 8.67838144e-01 6.05683863e-01 4.11479145e-01
5.31401277e-01 9.82951075e-02 -7.43352294e-01 5.50081488e-03
-1.10895312e+00 -9.10247117e-02 6.44494116e-01 1.04697466e+00
1.08995318e+00 9.08351988e-02 -2.33582407e-01 6.36641026e-01
8.13207924e-02 4.48984891e-01 3.29150885e-01 -1.27425587e+00
3.27869534e-01 5.61963856e-01 -2.86167651e-01 -5.87034941e-01
-1.34919301e-01 -4.26243037e-01 -9.96561468e-01 4.12495613e-01
3.36110979e-01 -1.51732527e-02 -1.35071516e+00 1.47111285e+00
-6.23634830e-03 5.68350554e-01 7.16193169e-02 7.06406176e-01
6.11924827e-01 5.71991265e-01 3.62324238e-01 9.57683846e-02
1.47660863e+00 -1.32014632e+00 -5.46300948e-01 -4.52715039e-01
5.37212253e-01 -7.80769408e-01 1.22554123e+00 4.33961332e-01
-1.45462811e+00 -7.68530190e-01 -1.06354463e+00 -5.56708455e-01
-2.09697545e-01 3.55200619e-01 7.64827490e-01 5.31238317e-01
-1.35808599e+00 8.15708101e-01 -1.07379043e+00 -2.65990168e-01
8.32489252e-01 4.57024902e-01 -2.36895487e-01 -2.97179148e-02
-6.03003919e-01 7.36708283e-01 7.07676709e-02 -4.21046987e-02
-1.20574570e+00 -1.07341921e+00 -7.39015281e-01 2.62878418e-01
7.20068887e-02 -6.68929338e-01 1.55012321e+00 -1.13047898e+00
-1.28229082e+00 8.77651453e-01 -4.34766233e-01 -1.04270804e+00
4.37329054e-01 -6.01285100e-01 1.08945914e-01 1.53648198e-01
3.77517869e-03 9.26226676e-01 1.22507775e+00 -9.79181707e-01
-8.12029660e-01 -3.66071552e-01 2.35465795e-01 8.10898691e-02
-5.78529537e-01 -1.99804902e-01 -1.12774730e+00 -6.91079676e-01
3.69067416e-02 -7.62679875e-01 -2.56926149e-01 2.25589231e-01
-3.27797771e-01 -6.10840283e-02 9.43974078e-01 -7.00396121e-01
8.44675839e-01 -2.32481194e+00 -2.00371314e-02 -1.32383173e-02
4.11398113e-01 4.39717144e-01 -3.46633613e-01 -1.77673966e-01
-3.47224362e-02 -4.78545353e-02 -2.33845249e-01 -9.10715759e-01
-1.79017812e-01 5.30887842e-01 -3.76021266e-01 3.23538244e-01
4.76843804e-01 1.07674062e+00 -6.42000258e-01 -5.33161461e-01
3.00926983e-01 4.59276378e-01 -7.18754470e-01 1.34384692e-01
-2.28000820e-01 -2.60064244e-01 -8.50664452e-02 6.81495130e-01
8.69512737e-01 -3.21714073e-01 -3.47223915e-02 -8.22500169e-01
-1.04450792e-01 4.06032354e-01 -9.51736987e-01 1.67031276e+00
-7.21957684e-01 7.73849845e-01 3.10641319e-01 -9.09959316e-01
5.14420271e-01 6.19979054e-02 1.64463222e-01 -8.92850041e-01
-2.50203833e-02 -2.42082346e-02 -4.20958877e-01 -3.01267743e-01
5.04159927e-01 4.67064977e-01 3.56804311e-01 3.59411955e-01
3.12537044e-01 9.96269099e-03 1.40068710e-01 2.66070276e-01
1.33083427e+00 2.14467924e-02 -1.43836111e-01 -3.46114099e-01
3.15545291e-01 1.46512121e-01 3.80840093e-01 8.85712028e-01
-1.21400274e-01 6.23973489e-01 7.26101279e-01 -3.99126291e-01
-1.23972726e+00 -1.09212708e+00 1.19100273e-01 1.17889571e+00
6.27142414e-02 -6.15673244e-01 -1.07033885e+00 -7.79848099e-01
2.81669367e-02 3.66810918e-01 -6.31161749e-01 -1.43941388e-01
-4.96227175e-01 -8.07589173e-01 8.96344125e-01 8.51110220e-01
9.79599714e-01 -8.25186729e-01 -6.33273542e-01 2.00697228e-01
2.43298523e-02 -1.30724108e+00 -4.02573705e-01 5.97537756e-01
-1.00106120e+00 -1.08288062e+00 -7.28661299e-01 -8.42231691e-01
7.22927332e-01 4.46610779e-01 1.40991938e+00 9.77651626e-02
-2.61077166e-01 4.08463567e-01 -5.77120632e-02 -3.65689814e-01
-2.92295933e-01 2.56477237e-01 -3.63384932e-01 -1.70208558e-01
2.12773144e-01 -3.92611116e-01 -6.44912481e-01 -7.70051256e-02
-7.50942945e-01 1.07298389e-01 7.96130836e-01 7.70577073e-01
5.58864295e-01 -9.36849043e-03 -2.31346916e-02 -7.54526734e-01
3.55779469e-01 -3.20638120e-02 -7.64483690e-01 2.38980159e-01
-4.31937516e-01 3.91898066e-01 5.74088991e-01 -3.35612148e-01
-9.40372407e-01 3.74587655e-01 -3.25080693e-01 -6.76935792e-01
4.08631600e-02 5.19827791e-02 -7.49285594e-02 -3.09981942e-01
4.68762040e-01 1.21324398e-01 -3.99533063e-02 -5.83828509e-01
5.06931365e-01 4.72634017e-01 1.04967523e+00 -3.50909680e-01
6.48384750e-01 6.06382668e-01 -1.10203281e-01 -8.39882314e-01
-1.01895654e+00 -4.06554818e-01 -2.94586122e-01 6.87556714e-02
9.92274225e-01 -1.30416548e+00 -8.16625893e-01 7.64908671e-01
-1.33285630e+00 -6.95171356e-01 -2.24950343e-01 1.62621036e-01
-2.79240370e-01 3.57929051e-01 -1.03186381e+00 -5.26390433e-01
-7.89000452e-01 -1.42057812e+00 1.14229953e+00 5.07806763e-02
5.46981469e-02 -6.23606384e-01 -3.69478643e-01 2.07758576e-01
3.50721538e-01 -3.47232014e-01 7.03269064e-01 -2.71443069e-01
-8.67822349e-01 2.62698948e-01 -8.13252151e-01 8.78004611e-01
-1.13440484e-01 1.60189383e-02 -1.27970541e+00 -2.11041510e-01
-1.79183528e-01 -4.35867488e-01 1.45997548e+00 4.51554030e-01
1.64618468e+00 -1.83264747e-01 -2.25149710e-02 1.04289472e+00
1.53402555e+00 -2.18992725e-01 9.36704576e-01 2.88497210e-01
8.73640954e-01 2.01848283e-01 1.93072334e-01 2.54609704e-01
4.82051879e-01 1.75022945e-01 7.01559305e-01 -5.72217762e-01
-5.12579739e-01 8.34002122e-02 6.19506121e-01 6.39621258e-01
7.22823218e-02 5.84306289e-03 -7.66607225e-01 7.28806376e-01
-1.67672014e+00 -6.63866639e-01 -2.07935631e-01 1.94006848e+00
7.62532711e-01 3.35593969e-01 -2.24095970e-01 2.04221368e-01
3.15039456e-01 8.57437849e-02 -5.24229705e-01 -4.99624699e-01
-8.18061158e-02 6.31945908e-01 1.12570262e+00 5.19086421e-01
-1.14091289e+00 1.18556201e+00 7.15835667e+00 7.84298301e-01
-1.05962825e+00 5.15911877e-02 9.36662078e-01 -1.65115744e-02
-3.49345244e-02 -1.28838673e-01 -7.37167716e-01 3.64242077e-01
9.35000658e-01 4.30470407e-01 4.76467818e-01 9.39125001e-01
-9.12475586e-02 -1.43582761e-01 -1.08607709e+00 8.74614358e-01
-1.23642728e-01 -1.58722126e+00 2.74078071e-01 -1.83695227e-01
5.99648833e-01 4.28833425e-01 1.49123058e-01 1.94008321e-01
5.69609523e-01 -8.62124324e-01 7.49906838e-01 2.94325829e-01
7.19001710e-01 -9.66668308e-01 7.77794838e-01 -1.31224394e-01
-1.27231121e+00 -3.84655893e-02 -4.90421593e-01 9.40622687e-02
-2.47294649e-01 7.45594680e-01 -7.01838553e-01 1.90292552e-01
1.14975715e+00 6.45735800e-01 -7.58347511e-01 8.70711744e-01
-2.38323510e-01 8.69432271e-01 -3.29951078e-01 3.57654989e-01
4.60063279e-01 1.57845184e-01 2.28572395e-02 1.50420940e+00
2.09177807e-01 -1.08705640e-01 -2.58227587e-01 5.60307205e-01
-3.61129761e-01 -3.92458797e-01 -1.93146393e-01 1.96946368e-01
3.09411258e-01 1.09814739e+00 -6.35872602e-01 -7.23255098e-01
-6.04282975e-01 1.57410800e+00 3.53105873e-01 2.85018504e-01
-1.06964731e+00 -3.37774932e-01 1.11784649e+00 1.88794062e-02
6.22427702e-01 -2.55137831e-01 -5.87639093e-01 -9.18999314e-01
4.53927852e-02 -8.60563993e-01 2.12565631e-01 -8.95598054e-01
-1.05015552e+00 4.46958810e-01 -3.69967848e-01 -7.07121372e-01
2.49089316e-01 -7.98319638e-01 -5.47502697e-01 7.01828122e-01
-1.76647007e+00 -1.28955090e+00 -5.78485727e-01 7.07302988e-01
5.26686370e-01 1.03702836e-01 5.39347172e-01 5.36441743e-01
-5.77875912e-01 8.38278592e-01 -9.83973071e-02 2.17518747e-01
7.46317565e-01 -1.35127890e+00 9.80726957e-01 1.02446485e+00
2.25477323e-01 3.81191909e-01 3.52335334e-01 -3.77421349e-01
-1.38458180e+00 -1.36110544e+00 7.15814054e-01 -1.54530883e-01
6.40800893e-01 -2.70226389e-01 -8.93264353e-01 7.79119670e-01
3.57914716e-01 7.89918974e-02 4.41568904e-02 -1.26306638e-01
-5.21485984e-01 -4.05191183e-01 -9.20523226e-01 6.99290931e-01
1.13407040e+00 -7.19937682e-01 -4.12985682e-01 1.58406541e-01
9.81046379e-01 -3.48782986e-01 -6.43307567e-01 3.80186915e-01
4.16483343e-01 -1.02807271e+00 1.32275307e+00 -3.24308306e-01
5.97698689e-01 -2.29200155e-01 -1.53716981e-01 -1.04085839e+00
-3.43163669e-01 -5.36162913e-01 -2.42112964e-01 1.26994467e+00
2.20862135e-01 -3.96083862e-01 1.02673686e+00 2.90989637e-01
-3.49260896e-01 -5.68455458e-01 -8.01889241e-01 -5.37363231e-01
-1.66291326e-01 -6.65252626e-01 6.28582895e-01 6.11246586e-01
-7.14700401e-01 2.28384450e-01 -3.08926582e-01 4.51509207e-01
9.49997663e-01 -3.08044165e-01 8.25606704e-01 -9.01289701e-01
-2.78700709e-01 -1.21578567e-01 -3.30319673e-01 -1.33419049e+00
1.16986275e-01 -6.57349765e-01 4.38645892e-02 -1.49324977e+00
2.49526091e-02 -2.53509223e-01 -3.36692005e-01 8.39445293e-01
-1.71371862e-01 6.77650630e-01 1.37593225e-01 1.02287702e-01
-6.94337845e-01 3.05520862e-01 1.08890319e+00 -3.10564190e-01
7.32261911e-02 -3.86711895e-01 -6.16184950e-01 8.99983585e-01
6.67631626e-01 -3.55437249e-01 -4.26215172e-01 -1.10496247e+00
-1.00004174e-01 -4.23682481e-01 7.19670653e-01 -1.42598343e+00
3.94333959e-01 3.50722402e-01 4.55564439e-01 -7.44742632e-01
2.85500973e-01 -9.04235125e-01 -2.21995145e-01 6.50462985e-01
-2.12141514e-01 1.94927618e-01 4.80922967e-01 3.79537612e-01
-2.67966669e-02 -7.08571300e-02 1.00991070e+00 -1.55368298e-01
-1.07919538e+00 3.18320841e-01 -4.23109233e-01 -4.97990884e-02
6.47669256e-01 -7.05559328e-02 -3.72571975e-01 -9.63785872e-02
-5.35612643e-01 9.64694470e-02 4.23976004e-01 7.96492957e-03
5.04265070e-01 -9.77625191e-01 -6.34460032e-01 3.04993391e-01
-1.94327757e-01 1.89658538e-01 1.40787661e-01 6.32251978e-01
-8.38031471e-01 7.88449198e-02 -2.24406034e-01 -6.96045637e-01
-1.35533690e+00 2.11083725e-01 3.73127937e-01 -3.21516305e-01
-6.86541259e-01 1.16803741e+00 1.61355525e-01 -5.04580289e-02
5.67962587e-01 -7.85131812e-01 2.85841644e-01 1.16917258e-03
5.27743518e-01 3.81491154e-01 3.40945125e-01 -2.29644045e-01
-2.94830680e-01 4.86643106e-01 -2.85643339e-01 8.69588777e-02
1.26733565e+00 -1.60120636e-01 -1.71348363e-01 5.39022572e-02
1.24123013e+00 -3.51284236e-01 -1.70660365e+00 -2.53520012e-01
-7.86441341e-02 -1.28036410e-01 5.95231771e-01 -5.35215557e-01
-1.43881083e+00 8.94681096e-01 7.75372803e-01 -6.81612492e-02
1.44457901e+00 -7.03686848e-02 1.07088482e+00 6.39875710e-01
5.89163937e-02 -1.10226429e+00 1.90849632e-01 7.41957068e-01
5.36066949e-01 -1.09891284e+00 6.98859841e-02 -3.55831653e-01
-4.37720776e-01 8.68899822e-01 6.76249862e-01 -3.14396679e-01
4.90256697e-01 8.88175547e-01 3.58955227e-02 6.27972037e-02
-8.58950317e-01 -4.21759903e-01 1.69911198e-02 5.92966557e-01
2.65403718e-01 -3.48812133e-01 2.58865982e-01 1.33906618e-01
-6.05939329e-02 5.64929917e-02 2.36317620e-01 7.26499557e-01
-2.42845014e-01 -8.18148851e-01 -2.97842592e-01 4.78266269e-01
-6.64797366e-01 -4.98956144e-01 -2.06228465e-01 6.23892009e-01
2.23441735e-01 7.64910221e-01 5.71232617e-01 -3.28350872e-01
2.96760470e-01 -1.17016442e-01 3.69029224e-01 -1.58785388e-01
-8.88964534e-01 -7.18047619e-02 7.65372291e-02 -1.03242254e+00
-3.91148120e-01 -2.62659192e-01 -1.13299668e+00 -5.94542801e-01
-1.62546888e-01 -2.45069206e-01 7.43637919e-01 8.20777059e-01
4.96345788e-01 8.23943734e-01 3.13464046e-01 -8.83219540e-01
-4.89531904e-01 -6.87603712e-01 -2.92698801e-01 2.77749985e-01
4.76462543e-01 -1.39424771e-01 -2.32844859e-01 4.53672349e-01] | [9.446131706237793, 1.3370610475540161] |
e216cf9a-1b9c-452e-9f79-1d15b56a1d60 | zero3d-semantic-driven-multi-category-3d | 2301.13591 | null | https://arxiv.org/abs/2301.13591v4 | https://arxiv.org/pdf/2301.13591v4.pdf | Zero3D: Semantic-Driven Multi-Category 3D Shape Generation | Semantic-driven 3D shape generation aims to generate 3D objects conditioned on text. Previous works face problems with single-category generation, low-frequency 3D details, and requiring a large number of paired datasets for training. To tackle these challenges, we propose a multi-category conditional diffusion model. Specifically, 1) to alleviate the problem of lack of large-scale paired data, we bridge the text, 2D image and 3D shape based on the pre-trained CLIP model, and 2) to obtain the multi-category 3D shape feature, we apply the conditional flow model to generate 3D shape vector conditioned on CLIP embedding. 3) to generate multi-category 3D shape, we employ the hidden-layer diffusion model conditioned on the multi-category shape vector, which greatly reduces the training time and memory consumption. | ['Yitong Fu', 'Yixuan Shen', 'Bo Han'] | 2023-01-31 | null | null | null | null | ['3d-shape-generation'] | ['computer-vision'] | [ 1.34421187e-02 -1.10996559e-01 1.04121834e-01 -1.30422980e-01
-6.78545713e-01 -5.96364379e-01 6.88275337e-01 -3.84668350e-01
-3.82416025e-02 3.65298063e-01 4.71502632e-01 -9.65058357e-02
4.41903993e-02 -1.14210010e+00 -7.85116315e-01 -6.04605973e-01
3.93398792e-01 4.17222887e-01 1.03829443e-01 5.68109080e-02
3.99844557e-01 6.00511074e-01 -1.57307136e+00 2.16295496e-01
8.85221124e-01 1.03618658e+00 3.61297369e-01 5.69195330e-01
-6.28814399e-01 2.19027638e-01 -3.98028582e-01 -2.43182123e-01
4.08248305e-01 -5.30320108e-01 -4.42457110e-01 4.63450879e-01
4.31508720e-01 -5.31032026e-01 -1.72712192e-01 8.97592485e-01
5.80407798e-01 6.22410476e-02 1.27993047e+00 -1.26525056e+00
-1.08140159e+00 1.54253200e-01 -5.83185673e-01 -4.70967770e-01
2.18012661e-01 3.02065998e-01 7.15487480e-01 -1.31795645e+00
1.02618802e+00 1.42732954e+00 4.36021328e-01 8.20196390e-01
-1.31419504e+00 -7.05034316e-01 1.24796242e-01 -2.34287336e-01
-1.45840347e+00 -5.13003580e-03 1.39273345e+00 -6.34647965e-01
5.81697702e-01 -2.17336733e-02 8.36060941e-01 1.09910870e+00
-8.06209296e-02 7.83132613e-01 7.73373187e-01 -2.30360448e-01
2.60952026e-01 -2.59757247e-02 -3.81382883e-01 6.18314207e-01
3.95536460e-02 1.80164531e-01 -1.74316838e-01 1.68081578e-02
1.15881205e+00 2.74301916e-01 1.84440240e-02 -6.12573266e-01
-1.19442928e+00 1.04700923e+00 3.82353425e-01 4.39016134e-01
-3.01102489e-01 3.72074246e-02 1.16635732e-01 5.95268831e-02
7.44531691e-01 1.94525093e-01 -2.72811472e-01 1.95030838e-01
-1.02113938e+00 5.08590996e-01 5.43781161e-01 1.39729583e+00
9.77544308e-01 2.15026587e-01 -3.18636149e-01 8.46636355e-01
4.88769591e-01 9.81468260e-01 3.86323005e-01 -8.58489633e-01
7.04547584e-01 8.37218761e-01 -6.45002499e-02 -1.11458862e+00
-6.56493604e-02 -1.06037773e-01 -1.17362618e+00 1.34269416e-01
2.49244913e-01 -7.50258788e-02 -1.19513285e+00 1.45857334e+00
4.60715026e-01 4.29369276e-03 1.50695831e-01 1.03654242e+00
9.02052104e-01 9.66318429e-01 -1.03049248e-01 9.88600701e-02
1.02236426e+00 -9.40640032e-01 -3.30410719e-01 1.17548011e-01
4.92806345e-01 -1.02544999e+00 1.00854182e+00 -1.83483794e-01
-1.12766743e+00 -8.47548068e-01 -7.76637256e-01 -4.73418832e-01
-4.70556498e-01 9.88088623e-02 3.38340610e-01 2.70932943e-01
-7.56005645e-01 3.50861847e-01 -3.90078932e-01 -6.18153848e-02
7.15273499e-01 -3.13188210e-02 -3.21011245e-01 -5.03413737e-01
-9.75405455e-01 4.58100677e-01 3.32502216e-01 -8.95237327e-02
-9.57365572e-01 -8.07573736e-01 -9.45468426e-01 2.57590655e-02
6.04526363e-02 -9.49526608e-01 6.16582155e-01 -3.61905187e-01
-1.39592350e+00 8.05946112e-01 -3.62986326e-02 1.93350703e-01
5.97543478e-01 -2.52853259e-02 7.64805898e-02 8.72379020e-02
1.81881353e-01 1.15772378e+00 1.13622773e+00 -1.48663616e+00
-2.91491300e-01 -3.10904503e-01 -1.95785657e-01 3.37958395e-01
-2.68151671e-01 -5.92595816e-01 -6.17487371e-01 -1.02609181e+00
2.97719449e-01 -7.12615669e-01 -4.16781902e-01 3.51633459e-01
-4.32736903e-01 -2.31649905e-01 1.04264736e+00 -6.04591310e-01
6.67555332e-01 -2.32340622e+00 4.03279781e-01 2.69897506e-02
1.80027351e-01 6.62647188e-02 -3.95956188e-01 1.86328575e-01
1.30589187e-01 3.03285509e-01 -5.18224299e-01 -5.16170681e-01
1.26768649e-01 1.68140098e-01 -2.92438835e-01 -5.22110350e-02
6.13890707e-01 1.07723272e+00 -8.99505317e-01 -6.95726097e-01
5.47952890e-01 6.82733536e-01 -9.08131063e-01 5.27707219e-01
-4.51674551e-01 5.83099663e-01 -7.86647260e-01 4.62644100e-01
1.16865849e+00 -1.25828460e-01 -4.09293503e-01 -5.49001813e-01
-1.10249184e-01 -1.53291434e-01 -1.09613097e+00 2.09709787e+00
-6.43554807e-01 1.11038752e-01 -1.84444115e-01 -9.45735574e-01
1.35572481e+00 1.38339728e-01 4.70276773e-01 -5.59909105e-01
1.55173540e-01 3.24715018e-01 -5.07635415e-01 -4.87821370e-01
3.65914732e-01 -3.27256769e-01 -2.03980863e-01 5.59646726e-01
1.66865930e-01 -8.37930083e-01 -2.65140161e-02 2.90050656e-01
5.69268882e-01 4.12594169e-01 -4.19791043e-01 -4.74812500e-02
5.12013853e-01 -1.00737534e-01 3.50161165e-01 2.68766761e-01
2.76365906e-01 1.16380000e+00 5.25722086e-01 -4.09927726e-01
-1.46383989e+00 -1.24706006e+00 8.17134529e-02 3.20495367e-01
2.22684890e-01 -2.81674601e-02 -5.87406695e-01 -9.15019393e-01
1.24492183e-01 6.61679506e-01 -5.29489160e-01 -1.72848225e-01
-6.06742203e-01 -3.96137387e-01 1.30601123e-01 4.83910829e-01
4.56196994e-01 -1.04314315e+00 -2.15944827e-01 1.97719231e-01
-1.10162646e-01 -8.82804930e-01 -8.54505658e-01 -1.89601198e-01
-1.05803978e+00 -8.98067474e-01 -1.36222398e+00 -1.19407403e+00
1.09005773e+00 2.63331413e-01 8.89342904e-01 -3.35056260e-02
-4.12382215e-01 2.79272109e-01 -4.26907152e-01 -8.26504156e-02
-2.93016583e-01 -4.48957644e-02 -3.49948645e-01 1.36600986e-01
1.00462832e-01 -7.57986367e-01 -5.04136622e-01 1.14425950e-01
-1.10720921e+00 4.57602859e-01 7.77763605e-01 9.77184594e-01
7.40497231e-01 5.39789945e-02 6.04413927e-01 -6.16585433e-01
2.79970497e-01 -4.23129112e-01 -4.66755331e-01 -4.85191010e-02
-3.97343695e-01 1.11338757e-01 6.92005575e-01 -6.87111020e-01
-1.10831010e+00 2.34998956e-01 -3.25379193e-01 -9.31513846e-01
-2.52999276e-01 1.08317800e-01 -4.11039829e-01 3.14361811e-01
3.61603320e-01 5.91757774e-01 1.10653274e-01 -8.54405880e-01
8.89216840e-01 5.67283154e-01 2.48427764e-01 -5.13774335e-01
1.19010997e+00 5.01448214e-01 9.24533084e-02 -6.69646859e-01
-8.10717702e-01 -6.39901161e-02 -9.87111986e-01 -7.58797582e-03
1.31579304e+00 -8.85227263e-01 -1.78979844e-01 5.52989423e-01
-1.43340182e+00 -2.38338992e-01 -4.75374758e-01 4.62249041e-01
-6.14462435e-01 4.00299668e-01 -5.51931441e-01 -6.72915518e-01
-5.98601401e-01 -9.27252948e-01 1.34968209e+00 2.05111325e-01
1.36386082e-01 -9.16232705e-01 -2.71964103e-01 1.87123835e-01
4.92543608e-01 3.07961166e-01 1.16421723e+00 -8.66106227e-02
-8.59135032e-01 -2.18358010e-01 -5.58116853e-01 4.97786582e-01
1.85686544e-01 -2.78826743e-01 -6.10411406e-01 -6.22554980e-02
-4.57578525e-02 -3.74524891e-01 5.69509387e-01 2.39291623e-01
1.14512420e+00 -2.32464552e-01 -1.15536861e-01 6.07326329e-01
1.19117320e+00 6.15626909e-02 3.42698216e-01 -3.47977757e-01
1.11680996e+00 7.30347753e-01 5.22138834e-01 3.68099004e-01
4.96019721e-01 4.18637514e-01 2.09529489e-01 -1.10677831e-01
-6.49039388e-01 -9.34127629e-01 -9.64629278e-02 1.27317166e+00
2.09635720e-01 -1.12279430e-01 -3.92056316e-01 7.05736458e-01
-1.50571668e+00 -7.29303956e-01 -1.33695662e-01 1.99587095e+00
6.45026863e-01 1.20778047e-01 -1.95965990e-01 -4.88869324e-02
8.11964512e-01 2.48213142e-01 -5.52014351e-01 1.38822556e-01
-1.54319733e-01 -8.36268440e-03 -9.77742206e-03 3.71331513e-01
-7.17127562e-01 1.04920065e+00 5.88875008e+00 1.14230204e+00
-1.24683750e+00 -4.92594950e-02 5.79261899e-01 -5.48409745e-02
-8.56565416e-01 1.05656542e-01 -8.65969777e-01 7.56159067e-01
1.88967437e-01 -1.49755031e-01 1.67266175e-01 8.08961332e-01
-1.21460393e-01 2.51947433e-01 -9.77434397e-01 1.21797752e+00
3.73978376e-01 -1.39892519e+00 6.59581184e-01 1.22045889e-01
9.52921748e-01 -4.75227535e-01 8.53116717e-03 3.65095586e-01
5.48128448e-02 -7.41624713e-01 7.79591501e-01 5.89579463e-01
1.26816738e+00 -7.11280942e-01 3.67154568e-01 4.49793339e-01
-1.24603057e+00 1.52484000e-01 -6.18426263e-01 3.64339113e-01
3.80041391e-01 8.28094363e-01 -3.19418728e-01 6.73127592e-01
5.31748414e-01 9.37873721e-01 -2.91414797e-01 8.72231901e-01
-1.85028151e-01 1.33919999e-01 -3.51642221e-01 -8.01388472e-02
2.18663365e-01 -4.16104823e-01 4.76943344e-01 8.77712607e-01
8.09983552e-01 -2.62912479e-03 2.20939457e-01 1.42457139e+00
-1.54282361e-01 8.23241472e-02 -8.32888722e-01 -2.44413242e-02
3.72665852e-01 1.12040961e+00 -7.22272515e-01 -4.85522747e-01
-4.14851636e-01 1.16937292e+00 9.71326679e-02 2.63038009e-01
-5.55558085e-01 -5.34754276e-01 1.81617215e-01 7.84166306e-02
5.82454562e-01 -4.82224107e-01 -4.78687972e-01 -1.16053212e+00
7.56427348e-02 -1.99594483e-01 4.77316082e-02 -1.08519113e+00
-1.48953414e+00 4.53987807e-01 -8.64193514e-02 -1.61998737e+00
-1.31289244e-01 -4.12558019e-01 -6.44492090e-01 1.08146942e+00
-1.50793982e+00 -1.41829312e+00 -2.72360563e-01 6.64195120e-01
6.87844098e-01 -3.42528298e-02 6.48048222e-01 5.42897940e-01
-1.42707005e-01 3.78349543e-01 -2.41927609e-01 1.21344887e-01
7.35745668e-01 -8.71799171e-01 5.78437865e-01 5.20173788e-01
-6.66085184e-02 2.12355301e-01 1.17535762e-01 -9.19357836e-01
-1.47318459e+00 -1.48148727e+00 1.07832694e+00 -4.44397390e-01
1.57510161e-01 -5.53090453e-01 -8.46300483e-01 2.27963954e-01
-2.68102229e-01 9.60889980e-02 3.65208477e-01 -3.54819119e-01
-4.19051319e-01 8.06954503e-02 -1.14029074e+00 7.03554928e-01
1.46481764e+00 -4.67338502e-01 -7.00301051e-01 1.26378983e-01
1.00820494e+00 -3.05382460e-01 -1.04149747e+00 3.29005718e-01
3.08820009e-01 -6.36227310e-01 1.01464498e+00 -2.25605667e-01
8.84485483e-01 -3.67440075e-01 -1.46977529e-01 -1.35727763e+00
-4.56389308e-01 -3.73738676e-01 -1.39827386e-01 1.34266794e+00
3.99820417e-01 -1.95198148e-01 9.71801281e-01 5.00227273e-01
-3.12284708e-01 -6.34418905e-01 -8.38235319e-01 -6.77720428e-01
4.86850500e-01 -3.41874510e-01 8.61103475e-01 8.95951509e-01
-6.24850214e-01 5.49534380e-01 -6.53290629e-01 -3.66968483e-01
6.43914640e-01 6.91248417e-01 9.25458729e-01 -1.04920161e+00
2.86092032e-02 -4.37316090e-01 -1.90188661e-01 -1.52514791e+00
-7.35756829e-02 -1.08099532e+00 2.16713604e-02 -1.64319730e+00
7.02335238e-02 -7.64016151e-01 1.84589520e-01 2.52508163e-01
-2.23586455e-01 4.54550743e-01 3.87500077e-01 4.42222916e-02
-2.22815141e-01 1.20763326e+00 2.02936840e+00 -2.94172078e-01
-1.90956563e-01 -2.38970354e-01 -7.26692498e-01 3.91359776e-01
3.51330578e-01 -4.90479052e-01 -4.81042713e-01 -7.25259066e-01
-1.18488394e-01 2.59079337e-01 3.45408201e-01 -8.18927646e-01
1.72444582e-01 -2.71033943e-01 8.37144315e-01 -1.09622550e+00
4.96399015e-01 -8.95665646e-01 -1.20852679e-01 4.02133614e-01
-1.66257516e-01 -4.57194209e-01 -8.22767615e-02 5.42128325e-01
-1.87670812e-01 -1.45786330e-01 7.71308482e-01 -1.86286807e-01
-3.15521866e-01 8.54402721e-01 -7.14904740e-02 1.06668144e-01
9.91312265e-01 -3.11957181e-01 -6.81636902e-03 -1.61875024e-01
-6.69488013e-01 3.07761461e-01 5.82334280e-01 8.77618492e-01
8.43172252e-01 -1.87477064e+00 -7.40978479e-01 7.37690091e-01
1.25081345e-01 6.47397041e-01 8.02482545e-01 2.06225574e-01
-3.42935979e-01 9.70890373e-02 -2.48876855e-01 -6.51992500e-01
-6.25983715e-01 8.87077987e-01 -5.86162470e-02 4.96548973e-02
-8.17378283e-01 6.92518950e-01 4.86992121e-01 -7.46302009e-01
-1.05193101e-01 3.02275252e-02 -1.21684350e-01 6.90914094e-02
3.45077574e-01 3.85314040e-02 -4.34276968e-01 -7.16728032e-01
2.63459478e-02 1.42012858e+00 1.12125225e-01 -1.01759762e-01
1.39478970e+00 -6.16516247e-02 9.88555998e-02 2.33373240e-01
1.64537621e+00 -7.22054690e-02 -1.64485741e+00 -2.34182596e-01
-5.17949879e-01 -6.25331402e-01 5.85421873e-03 -3.59712422e-01
-1.30410802e+00 1.21765208e+00 4.70126957e-01 -1.03084631e-01
8.57355773e-01 -6.45233989e-02 1.10607052e+00 -7.02166036e-02
1.73042938e-01 -9.27485347e-01 3.84051025e-01 5.16702116e-01
1.09573460e+00 -1.01175058e+00 -1.68419704e-01 -5.97041667e-01
-5.55192053e-01 1.01977003e+00 7.28815913e-01 -4.56094742e-01
1.13096690e+00 -2.01602597e-02 5.96624874e-02 -7.66860694e-02
-6.17740393e-01 -7.37413093e-02 4.38467115e-01 9.26308990e-01
5.43790460e-02 -1.41461724e-02 -2.52156585e-01 6.01326883e-01
-1.68527365e-01 2.97750812e-03 1.94030285e-01 7.31014729e-01
-8.46006498e-02 -1.21845579e+00 -2.18490899e-01 4.93109435e-01
4.27917764e-02 -1.54681252e-02 -1.50744349e-01 3.12423140e-01
3.18401426e-01 6.08997166e-01 3.27502042e-01 -5.31827629e-01
4.21928346e-01 1.11520730e-01 3.93905282e-01 -5.82927167e-01
5.17420135e-02 3.22919756e-01 -4.02799159e-01 -2.66546816e-01
-1.49733871e-01 -3.80355030e-01 -1.01152396e+00 -2.35180467e-01
-1.09240592e-01 4.56835330e-02 6.81518376e-01 6.83833659e-01
7.51148641e-01 3.07413667e-01 9.69812453e-01 -1.30769408e+00
-4.04703856e-01 -9.46342707e-01 -6.58239782e-01 8.20461452e-01
1.01608239e-01 -7.22855806e-01 -4.94716942e-01 2.99648792e-01] | [8.865668296813965, -3.608395576477051] |
0841f17f-3d4f-4141-9779-4209325c0584 | evaluating-mt-systems-a-theoretical-framework | 2202.05806 | null | https://arxiv.org/abs/2202.05806v1 | https://arxiv.org/pdf/2202.05806v1.pdf | Evaluating MT Systems: A Theoretical Framework | This paper outlines a theoretical framework using which different automatic metrics can be designed for evaluation of Machine Translation systems. It introduces the concept of {\em cognitive ease} which depends on {\em adequacy} and {\em lack of fluency}. Thus, cognitive ease becomes the main parameter to be measured rather than comprehensibility. The framework allows the components of cognitive ease to be broken up and computed based on different linguistic levels etc. Independence of dimensions and linearly combining them provides for a highly modular approach. The paper places the existing automatic methods in an overall framework, to understand them better and to improve upon them in future. It can also be used to evaluate the newer types of MT systems, such as speech to speech translation and discourse translation. | ['Rajeev Sangal'] | 2022-02-11 | null | null | null | null | ['speech-to-speech-translation'] | ['speech'] | [-7.17260092e-02 3.36148083e-01 -3.74936104e-01 -3.59917104e-01
-6.33686125e-01 -7.72239804e-01 9.43259358e-01 8.59917924e-02
-3.21059465e-01 8.26100588e-01 1.90047503e-01 -7.64300048e-01
-5.08024514e-01 -4.95582134e-01 1.06315307e-01 -2.76348114e-01
4.99811202e-01 7.68645644e-01 7.12222094e-03 -5.73329449e-01
4.56063598e-01 5.16775608e-01 -1.58591878e+00 1.22273350e-02
1.17081141e+00 3.85875463e-01 5.48351109e-01 4.74004865e-01
-3.94793630e-01 7.40324736e-01 -5.98012567e-01 -5.31247258e-01
-8.36634114e-02 -6.79122627e-01 -1.21795690e+00 -4.72895764e-02
-3.20981033e-02 -4.22803964e-03 3.93593013e-01 1.01039433e+00
5.82177103e-01 -1.51253015e-01 8.67309391e-01 -8.50732207e-01
-8.78263235e-01 7.18163550e-01 3.24653387e-01 4.48167533e-01
9.65036929e-01 -6.54514926e-03 5.84179878e-01 -8.66482139e-01
5.85950196e-01 1.23353684e+00 4.28792328e-01 3.37680817e-01
-1.02543759e+00 -1.54379040e-01 -2.58258343e-01 1.81392536e-01
-9.64725971e-01 -4.86109018e-01 3.24878067e-01 -5.87602198e-01
1.08321619e+00 5.75781763e-01 4.85697627e-01 8.37627888e-01
5.20274751e-02 1.91718921e-01 1.89528120e+00 -9.17632043e-01
1.22702204e-01 8.67811561e-01 3.66159797e-01 4.70560193e-01
2.99843460e-01 3.50897796e-02 -3.18621904e-01 1.60664096e-01
4.97040302e-01 -6.20510995e-01 -1.18298024e-01 1.89061254e-01
-1.23354924e+00 6.92221165e-01 -4.08669442e-01 9.36313570e-01
-1.94197103e-01 -5.81769168e-01 2.94867873e-01 8.07978570e-01
2.20928803e-01 6.56409502e-01 -4.45309848e-01 -7.35759676e-01
-8.53656411e-01 2.39246473e-01 1.12376165e+00 9.65147734e-01
4.80909228e-01 -3.28645185e-02 -2.59269923e-01 9.07070518e-01
2.61460483e-01 7.12609351e-01 8.23522449e-01 -8.35417688e-01
3.43502253e-01 6.79638088e-01 2.41239935e-01 -6.36909485e-01
-5.44827700e-01 -4.49263811e-01 -1.08089685e-01 2.30590314e-01
3.80069435e-01 -1.09752312e-01 -5.95029891e-01 1.62063181e+00
2.20736908e-03 -9.02265131e-01 1.22469682e-02 7.92995334e-01
7.80679166e-01 4.85563099e-01 1.06634766e-01 -6.71555996e-01
1.34685373e+00 -9.11528409e-01 -1.13284683e+00 1.45527110e-01
7.50995934e-01 -1.31565368e+00 1.43209255e+00 3.30230951e-01
-1.55894089e+00 -7.00720668e-01 -1.02191114e+00 -1.17555544e-01
-8.44009519e-01 8.80764425e-02 4.16108996e-01 1.09812808e+00
-1.63898396e+00 6.18817329e-01 -4.49981332e-01 -6.53330386e-01
-5.62643349e-01 4.82625633e-01 -6.02434650e-02 4.63122517e-01
-1.23045456e+00 1.70983851e+00 4.54646409e-01 -9.40913707e-02
-7.98792243e-02 6.52108565e-02 -5.99995673e-01 -1.14402696e-01
-9.66993123e-02 -8.37212503e-01 1.23030722e+00 -1.10526812e+00
-1.88129354e+00 1.05003726e+00 -1.35467961e-01 -2.09301319e-02
6.95125103e-01 -7.62992576e-02 -7.31183469e-01 6.40302375e-02
3.32747191e-01 3.43495131e-01 3.58431816e-01 -1.03079855e+00
-6.07761145e-01 -1.99664593e-01 2.76352078e-01 5.96951306e-01
-4.31324661e-01 5.47017574e-01 -9.31958556e-02 -5.07177293e-01
-6.14292780e-03 -9.32149649e-01 3.92508149e-01 -6.58595562e-01
4.21090089e-02 -4.11113441e-01 4.40045178e-01 -8.22318614e-01
1.68377519e+00 -1.55724919e+00 5.50017297e-01 -7.72207417e-03
9.70560983e-02 3.46581459e-01 1.03527501e-01 7.76881993e-01
-3.90919261e-02 3.08765471e-01 8.81787091e-02 1.10798605e-01
3.53055179e-01 7.42841437e-02 1.99668482e-01 -1.40718082e-02
3.44229713e-02 8.65294218e-01 -7.70569503e-01 -6.72660232e-01
3.80851805e-01 3.31547230e-01 7.95039162e-02 -4.80772071e-02
1.87639371e-01 1.91428423e-01 -6.34077013e-01 5.20473719e-01
4.04979140e-01 9.02709365e-03 7.66703337e-02 4.00884837e-01
-7.17731237e-01 5.62200606e-01 -8.27333450e-01 1.32239532e+00
-5.38789809e-01 5.48372090e-01 -2.85393059e-01 -7.48085558e-01
1.17370713e+00 7.10690498e-01 2.65299007e-02 -9.40584838e-01
3.55690271e-01 5.97978115e-01 1.50226370e-01 -7.05910623e-01
5.48119128e-01 -9.22150910e-02 3.81579176e-02 6.70181990e-01
4.02345471e-02 -1.53295279e-01 5.36018789e-01 -1.43014640e-01
6.05562031e-01 3.20500582e-01 6.28278375e-01 -8.48793864e-01
9.41385210e-01 1.26730815e-01 -2.19587367e-02 5.15532315e-01
-2.61163950e-01 -3.45144644e-02 3.24722379e-01 -2.34001726e-01
-1.21372116e+00 -9.92012143e-01 -3.58423144e-01 1.03690469e+00
-1.57875821e-01 -2.28403524e-01 -1.24292815e+00 -3.53581131e-01
-7.06650198e-01 8.44881117e-01 -3.02175969e-01 2.03776583e-01
-4.10844684e-01 -6.22710347e-01 3.94070297e-01 1.46982327e-01
4.58542794e-01 -1.07494938e+00 -6.85288846e-01 1.42894953e-01
-5.25030494e-01 -9.20248806e-01 1.75886959e-01 -6.93358853e-02
-1.04141700e+00 -4.15508240e-01 -5.70462942e-01 -9.30921853e-01
1.88298911e-01 2.86490619e-01 1.37776458e+00 3.29642296e-01
3.82481426e-01 3.51209998e-01 -7.92671740e-01 -5.23564577e-01
-8.78616154e-01 3.32472026e-01 2.31529489e-01 -7.74938762e-01
7.03384697e-01 -7.42713034e-01 -1.95653915e-01 4.69062239e-01
-7.64756083e-01 7.22656101e-02 7.14181364e-01 5.76513648e-01
1.72216535e-01 -2.76932061e-01 6.31293595e-01 -5.85329413e-01
1.31945562e+00 -1.55320212e-01 -3.91015038e-02 4.78772014e-01
-1.07838356e+00 3.60540859e-02 4.96532172e-01 -1.61462963e-01
-9.97298002e-01 -4.68869328e-01 -2.30011955e-01 4.15687621e-01
-2.72644669e-01 4.90300238e-01 -4.25409712e-02 -1.69983312e-01
5.71315587e-01 2.19588041e-01 4.43193391e-02 -4.66719449e-01
2.92710215e-01 9.98544812e-01 2.50489444e-01 -7.33872771e-01
6.15108013e-01 -2.85534620e-01 -1.74799785e-01 -8.81798327e-01
-3.68770838e-01 -1.28990725e-01 -8.57905805e-01 -4.44090843e-01
8.23502719e-01 -5.22762716e-01 -4.71316844e-01 5.02860807e-02
-1.06626022e+00 -1.70835897e-01 -4.27485928e-02 6.65591061e-01
-9.34286177e-01 5.67881703e-01 -6.50790930e-01 -9.01870728e-01
-3.75088543e-01 -1.20479715e+00 8.82722855e-01 2.51225531e-01
-7.10094392e-01 -1.28203511e+00 7.86222294e-02 6.75793707e-01
5.65487087e-01 -6.37601614e-02 1.21710515e+00 -6.90087974e-01
-8.29986706e-02 -1.14301264e-01 -9.89471097e-03 3.26807618e-01
5.99764362e-02 5.56899644e-02 -7.17271805e-01 -2.33460143e-02
5.88574946e-01 -1.51711240e-01 1.33244932e-01 1.58467278e-01
1.82755440e-01 -4.19103771e-01 -1.08894974e-01 -8.14399198e-02
1.36136532e+00 4.05572087e-01 8.28924417e-01 7.46325672e-01
1.16054431e-01 9.95324314e-01 7.04278767e-01 -8.96301493e-02
3.78900260e-01 9.53845024e-01 -2.59931803e-01 -2.59715430e-02
-1.46441191e-01 1.83087215e-01 6.41536891e-01 1.61960316e+00
-5.78390956e-01 -2.19050854e-01 -9.19349790e-01 1.46440551e-01
-1.55454469e+00 -1.03509116e+00 -2.45836586e-01 2.15244222e+00
6.42538846e-01 2.09318623e-01 4.30148005e-01 3.98736894e-01
8.17654848e-01 -4.10404921e-01 3.13487589e-01 -1.16458070e+00
-9.24078971e-02 4.03032333e-01 3.14210765e-02 9.42128479e-01
-6.18523955e-01 1.12079978e+00 7.50105143e+00 7.16921508e-01
-1.08102357e+00 2.79727370e-01 4.47569072e-01 2.70218879e-01
-3.18077445e-01 1.82200566e-01 -6.49782300e-01 4.99225527e-01
1.20184958e+00 -5.68614185e-01 4.54710394e-01 5.19445419e-01
5.19715071e-01 -1.88121945e-01 -1.09676576e+00 5.43333948e-01
6.34540021e-02 -7.50715494e-01 1.91874191e-01 1.77295119e-01
4.06707942e-01 -2.75349170e-01 1.17936201e-01 4.17172343e-01
8.05447176e-02 -9.85709012e-01 8.24074447e-01 5.67994952e-01
6.02591038e-01 -4.93686020e-01 8.95494580e-01 5.59045792e-01
-7.68074214e-01 1.26176789e-01 -3.36213291e-01 -5.90198159e-01
1.92603081e-01 1.36443481e-01 -6.13898814e-01 1.01225865e+00
3.27908218e-01 2.87685022e-02 -6.81154907e-01 4.16367710e-01
-1.42271042e-01 2.41779894e-01 -6.36194721e-02 -5.88483274e-01
2.90996104e-01 -5.56473672e-01 5.91262639e-01 1.41084635e+00
5.70116043e-01 -1.48364574e-01 -9.41176787e-02 7.52118289e-01
6.08426571e-01 7.96846390e-01 -6.42489731e-01 -7.27652088e-02
6.27613842e-01 1.24284720e+00 -9.18334544e-01 -4.96518254e-01
-5.00547588e-01 1.05209696e+00 2.21996158e-01 7.77106658e-02
-9.19994533e-01 -2.80558378e-01 1.59253031e-01 8.00976977e-02
-3.10320973e-01 -3.12021047e-01 -5.91274500e-01 -9.55376804e-01
2.58388460e-01 -1.17356634e+00 8.38756189e-02 -7.01915443e-01
-6.90468907e-01 1.04517639e+00 2.72918969e-01 -1.22033799e+00
-5.44469476e-01 -8.61363828e-01 -3.46089423e-01 1.25503123e+00
-1.09501171e+00 -8.64834368e-01 -9.33896899e-02 2.99504548e-01
6.24144375e-01 -2.96032459e-01 1.22506475e+00 2.18130499e-01
-4.26759541e-01 5.61003506e-01 -1.47282928e-01 -3.05287868e-01
4.99257207e-01 -1.25418544e+00 5.54988049e-02 7.79002845e-01
-1.20815836e-01 8.01113069e-01 8.32062364e-01 -3.76778096e-01
-1.04346502e+00 -3.97616416e-01 1.70062220e+00 -9.21967864e-01
6.29884899e-01 1.61856748e-02 -4.59367275e-01 1.99525014e-01
6.35715663e-01 -1.31487858e+00 7.46025741e-01 1.44144416e-01
1.01995975e-01 2.09692761e-01 -9.94209290e-01 7.46915817e-01
9.52412009e-01 -5.84711075e-01 -1.12006879e+00 3.46597165e-01
5.92865169e-01 -7.36965314e-02 -1.17672062e+00 3.84280890e-01
4.32731062e-01 -1.34572840e+00 6.48466885e-01 -2.76546478e-01
3.25032920e-01 -1.59360677e-01 4.05867547e-02 -1.12334049e+00
-5.72539449e-01 -8.19036961e-01 2.06074581e-01 1.29308283e+00
8.84720683e-01 -8.72239470e-01 4.43010628e-02 5.66501379e-01
-1.89948022e-01 -5.62229335e-01 -8.61657083e-01 -9.22383726e-01
3.60883534e-01 -3.11923742e-01 5.24309278e-01 1.09142947e+00
6.69428825e-01 7.20267415e-01 -1.56279832e-01 -4.13462996e-01
-2.07208470e-02 -2.35474423e-01 5.03187656e-01 -1.41193867e+00
-3.09989214e-01 -1.09185970e+00 -4.59976792e-01 -7.19098389e-01
-1.61286592e-01 -8.59239399e-01 -5.43022931e-01 -1.55272567e+00
2.15856835e-01 -3.76803428e-02 1.90471578e-02 1.21644616e-01
-1.25253245e-01 2.11419120e-01 2.10153475e-01 4.22684431e-01
-2.39818826e-01 5.88863641e-02 1.18663394e+00 3.38160098e-01
-1.85370207e-01 -1.49955601e-01 -8.95844340e-01 6.96524441e-01
1.01000166e+00 -2.11846739e-01 -5.08988619e-01 -5.03796458e-01
3.92348289e-01 2.57165670e-01 1.90297759e-03 -1.10530400e+00
-1.85389072e-01 -2.23022997e-01 2.28265539e-01 -1.52084649e-01
1.52679667e-01 -6.26406848e-01 2.94538230e-01 4.60107207e-01
-2.11466417e-01 8.32407773e-01 -3.92688029e-02 -2.65757471e-01
-2.87357897e-01 -4.84630674e-01 4.90295261e-01 -3.01695853e-01
-2.36559063e-01 -3.66142124e-01 -5.58923721e-01 -3.30100060e-01
9.22870874e-01 -7.09490836e-01 -2.31164023e-01 -3.77513975e-01
-8.50767791e-01 -5.34000136e-02 6.20770037e-01 6.01627290e-01
-5.25549129e-02 -1.30463266e+00 -7.10640967e-01 -2.56031573e-01
-1.22068428e-01 -9.20491457e-01 -1.02390237e-01 1.13764668e+00
-9.20525610e-01 9.95529771e-01 -5.77718914e-01 -3.12232673e-01
-1.20820010e+00 6.48958564e-01 2.62209892e-01 -3.67233306e-01
-2.90390819e-01 2.04762369e-01 -1.54225916e-01 -2.13426948e-01
-3.94075476e-02 -1.07175512e-02 -7.19848335e-01 1.88175607e-02
4.24479425e-01 6.87039614e-01 1.87534109e-01 -9.58783388e-01
-1.81651160e-01 7.08582163e-01 1.19118728e-01 -6.96636677e-01
7.60302961e-01 -6.41813278e-01 -5.72528780e-01 6.85567081e-01
5.90436399e-01 1.17866963e-01 -1.09047063e-01 1.51312783e-01
4.15993631e-01 -2.03184783e-01 -2.03616485e-01 -1.19366491e+00
-1.33373350e-01 8.48267496e-01 7.70257950e-01 6.00277126e-01
1.23188949e+00 -2.08355680e-01 1.27804533e-01 5.02130508e-01
4.67153668e-01 -1.67468321e+00 -4.06108677e-01 6.24778450e-01
8.01492989e-01 -8.26536000e-01 -2.55040884e-01 -5.59539080e-01
-4.98563796e-01 1.51459837e+00 3.33457947e-01 3.68512839e-01
2.99619704e-01 3.29359978e-01 3.20441067e-01 -1.02940230e-02
-4.36138451e-01 -4.40180779e-01 3.43654126e-01 6.19831204e-01
1.08902442e+00 4.11855131e-01 -1.64567697e+00 5.01427531e-01
-8.14520955e-01 4.79321592e-02 3.71523023e-01 7.10066140e-01
-7.97513902e-01 -1.64144039e+00 -3.70228887e-01 9.53552872e-02
-5.75436592e-01 -1.06076643e-01 -8.95295918e-01 1.01427019e+00
3.15971822e-01 1.32195270e+00 -4.49837655e-01 -6.72944844e-01
3.76163989e-01 5.43039441e-01 7.12617218e-01 -4.81513709e-01
-7.06112146e-01 1.09938085e-01 4.59328145e-01 -7.61525109e-02
-7.14238763e-01 -6.47113979e-01 -5.77751100e-01 -6.53876364e-01
-2.43769661e-01 5.13606369e-01 7.69836724e-01 1.23673832e+00
2.61675492e-02 2.21799463e-01 5.84797442e-01 -5.68389475e-01
-5.76840341e-01 -1.49294281e+00 -1.79542243e-01 1.76609278e-01
-2.14964703e-01 -5.65934181e-01 -2.70466119e-01 3.52311023e-02] | [11.17314624786377, 9.798202514648438] |
f552cdd8-5e51-4756-8632-763fcc98afd7 | controllable-radiance-fields-for-dynamic-face | 2210.05825 | null | https://arxiv.org/abs/2210.05825v1 | https://arxiv.org/pdf/2210.05825v1.pdf | Controllable Radiance Fields for Dynamic Face Synthesis | Recent work on 3D-aware image synthesis has achieved compelling results using advances in neural rendering. However, 3D-aware synthesis of face dynamics hasn't received much attention. Here, we study how to explicitly control generative model synthesis of face dynamics exhibiting non-rigid motion (e.g., facial expression change), while simultaneously ensuring 3D-awareness. For this we propose a Controllable Radiance Field (CoRF): 1) Motion control is achieved by embedding motion features within the layered latent motion space of a style-based generator; 2) To ensure consistency of background, motion features and subject-specific attributes such as lighting, texture, shapes, albedo, and identity, a face parsing net, a head regressor and an identity encoder are incorporated. On head image/video data we show that CoRFs are 3D-aware while enabling editing of identity, viewing directions, and motion. | ['Alexander G. Schwing', 'Oluwasanmi Koyejo', 'Liqian Ma', 'Peiye Zhuang'] | 2022-10-11 | null | null | null | null | ['face-parsing', 'face-generation', '3d-aware-image-synthesis'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 2.78062731e-01 2.26520255e-01 1.07816495e-01 -6.46977961e-01
-2.84292519e-01 -6.69497490e-01 9.88137245e-01 -8.28745663e-01
2.06233397e-01 4.17025149e-01 4.67183471e-01 1.07117057e-01
3.43078732e-01 -5.43862104e-01 -7.88467348e-01 -8.46828401e-01
1.40112579e-01 4.57233377e-02 -4.32671726e-01 -3.69851701e-02
-5.14389365e-04 8.22129667e-01 -1.58464682e+00 1.23666421e-01
2.87618071e-01 7.25322664e-01 -1.34799451e-01 9.88941014e-01
1.67529315e-01 8.79477978e-01 -3.41654837e-01 -1.38364226e-01
3.31023157e-01 -9.26895142e-01 -3.45016360e-01 6.57066286e-01
9.50175047e-01 -5.04737914e-01 -2.46648908e-01 7.67238379e-01
5.56460023e-01 2.10945785e-01 7.67581463e-01 -1.29643512e+00
-1.00067842e+00 -8.71323943e-02 -5.44642806e-01 -2.90367663e-01
3.36700529e-01 3.43308628e-01 6.05892479e-01 -6.54823780e-01
1.06173038e+00 1.69009173e+00 2.35667080e-01 1.20797515e+00
-1.58945930e+00 -5.15031993e-01 3.21391851e-01 -2.47982070e-01
-1.21893978e+00 -9.20762002e-01 1.01520157e+00 -5.71349263e-01
5.85893631e-01 4.72038746e-01 8.33992064e-01 1.52464783e+00
3.79705727e-01 3.57796133e-01 8.85205448e-01 -3.80386680e-01
3.37634176e-01 6.05711341e-02 -6.61374569e-01 9.32954848e-01
-2.82443762e-01 1.53413594e-01 -6.69581056e-01 -3.68832722e-02
1.39186740e+00 -5.09208083e-01 -2.75231123e-01 -7.45490551e-01
-9.75976348e-01 6.27915263e-01 -9.33137257e-03 -2.13326573e-01
-1.49596959e-01 5.20808697e-01 -1.36571392e-01 2.39816487e-01
8.02368164e-01 3.05351347e-01 -3.41879427e-01 1.27325714e-01
-8.35803568e-01 6.80913806e-01 5.96187353e-01 1.07389510e+00
6.63747549e-01 6.28745615e-01 -3.09830725e-01 6.09943151e-01
3.79960865e-01 9.31657732e-01 2.82549471e-01 -1.63135612e+00
-3.47476676e-02 8.99071768e-02 1.64803013e-01 -1.10322106e+00
-2.73031980e-01 -4.45302837e-02 -6.98024035e-01 5.53483903e-01
1.16062351e-01 -2.40544990e-01 -1.12376904e+00 2.44761181e+00
6.39015913e-01 3.28881264e-01 -1.27341121e-01 9.29638088e-01
4.88423645e-01 5.58854461e-01 2.91119441e-02 -3.65405709e-01
1.14245725e+00 -6.12216711e-01 -8.29684794e-01 -1.36017546e-01
2.41662294e-01 -8.19847703e-01 8.91595006e-01 1.46197537e-02
-1.48052096e+00 -5.86072326e-01 -6.72039211e-01 -2.44451195e-01
1.28510401e-01 1.00187488e-01 5.29316366e-01 6.60217464e-01
-1.44029558e+00 3.62168729e-01 -1.04156196e+00 -3.02110374e-01
1.56488091e-01 4.01254535e-01 -5.38866878e-01 2.93332011e-01
-7.87271380e-01 6.91325247e-01 -3.87813449e-01 -2.91264206e-02
-1.03031063e+00 -7.85034657e-01 -1.01326847e+00 -1.47527233e-01
-7.05556422e-02 -1.13514984e+00 1.12581015e+00 -1.39972770e+00
-2.25998211e+00 1.09589124e+00 -3.32754344e-01 2.30306834e-01
5.48334956e-01 -7.52107725e-02 -2.17918843e-01 8.39030370e-02
-1.47429585e-01 1.09475136e+00 1.45790601e+00 -1.30424070e+00
-2.07330525e-01 -3.27232122e-01 -8.02557319e-02 4.95158345e-01
-3.98968786e-01 7.11113364e-02 -5.99694610e-01 -7.72337556e-01
-1.17087821e-02 -1.27473998e+00 -8.19084942e-02 6.82545424e-01
-2.34086156e-01 3.59614313e-01 1.01529908e+00 -7.01892018e-01
7.60553658e-01 -2.25266290e+00 5.46332896e-01 3.36798318e-02
4.59043942e-02 -8.76816660e-02 -4.14363056e-01 -1.64800391e-01
-1.16781496e-01 2.03071982e-02 -1.81152955e-01 -7.36403108e-01
1.90704521e-02 2.27333456e-01 -2.76582330e-01 5.77209055e-01
5.24695277e-01 8.21384490e-01 -6.02448881e-01 -2.56430060e-01
3.09535384e-01 1.04401886e+00 -1.21922386e+00 4.43988651e-01
-6.37563944e-01 1.04023778e+00 -3.07556987e-01 4.38315123e-01
6.64843321e-01 -1.66531235e-01 2.81238139e-01 -8.00034478e-02
-1.39877021e-01 -7.52599761e-02 -1.01013649e+00 1.69766366e+00
-6.10933900e-01 7.79874623e-01 5.40759265e-01 -2.39474803e-01
8.33673596e-01 4.55125928e-01 6.00643873e-01 -4.68917817e-01
1.54823795e-01 -2.83087254e-01 -1.82991266e-01 -3.66633326e-01
2.92060733e-01 -1.56403854e-02 2.47600481e-01 4.00847644e-01
-5.63411191e-02 -4.20146853e-01 -4.23721313e-01 -9.93014574e-02
8.20747674e-01 6.66196406e-01 -2.12145135e-01 -3.35824549e-01
2.91358441e-01 -4.87997085e-01 5.50080001e-01 1.66872680e-01
1.17540322e-01 1.04867971e+00 5.50429642e-01 -2.60027319e-01
-1.38681698e+00 -1.06777942e+00 6.23860257e-03 9.47796583e-01
-2.21351013e-01 -9.08881873e-02 -1.18483925e+00 -1.25098154e-01
1.22074867e-02 5.43206334e-01 -7.79767692e-01 -2.59969383e-01
-9.27267611e-01 -6.59367204e-01 2.65841395e-01 3.29078555e-01
3.68871242e-01 -7.82795668e-01 -6.97492540e-01 9.86534208e-02
-2.69475318e-02 -1.16053271e+00 -1.01956296e+00 -3.53264064e-01
-6.22765064e-01 -4.94237751e-01 -7.57097602e-01 -5.48610747e-01
9.43690658e-01 1.07851416e-01 9.58183110e-01 -1.75451726e-01
-3.73843163e-01 5.72069764e-01 1.16303466e-01 1.78064611e-02
-5.32285333e-01 -1.54834554e-01 1.64575309e-01 5.11577785e-01
-3.93260956e-01 -8.47321868e-01 -7.25069106e-01 2.62106866e-01
-7.53072202e-01 4.38876629e-01 -1.54432461e-01 5.54065883e-01
4.31035429e-01 -5.38258910e-01 2.02172905e-01 -6.84328139e-01
2.50130385e-01 -1.12719961e-01 -8.42141688e-01 -5.84470257e-02
-4.70231891e-01 2.50962954e-02 3.64698172e-01 -7.51541615e-01
-1.47241926e+00 3.26780617e-01 -8.23853165e-02 -7.75814295e-01
-1.39925241e-01 -3.98507148e-01 -5.86684346e-01 -1.39947727e-01
6.06750250e-01 2.10647900e-02 9.29402485e-02 -2.50412196e-01
8.35526347e-01 2.92901337e-01 7.21485376e-01 -6.57323003e-01
8.49217057e-01 7.57340789e-01 2.41465345e-01 -1.07286084e+00
-4.81448233e-01 2.59422988e-01 -7.19993174e-01 -2.17621982e-01
1.22989643e+00 -1.06474733e+00 -7.62941420e-01 5.83020866e-01
-1.18205631e+00 -7.15985715e-01 -1.87595129e-01 3.37041527e-01
-9.97043610e-01 -1.24234870e-01 -5.59285522e-01 -9.15942311e-01
-1.44267663e-01 -1.08128035e+00 1.35310495e+00 2.12846905e-01
-5.56415021e-01 -1.00837791e+00 -1.58529561e-02 4.97234315e-02
5.53783834e-01 7.03157067e-01 1.06698334e+00 4.50244874e-01
-9.03812826e-01 2.31741920e-01 1.33700520e-01 1.88757151e-01
3.80325794e-01 4.06956971e-01 -1.15872681e+00 -2.94502437e-01
2.10181382e-02 -6.16989983e-03 2.98040748e-01 7.10055232e-01
1.06157422e+00 -5.94248891e-01 -1.47873610e-01 1.31251347e+00
9.04696584e-01 3.72389168e-01 4.66287166e-01 -2.18985960e-01
1.03590572e+00 7.45783627e-01 -1.56244829e-01 5.80709755e-01
3.06255430e-01 1.05185246e+00 3.04127276e-01 -1.02621384e-01
-6.42657340e-01 -3.52313787e-01 5.31786144e-01 5.98359883e-01
-1.75908104e-01 -3.57969850e-01 -3.51302564e-01 2.31527954e-01
-1.53068376e+00 -8.76286387e-01 3.15659046e-01 1.97238016e+00
8.30065608e-01 -3.37105602e-01 -8.44027624e-02 -3.95185024e-01
5.24363279e-01 4.18301642e-01 -7.54949510e-01 -3.45095277e-01
-2.84467250e-01 6.67812750e-02 1.52137488e-01 8.27813089e-01
-8.35826397e-01 1.14787900e+00 6.39710665e+00 1.65399268e-01
-1.47244096e+00 -3.38913761e-02 7.93345273e-01 -4.27201122e-01
-9.06257927e-01 7.72605045e-03 -7.84389257e-01 2.71855205e-01
5.83584249e-01 1.29003853e-01 6.73839509e-01 6.19784236e-01
6.56504333e-01 4.05378789e-01 -1.24061501e+00 8.79740119e-01
3.75107110e-01 -1.29939353e+00 2.39808351e-01 2.31112778e-01
8.81813705e-01 -6.53862894e-01 5.10024846e-01 -1.49032071e-01
2.81646162e-01 -1.01167607e+00 9.97221828e-01 7.85058677e-01
1.27467167e+00 -6.61670685e-01 -3.15166444e-01 2.27250438e-02
-8.46540570e-01 2.40580380e-01 6.68948004e-03 2.72109240e-01
2.90023953e-01 1.53825104e-01 -3.49071056e-01 -6.31428808e-02
4.02740717e-01 5.44142723e-01 -1.96885496e-01 1.44239530e-01
-2.43080750e-01 3.93166929e-01 -1.31627664e-01 4.82868314e-01
-4.81174849e-02 -3.31468284e-01 6.50453627e-01 8.52309704e-01
3.96270514e-01 3.61102074e-01 -1.70872912e-01 1.15304327e+00
-2.50155181e-01 -7.36326948e-02 -5.76902032e-01 3.26293185e-02
1.80607110e-01 1.13108504e+00 -3.61105204e-01 5.32931462e-02
-1.56850979e-01 1.34472799e+00 1.24298349e-01 5.38967669e-01
-8.00623000e-01 3.64265323e-01 1.44834721e+00 4.26175147e-01
1.84425727e-01 -4.40752387e-01 -1.34170577e-01 -1.29736161e+00
-2.45335087e-01 -7.73699820e-01 -2.41459906e-01 -1.08620262e+00
-8.55186403e-01 5.34265697e-01 -4.37090248e-02 -7.00989008e-01
-6.07080340e-01 -5.13048232e-01 -5.98370016e-01 9.21084166e-01
-1.09093654e+00 -1.45197821e+00 -1.59489155e-01 5.97719967e-01
5.41289687e-01 -2.68952012e-01 8.43552709e-01 1.52191058e-01
-6.55177474e-01 6.93989515e-01 -2.48946562e-01 -3.98254991e-02
7.18011260e-01 -8.91865969e-01 8.33491445e-01 6.93651378e-01
6.13899678e-02 6.09609485e-01 6.94633305e-01 -6.60450876e-01
-1.80272031e+00 -1.19979095e+00 6.32332385e-01 -7.18557477e-01
1.09144978e-01 -7.02842712e-01 -6.37718737e-01 8.22028041e-01
5.96971698e-02 1.70951545e-01 2.91443288e-01 -3.65455538e-01
-4.17755604e-01 -3.67314927e-02 -1.18032193e+00 1.01398826e+00
1.36748469e+00 -6.05365932e-01 2.54827976e-01 2.30954796e-01
6.78996265e-01 -7.37699151e-01 -6.16966844e-01 2.76329219e-01
8.14554274e-01 -7.70085514e-01 1.05630422e+00 -7.62193084e-01
4.15490508e-01 -2.59605259e-01 -3.96138698e-01 -1.31599247e+00
-4.63613093e-01 -1.21771276e+00 -2.10447595e-01 1.26326001e+00
9.47249960e-03 -1.35999411e-01 9.34992552e-01 9.19404268e-01
-5.00521250e-02 -3.92650843e-01 -7.49375045e-01 -3.89377922e-01
1.08456254e-01 -1.17273211e-01 8.14067841e-01 9.77941871e-01
-6.56625152e-01 2.57180631e-01 -8.77973378e-01 2.82721579e-01
5.17881095e-01 9.85035822e-02 9.82028902e-01 -7.76650190e-01
-3.59659344e-01 -3.78410816e-01 -2.07675129e-01 -1.02674031e+00
4.30928409e-01 -6.60873830e-01 -3.76339145e-02 -9.70943630e-01
-4.51002177e-03 -1.83732405e-01 4.24529880e-01 3.80348742e-01
1.46701490e-03 2.26865709e-01 2.42289603e-01 6.08029477e-02
9.90847126e-02 7.89271891e-01 1.67729354e+00 2.34529704e-01
-2.13793293e-01 -2.44908839e-01 -5.11053026e-01 7.32139111e-01
5.22674263e-01 -1.92249298e-01 -7.19306767e-01 -8.38562846e-01
6.92537948e-02 4.24700439e-01 3.81161541e-01 -7.31770635e-01
-7.57757574e-02 -4.95655060e-01 7.95335829e-01 5.66609427e-02
7.71729469e-01 -5.53128779e-01 5.81153989e-01 1.19106494e-01
-4.88519460e-01 2.65434861e-01 7.64902532e-02 4.41691309e-01
2.39455804e-01 4.65170562e-01 1.07681048e+00 -1.19050533e-01
-2.69763350e-01 6.93588972e-01 -4.08979535e-01 1.47428261e-02
6.32564247e-01 -6.79373071e-02 -3.28064449e-02 -7.37364233e-01
-7.29356945e-01 -1.95248201e-01 7.79025197e-01 5.87410271e-01
4.01292443e-01 -1.55097067e+00 -8.16929936e-01 7.24600911e-01
-1.42923504e-01 -5.12429960e-02 2.77682543e-01 2.27820233e-01
-5.21683574e-01 -3.70190255e-02 -2.44120136e-01 -7.36016929e-01
-1.27877617e+00 1.80856645e-01 5.74802041e-01 4.25004274e-01
-6.43760979e-01 8.95914614e-01 6.39259398e-01 -4.23878700e-01
3.03840581e-02 -5.03395610e-02 9.68845040e-02 -2.27770582e-01
4.13990676e-01 4.49013710e-03 -2.46319219e-01 -9.23242509e-01
-1.54915437e-01 9.57759619e-01 3.71325791e-01 -5.82099319e-01
1.14496851e+00 -4.08827722e-01 6.69131940e-03 1.22681454e-01
1.32734823e+00 6.28473982e-02 -2.11141324e+00 2.15440691e-01
-5.42401910e-01 -5.39005399e-01 3.07569206e-02 -4.43845153e-01
-1.38612819e+00 7.94189274e-01 6.11879945e-01 -3.05557549e-01
1.20062256e+00 -2.21947834e-01 5.13240516e-01 -9.54283029e-02
1.52331620e-01 -8.40985477e-01 1.95065945e-01 4.30208445e-01
9.86325204e-01 -8.05791259e-01 -2.52255082e-01 -4.13221329e-01
-5.06971836e-01 9.55963731e-01 6.92469656e-01 2.46061590e-02
6.87534988e-01 4.59309280e-01 2.50038058e-01 -3.84233706e-02
-9.76405859e-01 2.62192190e-01 3.57259661e-01 6.78857744e-01
5.45020461e-01 -4.50556129e-02 3.63530546e-01 -4.53113131e-02
-4.54381973e-01 -2.33293772e-01 3.10849220e-01 5.61797142e-01
-5.07701449e-02 -9.80304480e-01 -2.40965694e-01 -1.35363683e-01
-2.84125924e-01 1.48261815e-01 -1.80649251e-01 5.72471201e-01
3.01449060e-01 5.81420720e-01 4.37551469e-01 -2.18068913e-01
2.14487746e-01 9.24667791e-02 8.64873528e-01 -5.97765803e-01
-2.33502328e-01 3.99285525e-01 -5.94787560e-02 -5.95758080e-01
-2.94070125e-01 -7.84871757e-01 -9.62111115e-01 -4.81368035e-01
8.80902037e-02 -3.74754697e-01 8.03213894e-01 5.79960585e-01
5.22796810e-01 4.28252369e-01 9.60012317e-01 -9.63472486e-01
-1.73577145e-01 -5.05853653e-01 -5.63593030e-01 4.81277704e-01
5.40973723e-01 -5.11750340e-01 -4.11272466e-01 7.34926820e-01] | [12.72714900970459, -0.3902129828929901] |
d693c095-8ed8-4e7c-9914-e8ccf5f87020 | context-enhanced-stereo-transformer | 2210.11719 | null | https://arxiv.org/abs/2210.11719v1 | https://arxiv.org/pdf/2210.11719v1.pdf | Context-Enhanced Stereo Transformer | Stereo depth estimation is of great interest for computer vision research. However, existing methods struggles to generalize and predict reliably in hazardous regions, such as large uniform regions. To overcome these limitations, we propose Context Enhanced Path (CEP). CEP improves the generalization and robustness against common failure cases in existing solutions by capturing the long-range global information. We construct our stereo depth estimation model, Context Enhanced Stereo Transformer (CSTR), by plugging CEP into the state-of-the-art stereo depth estimation method Stereo Transformer. CSTR is examined on distinct public datasets, such as Scene Flow, Middlebury-2014, KITTI-2015, and MPI-Sintel. We find CSTR outperforms prior approaches by a large margin. For example, in the zero-shot synthetic-to-real setting, CSTR outperforms the best competing approaches on Middlebury-2014 dataset by 11%. Our extensive experiments demonstrate that the long-range information is critical for stereo matching task and CEP successfully captures such information. | ['Yingwei Li', 'Alan Yuille', 'Mathias Unberath', 'Russell H. Taylor', 'Zheng Wang', 'Yongkui Yang', 'Zhaoshuo Li', 'Weiyu Guo'] | 2022-10-21 | null | null | null | null | ['stereo-depth-estimation', 'stereo-matching-1'] | ['computer-vision', 'computer-vision'] | [ 1.77606493e-01 -4.32807148e-01 -2.61132587e-02 -3.44157159e-01
-9.17404830e-01 -5.01172006e-01 8.74628067e-01 -4.05859888e-01
-3.36971074e-01 8.06865513e-01 6.19013011e-01 -1.60590902e-01
2.57666886e-01 -7.85127521e-01 -5.84110916e-01 -4.55904305e-01
9.81923789e-02 2.30349913e-01 6.83741629e-01 -2.42955968e-01
6.30705774e-01 4.83236104e-01 -1.75433695e+00 3.33802670e-01
1.04030657e+00 9.89899993e-01 3.43790799e-01 7.30562389e-01
1.50321960e-01 1.08943033e+00 1.92991120e-05 -2.24018782e-01
7.19681144e-01 -1.60559550e-01 -8.81772280e-01 2.52591148e-02
1.11533380e+00 -8.53686452e-01 -7.94175446e-01 8.01932335e-01
7.14272022e-01 2.23201886e-01 4.87356633e-01 -1.21211267e+00
1.83068722e-01 -2.85783708e-01 -7.94970989e-01 3.47506642e-01
6.37701511e-01 6.07149065e-01 9.10003006e-01 -8.83269370e-01
1.06180477e+00 1.33524621e+00 8.01021814e-01 4.01526511e-01
-1.10156620e+00 -6.94707811e-01 8.79112631e-02 3.10774297e-01
-1.18005669e+00 -5.25380969e-01 5.33928335e-01 -7.60275662e-01
1.26999843e+00 -5.42347804e-02 4.65923697e-01 1.27967083e+00
8.80732313e-02 7.93508232e-01 1.28300750e+00 2.69780699e-02
1.39218777e-01 -4.94506031e-01 -1.44421652e-01 5.56516826e-01
2.77970284e-02 7.68418610e-01 -9.62699473e-01 8.11314397e-03
8.88010859e-01 -1.95678353e-01 -5.39817631e-01 -4.35541540e-01
-1.36940563e+00 6.08954012e-01 3.92709762e-01 -3.46094280e-01
-2.46682495e-01 6.57145679e-02 4.58702207e-01 1.70876831e-01
5.91476381e-01 1.84878647e-01 -4.05173093e-01 -5.31212807e-01
-1.04164016e+00 5.09372473e-01 6.48659408e-01 1.08697617e+00
1.00391126e+00 -1.21400803e-01 -1.31177098e-01 7.78577507e-01
9.36615542e-02 5.92155516e-01 3.47433500e-02 -1.56384039e+00
9.43892241e-01 2.82384664e-01 -1.05181271e-02 -5.98783612e-01
-1.90812945e-01 -2.93458849e-01 -8.00352216e-01 3.79678190e-01
5.78024507e-01 -7.86318779e-02 -9.16112065e-01 1.53096557e+00
3.34903896e-01 6.62152648e-01 5.21657057e-02 9.99340057e-01
9.46134031e-01 6.35054946e-01 -1.90516233e-01 2.16152728e-01
7.20733643e-01 -1.10593474e+00 -1.10218868e-01 -7.41640806e-01
4.88504350e-01 -8.25296938e-01 7.46870220e-01 3.88714939e-01
-8.53366852e-01 -5.08307099e-01 -8.65756452e-01 -2.93175727e-01
7.61970207e-02 -4.69543010e-01 6.49666488e-01 2.31828898e-01
-1.14411223e+00 6.35758400e-01 -6.17609143e-01 -4.25704807e-01
4.00458992e-01 -1.31610692e-01 -6.31214440e-01 -8.54723811e-01
-9.46435690e-01 6.81739569e-01 9.54503864e-02 -1.42340779e-01
-1.03847229e+00 -1.21890748e+00 -1.29409778e+00 -2.90803879e-01
2.32854590e-01 -1.03237724e+00 1.20100141e+00 -4.38838154e-01
-1.29686880e+00 1.04156876e+00 -4.44811076e-01 -4.27548438e-01
1.16203129e+00 -3.67873341e-01 -6.29471242e-02 2.70156980e-01
4.29165423e-01 1.13761997e+00 5.78013599e-01 -1.25368917e+00
-9.41053271e-01 -3.03332955e-01 1.04104578e-01 2.93596745e-01
3.29540640e-01 -3.87119710e-01 -7.43738413e-01 -4.03979719e-01
2.87601739e-01 -8.40430140e-01 -2.92632490e-01 2.65118062e-01
-3.36124450e-01 2.54239589e-01 6.88864827e-01 -5.08877218e-01
8.81218135e-01 -2.03622413e+00 -6.34343624e-02 9.76765435e-03
3.10244590e-01 4.68034297e-02 -8.62855837e-02 3.90262991e-01
9.48073789e-02 -5.05716145e-01 -4.29868102e-01 -5.94536781e-01
-1.75119132e-01 2.79622167e-01 -3.21327865e-01 6.41408324e-01
-8.64575356e-02 6.75980568e-01 -1.12506044e+00 -6.27412379e-01
8.32146108e-01 4.09757674e-01 -1.04850221e+00 2.82904476e-01
-1.32128270e-02 7.08957851e-01 -1.21443272e-01 8.13725412e-01
9.99270022e-01 -1.64417565e-01 -2.04714894e-01 -1.12032272e-01
-3.34921867e-01 4.00561661e-01 -1.16572809e+00 2.13538766e+00
-4.58424568e-01 1.06787896e+00 -2.57451653e-01 -5.10842204e-01
7.08663285e-01 -4.81338762e-02 5.38063407e-01 -1.12869179e+00
-2.93775022e-01 3.52085620e-01 -3.03660005e-01 -4.39590663e-01
7.85938263e-01 -7.68617988e-02 1.93050086e-01 -1.37750089e-01
-1.77044898e-01 -3.91625613e-01 1.79405957e-01 4.16896790e-01
1.42509913e+00 2.64604598e-01 2.52460301e-01 -2.24597842e-01
5.84888577e-01 1.76646262e-02 8.29962611e-01 7.85731614e-01
-5.70793509e-01 1.29294658e+00 3.43117267e-01 -2.21341610e-01
-1.09690213e+00 -1.24828124e+00 -2.62790501e-01 4.49345738e-01
5.53752780e-01 -4.82501656e-01 -2.40042940e-01 -5.84598124e-01
2.47750729e-01 4.41864312e-01 -4.56153512e-01 1.23556666e-01
-5.43905795e-01 -2.20723823e-01 3.82916361e-01 7.63785303e-01
1.05744278e+00 -8.89049709e-01 -5.92584014e-01 2.57272005e-01
-5.79763830e-01 -1.71044719e+00 -4.86588329e-01 -3.10414910e-01
-8.87462258e-01 -1.20036566e+00 -8.49362552e-01 -3.55444700e-01
1.40267178e-01 7.17712820e-01 1.50227225e+00 -1.89268157e-01
-3.12277436e-01 4.67340618e-01 -1.84683248e-01 1.83801606e-01
-1.15075499e-01 -3.24831903e-02 -4.29425269e-01 -1.92695320e-01
2.00171009e-01 -6.62082732e-01 -1.16585827e+00 5.45129538e-01
-5.75629830e-01 4.27387595e-01 2.26357102e-01 7.94782698e-01
3.90308201e-01 -4.81295168e-01 6.59315214e-02 -7.33782828e-01
-6.04074076e-02 -5.05021691e-01 -7.81856358e-01 -3.48827749e-01
-4.69884366e-01 -4.66633029e-02 2.80907780e-01 1.36727303e-01
-1.39727247e+00 -4.54721488e-02 -2.94112056e-01 -5.20523667e-01
-2.12543130e-01 6.94994703e-02 3.13921683e-02 -1.90831125e-01
7.18784213e-01 2.50414491e-01 -2.72056431e-01 -3.31016243e-01
2.37685628e-02 3.42079639e-01 9.69407678e-01 -6.98834777e-01
6.24617100e-01 1.04012537e+00 2.92109519e-01 -8.51401448e-01
-7.46729136e-01 -9.80355322e-01 -3.87209713e-01 -3.08366477e-01
6.81231260e-01 -1.44393969e+00 -4.34337258e-01 8.47994924e-01
-1.09392512e+00 -6.60046160e-01 1.30495513e-02 4.68341440e-01
-8.22025180e-01 6.78765774e-01 -7.14611530e-01 -5.23582339e-01
-1.79804906e-01 -1.29506683e+00 1.58341634e+00 -3.88608240e-02
-2.41303116e-01 -1.08357465e+00 3.49169314e-01 6.34551406e-01
3.69769692e-01 4.34601277e-01 3.39967340e-01 2.20559955e-01
-1.10221779e+00 2.11530775e-01 -7.12232172e-01 3.86202544e-01
-1.53688118e-01 -1.72372535e-02 -1.31425309e+00 -3.52248520e-01
-4.39949393e-01 -3.60814393e-01 1.35803282e+00 5.42466342e-01
1.09133494e+00 4.21985686e-01 -1.96365803e-01 1.26414585e+00
1.68918610e+00 -8.36904496e-02 1.22451198e+00 6.50778353e-01
9.39656854e-01 5.96201658e-01 9.26918447e-01 6.22101367e-01
7.43821621e-01 6.16669536e-01 5.42869151e-01 -1.05540708e-01
-4.10679072e-01 -3.30271363e-01 2.34449774e-01 2.57021934e-01
-2.39621535e-01 -1.32582337e-01 -1.11708069e+00 7.18755364e-01
-1.88806474e+00 -1.12880039e+00 -2.26387098e-01 2.18249679e+00
5.33908427e-01 3.14234078e-01 -2.61295259e-01 4.87643406e-02
3.88298959e-01 5.10069132e-01 -4.68966156e-01 1.68335184e-01
-3.83116782e-01 1.01853363e-01 7.32771277e-01 7.78900743e-01
-1.06196606e+00 1.18481970e+00 5.77879906e+00 7.56837845e-01
-1.19031084e+00 -9.63579565e-02 5.15749931e-01 -4.07380834e-02
-2.28069931e-01 2.44104102e-01 -7.51128256e-01 4.46381658e-01
4.00663376e-01 3.44651863e-02 2.34951168e-01 7.12088704e-01
3.28443080e-01 -6.14323318e-01 -1.27005351e+00 1.49678445e+00
-8.80813226e-02 -1.53953147e+00 -2.20781356e-01 2.15433210e-01
1.32747352e+00 6.34465039e-01 -1.41600832e-01 2.11731508e-01
3.31345558e-01 -9.40718353e-01 5.85207164e-01 4.28849518e-01
9.82160270e-01 -4.49644446e-01 6.46317065e-01 2.05526933e-01
-1.33188128e+00 2.45159879e-01 -1.92286119e-01 -1.20044723e-01
3.89665931e-01 6.93666339e-01 -4.58089501e-01 7.23080397e-01
7.64265418e-01 1.59149230e+00 -5.41810632e-01 1.40524304e+00
-1.78506091e-01 3.80353391e-01 -3.23927641e-01 8.13061833e-01
5.16743004e-01 -2.88339294e-02 6.44401610e-01 1.19651282e+00
3.23340058e-01 6.66255802e-02 3.29487100e-02 6.20849609e-01
7.75492284e-03 -3.99688572e-01 -8.53367507e-01 6.71208918e-01
4.97597218e-01 8.27409267e-01 -3.43380004e-01 -5.32070398e-01
-5.47579587e-01 1.09397292e+00 1.63367897e-01 4.21057403e-01
-7.84951687e-01 1.26587879e-03 1.18816316e+00 3.03597063e-01
2.04309866e-01 -1.96478203e-01 -2.75288910e-01 -1.53841043e+00
1.40109584e-01 -8.54392886e-01 3.48658413e-01 -9.50713396e-01
-1.23843086e+00 4.15580422e-01 8.27678218e-02 -1.63347673e+00
-4.57015812e-01 -5.43083489e-01 -5.13529539e-01 9.07309413e-01
-2.10916567e+00 -8.63051534e-01 -1.03906810e+00 9.10575747e-01
8.38812113e-01 3.03440820e-02 2.57162184e-01 4.70413774e-01
-2.07740307e-01 1.80847034e-01 -6.97543621e-02 -4.75512818e-02
1.08902621e+00 -1.11001384e+00 8.54435742e-01 1.12000513e+00
-3.28851014e-01 5.00376895e-02 7.27762043e-01 -5.74331403e-01
-1.13335407e+00 -1.15618122e+00 8.32312584e-01 -4.80403543e-01
3.75123918e-01 -1.27008826e-01 -7.88244486e-01 5.62988341e-01
-3.85359265e-02 4.35578167e-01 -1.53397873e-01 -2.29758546e-01
-5.90611994e-01 -1.23168297e-01 -1.14087892e+00 7.33208477e-01
1.75436687e+00 -7.64357567e-01 -3.67823839e-01 6.56422749e-02
4.88989115e-01 -8.33192766e-01 -5.84002852e-01 7.08405435e-01
5.98060429e-01 -1.91780889e+00 1.35897243e+00 -7.53326863e-02
1.06366861e+00 -2.09411696e-01 -4.73073125e-01 -1.24985063e+00
-3.80170979e-02 -6.25919461e-01 -2.22277008e-02 9.27491188e-01
-2.51609888e-02 -8.47598851e-01 1.00332689e+00 4.60239977e-01
-3.97685051e-01 -4.43514198e-01 -1.06285322e+00 -9.12970841e-01
4.87560816e-02 -7.76492655e-01 3.92801583e-01 8.37134063e-01
-3.48283380e-01 1.68393061e-01 -3.97935212e-01 5.00289574e-02
1.07723618e+00 8.76111761e-02 1.21996760e+00 -9.96768236e-01
-3.41448367e-01 -3.01720232e-01 -8.02385271e-01 -1.69285798e+00
2.10014343e-01 -4.88289356e-01 4.79948819e-02 -1.58781326e+00
2.04480276e-01 -5.13183177e-01 8.58031884e-02 -7.43507221e-02
-2.42884263e-01 2.75437623e-01 1.24725565e-01 8.18252936e-02
-5.40235519e-01 6.64902985e-01 1.46601260e+00 -5.65978400e-02
1.50407791e-01 -1.91897765e-01 -1.74467206e-01 7.58649290e-01
5.08337498e-01 -1.35370359e-01 -4.60624874e-01 -5.09955108e-01
-5.74784987e-02 4.81997222e-01 7.15963900e-01 -1.34082627e+00
1.80518582e-01 -2.07183436e-01 1.68664873e-01 -8.90935540e-01
5.94844818e-01 -6.46396101e-01 8.66589025e-02 2.78414220e-01
2.60301959e-02 -5.88976126e-03 1.73747733e-01 5.34165144e-01
-4.34767038e-01 3.15112293e-01 9.26002085e-01 1.54396454e-02
-1.48930669e+00 6.59035563e-01 9.13131461e-02 7.12251663e-01
7.96535969e-01 -6.02867067e-01 -6.94184661e-01 -5.59392929e-01
-5.54056205e-02 4.36403304e-01 8.26475263e-01 4.41324979e-01
7.75449216e-01 -1.09828591e+00 -7.49578476e-01 3.64967018e-01
6.56184971e-01 3.61391455e-01 4.90777165e-01 8.26704979e-01
-1.02735519e+00 3.66428941e-01 -2.57005095e-01 -1.11802590e+00
-1.11136127e+00 1.39564827e-01 4.82478142e-01 -1.68058783e-01
-1.10644162e+00 7.99336970e-01 6.85420930e-01 -2.39143699e-01
1.31269425e-01 -4.44629431e-01 2.93278992e-01 -4.51201051e-01
4.44976300e-01 5.59643209e-01 1.48662090e-01 -6.13035083e-01
-3.69198859e-01 8.24458897e-01 1.08446129e-01 -1.74839854e-01
1.07710528e+00 -3.89097571e-01 2.26987138e-01 -3.58501486e-02
1.43143594e+00 -1.08759195e-01 -2.07471418e+00 -4.40105855e-01
-3.76026809e-01 -1.26298189e+00 2.41830036e-01 -5.26835322e-01
-1.16133940e+00 1.00226593e+00 4.39963728e-01 -5.77554822e-01
1.02865887e+00 -1.73733115e-01 1.00160766e+00 8.01842883e-02
7.86658466e-01 -7.11600840e-01 3.23869810e-02 8.44951153e-01
6.65386617e-01 -1.81537056e+00 -1.13902286e-01 -6.71830654e-01
-6.36219025e-01 9.45497930e-01 8.81535053e-01 -1.23034909e-01
5.32380998e-01 2.71467090e-01 -1.84653501e-03 -1.12461947e-01
-8.67575169e-01 -4.08671945e-01 2.15436965e-01 6.66878283e-01
1.72270253e-01 -2.37779796e-01 3.37198764e-01 -3.17378342e-01
-2.49995351e-01 1.18066609e-01 4.85074103e-01 1.02821183e+00
-2.19046846e-01 -7.91531801e-01 -2.29765847e-01 1.36534840e-01
-1.25801107e-02 -3.99455309e-01 2.16000751e-02 8.72276664e-01
-7.75744393e-02 9.22733963e-01 1.95439190e-01 -3.71760219e-01
4.05872077e-01 -5.30402124e-01 6.22982502e-01 -3.18417728e-01
-1.96763054e-01 -1.11136347e-01 3.08645308e-01 -1.23917925e+00
-4.12752509e-01 -7.76057005e-01 -9.06851053e-01 -8.55212271e-01
2.84292072e-01 -5.46748757e-01 2.08124265e-01 9.02127087e-01
3.28124672e-01 3.59051287e-01 5.09794891e-01 -1.19437909e+00
-1.81410670e-01 -6.94258153e-01 -3.72212261e-01 7.36906469e-01
6.24958634e-01 -8.71573031e-01 -6.45019531e-01 -1.61142230e-01] | [8.618856430053711, -2.1087024211883545] |
52e88934-8b57-455e-9e18-d1d1b3912f47 | partially-relevant-video-retrieval | 2208.12510 | null | https://arxiv.org/abs/2208.12510v1 | https://arxiv.org/pdf/2208.12510v1.pdf | Partially Relevant Video Retrieval | Current methods for text-to-video retrieval (T2VR) are trained and tested on video-captioning oriented datasets such as MSVD, MSR-VTT and VATEX. A key property of these datasets is that videos are assumed to be temporally pre-trimmed with short duration, whilst the provided captions well describe the gist of the video content. Consequently, for a given paired video and caption, the video is supposed to be fully relevant to the caption. In reality, however, as queries are not known a priori, pre-trimmed video clips may not contain sufficient content to fully meet the query. This suggests a gap between the literature and the real world. To fill the gap, we propose in this paper a novel T2VR subtask termed Partially Relevant Video Retrieval (PRVR). An untrimmed video is considered to be partially relevant w.r.t. a given textual query if it contains a moment relevant to the query. PRVR aims to retrieve such partially relevant videos from a large collection of untrimmed videos. PRVR differs from single video moment retrieval and video corpus moment retrieval, as the latter two are to retrieve moments rather than untrimmed videos. We formulate PRVR as a multiple instance learning (MIL) problem, where a video is simultaneously viewed as a bag of video clips and a bag of video frames. Clips and frames represent video content at different time scales. We propose a Multi-Scale Similarity Learning (MS-SL) network that jointly learns clip-scale and frame-scale similarities for PRVR. Extensive experiments on three datasets (TVR, ActivityNet Captions, and Charades-STA) demonstrate the viability of the proposed method. We also show that our method can be used for improving video corpus moment retrieval. | ['Xun Wang', 'Xirong Li', 'ShuJie Chen', 'Xun Yang', 'Minsong Zhang', 'Xianke Chen', 'Jianfeng Dong'] | 2022-08-26 | null | null | null | null | ['moment-retrieval', 'partially-relevant-video-retrieval'] | ['computer-vision', 'computer-vision'] | [ 3.64316642e-01 -4.48728979e-01 -6.71551108e-01 -2.14311853e-01
-1.43514884e+00 -7.19128132e-01 5.61543643e-01 -4.50915471e-02
-2.77068436e-01 5.51992238e-01 3.65329325e-01 1.16266727e-01
-1.79871514e-01 -2.10486934e-01 -1.09540534e+00 -4.90592420e-01
-2.38895372e-01 3.54520470e-01 2.11684704e-01 -7.83844069e-02
2.43495420e-01 3.91234189e-01 -1.67899716e+00 7.26528823e-01
2.86215335e-01 1.16155088e+00 5.33736289e-01 8.37471664e-01
4.75024357e-02 1.06510878e+00 -6.20812774e-01 -1.94060951e-01
3.12878370e-01 -3.79095584e-01 -8.11781764e-01 3.18035573e-01
7.75172174e-01 -5.08733630e-01 -9.46990490e-01 7.27939785e-01
2.65873730e-01 4.88494754e-01 7.10732996e-01 -1.61358798e+00
-6.30729556e-01 4.15585726e-01 -5.15035391e-01 7.00096548e-01
9.12921965e-01 -2.45471388e-01 1.06450343e+00 -8.72686327e-01
1.07169819e+00 1.07387388e+00 2.80522294e-02 5.26799917e-01
-6.31393611e-01 -4.74378705e-01 3.80393982e-01 6.53011143e-01
-1.51527607e+00 -5.20890415e-01 8.62790704e-01 -4.45674896e-01
6.34762228e-01 4.67559487e-01 4.74922001e-01 1.32707322e+00
8.54816381e-03 1.14206064e+00 3.62844288e-01 -1.41929433e-01
9.05884281e-02 -6.40137643e-02 -2.20007941e-01 3.25983882e-01
-2.11274937e-01 -3.17054123e-01 -6.56799734e-01 -7.68660083e-02
7.04589665e-01 1.44276679e-01 -5.09402514e-01 -3.56990159e-01
-1.49238884e+00 6.86884999e-01 -4.49909233e-02 4.20159668e-01
-3.63923162e-01 1.16716631e-01 7.92980671e-01 4.88266855e-01
4.36869651e-01 5.77657223e-02 -2.28008896e-01 -2.27794841e-01
-1.23697186e+00 2.61043161e-01 6.05643332e-01 1.26945722e+00
6.04740202e-01 3.07770129e-02 -4.60313588e-01 7.10543513e-01
9.58164260e-02 7.04278052e-01 6.46215916e-01 -1.08558178e+00
8.68345678e-01 -1.10929012e-02 3.05865318e-01 -1.27312040e+00
3.04183811e-01 1.82817712e-01 -4.78678733e-01 -7.24484503e-01
1.29280001e-01 4.10004169e-01 -8.96773398e-01 1.63992441e+00
6.77648634e-02 7.25114882e-01 2.64297694e-01 1.34085047e+00
9.66521084e-01 1.32106197e+00 -5.30799106e-02 -5.91528416e-01
1.15806508e+00 -8.83439958e-01 -7.53583312e-01 -1.60286918e-01
2.73872137e-01 -8.07651103e-01 7.77681172e-01 3.02664548e-01
-1.18704021e+00 -5.62459052e-01 -8.77882659e-01 -6.57544881e-02
-2.21576601e-01 9.91989393e-03 1.31541878e-01 -3.78352702e-02
-1.03237331e+00 3.46291333e-01 -4.00095016e-01 -2.86175698e-01
6.32924214e-02 6.56954497e-02 -6.88601673e-01 -4.17132944e-01
-1.33295977e+00 5.58415234e-01 6.24841332e-01 1.00488262e-02
-1.36930847e+00 -5.16245604e-01 -9.87568974e-01 -1.18138436e-02
6.81081355e-01 -2.79683799e-01 1.29279089e+00 -1.21376991e+00
-8.31578732e-01 8.65621150e-01 -4.46437389e-01 -5.74284911e-01
4.23754185e-01 -3.09385687e-01 -7.40294278e-01 1.12188828e+00
1.72886029e-01 6.76387191e-01 1.32648349e+00 -1.23884904e+00
-7.10178375e-01 9.65424851e-02 3.00039738e-01 4.34588730e-01
-3.35112751e-01 4.55010802e-01 -1.26561689e+00 -8.07252884e-01
-1.12369567e-01 -8.39802146e-01 2.71615028e-01 -2.16747776e-01
-8.02144315e-03 -1.28045321e-01 1.22866154e+00 -1.01999307e+00
1.26800561e+00 -2.10280561e+00 3.71819615e-01 3.76303936e-03
-2.53334176e-02 2.12939173e-01 -6.55558884e-01 6.04501247e-01
-2.27794871e-01 5.45997582e-02 2.33697638e-01 -1.55304044e-01
-1.30495772e-01 3.39375883e-01 -7.76662469e-01 5.34455121e-01
-4.26105633e-02 9.01274502e-01 -1.14931202e+00 -9.83383000e-01
3.54266256e-01 6.21265590e-01 -1.78248867e-01 3.01822484e-01
-3.68354887e-01 2.57451445e-01 -5.00897765e-01 8.76070857e-01
2.72789389e-01 -1.73035175e-01 -8.90942439e-02 -5.25850773e-01
1.39234364e-01 -2.65972674e-01 -9.35221136e-01 1.76230156e+00
-2.89435744e-01 1.05650818e+00 -1.93968773e-01 -1.12125063e+00
6.07061863e-01 7.96435058e-01 9.74989355e-01 -8.82963836e-01
3.21525522e-02 1.81482881e-02 -5.87425411e-01 -9.32367086e-01
9.09626663e-01 2.30701879e-01 -1.77091017e-01 3.47111791e-01
6.48507103e-02 2.45405480e-01 5.86752415e-01 6.62729323e-01
9.34775651e-01 1.27265289e-01 8.73224139e-02 4.49163556e-01
5.12652874e-01 -1.39172226e-01 5.00632346e-01 7.34703660e-01
-3.52519572e-01 9.66665208e-01 4.22441155e-01 -1.02698036e-01
-1.18504417e+00 -9.53094125e-01 2.51166314e-01 1.28829300e+00
4.44775552e-01 -4.69670415e-01 -4.64298546e-01 -6.20463431e-01
-3.01960170e-01 2.56839514e-01 -4.57403362e-01 -5.95998615e-02
-7.69030035e-01 -1.27493128e-01 3.46578389e-01 3.55682045e-01
9.51224416e-02 -1.17707825e+00 -4.68362719e-01 9.98057872e-02
-9.36752677e-01 -1.64604270e+00 -1.10778844e+00 -5.30177534e-01
-6.93025768e-01 -1.13028836e+00 -1.03046083e+00 -9.67347205e-01
5.27361035e-01 8.31371248e-01 1.11507988e+00 -2.97612622e-02
-9.18059349e-02 9.43185627e-01 -9.38021183e-01 2.59227931e-01
-3.65915209e-01 -3.35018486e-01 2.29631379e-01 2.59352654e-01
2.06667706e-01 -2.71221280e-01 -4.24925566e-01 4.31431979e-01
-1.44150507e+00 -2.20060758e-02 5.41700721e-01 5.83481014e-01
9.39601839e-01 -1.04690762e-03 5.59641898e-01 -3.21739703e-01
3.49933714e-01 -8.61370265e-01 -2.70517498e-01 6.36542380e-01
5.93597330e-02 -3.10045660e-01 5.90802670e-01 -8.75178635e-01
-7.44822323e-01 -3.73370908e-02 3.67591947e-01 -1.44706953e+00
-6.01826236e-02 6.27214134e-01 -1.16112337e-01 1.04145825e-01
1.51936397e-01 5.55461287e-01 -3.45227271e-01 -2.34632686e-01
2.34355807e-01 6.39362276e-01 8.64344418e-01 -6.51298165e-01
9.66981471e-01 4.13107008e-01 -2.93302447e-01 -9.31634665e-01
-7.07970560e-01 -1.08119226e+00 -4.59427923e-01 -6.37219489e-01
7.85267889e-01 -1.18456686e+00 -4.85863477e-01 -5.07072322e-02
-1.10339355e+00 -6.13728119e-03 -3.00480742e-02 6.51900172e-01
-7.84411132e-01 7.69013882e-01 -5.56774497e-01 -6.54405177e-01
-3.55538994e-01 -1.09444690e+00 1.37295496e+00 4.55817953e-02
2.26210505e-02 -9.05241251e-01 -1.14747338e-01 5.88936865e-01
1.48281660e-02 2.87608594e-01 5.24906516e-01 -8.28755498e-01
-7.26682246e-01 -5.16267776e-01 -2.31483653e-01 3.54879588e-01
8.93743113e-02 1.27226830e-01 -7.14025021e-01 -5.88413060e-01
-1.43998429e-01 -4.53476578e-01 7.45496154e-01 4.65436965e-01
1.19356024e+00 -5.15716553e-01 -2.67970502e-01 3.13314825e-01
1.40555954e+00 5.34993708e-01 7.23264217e-01 2.47914523e-01
6.53713584e-01 3.85272056e-01 1.10760868e+00 5.15138268e-01
2.22515360e-01 8.69722307e-01 3.89623106e-01 2.09517285e-01
5.15548959e-02 -3.69779140e-01 8.14449489e-01 1.03201830e+00
2.92174090e-02 -5.87902486e-01 -5.19595206e-01 9.11732376e-01
-2.07246780e+00 -1.47815657e+00 2.60269940e-01 2.17688060e+00
4.71726626e-01 -2.31271893e-01 9.97919440e-02 -5.18398024e-02
9.70414698e-01 4.46932167e-01 -6.54136598e-01 -1.22522935e-01
-2.40717828e-01 -3.41150343e-01 1.30853593e-01 1.54558718e-01
-1.15078533e+00 7.10402131e-01 5.33527088e+00 1.11279488e+00
-1.01103175e+00 1.32271945e-01 3.25506657e-01 -3.32472801e-01
-1.00603387e-01 -1.62060261e-02 -4.80855376e-01 5.73589742e-01
1.01286185e+00 -4.43114161e-01 4.16076243e-01 8.03189993e-01
4.55341101e-01 3.17116976e-02 -1.43736434e+00 1.46865833e+00
7.27912426e-01 -1.26216435e+00 5.02038777e-01 -2.41633415e-01
7.07880020e-01 -3.06106191e-02 2.32664689e-01 4.67752367e-01
-5.46976328e-01 -7.09221721e-01 9.99273777e-01 5.54083645e-01
9.41630483e-01 -7.38693058e-01 6.59673274e-01 1.09653570e-01
-1.56844103e+00 -3.30240689e-02 -5.24483621e-01 4.62425888e-01
4.08433497e-01 1.38595119e-01 -7.04658329e-01 8.47609043e-01
7.37245619e-01 1.05595267e+00 -5.15225410e-01 1.11067033e+00
2.04115868e-01 2.82975763e-01 -6.36754930e-02 4.05023336e-01
2.97471613e-01 -1.48074046e-01 7.32524037e-01 1.27454197e+00
4.22460794e-01 1.45034358e-01 3.54070008e-01 3.16737890e-01
-2.97127008e-01 2.54783332e-01 -7.35034287e-01 -3.30876321e-01
5.11438608e-01 1.13909698e+00 -7.10848927e-01 -5.21610081e-01
-5.33753574e-01 1.14066052e+00 -2.25023627e-01 6.44704878e-01
-1.10912991e+00 -1.81764662e-01 4.02478516e-01 -1.61857486e-01
4.77035522e-01 5.98405227e-02 8.75807047e-01 -1.61855876e+00
3.29075247e-01 -9.95458186e-01 6.67967558e-01 -1.23048472e+00
-1.23874545e+00 6.06430948e-01 3.81668091e-01 -1.78092754e+00
-4.67792183e-01 -1.55441806e-01 -2.72123039e-01 3.04523885e-01
-1.56347871e+00 -1.12358224e+00 -2.17607439e-01 9.55187857e-01
1.11342072e+00 -1.77761436e-01 3.05615038e-01 5.25090337e-01
-3.29544485e-01 4.11818296e-01 2.26356253e-01 2.97334284e-01
8.81233692e-01 -7.87547886e-01 3.98391597e-02 9.63353276e-01
3.81883949e-01 3.62068474e-01 8.06688726e-01 -7.72211850e-01
-1.80257118e+00 -1.34661448e+00 6.58053458e-01 -4.45970416e-01
7.39879429e-01 -1.71213955e-01 -8.90805602e-01 8.72490287e-01
9.16970894e-02 1.35024086e-01 4.58335966e-01 -5.40031731e-01
-4.10821497e-01 -1.39269605e-01 -8.07827175e-01 5.24506927e-01
7.56698430e-01 -9.45397258e-01 -9.04277623e-01 6.28354192e-01
9.20891464e-01 -4.54055279e-01 -9.24542308e-01 3.53111804e-01
4.86764491e-01 -5.00373542e-01 1.27108955e+00 -5.72191000e-01
4.75412697e-01 -3.08261335e-01 -5.84753633e-01 -8.75526667e-01
3.28125149e-01 -6.66345775e-01 -6.31575942e-01 1.24697363e+00
-7.56134242e-02 1.84457660e-01 5.02366960e-01 4.26110566e-01
-2.43913606e-01 -2.89777845e-01 -1.05890882e+00 -1.10485017e+00
-4.30735767e-01 -7.19154596e-01 3.28564137e-01 9.50228512e-01
1.00008532e-01 7.48975053e-02 -9.85243738e-01 2.47129813e-01
4.70242798e-01 2.43733570e-01 5.57435393e-01 -8.35124969e-01
-1.14100330e-01 -8.19199011e-02 -4.60568249e-01 -1.12583172e+00
4.22266930e-01 -7.93257892e-01 2.35588357e-01 -1.44332886e+00
5.68854153e-01 1.34870216e-01 -3.30080032e-01 2.46731192e-01
1.54525533e-01 3.55701268e-01 4.13653046e-01 5.15421927e-01
-1.26838839e+00 3.47994506e-01 1.18873942e+00 -3.56461644e-01
-2.12299600e-01 -1.09146908e-01 -1.40880942e-01 3.64409775e-01
4.63081181e-01 -3.11809868e-01 -7.14226902e-01 -3.29902083e-01
2.25313395e-01 8.39403212e-01 4.19866979e-01 -7.93523788e-01
1.90457702e-01 -2.94117212e-01 1.84877127e-01 -8.83988917e-01
6.84096992e-01 -8.77866685e-01 3.21380764e-01 -1.04870528e-01
-5.60638845e-01 1.84279174e-01 9.41137224e-02 8.49909425e-01
-5.93180537e-01 -2.21348017e-01 2.57494450e-01 -1.33844819e-02
-1.16907942e+00 6.72471344e-01 -3.02869081e-01 1.72707066e-01
1.12338567e+00 -3.55137736e-01 -4.04999852e-01 -8.12313855e-01
-6.33639395e-01 2.91319221e-01 3.95777702e-01 8.83263052e-01
1.22924304e+00 -1.43295681e+00 -6.90334380e-01 -3.14359218e-01
4.96551961e-01 -3.90489161e-01 6.54068112e-01 6.69156253e-01
-3.16150486e-01 6.99024200e-01 2.31471919e-02 -6.59567058e-01
-1.40861511e+00 1.11567390e+00 -1.32724062e-01 5.52810803e-02
-7.32967198e-01 3.32905471e-01 2.25532204e-01 3.66680652e-01
4.71078902e-01 -2.55056977e-01 -3.60705823e-01 3.05906057e-01
8.10429096e-01 1.63079277e-01 -1.24940343e-01 -1.33025956e+00
-1.53657168e-01 6.46557987e-01 -2.93440163e-01 2.87672803e-02
1.20644367e+00 -3.66647393e-01 1.77369237e-01 4.59773928e-01
1.64287996e+00 -3.01658452e-01 -1.22450507e+00 -3.69307637e-01
-1.72395647e-01 -6.28656447e-01 -1.26971558e-01 -2.92533368e-01
-1.17371798e+00 4.53541249e-01 4.03634369e-01 -2.70894319e-02
1.22924244e+00 2.45162740e-01 1.12718070e+00 5.69002748e-01
3.21727037e-01 -1.18834317e+00 5.11608720e-01 2.44275346e-01
1.00688374e+00 -1.24375844e+00 -7.85796195e-02 1.51549419e-02
-8.35768759e-01 1.01552391e+00 3.96162540e-01 6.77858442e-02
1.96220860e-01 -2.55076617e-01 -2.36380249e-01 1.14629800e-02
-9.81715977e-01 -1.56735376e-01 6.38713241e-01 4.74423081e-01
-1.86556980e-01 -3.12412083e-01 3.83998081e-02 3.89244646e-01
2.57372975e-01 6.08773939e-02 7.47185230e-01 1.05385172e+00
-3.54642957e-01 -6.87440634e-01 -6.75259471e-01 3.16666663e-01
-5.74325621e-01 8.19997936e-02 -1.14748456e-01 7.91432679e-01
-3.55281919e-01 1.03451312e+00 7.52486736e-02 -4.28463638e-01
1.57988936e-01 -4.12757434e-02 3.56740206e-01 -4.49729174e-01
-1.55944228e-01 3.80167723e-01 -2.29263842e-01 -6.86570525e-01
-8.32299769e-01 -6.00137353e-01 -1.02307498e+00 -1.28050625e-01
-9.52352136e-02 4.66805577e-01 3.96772176e-01 1.04950178e+00
8.26550871e-02 2.74420172e-01 6.88984811e-01 -1.04317939e+00
-4.94478978e-02 -6.64882183e-01 -5.53168654e-01 7.49217987e-01
5.47500074e-01 -5.20325124e-01 -4.75163698e-01 6.73325121e-01] | [10.196479797363281, 0.8020896315574646] |
7de596aa-930f-4afe-941f-4faf2abb74f6 | artificial-life-properties-of-directed | 2005.06060 | null | https://arxiv.org/abs/2005.06060v1 | https://arxiv.org/pdf/2005.06060v1.pdf | Artificial life properties of directed interaction combinators vs. chemlambda | We provide a framework for experimentation at https://mbuliga.github.io/quinegraphs/ic-vs-chem.html#icvschem with two artificial chemistries: directed interaction combinators (dirIC, defined in section 2) and chemlambda. We are interested if these chemistries allow for artificial life behaviour: replication, metabolism and death. The main conclusion of these experiments is that graph rewrites systems which allow conflicting rewrites are better than those which don't, as concerns their artificial life properties. This is in contradiction with the search for good graph rewrite systems for decentralized computing, where non-conflicting graph rewrite systems are historically preferred. This continues the artificial chemistry experiments with chemlambda, lambda calculus or interaction combinators, available from the entry page at https://chemlambda.github.io/index.html and described in arXiv:2003.14332. | ['M. Buliga'] | 2020-05-12 | null | null | null | null | ['artificial-life'] | ['miscellaneous'] | [-6.24433815e-01 4.75603998e-01 1.14442788e-01 1.33144394e-01
1.82422072e-01 -1.08956873e+00 1.24046445e+00 2.78435767e-01
-1.42247111e-01 9.53359187e-01 -1.21212490e-02 -7.76397943e-01
-5.89180365e-02 -1.04589093e+00 -7.39686787e-01 -7.47211635e-01
-2.29717016e-01 6.01253688e-01 2.19317421e-01 -5.36371291e-01
2.67285913e-01 5.95783830e-01 -1.40151048e+00 -1.39272451e-01
7.56497145e-01 1.85853660e-01 -6.57780701e-03 1.17018425e+00
7.53030032e-02 8.93131673e-01 -4.44420636e-01 -5.09929240e-01
4.11130548e-01 -6.65774286e-01 -1.02792966e+00 -8.32927227e-01
1.93484630e-02 3.78954917e-01 -4.82340842e-01 1.11790872e+00
2.81881303e-01 -1.84270710e-01 6.95746779e-01 -1.73878920e+00
-6.39907479e-01 9.66247857e-01 -2.17114627e-01 8.26496556e-02
3.48238081e-01 6.46996021e-01 1.16148627e+00 -4.93290573e-01
8.81051838e-01 1.51539147e+00 3.04020286e-01 6.64099097e-01
-1.35869479e+00 -4.56582934e-01 -8.98181200e-02 -1.75337851e-01
-1.42640650e+00 -5.58243692e-01 -1.91037450e-03 -3.78940791e-01
1.24895108e+00 5.30873775e-01 7.76895165e-01 9.34695184e-01
9.25356984e-01 1.92055717e-01 1.08177221e+00 -4.94339138e-01
4.26177055e-01 1.67272761e-01 4.35976952e-01 8.54966164e-01
1.12258697e+00 2.26643547e-01 -3.43066663e-01 -4.69393820e-01
6.83724761e-01 -2.81943172e-01 -2.26153359e-01 -4.77669120e-01
-1.47167373e+00 6.44241512e-01 2.00514924e-02 4.74742889e-01
-9.33556333e-02 8.42037916e-01 3.93572658e-01 9.61506426e-01
1.42979920e-01 9.82402742e-01 -5.29145896e-01 5.20284772e-02
1.02371216e-01 5.44554532e-01 1.65033638e+00 1.30739808e+00
9.09599006e-01 2.38531339e-03 5.40601276e-02 9.34309959e-02
4.92485821e-01 4.99719113e-01 -8.04643556e-02 -1.49661863e+00
-3.50875139e-01 5.04461825e-01 -1.00225851e-01 -5.63612401e-01
-3.61891955e-01 -3.46302055e-02 -6.45889521e-01 3.95278275e-01
6.22289300e-01 -2.32001096e-01 -4.39868391e-01 1.55860460e+00
3.87477249e-01 -2.10038483e-01 4.22991455e-01 6.30083382e-01
8.26034725e-01 7.44659722e-01 -8.25373605e-02 -3.08947414e-01
1.27285099e+00 -9.05139685e-01 -4.83769625e-01 5.16354442e-01
1.22689044e+00 -7.16593981e-01 8.61388862e-01 6.06258869e-01
-1.11920452e+00 1.12916194e-01 -1.02677155e+00 -1.15391247e-01
-7.93446302e-01 -6.63761139e-01 1.29525411e+00 7.18490303e-01
-1.65737414e+00 4.90330279e-01 -8.53052795e-01 -9.52643871e-01
-4.85829651e-01 3.64066988e-01 -8.46705809e-02 1.69093534e-01
-1.22831666e+00 1.00097311e+00 2.22831219e-01 -3.63153338e-01
-1.04401350e+00 -4.98753905e-01 -3.87683094e-01 -1.13108151e-01
4.57385629e-01 -1.01431108e+00 1.33348596e+00 -6.19419634e-01
-1.58989811e+00 7.34331787e-01 3.25728804e-01 -5.30516386e-01
4.70973581e-01 2.75153160e-01 -1.00290626e-01 5.73351793e-03
-2.21772596e-01 2.62592524e-01 3.41588169e-01 -1.20646465e+00
-1.98048472e-01 -2.42076129e-01 6.90920591e-01 1.05623811e-01
1.99669480e-01 -5.08560650e-02 2.81543415e-02 -1.27363294e-01
-4.73963439e-01 -1.21827221e+00 -2.11013868e-01 -3.43982965e-01
-5.08119881e-01 -6.60647392e-01 4.26562250e-01 1.77189574e-01
8.98981214e-01 -1.61717319e+00 3.73884857e-01 3.76241267e-01
7.14784801e-01 -3.99690792e-02 8.55311006e-03 1.20458186e+00
-7.25808516e-02 6.66280687e-01 8.26729760e-02 2.72923440e-01
4.34457272e-01 8.12373459e-02 9.79346335e-02 6.83949530e-01
-3.87673020e-01 7.79228687e-01 -1.03130245e+00 -5.72281897e-01
6.52699843e-02 2.15357706e-01 -6.53126359e-01 -1.96464747e-01
-8.57223332e-01 4.09207314e-01 -5.10883629e-01 6.30230129e-01
4.63082224e-01 -4.39827114e-01 8.29015434e-01 2.24566400e-01
-4.92103338e-01 2.41843984e-01 -9.98288512e-01 1.55448413e+00
-4.77531627e-02 2.25078821e-01 3.94056052e-01 -4.88010705e-01
5.29889286e-01 3.01048636e-01 4.63298440e-01 -6.02731705e-01
2.77312487e-01 3.46570224e-01 4.94993091e-01 -4.99363244e-02
4.56551224e-01 2.30209857e-01 -2.12842762e-01 8.05393159e-01
-6.21318147e-02 -7.92072535e-01 7.10567057e-01 1.10972989e+00
1.77138233e+00 2.22071901e-01 2.25087970e-01 -1.18270624e+00
5.53760052e-01 2.16789797e-01 4.19288009e-01 1.00872433e+00
-2.74687819e-02 2.40329839e-02 9.50497508e-01 -3.67686450e-01
-1.30937874e+00 -1.05372679e+00 9.12009850e-02 9.42609847e-01
3.63245249e-01 -1.31567824e+00 -8.10191512e-01 -2.64890373e-01
5.69341844e-03 8.16751182e-01 -2.63405710e-01 -8.35478827e-02
-2.99655497e-01 -4.71033216e-01 8.11200976e-01 -3.29511881e-01
-2.00113188e-02 -9.92956221e-01 -3.13202798e-01 -3.02936304e-02
4.21096683e-01 -5.68508506e-01 -1.84760779e-01 1.65077165e-01
-4.75654989e-01 -1.14868546e+00 -1.22941643e-01 -2.46443495e-01
4.93124306e-01 2.02908665e-01 1.37309885e+00 9.29542303e-01
-4.61115390e-01 7.99426734e-01 -5.40393114e-01 -4.79990423e-01
-8.98551524e-01 3.00466061e-01 2.83844143e-01 -8.10387731e-01
-5.84742017e-02 -1.00165033e+00 -9.06621635e-01 4.11436930e-02
-8.40884149e-01 1.10984996e-01 2.53841728e-01 4.63568777e-01
-9.59529132e-02 -4.00750011e-01 2.38990337e-01 -1.09409249e+00
7.83546269e-01 -6.09913290e-01 -1.13008535e+00 4.29695427e-01
-1.08297312e+00 4.20809150e-01 8.66999626e-01 2.10736766e-01
-5.54614604e-01 -3.05094481e-01 3.44539344e-01 -2.41454262e-02
8.71303678e-02 3.65665913e-01 5.60182557e-02 8.48400348e-04
9.65786815e-01 -1.73417464e-01 3.17731708e-01 -9.00813937e-02
6.34347320e-01 2.36156851e-01 -4.48292233e-02 -1.34298027e+00
5.31284153e-01 1.58335567e-01 2.30798528e-01 -9.16069508e-01
2.03544721e-01 4.87990975e-02 -1.94725141e-01 -7.19098151e-02
6.86914861e-01 -6.75829768e-01 -1.41687977e+00 3.85374486e-01
-1.37593520e+00 -6.85482442e-01 -3.07069510e-01 2.78535992e-01
-5.97019076e-01 3.90523314e-01 -9.13132727e-01 -7.68075824e-01
-2.74952710e-01 -1.04640484e+00 6.52490675e-01 1.91657111e-01
-4.59992170e-01 -1.01419568e+00 4.88869309e-01 1.18766082e-02
3.27517062e-01 1.79643124e-01 1.18205142e+00 -7.33500719e-01
-1.01243436e+00 2.49940619e-01 2.65084598e-02 -1.28600806e-01
-2.23384708e-01 8.09760869e-01 -5.30828416e-01 -3.32144529e-01
-6.04844868e-01 -9.54749361e-02 2.40258276e-01 9.89426300e-03
5.99448740e-01 -4.99804437e-01 -4.73432273e-01 3.38890851e-01
1.48568082e+00 1.36731043e-01 7.93600917e-01 1.89939648e-01
5.35688698e-01 4.02999908e-01 1.82843894e-01 4.79948699e-01
4.66171473e-01 2.97075182e-01 3.01037192e-01 3.77990723e-01
-7.44520128e-02 1.22682571e-01 4.17190284e-01 1.07184136e+00
-2.62157053e-01 -6.34077311e-01 -1.18652260e+00 3.40380996e-01
-2.10073209e+00 -7.12225378e-01 -7.72241712e-01 2.10118890e+00
8.70681047e-01 -2.56438911e-01 2.39740565e-01 -4.53025579e-01
7.26552367e-01 -9.70581919e-02 -3.92349720e-01 -9.37361181e-01
1.08057663e-01 3.38557333e-01 8.05948079e-01 8.26467037e-01
-2.33318090e-01 1.18105221e+00 5.79758310e+00 6.22836471e-01
-8.02001894e-01 1.71366826e-01 2.45474800e-01 -2.63104830e-02
-7.49566138e-01 8.41053426e-01 -5.07591307e-01 4.14072275e-01
1.64233208e+00 -7.14779854e-01 1.12161267e+00 3.58121753e-01
2.21547425e-01 -3.06806266e-01 -1.20994461e+00 4.15972412e-01
-5.16774535e-01 -1.39798069e+00 -2.33442232e-01 3.11067015e-01
6.09339416e-01 2.71656811e-01 -4.82828259e-01 3.34673673e-01
1.42703724e+00 -9.77968872e-01 7.31726170e-01 4.88178730e-01
3.57996851e-01 -5.87436438e-01 1.62840977e-01 8.68187547e-02
-8.10553014e-01 3.27621162e-01 -2.42093384e-01 -2.70287633e-01
-3.52817535e-01 7.14595735e-01 -6.09583735e-01 8.52696002e-01
5.73611617e-01 4.23643500e-01 -5.59496641e-01 6.68524265e-01
-1.30132884e-01 3.58875185e-01 -5.49636841e-01 -5.71836114e-01
-1.79137811e-02 -8.32881868e-01 7.63400018e-01 1.04605067e+00
1.97794870e-01 2.24056870e-01 -1.43710047e-01 1.00755930e+00
-6.05111010e-02 5.52170649e-02 -1.06879592e+00 -5.25478840e-01
6.27845645e-01 1.36474013e+00 -1.01328564e+00 -5.55793285e-01
-1.08066596e-01 4.92673725e-01 1.16289273e-01 4.79528189e-01
-8.19846153e-01 -4.98899579e-01 7.78084755e-01 1.03185549e-01
-1.54843315e-01 -4.59688783e-01 -2.22818971e-01 -1.49493408e+00
-5.70455432e-01 -1.40793848e+00 3.52680832e-01 -8.31188619e-01
-1.16489601e+00 2.85467267e-01 1.31456554e-01 -4.18236017e-01
-1.39069349e-01 -4.33040410e-01 -6.04956269e-01 6.04685426e-01
-7.32868969e-01 -8.32426429e-01 7.53558874e-02 4.75057721e-01
-1.41687274e-01 1.27368778e-01 8.64657938e-01 -1.24519318e-01
-3.87154430e-01 1.24948353e-01 2.80354083e-01 -4.56728727e-01
7.90510356e-01 -1.36963892e+00 5.25538981e-01 7.03264713e-01
-2.47472435e-01 1.04943538e+00 1.29081750e+00 -6.97070599e-01
-2.37922502e+00 -5.44878185e-01 7.71600068e-01 -7.41112232e-01
1.00301659e+00 -8.64005983e-01 -4.07955885e-01 7.85088003e-01
8.93139660e-01 -3.45325053e-01 2.33658729e-03 1.69643655e-01
-4.19870347e-01 2.18786802e-02 -1.08542323e+00 1.12079847e+00
1.46437740e+00 -3.96773279e-01 2.71715820e-02 8.56358111e-01
1.16373348e+00 8.83044824e-02 -1.12385476e+00 1.11433148e-01
4.62269574e-01 -9.56111848e-01 6.21672809e-01 -3.61681879e-01
2.50772893e-01 -5.21323204e-01 -1.35067463e-01 -1.21046162e+00
-1.38400644e-01 -1.24780464e+00 -1.05566323e-01 1.04382110e+00
6.33616388e-01 -1.48996317e+00 3.29474419e-01 6.95413709e-01
-5.79817146e-02 -2.55193383e-01 -3.76078039e-01 -1.10031724e+00
4.41862196e-01 1.14280999e-01 5.67197323e-01 1.27213323e+00
5.24618208e-01 6.42752945e-01 2.76805967e-01 -2.06526428e-01
6.47458315e-01 2.23635044e-03 1.11808085e+00 -1.22224927e+00
-3.97883981e-01 -5.03214180e-01 -2.34947711e-01 -3.00042808e-01
1.26217544e-01 -1.30224609e+00 -1.81860939e-01 -1.60889137e+00
-2.61123739e-02 -5.25465250e-01 2.19623014e-01 5.34145057e-01
5.94085515e-01 -2.97118634e-01 3.02480698e-01 3.65493357e-01
-7.44937837e-01 3.47718269e-01 1.20158446e+00 8.13630372e-02
-5.71733974e-02 -4.53858614e-01 -9.65086341e-01 1.95350498e-01
1.04266346e+00 -5.39029777e-01 -4.42050397e-01 2.48602498e-02
8.05545270e-01 2.61415452e-01 3.14306676e-01 -7.35637367e-01
3.56325179e-01 -4.81944054e-01 -4.98609334e-01 9.63617116e-02
2.33459353e-01 -5.26671946e-01 8.30578864e-01 6.98117316e-01
-2.11364284e-01 5.87896943e-01 -7.34868571e-02 3.01741451e-01
4.77049589e-01 -1.93061829e-01 4.93954897e-01 -7.13195860e-01
-1.77045822e-01 1.43387482e-01 -8.73886347e-01 2.12197937e-02
1.07816434e+00 8.60749334e-02 -1.09716833e+00 -3.66704047e-01
-5.53400457e-01 4.13658738e-01 1.14168489e+00 9.09474790e-02
-1.16627954e-01 -6.77694976e-01 -6.03250921e-01 -4.00773108e-01
-4.23820913e-02 -2.09731400e-01 -3.01578194e-01 1.08373380e+00
-1.26623058e+00 4.76242453e-01 -1.37712300e-01 -3.03119361e-01
-1.10360372e+00 6.59725785e-01 3.88652325e-01 -7.00755641e-02
-4.77459580e-01 3.23314071e-01 3.85875165e-01 -7.57113218e-01
-1.85003012e-01 -3.60665470e-01 6.36867106e-01 -4.67595667e-01
-7.65034109e-02 4.53947067e-01 -4.67245616e-02 3.34610455e-02
-4.76751059e-01 3.96711305e-02 3.80020007e-03 -3.45888644e-01
1.33741951e+00 -9.13503617e-02 -1.11379874e+00 4.24415559e-01
8.30590487e-01 7.70061389e-02 -2.80122727e-01 3.79627943e-01
7.31176138e-02 -1.31683305e-01 -3.82439196e-01 -7.63383150e-01
-6.33201301e-01 3.38731885e-01 4.19540294e-02 7.32189000e-01
3.23254228e-01 1.09130040e-01 9.45459902e-02 8.49339068e-01
7.61223078e-01 -9.10116553e-01 -2.44822741e-01 7.25855768e-01
8.66883755e-01 -4.57433045e-01 3.73896152e-01 -4.74796653e-01
-3.25572222e-01 9.04833734e-01 7.34657586e-01 -3.25419724e-01
6.35832906e-01 4.33705181e-01 -3.23361367e-01 -7.40238428e-01
-1.50411773e+00 -1.76795557e-01 -9.05496895e-01 4.75885481e-01
7.12228715e-01 2.53619105e-01 -1.08422303e+00 -8.89977664e-02
-1.96982563e-01 -2.30164737e-01 1.24848938e+00 1.22694421e+00
-1.73846006e-01 -1.65411901e+00 -2.71227807e-01 2.25930393e-01
-3.49024326e-01 -3.88803482e-01 -1.10877287e+00 1.03711247e+00
-3.56508106e-01 9.35079992e-01 -1.91494569e-01 -7.64803961e-02
-1.23534493e-01 2.14580849e-01 7.71804810e-01 -6.69470906e-01
-9.74685252e-01 -3.40548933e-01 5.74312627e-01 -3.22346687e-01
-1.94653720e-01 -4.69112992e-01 -1.49440193e+00 -1.35304117e+00
-2.59708643e-01 8.45728338e-01 7.69412279e-01 3.89007926e-01
7.19666600e-01 3.23567837e-01 2.91759878e-01 -5.39723396e-01
-1.99662521e-01 -7.04730690e-01 -6.31121099e-01 -1.49299607e-01
-9.02326182e-02 -1.98780805e-01 -4.66790646e-01 9.28149447e-02] | [5.667630195617676, 4.306291103363037] |
50b4433b-04ee-4678-abfc-f6ffb78dacc4 | poetrydiffusion-towards-joint-semantic-and | 2306.08456 | null | https://arxiv.org/abs/2306.08456v1 | https://arxiv.org/pdf/2306.08456v1.pdf | PoetryDiffusion: Towards Joint Semantic and Metrical Manipulation in Poetry Generation | Poetry generation is a typical and popular task in natural language generation. While prior works have shown success in controlling either semantic or metrical aspects of poetry generation, there are still challenges in addressing both perspectives simultaneously. In this paper, we employ the Diffusion model to generate poetry in Sonnet and SongCi in Chinese for the first time to tackle such challenges. Different from autoregressive generation, our PoetryDiffusion model, based on Diffusion model, generates the complete sentence or poetry by taking into account the whole sentence information, resulting in improved semantic expression. Additionally, we incorporate a novel metrical controller to manipulate and evaluate metrics (format and rhythm). The denoising process in PoetryDiffusion allows for gradual enhancement of semantics and flexible integration of the metrical controller. Experimental results on two datasets demonstrate that our model outperforms existing models in terms of semantic, metrical and overall performance. | ['Bryan Hooi', 'Yue Feng', 'Chumin Liu', 'Zhiyuan Hu'] | 2023-06-14 | null | null | null | null | ['text-generation'] | ['natural-language-processing'] | [ 2.31156498e-01 8.81394222e-02 2.52813578e-01 -2.58331925e-01
-4.63376343e-01 -6.32005930e-01 9.77711797e-01 -1.18767016e-01
-2.15572253e-01 8.11008751e-01 7.55168557e-01 2.17426792e-01
-1.16782144e-01 -1.22760546e+00 -3.21914464e-01 -3.85227472e-01
6.08477473e-01 3.48073781e-01 -5.73386624e-02 -7.89099336e-01
4.81536984e-01 -2.57809628e-02 -1.44162929e+00 1.04941472e-01
1.33129704e+00 5.42533159e-01 3.83895963e-01 6.32187009e-01
-3.40277642e-01 1.00749540e+00 -1.20466745e+00 -5.71753681e-01
7.74837360e-02 -9.19330120e-01 -5.51675439e-01 -3.10464539e-02
1.16709255e-01 -1.55109251e-02 2.23125499e-02 9.39814985e-01
8.37408125e-01 3.01028192e-01 7.57255852e-01 -1.11623144e+00
-1.24788034e+00 1.26338005e+00 -1.94045290e-01 -3.29638273e-01
5.22857606e-01 -6.37768442e-03 1.16314530e+00 -8.48347604e-01
5.27641118e-01 1.15778399e+00 5.49558580e-01 6.73563778e-01
-1.08838356e+00 -4.47774351e-01 -2.53032446e-01 1.89226028e-02
-1.35581267e+00 -3.83131057e-01 9.62752342e-01 -2.88930833e-01
7.52597928e-01 4.94067132e-01 6.86279297e-01 9.81083870e-01
-8.96756276e-02 7.23773777e-01 9.53056395e-01 -5.48989713e-01
2.17557594e-01 8.94778147e-02 -2.97839254e-01 3.65434110e-01
-8.12190324e-02 -3.09804320e-01 -7.44921744e-01 2.30453849e-01
6.44891083e-01 -2.96949387e-01 -2.63761252e-01 2.99662501e-01
-1.36630499e+00 8.24355662e-01 1.23141967e-01 4.88000184e-01
-5.96423209e-01 2.87731081e-01 1.94670931e-01 1.82162955e-01
3.30050051e-01 9.52933371e-01 -7.75050605e-03 -5.98688900e-01
-1.29552507e+00 5.40301383e-01 9.31794584e-01 1.11452329e+00
4.94012505e-01 3.42389166e-01 -8.50815773e-01 1.03977191e+00
9.31661576e-03 5.14683962e-01 7.25584984e-01 -9.48753893e-01
4.67148691e-01 5.45559525e-01 7.12474808e-02 -1.08373773e+00
-1.30603850e-01 -2.86569357e-01 -9.01974142e-01 -1.75621912e-01
3.66502590e-02 -2.25166127e-01 -7.00017989e-01 1.68964732e+00
-5.70322238e-02 1.94906890e-02 2.37125419e-02 7.87859797e-01
9.61574733e-01 7.77470708e-01 1.94602415e-01 -2.01741047e-02
1.29837143e+00 -1.17963767e+00 -1.00174439e+00 -1.28841083e-02
3.90801996e-01 -1.02142847e+00 1.45303345e+00 4.80650663e-01
-1.36600637e+00 -7.22841501e-01 -1.18921626e+00 -4.17681992e-01
-2.91550040e-01 3.32778156e-01 4.51779574e-01 7.85007060e-01
-1.02459669e+00 7.84153342e-01 -3.77540469e-01 -1.95121512e-01
3.21960300e-02 -1.19304806e-02 1.47984847e-01 3.31110686e-01
-1.46332264e+00 6.43855691e-01 5.35705328e-01 -3.14543135e-02
-1.54683352e-01 -8.09258580e-01 -8.50101471e-01 2.45771095e-01
7.90890232e-02 -1.09015298e+00 1.23596859e+00 -5.98775387e-01
-2.05035472e+00 3.07296127e-01 4.43614870e-02 -3.64777118e-01
6.21003628e-01 -2.06599683e-01 -2.52413481e-01 -5.83133474e-02
1.45570174e-01 9.28516388e-01 6.37143195e-01 -9.50923026e-01
-3.75703484e-01 8.68684277e-02 1.74003374e-02 4.68943924e-01
-7.51902461e-01 6.75268844e-02 -3.33279133e-01 -1.13657093e+00
-4.05213349e-02 -5.05651534e-01 -7.65230209e-02 -4.48909998e-01
-4.00550663e-01 -2.39929199e-01 3.83689195e-01 -7.95320153e-01
1.81580472e+00 -1.85058761e+00 1.33133486e-01 -1.46157527e-02
-8.92105140e-03 1.49072707e-01 -1.50470749e-01 7.00978994e-01
4.04766738e-01 3.16741377e-01 -3.23630065e-01 -6.01648331e-01
4.04855400e-01 1.55378371e-01 -4.55249637e-01 -3.55858535e-01
3.34091306e-01 1.15027249e+00 -1.03983366e+00 -5.35295188e-01
1.04583845e-01 6.49007261e-01 -7.92649686e-01 1.61573850e-02
-3.96644205e-01 3.82971913e-01 -3.89424235e-01 2.85470575e-01
5.64158201e-01 6.34045750e-02 1.63284857e-02 1.40912354e-01
-1.63530082e-01 6.05610669e-01 -9.55633640e-01 1.88802469e+00
-8.09178591e-01 3.01152289e-01 -4.73251671e-01 -5.39247274e-01
1.48512733e+00 3.34706485e-01 2.56778657e-01 -7.16381848e-01
-1.16277047e-01 1.47907063e-01 -2.18338400e-01 -4.44724321e-01
1.38047564e+00 -3.13781768e-01 -3.35304976e-01 7.29641020e-01
-1.29670724e-01 -9.15978312e-01 5.94438434e-01 9.48704854e-02
9.08751905e-01 4.44579512e-01 2.85212338e-01 3.31601799e-02
4.87814188e-01 4.04384993e-02 4.48558092e-01 7.20366061e-01
2.38020331e-01 1.12801909e+00 3.60178679e-01 3.28271613e-02
-9.51736033e-01 -1.07718050e+00 3.79424721e-01 9.52354193e-01
-9.75336283e-02 -8.90565991e-01 -9.80460227e-01 -2.72671342e-01
-3.56739372e-01 1.27128696e+00 -2.14456022e-01 -1.52313858e-01
-6.81120157e-01 -7.31265247e-01 8.08050036e-01 4.99345839e-01
7.91411161e-01 -1.42529464e+00 -2.98399001e-01 4.52545255e-01
-5.45088768e-01 -1.01416540e+00 -6.82043433e-01 -6.47493303e-01
-5.90294778e-01 -4.12666887e-01 -6.50840700e-01 -6.37773037e-01
3.18410128e-01 9.19932798e-02 1.15455806e+00 -1.16227686e-01
-1.10461965e-01 1.20292820e-01 -6.08730376e-01 -5.73243022e-01
-6.83803082e-01 2.30995283e-01 -3.91235560e-01 -2.91410293e-02
2.24443227e-01 -7.33064294e-01 -4.37447697e-01 3.23679373e-02
-1.19845974e+00 3.17532122e-01 3.00701290e-01 7.74419665e-01
3.35554749e-01 1.49533361e-01 7.56290197e-01 -9.29913163e-01
1.40590787e+00 -4.44447100e-01 -2.05133364e-01 1.73487708e-01
-7.11396813e-01 4.31695674e-03 7.90598154e-01 -3.10425788e-01
-1.31520760e+00 -3.97990644e-01 -1.22275390e-01 1.56973630e-01
1.59385920e-01 7.86757171e-01 -1.99463740e-01 5.58121085e-01
6.92144752e-01 4.34175640e-01 -9.45889354e-02 -4.41247940e-01
5.71920633e-01 8.14840913e-01 5.35484016e-01 -6.75228953e-01
7.65277386e-01 3.57749462e-01 -1.60594925e-01 -7.98748195e-01
-6.28623307e-01 -1.41831204e-01 -4.18520123e-01 -8.55695456e-02
7.30078518e-01 -8.14867914e-01 -3.76911521e-01 5.32454371e-01
-1.36629009e+00 -1.32194296e-01 -7.94518471e-01 3.63882005e-01
-6.82157159e-01 3.35390836e-01 -5.48503518e-01 -6.47738397e-01
-7.13387728e-01 -7.64679313e-01 1.14600122e+00 3.54577541e-01
-8.18620563e-01 -1.13687050e+00 7.67126158e-02 5.16845345e-01
6.94022655e-01 1.29754856e-01 7.29683280e-01 -3.08015406e-01
-4.51849818e-01 -1.40301630e-01 -1.50836587e-01 5.35081148e-01
3.20136756e-01 -2.40759552e-02 -8.06956947e-01 4.16373223e-01
1.50548875e-01 -6.01938516e-02 7.65996993e-01 8.75013098e-02
8.81530225e-01 -3.95498872e-01 3.65909100e-01 6.42578781e-01
1.05664825e+00 -9.60769355e-02 9.92858708e-01 3.20707947e-01
7.20548689e-01 6.22282922e-01 6.12970054e-01 8.65251243e-01
7.78781056e-01 7.19583869e-01 -1.79245293e-01 1.26896307e-01
-6.12509489e-01 -7.32323349e-01 6.28988743e-01 1.28477728e+00
-3.29857051e-01 -4.85991329e-01 -6.15890145e-01 5.60828269e-01
-1.83291852e+00 -1.20121109e+00 -1.97930723e-01 1.81923640e+00
1.11143601e+00 -2.46512860e-01 -1.56742632e-01 2.96539426e-01
5.22835016e-01 3.56288016e-01 6.64637461e-02 -7.29803503e-01
-5.00204742e-01 4.96609986e-01 3.07372473e-02 4.74258095e-01
-7.12543011e-01 1.27169359e+00 6.28543234e+00 1.01908290e+00
-1.03670239e+00 -5.20851347e-04 2.43722826e-01 -4.11661901e-02
-8.45996559e-01 1.27391264e-01 -7.07638085e-01 6.01889908e-01
5.79273403e-01 -5.47418833e-01 5.38294077e-01 5.13532519e-01
8.02756786e-01 1.32585168e-01 -7.64634788e-01 8.47537577e-01
4.35224563e-01 -1.31684434e+00 3.41211557e-01 -2.47442558e-01
1.03707349e+00 -5.44462740e-01 2.62163226e-02 2.65923113e-01
4.07305598e-01 -1.06059635e+00 1.09316695e+00 9.27873075e-01
6.26289010e-01 -5.41876316e-01 4.78518516e-01 3.67307574e-01
-9.43916678e-01 2.37412900e-01 -1.34523898e-01 -3.72341037e-01
4.87384915e-01 4.91731644e-01 -9.93761599e-01 5.84968388e-01
2.77450025e-01 7.96229362e-01 -4.70063955e-01 8.17029059e-01
-9.57639217e-01 7.11232185e-01 -1.70062810e-01 -2.97417849e-01
8.64923671e-02 -6.01263523e-01 5.62876701e-01 1.27186799e+00
8.48077297e-01 -2.22358242e-01 -8.29299763e-02 1.45293522e+00
-1.05300263e-01 4.14823145e-01 -3.89579743e-01 -3.09203386e-01
5.67387164e-01 1.24656177e+00 -3.27130526e-01 -2.78708547e-01
1.15021713e-01 1.34131742e+00 -5.18251816e-03 4.27394509e-01
-8.65088761e-01 -6.28678203e-01 3.86803180e-01 1.86214939e-01
1.36496440e-01 -4.46989864e-01 -8.15295517e-01 -1.07502496e+00
3.14762115e-01 -6.08801663e-01 1.41198516e-01 -1.02783716e+00
-1.19836926e+00 4.63215917e-01 -1.28899455e-01 -1.12047875e+00
-6.38914824e-01 -1.02715433e-01 -8.88968349e-01 1.03982615e+00
-1.53961849e+00 -1.51251352e+00 -3.95852000e-01 2.55075008e-01
6.87787950e-01 -2.44786143e-01 9.12591517e-01 6.29845336e-02
-2.25213379e-01 4.41082925e-01 -1.47010237e-01 -9.72388089e-02
7.34257281e-01 -1.37635827e+00 5.91026664e-01 8.45031381e-01
2.16082275e-01 6.50189221e-01 6.36480093e-01 -6.39176250e-01
-1.00586665e+00 -1.13885653e+00 1.64729595e+00 -3.18639904e-01
6.72473311e-01 -3.22566956e-01 -4.73035574e-01 2.41058752e-01
3.86138141e-01 -7.73046076e-01 6.75171077e-01 -2.30083659e-01
-1.58225775e-01 3.93710211e-02 -9.89819407e-01 1.10141289e+00
1.12004638e+00 -2.46824890e-01 -5.76220036e-01 2.14192837e-01
9.14347410e-01 -1.28337502e-01 -8.73365521e-01 9.80630666e-02
3.58091801e-01 -9.96771812e-01 7.89224565e-01 -1.91909388e-01
8.60303819e-01 -5.06751716e-01 5.31696714e-04 -1.64421749e+00
-2.06480533e-01 -1.27920842e+00 9.18322336e-03 1.73392832e+00
3.51677507e-01 -3.67486805e-01 6.37718916e-01 4.30679500e-01
-3.26347321e-01 -3.18991363e-01 -4.58057731e-01 -7.32749879e-01
4.64298539e-02 -4.55078900e-01 1.11925638e+00 7.50367582e-01
7.15419054e-02 6.41637623e-01 -4.94773895e-01 -3.91215235e-01
1.81398377e-01 2.76610274e-02 9.03872550e-01 -9.41365480e-01
-4.74492669e-01 -8.03428292e-01 -2.03847572e-01 -1.04724884e+00
5.78657277e-02 -1.04941559e+00 1.10778064e-01 -1.87817109e+00
-1.22190893e-01 -4.13156271e-01 8.10892135e-02 2.79262424e-01
-3.61460477e-01 3.78482699e-01 5.12586355e-01 1.35811329e-01
-2.03260943e-01 8.51685822e-01 1.52780712e+00 -2.49596462e-02
-5.57750583e-01 -1.32045865e-01 -1.12784374e+00 5.14946699e-01
1.02604580e+00 -1.91498056e-01 -6.30146921e-01 -6.98649526e-01
3.97734493e-01 -2.40610048e-01 1.77332640e-01 -9.69232202e-01
3.15098822e-01 -2.25013688e-01 6.80628866e-02 -2.46681750e-01
5.02377391e-01 -1.08028144e-01 7.84347132e-02 6.78181425e-02
-3.14457864e-01 9.33406204e-02 -1.56151265e-01 9.27180722e-02
-4.44895059e-01 -4.22594815e-01 3.73436570e-01 -1.09265760e-01
-5.77210128e-01 -1.13088086e-01 -2.91851759e-01 1.60681874e-01
6.92756414e-01 -4.25233305e-01 -3.19135129e-01 -6.91816032e-01
-4.36001986e-01 4.40342240e-02 3.21579874e-01 6.13158941e-01
7.98726499e-01 -1.44535446e+00 -1.07912052e+00 1.44042104e-01
3.23788449e-02 -7.16228262e-02 9.78366286e-02 5.82043529e-01
-6.90533578e-01 1.58294067e-01 -1.78721081e-02 -8.19221511e-02
-8.35385859e-01 8.31327066e-02 -8.28323066e-02 -4.77057844e-01
-1.94853395e-01 6.13424957e-01 -1.38269112e-01 -7.14284301e-01
-1.85184762e-01 -1.66705057e-01 -4.07380849e-01 2.25251734e-01
5.26973128e-01 5.62786579e-01 -1.41764656e-01 -5.49104035e-01
3.19131225e-01 4.29520041e-01 2.64398664e-01 -6.36879802e-01
1.39306486e+00 -2.49716520e-01 -4.88903195e-01 4.17124689e-01
5.72033942e-01 4.30473655e-01 -8.68457735e-01 5.64319342e-02
1.53216019e-01 -3.09213400e-01 -1.28287539e-01 -9.97191727e-01
-6.81929111e-01 7.81589329e-01 -2.26973951e-01 1.27139613e-01
1.11245918e+00 -3.99273336e-01 1.25031543e+00 1.45460740e-01
1.00584917e-01 -1.43633521e+00 1.46586284e-01 7.59212554e-01
1.08416057e+00 -7.45379567e-01 -1.49614170e-01 -5.09000540e-01
-1.07059669e+00 1.11820328e+00 4.34479386e-01 3.98818590e-02
3.09697241e-01 1.22863837e-01 1.68746188e-01 1.01123065e-01
-6.48415267e-01 -8.40844288e-02 3.58164191e-01 7.29705632e-01
8.68536234e-01 2.72840798e-01 -8.10384989e-01 9.17898357e-01
-1.15115118e+00 1.76634759e-01 7.58461237e-01 6.99113905e-01
-4.33322638e-01 -1.46711946e+00 -2.39356831e-01 1.36168040e-02
-2.22698405e-01 -5.63428819e-01 -5.17198622e-01 3.91676277e-01
2.26462558e-01 1.20715249e+00 5.54836839e-02 -4.07850713e-01
3.20310980e-01 4.31414396e-02 4.22144473e-01 -8.40193570e-01
-9.14849997e-01 3.16028073e-02 2.47912377e-01 -1.26295701e-01
-2.41543368e-01 -5.57231903e-01 -1.39311743e+00 -3.08914393e-01
-1.03594281e-01 2.39954516e-01 1.02023554e+00 7.37344742e-01
4.61445898e-01 6.66862011e-01 6.63856685e-01 -4.54396546e-01
-7.58792400e-01 -1.28414392e+00 -6.70983851e-01 5.68081081e-01
-3.60020161e-01 -1.17537022e-01 -4.27686498e-02 3.56576413e-01] | [11.683601379394531, 9.17752456665039] |
05a90210-fa0e-446d-9a78-31104ab5da4c | on-the-apparent-conflict-between-individual | 1912.06883 | null | https://arxiv.org/abs/1912.06883v1 | https://arxiv.org/pdf/1912.06883v1.pdf | On the Apparent Conflict Between Individual and Group Fairness | A distinction has been drawn in fair machine learning research between `group' and `individual' fairness measures. Many technical research papers assume that both are important, but conflicting, and propose ways to minimise the trade-offs between these measures. This paper argues that this apparent conflict is based on a misconception. It draws on theoretical discussions from within the fair machine learning research, and from political and legal philosophy, to argue that individual and group fairness are not fundamentally in conflict. First, it outlines accounts of egalitarian fairness which encompass plausible motivations for both group and individual fairness, thereby suggesting that there need be no conflict in principle. Second, it considers the concept of individual justice, from legal philosophy and jurisprudence which seems similar but actually contradicts the notion of individual fairness as proposed in the fair machine learning literature. The conclusion is that the apparent conflict between individual and group fairness is more of an artifact of the blunt application of fairness measures, rather than a matter of conflicting principles. In practice, this conflict may be resolved by a nuanced consideration of the sources of `unfairness' in a particular deployment context, and the carefully justified application of measures to mitigate it. | ['Reuben Binns'] | 2019-12-14 | null | null | null | null | ['jurisprudence'] | ['miscellaneous'] | [ 9.05168653e-02 3.83384347e-01 -3.29775393e-01 -7.75282979e-01
-5.73291898e-01 -6.03748918e-01 7.80408204e-01 3.50137830e-01
-8.10261190e-01 5.88755369e-01 6.21966124e-01 -8.01772535e-01
-7.00493753e-01 -4.68411833e-01 4.94025722e-02 -5.15818715e-01
4.85394716e-01 1.52168408e-01 -2.83743829e-01 -2.87898570e-01
8.70204270e-01 4.19200808e-01 -1.32197428e+00 1.43388823e-01
8.47922981e-01 6.68633938e-01 -7.90536821e-01 4.59415138e-01
-3.81446600e-01 1.42079556e+00 -7.58762538e-01 -1.00296402e+00
5.78203678e-01 -7.47547686e-01 -1.18680179e+00 -1.66847527e-01
5.78522623e-01 -4.01648372e-01 1.26046538e-01 1.09055948e+00
5.04012525e-01 7.82588646e-02 6.52662337e-01 -1.49685848e+00
-6.78577185e-01 6.09049678e-01 -6.58888578e-01 4.35473204e-01
-2.90514380e-01 3.58632654e-01 1.30205619e+00 -5.19688614e-02
6.81026727e-02 1.26431870e+00 8.77967060e-01 6.14440441e-01
-1.10549116e+00 -7.82978594e-01 -6.08448461e-02 -1.10147178e-01
-1.11449087e+00 -7.87592411e-01 2.24058107e-01 -8.83791924e-01
6.42322600e-01 7.40417540e-01 5.86213171e-01 2.54172802e-01
4.05119836e-01 8.44060928e-02 1.37345386e+00 -8.24763536e-01
2.81350523e-01 2.55008996e-01 1.90032110e-01 2.82674521e-01
7.65351832e-01 4.68948185e-01 -1.43552870e-01 -6.18484557e-01
5.54325640e-01 -2.77602881e-01 -1.67697109e-02 2.03772336e-02
-7.10783005e-01 1.23156178e+00 1.58641070e-01 4.74660397e-01
-3.77261639e-01 1.19355097e-01 9.44243014e-01 4.40798670e-01
5.16489327e-01 5.60495496e-01 -2.31681824e-01 -4.28032160e-01
-9.87926781e-01 2.98625588e-01 9.24059272e-01 4.55065295e-02
4.50244009e-01 1.68408930e-01 -1.24251032e-02 6.36272013e-01
5.48983157e-01 2.88392883e-02 -3.13233444e-03 -1.66384244e+00
9.41607505e-02 2.70968646e-01 2.86819011e-01 -1.34165668e+00
-1.17966734e-01 -1.05208613e-01 -5.11978984e-01 8.99855435e-01
6.28805101e-01 -3.10773134e-01 -2.88170248e-01 1.79191720e+00
7.68572930e-03 -6.24540210e-01 -2.72964478e-01 9.32090998e-01
3.64499182e-01 1.27631709e-01 8.07768643e-01 -3.10620338e-01
9.91137207e-01 -2.27262855e-01 -7.81288683e-01 -4.29951042e-01
5.88080227e-01 -8.71885598e-01 9.11126435e-01 7.50493184e-02
-1.12145340e+00 -2.11444780e-01 -7.95820415e-01 -1.27140388e-01
-1.32903438e-02 -7.00786352e-01 8.41838241e-01 1.40504408e+00
-6.87954605e-01 7.41370082e-01 -1.90739110e-01 -4.00020093e-01
6.05695009e-01 -3.19253691e-02 -3.76854353e-02 3.93002093e-01
-9.55660224e-01 1.29779708e+00 1.77433819e-01 2.45186985e-01
-5.47448471e-02 -5.07436097e-01 -4.76959705e-01 1.10484466e-01
3.84525687e-01 -6.42390847e-01 1.09661269e+00 -1.77132237e+00
-8.77970636e-01 1.42040563e+00 4.67657149e-01 -3.33030045e-01
8.86334777e-01 -1.72228783e-01 -4.38067585e-01 -2.58482516e-01
2.50349611e-01 2.18983993e-01 4.75686461e-01 -1.36945987e+00
-9.30699706e-01 -4.65014368e-01 3.68387014e-01 3.59861642e-01
1.01839058e-01 7.95728564e-01 6.98461413e-01 -3.67770791e-01
-2.40388334e-01 -7.89603591e-01 -3.14523101e-01 1.04774363e-01
1.89763252e-02 -2.30483189e-01 3.58084589e-01 -3.81842554e-01
1.44113445e+00 -2.19826818e+00 -9.09964502e-01 1.88600391e-01
4.58233565e-01 2.91829109e-01 1.76140487e-01 4.85251456e-01
-2.07219586e-01 7.75712788e-01 -2.27249369e-01 3.97842765e-01
2.71850109e-01 3.36022764e-01 -3.59716594e-01 1.00609612e+00
-1.25180155e-01 5.91594934e-01 -7.74188399e-01 -6.47008598e-01
2.11596802e-01 1.10320605e-01 -4.73934293e-01 -2.07773402e-01
5.07019818e-01 5.16538173e-02 -2.15130463e-01 2.45721579e-01
6.81316137e-01 3.66506517e-01 5.16761184e-01 9.95901302e-02
-6.12851143e-01 5.42388916e-01 -1.08427596e+00 8.20244968e-01
2.17167854e-01 7.04113483e-01 2.88034558e-01 -1.13456368e+00
7.40808010e-01 3.66184026e-01 3.98160338e-01 -8.02507520e-01
6.32008493e-01 4.26099181e-01 5.83843291e-01 -4.56012219e-01
5.19380391e-01 -9.56370294e-01 -1.55304193e-01 1.05557442e+00
-4.39776897e-01 -2.84660727e-01 -2.78994858e-01 -1.46031842e-01
6.97837412e-01 -1.05406707e-02 6.30145013e-01 -8.30600977e-01
1.70490801e-01 -9.98502411e-03 8.76241565e-01 9.14359748e-01
-1.08539867e+00 3.38564515e-01 5.38893700e-01 -6.87197864e-01
-9.69374657e-01 -8.08234096e-01 -2.09909976e-01 1.16936028e+00
4.72896248e-02 -6.17541410e-02 -7.00250506e-01 -7.76530206e-01
1.20594054e-01 9.48698044e-01 -6.01792455e-01 -9.39732715e-02
-2.78159618e-01 -7.81111717e-01 7.41577268e-01 1.19454741e-01
3.99277866e-01 -8.78799736e-01 -1.56250381e+00 -9.16282907e-02
-4.53701723e-05 -6.94791436e-01 -3.25678848e-02 -5.64211495e-02
-7.17808485e-01 -1.24364924e+00 -7.27687627e-02 4.41509262e-02
1.50125295e-01 3.86992723e-01 1.22434664e+00 8.30104470e-01
2.08611345e-05 5.82456231e-01 -2.36457914e-01 -8.08001637e-01
-6.87466204e-01 -3.80921513e-01 -2.36089215e-01 -4.03011382e-01
1.03694189e+00 -3.80999655e-01 -4.43659544e-01 2.77389865e-02
-9.20204639e-01 -2.90312409e-01 4.40024167e-01 6.02412343e-01
-2.00232744e-01 1.21162131e-01 8.85730743e-01 -1.23841214e+00
1.08771992e+00 -5.58948636e-01 -1.41119257e-01 1.19146109e-01
-1.28352559e+00 -5.23268163e-01 1.90225542e-01 2.27023780e-01
-1.26973867e+00 -9.90954995e-01 5.05974777e-02 1.15822293e-01
-4.44977582e-01 2.65148580e-01 1.57002006e-02 -1.12804696e-01
1.14348435e+00 -5.72645903e-01 4.53582406e-01 -5.78507520e-02
4.36557919e-01 8.11549664e-01 1.32240817e-01 -8.47415745e-01
6.12841845e-01 4.60792542e-01 -2.29333252e-01 -7.23003268e-01
-8.31692815e-01 -1.51622221e-01 -5.88853180e-01 -4.27415103e-01
6.60502255e-01 -6.12367749e-01 -8.49804282e-01 -1.40454739e-01
-7.65318394e-01 -2.30429500e-01 -6.43891990e-01 5.59921443e-01
-6.36940122e-01 5.01803875e-01 -5.16950965e-01 -1.30845201e+00
-4.35125291e-01 -7.74865270e-01 9.76701751e-02 5.51735535e-02
-8.38430405e-01 -1.00853288e+00 -1.10365175e-01 7.16823161e-01
6.70458615e-01 5.72386742e-01 1.03721654e+00 -6.77659154e-01
4.62583788e-02 -5.51133513e-01 -3.68232936e-01 6.21819615e-01
5.28059304e-01 2.80572772e-01 -1.05792391e+00 -4.06223759e-02
4.46821868e-01 -3.09013009e-01 1.64802581e-01 2.62461871e-01
3.99734497e-01 -8.36309969e-01 4.18344676e-01 -8.75615999e-02
1.81245220e+00 3.07606459e-01 7.70736158e-01 5.37422478e-01
3.59767050e-01 1.32497454e+00 7.51460850e-01 6.38355255e-01
4.72259074e-01 4.41324770e-01 1.94632798e-01 -1.43234909e-01
2.77326286e-01 3.01818490e-01 -7.11322650e-02 4.06556696e-01
-4.78952378e-01 5.05989432e-01 -1.19505000e+00 4.80266869e-01
-1.92479265e+00 -1.62701166e+00 -6.31823123e-01 2.49745083e+00
6.33232236e-01 3.45063239e-01 6.76314771e-01 1.95416063e-01
8.04112315e-01 3.97068113e-01 -4.72583845e-02 -1.42510056e+00
3.67016137e-01 -8.73242617e-02 5.84513903e-01 7.09365427e-01
-7.68192053e-01 5.96501827e-01 7.14372301e+00 6.74571216e-01
-8.87032866e-01 3.36003095e-01 9.08534884e-01 -1.25225574e-01
-5.71447790e-01 2.87861139e-01 2.08259910e-01 4.78566915e-01
8.48204315e-01 -9.35261905e-01 1.39510885e-01 5.44099927e-01
3.92147541e-01 -2.78982490e-01 -7.61973917e-01 5.21765172e-01
-3.08271796e-01 -9.43852842e-01 -2.79348373e-01 3.52688879e-01
3.27903390e-01 -1.46222696e-01 6.24723248e-02 1.25610486e-01
6.18164003e-01 -1.21217906e+00 1.16119659e+00 2.38859549e-01
3.80447716e-01 -1.11232853e+00 8.61475289e-01 3.28456789e-01
-4.61478859e-01 -1.86678186e-01 -4.45932209e-01 -7.70051539e-01
-5.15143499e-02 6.38549089e-01 -9.98585075e-02 3.43828470e-01
6.79966986e-01 -3.70999090e-02 -1.50164589e-01 8.86121333e-01
-1.55202923e-02 5.55580914e-01 1.90475523e-01 5.00034869e-01
4.24481958e-01 -3.28608513e-01 2.81980753e-01 1.40498400e+00
-3.69445235e-01 3.50288033e-01 -2.24443376e-01 7.77837694e-01
4.75342870e-01 1.60343274e-01 -6.89449668e-01 -4.39454466e-02
7.56742239e-01 1.22513807e+00 -7.33765125e-01 -4.17580940e-02
-6.67482495e-01 3.78853321e-01 -9.79343057e-03 1.50063917e-01
-7.18712866e-01 -1.82169788e-02 8.45834851e-01 1.47393703e-01
-4.18956965e-01 1.27744451e-01 -8.83271337e-01 -7.08794355e-01
-2.52385259e-01 -1.27882612e+00 4.14192677e-01 4.07203697e-02
-1.26040494e+00 2.50179708e-01 -7.16314763e-02 -9.07625139e-01
7.65399411e-02 -2.09812522e-01 -7.29901195e-01 1.27458739e+00
-1.34722328e+00 -8.93654525e-01 1.27110615e-01 1.10086992e-01
-9.60588828e-02 2.72997320e-01 8.82695079e-01 1.94143325e-01
-1.86565652e-01 5.73790252e-01 -1.69358492e-01 -1.74184188e-01
7.94271588e-01 -1.20726204e+00 2.95237452e-01 7.57348061e-01
-3.99867713e-01 8.85739148e-01 9.39806521e-01 -4.03017908e-01
-6.79421902e-01 -6.03931785e-01 1.05731761e+00 -7.31209099e-01
3.92906427e-01 3.17413568e-01 -9.75455582e-01 6.81452572e-01
6.10506535e-01 -3.83188576e-01 1.50807726e+00 3.53238881e-01
-4.63503361e-01 1.12795532e-01 -1.67347789e+00 5.28405249e-01
8.72454107e-01 -7.31047869e-01 -7.27437615e-01 -2.35070866e-02
1.28422037e-01 1.60874128e-01 -8.10898244e-01 1.41527146e-01
9.55428898e-01 -1.43111420e+00 6.28245771e-01 -7.98031628e-01
3.39344352e-01 9.68027338e-02 -4.13643330e-01 -8.27544510e-01
-7.51089692e-01 -3.08873296e-01 8.14492822e-01 1.40223086e+00
1.09630540e-01 -8.88739884e-01 5.83514094e-01 1.56045926e+00
5.17193526e-02 -5.20264566e-01 -9.90512669e-01 -6.27775192e-01
8.26062620e-01 -5.32761037e-01 7.20303297e-01 1.78291535e+00
3.10029000e-01 2.99332589e-01 -2.96413094e-01 -2.95632720e-01
6.29645586e-01 -1.71595082e-01 7.52850652e-01 -1.61412621e+00
9.60417613e-02 -9.41322744e-01 -3.82527530e-01 1.63302813e-02
8.94641653e-02 -3.76380533e-01 2.73851417e-02 -1.50537086e+00
2.40025371e-01 -7.32138097e-01 -3.36112201e-01 4.26902056e-01
-2.01555386e-01 -1.50191262e-01 6.88484669e-01 6.33463740e-01
-1.50556043e-01 3.88211384e-02 7.80165792e-01 2.22848713e-01
-1.05174445e-02 -1.36042550e-01 -1.80108273e+00 8.77155542e-01
9.36788619e-01 -6.09020591e-01 -3.47142339e-01 -4.33051795e-01
4.50076580e-01 -3.33481848e-01 3.77884537e-01 -5.39522171e-01
4.74458840e-03 -9.80297804e-01 -5.55640832e-02 5.13329208e-01
-1.98633015e-01 -1.13677979e+00 2.46257141e-01 5.20924509e-01
-5.42017519e-01 -1.03097476e-01 2.86030471e-01 9.79446769e-02
1.00911476e-01 -4.22849238e-01 1.17080343e+00 -4.55674738e-01
-4.37112600e-01 -1.62250772e-01 -3.53503615e-01 1.93962947e-01
1.01029015e+00 -6.60556018e-01 -5.44682741e-01 -3.46894175e-01
-2.43641108e-01 -3.58728543e-02 8.73818338e-01 2.48329848e-01
9.88707319e-03 -1.13000762e+00 -8.53200197e-01 -4.67002213e-01
-1.06980622e-01 -7.50803649e-01 3.28558952e-01 9.35309231e-01
-5.27092397e-01 2.69310355e-01 -4.56197619e-01 1.38243854e-01
-1.09015882e+00 3.77687305e-01 7.96722114e-01 3.25758636e-01
-6.14608586e-01 4.82734650e-01 1.49071157e-01 -3.35557580e-01
4.27956805e-02 4.15978819e-01 -2.06189454e-01 2.53464937e-01
5.82082212e-01 7.38684118e-01 -2.59817302e-01 -1.09214425e+00
-4.68315750e-01 2.70967990e-01 9.81759056e-02 -3.25231373e-01
1.06918442e+00 -2.80595183e-01 -4.68791932e-01 4.42791134e-01
8.25889766e-01 2.19847962e-01 -8.20048213e-01 1.54069051e-01
2.40379140e-01 -1.28835964e+00 2.47097965e-02 -1.01370728e+00
-7.93581963e-01 9.61966336e-01 5.20896852e-01 7.93397605e-01
1.00589871e+00 -5.45884132e-01 1.15898587e-01 -3.44960153e-01
4.12479430e-01 -1.51752436e+00 -6.67531312e-01 -1.12562813e-01
9.30338025e-01 -1.06235778e+00 2.65921384e-01 -2.38612771e-01
-9.45946574e-01 1.00755143e+00 5.26002586e-01 1.41425384e-02
5.42882442e-01 -1.58588365e-01 4.75054115e-01 -4.01233017e-01
-4.92477924e-01 -5.79852127e-02 -1.95968583e-01 4.91107792e-01
9.65076864e-01 5.44141412e-01 -1.37364268e+00 4.89189297e-01
-3.03922683e-01 1.87053129e-01 5.91448009e-01 1.07028103e+00
-7.39651799e-01 -1.17122495e+00 -5.83241761e-01 6.52047336e-01
-1.11077476e+00 1.18326977e-01 -7.67906845e-01 1.04432011e+00
5.25111556e-01 1.26996839e+00 4.98272628e-02 -1.66121438e-01
3.19486976e-01 -1.66314095e-01 2.37181425e-01 -6.17547333e-01
-1.06941807e+00 -5.85856587e-02 4.11221594e-01 -3.20961148e-01
-6.13289356e-01 -5.66083491e-01 -8.68122876e-01 -1.28073597e+00
-3.23210478e-01 4.49534714e-01 3.56350899e-01 1.09257352e+00
8.21584389e-02 1.53000444e-01 4.70926791e-01 -2.40137428e-01
-8.34738374e-01 -4.38102752e-01 -8.08037758e-01 4.40332115e-01
3.10605139e-01 -2.19550297e-01 -5.33190429e-01 -4.48508292e-01] | [8.911038398742676, 5.673133850097656] |
6c0909f6-20ac-4601-8b2f-bd681146f26a | spatiotemporal-contrastive-video | 2008.03800 | null | https://arxiv.org/abs/2008.03800v4 | https://arxiv.org/pdf/2008.03800v4.pdf | Spatiotemporal Contrastive Video Representation Learning | We present a self-supervised Contrastive Video Representation Learning (CVRL) method to learn spatiotemporal visual representations from unlabeled videos. Our representations are learned using a contrastive loss, where two augmented clips from the same short video are pulled together in the embedding space, while clips from different videos are pushed away. We study what makes for good data augmentations for video self-supervised learning and find that both spatial and temporal information are crucial. We carefully design data augmentations involving spatial and temporal cues. Concretely, we propose a temporally consistent spatial augmentation method to impose strong spatial augmentations on each frame of the video while maintaining the temporal consistency across frames. We also propose a sampling-based temporal augmentation method to avoid overly enforcing invariance on clips that are distant in time. On Kinetics-600, a linear classifier trained on the representations learned by CVRL achieves 70.4% top-1 accuracy with a 3D-ResNet-50 (R3D-50) backbone, outperforming ImageNet supervised pre-training by 15.7% and SimCLR unsupervised pre-training by 18.8% using the same inflated R3D-50. The performance of CVRL can be further improved to 72.9% with a larger R3D-152 (2x filters) backbone, significantly closing the gap between unsupervised and supervised video representation learning. Our code and models will be available at https://github.com/tensorflow/models/tree/master/official/. | ['Ming-Hsuan Yang', 'Boqing Gong', 'Serge Belongie', 'Huisheng Wang', 'Tianjian Meng', 'Rui Qian', 'Yin Cui'] | 2020-08-09 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Qian_Spatiotemporal_Contrastive_Video_Representation_Learning_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Qian_Spatiotemporal_Contrastive_Video_Representation_Learning_CVPR_2021_paper.pdf | cvpr-2021-1 | ['self-supervised-action-recognition'] | ['computer-vision'] | [ 3.92200910e-02 -1.81290969e-01 -5.09003341e-01 -2.68531054e-01
-6.47378623e-01 -5.57819128e-01 5.23089588e-01 -1.46148270e-02
-5.26967347e-01 4.53777701e-01 6.34046257e-01 -5.03516272e-02
2.32363462e-01 -3.73835355e-01 -1.08415401e+00 -6.16650343e-01
-4.97317016e-01 -1.90785423e-01 2.83945113e-01 -9.81652364e-02
-1.36096328e-01 3.47567558e-01 -1.40745127e+00 5.21674037e-01
3.69315654e-01 9.69115615e-01 2.90315486e-02 7.06082702e-01
1.15439884e-01 1.11887610e+00 -3.50807130e-01 2.37320423e-01
4.45043594e-01 -4.46505219e-01 -8.21392298e-01 1.53573662e-01
8.24118555e-01 -4.78727400e-01 -1.01467478e+00 6.75784647e-01
2.52909452e-01 5.81686139e-01 4.95035589e-01 -1.31237292e+00
-9.59550738e-01 3.35949928e-01 -8.71647894e-01 7.30110943e-01
2.86828220e-01 2.67714024e-01 8.64877105e-01 -1.04586327e+00
7.49288440e-01 1.02947927e+00 4.84273583e-01 7.56662726e-01
-1.29470313e+00 -7.45205283e-01 6.82691813e-01 3.11287582e-01
-1.38205051e+00 -4.70152706e-01 7.35877514e-01 -6.50731027e-01
9.69833374e-01 6.90349117e-02 7.19375491e-01 1.44348955e+00
-1.03349760e-01 7.69097924e-01 7.41460800e-01 -1.95254117e-01
1.59224302e-01 -2.63235599e-01 1.59282088e-01 6.15676463e-01
-1.31687194e-01 1.84777424e-01 -7.32878983e-01 1.62753597e-01
1.17282617e+00 4.51289922e-01 -2.26745978e-01 -4.32507515e-01
-1.13975167e+00 7.62610495e-01 6.60515845e-01 3.76655549e-01
-2.62617469e-01 3.85476172e-01 6.70414209e-01 3.45693082e-01
5.83508730e-01 3.09260875e-01 -4.87509489e-01 -1.61187455e-01
-1.01381671e+00 1.16701446e-01 1.12806059e-01 1.00182033e+00
7.22026169e-01 4.02456462e-01 -3.37376148e-01 6.51260555e-01
1.05517991e-02 1.81923971e-01 6.61318421e-01 -1.16823137e+00
3.77776802e-01 3.34729642e-01 -1.22806944e-01 -8.80504668e-01
-1.86709493e-01 -3.99828643e-01 -7.37235010e-01 1.50924459e-01
3.79601657e-01 -1.42055064e-01 -1.13887703e+00 1.95143700e+00
1.12897390e-02 7.28454411e-01 -1.79757223e-01 1.06474447e+00
8.35932374e-01 9.45004225e-01 3.36837769e-01 -2.00265691e-01
1.06788278e+00 -1.27759421e+00 -6.50897920e-01 -1.39022201e-01
7.81335473e-01 -2.87108421e-01 1.22931099e+00 9.89836678e-02
-1.11943388e+00 -8.28205287e-01 -1.00598443e+00 -2.26865053e-01
-2.42549852e-01 4.93190922e-02 5.77209592e-01 -6.14303257e-03
-1.12441814e+00 5.85582495e-01 -1.00737274e+00 -3.25694710e-01
5.71341932e-01 1.91198364e-01 -7.62491465e-01 -1.39691502e-01
-1.07178962e+00 4.70815331e-01 1.56800374e-01 -2.11357981e-01
-1.28401458e+00 -1.04585803e+00 -1.00912786e+00 -1.67532369e-01
3.12736660e-01 -3.21570843e-01 9.67973769e-01 -1.36998749e+00
-1.11542833e+00 9.46608126e-01 -2.89449871e-01 -6.51014864e-01
3.82067204e-01 -5.30810893e-01 -4.68139559e-01 5.15756011e-01
2.56807357e-01 9.89217579e-01 8.97038519e-01 -1.09393537e+00
-2.16418087e-01 -1.83759704e-01 2.09717840e-01 2.66559005e-01
-5.03732026e-01 -8.28232467e-02 -7.84493327e-01 -1.16422558e+00
-1.08897693e-01 -1.00279534e+00 -2.73697615e-01 2.58257627e-01
-9.00116339e-02 9.72391963e-02 9.97664809e-01 -8.37094188e-01
1.17207193e+00 -2.35575294e+00 2.61949480e-01 -7.31225386e-02
2.27617383e-01 2.81522363e-01 -6.14652932e-01 1.58143297e-01
-5.55738449e-01 1.11709744e-01 -1.32218897e-01 -4.15594429e-01
-5.28258681e-01 1.65894255e-01 -4.02416259e-01 5.47200501e-01
4.31859702e-01 9.13273215e-01 -1.06608880e+00 -3.08819205e-01
2.83948362e-01 7.23588705e-01 -8.49591255e-01 3.17775488e-01
-1.47327021e-01 6.62513256e-01 -1.83392778e-01 5.29664874e-01
3.93022478e-01 -3.52348030e-01 6.24146536e-02 -3.07055831e-01
-1.04201689e-01 2.19961688e-01 -9.00145292e-01 2.20503855e+00
-2.03975558e-01 8.47199857e-01 -1.70714781e-01 -1.09649599e+00
7.33949542e-01 2.92380631e-01 9.11410153e-01 -8.42406034e-01
-1.25586033e-01 -4.17494535e-01 -2.85268277e-01 -5.23199677e-01
3.94547313e-01 1.70600936e-01 1.48095816e-01 3.04608911e-01
3.65785658e-01 5.17396092e-01 1.93045124e-01 4.87746805e-01
1.22771573e+00 5.04116476e-01 -1.20081916e-01 -1.29147768e-01
2.34527111e-01 -1.45189419e-01 7.15507984e-01 5.85026503e-01
-4.41145957e-01 8.52979362e-01 4.22370613e-01 -6.37131751e-01
-1.09812820e+00 -1.10848248e+00 1.13948748e-01 1.38673306e+00
8.45183153e-03 -7.75207698e-01 -4.86940891e-01 -8.82235646e-01
-1.09789111e-02 3.43315780e-01 -9.54069078e-01 -3.02695125e-01
-6.79748416e-01 -2.89955795e-01 4.55405802e-01 9.22656476e-01
4.20297593e-01 -8.03558052e-01 -3.02567154e-01 5.69690242e-02
-1.67587355e-01 -1.17299485e+00 -7.46297181e-01 9.56821516e-02
-9.42153752e-01 -9.29188907e-01 -8.14076900e-01 -7.66371846e-01
7.73893297e-01 5.88979125e-01 8.33034754e-01 1.13058612e-01
-2.49300569e-01 6.26608193e-01 -6.32476032e-01 2.04912007e-01
8.94690081e-02 -6.84283599e-02 3.14832747e-01 7.90814981e-02
2.78336704e-01 -7.70150244e-01 -7.92045057e-01 3.14754784e-01
-9.48082864e-01 1.34206131e-01 1.66303962e-01 7.81719267e-01
7.56570220e-01 -4.77940142e-01 3.30971777e-01 -5.43299079e-01
1.86772384e-02 -7.73194730e-01 -2.70712405e-01 -9.54307094e-02
-1.65641636e-01 -3.95199433e-02 5.35695255e-01 -8.63737524e-01
-6.44394636e-01 1.47436038e-01 6.94194660e-02 -1.32059336e+00
-1.16869144e-01 3.29467624e-01 1.16804346e-01 4.87927385e-02
7.26785243e-01 2.57583827e-01 2.41481677e-01 -3.93204242e-01
5.12976170e-01 1.03268199e-01 5.54919958e-01 -6.21460140e-01
7.88724184e-01 5.70146501e-01 -3.49368364e-01 -9.23559606e-01
-8.78884435e-01 -5.21775603e-01 -7.13331640e-01 -3.72012109e-01
1.02583492e+00 -1.35858297e+00 -3.41137409e-01 1.75515354e-01
-7.94264615e-01 -8.09566796e-01 -5.47208905e-01 6.41182959e-01
-5.90602994e-01 4.28024650e-01 -8.04369032e-01 -3.64155531e-01
2.43830848e-02 -8.70944262e-01 7.81974256e-01 1.00455500e-01
-4.03560549e-01 -8.43797982e-01 2.52591431e-01 3.73560339e-01
3.88891935e-01 2.71048099e-01 3.60325396e-01 -5.70213854e-01
-4.25269276e-01 8.57404023e-02 -1.10243067e-01 5.70405960e-01
2.42719099e-01 8.03992525e-02 -9.68174040e-01 -5.66026092e-01
-3.83072317e-01 -4.93105948e-01 1.20181811e+00 5.83274961e-01
1.49333930e+00 -4.32958424e-01 -1.49205118e-01 9.22670841e-01
1.16434801e+00 8.57117251e-02 7.26072311e-01 3.87150198e-01
9.62687612e-01 3.94010514e-01 4.98304695e-01 4.54666525e-01
1.68269545e-01 7.32775390e-01 4.50997293e-01 -2.79764742e-01
-3.55969816e-01 -4.76979852e-01 7.01737106e-01 6.39250219e-01
-3.94687116e-01 7.76201189e-02 -6.45864367e-01 6.12005651e-01
-1.87380946e+00 -1.29304755e+00 2.37428963e-01 2.15137196e+00
8.33225965e-01 3.93575914e-02 4.26814377e-01 4.62009571e-02
6.09016776e-01 6.13001347e-01 -6.11175120e-01 -1.46950223e-02
-1.53271928e-01 1.35531098e-01 4.45230544e-01 5.08553624e-01
-1.45334578e+00 1.08199990e+00 5.99394989e+00 5.81260681e-01
-1.31874084e+00 1.75417930e-01 7.10017920e-01 -5.99339545e-01
-1.61461577e-01 -5.49410656e-02 -5.03181636e-01 6.42919838e-01
9.27271664e-01 9.16598886e-02 4.10753399e-01 8.07894647e-01
5.03930509e-01 1.65027991e-01 -1.17154086e+00 1.07915628e+00
1.46958530e-01 -1.67454207e+00 2.01415882e-01 -9.34345797e-02
8.66165042e-01 1.20286308e-01 1.97715059e-01 4.41998482e-01
6.06837347e-02 -1.07067728e+00 6.82181239e-01 5.06682813e-01
9.06493723e-01 -4.01827425e-01 2.05518216e-01 -1.67894244e-01
-1.40152609e+00 -5.75944372e-02 -1.95722669e-01 -1.58336386e-01
9.10208598e-02 2.55371660e-01 -2.24721462e-01 2.29649931e-01
1.01256740e+00 1.28257215e+00 -5.57788432e-01 8.00608873e-01
-1.20474607e-01 7.58162081e-01 -9.95465517e-02 5.78505278e-01
3.26856881e-01 -1.69955730e-01 5.24130821e-01 1.41116190e+00
2.21809465e-02 1.66559696e-01 2.52741069e-01 6.51927888e-01
-2.48810679e-01 1.40492793e-03 -8.14186811e-01 -1.08918428e-01
3.98845941e-01 9.99835134e-01 -3.65412325e-01 -4.77316320e-01
-6.06786609e-01 1.23582399e+00 4.39382374e-01 7.79566705e-01
-1.09523487e+00 2.04172600e-02 9.93055165e-01 3.82524997e-01
3.45080435e-01 -4.84533340e-01 1.61063850e-01 -1.35514045e+00
2.74650771e-02 -7.63818502e-01 5.73207974e-01 -8.80048692e-01
-1.17722905e+00 4.71240103e-01 6.79992959e-02 -1.57238114e+00
-1.87573597e-01 -5.12351930e-01 -6.06711626e-01 4.49338287e-01
-1.36108613e+00 -1.00512648e+00 -3.88367534e-01 8.86063814e-01
7.50184894e-01 -2.06303313e-01 6.90197408e-01 4.81718004e-01
-6.81160569e-01 8.56156468e-01 -9.05509517e-02 5.21420300e-01
7.94398129e-01 -8.98079813e-01 1.85147539e-01 8.64955664e-01
3.18148285e-01 6.33815408e-01 4.07867044e-01 -4.54617828e-01
-1.35966074e+00 -1.45598888e+00 4.63367701e-01 -3.82693201e-01
7.05151737e-01 -3.02184194e-01 -1.24274409e+00 1.09477949e+00
2.32063964e-01 7.01160014e-01 7.62812138e-01 4.22536302e-03
-9.28791404e-01 -1.90817282e-01 -8.74896169e-01 6.60141587e-01
1.35700452e+00 -8.24156404e-01 -4.37380016e-01 3.80072504e-01
9.31329310e-01 -3.07116121e-01 -9.56604600e-01 3.42525840e-01
4.71103728e-01 -7.75320292e-01 1.07613146e+00 -1.00282967e+00
5.20818114e-01 -4.33146477e-01 -2.27913022e-01 -1.10147262e+00
-5.15284777e-01 -6.73609614e-01 -4.21466887e-01 9.56565738e-01
2.20037237e-01 -2.72141278e-01 8.68502617e-01 4.04804468e-01
-1.44425184e-01 -6.96428597e-01 -7.90104389e-01 -9.58029985e-01
1.74267039e-01 -4.91128653e-01 -1.50503740e-02 1.26674008e+00
1.42420158e-01 1.96886703e-01 -5.03751695e-01 1.18295521e-01
4.25460815e-01 -3.43619496e-01 5.93743682e-01 -6.61584854e-01
-2.57782608e-01 -2.67939270e-01 -5.61793268e-01 -1.23835480e+00
2.73925990e-01 -8.77006829e-01 -3.45942795e-01 -1.20096028e+00
3.28605443e-01 -2.39047900e-01 -6.86661839e-01 8.24983656e-01
3.95123009e-03 4.93290842e-01 3.29345435e-01 3.43831539e-01
-7.26252198e-01 7.09946573e-01 1.09745169e+00 -3.07939410e-01
-3.14024627e-01 -5.26767015e-01 -4.88723993e-01 5.63209414e-01
8.96100104e-01 -2.35781640e-01 -6.61958098e-01 -5.60967147e-01
-2.17511654e-01 -4.90028821e-02 5.15250862e-01 -1.06870961e+00
5.01082130e-02 -7.62885138e-02 7.04542637e-01 -3.27228785e-01
5.01679540e-01 -6.13679647e-01 2.41515823e-02 2.21506536e-01
-7.18352497e-01 2.10790366e-01 5.00218451e-01 5.85181773e-01
-3.11274648e-01 3.17889899e-01 7.92910755e-01 -6.21133037e-02
-9.95454669e-01 7.18569696e-01 -3.68356347e-01 1.52634680e-01
1.17380548e+00 -2.36083537e-01 -3.52497637e-01 -5.42525768e-01
-1.09678209e+00 1.64888456e-01 5.86994648e-01 7.02136278e-01
8.06327641e-01 -1.60654819e+00 -5.96199512e-01 3.40496957e-01
1.69069409e-01 -2.18012378e-01 5.96348763e-01 7.85997629e-01
-3.64495933e-01 1.80318087e-01 -4.35117066e-01 -6.14743233e-01
-1.12932003e+00 7.45561302e-01 3.19583505e-01 4.69965637e-02
-8.44025254e-01 9.11689878e-01 3.94015878e-01 9.32925940e-02
4.89905506e-01 -2.70793438e-01 -2.04881445e-01 -2.29170546e-02
7.84192443e-01 2.67466903e-01 -2.88105696e-01 -7.87925780e-01
-4.44474310e-01 5.96418083e-01 -3.26132119e-01 -1.04579404e-01
1.37371647e+00 8.41691345e-02 2.38325641e-01 4.17889446e-01
1.56195819e+00 -9.92543399e-02 -1.90498567e+00 -3.39358747e-01
-3.42769921e-01 -4.69081372e-01 4.44071554e-02 -3.79643887e-01
-1.45386827e+00 7.64102697e-01 6.91658437e-01 -1.77114397e-01
1.02910638e+00 7.45628327e-02 4.75011945e-01 3.33667211e-02
1.72162279e-02 -9.84167814e-01 6.79112017e-01 5.49482822e-01
1.12459528e+00 -1.13272774e+00 5.95142180e-03 -1.21747352e-01
-9.53691661e-01 9.00952220e-01 8.70054483e-01 -5.34825325e-01
6.81268334e-01 1.61801279e-01 -9.04334933e-02 -1.35896832e-01
-8.38613510e-01 -1.82692140e-01 3.06788832e-01 6.80458307e-01
6.49761617e-01 -2.06190526e-01 2.06945166e-01 4.12314415e-01
3.14717770e-01 -9.47996508e-03 3.79879594e-01 9.59857106e-01
-4.88138832e-02 -7.29478300e-01 -3.03598959e-02 3.12429100e-01
-3.10311198e-01 -1.36118278e-01 -3.57956393e-03 8.21287990e-01
3.66663933e-02 6.37357771e-01 4.84084398e-01 -6.52007878e-01
2.32354268e-01 3.27389836e-02 4.31202471e-01 -5.65055251e-01
-2.31943652e-01 2.01244891e-01 -5.13680242e-02 -9.45246935e-01
-5.79336762e-01 -6.60611153e-01 -1.36391556e+00 -2.78290480e-01
3.24610531e-01 -8.34925845e-03 1.57955363e-01 5.02119720e-01
6.43100023e-01 5.81985056e-01 6.97450817e-01 -1.21427226e+00
-9.92790535e-02 -8.79217386e-01 -4.64500785e-01 6.14693463e-01
5.53129792e-01 -7.92526245e-01 -4.35818970e-01 4.91558969e-01] | [8.81612777709961, 0.7590177655220032] |
dd29d3ea-c756-4da2-b7f5-06a33400c44d | boltzmann-exploration-expectationmaximisation | null | null | https://arxiv.org/abs/1912.08869 | https://arxiv.org/pdf/1912.08869.pdf | Boltzmann Exploration Expectation–Maximisation | We present a general method for fitting finite mixture models (FMM). Learning
in a mixture model consists of finding the most likely cluster assignment for each
data-point, as well as finding the parameters of the clusters themselves. In many
mixture models, this is difficult with current learning methods, where the most common approach is to employ monotone learning algorithms e.g. the conventional expectation-maximisation algorithm. While effective, the success of any
monotone algorithm is crucially dependant on good parameter initialisation, where
a common choice is K-means initialisation, commonly employed for Gaussian
mixture models.
For other types of mixture models, the path to good initialisation parameters is often
unclear and may require a problem-specific solution. To this end, we propose a
general heuristic learning algorithm that utilises Boltzmann exploration to assign
each observation to a specific base distribution within the mixture model, which we
call Boltzmann exploration expectation-maximisation (BEEM). With BEEM, hard
assignments allow straight forward parameter learning for each base distribution
by conditioning only on its assigned observations. Consequently, it can be applied
to mixtures of any base distribution where single component parameter learning is
tractable. The stochastic learning procedure is able to escape local optima and is
thus insensitive to parameter initialisation. We show competitive performance on a
number of synthetic benchmark cases as well as on real-world datasets. | ['Neil Dhir', 'Mathias Edman'] | 2019-12-18 | null | null | null | arxiv-2019-12 | ['iris-segmentation'] | ['medical'] | [ 3.00202012e-01 8.77906801e-04 -2.32961208e-01 -1.86030884e-04
-1.04777896e+00 -5.35130382e-01 7.38048911e-01 2.78567344e-01
-7.43934929e-01 6.24786258e-01 -3.67181122e-01 -4.20750797e-01
-4.77610916e-01 -7.45098054e-01 -6.74415290e-01 -1.31483138e+00
-1.00783177e-01 1.17668366e+00 1.85221031e-01 2.29248583e-01
2.59395421e-01 4.61557239e-01 -1.33740163e+00 -1.30721614e-01
8.59242201e-01 7.17637897e-01 3.60223591e-01 7.17362225e-01
-4.36399907e-01 1.46255881e-01 -4.91563231e-01 -3.39745849e-01
1.69332206e-01 -4.64956075e-01 -7.78489709e-01 3.07729691e-01
-2.83758074e-01 2.83326149e-01 4.15272802e-01 1.03763020e+00
3.73655111e-01 4.41476256e-01 1.28825986e+00 -1.38903368e+00
2.70942658e-01 5.17132521e-01 -7.71713674e-01 -1.05030961e-01
-1.17737345e-01 1.20586783e-01 7.53002048e-01 -5.26469529e-01
7.75028989e-02 1.13762653e+00 5.16947925e-01 5.13480425e-01
-1.79921925e+00 -4.88420039e-01 1.62257832e-02 1.94880292e-01
-1.58912289e+00 -5.08280814e-01 6.42263651e-01 -3.88584882e-01
6.28663838e-01 3.98533404e-01 6.12383902e-01 7.88888395e-01
7.83414096e-02 7.79052317e-01 1.28145671e+00 -6.47327960e-01
1.00301862e+00 2.10837662e-01 -1.83543846e-01 3.77139956e-01
1.07925097e-02 -1.81788012e-01 2.21085846e-02 -4.50765818e-01
6.23909831e-01 -1.53910339e-01 -2.32293904e-01 -9.98947501e-01
-8.81581664e-01 1.08781600e+00 1.73442334e-01 7.22825602e-02
-3.83620232e-01 4.52070087e-02 2.10175231e-01 5.19600399e-02
3.49746197e-01 4.18501288e-01 -4.14374501e-01 -2.41044596e-01
-1.25382626e+00 4.73606527e-01 9.47977543e-01 3.20515990e-01
9.27328110e-01 -4.08446431e-01 2.22708240e-01 1.08818305e+00
6.28878295e-01 2.79408753e-01 5.14519155e-01 -9.37319100e-01
1.13838449e-01 3.93785179e-01 2.76735485e-01 -4.73830789e-01
-2.45801657e-01 -3.74616355e-01 -1.20070887e+00 5.06234825e-01
7.43127882e-01 -1.50501192e-01 -1.07107294e+00 1.84536457e+00
6.02557421e-01 3.71581107e-01 1.04712375e-01 7.15359747e-01
7.23684803e-02 7.60834098e-01 -6.44375198e-03 -5.59733689e-01
9.52613533e-01 -7.22179949e-01 -2.89929748e-01 -3.92244041e-01
5.43865561e-01 -5.93898475e-01 7.76805460e-01 6.71014190e-01
-9.58504796e-01 -1.82022005e-01 -8.09002817e-01 6.49871767e-01
-2.82704324e-01 -4.26259160e-01 4.38130438e-01 8.54835093e-01
-9.33192670e-01 6.88000739e-01 -1.08239996e+00 -2.92389125e-01
4.05945033e-01 7.23297656e-01 -1.85342684e-01 -1.33122683e-01
-8.05944860e-01 8.64660859e-01 8.90042245e-01 1.14765257e-01
-7.35564053e-01 -4.17109013e-01 -6.85732782e-01 -9.88931730e-02
5.53406000e-01 -7.74468839e-01 1.23994517e+00 -7.59429514e-01
-1.74917352e+00 5.39386272e-01 -2.77153850e-01 -5.60543537e-01
5.93158603e-01 7.65021890e-02 -2.20970940e-02 -2.12172449e-01
-3.03974897e-01 8.29921067e-01 1.29986370e+00 -1.41617489e+00
-4.05406743e-01 9.89507418e-03 -3.59770894e-01 2.47670889e-01
-7.40314201e-02 -6.58405200e-02 -4.63193625e-01 -1.66738003e-01
2.92136043e-01 -1.11080337e+00 -8.27632129e-01 -4.74062502e-01
-3.70260358e-01 -3.44779521e-01 4.99761760e-01 -3.13539565e-01
1.29200447e+00 -1.99061215e+00 4.74665999e-01 6.68315053e-01
9.60943401e-02 1.68414533e-01 1.39560595e-01 3.66628498e-01
-1.21158212e-01 -1.26662448e-01 -7.39763975e-01 -5.98724484e-01
2.68788695e-01 3.78598392e-01 -2.05232725e-01 6.96703017e-01
4.65220772e-02 7.55357802e-01 -7.36871004e-01 -5.69379807e-01
4.25525069e-01 3.60156596e-01 -7.12170005e-01 2.79462457e-01
-4.20042217e-01 3.89664233e-01 -6.43110573e-02 9.99905616e-02
7.28872120e-01 -1.96790174e-01 1.85646638e-01 2.52395570e-01
5.00822403e-02 -2.37446930e-02 -1.86507630e+00 1.30484164e+00
-4.86576766e-01 1.37752429e-01 2.70515770e-01 -1.16216469e+00
8.76246870e-01 1.98838785e-01 4.89388674e-01 2.03000288e-02
-3.50319371e-02 4.32698488e-01 9.29791629e-02 -1.40142933e-01
5.17582744e-02 -7.02092052e-01 -8.17310903e-03 6.02757335e-01
-7.51774609e-02 -4.32360679e-01 7.57184550e-02 -7.69799203e-02
7.59047210e-01 9.56058875e-02 5.13047397e-01 -1.91879213e-01
3.31158757e-01 -2.71352202e-01 3.47260386e-01 8.82510185e-01
2.05222502e-01 5.09132862e-01 3.86565924e-01 -1.92947000e-01
-9.51300859e-01 -1.12965536e+00 -3.39679956e-01 8.77130687e-01
-6.48443773e-02 -5.04100502e-01 -1.06042302e+00 -5.60270905e-01
-3.35453182e-01 6.76717401e-01 -4.51004654e-01 -2.96913594e-01
-3.57956976e-01 -1.28547120e+00 -1.85030252e-02 4.50241119e-02
2.46558368e-01 -1.17708290e+00 -7.84130394e-01 4.27475989e-01
-6.55436441e-02 -4.58210021e-01 -1.17182210e-01 8.78837526e-01
-1.05512631e+00 -8.25715780e-01 -8.88640463e-01 -4.52575892e-01
5.12222588e-01 -1.82489276e-01 9.93461013e-01 -2.14299619e-01
-6.59101456e-02 1.57776549e-01 3.68462019e-02 -2.79060364e-01
-6.28660440e-01 4.51993674e-01 8.63365605e-02 1.41029269e-01
3.44044447e-01 -6.78375959e-01 -3.53841662e-01 3.51506084e-01
-1.05536127e+00 -3.75543982e-02 7.53354192e-01 6.77619576e-01
6.47437751e-01 6.47651374e-01 4.69253093e-01 -7.13411570e-01
6.37291789e-01 -6.92472875e-01 -4.95112062e-01 8.42508301e-02
-4.64514345e-01 1.73172474e-01 5.26107013e-01 -6.37654960e-01
-6.30124390e-01 2.98117369e-01 -2.60602176e-01 -4.53032374e-01
-7.19416916e-01 4.41873878e-01 -5.54718435e-01 2.20216542e-01
5.93790293e-01 3.42473716e-01 5.29986210e-02 -5.42356968e-01
3.75870377e-01 8.01603019e-01 5.91768682e-01 -8.19012523e-01
5.88312328e-01 1.61302641e-01 1.29933074e-01 -1.02811158e+00
-4.58334953e-01 -6.47021413e-01 -8.16418588e-01 -1.95908874e-01
7.07280636e-01 -4.75574344e-01 -5.90301275e-01 6.34379804e-01
-7.43594766e-01 -7.20244348e-01 -2.22301379e-01 3.57682228e-01
-9.35121179e-01 4.84960616e-01 -1.58088490e-01 -1.20433867e+00
7.68048614e-02 -1.25086212e+00 1.01040089e+00 7.25018457e-02
-5.45917153e-01 -1.26531386e+00 3.16541404e-01 1.80595681e-01
2.52840906e-01 7.17062131e-02 1.20771360e+00 -8.24772716e-01
-2.94481277e-01 -2.02381134e-01 3.76640648e-01 1.21187977e-01
5.82177825e-02 -8.93833872e-04 -8.16591561e-01 -5.03894687e-01
1.77438781e-01 -2.42601380e-01 1.02441049e+00 5.96948326e-01
1.06757033e+00 -3.96125019e-01 -6.20186388e-01 4.81560379e-01
1.23814261e+00 1.22162230e-01 6.01920307e-01 5.19952118e-01
4.40305024e-01 7.39769697e-01 2.65739113e-01 3.12912852e-01
1.30504727e-01 7.87657022e-01 4.44402933e-01 -7.94264860e-03
6.04891360e-01 3.43707087e-03 2.36328989e-01 5.66998482e-01
1.60966486e-01 -2.22967312e-01 -1.01059556e+00 4.88523155e-01
-2.00443006e+00 -7.93091893e-01 3.90818976e-02 2.58404422e+00
9.88666832e-01 3.90352428e-01 6.14882886e-01 4.68825281e-01
7.02079594e-01 -1.05638757e-01 -6.48353636e-01 -2.53992558e-01
8.79095867e-02 2.49122009e-01 1.70879826e-01 6.56484962e-01
-1.26910710e+00 7.45346844e-01 5.62169409e+00 1.40847290e+00
-7.97807217e-01 1.99213549e-02 6.67864144e-01 -1.89715177e-01
-4.05324660e-02 3.54277492e-02 -7.18873978e-01 7.17128098e-01
1.08412004e+00 2.18704209e-01 6.76925361e-01 5.82797706e-01
1.30996734e-01 -5.16890168e-01 -1.09218276e+00 1.12012899e+00
-3.86093140e-01 -9.25941944e-01 -2.99802542e-01 3.73796165e-01
6.06088817e-01 -2.40242541e-01 4.32092100e-02 3.44253719e-01
4.54907686e-01 -1.26331437e+00 4.86865848e-01 3.67306352e-01
4.34902936e-01 -1.24556184e+00 5.85416138e-01 1.03410375e+00
-7.44296193e-01 -4.08020578e-02 -4.53344643e-01 1.93098173e-01
2.12338194e-01 7.66089082e-01 -8.74765873e-01 1.82337448e-01
4.99642819e-01 3.10539324e-02 -2.83397585e-01 1.60474873e+00
-2.76853666e-02 8.26788604e-01 -9.28016186e-01 4.73631993e-02
3.91019166e-01 -5.71783304e-01 4.67855901e-01 1.36072934e+00
3.08570445e-01 -1.96229711e-01 2.38329560e-01 6.58469975e-01
2.97976792e-01 2.22245738e-01 -2.59618938e-01 1.59950659e-01
4.13098991e-01 1.21088743e+00 -1.25758576e+00 -1.72234267e-01
1.74328223e-01 8.84023428e-01 5.51797211e-01 4.25829738e-01
-5.97365081e-01 2.35728230e-02 5.56542635e-01 5.45584895e-02
5.27359426e-01 -8.90882760e-02 -1.42558351e-01 -6.75259531e-01
-2.28554890e-01 -9.64692831e-01 5.10626793e-01 -2.29083955e-01
-1.22523534e+00 3.75535131e-01 4.22339052e-01 -8.23042393e-01
-6.68592155e-01 -4.19297487e-01 -8.03199470e-01 9.59204555e-01
-1.14418721e+00 -6.76125050e-01 2.42351502e-01 4.88504350e-01
3.67385417e-01 7.19192326e-02 8.29647541e-01 -1.77943349e-01
-6.78640068e-01 2.77982533e-01 5.32060504e-01 -4.07955676e-01
5.11748374e-01 -1.58443189e+00 8.52909014e-02 5.17487586e-01
4.43355083e-01 5.80011547e-01 1.03073204e+00 -5.23955882e-01
-1.00125015e+00 -9.03466523e-01 4.63022113e-01 -2.53968894e-01
4.89970535e-01 -6.33239508e-01 -1.21491861e+00 4.31103587e-01
7.67198280e-02 -2.51440793e-01 7.96730638e-01 1.50659353e-01
1.15649551e-01 2.56906390e-01 -1.05152810e+00 7.05902934e-01
4.30655360e-01 -1.23552419e-01 -3.67402703e-01 2.51426160e-01
5.85726742e-03 -1.99743807e-01 -6.92661822e-01 3.13498437e-01
2.11609215e-01 -9.43254888e-01 8.90852988e-01 -5.47026217e-01
3.54708321e-02 -4.67618853e-01 -1.43971024e-02 -1.73962164e+00
-4.04098094e-01 -8.09053659e-01 -5.83733559e-01 1.29823112e+00
4.39667374e-01 -4.80441689e-01 8.54358852e-01 6.59800649e-01
4.71631676e-01 -1.16482460e+00 -1.16282904e+00 -7.69480228e-01
3.78437996e-01 -7.49485850e-01 6.54608309e-01 6.61494374e-01
-1.17190205e-01 4.66819435e-01 -3.21353436e-01 1.76493272e-01
8.38009477e-01 2.71440446e-01 8.87946367e-01 -1.22406065e+00
-7.92206705e-01 -9.00741279e-01 -1.77986622e-01 -1.08619964e+00
3.59208733e-01 -8.29061031e-01 3.42143476e-01 -1.37419724e+00
3.23632598e-01 -7.75111496e-01 -1.59937650e-01 2.20066637e-01
-3.80148411e-01 1.78979114e-01 6.32924810e-02 2.41958782e-01
-7.73507297e-01 5.42570770e-01 6.62376821e-01 4.02963953e-03
-5.14677048e-01 6.18545234e-01 -4.26047772e-01 8.26154947e-01
8.78185332e-01 -6.82627857e-01 -4.31150466e-01 2.69564897e-01
1.63434580e-01 -6.85240105e-02 3.18207949e-01 -9.08953428e-01
3.11769456e-01 -2.73600608e-01 4.15189892e-01 -4.41961497e-01
4.09493983e-01 -8.00148845e-01 5.27591765e-01 2.92734176e-01
-2.13923261e-01 -3.34970087e-01 1.56164512e-01 6.62996292e-01
1.62099883e-01 -9.02680576e-01 1.04249454e+00 -1.72190115e-01
-4.69375998e-01 1.31339252e-01 -6.27602816e-01 1.93130188e-02
1.05771232e+00 -3.79702508e-01 4.83566165e-01 -4.34539407e-01
-8.88101220e-01 2.16345906e-01 6.97236240e-01 -2.11697817e-02
3.61798435e-01 -1.15702224e+00 -5.26989400e-01 2.51805395e-01
-1.76211372e-01 5.28378129e-01 3.19072343e-02 9.26904380e-01
-4.12866138e-02 1.48610964e-01 2.73184508e-01 -8.91777456e-01
-1.01315677e+00 8.26729596e-01 5.29239297e-01 -3.30291122e-01
-4.16077286e-01 7.39987314e-01 1.84626564e-01 -5.15412867e-01
2.60136276e-01 -5.47283702e-02 -1.54120788e-01 7.08280951e-02
4.05805439e-01 4.21760350e-01 1.89871546e-02 -5.27290165e-01
-2.87783206e-01 4.58255112e-01 2.41996767e-03 -4.37772542e-01
1.28597224e+00 -8.86962265e-02 -1.90994576e-01 7.54010916e-01
1.12684667e+00 -1.88622683e-01 -1.51696444e+00 -8.64474177e-02
3.49009871e-01 -2.56212741e-01 7.86249191e-02 -5.21460891e-01
-7.46743858e-01 8.44191730e-01 4.53910500e-01 4.68032181e-01
9.67461109e-01 1.60647243e-01 4.08777863e-01 3.63624662e-01
3.11116695e-01 -9.05491829e-01 -2.46583089e-01 2.71305978e-01
5.03733158e-01 -1.15902090e+00 -1.68985844e-01 -4.94152568e-02
-4.49916512e-01 1.02112734e+00 3.70639473e-01 1.25653848e-01
7.49218702e-01 3.90421242e-01 1.73610151e-02 -4.38067466e-02
-6.71523690e-01 -8.17145780e-02 2.80123681e-01 4.90662843e-01
6.57815710e-02 1.27499074e-01 1.85409382e-01 3.53222281e-01
-3.31295609e-01 -3.02168339e-01 9.09319520e-02 7.97845066e-01
-5.83341897e-01 -1.29947305e+00 -7.60071754e-01 5.07724762e-01
-3.67393166e-01 7.93853998e-02 -2.43851677e-01 5.20940661e-01
1.05528355e-01 8.82234633e-01 1.26981661e-01 9.57513377e-02
-7.20541254e-02 5.40647388e-01 6.46535099e-01 -5.28400123e-01
-2.63216376e-01 5.55734277e-01 -3.36556405e-01 -1.73923478e-01
-3.84351403e-01 -9.72458661e-01 -1.09559011e+00 -1.75816968e-01
-4.93553340e-01 5.10525465e-01 6.84967101e-01 1.17689729e+00
-1.06490694e-01 7.73540959e-02 4.02318925e-01 -1.18886054e+00
-6.86809182e-01 -8.60454917e-01 -6.18222535e-01 1.03457570e-01
3.03222299e-01 -6.70248449e-01 -4.27941114e-01 -1.33881420e-01] | [6.646039962768555, 3.8023557662963867] |
aa70f926-cfe3-4054-952d-ddd8a32068d7 | visual-attention-methods-in-deep-learning-an | 2204.07756 | null | https://arxiv.org/abs/2204.07756v2 | https://arxiv.org/pdf/2204.07756v2.pdf | Visual Attention Methods in Deep Learning: An In-Depth Survey | Inspired by the human cognitive system, attention is a mechanism that imitates the human cognitive awareness about specific information, amplifying critical details to focus more on the essential aspects of data. Deep learning has employed attention to boost performance for many applications. Interestingly, the same attention design can suit processing different data modalities and can easily be incorporated into large networks. Furthermore, multiple complementary attention mechanisms can be incorporated in one network. Hence, attention techniques have become extremely attractive. However, the literature lacks a comprehensive survey specific to attention techniques to guide researchers in employing attention in their deep models. Note that, besides being demanding in terms of training data and computational resources, transformers only cover a single category in self-attention out of the many categories available. We fill this gap and provide an in-depth survey of 50 attention techniques categorizing them by their most prominent features. We initiate our discussion by introducing the fundamental concepts behind the success of attention mechanism. Next, we furnish some essentials such as the strengths and limitations of each attention category, describe their fundamental building blocks, basic formulations with primary usage, and applications specifically for computer vision. We also discuss the challenges and open questions related to attention mechanism in general. Finally, we recommend possible future research directions for deep attention. | ['Ajmal Mian', 'Fahad S Khan', 'Ibrahim Radwan', 'Saeed Anwar', 'Mohammed Hassanin'] | 2022-04-16 | null | null | null | null | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [ 1.27546825e-02 2.00562969e-01 -2.69563884e-01 -1.06036566e-01
-2.16931432e-01 -2.89586723e-01 2.64353871e-01 1.07652682e-03
-4.54962760e-01 3.51740003e-01 3.04023415e-01 -1.66172162e-01
-2.69305378e-01 -6.63919389e-01 -3.25174391e-01 -5.70360363e-01
6.35112077e-02 7.88249895e-02 -5.60694300e-02 -1.60823137e-01
6.13679171e-01 5.16353071e-01 -1.76243305e+00 1.31225392e-01
8.74832213e-01 1.07080388e+00 6.12320960e-01 4.31980222e-01
-3.23709965e-01 9.41166878e-01 -7.84693301e-01 -4.64306742e-01
-1.90084487e-01 -2.62274146e-01 -1.11175704e+00 3.51902246e-02
4.32747304e-01 -1.10667489e-01 -5.29496551e-01 1.09189439e+00
6.66866958e-01 5.07550478e-01 5.50919473e-01 -1.22088361e+00
-1.54856944e+00 2.82316625e-01 -4.46414977e-01 1.08346832e+00
1.46850482e-01 5.42231230e-03 1.24277532e+00 -1.10881162e+00
5.64013384e-02 1.24878955e+00 4.43698734e-01 6.75403476e-01
-5.96797466e-01 -3.17702860e-01 6.64566815e-01 8.28674197e-01
-1.10316098e+00 -1.62179843e-01 8.09777617e-01 -3.51978749e-01
1.31542099e+00 3.00006151e-01 6.89499676e-01 8.58034909e-01
2.85670698e-01 1.26966023e+00 7.38308787e-01 -4.87084925e-01
-1.95782259e-02 -1.83985140e-02 7.51212955e-01 4.00914907e-01
2.08287939e-01 -2.40160912e-01 -3.39001507e-01 2.76675552e-01
7.96646416e-01 3.27250540e-01 -2.02351958e-01 4.25719395e-02
-9.83551323e-01 1.00621903e+00 1.04114079e+00 5.49276471e-01
-7.22769558e-01 -2.78927833e-02 3.83752137e-01 4.49615829e-02
2.29680941e-01 6.70446873e-01 -3.30237269e-01 1.23480752e-01
-3.12194616e-01 -5.64262122e-02 1.72695085e-01 1.02076185e+00
7.00227022e-01 3.13263357e-01 -4.93293136e-01 9.78175581e-01
2.09641829e-01 2.01060727e-01 7.62468100e-01 -8.44400048e-01
2.02811614e-01 6.38860106e-01 -3.88625443e-01 -1.10765254e+00
-5.37424684e-01 -6.84207022e-01 -1.07725477e+00 -3.55253816e-02
-1.09077320e-01 -1.13948487e-01 -8.83833766e-01 1.70404732e+00
-8.65787193e-02 -2.75883555e-01 -2.07789853e-01 1.05324328e+00
1.28847194e+00 3.11444253e-01 4.78664577e-01 8.00592303e-02
1.87863362e+00 -1.32139838e+00 -1.01948118e+00 -7.22529590e-01
-8.31251070e-02 -5.75793743e-01 1.17295921e+00 7.92138204e-02
-1.31984937e+00 -9.32722569e-01 -8.02476346e-01 -5.93958855e-01
-6.55446172e-01 -9.31520760e-02 1.07208478e+00 3.19446921e-01
-1.20680499e+00 2.31482804e-01 -4.83422726e-01 -7.30829298e-01
6.68628991e-01 4.61884141e-01 6.64632395e-02 1.06987730e-01
-1.39700305e+00 1.19670951e+00 2.84954369e-01 2.28661790e-01
-7.08610654e-01 -4.82847720e-01 -8.05989742e-01 3.93583536e-01
3.10798913e-01 -9.80427861e-01 1.32718515e+00 -1.17691743e+00
-1.08972025e+00 7.65456259e-01 -3.68570000e-01 -4.22317475e-01
-1.98423147e-01 -4.08785611e-01 -3.49015117e-01 2.86154091e-01
1.07795605e-02 9.84534085e-01 6.52146101e-01 -8.17071199e-01
-7.94140935e-01 -2.39160478e-01 7.08763301e-01 4.14844215e-01
-6.58369541e-01 4.46818292e-01 -6.77553177e-01 -8.92418683e-01
-1.12879835e-01 -5.64853251e-01 -2.59349525e-01 -8.34707841e-02
-6.27870113e-02 -5.89737296e-01 7.49793410e-01 -2.91958362e-01
1.47872484e+00 -2.09952307e+00 2.05895677e-01 -4.40679848e-01
6.26631618e-01 3.93918723e-01 -9.96009409e-02 3.59782815e-01
-3.21775943e-01 2.83498406e-01 -1.27521520e-02 -3.03108960e-01
1.71857551e-02 2.81888336e-01 -3.50574553e-01 2.37742081e-01
3.53727847e-01 1.38386858e+00 -1.03423572e+00 -3.91774774e-01
2.85729080e-01 6.32939994e-01 -5.59686482e-01 2.70632058e-01
3.41670513e-01 2.03066513e-01 -5.08731604e-01 8.31819415e-01
3.25872093e-01 -4.68785346e-01 -3.27385902e-01 -1.94716036e-01
-1.18184417e-01 3.92583668e-01 -5.50645769e-01 1.35832107e+00
-3.33639085e-01 1.06213915e+00 1.16170406e-01 -1.14449000e+00
7.73151278e-01 3.79030198e-01 2.22183555e-01 -6.87489271e-01
4.06074911e-01 -2.41236061e-01 4.74063575e-01 -6.73629105e-01
7.11574614e-01 -7.12039620e-02 4.59934957e-02 5.25697291e-01
3.14110786e-01 4.22581375e-01 1.97078183e-01 1.20611288e-01
7.06679404e-01 -5.49580693e-01 7.17797220e-01 -4.51516300e-01
7.34381616e-01 -4.27406400e-01 3.64492118e-01 9.10106421e-01
-6.98042989e-01 5.57432532e-01 2.86663264e-01 -7.32198417e-01
-6.65412009e-01 -5.62240601e-01 -6.85854407e-04 1.71466434e+00
5.11338972e-02 -1.94502100e-01 -6.87265396e-01 -6.79991961e-01
-2.27990553e-01 3.56896818e-01 -1.19406986e+00 -2.68215626e-01
-3.23794812e-01 -7.22673893e-01 2.14890644e-01 1.09378302e+00
6.59681082e-01 -1.87541234e+00 -8.36725533e-01 -1.17213041e-01
-3.35318089e-01 -7.66578794e-01 -3.60174090e-01 1.68074980e-01
-8.62683773e-01 -1.07431471e+00 -8.85021448e-01 -1.04128253e+00
7.55849600e-01 8.66388440e-01 1.25122619e+00 4.93392169e-01
-1.73474327e-01 6.28487706e-01 -4.02122110e-01 -8.67068708e-01
2.75566071e-01 3.69856089e-01 6.11683764e-02 -8.63974318e-02
8.86416852e-01 -3.46112520e-01 -7.32327461e-01 4.05647643e-02
-7.68048644e-01 -1.99918345e-01 9.28331077e-01 6.48327231e-01
1.92637473e-01 -2.92985827e-01 7.96715200e-01 -4.93512183e-01
9.27901745e-01 -6.77004337e-01 1.09751880e-01 6.13387972e-02
-2.92616993e-01 -2.13100612e-01 3.88302714e-01 -4.82017785e-01
-9.10520077e-01 -4.61604595e-01 -2.10806236e-01 -4.57561523e-01
-2.66083091e-01 6.96139514e-01 -1.09654158e-01 -1.19307734e-01
4.22179908e-01 2.21526042e-01 -9.92898047e-02 -4.72638965e-01
3.10634464e-01 6.41519964e-01 4.09716696e-01 -4.05952722e-01
2.25749448e-01 2.98795134e-01 -4.37401474e-01 -9.59937155e-01
-1.23094392e+00 -4.91188735e-01 -7.62442350e-01 -2.36418769e-01
9.14526999e-01 -4.37601209e-01 -8.67026448e-01 3.99955004e-01
-1.27634013e+00 -3.06996028e-03 -2.87920296e-01 3.97430688e-01
-4.72042352e-01 3.22580487e-01 -7.47014105e-01 -7.54635870e-01
-5.49249828e-01 -1.17053699e+00 6.72295511e-01 6.19602144e-01
-4.33327287e-01 -1.29890180e+00 -2.76058763e-01 2.76785016e-01
7.53177702e-01 -2.90476441e-01 7.29596376e-01 -6.08054399e-01
-4.39349264e-01 -9.00283977e-02 -4.93872017e-01 2.12550282e-01
3.17524850e-01 -1.99673161e-01 -1.34165418e+00 -2.75628597e-01
1.92294717e-01 -3.49127412e-01 1.03323746e+00 7.41617799e-01
1.66792703e+00 -6.47653788e-02 -2.57677436e-01 4.96058375e-01
1.01874113e+00 3.05671096e-01 7.40986764e-01 5.78971982e-01
5.80493629e-01 7.02314436e-01 4.23542172e-01 9.52702239e-02
4.55300897e-01 2.89586067e-01 7.16409445e-01 -3.31167936e-01
-2.52952158e-01 2.01506570e-01 1.04804479e-01 9.32624519e-01
-5.97073615e-01 -3.30322415e-01 -6.91459060e-01 8.03230524e-01
-1.51079261e+00 -1.07955253e+00 6.98006004e-02 1.77336705e+00
4.50795501e-01 8.09880123e-02 1.47910910e-02 2.00370684e-01
1.04037702e+00 2.26015612e-01 -6.67490363e-01 -5.65669596e-01
-1.09817788e-01 5.48280254e-02 8.17373097e-02 4.44996178e-01
-1.21085882e+00 9.85419929e-01 7.85575819e+00 4.17021155e-01
-1.06237185e+00 1.96300056e-02 4.54784811e-01 7.35556260e-02
3.64621803e-02 -4.17924166e-01 -7.77586341e-01 3.77464503e-01
7.08749294e-01 -3.92204791e-01 2.09260300e-01 8.98664534e-01
-1.14220083e-01 1.42151147e-01 -8.97589266e-01 1.03161216e+00
2.92551875e-01 -1.23439407e+00 2.76012033e-01 8.28406140e-02
4.17385608e-01 1.66431293e-01 3.59511733e-01 5.77691853e-01
-2.72232834e-02 -1.00628793e+00 5.46515644e-01 4.35366571e-01
5.57846487e-01 -6.63945973e-01 8.07862639e-01 1.55921564e-01
-1.22669613e+00 -4.22102869e-01 -6.02368236e-01 -6.00989163e-01
1.95447262e-02 3.20195705e-02 -2.05850869e-01 3.72834474e-01
9.63549733e-01 8.70309711e-01 -6.82188511e-01 1.17590642e+00
-4.52734739e-01 3.13223243e-01 1.51850939e-01 -1.67520791e-01
4.60872978e-01 2.31153116e-01 4.02311593e-01 1.25568664e+00
1.09905593e-01 4.14831460e-01 -1.28223911e-01 8.57629955e-01
4.67750579e-02 -1.85535029e-01 -3.94356638e-01 -9.98354629e-02
5.40271759e-01 1.45315325e+00 -7.79917002e-01 -5.26723683e-01
-8.69374514e-01 7.97422171e-01 6.15198195e-01 4.06524658e-01
-9.18110132e-01 -6.23836339e-01 9.10456121e-01 -2.28310168e-01
3.60502928e-01 -9.81714576e-02 -3.63750279e-01 -9.54777181e-01
-3.27622145e-01 -7.24983692e-01 6.23570502e-01 -9.97958720e-01
-1.47835839e+00 9.05533850e-01 -2.08094522e-01 -1.00346661e+00
1.76298901e-01 -8.34224224e-01 -7.92362392e-01 1.03168750e+00
-1.57758296e+00 -9.63602662e-01 -6.34063601e-01 5.84972501e-01
1.04906261e+00 -9.43929181e-02 8.04881394e-01 2.75916606e-01
-7.95568883e-01 5.96906662e-01 -4.71852601e-01 2.36181825e-01
5.90720654e-01 -1.27581179e+00 5.64863443e-01 6.71681166e-01
-4.56084311e-02 1.12055492e+00 4.44883198e-01 -2.92959988e-01
-1.12399471e+00 -8.21390927e-01 1.03020477e+00 -5.30407727e-01
5.58099568e-01 -2.26310641e-02 -1.21649456e+00 9.35909331e-01
8.53267491e-01 -1.72375277e-01 8.11483264e-01 4.47015733e-01
-3.07282154e-02 1.20932452e-01 -8.58348131e-01 6.06039643e-01
9.60869372e-01 -4.06453699e-01 -1.04424977e+00 -4.61357869e-02
6.56056821e-01 -2.71878928e-01 -6.23450339e-01 1.50819123e-01
4.03932512e-01 -9.40865755e-01 9.70082521e-01 -9.04036045e-01
4.10911351e-01 -6.10341318e-02 2.71043666e-02 -1.12747204e+00
-1.32984018e+00 -4.14301634e-01 -4.52923298e-01 1.17636490e+00
-3.91706936e-02 -6.65963054e-01 5.26999593e-01 8.23473036e-01
-5.66255271e-01 -8.63292813e-01 -4.55422491e-01 -4.65161592e-01
1.63887709e-01 -4.87848520e-01 5.60365200e-01 9.56814706e-01
1.71526045e-01 6.60324931e-01 -1.73433885e-01 6.33093938e-02
3.25165749e-01 -6.27265871e-02 3.30424875e-01 -1.29027128e+00
1.52648583e-01 -8.40238571e-01 -4.41453427e-01 -1.33001685e+00
1.02417283e-01 -7.50171781e-01 -3.04560244e-01 -1.77393711e+00
5.86373866e-01 -9.28642228e-02 -8.26744974e-01 6.73835576e-01
-4.92511600e-01 3.89201075e-01 3.93434018e-01 2.49669746e-01
-6.86421096e-01 5.80844283e-01 1.39545357e+00 -1.06662884e-01
6.61641508e-02 -4.41153683e-02 -1.68054831e+00 9.61683750e-01
9.10178483e-01 7.46544898e-02 -4.64536756e-01 -9.31886792e-01
-1.04810263e-03 -6.28710389e-01 1.96384653e-01 -1.01394212e+00
5.11180639e-01 -1.96927190e-01 4.74727362e-01 -7.49164820e-01
2.84801930e-01 -7.61391640e-01 -5.50806105e-01 2.67253101e-01
-3.44239652e-01 3.24480683e-01 4.92685318e-01 3.40874404e-01
-3.94526392e-01 -3.02907765e-01 7.76807606e-01 -3.07845592e-01
-1.25627935e+00 4.99451250e-01 -6.98999763e-01 1.02601625e-01
9.62138414e-01 -3.48036647e-01 -2.86620021e-01 -4.04511482e-01
-8.27347457e-01 2.97336131e-01 -3.97875309e-02 7.73910940e-01
5.69815218e-01 -1.41566801e+00 -3.53444487e-01 2.99271166e-01
-1.15642920e-02 -1.71240885e-02 6.16867840e-01 7.70478249e-01
-7.36434106e-03 9.81056631e-01 -4.87580359e-01 -4.23771322e-01
-1.11358714e+00 1.22033739e+00 2.96287239e-01 1.07011333e-01
-2.99868733e-01 9.88560736e-01 8.94002676e-01 8.07125121e-02
4.03513253e-01 -4.49214429e-01 -8.21054697e-01 1.81450382e-01
1.07028127e+00 3.88234526e-01 1.12178937e-01 -6.37528837e-01
-5.83605051e-01 6.27787054e-01 -2.41535619e-01 4.42190588e-01
1.10499918e+00 -4.88413692e-01 -1.64985329e-01 2.89890945e-01
8.19409430e-01 -4.02626485e-01 -1.18699741e+00 -2.66907930e-01
-2.68454820e-01 -2.08215222e-01 1.87438294e-01 -7.64962196e-01
-1.39403951e+00 1.30368578e+00 2.24205106e-01 5.53780556e-01
1.46009970e+00 3.22947353e-01 4.04161364e-01 1.95180058e-01
3.55534777e-02 -9.39148188e-01 2.56787688e-01 9.94871736e-01
1.28850567e+00 -1.17831552e+00 2.83890287e-03 -1.68478712e-01
-7.58871436e-01 1.06790173e+00 1.03234327e+00 -2.46602386e-01
6.80846214e-01 -6.58823848e-02 1.96501054e-02 -3.89415562e-01
-8.04942727e-01 -5.67051113e-01 5.56380093e-01 8.15652370e-01
7.77495265e-01 -3.01941633e-01 -1.25744894e-01 8.65817368e-01
-1.48481056e-01 -4.61258516e-02 2.23435208e-01 7.83375263e-01
-7.26255536e-01 -7.46716499e-01 -4.59957004e-01 3.86014313e-01
-6.71999753e-01 -2.66532958e-01 -3.58826846e-01 8.23725224e-01
2.19564781e-01 8.65232587e-01 3.90773475e-01 -2.15559974e-01
2.70417124e-01 -1.73083916e-01 3.42989594e-01 -6.72105730e-01
-5.17378569e-01 -1.35882378e-01 -5.11734784e-01 -2.21158653e-01
-6.44378066e-01 -4.47006464e-01 -8.31856549e-01 -4.71159846e-01
-2.38424838e-01 6.65890276e-02 1.23244144e-01 9.42243159e-01
6.48671210e-01 9.94733393e-01 1.50790185e-01 -1.13924742e+00
-9.69753712e-02 -1.30141962e+00 -3.63135368e-01 1.28380999e-01
4.00168896e-01 -1.06917143e+00 -2.92426318e-01 -3.00173890e-02] | [9.900928497314453, 2.0220136642456055] |
5b3392de-f5f8-468a-bf8a-ae11d478435f | word-embeddings-via-causal-inference-gender | 2112.05194 | null | https://arxiv.org/abs/2112.05194v1 | https://arxiv.org/pdf/2112.05194v1.pdf | Word Embeddings via Causal Inference: Gender Bias Reducing and Semantic Information Preserving | With widening deployments of natural language processing (NLP) in daily life, inherited social biases from NLP models have become more severe and problematic. Previous studies have shown that word embeddings trained on human-generated corpora have strong gender biases that can produce discriminative results in downstream tasks. Previous debiasing methods focus mainly on modeling bias and only implicitly consider semantic information while completely overlooking the complex underlying causal structure among bias and semantic components. To address these issues, we propose a novel methodology that leverages a causal inference framework to effectively remove gender bias. The proposed method allows us to construct and analyze the complex causal mechanisms facilitating gender information flow while retaining oracle semantic information within word embeddings. Our comprehensive experiments show that the proposed method achieves state-of-the-art results in gender-debiasing tasks. In addition, our methods yield better performance in word similarity evaluation and various extrinsic downstream NLP tasks. | ['Bei Jiang', 'Yanchun Bao', 'Hongsheng Dai', 'Linglong Kong', 'Meichen Liu', 'Shenggang Hu', 'Wenxing Guo', 'Jinhan Xie', 'Dengdeng Yu', 'Lei Ding'] | 2021-12-09 | null | null | null | null | ['word-similarity'] | ['natural-language-processing'] | [-5.06195314e-02 6.30111098e-02 -6.82730854e-01 -6.72578812e-01
-3.84999931e-01 -4.53166425e-01 9.13774312e-01 5.50858200e-01
-6.75659776e-01 5.26475608e-01 8.29611957e-01 -3.15120310e-01
-1.09964319e-01 -8.78637731e-01 -2.85971135e-01 -5.16584039e-01
4.72020060e-01 4.44674462e-01 -1.08196594e-01 -3.86460185e-01
8.41410041e-01 1.26486361e-01 -1.45162296e+00 -2.57518947e-01
1.16561449e+00 5.81670105e-01 -1.93047315e-01 2.94584811e-01
-2.46934459e-01 3.60990614e-01 -5.85955083e-01 -9.62802827e-01
-1.87317967e-01 -5.92702366e-02 -6.77645147e-01 -8.06189656e-01
6.27271116e-01 -2.06323907e-01 -3.81658435e-01 1.28888273e+00
8.92425776e-01 1.49748042e-01 9.60611284e-01 -1.25602293e+00
-1.21781969e+00 7.35611141e-01 -7.25909412e-01 4.22609240e-01
1.50983691e-01 -9.26784649e-02 1.44613802e+00 -1.07365966e+00
3.53362530e-01 1.85350263e+00 6.33261561e-01 8.42101634e-01
-1.36821818e+00 -1.15101910e+00 2.00480044e-01 3.29588503e-01
-1.08336794e+00 -3.47922206e-01 7.45933354e-01 -2.34950110e-01
9.03705955e-01 6.00587912e-02 2.51359433e-01 1.66918314e+00
2.52882391e-01 7.66325653e-01 9.85847890e-01 -4.06386197e-01
-4.22892980e-02 1.16443157e-01 6.71333134e-01 4.92624491e-01
7.88724720e-01 8.43599811e-02 -1.00926745e+00 -2.83304304e-01
2.70303696e-01 -1.55615717e-01 1.68599829e-01 -2.30835788e-02
-1.07648885e+00 1.21241856e+00 2.04452612e-02 2.09643736e-01
-7.92983398e-02 3.91294926e-01 5.90210438e-01 2.22962573e-02
8.24016929e-01 7.19002366e-01 -3.34522009e-01 -3.69985402e-01
-9.90459204e-01 6.80268884e-01 5.97245395e-01 6.10564768e-01
5.00592232e-01 -3.23658556e-01 -6.29816294e-01 1.17218482e+00
3.15527081e-01 7.48919606e-01 7.46326447e-01 -8.46049666e-01
4.93358791e-01 4.79387403e-01 -1.05024204e-01 -1.57302690e+00
-1.92465499e-01 -2.17556164e-01 -5.55696428e-01 -4.99754936e-01
5.26891172e-01 1.26753822e-01 -7.48505652e-01 2.19151831e+00
3.20836008e-01 -1.44798711e-01 -2.20231041e-01 8.10227692e-01
6.50909841e-01 2.35442862e-01 7.32131064e-01 2.43173510e-01
1.67531812e+00 -7.16394901e-01 -8.68023098e-01 -5.28620541e-01
5.78482151e-01 -8.12483907e-01 1.53802776e+00 1.89944759e-01
-8.15778792e-01 -3.50598514e-01 -1.06809247e+00 -5.31671524e-01
-4.47302520e-01 9.70755070e-02 9.94763434e-01 1.00170207e+00
-5.24098039e-01 6.81494713e-01 -5.26686728e-01 -6.07601881e-01
6.39169812e-01 1.35247275e-01 -2.08622161e-02 -2.68776804e-01
-1.53436148e+00 1.08639824e+00 -2.35564969e-02 -3.31072181e-01
-7.94686854e-01 -1.22213650e+00 -8.33976924e-01 1.36558309e-01
1.72631070e-01 -8.68961155e-01 1.03118122e+00 -5.93538046e-01
-9.22107220e-01 1.06407750e+00 -5.16193867e-01 -1.92856088e-01
2.52946645e-01 -6.73478663e-01 -3.04994375e-01 -5.14115654e-02
4.36408699e-01 7.36479700e-01 1.05408287e+00 -8.92925203e-01
-4.30942029e-01 -7.03749597e-01 -2.08601460e-01 1.31719127e-01
-1.15812898e+00 2.53064871e-01 -1.72911230e-02 -1.07570934e+00
-2.03966931e-01 -5.82103074e-01 6.01414293e-02 -1.09597065e-01
-2.95834750e-01 -7.56405473e-01 6.08694255e-01 -5.78302264e-01
1.60317910e+00 -2.02891278e+00 2.51585633e-01 -6.26185983e-02
2.82317996e-01 2.17517495e-01 -1.26034141e-01 3.20162416e-01
-3.15224491e-02 3.25771153e-01 6.43464923e-03 -5.45470715e-01
4.91429955e-01 2.02133566e-01 -7.91611791e-01 3.52993846e-01
1.08300067e-01 8.89501333e-01 -1.07325149e+00 -5.93226910e-01
-9.73812789e-02 3.46087873e-01 -9.22299206e-01 2.21643284e-01
1.16483934e-01 -1.25335399e-02 -3.43785018e-01 6.62344337e-01
8.37562978e-01 1.56999156e-01 1.56875283e-01 -1.56746715e-01
6.43557385e-02 5.60726404e-01 -6.63415492e-01 1.65147829e+00
-5.07221043e-01 4.52074826e-01 -1.16071284e-01 -1.00835454e+00
9.02839065e-01 -2.85354197e-01 -5.07241338e-02 -7.14426696e-01
3.34915817e-01 1.65504918e-01 6.17188178e-02 -6.06103957e-01
9.81353760e-01 -5.40401816e-01 -4.69141781e-01 6.59852207e-01
2.98634440e-01 -1.04821675e-01 6.94183782e-02 5.04901052e-01
7.59196222e-01 -2.70687103e-01 9.63873565e-02 -7.52232075e-01
4.34597850e-01 -1.98060736e-01 7.03486562e-01 6.86909914e-01
-6.04098856e-01 4.59860831e-01 8.75760138e-01 -6.02749642e-03
-9.44892526e-01 -1.19611323e+00 -1.89981788e-01 1.67345083e+00
2.98431188e-01 -5.18050551e-01 -6.27093852e-01 -8.56203735e-01
5.25710404e-01 1.25807297e+00 -9.56560194e-01 -7.56517351e-01
-4.26225334e-01 -1.09134877e+00 1.08556485e+00 6.92024708e-01
2.04144374e-01 -7.48078942e-01 -7.11146891e-02 -1.83365703e-01
-2.08547711e-01 -8.42896163e-01 -5.93115389e-01 -2.37053409e-01
-7.05639362e-01 -1.02988291e+00 -4.58889037e-01 -5.27304888e-01
5.19066095e-01 1.66629478e-01 1.15357566e+00 -7.91480318e-02
-4.39685076e-01 2.89779425e-01 -9.69506651e-02 -7.25915432e-01
-6.99300393e-02 2.58998841e-01 3.54034185e-01 -2.20460221e-01
1.11499083e+00 -5.47320426e-01 -8.04647207e-01 -2.39725113e-02
-7.77975857e-01 -5.35513163e-01 3.87509435e-01 9.25014973e-01
-1.35900661e-01 -3.37003857e-01 9.85024154e-01 -1.01937163e+00
1.17266548e+00 -7.92019248e-01 -4.59559262e-02 5.12898788e-02
-1.17640996e+00 1.66251078e-01 2.91437477e-01 -4.53154176e-01
-1.47153246e+00 -8.85736108e-01 -7.55706429e-02 -1.09177239e-01
-2.82349940e-02 6.85319155e-02 -2.31016830e-01 4.49218661e-01
5.95232844e-01 -1.16299920e-01 -1.20431088e-01 -4.59100306e-01
6.67995870e-01 6.82113707e-01 3.62276912e-01 -1.21124756e+00
6.79164171e-01 5.89383125e-01 -2.24910453e-01 -7.02667236e-01
-1.15029132e+00 -4.39453959e-01 -2.81890631e-01 1.24046572e-01
9.23210561e-01 -7.58960247e-01 -6.50330544e-01 3.65483373e-01
-1.22701347e+00 1.79262966e-01 8.20287913e-02 3.71437699e-01
3.83932493e-03 4.98167127e-01 -5.67881346e-01 -6.86395168e-01
-4.93284672e-01 -7.12236345e-01 1.15252995e+00 1.16666816e-01
-8.81033361e-01 -1.12797368e+00 2.83737242e-01 5.73974967e-01
5.09530425e-01 -4.53882635e-01 1.43097341e+00 -8.57454896e-01
1.70778424e-01 3.12673897e-02 -6.95627034e-01 3.55180323e-01
2.34825350e-02 -7.09333420e-02 -1.05540562e+00 4.85121226e-03
-1.88604623e-01 -4.38164592e-01 1.20127773e+00 1.77302644e-01
1.33898211e+00 -1.23408273e-01 -4.15241241e-01 3.03945243e-01
1.12423229e+00 -3.92463923e-01 4.33155894e-01 1.42095521e-01
8.01947951e-01 1.14916611e+00 7.92363346e-01 4.57107246e-01
6.58981919e-01 2.21516892e-01 2.05519632e-01 2.34386638e-01
-8.81579369e-02 -8.26886952e-01 1.81533590e-01 5.40571630e-01
1.92969903e-01 -1.42600223e-01 -1.07624400e+00 1.02404714e+00
-1.66341233e+00 -8.22495341e-01 -1.61724463e-01 1.90749133e+00
1.02471554e+00 -3.07767857e-02 -2.40627751e-01 5.39508685e-02
8.82460535e-01 5.61127722e-01 -4.26035464e-01 -8.17170322e-01
8.51874892e-03 5.46025395e-01 4.84683067e-01 3.56460363e-01
-9.28671658e-01 1.30374920e+00 6.46151733e+00 1.02868533e+00
-8.12888920e-01 2.37460598e-01 5.42091131e-01 -2.39094898e-01
-8.11497331e-01 -2.01340348e-01 -1.01252997e+00 4.82712358e-01
6.57318532e-01 -4.78650481e-01 1.06990755e-01 7.51678407e-01
2.39449129e-01 -5.31993294e-03 -1.31269825e+00 9.51742589e-01
2.99315453e-01 -8.52074921e-01 4.00073439e-01 4.62843440e-02
6.33430004e-01 -3.74544561e-01 5.16774237e-01 5.00220895e-01
4.75631535e-01 -1.31353033e+00 6.38656855e-01 1.63781956e-01
6.41616464e-01 -9.59032655e-01 5.48132420e-01 9.54908803e-02
-3.21338087e-01 -3.33938837e-01 -5.58475018e-01 -3.94057721e-01
-6.80431277e-02 1.06143451e+00 -6.06681645e-01 1.41865775e-01
6.45668209e-01 7.14889169e-01 -6.28711760e-01 2.66626656e-01
-6.78454757e-01 8.05870056e-01 2.36138299e-01 -3.79987836e-01
-2.38192212e-02 -2.66365265e-03 5.12093663e-01 1.40395105e+00
2.49678701e-01 -1.59915537e-01 -5.79953313e-01 9.73840296e-01
-3.14926893e-01 3.70738238e-01 -6.78558648e-01 -4.24851716e-01
7.52537370e-01 1.26969266e+00 -4.48444128e-01 -2.91250259e-01
-3.42085630e-01 7.08915055e-01 6.19473100e-01 2.08638310e-01
-9.29816723e-01 -4.00132805e-01 1.37939417e+00 6.46786345e-03
1.17502525e-01 -2.69459635e-01 -8.14888179e-01 -1.33449781e+00
-2.33524203e-01 -7.28628993e-01 5.27487457e-01 -2.09357470e-01
-1.87115347e+00 -3.87260944e-01 4.78106365e-02 -3.87203127e-01
1.91303372e-01 -6.65110052e-01 -7.12592483e-01 7.27176189e-01
-1.69211483e+00 -1.06485283e+00 3.05631142e-02 2.41111875e-01
5.40774405e-01 -9.28415731e-02 6.66027427e-01 4.89813298e-01
-7.19354570e-01 9.40157712e-01 -2.53533691e-01 -1.18832495e-02
1.47691047e+00 -1.23223042e+00 1.13026597e-01 7.97358811e-01
-2.36725643e-01 1.27355742e+00 8.73636603e-01 -6.92515314e-01
-1.20560098e+00 -9.80222046e-01 1.67358470e+00 -7.39160597e-01
9.32503760e-01 -6.59026980e-01 -6.37487411e-01 3.89472723e-01
3.58602345e-01 -3.98214430e-01 9.67721522e-01 8.23625505e-01
-1.03610492e+00 -7.69963264e-02 -1.19266629e+00 6.95717275e-01
1.37846160e+00 -5.92431188e-01 -1.27687323e+00 -1.68805048e-02
7.92805851e-01 1.64504349e-01 -4.05269295e-01 2.54216105e-01
6.62116408e-01 -6.83535516e-01 1.08784330e+00 -9.62732375e-01
1.18678892e+00 2.30999947e-01 -1.88452199e-01 -1.41384542e+00
-3.27251881e-01 -2.59821713e-01 9.32477564e-02 1.69564784e+00
2.68614799e-01 -7.03023493e-01 7.15268850e-01 8.48991573e-01
1.65864944e-01 -4.59282160e-01 -8.12228143e-01 -5.64116538e-01
6.93238676e-01 -4.92456675e-01 6.13336802e-01 1.37939119e+00
1.53800264e-01 5.99739969e-01 -2.25885972e-01 -2.02686172e-02
8.18207145e-01 2.42841113e-02 5.30535042e-01 -1.29735112e+00
-1.36305597e-02 -5.61251342e-01 -7.16626570e-02 -8.16083729e-01
8.68972361e-01 -1.02149582e+00 -1.27561688e-01 -1.06035757e+00
5.81708014e-01 -3.34084868e-01 -3.40859830e-01 2.35421807e-01
-6.61583424e-01 2.14895472e-01 -1.66474089e-01 -1.61971956e-01
-3.64514738e-01 8.38721156e-01 9.65118766e-01 -2.21737921e-01
3.74818981e-01 -5.20224750e-01 -1.33650839e+00 8.51108909e-01
7.82700062e-01 -7.05496311e-01 -5.07961631e-01 -8.14471722e-01
6.00681484e-01 -7.56178975e-01 3.89765024e-01 -1.90807387e-01
8.41097757e-02 -2.18957648e-01 1.37284145e-01 -1.39863819e-01
1.46014303e-01 -3.10211003e-01 -8.77274692e-01 2.86003411e-01
-5.35417020e-01 5.48089519e-02 -3.71075654e-03 6.37636006e-01
-1.99047580e-01 -2.67048270e-01 5.97676277e-01 1.01237670e-01
-3.61527473e-01 7.68141598e-02 -2.24550232e-01 5.12287319e-01
6.67901814e-01 2.66386628e-01 -5.10969520e-01 -1.22496739e-01
-3.15643102e-01 2.33201563e-01 2.58157581e-01 9.03593123e-01
4.31121498e-01 -1.48331916e+00 -6.20140016e-01 1.62274837e-01
2.22512990e-01 -5.70966899e-01 1.56839043e-01 7.31517136e-01
-2.24693026e-02 5.10548115e-01 -4.65713814e-03 -1.10061668e-01
-1.29516804e+00 4.53445256e-01 -2.11897179e-01 -1.47919014e-01
5.70982993e-02 1.23594666e+00 4.38475817e-01 -6.14083886e-01
6.72145858e-02 -2.79735327e-01 -2.73821205e-01 7.75618255e-01
4.41364467e-01 7.83159018e-01 -3.31679620e-02 -4.85322446e-01
-4.23555136e-01 3.64947528e-01 -6.94204494e-02 -1.37723431e-01
1.31235898e+00 -2.21013278e-01 -5.09015799e-01 1.56199202e-01
1.26214480e+00 3.75370532e-01 -4.07013446e-01 -4.00469862e-02
2.69136101e-01 -9.42882597e-01 1.10583209e-01 -5.33069015e-01
-7.19928980e-01 1.28089297e+00 1.62448838e-01 -1.85108498e-01
6.23395145e-01 -9.88778844e-02 1.01607156e+00 8.39137379e-03
8.29954743e-02 -1.52750862e+00 3.32094640e-01 5.52043915e-01
5.36350071e-01 -1.18590903e+00 4.15468328e-02 -5.10752499e-01
-4.97394800e-01 8.82200420e-01 7.92793274e-01 -1.41279131e-01
5.51740587e-01 -3.04196216e-02 -1.26155406e-01 -2.40422130e-01
-7.87164032e-01 8.57715532e-02 1.29574090e-01 5.79404294e-01
6.65168524e-01 3.25195268e-02 -1.13545048e+00 1.28497541e+00
-3.72460902e-01 -3.97934735e-01 2.29936972e-01 4.05191004e-01
-3.08042437e-01 -1.38302124e+00 -2.71663129e-01 3.33060443e-01
-8.43220532e-01 -5.07568121e-01 -5.43325007e-01 5.60867071e-01
6.87310845e-02 8.82708490e-01 1.98077887e-01 -1.30087420e-01
2.98407078e-01 3.67616534e-01 6.84221923e-01 -6.36700451e-01
-4.35721904e-01 -5.20777524e-01 3.72950524e-01 -6.94491804e-01
-1.31977275e-01 -7.95480788e-01 -1.03102362e+00 -6.40392363e-01
-3.73263732e-02 1.21796511e-01 5.79568446e-01 7.94216335e-01
5.21351635e-01 2.30222702e-01 3.06261241e-01 -4.53735650e-01
-9.00483727e-01 -1.04079890e+00 -5.17263174e-01 6.74812138e-01
8.08918197e-03 -9.22278345e-01 -5.70328295e-01 -3.20974171e-01] | [9.364980697631836, 10.219145774841309] |
55e1333d-6fc8-4ae0-9086-90318bb0a4a9 | ditto-a-feature-representation-imitation | 2303.02357 | null | https://arxiv.org/abs/2303.02357v1 | https://arxiv.org/pdf/2303.02357v1.pdf | DiTTO: A Feature Representation Imitation Approach for Improving Cross-Lingual Transfer | Zero-shot cross-lingual transfer is promising, however has been shown to be sub-optimal, with inferior transfer performance across low-resource languages. In this work, we envision languages as domains for improving zero-shot transfer by jointly reducing the feature incongruity between the source and the target language and increasing the generalization capabilities of pre-trained multilingual transformers. We show that our approach, DiTTO, significantly outperforms the standard zero-shot fine-tuning method on multiple datasets across all languages using solely unlabeled instances in the target language. Empirical results show that jointly reducing feature incongruity for multiple target languages is vital for successful cross-lingual transfer. Moreover, our model enables better cross-lingual transfer than standard fine-tuning methods, even in the few-shot setting. | ['Monojit Choudhury', 'Sunayana Sitaram', 'Sandipan Dandapat', 'Abbaraju Soujanya', 'Shanu Kumar'] | 2023-03-04 | null | null | null | null | ['zero-shot-cross-lingual-transfer', 'cross-lingual-transfer'] | ['natural-language-processing', 'natural-language-processing'] | [-2.26218000e-01 -1.58305079e-01 -5.12668848e-01 -2.23761797e-01
-1.68145978e+00 -7.17690885e-01 8.48361790e-01 -2.80042410e-01
-6.81724727e-01 9.35303390e-01 3.97790521e-01 -1.59082964e-01
2.80734986e-01 -5.88088214e-01 -8.58091533e-01 -3.01833391e-01
1.91393733e-01 7.04616606e-01 1.90105036e-01 -6.08796418e-01
-2.05163971e-01 -1.32923722e-01 -1.33353889e+00 3.53207231e-01
1.15384495e+00 3.42349023e-01 1.89304784e-01 1.32892579e-01
-4.01739120e-01 2.54713774e-01 -3.76009375e-01 -6.61775529e-01
1.95207015e-01 -5.07321775e-01 -7.40675032e-01 -2.67888188e-01
8.57982635e-01 -1.43535152e-01 -1.60788491e-01 9.87637579e-01
7.45515049e-01 1.96725145e-01 7.79291570e-01 -9.41307127e-01
-1.13462031e+00 8.59861732e-01 -5.58467746e-01 1.75959840e-01
1.47782359e-02 2.63469875e-01 1.11021316e+00 -1.16219521e+00
7.52605677e-01 1.49201512e+00 8.44687700e-01 8.36030185e-01
-1.57469106e+00 -9.95315254e-01 4.55645025e-02 1.52339533e-01
-1.46282518e+00 -8.38766217e-01 3.28904599e-01 -5.83173692e-01
1.30773759e+00 -4.79989886e-01 2.78489560e-01 1.25772333e+00
-1.56794377e-02 7.46792376e-01 1.14068329e+00 -7.21102178e-01
-1.05979055e-01 4.95073229e-01 1.95406154e-02 5.06180584e-01
3.29097398e-02 7.04202950e-02 -8.25583816e-01 5.02348319e-02
4.52001125e-01 -4.31092173e-01 -1.45630270e-01 -5.19479454e-01
-1.21019149e+00 9.81111526e-01 2.97924250e-01 6.90581381e-01
-3.10877021e-02 1.55326694e-01 7.05561697e-01 7.79108405e-01
9.25713956e-01 4.56265152e-01 -6.64134502e-01 -1.03319488e-01
-8.11848938e-01 -2.22987533e-01 5.05980492e-01 1.16525900e+00
1.16729259e+00 1.67985708e-01 -4.76280302e-01 1.19631493e+00
-1.68998882e-01 4.79773462e-01 5.45051932e-01 -4.42597836e-01
7.68615007e-01 2.47700050e-01 -3.89629189e-04 -1.07415216e-02
8.94853398e-02 -2.49248296e-01 -3.17174107e-01 3.62940356e-02
4.03007507e-01 -5.04451871e-01 -8.54877770e-01 2.28483915e+00
1.28602147e-01 2.81269789e-01 3.60572070e-01 5.67525208e-01
5.13912916e-01 6.17823720e-01 5.57588279e-01 -8.51886198e-02
1.30510175e+00 -1.01071477e+00 -6.85746133e-01 -4.81826693e-01
1.18224847e+00 -8.54722798e-01 1.78494167e+00 -1.32268012e-01
-7.99736261e-01 -5.32249391e-01 -9.18213665e-01 -3.05364162e-01
-4.56543058e-01 -2.32327774e-01 5.83484054e-01 6.28664196e-01
-1.00652516e+00 4.09121156e-01 -4.40231860e-01 -6.63977146e-01
4.95279372e-01 1.60670638e-01 -5.49705148e-01 -5.71075201e-01
-1.72468519e+00 1.19629788e+00 4.00974959e-01 -8.56516540e-01
-9.89613175e-01 -1.37690175e+00 -1.06617320e+00 1.61147192e-01
1.70013338e-01 -5.16761839e-01 1.13322425e+00 -9.60887969e-01
-1.38321137e+00 1.11062813e+00 1.46466587e-02 -3.36695522e-01
5.00291705e-01 -3.37308019e-01 -2.69426048e-01 -3.44421208e-01
3.90778720e-01 9.65773284e-01 5.12813628e-01 -9.97723877e-01
-4.84072238e-01 -1.21862352e-01 -6.94021359e-02 4.79745746e-01
-8.46447825e-01 1.31678715e-01 -6.01805508e-01 -5.51525772e-01
-7.79895484e-01 -7.66137540e-01 4.74089123e-02 -2.14747131e-01
9.05131549e-02 -5.51607549e-01 7.11769700e-01 -4.62725282e-01
8.44478488e-01 -2.41014504e+00 1.87843398e-03 -4.24768120e-01
-4.01384562e-01 4.07653332e-01 -6.84699953e-01 5.12367785e-01
9.03549641e-02 7.31555074e-02 -1.20021783e-01 -4.11446810e-01
4.08958308e-02 6.74318150e-02 -3.05316120e-01 4.23036337e-01
3.55391741e-01 1.07635891e+00 -1.03892732e+00 -5.40206075e-01
1.54230416e-01 7.32340753e-01 -7.10215867e-01 8.77273008e-02
-1.49412036e-01 4.59501624e-01 -4.80163172e-02 2.45125532e-01
5.51216543e-01 -6.02381118e-02 1.98823288e-01 -5.75763471e-02
-4.04315144e-02 2.72149026e-01 -5.97784162e-01 2.20072198e+00
-9.62090254e-01 5.54036558e-01 -2.96437293e-01 -6.13054872e-01
8.53469551e-01 4.87934619e-01 3.40424001e-01 -1.07172406e+00
-1.10455170e-01 2.97204316e-01 6.78318888e-02 -1.92949131e-01
2.72761375e-01 -7.71572649e-01 -5.09279847e-01 6.12820327e-01
7.89215207e-01 -8.96025971e-02 2.23442510e-01 3.72114092e-01
6.18231595e-01 2.45823368e-01 3.61025006e-01 -7.03363597e-01
2.00781092e-01 5.03204651e-02 4.72708553e-01 6.08436942e-01
-3.70769590e-01 2.39560321e-01 2.05934420e-01 4.49919961e-02
-1.16617775e+00 -1.25462294e+00 -2.92655259e-01 1.86445248e+00
4.38143983e-02 -2.59561121e-01 -8.06903780e-01 -7.67547727e-01
1.77619457e-01 1.05351329e+00 -6.66823208e-01 -4.03846115e-01
-3.54279429e-01 -5.55466294e-01 7.47190356e-01 4.12713587e-01
9.69131812e-02 -7.14732885e-01 -5.18258382e-03 2.89564371e-01
-2.78131157e-01 -1.13493228e+00 -9.32728291e-01 3.17980319e-01
-4.93964046e-01 -5.35473585e-01 -1.01099420e+00 -9.54029322e-01
2.54332304e-01 4.28378016e-01 1.41251218e+00 -2.90791184e-01
-1.44346610e-01 2.40793839e-01 -1.83470652e-01 -2.29766175e-01
-3.11277092e-01 4.16007370e-01 3.44654441e-01 -2.55474269e-01
7.15091169e-01 -3.66799325e-01 -1.97848439e-01 2.94625044e-01
-4.88314122e-01 -1.71506628e-01 3.49130094e-01 9.10903156e-01
2.98756540e-01 -5.00278234e-01 9.84291315e-01 -1.07798302e+00
7.24749088e-01 -6.11426413e-01 -2.92502433e-01 6.68878973e-01
-4.03103322e-01 1.84198767e-01 5.65857649e-01 -6.32271707e-01
-1.32716835e+00 -3.47727656e-01 2.13007733e-01 -5.71185589e-01
6.09910861e-03 1.91061780e-01 -9.66970921e-02 -1.38192683e-01
1.05742991e+00 -2.36218832e-02 -2.83401817e-01 -5.74458897e-01
8.01339507e-01 5.78790605e-01 1.14503428e-01 -8.36251616e-01
5.25700927e-01 1.78686142e-01 -6.94308221e-01 -6.60835624e-01
-1.06788683e+00 -5.23561239e-01 -9.55254912e-01 -3.66766043e-02
7.88857579e-01 -1.50994468e+00 4.75377254e-02 1.92550376e-01
-1.07660651e+00 -6.44088268e-01 -2.90135652e-01 6.49904191e-01
-6.03011072e-01 -1.52420342e-01 -7.25096405e-01 -3.14027905e-01
-3.63766551e-01 -1.08275688e+00 1.14051056e+00 -1.13707222e-01
-1.27962276e-01 -1.33955359e+00 5.55759430e-01 1.17509760e-01
5.00215113e-01 -3.70219678e-01 1.11199152e+00 -7.29202569e-01
-3.04883510e-01 2.05894426e-01 -3.29291999e-01 1.30779773e-01
2.87692606e-01 -3.18455666e-01 -1.06463611e+00 -6.55197263e-01
-5.70760250e-01 -9.99345303e-01 9.37196195e-01 1.04123041e-01
3.23125869e-01 3.23282108e-02 -2.87725210e-01 6.27508342e-01
1.69618773e+00 -2.95395941e-01 3.61334592e-01 1.48128584e-01
7.64601588e-01 7.43195832e-01 7.18059242e-01 7.89393112e-02
4.53822523e-01 7.57104278e-01 -3.47175926e-01 -8.80728438e-02
-6.15619779e-01 -3.26488048e-01 7.89551914e-01 1.04921615e+00
2.03705221e-01 -5.05553074e-02 -1.10503745e+00 9.97085452e-01
-1.48685908e+00 -8.06864917e-01 2.83331394e-01 2.24859357e+00
1.20651841e+00 -1.52618572e-01 1.08959556e-01 -7.17600763e-01
8.75324547e-01 1.91445909e-02 -4.75557387e-01 -3.97365868e-01
-2.21991047e-01 3.54194343e-01 4.79551315e-01 7.61176884e-01
-7.86377907e-01 1.91349864e+00 6.84327269e+00 1.08842242e+00
-1.17742753e+00 8.28052938e-01 2.76271701e-01 -3.15311849e-01
-5.16525269e-01 -1.76443517e-01 -1.14710486e+00 2.23205134e-01
1.05358839e+00 -7.07182705e-01 4.51216012e-01 7.03248978e-01
-3.25883538e-01 3.54719579e-01 -1.27013135e+00 6.63079917e-01
3.65343094e-01 -1.01953685e+00 2.15970576e-01 -4.65428382e-02
1.14803219e+00 6.31394386e-01 7.56376237e-02 9.66879606e-01
8.51008356e-01 -7.99439073e-01 5.46721935e-01 2.08701752e-02
1.38752186e+00 -9.60855901e-01 3.14715892e-01 2.52280146e-01
-1.11464870e+00 3.36142540e-01 -5.09085298e-01 2.14820072e-01
2.16825619e-01 1.33857414e-01 -8.20428252e-01 2.38448232e-01
5.89695513e-01 6.91295028e-01 -5.07565498e-01 5.78653514e-01
-1.70114845e-01 4.35772002e-01 3.10834255e-02 2.13895500e-01
4.99700427e-01 8.07171613e-02 2.76337773e-01 1.56847513e+00
4.36631352e-01 -3.52755934e-01 2.41613373e-01 5.99616826e-01
-3.57218176e-01 5.34040332e-01 -1.05460262e+00 -2.73544025e-02
4.84023273e-01 9.44538832e-01 -3.42836082e-02 -5.89751184e-01
-7.66497314e-01 9.97984767e-01 1.06141865e+00 4.54634041e-01
-7.17435181e-01 -2.80732155e-01 8.87160003e-01 -8.46655890e-02
2.64570326e-01 -1.53088480e-01 -1.26744658e-01 -1.56267643e+00
-4.24939036e-01 -7.73522615e-01 5.73588848e-01 -3.26489747e-01
-1.74331570e+00 5.41155577e-01 -1.58042133e-01 -1.29071462e+00
-3.12296748e-01 -4.99008268e-01 -4.70952451e-01 1.17226803e+00
-1.66269934e+00 -1.58450651e+00 2.30571911e-01 1.06264067e+00
7.81684935e-01 -3.82427096e-01 1.08446097e+00 5.96066356e-01
-3.54630679e-01 1.14945567e+00 1.50068209e-01 4.75343727e-02
1.44463754e+00 -1.01766217e+00 4.39959496e-01 7.38023818e-01
1.66644022e-01 6.08152270e-01 5.72061241e-01 -6.69925392e-01
-1.31620026e+00 -1.20963180e+00 9.09015894e-01 -4.03632790e-01
1.12577677e+00 -6.59192860e-01 -1.19774354e+00 9.71930444e-01
7.31010914e-01 2.09905598e-02 1.07604134e+00 7.21853077e-01
-9.42741156e-01 -9.55526903e-03 -9.21766579e-01 6.34499371e-01
9.86771286e-01 -1.00015128e+00 -6.81291699e-01 4.11450565e-01
1.05677867e+00 2.60832757e-01 -1.07391429e+00 1.30730197e-01
3.38952839e-01 -4.09554780e-01 8.87991667e-01 -9.56781566e-01
3.33850682e-01 2.23775893e-01 -3.50135326e-01 -1.90304315e+00
-4.90188211e-01 -4.39880937e-01 3.19930077e-01 1.46535540e+00
6.44380033e-01 -4.90284652e-01 3.65377635e-01 2.14813113e-01
-1.88177869e-01 -1.55072704e-01 -1.05804360e+00 -1.25647008e+00
8.96525443e-01 -2.99106330e-01 2.14626387e-01 1.54568458e+00
1.45515531e-01 1.02461731e+00 -7.65954673e-01 -2.41297022e-01
8.92736077e-01 -1.67007059e-01 7.59820998e-01 -9.83413875e-01
-2.87179470e-01 -4.07706589e-01 -5.45102023e-02 -5.62972188e-01
6.72132373e-01 -1.34264839e+00 1.79400250e-01 -1.26202047e+00
4.21323568e-01 -5.93959093e-01 -6.28166080e-01 6.58345878e-01
-4.51279551e-01 4.90849614e-01 2.56322175e-01 1.73507869e-01
-5.97863078e-01 7.13145316e-01 1.25612676e+00 -1.83971673e-01
-1.69528693e-01 -6.08762562e-01 -7.79791832e-01 3.54796112e-01
6.06468379e-01 -5.34132063e-01 -3.82050127e-01 -8.44798923e-01
-7.72664621e-02 -2.35244036e-01 -3.02835256e-01 -7.52244413e-01
3.85430232e-02 -9.42551941e-02 -1.29918065e-02 -1.31671816e-01
3.79153460e-01 -5.00281692e-01 -3.17749828e-01 4.08992112e-01
-4.02393341e-01 -3.14452261e-01 4.73346531e-01 4.18372184e-01
-2.22304717e-01 9.86164287e-02 1.17588580e+00 -1.06517218e-01
-9.87791240e-01 4.40804720e-01 5.28718671e-03 6.80705309e-01
1.09619892e+00 2.98208654e-01 -4.59117144e-01 -3.98470126e-02
-5.60983241e-01 2.56312460e-01 5.78927696e-01 8.24812770e-01
4.17510085e-02 -1.76825488e+00 -1.10546803e+00 2.54865568e-02
8.18857431e-01 -6.85140789e-01 5.14352143e-01 7.77569115e-01
1.11456588e-01 4.50496167e-01 -4.92746711e-01 -6.83467150e-01
-1.19551921e+00 6.38628662e-01 3.41947317e-01 -2.85916775e-01
-4.53728139e-01 9.80066061e-01 6.57228231e-01 -8.23734701e-01
3.56346034e-02 4.04118896e-01 2.95018405e-02 4.71066117e-01
6.53618991e-01 4.74129692e-02 5.51291294e-02 -6.63641393e-01
-3.84284675e-01 7.05911994e-01 -4.43340272e-01 -4.10478890e-01
1.26562333e+00 -2.13817045e-01 1.98874906e-01 8.11734259e-01
1.40883863e+00 1.42719429e-02 -1.33955276e+00 -8.07209253e-01
-6.54560328e-02 -3.71302158e-01 -1.02177761e-01 -7.39799142e-01
-7.60149777e-01 1.24146056e+00 4.94529188e-01 -3.79929155e-01
6.16651118e-01 2.99905449e-01 7.49558628e-01 2.92497873e-01
7.23640025e-01 -1.16238213e+00 1.06961712e-01 8.28361273e-01
5.53802967e-01 -1.51078582e+00 -4.28840220e-01 -2.58238673e-01
-9.29182112e-01 5.67330718e-01 9.31511343e-01 -1.77773476e-01
4.56324935e-01 3.28835189e-01 2.35827997e-01 1.77655995e-01
-1.09576583e+00 -5.42102277e-01 4.10982519e-01 6.45234287e-01
8.71543288e-01 2.75487363e-01 -2.92479135e-02 3.63071144e-01
-6.58933073e-02 -1.07084982e-01 -7.58569911e-02 5.93677521e-01
-5.36745965e-01 -1.18830061e+00 -1.85402200e-01 -4.63297069e-02
-4.37867939e-01 -7.24530995e-01 -2.82416433e-01 1.01941764e+00
1.55695066e-01 5.60264528e-01 2.59653866e-01 -1.48494631e-01
3.54904592e-01 4.00621951e-01 8.71993899e-01 -9.71834481e-01
-7.29721844e-01 1.75301433e-01 1.73222482e-01 -3.27068925e-01
-1.00888439e-01 -5.88596165e-01 -8.68897259e-01 -2.39006042e-01
-1.56237856e-01 1.70379207e-01 4.80197102e-01 1.00968957e+00
4.12271649e-01 3.76844883e-01 4.71936792e-01 -5.88892639e-01
-5.47960699e-01 -1.19080329e+00 -3.59647930e-01 5.83289146e-01
2.53454328e-01 -7.80747831e-01 -2.58712053e-01 -2.20823232e-02] | [11.030548095703125, 9.751758575439453] |
360a6d59-18ba-4ab1-aec5-84e1b5af848b | is-gpt-4-a-good-data-analyst | 2305.15038 | null | https://arxiv.org/abs/2305.15038v1 | https://arxiv.org/pdf/2305.15038v1.pdf | Is GPT-4 a Good Data Analyst? | As large language models (LLMs) have demonstrated their powerful capabilities in plenty of domains and tasks, including context understanding, code generation, language generation, data storytelling, etc., many data analysts may raise concerns if their jobs will be replaced by AI. This controversial topic has drawn a lot of attention in public. However, we are still at a stage of divergent opinions without any definitive conclusion. Motivated by this, we raise the research question of "is GPT-4 a good data analyst?" in this work and aim to answer it by conducting head-to-head comparative studies. In detail, we regard GPT-4 as a data analyst to perform end-to-end data analysis with databases from a wide range of domains. We propose a framework to tackle the problems by carefully designing the prompts for GPT-4 to conduct experiments. We also design several task-specific evaluation metrics to systematically compare the performance between several professional human data analysts and GPT-4. Experimental results show that GPT-4 can achieve comparable performance to humans. We also provide in-depth discussions about our results to shed light on further studies before we reach the conclusion that GPT-4 can replace data analysts. | ['Lidong Bing', 'Xingxuan Li', 'Liying Cheng'] | 2023-05-24 | null | null | null | null | ['code-generation'] | ['computer-code'] | [-5.72498515e-02 1.19731188e-01 -1.61375016e-01 -5.10141790e-01
-8.87312651e-01 -4.20199990e-01 7.40093768e-01 4.37379986e-01
-3.04981381e-01 3.43219966e-01 2.46820211e-01 -6.43464983e-01
8.07727799e-02 -4.93643105e-01 -3.81735414e-01 -7.38414600e-02
-5.13960666e-04 6.14932060e-01 1.18218787e-01 -2.27717876e-01
6.41841710e-01 3.58217806e-02 -1.51990521e+00 7.07825005e-01
1.02995920e+00 4.88416851e-01 3.68116736e-01 2.94608802e-01
-4.23572898e-01 1.24297416e+00 -7.61494577e-01 -6.32721305e-01
3.10550407e-02 -1.90364122e-01 -1.19796348e+00 3.43931057e-02
-3.40596624e-02 1.75573155e-01 2.36766934e-01 1.01500940e+00
6.41202748e-01 -1.02786608e-01 2.14489967e-01 -1.61172497e+00
-7.77828455e-01 9.06465709e-01 -7.07584202e-01 4.72118109e-02
7.38453925e-01 2.73470998e-01 9.08840239e-01 -8.71770740e-01
6.72316611e-01 1.46997118e+00 4.38749939e-01 4.37229931e-01
-9.50612605e-01 -5.94783962e-01 3.48635256e-01 2.21072048e-01
-1.48134708e+00 -5.58079422e-01 6.68352485e-01 -7.60080516e-01
1.28873730e+00 2.99165934e-01 2.37073973e-01 1.06777990e+00
3.16959351e-01 1.00637305e+00 9.54009950e-01 -6.57212675e-01
3.19990396e-01 4.12665904e-01 2.79802978e-01 5.38925827e-01
2.26654246e-01 -2.10224628e-01 -7.51396894e-01 -4.61062163e-01
3.42375606e-01 -5.04162252e-01 7.45291710e-02 -1.11161515e-01
-1.37314177e+00 9.92761672e-01 6.99408501e-02 5.81515253e-01
-3.42619896e-01 -1.45317867e-01 5.54490626e-01 3.38417381e-01
4.80008125e-01 7.46018171e-01 -3.11597764e-01 -5.06874502e-01
-6.93096519e-01 6.86999261e-01 1.12736022e+00 1.32148659e+00
1.71603724e-01 -2.01880768e-01 -2.62590945e-01 8.26666892e-01
3.23826760e-01 2.45889947e-01 7.23370314e-01 -6.93852544e-01
6.34978414e-01 9.82834518e-01 1.53544173e-01 -1.05168974e+00
-3.46663892e-01 -1.13867298e-01 -8.04743826e-01 7.12743821e-03
2.29513288e-01 -3.56149793e-01 -3.72367710e-01 1.51654458e+00
-3.31732002e-03 -2.65682399e-01 1.91079259e-01 8.64992738e-01
8.95813704e-01 6.96641386e-01 2.61722296e-01 -1.42903164e-01
1.71323621e+00 -8.66097867e-01 -6.97336733e-01 -7.15864003e-01
1.02006233e+00 -9.63075161e-01 1.33746564e+00 5.77360094e-01
-1.03735137e+00 -7.12068260e-01 -7.85112619e-01 -2.94625640e-01
-2.89328068e-01 1.80630609e-01 6.90598190e-01 4.19096500e-01
-9.03217077e-01 8.02253559e-02 -8.53489459e-01 -6.89084113e-01
8.70802999e-02 -1.21890716e-01 -1.27286017e-01 9.19461995e-02
-1.27540863e+00 8.95339191e-01 5.15237510e-01 -1.50535479e-01
-6.48553610e-01 -4.23676401e-01 -6.56509280e-01 -2.61921078e-01
6.92874908e-01 -8.21257472e-01 1.69900537e+00 -4.02732939e-01
-9.54793513e-01 1.24207616e+00 -4.82558250e-01 -5.49930692e-01
5.05855024e-01 -3.01344126e-01 -6.57567859e-01 -6.30257308e-01
3.82958055e-01 4.04902190e-01 1.82155654e-01 -1.15067005e+00
-8.20655107e-01 -2.93257266e-01 1.88422173e-01 1.19916670e-01
-1.26831532e-01 6.78910315e-01 -5.27711153e-01 -7.23017335e-01
-2.80096978e-01 -9.44780588e-01 -5.08965671e-01 -3.76035064e-01
-5.39825320e-01 -5.98461270e-01 5.70556104e-01 -3.97072047e-01
1.81616700e+00 -2.15142107e+00 -1.24986231e-01 -1.22811787e-01
4.12021965e-01 3.69551510e-01 -5.03442660e-02 8.31579089e-01
-7.17195868e-02 5.02740085e-01 -2.24164590e-01 -2.59200752e-01
1.50099948e-01 1.31346183e-02 -5.32939255e-01 -3.75653133e-02
1.63619876e-01 9.48758304e-01 -8.48128974e-01 -5.89460671e-01
-1.87604472e-01 9.62379128e-02 -6.19059682e-01 4.37423468e-01
-5.72873890e-01 2.65097022e-01 -8.40566456e-01 5.33897638e-01
3.20454210e-01 -5.29906571e-01 7.40678161e-02 3.36745560e-01
-3.82585973e-01 3.34225625e-01 -1.04249012e+00 1.86306334e+00
-4.18243170e-01 7.54824221e-01 -1.57580376e-01 -8.54763448e-01
1.08401871e+00 3.04652721e-01 -1.43812792e-02 -6.61803424e-01
4.97460105e-02 1.47094876e-01 2.27590486e-01 -9.31162298e-01
7.24681616e-01 -1.98916420e-01 -6.06382847e-01 5.37832558e-01
-5.63436568e-01 -1.48443133e-01 4.33550835e-01 1.65883183e-01
9.80024934e-01 -1.74971342e-01 4.70455348e-01 -3.02118659e-01
5.02242863e-01 4.02662575e-01 4.05760914e-01 7.67427325e-01
3.87483351e-02 2.98182636e-01 8.35485935e-01 -4.89246249e-01
-6.66951776e-01 -4.42623973e-01 7.00394958e-02 1.34331203e+00
-1.46973440e-02 -9.53775108e-01 -8.78778458e-01 -5.46642661e-01
-8.28362703e-02 1.26538146e+00 -4.24032092e-01 4.72733229e-02
-4.61791277e-01 -8.04586887e-01 5.80227435e-01 4.69619453e-01
6.76519632e-01 -1.28750670e+00 -8.43239248e-01 2.80754000e-01
-5.40050030e-01 -1.16923273e+00 -2.32310832e-01 -3.96904834e-02
-6.19565010e-01 -9.41622078e-01 -2.57398725e-01 -8.45434368e-01
4.71188366e-01 3.45894486e-01 1.44716108e+00 1.74158201e-01
-7.10180551e-02 6.22135065e-02 -6.31724417e-01 -9.42711294e-01
-9.17392135e-01 2.25852862e-01 -1.94272518e-01 -4.38767612e-01
9.75691795e-01 -2.69185543e-01 -1.93562269e-01 6.13207221e-01
-1.00837171e+00 4.51866478e-01 6.25334263e-01 3.12586874e-01
1.91814765e-01 2.23221913e-01 6.07196331e-01 -1.29741275e+00
1.42667568e+00 -6.26804352e-01 -5.74674547e-01 5.64614832e-01
-6.80836856e-01 2.12921172e-01 6.00573421e-01 -1.89345881e-01
-1.11476171e+00 -4.07399595e-01 -5.44313788e-02 8.47596377e-02
-1.65664151e-01 1.16355026e+00 -2.89252847e-01 5.20310700e-01
1.00772750e+00 3.89780030e-02 -2.07037002e-01 -6.59175932e-01
3.81709754e-01 9.72821593e-01 4.16481435e-01 -9.49043810e-01
5.90260983e-01 2.94912737e-02 -7.90664732e-01 -7.43063211e-01
-7.03165233e-01 -3.41532916e-01 -3.83524418e-01 1.26376584e-01
6.94891453e-01 -9.76592064e-01 -6.03588700e-01 3.33057761e-01
-1.40326619e+00 -2.55997926e-01 -1.22033269e-03 1.03035733e-01
-3.61328125e-01 2.94065565e-01 -3.60976070e-01 -7.35073507e-01
-3.75207484e-01 -1.24627948e+00 1.01029074e+00 6.28826246e-02
-8.39337826e-01 -1.07764339e+00 6.55911043e-02 6.28180385e-01
4.07245338e-01 2.20882252e-01 1.05104828e+00 -8.71023118e-01
-4.39808637e-01 -3.90427023e-01 -1.71659395e-01 3.06803500e-04
-6.26825616e-02 5.36239967e-02 -9.72234070e-01 -2.71471798e-01
1.30647704e-01 -2.38377601e-01 3.12466413e-01 9.52731147e-02
1.43665266e+00 -3.30377638e-01 -5.35365582e-01 2.15498239e-01
1.08712935e+00 4.90808785e-01 4.35902983e-01 2.68991739e-01
5.57757676e-01 9.95416701e-01 8.72249663e-01 6.96276784e-01
6.45603538e-01 6.56532824e-01 3.43279727e-03 2.23536938e-02
1.29456356e-01 -4.62767005e-01 3.54985207e-01 8.02442431e-01
4.61978558e-03 -5.55331945e-01 -1.57445121e+00 6.01578772e-01
-2.00852990e+00 -6.71180487e-01 -2.93368667e-01 1.94190812e+00
8.50872636e-01 3.62328798e-01 1.19427279e-01 -1.22893289e-01
5.32572210e-01 1.27525181e-01 -3.80721360e-01 -4.51634258e-01
3.70691679e-02 -3.65373462e-01 -6.29950240e-02 1.67770967e-01
-7.20637321e-01 9.83667970e-01 6.18814373e+00 8.97101164e-01
-1.09655392e+00 -9.65960398e-02 6.58206999e-01 2.02299774e-01
-3.78996611e-01 3.94879341e-01 -7.93950021e-01 4.68561381e-01
1.10583961e+00 -1.02039599e+00 2.47661069e-01 1.05184054e+00
4.24308538e-01 -2.03312084e-01 -1.51169610e+00 1.07093763e+00
-5.53645752e-02 -1.12780821e+00 1.29725382e-01 1.36596173e-01
4.72005486e-01 -3.57194394e-02 -9.56854597e-02 6.24880016e-01
5.50147533e-01 -1.17728078e+00 8.07139516e-01 3.63130689e-01
5.23099124e-01 -5.12173414e-01 6.89788342e-01 8.81954074e-01
-1.00261724e+00 -3.38005871e-02 -3.68818760e-01 -4.19915140e-01
2.46935979e-01 7.56724656e-01 -1.04511154e+00 6.96574271e-01
3.48581195e-01 5.35841167e-01 -7.87438631e-01 8.80124867e-01
-9.54893827e-02 4.38444287e-01 1.24746546e-01 -1.36519626e-01
1.71735883e-02 2.12772533e-01 3.66894960e-01 1.31655562e+00
3.47967923e-01 1.74614593e-01 3.53969544e-01 1.12861276e+00
-3.39371897e-02 2.44943604e-01 -8.15209508e-01 -5.00778615e-01
5.86850405e-01 9.56205249e-01 -5.50981760e-01 -5.03927350e-01
-6.23766840e-01 5.34598649e-01 2.42570102e-01 2.04523176e-01
-5.37286103e-01 -4.66993451e-01 4.82317924e-01 4.15313780e-01
-4.15549040e-01 -2.48789087e-01 -3.78223270e-01 -1.24238443e+00
2.72150248e-01 -1.42578721e+00 4.77501214e-01 -9.62964535e-01
-1.37918627e+00 9.97677267e-01 3.76665652e-01 -1.28234243e+00
-5.12666047e-01 -4.24126685e-01 -5.55972695e-01 8.55448067e-01
-1.08512068e+00 -8.75866413e-01 -3.75085205e-01 3.26238573e-01
7.15331852e-01 -3.03418249e-01 8.53774607e-01 1.80968121e-01
-6.51354432e-01 3.70183647e-01 -3.85377735e-01 3.13217461e-01
7.22630680e-01 -1.10690761e+00 1.09828222e+00 9.78805721e-01
-1.96834770e-03 1.13131261e+00 1.05191290e+00 -8.09123635e-01
-1.33783352e+00 -9.83012676e-01 1.34060216e+00 -8.83817017e-01
6.88841581e-01 -4.87499923e-01 -1.09093821e+00 7.20466733e-01
4.46973205e-01 -3.65630865e-01 8.89397144e-01 2.53051430e-01
-1.18642494e-01 2.40978345e-01 -8.90566230e-01 5.77844203e-01
1.05212617e+00 -4.68966544e-01 -9.49541330e-01 3.94027531e-01
8.06240737e-01 -4.62343007e-01 -7.23317087e-01 2.01713353e-01
2.44465917e-01 -9.54714000e-01 5.79109073e-01 -6.61393404e-01
6.92638338e-01 -2.51086980e-01 -1.19983386e-02 -1.22063422e+00
-6.98434934e-02 -8.33621800e-01 4.03109372e-01 1.35255075e+00
4.67540234e-01 -6.12834930e-01 3.82578760e-01 1.13290393e+00
-1.19749956e-01 -5.15672445e-01 -4.93239164e-01 -7.38006175e-01
2.63570219e-01 -9.28387165e-01 7.61364460e-01 1.15529239e+00
4.62911546e-01 6.42528057e-01 -3.32834661e-01 3.62279527e-02
2.50004560e-01 4.00078028e-01 1.15296948e+00 -1.35173953e+00
-3.74721289e-02 -5.03675222e-01 7.39788041e-02 -1.14773178e+00
7.44078448e-03 -1.07984078e+00 -9.51574445e-02 -1.58058858e+00
2.85998821e-01 -3.09137851e-01 2.12491751e-01 5.96118391e-01
-1.76620334e-01 -5.65632761e-01 1.29700616e-01 3.66437376e-01
-7.79299021e-01 2.93656617e-01 1.12030780e+00 6.75645098e-02
-3.70129883e-01 1.84321105e-02 -1.37649012e+00 7.88620651e-01
6.85689628e-01 -4.20284897e-01 -7.05859303e-01 -5.63290417e-01
5.68427801e-01 2.79399127e-01 1.04658380e-01 -8.12241971e-01
4.42358911e-01 -5.07942379e-01 -1.76581651e-01 -3.74237597e-01
-3.18558216e-01 -6.61600769e-01 1.62422642e-01 2.60876685e-01
-6.48699522e-01 3.21102202e-01 4.44786489e-01 2.22595543e-01
-4.71380979e-01 -2.61030853e-01 3.20522904e-01 -2.48108998e-01
-9.68219042e-01 4.63806605e-03 -6.19409084e-01 3.67938757e-01
1.01763713e+00 -2.47056596e-02 -3.96386415e-01 -4.18738514e-01
-3.03272396e-01 6.90663457e-01 4.60490167e-01 9.19910729e-01
3.95384610e-01 -1.11242795e+00 -1.03702605e+00 3.24655712e-01
7.23347306e-01 1.61763906e-01 8.37848112e-02 5.02600670e-01
-4.44196105e-01 7.77668893e-01 1.04743741e-01 -4.67789799e-01
-1.29618752e+00 7.64291346e-01 -2.97489315e-02 -4.89976138e-01
-4.90901262e-01 6.92887187e-01 3.83795977e-01 -4.79949087e-01
2.50915438e-01 -6.58230484e-01 -1.30220786e-01 6.82041422e-02
7.86197305e-01 1.61826938e-01 1.27434209e-01 -3.12296003e-01
-4.14478689e-01 1.87500402e-01 -3.16662848e-01 4.06390727e-02
1.28321826e+00 -6.53410703e-02 -1.19403966e-01 6.55065715e-01
7.70154119e-01 2.22453363e-02 -6.23862386e-01 -4.18458313e-01
5.65805852e-01 -5.57668388e-01 -2.12855637e-01 -1.08019459e+00
-6.26782298e-01 9.00814474e-01 2.41130859e-01 6.90938711e-01
1.10769129e+00 2.32258379e-01 4.83097434e-01 4.82343644e-01
7.38337517e-01 -9.78400171e-01 -4.15759720e-03 6.02922142e-01
1.38017178e+00 -1.39965892e+00 -1.39357120e-01 -2.80408025e-01
-1.06122887e+00 8.89862537e-01 8.13413501e-01 3.66925776e-01
3.21176976e-01 3.77426744e-01 2.25513428e-01 -4.91797447e-01
-1.28075910e+00 6.83108196e-02 2.70707667e-01 4.07482415e-01
1.00888538e+00 4.84155640e-02 -4.20186192e-01 9.28616285e-01
-6.47476614e-01 3.49162310e-01 6.03711724e-01 1.22607529e+00
-4.37120557e-01 -1.49162591e+00 -4.14869189e-01 3.14751297e-01
-3.59836280e-01 -7.04227760e-02 -5.86669087e-01 9.09442306e-01
-7.53753707e-02 1.26337564e+00 -2.62528419e-01 -5.06081164e-01
6.37338042e-01 1.93054393e-01 -7.11003616e-02 -8.96517396e-01
-6.06409431e-01 -3.67447063e-02 3.91063988e-01 -3.71685416e-01
-4.02343094e-01 -5.36905468e-01 -1.17313421e+00 -5.93023062e-01
-2.57569947e-04 3.98386747e-01 4.41444129e-01 7.38359213e-01
5.96073687e-01 3.66866708e-01 2.29918331e-01 -1.16079800e-01
-3.82674754e-01 -1.17349362e+00 -2.86587298e-01 3.64033371e-01
-9.86918136e-02 -3.75586361e-01 -8.13584775e-02 2.27383375e-01] | [10.799031257629395, 8.70228385925293] |
a78a7dc4-ce51-42d9-9960-d529ad61b44e | three-stream-convolutional-neural-network | null | null | http://openaccess.thecvf.com/content_CVPRW_2019/html/PBVS/Liang_Three-Stream_Convolutional_Neural_Network_With_Multi-Task_and_Ensemble_Learning_for_CVPRW_2019_paper.html | http://openaccess.thecvf.com/content_CVPRW_2019/papers/PBVS/Liang_Three-Stream_Convolutional_Neural_Network_With_Multi-Task_and_Ensemble_Learning_for_CVPRW_2019_paper.pdf | Three-Stream Convolutional Neural Network With Multi-Task and Ensemble Learning for 3D Action Recognition | In this paper, we propose a three-stream convolutional neural network (3SCNN) for action recognition from skeleton sequences, which aims to thoroughly and fully exploit the skeleton data by extracting, learning, fusing and inferring multiple motion-related features, including 3D joint positions and joint displacements across adjacent frames as well as oriented bone segments. The proposed 3SCNN involves three sequential stages. The first stage enriches three independently extracted features by co-occurrence feature learning. The second stage involves multi-channel pairwise fusion to take advantage of the complementary and diverse nature among three features. The third stage is a multi-task and ensemble learning network to further improve the generalization ability of 3SCNN. Experimental results on the standard dataset show the effectiveness of our proposed multi-stream feature learning, fusion and inference method for skeleton-based 3D action recognition. | ['Hong Zhu', 'Duohan Liang', 'Wanjun Chen', 'Xiaorong Pan', 'Guoliang Fan', 'Guangfeng Lin'] | 2019-06-16 | null | null | null | the-ieee-conference-on-computer-vision-and-1 | ['3d-human-action-recognition'] | ['computer-vision'] | [ 6.66369200e-01 -4.11373138e-01 -2.37626091e-01 -3.07825565e-01
-7.89109707e-01 1.32003397e-01 4.40611005e-01 -2.45915353e-01
-5.09130299e-01 5.17153442e-01 5.97043753e-01 3.78532499e-01
-3.74166310e-01 -4.46942359e-01 -4.82929617e-01 -8.46239507e-01
-2.90349782e-01 1.39683187e-01 6.21740401e-01 -2.35720016e-02
2.10936189e-01 6.58104479e-01 -1.66752088e+00 5.02580583e-01
3.94901007e-01 1.28252041e+00 -1.05814703e-01 7.59359956e-01
3.15317214e-02 8.95207584e-01 -3.22429568e-01 1.08398478e-02
2.44413525e-01 -3.99446189e-01 -6.95723891e-01 4.46832031e-01
3.45175624e-01 -5.09312749e-01 -2.91428477e-01 6.62543595e-01
7.67634928e-01 1.65525153e-01 5.76917231e-01 -1.10191846e+00
-1.19072571e-02 3.96722764e-01 -7.69083798e-01 4.30482388e-01
4.51830477e-01 3.33114326e-01 7.74784267e-01 -8.57116878e-01
4.38156247e-01 1.26087523e+00 6.91270649e-01 3.34818244e-01
-8.11342299e-01 -6.30647838e-01 -5.99243306e-02 5.02713561e-01
-1.15561116e+00 -4.15243179e-01 8.96660924e-01 -3.76321763e-01
9.07901943e-01 -1.33537441e-01 1.00213671e+00 1.37089860e+00
3.69305998e-01 1.25269938e+00 1.06950021e+00 -2.49879286e-01
-3.77945714e-02 -9.36119616e-01 1.28946617e-01 9.95102108e-01
-6.48661479e-02 4.48825210e-02 -1.04390347e+00 1.02676824e-01
1.05090702e+00 2.28215039e-01 5.27165271e-02 -3.36774260e-01
-1.52310836e+00 4.05737847e-01 1.90384924e-01 4.08700675e-01
-6.69631898e-01 3.23167771e-01 5.78305185e-01 7.63477292e-03
2.96891809e-01 -3.92824471e-01 -5.27717531e-01 -5.44685900e-01
-9.73987162e-01 2.54330516e-01 1.52899027e-01 4.97514158e-01
6.60241008e-01 -1.24539584e-02 -3.51427913e-01 7.39814758e-01
4.36821789e-01 5.92337608e-01 9.09342945e-01 -1.10465884e+00
4.85628515e-01 7.03219593e-01 -4.17788118e-01 -8.42722595e-01
-6.60311401e-01 -2.52124041e-01 -9.11293089e-01 2.75236160e-01
3.77488166e-01 -6.26148656e-02 -9.70926046e-01 1.47255480e+00
6.36641204e-01 3.90773326e-01 -1.16424553e-01 9.92691576e-01
8.99477601e-01 3.41315418e-02 -2.52781436e-03 -3.70849011e-04
1.21187055e+00 -9.35520768e-01 -4.39069778e-01 2.03836113e-01
4.14580494e-01 -3.60820800e-01 3.48731905e-01 2.79835582e-01
-9.46738124e-01 -1.01239944e+00 -9.76751745e-01 3.44618820e-02
-4.71580811e-02 2.09985197e-01 7.23851025e-01 1.38097137e-01
-4.71938521e-01 5.94348192e-01 -1.35398066e+00 -1.34796470e-01
9.87645328e-01 5.87825060e-01 -7.70521283e-01 5.26617058e-02
-1.06195378e+00 5.18230677e-01 3.98717403e-01 3.46903056e-01
-8.62832546e-01 -3.00489038e-01 -9.98909891e-01 -2.09581181e-01
3.85157734e-01 -8.44923139e-01 1.03208137e+00 -9.10396159e-01
-1.64511001e+00 5.07841051e-01 -2.56562293e-01 -3.61361623e-01
4.22074884e-01 -4.90659326e-01 -1.77041501e-01 4.86609340e-01
1.91730902e-01 6.52652979e-01 9.75860953e-01 -6.77041411e-01
-7.68009186e-01 -9.54926312e-01 -2.78318733e-01 4.08067584e-01
-1.71270266e-01 -1.42031834e-01 -4.25544739e-01 -7.18683839e-01
3.95387024e-01 -7.27185786e-01 -2.37597823e-01 1.51215777e-01
-2.64932066e-01 -3.43733996e-01 8.27070475e-01 -5.22176683e-01
9.46886182e-01 -1.92096627e+00 6.21126533e-01 1.99729636e-01
8.55383426e-02 3.60730946e-01 -2.84641683e-02 1.27712727e-01
-1.07729249e-01 -6.02895379e-01 -4.05778676e-01 -3.43500406e-01
-2.24432439e-01 3.60328853e-01 5.02031446e-01 5.31028807e-01
5.82934558e-01 1.04558825e+00 -7.54903734e-01 -9.61805999e-01
4.79228765e-01 6.71239853e-01 -2.38096923e-01 -8.27972069e-02
1.00964412e-01 7.74423182e-01 -8.84765983e-01 1.12635195e+00
2.45770037e-01 -1.29063696e-01 -2.64727712e-01 -3.34201187e-01
1.82838753e-01 -3.12430948e-01 -1.34467137e+00 2.47317219e+00
-9.64823663e-02 2.49767974e-01 -3.23580593e-01 -1.33247554e+00
8.91609132e-01 2.55730093e-01 1.21487319e+00 -6.20758235e-01
3.55987340e-01 1.33183137e-01 -1.52602002e-01 -9.85412538e-01
2.86218561e-02 -1.06629193e-01 -7.75505304e-02 5.41618526e-01
3.68657053e-01 4.66135770e-01 6.23686910e-02 -1.59717068e-01
1.19081843e+00 7.83454120e-01 4.32674676e-01 2.67195612e-01
9.99829054e-01 -5.40062487e-01 8.34887385e-01 3.36364388e-01
-5.95331073e-01 4.82259721e-01 2.40022078e-01 -4.40147579e-01
-4.73201722e-01 -9.94201839e-01 8.17630664e-02 6.80973113e-01
-1.04627786e-02 -1.24758944e-01 -3.41681659e-01 -1.06078994e+00
3.23064536e-01 -1.97645143e-01 -7.64940560e-01 -2.14296713e-01
-7.91150451e-01 -4.77550507e-01 6.72798932e-01 1.04916465e+00
8.32167685e-01 -1.06778169e+00 -1.15674782e+00 2.19911978e-01
-2.72706896e-01 -1.06390464e+00 -3.00559461e-01 2.22846121e-01
-1.17096579e+00 -1.20976484e+00 -8.36031735e-01 -4.24995065e-01
3.10169846e-01 7.20142499e-02 5.19593596e-01 -7.08054304e-02
-5.11721730e-01 5.13578176e-01 -5.22287965e-01 -7.91634545e-02
1.35984316e-01 -1.95374474e-01 3.93239744e-02 5.02901495e-01
3.84960920e-01 -1.09715605e+00 -5.24277151e-01 8.99276435e-02
-8.49181890e-01 -8.62232670e-02 1.08339643e+00 9.21161830e-01
7.90260673e-01 -6.92027388e-03 4.89410371e-01 -3.16990197e-01
6.32667989e-02 -2.58560300e-01 2.91720498e-02 1.27086952e-01
-1.19083092e-01 1.77884966e-01 1.12296589e-01 -3.35888922e-01
-1.05192697e+00 5.85149229e-01 -1.66379124e-01 -5.02793431e-01
-5.46199083e-01 5.69125593e-01 -3.26321214e-01 2.13224106e-02
2.55565852e-01 5.66972971e-01 3.38175029e-01 -5.70585907e-01
2.66145527e-01 5.71449578e-01 7.89651752e-01 -5.62657654e-01
5.15100420e-01 6.79090202e-01 5.77124774e-01 -8.11191559e-01
-7.14448273e-01 -6.86345279e-01 -1.38027275e+00 -6.78717256e-01
1.19648826e+00 -9.29709494e-01 -6.17430925e-01 1.15455544e+00
-8.09817493e-01 -7.81893823e-03 -4.35814738e-01 8.88454616e-01
-8.17251980e-01 7.87239790e-01 -5.26842594e-01 -7.81330526e-01
-3.41139108e-01 -1.11585236e+00 1.56812203e+00 2.15539485e-01
-1.41908349e-02 -7.20245004e-01 1.69573545e-01 7.91699529e-01
-1.90791041e-02 8.49974513e-01 4.40218985e-01 -6.26206517e-01
-3.80386531e-01 -2.22825736e-01 -8.56567845e-02 4.28589940e-01
3.84934217e-01 -1.05789341e-01 -5.82725346e-01 3.49047147e-02
-3.21467608e-01 -6.90083861e-01 1.17283452e+00 6.16299570e-01
1.10406280e+00 3.88700098e-01 -2.06873700e-01 4.97379482e-01
1.07901835e+00 3.61187495e-02 5.03481984e-01 2.44379088e-01
9.34125781e-01 2.75683880e-01 6.55588806e-01 7.11020947e-01
3.53070587e-01 6.45549357e-01 4.69431221e-01 1.38775051e-01
-2.58647770e-01 4.27188911e-02 5.20715117e-01 8.58977020e-01
-7.72175431e-01 5.22661090e-01 -6.23213828e-01 3.61931324e-01
-1.95493162e+00 -1.19030583e+00 -8.82425979e-02 1.75850821e+00
5.93517244e-01 2.51936793e-01 4.44433749e-01 5.65938115e-01
4.62320238e-01 4.53199476e-01 -7.16194212e-01 2.34440133e-01
-1.14080951e-01 6.12080872e-01 2.86320299e-01 2.15600487e-02
-1.38656282e+00 5.79351485e-01 5.19184208e+00 7.92391300e-01
-8.46265912e-01 -5.16004451e-02 7.82471150e-02 -1.28324404e-01
3.02281529e-01 -2.98169106e-01 -6.60544038e-01 3.41807157e-01
4.90313232e-01 3.12566370e-01 -2.40856886e-01 4.70913500e-01
1.40800580e-01 -2.64174074e-01 -8.55113268e-01 8.66891563e-01
3.35554868e-01 -1.07326496e+00 2.00002268e-02 9.08004940e-02
5.58599532e-01 3.38114277e-02 -2.34421745e-01 -9.57856793e-03
-7.08333626e-02 -7.47158289e-01 6.14882886e-01 1.10134339e+00
3.82377267e-01 -8.79189432e-01 7.02794969e-01 3.50730270e-01
-1.47733009e+00 -2.71368414e-01 2.29924798e-01 -7.58041143e-02
3.73636812e-01 5.56465089e-01 -1.19870275e-01 1.00523961e+00
7.91191280e-01 1.57277906e+00 -6.40919089e-01 9.77028489e-01
-1.34959996e-01 3.35867047e-01 -3.07882369e-01 6.89487755e-02
3.15276623e-01 7.93808550e-02 6.02795541e-01 1.01284873e+00
9.86836031e-02 1.96956947e-01 2.62418777e-01 3.40004623e-01
3.43461245e-01 -1.72507152e-01 -3.52764100e-01 2.25493293e-02
-1.51448622e-01 1.19694567e+00 -8.26179922e-01 -2.74028063e-01
-4.22167301e-01 1.07813346e+00 5.98975681e-02 9.70561206e-02
-9.21676278e-01 -2.23334804e-01 5.59871316e-01 -3.56748581e-01
7.93363869e-01 -5.95377803e-01 -5.16322106e-02 -1.19440603e+00
1.96349040e-01 -7.73880422e-01 5.31278491e-01 -4.55986828e-01
-1.03023648e+00 5.74900173e-02 1.43367037e-01 -1.36500394e+00
-1.72673926e-01 -6.26918256e-01 -5.04297137e-01 4.18460548e-01
-1.41866887e+00 -1.64229977e+00 -2.58987635e-01 1.15781331e+00
8.28982532e-01 -3.46978962e-01 5.45112789e-01 3.55967224e-01
-7.77427614e-01 3.05628121e-01 -2.92588830e-01 3.54606926e-01
4.42962736e-01 -1.03907442e+00 3.99431400e-02 6.48658156e-01
2.75877267e-01 1.44687593e-01 -9.91178304e-02 -6.15182996e-01
-1.45012403e+00 -9.51718509e-01 5.33100903e-01 -2.29967758e-01
4.34848875e-01 2.17352316e-01 -5.43102860e-01 4.46451575e-01
-4.14554119e-01 3.04708481e-01 8.18050027e-01 -1.10086016e-01
-1.76149011e-01 -1.98852211e-01 -8.34184349e-01 9.67163071e-02
1.34320498e+00 -3.58282179e-01 -7.94648647e-01 7.51632228e-02
1.92889899e-01 -4.22756284e-01 -1.34589052e+00 7.36647785e-01
1.17511189e+00 -1.11753547e+00 1.20244169e+00 -7.29668021e-01
7.06710219e-01 -3.12614560e-01 -3.36543202e-01 -8.38728249e-01
-1.84695154e-01 -1.81581721e-01 -4.41179305e-01 8.15308213e-01
1.04350634e-01 -1.70821816e-01 9.46362138e-01 -3.95991094e-02
-2.59551913e-01 -9.77932572e-01 -1.32902825e+00 -6.31417274e-01
-2.29464903e-01 -7.96448827e-01 4.18741107e-01 5.30910730e-01
-1.12090498e-01 2.72632301e-01 -5.05661309e-01 -2.30530836e-02
7.89956450e-01 2.49189958e-01 8.69177163e-01 -1.28071249e+00
-4.49494958e-01 -4.02548134e-01 -1.07537127e+00 -1.17325163e+00
2.57930875e-01 -8.55165958e-01 -6.32969961e-02 -1.39075768e+00
2.34206304e-01 1.45385206e-01 -5.53324044e-01 6.71155870e-01
-2.35528499e-01 2.45692715e-01 -5.71789183e-02 5.94579279e-02
-9.33411181e-01 8.03772151e-01 1.58469129e+00 -1.77314669e-01
4.41717096e-02 2.15033293e-01 -1.76728144e-01 7.92470098e-01
3.79569829e-01 -2.20108494e-01 -3.42037529e-01 -2.86813021e-01
-4.25678790e-01 3.62140059e-01 7.78620362e-01 -1.38827395e+00
1.73143953e-01 -9.30233225e-02 9.67964768e-01 -1.01078320e+00
5.57341993e-01 -9.30700779e-01 -4.58361767e-02 7.25441933e-01
-2.16977984e-01 -2.71022052e-01 -1.01522930e-01 8.28839302e-01
-3.50744456e-01 1.65481254e-01 4.39989835e-01 -1.86634213e-01
-1.06403232e+00 5.62818646e-01 -3.49849701e-01 -2.10916832e-01
1.08645868e+00 -6.43178701e-01 2.48629510e-01 1.02012753e-02
-1.16695333e+00 1.76391825e-01 -1.67916074e-01 4.89230573e-01
9.87169504e-01 -1.73034561e+00 -7.32802749e-01 4.62062448e-01
7.71154836e-02 7.14464858e-02 5.35188317e-01 1.29959035e+00
-1.55035928e-01 1.97711363e-01 -7.13953078e-01 -1.09234297e+00
-1.47937477e+00 -1.81089621e-02 2.94693738e-01 -3.33255202e-01
-8.25602055e-01 9.15088236e-01 -5.10695755e-01 -6.55411407e-02
2.59658545e-01 -3.81439298e-01 -4.39457566e-01 1.40285119e-01
3.66735578e-01 7.14956880e-01 -7.01020434e-02 -1.16273868e+00
-6.34719610e-01 1.12380147e+00 1.93961516e-01 -1.18983798e-01
1.48336899e+00 1.99912596e-04 6.09797612e-02 5.72832704e-01
1.32518995e+00 -6.55446529e-01 -1.52605438e+00 -4.10942286e-01
3.40221077e-02 -4.94180650e-01 -7.59272054e-02 -5.38046300e-01
-1.42225301e+00 8.88587773e-01 6.14549220e-01 -3.86034280e-01
1.45228553e+00 -1.50105476e-01 1.19390762e+00 3.37844819e-01
3.16130340e-01 -1.16917932e+00 4.99864310e-01 3.97095889e-01
6.45765901e-01 -1.22212851e+00 2.83018798e-01 5.44379875e-02
-5.62595129e-01 1.40596282e+00 5.59316576e-01 -1.97380304e-01
9.47905362e-01 2.14469135e-02 -1.27412692e-01 -4.61115152e-01
-5.35991728e-01 -5.63787043e-01 4.05317098e-01 7.23104894e-01
2.11008295e-01 -1.76286519e-01 -4.55194443e-01 6.02272987e-01
3.10126990e-01 4.54353154e-01 -3.51208728e-03 1.34245384e+00
-4.05487090e-01 -1.12427747e+00 -2.46693939e-01 2.98544407e-01
-3.45011622e-01 4.90306735e-01 -3.44881684e-01 7.55851746e-01
6.31442308e-01 6.42966688e-01 -1.85248435e-01 -7.63715446e-01
4.67616498e-01 2.97914296e-01 8.58942688e-01 -2.51532853e-01
-4.61840123e-01 2.38592505e-01 4.66367416e-02 -1.13844740e+00
-1.32701218e+00 -1.18264556e+00 -1.46332514e+00 4.23663020e-01
-2.25372165e-01 -3.47600192e-01 4.92613852e-01 1.44005108e+00
2.39448443e-01 7.24679470e-01 6.28750980e-01 -1.20452559e+00
-5.74893296e-01 -1.01213086e+00 -5.73430181e-01 3.51158708e-01
4.33570862e-01 -1.19443262e+00 -5.85247464e-02 3.43366057e-01] | [7.842494487762451, 0.3504156470298767] |
fa9e7d98-4210-437a-a660-6d608a010737 | satellite-image-small-target-application | null | null | https://ieeexplore.ieee.org/abstract/document/9233819 | https://ieeexplore.ieee.org/abstract/document/9233819 | Satellite Image Small Target Application Based on Deep Segmented Residual Neural Network | This study employs a deep segmented residual neural network model to analyze the super-resolution of a single satellite image. A deep convolutional neural network model was analyzed, and its performance was improved. We proposed two residual layers to divide the deep network into two groups, the sum of the two residuals is the total residual, which can minimize the residual loss function and enhance the network performance. The experimental model achieved high peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) than networks without the proposed improvements when tested on satellite images. Considering these results, the application of this technology will be significant for further research on satellite images. | ['Yunqing Liu', 'Zikang Wei'] | 2020-10-26 | null | null | null | null | ['satellite-image-super-resolution'] | ['computer-vision'] | [ 0.21343476 -0.25466868 0.3234846 -0.2027449 -0.22552861 0.01287158
0.09424251 -0.6401011 -0.36904532 0.66104066 0.22465666 -0.01742836
-0.12709023 -1.0276184 -0.27693975 -1.0078496 -0.57237905 -0.58282715
0.534587 -0.4656535 0.24439514 0.7066484 -1.6808574 0.21914306
1.1048813 1.2632916 0.54975426 0.4773479 0.08073016 1.173664
-0.60471386 0.19213696 0.5060871 -0.3748418 -0.62723184 0.02185857
0.24512586 -0.86512715 -0.70691997 1.4836963 0.8539813 0.36300084
0.05562961 -0.53615516 -1.0025842 0.6303355 -0.7098294 0.8883578
-0.3027255 -0.06315171 0.6315891 -0.7959588 0.34044343 1.2975836
0.98361903 -0.08533245 -1.1398655 -0.79144084 -0.15749513 0.33233175
-1.6239251 -0.03377614 0.45609525 -0.18861674 0.62559766 0.24928772
0.37976024 0.22127277 0.38609588 0.3993767 1.1419442 0.08659534
-0.21324292 -0.18308607 0.22138694 0.38620105 0.2738513 0.33985153
0.37166485 0.5814146 1.2882404 0.01486472 -0.6693586 0.49352363
-0.7465528 0.8982314 1.1408995 0.7287859 -0.3745949 -0.17465106
0.3353473 0.35363367 0.6568362 -0.02529124 -0.29458454 0.60393846
-1.1723113 -0.04529687 0.14257959 0.2838805 0.7776009 0.7860797
-0.0784081 1.0812112 0.11634085 0.3122055 0.7782347 -1.1600657
0.10266055 0.55335855 -0.22871935 -1.590985 -0.55503494 -0.9815668
-1.4157326 0.5183465 -0.3093967 -0.11511286 -0.8846747 1.2870978
-0.04973686 0.29326487 0.22262187 1.4413033 1.4177537 0.98209
-0.23711239 -0.35014465 1.0068493 -0.91717654 -0.7705426 -0.12486362
0.22166151 -0.5877949 0.65621865 0.20021631 -1.0975735 -1.1543123
-1.281555 -0.03640606 -0.0732684 0.2751801 0.33034104 0.22058833
-1.3349204 1.2050306 -0.5687401 -0.01559949 0.325473 0.41263813
-0.35502997 0.2327176 -1.5279778 0.7965018 0.570873 0.8243108
-0.6924077 -0.29374442 -0.62979406 0.28946874 -0.11718887 -0.17290354
0.8742769 -1.2767205 -1.1606724 0.6210283 0.28483844 -0.4358692
0.26937243 0.16278586 -0.675542 0.24764606 0.06118002 0.33632183
0.3737057 -1.1899931 -0.84018236 -0.2926057 -0.11547776 0.18417554
-0.11454975 0.31233138 -0.24461313 -0.77331114 0.626895 -0.5431323
-0.45363173 -0.14400604 0.18085125 0.18796413 0.88298553 -1.3210306
1.086192 -2.1413074 0.08552485 -0.02555958 -0.05235198 0.68895656
-0.45879212 -0.29500565 -0.48494247 0.42326593 -0.20410657 0.40362385
-0.7314742 -0.03187161 -0.04140416 0.4723127 0.04345669 0.42357978
-0.5007776 -0.45241466 0.23209949 0.9100284 -0.0484791 0.32627282
0.41047174 0.4086786 -0.3299051 0.49422333 1.3710256 -0.01083313
-0.15132129 -0.6106787 -0.3832158 -0.23152567 -1.5206323 0.9904619
-0.16927008 0.96062994 0.53998065 -0.9574856 1.3139585 -0.06641332
0.51217955 -1.1457239 0.36865577 0.06724063 -0.28853813 -0.72034585
0.62526965 -0.18732913 0.61962414 -0.06402596 -0.38023448 0.35710934
-0.0527725 -0.1712236 0.43577832 0.06611296 -0.06858409 -0.40438795
0.86061853 -0.16146877 0.8106304 0.30398524 -0.22576201 0.7304933
0.13592741 -0.7349871 -1.0226593 -0.6132418 -0.28603816 0.8898228
0.47903076 0.4016855 -0.653037 -0.07485791 -0.31489447 0.13011667
-0.4019536 -0.02901545 -0.5930983 -0.876863 0.6416621 0.6039351
1.2964264 -1.1891574 -0.31921113 0.13916546 -0.4209286 -0.87037605
-0.18723257 -0.09119022 -1.2986475 -0.9141099 -0.9050632 -1.082172
0.3603586 0.7566836 0.7690939 0.4204639 -0.11711736 -0.44387218
-0.2477998 0.448894 -0.06815472 -0.28094628 -0.2110389 -0.2051059
-0.15787096 -0.6153745 -0.7912424 0.35581964 -1.1963445 -0.12567627
0.6598955 0.8280367 0.6211651 0.83216673 0.07079592 -0.14941892
0.68742263 -0.2810188 -0.69263315 -0.09189741 -0.66302 -0.15839206
0.6936488 -0.3147548 -1.2295595 -0.30161163 -0.12840222 -0.20873712
0.21743895 0.83228683 -0.07280532 -0.4185193 0.4109648 0.43209848
0.10157181 -0.69952124 -0.07463757 0.94412696 0.9021597 0.19767563
0.5237073 0.21136355 -0.05141283 -0.9523356 -0.4327908 -0.09482866
-0.38490275 -0.2147857 0.7769125 -1.3108453 -0.6182921 0.6795297
-0.92789805 -0.11497569 0.11675432 0.6383263 -0.01633407 0.7136013
-0.9568945 -0.6541271 -0.82064265 -1.2311852 0.56780225 0.87909704
0.78843915 -0.7932893 -0.10938783 0.1104837 0.8499336 0.3000207
0.28609926 -0.19461963 -0.65608495 -0.08711189 -0.93146247 1.0586505
-0.0749572 0.11247156 -0.6453409 -0.33396742 0.31691444 0.0209573
1.1976255 0.66047525 1.2491518 -0.39950657 0.13996273 1.2082818
1.9926821 0.23953533 1.3837818 0.9338619 0.4375999 0.44478425
0.4139239 0.29829207 0.05057864 0.42009357 0.8080895 -0.49804974
-0.27475837 0.34906724 0.4786783 0.8039103 -0.6052536 0.00804767
-0.49725702 0.59548753 -1.3794563 -1.2716863 -0.6322642 1.7076106
0.4766883 -0.15662743 -0.41025096 0.19719324 1.0511179 0.44646442
-0.400365 -0.16406097 -0.66108584 0.17341146 0.7449384 0.3966127
-1.2872326 0.8195342 7.003879 0.7440958 -1.4923445 -0.16324203
0.56942475 0.2861521 -0.03458267 -0.2784937 -0.49085078 0.54099417
0.90821606 -0.0310719 0.49860254 0.8371751 0.63566506 -0.01881171
0.04179163 0.6961066 -0.11748227 -1.1457812 0.16683444 -0.09734581
0.6324558 0.33019823 0.08183019 0.1826924 0.1996983 -1.170402
0.20660654 0.74013597 0.78619736 -0.8994039 1.4668368 0.10489284
-1.548554 -0.22564416 -0.8319644 -0.15880603 -0.18207352 0.46354365
-0.03245495 0.8420629 1.0907805 0.8045991 -0.66966945 1.1218963
0.01169072 0.37814784 -0.07153362 0.46075603 0.2707586 -0.55695885
0.38515157 1.2549012 0.5988302 0.39192358 0.07867072 0.620868
0.17086974 0.1564112 -0.1268477 0.20057279 0.23262894 1.362717
-0.6687301 -0.24508381 -0.3131379 0.9044225 -0.23309036 0.3703816
-0.86896193 -0.73871905 0.47709766 -0.09676393 0.44537416 -0.09745442
-0.41846535 -1.0933198 -0.16948287 -0.85802263 0.36352023 -1.3322307
-0.914888 1.0114282 -0.27875894 -1.3901527 0.27479443 -0.5411172
-0.532554 1.2215159 -1.7760909 -0.98775303 -0.91005576 0.41540813
0.36196485 -0.2159393 0.3807037 0.5017449 -0.8225954 0.203506
0.43011764 0.29950526 0.30146345 -0.6444062 0.1208628 1.2941768
-0.6631583 0.30205545 0.70775765 -0.47639394 -0.7413217 -1.1743568
0.42546675 0.7449427 0.43362775 0.7693469 -1.4245279 0.56773615
0.3223464 0.13556173 0.14374287 -0.61611307 -0.01064166 -0.46298912
-1.425552 0.27664077 0.6128725 -0.13052036 -0.56304854 -0.25304443
1.0008584 -0.23400071 -1.0809509 0.83253676 0.50566727 -1.2303152
1.1151335 -0.16981384 0.6429264 -0.77601975 -0.25070062 -1.2327505
-1.0340716 0.27565846 0.19521761 1.0858753 0.03183221 -0.37710634
0.4964015 0.2030189 -0.2643693 -0.41988775 -0.67102724 -0.7431987
-0.09805705 0.06648195 0.63788104 0.856957 -0.7272613 0.06665562
-0.59081095 0.5264745 0.6298958 -0.25116742 0.28340796 -1.3070785
0.35935426 -0.56385773 -0.49679968 -0.9898551 -0.11889941 -0.65328175
-0.02507143 -1.9717456 0.27686727 0.13535242 -0.51991415 0.24395974
-0.3176478 0.68943626 0.02461844 0.46345666 -0.07340495 0.638532
1.4771487 -0.2578882 -0.3485727 -0.02988704 -0.7831626 0.6276547
0.8121742 -0.08189992 0.06016371 -0.6860427 -0.08184346 0.1284634
0.18739139 -1.2292986 0.26892406 -0.19521032 0.6878551 -0.84993654
-0.07830819 -0.9609911 0.45208693 0.62635416 -0.32424182 -0.26646575
0.35852963 0.18970731 -0.55233127 -0.19488795 1.1782912 -0.28042847
-1.170845 0.24541211 -0.3024232 -0.53578645 0.77775866 -0.52322006
-0.4206563 -0.33035743 -0.63339144 0.13155462 0.19710976 0.1427876
1.0089079 -1.3895777 -1.1160115 0.256671 -0.5399794 0.0420302
0.8002107 0.71968025 -1.1285422 -0.05349025 -0.7716799 -0.4180419
-1.3623512 0.334606 1.0161189 -0.3227937 -0.67450696 0.6513264
-0.06401714 -0.32862568 0.21900378 -0.06818641 -0.8900674 0.08869028
0.980226 0.8343196 -0.12655176 -1.0574008 -0.19061838 0.75302225
0.13659541 0.17168449 1.8983662 -0.5221133 -0.6141039 -0.13009843
1.3125261 -0.31830296 -1.304626 -0.3850008 -0.36028117 -0.5751821
0.82927984 -0.61231583 -1.875515 0.7165839 1.2621833 0.3478892
1.7276452 -0.7240852 0.932357 0.091421 -0.02481922 -0.92631924
-0.07653293 0.69088906 0.9413049 -1.176377 0.37680304 0.04349159
-0.62271464 1.4437494 0.6237103 -0.4554542 0.5155812 0.26794085
0.06206724 0.16367052 -0.13280623 -0.592048 0.2263409 0.55850804
0.55041087 -0.1936932 -0.5484807 0.7570493 0.06756517 0.31748456
0.62734425 0.46210682 -1.007273 -0.33958647 -0.65769833 0.16285944
-0.81732714 -0.23972978 0.06551405 0.58698183 0.2876127 1.0352871
0.1512865 -0.885866 0.5706483 -0.6757595 -0.01766493 -0.02329264
-0.4481815 0.10525676 -0.27878967 -0.57216775 -0.76463914 0.02825495
-1.2326499 -0.673228 -0.508486 -0.01324049 0.60061073 0.6668412
0.13571618 0.9018022 1.0509318 -1.0615256 -0.35546812 -1.2940115
-0.9733238 0.08040583 0.46435192 -0.27016866 -0.6711874 0.0769964 ] | [10.318951606750488, -1.9022626876831055] |
27d004eb-67dc-40d3-a3bf-f19914671ccd | milliflow-scene-flow-estimation-on-mmwave | 2306.17010 | null | https://arxiv.org/abs/2306.17010v2 | https://arxiv.org/pdf/2306.17010v2.pdf | milliFlow: Scene Flow Estimation on mmWave Radar Point Cloud for Human Motion Sensing | Approaching the era of ubiquitous computing, human motion sensing plays a crucial role in smart systems for decision making, user interaction, and personalized services. Extensive research has been conducted on human tracking, pose estimation, gesture recognition, and activity recognition, which are predominantly based on cameras in traditional methods. However, the intrusive nature of cameras limits their use in smart home applications. To address this, mmWave radars have gained popularity due to their privacy-friendly features. In this work, we propose \textit{milliFlow}, a novel deep learning method for scene flow estimation as a complementary motion information for mmWave point cloud, serving as an intermediate level of features and directly benefiting downstream human motion sensing tasks. Experimental results demonstrate the superior performance of our method with an average 3D endpoint error of 4.6cm, significantly surpassing the competing approaches. Furthermore, by incorporating scene flow information, we achieve remarkable improvements in human activity recognition, human parsing, and human body part tracking. To foster further research in this area, we provide our codebase and dataset for open access. | ['Chris Xiaoxuan Lu', 'Peijun Zhao', 'Zhen Luo', 'Fangqiang Ding'] | 2023-06-29 | null | null | null | null | ['pose-estimation', 'activity-recognition', 'gesture-recognition', 'human-parsing', 'human-activity-recognition', 'scene-flow-estimation', 'decision-making', 'human-activity-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'reasoning', 'time-series'] | [ 2.31655076e-01 -2.79193282e-01 -3.44692260e-01 -2.50724852e-01
-5.48799455e-01 -4.29480702e-01 4.27310228e-01 -3.18734527e-01
-4.06223774e-01 4.88481164e-01 3.52204829e-01 -1.74224019e-01
1.54820755e-01 -6.28610194e-01 -1.62411630e-01 -9.14189816e-01
6.12251870e-02 -1.28232604e-02 2.84310311e-01 1.97342962e-01
-1.40239149e-01 5.22996008e-01 -1.29101276e+00 -6.23741075e-02
5.67221701e-01 1.24557316e+00 -1.03597030e-01 5.32974780e-01
2.49578968e-01 6.47458017e-01 -1.99014455e-01 -5.70776820e-01
3.23936701e-01 -3.72502655e-01 -1.92146540e-01 4.84050363e-02
4.59679544e-01 -7.85076499e-01 -8.31329048e-01 8.03468287e-01
8.44412565e-01 9.29709226e-02 1.46893606e-01 -1.21563554e+00
-7.23713860e-02 9.98524651e-02 -6.65440321e-01 1.84878513e-01
5.93709409e-01 3.49371523e-01 6.79390252e-01 -6.60142243e-01
3.68678063e-01 8.67215455e-01 5.52804887e-01 7.39105165e-01
-4.85489249e-01 -9.43777382e-01 6.01221025e-02 3.06527555e-01
-1.34309220e+00 -6.93665087e-01 7.59693563e-01 -3.44697565e-01
6.35141730e-01 3.71010691e-01 8.39778185e-01 1.27698970e+00
-1.16117500e-01 1.06941724e+00 5.35197914e-01 -4.85217795e-02
2.30971977e-01 -3.07828337e-01 -1.53803065e-01 8.95262182e-01
4.50980216e-01 9.91504341e-02 -6.82485104e-01 5.94929494e-02
9.58907664e-01 4.28416669e-01 -4.48073477e-01 -4.91154671e-01
-1.37169552e+00 3.90031457e-01 4.49016154e-01 2.68578410e-01
-4.10552531e-01 5.37045360e-01 9.13815945e-02 -2.32547775e-01
3.68797451e-01 -3.61028850e-01 -1.63863853e-01 -4.60869789e-01
-8.39590192e-01 6.87819272e-02 4.56865281e-01 1.01561368e+00
2.17172861e-01 4.36341353e-02 -1.90515339e-01 4.21258122e-01
6.66384757e-01 1.02144897e+00 2.38125741e-01 -9.77918863e-01
8.35053444e-01 5.94685972e-01 2.08331734e-01 -9.40819502e-01
-6.50782228e-01 -5.59033632e-01 -1.06640601e+00 -1.01333946e-01
4.24144924e-01 -2.93042600e-01 -6.69309318e-01 1.45675981e+00
5.37466526e-01 3.50261122e-01 -2.18451351e-01 1.33879733e+00
7.67922938e-01 3.71711999e-01 1.98499545e-01 -2.05017239e-01
1.35075629e+00 -7.40978181e-01 -5.93361914e-01 -3.66951287e-01
5.04771650e-01 -4.26237017e-01 6.27928853e-01 4.21562493e-01
-6.33426905e-01 -2.44547054e-01 -1.00457096e+00 8.00517760e-03
9.84591171e-02 1.53130308e-01 8.67197275e-01 1.05548286e+00
-5.09916604e-01 3.29830527e-01 -1.31903768e+00 -5.36337435e-01
7.57234097e-01 3.85126293e-01 -1.92683160e-01 -2.16820970e-01
-8.54115665e-01 2.93344408e-01 -1.10290915e-01 3.48852962e-01
-2.60745883e-01 -4.61273938e-01 -8.72210741e-01 -6.90019429e-02
2.00659186e-01 -9.48843479e-01 1.21449912e+00 -3.18465143e-01
-1.44807065e+00 7.13564098e-01 -1.42762735e-01 -3.28150421e-01
8.41082513e-01 -5.95642567e-01 -5.37320137e-01 2.78227061e-01
-3.09079625e-02 3.15496176e-01 7.31868684e-01 -6.64839089e-01
-7.89936185e-01 -8.19207013e-01 -6.60014227e-02 2.37329111e-01
-3.75199616e-01 2.35754307e-02 -7.16408730e-01 -6.00698292e-01
4.82154757e-01 -9.82641459e-01 -3.20198685e-01 3.42336863e-01
-2.82916635e-01 1.30380645e-01 6.62970901e-01 -6.11328125e-01
1.28491294e+00 -2.09052706e+00 -9.76600125e-02 5.45022637e-02
3.90793115e-01 3.39854717e-01 3.55579555e-01 4.30639796e-02
6.65095687e-01 -2.20139891e-01 -4.11011249e-01 -3.58536154e-01
9.16050151e-02 -2.10757151e-01 -4.34383713e-02 7.64465749e-01
-3.30242842e-01 1.06904328e+00 -8.60908747e-01 -5.24621248e-01
5.57430208e-01 5.89195549e-01 -2.97662139e-01 -6.69006854e-02
2.92533726e-01 9.31892753e-01 -8.18687916e-01 1.21591890e+00
5.95074117e-01 -2.04822421e-01 -2.65452336e-03 -2.40097567e-01
-3.40639018e-02 -5.38693592e-02 -1.18563569e+00 1.90917408e+00
-3.38470727e-01 6.13562822e-01 1.43916488e-01 -6.81908369e-01
6.84876502e-01 3.68721813e-01 1.04983926e+00 -8.17094564e-01
1.69945553e-01 6.27149194e-02 -3.41701448e-01 -7.55053639e-01
2.14240134e-01 -5.42396978e-02 -2.05468506e-01 2.60338306e-01
-4.02718037e-01 4.77622926e-01 -2.30964661e-01 -1.57100007e-01
1.38977194e+00 3.81522268e-01 4.11300004e-01 2.73395926e-01
4.98275608e-01 -1.97137713e-01 7.49066710e-01 5.51657856e-01
-7.86269665e-01 6.24456346e-01 -2.63174772e-01 -4.54259336e-01
-4.54055995e-01 -1.21208644e+00 -5.02546765e-02 7.71726191e-01
4.84097362e-01 -3.53995323e-01 -4.95198458e-01 -6.55632973e-01
-6.29158989e-02 8.08158815e-02 -2.46051010e-02 8.34074914e-02
-9.87118602e-01 -8.77321661e-01 6.50809288e-01 9.46840584e-01
1.05771291e+00 -8.02005708e-01 -1.23564816e+00 2.01137766e-01
-5.35522521e-01 -1.61014438e+00 -3.13877702e-01 -4.32911545e-01
-1.09121871e+00 -9.46906745e-01 -9.92869377e-01 -4.27291930e-01
3.08342457e-01 5.97606897e-01 6.14111304e-01 -3.00426520e-02
-3.98299068e-01 6.30810499e-01 -2.85031646e-01 -2.76798129e-01
4.48035955e-01 8.75325650e-02 2.27069810e-01 3.19265932e-01
5.80997586e-01 -7.61820853e-01 -1.20853555e+00 3.10817808e-01
-4.86583740e-01 2.57811276e-03 7.26924956e-01 1.84513405e-01
2.10881293e-01 -3.88763398e-01 1.78267911e-01 -4.10316616e-01
-6.26120120e-02 -3.67309630e-01 -4.97814059e-01 -1.05637126e-02
-3.37081492e-01 -3.14719766e-01 2.01464146e-01 -2.31774300e-02
-1.12047696e+00 5.08127213e-01 -1.23829342e-01 -3.44030321e-01
-3.58300269e-01 1.01615474e-01 -5.13399839e-01 -1.04415879e-01
2.49532774e-01 1.11786991e-01 -2.20993772e-01 -4.81708437e-01
3.22718352e-01 7.34408677e-01 7.71791220e-01 -4.50862013e-02
9.41983163e-01 9.35303748e-01 9.77163389e-02 -1.20226359e+00
-6.12745941e-01 -9.24013138e-01 -7.55762458e-01 -5.33534288e-01
1.14336348e+00 -1.19865203e+00 -9.71388996e-01 4.35231924e-01
-1.04386973e+00 7.42827402e-03 2.03920081e-01 9.23635006e-01
-4.25362498e-01 5.35708189e-01 -4.26629335e-01 -1.19246364e+00
-3.65214884e-01 -6.98830843e-01 1.33836293e+00 3.90458941e-01
-1.69905826e-01 -6.60008311e-01 -4.36273031e-02 8.85395169e-01
3.01207364e-01 5.71047127e-01 2.02944919e-01 3.95911839e-03
-1.18248355e+00 -4.30278242e-01 -1.41669288e-01 -2.54378170e-01
4.50074300e-03 -6.91569448e-01 -1.11124325e+00 -1.33411080e-01
-1.77350551e-01 1.43745512e-01 8.00775528e-01 5.42837977e-01
8.69705617e-01 5.66940084e-02 -7.91328132e-01 1.06179976e+00
1.15023088e+00 2.99172670e-01 6.46288574e-01 3.22271436e-01
1.05331671e+00 4.96683568e-01 5.89109838e-01 8.50155652e-01
3.51771295e-01 8.33369493e-01 5.04908860e-01 -2.58156639e-02
-1.24557860e-01 -2.10133910e-01 4.33659971e-01 5.62095582e-01
-4.78668302e-01 -1.75663903e-01 -9.25047874e-01 2.21498430e-01
-2.13468480e+00 -1.15261292e+00 -3.38558733e-01 2.15022755e+00
-1.53964134e-02 -2.18340605e-02 1.49719611e-01 7.55578652e-02
5.94168246e-01 3.67480338e-01 -6.63113534e-01 4.95941430e-01
1.49249017e-01 -6.92208530e-03 6.60825789e-01 8.88816118e-02
-1.54219854e+00 8.19372714e-01 5.11491728e+00 2.59036034e-01
-9.16807294e-01 1.66570023e-01 8.39034319e-02 -4.71272260e-01
8.56135711e-02 -3.65458965e-01 -8.90834332e-01 6.05086446e-01
4.43148524e-01 3.30780715e-01 1.12173989e-01 7.56049156e-01
5.13307095e-01 -2.36985981e-01 -8.63870621e-01 1.61124957e+00
6.57352135e-02 -1.03233171e+00 -4.58820760e-01 3.48963737e-01
3.14837813e-01 -6.58765733e-02 -6.39900118e-02 1.59619495e-01
-8.18779543e-02 -7.24265814e-01 5.80921829e-01 5.19248247e-01
6.85747027e-01 -6.56780183e-01 6.08871043e-01 5.18434644e-01
-1.58232009e+00 -1.20550156e-01 -1.00223638e-01 -2.36967102e-01
5.99553168e-01 6.20832562e-01 -3.58664334e-01 5.44819236e-01
8.40657651e-01 7.67331898e-01 -2.97214240e-01 1.21976137e+00
-1.54451743e-01 6.13076866e-01 -3.99870008e-01 -1.04233153e-01
-1.21130191e-01 -2.70558566e-01 6.78701162e-01 1.13143468e+00
4.82456774e-01 5.09274840e-01 2.60077924e-01 5.19656301e-01
-4.69365455e-02 -1.59814790e-01 -7.69044876e-01 1.01115160e-01
3.91003728e-01 1.35842383e+00 -8.82226884e-01 5.45680262e-02
-5.76057553e-01 1.27255154e+00 -3.48108917e-01 2.04972982e-01
-1.10424089e+00 -4.07817960e-01 9.91119921e-01 2.40101665e-01
3.10953408e-01 -7.24261045e-01 -3.34575027e-01 -1.42895603e+00
4.03744072e-01 -3.22056353e-01 3.14235717e-01 -4.15027648e-01
-7.94701159e-01 2.01431975e-01 -2.64340699e-01 -1.75196981e+00
-2.87531167e-01 -5.56618333e-01 -4.17700171e-01 3.56268764e-01
-1.19048643e+00 -1.33683968e+00 -8.01915526e-01 6.34258151e-01
4.10172135e-01 -7.77105242e-02 4.08077776e-01 7.16654360e-01
-7.70821631e-01 6.20781243e-01 -9.57478583e-02 4.95970309e-01
4.85151619e-01 -6.71533585e-01 5.00623584e-01 1.20581281e+00
4.02118325e-01 2.16420233e-01 3.57494503e-01 -7.16458976e-01
-1.76423585e+00 -1.12843394e+00 7.56017208e-01 -7.06723809e-01
3.18209141e-01 -3.36822063e-01 -2.76560247e-01 6.40279233e-01
-3.23930830e-01 1.62508413e-01 8.28727543e-01 -2.73114532e-01
1.19133256e-01 -2.81809449e-01 -9.20858264e-01 5.71257591e-01
1.79404294e+00 -2.53338873e-01 -2.31161833e-01 1.30273789e-01
3.52501512e-01 -4.13452417e-01 -6.11203313e-01 6.04987025e-01
9.57971096e-01 -9.87477303e-01 1.20027208e+00 -1.13552511e-01
1.73736475e-02 -4.21190739e-01 -2.55114436e-01 -4.47280675e-01
-3.89907628e-01 -7.03545868e-01 -7.07756221e-01 1.05322564e+00
-1.28478959e-01 -3.70483577e-01 1.44195747e+00 6.74486697e-01
5.94913103e-02 -5.14678895e-01 -1.06682062e+00 -8.03897023e-01
-5.53349137e-01 -7.83830881e-01 5.18368483e-01 6.92598104e-01
-1.41407773e-01 2.23239258e-01 -5.43032706e-01 4.71045434e-01
9.70309138e-01 2.77672976e-01 9.16046262e-01 -1.30002475e+00
-2.89362699e-01 -2.50170171e-01 -7.31462598e-01 -1.54650784e+00
-2.00805709e-01 -7.53965974e-01 1.27214612e-02 -1.69356859e+00
-5.51659539e-02 -1.10594213e-01 -3.30909565e-02 2.31946155e-01
4.10665013e-02 5.87363541e-01 2.25474015e-01 2.87597299e-01
-8.10971439e-01 5.36053538e-01 1.00762510e+00 -1.58374429e-01
-1.84904799e-01 3.88020515e-01 -5.55310011e-01 9.71518278e-01
7.63276875e-01 -1.74068257e-01 -2.17440650e-01 -6.09238505e-01
-1.00395225e-01 2.22592473e-01 6.84450269e-01 -1.61562300e+00
4.97520894e-01 -1.62383065e-01 7.28611231e-01 -6.73847377e-01
5.27459383e-01 -9.19770002e-01 3.33572030e-02 7.23328650e-01
1.95793837e-01 -9.71634984e-02 -1.93757191e-01 7.81884253e-01
8.71450230e-02 2.27681428e-01 3.91291618e-01 -5.67066036e-02
-1.07673395e+00 7.52773225e-01 -3.92396241e-01 -1.05468079e-01
1.03658867e+00 -4.85107839e-01 4.13050763e-02 -5.02667964e-01
-5.68812490e-01 2.06513524e-01 2.41749093e-01 5.08104503e-01
7.00402141e-01 -1.58910108e+00 -4.60700780e-01 1.65682092e-01
1.95785448e-01 -9.04987901e-02 2.25359231e-01 1.09198987e+00
-4.93224800e-01 6.85956359e-01 -5.18614203e-02 -8.33379567e-01
-1.20092416e+00 9.46663022e-02 1.10469870e-01 9.86891687e-02
-1.00794816e+00 6.10359907e-01 9.32564735e-02 2.78437301e-03
4.04453397e-01 -2.75436521e-01 -5.41742966e-02 -1.35508448e-01
6.35276496e-01 7.57215261e-01 2.48297136e-02 -6.32903755e-01
-7.84217358e-01 8.83388519e-01 4.27410156e-01 -2.91698635e-01
9.75038111e-01 -4.35583264e-01 4.81776804e-01 1.21583119e-01
9.06673789e-01 1.54306054e-01 -1.48271477e+00 -6.63366392e-02
3.31990719e-02 -5.57261288e-01 8.51459876e-02 -5.10108769e-01
-1.23089659e+00 9.91814256e-01 9.53566074e-01 -1.94116622e-01
1.20116282e+00 6.62629455e-02 1.25612366e+00 4.82832193e-01
8.82157862e-01 -7.82279134e-01 -1.23463057e-01 2.16792151e-01
2.86911547e-01 -1.45315921e+00 7.62769952e-02 -5.36827743e-01
-3.92872840e-01 9.54193115e-01 5.58970153e-01 3.41337502e-01
5.64666271e-01 1.40174612e-01 1.89669132e-01 -7.96647221e-02
-3.04754823e-02 -5.40359139e-01 2.14886591e-01 8.60196054e-01
4.02297586e-01 4.78051491e-02 -1.40052542e-01 6.74882233e-01
-1.22971557e-01 1.51354373e-01 -7.35575557e-02 1.12324214e+00
-4.93104964e-01 -7.48574436e-01 -3.88251334e-01 2.95529157e-01
-3.53703499e-01 2.14886770e-01 -2.56679028e-01 7.77700305e-01
1.01752870e-01 9.57193851e-01 -8.27085897e-02 -5.14292359e-01
3.86577964e-01 -1.30068272e-01 6.91724956e-01 -1.77491874e-01
-1.89537153e-01 -1.42104421e-02 -9.68039595e-03 -9.99825001e-01
-5.12904048e-01 -8.25661004e-01 -1.26937830e+00 -2.13091537e-01
9.59466323e-02 -3.90965521e-01 5.68154514e-01 1.04347944e+00
4.42465723e-01 1.98119253e-01 5.27006567e-01 -8.42474401e-01
-3.17869931e-01 -5.46774387e-01 -4.58809525e-01 2.13378176e-01
3.25140089e-01 -6.78642035e-01 -4.60489616e-02 1.05861127e-01] | [6.943568229675293, 0.3348081409931183] |
e0274b61-6348-437e-ad70-9dad37192af6 | neural-comprehension-language-models-with | 2304.01665 | null | https://arxiv.org/abs/2304.01665v2 | https://arxiv.org/pdf/2304.01665v2.pdf | Mastering Symbolic Operations: Augmenting Language Models with Compiled Neural Networks | Language models (LMs) proficiency in handling deterministic symbolic reasoning and rule-based tasks remains limited due to their dependency implicit learning on textual data. To enable fully rule comprehension ability, we explore how to incorporate compiled neural networks (CoNNs) which weight is specially designed into the architecture of LMs, to achieve high accuracy and robust performance. CoNNs are transformer-based neural networks that execute rules through artificially generated attention weights. Our method, which call "Neural Comprehension", by incorporating CoNN modules into the LM, the framework effectively tackles rule-intensive challenges. Our experiments on symbolic reasoning tasks and real-world arithmetic reasoning tasks demonstrate the superior performance of our method compared to existing techniques. Furthermore, our LM achieves flawless execution on symbolic operations tasks, highlighting the potential of our method in enabling LMs to possess true symbolic comprehension capabilities. Our code is publicly available at: https://github.com/WENGSYX/Neural-Comprehension. | ['Jun Zhao', 'Kang Liu', 'Shizhu He', 'Bin Li', 'Fei Xia', 'Minjun Zhu', 'Yixuan Weng'] | 2023-04-04 | null | null | null | null | ['arithmetic-reasoning'] | ['reasoning'] | [ 2.65590191e-01 4.07580376e-01 -1.58091322e-01 -2.77304441e-01
-4.10796791e-01 -4.47359324e-01 4.81762290e-01 -2.41821453e-01
-9.02496800e-02 4.10520405e-01 2.35889535e-02 -9.93274391e-01
-2.54331846e-02 -1.19394076e+00 -9.92495596e-01 8.86372104e-02
-3.84512842e-02 1.48551211e-01 2.22638011e-01 -5.26601791e-01
2.62593716e-01 2.88186282e-01 -1.48986053e+00 5.34651756e-01
1.33727658e+00 8.67667973e-01 1.75725460e-01 7.52777755e-01
-3.14766198e-01 1.79741764e+00 -3.77534389e-01 -4.38625455e-01
7.31272474e-02 -1.35247767e-01 -9.85570133e-01 -8.50361526e-01
3.05281132e-01 -5.47616124e-01 -2.48026207e-01 9.94458199e-01
-6.67646900e-02 2.17647955e-01 2.99395144e-01 -1.24983573e+00
-8.33101690e-01 1.31182015e+00 2.22410751e-03 1.91505216e-02
4.76450890e-01 5.53028882e-01 1.08109355e+00 -6.79527581e-01
2.84239769e-01 1.21827233e+00 7.52762556e-01 8.72109592e-01
-1.25470614e+00 -8.84894669e-01 4.28639412e-01 3.82701397e-01
-1.22607434e+00 -7.56701648e-01 6.25718117e-01 -2.48981342e-01
1.59896755e+00 2.00474873e-01 6.35083377e-01 1.05179775e+00
2.31097385e-01 1.02880132e+00 7.06600368e-01 -5.17826676e-01
1.07733943e-01 -2.88741440e-01 4.00228620e-01 1.11052632e+00
7.73453489e-02 2.79213160e-01 -6.70180559e-01 1.27806440e-01
7.73639500e-01 -1.84824750e-01 -1.91889182e-01 -8.42489973e-02
-1.11083746e+00 5.70292056e-01 4.36533779e-01 2.32382372e-01
-2.56797541e-02 6.62839949e-01 5.71019709e-01 4.60163802e-01
-1.50054723e-01 9.37816799e-01 -6.73427582e-01 -3.27967852e-01
-7.06907928e-01 2.51936644e-01 9.74634409e-01 1.14466178e+00
2.35379204e-01 5.53263903e-01 -6.27725571e-02 5.36556721e-01
1.51793614e-01 4.88219351e-01 4.31572348e-01 -1.30405188e+00
3.85556698e-01 9.81218040e-01 -3.63451689e-01 -8.26943099e-01
-4.57103729e-01 -4.62086886e-01 -4.49045330e-01 2.74205059e-01
4.32077378e-01 -8.82303491e-02 -6.45988464e-01 2.12869096e+00
-1.04029574e-01 3.87218624e-01 1.65265396e-01 5.46337366e-01
8.25524509e-01 6.06650770e-01 2.85878509e-01 2.50568956e-01
1.06202984e+00 -8.82169068e-01 -4.01073605e-01 -2.05457196e-01
9.73618567e-01 -3.82395349e-02 1.55154896e+00 6.30527496e-01
-1.46142924e+00 -6.33639634e-01 -1.19590950e+00 -3.12024087e-01
-4.75168914e-01 -5.81783354e-02 1.12572896e+00 2.58360893e-01
-1.06005776e+00 5.75917184e-01 -1.06458664e+00 1.55994907e-01
5.75791121e-01 5.79704463e-01 4.65008756e-03 1.56070486e-01
-1.36959898e+00 9.30997550e-01 7.55094469e-01 1.93166435e-01
-9.78101075e-01 -9.97066975e-01 -1.25468600e+00 5.27178943e-01
8.06657076e-01 -7.79245079e-01 1.84358406e+00 -9.64219749e-01
-1.75790679e+00 4.74436760e-01 -1.43480241e-01 -5.76442957e-01
1.20433949e-01 -3.39076281e-01 -3.20339441e-01 9.48770121e-02
-2.27538228e-01 6.86740339e-01 4.25550491e-01 -9.62578118e-01
-3.35138887e-01 -3.45983426e-03 5.48478544e-01 -1.44513816e-01
-1.82878673e-01 -2.75886506e-02 -7.41231069e-02 -5.75814903e-01
-1.32693335e-01 -6.04510367e-01 1.50790393e-01 -1.63369969e-01
-3.49432558e-01 -4.49115664e-01 3.74233276e-01 -6.69902623e-01
1.39069402e+00 -1.89971769e+00 1.03350364e-01 2.54607320e-01
4.38008636e-01 5.26975214e-01 -2.25826457e-01 1.60727128e-01
-4.02857251e-02 2.35506102e-01 -2.50609487e-01 2.55400062e-01
3.85336488e-01 1.88096449e-01 -7.16836274e-01 -3.36881489e-01
5.64405918e-01 1.48861456e+00 -1.03885543e+00 -3.70618016e-01
5.22471033e-02 4.98549966e-03 -8.43064845e-01 2.48619959e-01
-9.80461895e-01 1.32786259e-01 -4.36566323e-01 7.70528972e-01
3.24409872e-01 -3.73930365e-01 4.56980705e-01 2.13899374e-01
4.21746410e-02 5.87779760e-01 -6.76204860e-01 1.80741918e+00
-6.85394287e-01 5.23544967e-01 -1.89860120e-01 -8.63720596e-01
5.87572873e-01 1.88810393e-01 -3.17796797e-01 -8.59281540e-01
2.40265623e-01 1.75261974e-01 4.80124623e-01 -4.75856096e-01
3.07497859e-01 9.32574868e-02 -1.24429248e-01 3.86150360e-01
8.23077559e-03 -1.63113803e-01 3.84911269e-01 4.67267722e-01
1.26241028e+00 7.60472476e-01 2.93537259e-01 -2.95905676e-02
6.77004278e-01 2.17271388e-01 5.13499379e-01 9.78545845e-01
1.65527642e-01 -1.75117850e-01 8.70254695e-01 -4.34663773e-01
-7.80806243e-01 -1.01712668e+00 2.02798441e-01 1.51205409e+00
-4.30795044e-01 -5.02002478e-01 -6.84164345e-01 -3.60017389e-01
2.14043017e-02 1.27654159e+00 -5.08233488e-01 -5.60236156e-01
-9.28595364e-01 -1.03282921e-01 1.23016167e+00 9.59529579e-01
7.17225254e-01 -1.42370129e+00 -7.67241776e-01 1.40570715e-01
-3.10199652e-02 -9.96400416e-01 -8.64432454e-02 1.72521874e-01
-8.00694287e-01 -1.09618735e+00 8.62118602e-02 -6.80006981e-01
5.17592430e-01 -2.62937158e-01 1.28269506e+00 5.38132846e-01
1.34455913e-03 1.02114461e-01 -1.92235172e-01 -3.68927509e-01
-5.75581014e-01 3.64120454e-01 -7.84940720e-02 -7.43903935e-01
4.23697740e-01 -8.23351622e-01 -4.81028147e-02 -6.73135668e-02
-7.03639925e-01 5.98083675e-01 4.44620937e-01 7.58161068e-01
7.96932504e-02 -1.94209829e-01 6.84160888e-01 -1.05053365e+00
8.66786778e-01 -3.87177825e-01 -8.83848011e-01 4.41219449e-01
-6.52094305e-01 1.80515334e-01 1.06717873e+00 -2.65017182e-01
-1.05705023e+00 -2.83998907e-01 -1.41581699e-01 -2.37254664e-01
-1.86167374e-01 8.27477753e-01 7.85700139e-03 -2.24652793e-02
7.39585400e-01 4.55261469e-01 2.14397535e-01 -5.18923216e-02
4.43107456e-01 1.86846808e-01 8.51571560e-01 -1.45978320e+00
6.97780669e-01 -2.00913623e-01 -1.38963953e-01 -1.33481100e-01
-9.15542305e-01 3.48485917e-01 -2.84303606e-01 5.51832132e-02
5.02733052e-01 -7.83802032e-01 -1.49826527e+00 4.24965709e-01
-1.14895439e+00 -1.20900667e+00 -1.66406438e-01 3.57460417e-03
-7.88651943e-01 -3.45166842e-03 -9.31960523e-01 -7.51954317e-01
-4.96393114e-01 -1.12151337e+00 5.97187221e-01 7.03750029e-02
-6.23893321e-01 -1.06366169e+00 -2.65155196e-01 3.03994864e-01
6.79403067e-01 1.44715801e-01 1.56292069e+00 -7.95232475e-01
-6.38639987e-01 9.23181176e-02 -3.18745136e-01 3.87940645e-01
-4.08371329e-01 5.64077757e-02 -8.41742516e-01 2.08090857e-01
-3.00806671e-01 -7.86575317e-01 6.61663949e-01 5.20245247e-02
1.72331059e+00 -4.85207886e-01 -8.89580250e-02 8.51415396e-01
1.08246005e+00 2.48186171e-01 6.26800776e-01 3.06739002e-01
5.89209139e-01 2.46876001e-01 1.39893249e-01 1.46738127e-01
6.87833846e-01 3.27736527e-01 3.25134933e-01 2.41868064e-01
-1.42764732e-01 -5.41889489e-01 5.85448086e-01 8.66531909e-01
-1.84499145e-01 -2.33251452e-02 -1.56777477e+00 1.44500479e-01
-1.87448514e+00 -9.78139937e-01 3.82397808e-02 1.68016803e+00
1.30943215e+00 3.63740414e-01 -1.57984108e-01 2.60356456e-01
1.00055687e-01 -2.29071364e-01 -8.04413617e-01 -9.12915707e-01
1.35415003e-01 7.42600262e-01 -7.04586506e-02 6.24317884e-01
-7.57809341e-01 1.36368883e+00 6.30289650e+00 8.20819438e-01
-1.07468939e+00 -1.18828036e-01 2.37962887e-01 -1.21830367e-01
-3.70969445e-01 -1.07841648e-01 -6.60918534e-01 1.61925316e-01
1.20674527e+00 -7.02572390e-02 8.61870289e-01 7.22812355e-01
2.66764965e-02 4.78063524e-03 -1.38400233e+00 3.02373916e-01
-1.67796016e-01 -1.53981030e+00 2.57150203e-01 -3.95446062e-01
4.34101194e-01 -1.15613796e-01 1.53130829e-01 1.13756275e+00
7.34947264e-01 -1.49559307e+00 9.10649776e-01 7.46331453e-01
8.62092495e-01 -7.83391953e-01 4.98469591e-01 5.11897147e-01
-9.60593760e-01 -3.27046245e-01 -2.35842224e-02 -6.45716608e-01
-2.54811257e-01 1.73170388e-01 -7.18071938e-01 3.04730207e-01
2.85776585e-01 7.31418431e-01 -6.82084143e-01 4.31697994e-01
-8.09060574e-01 8.56485903e-01 -1.59411952e-01 -1.65067300e-01
1.17669880e-01 2.65445292e-01 1.59578100e-01 1.17157686e+00
-4.19452228e-02 3.96018118e-01 5.55055439e-02 1.55314016e+00
-1.80913433e-02 -2.73242861e-01 -5.68903029e-01 -2.94140369e-01
5.85171521e-01 8.10424507e-01 -3.21533173e-01 -4.51892614e-01
-2.78057069e-01 4.04646248e-01 8.64147842e-01 4.94686604e-01
-9.96665537e-01 -5.81717253e-01 4.14733499e-01 -6.97005615e-02
8.75854269e-02 -4.52497572e-01 -8.03253710e-01 -1.30886245e+00
1.95769407e-02 -1.30956948e+00 1.96627989e-01 -1.06668234e+00
-8.50279868e-01 3.82262260e-01 2.17613146e-01 -5.33589423e-01
-3.66127968e-01 -1.01459086e+00 -8.28120470e-01 6.96465492e-01
-1.36433506e+00 -1.22911870e+00 -2.48140693e-01 5.54892242e-01
3.74929041e-01 -3.04695457e-01 9.87322628e-01 -9.92751587e-03
-7.43386745e-01 9.13392067e-01 -4.93716478e-01 4.11997914e-01
1.69287995e-03 -1.13205671e+00 5.30151069e-01 8.37485731e-01
-2.99270302e-01 1.40515888e+00 3.49459797e-01 -5.50131381e-01
-1.64101827e+00 -9.11901891e-01 6.79839611e-01 -5.00777721e-01
9.95010853e-01 -4.74635363e-01 -1.05253232e+00 1.17215073e+00
1.44745335e-01 -3.30972821e-01 6.40134215e-01 2.78523833e-01
-7.44557858e-01 1.02962852e-01 -7.65723944e-01 9.48604584e-01
1.40325677e+00 -5.59121013e-01 -1.06496203e+00 1.04804344e-01
1.00181520e+00 -6.61164701e-01 -8.59313190e-01 4.85304981e-01
6.87441826e-01 -7.14592934e-01 9.03225422e-01 -8.84962738e-01
1.08629167e+00 -2.65634030e-01 5.74898953e-03 -1.06433833e+00
-3.00100058e-01 -4.57241327e-01 -8.28166068e-01 9.63866413e-01
6.21238649e-01 -9.10136938e-01 4.15633827e-01 8.18027914e-01
-3.83536577e-01 -1.11827528e+00 -5.12708843e-01 -7.88014650e-01
4.83753800e-01 -9.76164997e-01 9.18661118e-01 9.15432274e-01
4.78718162e-01 6.69127777e-02 1.37883231e-01 -5.29060178e-02
1.70808509e-01 1.74630955e-01 6.44480228e-01 -9.81754124e-01
-5.47419667e-01 -1.02998638e+00 1.06648482e-01 -9.71200645e-01
8.75134230e-01 -1.51836586e+00 -8.25424194e-02 -1.25712955e+00
-3.69700380e-02 -4.11752373e-01 -2.11386323e-01 1.27149832e+00
-3.64382267e-02 -1.62634626e-01 3.06810647e-01 -1.24375820e-01
-6.55065358e-01 4.37628508e-01 1.15893209e+00 -1.59150437e-01
1.04203075e-02 -4.15423065e-01 -8.47887278e-01 8.15129042e-01
1.03689110e+00 -1.55974343e-01 -5.43426216e-01 -8.47993374e-01
6.52198851e-01 4.97183613e-02 4.95605707e-01 -1.16660750e+00
4.91678804e-01 -4.19484228e-01 2.03000069e-01 -1.37517855e-01
3.38919461e-03 -5.49383044e-01 -1.63912714e-01 6.94548845e-01
-8.53251934e-01 8.15403834e-02 6.60055399e-01 -2.28197109e-02
-1.36593103e-01 -1.93607926e-01 3.67976516e-01 -3.21478128e-01
-7.84162343e-01 -1.16128720e-01 -5.04608154e-01 2.03401223e-01
7.31241524e-01 -1.23130165e-01 -6.23850644e-01 -2.22510561e-01
-7.71547496e-01 6.51376486e-01 1.86809853e-01 2.15270385e-01
7.86447704e-01 -1.09750569e+00 -4.68601137e-01 3.17922980e-01
1.02874875e-01 2.25681111e-01 -1.73454151e-01 1.00379980e+00
-7.84426212e-01 6.59517646e-01 -2.57339686e-01 -1.03541367e-01
-7.49793530e-01 5.38849175e-01 5.20452678e-01 -3.24321032e-01
-5.67663491e-01 9.50355828e-01 1.74079984e-02 -9.07570660e-01
2.98458099e-01 -8.92398179e-01 9.11667347e-02 -6.65651321e-01
5.98461866e-01 2.10131779e-01 -1.66159302e-01 2.90048808e-01
-2.19214991e-01 1.59297988e-01 -1.16476178e-01 1.92291185e-01
1.27520072e+00 4.97200996e-01 -4.29356098e-01 4.62670803e-01
5.54064572e-01 -1.84791207e-01 -9.85653222e-01 -2.08185151e-01
8.93266499e-02 1.78609505e-01 -1.01506531e-01 -1.27031839e+00
-7.50206470e-01 9.13403034e-01 -2.49130949e-01 -1.01764686e-01
1.13281786e+00 -4.07163143e-01 7.68767118e-01 9.94701803e-01
3.91615480e-01 -8.31067383e-01 -1.40878499e-01 1.23547542e+00
9.04794514e-01 -8.48188341e-01 -2.54581451e-01 -3.00487608e-01
-3.02166760e-01 1.39153945e+00 1.14679146e+00 -3.21280479e-01
1.60390064e-01 6.92286432e-01 -2.74296135e-01 -9.29097161e-02
-1.30196917e+00 -5.42923696e-02 3.43176462e-02 4.27337140e-01
6.19478941e-01 -1.39838653e-02 2.83020269e-02 1.02400982e+00
-6.67845070e-01 5.84215820e-01 3.16645950e-01 1.04505610e+00
-2.91035056e-01 -9.12219524e-01 -1.30675316e-01 5.74917734e-01
-1.90721154e-01 -6.26760900e-01 -2.44366169e-01 8.60511124e-01
1.79129109e-01 6.79981709e-01 -4.98537607e-02 -4.93854880e-01
2.91761726e-01 6.31991327e-01 7.11747169e-01 -7.20382869e-01
-9.65110302e-01 -7.20654070e-01 2.97393233e-01 -8.40600014e-01
1.46751344e-01 -4.44944412e-01 -1.80836523e+00 -6.86446667e-01
9.26224813e-02 -1.40124440e-01 1.67620227e-01 9.78841782e-01
3.69014591e-01 9.91917312e-01 -1.46264538e-01 -4.30689007e-01
-8.12356293e-01 -7.09532917e-01 4.16211672e-02 4.52893078e-02
1.25270963e-01 -4.04612452e-01 -2.26720255e-02 -4.34268415e-02] | [9.387323379516602, 7.293854713439941] |
f4dc239f-fc41-45e7-b381-ad68f031e124 | quantum-machine-learning-for-malware | 2305.09674 | null | https://arxiv.org/abs/2305.09674v3 | https://arxiv.org/pdf/2305.09674v3.pdf | Quantum Machine Learning for Malware Classification | In a context of malicious software detection, machine learning (ML) is widely used to generalize to new malware. However, it has been demonstrated that ML models can be fooled or may have generalization problems on malware that has never been seen. We investigate the possible benefits of quantum algorithms for classification tasks. We implement two models of Quantum Machine Learning algorithms, and we compare them to classical models for the classification of a dataset composed of malicious and benign executable files. We try to optimize our algorithms based on methods found in the literature, and analyze our results in an exploratory way, to identify the most interesting directions to explore for the future. | ['Tony Quertier', 'Grégoire Barrué'] | 2023-05-09 | null | null | null | null | ['malware-classification'] | ['miscellaneous'] | [ 3.39237303e-01 -4.62402366e-02 -3.72178078e-01 -6.46508485e-02
-2.30539739e-01 -7.60283470e-01 9.46062088e-01 1.05947599e-01
-2.23213121e-01 4.37253267e-01 -5.08102119e-01 -1.06077468e+00
9.13762078e-02 -8.78181100e-01 -5.67055047e-01 -7.10753679e-01
-4.75496083e-01 4.33195889e-01 4.00791913e-01 -4.04371768e-01
7.06221700e-01 5.90430200e-01 -1.26984894e+00 2.94835538e-01
5.93374431e-01 4.54723209e-01 -3.42009962e-01 9.54227924e-01
2.65875161e-01 8.71861160e-01 -5.88653505e-01 -5.87332785e-01
5.04890501e-01 -4.43914682e-01 -1.08180749e+00 -2.41291642e-01
1.27462253e-01 -1.80925682e-01 -5.01441121e-01 1.49467099e+00
-1.97592035e-01 -6.93306103e-02 8.68626714e-01 -1.45920765e+00
-6.72718406e-01 5.27891874e-01 1.35116803e-03 4.82086241e-01
1.95829913e-01 5.47607362e-01 9.69259202e-01 -6.24623857e-02
7.93600559e-01 1.27538121e+00 5.86618364e-01 1.02011490e+00
-1.32728732e+00 -6.35066092e-01 -6.55247808e-01 6.94034874e-01
-1.10693216e+00 -1.46604791e-01 4.92798269e-01 -4.94438946e-01
1.13802135e+00 1.27968967e-01 3.96476746e-01 1.56630695e+00
7.87855208e-01 3.61721426e-01 1.45415747e+00 -3.82124782e-01
3.88682753e-01 3.74088079e-01 4.31219220e-01 1.05101168e+00
6.94359541e-01 7.66846478e-01 -2.06542000e-01 -7.27903247e-01
4.52999286e-02 -2.41209373e-01 -2.14953974e-01 -2.72030681e-01
-8.96258712e-01 1.40355086e+00 3.14099878e-01 4.70503569e-01
8.38906318e-02 3.67624670e-01 6.70306623e-01 5.46871781e-01
3.20596337e-01 7.93286800e-01 -4.87762839e-01 -4.07243192e-01
-6.00840449e-01 1.16059974e-01 1.24450779e+00 5.49446821e-01
9.15297389e-01 -4.13029306e-02 5.17372251e-01 -2.36292928e-01
2.92519093e-01 5.13110995e-01 2.80788809e-01 -9.01903808e-01
-6.19721673e-02 -9.16494336e-03 -3.10308009e-01 -5.78117728e-01
-3.68265301e-01 -2.19071731e-01 -3.59786034e-01 3.03090602e-01
3.00119728e-01 -6.78344071e-02 -3.65789503e-01 1.34438086e+00
-1.27253532e-01 4.87103492e-01 3.88666570e-01 2.08824441e-01
2.94039160e-01 5.10698915e-01 -1.05511069e-01 -5.04783332e-01
8.05800140e-01 -5.03285170e-01 -3.52790058e-01 3.84940840e-02
1.33161187e+00 -5.25722444e-01 7.58362412e-01 6.56005681e-01
-1.13875285e-01 -1.44059118e-02 -1.08862758e+00 3.63283455e-01
-4.96103793e-01 -4.66346204e-01 9.58088219e-01 1.45000124e+00
-8.91863286e-01 1.27299058e+00 -1.13543749e+00 -5.54657042e-01
4.00124311e-01 4.03188258e-01 9.83952265e-03 2.90355891e-01
-1.17884338e+00 1.03051686e+00 5.93633652e-01 -5.09828806e-01
-1.23353374e+00 -1.52143948e-02 -4.70496207e-01 -1.56030789e-01
5.20194769e-01 -4.67695326e-01 1.20250762e+00 -6.24658167e-01
-1.54844129e+00 7.10483670e-01 -1.25328541e-01 -7.74389684e-01
1.06651038e-02 4.10370409e-01 -3.98278296e-01 2.79360265e-01
-3.18831295e-01 -8.11729580e-02 9.86121714e-01 -9.32852983e-01
-1.75402641e-01 -5.71795225e-01 4.28281158e-01 -6.54787540e-01
-3.16384643e-01 6.08230941e-02 6.85604870e-01 -7.36165121e-02
-2.37228692e-01 -1.49594307e+00 -2.15484381e-01 -9.29219663e-01
-4.72329855e-01 -2.87170738e-01 9.25157607e-01 -2.92263240e-01
1.08301926e+00 -2.12006044e+00 1.90589592e-01 1.45232424e-01
1.13494180e-01 2.40923882e-01 1.28994018e-01 4.69112009e-01
3.52677703e-02 7.20379889e-01 -1.44864887e-01 1.17225029e-01
1.30185932e-02 3.88715267e-01 -6.29184067e-01 7.35172451e-01
1.23394914e-01 8.97291839e-01 -8.83536935e-01 -2.95639396e-01
-6.13232851e-02 7.78627023e-02 -7.64854014e-01 -1.47754833e-01
-4.94484335e-01 5.37461936e-01 -5.89117467e-01 6.15617454e-01
3.85416240e-01 -4.89821285e-01 9.62281078e-02 2.03010559e-01
6.27341345e-02 6.29216373e-01 -4.79293108e-01 1.00674593e+00
-3.35755110e-01 8.29356313e-01 -3.33767384e-01 -1.05327451e+00
4.24027979e-01 6.07086867e-02 9.72234756e-02 -6.52297735e-02
5.36001980e-01 3.28190029e-01 8.15915465e-01 -6.76594317e-01
3.37535739e-01 -3.98339331e-01 -6.10666871e-02 8.84018779e-01
3.19810770e-02 -3.48850101e-01 1.67882681e-01 2.14848995e-01
1.44673741e+00 -1.43323436e-01 4.04028624e-01 -2.49559358e-01
7.14782000e-01 3.50685984e-01 -2.19365545e-02 1.13774848e+00
-5.44865370e-01 -3.07717413e-01 8.14513206e-01 -2.90347338e-01
-9.27766085e-01 -9.76410389e-01 -4.65172052e-01 1.00816989e+00
1.25857785e-01 -8.27731490e-01 -9.76261735e-01 -9.61046875e-01
-2.49372467e-01 9.95274365e-01 -5.62589586e-01 -7.04811215e-01
-5.60741127e-01 -1.25880730e+00 8.44220996e-01 -6.99053183e-02
7.11591691e-02 -7.29877234e-01 -6.77728295e-01 -1.70880154e-01
4.65197295e-01 -1.19462860e+00 3.53250027e-01 5.01925051e-01
-9.97347832e-01 -1.41648948e+00 3.62234592e-01 -1.83327287e-01
2.84074575e-01 1.67977363e-02 7.85287201e-01 4.10946071e-01
-3.03885311e-01 5.69048107e-01 -6.83100402e-01 -3.25943917e-01
-1.62661338e+00 2.50891119e-01 5.12376726e-01 -3.94675732e-01
7.84416497e-01 -2.12454468e-01 7.93115199e-02 -7.93044493e-02
-1.03695321e+00 -6.35876656e-01 4.33327734e-01 8.60226810e-01
-1.82882890e-01 2.31816903e-01 1.77880451e-01 -1.10212982e+00
6.13932073e-01 -5.96811712e-01 -6.99285209e-01 1.74010664e-01
-8.75269949e-01 4.10195291e-01 1.13545752e+00 -5.21159708e-01
-4.61804837e-01 -2.50071138e-01 -2.57056355e-01 -2.51687229e-01
-2.10546792e-01 2.16237247e-01 2.23239750e-01 -8.03898275e-01
1.14400601e+00 3.51290405e-01 1.35414004e-01 -4.31409404e-02
4.68355983e-01 8.84493351e-01 -1.79832384e-01 -6.00782871e-01
1.21491957e+00 5.74388027e-01 6.41058028e-01 -1.14448917e+00
-6.88732922e-01 -1.57511141e-03 -6.53079450e-01 9.21885818e-02
8.38676274e-01 -1.55607715e-01 -8.35928738e-01 1.29426345e-01
-1.16617501e+00 -1.28634661e-01 3.82309780e-02 5.22798359e-01
-6.88951969e-01 9.60181713e-01 -8.19099486e-01 -7.55674243e-01
2.24694703e-02 -1.54047167e+00 6.50835931e-01 6.75263703e-02
-5.73459230e-02 -1.26940000e+00 2.56037742e-01 4.46561389e-02
3.27120453e-01 -1.24142371e-01 1.32250035e+00 -1.42602456e+00
-7.56547749e-01 -2.15285018e-01 1.82286069e-01 4.01738077e-01
-8.12681168e-02 3.50680232e-01 -9.40611660e-01 -5.43155909e-01
4.58646506e-01 -2.62804836e-01 9.35886085e-01 -2.51136363e-01
1.11889982e+00 -3.28979373e-01 -3.69473100e-01 5.49137414e-01
1.31261086e+00 2.85958618e-01 3.15532357e-01 4.04170364e-01
3.92674714e-01 3.16313505e-01 2.41297990e-01 2.53644854e-01
-1.39638856e-01 4.21743512e-01 7.31972277e-01 9.68639970e-01
4.19194102e-01 2.41102174e-01 6.59750998e-01 8.65184307e-01
-1.83592752e-01 1.43977433e-01 -1.02697003e+00 -1.84836775e-01
-1.30386162e+00 -1.22992063e+00 -4.05297577e-01 2.20079446e+00
4.24152851e-01 4.84797925e-01 1.88477963e-01 -8.83912574e-03
5.52134931e-01 3.41587961e-02 -3.55519235e-01 -9.43348348e-01
3.14118974e-02 7.61066854e-01 6.32583201e-01 4.06839758e-01
-1.22828460e+00 1.10659599e+00 7.22579193e+00 9.38730896e-01
-1.17403626e+00 5.06188571e-01 3.62487435e-01 3.23149979e-01
-7.86182955e-02 7.50190854e-01 -6.43326163e-01 2.01626480e-01
1.54836631e+00 -5.47213316e-01 1.03587306e+00 8.82298708e-01
-2.24081293e-01 5.39541654e-02 -1.52107859e+00 6.83370650e-01
1.85829848e-01 -1.19501579e+00 6.73121065e-02 4.16041732e-01
7.22369075e-01 4.04606491e-01 2.02848256e-01 7.40660906e-01
-8.27197731e-03 -1.04265308e+00 1.34667978e-01 1.44988641e-01
1.28108680e-01 -6.07358217e-01 6.48454189e-01 7.32894182e-01
-5.35270095e-01 -4.53544885e-01 -4.44624305e-01 -4.41467732e-01
-4.83215570e-01 1.70952398e-02 -1.20274770e+00 5.38984656e-01
2.69664377e-01 6.47641063e-01 -1.03062022e+00 6.54481053e-01
-2.62985229e-01 8.90251994e-01 -1.06816679e-01 -7.07008660e-01
3.58422697e-01 -3.64823729e-01 8.44937563e-01 9.91967738e-01
1.71735659e-01 -2.51817346e-01 1.12540886e-01 1.13751030e+00
2.93174148e-01 8.48372430e-02 -1.07416928e+00 -7.90151060e-01
-6.45056814e-02 1.14491749e+00 -8.75431061e-01 -4.52016622e-01
-3.10521752e-01 7.70724177e-01 -4.76907156e-02 2.04522070e-02
-8.13564539e-01 -3.15952986e-01 3.07919174e-01 -1.34152114e-01
-1.22045474e-02 -4.81917232e-01 -2.15912730e-01 -1.50023770e+00
-5.65490544e-01 -1.02151585e+00 1.46057531e-01 -1.40144184e-01
-1.33085907e+00 5.86884618e-01 -3.06167230e-02 -1.02133429e+00
-6.38398409e-01 -1.25496566e+00 -7.36508429e-01 2.94463336e-01
-9.34083700e-01 -6.65387511e-01 3.87806445e-01 2.27127492e-01
4.84963134e-02 -4.44416076e-01 9.08583641e-01 -2.10758388e-01
-4.35077339e-01 3.07094097e-01 4.78492618e-01 -4.28503985e-03
2.94104159e-01 -1.08452475e+00 3.89289051e-01 7.81488359e-01
4.67855215e-01 9.12336349e-01 1.12539911e+00 -7.17797637e-01
-1.96739304e+00 -7.41282761e-01 4.31747079e-01 -9.55879748e-01
1.36618030e+00 -3.74455839e-01 -8.01397622e-01 8.70873988e-01
-5.92552274e-02 -2.50282109e-01 7.57910311e-01 1.19940653e-01
-6.14550591e-01 4.38844919e-01 -1.00563776e+00 5.57642937e-01
7.63659477e-01 -1.04333317e+00 -6.27816617e-01 7.75610566e-01
6.88438177e-01 2.03621000e-01 -5.85363090e-01 3.59960228e-01
3.61003548e-01 -1.22235143e+00 6.02594912e-01 -1.11174512e+00
3.00952792e-01 -6.81277439e-02 -3.15367401e-01 -1.24423909e+00
7.23079741e-02 -7.85506010e-01 -3.14539135e-01 4.51580197e-01
3.14579189e-01 -9.65264976e-01 6.55372620e-01 1.78410653e-02
1.13442324e-01 -6.15019321e-01 -1.33080816e+00 -1.37597656e+00
6.24761760e-01 -6.92532301e-01 1.79186717e-01 7.72899091e-01
4.29227620e-01 3.21828038e-01 -1.33562550e-01 2.80487388e-01
8.61924887e-01 2.16067061e-01 7.59834409e-01 -1.32229817e+00
-8.27049792e-01 -7.88928270e-01 -8.49953353e-01 -6.24444664e-01
8.52888703e-01 -1.42705119e+00 -3.29766631e-01 -4.25650597e-01
6.67194009e-01 -1.40542299e-01 5.36295362e-02 -2.19149832e-02
1.35572642e-01 2.76501834e-01 1.87210530e-01 3.50759506e-01
-5.09270251e-01 1.75741211e-01 8.91450763e-01 -2.87340432e-01
9.25468951e-02 2.54651189e-01 -2.84631848e-01 8.45116913e-01
9.45296049e-01 -7.42701590e-01 3.08411079e-04 3.21912318e-01
3.91910464e-01 -1.42261595e-01 4.82308924e-01 -9.54916656e-01
-1.96543798e-01 -3.18948746e-01 -2.20643863e-01 -1.51547045e-01
1.81614742e-01 -5.01051426e-01 -2.18691260e-01 1.33553135e+00
-1.11012690e-01 -1.91386402e-01 -3.13871622e-01 5.64813793e-01
3.24472547e-01 -1.12963557e+00 8.55835676e-01 -2.81562001e-01
-4.01515633e-01 2.47499645e-01 -7.04414010e-01 9.42986179e-03
1.29740274e+00 2.39425361e-01 -6.85102284e-01 -2.84221470e-01
-7.18987286e-01 -4.94900376e-01 1.04495907e+00 2.53368527e-01
2.25453094e-01 -7.61916041e-01 -3.29346389e-01 -3.63141708e-02
1.66142583e-01 -1.16788459e+00 -7.25084022e-02 9.90234613e-01
-8.01306963e-01 8.18667531e-01 3.48568410e-02 -6.52473927e-01
-1.16950715e+00 1.31431067e+00 2.94631243e-01 -1.85157001e-01
-2.10867330e-01 2.12357536e-01 -2.28912473e-01 -3.05537015e-01
-3.36623073e-01 -9.76232961e-02 8.40908587e-02 -4.40045804e-01
3.75668555e-01 5.35663366e-01 -8.32983777e-02 -8.89619648e-01
-4.53238666e-01 3.05262715e-01 2.83073690e-02 1.42090112e-01
8.84729505e-01 2.93925047e-01 -6.19615614e-01 7.57966042e-01
1.39717650e+00 6.60997108e-02 -3.23857158e-01 1.27469495e-01
4.34107691e-01 -4.26328301e-01 -3.90209228e-01 -2.55545139e-01
-1.79885179e-01 1.13463199e+00 7.34174967e-01 9.29225922e-01
6.79668665e-01 3.22010368e-01 5.94670355e-01 1.06173646e+00
8.89125824e-01 -6.35457277e-01 1.78849190e-01 9.83638406e-01
1.20724933e-02 -1.22120082e+00 1.06139630e-02 -4.18922693e-01
-6.70526922e-02 1.66400194e+00 2.04601213e-01 -4.70093071e-01
6.12855375e-01 5.41411862e-02 -3.85824978e-01 -2.17186093e-01
-9.56714451e-01 -2.45219216e-01 -1.35196056e-02 5.52070677e-01
2.54026085e-01 2.78511226e-01 -4.52853411e-01 1.01074859e-01
-3.17489266e-01 -3.16012591e-01 1.39291871e+00 1.05789351e+00
-6.31618202e-01 -1.47911465e+00 -5.36304057e-01 6.30891919e-01
-6.78904355e-01 -2.59005487e-01 -6.47663653e-01 7.77024448e-01
1.75439045e-01 1.12989306e+00 -4.06082243e-01 -9.66261506e-01
-5.51175952e-01 4.52988446e-01 8.62186432e-01 -8.76513600e-01
-4.65097994e-01 -4.05237645e-01 -1.28461838e-01 -2.54257053e-01
-3.12290341e-01 -6.89139903e-01 -1.16994786e+00 -4.59073693e-01
-6.13578796e-01 1.93764433e-01 7.86233723e-01 1.27605629e+00
2.54192725e-02 3.49120498e-01 8.75034869e-01 -7.30364323e-01
-1.30855525e+00 -8.11911643e-01 -5.31283975e-01 2.62772381e-01
2.33027384e-01 -5.81304371e-01 -9.53120232e-01 -2.38629475e-01] | [5.576709747314453, 5.134101390838623] |
d0826e0c-f866-4d2c-a4b3-c29a863f4df2 | is-style-all-you-need-dependencies-between | 2211.08213 | null | https://arxiv.org/abs/2211.08213v1 | https://arxiv.org/pdf/2211.08213v1.pdf | Is Style All You Need? Dependencies Between Emotion and GST-based Speaker Recognition | In this work, we study the hypothesis that speaker identity embeddings extracted from speech samples may be used for detection and classification of emotion. In particular, we show that emotions can be effectively identified by learning speaker identities by use of a 1-D Triplet Convolutional Neural Network (CNN) & Global Style Token (GST) scheme (e.g., DeepTalk Network) and reusing the trained speaker recognition model weights to generate features in the emotion classification domain. The automatic speaker recognition (ASR) network is trained with VoxCeleb1, VoxCeleb2, and Librispeech datasets with a triplet training loss function using speaker identity labels. Using an Support Vector Machine (SVM) classifier, we map speaker identity embeddings into discrete emotion categories from the CREMA-D, IEMOCAP, and MSP-Podcast datasets. On the task of speech emotion detection, we obtain 80.8% ACC with acted emotion samples from CREMA-D, 81.2% ACC with semi-natural emotion samples in IEMOCAP, and 66.9% ACC with natural emotion samples in MSP-Podcast. We also propose a novel two-stage hierarchical classifier (HC) approach which demonstrates +2% ACC improvement on CREMA-D emotion samples. Through this work, we seek to convey the importance of holistically modeling intra-user variation within audio samples | ['Arun Ross', 'Morgan Sandler'] | 2022-11-15 | null | null | null | null | ['emotion-classification', 'emotion-classification', 'speaker-recognition'] | ['computer-vision', 'natural-language-processing', 'speech'] | [-4.87138703e-02 1.51556402e-01 1.51391432e-01 -7.22750843e-01
-9.24991727e-01 -4.58918273e-01 4.94655401e-01 7.45511502e-02
-3.83147210e-01 3.12968880e-01 3.19321781e-01 8.51080269e-02
2.73682088e-01 -3.34050804e-01 -2.53102034e-01 -4.98189062e-01
-1.25264734e-01 4.54243347e-02 -5.84997296e-01 -3.32777113e-01
-1.14335984e-01 5.38321793e-01 -1.80137384e+00 5.24564981e-01
3.50051701e-01 1.73618090e+00 -6.59313321e-01 1.04933381e+00
-2.27029081e-02 9.08772767e-01 -9.73614097e-01 -3.53894651e-01
8.78858045e-02 -3.45977277e-01 -8.55139852e-01 -7.10214674e-02
5.20832837e-01 -2.31749080e-02 -1.80387020e-01 6.11655295e-01
8.59598458e-01 2.76574850e-01 6.91912472e-01 -1.68628192e+00
-4.18843597e-01 6.89362109e-01 6.48543984e-02 -1.45070609e-02
4.58825886e-01 -2.45611489e-01 1.09237778e+00 -1.21725154e+00
2.45789409e-01 1.45956492e+00 8.32300961e-01 6.84329391e-01
-1.03041375e+00 -9.10111427e-01 -5.86802848e-02 2.29381680e-01
-1.71324444e+00 -9.71249104e-01 8.48060548e-01 -4.29068208e-01
1.06092608e+00 4.72039819e-01 3.61798406e-01 1.52869081e+00
-3.63856852e-01 8.68694246e-01 8.90321016e-01 -4.42589641e-01
3.44159901e-01 6.77553177e-01 2.72590965e-01 4.39483225e-01
-8.80412519e-01 6.00584410e-02 -1.04964352e+00 -3.81520361e-01
1.09490700e-01 -3.86944652e-01 -2.09184825e-01 2.40107387e-01
-7.46445477e-01 9.29564774e-01 1.20346501e-01 3.36853325e-01
-3.34198862e-01 -1.42701715e-01 9.42141414e-01 6.48447335e-01
8.46490681e-01 3.24848861e-01 -5.46876192e-01 -5.05743146e-01
-9.40531731e-01 -5.28222993e-02 9.66797769e-01 7.83688784e-01
5.90542972e-01 5.78118980e-01 -2.30582319e-02 1.36864781e+00
1.94120079e-01 2.54834950e-01 8.48192096e-01 -6.65846109e-01
1.34222806e-02 2.59324074e-01 -1.71516597e-01 -8.63834083e-01
-3.65923166e-01 -2.09309086e-01 -6.61768615e-01 4.46926840e-02
-1.07901990e-01 -4.01222020e-01 -5.37262738e-01 1.75532293e+00
2.36825421e-01 2.18856737e-01 5.59294105e-01 7.37994254e-01
1.01321208e+00 7.30328083e-01 9.92343575e-02 1.95841357e-01
1.36313486e+00 -7.91084230e-01 -5.89929461e-01 1.63987920e-01
6.20005131e-01 -5.46554685e-01 1.05064011e+00 4.39313740e-01
-8.13519776e-01 -7.95423448e-01 -9.33198333e-01 -2.38118749e-02
-6.79849029e-01 5.10835946e-01 1.63600177e-01 1.15871608e+00
-1.12970209e+00 2.22434133e-01 -3.62821549e-01 -3.91179889e-01
2.44513467e-01 3.88735741e-01 -5.04071653e-01 5.19426763e-01
-1.37659073e+00 6.30841255e-01 -2.35366467e-02 5.96597698e-03
-1.21501648e+00 -8.77969623e-01 -1.09633267e+00 2.02705577e-01
-4.04894829e-01 1.11357830e-01 1.21517086e+00 -1.44662750e+00
-2.02131796e+00 8.94904971e-01 -1.22166000e-01 -6.06037438e-01
2.96196252e-01 -1.35632366e-01 -1.00241506e+00 3.75636101e-01
-2.10230023e-01 8.23662043e-01 1.10490775e+00 -1.08490169e+00
-7.01970935e-01 -2.79387951e-01 -3.74687374e-01 -8.67006332e-02
-8.01087737e-01 3.60863298e-01 2.67848402e-01 -4.07471597e-01
-1.84887528e-01 -8.01966548e-01 4.65127200e-01 -2.12390020e-01
-4.26719993e-01 -5.72951317e-01 1.03715074e+00 -8.52660239e-01
9.94536161e-01 -2.69174814e+00 -2.15764016e-01 3.74118805e-01
8.84619430e-02 2.08959773e-01 -2.42968872e-01 1.77121893e-01
-3.82272899e-01 1.71726510e-01 6.19719853e-04 -7.55365372e-01
4.35400754e-01 -1.92324236e-01 -4.00050044e-01 3.31673741e-01
5.52640438e-01 3.61738861e-01 -3.77523899e-01 -2.02852339e-01
9.35351178e-02 7.66981125e-01 -4.50572014e-01 3.99302781e-01
2.66469002e-01 2.89840112e-03 6.07310534e-02 6.84272051e-01
5.13741434e-01 5.04102707e-01 -1.92395149e-04 -1.84689358e-01
-2.24519566e-01 4.78577763e-01 -9.99971271e-01 1.24079478e+00
-9.95865762e-01 9.82791960e-01 5.26072860e-01 -1.04186809e+00
1.42960918e+00 8.60121429e-01 3.49175215e-01 -4.29304093e-01
3.12413156e-01 9.56446975e-02 -2.80689567e-01 -5.06506085e-01
6.60250127e-01 -3.74491692e-01 -3.98853570e-01 2.44038075e-01
4.56706792e-01 6.29794374e-02 -5.04616439e-01 -4.64646555e-02
8.20535839e-01 -5.67174852e-01 -1.03654273e-01 -1.32353112e-01
6.98328793e-01 -3.33219767e-01 3.83377522e-01 4.95977104e-01
-5.86567998e-01 4.91759479e-01 4.79637265e-01 -1.80011362e-01
-7.11979389e-01 -9.83299673e-01 -3.00020754e-01 1.57235765e+00
-5.67239523e-01 -2.53187716e-01 -6.43624902e-01 -6.03724957e-01
-1.67553057e-03 7.68287241e-01 -7.18193471e-01 -3.49362850e-01
-2.15650037e-01 -3.15418124e-01 1.28846169e+00 3.97529751e-01
3.63437086e-01 -1.12260032e+00 -1.18031792e-01 2.56897897e-01
-3.95579860e-02 -1.27340078e+00 -4.63280290e-01 3.76318783e-01
-1.17468238e-01 -5.58945477e-01 -6.13925636e-01 -9.97744381e-01
7.76262879e-02 -2.55801558e-01 8.42767179e-01 -5.73683679e-01
-2.88728416e-01 8.23056400e-01 -5.74962020e-01 -3.82542759e-01
-5.58887899e-01 1.02537781e-01 4.71149713e-01 7.31410205e-01
5.45192361e-01 -4.99919832e-01 -1.93506226e-01 2.66729027e-01
-5.00267208e-01 -5.10445654e-01 2.23427657e-02 8.23078454e-01
-4.80392836e-02 -2.20514297e-01 1.17407525e+00 -3.01940650e-01
6.98103964e-01 -5.76900721e-01 -6.42905682e-02 1.92759018e-02
-1.54204905e-01 -2.96248436e-01 6.89603269e-01 -5.38274825e-01
-9.63051379e-01 6.34427443e-02 -5.11081159e-01 -7.42909074e-01
-5.94895780e-01 2.40190864e-01 -3.16150397e-01 6.67512342e-02
7.61595368e-01 2.04989865e-01 1.11664154e-01 -3.16322893e-01
2.49756664e-01 1.61216080e+00 3.80074114e-01 -5.15732646e-01
2.30917364e-01 2.47122228e-01 -7.72308350e-01 -1.43126225e+00
-4.75064367e-01 -5.77315629e-01 -2.09303364e-01 -2.77987540e-01
9.84363198e-01 -1.23145521e+00 -1.11519086e+00 5.37944078e-01
-9.13127184e-01 -3.27211082e-01 -2.42981791e-01 5.31584978e-01
-3.63675773e-01 8.60567484e-03 -6.80478752e-01 -1.27755594e+00
-5.45765936e-01 -8.86970460e-01 1.14509869e+00 7.52221793e-02
-7.34514415e-01 -7.65285969e-01 -1.15814678e-01 2.47257397e-01
5.09750605e-01 7.72128925e-02 6.16412520e-01 -1.26597524e+00
2.84476638e-01 -4.20904100e-01 -1.30019784e-01 9.88863051e-01
8.12158883e-02 7.85527453e-02 -1.74168408e+00 -7.38762319e-02
3.55778635e-02 -6.89004958e-01 4.92049575e-01 1.71864200e-02
1.20615244e+00 -3.63887966e-01 2.81693637e-01 4.25238103e-01
8.29832852e-01 1.02657177e-01 3.01582128e-01 -1.21994928e-01
4.30018723e-01 9.95716274e-01 2.05675825e-01 6.79351807e-01
4.73953426e-01 6.46338105e-01 1.41576231e-01 1.18349627e-01
7.62616401e-04 1.92589723e-02 7.77216375e-01 9.77511644e-01
5.44443548e-01 -1.13532215e-01 -8.93382192e-01 8.03346932e-01
-1.24583340e+00 -9.31727111e-01 1.03741691e-01 1.93159592e+00
7.83806384e-01 -2.28759378e-01 4.00080979e-01 3.69270176e-01
7.66439259e-01 8.14357474e-02 -2.35455662e-01 -9.78719950e-01
-1.22590162e-01 4.63548809e-01 -3.21679004e-02 5.16465664e-01
-1.19515610e+00 7.45857537e-01 5.51935244e+00 6.56055629e-01
-1.67999852e+00 1.74075752e-01 8.01366806e-01 -3.56285363e-01
2.16520056e-01 -6.28083646e-01 -7.83674479e-01 4.30399120e-01
1.64783418e+00 -1.39488548e-01 2.40689546e-01 1.09814501e+00
2.35514328e-01 4.55079228e-01 -1.19934082e+00 1.36028910e+00
4.44708586e-01 -8.84147406e-01 -3.17214102e-01 -2.86984980e-01
2.97428340e-01 -7.94566178e-04 2.77380526e-01 7.99857080e-01
7.18301311e-02 -1.00091815e+00 8.92202616e-01 1.66404843e-01
8.36602211e-01 -1.05322313e+00 6.31138146e-01 -1.07769579e-01
-1.11112905e+00 -2.11873531e-01 -5.07651009e-02 7.96192437e-02
-7.52028525e-02 3.79981905e-01 -1.16977108e+00 3.69351268e-01
9.05494988e-01 6.10054731e-01 -1.37459442e-01 4.42320049e-01
2.81528682e-01 9.30567861e-01 -4.48500007e-01 -1.32580400e-01
2.20492825e-01 2.93236941e-01 3.62149388e-01 1.53441596e+00
4.05814111e-01 -1.62209332e-01 -2.29537830e-01 6.22826219e-01
-4.04280275e-01 3.02931041e-01 -2.96636730e-01 -2.40061551e-01
5.13581455e-01 1.34564388e+00 6.57766610e-02 -5.15821397e-01
-5.74097335e-02 1.07370257e+00 2.44065553e-01 3.56472820e-01
-7.40627110e-01 -7.58216441e-01 1.30630147e+00 -3.98128062e-01
3.31893891e-01 9.86493453e-02 -8.29497650e-02 -9.93733287e-01
-2.55533755e-01 -9.99680459e-01 1.75133467e-01 -6.75404727e-01
-1.59213769e+00 9.23961580e-01 -4.58931535e-01 -1.13242173e+00
-2.98586130e-01 -5.99674702e-01 -9.06851530e-01 8.57450068e-01
-1.23038447e+00 -9.67902780e-01 -2.85670847e-01 6.91659033e-01
4.80388045e-01 -5.38890779e-01 1.15296113e+00 4.89850581e-01
-7.18461931e-01 1.14369631e+00 9.79518071e-02 4.79855299e-01
8.75797510e-01 -1.16960013e+00 1.81030437e-01 1.47479147e-01
-5.07397763e-03 2.34257072e-01 3.92119855e-01 8.00753087e-02
-1.00869644e+00 -1.26961744e+00 1.07669687e+00 -1.36267230e-01
6.55058384e-01 -8.56612027e-01 -7.48499453e-01 4.71715182e-01
2.79878616e-01 -7.92165473e-03 1.23101425e+00 3.41630846e-01
-6.81838214e-01 -2.09551707e-01 -1.42280650e+00 3.65484208e-01
4.37221825e-01 -1.29164720e+00 -3.96805257e-01 1.01920389e-01
8.13891649e-01 2.35604886e-02 -1.24254155e+00 2.60729343e-02
6.40109181e-01 -7.43199527e-01 9.21105087e-01 -6.93048775e-01
3.66126597e-01 1.90536171e-01 -5.40184140e-01 -1.38733745e+00
2.55701423e-01 -5.65146267e-01 1.85293779e-01 1.75915575e+00
4.69904423e-01 -7.94352829e-01 5.67880273e-01 6.00563705e-01
-4.04678524e-01 -3.83895218e-01 -1.29623985e+00 -7.24605381e-01
1.35714546e-01 -8.20747495e-01 5.82334459e-01 1.37978423e+00
2.47289687e-01 2.86456853e-01 -4.33164507e-01 2.98068434e-01
2.67347038e-01 -3.23460013e-01 7.56904721e-01 -1.04430711e+00
6.30255556e-03 -3.39553386e-01 -6.26858950e-01 -5.62477887e-01
6.78906083e-01 -9.69556391e-01 1.36857644e-01 -7.15917766e-01
-5.96947491e-01 -3.20010573e-01 -4.29444134e-01 4.52017367e-01
3.13328892e-01 2.57054746e-01 1.10081717e-01 -2.61957973e-01
-3.19020540e-01 9.14096832e-01 2.79194742e-01 -1.74196929e-01
-2.37377077e-01 -6.52101710e-02 -4.72051203e-01 3.62378508e-01
1.02790904e+00 -4.31980312e-01 2.66004377e-03 1.96003824e-01
-2.61757731e-01 6.96290806e-02 5.71395516e-01 -1.03012156e+00
4.57694903e-02 5.21963239e-01 2.63043523e-01 -2.65099317e-01
7.88142025e-01 -6.55391634e-01 -3.64957273e-01 2.12648120e-02
-7.65044928e-01 -2.88015395e-01 4.59224433e-01 1.93410903e-01
-5.77312589e-01 -1.19355850e-01 6.41145647e-01 3.67493659e-01
-4.45328236e-01 3.84897962e-02 -8.65083039e-01 -3.00957561e-02
6.01036727e-01 1.59785370e-04 -8.71098042e-02 -5.53578019e-01
-1.12522340e+00 4.81700487e-02 -1.05987199e-01 7.18922794e-01
4.89427000e-01 -1.55521822e+00 -8.19274426e-01 5.03143072e-01
3.75663310e-01 -5.33603787e-01 5.85286558e-01 6.67207897e-01
-1.64992511e-02 2.33661935e-01 -6.93769455e-02 -5.80935419e-01
-1.46201313e+00 1.42489141e-03 5.39527833e-01 3.50045651e-01
-4.22552340e-02 1.21038187e+00 -3.54656258e-05 -1.02401626e+00
4.52958554e-01 -1.86323658e-01 -5.90429530e-02 6.00210667e-01
5.68288803e-01 4.12000567e-01 2.36743599e-01 -9.05907750e-01
-5.00986755e-01 2.30679195e-02 -8.14803913e-02 -3.78228575e-01
1.26548231e+00 -8.30145031e-02 7.07343668e-02 5.88534236e-01
1.95166719e+00 1.22816429e-01 -9.32976663e-01 -4.80209589e-02
-7.96802640e-02 4.92035486e-02 2.28452861e-01 -7.14325428e-01
-8.52307141e-01 1.17820835e+00 1.09885728e+00 5.46855867e-01
8.89147460e-01 2.46973999e-04 7.11955070e-01 1.64416835e-01
-2.05295488e-01 -1.38054132e+00 1.14826329e-01 6.25263155e-01
9.93489921e-01 -1.24074662e+00 -8.30643535e-01 1.86138265e-02
-8.98198009e-01 1.10568893e+00 5.00737131e-01 6.54331893e-02
7.71506011e-01 1.25506490e-01 5.10910571e-01 -1.04296580e-01
-7.58724928e-01 4.26806957e-02 7.90619254e-02 6.10376060e-01
5.01565576e-01 3.02718997e-01 5.34482777e-01 1.01654267e+00
-5.17413795e-01 -4.63118851e-01 3.73578489e-01 6.08073354e-01
-3.08413118e-01 -8.22721720e-01 -4.94524091e-01 2.48058155e-01
-4.13872272e-01 -1.77726462e-01 -6.69837713e-01 4.36088294e-01
-6.93414360e-02 1.29426086e+00 3.48385572e-01 -6.12179875e-01
3.82299274e-01 8.33020210e-01 -1.84507117e-01 -4.35990334e-01
-1.15495420e+00 -1.37355804e-01 4.59481478e-01 -3.02862257e-01
-1.55524909e-01 -6.92177474e-01 -1.04583347e+00 -1.82995453e-01
4.75090276e-03 3.39398712e-01 1.02321088e+00 6.58709347e-01
6.22069657e-01 6.57949924e-01 1.06374240e+00 -8.16829443e-01
-4.78740364e-01 -1.17952752e+00 -6.84247553e-01 4.03203666e-01
5.88129222e-01 -3.39143902e-01 -8.69384646e-01 3.97825018e-02] | [13.737884521484375, 5.838863372802734] |
f28ae56a-d560-4cdf-865c-6933dd293831 | fusion-of-hyperspectral-and-ground | 1804.05273 | null | http://arxiv.org/abs/1804.05273v3 | http://arxiv.org/pdf/1804.05273v3.pdf | Fusion of hyperspectral and ground penetrating radar to estimate soil moisture | In this contribution, we investigate the potential of hyperspectral data
combined with either simulated ground penetrating radar (GPR) or simulated
(sensor-like) soil-moisture data to estimate soil moisture. We propose two
simulation approaches to extend a given multi-sensor dataset which contains
sparse GPR data. In the first approach, simulated GPR data is generated either
by an interpolation along the time axis or by a machine learning model. The
second approach includes the simulation of soil-moisture along the GPR profile.
The soil-moisture estimation is improved significantly by the fusion of
hyperspectral and GPR data. In contrast, the combination of simulated,
sensor-like soil-moisture values and hyperspectral data achieves the worst
regression performance. In conclusion, the estimation of soil moisture with
hyperspectral and GPR data engages further investigations. | ['Sina Keller', 'Felix M. Riese'] | 2018-04-14 | null | null | null | null | ['soil-moisture-estimation'] | ['computer-vision'] | [ 6.28527641e-01 -5.36798649e-02 3.50241959e-01 -2.61303395e-01
-7.88539767e-01 -2.71954447e-01 3.79248619e-01 3.32019061e-01
-1.23141319e-01 1.23397028e+00 -1.71593547e-01 -8.56628776e-01
-3.74851227e-01 -1.71662915e+00 -3.96165401e-01 -9.71053302e-01
-9.92534962e-03 3.14106971e-01 -3.28141116e-02 -5.05155146e-01
1.33073688e-01 7.93428123e-01 -1.77060640e+00 1.14112580e-02
1.22826755e+00 1.02713406e+00 7.01874435e-01 4.96603578e-01
1.63864523e-01 -5.49696051e-02 6.56257644e-02 2.96746433e-01
3.26049954e-01 1.79815141e-03 -2.34815329e-01 -3.69904637e-02
2.49427795e-01 -3.68106395e-01 1.52479842e-01 1.08618462e+00
5.37577212e-01 -7.13678896e-02 5.76203644e-01 -6.68816328e-01
3.04587930e-01 5.36620080e-01 -1.08120108e+00 -3.67173582e-01
4.36891884e-01 -1.74117610e-01 3.18536431e-01 -6.18245065e-01
5.92138395e-02 8.95726621e-01 9.63335872e-01 -3.88915271e-01
-1.19410276e+00 -3.94863635e-01 -2.77367622e-01 -8.00934359e-02
-1.32350564e+00 -4.61073406e-02 5.95386744e-01 -4.64066595e-01
6.90992236e-01 3.55974168e-01 1.03466463e+00 7.33312666e-01
2.32876271e-01 6.66941851e-02 1.42156863e+00 -7.10053325e-01
4.32128400e-01 -4.31956053e-02 1.52744845e-01 1.20580427e-01
1.01638544e+00 7.77781010e-01 1.31178007e-01 -5.27412415e-01
4.91736978e-01 -4.20796201e-02 -4.77876455e-01 -7.30902478e-02
-4.64398205e-01 1.05534983e+00 1.61889598e-01 1.94173634e-01
-1.16960001e+00 -3.45537603e-01 7.44279996e-02 4.53507900e-02
7.18899071e-01 3.95826280e-01 -5.86806059e-01 3.95791441e-01
-1.69819629e+00 5.24661779e-01 8.09112251e-01 5.48307657e-01
1.09082234e+00 4.73115444e-01 3.29087019e-01 5.95654011e-01
7.16716409e-01 1.30928850e+00 1.69181153e-01 -5.06997943e-01
2.45639518e-01 -4.17698398e-02 5.82132161e-01 -1.06700253e+00
-5.54214716e-01 -3.09945375e-01 -7.49770522e-01 2.29749888e-01
7.63742551e-02 -8.13143551e-01 -7.24301636e-01 1.11130428e+00
2.44767427e-01 1.40391216e-01 5.70164621e-01 5.30673563e-01
8.07588756e-01 9.44208741e-01 4.02972400e-01 -4.86014992e-01
1.12739074e+00 -1.94979742e-01 -5.87273240e-01 -2.23149717e-01
5.05643189e-01 -5.65321624e-01 1.50064901e-01 5.19978404e-01
-6.95756972e-01 -3.73360425e-01 -1.08732748e+00 1.01439309e+00
-6.30894184e-01 2.96241879e-01 7.39650726e-01 8.19473147e-01
-1.02072334e+00 8.76295030e-01 -8.70090187e-01 -5.60084760e-01
-1.39471188e-01 -4.75478023e-02 -5.10154292e-02 -1.67554542e-01
-1.41755724e+00 9.99508679e-01 5.07882655e-01 5.25139213e-01
-1.57660574e-01 -8.07069778e-01 -1.08056080e+00 -8.86287242e-02
-3.53338003e-01 -5.29861033e-01 7.22151637e-01 -3.26528460e-01
-1.50073063e+00 4.55811054e-01 3.51340026e-02 -5.26961684e-01
4.82039034e-01 -1.30299255e-01 -6.60167813e-01 3.88700247e-01
1.28570139e-01 1.47285834e-01 6.10693395e-01 -1.73212564e+00
-6.11864150e-01 -6.05540752e-01 -5.53938508e-01 7.68701211e-02
3.69462043e-01 -3.76398653e-01 6.47080004e-01 -9.22790989e-02
7.85414934e-01 -6.74299181e-01 -6.12693071e-01 -4.95369107e-01
-2.81782359e-01 6.77991450e-01 8.12397480e-01 -9.58455741e-01
6.87191904e-01 -1.85181725e+00 -5.78225136e-01 7.04678833e-01
-4.05850530e-01 2.58990377e-01 -1.58300251e-02 5.68388820e-01
-4.85582769e-01 -4.71203029e-02 -5.03703713e-01 5.74705452e-02
-3.99749607e-01 1.45501792e-01 -5.46584308e-01 6.26748145e-01
1.47370510e-02 3.96460593e-01 -6.38925433e-01 -2.07772013e-02
5.23604095e-01 4.24034834e-01 -3.06709949e-02 1.58653352e-02
-8.33003223e-02 4.41042870e-01 -6.77073300e-01 6.47524178e-01
2.07297039e+00 3.91657054e-01 4.97059494e-01 -5.08414209e-01
-6.30793154e-01 -1.80235967e-01 -1.42093384e+00 1.05706334e+00
-5.44899106e-01 6.51475042e-02 3.96133870e-01 -1.39418519e+00
1.60957766e+00 3.81414860e-01 5.81547737e-01 -8.72912884e-01
-2.36672491e-01 4.34046119e-01 -4.12321299e-01 -5.50708354e-01
9.21832144e-01 -6.02150977e-01 3.26876491e-01 5.61046824e-02
-4.53458279e-01 -8.25842083e-01 -2.81635612e-01 -3.62464190e-01
4.44923967e-01 5.87375879e-01 2.18152821e-01 -6.10410511e-01
4.46623087e-01 2.32233837e-01 2.41566841e-02 7.20245540e-01
3.73909891e-01 4.90887552e-01 6.60494110e-03 4.32357714e-02
-8.94160688e-01 -9.78572845e-01 -8.13060284e-01 5.84466696e-01
9.71835181e-02 3.56179923e-01 -1.97458580e-01 2.12893397e-01
4.16180521e-01 1.09747589e+00 -3.69451761e-01 7.89253339e-02
1.71746418e-01 -1.46177495e+00 4.80110109e-01 2.38845095e-01
7.68320799e-01 -7.76416302e-01 -6.51788473e-01 5.35857797e-01
-3.24121892e-01 -9.79422450e-01 1.11668086e+00 4.57943141e-01
-1.40330398e+00 -8.18271637e-01 -6.08635545e-01 4.40921411e-02
2.26922080e-01 2.90660113e-01 8.72737825e-01 -3.60455662e-01
-6.35270998e-02 4.28991288e-01 -8.68046522e-01 -4.80092257e-01
-2.96350598e-01 -4.48282748e-01 -4.22062039e-01 -1.21065117e-01
2.57778525e-01 -8.69738400e-01 -3.61673743e-01 -3.97248343e-02
-8.98666680e-01 5.02288826e-02 4.44889814e-01 3.19237798e-01
5.70311368e-01 4.16890025e-01 5.33099890e-01 -7.80641913e-01
2.87076443e-01 -1.06626272e+00 -7.04361737e-01 1.45227000e-01
-3.06064844e-01 -2.77300745e-01 8.53648118e-04 -2.69732866e-02
-1.41213226e+00 4.89055971e-03 -2.95335412e-01 3.04754823e-01
-6.36312962e-01 1.32490337e+00 -9.52218547e-02 -2.36152038e-01
8.19643199e-01 3.30725998e-01 -9.09035876e-02 -5.83441496e-01
-1.22503504e-01 7.34195888e-01 5.60407162e-01 -7.61577547e-01
6.01259530e-01 4.47383285e-01 3.36000144e-01 -1.61788845e+00
-5.25916159e-01 -5.44399083e-01 -3.58740270e-01 -2.14278862e-01
3.79580855e-01 -1.02421558e+00 9.41941887e-03 8.22245717e-01
-8.94924164e-01 -4.14701849e-01 -1.96333542e-01 9.90621984e-01
-6.55224562e-01 7.13101149e-01 -6.46656901e-02 -1.44716191e+00
-6.86795354e-01 -5.78840733e-01 1.06220210e+00 1.28789231e-01
1.81711927e-01 -1.01884115e+00 2.03825772e-01 -4.29757722e-02
5.46284556e-01 9.95369792e-01 4.72102940e-01 -1.21480346e-01
1.83213651e-02 -4.70222622e-01 -5.86346924e-01 8.49993303e-02
7.73757696e-02 3.27573985e-01 -1.02020717e+00 -2.56488860e-01
2.38498494e-01 -8.41468647e-02 9.43487525e-01 1.01356828e+00
8.63199949e-01 1.78295657e-01 -2.76938170e-01 7.56120026e-01
2.22309446e+00 1.17346913e-01 1.07012618e+00 5.34713745e-01
4.13310565e-02 5.60149193e-01 1.07793331e+00 1.13954711e+00
5.73770106e-02 4.52381968e-02 8.39945257e-01 -3.65946025e-01
5.13100684e-01 7.72181600e-02 -8.00290108e-02 1.20457271e-02
-6.51671648e-01 -1.92107365e-01 -1.03953183e+00 3.40620577e-01
-1.70427477e+00 -1.26932251e+00 -8.69554400e-01 2.51892018e+00
3.78763050e-01 -2.98832297e-01 -2.66160905e-01 5.11558473e-01
9.44985330e-01 7.38150477e-02 -1.85657740e-01 -2.00416893e-01
-4.40730423e-01 8.09072375e-01 1.14903522e+00 4.29903030e-01
-9.04423475e-01 4.75265831e-01 6.40347338e+00 1.94208086e-01
-1.42471433e+00 -2.28015646e-01 -3.47676054e-02 7.42396355e-01
-6.64746344e-01 1.44460678e-01 -7.96198964e-01 2.93786645e-01
1.17537653e+00 -1.48569131e-02 -3.68327796e-02 6.01141930e-01
7.26430357e-01 -1.10316062e+00 -2.12846979e-01 4.88311768e-01
-5.25518477e-01 -1.00152338e+00 -6.12667501e-02 2.25154281e-01
6.07848287e-01 2.41509318e-01 -2.92699158e-01 8.48349482e-02
4.54000145e-01 -6.07785702e-01 4.09676701e-01 1.04577446e+00
9.46874559e-01 -5.72291672e-01 1.04857469e+00 3.02122772e-01
-1.24217343e+00 1.19397633e-01 -5.95510900e-01 -1.15175746e-01
1.79927275e-01 1.29554737e+00 -4.04301912e-01 1.53738749e+00
5.10862768e-01 7.87945747e-01 -1.90410033e-01 1.50137663e+00
-7.61896819e-02 7.78433204e-01 -7.69940674e-01 4.31554943e-01
3.06943327e-01 -8.39791179e-01 3.33073169e-01 1.29507792e+00
1.26102591e+00 3.60746205e-01 -4.72886814e-03 7.63013661e-01
1.11636221e+00 2.06970453e-01 -7.62345195e-01 2.21882552e-01
5.16344011e-01 1.28173530e+00 -1.32426381e-01 -2.01789826e-01
-2.33753815e-01 2.81786472e-01 -7.34398484e-01 6.13602638e-01
-6.13298059e-01 -2.03614011e-01 3.51043314e-01 4.84783471e-01
3.92609775e-01 -2.48138338e-01 -4.29280818e-01 -7.04942107e-01
-3.89899880e-01 -4.01496559e-01 1.23348214e-01 -1.32087147e+00
-9.78896797e-01 2.35072255e-01 2.32821167e-01 -1.16431105e+00
-2.90715158e-01 -5.24802566e-01 -8.20580125e-01 1.51124692e+00
-2.01694870e+00 -9.43972588e-01 -1.02148163e+00 3.94065589e-01
-4.91318882e-01 7.52995536e-02 1.18196452e+00 -3.50568965e-02
-1.42315954e-01 -2.52979040e-01 6.65129781e-01 -7.97143757e-01
2.16564387e-01 -9.64669764e-01 4.56022471e-02 4.42503273e-01
-9.71973300e-01 -1.38926774e-01 1.19415402e+00 -9.96134520e-01
-1.11587489e+00 -1.33645833e+00 4.94766146e-01 8.10071647e-01
5.62201440e-01 6.26843452e-01 -1.05617023e+00 2.94514805e-01
-2.06979975e-01 -3.47509570e-02 6.10304892e-01 7.90135190e-03
4.08663973e-02 -4.42809612e-02 -1.77067804e+00 -8.66845157e-03
1.01948611e-01 -2.92959750e-01 -5.67923307e-01 1.83184013e-01
-2.55841743e-02 -1.32154673e-01 -1.37939191e+00 1.16140807e+00
7.42264986e-01 -1.15689981e+00 9.08951461e-01 5.46933301e-02
4.60211039e-01 -3.55762616e-02 -8.36868882e-01 -1.53433931e+00
-2.88916439e-01 1.57844536e-02 4.00964200e-01 7.31642842e-01
4.43262190e-01 -8.23504031e-01 7.26334572e-01 3.45393658e-01
5.14204539e-02 -2.15455979e-01 -5.49070537e-01 -5.40545642e-01
3.11990201e-01 -4.03782427e-01 5.87707877e-01 8.02080810e-01
-1.32970676e-01 -4.18242604e-01 1.09926850e-01 8.25953901e-01
8.51205409e-01 3.13281924e-01 4.00916666e-01 -1.89135814e+00
8.59961808e-02 -7.52756968e-02 -1.78744420e-01 -4.00881737e-01
7.00359838e-03 -3.20639461e-01 1.53002337e-01 -1.78742898e+00
-8.15650895e-02 -7.09216118e-01 7.76206553e-02 2.58208036e-01
4.87455949e-02 -1.43888891e-01 -3.57550859e-01 -2.87041008e-01
5.97731411e-01 5.37127078e-01 1.04362035e+00 1.40050307e-01
-4.31294113e-01 3.28484803e-01 -2.28083074e-01 5.70783138e-01
1.06463420e+00 -2.09817588e-01 -1.05024442e-01 7.14318976e-02
3.39313388e-01 9.18805063e-01 5.31703770e-01 -1.25128388e+00
-2.53567249e-01 -3.32133234e-01 3.88080180e-01 -1.25150132e+00
3.52934688e-01 -7.61776149e-01 6.57860696e-01 4.21407938e-01
5.48751235e-01 -6.09891832e-01 7.39926577e-01 2.91532874e-01
-1.92259088e-01 -6.57033205e-01 1.05349851e+00 -1.85938880e-01
-6.16349280e-01 1.53367713e-01 -6.85171008e-01 -6.83586240e-01
6.95421875e-01 -5.85701883e-01 -3.85284096e-01 -2.54257053e-01
-1.00245047e+00 -1.58073902e-01 3.12003791e-01 -5.47821224e-01
5.35551965e-01 -9.33336318e-01 -1.03931773e+00 3.65147769e-01
-7.58413225e-02 -1.14553809e-01 4.73472178e-01 6.93289876e-01
-5.82208395e-01 2.15262592e-01 -4.46745127e-01 -4.99107778e-01
-8.25772941e-01 5.19564673e-02 6.14209652e-01 -2.67966419e-01
-2.63495266e-01 2.58104593e-01 -4.23698246e-01 -8.51566672e-01
-5.91385663e-01 -1.88379318e-01 -4.89898980e-01 5.44668138e-01
2.39347145e-01 4.28562462e-01 3.64273697e-01 -4.34778959e-01
8.95096455e-03 6.71246588e-01 6.74932003e-01 -1.80985466e-01
1.82687545e+00 -1.82552025e-01 -1.44917786e-01 2.80593306e-01
6.76868379e-01 -1.31028533e-01 -9.53331530e-01 -2.08649173e-01
-2.84012228e-01 -2.18403295e-01 6.36828899e-01 -6.48782730e-01
-1.00941670e+00 6.84543848e-01 4.45768178e-01 4.59997982e-01
1.21157920e+00 -6.73679709e-01 2.16485634e-01 4.77619529e-01
5.70847809e-01 -9.55418169e-01 -7.46344745e-01 4.88038063e-01
9.65551138e-01 -1.09780729e+00 4.06428695e-01 -6.20555401e-01
-2.23876983e-01 1.40752029e+00 2.11073738e-02 -1.39187202e-01
1.13425934e+00 4.57718849e-01 -2.78791487e-01 -4.15602252e-02
-8.37261826e-02 -4.72980052e-01 -4.42153215e-01 9.96021748e-01
2.95574009e-01 4.69883800e-01 -3.91804665e-01 5.29354632e-01
-3.12662691e-01 4.87686813e-01 7.22433805e-01 1.03713012e+00
-8.18581343e-01 -7.15897739e-01 -9.49180663e-01 7.24542379e-01
1.89241618e-01 -3.25295150e-01 5.34904122e-01 6.30334496e-01
-8.35358948e-02 1.14519203e+00 1.19191885e-01 3.89460735e-02
2.47024730e-01 -6.97276443e-02 4.08377051e-01 -5.04299164e-01
7.04357401e-02 4.69943173e-02 3.01799834e-01 -2.30850771e-01
-6.83746517e-01 -1.01907003e+00 -8.33326995e-01 -4.72424626e-01
-5.26992440e-01 3.33064109e-01 1.25620365e+00 8.62041593e-01
-1.20525360e-01 2.44403869e-01 8.38105559e-01 -1.34141195e+00
-8.03287208e-01 -1.13525379e+00 -1.77625191e+00 -3.59001070e-01
2.22160816e-01 -8.36126685e-01 -7.37990201e-01 -4.25382435e-01] | [9.425556182861328, -1.6078144311904907] |
92cb253e-1056-4f36-96e6-9ad54c951966 | temporal-question-generation-from-history | null | null | https://aclanthology.org/2021.icon-main.49 | https://aclanthology.org/2021.icon-main.49.pdf | Temporal Question Generation from History Text | Temporal analysis of history text has always held special significance to students, historians and the Social Sciences community in general. We observe from experimental data that existing deep learning (DL) models of ProphetNet and UniLM for question generation (QG) task do not perform satisfactorily when used directly for temporal QG from history text. We propose linguistically motivated templates for generating temporal questions that probe different aspects of history text and show that finetuning the DL models using the temporal questions significantly improves their performance on temporal QG task. Using automated metrics as well as human expert evaluation, we show that performance of the DL models finetuned with the template-based questions is better than finetuning done with temporal questions from SQuAD. | ['Girish Palshikar', 'Sangameshwar Patil', 'Harsimran Bedi'] | null | null | null | null | icon-2021-12 | ['question-generation'] | ['natural-language-processing'] | [-2.95448661e-01 3.34253937e-01 -1.11510754e-01 -8.77668634e-02
-7.76425540e-01 -9.35293615e-01 1.24090958e+00 2.42689520e-01
-5.28768063e-01 7.28397489e-01 7.73076057e-01 -7.01570094e-01
-2.62347311e-01 -9.45647895e-01 -3.50440115e-01 -2.41700709e-01
-1.27632171e-01 7.53116190e-01 5.93839705e-01 -8.03332627e-01
4.07232165e-01 2.90763546e-02 -1.09483242e+00 3.78834099e-01
8.92906666e-01 6.46066248e-01 -2.09407866e-01 8.83856356e-01
-2.40047336e-01 1.16754961e+00 -1.06691563e+00 -6.57235622e-01
8.72275755e-02 -8.69726956e-01 -1.54018176e+00 -5.82961619e-01
7.07857430e-01 -3.03524405e-01 -5.85692942e-01 4.78502542e-01
4.12943929e-01 4.95859712e-01 6.86507225e-01 -1.01350391e+00
-9.28002477e-01 1.17839551e+00 1.53081298e-01 9.14302886e-01
5.51882446e-01 3.10435683e-01 1.51962388e+00 -4.63682055e-01
7.64809549e-01 1.46072829e+00 6.67421937e-01 5.49943507e-01
-1.18297625e+00 -4.34089005e-01 5.16470484e-02 8.47880602e-01
-8.95646036e-01 -8.87296200e-02 7.70985126e-01 -4.46307063e-01
1.21398294e+00 2.55180657e-01 7.86055565e-01 1.37352490e+00
6.21600091e-01 7.55807817e-01 8.58458340e-01 -2.82290012e-01
1.42739937e-01 -3.36555600e-01 9.31965709e-02 5.59139013e-01
-3.77127856e-01 4.13841337e-01 -6.81796968e-01 1.41653106e-01
6.26419067e-01 -6.62445664e-01 1.06907330e-01 4.59785104e-01
-1.37009108e+00 1.15648150e+00 3.78516495e-01 8.45631242e-01
-9.50116441e-02 5.51836610e-01 4.34756458e-01 6.83260679e-01
4.75462139e-01 1.03139973e+00 -6.93833113e-01 -3.38528931e-01
-1.12200463e+00 7.66163468e-01 9.32008147e-01 6.64946675e-01
2.99751997e-01 1.60688534e-01 -7.93136537e-01 4.03269500e-01
-9.19049010e-02 3.67484927e-01 8.46929073e-01 -1.07178867e+00
3.11879784e-01 5.29328108e-01 -3.31525924e-03 -1.00869989e+00
-7.09275067e-01 -4.79624301e-01 -4.14234549e-01 -3.87620270e-01
8.90639544e-01 -1.94374129e-01 -9.42934930e-01 1.89778829e+00
1.07240528e-01 -1.58796489e-01 -4.22867872e-02 5.11408567e-01
1.20960927e+00 1.08162355e+00 1.73991874e-01 -1.28951997e-01
1.38056266e+00 -7.67984986e-01 -6.85546279e-01 -2.89880395e-01
5.71721792e-01 -6.29742980e-01 1.42949688e+00 4.30258900e-01
-1.00718713e+00 -5.77640891e-01 -9.10845637e-01 -4.81208056e-01
-2.60593295e-01 -3.18293512e-01 3.29223871e-01 2.98215359e-01
-1.09355485e+00 1.08696091e+00 -6.37566745e-01 -5.96553922e-01
-1.08930424e-01 7.09677711e-02 2.25677699e-01 4.13506210e-01
-1.69174027e+00 1.16401422e+00 5.25445521e-01 -2.63629645e-01
-1.14861071e+00 -7.20894516e-01 -7.69303322e-01 -8.01025480e-02
3.69330794e-01 -5.88363290e-01 1.88909018e+00 -4.10604715e-01
-1.54903102e+00 7.90853262e-01 1.73645183e-01 -7.58647561e-01
4.55765873e-01 -1.06695697e-01 -4.03709263e-01 4.10636455e-01
2.27547854e-01 6.31307960e-01 6.33513272e-01 -4.41133618e-01
-3.44854534e-01 -4.00297828e-02 3.86893004e-01 -1.34774595e-01
-3.01906139e-01 -5.08313924e-02 9.75973606e-02 -8.74437749e-01
-3.04540008e-01 -6.90601110e-01 -2.57397175e-01 -3.31429601e-01
-3.22813928e-01 -9.35860336e-01 6.70483351e-01 -1.03439128e+00
1.44726408e+00 -1.40964806e+00 3.50557938e-02 -2.40996853e-01
-2.85283960e-02 8.43510926e-02 -2.35163808e-01 6.48273587e-01
1.06099844e-01 2.29848653e-01 -6.10516518e-02 7.27762654e-02
1.94723323e-01 3.02744806e-01 -4.94051814e-01 8.55388343e-02
-8.36654827e-02 1.29695964e+00 -1.33326447e+00 -6.69739485e-01
-9.20438543e-02 -1.62951812e-01 -7.55935848e-01 2.67669439e-01
-1.05927157e+00 5.51509321e-01 -3.57680798e-01 1.48318708e-01
-1.40347391e-01 -4.89648789e-01 1.32631063e-01 2.12048758e-02
2.74192709e-02 9.89112079e-01 -2.52328783e-01 1.74888599e+00
-5.50732255e-01 9.31172252e-01 -7.49944627e-01 -5.27312636e-01
7.95985043e-01 5.77472210e-01 2.16843337e-01 -1.11828792e+00
2.21486419e-01 -9.61119309e-03 2.29080230e-01 -8.02641332e-01
8.22918415e-01 -6.85809731e-01 -4.43343550e-01 9.68164921e-01
3.10365528e-01 -4.32094753e-01 5.10235012e-01 4.16230470e-01
1.18726397e+00 3.75637978e-01 2.94871956e-01 -3.20989877e-01
4.73332852e-01 3.66849720e-01 3.26029599e-01 7.29049265e-01
-2.19416842e-01 3.95468026e-01 5.78308165e-01 -6.49681151e-01
-1.22968626e+00 -1.11953735e+00 1.85732692e-01 1.43979192e+00
-1.91148162e-01 -8.42330217e-01 -4.86398816e-01 -9.21793878e-01
-2.94860929e-01 1.22098768e+00 -8.74709368e-01 -2.11365208e-01
-1.02864742e+00 -5.23501277e-01 6.54381096e-01 5.31346381e-01
3.89682680e-01 -1.60539496e+00 -7.46404767e-01 5.98922312e-01
-5.61925650e-01 -1.01428699e+00 -6.21833324e-01 -1.32464662e-01
-8.06088388e-01 -9.53696549e-01 -6.06141031e-01 -6.10987782e-01
-1.51508719e-01 -4.19200569e-01 1.48774529e+00 -3.21012251e-02
7.30774924e-02 3.17166805e-01 -5.37441909e-01 -1.72969431e-01
-8.10757875e-01 5.24994373e-01 -1.92622125e-01 -5.49538195e-01
1.07753992e-01 -8.72642279e-01 -5.75775981e-01 2.48597562e-01
-9.61992204e-01 -1.63452044e-01 5.43955863e-02 5.80067337e-01
-3.41024697e-01 -1.25329539e-01 9.58401501e-01 -7.48585880e-01
8.44948351e-01 -5.63007712e-01 -4.88503426e-01 1.20789558e-01
-6.63473368e-01 5.37086308e-01 8.00554156e-01 -6.18879378e-01
-1.03756416e+00 -7.72074521e-01 -4.53540057e-01 6.67071119e-02
3.25413495e-01 5.65029502e-01 2.23857909e-01 5.35111785e-01
1.37792456e+00 1.33018643e-01 -4.15765673e-01 -4.71526086e-01
6.58475935e-01 6.75665215e-02 6.98018849e-01 -9.03910637e-01
1.05733597e+00 2.79949725e-01 -7.86792785e-02 -6.07121587e-01
-1.31495762e+00 -1.25847161e-01 -3.64944994e-01 -3.04219186e-01
1.09300828e+00 -3.79235268e-01 -7.25724697e-01 1.15796693e-01
-1.40144181e+00 -6.51836872e-01 -5.42762399e-01 2.34584417e-02
-7.45790720e-01 3.56799841e-01 -7.72595823e-01 -5.40861249e-01
-5.57850361e-01 -6.25465453e-01 8.46820891e-01 2.15602696e-01
-7.10838675e-01 -1.44737542e+00 4.11265284e-01 2.95168579e-01
5.25713980e-01 4.88261104e-01 1.30644393e+00 -7.84729958e-01
-4.88221437e-01 1.52495652e-01 3.25916499e-01 1.56114861e-01
-4.97715373e-04 -2.80647069e-01 -8.43228102e-01 6.23821504e-02
-1.72779765e-02 -5.00690699e-01 8.39149535e-01 4.22148816e-02
6.69133782e-01 -9.96897519e-01 5.14946915e-02 1.62351146e-01
1.04643655e+00 1.27592385e-01 5.98942637e-01 4.07098949e-01
4.52540070e-01 7.35039651e-01 4.02936667e-01 5.53550780e-01
7.24231720e-01 5.76530278e-01 2.06868991e-01 5.64637721e-01
-2.28064880e-01 -5.80079734e-01 6.11484289e-01 7.44956791e-01
2.10382342e-01 -3.32548201e-01 -1.07863903e+00 1.17602885e+00
-1.79797769e+00 -1.27201593e+00 -1.33019850e-01 1.57393885e+00
1.42301059e+00 2.93612212e-01 4.91583198e-01 3.38310033e-01
2.16202691e-01 5.07669330e-01 -1.94486573e-01 -4.36844885e-01
-2.55348738e-02 6.24004006e-01 -1.39607191e-01 6.62030518e-01
-8.00709784e-01 1.16530812e+00 7.27730179e+00 8.66086245e-01
-8.69365156e-01 2.26951718e-01 3.71417344e-01 -3.29879746e-02
-6.89262569e-01 3.68797213e-01 -5.82809567e-01 1.32122487e-01
1.39383781e+00 -7.63328195e-01 3.75144392e-01 4.53449070e-01
2.43123740e-01 1.46870434e-01 -1.40468514e+00 5.85399389e-01
-1.02068409e-02 -1.72588360e+00 1.43533215e-01 -2.39831150e-01
7.86464334e-01 -2.21863374e-01 -1.18496701e-01 6.40738666e-01
7.38313079e-01 -1.11271703e+00 1.04839063e+00 4.55447733e-01
2.92145580e-01 -3.31210971e-01 3.90240699e-01 5.80869973e-01
-8.55484307e-01 -2.31992185e-01 -8.39827657e-02 -2.67033160e-01
4.05543447e-01 2.63281286e-01 -1.40550244e+00 4.87033725e-01
3.96571845e-01 5.82410216e-01 -1.05582428e+00 8.28669190e-01
-8.15345943e-01 1.06093395e+00 -1.56492695e-01 -3.63211989e-01
4.51631367e-01 2.37151951e-01 7.11524904e-01 1.09939611e+00
1.79730907e-01 -1.81685691e-03 -1.21790200e-01 9.10674214e-01
7.18162209e-02 1.12414129e-01 -4.01001632e-01 -3.51637870e-01
1.41251191e-01 1.03445733e+00 -5.29585660e-01 -3.78285199e-01
1.29367784e-01 5.98483145e-01 9.12857279e-02 1.23588361e-01
-7.01073468e-01 -1.56452984e-01 1.24058174e-02 5.16838372e-01
2.84871787e-01 -5.09869933e-01 -5.69199584e-02 -1.05059934e+00
-2.34327048e-01 -7.56295621e-01 9.51558292e-01 -9.87351179e-01
-1.38565707e+00 7.14993536e-01 3.73808652e-01 -9.02049065e-01
-9.02791321e-01 -3.84804666e-01 -8.28229666e-01 5.32107532e-01
-1.03347135e+00 -1.21134257e+00 8.28774571e-02 5.52743554e-01
5.24149120e-01 4.41418253e-02 4.64511037e-01 -9.96757671e-03
2.30032280e-02 6.02863312e-01 -4.86990511e-01 2.53931135e-01
4.27889705e-01 -1.49548209e+00 9.06751394e-01 9.21386421e-01
4.09422100e-01 5.50842464e-01 1.15738881e+00 -5.98516107e-01
-9.35372055e-01 -1.06885719e+00 1.51602578e+00 -9.22950089e-01
1.12311697e+00 -2.36760065e-01 -7.70112753e-01 6.69406414e-01
7.47574270e-01 -6.86207294e-01 3.72621268e-01 5.93001880e-02
-4.63295430e-01 1.86016578e-02 -9.14452732e-01 6.98573411e-01
9.78011191e-01 -7.26812124e-01 -1.34714139e+00 7.20016181e-01
1.19493151e+00 -1.02691211e-01 -1.03463829e+00 7.91663006e-02
3.98075223e-01 -6.77739501e-01 7.03894198e-01 -8.87414515e-01
8.01597834e-01 -1.32806838e-01 7.18684196e-02 -1.13337326e+00
-3.04090291e-01 -1.13497055e+00 -1.54146716e-01 1.07878935e+00
4.88673896e-01 -2.49355897e-01 7.33411193e-01 9.64646414e-02
-7.85309970e-02 -2.50854611e-01 -1.14800620e+00 -9.89309728e-01
7.78005540e-01 -5.66098750e-01 4.25919741e-01 7.08898902e-01
1.07831337e-01 1.00817883e+00 -3.30564827e-01 -4.20090079e-01
2.72614032e-01 1.91895977e-01 5.97624481e-01 -1.12850392e+00
-5.68503618e-01 -6.07719421e-01 -5.68248928e-02 -9.79706407e-01
1.10010110e-01 -1.04992843e+00 -9.85059328e-03 -1.81763542e+00
-3.28170627e-01 1.40727591e-02 1.33558467e-01 5.69910228e-01
-2.20709845e-01 2.69657552e-01 2.43577316e-01 -6.89724982e-02
-5.24370968e-01 8.35099101e-01 1.61745715e+00 -1.14363529e-01
-9.61050838e-02 -5.87387197e-02 -7.34368145e-01 3.48397970e-01
7.48848736e-01 -5.88361144e-01 -5.09453356e-01 -3.35333049e-01
9.06330466e-01 4.64945763e-01 3.66601855e-01 -8.88826191e-01
1.43354982e-01 -3.85745972e-01 1.68097213e-01 -7.38075733e-01
7.41347373e-02 -1.36023566e-01 -1.61417291e-01 6.35883570e-01
-6.12399459e-01 4.13550019e-01 1.98155984e-01 3.21704298e-01
-1.00696357e-02 -3.12171280e-01 6.49840117e-01 -4.81658071e-01
-4.63506579e-01 2.44487509e-01 -6.14093006e-01 6.33609474e-01
5.01821458e-01 1.39552519e-01 -5.25761724e-01 -8.53407919e-01
-8.47268760e-01 3.46836001e-01 -5.57474494e-02 6.04568601e-01
4.26164746e-01 -1.31713688e+00 -9.87475574e-01 -4.99830693e-01
1.93943884e-02 -2.70893335e-01 1.47124473e-02 6.43431187e-01
-5.47922730e-01 6.37018144e-01 4.41511832e-02 -3.13702196e-01
-6.31008804e-01 7.00751722e-01 3.92576665e-01 -8.52588773e-01
-6.80727541e-01 8.17000747e-01 -5.43111973e-02 -5.09969771e-01
-2.28866767e-02 -7.03801215e-01 -4.36764181e-01 3.40698808e-01
4.12299901e-01 3.00418943e-01 -8.19444563e-03 -2.49597535e-01
-1.26593694e-01 3.50387275e-01 -9.61135998e-02 -7.27328002e-01
1.41363990e+00 1.98845133e-01 8.77320841e-02 6.36220336e-01
8.81482184e-01 -8.59546810e-02 -8.88860464e-01 -5.17019808e-01
4.88661975e-01 1.54892042e-01 -3.95463437e-01 -1.02478921e+00
-5.57734311e-01 8.96443129e-01 -4.80553322e-02 6.54602349e-01
9.37556863e-01 2.24386156e-01 1.14698434e+00 4.49426502e-01
2.48373121e-01 -9.77248967e-01 8.98035526e-01 1.05900180e+00
1.27635491e+00 -5.82602322e-01 -4.63340506e-02 3.25061888e-01
-4.72076327e-01 1.11507666e+00 6.16117179e-01 -2.29900286e-01
5.29702902e-01 -2.71933407e-01 9.36813876e-02 -3.42677027e-01
-1.24255991e+00 -1.57591969e-01 4.63977635e-01 3.08195859e-01
4.21802908e-01 -2.24239603e-01 -5.29645920e-01 5.36248684e-01
-1.06873250e+00 -1.41925409e-01 7.63402581e-01 7.71471024e-01
-5.25363088e-01 -1.26350892e+00 -3.15996706e-01 1.72808558e-01
-4.40715522e-01 -3.21187556e-01 -5.51149309e-01 7.56029427e-01
9.89330038e-02 1.04703665e+00 -7.69750997e-02 -5.64413846e-01
5.82535341e-02 2.57082254e-01 8.59864414e-01 -7.68243372e-01
-1.27120101e+00 -2.12541565e-01 4.98520344e-01 -3.22070986e-01
-1.48031697e-01 -6.01405442e-01 -1.16426802e+00 -4.03779238e-01
1.88521326e-01 4.64822114e-01 2.85176396e-01 1.33279240e+00
-1.49321184e-01 4.78592634e-01 3.77286911e-01 -3.04896653e-01
-8.41042042e-01 -1.42218912e+00 -1.21664971e-01 4.06896323e-01
3.67452115e-01 -3.07060808e-01 -3.42935562e-01 1.72107682e-01] | [11.29973316192627, 8.770824432373047] |
c7ae1d71-04f9-49ff-9d73-33792f3bfe56 | argan-attentive-recurrent-generative | 1908.01323 | null | https://arxiv.org/abs/1908.01323v1 | https://arxiv.org/pdf/1908.01323v1.pdf | ARGAN: Attentive Recurrent Generative Adversarial Network for Shadow Detection and Removal | In this paper we propose an attentive recurrent generative adversarial network (ARGAN) to detect and remove shadows in an image. The generator consists of multiple progressive steps. At each step a shadow attention detector is firstly exploited to generate an attention map which specifies shadow regions in the input image.Given the attention map, a negative residual by a shadow remover encoder will recover a shadow-lighter or even a shadow-free image. A discriminator is designed to classify whether the output image in the last progressive step is real or fake. Moreover, ARGAN is suitable to be trained with a semi-supervised strategy to make full use of sufficient unsupervised data. The experiments on four public datasets have demonstrated that our ARGAN is robust to detect both simple and complex shadows and to produce more realistic shadow removal results. It outperforms the state-of-the-art methods, especially in detail of recovering shadow areas. | ['Chengjiang Long', 'Chunxia Xiao', 'Ling Zhang', 'Bin Ding'] | 2019-08-04 | argan-attentive-recurrent-generative-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Ding_ARGAN_Attentive_Recurrent_Generative_Adversarial_Network_for_Shadow_Detection_and_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Ding_ARGAN_Attentive_Recurrent_Generative_Adversarial_Network_for_Shadow_Detection_and_ICCV_2019_paper.pdf | iccv-2019-10 | ['shadow-removal', 'shadow-detection-and-removal', 'shadow-detection'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 9.07640755e-01 3.59552532e-01 3.54049474e-01 -2.94473201e-01
-6.26713336e-01 -3.11596245e-01 5.29494762e-01 -8.31558704e-01
-2.99781412e-02 9.73049223e-01 7.12504312e-02 -3.91701609e-01
6.20326698e-01 -8.77003014e-01 -7.49560714e-01 -1.06944227e+00
3.53513122e-01 2.46366024e-01 5.10500491e-01 -2.80249327e-01
6.35120347e-02 5.81893563e-01 -1.35030794e+00 1.85275912e-01
9.95995164e-01 8.47682178e-01 7.71663129e-01 7.82454431e-01
1.82352230e-01 9.55335200e-01 -1.05221379e+00 -1.64899409e-01
3.82210106e-01 -9.47314918e-01 -2.09317848e-01 1.13697194e-01
1.67699113e-01 -6.19771004e-01 -3.59621227e-01 9.31958735e-01
6.47099793e-01 5.48159294e-02 6.49376988e-01 -1.25322258e+00
-7.04067588e-01 1.23136267e-01 -3.71335357e-01 6.67466670e-02
2.15005010e-01 3.98705512e-01 5.25548935e-01 -1.03522611e+00
5.76135278e-01 1.12488663e+00 4.44295585e-01 4.68693793e-01
-9.23512757e-01 -7.47856557e-01 -7.66122043e-02 2.09824312e-02
-1.16713464e+00 -4.23200965e-01 1.04160881e+00 8.20714533e-02
2.75587767e-01 4.43816036e-01 5.70598483e-01 1.41763616e+00
5.35164773e-01 8.86935651e-01 1.43280590e+00 -4.43203151e-01
2.67142355e-01 2.24823281e-01 -3.78686398e-01 8.80629241e-01
1.19976290e-01 2.74956256e-01 -3.28224689e-01 4.95996065e-02
6.58906937e-01 8.15417767e-02 -5.87176919e-01 -1.28803670e-01
-8.62850785e-01 6.99133694e-01 8.84378552e-01 7.04145506e-02
-5.56210935e-01 6.05718754e-02 -8.86865854e-02 3.08873523e-02
4.37283516e-01 3.52438480e-01 1.57513708e-01 6.33707941e-01
-1.01619291e+00 -8.26841146e-02 6.00187242e-01 7.17641175e-01
8.86185050e-01 4.77855891e-01 -5.22350729e-01 4.87828493e-01
2.00423524e-01 1.12773013e+00 3.41182888e-01 -7.12902606e-01
2.59948313e-01 3.67772758e-01 2.77112722e-01 -9.77278352e-01
3.41765955e-02 -3.33904266e-01 -8.40107560e-01 7.96218872e-01
-4.43757884e-02 -2.45794609e-01 -1.50553548e+00 1.40578163e+00
1.40368342e-01 4.28718746e-01 3.48130584e-01 1.04544544e+00
1.00732172e+00 9.06139553e-01 -2.17544436e-01 -1.86086476e-01
1.15216815e+00 -1.03199244e+00 -8.70318890e-01 -7.11286128e-01
-3.01655293e-01 -7.90044367e-01 1.08805597e+00 1.31679311e-01
-7.44617343e-01 -6.97581768e-01 -1.30906010e+00 5.82939275e-02
-3.48829240e-01 3.52702439e-01 4.88327920e-01 6.75942957e-01
-8.37381184e-01 2.03819975e-01 -4.59762156e-01 -1.08359963e-01
4.44462568e-01 6.14660643e-02 -1.16999121e-02 -8.97085145e-02
-1.18696821e+00 7.37661362e-01 2.35112265e-01 5.53667903e-01
-1.55388796e+00 -2.68670410e-01 -9.63283598e-01 9.67216566e-02
4.36199188e-01 -4.11995828e-01 8.79165888e-01 -1.61650085e+00
-1.56186306e+00 6.73882842e-01 -3.76203686e-01 -5.31740904e-01
7.04406917e-01 -1.91573426e-01 -4.11383063e-01 1.06336109e-01
1.84995219e-01 4.00093287e-01 1.48156404e+00 -1.92974365e+00
-2.91801900e-01 4.21154872e-02 -1.29412608e-02 4.53960955e-01
1.38254449e-01 -1.41723916e-01 -4.47968274e-01 -7.57850707e-01
-9.43331867e-02 -1.05031455e+00 -1.84375763e-01 -1.66532919e-01
-9.86563563e-01 4.55207646e-01 1.31738877e+00 -7.55747974e-01
8.97314608e-01 -2.03130031e+00 -1.35680795e-01 2.89504588e-01
4.22746018e-02 5.55803716e-01 -9.27916095e-02 3.13362122e-01
1.44146994e-01 -1.95019215e-01 -8.02998364e-01 -5.84106088e-01
-2.72492409e-01 3.36362749e-01 -9.22746956e-01 5.62620699e-01
2.30387598e-01 1.00917614e+00 -8.18030238e-01 -4.51552331e-01
3.10847819e-01 6.54532552e-01 1.79768428e-01 7.29373574e-01
-2.14621022e-01 4.46143568e-01 -4.58003461e-01 8.63197386e-01
9.70396221e-01 -1.80313274e-01 -1.93767902e-03 5.09623699e-02
2.30411310e-02 -6.90661073e-02 -7.86227465e-01 1.06338906e+00
-4.83779371e-01 1.13530850e+00 -3.21759656e-02 -3.25260937e-01
1.02515173e+00 3.60051133e-02 -2.55186617e-01 -7.07604170e-01
1.65312156e-01 1.28141031e-01 -2.01176345e-01 -4.43288445e-01
6.01917088e-01 -9.93193537e-02 -1.00304164e-01 5.29092312e-01
-4.03842956e-01 -2.31421947e-01 -3.69971782e-01 2.24130854e-01
1.06495297e+00 2.17296481e-01 -1.83031633e-02 5.75241186e-02
6.58902884e-01 -3.36097896e-01 6.47410095e-01 9.13056552e-01
-4.52184565e-02 1.14304566e+00 3.44015479e-01 -1.65730551e-01
-8.00707519e-01 -1.13487506e+00 1.99105069e-01 1.05109632e+00
5.70466876e-01 2.55240291e-01 -8.44771028e-01 -9.50857401e-01
-1.21695593e-01 9.89761412e-01 -8.74643803e-01 -2.43211165e-01
-5.33804953e-01 -5.87190866e-01 4.70111191e-01 3.39931071e-01
9.51307476e-01 -1.94257855e+00 -1.02442372e+00 -1.93069130e-01
-1.87999398e-01 -9.52230155e-01 -3.84225368e-01 3.26975286e-01
-2.47315928e-01 -1.00489128e+00 -9.09412682e-01 -7.36897111e-01
9.76963878e-01 5.11579633e-01 8.87106121e-01 1.74008355e-01
-3.92748505e-01 -2.86125481e-01 -3.61791968e-01 -6.47695541e-01
-6.90657854e-01 -2.37751350e-01 -4.53257680e-01 6.04789317e-01
-2.83895910e-01 -3.50719810e-01 -9.26778018e-01 2.38171607e-01
-1.05539477e+00 2.76637107e-01 1.13222206e+00 7.30472445e-01
3.62542152e-01 6.70412928e-02 3.37335110e-01 -1.34351170e+00
5.80357611e-01 -3.90357316e-01 -5.24477124e-01 2.80866116e-01
-6.94450855e-01 1.57396764e-01 8.82698715e-01 -2.89192766e-01
-1.62778711e+00 1.74688607e-01 1.45395845e-01 -4.69285727e-01
-1.25955030e-01 -1.37910098e-01 -4.01477844e-01 -8.61887708e-02
6.79388583e-01 5.61165571e-01 -2.84316987e-01 -1.71070565e-02
2.47031003e-01 5.74860990e-01 7.35100865e-01 6.24344535e-02
1.37856519e+00 8.19589913e-01 -1.12986162e-01 -6.86618924e-01
-9.25182879e-01 -4.60067503e-02 -4.07847792e-01 -3.13431472e-01
8.64454448e-01 -6.01701736e-01 -1.54342547e-01 7.90129125e-01
-1.07746530e+00 -8.28212798e-01 -2.08062619e-01 -2.11926863e-01
-2.75042087e-01 2.97836244e-01 -2.32929975e-01 -1.03925622e+00
-5.44662297e-01 -1.08198154e+00 1.25545776e+00 6.03631318e-01
4.17400062e-01 -7.08410442e-01 -6.12535216e-02 2.53179461e-01
4.94158596e-01 5.23737133e-01 5.14685690e-01 -1.41669303e-01
-9.28224444e-01 -7.46586099e-02 -3.82380038e-01 6.20689332e-01
2.50573456e-01 -1.65843934e-01 -1.42147529e+00 -2.40775034e-01
-1.08461371e-02 -3.04798454e-01 1.15875566e+00 1.49166971e-01
1.15885794e+00 -4.59135026e-01 -3.51094395e-01 7.07721591e-01
1.51576900e+00 2.85683364e-01 1.24841380e+00 1.37963384e-01
8.27109575e-01 2.00702012e-01 8.77607942e-01 7.90211409e-02
-6.33060485e-02 2.95561373e-01 7.44594216e-01 -8.56126368e-01
-5.92009366e-01 -3.20671916e-01 5.68428278e-01 -1.00622073e-01
-1.17180198e-01 -8.87800395e-01 -4.33685064e-01 3.52854162e-01
-1.53019178e+00 -1.12025285e+00 -5.85936457e-02 2.13249946e+00
5.46090662e-01 2.80258417e-01 -5.02937317e-01 1.34599045e-01
6.92838430e-01 8.64270270e-01 -7.49204218e-01 -2.09886536e-01
-4.36211616e-01 3.45542163e-01 7.83166647e-01 6.38565421e-01
-8.73414874e-01 1.43958342e+00 5.65768385e+00 6.82478964e-01
-1.17607915e+00 -1.03098124e-01 5.78720450e-01 4.23339278e-01
-5.67204118e-01 1.09476201e-01 -4.08728719e-01 6.96144342e-01
5.64556479e-01 4.32436168e-01 4.79348660e-01 6.88323796e-01
2.19363302e-01 -5.04706502e-01 -3.23724419e-01 5.85266769e-01
5.88998854e-01 -9.63948011e-01 -5.33548743e-02 1.15130190e-02
9.26467717e-01 -5.11475056e-02 2.05503285e-01 2.86526263e-01
3.22816402e-01 -1.14974189e+00 6.72170997e-01 8.89126003e-01
1.04970396e+00 -7.58713901e-01 9.41588521e-01 2.91928560e-01
-1.04012311e+00 -9.47251637e-03 -2.58942813e-01 2.86995679e-01
1.02269799e-01 5.47871411e-01 -1.39454305e+00 4.55536067e-01
4.78609502e-01 1.14477158e-01 -6.83176816e-01 6.41334176e-01
-1.04366541e+00 6.41774178e-01 9.20365527e-02 5.35250530e-02
8.60714167e-02 -2.18990609e-01 5.75558782e-01 1.35026944e+00
1.27921119e-01 1.26289606e-01 -2.67292038e-02 8.81841004e-01
-1.54846027e-01 -4.15672302e-01 -7.35808015e-01 2.47752279e-01
3.45749795e-01 1.38092816e+00 -9.41901088e-01 -3.88554901e-01
1.08329006e-01 1.91666746e+00 -1.03386246e-01 7.53689408e-01
-1.22092211e+00 -6.80650651e-01 2.30558723e-01 7.14347214e-02
4.29729044e-01 1.57841727e-01 -1.45069838e-01 -7.48783588e-01
-3.82044613e-02 -6.99338377e-01 -3.55361290e-02 -1.50071073e+00
-7.53321886e-01 9.77493465e-01 -4.51683611e-01 -1.10941696e+00
-2.44373202e-01 -2.81750530e-01 -1.12392175e+00 1.03429043e+00
-1.89408886e+00 -1.32333684e+00 -1.05183804e+00 6.33979380e-01
7.94553876e-01 -9.32928994e-02 7.48938262e-01 -1.48405775e-01
-4.77506876e-01 4.15448308e-01 -5.71597517e-02 2.08933190e-01
7.23449886e-01 -1.33880866e+00 4.37831759e-01 1.38859951e+00
-1.15471222e-01 1.42566592e-01 9.26759601e-01 -8.66152942e-01
-1.13617980e+00 -1.31894994e+00 6.77874148e-01 -2.97450006e-01
6.12499528e-02 -5.14014065e-01 -9.05409515e-01 5.35614908e-01
3.73561412e-01 1.35988444e-01 1.34796232e-01 -6.56535864e-01
-2.04814494e-01 -3.04466009e-01 -1.30577278e+00 5.59685588e-01
8.10120642e-01 -5.64002335e-01 -4.12996024e-01 4.20185745e-01
6.49445176e-01 -6.75333619e-01 7.98481628e-02 4.74568695e-01
3.87870789e-01 -1.39369714e+00 8.30732882e-01 1.13574058e-01
4.54810351e-01 -5.40354013e-01 4.20592353e-02 -1.20273578e+00
-1.17802046e-01 -8.40671897e-01 -1.13316730e-01 1.10618854e+00
2.35241577e-01 -7.21274316e-01 7.29504466e-01 -2.22143554e-03
-3.84096950e-01 -6.87553763e-01 -4.32539910e-01 -4.12290275e-01
-7.06820607e-01 -9.49547440e-02 5.49982965e-01 3.74959648e-01
-1.00373304e+00 2.35613868e-01 -7.60518551e-01 6.39089942e-01
6.36992991e-01 6.38785660e-01 1.02643478e+00 -7.93340266e-01
-3.46637756e-01 -2.63741743e-02 2.01750938e-02 -8.18828940e-01
3.36973071e-01 -5.04396260e-01 6.30479932e-01 -1.65320182e+00
3.17568779e-01 -3.96622926e-01 -2.59780139e-01 5.62809408e-01
-4.60801333e-01 7.27505744e-01 1.80545151e-01 3.28857958e-01
-4.31905121e-01 7.72104025e-01 1.32674897e+00 -5.10420091e-03
-3.35137069e-01 4.79587108e-01 -5.63256085e-01 6.23532534e-01
8.48700225e-01 -5.19065619e-01 -5.14384925e-01 -6.26894757e-02
-2.74870962e-01 7.46291205e-02 7.76202440e-01 -9.72954273e-01
-1.33517087e-01 -2.32216805e-01 7.76734531e-01 -6.64294362e-01
5.98648787e-01 -6.81963325e-01 1.09396562e-01 5.80259204e-01
-1.02150075e-01 -3.86859864e-01 -1.46563321e-01 7.29102850e-01
1.04698077e-01 -5.70459254e-02 9.50031877e-01 -6.82033747e-02
-7.07731009e-01 1.27534673e-01 -2.36521363e-01 -2.13091597e-01
1.04188156e+00 -1.62820563e-01 -3.95417899e-01 -6.60849631e-01
-2.33802304e-01 -3.41132320e-02 5.67444384e-01 2.79723734e-01
9.43435848e-01 -9.94288445e-01 -6.98856652e-01 5.26565075e-01
-6.48558438e-02 3.57170962e-02 1.97135121e-01 2.44300589e-01
-6.37126565e-01 1.07829701e-02 -1.86029613e-01 -2.58207470e-01
-1.34742379e+00 4.41538662e-01 4.05179024e-01 -1.49853468e-01
-8.31944525e-01 6.75978124e-01 5.99405646e-01 3.71380933e-02
2.63433486e-01 -1.18364818e-01 -7.22740171e-03 -3.29113692e-01
3.39003652e-01 1.26996130e-01 -1.31924048e-01 -6.87425971e-01
-1.60328686e-01 1.88058048e-01 4.23489481e-01 -3.14688653e-01
1.06489170e+00 -3.26207168e-02 6.24690019e-03 3.15494359e-01
9.49816465e-01 4.44109857e-01 -1.87166703e+00 -1.96222253e-02
-6.59951150e-01 -6.06129408e-01 3.39118391e-03 -1.11209846e+00
-1.24283767e+00 5.06818771e-01 7.62495458e-01 2.45547801e-01
1.51876080e+00 -3.80734950e-02 9.45563853e-01 9.25086364e-02
1.40254602e-01 -7.82289803e-01 2.50206172e-01 1.78169131e-01
1.27326822e+00 -1.31514692e+00 4.15753275e-02 -2.76653379e-01
-9.82596636e-01 8.82272720e-01 4.60351855e-01 -4.28008527e-01
2.41292641e-01 3.05818588e-01 2.01269418e-01 -1.32592008e-01
-1.65849224e-01 -4.44132030e-01 3.63380402e-01 7.29757488e-01
-1.87405393e-01 5.60959317e-02 1.16566218e-01 1.07527167e-01
-2.24627733e-01 -4.00748342e-01 5.90550244e-01 7.33221292e-01
-6.26387000e-01 -8.13404858e-01 -6.14601374e-01 7.30196312e-02
-1.02601349e-01 -1.52031586e-01 -8.19405019e-01 8.09918821e-01
3.03642541e-01 1.00327742e+00 -1.56138167e-01 -3.63399148e-01
7.14904144e-02 -1.51769355e-01 2.13631049e-01 -6.06527269e-01
-3.30985725e-01 -5.89109957e-02 -1.02170244e-01 -6.87014043e-01
-9.37112495e-02 -4.28954244e-01 -1.41609967e+00 1.19901702e-01
-3.21541905e-01 -4.65030316e-03 6.82340860e-01 7.37989604e-01
2.16956913e-01 7.90329576e-01 1.05772173e+00 -1.09628737e+00
-2.22334042e-02 -9.74128544e-01 -4.44823235e-01 2.31770337e-01
7.94865370e-01 -4.76559013e-01 -6.15641773e-01 2.43786070e-02] | [10.845498085021973, -4.103166103363037] |
34e1dd8e-c018-4be2-9f4f-14af2be1b21e | memorization-capacity-of-neural-networks-with | 2303.11247 | null | https://arxiv.org/abs/2303.11247v1 | https://arxiv.org/pdf/2303.11247v1.pdf | Memorization Capacity of Neural Networks with Conditional Computation | Many empirical studies have demonstrated the performance benefits of conditional computation in neural networks, including reduced inference time and power consumption. We study the fundamental limits of neural conditional computation from the perspective of memorization capacity. For Rectified Linear Unit (ReLU) networks without conditional computation, it is known that memorizing a collection of $n$ input-output relationships can be accomplished via a neural network with $O(\sqrt{n})$ neurons. Calculating the output of this neural network can be accomplished using $O(\sqrt{n})$ elementary arithmetic operations of additions, multiplications and comparisons for each input. Using a conditional ReLU network, we show that the same task can be accomplished using only $O(\log n)$ operations per input. This represents an almost exponential improvement as compared to networks without conditional computation. We also show that the $\Theta(\log n)$ rate is the best possible. Our achievability result utilizes a general methodology to synthesize a conditional network out of an unconditional network in a computationally-efficient manner, bridging the gap between unconditional and conditional architectures. | ['Erdem Koyuncu'] | 2023-03-20 | null | null | null | null | ['memorization'] | ['natural-language-processing'] | [ 7.91292131e-01 2.82738984e-01 -1.10508375e-01 -4.66404051e-01
-6.51102841e-01 -4.37172055e-01 1.92341939e-01 2.52011478e-01
-1.08664548e+00 9.18025911e-01 -4.48801607e-01 -9.73722160e-01
1.08647346e-01 -1.13835359e+00 -1.20653403e+00 -7.94945002e-01
-5.77855587e-01 -6.72088116e-02 2.06398070e-01 -1.96217358e-01
3.47722560e-01 3.73063922e-01 -1.40716636e+00 7.12866941e-03
3.93423885e-01 1.37903500e+00 3.52399908e-02 1.08404851e+00
1.50153846e-01 8.59007835e-01 -5.80215096e-01 -4.04694408e-01
3.84667695e-01 -4.95916456e-01 -7.33866572e-01 -8.30130041e-01
3.10220152e-01 -3.79515558e-01 -5.28084457e-01 1.21394348e+00
3.07852566e-01 1.61540598e-01 4.79801774e-01 -7.29346454e-01
-5.70224881e-01 1.23719180e+00 -3.43003243e-01 2.24086031e-01
3.69762555e-02 -1.90005198e-01 1.15649295e+00 -5.93283117e-01
1.38326541e-01 9.12959874e-01 5.61234415e-01 4.67325807e-01
-1.49570584e+00 -1.19270301e+00 -6.75803516e-03 -1.67582542e-01
-1.91365552e+00 -5.65367460e-01 3.39475662e-01 2.24268109e-01
1.41604984e+00 1.67030871e-01 5.32207906e-01 1.34968579e-01
4.45343137e-01 5.08686125e-01 6.80622578e-01 -7.86215961e-01
3.44508380e-01 -1.58973992e-01 2.55119354e-01 1.04932106e+00
5.37003577e-01 -1.66512340e-01 -6.59619689e-01 2.72122711e-01
9.78203595e-01 -7.40783960e-02 -3.12878639e-01 3.17327321e-01
-9.69472229e-01 6.62248850e-01 6.56468451e-01 3.62600297e-01
-1.36499833e-02 1.19443476e+00 2.28200540e-01 6.55212164e-01
-1.58894304e-02 1.11563668e-01 -4.03696835e-01 1.21779464e-01
-1.14157152e+00 -9.92877185e-02 8.00407290e-01 1.16522980e+00
9.92277324e-01 5.61640561e-01 2.36953750e-01 2.15533271e-01
-2.97362753e-03 7.93299079e-01 2.02365726e-01 -8.04790914e-01
6.29779637e-01 2.30973467e-01 -3.00935119e-01 -6.53687298e-01
-3.57832730e-01 -4.42091793e-01 -1.45842838e+00 1.89033344e-01
3.85478854e-01 -6.21649086e-01 -7.80828714e-01 2.25843906e+00
-5.43997228e-01 -9.12293941e-02 1.52259797e-01 2.74004608e-01
2.84119040e-01 9.48920131e-01 -1.01272158e-01 -3.57130140e-01
1.17464995e+00 -3.16945225e-01 -4.55418378e-01 -2.17689320e-01
9.61850107e-01 -4.63412642e-01 5.96775651e-01 3.88729095e-01
-1.78109121e+00 -4.37206864e-01 -1.67700350e+00 -1.40515238e-01
-4.01999652e-01 9.41621885e-02 9.63709652e-01 9.96350348e-01
-1.57088435e+00 4.16390598e-01 -9.11423683e-01 4.37481850e-01
2.89890885e-01 1.06099379e+00 -2.44626001e-01 7.67795518e-02
-1.29213631e+00 6.94751501e-01 5.94422996e-01 4.15861785e-01
-6.47716165e-01 -4.23233092e-01 -9.09251451e-01 5.42280853e-01
2.83740520e-01 -3.91257048e-01 9.94995356e-01 -7.15448201e-01
-1.24300611e+00 3.80854458e-01 -4.20946628e-01 -1.11457646e+00
-1.48192286e-01 1.76804990e-01 -1.68153644e-01 1.68237194e-01
-5.15219390e-01 7.32405186e-01 5.57464123e-01 -6.20587409e-01
-5.69080353e-01 -2.80761093e-01 2.25894600e-01 -3.35126340e-01
-3.86230737e-01 -2.23317012e-01 -2.57195920e-01 -4.10735160e-01
5.28128564e-01 -8.91320944e-01 -3.03514421e-01 1.94206536e-01
-4.92141396e-01 -1.89264238e-01 7.74580464e-02 -1.16428927e-01
1.25585520e+00 -2.11021256e+00 -2.15681300e-01 6.97604001e-01
3.11747104e-01 2.19643965e-01 2.43593723e-01 -1.03843607e-01
-1.35972816e-02 1.80570707e-01 -4.13004935e-01 -1.81973904e-01
-1.53407782e-01 2.14895278e-01 -4.59276557e-01 4.88678813e-01
7.41029009e-02 9.77653086e-01 -4.14939761e-01 -3.14496279e-01
-2.98276722e-01 5.58263123e-01 -7.36546338e-01 -2.58605897e-01
4.76687541e-03 -4.44891155e-01 -1.94472019e-02 2.76804864e-01
7.21404254e-01 -3.84626508e-01 6.20536923e-01 -2.26643905e-02
1.02802325e-04 6.02202415e-01 -1.16867065e+00 1.56249416e+00
-7.94071615e-01 1.14546394e+00 1.21854916e-01 -1.19568491e+00
9.51031983e-01 3.82955551e-01 -2.49527410e-01 -8.10722947e-01
5.15248835e-01 2.85464376e-01 7.07804188e-02 3.35333049e-01
6.41119719e-01 -3.93251330e-01 -3.97583008e-01 9.09684002e-01
3.34467590e-01 1.96768001e-01 5.21294661e-02 4.68882114e-01
1.04488349e+00 -3.44995469e-01 2.19093949e-01 -4.46210295e-01
3.75509471e-01 -6.80230200e-01 5.66866398e-01 1.05279386e+00
9.03425664e-02 -3.59024256e-02 7.33724117e-01 -2.68141001e-01
-7.80625403e-01 -1.58644640e+00 -9.15247798e-02 1.25638115e+00
-2.90898774e-02 -3.97653610e-01 -7.77881563e-01 2.11395279e-01
-5.72952151e-01 6.48385167e-01 -7.20408261e-01 -2.96892047e-01
-8.76268208e-01 -6.98024273e-01 1.35547388e+00 9.51593459e-01
7.99396157e-01 -5.82139790e-01 -1.15238333e+00 6.97572008e-02
1.47142470e-01 -8.95131052e-01 -2.29990929e-01 1.14546871e+00
-1.05092883e+00 -4.26836908e-01 -4.00611490e-01 -1.17661512e+00
1.15303111e+00 3.49792317e-02 8.28676164e-01 1.45623192e-01
-1.92948788e-01 -2.09200069e-01 3.04279894e-01 -2.20802173e-01
-2.08016828e-01 1.18188560e-01 3.25710565e-01 -4.53718364e-01
3.73555362e-01 -9.48351264e-01 -6.15225434e-01 -2.40186587e-01
-1.01727772e+00 7.09439963e-02 9.15160716e-01 6.66181982e-01
5.87196946e-01 8.63363221e-02 5.63757777e-01 -8.59653413e-01
4.23112839e-01 -5.70919514e-02 -8.49007249e-01 3.06603730e-01
-5.91716945e-01 6.57118678e-01 8.24077368e-01 -3.37490290e-01
-8.92351329e-01 2.74077505e-01 -2.30220228e-01 1.64439857e-01
4.64320421e-01 5.16125619e-01 2.78119832e-01 -1.62941307e-01
8.03302050e-01 6.33880198e-01 -2.12364838e-01 9.99559611e-02
4.50235039e-01 2.17680544e-01 8.97701323e-01 -6.47867978e-01
4.84243155e-01 3.21678162e-01 6.75961018e-01 -7.14599550e-01
-3.13933283e-01 3.17434609e-01 -5.89357615e-01 1.46071255e-01
5.91700375e-01 -9.72606063e-01 -1.54372251e+00 6.48721168e-03
-1.12239909e+00 -4.32265699e-01 -2.80953050e-01 4.02239859e-01
-2.97400087e-01 1.30404457e-01 -8.02070200e-01 -1.31176019e+00
-6.48919702e-01 -9.25157666e-01 1.30688474e-01 1.51921555e-01
-1.25804111e-01 -7.16168940e-01 -7.65101731e-01 -4.04238731e-01
7.00302243e-01 -2.86969002e-02 1.30544245e+00 -2.14125857e-01
-9.04494166e-01 -4.85502452e-01 -5.03339231e-01 4.88767684e-01
-3.19319308e-01 -3.82802069e-01 -8.08597803e-01 -2.81292886e-01
-7.44857490e-02 -5.81770539e-01 1.38511026e+00 1.74412146e-01
1.28588867e+00 -5.32030702e-01 -1.68987140e-01 4.93467629e-01
1.67981696e+00 2.24555239e-01 9.06018198e-01 -5.87999046e-01
1.50525853e-01 -2.24050675e-02 -2.73485392e-01 3.58079433e-01
2.04728410e-01 -1.05189614e-01 3.37082386e-01 1.31811336e-01
1.68254554e-01 -1.71666428e-01 2.32290864e-01 8.95092487e-01
-3.44638437e-01 -1.73793018e-01 -6.83595359e-01 5.38177431e-01
-1.30761373e+00 -7.08458066e-01 1.93013951e-01 2.57080746e+00
1.14411914e+00 8.40904772e-01 -5.51960230e-01 6.21091127e-01
4.00258631e-01 -9.19564888e-02 -3.97291899e-01 -9.16616857e-01
-7.34122097e-02 1.25113916e+00 1.01550913e+00 6.11913502e-01
-4.72632080e-01 7.24271655e-01 6.80273867e+00 9.43632782e-01
-1.07000613e+00 -2.83539537e-02 9.33981001e-01 -3.50924611e-01
-2.79348284e-01 -7.70089850e-02 -7.49548197e-01 -5.18400073e-02
1.56383491e+00 1.42416760e-01 4.34965640e-01 5.87148070e-01
-5.99696457e-01 -4.83652085e-01 -1.38039589e+00 9.39735174e-01
5.82057796e-02 -1.53435636e+00 1.59208581e-01 -7.73938447e-02
6.97086751e-01 -1.69659942e-01 2.60992497e-01 1.60643056e-01
3.43461543e-01 -1.21813524e+00 4.88294244e-01 1.74274683e-01
1.26360345e+00 -1.31836367e+00 5.59920788e-01 2.71299899e-01
-1.52015078e+00 -2.75018532e-02 -5.30576348e-01 -4.52288985e-01
3.02167318e-04 6.60716891e-01 -5.46462059e-01 -1.22963777e-02
6.11703992e-01 -2.81690836e-01 -8.45448822e-02 3.96367669e-01
-3.11781764e-01 6.16737127e-01 -7.38970697e-01 -5.79179883e-01
2.20213652e-01 4.28791679e-02 -2.43877858e-01 1.53850818e+00
4.28708583e-01 3.79494429e-01 -5.89929283e-01 7.94569612e-01
-8.02081466e-01 -3.13077629e-01 -3.48294199e-01 1.91771418e-01
7.32049823e-01 6.46936297e-01 -9.32331979e-01 -5.19174099e-01
3.94021533e-03 9.14172769e-01 5.55083215e-01 1.46019235e-01
-7.01093316e-01 -1.25269997e+00 2.25673437e-01 -2.33324558e-01
5.40505469e-01 -7.68769324e-01 -8.56735289e-01 -7.21710622e-01
1.03112847e-01 6.84349984e-02 1.29102513e-01 -3.58336866e-01
-2.28849903e-01 4.92313981e-01 -1.80030242e-01 -7.30981350e-01
-3.29579324e-01 -5.30006349e-01 -3.07658017e-01 8.36829960e-01
-1.29403174e+00 -4.15050268e-01 2.93413341e-01 6.11111820e-01
-1.66192248e-01 1.38826281e-01 1.24893534e+00 1.95155278e-01
-2.33487695e-01 1.34985709e+00 -1.40086683e-02 3.97965878e-01
1.49631307e-01 -8.87759268e-01 3.84173483e-01 1.11032796e+00
6.58380091e-02 1.32306325e+00 5.12236059e-01 3.88680845e-02
-1.85547066e+00 -8.78432155e-01 1.33606362e+00 3.22497159e-01
4.27545369e-01 -7.37610281e-01 -5.19248664e-01 9.88919854e-01
2.94055730e-01 -2.46350896e-02 9.21023667e-01 7.33655272e-03
-8.59234214e-01 -3.89763266e-01 -1.05585635e+00 8.31541240e-01
1.00131452e+00 -7.18577564e-01 -2.07497165e-01 -1.16934747e-01
7.72495508e-01 -3.03901225e-01 -7.11763680e-01 7.55964741e-02
9.48733389e-01 -7.27782488e-01 9.57900226e-01 -3.48518118e-02
3.64035428e-01 -1.49035335e-01 -5.12581527e-01 -6.31130099e-01
7.65720382e-02 -7.58421123e-01 -2.84422219e-01 5.28735399e-01
1.04310489e+00 -9.40458953e-01 7.43351221e-01 6.87795579e-01
1.63901567e-01 -7.88646638e-01 -1.36638021e+00 -4.94055808e-01
3.80140334e-01 -9.58758950e-01 2.37751260e-01 4.40461904e-01
3.74255121e-01 4.49650615e-01 -4.27471906e-01 2.15029083e-02
3.41992885e-01 2.57726721e-02 3.18958938e-01 -8.82989943e-01
-5.02792001e-01 -4.80199695e-01 -5.28161943e-01 -1.79251015e+00
-1.49016157e-01 -1.00813174e+00 4.12952542e-01 -1.16870308e+00
-9.40543488e-02 -6.64827585e-01 -6.43368483e-01 6.81448758e-01
3.70373636e-01 7.66020000e-01 1.68850482e-01 -3.01596344e-01
-5.21297276e-01 4.19882797e-02 7.13335752e-01 -9.40357074e-02
7.93650001e-02 -1.44360676e-01 -8.66912067e-01 5.57041824e-01
9.36225653e-01 -4.41702515e-01 -4.80348676e-01 -7.50240266e-01
1.06065512e+00 4.06134158e-01 5.64888567e-02 -1.40232766e+00
9.54483449e-01 3.50865483e-01 4.16802794e-01 -7.59566188e-01
6.11592889e-01 -5.66670477e-01 -4.03046489e-01 1.07856858e+00
-7.86021531e-01 2.49489546e-01 2.27434859e-01 5.15062153e-01
1.71532080e-01 -3.86944979e-01 7.07255602e-01 -8.47673491e-02
-2.20372602e-01 5.59154944e-03 -8.92206132e-01 -1.80871472e-01
7.08787322e-01 -8.72313604e-02 -2.70046979e-01 -4.49031919e-01
-5.56623995e-01 -6.61278218e-02 -4.81033802e-01 -4.46077585e-01
8.97045732e-01 -8.82333398e-01 -4.30475414e-01 3.97047311e-01
-5.74267089e-01 2.70003438e-01 1.93816990e-01 4.54188675e-01
-6.75363958e-01 8.47050726e-01 -1.29318625e-01 -2.04196885e-01
-8.52531791e-01 2.29690522e-01 1.57157704e-01 -2.77203351e-01
-1.54520258e-01 1.40947759e+00 -1.92351222e-01 -1.14696147e-02
3.68439257e-01 -7.09391832e-01 5.38802147e-01 -3.67948234e-01
8.17313433e-01 3.02758455e-01 -7.33661205e-02 -1.62280411e-01
-2.97738999e-01 3.39452446e-01 -2.43672192e-01 -4.62487787e-01
9.47245121e-01 1.78356603e-01 -5.50362408e-01 3.57071996e-01
1.61062074e+00 -2.57990748e-01 -7.99901307e-01 -4.91461694e-01
-4.39208776e-01 2.56583393e-01 5.55003472e-02 -3.60972315e-01
-1.15405262e+00 1.19283497e+00 5.86237907e-01 1.38064951e-01
1.43718505e+00 -3.56776685e-01 8.13760877e-01 1.27321637e+00
7.22080708e-01 -1.19909704e+00 -1.56959042e-01 7.40978122e-01
1.70822427e-01 -6.47830725e-01 3.51151824e-02 -3.33047867e-01
1.54849246e-01 1.35089231e+00 2.92786270e-01 -5.93552589e-01
1.08059597e+00 7.61346877e-01 -6.24640346e-01 2.57406861e-01
-9.45631325e-01 1.57646641e-01 -2.60081768e-01 1.54121712e-01
5.55200636e-01 4.11859393e-01 -4.63047504e-01 4.82453376e-01
-5.66589177e-01 5.39518222e-02 6.59116149e-01 1.25256443e+00
-6.82506084e-01 -9.05994356e-01 1.13066390e-01 6.01338565e-01
-7.53487825e-01 -6.46327913e-01 3.42489600e-01 3.46960038e-01
-6.33787587e-02 9.76940572e-01 4.82968479e-01 -5.38350999e-01
-3.68201464e-01 3.21858585e-01 6.58023238e-01 -1.75317332e-01
-3.73368442e-01 -4.25040871e-01 -1.61072034e-02 -2.29258910e-01
-1.80200204e-01 -5.37743568e-02 -1.97503340e+00 -7.30714083e-01
-3.09003502e-01 2.32496615e-02 9.27992046e-01 9.37288523e-01
1.08132511e-01 7.12243438e-01 3.24234247e-01 -5.93668818e-01
-3.46639425e-01 -5.66992164e-01 -3.85422289e-01 -6.03883445e-01
4.52653497e-01 1.24654032e-01 -2.28512779e-01 6.61172196e-02] | [8.416016578674316, 3.1804044246673584] |
d9a827f9-f6d9-403f-92ce-09ddcba38c21 | universal-model-for-multi-domain-medical | 2007.08628 | null | https://arxiv.org/abs/2007.08628v1 | https://arxiv.org/pdf/2007.08628v1.pdf | Universal Model for Multi-Domain Medical Image Retrieval | Medical Image Retrieval (MIR) helps doctors quickly find similar patients' data, which can considerably aid the diagnosis process. MIR is becoming increasingly helpful due to the wide use of digital imaging modalities and the growth of the medical image repositories. However, the popularity of various digital imaging modalities in hospitals also poses several challenges to MIR. Usually, one image retrieval model is only trained to handle images from one modality or one source. When there are needs to retrieve medical images from several sources or domains, multiple retrieval models need to be maintained, which is cost ineffective. In this paper, we study an important but unexplored task: how to train one MIR model that is applicable to medical images from multiple domains? Simply fusing the training data from multiple domains cannot solve this problem because some domains become over-fit sooner when trained together using existing methods. Therefore, we propose to distill the knowledge in multiple specialist MIR models into a single multi-domain MIR model via universal embedding to solve this problem. Using skin disease, x-ray, and retina image datasets, we validate that our proposed universal model can effectively accomplish multi-domain MIR. | ['Yang Feng', 'Jiebo Luo', 'Yubao Liu'] | 2020-07-14 | null | null | null | null | ['medical-image-retrieval', 'medical-image-retrieval'] | ['computer-vision', 'medical'] | [ 1.24656014e-01 -1.93414047e-01 -2.71402836e-01 -7.29762912e-02
-1.07257652e+00 -2.64804065e-01 2.94036806e-01 4.72901911e-01
-4.63223279e-01 5.14598668e-01 2.57811785e-01 -1.45211071e-01
-3.22498709e-01 -6.22801423e-01 -2.32265502e-01 -6.27080977e-01
4.47321773e-01 5.37904620e-01 2.85806060e-01 -9.20001864e-02
-9.74408090e-02 2.72352010e-01 -1.31300437e+00 4.24866915e-01
9.78184998e-01 7.70804167e-01 5.46389818e-01 4.42856401e-01
-2.97055870e-01 8.20024729e-01 -6.01491451e-01 -3.93588275e-01
3.12628061e-01 -4.76839066e-01 -8.48774552e-01 8.44776258e-02
3.09146285e-01 -5.09082019e-01 -6.34917617e-01 1.08145511e+00
8.65995526e-01 -8.44039917e-02 7.00345218e-01 -8.50752056e-01
-1.20647442e+00 8.06559324e-02 -6.28135562e-01 2.03021735e-01
2.02272132e-01 -2.32831955e-01 6.68658435e-01 -5.52182376e-01
7.14522779e-01 1.05163872e+00 2.96805084e-01 6.89134955e-01
-7.03103840e-01 -4.87552732e-01 -1.83261305e-01 2.40611091e-01
-1.38479948e+00 -5.00631332e-02 5.53245008e-01 -3.45989615e-01
3.21907043e-01 1.93638712e-01 5.41320205e-01 7.40859866e-01
3.72429490e-02 1.01558447e+00 9.46675777e-01 -2.91320086e-01
-1.10448383e-01 4.14002091e-01 6.64284974e-02 5.31145751e-01
3.48103672e-01 -4.19230610e-01 -2.08590150e-01 -2.40398809e-01
9.69613671e-01 6.30516589e-01 -4.40250069e-01 -7.03437850e-02
-1.28397012e+00 6.84481680e-01 6.20684981e-01 6.18662536e-01
-5.09243727e-01 -2.56470710e-01 3.13913614e-01 4.17243958e-01
3.81042957e-01 5.58423936e-01 -2.19001874e-01 1.78354159e-01
-8.51315975e-01 3.10054775e-02 5.08735001e-01 7.42725015e-01
5.59882641e-01 -4.92391825e-01 6.67613670e-02 1.18106318e+00
3.46925765e-01 6.11559391e-01 1.00247419e+00 -4.68011856e-01
2.42464080e-01 9.87064242e-01 -1.88006498e-02 -1.20078540e+00
-1.95841223e-01 -3.00036997e-01 -1.06691647e+00 -4.85314608e-01
1.54604018e-01 2.12644309e-01 -9.85591769e-01 1.36716771e+00
4.27134871e-01 2.31463715e-01 3.63945931e-01 1.04115856e+00
1.19493401e+00 5.09079993e-01 3.34519558e-02 -1.10227749e-01
1.48978019e+00 -1.02213335e+00 -7.47868478e-01 -1.96954489e-01
5.66151619e-01 -1.04208672e+00 7.45210528e-01 2.00121060e-01
-8.86071563e-01 -5.24454951e-01 -9.21566904e-01 -2.22742185e-01
-3.72779518e-01 2.43525282e-01 2.67778903e-01 1.72867253e-01
-9.13043916e-01 1.89083949e-01 -5.26003897e-01 -5.44194937e-01
4.70884323e-01 2.37152293e-01 -5.87088585e-01 -8.58575106e-01
-1.33411324e+00 1.08115900e+00 1.56184152e-01 -1.26991138e-01
-6.05376542e-01 -5.37300825e-01 -6.14428759e-01 -2.79668242e-01
2.40629598e-01 -7.80196071e-01 9.38322902e-01 -9.99069989e-01
-8.88958693e-01 1.08095467e+00 1.35550192e-02 -7.65277520e-02
3.60512882e-01 -2.51174986e-01 -6.44650102e-01 6.41508818e-01
2.51992464e-01 5.90824902e-01 9.10144389e-01 -1.14644551e+00
-4.34682608e-01 -3.34980041e-01 1.76865891e-01 3.06629628e-01
-7.87668347e-01 9.34337848e-04 -9.35757697e-01 -7.25619912e-01
1.51234075e-01 -8.66978943e-01 -2.88461387e-01 2.90278375e-01
-1.24740817e-01 -1.98913157e-01 7.08724260e-01 -8.35040927e-01
1.31817281e+00 -2.43380237e+00 2.34902486e-01 -2.74177268e-02
4.03228492e-01 5.44572413e-01 -4.05850798e-01 3.07074577e-01
-1.40336936e-03 1.13811970e-01 -2.22409472e-01 -1.43598527e-01
-5.04118681e-01 4.22642142e-01 9.62648988e-02 2.31038690e-01
2.36868724e-01 8.42398942e-01 -9.75052416e-01 -9.48965788e-01
1.61409914e-01 5.55519819e-01 -3.52751404e-01 3.10899645e-01
1.73181936e-01 4.92035121e-01 -7.76243925e-01 8.19587588e-01
6.50966644e-01 -8.37351620e-01 1.44541591e-01 -3.02778363e-01
4.54110563e-01 -1.54960483e-01 -9.81143594e-01 1.80255604e+00
-4.27556992e-01 3.52759689e-01 -2.53983676e-01 -1.13730609e+00
6.42740130e-01 5.58883727e-01 8.47081959e-01 -8.79119873e-01
2.77709179e-02 4.76565123e-01 -1.23202600e-01 -8.57038975e-01
1.23227648e-01 -2.87005305e-01 7.36317933e-02 5.43039143e-01
-1.17502011e-01 5.95503720e-04 1.67543639e-03 2.54019022e-01
1.13731039e+00 -4.86210048e-01 3.40806872e-01 1.68362051e-01
4.92308825e-01 1.61725059e-01 3.91293854e-01 5.02413154e-01
-2.21820518e-01 8.56897295e-01 1.38426023e-02 -4.74792600e-01
-8.23262930e-01 -1.01575005e+00 -3.57719541e-01 5.81822395e-01
5.07053256e-01 -2.34252870e-01 -2.19297424e-01 -7.14871049e-01
6.38299212e-02 -1.69166192e-01 -5.02166212e-01 -3.82925123e-01
-4.15281653e-01 -6.94809735e-01 3.44712585e-01 3.56037021e-01
5.54972112e-01 -8.69109631e-01 -5.21866500e-01 2.10902661e-01
-3.75606328e-01 -1.04689169e+00 -4.62149203e-01 -3.54726344e-01
-8.85112047e-01 -1.27229118e+00 -1.55065012e+00 -1.05626321e+00
9.51057911e-01 6.97458088e-01 1.00010788e+00 3.86440545e-01
-7.23755240e-01 6.05331063e-01 -5.62020838e-01 -3.24575543e-01
-5.22797942e-01 -2.51946151e-02 2.48829108e-02 -1.07613631e-01
3.99669439e-01 -6.38097823e-02 -8.90173674e-01 2.22834781e-01
-1.48436487e+00 9.32028741e-02 8.18085849e-01 9.39883053e-01
6.78571522e-01 7.84567595e-02 7.59455621e-01 -8.45028639e-01
7.29894757e-01 -6.93322837e-01 1.56507362e-02 6.93005860e-01
-4.43503410e-01 -8.00220817e-02 3.11516821e-01 -6.18916214e-01
-7.21423447e-01 -1.91702560e-01 -8.59308541e-02 -7.42151082e-01
7.15767816e-02 8.95866275e-01 2.47907072e-01 5.01321293e-02
6.05022371e-01 2.15725228e-01 2.98746884e-01 -5.85626483e-01
1.22927599e-01 1.03518283e+00 2.71647692e-01 -2.83129215e-01
6.16524398e-01 3.54708850e-01 -2.28995502e-01 -8.01090717e-01
-8.66551697e-01 -9.15357471e-01 -4.40087676e-01 -5.56579009e-02
9.58013833e-01 -1.10105395e+00 -7.58999661e-02 2.27848217e-01
-1.10392654e+00 3.60147774e-01 -1.05444655e-01 5.27963281e-01
1.09196000e-01 6.63371027e-01 -5.72100341e-01 -4.09727335e-01
-3.68594319e-01 -1.23320580e+00 1.03965628e+00 3.08650702e-01
2.36846153e-02 -9.47576165e-01 9.78577286e-02 4.83428895e-01
4.15412664e-01 1.16601279e-02 9.52717066e-01 -6.45272255e-01
-5.38846731e-01 -5.51633775e-01 -4.76821393e-01 5.73855460e-01
7.01211095e-01 -3.19162667e-01 -6.94773495e-01 -3.12027246e-01
7.58974776e-02 -4.37639534e-01 8.68596911e-01 2.01385200e-01
1.12054360e+00 -1.78398445e-01 -4.29512501e-01 3.06764513e-01
1.53118646e+00 2.13849351e-01 4.52703506e-01 4.40574378e-01
6.57742143e-01 6.56262100e-01 7.49953330e-01 2.59944439e-01
4.66248393e-01 3.33197176e-01 2.26334766e-01 -4.13581252e-01
-1.25810519e-01 3.47579122e-02 6.38522254e-03 1.31618834e+00
7.78399855e-02 -1.25795856e-01 -1.04827154e+00 8.24838340e-01
-1.90132236e+00 -6.15532339e-01 2.28594601e-01 2.08280134e+00
9.61913526e-01 -3.93210799e-01 -2.18832538e-01 -1.88467070e-01
5.79654872e-01 -1.06961586e-01 -5.49400151e-01 1.35256439e-01
-9.61143523e-03 2.29429007e-01 1.69742033e-01 -7.63615891e-02
-9.96088803e-01 5.39823234e-01 6.22660446e+00 7.60204673e-01
-1.29482126e+00 2.83823878e-01 5.12766778e-01 1.61739495e-02
-4.74341184e-01 -3.33101273e-01 -3.73863816e-01 4.94029671e-01
6.47641122e-01 -2.85803467e-01 -1.39688272e-02 7.07384169e-01
-2.63291240e-01 -9.79193673e-02 -1.07511508e+00 1.40726399e+00
3.35371852e-01 -1.38079762e+00 4.92588818e-01 -2.73141786e-02
7.01796055e-01 -5.13188243e-02 3.71477127e-01 1.54372320e-01
-1.08901136e-01 -1.00467765e+00 -6.87167719e-02 5.19028485e-01
8.68114889e-01 -4.13392574e-01 9.62391913e-01 3.13644201e-01
-1.02399790e+00 9.32401046e-02 -5.18951058e-01 5.54712176e-01
-9.58319977e-02 5.26601613e-01 -8.18425894e-01 9.48241949e-01
7.04757750e-01 9.73850131e-01 -6.20589256e-01 1.30054069e+00
9.39635336e-02 3.15233395e-02 -1.90757468e-01 2.76068658e-01
1.77349880e-01 -7.42908791e-02 1.43232018e-01 8.54845941e-01
5.23769200e-01 2.07202643e-01 2.75159240e-01 3.07116061e-01
-1.94565862e-01 3.71923417e-01 -7.89053380e-01 -2.88228363e-01
2.39852279e-01 1.09325933e+00 -4.00857896e-01 -3.85339856e-01
-7.98180878e-01 1.17144084e+00 -2.32570097e-02 2.86462575e-01
-5.97170115e-01 -1.68921039e-01 5.82726300e-01 7.44473189e-02
-2.36922085e-01 5.10118110e-03 1.97415531e-01 -1.34758508e+00
-8.60110819e-02 -1.04952407e+00 6.06699765e-01 -7.23028779e-01
-1.70296514e+00 6.65792763e-01 -2.37751499e-01 -1.78525174e+00
-8.87655765e-02 -4.18704510e-01 -6.78821132e-02 8.43042850e-01
-2.04953623e+00 -1.10210943e+00 -2.53513545e-01 9.33436096e-01
5.37255287e-01 -2.61631280e-01 9.62919116e-01 8.44288886e-01
-4.42818880e-01 5.32190382e-01 3.46037447e-01 2.78146416e-01
1.23136687e+00 -8.78992260e-01 -3.39198232e-01 5.24199426e-01
-8.03477503e-03 7.78666377e-01 1.64180189e-01 -5.18671215e-01
-1.38353086e+00 -1.02413893e+00 7.56469607e-01 -3.09005767e-01
4.65263069e-01 4.63894814e-01 -1.14037943e+00 3.29677105e-01
1.08343512e-01 1.71749413e-01 1.14845598e+00 -2.49414042e-01
-2.33400643e-01 -1.56080663e-01 -1.15575695e+00 5.66234529e-01
5.44317305e-01 -7.41382122e-01 -7.09536374e-01 5.81113338e-01
7.43827999e-01 -2.16316760e-01 -1.10163212e+00 4.17708158e-01
3.76365423e-01 -4.13736433e-01 1.13368094e+00 -5.68501174e-01
6.86551630e-01 -2.59432018e-01 -2.60838687e-01 -1.12524831e+00
-1.49372676e-02 -1.51713798e-03 1.09410733e-01 8.52142930e-01
1.27419457e-01 -7.19369113e-01 3.67638141e-01 5.93316734e-01
1.64366901e-01 -7.89044380e-01 -8.42767417e-01 -4.75862384e-01
9.65169165e-03 3.10617983e-02 4.23031747e-01 1.30246723e+00
-1.37846291e-01 2.22512305e-01 -3.68723273e-01 2.02484861e-01
4.18517798e-01 2.85835385e-01 5.94965458e-01 -1.28384352e+00
-1.70439497e-01 -1.14078365e-01 -4.62952375e-01 -1.09721363e+00
-1.66552767e-01 -9.74418104e-01 -1.94012731e-01 -2.02631283e+00
6.65758610e-01 -6.79790437e-01 -6.74190462e-01 5.64779997e-01
-4.16884184e-01 3.93693298e-01 1.50021732e-01 7.79872835e-01
-6.41265452e-01 2.96120614e-01 1.72383523e+00 -3.68769288e-01
1.24727786e-01 -2.64751434e-01 -9.27263081e-01 6.29082263e-01
5.80624104e-01 -3.98772717e-01 -6.21136189e-01 -8.22418749e-01
2.68598795e-01 2.54900724e-01 1.83483183e-01 -9.16444182e-01
5.02158105e-01 5.09243570e-02 3.39144915e-01 -4.03680891e-01
3.94123822e-01 -1.05930305e+00 2.57241112e-07 4.45439488e-01
-8.69026184e-02 1.12788692e-01 1.80320710e-01 6.37406588e-01
-7.54872620e-01 -2.23756850e-01 7.82763720e-01 -3.84684086e-01
-7.54822195e-01 5.93064308e-01 -1.44802898e-01 -1.11415926e-02
1.09840095e+00 -1.06433302e-01 -3.64736736e-01 -2.90790439e-01
-7.17068136e-01 4.82965946e-01 3.90306234e-01 5.99966347e-01
1.06642818e+00 -1.36310387e+00 -7.00013518e-01 1.21401533e-01
4.07225162e-01 2.21592933e-01 5.23853600e-01 9.03368115e-01
-6.69111431e-01 2.96929002e-01 -1.56103045e-01 -7.40806997e-01
-1.49052775e+00 6.37354910e-01 8.52169245e-02 -4.42916691e-01
-6.42370701e-01 7.62017071e-01 3.19475144e-01 -3.16298693e-01
4.56429794e-02 -1.33075953e-01 -2.39418626e-01 2.33005285e-01
6.49489999e-01 -3.90895158e-02 -2.09784489e-02 -5.97708344e-01
-3.08359921e-01 8.56296480e-01 -4.67624396e-01 1.23021998e-01
1.28190804e+00 -1.78166613e-01 -2.84830123e-01 3.88284355e-01
1.40745521e+00 -3.98938388e-01 -4.92412448e-01 -5.72088122e-01
-2.69989371e-01 -5.36590993e-01 4.93906252e-02 -6.24584377e-01
-1.13310933e+00 9.99556005e-01 8.32809865e-01 1.02361128e-01
1.34507799e+00 2.30228409e-01 1.17094743e+00 3.48448396e-01
3.86986911e-01 -9.80930507e-01 5.16768813e-01 1.19168483e-01
7.43633449e-01 -1.64084649e+00 6.31241202e-02 -1.76029980e-01
-8.84636641e-01 9.29373205e-01 4.39218521e-01 8.82385671e-02
8.66094887e-01 -2.38259554e-01 3.86153609e-01 -2.68432111e-01
-4.25542116e-01 -2.65835583e-01 3.77438992e-01 3.54529470e-01
3.03370386e-01 -3.62616517e-02 -3.39665920e-01 3.91961783e-01
4.29537177e-01 2.00114116e-01 1.73332229e-01 1.10845804e+00
-3.46303225e-01 -1.47455561e+00 -3.09462726e-01 7.09965348e-01
-7.29852378e-01 -1.67445838e-01 -1.97477207e-01 5.80491722e-01
1.66332081e-01 8.04041922e-01 -1.70274764e-01 -2.48154014e-01
2.15285107e-01 -1.00665711e-01 4.44563627e-01 -8.70650828e-01
-3.29538047e-01 1.29527614e-01 -4.03573334e-01 -1.63784802e-01
-8.39846313e-01 -3.42561990e-01 -1.10672474e+00 2.63563846e-03
-2.35527799e-01 6.46448433e-02 4.91056681e-01 7.47636199e-01
3.76689672e-01 6.21326387e-01 4.97529000e-01 -2.10979447e-01
-4.58398581e-01 -6.65583670e-01 -4.98103231e-01 6.23107970e-01
4.59164262e-01 -5.03495276e-01 -1.15887366e-01 8.78760740e-02] | [14.449727058410645, -1.6606078147888184] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.